-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/boot/Makefile linux-2.6.18-xen.hg/arch/i386/boot/Makefile
---- linux-2.6.18/arch/i386/boot/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/boot/Makefile 2007-12-23 11:14:54.372518519 +0100
-@@ -26,7 +26,7 @@
- #RAMDISK := -DRAMDISK=512
-
- targets := vmlinux.bin bootsect bootsect.o \
-- setup setup.o zImage bzImage
-+ setup setup.o zImage bzImage vmlinuz vmlinux-stripped
- subdir- := compressed
-
- hostprogs-y := tools/build
-@@ -133,5 +133,13 @@
- cp System.map $(INSTALL_PATH)/
- if [ -x /sbin/lilo ]; then /sbin/lilo; else /etc/lilo/install; fi
-
-+$(obj)/vmlinuz: $(obj)/vmlinux-stripped FORCE
-+ $(call if_changed,gzip)
-+ @echo 'Kernel: $@ is ready' ' (#'`cat .version`')'
-+
-+$(obj)/vmlinux-stripped: OBJCOPYFLAGS := -g --strip-unneeded
-+$(obj)/vmlinux-stripped: vmlinux FORCE
-+ $(call if_changed,objcopy)
-+
- install:
- sh $(srctree)/$(src)/install.sh $(KERNELRELEASE) $(BOOTIMAGE) System.map "$(INSTALL_PATH)"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/Kconfig linux-2.6.18-xen.hg/arch/i386/Kconfig
---- linux-2.6.18/arch/i386/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/Kconfig 2007-12-23 11:14:54.369185012 +0100
+--- linux-2.6.18.8/Documentation/i2c/busses/i2c-i801 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/Documentation/i2c/busses/i2c-i801 2008-05-19 00:32:37.654858195 +0300
+@@ -10,6 +10,11 @@
+ * Intel 6300ESB
+ * Intel 82801FB/FR/FW/FRW (ICH6)
+ * Intel ICH7
++ * Intel 82801G (ICH7)
++ * Intel 82801H (ICH9)
++ * Intel 82801I (ICH9)
++ * Intel Tolapai
++ * Intel ICH10
+ Datasheets: Publicly available at the Intel website
+
+ Authors:
+--- linux-2.6.18.8/MAINTAINERS 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/MAINTAINERS 2008-05-19 00:32:43.775210968 +0300
+@@ -2558,6 +2558,13 @@
+ L: linux-ia64@vger.kernel.org
+ S: Supported
+
++SFC NETWORK DRIVER
++P: Steve Hodgson
++P: Ben Hutchings
++P: Robert Stonehouse
++M: linux-net-drivers@solarflare.com
++S: Supported
++
+ SGI VISUAL WORKSTATION 320 AND 540
+ P: Andrey Panin
+ M: pazke@donpac.ru
+--- linux-2.6.18.8/arch/i386/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/Kconfig 2008-05-19 00:33:01.816250845 +0300
@@ -16,6 +16,7 @@
config GENERIC_TIME
default y
config LOCKDEP_SUPPORT
-@@ -103,6 +104,15 @@
+@@ -103,6 +104,16 @@
help
Choose this option if your computer is a standard PC or compatible.
+config X86_XEN
+ bool "Xen-compatible"
++ select XEN
+ select X86_UP_APIC if !SMP && XEN_PRIVILEGED_GUEST
+ select X86_UP_IOAPIC if !SMP && XEN_PRIVILEGED_GUEST
+ select SWIOTLB
config X86_ELAN
bool "AMD Elan"
help
-@@ -213,6 +223,7 @@
+@@ -213,6 +224,7 @@
config HPET_TIMER
bool "HPET Timer Support"
help
This enables the use of the HPET for the kernel's internal timer.
HPET is the next generation timer replacing legacy 8254s.
-@@ -263,7 +274,7 @@
+@@ -263,7 +275,7 @@
config X86_UP_APIC
bool "Local APIC support on uniprocessors"
help
A local APIC (Advanced Programmable Interrupt Controller) is an
integrated interrupt controller in the CPU. If you have a single-CPU
-@@ -288,12 +299,12 @@
+@@ -288,12 +300,12 @@
config X86_LOCAL_APIC
bool
default y
config X86_VISWS_APIC
-@@ -303,7 +314,7 @@
+@@ -303,7 +315,7 @@
config X86_MCE
bool "Machine Check Exception"
---help---
Machine Check Exception support allows the processor to notify the
kernel if it detects a problem (e.g. overheating, component failure).
-@@ -402,6 +413,7 @@
+@@ -402,6 +414,7 @@
config MICROCODE
tristate "/dev/cpu/microcode - Intel IA32 CPU microcode support"
---help---
If you say Y here and also to "/dev file system support" in the
'File systems' section, you will be able to update the microcode on
-@@ -434,6 +446,10 @@
+@@ -434,6 +447,10 @@
with major 203 and minors 0 to 31 for /dev/cpu/0/cpuid to
/dev/cpu/31/cpuid.
source "drivers/firmware/Kconfig"
choice
-@@ -616,6 +632,7 @@
+@@ -616,6 +633,7 @@
config MATH_EMULATION
bool "Math emulation"
---help---
Linux can emulate a math coprocessor (used for floating point
operations) if you don't have one. 486DX and Pentium processors have
-@@ -641,6 +658,8 @@
+@@ -641,6 +659,8 @@
config MTRR
bool "MTRR (Memory Type Range Register) support"
---help---
On Intel P6 family processors (Pentium Pro, Pentium II and later)
the Memory Type Range Registers (MTRRs) may be used to control
-@@ -675,7 +694,7 @@
+@@ -675,7 +695,7 @@
config EFI
bool "Boot from EFI support"
default n
---help---
This enables the the kernel to boot on EFI platforms using
-@@ -693,7 +712,7 @@
+@@ -693,7 +713,7 @@
config IRQBALANCE
bool "Enable kernel irq balancing"
default y
help
The default yes will allow the kernel to do irq load balancing.
-@@ -741,7 +760,7 @@
+@@ -741,7 +761,7 @@
config KEXEC
bool "kexec system call (EXPERIMENTAL)"
help
kexec is a system call that implements the ability to shutdown your
current kernel, and to start another kernel. It is like a reboot
-@@ -793,6 +812,7 @@
+@@ -793,6 +813,7 @@
config COMPAT_VDSO
bool "Compat VDSO support"
default y
help
Map the VDSO to the predictable old-style address too.
-@@ -810,18 +830,18 @@
+@@ -810,18 +831,18 @@
depends on HIGHMEM
menu "Power management options (ACPI, APM)"
---help---
APM is a BIOS specification for saving power using several different
techniques. This is mostly useful for battery powered laptops with
-@@ -1006,6 +1026,7 @@
+@@ -1006,6 +1027,7 @@
config PCI_GOBIOS
bool "BIOS"
config PCI_GOMMCONFIG
bool "MMConfig"
-@@ -1013,6 +1034,13 @@
+@@ -1013,6 +1035,13 @@
config PCI_GODIRECT
bool "Direct"
config PCI_GOANY
bool "Any"
-@@ -1020,7 +1048,7 @@
+@@ -1020,7 +1049,7 @@
config PCI_BIOS
bool
default y
config PCI_DIRECT
-@@ -1033,6 +1061,18 @@
+@@ -1033,6 +1062,19 @@
depends on PCI && ACPI && (PCI_GOMMCONFIG || PCI_GOANY)
default y
+config XEN_PCIDEV_FRONTEND
+ bool
+ depends on PCI && X86_XEN && (PCI_GOXEN_FE || PCI_GOANY)
++ select HOTPLUG
+ default y
+
+config XEN_PCIDEV_FE_DEBUG
source "drivers/pci/pcie/Kconfig"
source "drivers/pci/Kconfig"
-@@ -1043,7 +1083,7 @@
+@@ -1043,7 +1085,7 @@
config ISA
bool "ISA support"
help
Find out whether you have ISA slots on your motherboard. ISA is the
name of a bus system, i.e. the way the CPU talks to the other stuff
-@@ -1070,7 +1110,7 @@
+@@ -1070,7 +1112,7 @@
source "drivers/eisa/Kconfig"
config MCA
default y if X86_VOYAGER
help
MicroChannel Architecture is found in some IBM PS/2 machines and
-@@ -1146,6 +1186,8 @@
+@@ -1146,6 +1188,8 @@
source "crypto/Kconfig"
source "lib/Kconfig"
#
-@@ -1171,7 +1213,7 @@
+@@ -1171,7 +1215,7 @@
config X86_HT
bool
default y
config X86_BIOS_REBOOT
-@@ -1184,6 +1226,16 @@
+@@ -1182,6 +1226,17 @@
+ config X86_TRAMPOLINE
+ bool
depends on X86_SMP || (X86_VOYAGER && SMP)
- default y
-
++ depends on !XEN
++ default y
++
+config X86_NO_TSS
+ bool
+ depends on X86_XEN
+config X86_NO_IDT
+ bool
+ depends on X86_XEN
-+ default y
-+
- config KTIME_SCALAR
- bool
default y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/Kconfig.cpu linux-2.6.18-xen.hg/arch/i386/Kconfig.cpu
---- linux-2.6.18/arch/i386/Kconfig.cpu 2007-12-23 11:26:51.000120088 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/Kconfig.cpu 2007-12-23 11:14:54.369185012 +0100
+
+ config KTIME_SCALAR
+--- linux-2.6.18.8/arch/i386/Kconfig.cpu 2008-05-19 00:42:33.777222817 +0300
++++ linux-2.6.18-xen.hg/arch/i386/Kconfig.cpu 2008-05-19 00:33:01.816250845 +0300
@@ -252,7 +252,7 @@
config X86_F00F_BUG
- depends on (MWINCHIP3D || MWINCHIP2 || MCRUSOE || MEFFICEON || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMM || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MK8 || MVIAC3_2 || MGEODEGX1 || MGEODE_LX) && !X86_NUMAQ
+ depends on (MWINCHIP3D || MWINCHIP2 || MCRUSOE || MEFFICEON || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMM || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MK8 || MVIAC3_2 || MGEODEGX1 || MGEODE_LX) && !X86_NUMAQ && !X86_XEN
default y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/Kconfig.debug linux-2.6.18-xen.hg/arch/i386/Kconfig.debug
---- linux-2.6.18/arch/i386/Kconfig.debug 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/Kconfig.debug 2007-12-23 11:14:54.369185012 +0100
+--- linux-2.6.18.8/arch/i386/Kconfig.debug 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/Kconfig.debug 2008-05-19 00:33:01.864253612 +0300
@@ -79,6 +79,7 @@
config DOUBLEFAULT
default y
help
This option allows trapping of rare doublefault exceptions that
would otherwise cause a system to silently reboot. Disabling this
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/acpi/boot-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/acpi/boot-xen.c
---- linux-2.6.18/arch/i386/kernel/acpi/boot-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/boot-xen.c 2007-12-23 11:14:54.382519030 +0100
+--- linux-2.6.18.8/arch/i386/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/Makefile 2008-05-19 00:33:01.868253842 +0300
+@@ -71,6 +71,10 @@
+ mflags-$(CONFIG_X86_SUMMIT) := -Iinclude/asm-i386/mach-summit
+ mcore-$(CONFIG_X86_SUMMIT) := mach-default
+
++# Xen subarch support
++mflags-$(CONFIG_X86_XEN) := -Iinclude/asm-i386/mach-xen
++mcore-$(CONFIG_X86_XEN) := mach-xen
++
+ # generic subarchitecture
+ mflags-$(CONFIG_X86_GENERICARCH) := -Iinclude/asm-i386/mach-generic
+ mcore-$(CONFIG_X86_GENERICARCH) := mach-default
+@@ -102,9 +106,20 @@
+
+ boot := arch/i386/boot
+
+-PHONY += zImage bzImage compressed zlilo bzlilo \
++PHONY += zImage bzImage vmlinuz compressed zlilo bzlilo \
+ zdisk bzdisk fdimage fdimage144 fdimage288 isoimage install
+
++ifdef CONFIG_XEN
++CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
++ -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(CPPFLAGS)
++all: vmlinuz
++
++# KBUILD_IMAGE specifies the target image being built
++KBUILD_IMAGE := $(boot)/vmlinuz
++
++vmlinuz: vmlinux
++ $(Q)$(MAKE) $(build)=$(boot) $(KBUILD_IMAGE)
++else
+ all: bzImage
+
+ # KBUILD_IMAGE specify target image being built
+@@ -124,6 +139,7 @@
+
+ fdimage fdimage144 fdimage288 isoimage: vmlinux
+ $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(KBUILD_IMAGE) $@
++endif
+
+ install:
+ $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(KBUILD_IMAGE) install
+--- linux-2.6.18.8/arch/i386/boot/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/boot/Makefile 2008-05-19 00:33:01.952258684 +0300
+@@ -26,7 +26,7 @@
+ #RAMDISK := -DRAMDISK=512
+
+ targets := vmlinux.bin bootsect bootsect.o \
+- setup setup.o zImage bzImage
++ setup setup.o zImage bzImage vmlinuz vmlinux-stripped
+ subdir- := compressed
+
+ hostprogs-y := tools/build
+@@ -133,5 +133,13 @@
+ cp System.map $(INSTALL_PATH)/
+ if [ -x /sbin/lilo ]; then /sbin/lilo; else /etc/lilo/install; fi
+
++$(obj)/vmlinuz: $(obj)/vmlinux-stripped FORCE
++ $(call if_changed,gzip)
++ @echo 'Kernel: $@ is ready' ' (#'`cat .version`')'
++
++$(obj)/vmlinux-stripped: OBJCOPYFLAGS := -g --strip-unneeded
++$(obj)/vmlinux-stripped: vmlinux FORCE
++ $(call if_changed,objcopy)
++
+ install:
+ sh $(srctree)/$(src)/install.sh $(KERNELRELEASE) $(BOOTIMAGE) System.map "$(INSTALL_PATH)"
+--- linux-2.6.18.8/arch/i386/kernel/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/Makefile 2008-05-19 00:33:02.540292579 +0300
+@@ -43,6 +43,7 @@
+ EXTRA_AFLAGS := -traditional
+
+ obj-$(CONFIG_SCx200) += scx200.o
++obj-$(CONFIG_XEN) += fixup.o
+
+ # vsyscall.o contains the vsyscall DSO images as __initdata.
+ # We must build both images before we can assemble it.
+@@ -80,5 +81,8 @@
+ $(obj)/vsyscall-sysenter.o $(obj)/vsyscall-note.o FORCE
+ $(call if_changed,syscall)
+
++early_printk-y += ../../x86_64/kernel/early_printk.o
+ k8-y += ../../x86_64/kernel/k8.o
+
++disabled-obj-$(CONFIG_XEN) := i8259.o reboot.o smpboot.o
++%/head.o %/head.s: $(if $(CONFIG_XEN),EXTRA_AFLAGS,dummy) :=
+--- linux-2.6.18.8/arch/i386/kernel/acpi/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/Makefile 2008-05-19 00:33:02.540292579 +0300
+@@ -4,5 +4,7 @@
+
+ ifneq ($(CONFIG_ACPI_PROCESSOR),)
+ obj-y += cstate.o processor.o
++obj-$(CONFIG_XEN) += processor_extcntl_xen.o
+ endif
+
++disabled-obj-$(CONFIG_XEN) := cstate.o wakeup.o
+--- linux-2.6.18.8/arch/i386/kernel/acpi/boot-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/boot-xen.c 2008-05-19 00:33:02.560293731 +0300
@@ -0,0 +1,1168 @@
+/*
+ * boot.c - Architecture-Specific Low-Level ACPI Boot Support
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/acpi/Makefile linux-2.6.18-xen.hg/arch/i386/kernel/acpi/Makefile
---- linux-2.6.18/arch/i386/kernel/acpi/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/Makefile 2007-12-23 11:14:54.382519030 +0100
-@@ -6,3 +6,4 @@
- obj-y += cstate.o processor.o
- endif
+--- linux-2.6.18.8/arch/i386/kernel/acpi/processor.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/processor.c 2008-05-19 00:33:02.676300418 +0300
+@@ -62,7 +62,18 @@
+ /* Initialize _PDC data based on the CPU vendor */
+ void arch_acpi_processor_init_pdc(struct acpi_processor *pr)
+ {
++#ifdef CONFIG_XEN
++ /*
++ * As a work-around, just use cpu0's cpuinfo for all processors.
++ * Further work is required to expose xen hypervisor interface of
++ * getting physical cpuinfo to dom0 kernel and then
++ * arch_acpi_processor_init_pdc can set _PDC parameters according
++ * to Xen's phys information.
++ */
++ unsigned int cpu = 0;
++#else
+ unsigned int cpu = pr->id;
++#endif /* CONFIG_XEN */
+ struct cpuinfo_x86 *c = cpu_data + cpu;
-+disabled-obj-$(CONFIG_XEN) := cstate.o wakeup.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/acpi/sleep-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/acpi/sleep-xen.c
---- linux-2.6.18/arch/i386/kernel/acpi/sleep-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/sleep-xen.c 2007-12-23 11:14:54.385852540 +0100
-@@ -0,0 +1,134 @@
+ pr->pdc = NULL;
+--- linux-2.6.18.8/arch/i386/kernel/acpi/processor_extcntl_xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/processor_extcntl_xen.c 2008-05-19 00:33:02.676300418 +0300
+@@ -0,0 +1,235 @@
++/*
++ * processor_extcntl_xen.c - interface to notify Xen
++ *
++ * Copyright (C) 2008, Intel corporation
++ *
++ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or (at
++ * your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful, but
++ * WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
++ * General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License along
++ * with this program; if not, write to the Free Software Foundation, Inc.,
++ * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
++ *
++ */
++
++#include <linux/kernel.h>
++#include <linux/init.h>
++#include <linux/types.h>
++#include <linux/acpi.h>
++#include <linux/pm.h>
++#include <linux/cpu.h>
++
++#include <linux/cpufreq.h>
++#include <acpi/processor.h>
++#include <asm/hypercall.h>
++
++static int xen_processor_pmbits;
++static int __init set_xen_processor_pmbits(char *str)
++{
++ get_option(&str, &xen_processor_pmbits);
++
++ return 1;
++}
++__setup("xen_processor_pmbits=", set_xen_processor_pmbits);
++EXPORT_SYMBOL(xen_processor_pmbits);
++
++static int xen_cx_notifier(struct acpi_processor *pr, int action)
++{
++ int ret, count = 0, i;
++ xen_platform_op_t op = {
++ .cmd = XENPF_set_processor_pminfo,
++ .interface_version = XENPF_INTERFACE_VERSION,
++ .u.set_pminfo.id = pr->acpi_id,
++ .u.set_pminfo.type = XEN_PM_CX,
++ };
++ struct xen_processor_cx *data, *buf;
++ struct acpi_processor_cx *cx;
++
++ if (action == PROCESSOR_PM_CHANGE)
++ return -EINVAL;
++
++ /* Convert to Xen defined structure and hypercall */
++ buf = kzalloc(pr->power.count * sizeof(struct xen_processor_cx),
++ GFP_KERNEL);
++ if (!buf)
++ return -ENOMEM;
++
++ data = buf;
++ for (i = 1; i <= pr->power.count; i++) {
++ cx = &pr->power.states[i];
++ /* Skip invalid cstate entry */
++ if (!cx->valid)
++ continue;
++
++ data->type = cx->type;
++ data->latency = cx->latency;
++ data->power = cx->power;
++ data->reg.space_id = cx->reg.space_id;
++ data->reg.bit_width = cx->reg.bit_width;
++ data->reg.bit_offset = cx->reg.bit_offset;
++ data->reg.access_size = cx->reg.reserved;
++ data->reg.address = cx->reg.address;
++
++ /* Get dependency relationships */
++ if (cx->csd_count) {
++ printk("Wow! _CSD is found. Not support for now!\n");
++ kfree(buf);
++ return -EINVAL;
++ } else {
++ data->dpcnt = 0;
++ set_xen_guest_handle(data->dp, NULL);
++ }
++
++ data++;
++ count++;
++ }
++
++ if (!count) {
++ printk("No available Cx info for cpu %d\n", pr->acpi_id);
++ kfree(buf);
++ return -EINVAL;
++ }
++
++ op.u.set_pminfo.power.count = count;
++ op.u.set_pminfo.power.flags.bm_control = pr->flags.bm_control;
++ op.u.set_pminfo.power.flags.bm_check = pr->flags.bm_check;
++ op.u.set_pminfo.power.flags.has_cst = pr->flags.has_cst;
++ op.u.set_pminfo.power.flags.power_setup_done = pr->flags.power_setup_done;
++
++ set_xen_guest_handle(op.u.set_pminfo.power.states, buf);
++ ret = HYPERVISOR_platform_op(&op);
++ kfree(buf);
++ return ret;
++}
++
++static void convert_pct_reg(struct xen_pct_register *xpct,
++ struct acpi_pct_register *apct)
++{
++ xpct->descriptor = apct->descriptor;
++ xpct->length = apct->length;
++ xpct->space_id = apct->space_id;
++ xpct->bit_width = apct->bit_width;
++ xpct->bit_offset = apct->bit_offset;
++ xpct->reserved = apct->reserved;
++ xpct->address = apct->address;
++}
++
++static void convert_pss_states(struct xen_processor_px *xpss,
++ struct acpi_processor_px *apss, int state_count)
++{
++ int i;
++ for(i=0; i<state_count; i++) {
++ xpss->core_frequency = apss->core_frequency;
++ xpss->power = apss->power;
++ xpss->transition_latency = apss->transition_latency;
++ xpss->bus_master_latency = apss->bus_master_latency;
++ xpss->control = apss->control;
++ xpss->status = apss->status;
++ xpss++;
++ apss++;
++ }
++}
++
++static void convert_psd_pack(struct xen_psd_package *xpsd,
++ struct acpi_psd_package *apsd)
++{
++ xpsd->num_entries = apsd->num_entries;
++ xpsd->revision = apsd->revision;
++ xpsd->domain = apsd->domain;
++ xpsd->coord_type = apsd->coord_type;
++ xpsd->num_processors = apsd->num_processors;
++}
++
++static int xen_px_notifier(struct acpi_processor *pr, int action)
++{
++ int ret;
++ xen_platform_op_t op = {
++ .cmd = XENPF_set_processor_pminfo,
++ .interface_version = XENPF_INTERFACE_VERSION,
++ .u.set_pminfo.id = pr->acpi_id,
++ .u.set_pminfo.type = XEN_PM_PX,
++ };
++ struct xen_processor_performance *perf;
++ struct xen_processor_px *states = NULL;
++ struct acpi_processor_performance *px;
++ struct acpi_psd_package *pdomain;
++
++ /* leave dynamic ppc handle in the future */
++ if (action == PROCESSOR_PM_CHANGE)
++ return 0;
++
++ perf = &op.u.set_pminfo.perf;
++ px = pr->performance;
++
++ perf->flags = XEN_PX_PPC |
++ XEN_PX_PCT |
++ XEN_PX_PSS |
++ XEN_PX_PSD;
++
++ /* ppc */
++ perf->ppc = pr->performance_platform_limit;
++
++ /* pct */
++ convert_pct_reg(&perf->control_register, &px->control_register);
++ convert_pct_reg(&perf->status_register, &px->status_register);
++
++ /* pss */
++ perf->state_count = px->state_count;
++ states = kzalloc(px->state_count*sizeof(xen_processor_px_t),GFP_KERNEL);
++ if (!states)
++ return -ENOMEM;
++ convert_pss_states(states, px->states, px->state_count);
++ set_xen_guest_handle(perf->states, states);
++
++ /* psd */
++ pdomain = &px->domain_info;
++ convert_psd_pack(&perf->domain_info, pdomain);
++ if (perf->domain_info.num_processors) {
++ if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ALL)
++ perf->shared_type = CPUFREQ_SHARED_TYPE_ALL;
++ else if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ANY)
++ perf->shared_type = CPUFREQ_SHARED_TYPE_ANY;
++ else if (pdomain->coord_type == DOMAIN_COORD_TYPE_HW_ALL)
++ perf->shared_type = CPUFREQ_SHARED_TYPE_HW;
++ } else
++ perf->shared_type = CPUFREQ_SHARED_TYPE_NONE;
++
++ ret = HYPERVISOR_platform_op(&op);
++ kfree(states);
++ return ret;
++}
++
++static int xen_tx_notifier(struct acpi_processor *pr, int action)
++{
++ return -EINVAL;
++}
++static int xen_hotplug_notifier(struct acpi_processor *pr, int event)
++{
++ return -EINVAL;
++}
++
++static struct processor_extcntl_ops xen_extcntl_ops = {
++ .hotplug = xen_hotplug_notifier,
++};
++
++static int __cpuinit xen_init_processor_extcntl(void)
++{
++ if (xen_processor_pmbits & XEN_PROCESSOR_PM_CX)
++ xen_extcntl_ops.pm_ops[PM_TYPE_IDLE] = xen_cx_notifier;
++ if (xen_processor_pmbits & XEN_PROCESSOR_PM_PX)
++ xen_extcntl_ops.pm_ops[PM_TYPE_PERF] = xen_px_notifier;
++ if (xen_processor_pmbits & XEN_PROCESSOR_PM_TX)
++ xen_extcntl_ops.pm_ops[PM_TYPE_THR] = xen_tx_notifier;
++
++ return processor_register_extcntl(&xen_extcntl_ops);
++}
++core_initcall(xen_init_processor_extcntl);
+--- linux-2.6.18.8/arch/i386/kernel/acpi/sleep-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/acpi/sleep-xen.c 2008-05-19 00:33:02.676300418 +0300
+@@ -0,0 +1,113 @@
+/*
+ * sleep.c - x86-specific ACPI sleep support.
+ *
+}
+
+core_initcall(acpisleep_dmi_init);
-+
-+#else /* CONFIG_ACPI_PV_SLEEP */
-+#include <asm/hypervisor.h>
-+#include <xen/interface/platform.h>
-+int acpi_notify_hypervisor_state(u8 sleep_state,
-+ u32 pm1a_cnt, u32 pm1b_cnt)
-+{
-+ struct xen_platform_op op = {
-+ .cmd = XENPF_enter_acpi_sleep,
-+ .interface_version = XENPF_INTERFACE_VERSION,
-+ .u = {
-+ .enter_acpi_sleep = {
-+ .pm1a_cnt_val = (u16)pm1a_cnt,
-+ .pm1b_cnt_val = (u16)pm1b_cnt,
-+ .sleep_state = sleep_state,
-+ },
-+ },
-+ };
-+
-+ return HYPERVISOR_platform_op(&op);
-+}
+#endif /* CONFIG_ACPI_PV_SLEEP */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/apic-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/apic-xen.c
---- linux-2.6.18/arch/i386/kernel/apic-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/apic-xen.c 2007-12-23 11:14:54.389186058 +0100
+--- linux-2.6.18.8/arch/i386/kernel/apic-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/apic-xen.c 2008-05-19 00:33:02.688301110 +0300
@@ -0,0 +1,155 @@
+/*
+ * Local APIC handling, local APIC timers
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/asm-offsets.c linux-2.6.18-xen.hg/arch/i386/kernel/asm-offsets.c
---- linux-2.6.18/arch/i386/kernel/asm-offsets.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/asm-offsets.c 2007-12-23 11:14:54.392519572 +0100
+--- linux-2.6.18.8/arch/i386/kernel/asm-offsets.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/asm-offsets.c 2008-05-19 00:33:02.696301571 +0300
@@ -66,9 +66,14 @@
OFFSET(pbe_orig_address, pbe, orig_address);
OFFSET(pbe_next, pbe, next);
DEFINE(PAGE_SIZE_asm, PAGE_SIZE);
DEFINE(VDSO_PRELINK, VDSO_PRELINK);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/common-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/cpu/common-xen.c
---- linux-2.6.18/arch/i386/kernel/cpu/common-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/common-xen.c 2007-12-23 11:14:54.392519572 +0100
+--- linux-2.6.18.8/arch/i386/kernel/cpu/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/Makefile 2008-05-19 00:33:02.696301571 +0300
+@@ -17,3 +17,4 @@
+
+ obj-$(CONFIG_MTRR) += mtrr/
+ obj-$(CONFIG_CPU_FREQ) += cpufreq/
++
+--- linux-2.6.18.8/arch/i386/kernel/cpu/common-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/common-xen.c 2008-05-19 00:33:02.712302493 +0300
@@ -0,0 +1,743 @@
+#include <linux/init.h>
+#include <linux/string.h>
+ per_cpu(cpu_tlbstate, cpu).active_mm = &init_mm;
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/cpufreq/powernow-k8.c linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.c
---- linux-2.6.18/arch/i386/kernel/cpu/cpufreq/powernow-k8.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.c 2007-12-23 11:14:54.402520090 +0100
+--- linux-2.6.18.8/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c 2008-05-19 00:42:33.777222817 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c 2008-05-19 00:33:02.720302954 +0300
+@@ -568,6 +568,17 @@
+ {
+ dprintk("acpi_cpufreq_init\n");
+
++#ifdef CONFIG_XEN
++ /*
++ * This effectively blocks in-kernel cpufreq driver to interfere
++ * external control logic
++ */
++ if (processor_pmperf_external()) {
++ printk("CPUFREQ is controllerd externally...exit then!\n");
++ return -1;
++ }
++#endif /* CONFIG_XEN */
++
+ acpi_cpufreq_early_init_acpi();
+
+ return cpufreq_register_driver(&acpi_cpufreq_driver);
+--- linux-2.6.18.8/arch/i386/kernel/cpu/cpufreq/powernow-k8.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.c 2008-05-19 00:33:02.804307796 +0300
@@ -46,7 +46,7 @@
#define PFX "powernow-k8: "
printk(KERN_INFO PFX "Found %d %s "
"processors (" VERSION ")\n", supported_cpus,
boot_cpu_data.x86_model_id);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/cpufreq/powernow-k8.h linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.h
---- linux-2.6.18/arch/i386/kernel/cpu/cpufreq/powernow-k8.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.h 2007-12-23 11:14:54.405853593 +0100
+--- linux-2.6.18.8/arch/i386/kernel/cpu/cpufreq/powernow-k8.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/cpufreq/powernow-k8.h 2008-05-19 00:33:02.804307796 +0300
@@ -1,5 +1,5 @@
/*
- * (c) 2003-2006 Advanced Micro Devices, Inc.
/* define the two driver architectures */
#define CPU_OPTERON 0
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/Makefile linux-2.6.18-xen.hg/arch/i386/kernel/cpu/Makefile
---- linux-2.6.18/arch/i386/kernel/cpu/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/Makefile 2007-12-23 11:14:54.392519572 +0100
-@@ -17,3 +17,4 @@
+--- linux-2.6.18.8/arch/i386/kernel/cpu/mtrr/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/Makefile 2008-05-19 00:33:03.428343766 +0300
+@@ -3,3 +3,4 @@
+ obj-y += cyrix.o
+ obj-y += centaur.o
- obj-$(CONFIG_MTRR) += mtrr/
- obj-$(CONFIG_CPU_FREQ) += cpufreq/
-+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/mtrr/main-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/main-xen.c
---- linux-2.6.18/arch/i386/kernel/cpu/mtrr/main-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/main-xen.c 2007-12-23 11:14:54.419187634 +0100
-@@ -0,0 +1,197 @@
++obj-$(CONFIG_XEN) := main.o if.o
+--- linux-2.6.18.8/arch/i386/kernel/cpu/mtrr/main-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/main-xen.c 2008-05-19 00:33:03.476346532 +0300
+@@ -0,0 +1,198 @@
+#include <linux/init.h>
+#include <linux/proc_fs.h>
+#include <linux/ctype.h>
+
+ op.cmd = XENPF_read_memtype;
+ op.u.read_memtype.reg = reg;
-+ (void)HYPERVISOR_platform_op(&op);
++ if (unlikely(HYPERVISOR_platform_op(&op)))
++ memset(&op.u.read_memtype, 0, sizeof(op.u.read_memtype));
+
+ *size = op.u.read_memtype.nr_mfns;
+ *base = op.u.read_memtype.mfn;
+}
+
+subsys_initcall(mtrr_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/cpu/mtrr/Makefile linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/Makefile
---- linux-2.6.18/arch/i386/kernel/cpu/mtrr/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/cpu/mtrr/Makefile 2007-12-23 11:14:54.415854123 +0100
-@@ -3,3 +3,4 @@
- obj-y += cyrix.o
- obj-y += centaur.o
-
-+obj-$(CONFIG_XEN) := main.o if.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/crash.c linux-2.6.18-xen.hg/arch/i386/kernel/crash.c
---- linux-2.6.18/arch/i386/kernel/crash.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/crash.c 2007-12-23 11:14:54.422521143 +0100
+--- linux-2.6.18.8/arch/i386/kernel/crash.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/crash.c 2008-05-19 00:33:03.540350222 +0300
@@ -90,6 +90,7 @@
crash_save_this_cpu(regs, cpu);
}
+#endif /* CONFIG_XEN */
crash_save_self(regs);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/early_printk-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/early_printk-xen.c
---- linux-2.6.18/arch/i386/kernel/early_printk-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/early_printk-xen.c 2007-12-23 11:14:54.422521143 +0100
+--- linux-2.6.18.8/arch/i386/kernel/early_printk-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/early_printk-xen.c 2008-05-19 00:33:03.580352527 +0300
@@ -0,0 +1,2 @@
+
+#include "../../x86_64/kernel/early_printk-xen.c"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/entry.S linux-2.6.18-xen.hg/arch/i386/kernel/entry.S
---- linux-2.6.18/arch/i386/kernel/entry.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/entry.S 2007-12-23 11:14:54.425854647 +0100
-@@ -269,7 +269,7 @@
- CFI_STARTPROC simple
- CFI_DEF_CFA esp, 0
- CFI_REGISTER esp, ebp
-- movl TSS_sysenter_esp0(%esp),%esp
-+ movl SYSENTER_stack_esp0(%esp),%esp
- sysenter_past_esp:
- /*
- * No need to follow this irqs on/off section: the syscall
-@@ -689,7 +689,7 @@
- * that sets up the real kernel stack. Check here, since we can't
- * allow the wrong stack to be used.
- *
-- * "TSS_sysenter_esp0+12" is because the NMI/debug handler will have
-+ * "SYSENTER_stack_esp0+12" is because the NMI/debug handler will have
- * already pushed 3 words if it hits on the sysenter instruction:
- * eflags, cs and eip.
- *
-@@ -701,7 +701,7 @@
- cmpw $__KERNEL_CS,4(%esp); \
- jne ok; \
- label: \
-- movl TSS_sysenter_esp0+offset(%esp),%esp; \
-+ movl SYSENTER_stack_esp0+offset(%esp),%esp; \
- pushfl; \
- pushl $__KERNEL_CS; \
- pushl $sysenter_past_esp
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/entry-xen.S linux-2.6.18-xen.hg/arch/i386/kernel/entry-xen.S
---- linux-2.6.18/arch/i386/kernel/entry-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/entry-xen.S 2007-12-23 11:14:54.425854647 +0100
+--- linux-2.6.18.8/arch/i386/kernel/entry-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/entry-xen.S 2008-05-19 00:33:03.584352758 +0300
@@ -0,0 +1,1238 @@
+/*
+ * linux/arch/i386/entry.S
+#include "syscall_table.S"
+
+syscall_table_size=(.-sys_call_table)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/fixup.c linux-2.6.18-xen.hg/arch/i386/kernel/fixup.c
---- linux-2.6.18/arch/i386/kernel/fixup.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/fixup.c 2007-12-23 11:14:54.429188153 +0100
+--- linux-2.6.18.8/arch/i386/kernel/entry.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/entry.S 2008-05-19 00:33:03.584352758 +0300
+@@ -269,7 +269,7 @@
+ CFI_STARTPROC simple
+ CFI_DEF_CFA esp, 0
+ CFI_REGISTER esp, ebp
+- movl TSS_sysenter_esp0(%esp),%esp
++ movl SYSENTER_stack_esp0(%esp),%esp
+ sysenter_past_esp:
+ /*
+ * No need to follow this irqs on/off section: the syscall
+@@ -689,7 +689,7 @@
+ * that sets up the real kernel stack. Check here, since we can't
+ * allow the wrong stack to be used.
+ *
+- * "TSS_sysenter_esp0+12" is because the NMI/debug handler will have
++ * "SYSENTER_stack_esp0+12" is because the NMI/debug handler will have
+ * already pushed 3 words if it hits on the sysenter instruction:
+ * eflags, cs and eip.
+ *
+@@ -701,7 +701,7 @@
+ cmpw $__KERNEL_CS,4(%esp); \
+ jne ok; \
+ label: \
+- movl TSS_sysenter_esp0+offset(%esp),%esp; \
++ movl SYSENTER_stack_esp0+offset(%esp),%esp; \
+ pushfl; \
+ pushl $__KERNEL_CS; \
+ pushl $sysenter_past_esp
+--- linux-2.6.18.8/arch/i386/kernel/fixup.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/fixup.c 2008-05-19 00:33:03.584352758 +0300
@@ -0,0 +1,88 @@
+/******************************************************************************
+ * fixup.c
+ if (current->tgid == 1)
+ return;
+
-+ HYPERVISOR_vm_assist(
-+ VMASST_CMD_disable, VMASST_TYPE_4gb_segments_notify);
++ VOID(HYPERVISOR_vm_assist(VMASST_CMD_disable,
++ VMASST_TYPE_4gb_segments_notify));
+
+ if (test_and_set_bit(0, &printed))
+ return;
+
+static int __init fixup_init(void)
+{
-+ HYPERVISOR_vm_assist(
-+ VMASST_CMD_enable, VMASST_TYPE_4gb_segments_notify);
++ WARN_ON(HYPERVISOR_vm_assist(VMASST_CMD_enable,
++ VMASST_TYPE_4gb_segments_notify));
+ return 0;
+}
+__initcall(fixup_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/head-xen.S linux-2.6.18-xen.hg/arch/i386/kernel/head-xen.S
---- linux-2.6.18/arch/i386/kernel/head-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/head-xen.S 2007-12-23 11:14:54.429188153 +0100
+--- linux-2.6.18.8/arch/i386/kernel/head-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/head-xen.S 2008-05-19 00:33:03.588352989 +0300
@@ -0,0 +1,207 @@
+
+
+#endif
+ ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic")
+ ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/init_task-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/init_task-xen.c
---- linux-2.6.18/arch/i386/kernel/init_task-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/init_task-xen.c 2007-12-23 11:14:54.432521660 +0100
+--- linux-2.6.18.8/arch/i386/kernel/init_task-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/init_task-xen.c 2008-05-19 00:33:03.596353450 +0300
@@ -0,0 +1,51 @@
+#include <linux/mm.h>
+#include <linux/module.h>
+DEFINE_PER_CPU(struct tss_struct, init_tss) ____cacheline_internodealigned_in_smp = INIT_TSS;
+#endif
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/io_apic-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/io_apic-xen.c
---- linux-2.6.18/arch/i386/kernel/io_apic-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/io_apic-xen.c 2007-12-23 11:14:54.435855171 +0100
+--- linux-2.6.18.8/arch/i386/kernel/io_apic-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/io_apic-xen.c 2008-05-19 00:33:03.600353680 +0300
@@ -0,0 +1,2770 @@
+/*
+ * Intel IO-APIC support for multi-Pentium hosts.
+ apic_op.apic_physbase = mp_ioapics[apic].mpc_apicaddr;
+ apic_op.reg = reg;
+ apic_op.value = value;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_apic_write, &apic_op);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_apic_write, &apic_op));
+}
+
+#define io_apic_read(a,r) xen_io_apic_read(a,r)
+ struct xen_platform_op op = { .cmd = XENPF_platform_quirk };
+ op.u.platform_quirk.quirk_id = sis_apic_bug ?
+ QUIRK_IOAPIC_BAD_REGSEL : QUIRK_IOAPIC_GOOD_REGSEL;
-+ HYPERVISOR_platform_op(&op);
++ VOID(HYPERVISOR_platform_op(&op));
+ }
+ return 0;
+}
+}
+
+#endif /* CONFIG_ACPI */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/ioport-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/ioport-xen.c
---- linux-2.6.18/arch/i386/kernel/ioport-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/ioport-xen.c 2007-12-23 11:14:54.435855171 +0100
-@@ -0,0 +1,122 @@
+--- linux-2.6.18.8/arch/i386/kernel/ioport-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/ioport-xen.c 2008-05-19 00:33:03.604353911 +0300
+@@ -0,0 +1,123 @@
+/*
+ * linux/arch/i386/kernel/ioport.c
+ *
+
+ set_xen_guest_handle(set_iobitmap.bitmap, (char *)bitmap);
+ set_iobitmap.nr_ports = IO_BITMAP_BITS;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &set_iobitmap);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap,
++ &set_iobitmap));
+ }
+
+ set_bitmap(t->io_bitmap_ptr, from, num, !turn_on);
+ set_iopl_mask(t->iopl);
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/irq-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/irq-xen.c
---- linux-2.6.18/arch/i386/kernel/irq-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/irq-xen.c 2007-12-23 11:14:54.435855171 +0100
+--- linux-2.6.18.8/arch/i386/kernel/irq-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/irq-xen.c 2008-05-19 00:33:03.604353911 +0300
@@ -0,0 +1,324 @@
+/*
+ * linux/arch/i386/kernel/irq.c
+}
+#endif
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/ldt-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/ldt-xen.c
---- linux-2.6.18/arch/i386/kernel/ldt-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/ldt-xen.c 2007-12-23 11:14:54.439188683 +0100
+--- linux-2.6.18.8/arch/i386/kernel/ldt-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/ldt-xen.c 2008-05-19 00:33:03.608354141 +0300
@@ -0,0 +1,270 @@
+/*
+ * linux/kernel/ldt.c
+ }
+ return ret;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/machine_kexec.c linux-2.6.18-xen.hg/arch/i386/kernel/machine_kexec.c
---- linux-2.6.18/arch/i386/kernel/machine_kexec.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/machine_kexec.c 2007-12-23 11:14:54.439188683 +0100
-@@ -19,123 +19,52 @@
+--- linux-2.6.18.8/arch/i386/kernel/machine_kexec.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/machine_kexec.c 2008-05-19 00:33:03.612354372 +0300
+@@ -19,123 +19,67 @@
#include <asm/desc.h>
#include <asm/system.h>
-#ifndef CONFIG_X86_PAE
-#define LEVEL1_SIZE (1UL << 22UL)
-static u32 pgtable_level1[1024] PAGE_ALIGNED;
--
++#ifdef CONFIG_XEN
++#include <xen/interface/kexec.h>
++#endif
+
-static void identity_map_page(unsigned long address)
-{
- unsigned long level1_index, level2_index;
- u32 *pgtable_level2;
--
++#define PAGE_ALIGNED __attribute__ ((__aligned__(PAGE_SIZE)))
++static u32 kexec_pgd[1024] PAGE_ALIGNED;
++#ifdef CONFIG_X86_PAE
++static u32 kexec_pmd0[1024] PAGE_ALIGNED;
++static u32 kexec_pmd1[1024] PAGE_ALIGNED;
++#endif
++static u32 kexec_pte0[1024] PAGE_ALIGNED;
++static u32 kexec_pte1[1024] PAGE_ALIGNED;
+
- /* Find the current page table */
- pgtable_level2 = __va(read_cr3());
--
++#ifdef CONFIG_XEN
+
- /* Find the indexes of the physical address to identity map */
- level1_index = (address % LEVEL1_SIZE)/LEVEL0_SIZE;
- level2_index = address / LEVEL1_SIZE;
- /* Identity map the page table entry */
- pgtable_level1[level1_index] = address | L0_ATTR;
- pgtable_level2[level2_index] = __pa(pgtable_level1) | L1_ATTR;
--
++#define __ma(x) (pfn_to_mfn(__pa((x)) >> PAGE_SHIFT) << PAGE_SHIFT)
+
- /* Flush the tlb so the new mapping takes effect.
- * Global tlb entries are not flushed but that is not an issue.
- */
- load_cr3(pgtable_level2);
-}
--
++#if PAGES_NR > KEXEC_XEN_NO_PAGES
++#error PAGES_NR is greater than KEXEC_XEN_NO_PAGES - Xen support will break
++#endif
+
-#else
-#define LEVEL1_SIZE (1UL << 21UL)
-#define LEVEL2_SIZE (1UL << 30UL)
-static u64 pgtable_level1[512] PAGE_ALIGNED;
-static u64 pgtable_level2[512] PAGE_ALIGNED;
--
++#if PA_CONTROL_PAGE != 0
++#error PA_CONTROL_PAGE is non zero - Xen support will break
++#endif
+
-static void identity_map_page(unsigned long address)
--{
++void machine_kexec_setup_load_arg(xen_kexec_image_t *xki, struct kimage *image)
+ {
- unsigned long level1_index, level2_index, level3_index;
- u64 *pgtable_level3;
--
++ void *control_page;
+
- /* Find the current page table */
- pgtable_level3 = __va(read_cr3());
--
++ memset(xki->page_list, 0, sizeof(xki->page_list));
+
- /* Find the indexes of the physical address to identity map */
- level1_index = (address % LEVEL1_SIZE)/LEVEL0_SIZE;
- level2_index = (address % LEVEL2_SIZE)/LEVEL1_SIZE;
- pgtable_level2[level2_index] = __pa(pgtable_level1) | L1_ATTR;
- set_64bit(&pgtable_level3[level3_index],
- __pa(pgtable_level2) | L2_ATTR);
-+#ifdef CONFIG_XEN
-+#include <xen/interface/kexec.h>
-+#endif
++ control_page = page_address(image->control_code_page);
++ memcpy(control_page, relocate_kernel, PAGE_SIZE);
- /* Flush the tlb so the new mapping takes effect.
- * Global tlb entries are not flushed but that is not an issue.
- */
- load_cr3(pgtable_level3);
-}
-+#define PAGE_ALIGNED __attribute__ ((__aligned__(PAGE_SIZE)))
-+static u32 kexec_pgd[1024] PAGE_ALIGNED;
++ xki->page_list[PA_CONTROL_PAGE] = __ma(control_page);
++ xki->page_list[PA_PGD] = __ma(kexec_pgd);
+#ifdef CONFIG_X86_PAE
-+static u32 kexec_pmd0[1024] PAGE_ALIGNED;
-+static u32 kexec_pmd1[1024] PAGE_ALIGNED;
++ xki->page_list[PA_PMD_0] = __ma(kexec_pmd0);
++ xki->page_list[PA_PMD_1] = __ma(kexec_pmd1);
#endif
-+static u32 kexec_pte0[1024] PAGE_ALIGNED;
-+static u32 kexec_pte1[1024] PAGE_ALIGNED;
++ xki->page_list[PA_PTE_0] = __ma(kexec_pte0);
++ xki->page_list[PA_PTE_1] = __ma(kexec_pte1);
-static void set_idt(void *newidt, __u16 limit)
-{
- struct Xgt_desc_struct curidt;
-+#ifdef CONFIG_XEN
-
+-
- /* ia32 supports unaliged loads & stores */
- curidt.size = limit;
- curidt.address = (unsigned long)newidt;
-+#define __ma(x) (pfn_to_mfn(__pa((x)) >> PAGE_SHIFT) << PAGE_SHIFT)
-
+-
- load_idt(&curidt);
-};
-+#if PAGES_NR > KEXEC_XEN_NO_PAGES
-+#error PAGES_NR is greater than KEXEC_XEN_NO_PAGES - Xen support will break
-+#endif
-
-+#if PA_CONTROL_PAGE != 0
-+#error PA_CONTROL_PAGE is non zero - Xen support will break
-+#endif
+-
++}
-static void set_gdt(void *newgdt, __u16 limit)
-+void machine_kexec_setup_load_arg(xen_kexec_image_t *xki, struct kimage *image)
++int __init machine_kexec_setup_resources(struct resource *hypervisor,
++ struct resource *phys_cpus,
++ int nr_phys_cpus)
{
- struct Xgt_desc_struct curgdt;
-+ void *control_page;
-
+-
- /* ia32 supports unaligned loads & stores */
- curgdt.size = limit;
- curgdt.address = (unsigned long)newgdt;
-+ memset(xki->page_list, 0, sizeof(xki->page_list));
++ int k;
- load_gdt(&curgdt);
-};
-+ control_page = page_address(image->control_code_page);
-+ memcpy(control_page, relocate_kernel, PAGE_SIZE);
++ /* The per-cpu crash note resources belong to the hypervisor resource */
++ for (k = 0; k < nr_phys_cpus; k++)
++ request_resource(hypervisor, phys_cpus + k);
-static void load_segments(void)
-{
-#define __STR(X) #X
-#define STR(X) __STR(X)
-+ xki->page_list[PA_CONTROL_PAGE] = __ma(control_page);
-+ xki->page_list[PA_PGD] = __ma(kexec_pgd);
-+#ifdef CONFIG_X86_PAE
-+ xki->page_list[PA_PMD_0] = __ma(kexec_pmd0);
-+ xki->page_list[PA_PMD_1] = __ma(kexec_pmd1);
-+#endif
-+ xki->page_list[PA_PTE_0] = __ma(kexec_pte0);
-+ xki->page_list[PA_PTE_1] = __ma(kexec_pte1);
-
+-
- __asm__ __volatile__ (
- "\tljmp $"STR(__KERNEL_CS)",$1f\n"
- "\t1:\n"
- ::: "eax", "memory");
-#undef STR
-#undef __STR
++ return 0;
}
-typedef asmlinkage NORET_TYPE void (*relocate_new_kernel_t)(
-extern const unsigned char relocate_new_kernel[];
-extern void relocate_new_kernel_end(void);
-extern const unsigned int relocate_new_kernel_size;
++void machine_kexec_register_resources(struct resource *res) { ; }
++
+#endif /* CONFIG_XEN */
/*
* A architecture hook called to validate the
-@@ -163,49 +92,38 @@
+@@ -163,49 +107,38 @@
{
}
- set_idt(phys_to_virt(0),0);
+ control_page = page_address(image->control_code_page);
+ memcpy(control_page, relocate_kernel, PAGE_SIZE);
-+
+
+- /* now call it */
+- rnk = (relocate_new_kernel_t) reboot_code_buffer;
+- (*rnk)(page_list, reboot_code_buffer, image->start, cpu_has_pae);
+ page_list[PA_CONTROL_PAGE] = __pa(control_page);
+ page_list[VA_CONTROL_PAGE] = (unsigned long)relocate_kernel;
+ page_list[PA_PGD] = __pa(kexec_pgd);
+ page_list[VA_PTE_0] = (unsigned long)kexec_pte0;
+ page_list[PA_PTE_1] = __pa(kexec_pte1);
+ page_list[VA_PTE_1] = (unsigned long)kexec_pte1;
-
-- /* now call it */
-- rnk = (relocate_new_kernel_t) reboot_code_buffer;
-- (*rnk)(page_list, reboot_code_buffer, image->start, cpu_has_pae);
++
+ relocate_kernel((unsigned long)image->head, (unsigned long)page_list,
+ image->start, cpu_has_pae);
}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/Makefile linux-2.6.18-xen.hg/arch/i386/kernel/Makefile
---- linux-2.6.18/arch/i386/kernel/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/Makefile 2007-12-23 11:14:54.382519030 +0100
-@@ -43,6 +43,7 @@
- EXTRA_AFLAGS := -traditional
-
- obj-$(CONFIG_SCx200) += scx200.o
-+obj-$(CONFIG_XEN) += fixup.o
-
- # vsyscall.o contains the vsyscall DSO images as __initdata.
- # We must build both images before we can assemble it.
-@@ -80,5 +81,8 @@
- $(obj)/vsyscall-sysenter.o $(obj)/vsyscall-note.o FORCE
- $(call if_changed,syscall)
-
-+early_printk-y += ../../x86_64/kernel/early_printk.o
- k8-y += ../../x86_64/kernel/k8.o
-
-+disabled-obj-$(CONFIG_XEN) := i8259.o reboot.o smpboot.o trampoline.o
-+%/head.o %/head.s: $(if $(CONFIG_XEN),EXTRA_AFLAGS,dummy) :=
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/microcode-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/microcode-xen.c
---- linux-2.6.18/arch/i386/kernel/microcode-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/microcode-xen.c 2007-12-23 11:14:54.439188683 +0100
+--- linux-2.6.18.8/arch/i386/kernel/microcode-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/microcode-xen.c 2008-05-19 00:33:03.612354372 +0300
@@ -0,0 +1,144 @@
+/*
+ * Intel CPU Microcode Update Driver for Linux
+module_init(microcode_init)
+module_exit(microcode_exit)
+MODULE_ALIAS_MISCDEV(MICROCODE_MINOR);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/mpparse-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/mpparse-xen.c
---- linux-2.6.18/arch/i386/kernel/mpparse-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/mpparse-xen.c 2007-12-23 11:14:54.442522194 +0100
+--- linux-2.6.18.8/arch/i386/kernel/mpparse-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/mpparse-xen.c 2008-05-19 00:33:03.616354603 +0300
@@ -0,0 +1,1185 @@
+/*
+ * Intel Multiprocessor Specification 1.1 and 1.4
+
+#endif /* CONFIG_X86_IO_APIC */
+#endif /* CONFIG_ACPI */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/pci-dma-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/pci-dma-xen.c
---- linux-2.6.18/arch/i386/kernel/pci-dma-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/pci-dma-xen.c 2007-12-23 11:14:54.445855703 +0100
-@@ -0,0 +1,382 @@
+--- linux-2.6.18.8/arch/i386/kernel/pci-dma-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/pci-dma-xen.c 2008-05-19 00:33:03.644356217 +0300
+@@ -0,0 +1,415 @@
+/*
+ * Dynamic DMA mapping support.
+ *
+ } \
+} while (0)
+
++static int check_pages_physically_contiguous(unsigned long pfn,
++ unsigned int offset,
++ size_t length)
++{
++ unsigned long next_mfn;
++ int i;
++ int nr_pages;
++
++ next_mfn = pfn_to_mfn(pfn);
++ nr_pages = (offset + length + PAGE_SIZE-1) >> PAGE_SHIFT;
++
++ for (i = 1; i < nr_pages; i++) {
++ if (pfn_to_mfn(++pfn) != ++next_mfn)
++ return 0;
++ }
++ return 1;
++}
++
++int range_straddles_page_boundary(paddr_t p, size_t size)
++{
++ extern unsigned long *contiguous_bitmap;
++ unsigned long pfn = p >> PAGE_SHIFT;
++ unsigned int offset = p & ~PAGE_MASK;
++
++ if (offset + size <= PAGE_SIZE)
++ return 0;
++ if (test_bit(pfn, contiguous_bitmap))
++ return 0;
++ if (check_pages_physically_contiguous(pfn, offset, size))
++ return 0;
++ return 1;
++}
++
+int
+dma_map_sg(struct device *hwdev, struct scatterlist *sg, int nents,
+ enum dma_data_direction direction)
+ swiotlb_sync_single_for_device(dev, dma_handle, size, direction);
+}
+EXPORT_SYMBOL(dma_sync_single_for_device);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/process-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/process-xen.c
---- linux-2.6.18/arch/i386/kernel/process-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/process-xen.c 2007-12-23 11:14:54.445855703 +0100
-@@ -0,0 +1,853 @@
+--- linux-2.6.18.8/arch/i386/kernel/process-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/process-xen.c 2008-05-19 00:33:03.700359445 +0300
+@@ -0,0 +1,856 @@
+/*
+ * linux/arch/i386/kernel/process.c
+ *
+ local_irq_disable();
+ cpu_clear(smp_processor_id(), cpu_initialized);
+ preempt_enable_no_resched();
-+ HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
++ VOID(HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL));
+ cpu_bringup();
+}
+#else
+ struct thread_struct *t = &tsk->thread;
+ struct physdev_set_iobitmap set_iobitmap;
+ memset(&set_iobitmap, 0, sizeof(set_iobitmap));
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &set_iobitmap);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap,
++ &set_iobitmap));
+ kfree(t->io_bitmap_ptr);
+ t->io_bitmap_ptr = NULL;
+ clear_thread_flag(TIF_IO_BITMAP);
+ mcl++;
+ }
+
-+ (void)HYPERVISOR_multicall(_mcl, mcl - _mcl);
++ BUG_ON(mcl > _mcl + ARRAY_SIZE(_mcl));
++ if (unlikely(HYPERVISOR_multicall_check(_mcl, mcl - _mcl, NULL)))
++ BUG();
+
+ /*
+ * Restore %fs and %gs if needed.
+ sp -= get_random_int() % 8192;
+ return sp & ~0xf;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/quirks-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/quirks-xen.c
---- linux-2.6.18/arch/i386/kernel/quirks-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/quirks-xen.c 2007-12-23 11:14:54.449189209 +0100
+--- linux-2.6.18.8/arch/i386/kernel/quirks-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/quirks-xen.c 2008-05-19 00:33:03.704359675 +0300
@@ -0,0 +1,47 @@
+/*
+ * This file contains work-arounds for x86 and x86_64 platform bugs.
+ printk(KERN_INFO "Disabling irq balancing and affinity\n");
+ op.cmd = XENPF_platform_quirk;
+ op.u.platform_quirk.quirk_id = QUIRK_NOIRQBALANCING;
-+ (void)HYPERVISOR_platform_op(&op);
++ WARN_ON(HYPERVISOR_platform_op(&op));
+ }
+
+ /* put back the original value for config space*/
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7525_MCH, quirk_intel_irqbalance);
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7520_MCH, quirk_intel_irqbalance);
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/relocate_kernel.S linux-2.6.18-xen.hg/arch/i386/kernel/relocate_kernel.S
---- linux-2.6.18/arch/i386/kernel/relocate_kernel.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/relocate_kernel.S 2007-12-23 11:14:54.452522713 +0100
+--- linux-2.6.18.8/arch/i386/kernel/relocate_kernel.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/relocate_kernel.S 2008-05-19 00:33:03.708359906 +0300
@@ -7,16 +7,138 @@
*/
+idt_48:
+ .word 0 /* limit */
+ .long 0 /* base */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/setup.c linux-2.6.18-xen.hg/arch/i386/kernel/setup.c
---- linux-2.6.18/arch/i386/kernel/setup.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/setup.c 2007-12-23 11:14:54.455856217 +0100
-@@ -956,6 +956,28 @@
- return 0;
- }
-
-+/*
-+ * This function checks if any part of the range <start,end> is mapped
-+ * with type.
-+ */
-+int
-+e820_any_mapped(u64 start, u64 end, unsigned type)
-+{
-+ int i;
-+
-+ for (i = 0; i < e820.nr_map; i++) {
-+ const struct e820entry *ei = &e820.map[i];
-+
-+ if (type && ei->type != type)
-+ continue;
-+ if (ei->addr >= end || ei->addr + ei->size <= start)
-+ continue;
-+ return 1;
-+ }
-+ return 0;
-+}
-+EXPORT_SYMBOL_GPL(e820_any_mapped);
-+
- /*
- * This function checks if the entire range <start,end> is mapped with type.
- *
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/setup-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/setup-xen.c
---- linux-2.6.18/arch/i386/kernel/setup-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/setup-xen.c 2007-12-23 11:14:54.452522713 +0100
-@@ -0,0 +1,1918 @@
+--- linux-2.6.18.8/arch/i386/kernel/setup-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/setup-xen.c 2008-05-19 00:33:03.792364748 +0300
+@@ -0,0 +1,1919 @@
+/*
+ * linux/arch/i386/kernel/setup.c
+ *
+#endif
+ add_memory_region(start, size, type);
+ } while (biosmap++,--nr_map);
++
++#ifdef CONFIG_XEN
++ if (is_initial_xendomain()) {
++ struct xen_memory_map memmap;
++
++ memmap.nr_entries = E820MAX;
++ set_xen_guest_handle(memmap.buffer, machine_e820.map);
++
++ if (HYPERVISOR_memory_op(XENMEM_machine_memory_map, &memmap))
++ BUG();
++ machine_e820.nr_map = memmap.nr_entries;
++ } else
++ machine_e820 = e820;
++#endif
++
+ return 0;
+}
+
+ crashk_res.end - crashk_res.start + 1);
+#endif
+#endif
-+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap))
-+ phys_to_machine_mapping =
-+ (unsigned long *)xen_start_info->mfn_list;
+}
+
+/*
+static void __init register_memory(void)
+{
+#ifdef CONFIG_XEN
-+ if (is_initial_xendomain()) {
-+ struct xen_memory_map memmap;
-+
-+ memmap.nr_entries = E820MAX;
-+ set_xen_guest_handle(memmap.buffer, machine_e820.map);
-+
-+ if (HYPERVISOR_memory_op(XENMEM_machine_memory_map, &memmap))
-+ BUG();
-+
-+ machine_e820.nr_map = memmap.nr_entries;
++ if (is_initial_xendomain())
+ e820_setup_gap(machine_e820.map, machine_e820.nr_map);
-+ }
+ else
+#endif
+ e820_setup_gap(e820.map, e820.nr_map);
+ /* Register a call for panic conditions. */
+ atomic_notifier_chain_register(&panic_notifier_list, &xen_panic_block);
+
-+ HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_4gb_segments);
-+ HYPERVISOR_vm_assist(VMASST_CMD_enable,
-+ VMASST_TYPE_writable_pagetables);
++ WARN_ON(HYPERVISOR_vm_assist(VMASST_CMD_enable,
++ VMASST_TYPE_4gb_segments));
++ WARN_ON(HYPERVISOR_vm_assist(VMASST_CMD_enable,
++ VMASST_TYPE_writable_pagetables));
+
+ memcpy(&boot_cpu_data, &new_cpu_data, sizeof(new_cpu_data));
++ pre_setup_arch_hook();
+ early_cpu_init();
+#ifdef CONFIG_SMP
+ prefill_possible_map();
+ rd_doload = ((RAMDISK_FLAGS & RAMDISK_LOAD_FLAG) != 0);
+#endif
+
-+ setup_xen_features();
-+
+ ARCH_SETUP
+ if (efi_enabled)
+ efi_init();
+ efi_map_memmap();
+
+ set_iopl.iopl = 1;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl));
+
+#ifdef CONFIG_ACPI
+ if (!is_initial_xendomain()) {
+ * c-basic-offset:8
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/smp-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/smp-xen.c
---- linux-2.6.18/arch/i386/kernel/smp-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/smp-xen.c 2007-12-23 11:14:54.455856217 +0100
+--- linux-2.6.18.8/arch/i386/kernel/setup.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/setup.c 2008-05-19 00:33:03.796364978 +0300
+@@ -956,6 +956,28 @@
+ return 0;
+ }
+
++/*
++ * This function checks if any part of the range <start,end> is mapped
++ * with type.
++ */
++int
++e820_any_mapped(u64 start, u64 end, unsigned type)
++{
++ int i;
++
++ for (i = 0; i < e820.nr_map; i++) {
++ const struct e820entry *ei = &e820.map[i];
++
++ if (type && ei->type != type)
++ continue;
++ if (ei->addr >= end || ei->addr + ei->size <= start)
++ continue;
++ return 1;
++ }
++ return 0;
++}
++EXPORT_SYMBOL_GPL(e820_any_mapped);
++
+ /*
+ * This function checks if the entire range <start,end> is mapped with type.
+ *
+--- linux-2.6.18.8/arch/i386/kernel/smp-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/smp-xen.c 2008-05-19 00:33:03.800365209 +0300
@@ -0,0 +1,605 @@
+/*
+ * Intel SMP support routines.
+ return IRQ_HANDLED;
+}
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/sysenter.c linux-2.6.18-xen.hg/arch/i386/kernel/sysenter.c
---- linux-2.6.18/arch/i386/kernel/sysenter.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/sysenter.c 2007-12-23 11:14:54.462523240 +0100
+--- linux-2.6.18.8/arch/i386/kernel/sysenter.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/sysenter.c 2008-05-19 00:33:03.808365670 +0300
@@ -23,6 +23,10 @@
#include <asm/pgtable.h>
#include <asm/unistd.h>
put_cpu();
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/time-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/time-xen.c
---- linux-2.6.18/arch/i386/kernel/time-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/time-xen.c 2007-12-23 11:14:54.462523240 +0100
-@@ -0,0 +1,1170 @@
+@@ -75,11 +109,6 @@
+ #ifdef CONFIG_COMPAT_VDSO
+ __set_fixmap(FIX_VDSO, __pa(syscall_page), PAGE_READONLY);
+ printk("Compat vDSO mapped to %08lx.\n", __fix_to_virt(FIX_VDSO));
+-#else
+- /*
+- * In the non-compat case the ELF coredumping code needs the fixmap:
+- */
+- __set_fixmap(FIX_VDSO, __pa(syscall_page), PAGE_KERNEL_RO);
+ #endif
+
+ if (!boot_cpu_has(X86_FEATURE_SEP)) {
+@@ -142,6 +171,13 @@
+ vma->vm_end = addr + PAGE_SIZE;
+ /* MAYWRITE to allow gdb to COW and set breakpoints */
+ vma->vm_flags = VM_READ|VM_EXEC|VM_MAYREAD|VM_MAYEXEC|VM_MAYWRITE;
++ /*
++ * Make sure the vDSO gets into every core dump.
++ * Dumping its contents makes post-mortem fully interpretable later
++ * without matching up the same kernel and hardware config to see
++ * what PC values meant.
++ */
++ vma->vm_flags |= VM_ALWAYSDUMP;
+ vma->vm_flags |= mm->def_flags;
+ vma->vm_page_prot = protection_map[vma->vm_flags & 7];
+ vma->vm_ops = &syscall_vm_ops;
+--- linux-2.6.18.8/arch/i386/kernel/time-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/time-xen.c 2008-05-19 00:33:03.808365670 +0300
+@@ -0,0 +1,1194 @@
+/*
+ * linux/arch/i386/kernel/time.c
+ *
+ * Reads a consistent set of time-base values from Xen, into a shadow data
+ * area.
+ */
-+static void get_time_values_from_xen(int cpu)
++static void get_time_values_from_xen(unsigned int cpu)
+{
+ struct vcpu_time_info *src;
+ struct shadow_time_info *dst;
+ dst->tsc_to_usec_mul = dst->tsc_to_nsec_mul / 1000;
+}
+
-+static inline int time_values_up_to_date(int cpu)
++static inline int time_values_up_to_date(unsigned int cpu)
+{
+ struct vcpu_time_info *src;
+ struct shadow_time_info *dst;
+ op.u.settime.secs = sec;
+ op.u.settime.nsecs = nsec;
+ op.u.settime.system_time = shadow->system_timestamp;
-+ HYPERVISOR_platform_op(&op);
++ WARN_ON(HYPERVISOR_platform_op(&op));
+ update_wallclock();
+ } else if (independent_wallclock) {
+ nsec -= shadow->system_timestamp;
+ op.u.settime.secs = sec;
+ op.u.settime.nsecs = nsec;
+ op.u.settime.system_time = processed_system_time;
-+ HYPERVISOR_platform_op(&op);
++ WARN_ON(HYPERVISOR_platform_op(&op));
+
+ update_wallclock();
+
+ */
+unsigned long long monotonic_clock(void)
+{
-+ int cpu = get_cpu();
++ unsigned int cpu = get_cpu();
+ struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu);
+ u64 time;
+ u32 local_time_version;
+{
+ s64 delta, delta_cpu, stolen, blocked;
+ u64 sched_time;
-+ int i, cpu = smp_processor_id();
++ unsigned int i, cpu = smp_processor_id();
+ struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu);
+ struct vcpu_runstate_info *runstate = &per_cpu(runstate, cpu);
+
+ if ((unlikely(delta < -(s64)permitted_clock_jitter) ||
+ unlikely(delta_cpu < -(s64)permitted_clock_jitter))
+ && printk_ratelimit()) {
-+ printk("Timer ISR/%d: Time went backwards: "
++ printk("Timer ISR/%u: Time went backwards: "
+ "delta=%lld delta_cpu=%lld shadow=%lld "
+ "off=%lld processed=%lld cpu_processed=%lld\n",
+ cpu, delta, delta_cpu, shadow->system_timestamp,
+ return IRQ_HANDLED;
+}
+
-+static void init_missing_ticks_accounting(int cpu)
++static void init_missing_ticks_accounting(unsigned int cpu)
+{
+ struct vcpu_register_runstate_memory_area area;
+ struct vcpu_runstate_info *runstate = &per_cpu(runstate, cpu);
+ }
+#endif
+
-+ HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, 0,
-+ &xen_set_periodic_tick);
++ switch (HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, 0,
++ &xen_set_periodic_tick)) {
++ case 0:
++#if CONFIG_XEN_COMPAT <= 0x030004
++ case -ENOSYS:
++#endif
++ break;
++ default:
++ BUG();
++ }
+
+ get_time_values_from_xen(0);
+
+void halt(void)
+{
+ if (irqs_disabled())
-+ HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
++ VOID(HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL));
+}
+EXPORT_SYMBOL(halt);
+
+ init_cpu_khz();
+
+ for_each_online_cpu(cpu) {
-+ HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, cpu,
-+ &xen_set_periodic_tick);
++ switch (HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, cpu,
++ &xen_set_periodic_tick)) {
++ case 0:
++#if CONFIG_XEN_COMPAT <= 0x030004
++ case -ENOSYS:
++#endif
++ break;
++ default:
++ BUG();
++ }
+ get_time_values_from_xen(cpu);
+ per_cpu(processed_system_time, cpu) =
+ per_cpu(shadow_time, 0).system_timestamp;
+
+ BUG_ON(cpu == 0);
+
-+ HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, cpu,
-+ &xen_set_periodic_tick);
++ switch (HYPERVISOR_vcpu_op(VCPUOP_set_periodic_timer, cpu,
++ &xen_set_periodic_tick)) {
++ case 0:
++#if CONFIG_XEN_COMPAT <= 0x030004
++ case -ENOSYS:
++#endif
++ break;
++ default:
++ BUG();
++ }
+
+ do {
+ seq = read_seqbegin(&xtime_lock);
+ init_missing_ticks_accounting(cpu);
+ } while (read_seqretry(&xtime_lock, seq));
+
-+ sprintf(timer_name[cpu], "timer%d", cpu);
++ sprintf(timer_name[cpu], "timer%u", cpu);
+ irq = bind_virq_to_irqhandler(VIRQ_TIMER,
+ cpu,
+ timer_interrupt,
+ op.u.change_freq.flags = 0;
+ op.u.change_freq.cpu = freq->cpu;
+ op.u.change_freq.freq = (u64)freq->new * 1000;
-+ HYPERVISOR_platform_op(&op);
++ WARN_ON(HYPERVISOR_platform_op(&op));
+
+ return 0;
+}
+ return 0;
+}
+__initcall(xen_sysctl_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/traps.c linux-2.6.18-xen.hg/arch/i386/kernel/traps.c
---- linux-2.6.18/arch/i386/kernel/traps.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/traps.c 2007-12-23 11:14:54.465856754 +0100
-@@ -642,18 +642,11 @@
-
- static void io_check_error(unsigned char reason, struct pt_regs * regs)
- {
-- unsigned long i;
--
- printk(KERN_EMERG "NMI: IOCK error (debug interrupt?)\n");
- show_registers(regs);
-
- /* Re-enable the IOCK line, wait for a few seconds */
-- reason = (reason & 0xf) | 8;
-- outb(reason, 0x61);
-- i = 2000;
-- while (--i) udelay(1000);
-- reason &= ~8;
-- outb(reason, 0x61);
-+ clear_io_check_error(reason);
- }
-
- static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/traps-xen.c linux-2.6.18-xen.hg/arch/i386/kernel/traps-xen.c
---- linux-2.6.18/arch/i386/kernel/traps-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/traps-xen.c 2007-12-23 11:14:54.465856754 +0100
-@@ -0,0 +1,1186 @@
+--- linux-2.6.18.8/arch/i386/kernel/traps-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/traps-xen.c 2008-05-19 00:33:03.812365901 +0300
+@@ -0,0 +1,1190 @@
+/*
+ * linux/arch/i386/traps.c
+ *
+ * NB. All these are "trap gates" (i.e. events_mask isn't set) except
+ * for those that specify <dpl>|4 in the second field.
+ */
-+static trap_info_t trap_table[] = {
++static trap_info_t __cpuinitdata trap_table[] = {
+ { 0, 0, __KERNEL_CS, (unsigned long)divide_error },
+ { 1, 0|4, __KERNEL_CS, (unsigned long)debug },
+ { 3, 3|4, __KERNEL_CS, (unsigned long)int3 },
+
+void __init trap_init(void)
+{
-+ HYPERVISOR_set_trap_table(trap_table);
++ int ret;
++
++ ret = HYPERVISOR_set_trap_table(trap_table);
++ if (ret)
++ printk("HYPERVISOR_set_trap_table failed: error %d\n", ret);
+
+ if (cpu_has_fxsr) {
+ /*
+ cpu_init();
+}
+
-+void smp_trap_init(trap_info_t *trap_ctxt)
++void __cpuinit smp_trap_init(trap_info_t *trap_ctxt)
+{
-+ trap_info_t *t = trap_table;
++ const trap_info_t *t = trap_table;
+
+ for (t = trap_table; t->address; t++) {
+ trap_ctxt[t->vector].flags = t->flags;
+}
+__setup("call_trace=", call_trace_setup);
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/vm86.c linux-2.6.18-xen.hg/arch/i386/kernel/vm86.c
---- linux-2.6.18/arch/i386/kernel/vm86.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/vm86.c 2007-12-23 11:14:54.465856754 +0100
+--- linux-2.6.18.8/arch/i386/kernel/traps.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/traps.c 2008-05-19 00:33:03.816366131 +0300
+@@ -642,18 +642,11 @@
+
+ static void io_check_error(unsigned char reason, struct pt_regs * regs)
+ {
+- unsigned long i;
+-
+ printk(KERN_EMERG "NMI: IOCK error (debug interrupt?)\n");
+ show_registers(regs);
+
+ /* Re-enable the IOCK line, wait for a few seconds */
+- reason = (reason & 0xf) | 8;
+- outb(reason, 0x61);
+- i = 2000;
+- while (--i) udelay(1000);
+- reason &= ~8;
+- outb(reason, 0x61);
++ clear_io_check_error(reason);
+ }
+
+ static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs)
+--- linux-2.6.18.8/arch/i386/kernel/vm86.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/vm86.c 2008-05-19 00:33:03.816366131 +0300
@@ -97,7 +97,9 @@
struct pt_regs * FASTCALL(save_v86_state(struct kernel_vm86_regs * regs));
struct pt_regs * fastcall save_v86_state(struct kernel_vm86_regs * regs)
tsk->thread.screen_bitmap = info->screen_bitmap;
if (info->flags & VM86_SCREEN_BITMAP)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/vmlinux.lds.S linux-2.6.18-xen.hg/arch/i386/kernel/vmlinux.lds.S
---- linux-2.6.18/arch/i386/kernel/vmlinux.lds.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/vmlinux.lds.S 2007-12-23 11:14:54.469190263 +0100
+--- linux-2.6.18.8/arch/i386/kernel/vmlinux.lds.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/vmlinux.lds.S 2008-05-19 00:33:03.820366362 +0300
@@ -13,6 +13,12 @@
OUTPUT_ARCH(i386)
ENTRY(phys_startup_32)
+
+ NOTES
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/kernel/vsyscall-note-xen.S linux-2.6.18-xen.hg/arch/i386/kernel/vsyscall-note-xen.S
---- linux-2.6.18/arch/i386/kernel/vsyscall-note-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/kernel/vsyscall-note-xen.S 2007-12-23 11:14:54.469190263 +0100
+--- linux-2.6.18.8/arch/i386/kernel/vsyscall-note-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/kernel/vsyscall-note-xen.S 2008-05-19 00:33:03.820366362 +0300
@@ -0,0 +1,32 @@
+/*
+ * This supplies .note.* sections to go into the PT_NOTE inside the vDSO text.
+NOTE_KERNELCAP_BEGIN(1, 1)
+NOTE_KERNELCAP(0, "nosegneg")
+NOTE_KERNELCAP_END
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mach-xen/Makefile linux-2.6.18-xen.hg/arch/i386/mach-xen/Makefile
---- linux-2.6.18/arch/i386/mach-xen/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mach-xen/Makefile 2007-12-23 11:14:54.489191309 +0100
+--- linux-2.6.18.8/arch/i386/lib/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/lib/Makefile 2008-05-19 00:33:03.824366592 +0300
+@@ -7,3 +7,4 @@
+ bitops.o
+
+ lib-$(CONFIG_X86_USE_3DNOW) += mmx.o
++lib-$(CONFIG_XEN_SCRUB_PAGES) += scrub.o
+--- linux-2.6.18.8/arch/i386/lib/scrub.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/lib/scrub.c 2008-05-19 00:33:03.848367976 +0300
+@@ -0,0 +1,21 @@
++#include <asm/cpufeature.h>
++#include <asm/page.h>
++#include <asm/processor.h>
++
++void scrub_pages(void *v, unsigned int count)
++{
++ if (likely(cpu_has_xmm2)) {
++ unsigned long n = count * (PAGE_SIZE / sizeof(long) / 4);
++
++ for (; n--; v += sizeof(long) * 4)
++ asm("movnti %1,(%0)\n\t"
++ "movnti %1,%c2(%0)\n\t"
++ "movnti %1,2*%c2(%0)\n\t"
++ "movnti %1,3*%c2(%0)\n\t"
++ : : "r" (v), "r" (0L), "i" (sizeof(long))
++ : "memory");
++ asm volatile("sfence" : : : "memory");
++ } else
++ for (; count--; v += PAGE_SIZE)
++ clear_page(v);
++}
+--- linux-2.6.18.8/arch/i386/mach-xen/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mach-xen/Makefile 2008-05-19 00:33:03.900370973 +0300
@@ -0,0 +1,5 @@
+#
+# Makefile for the linux kernel.
+#
+
+obj-y := setup.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mach-xen/setup.c linux-2.6.18-xen.hg/arch/i386/mach-xen/setup.c
---- linux-2.6.18/arch/i386/mach-xen/setup.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mach-xen/setup.c 2007-12-23 11:14:54.489191309 +0100
-@@ -0,0 +1,147 @@
+--- linux-2.6.18.8/arch/i386/mach-xen/setup.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mach-xen/setup.c 2008-05-19 00:33:03.900370973 +0300
+@@ -0,0 +1,158 @@
+/*
+ * Machine specific setup for generic
+ */
+
++#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+extern void failsafe_callback(void);
+extern void nmi(void);
+
-+unsigned long *machine_to_phys_mapping;
++unsigned long *machine_to_phys_mapping = (void *)MACH2PHYS_VIRT_START;
+EXPORT_SYMBOL(machine_to_phys_mapping);
+unsigned int machine_to_phys_order;
+EXPORT_SYMBOL(machine_to_phys_order);
+
-+void __init machine_specific_arch_setup(void)
++void __init pre_setup_arch_hook(void)
+{
-+ int ret;
+ struct xen_machphys_mapping mapping;
+ unsigned long machine_to_phys_nr_ents;
+ struct xen_platform_parameters pp;
++
++ init_mm.pgd = swapper_pg_dir = (pgd_t *)xen_start_info->pt_base;
++
++ setup_xen_features();
++
++ if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
++ set_fixaddr_top(pp.virt_start);
++
++ if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
++ machine_to_phys_mapping = (unsigned long *)mapping.v_start;
++ machine_to_phys_nr_ents = mapping.max_mfn + 1;
++ } else
++ machine_to_phys_nr_ents = MACH2PHYS_NR_ENTRIES;
++ machine_to_phys_order = fls(machine_to_phys_nr_ents - 1);
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap))
++ phys_to_machine_mapping =
++ (unsigned long *)xen_start_info->mfn_list;
++}
++
++void __init machine_specific_arch_setup(void)
++{
++ int ret;
+ static struct callback_register __initdata event = {
+ .type = CALLBACKTYPE_event,
+ .address = { __KERNEL_CS, (unsigned long)hypervisor_callback },
+ HYPERVISOR_nmi_op(XENNMI_register_callback, &cb);
+ }
+#endif
-+
-+ if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
-+ set_fixaddr_top(pp.virt_start);
-+
-+ machine_to_phys_mapping = (unsigned long *)MACH2PHYS_VIRT_START;
-+ machine_to_phys_nr_ents = MACH2PHYS_NR_ENTRIES;
-+ if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
-+ machine_to_phys_mapping = (unsigned long *)mapping.v_start;
-+ machine_to_phys_nr_ents = mapping.max_mfn + 1;
-+ }
-+ while ((1UL << machine_to_phys_order) < machine_to_phys_nr_ents )
-+ machine_to_phys_order++;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/Makefile linux-2.6.18-xen.hg/arch/i386/Makefile
---- linux-2.6.18/arch/i386/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/Makefile 2007-12-23 11:14:54.369185012 +0100
-@@ -71,6 +71,10 @@
- mflags-$(CONFIG_X86_SUMMIT) := -Iinclude/asm-i386/mach-summit
- mcore-$(CONFIG_X86_SUMMIT) := mach-default
-
-+# Xen subarch support
-+mflags-$(CONFIG_X86_XEN) := -Iinclude/asm-i386/mach-xen
-+mcore-$(CONFIG_X86_XEN) := mach-xen
-+
- # generic subarchitecture
- mflags-$(CONFIG_X86_GENERICARCH) := -Iinclude/asm-i386/mach-generic
- mcore-$(CONFIG_X86_GENERICARCH) := mach-default
-@@ -102,9 +106,20 @@
-
- boot := arch/i386/boot
-
--PHONY += zImage bzImage compressed zlilo bzlilo \
-+PHONY += zImage bzImage vmlinuz compressed zlilo bzlilo \
- zdisk bzdisk fdimage fdimage144 fdimage288 isoimage install
-
-+ifdef CONFIG_XEN
-+CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
-+ -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(CPPFLAGS)
-+all: vmlinuz
-+
-+# KBUILD_IMAGE specifies the target image being built
-+KBUILD_IMAGE := $(boot)/vmlinuz
-+
-+vmlinuz: vmlinux
-+ $(Q)$(MAKE) $(build)=$(boot) $(KBUILD_IMAGE)
-+else
- all: bzImage
-
- # KBUILD_IMAGE specify target image being built
-@@ -124,6 +139,7 @@
-
- fdimage fdimage144 fdimage288 isoimage: vmlinux
- $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(KBUILD_IMAGE) $@
-+endif
-
- install:
- $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(KBUILD_IMAGE) install
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/fault-xen.c linux-2.6.18-xen.hg/arch/i386/mm/fault-xen.c
---- linux-2.6.18/arch/i386/mm/fault-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/fault-xen.c 2007-12-23 11:14:54.519192886 +0100
+--- linux-2.6.18.8/arch/i386/mm/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/Makefile 2008-05-19 00:33:03.956374201 +0300
+@@ -8,3 +8,4 @@
+ obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
+ obj-$(CONFIG_HIGHMEM) += highmem.o
+ obj-$(CONFIG_BOOT_IOREMAP) += boot_ioremap.o
++obj-$(CONFIG_XEN) += hypervisor.o
+--- linux-2.6.18.8/arch/i386/mm/fault-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/fault-xen.c 2008-05-19 00:33:03.960374432 +0300
@@ -0,0 +1,779 @@
+/*
+ * linux/arch/i386/mm/fault.c
+ }
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/highmem-xen.c linux-2.6.18-xen.hg/arch/i386/mm/highmem-xen.c
---- linux-2.6.18/arch/i386/mm/highmem-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/highmem-xen.c 2007-12-23 11:14:54.519192886 +0100
+--- linux-2.6.18.8/arch/i386/mm/highmem-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/highmem-xen.c 2008-05-19 00:33:03.964374662 +0300
@@ -0,0 +1,136 @@
+#include <linux/highmem.h>
+#include <linux/module.h>
+EXPORT_SYMBOL(kmap_atomic_pte);
+EXPORT_SYMBOL(kunmap_atomic);
+EXPORT_SYMBOL(kmap_atomic_to_page);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/hypervisor.c linux-2.6.18-xen.hg/arch/i386/mm/hypervisor.c
---- linux-2.6.18/arch/i386/mm/hypervisor.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/hypervisor.c 2007-12-23 11:14:54.522526395 +0100
-@@ -0,0 +1,550 @@
+--- linux-2.6.18.8/arch/i386/mm/hypervisor.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/hypervisor.c 2008-05-19 00:33:03.964374662 +0300
+@@ -0,0 +1,602 @@
+/******************************************************************************
+ * mm/hypervisor.c
+ *
+ BUG_ON(HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0);
+}
+
-+void xen_set_ldt(unsigned long ptr, unsigned long len)
++void xen_set_ldt(const void *ptr, unsigned int ents)
+{
+ struct mmuext_op op;
+ op.cmd = MMUEXT_SET_LDT;
-+ op.arg1.linear_addr = ptr;
-+ op.arg2.nr_ents = len;
++ op.arg1.linear_addr = (unsigned long)ptr;
++ op.arg2.nr_ents = ents;
+ BUG_ON(HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0);
+}
+
+ unsigned long vstart, unsigned int order, unsigned int address_bits)
+{
+ unsigned long *in_frames = discontig_frames, out_frame;
-+ unsigned long frame, i, flags;
-+ long rc;
-+ int success;
++ unsigned long frame, flags;
++ unsigned int i;
++ int rc, success;
+ struct xen_memory_exchange exchange = {
+ .in = {
+ .nr_extents = 1UL << order,
+ set_xen_guest_handle(exchange.in.extent_start, in_frames);
+ set_xen_guest_handle(exchange.out.extent_start, &out_frame);
+
-+ scrub_pages(vstart, 1 << order);
++ scrub_pages((void *)vstart, 1 << order);
+
+ balloon_lock(flags);
+
+ /* 1. Zap current PTEs, remembering MFNs. */
-+ for (i = 0; i < (1UL<<order); i++) {
++ for (i = 0; i < (1U<<order); i++) {
+ in_frames[i] = pfn_to_mfn((__pa(vstart) >> PAGE_SHIFT) + i);
+ MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
+ __pte_ma(0), 0);
+ &exchange.out) == 1);
+ if (!success) {
+ /* Couldn't get special memory: fall back to normal. */
-+ for (i = 0; i < (1UL<<order); i++)
++ for (i = 0; i < (1U<<order); i++)
+ in_frames[i] = (__pa(vstart)>>PAGE_SHIFT) + i;
+ if (HYPERVISOR_memory_op(XENMEM_populate_physmap,
+ &exchange.in) != (1UL<<order))
+#endif
+
+ /* 3. Map the new extent in place of old pages. */
-+ for (i = 0; i < (1UL<<order); i++) {
++ for (i = 0; i < (1U<<order); i++) {
+ frame = success ? (out_frame + i) : in_frames[i];
+ MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
+ pfn_pte_ma(frame, PAGE_KERNEL), 0);
+void xen_destroy_contiguous_region(unsigned long vstart, unsigned int order)
+{
+ unsigned long *out_frames = discontig_frames, in_frame;
-+ unsigned long frame, i, flags;
-+ long rc;
-+ int success;
++ unsigned long frame, flags;
++ unsigned int i;
++ int rc, success;
+ struct xen_memory_exchange exchange = {
+ .in = {
+ .nr_extents = 1,
+ set_xen_guest_handle(exchange.in.extent_start, &in_frame);
+ set_xen_guest_handle(exchange.out.extent_start, out_frames);
+
-+ scrub_pages(vstart, 1 << order);
++ scrub_pages((void *)vstart, 1 << order);
+
+ balloon_lock(flags);
+
+ in_frame = pfn_to_mfn(__pa(vstart) >> PAGE_SHIFT);
+
+ /* 2. Zap current PTEs. */
-+ for (i = 0; i < (1UL<<order); i++) {
++ for (i = 0; i < (1U<<order); i++) {
+ MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
+ __pte_ma(0), 0);
+ set_phys_to_machine((__pa(vstart)>>PAGE_SHIFT)+i,
+#endif
+
+ /* 4. Map new pages in place of old pages. */
-+ for (i = 0; i < (1UL<<order); i++) {
++ for (i = 0; i < (1U<<order); i++) {
+ frame = success ? out_frames[i] : (in_frame + i);
+ MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
+ pfn_pte_ma(frame, PAGE_KERNEL), 0);
+{
+ unsigned long flags, frame;
+ unsigned long *in_frames = discontig_frames, *out_frames = limited_frames;
-+ void *v;
+ struct page *page;
-+ int i, nr_mcl, rc, success;
++ unsigned int i, n, nr_mcl;
++ int rc, success;
++ DECLARE_BITMAP(limit_map, 1 << MAX_CONTIG_ORDER);
+
+ struct xen_memory_exchange exchange = {
+ .in = {
-+ .nr_extents = 1UL << order,
+ .extent_order = 0,
+ .domid = DOMID_SELF
+ },
+ .out = {
-+ .nr_extents = 1UL << order,
+ .extent_order = 0,
+ .address_bits = address_bits,
+ .domid = DOMID_SELF
+ if (unlikely(order > MAX_CONTIG_ORDER))
+ return -ENOMEM;
+
++ bitmap_zero(limit_map, 1U << order);
+ set_xen_guest_handle(exchange.in.extent_start, in_frames);
+ set_xen_guest_handle(exchange.out.extent_start, out_frames);
+
+ /* 0. Scrub the pages. */
-+ for ( i = 0 ; i < 1UL<<order ; i++ ) {
++ for (i = 0, n = 0; i < 1U<<order ; i++) {
+ page = &pages[i];
++ if (!(pfn_to_mfn(page_to_pfn(page)) >> (address_bits - PAGE_SHIFT)))
++ continue;
++ __set_bit(i, limit_map);
+
-+ if (!PageHighMem(page)) {
-+ v = page_address(page);
-+ scrub_pages(v, 1);
-+ } else {
-+ v = kmap(page);
-+ scrub_pages(v, 1);
++ if (!PageHighMem(page))
++ scrub_pages(page_address(page), 1);
++#ifdef CONFIG_XEN_SCRUB_PAGES
++ else {
++ scrub_pages(kmap(page), 1);
+ kunmap(page);
++ ++n;
+ }
++#endif
+ }
++ if (bitmap_empty(limit_map, 1U << order))
++ return 0;
+
-+ kmap_flush_unused();
++ if (n)
++ kmap_flush_unused();
+
+ balloon_lock(flags);
+
+ /* 1. Zap current PTEs (if any), remembering MFNs. */
-+ for (i = 0, nr_mcl = 0; i < (1UL<<order); i++) {
++ for (i = 0, n = 0, nr_mcl = 0; i < (1U<<order); i++) {
++ if(!test_bit(i, limit_map))
++ continue;
+ page = &pages[i];
+
-+ out_frames[i] = page_to_pfn(page);
-+ in_frames[i] = pfn_to_mfn(out_frames[i]);
++ out_frames[n] = page_to_pfn(page);
++ in_frames[n] = pfn_to_mfn(out_frames[n]);
+
+ if (!PageHighMem(page))
+ MULTI_update_va_mapping(cr_mcl + nr_mcl++,
+ (unsigned long)page_address(page),
+ __pte_ma(0), 0);
+
-+ set_phys_to_machine(out_frames[i], INVALID_P2M_ENTRY);
++ set_phys_to_machine(out_frames[n], INVALID_P2M_ENTRY);
++ ++n;
+ }
-+ if (HYPERVISOR_multicall_check(cr_mcl, nr_mcl, NULL))
++ if (nr_mcl && HYPERVISOR_multicall_check(cr_mcl, nr_mcl, NULL))
+ BUG();
+
+ /* 2. Get new memory below the required limit. */
++ exchange.in.nr_extents = n;
++ exchange.out.nr_extents = n;
+ rc = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
-+ success = (exchange.nr_exchanged == (1UL << order));
++ success = (exchange.nr_exchanged == n);
+ BUG_ON(!success && ((exchange.nr_exchanged != 0) || (rc == 0)));
+ BUG_ON(success && (rc != 0));
+#if CONFIG_XEN_COMPAT <= 0x030002
+ if (unlikely(rc == -ENOSYS)) {
+ /* Compatibility when XENMEM_exchange is unsupported. */
+ if (HYPERVISOR_memory_op(XENMEM_decrease_reservation,
-+ &exchange.in) != (1UL << order))
++ &exchange.in) != n)
+ BUG();
-+ success = (HYPERVISOR_memory_op(XENMEM_populate_physmap,
-+ &exchange.out) != (1UL <<order));
++ if (HYPERVISOR_memory_op(XENMEM_populate_physmap,
++ &exchange.out) != n)
++ BUG();
++ success = 1;
+ }
+#endif
+
+ /* 3. Map the new pages in place of old pages. */
-+ for (i = 0, nr_mcl = 0; i < (1UL<<order); i++) {
-+ unsigned long pfn;
++ for (i = 0, n = 0, nr_mcl = 0; i < (1U<<order); i++) {
++ if(!test_bit(i, limit_map))
++ continue;
+ page = &pages[i];
-+ pfn = page_to_pfn(page);
+
-+ frame = success ? out_frames[i] : in_frames[i];
++ frame = success ? out_frames[n] : in_frames[n];
+
+ if (!PageHighMem(page))
+ MULTI_update_va_mapping(cr_mcl + nr_mcl++,
+ (unsigned long)page_address(page),
+ pfn_pte_ma(frame, PAGE_KERNEL), 0);
+
-+ set_phys_to_machine(pfn, frame);
++ set_phys_to_machine(page_to_pfn(page), frame);
++ ++n;
++ }
++ if (nr_mcl) {
++ cr_mcl[nr_mcl - 1].args[MULTI_UVMFLAGS_INDEX] = order
++ ? UVMF_TLB_FLUSH|UVMF_ALL
++ : UVMF_INVLPG|UVMF_ALL;
++ if (HYPERVISOR_multicall_check(cr_mcl, nr_mcl, NULL))
++ BUG();
+ }
-+ cr_mcl[nr_mcl - 1].args[MULTI_UVMFLAGS_INDEX] = order
-+ ? UVMF_TLB_FLUSH|UVMF_ALL
-+ : UVMF_INVLPG|UVMF_ALL;
-+ if (HYPERVISOR_multicall_check(cr_mcl, nr_mcl, NULL))
-+ BUG();
+
+ balloon_unlock(flags);
+
+ mach_lp, (u64)entry_a | ((u64)entry_b<<32));
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/init-xen.c linux-2.6.18-xen.hg/arch/i386/mm/init-xen.c
---- linux-2.6.18/arch/i386/mm/init-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/init-xen.c 2007-12-23 11:14:54.522526395 +0100
-@@ -0,0 +1,850 @@
++
++#define MAX_BATCHED_FULL_PTES 32
++
++int xen_change_pte_range(struct mm_struct *mm, pmd_t *pmd,
++ unsigned long addr, unsigned long end, pgprot_t newprot)
++{
++ int rc = 0, i = 0;
++ mmu_update_t u[MAX_BATCHED_FULL_PTES];
++ pte_t *pte;
++ spinlock_t *ptl;
++
++ if (!xen_feature(XENFEAT_mmu_pt_update_preserve_ad))
++ return 0;
++
++ pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
++ do {
++ if (pte_present(*pte)) {
++ u[i].ptr = (__pmd_val(*pmd) & PHYSICAL_PAGE_MASK)
++ | ((unsigned long)pte & ~PAGE_MASK)
++ | MMU_PT_UPDATE_PRESERVE_AD;
++ u[i].val = __pte_val(pte_modify(*pte, newprot));
++ if (++i == MAX_BATCHED_FULL_PTES) {
++ if ((rc = HYPERVISOR_mmu_update(
++ &u[0], i, NULL, DOMID_SELF)) != 0)
++ break;
++ i = 0;
++ }
++ }
++ } while (pte++, addr += PAGE_SIZE, addr != end);
++ if (i)
++ rc = HYPERVISOR_mmu_update( &u[0], i, NULL, DOMID_SELF);
++ pte_unmap_unlock(pte - 1, ptl);
++ BUG_ON(rc && rc != -ENOSYS);
++ return !rc;
++}
+--- linux-2.6.18.8/arch/i386/mm/init-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/init-xen.c 2008-05-19 00:33:03.968374893 +0300
+@@ -0,0 +1,847 @@
+/*
+ * linux/arch/i386/mm/init.c
+ *
+ unsigned long vaddr;
+ pgd_t *pgd_base = (pgd_t *)xen_start_info->pt_base;
+
-+ swapper_pg_dir = pgd_base;
-+ init_mm.pgd = pgd_base;
-+
+ /* Enable PSE if available */
+ if (cpu_has_pse) {
+ set_in_cr4(X86_CR4_PSE);
+}
+#endif
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/ioremap-xen.c linux-2.6.18-xen.hg/arch/i386/mm/ioremap-xen.c
---- linux-2.6.18/arch/i386/mm/ioremap-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/ioremap-xen.c 2007-12-23 11:14:54.522526395 +0100
+--- linux-2.6.18.8/arch/i386/mm/ioremap-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/ioremap-xen.c 2008-05-19 00:33:03.968374893 +0300
@@ -0,0 +1,443 @@
+/*
+ * arch/i386/mm/ioremap.c
+ * Fill in the machine address: PTE ptr is done later by
+ * apply_to_page_range().
+ */
-+ v->val = __pte_val(pfn_pte_ma(mfn, prot));
++ v->val = __pte_val(pfn_pte_ma(mfn, prot)) | _PAGE_IO;
+
+ mfn++;
+ address += PAGE_SIZE;
+ --nrpages;
+ }
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/Makefile linux-2.6.18-xen.hg/arch/i386/mm/Makefile
---- linux-2.6.18/arch/i386/mm/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/mm/Makefile 2007-12-23 11:14:54.515859375 +0100
-@@ -8,3 +8,4 @@
- obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
- obj-$(CONFIG_HIGHMEM) += highmem.o
- obj-$(CONFIG_BOOT_IOREMAP) += boot_ioremap.o
-+obj-$(CONFIG_XEN) += hypervisor.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/pageattr.c linux-2.6.18-xen.hg/arch/i386/mm/pageattr.c
---- linux-2.6.18/arch/i386/mm/pageattr.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/mm/pageattr.c 2007-12-23 11:14:54.525859899 +0100
+--- linux-2.6.18.8/arch/i386/mm/pageattr.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/pageattr.c 2008-05-19 00:33:03.972375123 +0300
@@ -84,7 +84,7 @@
unsigned long flags;
return;
spin_lock_irqsave(&pgd_lock, flags);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/pgtable.c linux-2.6.18-xen.hg/arch/i386/mm/pgtable.c
---- linux-2.6.18/arch/i386/mm/pgtable.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/mm/pgtable.c 2007-12-23 11:14:54.525859899 +0100
-@@ -12,6 +12,7 @@
- #include <linux/slab.h>
- #include <linux/pagemap.h>
- #include <linux/spinlock.h>
-+#include <linux/module.h>
-
- #include <asm/system.h>
- #include <asm/pgtable.h>
-@@ -137,6 +138,10 @@
- __flush_tlb_one(vaddr);
- }
-
-+static int nr_fixmaps = 0;
-+unsigned long __FIXADDR_TOP = 0xfffff000;
-+EXPORT_SYMBOL(__FIXADDR_TOP);
-+
- void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags)
- {
- unsigned long address = __fix_to_virt(idx);
-@@ -146,6 +151,13 @@
- return;
- }
- set_pte_pfn(address, phys >> PAGE_SHIFT, flags);
-+ nr_fixmaps++;
-+}
-+
-+void set_fixaddr_top(unsigned long top)
-+{
-+ BUG_ON(nr_fixmaps > 0);
-+ __FIXADDR_TOP = top - PAGE_SIZE;
- }
-
- pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
-@@ -214,6 +226,7 @@
- spin_lock_irqsave(&pgd_lock, flags);
- }
-
-+ if (PTRS_PER_PMD == 1 || HAVE_SHARED_KERNEL_PMD)
- clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
- swapper_pg_dir + USER_PTRS_PER_PGD,
- KERNEL_PGD_PTRS);
-@@ -248,6 +261,30 @@
- goto out_oom;
- set_pgd(&pgd[i], __pgd(1 + __pa(pmd)));
- }
-+
-+ if (!HAVE_SHARED_KERNEL_PMD) {
-+ unsigned long flags;
-+
-+ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
-+ pmd_t *pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL);
-+ if (!pmd)
-+ goto out_oom;
-+ set_pgd(&pgd[USER_PTRS_PER_PGD], __pgd(1 + __pa(pmd)));
-+ }
-+
-+ spin_lock_irqsave(&pgd_lock, flags);
-+ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
-+ unsigned long v = (unsigned long)i << PGDIR_SHIFT;
-+ pgd_t *kpgd = pgd_offset_k(v);
-+ pud_t *kpud = pud_offset(kpgd, v);
-+ pmd_t *kpmd = pmd_offset(kpud, v);
-+ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
-+ memcpy(pmd, kpmd, PAGE_SIZE);
-+ }
-+ pgd_list_add(pgd);
-+ spin_unlock_irqrestore(&pgd_lock, flags);
-+ }
-+
- return pgd;
-
- out_oom:
-@@ -262,9 +299,23 @@
- int i;
-
- /* in the PAE case user pgd entries are overwritten before usage */
-- if (PTRS_PER_PMD > 1)
-- for (i = 0; i < USER_PTRS_PER_PGD; ++i)
-- kmem_cache_free(pmd_cache, (void *)__va(pgd_val(pgd[i])-1));
-+ if (PTRS_PER_PMD > 1) {
-+ for (i = 0; i < USER_PTRS_PER_PGD; ++i) {
-+ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
-+ kmem_cache_free(pmd_cache, pmd);
-+ }
-+ if (!HAVE_SHARED_KERNEL_PMD) {
-+ unsigned long flags;
-+ spin_lock_irqsave(&pgd_lock, flags);
-+ pgd_list_del(pgd);
-+ spin_unlock_irqrestore(&pgd_lock, flags);
-+ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
-+ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
-+ memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
-+ kmem_cache_free(pmd_cache, pmd);
-+ }
-+ }
-+ }
- /* in the non-PAE case, free_pgtables() clears user pgd entries */
- kmem_cache_free(pgd_cache, pgd);
- }
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/mm/pgtable-xen.c linux-2.6.18-xen.hg/arch/i386/mm/pgtable-xen.c
---- linux-2.6.18/arch/i386/mm/pgtable-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/mm/pgtable-xen.c 2007-12-23 11:14:54.525859899 +0100
+--- linux-2.6.18.8/arch/i386/mm/pgtable-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/pgtable-xen.c 2008-05-19 00:33:03.972375123 +0300
@@ -0,0 +1,725 @@
+/*
+ * linux/arch/i386/mm/pgtable.c
+ !mm->context.has_foreign_mappings)
+ mm_unpin(mm);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/oprofile/Makefile linux-2.6.18-xen.hg/arch/i386/oprofile/Makefile
---- linux-2.6.18/arch/i386/oprofile/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/oprofile/Makefile 2007-12-23 11:14:54.529193404 +0100
+--- linux-2.6.18.8/arch/i386/mm/pgtable.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/mm/pgtable.c 2008-05-19 00:33:03.972375123 +0300
+@@ -12,6 +12,7 @@
+ #include <linux/slab.h>
+ #include <linux/pagemap.h>
+ #include <linux/spinlock.h>
++#include <linux/module.h>
+
+ #include <asm/system.h>
+ #include <asm/pgtable.h>
+@@ -137,6 +138,10 @@
+ __flush_tlb_one(vaddr);
+ }
+
++static int nr_fixmaps = 0;
++unsigned long __FIXADDR_TOP = 0xfffff000;
++EXPORT_SYMBOL(__FIXADDR_TOP);
++
+ void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags)
+ {
+ unsigned long address = __fix_to_virt(idx);
+@@ -146,6 +151,13 @@
+ return;
+ }
+ set_pte_pfn(address, phys >> PAGE_SHIFT, flags);
++ nr_fixmaps++;
++}
++
++void set_fixaddr_top(unsigned long top)
++{
++ BUG_ON(nr_fixmaps > 0);
++ __FIXADDR_TOP = top - PAGE_SIZE;
+ }
+
+ pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
+@@ -214,6 +226,7 @@
+ spin_lock_irqsave(&pgd_lock, flags);
+ }
+
++ if (PTRS_PER_PMD == 1 || HAVE_SHARED_KERNEL_PMD)
+ clone_pgd_range((pgd_t *)pgd + USER_PTRS_PER_PGD,
+ swapper_pg_dir + USER_PTRS_PER_PGD,
+ KERNEL_PGD_PTRS);
+@@ -248,6 +261,30 @@
+ goto out_oom;
+ set_pgd(&pgd[i], __pgd(1 + __pa(pmd)));
+ }
++
++ if (!HAVE_SHARED_KERNEL_PMD) {
++ unsigned long flags;
++
++ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
++ pmd_t *pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL);
++ if (!pmd)
++ goto out_oom;
++ set_pgd(&pgd[USER_PTRS_PER_PGD], __pgd(1 + __pa(pmd)));
++ }
++
++ spin_lock_irqsave(&pgd_lock, flags);
++ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
++ unsigned long v = (unsigned long)i << PGDIR_SHIFT;
++ pgd_t *kpgd = pgd_offset_k(v);
++ pud_t *kpud = pud_offset(kpgd, v);
++ pmd_t *kpmd = pmd_offset(kpud, v);
++ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
++ memcpy(pmd, kpmd, PAGE_SIZE);
++ }
++ pgd_list_add(pgd);
++ spin_unlock_irqrestore(&pgd_lock, flags);
++ }
++
+ return pgd;
+
+ out_oom:
+@@ -262,9 +299,23 @@
+ int i;
+
+ /* in the PAE case user pgd entries are overwritten before usage */
+- if (PTRS_PER_PMD > 1)
+- for (i = 0; i < USER_PTRS_PER_PGD; ++i)
+- kmem_cache_free(pmd_cache, (void *)__va(pgd_val(pgd[i])-1));
++ if (PTRS_PER_PMD > 1) {
++ for (i = 0; i < USER_PTRS_PER_PGD; ++i) {
++ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
++ kmem_cache_free(pmd_cache, pmd);
++ }
++ if (!HAVE_SHARED_KERNEL_PMD) {
++ unsigned long flags;
++ spin_lock_irqsave(&pgd_lock, flags);
++ pgd_list_del(pgd);
++ spin_unlock_irqrestore(&pgd_lock, flags);
++ for (i = USER_PTRS_PER_PGD; i < PTRS_PER_PGD; i++) {
++ pmd_t *pmd = (void *)__va(pgd_val(pgd[i])-1);
++ memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
++ kmem_cache_free(pmd_cache, pmd);
++ }
++ }
++ }
+ /* in the non-PAE case, free_pgtables() clears user pgd entries */
+ kmem_cache_free(pgd_cache, pgd);
+ }
+--- linux-2.6.18.8/arch/i386/oprofile/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/oprofile/Makefile 2008-05-19 00:33:03.976375354 +0300
@@ -6,7 +6,14 @@
oprofilefs.o oprofile_stats.o \
timer_int.o )
op_model_ppro.o op_model_p4.o
oprofile-$(CONFIG_X86_IO_APIC) += nmi_timer_int.o
+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/oprofile/xenoprof.c linux-2.6.18-xen.hg/arch/i386/oprofile/xenoprof.c
---- linux-2.6.18/arch/i386/oprofile/xenoprof.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/oprofile/xenoprof.c 2007-12-23 11:14:54.532526912 +0100
+--- linux-2.6.18.8/arch/i386/oprofile/xenoprof.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/oprofile/xenoprof.c 2008-05-19 00:33:03.984375815 +0300
@@ -0,0 +1,179 @@
+/**
+ * @file xenoprof.c
+ counter.kernel = (uint32_t)counter_config[i].kernel;
+ counter.user = (uint32_t)counter_config[i].user;
+ counter.unit_mask = (uint64_t)counter_config[i].unit_mask;
-+ HYPERVISOR_xenoprof_op(XENOPROF_counter,
-+ &counter);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_counter,
++ &counter));
+ }
+}
+
+{
+ xenoprofile_exit();
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/pci/irq.c linux-2.6.18-xen.hg/arch/i386/pci/irq.c
---- linux-2.6.18/arch/i386/pci/irq.c 2007-12-23 11:26:51.003453595 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/pci/irq.c 2007-12-23 11:14:54.535860422 +0100
-@@ -543,6 +543,12 @@
- case PCI_DEVICE_ID_INTEL_ICH8_2:
- case PCI_DEVICE_ID_INTEL_ICH8_3:
- case PCI_DEVICE_ID_INTEL_ICH8_4:
-+ case PCI_DEVICE_ID_INTEL_ICH9_0:
-+ case PCI_DEVICE_ID_INTEL_ICH9_1:
-+ case PCI_DEVICE_ID_INTEL_ICH9_2:
-+ case PCI_DEVICE_ID_INTEL_ICH9_3:
-+ case PCI_DEVICE_ID_INTEL_ICH9_4:
-+ case PCI_DEVICE_ID_INTEL_ICH9_5:
- r->name = "PIIX/ICH";
- r->get = pirq_piix_get;
- r->set = pirq_piix_set;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/pci/irq-xen.c linux-2.6.18-xen.hg/arch/i386/pci/irq-xen.c
---- linux-2.6.18/arch/i386/pci/irq-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/pci/irq-xen.c 2007-12-23 11:14:54.535860422 +0100
-@@ -0,0 +1,1205 @@
+--- linux-2.6.18.8/arch/i386/pci/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/pci/Makefile 2008-05-19 00:33:03.984375815 +0300
+@@ -4,6 +4,10 @@
+ obj-$(CONFIG_PCI_MMCONFIG) += mmconfig.o direct.o
+ obj-$(CONFIG_PCI_DIRECT) += direct.o
+
++# pcifront should be after pcbios.o, mmconfig.o, and direct.o as it should only
++# take over if direct access to the PCI bus is unavailable
++obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += pcifront.o
++
+ pci-y := fixup.o
+ pci-$(CONFIG_ACPI) += acpi.o
+ pci-y += legacy.o irq.o
+--- linux-2.6.18.8/arch/i386/pci/irq-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/pci/irq-xen.c 2008-05-19 00:33:03.988376046 +0300
+@@ -0,0 +1,1215 @@
+/*
+ * Low-Level PCI Support for PC -- Routing of Interrupts
+ *
+ */
+static int pirq_via586_get(struct pci_dev *router, struct pci_dev *dev, int pirq)
+{
-+ static const unsigned int pirqmap[4] = { 3, 2, 5, 1 };
++ static const unsigned int pirqmap[5] = { 3, 2, 5, 1, 1 };
+ return read_config_nybble(router, 0x55, pirqmap[pirq-1]);
+}
+
+static int pirq_via586_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq)
+{
-+ static const unsigned int pirqmap[4] = { 3, 2, 5, 1 };
++ static const unsigned int pirqmap[5] = { 3, 2, 5, 1, 1 };
+ write_config_nybble(router, 0x55, pirqmap[pirq-1], irq);
+ return 1;
+}
+ case PCI_DEVICE_ID_INTEL_ICH8_2:
+ case PCI_DEVICE_ID_INTEL_ICH8_3:
+ case PCI_DEVICE_ID_INTEL_ICH8_4:
++ case PCI_DEVICE_ID_INTEL_ICH9_0:
++ case PCI_DEVICE_ID_INTEL_ICH9_1:
++ case PCI_DEVICE_ID_INTEL_ICH9_2:
++ case PCI_DEVICE_ID_INTEL_ICH9_3:
++ case PCI_DEVICE_ID_INTEL_ICH9_4:
++ case PCI_DEVICE_ID_INTEL_ICH9_5:
++ case PCI_DEVICE_ID_INTEL_ICH10_0:
++ case PCI_DEVICE_ID_INTEL_ICH10_1:
++ case PCI_DEVICE_ID_INTEL_ICH10_2:
++ case PCI_DEVICE_ID_INTEL_ICH10_3:
+ r->name = "PIIX/ICH";
+ r->get = pirq_piix_get;
+ r->set = pirq_piix_set;
+
+ return count;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/pci/Makefile linux-2.6.18-xen.hg/arch/i386/pci/Makefile
---- linux-2.6.18/arch/i386/pci/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/pci/Makefile 2007-12-23 11:14:54.532526912 +0100
-@@ -4,6 +4,10 @@
- obj-$(CONFIG_PCI_MMCONFIG) += mmconfig.o direct.o
- obj-$(CONFIG_PCI_DIRECT) += direct.o
-
-+# pcifront should be after pcbios.o, mmconfig.o, and direct.o as it should only
-+# take over if direct access to the PCI bus is unavailable
-+obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += pcifront.o
-+
- pci-y := fixup.o
- pci-$(CONFIG_ACPI) += acpi.o
- pci-y += legacy.o irq.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/pci/pcifront.c linux-2.6.18-xen.hg/arch/i386/pci/pcifront.c
---- linux-2.6.18/arch/i386/pci/pcifront.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/i386/pci/pcifront.c 2007-12-23 11:14:54.539193935 +0100
+--- linux-2.6.18.8/arch/i386/pci/irq.c 2008-05-19 00:42:33.781223047 +0300
++++ linux-2.6.18-xen.hg/arch/i386/pci/irq.c 2008-05-19 00:33:03.992376276 +0300
+@@ -543,6 +543,16 @@
+ case PCI_DEVICE_ID_INTEL_ICH8_2:
+ case PCI_DEVICE_ID_INTEL_ICH8_3:
+ case PCI_DEVICE_ID_INTEL_ICH8_4:
++ case PCI_DEVICE_ID_INTEL_ICH9_0:
++ case PCI_DEVICE_ID_INTEL_ICH9_1:
++ case PCI_DEVICE_ID_INTEL_ICH9_2:
++ case PCI_DEVICE_ID_INTEL_ICH9_3:
++ case PCI_DEVICE_ID_INTEL_ICH9_4:
++ case PCI_DEVICE_ID_INTEL_ICH9_5:
++ case PCI_DEVICE_ID_INTEL_ICH10_0:
++ case PCI_DEVICE_ID_INTEL_ICH10_1:
++ case PCI_DEVICE_ID_INTEL_ICH10_2:
++ case PCI_DEVICE_ID_INTEL_ICH10_3:
+ r->name = "PIIX/ICH";
+ r->get = pirq_piix_get;
+ r->set = pirq_piix_set;
+--- linux-2.6.18.8/arch/i386/pci/pcifront.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/pci/pcifront.c 2008-05-19 00:33:03.996376507 +0300
@@ -0,0 +1,55 @@
+/*
+ * PCI Frontend Stub - puts some "dummy" functions in to the Linux x86 PCI core
+}
+
+arch_initcall(pcifront_x86_stub_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/power/cpu.c linux-2.6.18-xen.hg/arch/i386/power/cpu.c
---- linux-2.6.18/arch/i386/power/cpu.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/power/cpu.c 2007-12-23 11:14:54.542527446 +0100
+--- linux-2.6.18.8/arch/i386/power/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/power/Makefile 2008-05-19 00:33:03.996376507 +0300
+@@ -1,2 +1,4 @@
+-obj-$(CONFIG_PM) += cpu.o
++obj-$(subst m,y,$(CONFIG_APM)) += cpu.o
++obj-$(CONFIG_SOFTWARE_SUSPEND) += cpu.o
++obj-$(CONFIG_ACPI_SLEEP) += cpu.o
+ obj-$(CONFIG_SOFTWARE_SUSPEND) += swsusp.o
+--- linux-2.6.18.8/arch/i386/power/cpu.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/i386/power/cpu.c 2008-05-19 00:33:04.000376737 +0300
@@ -62,11 +62,12 @@
static void fix_processor_context(void)
load_TR_desc(); /* This does ltr */
load_LDT(¤t->active_mm->context); /* This does lldt */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/i386/power/Makefile linux-2.6.18-xen.hg/arch/i386/power/Makefile
---- linux-2.6.18/arch/i386/power/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/i386/power/Makefile 2007-12-23 11:14:54.542527446 +0100
-@@ -1,2 +1,4 @@
--obj-$(CONFIG_PM) += cpu.o
-+obj-$(subst m,y,$(CONFIG_APM)) += cpu.o
-+obj-$(CONFIG_SOFTWARE_SUSPEND) += cpu.o
-+obj-$(CONFIG_ACPI_SLEEP) += cpu.o
- obj-$(CONFIG_SOFTWARE_SUSPEND) += swsusp.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/hp/common/sba_iommu.c linux-2.6.18-xen.hg/arch/ia64/hp/common/sba_iommu.c
---- linux-2.6.18/arch/ia64/hp/common/sba_iommu.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/hp/common/sba_iommu.c 2007-12-23 11:14:54.552527964 +0100
+--- linux-2.6.18.8/arch/ia64/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/Kconfig 2008-05-19 00:33:04.000376737 +0300
+@@ -58,6 +58,28 @@
+ bool
+ default y
+
++config XEN
++ bool "Xen hypervisor support"
++ default y
++ select XEN_XENCOMM
++ help
++ Enable Xen hypervisor support. Resulting kernel runs
++ both as a guest OS on Xen and natively on hardware.
++
++config XEN_IA64_EXPOSE_P2M
++ bool "Xen/IA64 exposure p2m table"
++ depends on XEN
++ default y
++ help
++ expose p2m from xen
++
++config XEN_IA64_EXPOSE_P2M_USE_DTR
++ bool "Xen/IA64 map p2m table with dtr"
++ depends on XEN_IA64_EXPOSE_P2M
++ default y
++ help
++ use dtr to map the exposed p2m table
++
+ config SCHED_NO_NO_OMIT_FRAME_POINTER
+ bool
+ default y
+@@ -133,6 +155,10 @@
+ config IA64_HP_SIM
+ bool "Ski-simulator"
+
++config IA64_XEN
++ bool "Xen guest"
++ depends on XEN
++
+ endchoice
+
+ choice
+@@ -431,6 +457,29 @@
+
+ source "drivers/sn/Kconfig"
+
++config KEXEC
++ bool "kexec system call (EXPERIMENTAL)"
++ depends on EXPERIMENTAL && !IA64_HP_SIM && (!SMP || HOTPLUG_CPU) && !XEN_UNPRIVILEGED_GUEST
++ help
++ kexec is a system call that implements the ability to shutdown your
++ current kernel, and to start another kernel. It is like a reboot
++ but it is indepedent of the system firmware. And like a reboot
++ you can start any kernel with it, not just Linux.
++
++ The name comes from the similiarity to the exec system call.
++
++ It is an ongoing process to be certain the hardware in a machine
++ is properly shutdown, so do not be surprised if this code does not
++ initially work for you. It may help to enable device hotplugging
++ support. As of this writing the exact hardware interface is
++ strongly in flux, so no good recommendation can be made.
++
++config CRASH_DUMP
++ bool "kernel crash dumps (EXPERIMENTAL)"
++ depends on EXPERIMENTAL && IA64_MCA_RECOVERY && !IA64_HP_SIM && (!SMP || HOTPLUG_CPU)
++ help
++ Generate crash dump after being started by kexec.
++
+ source "drivers/firmware/Kconfig"
+
+ source "fs/Kconfig.binfmt"
+@@ -465,6 +514,21 @@
+ bool
+ default PCI
+
++config XEN_PCIDEV_FRONTEND
++ bool "Xen PCI Frontend"
++ depends on PCI && XEN
++ default y
++ help
++ The PCI device frontend driver allows the kernel to import arbitrary
++ PCI devices from a PCI backend to support PCI driver domains.
++
++config XEN_PCIDEV_FE_DEBUG
++ bool "Xen PCI Frontend Debugging"
++ depends on XEN_PCIDEV_FRONTEND
++ default n
++ help
++ Enables some debug statements within the PCI Frontend.
++
+ source "drivers/pci/pcie/Kconfig"
+
+ source "drivers/pci/Kconfig"
+@@ -528,3 +592,16 @@
+ source "security/Kconfig"
+
+ source "crypto/Kconfig"
++
++#
++# override default values of drivers/xen/Kconfig
++#
++if XEN
++config XEN_SMPBOOT
++ default n
++
++config XEN_DEVMEM
++ default n
++endif
++
++source "drivers/xen/Kconfig"
+--- linux-2.6.18.8/arch/ia64/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/Makefile 2008-05-19 00:33:04.000376737 +0300
+@@ -45,6 +45,12 @@
+ endif
+
+ CFLAGS += $(cflags-y)
++
++cppflags-$(CONFIG_XEN) += \
++ -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION)
++
++CPPFLAGS += $(cppflags-y)
++
+ head-y := arch/ia64/kernel/head.o arch/ia64/kernel/init_task.o
+
+ libs-y += arch/ia64/lib/
+@@ -54,7 +60,9 @@
+ core-$(CONFIG_IA64_GENERIC) += arch/ia64/dig/
+ core-$(CONFIG_IA64_HP_ZX1) += arch/ia64/dig/
+ core-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/dig/
++core-$(CONFIG_IA64_XEN) += arch/ia64/dig/
+ core-$(CONFIG_IA64_SGI_SN2) += arch/ia64/sn/
++core-$(CONFIG_XEN) += arch/ia64/xen/
+
+ drivers-$(CONFIG_PCI) += arch/ia64/pci/
+ drivers-$(CONFIG_IA64_HP_SIM) += arch/ia64/hp/sim/
+@@ -87,8 +95,8 @@
+ boot: lib/lib.a vmlinux
+ $(Q)$(MAKE) $(build)=$(boot) $@
+
+-install: vmlinux.gz
+- sh $(srctree)/arch/ia64/install.sh $(KERNELRELEASE) $< System.map "$(INSTALL_PATH)"
++install:
++ -yes | sh $(srctree)/arch/ia64/install.sh $(KERNELRELEASE) vmlinux.gz System.map "$(INSTALL_PATH)"
+
+ define archhelp
+ echo '* compressed - Build compressed kernel image'
+--- linux-2.6.18.8/arch/ia64/hp/common/sba_iommu.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/hp/common/sba_iommu.c 2008-05-19 00:33:04.016377660 +0300
@@ -42,6 +42,11 @@
#include <asm/system.h> /* wmb() */
DBG_INIT("%s() prefetch spill addr: 0x%lx\n", __FUNCTION__, prefetch_spill_page);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/Kconfig linux-2.6.18-xen.hg/arch/ia64/Kconfig
---- linux-2.6.18/arch/ia64/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/Kconfig 2007-12-23 11:14:54.542527446 +0100
-@@ -58,6 +58,28 @@
- bool
- default y
-
-+config XEN
-+ bool "Xen hypervisor support"
-+ default y
-+ select XEN_XENCOMM
-+ help
-+ Enable Xen hypervisor support. Resulting kernel runs
-+ both as a guest OS on Xen and natively on hardware.
-+
-+config XEN_IA64_EXPOSE_P2M
-+ bool "Xen/IA64 exposure p2m table"
-+ depends on XEN
-+ default y
-+ help
-+ expose p2m from xen
-+
-+config XEN_IA64_EXPOSE_P2M_USE_DTR
-+ bool "Xen/IA64 map p2m table with dtr"
-+ depends on XEN_IA64_EXPOSE_P2M
-+ default y
-+ help
-+ use dtr to map the exposed p2m table
+--- linux-2.6.18.8/arch/ia64/kernel/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/Makefile 2008-05-19 00:33:04.320395183 +0300
+@@ -28,6 +28,8 @@
+ obj-$(CONFIG_CPU_FREQ) += cpufreq/
+ obj-$(CONFIG_IA64_MCA_RECOVERY) += mca_recovery.o
+ obj-$(CONFIG_KPROBES) += kprobes.o jprobes.o
++obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
++obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
+ obj-$(CONFIG_IA64_UNCACHED_ALLOCATOR) += uncached.o
+ obj-$(CONFIG_AUDIT) += audit.o
+ mca_recovery-y += mca_drv.o mca_drv_asm.o
+@@ -61,3 +63,61 @@
+ # We must build gate.so before we can assemble it.
+ # Note: kbuild does not track this dependency due to usage of .incbin
+ $(obj)/gate-data.o: $(obj)/gate.so
+
- config SCHED_NO_NO_OMIT_FRAME_POINTER
- bool
- default y
-@@ -133,6 +155,10 @@
- config IA64_HP_SIM
- bool "Ski-simulator"
-
-+config IA64_XEN
-+ bool "Xen guest"
-+ depends on XEN
++#
++# gate page paravirtualization for xen
++#
++obj-$(CONFIG_XEN) += xengate-data.o
+
- endchoice
-
- choice
-@@ -431,6 +457,29 @@
-
- source "drivers/sn/Kconfig"
-
-+config KEXEC
-+ bool "kexec system call (EXPERIMENTAL)"
-+ depends on EXPERIMENTAL && !IA64_HP_SIM && (!SMP || HOTPLUG_CPU)
-+ help
-+ kexec is a system call that implements the ability to shutdown your
-+ current kernel, and to start another kernel. It is like a reboot
-+ but it is indepedent of the system firmware. And like a reboot
-+ you can start any kernel with it, not just Linux.
++ifeq ($(CONFIG_XEN), y)
++# The gate DSO image is built using a special linker script.
++targets += xengate.so xengate-syms.o
++endif
+
-+ The name comes from the similiarity to the exec system call.
++extra-$(CONFIG_XEN) += xengate.so xengate.lds xengate.o
+
-+ It is an ongoing process to be certain the hardware in a machine
-+ is properly shutdown, so do not be surprised if this code does not
-+ initially work for you. It may help to enable device hotplugging
-+ support. As of this writing the exact hardware interface is
-+ strongly in flux, so no good recommendation can be made.
++AFLAGS_xengate.o += -D__XEN_IA64_VDSO_PARAVIRT
++$(obj)/xengate.o: $(src)/gate.S FORCE
++ $(call if_changed_dep,as_o_S)
+
-+config CRASH_DUMP
-+ bool "kernel crash dumps (EXPERIMENTAL)"
-+ depends on EXPERIMENTAL && IA64_MCA_RECOVERY && !IA64_HP_SIM && (!SMP || HOTPLUG_CPU)
-+ help
-+ Generate crash dump after being started by kexec.
++CPPFLAGS_xengate.lds := -P -C -U$(ARCH) -D__XEN_IA64_VDSO_PARAVIRT
++$(obj)/xengate.lds: $(src)/gate.lds.S
++ $(call if_changed_dep,cpp_lds_S)
+
- source "drivers/firmware/Kconfig"
-
- source "fs/Kconfig.binfmt"
-@@ -465,6 +514,21 @@
- bool
- default PCI
-
-+config XEN_PCIDEV_FRONTEND
-+ bool "Xen PCI Frontend"
-+ depends on PCI && XEN
-+ default y
-+ help
-+ The PCI device frontend driver allows the kernel to import arbitrary
-+ PCI devices from a PCI backend to support PCI driver domains.
++GATECFLAGS_xengate.so = -shared -s -Wl,-soname=linux-gate.so.1 \
++ $(call ld-option, -Wl$(comma)--hash-style=sysv)
++$(obj)/xengate.so: $(obj)/xengate.lds $(obj)/xengate.o FORCE
++ $(call if_changed,gate)
+
-+config XEN_PCIDEV_FE_DEBUG
-+ bool "Xen PCI Frontend Debugging"
-+ depends on XEN_PCIDEV_FRONTEND
-+ default n
-+ help
-+ Enables some debug statements within the PCI Frontend.
++ifeq ($(CONFIG_XEN), y)
++$(obj)/built-in.o: $(obj)/xengate-syms.o
++$(obj)/built-in.o: ld_flags += -R $(obj)/xengate-syms.o
++$(obj)/mca_recovery.o: $(obj)/gate-syms.o $(obj)/xengate-syms.o
++endif
+
- source "drivers/pci/pcie/Kconfig"
-
- source "drivers/pci/Kconfig"
-@@ -528,3 +592,16 @@
- source "security/Kconfig"
-
- source "crypto/Kconfig"
++GATECFLAGS_xengate-syms.o = -r
++$(obj)/xengate-syms.o: $(obj)/xengate.lds $(obj)/xengate.o FORCE
++ $(call if_changed,gate)
++$(obj)/xengate-data.o: $(obj)/xengate.so
+
+#
-+# override default values of drivers/xen/Kconfig
++# .tmp_gate.o to calculate padding size for __kernel_syscall_via_epc
+#
-+if XEN
-+config XEN_SMPBOOT
-+ default n
++extra-$(CONFIG_XEN) += gate-skip.s .tmp_gate.o
+
-+config XEN_DEVMEM
-+ default n
++ifeq ($(CONFIG_XEN), y)
++AFLAGS_gate.o += -D__KERNEL_SYSCALL_VIA_EPC_PADDING
++$(obj)/gate.o: $(obj)/gate-skip.s FORCE
+endif
+
-+source "drivers/xen/Kconfig"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/acpi.c linux-2.6.18-xen.hg/arch/ia64/kernel/acpi.c
---- linux-2.6.18/arch/ia64/kernel/acpi.c 2007-12-23 11:26:51.003453595 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/acpi.c 2007-12-23 11:14:54.572529022 +0100
++$(obj)/.tmp_gate.o: $(src)/gate.S FORCE
++ $(call if_changed_dep,as_o_S)
++
++quiet_cmd_gate_size = GATE_SIZE $@
++ cmd_gate_size = $(NM) --extern-only --print-size $(obj)/xengate.o | \
++ $(AWK) '/__kernel_syscall_via_epc/{printf "\t.skip 0x"$$2" - "}' > $@; \
++ $(NM) --extern-only --print-size $(obj)/.tmp_gate.o | \
++ $(AWK) '/__kernel_syscall_via_epc/{printf "0x"$$2"\n"}' >> $@
++
++$(obj)/gate-skip.s: $(obj)/xengate.o $(obj)/.tmp_gate.o FORCE
++ $(call if_changed,gate_size)
+--- linux-2.6.18.8/arch/ia64/kernel/acpi.c 2008-05-19 00:42:33.781223047 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/acpi.c 2008-05-19 00:33:04.324395414 +0300
@@ -109,6 +109,10 @@
return "hpzx1";
} else if (!strcmp(hdr->oem_id, "SGI")) {
# else
# error Unknown platform. Fix acpi.c.
# endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/asm-offsets.c linux-2.6.18-xen.hg/arch/ia64/kernel/asm-offsets.c
---- linux-2.6.18/arch/ia64/kernel/asm-offsets.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/asm-offsets.c 2007-12-23 11:14:54.572529022 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/asm-offsets.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/asm-offsets.c 2008-05-19 00:33:04.328395644 +0300
@@ -268,4 +268,29 @@
DEFINE(IA64_TIME_SOURCE_MMIO64, TIME_SOURCE_MMIO64);
DEFINE(IA64_TIME_SOURCE_MMIO32, TIME_SOURCE_MMIO32);
+ DEFINE_MAPPED_REG_OFS(XSI_B1NATS_OFS, vnat);
+#endif /* CONFIG_XEN */
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/crash.c linux-2.6.18-xen.hg/arch/ia64/kernel/crash.c
---- linux-2.6.18/arch/ia64/kernel/crash.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/crash.c 2007-12-23 11:14:54.575862524 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/crash.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/crash.c 2008-05-19 00:33:04.336396106 +0300
@@ -0,0 +1,222 @@
+/*
+ * arch/ia64/kernel/crash.c
+
+__initcall(machine_crash_setup);
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/crash_dump.c linux-2.6.18-xen.hg/arch/ia64/kernel/crash_dump.c
---- linux-2.6.18/arch/ia64/kernel/crash_dump.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/crash_dump.c 2007-12-23 11:14:54.575862524 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/crash_dump.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/crash_dump.c 2008-05-19 00:33:04.336396106 +0300
@@ -0,0 +1,48 @@
+/*
+ * kernel/crash_dump.c - Memory preserving reboot related code.
+ return csize;
+}
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/efi.c linux-2.6.18-xen.hg/arch/ia64/kernel/efi.c
---- linux-2.6.18/arch/ia64/kernel/efi.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/efi.c 2007-12-23 11:14:54.579196027 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/efi.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/efi.c 2008-05-19 00:33:04.336396106 +0300
@@ -26,6 +26,7 @@
#include <linux/types.h>
#include <linux/time.h>
+ return ~0UL;
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/entry.S linux-2.6.18-xen.hg/arch/ia64/kernel/entry.S
---- linux-2.6.18/arch/ia64/kernel/entry.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/entry.S 2007-12-23 11:14:54.579196027 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/entry.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/entry.S 2008-05-19 00:33:04.340396336 +0300
@@ -180,7 +180,7 @@
* called. The code starting at .map relies on this. The rest of the code
* doesn't care about the interrupt masking status.
data8 sys_ni_syscall // reserved for vserver
data8 sys_waitid // 1270
data8 sys_add_key
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/fsys.S linux-2.6.18-xen.hg/arch/ia64/kernel/fsys.S
---- linux-2.6.18/arch/ia64/kernel/fsys.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/fsys.S 2007-12-23 11:14:54.582529532 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/fsys.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/fsys.S 2008-05-19 00:33:04.340396336 +0300
@@ -516,11 +516,34 @@
adds r17=-1024,r15
movl r14=sys_call_table
cmp.eq p8,p0=r3,r0 // A
(p10) br.cond.spnt.many ia64_ret_from_syscall // B return if bad call-frame or r15 is a NaT
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/gate.lds.S linux-2.6.18-xen.hg/arch/ia64/kernel/gate.lds.S
---- linux-2.6.18/arch/ia64/kernel/gate.lds.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/gate.lds.S 2007-12-23 11:14:54.582529532 +0100
-@@ -28,6 +28,24 @@
- . = GATE_ADDR + 0x500;
-
- .data.patch : {
-+#ifdef __XEN_IA64_VDSO_PARAVIRT
-+#define __start_gate_mckinley_e9_patchlist \
-+ __start_gate_mckinley_e9_patchlist_xen
-+#define __end_gate_mckinley_e9_patchlist \
-+ __end_gate_mckinley_e9_patchlist_xen
-+#define __start_gate_vtop_patchlist \
-+ __start_gate_vtop_patchlist_xen
-+#define __end_gate_vtop_patchlist \
-+ __end_gate_vtop_patchlist_xen
-+#define __start_gate_fsyscall_patchlist \
-+ __start_gate_fsyscall_patchlist_xen
-+#define __end_gate_fsyscall_patchlist \
-+ __end_gate_fsyscall_patchlist_xen
-+#define __start_gate_brl_fsys_bubble_down_patchlist \
-+ __start_gate_brl_fsys_bubble_down_patchlist_xen
-+#define __end_gate_brl_fsys_bubble_down_patchlist \
-+ __end_gate_brl_fsys_bubble_down_patchlist_xen
-+#endif
- __start_gate_mckinley_e9_patchlist = .;
- *(.data.patch.mckinley_e9)
- __end_gate_mckinley_e9_patchlist = .;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/gate.S linux-2.6.18-xen.hg/arch/ia64/kernel/gate.S
---- linux-2.6.18/arch/ia64/kernel/gate.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/gate.S 2007-12-23 11:14:54.582529532 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/gate.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/gate.S 2008-05-19 00:33:04.344396567 +0300
@@ -32,102 +32,6 @@
[1:](pr)brl.cond.sptk 0; \
.xdata4 ".data.patch.brl_fsys_bubble_down", 1b-.
+.include "arch/ia64/kernel/gate-skip.s"
+#endif
+END(__kernel_syscall_via_epc)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/head.S linux-2.6.18-xen.hg/arch/ia64/kernel/head.S
---- linux-2.6.18/arch/ia64/kernel/head.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/head.S 2007-12-23 11:14:54.585863043 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/gate.lds.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/gate.lds.S 2008-05-19 00:33:04.344396567 +0300
+@@ -28,6 +28,24 @@
+ . = GATE_ADDR + 0x500;
+
+ .data.patch : {
++#ifdef __XEN_IA64_VDSO_PARAVIRT
++#define __start_gate_mckinley_e9_patchlist \
++ __start_gate_mckinley_e9_patchlist_xen
++#define __end_gate_mckinley_e9_patchlist \
++ __end_gate_mckinley_e9_patchlist_xen
++#define __start_gate_vtop_patchlist \
++ __start_gate_vtop_patchlist_xen
++#define __end_gate_vtop_patchlist \
++ __end_gate_vtop_patchlist_xen
++#define __start_gate_fsyscall_patchlist \
++ __start_gate_fsyscall_patchlist_xen
++#define __end_gate_fsyscall_patchlist \
++ __end_gate_fsyscall_patchlist_xen
++#define __start_gate_brl_fsys_bubble_down_patchlist \
++ __start_gate_brl_fsys_bubble_down_patchlist_xen
++#define __end_gate_brl_fsys_bubble_down_patchlist \
++ __end_gate_brl_fsys_bubble_down_patchlist_xen
++#endif
+ __start_gate_mckinley_e9_patchlist = .;
+ *(.data.patch.mckinley_e9)
+ __end_gate_mckinley_e9_patchlist = .;
+--- linux-2.6.18.8/arch/ia64/kernel/head.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/head.S 2008-05-19 00:33:04.348396797 +0300
@@ -367,6 +367,12 @@
;;
(isBP) st8 [r2]=r28 // save the address of the boot param area passed by the bootloader
#ifdef CONFIG_SMP
(isAP) br.call.sptk.many rp=start_secondary
.ret0:
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/iosapic.c linux-2.6.18-xen.hg/arch/ia64/kernel/iosapic.c
---- linux-2.6.18/arch/ia64/kernel/iosapic.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/iosapic.c 2007-12-23 11:14:54.585863043 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/iosapic.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/iosapic.c 2008-05-19 00:33:04.348396797 +0300
@@ -159,6 +159,75 @@
static int iosapic_kmalloc_ok;
static LIST_HEAD(free_rte_list);
if (pcat_compat) {
/*
* Disable the compatibility mode interrupts (8259 style),
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/irq_ia64.c linux-2.6.18-xen.hg/arch/ia64/kernel/irq_ia64.c
---- linux-2.6.18/arch/ia64/kernel/irq_ia64.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/irq_ia64.c 2007-12-23 11:14:54.585863043 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/irq_ia64.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/irq_ia64.c 2008-05-19 00:33:04.352397028 +0300
@@ -30,6 +30,9 @@
#include <linux/smp_lock.h>
#include <linux/threads.h>
#ifdef CONFIG_SMP
phys_cpu_id = cpu_physical_id(cpu);
#else
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/machine_kexec.c linux-2.6.18-xen.hg/arch/ia64/kernel/machine_kexec.c
---- linux-2.6.18/arch/ia64/kernel/machine_kexec.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/machine_kexec.c 2007-12-23 11:14:54.589196561 +0100
-@@ -0,0 +1,145 @@
+--- linux-2.6.18.8/arch/ia64/kernel/machine_kexec.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/machine_kexec.c 2008-05-19 00:33:04.356397258 +0300
+@@ -0,0 +1,204 @@
+/*
+ * arch/ia64/kernel/machine_kexec.c
+ *
+ xki->reboot_code_buffer =
+ kexec_page_to_pfn(image->control_code_page) << PAGE_SHIFT;
+}
-+#endif /* CONFIG_XEN */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/Makefile linux-2.6.18-xen.hg/arch/ia64/kernel/Makefile
---- linux-2.6.18/arch/ia64/kernel/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/Makefile 2007-12-23 11:14:54.572529022 +0100
-@@ -28,6 +28,8 @@
- obj-$(CONFIG_CPU_FREQ) += cpufreq/
- obj-$(CONFIG_IA64_MCA_RECOVERY) += mca_recovery.o
- obj-$(CONFIG_KPROBES) += kprobes.o jprobes.o
-+obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
-+obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
- obj-$(CONFIG_IA64_UNCACHED_ALLOCATOR) += uncached.o
- obj-$(CONFIG_AUDIT) += audit.o
- mca_recovery-y += mca_drv.o mca_drv_asm.o
-@@ -61,3 +63,61 @@
- # We must build gate.so before we can assemble it.
- # Note: kbuild does not track this dependency due to usage of .incbin
- $(obj)/gate-data.o: $(obj)/gate.so
+
-+#
-+# gate page paravirtualization for xen
-+#
-+obj-$(CONFIG_XEN) += xengate-data.o
++static struct resource xen_hypervisor_heap_res;
+
-+ifeq ($(CONFIG_XEN), y)
-+# The gate DSO image is built using a special linker script.
-+targets += xengate.so xengate-syms.o
-+endif
++int __init machine_kexec_setup_resources(struct resource *hypervisor,
++ struct resource *phys_cpus,
++ int nr_phys_cpus)
++{
++ xen_kexec_range_t range;
++ int k;
+
-+extra-$(CONFIG_XEN) += xengate.so xengate.lds xengate.o
++ /* fill in xen_hypervisor_heap_res with hypervisor heap
++ * machine address range
++ */
+
-+AFLAGS_xengate.o += -D__XEN_IA64_VDSO_PARAVIRT
-+$(obj)/xengate.o: $(src)/gate.S FORCE
-+ $(call if_changed_dep,as_o_S)
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_XENHEAP;
+
-+CPPFLAGS_xengate.lds := -P -C -U$(ARCH) -D__XEN_IA64_VDSO_PARAVIRT
-+$(obj)/xengate.lds: $(src)/gate.lds.S
-+ $(call if_changed_dep,cpp_lds_S)
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ return -1;
+
-+GATECFLAGS_xengate.so = -shared -s -Wl,-soname=linux-gate.so.1 \
-+ $(call ld-option, -Wl$(comma)--hash-style=sysv)
-+$(obj)/xengate.so: $(obj)/xengate.lds $(obj)/xengate.o FORCE
-+ $(call if_changed,gate)
++ xen_hypervisor_heap_res.name = "Hypervisor heap";
++ xen_hypervisor_heap_res.start = range.start;
++ xen_hypervisor_heap_res.end = range.start + range.size - 1;
++ xen_hypervisor_heap_res.flags = IORESOURCE_BUSY | IORESOURCE_MEM;
+
-+ifeq ($(CONFIG_XEN), y)
-+$(obj)/built-in.o: $(obj)/xengate-syms.o
-+$(obj)/built-in.o: ld_flags += -R $(obj)/xengate-syms.o
-+$(obj)/mca_recovery.o: $(obj)/gate-syms.o $(obj)/xengate-syms.o
-+endif
++ /* The per-cpu crash note resources belong inside the
++ * hypervisor heap resource */
++ for (k = 0; k < nr_phys_cpus; k++)
++ request_resource(&xen_hypervisor_heap_res, phys_cpus + k);
+
-+GATECFLAGS_xengate-syms.o = -r
-+$(obj)/xengate-syms.o: $(obj)/xengate.lds $(obj)/xengate.o FORCE
-+ $(call if_changed,gate)
-+$(obj)/xengate-data.o: $(obj)/xengate.so
++ /* fill in efi_memmap_res with EFI memmap machine address range */
+
-+#
-+# .tmp_gate.o to calculate padding size for __kernel_syscall_via_epc
-+#
-+extra-$(CONFIG_XEN) += gate-skip.s .tmp_gate.o
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_EFI_MEMMAP;
+
-+ifeq ($(CONFIG_XEN), y)
-+AFLAGS_gate.o += -D__KERNEL_SYSCALL_VIA_EPC_PADDING
-+$(obj)/gate.o: $(obj)/gate-skip.s FORCE
-+endif
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ return -1;
+
-+$(obj)/.tmp_gate.o: $(src)/gate.S FORCE
-+ $(call if_changed_dep,as_o_S)
++ efi_memmap_res.start = range.start;
++ efi_memmap_res.end = range.start + range.size - 1;
+
-+quiet_cmd_gate_size = GATE_SIZE $@
-+ cmd_gate_size = $(NM) --extern-only --print-size $(obj)/xengate.o | \
-+ $(AWK) '/__kernel_syscall_via_epc/{printf "\t.skip 0x"$$2" - "}' > $@; \
-+ $(NM) --extern-only --print-size $(obj)/.tmp_gate.o | \
-+ $(AWK) '/__kernel_syscall_via_epc/{printf "0x"$$2"\n"}' >> $@
++ /* fill in boot_param_res with boot parameter machine address range */
+
-+$(obj)/gate-skip.s: $(obj)/xengate.o $(obj)/.tmp_gate.o FORCE
-+ $(call if_changed,gate_size)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/mca.c linux-2.6.18-xen.hg/arch/ia64/kernel/mca.c
---- linux-2.6.18/arch/ia64/kernel/mca.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/mca.c 2007-12-23 11:14:54.592530075 +0100
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_BOOT_PARAM;
++
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ return -1;
++
++ boot_param_res.start = range.start;
++ boot_param_res.end = range.start + range.size - 1;
++
++ return 0;
++}
++
++void machine_kexec_register_resources(struct resource *res)
++{
++ request_resource(res, &xen_hypervisor_heap_res);
++}
++#endif /* CONFIG_XEN */
+--- linux-2.6.18.8/arch/ia64/kernel/mca.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/mca.c 2008-05-19 00:33:04.360397489 +0300
@@ -79,6 +79,7 @@
#include <asm/system.h>
#include <asm/sal.h>
}
if (notify_die(DIE_MCA_MONARCH_LEAVE, "MCA", regs, (long)&nd, 0, recover)
== NOTIFY_STOP)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/pal.S linux-2.6.18-xen.hg/arch/ia64/kernel/pal.S
---- linux-2.6.18/arch/ia64/kernel/pal.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/pal.S 2007-12-23 11:14:54.595863584 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/pal.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/pal.S 2008-05-19 00:33:04.388399103 +0300
@@ -16,6 +16,7 @@
#include <asm/processor.h>
/*
* Make a PAL call using the stacked registers calling convention.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/patch.c linux-2.6.18-xen.hg/arch/ia64/kernel/patch.c
---- linux-2.6.18/arch/ia64/kernel/patch.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/patch.c 2007-12-23 11:14:54.599197091 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/patch.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/patch.c 2008-05-19 00:33:04.392399334 +0300
@@ -184,9 +184,37 @@
ia64_srlz_i();
}
# define START(name) ((unsigned long) __start_gate_##name##_patchlist)
# define END(name) ((unsigned long)__end_gate_##name##_patchlist)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/perfmon.c linux-2.6.18-xen.hg/arch/ia64/kernel/perfmon.c
---- linux-2.6.18/arch/ia64/kernel/perfmon.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/perfmon.c 2007-12-23 11:14:54.599197091 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/perfmon.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/perfmon.c 2008-05-19 00:33:04.396399564 +0300
@@ -52,6 +52,31 @@
#include <asm/delay.h>
DPRINT(("ctx_state=%d task [%d]\n", ctx->ctx_state, task ? task->pid : -1));
prev_state = ctx->ctx_state;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/relocate_kernel.S linux-2.6.18-xen.hg/arch/ia64/kernel/relocate_kernel.S
---- linux-2.6.18/arch/ia64/kernel/relocate_kernel.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/relocate_kernel.S 2007-12-23 11:14:54.605864096 +0100
-@@ -0,0 +1,380 @@
+--- linux-2.6.18.8/arch/ia64/kernel/relocate_kernel.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/relocate_kernel.S 2008-05-19 00:33:04.452402792 +0300
+@@ -0,0 +1,346 @@
+/*
+ * arch/ia64/kernel/relocate_kernel.S
+ *
+ */
+GLOBAL_ENTRY(relocate_new_kernel)
+ .prologue
-+#ifdef CONFIG_XEN
-+ alloc r31=ar.pfs,8,0,0,0
-+#else
+ alloc r31=ar.pfs,4,0,0,0
-+#endif
+ .body
+.reloc_entry:
+{
+ srlz.i
+}
+ ;;
-+#ifdef CONFIG_XEN
-+ dep r2=0,r2,60,4 //to physical address
-+#else
+ dep r2=0,r2,61,3 //to physical address
-+#endif
+ ;;
+ //first switch to physical mode
+ add r3=1f-.reloc_entry, r2
+ //physical mode code begin
+ mov b6=in1
+#ifdef CONFIG_XEN
-+ dep r28=0,in2,60,4 //to physical address
++ mov r28=in2 //already a physical address
+#else
+ dep r28=0,in2,61,3 //to physical address
-+#endif
+
+ // purge all TC entries
+#define O(member) IA64_CPUINFO_##member##_OFFSET
-+#ifdef CONFIG_XEN
-+ mov r2=in4 // load phys addr of cpu_info into r2
-+#else
+ GET_THIS_PADDR(r2, cpu_info) // load phys addr of cpu_info into r2
-+#endif
+ ;;
+ addl r17=O(PTCE_STRIDE),r2
+ addl r2=O(PTCE_BASE),r2
+ srlz.i
+ ;;
+ //purge TR entry for kernel text and data
-+#ifdef CONFIG_XEN
-+ mov r16=in5
-+#else
+ movl r16=KERNEL_START
-+#endif
+ mov r18=KERNEL_TR_PAGE_SHIFT<<2
+ ;;
+ ptr.i r16, r18
+ mov r16=IA64_KR(CURRENT_STACK)
+ ;;
+ shl r16=r16,IA64_GRANULE_SHIFT
-+#ifdef CONFIG_XEN
-+ mov r19=in6
-+#else
+ movl r19=PAGE_OFFSET
-+#endif
+ ;;
+ add r16=r19,r16
+ mov r18=IA64_GRANULE_SHIFT<<2
+ ;;
+ srlz.i
+ ;;
-+
-+#ifdef XEN
-+ /* XXX: Is this neccessary ??? */
-+ // purge TR entry for VHPT
-+ mov r16=in7
-+ ;;
-+ dep r16=0,r16,0,IA64_GRANULE_SHIFT
-+ mov r18=IA64_GRANULE_SHIFT<<2
-+ ;;
-+ ptr.d r16,r18
-+ ;;
-+ srlz.i
-+ ;;
-+#endif
++#endif /* ! CONFIG_XEN */
+
+ //copy segments
+ movl r16=PAGE_MASK
+END(ia64_dump_cpu_regs)
+
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/salinfo.c linux-2.6.18-xen.hg/arch/ia64/kernel/salinfo.c
---- linux-2.6.18/arch/ia64/kernel/salinfo.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/salinfo.c 2007-12-23 11:14:54.609197603 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/salinfo.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/salinfo.c 2008-05-19 00:33:04.452402792 +0300
@@ -375,6 +375,25 @@
data->open = 0;
return -ENOMEM;
vfree(data->log_buffer);
vfree(data->oemdata);
data->log_buffer = NULL;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/setup.c linux-2.6.18-xen.hg/arch/ia64/kernel/setup.c
---- linux-2.6.18/arch/ia64/kernel/setup.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/setup.c 2007-12-23 11:14:54.609197603 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/setup.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/setup.c 2008-05-19 00:33:04.456403023 +0300
@@ -43,6 +43,8 @@
#include <linux/initrd.h>
#include <linux/pm.h>
+ }
+ }
+ }
-+#ifdef CONFIG_XEN
-+ }
-+#endif
+ efi_memmap_res.start = ia64_boot_param->efi_memmap;
+ efi_memmap_res.end = efi_memmap_res.start +
+ ia64_boot_param->efi_memmap_size;
+ boot_param_res.start = kexec_virt_to_phys(ia64_boot_param);
+ boot_param_res.end = boot_param_res.start +
+ sizeof(*ia64_boot_param);
++#ifdef CONFIG_XEN
++ }
++#endif
+ }
+#endif
/* end of memory marker */
pm_idle = default_idle;
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/smp.c linux-2.6.18-xen.hg/arch/ia64/kernel/smp.c
---- linux-2.6.18/arch/ia64/kernel/smp.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/smp.c 2007-12-23 11:14:54.612531114 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/smp.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/smp.c 2008-05-19 00:33:04.460403253 +0300
@@ -30,6 +30,7 @@
#include <linux/delay.h>
#include <linux/efi.h>
call_data = &data;
mb(); /* ensure store to call_data precedes setting of IPI_CALL_FUNC */
send_IPI_allbutself(IPI_CALL_FUNC);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/time.c linux-2.6.18-xen.hg/arch/ia64/kernel/time.c
---- linux-2.6.18/arch/ia64/kernel/time.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/time.c 2007-12-23 11:14:54.612531114 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/time.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/time.c 2008-05-19 00:33:04.464403484 +0300
@@ -29,6 +29,14 @@
#include <asm/sections.h>
#include <asm/system.h>
/* Setup the CPU local timer tick */
ia64_cpu_local_tick();
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/vmlinux.lds.S linux-2.6.18-xen.hg/arch/ia64/kernel/vmlinux.lds.S
---- linux-2.6.18/arch/ia64/kernel/vmlinux.lds.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/vmlinux.lds.S 2007-12-23 11:14:54.619198137 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/vmlinux.lds.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/vmlinux.lds.S 2008-05-19 00:33:04.472403945 +0300
@@ -183,6 +183,12 @@
__start_gate_section = .;
*(.data.gate)
}
. = ALIGN(PAGE_SIZE); /* make sure the gate page doesn't expose kernel data */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/kernel/xengate-data.S linux-2.6.18-xen.hg/arch/ia64/kernel/xengate-data.S
---- linux-2.6.18/arch/ia64/kernel/xengate-data.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/kernel/xengate-data.S 2007-12-23 11:14:54.619198137 +0100
+--- linux-2.6.18.8/arch/ia64/kernel/xengate-data.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/kernel/xengate-data.S 2008-05-19 00:33:04.472403945 +0300
@@ -0,0 +1,3 @@
+ .section .data.gate.xen, "aw"
+
+ .incbin "arch/ia64/kernel/xengate.so"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/Makefile linux-2.6.18-xen.hg/arch/ia64/Makefile
---- linux-2.6.18/arch/ia64/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/Makefile 2007-12-23 11:14:54.542527446 +0100
-@@ -45,6 +45,12 @@
- endif
-
- CFLAGS += $(cflags-y)
-+
-+cppflags-$(CONFIG_XEN) += \
-+ -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION)
-+
-+CPPFLAGS += $(cppflags-y)
-+
- head-y := arch/ia64/kernel/head.o arch/ia64/kernel/init_task.o
-
- libs-y += arch/ia64/lib/
-@@ -54,7 +60,9 @@
- core-$(CONFIG_IA64_GENERIC) += arch/ia64/dig/
- core-$(CONFIG_IA64_HP_ZX1) += arch/ia64/dig/
- core-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/dig/
-+core-$(CONFIG_IA64_XEN) += arch/ia64/dig/
- core-$(CONFIG_IA64_SGI_SN2) += arch/ia64/sn/
-+core-$(CONFIG_XEN) += arch/ia64/xen/
-
- drivers-$(CONFIG_PCI) += arch/ia64/pci/
- drivers-$(CONFIG_IA64_HP_SIM) += arch/ia64/hp/sim/
-@@ -87,8 +95,8 @@
- boot: lib/lib.a vmlinux
- $(Q)$(MAKE) $(build)=$(boot) $@
-
--install: vmlinux.gz
-- sh $(srctree)/arch/ia64/install.sh $(KERNELRELEASE) $< System.map "$(INSTALL_PATH)"
-+install:
-+ -yes | sh $(srctree)/arch/ia64/install.sh $(KERNELRELEASE) vmlinux.gz System.map "$(INSTALL_PATH)"
-
- define archhelp
- echo '* compressed - Build compressed kernel image'
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/mm/contig.c linux-2.6.18-xen.hg/arch/ia64/mm/contig.c
---- linux-2.6.18/arch/ia64/mm/contig.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/mm/contig.c 2007-12-23 11:14:54.632532163 +0100
+--- linux-2.6.18.8/arch/ia64/mm/contig.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/mm/contig.c 2008-05-19 00:33:04.492405098 +0300
@@ -18,6 +18,9 @@
#include <linux/efi.h>
#include <linux/mm.h>
#include <linux/swap.h>
-+#ifdef CONFIG_XEN
++#if defined(CONFIG_XEN) && defined(CONFIG_KEXEC)
+#include <linux/kexec.h>
+#endif
}
#ifdef CONFIG_SMP
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/mm/init.c linux-2.6.18-xen.hg/arch/ia64/mm/init.c
---- linux-2.6.18/arch/ia64/mm/init.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/mm/init.c 2007-12-23 11:14:54.635865674 +0100
+--- linux-2.6.18.8/arch/ia64/mm/init.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/mm/init.c 2008-05-19 00:33:04.496405328 +0300
@@ -303,16 +303,34 @@
setup_gate (void)
{
put_kernel_page(page, GATE_ADDR + PAGE_SIZE, PAGE_GATE);
#else
put_kernel_page(page, GATE_ADDR + PERCPU_PAGE_SIZE, PAGE_GATE);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/mm/ioremap.c linux-2.6.18-xen.hg/arch/ia64/mm/ioremap.c
---- linux-2.6.18/arch/ia64/mm/ioremap.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/mm/ioremap.c 2007-12-23 11:14:54.635865674 +0100
+--- linux-2.6.18.8/arch/ia64/mm/ioremap.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/mm/ioremap.c 2008-05-19 00:33:04.500405559 +0300
@@ -16,6 +16,9 @@
static inline void __iomem *
__ioremap (unsigned long offset, unsigned long size)
return (void __iomem *) (__IA64_UNCACHED_OFFSET | offset);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/oprofile/init.c linux-2.6.18-xen.hg/arch/ia64/oprofile/init.c
---- linux-2.6.18/arch/ia64/oprofile/init.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/oprofile/init.c 2007-12-23 11:14:54.639199186 +0100
+--- linux-2.6.18.8/arch/ia64/oprofile/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/oprofile/Makefile 2008-05-19 00:33:04.520406712 +0300
+@@ -8,3 +8,7 @@
+
+ oprofile-y := $(DRIVER_OBJS) init.o backtrace.o
+ oprofile-$(CONFIG_PERFMON) += perfmon.o
++ifeq ($(CONFIG_XEN), y)
++oprofile-$(CONFIG_PERFMON) += xenoprof.o \
++ ../../../drivers/xen/xenoprof/xenoprofile.o
++endif
+--- linux-2.6.18.8/arch/ia64/oprofile/init.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/oprofile/init.c 2008-05-19 00:33:04.520406712 +0300
@@ -11,6 +11,8 @@
#include <linux/oprofile.h>
#include <linux/init.h>
#ifdef CONFIG_PERFMON
perfmon_exit();
#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/oprofile/Makefile linux-2.6.18-xen.hg/arch/ia64/oprofile/Makefile
---- linux-2.6.18/arch/ia64/oprofile/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/oprofile/Makefile 2007-12-23 11:14:54.635865674 +0100
-@@ -8,3 +8,7 @@
-
- oprofile-y := $(DRIVER_OBJS) init.o backtrace.o
- oprofile-$(CONFIG_PERFMON) += perfmon.o
-+ifeq ($(CONFIG_XEN), y)
-+oprofile-$(CONFIG_PERFMON) += xenoprof.o \
-+ ../../../drivers/xen/xenoprof/xenoprofile.o
-+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/oprofile/oprofile_perfmon.h linux-2.6.18-xen.hg/arch/ia64/oprofile/oprofile_perfmon.h
---- linux-2.6.18/arch/ia64/oprofile/oprofile_perfmon.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/oprofile/oprofile_perfmon.h 2007-12-23 11:14:54.639199186 +0100
+--- linux-2.6.18.8/arch/ia64/oprofile/oprofile_perfmon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/oprofile/oprofile_perfmon.h 2008-05-19 00:33:04.520406712 +0300
@@ -0,0 +1,30 @@
+#ifndef OPROFILE_PERFMON_H
+#define OPROFILE_PERFMON_H
+#endif /* CONFIG_XEN */
+
+#endif /* OPROFILE_PERFMON_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/oprofile/perfmon.c linux-2.6.18-xen.hg/arch/ia64/oprofile/perfmon.c
---- linux-2.6.18/arch/ia64/oprofile/perfmon.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/oprofile/perfmon.c 2007-12-23 11:14:54.639199186 +0100
+--- linux-2.6.18.8/arch/ia64/oprofile/perfmon.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/oprofile/perfmon.c 2008-05-19 00:33:04.520406712 +0300
@@ -13,6 +13,7 @@
#include <asm/perfmon.h>
#include <asm/ptrace.h>
- pfm_unregister_buffer_fmt(oprofile_fmt.fmt_uuid);
+ __perfmon_exit();
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/oprofile/xenoprof.c linux-2.6.18-xen.hg/arch/ia64/oprofile/xenoprof.c
---- linux-2.6.18/arch/ia64/oprofile/xenoprof.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/oprofile/xenoprof.c 2007-12-23 11:14:54.639199186 +0100
+--- linux-2.6.18.8/arch/ia64/oprofile/xenoprof.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/oprofile/xenoprof.c 2008-05-19 00:33:04.524406942 +0300
@@ -0,0 +1,142 @@
+/******************************************************************************
+ * xenoprof ia64 specific part
+}
+
+/* XXX move them to an appropriate header file. */
-+struct resource* xen_ia64_allocate_resource(unsigned long size);
-+void xen_ia64_release_resource(struct resource* res);
-+void xen_ia64_unmap_resource(struct resource* res);
++struct resource* xen_ia64_allocate_resource(unsigned long size);
++void xen_ia64_release_resource(struct resource *res);
++void xen_ia64_unmap_resource(struct resource *res);
+
+struct resource*
+xenoprof_ia64_allocate_resource(int32_t max_samples)
+ return xen_ia64_allocate_resource(bufsize);
+}
+
-+void xenoprof_arch_unmap_shared_buffer(struct xenoprof_shared_buffer* sbuf)
++void xenoprof_arch_unmap_shared_buffer(struct xenoprof_shared_buffer *sbuf)
+{
+ if (sbuf->buffer) {
+ xen_ia64_unmap_resource(sbuf->arch.res);
+ }
+}
+
-+int xenoprof_arch_map_shared_buffer(struct xenoprof_get_buffer* get_buffer,
-+ struct xenoprof_shared_buffer* sbuf)
++int xenoprof_arch_map_shared_buffer(struct xenoprof_get_buffer *get_buffer,
++ struct xenoprof_shared_buffer *sbuf)
+{
+ int ret;
-+ struct resource* res;
++ struct resource *res;
+
+ sbuf->buffer = NULL;
+ sbuf->arch.res = NULL;
+ return ret;
+}
+
-+int xenoprof_arch_set_passive(struct xenoprof_passive* pdomain,
-+ struct xenoprof_shared_buffer* sbuf)
++int xenoprof_arch_set_passive(struct xenoprof_passive *pdomain,
++ struct xenoprof_shared_buffer *sbuf)
+{
+ int ret;
-+ struct resource* res;
++ struct resource *res;
+
+ sbuf->buffer = NULL;
+ sbuf->arch.res = NULL;
+
+ return ret;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/pci/pci.c linux-2.6.18-xen.hg/arch/ia64/pci/pci.c
---- linux-2.6.18/arch/ia64/pci/pci.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/pci/pci.c 2007-12-23 11:14:54.639199186 +0100
+--- linux-2.6.18.8/arch/ia64/pci/pci.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/pci/pci.c 2008-05-19 00:33:04.524406942 +0300
@@ -30,6 +30,15 @@
#include <asm/irq.h>
#include <asm/hw_irq.h>
+}
+EXPORT_SYMBOL(xen_pcibios_setup_root_windows);
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/sn/kernel/setup.c linux-2.6.18-xen.hg/arch/ia64/sn/kernel/setup.c
---- linux-2.6.18/arch/ia64/sn/kernel/setup.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/sn/kernel/setup.c 2007-12-23 11:14:54.655866720 +0100
+--- linux-2.6.18.8/arch/ia64/sn/kernel/setup.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/sn/kernel/setup.c 2008-05-19 00:33:04.968432536 +0300
@@ -763,5 +763,13 @@
return 0;
return test_bit(id, sn_prom_features);
+}
EXPORT_SYMBOL(sn_prom_feature_available);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/sn/pci/pcibr/pcibr_provider.c linux-2.6.18-xen.hg/arch/ia64/sn/pci/pcibr/pcibr_provider.c
---- linux-2.6.18/arch/ia64/sn/pci/pcibr/pcibr_provider.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/pcibr/pcibr_provider.c 2007-12-23 11:14:54.669200766 +0100
+--- linux-2.6.18.8/arch/ia64/sn/pci/pcibr/pcibr_provider.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/pcibr/pcibr_provider.c 2008-05-19 00:33:05.004434611 +0300
@@ -15,6 +15,7 @@
#include <asm/sn/pcibus_provider_defs.h>
#include <asm/sn/pcidev.h>
spin_lock_init(&soft->pbi_lock);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/sn/pci/tioca_provider.c linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioca_provider.c
---- linux-2.6.18/arch/ia64/sn/pci/tioca_provider.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioca_provider.c 2007-12-23 11:14:54.669200766 +0100
+--- linux-2.6.18.8/arch/ia64/sn/pci/tioca_provider.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioca_provider.c 2008-05-19 00:33:05.004434611 +0300
@@ -611,7 +611,9 @@
return NULL;
/* init kernel-private area */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/sn/pci/tioce_provider.c linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioce_provider.c
---- linux-2.6.18/arch/ia64/sn/pci/tioce_provider.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioce_provider.c 2007-12-23 11:14:54.669200766 +0100
+--- linux-2.6.18.8/arch/ia64/sn/pci/tioce_provider.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/sn/pci/tioce_provider.c 2008-05-19 00:33:05.008434842 +0300
@@ -1006,7 +1006,9 @@
return NULL;
tioce_kern = tioce_kern_init(tioce_common);
if (tioce_kern == NULL) {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/hypercall.S linux-2.6.18-xen.hg/arch/ia64/xen/hypercall.S
---- linux-2.6.18/arch/ia64/xen/hypercall.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/hypercall.S 2007-12-23 11:14:54.672534273 +0100
+--- linux-2.6.18.8/arch/ia64/xen/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/Makefile 2008-05-19 00:33:05.008434842 +0300
+@@ -0,0 +1,9 @@
++#
++# Makefile for Xen components
++#
++
++obj-y := hypercall.o xenivt.o xenentry.o xensetup.o xenpal.o \
++ hypervisor.o util.o xencomm.o xcom_hcall.o \
++ xcom_privcmd.o xcom_asm.o xen_dma.o
++
++obj-$(CONFIG_IA64_GENERIC) += machvec.o
+--- linux-2.6.18.8/arch/ia64/xen/hypercall.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/hypercall.S 2008-05-19 00:33:05.008434842 +0300
@@ -0,0 +1,141 @@
+/*
+ * Support routines for Xen hypercalls
+GLOBAL_ENTRY(xen_get_psr)
+ XEN_HYPER_GET_PSR
+ br.ret.sptk.many rp
-+ ;;
++ ;;
+END(xen_get_psr)
+
+GLOBAL_ENTRY(xen_get_ivr)
+#endif /* ASM_SUPPORTED */
+
+GLOBAL_ENTRY(xen_send_ipi)
-+ mov r14=r32
-+ mov r15=r33
-+ mov r2=0x400
-+ break 0x1000
-+ ;;
-+ br.ret.sptk.many rp
-+ ;;
++ mov r14=r32
++ mov r15=r33
++ mov r2=0x400
++ break 0x1000
++ ;;
++ br.ret.sptk.many rp
++ ;;
+END(xen_send_ipi)
+
+GLOBAL_ENTRY(__hypercall)
+ br.ret.sptk.many b0
+ ;;
+END(__hypercall)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/hypervisor.c linux-2.6.18-xen.hg/arch/ia64/xen/hypervisor.c
---- linux-2.6.18/arch/ia64/xen/hypervisor.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/hypervisor.c 2007-12-23 11:14:54.672534273 +0100
-@@ -0,0 +1,1505 @@
+--- linux-2.6.18.8/arch/ia64/xen/hypervisor.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/hypervisor.c 2008-05-19 00:33:05.012435072 +0300
+@@ -0,0 +1,1526 @@
+/******************************************************************************
+ * include/asm-ia64/shadow.h
+ *
+ *
+ */
+
-+//#include <linux/kernel.h>
+#include <linux/spinlock.h>
+#include <linux/bootmem.h>
+#include <linux/module.h>
+#include <xen/xencons.h>
+#include <xen/balloon.h>
+
-+shared_info_t *HYPERVISOR_shared_info __read_mostly = (shared_info_t *)XSI_BASE;
++shared_info_t *HYPERVISOR_shared_info __read_mostly =
++ (shared_info_t *)XSI_BASE;
+EXPORT_SYMBOL(HYPERVISOR_shared_info);
+
+start_info_t *xen_start_info;
+
+ if (ia64_platform_is("xen"))
+ dig_setup(cmdline_p);
-+
++
+ if (!is_running_on_xen() || !is_initial_xendomain())
+ return;
+
+ xen_smp_intr_init();
+}
+
-+//XXX same as i386, x86_64 contiguous_bitmap_set(), contiguous_bitmap_clear()
-+// move those to lib/contiguous_bitmap?
-+//XXX discontigmem/sparsemem
++/*
++ *XXX same as i386, x86_64 contiguous_bitmap_set(), contiguous_bitmap_clear()
++ * move those to lib/contiguous_bitmap?
++ *XXX discontigmem/sparsemem
++ */
+
+/*
+ * Bitmap is indexed by page number. If bit is set, the page is part of a
+ pte_t *pte;
+
+ bitmap_start = (unsigned long)contiguous_bitmap +
-+ ((__pa(start) >> PAGE_SHIFT) >> 3);
++ ((__pa(start) >> PAGE_SHIFT) >> 3);
+ bitmap_end = (unsigned long)contiguous_bitmap +
-+ (((__pa(end) >> PAGE_SHIFT) + 2 * BITS_PER_LONG) >> 3);
++ (((__pa(end) >> PAGE_SHIFT) + 2 * BITS_PER_LONG) >> 3);
+
+ start_page = bitmap_start & PAGE_MASK;
+ end_page = PAGE_ALIGN(bitmap_end);
+ node = paddr_to_nid(__pa(start));
+
+ bitmap = alloc_bootmem_pages_node(NODE_DATA(node),
-+ end_page - start_page);
++ end_page - start_page);
+ BUG_ON(!bitmap);
+ memset(bitmap, 0, end_page - start_page);
+
+ pgd = pgd_offset_k(address);
+ if (pgd_none(*pgd))
+ pgd_populate(&init_mm, pgd,
-+ alloc_bootmem_pages_node(NODE_DATA(node),
-+ PAGE_SIZE));
++ alloc_bootmem_pages_node(NODE_DATA(node),
++ PAGE_SIZE));
+ pud = pud_offset(pgd, address);
+
+ if (pud_none(*pud))
+ pud_populate(&init_mm, pud,
-+ alloc_bootmem_pages_node(NODE_DATA(node),
-+ PAGE_SIZE));
++ alloc_bootmem_pages_node(NODE_DATA(node),
++ PAGE_SIZE));
+ pmd = pmd_offset(pud, address);
+
+ if (pmd_none(*pmd))
+ pmd_populate_kernel(&init_mm, pmd,
-+ alloc_bootmem_pages_node
-+ (NODE_DATA(node), PAGE_SIZE));
++ alloc_bootmem_pages_node
++ (NODE_DATA(node), PAGE_SIZE));
+ pte = pte_offset_kernel(pmd, address);
+
+ if (pte_none(*pte))
+ set_pte(pte,
-+ pfn_pte(__pa(bitmap + (address - start_page))
-+ >> PAGE_SHIFT, PAGE_KERNEL));
++ pfn_pte(__pa(bitmap + (address - start_page))
++ >> PAGE_SHIFT, PAGE_KERNEL));
+ }
+ return 0;
+}
+ }
+}
+
-+// __xen_create_contiguous_region(), __xen_destroy_contiguous_region()
-+// are based on i386 xen_create_contiguous_region(),
-+// xen_destroy_contiguous_region()
++/*
++ * __xen_create_contiguous_region(), __xen_destroy_contiguous_region()
++ * are based on i386 xen_create_contiguous_region(),
++ * xen_destroy_contiguous_region()
++ */
+
+/* Protected by balloon_lock. */
+#define MAX_CONTIG_ORDER 7
+ balloon_lock(flags);
+
+ /* Get a new contiguous memory extent. */
-+ for (i = 0; i < num_gpfn; i++) {
++ for (i = 0; i < num_gpfn; i++)
+ in_frames[i] = start_gpfn + i;
-+ }
+ out_frame = start_gpfn;
+ error = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
+ success = (exchange.nr_exchanged == num_gpfn);
+ .domid = DOMID_SELF
+ },
+ .nr_exchanged = 0
-+ };
++ };
+
+
+ if (!test_bit(start_gpfn, contiguous_bitmap))
+
+ contiguous_bitmap_clear(start_gpfn, num_gpfn);
+
-+ /* Do the exchange for non-contiguous MFNs. */
++ /* Do the exchange for non-contiguous MFNs. */
+ in_frame = start_gpfn;
-+ for (i = 0; i < num_gpfn; i++) {
++ for (i = 0; i < num_gpfn; i++)
+ out_frames[i] = start_gpfn + i;
-+ }
+ error = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
+ success = (exchange.nr_exchanged == 1);
+ BUG_ON(!success && ((exchange.nr_exchanged != 0) || (error == 0)));
+ BUG_ON(success && (error != 0));
+ if (unlikely(error == -ENOSYS)) {
-+ /* Compatibility when XENMEM_exchange is unsupported. */
++ /* Compatibility when XENMEM_exchange is unsupported. */
+ error = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
+ &exchange.in);
+ BUG_ON(error != 1);
+ order, address_bits);
+}
+
-+
-+///////////////////////////////////////////////////////////////////////////
-+// grant table hack
-+// cmd: GNTTABOP_xxx
-+
++/****************************************************************************
++ * grant table hack
++ * cmd: GNTTABOP_xxx
++ */
+#include <linux/mm.h>
+#include <xen/interface/xen.h>
+#include <xen/gnttab.h>
+
+ if (flags & GNTMAP_host_map) {
+ if (flags & GNTMAP_application_map) {
-+ xprintd("GNTMAP_application_map is not supported yet: flags 0x%x\n", flags);
++ xprintd("GNTMAP_application_map is not supported yet:"
++ " flags 0x%x\n", flags);
+ BUG();
+ }
+ if (flags & GNTMAP_contains_pte) {
-+ xprintd("GNTMAP_contains_pte is not supported yet flags 0x%x\n", flags);
++ xprintd("GNTMAP_contains_pte is not supported yet"
++ " flags 0x%x\n", flags);
+ BUG();
+ }
+ } else if (flags & GNTMAP_device_map) {
-+ xprintd("GNTMAP_device_map is not supported yet 0x%x\n", flags);
-+ BUG();//XXX not yet. actually this flag is not used.
++ xprintd("GNTMAP_device_map is not supported yet 0x%x\n",
++ flags);
++ BUG(); /* XXX not yet. actually this flag is not used. */
+ } else {
+ BUG();
+ }
+}
+EXPORT_SYMBOL(HYPERVISOR_grant_table_op);
+
-+///////////////////////////////////////////////////////////////////////////
-+// foreign mapping
++/**************************************************************************
++ * foreign mapping
++ */
+#include <linux/efi.h>
-+#include <asm/meminit.h> // for IA64_GRANULE_SIZE, GRANULEROUND{UP,DOWN}()
++#include <asm/meminit.h> /* for IA64_GRANULE_SIZE, GRANULEROUND{UP,DOWN}() */
+
+static unsigned long privcmd_resource_min = 0;
-+// Xen/ia64 currently can handle pseudo physical address bits up to
-+// (PAGE_SHIFT * 3)
-+static unsigned long privcmd_resource_max = GRANULEROUNDDOWN((1UL << (PAGE_SHIFT * 3)) - 1);
++/* Xen/ia64 currently can handle pseudo physical address bits up to
++ * (PAGE_SHIFT * 3) */
++static unsigned long privcmd_resource_max =
++ GRANULEROUNDDOWN((1UL << (PAGE_SHIFT * 3)) - 1);
+static unsigned long privcmd_resource_align = IA64_GRANULE_SIZE;
+
+static unsigned long
+ efi_map_end = efi_map_start + ia64_boot_param->efi_memmap_size;
+ efi_desc_size = ia64_boot_param->efi_memdesc_size;
+
-+ // at first check the used highest address
++ /* at first check the used highest address */
+ for (p = efi_map_start; p < efi_map_end; p += efi_desc_size) {
-+ // nothing
++ /* nothing */;
+ }
+ md = p - efi_desc_size;
+ privcmd_resource_min = GRANULEROUNDUP(md_end_addr(md));
+ if (xen_ia64_privcmd_check_size(privcmd_resource_min,
-+ privcmd_resource_max)) {
++ privcmd_resource_max))
+ goto out;
-+ }
+
-+ // the used highest address is too large. try to find the largest gap.
++ /* the used highest address is too large.
++ * try to find the largest gap. */
+ tmp_min = privcmd_resource_max;
+ tmp_max = 0;
+ gap_size = 0;
+
+ md = p;
+ end = md_end_addr(md);
-+ if (end > privcmd_resource_max) {
++ if (end > privcmd_resource_max)
+ break;
-+ }
+ if (end < prev_end) {
-+ // work around.
-+ // Xen may pass incompletely sorted memory
-+ // descriptors like
-+ // [x, x + length]
-+ // [x, x]
-+ // this order should be reversed.
++ /* work around.
++ * Xen may pass incompletely sorted memory
++ * descriptors like
++ * [x, x + length]
++ * [x, x]
++ * this order should be reversed. */
+ continue;
+ }
+ next = p + efi_desc_size;
+ next_start = next->phys_addr;
-+ if (next_start > privcmd_resource_max) {
++ if (next_start > privcmd_resource_max)
+ next_start = privcmd_resource_max;
-+ }
+ if (end < next_start && gap_size < (next_start - end)) {
+ tmp_min = end;
+ tmp_max = next_start;
+ privcmd_resource_max = tmp_max;
+ if (!xen_ia64_privcmd_check_size(privcmd_resource_min,
+ privcmd_resource_max)) {
-+ // Any large enough gap isn't found.
-+ // go ahead anyway with the warning hoping that large region
-+ // won't be requested.
-+ printk(KERN_WARNING "xen privcmd: large enough region for privcmd mmap is not found.\n");
++ /* Any large enough gap isn't found.
++ * go ahead anyway with the warning hoping that large region
++ * won't be requested. */
++ printk(KERN_WARNING "xen privcmd: "
++ "large enough region for privcmd mmap is not found.\n");
+ }
+
+out:
-+ printk(KERN_INFO "xen privcmd uses pseudo physical addr range [0x%lx, 0x%lx] (%ldMB)\n",
++ printk(KERN_INFO "xen privcmd uses pseudo physical addr range "
++ "[0x%lx, 0x%lx] (%ldMB)\n",
+ privcmd_resource_min, privcmd_resource_max,
+ (privcmd_resource_max - privcmd_resource_min) >> 20);
+ BUG_ON(privcmd_resource_min >= privcmd_resource_max);
+
-+ // XXX this should be somewhere appropriate
++ /* XXX this should be somewhere appropriate */
+ (void)p2m_expose_init();
+
+ return 0;
+
+struct xen_ia64_privcmd_range {
+ atomic_t ref_count;
-+ unsigned long pgoff; // in PAGE_SIZE
-+ struct resource* res;
++ unsigned long pgoff; /* in PAGE_SIZE */
++ struct resource *res;
+
-+ // for foreign domain p2m mapping
-+ void* private;
-+ void (*callback)(struct xen_ia64_privcmd_range* range, void* arg);
++ /* for foreign domain p2m mapping */
++ void *private;
++ void (*callback)(struct xen_ia64_privcmd_range *range, void *arg);
+
+ unsigned long num_entries;
+ struct xen_ia64_privcmd_entry entries[0];
+
+struct xen_ia64_privcmd_vma {
+ int is_privcmd_mmapped;
-+ struct xen_ia64_privcmd_range* range;
++ struct xen_ia64_privcmd_range *range;
+
+ unsigned long num_entries;
-+ struct xen_ia64_privcmd_entry* entries;
++ struct xen_ia64_privcmd_entry *entries;
+};
+
+static void
-+xen_ia64_privcmd_init_entry(struct xen_ia64_privcmd_entry* entry)
++xen_ia64_privcmd_init_entry(struct xen_ia64_privcmd_entry *entry)
+{
+ atomic_set(&entry->map_count, 0);
+ entry->gpfn = INVALID_GPFN;
+}
+
+static int
-+xen_ia64_privcmd_entry_mmap(struct vm_area_struct* vma,
++xen_ia64_privcmd_entry_mmap(struct vm_area_struct *vma,
+ unsigned long addr,
-+ struct xen_ia64_privcmd_range* privcmd_range,
++ struct xen_ia64_privcmd_range *privcmd_range,
+ int i,
+ unsigned long gmfn,
+ pgprot_t prot,
+ domid_t domid)
+{
+ int error = 0;
-+ struct xen_ia64_privcmd_entry* entry = &privcmd_range->entries[i];
++ struct xen_ia64_privcmd_entry *entry = &privcmd_range->entries[i];
+ unsigned long gpfn;
+ unsigned long flags;
+
+ gpfn = (privcmd_range->res->start >> PAGE_SHIFT) + i;
+
+ flags = ASSIGN_writable;
-+ if (pgprot_val(prot) == PROT_READ) {
++ if (pgprot_val(prot) == PROT_READ)
+ flags = ASSIGN_readonly;
-+ }
+ error = HYPERVISOR_add_physmap_with_gmfn(gpfn, gmfn, flags, domid);
-+ if (error != 0) {
++ if (error != 0)
+ goto out;
-+ }
+
+ prot = vma->vm_page_prot;
+ error = remap_pfn_range(vma, addr, gpfn, 1 << PAGE_SHIFT, prot);
++ /*
++ * VM_PFNMAP is set in remap_pfn_range().
++ * Reset the flag to avoid BUG_ON() in do_no_page().
++ */
++ vma->vm_flags &= ~VM_PFNMAP;
++
+ if (error != 0) {
+ error = HYPERVISOR_zap_physmap(gpfn, 0);
-+ if (error) {
-+ BUG();//XXX
-+ }
++ if (error)
++ BUG(); /* XXX */
+ } else {
+ atomic_inc(&entry->map_count);
+ entry->gpfn = gpfn;
+}
+
+static void
-+xen_ia64_privcmd_entry_munmap(struct xen_ia64_privcmd_range* privcmd_range,
++xen_ia64_privcmd_entry_munmap(struct xen_ia64_privcmd_range *privcmd_range,
+ int i)
+{
-+ struct xen_ia64_privcmd_entry* entry = &privcmd_range->entries[i];
++ struct xen_ia64_privcmd_entry *entry = &privcmd_range->entries[i];
+ unsigned long gpfn = entry->gpfn;
-+ //gpfn = (privcmd_range->res->start >> PAGE_SHIFT) +
-+ // (vma->vm_pgoff - privcmd_range->pgoff);
++ /* gpfn = (privcmd_range->res->start >> PAGE_SHIFT) +
++ (vma->vm_pgoff - privcmd_range->pgoff); */
+ int error;
+
+ error = HYPERVISOR_zap_physmap(gpfn, 0);
-+ if (error) {
-+ BUG();//XXX
-+ }
++ if (error)
++ BUG(); /* XXX */
+ entry->gpfn = INVALID_GPFN;
+}
+
+static void
-+xen_ia64_privcmd_entry_open(struct xen_ia64_privcmd_range* privcmd_range,
++xen_ia64_privcmd_entry_open(struct xen_ia64_privcmd_range *privcmd_range,
+ int i)
+{
-+ struct xen_ia64_privcmd_entry* entry = &privcmd_range->entries[i];
-+ if (entry->gpfn != INVALID_GPFN) {
++ struct xen_ia64_privcmd_entry *entry = &privcmd_range->entries[i];
++ if (entry->gpfn != INVALID_GPFN)
+ atomic_inc(&entry->map_count);
-+ } else {
++ else
+ BUG_ON(atomic_read(&entry->map_count) != 0);
-+ }
+}
+
+static void
-+xen_ia64_privcmd_entry_close(struct xen_ia64_privcmd_range* privcmd_range,
++xen_ia64_privcmd_entry_close(struct xen_ia64_privcmd_range *privcmd_range,
+ int i)
+{
-+ struct xen_ia64_privcmd_entry* entry = &privcmd_range->entries[i];
++ struct xen_ia64_privcmd_entry *entry = &privcmd_range->entries[i];
+ if (entry->gpfn != INVALID_GPFN &&
-+ atomic_dec_and_test(&entry->map_count)) {
++ atomic_dec_and_test(&entry->map_count))
+ xen_ia64_privcmd_entry_munmap(privcmd_range, i);
-+ }
+}
+
-+static void xen_ia64_privcmd_vma_open(struct vm_area_struct* vma);
-+static void xen_ia64_privcmd_vma_close(struct vm_area_struct* vma);
++static void xen_ia64_privcmd_vma_open(struct vm_area_struct *vma);
++static void xen_ia64_privcmd_vma_close(struct vm_area_struct *vma);
++
++static struct page *
++xen_ia64_privcmd_vma_nopage(struct vm_area_struct *vma,
++ unsigned long address,
++ int *type)
++{
++ return NOPAGE_SIGBUS;
++}
+
+struct vm_operations_struct xen_ia64_privcmd_vm_ops = {
-+ .open = &xen_ia64_privcmd_vma_open,
-+ .close = &xen_ia64_privcmd_vma_close,
++ .open = xen_ia64_privcmd_vma_open,
++ .close = xen_ia64_privcmd_vma_close,
++ .nopage = xen_ia64_privcmd_vma_nopage
+};
+
+static void
-+__xen_ia64_privcmd_vma_open(struct vm_area_struct* vma,
-+ struct xen_ia64_privcmd_vma* privcmd_vma,
-+ struct xen_ia64_privcmd_range* privcmd_range)
++__xen_ia64_privcmd_vma_open(struct vm_area_struct *vma,
++ struct xen_ia64_privcmd_vma *privcmd_vma,
++ struct xen_ia64_privcmd_range *privcmd_range)
+{
+ unsigned long entry_offset = vma->vm_pgoff - privcmd_range->pgoff;
-+ unsigned long num_entries = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
++ unsigned long num_entries =
++ (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
+ unsigned long i;
+
+ BUG_ON(entry_offset < 0);
+ privcmd_vma->num_entries = num_entries;
+ privcmd_vma->entries = &privcmd_range->entries[entry_offset];
+ vma->vm_private_data = privcmd_vma;
-+ for (i = 0; i < privcmd_vma->num_entries; i++) {
++ for (i = 0; i < privcmd_vma->num_entries; i++)
+ xen_ia64_privcmd_entry_open(privcmd_range, entry_offset + i);
-+ }
+
+ vma->vm_private_data = privcmd_vma;
+ vma->vm_ops = &xen_ia64_privcmd_vm_ops;
+}
+
+static void
-+xen_ia64_privcmd_vma_open(struct vm_area_struct* vma)
++xen_ia64_privcmd_vma_open(struct vm_area_struct *vma)
+{
-+ struct xen_ia64_privcmd_vma* old_privcmd_vma = (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
-+ struct xen_ia64_privcmd_vma* privcmd_vma = (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
-+ struct xen_ia64_privcmd_range* privcmd_range = privcmd_vma->range;
++ struct xen_ia64_privcmd_vma *old_privcmd_vma =
++ (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
++ struct xen_ia64_privcmd_vma *privcmd_vma =
++ (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
++ struct xen_ia64_privcmd_range *privcmd_range = privcmd_vma->range;
+
+ atomic_inc(&privcmd_range->ref_count);
-+ // vm_op->open() can't fail.
++ /* vm_op->open() can't fail. */
+ privcmd_vma = kmalloc(sizeof(*privcmd_vma), GFP_KERNEL | __GFP_NOFAIL);
-+ // copy original value if necessary
++ /* copy original value if necessary */
+ privcmd_vma->is_privcmd_mmapped = old_privcmd_vma->is_privcmd_mmapped;
+
+ __xen_ia64_privcmd_vma_open(vma, privcmd_vma, privcmd_range);
+}
+
+static void
-+xen_ia64_privcmd_vma_close(struct vm_area_struct* vma)
++xen_ia64_privcmd_vma_close(struct vm_area_struct *vma)
+{
-+ struct xen_ia64_privcmd_vma* privcmd_vma =
++ struct xen_ia64_privcmd_vma *privcmd_vma =
+ (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
-+ struct xen_ia64_privcmd_range* privcmd_range = privcmd_vma->range;
++ struct xen_ia64_privcmd_range *privcmd_range = privcmd_vma->range;
+ unsigned long entry_offset = vma->vm_pgoff - privcmd_range->pgoff;
+ unsigned long i;
+
+ if (atomic_dec_and_test(&privcmd_range->ref_count)) {
+#if 1
+ for (i = 0; i < privcmd_range->num_entries; i++) {
-+ struct xen_ia64_privcmd_entry* entry =
++ struct xen_ia64_privcmd_entry *entry =
+ &privcmd_range->entries[i];
+ BUG_ON(atomic_read(&entry->map_count) != 0);
+ BUG_ON(entry->gpfn != INVALID_GPFN);
+int
+privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
+{
-+ struct xen_ia64_privcmd_vma* privcmd_vma =
++ struct xen_ia64_privcmd_vma *privcmd_vma =
+ (struct xen_ia64_privcmd_vma *)vma->vm_private_data;
+ return (xchg(&privcmd_vma->is_privcmd_mmapped, 1) == 0);
+}
+ int error;
+ unsigned long size = vma->vm_end - vma->vm_start;
+ unsigned long num_entries = size >> PAGE_SHIFT;
-+ struct xen_ia64_privcmd_range* privcmd_range = NULL;
-+ struct xen_ia64_privcmd_vma* privcmd_vma = NULL;
-+ struct resource* res = NULL;
++ struct xen_ia64_privcmd_range *privcmd_range = NULL;
++ struct xen_ia64_privcmd_vma *privcmd_vma = NULL;
++ struct resource *res = NULL;
+ unsigned long i;
+ BUG_ON(!is_running_on_xen());
+
+ privcmd_range =
+ vmalloc(sizeof(*privcmd_range) +
+ sizeof(privcmd_range->entries[0]) * num_entries);
-+ if (privcmd_range == NULL) {
++ if (privcmd_range == NULL)
+ goto out_enomem0;
-+ }
+ privcmd_vma = kmalloc(sizeof(*privcmd_vma), GFP_KERNEL);
-+ if (privcmd_vma == NULL) {
++ if (privcmd_vma == NULL)
+ goto out_enomem1;
-+ }
+ privcmd_vma->is_privcmd_mmapped = 0;
+
+ res = kzalloc(sizeof(*res), GFP_KERNEL);
-+ if (res == NULL) {
++ if (res == NULL)
+ goto out_enomem1;
-+ }
+ res->name = "Xen privcmd mmap";
+ error = allocate_resource(&iomem_resource, res, size,
+ privcmd_resource_min, privcmd_resource_max,
+ privcmd_resource_align, NULL, NULL);
-+ if (error) {
++ if (error)
+ goto out_enomem1;
-+ }
+ privcmd_range->res = res;
+
+ /* DONTCOPY is essential for Xen as copy_page_range is broken. */
-+ vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY | VM_PFNMAP;
++ vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY;
+
+ atomic_set(&privcmd_range->ref_count, 1);
+ privcmd_range->pgoff = vma->vm_pgoff;
+ privcmd_range->num_entries = num_entries;
+ privcmd_range->private = NULL;
+ privcmd_range->callback = NULL;
-+ for (i = 0; i < privcmd_range->num_entries; i++) {
++ for (i = 0; i < privcmd_range->num_entries; i++)
+ xen_ia64_privcmd_init_entry(&privcmd_range->entries[i]);
-+ }
+
+ __xen_ia64_privcmd_vma_open(vma, privcmd_vma, privcmd_range);
+ return 0;
+
+int
+direct_remap_pfn_range(struct vm_area_struct *vma,
-+ unsigned long address, // process virtual address
-+ unsigned long gmfn, // gmfn, gmfn + 1, ... gmfn + size/PAGE_SIZE
++ unsigned long address, /* process virtual address */
++ unsigned long gmfn, /* gmfn, gmfn + 1, ... gmfn + size/PAGE_SIZE */
+ unsigned long size,
+ pgprot_t prot,
-+ domid_t domid) // target domain
++ domid_t domid) /* target domain */
+{
-+ struct xen_ia64_privcmd_vma* privcmd_vma =
++ struct xen_ia64_privcmd_vma *privcmd_vma =
+ (struct xen_ia64_privcmd_vma*)vma->vm_private_data;
-+ struct xen_ia64_privcmd_range* privcmd_range = privcmd_vma->range;
++ struct xen_ia64_privcmd_range *privcmd_range = privcmd_vma->range;
+ unsigned long entry_offset = vma->vm_pgoff - privcmd_range->pgoff;
+
+ unsigned long i;
+ BUG_ON(!is_running_on_xen());
+
+#if 0
-+ if (prot != vm->vm_page_prot) {
++ if (prot != vm->vm_page_prot)
+ return -EINVAL;
-+ }
+#endif
+
+ i = (address - vma->vm_start) >> PAGE_SHIFT;
+ for (offset = 0; offset < size; offset += PAGE_SIZE) {
+ error = xen_ia64_privcmd_entry_mmap(vma, (address + offset) & PAGE_MASK, privcmd_range, entry_offset + i, gmfn, prot, domid);
-+ if (error != 0) {
++ if (error != 0)
+ break;
-+ }
+
+ i++;
+ gmfn++;
-+ }
++ }
+
+ return error;
+}
+
+
-+///////////////////////////////////////////////////////////////////////////
-+// expose p2m table
++/**************************************************************************
++ * expose p2m table
++ */
+#ifdef CONFIG_XEN_IA64_EXPOSE_P2M
+#include <linux/cpu.h>
+#include <asm/uaccess.h>
+};
+static unsigned long p2m_assign_start_pfn __read_mostly;
+static unsigned long p2m_assign_end_pfn __read_mostly;
-+static unsigned long p2m_expose_size; // this is referenced only when resume.
-+ // so __read_mostly doesn't make sense.
-+volatile const pte_t* p2m_pte __read_mostly;
++static unsigned long p2m_expose_size; /* this is referenced only when resume.
++ * so __read_mostly doesn't make sense.
++ */
++volatile const pte_t *p2m_pte __read_mostly;
+
+#define GRANULE_PFN PTRS_PER_PTE
+static unsigned long p2m_granule_pfn __read_mostly = GRANULE_PFN;
+static int xen_ia64_p2m_expose __read_mostly = 1;
+module_param(xen_ia64_p2m_expose, int, 0);
+MODULE_PARM_DESC(xen_ia64_p2m_expose,
-+ "enable/disable xen/ia64 p2m exposure optimization\n");
++ "enable/disable xen/ia64 p2m exposure optimization\n");
+
+#ifdef CONFIG_XEN_IA64_EXPOSE_P2M_USE_DTR
+static int xen_ia64_p2m_expose_use_dtr __read_mostly = 1;
+module_param(xen_ia64_p2m_expose_use_dtr, int, 0);
+MODULE_PARM_DESC(xen_ia64_p2m_expose_use_dtr,
-+ "use/unuse dtr to map exposed p2m table\n");
++ "use/unuse dtr to map exposed p2m table\n");
+
+static const int p2m_page_shifts[] = {
+ _PAGE_SIZE_4K,
+};
+static struct p2m_itr_arg p2m_itr_arg __read_mostly;
+
-+// This should be in asm-ia64/kregs.h
++/* This should be in asm-ia64/kregs.h */
+#define IA64_TR_P2M_TABLE 3
+
+static void
-+p2m_itr(void* info)
++p2m_itr(void *info)
+{
-+ struct p2m_itr_arg* arg = (struct p2m_itr_arg*)info;
++ struct p2m_itr_arg *arg = (struct p2m_itr_arg*)info;
+ ia64_itr(0x2, IA64_TR_P2M_TABLE,
-+ arg->vaddr, arg->pteval, arg->log_page_size);
++ arg->vaddr, arg->pteval, arg->log_page_size);
+ ia64_srlz_d();
+}
+
+static int
+p2m_expose_dtr_call(struct notifier_block *self,
-+ unsigned long event, void* ptr)
++ unsigned long event, void *ptr)
+{
+ unsigned int cpu = (unsigned int)(long)ptr;
+ if (event != CPU_ONLINE)
+ continue;
+
+ granule_pfn = max(page_size >> PAGE_SHIFT,
-+ p2m_granule_pfn);
++ p2m_granule_pfn);
+ p2m_convert_min_pfn = ROUNDDOWN(p2m_min_low_pfn,
-+ granule_pfn);
++ granule_pfn);
+ p2m_convert_max_pfn = ROUNDUP(p2m_max_low_pfn,
-+ granule_pfn);
++ granule_pfn);
+ num_pfn = p2m_convert_max_pfn - p2m_convert_min_pfn;
+ p2m_expose_size = num_pfn << PAGE_SHIFT;
+ p2m_size = p2m_table_size(num_pfn);
-+ p2m_size = ROUNDUP(p2m_size, granule_pfn << PAGE_SHIFT);
++ p2m_size = ROUNDUP(p2m_size,
++ granule_pfn << PAGE_SHIFT);
+ if (p2m_size == page_size)
+ break;
+ }
+ {
+ BUG_ON(p2m_granule_pfn & (p2m_granule_pfn - 1));
+ p2m_convert_min_pfn = ROUNDDOWN(p2m_min_low_pfn,
-+ p2m_granule_pfn);
-+ p2m_convert_max_pfn = ROUNDUP(p2m_max_low_pfn, p2m_granule_pfn);
++ p2m_granule_pfn);
++ p2m_convert_max_pfn = ROUNDUP(p2m_max_low_pfn,
++ p2m_granule_pfn);
+ num_pfn = p2m_convert_max_pfn - p2m_convert_min_pfn;
+ p2m_expose_size = num_pfn << PAGE_SHIFT;
+ p2m_size = p2m_table_size(num_pfn);
+ p2m_size = ROUNDUP(p2m_size, p2m_granule_pfn << PAGE_SHIFT);
+ align = max(privcmd_resource_align,
-+ p2m_granule_pfn << PAGE_SHIFT);
++ p2m_granule_pfn << PAGE_SHIFT);
+ }
+
-+ // use privcmd region
++ /* use privcmd region */
+ error = allocate_resource(&iomem_resource, &p2m_resource, p2m_size,
-+ privcmd_resource_min, privcmd_resource_max,
-+ align, NULL, NULL);
++ privcmd_resource_min, privcmd_resource_max,
++ align, NULL, NULL);
+ if (error) {
+ printk(KERN_ERR P2M_PREFIX
+ "can't allocate region for p2m exposure "
+ p2m_assign_end_pfn = p2m_resource.end >> PAGE_SHIFT;
+
+ error = HYPERVISOR_expose_p2m(p2m_convert_min_pfn,
-+ p2m_assign_start_pfn,
-+ p2m_expose_size, p2m_granule_pfn);
++ p2m_assign_start_pfn,
++ p2m_expose_size, p2m_granule_pfn);
+ if (error) {
+ printk(KERN_ERR P2M_PREFIX "failed expose p2m hypercall %d\n",
+ error);
+#ifdef CONFIG_XEN_IA64_EXPOSE_P2M_USE_DTR
+ if (xen_ia64_p2m_expose_use_dtr) {
+ p2m_itr_arg.vaddr = (unsigned long)__va(p2m_assign_start_pfn
-+ << PAGE_SHIFT);
++ << PAGE_SHIFT);
+ p2m_itr_arg.pteval = pte_val(pfn_pte(p2m_assign_start_pfn,
-+ PAGE_KERNEL));
++ PAGE_KERNEL));
+ p2m_itr_arg.log_page_size = log_page_size;
+ smp_mb();
+ smp_call_function(&p2m_itr, &p2m_itr_arg, 1, 1);
+ * interrupts are masked when resume.
+ */
+ error = HYPERVISOR_expose_p2m(p2m_convert_min_pfn,
-+ p2m_assign_start_pfn,
-+ p2m_expose_size, p2m_granule_pfn);
++ p2m_assign_start_pfn,
++ p2m_expose_size, p2m_granule_pfn);
+ if (error) {
+ printk(KERN_ERR P2M_PREFIX "failed expose p2m hypercall %d\n",
+ error);
+ }
+}
+
-+//XXX inlinize?
++/* XXX inlinize? */
+unsigned long
+p2m_phystomach(unsigned long gpfn)
+{
-+ volatile const pte_t* pte;
++ volatile const pte_t *pte;
+ unsigned long mfn;
+ unsigned long pteval;
+
+
+ mfn = INVALID_MFN;
+ if (likely(__get_user(pteval, (unsigned long __user *)pte) == 0 &&
-+ pte_present(__pte(pteval)) &&
-+ pte_pfn(__pte(pteval)) != (INVALID_MFN >> PAGE_SHIFT)))
++ pte_present(__pte(pteval)) &&
++ pte_pfn(__pte(pteval)) != (INVALID_MFN >> PAGE_SHIFT)))
+ mfn = (pteval & _PFN_MASK) >> PAGE_SHIFT;
+
+ return mfn;
+EXPORT_SYMBOL_GPL(p2m_pte);
+EXPORT_SYMBOL_GPL(p2m_phystomach);
+
-+///////////////////////////////////////////////////////////////////////////
-+// foreign domain p2m mapping
++/**************************************************************************
++ * foreign domain p2m mapping
++ */
+#include <asm/xen/xencomm.h>
+#include <xen/public/privcmd.h>
+
+};
+
+static void
-+xen_foreign_p2m_unexpose(struct xen_ia64_privcmd_range* privcmd_range,
-+ void* arg)
++xen_foreign_p2m_unexpose(struct xen_ia64_privcmd_range *privcmd_range,
++ void *arg)
+{
-+ struct foreign_p2m_private* private = (struct foreign_p2m_private*)arg;
++ struct foreign_p2m_private *private = (struct foreign_p2m_private*)arg;
+ int ret;
+
+ privcmd_range->private = NULL;
+}
+
+int
-+xen_foreign_p2m_expose(privcmd_hypercall_t* hypercall)
++xen_foreign_p2m_expose(privcmd_hypercall_t *hypercall)
+{
-+ // hypercall->
-+ // arg0: cmd = IA64_DOM0VP_expose_foreign_p2m
-+ // arg1: va
-+ // arg2: domid
-+ // arg3: __user* memmap_info
-+ // arg4: flags
++ /*
++ * hypercall->
++ * arg0: cmd = IA64_DOM0VP_expose_foreign_p2m
++ * arg1: va
++ * arg2: domid
++ * arg3: __user* memmap_info
++ * arg4: flags
++ */
+
+ int ret = 0;
-+ struct mm_struct* mm = current->mm;
++ struct mm_struct *mm = current->mm;
+
+ unsigned long vaddr = hypercall->arg[1];
+ domid_t domid = hypercall->arg[2];
+
+ struct xen_ia64_memmap_info memmap_info;
+ size_t memmap_size;
-+ struct xen_ia64_memmap_info* k_memmap_info = NULL;
++ struct xen_ia64_memmap_info *k_memmap_info = NULL;
+ unsigned long max_gpfn;
+ unsigned long p2m_size;
-+ struct resource* res;
++ struct resource *res;
+ unsigned long gpfn;
+
-+ struct vm_area_struct* vma;
-+ void* p;
++ struct vm_area_struct *vma;
++ void *p;
+ unsigned long prev_src_gpfn_end;
+
-+ struct xen_ia64_privcmd_vma* privcmd_vma;
-+ struct xen_ia64_privcmd_range* privcmd_range;
-+ struct foreign_p2m_private* private = NULL;
++ struct xen_ia64_privcmd_vma *privcmd_vma;
++ struct xen_ia64_privcmd_range *privcmd_range;
++ struct foreign_p2m_private *private = NULL;
+
+ BUG_ON(hypercall->arg[0] != IA64_DOM0VP_expose_foreign_p2m);
+
+ }
+
+ gpfn = res->start >> PAGE_SHIFT;
-+ // arg0: dest_gpfn
-+ // arg1: domid
-+ // arg2: XEN_GUEST_HANDLE(char) buffer: memmap_info
-+ // arg3: flags
-+ // The hypercall checks its intergirty/simplfies it and
-+ // copy it back for us.
++ /*
++ * arg0: dest_gpfn
++ * arg1: domid
++ * arg2: XEN_GUEST_HANDLE(char) buffer: memmap_info
++ * arg3: flags
++ * The hypercall checks its intergirty/simplfies it and
++ * copy it back for us.
++ */
+ ret = xencomm_arch_expose_foreign_p2m(gpfn, domid,
+ xencomm_map_no_alloc(k_memmap_info, memmap_size),
+ hypercall->arg[4]);
+ vma->vm_page_prot);
+ if (ret) {
+ for (i = 0; i < gpfn + gpfn_offset; i++) {
-+ struct xen_ia64_privcmd_entry* entry =
++ struct xen_ia64_privcmd_entry *entry =
+ &privcmd_range->entries[i];
+ BUG_ON(atomic_read(&entry->map_count) != 1 &&
+ atomic_read(&entry->map_count) != 0);
+ for (i = gpfn_offset;
+ i < gpfn_offset + (size >> PAGE_SHIFT);
+ i++) {
-+ struct xen_ia64_privcmd_entry* entry =
++ struct xen_ia64_privcmd_entry *entry =
+ &privcmd_range->entries[i];
+ BUG_ON(atomic_read(&entry->map_count) != 0);
+ BUG_ON(entry->gpfn != INVALID_GPFN);
+}
+#endif
+
-+///////////////////////////////////////////////////////////////////////////
-+// for xenoprof
-+
++/**************************************************************************
++ * for xenoprof
++ */
+struct resource*
+xen_ia64_allocate_resource(unsigned long size)
+{
-+ struct resource* res;
++ struct resource *res;
+ int error;
+
+ res = kzalloc(sizeof(*res), GFP_KERNEL);
+ res->name = "Xen";
+ res->flags = IORESOURCE_MEM;
+ error = allocate_resource(&iomem_resource, res, PAGE_ALIGN(size),
-+ privcmd_resource_min, privcmd_resource_max,
-+ IA64_GRANULE_SIZE, NULL, NULL);
++ privcmd_resource_min, privcmd_resource_max,
++ IA64_GRANULE_SIZE, NULL, NULL);
+ if (error) {
+ kfree(res);
+ return ERR_PTR(error);
+EXPORT_SYMBOL_GPL(xen_ia64_allocate_resource);
+
+void
-+xen_ia64_release_resource(struct resource* res)
++xen_ia64_release_resource(struct resource *res)
+{
+ release_resource(res);
+ kfree(res);
+EXPORT_SYMBOL_GPL(xen_ia64_release_resource);
+
+void
-+xen_ia64_unmap_resource(struct resource* res)
++xen_ia64_unmap_resource(struct resource *res)
+{
+ unsigned long gpfn = res->start >> PAGE_SHIFT;
+ unsigned long nr_pages = (res->end - res->start) >> PAGE_SHIFT;
+}
+EXPORT_SYMBOL_GPL(xen_ia64_unmap_resource);
+
-+///////////////////////////////////////////////////////////////////////////
-+// opt feature
++/**************************************************************************
++ * opt feature
++ */
+void
+xen_ia64_enable_opt_feature(void)
+{
+ HYPERVISOR_opt_feature(&optf);
+}
+
-+///////////////////////////////////////////////////////////////////////////
-+// suspend/resume
++/**************************************************************************
++ * suspend/resume
++ */
+void
+xen_post_suspend(int suspend_cancelled)
+{
+ xen_ia64_enable_opt_feature();
+ /* add more if necessary */
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/machvec.c linux-2.6.18-xen.hg/arch/ia64/xen/machvec.c
---- linux-2.6.18/arch/ia64/xen/machvec.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/machvec.c 2007-12-23 11:14:54.672534273 +0100
+--- linux-2.6.18.8/arch/ia64/xen/machvec.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/machvec.c 2008-05-19 00:33:05.032436225 +0300
@@ -0,0 +1,4 @@
+#define MACHVEC_PLATFORM_NAME xen
+#define MACHVEC_PLATFORM_HEADER <asm/machvec_xen.h>
+#include <asm/machvec_init.h>
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/Makefile linux-2.6.18-xen.hg/arch/ia64/xen/Makefile
---- linux-2.6.18/arch/ia64/xen/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/Makefile 2007-12-23 11:14:54.672534273 +0100
-@@ -0,0 +1,9 @@
-+#
-+# Makefile for Xen components
-+#
-+
-+obj-y := hypercall.o xenivt.o xenentry.o xensetup.o xenpal.o xenhpski.o \
-+ hypervisor.o util.o xencomm.o xcom_hcall.o \
-+ xcom_privcmd.o xen_dma.o
-+
-+obj-$(CONFIG_IA64_GENERIC) += machvec.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/swiotlb.c linux-2.6.18-xen.hg/arch/ia64/xen/swiotlb.c
---- linux-2.6.18/arch/ia64/xen/swiotlb.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/swiotlb.c 2007-12-23 11:14:54.672534273 +0100
+--- linux-2.6.18.8/arch/ia64/xen/swiotlb.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/swiotlb.c 2008-05-19 00:33:05.036436456 +0300
@@ -0,0 +1,906 @@
+/*
+ * Dynamic DMA mapping support.
+EXPORT_SYMBOL(swiotlb_alloc_coherent);
+EXPORT_SYMBOL(swiotlb_free_coherent);
+EXPORT_SYMBOL(swiotlb_dma_supported);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/util.c linux-2.6.18-xen.hg/arch/ia64/xen/util.c
---- linux-2.6.18/arch/ia64/xen/util.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/util.c 2007-12-23 11:14:54.675867776 +0100
-@@ -0,0 +1,105 @@
+--- linux-2.6.18.8/arch/ia64/xen/util.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/util.c 2008-05-19 00:33:05.036436456 +0300
+@@ -0,0 +1,102 @@
+/******************************************************************************
+ * arch/ia64/xen/util.c
+ * This file is the ia64 counterpart of drivers/xen/util.c
+ int order;
+ unsigned long virt;
+ unsigned long nr_pages;
-+ struct vm_struct* area;
-+
++ struct vm_struct *area;
++
+ order = get_order(size);
+ virt = __get_free_pages(GFP_KERNEL, order);
-+ if (virt == 0) {
++ if (virt == 0)
+ goto err0;
-+ }
+ nr_pages = 1 << order;
+ scrub_pages(virt, nr_pages);
-+
++
+ area = kmalloc(sizeof(*area), GFP_KERNEL);
-+ if (area == NULL) {
++ if (area == NULL)
+ goto err1;
-+ }
-+
-+ area->flags = VM_IOREMAP;//XXX
++
++ area->flags = VM_IOREMAP; /* XXX */
+ area->addr = (void*)virt;
+ area->size = size;
-+ area->pages = NULL; //XXX
++ area->pages = NULL; /* XXX */
+ area->nr_pages = nr_pages;
+ area->phys_addr = 0; /* xenbus_map_ring_valloc uses this field! */
+
+ free_pages(virt, order);
+err0:
+ return NULL;
-+
+}
+EXPORT_SYMBOL_GPL(alloc_vm_area);
+
+ unsigned long i;
+ unsigned long phys_addr = __pa(area->addr);
+
-+ // This area is used for foreign page mappping.
-+ // So underlying machine page may not be assigned.
++ /* This area is used for foreign page mappping.
++ * So underlying machine page may not be assigned. */
+ for (i = 0; i < (1 << order); i++) {
+ unsigned long ret;
+ unsigned long gpfn = (phys_addr >> PAGE_SHIFT) + i;
+ * tab-width: 8
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xcom_hcall.c linux-2.6.18-xen.hg/arch/ia64/xen/xcom_hcall.c
---- linux-2.6.18/arch/ia64/xen/xcom_hcall.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xcom_hcall.c 2007-12-23 11:14:54.675867776 +0100
-@@ -0,0 +1,663 @@
+--- linux-2.6.18.8/arch/ia64/xen/xcom_asm.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xcom_asm.S 2008-05-19 00:33:05.036436456 +0300
+@@ -0,0 +1,27 @@
++/*
++ * xencomm suspend support
++ * Support routines for Xen
++ *
++ * Copyright (C) 2005 Dan Magenheimer <dan.magenheimer@hp.com>
++ */
++#include <asm/asmmacro.h>
++#include <xen/interface/xen.h>
++
++/*
++ * Stub for suspend.
++ * Just force the stacked registers to be written in memory.
++ */
++GLOBAL_ENTRY(xencomm_arch_hypercall_suspend)
++ ;;
++ alloc r20=ar.pfs,0,0,6,0
++ mov r2=__HYPERVISOR_sched_op
++ ;;
++ /* We don't want to deal with RSE. */
++ flushrs
++ mov r33=r32
++ mov r32=2 // SCHEDOP_shutdown
++ ;;
++ break 0x1000
++ ;;
++ br.ret.sptk.many b0
++END(xencomm_arch_hypercall_suspend)
+--- linux-2.6.18.8/arch/ia64/xen/xcom_hcall.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xcom_hcall.c 2008-05-19 00:33:05.036436456 +0300
+@@ -0,0 +1,661 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+}
+EXPORT_SYMBOL_GPL(xencomm_hypercall_hvm_op);
+
-+#ifndef CONFIG_VMX_GUEST
+int
+xencomm_hypercall_suspend(unsigned long srec)
+{
+ return xencomm_arch_hypercall_suspend(
+ xencomm_map_no_alloc(&arg, sizeof(arg)));
+}
-+#endif
+
+int
+xencomm_hypercall_xenoprof_op(int op, void *arg)
+
+ return xencomm_arch_hypercall_kexec_op(cmd, desc);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xcom_privcmd.c linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c
---- linux-2.6.18/arch/ia64/xen/xcom_privcmd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c 2007-12-23 11:14:54.675867776 +0100
-@@ -0,0 +1,823 @@
+--- linux-2.6.18.8/arch/ia64/xen/xcom_privcmd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c 2008-05-19 00:33:05.040436686 +0300
+@@ -0,0 +1,824 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ break;
+ case XEN_SYSCTL_tbuf_op:
+ case XEN_SYSCTL_sched_id:
++ case XEN_SYSCTL_availheap:
+ break;
+ case XEN_SYSCTL_perfc_op:
+ {
+ .interface_version = XEN_SYSCTL_INTERFACE_VERSION,
+ .u.perfc_op = {
+ .cmd = XEN_SYSCTL_PERFCOP_query,
-+ // .desc.p = NULL,
-+ // .val.p = NULL,
++ /* .desc.p = NULL, */
++ /* .val.p = NULL, */
+ },
+ };
+
+ }
+}
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xencomm.c linux-2.6.18-xen.hg/arch/ia64/xen/xencomm.c
---- linux-2.6.18/arch/ia64/xen/xencomm.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xencomm.c 2007-12-23 11:14:54.675867776 +0100
-@@ -0,0 +1,109 @@
-+/*
-+ * Copyright (C) 2006 Hollis Blanchard <hollisb@us.ibm.com>, IBM Corporation
-+ *
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License as published by
-+ * the Free Software Foundation; either version 2 of the License, or
-+ * (at your option) any later version.
-+ *
-+ * This program is distributed in the hope that it will be useful,
-+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
-+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-+ * GNU General Public License for more details.
-+ *
-+ * You should have received a copy of the GNU General Public License
-+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+ */
-+
-+#include <linux/gfp.h>
-+#include <linux/mm.h>
-+#include <xen/interface/xen.h>
-+#include <asm/page.h>
-+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
-+
-+#include <asm/xen/xencomm.h>
-+
-+static unsigned long kernel_start_pa;
-+
-+void
-+xencomm_initialize (void)
-+{
-+ kernel_start_pa = KERNEL_START - ia64_tpa(KERNEL_START);
-+}
-+
-+/* Translate virtual address to physical address. */
-+unsigned long
-+xencomm_vtop(unsigned long vaddr)
-+{
-+#ifndef CONFIG_VMX_GUEST
-+ struct page *page;
-+ struct vm_area_struct *vma;
-+#endif
-+
-+ if (vaddr == 0)
-+ return 0;
-+
-+#ifdef __ia64__
-+ if (REGION_NUMBER(vaddr) == 5) {
-+ pgd_t *pgd;
-+ pud_t *pud;
-+ pmd_t *pmd;
-+ pte_t *ptep;
-+
-+ /* On ia64, TASK_SIZE refers to current. It is not initialized
-+ during boot.
-+ Furthermore the kernel is relocatable and __pa() doesn't
-+ work on addresses. */
-+ if (vaddr >= KERNEL_START
-+ && vaddr < (KERNEL_START + KERNEL_TR_PAGE_SIZE)) {
-+ return vaddr - kernel_start_pa;
-+ }
-+
-+ /* In kernel area -- virtually mapped. */
-+ pgd = pgd_offset_k(vaddr);
-+ if (pgd_none(*pgd) || pgd_bad(*pgd))
-+ return ~0UL;
-+
-+ pud = pud_offset(pgd, vaddr);
-+ if (pud_none(*pud) || pud_bad(*pud))
-+ return ~0UL;
-+
-+ pmd = pmd_offset(pud, vaddr);
-+ if (pmd_none(*pmd) || pmd_bad(*pmd))
-+ return ~0UL;
-+
-+ ptep = pte_offset_kernel(pmd, vaddr);
-+ if (!ptep)
-+ return ~0UL;
-+
-+ return (pte_val(*ptep) & _PFN_MASK) | (vaddr & ~PAGE_MASK);
-+ }
-+#endif
-+
-+ if (vaddr > TASK_SIZE) {
-+ /* kernel address */
-+ return __pa(vaddr);
-+ }
-+
-+
-+#ifdef CONFIG_VMX_GUEST
-+ /* No privcmd within vmx guest. */
-+ return ~0UL;
-+#else
-+ /* XXX double-check (lack of) locking */
-+ vma = find_extend_vma(current->mm, vaddr);
-+ if (!vma)
-+ return ~0UL;
-+
-+ /* We assume the page is modified. */
-+ page = follow_page(vma, vaddr, FOLL_WRITE | FOLL_TOUCH);
-+ if (!page)
-+ return ~0UL;
-+
-+ return (page_to_pfn(page) << PAGE_SHIFT) | (vaddr & ~PAGE_MASK);
-+#endif
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xen_dma.c linux-2.6.18-xen.hg/arch/ia64/xen/xen_dma.c
---- linux-2.6.18/arch/ia64/xen/xen_dma.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xen_dma.c 2007-12-23 11:14:54.675867776 +0100
+--- linux-2.6.18.8/arch/ia64/xen/xen_dma.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xen_dma.c 2008-05-19 00:33:05.044436917 +0300
@@ -0,0 +1,155 @@
+/*
+ * Copyright (C) 2007 Hewlett-Packard Development Company, L.P.
+ gnttab_dma_unmap_page(dma_addr);
+}
+EXPORT_SYMBOL(xen_unmap_single);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xenentry.S linux-2.6.18-xen.hg/arch/ia64/xen/xenentry.S
---- linux-2.6.18/arch/ia64/xen/xenentry.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xenentry.S 2007-12-23 11:14:54.679201278 +0100
+--- linux-2.6.18.8/arch/ia64/xen/xencomm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xencomm.c 2008-05-19 00:33:05.044436917 +0300
+@@ -0,0 +1,109 @@
++/*
++ * Copyright (C) 2006 Hollis Blanchard <hollisb@us.ibm.com>, IBM Corporation
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ */
++
++#include <linux/gfp.h>
++#include <linux/mm.h>
++#include <xen/interface/xen.h>
++#include <asm/page.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++#include <asm/xen/xencomm.h>
++
++static unsigned long kernel_start_pa;
++
++void
++xencomm_initialize (void)
++{
++ kernel_start_pa = KERNEL_START - ia64_tpa(KERNEL_START);
++}
++
++/* Translate virtual address to physical address. */
++unsigned long
++xencomm_vtop(unsigned long vaddr)
++{
++#ifndef CONFIG_VMX_GUEST
++ struct page *page;
++ struct vm_area_struct *vma;
++#endif
++
++ if (vaddr == 0)
++ return 0;
++
++#ifdef __ia64__
++ if (REGION_NUMBER(vaddr) == 5) {
++ pgd_t *pgd;
++ pud_t *pud;
++ pmd_t *pmd;
++ pte_t *ptep;
++
++ /* On ia64, TASK_SIZE refers to current. It is not initialized
++ during boot.
++ Furthermore the kernel is relocatable and __pa() doesn't
++ work on addresses. */
++ if (vaddr >= KERNEL_START
++ && vaddr < (KERNEL_START + KERNEL_TR_PAGE_SIZE)) {
++ return vaddr - kernel_start_pa;
++ }
++
++ /* In kernel area -- virtually mapped. */
++ pgd = pgd_offset_k(vaddr);
++ if (pgd_none(*pgd) || pgd_bad(*pgd))
++ return ~0UL;
++
++ pud = pud_offset(pgd, vaddr);
++ if (pud_none(*pud) || pud_bad(*pud))
++ return ~0UL;
++
++ pmd = pmd_offset(pud, vaddr);
++ if (pmd_none(*pmd) || pmd_bad(*pmd))
++ return ~0UL;
++
++ ptep = pte_offset_kernel(pmd, vaddr);
++ if (!ptep)
++ return ~0UL;
++
++ return (pte_val(*ptep) & _PFN_MASK) | (vaddr & ~PAGE_MASK);
++ }
++#endif
++
++ if (vaddr > TASK_SIZE) {
++ /* kernel address */
++ return __pa(vaddr);
++ }
++
++
++#ifdef CONFIG_VMX_GUEST
++ /* No privcmd within vmx guest. */
++ return ~0UL;
++#else
++ /* XXX double-check (lack of) locking */
++ vma = find_extend_vma(current->mm, vaddr);
++ if (!vma)
++ return ~0UL;
++
++ /* We assume the page is modified. */
++ page = follow_page(vma, vaddr, FOLL_WRITE | FOLL_TOUCH);
++ if (!page)
++ return ~0UL;
++
++ return (page_to_pfn(page) << PAGE_SHIFT) | (vaddr & ~PAGE_MASK);
++#endif
++}
+--- linux-2.6.18.8/arch/ia64/xen/xenentry.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xenentry.S 2008-05-19 00:33:05.044436917 +0300
@@ -0,0 +1,931 @@
+/*
+ * ia64/xen/entry.S
+#else
+END(ia64_leave_kernel)
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xenhpski.c linux-2.6.18-xen.hg/arch/ia64/xen/xenhpski.c
---- linux-2.6.18/arch/ia64/xen/xenhpski.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xenhpski.c 2007-12-23 11:14:54.679201278 +0100
-@@ -0,0 +1,19 @@
-+#include <linux/kernel.h>
-+#include <asm/hypervisor.h>
-+
-+int
-+running_on_sim(void)
-+{
-+ int i;
-+ long cpuid[6];
-+
-+ for (i = 0; i < 5; ++i)
-+ cpuid[i] = xen_get_cpuid(i);
-+ if ((cpuid[0] & 0xff) != 'H') return 0;
-+ if ((cpuid[3] & 0xff) != 0x4) return 0;
-+ if (((cpuid[3] >> 8) & 0xff) != 0x0) return 0;
-+ if (((cpuid[3] >> 16) & 0xff) != 0x0) return 0;
-+ if (((cpuid[3] >> 24) & 0x7) != 0x7) return 0;
-+ return 1;
-+}
-+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xenivt.S linux-2.6.18-xen.hg/arch/ia64/xen/xenivt.S
---- linux-2.6.18/arch/ia64/xen/xenivt.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xenivt.S 2007-12-23 11:14:54.679201278 +0100
-@@ -0,0 +1,2177 @@
+--- linux-2.6.18.8/arch/ia64/xen/xenivt.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xenivt.S 2008-05-19 00:33:05.048437147 +0300
+@@ -0,0 +1,2178 @@
+/*
+ * arch/ia64/xen/ivt.S
+ *
+ st4 [r3]=r14 // vpsr.ic = 1
+ adds r3=8,r2 // set up second base pointer
+ ;;
++ sum PSR_DEFAULT_BITS
+#else
+ mov out0=cr.ifa
+ mov out1=cr.isr
+ adds r3=8,r2 // set up second base pointer
+ ;;
+ ssm psr.ic | PSR_DEFAULT_BITS
++#endif
+ ;;
+ srlz.i // guarantee that interruption collectin is on
+ ;;
-+#endif
+#ifdef CONFIG_XEN
+
+#define MASK_TO_PEND_OFS (-1)
+
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xenminstate.h linux-2.6.18-xen.hg/arch/ia64/xen/xenminstate.h
---- linux-2.6.18/arch/ia64/xen/xenminstate.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xenminstate.h 2007-12-23 11:14:54.679201278 +0100
+--- linux-2.6.18.8/arch/ia64/xen/xenminstate.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xenminstate.h 2008-05-19 00:33:05.068438300 +0300
@@ -0,0 +1,358 @@
+#include <asm/cache.h>
+
+#else
+#define SAVE_MIN DO_SAVE_MIN( , mov r30=r0, )
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xenpal.S linux-2.6.18-xen.hg/arch/ia64/xen/xenpal.S
---- linux-2.6.18/arch/ia64/xen/xenpal.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xenpal.S 2007-12-23 11:14:54.679201278 +0100
+--- linux-2.6.18.8/arch/ia64/xen/xenpal.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xenpal.S 2008-05-19 00:33:05.072438531 +0300
@@ -0,0 +1,85 @@
+/*
+ * ia64/xen/xenpal.S
+ srlz.d // seralize restoration of psr.l
+ br.ret.sptk.many b0
+END(xen_pal_call_static)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/ia64/xen/xensetup.S linux-2.6.18-xen.hg/arch/ia64/xen/xensetup.S
---- linux-2.6.18/arch/ia64/xen/xensetup.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/ia64/xen/xensetup.S 2007-12-23 11:14:54.682534784 +0100
-@@ -0,0 +1,59 @@
+--- linux-2.6.18.8/arch/ia64/xen/xensetup.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/ia64/xen/xensetup.S 2008-05-19 00:33:05.072438531 +0300
+@@ -0,0 +1,40 @@
+/*
+ * Support routines for Xen
+ *
+ br.ret.sptk.many rp
+ ;;
+END(early_xen_setup)
+--- linux-2.6.18.8/arch/powerpc/Kconfig 2008-05-19 00:42:33.785223278 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/Kconfig 2008-05-19 00:33:12.688877542 +0300
+@@ -439,6 +439,17 @@
+ bool
+ default n
+
++config PPC_XEN
++ bool "Enable Xen compatible kernel"
++ depends on PPC_MULTIPLATFORM && PPC64 && PPC_MAPLE && PPC_PSERIES && SMP
++ select XEN
++ select XEN_PRIVILEGED_GUEST
++ select XEN_UNPRIVILEGED_GUEST
++ select XEN_XENCOMM
++
++ help
++ This option will compile a kernel compatible with Xen hypervisor
+
-+#include <xen/interface/xen.h>
+ config XICS
+ depends on PPC_PSERIES
+ bool
+@@ -1080,6 +1091,8 @@
+
+ source "security/Kconfig"
+
++source "drivers/xen/Kconfig"
+
-+/* Stub for suspend.
-+ Just force the stacked registers to be written in memory. */
-+GLOBAL_ENTRY(xencomm_arch_hypercall_suspend)
-+ ;;
-+ alloc r20=ar.pfs,0,0,6,0
-+ mov r2=__HYPERVISOR_sched_op
-+ ;;
-+ /* We don't want to deal with RSE. */
-+ flushrs
-+ mov r33=r32
-+ mov r32=2 // SCHEDOP_shutdown
-+ ;;
-+ break 0x1000
-+ ;;
-+ br.ret.sptk.many b0
-+END(xencomm_arch_hypercall_suspend)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/boot/Makefile linux-2.6.18-xen.hg/arch/powerpc/boot/Makefile
---- linux-2.6.18/arch/powerpc/boot/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/boot/Makefile 2007-12-23 11:14:58.952759023 +0100
+ config KEYS_COMPAT
+ bool
+ depends on COMPAT && KEYS
+--- linux-2.6.18.8/arch/powerpc/Kconfig.debug 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/Kconfig.debug 2008-05-19 00:33:12.688877542 +0300
+@@ -160,6 +160,20 @@
+ Select this to enable early debugging for legacy iSeries. You need
+ to hit "Ctrl-x Ctrl-x" to see the messages on the console.
+
++config PPC_EARLY_DEBUG_XEN_DOM0
++ bool "Xen Dom0 Console"
++ depends on PPC_XEN
++ help
++ Select this to enable early debugging for Xen Dom0. Setting
++ this will result in a kernel that may not work as a DomU.
++
++config PPC_EARLY_DEBUG_XEN_DOMU
++ bool "Xen DomU Console"
++ depends on PPC_XEN && XEN_UNPRIVILEGED_GUEST
++ help
++ Select this to enable early debugging for Xen DomU. Setting
++ this will result in a kernel that may not work as a Dom0.
++
+ endchoice
+
+ endmenu
+--- linux-2.6.18.8/arch/powerpc/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/Makefile 2008-05-19 00:33:12.688877542 +0300
+@@ -65,6 +65,7 @@
+ AFLAGS-$(CONFIG_PPC32) := -Iarch/$(ARCH)
+ CFLAGS-$(CONFIG_PPC64) := -mminimal-toc -mtraceback=none -mcall-aixdesc
+ CFLAGS-$(CONFIG_PPC32) := -Iarch/$(ARCH) -ffixed-r2 -mmultiple
++CFLAGS-$(CONFIG_PPC_XEN) += -Iinclude/asm-$(ARCH)/xen
+ CPPFLAGS += $(CPPFLAGS-y)
+ AFLAGS += $(AFLAGS-y)
+ CFLAGS += -msoft-float -pipe $(CFLAGS-y)
+--- linux-2.6.18.8/arch/powerpc/boot/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/boot/Makefile 2008-05-19 00:33:12.692877773 +0300
@@ -36,8 +36,11 @@
$(addprefix $(obj)/,$(zlib) main.o): $(addprefix $(obj)/,$(zliblinuxheader)) $(addprefix $(obj)/,$(zlibheader))
#$(addprefix $(obj)/,main.o): $(addprefix $(obj)/,zlib.h)
src-boot := $(addprefix $(obj)/, $(src-boot))
obj-boot := $(addsuffix .o, $(basename $(src-boot)))
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/boot/xen_guest.S linux-2.6.18-xen.hg/arch/powerpc/boot/xen_guest.S
---- linux-2.6.18/arch/powerpc/boot/xen_guest.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/boot/xen_guest.S 2007-12-23 11:14:58.966093066 +0100
+--- linux-2.6.18.8/arch/powerpc/boot/xen_guest.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/boot/xen_guest.S 2008-05-19 00:33:12.712878926 +0300
@@ -0,0 +1,7 @@
+ .section __xen_guest
+ .ascii "GUEST_OS=linux"
+ .ascii ",VIRT_BASE=0x0"
+ .ascii ",LOADER=generic"
+ .byte 0
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/configs/xen_maple_defconfig linux-2.6.18-xen.hg/arch/powerpc/configs/xen_maple_defconfig
---- linux-2.6.18/arch/powerpc/configs/xen_maple_defconfig 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/configs/xen_maple_defconfig 2007-12-23 11:14:58.979427095 +0100
+--- linux-2.6.18.8/arch/powerpc/configs/xen_maple_defconfig 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/configs/xen_maple_defconfig 2008-05-19 00:33:12.732880079 +0300
@@ -0,0 +1,1342 @@
+#
+# Automatically generated make config: don't edit
+#
+# Hardware crypto devices
+#
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/Kconfig linux-2.6.18-xen.hg/arch/powerpc/Kconfig
---- linux-2.6.18/arch/powerpc/Kconfig 2007-12-23 11:26:51.006787102 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/Kconfig 2007-12-23 11:14:58.952759023 +0100
-@@ -439,6 +439,17 @@
- bool
- default n
-
-+config PPC_XEN
-+ bool "Enable Xen compatible kernel"
-+ depends on PPC_MULTIPLATFORM && PPC64 && PPC_MAPLE && PPC_PSERIES && SMP
-+ select XEN
-+ select XEN_PRIVILEGED_GUEST
-+ select XEN_UNPRIVILEGED_GUEST
-+ select XEN_XENCOMM
-+
-+ help
-+ This option will compile a kernel compatible with Xen hypervisor
-+
- config XICS
- depends on PPC_PSERIES
- bool
-@@ -1080,6 +1091,8 @@
-
- source "security/Kconfig"
-
-+source "drivers/xen/Kconfig"
-+
- config KEYS_COMPAT
- bool
- depends on COMPAT && KEYS
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/Kconfig.debug linux-2.6.18-xen.hg/arch/powerpc/Kconfig.debug
---- linux-2.6.18/arch/powerpc/Kconfig.debug 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/Kconfig.debug 2007-12-23 11:14:58.952759023 +0100
-@@ -160,6 +160,20 @@
- Select this to enable early debugging for legacy iSeries. You need
- to hit "Ctrl-x Ctrl-x" to see the messages on the console.
-
-+config PPC_EARLY_DEBUG_XEN_DOM0
-+ bool "Xen Dom0 Console"
-+ depends on PPC_XEN
-+ help
-+ Select this to enable early debugging for Xen Dom0. Setting
-+ this will result in a kernel that may not work as a DomU.
-+
-+config PPC_EARLY_DEBUG_XEN_DOMU
-+ bool "Xen DomU Console"
-+ depends on PPC_XEN && XEN_UNPRIVILEGED_GUEST
-+ help
-+ Select this to enable early debugging for Xen DomU. Setting
-+ this will result in a kernel that may not work as a Dom0.
-+
- endchoice
-
- endmenu
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/kernel/cpu_setup_power4.S linux-2.6.18-xen.hg/arch/powerpc/kernel/cpu_setup_power4.S
---- linux-2.6.18/arch/powerpc/kernel/cpu_setup_power4.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/kernel/cpu_setup_power4.S 2007-12-23 11:14:58.982760606 +0100
+--- linux-2.6.18.8/arch/powerpc/kernel/cpu_setup_power4.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/kernel/cpu_setup_power4.S 2008-05-19 00:33:12.740880540 +0300
@@ -73,6 +73,13 @@
blr
mfspr r0,SPRN_HID0
li r11,5 /* clear DOZE and SLEEP */
rldimi r0,r11,52,8 /* set NAP and DPM */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/kernel/prom_init.c linux-2.6.18-xen.hg/arch/powerpc/kernel/prom_init.c
---- linux-2.6.18/arch/powerpc/kernel/prom_init.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/kernel/prom_init.c 2007-12-23 11:14:59.016095685 +0100
+--- linux-2.6.18.8/arch/powerpc/kernel/prom_init.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/kernel/prom_init.c 2008-05-19 00:33:12.852886996 +0300
@@ -188,6 +188,7 @@
#define PLATFORM_LPAR 0x0001
#define PLATFORM_POWERMAC 0x0400
/*
* On non-powermacs, try to instantiate RTAS and puts all CPUs
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/kernel/setup-common.c linux-2.6.18-xen.hg/arch/powerpc/kernel/setup-common.c
---- linux-2.6.18/arch/powerpc/kernel/setup-common.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/kernel/setup-common.c 2007-12-23 11:14:59.022762705 +0100
+--- linux-2.6.18.8/arch/powerpc/kernel/setup-common.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/kernel/setup-common.c 2008-05-19 00:33:12.864887688 +0300
@@ -387,6 +387,12 @@
}
}
#ifdef CONFIG_PPC64
/*
* On pSeries LPAR, we need to know how many cpus
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/kernel/udbg.c linux-2.6.18-xen.hg/arch/powerpc/kernel/udbg.c
---- linux-2.6.18/arch/powerpc/kernel/udbg.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/kernel/udbg.c 2007-12-23 11:14:59.032763228 +0100
+--- linux-2.6.18.8/arch/powerpc/kernel/udbg.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/kernel/udbg.c 2008-05-19 00:33:12.896889532 +0300
@@ -45,6 +45,9 @@
#elif defined(CONFIG_PPC_EARLY_DEBUG_ISERIES)
/* For iSeries - hit Ctrl-x Ctrl-x to see the output */
#endif
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/Makefile linux-2.6.18-xen.hg/arch/powerpc/Makefile
---- linux-2.6.18/arch/powerpc/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/Makefile 2007-12-23 11:14:58.952759023 +0100
-@@ -65,6 +65,7 @@
- AFLAGS-$(CONFIG_PPC32) := -Iarch/$(ARCH)
- CFLAGS-$(CONFIG_PPC64) := -mminimal-toc -mtraceback=none -mcall-aixdesc
- CFLAGS-$(CONFIG_PPC32) := -Iarch/$(ARCH) -ffixed-r2 -mmultiple
-+CFLAGS-$(CONFIG_PPC_XEN) += -Iinclude/asm-$(ARCH)/xen
- CPPFLAGS += $(CPPFLAGS-y)
- AFLAGS += $(AFLAGS-y)
- CFLAGS += -msoft-float -pipe $(CFLAGS-y)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/mm/slb_low.S linux-2.6.18-xen.hg/arch/powerpc/mm/slb_low.S
---- linux-2.6.18/arch/powerpc/mm/slb_low.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/mm/slb_low.S 2007-12-23 11:14:59.089432866 +0100
+--- linux-2.6.18.8/arch/powerpc/kernel/vdso.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/kernel/vdso.c 2008-05-19 00:33:12.896889532 +0300
+@@ -282,6 +282,13 @@
+ * pages though
+ */
+ vma->vm_flags = VM_READ|VM_EXEC|VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC;
++ /*
++ * Make sure the vDSO gets into every core dump.
++ * Dumping its contents makes post-mortem fully interpretable later
++ * without matching up the same kernel and hardware config to see
++ * what PC values meant.
++ */
++ vma->vm_flags |= VM_ALWAYSDUMP;
+ vma->vm_flags |= mm->def_flags;
+ vma->vm_page_prot = protection_map[vma->vm_flags & 0x7];
+ vma->vm_ops = &vdso_vmops;
+--- linux-2.6.18.8/arch/powerpc/mm/slb_low.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/mm/slb_low.S 2008-05-19 00:33:12.992895066 +0300
@@ -51,6 +51,23 @@
*/
bne cr7,1f
/* Linear mapping encoding bits, the "li" instruction below will
* be patched by the kernel at boot
*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/Makefile linux-2.6.18-xen.hg/arch/powerpc/platforms/Makefile
---- linux-2.6.18/arch/powerpc/platforms/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/Makefile 2007-12-23 11:14:59.109433918 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/Makefile 2008-05-19 00:33:13.064899216 +0300
@@ -12,6 +12,8 @@
obj-$(CONFIG_PPC_86xx) += 86xx/
obj-$(CONFIG_PPC_PSERIES) += pseries/
obj-$(CONFIG_PPC_MAPLE) += maple/
obj-$(CONFIG_PPC_CELL) += cell/
obj-$(CONFIG_EMBEDDED6xx) += embedded6xx/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/pseries/iommu.c linux-2.6.18-xen.hg/arch/powerpc/platforms/pseries/iommu.c
---- linux-2.6.18/arch/powerpc/platforms/pseries/iommu.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/pseries/iommu.c 2007-12-23 11:14:59.179437597 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/pseries/iommu.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/pseries/iommu.c 2008-05-19 00:33:13.728937491 +0300
@@ -531,6 +531,17 @@
* already allocated.
*/
for (pdn = dn; pdn && PCI_DN(pdn) && !PCI_DN(pdn)->iommu_table;
pdn = pdn->parent) {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/balloon.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/balloon.c
---- linux-2.6.18/arch/powerpc/platforms/xen/balloon.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/balloon.c 2007-12-23 11:14:59.186104615 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/Makefile 2008-05-19 00:33:13.804941872 +0300
+@@ -0,0 +1,20 @@
++obj-y += gnttab.o
++obj-y += hcall.o
++obj-y += reboot.o
++obj-y += setup.o
++obj-y += smp.o
++obj-y += time.o
++obj-y += udbg_xen.o
++obj-y += xen_guest.o
++obj-y += xencomm.o
++
++# we need the latest __XEN_INTERFACE_VERSION__ (see xen-compat.h)
++CFLAGS_hcall.o += -D__XEN_TOOLS__
++
++ifndef CONFIG_XEN_BALLOON
++obj-y += balloon.o
++endif
++
++ifndef CONFIG_XEN_UTIL
++obj-y += util.o
++endif
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/balloon.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/balloon.c 2008-05-19 00:33:13.804941872 +0300
@@ -0,0 +1,82 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+EXPORT_SYMBOL_GPL(alloc_empty_pages_and_pagevec);
+EXPORT_SYMBOL_GPL(free_empty_pages_and_pagevec);
+EXPORT_SYMBOL_GPL(balloon_release_driver_page);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/gnttab.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/gnttab.c
---- linux-2.6.18/arch/powerpc/platforms/xen/gnttab.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/gnttab.c 2007-12-23 11:14:59.186104615 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/gnttab.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/gnttab.c 2008-05-19 00:33:13.804941872 +0300
@@ -0,0 +1,468 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+
+ return shared;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/hcall.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/hcall.c
---- linux-2.6.18/arch/powerpc/platforms/xen/hcall.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/hcall.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/hcall.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/hcall.c 2008-05-19 00:33:13.804941872 +0300
@@ -0,0 +1,891 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+
+ return rc;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/Makefile linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/Makefile
---- linux-2.6.18/arch/powerpc/platforms/xen/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/Makefile 2007-12-23 11:14:59.186104615 +0100
-@@ -0,0 +1,20 @@
-+obj-y += gnttab.o
-+obj-y += hcall.o
-+obj-y += reboot.o
-+obj-y += setup.o
-+obj-y += smp.o
-+obj-y += time.o
-+obj-y += udbg_xen.o
-+obj-y += xen_guest.o
-+obj-y += xencomm.o
-+
-+# we need the latest __XEN_INTERFACE_VERSION__ (see xen-compat.h)
-+CFLAGS_hcall.o += -D__XEN_TOOLS__
-+
-+ifndef CONFIG_XEN_BALLOON
-+obj-y += balloon.o
-+endif
-+
-+ifndef CONFIG_XEN_UTIL
-+obj-y += util.o
-+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/reboot.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/reboot.c
---- linux-2.6.18/arch/powerpc/platforms/xen/reboot.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/reboot.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/reboot.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/reboot.c 2008-05-19 00:33:13.808942103 +0300
@@ -0,0 +1,53 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ ppc_md.halt = domain_machine_power_off;
+ }
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/setup.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.c
---- linux-2.6.18/arch/powerpc/platforms/xen/setup.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/setup.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.c 2008-05-19 00:33:13.808942103 +0300
@@ -0,0 +1,336 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ .machine_crash_shutdown = xen_machine_crash_shutdown,
+#endif
+};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/setup.h linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.h
---- linux-2.6.18/arch/powerpc/platforms/xen/setup.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.h 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/setup.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/setup.h 2008-05-19 00:33:13.808942103 +0300
@@ -0,0 +1,47 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+
+extern void __init xen_setup_time(struct machdep_calls *host_md);
+extern void xen_setup_smp(void);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/smp.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/smp.c
---- linux-2.6.18/arch/powerpc/platforms/xen/smp.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/smp.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/smp.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/smp.c 2008-05-19 00:33:13.848944409 +0300
@@ -0,0 +1,444 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ xen_boot_secondary_vcpus();
+ smp_release_cpus();
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/time.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/time.c
---- linux-2.6.18/arch/powerpc/platforms/xen/time.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/time.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/time.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/time.c 2008-05-19 00:33:13.848944409 +0300
@@ -0,0 +1,114 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ ppc_md.get_rtc_time = xen_get_rtc_time;
+ host_md_get_rtc_time = host_md->get_rtc_time;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/udbg_xen.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/udbg_xen.c
---- linux-2.6.18/arch/powerpc/platforms/xen/udbg_xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/udbg_xen.c 2007-12-23 11:14:59.189438125 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/udbg_xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/udbg_xen.c 2008-05-19 00:33:13.852944639 +0300
@@ -0,0 +1,164 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ intf = (struct xencons_interface *)mfn_to_virt(__console_mfn);
+ }
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/util.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/util.c
---- linux-2.6.18/arch/powerpc/platforms/xen/util.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/util.c 2007-12-23 11:14:59.192771633 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/util.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/util.c 2008-05-19 00:33:13.852944639 +0300
@@ -0,0 +1,70 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ preempt_enable();
+}
+EXPORT_SYMBOL_GPL(unlock_vm_area);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/xencomm.c linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xencomm.c
---- linux-2.6.18/arch/powerpc/platforms/xen/xencomm.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xencomm.c 2007-12-23 11:14:59.192771633 +0100
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/xen_guest.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xen_guest.S 2008-05-19 00:33:13.852944639 +0300
+@@ -0,0 +1,27 @@
++/*
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ * Copyright (C) IBM Corp. 2006
++ *
++ * Authors: Jimi Xenidis <jimix@watson.ibm.com>
++ */
++
++ .section __xen_guest
++ .ascii "GUEST_OS=linux"
++ .ascii ",GUEST_VER=xen-3.0"
++ .ascii ",XEN_VER=xen-3.0"
++ .ascii ",VIRT_BASE=0xC000000000000000"
++ .ascii ",LOADER=generic"
++ .byte 0
+--- linux-2.6.18.8/arch/powerpc/platforms/xen/xencomm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xencomm.c 2008-05-19 00:33:13.852944639 +0300
@@ -0,0 +1,54 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+
+ return (page_to_pfn(page) << PAGE_SHIFT) | (vaddr & ~PAGE_MASK);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/platforms/xen/xen_guest.S linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xen_guest.S
---- linux-2.6.18/arch/powerpc/platforms/xen/xen_guest.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/powerpc/platforms/xen/xen_guest.S 2007-12-23 11:14:59.192771633 +0100
-@@ -0,0 +1,27 @@
-+/*
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License as published by
-+ * the Free Software Foundation; either version 2 of the License, or
-+ * (at your option) any later version.
-+ *
-+ * This program is distributed in the hope that it will be useful,
-+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
-+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-+ * GNU General Public License for more details.
-+ *
-+ * You should have received a copy of the GNU General Public License
-+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+ *
-+ * Copyright (C) IBM Corp. 2006
-+ *
-+ * Authors: Jimi Xenidis <jimix@watson.ibm.com>
-+ */
-+
-+ .section __xen_guest
-+ .ascii "GUEST_OS=linux"
-+ .ascii ",GUEST_VER=xen-3.0"
-+ .ascii ",XEN_VER=xen-3.0"
-+ .ascii ",VIRT_BASE=0xC000000000000000"
-+ .ascii ",LOADER=generic"
-+ .byte 0
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/sysdev/mpic.c linux-2.6.18-xen.hg/arch/powerpc/sysdev/mpic.c
---- linux-2.6.18/arch/powerpc/sysdev/mpic.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/sysdev/mpic.c 2007-12-23 11:14:59.199438646 +0100
+--- linux-2.6.18.8/arch/powerpc/sysdev/mpic.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/sysdev/mpic.c 2008-05-19 00:33:13.904947637 +0300
@@ -765,6 +765,9 @@
else if (hw >= MPIC_VEC_IPI_0) {
WARN_ON(!(mpic->flags & MPIC_PRIMARY));
/* Initialize interrupt sources */
if (mpic->irq_count == 0)
mpic->irq_count = mpic->num_sources;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/powerpc/xmon/xmon.c linux-2.6.18-xen.hg/arch/powerpc/xmon/xmon.c
---- linux-2.6.18/arch/powerpc/xmon/xmon.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/powerpc/xmon/xmon.c 2007-12-23 11:14:59.206105666 +0100
+--- linux-2.6.18.8/arch/powerpc/xmon/xmon.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/powerpc/xmon/xmon.c 2008-05-19 00:33:13.916948328 +0300
@@ -752,6 +752,9 @@
cmd = inchar();
}
case 'm':
cmd = inchar();
switch (cmd) {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/boot/Makefile linux-2.6.18-xen.hg/arch/x86_64/boot/Makefile
---- linux-2.6.18/arch/x86_64/boot/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/boot/Makefile 2007-12-23 11:15:00.042816270 +0100
+--- linux-2.6.18.8/arch/x86_64/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/Kconfig 2008-05-19 00:33:18.825231243 +0300
+@@ -135,6 +135,23 @@
+
+ endchoice
+
++config X86_64_XEN
++ bool "Enable Xen compatible kernel"
++ select XEN
++ select SWIOTLB
++ help
++ This option will compile a kernel compatible with Xen hypervisor
++
++config X86_NO_TSS
++ bool
++ depends on X86_64_XEN
++ default y
++
++config X86_NO_IDT
++ bool
++ depends on X86_64_XEN
++ default y
++
+ #
+ # Define implied options from the CPU selection here
+ #
+@@ -155,6 +172,7 @@
+
+ config X86_TSC
+ bool
++ depends on !X86_64_XEN
+ default y
+
+ config X86_GOOD_APIC
+@@ -197,7 +215,7 @@
+
+ config X86_HT
+ bool
+- depends on SMP && !MK8
++ depends on SMP && !MK8 && !X86_64_XEN
+ default y
+
+ config MATH_EMULATION
+@@ -211,14 +229,22 @@
+
+ config X86_IO_APIC
+ bool
++ depends !XEN_UNPRIVILEGED_GUEST
++ default y
++
++config X86_XEN_GENAPIC
++ bool
++ depends on X86_64_XEN
+ default y
+
+ config X86_LOCAL_APIC
+ bool
++ depends !XEN_UNPRIVILEGED_GUEST
+ default y
+
+ config MTRR
+ bool "MTRR (Memory Type Range Register) support"
++ depends on !XEN_UNPRIVILEGED_GUEST
+ ---help---
+ On Intel P6 family processors (Pentium Pro, Pentium II and later)
+ the Memory Type Range Registers (MTRRs) may be used to control
+@@ -259,7 +285,7 @@
+
+ config SCHED_SMT
+ bool "SMT (Hyperthreading) scheduler support"
+- depends on SMP
++ depends on SMP && !X86_64_XEN
+ default n
+ help
+ SMT scheduler support improves the CPU scheduler's decision making
+@@ -269,7 +295,7 @@
+
+ config SCHED_MC
+ bool "Multi-core scheduler support"
+- depends on SMP
++ depends on SMP && !X86_64_XEN
+ default y
+ help
+ Multi-core scheduler support improves the CPU scheduler's decision
+@@ -280,7 +306,7 @@
+
+ config NUMA
+ bool "Non Uniform Memory Access (NUMA) Support"
+- depends on SMP
++ depends on SMP && !X86_64_XEN
+ help
+ Enable NUMA (Non Uniform Memory Access) support. The kernel
+ will try to allocate memory used by a CPU on the local memory
+@@ -341,7 +367,7 @@
+
+ config ARCH_SPARSEMEM_ENABLE
+ def_bool y
+- depends on (NUMA || EXPERIMENTAL)
++ depends on (NUMA || EXPERIMENTAL) && !X86_64_XEN
+
+ config ARCH_MEMORY_PROBE
+ def_bool y
+@@ -365,6 +391,7 @@
+ int "Maximum number of CPUs (2-256)"
+ range 2 255
+ depends on SMP
++ default "16" if X86_64_XEN
+ default "8"
+ help
+ This allows you to specify the maximum number of CPUs which this
+@@ -387,6 +414,7 @@
+
+ config HPET_TIMER
+ bool
++ depends on !X86_64_XEN
+ default y
+ help
+ Use the IA-PC HPET (High Precision Event Timer) to manage
+@@ -407,7 +435,7 @@
+ default y
+ select SWIOTLB
+ select AGP
+- depends on PCI
++ depends on PCI && !X86_64_XEN
+ help
+ Support for full DMA access of devices with 32bit memory access only
+ on systems with more than 3GB. This is usually needed for USB,
+@@ -423,7 +451,7 @@
+ bool "IBM Calgary IOMMU support"
+ default y
+ select SWIOTLB
+- depends on PCI && EXPERIMENTAL
++ depends on PCI && !X86_64_XEN && EXPERIMENTAL
+ help
+ Support for hardware IOMMUs in IBM's xSeries x366 and x460
+ systems. Needed to run systems with more than 3GB of memory
+@@ -444,6 +472,7 @@
+
+ config X86_MCE
+ bool "Machine check support" if EMBEDDED
++ depends on !X86_64_XEN
+ default y
+ help
+ Include a machine check error handler to report hardware errors.
+@@ -469,7 +498,7 @@
+
+ config KEXEC
+ bool "kexec system call (EXPERIMENTAL)"
+- depends on EXPERIMENTAL
++ depends on EXPERIMENTAL && !XEN_UNPRIVILEGED_GUEST
+ help
+ kexec is a system call that implements the ability to shutdown your
+ current kernel, and to start another kernel. It is like a reboot
+@@ -564,8 +593,9 @@
+ default y
+
+ menu "Power management options"
++ depends on !XEN_UNPRIVILEGED_GUEST
+
+-source kernel/power/Kconfig
++source "kernel/power/Kconfig"
+
+ source "drivers/acpi/Kconfig"
+
+@@ -588,6 +618,22 @@
+ bool "Support mmconfig PCI config space access"
+ depends on PCI && ACPI
+
++config XEN_PCIDEV_FRONTEND
++ bool "Xen PCI Frontend"
++ depends on PCI && X86_64_XEN
++ select HOTPLUG
++ default y
++ help
++ The PCI device frontend driver allows the kernel to import arbitrary
++ PCI devices from a PCI backend to support PCI driver domains.
++
++config XEN_PCIDEV_FE_DEBUG
++ bool "Xen PCI Frontend Debugging"
++ depends on XEN_PCIDEV_FRONTEND
++ default n
++ help
++ Enables some debug statements within the PCI Frontend.
++
+ source "drivers/pci/pcie/Kconfig"
+
+ source "drivers/pci/Kconfig"
+@@ -658,4 +704,6 @@
+
+ source "crypto/Kconfig"
+
++source "drivers/xen/Kconfig"
++
+ source "lib/Kconfig"
+--- linux-2.6.18.8/arch/x86_64/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/Makefile 2008-05-19 00:33:18.825231243 +0300
+@@ -71,9 +71,22 @@
+
+ boot := arch/x86_64/boot
+
+-PHONY += bzImage bzlilo install archmrproper \
++PHONY += bzImage bzlilo vmlinuz install archmrproper \
+ fdimage fdimage144 fdimage288 isoimage archclean
+
++ifdef CONFIG_XEN
++CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
++ -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(CPPFLAGS)
++LDFLAGS_vmlinux := -e startup_64
++#Default target when executing "make"
++all: vmlinuz
++
++BOOTIMAGE := $(boot)/vmlinuz
++KBUILD_IMAGE := $(BOOTIMAGE)
++
++vmlinuz: vmlinux
++ $(Q)$(MAKE) $(build)=$(boot) $(BOOTIMAGE)
++else
+ #Default target when executing "make"
+ all: bzImage
+
+@@ -91,6 +104,7 @@
+
+ fdimage fdimage144 fdimage288 isoimage: vmlinux
+ $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(BOOTIMAGE) $@
++endif
+
+ install:
+ $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(BOOTIMAGE) $@
+--- linux-2.6.18.8/arch/x86_64/boot/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/boot/Makefile 2008-05-19 00:33:18.829231473 +0300
@@ -26,7 +26,7 @@
#RAMDISK := -DRAMDISK=512
+
install:
sh $(srctree)/$(src)/install.sh $(KERNELRELEASE) $(BOOTIMAGE) System.map "$(INSTALL_PATH)"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/ia32entry-xen.S linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32entry-xen.S
---- linux-2.6.18/arch/x86_64/ia32/ia32entry-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32entry-xen.S 2007-12-23 11:15:00.056150298 +0100
-@@ -0,0 +1,668 @@
+--- linux-2.6.18.8/arch/x86_64/ia32/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/Makefile 2008-05-19 00:33:18.841232165 +0300
+@@ -14,11 +14,14 @@
+ audit-class-$(CONFIG_AUDIT) := audit.o
+ obj-$(CONFIG_IA32_EMULATION) += $(audit-class-y)
+
++syscall32-types-y := sysenter syscall
++syscall32-types-$(subst 1,$(CONFIG_XEN),$(shell expr $(CONFIG_XEN_COMPAT)0 '<' 0x0302000)) += int80
++
+ $(obj)/syscall32_syscall.o: \
+- $(foreach F,sysenter syscall,$(obj)/vsyscall-$F.so)
++ $(foreach F,$(syscall32-types-y),$(obj)/vsyscall-$F.so)
+
+ # Teach kbuild about targets
+-targets := $(foreach F,sysenter syscall,vsyscall-$F.o vsyscall-$F.so)
++targets := $(foreach F,$(syscall32-types-y),vsyscall-$F.o vsyscall-$F.so)
+
+ # The DSO images are built using a special linker script
+ quiet_cmd_syscall = SYSCALL $@
+@@ -27,9 +30,10 @@
+ -Wl,-soname=linux-gate.so.1 -o $@ \
+ -Wl,-T,$(filter-out FORCE,$^)
+
+-$(obj)/vsyscall-sysenter.so $(obj)/vsyscall-syscall.so: \
++$(foreach F,$(syscall32-types-y),$(obj)/vsyscall-$F.so): \
+ $(obj)/vsyscall-%.so: $(src)/vsyscall.lds $(obj)/vsyscall-%.o FORCE
+ $(call if_changed,syscall)
+
+-AFLAGS_vsyscall-sysenter.o = -m32 -Wa,-32
+-AFLAGS_vsyscall-syscall.o = -m32 -Wa,-32
++AFLAGS_vsyscall-sysenter.o = -m32 -Wa,-32 -Iarch/i386/kernel
++AFLAGS_vsyscall-syscall.o = -m32 -Wa,-32 -Iarch/i386/kernel
++AFLAGS_vsyscall-int80.o = -m32 -Wa,-32 -Iarch/i386/kernel
+--- linux-2.6.18.8/arch/x86_64/ia32/ia32_binfmt.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32_binfmt.c 2008-05-19 00:33:18.845232396 +0300
+@@ -65,55 +65,6 @@
+ #define ELF_NGREG (sizeof (struct user_regs_struct32) / sizeof(elf_greg_t))
+ typedef elf_greg_t elf_gregset_t[ELF_NGREG];
+
+-/*
+- * These macros parameterize elf_core_dump in fs/binfmt_elf.c to write out
+- * extra segments containing the vsyscall DSO contents. Dumping its
+- * contents makes post-mortem fully interpretable later without matching up
+- * the same kernel and hardware config to see what PC values meant.
+- * Dumping its extra ELF program headers includes all the other information
+- * a debugger needs to easily find how the vsyscall DSO was being used.
+- */
+-#define ELF_CORE_EXTRA_PHDRS (find_vma(current->mm, VSYSCALL32_BASE) ? \
+- (VSYSCALL32_EHDR->e_phnum) : 0)
+-#define ELF_CORE_WRITE_EXTRA_PHDRS \
+-do { \
+- if (find_vma(current->mm, VSYSCALL32_BASE)) { \
+- const struct elf32_phdr *const vsyscall_phdrs = \
+- (const struct elf32_phdr *) (VSYSCALL32_BASE \
+- + VSYSCALL32_EHDR->e_phoff);\
+- int i; \
+- Elf32_Off ofs = 0; \
+- for (i = 0; i < VSYSCALL32_EHDR->e_phnum; ++i) { \
+- struct elf32_phdr phdr = vsyscall_phdrs[i]; \
+- if (phdr.p_type == PT_LOAD) { \
+- BUG_ON(ofs != 0); \
+- ofs = phdr.p_offset = offset; \
+- phdr.p_memsz = PAGE_ALIGN(phdr.p_memsz); \
+- phdr.p_filesz = phdr.p_memsz; \
+- offset += phdr.p_filesz; \
+- } \
+- else \
+- phdr.p_offset += ofs; \
+- phdr.p_paddr = 0; /* match other core phdrs */ \
+- DUMP_WRITE(&phdr, sizeof(phdr)); \
+- } \
+- } \
+-} while (0)
+-#define ELF_CORE_WRITE_EXTRA_DATA \
+-do { \
+- if (find_vma(current->mm, VSYSCALL32_BASE)) { \
+- const struct elf32_phdr *const vsyscall_phdrs = \
+- (const struct elf32_phdr *) (VSYSCALL32_BASE \
+- + VSYSCALL32_EHDR->e_phoff); \
+- int i; \
+- for (i = 0; i < VSYSCALL32_EHDR->e_phnum; ++i) { \
+- if (vsyscall_phdrs[i].p_type == PT_LOAD) \
+- DUMP_WRITE((void *) (u64) vsyscall_phdrs[i].p_vaddr,\
+- PAGE_ALIGN(vsyscall_phdrs[i].p_memsz)); \
+- } \
+- } \
+-} while (0)
+-
+ struct elf_siginfo
+ {
+ int si_signo; /* signal number */
+--- linux-2.6.18.8/arch/x86_64/ia32/ia32_signal.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32_signal.c 2008-05-19 00:33:18.845232396 +0300
+@@ -113,25 +113,19 @@
+ }
+
+ asmlinkage long
+-sys32_sigsuspend(int history0, int history1, old_sigset_t mask,
+- struct pt_regs *regs)
++sys32_sigsuspend(int history0, int history1, old_sigset_t mask)
+ {
+- sigset_t saveset;
+-
+ mask &= _BLOCKABLE;
+ spin_lock_irq(¤t->sighand->siglock);
+- saveset = current->blocked;
++ current->saved_sigmask = current->blocked;
+ siginitset(¤t->blocked, mask);
+ recalc_sigpending();
+ spin_unlock_irq(¤t->sighand->siglock);
+
+- regs->rax = -EINTR;
+- while (1) {
+ current->state = TASK_INTERRUPTIBLE;
+ schedule();
+- if (do_signal(regs, &saveset))
+- return -EINTR;
+- }
++ set_thread_flag(TIF_RESTORE_SIGMASK);
++ return -ERESTARTNOHAND;
+ }
+
+ asmlinkage long
+@@ -508,11 +502,11 @@
+ current->comm, current->pid, frame, regs->rip, frame->pretcode);
+ #endif
+
+- return 1;
++ return 0;
+
+ give_sigsegv:
+ force_sigsegv(sig, current);
+- return 0;
++ return -EFAULT;
+ }
+
+ int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
+@@ -604,9 +598,9 @@
+ current->comm, current->pid, frame, regs->rip, frame->pretcode);
+ #endif
+
+- return 1;
++ return 0;
+
+ give_sigsegv:
+ force_sigsegv(sig, current);
+- return 0;
++ return -EFAULT;
+ }
+--- linux-2.6.18.8/arch/x86_64/ia32/ia32entry-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32entry-xen.S 2008-05-19 00:33:18.849232626 +0300
+@@ -0,0 +1,666 @@
+/*
+ * Compatibility mode system call entry point for x86-64.
+ *
+ movl \offset+72(%rsp),%eax
+ .endm
+
-+#include "../kernel/xen_entry.S"
-+
+ .macro CFI_STARTPROC32 simple
+ CFI_STARTPROC \simple
+ CFI_UNDEFINED r8
+ .quad sys_readlinkat /* 305 */
+ .quad sys_fchmodat
+ .quad sys_faccessat
-+ .quad quiet_ni_syscall /* pselect6 for now */
-+ .quad quiet_ni_syscall /* ppoll for now */
++ .quad compat_sys_pselect6
++ .quad compat_sys_ppoll
+ .quad sys_unshare /* 310 */
+ .quad compat_sys_set_robust_list
+ .quad compat_sys_get_robust_list
+ .quad compat_sys_vmsplice
+ .quad compat_sys_move_pages
+ia32_syscall_end:
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/Makefile linux-2.6.18-xen.hg/arch/x86_64/ia32/Makefile
---- linux-2.6.18/arch/x86_64/ia32/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/Makefile 2007-12-23 11:15:00.052816796 +0100
-@@ -14,11 +14,14 @@
- audit-class-$(CONFIG_AUDIT) := audit.o
- obj-$(CONFIG_IA32_EMULATION) += $(audit-class-y)
-
-+syscall32-types-y := sysenter syscall
-+syscall32-types-$(subst 1,$(CONFIG_XEN),$(shell expr $(CONFIG_XEN_COMPAT)0 '<' 0x0302000)) += int80
-+
- $(obj)/syscall32_syscall.o: \
-- $(foreach F,sysenter syscall,$(obj)/vsyscall-$F.so)
-+ $(foreach F,$(syscall32-types-y),$(obj)/vsyscall-$F.so)
-
- # Teach kbuild about targets
--targets := $(foreach F,sysenter syscall,vsyscall-$F.o vsyscall-$F.so)
-+targets := $(foreach F,$(syscall32-types-y),vsyscall-$F.o vsyscall-$F.so)
-
- # The DSO images are built using a special linker script
- quiet_cmd_syscall = SYSCALL $@
-@@ -27,9 +30,10 @@
- -Wl,-soname=linux-gate.so.1 -o $@ \
- -Wl,-T,$(filter-out FORCE,$^)
-
--$(obj)/vsyscall-sysenter.so $(obj)/vsyscall-syscall.so: \
-+$(foreach F,$(syscall32-types-y),$(obj)/vsyscall-$F.so): \
- $(obj)/vsyscall-%.so: $(src)/vsyscall.lds $(obj)/vsyscall-%.o FORCE
- $(call if_changed,syscall)
-
--AFLAGS_vsyscall-sysenter.o = -m32 -Wa,-32
--AFLAGS_vsyscall-syscall.o = -m32 -Wa,-32
-+AFLAGS_vsyscall-sysenter.o = -m32 -Wa,-32 -Iarch/i386/kernel
-+AFLAGS_vsyscall-syscall.o = -m32 -Wa,-32 -Iarch/i386/kernel
-+AFLAGS_vsyscall-int80.o = -m32 -Wa,-32 -Iarch/i386/kernel
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/syscall32_syscall-xen.S linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32_syscall-xen.S
---- linux-2.6.18/arch/x86_64/ia32/syscall32_syscall-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32_syscall-xen.S 2007-12-23 11:15:00.059483805 +0100
-@@ -0,0 +1,28 @@
-+/* 32bit VDSOs mapped into user space. */
-+
-+ .section ".init.data","aw"
-+
-+#if CONFIG_XEN_COMPAT < 0x030200
-+
-+ .globl syscall32_int80
-+ .globl syscall32_int80_end
-+
-+syscall32_int80:
-+ .incbin "arch/x86_64/ia32/vsyscall-int80.so"
-+syscall32_int80_end:
-+
-+#endif
-+
-+ .globl syscall32_syscall
-+ .globl syscall32_syscall_end
-+
-+syscall32_syscall:
-+ .incbin "arch/x86_64/ia32/vsyscall-syscall.so"
-+syscall32_syscall_end:
-+
-+ .globl syscall32_sysenter
-+ .globl syscall32_sysenter_end
-+
-+syscall32_sysenter:
-+ .incbin "arch/x86_64/ia32/vsyscall-sysenter.so"
-+syscall32_sysenter_end:
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/syscall32-xen.c linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32-xen.c
---- linux-2.6.18/arch/x86_64/ia32/syscall32-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32-xen.c 2007-12-23 11:15:00.059483805 +0100
+--- linux-2.6.18.8/arch/x86_64/ia32/ia32entry.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/ia32entry.S 2008-05-19 00:33:18.869233779 +0300
+@@ -703,8 +703,8 @@
+ .quad sys_readlinkat /* 305 */
+ .quad sys_fchmodat
+ .quad sys_faccessat
+- .quad quiet_ni_syscall /* pselect6 for now */
+- .quad quiet_ni_syscall /* ppoll for now */
++ .quad compat_sys_pselect6
++ .quad compat_sys_ppoll
+ .quad sys_unshare /* 310 */
+ .quad compat_sys_set_robust_list
+ .quad compat_sys_get_robust_list
+--- linux-2.6.18.8/arch/x86_64/ia32/syscall32-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32-xen.c 2008-05-19 00:33:18.873234010 +0300
@@ -0,0 +1,137 @@
+/* Copyright 2002,2003 Andi Kleen, SuSE Labs */
+
+/* May not be __init: called during resume */
+void syscall32_cpu_init(void)
+{
-+ static struct callback_register cstar = {
++ static const struct callback_register cstar = {
+ .type = CALLBACKTYPE_syscall32,
+ .address = (unsigned long)ia32_cstar_target
+ };
-+ static struct callback_register sysenter = {
++ static const struct callback_register sysenter = {
+ .type = CALLBACKTYPE_sysenter,
+ .address = (unsigned long)ia32_sysenter_target
+ };
+ if (use_sysenter < 0)
+ use_sysenter = (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/vsyscall-int80.S linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-int80.S
---- linux-2.6.18/arch/x86_64/ia32/vsyscall-int80.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-int80.S 2007-12-23 11:15:00.059483805 +0100
+--- linux-2.6.18.8/arch/x86_64/ia32/syscall32.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32.c 2008-05-19 00:33:18.877234240 +0300
+@@ -59,6 +59,13 @@
+ vma->vm_end = VSYSCALL32_END;
+ /* MAYWRITE to allow gdb to COW and set breakpoints */
+ vma->vm_flags = VM_READ|VM_EXEC|VM_MAYREAD|VM_MAYEXEC|VM_MAYWRITE;
++ /*
++ * Make sure the vDSO gets into every core dump.
++ * Dumping its contents makes post-mortem fully interpretable later
++ * without matching up the same kernel and hardware config to see
++ * what PC values meant.
++ */
++ vma->vm_flags |= VM_ALWAYSDUMP;
+ vma->vm_flags |= mm->def_flags;
+ vma->vm_page_prot = protection_map[vma->vm_flags & 7];
+ vma->vm_ops = &syscall32_vm_ops;
+--- linux-2.6.18.8/arch/x86_64/ia32/syscall32_syscall-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/syscall32_syscall-xen.S 2008-05-19 00:33:18.877234240 +0300
+@@ -0,0 +1,28 @@
++/* 32bit VDSOs mapped into user space. */
++
++ .section ".init.data","aw"
++
++#if CONFIG_XEN_COMPAT < 0x030200
++
++ .globl syscall32_int80
++ .globl syscall32_int80_end
++
++syscall32_int80:
++ .incbin "arch/x86_64/ia32/vsyscall-int80.so"
++syscall32_int80_end:
++
++#endif
++
++ .globl syscall32_syscall
++ .globl syscall32_syscall_end
++
++syscall32_syscall:
++ .incbin "arch/x86_64/ia32/vsyscall-syscall.so"
++syscall32_syscall_end:
++
++ .globl syscall32_sysenter
++ .globl syscall32_sysenter_end
++
++syscall32_sysenter:
++ .incbin "arch/x86_64/ia32/vsyscall-sysenter.so"
++syscall32_sysenter_end:
+--- linux-2.6.18.8/arch/x86_64/ia32/vsyscall-int80.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-int80.S 2008-05-19 00:33:18.877234240 +0300
@@ -0,0 +1,58 @@
+/*
+ * Code for the vsyscall page. This version uses the old int $0x80 method.
+ */
+#define SYSCALL_ENTER_KERNEL int $0x80
+#include "vsyscall-sigreturn.S"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/ia32/vsyscall-sigreturn.S linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-sigreturn.S
---- linux-2.6.18/arch/x86_64/ia32/vsyscall-sigreturn.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-sigreturn.S 2007-12-23 11:15:00.059483805 +0100
+--- linux-2.6.18.8/arch/x86_64/ia32/vsyscall-sigreturn.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/ia32/vsyscall-sigreturn.S 2008-05-19 00:33:18.877234240 +0300
@@ -139,5 +139,5 @@
.align 4
.LENDFDE3:
-#include "../../i386/kernel/vsyscall-note.S"
+#include <vsyscall-note.S>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/Kconfig linux-2.6.18-xen.hg/arch/x86_64/Kconfig
---- linux-2.6.18/arch/x86_64/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/Kconfig 2007-12-23 11:15:00.039482758 +0100
-@@ -135,6 +135,22 @@
-
- endchoice
-
-+config X86_64_XEN
-+ bool "Enable Xen compatible kernel"
-+ select SWIOTLB
-+ help
-+ This option will compile a kernel compatible with Xen hypervisor
-+
-+config X86_NO_TSS
-+ bool
-+ depends on X86_64_XEN
-+ default y
-+
-+config X86_NO_IDT
-+ bool
-+ depends on X86_64_XEN
-+ default y
-+
- #
- # Define implied options from the CPU selection here
- #
-@@ -155,6 +171,7 @@
-
- config X86_TSC
- bool
-+ depends on !X86_64_XEN
- default y
-
- config X86_GOOD_APIC
-@@ -197,7 +214,7 @@
-
- config X86_HT
- bool
-- depends on SMP && !MK8
-+ depends on SMP && !MK8 && !X86_64_XEN
- default y
-
- config MATH_EMULATION
-@@ -211,14 +228,22 @@
-
- config X86_IO_APIC
- bool
-+ depends !XEN_UNPRIVILEGED_GUEST
-+ default y
-+
-+config X86_XEN_GENAPIC
-+ bool
-+ depends on X86_64_XEN
- default y
-
- config X86_LOCAL_APIC
- bool
-+ depends !XEN_UNPRIVILEGED_GUEST
- default y
-
- config MTRR
- bool "MTRR (Memory Type Range Register) support"
-+ depends on !XEN_UNPRIVILEGED_GUEST
- ---help---
- On Intel P6 family processors (Pentium Pro, Pentium II and later)
- the Memory Type Range Registers (MTRRs) may be used to control
-@@ -259,7 +284,7 @@
-
- config SCHED_SMT
- bool "SMT (Hyperthreading) scheduler support"
-- depends on SMP
-+ depends on SMP && !X86_64_XEN
- default n
- help
- SMT scheduler support improves the CPU scheduler's decision making
-@@ -269,7 +294,7 @@
-
- config SCHED_MC
- bool "Multi-core scheduler support"
-- depends on SMP
-+ depends on SMP && !X86_64_XEN
- default y
- help
- Multi-core scheduler support improves the CPU scheduler's decision
-@@ -280,7 +305,7 @@
-
- config NUMA
- bool "Non Uniform Memory Access (NUMA) Support"
-- depends on SMP
-+ depends on SMP && !X86_64_XEN
- help
- Enable NUMA (Non Uniform Memory Access) support. The kernel
- will try to allocate memory used by a CPU on the local memory
-@@ -341,7 +366,7 @@
-
- config ARCH_SPARSEMEM_ENABLE
- def_bool y
-- depends on (NUMA || EXPERIMENTAL)
-+ depends on (NUMA || EXPERIMENTAL) && !X86_64_XEN
-
- config ARCH_MEMORY_PROBE
- def_bool y
-@@ -365,6 +390,7 @@
- int "Maximum number of CPUs (2-256)"
- range 2 255
- depends on SMP
-+ default "16" if X86_64_XEN
- default "8"
- help
- This allows you to specify the maximum number of CPUs which this
-@@ -387,6 +413,7 @@
-
- config HPET_TIMER
- bool
-+ depends on !X86_64_XEN
- default y
- help
- Use the IA-PC HPET (High Precision Event Timer) to manage
-@@ -407,7 +434,7 @@
- default y
- select SWIOTLB
- select AGP
-- depends on PCI
-+ depends on PCI && !X86_64_XEN
- help
- Support for full DMA access of devices with 32bit memory access only
- on systems with more than 3GB. This is usually needed for USB,
-@@ -423,7 +450,7 @@
- bool "IBM Calgary IOMMU support"
- default y
- select SWIOTLB
-- depends on PCI && EXPERIMENTAL
-+ depends on PCI && !X86_64_XEN && EXPERIMENTAL
- help
- Support for hardware IOMMUs in IBM's xSeries x366 and x460
- systems. Needed to run systems with more than 3GB of memory
-@@ -444,6 +471,7 @@
-
- config X86_MCE
- bool "Machine check support" if EMBEDDED
-+ depends on !X86_64_XEN
- default y
- help
- Include a machine check error handler to report hardware errors.
-@@ -469,7 +497,7 @@
-
- config KEXEC
- bool "kexec system call (EXPERIMENTAL)"
-- depends on EXPERIMENTAL
-+ depends on EXPERIMENTAL && !XEN_UNPRIVILEGED_GUEST
- help
- kexec is a system call that implements the ability to shutdown your
- current kernel, and to start another kernel. It is like a reboot
-@@ -564,8 +592,9 @@
- default y
-
- menu "Power management options"
-+ depends on !XEN_UNPRIVILEGED_GUEST
-
--source kernel/power/Kconfig
-+source "kernel/power/Kconfig"
-
- source "drivers/acpi/Kconfig"
-
-@@ -588,6 +617,21 @@
- bool "Support mmconfig PCI config space access"
- depends on PCI && ACPI
-
-+config XEN_PCIDEV_FRONTEND
-+ bool "Xen PCI Frontend"
-+ depends on PCI && X86_64_XEN
-+ default y
-+ help
-+ The PCI device frontend driver allows the kernel to import arbitrary
-+ PCI devices from a PCI backend to support PCI driver domains.
-+
-+config XEN_PCIDEV_FE_DEBUG
-+ bool "Xen PCI Frontend Debugging"
-+ depends on XEN_PCIDEV_FRONTEND
-+ default n
-+ help
-+ Enables some debug statements within the PCI Frontend.
-+
- source "drivers/pci/pcie/Kconfig"
-
- source "drivers/pci/Kconfig"
-@@ -658,4 +702,6 @@
-
- source "crypto/Kconfig"
+--- linux-2.6.18.8/arch/x86_64/kernel/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/Makefile 2008-05-19 00:33:18.921236777 +0300
+@@ -21,11 +21,13 @@
+ obj-$(CONFIG_X86_CPUID) += cpuid.o
+ obj-$(CONFIG_SMP) += smp.o smpboot.o trampoline.o
+ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o
++obj-$(CONFIG_X86_XEN_GENAPIC) += genapic.o genapic_xen.o
+ obj-$(CONFIG_X86_IO_APIC) += io_apic.o mpparse.o \
+ genapic.o genapic_cluster.o genapic_flat.o
+ obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
+ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
+-obj-$(CONFIG_PM) += suspend.o
++obj-$(CONFIG_SOFTWARE_SUSPEND) += suspend.o
++obj-$(CONFIG_ACPI_SLEEP) += suspend.o
+ obj-$(CONFIG_SOFTWARE_SUSPEND) += suspend_asm.o
+ obj-$(CONFIG_CPU_FREQ) += cpufreq/
+ obj-$(CONFIG_EARLY_PRINTK) += early_printk.o
+@@ -55,3 +57,8 @@
+ msr-$(subst m,y,$(CONFIG_X86_MSR)) += ../../i386/kernel/msr.o
+ alternative-y += ../../i386/kernel/alternative.o
-+source "drivers/xen/Kconfig"
++time-$(CONFIG_XEN) += ../../i386/kernel/time.o
++pci-dma-$(CONFIG_XEN) += ../../i386/kernel/pci-dma.o
+
- source "lib/Kconfig"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/acpi/Makefile linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/Makefile
---- linux-2.6.18/arch/x86_64/kernel/acpi/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/Makefile 2007-12-23 11:15:00.062817316 +0100
-@@ -5,5 +5,7 @@
++disabled-obj-$(CONFIG_XEN) := i8259.o reboot.o smpboot.o trampoline.o
++%/head.o %/head.s: $(if $(CONFIG_XEN),EXTRA_AFLAGS,dummy) :=
+--- linux-2.6.18.8/arch/x86_64/kernel/acpi/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/Makefile 2008-05-19 00:33:18.921236777 +0300
+@@ -5,5 +5,9 @@
ifneq ($(CONFIG_ACPI_PROCESSOR),)
obj-y += processor.o
processor-y := ../../../i386/kernel/acpi/processor.o ../../../i386/kernel/acpi/cstate.o
+processor-$(CONFIG_XEN) := ../../../i386/kernel/acpi/processor.o
endif
++obj-$(CONFIG_XEN) += processor_extcnt_xen.o
++processor_extcnt_xen-$(CONFIG_XEN) := ../../../i386/kernel/acpi/processor_extcntl_xen.o
+disabled-obj-$(CONFIG_XEN) := wakeup.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/acpi/sleep-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/sleep-xen.c
---- linux-2.6.18/arch/x86_64/kernel/acpi/sleep-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/sleep-xen.c 2007-12-23 11:15:00.066150829 +0100
-@@ -0,0 +1,167 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/acpi/sleep-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/acpi/sleep-xen.c 2008-05-19 00:33:18.921236777 +0300
+@@ -0,0 +1,146 @@
+/*
+ * acpi.c - Architecture-Specific Low-Level ACPI Support
+ *
+}
+
+__setup("acpi_sleep=", acpi_sleep_setup);
-+
-+#else /* CONFIG_ACPI_PV_SLEEP */
-+#include <asm/hypervisor.h>
-+#include <xen/interface/platform.h>
-+int acpi_notify_hypervisor_state(u8 sleep_state,
-+ u32 pm1a_cnt, u32 pm1b_cnt)
-+{
-+ struct xen_platform_op op = {
-+ .cmd = XENPF_enter_acpi_sleep,
-+ .interface_version = XENPF_INTERFACE_VERSION,
-+ .u = {
-+ .enter_acpi_sleep = {
-+ .pm1a_cnt_val = (u16)pm1a_cnt,
-+ .pm1b_cnt_val = (u16)pm1b_cnt,
-+ .sleep_state = sleep_state,
-+ },
-+ },
-+ };
-+
-+ return HYPERVISOR_platform_op(&op);
-+}
+#endif /* CONFIG_ACPI_PV_SLEEP */
+
+#endif /*CONFIG_ACPI_SLEEP */
+void acpi_pci_link_exit(void)
+{
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/apic-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/apic-xen.c
---- linux-2.6.18/arch/x86_64/kernel/apic-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/apic-xen.c 2007-12-23 11:15:00.066150829 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/apic-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/apic-xen.c 2008-05-19 00:33:18.925237007 +0300
@@ -0,0 +1,197 @@
+/*
+ * Local APIC handling, local APIC timers
+
+ return 1;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/asm-offsets.c linux-2.6.18-xen.hg/arch/x86_64/kernel/asm-offsets.c
---- linux-2.6.18/arch/x86_64/kernel/asm-offsets.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/asm-offsets.c 2007-12-23 11:15:00.069484339 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/asm-offsets.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/asm-offsets.c 2008-05-19 00:33:18.929237238 +0300
@@ -67,8 +67,10 @@
DEFINE(pbe_address, offsetof(struct pbe, address));
DEFINE(pbe_orig_address, offsetof(struct pbe, orig_address));
BLANK();
DEFINE(crypto_tfm_ctx_offset, offsetof(struct crypto_tfm, __crt_ctx));
return 0;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/crash.c linux-2.6.18-xen.hg/arch/x86_64/kernel/crash.c
---- linux-2.6.18/arch/x86_64/kernel/crash.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/crash.c 2007-12-23 11:15:00.069484339 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/crash.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/crash.c 2008-05-19 00:33:18.933237468 +0300
@@ -92,6 +92,7 @@
crash_save_this_cpu(regs, cpu);
}
+#endif /* CONFIG_XEN */
crash_save_self(regs);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/e820.c linux-2.6.18-xen.hg/arch/x86_64/kernel/e820.c
---- linux-2.6.18/arch/x86_64/kernel/e820.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/e820.c 2007-12-23 11:15:00.072817848 +0100
-@@ -93,7 +93,7 @@
- * This function checks if any part of the range <start,end> is mapped
- * with type.
- */
--int __meminit
-+int
- e820_any_mapped(unsigned long start, unsigned long end, unsigned type)
- {
- int i;
-@@ -107,6 +107,7 @@
- }
- return 0;
- }
-+EXPORT_SYMBOL_GPL(e820_any_mapped);
-
- /*
- * This function checks if the entire range <start,end> is mapped with type.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/e820-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/e820-xen.c
---- linux-2.6.18/arch/x86_64/kernel/e820-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/e820-xen.c 2007-12-23 11:15:00.069484339 +0100
-@@ -0,0 +1,783 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/e820-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/e820-xen.c 2008-05-19 00:33:18.933237468 +0300
+@@ -0,0 +1,798 @@
+/*
+ * Handle the memory map.
+ * The functions here do the job until bootmem takes over.
+
+extern struct resource code_resource, data_resource;
+
++#ifdef CONFIG_XEN
++extern struct e820map machine_e820;
++#endif
++
+/* Check for some hardcoded bad areas that early boot is not allowed to touch */
+static inline int bad_addr(unsigned long *addrp, unsigned long size)
+{
+ for (i = 0; i < e820.nr_map; i++) {
+ struct e820entry *ei = &e820.map[i];
+#else
-+ extern struct e820map machine_e820;
-+
+ if (!is_initial_xendomain())
+ return 0;
+ for (i = 0; i < machine_e820.nr_map; i++) {
+ for (i = 0; i < e820.nr_map; i++) {
+ struct e820entry *ei = &e820.map[i];
+#else
-+ extern struct e820map machine_e820;
-+
+ if (!is_initial_xendomain())
+ return 0;
+ for (i = 0; i < machine_e820.nr_map; i++) {
+
+ add_memory_region(start, size, type);
+ } while (biosmap++,--nr_map);
++
++#ifdef CONFIG_XEN
++ if (is_initial_xendomain()) {
++ struct xen_memory_map memmap;
++
++ memmap.nr_entries = E820MAX;
++ set_xen_guest_handle(memmap.buffer, machine_e820.map);
++
++ if (HYPERVISOR_memory_op(XENMEM_machine_memory_map, &memmap))
++ BUG();
++ machine_e820.nr_map = memmap.nr_entries;
++ } else
++ machine_e820 = e820;
++#endif
++
+ return 0;
+}
+
+ printk(KERN_INFO "Allocating PCI resources starting at %lx (gap: %lx:%lx)\n",
+ pci_mem_start, gapstart, gapsize);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/early_printk-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/early_printk-xen.c
---- linux-2.6.18/arch/x86_64/kernel/early_printk-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/early_printk-xen.c 2007-12-23 11:15:00.072817848 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/e820.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/e820.c 2008-05-19 00:33:18.933237468 +0300
+@@ -93,7 +93,7 @@
+ * This function checks if any part of the range <start,end> is mapped
+ * with type.
+ */
+-int __meminit
++int
+ e820_any_mapped(unsigned long start, unsigned long end, unsigned type)
+ {
+ int i;
+@@ -107,6 +107,7 @@
+ }
+ return 0;
+ }
++EXPORT_SYMBOL_GPL(e820_any_mapped);
+
+ /*
+ * This function checks if the entire range <start,end> is mapped with type.
+--- linux-2.6.18.8/arch/x86_64/kernel/early_printk-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/early_printk-xen.c 2008-05-19 00:33:18.937237699 +0300
@@ -0,0 +1,302 @@
+#include <linux/console.h>
+#include <linux/kernel.h>
+}
+
+__setup("earlyprintk=", setup_early_printk);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/entry-xen.S linux-2.6.18-xen.hg/arch/x86_64/kernel/entry-xen.S
---- linux-2.6.18/arch/x86_64/kernel/entry-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/entry-xen.S 2007-12-23 11:15:00.072817848 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/entry-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/entry-xen.S 2008-05-19 00:33:18.937237699 +0300
@@ -0,0 +1,1322 @@
+/*
+ * linux/arch/x86_64/entry.S
+ CFI_ENDPROC
+ENDPROC(arch_unwind_init_running)
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/genapic_xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic_xen.c
---- linux-2.6.18/arch/x86_64/kernel/genapic_xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic_xen.c 2007-12-23 11:15:00.079484858 +0100
-@@ -0,0 +1,161 @@
-+/*
-+ * Copyright 2004 James Cleverdon, IBM.
-+ * Subject to the GNU Public License, v.2
-+ *
-+ * Xen APIC subarch code. Maximum 8 CPUs, logical delivery.
-+ *
-+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
-+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
-+ * James Cleverdon.
-+ *
-+ * Hacked to pieces for Xen by Chris Wright.
-+ */
-+#include <linux/threads.h>
-+#include <linux/cpumask.h>
-+#include <linux/string.h>
-+#include <linux/kernel.h>
-+#include <linux/ctype.h>
-+#include <linux/init.h>
-+#ifdef CONFIG_XEN_PRIVILEGED_GUEST
-+#include <asm/smp.h>
-+#include <asm/ipi.h>
-+#else
-+#include <asm/apic.h>
-+#include <asm/apicdef.h>
-+#include <asm/genapic.h>
-+#endif
-+#include <xen/evtchn.h>
-+
-+DECLARE_PER_CPU(int, ipi_to_irq[NR_IPIS]);
-+
-+static inline void __send_IPI_one(unsigned int cpu, int vector)
-+{
-+ int irq = per_cpu(ipi_to_irq, cpu)[vector];
-+ BUG_ON(irq < 0);
-+ notify_remote_via_irq(irq);
-+}
-+
-+void xen_send_IPI_shortcut(unsigned int shortcut, int vector, unsigned int dest)
-+{
-+ int cpu;
-+
-+ switch (shortcut) {
-+ case APIC_DEST_SELF:
-+ __send_IPI_one(smp_processor_id(), vector);
-+ break;
-+ case APIC_DEST_ALLBUT:
-+ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
-+ if (cpu == smp_processor_id())
-+ continue;
-+ if (cpu_isset(cpu, cpu_online_map)) {
-+ __send_IPI_one(cpu, vector);
-+ }
-+ }
-+ break;
-+ case APIC_DEST_ALLINC:
-+ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
-+ if (cpu_isset(cpu, cpu_online_map)) {
-+ __send_IPI_one(cpu, vector);
-+ }
-+ }
-+ break;
-+ default:
-+ printk("XXXXXX __send_IPI_shortcut %08x vector %d\n", shortcut,
-+ vector);
-+ break;
-+ }
-+}
-+
-+static cpumask_t xen_target_cpus(void)
-+{
-+ return cpu_online_map;
-+}
-+
-+/*
-+ * Set up the logical destination ID.
-+ * Do nothing, not called now.
-+ */
-+static void xen_init_apic_ldr(void)
-+{
-+ Dprintk("%s\n", __FUNCTION__);
-+ return;
-+}
-+
-+static void xen_send_IPI_allbutself(int vector)
-+{
-+ /*
-+ * if there are no other CPUs in the system then
-+ * we get an APIC send error if we try to broadcast.
-+ * thus we have to avoid sending IPIs in this case.
-+ */
-+ Dprintk("%s\n", __FUNCTION__);
-+ if (num_online_cpus() > 1)
-+ xen_send_IPI_shortcut(APIC_DEST_ALLBUT, vector, APIC_DEST_LOGICAL);
-+}
-+
-+static void xen_send_IPI_all(int vector)
-+{
-+ Dprintk("%s\n", __FUNCTION__);
-+ xen_send_IPI_shortcut(APIC_DEST_ALLINC, vector, APIC_DEST_LOGICAL);
-+}
-+
-+static void xen_send_IPI_mask(cpumask_t cpumask, int vector)
-+{
-+ unsigned long mask = cpus_addr(cpumask)[0];
-+ unsigned int cpu;
-+ unsigned long flags;
-+
-+ Dprintk("%s\n", __FUNCTION__);
-+ local_irq_save(flags);
-+ WARN_ON(mask & ~cpus_addr(cpu_online_map)[0]);
-+
-+ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
-+ if (cpu_isset(cpu, cpumask)) {
-+ __send_IPI_one(cpu, vector);
-+ }
-+ }
-+ local_irq_restore(flags);
-+}
-+
-+#ifdef CONFIG_XEN_PRIVILEGED_GUEST
-+static int xen_apic_id_registered(void)
-+{
-+ /* better be set */
-+ Dprintk("%s\n", __FUNCTION__);
-+ return physid_isset(smp_processor_id(), phys_cpu_present_map);
-+}
-+#endif
-+
-+static unsigned int xen_cpu_mask_to_apicid(cpumask_t cpumask)
-+{
-+ Dprintk("%s\n", __FUNCTION__);
-+ return cpus_addr(cpumask)[0] & APIC_ALL_CPUS;
-+}
-+
-+static unsigned int phys_pkg_id(int index_msb)
-+{
-+ u32 ebx;
-+
-+ Dprintk("%s\n", __FUNCTION__);
-+ ebx = cpuid_ebx(1);
-+ return ((ebx >> 24) & 0xFF) >> index_msb;
-+}
-+
-+struct genapic apic_xen = {
-+ .name = "xen",
-+#ifdef CONFIG_XEN_PRIVILEGED_GUEST
-+ .int_delivery_mode = dest_LowestPrio,
-+#endif
-+ .int_dest_mode = (APIC_DEST_LOGICAL != 0),
-+ .int_delivery_dest = APIC_DEST_LOGICAL | APIC_DM_LOWEST,
-+ .target_cpus = xen_target_cpus,
-+#ifdef CONFIG_XEN_PRIVILEGED_GUEST
-+ .apic_id_registered = xen_apic_id_registered,
-+#endif
-+ .init_apic_ldr = xen_init_apic_ldr,
-+ .send_IPI_all = xen_send_IPI_all,
-+ .send_IPI_allbutself = xen_send_IPI_allbutself,
-+ .send_IPI_mask = xen_send_IPI_mask,
-+ .cpu_mask_to_apicid = xen_cpu_mask_to_apicid,
-+ .phys_pkg_id = phys_pkg_id,
-+};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/genapic-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic-xen.c
---- linux-2.6.18/arch/x86_64/kernel/genapic-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic-xen.c 2007-12-23 11:15:00.076151352 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/genapic-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic-xen.c 2008-05-19 00:33:18.941237929 +0300
@@ -0,0 +1,143 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ xen_send_IPI_shortcut(APIC_DEST_SELF, vector, APIC_DEST_PHYSICAL);
+#endif
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/head64-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/head64-xen.c
---- linux-2.6.18/arch/x86_64/kernel/head64-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/head64-xen.c 2007-12-23 11:15:00.079484858 +0100
-@@ -0,0 +1,162 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/genapic_xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/genapic_xen.c 2008-05-19 00:33:18.945238160 +0300
+@@ -0,0 +1,161 @@
+/*
-+ * linux/arch/x86_64/kernel/head64.c -- prepare to run common code
++ * Copyright 2004 James Cleverdon, IBM.
++ * Subject to the GNU Public License, v.2
+ *
-+ * Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
++ * Xen APIC subarch code. Maximum 8 CPUs, logical delivery.
+ *
-+ * Jun Nakajima <jun.nakajima@intel.com>
-+ * Modified for Xen.
++ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
++ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
++ * James Cleverdon.
++ *
++ * Hacked to pieces for Xen by Chris Wright.
+ */
-+
-+#include <linux/init.h>
-+#include <linux/linkage.h>
-+#include <linux/types.h>
-+#include <linux/kernel.h>
++#include <linux/threads.h>
++#include <linux/cpumask.h>
+#include <linux/string.h>
-+#include <linux/percpu.h>
-+#include <linux/module.h>
-+
-+#include <asm/processor.h>
-+#include <asm/proto.h>
++#include <linux/kernel.h>
++#include <linux/ctype.h>
++#include <linux/init.h>
++#ifdef CONFIG_XEN_PRIVILEGED_GUEST
+#include <asm/smp.h>
-+#include <asm/bootsetup.h>
-+#include <asm/setup.h>
-+#include <asm/desc.h>
-+#include <asm/pgtable.h>
-+#include <asm/sections.h>
++#include <asm/ipi.h>
++#else
++#include <asm/apic.h>
++#include <asm/apicdef.h>
++#include <asm/genapic.h>
++#endif
++#include <xen/evtchn.h>
+
-+unsigned long start_pfn;
++DECLARE_PER_CPU(int, ipi_to_irq[NR_IPIS]);
+
-+/* Don't add a printk in there. printk relies on the PDA which is not initialized
-+ yet. */
-+#if 0
-+static void __init clear_bss(void)
++static inline void __send_IPI_one(unsigned int cpu, int vector)
+{
-+ memset(__bss_start, 0,
-+ (unsigned long) __bss_stop - (unsigned long) __bss_start);
++ int irq = per_cpu(ipi_to_irq, cpu)[vector];
++ BUG_ON(irq < 0);
++ notify_remote_via_irq(irq);
+}
-+#endif
-+
-+#define NEW_CL_POINTER 0x228 /* Relative to real mode data */
-+#define OLD_CL_MAGIC_ADDR 0x90020
-+#define OLD_CL_MAGIC 0xA33F
-+#define OLD_CL_BASE_ADDR 0x90000
-+#define OLD_CL_OFFSET 0x90022
-+
-+extern char saved_command_line[];
+
-+static void __init copy_bootdata(char *real_mode_data)
++void xen_send_IPI_shortcut(unsigned int shortcut, int vector, unsigned int dest)
+{
-+#ifndef CONFIG_XEN
-+ int new_data;
-+ char * command_line;
++ int cpu;
+
-+ memcpy(x86_boot_params, real_mode_data, BOOT_PARAM_SIZE);
-+ new_data = *(int *) (x86_boot_params + NEW_CL_POINTER);
-+ if (!new_data) {
-+ if (OLD_CL_MAGIC != * (u16 *) OLD_CL_MAGIC_ADDR) {
-+ printk("so old bootloader that it does not support commandline?!\n");
-+ return;
++ switch (shortcut) {
++ case APIC_DEST_SELF:
++ __send_IPI_one(smp_processor_id(), vector);
++ break;
++ case APIC_DEST_ALLBUT:
++ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
++ if (cpu == smp_processor_id())
++ continue;
++ if (cpu_isset(cpu, cpu_online_map)) {
++ __send_IPI_one(cpu, vector);
++ }
+ }
-+ new_data = OLD_CL_BASE_ADDR + * (u16 *) OLD_CL_OFFSET;
-+ printk("old bootloader convention, maybe loadlin?\n");
++ break;
++ case APIC_DEST_ALLINC:
++ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
++ if (cpu_isset(cpu, cpu_online_map)) {
++ __send_IPI_one(cpu, vector);
++ }
++ }
++ break;
++ default:
++ printk("XXXXXX __send_IPI_shortcut %08x vector %d\n", shortcut,
++ vector);
++ break;
+ }
-+ command_line = (char *) ((u64)(new_data));
-+ memcpy(saved_command_line, command_line, COMMAND_LINE_SIZE);
-+#else
-+ int max_cmdline;
-+
-+ if ((max_cmdline = MAX_GUEST_CMDLINE) > COMMAND_LINE_SIZE)
-+ max_cmdline = COMMAND_LINE_SIZE;
-+ memcpy(saved_command_line, xen_start_info->cmd_line, max_cmdline);
-+ saved_command_line[max_cmdline-1] = '\0';
-+#endif
-+ printk("Bootdata ok (command line is %s)\n", saved_command_line);
+}
+
-+static void __init setup_boot_cpu_data(void)
++static cpumask_t xen_target_cpus(void)
+{
-+ unsigned int dummy, eax;
-+
-+ /* get vendor info */
-+ cpuid(0, (unsigned int *)&boot_cpu_data.cpuid_level,
-+ (unsigned int *)&boot_cpu_data.x86_vendor_id[0],
-+ (unsigned int *)&boot_cpu_data.x86_vendor_id[8],
-+ (unsigned int *)&boot_cpu_data.x86_vendor_id[4]);
++ return cpu_online_map;
++}
+
-+ /* get cpu type */
-+ cpuid(1, &eax, &dummy, &dummy,
-+ (unsigned int *) &boot_cpu_data.x86_capability);
-+ boot_cpu_data.x86 = (eax >> 8) & 0xf;
-+ boot_cpu_data.x86_model = (eax >> 4) & 0xf;
-+ boot_cpu_data.x86_mask = eax & 0xf;
++/*
++ * Set up the logical destination ID.
++ * Do nothing, not called now.
++ */
++static void xen_init_apic_ldr(void)
++{
++ Dprintk("%s\n", __FUNCTION__);
++ return;
+}
+
-+#include <xen/interface/memory.h>
-+unsigned long *machine_to_phys_mapping;
-+EXPORT_SYMBOL(machine_to_phys_mapping);
-+unsigned int machine_to_phys_order;
-+EXPORT_SYMBOL(machine_to_phys_order);
++static void xen_send_IPI_allbutself(int vector)
++{
++ /*
++ * if there are no other CPUs in the system then
++ * we get an APIC send error if we try to broadcast.
++ * thus we have to avoid sending IPIs in this case.
++ */
++ Dprintk("%s\n", __FUNCTION__);
++ if (num_online_cpus() > 1)
++ xen_send_IPI_shortcut(APIC_DEST_ALLBUT, vector, APIC_DEST_LOGICAL);
++}
+
-+void __init x86_64_start_kernel(char * real_mode_data)
++static void xen_send_IPI_all(int vector)
+{
-+ struct xen_machphys_mapping mapping;
-+ unsigned long machine_to_phys_nr_ents;
-+ char *s;
-+ int i;
++ Dprintk("%s\n", __FUNCTION__);
++ xen_send_IPI_shortcut(APIC_DEST_ALLINC, vector, APIC_DEST_LOGICAL);
++}
+
-+ setup_xen_features();
++static void xen_send_IPI_mask(cpumask_t cpumask, int vector)
++{
++ unsigned long mask = cpus_addr(cpumask)[0];
++ unsigned int cpu;
++ unsigned long flags;
+
-+ xen_start_info = (struct start_info *)real_mode_data;
-+ if (!xen_feature(XENFEAT_auto_translated_physmap))
-+ phys_to_machine_mapping =
-+ (unsigned long *)xen_start_info->mfn_list;
-+ start_pfn = (__pa(xen_start_info->pt_base) >> PAGE_SHIFT) +
-+ xen_start_info->nr_pt_frames;
++ Dprintk("%s\n", __FUNCTION__);
++ local_irq_save(flags);
++ WARN_ON(mask & ~cpus_addr(cpu_online_map)[0]);
+
-+ machine_to_phys_mapping = (unsigned long *)MACH2PHYS_VIRT_START;
-+ machine_to_phys_nr_ents = MACH2PHYS_NR_ENTRIES;
-+ if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
-+ machine_to_phys_mapping = (unsigned long *)mapping.v_start;
-+ machine_to_phys_nr_ents = mapping.max_mfn + 1;
++ for (cpu = 0; cpu < NR_CPUS; ++cpu) {
++ if (cpu_isset(cpu, cpumask)) {
++ __send_IPI_one(cpu, vector);
++ }
+ }
-+ while ((1UL << machine_to_phys_order) < machine_to_phys_nr_ents )
-+ machine_to_phys_order++;
++ local_irq_restore(flags);
++}
+
-+#if 0
-+ for (i = 0; i < 256; i++)
-+ set_intr_gate(i, early_idt_handler);
-+ asm volatile("lidt %0" :: "m" (idt_descr));
++#ifdef CONFIG_XEN_PRIVILEGED_GUEST
++static int xen_apic_id_registered(void)
++{
++ /* better be set */
++ Dprintk("%s\n", __FUNCTION__);
++ return physid_isset(smp_processor_id(), phys_cpu_present_map);
++}
+#endif
+
-+ /*
-+ * This must be called really, really early:
-+ */
-+ lockdep_init();
++static unsigned int xen_cpu_mask_to_apicid(cpumask_t cpumask)
++{
++ Dprintk("%s\n", __FUNCTION__);
++ return cpus_addr(cpumask)[0] & APIC_ALL_CPUS;
++}
+
-+ for (i = 0; i < NR_CPUS; i++)
-+ cpu_pda(i) = &boot_cpu_pda[i];
++static unsigned int phys_pkg_id(int index_msb)
++{
++ u32 ebx;
+
-+ pda_init(0);
-+ copy_bootdata(real_mode_data);
-+#ifdef CONFIG_SMP
-+ cpu_set(0, cpu_online_map);
-+#endif
-+ s = strstr(saved_command_line, "earlyprintk=");
-+ if (s != NULL)
-+ setup_early_printk(strchr(s, '=') + 1);
-+#ifdef CONFIG_NUMA
-+ s = strstr(saved_command_line, "numa=");
-+ if (s != NULL)
-+ numa_setup(s+5);
++ Dprintk("%s\n", __FUNCTION__);
++ ebx = cpuid_ebx(1);
++ return ((ebx >> 24) & 0xFF) >> index_msb;
++}
++
++struct genapic apic_xen = {
++ .name = "xen",
++#ifdef CONFIG_XEN_PRIVILEGED_GUEST
++ .int_delivery_mode = dest_LowestPrio,
+#endif
-+#ifdef CONFIG_X86_IO_APIC
-+ if (strstr(saved_command_line, "disableapic"))
-+ disable_apic = 1;
++ .int_dest_mode = (APIC_DEST_LOGICAL != 0),
++ .int_delivery_dest = APIC_DEST_LOGICAL | APIC_DM_LOWEST,
++ .target_cpus = xen_target_cpus,
++#ifdef CONFIG_XEN_PRIVILEGED_GUEST
++ .apic_id_registered = xen_apic_id_registered,
+#endif
-+ /* You need early console to see that */
-+ if (__pa_symbol(&_end) >= KERNEL_TEXT_SIZE)
-+ panic("Kernel too big for kernel mapping\n");
-+
-+ setup_boot_cpu_data();
-+ start_kernel();
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/head-xen.S linux-2.6.18-xen.hg/arch/x86_64/kernel/head-xen.S
---- linux-2.6.18/arch/x86_64/kernel/head-xen.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/head-xen.S 2007-12-23 11:15:00.079484858 +0100
++ .init_apic_ldr = xen_init_apic_ldr,
++ .send_IPI_all = xen_send_IPI_all,
++ .send_IPI_allbutself = xen_send_IPI_allbutself,
++ .send_IPI_mask = xen_send_IPI_mask,
++ .cpu_mask_to_apicid = xen_cpu_mask_to_apicid,
++ .phys_pkg_id = phys_pkg_id,
++};
+--- linux-2.6.18.8/arch/x86_64/kernel/head-xen.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/head-xen.S 2008-05-19 00:33:18.945238160 +0300
@@ -0,0 +1,214 @@
+/*
+ * linux/arch/x86_64/kernel/head.S -- start in 32bit and switch to 64bit
+ ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz, "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel")
+ ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic")
+ ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/init_task.c linux-2.6.18-xen.hg/arch/x86_64/kernel/init_task.c
---- linux-2.6.18/arch/x86_64/kernel/init_task.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/init_task.c 2007-12-23 11:15:00.082818365 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/head64-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/head64-xen.c 2008-05-19 00:33:18.945238160 +0300
+@@ -0,0 +1,162 @@
++/*
++ * linux/arch/x86_64/kernel/head64.c -- prepare to run common code
++ *
++ * Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
++ *
++ * Jun Nakajima <jun.nakajima@intel.com>
++ * Modified for Xen.
++ */
++
++#include <linux/init.h>
++#include <linux/linkage.h>
++#include <linux/types.h>
++#include <linux/kernel.h>
++#include <linux/string.h>
++#include <linux/percpu.h>
++#include <linux/module.h>
++
++#include <asm/processor.h>
++#include <asm/proto.h>
++#include <asm/smp.h>
++#include <asm/bootsetup.h>
++#include <asm/setup.h>
++#include <asm/desc.h>
++#include <asm/pgtable.h>
++#include <asm/sections.h>
++
++unsigned long start_pfn;
++
++/* Don't add a printk in there. printk relies on the PDA which is not initialized
++ yet. */
++#if 0
++static void __init clear_bss(void)
++{
++ memset(__bss_start, 0,
++ (unsigned long) __bss_stop - (unsigned long) __bss_start);
++}
++#endif
++
++#define NEW_CL_POINTER 0x228 /* Relative to real mode data */
++#define OLD_CL_MAGIC_ADDR 0x90020
++#define OLD_CL_MAGIC 0xA33F
++#define OLD_CL_BASE_ADDR 0x90000
++#define OLD_CL_OFFSET 0x90022
++
++extern char saved_command_line[];
++
++static void __init copy_bootdata(char *real_mode_data)
++{
++#ifndef CONFIG_XEN
++ int new_data;
++ char * command_line;
++
++ memcpy(x86_boot_params, real_mode_data, BOOT_PARAM_SIZE);
++ new_data = *(int *) (x86_boot_params + NEW_CL_POINTER);
++ if (!new_data) {
++ if (OLD_CL_MAGIC != * (u16 *) OLD_CL_MAGIC_ADDR) {
++ printk("so old bootloader that it does not support commandline?!\n");
++ return;
++ }
++ new_data = OLD_CL_BASE_ADDR + * (u16 *) OLD_CL_OFFSET;
++ printk("old bootloader convention, maybe loadlin?\n");
++ }
++ command_line = (char *) ((u64)(new_data));
++ memcpy(saved_command_line, command_line, COMMAND_LINE_SIZE);
++#else
++ int max_cmdline;
++
++ if ((max_cmdline = MAX_GUEST_CMDLINE) > COMMAND_LINE_SIZE)
++ max_cmdline = COMMAND_LINE_SIZE;
++ memcpy(saved_command_line, xen_start_info->cmd_line, max_cmdline);
++ saved_command_line[max_cmdline-1] = '\0';
++#endif
++ printk("Bootdata ok (command line is %s)\n", saved_command_line);
++}
++
++static void __init setup_boot_cpu_data(void)
++{
++ unsigned int dummy, eax;
++
++ /* get vendor info */
++ cpuid(0, (unsigned int *)&boot_cpu_data.cpuid_level,
++ (unsigned int *)&boot_cpu_data.x86_vendor_id[0],
++ (unsigned int *)&boot_cpu_data.x86_vendor_id[8],
++ (unsigned int *)&boot_cpu_data.x86_vendor_id[4]);
++
++ /* get cpu type */
++ cpuid(1, &eax, &dummy, &dummy,
++ (unsigned int *) &boot_cpu_data.x86_capability);
++ boot_cpu_data.x86 = (eax >> 8) & 0xf;
++ boot_cpu_data.x86_model = (eax >> 4) & 0xf;
++ boot_cpu_data.x86_mask = eax & 0xf;
++}
++
++#include <xen/interface/memory.h>
++unsigned long *machine_to_phys_mapping;
++EXPORT_SYMBOL(machine_to_phys_mapping);
++unsigned int machine_to_phys_order;
++EXPORT_SYMBOL(machine_to_phys_order);
++
++void __init x86_64_start_kernel(char * real_mode_data)
++{
++ struct xen_machphys_mapping mapping;
++ unsigned long machine_to_phys_nr_ents;
++ char *s;
++ int i;
++
++ setup_xen_features();
++
++ xen_start_info = (struct start_info *)real_mode_data;
++ if (!xen_feature(XENFEAT_auto_translated_physmap))
++ phys_to_machine_mapping =
++ (unsigned long *)xen_start_info->mfn_list;
++ start_pfn = (__pa(xen_start_info->pt_base) >> PAGE_SHIFT) +
++ xen_start_info->nr_pt_frames;
++
++ machine_to_phys_mapping = (unsigned long *)MACH2PHYS_VIRT_START;
++ machine_to_phys_nr_ents = MACH2PHYS_NR_ENTRIES;
++ if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
++ machine_to_phys_mapping = (unsigned long *)mapping.v_start;
++ machine_to_phys_nr_ents = mapping.max_mfn + 1;
++ }
++ while ((1UL << machine_to_phys_order) < machine_to_phys_nr_ents )
++ machine_to_phys_order++;
++
++#if 0
++ for (i = 0; i < 256; i++)
++ set_intr_gate(i, early_idt_handler);
++ asm volatile("lidt %0" :: "m" (idt_descr));
++#endif
++
++ /*
++ * This must be called really, really early:
++ */
++ lockdep_init();
++
++ for (i = 0; i < NR_CPUS; i++)
++ cpu_pda(i) = &boot_cpu_pda[i];
++
++ pda_init(0);
++ copy_bootdata(real_mode_data);
++#ifdef CONFIG_SMP
++ cpu_set(0, cpu_online_map);
++#endif
++ s = strstr(saved_command_line, "earlyprintk=");
++ if (s != NULL)
++ setup_early_printk(strchr(s, '=') + 1);
++#ifdef CONFIG_NUMA
++ s = strstr(saved_command_line, "numa=");
++ if (s != NULL)
++ numa_setup(s+5);
++#endif
++#ifdef CONFIG_X86_IO_APIC
++ if (strstr(saved_command_line, "disableapic"))
++ disable_apic = 1;
++#endif
++ /* You need early console to see that */
++ if (__pa_symbol(&_end) >= KERNEL_TEXT_SIZE)
++ panic("Kernel too big for kernel mapping\n");
++
++ setup_boot_cpu_data();
++ start_kernel();
++}
+--- linux-2.6.18.8/arch/x86_64/kernel/init_task.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/init_task.c 2008-05-19 00:33:18.961239082 +0300
@@ -37,6 +37,8 @@
struct task_struct init_task = INIT_TASK(init_task);
+#endif
#define ALIGN_TO_4K __attribute__((section(".data.init_task")))
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/io_apic-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/io_apic-xen.c
---- linux-2.6.18/arch/x86_64/kernel/io_apic-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/io_apic-xen.c 2007-12-23 11:15:00.082818365 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/io_apic-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/io_apic-xen.c 2008-05-19 00:33:19.005241619 +0300
@@ -0,0 +1,2262 @@
+/*
+ * Intel IO-APIC support for multi-Pentium hosts.
+ apic_op.apic_physbase = mp_ioapics[apic].mpc_apicaddr;
+ apic_op.reg = reg;
+ apic_op.value = value;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_apic_write, &apic_op);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_apic_write, &apic_op));
+}
+
+#define io_apic_read(a,r) xen_io_apic_read(a,r)
+}
+#endif
+#endif /* !CONFIG_XEN */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/ioport-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/ioport-xen.c
---- linux-2.6.18/arch/x86_64/kernel/ioport-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/ioport-xen.c 2007-12-23 11:15:00.086151878 +0100
-@@ -0,0 +1,99 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/ioport-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/ioport-xen.c 2008-05-19 00:33:19.009241849 +0300
+@@ -0,0 +1,100 @@
+/*
+ * linux/arch/x86_64/kernel/ioport.c
+ *
+
+ set_xen_guest_handle(set_iobitmap.bitmap, (char *)bitmap);
+ set_iobitmap.nr_ports = IO_BITMAP_BITS;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &set_iobitmap);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap,
++ &set_iobitmap));
+ }
+
+ set_bitmap(t->io_bitmap_ptr, from, num, !turn_on);
+
+ /* Force the change at ring 0. */
+ set_iopl.iopl = (new_iopl == 0) ? 1 : new_iopl;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl));
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/irq-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/irq-xen.c
---- linux-2.6.18/arch/x86_64/kernel/irq-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/irq-xen.c 2007-12-23 11:15:00.086151878 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/irq-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/irq-xen.c 2008-05-19 00:33:19.009241849 +0300
@@ -0,0 +1,197 @@
+/*
+ * linux/arch/x86_64/kernel/irq.c
+ printk("unexpected IRQ trap at vector %02x\n", irq);
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/ldt-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/ldt-xen.c
---- linux-2.6.18/arch/x86_64/kernel/ldt-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/ldt-xen.c 2007-12-23 11:15:00.086151878 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/ldt-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/ldt-xen.c 2008-05-19 00:33:19.013242080 +0300
@@ -0,0 +1,282 @@
+/*
+ * linux/arch/x86_64/kernel/ldt.c
+ }
+ return ret;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/machine_kexec.c linux-2.6.18-xen.hg/arch/x86_64/kernel/machine_kexec.c
---- linux-2.6.18/arch/x86_64/kernel/machine_kexec.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/machine_kexec.c 2007-12-23 11:15:00.089485388 +0100
-@@ -15,6 +15,113 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/machine_kexec.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/machine_kexec.c 2008-05-19 00:33:19.013242080 +0300
+@@ -15,6 +15,128 @@
#include <asm/mmu_context.h>
#include <asm/io.h>
+ xki->page_list[PA_PTE_1] = __ma(kexec_pte1);
+}
+
++int __init machine_kexec_setup_resources(struct resource *hypervisor,
++ struct resource *phys_cpus,
++ int nr_phys_cpus)
++{
++ int k;
++
++ /* The per-cpu crash note resources belong to the hypervisor resource */
++ for (k = 0; k < nr_phys_cpus; k++)
++ request_resource(hypervisor, phys_cpus + k);
++
++ return 0;
++}
++
++void machine_kexec_register_resources(struct resource *res) { ; }
++
+#else /* CONFIG_XEN */
+
+#define x__pmd(x) __pmd(x)
static void init_level2_page(pmd_t *level2p, unsigned long addr)
{
unsigned long end_addr;
-@@ -22,7 +129,7 @@
+@@ -22,7 +144,7 @@
addr &= PAGE_MASK;
end_addr = addr + PUD_SIZE;
while (addr < end_addr) {
addr += PMD_SIZE;
}
}
-@@ -47,12 +154,12 @@
+@@ -47,12 +169,12 @@
}
level2p = (pmd_t *)page_address(page);
init_level2_page(level2p, addr);
addr += PUD_SIZE;
}
out:
-@@ -83,12 +190,12 @@
+@@ -83,12 +205,12 @@
if (result) {
goto out;
}
addr += PGDIR_SIZE;
}
out:
-@@ -99,77 +206,29 @@
+@@ -99,77 +221,29 @@
static int init_pgtable(struct kimage *image, unsigned long start_pgtable)
{
pgd_t *level4p;
- /* x86-64 supports unaliged loads & stores */
- curidt.size = limit;
- curidt.address = (unsigned long)newidt;
-+ unsigned long x_end_pfn = end_pfn;
-
+-
- __asm__ __volatile__ (
- "lidtq %0\n"
- : : "m" (curidt)
- );
-};
--
--
--static void set_gdt(void *newgdt, u16 limit)
--{
-- struct desc_ptr curgdt;
++ unsigned long x_end_pfn = end_pfn;
+
+#ifdef CONFIG_XEN
+ x_end_pfn = HYPERVISOR_memory_op(XENMEM_maximum_ram_page, NULL);
+#endif
+-static void set_gdt(void *newgdt, u16 limit)
+-{
+- struct desc_ptr curgdt;
+-
- /* x86-64 supports unaligned loads & stores */
- curgdt.size = limit;
- curgdt.address = (unsigned long)newgdt;
return 0;
}
-@@ -178,51 +237,43 @@
+@@ -178,51 +252,43 @@
return;
}
+ image->start);
}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/Makefile linux-2.6.18-xen.hg/arch/x86_64/kernel/Makefile
---- linux-2.6.18/arch/x86_64/kernel/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/Makefile 2007-12-23 11:15:00.062817316 +0100
-@@ -21,11 +21,13 @@
- obj-$(CONFIG_X86_CPUID) += cpuid.o
- obj-$(CONFIG_SMP) += smp.o smpboot.o trampoline.o
- obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o
-+obj-$(CONFIG_X86_XEN_GENAPIC) += genapic.o genapic_xen.o
- obj-$(CONFIG_X86_IO_APIC) += io_apic.o mpparse.o \
- genapic.o genapic_cluster.o genapic_flat.o
- obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
- obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
--obj-$(CONFIG_PM) += suspend.o
-+obj-$(CONFIG_SOFTWARE_SUSPEND) += suspend.o
-+obj-$(CONFIG_ACPI_SLEEP) += suspend.o
- obj-$(CONFIG_SOFTWARE_SUSPEND) += suspend_asm.o
- obj-$(CONFIG_CPU_FREQ) += cpufreq/
- obj-$(CONFIG_EARLY_PRINTK) += early_printk.o
-@@ -55,3 +57,8 @@
- msr-$(subst m,y,$(CONFIG_X86_MSR)) += ../../i386/kernel/msr.o
- alternative-y += ../../i386/kernel/alternative.o
-
-+time-$(CONFIG_XEN) += ../../i386/kernel/time.o
-+pci-dma-$(CONFIG_XEN) += ../../i386/kernel/pci-dma.o
-+
-+disabled-obj-$(CONFIG_XEN) := i8259.o reboot.o smpboot.o trampoline.o
-+%/head.o %/head.s: $(if $(CONFIG_XEN),EXTRA_AFLAGS,dummy) :=
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/mpparse-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/mpparse-xen.c
---- linux-2.6.18/arch/x86_64/kernel/mpparse-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/mpparse-xen.c 2007-12-23 11:15:00.092818896 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/mpparse-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/mpparse-xen.c 2008-05-19 00:33:19.021242541 +0300
@@ -0,0 +1,1011 @@
+/*
+ * Intel Multiprocessor Specification 1.1 and 1.4
+
+#endif /*CONFIG_X86_IO_APIC*/
+#endif /*CONFIG_ACPI*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/pci-swiotlb-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/pci-swiotlb-xen.c
---- linux-2.6.18/arch/x86_64/kernel/pci-swiotlb-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/pci-swiotlb-xen.c 2007-12-23 11:15:00.096152403 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/pci-swiotlb-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/pci-swiotlb-xen.c 2008-05-19 00:33:19.025242771 +0300
@@ -0,0 +1,55 @@
+/* Glue code to lib/swiotlb.c */
+
+ }
+#endif
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/process-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/process-xen.c
---- linux-2.6.18/arch/x86_64/kernel/process-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/process-xen.c 2007-12-23 11:15:00.099485909 +0100
-@@ -0,0 +1,829 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/process-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/process-xen.c 2008-05-19 00:33:19.029243002 +0300
+@@ -0,0 +1,833 @@
+/*
+ * linux/arch/x86-64/kernel/process.c
+ *
+ local_irq_disable();
+ cpu_clear(smp_processor_id(), cpu_initialized);
+ preempt_enable_no_resched();
-+ HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL);
++ VOID(HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL));
+ cpu_bringup();
+}
+#else
+ put_cpu();
+#endif
+#ifdef CONFIG_XEN
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &iobmp_op);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap,
++ &iobmp_op));
+#endif
+ t->io_bitmap_max = 0;
+ }
+
+void load_gs_index(unsigned gs)
+{
-+ HYPERVISOR_set_segment_base(SEGBASE_GS_USER_SEL, gs);
++ WARN_ON(HYPERVISOR_set_segment_base(SEGBASE_GS_USER_SEL, gs));
+}
+
+void flush_thread(void)
+ mcl++;
+ }
+
-+ (void)HYPERVISOR_multicall(_mcl, mcl - _mcl);
++ BUG_ON(mcl > _mcl + ARRAY_SIZE(_mcl));
++ if (unlikely(HYPERVISOR_multicall_check(_mcl, mcl - _mcl, NULL)))
++ BUG();
++
+ /*
+ * Switch DS and ES.
+ * This won't pick up thread selector changes, but I guess that is ok.
+ loadsegment(fs, next->fsindex);
+
+ if (next->fs)
-+ HYPERVISOR_set_segment_base(SEGBASE_FS, next->fs);
++ WARN_ON(HYPERVISOR_set_segment_base(SEGBASE_FS, next->fs));
+
+ if (unlikely(next->gsindex))
+ load_gs_index(next->gsindex);
+
+ if (next->gs)
-+ HYPERVISOR_set_segment_base(SEGBASE_GS_USER, next->gs);
++ WARN_ON(HYPERVISOR_set_segment_base(SEGBASE_GS_USER, next->gs));
+
+ /*
+ * Switch the PDA context.
+{
+}
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/relocate_kernel.S linux-2.6.18-xen.hg/arch/x86_64/kernel/relocate_kernel.S
---- linux-2.6.18/arch/x86_64/kernel/relocate_kernel.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/relocate_kernel.S 2007-12-23 11:15:00.099485909 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/relocate_kernel.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/relocate_kernel.S 2008-05-19 00:33:19.033243233 +0300
@@ -7,31 +7,195 @@
*/
+idt_80:
+ .word 0 /* limit */
+ .quad 0 /* base */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/setup64-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/setup64-xen.c
---- linux-2.6.18/arch/x86_64/kernel/setup64-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/setup64-xen.c 2007-12-23 11:15:00.102819418 +0100
-@@ -0,0 +1,365 @@
-+/*
-+ * X86-64 specific CPU setup.
-+ * Copyright (C) 1995 Linus Torvalds
-+ * Copyright 2001, 2002, 2003 SuSE Labs / Andi Kleen.
-+ * See setup.c for older changelog.
-+ *
-+ * Jun Nakajima <jun.nakajima@intel.com>
-+ * Modified for Xen
-+ *
-+ */
-+#include <linux/init.h>
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/string.h>
-+#include <linux/bootmem.h>
-+#include <linux/bitops.h>
-+#include <linux/module.h>
-+#include <asm/bootsetup.h>
-+#include <asm/pda.h>
-+#include <asm/pgtable.h>
-+#include <asm/processor.h>
-+#include <asm/desc.h>
-+#include <asm/atomic.h>
-+#include <asm/mmu_context.h>
-+#include <asm/smp.h>
-+#include <asm/i387.h>
-+#include <asm/percpu.h>
-+#include <asm/proto.h>
-+#include <asm/sections.h>
-+#ifdef CONFIG_XEN
-+#include <asm/hypervisor.h>
-+#endif
-+
-+char x86_boot_params[BOOT_PARAM_SIZE] __initdata = {0,};
-+
-+cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
-+
-+struct x8664_pda *_cpu_pda[NR_CPUS] __read_mostly;
-+EXPORT_SYMBOL(_cpu_pda);
-+struct x8664_pda boot_cpu_pda[NR_CPUS] __cacheline_aligned;
-+
-+#ifndef CONFIG_X86_NO_IDT
-+struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table };
-+#endif
-+
-+char boot_cpu_stack[IRQSTACKSIZE] __attribute__((section(".bss.page_aligned")));
-+
-+unsigned long __supported_pte_mask __read_mostly = ~0UL;
-+EXPORT_SYMBOL(__supported_pte_mask);
-+static int do_not_nx __cpuinitdata = 0;
-+
-+/* noexec=on|off
-+Control non executable mappings for 64bit processes.
-+
-+on Enable(default)
-+off Disable
-+*/
-+int __init nonx_setup(char *str)
-+{
-+ if (!strncmp(str, "on", 2)) {
-+ __supported_pte_mask |= _PAGE_NX;
-+ do_not_nx = 0;
-+ } else if (!strncmp(str, "off", 3)) {
-+ do_not_nx = 1;
-+ __supported_pte_mask &= ~_PAGE_NX;
-+ }
-+ return 1;
-+}
-+__setup("noexec=", nonx_setup); /* parsed early actually */
-+
-+int force_personality32 = 0;
-+
-+/* noexec32=on|off
-+Control non executable heap for 32bit processes.
-+To control the stack too use noexec=off
-+
-+on PROT_READ does not imply PROT_EXEC for 32bit processes
-+off PROT_READ implies PROT_EXEC (default)
-+*/
-+static int __init nonx32_setup(char *str)
-+{
-+ if (!strcmp(str, "on"))
-+ force_personality32 &= ~READ_IMPLIES_EXEC;
-+ else if (!strcmp(str, "off"))
-+ force_personality32 |= READ_IMPLIES_EXEC;
-+ return 1;
-+}
-+__setup("noexec32=", nonx32_setup);
-+
-+/*
-+ * Great future plan:
-+ * Declare PDA itself and support (irqstack,tss,pgd) as per cpu data.
-+ * Always point %gs to its beginning
-+ */
-+void __init setup_per_cpu_areas(void)
-+{
-+ int i;
-+ unsigned long size;
-+
-+#ifdef CONFIG_HOTPLUG_CPU
-+ prefill_possible_map();
-+#endif
-+
-+ /* Copy section for each CPU (we discard the original) */
-+ size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
-+#ifdef CONFIG_MODULES
-+ if (size < PERCPU_ENOUGH_ROOM)
-+ size = PERCPU_ENOUGH_ROOM;
-+#endif
-+
-+ for_each_cpu_mask (i, cpu_possible_map) {
-+ char *ptr;
-+
-+ if (!NODE_DATA(cpu_to_node(i))) {
-+ printk("cpu with no node %d, num_online_nodes %d\n",
-+ i, num_online_nodes());
-+ ptr = alloc_bootmem(size);
-+ } else {
-+ ptr = alloc_bootmem_node(NODE_DATA(cpu_to_node(i)), size);
-+ }
-+ if (!ptr)
-+ panic("Cannot allocate cpu data for CPU %d\n", i);
-+ cpu_pda(i)->data_offset = ptr - __per_cpu_start;
-+ memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
-+ }
-+}
-+
-+#ifdef CONFIG_XEN
-+static void switch_pt(void)
-+{
-+ xen_pt_switch(__pa_symbol(init_level4_pgt));
-+ xen_new_user_pt(__pa_symbol(__user_pgd(init_level4_pgt)));
-+}
-+
-+static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
-+{
-+ unsigned long frames[16];
-+ unsigned long va;
-+ int f;
-+
-+ for (va = gdt_descr->address, f = 0;
-+ va < gdt_descr->address + gdt_descr->size;
-+ va += PAGE_SIZE, f++) {
-+ frames[f] = virt_to_mfn(va);
-+ make_page_readonly(
-+ (void *)va, XENFEAT_writable_descriptor_tables);
-+ }
-+ if (HYPERVISOR_set_gdt(frames, (gdt_descr->size + 1) /
-+ sizeof (struct desc_struct)))
-+ BUG();
-+}
-+#else
-+static void switch_pt(void)
-+{
-+ asm volatile("movq %0,%%cr3" :: "r" (__pa_symbol(&init_level4_pgt)));
-+}
-+
-+static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
-+{
-+ asm volatile("lgdt %0" :: "m" (*gdt_descr));
-+ asm volatile("lidt %0" :: "m" (idt_descr));
-+}
-+#endif
-+
-+void pda_init(int cpu)
-+{
-+ struct x8664_pda *pda = cpu_pda(cpu);
-+
-+ /* Setup up data that may be needed in __get_free_pages early */
-+ asm volatile("movl %0,%%fs ; movl %0,%%gs" :: "r" (0));
-+#ifndef CONFIG_XEN
-+ wrmsrl(MSR_GS_BASE, pda);
-+#else
-+ HYPERVISOR_set_segment_base(SEGBASE_GS_KERNEL, (unsigned long)pda);
-+#endif
-+ pda->cpunumber = cpu;
-+ pda->irqcount = -1;
-+ pda->kernelstack =
-+ (unsigned long)stack_thread_info() - PDA_STACKOFFSET + THREAD_SIZE;
-+ pda->active_mm = &init_mm;
-+ pda->mmu_state = 0;
-+
-+ if (cpu == 0) {
-+#ifdef CONFIG_XEN
-+ xen_init_pt();
-+#endif
-+ /* others are initialized in smpboot.c */
-+ pda->pcurrent = &init_task;
-+ pda->irqstackptr = boot_cpu_stack;
-+ } else {
-+ pda->irqstackptr = (char *)
-+ __get_free_pages(GFP_ATOMIC, IRQSTACK_ORDER);
-+ if (!pda->irqstackptr)
-+ panic("cannot allocate irqstack for cpu %d", cpu);
-+ }
-+
-+ switch_pt();
-+
-+ pda->irqstackptr += IRQSTACKSIZE-64;
-+}
-+
-+#ifndef CONFIG_X86_NO_TSS
-+char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]
-+__attribute__((section(".bss.page_aligned")));
-+#endif
-+
-+/* May not be marked __init: used by software suspend */
-+void syscall_init(void)
-+{
-+#ifndef CONFIG_XEN
-+ /*
-+ * LSTAR and STAR live in a bit strange symbiosis.
-+ * They both write to the same internal register. STAR allows to set CS/DS
-+ * but only a 32bit target. LSTAR sets the 64bit rip.
-+ */
-+ wrmsrl(MSR_STAR, ((u64)__USER32_CS)<<48 | ((u64)__KERNEL_CS)<<32);
-+ wrmsrl(MSR_LSTAR, system_call);
-+
-+ /* Flags to clear on syscall */
-+ wrmsrl(MSR_SYSCALL_MASK, EF_TF|EF_DF|EF_IE|0x3000);
-+#endif
-+#ifdef CONFIG_IA32_EMULATION
-+ syscall32_cpu_init ();
-+#endif
-+}
-+
-+void __cpuinit check_efer(void)
-+{
-+ unsigned long efer;
-+
-+ rdmsrl(MSR_EFER, efer);
-+ if (!(efer & EFER_NX) || do_not_nx) {
-+ __supported_pte_mask &= ~_PAGE_NX;
-+ }
-+}
-+
-+unsigned long kernel_eflags;
-+
-+/*
-+ * cpu_init() initializes state that is per-CPU. Some data is already
-+ * initialized (naturally) in the bootstrap process, such as the GDT
-+ * and IDT. We reload them nevertheless, this function acts as a
-+ * 'CPU state barrier', nothing should get across.
-+ * A lot of state is already set up in PDA init.
-+ */
-+void __cpuinit cpu_init (void)
-+{
-+ int cpu = stack_smp_processor_id();
-+#ifndef CONFIG_X86_NO_TSS
-+ struct tss_struct *t = &per_cpu(init_tss, cpu);
-+ struct orig_ist *orig_ist = &per_cpu(orig_ist, cpu);
-+ unsigned long v;
-+ char *estacks = NULL;
-+ unsigned i;
-+#endif
-+ struct task_struct *me;
-+
-+ /* CPU 0 is initialised in head64.c */
-+ if (cpu != 0) {
-+ pda_init(cpu);
-+ zap_low_mappings(cpu);
-+ }
-+#ifndef CONFIG_X86_NO_TSS
-+ else
-+ estacks = boot_exception_stacks;
-+#endif
-+
-+ me = current;
-+
-+ if (cpu_test_and_set(cpu, cpu_initialized))
-+ panic("CPU#%d already initialized!\n", cpu);
-+
-+ printk("Initializing CPU#%d\n", cpu);
-+
-+ clear_in_cr4(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
-+
-+ /*
-+ * Initialize the per-CPU GDT with the boot GDT,
-+ * and set up the GDT descriptor:
-+ */
-+#ifndef CONFIG_XEN
-+ if (cpu)
-+ memcpy(cpu_gdt(cpu), cpu_gdt_table, GDT_SIZE);
-+#endif
-+
-+ cpu_gdt_descr[cpu].size = GDT_SIZE;
-+ cpu_gdt_init(&cpu_gdt_descr[cpu]);
-+
-+ memset(me->thread.tls_array, 0, GDT_ENTRY_TLS_ENTRIES * 8);
-+ syscall_init();
-+
-+ wrmsrl(MSR_FS_BASE, 0);
-+ wrmsrl(MSR_KERNEL_GS_BASE, 0);
-+ barrier();
-+
-+ check_efer();
-+
-+#ifndef CONFIG_X86_NO_TSS
-+ /*
-+ * set up and load the per-CPU TSS
-+ */
-+ for (v = 0; v < N_EXCEPTION_STACKS; v++) {
-+ if (cpu) {
-+ static const unsigned int order[N_EXCEPTION_STACKS] = {
-+ [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STACK_ORDER,
-+ [DEBUG_STACK - 1] = DEBUG_STACK_ORDER
-+ };
-+
-+ estacks = (char *)__get_free_pages(GFP_ATOMIC, order[v]);
-+ if (!estacks)
-+ panic("Cannot allocate exception stack %ld %d\n",
-+ v, cpu);
-+ }
-+ switch (v + 1) {
-+#if DEBUG_STKSZ > EXCEPTION_STKSZ
-+ case DEBUG_STACK:
-+ cpu_pda(cpu)->debugstack = (unsigned long)estacks;
-+ estacks += DEBUG_STKSZ;
-+ break;
-+#endif
-+ default:
-+ estacks += EXCEPTION_STKSZ;
-+ break;
-+ }
-+ orig_ist->ist[v] = t->ist[v] = (unsigned long)estacks;
-+ }
-+
-+ t->io_bitmap_base = offsetof(struct tss_struct, io_bitmap);
-+ /*
-+ * <= is required because the CPU will access up to
-+ * 8 bits beyond the end of the IO permission bitmap.
-+ */
-+ for (i = 0; i <= IO_BITMAP_LONGS; i++)
-+ t->io_bitmap[i] = ~0UL;
-+#endif
-+
-+ atomic_inc(&init_mm.mm_count);
-+ me->active_mm = &init_mm;
-+ if (me->mm)
-+ BUG();
-+ enter_lazy_tlb(&init_mm, me);
-+
-+#ifndef CONFIG_X86_NO_TSS
-+ set_tss_desc(cpu, t);
-+#endif
-+#ifndef CONFIG_XEN
-+ load_TR_desc();
-+#endif
-+ load_LDT(&init_mm.context);
-+
-+ /*
-+ * Clear all 6 debug registers:
-+ */
-+
-+ set_debugreg(0UL, 0);
-+ set_debugreg(0UL, 1);
-+ set_debugreg(0UL, 2);
-+ set_debugreg(0UL, 3);
-+ set_debugreg(0UL, 6);
-+ set_debugreg(0UL, 7);
-+
-+ fpu_init();
-+
-+ raw_local_save_flags(kernel_eflags);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/setup-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/setup-xen.c
---- linux-2.6.18/arch/x86_64/kernel/setup-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/setup-xen.c 2007-12-23 11:15:00.102819418 +0100
-@@ -0,0 +1,1662 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/setup-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/setup-xen.c 2008-05-19 00:33:19.037243463 +0300
+@@ -0,0 +1,1652 @@
+/*
+ * linux/arch/x86-64/kernel/setup.c
+ *
+
+ copy_edid();
+
-+ HYPERVISOR_vm_assist(VMASST_CMD_enable,
-+ VMASST_TYPE_writable_pagetables);
++ WARN_ON(HYPERVISOR_vm_assist(VMASST_CMD_enable,
++ VMASST_TYPE_writable_pagetables));
+
+ ARCH_SETUP
+#else
+ */
+ probe_roms();
+#ifdef CONFIG_XEN
-+ if (is_initial_xendomain()) {
-+ struct xen_memory_map memmap;
-+
-+ memmap.nr_entries = E820MAX;
-+ set_xen_guest_handle(memmap.buffer, machine_e820.map);
-+
-+ if (HYPERVISOR_memory_op(XENMEM_machine_memory_map, &memmap))
-+ BUG();
-+ machine_e820.nr_map = memmap.nr_entries;
-+
++ if (is_initial_xendomain())
+ e820_reserve_resources(machine_e820.map, machine_e820.nr_map);
-+ }
+#else
+ e820_reserve_resources(e820.map, e820.nr_map);
+#endif
+ struct physdev_set_iopl set_iopl;
+
+ set_iopl.iopl = 1;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl));
+
+ if (is_initial_xendomain()) {
+#ifdef CONFIG_VT
+}
+device_initcall(add_pcspkr);
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/smp-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/smp-xen.c
---- linux-2.6.18/arch/x86_64/kernel/smp-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/smp-xen.c 2007-12-23 11:15:00.106152927 +0100
-@@ -0,0 +1,577 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/setup64-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/setup64-xen.c 2008-05-19 00:33:19.037243463 +0300
+@@ -0,0 +1,367 @@
++/*
++ * X86-64 specific CPU setup.
++ * Copyright (C) 1995 Linus Torvalds
++ * Copyright 2001, 2002, 2003 SuSE Labs / Andi Kleen.
++ * See setup.c for older changelog.
++ *
++ * Jun Nakajima <jun.nakajima@intel.com>
++ * Modified for Xen
++ *
++ */
++#include <linux/init.h>
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/string.h>
++#include <linux/bootmem.h>
++#include <linux/bitops.h>
++#include <linux/module.h>
++#include <asm/bootsetup.h>
++#include <asm/pda.h>
++#include <asm/pgtable.h>
++#include <asm/processor.h>
++#include <asm/desc.h>
++#include <asm/atomic.h>
++#include <asm/mmu_context.h>
++#include <asm/smp.h>
++#include <asm/i387.h>
++#include <asm/percpu.h>
++#include <asm/proto.h>
++#include <asm/sections.h>
++#ifdef CONFIG_XEN
++#include <asm/hypervisor.h>
++#endif
++
++char x86_boot_params[BOOT_PARAM_SIZE] __initdata = {0,};
++
++cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
++
++struct x8664_pda *_cpu_pda[NR_CPUS] __read_mostly;
++EXPORT_SYMBOL(_cpu_pda);
++struct x8664_pda boot_cpu_pda[NR_CPUS] __cacheline_aligned;
++
++#ifndef CONFIG_X86_NO_IDT
++struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table };
++#endif
++
++char boot_cpu_stack[IRQSTACKSIZE] __attribute__((section(".bss.page_aligned")));
++
++unsigned long __supported_pte_mask __read_mostly = ~0UL;
++EXPORT_SYMBOL(__supported_pte_mask);
++static int do_not_nx __cpuinitdata = 0;
++
++/* noexec=on|off
++Control non executable mappings for 64bit processes.
++
++on Enable(default)
++off Disable
++*/
++int __init nonx_setup(char *str)
++{
++ if (!strncmp(str, "on", 2)) {
++ __supported_pte_mask |= _PAGE_NX;
++ do_not_nx = 0;
++ } else if (!strncmp(str, "off", 3)) {
++ do_not_nx = 1;
++ __supported_pte_mask &= ~_PAGE_NX;
++ }
++ return 1;
++}
++__setup("noexec=", nonx_setup); /* parsed early actually */
++
++int force_personality32 = 0;
++
++/* noexec32=on|off
++Control non executable heap for 32bit processes.
++To control the stack too use noexec=off
++
++on PROT_READ does not imply PROT_EXEC for 32bit processes
++off PROT_READ implies PROT_EXEC (default)
++*/
++static int __init nonx32_setup(char *str)
++{
++ if (!strcmp(str, "on"))
++ force_personality32 &= ~READ_IMPLIES_EXEC;
++ else if (!strcmp(str, "off"))
++ force_personality32 |= READ_IMPLIES_EXEC;
++ return 1;
++}
++__setup("noexec32=", nonx32_setup);
++
++/*
++ * Great future plan:
++ * Declare PDA itself and support (irqstack,tss,pgd) as per cpu data.
++ * Always point %gs to its beginning
++ */
++void __init setup_per_cpu_areas(void)
++{
++ int i;
++ unsigned long size;
++
++#ifdef CONFIG_HOTPLUG_CPU
++ prefill_possible_map();
++#endif
++
++ /* Copy section for each CPU (we discard the original) */
++ size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
++#ifdef CONFIG_MODULES
++ if (size < PERCPU_ENOUGH_ROOM)
++ size = PERCPU_ENOUGH_ROOM;
++#endif
++
++ for_each_cpu_mask (i, cpu_possible_map) {
++ char *ptr;
++
++ if (!NODE_DATA(cpu_to_node(i))) {
++ printk("cpu with no node %d, num_online_nodes %d\n",
++ i, num_online_nodes());
++ ptr = alloc_bootmem(size);
++ } else {
++ ptr = alloc_bootmem_node(NODE_DATA(cpu_to_node(i)), size);
++ }
++ if (!ptr)
++ panic("Cannot allocate cpu data for CPU %d\n", i);
++ cpu_pda(i)->data_offset = ptr - __per_cpu_start;
++ memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
++ }
++}
++
++#ifdef CONFIG_XEN
++static void switch_pt(void)
++{
++ xen_pt_switch(__pa_symbol(init_level4_pgt));
++ xen_new_user_pt(__pa_symbol(__user_pgd(init_level4_pgt)));
++}
++
++static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
++{
++ unsigned long frames[16];
++ unsigned long va;
++ int f;
++
++ for (va = gdt_descr->address, f = 0;
++ va < gdt_descr->address + gdt_descr->size;
++ va += PAGE_SIZE, f++) {
++ frames[f] = virt_to_mfn(va);
++ make_page_readonly(
++ (void *)va, XENFEAT_writable_descriptor_tables);
++ }
++ if (HYPERVISOR_set_gdt(frames, (gdt_descr->size + 1) /
++ sizeof (struct desc_struct)))
++ BUG();
++}
++#else
++static void switch_pt(void)
++{
++ asm volatile("movq %0,%%cr3" :: "r" (__pa_symbol(&init_level4_pgt)));
++}
++
++static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
++{
++ asm volatile("lgdt %0" :: "m" (*gdt_descr));
++ asm volatile("lidt %0" :: "m" (idt_descr));
++}
++#endif
++
++void pda_init(int cpu)
++{
++ struct x8664_pda *pda = cpu_pda(cpu);
++
++ /* Setup up data that may be needed in __get_free_pages early */
++ asm volatile("movl %0,%%fs ; movl %0,%%gs" :: "r" (0));
++#ifndef CONFIG_XEN
++ wrmsrl(MSR_GS_BASE, pda);
++#else
++ if (HYPERVISOR_set_segment_base(SEGBASE_GS_KERNEL,
++ (unsigned long)pda))
++ BUG();
++#endif
++ pda->cpunumber = cpu;
++ pda->irqcount = -1;
++ pda->kernelstack =
++ (unsigned long)stack_thread_info() - PDA_STACKOFFSET + THREAD_SIZE;
++ pda->active_mm = &init_mm;
++ pda->mmu_state = 0;
++
++ if (cpu == 0) {
++#ifdef CONFIG_XEN
++ xen_init_pt();
++#endif
++ /* others are initialized in smpboot.c */
++ pda->pcurrent = &init_task;
++ pda->irqstackptr = boot_cpu_stack;
++ } else {
++ pda->irqstackptr = (char *)
++ __get_free_pages(GFP_ATOMIC, IRQSTACK_ORDER);
++ if (!pda->irqstackptr)
++ panic("cannot allocate irqstack for cpu %d", cpu);
++ }
++
++ switch_pt();
++
++ pda->irqstackptr += IRQSTACKSIZE-64;
++}
++
++#ifndef CONFIG_X86_NO_TSS
++char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]
++__attribute__((section(".bss.page_aligned")));
++#endif
++
++/* May not be marked __init: used by software suspend */
++void syscall_init(void)
++{
++#ifndef CONFIG_XEN
++ /*
++ * LSTAR and STAR live in a bit strange symbiosis.
++ * They both write to the same internal register. STAR allows to set CS/DS
++ * but only a 32bit target. LSTAR sets the 64bit rip.
++ */
++ wrmsrl(MSR_STAR, ((u64)__USER32_CS)<<48 | ((u64)__KERNEL_CS)<<32);
++ wrmsrl(MSR_LSTAR, system_call);
++
++ /* Flags to clear on syscall */
++ wrmsrl(MSR_SYSCALL_MASK, EF_TF|EF_DF|EF_IE|0x3000);
++#endif
++#ifdef CONFIG_IA32_EMULATION
++ syscall32_cpu_init ();
++#endif
++}
++
++void __cpuinit check_efer(void)
++{
++ unsigned long efer;
++
++ rdmsrl(MSR_EFER, efer);
++ if (!(efer & EFER_NX) || do_not_nx) {
++ __supported_pte_mask &= ~_PAGE_NX;
++ }
++}
++
++unsigned long kernel_eflags;
++
++/*
++ * cpu_init() initializes state that is per-CPU. Some data is already
++ * initialized (naturally) in the bootstrap process, such as the GDT
++ * and IDT. We reload them nevertheless, this function acts as a
++ * 'CPU state barrier', nothing should get across.
++ * A lot of state is already set up in PDA init.
++ */
++void __cpuinit cpu_init (void)
++{
++ int cpu = stack_smp_processor_id();
++#ifndef CONFIG_X86_NO_TSS
++ struct tss_struct *t = &per_cpu(init_tss, cpu);
++ struct orig_ist *orig_ist = &per_cpu(orig_ist, cpu);
++ unsigned long v;
++ char *estacks = NULL;
++ unsigned i;
++#endif
++ struct task_struct *me;
++
++ /* CPU 0 is initialised in head64.c */
++ if (cpu != 0) {
++ pda_init(cpu);
++ zap_low_mappings(cpu);
++ }
++#ifndef CONFIG_X86_NO_TSS
++ else
++ estacks = boot_exception_stacks;
++#endif
++
++ me = current;
++
++ if (cpu_test_and_set(cpu, cpu_initialized))
++ panic("CPU#%d already initialized!\n", cpu);
++
++ printk("Initializing CPU#%d\n", cpu);
++
++ clear_in_cr4(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
++
++ /*
++ * Initialize the per-CPU GDT with the boot GDT,
++ * and set up the GDT descriptor:
++ */
++#ifndef CONFIG_XEN
++ if (cpu)
++ memcpy(cpu_gdt(cpu), cpu_gdt_table, GDT_SIZE);
++#endif
++
++ cpu_gdt_descr[cpu].size = GDT_SIZE;
++ cpu_gdt_init(&cpu_gdt_descr[cpu]);
++
++ memset(me->thread.tls_array, 0, GDT_ENTRY_TLS_ENTRIES * 8);
++ syscall_init();
++
++ wrmsrl(MSR_FS_BASE, 0);
++ wrmsrl(MSR_KERNEL_GS_BASE, 0);
++ barrier();
++
++ check_efer();
++
++#ifndef CONFIG_X86_NO_TSS
++ /*
++ * set up and load the per-CPU TSS
++ */
++ for (v = 0; v < N_EXCEPTION_STACKS; v++) {
++ if (cpu) {
++ static const unsigned int order[N_EXCEPTION_STACKS] = {
++ [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STACK_ORDER,
++ [DEBUG_STACK - 1] = DEBUG_STACK_ORDER
++ };
++
++ estacks = (char *)__get_free_pages(GFP_ATOMIC, order[v]);
++ if (!estacks)
++ panic("Cannot allocate exception stack %ld %d\n",
++ v, cpu);
++ }
++ switch (v + 1) {
++#if DEBUG_STKSZ > EXCEPTION_STKSZ
++ case DEBUG_STACK:
++ cpu_pda(cpu)->debugstack = (unsigned long)estacks;
++ estacks += DEBUG_STKSZ;
++ break;
++#endif
++ default:
++ estacks += EXCEPTION_STKSZ;
++ break;
++ }
++ orig_ist->ist[v] = t->ist[v] = (unsigned long)estacks;
++ }
++
++ t->io_bitmap_base = offsetof(struct tss_struct, io_bitmap);
++ /*
++ * <= is required because the CPU will access up to
++ * 8 bits beyond the end of the IO permission bitmap.
++ */
++ for (i = 0; i <= IO_BITMAP_LONGS; i++)
++ t->io_bitmap[i] = ~0UL;
++#endif
++
++ atomic_inc(&init_mm.mm_count);
++ me->active_mm = &init_mm;
++ if (me->mm)
++ BUG();
++ enter_lazy_tlb(&init_mm, me);
++
++#ifndef CONFIG_X86_NO_TSS
++ set_tss_desc(cpu, t);
++#endif
++#ifndef CONFIG_XEN
++ load_TR_desc();
++#endif
++ load_LDT(&init_mm.context);
++
++ /*
++ * Clear all 6 debug registers:
++ */
++
++ set_debugreg(0UL, 0);
++ set_debugreg(0UL, 1);
++ set_debugreg(0UL, 2);
++ set_debugreg(0UL, 3);
++ set_debugreg(0UL, 6);
++ set_debugreg(0UL, 7);
++
++ fpu_init();
++
++ raw_local_save_flags(kernel_eflags);
++}
+--- linux-2.6.18.8/arch/x86_64/kernel/signal.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/signal.c 2008-05-19 00:33:19.041243694 +0300
+@@ -38,37 +38,6 @@
+ sigset_t *set, struct pt_regs * regs);
+
+ asmlinkage long
+-sys_rt_sigsuspend(sigset_t __user *unewset, size_t sigsetsize, struct pt_regs *regs)
+-{
+- sigset_t saveset, newset;
+-
+- /* XXX: Don't preclude handling different sized sigset_t's. */
+- if (sigsetsize != sizeof(sigset_t))
+- return -EINVAL;
+-
+- if (copy_from_user(&newset, unewset, sizeof(newset)))
+- return -EFAULT;
+- sigdelsetmask(&newset, ~_BLOCKABLE);
+-
+- spin_lock_irq(¤t->sighand->siglock);
+- saveset = current->blocked;
+- current->blocked = newset;
+- recalc_sigpending();
+- spin_unlock_irq(¤t->sighand->siglock);
+-#ifdef DEBUG_SIG
+- printk("rt_sigsuspend savset(%lx) newset(%lx) regs(%p) rip(%lx)\n",
+- saveset, newset, regs, regs->rip);
+-#endif
+- regs->rax = -EINTR;
+- while (1) {
+- current->state = TASK_INTERRUPTIBLE;
+- schedule();
+- if (do_signal(regs, &saveset))
+- return -EINTR;
+- }
+-}
+-
+-asmlinkage long
+ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss,
+ struct pt_regs *regs)
+ {
+@@ -341,11 +310,11 @@
+ current->comm, current->pid, frame, regs->rip, frame->pretcode);
+ #endif
+
+- return 1;
++ return 0;
+
+ give_sigsegv:
+ force_sigsegv(sig, current);
+- return 0;
++ return -EFAULT;
+ }
+
+ /*
+@@ -408,7 +377,7 @@
+ #endif
+ ret = setup_rt_frame(sig, ka, info, oldset, regs);
+
+- if (ret) {
++ if (ret == 0) {
+ spin_lock_irq(¤t->sighand->siglock);
+ sigorsets(¤t->blocked,¤t->blocked,&ka->sa.sa_mask);
+ if (!(ka->sa.sa_flags & SA_NODEFER))
+@@ -425,11 +394,12 @@
+ * want to handle. Thus you cannot kill init even with a SIGKILL even by
+ * mistake.
+ */
+-int do_signal(struct pt_regs *regs, sigset_t *oldset)
++static void do_signal(struct pt_regs *regs)
+ {
+ struct k_sigaction ka;
+ siginfo_t info;
+ int signr;
++ sigset_t *oldset;
+
+ /*
+ * We want the common case to go fast, which
+@@ -438,9 +408,11 @@
+ * if so.
+ */
+ if (!user_mode(regs))
+- return 1;
++ return;
+
+- if (!oldset)
++ if (test_thread_flag(TIF_RESTORE_SIGMASK))
++ oldset = ¤t->saved_sigmask;
++ else
+ oldset = ¤t->blocked;
+
+ signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+@@ -454,30 +426,46 @@
+ set_debugreg(current->thread.debugreg7, 7);
+
+ /* Whee! Actually deliver the signal. */
+- return handle_signal(signr, &info, &ka, oldset, regs);
++ if (handle_signal(signr, &info, &ka, oldset, regs) == 0) {
++ /* a signal was successfully delivered; the saved
++ * sigmask will have been stored in the signal frame,
++ * and will be restored by sigreturn, so we can simply
++ * clear the TIF_RESTORE_SIGMASK flag */
++ clear_thread_flag(TIF_RESTORE_SIGMASK);
++ }
++ return;
+ }
+
+ /* Did we come from a system call? */
+ if ((long)regs->orig_rax >= 0) {
+ /* Restart the system call - no handlers present */
+ long res = regs->rax;
+- if (res == -ERESTARTNOHAND ||
+- res == -ERESTARTSYS ||
+- res == -ERESTARTNOINTR) {
++ switch (res) {
++ case -ERESTARTNOHAND:
++ case -ERESTARTSYS:
++ case -ERESTARTNOINTR:
+ regs->rax = regs->orig_rax;
+ regs->rip -= 2;
+- }
+- if (regs->rax == (unsigned long)-ERESTART_RESTARTBLOCK) {
++ break;
++ case -ERESTART_RESTARTBLOCK:
+ regs->rax = test_thread_flag(TIF_IA32) ?
+ __NR_ia32_restart_syscall :
+ __NR_restart_syscall;
+ regs->rip -= 2;
++ break;
+ }
+ }
+- return 0;
++
++ /* if there's no signal to deliver, we just put the saved sigmask
++ back. */
++ if (test_thread_flag(TIF_RESTORE_SIGMASK)) {
++ clear_thread_flag(TIF_RESTORE_SIGMASK);
++ sigprocmask(SIG_SETMASK, ¤t->saved_sigmask, NULL);
++ }
+ }
+
+-void do_notify_resume(struct pt_regs *regs, sigset_t *oldset, __u32 thread_info_flags)
++void
++do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags)
+ {
+ #ifdef DEBUG_SIG
+ printk("do_notify_resume flags:%x rip:%lx rsp:%lx caller:%lx pending:%lx\n",
+@@ -491,8 +479,8 @@
+ }
+
+ /* deal with pending signal delivery */
+- if (thread_info_flags & _TIF_SIGPENDING)
+- do_signal(regs,oldset);
++ if (thread_info_flags & (_TIF_SIGPENDING|_TIF_RESTORE_SIGMASK))
++ do_signal(regs);
+ }
+
+ void signal_fault(struct pt_regs *regs, void __user *frame, char *where)
+--- linux-2.6.18.8/arch/x86_64/kernel/smp-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/smp-xen.c 2008-05-19 00:33:19.041243694 +0300
+@@ -0,0 +1,575 @@
+/*
+ * Intel SMP support routines.
+ *
+ to a full cache line because other CPUs can access it and we don't
+ want false sharing in the per cpu data segment. */
+static DEFINE_PER_CPU(union smp_flush_state, flush_state);
-+#endif
+
+/*
+ * We cannot call mmdrop() because we are in interrupt context,
+ load_cr3(swapper_pg_dir);
+}
+
-+#ifndef CONFIG_XEN
+/*
+ *
+ * The flush IPI assumes that a thread switch happens in this order:
+ return 0; /* Should not happen */
+#endif
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/suspend.c linux-2.6.18-xen.hg/arch/x86_64/kernel/suspend.c
---- linux-2.6.18/arch/x86_64/kernel/suspend.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/suspend.c 2007-12-23 11:15:00.106152927 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/suspend.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/suspend.c 2008-05-19 00:33:19.045243924 +0300
@@ -114,12 +114,14 @@
void fix_processor_context(void)
syscall_init(); /* This sets MSR_*STAR and related */
load_TR_desc(); /* This does ltr */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/traps-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/traps-xen.c
---- linux-2.6.18/arch/x86_64/kernel/traps-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/traps-xen.c 2007-12-23 11:15:00.112819947 +0100
-@@ -0,0 +1,1175 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/traps-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/traps-xen.c 2008-05-19 00:33:19.053244386 +0300
+@@ -0,0 +1,1173 @@
+/*
+ * linux/arch/x86-64/traps.c
+ *
+ * NB. All these are "interrupt gates" (i.e. events_mask is set) because we
+ * specify <dpl>|4 in the second field.
+ */
-+static trap_info_t trap_table[] = {
++static trap_info_t __cpuinitdata trap_table[] = {
+ { 0, 0|4, __KERNEL_CS, (unsigned long)divide_error },
+ { 1, 0|4, __KERNEL_CS, (unsigned long)debug },
+ { 3, 3|4, __KERNEL_CS, (unsigned long)int3 },
+ int ret;
+
+ ret = HYPERVISOR_set_trap_table(trap_table);
-+
+ if (ret)
-+ printk("HYPERVISOR_set_trap_table faild: error %d\n",
-+ ret);
++ printk("HYPERVISOR_set_trap_table failed: error %d\n", ret);
+
+ /*
+ * Should be a barrier for any external CPU state.
+ cpu_init();
+}
+
-+void smp_trap_init(trap_info_t *trap_ctxt)
++void __cpuinit smp_trap_init(trap_info_t *trap_ctxt)
+{
-+ trap_info_t *t = trap_table;
++ const trap_info_t *t = trap_table;
+
+ for (t = trap_table; t->address; t++) {
+ trap_ctxt[t->vector].flags = t->flags;
+}
+__setup("call_trace=", call_trace_setup);
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/vmlinux.lds.S linux-2.6.18-xen.hg/arch/x86_64/kernel/vmlinux.lds.S
---- linux-2.6.18/arch/x86_64/kernel/vmlinux.lds.S 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/vmlinux.lds.S 2007-12-23 11:15:00.112819947 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/vmlinux.lds.S 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/vmlinux.lds.S 2008-05-19 00:33:19.053244386 +0300
@@ -13,6 +13,13 @@
OUTPUT_ARCH(i386:x86-64)
ENTRY(phys_startup_64)
+
+ NOTES
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/vsyscall-xen.c linux-2.6.18-xen.hg/arch/x86_64/kernel/vsyscall-xen.c
---- linux-2.6.18/arch/x86_64/kernel/vsyscall-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/vsyscall-xen.c 2007-12-23 11:15:00.112819947 +0100
+--- linux-2.6.18.8/arch/x86_64/kernel/vsyscall-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/vsyscall-xen.c 2008-05-19 00:33:19.057244616 +0300
@@ -0,0 +1,227 @@
+/*
+ * linux/arch/x86_64/kernel/vsyscall.c
+}
+
+__initcall(vsyscall_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/kernel/xen_entry.S linux-2.6.18-xen.hg/arch/x86_64/kernel/xen_entry.S
---- linux-2.6.18/arch/x86_64/kernel/xen_entry.S 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/kernel/xen_entry.S 2007-12-23 11:15:00.116153454 +0100
-@@ -0,0 +1,40 @@
+--- linux-2.6.18.8/arch/x86_64/kernel/xen_entry.S 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/kernel/xen_entry.S 2008-05-19 00:33:19.057244616 +0300
+@@ -0,0 +1,36 @@
+/*
+ * Copied from arch/xen/i386/kernel/entry.S
+ */
+ XEN_LOCKED_UNBLOCK_EVENTS(reg) ; \
+ XEN_PUT_VCPU_INFO(reg)
+#define XEN_TEST_PENDING(reg) testb $0xFF,evtchn_upcall_pending(reg)
-+
-+VGCF_IN_SYSCALL = (1<<8)
-+
-+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/Makefile linux-2.6.18-xen.hg/arch/x86_64/Makefile
---- linux-2.6.18/arch/x86_64/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/Makefile 2007-12-23 11:15:00.042816270 +0100
-@@ -71,9 +71,22 @@
-
- boot := arch/x86_64/boot
-
--PHONY += bzImage bzlilo install archmrproper \
-+PHONY += bzImage bzlilo vmlinuz install archmrproper \
- fdimage fdimage144 fdimage288 isoimage archclean
-
-+ifdef CONFIG_XEN
-+CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
-+ -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(CPPFLAGS)
-+LDFLAGS_vmlinux := -e startup_64
-+#Default target when executing "make"
-+all: vmlinuz
-+
-+BOOTIMAGE := $(boot)/vmlinuz
-+KBUILD_IMAGE := $(BOOTIMAGE)
-+
-+vmlinuz: vmlinux
-+ $(Q)$(MAKE) $(build)=$(boot) $(BOOTIMAGE)
-+else
- #Default target when executing "make"
- all: bzImage
-
-@@ -91,6 +104,7 @@
-
- fdimage fdimage144 fdimage288 isoimage: vmlinux
- $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(BOOTIMAGE) $@
-+endif
+--- linux-2.6.18.8/arch/x86_64/lib/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/lib/Makefile 2008-05-19 00:33:19.061244847 +0300
+@@ -10,3 +10,4 @@
+ usercopy.o getuser.o putuser.o \
+ thunk.o clear_page.o copy_page.o bitstr.o bitops.o
+ lib-y += memcpy.o memmove.o memset.o copy_user.o
++lib-$(CONFIG_XEN_SCRUB_PAGES) += scrub.o
+--- linux-2.6.18.8/arch/x86_64/lib/scrub.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/lib/scrub.c 2008-05-19 00:33:19.097246922 +0300
+@@ -0,0 +1 @@
++#include "../../i386/lib/scrub.c"
+--- linux-2.6.18.8/arch/x86_64/mm/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/mm/Makefile 2008-05-19 00:33:19.101247152 +0300
+@@ -7,5 +7,8 @@
+ obj-$(CONFIG_NUMA) += numa.o
+ obj-$(CONFIG_K8_NUMA) += k8topology.o
+ obj-$(CONFIG_ACPI_NUMA) += srat.o
++obj-$(CONFIG_XEN) += hypervisor.o
- install:
- $(Q)$(MAKE) $(build)=$(boot) BOOTIMAGE=$(BOOTIMAGE) $@
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/mm/fault-xen.c linux-2.6.18-xen.hg/arch/x86_64/mm/fault-xen.c
---- linux-2.6.18/arch/x86_64/mm/fault-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/mm/fault-xen.c 2007-12-23 11:15:00.126153975 +0100
+ hugetlbpage-y = ../../i386/mm/hugetlbpage.o
++ioremap-$(CONFIG_XEN) := ../../i386/mm/ioremap.o
++hypervisor-y := ../../i386/mm/hypervisor.o
+--- linux-2.6.18.8/arch/x86_64/mm/fault-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/mm/fault-xen.c 2008-05-19 00:33:19.101247152 +0300
@@ -0,0 +1,724 @@
+/*
+ * linux/arch/x86-64/mm/fault.c
+ return 1;
+}
+__setup("pagefaulttrace", enable_pagefaulttrace);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/mm/init-xen.c linux-2.6.18-xen.hg/arch/x86_64/mm/init-xen.c
---- linux-2.6.18/arch/x86_64/mm/init-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/mm/init-xen.c 2007-12-23 11:15:00.126153975 +0100
+--- linux-2.6.18.8/arch/x86_64/mm/init-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/mm/init-xen.c 2008-05-19 00:33:19.105247383 +0300
@@ -0,0 +1,1205 @@
+/*
+ * linux/arch/x86_64/mm/init.c
+EXPORT_SYMBOL(__kernel_page_user);
+#endif
+
++int after_bootmem;
++
+extern unsigned long *contiguous_bitmap;
+
+static unsigned long dma_reserve __initdata;
+ pte_t pte, *ptep;
+ unsigned long *page = (unsigned long *) init_level4_pgt;
+
++ BUG_ON(after_bootmem);
++
+ if (xen_feature(feature))
+ return;
+
+ printk(KERN_INFO "%lu pages swap cached\n",cached);
+}
+
-+int after_bootmem;
+
+static __init void *spp_getpage(void)
+{
+
+#define PTE_SIZE PAGE_SIZE
+
-+static inline void __set_pte(pte_t *dst, pte_t val)
-+{
-+ *dst = val;
-+}
-+
+static inline int make_readonly(unsigned long paddr)
+{
+ extern char __vsyscall_0;
+ unsigned long pte_phys;
+ pte_t *pte, *pte_save;
+
-+ if (address >= end) {
-+ if (!after_bootmem)
-+ for (; i < PTRS_PER_PMD; i++, pmd++)
-+ set_pmd(pmd, __pmd(0));
++ if (address >= end)
+ break;
-+ }
+ pte = alloc_static_page(&pte_phys);
+ pte_save = pte;
+ for (k = 0; k < PTRS_PER_PTE; pte++, k++, address += PTE_SIZE) {
+ unsigned long pteval = address | _PAGE_NX | _KERNPG_TABLE;
+
-+ if ((address >= end) ||
-+ ((address >> PAGE_SHIFT) >=
-+ xen_start_info->nr_pages))
++ if (address >= (after_bootmem
++ ? end
++ : xen_start_info->nr_pages << PAGE_SHIFT))
+ pteval = 0;
+ else if (make_readonly(address))
+ pteval &= ~_PAGE_RW;
-+ __set_pte(pte, __pte(pteval & __supported_pte_mask));
++ set_pte(pte, __pte(pteval & __supported_pte_mask));
++ }
++ if (!after_bootmem) {
++ early_make_page_readonly(pte_save, XENFEAT_writable_page_tables);
++ *pmd = __pmd(pte_phys | _KERNPG_TABLE);
++ } else {
++ make_page_readonly(pte_save, XENFEAT_writable_page_tables);
++ set_pmd(pmd, __pmd(pte_phys | _KERNPG_TABLE));
+ }
-+ pte = pte_save;
-+ early_make_page_readonly(pte, XENFEAT_writable_page_tables);
-+ set_pmd(pmd, __pmd(pte_phys | _KERNPG_TABLE));
+ }
+}
+
+ break;
+
+ pmd = alloc_static_page(&pmd_phys);
-+ early_make_page_readonly(pmd, XENFEAT_writable_page_tables);
++
+ spin_lock(&init_mm.page_table_lock);
-+ set_pud(pud, __pud(pmd_phys | _KERNPG_TABLE));
++ *pud = __pud(pmd_phys | _KERNPG_TABLE);
+ phys_pmd_init(pmd, paddr, end);
+ spin_unlock(&init_mm.page_table_lock);
++
++ early_make_page_readonly(pmd, XENFEAT_writable_page_tables);
+ }
+ __flush_tlb();
+}
+
+ /* Kill mapping of low 1MB. */
+ while (va < (unsigned long)&_text) {
-+ HYPERVISOR_update_va_mapping(va, __pte_ma(0), 0);
++ if (HYPERVISOR_update_va_mapping(va, __pte_ma(0), 0))
++ BUG();
+ va += PAGE_SIZE;
+ }
+
+ pmd = (pmd_t *)&page[pmd_index(va)];
+ if (pmd_none(*pmd))
+ break;
-+ HYPERVISOR_update_va_mapping(va, __pte_ma(0), 0);
++ if (HYPERVISOR_update_va_mapping(va, __pte_ma(0), 0))
++ BUG();
+ va += PAGE_SIZE;
+ }
+}
+ start = PAGE_ALIGN((unsigned long)_end);
+ end = __START_KERNEL_map + (table_end << PAGE_SHIFT);
+ for (; start < end; start += PAGE_SIZE)
-+ WARN_ON(HYPERVISOR_update_va_mapping(
-+ start, __pte_ma(0), 0));
++ if (HYPERVISOR_update_va_mapping(start, __pte_ma(0), 0))
++ BUG();
+
+ /* Allocate pte's for initial fixmaps from 'start_pfn' allocator. */
+ table_end = ~0UL;
+ pgd_t *pgd = pgd_offset_k(start);
+ pud_t *pud;
+
-+ if (after_bootmem) {
++ if (after_bootmem)
+ pud = pud_offset(pgd, start & PGDIR_MASK);
-+ make_page_readonly(pud, XENFEAT_writable_page_tables);
-+ pud_phys = __pa(pud);
-+ } else {
++ else
+ pud = alloc_static_page(&pud_phys);
-+ early_make_page_readonly(pud, XENFEAT_writable_page_tables);
-+ }
+ next = start + PGDIR_SIZE;
+ if (next > end)
+ next = end;
+ phys_pud_init(pud, __pa(start), __pa(next));
-+ if (!after_bootmem)
++ if (!after_bootmem) {
++ early_make_page_readonly(pud, XENFEAT_writable_page_tables);
+ set_pgd(pgd_offset_k(start), mk_kernel_pgd(pud_phys));
++ }
+ }
+
+ if (!after_bootmem) {
+{
+ return (addr >= VSYSCALL_START) && (addr < VSYSCALL_END);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/mm/Makefile linux-2.6.18-xen.hg/arch/x86_64/mm/Makefile
---- linux-2.6.18/arch/x86_64/mm/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/mm/Makefile 2007-12-23 11:15:00.126153975 +0100
-@@ -7,5 +7,8 @@
- obj-$(CONFIG_NUMA) += numa.o
- obj-$(CONFIG_K8_NUMA) += k8topology.o
- obj-$(CONFIG_ACPI_NUMA) += srat.o
-+obj-$(CONFIG_XEN) += hypervisor.o
-
- hugetlbpage-y = ../../i386/mm/hugetlbpage.o
-+ioremap-$(CONFIG_XEN) := ../../i386/mm/ioremap.o
-+hypervisor-y := ../../i386/mm/hypervisor.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/mm/pageattr-xen.c linux-2.6.18-xen.hg/arch/x86_64/mm/pageattr-xen.c
---- linux-2.6.18/arch/x86_64/mm/pageattr-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/arch/x86_64/mm/pageattr-xen.c 2007-12-23 11:15:00.129487487 +0100
+--- linux-2.6.18.8/arch/x86_64/mm/pageattr-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/mm/pageattr-xen.c 2008-05-19 00:33:19.553273207 +0300
@@ -0,0 +1,502 @@
+/*
+ * Copyright 2002 Andi Kleen, SuSE Labs.
+}
+#endif /* CONFIG_XEN */
+
-+static inline pte_t *lookup_address(unsigned long address)
++pte_t *lookup_address(unsigned long address)
+{
+ pgd_t *pgd = pgd_offset_k(address);
+ pud_t *pud;
+
+EXPORT_SYMBOL(change_page_attr);
+EXPORT_SYMBOL(global_flush_tlb);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/oprofile/Makefile linux-2.6.18-xen.hg/arch/x86_64/oprofile/Makefile
---- linux-2.6.18/arch/x86_64/oprofile/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/oprofile/Makefile 2007-12-23 11:15:00.132820998 +0100
+--- linux-2.6.18.8/arch/x86_64/oprofile/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/oprofile/Makefile 2008-05-19 00:33:19.557273438 +0300
@@ -11,9 +11,15 @@
oprofilefs.o oprofile_stats.o \
timer_int.o )
+endif
+oprofile-y = $(DRIVER_OBJS) $(XENOPROF_COMMON_OBJS) \
+ $(addprefix ../../i386/oprofile/, $(OPROFILE-y))
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/arch/x86_64/pci/Makefile linux-2.6.18-xen.hg/arch/x86_64/pci/Makefile
---- linux-2.6.18/arch/x86_64/pci/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/arch/x86_64/pci/Makefile 2007-12-23 11:15:00.132820998 +0100
+--- linux-2.6.18.8/arch/x86_64/pci/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/arch/x86_64/pci/Makefile 2008-05-19 00:33:19.557273438 +0300
@@ -15,8 +15,13 @@
obj-$(CONFIG_NUMA) += k8-bus.o
legacy-y += ../../i386/pci/legacy.o
irq-y += ../../i386/pci/irq.o
common-y += ../../i386/pci/common.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/conf.linux-native/00_xen_to_native linux-2.6.18-xen.hg/buildconfigs/conf.linux-native/00_xen_to_native
---- linux-2.6.18/buildconfigs/conf.linux-native/00_xen_to_native 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/conf.linux-native/00_xen_to_native 2007-12-23 11:15:00.629513738 +0100
+--- linux-2.6.18.8/block/elevator.c 2008-05-19 00:42:33.885229044 +0300
++++ linux-2.6.18-xen.hg/block/elevator.c 2008-05-19 00:33:19.697281508 +0300
+@@ -493,6 +493,16 @@
+ int ret;
+
+ while ((rq = __elv_next_request(q)) != NULL) {
++ /*
++ * Kill the empty barrier place holder, the driver must
++ * not ever see it.
++ */
++ if (blk_empty_barrier(rq)) {
++ blkdev_dequeue_request(rq);
++ end_that_request_chunk(rq, 1, 0);
++ end_that_request_last(rq, 1);
++ continue;
++ }
+ if (!(rq->flags & REQ_STARTED)) {
+ elevator_t *e = q->elevator;
+
+--- linux-2.6.18.8/block/ll_rw_blk.c 2008-05-19 00:42:33.885229044 +0300
++++ linux-2.6.18-xen.hg/block/ll_rw_blk.c 2008-05-19 00:33:19.701281739 +0300
+@@ -483,9 +483,12 @@
+ * Queue ordered sequence. As we stack them at the head, we
+ * need to queue in reverse order. Note that we rely on that
+ * no fs request uses ELEVATOR_INSERT_FRONT and thus no fs
+- * request gets inbetween ordered sequence.
++ * request gets inbetween ordered sequence. If this request is
++ * an empty barrier, we don't need to do a postflush ever since
++ * there will be no data written between the pre and post flush.
++ * Hence a single flush will suffice.
+ */
+- if (q->ordered & QUEUE_ORDERED_POSTFLUSH)
++ if ((q->ordered & QUEUE_ORDERED_POSTFLUSH) && !blk_empty_barrier(rq))
+ queue_flush(q, QUEUE_ORDERED_POSTFLUSH);
+ else
+ q->ordseq |= QUEUE_ORDSEQ_POSTFLUSH;
+@@ -2967,7 +2970,7 @@
+ {
+ struct block_device *bdev = bio->bi_bdev;
+
+- if (bdev != bdev->bd_contains) {
++ if (bio_sectors(bio) && bdev != bdev->bd_contains) {
+ struct hd_struct *p = bdev->bd_part;
+ const int rw = bio_data_dir(bio);
+
+@@ -3028,7 +3031,7 @@
+ might_sleep();
+ /* Test device or partition size, when known. */
+ maxsector = bio->bi_bdev->bd_inode->i_size >> 9;
+- if (maxsector) {
++ if (maxsector && nr_sectors) {
+ sector_t sector = bio->bi_sector;
+
+ if (maxsector < nr_sectors || maxsector - nr_sectors < sector) {
+@@ -3094,7 +3097,7 @@
+ old_dev = bio->bi_bdev->bd_dev;
+
+ maxsector = bio->bi_bdev->bd_inode->i_size >> 9;
+- if (maxsector) {
++ if (maxsector && nr_sectors) {
+ sector_t sector = bio->bi_sector;
+
+ if (maxsector < nr_sectors || maxsector - nr_sectors < sector) {
+@@ -3128,9 +3131,12 @@
+ {
+ int count = bio_sectors(bio);
+
++ bio->bi_rw |= rw;
++
++ if (!bio_empty_barrier(bio)) {
+ BIO_BUG_ON(!bio->bi_size);
+ BIO_BUG_ON(!bio->bi_io_vec);
+- bio->bi_rw |= rw;
++
+ if (rw & WRITE)
+ count_vm_events(PGPGOUT, count);
+ else
+@@ -3144,6 +3150,7 @@
+ (unsigned long long)bio->bi_sector,
+ bdevname(bio->bi_bdev,b));
+ }
++ }
+
+ generic_make_request(bio);
+ }
+@@ -3260,6 +3267,13 @@
+ while ((bio = req->bio) != NULL) {
+ int nbytes;
+
++ /* For an empty barrier request, the low level driver must
++ * store a potential error location in ->sector. We pass
++ * that back up in ->bi_sector
++ */
++ if (blk_empty_barrier(req))
++ bio->bi_sector = req->sector;
++
+ if (nr_bytes >= bio->bi_size) {
+ req->bio = bio->bi_next;
+ nbytes = bio->bi_size;
+--- linux-2.6.18.8/buildconfigs/conf.linux-native/00_xen_to_native 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/conf.linux-native/00_xen_to_native 2008-05-19 00:33:19.705281969 +0300
@@ -0,0 +1,86 @@
+# Linux kernel version: 2.6.16.13-native
+# Mon May 15 10:59:54 2006
+CONFIG_X86_HT=y
+# CONFIG_X86_NO_TSS is not set
+# CONFIG_X86_NO_IDT is not set
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/create_config.sh linux-2.6.18-xen.hg/buildconfigs/create_config.sh
---- linux-2.6.18/buildconfigs/create_config.sh 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/create_config.sh 2007-12-23 11:15:00.629513738 +0100
+--- linux-2.6.18.8/buildconfigs/create_config.sh 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/create_config.sh 2008-05-19 00:33:19.705281969 +0300
@@ -0,0 +1,54 @@
+#!/bin/sh
+set -e
+ done
+ fi
+done
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen0_ia64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_ia64
---- linux-2.6.18/buildconfigs/linux-defconfig_xen0_ia64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_ia64 2007-12-23 11:15:00.636180766 +0100
-@@ -0,0 +1,1697 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen0_ia64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_ia64 2008-05-19 00:33:19.709282200 +0300
+@@ -0,0 +1,1701 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Mon Dec 17 13:38:33 2007
++# Tue Feb 19 11:20:09 2008
+#
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+# CONFIG_IXGB is not set
+# CONFIG_S2IO is not set
+# CONFIG_MYRI10GE is not set
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+CONFIG_XEN_BLKDEV_TAP=y
+CONFIG_XEN_NETDEV_BACKEND=y
+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
+# CONFIG_XEN_NETDEV_LOOPBACK is not set
+CONFIG_XEN_PCIDEV_BACKEND=y
+# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
+CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
+CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_KEYBOARD=y
+# CONFIG_XEN_SCRUB_PAGES is not set
+CONFIG_NO_IDLE_HZ=y
+CONFIG_XEN_BALLOON=y
+CONFIG_XEN_XENCOMM=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen0_x86_32 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_32
---- linux-2.6.18/buildconfigs/linux-defconfig_xen0_x86_32 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_32 2007-12-23 11:15:00.659515321 +0100
-@@ -0,0 +1,1451 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen0_x86_32 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_32 2008-05-19 00:33:19.709282200 +0300
+@@ -0,0 +1,1456 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
+CONFIG_XEN_PCIDEV_FRONTEND=y
+# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
+# CONFIG_PCIEPORTBUS is not set
++# CONFIG_PCI_MSI is not set
+# CONFIG_PCI_DEBUG is not set
+CONFIG_ISA_DMA_API=y
+# CONFIG_SCx200 is not set
+# CONFIG_IXGB is not set
+# CONFIG_S2IO is not set
+# CONFIG_MYRI10GE is not set
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+#
+# CONFIG_CRYPTO_DEV_PADLOCK is not set
+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+CONFIG_XEN_BLKDEV_TAP=y
+CONFIG_XEN_NETDEV_BACKEND=y
+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
+# CONFIG_XEN_NETDEV_LOOPBACK is not set
+CONFIG_XEN_PCIDEV_BACKEND=y
+# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
+CONFIG_XEN_TPMDEV_BACKEND=m
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
++CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
+CONFIG_XEN_SCRUB_PAGES=y
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_SMPBOOT=y
+CONFIG_XEN_BALLOON=y
+CONFIG_XEN_DEVMEM=y
-+CONFIG_XEN_GRANT_DEV=y
+
+#
+# Library routines
+CONFIG_X86_NO_TSS=y
+CONFIG_X86_NO_IDT=y
+CONFIG_KTIME_SCALAR=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen0_x86_64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_64
---- linux-2.6.18/buildconfigs/linux-defconfig_xen0_x86_64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_64 2007-12-23 11:15:00.659515321 +0100
-@@ -0,0 +1,1389 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen0_x86_64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen0_x86_64 2008-05-19 00:33:19.709282200 +0300
+@@ -0,0 +1,1388 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Tue Oct 16 09:32:39 2007
++# Mon Feb 18 10:41:04 2008
+#
+CONFIG_X86_64=y
+CONFIG_64BIT=y
+# Code maturity level options
+#
+CONFIG_EXPERIMENTAL=y
-+CONFIG_BROKEN_ON_SMP=y
++CONFIG_LOCK_KERNEL=y
+CONFIG_INIT_ENV_ARG_LIMIT=32
+
+#
+# CONFIG_TASKSTATS is not set
+# CONFIG_AUDIT is not set
+# CONFIG_IKCONFIG is not set
++# CONFIG_CPUSETS is not set
+# CONFIG_RELAY is not set
+CONFIG_INITRAMFS_SOURCE=""
+# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
+# CONFIG_MODVERSIONS is not set
+# CONFIG_MODULE_SRCVERSION_ALL is not set
+CONFIG_KMOD=y
++CONFIG_STOP_MACHINE=y
+
+#
+# Block layer
+CONFIG_X86_INTERNODE_CACHE_BYTES=128
+CONFIG_X86_GOOD_APIC=y
+CONFIG_MICROCODE=y
-+# CONFIG_X86_MSR is not set
-+# CONFIG_X86_CPUID is not set
++CONFIG_X86_MSR=y
++CONFIG_X86_CPUID=y
+CONFIG_X86_IO_APIC=y
+CONFIG_X86_XEN_GENAPIC=y
+CONFIG_X86_LOCAL_APIC=y
+CONFIG_MTRR=y
-+# CONFIG_SMP is not set
++CONFIG_SMP=y
+CONFIG_PREEMPT_NONE=y
+# CONFIG_PREEMPT_VOLUNTARY is not set
+# CONFIG_PREEMPT is not set
++CONFIG_PREEMPT_BKL=y
+CONFIG_ARCH_FLATMEM_ENABLE=y
+CONFIG_SELECT_MEMORY_MODEL=y
+CONFIG_FLATMEM_MANUAL=y
+# CONFIG_SPARSEMEM_STATIC is not set
+CONFIG_SPLIT_PTLOCK_CPUS=4
+CONFIG_RESOURCES_64BIT=y
++CONFIG_NR_CPUS=8
++CONFIG_HOTPLUG_CPU=y
+CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
+CONFIG_SWIOTLB=y
+CONFIG_KEXEC=y
+# CONFIG_HZ_1000 is not set
+CONFIG_HZ=100
+# CONFIG_REORDER is not set
-+CONFIG_K8_NB=y
+CONFIG_GENERIC_HARDIRQS=y
+CONFIG_GENERIC_IRQ_PROBE=y
+CONFIG_ISA_DMA_API=y
++CONFIG_GENERIC_PENDING_IRQ=y
+
+#
+# Power management options
+CONFIG_PM_LEGACY=y
+# CONFIG_PM_DEBUG is not set
+# CONFIG_SOFTWARE_SUSPEND is not set
++CONFIG_SUSPEND_SMP=y
+
+#
+# ACPI (Advanced Configuration and Power Interface) Support
+CONFIG_ACPI_BATTERY=m
+CONFIG_ACPI_BUTTON=m
+CONFIG_ACPI_VIDEO=m
-+# CONFIG_ACPI_HOTKEY is not set
++CONFIG_ACPI_HOTKEY=m
+CONFIG_ACPI_FAN=m
+CONFIG_ACPI_DOCK=m
+CONFIG_ACPI_PROCESSOR=m
++CONFIG_ACPI_HOTPLUG_CPU=y
+CONFIG_ACPI_THERMAL=m
+CONFIG_ACPI_ASUS=m
+CONFIG_ACPI_IBM=m
+CONFIG_ACPI_TOSHIBA=m
++# CONFIG_ACPI_CUSTOM_DSDT is not set
+CONFIG_ACPI_BLACKLIST_YEAR=0
+# CONFIG_ACPI_DEBUG is not set
+CONFIG_ACPI_EC=y
+CONFIG_ACPI_POWER=y
+CONFIG_ACPI_SYSTEM=y
-+# CONFIG_ACPI_CONTAINER is not set
++CONFIG_ACPI_CONTAINER=m
+CONFIG_ACPI_PV_SLEEP=y
+
+#
+#
+CONFIG_PCI=y
+CONFIG_PCI_DIRECT=y
-+# CONFIG_PCI_MMCONFIG is not set
++CONFIG_PCI_MMCONFIG=y
+CONFIG_XEN_PCIDEV_FRONTEND=y
+# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
+# CONFIG_PCIEPORTBUS is not set
++# CONFIG_PCI_MSI is not set
+# CONFIG_PCI_DEBUG is not set
+
+#
+# Executable file formats / Emulations
+#
+CONFIG_BINFMT_ELF=y
-+CONFIG_BINFMT_MISC=y
++# CONFIG_BINFMT_MISC is not set
+CONFIG_IA32_EMULATION=y
-+# CONFIG_IA32_AOUT is not set
++CONFIG_IA32_AOUT=y
+CONFIG_COMPAT=y
+CONFIG_SYSVIPC_COMPAT=y
+
+CONFIG_PACKET=y
+# CONFIG_PACKET_MMAP is not set
+CONFIG_UNIX=y
++CONFIG_XFRM=y
++# CONFIG_XFRM_USER is not set
+# CONFIG_NET_KEY is not set
+CONFIG_INET=y
+# CONFIG_IP_MULTICAST is not set
+# CONFIG_INET_IPCOMP is not set
+# CONFIG_INET_XFRM_TUNNEL is not set
+# CONFIG_INET_TUNNEL is not set
-+# CONFIG_INET_XFRM_MODE_TRANSPORT is not set
-+# CONFIG_INET_XFRM_MODE_TUNNEL is not set
++CONFIG_INET_XFRM_MODE_TRANSPORT=y
++CONFIG_INET_XFRM_MODE_TUNNEL=y
+# CONFIG_INET_DIAG is not set
+# CONFIG_TCP_CONG_ADVANCED is not set
+CONFIG_TCP_CONG_BIC=y
+#
+# Generic Driver Options
+#
-+CONFIG_STANDALONE=y
-+# CONFIG_PREVENT_FIRMWARE_BUILD is not set
++# CONFIG_STANDALONE is not set
++CONFIG_PREVENT_FIRMWARE_BUILD=y
+# CONFIG_FW_LOADER is not set
+# CONFIG_DEBUG_DRIVER is not set
+CONFIG_SYS_HYPERVISOR=y
+# CONFIG_BLK_DEV_UB is not set
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_RAM_COUNT=16
-+CONFIG_BLK_DEV_RAM_SIZE=16384
++CONFIG_BLK_DEV_RAM_SIZE=4096
+CONFIG_BLK_DEV_RAM_BLOCKSIZE=1024
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_CDROM_PKTCDVD is not set
+#
+CONFIG_SCSI_SPI_ATTRS=y
+# CONFIG_SCSI_FC_ATTRS is not set
-+CONFIG_SCSI_ISCSI_ATTRS=y
++# CONFIG_SCSI_ISCSI_ATTRS is not set
+# CONFIG_SCSI_SAS_ATTRS is not set
+
+#
+CONFIG_SCSI_AIC7XXX=y
+CONFIG_AIC7XXX_CMDS_PER_DEVICE=32
+CONFIG_AIC7XXX_RESET_DELAY_MS=15000
-+# CONFIG_AIC7XXX_BUILD_FIRMWARE is not set
+CONFIG_AIC7XXX_DEBUG_ENABLE=y
+CONFIG_AIC7XXX_DEBUG_MASK=0
+CONFIG_AIC7XXX_REG_PRETTY_PRINT=y
+CONFIG_SCSI_AIC79XX=y
+CONFIG_AIC79XX_CMDS_PER_DEVICE=32
+CONFIG_AIC79XX_RESET_DELAY_MS=15000
-+# CONFIG_AIC79XX_BUILD_FIRMWARE is not set
+# CONFIG_AIC79XX_ENABLE_RD_STRM is not set
+CONFIG_AIC79XX_DEBUG_ENABLE=y
+CONFIG_AIC79XX_DEBUG_MASK=0
+# CONFIG_SCSI_SATA_VIA is not set
+# CONFIG_SCSI_SATA_VITESSE is not set
+CONFIG_SCSI_SATA_INTEL_COMBINED=y
-+CONFIG_SCSI_BUSLOGIC=y
-+# CONFIG_SCSI_OMIT_FLASHPOINT is not set
++# CONFIG_SCSI_BUSLOGIC is not set
+# CONFIG_SCSI_DMX3191D is not set
+# CONFIG_SCSI_EATA is not set
+# CONFIG_SCSI_FUTURE_DOMAIN is not set
+#
+CONFIG_MD=y
+CONFIG_BLK_DEV_MD=y
-+CONFIG_MD_LINEAR=y
++# CONFIG_MD_LINEAR is not set
+CONFIG_MD_RAID0=y
+CONFIG_MD_RAID1=y
+# CONFIG_MD_RAID10 is not set
-+CONFIG_MD_RAID456=y
-+# CONFIG_MD_RAID5_RESHAPE is not set
-+CONFIG_MD_MULTIPATH=y
++# CONFIG_MD_RAID456 is not set
++# CONFIG_MD_MULTIPATH is not set
+# CONFIG_MD_FAULTY is not set
+CONFIG_BLK_DEV_DM=y
-+CONFIG_DM_CRYPT=y
++# CONFIG_DM_CRYPT is not set
+CONFIG_DM_SNAPSHOT=y
+CONFIG_DM_MIRROR=y
+# CONFIG_DM_ZERO is not set
-+CONFIG_DM_MULTIPATH=y
-+CONFIG_DM_MULTIPATH_EMC=y
++# CONFIG_DM_MULTIPATH is not set
+
+#
+# Fusion MPT device support
+# CONFIG_IXGB is not set
+# CONFIG_S2IO is not set
+# CONFIG_MYRI10GE is not set
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+CONFIG_VT=y
+CONFIG_VT_CONSOLE=y
+CONFIG_HW_CONSOLE=y
-+# CONFIG_VT_HW_CONSOLE_BINDING is not set
++CONFIG_VT_HW_CONSOLE_BINDING=y
+# CONFIG_SERIAL_NONSTANDARD is not set
+
+#
+# CONFIG_WATCHDOG is not set
+# CONFIG_HW_RANDOM is not set
+# CONFIG_NVRAM is not set
-+CONFIG_RTC=y
++# CONFIG_RTC is not set
++# CONFIG_GEN_RTC is not set
+# CONFIG_DTLK is not set
+# CONFIG_R3964 is not set
+# CONFIG_APPLICOM is not set
+#
+# Ftape, the floppy tape device driver
+#
-+# CONFIG_FTAPE is not set
+CONFIG_AGP=m
-+CONFIG_AGP_AMD64=m
-+# CONFIG_AGP_INTEL is not set
++# CONFIG_AGP_AMD64 is not set
++CONFIG_AGP_INTEL=m
+CONFIG_AGP_SIS=m
+CONFIG_AGP_VIA=m
+CONFIG_DRM=m
+CONFIG_DRM_TDFX=m
+CONFIG_DRM_R128=m
+CONFIG_DRM_RADEON=m
++CONFIG_DRM_I810=m
++CONFIG_DRM_I830=m
++CONFIG_DRM_I915=m
+CONFIG_DRM_MGA=m
+CONFIG_DRM_SIS=m
+# CONFIG_DRM_VIA is not set
+#
+# LED devices
+#
-+CONFIG_NEW_LEDS=y
-+CONFIG_LEDS_CLASS=m
++# CONFIG_NEW_LEDS is not set
+
+#
+# LED drivers
+#
+# LED Triggers
+#
-+CONFIG_LEDS_TRIGGERS=y
-+CONFIG_LEDS_TRIGGER_TIMER=y
-+CONFIG_LEDS_TRIGGER_IDE_DISK=y
-+CONFIG_LEDS_TRIGGER_HEARTBEAT=y
+
+#
+# InfiniBand support
+#
-+CONFIG_INFINIBAND=y
-+# CONFIG_INFINIBAND_USER_MAD is not set
-+# CONFIG_INFINIBAND_USER_ACCESS is not set
-+CONFIG_INFINIBAND_ADDR_TRANS=y
-+CONFIG_INFINIBAND_MTHCA=y
-+CONFIG_INFINIBAND_MTHCA_DEBUG=y
-+CONFIG_INFINIBAND_IPOIB=y
-+CONFIG_INFINIBAND_IPOIB_DEBUG=y
-+CONFIG_INFINIBAND_IPOIB_DEBUG_DATA=y
-+CONFIG_INFINIBAND_SRP=y
-+CONFIG_INFINIBAND_ISER=y
++# CONFIG_INFINIBAND is not set
+
+#
+# EDAC - error detection and reporting (RAS) (EXPERIMENTAL)
+#
+# RTC drivers
+#
-+CONFIG_RTC_DRV_DS1553=m
-+CONFIG_RTC_DRV_DS1742=m
++# CONFIG_RTC_DRV_DS1553 is not set
++# CONFIG_RTC_DRV_DS1742 is not set
+CONFIG_RTC_DRV_M48T86=m
+CONFIG_RTC_DRV_TEST=m
-+CONFIG_RTC_DRV_V3020=m
++# CONFIG_RTC_DRV_V3020 is not set
+
+#
+# DMA Engine support
+#
+# DOS/FAT/NT Filesystems
+#
-+CONFIG_FAT_FS=y
-+CONFIG_MSDOS_FS=y
-+CONFIG_VFAT_FS=y
++CONFIG_FAT_FS=m
++CONFIG_MSDOS_FS=m
++CONFIG_VFAT_FS=m
+CONFIG_FAT_DEFAULT_CODEPAGE=437
+CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1"
+# CONFIG_NTFS_FS is not set
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_UNUSED_SYMBOLS=y
+CONFIG_DEBUG_KERNEL=y
-+CONFIG_LOG_BUF_SHIFT=15
++CONFIG_LOG_BUF_SHIFT=14
+CONFIG_DETECT_SOFTLOCKUP=y
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_DEBUG_SLAB is not set
+# CONFIG_DEBUG_SPINLOCK_SLEEP is not set
+# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
+# CONFIG_DEBUG_KOBJECT is not set
-+# CONFIG_DEBUG_INFO is not set
++CONFIG_DEBUG_INFO=y
+# CONFIG_DEBUG_FS is not set
+# CONFIG_DEBUG_VM is not set
+CONFIG_FRAME_POINTER=y
+# Hardware crypto devices
+#
+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+CONFIG_XEN_BLKDEV_TAP=y
+CONFIG_XEN_NETDEV_BACKEND=y
+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
+# CONFIG_XEN_NETDEV_LOOPBACK is not set
+CONFIG_XEN_PCIDEV_BACKEND=y
+# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
+# CONFIG_XEN_PCIDEV_BACKEND_SLOT is not set
+# CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER is not set
+# CONFIG_XEN_PCIDEV_BE_DEBUG is not set
-+CONFIG_XEN_TPMDEV_BACKEND=m
++# CONFIG_XEN_TPMDEV_BACKEND is not set
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
++CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
+CONFIG_XEN_SCRUB_PAGES=y
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_COMPAT=0x030002
+CONFIG_HAVE_IRQ_IGNORE_UNHANDLED=y
+CONFIG_NO_IDLE_HZ=y
++CONFIG_XEN_SMPBOOT=y
+CONFIG_XEN_BALLOON=y
+CONFIG_XEN_DEVMEM=y
-+CONFIG_XEN_GRANT_DEV=y
+
+#
+# Library routines
+# CONFIG_CRC_CCITT is not set
+# CONFIG_CRC16 is not set
+CONFIG_CRC32=y
-+CONFIG_LIBCRC32C=m
++CONFIG_LIBCRC32C=y
+CONFIG_ZLIB_INFLATE=y
+CONFIG_PLIST=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen_ia64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_ia64
---- linux-2.6.18/buildconfigs/linux-defconfig_xen_ia64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_ia64 2007-12-23 11:15:00.709517947 +0100
-@@ -0,0 +1,1697 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xenU_ia64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_ia64 2008-05-19 00:33:19.713282430 +0300
+@@ -0,0 +1,1510 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Mon Dec 17 13:38:29 2007
++# Tue Feb 19 11:20:14 2008
+#
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+CONFIG_LOCALVERSION_AUTO=y
+CONFIG_SWAP=y
+CONFIG_SYSVIPC=y
-+CONFIG_POSIX_MQUEUE=y
++# CONFIG_POSIX_MQUEUE is not set
+CONFIG_BSD_PROCESS_ACCT=y
+# CONFIG_BSD_PROCESS_ACCT_V3 is not set
+# CONFIG_TASKSTATS is not set
+# CONFIG_AUDIT is not set
-+CONFIG_IKCONFIG=y
-+CONFIG_IKCONFIG_PROC=y
++# CONFIG_IKCONFIG is not set
+# CONFIG_CPUSETS is not set
+# CONFIG_RELAY is not set
+CONFIG_INITRAMFS_SOURCE=""
+# CONFIG_EMBEDDED is not set
+CONFIG_SYSCTL=y
+CONFIG_KALLSYMS=y
-+CONFIG_KALLSYMS_ALL=y
-+CONFIG_KALLSYMS_EXTRA_PASS=y
++# CONFIG_KALLSYMS_ALL is not set
++# CONFIG_KALLSYMS_EXTRA_PASS is not set
+CONFIG_HOTPLUG=y
+CONFIG_PRINTK=y
+CONFIG_BUG=y
+# Loadable module support
+#
+CONFIG_MODULES=y
-+CONFIG_MODULE_UNLOAD=y
-+# CONFIG_MODULE_FORCE_UNLOAD is not set
-+CONFIG_MODVERSIONS=y
-+CONFIG_MODULE_SRCVERSION_ALL=y
-+CONFIG_KMOD=y
++# CONFIG_MODULE_UNLOAD is not set
++# CONFIG_MODVERSIONS is not set
++# CONFIG_MODULE_SRCVERSION_ALL is not set
++# CONFIG_KMOD is not set
+CONFIG_STOP_MACHINE=y
+
+#
+CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER=y
+CONFIG_DMA_IS_DMA32=y
+CONFIG_AUDIT_ARCH=y
-+CONFIG_IA64_GENERIC=y
++# CONFIG_IA64_GENERIC is not set
+# CONFIG_IA64_DIG is not set
+# CONFIG_IA64_HP_ZX1 is not set
+# CONFIG_IA64_HP_ZX1_SWIOTLB is not set
+# CONFIG_IA64_SGI_SN2 is not set
+# CONFIG_IA64_HP_SIM is not set
-+# CONFIG_IA64_XEN is not set
++CONFIG_IA64_XEN=y
+# CONFIG_ITANIUM is not set
+CONFIG_MCKINLEY=y
+# CONFIG_IA64_PAGE_SIZE_4KB is not set
+# CONFIG_HZ_1000 is not set
+CONFIG_HZ=100
+CONFIG_IA64_L1_CACHE_SHIFT=7
-+CONFIG_IA64_CYCLONE=y
++# CONFIG_IA64_CYCLONE is not set
+CONFIG_IOSAPIC=y
-+# CONFIG_IA64_SGI_SN_XP is not set
+CONFIG_FORCE_MAX_ZONEORDER=11
+CONFIG_SMP=y
+CONFIG_NR_CPUS=16
+# CONFIG_PERMIT_BSP_REMOVE is not set
+# CONFIG_PREEMPT is not set
+CONFIG_SELECT_MEMORY_MODEL=y
-+# CONFIG_FLATMEM_MANUAL is not set
-+CONFIG_DISCONTIGMEM_MANUAL=y
++CONFIG_FLATMEM_MANUAL=y
++# CONFIG_DISCONTIGMEM_MANUAL is not set
+# CONFIG_SPARSEMEM_MANUAL is not set
-+CONFIG_DISCONTIGMEM=y
++CONFIG_FLATMEM=y
+CONFIG_FLAT_NODE_MEM_MAP=y
-+CONFIG_NEED_MULTIPLE_NODES=y
+# CONFIG_SPARSEMEM_STATIC is not set
+CONFIG_SPLIT_PTLOCK_CPUS=4
-+# CONFIG_MIGRATION is not set
+CONFIG_RESOURCES_64BIT=y
+CONFIG_ARCH_SELECT_MEMORY_MODEL=y
+CONFIG_ARCH_DISCONTIGMEM_ENABLE=y
+CONFIG_ARCH_FLATMEM_ENABLE=y
+CONFIG_ARCH_SPARSEMEM_ENABLE=y
-+CONFIG_ARCH_DISCONTIGMEM_DEFAULT=y
-+CONFIG_NUMA=y
-+CONFIG_NODES_SHIFT=10
+CONFIG_VIRTUAL_MEM_MAP=y
+CONFIG_HOLES_IN_ZONE=y
-+CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID=y
-+CONFIG_HAVE_ARCH_NODEDATA_EXTENSION=y
+# CONFIG_IA32_SUPPORT is not set
-+# CONFIG_IA64_MCA_RECOVERY is not set
++CONFIG_IA64_MCA_RECOVERY=y
+CONFIG_PERFMON=y
+CONFIG_IA64_PALINFO=y
-+CONFIG_SGI_SN=y
-+
-+#
-+# SN Devices
-+#
-+# CONFIG_SGI_IOC3 is not set
-+CONFIG_KEXEC=y
++# CONFIG_KEXEC is not set
++# CONFIG_CRASH_DUMP is not set
+
+#
+# Firmware Drivers
+CONFIG_ACPI_PROCESSOR=y
+CONFIG_ACPI_HOTPLUG_CPU=y
+CONFIG_ACPI_THERMAL=y
-+CONFIG_ACPI_NUMA=y
+CONFIG_ACPI_BLACKLIST_YEAR=0
+# CONFIG_ACPI_DEBUG is not set
+CONFIG_ACPI_EC=y
+# CONFIG_HOTPLUG_PCI_ACPI_IBM is not set
+# CONFIG_HOTPLUG_PCI_CPCI is not set
+# CONFIG_HOTPLUG_PCI_SHPC is not set
-+# CONFIG_HOTPLUG_PCI_SGI is not set
+
+#
+# PCCARD (PCMCIA/CardBus) support
+# CONFIG_NET_IPIP is not set
+# CONFIG_NET_IPGRE is not set
+# CONFIG_IP_MROUTE is not set
-+CONFIG_ARPD=y
-+CONFIG_SYN_COOKIES=y
++# CONFIG_ARPD is not set
++# CONFIG_SYN_COOKIES is not set
+# CONFIG_INET_AH is not set
+# CONFIG_INET_ESP is not set
+# CONFIG_INET_IPCOMP is not set
+# CONFIG_NETWORK_SECMARK is not set
+CONFIG_NETFILTER=y
+# CONFIG_NETFILTER_DEBUG is not set
-+CONFIG_BRIDGE_NETFILTER=y
+
+#
+# Core Netfilter Configuration
+# CONFIG_IP_NF_QUEUE is not set
+
+#
-+# Bridge: Netfilter Configuration
-+#
-+# CONFIG_BRIDGE_NF_EBTABLES is not set
-+
-+#
+# DCCP Configuration (EXPERIMENTAL)
+#
+# CONFIG_IP_DCCP is not set
+#
+# CONFIG_TIPC is not set
+# CONFIG_ATM is not set
-+CONFIG_BRIDGE=y
++# CONFIG_BRIDGE is not set
+# CONFIG_VLAN_8021Q is not set
+# CONFIG_DECNET is not set
-+CONFIG_LLC=y
+# CONFIG_LLC2 is not set
+# CONFIG_IPX is not set
+# CONFIG_ATALK is not set
+#
+CONFIG_STANDALONE=y
+CONFIG_PREVENT_FIRMWARE_BUILD=y
-+CONFIG_FW_LOADER=y
++# CONFIG_FW_LOADER is not set
+# CONFIG_DEBUG_DRIVER is not set
+CONFIG_SYS_HYPERVISOR=y
+
+# Block devices
+#
+# CONFIG_BLK_CPQ_DA is not set
-+CONFIG_BLK_CPQ_CISS_DA=y
-+# CONFIG_CISS_SCSI_TAPE is not set
++# CONFIG_BLK_CPQ_CISS_DA is not set
+# CONFIG_BLK_DEV_DAC960 is not set
+# CONFIG_BLK_DEV_UMEM is not set
+# CONFIG_BLK_DEV_COW_COMMON is not set
+CONFIG_BLK_DEV_LOOP=y
-+CONFIG_BLK_DEV_CRYPTOLOOP=y
-+CONFIG_BLK_DEV_NBD=m
++# CONFIG_BLK_DEV_CRYPTOLOOP is not set
++# CONFIG_BLK_DEV_NBD is not set
+# CONFIG_BLK_DEV_SX8 is not set
+# CONFIG_BLK_DEV_UB is not set
+CONFIG_BLK_DEV_RAM=y
+#
+# ATA/ATAPI/MFM/RLL support
+#
-+CONFIG_IDE=y
-+CONFIG_IDE_MAX_HWIFS=4
-+CONFIG_BLK_DEV_IDE=y
-+
-+#
-+# Please see Documentation/ide.txt for help/info on IDE drives
-+#
-+# CONFIG_BLK_DEV_IDE_SATA is not set
-+CONFIG_BLK_DEV_IDEDISK=y
-+# CONFIG_IDEDISK_MULTI_MODE is not set
-+CONFIG_BLK_DEV_IDECD=y
-+# CONFIG_BLK_DEV_IDETAPE is not set
-+CONFIG_BLK_DEV_IDEFLOPPY=y
-+CONFIG_BLK_DEV_IDESCSI=y
-+# CONFIG_IDE_TASK_IOCTL is not set
-+
-+#
-+# IDE chipset support/bugfixes
-+#
-+# CONFIG_IDE_GENERIC is not set
-+CONFIG_BLK_DEV_IDEPCI=y
-+# CONFIG_IDEPCI_SHARE_IRQ is not set
-+# CONFIG_BLK_DEV_OFFBOARD is not set
-+CONFIG_BLK_DEV_GENERIC=y
-+# CONFIG_BLK_DEV_OPTI621 is not set
-+CONFIG_BLK_DEV_IDEDMA_PCI=y
-+# CONFIG_BLK_DEV_IDEDMA_FORCED is not set
-+CONFIG_IDEDMA_PCI_AUTO=y
-+# CONFIG_IDEDMA_ONLYDISK is not set
-+# CONFIG_BLK_DEV_AEC62XX is not set
-+# CONFIG_BLK_DEV_ALI15X3 is not set
-+# CONFIG_BLK_DEV_AMD74XX is not set
-+CONFIG_BLK_DEV_CMD64X=y
-+# CONFIG_BLK_DEV_TRIFLEX is not set
-+# CONFIG_BLK_DEV_CY82C693 is not set
-+# CONFIG_BLK_DEV_CS5520 is not set
-+# CONFIG_BLK_DEV_CS5530 is not set
-+# CONFIG_BLK_DEV_HPT34X is not set
-+# CONFIG_BLK_DEV_HPT366 is not set
-+# CONFIG_BLK_DEV_SC1200 is not set
-+CONFIG_BLK_DEV_PIIX=y
-+# CONFIG_BLK_DEV_IT821X is not set
-+# CONFIG_BLK_DEV_NS87415 is not set
-+# CONFIG_BLK_DEV_PDC202XX_OLD is not set
-+# CONFIG_BLK_DEV_PDC202XX_NEW is not set
-+# CONFIG_BLK_DEV_SVWKS is not set
-+# CONFIG_BLK_DEV_SIIMAGE is not set
-+# CONFIG_BLK_DEV_SLC90E66 is not set
-+# CONFIG_BLK_DEV_TRM290 is not set
-+# CONFIG_BLK_DEV_VIA82CXXX is not set
-+# CONFIG_IDE_ARM is not set
-+CONFIG_BLK_DEV_IDEDMA=y
-+# CONFIG_IDEDMA_IVB is not set
-+CONFIG_IDEDMA_AUTO=y
-+# CONFIG_BLK_DEV_HD is not set
++# CONFIG_IDE is not set
+
+#
+# SCSI device support
+# SCSI Transport Attributes
+#
+CONFIG_SCSI_SPI_ATTRS=y
-+CONFIG_SCSI_FC_ATTRS=y
++# CONFIG_SCSI_FC_ATTRS is not set
+# CONFIG_SCSI_ISCSI_ATTRS is not set
+CONFIG_SCSI_SAS_ATTRS=y
+
+#
+# Multi-device support (RAID and LVM)
+#
-+CONFIG_MD=y
-+# CONFIG_BLK_DEV_MD is not set
-+CONFIG_BLK_DEV_DM=y
-+CONFIG_DM_CRYPT=m
-+CONFIG_DM_SNAPSHOT=y
-+CONFIG_DM_MIRROR=m
-+CONFIG_DM_ZERO=m
-+CONFIG_DM_MULTIPATH=m
-+CONFIG_DM_MULTIPATH_EMC=m
++# CONFIG_MD is not set
+
+#
+# Fusion MPT device support
+CONFIG_DUMMY=y
+# CONFIG_BONDING is not set
+# CONFIG_EQUALIZER is not set
-+CONFIG_TUN=y
++# CONFIG_TUN is not set
+
+#
+# ARCnet devices
+#
-+CONFIG_ARCNET=y
-+# CONFIG_ARCNET_1201 is not set
-+# CONFIG_ARCNET_1051 is not set
-+# CONFIG_ARCNET_RAW is not set
-+# CONFIG_ARCNET_CAP is not set
-+# CONFIG_ARCNET_COM90xx is not set
-+# CONFIG_ARCNET_COM90xxIO is not set
-+# CONFIG_ARCNET_RIM_I is not set
-+# CONFIG_ARCNET_COM20020 is not set
++# CONFIG_ARCNET is not set
+
+#
+# PHY device support
+# CONFIG_B44 is not set
+# CONFIG_FORCEDETH is not set
+# CONFIG_DGRS is not set
-+CONFIG_EEPRO100=y
++# CONFIG_EEPRO100 is not set
+CONFIG_E100=y
+# CONFIG_FEALNX is not set
+# CONFIG_NATSEMI is not set
+# CONFIG_IXGB is not set
+# CONFIG_S2IO is not set
+# CONFIG_MYRI10GE is not set
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+# CONFIG_SLIP is not set
+# CONFIG_NET_FC is not set
+# CONFIG_SHAPER is not set
-+CONFIG_NETCONSOLE=y
-+CONFIG_NETPOLL=y
-+# CONFIG_NETPOLL_RX is not set
-+# CONFIG_NETPOLL_TRAP is not set
-+CONFIG_NET_POLL_CONTROLLER=y
++# CONFIG_NETCONSOLE is not set
++# CONFIG_NETPOLL is not set
++# CONFIG_NET_POLL_CONTROLLER is not set
+
+#
+# ISDN subsystem
+#
-+CONFIG_ISDN=m
-+
-+#
-+# Old ISDN4Linux
-+#
-+# CONFIG_ISDN_I4L is not set
-+
-+#
-+# CAPI subsystem
-+#
-+# CONFIG_ISDN_CAPI is not set
++# CONFIG_ISDN is not set
+
+#
+# Telephony Support
+#
+# Input Device Drivers
+#
-+CONFIG_INPUT_KEYBOARD=y
-+CONFIG_KEYBOARD_ATKBD=y
-+# CONFIG_KEYBOARD_SUNKBD is not set
-+# CONFIG_KEYBOARD_LKKBD is not set
-+# CONFIG_KEYBOARD_XTKBD is not set
-+# CONFIG_KEYBOARD_NEWTON is not set
-+CONFIG_INPUT_MOUSE=y
-+CONFIG_MOUSE_PS2=y
-+# CONFIG_MOUSE_SERIAL is not set
-+# CONFIG_MOUSE_VSXXXAA is not set
++# CONFIG_INPUT_KEYBOARD is not set
++# CONFIG_INPUT_MOUSE is not set
+# CONFIG_INPUT_JOYSTICK is not set
+# CONFIG_INPUT_TOUCHSCREEN is not set
+# CONFIG_INPUT_MISC is not set
+# Hardware I/O ports
+#
+CONFIG_SERIO=y
-+CONFIG_SERIO_I8042=y
++# CONFIG_SERIO_I8042 is not set
+# CONFIG_SERIO_SERPORT is not set
+# CONFIG_SERIO_PCIPS2 is not set
-+CONFIG_SERIO_LIBPS2=y
+# CONFIG_SERIO_RAW is not set
-+CONFIG_GAMEPORT=y
-+# CONFIG_GAMEPORT_NS558 is not set
-+# CONFIG_GAMEPORT_L4 is not set
-+# CONFIG_GAMEPORT_EMU10K1 is not set
-+# CONFIG_GAMEPORT_FM801 is not set
++# CONFIG_GAMEPORT is not set
+
+#
+# Character devices
+CONFIG_VT_CONSOLE=y
+CONFIG_HW_CONSOLE=y
+# CONFIG_VT_HW_CONSOLE_BINDING is not set
-+CONFIG_SERIAL_NONSTANDARD=y
-+# CONFIG_COMPUTONE is not set
-+# CONFIG_ROCKETPORT is not set
-+# CONFIG_CYCLADES is not set
-+# CONFIG_DIGIEPCA is not set
-+# CONFIG_MOXA_INTELLIO is not set
-+# CONFIG_MOXA_SMARTIO is not set
-+# CONFIG_ISI is not set
-+# CONFIG_SYNCLINKMP is not set
-+# CONFIG_SYNCLINK_GT is not set
-+# CONFIG_N_HDLC is not set
-+# CONFIG_SPECIALIX is not set
-+# CONFIG_SX is not set
-+# CONFIG_RIO is not set
-+# CONFIG_STALDRV is not set
-+# CONFIG_SGI_SNSC is not set
-+# CONFIG_SGI_TIOCX is not set
++# CONFIG_SERIAL_NONSTANDARD is not set
+
+#
+# Serial drivers
+#
++CONFIG_SERIAL_8250=y
++CONFIG_SERIAL_8250_CONSOLE=y
++CONFIG_SERIAL_8250_PCI=y
++CONFIG_SERIAL_8250_NR_UARTS=8
++CONFIG_SERIAL_8250_RUNTIME_UARTS=4
++CONFIG_SERIAL_8250_EXTENDED=y
++CONFIG_SERIAL_8250_SHARE_IRQ=y
++# CONFIG_SERIAL_8250_DETECT_IRQ is not set
++# CONFIG_SERIAL_8250_RSA is not set
+
+#
+# Non-8250 serial port support
+#
-+# CONFIG_SERIAL_SGI_L1_CONSOLE is not set
++CONFIG_SERIAL_CORE=y
++CONFIG_SERIAL_CORE_CONSOLE=y
+# CONFIG_SERIAL_JSM is not set
+CONFIG_UNIX98_PTYS=y
+CONFIG_LEGACY_PTYS=y
+CONFIG_AGP=y
+# CONFIG_AGP_SIS is not set
+# CONFIG_AGP_VIA is not set
-+CONFIG_AGP_I460=y
-+# CONFIG_AGP_HP_ZX1 is not set
-+# CONFIG_AGP_SGI_TIOCA is not set
+CONFIG_DRM=y
+# CONFIG_DRM_TDFX is not set
+# CONFIG_DRM_R128 is not set
-+# CONFIG_DRM_RADEON is not set
++CONFIG_DRM_RADEON=y
+# CONFIG_DRM_MGA is not set
+# CONFIG_DRM_SIS is not set
+# CONFIG_DRM_VIA is not set
+# CONFIG_RAW_DRIVER is not set
+# CONFIG_HPET is not set
+# CONFIG_HANGCHECK_TIMER is not set
-+# CONFIG_MMTIMER is not set
+
+#
+# TPM devices
+CONFIG_SND_HWDEP=y
+CONFIG_SND_RAWMIDI=y
+CONFIG_SND_SEQUENCER=y
-+CONFIG_SND_SEQ_DUMMY=y
++# CONFIG_SND_SEQ_DUMMY is not set
+CONFIG_SND_OSSEMUL=y
+CONFIG_SND_MIXER_OSS=y
+CONFIG_SND_PCM_OSS=y
+CONFIG_SND_OPL3_LIB=y
+CONFIG_SND_AC97_CODEC=y
+CONFIG_SND_AC97_BUS=y
-+CONFIG_SND_DUMMY=y
-+CONFIG_SND_VIRMIDI=y
++# CONFIG_SND_DUMMY is not set
++# CONFIG_SND_VIRMIDI is not set
+# CONFIG_SND_MTPAV is not set
-+CONFIG_SND_SERIAL_U16550=y
-+CONFIG_SND_MPU401=y
++# CONFIG_SND_SERIAL_U16550 is not set
++# CONFIG_SND_MPU401 is not set
+
+#
+# PCI devices
+# CONFIG_SND_AD1889 is not set
+# CONFIG_SND_ALS300 is not set
+# CONFIG_SND_ALI5451 is not set
-+CONFIG_SND_ATIIXP=y
++# CONFIG_SND_ATIIXP is not set
+# CONFIG_SND_ATIIXP_MODEM is not set
+# CONFIG_SND_AU8810 is not set
+# CONFIG_SND_AU8820 is not set
+# CONFIG_SND_CMIPCI is not set
+# CONFIG_SND_CS4281 is not set
+# CONFIG_SND_CS46XX is not set
-+# CONFIG_SND_DARLA20 is not set
-+# CONFIG_SND_GINA20 is not set
-+# CONFIG_SND_LAYLA20 is not set
-+# CONFIG_SND_DARLA24 is not set
-+# CONFIG_SND_GINA24 is not set
-+# CONFIG_SND_LAYLA24 is not set
-+# CONFIG_SND_MONA is not set
-+# CONFIG_SND_MIA is not set
-+# CONFIG_SND_ECHO3G is not set
-+# CONFIG_SND_INDIGO is not set
-+# CONFIG_SND_INDIGOIO is not set
-+# CONFIG_SND_INDIGODJ is not set
+# CONFIG_SND_EMU10K1 is not set
+# CONFIG_SND_EMU10K1X is not set
+# CONFIG_SND_ENS1370 is not set
+# CONFIG_SND_MIXART is not set
+# CONFIG_SND_NM256 is not set
+# CONFIG_SND_PCXHR is not set
-+# CONFIG_SND_RIPTIDE is not set
+# CONFIG_SND_RME32 is not set
+# CONFIG_SND_RME96 is not set
+# CONFIG_SND_RME9652 is not set
+#
+# Open Sound System
+#
-+CONFIG_SOUND_PRIME=y
-+# CONFIG_OSS_OBSOLETE_DRIVER is not set
-+# CONFIG_SOUND_BT878 is not set
-+# CONFIG_SOUND_ES1371 is not set
-+# CONFIG_SOUND_ICH is not set
-+# CONFIG_SOUND_TRIDENT is not set
-+# CONFIG_SOUND_MSNDCLAS is not set
-+# CONFIG_SOUND_MSNDPIN is not set
-+# CONFIG_SOUND_VIA82CXXX is not set
-+# CONFIG_SOUND_TVMIXER is not set
++# CONFIG_SOUND_PRIME is not set
+
+#
+# USB support
+#
+# Miscellaneous USB options
+#
-+CONFIG_USB_DEVICEFS=y
++# CONFIG_USB_DEVICEFS is not set
+CONFIG_USB_BANDWIDTH=y
+# CONFIG_USB_DYNAMIC_MINORS is not set
+# CONFIG_USB_SUSPEND is not set
+# CONFIG_USB_STORAGE_DEBUG is not set
+# CONFIG_USB_STORAGE_DATAFAB is not set
+# CONFIG_USB_STORAGE_FREECOM is not set
-+# CONFIG_USB_STORAGE_ISD200 is not set
+# CONFIG_USB_STORAGE_DPCM is not set
+# CONFIG_USB_STORAGE_USBAT is not set
+# CONFIG_USB_STORAGE_SDDR09 is not set
+# CONFIG_USB_APPLEDISPLAY is not set
+# CONFIG_USB_SISUSBVGA is not set
+# CONFIG_USB_LD is not set
-+# CONFIG_USB_TEST is not set
+
+#
+# USB DSL modem support
+#
+CONFIG_EXT2_FS=y
+CONFIG_EXT2_FS_XATTR=y
-+CONFIG_EXT2_FS_POSIX_ACL=y
-+CONFIG_EXT2_FS_SECURITY=y
++# CONFIG_EXT2_FS_POSIX_ACL is not set
++# CONFIG_EXT2_FS_SECURITY is not set
+# CONFIG_EXT2_FS_XIP is not set
+CONFIG_EXT3_FS=y
+CONFIG_EXT3_FS_XATTR=y
-+CONFIG_EXT3_FS_POSIX_ACL=y
-+CONFIG_EXT3_FS_SECURITY=y
++# CONFIG_EXT3_FS_POSIX_ACL is not set
++# CONFIG_EXT3_FS_SECURITY is not set
+CONFIG_JBD=y
+# CONFIG_JBD_DEBUG is not set
+CONFIG_FS_MBCACHE=y
-+CONFIG_REISERFS_FS=y
-+# CONFIG_REISERFS_CHECK is not set
-+# CONFIG_REISERFS_PROC_INFO is not set
-+CONFIG_REISERFS_FS_XATTR=y
-+CONFIG_REISERFS_FS_POSIX_ACL=y
-+CONFIG_REISERFS_FS_SECURITY=y
++# CONFIG_REISERFS_FS is not set
+# CONFIG_JFS_FS is not set
-+CONFIG_FS_POSIX_ACL=y
-+CONFIG_XFS_FS=y
-+# CONFIG_XFS_QUOTA is not set
-+# CONFIG_XFS_SECURITY is not set
-+# CONFIG_XFS_POSIX_ACL is not set
-+# CONFIG_XFS_RT is not set
++# CONFIG_FS_POSIX_ACL is not set
++# CONFIG_XFS_FS is not set
+# CONFIG_OCFS2_FS is not set
+# CONFIG_MINIX_FS is not set
+# CONFIG_ROMFS_FS is not set
+# CONFIG_QUOTA is not set
+CONFIG_DNOTIFY=y
+CONFIG_AUTOFS_FS=y
-+CONFIG_AUTOFS4_FS=y
++# CONFIG_AUTOFS4_FS is not set
+# CONFIG_FUSE_FS is not set
+
+#
+#
+CONFIG_PROC_FS=y
+CONFIG_PROC_KCORE=y
-+CONFIG_PROC_IOMEM_MACHINE=y
+CONFIG_SYSFS=y
+CONFIG_TMPFS=y
+# CONFIG_HUGETLB_PAGE is not set
+CONFIG_NFS_V3=y
+# CONFIG_NFS_V3_ACL is not set
+CONFIG_NFS_V4=y
-+CONFIG_NFS_DIRECTIO=y
++# CONFIG_NFS_DIRECTIO is not set
+CONFIG_NFSD=y
+CONFIG_NFSD_V3=y
+# CONFIG_NFSD_V3_ACL is not set
-+CONFIG_NFSD_V4=y
-+CONFIG_NFSD_TCP=y
++# CONFIG_NFSD_V4 is not set
++# CONFIG_NFSD_TCP is not set
+CONFIG_ROOT_NFS=y
+CONFIG_LOCKD=y
+CONFIG_LOCKD_V4=y
+CONFIG_SUNRPC_GSS=y
+CONFIG_RPCSEC_GSS_KRB5=y
+# CONFIG_RPCSEC_GSS_SPKM3 is not set
-+CONFIG_SMB_FS=y
-+CONFIG_SMB_NLS_DEFAULT=y
-+CONFIG_SMB_NLS_REMOTE="cp437"
-+CONFIG_CIFS=y
-+# CONFIG_CIFS_STATS is not set
-+# CONFIG_CIFS_WEAK_PW_HASH is not set
-+# CONFIG_CIFS_XATTR is not set
-+# CONFIG_CIFS_DEBUG2 is not set
-+# CONFIG_CIFS_EXPERIMENTAL is not set
++# CONFIG_SMB_FS is not set
++# CONFIG_CIFS is not set
+# CONFIG_NCP_FS is not set
+# CONFIG_CODA_FS is not set
+# CONFIG_AFS_FS is not set
+# CONFIG_SOLARIS_X86_PARTITION is not set
+# CONFIG_UNIXWARE_DISKLABEL is not set
+# CONFIG_LDM_PARTITION is not set
-+CONFIG_SGI_PARTITION=y
++# CONFIG_SGI_PARTITION is not set
+# CONFIG_ULTRIX_PARTITION is not set
+# CONFIG_SUN_PARTITION is not set
+# CONFIG_KARMA_PARTITION is not set
+CONFIG_IRQ_PER_CPU=y
+
+#
-+# HP Simulator drivers
-+#
-+# CONFIG_HP_SIMETH is not set
-+# CONFIG_HP_SIMSERIAL is not set
-+# CONFIG_HP_SIMSCSI is not set
-+
-+#
+# Instrumentation Support
+#
+# CONFIG_PROFILING is not set
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_UNUSED_SYMBOLS=y
+CONFIG_DEBUG_KERNEL=y
-+CONFIG_LOG_BUF_SHIFT=20
++CONFIG_LOG_BUF_SHIFT=17
+CONFIG_DETECT_SOFTLOCKUP=y
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_DEBUG_SLAB is not set
+#
+# XEN
+#
-+CONFIG_XEN_PRIVILEGED_GUEST=y
-+# CONFIG_XEN_UNPRIVILEGED_GUEST is not set
++# CONFIG_XEN_PRIVILEGED_GUEST is not set
++CONFIG_XEN_UNPRIVILEGED_GUEST=y
+CONFIG_XEN_PRIVCMD=y
+CONFIG_XEN_XENBUS_DEV=y
-+CONFIG_XEN_BACKEND=y
-+CONFIG_XEN_BLKDEV_BACKEND=y
-+CONFIG_XEN_BLKDEV_TAP=y
-+CONFIG_XEN_NETDEV_BACKEND=y
-+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
-+# CONFIG_XEN_NETDEV_LOOPBACK is not set
-+CONFIG_XEN_PCIDEV_BACKEND=y
-+# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
-+# CONFIG_XEN_PCIDEV_BACKEND_PASS is not set
-+# CONFIG_XEN_PCIDEV_BACKEND_SLOT is not set
-+CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER=y
-+# CONFIG_XEN_PCIDEV_BE_DEBUG is not set
-+CONFIG_XEN_TPMDEV_BACKEND=m
++# CONFIG_XEN_BACKEND is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
-+CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_GRANT_DEV is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
+CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_KEYBOARD=y
+# CONFIG_XEN_SCRUB_PAGES is not set
-+CONFIG_XEN_DISABLE_SERIAL=y
++# CONFIG_XEN_DISABLE_SERIAL is not set
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_COMPAT_030002_AND_LATER=y
+# CONFIG_XEN_COMPAT_030004_AND_LATER is not set
+CONFIG_NO_IDLE_HZ=y
+CONFIG_XEN_BALLOON=y
+CONFIG_XEN_XENCOMM=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xenU_ia64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_ia64
---- linux-2.6.18/buildconfigs/linux-defconfig_xenU_ia64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_ia64 2007-12-23 11:15:00.662848831 +0100
-@@ -0,0 +1,1507 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xenU_x86_32 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_32 2008-05-19 00:33:19.713282430 +0300
+@@ -0,0 +1,952 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Mon Dec 17 13:38:36 2007
++# Tue Oct 16 09:31:29 2007
+#
++CONFIG_X86_32=y
++CONFIG_LOCKDEP_SUPPORT=y
++CONFIG_STACKTRACE_SUPPORT=y
++CONFIG_SEMAPHORE_SLEEPERS=y
++CONFIG_X86=y
++CONFIG_MMU=y
++CONFIG_GENERIC_ISA_DMA=y
++CONFIG_GENERIC_IOMAP=y
++CONFIG_GENERIC_HWEIGHT=y
++CONFIG_ARCH_MAY_HAVE_PC_FDC=y
++CONFIG_DMI=y
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+#
+# General setup
+#
+CONFIG_LOCALVERSION=""
-+CONFIG_LOCALVERSION_AUTO=y
++# CONFIG_LOCALVERSION_AUTO is not set
+CONFIG_SWAP=y
+CONFIG_SYSVIPC=y
+# CONFIG_POSIX_MQUEUE is not set
-+CONFIG_BSD_PROCESS_ACCT=y
-+# CONFIG_BSD_PROCESS_ACCT_V3 is not set
++# CONFIG_BSD_PROCESS_ACCT is not set
+# CONFIG_TASKSTATS is not set
+# CONFIG_AUDIT is not set
+# CONFIG_IKCONFIG is not set
+# CONFIG_CPUSETS is not set
+# CONFIG_RELAY is not set
+CONFIG_INITRAMFS_SOURCE=""
-+CONFIG_CC_OPTIMIZE_FOR_SIZE=y
++# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
+# CONFIG_EMBEDDED is not set
++CONFIG_UID16=y
+CONFIG_SYSCTL=y
+CONFIG_KALLSYMS=y
+# CONFIG_KALLSYMS_ALL is not set
+# Loadable module support
+#
+CONFIG_MODULES=y
-+# CONFIG_MODULE_UNLOAD is not set
++CONFIG_MODULE_UNLOAD=y
++# CONFIG_MODULE_FORCE_UNLOAD is not set
+# CONFIG_MODVERSIONS is not set
+# CONFIG_MODULE_SRCVERSION_ALL is not set
-+# CONFIG_KMOD is not set
++CONFIG_KMOD=y
+CONFIG_STOP_MACHINE=y
+
+#
+# Block layer
+#
++# CONFIG_LBD is not set
+# CONFIG_BLK_DEV_IO_TRACE is not set
++CONFIG_LSF=y
+
+#
+# IO Schedulers
+#
+# Processor type and features
+#
-+CONFIG_IA64=y
-+CONFIG_64BIT=y
-+CONFIG_MMU=y
-+CONFIG_SWIOTLB=y
++CONFIG_SMP=y
++# CONFIG_X86_PC is not set
++CONFIG_X86_XEN=y
++# CONFIG_X86_ELAN is not set
++# CONFIG_X86_VOYAGER is not set
++# CONFIG_X86_NUMAQ is not set
++# CONFIG_X86_SUMMIT is not set
++# CONFIG_X86_BIGSMP is not set
++# CONFIG_X86_VISWS is not set
++# CONFIG_X86_GENERICARCH is not set
++# CONFIG_X86_ES7000 is not set
++# CONFIG_M386 is not set
++# CONFIG_M486 is not set
++# CONFIG_M586 is not set
++# CONFIG_M586TSC is not set
++# CONFIG_M586MMX is not set
++CONFIG_M686=y
++# CONFIG_MPENTIUMII is not set
++# CONFIG_MPENTIUMIII is not set
++# CONFIG_MPENTIUMM is not set
++# CONFIG_MPENTIUM4 is not set
++# CONFIG_MK6 is not set
++# CONFIG_MK7 is not set
++# CONFIG_MK8 is not set
++# CONFIG_MCRUSOE is not set
++# CONFIG_MEFFICEON is not set
++# CONFIG_MWINCHIPC6 is not set
++# CONFIG_MWINCHIP2 is not set
++# CONFIG_MWINCHIP3D is not set
++# CONFIG_MGEODEGX1 is not set
++# CONFIG_MGEODE_LX is not set
++# CONFIG_MCYRIXIII is not set
++# CONFIG_MVIAC3_2 is not set
++# CONFIG_X86_GENERIC is not set
++CONFIG_X86_CMPXCHG=y
++CONFIG_X86_XADD=y
++CONFIG_X86_L1_CACHE_SHIFT=5
+CONFIG_RWSEM_XCHGADD_ALGORITHM=y
-+CONFIG_GENERIC_FIND_NEXT_BIT=y
+CONFIG_GENERIC_CALIBRATE_DELAY=y
-+CONFIG_TIME_INTERPOLATION=y
-+CONFIG_DMI=y
-+CONFIG_EFI=y
-+CONFIG_GENERIC_IOMAP=y
-+CONFIG_XEN=y
-+CONFIG_XEN_IA64_EXPOSE_P2M=y
-+CONFIG_XEN_IA64_EXPOSE_P2M_USE_DTR=y
-+CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER=y
-+CONFIG_DMA_IS_DMA32=y
-+CONFIG_AUDIT_ARCH=y
-+# CONFIG_IA64_GENERIC is not set
-+# CONFIG_IA64_DIG is not set
-+# CONFIG_IA64_HP_ZX1 is not set
-+# CONFIG_IA64_HP_ZX1_SWIOTLB is not set
-+# CONFIG_IA64_SGI_SN2 is not set
-+# CONFIG_IA64_HP_SIM is not set
-+CONFIG_IA64_XEN=y
-+# CONFIG_ITANIUM is not set
-+CONFIG_MCKINLEY=y
-+# CONFIG_IA64_PAGE_SIZE_4KB is not set
-+# CONFIG_IA64_PAGE_SIZE_8KB is not set
-+CONFIG_IA64_PAGE_SIZE_16KB=y
-+# CONFIG_IA64_PAGE_SIZE_64KB is not set
-+CONFIG_PGTABLE_3=y
-+# CONFIG_PGTABLE_4 is not set
-+CONFIG_HZ_100=y
-+# CONFIG_HZ_250 is not set
-+# CONFIG_HZ_1000 is not set
-+CONFIG_HZ=100
-+CONFIG_IA64_L1_CACHE_SHIFT=7
-+# CONFIG_IA64_CYCLONE is not set
-+CONFIG_IOSAPIC=y
-+CONFIG_FORCE_MAX_ZONEORDER=11
-+CONFIG_SMP=y
-+CONFIG_NR_CPUS=16
-+CONFIG_HOTPLUG_CPU=y
-+CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
-+# CONFIG_SCHED_SMT is not set
-+# CONFIG_PERMIT_BSP_REMOVE is not set
++CONFIG_X86_PPRO_FENCE=y
++CONFIG_X86_WP_WORKS_OK=y
++CONFIG_X86_INVLPG=y
++CONFIG_X86_BSWAP=y
++CONFIG_X86_POPAD_OK=y
++CONFIG_X86_CMPXCHG64=y
++CONFIG_X86_GOOD_APIC=y
++CONFIG_X86_USE_PPRO_CHECKSUM=y
++CONFIG_NR_CPUS=8
++CONFIG_PREEMPT_NONE=y
++# CONFIG_PREEMPT_VOLUNTARY is not set
+# CONFIG_PREEMPT is not set
++CONFIG_PREEMPT_BKL=y
++CONFIG_VM86=y
++# CONFIG_TOSHIBA is not set
++# CONFIG_I8K is not set
++# CONFIG_X86_REBOOTFIXUPS is not set
++# CONFIG_X86_MSR is not set
++CONFIG_X86_CPUID=y
++CONFIG_SWIOTLB=y
++
++#
++# Firmware Drivers
++#
++# CONFIG_EDD is not set
++# CONFIG_DELL_RBU is not set
++# CONFIG_DCDBAS is not set
++# CONFIG_NOHIGHMEM is not set
++CONFIG_HIGHMEM4G=y
++# CONFIG_HIGHMEM64G is not set
++CONFIG_PAGE_OFFSET=0xC0000000
++CONFIG_HIGHMEM=y
+CONFIG_SELECT_MEMORY_MODEL=y
+CONFIG_FLATMEM_MANUAL=y
+# CONFIG_DISCONTIGMEM_MANUAL is not set
+# CONFIG_SPARSEMEM_STATIC is not set
+CONFIG_SPLIT_PTLOCK_CPUS=4
+CONFIG_RESOURCES_64BIT=y
-+CONFIG_ARCH_SELECT_MEMORY_MODEL=y
-+CONFIG_ARCH_DISCONTIGMEM_ENABLE=y
-+CONFIG_ARCH_FLATMEM_ENABLE=y
-+CONFIG_ARCH_SPARSEMEM_ENABLE=y
-+CONFIG_VIRTUAL_MEM_MAP=y
-+CONFIG_HOLES_IN_ZONE=y
-+# CONFIG_IA32_SUPPORT is not set
-+CONFIG_IA64_MCA_RECOVERY=y
-+CONFIG_PERFMON=y
-+CONFIG_IA64_PALINFO=y
-+# CONFIG_KEXEC is not set
++# CONFIG_HIGHPTE is not set
++# CONFIG_REGPARM is not set
++CONFIG_SECCOMP=y
++CONFIG_HZ_100=y
++# CONFIG_HZ_250 is not set
++# CONFIG_HZ_1000 is not set
++CONFIG_HZ=100
+# CONFIG_CRASH_DUMP is not set
++CONFIG_PHYSICAL_START=0x100000
++CONFIG_HOTPLUG_CPU=y
++CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
+
+#
-+# Firmware Drivers
-+#
-+CONFIG_EFI_VARS=y
-+CONFIG_EFI_PCDP=y
-+CONFIG_BINFMT_ELF=y
-+CONFIG_BINFMT_MISC=y
-+
-+#
-+# Power management and ACPI
-+#
-+CONFIG_PM=y
-+CONFIG_PM_LEGACY=y
-+# CONFIG_PM_DEBUG is not set
-+
-+#
-+# ACPI (Advanced Configuration and Power Interface) Support
-+#
-+CONFIG_ACPI=y
-+CONFIG_ACPI_BUTTON=y
-+CONFIG_ACPI_FAN=y
-+# CONFIG_ACPI_DOCK is not set
-+CONFIG_ACPI_PROCESSOR=y
-+CONFIG_ACPI_HOTPLUG_CPU=y
-+CONFIG_ACPI_THERMAL=y
-+CONFIG_ACPI_BLACKLIST_YEAR=0
-+# CONFIG_ACPI_DEBUG is not set
-+CONFIG_ACPI_EC=y
-+CONFIG_ACPI_POWER=y
-+CONFIG_ACPI_SYSTEM=y
-+CONFIG_ACPI_CONTAINER=y
-+
-+#
-+# CPU Frequency scaling
++# Bus options (PCI, PCMCIA, EISA, MCA, ISA)
+#
-+# CONFIG_CPU_FREQ is not set
++# CONFIG_PCI is not set
++CONFIG_ISA_DMA_API=y
++# CONFIG_SCx200 is not set
+
+#
-+# Bus options (PCI, PCMCIA)
++# PCCARD (PCMCIA/CardBus) support
+#
-+CONFIG_PCI=y
-+CONFIG_PCI_DOMAINS=y
-+CONFIG_XEN_PCIDEV_FRONTEND=y
-+# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
-+# CONFIG_PCIEPORTBUS is not set
-+# CONFIG_PCI_DEBUG is not set
++# CONFIG_PCCARD is not set
+
+#
+# PCI Hotplug Support
+#
-+CONFIG_HOTPLUG_PCI=y
-+# CONFIG_HOTPLUG_PCI_FAKE is not set
-+CONFIG_HOTPLUG_PCI_ACPI=y
-+# CONFIG_HOTPLUG_PCI_ACPI_IBM is not set
-+# CONFIG_HOTPLUG_PCI_CPCI is not set
-+# CONFIG_HOTPLUG_PCI_SHPC is not set
+
+#
-+# PCCARD (PCMCIA/CardBus) support
++# Executable file formats
+#
-+# CONFIG_PCCARD is not set
++CONFIG_BINFMT_ELF=y
++# CONFIG_BINFMT_AOUT is not set
++# CONFIG_BINFMT_MISC is not set
+
+#
+# Networking
+# CONFIG_XFRM_USER is not set
+# CONFIG_NET_KEY is not set
+CONFIG_INET=y
-+CONFIG_IP_MULTICAST=y
++# CONFIG_IP_MULTICAST is not set
+# CONFIG_IP_ADVANCED_ROUTER is not set
+CONFIG_IP_FIB_HASH=y
+CONFIG_IP_PNP=y
-+CONFIG_IP_PNP_DHCP=y
++# CONFIG_IP_PNP_DHCP is not set
+# CONFIG_IP_PNP_BOOTP is not set
+# CONFIG_IP_PNP_RARP is not set
+# CONFIG_NET_IPIP is not set
+# CONFIG_NET_IPGRE is not set
-+# CONFIG_IP_MROUTE is not set
+# CONFIG_ARPD is not set
+# CONFIG_SYN_COOKIES is not set
+# CONFIG_INET_AH is not set
+# CONFIG_INET_TUNNEL is not set
+CONFIG_INET_XFRM_MODE_TRANSPORT=y
+CONFIG_INET_XFRM_MODE_TUNNEL=y
-+CONFIG_INET_DIAG=y
-+CONFIG_INET_TCP_DIAG=y
++# CONFIG_INET_DIAG is not set
+# CONFIG_TCP_CONG_ADVANCED is not set
+CONFIG_TCP_CONG_BIC=y
-+
-+#
-+# IP: Virtual Server Configuration
-+#
-+# CONFIG_IP_VS is not set
+# CONFIG_IPV6 is not set
+# CONFIG_INET6_XFRM_TUNNEL is not set
+# CONFIG_INET6_TUNNEL is not set
+# CONFIG_NETWORK_SECMARK is not set
-+CONFIG_NETFILTER=y
-+# CONFIG_NETFILTER_DEBUG is not set
-+
-+#
-+# Core Netfilter Configuration
-+#
-+# CONFIG_NETFILTER_NETLINK is not set
-+# CONFIG_NF_CONNTRACK is not set
-+# CONFIG_NETFILTER_XTABLES is not set
-+
-+#
-+# IP: Netfilter Configuration
-+#
-+# CONFIG_IP_NF_CONNTRACK is not set
-+# CONFIG_IP_NF_QUEUE is not set
++# CONFIG_NETFILTER is not set
+
+#
+# DCCP Configuration (EXPERIMENTAL)
+#
+# Plug and Play support
+#
-+# CONFIG_PNP is not set
+
+#
+# Block devices
+#
-+# CONFIG_BLK_CPQ_DA is not set
-+# CONFIG_BLK_CPQ_CISS_DA is not set
-+# CONFIG_BLK_DEV_DAC960 is not set
-+# CONFIG_BLK_DEV_UMEM is not set
++# CONFIG_BLK_DEV_FD is not set
+# CONFIG_BLK_DEV_COW_COMMON is not set
-+CONFIG_BLK_DEV_LOOP=y
++CONFIG_BLK_DEV_LOOP=m
+# CONFIG_BLK_DEV_CRYPTOLOOP is not set
-+# CONFIG_BLK_DEV_NBD is not set
-+# CONFIG_BLK_DEV_SX8 is not set
-+# CONFIG_BLK_DEV_UB is not set
++CONFIG_BLK_DEV_NBD=m
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_RAM_COUNT=16
+CONFIG_BLK_DEV_RAM_SIZE=4096
+# SCSI device support
+#
+# CONFIG_RAID_ATTRS is not set
-+CONFIG_SCSI=y
++CONFIG_SCSI=m
+CONFIG_SCSI_PROC_FS=y
+
+#
+# SCSI support type (disk, tape, CD-ROM)
+#
-+CONFIG_BLK_DEV_SD=y
-+CONFIG_CHR_DEV_ST=y
-+CONFIG_CHR_DEV_OSST=y
-+CONFIG_BLK_DEV_SR=y
-+CONFIG_BLK_DEV_SR_VENDOR=y
-+CONFIG_CHR_DEV_SG=y
++CONFIG_BLK_DEV_SD=m
++# CONFIG_CHR_DEV_ST is not set
++# CONFIG_CHR_DEV_OSST is not set
++# CONFIG_BLK_DEV_SR is not set
++# CONFIG_CHR_DEV_SG is not set
+# CONFIG_CHR_DEV_SCH is not set
+
+#
+# Some SCSI devices (e.g. CD jukebox) support multiple LUNs
+#
-+CONFIG_SCSI_MULTI_LUN=y
-+CONFIG_SCSI_CONSTANTS=y
-+CONFIG_SCSI_LOGGING=y
++# CONFIG_SCSI_MULTI_LUN is not set
++# CONFIG_SCSI_CONSTANTS is not set
++# CONFIG_SCSI_LOGGING is not set
+
+#
+# SCSI Transport Attributes
+#
-+CONFIG_SCSI_SPI_ATTRS=y
++# CONFIG_SCSI_SPI_ATTRS is not set
+# CONFIG_SCSI_FC_ATTRS is not set
+# CONFIG_SCSI_ISCSI_ATTRS is not set
-+CONFIG_SCSI_SAS_ATTRS=y
++# CONFIG_SCSI_SAS_ATTRS is not set
+
+#
+# SCSI low-level drivers
+#
+# CONFIG_ISCSI_TCP is not set
-+# CONFIG_BLK_DEV_3W_XXXX_RAID is not set
-+# CONFIG_SCSI_3W_9XXX is not set
-+# CONFIG_SCSI_ACARD is not set
-+# CONFIG_SCSI_AACRAID is not set
-+# CONFIG_SCSI_AIC7XXX is not set
-+# CONFIG_SCSI_AIC7XXX_OLD is not set
-+# CONFIG_SCSI_AIC79XX is not set
-+# CONFIG_MEGARAID_NEWGEN is not set
-+# CONFIG_MEGARAID_LEGACY is not set
-+# CONFIG_MEGARAID_SAS is not set
+# CONFIG_SCSI_SATA is not set
-+# CONFIG_SCSI_HPTIOP is not set
-+# CONFIG_SCSI_DMX3191D is not set
-+# CONFIG_SCSI_FUTURE_DOMAIN is not set
-+# CONFIG_SCSI_IPS is not set
-+# CONFIG_SCSI_INITIO is not set
-+# CONFIG_SCSI_INIA100 is not set
-+CONFIG_SCSI_SYM53C8XX_2=y
-+CONFIG_SCSI_SYM53C8XX_DMA_ADDRESSING_MODE=1
-+CONFIG_SCSI_SYM53C8XX_DEFAULT_TAGS=16
-+CONFIG_SCSI_SYM53C8XX_MAX_TAGS=64
-+CONFIG_SCSI_SYM53C8XX_MMIO=y
-+# CONFIG_SCSI_IPR is not set
-+CONFIG_SCSI_QLOGIC_1280=y
-+# CONFIG_SCSI_QLA_FC is not set
-+# CONFIG_SCSI_LPFC is not set
-+# CONFIG_SCSI_DC395x is not set
-+# CONFIG_SCSI_DC390T is not set
+# CONFIG_SCSI_DEBUG is not set
+
+#
+#
+# Fusion MPT device support
+#
-+CONFIG_FUSION=y
-+CONFIG_FUSION_SPI=y
-+# CONFIG_FUSION_FC is not set
-+CONFIG_FUSION_SAS=y
-+CONFIG_FUSION_MAX_SGE=128
-+# CONFIG_FUSION_CTL is not set
++# CONFIG_FUSION is not set
+
+#
+# IEEE 1394 (FireWire) support
+#
-+# CONFIG_IEEE1394 is not set
+
+#
+# I2O device support
+#
-+# CONFIG_I2O is not set
+
+#
+# Network device support
+#
+CONFIG_NETDEVICES=y
-+CONFIG_DUMMY=y
++# CONFIG_DUMMY is not set
+# CONFIG_BONDING is not set
+# CONFIG_EQUALIZER is not set
+# CONFIG_TUN is not set
+
+#
-+# ARCnet devices
-+#
-+# CONFIG_ARCNET is not set
-+
-+#
+# PHY device support
+#
-+# CONFIG_PHYLIB is not set
+
+#
+# Ethernet (10 or 100Mbit)
+#
-+CONFIG_NET_ETHERNET=y
-+CONFIG_MII=y
-+# CONFIG_HAPPYMEAL is not set
-+# CONFIG_SUNGEM is not set
-+# CONFIG_CASSINI is not set
-+# CONFIG_NET_VENDOR_3COM is not set
-+
-+#
-+# Tulip family network device support
-+#
-+CONFIG_NET_TULIP=y
-+# CONFIG_DE2104X is not set
-+CONFIG_TULIP=y
-+CONFIG_TULIP_MWI=y
-+CONFIG_TULIP_MMIO=y
-+CONFIG_TULIP_NAPI=y
-+CONFIG_TULIP_NAPI_HW_MITIGATION=y
-+# CONFIG_DE4X5 is not set
-+# CONFIG_WINBOND_840 is not set
-+# CONFIG_DM9102 is not set
-+# CONFIG_ULI526X is not set
-+# CONFIG_HP100 is not set
-+CONFIG_NET_PCI=y
-+# CONFIG_PCNET32 is not set
-+# CONFIG_AMD8111_ETH is not set
-+# CONFIG_ADAPTEC_STARFIRE is not set
-+# CONFIG_B44 is not set
-+# CONFIG_FORCEDETH is not set
-+# CONFIG_DGRS is not set
-+# CONFIG_EEPRO100 is not set
-+CONFIG_E100=y
-+# CONFIG_FEALNX is not set
-+# CONFIG_NATSEMI is not set
-+# CONFIG_NE2K_PCI is not set
-+# CONFIG_8139CP is not set
-+# CONFIG_8139TOO is not set
-+# CONFIG_SIS900 is not set
-+# CONFIG_EPIC100 is not set
-+# CONFIG_SUNDANCE is not set
-+# CONFIG_VIA_RHINE is not set
++# CONFIG_NET_ETHERNET is not set
+
+#
+# Ethernet (1000 Mbit)
+#
-+# CONFIG_ACENIC is not set
-+# CONFIG_DL2K is not set
-+CONFIG_E1000=y
-+# CONFIG_E1000_NAPI is not set
-+# CONFIG_E1000_DISABLE_PACKET_SPLIT is not set
-+# CONFIG_NS83820 is not set
-+# CONFIG_HAMACHI is not set
-+# CONFIG_YELLOWFIN is not set
-+# CONFIG_R8169 is not set
-+# CONFIG_SIS190 is not set
-+# CONFIG_SKGE is not set
-+# CONFIG_SKY2 is not set
-+# CONFIG_SK98LIN is not set
-+# CONFIG_VIA_VELOCITY is not set
-+CONFIG_TIGON3=y
-+# CONFIG_BNX2 is not set
+
+#
+# Ethernet (10000 Mbit)
+#
-+# CONFIG_CHELSIO_T1 is not set
-+# CONFIG_IXGB is not set
-+# CONFIG_S2IO is not set
-+# CONFIG_MYRI10GE is not set
+
+#
+# Token Ring devices
+#
-+# CONFIG_TR is not set
+
+#
+# Wireless LAN (non-hamradio)
+# Wan interfaces
+#
+# CONFIG_WAN is not set
-+# CONFIG_FDDI is not set
-+# CONFIG_HIPPI is not set
+# CONFIG_PPP is not set
+# CONFIG_SLIP is not set
-+# CONFIG_NET_FC is not set
+# CONFIG_SHAPER is not set
+# CONFIG_NETCONSOLE is not set
+# CONFIG_NETPOLL is not set
+# Userland interfaces
+#
+CONFIG_INPUT_MOUSEDEV=y
-+CONFIG_INPUT_MOUSEDEV_PSAUX=y
++# CONFIG_INPUT_MOUSEDEV_PSAUX is not set
+CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024
+CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768
-+CONFIG_INPUT_JOYDEV=y
++# CONFIG_INPUT_JOYDEV is not set
+# CONFIG_INPUT_TSDEV is not set
-+CONFIG_INPUT_EVDEV=y
++# CONFIG_INPUT_EVDEV is not set
+# CONFIG_INPUT_EVBUG is not set
+
+#
+# Input Device Drivers
+#
-+# CONFIG_INPUT_KEYBOARD is not set
++CONFIG_INPUT_KEYBOARD=y
++# CONFIG_KEYBOARD_ATKBD is not set
++# CONFIG_KEYBOARD_SUNKBD is not set
++# CONFIG_KEYBOARD_LKKBD is not set
++# CONFIG_KEYBOARD_XTKBD is not set
++# CONFIG_KEYBOARD_NEWTON is not set
+# CONFIG_INPUT_MOUSE is not set
+# CONFIG_INPUT_JOYSTICK is not set
+# CONFIG_INPUT_TOUCHSCREEN is not set
+# Hardware I/O ports
+#
+CONFIG_SERIO=y
-+# CONFIG_SERIO_I8042 is not set
-+# CONFIG_SERIO_SERPORT is not set
-+# CONFIG_SERIO_PCIPS2 is not set
++CONFIG_SERIO_I8042=y
++CONFIG_SERIO_SERPORT=y
++# CONFIG_SERIO_CT82C710 is not set
+# CONFIG_SERIO_RAW is not set
+# CONFIG_GAMEPORT is not set
+
+#
+# Serial drivers
+#
-+CONFIG_SERIAL_8250=y
-+CONFIG_SERIAL_8250_CONSOLE=y
-+CONFIG_SERIAL_8250_PCI=y
-+CONFIG_SERIAL_8250_NR_UARTS=8
-+CONFIG_SERIAL_8250_RUNTIME_UARTS=4
-+CONFIG_SERIAL_8250_EXTENDED=y
-+CONFIG_SERIAL_8250_SHARE_IRQ=y
-+# CONFIG_SERIAL_8250_DETECT_IRQ is not set
-+# CONFIG_SERIAL_8250_RSA is not set
+
+#
+# Non-8250 serial port support
+#
-+CONFIG_SERIAL_CORE=y
-+CONFIG_SERIAL_CORE_CONSOLE=y
-+# CONFIG_SERIAL_JSM is not set
+CONFIG_UNIX98_PTYS=y
+CONFIG_LEGACY_PTYS=y
+CONFIG_LEGACY_PTY_COUNT=256
+# Watchdog Cards
+#
+# CONFIG_WATCHDOG is not set
-+# CONFIG_HW_RANDOM is not set
-+CONFIG_EFI_RTC=y
++CONFIG_HW_RANDOM=y
++CONFIG_HW_RANDOM_VIA=y
++# CONFIG_NVRAM is not set
++# CONFIG_RTC is not set
++# CONFIG_GEN_RTC is not set
+# CONFIG_DTLK is not set
+# CONFIG_R3964 is not set
-+# CONFIG_APPLICOM is not set
+
+#
+# Ftape, the floppy tape device driver
+#
-+CONFIG_AGP=y
-+# CONFIG_AGP_SIS is not set
-+# CONFIG_AGP_VIA is not set
-+CONFIG_DRM=y
-+# CONFIG_DRM_TDFX is not set
-+# CONFIG_DRM_R128 is not set
-+CONFIG_DRM_RADEON=y
-+# CONFIG_DRM_MGA is not set
-+# CONFIG_DRM_SIS is not set
-+# CONFIG_DRM_VIA is not set
-+# CONFIG_DRM_SAVAGE is not set
++# CONFIG_MWAVE is not set
++# CONFIG_PC8736x_GPIO is not set
++# CONFIG_NSC_GPIO is not set
++# CONFIG_CS5535_GPIO is not set
+# CONFIG_RAW_DRIVER is not set
-+# CONFIG_HPET is not set
+# CONFIG_HANGCHECK_TIMER is not set
+
+#
+#
+# I2C support
+#
-+CONFIG_I2C=y
-+CONFIG_I2C_CHARDEV=y
-+
-+#
-+# I2C Algorithms
-+#
-+CONFIG_I2C_ALGOBIT=y
-+CONFIG_I2C_ALGOPCF=y
-+# CONFIG_I2C_ALGOPCA is not set
-+
-+#
-+# I2C Hardware Bus support
-+#
-+# CONFIG_I2C_ALI1535 is not set
-+# CONFIG_I2C_ALI1563 is not set
-+# CONFIG_I2C_ALI15X3 is not set
-+# CONFIG_I2C_AMD756 is not set
-+# CONFIG_I2C_AMD8111 is not set
-+# CONFIG_I2C_I801 is not set
-+# CONFIG_I2C_I810 is not set
-+# CONFIG_I2C_PIIX4 is not set
-+# CONFIG_I2C_NFORCE2 is not set
-+# CONFIG_I2C_OCORES is not set
-+# CONFIG_I2C_PARPORT_LIGHT is not set
-+# CONFIG_I2C_PROSAVAGE is not set
-+# CONFIG_I2C_SAVAGE4 is not set
-+# CONFIG_I2C_SIS5595 is not set
-+# CONFIG_I2C_SIS630 is not set
-+# CONFIG_I2C_SIS96X is not set
-+# CONFIG_I2C_STUB is not set
-+# CONFIG_I2C_VIA is not set
-+# CONFIG_I2C_VIAPRO is not set
-+# CONFIG_I2C_VOODOO3 is not set
-+# CONFIG_I2C_PCA_ISA is not set
-+
-+#
-+# Miscellaneous I2C Chip support
-+#
-+# CONFIG_SENSORS_DS1337 is not set
-+# CONFIG_SENSORS_DS1374 is not set
-+# CONFIG_SENSORS_EEPROM is not set
-+# CONFIG_SENSORS_PCF8574 is not set
-+# CONFIG_SENSORS_PCA9539 is not set
-+# CONFIG_SENSORS_PCF8591 is not set
-+# CONFIG_SENSORS_MAX6875 is not set
-+# CONFIG_I2C_DEBUG_CORE is not set
-+# CONFIG_I2C_DEBUG_ALGO is not set
-+# CONFIG_I2C_DEBUG_BUS is not set
-+# CONFIG_I2C_DEBUG_CHIP is not set
++# CONFIG_I2C is not set
+
+#
+# SPI support
+#
+# Hardware Monitoring support
+#
-+CONFIG_HWMON=y
++# CONFIG_HWMON is not set
+# CONFIG_HWMON_VID is not set
-+# CONFIG_SENSORS_ABITUGURU is not set
-+# CONFIG_SENSORS_ADM1021 is not set
-+# CONFIG_SENSORS_ADM1025 is not set
-+# CONFIG_SENSORS_ADM1026 is not set
-+# CONFIG_SENSORS_ADM1031 is not set
-+# CONFIG_SENSORS_ADM9240 is not set
-+# CONFIG_SENSORS_ASB100 is not set
-+# CONFIG_SENSORS_ATXP1 is not set
-+# CONFIG_SENSORS_DS1621 is not set
-+# CONFIG_SENSORS_F71805F is not set
-+# CONFIG_SENSORS_FSCHER is not set
-+# CONFIG_SENSORS_FSCPOS is not set
-+# CONFIG_SENSORS_GL518SM is not set
-+# CONFIG_SENSORS_GL520SM is not set
-+# CONFIG_SENSORS_IT87 is not set
-+# CONFIG_SENSORS_LM63 is not set
-+# CONFIG_SENSORS_LM75 is not set
-+# CONFIG_SENSORS_LM77 is not set
-+# CONFIG_SENSORS_LM78 is not set
-+# CONFIG_SENSORS_LM80 is not set
-+# CONFIG_SENSORS_LM83 is not set
-+# CONFIG_SENSORS_LM85 is not set
-+# CONFIG_SENSORS_LM87 is not set
-+# CONFIG_SENSORS_LM90 is not set
-+# CONFIG_SENSORS_LM92 is not set
-+# CONFIG_SENSORS_MAX1619 is not set
-+# CONFIG_SENSORS_PC87360 is not set
-+# CONFIG_SENSORS_SIS5595 is not set
-+# CONFIG_SENSORS_SMSC47M1 is not set
-+# CONFIG_SENSORS_SMSC47M192 is not set
-+# CONFIG_SENSORS_SMSC47B397 is not set
-+# CONFIG_SENSORS_VIA686A is not set
-+# CONFIG_SENSORS_VT8231 is not set
-+# CONFIG_SENSORS_W83781D is not set
-+# CONFIG_SENSORS_W83791D is not set
-+# CONFIG_SENSORS_W83792D is not set
-+# CONFIG_SENSORS_W83L785TS is not set
-+# CONFIG_SENSORS_W83627HF is not set
-+# CONFIG_SENSORS_W83627EHF is not set
-+# CONFIG_HWMON_DEBUG_CHIP is not set
+
+#
+# Misc devices
+#
+# Multimedia devices
+#
-+CONFIG_VIDEO_DEV=y
-+CONFIG_VIDEO_V4L1=y
-+CONFIG_VIDEO_V4L1_COMPAT=y
-+CONFIG_VIDEO_V4L2=y
-+
-+#
-+# Video Capture Adapters
-+#
-+
-+#
-+# Video Capture Adapters
-+#
-+# CONFIG_VIDEO_ADV_DEBUG is not set
-+# CONFIG_VIDEO_VIVI is not set
-+# CONFIG_VIDEO_BT848 is not set
-+# CONFIG_VIDEO_CPIA is not set
-+# CONFIG_VIDEO_CPIA2 is not set
-+# CONFIG_VIDEO_SAA5246A is not set
-+# CONFIG_VIDEO_SAA5249 is not set
-+# CONFIG_TUNER_3036 is not set
-+# CONFIG_VIDEO_STRADIS is not set
-+# CONFIG_VIDEO_ZORAN is not set
-+# CONFIG_VIDEO_SAA7134 is not set
-+# CONFIG_VIDEO_MXB is not set
-+# CONFIG_VIDEO_DPC is not set
-+# CONFIG_VIDEO_HEXIUM_ORION is not set
-+# CONFIG_VIDEO_HEXIUM_GEMINI is not set
-+# CONFIG_VIDEO_CX88 is not set
-+
-+#
-+# Encoders and Decoders
-+#
-+# CONFIG_VIDEO_MSP3400 is not set
-+# CONFIG_VIDEO_CS53L32A is not set
-+# CONFIG_VIDEO_TLV320AIC23B is not set
-+# CONFIG_VIDEO_WM8775 is not set
-+# CONFIG_VIDEO_WM8739 is not set
-+# CONFIG_VIDEO_CX2341X is not set
-+# CONFIG_VIDEO_CX25840 is not set
-+# CONFIG_VIDEO_SAA711X is not set
-+# CONFIG_VIDEO_SAA7127 is not set
-+# CONFIG_VIDEO_UPD64031A is not set
-+# CONFIG_VIDEO_UPD64083 is not set
-+
-+#
-+# V4L USB devices
-+#
-+# CONFIG_VIDEO_PVRUSB2 is not set
-+# CONFIG_VIDEO_EM28XX is not set
-+# CONFIG_USB_VICAM is not set
-+# CONFIG_USB_IBMCAM is not set
-+# CONFIG_USB_KONICAWC is not set
-+# CONFIG_USB_QUICKCAM_MESSENGER is not set
-+# CONFIG_USB_ET61X251 is not set
-+# CONFIG_VIDEO_OVCAMCHIP is not set
-+# CONFIG_USB_W9968CF is not set
-+# CONFIG_USB_OV511 is not set
-+# CONFIG_USB_SE401 is not set
-+# CONFIG_USB_SN9C102 is not set
-+# CONFIG_USB_STV680 is not set
-+# CONFIG_USB_ZC0301 is not set
-+# CONFIG_USB_PWC is not set
-+
-+#
-+# Radio Adapters
-+#
-+# CONFIG_RADIO_GEMTEK_PCI is not set
-+# CONFIG_RADIO_MAXIRADIO is not set
-+# CONFIG_RADIO_MAESTRO is not set
-+# CONFIG_USB_DSBR is not set
++# CONFIG_VIDEO_DEV is not set
+
+#
+# Digital Video Broadcasting Devices
+#
+# CONFIG_DVB is not set
-+# CONFIG_USB_DABUSB is not set
+
+#
+# Graphics support
+#
+CONFIG_FIRMWARE_EDID=y
-+CONFIG_FB=y
-+CONFIG_FB_CFB_FILLRECT=y
-+CONFIG_FB_CFB_COPYAREA=y
-+CONFIG_FB_CFB_IMAGEBLIT=y
-+# CONFIG_FB_MACMODES is not set
-+# CONFIG_FB_BACKLIGHT is not set
-+CONFIG_FB_MODE_HELPERS=y
-+# CONFIG_FB_TILEBLITTING is not set
-+# CONFIG_FB_CIRRUS is not set
-+# CONFIG_FB_PM2 is not set
-+# CONFIG_FB_CYBER2000 is not set
-+# CONFIG_FB_ASILIANT is not set
-+# CONFIG_FB_IMSTT is not set
-+# CONFIG_FB_S1D13XXX is not set
-+# CONFIG_FB_NVIDIA is not set
-+# CONFIG_FB_RIVA is not set
-+# CONFIG_FB_MATROX is not set
-+# CONFIG_FB_RADEON is not set
-+# CONFIG_FB_ATY128 is not set
-+# CONFIG_FB_ATY is not set
-+# CONFIG_FB_SAVAGE is not set
-+# CONFIG_FB_SIS is not set
-+# CONFIG_FB_NEOMAGIC is not set
-+# CONFIG_FB_KYRO is not set
-+# CONFIG_FB_3DFX is not set
-+# CONFIG_FB_VOODOO1 is not set
-+# CONFIG_FB_TRIDENT is not set
-+# CONFIG_FB_VIRTUAL is not set
++# CONFIG_FB is not set
+
+#
+# Console display driver support
+CONFIG_VGA_CONSOLE=y
+# CONFIG_VGACON_SOFT_SCROLLBACK is not set
+CONFIG_DUMMY_CONSOLE=y
-+CONFIG_FRAMEBUFFER_CONSOLE=y
-+# CONFIG_FRAMEBUFFER_CONSOLE_ROTATION is not set
-+# CONFIG_FONTS is not set
-+CONFIG_FONT_8x8=y
-+CONFIG_FONT_8x16=y
-+
-+#
-+# Logo configuration
-+#
-+CONFIG_LOGO=y
-+# CONFIG_LOGO_LINUX_MONO is not set
-+# CONFIG_LOGO_LINUX_VGA16 is not set
-+CONFIG_LOGO_LINUX_CLUT224=y
+# CONFIG_BACKLIGHT_LCD_SUPPORT is not set
+
+#
+# Sound
+#
-+CONFIG_SOUND=y
-+
-+#
-+# Advanced Linux Sound Architecture
-+#
-+CONFIG_SND=y
-+CONFIG_SND_TIMER=y
-+CONFIG_SND_PCM=y
-+CONFIG_SND_HWDEP=y
-+CONFIG_SND_RAWMIDI=y
-+CONFIG_SND_SEQUENCER=y
-+# CONFIG_SND_SEQ_DUMMY is not set
-+CONFIG_SND_OSSEMUL=y
-+CONFIG_SND_MIXER_OSS=y
-+CONFIG_SND_PCM_OSS=y
-+CONFIG_SND_PCM_OSS_PLUGINS=y
-+CONFIG_SND_SEQUENCER_OSS=y
-+# CONFIG_SND_DYNAMIC_MINORS is not set
-+CONFIG_SND_SUPPORT_OLD_API=y
-+CONFIG_SND_VERBOSE_PROCFS=y
-+# CONFIG_SND_VERBOSE_PRINTK is not set
-+# CONFIG_SND_DEBUG is not set
-+
-+#
-+# Generic devices
-+#
-+CONFIG_SND_MPU401_UART=y
-+CONFIG_SND_OPL3_LIB=y
-+CONFIG_SND_AC97_CODEC=y
-+CONFIG_SND_AC97_BUS=y
-+# CONFIG_SND_DUMMY is not set
-+# CONFIG_SND_VIRMIDI is not set
-+# CONFIG_SND_MTPAV is not set
-+# CONFIG_SND_SERIAL_U16550 is not set
-+# CONFIG_SND_MPU401 is not set
-+
-+#
-+# PCI devices
-+#
-+# CONFIG_SND_AD1889 is not set
-+# CONFIG_SND_ALS300 is not set
-+# CONFIG_SND_ALI5451 is not set
-+# CONFIG_SND_ATIIXP is not set
-+# CONFIG_SND_ATIIXP_MODEM is not set
-+# CONFIG_SND_AU8810 is not set
-+# CONFIG_SND_AU8820 is not set
-+# CONFIG_SND_AU8830 is not set
-+# CONFIG_SND_AZT3328 is not set
-+# CONFIG_SND_BT87X is not set
-+# CONFIG_SND_CA0106 is not set
-+# CONFIG_SND_CMIPCI is not set
-+# CONFIG_SND_CS4281 is not set
-+# CONFIG_SND_CS46XX is not set
-+# CONFIG_SND_EMU10K1 is not set
-+# CONFIG_SND_EMU10K1X is not set
-+# CONFIG_SND_ENS1370 is not set
-+# CONFIG_SND_ENS1371 is not set
-+# CONFIG_SND_ES1938 is not set
-+# CONFIG_SND_ES1968 is not set
-+CONFIG_SND_FM801=y
-+# CONFIG_SND_FM801_TEA575X_BOOL is not set
-+# CONFIG_SND_HDA_INTEL is not set
-+# CONFIG_SND_HDSP is not set
-+# CONFIG_SND_HDSPM is not set
-+# CONFIG_SND_ICE1712 is not set
-+# CONFIG_SND_ICE1724 is not set
-+# CONFIG_SND_INTEL8X0 is not set
-+# CONFIG_SND_INTEL8X0M is not set
-+# CONFIG_SND_KORG1212 is not set
-+# CONFIG_SND_MAESTRO3 is not set
-+# CONFIG_SND_MIXART is not set
-+# CONFIG_SND_NM256 is not set
-+# CONFIG_SND_PCXHR is not set
-+# CONFIG_SND_RME32 is not set
-+# CONFIG_SND_RME96 is not set
-+# CONFIG_SND_RME9652 is not set
-+# CONFIG_SND_SONICVIBES is not set
-+# CONFIG_SND_TRIDENT is not set
-+# CONFIG_SND_VIA82XX is not set
-+# CONFIG_SND_VIA82XX_MODEM is not set
-+# CONFIG_SND_VX222 is not set
-+# CONFIG_SND_YMFPCI is not set
-+
-+#
-+# USB devices
-+#
-+# CONFIG_SND_USB_AUDIO is not set
-+
-+#
-+# Open Sound System
-+#
-+# CONFIG_SOUND_PRIME is not set
++# CONFIG_SOUND is not set
+
+#
+# USB support
+#
-+CONFIG_USB_ARCH_HAS_HCD=y
-+CONFIG_USB_ARCH_HAS_OHCI=y
-+CONFIG_USB_ARCH_HAS_EHCI=y
-+CONFIG_USB=y
-+# CONFIG_USB_DEBUG is not set
-+
-+#
-+# Miscellaneous USB options
-+#
-+# CONFIG_USB_DEVICEFS is not set
-+CONFIG_USB_BANDWIDTH=y
-+# CONFIG_USB_DYNAMIC_MINORS is not set
-+# CONFIG_USB_SUSPEND is not set
-+# CONFIG_USB_OTG is not set
-+
-+#
-+# USB Host Controller Drivers
-+#
-+CONFIG_USB_EHCI_HCD=y
-+# CONFIG_USB_EHCI_SPLIT_ISO is not set
-+# CONFIG_USB_EHCI_ROOT_HUB_TT is not set
-+# CONFIG_USB_EHCI_TT_NEWSCHED is not set
-+# CONFIG_USB_ISP116X_HCD is not set
-+CONFIG_USB_OHCI_HCD=y
-+# CONFIG_USB_OHCI_BIG_ENDIAN is not set
-+CONFIG_USB_OHCI_LITTLE_ENDIAN=y
-+CONFIG_USB_UHCI_HCD=y
-+# CONFIG_USB_SL811_HCD is not set
-+
-+#
-+# USB Device Class drivers
-+#
-+# CONFIG_USB_ACM is not set
-+# CONFIG_USB_PRINTER is not set
++# CONFIG_USB_ARCH_HAS_HCD is not set
++# CONFIG_USB_ARCH_HAS_OHCI is not set
++# CONFIG_USB_ARCH_HAS_EHCI is not set
+
+#
+# NOTE: USB_STORAGE enables SCSI, and 'SCSI disk support'
+#
+
+#
-+# may also be needed; see USB_STORAGE Help for more information
-+#
-+CONFIG_USB_STORAGE=y
-+# CONFIG_USB_STORAGE_DEBUG is not set
-+# CONFIG_USB_STORAGE_DATAFAB is not set
-+# CONFIG_USB_STORAGE_FREECOM is not set
-+# CONFIG_USB_STORAGE_DPCM is not set
-+# CONFIG_USB_STORAGE_USBAT is not set
-+# CONFIG_USB_STORAGE_SDDR09 is not set
-+# CONFIG_USB_STORAGE_SDDR55 is not set
-+# CONFIG_USB_STORAGE_JUMPSHOT is not set
-+# CONFIG_USB_STORAGE_ALAUDA is not set
-+# CONFIG_USB_LIBUSUAL is not set
-+
-+#
-+# USB Input Devices
-+#
-+CONFIG_USB_HID=y
-+CONFIG_USB_HIDINPUT=y
-+# CONFIG_USB_HIDINPUT_POWERBOOK is not set
-+# CONFIG_HID_FF is not set
-+CONFIG_USB_HIDDEV=y
-+# CONFIG_USB_AIPTEK is not set
-+# CONFIG_USB_WACOM is not set
-+# CONFIG_USB_ACECAD is not set
-+# CONFIG_USB_KBTAB is not set
-+# CONFIG_USB_POWERMATE is not set
-+# CONFIG_USB_TOUCHSCREEN is not set
-+# CONFIG_USB_YEALINK is not set
-+# CONFIG_USB_XPAD is not set
-+# CONFIG_USB_ATI_REMOTE is not set
-+# CONFIG_USB_ATI_REMOTE2 is not set
-+# CONFIG_USB_KEYSPAN_REMOTE is not set
-+# CONFIG_USB_APPLETOUCH is not set
-+
-+#
-+# USB Imaging devices
-+#
-+# CONFIG_USB_MDC800 is not set
-+# CONFIG_USB_MICROTEK is not set
-+
-+#
-+# USB Network Adapters
-+#
-+# CONFIG_USB_CATC is not set
-+# CONFIG_USB_KAWETH is not set
-+# CONFIG_USB_PEGASUS is not set
-+# CONFIG_USB_RTL8150 is not set
-+# CONFIG_USB_USBNET is not set
-+CONFIG_USB_MON=y
-+
-+#
-+# USB port drivers
-+#
-+
-+#
-+# USB Serial Converter support
-+#
-+# CONFIG_USB_SERIAL is not set
-+
-+#
-+# USB Miscellaneous drivers
-+#
-+# CONFIG_USB_EMI62 is not set
-+# CONFIG_USB_EMI26 is not set
-+# CONFIG_USB_AUERSWALD is not set
-+# CONFIG_USB_RIO500 is not set
-+# CONFIG_USB_LEGOTOWER is not set
-+# CONFIG_USB_LCD is not set
-+# CONFIG_USB_LED is not set
-+# CONFIG_USB_CYPRESS_CY7C63 is not set
-+# CONFIG_USB_CYTHERM is not set
-+# CONFIG_USB_PHIDGETKIT is not set
-+# CONFIG_USB_PHIDGETSERVO is not set
-+# CONFIG_USB_IDMOUSE is not set
-+# CONFIG_USB_APPLEDISPLAY is not set
-+# CONFIG_USB_SISUSBVGA is not set
-+# CONFIG_USB_LD is not set
-+
-+#
-+# USB DSL modem support
-+#
-+
-+#
+# USB Gadget Support
+#
+# CONFIG_USB_GADGET is not set
+#
+# InfiniBand support
+#
-+# CONFIG_INFINIBAND is not set
+
+#
+# EDAC - error detection and reporting (RAS) (EXPERIMENTAL)
+#
++# CONFIG_EDAC is not set
+
+#
+# Real Time Clock
+# File systems
+#
+CONFIG_EXT2_FS=y
-+CONFIG_EXT2_FS_XATTR=y
-+# CONFIG_EXT2_FS_POSIX_ACL is not set
-+# CONFIG_EXT2_FS_SECURITY is not set
++# CONFIG_EXT2_FS_XATTR is not set
+# CONFIG_EXT2_FS_XIP is not set
+CONFIG_EXT3_FS=y
+CONFIG_EXT3_FS_XATTR=y
+CONFIG_JBD=y
+# CONFIG_JBD_DEBUG is not set
+CONFIG_FS_MBCACHE=y
-+# CONFIG_REISERFS_FS is not set
++CONFIG_REISERFS_FS=y
++# CONFIG_REISERFS_CHECK is not set
++# CONFIG_REISERFS_PROC_INFO is not set
++# CONFIG_REISERFS_FS_XATTR is not set
+# CONFIG_JFS_FS is not set
+# CONFIG_FS_POSIX_ACL is not set
+# CONFIG_XFS_FS is not set
+# CONFIG_QUOTA is not set
+CONFIG_DNOTIFY=y
+CONFIG_AUTOFS_FS=y
-+# CONFIG_AUTOFS4_FS is not set
++CONFIG_AUTOFS4_FS=y
+# CONFIG_FUSE_FS is not set
+
+#
+#
+CONFIG_ISO9660_FS=y
+CONFIG_JOLIET=y
-+# CONFIG_ZISOFS is not set
-+CONFIG_UDF_FS=y
-+CONFIG_UDF_NLS=y
++CONFIG_ZISOFS=y
++CONFIG_ZISOFS_FS=y
++# CONFIG_UDF_FS is not set
+
+#
+# DOS/FAT/NT Filesystems
+#
-+CONFIG_FAT_FS=y
-+CONFIG_MSDOS_FS=y
-+CONFIG_VFAT_FS=y
++CONFIG_FAT_FS=m
++CONFIG_MSDOS_FS=m
++CONFIG_VFAT_FS=m
+CONFIG_FAT_DEFAULT_CODEPAGE=437
+CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1"
+# CONFIG_NTFS_FS is not set
+# CONFIG_BEFS_FS is not set
+# CONFIG_BFS_FS is not set
+# CONFIG_EFS_FS is not set
-+# CONFIG_CRAMFS is not set
++CONFIG_CRAMFS=y
+# CONFIG_VXFS_FS is not set
+# CONFIG_HPFS_FS is not set
+# CONFIG_QNX4FS_FS is not set
+CONFIG_NFS_FS=y
+CONFIG_NFS_V3=y
+# CONFIG_NFS_V3_ACL is not set
-+CONFIG_NFS_V4=y
++# CONFIG_NFS_V4 is not set
+# CONFIG_NFS_DIRECTIO is not set
-+CONFIG_NFSD=y
-+CONFIG_NFSD_V3=y
-+# CONFIG_NFSD_V3_ACL is not set
-+# CONFIG_NFSD_V4 is not set
-+# CONFIG_NFSD_TCP is not set
++# CONFIG_NFSD is not set
+CONFIG_ROOT_NFS=y
+CONFIG_LOCKD=y
+CONFIG_LOCKD_V4=y
-+CONFIG_EXPORTFS=y
+CONFIG_NFS_COMMON=y
+CONFIG_SUNRPC=y
-+CONFIG_SUNRPC_GSS=y
-+CONFIG_RPCSEC_GSS_KRB5=y
++# CONFIG_RPCSEC_GSS_KRB5 is not set
+# CONFIG_RPCSEC_GSS_SPKM3 is not set
+# CONFIG_SMB_FS is not set
+# CONFIG_CIFS is not set
+#
+# Partition Types
+#
-+CONFIG_PARTITION_ADVANCED=y
-+# CONFIG_ACORN_PARTITION is not set
-+# CONFIG_OSF_PARTITION is not set
-+# CONFIG_AMIGA_PARTITION is not set
-+# CONFIG_ATARI_PARTITION is not set
-+# CONFIG_MAC_PARTITION is not set
++# CONFIG_PARTITION_ADVANCED is not set
+CONFIG_MSDOS_PARTITION=y
-+# CONFIG_BSD_DISKLABEL is not set
-+# CONFIG_MINIX_SUBPARTITION is not set
-+# CONFIG_SOLARIS_X86_PARTITION is not set
-+# CONFIG_UNIXWARE_DISKLABEL is not set
-+# CONFIG_LDM_PARTITION is not set
-+# CONFIG_SGI_PARTITION is not set
-+# CONFIG_ULTRIX_PARTITION is not set
-+# CONFIG_SUN_PARTITION is not set
-+# CONFIG_KARMA_PARTITION is not set
-+CONFIG_EFI_PARTITION=y
+
+#
+# Native Language Support
+CONFIG_NLS=y
+CONFIG_NLS_DEFAULT="iso8859-1"
+CONFIG_NLS_CODEPAGE_437=y
-+CONFIG_NLS_CODEPAGE_737=y
-+CONFIG_NLS_CODEPAGE_775=y
-+CONFIG_NLS_CODEPAGE_850=y
-+CONFIG_NLS_CODEPAGE_852=y
-+CONFIG_NLS_CODEPAGE_855=y
-+CONFIG_NLS_CODEPAGE_857=y
-+CONFIG_NLS_CODEPAGE_860=y
-+CONFIG_NLS_CODEPAGE_861=y
-+CONFIG_NLS_CODEPAGE_862=y
-+CONFIG_NLS_CODEPAGE_863=y
-+CONFIG_NLS_CODEPAGE_864=y
-+CONFIG_NLS_CODEPAGE_865=y
-+CONFIG_NLS_CODEPAGE_866=y
-+CONFIG_NLS_CODEPAGE_869=y
-+CONFIG_NLS_CODEPAGE_936=y
-+CONFIG_NLS_CODEPAGE_950=y
-+CONFIG_NLS_CODEPAGE_932=y
-+CONFIG_NLS_CODEPAGE_949=y
-+CONFIG_NLS_CODEPAGE_874=y
-+CONFIG_NLS_ISO8859_8=y
++# CONFIG_NLS_CODEPAGE_737 is not set
++# CONFIG_NLS_CODEPAGE_775 is not set
++# CONFIG_NLS_CODEPAGE_850 is not set
++# CONFIG_NLS_CODEPAGE_852 is not set
++# CONFIG_NLS_CODEPAGE_855 is not set
++# CONFIG_NLS_CODEPAGE_857 is not set
++# CONFIG_NLS_CODEPAGE_860 is not set
++# CONFIG_NLS_CODEPAGE_861 is not set
++# CONFIG_NLS_CODEPAGE_862 is not set
++# CONFIG_NLS_CODEPAGE_863 is not set
++# CONFIG_NLS_CODEPAGE_864 is not set
++# CONFIG_NLS_CODEPAGE_865 is not set
++# CONFIG_NLS_CODEPAGE_866 is not set
++# CONFIG_NLS_CODEPAGE_869 is not set
++# CONFIG_NLS_CODEPAGE_936 is not set
++# CONFIG_NLS_CODEPAGE_950 is not set
++# CONFIG_NLS_CODEPAGE_932 is not set
++# CONFIG_NLS_CODEPAGE_949 is not set
++# CONFIG_NLS_CODEPAGE_874 is not set
++# CONFIG_NLS_ISO8859_8 is not set
+# CONFIG_NLS_CODEPAGE_1250 is not set
-+CONFIG_NLS_CODEPAGE_1251=y
++# CONFIG_NLS_CODEPAGE_1251 is not set
+# CONFIG_NLS_ASCII is not set
+CONFIG_NLS_ISO8859_1=y
-+CONFIG_NLS_ISO8859_2=y
-+CONFIG_NLS_ISO8859_3=y
-+CONFIG_NLS_ISO8859_4=y
-+CONFIG_NLS_ISO8859_5=y
-+CONFIG_NLS_ISO8859_6=y
-+CONFIG_NLS_ISO8859_7=y
-+CONFIG_NLS_ISO8859_9=y
-+CONFIG_NLS_ISO8859_13=y
-+CONFIG_NLS_ISO8859_14=y
-+CONFIG_NLS_ISO8859_15=y
-+CONFIG_NLS_KOI8_R=y
-+CONFIG_NLS_KOI8_U=y
-+CONFIG_NLS_UTF8=y
++# CONFIG_NLS_ISO8859_2 is not set
++# CONFIG_NLS_ISO8859_3 is not set
++# CONFIG_NLS_ISO8859_4 is not set
++# CONFIG_NLS_ISO8859_5 is not set
++# CONFIG_NLS_ISO8859_6 is not set
++# CONFIG_NLS_ISO8859_7 is not set
++# CONFIG_NLS_ISO8859_9 is not set
++# CONFIG_NLS_ISO8859_13 is not set
++# CONFIG_NLS_ISO8859_14 is not set
++# CONFIG_NLS_ISO8859_15 is not set
++# CONFIG_NLS_KOI8_R is not set
++# CONFIG_NLS_KOI8_U is not set
++# CONFIG_NLS_UTF8 is not set
+
+#
-+# Library routines
-+#
-+# CONFIG_CRC_CCITT is not set
-+# CONFIG_CRC16 is not set
-+CONFIG_CRC32=y
-+# CONFIG_LIBCRC32C is not set
-+CONFIG_PLIST=y
-+CONFIG_GENERIC_HARDIRQS=y
-+CONFIG_GENERIC_IRQ_PROBE=y
-+CONFIG_GENERIC_PENDING_IRQ=y
-+CONFIG_IRQ_PER_CPU=y
-+
-+#
-+# Instrumentation Support
++# Instrumentation Support
+#
+# CONFIG_PROFILING is not set
+# CONFIG_KPROBES is not set
+#
+# Kernel hacking
+#
++CONFIG_TRACE_IRQFLAGS_SUPPORT=y
+# CONFIG_PRINTK_TIME is not set
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_UNUSED_SYMBOLS=y
+CONFIG_DEBUG_KERNEL=y
-+CONFIG_LOG_BUF_SHIFT=17
++CONFIG_LOG_BUF_SHIFT=14
+CONFIG_DETECT_SOFTLOCKUP=y
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_DEBUG_SLAB is not set
+# CONFIG_DEBUG_RT_MUTEXES is not set
+# CONFIG_RT_MUTEX_TESTER is not set
+# CONFIG_DEBUG_SPINLOCK is not set
-+CONFIG_DEBUG_MUTEXES=y
++# CONFIG_DEBUG_MUTEXES is not set
+# CONFIG_DEBUG_RWSEMS is not set
++# CONFIG_DEBUG_LOCK_ALLOC is not set
++# CONFIG_PROVE_LOCKING is not set
+# CONFIG_DEBUG_SPINLOCK_SLEEP is not set
+# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
+# CONFIG_DEBUG_KOBJECT is not set
-+# CONFIG_DEBUG_INFO is not set
++# CONFIG_DEBUG_HIGHMEM is not set
++CONFIG_DEBUG_BUGVERBOSE=y
++CONFIG_DEBUG_INFO=y
+# CONFIG_DEBUG_FS is not set
+# CONFIG_DEBUG_VM is not set
++CONFIG_FRAME_POINTER=y
++# CONFIG_UNWIND_INFO is not set
+CONFIG_FORCED_INLINING=y
+# CONFIG_RCU_TORTURE_TEST is not set
-+CONFIG_IA64_GRANULE_16MB=y
-+# CONFIG_IA64_GRANULE_64MB is not set
-+CONFIG_IA64_PRINT_HAZARDS=y
-+# CONFIG_DISABLE_VHPT is not set
-+# CONFIG_IA64_DEBUG_CMPXCHG is not set
-+# CONFIG_IA64_DEBUG_IRQ is not set
++CONFIG_EARLY_PRINTK=y
++# CONFIG_DEBUG_STACKOVERFLOW is not set
++# CONFIG_DEBUG_STACK_USAGE is not set
++# CONFIG_DEBUG_PAGEALLOC is not set
++# CONFIG_DEBUG_RODATA is not set
++# CONFIG_4KSTACKS is not set
+
+#
+# Security options
+# CONFIG_CRYPTO_HMAC is not set
+# CONFIG_CRYPTO_NULL is not set
+# CONFIG_CRYPTO_MD4 is not set
-+CONFIG_CRYPTO_MD5=y
++CONFIG_CRYPTO_MD5=m
+# CONFIG_CRYPTO_SHA1 is not set
+# CONFIG_CRYPTO_SHA256 is not set
+# CONFIG_CRYPTO_SHA512 is not set
+# CONFIG_CRYPTO_WP512 is not set
+# CONFIG_CRYPTO_TGR192 is not set
-+CONFIG_CRYPTO_DES=y
++# CONFIG_CRYPTO_DES is not set
+# CONFIG_CRYPTO_BLOWFISH is not set
+# CONFIG_CRYPTO_TWOFISH is not set
+# CONFIG_CRYPTO_SERPENT is not set
+# CONFIG_CRYPTO_AES is not set
++# CONFIG_CRYPTO_AES_586 is not set
+# CONFIG_CRYPTO_CAST5 is not set
+# CONFIG_CRYPTO_CAST6 is not set
+# CONFIG_CRYPTO_TEA is not set
+# CONFIG_CRYPTO_ANUBIS is not set
+# CONFIG_CRYPTO_DEFLATE is not set
+# CONFIG_CRYPTO_MICHAEL_MIC is not set
-+# CONFIG_CRYPTO_CRC32C is not set
++CONFIG_CRYPTO_CRC32C=m
+# CONFIG_CRYPTO_TEST is not set
+
+#
+# Hardware crypto devices
+#
-+# CONFIG_XEN_SMPBOOT is not set
-+# CONFIG_XEN_DEVMEM is not set
++# CONFIG_CRYPTO_DEV_PADLOCK is not set
++CONFIG_XEN=y
+CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+CONFIG_XEN_PRIVCMD=y
+CONFIG_XEN_XENBUS_DEV=y
+# CONFIG_XEN_BACKEND is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
+# CONFIG_XEN_GRANT_DEV is not set
-+CONFIG_XEN_FRAMEBUFFER=y
-+CONFIG_XEN_KEYBOARD=y
-+# CONFIG_XEN_SCRUB_PAGES is not set
-+# CONFIG_XEN_DISABLE_SERIAL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
++CONFIG_XEN_SCRUB_PAGES=y
++CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_COMPAT_030002_AND_LATER=y
+# CONFIG_XEN_COMPAT_030004_AND_LATER is not set
+CONFIG_XEN_COMPAT=0x030002
+CONFIG_HAVE_IRQ_IGNORE_UNHANDLED=y
+CONFIG_NO_IDLE_HZ=y
++CONFIG_XEN_SMPBOOT=y
+CONFIG_XEN_BALLOON=y
-+CONFIG_XEN_XENCOMM=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xenU_x86_32 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_32
---- linux-2.6.18/buildconfigs/linux-defconfig_xenU_x86_32 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_32 2007-12-23 11:15:00.706184434 +0100
-@@ -0,0 +1,950 @@
++CONFIG_XEN_DEVMEM=y
++
++#
++# Library routines
++#
++# CONFIG_CRC_CCITT is not set
++# CONFIG_CRC16 is not set
++# CONFIG_CRC32 is not set
++CONFIG_LIBCRC32C=m
++CONFIG_ZLIB_INFLATE=y
++CONFIG_PLIST=y
++CONFIG_GENERIC_HARDIRQS=y
++CONFIG_GENERIC_IRQ_PROBE=y
++CONFIG_GENERIC_PENDING_IRQ=y
++CONFIG_X86_SMP=y
++CONFIG_X86_BIOS_REBOOT=y
++CONFIG_X86_TRAMPOLINE=y
++CONFIG_X86_NO_TSS=y
++CONFIG_X86_NO_IDT=y
++CONFIG_KTIME_SCALAR=y
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xenU_x86_64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_64 2008-05-19 00:33:19.713282430 +0300
+@@ -0,0 +1,905 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Tue Oct 16 09:31:29 2007
++# Mon Feb 18 10:42:42 2008
+#
-+CONFIG_X86_32=y
++CONFIG_X86_64=y
++CONFIG_64BIT=y
++CONFIG_X86=y
+CONFIG_LOCKDEP_SUPPORT=y
+CONFIG_STACKTRACE_SUPPORT=y
+CONFIG_SEMAPHORE_SLEEPERS=y
-+CONFIG_X86=y
+CONFIG_MMU=y
++CONFIG_RWSEM_GENERIC_SPINLOCK=y
++CONFIG_GENERIC_HWEIGHT=y
++CONFIG_GENERIC_CALIBRATE_DELAY=y
++CONFIG_X86_CMPXCHG=y
++CONFIG_EARLY_PRINTK=y
+CONFIG_GENERIC_ISA_DMA=y
+CONFIG_GENERIC_IOMAP=y
-+CONFIG_GENERIC_HWEIGHT=y
+CONFIG_ARCH_MAY_HAVE_PC_FDC=y
+CONFIG_DMI=y
++CONFIG_AUDIT_ARCH=y
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+#
+#
+# Processor type and features
+#
-+CONFIG_SMP=y
-+# CONFIG_X86_PC is not set
-+CONFIG_X86_XEN=y
-+# CONFIG_X86_ELAN is not set
-+# CONFIG_X86_VOYAGER is not set
-+# CONFIG_X86_NUMAQ is not set
-+# CONFIG_X86_SUMMIT is not set
-+# CONFIG_X86_BIGSMP is not set
-+# CONFIG_X86_VISWS is not set
-+# CONFIG_X86_GENERICARCH is not set
-+# CONFIG_X86_ES7000 is not set
-+# CONFIG_M386 is not set
-+# CONFIG_M486 is not set
-+# CONFIG_M586 is not set
-+# CONFIG_M586TSC is not set
-+# CONFIG_M586MMX is not set
-+CONFIG_M686=y
-+# CONFIG_MPENTIUMII is not set
-+# CONFIG_MPENTIUMIII is not set
-+# CONFIG_MPENTIUMM is not set
-+# CONFIG_MPENTIUM4 is not set
-+# CONFIG_MK6 is not set
-+# CONFIG_MK7 is not set
++CONFIG_X86_PC=y
++# CONFIG_X86_VSMP is not set
+# CONFIG_MK8 is not set
-+# CONFIG_MCRUSOE is not set
-+# CONFIG_MEFFICEON is not set
-+# CONFIG_MWINCHIPC6 is not set
-+# CONFIG_MWINCHIP2 is not set
-+# CONFIG_MWINCHIP3D is not set
-+# CONFIG_MGEODEGX1 is not set
-+# CONFIG_MGEODE_LX is not set
-+# CONFIG_MCYRIXIII is not set
-+# CONFIG_MVIAC3_2 is not set
-+# CONFIG_X86_GENERIC is not set
-+CONFIG_X86_CMPXCHG=y
-+CONFIG_X86_XADD=y
-+CONFIG_X86_L1_CACHE_SHIFT=5
-+CONFIG_RWSEM_XCHGADD_ALGORITHM=y
-+CONFIG_GENERIC_CALIBRATE_DELAY=y
-+CONFIG_X86_PPRO_FENCE=y
-+CONFIG_X86_WP_WORKS_OK=y
-+CONFIG_X86_INVLPG=y
-+CONFIG_X86_BSWAP=y
-+CONFIG_X86_POPAD_OK=y
-+CONFIG_X86_CMPXCHG64=y
++# CONFIG_MPSC is not set
++CONFIG_GENERIC_CPU=y
++CONFIG_X86_64_XEN=y
++CONFIG_X86_NO_TSS=y
++CONFIG_X86_NO_IDT=y
++CONFIG_X86_L1_CACHE_BYTES=128
++CONFIG_X86_L1_CACHE_SHIFT=7
++CONFIG_X86_INTERNODE_CACHE_BYTES=128
+CONFIG_X86_GOOD_APIC=y
-+CONFIG_X86_USE_PPRO_CHECKSUM=y
-+CONFIG_NR_CPUS=8
++# CONFIG_MICROCODE is not set
++# CONFIG_X86_MSR is not set
++CONFIG_X86_CPUID=y
++CONFIG_X86_XEN_GENAPIC=y
++CONFIG_SMP=y
+CONFIG_PREEMPT_NONE=y
+# CONFIG_PREEMPT_VOLUNTARY is not set
+# CONFIG_PREEMPT is not set
+CONFIG_PREEMPT_BKL=y
-+CONFIG_VM86=y
-+# CONFIG_TOSHIBA is not set
-+# CONFIG_I8K is not set
-+# CONFIG_X86_REBOOTFIXUPS is not set
-+# CONFIG_X86_MSR is not set
-+CONFIG_X86_CPUID=y
-+CONFIG_SWIOTLB=y
-+
-+#
-+# Firmware Drivers
-+#
-+# CONFIG_EDD is not set
-+# CONFIG_DELL_RBU is not set
-+# CONFIG_DCDBAS is not set
-+# CONFIG_NOHIGHMEM is not set
-+CONFIG_HIGHMEM4G=y
-+# CONFIG_HIGHMEM64G is not set
-+CONFIG_PAGE_OFFSET=0xC0000000
-+CONFIG_HIGHMEM=y
++CONFIG_ARCH_FLATMEM_ENABLE=y
+CONFIG_SELECT_MEMORY_MODEL=y
+CONFIG_FLATMEM_MANUAL=y
+# CONFIG_DISCONTIGMEM_MANUAL is not set
+# CONFIG_SPARSEMEM_STATIC is not set
+CONFIG_SPLIT_PTLOCK_CPUS=4
+CONFIG_RESOURCES_64BIT=y
-+# CONFIG_HIGHPTE is not set
-+# CONFIG_REGPARM is not set
++CONFIG_NR_CPUS=8
++CONFIG_HOTPLUG_CPU=y
++CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
++CONFIG_SWIOTLB=y
++# CONFIG_CRASH_DUMP is not set
++CONFIG_PHYSICAL_START=0x200000
+CONFIG_SECCOMP=y
+CONFIG_HZ_100=y
+# CONFIG_HZ_250 is not set
+# CONFIG_HZ_1000 is not set
+CONFIG_HZ=100
-+# CONFIG_CRASH_DUMP is not set
-+CONFIG_PHYSICAL_START=0x100000
-+CONFIG_HOTPLUG_CPU=y
-+CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
++# CONFIG_REORDER is not set
++CONFIG_GENERIC_HARDIRQS=y
++CONFIG_GENERIC_IRQ_PROBE=y
++CONFIG_ISA_DMA_API=y
++CONFIG_GENERIC_PENDING_IRQ=y
+
+#
-+# Bus options (PCI, PCMCIA, EISA, MCA, ISA)
++# Bus options (PCI etc.)
+#
+# CONFIG_PCI is not set
-+CONFIG_ISA_DMA_API=y
-+# CONFIG_SCx200 is not set
+
+#
+# PCCARD (PCMCIA/CardBus) support
+#
+
+#
-+# Executable file formats
++# Executable file formats / Emulations
+#
+CONFIG_BINFMT_ELF=y
-+# CONFIG_BINFMT_AOUT is not set
+# CONFIG_BINFMT_MISC is not set
++CONFIG_IA32_EMULATION=y
++CONFIG_IA32_AOUT=y
++CONFIG_COMPAT=y
++CONFIG_SYSVIPC_COMPAT=y
+
+#
+# Networking
+# Input Device Drivers
+#
+CONFIG_INPUT_KEYBOARD=y
-+# CONFIG_KEYBOARD_ATKBD is not set
++CONFIG_KEYBOARD_ATKBD=y
+# CONFIG_KEYBOARD_SUNKBD is not set
+# CONFIG_KEYBOARD_LKKBD is not set
+# CONFIG_KEYBOARD_XTKBD is not set
+CONFIG_SERIO_I8042=y
+CONFIG_SERIO_SERPORT=y
+# CONFIG_SERIO_CT82C710 is not set
++CONFIG_SERIO_LIBPS2=y
+# CONFIG_SERIO_RAW is not set
+# CONFIG_GAMEPORT is not set
+
+#
+# CONFIG_WATCHDOG is not set
+CONFIG_HW_RANDOM=y
-+CONFIG_HW_RANDOM_VIA=y
+# CONFIG_NVRAM is not set
+# CONFIG_RTC is not set
+# CONFIG_GEN_RTC is not set
+#
+# CONFIG_MWAVE is not set
+# CONFIG_PC8736x_GPIO is not set
-+# CONFIG_NSC_GPIO is not set
-+# CONFIG_CS5535_GPIO is not set
+# CONFIG_RAW_DRIVER is not set
+# CONFIG_HANGCHECK_TIMER is not set
+
+#
+
+#
++# Firmware Drivers
++#
++# CONFIG_EDD is not set
++# CONFIG_DELL_RBU is not set
++# CONFIG_DCDBAS is not set
++
++#
+# File systems
+#
+CONFIG_EXT2_FS=y
+# CONFIG_DEBUG_SPINLOCK_SLEEP is not set
+# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
+# CONFIG_DEBUG_KOBJECT is not set
-+# CONFIG_DEBUG_HIGHMEM is not set
-+CONFIG_DEBUG_BUGVERBOSE=y
+CONFIG_DEBUG_INFO=y
+# CONFIG_DEBUG_FS is not set
+# CONFIG_DEBUG_VM is not set
+# CONFIG_UNWIND_INFO is not set
+CONFIG_FORCED_INLINING=y
+# CONFIG_RCU_TORTURE_TEST is not set
-+CONFIG_EARLY_PRINTK=y
++# CONFIG_DEBUG_RODATA is not set
+# CONFIG_DEBUG_STACKOVERFLOW is not set
+# CONFIG_DEBUG_STACK_USAGE is not set
-+# CONFIG_DEBUG_PAGEALLOC is not set
-+# CONFIG_DEBUG_RODATA is not set
-+# CONFIG_4KSTACKS is not set
+
+#
+# Security options
+# CONFIG_CRYPTO_TWOFISH is not set
+# CONFIG_CRYPTO_SERPENT is not set
+# CONFIG_CRYPTO_AES is not set
-+# CONFIG_CRYPTO_AES_586 is not set
++# CONFIG_CRYPTO_AES_X86_64 is not set
+# CONFIG_CRYPTO_CAST5 is not set
+# CONFIG_CRYPTO_CAST6 is not set
+# CONFIG_CRYPTO_TEA is not set
+#
+# Hardware crypto devices
+#
-+# CONFIG_CRYPTO_DEV_PADLOCK is not set
+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+CONFIG_XEN_PRIVCMD=y
+CONFIG_XEN_XENBUS_DEV=y
+# CONFIG_XEN_BACKEND is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
++CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
+CONFIG_XEN_SCRUB_PAGES=y
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_SMPBOOT=y
+CONFIG_XEN_BALLOON=y
+CONFIG_XEN_DEVMEM=y
-+# CONFIG_XEN_GRANT_DEV is not set
+
+#
+# Library routines
+CONFIG_LIBCRC32C=m
+CONFIG_ZLIB_INFLATE=y
+CONFIG_PLIST=y
-+CONFIG_GENERIC_HARDIRQS=y
-+CONFIG_GENERIC_IRQ_PROBE=y
-+CONFIG_GENERIC_PENDING_IRQ=y
-+CONFIG_X86_SMP=y
-+CONFIG_X86_BIOS_REBOOT=y
-+CONFIG_X86_TRAMPOLINE=y
-+CONFIG_X86_NO_TSS=y
-+CONFIG_X86_NO_IDT=y
-+CONFIG_KTIME_SCALAR=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xenU_x86_64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_64
---- linux-2.6.18/buildconfigs/linux-defconfig_xenU_x86_64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xenU_x86_64 2007-12-23 11:15:00.706184434 +0100
-@@ -0,0 +1,1241 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen_ia64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_ia64 2008-05-19 00:33:19.717282661 +0300
+@@ -0,0 +1,1701 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Tue Oct 16 09:32:52 2007
++# Tue Feb 19 11:20:00 2008
+#
-+CONFIG_X86_64=y
-+CONFIG_64BIT=y
-+CONFIG_X86=y
-+CONFIG_LOCKDEP_SUPPORT=y
-+CONFIG_STACKTRACE_SUPPORT=y
-+CONFIG_SEMAPHORE_SLEEPERS=y
-+CONFIG_MMU=y
-+CONFIG_RWSEM_GENERIC_SPINLOCK=y
-+CONFIG_GENERIC_HWEIGHT=y
-+CONFIG_GENERIC_CALIBRATE_DELAY=y
-+CONFIG_X86_CMPXCHG=y
-+CONFIG_EARLY_PRINTK=y
-+CONFIG_GENERIC_ISA_DMA=y
-+CONFIG_GENERIC_IOMAP=y
-+CONFIG_ARCH_MAY_HAVE_PC_FDC=y
-+CONFIG_DMI=y
-+CONFIG_AUDIT_ARCH=y
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+#
+# General setup
+#
+CONFIG_LOCALVERSION=""
-+# CONFIG_LOCALVERSION_AUTO is not set
++CONFIG_LOCALVERSION_AUTO=y
+CONFIG_SWAP=y
+CONFIG_SYSVIPC=y
+CONFIG_POSIX_MQUEUE=y
+CONFIG_BSD_PROCESS_ACCT=y
+# CONFIG_BSD_PROCESS_ACCT_V3 is not set
+# CONFIG_TASKSTATS is not set
-+CONFIG_AUDIT=y
-+CONFIG_AUDITSYSCALL=y
-+# CONFIG_IKCONFIG is not set
++# CONFIG_AUDIT is not set
++CONFIG_IKCONFIG=y
++CONFIG_IKCONFIG_PROC=y
+# CONFIG_CPUSETS is not set
+# CONFIG_RELAY is not set
+CONFIG_INITRAMFS_SOURCE=""
-+# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
++CONFIG_CC_OPTIMIZE_FOR_SIZE=y
+# CONFIG_EMBEDDED is not set
-+CONFIG_UID16=y
+CONFIG_SYSCTL=y
+CONFIG_KALLSYMS=y
-+# CONFIG_KALLSYMS_ALL is not set
++CONFIG_KALLSYMS_ALL=y
+CONFIG_KALLSYMS_EXTRA_PASS=y
+CONFIG_HOTPLUG=y
+CONFIG_PRINTK=y
+CONFIG_MODULE_UNLOAD=y
+# CONFIG_MODULE_FORCE_UNLOAD is not set
+CONFIG_MODVERSIONS=y
-+# CONFIG_MODULE_SRCVERSION_ALL is not set
++CONFIG_MODULE_SRCVERSION_ALL=y
+CONFIG_KMOD=y
+CONFIG_STOP_MACHINE=y
+
+#
+# Block layer
+#
-+CONFIG_LBD=y
+# CONFIG_BLK_DEV_IO_TRACE is not set
-+CONFIG_LSF=y
+
+#
+# IO Schedulers
+#
+# Processor type and features
+#
-+CONFIG_X86_PC=y
-+# CONFIG_X86_VSMP is not set
-+# CONFIG_MK8 is not set
-+CONFIG_MPSC=y
-+# CONFIG_GENERIC_CPU is not set
-+CONFIG_X86_64_XEN=y
-+CONFIG_X86_NO_TSS=y
-+CONFIG_X86_NO_IDT=y
-+CONFIG_X86_L1_CACHE_BYTES=128
-+CONFIG_X86_L1_CACHE_SHIFT=7
-+CONFIG_X86_INTERNODE_CACHE_BYTES=128
-+CONFIG_X86_GOOD_APIC=y
-+# CONFIG_MICROCODE is not set
-+# CONFIG_X86_MSR is not set
-+CONFIG_X86_CPUID=y
-+CONFIG_X86_XEN_GENAPIC=y
++CONFIG_IA64=y
++CONFIG_64BIT=y
++CONFIG_MMU=y
++CONFIG_SWIOTLB=y
++CONFIG_RWSEM_XCHGADD_ALGORITHM=y
++CONFIG_GENERIC_FIND_NEXT_BIT=y
++CONFIG_GENERIC_CALIBRATE_DELAY=y
++CONFIG_TIME_INTERPOLATION=y
++CONFIG_DMI=y
++CONFIG_EFI=y
++CONFIG_GENERIC_IOMAP=y
++CONFIG_XEN=y
++CONFIG_XEN_IA64_EXPOSE_P2M=y
++CONFIG_XEN_IA64_EXPOSE_P2M_USE_DTR=y
++CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER=y
++CONFIG_DMA_IS_DMA32=y
++CONFIG_AUDIT_ARCH=y
++CONFIG_IA64_GENERIC=y
++# CONFIG_IA64_DIG is not set
++# CONFIG_IA64_HP_ZX1 is not set
++# CONFIG_IA64_HP_ZX1_SWIOTLB is not set
++# CONFIG_IA64_SGI_SN2 is not set
++# CONFIG_IA64_HP_SIM is not set
++# CONFIG_IA64_XEN is not set
++# CONFIG_ITANIUM is not set
++CONFIG_MCKINLEY=y
++# CONFIG_IA64_PAGE_SIZE_4KB is not set
++# CONFIG_IA64_PAGE_SIZE_8KB is not set
++CONFIG_IA64_PAGE_SIZE_16KB=y
++# CONFIG_IA64_PAGE_SIZE_64KB is not set
++CONFIG_PGTABLE_3=y
++# CONFIG_PGTABLE_4 is not set
++CONFIG_HZ_100=y
++# CONFIG_HZ_250 is not set
++# CONFIG_HZ_1000 is not set
++CONFIG_HZ=100
++CONFIG_IA64_L1_CACHE_SHIFT=7
++CONFIG_IA64_CYCLONE=y
++CONFIG_IOSAPIC=y
++# CONFIG_IA64_SGI_SN_XP is not set
++CONFIG_FORCE_MAX_ZONEORDER=11
+CONFIG_SMP=y
-+CONFIG_PREEMPT_NONE=y
-+# CONFIG_PREEMPT_VOLUNTARY is not set
++CONFIG_NR_CPUS=16
++CONFIG_HOTPLUG_CPU=y
++CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
++# CONFIG_SCHED_SMT is not set
++# CONFIG_PERMIT_BSP_REMOVE is not set
+# CONFIG_PREEMPT is not set
-+CONFIG_PREEMPT_BKL=y
-+CONFIG_ARCH_FLATMEM_ENABLE=y
+CONFIG_SELECT_MEMORY_MODEL=y
-+CONFIG_FLATMEM_MANUAL=y
-+# CONFIG_DISCONTIGMEM_MANUAL is not set
++# CONFIG_FLATMEM_MANUAL is not set
++CONFIG_DISCONTIGMEM_MANUAL=y
+# CONFIG_SPARSEMEM_MANUAL is not set
-+CONFIG_FLATMEM=y
++CONFIG_DISCONTIGMEM=y
+CONFIG_FLAT_NODE_MEM_MAP=y
++CONFIG_NEED_MULTIPLE_NODES=y
+# CONFIG_SPARSEMEM_STATIC is not set
+CONFIG_SPLIT_PTLOCK_CPUS=4
++# CONFIG_MIGRATION is not set
+CONFIG_RESOURCES_64BIT=y
-+CONFIG_NR_CPUS=16
-+# CONFIG_HOTPLUG_CPU is not set
-+CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
-+CONFIG_SWIOTLB=y
-+# CONFIG_CRASH_DUMP is not set
-+CONFIG_PHYSICAL_START=0x200000
-+CONFIG_SECCOMP=y
-+CONFIG_HZ_100=y
-+# CONFIG_HZ_250 is not set
-+# CONFIG_HZ_1000 is not set
-+CONFIG_HZ=100
-+# CONFIG_REORDER is not set
-+CONFIG_GENERIC_HARDIRQS=y
-+CONFIG_GENERIC_IRQ_PROBE=y
-+CONFIG_ISA_DMA_API=y
-+CONFIG_GENERIC_PENDING_IRQ=y
++CONFIG_ARCH_SELECT_MEMORY_MODEL=y
++CONFIG_ARCH_DISCONTIGMEM_ENABLE=y
++CONFIG_ARCH_FLATMEM_ENABLE=y
++CONFIG_ARCH_SPARSEMEM_ENABLE=y
++CONFIG_ARCH_DISCONTIGMEM_DEFAULT=y
++CONFIG_NUMA=y
++CONFIG_NODES_SHIFT=10
++CONFIG_VIRTUAL_MEM_MAP=y
++CONFIG_HOLES_IN_ZONE=y
++CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID=y
++CONFIG_HAVE_ARCH_NODEDATA_EXTENSION=y
++# CONFIG_IA32_SUPPORT is not set
++# CONFIG_IA64_MCA_RECOVERY is not set
++CONFIG_PERFMON=y
++CONFIG_IA64_PALINFO=y
++CONFIG_SGI_SN=y
+
+#
-+# Bus options (PCI etc.)
++# SN Devices
+#
-+# CONFIG_PCI is not set
++# CONFIG_SGI_IOC3 is not set
++CONFIG_KEXEC=y
+
+#
-+# PCCARD (PCMCIA/CardBus) support
++# Firmware Drivers
+#
-+# CONFIG_PCCARD is not set
++CONFIG_EFI_VARS=y
++CONFIG_EFI_PCDP=y
++CONFIG_BINFMT_ELF=y
++CONFIG_BINFMT_MISC=y
++
++#
++# Power management and ACPI
++#
++CONFIG_PM=y
++CONFIG_PM_LEGACY=y
++# CONFIG_PM_DEBUG is not set
++
++#
++# ACPI (Advanced Configuration and Power Interface) Support
++#
++CONFIG_ACPI=y
++CONFIG_ACPI_BUTTON=y
++CONFIG_ACPI_FAN=y
++# CONFIG_ACPI_DOCK is not set
++CONFIG_ACPI_PROCESSOR=y
++CONFIG_ACPI_HOTPLUG_CPU=y
++CONFIG_ACPI_THERMAL=y
++CONFIG_ACPI_NUMA=y
++CONFIG_ACPI_BLACKLIST_YEAR=0
++# CONFIG_ACPI_DEBUG is not set
++CONFIG_ACPI_EC=y
++CONFIG_ACPI_POWER=y
++CONFIG_ACPI_SYSTEM=y
++CONFIG_ACPI_CONTAINER=y
++
++#
++# CPU Frequency scaling
++#
++# CONFIG_CPU_FREQ is not set
++
++#
++# Bus options (PCI, PCMCIA)
++#
++CONFIG_PCI=y
++CONFIG_PCI_DOMAINS=y
++CONFIG_XEN_PCIDEV_FRONTEND=y
++# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
++# CONFIG_PCIEPORTBUS is not set
++# CONFIG_PCI_DEBUG is not set
+
+#
+# PCI Hotplug Support
+#
++CONFIG_HOTPLUG_PCI=y
++# CONFIG_HOTPLUG_PCI_FAKE is not set
++CONFIG_HOTPLUG_PCI_ACPI=y
++# CONFIG_HOTPLUG_PCI_ACPI_IBM is not set
++# CONFIG_HOTPLUG_PCI_CPCI is not set
++# CONFIG_HOTPLUG_PCI_SHPC is not set
++# CONFIG_HOTPLUG_PCI_SGI is not set
+
+#
-+# Executable file formats / Emulations
++# PCCARD (PCMCIA/CardBus) support
+#
-+CONFIG_BINFMT_ELF=y
-+CONFIG_BINFMT_MISC=y
-+CONFIG_IA32_EMULATION=y
-+# CONFIG_IA32_AOUT is not set
-+CONFIG_COMPAT=y
-+CONFIG_SYSVIPC_COMPAT=y
++# CONFIG_PCCARD is not set
+
+#
+# Networking
+#
+# CONFIG_NETDEBUG is not set
+CONFIG_PACKET=y
-+CONFIG_PACKET_MMAP=y
++# CONFIG_PACKET_MMAP is not set
+CONFIG_UNIX=y
+CONFIG_XFRM=y
-+CONFIG_XFRM_USER=y
-+CONFIG_NET_KEY=m
++# CONFIG_XFRM_USER is not set
++# CONFIG_NET_KEY is not set
+CONFIG_INET=y
+CONFIG_IP_MULTICAST=y
-+CONFIG_IP_ADVANCED_ROUTER=y
-+CONFIG_ASK_IP_FIB_HASH=y
-+# CONFIG_IP_FIB_TRIE is not set
++# CONFIG_IP_ADVANCED_ROUTER is not set
+CONFIG_IP_FIB_HASH=y
-+CONFIG_IP_MULTIPLE_TABLES=y
-+CONFIG_IP_ROUTE_FWMARK=y
-+CONFIG_IP_ROUTE_MULTIPATH=y
-+# CONFIG_IP_ROUTE_MULTIPATH_CACHED is not set
-+CONFIG_IP_ROUTE_VERBOSE=y
+CONFIG_IP_PNP=y
+CONFIG_IP_PNP_DHCP=y
-+CONFIG_IP_PNP_BOOTP=y
-+CONFIG_IP_PNP_RARP=y
-+CONFIG_NET_IPIP=m
-+CONFIG_NET_IPGRE=m
-+CONFIG_NET_IPGRE_BROADCAST=y
-+CONFIG_IP_MROUTE=y
-+CONFIG_IP_PIMSM_V1=y
-+CONFIG_IP_PIMSM_V2=y
-+# CONFIG_ARPD is not set
++# CONFIG_IP_PNP_BOOTP is not set
++# CONFIG_IP_PNP_RARP is not set
++# CONFIG_NET_IPIP is not set
++# CONFIG_NET_IPGRE is not set
++# CONFIG_IP_MROUTE is not set
++CONFIG_ARPD=y
+CONFIG_SYN_COOKIES=y
-+CONFIG_INET_AH=m
-+CONFIG_INET_ESP=m
-+CONFIG_INET_IPCOMP=m
-+CONFIG_INET_XFRM_TUNNEL=m
-+CONFIG_INET_TUNNEL=m
++# CONFIG_INET_AH is not set
++# CONFIG_INET_ESP is not set
++# CONFIG_INET_IPCOMP is not set
++# CONFIG_INET_XFRM_TUNNEL is not set
++# CONFIG_INET_TUNNEL is not set
+CONFIG_INET_XFRM_MODE_TRANSPORT=y
+CONFIG_INET_XFRM_MODE_TUNNEL=y
-+# CONFIG_INET_DIAG is not set
++CONFIG_INET_DIAG=y
++CONFIG_INET_TCP_DIAG=y
+# CONFIG_TCP_CONG_ADVANCED is not set
+CONFIG_TCP_CONG_BIC=y
+
+#
+# IP: Virtual Server Configuration
+#
-+CONFIG_IP_VS=m
-+# CONFIG_IP_VS_DEBUG is not set
-+CONFIG_IP_VS_TAB_BITS=12
-+
-+#
-+# IPVS transport protocol load balancing support
-+#
-+CONFIG_IP_VS_PROTO_TCP=y
-+CONFIG_IP_VS_PROTO_UDP=y
-+CONFIG_IP_VS_PROTO_ESP=y
-+CONFIG_IP_VS_PROTO_AH=y
-+
-+#
-+# IPVS scheduler
-+#
-+CONFIG_IP_VS_RR=m
-+CONFIG_IP_VS_WRR=m
-+CONFIG_IP_VS_LC=m
-+CONFIG_IP_VS_WLC=m
-+CONFIG_IP_VS_LBLC=m
-+CONFIG_IP_VS_LBLCR=m
-+CONFIG_IP_VS_DH=m
-+CONFIG_IP_VS_SH=m
-+CONFIG_IP_VS_SED=m
-+CONFIG_IP_VS_NQ=m
-+
-+#
-+# IPVS application helper
-+#
-+CONFIG_IP_VS_FTP=m
-+CONFIG_IPV6=m
-+CONFIG_IPV6_PRIVACY=y
-+# CONFIG_IPV6_ROUTER_PREF is not set
-+CONFIG_INET6_AH=m
-+CONFIG_INET6_ESP=m
-+CONFIG_INET6_IPCOMP=m
-+CONFIG_INET6_XFRM_TUNNEL=m
-+CONFIG_INET6_TUNNEL=m
-+CONFIG_INET6_XFRM_MODE_TRANSPORT=m
-+CONFIG_INET6_XFRM_MODE_TUNNEL=m
-+CONFIG_IPV6_TUNNEL=m
++# CONFIG_IP_VS is not set
++# CONFIG_IPV6 is not set
++# CONFIG_INET6_XFRM_TUNNEL is not set
++# CONFIG_INET6_TUNNEL is not set
+# CONFIG_NETWORK_SECMARK is not set
+CONFIG_NETFILTER=y
+# CONFIG_NETFILTER_DEBUG is not set
+# Core Netfilter Configuration
+#
+# CONFIG_NETFILTER_NETLINK is not set
++# CONFIG_NF_CONNTRACK is not set
+# CONFIG_NETFILTER_XTABLES is not set
+
+#
+# IP: Netfilter Configuration
+#
-+CONFIG_IP_NF_CONNTRACK=m
-+CONFIG_IP_NF_CT_ACCT=y
-+# CONFIG_IP_NF_CONNTRACK_MARK is not set
-+# CONFIG_IP_NF_CONNTRACK_EVENTS is not set
-+CONFIG_IP_NF_CT_PROTO_SCTP=m
-+CONFIG_IP_NF_FTP=m
-+CONFIG_IP_NF_IRC=m
-+# CONFIG_IP_NF_NETBIOS_NS is not set
-+CONFIG_IP_NF_TFTP=m
-+CONFIG_IP_NF_AMANDA=m
-+# CONFIG_IP_NF_PPTP is not set
-+# CONFIG_IP_NF_H323 is not set
-+# CONFIG_IP_NF_SIP is not set
-+CONFIG_IP_NF_QUEUE=m
-+
-+#
-+# IPv6: Netfilter Configuration (EXPERIMENTAL)
-+#
-+# CONFIG_IP6_NF_QUEUE is not set
++# CONFIG_IP_NF_CONNTRACK is not set
++# CONFIG_IP_NF_QUEUE is not set
+
+#
+# Bridge: Netfilter Configuration
+#
-+CONFIG_BRIDGE_NF_EBTABLES=m
-+CONFIG_BRIDGE_EBT_BROUTE=m
-+CONFIG_BRIDGE_EBT_T_FILTER=m
-+CONFIG_BRIDGE_EBT_T_NAT=m
-+CONFIG_BRIDGE_EBT_802_3=m
-+CONFIG_BRIDGE_EBT_AMONG=m
-+CONFIG_BRIDGE_EBT_ARP=m
-+CONFIG_BRIDGE_EBT_IP=m
-+CONFIG_BRIDGE_EBT_LIMIT=m
-+CONFIG_BRIDGE_EBT_MARK=m
-+CONFIG_BRIDGE_EBT_PKTTYPE=m
-+CONFIG_BRIDGE_EBT_STP=m
-+CONFIG_BRIDGE_EBT_VLAN=m
-+CONFIG_BRIDGE_EBT_ARPREPLY=m
-+CONFIG_BRIDGE_EBT_DNAT=m
-+CONFIG_BRIDGE_EBT_MARK_T=m
-+CONFIG_BRIDGE_EBT_REDIRECT=m
-+CONFIG_BRIDGE_EBT_SNAT=m
-+CONFIG_BRIDGE_EBT_LOG=m
-+# CONFIG_BRIDGE_EBT_ULOG is not set
++# CONFIG_BRIDGE_NF_EBTABLES is not set
+
+#
+# DCCP Configuration (EXPERIMENTAL)
+#
+# SCTP Configuration (EXPERIMENTAL)
+#
-+CONFIG_IP_SCTP=m
-+# CONFIG_SCTP_DBG_MSG is not set
-+# CONFIG_SCTP_DBG_OBJCNT is not set
-+# CONFIG_SCTP_HMAC_NONE is not set
-+# CONFIG_SCTP_HMAC_SHA1 is not set
-+CONFIG_SCTP_HMAC_MD5=y
++# CONFIG_IP_SCTP is not set
+
+#
+# TIPC Configuration (EXPERIMENTAL)
+#
+# CONFIG_TIPC is not set
-+CONFIG_ATM=m
-+CONFIG_ATM_CLIP=m
-+# CONFIG_ATM_CLIP_NO_ICMP is not set
-+CONFIG_ATM_LANE=m
-+# CONFIG_ATM_MPOA is not set
-+CONFIG_ATM_BR2684=m
-+# CONFIG_ATM_BR2684_IPFILTER is not set
-+CONFIG_BRIDGE=m
-+CONFIG_VLAN_8021Q=m
++# CONFIG_ATM is not set
++CONFIG_BRIDGE=y
++# CONFIG_VLAN_8021Q is not set
+# CONFIG_DECNET is not set
-+CONFIG_LLC=m
++CONFIG_LLC=y
+# CONFIG_LLC2 is not set
-+CONFIG_IPX=m
-+# CONFIG_IPX_INTERN is not set
-+CONFIG_ATALK=m
-+CONFIG_DEV_APPLETALK=m
-+CONFIG_IPDDP=m
-+CONFIG_IPDDP_ENCAP=y
-+CONFIG_IPDDP_DECAP=y
++# CONFIG_IPX is not set
++# CONFIG_ATALK is not set
+# CONFIG_X25 is not set
+# CONFIG_LAPB is not set
+# CONFIG_ECONET is not set
-+CONFIG_WAN_ROUTER=m
++# CONFIG_WAN_ROUTER is not set
+
+#
+# QoS and/or fair queueing
+#
-+CONFIG_NET_SCHED=y
-+CONFIG_NET_SCH_CLK_JIFFIES=y
-+# CONFIG_NET_SCH_CLK_GETTIMEOFDAY is not set
-+# CONFIG_NET_SCH_CLK_CPU is not set
-+
-+#
-+# Queueing/Scheduling
-+#
-+CONFIG_NET_SCH_CBQ=m
-+CONFIG_NET_SCH_HTB=m
-+CONFIG_NET_SCH_HFSC=m
-+CONFIG_NET_SCH_ATM=m
-+CONFIG_NET_SCH_PRIO=m
-+CONFIG_NET_SCH_RED=m
-+CONFIG_NET_SCH_SFQ=m
-+CONFIG_NET_SCH_TEQL=m
-+CONFIG_NET_SCH_TBF=m
-+CONFIG_NET_SCH_GRED=m
-+CONFIG_NET_SCH_DSMARK=m
-+CONFIG_NET_SCH_NETEM=m
-+CONFIG_NET_SCH_INGRESS=m
-+
-+#
-+# Classification
-+#
-+CONFIG_NET_CLS=y
-+# CONFIG_NET_CLS_BASIC is not set
-+CONFIG_NET_CLS_TCINDEX=m
-+CONFIG_NET_CLS_ROUTE4=m
-+CONFIG_NET_CLS_ROUTE=y
-+CONFIG_NET_CLS_FW=m
-+CONFIG_NET_CLS_U32=m
-+CONFIG_CLS_U32_PERF=y
-+# CONFIG_CLS_U32_MARK is not set
-+CONFIG_NET_CLS_RSVP=m
-+CONFIG_NET_CLS_RSVP6=m
-+# CONFIG_NET_EMATCH is not set
-+# CONFIG_NET_CLS_ACT is not set
-+CONFIG_NET_CLS_POLICE=y
-+CONFIG_NET_CLS_IND=y
-+CONFIG_NET_ESTIMATOR=y
++# CONFIG_NET_SCHED is not set
+
+#
+# Network testing
+#
+# CONFIG_NET_PKTGEN is not set
+# CONFIG_HAMRADIO is not set
-+CONFIG_IRDA=m
-+
-+#
-+# IrDA protocols
-+#
-+CONFIG_IRLAN=m
-+CONFIG_IRNET=m
-+CONFIG_IRCOMM=m
-+# CONFIG_IRDA_ULTRA is not set
-+
-+#
-+# IrDA options
-+#
-+CONFIG_IRDA_CACHE_LAST_LSAP=y
-+CONFIG_IRDA_FAST_RR=y
-+# CONFIG_IRDA_DEBUG is not set
-+
-+#
-+# Infrared-port device drivers
-+#
-+
-+#
-+# SIR device drivers
-+#
-+CONFIG_IRTTY_SIR=m
-+
-+#
-+# Dongle support
-+#
-+CONFIG_DONGLE=y
-+CONFIG_ESI_DONGLE=m
-+CONFIG_ACTISYS_DONGLE=m
-+CONFIG_TEKRAM_DONGLE=m
-+# CONFIG_TOIM3232_DONGLE is not set
-+CONFIG_LITELINK_DONGLE=m
-+CONFIG_MA600_DONGLE=m
-+CONFIG_GIRBIL_DONGLE=m
-+CONFIG_MCP2120_DONGLE=m
-+CONFIG_OLD_BELKIN_DONGLE=m
-+CONFIG_ACT200L_DONGLE=m
-+
-+#
-+# Old SIR device drivers
-+#
-+
-+#
-+# Old Serial dongle support
-+#
-+
-+#
-+# FIR device drivers
-+#
-+# CONFIG_NSC_FIR is not set
-+# CONFIG_WINBOND_FIR is not set
-+# CONFIG_SMC_IRCC_FIR is not set
-+# CONFIG_ALI_FIR is not set
-+# CONFIG_VIA_FIR is not set
-+CONFIG_BT=m
-+CONFIG_BT_L2CAP=m
-+CONFIG_BT_SCO=m
-+CONFIG_BT_RFCOMM=m
-+CONFIG_BT_RFCOMM_TTY=y
-+CONFIG_BT_BNEP=m
-+CONFIG_BT_BNEP_MC_FILTER=y
-+CONFIG_BT_BNEP_PROTO_FILTER=y
-+CONFIG_BT_HIDP=m
-+
-+#
-+# Bluetooth device drivers
-+#
-+CONFIG_BT_HCIUART=m
-+CONFIG_BT_HCIUART_H4=y
-+CONFIG_BT_HCIUART_BCSP=y
-+CONFIG_BT_HCIVHCI=m
++# CONFIG_IRDA is not set
++# CONFIG_BT is not set
+# CONFIG_IEEE80211 is not set
-+CONFIG_WIRELESS_EXT=y
+
+#
+# Device Drivers
+#
+# Plug and Play support
+#
++# CONFIG_PNP is not set
+
+#
+# Block devices
+#
-+CONFIG_BLK_DEV_FD=m
++# CONFIG_BLK_CPQ_DA is not set
++CONFIG_BLK_CPQ_CISS_DA=y
++# CONFIG_CISS_SCSI_TAPE is not set
++# CONFIG_BLK_DEV_DAC960 is not set
++# CONFIG_BLK_DEV_UMEM is not set
+# CONFIG_BLK_DEV_COW_COMMON is not set
-+CONFIG_BLK_DEV_LOOP=m
-+CONFIG_BLK_DEV_CRYPTOLOOP=m
++CONFIG_BLK_DEV_LOOP=y
++CONFIG_BLK_DEV_CRYPTOLOOP=y
+CONFIG_BLK_DEV_NBD=m
++# CONFIG_BLK_DEV_SX8 is not set
++# CONFIG_BLK_DEV_UB is not set
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_RAM_COUNT=16
-+CONFIG_BLK_DEV_RAM_SIZE=16384
++CONFIG_BLK_DEV_RAM_SIZE=4096
+CONFIG_BLK_DEV_RAM_BLOCKSIZE=1024
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_CDROM_PKTCDVD is not set
+#
+# ATA/ATAPI/MFM/RLL support
+#
-+# CONFIG_IDE is not set
++CONFIG_IDE=y
++CONFIG_IDE_MAX_HWIFS=4
++CONFIG_BLK_DEV_IDE=y
++
++#
++# Please see Documentation/ide.txt for help/info on IDE drives
++#
++# CONFIG_BLK_DEV_IDE_SATA is not set
++CONFIG_BLK_DEV_IDEDISK=y
++# CONFIG_IDEDISK_MULTI_MODE is not set
++CONFIG_BLK_DEV_IDECD=y
++# CONFIG_BLK_DEV_IDETAPE is not set
++CONFIG_BLK_DEV_IDEFLOPPY=y
++CONFIG_BLK_DEV_IDESCSI=y
++# CONFIG_IDE_TASK_IOCTL is not set
++
++#
++# IDE chipset support/bugfixes
++#
++# CONFIG_IDE_GENERIC is not set
++CONFIG_BLK_DEV_IDEPCI=y
++# CONFIG_IDEPCI_SHARE_IRQ is not set
++# CONFIG_BLK_DEV_OFFBOARD is not set
++CONFIG_BLK_DEV_GENERIC=y
++# CONFIG_BLK_DEV_OPTI621 is not set
++CONFIG_BLK_DEV_IDEDMA_PCI=y
++# CONFIG_BLK_DEV_IDEDMA_FORCED is not set
++CONFIG_IDEDMA_PCI_AUTO=y
++# CONFIG_IDEDMA_ONLYDISK is not set
++# CONFIG_BLK_DEV_AEC62XX is not set
++# CONFIG_BLK_DEV_ALI15X3 is not set
++# CONFIG_BLK_DEV_AMD74XX is not set
++CONFIG_BLK_DEV_CMD64X=y
++# CONFIG_BLK_DEV_TRIFLEX is not set
++# CONFIG_BLK_DEV_CY82C693 is not set
++# CONFIG_BLK_DEV_CS5520 is not set
++# CONFIG_BLK_DEV_CS5530 is not set
++# CONFIG_BLK_DEV_HPT34X is not set
++# CONFIG_BLK_DEV_HPT366 is not set
++# CONFIG_BLK_DEV_SC1200 is not set
++CONFIG_BLK_DEV_PIIX=y
++# CONFIG_BLK_DEV_IT821X is not set
++# CONFIG_BLK_DEV_NS87415 is not set
++# CONFIG_BLK_DEV_PDC202XX_OLD is not set
++# CONFIG_BLK_DEV_PDC202XX_NEW is not set
++# CONFIG_BLK_DEV_SVWKS is not set
++# CONFIG_BLK_DEV_SIIMAGE is not set
++# CONFIG_BLK_DEV_SLC90E66 is not set
++# CONFIG_BLK_DEV_TRM290 is not set
++# CONFIG_BLK_DEV_VIA82CXXX is not set
++# CONFIG_IDE_ARM is not set
++CONFIG_BLK_DEV_IDEDMA=y
++# CONFIG_IDEDMA_IVB is not set
++CONFIG_IDEDMA_AUTO=y
++# CONFIG_BLK_DEV_HD is not set
+
+#
+# SCSI device support
+#
+# CONFIG_RAID_ATTRS is not set
-+CONFIG_SCSI=m
++CONFIG_SCSI=y
+CONFIG_SCSI_PROC_FS=y
+
+#
+# SCSI support type (disk, tape, CD-ROM)
+#
-+CONFIG_BLK_DEV_SD=m
-+CONFIG_CHR_DEV_ST=m
-+CONFIG_CHR_DEV_OSST=m
-+CONFIG_BLK_DEV_SR=m
++CONFIG_BLK_DEV_SD=y
++CONFIG_CHR_DEV_ST=y
++CONFIG_CHR_DEV_OSST=y
++CONFIG_BLK_DEV_SR=y
+CONFIG_BLK_DEV_SR_VENDOR=y
-+CONFIG_CHR_DEV_SG=m
++CONFIG_CHR_DEV_SG=y
+# CONFIG_CHR_DEV_SCH is not set
+
+#
+# Some SCSI devices (e.g. CD jukebox) support multiple LUNs
+#
-+# CONFIG_SCSI_MULTI_LUN is not set
++CONFIG_SCSI_MULTI_LUN=y
+CONFIG_SCSI_CONSTANTS=y
+CONFIG_SCSI_LOGGING=y
+
+#
+# SCSI Transport Attributes
+#
-+CONFIG_SCSI_SPI_ATTRS=m
-+CONFIG_SCSI_FC_ATTRS=m
++CONFIG_SCSI_SPI_ATTRS=y
++CONFIG_SCSI_FC_ATTRS=y
+# CONFIG_SCSI_ISCSI_ATTRS is not set
-+# CONFIG_SCSI_SAS_ATTRS is not set
++CONFIG_SCSI_SAS_ATTRS=y
+
+#
+# SCSI low-level drivers
+#
+# CONFIG_ISCSI_TCP is not set
-+CONFIG_SCSI_SATA=m
++# CONFIG_BLK_DEV_3W_XXXX_RAID is not set
++# CONFIG_SCSI_3W_9XXX is not set
++# CONFIG_SCSI_ACARD is not set
++# CONFIG_SCSI_AACRAID is not set
++# CONFIG_SCSI_AIC7XXX is not set
++# CONFIG_SCSI_AIC7XXX_OLD is not set
++# CONFIG_SCSI_AIC79XX is not set
++# CONFIG_MEGARAID_NEWGEN is not set
++# CONFIG_MEGARAID_LEGACY is not set
++# CONFIG_MEGARAID_SAS is not set
++# CONFIG_SCSI_SATA is not set
++# CONFIG_SCSI_HPTIOP is not set
++# CONFIG_SCSI_DMX3191D is not set
++# CONFIG_SCSI_FUTURE_DOMAIN is not set
++# CONFIG_SCSI_IPS is not set
++# CONFIG_SCSI_INITIO is not set
++# CONFIG_SCSI_INIA100 is not set
++CONFIG_SCSI_SYM53C8XX_2=y
++CONFIG_SCSI_SYM53C8XX_DMA_ADDRESSING_MODE=1
++CONFIG_SCSI_SYM53C8XX_DEFAULT_TAGS=16
++CONFIG_SCSI_SYM53C8XX_MAX_TAGS=64
++CONFIG_SCSI_SYM53C8XX_MMIO=y
++# CONFIG_SCSI_IPR is not set
++CONFIG_SCSI_QLOGIC_1280=y
++# CONFIG_SCSI_QLA_FC is not set
++# CONFIG_SCSI_LPFC is not set
++# CONFIG_SCSI_DC395x is not set
++# CONFIG_SCSI_DC390T is not set
+# CONFIG_SCSI_DEBUG is not set
+
+#
+# Multi-device support (RAID and LVM)
+#
+CONFIG_MD=y
-+CONFIG_BLK_DEV_MD=y
-+CONFIG_MD_LINEAR=m
-+CONFIG_MD_RAID0=m
-+CONFIG_MD_RAID1=m
-+CONFIG_MD_RAID10=m
-+# CONFIG_MD_RAID456 is not set
-+CONFIG_MD_MULTIPATH=m
-+# CONFIG_MD_FAULTY is not set
-+CONFIG_BLK_DEV_DM=m
++# CONFIG_BLK_DEV_MD is not set
++CONFIG_BLK_DEV_DM=y
+CONFIG_DM_CRYPT=m
-+CONFIG_DM_SNAPSHOT=m
++CONFIG_DM_SNAPSHOT=y
+CONFIG_DM_MIRROR=m
+CONFIG_DM_ZERO=m
-+# CONFIG_DM_MULTIPATH is not set
++CONFIG_DM_MULTIPATH=m
++CONFIG_DM_MULTIPATH_EMC=m
+
+#
+# Fusion MPT device support
+#
-+# CONFIG_FUSION is not set
++CONFIG_FUSION=y
++CONFIG_FUSION_SPI=y
++# CONFIG_FUSION_FC is not set
++CONFIG_FUSION_SAS=y
++CONFIG_FUSION_MAX_SGE=128
++# CONFIG_FUSION_CTL is not set
+
+#
+# IEEE 1394 (FireWire) support
+#
++# CONFIG_IEEE1394 is not set
+
+#
+# I2O device support
+#
++# CONFIG_I2O is not set
+
+#
+# Network device support
+#
+CONFIG_NETDEVICES=y
-+CONFIG_DUMMY=m
-+CONFIG_BONDING=m
-+CONFIG_EQUALIZER=m
-+CONFIG_TUN=m
++CONFIG_DUMMY=y
++# CONFIG_BONDING is not set
++# CONFIG_EQUALIZER is not set
++CONFIG_TUN=y
++
++#
++# ARCnet devices
++#
++CONFIG_ARCNET=y
++# CONFIG_ARCNET_1201 is not set
++# CONFIG_ARCNET_1051 is not set
++# CONFIG_ARCNET_RAW is not set
++# CONFIG_ARCNET_CAP is not set
++# CONFIG_ARCNET_COM90xx is not set
++# CONFIG_ARCNET_COM90xxIO is not set
++# CONFIG_ARCNET_RIM_I is not set
++# CONFIG_ARCNET_COM20020 is not set
+
+#
+# PHY device support
+# Ethernet (10 or 100Mbit)
+#
+CONFIG_NET_ETHERNET=y
-+CONFIG_MII=m
++CONFIG_MII=y
++# CONFIG_HAPPYMEAL is not set
++# CONFIG_SUNGEM is not set
++# CONFIG_CASSINI is not set
++# CONFIG_NET_VENDOR_3COM is not set
++
++#
++# Tulip family network device support
++#
++CONFIG_NET_TULIP=y
++# CONFIG_DE2104X is not set
++CONFIG_TULIP=y
++CONFIG_TULIP_MWI=y
++CONFIG_TULIP_MMIO=y
++CONFIG_TULIP_NAPI=y
++CONFIG_TULIP_NAPI_HW_MITIGATION=y
++# CONFIG_DE4X5 is not set
++# CONFIG_WINBOND_840 is not set
++# CONFIG_DM9102 is not set
++# CONFIG_ULI526X is not set
++# CONFIG_HP100 is not set
++CONFIG_NET_PCI=y
++# CONFIG_PCNET32 is not set
++# CONFIG_AMD8111_ETH is not set
++# CONFIG_ADAPTEC_STARFIRE is not set
++# CONFIG_B44 is not set
++# CONFIG_FORCEDETH is not set
++# CONFIG_DGRS is not set
++CONFIG_EEPRO100=y
++CONFIG_E100=y
++# CONFIG_FEALNX is not set
++# CONFIG_NATSEMI is not set
++# CONFIG_NE2K_PCI is not set
++# CONFIG_8139CP is not set
++# CONFIG_8139TOO is not set
++# CONFIG_SIS900 is not set
++# CONFIG_EPIC100 is not set
++# CONFIG_SUNDANCE is not set
++# CONFIG_VIA_RHINE is not set
+
+#
+# Ethernet (1000 Mbit)
+#
++# CONFIG_ACENIC is not set
++# CONFIG_DL2K is not set
++CONFIG_E1000=y
++# CONFIG_E1000_NAPI is not set
++# CONFIG_E1000_DISABLE_PACKET_SPLIT is not set
++# CONFIG_NS83820 is not set
++# CONFIG_HAMACHI is not set
++# CONFIG_YELLOWFIN is not set
++# CONFIG_R8169 is not set
++# CONFIG_SIS190 is not set
++# CONFIG_SKGE is not set
++# CONFIG_SKY2 is not set
++# CONFIG_SK98LIN is not set
++# CONFIG_VIA_VELOCITY is not set
++CONFIG_TIGON3=y
++# CONFIG_BNX2 is not set
+
+#
+# Ethernet (10000 Mbit)
+#
++# CONFIG_CHELSIO_T1 is not set
++# CONFIG_IXGB is not set
++# CONFIG_S2IO is not set
++# CONFIG_MYRI10GE is not set
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+#
++# CONFIG_TR is not set
+
+#
+# Wireless LAN (non-hamradio)
+#
-+CONFIG_NET_RADIO=y
-+# CONFIG_NET_WIRELESS_RTNETLINK is not set
-+
-+#
-+# Obsolete Wireless cards support (pre-802.11)
-+#
-+# CONFIG_STRIP is not set
-+# CONFIG_HOSTAP is not set
++# CONFIG_NET_RADIO is not set
+
+#
+# Wan interfaces
+#
+# CONFIG_WAN is not set
-+
-+#
-+# ATM drivers
-+#
-+# CONFIG_ATM_DUMMY is not set
-+CONFIG_ATM_TCP=m
-+CONFIG_PPP=m
-+CONFIG_PPP_MULTILINK=y
-+CONFIG_PPP_FILTER=y
-+CONFIG_PPP_ASYNC=m
-+CONFIG_PPP_SYNC_TTY=m
-+CONFIG_PPP_DEFLATE=m
-+# CONFIG_PPP_BSDCOMP is not set
-+# CONFIG_PPP_MPPE is not set
-+CONFIG_PPPOE=m
-+CONFIG_PPPOATM=m
++# CONFIG_FDDI is not set
++# CONFIG_HIPPI is not set
++# CONFIG_PPP is not set
+# CONFIG_SLIP is not set
++# CONFIG_NET_FC is not set
+# CONFIG_SHAPER is not set
-+CONFIG_NETCONSOLE=m
++CONFIG_NETCONSOLE=y
+CONFIG_NETPOLL=y
+# CONFIG_NETPOLL_RX is not set
-+CONFIG_NETPOLL_TRAP=y
++# CONFIG_NETPOLL_TRAP is not set
+CONFIG_NET_POLL_CONTROLLER=y
+
+#
+# ISDN subsystem
+#
-+# CONFIG_ISDN is not set
++CONFIG_ISDN=m
++
++#
++# Old ISDN4Linux
++#
++# CONFIG_ISDN_I4L is not set
++
++#
++# CAPI subsystem
++#
++# CONFIG_ISDN_CAPI is not set
+
+#
+# Telephony Support
+# Userland interfaces
+#
+CONFIG_INPUT_MOUSEDEV=y
-+# CONFIG_INPUT_MOUSEDEV_PSAUX is not set
++CONFIG_INPUT_MOUSEDEV_PSAUX=y
+CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024
+CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768
-+# CONFIG_INPUT_JOYDEV is not set
++CONFIG_INPUT_JOYDEV=y
+# CONFIG_INPUT_TSDEV is not set
-+# CONFIG_INPUT_EVDEV is not set
++CONFIG_INPUT_EVDEV=y
+# CONFIG_INPUT_EVBUG is not set
+
+#
+# CONFIG_KEYBOARD_LKKBD is not set
+# CONFIG_KEYBOARD_XTKBD is not set
+# CONFIG_KEYBOARD_NEWTON is not set
-+# CONFIG_INPUT_MOUSE is not set
++CONFIG_INPUT_MOUSE=y
++CONFIG_MOUSE_PS2=y
++# CONFIG_MOUSE_SERIAL is not set
++# CONFIG_MOUSE_VSXXXAA is not set
+# CONFIG_INPUT_JOYSTICK is not set
+# CONFIG_INPUT_TOUCHSCREEN is not set
+# CONFIG_INPUT_MISC is not set
+#
+CONFIG_SERIO=y
+CONFIG_SERIO_I8042=y
-+CONFIG_SERIO_SERPORT=y
-+# CONFIG_SERIO_CT82C710 is not set
++# CONFIG_SERIO_SERPORT is not set
++# CONFIG_SERIO_PCIPS2 is not set
+CONFIG_SERIO_LIBPS2=y
+# CONFIG_SERIO_RAW is not set
-+# CONFIG_GAMEPORT is not set
++CONFIG_GAMEPORT=y
++# CONFIG_GAMEPORT_NS558 is not set
++# CONFIG_GAMEPORT_L4 is not set
++# CONFIG_GAMEPORT_EMU10K1 is not set
++# CONFIG_GAMEPORT_FM801 is not set
+
+#
+# Character devices
+CONFIG_VT_CONSOLE=y
+CONFIG_HW_CONSOLE=y
+# CONFIG_VT_HW_CONSOLE_BINDING is not set
-+# CONFIG_SERIAL_NONSTANDARD is not set
++CONFIG_SERIAL_NONSTANDARD=y
++# CONFIG_COMPUTONE is not set
++# CONFIG_ROCKETPORT is not set
++# CONFIG_CYCLADES is not set
++# CONFIG_DIGIEPCA is not set
++# CONFIG_MOXA_INTELLIO is not set
++# CONFIG_MOXA_SMARTIO is not set
++# CONFIG_ISI is not set
++# CONFIG_SYNCLINKMP is not set
++# CONFIG_SYNCLINK_GT is not set
++# CONFIG_N_HDLC is not set
++# CONFIG_SPECIALIX is not set
++# CONFIG_SX is not set
++# CONFIG_RIO is not set
++# CONFIG_STALDRV is not set
++# CONFIG_SGI_SNSC is not set
++# CONFIG_SGI_TIOCX is not set
+
+#
+# Serial drivers
+#
+# Non-8250 serial port support
+#
++# CONFIG_SERIAL_SGI_L1_CONSOLE is not set
++# CONFIG_SERIAL_JSM is not set
+CONFIG_UNIX98_PTYS=y
+CONFIG_LEGACY_PTYS=y
+CONFIG_LEGACY_PTY_COUNT=256
+# Watchdog Cards
+#
+# CONFIG_WATCHDOG is not set
-+CONFIG_HW_RANDOM=y
-+# CONFIG_NVRAM is not set
-+# CONFIG_RTC is not set
-+# CONFIG_GEN_RTC is not set
++# CONFIG_HW_RANDOM is not set
++CONFIG_EFI_RTC=y
+# CONFIG_DTLK is not set
+# CONFIG_R3964 is not set
++# CONFIG_APPLICOM is not set
+
+#
+# Ftape, the floppy tape device driver
+#
-+# CONFIG_MWAVE is not set
-+# CONFIG_PC8736x_GPIO is not set
-+# CONFIG_RAW_DRIVER is not set
-+# CONFIG_HANGCHECK_TIMER is not set
-+
-+#
-+# TPM devices
-+#
-+# CONFIG_TCG_TPM is not set
-+# CONFIG_TELCLOCK is not set
++CONFIG_AGP=y
++# CONFIG_AGP_SIS is not set
++# CONFIG_AGP_VIA is not set
++CONFIG_AGP_I460=y
++# CONFIG_AGP_HP_ZX1 is not set
++# CONFIG_AGP_SGI_TIOCA is not set
++CONFIG_DRM=y
++# CONFIG_DRM_TDFX is not set
++# CONFIG_DRM_R128 is not set
++# CONFIG_DRM_RADEON is not set
++# CONFIG_DRM_MGA is not set
++# CONFIG_DRM_SIS is not set
++# CONFIG_DRM_VIA is not set
++# CONFIG_DRM_SAVAGE is not set
++# CONFIG_RAW_DRIVER is not set
++# CONFIG_HPET is not set
++# CONFIG_HANGCHECK_TIMER is not set
++# CONFIG_MMTIMER is not set
++
++#
++# TPM devices
++#
++# CONFIG_TCG_TPM is not set
++# CONFIG_TELCLOCK is not set
+
+#
+# I2C support
+#
-+# CONFIG_I2C is not set
++CONFIG_I2C=y
++CONFIG_I2C_CHARDEV=y
++
++#
++# I2C Algorithms
++#
++CONFIG_I2C_ALGOBIT=y
++CONFIG_I2C_ALGOPCF=y
++# CONFIG_I2C_ALGOPCA is not set
++
++#
++# I2C Hardware Bus support
++#
++# CONFIG_I2C_ALI1535 is not set
++# CONFIG_I2C_ALI1563 is not set
++# CONFIG_I2C_ALI15X3 is not set
++# CONFIG_I2C_AMD756 is not set
++# CONFIG_I2C_AMD8111 is not set
++# CONFIG_I2C_I801 is not set
++# CONFIG_I2C_I810 is not set
++# CONFIG_I2C_PIIX4 is not set
++# CONFIG_I2C_NFORCE2 is not set
++# CONFIG_I2C_OCORES is not set
++# CONFIG_I2C_PARPORT_LIGHT is not set
++# CONFIG_I2C_PROSAVAGE is not set
++# CONFIG_I2C_SAVAGE4 is not set
++# CONFIG_I2C_SIS5595 is not set
++# CONFIG_I2C_SIS630 is not set
++# CONFIG_I2C_SIS96X is not set
++# CONFIG_I2C_STUB is not set
++# CONFIG_I2C_VIA is not set
++# CONFIG_I2C_VIAPRO is not set
++# CONFIG_I2C_VOODOO3 is not set
++# CONFIG_I2C_PCA_ISA is not set
++
++#
++# Miscellaneous I2C Chip support
++#
++# CONFIG_SENSORS_DS1337 is not set
++# CONFIG_SENSORS_DS1374 is not set
++# CONFIG_SENSORS_EEPROM is not set
++# CONFIG_SENSORS_PCF8574 is not set
++# CONFIG_SENSORS_PCA9539 is not set
++# CONFIG_SENSORS_PCF8591 is not set
++# CONFIG_SENSORS_MAX6875 is not set
++# CONFIG_I2C_DEBUG_CORE is not set
++# CONFIG_I2C_DEBUG_ALGO is not set
++# CONFIG_I2C_DEBUG_BUS is not set
++# CONFIG_I2C_DEBUG_CHIP is not set
+
+#
+# SPI support
+#
+# Hardware Monitoring support
+#
-+# CONFIG_HWMON is not set
++CONFIG_HWMON=y
+# CONFIG_HWMON_VID is not set
++# CONFIG_SENSORS_ABITUGURU is not set
++# CONFIG_SENSORS_ADM1021 is not set
++# CONFIG_SENSORS_ADM1025 is not set
++# CONFIG_SENSORS_ADM1026 is not set
++# CONFIG_SENSORS_ADM1031 is not set
++# CONFIG_SENSORS_ADM9240 is not set
++# CONFIG_SENSORS_ASB100 is not set
++# CONFIG_SENSORS_ATXP1 is not set
++# CONFIG_SENSORS_DS1621 is not set
++# CONFIG_SENSORS_F71805F is not set
++# CONFIG_SENSORS_FSCHER is not set
++# CONFIG_SENSORS_FSCPOS is not set
++# CONFIG_SENSORS_GL518SM is not set
++# CONFIG_SENSORS_GL520SM is not set
++# CONFIG_SENSORS_IT87 is not set
++# CONFIG_SENSORS_LM63 is not set
++# CONFIG_SENSORS_LM75 is not set
++# CONFIG_SENSORS_LM77 is not set
++# CONFIG_SENSORS_LM78 is not set
++# CONFIG_SENSORS_LM80 is not set
++# CONFIG_SENSORS_LM83 is not set
++# CONFIG_SENSORS_LM85 is not set
++# CONFIG_SENSORS_LM87 is not set
++# CONFIG_SENSORS_LM90 is not set
++# CONFIG_SENSORS_LM92 is not set
++# CONFIG_SENSORS_MAX1619 is not set
++# CONFIG_SENSORS_PC87360 is not set
++# CONFIG_SENSORS_SIS5595 is not set
++# CONFIG_SENSORS_SMSC47M1 is not set
++# CONFIG_SENSORS_SMSC47M192 is not set
++# CONFIG_SENSORS_SMSC47B397 is not set
++# CONFIG_SENSORS_VIA686A is not set
++# CONFIG_SENSORS_VT8231 is not set
++# CONFIG_SENSORS_W83781D is not set
++# CONFIG_SENSORS_W83791D is not set
++# CONFIG_SENSORS_W83792D is not set
++# CONFIG_SENSORS_W83L785TS is not set
++# CONFIG_SENSORS_W83627HF is not set
++# CONFIG_SENSORS_W83627EHF is not set
++# CONFIG_HWMON_DEBUG_CHIP is not set
+
+#
+# Misc devices
+#
+# Multimedia devices
+#
-+# CONFIG_VIDEO_DEV is not set
++CONFIG_VIDEO_DEV=y
++CONFIG_VIDEO_V4L1=y
++CONFIG_VIDEO_V4L1_COMPAT=y
++CONFIG_VIDEO_V4L2=y
++
++#
++# Video Capture Adapters
++#
++
++#
++# Video Capture Adapters
++#
++# CONFIG_VIDEO_ADV_DEBUG is not set
++# CONFIG_VIDEO_VIVI is not set
++# CONFIG_VIDEO_BT848 is not set
++# CONFIG_VIDEO_CPIA is not set
++# CONFIG_VIDEO_CPIA2 is not set
++# CONFIG_VIDEO_SAA5246A is not set
++# CONFIG_VIDEO_SAA5249 is not set
++# CONFIG_TUNER_3036 is not set
++# CONFIG_VIDEO_STRADIS is not set
++# CONFIG_VIDEO_ZORAN is not set
++# CONFIG_VIDEO_SAA7134 is not set
++# CONFIG_VIDEO_MXB is not set
++# CONFIG_VIDEO_DPC is not set
++# CONFIG_VIDEO_HEXIUM_ORION is not set
++# CONFIG_VIDEO_HEXIUM_GEMINI is not set
++# CONFIG_VIDEO_CX88 is not set
++
++#
++# Encoders and Decoders
++#
++# CONFIG_VIDEO_MSP3400 is not set
++# CONFIG_VIDEO_CS53L32A is not set
++# CONFIG_VIDEO_TLV320AIC23B is not set
++# CONFIG_VIDEO_WM8775 is not set
++# CONFIG_VIDEO_WM8739 is not set
++# CONFIG_VIDEO_CX2341X is not set
++# CONFIG_VIDEO_CX25840 is not set
++# CONFIG_VIDEO_SAA711X is not set
++# CONFIG_VIDEO_SAA7127 is not set
++# CONFIG_VIDEO_UPD64031A is not set
++# CONFIG_VIDEO_UPD64083 is not set
++
++#
++# V4L USB devices
++#
++# CONFIG_VIDEO_PVRUSB2 is not set
++# CONFIG_VIDEO_EM28XX is not set
++# CONFIG_USB_VICAM is not set
++# CONFIG_USB_IBMCAM is not set
++# CONFIG_USB_KONICAWC is not set
++# CONFIG_USB_QUICKCAM_MESSENGER is not set
++# CONFIG_USB_ET61X251 is not set
++# CONFIG_VIDEO_OVCAMCHIP is not set
++# CONFIG_USB_W9968CF is not set
++# CONFIG_USB_OV511 is not set
++# CONFIG_USB_SE401 is not set
++# CONFIG_USB_SN9C102 is not set
++# CONFIG_USB_STV680 is not set
++# CONFIG_USB_ZC0301 is not set
++# CONFIG_USB_PWC is not set
++
++#
++# Radio Adapters
++#
++# CONFIG_RADIO_GEMTEK_PCI is not set
++# CONFIG_RADIO_MAXIRADIO is not set
++# CONFIG_RADIO_MAESTRO is not set
++# CONFIG_USB_DSBR is not set
+
+#
+# Digital Video Broadcasting Devices
+#
+# CONFIG_DVB is not set
++# CONFIG_USB_DABUSB is not set
+
+#
+# Graphics support
+#
+CONFIG_FIRMWARE_EDID=y
-+# CONFIG_FB is not set
++CONFIG_FB=y
++CONFIG_FB_CFB_FILLRECT=y
++CONFIG_FB_CFB_COPYAREA=y
++CONFIG_FB_CFB_IMAGEBLIT=y
++# CONFIG_FB_MACMODES is not set
++# CONFIG_FB_BACKLIGHT is not set
++CONFIG_FB_MODE_HELPERS=y
++# CONFIG_FB_TILEBLITTING is not set
++# CONFIG_FB_CIRRUS is not set
++# CONFIG_FB_PM2 is not set
++# CONFIG_FB_CYBER2000 is not set
++# CONFIG_FB_ASILIANT is not set
++# CONFIG_FB_IMSTT is not set
++# CONFIG_FB_S1D13XXX is not set
++# CONFIG_FB_NVIDIA is not set
++# CONFIG_FB_RIVA is not set
++# CONFIG_FB_MATROX is not set
++# CONFIG_FB_RADEON is not set
++# CONFIG_FB_ATY128 is not set
++# CONFIG_FB_ATY is not set
++# CONFIG_FB_SAVAGE is not set
++# CONFIG_FB_SIS is not set
++# CONFIG_FB_NEOMAGIC is not set
++# CONFIG_FB_KYRO is not set
++# CONFIG_FB_3DFX is not set
++# CONFIG_FB_VOODOO1 is not set
++# CONFIG_FB_TRIDENT is not set
++# CONFIG_FB_VIRTUAL is not set
+
+#
+# Console display driver support
+CONFIG_VGA_CONSOLE=y
+# CONFIG_VGACON_SOFT_SCROLLBACK is not set
+CONFIG_DUMMY_CONSOLE=y
++CONFIG_FRAMEBUFFER_CONSOLE=y
++# CONFIG_FRAMEBUFFER_CONSOLE_ROTATION is not set
++# CONFIG_FONTS is not set
++CONFIG_FONT_8x8=y
++CONFIG_FONT_8x16=y
++
++#
++# Logo configuration
++#
++CONFIG_LOGO=y
++# CONFIG_LOGO_LINUX_MONO is not set
++# CONFIG_LOGO_LINUX_VGA16 is not set
++CONFIG_LOGO_LINUX_CLUT224=y
+# CONFIG_BACKLIGHT_LCD_SUPPORT is not set
+
+#
+# Sound
+#
-+# CONFIG_SOUND is not set
++CONFIG_SOUND=y
++
++#
++# Advanced Linux Sound Architecture
++#
++CONFIG_SND=y
++CONFIG_SND_TIMER=y
++CONFIG_SND_PCM=y
++CONFIG_SND_HWDEP=y
++CONFIG_SND_RAWMIDI=y
++CONFIG_SND_SEQUENCER=y
++CONFIG_SND_SEQ_DUMMY=y
++CONFIG_SND_OSSEMUL=y
++CONFIG_SND_MIXER_OSS=y
++CONFIG_SND_PCM_OSS=y
++CONFIG_SND_PCM_OSS_PLUGINS=y
++CONFIG_SND_SEQUENCER_OSS=y
++# CONFIG_SND_DYNAMIC_MINORS is not set
++CONFIG_SND_SUPPORT_OLD_API=y
++CONFIG_SND_VERBOSE_PROCFS=y
++# CONFIG_SND_VERBOSE_PRINTK is not set
++# CONFIG_SND_DEBUG is not set
++
++#
++# Generic devices
++#
++CONFIG_SND_MPU401_UART=y
++CONFIG_SND_OPL3_LIB=y
++CONFIG_SND_AC97_CODEC=y
++CONFIG_SND_AC97_BUS=y
++CONFIG_SND_DUMMY=y
++CONFIG_SND_VIRMIDI=y
++# CONFIG_SND_MTPAV is not set
++CONFIG_SND_SERIAL_U16550=y
++CONFIG_SND_MPU401=y
++
++#
++# PCI devices
++#
++# CONFIG_SND_AD1889 is not set
++# CONFIG_SND_ALS300 is not set
++# CONFIG_SND_ALI5451 is not set
++CONFIG_SND_ATIIXP=y
++# CONFIG_SND_ATIIXP_MODEM is not set
++# CONFIG_SND_AU8810 is not set
++# CONFIG_SND_AU8820 is not set
++# CONFIG_SND_AU8830 is not set
++# CONFIG_SND_AZT3328 is not set
++# CONFIG_SND_BT87X is not set
++# CONFIG_SND_CA0106 is not set
++# CONFIG_SND_CMIPCI is not set
++# CONFIG_SND_CS4281 is not set
++# CONFIG_SND_CS46XX is not set
++# CONFIG_SND_DARLA20 is not set
++# CONFIG_SND_GINA20 is not set
++# CONFIG_SND_LAYLA20 is not set
++# CONFIG_SND_DARLA24 is not set
++# CONFIG_SND_GINA24 is not set
++# CONFIG_SND_LAYLA24 is not set
++# CONFIG_SND_MONA is not set
++# CONFIG_SND_MIA is not set
++# CONFIG_SND_ECHO3G is not set
++# CONFIG_SND_INDIGO is not set
++# CONFIG_SND_INDIGOIO is not set
++# CONFIG_SND_INDIGODJ is not set
++# CONFIG_SND_EMU10K1 is not set
++# CONFIG_SND_EMU10K1X is not set
++# CONFIG_SND_ENS1370 is not set
++# CONFIG_SND_ENS1371 is not set
++# CONFIG_SND_ES1938 is not set
++# CONFIG_SND_ES1968 is not set
++CONFIG_SND_FM801=y
++# CONFIG_SND_FM801_TEA575X_BOOL is not set
++# CONFIG_SND_HDA_INTEL is not set
++# CONFIG_SND_HDSP is not set
++# CONFIG_SND_HDSPM is not set
++# CONFIG_SND_ICE1712 is not set
++# CONFIG_SND_ICE1724 is not set
++# CONFIG_SND_INTEL8X0 is not set
++# CONFIG_SND_INTEL8X0M is not set
++# CONFIG_SND_KORG1212 is not set
++# CONFIG_SND_MAESTRO3 is not set
++# CONFIG_SND_MIXART is not set
++# CONFIG_SND_NM256 is not set
++# CONFIG_SND_PCXHR is not set
++# CONFIG_SND_RIPTIDE is not set
++# CONFIG_SND_RME32 is not set
++# CONFIG_SND_RME96 is not set
++# CONFIG_SND_RME9652 is not set
++# CONFIG_SND_SONICVIBES is not set
++# CONFIG_SND_TRIDENT is not set
++# CONFIG_SND_VIA82XX is not set
++# CONFIG_SND_VIA82XX_MODEM is not set
++# CONFIG_SND_VX222 is not set
++# CONFIG_SND_YMFPCI is not set
++
++#
++# USB devices
++#
++# CONFIG_SND_USB_AUDIO is not set
++
++#
++# Open Sound System
++#
++CONFIG_SOUND_PRIME=y
++# CONFIG_OSS_OBSOLETE_DRIVER is not set
++# CONFIG_SOUND_BT878 is not set
++# CONFIG_SOUND_ES1371 is not set
++# CONFIG_SOUND_ICH is not set
++# CONFIG_SOUND_TRIDENT is not set
++# CONFIG_SOUND_MSNDCLAS is not set
++# CONFIG_SOUND_MSNDPIN is not set
++# CONFIG_SOUND_VIA82CXXX is not set
++# CONFIG_SOUND_TVMIXER is not set
+
+#
+# USB support
+#
-+# CONFIG_USB_ARCH_HAS_HCD is not set
-+# CONFIG_USB_ARCH_HAS_OHCI is not set
-+# CONFIG_USB_ARCH_HAS_EHCI is not set
++CONFIG_USB_ARCH_HAS_HCD=y
++CONFIG_USB_ARCH_HAS_OHCI=y
++CONFIG_USB_ARCH_HAS_EHCI=y
++CONFIG_USB=y
++# CONFIG_USB_DEBUG is not set
++
++#
++# Miscellaneous USB options
++#
++CONFIG_USB_DEVICEFS=y
++CONFIG_USB_BANDWIDTH=y
++# CONFIG_USB_DYNAMIC_MINORS is not set
++# CONFIG_USB_SUSPEND is not set
++# CONFIG_USB_OTG is not set
++
++#
++# USB Host Controller Drivers
++#
++CONFIG_USB_EHCI_HCD=y
++# CONFIG_USB_EHCI_SPLIT_ISO is not set
++# CONFIG_USB_EHCI_ROOT_HUB_TT is not set
++# CONFIG_USB_EHCI_TT_NEWSCHED is not set
++# CONFIG_USB_ISP116X_HCD is not set
++CONFIG_USB_OHCI_HCD=y
++# CONFIG_USB_OHCI_BIG_ENDIAN is not set
++CONFIG_USB_OHCI_LITTLE_ENDIAN=y
++CONFIG_USB_UHCI_HCD=y
++# CONFIG_USB_SL811_HCD is not set
++
++#
++# USB Device Class drivers
++#
++# CONFIG_USB_ACM is not set
++# CONFIG_USB_PRINTER is not set
+
+#
+# NOTE: USB_STORAGE enables SCSI, and 'SCSI disk support'
+#
+
+#
++# may also be needed; see USB_STORAGE Help for more information
++#
++CONFIG_USB_STORAGE=y
++# CONFIG_USB_STORAGE_DEBUG is not set
++# CONFIG_USB_STORAGE_DATAFAB is not set
++# CONFIG_USB_STORAGE_FREECOM is not set
++# CONFIG_USB_STORAGE_ISD200 is not set
++# CONFIG_USB_STORAGE_DPCM is not set
++# CONFIG_USB_STORAGE_USBAT is not set
++# CONFIG_USB_STORAGE_SDDR09 is not set
++# CONFIG_USB_STORAGE_SDDR55 is not set
++# CONFIG_USB_STORAGE_JUMPSHOT is not set
++# CONFIG_USB_STORAGE_ALAUDA is not set
++# CONFIG_USB_LIBUSUAL is not set
++
++#
++# USB Input Devices
++#
++CONFIG_USB_HID=y
++CONFIG_USB_HIDINPUT=y
++# CONFIG_USB_HIDINPUT_POWERBOOK is not set
++# CONFIG_HID_FF is not set
++CONFIG_USB_HIDDEV=y
++# CONFIG_USB_AIPTEK is not set
++# CONFIG_USB_WACOM is not set
++# CONFIG_USB_ACECAD is not set
++# CONFIG_USB_KBTAB is not set
++# CONFIG_USB_POWERMATE is not set
++# CONFIG_USB_TOUCHSCREEN is not set
++# CONFIG_USB_YEALINK is not set
++# CONFIG_USB_XPAD is not set
++# CONFIG_USB_ATI_REMOTE is not set
++# CONFIG_USB_ATI_REMOTE2 is not set
++# CONFIG_USB_KEYSPAN_REMOTE is not set
++# CONFIG_USB_APPLETOUCH is not set
++
++#
++# USB Imaging devices
++#
++# CONFIG_USB_MDC800 is not set
++# CONFIG_USB_MICROTEK is not set
++
++#
++# USB Network Adapters
++#
++# CONFIG_USB_CATC is not set
++# CONFIG_USB_KAWETH is not set
++# CONFIG_USB_PEGASUS is not set
++# CONFIG_USB_RTL8150 is not set
++# CONFIG_USB_USBNET is not set
++CONFIG_USB_MON=y
++
++#
++# USB port drivers
++#
++
++#
++# USB Serial Converter support
++#
++# CONFIG_USB_SERIAL is not set
++
++#
++# USB Miscellaneous drivers
++#
++# CONFIG_USB_EMI62 is not set
++# CONFIG_USB_EMI26 is not set
++# CONFIG_USB_AUERSWALD is not set
++# CONFIG_USB_RIO500 is not set
++# CONFIG_USB_LEGOTOWER is not set
++# CONFIG_USB_LCD is not set
++# CONFIG_USB_LED is not set
++# CONFIG_USB_CYPRESS_CY7C63 is not set
++# CONFIG_USB_CYTHERM is not set
++# CONFIG_USB_PHIDGETKIT is not set
++# CONFIG_USB_PHIDGETSERVO is not set
++# CONFIG_USB_IDMOUSE is not set
++# CONFIG_USB_APPLEDISPLAY is not set
++# CONFIG_USB_SISUSBVGA is not set
++# CONFIG_USB_LD is not set
++# CONFIG_USB_TEST is not set
++
++#
++# USB DSL modem support
++#
++
++#
+# USB Gadget Support
+#
+# CONFIG_USB_GADGET is not set
+#
+# InfiniBand support
+#
++# CONFIG_INFINIBAND is not set
+
+#
+# EDAC - error detection and reporting (RAS) (EXPERIMENTAL)
+#
-+# CONFIG_EDAC is not set
+
+#
+# Real Time Clock
+#
+
+#
-+# Firmware Drivers
-+#
-+# CONFIG_EDD is not set
-+# CONFIG_DELL_RBU is not set
-+# CONFIG_DCDBAS is not set
-+
-+#
+# File systems
+#
+CONFIG_EXT2_FS=y
+# CONFIG_EXT2_FS_XIP is not set
+CONFIG_EXT3_FS=y
+CONFIG_EXT3_FS_XATTR=y
-+# CONFIG_EXT3_FS_POSIX_ACL is not set
-+# CONFIG_EXT3_FS_SECURITY is not set
++CONFIG_EXT3_FS_POSIX_ACL=y
++CONFIG_EXT3_FS_SECURITY=y
+CONFIG_JBD=y
+# CONFIG_JBD_DEBUG is not set
+CONFIG_FS_MBCACHE=y
+CONFIG_REISERFS_FS=y
+# CONFIG_REISERFS_CHECK is not set
-+CONFIG_REISERFS_PROC_INFO=y
++# CONFIG_REISERFS_PROC_INFO is not set
+CONFIG_REISERFS_FS_XATTR=y
+CONFIG_REISERFS_FS_POSIX_ACL=y
+CONFIG_REISERFS_FS_SECURITY=y
-+CONFIG_JFS_FS=m
-+CONFIG_JFS_POSIX_ACL=y
-+# CONFIG_JFS_SECURITY is not set
-+# CONFIG_JFS_DEBUG is not set
-+# CONFIG_JFS_STATISTICS is not set
++# CONFIG_JFS_FS is not set
+CONFIG_FS_POSIX_ACL=y
-+CONFIG_XFS_FS=m
++CONFIG_XFS_FS=y
+# CONFIG_XFS_QUOTA is not set
-+CONFIG_XFS_SECURITY=y
-+CONFIG_XFS_POSIX_ACL=y
++# CONFIG_XFS_SECURITY is not set
++# CONFIG_XFS_POSIX_ACL is not set
+# CONFIG_XFS_RT is not set
+# CONFIG_OCFS2_FS is not set
-+CONFIG_MINIX_FS=m
-+CONFIG_ROMFS_FS=m
++# CONFIG_MINIX_FS is not set
++# CONFIG_ROMFS_FS is not set
+CONFIG_INOTIFY=y
+CONFIG_INOTIFY_USER=y
-+CONFIG_QUOTA=y
-+# CONFIG_QFMT_V1 is not set
-+CONFIG_QFMT_V2=y
-+CONFIG_QUOTACTL=y
++# CONFIG_QUOTA is not set
+CONFIG_DNOTIFY=y
-+CONFIG_AUTOFS_FS=m
-+CONFIG_AUTOFS4_FS=m
++CONFIG_AUTOFS_FS=y
++CONFIG_AUTOFS4_FS=y
+# CONFIG_FUSE_FS is not set
+
+#
+#
+CONFIG_ISO9660_FS=y
+CONFIG_JOLIET=y
-+CONFIG_ZISOFS=y
-+CONFIG_ZISOFS_FS=y
-+CONFIG_UDF_FS=m
++# CONFIG_ZISOFS is not set
++CONFIG_UDF_FS=y
+CONFIG_UDF_NLS=y
+
+#
+# DOS/FAT/NT Filesystems
+#
-+CONFIG_FAT_FS=m
-+CONFIG_MSDOS_FS=m
-+CONFIG_VFAT_FS=m
++CONFIG_FAT_FS=y
++CONFIG_MSDOS_FS=y
++CONFIG_VFAT_FS=y
+CONFIG_FAT_DEFAULT_CODEPAGE=437
-+CONFIG_FAT_DEFAULT_IOCHARSET="ascii"
++CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1"
+# CONFIG_NTFS_FS is not set
+
+#
+#
+CONFIG_PROC_FS=y
+CONFIG_PROC_KCORE=y
++CONFIG_PROC_IOMEM_MACHINE=y
+CONFIG_SYSFS=y
+CONFIG_TMPFS=y
+# CONFIG_HUGETLB_PAGE is not set
+# Miscellaneous filesystems
+#
+# CONFIG_ADFS_FS is not set
-+CONFIG_AFFS_FS=m
-+CONFIG_HFS_FS=m
-+CONFIG_HFSPLUS_FS=m
-+CONFIG_BEFS_FS=m
-+# CONFIG_BEFS_DEBUG is not set
-+CONFIG_BFS_FS=m
-+CONFIG_EFS_FS=m
-+CONFIG_CRAMFS=y
-+CONFIG_VXFS_FS=m
++# CONFIG_AFFS_FS is not set
++# CONFIG_HFS_FS is not set
++# CONFIG_HFSPLUS_FS is not set
++# CONFIG_BEFS_FS is not set
++# CONFIG_BFS_FS is not set
++# CONFIG_EFS_FS is not set
++# CONFIG_CRAMFS is not set
++# CONFIG_VXFS_FS is not set
+# CONFIG_HPFS_FS is not set
-+CONFIG_QNX4FS_FS=m
-+CONFIG_SYSV_FS=m
-+CONFIG_UFS_FS=m
-+# CONFIG_UFS_FS_WRITE is not set
-+# CONFIG_UFS_DEBUG is not set
++# CONFIG_QNX4FS_FS is not set
++# CONFIG_SYSV_FS is not set
++# CONFIG_UFS_FS is not set
+
+#
+# Network File Systems
+#
-+CONFIG_NFS_FS=m
++CONFIG_NFS_FS=y
+CONFIG_NFS_V3=y
+# CONFIG_NFS_V3_ACL is not set
+CONFIG_NFS_V4=y
+CONFIG_NFS_DIRECTIO=y
-+CONFIG_NFSD=m
++CONFIG_NFSD=y
+CONFIG_NFSD_V3=y
+# CONFIG_NFSD_V3_ACL is not set
+CONFIG_NFSD_V4=y
+CONFIG_NFSD_TCP=y
-+CONFIG_LOCKD=m
++CONFIG_ROOT_NFS=y
++CONFIG_LOCKD=y
+CONFIG_LOCKD_V4=y
-+CONFIG_EXPORTFS=m
++CONFIG_EXPORTFS=y
+CONFIG_NFS_COMMON=y
-+CONFIG_SUNRPC=m
-+CONFIG_SUNRPC_GSS=m
-+CONFIG_RPCSEC_GSS_KRB5=m
-+CONFIG_RPCSEC_GSS_SPKM3=m
-+CONFIG_SMB_FS=m
-+# CONFIG_SMB_NLS_DEFAULT is not set
-+CONFIG_CIFS=m
++CONFIG_SUNRPC=y
++CONFIG_SUNRPC_GSS=y
++CONFIG_RPCSEC_GSS_KRB5=y
++# CONFIG_RPCSEC_GSS_SPKM3 is not set
++CONFIG_SMB_FS=y
++CONFIG_SMB_NLS_DEFAULT=y
++CONFIG_SMB_NLS_REMOTE="cp437"
++CONFIG_CIFS=y
+# CONFIG_CIFS_STATS is not set
+# CONFIG_CIFS_WEAK_PW_HASH is not set
-+CONFIG_CIFS_XATTR=y
-+CONFIG_CIFS_POSIX=y
++# CONFIG_CIFS_XATTR is not set
+# CONFIG_CIFS_DEBUG2 is not set
+# CONFIG_CIFS_EXPERIMENTAL is not set
-+CONFIG_NCP_FS=m
-+CONFIG_NCPFS_PACKET_SIGNING=y
-+CONFIG_NCPFS_IOCTL_LOCKING=y
-+CONFIG_NCPFS_STRONG=y
-+CONFIG_NCPFS_NFS_NS=y
-+CONFIG_NCPFS_OS2_NS=y
-+CONFIG_NCPFS_SMALLDOS=y
-+CONFIG_NCPFS_NLS=y
-+CONFIG_NCPFS_EXTRAS=y
++# CONFIG_NCP_FS is not set
+# CONFIG_CODA_FS is not set
+# CONFIG_AFS_FS is not set
+# CONFIG_9P_FS is not set
+#
+CONFIG_PARTITION_ADVANCED=y
+# CONFIG_ACORN_PARTITION is not set
-+CONFIG_OSF_PARTITION=y
++# CONFIG_OSF_PARTITION is not set
+# CONFIG_AMIGA_PARTITION is not set
+# CONFIG_ATARI_PARTITION is not set
-+CONFIG_MAC_PARTITION=y
++# CONFIG_MAC_PARTITION is not set
+CONFIG_MSDOS_PARTITION=y
-+CONFIG_BSD_DISKLABEL=y
-+CONFIG_MINIX_SUBPARTITION=y
-+CONFIG_SOLARIS_X86_PARTITION=y
-+CONFIG_UNIXWARE_DISKLABEL=y
++# CONFIG_BSD_DISKLABEL is not set
++# CONFIG_MINIX_SUBPARTITION is not set
++# CONFIG_SOLARIS_X86_PARTITION is not set
++# CONFIG_UNIXWARE_DISKLABEL is not set
+# CONFIG_LDM_PARTITION is not set
+CONFIG_SGI_PARTITION=y
+# CONFIG_ULTRIX_PARTITION is not set
-+CONFIG_SUN_PARTITION=y
++# CONFIG_SUN_PARTITION is not set
+# CONFIG_KARMA_PARTITION is not set
+CONFIG_EFI_PARTITION=y
+
+# Native Language Support
+#
+CONFIG_NLS=y
-+CONFIG_NLS_DEFAULT="utf8"
++CONFIG_NLS_DEFAULT="iso8859-1"
+CONFIG_NLS_CODEPAGE_437=y
-+CONFIG_NLS_CODEPAGE_737=m
-+CONFIG_NLS_CODEPAGE_775=m
-+CONFIG_NLS_CODEPAGE_850=m
-+CONFIG_NLS_CODEPAGE_852=m
-+CONFIG_NLS_CODEPAGE_855=m
-+CONFIG_NLS_CODEPAGE_857=m
-+CONFIG_NLS_CODEPAGE_860=m
-+CONFIG_NLS_CODEPAGE_861=m
-+CONFIG_NLS_CODEPAGE_862=m
-+CONFIG_NLS_CODEPAGE_863=m
-+CONFIG_NLS_CODEPAGE_864=m
-+CONFIG_NLS_CODEPAGE_865=m
-+CONFIG_NLS_CODEPAGE_866=m
-+CONFIG_NLS_CODEPAGE_869=m
-+CONFIG_NLS_CODEPAGE_936=m
-+CONFIG_NLS_CODEPAGE_950=m
-+CONFIG_NLS_CODEPAGE_932=m
-+CONFIG_NLS_CODEPAGE_949=m
-+CONFIG_NLS_CODEPAGE_874=m
-+CONFIG_NLS_ISO8859_8=m
-+CONFIG_NLS_CODEPAGE_1250=m
-+CONFIG_NLS_CODEPAGE_1251=m
-+CONFIG_NLS_ASCII=y
-+CONFIG_NLS_ISO8859_1=m
-+CONFIG_NLS_ISO8859_2=m
-+CONFIG_NLS_ISO8859_3=m
-+CONFIG_NLS_ISO8859_4=m
-+CONFIG_NLS_ISO8859_5=m
-+CONFIG_NLS_ISO8859_6=m
-+CONFIG_NLS_ISO8859_7=m
-+CONFIG_NLS_ISO8859_9=m
-+CONFIG_NLS_ISO8859_13=m
-+CONFIG_NLS_ISO8859_14=m
-+CONFIG_NLS_ISO8859_15=m
-+CONFIG_NLS_KOI8_R=m
-+CONFIG_NLS_KOI8_U=m
-+CONFIG_NLS_UTF8=m
++CONFIG_NLS_CODEPAGE_737=y
++CONFIG_NLS_CODEPAGE_775=y
++CONFIG_NLS_CODEPAGE_850=y
++CONFIG_NLS_CODEPAGE_852=y
++CONFIG_NLS_CODEPAGE_855=y
++CONFIG_NLS_CODEPAGE_857=y
++CONFIG_NLS_CODEPAGE_860=y
++CONFIG_NLS_CODEPAGE_861=y
++CONFIG_NLS_CODEPAGE_862=y
++CONFIG_NLS_CODEPAGE_863=y
++CONFIG_NLS_CODEPAGE_864=y
++CONFIG_NLS_CODEPAGE_865=y
++CONFIG_NLS_CODEPAGE_866=y
++CONFIG_NLS_CODEPAGE_869=y
++CONFIG_NLS_CODEPAGE_936=y
++CONFIG_NLS_CODEPAGE_950=y
++CONFIG_NLS_CODEPAGE_932=y
++CONFIG_NLS_CODEPAGE_949=y
++CONFIG_NLS_CODEPAGE_874=y
++CONFIG_NLS_ISO8859_8=y
++# CONFIG_NLS_CODEPAGE_1250 is not set
++CONFIG_NLS_CODEPAGE_1251=y
++# CONFIG_NLS_ASCII is not set
++CONFIG_NLS_ISO8859_1=y
++CONFIG_NLS_ISO8859_2=y
++CONFIG_NLS_ISO8859_3=y
++CONFIG_NLS_ISO8859_4=y
++CONFIG_NLS_ISO8859_5=y
++CONFIG_NLS_ISO8859_6=y
++CONFIG_NLS_ISO8859_7=y
++CONFIG_NLS_ISO8859_9=y
++CONFIG_NLS_ISO8859_13=y
++CONFIG_NLS_ISO8859_14=y
++CONFIG_NLS_ISO8859_15=y
++CONFIG_NLS_KOI8_R=y
++CONFIG_NLS_KOI8_U=y
++CONFIG_NLS_UTF8=y
++
++#
++# Library routines
++#
++# CONFIG_CRC_CCITT is not set
++# CONFIG_CRC16 is not set
++CONFIG_CRC32=y
++# CONFIG_LIBCRC32C is not set
++CONFIG_PLIST=y
++CONFIG_GENERIC_HARDIRQS=y
++CONFIG_GENERIC_IRQ_PROBE=y
++CONFIG_GENERIC_PENDING_IRQ=y
++CONFIG_IRQ_PER_CPU=y
++
++#
++# HP Simulator drivers
++#
++# CONFIG_HP_SIMETH is not set
++# CONFIG_HP_SIMSERIAL is not set
++# CONFIG_HP_SIMSCSI is not set
+
+#
+# Instrumentation Support
+#
+# Kernel hacking
+#
-+CONFIG_TRACE_IRQFLAGS_SUPPORT=y
+# CONFIG_PRINTK_TIME is not set
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_UNUSED_SYMBOLS=y
+CONFIG_DEBUG_KERNEL=y
-+CONFIG_LOG_BUF_SHIFT=15
++CONFIG_LOG_BUF_SHIFT=20
+CONFIG_DETECT_SOFTLOCKUP=y
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_DEBUG_SLAB is not set
+# CONFIG_DEBUG_RT_MUTEXES is not set
+# CONFIG_RT_MUTEX_TESTER is not set
+# CONFIG_DEBUG_SPINLOCK is not set
-+# CONFIG_DEBUG_MUTEXES is not set
++CONFIG_DEBUG_MUTEXES=y
+# CONFIG_DEBUG_RWSEMS is not set
-+# CONFIG_DEBUG_LOCK_ALLOC is not set
-+# CONFIG_PROVE_LOCKING is not set
+# CONFIG_DEBUG_SPINLOCK_SLEEP is not set
+# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
+# CONFIG_DEBUG_KOBJECT is not set
-+CONFIG_DEBUG_INFO=y
++# CONFIG_DEBUG_INFO is not set
+# CONFIG_DEBUG_FS is not set
+# CONFIG_DEBUG_VM is not set
-+CONFIG_FRAME_POINTER=y
-+# CONFIG_UNWIND_INFO is not set
+CONFIG_FORCED_INLINING=y
+# CONFIG_RCU_TORTURE_TEST is not set
-+# CONFIG_DEBUG_RODATA is not set
-+# CONFIG_DEBUG_STACKOVERFLOW is not set
-+# CONFIG_DEBUG_STACK_USAGE is not set
++CONFIG_IA64_GRANULE_16MB=y
++# CONFIG_IA64_GRANULE_64MB is not set
++CONFIG_IA64_PRINT_HAZARDS=y
++# CONFIG_DISABLE_VHPT is not set
++# CONFIG_IA64_DEBUG_CMPXCHG is not set
++# CONFIG_IA64_DEBUG_IRQ is not set
+
+#
+# Security options
+# Cryptographic options
+#
+CONFIG_CRYPTO=y
-+CONFIG_CRYPTO_HMAC=y
-+CONFIG_CRYPTO_NULL=m
-+CONFIG_CRYPTO_MD4=m
++# CONFIG_CRYPTO_HMAC is not set
++# CONFIG_CRYPTO_NULL is not set
++# CONFIG_CRYPTO_MD4 is not set
+CONFIG_CRYPTO_MD5=y
-+CONFIG_CRYPTO_SHA1=y
-+CONFIG_CRYPTO_SHA256=m
-+CONFIG_CRYPTO_SHA512=m
-+CONFIG_CRYPTO_WP512=m
++# CONFIG_CRYPTO_SHA1 is not set
++# CONFIG_CRYPTO_SHA256 is not set
++# CONFIG_CRYPTO_SHA512 is not set
++# CONFIG_CRYPTO_WP512 is not set
+# CONFIG_CRYPTO_TGR192 is not set
-+CONFIG_CRYPTO_DES=m
-+CONFIG_CRYPTO_BLOWFISH=m
-+CONFIG_CRYPTO_TWOFISH=m
-+CONFIG_CRYPTO_SERPENT=m
++CONFIG_CRYPTO_DES=y
++# CONFIG_CRYPTO_BLOWFISH is not set
++# CONFIG_CRYPTO_TWOFISH is not set
++# CONFIG_CRYPTO_SERPENT is not set
+# CONFIG_CRYPTO_AES is not set
-+# CONFIG_CRYPTO_AES_X86_64 is not set
-+CONFIG_CRYPTO_CAST5=m
-+CONFIG_CRYPTO_CAST6=m
-+CONFIG_CRYPTO_TEA=m
-+CONFIG_CRYPTO_ARC4=m
-+CONFIG_CRYPTO_KHAZAD=m
++# CONFIG_CRYPTO_CAST5 is not set
++# CONFIG_CRYPTO_CAST6 is not set
++# CONFIG_CRYPTO_TEA is not set
++# CONFIG_CRYPTO_ARC4 is not set
++# CONFIG_CRYPTO_KHAZAD is not set
+# CONFIG_CRYPTO_ANUBIS is not set
-+CONFIG_CRYPTO_DEFLATE=m
-+CONFIG_CRYPTO_MICHAEL_MIC=m
-+CONFIG_CRYPTO_CRC32C=m
++# CONFIG_CRYPTO_DEFLATE is not set
++# CONFIG_CRYPTO_MICHAEL_MIC is not set
++# CONFIG_CRYPTO_CRC32C is not set
+# CONFIG_CRYPTO_TEST is not set
+
+#
+# Hardware crypto devices
+#
-+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++# CONFIG_XEN_SMPBOOT is not set
++# CONFIG_XEN_DEVMEM is not set
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+#
-+# CONFIG_XEN_PRIVILEGED_GUEST is not set
-+CONFIG_XEN_UNPRIVILEGED_GUEST=y
++CONFIG_XEN_PRIVILEGED_GUEST=y
++# CONFIG_XEN_UNPRIVILEGED_GUEST is not set
+CONFIG_XEN_PRIVCMD=y
+CONFIG_XEN_XENBUS_DEV=y
-+# CONFIG_XEN_BACKEND is not set
++CONFIG_XEN_BACKEND=y
++CONFIG_XEN_BLKDEV_BACKEND=y
++CONFIG_XEN_BLKDEV_TAP=y
++CONFIG_XEN_NETDEV_BACKEND=y
++# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
++# CONFIG_XEN_NETDEV_LOOPBACK is not set
++CONFIG_XEN_PCIDEV_BACKEND=y
++# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
++# CONFIG_XEN_PCIDEV_BACKEND_PASS is not set
++# CONFIG_XEN_PCIDEV_BACKEND_SLOT is not set
++CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER=y
++# CONFIG_XEN_PCIDEV_BE_DEBUG is not set
++CONFIG_XEN_TPMDEV_BACKEND=m
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
-+CONFIG_XEN_SCRUB_PAGES=y
++CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
++CONFIG_XEN_FRAMEBUFFER=y
++CONFIG_XEN_KEYBOARD=y
++# CONFIG_XEN_SCRUB_PAGES is not set
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_XEN_SYSFS=y
+CONFIG_XEN_COMPAT_030002_AND_LATER=y
+CONFIG_XEN_COMPAT=0x030002
+CONFIG_HAVE_IRQ_IGNORE_UNHANDLED=y
+CONFIG_NO_IDLE_HZ=y
-+CONFIG_XEN_SMPBOOT=y
+CONFIG_XEN_BALLOON=y
-+CONFIG_XEN_DEVMEM=y
-+# CONFIG_XEN_GRANT_DEV is not set
-+
-+#
-+# Library routines
-+#
-+CONFIG_CRC_CCITT=m
-+# CONFIG_CRC16 is not set
-+CONFIG_CRC32=y
-+CONFIG_LIBCRC32C=m
-+CONFIG_ZLIB_INFLATE=y
-+CONFIG_ZLIB_DEFLATE=m
-+CONFIG_TEXTSEARCH=y
-+CONFIG_TEXTSEARCH_KMP=m
-+CONFIG_PLIST=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen_x86_32 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_32
---- linux-2.6.18/buildconfigs/linux-defconfig_xen_x86_32 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_32 2007-12-23 11:15:00.746186534 +0100
-@@ -0,0 +1,3318 @@
++CONFIG_XEN_XENCOMM=y
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen_x86_32 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_32 2008-05-19 00:33:19.717282661 +0300
+@@ -0,0 +1,3324 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
+CONFIG_XEN_PCIDEV_FRONTEND=y
+# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
+# CONFIG_PCIEPORTBUS is not set
++# CONFIG_PCI_MSI is not set
+# CONFIG_PCI_DEBUG is not set
+CONFIG_ISA_DMA_API=y
+CONFIG_SCx200=m
+CONFIG_I2O=m
+CONFIG_I2O_LCT_NOTIFY_ON_CHANGES=y
+CONFIG_I2O_EXT_ADAPTEC=y
++CONFIG_I2O_EXT_ADAPTEC_DMA64=y
+CONFIG_I2O_CONFIG=m
+CONFIG_I2O_CONFIG_OLD_IOCTL=y
+CONFIG_I2O_BUS=m
+CONFIG_S2IO=m
+CONFIG_S2IO_NAPI=y
+CONFIG_MYRI10GE=m
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+#
+# CONFIG_CRYPTO_DEV_PADLOCK is not set
+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+CONFIG_XEN_BLKDEV_TAP=y
+CONFIG_XEN_NETDEV_BACKEND=y
+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
+# CONFIG_XEN_NETDEV_LOOPBACK is not set
+CONFIG_XEN_PCIDEV_BACKEND=m
+CONFIG_XEN_PCIDEV_BACKEND_VPCI=y
+CONFIG_XEN_TPMDEV_BACKEND=m
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
-+CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
++CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_KEYBOARD=y
+CONFIG_XEN_SCRUB_PAGES=y
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_X86_NO_TSS=y
+CONFIG_X86_NO_IDT=y
+CONFIG_KTIME_SCALAR=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/buildconfigs/linux-defconfig_xen_x86_64 linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_64
---- linux-2.6.18/buildconfigs/linux-defconfig_xen_x86_64 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_64 2007-12-23 11:15:00.746186534 +0100
-@@ -0,0 +1,3139 @@
+--- linux-2.6.18.8/buildconfigs/linux-defconfig_xen_x86_64 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/buildconfigs/linux-defconfig_xen_x86_64 2008-05-19 00:33:19.769285658 +0300
+@@ -0,0 +1,3144 @@
+#
+# Automatically generated make config: don't edit
+# Linux kernel version: 2.6.18.8
-+# Tue Oct 16 09:32:25 2007
++# Mon Feb 18 10:38:24 2008
+#
+CONFIG_X86_64=y
+CONFIG_64BIT=y
+CONFIG_XEN_PCIDEV_FRONTEND=y
+# CONFIG_XEN_PCIDEV_FE_DEBUG is not set
+# CONFIG_PCIEPORTBUS is not set
++# CONFIG_PCI_MSI is not set
+# CONFIG_PCI_DEBUG is not set
+
+#
+CONFIG_S2IO=m
+CONFIG_S2IO_NAPI=y
+CONFIG_MYRI10GE=m
++# CONFIG_SFC is not set
+
+#
+# Token Ring devices
+# Hardware crypto devices
+#
+CONFIG_XEN=y
-+CONFIG_XEN_INTERFACE_VERSION=0x00030206
++CONFIG_XEN_INTERFACE_VERSION=0x00030207
+
+#
+# XEN
+CONFIG_XEN_BLKDEV_TAP=y
+CONFIG_XEN_NETDEV_BACKEND=y
+# CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL is not set
++# CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND is not set
+# CONFIG_XEN_NETDEV_LOOPBACK is not set
+CONFIG_XEN_PCIDEV_BACKEND=m
+# CONFIG_XEN_PCIDEV_BACKEND_VPCI is not set
+CONFIG_XEN_TPMDEV_BACKEND=m
+CONFIG_XEN_BLKDEV_FRONTEND=y
+CONFIG_XEN_NETDEV_FRONTEND=y
-+CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_GRANT_DEV=y
++# CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND is not set
++CONFIG_XEN_FRAMEBUFFER=y
+CONFIG_XEN_KEYBOARD=y
+CONFIG_XEN_SCRUB_PAGES=y
+CONFIG_XEN_DISABLE_SERIAL=y
+CONFIG_TEXTSEARCH_BM=m
+CONFIG_TEXTSEARCH_FSM=m
+CONFIG_PLIST=y
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/acpi/hardware/hwsleep.c linux-2.6.18-xen.hg/drivers/acpi/hardware/hwsleep.c
---- linux-2.6.18/drivers/acpi/hardware/hwsleep.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/acpi/hardware/hwsleep.c 2007-12-23 11:15:01.949583064 +0100
-@@ -327,6 +327,7 @@
-
- ACPI_FLUSH_CPU_CACHE();
-
-+#ifndef CONFIG_ACPI_PV_SLEEP
- status = acpi_hw_register_write(ACPI_MTX_DO_NOT_LOCK,
- ACPI_REGISTER_PM1A_CONTROL,
- PM1Acontrol);
-@@ -378,6 +379,15 @@
- } while (!in_value);
-
- return_ACPI_STATUS(AE_OK);
-+#else
-+ /* PV ACPI just need check hypercall return value */
-+ status = acpi_notify_hypervisor_state(sleep_state,
-+ PM1Acontrol, PM1Bcontrol);
-+ if (ACPI_FAILURE(status))
-+ return_ACPI_STATUS(status);
-+ else
-+ return_ACPI_STATUS(AE_OK);
-+#endif
- }
-
- ACPI_EXPORT_SYMBOL(acpi_enter_sleep_state)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/acpi/Kconfig linux-2.6.18-xen.hg/drivers/acpi/Kconfig
---- linux-2.6.18/drivers/acpi/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/acpi/Kconfig 2007-12-23 11:15:01.496225920 +0100
+--- linux-2.6.18.8/drivers/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/Makefile 2008-05-19 00:33:19.861290962 +0300
+@@ -31,6 +31,7 @@
+ obj-$(CONFIG_NUBUS) += nubus/
+ obj-$(CONFIG_ATM) += atm/
+ obj-$(CONFIG_PPC_PMAC) += macintosh/
++obj-$(CONFIG_XEN) += xen/
+ obj-$(CONFIG_IDE) += ide/
+ obj-$(CONFIG_FC4) += fc4/
+ obj-$(CONFIG_SCSI) += scsi/
+--- linux-2.6.18.8/drivers/acpi/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/Kconfig 2008-05-19 00:33:19.877291884 +0300
@@ -305,6 +305,7 @@
config X86_PM_TIMER
bool "Power Management Timer Support" if EMBEDDED
default y
help
The Power Management Timer is available on all ACPI-capable,
-@@ -362,6 +363,10 @@
+@@ -362,6 +363,16 @@
A "Smart Battery" is quite old and quite rare compared
to today's ACPI "Control Method" battery.
+config ACPI_PV_SLEEP
+ bool
+ depends on X86 && XEN && ACPI_SLEEP
++ default y
++
++config PROCESSOR_EXTERNAL_CONTROL
++ bool
++ depends on X86 && XEN
++ select ACPI_PROCESSOR
+ default y
endif # ACPI
endmenu
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/acpi/sleep/main.c linux-2.6.18-xen.hg/drivers/acpi/sleep/main.c
---- linux-2.6.18/drivers/acpi/sleep/main.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/acpi/sleep/main.c 2007-12-23 11:15:01.989585156 +0100
+--- linux-2.6.18.8/drivers/acpi/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/Makefile 2008-05-19 00:33:19.877291884 +0300
+@@ -34,6 +34,9 @@
+ ifdef CONFIG_CPU_FREQ
+ processor-objs += processor_perflib.o
+ endif
++ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++processor-objs += processor_extcntl.o
++endif
+
+ obj-y += sleep/
+ obj-y += bus.o glue.o
+--- linux-2.6.18.8/drivers/acpi/dispatcher/dsobject.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/dispatcher/dsobject.c 2008-05-19 00:33:19.893292806 +0300
+@@ -137,6 +137,71 @@
+ return_ACPI_STATUS(status);
+ }
+ }
++
++ /* Special object resolution for elements of a package */
++
++ if ((op->common.parent->common.aml_opcode == AML_PACKAGE_OP) ||
++ (op->common.parent->common.aml_opcode ==
++ AML_VAR_PACKAGE_OP)) {
++ /*
++ * Attempt to resolve the node to a value before we insert it into
++ * the package. If this is a reference to a common data type,
++ * resolve it immediately. According to the ACPI spec, package
++ * elements can only be "data objects" or method references.
++ * Attempt to resolve to an Integer, Buffer, String or Package.
++ * If cannot, return the named reference (for things like Devices,
++ * Methods, etc.) Buffer Fields and Fields will resolve to simple
++ * objects (int/buf/str/pkg).
++ *
++ * NOTE: References to things like Devices, Methods, Mutexes, etc.
++ * will remain as named references. This behavior is not described
++ * in the ACPI spec, but it appears to be an oversight.
++ */
++ obj_desc = (union acpi_operand_object *)op->common.node;
++
++ status =
++ acpi_ex_resolve_node_to_value(ACPI_CAST_INDIRECT_PTR
++ (struct
++ acpi_namespace_node,
++ &obj_desc),
++ walk_state);
++ if (ACPI_FAILURE(status)) {
++ return_ACPI_STATUS(status);
++ }
++
++ switch (op->common.node->type) {
++ /*
++ * For these types, we need the actual node, not the subobject.
++ * However, the subobject got an extra reference count above.
++ */
++ case ACPI_TYPE_MUTEX:
++ case ACPI_TYPE_METHOD:
++ case ACPI_TYPE_POWER:
++ case ACPI_TYPE_PROCESSOR:
++ case ACPI_TYPE_EVENT:
++ case ACPI_TYPE_REGION:
++ case ACPI_TYPE_DEVICE:
++ case ACPI_TYPE_THERMAL:
++
++ obj_desc =
++ (union acpi_operand_object *)op->common.
++ node;
++ break;
++
++ default:
++ break;
++ }
++
++ /*
++ * If above resolved to an operand object, we are done. Otherwise,
++ * we have a NS node, we must create the package entry as a named
++ * reference.
++ */
++ if (ACPI_GET_DESCRIPTOR_TYPE(obj_desc) !=
++ ACPI_DESC_TYPE_NAMED) {
++ goto exit;
++ }
++ }
+ }
+
+ /* Create and init a new internal ACPI object */
+@@ -156,6 +221,7 @@
+ return_ACPI_STATUS(status);
+ }
+
++ exit:
+ *obj_desc_ptr = obj_desc;
+ return_ACPI_STATUS(AE_OK);
+ }
+@@ -358,12 +424,25 @@
+ arg = arg->common.next;
+ for (i = 0; arg; i++) {
+ if (arg->common.aml_opcode == AML_INT_RETURN_VALUE_OP) {
+-
+- /* Object (package or buffer) is already built */
++ if (arg->common.node->type == ACPI_TYPE_METHOD) {
++ /*
++ * A method reference "looks" to the parser to be a method
++ * invocation, so we special case it here
++ */
++ arg->common.aml_opcode = AML_INT_NAMEPATH_OP;
++ status =
++ acpi_ds_build_internal_object(walk_state,
++ arg,
++ &obj_desc->
++ package.
++ elements[i]);
++ } else {
++ /* This package element is already built, just get it */
+
+ obj_desc->package.elements[i] =
+ ACPI_CAST_PTR(union acpi_operand_object,
+ arg->common.node);
++ }
+ } else {
+ status = acpi_ds_build_internal_object(walk_state, arg,
+ &obj_desc->
+--- linux-2.6.18.8/drivers/acpi/hardware/hwsleep.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/hardware/hwsleep.c 2008-05-19 00:33:19.949296034 +0300
+@@ -227,7 +227,11 @@
+ u32 PM1Bcontrol;
+ struct acpi_bit_register_info *sleep_type_reg_info;
+ struct acpi_bit_register_info *sleep_enable_reg_info;
++#if !(defined(CONFIG_XEN) && defined(CONFIG_X86))
+ u32 in_value;
++#else
++ int err;
++#endif
+ acpi_status status;
+
+ ACPI_FUNCTION_TRACE(acpi_enter_sleep_state);
+@@ -327,6 +331,7 @@
+
+ ACPI_FLUSH_CPU_CACHE();
+
++#if !(defined(CONFIG_XEN) && defined(CONFIG_X86))
+ status = acpi_hw_register_write(ACPI_MTX_DO_NOT_LOCK,
+ ACPI_REGISTER_PM1A_CONTROL,
+ PM1Acontrol);
+@@ -376,6 +381,16 @@
+ /* Spin until we wake */
+
+ } while (!in_value);
++#else
++ /* PV ACPI just need check hypercall return value */
++ err = acpi_notify_hypervisor_state(sleep_state,
++ PM1Acontrol, PM1Bcontrol);
++ if (err) {
++ ACPI_DEBUG_PRINT((ACPI_DB_ERROR,
++ "Hypervisor failure [%d]\n", err));
++ return_ACPI_STATUS(AE_ERROR);
++ }
++#endif
+
+ return_ACPI_STATUS(AE_OK);
+ }
+--- linux-2.6.18.8/drivers/acpi/processor_core.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/processor_core.c 2008-05-19 00:33:20.461325548 +0300
+@@ -474,8 +474,14 @@
+ * they are physically not present.
+ */
+ if (cpu_index == -1) {
++#ifdef CONFIG_XEN
++ if (ACPI_FAILURE
++ (acpi_processor_hotadd_init(pr->handle, &pr->id)) &&
++ !processor_cntl_external()) {
++#else
+ if (ACPI_FAILURE
+ (acpi_processor_hotadd_init(pr->handle, &pr->id))) {
++#endif /* CONFIG_XEN */
+ printk(KERN_ERR PREFIX
+ "Getting cpuindex for acpiid 0x%x\n",
+ pr->acpi_id);
+@@ -517,7 +523,11 @@
+ return 0;
+ }
+
++#ifdef CONFIG_XEN
++static void *processor_device_array[NR_ACPI_CPUS];
++#else
+ static void *processor_device_array[NR_CPUS];
++#endif /* CONFIG_XEN */
+
+ static int acpi_processor_start(struct acpi_device *device)
+ {
+@@ -529,27 +539,48 @@
+ pr = acpi_driver_data(device);
+
+ result = acpi_processor_get_info(pr);
++#ifdef CONFIG_XEN
++ if (result ||
++ ((pr->id == -1) && !processor_cntl_external())) {
++#else
+ if (result) {
++#endif /* CONFIG_XEN */
+ /* Processor is physically not present */
+ return 0;
+ }
+
++#ifdef CONFIG_XEN
++ BUG_ON(!processor_cntl_external() &&
++ ((pr->id >= NR_CPUS) || (pr->id < 0)));
++#else
+ BUG_ON((pr->id >= NR_CPUS) || (pr->id < 0));
++#endif /* CONFIG_XEN */
+
+ /*
+ * Buggy BIOS check
+ * ACPI id of processors can be reported wrongly by the BIOS.
+ * Don't trust it blindly
+ */
++#ifdef CONFIG_XEN
++ if (processor_device_array[pr->acpi_id] != NULL &&
++ processor_device_array[pr->acpi_id] != (void *)device) {
++#else
+ if (processor_device_array[pr->id] != NULL &&
+ processor_device_array[pr->id] != (void *)device) {
++#endif /* CONFIG_XEN */
+ printk(KERN_WARNING "BIOS reported wrong ACPI id"
+ "for the processor\n");
+ return -ENODEV;
+ }
++#ifdef CONFIG_XEN
++ processor_device_array[pr->acpi_id] = (void *)device;
++ if (pr->id != -1)
++ processors[pr->id] = pr;
++#else
+ processor_device_array[pr->id] = (void *)device;
+
+ processors[pr->id] = pr;
++#endif /* CONFIG_XEN */
+
+ result = acpi_processor_add_fs(device);
+ if (result)
+@@ -564,6 +595,10 @@
+
+ acpi_processor_power_init(pr, device);
+
++#ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++ processor_extcntl_init(pr);
++#endif
++
+ if (pr->flags.throttling) {
+ printk(KERN_INFO PREFIX "%s [%s] (supports",
+ acpi_device_name(device), acpi_device_bid(device));
+@@ -656,7 +691,13 @@
+
+ acpi_processor_remove_fs(device);
+
++#ifdef CONFIG_XEN
++ if (pr->id != -1)
+ processors[pr->id] = NULL;
++#else
++ processors[pr->id] = NULL;
++#endif /* CONFIG_XEN */
++
+
+ kfree(pr);
+
+@@ -710,6 +751,12 @@
+ if (!pr)
+ return -ENODEV;
+
++#ifdef CONFIG_XEN
++ if (processor_cntl_external())
++ processor_notify_external(pr,
++ PROCESSOR_HOTPLUG, HOTPLUG_TYPE_ADD);
++#endif /* CONFIG_XEN */
++
+ if ((pr->id >= 0) && (pr->id < NR_CPUS)) {
+ kobject_uevent(&(*device)->kobj, KOBJ_ONLINE);
+ }
+@@ -748,6 +795,12 @@
+ break;
+ }
+
++#ifdef CONFIG_XEN
++ if (processor_cntl_external())
++ processor_notify_external(pr,
++ PROCESSOR_HOTPLUG, HOTPLUG_TYPE_ADD);
++#endif /* CONFIG_XEN */
++
+ if (pr->id >= 0 && (pr->id < NR_CPUS)) {
+ kobject_uevent(&device->kobj, KOBJ_OFFLINE);
+ break;
+@@ -777,8 +830,20 @@
+ return;
+ }
+
++#ifdef CONFIG_XEN
++ if ((pr->id >= 0) && (pr->id < NR_CPUS)
++ && (cpu_present(pr->id)))
++#else
+ if ((pr->id < NR_CPUS) && (cpu_present(pr->id)))
++#endif /* CONFIG_XEN */
+ kobject_uevent(&device->kobj, KOBJ_OFFLINE);
++
++#ifdef CONFIG_XEN
++ if (processor_cntl_external())
++ processor_notify_external(pr, PROCESSOR_HOTPLUG,
++ HOTPLUG_TYPE_REMOVE);
++#endif /* CONFIG_XEN */
++
+ break;
+ default:
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO,
+@@ -843,6 +908,11 @@
+
+ static int acpi_processor_handle_eject(struct acpi_processor *pr)
+ {
++#ifdef CONFIG_XEN
++ if (pr->id == -1)
++ return (0);
++#endif /* CONFIG_XEN */
++
+ if (cpu_online(pr->id)) {
+ return (-EINVAL);
+ }
+--- linux-2.6.18.8/drivers/acpi/processor_extcntl.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/processor_extcntl.c 2008-05-19 00:33:20.465325778 +0300
+@@ -0,0 +1,248 @@
++/*
++ * processor_extcntl.c - channel to external control logic
++ *
++ * Copyright (C) 2008, Intel corporation
++ *
++ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or (at
++ * your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful, but
++ * WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
++ * General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License along
++ * with this program; if not, write to the Free Software Foundation, Inc.,
++ * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
++ *
++ */
++
++#include <linux/kernel.h>
++#include <linux/init.h>
++#include <linux/types.h>
++#include <linux/acpi.h>
++#include <linux/pm.h>
++#include <linux/cpu.h>
++
++#include <acpi/processor.h>
++
++static int processor_extcntl_parse_csd(struct acpi_processor *pr);
++static int processor_extcntl_get_performance(struct acpi_processor *pr);
++/*
++ * External processor control logic may register with its own set of
++ * ops to get ACPI related notification. One example is like VMM.
++ */
++struct processor_extcntl_ops *processor_extcntl_ops;
++
++static int processor_notify_smm(void)
++{
++ acpi_status status;
++ static int is_done = 0;
++
++ /* only need successfully notify BIOS once */
++ /* avoid double notification which may lead to unexpected result */
++ if (is_done)
++ return 0;
++
++ /* Can't write pstate_cnt to smi_cmd if either value is zero */
++ if ((!acpi_fadt.smi_cmd) || (!acpi_fadt.pstate_cnt)) {
++ ACPI_DEBUG_PRINT((ACPI_DB_INFO,"No SMI port or pstate_cnt\n"));
++ return 0;
++ }
++
++ ACPI_DEBUG_PRINT((ACPI_DB_INFO,
++ "Writing pstate_cnt [0x%x] to smi_cmd [0x%x]\n",
++ acpi_fadt.pstate_cnt, acpi_fadt.smi_cmd));
++
++ /* FADT v1 doesn't support pstate_cnt, many BIOS vendors use
++ * it anyway, so we need to support it... */
++ if (acpi_fadt_is_v1) {
++ ACPI_DEBUG_PRINT((ACPI_DB_INFO,
++ "Using v1.0 FADT reserved value for pstate_cnt\n"));
++ }
++
++ status = acpi_os_write_port(acpi_fadt.smi_cmd,
++ (u32) acpi_fadt.pstate_cnt, 8);
++ if (ACPI_FAILURE(status))
++ return status;
++
++ is_done = 1;
++
++ return 0;
++}
++
++int processor_notify_external(struct acpi_processor *pr, int event, int type)
++{
++ int ret = -EINVAL;
++
++ if (!processor_cntl_external())
++ return -EINVAL;
++
++ switch (event) {
++ case PROCESSOR_PM_INIT:
++ case PROCESSOR_PM_CHANGE:
++ if ((type >= PM_TYPE_MAX) ||
++ !processor_extcntl_ops->pm_ops[type])
++ break;
++
++ ret = processor_extcntl_ops->pm_ops[type](pr, event);
++ break;
++ case PROCESSOR_HOTPLUG:
++ if (processor_extcntl_ops->hotplug)
++ ret = processor_extcntl_ops->hotplug(pr, type);
++ break;
++ default:
++ printk(KERN_ERR "Unsupport processor events %d.\n", event);
++ break;
++ }
++
++ return ret;
++}
++
++/*
++ * External control logic can decide to grab full or part of physical
++ * processor control bits. Take a VMM for example, physical processors
++ * are owned by VMM and thus existence information like hotplug is
++ * always required to be notified to VMM. Similar is processor idle
++ * state which is also necessarily controlled by VMM. But for other
++ * control bits like performance/throttle states, VMM may choose to
++ * control or not upon its own policy.
++ *
++ * Such ownership is unlikely to be switched in the fly, and thus
++ * not sure unregister interface is required or not. Being same reason,
++ * lock is not forced now.
++ */
++int processor_register_extcntl(struct processor_extcntl_ops *ops)
++{
++ if (!ops)
++ return -EINVAL;
++
++ if (processor_extcntl_ops &&
++ (processor_extcntl_ops != ops))
++ return -EBUSY;
++
++ processor_extcntl_ops = ops;
++ return 0;
++}
++
++int processor_unregister_extcntl(struct processor_extcntl_ops *ops)
++{
++ if (processor_extcntl_ops == ops)
++ processor_extcntl_ops = NULL;
++
++ return 0;
++}
++
++/*
++ * This is called from ACPI processor init, and targeted to hold
++ * some tricky housekeeping jobs to satisfy external control model.
++ * For example, we may put dependency parse stub here for idle
++ * and performance state. Those information may be not available
++ * if splitting from dom0 control logic like cpufreq driver.
++ */
++int processor_extcntl_init(struct acpi_processor *pr)
++{
++ /* parse cstate dependency information */
++ if (processor_pm_external())
++ processor_extcntl_parse_csd(pr);
++
++ /* Initialize performance states */
++ if (processor_pmperf_external())
++ processor_extcntl_get_performance(pr);
++
++ return 0;
++}
++
++/*
++ * Currently no _CSD is implemented which is why existing ACPI code
++ * doesn't parse _CSD at all. But to keep interface complete with
++ * external control logic, we put a placeholder here for future
++ * compatibility.
++ */
++static int processor_extcntl_parse_csd(struct acpi_processor *pr)
++{
++ int i;
++
++ for (i = 0; i < pr->power.count; i++) {
++ if (!pr->power.states[i].valid)
++ continue;
++
++ /* No dependency by default */
++ pr->power.states[i].domain_info = NULL;
++ pr->power.states[i].csd_count = 0;
++ }
++
++ return 0;
++}
++
++/*
++ * Existing ACPI module does parse performance states at some point,
++ * when acpi-cpufreq driver is loaded which however is something
++ * we'd like to disable to avoid confliction with external control
++ * logic. So we have to collect raw performance information here
++ * when ACPI processor object is found and started.
++ */
++#ifdef CONFIG_CPU_FREQ
++static int processor_extcntl_get_performance(struct acpi_processor *pr)
++{
++ int ret;
++ struct acpi_processor_performance *perf;
++ struct acpi_psd_package *pdomain;
++
++ if (pr->performance)
++ return -EBUSY;
++
++ perf = kzalloc(sizeof(struct acpi_processor_performance), GFP_KERNEL);
++ if (!perf)
++ return -ENOMEM;
++
++ pr->performance = perf;
++ /* Get basic performance state information */
++ ret = acpi_processor_get_performance_info(pr);
++ if (ret < 0)
++ goto err_out;
++
++ /*
++ * Well, here we need retrieve performance dependency information
++ * from _PSD object. The reason why existing interface is not used
++ * is due to the reason that existing interface sticks to Linux cpu
++ * id to construct some bitmap, however we want to split ACPI
++ * processor objects from Linux cpu id logic. For example, even
++ * when Linux is configured as UP, we still want to parse all ACPI
++ * processor objects to external logic. In this case, it's preferred
++ * to use ACPI ID instead.
++ */
++ pr->performance->domain_info.num_processors = 0;
++ ret = acpi_processor_get_psd(pr);
++ if (ret < 0)
++ goto err_out;
++
++ /* Some sanity check */
++ pdomain = &pr->performance->domain_info;
++ if ((pdomain->revision != ACPI_PSD_REV0_REVISION) ||
++ (pdomain->num_entries != ACPI_PSD_REV0_ENTRIES) ||
++ ((pdomain->coord_type != DOMAIN_COORD_TYPE_SW_ALL) &&
++ (pdomain->coord_type != DOMAIN_COORD_TYPE_SW_ANY) &&
++ (pdomain->coord_type != DOMAIN_COORD_TYPE_HW_ALL))) {
++ ret = -EINVAL;
++ goto err_out;
++ }
++
++ /* Last step is to notify BIOS that external logic exists */
++ processor_notify_smm();
++
++ processor_notify_external(pr, PROCESSOR_PM_INIT, PM_TYPE_PERF);
++
++ return 0;
++err_out:
++ pr->performance = NULL;
++ kfree(perf);
++ return ret;
++}
++#else
++static int processor_extcntl_get_performance(struct acpi_processor *pr) { return 0; }
++#endif
+--- linux-2.6.18.8/drivers/acpi/processor_idle.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/processor_idle.c 2008-05-19 00:33:20.465325778 +0300
+@@ -714,8 +714,17 @@
+ (reg->space_id != ACPI_ADR_SPACE_FIXED_HARDWARE))
+ continue;
+
++#ifdef CONFIG_XEN
++ if (!processor_pm_external())
++ cx.address = (reg->space_id ==
++ ACPI_ADR_SPACE_FIXED_HARDWARE) ?
++ 0 : reg->address;
++ else
++ cx.address = reg->address;
++#else
+ cx.address = (reg->space_id == ACPI_ADR_SPACE_FIXED_HARDWARE) ?
+ 0 : reg->address;
++#endif /* CONFIG_XEN */
+
+ /* There should be an easy way to extract an integer... */
+ obj = (union acpi_object *)&(element->package.elements[1]);
+@@ -724,9 +733,17 @@
+
+ cx.type = obj->integer.value;
+
++#ifdef CONFIG_XEN
++ /* Following check doesn't apply to external control case */
++ if (!processor_pm_external())
++ if ((cx.type != ACPI_STATE_C1) &&
++ (reg->space_id != ACPI_ADR_SPACE_SYSTEM_IO))
++ continue;
++#else
+ if ((cx.type != ACPI_STATE_C1) &&
+ (reg->space_id != ACPI_ADR_SPACE_SYSTEM_IO))
+ continue;
++#endif /* CONFIG_XEN */
+
+ if ((cx.type < ACPI_STATE_C2) || (cx.type > ACPI_STATE_C3))
+ continue;
+@@ -743,6 +760,12 @@
+
+ cx.power = obj->integer.value;
+
++#ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++ /* cache control methods to notify external logic */
++ if (processor_pm_external())
++ memcpy(&cx.reg, reg, sizeof(*reg));
++#endif
++
+ current_count++;
+ memcpy(&(pr->power.states[current_count]), &cx, sizeof(cx));
+
+@@ -985,12 +1008,24 @@
+ return -ENODEV;
+
+ /* Fall back to the default idle loop */
++#ifdef CONFIG_XEN
++ if (!processor_pm_external())
++ pm_idle = pm_idle_save;
++#else
+ pm_idle = pm_idle_save;
++#endif /* CONFIG_XEN */
+ synchronize_sched(); /* Relies on interrupts forcing exit from idle. */
+
+ pr->flags.power = 0;
+ result = acpi_processor_get_power_info(pr);
++#ifdef CONFIG_XEN
++ if (processor_pm_external())
++ processor_notify_external(pr,
++ PROCESSOR_PM_CHANGE, PM_TYPE_IDLE);
++ else if ((pr->flags.power == 1) && (pr->flags.power_setup_done))
++#else
+ if ((pr->flags.power == 1) && (pr->flags.power_setup_done))
++#endif /* CONFIG_XEN */
+ pm_idle = acpi_processor_idle;
+
+ return result;
+@@ -1122,7 +1157,11 @@
+ pr->power.states[i].type);
+ printk(")\n");
+
++#ifdef CONFIG_XEN
++ if (!processor_pm_external() && (pr->id == 0)) {
++#else
+ if (pr->id == 0) {
++#endif /* CONFIG_XEN */
+ pm_idle_save = pm_idle;
+ pm_idle = acpi_processor_idle;
+ }
+@@ -1141,6 +1180,11 @@
+
+ pr->flags.power_setup_done = 1;
+
++#ifdef CONFIG_XEN
++ if (processor_pm_external())
++ processor_notify_external(pr,
++ PROCESSOR_PM_INIT, PM_TYPE_IDLE);
++#endif /* CONFIG_XEN */
+ return 0;
+ }
+
+--- linux-2.6.18.8/drivers/acpi/processor_perflib.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/processor_perflib.c 2008-05-19 00:33:20.465325778 +0300
+@@ -136,6 +136,11 @@
+ int ret = acpi_processor_get_platform_limit(pr);
+ if (ret < 0)
+ return (ret);
++#ifdef CONFIG_XEN
++ else if (processor_pmperf_external())
++ return processor_notify_external(pr,
++ PROCESSOR_PM_CHANGE, PM_TYPE_PERF);
++#endif /* CONFIG_XEN */
+ else
+ return cpufreq_update_policy(pr->id);
+ }
+@@ -299,7 +304,11 @@
+ return result;
+ }
+
++#ifndef CONFIG_PROCESSOR_EXTERNAL_CONTROL
+ static int acpi_processor_get_performance_info(struct acpi_processor *pr)
++#else
++int acpi_processor_get_performance_info(struct acpi_processor *pr)
++#endif
+ {
+ int result = 0;
+ acpi_status status = AE_OK;
+@@ -538,7 +547,11 @@
+ }
+ #endif /* CONFIG_X86_ACPI_CPUFREQ_PROC_INTF */
+
++#ifndef CONFIG_PROCESSOR_EXTERNAL_CONTROL
+ static int acpi_processor_get_psd(struct acpi_processor *pr)
++#else
++int acpi_processor_get_psd(struct acpi_processor *pr)
++#endif
+ {
+ int result = 0;
+ acpi_status status = AE_OK;
+--- linux-2.6.18.8/drivers/acpi/resources/rsxface.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/resources/rsxface.c 2008-05-19 00:33:20.481326701 +0300
+@@ -476,8 +476,6 @@
+ return (AE_CTRL_TERMINATE);
+ }
+
+-ACPI_EXPORT_SYMBOL(acpi_rs_match_vendor_resource)
+-
+ /*******************************************************************************
+ *
+ * FUNCTION: acpi_walk_resources
+--- linux-2.6.18.8/drivers/acpi/sleep/main.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/sleep/main.c 2008-05-19 00:33:20.485326931 +0300
@@ -91,7 +91,14 @@
break;
return 0;
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/acpi/sleep/poweroff.c linux-2.6.18-xen.hg/drivers/acpi/sleep/poweroff.c
---- linux-2.6.18/drivers/acpi/sleep/poweroff.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/acpi/sleep/poweroff.c 2007-12-23 11:15:01.989585156 +0100
+--- linux-2.6.18.8/drivers/acpi/sleep/poweroff.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/acpi/sleep/poweroff.c 2008-05-19 00:33:20.485326931 +0300
@@ -20,6 +20,7 @@
int acpi_sleep_prepare(u32 acpi_state)
{
ACPI_FLUSH_CPU_CACHE();
acpi_enable_wakeup_device_prep(acpi_state);
#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/block/floppy.c linux-2.6.18-xen.hg/drivers/block/floppy.c
---- linux-2.6.18/drivers/block/floppy.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/block/floppy.c 2007-12-23 11:15:02.082923396 +0100
+--- linux-2.6.18.8/drivers/block/floppy.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/block/floppy.c 2008-05-19 00:33:20.765343071 +0300
@@ -4392,12 +4392,16 @@
if (fd_request_dma()) {
DPRINT("Unable to grab DMA%d for the floppy driver\n",
for (fdc = 0; fdc < N_FDC; fdc++) {
if (FDCS->address != -1) {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/agp/amd64-agp.c linux-2.6.18-xen.hg/drivers/char/agp/amd64-agp.c
---- linux-2.6.18/drivers/char/agp/amd64-agp.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/char/agp/amd64-agp.c 2007-12-23 11:15:02.149593567 +0100
+--- linux-2.6.18.8/drivers/char/agp/amd64-agp.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/agp/amd64-agp.c 2008-05-19 00:33:21.213368896 +0300
@@ -15,6 +15,7 @@
#include <linux/mmzone.h>
#include <asm/page.h> /* PAGE_SIZE */
/* Request the Aperture. This catches cases when someone else
already put a mapping in there - happens with some very broken BIOS
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/agp/generic.c linux-2.6.18-xen.hg/drivers/char/agp/generic.c
---- linux-2.6.18/drivers/char/agp/generic.c 2007-12-23 11:26:51.070123757 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/agp/generic.c 2007-12-23 11:15:02.152927074 +0100
+--- linux-2.6.18.8/drivers/char/agp/generic.c 2008-05-19 00:42:33.913230658 +0300
++++ linux-2.6.18-xen.hg/drivers/char/agp/generic.c 2008-05-19 00:33:21.305374199 +0300
@@ -51,28 +51,6 @@
*/
EXPORT_SYMBOL_GPL(agp_memory_reserved);
/*
* Generic routines for handling agp_memory structures -
* They use the basic page allocation routines to do the brunt of the work.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/agp/intel-agp.c linux-2.6.18-xen.hg/drivers/char/agp/intel-agp.c
---- linux-2.6.18/drivers/char/agp/intel-agp.c 2007-12-23 11:26:51.070123757 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/agp/intel-agp.c 2007-12-23 11:15:02.152927074 +0100
+--- linux-2.6.18.8/drivers/char/agp/intel-agp.c 2008-05-19 00:42:33.965233656 +0300
++++ linux-2.6.18-xen.hg/drivers/char/agp/intel-agp.c 2008-05-19 00:33:21.309374430 +0300
@@ -164,9 +164,17 @@
if (page == NULL)
return NULL;
atomic_dec(&agp_bridge->current_memory_agp);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/mem.c linux-2.6.18-xen.hg/drivers/char/mem.c
---- linux-2.6.18/drivers/char/mem.c 2007-12-23 11:26:51.073457266 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/mem.c 2007-12-23 11:15:04.043026325 +0100
+--- linux-2.6.18.8/drivers/char/mem.c 2008-05-19 00:42:33.969233887 +0300
++++ linux-2.6.18-xen.hg/drivers/char/mem.c 2008-05-19 00:33:22.273429998 +0300
@@ -101,6 +101,7 @@
}
#endif
#ifndef __HAVE_PHYS_MEM_ACCESS_PROT
static pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,
-@@ -780,6 +782,7 @@
+@@ -262,6 +264,9 @@
+ static int mmap_kmem(struct file * file, struct vm_area_struct * vma)
+ {
+ unsigned long pfn;
++#ifdef CONFIG_XEN
++ unsigned long i, count;
++#endif
+
+ /* Turn a kernel-virtual address into a physical page frame */
+ pfn = __pa((u64)vma->vm_pgoff << PAGE_SHIFT) >> PAGE_SHIFT;
+@@ -276,6 +281,13 @@
+ if (!pfn_valid(pfn))
+ return -EIO;
+
++#ifdef CONFIG_XEN
++ count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
++ for (i = 0; i < count; i++)
++ if ((pfn + i) != mfn_to_local_pfn(pfn_to_mfn(pfn + i)))
++ return -EIO;
++#endif
++
+ vma->vm_pgoff = pfn;
+ return mmap_mem(file, vma);
+ }
+@@ -780,6 +792,7 @@
#define open_kmem open_mem
#define open_oldmem open_mem
static const struct file_operations mem_fops = {
.llseek = memory_lseek,
.read = read_mem,
-@@ -787,6 +790,9 @@
+@@ -787,6 +800,9 @@
.mmap = mmap_mem,
.open = open_mem,
};
static const struct file_operations kmem_fops = {
.llseek = memory_lseek,
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/Kconfig linux-2.6.18-xen.hg/drivers/char/tpm/Kconfig
---- linux-2.6.18/drivers/char/tpm/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/Kconfig 2007-12-23 11:15:05.833120324 +0100
+--- linux-2.6.18.8/drivers/char/tpm/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/Kconfig 2008-05-19 00:33:22.477441757 +0300
@@ -58,5 +58,13 @@
Further information on this driver and the supported hardware
can be found at http://www.prosec.rub.de/tpm
+ will be called tpm_xenu.
+endmenu
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/Makefile linux-2.6.18-xen.hg/drivers/char/tpm/Makefile
---- linux-2.6.18/drivers/char/tpm/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/Makefile 2007-12-23 11:15:05.836453833 +0100
+--- linux-2.6.18.8/drivers/char/tpm/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/Makefile 2008-05-19 00:33:22.477441757 +0300
@@ -9,3 +9,5 @@
obj-$(CONFIG_TCG_NSC) += tpm_nsc.o
obj-$(CONFIG_TCG_ATMEL) += tpm_atmel.o
obj-$(CONFIG_TCG_INFINEON) += tpm_infineon.o
+obj-$(CONFIG_TCG_XEN) += tpm_xenu.o
+tpm_xenu-y = tpm_xen.o tpm_vtpm.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/tpm.h linux-2.6.18-xen.hg/drivers/char/tpm/tpm.h
---- linux-2.6.18/drivers/char/tpm/tpm.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm.h 2007-12-23 11:15:05.836453833 +0100
+--- linux-2.6.18.8/drivers/char/tpm/tpm.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm.h 2008-05-19 00:33:22.481441988 +0300
@@ -105,6 +105,9 @@
struct dentry **bios_dir;
extern void tpm_get_timeouts(struct tpm_chip *);
extern void tpm_gen_interrupt(struct tpm_chip *);
extern void tpm_continue_selftest(struct tpm_chip *);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/tpm_vtpm.c linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.c
---- linux-2.6.18/drivers/char/tpm/tpm_vtpm.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.c 2007-12-23 11:15:05.839787342 +0100
+--- linux-2.6.18.8/drivers/char/tpm/tpm_vtpm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.c 2008-05-19 00:33:22.485442218 +0300
@@ -0,0 +1,542 @@
+/*
+ * Copyright (C) 2006 IBM Corporation
+ tpm_remove_hardware(dev);
+ kfree(vtpms);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/tpm_vtpm.h linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.h
---- linux-2.6.18/drivers/char/tpm/tpm_vtpm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.h 2007-12-23 11:15:05.839787342 +0100
+--- linux-2.6.18.8/drivers/char/tpm/tpm_vtpm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_vtpm.h 2008-05-19 00:33:22.485442218 +0300
@@ -0,0 +1,55 @@
+#ifndef TPM_VTPM_H
+#define TPM_VTPM_H
+}
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tpm/tpm_xen.c linux-2.6.18-xen.hg/drivers/char/tpm/tpm_xen.c
---- linux-2.6.18/drivers/char/tpm/tpm_xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_xen.c 2007-12-23 11:15:05.839787342 +0100
-@@ -0,0 +1,719 @@
+--- linux-2.6.18.8/drivers/char/tpm/tpm_xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tpm/tpm_xen.c 2008-05-19 00:33:22.489442449 +0300
+@@ -0,0 +1,722 @@
+/*
+ * Copyright (c) 2005, IBM Corporation
+ *
+ case XenbusStateInitialising:
+ case XenbusStateInitWait:
+ case XenbusStateInitialised:
++ case XenbusStateReconfiguring:
++ case XenbusStateReconfigured:
+ case XenbusStateUnknown:
+ break;
+
+ tx = &tp->tx->ring[i].req;
+ tx->addr = virt_to_machine(txb->data);
+ tx->size = txb->len;
++ tx->unused = 0;
+
+ DPRINTK("First 4 characters sent by TPM-FE are "
+ "0x%02x 0x%02x 0x%02x 0x%02x\n",
+module_init(tpmif_init);
+
+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/char/tty_io.c linux-2.6.18-xen.hg/drivers/char/tty_io.c
---- linux-2.6.18/drivers/char/tty_io.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/char/tty_io.c 2007-12-23 11:15:05.843120852 +0100
+--- linux-2.6.18.8/drivers/char/tty_io.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/char/tty_io.c 2008-05-19 00:33:22.489442449 +0300
@@ -130,6 +130,8 @@
vt.c for deeply disgusting hack reasons */
DEFINE_MUTEX(tty_mutex);
#endif
return 0;
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/cpufreq/cpufreq_ondemand.c linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_ondemand.c
---- linux-2.6.18/drivers/cpufreq/cpufreq_ondemand.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_ondemand.c 2007-12-23 11:15:06.203139758 +0100
+--- linux-2.6.18.8/drivers/cpufreq/cpufreq_ondemand.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_ondemand.c 2008-05-19 00:33:22.565446830 +0300
@@ -96,6 +96,7 @@
return retval;
}
/* Check for frequency increase */
if (load > dbs_tuners_ins.up_threshold) {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/cpufreq/cpufreq_stats.c linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_stats.c
---- linux-2.6.18/drivers/cpufreq/cpufreq_stats.c 2007-12-23 11:26:51.073457266 +0100
-+++ linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_stats.c 2007-12-23 11:15:06.356481136 +0100
+--- linux-2.6.18.8/drivers/cpufreq/cpufreq_stats.c 2008-05-19 00:42:33.973234117 +0300
++++ linux-2.6.18-xen.hg/drivers/cpufreq/cpufreq_stats.c 2008-05-19 00:33:22.569447060 +0300
@@ -292,6 +292,9 @@
if (old_index == new_index)
return 0;
spin_lock(&cpufreq_stats_lock);
stat->last_index = new_index;
#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/i2c/busses/i2c-i801.c linux-2.6.18-xen.hg/drivers/i2c/busses/i2c-i801.c
---- linux-2.6.18/drivers/i2c/busses/i2c-i801.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/i2c/busses/i2c-i801.c 2007-12-23 11:15:07.589879240 +0100
-@@ -33,6 +33,7 @@
+--- linux-2.6.18.8/drivers/i2c/busses/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/i2c/busses/Kconfig 2008-05-19 00:33:22.929467812 +0300
+@@ -125,6 +125,8 @@
+ ICH7
+ ESB2
+ ICH8
++ ICH9
++ ICH10
+
+ This driver can also be built as a module. If so, the module
+ will be called i2c-i801.
+--- linux-2.6.18.8/drivers/i2c/busses/i2c-i801.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/i2c/busses/i2c-i801.c 2008-05-19 00:33:22.953469196 +0300
+@@ -33,6 +33,9 @@
ICH7 27DA
ESB2 269B
ICH8 283E
+ ICH9 2930
++ ICH10 3A30
++ ICH10 3A60
This driver supports several versions of Intel's I/O Controller Hubs (ICH).
For SMBus support, they are similar to the PIIX4 and are part
of Intel's '810' and other chipsets.
-@@ -457,6 +458,7 @@
+@@ -457,6 +460,9 @@
{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_17) },
{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ESB2_17) },
{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_5) },
+ { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH9_6) },
++ { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH10_4) },
++ { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH10_5) },
{ 0, }
};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/i2c/busses/Kconfig linux-2.6.18-xen.hg/drivers/i2c/busses/Kconfig
---- linux-2.6.18/drivers/i2c/busses/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/i2c/busses/Kconfig 2007-12-23 11:15:07.583212226 +0100
-@@ -125,6 +125,7 @@
- ICH7
- ESB2
- ICH8
-+ ICH9
-
- This driver can also be built as a module. If so, the module
- will be called i2c-i801.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/ide/ide-lib.c linux-2.6.18-xen.hg/drivers/ide/ide-lib.c
---- linux-2.6.18/drivers/ide/ide-lib.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/ide/ide-lib.c 2007-12-23 11:15:07.633214851 +0100
+--- linux-2.6.18.8/drivers/ide/ide-lib.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/ide/ide-lib.c 2008-05-19 00:33:23.037474038 +0300
@@ -408,10 +408,10 @@
{
u64 addr = BLK_BOUNCE_HIGH; /* dma64_addr_t */
addr = HWIF(drive)->pci_dev->dma_mask;
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/Makefile linux-2.6.18-xen.hg/drivers/Makefile
---- linux-2.6.18/drivers/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/Makefile 2007-12-23 11:15:01.252879806 +0100
-@@ -31,6 +31,7 @@
- obj-$(CONFIG_NUBUS) += nubus/
- obj-$(CONFIG_ATM) += atm/
- obj-$(CONFIG_PPC_PMAC) += macintosh/
-+obj-$(CONFIG_XEN) += xen/
- obj-$(CONFIG_IDE) += ide/
- obj-$(CONFIG_FC4) += fc4/
- obj-$(CONFIG_SCSI) += scsi/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/net/r8169.c linux-2.6.18-xen.hg/drivers/net/r8169.c
---- linux-2.6.18/drivers/net/r8169.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/net/r8169.c 2007-12-23 11:15:19.093816661 +0100
+--- linux-2.6.18.8/drivers/input/gameport/gameport.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/input/gameport/gameport.c 2008-05-19 00:33:23.781516925 +0300
+@@ -37,8 +37,6 @@
+ EXPORT_SYMBOL(gameport_open);
+ EXPORT_SYMBOL(gameport_close);
+ EXPORT_SYMBOL(gameport_rescan);
+-EXPORT_SYMBOL(gameport_cooked_read);
+-EXPORT_SYMBOL(gameport_set_name);
+ EXPORT_SYMBOL(gameport_set_phys);
+ EXPORT_SYMBOL(gameport_start_polling);
+ EXPORT_SYMBOL(gameport_stop_polling);
+--- linux-2.6.18.8/drivers/net/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/Kconfig 2008-05-19 00:33:28.081764793 +0300
+@@ -2399,6 +2399,8 @@
+ <file:Documentation/networking/net-modules.txt>. The module
+ will be called myri10ge.
+
++source "drivers/net/sfc/Kconfig"
++
+ endmenu
+
+ source "drivers/net/tokenring/Kconfig"
+--- linux-2.6.18.8/drivers/net/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/Makefile 2008-05-19 00:33:28.085765023 +0300
+@@ -219,3 +219,5 @@
+
+ obj-$(CONFIG_FS_ENET) += fs_enet/
+
++obj-$(CONFIG_SFC) += sfc/
++
+--- linux-2.6.18.8/drivers/net/r8169.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/r8169.c 2008-05-19 00:33:28.801806296 +0300
@@ -81,7 +81,7 @@
#ifdef RTL8169_DEBUG
}
static void __exit
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/buffer_sync.c linux-2.6.18-xen.hg/drivers/oprofile/buffer_sync.c
---- linux-2.6.18/drivers/oprofile/buffer_sync.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/buffer_sync.c 2007-12-23 11:15:23.417377030 +0100
-@@ -6,6 +6,10 @@
- *
- * @author John Levon <levon@movementarian.org>
- *
-+ * Modified by Aravind Menon for Xen
-+ * These modifications are:
-+ * Copyright (C) 2005 Hewlett-Packard Co.
+--- linux-2.6.18.8/drivers/net/sfc/Kconfig 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/Kconfig 2008-05-19 00:33:28.825807680 +0300
+@@ -0,0 +1,35 @@
++config SFC
++ tristate "Solarflare Solarstorm SFC4000 support"
++ depends on PCI && INET
++ select MII
++ help
++ This driver supports 10-gigabit Ethernet cards based on
++ the Solarflare Communications Solarstorm SFC4000 controller.
++
++ To compile this driver as a module, choose M here. The module
++ will be called sfc.
++
++config SFC_DEBUGFS
++ bool "Solarflare Solarstorm SFC4000 debugging support"
++ depends on SFC && DEBUG_FS
++ default N
++ help
++ This option creates an "sfc" subdirectory of debugfs with
++ debugging information for the SFC4000 driver.
++
++ If unsure, say N.
++
++config SFC_MTD
++ depends on SFC && MTD && MTD_PARTITIONS
++ tristate "Solarflare Solarstorm SFC4000 flash/EEPROM support"
++ help
++ This module exposes the on-board flash and/or EEPROM memory as
++ MTD devices (e.g. /dev/mtd1). This makes it possible to upload a
++ new boot ROM to the NIC.
++
++config SFC_RESOURCE
++ depends on SFC && X86
++ tristate "Solarflare Solarstorm SFC4000 resource driver"
++ help
++ This module provides the SFC resource manager driver.
++
+--- linux-2.6.18.8/drivers/net/sfc/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/Makefile 2008-05-19 00:33:28.825807680 +0300
+@@ -0,0 +1,43 @@
++
++# Final objects
++sfc_o = sfc.o
++sfc_mtd_o = sfc_mtd.o
++
++# Constituent objects
++sfc_elements_o :=
++sfc_elements_o += efx.o
++sfc_elements_o += falcon.o
++sfc_elements_o += tx.o
++sfc_elements_o += rx.o
++sfc_elements_o += mentormac.o
++sfc_elements_o += falcon_gmac.o
++sfc_elements_o += falcon_xmac.o
++sfc_elements_o += alaska.o
++sfc_elements_o += i2c-direct.o
++sfc_elements_o += selftest.o
++sfc_elements_o += driverlink.o
++ifeq ($(CONFIG_SFC_DEBUGFS),y)
++sfc_elements_o += debugfs.o
++endif
++sfc_elements_o += ethtool.o
++sfc_elements_o += xfp_phy.o
++sfc_elements_o += mdio_10g.o
++sfc_elements_o += txc43128_phy.o
++sfc_elements_o += tenxpress.o
++sfc_elements_o += lm87_support.o
++sfc_elements_o += boards.o
++sfc_elements_o += sfe4001.o
++sfc_elements_o += pm8358_phy.o
++sfc_elements_o += null_phy.o
++sfc_elements_o += phy.o
++sfc_elements_o += kernel_compat.o
++
++sfc_mtd_elements_o := mtd.o
++
++obj-$(CONFIG_SFC) += $(sfc_o)
++obj-$(CONFIG_SFC_MTD) += $(sfc_mtd_o)
++
++sfc-objs = $(sfc_elements_o)
++sfc_mtd-objs = $(sfc_mtd_elements_o)
++
++obj-$(CONFIG_SFC_RESOURCE) += sfc_resource/
+--- linux-2.6.18.8/drivers/net/sfc/alaska.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/alaska.c 2008-05-19 00:33:28.825807680 +0300
+@@ -0,0 +1,159 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
+ *
- * This is the core of the buffer management. Each
- * CPU buffer is processed and entered into the
- * global event buffer. Such processing is necessary
-@@ -38,6 +42,7 @@
- static DEFINE_SPINLOCK(task_mortuary);
- static void process_task_mortuary(void);
-
-+static int cpu_current_domain[NR_CPUS];
-
- /* Take ownership of the task struct and place it on the
- * list for processing. Only after two full buffer syncs
-@@ -146,6 +151,11 @@
- int sync_start(void)
- {
- int err;
-+ int i;
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ for (i = 0; i < NR_CPUS; i++) {
-+ cpu_current_domain[i] = COORDINATOR_DOMAIN;
++#include "net_driver.h"
++#include <linux/ethtool.h>
++#include "gmii.h"
++#include "phy.h"
++
++/* Marvell 88E1111 "Alaska" PHY control */
++#define ALASKA_PHY_SPECIFIC 16
++#define ALASKA_ALLOW_SLEEP 0x0200
++
++#define ALASKA_EXTENDED_CONTROL 20
++#define EXTENDED_LINE_LOOPBACK 0x8000
++
++#define ALASKA_LED_CONTROL 24
++#define LED_BLINK_MASK 0x0700
++#define LED_BLINK_FAST 0x0100
++#define LED_BLINK_SLOW 0x0300
++#define LED_TX_CTRL_MASK 0x0041
++#define LED_TX_CTRL_LINK_AND_ACTIVITY 0x0001
++
++#define ALASKA_LED_OVERRIDE 25
++#define LED_LINK1000_MASK 0x0030
++#define LED_LINK1000_BLINK 0x0010
++#define LED_TX_MASK 0x0003
++#define LED_TX_BLINK 0x0001
++
++static void alaska_reconfigure(struct efx_nic *efx)
++{
++ struct mii_if_info *gmii = &efx->mii;
++ u32 bmcr, phy_ext;
++
++ /* Configure line loopback if requested */
++ phy_ext = gmii->mdio_read(gmii->dev, gmii->phy_id,
++ ALASKA_EXTENDED_CONTROL);
++ if (efx->loopback_mode == LOOPBACK_NETWORK)
++ phy_ext |= EXTENDED_LINE_LOOPBACK;
++ else
++ phy_ext &= ~EXTENDED_LINE_LOOPBACK;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, ALASKA_EXTENDED_CONTROL,
++ phy_ext);
++
++ /* Configure PHY loopback if requested */
++ bmcr = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_BMCR);
++ if (efx->loopback_mode == LOOPBACK_PHY)
++ bmcr |= BMCR_LOOPBACK;
++ else
++ bmcr &= ~BMCR_LOOPBACK;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, MII_BMCR, bmcr);
++
++ /* Read link up status */
++ if (efx->loopback_mode == LOOPBACK_NONE)
++ efx->link_up = mii_link_ok(gmii);
++ else
++ efx->link_up = 1;
++
++ /* Determine link options from PHY */
++ if (gmii->force_media) {
++ efx->link_options = gmii_forced_result(bmcr);
++ } else {
++ int lpa = gmii_lpa(gmii);
++ int adv = gmii_advertised(gmii);
++ efx->link_options = gmii_nway_result(adv & lpa);
+ }
-
- start_cpu_work();
-
-@@ -275,13 +285,29 @@
- last_cookie = INVALID_COOKIE;
- }
-
--static void add_kernel_ctx_switch(unsigned int in_kernel)
-+static void add_cpu_mode_switch(unsigned int cpu_mode)
- {
- add_event_entry(ESCAPE_CODE);
-- if (in_kernel)
-+ switch (cpu_mode) {
-+ case CPU_MODE_USER:
-+ add_event_entry(USER_ENTER_SWITCH_CODE);
-+ break;
-+ case CPU_MODE_KERNEL:
- add_event_entry(KERNEL_ENTER_SWITCH_CODE);
-- else
-- add_event_entry(KERNEL_EXIT_SWITCH_CODE);
-+ break;
-+ case CPU_MODE_XEN:
-+ add_event_entry(XEN_ENTER_SWITCH_CODE);
-+ break;
-+ default:
-+ break;
++}
++
++static void alaska_clear_interrupt(struct efx_nic *efx)
++{
++ struct mii_if_info *gmii = &efx->mii;
++
++ /* Read interrupt status register to clear */
++ gmii->mdio_read(gmii->dev, gmii->phy_id, GMII_ISR);
++}
++
++static int alaska_init(struct efx_nic *efx)
++{
++ struct mii_if_info *gmii = &efx->mii;
++ u32 ier, leds, ctrl_1g, phy_spec;
++
++ /* Read ISR to clear any outstanding PHY interrupts */
++ gmii->mdio_read(gmii->dev, gmii->phy_id, GMII_ISR);
++
++ /* Enable PHY interrupts */
++ ier = gmii->mdio_read(gmii->dev, gmii->phy_id, GMII_IER);
++ ier |= IER_LINK_CHG;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, GMII_IER, ier);
++
++ /* Remove 1G half-duplex as unsupported in Mentor MAC */
++ ctrl_1g = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_CTRL1000);
++ ctrl_1g &= ~(ADVERTISE_1000HALF);
++ gmii->mdio_write(gmii->dev, gmii->phy_id, MII_CTRL1000, ctrl_1g);
++
++ /*
++ * The PHY can save power when there is no external connection
++ * (sleep mode). However, this is incompatible with PHY
++ * loopback, and if enable and disable it quickly the PHY can
++ * go to sleep even when sleep mode is disabled. (SFC bug
++ * 9309.) Therefore we disable it all the time.
++ */
++ phy_spec = gmii->mdio_read(gmii->dev, gmii->phy_id,
++ ALASKA_PHY_SPECIFIC);
++ phy_spec &= ~ALASKA_ALLOW_SLEEP;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, ALASKA_PHY_SPECIFIC,
++ phy_spec);
++
++ /* Configure LEDs */
++ leds = gmii->mdio_read(gmii->dev, gmii->phy_id, ALASKA_LED_CONTROL);
++ leds &= ~(LED_BLINK_MASK | LED_TX_CTRL_MASK);
++ leds |= (LED_BLINK_FAST | LED_TX_CTRL_LINK_AND_ACTIVITY);
++ gmii->mdio_write(gmii->dev, gmii->phy_id, ALASKA_LED_CONTROL, leds);
++
++ return 0;
++}
++
++static void alaska_fini(struct efx_nic *efx)
++{
++ struct mii_if_info *gmii = &efx->mii;
++ u32 ier;
++
++ /* Disable PHY interrupts */
++ ier = gmii->mdio_read(gmii->dev, gmii->phy_id, GMII_IER);
++ ier &= ~IER_LINK_CHG;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, GMII_IER, ier);
++}
++
++
++struct efx_phy_operations alaska_phy_operations = {
++ .init = alaska_init,
++ .fini = alaska_fini,
++ .reconfigure = alaska_reconfigure,
++ .clear_interrupt = alaska_clear_interrupt,
++ .loopbacks = (1 << LOOPBACK_PHY) | (1 << LOOPBACK_NETWORK),
++ .startup_loopback = LOOPBACK_PHY,
++};
+--- linux-2.6.18.8/drivers/net/sfc/bitfield.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/bitfield.h 2008-05-19 00:33:28.829807910 +0300
+@@ -0,0 +1,544 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_BITFIELD_H
++#define EFX_BITFIELD_H
++
++/*
++ * Efx bitfield access
++ *
++ * Efx NICs make extensive use of bitfields up to 128 bits
++ * wide. Since there is no native 128-bit datatype on most systems,
++ * and since 64-bit datatypes are inefficient on 32-bit systems and
++ * vice versa, we wrap accesses in a way that uses the most efficient
++ * datatype.
++ *
++ * The NICs are PCI devices and therefore little-endian. Since most
++ * of the quantities that we deal with are DMAed to/from host memory,
++ * we define our datatypes (efx_oword_t, efx_qword_t and
++ * efx_dword_t) to be little-endian.
++ */
++
++/* Lowest bit numbers and widths */
++#define EFX_DUMMY_FIELD_LBN 0
++#define EFX_DUMMY_FIELD_WIDTH 0
++#define EFX_DWORD_0_LBN 0
++#define EFX_DWORD_0_WIDTH 32
++#define EFX_DWORD_1_LBN 32
++#define EFX_DWORD_1_WIDTH 32
++#define EFX_DWORD_2_LBN 64
++#define EFX_DWORD_2_WIDTH 32
++#define EFX_DWORD_3_LBN 96
++#define EFX_DWORD_3_WIDTH 32
++
++#define EFX_BYTE 1
++#define EFX_WORD 2
++#define EFX_DWORD 4
++#define EFX_OWORD 8
++
++/* Specified attribute (e.g. LBN) of the specified field */
++#define EFX_VAL(field, attribute) field ## _ ## attribute
++/* Low bit number of the specified field */
++#define EFX_LOW_BIT(field) EFX_VAL(field, LBN)
++/* Bit width of the specified field */
++#define EFX_WIDTH(field) EFX_VAL(field, WIDTH)
++/* High bit number of the specified field */
++#define EFX_HIGH_BIT(field) (EFX_LOW_BIT(field) + EFX_WIDTH(field) - 1)
++/* Mask equal in width to the specified field.
++ *
++ * For example, a field with width 5 would have a mask of 0x1f.
++ *
++ * The maximum width mask that can be generated is 64 bits.
++ */
++#define EFX_MASK64(field) \
++ (EFX_WIDTH(field) == 64 ? ~((u64) 0) : \
++ (((((u64) 1) << EFX_WIDTH(field))) - 1))
++
++/* Mask equal in width to the specified field.
++ *
++ * For example, a field with width 5 would have a mask of 0x1f.
++ *
++ * The maximum width mask that can be generated is 32 bits. Use
++ * EFX_MASK64 for higher width fields.
++ */
++#define EFX_MASK32(field) \
++ (EFX_WIDTH(field) == 32 ? ~((u32) 0) : \
++ (((((u32) 1) << EFX_WIDTH(field))) - 1))
++
++/* A doubleword (i.e. 4 byte) datatype - little-endian in HW */
++typedef union efx_dword {
++ __le32 u32[1];
++} efx_dword_t;
++
++/* A quadword (i.e. 8 byte) datatype - little-endian in HW */
++typedef union efx_qword {
++ __le64 u64[1];
++ __le32 u32[2];
++ efx_dword_t dword[2];
++} efx_qword_t;
++
++/* An octword (eight-word, i.e. 16 byte) datatype - little-endian in HW */
++typedef union efx_oword {
++ __le64 u64[2];
++ efx_qword_t qword[2];
++ __le32 u32[4];
++ efx_dword_t dword[4];
++} efx_oword_t;
++
++/* Format string and value expanders for printk */
++#define EFX_DWORD_FMT "%08x"
++#define EFX_QWORD_FMT "%08x:%08x"
++#define EFX_OWORD_FMT "%08x:%08x:%08x:%08x"
++#define EFX_DWORD_VAL(dword) \
++ ((unsigned int) le32_to_cpu((dword).u32[0]))
++#define EFX_QWORD_VAL(qword) \
++ ((unsigned int) le32_to_cpu((qword).u32[1])), \
++ ((unsigned int) le32_to_cpu((qword).u32[0]))
++#define EFX_OWORD_VAL(oword) \
++ ((unsigned int) le32_to_cpu((oword).u32[3])), \
++ ((unsigned int) le32_to_cpu((oword).u32[2])), \
++ ((unsigned int) le32_to_cpu((oword).u32[1])), \
++ ((unsigned int) le32_to_cpu((oword).u32[0]))
++
++/*
++ * Extract bit field portion [low,high) from the native-endian element
++ * which contains bits [min,max).
++ *
++ * For example, suppose "element" represents the high 32 bits of a
++ * 64-bit value, and we wish to extract the bits belonging to the bit
++ * field occupying bits 28-45 of this 64-bit value.
++ *
++ * Then EFX_EXTRACT ( element, 32, 63, 28, 45 ) would give
++ *
++ * ( element ) << 4
++ *
++ * The result will contain the relevant bits filled in in the range
++ * [0,high-low), with garbage in bits [high-low+1,...).
++ */
++#define EFX_EXTRACT_NATIVE(native_element, min, max, low, high) \
++ (((low > max) || (high < min)) ? 0 : \
++ ((low > min) ? \
++ ((native_element) >> (low - min)) : \
++ ((native_element) << (min - low))))
++
++/*
++ * Extract bit field portion [low,high) from the 64-bit little-endian
++ * element which contains bits [min,max)
++ */
++#define EFX_EXTRACT64(element, min, max, low, high) \
++ EFX_EXTRACT_NATIVE(le64_to_cpu(element), min, max, low, high)
++
++/*
++ * Extract bit field portion [low,high) from the 32-bit little-endian
++ * element which contains bits [min,max)
++ */
++#define EFX_EXTRACT32(element, min, max, low, high) \
++ EFX_EXTRACT_NATIVE(le32_to_cpu(element), min, max, low, high)
++
++#define EFX_EXTRACT_OWORD64(oword, low, high) \
++ (EFX_EXTRACT64((oword).u64[0], 0, 63, low, high) | \
++ EFX_EXTRACT64((oword).u64[1], 64, 127, low, high))
++
++#define EFX_EXTRACT_QWORD64(qword, low, high) \
++ EFX_EXTRACT64((qword).u64[0], 0, 63, low, high)
++
++#define EFX_EXTRACT_OWORD32(oword, low, high) \
++ (EFX_EXTRACT32((oword).u32[0], 0, 31, low, high) | \
++ EFX_EXTRACT32((oword).u32[1], 32, 63, low, high) | \
++ EFX_EXTRACT32((oword).u32[2], 64, 95, low, high) | \
++ EFX_EXTRACT32((oword).u32[3], 96, 127, low, high))
++
++#define EFX_EXTRACT_QWORD32(qword, low, high) \
++ (EFX_EXTRACT32((qword).u32[0], 0, 31, low, high) | \
++ EFX_EXTRACT32((qword).u32[1], 32, 63, low, high))
++
++#define EFX_EXTRACT_DWORD(dword, low, high) \
++ EFX_EXTRACT32((dword).u32[0], 0, 31, low, high)
++
++#define EFX_OWORD_FIELD64(oword, field) \
++ (EFX_EXTRACT_OWORD64(oword, EFX_LOW_BIT(field), EFX_HIGH_BIT(field)) \
++ & EFX_MASK64(field))
++
++#define EFX_QWORD_FIELD64(qword, field) \
++ (EFX_EXTRACT_QWORD64(qword, EFX_LOW_BIT(field), EFX_HIGH_BIT(field)) \
++ & EFX_MASK64(field))
++
++#define EFX_OWORD_FIELD32(oword, field) \
++ (EFX_EXTRACT_OWORD32(oword, EFX_LOW_BIT(field), EFX_HIGH_BIT(field)) \
++ & EFX_MASK32(field))
++
++#define EFX_QWORD_FIELD32(qword, field) \
++ (EFX_EXTRACT_QWORD32(qword, EFX_LOW_BIT(field), EFX_HIGH_BIT(field)) \
++ & EFX_MASK32(field))
++
++#define EFX_DWORD_FIELD(dword, field) \
++ (EFX_EXTRACT_DWORD(dword, EFX_LOW_BIT(field), EFX_HIGH_BIT(field)) \
++ & EFX_MASK32(field))
++
++#define EFX_OWORD_IS_ZERO64(oword) \
++ (((oword).u64[0] | (oword).u64[1]) == (__force __le64) 0)
++
++#define EFX_QWORD_IS_ZERO64(qword) \
++ (((qword).u64[0]) == (__force __le64) 0)
++
++#define EFX_OWORD_IS_ZERO32(oword) \
++ (((oword).u32[0] | (oword).u32[1] | (oword).u32[2] | (oword).u32[3]) \
++ == (__force __le32) 0)
++
++#define EFX_QWORD_IS_ZERO32(qword) \
++ (((qword).u32[0] | (qword).u32[1]) == (__force __le32) 0)
++
++#define EFX_DWORD_IS_ZERO(dword) \
++ (((dword).u32[0]) == (__force __le32) 0)
++
++#define EFX_OWORD_IS_ALL_ONES64(oword) \
++ (((oword).u64[0] & (oword).u64[1]) == ~((__force __le64) 0))
++
++#define EFX_QWORD_IS_ALL_ONES64(qword) \
++ ((qword).u64[0] == ~((__force __le64) 0))
++
++#define EFX_OWORD_IS_ALL_ONES32(oword) \
++ (((oword).u32[0] & (oword).u32[1] & (oword).u32[2] & (oword).u32[3]) \
++ == ~((__force __le32) 0))
++
++#define EFX_QWORD_IS_ALL_ONES32(qword) \
++ (((qword).u32[0] & (qword).u32[1]) == ~((__force __le32) 0))
++
++#define EFX_DWORD_IS_ALL_ONES(dword) \
++ ((dword).u32[0] == ~((__force __le32) 0))
++
++#if BITS_PER_LONG == 64
++#define EFX_OWORD_FIELD EFX_OWORD_FIELD64
++#define EFX_QWORD_FIELD EFX_QWORD_FIELD64
++#define EFX_OWORD_IS_ZERO EFX_OWORD_IS_ZERO64
++#define EFX_QWORD_IS_ZERO EFX_QWORD_IS_ZERO64
++#define EFX_OWORD_IS_ALL_ONES EFX_OWORD_IS_ALL_ONES64
++#define EFX_QWORD_IS_ALL_ONES EFX_QWORD_IS_ALL_ONES64
++#else
++#define EFX_OWORD_FIELD EFX_OWORD_FIELD32
++#define EFX_QWORD_FIELD EFX_QWORD_FIELD32
++#define EFX_OWORD_IS_ZERO EFX_OWORD_IS_ZERO32
++#define EFX_QWORD_IS_ZERO EFX_QWORD_IS_ZERO32
++#define EFX_OWORD_IS_ALL_ONES EFX_OWORD_IS_ALL_ONES32
++#define EFX_QWORD_IS_ALL_ONES EFX_QWORD_IS_ALL_ONES32
++#endif
++
++/*
++ * Construct bit field portion
++ *
++ * Creates the portion of the bit field [low,high) that lies within
++ * the range [min,max).
++ */
++#define EFX_INSERT_NATIVE64(min, max, low, high, value) \
++ (((low > max) || (high < min)) ? 0 : \
++ ((low > min) ? \
++ (((u64) (value)) << (low - min)) : \
++ (((u64) (value)) >> (min - low))))
++
++#define EFX_INSERT_NATIVE32(min, max, low, high, value) \
++ (((low > max) || (high < min)) ? 0 : \
++ ((low > min) ? \
++ (((u32) (value)) << (low - min)) : \
++ (((u32) (value)) >> (min - low))))
++
++#define EFX_INSERT_NATIVE(min, max, low, high, value) \
++ ((((max - min) >= 32) || ((high - low) >= 32)) ? \
++ EFX_INSERT_NATIVE64(min, max, low, high, value) : \
++ EFX_INSERT_NATIVE32(min, max, low, high, value))
++
++/*
++ * Construct bit field portion
++ *
++ * Creates the portion of the named bit field that lies within the
++ * range [min,max).
++ */
++#define EFX_INSERT_FIELD_NATIVE(min, max, field, value) \
++ EFX_INSERT_NATIVE(min, max, EFX_LOW_BIT(field), \
++ EFX_HIGH_BIT(field), value)
++
++/*
++ * Construct bit field
++ *
++ * Creates the portion of the named bit fields that lie within the
++ * range [min,max).
++ */
++#define EFX_INSERT_FIELDS_NATIVE(min, max, \
++ field1, value1, \
++ field2, value2, \
++ field3, value3, \
++ field4, value4, \
++ field5, value5, \
++ field6, value6, \
++ field7, value7, \
++ field8, value8, \
++ field9, value9, \
++ field10, value10) \
++ (EFX_INSERT_FIELD_NATIVE((min), (max), field1, (value1)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field2, (value2)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field3, (value3)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field4, (value4)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field5, (value5)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field6, (value6)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field7, (value7)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field8, (value8)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field9, (value9)) | \
++ EFX_INSERT_FIELD_NATIVE((min), (max), field10, (value10)))
++
++#define EFX_INSERT_FIELDS64(...) \
++ cpu_to_le64(EFX_INSERT_FIELDS_NATIVE(__VA_ARGS__))
++
++#define EFX_INSERT_FIELDS32(...) \
++ cpu_to_le32(EFX_INSERT_FIELDS_NATIVE(__VA_ARGS__))
++
++#define EFX_POPULATE_OWORD64(oword, ...) do { \
++ (oword).u64[0] = EFX_INSERT_FIELDS64(0, 63, __VA_ARGS__); \
++ (oword).u64[1] = EFX_INSERT_FIELDS64(64, 127, __VA_ARGS__); \
++ } while (0)
++
++#define EFX_POPULATE_QWORD64(qword, ...) do { \
++ (qword).u64[0] = EFX_INSERT_FIELDS64(0, 63, __VA_ARGS__); \
++ } while (0)
++
++#define EFX_POPULATE_OWORD32(oword, ...) do { \
++ (oword).u32[0] = EFX_INSERT_FIELDS32(0, 31, __VA_ARGS__); \
++ (oword).u32[1] = EFX_INSERT_FIELDS32(32, 63, __VA_ARGS__); \
++ (oword).u32[2] = EFX_INSERT_FIELDS32(64, 95, __VA_ARGS__); \
++ (oword).u32[3] = EFX_INSERT_FIELDS32(96, 127, __VA_ARGS__); \
++ } while (0)
++
++#define EFX_POPULATE_QWORD32(qword, ...) do { \
++ (qword).u32[0] = EFX_INSERT_FIELDS32(0, 31, __VA_ARGS__); \
++ (qword).u32[1] = EFX_INSERT_FIELDS32(32, 63, __VA_ARGS__); \
++ } while (0)
++
++#define EFX_POPULATE_DWORD(dword, ...) do { \
++ (dword).u32[0] = EFX_INSERT_FIELDS32(0, 31, __VA_ARGS__); \
++ } while (0)
++
++#if BITS_PER_LONG == 64
++#define EFX_POPULATE_OWORD EFX_POPULATE_OWORD64
++#define EFX_POPULATE_QWORD EFX_POPULATE_QWORD64
++#else
++#define EFX_POPULATE_OWORD EFX_POPULATE_OWORD32
++#define EFX_POPULATE_QWORD EFX_POPULATE_QWORD32
++#endif
++
++/* Populate an octword field with various numbers of arguments */
++#define EFX_POPULATE_OWORD_10 EFX_POPULATE_OWORD
++#define EFX_POPULATE_OWORD_9(oword, ...) \
++ EFX_POPULATE_OWORD_10(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_8(oword, ...) \
++ EFX_POPULATE_OWORD_9(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_7(oword, ...) \
++ EFX_POPULATE_OWORD_8(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_6(oword, ...) \
++ EFX_POPULATE_OWORD_7(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_5(oword, ...) \
++ EFX_POPULATE_OWORD_6(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_4(oword, ...) \
++ EFX_POPULATE_OWORD_5(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_3(oword, ...) \
++ EFX_POPULATE_OWORD_4(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_2(oword, ...) \
++ EFX_POPULATE_OWORD_3(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_OWORD_1(oword, ...) \
++ EFX_POPULATE_OWORD_2(oword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_ZERO_OWORD(oword) \
++ EFX_POPULATE_OWORD_1(oword, EFX_DUMMY_FIELD, 0)
++#define EFX_SET_OWORD(oword) \
++ EFX_POPULATE_OWORD_4(oword, \
++ EFX_DWORD_0, 0xffffffff, \
++ EFX_DWORD_1, 0xffffffff, \
++ EFX_DWORD_2, 0xffffffff, \
++ EFX_DWORD_3, 0xffffffff)
++
++/* Populate a quadword field with various numbers of arguments */
++#define EFX_POPULATE_QWORD_10 EFX_POPULATE_QWORD
++#define EFX_POPULATE_QWORD_9(qword, ...) \
++ EFX_POPULATE_QWORD_10(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_8(qword, ...) \
++ EFX_POPULATE_QWORD_9(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_7(qword, ...) \
++ EFX_POPULATE_QWORD_8(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_6(qword, ...) \
++ EFX_POPULATE_QWORD_7(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_5(qword, ...) \
++ EFX_POPULATE_QWORD_6(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_4(qword, ...) \
++ EFX_POPULATE_QWORD_5(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_3(qword, ...) \
++ EFX_POPULATE_QWORD_4(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_2(qword, ...) \
++ EFX_POPULATE_QWORD_3(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_QWORD_1(qword, ...) \
++ EFX_POPULATE_QWORD_2(qword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_ZERO_QWORD(qword) \
++ EFX_POPULATE_QWORD_1(qword, EFX_DUMMY_FIELD, 0)
++#define EFX_SET_QWORD(qword) \
++ EFX_POPULATE_QWORD_2(qword, \
++ EFX_DWORD_0, 0xffffffff, \
++ EFX_DWORD_1, 0xffffffff)
++
++/* Populate a dword field with various numbers of arguments */
++#define EFX_POPULATE_DWORD_10 EFX_POPULATE_DWORD
++#define EFX_POPULATE_DWORD_9(dword, ...) \
++ EFX_POPULATE_DWORD_10(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_8(dword, ...) \
++ EFX_POPULATE_DWORD_9(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_7(dword, ...) \
++ EFX_POPULATE_DWORD_8(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_6(dword, ...) \
++ EFX_POPULATE_DWORD_7(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_5(dword, ...) \
++ EFX_POPULATE_DWORD_6(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_4(dword, ...) \
++ EFX_POPULATE_DWORD_5(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_3(dword, ...) \
++ EFX_POPULATE_DWORD_4(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_2(dword, ...) \
++ EFX_POPULATE_DWORD_3(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_POPULATE_DWORD_1(dword, ...) \
++ EFX_POPULATE_DWORD_2(dword, EFX_DUMMY_FIELD, 0, __VA_ARGS__)
++#define EFX_ZERO_DWORD(dword) \
++ EFX_POPULATE_DWORD_1(dword, EFX_DUMMY_FIELD, 0)
++#define EFX_SET_DWORD(dword) \
++ EFX_POPULATE_DWORD_1(dword, EFX_DWORD_0, 0xffffffff)
++
++/*
++ * Modify a named field within an already-populated structure. Used
++ * for read-modify-write operations.
++ *
++ */
++
++#define EFX_INVERT_OWORD(oword) do { \
++ (oword).u64[0] = ~((oword).u64[0]); \
++ (oword).u64[1] = ~((oword).u64[1]); \
++ } while (0)
++
++#define EFX_INSERT_FIELD64(...) \
++ cpu_to_le64(EFX_INSERT_FIELD_NATIVE(__VA_ARGS__))
++
++#define EFX_INSERT_FIELD32(...) \
++ cpu_to_le32(EFX_INSERT_FIELD_NATIVE(__VA_ARGS__))
++
++#define EFX_INPLACE_MASK64(min, max, field) \
++ EFX_INSERT_FIELD64(min, max, field, EFX_MASK64(field))
++
++#define EFX_INPLACE_MASK32(min, max, field) \
++ EFX_INSERT_FIELD32(min, max, field, EFX_MASK32(field))
++
++#define EFX_SET_OWORD_FIELD64(oword, field, value) do { \
++ (oword).u64[0] = (((oword).u64[0] \
++ & ~EFX_INPLACE_MASK64(0, 63, field)) \
++ | EFX_INSERT_FIELD64(0, 63, field, value)); \
++ (oword).u64[1] = (((oword).u64[1] \
++ & ~EFX_INPLACE_MASK64(64, 127, field)) \
++ | EFX_INSERT_FIELD64(64, 127, field, value)); \
++ } while (0)
++
++#define EFX_SET_QWORD_FIELD64(qword, field, value) do { \
++ (qword).u64[0] = (((qword).u64[0] \
++ & ~EFX_INPLACE_MASK64(0, 63, field)) \
++ | EFX_INSERT_FIELD64(0, 63, field, value)); \
++ } while (0)
++
++#define EFX_SET_OWORD_FIELD32(oword, field, value) do { \
++ (oword).u32[0] = (((oword).u32[0] \
++ & ~EFX_INPLACE_MASK32(0, 31, field)) \
++ | EFX_INSERT_FIELD32(0, 31, field, value)); \
++ (oword).u32[1] = (((oword).u32[1] \
++ & ~EFX_INPLACE_MASK32(32, 63, field)) \
++ | EFX_INSERT_FIELD32(32, 63, field, value)); \
++ (oword).u32[2] = (((oword).u32[2] \
++ & ~EFX_INPLACE_MASK32(64, 95, field)) \
++ | EFX_INSERT_FIELD32(64, 95, field, value)); \
++ (oword).u32[3] = (((oword).u32[3] \
++ & ~EFX_INPLACE_MASK32(96, 127, field)) \
++ | EFX_INSERT_FIELD32(96, 127, field, value)); \
++ } while (0)
++
++#define EFX_SET_QWORD_FIELD32(qword, field, value) do { \
++ (qword).u32[0] = (((qword).u32[0] \
++ & ~EFX_INPLACE_MASK32(0, 31, field)) \
++ | EFX_INSERT_FIELD32(0, 31, field, value)); \
++ (qword).u32[1] = (((qword).u32[1] \
++ & ~EFX_INPLACE_MASK32(32, 63, field)) \
++ | EFX_INSERT_FIELD32(32, 63, field, value)); \
++ } while (0)
++
++#define EFX_SET_DWORD_FIELD(dword, field, value) do { \
++ (dword).u32[0] = (((dword).u32[0] \
++ & ~EFX_INPLACE_MASK32(0, 31, field)) \
++ | EFX_INSERT_FIELD32(0, 31, field, value)); \
++ } while (0)
++
++#if BITS_PER_LONG == 64
++#define EFX_SET_OWORD_FIELD EFX_SET_OWORD_FIELD64
++#define EFX_SET_QWORD_FIELD EFX_SET_QWORD_FIELD64
++#else
++#define EFX_SET_OWORD_FIELD EFX_SET_OWORD_FIELD32
++#define EFX_SET_QWORD_FIELD EFX_SET_QWORD_FIELD32
++#endif
++
++#define EFX_SET_OWORD_FIELD_VER(efx, oword, field, value) do { \
++ if (FALCON_REV(efx) == FALCON_REV_B0) { \
++ EFX_SET_OWORD_FIELD((oword), field##_B0, (value)); \
++ } else { \
++ EFX_SET_OWORD_FIELD((oword), field##_A1, (value)); \
++ } \
++} while (0)
++
++#define EFX_QWORD_FIELD_VER(efx, qword, field) \
++ (FALCON_REV(efx) == FALCON_REV_B0 ? \
++ EFX_QWORD_FIELD((qword), field##_B0) : \
++ EFX_QWORD_FIELD((qword), field##_A1))
++
++/* Used to avoid compiler warnings about shift range exceeding width
++ * of the data types when dma_addr_t is only 32 bits wide.
++ */
++#define DMA_ADDR_T_WIDTH (8 * sizeof(dma_addr_t))
++#define EFX_DMA_TYPE_WIDTH(width) \
++ (((width) < DMA_ADDR_T_WIDTH) ? (width) : DMA_ADDR_T_WIDTH)
++#define EFX_DMA_MAX_MASK ((DMA_ADDR_T_WIDTH == 64) ? \
++ ~((u64) 0) : ~((u32) 0))
++#define EFX_DMA_MASK(mask) ((mask) & EFX_DMA_MAX_MASK)
++
++/*
++ * Determine if a DMA address is over the 4GB threshold
++ *
++ * Defined in a slightly tortuous way to avoid compiler warnings.
++ */
++static inline int efx_is_over_4gb(dma_addr_t address)
++{
++ if (DMA_ADDR_T_WIDTH > 32)
++ return (((u64) address) >> 32) ? 1 : 0;
++ else
++ /* Can never be true */
++ return 0;
++}
++
++#endif /* EFX_BITFIELD_H */
+--- linux-2.6.18.8/drivers/net/sfc/boards.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/boards.c 2008-05-19 00:33:28.829807910 +0300
+@@ -0,0 +1,528 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "net_driver.h"
++#include "phy.h"
++#include "lm87_support.h"
++#include "boards.h"
++#include "efx.h"
++
++/* Macros for unpacking the board revision */
++/* The revision info is in host byte order. */
++#define BOARD_TYPE(_rev) (_rev >> 8)
++#define BOARD_MAJOR(_rev) ((_rev >> 4) & 0xf)
++#define BOARD_MINOR(_rev) (_rev & 0xf)
++
++/* Blink support. If the PHY has no auto-blink mode so we hang it off a timer */
++#define BLINK_INTERVAL (HZ/2)
++
++static void blink_led_timer(unsigned long context)
++{
++ struct efx_nic *efx = (struct efx_nic *)context;
++ struct efx_blinker *bl = &efx->board_info.blinker;
++ efx->board_info.set_fault_led(efx, bl->state);
++ bl->state = !bl->state;
++ if (bl->resubmit) {
++ bl->timer.expires = jiffies + BLINK_INTERVAL;
++ add_timer(&bl->timer);
+ }
+}
+
-+static void add_domain_switch(unsigned long domain_id)
++static void board_blink(struct efx_nic *efx, int blink)
+{
-+ add_event_entry(ESCAPE_CODE);
-+ add_event_entry(DOMAIN_SWITCH_CODE);
-+ add_event_entry(domain_id);
- }
-
- static void
-@@ -348,9 +374,9 @@
- * for later lookup from userspace.
- */
- static int
--add_sample(struct mm_struct * mm, struct op_sample * s, int in_kernel)
-+add_sample(struct mm_struct * mm, struct op_sample * s, int cpu_mode)
- {
-- if (in_kernel) {
-+ if (cpu_mode >= CPU_MODE_KERNEL) {
- add_sample_entry(s->eip, s->event);
- return 1;
- } else if (mm) {
-@@ -496,15 +522,21 @@
- struct mm_struct *mm = NULL;
- struct task_struct * new;
- unsigned long cookie = 0;
-- int in_kernel = 1;
-+ int cpu_mode = 1;
- unsigned int i;
- sync_buffer_state state = sb_buffer_start;
- unsigned long available;
-+ int domain_switch = 0;
-
- mutex_lock(&buffer_mutex);
-
- add_cpu_switch(cpu);
-
-+ /* We need to assign the first samples in this CPU buffer to the
-+ same domain that we were processing at the last sync_buffer */
-+ if (cpu_current_domain[cpu] != COORDINATOR_DOMAIN) {
-+ add_domain_switch(cpu_current_domain[cpu]);
++ struct efx_blinker *blinker = &efx->board_info.blinker;
++
++ /* The rtnl mutex serialises all ethtool ioctls, so
++ * nothing special needs doing here. */
++ if (blink) {
++ blinker->resubmit = 1;
++ blinker->state = 0;
++ setup_timer(&blinker->timer, blink_led_timer,
++ (unsigned long)efx);
++ blinker->timer.expires = jiffies + BLINK_INTERVAL;
++ add_timer(&blinker->timer);
++ } else {
++ blinker->resubmit = 0;
++ if (blinker->timer.function)
++ del_timer_sync(&blinker->timer);
++ efx->board_info.set_fault_led(efx, 0);
+ }
- /* Remember, only we can modify tail_pos */
-
- available = get_slots(cpu_buf);
-@@ -512,16 +544,18 @@
- for (i = 0; i < available; ++i) {
- struct op_sample * s = &cpu_buf->buffer[cpu_buf->tail_pos];
-
-- if (is_code(s->eip)) {
-- if (s->event <= CPU_IS_KERNEL) {
-- /* kernel/userspace switch */
-- in_kernel = s->event;
-+ if (is_code(s->eip) && !domain_switch) {
-+ if (s->event <= CPU_MODE_XEN) {
-+ /* xen/kernel/userspace switch */
-+ cpu_mode = s->event;
- if (state == sb_buffer_start)
- state = sb_sample_start;
-- add_kernel_ctx_switch(s->event);
-+ add_cpu_mode_switch(s->event);
- } else if (s->event == CPU_TRACE_BEGIN) {
- state = sb_bt_start;
- add_trace_begin();
-+ } else if (s->event == CPU_DOMAIN_SWITCH) {
-+ domain_switch = 1;
- } else {
- struct mm_struct * oldmm = mm;
-
-@@ -535,19 +569,34 @@
- add_user_ctx_switch(new, cookie);
- }
- } else {
-- if (state >= sb_bt_start &&
-- !add_sample(mm, s, in_kernel)) {
-+ if (domain_switch) {
-+ cpu_current_domain[cpu] = s->eip;
-+ add_domain_switch(s->eip);
-+ domain_switch = 0;
-+ } else {
-+ if (cpu_current_domain[cpu] !=
-+ COORDINATOR_DOMAIN) {
-+ add_sample_entry(s->eip, s->event);
-+ }
-+ else if (state >= sb_bt_start &&
-+ !add_sample(mm, s, cpu_mode)) {
- if (state == sb_bt_start) {
- state = sb_bt_ignore;
- atomic_inc(&oprofile_stats.bt_lost_no_mapping);
- }
- }
- }
++}
++
++
++struct sensor_conf {
++ const char *name;
++ const unsigned high;
++ const unsigned low;
++};
++
++#define NO_LIMIT ((unsigned)-1)
++
++#define LM87_SENSOR_BYTES (18)
++
++static int sensor_limits_to_bytes(const struct sensor_conf *limits,
++ int nlimits, u8 *bytes, int maxbytes)
++{
++ int i, nbytes;
++ nbytes = 0;
++ for (i = 0; i < nlimits; i++) {
++ bytes[nbytes++] = limits[i].high;
++ if (limits[i].low != NO_LIMIT)
++ bytes[nbytes++] = limits[i].low;
++ /* We may have overrun by one at this point, but this test
++ * should only trigger in development drivers as the sizes
++ * are not dynamic. */
++ if (nbytes > maxbytes) {
++ printk(KERN_ERR "%s: out of space!\n", __func__);
++ break;
+ }
-
- increment_tail(cpu_buf);
- }
- release_mm(mm);
-
-+ /* We reset domain to COORDINATOR at each CPU switch */
-+ if (cpu_current_domain[cpu] != COORDINATOR_DOMAIN) {
-+ add_domain_switch(COORDINATOR_DOMAIN);
+ }
++ return nbytes;
++}
+
- mark_done(cpu);
-
- mutex_unlock(&buffer_mutex);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/cpu_buffer.c linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.c
---- linux-2.6.18/drivers/oprofile/cpu_buffer.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.c 2007-12-23 11:15:23.420710539 +0100
-@@ -6,6 +6,10 @@
- *
- * @author John Levon <levon@movementarian.org>
- *
-+ * Modified by Aravind Menon for Xen
-+ * These modifications are:
-+ * Copyright (C) 2005 Hewlett-Packard Co.
-+ *
- * Each CPU has a local buffer that stores PC value/event
- * pairs. We also log context switches when we notice them.
- * Eventually each CPU's buffer is processed into the global
-@@ -34,6 +38,8 @@
- #define DEFAULT_TIMER_EXPIRE (HZ / 10)
- static int work_enabled;
-
-+static int32_t current_domain = COORDINATOR_DOMAIN;
++/*****************************************************************************
++ * Support for the SFE4002
++ *
++ */
++/* LM87 configuration data for the sensor on the SFE4002 board */
++static const struct sensor_conf sfe4002_lm87_limits[] = {
++ {"1.8V line", 0x91, 0x83}, /* 2.5V sensor, scaled for 1.8V */
++ {"1.2V line", 0x5a, 0x51}, /* Vccp1 */
++ {"3.3V line", 0xca, 0xb6},
++ {"5V line", 0xc9, 0xb6},
++ {"12V line", 0xe0, 0xb0},
++ {"1V line", 0x4b, 0x44}, /* vccp2 */
++ {"Ext. temp.", 0x46, 0x0a}, /* ASIC temp. */
++ {"Int. temp.", 0x3c, 0x0a}, /* Board temp. */
++ {"1.66V line", 0xb2, NO_LIMIT}, /* AIN1 only takes 1 value */
++ {"1.5V line", 0xa1, NO_LIMIT} /* AIN2 only takes 1 value */
++};
++
++static const int sfe4002_lm87_nlimits = ARRAY_SIZE(sfe4002_lm87_limits);
++
++static u16 sfe4002_lm87_irq_mask = EFX_LM87_NO_INTS;
++
++/* I2C ID of the onboard LM87 chip. This is board-specific as the bottom two
++ * bits are set by strap pins */
++#define SFE4002_LM87_I2C_ID (0x2e)
++
++/****************************************************************************/
++/* LED allocations. Note that on rev A0 boards the schematic and the reality
++ * differ: red and green are swapped. Below is the fixed (A1) layout (there
++ * are only 3 A0 boards in existence, so no real reason to make this
++ * conditional).
++ */
++#define SFE4002_FAULT_LED (2) /* Red */
++#define SFE4002_RX_LED (0) /* Green */
++#define SFE4002_TX_LED (1) /* Amber */
++
++static int sfe4002_init_leds(struct efx_nic *efx)
++{
++ /* Set the TX and RX LEDs to reflect status and activity, and the
++ * fault LED off */
++ xfp_set_led(efx, SFE4002_TX_LED,
++ QUAKE_LED_TXLINK | QUAKE_LED_LINK_ACTSTAT);
++ xfp_set_led(efx, SFE4002_RX_LED,
++ QUAKE_LED_RXLINK | QUAKE_LED_LINK_ACTSTAT);
++ xfp_set_led(efx, SFE4002_FAULT_LED, QUAKE_LED_OFF);
++ efx->board_info.blinker.led_num = SFE4002_FAULT_LED;
++ return 0;
++}
+
- void free_cpu_buffers(void)
- {
- int i;
-@@ -57,7 +63,7 @@
- goto fail;
-
- b->last_task = NULL;
-- b->last_is_kernel = -1;
-+ b->last_cpu_mode = -1;
- b->tracing = 0;
- b->buffer_size = buffer_size;
- b->tail_pos = 0;
-@@ -113,7 +119,7 @@
- * collected will populate the buffer with proper
- * values to initialize the buffer
- */
-- cpu_buf->last_is_kernel = -1;
-+ cpu_buf->last_cpu_mode = -1;
- cpu_buf->last_task = NULL;
- }
-
-@@ -163,13 +169,13 @@
- * because of the head/tail separation of the writer and reader
- * of the CPU buffer.
- *
-- * is_kernel is needed because on some architectures you cannot
-+ * cpu_mode is needed because on some architectures you cannot
- * tell if you are in kernel or user space simply by looking at
-- * pc. We tag this in the buffer by generating kernel enter/exit
-- * events whenever is_kernel changes
-+ * pc. We tag this in the buffer by generating kernel/user (and xen)
-+ * enter events whenever cpu_mode changes
- */
- static int log_sample(struct oprofile_cpu_buffer * cpu_buf, unsigned long pc,
-- int is_kernel, unsigned long event)
-+ int cpu_mode, unsigned long event)
- {
- struct task_struct * task;
-
-@@ -180,18 +186,18 @@
- return 0;
- }
-
-- is_kernel = !!is_kernel;
--
- task = current;
-
- /* notice a switch from user->kernel or vice versa */
-- if (cpu_buf->last_is_kernel != is_kernel) {
-- cpu_buf->last_is_kernel = is_kernel;
-- add_code(cpu_buf, is_kernel);
-+ if (cpu_buf->last_cpu_mode != cpu_mode) {
-+ cpu_buf->last_cpu_mode = cpu_mode;
-+ add_code(cpu_buf, cpu_mode);
- }
-
- /* notice a task switch */
-- if (cpu_buf->last_task != task) {
-+ /* if not processing other domain samples */
-+ if ((cpu_buf->last_task != task) &&
-+ (current_domain == COORDINATOR_DOMAIN)) {
- cpu_buf->last_task = task;
- add_code(cpu_buf, (unsigned long)task);
- }
-@@ -275,6 +281,25 @@
- add_sample(cpu_buf, pc, 0);
- }
-
-+int oprofile_add_domain_switch(int32_t domain_id)
++static void sfe4002_fault_led(struct efx_nic *efx, int state)
+{
-+ struct oprofile_cpu_buffer * cpu_buf = &cpu_buffer[smp_processor_id()];
++ xfp_set_led(efx, SFE4002_FAULT_LED, state ? QUAKE_LED_ON :
++ QUAKE_LED_OFF);
++}
+
-+ /* should have space for switching into and out of domain
-+ (2 slots each) plus one sample and one cpu mode switch */
-+ if (((nr_available_slots(cpu_buf) < 6) &&
-+ (domain_id != COORDINATOR_DOMAIN)) ||
-+ (nr_available_slots(cpu_buf) < 2))
-+ return 0;
++static int sfe4002_sensor_meaning(struct efx_nic *efx, int limit_num,
++ unsigned val)
++{
++ const struct sensor_conf *lim = &sfe4002_lm87_limits[limit_num];
++ if (lim->low == NO_LIMIT)
++ EFX_ERR(efx, "%10s 0x%02x (nominal value 0x%02x)\n", lim->name,
++ val, lim->high);
++ else
++ EFX_ERR(efx, "%10s 0x%02x (nominal range 0x%02x - 0x%02x)\n",
++ lim->name, val, lim->high, lim->low);
++ return 1;
++}
+
-+ add_code(cpu_buf, CPU_DOMAIN_SWITCH);
-+ add_sample(cpu_buf, domain_id, 0);
++static int sfe4002_check_hw(struct efx_nic *efx)
++{
++ int rc;
+
-+ current_domain = domain_id;
++ /* A0 board rev. 4002s report a temperature fault the whole time
++ * (bad sensor) so we mask it out. */
++ unsigned alarm_mask = (efx->board_info.minor > 0) ?
++ 0 : ~EFX_LM87_ETMP_INT;
+
-+ return 1;
++ /* Check the sensor (NOP if not present). */
++ rc = efx_check_lm87(efx, alarm_mask);
++
++ /* We treat both lm87 interrupts and failure to talk to the lm87
++ * as problems (since failure will only be reported if we did
++ * find the sensor at probe time. */
++ if (rc)
++ EFX_ERR(efx, "sensor alert!\n");
++ return rc;
+}
+
- /*
- * This serves to avoid cpu buffer overflow, and makes sure
- * the task mortuary progresses
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/cpu_buffer.h linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.h
---- linux-2.6.18/drivers/oprofile/cpu_buffer.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.h 2007-12-23 11:15:23.420710539 +0100
-@@ -36,7 +36,7 @@
- volatile unsigned long tail_pos;
- unsigned long buffer_size;
- struct task_struct * last_task;
-- int last_is_kernel;
-+ int last_cpu_mode;
- int tracing;
- struct op_sample * buffer;
- unsigned long sample_received;
-@@ -51,7 +51,10 @@
- void cpu_buffer_reset(struct oprofile_cpu_buffer * cpu_buf);
-
- /* transient events for the CPU buffer -> event buffer */
--#define CPU_IS_KERNEL 1
--#define CPU_TRACE_BEGIN 2
-+#define CPU_MODE_USER 0
-+#define CPU_MODE_KERNEL 1
-+#define CPU_MODE_XEN 2
-+#define CPU_TRACE_BEGIN 3
-+#define CPU_DOMAIN_SWITCH 4
-
- #endif /* OPROFILE_CPU_BUFFER_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/event_buffer.h linux-2.6.18-xen.hg/drivers/oprofile/event_buffer.h
---- linux-2.6.18/drivers/oprofile/event_buffer.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/event_buffer.h 2007-12-23 11:15:23.420710539 +0100
-@@ -29,15 +29,20 @@
- #define CPU_SWITCH_CODE 2
- #define COOKIE_SWITCH_CODE 3
- #define KERNEL_ENTER_SWITCH_CODE 4
--#define KERNEL_EXIT_SWITCH_CODE 5
-+#define USER_ENTER_SWITCH_CODE 5
- #define MODULE_LOADED_CODE 6
- #define CTX_TGID_CODE 7
- #define TRACE_BEGIN_CODE 8
- #define TRACE_END_CODE 9
-+#define XEN_ENTER_SWITCH_CODE 10
-+#define DOMAIN_SWITCH_CODE 11
-
- #define INVALID_COOKIE ~0UL
- #define NO_COOKIE 0UL
-
-+/* Constant used to refer to coordinator domain (Xen) */
-+#define COORDINATOR_DOMAIN -1
++static int sfe4002_init(struct efx_nic *efx)
++{
++ u8 lm87_bytes[LM87_SENSOR_BYTES];
++ int nbytes;
++ int rc;
+
- /* add data to the event buffer */
- void add_event_entry(unsigned long data);
-
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/oprof.c linux-2.6.18-xen.hg/drivers/oprofile/oprof.c
---- linux-2.6.18/drivers/oprofile/oprof.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/oprof.c 2007-12-23 11:15:23.420710539 +0100
-@@ -5,6 +5,10 @@
- * @remark Read the file COPYING
- *
- * @author John Levon <levon@movementarian.org>
++ efx->board_info.monitor = sfe4002_check_hw;
++ efx->board_info.interpret_sensor = sfe4002_sensor_meaning;
++ efx->board_info.init_leds = sfe4002_init_leds;
++ efx->board_info.set_fault_led = sfe4002_fault_led;
++ efx->board_info.blink = board_blink;
++ /* To clean up shut down the lm87 (NOP if not present) */
++ efx->board_info.fini = efx_remove_lm87;
++
++ nbytes = sensor_limits_to_bytes(sfe4002_lm87_limits,
++ sfe4002_lm87_nlimits, lm87_bytes,
++ LM87_SENSOR_BYTES);
++
++ /* Activate the lm87 sensor if present (succeeds if nothing there) */
++ rc = efx_probe_lm87(efx, SFE4002_LM87_I2C_ID,
++ lm87_bytes, nbytes, sfe4002_lm87_irq_mask);
++
++ return rc;
++}
++
++/*****************************************************************************
++ * Support for the SFE4003
+ *
-+ * Modified by Aravind Menon for Xen
-+ * These modifications are:
-+ * Copyright (C) 2005 Hewlett-Packard Co.
- */
-
- #include <linux/kernel.h>
-@@ -33,6 +37,32 @@
- */
- static int timer = 0;
-
-+int oprofile_set_active(int active_domains[], unsigned int adomains)
-+{
-+ int err;
++ */
++/* LM87 configuration data for the sensor on the SFE4003 board */
++static const struct sensor_conf sfe4003_lm87_limits[] = {
++ {"1.5V line", 0x78, 0x6d}, /* 2.5V input, values scaled for 1.5V */
++ {"1.2V line", 0x5a, 0x51}, /* Vccp1 */
++ {"3.3V line", 0xca, 0xb6},
++ {"5V line", 0xc0, 0x00}, /* Sensor not connected. */
++ {"12V line", 0xe0, 0xb0},
++ {"1V line", 0x4b, 0x44}, /* Vccp2 */
++ {"Ext. temp.", 0x46, 0x0a}, /* ASIC temp. */
++ {"Int. temp.", 0x3c, 0x0a}, /* Board temp. */
++ {"", 0xff, NO_LIMIT}, /* FAN1/AIN1 unused */
++ {"", 0xff, NO_LIMIT} /* FAN2/AIN2 unused */
++};
+
-+ if (!oprofile_ops.set_active)
-+ return -EINVAL;
++static const int sfe4003_lm87_nlimits = ARRAY_SIZE(sfe4003_lm87_limits);
+
-+ mutex_lock(&start_mutex);
-+ err = oprofile_ops.set_active(active_domains, adomains);
-+ mutex_unlock(&start_mutex);
-+ return err;
++static u16 sfe4003_lm87_irq_mask = EFX_LM87_NO_INTS;
++
++
++static int sfe4003_sensor_meaning(struct efx_nic *efx, int limit_num,
++ unsigned val)
++{
++ const struct sensor_conf *lim = &sfe4003_lm87_limits[limit_num];
++ if (lim->low == NO_LIMIT)
++ return 0; /* Neither AIN1 nor AIN2 mean anything to us */
++ else
++ EFX_ERR(efx, "%10s 0x%02x (nominal range 0x%02x - 0x%02x)\n",
++ lim->name, val, lim->high, lim->low);
++ return 1;
+}
+
-+int oprofile_set_passive(int passive_domains[], unsigned int pdomains)
++/* I2C ID of the onboard LM87 chip. This is board-specific as the bottom two
++ * bits are set by strap pins */
++#define SFE4003_LM87_I2C_ID (0x2e)
++
++/* Board-specific LED info. */
++#define SFE4003_RED_LED_GPIO (11)
++#define SFE4003_LED_ON (1)
++#define SFE4003_LED_OFF (0)
++
++static void sfe4003_fault_led(struct efx_nic *efx, int state)
+{
-+ int err;
++ /* The LEDs were not wired to GPIOs before A3 */
++ if (efx->board_info.minor < 3 && efx->board_info.major == 0)
++ return;
+
-+ if (!oprofile_ops.set_passive)
-+ return -EINVAL;
++ txc_set_gpio_val(efx, SFE4003_RED_LED_GPIO,
++ state ? SFE4003_LED_ON : SFE4003_LED_OFF);
++}
+
-+ mutex_lock(&start_mutex);
-+ err = oprofile_ops.set_passive(passive_domains, pdomains);
-+ mutex_unlock(&start_mutex);
-+ return err;
++static int sfe4003_init_leds(struct efx_nic *efx)
++{
++ /* The LEDs were not wired to GPIOs before A3 */
++ if (efx->board_info.minor < 3 && efx->board_info.major == 0)
++ return 0;
++
++ txc_set_gpio_dir(efx, SFE4003_RED_LED_GPIO, TXC_GPIO_DIR_OUTPUT);
++ txc_set_gpio_val(efx, SFE4003_RED_LED_GPIO, SFE4003_LED_OFF);
++ return 0;
+}
+
- int oprofile_setup(void)
- {
- int err;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/oprof.h linux-2.6.18-xen.hg/drivers/oprofile/oprof.h
---- linux-2.6.18/drivers/oprofile/oprof.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/oprof.h 2007-12-23 11:15:23.420710539 +0100
-@@ -36,4 +36,7 @@
-
- int oprofile_set_backtrace(unsigned long depth);
-
-+int oprofile_set_active(int active_domains[], unsigned int adomains);
-+int oprofile_set_passive(int passive_domains[], unsigned int pdomains);
-+
- #endif /* OPROF_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/oprofile/oprofile_files.c linux-2.6.18-xen.hg/drivers/oprofile/oprofile_files.c
---- linux-2.6.18/drivers/oprofile/oprofile_files.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/oprofile/oprofile_files.c 2007-12-23 11:15:23.420710539 +0100
-@@ -5,10 +5,16 @@
- * @remark Read the file COPYING
- *
- * @author John Levon <levon@movementarian.org>
-+ *
-+ * Modified by Aravind Menon for Xen
-+ * These modifications are:
-+ * Copyright (C) 2005 Hewlett-Packard Co.
- */
-
- #include <linux/fs.h>
- #include <linux/oprofile.h>
-+#include <asm/uaccess.h>
-+#include <linux/ctype.h>
-
- #include "event_buffer.h"
- #include "oprofile_stats.h"
-@@ -118,10 +124,201 @@
- .write = dump_write,
- };
-
-+#define TMPBUFSIZE 512
++static int sfe4003_check_hw(struct efx_nic *efx)
++{
++ int rc;
++ /* A0/A1/A2 board rev. 4003s report a temperature fault the whole time
++ * (bad sensor) so we mask it out. */
++ unsigned alarm_mask =
++ ~(EFX_LM87_ETMP_INT | EFX_LM87_FAN1_INT | EFX_LM87_FAN2_INT);
+
-+static unsigned int adomains = 0;
-+static int active_domains[MAX_OPROF_DOMAINS + 1];
-+static DEFINE_MUTEX(adom_mutex);
++ /* Check the sensor (NOP if not present). */
+
-+static ssize_t adomain_write(struct file * file, char const __user * buf,
-+ size_t count, loff_t * offset)
++ rc = efx_check_lm87(efx, alarm_mask);
++ /* We treat both lm87 interrupts and failure to talk to the lm87
++ * as problems (since failure will only be reported if we did
++ * find the sensor at probe time. */
++ if (rc)
++ EFX_ERR(efx, "sensor alert!\n");
++
++ return rc;
++}
++
++static int sfe4003_init(struct efx_nic *efx)
+{
-+ char *tmpbuf;
-+ char *startp, *endp;
-+ int i;
-+ unsigned long val;
-+ ssize_t retval = count;
-+
-+ if (*offset)
-+ return -EINVAL;
-+ if (count > TMPBUFSIZE - 1)
-+ return -EINVAL;
++ u8 lm87_bytes[LM87_SENSOR_BYTES];
++ int nbytes;
++ int rc;
++ efx->board_info.monitor = sfe4003_check_hw;
++ efx->board_info.interpret_sensor = sfe4003_sensor_meaning;
++ efx->board_info.init_leds = sfe4003_init_leds;
++ efx->board_info.set_fault_led = sfe4003_fault_led;
++ efx->board_info.blink = board_blink;
++ /* To clean up shut down the lm87 (NOP if not present) */
++ efx->board_info.fini = efx_remove_lm87;
++
++ nbytes = sensor_limits_to_bytes(sfe4003_lm87_limits,
++ sfe4003_lm87_nlimits, lm87_bytes,
++ LM87_SENSOR_BYTES);
++
++ /* Activate the lm87 sensor if present (succeeds if nothing there) */
++ rc = efx_probe_lm87(efx, SFE4003_LM87_I2C_ID,
++ lm87_bytes, nbytes, sfe4003_lm87_irq_mask);
+
-+ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
-+ return -ENOMEM;
++ if (rc < 0)
++ EFX_ERR(efx, "Temperature sensor probe failure: "
++ "please check the jumper position\n");
++ return rc;
++}
+
-+ if (copy_from_user(tmpbuf, buf, count)) {
-+ kfree(tmpbuf);
-+ return -EFAULT;
++/*****************************************************************************
++ * Support for the SFE4005
++ *
++ */
++/* LM87 configuration data for the sensor on the SFE4005 board */
++static const u8 sfe4005_lm87_limits[] = {
++ 0x51, /* 2.5V high lim. (actually monitor 1.0V line, so 1050mV) */
++ 0x49, /* 2.5V low lim. (950mV) */
++ 0xf6, /* Vccp1 high lim. (3.3V rail, 3465 mV) */
++ 0xde, /* Vcpp1 low lim. (3.3V rail, 3135 mV) */
++ 0xca, /* 3.3V AUX high lim. (3465 mV) */
++ 0xb6, /* 3.3V AUX low lim. (3135mV) */
++ 0xc0, /* 5V high lim. not connected) */
++ 0x00, /* 5V low lim. (not connected) */
++ 0xd0, /* 12V high lim. (13000mV) */
++ 0xb0, /* 12V low lim. (11000mV) */
++ 0xc0, /* Vccp2 high lim. (unused) */
++ 0x00, /* Vccp2 low lim. (unused) */
++ 0x46, /* Ext temp 1 (ASIC) high lim. */
++ 0x0a, /* Ext temp 1 low lim. */
++ 0x3c, /* Int temp (board) high lim. */
++ 0x0a, /* Int temp 1 low lim. */
++ 0xff, /* Fan 1 high (unused) */
++ 0xff, /* Fan 2 high (unused) */
++};
++
++#define SFE4005_LM87_I2C_ID (0x2e)
++
++/* Until the LM87 monitoring is interrupt driven. */
++#define SFE4005_LM87_IRQMASK EFX_LM87_NO_INTS
++
++#define SFE4005_PCF8575_I2C_ID (0x20)
++/* Definitions for the I/O expander that controls the CX4 chip:
++ * which PCF8575 pin maps to which function */
++#define SFE4005_PORT0_EXTLOOP (1 << 0)
++#define SFE4005_PORT1_EXTLOOP (1 << 1)
++#define SFE4005_HOSTPROT_LOOP (1 << 2)
++#define SFE4005_BCAST (1 << 3) /* TX on both ports */
++#define SFE4005_PORT0_EQ (1 << 4)
++#define SFE4005_PORT1_EQ (1 << 5)
++#define SFE4005_HOSTPORT_EQ (1 << 6)
++#define SFE4005_PORTSEL (1 << 7) /* Which port (for RX in BCAST mode) */
++#define SFE4005_PORT0_PRE_LBN (8) /* Preemphasis on port 0 (2 bits)*/
++#define SFE4005_PORT1_PRE_LBN (10) /* Preemphasis on port 1 (2 bits)*/
++#define SFE4005_HOSTPORT_PRE_LBN (12) /* Preemphasis on host port (2 bits) */
++#define SFE4005_UNUSED (1 << 14)
++#define SFE4005_CX4uC_nRESET (1 << 15) /* Reset the controller on CX4 chip */
++
++
++/* By default only turn on host port EQ. Can also OR in SFE4005_PORT0_EQ,
++ * SFE4005_PORT1_EQ but this hasn't been seen to make a difference. */
++#define SFE4005_CX4_DEFAULTS (SFE4005_CX4uC_nRESET | SFE4005_HOSTPORT_EQ)
++
++static int sfe4005_write_ioexpander(struct efx_nic *efx)
++{
++ unsigned long iobits = (unsigned long)efx->phy_data;
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ u8 send[2], check[2];
++ int rc;
++ /* Do not, EVER, deassert nRESET as that will reset Falcon too,
++ * and the driver won't know to repush the configuration, so
++ * nothing will work until the next power cycle. */
++ BUG_ON(!(iobits & SFE4005_CX4uC_nRESET));
++ send[0] = (iobits & 0xff);
++ send[1] = ((iobits >> 8) & 0xff);
++ rc = efx_i2c_send_bytes(i2c, SFE4005_PCF8575_I2C_ID, send, 2);
++ if (rc) {
++ EFX_ERR(efx, "failed to write to I/O expander: %d\n", rc);
++ return rc;
+ }
-+ tmpbuf[count] = 0;
++ /* Paranoia: just check what the I/O expander reads back */
++ rc = efx_i2c_recv_bytes(i2c, SFE4005_PCF8575_I2C_ID, check, 2);
++ if (rc)
++ EFX_ERR(efx, "failed to read back from I/O expander: %d\n", rc);
++ else if (check[0] != send[0] || check[1] != send[1])
++ EFX_ERR(efx, "read back wrong value from I/O expander: "
++ "wanted %.2x%.2x, got %.2x%.2x\n",
++ send[1], send[0], check[1], check[0]);
++ return rc;
++}
+
-+ mutex_lock(&adom_mutex);
++static int sfe4005_init(struct efx_nic *efx)
++{
++ unsigned long iobits = SFE4005_CX4_DEFAULTS;
++ int rc;
+
-+ startp = tmpbuf;
-+ /* Parse one more than MAX_OPROF_DOMAINS, for easy error checking */
-+ for (i = 0; i <= MAX_OPROF_DOMAINS; i++) {
-+ val = simple_strtoul(startp, &endp, 0);
-+ if (endp == startp)
++ /* There is no PHY as such on the SFE4005 so phy_data is ours. */
++ efx->phy_data = (void *)iobits;
++
++ /* Push the values */
++ rc = sfe4005_write_ioexpander(efx);
++ if (rc)
++ return rc;
++
++ /* Activate the lm87 sensor if present (succeeds if nothing there) */
++ rc = efx_probe_lm87(efx, SFE4005_LM87_I2C_ID,
++ sfe4005_lm87_limits,
++ sizeof(sfe4005_lm87_limits), SFE4005_LM87_IRQMASK);
++
++ /* To clean up shut down the lm87 (NOP if not present) */
++ efx->board_info.fini = efx_remove_lm87;
++
++ return rc;
++}
++
++/* This will get expanded as board-specific details get moved out of the
++ * PHY drivers. */
++struct efx_board_data {
++ const char *ref_model;
++ const char *gen_type;
++ int (*init) (struct efx_nic *nic);
++ unsigned mwatts;
++};
++
++static void dummy_fini(struct efx_nic *nic)
++{
++}
++
++static int dummy_init(struct efx_nic *nic)
++{
++ nic->board_info.fini = dummy_fini;
++ return 0;
++}
++
++/* Maximum board power (mW)
++ * Falcon controller ASIC accounts for 2.2W
++ * 10Xpress PHY accounts for 12W
++ *
++ */
++#define SFE4001_POWER 18000
++#define SFE4002_POWER 7500
++#define SFE4003_POWER 4500
++#define SFE4005_POWER 4500
++
++static struct efx_board_data board_data[] = {
++ [EFX_BOARD_INVALID] =
++ {NULL, NULL, dummy_init, 0},
++ [EFX_BOARD_SFE4001] =
++ {"SFE4001", "10GBASE-T adapter", sfe4001_poweron, SFE4001_POWER },
++ [EFX_BOARD_SFE4002] =
++ {"SFE4002", "XFP adapter", sfe4002_init, SFE4002_POWER },
++ [EFX_BOARD_SFE4003] =
++ {"SFE4003", "10GBASE-CX4 adapter", sfe4003_init, SFE4003_POWER },
++ [EFX_BOARD_SFE4005] =
++ {"SFE4005", "10G blade adapter", sfe4005_init, SFE4005_POWER },
++};
++
++int efx_set_board_info(struct efx_nic *efx, u16 revision_info)
++{
++ int rc = 0;
++ struct efx_board_data *data;
++
++ if (BOARD_TYPE(revision_info) >= EFX_BOARD_MAX) {
++ EFX_ERR(efx, "squashing unknown board type %d\n",
++ BOARD_TYPE(revision_info));
++ revision_info = 0;
++ }
++
++ if (BOARD_TYPE(revision_info) == 0) {
++ efx->board_info.major = 0;
++ efx->board_info.minor = 0;
++ /* For early boards that don't have revision info. there is
++ * only 1 board for each PHY type, so we can work it out, with
++ * the exception of the PHY-less boards. */
++ switch (efx->phy_type) {
++ case PHY_TYPE_10XPRESS:
++ efx->board_info.type = EFX_BOARD_SFE4001;
+ break;
-+ while (ispunct(*endp) || isspace(*endp))
-+ endp++;
-+ active_domains[i] = val;
-+ if (active_domains[i] != val)
-+ /* Overflow, force error below */
-+ i = MAX_OPROF_DOMAINS + 1;
-+ startp = endp;
++ case PHY_TYPE_XFP:
++ efx->board_info.type = EFX_BOARD_SFE4002;
++ break;
++ case PHY_TYPE_CX4_RTMR:
++ efx->board_info.type = EFX_BOARD_SFE4003;
++ break;
++ default:
++ efx->board_info.type = 0;
++ break;
++ }
++ } else {
++ efx->board_info.type = BOARD_TYPE(revision_info);
++ efx->board_info.major = BOARD_MAJOR(revision_info);
++ efx->board_info.minor = BOARD_MINOR(revision_info);
+ }
-+ /* Force error on trailing junk */
-+ adomains = *startp ? MAX_OPROF_DOMAINS + 1 : i;
+
-+ kfree(tmpbuf);
++ data = &board_data[efx->board_info.type];
+
-+ if (adomains > MAX_OPROF_DOMAINS
-+ || oprofile_set_active(active_domains, adomains)) {
-+ adomains = 0;
-+ retval = -EINVAL;
-+ }
++ /* Report the board model number or generic type for recognisable
++ * boards. */
++ if (efx->board_info.type != 0)
++ EFX_INFO(efx, "board is %s rev %c%d\n",
++ (efx->pci_dev->subsystem_vendor == EFX_VENDID_SFC)
++ ? data->ref_model : data->gen_type,
++ 'A' + efx->board_info.major, efx->board_info.minor);
+
-+ mutex_unlock(&adom_mutex);
-+ return retval;
++ efx->board_info.init = data->init;
++ efx->board_info.mwatts = data->mwatts;
++
++ return rc;
+}
+--- linux-2.6.18.8/drivers/net/sfc/boards.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/boards.h 2008-05-19 00:33:28.829807910 +0300
+@@ -0,0 +1,51 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static ssize_t adomain_read(struct file * file, char __user * buf,
-+ size_t count, loff_t * offset)
-+{
-+ char * tmpbuf;
-+ size_t len;
-+ int i;
-+ ssize_t retval;
++#ifndef EFX_BOARDS_H
++#define EFX_BOARDS_H
+
-+ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
-+ return -ENOMEM;
++/* Board IDs (must fit in 8 bits). Note that 0 must never be assigned because
++ * on early boards it means there is no revision info. Board types pre 400x
++ * are not covered here, but this is not a problem because:
++ * - the early Falcon boards (FPGA, 401, 403) don't have any extra H/W we
++ * need care about and aren't being updated.
++ */
++enum efx_board_type {
++ EFX_BOARD_INVALID = 0, /* Early boards do not have board rev. info. */
++ EFX_BOARD_SFE4001 = 1,
++ EFX_BOARD_SFE4002 = 2,
++ EFX_BOARD_SFE4003 = 3,
++ EFX_BOARD_SFE4005 = 4,
++ /* Insert new types before here */
++ EFX_BOARD_MAX
++};
+
-+ mutex_lock(&adom_mutex);
++extern int efx_set_board_info(struct efx_nic *efx, u16 revision_info);
+
-+ len = 0;
-+ for (i = 0; i < adomains; i++)
-+ len += snprintf(tmpbuf + len,
-+ len < TMPBUFSIZE ? TMPBUFSIZE - len : 0,
-+ "%u ", active_domains[i]);
-+ WARN_ON(len > TMPBUFSIZE);
-+ if (len != 0 && len <= TMPBUFSIZE)
-+ tmpbuf[len-1] = '\n';
++/* SFE4001 (10GBASE-T) */
++extern int sfe4001_poweron(struct efx_nic *efx);
++extern void sfe4001_poweroff(struct efx_nic *efx);
+
-+ mutex_unlock(&adom_mutex);
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/config.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/config.h 2008-05-19 00:33:28.829807910 +0300
+@@ -0,0 +1 @@
++/* SFC config options can go here */
+--- linux-2.6.18.8/drivers/net/sfc/debugfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/debugfs.c 2008-05-19 00:33:28.829807910 +0300
+@@ -0,0 +1,924 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ retval = simple_read_from_buffer(buf, count, offset, tmpbuf, len);
++#include <linux/module.h>
++#include <linux/pci.h>
++/* For out-of-tree builds we always need procfs, if only for a compatibility
++ * symlink.
++ */
++#include <linux/proc_fs.h>
++#include <linux/dcache.h>
++#include <linux/seq_file.h>
++#include "net_driver.h"
++#include "efx.h"
++#include "debugfs.h"
++#include "falcon.h"
+
-+ kfree(tmpbuf);
-+ return retval;
++/* EFX_USE_DEBUGFS is defined by kernel_compat.h so we can't decide whether to
++ * include this earlier.
++ */
++#ifdef EFX_USE_DEBUGFS
++#include <linux/debugfs.h>
++#endif
++
++#ifndef PRIu64
++# if (BITS_PER_LONG == 64)
++# define PRIu64 "lu"
++# else
++# define PRIu64 "llu"
++# endif
++#endif
++
++#ifndef EFX_USE_DEBUGFS
++
++static void efx_debugfs_remove(struct proc_dir_entry *entry)
++{
++ if (entry)
++ remove_proc_entry(entry->name, entry->parent);
+}
++#define debugfs_remove efx_debugfs_remove
+
++#define debugfs_create_dir proc_mkdir
++#define debugfs_create_symlink proc_symlink
+
-+static struct file_operations active_domain_ops = {
-+ .read = adomain_read,
-+ .write = adomain_write,
++#endif /* !EFX_USE_DEBUGFS */
++
++/* Parameter definition bound to a structure - each file has one of these */
++struct efx_debugfs_bound_param {
++ const struct efx_debugfs_parameter *param;
++ void *structure;
+};
+
-+static unsigned int pdomains = 0;
-+static int passive_domains[MAX_OPROF_DOMAINS];
-+static DEFINE_MUTEX(pdom_mutex);
+
-+static ssize_t pdomain_write(struct file * file, char const __user * buf,
-+ size_t count, loff_t * offset)
++/* Maximum length for a name component or symlink target */
++#define EFX_DEBUGFS_NAME_LEN 32
++
++
++/* Top-level debug directory ([/sys/kernel]/debug/sfc) */
++static struct dentry *efx_debug_root;
++
++/* "cards" directory ([/sys/kernel]/debug/sfc/cards) */
++static struct dentry *efx_debug_cards;
++
++
++/* Sequential file interface to bound parameters */
++
++#if defined(EFX_USE_DEBUGFS)
++
++static int efx_debugfs_seq_show(struct seq_file *file, void *v)
+{
-+ char *tmpbuf;
-+ char *startp, *endp;
-+ int i;
-+ unsigned long val;
-+ ssize_t retval = count;
-+
-+ if (*offset)
-+ return -EINVAL;
-+ if (count > TMPBUFSIZE - 1)
-+ return -EINVAL;
++ struct efx_debugfs_bound_param *binding =
++ (struct efx_debugfs_bound_param *)file->private;
+
-+ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
-+ return -ENOMEM;
++ return binding->param->reader(file,
++ binding->structure +
++ binding->param->offset);
++}
+
-+ if (copy_from_user(tmpbuf, buf, count)) {
-+ kfree(tmpbuf);
-+ return -EFAULT;
-+ }
-+ tmpbuf[count] = 0;
++static int efx_debugfs_open(struct inode *inode, struct file *file)
++{
++ return single_open(file, efx_debugfs_seq_show, inode->i_private);
++}
+
-+ mutex_lock(&pdom_mutex);
++#else /* EFX_NOT_UPSTREAM && !EFX_USE_DEBUGFS */
+
-+ startp = tmpbuf;
-+ /* Parse one more than MAX_OPROF_DOMAINS, for easy error checking */
-+ for (i = 0; i <= MAX_OPROF_DOMAINS; i++) {
-+ val = simple_strtoul(startp, &endp, 0);
-+ if (endp == startp)
-+ break;
-+ while (ispunct(*endp) || isspace(*endp))
-+ endp++;
-+ passive_domains[i] = val;
-+ if (passive_domains[i] != val)
-+ /* Overflow, force error below */
-+ i = MAX_OPROF_DOMAINS + 1;
-+ startp = endp;
++static int efx_debugfs_seq_show(struct seq_file *file, void *v)
++{
++ struct proc_dir_entry *entry = (struct proc_dir_entry *)file->private;
++ struct efx_debugfs_parameter *param =
++ (struct efx_debugfs_parameter *)entry->data;
++ void *structure = (void *)entry->read_proc;
++
++ if (!structure)
++ return -EIO;
++
++ return param->reader(file, structure + param->offset);
++}
++
++static int efx_debugfs_open(struct inode *inode, struct file *file)
++{
++ return single_open(file, efx_debugfs_seq_show, PROC_I(inode)->pde);
++}
++
++#endif /* !EFX_NOT_UPSTREAM || EFX_USE_DEBUGFS */
++
++
++static struct file_operations efx_debugfs_file_ops = {
++ .owner = THIS_MODULE,
++ .open = efx_debugfs_open,
++ .read = seq_read,
++ .llseek = seq_lseek,
++ .release = seq_release
++};
++
++
++#if defined(EFX_USE_DEBUGFS)
++
++/**
++ * efx_fini_debugfs_child - remove a named child of a debugfs directory
++ * @dir: Directory
++ * @name: Name of child
++ *
++ * This removes the named child from the directory, if it exists.
++ */
++void efx_fini_debugfs_child(struct dentry *dir, const char *name)
++{
++ struct qstr child_name;
++ struct dentry *child;
++
++ child_name.len = strlen(name);
++ child_name.name = name;
++ child_name.hash = full_name_hash(child_name.name, child_name.len);
++ child = d_lookup(dir, &child_name);
++ if (child) {
++ /* If it's a "regular" file, free its parameter binding */
++ if (S_ISREG(child->d_inode->i_mode))
++ kfree(child->d_inode->i_private);
++ debugfs_remove(child);
++ dput(child);
+ }
-+ /* Force error on trailing junk */
-+ pdomains = *startp ? MAX_OPROF_DOMAINS + 1 : i;
++}
+
-+ kfree(tmpbuf);
++#else /* EFX_NOT_UPSTREAM && !EFX_USE_DEBUGFS */
+
-+ if (pdomains > MAX_OPROF_DOMAINS
-+ || oprofile_set_passive(passive_domains, pdomains)) {
-+ pdomains = 0;
-+ retval = -EINVAL;
++void efx_fini_debugfs_child(struct proc_dir_entry *dir, const char *name)
++{
++ remove_proc_entry(name, dir);
++}
++
++#endif /* !EFX_NOT_UPSTREAM || EFX_USE_DEBUGFS */
++
++/*
++ * Remove a debugfs directory.
++ *
++ * This removes the named parameter-files and sym-links from the
++ * directory, and the directory itself. It does not do any recursion
++ * to subdirectories.
++ */
++static void efx_fini_debugfs_dir(struct dentry *dir,
++ struct efx_debugfs_parameter *params,
++ const char *const *symlink_names)
++{
++ if (!dir)
++ return;
++
++ while (params->name) {
++ efx_fini_debugfs_child(dir, params->name);
++ params++;
+ }
++ while (symlink_names && *symlink_names) {
++ efx_fini_debugfs_child(dir, *symlink_names);
++ symlink_names++;
++ }
++ debugfs_remove(dir);
++}
+
-+ mutex_unlock(&pdom_mutex);
-+ return retval;
++/* Functions for printing various types of parameter. */
++
++int efx_debugfs_read_uint(struct seq_file *file, void *data)
++{
++ return seq_printf(file, "%#x\n", *(unsigned int *)data);
+}
+
-+static ssize_t pdomain_read(struct file * file, char __user * buf,
-+ size_t count, loff_t * offset)
++int efx_debugfs_read_int(struct seq_file *file, void *data)
+{
-+ char * tmpbuf;
-+ size_t len;
-+ int i;
-+ ssize_t retval;
++ return seq_printf(file, "%d\n", *(int *)data);
++}
+
-+ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
-+ return -ENOMEM;
++int efx_debugfs_read_atomic(struct seq_file *file, void *data)
++{
++ unsigned int value = atomic_read((atomic_t *) data);
+
-+ mutex_lock(&pdom_mutex);
++ return seq_printf(file, "%#x\n", value);
++}
+
-+ len = 0;
-+ for (i = 0; i < pdomains; i++)
-+ len += snprintf(tmpbuf + len,
-+ len < TMPBUFSIZE ? TMPBUFSIZE - len : 0,
-+ "%u ", passive_domains[i]);
-+ WARN_ON(len > TMPBUFSIZE);
-+ if (len != 0 && len <= TMPBUFSIZE)
-+ tmpbuf[len-1] = '\n';
++int efx_debugfs_read_dword(struct seq_file *file, void *data)
++{
++ unsigned int value = EFX_DWORD_FIELD(*(efx_dword_t *) data,
++ EFX_DWORD_0);
+
-+ mutex_unlock(&pdom_mutex);
++ return seq_printf(file, "%#x\n", value);
++}
+
-+ retval = simple_read_from_buffer(buf, count, offset, tmpbuf, len);
++static int efx_debugfs_read_int_mode(struct seq_file *file, void *data)
++{
++ unsigned int value = *(enum efx_int_mode *) data;
+
-+ kfree(tmpbuf);
-+ return retval;
++ return seq_printf(file, "%d => %s\n", value,
++ STRING_TABLE_LOOKUP(value, efx_interrupt_mode));
+}
+
-+static struct file_operations passive_domain_ops = {
-+ .read = pdomain_read,
-+ .write = pdomain_write,
-+};
++#define EFX_INT_MODE_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ enum efx_int_mode, efx_debugfs_read_int_mode)
+
- void oprofile_create_files(struct super_block * sb, struct dentry * root)
- {
- oprofilefs_create_file(sb, root, "enable", &enable_fops);
- oprofilefs_create_file_perm(sb, root, "dump", &dump_fops, 0666);
-+ oprofilefs_create_file(sb, root, "active_domains", &active_domain_ops);
-+ oprofilefs_create_file(sb, root, "passive_domains", &passive_domain_ops);
- oprofilefs_create_file(sb, root, "buffer", &event_buffer_fops);
- oprofilefs_create_ulong(sb, root, "buffer_size", &fs_buffer_size);
- oprofilefs_create_ulong(sb, root, "buffer_watershed", &fs_buffer_watershed);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/pci/bus.c linux-2.6.18-xen.hg/drivers/pci/bus.c
---- linux-2.6.18/drivers/pci/bus.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/pci/bus.c 2007-12-23 11:15:23.454045623 +0100
-@@ -17,6 +17,8 @@
-
- #include "pci.h"
-
-+extern int pci_mem_align;
++static int efx_debugfs_read_loop_mode(struct seq_file *file, void *data)
++{
++ unsigned int value = *(enum efx_loopback_mode *)data;
+
- /**
- * pci_bus_alloc_resource - allocate a resource from a parent bus
- * @bus: PCI bus
-@@ -44,6 +46,11 @@
-
- type_mask |= IORESOURCE_IO | IORESOURCE_MEM;
-
-+ /* If the boot parameter 'pci-mem-align' was specified then we need to
-+ align the memory addresses, at page size alignment. */
-+ if (pci_mem_align && (align < (PAGE_SIZE-1)))
-+ align = PAGE_SIZE - 1;
++ return seq_printf(file, "%d => %s\n", value,
++ STRING_TABLE_LOOKUP(value, efx_loopback_mode));
++}
+
- for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
- struct resource *r = bus->resource[i];
- if (!r)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/pci/Kconfig linux-2.6.18-xen.hg/drivers/pci/Kconfig
---- linux-2.6.18/drivers/pci/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/pci/Kconfig 2007-12-23 11:15:23.454045623 +0100
-@@ -5,6 +5,7 @@
- bool "Message Signaled Interrupts (MSI and MSI-X)"
- depends on PCI
- depends on (X86_LOCAL_APIC && X86_IO_APIC) || IA64
-+ depends on !XEN
- help
- This allows device drivers to enable MSI (Message Signaled
- Interrupts). Message Signaled Interrupts enable a device to
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/pci/quirks.c linux-2.6.18-xen.hg/drivers/pci/quirks.c
---- linux-2.6.18/drivers/pci/quirks.c 2007-12-23 11:26:51.163461987 +0100
-+++ linux-2.6.18-xen.hg/drivers/pci/quirks.c 2007-12-23 11:15:23.664056653 +0100
-@@ -23,6 +23,40 @@
- #include <linux/acpi.h>
- #include "pci.h"
-
-+/* A global flag which signals if we should page-align PCI mem windows. */
-+int pci_mem_align = 0;
++#define EFX_LOOPBACK_MODE_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ enum efx_loopback_mode, efx_debugfs_read_loop_mode)
+
-+static int __init set_pci_mem_align(char *str)
++static int efx_debugfs_read_phy_type(struct seq_file *file, void *data)
+{
-+ pci_mem_align = 1;
-+ return 1;
++ unsigned int value = *(enum phy_type *) data;
++
++ return seq_printf(file, "%d => %s\n", value,
++ STRING_TABLE_LOOKUP(value, efx_phy_type));
+}
-+__setup("pci-mem-align", set_pci_mem_align);
+
-+/* This quirk function enables us to force all memory resources which are
-+ * assigned to PCI devices, to be page-aligned.
++#define EFX_PHY_TYPE_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ enum phy_type, efx_debugfs_read_phy_type)
++
++int efx_debugfs_read_string(struct seq_file *file, void *data)
++{
++ return seq_puts(file, (const char *)data);
++}
++
++
++/**
++ * efx_init_debugfs_files - create parameter-files in a debugfs directory
++ * @parent: Containing directory
++ * @params: Pointer to zero-terminated parameter definition array
++ * @structure: Structure containing parameters
++ *
++ * Add parameter-files to the given debugfs directory. Return a
++ * negative error code or 0 on success.
+ */
-+static void __devinit quirk_align_mem_resources(struct pci_dev *dev)
++static int efx_init_debugfs_files(struct dentry *parent,
++ struct efx_debugfs_parameter *params,
++ void *structure)
+{
-+ int i;
-+ struct resource *r;
-+ resource_size_t old_start;
++ struct efx_debugfs_parameter *param = params;
+
-+ if (!pci_mem_align)
-+ return;
++ while (param->name) {
++ struct dentry *entry;
++#if defined(EFX_USE_DEBUGFS)
++ struct efx_debugfs_bound_param *binding;
+
-+ for (i=0; i < DEVICE_COUNT_RESOURCE; i++) {
-+ r = &dev->resource[i];
-+ if ((r == NULL) || !(r->flags & IORESOURCE_MEM))
-+ continue;
++ binding = kmalloc(sizeof(*binding), GFP_KERNEL);
++ if (!binding)
++ goto err;
++ binding->param = param;
++ binding->structure = structure;
+
-+ old_start = r->start;
-+ r->start = (r->start + PAGE_SIZE - 1) & ~(PAGE_SIZE - 1);
-+ r->end = r->end - (old_start - r->start);
++ entry = debugfs_create_file(param->name, S_IRUGO, parent,
++ binding, &efx_debugfs_file_ops);
++ if (!entry) {
++ kfree(binding);
++ goto err;
++ }
++#else
++ entry = create_proc_entry(param->name, S_IRUGO, parent);
++ if (!entry)
++ goto err;
++ /*
++ * We have no good way to free a binding created here.
++ * However, once we install our file_operations the
++ * read_proc pointer becomes redundant and we can
++ * abuse it as a structure pointer.
++ */
++ entry->data = param;
++ entry->read_proc = NULL;
++ smp_wmb();
++ entry->proc_fops = &efx_debugfs_file_ops;
++ smp_wmb();
++ entry->read_proc = (read_proc_t *) structure;
++#endif
++
++ param++;
++ }
++
++ return 0;
++
++ err:
++ while (param != params) {
++ param--;
++ efx_fini_debugfs_child(parent, param->name);
+ }
++ return -ENOMEM;
+}
-+DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, quirk_align_mem_resources);
+
- /* The Mellanox Tavor device gives false positive parity errors
- * Mark this device with a broken_parity_status, to allow
- * PCI scanning code to "skip" this now blacklisted device.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/pnp/manager.c linux-2.6.18-xen.hg/drivers/pnp/manager.c
---- linux-2.6.18/drivers/pnp/manager.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/pnp/manager.c 2007-12-23 11:15:24.564103912 +0100
-@@ -168,7 +168,7 @@
- return 0;
- }
-
--static int pnp_assign_dma(struct pnp_dev *dev, struct pnp_dma *rule, int idx)
-+static void pnp_assign_dma(struct pnp_dev *dev, struct pnp_dma *rule, int idx)
- {
- resource_size_t *start, *end;
- unsigned long *flags;
-@@ -179,18 +179,14 @@
- 1, 3, 5, 6, 7, 0, 2, 4
- };
-
-- if (!dev || !rule)
-- return -EINVAL;
--
- if (idx >= PNP_MAX_DMA) {
- pnp_err("More than 2 dmas is incompatible with pnp specifications.");
-- /* pretend we were successful so at least the manager won't try again */
-- return 1;
-+ return;
- }
-
- /* check if this resource has been manually set, if so skip */
- if (!(dev->res.dma_resource[idx].flags & IORESOURCE_AUTO))
-- return 1;
-+ return;
-
- start = &dev->res.dma_resource[idx].start;
- end = &dev->res.dma_resource[idx].end;
-@@ -200,19 +196,17 @@
- *flags |= rule->flags | IORESOURCE_DMA;
- *flags &= ~IORESOURCE_UNSET;
-
-- if (!rule->map) {
-- *flags |= IORESOURCE_DISABLED;
-- return 1; /* skip disabled resource requests */
-- }
--
- for (i = 0; i < 8; i++) {
- if(rule->map & (1<<xtab[i])) {
- *start = *end = xtab[i];
- if(pnp_check_dma(dev, idx))
-- return 1;
-+ return;
- }
- }
-- return 0;
-+#ifdef MAX_DMA_CHANNELS
-+ *start = *end = MAX_DMA_CHANNELS;
-+#endif
-+ *flags |= IORESOURCE_UNSET | IORESOURCE_DISABLED;
- }
-
- /**
-@@ -331,8 +325,7 @@
- irq = irq->next;
- }
- while (dma) {
-- if (!pnp_assign_dma(dev, dma, ndma))
-- goto fail;
-+ pnp_assign_dma(dev, dma, ndma);
- ndma++;
- dma = dma->next;
- }
-@@ -367,8 +360,7 @@
- irq = irq->next;
- }
- while (dma) {
-- if (!pnp_assign_dma(dev, dma, ndma))
-- goto fail;
-+ pnp_assign_dma(dev, dma, ndma);
- ndma++;
- dma = dma->next;
- }
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/scsi/ahci.c linux-2.6.18-xen.hg/drivers/scsi/ahci.c
---- linux-2.6.18/drivers/scsi/ahci.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/scsi/ahci.c 2007-12-23 11:15:25.537488366 +0100
-@@ -317,6 +317,28 @@
- board_ahci }, /* ICH8M */
- { PCI_VENDOR_ID_INTEL, 0x282a, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
- board_ahci }, /* ICH8M */
-+ { PCI_VENDOR_ID_INTEL, 0x2922, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x2923, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x2924, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x2925, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x2927, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x2929, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9M */
-+ { PCI_VENDOR_ID_INTEL, 0x292a, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9M */
-+ { PCI_VENDOR_ID_INTEL, 0x292b, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9M */
-+ { PCI_VENDOR_ID_INTEL, 0x292f, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9M */
-+ { PCI_VENDOR_ID_INTEL, 0x294d, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9 */
-+ { PCI_VENDOR_ID_INTEL, 0x294e, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-+ board_ahci }, /* ICH9M */
-
- /* JMicron */
- { 0x197b, 0x2360, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/scsi/ata_piix.c linux-2.6.18-xen.hg/drivers/scsi/ata_piix.c
---- linux-2.6.18/drivers/scsi/ata_piix.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/scsi/ata_piix.c 2007-12-23 11:15:26.444202639 +0100
-@@ -125,6 +125,7 @@
- ich6m_sata_ahci = 6,
- ich7m_sata_ahci = 7,
- ich8_sata_ahci = 8,
-+ ich9_sata_ahci = 9,
-
- /* constants for mapping table */
- P0 = 0, /* port 0 */
-@@ -198,6 +199,18 @@
- { 0x8086, 0x2825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
- /* Mobile SATA Controller IDE (ICH8M, ditto) */
- { 0x8086, 0x2828, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
-+ /* SATA Controller 1 IDE (ICH9) */
-+ { 0x8086, 0x2920, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-+ /* SATA Controller 1 IDE (ICH9) */
-+ { 0x8086, 0x2921, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-+ /* SATA Controller 2 IDE (ICH9) */
-+ { 0x8086, 0x2926, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-+ /* Mobile SATA Controller 1 IDE (ICH9M) */
-+ { 0x8086, 0x2928, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-+ /* Mobile SATA Controller 2 IDE (ICH9M) */
-+ { 0x8086, 0x292d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-+ /* Mobile SATA Controller 2 IDE (ICH9M) */
-+ { 0x8086, 0x292e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
-
- { } /* terminate list */
- };
-@@ -361,9 +374,22 @@
- .present_shift = 8,
- .map = {
- /* PM PS SM SS MAP */
-- { P0, NA, P1, NA }, /* 00b (hardwired) */
-+ { P0, P2, P1, P3 }, /* 00b (hardwired when in AHCI) */
- { RV, RV, RV, RV },
-- { RV, RV, RV, RV }, /* 10b (never) */
-+ { IDE, IDE, NA, NA }, /* 10b (IDE mode) */
-+ { RV, RV, RV, RV },
-+ },
-+};
++/**
++ * efx_init_debugfs_netdev - create debugfs sym-links for net device
++ * @net_dev: Net device
++ *
++ * Create sym-links named after @net_dev to the debugfs directories for
++ * the corresponding NIC and port. Return a negative error code or 0 on
++ * success. The sym-links must be cleaned up using
++ * efx_fini_debugfs_netdev().
++ */
++int efx_init_debugfs_netdev(struct net_device *net_dev)
++{
++ struct efx_nic *efx = net_dev->priv;
++ char name[EFX_DEBUGFS_NAME_LEN];
++ char target[EFX_DEBUGFS_NAME_LEN];
++ size_t len;
+
-+static const struct piix_map_db ich9_map_db = {
-+ .mask = 0x3,
-+ .port_enable = 0x3,
-+ .present_shift = 8,
-+ .map = {
-+ /* PM PS SM SS MAP */
-+ { P0, P2, P1, P3 }, /* 00b (hardwired when in AHCI) */
-+ { RV, RV, RV, RV },
-+ { IDE, IDE, NA, NA }, /* 10b (IDE mode) */
- { RV, RV, RV, RV },
- },
- };
-@@ -376,6 +402,7 @@
- [ich6m_sata_ahci] = &ich6m_map_db,
- [ich7m_sata_ahci] = &ich7m_map_db,
- [ich8_sata_ahci] = &ich8_map_db,
-+ [ich9_sata_ahci] = &ich9_map_db,
- };
-
- static struct ata_port_info piix_port_info[] = {
-@@ -487,6 +514,18 @@
- .udma_mask = 0x7f, /* udma0-6 */
- .port_ops = &piix_sata_ops,
- },
++ if (snprintf(name, sizeof(name), "nic_%s", net_dev->name) >=
++ sizeof(name))
++ return -ENAMETOOLONG;
++ if (snprintf(target, sizeof(target), "cards/%s", pci_name(efx->pci_dev))
++ >= sizeof(target))
++ return -ENAMETOOLONG;
++ efx->debug_symlink = debugfs_create_symlink(name,
++ efx_debug_root, target);
++ if (!efx->debug_symlink)
++ return -ENOMEM;
+
-+ /* ich9_sata_ahci */
-+ {
-+ .sht = &piix_sht,
-+ .host_flags = ATA_FLAG_SATA |
-+ PIIX_FLAG_CHECKINTR | PIIX_FLAG_SCR |
-+ PIIX_FLAG_AHCI,
-+ .pio_mask = 0x1f, /* pio0-4 */
-+ .mwdma_mask = 0x07, /* mwdma0-2 */
-+ .udma_mask = 0x7f, /* udma0-6 */
-+ .port_ops = &piix_sata_ops,
-+ },
- };
-
- static struct pci_bits piix_enable_bits[] = {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/serial/Kconfig linux-2.6.18-xen.hg/drivers/serial/Kconfig
---- linux-2.6.18/drivers/serial/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/serial/Kconfig 2007-12-23 11:15:29.281018269 +0100
-@@ -11,6 +11,7 @@
- config SERIAL_8250
- tristate "8250/16550 and compatible serial support"
- depends on (BROKEN || !SPARC)
-+ depends on !XEN_DISABLE_SERIAL
- select SERIAL_CORE
- ---help---
- This selects whether you want to include the driver for the standard
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/video/console/Kconfig linux-2.6.18-xen.hg/drivers/video/console/Kconfig
---- linux-2.6.18/drivers/video/console/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/drivers/video/console/Kconfig 2007-12-23 11:15:32.451184741 +0100
-@@ -53,6 +53,7 @@
- config VIDEO_SELECT
- bool "Video mode selection support"
- depends on X86 && VGA_CONSOLE
-+ depends on !XEN
- ---help---
- This enables support for text mode selection on kernel startup. If
- you want to take advantage of some high-resolution text mode your
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/balloon/balloon.c linux-2.6.18-xen.hg/drivers/xen/balloon/balloon.c
---- linux-2.6.18/drivers/xen/balloon/balloon.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/balloon/balloon.c 2007-12-23 11:15:33.537908474 +0100
-@@ -0,0 +1,675 @@
-+/******************************************************************************
-+ * balloon.c
++ if (snprintf(name, sizeof(name), "if_%s", net_dev->name) >=
++ sizeof(name))
++ return -ENAMETOOLONG;
++ len = snprintf(target, sizeof(target),
++ "cards/%s/port0", pci_name(efx->pci_dev));
++ if (len >= sizeof(target))
++ return -ENAMETOOLONG;
++ efx->debug_port_symlink = debugfs_create_symlink(name,
++ efx_debug_root,
++ target);
++ if (!efx->debug_port_symlink)
++ return -ENOMEM;
++
++ return 0;
++}
++
++/**
++ * efx_fini_debugfs_netdev - remove debugfs sym-links for net device
++ * @net_dev: Net device
+ *
-+ * Xen balloon driver - enables returning/claiming memory to/from Xen.
++ * Remove sym-links created for @net_dev by efx_init_debugfs_netdev().
++ */
++void efx_fini_debugfs_netdev(struct net_device *net_dev)
++{
++ struct efx_nic *efx = net_dev->priv;
++
++ debugfs_remove(efx->debug_port_symlink);
++ efx->debug_port_symlink = NULL;
++ debugfs_remove(efx->debug_symlink);
++ efx->debug_symlink = NULL;
++}
++
++/* Per-port parameters */
++static struct efx_debugfs_parameter efx_debugfs_port_parameters[] = {
++ EFX_NAMED_PARAMETER(enabled, struct efx_nic, port_enabled,
++ int, efx_debugfs_read_int),
++ EFX_INT_PARAMETER(struct efx_nic, net_dev_registered),
++ EFX_INT_PARAMETER(struct efx_nic, rx_checksum_enabled),
++ EFX_ATOMIC_PARAMETER(struct efx_nic, netif_stop_count),
++ EFX_INT_PARAMETER(struct efx_nic, link_up),
++ EFX_UINT_PARAMETER(struct efx_nic, link_options),
++ EFX_INT_PARAMETER(struct efx_nic, promiscuous),
++ EFX_UINT_PARAMETER(struct efx_nic, loopback_modes),
++ EFX_LOOPBACK_MODE_PARAMETER(struct efx_nic, loopback_mode),
++ EFX_PHY_TYPE_PARAMETER(struct efx_nic, phy_type),
++ EFX_NAMED_PARAMETER(phy_id, struct efx_nic, mii.phy_id,
++ int, efx_debugfs_read_int),
++ EFX_UINT_PARAMETER(struct efx_nic, n_link_state_changes),
++ {NULL},
++};
++
++/**
++ * efx_init_debugfs_port - create debugfs directory for port
++ * @efx: Efx NIC
+ *
-+ * Copyright (c) 2003, B Dragovic
-+ * Copyright (c) 2003-2004, M Williamson, K Fraser
-+ * Copyright (c) 2005 Dan M. Smith, IBM Corporation
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * Create a debugfs directory containing parameter-files for @efx.
++ * Return a negative error code or 0 on success. The directory must be
++ * cleaned up using efx_fini_debugfs_port().
+ */
++int efx_init_debugfs_port(struct efx_nic *efx)
++{
++ int rc;
+
-+#include <linux/kernel.h>
-+#include <linux/module.h>
-+#include <linux/sched.h>
-+#include <linux/errno.h>
-+#include <linux/mm.h>
-+#include <linux/mman.h>
-+#include <linux/smp_lock.h>
-+#include <linux/pagemap.h>
-+#include <linux/bootmem.h>
-+#include <linux/highmem.h>
-+#include <linux/vmalloc.h>
-+#include <linux/mutex.h>
-+#include <xen/xen_proc.h>
-+#include <asm/hypervisor.h>
-+#include <xen/balloon.h>
-+#include <xen/interface/memory.h>
-+#include <asm/maddr.h>
-+#include <asm/page.h>
-+#include <asm/pgalloc.h>
-+#include <asm/pgtable.h>
-+#include <asm/uaccess.h>
-+#include <asm/tlb.h>
-+#include <linux/highmem.h>
-+#include <linux/list.h>
-+#include <xen/xenbus.h>
-+#include "common.h"
++ /* Create directory */
++ efx->debug_port_dir = debugfs_create_dir("port0", efx->debug_dir);
++ if (!efx->debug_port_dir)
++ return -ENOMEM;
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
++ /* Create files */
++ rc = efx_init_debugfs_files(efx->debug_port_dir,
++ efx_debugfs_port_parameters,
++ (void *)efx);
++ if (rc)
++ efx_fini_debugfs_port(efx);
+
-+#ifdef CONFIG_PROC_FS
-+static struct proc_dir_entry *balloon_pde;
-+#endif
++ return rc;
++}
+
-+static DEFINE_MUTEX(balloon_mutex);
++/**
++ * efx_fini_debugfs_port - remove debugfs directory for port
++ * @efx: Efx NIC
++ *
++ * Remove directory created for @efx by efx_init_debugfs_port().
++ */
++void efx_fini_debugfs_port(struct efx_nic *efx)
++{
++ efx_fini_debugfs_dir(efx->debug_port_dir,
++ efx_debugfs_port_parameters, NULL);
++ efx->debug_port_dir = NULL;
++}
+
-+/*
-+ * Protects atomic reservation decrease/increase against concurrent increases.
-+ * Also protects non-atomic updates of current_pages and driver_pages, and
-+ * balloon lists.
++/**
++ * efx_extend_debugfs_port - add parameter-files to directory for port
++ * @efx: Efx NIC
++ * @structure: Structure containing parameters
++ * @params: Pointer to zero-terminated parameter definition array
++ *
++ * Add parameter-files to the debugfs directory for @efx. Return
++ * a negative error code or 0 on success. This is intended for
++ * PHY-specific parameters. The files must be cleaned up using
++ * efx_trim_debugfs_port().
+ */
-+DEFINE_SPINLOCK(balloon_lock);
++int efx_extend_debugfs_port(struct efx_nic *efx,
++ void *structure,
++ struct efx_debugfs_parameter *params)
++{
++ return efx_init_debugfs_files(efx->debug_port_dir, params, structure);
++}
+
-+struct balloon_stats balloon_stats;
++/**
++ * efx_trim_debugfs_port - remove parameter-files from directory for port
++ * @efx: Efx NIC
++ * @params: Pointer to zero-terminated parameter definition array
++ *
++ * Remove parameter-files previously added to the debugfs directory
++ * for @efx using efx_extend_debugfs_port().
++ */
++void efx_trim_debugfs_port(struct efx_nic *efx,
++ struct efx_debugfs_parameter *params)
++{
++ struct dentry *dir = efx->debug_port_dir;
+
-+/* We increase/decrease in batches which fit in a page */
-+static unsigned long frame_list[PAGE_SIZE / sizeof(unsigned long)];
++ if (dir) {
++ struct efx_debugfs_parameter *field;
++ for (field = params; field->name; field++)
++ efx_fini_debugfs_child(dir, field->name);
++ }
++}
+
-+/* VM /proc information for memory */
-+extern unsigned long totalram_pages;
++/* Per-TX-queue parameters */
++static struct efx_debugfs_parameter efx_debugfs_tx_queue_parameters[] = {
++ EFX_UINT_PARAMETER(struct efx_tx_queue, insert_count),
++ EFX_UINT_PARAMETER(struct efx_tx_queue, write_count),
++ EFX_UINT_PARAMETER(struct efx_tx_queue, read_count),
++ EFX_INT_PARAMETER(struct efx_tx_queue, stopped),
++ {NULL},
++};
+
-+#ifndef MODULE
-+extern unsigned long totalhigh_pages;
-+#define inc_totalhigh_pages() (totalhigh_pages++)
-+#define dec_totalhigh_pages() (totalhigh_pages--)
-+#else
-+#define inc_totalhigh_pages() ((void)0)
-+#define dec_totalhigh_pages() ((void)0)
-+#endif
++static void efx_fini_debugfs_tx_queue(struct efx_tx_queue *tx_queue);
+
-+/* List of ballooned pages, threaded through the mem_map array. */
-+static LIST_HEAD(ballooned_pages);
++/**
++ * efx_init_debugfs_tx_queue - create debugfs directory for TX queue
++ * @tx_queue: Efx TX queue
++ *
++ * Create a debugfs directory containing parameter-files for @tx_queue.
++ * Return a negative error code or 0 on success. The directory must be
++ * cleaned up using efx_fini_debugfs_tx_queue().
++ */
++static int efx_init_debugfs_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ char name[EFX_DEBUGFS_NAME_LEN];
++ char target[EFX_DEBUGFS_NAME_LEN];
++ int rc;
+
-+/* Main work function, always executed in process context. */
-+static void balloon_process(void *unused);
-+static DECLARE_WORK(balloon_worker, balloon_process, NULL);
-+static struct timer_list balloon_timer;
++ /* Create directory */
++ if (snprintf(name, sizeof(name), EFX_TX_QUEUE_NAME(tx_queue))
++ >= sizeof(name))
++ goto err_len;
++ tx_queue->debug_dir = debugfs_create_dir(name,
++ tx_queue->efx->debug_dir);
++ if (!tx_queue->debug_dir)
++ goto err_mem;
++
++ /* Create files */
++ rc = efx_init_debugfs_files(tx_queue->debug_dir,
++ efx_debugfs_tx_queue_parameters,
++ (void *)tx_queue);
++ if (rc)
++ goto err;
+
-+/* When ballooning out (allocating memory to return to Xen) we don't really
-+ want the kernel to try too hard since that can trigger the oom killer. */
-+#define GFP_BALLOON \
-+ (GFP_HIGHUSER | __GFP_NOWARN | __GFP_NORETRY | __GFP_NOMEMALLOC)
++ /* Create symlink to channel */
++ if (snprintf(target, sizeof(target),
++ "../" EFX_CHANNEL_NAME(tx_queue->channel)) >=
++ sizeof(target))
++ goto err_len;
++ if (!debugfs_create_symlink("channel", tx_queue->debug_dir, target))
++ goto err_mem;
+
-+#define PAGE_TO_LIST(p) (&(p)->lru)
-+#define LIST_TO_PAGE(l) list_entry((l), struct page, lru)
-+#define UNLIST_PAGE(p) \
-+ do { \
-+ list_del(PAGE_TO_LIST(p)); \
-+ PAGE_TO_LIST(p)->next = NULL; \
-+ PAGE_TO_LIST(p)->prev = NULL; \
-+ } while(0)
++ /* Create symlink to port */
++ if (!debugfs_create_symlink("port", tx_queue->debug_dir, "../port0"))
++ goto err_mem;
+
-+#define IPRINTK(fmt, args...) \
-+ printk(KERN_INFO "xen_mem: " fmt, ##args)
-+#define WPRINTK(fmt, args...) \
-+ printk(KERN_WARNING "xen_mem: " fmt, ##args)
++ return 0;
+
-+/* balloon_append: add the given page to the balloon. */
-+static void balloon_append(struct page *page)
++ err_len:
++ rc = -ENAMETOOLONG;
++ goto err;
++ err_mem:
++ rc = -ENOMEM;
++ err:
++ efx_fini_debugfs_tx_queue(tx_queue);
++ return rc;
++}
++
++/**
++ * efx_fini_debugfs_tx_queue - remove debugfs directory for TX queue
++ * @tx_queue: Efx TX queue
++ *
++ * Remove directory created for @tx_queue by efx_init_debugfs_tx_queue().
++ */
++static void efx_fini_debugfs_tx_queue(struct efx_tx_queue *tx_queue)
+{
-+ /* Lowmem is re-populated first, so highmem pages go at list tail. */
-+ if (PageHighMem(page)) {
-+ list_add_tail(PAGE_TO_LIST(page), &ballooned_pages);
-+ bs.balloon_high++;
-+ dec_totalhigh_pages();
-+ } else {
-+ list_add(PAGE_TO_LIST(page), &ballooned_pages);
-+ bs.balloon_low++;
-+ }
++ static const char *const symlink_names[] = {
++ "channel", "port", NULL
++ };
++
++ efx_fini_debugfs_dir(tx_queue->debug_dir,
++ efx_debugfs_tx_queue_parameters, symlink_names);
++ tx_queue->debug_dir = NULL;
+}
+
-+/* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
-+static struct page *balloon_retrieve(void)
++/* Per-RX-queue parameters */
++static struct efx_debugfs_parameter efx_debugfs_rx_queue_parameters[] = {
++ EFX_INT_PARAMETER(struct efx_rx_queue, added_count),
++ EFX_INT_PARAMETER(struct efx_rx_queue, removed_count),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, max_fill),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, fast_fill_trigger),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, fast_fill_limit),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, min_fill),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, min_overfill),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, alloc_page_count),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, alloc_skb_count),
++ EFX_UINT_PARAMETER(struct efx_rx_queue, slow_fill_count),
++ {NULL},
++};
++
++static void efx_fini_debugfs_rx_queue(struct efx_rx_queue *rx_queue);
++
++/**
++ * efx_init_debugfs_rx_queue - create debugfs directory for RX queue
++ * @rx_queue: Efx RX queue
++ *
++ * Create a debugfs directory containing parameter-files for @rx_queue.
++ * Return a negative error code or 0 on success. The directory must be
++ * cleaned up using efx_fini_debugfs_rx_queue().
++ */
++static int efx_init_debugfs_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ struct page *page;
++ char name[EFX_DEBUGFS_NAME_LEN];
++ char target[EFX_DEBUGFS_NAME_LEN];
++ int rc;
+
-+ if (list_empty(&ballooned_pages))
-+ return NULL;
++ /* Create directory */
++ if (snprintf(name, sizeof(name), EFX_RX_QUEUE_NAME(rx_queue))
++ >= sizeof(name))
++ goto err_len;
++ rx_queue->debug_dir = debugfs_create_dir(name,
++ rx_queue->efx->debug_dir);
++ if (!rx_queue->debug_dir)
++ goto err_mem;
++
++ /* Create files */
++ rc = efx_init_debugfs_files(rx_queue->debug_dir,
++ efx_debugfs_rx_queue_parameters,
++ (void *)rx_queue);
++ if (rc)
++ goto err;
+
-+ page = LIST_TO_PAGE(ballooned_pages.next);
-+ UNLIST_PAGE(page);
++ /* Create symlink to channel */
++ if (snprintf(target, sizeof(target),
++ "../" EFX_CHANNEL_NAME(rx_queue->channel)) >=
++ sizeof(target))
++ goto err_len;
++ if (!debugfs_create_symlink("channel", rx_queue->debug_dir, target))
++ goto err_mem;
+
-+ if (PageHighMem(page)) {
-+ bs.balloon_high--;
-+ inc_totalhigh_pages();
-+ }
-+ else
-+ bs.balloon_low--;
++ return 0;
+
-+ return page;
++ err_len:
++ rc = -ENAMETOOLONG;
++ goto err;
++ err_mem:
++ rc = -ENOMEM;
++ err:
++ efx_fini_debugfs_rx_queue(rx_queue);
++ return rc;
+}
+
-+static struct page *balloon_first_page(void)
++/**
++ * efx_fini_debugfs_rx_queue - remove debugfs directory for RX queue
++ * @rx_queue: Efx RX queue
++ *
++ * Remove directory created for @rx_queue by efx_init_debugfs_rx_queue().
++ */
++static void efx_fini_debugfs_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ if (list_empty(&ballooned_pages))
-+ return NULL;
-+ return LIST_TO_PAGE(ballooned_pages.next);
-+}
++ const char *const symlink_names[] = {
++ "channel", NULL
++ };
+
-+static struct page *balloon_next_page(struct page *page)
-+{
-+ struct list_head *next = PAGE_TO_LIST(page)->next;
-+ if (next == &ballooned_pages)
-+ return NULL;
-+ return LIST_TO_PAGE(next);
-+}
++ efx_fini_debugfs_dir(rx_queue->debug_dir,
++ efx_debugfs_rx_queue_parameters, symlink_names);
++ rx_queue->debug_dir = NULL;
++}
++
++/* Per-channel parameters */
++static struct efx_debugfs_parameter efx_debugfs_channel_parameters[] = {
++ EFX_INT_PARAMETER(struct efx_channel, enabled),
++ EFX_INT_PARAMETER(struct efx_channel, irq),
++ EFX_UINT_PARAMETER(struct efx_channel, has_interrupt),
++ EFX_UINT_PARAMETER(struct efx_channel, irq_moderation),
++ EFX_UINT_PARAMETER(struct efx_channel, eventq_read_ptr),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_tobe_disc),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_ip_frag_err),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_ip_hdr_chksum_err),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_tcp_udp_chksum_err),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_frm_trunc),
++ EFX_UINT_PARAMETER(struct efx_channel, n_rx_overlength),
++ EFX_UINT_PARAMETER(struct efx_channel, n_skbuff_leaks),
++ EFX_INT_PARAMETER(struct efx_channel, rx_alloc_level),
++ EFX_INT_PARAMETER(struct efx_channel, rx_alloc_push_pages),
++ EFX_INT_PARAMETER(struct efx_channel, rx_alloc_pop_pages),
++ {NULL},
++};
++
++static void efx_fini_debugfs_channel(struct efx_channel *channel);
+
-+static void balloon_alarm(unsigned long unused)
++/**
++ * efx_init_debugfs_channel - create debugfs directory for channel
++ * @channel: Efx channel
++ *
++ * Create a debugfs directory containing parameter-files for @channel.
++ * Return a negative error code or 0 on success. The directory must be
++ * cleaned up using efx_fini_debugfs_channel().
++ */
++static int efx_init_debugfs_channel(struct efx_channel *channel)
+{
-+ schedule_work(&balloon_worker);
++ char name[EFX_DEBUGFS_NAME_LEN];
++ int rc;
++
++ /* Create directory */
++ if (snprintf(name, sizeof(name), EFX_CHANNEL_NAME(channel))
++ >= sizeof(name))
++ goto err_len;
++ channel->debug_dir = debugfs_create_dir(name, channel->efx->debug_dir);
++ if (!channel->debug_dir)
++ goto err_mem;
++
++ /* Create files */
++ rc = efx_init_debugfs_files(channel->debug_dir,
++ efx_debugfs_channel_parameters,
++ (void *)channel);
++ if (rc)
++ goto err;
++
++ return 0;
++
++ err_len:
++ rc = -ENAMETOOLONG;
++ goto err;
++ err_mem:
++ rc = -ENOMEM;
++ err:
++ efx_fini_debugfs_channel(channel);
++ return rc;
+}
+
-+static unsigned long current_target(void)
++/**
++ * efx_fini_debugfs_channel - remove debugfs directory for channel
++ * @channel: Efx channel
++ *
++ * Remove directory created for @channel by efx_init_debugfs_channel().
++ */
++static void efx_fini_debugfs_channel(struct efx_channel *channel)
+{
-+ unsigned long target = min(bs.target_pages, bs.hard_limit);
-+ if (target > (bs.current_pages + bs.balloon_low + bs.balloon_high))
-+ target = bs.current_pages + bs.balloon_low + bs.balloon_high;
-+ return target;
++ efx_fini_debugfs_dir(channel->debug_dir,
++ efx_debugfs_channel_parameters, NULL);
++ channel->debug_dir = NULL;
+}
+
-+static int increase_reservation(unsigned long nr_pages)
-+{
-+ unsigned long pfn, i, flags;
-+ struct page *page;
-+ long rc;
-+ struct xen_memory_reservation reservation = {
-+ .address_bits = 0,
-+ .extent_order = 0,
-+ .domid = DOMID_SELF
-+ };
++/* Per-NIC parameters */
++static struct efx_debugfs_parameter efx_debugfs_nic_parameters[] = {
++ EFX_INT_PARAMETER(struct efx_nic, legacy_irq),
++ EFX_INT_PARAMETER(struct efx_nic, rss_queues),
++ EFX_UINT_PARAMETER(struct efx_nic, rx_buffer_len),
++ EFX_INT_MODE_PARAMETER(struct efx_nic, interrupt_mode),
++ {.name = "hardware_desc",
++ .offset = 0,
++ .reader = falcon_debugfs_read_hardware_desc},
++ {NULL},
++};
+
-+ if (nr_pages > ARRAY_SIZE(frame_list))
-+ nr_pages = ARRAY_SIZE(frame_list);
++/* Per-NIC error counts */
++static struct efx_debugfs_parameter efx_debugfs_nic_error_parameters[] = {
++ EFX_ATOMIC_PARAMETER(struct efx_nic_errors, missing_event),
++ EFX_ATOMIC_PARAMETER(struct efx_nic_errors, rx_reset),
++ EFX_ATOMIC_PARAMETER(struct efx_nic_errors, rx_desc_fetch),
++ EFX_ATOMIC_PARAMETER(struct efx_nic_errors, tx_desc_fetch),
++ EFX_ATOMIC_PARAMETER(struct efx_nic_errors, spurious_tx),
++ {NULL},
++};
+
-+ balloon_lock(flags);
++/**
++ * efx_init_debugfs_channels - create debugfs directories for NIC channels
++ * @efx: Efx NIC
++ *
++ * Create subdirectories of @efx's debugfs directory for all the
++ * channels, RX queues and TX queues used by this driver. Return a
++ * negative error code or 0 on success. The subdirectories must be
++ * cleaned up using efx_fini_debugfs_channels().
++ */
++int efx_init_debugfs_channels(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++ struct efx_rx_queue *rx_queue;
++ struct efx_tx_queue *tx_queue;
++ int rc;
+
-+ page = balloon_first_page();
-+ for (i = 0; i < nr_pages; i++) {
-+ BUG_ON(page == NULL);
-+ frame_list[i] = page_to_pfn(page);;
-+ page = balloon_next_page(page);
++ efx_for_each_channel(channel, efx) {
++ rc = efx_init_debugfs_channel(channel);
++ if (rc)
++ goto err;
+ }
+
-+ set_xen_guest_handle(reservation.extent_start, frame_list);
-+ reservation.nr_extents = nr_pages;
-+ rc = HYPERVISOR_memory_op(
-+ XENMEM_populate_physmap, &reservation);
-+ if (rc < nr_pages) {
-+ if (rc > 0) {
-+ int ret;
-+
-+ /* We hit the Xen hard limit: reprobe. */
-+ reservation.nr_extents = rc;
-+ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
-+ &reservation);
-+ BUG_ON(ret != rc);
-+ }
-+ if (rc >= 0)
-+ bs.hard_limit = (bs.current_pages + rc -
-+ bs.driver_pages);
-+ goto out;
++ efx_for_each_rx_queue(rx_queue, efx) {
++ rc = efx_init_debugfs_rx_queue(rx_queue);
++ if (rc)
++ goto err;
+ }
+
-+ for (i = 0; i < nr_pages; i++) {
-+ page = balloon_retrieve();
-+ BUG_ON(page == NULL);
-+
-+ pfn = page_to_pfn(page);
-+ BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap) &&
-+ phys_to_machine_mapping_valid(pfn));
++ efx_for_each_tx_queue(tx_queue, efx) {
++ rc = efx_init_debugfs_tx_queue(tx_queue);
++ if (rc)
++ goto err;
++ }
+
-+ set_phys_to_machine(pfn, frame_list[i]);
++ return 0;
+
-+#ifdef CONFIG_XEN
-+ /* Link back into the page tables if not highmem. */
-+ if (pfn < max_low_pfn) {
-+ int ret;
-+ ret = HYPERVISOR_update_va_mapping(
-+ (unsigned long)__va(pfn << PAGE_SHIFT),
-+ pfn_pte_ma(frame_list[i], PAGE_KERNEL),
-+ 0);
-+ BUG_ON(ret);
-+ }
-+#endif
++ err:
++ efx_fini_debugfs_channels(efx);
++ return rc;
++}
+
-+ /* Relinquish the page back to the allocator. */
-+ ClearPageReserved(page);
-+ init_page_count(page);
-+ __free_page(page);
-+ }
++/**
++ * efx_fini_debugfs_channels - remove debugfs directories for NIC queues
++ * @efx: Efx NIC
++ *
++ * Remove subdirectories of @efx's debugfs directory created by
++ * efx_init_debugfs_channels().
++ */
++void efx_fini_debugfs_channels(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++ struct efx_rx_queue *rx_queue;
++ struct efx_tx_queue *tx_queue;
+
-+ bs.current_pages += nr_pages;
-+ totalram_pages = bs.current_pages;
++ efx_for_each_tx_queue(tx_queue, efx)
++ efx_fini_debugfs_tx_queue(tx_queue);
+
-+ out:
-+ balloon_unlock(flags);
++ efx_for_each_rx_queue(rx_queue, efx)
++ efx_fini_debugfs_rx_queue(rx_queue);
+
-+ return 0;
++ efx_for_each_channel(channel, efx)
++ efx_fini_debugfs_channel(channel);
+}
+
-+static int decrease_reservation(unsigned long nr_pages)
++/**
++ * efx_init_debugfs_nic - create debugfs directory for NIC
++ * @efx: Efx NIC
++ *
++ * Create debugfs directory containing parameter-files for @efx,
++ * and a subdirectory "errors" containing per-NIC error counts.
++ * Return a negative error code or 0 on success. The directories
++ * must be cleaned up using efx_fini_debugfs_nic().
++ */
++int efx_init_debugfs_nic(struct efx_nic *efx)
+{
-+ unsigned long pfn, i, flags;
-+ struct page *page;
-+ void *v;
-+ int need_sleep = 0;
-+ int ret;
-+ struct xen_memory_reservation reservation = {
-+ .address_bits = 0,
-+ .extent_order = 0,
-+ .domid = DOMID_SELF
-+ };
++ int rc;
+
-+ if (nr_pages > ARRAY_SIZE(frame_list))
-+ nr_pages = ARRAY_SIZE(frame_list);
++ /* Create directory */
++ efx->debug_dir = debugfs_create_dir(pci_name(efx->pci_dev),
++ efx_debug_cards);
++ if (!efx->debug_dir)
++ goto err_mem;
+
-+ for (i = 0; i < nr_pages; i++) {
-+ if ((page = alloc_page(GFP_BALLOON)) == NULL) {
-+ nr_pages = i;
-+ need_sleep = 1;
-+ break;
-+ }
++ /* Create errors directory */
++ efx->errors.debug_dir = debugfs_create_dir("errors", efx->debug_dir);
++ if (!efx->errors.debug_dir)
++ goto err_mem;
+
-+ pfn = page_to_pfn(page);
-+ frame_list[i] = pfn_to_mfn(pfn);
++ /* Create files */
++ rc = efx_init_debugfs_files(efx->debug_dir,
++ efx_debugfs_nic_parameters, (void *)efx);
++ if (rc)
++ goto err;
++ rc = efx_init_debugfs_files(efx->errors.debug_dir,
++ efx_debugfs_nic_error_parameters,
++ (void *)&efx->errors);
++ if (rc)
++ goto err;
+
-+ if (!PageHighMem(page)) {
-+ v = phys_to_virt(pfn << PAGE_SHIFT);
-+ scrub_pages(v, 1);
-+#ifdef CONFIG_XEN
-+ ret = HYPERVISOR_update_va_mapping(
-+ (unsigned long)v, __pte_ma(0), 0);
-+ BUG_ON(ret);
-+#endif
-+ }
-+#ifdef CONFIG_XEN_SCRUB_PAGES
-+ else {
-+ v = kmap(page);
-+ scrub_pages(v, 1);
-+ kunmap(page);
-+ }
++ return 0;
++
++ err_mem:
++ rc = -ENOMEM;
++ err:
++ efx_fini_debugfs_nic(efx);
++ return rc;
++}
++
++/**
++ * efx_fini_debugfs_nic - remove debugfs directories for NIC
++ * @efx: Efx NIC
++ *
++ * Remove debugfs directories created for @efx by efx_init_debugfs_nic().
++ */
++void efx_fini_debugfs_nic(struct efx_nic *efx)
++{
++ efx_fini_debugfs_dir(efx->errors.debug_dir,
++ efx_debugfs_nic_error_parameters, NULL);
++ efx->errors.debug_dir = NULL;
++ efx_fini_debugfs_dir(efx->debug_dir, efx_debugfs_nic_parameters, NULL);
++ efx->debug_dir = NULL;
++}
++
++/**
++ * efx_init_debugfs - create debugfs directories for sfc driver
++ *
++ * Create debugfs directories "sfc" and "sfc/cards". This must be
++ * called before any of the other functions that create debugfs
++ * directories. Return a negative error code or 0 on success. The
++ * directories must be cleaned up using efx_fini_debugfs().
++ */
++int efx_init_debugfs(void)
++{
++ /* Create top-level directory */
++#if defined(EFX_USE_DEBUGFS)
++ efx_debug_root = debugfs_create_dir("sfc", NULL);
++#else
++ efx_debug_root = proc_mkdir("sfc", proc_root_driver);
+#endif
-+ }
++ if (!efx_debug_root)
++ goto err;
+
-+#ifdef CONFIG_XEN
-+ /* Ensure that ballooned highmem pages don't have kmaps. */
-+ kmap_flush_unused();
-+ flush_tlb_all();
++ /* Create "cards" directory */
++ efx_debug_cards = debugfs_create_dir("cards", efx_debug_root);
++ if (!efx_debug_cards)
++ goto err;
++
++#if defined(EFX_USE_DEBUGFS)
++ /* Create compatibility sym-link */
++ if (!proc_symlink("sfc", proc_root_driver, "/sys/kernel/debug/sfc"))
++ goto err;
+#endif
++ return 0;
+
-+ balloon_lock(flags);
++ err:
++ efx_fini_debugfs();
++ return -ENOMEM;
++}
+
-+ /* No more mappings: invalidate P2M and add to balloon. */
-+ for (i = 0; i < nr_pages; i++) {
-+ pfn = mfn_to_pfn(frame_list[i]);
-+ set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
-+ balloon_append(pfn_to_page(pfn));
-+ }
++/**
++ * efx_fini_debugfs - remove debugfs directories for sfc driver
++ *
++ * Remove directories created by efx_init_debugfs().
++ */
++void efx_fini_debugfs(void)
++{
++#if defined(EFX_USE_DEBUGFS)
++ remove_proc_entry("sfc", proc_root_driver);
++#endif
++ debugfs_remove(efx_debug_cards);
++ efx_debug_cards = NULL;
++ debugfs_remove(efx_debug_root);
++ efx_debug_root = NULL;
++}
+--- linux-2.6.18.8/drivers/net/sfc/debugfs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/debugfs.h 2008-05-19 00:33:28.833808141 +0300
+@@ -0,0 +1,172 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ set_xen_guest_handle(reservation.extent_start, frame_list);
-+ reservation.nr_extents = nr_pages;
-+ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
-+ BUG_ON(ret != nr_pages);
++#ifndef EFX_DEBUGFS_H
++#define EFX_DEBUGFS_H
+
-+ bs.current_pages -= nr_pages;
-+ totalram_pages = bs.current_pages;
++#ifdef CONFIG_SFC_DEBUGFS
+
-+ balloon_unlock(flags);
++struct seq_file;
+
-+ return need_sleep;
++struct efx_debugfs_parameter {
++ const char *name;
++ size_t offset;
++ int (*reader)(struct seq_file *, void *);
++};
++
++extern void efx_fini_debugfs_child(struct dentry *dir, const char *name);
++extern int efx_init_debugfs_netdev(struct net_device *net_dev);
++extern void efx_fini_debugfs_netdev(struct net_device *net_dev);
++extern int efx_init_debugfs_port(struct efx_nic *efx);
++extern void efx_fini_debugfs_port(struct efx_nic *efx);
++extern int efx_init_debugfs_nic(struct efx_nic *efx);
++extern void efx_fini_debugfs_nic(struct efx_nic *efx);
++extern int efx_init_debugfs_channels(struct efx_nic *efx);
++extern void efx_fini_debugfs_channels(struct efx_nic *efx);
++extern int efx_init_debugfs(void);
++extern void efx_fini_debugfs(void);
++extern int efx_extend_debugfs_port(struct efx_nic *efx,
++ void *context,
++ struct efx_debugfs_parameter *params);
++extern void efx_trim_debugfs_port(struct efx_nic *efx,
++ struct efx_debugfs_parameter *params);
++
++/* Helpers for handling debugfs entry reads */
++extern int efx_debugfs_read_uint(struct seq_file *, void *);
++extern int efx_debugfs_read_string(struct seq_file *, void *);
++extern int efx_debugfs_read_int(struct seq_file *, void *);
++extern int efx_debugfs_read_atomic(struct seq_file *, void *);
++extern int efx_debugfs_read_dword(struct seq_file *, void *);
++
++/* Handy macros for filling out parameters */
++
++/* Initialiser for a struct efx_debugfs_parameter with type-checking */
++#define EFX_PARAMETER(container_type, parameter, field_type, \
++ reader_function) { \
++ .name = #parameter, \
++ .offset = ((((field_type *) 0) == \
++ &((container_type *) 0)->parameter) ? \
++ offsetof(container_type, parameter) : \
++ offsetof(container_type, parameter)), \
++ .reader = reader_function, \
++}
++
++/* Likewise, but the file name is not taken from the field name */
++#define EFX_NAMED_PARAMETER(_name, container_type, parameter, field_type, \
++ reader_function) { \
++ .name = #_name, \
++ .offset = ((((field_type *) 0) == \
++ &((container_type *) 0)->parameter) ? \
++ offsetof(container_type, parameter) : \
++ offsetof(container_type, parameter)), \
++ .reader = reader_function, \
++}
++
++/* Likewise, but with one file for each of 4 lanes */
++#define EFX_PER_LANE_PARAMETER(prefix, suffix, container_type, parameter, \
++ field_type, reader_function) { \
++ .name = prefix "0" suffix, \
++ .offset = ((((field_type *) 0) == \
++ ((container_type *) 0)->parameter) ? \
++ offsetof(container_type, parameter[0]) : \
++ offsetof(container_type, parameter[0])), \
++ .reader = reader_function, \
++}, { \
++ .name = prefix "1" suffix, \
++ .offset = offsetof(container_type, parameter[1]), \
++ .reader = reader_function, \
++}, { \
++ .name = prefix "2" suffix, \
++ .offset = offsetof(container_type, parameter[2]), \
++ .reader = reader_function, \
++}, { \
++ .name = prefix "3" suffix, \
++ .offset = offsetof(container_type, parameter[3]), \
++ .reader = reader_function, \
++}
++
++/* A string parameter (string embedded in the structure) */
++#define EFX_STRING_PARAMETER(container_type, parameter) { \
++ .name = #parameter, \
++ .offset = ((((char *) 0) == \
++ ((container_type *) 0)->parameter) ? \
++ offsetof(container_type, parameter) : \
++ offsetof(container_type, parameter)), \
++ .reader = efx_debugfs_read_string, \
++}
++
++/* An unsigned integer parameter */
++#define EFX_UINT_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ unsigned int, efx_debugfs_read_uint)
++
++/* A dword parameter */
++#define EFX_DWORD_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ efx_dword_t, efx_debugfs_read_dword)
++
++/* An atomic_t parameter */
++#define EFX_ATOMIC_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ atomic_t, efx_debugfs_read_atomic)
++
++/* An integer parameter */
++#define EFX_INT_PARAMETER(container_type, parameter) \
++ EFX_PARAMETER(container_type, parameter, \
++ int, efx_debugfs_read_int)
++
++#else /* !CONFIG_SFC_DEBUGFS */
++
++static inline int efx_init_debugfs_netdev(struct net_device *net_dev)
++{
++ return 0;
++}
++static inline void efx_fini_debugfs_netdev(struct net_device *net_dev) {}
++static inline int efx_init_debugfs_port(struct efx_nic *efx)
++{
++ return 0;
++}
++static inline void efx_fini_debugfs_port(struct efx_nic *efx) {}
++static inline int efx_init_debugfs_nic(struct efx_nic *efx)
++{
++ return 0;
++}
++static inline void efx_fini_debugfs_nic(struct efx_nic *efx) {}
++static inline int efx_init_debugfs_channels(struct efx_nic *efx)
++{
++ return 0;
++}
++static inline void efx_fini_debugfs_channels(struct efx_nic *efx) {}
++static inline int efx_init_debugfs(void)
++{
++ return 0;
+}
++static inline void efx_fini_debugfs(void) {}
+
-+/*
-+ * We avoid multiple worker processes conflicting via the balloon mutex.
-+ * We may of course race updates of the target counts (which are protected
-+ * by the balloon lock), or with changes to the Xen hard limit, but we will
-+ * recover from these in time.
++#endif /* CONFIG_SFC_DEBUGFS */
++
++#endif /* EFX_DEBUGFS_H */
+--- linux-2.6.18.8/drivers/net/sfc/driverlink.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/driverlink.c 2008-05-19 00:33:28.833808141 +0300
+@@ -0,0 +1,544 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static void balloon_process(void *unused)
-+{
-+ int need_sleep = 0;
-+ long credit;
+
-+ mutex_lock(&balloon_mutex);
++#include <linux/module.h>
++#include <linux/list.h>
++#include <linux/skbuff.h>
++#include <linux/rtnetlink.h>
++#include "net_driver.h"
++#include "efx.h"
++#include "driverlink.h"
+
-+ do {
-+ credit = current_target() - bs.current_pages;
-+ if (credit > 0)
-+ need_sleep = (increase_reservation(credit) != 0);
-+ if (credit < 0)
-+ need_sleep = (decrease_reservation(-credit) != 0);
++/* Driverlink semaphore
++ * This semaphore must be held for any operation that modifies any of
++ * the driverlink lists.
++ */
++static DEFINE_MUTEX(efx_driverlink_lock);
+
-+#ifndef CONFIG_PREEMPT
-+ if (need_resched())
-+ schedule();
-+#endif
-+ } while ((credit != 0) && !need_sleep);
++/* List of all registered drivers */
++static LIST_HEAD(efx_driver_list);
+
-+ /* Schedule more work if there is some still to be done. */
-+ if (current_target() != bs.current_pages)
-+ mod_timer(&balloon_timer, jiffies + HZ);
++/* List of all registered Efx ports */
++static LIST_HEAD(efx_port_list);
+
-+ mutex_unlock(&balloon_mutex);
-+}
++/* Driver link handle used internally to track devices */
++struct efx_dl_handle {
++ /* The efx_dl_device consumers see */
++ struct efx_dl_device efx_dev;
++ /* The efx_nic providers provide */
++ struct efx_nic *efx;
++ /* Per-device list */
++ struct list_head port_node;
++ /* Per-driver list */
++ struct list_head driver_node;
++};
+
-+/* Resets the Xen limit, sets new target, and kicks off processing. */
-+void balloon_set_new_target(unsigned long target)
++/* Get the handle for an efx_dl_device */
++static struct efx_dl_handle *efx_dl_handle(struct efx_dl_device *efx_dev)
+{
-+ /* No need for lock. Not read-modify-write updates. */
-+ bs.hard_limit = ~0UL;
-+ bs.target_pages = target;
-+ schedule_work(&balloon_worker);
++ return container_of(efx_dev, struct efx_dl_handle, efx_dev);
+}
+
-+static struct xenbus_watch target_watch =
++/* Remove an Efx device
++ * You must hold the efx_driverlink_lock before calling this
++ * function.
++ */
++static void efx_dl_del_device(struct efx_dl_device *efx_dev)
+{
-+ .node = "memory/target"
-+};
++ struct efx_dl_handle *efx_handle = efx_dl_handle(efx_dev);
+
-+/* React to a change in the target key */
-+static void watch_target(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
-+{
-+ unsigned long long new_target;
-+ int err;
++ EFX_INFO(efx_handle->efx, "%s driverlink client unregistering\n",
++ efx_dev->driver->name);
+
-+ err = xenbus_scanf(XBT_NIL, "memory", "target", "%llu", &new_target);
-+ if (err != 1) {
-+ /* This is ok (for domain0 at least) - so just return */
-+ return;
-+ }
++ /* Call driver's remove() routine */
++ if (efx_dev->driver->remove)
++ efx_dev->driver->remove(efx_dev);
+
-+ /* The given memory/target value is in KiB, so it needs converting to
-+ * pages. PAGE_SHIFT converts bytes to pages, hence PAGE_SHIFT - 10.
-+ */
-+ balloon_set_new_target(new_target >> (PAGE_SHIFT - 10));
++ /* Remove handle from per-driver and per-NIC lists */
++ list_del(&efx_handle->driver_node);
++ list_del(&efx_handle->port_node);
++
++ /* Free efx_handle structure */
++ kfree(efx_handle);
+}
+
-+static int balloon_init_watcher(struct notifier_block *notifier,
-+ unsigned long event,
-+ void *data)
++/* Try to add an Efx device
++ * Attempt to probe the given device with the driver, creating a
++ * new efx_dl_device. If the probe routine fails, because the driver
++ * doesn't support this port, then the efx_dl_device is destroyed,
++ */
++static void efx_dl_try_add_device(struct efx_nic *efx,
++ struct efx_dl_driver *driver)
+{
-+ int err;
++ struct efx_dl_handle *efx_handle;
++ struct efx_dl_device *efx_dev;
++ int rc;
+
-+ err = register_xenbus_watch(&target_watch);
-+ if (err)
-+ printk(KERN_ERR "Failed to set balloon watcher\n");
++ /* Allocate and initialise new efx_dl_device structure */
++ efx_handle = kzalloc(sizeof(*efx_handle), GFP_KERNEL);
++ efx_dev = &efx_handle->efx_dev;
++ efx_handle->efx = efx;
++ efx_dev->driver = driver;
++ efx_dev->pci_dev = efx->pci_dev;
++ INIT_LIST_HEAD(&efx_handle->port_node);
++ INIT_LIST_HEAD(&efx_handle->driver_node);
++
++ /* Attempt driver probe */
++ rc = driver->probe(efx_dev, efx->net_dev,
++ efx->dl_info, efx->silicon_rev);
++ if (rc)
++ goto fail;
+
-+ return NOTIFY_DONE;
++ /* Add device to per-driver and per-NIC lists */
++ list_add_tail(&efx_handle->driver_node, &driver->device_list);
++ list_add_tail(&efx_handle->port_node, &efx->dl_device_list);
++
++ EFX_INFO(efx, "%s driverlink client registered\n", driver->name);
++ return;
++
++ fail:
++ EFX_INFO(efx, "%s driverlink client skipped\n", driver->name);
++
++ kfree(efx_dev);
+}
+
-+#ifdef CONFIG_PROC_FS
-+static int balloon_write(struct file *file, const char __user *buffer,
-+ unsigned long count, void *data)
++/**
++ * efx_dl_unregister_driver - unregister an Efx device driver
++ * @driver: Efx driverlink driver
++ *
++ * Unregisters an Efx driver. The driver's remove() method will be
++ * called for all Efx devices currently claimed by the driver.
++ */
++void efx_dl_unregister_driver(struct efx_dl_driver *driver)
+{
-+ char memstring[64], *endchar;
-+ unsigned long long target_bytes;
++ struct efx_dl_handle *efx_handle, *efx_handle_n;
+
-+ if (!capable(CAP_SYS_ADMIN))
-+ return -EPERM;
++ printk(KERN_INFO "Efx driverlink unregistering %s driver\n",
++ driver->name);
+
-+ if (count <= 1)
-+ return -EBADMSG; /* runt */
-+ if (count > sizeof(memstring))
-+ return -EFBIG; /* too long */
++ /* Acquire lock. We can't return failure, so have to use
++ * down() instead of down_interruptible()
++ */
++ mutex_lock(&efx_driverlink_lock);
+
-+ if (copy_from_user(memstring, buffer, count))
-+ return -EFAULT;
-+ memstring[sizeof(memstring)-1] = '\0';
++ /* Remove all devices claimed by the driver */
++ list_for_each_entry_safe(efx_handle, efx_handle_n,
++ &driver->device_list, driver_node)
++ efx_dl_del_device(&efx_handle->efx_dev);
+
-+ target_bytes = memparse(memstring, &endchar);
-+ balloon_set_new_target(target_bytes >> PAGE_SHIFT);
++ /* Remove driver from driver list */
++ list_del(&driver->node);
+
-+ return count;
++ /* Release lock */
++ mutex_unlock(&efx_driverlink_lock);
+}
++EXPORT_SYMBOL(efx_dl_unregister_driver);
+
-+static int balloon_read(char *page, char **start, off_t off,
-+ int count, int *eof, void *data)
++/**
++ * efx_dl_register_driver - register an Efx device driver
++ * @driver: Efx driverlink driver
++ *
++ * Registers a new Efx driver. The driver's probe() method will be
++ * called for all Efx NICs currently registered.
++ *
++ * Return a negative error code or 0 on success.
++ */
++int efx_dl_register_driver(struct efx_dl_driver *driver)
+{
-+ int len;
++ struct efx_nic *efx;
++ int rc;
+
-+ len = sprintf(
-+ page,
-+ "Current allocation: %8lu kB\n"
-+ "Requested target: %8lu kB\n"
-+ "Low-mem balloon: %8lu kB\n"
-+ "High-mem balloon: %8lu kB\n"
-+ "Driver pages: %8lu kB\n"
-+ "Xen hard limit: ",
-+ PAGES2KB(bs.current_pages), PAGES2KB(bs.target_pages),
-+ PAGES2KB(bs.balloon_low), PAGES2KB(bs.balloon_high),
-+ PAGES2KB(bs.driver_pages));
++ printk(KERN_INFO "Efx driverlink registering %s driver\n",
++ driver->name);
+
-+ if (bs.hard_limit != ~0UL)
-+ len += sprintf(page + len, "%8lu kB\n",
-+ PAGES2KB(bs.hard_limit));
-+ else
-+ len += sprintf(page + len, " ??? kB\n");
++ /* Initialise driver list structures */
++ INIT_LIST_HEAD(&driver->node);
++ INIT_LIST_HEAD(&driver->device_list);
+
-+ *eof = 1;
-+ return len;
-+}
-+#endif
++ /* Acquire lock */
++ rc = mutex_lock_interruptible(&efx_driverlink_lock);
++ if (rc)
++ return rc;
+
-+static struct notifier_block xenstore_notifier;
++ /* Add driver to driver list */
++ list_add_tail(&driver->node, &efx_driver_list);
+
-+static int __init balloon_init(void)
++ /* Feed all existing devices to driver */
++ list_for_each_entry(efx, &efx_port_list, dl_node)
++ efx_dl_try_add_device(efx, driver);
++
++ /* Release locks */
++ mutex_unlock(&efx_driverlink_lock);
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_dl_register_driver);
++
++void efx_dl_unregister_nic(struct efx_nic *efx)
+{
-+#if defined(CONFIG_X86) && defined(CONFIG_XEN)
-+ unsigned long pfn;
-+ struct page *page;
-+#endif
++ struct efx_dl_handle *efx_handle, *efx_handle_n;
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ if (!efx)
++ return;
+
-+ IPRINTK("Initialising balloon driver.\n");
++ /* Acquire lock. We can't return failure, so have to use
++ * down() instead of down_interruptible()
++ */
++ mutex_lock(&efx_driverlink_lock);
+
-+#ifdef CONFIG_XEN
-+ bs.current_pages = min(xen_start_info->nr_pages, max_pfn);
-+ totalram_pages = bs.current_pages;
-+#else
-+ bs.current_pages = totalram_pages;
-+#endif
-+ bs.target_pages = bs.current_pages;
-+ bs.balloon_low = 0;
-+ bs.balloon_high = 0;
-+ bs.driver_pages = 0UL;
-+ bs.hard_limit = ~0UL;
++ /* Remove all devices related to this NIC */
++ list_for_each_entry_safe_reverse(efx_handle, efx_handle_n,
++ &efx->dl_device_list,
++ port_node)
++ efx_dl_del_device(&efx_handle->efx_dev);
+
-+ init_timer(&balloon_timer);
-+ balloon_timer.data = 0;
-+ balloon_timer.function = balloon_alarm;
-+
-+#ifdef CONFIG_PROC_FS
-+ if ((balloon_pde = create_xen_proc_entry("balloon", 0644)) == NULL) {
-+ WPRINTK("Unable to create /proc/xen/balloon.\n");
-+ return -1;
-+ }
++ /* Remove port from port list */
++ list_del(&efx->dl_node);
+
-+ balloon_pde->read_proc = balloon_read;
-+ balloon_pde->write_proc = balloon_write;
-+#endif
-+ balloon_sysfs_init();
++ /* Release lock */
++ mutex_unlock(&efx_driverlink_lock);
++}
+
-+#if defined(CONFIG_X86) && defined(CONFIG_XEN)
-+ /* Initialise the balloon with excess memory space. */
-+ for (pfn = xen_start_info->nr_pages; pfn < max_pfn; pfn++) {
-+ page = pfn_to_page(pfn);
-+ if (!PageReserved(page))
-+ balloon_append(page);
-+ }
-+#endif
++int efx_dl_register_nic(struct efx_nic *efx)
++{
++ struct efx_dl_driver *driver;
++ int rc;
+
-+ target_watch.callback = watch_target;
-+ xenstore_notifier.notifier_call = balloon_init_watcher;
++ /* Acquire lock */
++ rc = mutex_lock_interruptible(&efx_driverlink_lock);
++ if (rc)
++ return rc;
++
++ /* Add port to port list */
++ list_add_tail(&efx->dl_node, &efx_port_list);
++
++ /* Feed port to all existing drivers */
++ list_for_each_entry(driver, &efx_driver_list, node)
++ efx_dl_try_add_device(efx, driver);
++
++ /* Release lock */
++ mutex_unlock(&efx_driverlink_lock);
+
-+ register_xenstore_notifier(&xenstore_notifier);
-+
+ return 0;
+}
+
-+subsys_initcall(balloon_init);
-+
-+static void __exit balloon_exit(void)
++/*
++ * Dummy callback implementations.
++ *
++ * To avoid a branch point on the fast-path, the callbacks are always
++ * implemented - they are never NULL.
++ */
++#if defined(EFX_USE_FASTCALL)
++static enum efx_veto fastcall
++#else
++static enum efx_veto
++#endif
++efx_dummy_tx_packet_callback(struct efx_dl_device *efx_dev, struct sk_buff *skb)
+{
-+ /* XXX - release balloon here */
-+ return;
++ /* Never veto the packet */
++ return EFX_ALLOW_PACKET;
+}
+
-+module_exit(balloon_exit);
-+
-+void balloon_update_driver_allowance(long delta)
++#if defined(EFX_USE_FASTCALL)
++static enum efx_veto fastcall
++#else
++static enum efx_veto
++#endif
++efx_dummy_rx_packet_callback(struct efx_dl_device *efx_dev,
++ const char *pkt_buf, int len)
+{
-+ unsigned long flags;
++ /* Never veto the packet */
++ return EFX_ALLOW_PACKET;
++}
+
-+ balloon_lock(flags);
-+ bs.driver_pages += delta;
-+ balloon_unlock(flags);
++static void
++efx_dummy_link_change_callback(struct efx_dl_device *efx_dev, int link_up)
++{
+}
+
-+#ifdef CONFIG_XEN
-+static int dealloc_pte_fn(
-+ pte_t *pte, struct page *pmd_page, unsigned long addr, void *data)
++static int
++efx_dummy_request_mtu_callback(struct efx_dl_device *efx_dev, int new_mtu)
+{
-+ unsigned long mfn = pte_mfn(*pte);
-+ int ret;
-+ struct xen_memory_reservation reservation = {
-+ .nr_extents = 1,
-+ .extent_order = 0,
-+ .domid = DOMID_SELF
-+ };
-+ set_xen_guest_handle(reservation.extent_start, &mfn);
-+ set_pte_at(&init_mm, addr, pte, __pte_ma(0));
-+ set_phys_to_machine(__pa(addr) >> PAGE_SHIFT, INVALID_P2M_ENTRY);
-+ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
-+ BUG_ON(ret != 1);
++ /* Always allow */
+ return 0;
+}
-+#endif
+
-+struct page **alloc_empty_pages_and_pagevec(int nr_pages)
++static void
++efx_dummy_mtu_changed_callback(struct efx_dl_device *efx_dev, int mtu)
+{
-+ unsigned long vaddr, flags;
-+ struct page *page, **pagevec;
-+ int i, ret;
++ return;
++}
+
-+ pagevec = kmalloc(sizeof(page) * nr_pages, GFP_KERNEL);
-+ if (pagevec == NULL)
-+ return NULL;
++static void efx_dummy_event_callback(struct efx_dl_device *efx_dev, void *event)
++{
++ return;
++}
+
-+ for (i = 0; i < nr_pages; i++) {
-+ page = pagevec[i] = alloc_page(GFP_KERNEL);
-+ if (page == NULL)
-+ goto err;
++struct efx_dl_callbacks efx_default_callbacks = {
++ .tx_packet = efx_dummy_tx_packet_callback,
++ .rx_packet = efx_dummy_rx_packet_callback,
++ .link_change = efx_dummy_link_change_callback,
++ .request_mtu = efx_dummy_request_mtu_callback,
++ .mtu_changed = efx_dummy_mtu_changed_callback,
++ .event = efx_dummy_event_callback,
++};
+
-+ vaddr = (unsigned long)page_address(page);
++#define EFX_DL_UNREGISTER_CALLBACK(_port, _dev, _member) \
++ do { \
++ BUG_ON((_port)->dl_cb_dev._member != (_dev)); \
++ (_port)->dl_cb._member = \
++ efx_default_callbacks._member; \
++ (_port)->dl_cb_dev._member = NULL; \
++ } while (0)
+
-+ scrub_pages(vaddr, 1);
+
-+ balloon_lock(flags);
++#define EFX_DL_REGISTER_CALLBACK(_port, _dev, _from, _member) \
++ if ((_from)->_member) { \
++ BUG_ON((_port)->dl_cb_dev._member != NULL); \
++ (_port)->dl_cb._member = (_from)->_member; \
++ (_port)->dl_cb_dev._member = _dev; \
++ }
+
-+ if (xen_feature(XENFEAT_auto_translated_physmap)) {
-+ unsigned long gmfn = page_to_pfn(page);
-+ struct xen_memory_reservation reservation = {
-+ .nr_extents = 1,
-+ .extent_order = 0,
-+ .domid = DOMID_SELF
-+ };
-+ set_xen_guest_handle(reservation.extent_start, &gmfn);
-+ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
-+ &reservation);
-+ if (ret == 1)
-+ ret = 0; /* success */
-+ } else {
-+#ifdef CONFIG_XEN
-+ ret = apply_to_page_range(&init_mm, vaddr, PAGE_SIZE,
-+ dealloc_pte_fn, NULL);
-+#else
-+ /* Cannot handle non-auto translate mode. */
-+ ret = 1;
-+#endif
-+ }
++/**
++ * efx_dl_unregister_callbacks - unregister callbacks for an Efx NIC
++ * @efx_dev: Efx driverlink device
++ * @callbacks: Callback list
++ *
++ * This removes a set of callbacks registered with
++ * efx_dl_register_callbacks(). It should be called as part of the
++ * client's remove() method.
++ *
++ * The net driver will ensure that all callback functions have
++ * returned to the net driver before efx_dl_unregister_callbacks()
++ * returns. Note that the device itself may still be running when the
++ * client's remove() method is called. The client must therefore
++ * unhook its callbacks using efx_dl_unregister_callbacks() and only
++ * then ensure that any delayed tasks triggered by callback methods
++ * (e.g. scheduled tasklets) have completed.
++ */
++void efx_dl_unregister_callbacks(struct efx_dl_device *efx_dev,
++ struct efx_dl_callbacks *callbacks)
++{
++ struct efx_dl_handle *efx_handle = efx_dl_handle(efx_dev);
++ struct efx_nic *efx = efx_handle->efx;
+
-+ if (ret != 0) {
-+ balloon_unlock(flags);
-+ __free_page(page);
-+ goto err;
-+ }
++ /* Suspend net driver operations */
++ efx_suspend(efx);
+
-+ totalram_pages = --bs.current_pages;
++ EFX_INFO(efx, "removing callback hooks into %s driver\n",
++ efx_dev->driver->name);
+
-+ balloon_unlock(flags);
-+ }
++ if (callbacks->tx_packet)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, tx_packet);
+
-+ out:
-+ schedule_work(&balloon_worker);
-+#ifdef CONFIG_XEN
-+ flush_tlb_all();
-+#endif
-+ return pagevec;
++ if (callbacks->rx_packet)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, rx_packet);
+
-+ err:
-+ balloon_lock(flags);
-+ while (--i >= 0)
-+ balloon_append(pagevec[i]);
-+ balloon_unlock(flags);
-+ kfree(pagevec);
-+ pagevec = NULL;
-+ goto out;
++ if (callbacks->link_change)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, link_change);
++
++ if (callbacks->request_mtu)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, request_mtu);
++
++ if (callbacks->mtu_changed)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, mtu_changed);
++
++ if (callbacks->event)
++ EFX_DL_UNREGISTER_CALLBACK(efx, efx_dev, event);
++
++ /* Resume net driver operations */
++ efx_resume(efx);
+}
++EXPORT_SYMBOL(efx_dl_unregister_callbacks);
+
-+void free_empty_pages_and_pagevec(struct page **pagevec, int nr_pages)
++/**
++ * efx_dl_register_callbacks - register callbacks for an Efx NIC
++ * @efx_dev: Efx driverlink device
++ * @callbacks: Callback list
++ *
++ * This registers a set of callback functions with the net driver.
++ * These functions will be called at various key points to allow
++ * external code to monitor and/or modify the behaviour of the network
++ * driver. Any of the callback function pointers may be %NULL if a
++ * callback is not required. The intended user of this mechanism is
++ * the SFC char driver.
++ *
++ * This client should call efx_dl_register_callbacks() during its
++ * probe() method. The client must ensure that it also calls
++ * efx_dl_unregister_callbacks() as part of its remove() method.
++ *
++ * Only one function may be registered for each callback per NIC.
++ * If a requested callback is already registered for this NIC, this
++ * function will return -%EBUSY.
++ *
++ * The device may already be running, so the client must be prepared
++ * for callbacks to be triggered immediately after calling
++ * efx_dl_register_callbacks().
++ *
++ * Return a negative error code or 0 on success.
++ */
++int efx_dl_register_callbacks(struct efx_dl_device *efx_dev,
++ struct efx_dl_callbacks *callbacks)
+{
-+ unsigned long flags;
-+ int i;
++ struct efx_dl_handle *efx_handle = efx_dl_handle(efx_dev);
++ struct efx_nic *efx = efx_handle->efx;
++ int rc = 0;
+
-+ if (pagevec == NULL)
-+ return;
++ /* Suspend net driver operations */
++ efx_suspend(efx);
+
-+ balloon_lock(flags);
-+ for (i = 0; i < nr_pages; i++) {
-+ BUG_ON(page_count(pagevec[i]) != 1);
-+ balloon_append(pagevec[i]);
++ /* Check that the requested callbacks are not already hooked. */
++ if ((callbacks->tx_packet && efx->dl_cb_dev.tx_packet) ||
++ (callbacks->rx_packet && efx->dl_cb_dev.rx_packet) ||
++ (callbacks->link_change && efx->dl_cb_dev.link_change) ||
++ (callbacks->request_mtu && efx->dl_cb_dev.request_mtu) ||
++ (callbacks->mtu_changed && efx->dl_cb_dev.mtu_changed) ||
++ (callbacks->event && efx->dl_cb_dev.event)) {
++ rc = -EBUSY;
++ goto out;
+ }
-+ balloon_unlock(flags);
+
-+ kfree(pagevec);
++ EFX_INFO(efx, "adding callback hooks to %s driver\n",
++ efx_dev->driver->name);
+
-+ schedule_work(&balloon_worker);
++ /* Hook in callbacks. For maximum speed, we never check to
++ * see whether these are NULL before calling; therefore we
++ * must ensure that they are never NULL. If the set we're
++ * being asked to hook in is sparse, we leave the default
++ * values in place for the empty hooks.
++ */
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, tx_packet);
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, rx_packet);
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, link_change);
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, request_mtu);
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, mtu_changed);
++ EFX_DL_REGISTER_CALLBACK(efx, efx_dev, callbacks, event);
++
++ out:
++ /* Resume net driver operations */
++ efx_resume(efx);
++
++ return rc;
+}
++EXPORT_SYMBOL(efx_dl_register_callbacks);
+
-+void balloon_release_driver_page(struct page *page)
++/**
++ * efx_dl_schedule_reset - schedule an Efx NIC reset
++ * @efx_dev: Efx driverlink device
++ *
++ * This schedules a hardware reset for a short time in the future. It
++ * can be called from any context, and so can be used when
++ * efx_dl_reset() cannot be called.
++ */
++void efx_dl_schedule_reset(struct efx_dl_device *efx_dev)
+{
-+ unsigned long flags;
-+
-+ balloon_lock(flags);
-+ balloon_append(page);
-+ bs.driver_pages--;
-+ balloon_unlock(flags);
++ struct efx_dl_handle *efx_handle = efx_dl_handle(efx_dev);
++ struct efx_nic *efx = efx_handle->efx;
+
-+ schedule_work(&balloon_worker);
++ efx_schedule_reset(efx, RESET_TYPE_ALL);
+}
++EXPORT_SYMBOL(efx_dl_schedule_reset);
+
-+EXPORT_SYMBOL_GPL(balloon_update_driver_allowance);
-+EXPORT_SYMBOL_GPL(alloc_empty_pages_and_pagevec);
-+EXPORT_SYMBOL_GPL(free_empty_pages_and_pagevec);
-+EXPORT_SYMBOL_GPL(balloon_release_driver_page);
-+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/balloon/common.h linux-2.6.18-xen.hg/drivers/xen/balloon/common.h
---- linux-2.6.18/drivers/xen/balloon/common.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/balloon/common.h 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,58 @@
-+/******************************************************************************
-+ * balloon/common.h
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/*
++ * Lock the driverlink layer before a reset
++ * To avoid deadlock, efx_driverlink_lock needs to be acquired before
++ * efx->suspend_lock.
+ */
++void efx_dl_reset_lock(void)
++{
++ /* Acquire lock */
++ mutex_lock(&efx_driverlink_lock);
++}
+
-+#ifndef __XEN_BALLOON_COMMON_H__
-+#define __XEN_BALLOON_COMMON_H__
++/*
++ * Unlock the driverlink layer after a reset
++ * This call must be matched against efx_dl_reset_lock.
++ */
++void efx_dl_reset_unlock(void)
++{
++ /* Acquire lock */
++ mutex_unlock(&efx_driverlink_lock);
++}
+
-+#define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
++/*
++ * Suspend ready for reset
++ * This calls the reset_suspend method of all drivers registered to
++ * the specified NIC. It must only be called between
++ * efx_dl_reset_lock and efx_dl_reset_unlock.
++ */
++void efx_dl_reset_suspend(struct efx_nic *efx)
++{
++ struct efx_dl_handle *efx_handle;
++ struct efx_dl_device *efx_dev;
+
-+struct balloon_stats {
-+ /* We aim for 'current allocation' == 'target allocation'. */
-+ unsigned long current_pages;
-+ unsigned long target_pages;
-+ /* We may hit the hard limit in Xen. If we do then we remember it. */
-+ unsigned long hard_limit;
-+ /*
-+ * Drivers may alter the memory reservation independently, but they
-+ * must inform the balloon driver so we avoid hitting the hard limit.
-+ */
-+ unsigned long driver_pages;
-+ /* Number of pages in high- and low-memory balloons. */
-+ unsigned long balloon_low;
-+ unsigned long balloon_high;
-+};
++ BUG_ON(!mutex_is_locked(&efx_driverlink_lock));
+
-+extern struct balloon_stats balloon_stats;
-+#define bs balloon_stats
++ /* Call suspend method of each driver in turn */
++ list_for_each_entry_reverse(efx_handle,
++ &efx->dl_device_list,
++ port_node) {
++ efx_dev = &efx_handle->efx_dev;
++ if (efx_dev->driver->reset_suspend)
++ efx_dev->driver->reset_suspend(efx_dev);
++ }
++}
+
-+int balloon_sysfs_init(void);
-+void balloon_sysfs_exit(void);
++/*
++ * Resume after a reset
++ * This calls the reset_resume method of all drivers registered to the
++ * specified NIC. It must only be called between efx_dl_reset_lock
++ * and efx_dl_reset_unlock.
++ */
++void efx_dl_reset_resume(struct efx_nic *efx, int ok)
++{
++ struct efx_dl_handle *efx_handle;
++ struct efx_dl_device *efx_dev;
+
-+void balloon_set_new_target(unsigned long target);
++ BUG_ON(!mutex_is_locked(&efx_driverlink_lock));
+
-+#endif /* __XEN_BALLOON_COMMON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/balloon/Makefile linux-2.6.18-xen.hg/drivers/xen/balloon/Makefile
---- linux-2.6.18/drivers/xen/balloon/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/balloon/Makefile 2007-12-23 11:15:33.537908474 +0100
-@@ -0,0 +1,2 @@
++ /* Call resume method of each driver in turn */
++ list_for_each_entry(efx_handle, &efx->dl_device_list,
++ port_node) {
++ efx_dev = &efx_handle->efx_dev;
++ if (efx_dev->driver->reset_resume)
++ efx_dev->driver->reset_resume(efx_dev, ok);
++ }
++}
+
-+obj-y := balloon.o sysfs.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/balloon/sysfs.c linux-2.6.18-xen.hg/drivers/xen/balloon/sysfs.c
---- linux-2.6.18/drivers/xen/balloon/sysfs.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/balloon/sysfs.c 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,170 @@
-+/******************************************************************************
-+ * balloon/sysfs.c
++/**
++ * efx_dl_get_nic - obtain the Efx NIC for the given driverlink device
++ * @efx_dev: Efx driverlink device
+ *
-+ * Xen balloon driver - sysfs interfaces.
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * Get a pointer to the &struct efx_nic corresponding to
++ * @efx_dev. This can be used by driverlink clients built along with
++ * the sfc driver, which may have intimate knowledge of its internals.
++ */
++struct efx_nic *efx_dl_get_nic(struct efx_dl_device *efx_dev)
++{
++ return efx_dl_handle(efx_dev)->efx;
++}
++EXPORT_SYMBOL(efx_dl_get_nic);
+--- linux-2.6.18.8/drivers/net/sfc/driverlink.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/driverlink.h 2008-05-19 00:33:28.833808141 +0300
+@@ -0,0 +1,93 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/capability.h>
-+#include <linux/errno.h>
-+#include <linux/stat.h>
-+#include <linux/string.h>
-+#include <linux/sysdev.h>
-+#include "common.h"
++#ifndef EFX_DRIVERLINK_H
++#define EFX_DRIVERLINK_H
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
++/* Forward declarations */
++struct efx_dl_device;
++struct efx_nic;
++
++/*
++ * Efx driverlink
++ *
++ * This header file defines the portions of the Efx driverlink
++ * interface that are used only within the sfc module. It also
++ * declares efx_dl_get_nic(), which may be used by sfc_mtd
++ * and any other module built along with sfc.
++ */
++
++
++/* Efx callback devices
++ *
++ * A list of the devices that own each callback. The partner to
++ * struct efx_dl_callbacks
++ */
++struct efx_dl_cb_devices {
++ /* Device owning the tx_packet callback */
++ struct efx_dl_device *tx_packet;
++ /* Device owning the rx_packet callback */
++ struct efx_dl_device *rx_packet;
++ /* Device owning the link_change callback. */
++ struct efx_dl_device *link_change;
++ /* Device owning the request_mtu callback. */
++ struct efx_dl_device *request_mtu;
++ /* Device owning the mtu_changed callback. */
++ struct efx_dl_device *mtu_changed;
++ /* Device owning the event callback. */
++ struct efx_dl_device *event;
++};
++
++/* No-op callbacks used for initialisation */
++extern struct efx_dl_callbacks efx_default_callbacks;
++
++/* Macro used to invoke callbacks */
++#define EFX_DL_CALLBACK(_port, _name, ...) \
++ (_port)->dl_cb._name((_port)->dl_cb_dev._name, __VA_ARGS__)
++
++/* Register an Efx NIC */
++extern int efx_dl_register_nic(struct efx_nic *efx);
++
++/* Unregister an Efx NIC */
++extern void efx_dl_unregister_nic(struct efx_nic *efx);
++
++/* Lock the driverlink layer prior to a reset */
++extern void efx_dl_reset_lock(void);
++
++/* Unlock the driverlink layer following a reset */
++extern void efx_dl_reset_unlock(void);
++
++/* Suspend all drivers prior to a hardware reset */
++extern void efx_dl_reset_suspend(struct efx_nic *efx);
++
++/* Resume all drivers after a hardware reset */
++extern void efx_dl_reset_resume(struct efx_nic *efx, int ok);
++
++/* Obtain the Efx NIC for the given driverlink device. */
++extern struct efx_nic *efx_dl_get_nic(struct efx_dl_device *efx_dev);
++
++#endif /* EFX_DRIVERLINK_H */
+--- linux-2.6.18.8/drivers/net/sfc/driverlink_api.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/driverlink_api.h 2008-05-19 00:33:28.833808141 +0300
+@@ -0,0 +1,612 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_DRIVERLINK_API_H
++#define EFX_DRIVERLINK_API_H
++
++#include <linux/list.h> /* for struct list_head */
++#if !defined(EFX_USE_FASTCALL)
++ #include <linux/version.h>
++ #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)
++ #define EFX_USE_FASTCALL yes
++ #include <linux/linkage.h>
++ #endif
+#endif
+
-+#define BALLOON_CLASS_NAME "memory"
++/**
++ * DOC: Efx driverlink API
++ *
++ * This file must be included by any driver that wishes to attach to
++ * devices claimed by the Solarflare NIC driver (sfc). It allows separate
++ * kernel modules to expose other functionality offered by the NIC, with
++ * the sfc driver remaining in overall control.
++ *
++ * Overview:
++ *
++ * Driverlink clients define a &struct efx_dl_driver, and register
++ * this structure with the driverlink layer using
++ * efx_dl_register_driver(), which is exported by the sfc driver.
++ *
++ * The probe() routine of each driverlink client driver is called by
++ * the driverlink layer for each physical port in the system, after
++ * the sfc driver has performed start-of-day hardware initialisation
++ * and self-test. If ports are added or removed via pci hotplug then
++ * the &struct efx_dl_driver probe() or remove() routines are called
++ * as appropriate.
++ *
++ * If the port doesn't provide the necessary hardware resources for a
++ * client, then that client can return failure from its probe()
++ * routine. Information provided to the client driver at probe time
++ * includes
++ *
++ * Each probe() routine is given a unique &struct efx_dl_device per
++ * port, which means it can safely use the @priv member to store any
++ * useful state it needs. The probe routine also has the opportunity
++ * to provide a &struct efx_dl_callbacks via
++ * efx_dl_register_callbacks(), which allows the client to intercept
++ * the sfc driver's operations at strategic points.
++ *
++ * Occasionally, the underlying Efx device may need to be reset to
++ * recover from an error condition. The client's reset_suspend() and
++ * reset_resume() methods [if provided] will be called to enable the
++ * client to suspend operations and preserve any state before the
++ * reset. The client can itself request a reset using efx_dl_reset()
++ * or efx_dl_schedule_reset(), should it detect an error condition
++ * necessitating a reset.
++ *
++ * Example:
++ *
++ * The MTD driver (mtd.c) uses the driverlink layer.
++ */
+
-+#define BALLOON_SHOW(name, format, args...) \
-+ static ssize_t show_##name(struct sys_device *dev, \
-+ char *buf) \
-+ { \
-+ return sprintf(buf, format, ##args); \
-+ } \
-+ static SYSDEV_ATTR(name, S_IRUGO, show_##name, NULL)
++/* Forward declarations */
++struct pci_dev;
++struct net_device;
++struct sk_buff;
++struct efx_dl_device;
++struct efx_dl_device_info;
+
-+BALLOON_SHOW(current_kb, "%lu\n", PAGES2KB(bs.current_pages));
-+BALLOON_SHOW(low_kb, "%lu\n", PAGES2KB(bs.balloon_low));
-+BALLOON_SHOW(high_kb, "%lu\n", PAGES2KB(bs.balloon_high));
-+BALLOON_SHOW(hard_limit_kb,
-+ (bs.hard_limit!=~0UL) ? "%lu\n" : "???\n",
-+ (bs.hard_limit!=~0UL) ? PAGES2KB(bs.hard_limit) : 0);
-+BALLOON_SHOW(driver_kb, "%lu\n", PAGES2KB(bs.driver_pages));
++/*
++ * This is used to guard against the registration of driverlink
++ * clients using an incorrect version of the API.
++ */
++#define EFX_DRIVERLINK_API_VERSION 1
+
-+static ssize_t show_target_kb(struct sys_device *dev, char *buf)
-+{
-+ return sprintf(buf, "%lu\n", PAGES2KB(bs.target_pages));
-+}
+
-+static ssize_t store_target_kb(struct sys_device *dev,
-+ const char *buf,
-+ size_t count)
-+{
-+ char memstring[64], *endchar;
-+ unsigned long long target_bytes;
++/**
++ * struct efx_dl_driver - An Efx driverlink device driver
++ *
++ * This is the analogue of a struct pci_driver for a normal PCI
++ * driver. Driverlink clients should register themselves using
++ * efx_dl_register_driver() at module initialisation, and deregister
++ * themselves using efx_dl_unregister_driver() at module exit.
++ *
++ * All calls to members of efx_dl_driver are serialised by a single
++ * semaphore, so you are allowed to sleep in these functions. Take care
++ * to not call driverlink methods from within these callbacks, otherwise
++ * a deadlock is possible.
++ *
++ * @name: Name of the driver
++ * @probe: Called when device added
++ * @remove: Called when device removed
++ * @reset_suspend: Called before device is reset
++ * @reset_resume: Called after device is reset
++ */
++struct efx_dl_driver {
++ const char *name;
+
-+ if (!capable(CAP_SYS_ADMIN))
-+ return -EPERM;
-+
-+ if (count <= 1)
-+ return -EBADMSG; /* runt */
-+ if (count > sizeof(memstring))
-+ return -EFBIG; /* too long */
-+ strcpy(memstring, buf);
-+
-+ target_bytes = memparse(memstring, &endchar);
-+ balloon_set_new_target(target_bytes >> PAGE_SHIFT);
-+
-+ return count;
-+}
++ /*
++ * probe - Handle device addition.
++ * @efx_dev: Efx driverlink device
++ * @net_dev: The net_dev relevant to this port
++ * @dev_info: A linked list of device information.
++ * @silicon_rev: Silicon revision name.
++ *
++ * This will be called after driverlink client registration for
++ * every port on the system, and for every port that appears
++ * thereafter via hotplug.
++ *
++ * The client may use either @efx_dev->pci_dev, the dev_info linked
++ * list of available driver information, or the silicon revision
++ * name to determine if they can support this port. If they can,
++ * they should return 0 to indicate the probe was successful. Any
++ * other return code indicates that the probe failed, and the
++ * @efx_dl_dev will be invalidated.
++ *
++ * The client should perform whatever initialisation it
++ * requires, and store a pointer to its private data in
++ * @efx_dl_dev->priv (which is not shared between clients).
++ * It may also wish to hook in a callbacks table using
++ * efx_dl_register_callbacks().
++ *
++ * Return a negative error code or 0 on success.
++ */
++ int (*probe) (struct efx_dl_device *efx_dl_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char *silicon_rev);
+
-+static SYSDEV_ATTR(target_kb, S_IRUGO | S_IWUSR,
-+ show_target_kb, store_target_kb);
++ /*
++ * remove - Handle device removal.
++ * @efx_dev: Efx driverlink device
++ *
++ * This will be called at driver exit (or hotplug removal) for
++ * each registered driverlink client.
++ *
++ * The client must ensure that it has finished all operations
++ * using this device before returning from this method. If it
++ * has hooked in a callbacks table using
++ * efx_dl_register_callbacks(), it must unhook it using
++ * efx_dl_unregister_callbacks(), and then ensure that all
++ * callback-triggered operations (e.g. scheduled tasklets)
++ * have completed before returning. (It does not need to
++ * explicitly wait for callback methods to finish executing,
++ * since efx_dl_unregister_callbacks() will sleep until all
++ * callbacks have returned anyway.)
++ *
++ * Note that the device itself may not have been removed; it
++ * may be simply that the client is being unloaded
++ * via efx_dl_unregister_driver(). In this case other clients
++ * (and the sfc driver itself) will still be using the device,
++ * so the client cannot assume that the device itself is quiescent.
++ * In particular, callbacks may continue to be triggered at any
++ * point until efx_dl_unregister_callbacks() is called.
++ */
++ void (*remove) (struct efx_dl_device *efx_dev);
+
-+static struct sysdev_attribute *balloon_attrs[] = {
-+ &attr_target_kb,
++ /*
++ * reset_suspend - Suspend ready for reset.
++ * @efx_dev: Efx driverlink device
++ *
++ * This method will be called immediately before a hardware
++ * reset (which may or may not have been initiated by the
++ * driverlink client). This client must save any state that it
++ * will need to restore after the reset, and suspend all
++ * operations that might access the hardware. It must not
++ * return until the client can guarantee to have stopped
++ * touching the hardware.
++ *
++ * It is guaranteed that callbacks will be inactive by the
++ * time this method is called; the driverlink layer will
++ * already have prevented new callbacks being made and waited
++ * for all callbacks functions to return before calling
++ * reset_suspend(). However, any delayed work scheduled by
++ * the callback functions (e.g. tasklets) may not yet have
++ * completed.
++ *
++ * This method is allowed to sleep, so waiting on tasklets,
++ * work queues etc. is permitted. There will always be a
++ * corresponding call to the reset_resume() method, so it is
++ * safe to e.g. down a semaphore within reset_suspend() and up
++ * it within reset_resume(). (However, you obviously cannot
++ * do the same with a spinlock).
++ *
++ * Note that the reset operation may be being carried out in
++ * the context of scheduled work, so you cannot use
++ * flush_scheduled_work() to ensure that any work you may have
++ * scheduled has completed.
++ *
++ * During hardware reset, there is a chance of receiving
++ * spurious interrupts, so the client's ISR (if any) should be
++ * unhooked or otherwise disabled.
++ */
++ void (*reset_suspend) (struct efx_dl_device *efx_dev);
++
++ /*
++ * reset_resume - Restore after a reset.
++ * @efx_dev: Efx driverlink device
++ * @ok: Reset success indicator
++ *
++ * This method will be called after a hardware reset. There
++ * will always have been a corresponding call to the
++ * reset_suspend() method beforehand.
++ *
++ * If @ok is non-zero, the client should restore the state
++ * that it saved during the call to reset_suspend() and resume
++ * normal operations.
++ *
++ * If @ok is zero, the reset operation has failed and the
++ * hardware is currently in an unusable state. In this case,
++ * the client should release any locks taken out by
++ * reset_suspend(), but should not take any other action; in
++ * particular, it must not access the hardware, nor resume
++ * normal operations. The hardware is effectively dead at
++ * this point, and our sole aim is to avoid deadlocking or
++ * crashing the host.
++ *
++ * The driverlink layer will still be locked when
++ * reset_resume() is called, so the client may not call
++ * driverlink functions. In particular, if the reset failed,
++ * the client must not call efx_dl_unregister_callbacks() at
++ * this point; it should wait until remove() is called.
++ */
++ void (*reset_resume) (struct efx_dl_device *efx_dev, int ok);
++
++/* private: */
++ struct list_head node;
++ struct list_head device_list;
+};
+
-+static struct attribute *balloon_info_attrs[] = {
-+ &attr_current_kb.attr,
-+ &attr_low_kb.attr,
-+ &attr_high_kb.attr,
-+ &attr_hard_limit_kb.attr,
-+ &attr_driver_kb.attr,
-+ NULL
++/**
++ * DOC: Efx driverlink device information
++ *
++ * Each &struct efx_dl_device makes certain hardware resources visible
++ * to driverlink clients, and they describe which resources are
++ * available by passing a linked list of &struct efx_dl_device_info
++ * into the probe() routine.
++ *
++ * The driverlink client's probe function can iterate through the linked list,
++ * and provided that it understands the resources that are exported, it can
++ * choose to make use of them through an external interface.
++ */
++
++/**
++ * enum efx_dl_device_info_type - Device information identifier.
++ *
++ * Each distinct hardware resource API will have a member in this
++ * enumeration.
++ *
++ * @EFX_DL_FALCON_RESOURCES: Information type is &struct efx_dl_falcon_resources
++ */
++enum efx_dl_device_info_type {
++ /** Falcon resources available for export */
++ EFX_DL_FALCON_RESOURCES = 0,
+};
+
-+static struct attribute_group balloon_info_group = {
-+ .name = "info",
-+ .attrs = balloon_info_attrs,
++/**
++ * struct efx_dl_device_info - device information structure
++ * @next: Link to next structure, if any
++ * @type: Type code for this structure
++ *
++ * This structure is embedded in other structures provided by the
++ * driverlink device provider, and implements a linked list of
++ * resources pertinent to a driverlink client.
++ *
++ * Example: &struct efx_dl_falcon_resources
++ */
++struct efx_dl_device_info {
++ struct efx_dl_device_info *next;
++ enum efx_dl_device_info_type type;
+};
+
-+static struct sysdev_class balloon_sysdev_class = {
-+ set_kset_name(BALLOON_CLASS_NAME),
++/**
++ * enum efx_dl_falcon_resource_flags - Falcon resource information flags.
++ *
++ * Flags that describe hardware variations for the described Falcon based port.
++ *
++ * @EFX_DL_FALCON_DUAL_FUNC: Port is dual-function.
++ * Certain silicon revisions have two pci functions, and require
++ * certain hardware resources to be accessed via the secondary
++ * function. See the discussion of @pci_dev in &struct efx_dl_device
++ * below.
++ * @EFX_DL_FALCON_USE_MSI: Port is initialised to use MSI/MSI-X interrupts.
++ * Falcon supports traditional legacy interrupts and MSI/MSI-X
++ * interrupts. Since the sfc driver supports either, as a run
++ * time configuration, driverlink drivers need to be aware of which
++ * one to use for their interrupting resources.
++ */
++enum efx_dl_falcon_resource_flags {
++ EFX_DL_FALCON_DUAL_FUNC = 0x1,
++ EFX_DL_FALCON_USE_MSI = 0x2,
+};
+
-+static struct sys_device balloon_sysdev;
++/**
++ * struct efx_dl_falcon_resources - Falcon resource information.
++ *
++ * This structure describes Falcon hardware resources available for
++ * use by a driverlink driver.
++ *
++ * @hdr: Resource linked list header
++ * @biu_lock: Register access lock.
++ * Some Falcon revisions require register access for configuration
++ * registers to be serialised between ports and PCI functions.
++ * The sfc driver will provide the appropriate lock semantics for
++ * the underlying hardware.
++ * @buffer_table_min: First available buffer table entry
++ * @buffer_table_max: Last available buffer table entry + 1
++ * @evq_timer_min: First available event queue with timer
++ * @evq_timer_max: Last available event queue with timer + 1
++ * @evq_int_min: First available event queue with interrupt
++ * @evq_int_max: Last available event queue with interrupt + 1
++ * @rxq_min: First available RX queue
++ * @rxq_max: Last available RX queue + 1
++ * @txq_min: First available TX queue
++ * @txq_max: Last available TX queue + 1
++ * @flags: Hardware variation flags
++ */
++struct efx_dl_falcon_resources {
++ struct efx_dl_device_info hdr;
++ spinlock_t *biu_lock;
++ unsigned buffer_table_min, buffer_table_max;
++ unsigned evq_timer_min, evq_timer_max;
++ unsigned evq_int_min, evq_int_max;
++ unsigned rxq_min, rxq_max;
++ unsigned txq_min, txq_max;
++ enum efx_dl_falcon_resource_flags flags;
++};
+
-+static int register_balloon(struct sys_device *sysdev)
-+{
-+ int i, error;
++/**
++ * struct efx_dl_device - An Efx driverlink device.
++ *
++ * @pci_dev: Underlying PCI device.
++ * This is the PCI device used by the sfc driver. It will
++ * already have been enabled for bus-mastering DMA etc.
++ * @priv: Driver private data
++ * Driverlink clients can use this to store a pointer to their
++ * internal per-device data structure. Each (driver, device)
++ * tuple has a separate &struct efx_dl_device, so clients can use
++ * this @priv field independently.
++ * @driver: Efx driverlink driver for this device
++ */
++struct efx_dl_device {
++ struct pci_dev *pci_dev;
++ void *priv;
++ struct efx_dl_driver *driver;
++};
+
-+ error = sysdev_class_register(&balloon_sysdev_class);
-+ if (error)
-+ return error;
++/**
++ * enum efx_veto - Packet veto request flag.
++ *
++ * This is the return type for the rx_packet() and tx_packet() methods
++ * in &struct efx_dl_callbacks.
++ *
++ * @EFX_ALLOW_PACKET: Packet may be transmitted/received
++ * @EFX_VETO_PACKET: Packet must not be transmitted/received
++ */
++enum efx_veto {
++ EFX_ALLOW_PACKET = 0,
++ EFX_VETO_PACKET = 1,
++};
+
-+ sysdev->id = 0;
-+ sysdev->cls = &balloon_sysdev_class;
++/**
++ * struct efx_dl_callbacks - Efx callbacks
++ *
++ * These methods can be hooked in to the sfc driver via
++ * efx_dl_register_callbacks(). They allow clients to intercept and/or
++ * modify the behaviour of the sfc driver at predetermined points.
++ *
++ * For efficiency, only one client can hook each callback.
++ *
++ * Since these callbacks are called on packet transmit and reception
++ * paths, clients should avoid acquiring locks or allocating memory.
++ *
++ * @tx_packet: Called when packet is about to be transmitted
++ * @rx_packet: Called when packet is received
++ * @link_change: Called when link status has changed
++ * @request_mtu: Called to request MTU change
++ * @mtu_changed: Called when MTU has been changed
++ * @event: Called when NIC event is not handled by the sfc driver
++ */
++struct efx_dl_callbacks {
++ /*
++ * tx_packet - Packet about to be transmitted.
++ * @efx_dev: Efx driverlink device
++ * @skb: Socket buffer containing the packet to be sent
++ *
++ * This method is called for every packet about to be
++ * transmitted. It allows the client to snoop on traffic sent
++ * via the kernel queues.
++ *
++ * The method may return %EFX_VETO_PACKET in order to prevent
++ * the sfc driver from transmitting the packet. The net
++ * driver will then discard the packet. If the client wishes
++ * to retain a reference to the packet data after returning
++ * %EFX_VETO_PACKET, it must obtain its own copy of the
++ * packet (e.g. by calling skb_get(), or by copying out the
++ * packet data to an external buffer).
++ *
++ * This method must return quickly, since it will have a
++ * direct performance impact upon the sfc driver. It will be
++ * called with interrupts disabled (and may be called in
++ * interrupt context), so may not sleep. Since the sfc driver
++ * may have multiple TX queues, running in parallel, please avoid
++ * the need for locking if it all possible.
++ */
++#if defined(EFX_USE_FASTCALL)
++ enum efx_veto fastcall (*tx_packet) (struct efx_dl_device *efx_dev,
++ struct sk_buff *skb);
++#else
++ enum efx_veto (*tx_packet) (struct efx_dl_device *efx_dev,
++ struct sk_buff *skb);
++#endif
+
-+ error = sysdev_register(sysdev);
-+ if (error) {
-+ sysdev_class_unregister(&balloon_sysdev_class);
-+ return error;
-+ }
++ /*
++ * rx_packet - Packet received.
++ * @efx_dev: Efx driverlink device
++ * @pkt_hdr: Pointer to received packet
++ * @pkt_len: Length of received packet
++ *
++ * This method is called for every received packet. It allows
++ * the client to snoop on traffic received by the kernel
++ * queues.
++ *
++ * The method may return %EFX_VETO_PACKET in order to prevent
++ * the sfc driver from passing the packet to the kernel. The net
++ * driver will then discard the packet.
++ *
++ * This method must return quickly, since it will have a
++ * direct performance impact upon the sfc driver. It is
++ * called in tasklet context, so may not sleep. Note that
++ * there are per-channel tasklets in the sfc driver, so
++ * rx_packet() may be called simultaneously on different CPUs
++ * and must lock appropriately. The design of the sfc driver
++ * allows for lockless operation between receive channels, so
++ * please avoid the need for locking if at all possible.
++ */
++#if defined(EFX_USE_FASTCALL)
++ enum efx_veto fastcall (*rx_packet) (struct efx_dl_device *efx_dev,
++ const char *pkt_hdr, int pkt_len);
++#else
++ enum efx_veto (*rx_packet) (struct efx_dl_device *efx_dev,
++ const char *pkt_hdr, int pkt_len);
++#endif
+
-+ for (i = 0; i < ARRAY_SIZE(balloon_attrs); i++) {
-+ error = sysdev_create_file(sysdev, balloon_attrs[i]);
-+ if (error)
-+ goto fail;
-+ }
++ /*
++ * link_change - Link status change.
++ * @efx_dev: Efx driverlink device
++ * @link_up: Link up indicator
++ *
++ * This method is called to inform the driverlink client
++ * whenever the PHY link status changes. By the time this
++ * function is called, the MAC has already been reconfigured
++ * with the new autonegotiation settings from the PHY.
++ *
++ * This method is called from tasklet context and may not
++ * sleep.
++ */
++ void (*link_change) (struct efx_dl_device *efx_dev, int link_up);
+
-+ error = sysfs_create_group(&sysdev->kobj, &balloon_info_group);
-+ if (error)
-+ goto fail;
-+
-+ return 0;
++ /*
++ * request_mtu: Request MTU change.
++ * @efx_dev: Efx driverlink device
++ * @new_mtu: Requested new MTU
++ *
++ * This method is called whenever the user requests an MTU
++ * change on an interface. The client may return an error, in
++ * which case the MTU change request will be denied. If the
++ * client returns success, the MAC will be reconfigured with a
++ * new maxmimum frame length equal to
++ * EFX_MAX_FRAME_LEN(new_mtu). The client will be notified
++ * via the mtu_changed() method once the MAC has been
++ * reconfigured.
++ *
++ * The current MTU for the port can be obtained via
++ * efx_dl_get_netdev(efx_dl_device)->mtu.
++ *
++ * The sfc driver guarantees that no other callback functions
++ * are in progress when this method is called. This function
++ * is called in process context and may sleep.
++ *
++ * Return a negative error code or 0 on success.
++ */
++ int (*request_mtu) (struct efx_dl_device *efx_dev, int new_mtu);
+
-+ fail:
-+ while (--i >= 0)
-+ sysdev_remove_file(sysdev, balloon_attrs[i]);
-+ sysdev_unregister(sysdev);
-+ sysdev_class_unregister(&balloon_sysdev_class);
-+ return error;
-+}
++ /*
++ * mtu_changed - MTU has been changed.
++ * @efx_dev: Efx driverlink device
++ * @mtu: The new MTU
++ *
++ * This method is called once the MAC has been reconfigured
++ * with a new MTU. There will have been a preceding call to
++ * request_mtu().
++ *
++ * The sfc driver guarantees that no other callback functions
++ * are in progress when this method is called. This function
++ * is called in process context and may sleep.
++ */
++ void (*mtu_changed) (struct efx_dl_device *efx_dev, int mtu);
+
-+static void unregister_balloon(struct sys_device *sysdev)
-+{
-+ int i;
++ /*
++ * event - Event callback.
++ * @efx_dev: Efx driverlink device
++ * @p_event: Pointer to event
++ *
++ * This method is called for each event that is not handled by the
++ * sfc driver.
++ */
++ void (*event) (struct efx_dl_device *efx_dev, void *p_event);
++};
+
-+ sysfs_remove_group(&sysdev->kobj, &balloon_info_group);
-+ for (i = 0; i < ARRAY_SIZE(balloon_attrs); i++)
-+ sysdev_remove_file(sysdev, balloon_attrs[i]);
-+ sysdev_unregister(sysdev);
-+ sysdev_class_unregister(&balloon_sysdev_class);
-+}
++/* Include API version number in symbol used for efx_dl_register_driver */
++#define efx_dl_stringify_1(x, y) x ## y
++#define efx_dl_stringify_2(x, y) efx_dl_stringify_1(x, y)
++#define efx_dl_register_driver \
++ efx_dl_stringify_2(efx_dl_register_driver_api_ver_, \
++ EFX_DRIVERLINK_API_VERSION)
+
-+int balloon_sysfs_init(void)
-+{
-+ return register_balloon(&balloon_sysdev);
-+}
++extern int efx_dl_register_driver(struct efx_dl_driver *driver);
+
-+void balloon_sysfs_exit(void)
-+{
-+ unregister_balloon(&balloon_sysdev);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/blkback.c linux-2.6.18-xen.hg/drivers/xen/blkback/blkback.c
---- linux-2.6.18/drivers/xen/blkback/blkback.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/blkback.c 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,617 @@
-+/******************************************************************************
-+ * arch/xen/drivers/blkif/backend/main.c
-+ *
-+ * Back-end of the driver for virtual block devices. This portion of the
-+ * driver exports a 'unified' block-device interface that can be accessed
-+ * by any operating system that implements a compatible front end. A
-+ * reference front-end implementation can be found in:
-+ * arch/xen/drivers/blkif/frontend
-+ *
-+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
-+ * Copyright (c) 2005, Christopher Clark
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++extern void efx_dl_unregister_driver(struct efx_dl_driver *driver);
++
++extern int efx_dl_register_callbacks(struct efx_dl_device *efx_dev,
++ struct efx_dl_callbacks *callbacks);
++
++extern void efx_dl_unregister_callbacks(struct efx_dl_device *efx_dev,
++ struct efx_dl_callbacks *callbacks);
++
++extern void efx_dl_schedule_reset(struct efx_dl_device *efx_dev);
++
++/**
++ * efx_dl_for_each_device_info_matching - iterate an efx_dl_device_info list
++ * @_dev_info: Pointer to first &struct efx_dl_device_info
++ * @_type: Type code to look for
++ * @_info_type: Structure type corresponding to type code
++ * @_field: Name of &struct efx_dl_device_info field in the type
++ * @_p: Iterator variable
++ *
++ * Example:
++ *
++ * static int driver_dl_probe(... const struct efx_dl_device_info *dev_info ...)
++ * {
++ * struct efx_dl_falcon_resources *res;
++ *
++ * efx_dl_for_each_device_info_matching(dev_info,EFX_DL_FALCON_RESOURCES,
++ * struct efx_dl_falcon_resources,
++ * hdr, res) {
++ * if (res->flags & EFX_DL_FALCON_DUAL_FUNC) {
++ * .....
++ * }
++ * }
++ * }
++ */
++#define efx_dl_for_each_device_info_matching(_dev_info, _type, \
++ _info_type, _field, _p) \
++ for ((_p) = container_of((_dev_info), _info_type, _field); \
++ (_p) != NULL; \
++ (_p) = container_of((_p)->_field.next, _info_type, _field))\
++ if ((_p)->_field.type != _type) \
++ continue; \
++ else
++
++/**
++ * efx_dl_search_device_info - search an efx_dl_device_info list
++ * @_dev_info: Pointer to first &struct efx_dl_device_info
++ * @_type: Type code to look for
++ * @_info_type: Structure type corresponding to type code
++ * @_field: Name of &struct efx_dl_device_info member in this type
++ * @_p: Result variable
++ *
++ * Example:
++ *
++ * static int driver_dl_probe(... const struct efx_dl_device_info *dev_info ...)
++ * {
++ * struct efx_dl_falcon_resources *res;
++ *
++ * efx_dl_search_device_info(dev_info, EFX_DL_FALCON_RESOURCES,
++ * struct efx_dl_falcon_resources, hdr, res);
++ * if (res != NULL) {
++ * ....
++ * }
++ * }
++ */
++#define efx_dl_search_device_info(_dev_info, _type, _info_type, \
++ _field, _p) \
++ efx_dl_for_each_device_info_matching((_dev_info), (_type), \
++ _info_type, _field, (_p)) \
++ break;
++
++#endif /* EFX_DRIVERLINK_API_H */
+--- linux-2.6.18.8/drivers/net/sfc/efx.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/efx.c 2008-05-19 00:33:28.837808372 +0300
+@@ -0,0 +1,2783 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/spinlock.h>
-+#include <linux/kthread.h>
-+#include <linux/list.h>
-+#include <xen/balloon.h>
-+#include <asm/hypervisor.h>
-+#include "common.h"
++#include <linux/module.h>
++#include <linux/pci.h>
++#include <linux/netdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/delay.h>
++#include <linux/notifier.h>
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/in.h>
++#include <linux/crc32.h>
++#include <linux/ethtool.h>
++#include <asm/uaccess.h>
++#include "net_driver.h"
++#include "gmii.h"
++#include "driverlink.h"
++#include "selftest.h"
++#include "debugfs.h"
++#include "ethtool.h"
++#include "tx.h"
++#include "rx.h"
++#include "efx.h"
++#include "mdio_10g.h"
++#include "falcon.h"
++#include "workarounds.h"
+
-+/*
-+ * These are rather arbitrary. They are fairly large because adjacent requests
-+ * pulled from a communication ring are quite likely to end up being part of
-+ * the same scatter/gather request at the disc.
-+ *
-+ * ** TRY INCREASING 'blkif_reqs' IF WRITE SPEEDS SEEM TOO LOW **
-+ *
-+ * This will increase the chances of being able to write whole tracks.
-+ * 64 should be enough to keep us competitive with Linux.
++/**************************************************************************
++ *
++ * Type name strings
++ *
++ **************************************************************************
+ */
-+static int blkif_reqs = 64;
-+module_param_named(reqs, blkif_reqs, int, 0);
-+MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate");
+
-+/* Run-time switchable: /sys/module/blkback/parameters/ */
-+static unsigned int log_stats = 0;
-+static unsigned int debug_lvl = 0;
-+module_param(log_stats, int, 0644);
-+module_param(debug_lvl, int, 0644);
++/* Loopback mode names (see LOOPBACK_MODE()) */
++const unsigned int efx_loopback_mode_max = LOOPBACK_MAX;
++const char *efx_loopback_mode_names[] = {
++ [LOOPBACK_NONE] = "NONE",
++ [LOOPBACK_MAC] = "MAC",
++ [LOOPBACK_XGMII] = "XGMII",
++ [LOOPBACK_XGXS] = "XGXS",
++ [LOOPBACK_XAUI] = "XAUI",
++ [LOOPBACK_PHY] = "PHY",
++ [LOOPBACK_PHYXS] = "PHY(XS)",
++ [LOOPBACK_PCS] = "PHY(PCS)",
++ [LOOPBACK_PMAPMD] = "PHY(PMAPMD)",
++ [LOOPBACK_NETWORK] = "NETWORK",
++};
++
++/* Interrupt mode names (see INT_MODE())) */
++const unsigned int efx_interrupt_mode_max = EFX_INT_MODE_MAX;
++const char *efx_interrupt_mode_names[] = {
++ [EFX_INT_MODE_MSIX] = "MSI-X",
++ [EFX_INT_MODE_MSI] = "MSI",
++ [EFX_INT_MODE_LEGACY] = "legacy",
++};
++
++/* PHY type names (see PHY_TYPE())) */
++const unsigned int efx_phy_type_max = PHY_TYPE_MAX;
++const char *efx_phy_type_names[] = {
++ [PHY_TYPE_NONE] = "none",
++ [PHY_TYPE_CX4_RTMR] = "Mysticom CX4",
++ [PHY_TYPE_1G_ALASKA] = "1G Alaska",
++ [PHY_TYPE_10XPRESS] = "SFC 10Xpress",
++ [PHY_TYPE_XFP] = "Quake XFP",
++ [PHY_TYPE_PM8358] = "PM8358 XAUI",
++};
++
++const unsigned int efx_reset_type_max = RESET_TYPE_MAX;
++const char *efx_reset_type_names[] = {
++ [RESET_TYPE_INVISIBLE] = "INVISIBLE",
++ [RESET_TYPE_ALL] = "ALL",
++ [RESET_TYPE_WORLD] = "WORLD",
++ [RESET_TYPE_DISABLE] = "DISABLE",
++ [RESET_TYPE_MONITOR] = "MONITOR",
++ [RESET_TYPE_INT_ERROR] = "INT_ERROR",
++ [RESET_TYPE_RX_RECOVERY] = "RX_RECOVERY",
++};
++
++const unsigned int efx_nic_state_max = STATE_MAX;
++const char *efx_nic_state_names[] = {
++ [STATE_INIT] = "INIT",
++ [STATE_RUNNING] = "RUNNING",
++ [STATE_FINI] = "FINI",
++ [STATE_RESETTING] = "RESETTING",
++ [STATE_DISABLED] = "DISABLED",
++};
++
++#define EFX_MAX_MTU (9 * 1024)
++
++
++/**************************************************************************
++ *
++ * Configurable values
++ *
++ *************************************************************************/
+
+/*
-+ * Each outstanding request that we've passed to the lower device layers has a
-+ * 'pending_req' allocated to it. Each buffer_head that completes decrements
-+ * the pendcnt towards zero. When it hits zero, the specified domain has a
-+ * response queued for it, with the saved 'id' passed back.
++ * Use separate channels for TX and RX events
++ *
++ * Set this to 1 to use separate channels for TX and RX. It allows us to
++ * apply a higher level of interrupt moderation to TX events.
++ *
++ * This is forced to 0 for MSI interrupt mode as the interrupt vector
++ * is not written
+ */
-+typedef struct {
-+ blkif_t *blkif;
-+ u64 id;
-+ int nr_pages;
-+ atomic_t pendcnt;
-+ unsigned short operation;
-+ int status;
-+ struct list_head free_list;
-+} pending_req_t;
++static unsigned int separate_tx_and_rx_channels = 1;
+
-+static pending_req_t *pending_reqs;
-+static struct list_head pending_free;
-+static DEFINE_SPINLOCK(pending_free_lock);
-+static DECLARE_WAIT_QUEUE_HEAD(pending_free_wq);
++/* This is the weight assigned to each of the (per-channel) virtual
++ * NAPI devices.
++ */
++static int napi_weight = 64;
+
-+#define BLKBACK_INVALID_HANDLE (~0)
++/* This is the time (in jiffies) between invocations of the hardware
++ * monitor, which checks for known hardware bugs and resets the
++ * hardware and driver as necessary.
++ */
++unsigned int efx_monitor_interval = 1 * HZ;
+
-+static struct page **pending_pages;
-+static grant_handle_t *pending_grant_handles;
++/* This controls whether or not the hardware monitor will trigger a
++ * reset when it detects an error condition.
++ */
++static unsigned int monitor_reset = 1;
+
-+static inline int vaddr_pagenr(pending_req_t *req, int seg)
++/* This controls whether or not the driver will initialise devices
++ * with invalid MAC addresses stored in the EEPROM or flash. If true,
++ * such devices will be initialised with a random locally-generated
++ * MAC address. This allows for loading the efx_mtd driver to
++ * reprogram the flash, even if the flash contents (including the MAC
++ * address) have previously been erased.
++ */
++static unsigned int allow_bad_hwaddr;
++
++/* Initial interrupt moderation settings. They can be modified after
++ * module load with ethtool.
++ *
++ * The default for RX should strike a balance between increasing the
++ * round-trip latency and reducing overhead.
++ */
++static unsigned int rx_irq_mod_usec = 60;
++
++/* Initial interrupt moderation settings. They can be modified after
++ * module load with ethtool.
++ *
++ * This default is chosen to ensure that a 10G link does not go idle
++ * while a TX queue is stopped after it has become full. A queue is
++ * restarted when it drops below half full. The time this takes (assuming
++ * worst case 3 descriptors per packet and 1024 descriptors) is
++ * 512 / 3 * 1.2 = 205 usec.
++ */
++static unsigned int tx_irq_mod_usec = 150;
++
++/* Ignore online self-test failures at load
++ *
++ * If set to 1, then the driver will not fail to load
++ * if the online self-test fails. Useful only during testing
++ */
++static unsigned int allow_load_on_failure;
++
++/* Set to 1 to enable the use of Message-Signalled Interrupts (MSI).
++ * MSI will not work on some motherboards due to limitations of the
++ * chipset, so the default is off.
++ *
++ * This is the highest capability interrupt mode to use
++ * 0 => MSI-X
++ * 1 => MSI
++ * 2 => legacy
++ */
++static unsigned int interrupt_mode;
++
++/* If set to 1, then the driver will perform an offline self test
++ * when each interface first comes up. This will appear like the
++ * interface bounces up and down
++ */
++static unsigned int onload_offline_selftest = 1;
++
++/* This is the requested number of CPUs to use for Receive-Side Scaling (RSS),
++ * i.e. the number of CPUs among which we may distribute simultaneous
++ * interrupt handling.
++ *
++ * Cards without MSI-X will only target one CPU
++ *
++ * Default (0) means to use all CPUs in the system. This parameter
++ * can be set using "rss_cpus=xxx" when loading the module.
++ */
++static unsigned int rss_cpus;
++module_param(rss_cpus, uint, 0444);
++MODULE_PARM_DESC(rss_cpus, "Number of CPUs to use for Receive-Side Scaling");
++
++/**************************************************************************
++ *
++ * Utility functions and prototypes
++ *
++ *************************************************************************/
++static void efx_remove_channel(struct efx_channel *channel);
++static void efx_remove_port(struct efx_nic *efx);
++static void efx_fini_napi(struct efx_nic *efx);
++static void efx_fini_channels(struct efx_nic *efx);
++
++/**************************************************************************
++ *
++ * Event queue processing
++ *
++ *************************************************************************/
++
++/* Process channel's event queue
++ *
++ * This function is responsible for processing the event queue of a
++ * single channel. The caller must guarantee that this function will
++ * never be concurrently called more than once on the same channel,
++ * though different channels may be being processed concurrently.
++ */
++static inline int efx_process_channel(struct efx_channel *channel, int rx_quota)
+{
-+ return (req - pending_reqs) * BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
++ int rxdmaqs;
++ struct efx_rx_queue *rx_queue;
++
++ if (unlikely(channel->efx->reset_pending != RESET_TYPE_NONE ||
++ !channel->enabled))
++ return rx_quota;
++
++ rxdmaqs = falcon_process_eventq(channel, &rx_quota);
++
++ /* Deliver last RX packet. */
++ if (channel->rx_pkt) {
++ __efx_rx_packet(channel, channel->rx_pkt,
++ channel->rx_pkt_csummed);
++ channel->rx_pkt = NULL;
++ }
++
++ efx_rx_strategy(channel);
++
++ /* Refill descriptor rings as necessary */
++ rx_queue = &channel->efx->rx_queue[0];
++ while (rxdmaqs) {
++ if (rxdmaqs & 0x01)
++ efx_fast_push_rx_descriptors(rx_queue);
++ rx_queue++;
++ rxdmaqs >>= 1;
++ }
++
++ return rx_quota;
+}
+
-+static inline unsigned long vaddr(pending_req_t *req, int seg)
++/* Mark channel as finished processing
++ *
++ * Note that since we will not receive further interrupts for this
++ * channel before we finish processing and call the eventq_read_ack()
++ * method, there is no need to use the interrupt hold-off timers.
++ */
++static inline void efx_channel_processed(struct efx_channel *channel)
+{
-+ unsigned long pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, seg)]);
-+ return (unsigned long)pfn_to_kaddr(pfn);
++ /* Write to EVQ_RPTR_REG. If a new event arrived in a race
++ * with finishing processing, a new interrupt will be raised.
++ */
++ channel->work_pending = 0;
++ smp_wmb(); /* Ensure channel updated before any new interrupt. */
++ falcon_eventq_read_ack(channel);
+}
+
-+#define pending_handle(_req, _seg) \
-+ (pending_grant_handles[vaddr_pagenr(_req, _seg)])
++/* NAPI poll handler
++ *
++ * NAPI guarantees serialisation of polls of the same device, which
++ * provides the guarantee required by efx_process_channel().
++ */
++#if !defined(EFX_HAVE_OLD_NAPI)
++static int efx_poll(struct napi_struct *napi, int budget)
++{
++ struct efx_channel *channel =
++ container_of(napi, struct efx_channel, napi_str);
++ struct net_device *napi_dev = channel->napi_dev;
++#else
++static int efx_poll(struct net_device *napi, int *budget_ret)
++{
++ struct net_device *napi_dev = napi;
++ struct efx_channel *channel = napi_dev->priv;
++ int budget = min(napi_dev->quota, *budget_ret);
++#endif
++ int unused;
++ int rx_packets;
+
++ EFX_TRACE(channel->efx, "channel %d NAPI poll executing on CPU %d\n",
++ channel->channel, raw_smp_processor_id());
+
-+static int do_block_io_op(blkif_t *blkif);
-+static void dispatch_rw_block_io(blkif_t *blkif,
-+ blkif_request_t *req,
-+ pending_req_t *pending_req);
-+static void make_response(blkif_t *blkif, u64 id,
-+ unsigned short op, int st);
++ unused = efx_process_channel(channel, budget);
++ rx_packets = (budget - unused);
++#if defined(EFX_HAVE_OLD_NAPI)
++ napi_dev->quota -= rx_packets;
++ *budget_ret -= rx_packets;
++#endif
+
-+/******************************************************************
-+ * misc small helpers
++ if (rx_packets < budget) {
++ /* There is no race here; although napi_disable() will
++ * only wait for netif_rx_complete(), this isn't a problem
++ * since efx_channel_processed() will have no effect if
++ * interrupts have already been disabled.
++ */
++ netif_rx_complete(napi_dev, napi);
++ efx_channel_processed(channel);
++ }
++
++#if !defined(EFX_HAVE_OLD_NAPI)
++ return rx_packets;
++#else
++ return (rx_packets >= budget);
++#endif
++}
++
++/* Process the eventq of the specified channel immediately on this CPU
++ *
++ * Disable hardware generated interrupts, wait for any existing
++ * processing to finish, then directly poll (and ack ) the eventq.
++ * Finally reenable NAPI and interrupts.
++ *
++ * Since we are touching interrupts the caller should hold the suspend lock
+ */
-+static pending_req_t* alloc_req(void)
++void efx_process_channel_now(struct efx_channel *channel)
+{
-+ pending_req_t *req = NULL;
-+ unsigned long flags;
++ struct efx_nic *efx = channel->efx;
+
-+ spin_lock_irqsave(&pending_free_lock, flags);
-+ if (!list_empty(&pending_free)) {
-+ req = list_entry(pending_free.next, pending_req_t, free_list);
-+ list_del(&req->free_list);
-+ }
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
-+ return req;
++ BUG_ON(!channel->used_flags);
++ BUG_ON(!channel->enabled);
++
++ /* Disable interrupts and wait for ISRs to complete */
++ falcon_disable_interrupts(efx);
++ if (efx->legacy_irq)
++ synchronize_irq(efx->legacy_irq);
++ if (channel->has_interrupt && channel->irq)
++ synchronize_irq(channel->irq);
++
++ /* Wait for any NAPI processing to complete */
++ napi_disable(&channel->napi_str);
++
++ /* Poll the channel */
++ (void) efx_process_channel(channel, efx->type->evq_size);
++
++ /* Ack the eventq. This may cause an interrupt to be generated
++ * when they are reenabled */
++ efx_channel_processed(channel);
++
++ /* Reenable NAPI polling */
++ napi_enable(&channel->napi_str);
++
++ /* Reenable interrupts */
++ falcon_enable_interrupts(efx);
+}
+
-+static void free_req(pending_req_t *req)
++/* Create event queue
++ * Event queue memory allocations are done only once. If the channel
++ * is reset, the memory buffer will be reused; this guards against
++ * errors during channel reset and also simplifies interrupt handling.
++ */
++static int efx_probe_eventq(struct efx_channel *channel)
+{
-+ unsigned long flags;
-+ int was_empty;
++ EFX_LOG(channel->efx, "chan %d create event queue\n", channel->channel);
+
-+ spin_lock_irqsave(&pending_free_lock, flags);
-+ was_empty = list_empty(&pending_free);
-+ list_add(&req->free_list, &pending_free);
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
-+ if (was_empty)
-+ wake_up(&pending_free_wq);
++ return falcon_probe_eventq(channel);
+}
+
-+static void unplug_queue(blkif_t *blkif)
++/* Prepare channel's event queue */
++static int efx_init_eventq(struct efx_channel *channel)
+{
-+ if (blkif->plug == NULL)
-+ return;
-+ if (blkif->plug->unplug_fn)
-+ blkif->plug->unplug_fn(blkif->plug);
-+ blk_put_queue(blkif->plug);
-+ blkif->plug = NULL;
++ EFX_LOG(channel->efx, "chan %d init event queue\n", channel->channel);
++
++ ASSERT_RTNL();
++
++ /* Initialise fields */
++ channel->eventq_read_ptr = 0;
++
++ return falcon_init_eventq(channel);
+}
+
-+static void plug_queue(blkif_t *blkif, struct bio *bio)
++static void efx_fini_eventq(struct efx_channel *channel)
+{
-+ request_queue_t *q = bdev_get_queue(bio->bi_bdev);
++ EFX_LOG(channel->efx, "chan %d fini event queue\n", channel->channel);
+
-+ if (q == blkif->plug)
-+ return;
-+ unplug_queue(blkif);
-+ blk_get_queue(q);
-+ blkif->plug = q;
++ ASSERT_RTNL();
++
++ falcon_fini_eventq(channel);
+}
+
-+static void fast_flush_area(pending_req_t *req)
++static void efx_remove_eventq(struct efx_channel *channel)
+{
-+ struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-+ unsigned int i, invcount = 0;
-+ grant_handle_t handle;
-+ int ret;
++ EFX_LOG(channel->efx, "chan %d remove event queue\n", channel->channel);
+
-+ for (i = 0; i < req->nr_pages; i++) {
-+ handle = pending_handle(req, i);
-+ if (handle == BLKBACK_INVALID_HANDLE)
-+ continue;
-+ gnttab_set_unmap_op(&unmap[i], vaddr(req, i), GNTMAP_host_map,
-+ handle);
-+ pending_handle(req, i) = BLKBACK_INVALID_HANDLE;
-+ invcount++;
-+ }
-+
-+ ret = HYPERVISOR_grant_table_op(
-+ GNTTABOP_unmap_grant_ref, unmap, invcount);
-+ BUG_ON(ret);
++ falcon_remove_eventq(channel);
+}
+
-+/******************************************************************
-+ * SCHEDULER FUNCTIONS
++/**************************************************************************
++ *
++ * Channel handling
++ *
++ *************************************************************************/
++
++/* Setup per-NIC RX buffer parameters.
++ * Calculate the rx buffer allocation parameters required to support
++ * the current MTU, including padding for header alignment and overruns.
+ */
++static void efx_calc_rx_buffer_params(struct efx_nic *efx)
++{
++ unsigned int order, len;
+
-+static void print_stats(blkif_t *blkif)
++ len = (max(EFX_PAGE_IP_ALIGN, NET_IP_ALIGN) +
++ EFX_MAX_FRAME_LEN(efx->net_dev->mtu) +
++ efx->type->rx_buffer_padding);
++
++ /* Page-based allocation page-order */
++ for (order = 0; ((1u << order) * PAGE_SIZE) < len; ++order)
++ ;
++
++ efx->rx_buffer_len = len;
++ efx->rx_buffer_order = order;
++}
++
++static int efx_probe_channel(struct efx_channel *channel)
+{
-+ printk(KERN_DEBUG "%s: oo %3d | rd %4d | wr %4d | br %4d\n",
-+ current->comm, blkif->st_oo_req,
-+ blkif->st_rd_req, blkif->st_wr_req, blkif->st_br_req);
-+ blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
-+ blkif->st_rd_req = 0;
-+ blkif->st_wr_req = 0;
-+ blkif->st_oo_req = 0;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++ int rc;
++
++ EFX_LOG(channel->efx, "creating channel %d\n", channel->channel);
++
++ rc = efx_probe_eventq(channel);
++ if (rc)
++ goto fail1;
++
++ efx_for_each_channel_tx_queue(tx_queue, channel) {
++ rc = efx_probe_tx_queue(tx_queue);
++ if (rc)
++ goto fail2;
++ }
++
++ efx_for_each_channel_rx_queue(rx_queue, channel) {
++ rc = efx_probe_rx_queue(rx_queue);
++ if (rc)
++ goto fail3;
++ }
++
++ channel->n_rx_frm_trunc = 0;
++
++ return 0;
++
++ fail3:
++ efx_for_each_channel_rx_queue(rx_queue, channel)
++ efx_remove_rx_queue(rx_queue);
++ fail2:
++ efx_for_each_channel_tx_queue(tx_queue, channel)
++ efx_remove_tx_queue(tx_queue);
++ fail1:
++ return rc;
+}
+
-+int blkif_schedule(void *arg)
++
++/* Channels are shutdown and reinitialised whilst the NIC is running
++ * to propagate configuration changes (mtu, checksum offload), or
++ * to clear hardware error conditions
++ */
++static int efx_init_channels(struct efx_nic *efx)
+{
-+ blkif_t *blkif = arg;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++ struct efx_channel *channel;
++ int rc = 0;
+
-+ blkif_get(blkif);
++ /* Recalculate the rx buffer parameters */
++ efx_calc_rx_buffer_params(efx);
+
-+ if (debug_lvl)
-+ printk(KERN_DEBUG "%s: started\n", current->comm);
++ /* Initialise the channels */
++ efx_for_each_channel(channel, efx) {
++ EFX_LOG(channel->efx, "init chan %d\n", channel->channel);
+
-+ while (!kthread_should_stop()) {
-+ if (try_to_freeze())
-+ continue;
++ rc = efx_init_eventq(channel);
++ if (rc)
++ goto err;
+
-+ wait_event_interruptible(
-+ blkif->wq,
-+ blkif->waiting_reqs || kthread_should_stop());
-+ wait_event_interruptible(
-+ pending_free_wq,
-+ !list_empty(&pending_free) || kthread_should_stop());
++ efx_for_each_channel_tx_queue(tx_queue, channel) {
++ rc = efx_init_tx_queue(tx_queue);
++ if (rc)
++ goto err;
++ }
+
-+ blkif->waiting_reqs = 0;
-+ smp_mb(); /* clear flag *before* checking for work */
++ /* The rx buffer allocation strategy is MTU dependent */
++ efx_rx_strategy(channel);
+
-+ if (do_block_io_op(blkif))
-+ blkif->waiting_reqs = 1;
-+ unplug_queue(blkif);
++ efx_for_each_channel_rx_queue(rx_queue, channel) {
++ rc = efx_init_rx_queue(rx_queue);
++ if (rc)
++ goto err;
++ }
+
-+ if (log_stats && time_after(jiffies, blkif->st_print))
-+ print_stats(blkif);
++ WARN_ON(channel->rx_pkt != NULL);
++ efx_rx_strategy(channel);
+ }
+
-+ if (log_stats)
-+ print_stats(blkif);
-+ if (debug_lvl)
-+ printk(KERN_DEBUG "%s: exiting\n", current->comm);
++ return 0;
+
-+ blkif->xenblkd = NULL;
-+ blkif_put(blkif);
++ err:
++ EFX_ERR(efx, "failed to initialise channel %d\n",
++ channel ? channel->channel : -1);
++ efx_fini_channels(efx);
++ return rc;
++}
+
-+ return 0;
++/* This enables event queue processing and packet transmission.
++ *
++ * Note that this function is not allowed to fail, since that would
++ * introduce too much complexity into the suspend/resume path.
++ */
++static void efx_start_channel(struct efx_channel *channel)
++{
++ struct efx_rx_queue *rx_queue;
++
++ EFX_LOG(channel->efx, "starting chan %d\n", channel->channel);
++
++ if (!(channel->efx->net_dev->flags & IFF_UP))
++ netif_napi_add(channel->napi_dev, &channel->napi_str,
++ efx_poll, napi_weight);
++
++ /* Mark channel as enabled */
++ channel->work_pending = 0;
++ channel->enabled = 1;
++ smp_wmb(); /* ensure channel updated before first interrupt */
++
++ /* Enable NAPI poll handler */
++ napi_enable(&channel->napi_str);
++
++ /* Load up RX descriptors */
++ efx_for_each_channel_rx_queue(rx_queue, channel)
++ efx_fast_push_rx_descriptors(rx_queue);
+}
+
-+/******************************************************************
-+ * COMPLETION CALLBACK -- Called as bh->b_end_io()
++/* This disables event queue processing and packet transmission.
++ * This function does not guarantee that all queue processing
++ * (e.g. RX refill) is complete.
+ */
++static void efx_stop_channel(struct efx_channel *channel)
++{
++ struct efx_rx_queue *rx_queue;
+
-+static void __end_block_io_op(pending_req_t *pending_req, int error)
++ if (!channel->enabled)
++ return;
++
++ EFX_LOG(channel->efx, "stop chan %d\n", channel->channel);
++
++ /* Mark channel as disabled */
++ channel->enabled = 0;
++
++ /* Wait for any NAPI processing to complete */
++ napi_disable(&channel->napi_str);
++
++ /* Ensure that any worker threads have exited or will be
++ * no-ops.
++ */
++ efx_for_each_channel_rx_queue(rx_queue, channel) {
++ spin_lock_bh(&rx_queue->add_lock);
++ spin_unlock_bh(&rx_queue->add_lock);
++ }
++}
++
++static void efx_fini_channels(struct efx_nic *efx)
+{
-+ /* An error fails the entire request. */
-+ if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) &&
-+ (error == -EOPNOTSUPP)) {
-+ DPRINTK("blkback: write barrier op failed, not supported\n");
-+ blkback_barrier(XBT_NIL, pending_req->blkif->be, 0);
-+ pending_req->status = BLKIF_RSP_EOPNOTSUPP;
-+ } else if (error) {
-+ DPRINTK("Buffer not up-to-date at end of operation, "
-+ "error=%d\n", error);
-+ pending_req->status = BLKIF_RSP_ERROR;
++ struct efx_channel *channel;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++
++ ASSERT_RTNL();
++
++ efx_for_each_channel(channel, efx) {
++ EFX_LOG(channel->efx, "shut down chan %d\n", channel->channel);
++
++ efx_for_each_channel_rx_queue(rx_queue, channel)
++ efx_fini_rx_queue(rx_queue);
++ efx_for_each_channel_tx_queue(tx_queue, channel)
++ efx_fini_tx_queue(tx_queue);
+ }
+
-+ if (atomic_dec_and_test(&pending_req->pendcnt)) {
-+ fast_flush_area(pending_req);
-+ make_response(pending_req->blkif, pending_req->id,
-+ pending_req->operation, pending_req->status);
-+ blkif_put(pending_req->blkif);
-+ free_req(pending_req);
++ /* Do the event queues last so that we can handle flush events
++ * for all DMA queues. */
++ efx_for_each_channel(channel, efx) {
++ EFX_LOG(channel->efx, "shut down evq %d\n", channel->channel);
++
++ efx_fini_eventq(channel);
+ }
+}
+
-+static int end_block_io_op(struct bio *bio, unsigned int done, int error)
++static void efx_remove_channel(struct efx_channel *channel)
+{
-+ if (bio->bi_size != 0)
-+ return 1;
-+ __end_block_io_op(bio->bi_private, error);
-+ bio_put(bio);
-+ return error;
-+}
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
+
++ EFX_LOG(channel->efx, "destroy chan %d\n", channel->channel);
+
-+/******************************************************************************
-+ * NOTIFICATION FROM GUEST OS.
-+ */
++ efx_for_each_channel_rx_queue(rx_queue, channel)
++ efx_remove_rx_queue(rx_queue);
++ efx_for_each_channel_tx_queue(tx_queue, channel)
++ efx_remove_tx_queue(tx_queue);
++ efx_remove_eventq(channel);
+
-+static void blkif_notify_work(blkif_t *blkif)
-+{
-+ blkif->waiting_reqs = 1;
-+ wake_up(&blkif->wq);
++ channel->used_flags = 0;
+}
+
-+irqreturn_t blkif_be_int(int irq, void *dev_id, struct pt_regs *regs)
++/**************************************************************************
++ *
++ * Port handling
++ *
++ **************************************************************************/
++
++/* This ensures that the kernel is kept informed (via
++ * netif_carrier_on/off) of the link status, and also maintains the
++ * link status's stop on the port's TX queue.
++ */
++static void efx_link_status_changed(struct efx_nic *efx)
+{
-+ blkif_notify_work(dev_id);
-+ return IRQ_HANDLED;
-+}
++ unsigned long flags __attribute__ ((unused));
++ int carrier_ok;
+
++ /* Ensure no link status notifications get sent to the OS after the net
++ * device has been unregistered. */
++ if (!efx->net_dev_registered)
++ return;
+
++ carrier_ok = netif_carrier_ok(efx->net_dev) ? 1 : 0;
++ if (efx->link_up != carrier_ok) {
++ efx->n_link_state_changes++;
+
-+/******************************************************************
-+ * DOWNWARD CALLS -- These interface with the block-device layer proper.
-+ */
++ if (efx->link_up)
++ netif_carrier_on(efx->net_dev);
++ else
++ netif_carrier_off(efx->net_dev);
++ }
+
-+static int do_block_io_op(blkif_t *blkif)
++ /* Inform driverlink client */
++ EFX_DL_CALLBACK(efx, link_change, efx->link_up);
++
++ /* Status message for kernel log */
++ if (efx->link_up) {
++ struct mii_if_info *gmii = &efx->mii;
++ unsigned adv, lpa;
++ /* NONE here means direct XAUI from the controller, with no
++ * MDIO-attached device we can query. */
++ if (efx->phy_type != PHY_TYPE_NONE) {
++ adv = gmii_advertised(gmii);
++ lpa = gmii_lpa(gmii);
++ } else {
++ lpa = GM_LPA_10000 | LPA_DUPLEX;
++ adv = lpa;
++ }
++ EFX_INFO(efx, "link up at %dMbps %s-duplex "
++ "(adv %04x lpa %04x) (MTU %d)%s%s%s%s\n",
++ (efx->link_options & GM_LPA_10000 ? 10000 :
++ (efx->link_options & GM_LPA_1000 ? 1000 :
++ (efx->link_options & GM_LPA_100 ? 100 :
++ 10))),
++ (efx->link_options & GM_LPA_DUPLEX ?
++ "full" : "half"),
++ adv, lpa,
++ efx->net_dev->mtu,
++ (efx->loopback_mode ? " [" : ""),
++ (efx->loopback_mode ? LOOPBACK_MODE(efx) : ""),
++ (efx->loopback_mode ? " LOOPBACK]" : ""),
++ (efx->promiscuous ? " [PROMISC]" : ""));
++ } else {
++ EFX_INFO(efx, "link down\n");
++ }
++
++}
++
++/* This call reinitialises the MAC to pick up new PHY settings
++ * To call from a context that cannot sleep use reconfigure_work work item
++ * For on_disabled=1 the caller must be serialised against efx_reset,
++ * ideally by holding the rtnl lock.
++ */
++void efx_reconfigure_port(struct efx_nic *efx, int on_disabled)
+{
-+ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
-+ blkif_request_t req;
-+ pending_req_t *pending_req;
-+ RING_IDX rc, rp;
-+ int more_to_do = 0;
++ mutex_lock(&efx->mac_lock);
+
-+ rc = blk_rings->common.req_cons;
-+ rp = blk_rings->common.sring->req_prod;
-+ rmb(); /* Ensure we see queued requests up to 'rp'. */
++ EFX_LOG(efx, "reconfiguring MAC from PHY settings\n");
+
-+ while ((rc != rp)) {
++ if (on_disabled)
++ ASSERT_RTNL();
++ else if (!efx->port_enabled)
++ goto out;
+
-+ if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc))
-+ break;
++ efx->mac_op->reconfigure(efx);
+
-+ pending_req = alloc_req();
-+ if (NULL == pending_req) {
-+ blkif->st_oo_req++;
-+ more_to_do = 1;
-+ break;
-+ }
++out:
++ /* Inform kernel of loss/gain of carrier */
++ efx_link_status_changed(efx);
+
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req));
-+ break;
-+ case BLKIF_PROTOCOL_X86_32:
-+ blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
-+ break;
-+ case BLKIF_PROTOCOL_X86_64:
-+ blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
-+ break;
-+ default:
-+ BUG();
-+ }
-+ blk_rings->common.req_cons = ++rc; /* before make_response() */
++ mutex_unlock(&efx->mac_lock);
++}
+
-+ switch (req.operation) {
-+ case BLKIF_OP_READ:
-+ blkif->st_rd_req++;
-+ dispatch_rw_block_io(blkif, &req, pending_req);
-+ break;
-+ case BLKIF_OP_WRITE_BARRIER:
-+ blkif->st_br_req++;
-+ /* fall through */
-+ case BLKIF_OP_WRITE:
-+ blkif->st_wr_req++;
-+ dispatch_rw_block_io(blkif, &req, pending_req);
-+ break;
-+ default:
-+ DPRINTK("error: unknown block io operation [%d]\n",
-+ req.operation);
-+ make_response(blkif, req.id, req.operation,
-+ BLKIF_RSP_ERROR);
-+ free_req(pending_req);
-+ break;
-+ }
-+ }
-+ return more_to_do;
++static void efx_reconfigure_work(struct work_struct *data)
++{
++ struct efx_nic *efx = container_of(data, struct efx_nic,
++ reconfigure_work);
++
++ EFX_LOG(efx, "MAC reconfigure executing on CPU %d\n",
++ raw_smp_processor_id());
++
++ /* Reinitialise MAC to activate new PHY parameters */
++ efx_reconfigure_port(efx, 0);
+}
+
-+static void dispatch_rw_block_io(blkif_t *blkif,
-+ blkif_request_t *req,
-+ pending_req_t *pending_req)
++static int efx_probe_port(struct efx_nic *efx)
+{
-+ extern void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]);
-+ struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-+ struct phys_req preq;
-+ struct {
-+ unsigned long buf; unsigned int nsec;
-+ } seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-+ unsigned int nseg;
-+ struct bio *bio = NULL, *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-+ int ret, i, nbio = 0;
-+ int operation;
++ unsigned char *dev_addr;
++ int rc;
+
-+ switch (req->operation) {
-+ case BLKIF_OP_READ:
-+ operation = READ;
-+ break;
-+ case BLKIF_OP_WRITE:
-+ operation = WRITE;
-+ break;
-+ case BLKIF_OP_WRITE_BARRIER:
-+ operation = WRITE_BARRIER;
-+ break;
-+ default:
-+ operation = 0; /* make gcc happy */
-+ BUG();
-+ }
++ EFX_LOG(efx, "create port\n");
+
-+ /* Check that number of segments is sane. */
-+ nseg = req->nr_segments;
-+ if (unlikely(nseg == 0) ||
-+ unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
-+ DPRINTK("Bad number of segments in request (%d)\n", nseg);
-+ goto fail_response;
++ /* Connect up MAC/PHY operations table and read MAC address */
++ rc = falcon_probe_port(efx);
++ if (rc)
++ goto err;
++
++ /* Sanity check MAC address */
++ dev_addr = efx->mac_address;
++ if (!is_valid_ether_addr(dev_addr)) {
++ DECLARE_MAC_BUF(mac);
++
++ EFX_ERR(efx, "invalid MAC address %s\n",
++ print_mac(mac, dev_addr));
++ if (!allow_bad_hwaddr) {
++ rc = -EINVAL;
++ goto err;
++ }
++ random_ether_addr(dev_addr);
++ EFX_INFO(efx, "using locally-generated MAC %s\n",
++ print_mac(mac, dev_addr));
+ }
+
-+ preq.dev = req->handle;
-+ preq.sector_number = req->sector_number;
-+ preq.nr_sects = 0;
++ /* Register debugfs entries */
++ rc = efx_init_debugfs_port(efx);
++ if (rc)
++ goto err;
+
-+ pending_req->blkif = blkif;
-+ pending_req->id = req->id;
-+ pending_req->operation = req->operation;
-+ pending_req->status = BLKIF_RSP_OKAY;
-+ pending_req->nr_pages = nseg;
++ return 0;
+
-+ for (i = 0; i < nseg; i++) {
-+ uint32_t flags;
++ err:
++ efx_remove_port(efx);
++ return rc;
++}
+
-+ seg[i].nsec = req->seg[i].last_sect -
-+ req->seg[i].first_sect + 1;
++static int efx_init_port(struct efx_nic *efx)
++{
++ int rc;
+
-+ if ((req->seg[i].last_sect >= (PAGE_SIZE >> 9)) ||
-+ (req->seg[i].last_sect < req->seg[i].first_sect))
-+ goto fail_response;
-+ preq.nr_sects += seg[i].nsec;
++ EFX_LOG(efx, "init port\n");
+
-+ flags = GNTMAP_host_map;
-+ if (operation != READ)
-+ flags |= GNTMAP_readonly;
-+ gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
-+ req->seg[i].gref, blkif->domid);
-+ }
++ /* The default power state is ON */
++ efx->phy_powered = 1;
+
-+ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg);
-+ BUG_ON(ret);
++ /* Initialise the MAC and PHY */
++ rc = efx->mac_op->init(efx);
++ if (rc)
++ return rc;
+
-+ for (i = 0; i < nseg; i++) {
-+ if (unlikely(map[i].status != 0)) {
-+ DPRINTK("invalid buffer -- could not remap it\n");
-+ map[i].handle = BLKBACK_INVALID_HANDLE;
-+ ret |= 1;
-+ }
++ efx->port_initialized = 1;
+
-+ pending_handle(pending_req, i) = map[i].handle;
++ /* Reconfigure port to program MAC registers */
++ efx->mac_op->reconfigure(efx);
+
-+ if (ret)
-+ continue;
++ return 0;
++}
+
-+ set_phys_to_machine(__pa(vaddr(
-+ pending_req, i)) >> PAGE_SHIFT,
-+ FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
-+ seg[i].buf = map[i].dev_bus_addr |
-+ (req->seg[i].first_sect << 9);
-+ }
++/* Allow efx_reconfigure_port() to run, and propagate delayed changes
++ * to the promiscuous flag to the MAC if needed */
++static void efx_start_port(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "start port\n");
++ ASSERT_RTNL();
+
-+ if (ret)
-+ goto fail_flush;
++ BUG_ON(efx->port_enabled);
+
-+ if (vbd_translate(&preq, blkif, operation) != 0) {
-+ DPRINTK("access denied: %s of [%llu,%llu] on dev=%04x\n",
-+ operation == READ ? "read" : "write",
-+ preq.sector_number,
-+ preq.sector_number + preq.nr_sects, preq.dev);
-+ goto fail_flush;
-+ }
++ mutex_lock(&efx->mac_lock);
++ efx->port_enabled = 1;
++ mutex_unlock(&efx->mac_lock);
+
-+ for (i = 0; i < nseg; i++) {
-+ if (((int)preq.sector_number|(int)seg[i].nsec) &
-+ ((bdev_hardsect_size(preq.bdev) >> 9) - 1)) {
-+ DPRINTK("Misaligned I/O request from domain %d",
-+ blkif->domid);
-+ goto fail_put_bio;
++ if (efx->net_dev_registered) {
++ int promiscuous;
++
++ netif_tx_lock_bh(efx->net_dev);
++ promiscuous = (efx->net_dev->flags & IFF_PROMISC) ? 1 : 0;
++ if (efx->promiscuous != promiscuous) {
++ efx->promiscuous = promiscuous;
++ queue_work(efx->workqueue, &efx->reconfigure_work);
+ }
++ netif_tx_unlock_bh(efx->net_dev);
++ }
++}
+
-+ while ((bio == NULL) ||
-+ (bio_add_page(bio,
-+ virt_to_page(vaddr(pending_req, i)),
-+ seg[i].nsec << 9,
-+ seg[i].buf & ~PAGE_MASK) == 0)) {
-+ bio = biolist[nbio++] = bio_alloc(GFP_KERNEL, nseg-i);
-+ if (unlikely(bio == NULL))
-+ goto fail_put_bio;
++/* Prevents efx_reconfigure_port() from executing, and prevents
++ * efx_set_multicast_list() from scheduling efx_reconfigure_work.
++ * efx_reconfigure_work can still be scheduled via NAPI processing
++ * until efx_flush_all() is called */
++static void efx_stop_port(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "stop port\n");
++ ASSERT_RTNL();
+
-+ bio->bi_bdev = preq.bdev;
-+ bio->bi_private = pending_req;
-+ bio->bi_end_io = end_block_io_op;
-+ bio->bi_sector = preq.sector_number;
-+ }
++ mutex_lock(&efx->mac_lock);
++ efx->port_enabled = 0;
++ mutex_unlock(&efx->mac_lock);
+
-+ preq.sector_number += seg[i].nsec;
++ /* Serialise against efx_set_multicast_list() */
++ if (efx->net_dev_registered) {
++ netif_tx_lock_bh(efx->net_dev);
++ netif_tx_unlock_bh(efx->net_dev);
+ }
++}
+
-+ plug_queue(blkif, bio);
-+ atomic_set(&pending_req->pendcnt, nbio);
-+ blkif_get(blkif);
++static void efx_fini_port(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "shut down port\n");
+
-+ for (i = 0; i < nbio; i++)
-+ submit_bio(operation, biolist[i]);
++ if (!efx->port_initialized)
++ return;
+
-+ if (operation == READ)
-+ blkif->st_rd_sect += preq.nr_sects;
-+ else if (operation == WRITE)
-+ blkif->st_wr_sect += preq.nr_sects;
++ efx->mac_op->fini(efx);
++ efx->port_initialized = 0;
+
-+ return;
++ /* Mark the link down */
++ efx->link_up = 0;
++ efx_link_status_changed(efx);
++}
+
-+ fail_put_bio:
-+ for (i = 0; i < (nbio-1); i++)
-+ bio_put(biolist[i]);
-+ fail_flush:
-+ fast_flush_area(pending_req);
-+ fail_response:
-+ make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
-+ free_req(pending_req);
-+}
++static void efx_remove_port(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "destroying port\n");
+
++ efx_fini_debugfs_port(efx);
++ falcon_remove_port(efx);
++}
+
++/**************************************************************************
++ *
++ * NIC handling
++ *
++ **************************************************************************/
+
-+/******************************************************************
-+ * MISCELLANEOUS SETUP / TEARDOWN / DEBUGGING
-+ */
++/* This configures the PCI device to enable I/O and DMA. */
++static int efx_init_io(struct efx_nic *efx)
++{
++ struct pci_dev *pci_dev = efx->pci_dev;
++ int rc;
+
++ EFX_LOG(efx, "initialising I/O\n");
+
-+static void make_response(blkif_t *blkif, u64 id,
-+ unsigned short op, int st)
-+{
-+ blkif_response_t resp;
-+ unsigned long flags;
-+ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
-+ int more_to_do = 0;
-+ int notify;
++ /* Generic device-enabling code */
++ rc = pci_enable_device(pci_dev);
++ if (rc) {
++ EFX_ERR(efx, "failed to enable PCI device\n");
++ goto fail1;
++ }
+
-+ resp.id = id;
-+ resp.operation = op;
-+ resp.status = st;
++ pci_set_master(pci_dev);
+
-+ spin_lock_irqsave(&blkif->blk_ring_lock, flags);
-+ /* Place on the response ring for the relevant domain. */
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ case BLKIF_PROTOCOL_X86_32:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ case BLKIF_PROTOCOL_X86_64:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ default:
-+ BUG();
++ /* Set the PCI DMA mask. Try all possibilities from our
++ * genuine mask down to 32 bits, because some architectures
++ * (e.g. x86_64 with iommu_sac_force set) will allow 40 bit
++ * masks event though they reject 46 bit masks.
++ */
++ efx->dma_mask = efx->type->max_dma_mask;
++ while (efx->dma_mask > 0x7fffffffUL) {
++ if (pci_dma_supported(pci_dev, efx->dma_mask) &&
++ ((rc = pci_set_dma_mask(pci_dev, efx->dma_mask)) == 0))
++ break;
++ efx->dma_mask >>= 1;
+ }
-+ blk_rings->common.rsp_prod_pvt++;
-+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
-+ if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) {
-+ /*
-+ * Tail check for pending requests. Allows frontend to avoid
-+ * notifications if requests are already in flight (lower
-+ * overheads and promotes batching).
++ if (rc) {
++ EFX_ERR(efx, "could not find a suitable DMA mask\n");
++ goto fail2;
++ }
++ EFX_LOG(efx, "using DMA mask %llx\n",
++ (unsigned long long)efx->dma_mask);
++ rc = pci_set_consistent_dma_mask(pci_dev, efx->dma_mask);
++ if (rc) {
++ /* pci_set_consistent_dma_mask() is not *allowed* to
++ * fail with a mask that pci_set_dma_mask() accepted,
++ * but just in case...
+ */
-+ RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
++ EFX_ERR(efx, "failed to set consistent DMA mask\n");
++ goto fail2;
++ }
+
-+ } else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) {
-+ more_to_do = 1;
++ /* Get memory base address */
++ efx->membase_phys = pci_resource_start(efx->pci_dev,
++ efx->type->mem_bar);
++#if !defined(EFX_HAVE_MSIX_TABLE_RESERVED)
++ rc = pci_request_region(pci_dev, efx->type->mem_bar, "sfc");
++#else
++ if (!request_mem_region(efx->membase_phys, efx->type->mem_map_size,
++ "sfc"))
++ rc = -EIO;
++#endif
++ if (rc) {
++ EFX_ERR(efx, "request for memory BAR failed\n");
++ rc = -EIO;
++ goto fail3;
++ }
++ efx->membase = ioremap_nocache(efx->membase_phys,
++ efx->type->mem_map_size);
++ if (!efx->membase) {
++ EFX_ERR(efx, "could not map memory BAR %d at %lx+%x\n",
++ efx->type->mem_bar, efx->membase_phys,
++ efx->type->mem_map_size);
++ rc = -ENOMEM;
++ goto fail4;
+ }
++ EFX_LOG(efx, "memory BAR %u at %lx+%x (virtual %p)\n",
++ efx->type->mem_bar, efx->membase_phys, efx->type->mem_map_size,
++ efx->membase);
+
-+ spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
++ return 0;
+
-+ if (more_to_do)
-+ blkif_notify_work(blkif);
-+ if (notify)
-+ notify_remote_via_irq(blkif->irq);
++ fail4:
++ release_mem_region(efx->membase_phys, efx->type->mem_map_size);
++ fail3:
++ efx->membase_phys = 0UL;
++ /* fall-thru */
++ fail2:
++ pci_disable_device(efx->pci_dev);
++ fail1:
++ return rc;
+}
+
-+static int __init blkif_init(void)
++static void efx_fini_io(struct efx_nic *efx)
+{
-+ int i, mmap_pages;
-+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ EFX_LOG(efx, "shutting down I/O\n");
+
-+ mmap_pages = blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
++ if (efx->membase) {
++ iounmap(efx->membase);
++ efx->membase = NULL;
++ }
+
-+ pending_reqs = kmalloc(sizeof(pending_reqs[0]) *
-+ blkif_reqs, GFP_KERNEL);
-+ pending_grant_handles = kmalloc(sizeof(pending_grant_handles[0]) *
-+ mmap_pages, GFP_KERNEL);
-+ pending_pages = alloc_empty_pages_and_pagevec(mmap_pages);
++ if (efx->membase_phys) {
++#if !defined(EFX_HAVE_MSIX_TABLE_RESERVED)
++ pci_release_region(efx->pci_dev, efx->type->mem_bar);
++#else
++ release_mem_region(efx->membase_phys, efx->type->mem_map_size);
++#endif
++ efx->membase_phys = 0UL;
++ }
+
-+ if (!pending_reqs || !pending_grant_handles || !pending_pages)
-+ goto out_of_memory;
++ pci_disable_device(efx->pci_dev);
++}
+
-+ for (i = 0; i < mmap_pages; i++)
-+ pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
++/* Probe the number and type of interrupts we are able to obtain. */
++static int efx_probe_interrupts(struct efx_nic *efx)
++{
++ struct msix_entry xentries[EFX_MAX_CHANNELS];
++ int rc, i;
+
-+ blkif_interface_init();
++ /* Select number of used RSS queues */
++ /* TODO: Can we react to CPU hotplug? */
++ if (rss_cpus == 0)
++ rss_cpus = num_online_cpus();
+
-+ memset(pending_reqs, 0, sizeof(pending_reqs));
-+ INIT_LIST_HEAD(&pending_free);
++ efx->rss_queues = 1;
++ if (efx->interrupt_mode == EFX_INT_MODE_MSIX) {
++ unsigned int max_channel = efx->type->phys_addr_channels - 1;
+
-+ for (i = 0; i < blkif_reqs; i++)
-+ list_add_tail(&pending_reqs[i].free_list, &pending_free);
++ BUG_ON(!pci_find_capability(efx->pci_dev, PCI_CAP_ID_MSIX));
++ efx->rss_queues = min(max_channel + 1, rss_cpus);
++ efx->rss_queues = min(efx->rss_queues, EFX_MAX_CHANNELS);
++ }
+
-+ blkif_xenbus_init();
++ /* Determine how many RSS queues we can use, and mark channels
++ * with the appropriate interrupt state */
++ if (efx->interrupt_mode == EFX_INT_MODE_MSIX) {
++ /* Build MSI request structure */
++ for (i = 0; i < efx->rss_queues; i++)
++ xentries[i].entry = i;
+
-+ return 0;
++ /* Request maximum number of MSI interrupts */
++ rc = pci_enable_msix(efx->pci_dev, xentries, efx->rss_queues);
++ if (rc > 0) {
++ EFX_BUG_ON_PARANOID(rc >= efx->rss_queues);
++ efx->rss_queues = rc;
++ rc = pci_enable_msix(efx->pci_dev, xentries,
++ efx->rss_queues);
++ }
++ if (rc == 0) {
++ for (i = 0; i < efx->rss_queues; i++) {
++ efx->channel[i].has_interrupt = 1;
++ efx->channel[i].irq = xentries[i].vector;
++ }
++ } else {
++ /* Fall back to single channel MSI */
++ efx->interrupt_mode = EFX_INT_MODE_MSI;
++ EFX_ERR(efx, "could not enable MSI-X\n");
++ }
++ }
+
-+ out_of_memory:
-+ kfree(pending_reqs);
-+ kfree(pending_grant_handles);
-+ free_empty_pages_and_pagevec(pending_pages, mmap_pages);
-+ printk("%s: out of memory\n", __FUNCTION__);
-+ return -ENOMEM;
++ /* Try single interrupt MSI */
++ if (efx->interrupt_mode == EFX_INT_MODE_MSI) {
++ efx->rss_queues = 1;
++ rc = pci_enable_msi(efx->pci_dev);
++ if (rc == 0) {
++ efx->channel[0].irq = efx->pci_dev->irq;
++ efx->channel[0].has_interrupt = 1;
++ } else {
++ EFX_ERR(efx, "could not enable MSI\n");
++ efx->interrupt_mode = EFX_INT_MODE_LEGACY;
++ }
++ }
++
++ /* Assume legacy interrupts */
++ if (efx->interrupt_mode == EFX_INT_MODE_LEGACY) {
++ /* Every channel is interruptible */
++ for (i = 0; i < EFX_MAX_CHANNELS; i++)
++ efx->channel[i].has_interrupt = 1;
++ efx->legacy_irq = efx->pci_dev->irq;
++ }
++
++ return 0;
+}
+
-+module_init(blkif_init);
++static void efx_remove_interrupts(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/common.h linux-2.6.18-xen.hg/drivers/xen/blkback/common.h
---- linux-2.6.18/drivers/xen/blkback/common.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/common.h 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,139 @@
-+/*
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ /* Remove MSI/MSI-X interrupts */
++ efx_for_each_channel_with_interrupt(channel, efx)
++ channel->irq = 0;
++ pci_disable_msi(efx->pci_dev);
++ pci_disable_msix(efx->pci_dev);
++
++ /* Remove legacy interrupt */
++ efx->legacy_irq = 0;
++}
++
++/* Select number of used resources
++ * Should be called after probe_interrupts()
+ */
++static int efx_select_used(struct efx_nic *efx)
++{
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++ int i;
+
-+#ifndef __BLKIF__BACKEND__COMMON_H__
-+#define __BLKIF__BACKEND__COMMON_H__
++ /* TX queues. One per port per channel with TX capability
++ * (more than one per port won't work on Linux, due to out
++ * of order issues... but will be fine on Solaris)
++ */
++ tx_queue = &efx->tx_queue[0];
+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/interrupt.h>
-+#include <linux/slab.h>
-+#include <linux/blkdev.h>
-+#include <linux/vmalloc.h>
-+#include <linux/wait.h>
-+#include <asm/io.h>
-+#include <asm/setup.h>
-+#include <asm/pgalloc.h>
-+#include <xen/evtchn.h>
-+#include <asm/hypervisor.h>
-+#include <xen/blkif.h>
-+#include <xen/gnttab.h>
-+#include <xen/driver_util.h>
-+#include <xen/xenbus.h>
++ /* Perform this for each channel with TX capabilities.
++ * At the moment, we only support a single TX queue
++ */
++ tx_queue->used = 1;
++ if ((!EFX_INT_MODE_USE_MSI(efx)) && separate_tx_and_rx_channels)
++ tx_queue->channel = &efx->channel[1];
++ else
++ tx_queue->channel = &efx->channel[0];
++ tx_queue->channel->used_flags |= EFX_USED_BY_TX;
++ tx_queue++;
++
++ /* RX queues. Each has a dedicated channel. */
++ for (i = 0; i < EFX_MAX_RX_QUEUES; i++) {
++ rx_queue = &efx->rx_queue[i];
++
++ if (i < efx->rss_queues) {
++ rx_queue->used = 1;
++ /* If we allow multiple RX queues per channel
++ * we need to decide that here
++ */
++ rx_queue->channel = &efx->channel[rx_queue->queue];
++ rx_queue->channel->used_flags |= EFX_USED_BY_RX;
++ rx_queue++;
++ }
++ }
++ return 0;
++}
+
-+#define DPRINTK(_f, _a...) \
-+ pr_debug("(file=%s, line=%d) " _f, \
-+ __FILE__ , __LINE__ , ## _a )
++static int efx_probe_nic(struct efx_nic *efx)
++{
++ int rc;
+
-+struct vbd {
-+ blkif_vdev_t handle; /* what the domain refers to this vbd as */
-+ unsigned char readonly; /* Non-zero -> read-only */
-+ unsigned char type; /* VDISK_xxx */
-+ u32 pdevice; /* phys device that this vbd maps to */
-+ struct block_device *bdev;
-+};
++ EFX_LOG(efx, "creating NIC\n");
+
-+struct backend_info;
++ /* Carry out hardware-type specific initialisation */
++ rc = falcon_probe_nic(efx);
++ if (rc)
++ goto fail1;
+
-+typedef struct blkif_st {
-+ /* Unique identifier for this interface. */
-+ domid_t domid;
-+ unsigned int handle;
-+ /* Physical parameters of the comms window. */
-+ unsigned int irq;
-+ /* Comms information. */
-+ enum blkif_protocol blk_protocol;
-+ blkif_back_rings_t blk_rings;
-+ struct vm_struct *blk_ring_area;
-+ /* The VBD attached to this interface. */
-+ struct vbd vbd;
-+ /* Back pointer to the backend_info. */
-+ struct backend_info *be;
-+ /* Private fields. */
-+ spinlock_t blk_ring_lock;
-+ atomic_t refcnt;
++ /* Determine the number of channels and RX queues by trying to hook
++ * in MSI-X interrupts. */
++ rc = efx_probe_interrupts(efx);
++ if (rc)
++ goto fail2;
+
-+ wait_queue_head_t wq;
-+ struct task_struct *xenblkd;
-+ unsigned int waiting_reqs;
-+ request_queue_t *plug;
++ /* Determine number of RX queues and TX queues */
++ rc = efx_select_used(efx);
++ if (rc)
++ goto fail3;
+
-+ /* statistics */
-+ unsigned long st_print;
-+ int st_rd_req;
-+ int st_wr_req;
-+ int st_oo_req;
-+ int st_br_req;
-+ int st_rd_sect;
-+ int st_wr_sect;
++ /* Register debugfs entries */
++ rc = efx_init_debugfs_nic(efx);
++ if (rc)
++ goto fail4;
++ /* Initialise the interrupt moderation settings */
++ efx_init_irq_moderation(efx, tx_irq_mod_usec, rx_irq_mod_usec);
+
-+ wait_queue_head_t waiting_to_free;
++ return 0;
+
-+ grant_handle_t shmem_handle;
-+ grant_ref_t shmem_ref;
-+} blkif_t;
++ fail4:
++ /* fall-thru */
++ fail3:
++ efx_remove_interrupts(efx);
++ fail2:
++ falcon_remove_nic(efx);
++ fail1:
++ return rc;
++}
+
-+blkif_t *blkif_alloc(domid_t domid);
-+void blkif_disconnect(blkif_t *blkif);
-+void blkif_free(blkif_t *blkif);
-+int blkif_map(blkif_t *blkif, unsigned long shared_page, unsigned int evtchn);
++static void efx_remove_nic(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "destroying NIC\n");
+
-+#define blkif_get(_b) (atomic_inc(&(_b)->refcnt))
-+#define blkif_put(_b) \
-+ do { \
-+ if (atomic_dec_and_test(&(_b)->refcnt)) \
-+ wake_up(&(_b)->waiting_to_free);\
-+ } while (0)
++ efx_remove_interrupts(efx);
++ falcon_remove_nic(efx);
+
-+/* Create a vbd. */
-+int vbd_create(blkif_t *blkif, blkif_vdev_t vdevice, unsigned major,
-+ unsigned minor, int readonly);
-+void vbd_free(struct vbd *vbd);
++ efx_fini_debugfs_nic(efx);
++}
+
-+unsigned long long vbd_size(struct vbd *vbd);
-+unsigned int vbd_info(struct vbd *vbd);
-+unsigned long vbd_secsize(struct vbd *vbd);
++/**************************************************************************
++ *
++ * NIC startup/shutdown
++ *
++ *************************************************************************/
+
-+struct phys_req {
-+ unsigned short dev;
-+ unsigned short nr_sects;
-+ struct block_device *bdev;
-+ blkif_sector_t sector_number;
-+};
++static int efx_probe_all(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++ int rc;
+
-+int vbd_translate(struct phys_req *req, blkif_t *blkif, int operation);
++ /* Create NIC */
++ rc = efx_probe_nic(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to create NIC\n");
++ goto fail1;
++ }
+
-+void blkif_interface_init(void);
++ /* Create port */
++ rc = efx_probe_port(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to create port\n");
++ goto fail2;
++ }
+
-+void blkif_xenbus_init(void);
++ /* Create channels */
++ efx_for_each_channel(channel, efx) {
++ rc = efx_probe_channel(channel);
++ if (rc) {
++ EFX_ERR(efx, "failed to create channel %d\n",
++ channel->channel);
++ goto fail3;
++ }
++ }
+
-+irqreturn_t blkif_be_int(int irq, void *dev_id, struct pt_regs *regs);
-+int blkif_schedule(void *arg);
++ return 0;
+
-+int blkback_barrier(struct xenbus_transaction xbt,
-+ struct backend_info *be, int state);
++ fail3:
++ efx_for_each_channel(channel, efx)
++ efx_remove_channel(channel);
++ fail2:
++ efx_remove_port(efx);
++ fail1:
++ return rc;
++}
+
-+#endif /* __BLKIF__BACKEND__COMMON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/interface.c linux-2.6.18-xen.hg/drivers/xen/blkback/interface.c
---- linux-2.6.18/drivers/xen/blkback/interface.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/interface.c 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,181 @@
-+/******************************************************************************
-+ * arch/xen/drivers/blkif/backend/interface.c
-+ *
-+ * Block-device interface management.
-+ *
-+ * Copyright (c) 2004, Keir Fraser
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++/* Called after previous invocation(s) of efx_stop_all, restarts the
++ * port, kernel transmit queue, NAPI processing and hardware interrupts.
++ * This function is safe to call multiple times when the NIC is in any
++ * state. */
++static void efx_start_all(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
+
-+#include "common.h"
-+#include <xen/evtchn.h>
-+#include <linux/kthread.h>
++ ASSERT_RTNL();
+
-+static kmem_cache_t *blkif_cachep;
++ /* Check that it is appropriate to restart the interface. All
++ * of these flags are safe to read under just the rtnl lock */
++ if (efx->port_enabled)
++ return;
++ if ((efx->state != STATE_RUNNING) && (efx->state != STATE_INIT))
++ return;
++ if (efx->net_dev_registered && !netif_running(efx->net_dev))
++ return;
+
-+blkif_t *blkif_alloc(domid_t domid)
++ /* Mark the port as enabled so port reconfigurations can start, then
++ * restart the transmit interface early so the watchdog timer stops */
++ efx_start_port(efx);
++ efx_wake_queue(efx);
++
++ efx_for_each_channel(channel, efx)
++ efx_start_channel(channel);
++
++ falcon_enable_interrupts(efx);
++
++ /* Start hardware monitor if we're in RUNNING */
++ if (efx->state == STATE_RUNNING)
++ queue_delayed_work(efx->workqueue, &efx->monitor_work,
++ efx_monitor_interval);
++}
++
++/* Flush all delayed work. Should only be called when no more delayed work
++ * will be scheduled. This doesn't flush pending online resets (efx_reset),
++ * since we're holding the rtnl_lock at this point. */
++static void efx_flush_all(struct efx_nic *efx)
+{
-+ blkif_t *blkif;
++#if defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ struct efx_rx_queue *rx_queue;
+
-+ blkif = kmem_cache_alloc(blkif_cachep, GFP_KERNEL);
-+ if (!blkif)
-+ return ERR_PTR(-ENOMEM);
++ /* Make sure the hardware monitor is stopped */
++ cancel_delayed_work_sync(&efx->monitor_work);
+
-+ memset(blkif, 0, sizeof(*blkif));
-+ blkif->domid = domid;
-+ spin_lock_init(&blkif->blk_ring_lock);
-+ atomic_set(&blkif->refcnt, 1);
-+ init_waitqueue_head(&blkif->wq);
-+ blkif->st_print = jiffies;
-+ init_waitqueue_head(&blkif->waiting_to_free);
++ /* Ensure that all RX slow refills are complete. */
++ efx_for_each_rx_queue(rx_queue, efx) {
++ cancel_delayed_work_sync(&rx_queue->work);
++ }
++#endif
+
-+ return blkif;
++#if defined(EFX_USE_CANCEL_WORK_SYNC)
++ /* Stop scheduled port reconfigurations */
++ cancel_work_sync(&efx->reconfigure_work);
++#endif
++
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ /* Ensure that the hardware monitor and asynchronous port
++ * reconfigurations are complete, which are the only two consumers
++ * of efx->workqueue. Since the hardware monitor runs on a long period,
++ * we put in some effort to cancel the delayed work safely rather
++ * than just flushing the queue twice (which is guaranteed to flush
++ * all the work since both efx_monitor and efx_reconfigure_work disarm
++ * if !efx->port_enabled. */
++ if (timer_pending(&efx->monitor_work.timer))
++ cancel_delayed_work(&efx->monitor_work);
++ flush_workqueue(efx->workqueue);
++ if (timer_pending(&efx->monitor_work.timer))
++ cancel_delayed_work(&efx->monitor_work);
++ flush_workqueue(efx->workqueue);
++
++ /* efx_rx_work will disarm if !channel->enabled, so we can just
++ * flush the refill workqueue twice as well. */
++ flush_workqueue(efx->refill_workqueue);
++ flush_workqueue(efx->refill_workqueue);
++#endif
+}
+
-+static int map_frontend_page(blkif_t *blkif, unsigned long shared_page)
++/* Quiesce hardware and software without bringing the link down.
++ * Safe to call multiple times, when the nic and interface is in any
++ * state. The caller is guaranteed to subsequently be in a position
++ * to modify any hardware and software state they see fit without
++ * taking locks. */
++static void efx_stop_all(struct efx_nic *efx)
+{
-+ struct gnttab_map_grant_ref op;
++ struct efx_channel *channel;
+
-+ gnttab_set_map_op(&op, (unsigned long)blkif->blk_ring_area->addr,
-+ GNTMAP_host_map, shared_page, blkif->domid);
++ ASSERT_RTNL();
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
-+ BUG();
++ /* port_enabled can be read safely under the rtnl lock */
++ if (!efx->port_enabled)
++ return;
+
-+ if (op.status) {
-+ DPRINTK(" Grant table operation failure !\n");
-+ return op.status;
++ /* Disable interrupts and wait for ISR to complete */
++ falcon_disable_interrupts(efx);
++ if (efx->legacy_irq)
++ synchronize_irq(efx->legacy_irq);
++ efx_for_each_channel_with_interrupt(channel, efx)
++ if (channel->irq)
++ synchronize_irq(channel->irq);
++
++ /* Stop all synchronous port reconfigurations. */
++ efx_stop_port(efx);
++
++ /* Stop all NAPI processing and synchronous rx refills */
++ efx_for_each_channel(channel, efx)
++ efx_stop_channel(channel);
++
++ /* Flush reconfigure_work, refill_workqueue, monitor_work */
++ efx_flush_all(efx);
++
++ /* Stop the kernel transmit interface late, so the watchdog
++ * timer isn't ticking over the flush */
++ efx_stop_queue(efx);
++ if (efx->net_dev_registered) {
++ netif_tx_lock_bh(efx->net_dev);
++ netif_tx_unlock_bh(efx->net_dev);
+ }
++}
+
-+ blkif->shmem_ref = shared_page;
-+ blkif->shmem_handle = op.handle;
++static void efx_remove_all(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++
++ efx_for_each_channel(channel, efx)
++ efx_remove_channel(channel);
++ efx_remove_port(efx);
++ efx_remove_nic(efx);
++}
++
++static int efx_run_selftests(struct efx_nic *efx)
++{
++ struct efx_self_tests tests;
++ unsigned modes = efx->startup_loopbacks & efx->loopback_modes;
++ int rc;
++
++ rc = efx_online_test(efx, &tests);
++ if (rc) {
++ EFX_ERR(efx, "failed self-tests with interrupt_mode of %s\n",
++ INT_MODE(efx));
++ goto fail;
++ }
++
++ if (onload_offline_selftest && modes) {
++ /* Run offline self test */
++ EFX_LOG(efx, "performing on-load offline self-tests\n");
++ rc = efx_offline_test(efx, &tests, modes);
++ EFX_LOG(efx, "%s on-load offline self-tests\n",
++ rc ? "FAILED" : "PASSED");
++ if (rc)
++ goto fail;
++ }
+
+ return 0;
++
++ fail:
++ EFX_ERR(efx, "self-tests failed. Given up!\n");
++ if (allow_load_on_failure)
++ rc = 0;
++
++ return rc;
+}
+
-+static void unmap_frontend_page(blkif_t *blkif)
++int efx_flush_queues(struct efx_nic *efx)
+{
-+ struct gnttab_unmap_grant_ref op;
++ int rc;
+
-+ gnttab_set_unmap_op(&op, (unsigned long)blkif->blk_ring_area->addr,
-+ GNTMAP_host_map, blkif->shmem_handle);
++ ASSERT_RTNL();
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
-+ BUG();
++ efx_stop_all(efx);
++
++ /* We can't just flush the tx queues because the event queues
++ * may contain tx completions from that queue. Just flush everything */
++ efx_fini_channels(efx);
++ rc = efx_init_channels(efx);
++ if (rc) {
++ efx_schedule_reset(efx, RESET_TYPE_DISABLE);
++ return rc;
++ }
++
++ efx_start_all(efx);
++
++ return 0;
+}
+
-+int blkif_map(blkif_t *blkif, unsigned long shared_page, unsigned int evtchn)
++/**************************************************************************
++ *
++ * Interrupt moderation
++ *
++ **************************************************************************/
++
++/* Set interrupt moderation parameters */
++void efx_init_irq_moderation(struct efx_nic *efx, int tx_usecs, int rx_usecs)
+{
-+ int err;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
+
-+ /* Already connected through? */
-+ if (blkif->irq)
-+ return 0;
++ ASSERT_RTNL();
+
-+ if ( (blkif->blk_ring_area = alloc_vm_area(PAGE_SIZE)) == NULL )
-+ return -ENOMEM;
++ efx_for_each_tx_queue(tx_queue, efx)
++ tx_queue->channel->irq_moderation = tx_usecs;
+
-+ err = map_frontend_page(blkif, shared_page);
-+ if (err) {
-+ free_vm_area(blkif->blk_ring_area);
-+ return err;
++ efx_for_each_rx_queue(rx_queue, efx)
++ rx_queue->channel->irq_moderation = rx_usecs;
++}
++
++/**************************************************************************
++ *
++ * Hardware monitor
++ *
++ **************************************************************************/
++
++/* Run periodically off the general workqueue. Serialised against
++ * efx_reconfigure_port via the mac_lock */
++static void efx_monitor(struct work_struct *data)
++{
++#if !defined(EFX_NEED_WORK_API_WRAPPERS)
++ struct efx_nic *efx = container_of(data, struct efx_nic,
++ monitor_work.work);
++#else
++ struct efx_nic *efx = container_of(data, struct efx_nic,
++ monitor_work);
++#endif
++ int rc = 0;
++
++ EFX_TRACE(efx, "hardware monitor executing on CPU %d\n",
++ raw_smp_processor_id());
++
++
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ /* Without cancel_delayed_work_sync(), we have to make sure that
++ * we don't rearm when port_enabled == 0 */
++ mutex_lock(&efx->mac_lock);
++ if (!efx->port_enabled) {
++ mutex_unlock(&efx->mac_lock);
++ return;
+ }
+
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ {
-+ blkif_sring_t *sring;
-+ sring = (blkif_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.native, sring, PAGE_SIZE);
-+ break;
++ rc = efx->mac_op->check_hw(efx);
++#else
++ /* If the mac_lock is already held then it is likely a port
++ * reconfiguration is already in place, which will likely do
++ * most of the work of check_hw() anyway. */
++ if (!mutex_trylock(&efx->mac_lock)) {
++ queue_delayed_work(efx->workqueue, &efx->monitor_work,
++ efx_monitor_interval);
++ return;
+ }
-+ case BLKIF_PROTOCOL_X86_32:
-+ {
-+ blkif_x86_32_sring_t *sring_x86_32;
-+ sring_x86_32 = (blkif_x86_32_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.x86_32, sring_x86_32, PAGE_SIZE);
-+ break;
++
++ if (efx->port_enabled)
++ rc = efx->mac_op->check_hw(efx);
++#endif
++ mutex_unlock(&efx->mac_lock);
++
++ if (rc) {
++ if (monitor_reset) {
++ EFX_ERR(efx, "hardware monitor detected a fault: "
++ "triggering reset\n");
++ efx_schedule_reset(efx, RESET_TYPE_MONITOR);
++ } else {
++ EFX_ERR(efx, "hardware monitor detected a fault, "
++ "skipping reset\n");
++ }
+ }
-+ case BLKIF_PROTOCOL_X86_64:
-+ {
-+ blkif_x86_64_sring_t *sring_x86_64;
-+ sring_x86_64 = (blkif_x86_64_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.x86_64, sring_x86_64, PAGE_SIZE);
++
++ queue_delayed_work(efx->workqueue, &efx->monitor_work,
++ efx_monitor_interval);
++}
++
++/**************************************************************************
++ *
++ * ioctls
++ *
++ *************************************************************************/
++
++/* Net device ioctl
++ * Context: process, rtnl_lock() held.
++ */
++static int efx_ioctl(struct net_device *net_dev, struct ifreq *ifr, int cmd)
++{
++ struct efx_nic *efx = net_dev->priv;
++ int rc;
++
++ ASSERT_RTNL();
++
++ switch (cmd) {
++ case SIOCGMIIPHY:
++ case SIOCGMIIREG:
++ rc = generic_mii_ioctl(&efx->mii, if_mii(ifr), cmd, NULL);
++ break;
++ case SIOCSMIIREG:
++ rc = generic_mii_ioctl(&efx->mii, if_mii(ifr), cmd, NULL);
++ efx_reconfigure_port(efx, 0);
+ break;
-+ }
+ default:
-+ BUG();
++ rc = -EOPNOTSUPP;
+ }
+
-+ err = bind_interdomain_evtchn_to_irqhandler(
-+ blkif->domid, evtchn, blkif_be_int, 0, "blkif-backend", blkif);
-+ if (err < 0)
-+ {
-+ unmap_frontend_page(blkif);
-+ free_vm_area(blkif->blk_ring_area);
-+ blkif->blk_rings.common.sring = NULL;
-+ return err;
++ return rc;
++}
++
++/**************************************************************************
++ *
++ * NAPI interface
++ *
++ **************************************************************************/
++
++/* Allocate the NAPI dev's.
++ * Called after we know how many channels there are.
++ */
++static int efx_init_napi(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++ int rc;
++
++ ASSERT_RTNL();
++
++ /* Allocate the NAPI dev for the port */
++ efx->net_dev = alloc_etherdev(0);
++ if (!efx->net_dev) {
++ rc = -ENOMEM;
++ goto err;
++ }
++ efx->net_dev->priv = efx;
++ efx->mii.dev = efx->net_dev;
++
++ /* Set features based on module parameters and DMA mask.
++ * Enable DMA to ZONE_HIGHMEM if the NIC can access all memory
++ * directly. This only has an effect on 32-bit systems and
++ * PAE on x86 limits memory to 64GB so 40 bits is plenty to
++ * address everything. If the device can't address 40 bits
++ * then it's safest to turn NETIF_F_HIGHDMA off because this
++ * might be a PAE system with more than 4G of RAM and a 32-bit
++ * NIC. The use of EFX_DMA_MASK is to eliminate compiler
++ * warnings on platforms where dma_addr_t is 32-bit. We
++ * assume that in those cases we can access all memory
++ * directly if our DMA mask is all ones. */
++ efx->net_dev->features |= NETIF_F_IP_CSUM | NETIF_F_SG;
++ if (efx->dma_mask >= EFX_DMA_MASK(DMA_40BIT_MASK))
++ efx->net_dev->features |= NETIF_F_HIGHDMA;
++
++ /* Copy MAC address */
++ memcpy(&efx->net_dev->dev_addr, efx->mac_address, ETH_ALEN);
++
++ /* Allocate the per channel devs */
++ efx_for_each_channel(channel, efx) {
++#if !defined(EFX_HAVE_OLD_NAPI)
++ channel->napi_dev = efx->net_dev;
++#else
++ channel->napi_dev = alloc_etherdev(0);
++ if (!channel->napi_dev) {
++ rc = -ENOMEM;
++ goto err;
++ }
++ channel->napi_dev->priv = channel;
++ atomic_set(&channel->napi_dev->refcnt, 1);
++#endif
+ }
-+ blkif->irq = err;
+
+ return 0;
++ err:
++ efx_fini_napi(efx);
++ return rc;
+}
+
-+void blkif_disconnect(blkif_t *blkif)
++/* Free the NAPI state for the port and channels */
++static void efx_fini_napi(struct efx_nic *efx)
+{
-+ if (blkif->xenblkd) {
-+ kthread_stop(blkif->xenblkd);
-+ blkif->xenblkd = NULL;
-+ }
++ struct efx_channel *channel;
+
-+ atomic_dec(&blkif->refcnt);
-+ wait_event(blkif->waiting_to_free, atomic_read(&blkif->refcnt) == 0);
-+ atomic_inc(&blkif->refcnt);
++ ASSERT_RTNL();
+
-+ if (blkif->irq) {
-+ unbind_from_irqhandler(blkif->irq, blkif);
-+ blkif->irq = 0;
++ efx_for_each_channel(channel, efx) {
++ /* Finish per channel NAPI */
++#if defined(EFX_HAVE_OLD_NAPI)
++ if (channel->napi_dev) {
++ channel->napi_dev->priv = NULL;
++ free_netdev(channel->napi_dev);
++ }
++#endif
++ channel->napi_dev = NULL;
+ }
+
-+ if (blkif->blk_rings.common.sring) {
-+ unmap_frontend_page(blkif);
-+ free_vm_area(blkif->blk_ring_area);
-+ blkif->blk_rings.common.sring = NULL;
++ /* Finish port NAPI */
++ if (efx->net_dev) {
++ efx->net_dev->priv = NULL;
++ free_netdev(efx->net_dev);
++ efx->net_dev = NULL;
+ }
+}
+
-+void blkif_free(blkif_t *blkif)
++/**************************************************************************
++ *
++ * Kernel netpoll interface
++ *
++ *************************************************************************/
++
++#ifdef CONFIG_NET_POLL_CONTROLLER
++
++/* Although in the common case interrupts will be disabled, this is not
++ * guaranteed. However, all our work happens inside the NAPI callback,
++ * so no locking is required.
++ */
++static void efx_netpoll(struct net_device *net_dev)
+{
-+ if (!atomic_dec_and_test(&blkif->refcnt))
-+ BUG();
-+ kmem_cache_free(blkif_cachep, blkif);
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_channel *channel;
++
++ efx_for_each_channel_with_interrupt(channel, efx)
++ efx_schedule_channel(channel);
+}
+
-+void __init blkif_interface_init(void)
++#endif
++
++/**************************************************************************
++ *
++ * Kernel net device interface
++ *
++ *************************************************************************/
++
++/* Context: process, rtnl_lock() held. */
++static int efx_net_open(struct net_device *net_dev)
+{
-+ blkif_cachep = kmem_cache_create("blkif_cache", sizeof(blkif_t),
-+ 0, 0, NULL, NULL);
++ struct efx_nic *efx = net_dev->priv;
++ ASSERT_RTNL();
++
++ EFX_LOG(efx, "opening device %s on CPU %d\n", net_dev->name,
++ raw_smp_processor_id());
++ efx_start_all(efx);
++ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/Makefile linux-2.6.18-xen.hg/drivers/xen/blkback/Makefile
---- linux-2.6.18/drivers/xen/blkback/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/Makefile 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,3 @@
-+obj-$(CONFIG_XEN_BLKDEV_BACKEND) := blkbk.o
+
-+blkbk-y := blkback.o xenbus.o interface.o vbd.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/vbd.c linux-2.6.18-xen.hg/drivers/xen/blkback/vbd.c
---- linux-2.6.18/drivers/xen/blkback/vbd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/vbd.c 2007-12-23 11:15:33.541241984 +0100
-@@ -0,0 +1,118 @@
-+/******************************************************************************
-+ * blkback/vbd.c
-+ *
-+ * Routines for managing virtual block devices (VBDs).
-+ *
-+ * Copyright (c) 2003-2005, Keir Fraser & Steve Hand
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/* Context: process, rtnl_lock() held.
++ * Note that the kernel will ignore our return code; this method
++ * should really be a void.
+ */
++static int efx_net_stop(struct net_device *net_dev)
++{
++ struct efx_nic *efx = net_dev->priv;
++ int rc;
+
-+#include "common.h"
++ EFX_LOG(efx, "closing %s on CPU %d\n", net_dev->name,
++ raw_smp_processor_id());
+
-+#define vbd_sz(_v) ((_v)->bdev->bd_part ? \
-+ (_v)->bdev->bd_part->nr_sects : (_v)->bdev->bd_disk->capacity)
++ /* Stop device and flush all the channels */
++ efx_stop_all(efx);
++ efx_fini_channels(efx);
++ rc = efx_init_channels(efx);
++ if (rc)
++ efx_schedule_reset(efx, RESET_TYPE_DISABLE);
+
-+unsigned long long vbd_size(struct vbd *vbd)
-+{
-+ return vbd_sz(vbd);
++ return 0;
+}
+
-+unsigned int vbd_info(struct vbd *vbd)
++/* Context: process, dev_base_lock held, non-blocking.
++ * Statistics are taken directly from the MAC.
++ */
++static struct net_device_stats *efx_net_stats(struct net_device *net_dev)
+{
-+ return vbd->type | (vbd->readonly?VDISK_READONLY:0);
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_mac_stats *mac_stats = &efx->mac_stats;
++ struct net_device_stats *stats = &efx->stats;
++
++ if (!spin_trylock(&efx->stats_lock))
++ return stats;
++ if (efx->state == STATE_RUNNING)
++ efx->mac_op->update_stats(efx);
++ spin_unlock(&efx->stats_lock);
++
++ stats->rx_packets = mac_stats->rx_packets;
++ stats->tx_packets = mac_stats->tx_packets;
++ stats->rx_bytes = mac_stats->rx_bytes;
++ stats->tx_bytes = mac_stats->tx_bytes;
++ stats->tx_errors = mac_stats->tx_bad;
++ stats->multicast = mac_stats->rx_multicast;
++ stats->collisions = mac_stats->tx_collision;
++ stats->rx_length_errors = mac_stats->rx_gtjumbo;
++ stats->rx_over_errors = mac_stats->rx_overflow;
++ stats->rx_crc_errors = mac_stats->rx_bad;
++ stats->rx_frame_errors = mac_stats->rx_align_error;
++ stats->rx_fifo_errors = 0;
++ stats->rx_missed_errors = mac_stats->rx_missed;
++ stats->rx_errors = (stats->rx_length_errors +
++ stats->rx_over_errors +
++ stats->rx_crc_errors +
++ stats->rx_frame_errors +
++ stats->rx_fifo_errors +
++ stats->rx_missed_errors +
++ mac_stats->rx_symbol_error);
++ stats->tx_aborted_errors = 0;
++ stats->tx_carrier_errors = 0;
++ stats->tx_fifo_errors = 0;
++ stats->tx_heartbeat_errors = 0;
++ stats->tx_window_errors = 0;
++
++ return stats;
+}
+
-+unsigned long vbd_secsize(struct vbd *vbd)
++/* Context: netif_tx_lock held, BHs disabled. */
++static void efx_watchdog(struct net_device *net_dev)
+{
-+ return bdev_hardsect_size(vbd->bdev);
++ struct efx_nic *efx = net_dev->priv;
++
++ EFX_ERR(efx, "TX stuck with stop_count=%d port_enabled=%d: %s\n",
++ atomic_read(&efx->netif_stop_count), efx->port_enabled,
++ monitor_reset ? "resetting channels" : "skipping reset");
++
++ if (monitor_reset)
++ efx_schedule_reset(efx, RESET_TYPE_MONITOR);
+}
+
-+int vbd_create(blkif_t *blkif, blkif_vdev_t handle, unsigned major,
-+ unsigned minor, int readonly)
++
++/* Context: process, rtnl_lock() held. */
++static int efx_change_mtu(struct net_device *net_dev, int new_mtu)
+{
-+ struct vbd *vbd;
-+ struct block_device *bdev;
++ struct efx_nic *efx = net_dev->priv;
++ int rc = 0;
+
-+ vbd = &blkif->vbd;
-+ vbd->handle = handle;
-+ vbd->readonly = readonly;
-+ vbd->type = 0;
++ ASSERT_RTNL();
+
-+ vbd->pdevice = MKDEV(major, minor);
++ if (new_mtu > EFX_MAX_MTU)
++ return -EINVAL;
+
-+ bdev = open_by_devnum(vbd->pdevice,
-+ vbd->readonly ? FMODE_READ : FMODE_WRITE);
++ efx_stop_all(efx);
+
-+ if (IS_ERR(bdev)) {
-+ DPRINTK("vbd_creat: device %08x could not be opened.\n",
-+ vbd->pdevice);
-+ return -ENOENT;
++ /* Ask driverlink client if we can change MTU */
++ rc = EFX_DL_CALLBACK(efx, request_mtu, new_mtu);
++ if (rc) {
++ EFX_ERR(efx, "MTU change vetoed by driverlink %s driver\n",
++ efx->dl_cb_dev.request_mtu->driver->name);
++ goto out;
+ }
+
-+ vbd->bdev = bdev;
++ EFX_LOG(efx, "changing MTU to %d\n", new_mtu);
+
-+ if (vbd->bdev->bd_disk == NULL) {
-+ DPRINTK("vbd_creat: device %08x doesn't exist.\n",
-+ vbd->pdevice);
-+ vbd_free(vbd);
-+ return -ENOENT;
-+ }
++ efx_fini_channels(efx);
++ net_dev->mtu = new_mtu;
++ rc = efx_init_channels(efx);
++ if (rc)
++ goto fail;
+
-+ if (vbd->bdev->bd_disk->flags & GENHD_FL_CD)
-+ vbd->type |= VDISK_CDROM;
-+ if (vbd->bdev->bd_disk->flags & GENHD_FL_REMOVABLE)
-+ vbd->type |= VDISK_REMOVABLE;
++ /* Reconfigure the MAC */
++ efx_reconfigure_port(efx, 1);
+
-+ DPRINTK("Successful creation of handle=%04x (dom=%u)\n",
-+ handle, blkif->domid);
-+ return 0;
++ /* Notify driverlink client of new MTU */
++ EFX_DL_CALLBACK(efx, mtu_changed, new_mtu);
++
++ efx_start_all(efx);
++
++ out:
++ return rc;
++
++ fail:
++ efx_schedule_reset(efx, RESET_TYPE_DISABLE);
++ return rc;
+}
+
-+void vbd_free(struct vbd *vbd)
++static int efx_set_mac_address(struct net_device *net_dev, void *data)
+{
-+ if (vbd->bdev)
-+ blkdev_put(vbd->bdev);
-+ vbd->bdev = NULL;
++ struct efx_nic *efx = net_dev->priv;
++ struct sockaddr *addr = data;
++ char *new_addr = addr->sa_data;
++
++ ASSERT_RTNL();
++
++ if (!is_valid_ether_addr(new_addr)) {
++ DECLARE_MAC_BUF(mac);
++ EFX_ERR(efx, "invalid ethernet MAC address requested: %s\n",
++ print_mac(mac, new_addr));
++ return -EINVAL;
++ }
++
++ memcpy(net_dev->dev_addr, new_addr, net_dev->addr_len);
++
++ /* Reconfigure the MAC */
++ efx_reconfigure_port(efx, 1);
++
++ return 0;
+}
+
-+int vbd_translate(struct phys_req *req, blkif_t *blkif, int operation)
++/* Context: netif_tx_lock held, BHs disabled. */
++static void efx_set_multicast_list(struct net_device *net_dev)
+{
-+ struct vbd *vbd = &blkif->vbd;
-+ int rc = -EACCES;
-+
-+ if ((operation != READ) && vbd->readonly)
-+ goto out;
++ struct efx_nic *efx = net_dev->priv;
++ struct dev_mc_list *mc_list = net_dev->mc_list;
++ union efx_multicast_hash *mc_hash = &efx->multicast_hash;
++ unsigned long flags __attribute__ ((unused));
++ int promiscuous;
++ u32 crc;
++ int bit;
++ int i;
+
-+ if (unlikely((req->sector_number + req->nr_sects) > vbd_sz(vbd)))
-+ goto out;
++ /* Set per-MAC promiscuity flag and reconfigure MAC if necessary */
++ promiscuous = (net_dev->flags & IFF_PROMISC) ? 1 : 0;
++ if (efx->promiscuous != promiscuous) {
++ if (efx->port_enabled) {
++ efx->promiscuous = promiscuous;
++ queue_work(efx->workqueue, &efx->reconfigure_work);
++ }
++ }
+
-+ req->dev = vbd->pdevice;
-+ req->bdev = vbd->bdev;
-+ rc = 0;
++ /* Build multicast hash table */
++ if (promiscuous || (net_dev->flags & IFF_ALLMULTI)) {
++ memset(mc_hash, 0xff, sizeof(*mc_hash));
++ } else {
++ memset(mc_hash, 0x00, sizeof(*mc_hash));
++ for (i = 0; i < net_dev->mc_count; i++) {
++ crc = ether_crc_le(ETH_ALEN, mc_list->dmi_addr);
++ bit = (crc & ((1 << EFX_MCAST_HASH_BITS) - 1));
++ set_bit_le(bit, (void *)mc_hash);
++ mc_list = mc_list->next;
++ }
++ }
+
-+ out:
-+ return rc;
++ /* Create and activate new global multicast hash table */
++ falcon_set_multicast_hash(efx);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkback/xenbus.c linux-2.6.18-xen.hg/drivers/xen/blkback/xenbus.c
---- linux-2.6.18/drivers/xen/blkback/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkback/xenbus.c 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,533 @@
-+/* Xenbus code for blkif backend
-+ Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
-+ Copyright (C) 2005 XenSource Ltd
+
-+ This program is free software; you can redistribute it and/or modify
-+ it under the terms of the GNU General Public License as published by
-+ the Free Software Foundation; either version 2 of the License, or
-+ (at your option) any later version.
++/* Handle net device notifier events */
++static int efx_netdev_event(struct notifier_block *this,
++ unsigned long event, void *ptr)
++{
++ struct net_device *net_dev = (struct net_device *)ptr;
+
-+ This program is distributed in the hope that it will be useful,
-+ but WITHOUT ANY WARRANTY; without even the implied warranty of
-+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-+ GNU General Public License for more details.
++ if (net_dev->open == efx_net_open && event == NETDEV_CHANGENAME) {
++ struct efx_nic *efx = net_dev->priv;
+
-+ You should have received a copy of the GNU General Public License
-+ along with this program; if not, write to the Free Software
-+ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+*/
++ strcpy(efx->name, net_dev->name);
++ efx_fini_debugfs_netdev(net_dev);
++ efx_init_debugfs_netdev(net_dev);
++ }
+
-+#include <stdarg.h>
-+#include <linux/module.h>
-+#include <linux/kthread.h>
-+#include "common.h"
++ return NOTIFY_DONE;
++}
+
-+#undef DPRINTK
-+#define DPRINTK(fmt, args...) \
-+ pr_debug("blkback/xenbus (%s:%d) " fmt ".\n", \
-+ __FUNCTION__, __LINE__, ##args)
++static struct notifier_block efx_netdev_notifier = {
++ .notifier_call = efx_netdev_event,
++};
+
-+struct backend_info
++static int efx_register_netdev(struct efx_nic *efx)
+{
-+ struct xenbus_device *dev;
-+ blkif_t *blkif;
-+ struct xenbus_watch backend_watch;
-+ unsigned major;
-+ unsigned minor;
-+ char *mode;
-+};
++ struct net_device *net_dev = efx->net_dev;
++ int rc;
+
-+static void connect(struct backend_info *);
-+static int connect_ring(struct backend_info *);
-+static void backend_changed(struct xenbus_watch *, const char **,
-+ unsigned int);
++ net_dev->watchdog_timeo = 5 * HZ;
++ net_dev->irq = efx->pci_dev->irq;
++ net_dev->open = efx_net_open;
++ net_dev->stop = efx_net_stop;
++ net_dev->get_stats = efx_net_stats;
++ net_dev->tx_timeout = &efx_watchdog;
++ net_dev->hard_start_xmit = efx_hard_start_xmit;
++ net_dev->do_ioctl = efx_ioctl;
++ net_dev->change_mtu = efx_change_mtu;
++ net_dev->set_mac_address = efx_set_mac_address;
++ net_dev->set_multicast_list = efx_set_multicast_list;
++#ifdef CONFIG_NET_POLL_CONTROLLER
++ net_dev->poll_controller = efx_netpoll;
++#endif
++ SET_NETDEV_DEV(net_dev, &efx->pci_dev->dev);
++ SET_ETHTOOL_OPS(net_dev, &efx_ethtool_ops);
++
++ /* Always start with carrier off; PHY events will detect the link */
++ netif_carrier_off(efx->net_dev);
++
++ BUG_ON(efx->net_dev_registered);
++
++ /* Clear MAC statistics */
++ efx->mac_op->update_stats(efx);
++ memset(&efx->mac_stats, 0, sizeof(efx->mac_stats));
++
++ rc = register_netdev(net_dev);
++ if (rc) {
++ EFX_ERR(efx, "could not register net dev\n");
++ return rc;
++ }
++ strcpy(efx->name, net_dev->name);
+
-+static int blkback_name(blkif_t *blkif, char *buf)
-+{
-+ char *devpath, *devname;
-+ struct xenbus_device *dev = blkif->be->dev;
++ /* Create debugfs symlinks */
++ rc = efx_init_debugfs_netdev(net_dev);
++ if (rc) {
++ EFX_ERR(efx, "failed to init net dev debugfs\n");
++ unregister_netdev(efx->net_dev);
++ return rc;
++ }
+
-+ devpath = xenbus_read(XBT_NIL, dev->nodename, "dev", NULL);
-+ if (IS_ERR(devpath))
-+ return PTR_ERR(devpath);
-+
-+ if ((devname = strstr(devpath, "/dev/")) != NULL)
-+ devname += strlen("/dev/");
-+ else
-+ devname = devpath;
++ /* Allow link change notifications to be sent to the operating
++ * system. The must happen after register_netdev so that
++ * there are no outstanding link changes if that call fails.
++ * It must happen before efx_reconfigure_port so that the
++ * initial state of the link is reported. */
++ mutex_lock(&efx->mac_lock);
++ efx->net_dev_registered = 1;
++ mutex_unlock(&efx->mac_lock);
++
++ /* Safety net: in case we don't get a PHY event */
++ rtnl_lock();
++ efx_reconfigure_port(efx, 1);
++ rtnl_unlock();
++
++ EFX_LOG(efx, "registered\n");
+
-+ snprintf(buf, TASK_COMM_LEN, "blkback.%d.%s", blkif->domid, devname);
-+ kfree(devpath);
-+
+ return 0;
+}
+
-+static void update_blkif_status(blkif_t *blkif)
-+{
-+ int err;
-+ char name[TASK_COMM_LEN];
++static void efx_unregister_netdev(struct efx_nic *efx)
++{
++ int was_registered = efx->net_dev_registered;
++ struct efx_tx_queue *tx_queue;
+
-+ /* Not ready to connect? */
-+ if (!blkif->irq || !blkif->vbd.bdev)
++ if (!efx->net_dev)
+ return;
+
-+ /* Already connected? */
-+ if (blkif->be->dev->state == XenbusStateConnected)
-+ return;
++ BUG_ON(efx->net_dev->priv != efx);
+
-+ /* Attempt to connect: exit if we fail to. */
-+ connect(blkif->be);
-+ if (blkif->be->dev->state != XenbusStateConnected)
-+ return;
++ /* SFC Bug 5356: Ensure that no more link status notifications get
++ * sent to the stack. Bad things happen if there's an
++ * outstanding notification after the net device is freed, but
++ * they only get flushed out by unregister_netdev, not by
++ * free_netdev. */
++ mutex_lock(&efx->mac_lock);
++ efx->net_dev_registered = 0;
++ mutex_unlock(&efx->mac_lock);
+
-+ err = blkback_name(blkif, name);
-+ if (err) {
-+ xenbus_dev_error(blkif->be->dev, err, "get blkback dev name");
-+ return;
-+ }
++ /* Free up any skbs still remaining. This has to happen before
++ * we try to unregister the netdev as running their destructors
++ * may be needed to get the device ref. count to 0. */
++ efx_for_each_tx_queue(tx_queue, efx)
++ efx_release_tx_buffers(tx_queue);
+
-+ blkif->xenblkd = kthread_run(blkif_schedule, blkif, name);
-+ if (IS_ERR(blkif->xenblkd)) {
-+ err = PTR_ERR(blkif->xenblkd);
-+ blkif->xenblkd = NULL;
-+ xenbus_dev_error(blkif->be->dev, err, "start xenblkd");
++ if (was_registered) {
++ strlcpy(efx->name, pci_name(efx->pci_dev), sizeof(efx->name));
++ efx_fini_debugfs_netdev(efx->net_dev);
++ unregister_netdev(efx->net_dev);
+ }
+}
+
++/**************************************************************************
++ *
++ * Device reset and suspend
++ *
++ **************************************************************************/
+
-+/****************************************************************
-+ * sysfs interface for VBD I/O requests
++/* This suspends the device (and acquires the suspend lock) without
++ * flushing the descriptor queues. It is included for the convenience
++ * of the driverlink layer.
+ */
++void efx_suspend(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "suspending operations\n");
+
-+#define VBD_SHOW(name, format, args...) \
-+ static ssize_t show_##name(struct device *_dev, \
-+ struct device_attribute *attr, \
-+ char *buf) \
-+ { \
-+ struct xenbus_device *dev = to_xenbus_device(_dev); \
-+ struct backend_info *be = dev->dev.driver_data; \
-+ \
-+ return sprintf(buf, format, ##args); \
-+ } \
-+ static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL)
++ down(&efx->suspend_lock);
+
-+VBD_SHOW(oo_req, "%d\n", be->blkif->st_oo_req);
-+VBD_SHOW(rd_req, "%d\n", be->blkif->st_rd_req);
-+VBD_SHOW(wr_req, "%d\n", be->blkif->st_wr_req);
-+VBD_SHOW(br_req, "%d\n", be->blkif->st_br_req);
-+VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect);
-+VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect);
++ rtnl_lock();
++ efx_stop_all(efx);
++}
+
-+static struct attribute *vbdstat_attrs[] = {
-+ &dev_attr_oo_req.attr,
-+ &dev_attr_rd_req.attr,
-+ &dev_attr_wr_req.attr,
-+ &dev_attr_br_req.attr,
-+ &dev_attr_rd_sect.attr,
-+ &dev_attr_wr_sect.attr,
-+ NULL
-+};
++void efx_resume(struct efx_nic *efx)
++{
++ EFX_LOG(efx, "resuming operations\n");
+
-+static struct attribute_group vbdstat_group = {
-+ .name = "statistics",
-+ .attrs = vbdstat_attrs,
-+};
++ efx_start_all(efx);
++ rtnl_unlock();
+
-+VBD_SHOW(physical_device, "%x:%x\n", be->major, be->minor);
-+VBD_SHOW(mode, "%s\n", be->mode);
++ up(&efx->suspend_lock);
++}
+
-+int xenvbd_sysfs_addif(struct xenbus_device *dev)
++/* The final hardware and software finalisation before reset.
++ * This function does not handle serialisation with the kernel, it
++ * assumes the caller has done this */
++static int efx_reset_down(struct efx_nic *efx, struct ethtool_cmd *ecmd)
+{
-+ int error;
-+
-+ error = device_create_file(&dev->dev, &dev_attr_physical_device);
-+ if (error)
-+ goto fail1;
++ int rc;
+
-+ error = device_create_file(&dev->dev, &dev_attr_mode);
-+ if (error)
-+ goto fail2;
++ ASSERT_RTNL();
+
-+ error = sysfs_create_group(&dev->dev.kobj, &vbdstat_group);
-+ if (error)
-+ goto fail3;
++ rc = efx->mac_op->get_settings(efx, ecmd);
++ if (rc) {
++ EFX_ERR(efx, "could not back up PHY settings\n");
++ goto fail;
++ }
+
++ efx_fini_channels(efx);
+ return 0;
+
-+fail3: sysfs_remove_group(&dev->dev.kobj, &vbdstat_group);
-+fail2: device_remove_file(&dev->dev, &dev_attr_mode);
-+fail1: device_remove_file(&dev->dev, &dev_attr_physical_device);
-+ return error;
-+}
-+
-+void xenvbd_sysfs_delif(struct xenbus_device *dev)
-+{
-+ sysfs_remove_group(&dev->dev.kobj, &vbdstat_group);
-+ device_remove_file(&dev->dev, &dev_attr_mode);
-+ device_remove_file(&dev->dev, &dev_attr_physical_device);
++ fail:
++ return rc;
+}
+
-+static int blkback_remove(struct xenbus_device *dev)
++/* The first part of software initialisation after a hardware reset
++ * This function does not handle serialisation with the kernel, it
++ * assumes the caller has done this */
++static int efx_reset_up(struct efx_nic *efx, struct ethtool_cmd *ecmd)
+{
-+ struct backend_info *be = dev->dev.driver_data;
-+
-+ DPRINTK("");
++ int rc;
+
-+ if (be->major || be->minor)
-+ xenvbd_sysfs_delif(dev);
++ rc = efx_init_channels(efx);
++ if (rc)
++ goto fail1;
+
-+ if (be->backend_watch.node) {
-+ unregister_xenbus_watch(&be->backend_watch);
-+ kfree(be->backend_watch.node);
-+ be->backend_watch.node = NULL;
-+ }
++ /* In an INVISIBLE_RESET there might not be a link state transition,
++ * so we push the multicast list here. */
++ falcon_set_multicast_hash(efx);
+
-+ if (be->blkif) {
-+ blkif_disconnect(be->blkif);
-+ vbd_free(&be->blkif->vbd);
-+ blkif_free(be->blkif);
-+ be->blkif = NULL;
++ /* Restore MAC and PHY settings. */
++ rc = efx->mac_op->set_settings(efx, ecmd);
++ if (rc) {
++ EFX_ERR(efx, "could not restore PHY settings\n");
++ goto fail2;
+ }
+
-+ kfree(be);
-+ dev->dev.driver_data = NULL;
+ return 0;
-+}
-+
-+int blkback_barrier(struct xenbus_transaction xbt,
-+ struct backend_info *be, int state)
-+{
-+ struct xenbus_device *dev = be->dev;
-+ int err;
-+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-barrier",
-+ "%d", state);
-+ if (err)
-+ xenbus_dev_fatal(dev, err, "writing feature-barrier");
+
-+ return err;
++ fail2:
++ efx_fini_channels(efx);
++ fail1:
++ return rc;
+}
+
-+/**
-+ * Entry point to this code when a new device is created. Allocate the basic
-+ * structures, and watch the store waiting for the hotplug scripts to tell us
-+ * the device's physical major and minor numbers. Switch to InitWait.
++/* Reset the NIC as transparently as possible. Do not reset the PHY
++ * Note that the reset may fail, in which case the card will be left
++ * in a most-probably-unusable state.
++ *
++ * This function will sleep. You cannot reset from within an atomic
++ * state; use efx_schedule_reset() instead.
+ */
-+static int blkback_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
++static int efx_reset(struct efx_nic *efx)
+{
-+ int err;
-+ struct backend_info *be = kzalloc(sizeof(struct backend_info),
-+ GFP_KERNEL);
-+ if (!be) {
-+ xenbus_dev_fatal(dev, -ENOMEM,
-+ "allocating backend structure");
-+ return -ENOMEM;
++ struct ethtool_cmd ecmd;
++ unsigned long flags __attribute__ ((unused));
++ enum reset_type method = efx->reset_pending;
++ int rc;
++
++ efx_dl_reset_lock();
++
++ rc = down_interruptible(&efx->suspend_lock);
++ if (rc) {
++ EFX_ERR(efx, "reset aborted by signal\n");
++ goto unlock_dl_lock;
+ }
-+ be->dev = dev;
-+ dev->dev.driver_data = be;
+
-+ be->blkif = blkif_alloc(dev->otherend_id);
-+ if (IS_ERR(be->blkif)) {
-+ err = PTR_ERR(be->blkif);
-+ be->blkif = NULL;
-+ xenbus_dev_fatal(dev, err, "creating block interface");
-+ goto fail;
++ /* We've got suspend_lock, which means we can only be in
++ * STATE_RUNNING or STATE_FINI. Don't clear
++ * efx->reset_pending, since this flag indicates that we
++ * should retry device initialisation.
++ */
++ if (efx->state != STATE_RUNNING) {
++ EFX_INFO(efx, "scheduled reset quenched. NIC not RUNNING\n");
++ goto unlock_suspend_lock;
+ }
+
-+ /* setup back pointer */
-+ be->blkif->be = be;
++ /* Notify driverlink clients of imminent reset. */
++ efx_dl_reset_suspend(efx);
++ rtnl_lock();
+
-+ err = xenbus_watch_path2(dev, dev->nodename, "physical-device",
-+ &be->backend_watch, backend_changed);
-+ if (err)
-+ goto fail;
++ efx->state = STATE_RESETTING;
++ EFX_INFO(efx, "resetting (%s)\n", RESET_TYPE(method));
+
-+ err = xenbus_switch_state(dev, XenbusStateInitWait);
-+ if (err)
-+ goto fail;
++ /* The net_dev->get_stats handler is quite slow, and will fail
++ * if a fetch is pending over reset. Serialise against it. */
++ spin_lock(&efx->stats_lock);
++ spin_unlock(&efx->stats_lock);
+
-+ return 0;
++ efx_stop_all(efx);
++ mutex_lock(&efx->mac_lock);
+
-+fail:
-+ DPRINTK("failed");
-+ blkback_remove(dev);
-+ return err;
-+}
++ rc = efx_reset_down(efx, &ecmd);
++ if (rc)
++ goto fail1;
++ falcon_fini_nic(efx);
+
++ rc = falcon_reset_hw(efx, method);
++ if (rc) {
++ EFX_ERR(efx, "failed to reset hardware\n");
++ goto fail2;
++ }
+
-+/**
-+ * Callback received when the hotplug scripts have placed the physical-device
-+ * node. Read it and the mode node, and create a vbd. If the frontend is
-+ * ready, connect.
-+ */
-+static void backend_changed(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
-+{
-+ int err;
-+ unsigned major;
-+ unsigned minor;
-+ struct backend_info *be
-+ = container_of(watch, struct backend_info, backend_watch);
-+ struct xenbus_device *dev = be->dev;
++ /* Allow resets to be rescheduled. */
++ efx->reset_pending = RESET_TYPE_NONE;
+
-+ DPRINTK("");
++ /* Reinitialise bus-mastering, which may have been turned off before
++ * the reset was scheduled. This is still appropriate, even in the
++ * RESET_TYPE_DISABLE since this driver generally assumes the hardware
++ * can respond to requests. */
++ pci_set_master(efx->pci_dev);
+
-+ err = xenbus_scanf(XBT_NIL, dev->nodename, "physical-device", "%x:%x",
-+ &major, &minor);
-+ if (XENBUS_EXIST_ERR(err)) {
-+ /* Since this watch will fire once immediately after it is
-+ registered, we expect this. Ignore it, and wait for the
-+ hotplug scripts. */
-+ return;
-+ }
-+ if (err != 2) {
-+ xenbus_dev_fatal(dev, err, "reading physical-device");
-+ return;
++ /* Reinitialise device. This is appropriate in the RESET_TYPE_DISABLE
++ * case so the driver can talk to external SRAM */
++ rc = falcon_init_nic(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to initialise NIC\n");
++ goto fail3;
+ }
+
-+ if ((be->major || be->minor) &&
-+ ((be->major != major) || (be->minor != minor))) {
-+ printk(KERN_WARNING
-+ "blkback: changing physical device (from %x:%x to "
-+ "%x:%x) not supported.\n", be->major, be->minor,
-+ major, minor);
-+ return;
++ /* Leave device stopped if necessary */
++ if (method == RESET_TYPE_DISABLE) {
++ /* Reinitialise the device anyway so the driver unload sequence
++ * can talk to the external SRAM */
++ (void) falcon_init_nic(efx);
++ rc = -EIO;
++ goto fail4;
+ }
+
-+ be->mode = xenbus_read(XBT_NIL, dev->nodename, "mode", NULL);
-+ if (IS_ERR(be->mode)) {
-+ err = PTR_ERR(be->mode);
-+ be->mode = NULL;
-+ xenbus_dev_fatal(dev, err, "reading mode");
-+ return;
-+ }
++ rc = efx_reset_up(efx, &ecmd);
++ if (rc)
++ goto fail5;
+
-+ if (be->major == 0 && be->minor == 0) {
-+ /* Front end dir is a number, which is used as the handle. */
++ mutex_unlock(&efx->mac_lock);
++ efx_reconfigure_port(efx, 1);
++ EFX_LOG(efx, "reset complete\n");
+
-+ char *p = strrchr(dev->otherend, '/') + 1;
-+ long handle = simple_strtoul(p, NULL, 0);
++ efx->state = STATE_RUNNING;
++ efx_start_all(efx);
+
-+ be->major = major;
-+ be->minor = minor;
++ rtnl_unlock();
+
-+ err = vbd_create(be->blkif, handle, major, minor,
-+ (NULL == strchr(be->mode, 'w')));
-+ if (err) {
-+ be->major = be->minor = 0;
-+ xenbus_dev_fatal(dev, err, "creating vbd structure");
-+ return;
-+ }
++ goto notify;
+
-+ err = xenvbd_sysfs_addif(dev);
-+ if (err) {
-+ vbd_free(&be->blkif->vbd);
-+ be->major = be->minor = 0;
-+ xenbus_dev_fatal(dev, err, "creating sysfs entries");
-+ return;
-+ }
++ fail5:
++ fail4:
++ fail3:
++ fail2:
++ fail1:
++ EFX_ERR(efx, "has been disabled\n");
++ efx->state = STATE_DISABLED;
+
-+ /* We're potentially connected now */
-+ update_blkif_status(be->blkif);
-+ }
-+}
++ /* Remove the net_dev */
++ mutex_unlock(&efx->mac_lock);
++ rtnl_unlock();
++ efx_unregister_netdev(efx);
++ efx_fini_port(efx);
+
++ notify:
++ /* Notify driverlink clients of completed reset */
++ efx_dl_reset_resume(efx, (rc == 0));
+
-+/**
-+ * Callback received when the frontend's state changes.
++ unlock_suspend_lock:
++ up(&efx->suspend_lock);
++
++ unlock_dl_lock:
++ efx_dl_reset_unlock();
++
++ return rc;
++}
++
++/* The worker thread exists so that code that cannot sleep can
++ * schedule a reset for later.
+ */
-+static void frontend_changed(struct xenbus_device *dev,
-+ enum xenbus_state frontend_state)
++static void efx_reset_work(struct work_struct *data)
+{
-+ struct backend_info *be = dev->dev.driver_data;
-+ int err;
-+
-+ DPRINTK("%s", xenbus_strstate(frontend_state));
++ struct efx_nic *nic = container_of(data, struct efx_nic, reset_work);
+
-+ switch (frontend_state) {
-+ case XenbusStateInitialising:
-+ if (dev->state == XenbusStateClosed) {
-+ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
-+ __FUNCTION__, dev->nodename);
-+ xenbus_switch_state(dev, XenbusStateInitWait);
-+ }
-+ break;
++ efx_reset(nic);
++}
+
-+ case XenbusStateInitialised:
-+ case XenbusStateConnected:
-+ /* Ensure we connect even when two watches fire in
-+ close successsion and we miss the intermediate value
-+ of frontend_state. */
-+ if (dev->state == XenbusStateConnected)
-+ break;
++void efx_schedule_reset(struct efx_nic *efx, enum reset_type type)
++{
++ enum reset_type method;
+
-+ err = connect_ring(be);
-+ if (err)
-+ break;
-+ update_blkif_status(be->blkif);
-+ break;
++ if (efx->reset_pending != RESET_TYPE_NONE) {
++ EFX_INFO(efx, "quenching already scheduled reset\n");
++ return;
++ }
+
-+ case XenbusStateClosing:
-+ blkif_disconnect(be->blkif);
-+ xenbus_switch_state(dev, XenbusStateClosing);
++ switch (type) {
++ case RESET_TYPE_INVISIBLE:
++ case RESET_TYPE_ALL:
++ case RESET_TYPE_WORLD:
++ case RESET_TYPE_DISABLE:
++ method = type;
+ break;
-+
-+ case XenbusStateClosed:
-+ xenbus_switch_state(dev, XenbusStateClosed);
-+ if (xenbus_dev_is_online(dev))
-+ break;
-+ /* fall through if not online */
-+ case XenbusStateUnknown:
-+ device_unregister(&dev->dev);
++ case RESET_TYPE_RX_RECOVERY:
++ case RESET_TYPE_RX_DESC_FETCH:
++ case RESET_TYPE_TX_DESC_FETCH:
++ method = RESET_TYPE_INVISIBLE;
+ break;
-+
+ default:
-+ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
-+ frontend_state);
++ method = RESET_TYPE_ALL;
+ break;
+ }
++
++ if (method != type)
++ EFX_LOG(efx, "scheduling %s reset for %s\n",
++ RESET_TYPE(method), RESET_TYPE(type));
++ else
++ EFX_LOG(efx, "scheduling %s reset\n", RESET_TYPE(method));
++
++ efx->reset_pending = method;
++
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ queue_work(efx->reset_workqueue, &efx->reset_work);
++#else
++ queue_work(efx->workqueue, &efx->reset_work);
++#endif
+}
+
++/**************************************************************************
++ *
++ * List of NICs we support
++ *
++ **************************************************************************/
++
++enum efx_type_index {
++ EFX_TYPE_FALCON_A = 0,
++ EFX_TYPE_FALCON_B = 1,
++};
+
-+/* ** Connection ** */
++static struct efx_nic_type *efx_nic_types[] = {
++ [EFX_TYPE_FALCON_A] = &falcon_a_nic_type,
++ [EFX_TYPE_FALCON_B] = &falcon_b_nic_type,
++};
+
+
-+/**
-+ * Write the physical details regarding the block device to the store, and
-+ * switch to Connected state.
-+ */
-+static void connect(struct backend_info *be)
++/* PCI device ID table */
++static struct pci_device_id efx_pci_table[] __devinitdata = {
++ {EFX_VENDID_SFC, FALCON_A_P_DEVID, PCI_ANY_ID, PCI_ANY_ID,
++ 0, 0, EFX_TYPE_FALCON_A},
++ {EFX_VENDID_SFC, FALCON_B_P_DEVID, PCI_ANY_ID, PCI_ANY_ID,
++ 0, 0, EFX_TYPE_FALCON_B},
++ {0} /* end of list */
++};
++
++/**************************************************************************
++ *
++ * Dummy PHY/MAC/Board operations
++ *
++ * Can be used where the MAC does not implement this operation
++ * Needed so all function pointers are valid and do not have to be tested
++ * before use
++ *
++ **************************************************************************/
++int efx_port_dummy_op_int(struct efx_nic *efx)
+{
-+ struct xenbus_transaction xbt;
-+ int err;
-+ struct xenbus_device *dev = be->dev;
++ return 0;
++}
++void efx_port_dummy_op_void(struct efx_nic *efx) {}
++void efx_port_dummy_op_blink(struct efx_nic *efx, int blink) {}
+
-+ DPRINTK("%s", dev->otherend);
++static struct efx_mac_operations efx_dummy_mac_operations = {
++ .init = efx_port_dummy_op_int,
++ .reconfigure = efx_port_dummy_op_void,
++ .fini = efx_port_dummy_op_void,
++};
+
-+ /* Supply the information about the device the frontend needs */
-+again:
-+ err = xenbus_transaction_start(&xbt);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "starting transaction");
-+ return;
-+ }
++static struct efx_phy_operations efx_dummy_phy_operations = {
++ .init = efx_port_dummy_op_int,
++ .reconfigure = efx_port_dummy_op_void,
++ .check_hw = efx_port_dummy_op_int,
++ .fini = efx_port_dummy_op_void,
++ .clear_interrupt = efx_port_dummy_op_void,
++ .reset_xaui = efx_port_dummy_op_void,
++};
+
-+ err = blkback_barrier(xbt, be, 1);
-+ if (err)
-+ goto abort;
++/* Dummy board operations */
++static int efx_nic_dummy_op_int(struct efx_nic *nic)
++{
++ return 0;
++}
+
-+ err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
-+ vbd_size(&be->blkif->vbd));
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "writing %s/sectors",
-+ dev->nodename);
-+ goto abort;
++static void efx_nic_dummy_op_void(struct efx_nic *nic) {}
++
++static struct efx_board efx_dummy_board_info = {
++ .init = efx_nic_dummy_op_int,
++ .init_leds = efx_port_dummy_op_int,
++ .set_fault_led = efx_port_dummy_op_blink,
++ .monitor = efx_nic_dummy_op_int,
++ .blink = efx_port_dummy_op_blink,
++ .fini = efx_nic_dummy_op_void,
++};
++
++/**************************************************************************
++ *
++ * Data housekeeping
++ *
++ **************************************************************************/
++
++/* This zeroes out and then fills in the invariants in a struct
++ * efx_nic (including all sub-structures).
++ */
++static int efx_init_struct(struct efx_nic *efx, enum efx_type_index type,
++ struct pci_dev *pci_dev)
++{
++ struct efx_channel *channel;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++ int i, rc;
++
++ /* Initialise common structures */
++ memset(efx, 0, sizeof(*efx));
++ spin_lock_init(&efx->biu_lock);
++ spin_lock_init(&efx->phy_lock);
++ mutex_init(&efx->spi_lock);
++ sema_init(&efx->suspend_lock, 1);
++ INIT_WORK(&efx->reset_work, efx_reset_work);
++ INIT_DELAYED_WORK(&efx->monitor_work, efx_monitor);
++ efx->pci_dev = pci_dev;
++ efx->state = STATE_INIT;
++ efx->reset_pending = RESET_TYPE_NONE;
++ strlcpy(efx->name, pci_name(pci_dev), sizeof(efx->name));
++ efx->board_info = efx_dummy_board_info;
++
++ efx->rx_checksum_enabled = 1;
++ spin_lock_init(&efx->netif_stop_lock);
++ spin_lock_init(&efx->stats_lock);
++ mutex_init(&efx->mac_lock);
++ efx->mac_op = &efx_dummy_mac_operations;
++ efx->phy_op = &efx_dummy_phy_operations;
++ INIT_LIST_HEAD(&efx->dl_node);
++ INIT_LIST_HEAD(&efx->dl_device_list);
++ efx->dl_cb = efx_default_callbacks;
++ INIT_WORK(&efx->reconfigure_work, efx_reconfigure_work);
++ atomic_set(&efx->netif_stop_count, 1);
++
++ for (i = 0; i < EFX_MAX_CHANNELS; i++) {
++ channel = &efx->channel[i];
++ channel->efx = efx;
++ channel->channel = i;
++ channel->evqnum = i;
++ channel->work_pending = 0;
++ }
++ for (i = 0; i < EFX_MAX_TX_QUEUES; i++) {
++ tx_queue = &efx->tx_queue[i];
++ tx_queue->efx = efx;
++ tx_queue->queue = i;
++ tx_queue->buffer = NULL;
++ tx_queue->channel = &efx->channel[0]; /* for safety */
++ }
++ for (i = 0; i < EFX_MAX_RX_QUEUES; i++) {
++ rx_queue = &efx->rx_queue[i];
++ rx_queue->efx = efx;
++ rx_queue->queue = i;
++ rx_queue->channel = &efx->channel[0]; /* for safety */
++ rx_queue->buffer = NULL;
++ spin_lock_init(&rx_queue->add_lock);
++ INIT_DELAYED_WORK(&rx_queue->work, efx_rx_work);
++ }
++
++ efx->type = efx_nic_types[type];
++
++ /* Sanity-check NIC type */
++ EFX_BUG_ON_PARANOID(efx->type->txd_ring_mask &
++ (efx->type->txd_ring_mask + 1));
++ EFX_BUG_ON_PARANOID(efx->type->rxd_ring_mask &
++ (efx->type->rxd_ring_mask + 1));
++ EFX_BUG_ON_PARANOID(efx->type->evq_size &
++ (efx->type->evq_size - 1));
++ /* As close as we can get to guaranteeing that we don't overflow */
++ EFX_BUG_ON_PARANOID(efx->type->evq_size <
++ (efx->type->txd_ring_mask + 1 +
++ efx->type->rxd_ring_mask + 1));
++
++ EFX_BUG_ON_PARANOID(efx->type->phys_addr_channels > EFX_MAX_CHANNELS);
++
++ /* Higher numbered interrupt modes are less capable! */
++ efx->interrupt_mode = max(efx->type->max_interrupt_mode,
++ interrupt_mode);
++#if defined(EFX_NEED_DUMMY_MSIX)
++ if (efx->interrupt_mode == EFX_INT_MODE_MSIX)
++ efx->interrupt_mode = EFX_INT_MODE_MSI;
++#endif
++
++ /* Tasks that can fail are last */
++ efx->refill_workqueue = create_workqueue("sfc_refill");
++ if (!efx->refill_workqueue) {
++ rc = -ENOMEM;
++ goto fail1;
+ }
+
-+ /* FIXME: use a typename instead */
-+ err = xenbus_printf(xbt, dev->nodename, "info", "%u",
-+ vbd_info(&be->blkif->vbd));
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "writing %s/info",
-+ dev->nodename);
-+ goto abort;
++ efx->workqueue = create_singlethread_workqueue("sfc_work");
++ if (!efx->workqueue) {
++ rc = -ENOMEM;
++ goto fail2;
+ }
-+ err = xenbus_printf(xbt, dev->nodename, "sector-size", "%lu",
-+ vbd_secsize(&be->blkif->vbd));
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "writing %s/sector-size",
-+ dev->nodename);
-+ goto abort;
++
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ efx->reset_workqueue = create_singlethread_workqueue("sfc_reset");
++ if (!efx->reset_workqueue) {
++ rc = -ENOMEM;
++ goto fail3;
+ }
++#endif
+
-+ err = xenbus_transaction_end(xbt, 0);
-+ if (err == -EAGAIN)
-+ goto again;
-+ if (err)
-+ xenbus_dev_fatal(dev, err, "ending transaction");
++ return 0;
+
-+ err = xenbus_switch_state(dev, XenbusStateConnected);
-+ if (err)
-+ xenbus_dev_fatal(dev, err, "switching to Connected state",
-+ dev->nodename);
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ fail3:
++ destroy_workqueue(efx->workqueue);
++ efx->workqueue = NULL;
++#endif
+
-+ return;
-+ abort:
-+ xenbus_transaction_end(xbt, 1);
++ fail2:
++ destroy_workqueue(efx->refill_workqueue);
++ efx->refill_workqueue = NULL;
++ fail1:
++ return rc;
+}
+
-+
-+static int connect_ring(struct backend_info *be)
++static void efx_fini_struct(struct efx_nic *efx)
+{
-+ struct xenbus_device *dev = be->dev;
-+ unsigned long ring_ref;
-+ unsigned int evtchn;
-+ char protocol[64] = "";
-+ int err;
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ if (efx->reset_workqueue) {
++ destroy_workqueue(efx->reset_workqueue);
++ efx->reset_workqueue = NULL;
++ }
++#endif
++ if (efx->workqueue) {
++ destroy_workqueue(efx->workqueue);
++ efx->workqueue = NULL;
++ }
++ if (efx->refill_workqueue) {
++ destroy_workqueue(efx->refill_workqueue);
++ efx->refill_workqueue = NULL;
++ }
++}
+
-+ DPRINTK("%s", dev->otherend);
-+
-+ err = xenbus_gather(XBT_NIL, dev->otherend, "ring-ref", "%lu", &ring_ref,
-+ "event-channel", "%u", &evtchn, NULL);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err,
-+ "reading %s/ring-ref and event-channel",
-+ dev->otherend);
-+ return err;
-+ }
++/**************************************************************************
++ *
++ * PCI interface
++ *
++ **************************************************************************/
+
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
-+ err = xenbus_gather(XBT_NIL, dev->otherend, "protocol",
-+ "%63s", protocol, NULL);
-+ if (err)
-+ strcpy(protocol, "unspecified, assuming native");
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_NATIVE))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_32))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_32;
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_64))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_64;
-+ else {
-+ xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol);
-+ return -1;
-+ }
-+ printk(KERN_INFO
-+ "blkback: ring-ref %ld, event-channel %d, protocol %d (%s)\n",
-+ ring_ref, evtchn, be->blkif->blk_protocol, protocol);
++/* Main body of final NIC shutdown code
++ * This is called only at module unload (or hotplug removal).
++ */
++static void efx_pci_remove_main(struct efx_nic *efx)
++{
++ ASSERT_RTNL();
+
-+ /* Map the shared frame, irq etc. */
-+ err = blkif_map(be->blkif, ring_ref, evtchn);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "mapping ring-ref %lu port %u",
-+ ring_ref, evtchn);
-+ return err;
-+ }
++ /* Skip everything if we never obtained a valid membase */
++ if (!efx->membase)
++ return;
+
-+ return 0;
-+}
++ efx_fini_channels(efx);
++ efx_fini_port(efx);
+
++ /* Shutdown the board, then the NIC and board state */
++ efx->board_info.fini(efx);
++ falcon_fini_nic(efx);
++ falcon_fini_interrupt(efx);
++ efx->board_info.fini(efx);
+
-+/* ** Driver Registration ** */
++ /* Tear down NAPI and LRO */
++ efx_fini_napi(efx);
++ efx_remove_all(efx);
++}
+
++/* Final NIC shutdown
++ * This is called only at module unload (or hotplug removal).
++ */
++static void efx_pci_remove(struct pci_dev *pci_dev)
++{
++ struct efx_nic *efx;
+
-+static struct xenbus_device_id blkback_ids[] = {
-+ { "vbd" },
-+ { "" }
-+};
++ efx = pci_get_drvdata(pci_dev);
++ if (!efx)
++ return;
+
++ /* Unregister driver from driverlink layer */
++ efx_dl_unregister_nic(efx);
+
-+static struct xenbus_driver blkback = {
-+ .name = "vbd",
-+ .owner = THIS_MODULE,
-+ .ids = blkback_ids,
-+ .probe = blkback_probe,
-+ .remove = blkback_remove,
-+ .otherend_changed = frontend_changed
-+};
++ /* Mark the NIC as fini under both suspend_lock and
++ * rtnl_lock */
++ down(&efx->suspend_lock);
++ rtnl_lock();
++ efx->state = STATE_FINI;
++ up(&efx->suspend_lock);
+
++ if (efx->membase) {
++ /* Stop the NIC. Since we're in STATE_FINI, this
++ * won't be reversed. */
++ if (efx->net_dev_registered)
++ dev_close(efx->net_dev);
+
-+void blkif_xenbus_init(void)
-+{
-+ xenbus_register_backend(&blkback);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkfront/blkfront.c linux-2.6.18-xen.hg/drivers/xen/blkfront/blkfront.c
---- linux-2.6.18/drivers/xen/blkfront/blkfront.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkfront/blkfront.c 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,920 @@
-+/******************************************************************************
-+ * blkfront.c
-+ *
-+ * XenLinux virtual block-device driver.
-+ *
-+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
-+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
-+ * Copyright (c) 2004, Christian Limpach
-+ * Copyright (c) 2004, Andrew Warfield
-+ * Copyright (c) 2005, Christopher Clark
-+ * Copyright (c) 2005, XenSource Ltd
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++ /* Release the rtnl lock. Any queued efx_resets()
++ * can now return early [we're in STATE_FINI]. */
++ rtnl_unlock();
+
-+#include <linux/version.h>
-+#include "block.h"
-+#include <linux/cdrom.h>
-+#include <linux/sched.h>
-+#include <linux/interrupt.h>
-+#include <scsi/scsi.h>
-+#include <xen/evtchn.h>
-+#include <xen/xenbus.h>
-+#include <xen/interface/grant_table.h>
-+#include <xen/interface/io/protocols.h>
-+#include <xen/gnttab.h>
-+#include <asm/hypervisor.h>
-+#include <asm/maddr.h>
++ efx_unregister_netdev(efx);
++ efx_fini_debugfs_channels(efx);
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
++ /* Wait for any scheduled resets to complete. No more will be
++ * scheduled from this point because efx_stop_all() has been
++ * called, we are no longer registered with driverlink, and
++ * the net_device's have been removed. */
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ flush_workqueue(efx->reset_workqueue);
++#else
++ flush_workqueue(efx->workqueue);
+#endif
+
-+#define BLKIF_STATE_DISCONNECTED 0
-+#define BLKIF_STATE_CONNECTED 1
-+#define BLKIF_STATE_SUSPENDED 2
-+
-+#define MAXIMUM_OUTSTANDING_BLOCK_REQS \
-+ (BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
-+#define GRANT_INVALID_REF 0
-+
-+static void connect(struct blkfront_info *);
-+static void blkfront_closing(struct xenbus_device *);
-+static int blkfront_remove(struct xenbus_device *);
-+static int talk_to_backend(struct xenbus_device *, struct blkfront_info *);
-+static int setup_blkring(struct xenbus_device *, struct blkfront_info *);
++ /* Fini and remove all the software state */
++ rtnl_lock();
++ efx_pci_remove_main(efx);
++ }
+
-+static void kick_pending_request_queues(struct blkfront_info *);
++ rtnl_unlock();
+
-+static irqreturn_t blkif_int(int irq, void *dev_id, struct pt_regs *ptregs);
-+static void blkif_restart_queue(void *arg);
-+static void blkif_recover(struct blkfront_info *);
-+static void blkif_completion(struct blk_shadow *);
-+static void blkif_free(struct blkfront_info *, int);
++ efx_fini_io(efx);
++ EFX_LOG(efx, "shutdown successful\n");
+
++ pci_set_drvdata(pci_dev, NULL);
++ efx_fini_struct(efx);
++ kfree(efx);
++};
+
-+/**
-+ * Entry point to this code when a new device is created. Allocate the basic
-+ * structures and the ring buffer for communication with the backend, and
-+ * inform the backend of the appropriate details for those. Switch to
-+ * Initialised state.
++/* Main body of NIC initialisation
++ * This is called at module load (or hotplug insertion, theoretically).
+ */
-+static int blkfront_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
++static int efx_pci_probe_main(struct efx_nic *efx)
+{
-+ int err, vdevice, i;
-+ struct blkfront_info *info;
++ int rc;
+
-+ /* FIXME: Use dynamic device id if this is not set. */
-+ err = xenbus_scanf(XBT_NIL, dev->nodename,
-+ "virtual-device", "%i", &vdevice);
-+ if (err != 1) {
-+ xenbus_dev_fatal(dev, err, "reading virtual-device");
-+ return err;
++ /* Do start-of-day initialisation */
++ rc = efx_probe_all(efx);
++ if (rc)
++ goto fail1;
++
++ /* Initialise port/channel net_dev's */
++ rc = efx_init_napi(efx);
++ if (rc)
++ goto fail2;
++
++ /* Initialise the board */
++ rc = efx->board_info.init(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to initialise board\n");
++ goto fail3;
+ }
+
-+ info = kzalloc(sizeof(*info), GFP_KERNEL);
-+ if (!info) {
-+ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
-+ return -ENOMEM;
++ /* Initialise device */
++ rc = falcon_init_nic(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to initialise NIC\n");
++ goto fail4;
+ }
+
-+ info->xbdev = dev;
-+ info->vdevice = vdevice;
-+ info->connected = BLKIF_STATE_DISCONNECTED;
-+ INIT_WORK(&info->work, blkif_restart_queue, (void *)info);
++ /* Initialise port */
++ rc = efx_init_port(efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to initialise port\n");
++ goto fail5;
++ }
+
-+ for (i = 0; i < BLK_RING_SIZE; i++)
-+ info->shadow[i].req.id = i+1;
-+ info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
++ /* Initialise channels */
++ rc = efx_init_channels(efx);
++ if (rc)
++ goto fail6;
+
-+ /* Front end dir is a number, which is used as the id. */
-+ info->handle = simple_strtoul(strrchr(dev->nodename,'/')+1, NULL, 0);
-+ dev->dev.driver_data = info;
++ rc = falcon_init_interrupt(efx);
++ if (rc)
++ goto fail7;
+
-+ err = talk_to_backend(dev, info);
-+ if (err) {
-+ kfree(info);
-+ dev->dev.driver_data = NULL;
-+ return err;
-+ }
++ /* Start up device - interrupts can occur from this point */
++ efx_start_all(efx);
++
++ /* Check basic functionality and set interrupt mode */
++ rc = efx_run_selftests(efx);
++ if (rc)
++ goto fail8;
++
++ /* Stop the NIC */
++ efx_stop_all(efx);
+
+ return 0;
-+}
+
++ fail8:
++ efx_stop_all(efx);
++ falcon_fini_interrupt(efx);
++ fail7:
++ efx_fini_channels(efx);
++ fail6:
++ efx_fini_port(efx);
++ fail5:
++ falcon_fini_nic(efx);
++ fail4:
++ efx->board_info.fini(efx);
++ fail3:
++ efx_fini_napi(efx);
++ fail2:
++ efx_remove_all(efx);
++ fail1:
++ return rc;
++}
+
-+/**
-+ * We are reconnecting to the backend, due to a suspend/resume, or a backend
-+ * driver restart. We tear down our blkif structure and recreate it, but
-+ * leave the device-layer structures intact so that this is transparent to the
-+ * rest of the kernel.
++/* NIC initialisation
++ *
++ * This is called at module load (or hotplug insertion,
++ * theoretically). It sets up PCI mappings, tests and resets the NIC,
++ * sets up and registers the network devices with the kernel and hooks
++ * the interrupt service routine. It does not prepare the device for
++ * transmission; this is left to the first time one of the network
++ * interfaces is brought up (i.e. efx_net_open).
+ */
-+static int blkfront_resume(struct xenbus_device *dev)
++static int __devinit efx_pci_probe(struct pci_dev *pci_dev,
++ const struct pci_device_id *entry)
+{
-+ struct blkfront_info *info = dev->dev.driver_data;
-+ int err;
++ struct efx_nic *efx;
++ enum efx_type_index type = entry->driver_data;
++ int i, rc;
+
-+ DPRINTK("blkfront_resume: %s\n", dev->nodename);
++ /* Allocate and initialise a struct efx_nic */
++ efx = kmalloc(sizeof(*efx), GFP_KERNEL);
++ if (!efx) {
++ rc = -ENOMEM;
++ goto fail1;
++ }
++ pci_set_drvdata(pci_dev, efx);
++ rc = efx_init_struct(efx, type, pci_dev);
++ if (rc)
++ goto fail2;
+
-+ blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
++ EFX_INFO(efx, "Solarflare Communications NIC detected\n");
+
-+ err = talk_to_backend(dev, info);
-+ if (info->connected == BLKIF_STATE_SUSPENDED && !err)
-+ blkif_recover(info);
++ /* Set up basic I/O (BAR mappings etc) */
++ rc = efx_init_io(efx);
++ if (rc)
++ goto fail3;
+
-+ return err;
-+}
++ /* From this point on we begin to expose the driver to the OS
++ * to varying degrees, so lets grab the suspend_lock and
++ * rtnl_lock to serialise against efx_reset() and
++ * friends. efx->state is not STATE_RUNNING yet, but we don't
++ * want these tasks to fail, just to block until we drop the
++ * lock
++ */
++ rc = down_interruptible(&efx->suspend_lock);
++ if (rc) {
++ EFX_ERR(efx, "suspend interrupted - aborting\n");
++ goto fail4;
++ }
+
++ rtnl_lock();
+
-+/* Common code used when first setting up, and when resuming. */
-+static int talk_to_backend(struct xenbus_device *dev,
-+ struct blkfront_info *info)
-+{
-+ const char *message = NULL;
-+ struct xenbus_transaction xbt;
-+ int err;
++ /* Probe, initialise and start everything. Run self-test */
++ for (i = 0; i < 5; i++) {
++ rc = efx_pci_probe_main(efx);
++ if (rc == 0)
++ break;
+
-+ /* Create shared ring, alloc event channel. */
-+ err = setup_blkring(dev, info);
-+ if (err)
-+ goto out;
++ /* Retry if a recoverably reset event has been scheduled */
++ if ((efx->reset_pending != RESET_TYPE_INVISIBLE) &&
++ (efx->reset_pending != RESET_TYPE_ALL))
++ goto fail5;
+
-+again:
-+ err = xenbus_transaction_start(&xbt);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "starting transaction");
-+ goto destroy_blkring;
-+ }
++ /* Serialise against efx_reset(). No more resets will be
++ * scheduled since efx_stop_all() has been called, and we
++ * have not and never have been registered with either
++ * the rtnetlink or driverlink layers. */
++ rtnl_unlock();
++ up(&efx->suspend_lock);
+
-+ err = xenbus_printf(xbt, dev->nodename,
-+ "ring-ref","%u", info->ring_ref);
-+ if (err) {
-+ message = "writing ring-ref";
-+ goto abort_transaction;
-+ }
-+ err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
-+ irq_to_evtchn_port(info->irq));
-+ if (err) {
-+ message = "writing event-channel";
-+ goto abort_transaction;
-+ }
-+ err = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
-+ XEN_IO_PROTO_ABI_NATIVE);
-+ if (err) {
-+ message = "writing protocol";
-+ goto abort_transaction;
-+ }
++#if defined(EFX_USE_CANCEL_WORK_SYNC)
++ cancel_work_sync(&efx->reset_work);
++#else
++ flush_workqueue(efx->reset_workqueue);
++#endif
+
-+ err = xenbus_transaction_end(xbt, 0);
-+ if (err) {
-+ if (err == -EAGAIN)
-+ goto again;
-+ xenbus_dev_fatal(dev, err, "completing transaction");
-+ goto destroy_blkring;
++ down(&efx->suspend_lock);
++ rtnl_lock();
++
++ efx->reset_pending = RESET_TYPE_NONE;
++ };
++ if (rc) {
++ EFX_ERR(efx, "Could not reset NIC\n");
++ goto fail5;
+ }
+
-+ xenbus_switch_state(dev, XenbusStateInitialised);
++ /* Self-tests have all passed */
++ rc = efx_init_debugfs_channels(efx);
++ if (rc)
++ goto fail6;
++
++ /* Switch to the running state before we expose the device to
++ * the OS. This is to ensure that the initial gathering of
++ * MAC stats succeeds. */
++ efx->state = STATE_RUNNING;
++
++ rtnl_unlock();
++
++ rc = efx_register_netdev(efx);
++ if (rc)
++ goto fail7;
++
++ up(&efx->suspend_lock);
++
++ EFX_LOG(efx, "initialisation successful\n");
++
++ /* Register with driverlink layer */
++ rc = efx_dl_register_nic(efx);
++ if (rc)
++ goto fail8;
+
+ return 0;
+
-+ abort_transaction:
-+ xenbus_transaction_end(xbt, 1);
-+ if (message)
-+ xenbus_dev_fatal(dev, err, "%s", message);
-+ destroy_blkring:
-+ blkif_free(info, 0);
-+ out:
-+ return err;
++ fail8:
++ down(&efx->suspend_lock);
++ efx_unregister_netdev(efx);
++ fail7:
++ /* Re-acquire the rtnl lock around pci_remove_main() */
++ rtnl_lock();
++ efx_fini_debugfs_channels(efx);
++ fail6:
++ efx_pci_remove_main(efx);
++ fail5:
++ /* Drop the locks before fini */
++ rtnl_unlock();
++ up(&efx->suspend_lock);
++ fail4:
++ efx_fini_io(efx);
++ fail3:
++ efx_fini_struct(efx);
++ fail2:
++ kfree(efx);
++ fail1:
++ EFX_LOG(efx, "initialisation failed. rc=%d\n", rc);
++ return rc;
+}
+
++/* PCI driver definition */
++static struct pci_driver efx_pci_driver = {
++ .name = EFX_DRIVER_NAME,
++ .id_table = efx_pci_table,
++ .probe = efx_pci_probe,
++ .remove = efx_pci_remove,
++};
+
-+static int setup_blkring(struct xenbus_device *dev,
-+ struct blkfront_info *info)
++/**************************************************************************
++ *
++ * Kernel module interface
++ *
++ *************************************************************************/
++
++module_param(interrupt_mode, uint, 0444);
++MODULE_PARM_DESC(interrupt_mode,
++ "Interrupt mode (0=>MSIX 1=>MSI 2=>legacy)");
++
++module_param(onload_offline_selftest, uint, 0444);
++MODULE_PARM_DESC(onload_offline_selftest, "Perform offline selftest on load");
++
++static int __init efx_init_module(void)
+{
-+ blkif_sring_t *sring;
-+ int err;
++ int rc;
+
-+ info->ring_ref = GRANT_INVALID_REF;
++ printk(KERN_INFO "Solarflare NET driver v" EFX_DRIVER_VERSION "\n");
+
-+ sring = (blkif_sring_t *)__get_free_page(GFP_KERNEL);
-+ if (!sring) {
-+ xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
-+ return -ENOMEM;
-+ }
-+ SHARED_RING_INIT(sring);
-+ FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
++ rc = efx_init_debugfs();
++ if (rc)
++ goto err_debugfs;
+
-+ err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
-+ if (err < 0) {
-+ free_page((unsigned long)sring);
-+ info->ring.sring = NULL;
-+ goto fail;
-+ }
-+ info->ring_ref = err;
++ rc = register_netdevice_notifier(&efx_netdev_notifier);
++ if (rc)
++ goto err_notifier;
+
-+ err = bind_listening_port_to_irqhandler(
-+ dev->otherend_id, blkif_int, SA_SAMPLE_RANDOM, "blkif", info);
-+ if (err <= 0) {
-+ xenbus_dev_fatal(dev, err,
-+ "bind_listening_port_to_irqhandler");
-+ goto fail;
-+ }
-+ info->irq = err;
++ rc = pci_register_driver(&efx_pci_driver);
++ if (rc < 0)
++ goto err_pci;
+
+ return 0;
-+fail:
-+ blkif_free(info, 0);
-+ return err;
++
++ err_pci:
++ unregister_netdevice_notifier(&efx_netdev_notifier);
++ err_notifier:
++ efx_fini_debugfs();
++ err_debugfs:
++ return rc;
+}
+
++static void __exit efx_exit_module(void)
++{
++ printk(KERN_INFO "Solarflare NET driver unloading\n");
+
-+/**
-+ * Callback received when the backend's state changes.
++ pci_unregister_driver(&efx_pci_driver);
++ unregister_netdevice_notifier(&efx_netdev_notifier);
++ efx_fini_debugfs();
++
++}
++
++module_init(efx_init_module);
++module_exit(efx_exit_module);
++
++MODULE_AUTHOR("Michael Brown <mbrown@fensystems.co.uk> and "
++ "Solarflare Communications");
++MODULE_DESCRIPTION("Solarflare Communications network driver");
++MODULE_LICENSE("GPL");
++MODULE_DEVICE_TABLE(pci, efx_pci_table);
+--- linux-2.6.18.8/drivers/net/sfc/efx.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/efx.h 2008-05-19 00:33:28.837808372 +0300
+@@ -0,0 +1,103 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static void backend_changed(struct xenbus_device *dev,
-+ enum xenbus_state backend_state)
-+{
-+ struct blkfront_info *info = dev->dev.driver_data;
-+ struct block_device *bd;
+
-+ DPRINTK("blkfront:backend_changed.\n");
++#ifndef EFX_EFX_H
++#define EFX_EFX_H
+
-+ switch (backend_state) {
-+ case XenbusStateInitialising:
-+ case XenbusStateInitWait:
-+ case XenbusStateInitialised:
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ break;
++#include "net_driver.h"
+
-+ case XenbusStateConnected:
-+ connect(info);
-+ break;
++/* PCI IDs */
++#define EFX_VENDID_SFC 0x1924
++#define FALCON_A_P_DEVID 0x0703
++#define FALCON_A_S_DEVID 0x6703
++#define FALCON_B_P_DEVID 0x0710
+
-+ case XenbusStateClosing:
-+ bd = bdget(info->dev);
-+ if (bd == NULL)
-+ xenbus_dev_fatal(dev, -ENODEV, "bdget failed");
++/* TX */
++extern int efx_xmit(struct efx_nic *efx,
++ struct efx_tx_queue *tx_queue, struct sk_buff *skb);
++extern void efx_stop_queue(struct efx_nic *efx);
++extern void efx_wake_queue(struct efx_nic *efx);
+
-+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
-+ down(&bd->bd_sem);
++/* RX */
++#if defined(EFX_USE_FASTCALL)
++extern void fastcall efx_xmit_done(struct efx_tx_queue *tx_queue,
++ unsigned int index);
+#else
-+ mutex_lock(&bd->bd_mutex);
++extern void efx_xmit_done(struct efx_tx_queue *tx_queue, unsigned int index);
+#endif
-+ if (info->users > 0)
-+ xenbus_dev_error(dev, -EBUSY,
-+ "Device in use; refusing to close");
-+ else
-+ blkfront_closing(dev);
-+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
-+ up(&bd->bd_sem);
++#if defined(EFX_USE_FASTCALL)
++extern void fastcall efx_rx_packet(struct efx_rx_queue *rx_queue,
++ unsigned int index, unsigned int len,
++ int checksummed, int discard);
+#else
-+ mutex_unlock(&bd->bd_mutex);
++extern void efx_rx_packet(struct efx_rx_queue *rx_queue, unsigned int index,
++ unsigned int len, int checksummed, int discard);
+#endif
-+ bdput(bd);
-+ break;
-+ }
-+}
-+
++extern void efx_fini_rx_buffer(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *rx_buf);
+
-+/* ** Connection ** */
++/* Channels */
++extern void efx_process_channel_now(struct efx_channel *channel);
++extern int efx_flush_queues(struct efx_nic *efx);
+
++/* Ports */
++extern void efx_reconfigure_port(struct efx_nic *efx,
++ int on_disabled);
+
-+/*
-+ * Invoked when the backend is finally 'ready' (and has told produced
-+ * the details about the physical device - #sectors, size, etc).
-+ */
-+static void connect(struct blkfront_info *info)
-+{
-+ unsigned long long sectors;
-+ unsigned long sector_size;
-+ unsigned int binfo;
-+ int err;
++/* Global */
++extern void efx_schedule_reset(struct efx_nic *efx, enum reset_type type);
++extern void efx_suspend(struct efx_nic *efx);
++extern void efx_resume(struct efx_nic *efx);
++extern void efx_init_irq_moderation(struct efx_nic *efx, int tx_usecs,
++ int rx_usecs);
++extern int efx_request_power(struct efx_nic *efx, int mw, const char *name);
++extern void efx_hex_dump(const u8 *, unsigned int, const char *);
+
-+ if ((info->connected == BLKIF_STATE_CONNECTED) ||
-+ (info->connected == BLKIF_STATE_SUSPENDED) )
-+ return;
++/* Dummy PHY ops for PHY drivers */
++extern int efx_port_dummy_op_int(struct efx_nic *efx);
++extern void efx_port_dummy_op_void(struct efx_nic *efx);
++extern void efx_port_dummy_op_blink(struct efx_nic *efx, int blink);
+
-+ DPRINTK("blkfront.c:connect:%s.\n", info->xbdev->otherend);
+
-+ err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
-+ "sectors", "%Lu", §ors,
-+ "info", "%u", &binfo,
-+ "sector-size", "%lu", §or_size,
-+ NULL);
-+ if (err) {
-+ xenbus_dev_fatal(info->xbdev, err,
-+ "reading backend fields at %s",
-+ info->xbdev->otherend);
-+ return;
-+ }
++extern unsigned int efx_monitor_interval;
+
-+ err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
-+ "feature-barrier", "%lu", &info->feature_barrier,
-+ NULL);
-+ if (err)
-+ info->feature_barrier = 0;
++static inline void efx_schedule_channel(struct efx_channel *channel)
++{
++ EFX_TRACE(channel->efx, "channel %d scheduling NAPI poll on CPU%d\n",
++ channel->channel, raw_smp_processor_id());
++ channel->work_pending = 1;
+
-+ err = xlvbd_add(sectors, info->vdevice, binfo, sector_size, info);
-+ if (err) {
-+ xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
-+ info->xbdev->otherend);
-+ return;
-+ }
++#if defined(EFX_HAVE_OLD_NAPI)
++ if (!test_and_set_bit(__LINK_STATE_RX_SCHED, &channel->napi_dev->state))
++ __netif_rx_schedule(channel->napi_dev);
++#else
++ netif_rx_schedule(channel->napi_dev, &channel->napi_str);
++#endif
++}
+
-+ (void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
+
-+ /* Kick pending requests. */
-+ spin_lock_irq(&blkif_io_lock);
-+ info->connected = BLKIF_STATE_CONNECTED;
-+ kick_pending_request_queues(info);
-+ spin_unlock_irq(&blkif_io_lock);
++#endif /* EFX_EFX_H */
+--- linux-2.6.18.8/drivers/net/sfc/enum.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/enum.h 2008-05-19 00:33:28.837808372 +0300
+@@ -0,0 +1,117 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ add_disk(info->gd);
++#ifndef EFX_ENUM_H
++#define EFX_ENUM_H
+
-+ info->is_ready = 1;
-+}
++/**
++ * enum efx_loopback_mode - loopback modes
++ * @LOOPBACK_NONE: no loopback
++ * @LOOPBACK_NEAR: loopback nearest to bus
++ * @LOOPBACK_MAC: loopback within MAC unspecified level
++ * @LOOPBACK_XGMII: loopback within MAC at XGMII level
++ * @LOOPBACK_XGXS: loopback within MAC at XGXS level
++ * @LOOPBACK_XAUI: loopback within MAC at XAUI level
++ * @LOOPBACK_PHY: loopback within PHY unspecified level
++ * @LOOPBACK_PHYXS: loopback within PHY at PHYXS level
++ * @LOOPBACK_PCS: loopback within PHY at PCS level
++ * @LOOPBACK_PMAPMD: loopback within PHY at PMAPMD level
++ * @LOOPBACK_FAR: loopback furthest from bus
++ * @LOOPBACK_NETWORK: reflecting loopback (even further than furthest!)
++ */
++/* Please keep in order and up-to-date w.r.t the following two #defines */
++enum efx_loopback_mode {
++ LOOPBACK_NONE = 0,
++ LOOPBACK_NEAR = 1,
++ LOOPBACK_MAC = 2,
++ LOOPBACK_XGMII = 3,
++ LOOPBACK_XGXS = 4,
++ LOOPBACK_XAUI = 5,
++ LOOPBACK_PHY = 6,
++ LOOPBACK_PHYXS = 7,
++ LOOPBACK_PCS = 8,
++ LOOPBACK_PMAPMD = 9,
++ LOOPBACK_FAR = 10,
++ LOOPBACK_NETWORK = 11,
++ LOOPBACK_MAX
++};
++#define LOOPBACK_TEST_MAX LOOPBACK_FAR
++
++/* These loopbacks occur within the controller */
++#define LOOPBACKS_10G_INTERNAL ((1 << LOOPBACK_XGMII)| \
++ (1 << LOOPBACK_XGXS) | \
++ (1 << LOOPBACK_XAUI))
++
++#define LOOPBACKS_1G_INTERNAL (1 << LOOPBACK_MAC)
++
++#define LOOPBACK_MASK(_efx) \
++ (1 << (_efx)->loopback_mode)
++
++#define LOOPBACK_INTERNAL(_efx) \
++ (((LOOPBACKS_10G_INTERNAL | LOOPBACKS_1G_INTERNAL) & \
++ LOOPBACK_MASK(_efx)) ? 1 : 0)
++
++#define LOOPBACK_CHANGED(_from, _to, _mask) \
++ ((LOOPBACK_MASK(_from) ^ LOOPBACK_MASK(_to)) & \
++ (_mask) ? 1 : 0)
++
++#define LOOPBACK_OUT_OF(_from, _to, _mask) \
++ (((LOOPBACK_MASK(_from) & (_mask)) && \
++ ((LOOPBACK_MASK(_to) & (_mask)) == 0)) ? 1 : 0)
++
++/*****************************************************************************/
+
+/**
-+ * Handle the change of state of the backend to Closing. We must delete our
-+ * device-layer structures now, to ensure that writes are flushed through to
-+ * the backend. Once is this done, we can switch to Closed in
-+ * acknowledgement.
++ * enum reset_type - reset types
++ *
++ * %RESET_TYPE_INVSIBLE, %RESET_TYPE_ALL, %RESET_TYPE_WORLD and
++ * %RESET_TYPE_DISABLE specify the method/scope of the reset. The
++ * other valuesspecify reasons, which efx_schedule_reset() will choose
++ * a method for.
++ *
++ * @RESET_TYPE_INVISIBLE: don't reset the PHYs or interrupts
++ * @RESET_TYPE_ALL: reset everything but PCI core blocks
++ * @RESET_TYPE_WORLD: reset everything, save & restore PCI config
++ * @RESET_TYPE_DISABLE: disable NIC
++ * @RESET_TYPE_MONITOR: reset due to hardware monitor
++ * @RESET_TYPE_INT_ERROR: reset due to internal error
++ * @RESET_TYPE_RX_RECOVERY: reset to recover from RX datapath errors
++ */
++enum reset_type {
++ RESET_TYPE_NONE = -1,
++ RESET_TYPE_INVISIBLE = 0,
++ RESET_TYPE_ALL = 1,
++ RESET_TYPE_WORLD = 2,
++ RESET_TYPE_DISABLE = 3,
++ RESET_TYPE_MAX_METHOD,
++ RESET_TYPE_MONITOR,
++ RESET_TYPE_INT_ERROR,
++ RESET_TYPE_RX_RECOVERY,
++ RESET_TYPE_RX_DESC_FETCH,
++ RESET_TYPE_TX_DESC_FETCH,
++ RESET_TYPE_MAX,
++};
++
++#endif /* EFX_ENUM_H */
+--- linux-2.6.18.8/drivers/net/sfc/ethtool.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/ethtool.c 2008-05-19 00:33:28.841808602 +0300
+@@ -0,0 +1,734 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static void blkfront_closing(struct xenbus_device *dev)
-+{
-+ struct blkfront_info *info = dev->dev.driver_data;
-+ unsigned long flags;
+
-+ DPRINTK("blkfront_closing: %s removed\n", dev->nodename);
++#include <linux/netdevice.h>
++#include <linux/ethtool.h>
++#include <linux/rtnetlink.h>
++#include <asm/uaccess.h>
++#include "net_driver.h"
++#include "selftest.h"
++#include "efx.h"
++#include "ethtool.h"
++#include "falcon.h"
++#include "gmii.h"
+
-+ if (info->rq == NULL)
-+ goto out;
++static int efx_ethtool_set_tx_csum(struct net_device *net_dev, u32 enable);
+
-+ spin_lock_irqsave(&blkif_io_lock, flags);
-+ /* No more blkif_request(). */
-+ blk_stop_queue(info->rq);
-+ /* No more gnttab callback work. */
-+ gnttab_cancel_free_callback(&info->callback);
-+ spin_unlock_irqrestore(&blkif_io_lock, flags);
++struct ethtool_string {
++ char name[ETH_GSTRING_LEN];
++};
+
-+ /* Flush gnttab callback work. Must be done with no locks held. */
-+ flush_scheduled_work();
++struct efx_ethtool_stat {
++ const char *name;
++ enum {
++ EFX_ETHTOOL_STAT_SOURCE_mac_stats,
++ EFX_ETHTOOL_STAT_SOURCE_nic,
++ EFX_ETHTOOL_STAT_SOURCE_channel
++ } source;
++ unsigned offset;
++ u64(*get_stat) (void *field); /* Reader function */
++};
++
++/* Initialiser for a struct #efx_ethtool_stat with type-checking */
++#define EFX_ETHTOOL_STAT(stat_name, source_name, field, field_type, \
++ get_stat_function) { \
++ .name = #stat_name, \
++ .source = EFX_ETHTOOL_STAT_SOURCE_##source_name, \
++ .offset = ((((field_type *) 0) == \
++ &((struct efx_##source_name *)0)->field) ? \
++ offsetof(struct efx_##source_name, field) : \
++ offsetof(struct efx_##source_name, field)), \
++ .get_stat = get_stat_function, \
++}
++
++static u64 efx_get_uint_stat(void *field)
++{
++ return *(unsigned int *)field;
++}
++
++static u64 efx_get_ulong_stat(void *field)
++{
++ return *(unsigned long *)field;
++}
++
++static u64 efx_get_u64_stat(void *field)
++{
++ return *(u64 *) field;
++}
++
++static u64 efx_get_atomic_stat(void *field)
++{
++ return atomic_read((atomic_t *) field);
++}
++
++#define EFX_ETHTOOL_ULONG_MAC_STAT(field) \
++ EFX_ETHTOOL_STAT(field, mac_stats, field, \
++ unsigned long, efx_get_ulong_stat)
++
++#define EFX_ETHTOOL_U64_MAC_STAT(field) \
++ EFX_ETHTOOL_STAT(field, mac_stats, field, \
++ u64, efx_get_u64_stat)
++
++#define EFX_ETHTOOL_UINT_NIC_STAT(name) \
++ EFX_ETHTOOL_STAT(name, nic, n_##name, \
++ unsigned int, efx_get_uint_stat)
++
++#define EFX_ETHTOOL_ATOMIC_NIC_ERROR_STAT(field) \
++ EFX_ETHTOOL_STAT(field, nic, errors.field, \
++ atomic_t, efx_get_atomic_stat)
++
++#define EFX_ETHTOOL_UINT_CHANNEL_STAT(field) \
++ EFX_ETHTOOL_STAT(field, channel, n_##field, \
++ unsigned int, efx_get_uint_stat)
++
++static struct efx_ethtool_stat efx_ethtool_stats[] = {
++ EFX_ETHTOOL_U64_MAC_STAT(tx_bytes),
++ EFX_ETHTOOL_U64_MAC_STAT(tx_good_bytes),
++ EFX_ETHTOOL_U64_MAC_STAT(tx_bad_bytes),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_packets),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_bad),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_pause),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_control),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_unicast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_multicast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_broadcast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_lt64),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_64),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_65_to_127),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_128_to_255),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_256_to_511),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_512_to_1023),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_1024_to_15xx),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_15xx_to_jumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_gtjumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_collision),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_single_collision),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_multiple_collision),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_excessive_collision),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_deferred),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_late_collision),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_excessive_deferred),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_non_tcpudp),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_mac_src_error),
++ EFX_ETHTOOL_ULONG_MAC_STAT(tx_ip_src_error),
++ EFX_ETHTOOL_U64_MAC_STAT(rx_bytes),
++ EFX_ETHTOOL_U64_MAC_STAT(rx_good_bytes),
++ EFX_ETHTOOL_U64_MAC_STAT(rx_bad_bytes),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_packets),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_good),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_bad),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_pause),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_control),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_unicast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_multicast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_broadcast),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_lt64),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_64),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_65_to_127),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_128_to_255),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_256_to_511),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_512_to_1023),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_1024_to_15xx),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_15xx_to_jumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_gtjumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_bad_lt64),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_bad_64_to_15xx),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_bad_15xx_to_jumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_bad_gtjumbo),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_overflow),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_missed),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_false_carrier),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_symbol_error),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_align_error),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_length_error),
++ EFX_ETHTOOL_ULONG_MAC_STAT(rx_internal_error),
++ EFX_ETHTOOL_UINT_NIC_STAT(rx_nodesc_drop_cnt),
++ EFX_ETHTOOL_ATOMIC_NIC_ERROR_STAT(rx_reset),
++ EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_tobe_disc),
++ EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_ip_hdr_chksum_err),
++ EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_tcp_udp_chksum_err),
++ EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_frm_trunc),
++};
++
++/* Number of ethtool statistics */
++#define EFX_ETHTOOL_NUM_STATS ARRAY_SIZE(efx_ethtool_stats)
++
++/**************************************************************************
++ *
++ * Ethtool operations
++ *
++ **************************************************************************
++ */
++
++/* Identify device by flashing LEDs */
++static int efx_ethtool_phys_id(struct net_device *net_dev, u32 seconds)
++{
++ struct efx_nic *efx = net_dev->priv;
++
++ efx->board_info.blink(efx, 1);
++ schedule_timeout_interruptible(seconds * HZ);
++ efx->board_info.blink(efx, 0);
++ return 0;
++}
+
-+ xlvbd_del(info);
++/* This must be called with rtnl_lock held. */
++int efx_ethtool_get_settings(struct net_device *net_dev,
++ struct ethtool_cmd *ecmd)
++{
++ struct efx_nic *efx = net_dev->priv;
+
-+ out:
-+ xenbus_frontend_closed(dev);
++ return efx->mac_op->get_settings(efx, ecmd);
+}
+
-+
-+static int blkfront_remove(struct xenbus_device *dev)
++/* This must be called with rtnl_lock held. */
++int efx_ethtool_set_settings(struct net_device *net_dev,
++ struct ethtool_cmd *ecmd)
+{
-+ struct blkfront_info *info = dev->dev.driver_data;
-+
-+ DPRINTK("blkfront_remove: %s removed\n", dev->nodename);
++ struct efx_nic *efx = net_dev->priv;
++ int rc;
+
-+ blkif_free(info, 0);
++ rc = efx->mac_op->set_settings(efx, ecmd);
++ if (rc)
++ return rc;
+
-+ kfree(info);
++ /* Push the settings to the MAC */
++ efx_reconfigure_port(efx, 0);
+
+ return 0;
+}
+
-+
-+static inline int GET_ID_FROM_FREELIST(
-+ struct blkfront_info *info)
++static void efx_ethtool_get_drvinfo(struct net_device *net_dev,
++ struct ethtool_drvinfo *info)
+{
-+ unsigned long free = info->shadow_free;
-+ BUG_ON(free > BLK_RING_SIZE);
-+ info->shadow_free = info->shadow[free].req.id;
-+ info->shadow[free].req.id = 0x0fffffee; /* debug */
-+ return free;
-+}
++ struct efx_nic *efx = net_dev->priv;
+
-+static inline void ADD_ID_TO_FREELIST(
-+ struct blkfront_info *info, unsigned long id)
-+{
-+ info->shadow[id].req.id = info->shadow_free;
-+ info->shadow[id].request = 0;
-+ info->shadow_free = id;
++ strlcpy(info->driver, EFX_DRIVER_NAME, sizeof(info->driver));
++ strlcpy(info->version, EFX_DRIVER_VERSION, sizeof(info->version));
++ strlcpy(info->bus_info, pci_name(efx->pci_dev), sizeof(info->bus_info));
+}
+
-+static inline void flush_requests(struct blkfront_info *info)
++/**
++ * efx_fill_test - fill in an individual self-test entry
++ * @test_index: Index of the test
++ * @strings: Ethtool strings, or %NULL
++ * @data: Ethtool test results, or %NULL
++ * @test: Pointer to test result (used only if data != %NULL)
++ * @unit_format: Unit name format (e.g. "channel\%d")
++ * @unit_id: Unit id (e.g. 0 for "channel0")
++ * @test_format: Test name format (e.g. "loopback.\%s.tx.sent")
++ * @test_id: Test id (e.g. "PHY" for "loopback.PHY.tx_sent")
++ *
++ * Fill in an individual self-test entry.
++ */
++static void efx_fill_test(unsigned int test_index,
++ struct ethtool_string *strings, u64 *data,
++ int *test, const char *unit_format, int unit_id,
++ const char *test_format, const char *test_id)
+{
-+ int notify;
++ struct ethtool_string unit_str, test_str;
+
-+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&info->ring, notify);
++ /* Fill data value, if applicable */
++ if (data)
++ data[test_index] = *test;
+
-+ if (notify)
-+ notify_remote_via_irq(info->irq);
++ /* Fill string, if applicable */
++ if (strings) {
++ snprintf(unit_str.name, sizeof(unit_str.name),
++ unit_format, unit_id);
++ snprintf(test_str.name, sizeof(test_str.name),
++ test_format, test_id);
++ snprintf(strings[test_index].name,
++ sizeof(strings[test_index].name),
++ "%-9s%-17s", unit_str.name, test_str.name);
++ }
+}
+
-+static void kick_pending_request_queues(struct blkfront_info *info)
-+{
-+ if (!RING_FULL(&info->ring)) {
-+ /* Re-enable calldowns. */
-+ blk_start_queue(info->rq);
-+ /* Kick things off immediately. */
-+ do_blkif_request(info->rq);
-+ }
++#define EFX_PORT_NAME "port%d", 0
++
++/**
++ * efx_fill_loopback_test - fill in a block of loopback self-test entries
++ * @efx: Efx NIC
++ * @lb_tests: Efx loopback self-test results structure
++ * @mode: Loopback test mode
++ * @test_index: Starting index of the test
++ * @strings: Ethtool strings, or %NULL
++ * @data: Ethtool test results, or %NULL
++ *
++ * Fill in a block of loopback self-test entries. Return new test
++ * index.
++ */
++static int efx_fill_loopback_test(struct efx_nic *efx,
++ struct efx_loopback_self_tests *lb_tests,
++ enum efx_loopback_mode mode,
++ unsigned int test_index,
++ struct ethtool_string *strings, u64 *data)
++{
++ struct efx_tx_queue *tx_queue;
++
++ efx_for_each_tx_queue(tx_queue, efx) {
++ efx_fill_test(test_index++, strings, data,
++ &lb_tests->tx_sent[tx_queue->queue],
++ EFX_TX_QUEUE_NAME(tx_queue),
++ "loopback.%s.tx_sent",
++ efx_loopback_mode_names[mode]);
++ efx_fill_test(test_index++, strings, data,
++ &lb_tests->tx_done[tx_queue->queue],
++ EFX_TX_QUEUE_NAME(tx_queue),
++ "loopback.%s.tx_done",
++ efx_loopback_mode_names[mode]);
++ }
++ efx_fill_test(test_index++, strings, data,
++ &lb_tests->rx_good,
++ EFX_PORT_NAME,
++ "loopback.%s.rx_good",
++ efx_loopback_mode_names[mode]);
++ efx_fill_test(test_index++, strings, data,
++ &lb_tests->rx_bad,
++ EFX_PORT_NAME,
++ "loopback.%s.rx_bad",
++ efx_loopback_mode_names[mode]);
++
++ return test_index;
+}
+
-+static void blkif_restart_queue(void *arg)
-+{
-+ struct blkfront_info *info = (struct blkfront_info *)arg;
-+ spin_lock_irq(&blkif_io_lock);
-+ if (info->connected == BLKIF_STATE_CONNECTED)
-+ kick_pending_request_queues(info);
-+ spin_unlock_irq(&blkif_io_lock);
++/**
++ * efx_ethtool_fill_self_tests - get self-test details
++ * @efx: Efx NIC
++ * @tests: Efx self-test results structure, or %NULL
++ * @strings: Ethtool strings, or %NULL
++ * @data: Ethtool test results, or %NULL
++ *
++ * Get self-test number of strings, strings, and/or test results.
++ * Return number of strings (== number of test results).
++ *
++ * The reason for merging these three functions is to make sure that
++ * they can never be inconsistent.
++ */
++static int efx_ethtool_fill_self_tests(struct efx_nic *efx,
++ struct efx_self_tests *tests,
++ struct ethtool_string *strings,
++ u64 *data)
++{
++ struct efx_channel *channel;
++ unsigned int n = 0;
++ enum efx_loopback_mode mode;
++
++ /* Interrupt */
++ efx_fill_test(n++, strings, data, &tests->interrupt,
++ "core", 0, "interrupt", NULL);
++
++ /* Event queues */
++ efx_for_each_channel(channel, efx) {
++ efx_fill_test(n++, strings, data,
++ &tests->eventq_dma[channel->channel],
++ EFX_CHANNEL_NAME(channel),
++ "eventq.dma", NULL);
++ efx_fill_test(n++, strings, data,
++ &tests->eventq_int[channel->channel],
++ EFX_CHANNEL_NAME(channel),
++ "eventq.int", NULL);
++ efx_fill_test(n++, strings, data,
++ &tests->eventq_poll[channel->channel],
++ EFX_CHANNEL_NAME(channel),
++ "eventq.poll", NULL);
++ }
++
++ /* PHY presence */
++ efx_fill_test(n++, strings, data, &tests->phy_ok,
++ EFX_PORT_NAME, "phy_ok", NULL);
++
++ /* Loopback tests */
++ efx_fill_test(n++, strings, data, &tests->loopback_speed,
++ EFX_PORT_NAME, "loopback.speed", NULL);
++ efx_fill_test(n++, strings, data, &tests->loopback_full_duplex,
++ EFX_PORT_NAME, "loopback.full_duplex", NULL);
++ for (mode = LOOPBACK_NONE; mode < LOOPBACK_TEST_MAX; mode++) {
++ if (!(efx->loopback_modes & (1 << mode)))
++ continue;
++ n = efx_fill_loopback_test(efx,
++ &tests->loopback[mode], mode, n,
++ strings, data);
++ }
++
++ return n;
+}
+
-+static void blkif_restart_queue_callback(void *arg)
++static int efx_ethtool_get_stats_count(struct net_device *net_dev)
+{
-+ struct blkfront_info *info = (struct blkfront_info *)arg;
-+ schedule_work(&info->work);
++ return EFX_ETHTOOL_NUM_STATS;
+}
+
-+int blkif_open(struct inode *inode, struct file *filep)
++static int efx_ethtool_self_test_count(struct net_device *net_dev)
+{
-+ struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
-+ info->users++;
-+ return 0;
-+}
++ struct efx_nic *efx = net_dev->priv;
+
++ return efx_ethtool_fill_self_tests(efx, NULL, NULL, NULL);
++}
+
-+int blkif_release(struct inode *inode, struct file *filep)
++static void efx_ethtool_get_strings(struct net_device *net_dev,
++ u32 string_set, u8 *strings)
+{
-+ struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
-+ info->users--;
-+ if (info->users == 0) {
-+ /* Check whether we have been instructed to close. We will
-+ have ignored this request initially, as the device was
-+ still mounted. */
-+ struct xenbus_device * dev = info->xbdev;
-+ enum xenbus_state state = xenbus_read_driver_state(dev->otherend);
++ struct efx_nic *efx = net_dev->priv;
++ struct ethtool_string *ethtool_strings =
++ (struct ethtool_string *)strings;
++ int i;
+
-+ if (state == XenbusStateClosing && info->is_ready)
-+ blkfront_closing(dev);
++ switch (string_set) {
++ case ETH_SS_STATS:
++ for (i = 0; i < EFX_ETHTOOL_NUM_STATS; i++)
++ strncpy(ethtool_strings[i].name,
++ efx_ethtool_stats[i].name,
++ sizeof(ethtool_strings[i].name));
++ break;
++ case ETH_SS_TEST:
++ efx_ethtool_fill_self_tests(efx, NULL,
++ ethtool_strings, NULL);
++ break;
++ default:
++ /* No other string sets */
++ break;
+ }
-+ return 0;
+}
+
-+
-+int blkif_ioctl(struct inode *inode, struct file *filep,
-+ unsigned command, unsigned long argument)
++static void efx_ethtool_get_stats(struct net_device *net_dev,
++ struct ethtool_stats *stats
++ __attribute__ ((unused)), u64 *data)
+{
++ unsigned long flags __attribute__ ((unused));
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_mac_stats *mac_stats = &efx->mac_stats;
++ struct efx_ethtool_stat *stat;
++ struct efx_channel *channel;
+ int i;
+
-+ DPRINTK_IOCTL("command: 0x%x, argument: 0x%lx, dev: 0x%04x\n",
-+ command, (long)argument, inode->i_rdev);
++ EFX_BUG_ON_PARANOID(stats->n_stats != EFX_ETHTOOL_NUM_STATS);
+
-+ switch (command) {
-+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16)
-+ case HDIO_GETGEO: {
-+ struct block_device *bd = inode->i_bdev;
-+ struct hd_geometry geo;
-+ int ret;
++ /* Update MAC and NIC statistics */
++ net_dev->get_stats(net_dev);
++ falcon_update_nic_stats(efx);
+
-+ if (!argument)
-+ return -EINVAL;
++ /* Fill detailed statistics buffer */
++ for (i = 0; i < EFX_ETHTOOL_NUM_STATS; i++) {
++ stat = &efx_ethtool_stats[i];
++ switch (stat->source) {
++ case EFX_ETHTOOL_STAT_SOURCE_mac_stats:
++ data[i] = stat->get_stat((void *)mac_stats +
++ stat->offset);
++ break;
++ case EFX_ETHTOOL_STAT_SOURCE_nic:
++ data[i] = stat->get_stat((void *)efx + stat->offset);
++ break;
++ case EFX_ETHTOOL_STAT_SOURCE_channel:
++ data[i] = 0;
++ efx_for_each_channel(channel, efx)
++ data[i] += stat->get_stat((void *)channel +
++ stat->offset);
++ break;
++ }
++ }
++}
+
-+ geo.start = get_start_sect(bd);
-+ ret = blkif_getgeo(bd, &geo);
-+ if (ret)
-+ return ret;
++static int efx_ethtool_set_tx_csum(struct net_device *net_dev, u32 enable)
++{
++ struct efx_nic *efx = net_dev->priv;
++ int rc;
+
-+ if (copy_to_user((struct hd_geometry __user *)argument, &geo,
-+ sizeof(geo)))
-+ return -EFAULT;
++ rc = ethtool_op_set_tx_csum(net_dev, enable);
++ if (rc)
++ return rc;
+
-+ return 0;
-+ }
-+#endif
-+ case CDROMMULTISESSION:
-+ DPRINTK("FIXME: support multisession CDs later\n");
-+ for (i = 0; i < sizeof(struct cdrom_multisession); i++)
-+ if (put_user(0, (char __user *)(argument + i)))
-+ return -EFAULT;
-+ return 0;
+
-+ case CDROM_GET_CAPABILITY: {
-+ struct blkfront_info *info =
-+ inode->i_bdev->bd_disk->private_data;
-+ struct gendisk *gd = info->gd;
-+ if (gd->flags & GENHD_FL_CD)
-+ return 0;
-+ return -EINVAL;
-+ }
-+ default:
-+ /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
-+ command);*/
-+ return -EINVAL; /* same return as native Linux */
-+ }
++ efx_flush_queues(efx);
+
+ return 0;
+}
+
-+
-+int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
++static int efx_ethtool_set_rx_csum(struct net_device *net_dev, u32 enable)
+{
-+ /* We don't have real geometry info, but let's at least return
-+ values consistent with the size of the device */
-+ sector_t nsect = get_capacity(bd->bd_disk);
-+ sector_t cylinders = nsect;
++ struct efx_nic *efx = net_dev->priv;
++
++ /* No way to stop the hardware doing the checks; we just
++ * ignore the result.
++ */
++ efx->rx_checksum_enabled = (enable ? 1 : 0);
+
-+ hg->heads = 0xff;
-+ hg->sectors = 0x3f;
-+ sector_div(cylinders, hg->heads * hg->sectors);
-+ hg->cylinders = cylinders;
-+ if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
-+ hg->cylinders = 0xffff;
+ return 0;
+}
+
-+
-+/*
-+ * blkif_queue_request
-+ *
-+ * request block io
-+ *
-+ * id: for guest use only.
-+ * operation: BLKIF_OP_{READ,WRITE,PROBE}
-+ * buffer: buffer to read/write into. this should be a
-+ * virtual address in the guest os.
-+ */
-+static int blkif_queue_request(struct request *req)
++static u32 efx_ethtool_get_rx_csum(struct net_device *net_dev)
+{
-+ struct blkfront_info *info = req->rq_disk->private_data;
-+ unsigned long buffer_mfn;
-+ blkif_request_t *ring_req;
-+ struct bio *bio;
-+ struct bio_vec *bvec;
-+ int idx;
-+ unsigned long id;
-+ unsigned int fsect, lsect;
-+ int ref;
-+ grant_ref_t gref_head;
++ struct efx_nic *efx = net_dev->priv;
+
-+ if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
-+ return 1;
++ return efx->rx_checksum_enabled;
++}
+
-+ if (gnttab_alloc_grant_references(
-+ BLKIF_MAX_SEGMENTS_PER_REQUEST, &gref_head) < 0) {
-+ gnttab_request_free_callback(
-+ &info->callback,
-+ blkif_restart_queue_callback,
-+ info,
-+ BLKIF_MAX_SEGMENTS_PER_REQUEST);
-+ return 1;
-+ }
++static void efx_ethtool_self_test(struct net_device *net_dev,
++ struct ethtool_test *test, u64 *data)
++{
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_self_tests efx_tests;
++ int offline, already_up;
++ int rc;
+
-+ /* Fill out a communications ring structure. */
-+ ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
-+ id = GET_ID_FROM_FREELIST(info);
-+ info->shadow[id].request = (unsigned long)req;
++ /* Make sure we've got rtnl lock since we're playing with
++ * interrupts, and calling efx_process_channel_now and others
++ */
++ ASSERT_RTNL();
+
-+ ring_req->id = id;
-+ ring_req->sector_number = (blkif_sector_t)req->sector;
-+ ring_req->handle = info->handle;
++ /* If the NIC isn't in the RUNNING state then exit */
++ if (efx->state != STATE_RUNNING) {
++ rc = -EIO;
++ goto fail1;
++ }
+
-+ ring_req->operation = rq_data_dir(req) ?
-+ BLKIF_OP_WRITE : BLKIF_OP_READ;
-+ if (blk_barrier_rq(req))
-+ ring_req->operation = BLKIF_OP_WRITE_BARRIER;
++ /* Make sure the interface is up. We need interrupts, NAPI
++ * and some RX buffers so this is helpful. NB. The caller has
++ * rtnl_lock so nobody else can call dev_open. */
++ already_up = (efx->net_dev->flags & IFF_UP);
++ if (!already_up) {
++ rc = dev_open(efx->net_dev);
++ if (rc) {
++ EFX_ERR(efx, "failed opening device.\n");
++ goto fail2;
++ }
++ }
+
-+ ring_req->nr_segments = 0;
-+ rq_for_each_bio (bio, req) {
-+ bio_for_each_segment (bvec, bio, idx) {
-+ BUG_ON(ring_req->nr_segments
-+ == BLKIF_MAX_SEGMENTS_PER_REQUEST);
-+ buffer_mfn = page_to_phys(bvec->bv_page) >> PAGE_SHIFT;
-+ fsect = bvec->bv_offset >> 9;
-+ lsect = fsect + (bvec->bv_len >> 9) - 1;
-+ /* install a grant reference. */
-+ ref = gnttab_claim_grant_reference(&gref_head);
-+ BUG_ON(ref == -ENOSPC);
++ memset(&efx_tests, 0, sizeof(efx_tests));
++ offline = (test->flags & ETH_TEST_FL_OFFLINE);
+
-+ gnttab_grant_foreign_access_ref(
-+ ref,
-+ info->xbdev->otherend_id,
-+ buffer_mfn,
-+ rq_data_dir(req) ? GTF_readonly : 0 );
++ /* Perform online self tests first */
++ rc = efx_online_test(efx, &efx_tests);
++ if (rc)
++ goto out;
+
-+ info->shadow[id].frame[ring_req->nr_segments] =
-+ mfn_to_pfn(buffer_mfn);
++ /* Perform offline tests only if online tests passed */
++ if (offline) {
++ /* Stop the kernel from sending packets during the test. The
++ * selftest will be consistently bringing the port up and down
++ * as it moves between loopback modes, so the watchdog timer
++ * probably won't run anyway */
++ efx_stop_queue(efx);
+
-+ ring_req->seg[ring_req->nr_segments] =
-+ (struct blkif_request_segment) {
-+ .gref = ref,
-+ .first_sect = fsect,
-+ .last_sect = lsect };
++ rc = efx_flush_queues(efx);
++ if (rc != 0)
++ goto out_offline;
+
-+ ring_req->nr_segments++;
-+ }
++ rc = efx_offline_test(efx, &efx_tests,
++ efx->loopback_modes);
++ out_offline:
++ efx_wake_queue(efx);
+ }
+
-+ info->ring.req_prod_pvt++;
-+
-+ /* Keep a private copy so we can reissue requests when recovering. */
-+ info->shadow[id].req = *ring_req;
++ /* fall-thru */
++ out:
++ if (!already_up)
++ dev_close(efx->net_dev);
+
-+ gnttab_free_grant_references(gref_head);
++ EFX_LOG(efx, "%s all %sline self-tests\n",
++ rc == 0 ? "passed" : "failed", offline ? "off" : "on");
+
-+ return 0;
++ fail2:
++ fail1:
++ /* Fill ethtool results structures */
++ efx_ethtool_fill_self_tests(efx, &efx_tests, NULL, data);
++ if (rc)
++ test->flags |= ETH_TEST_FL_FAILED;
+}
+
-+/*
-+ * do_blkif_request
-+ * read a block; request is in a request queue
-+ */
-+void do_blkif_request(request_queue_t *rq)
++/* Restart autonegotiation */
++static int efx_ethtool_nway_reset(struct net_device *net_dev)
+{
-+ struct blkfront_info *info = NULL;
-+ struct request *req;
-+ int queued;
-+
-+ DPRINTK("Entered do_blkif_request\n");
++ struct efx_nic *efx = net_dev->priv;
+
-+ queued = 0;
++ return mii_nway_restart(&efx->mii);
++}
+
-+ while ((req = elv_next_request(rq)) != NULL) {
-+ info = req->rq_disk->private_data;
-+ if (!blk_fs_request(req)) {
-+ end_request(req, 0);
-+ continue;
-+ }
++static u32 efx_ethtool_get_link(struct net_device *net_dev)
++{
++ struct efx_nic *efx = net_dev->priv;
+
-+ if (RING_FULL(&info->ring))
-+ goto wait;
++ return efx->link_up;
++}
+
-+ DPRINTK("do_blk_req %p: cmd %p, sec %llx, "
-+ "(%u/%li) buffer:%p [%s]\n",
-+ req, req->cmd, (long long)req->sector,
-+ req->current_nr_sectors,
-+ req->nr_sectors, req->buffer,
-+ rq_data_dir(req) ? "write" : "read");
++static int efx_ethtool_get_coalesce(struct net_device *net_dev,
++ struct ethtool_coalesce *coalesce)
++{
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_tx_queue *tx_queue;
++ struct efx_rx_queue *rx_queue;
++ struct efx_channel *channel;
+
++ memset(coalesce, 0, sizeof(*coalesce));
+
-+ blkdev_dequeue_request(req);
-+ if (blkif_queue_request(req)) {
-+ blk_requeue_request(rq, req);
-+ wait:
-+ /* Avoid pointless unplugs. */
-+ blk_stop_queue(rq);
-+ break;
++ /* Find lowest IRQ moderation across all used TX queues */
++ coalesce->tx_coalesce_usecs_irq = ~((u32) 0);
++ efx_for_each_tx_queue(tx_queue, efx) {
++ channel = tx_queue->channel;
++ if (channel->irq_moderation < coalesce->tx_coalesce_usecs_irq) {
++ if (channel->used_flags != EFX_USED_BY_RX_TX)
++ coalesce->tx_coalesce_usecs_irq =
++ channel->irq_moderation;
++ else
++ coalesce->tx_coalesce_usecs_irq = 0;
+ }
++ }
+
-+ queued++;
++ /* Find lowest IRQ moderation across all used RX queues */
++ coalesce->rx_coalesce_usecs_irq = ~((u32) 0);
++ efx_for_each_rx_queue(rx_queue, efx) {
++ channel = rx_queue->channel;
++ if (channel->irq_moderation < coalesce->rx_coalesce_usecs_irq)
++ coalesce->rx_coalesce_usecs_irq =
++ channel->irq_moderation;
+ }
+
-+ if (queued != 0)
-+ flush_requests(info);
++ return 0;
+}
+
-+
-+static irqreturn_t blkif_int(int irq, void *dev_id, struct pt_regs *ptregs)
++/* Set coalescing parameters
++ * The difficulties occur for shared channels
++ */
++static int efx_ethtool_set_coalesce(struct net_device *net_dev,
++ struct ethtool_coalesce *coalesce)
+{
-+ struct request *req;
-+ blkif_response_t *bret;
-+ RING_IDX i, rp;
-+ unsigned long flags;
-+ struct blkfront_info *info = (struct blkfront_info *)dev_id;
-+ int uptodate;
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_channel *channel;
++ struct efx_tx_queue *tx_queue;
++ unsigned tx_usecs, rx_usecs;
+
-+ spin_lock_irqsave(&blkif_io_lock, flags);
++ if (coalesce->use_adaptive_rx_coalesce ||
++ coalesce->use_adaptive_tx_coalesce)
++ return -EOPNOTSUPP;
+
-+ if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) {
-+ spin_unlock_irqrestore(&blkif_io_lock, flags);
-+ return IRQ_HANDLED;
++ if (coalesce->rx_coalesce_usecs || coalesce->tx_coalesce_usecs) {
++ EFX_ERR(efx, "invalid coalescing setting. "
++ "Only rx/tx_coalesce_usecs_irq are supported\n");
++ return -EOPNOTSUPP;
+ }
+
-+ again:
-+ rp = info->ring.sring->rsp_prod;
-+ rmb(); /* Ensure we see queued responses up to 'rp'. */
++ rx_usecs = coalesce->rx_coalesce_usecs_irq;
++ tx_usecs = coalesce->tx_coalesce_usecs_irq;
+
-+ for (i = info->ring.rsp_cons; i != rp; i++) {
-+ unsigned long id;
-+ int ret;
++ /* If the channel is shared only allow RX parameters to be set */
++ efx_for_each_tx_queue(tx_queue, efx) {
++ if ((tx_queue->channel->used_flags == EFX_USED_BY_RX_TX) &&
++ tx_usecs) {
++ EFX_ERR(efx, "Channel is shared. "
++ "Only RX coalescing may be set\n");
++ return -EOPNOTSUPP;
++ }
++ }
+
-+ bret = RING_GET_RESPONSE(&info->ring, i);
-+ id = bret->id;
-+ req = (struct request *)info->shadow[id].request;
++ efx_init_irq_moderation(efx, tx_usecs, rx_usecs);
+
-+ blkif_completion(&info->shadow[id]);
++ /* Reset channel to pick up new moderation value. Note that
++ * this may change the value of the irq_moderation field
++ * (e.g. to allow for hardware timer granularity).
++ */
++ efx_for_each_channel(channel, efx)
++ falcon_set_int_moderation(channel);
+
-+ ADD_ID_TO_FREELIST(info, id);
++ return 0;
++}
+
-+ uptodate = (bret->status == BLKIF_RSP_OKAY);
-+ switch (bret->operation) {
-+ case BLKIF_OP_WRITE_BARRIER:
-+ if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
-+ printk("blkfront: %s: write barrier op failed\n",
-+ info->gd->disk_name);
-+ uptodate = -EOPNOTSUPP;
-+ info->feature_barrier = 0;
-+ xlvbd_barrier(info);
-+ }
-+ /* fall through */
-+ case BLKIF_OP_READ:
-+ case BLKIF_OP_WRITE:
-+ if (unlikely(bret->status != BLKIF_RSP_OKAY))
-+ DPRINTK("Bad return from blkdev data "
-+ "request: %x\n", bret->status);
++static int efx_ethtool_set_pauseparam(struct net_device *net_dev,
++ struct ethtool_pauseparam *pause)
++{
++ struct efx_nic *efx = net_dev->priv;
++ enum efx_fc_type flow_control = efx->flow_control;
++ int rc;
+
-+ ret = end_that_request_first(req, uptodate,
-+ req->hard_nr_sectors);
-+ BUG_ON(ret);
-+ end_that_request_last(req, uptodate);
-+ break;
-+ default:
-+ BUG();
-+ }
-+ }
++ flow_control &= ~(EFX_FC_RX | EFX_FC_TX | EFX_FC_AUTO);
++ flow_control |= pause->rx_pause ? EFX_FC_RX : 0;
++ flow_control |= pause->tx_pause ? EFX_FC_TX : 0;
++ flow_control |= pause->autoneg ? EFX_FC_AUTO : 0;
+
-+ info->ring.rsp_cons = i;
++ /* Try to push the pause parameters */
++ rc = efx->mac_op->set_pause(efx, flow_control);
++ if (rc)
++ return rc;
+
-+ if (i != info->ring.req_prod_pvt) {
-+ int more_to_do;
-+ RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
-+ if (more_to_do)
-+ goto again;
-+ } else
-+ info->ring.sring->rsp_event = i + 1;
++ /* Push the settings to the MAC */
++ efx_reconfigure_port(efx, 0);
+
-+ kick_pending_request_queues(info);
++ return 0;
++}
+
-+ spin_unlock_irqrestore(&blkif_io_lock, flags);
++static void efx_ethtool_get_pauseparam(struct net_device *net_dev,
++ struct ethtool_pauseparam *pause)
++{
++ struct efx_nic *efx = net_dev->priv;
+
-+ return IRQ_HANDLED;
++ pause->rx_pause = (efx->flow_control & EFX_FC_RX) ? 1 : 0;
++ pause->tx_pause = (efx->flow_control & EFX_FC_TX) ? 1 : 0;
++ pause->autoneg = (efx->flow_control & EFX_FC_AUTO) ? 1 : 0;
+}
+
-+static void blkif_free(struct blkfront_info *info, int suspend)
++
++#if defined(EFX_USE_ETHTOOL_GET_PERM_ADDR)
++static int efx_ethtool_op_get_perm_addr(struct net_device *net_dev,
++ struct ethtool_perm_addr *addr,
++ u8 *data)
+{
-+ /* Prevent new requests being issued until we fix things up. */
-+ spin_lock_irq(&blkif_io_lock);
-+ info->connected = suspend ?
-+ BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
-+ /* No more blkif_request(). */
-+ if (info->rq)
-+ blk_stop_queue(info->rq);
-+ /* No more gnttab callback work. */
-+ gnttab_cancel_free_callback(&info->callback);
-+ spin_unlock_irq(&blkif_io_lock);
++ struct efx_nic *efx = net_dev->priv;
+
-+ /* Flush gnttab callback work. Must be done with no locks held. */
-+ flush_scheduled_work();
++ memcpy(data, efx->mac_address, ETH_ALEN);
+
-+ /* Free resources associated with old device channel. */
-+ if (info->ring_ref != GRANT_INVALID_REF) {
-+ gnttab_end_foreign_access(info->ring_ref,
-+ (unsigned long)info->ring.sring);
-+ info->ring_ref = GRANT_INVALID_REF;
-+ info->ring.sring = NULL;
-+ }
-+ if (info->irq)
-+ unbind_from_irqhandler(info->irq, info);
-+ info->irq = 0;
++ return 0;
+}
++#endif
+
-+static void blkif_completion(struct blk_shadow *s)
-+{
-+ int i;
-+ for (i = 0; i < s->req.nr_segments; i++)
-+ gnttab_end_foreign_access(s->req.seg[i].gref, 0UL);
-+}
++struct ethtool_ops efx_ethtool_ops = {
++ .get_settings = efx_ethtool_get_settings,
++ .set_settings = efx_ethtool_set_settings,
++ .get_drvinfo = efx_ethtool_get_drvinfo,
++ .nway_reset = efx_ethtool_nway_reset,
++ .get_link = efx_ethtool_get_link,
++ .get_coalesce = efx_ethtool_get_coalesce,
++ .set_coalesce = efx_ethtool_set_coalesce,
++ .get_pauseparam = efx_ethtool_get_pauseparam,
++ .set_pauseparam = efx_ethtool_set_pauseparam,
++ .get_rx_csum = efx_ethtool_get_rx_csum,
++ .set_rx_csum = efx_ethtool_set_rx_csum,
++ .get_tx_csum = ethtool_op_get_tx_csum,
++ .set_tx_csum = efx_ethtool_set_tx_csum,
++ .get_sg = ethtool_op_get_sg,
++ .set_sg = ethtool_op_set_sg,
++#if defined(EFX_USE_ETHTOOL_FLAGS)
++ .get_flags = ethtool_op_get_flags,
++ .set_flags = ethtool_op_set_flags,
++#endif
++ .self_test_count = efx_ethtool_self_test_count,
++ .self_test = efx_ethtool_self_test,
++ .get_strings = efx_ethtool_get_strings,
++ .phys_id = efx_ethtool_phys_id,
++ .get_stats_count = efx_ethtool_get_stats_count,
++ .get_ethtool_stats = efx_ethtool_get_stats,
++#if defined(EFX_USE_ETHTOOL_GET_PERM_ADDR)
++ .get_perm_addr = efx_ethtool_op_get_perm_addr,
++#endif
++};
+--- linux-2.6.18.8/drivers/net/sfc/ethtool.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/ethtool.h 2008-05-19 00:33:28.841808602 +0300
+@@ -0,0 +1,44 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static void blkif_recover(struct blkfront_info *info)
-+{
-+ int i;
-+ blkif_request_t *req;
-+ struct blk_shadow *copy;
-+ int j;
++#ifndef EFX_ETHTOOL_H
++#define EFX_ETHTOOL_H
+
-+ /* Stage 1: Make a safe copy of the shadow state. */
-+ copy = kmalloc(sizeof(info->shadow), GFP_KERNEL | __GFP_NOFAIL);
-+ memcpy(copy, info->shadow, sizeof(info->shadow));
++#include "net_driver.h"
+
-+ /* Stage 2: Set up free list. */
-+ memset(&info->shadow, 0, sizeof(info->shadow));
-+ for (i = 0; i < BLK_RING_SIZE; i++)
-+ info->shadow[i].req.id = i+1;
-+ info->shadow_free = info->ring.req_prod_pvt;
-+ info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
++/*
++ * Ethtool support
++ */
+
-+ /* Stage 3: Find pending requests and requeue them. */
-+ for (i = 0; i < BLK_RING_SIZE; i++) {
-+ /* Not in use? */
-+ if (copy[i].request == 0)
-+ continue;
++extern int efx_ethtool_get_settings(struct net_device *net_dev,
++ struct ethtool_cmd *ecmd);
++extern int efx_ethtool_set_settings(struct net_device *net_dev,
++ struct ethtool_cmd *ecmd);
+
-+ /* Grab a request slot and copy shadow state into it. */
-+ req = RING_GET_REQUEST(
-+ &info->ring, info->ring.req_prod_pvt);
-+ *req = copy[i].req;
++extern struct ethtool_ops efx_ethtool_ops;
+
-+ /* We get a new request id, and must reset the shadow state. */
-+ req->id = GET_ID_FROM_FREELIST(info);
-+ memcpy(&info->shadow[req->id], ©[i], sizeof(copy[i]));
++#endif /* EFX_ETHTOOL_H */
+--- linux-2.6.18.8/drivers/net/sfc/extraversion.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/extraversion.h 2008-05-19 00:33:28.841808602 +0300
+@@ -0,0 +1,4 @@
++/*
++ * If compiling on kernels with backported features you may need to
++ * define EFX_DIST_KVER_ symbols here
++ */
+--- linux-2.6.18.8/drivers/net/sfc/falcon.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon.c 2008-05-19 00:33:28.845808833 +0300
+@@ -0,0 +1,3708 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* Rewrite any grant references invalidated by susp/resume. */
-+ for (j = 0; j < req->nr_segments; j++)
-+ gnttab_grant_foreign_access_ref(
-+ req->seg[j].gref,
-+ info->xbdev->otherend_id,
-+ pfn_to_mfn(info->shadow[req->id].frame[j]),
-+ rq_data_dir((struct request *)
-+ info->shadow[req->id].request) ?
-+ GTF_readonly : 0);
-+ info->shadow[req->id].req = *req;
++#include <asm/io.h>
++#include <asm/bitops.h>
++#include <linux/delay.h>
++#include <linux/pci.h>
++#include <linux/module.h>
++#include <linux/seq_file.h>
++#include "net_driver.h"
++#include "bitfield.h"
++#include "efx.h"
++#include "mac.h"
++#include "gmii.h"
++#include "spi.h"
++#include "falcon.h"
++#include "falcon_hwdefs.h"
++#include "falcon_io.h"
++#include "mdio_10g.h"
++#include "phy.h"
++#include "boards.h"
++#include "driverlink.h"
++#include "workarounds.h"
+
-+ info->ring.req_prod_pvt++;
-+ }
++/* Falcon hardware control.
++ * Falcon is the internal codename for the SFC4000 controller that is
++ * present in SFE400X evaluation boards
++ */
+
-+ kfree(copy);
++struct falcon_nic_data {
++ /* Number of entries in each TX queue descriptor cache. */
++ unsigned tx_dc_entries;
++ /* Number of entries in each RX queue descriptor cache. */
++ unsigned rx_dc_entries;
++ /* Base address in SRAM of TX queue descriptor caches. */
++ unsigned tx_dc_base;
++ /* Base address in SRAM of RX queue descriptor caches. */
++ unsigned rx_dc_base;
+
-+ (void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
++ /* Previous loopback mode used in deconfigure_mac_wrapper */
++ enum efx_loopback_mode old_loopback_mode;
+
-+ spin_lock_irq(&blkif_io_lock);
++ /* Driverlink parameters */
++ struct efx_dl_falcon_resources resources;
++};
+
-+ /* Now safe for us to use the shared ring */
-+ info->connected = BLKIF_STATE_CONNECTED;
++/**************************************************************************
++ *
++ * Configurable values
++ *
++ **************************************************************************
++ */
+
-+ /* Send off requeued requests */
-+ flush_requests(info);
++static int disable_dma_stats;
+
-+ /* Kick any other new requests queued since we resumed */
-+ kick_pending_request_queues(info);
++/* Specify the size of the RX descriptor cache */
++static int descriptor_cache_size = 64;
+
-+ spin_unlock_irq(&blkif_io_lock);
-+}
++/*
++ * Override EEPROM/flash type from non-volatile configuration or GPIO;
++ * may need to be specified if bootstrapping from blank flash.
++ */
++static unsigned int eeprom_type = -1;
++static unsigned int flash_type = -1;
+
-+int blkfront_is_ready(struct xenbus_device *dev)
-+{
-+ struct blkfront_info *info = dev->dev.driver_data;
++/* RX FIFO XOFF watermark
++ *
++ * When the amount of the RX FIFO increases used increases past this
++ * watermark send XOFF. Only used if RX flow control is enabled (ethtool -A)
++ * This also has an effect on RX/TX arbitration
++ */
++static int rx_xoff_thresh_bytes = -1;
++module_param(rx_xoff_thresh_bytes, int, 0644);
++MODULE_PARM_DESC(rx_xoff_thresh_bytes, "RX fifo XOFF threshold");
+
-+ return info->is_ready;
-+}
++/* RX FIFO XON watermark
++ *
++ * When the amount of the RX FIFO used decreases below this
++ * watermark send XON. Only used if TX flow control is enabled (ethtool -A)
++ * This also has an effect on RX/TX arbitration
++ */
++static int rx_xon_thresh_bytes = -1;
++module_param(rx_xon_thresh_bytes, int, 0644);
++MODULE_PARM_DESC(rx_xon_thresh_bytes, "RX fifo XON threshold");
+
++/* TX descriptor ring size - min 512 max 4k */
++#define FALCON_TXD_RING_ORDER TX_DESCQ_SIZE_1K
++#define FALCON_TXD_RING_SIZE 1024
++#define FALCON_TXD_RING_MASK (FALCON_TXD_RING_SIZE - 1)
+
-+/* ** Driver Registration ** */
++/* RX descriptor ring size - min 512 max 4k */
++#define FALCON_RXD_RING_ORDER RX_DESCQ_SIZE_1K
++#define FALCON_RXD_RING_SIZE 1024
++#define FALCON_RXD_RING_MASK (FALCON_RXD_RING_SIZE - 1)
+
++/* Event queue size - max 32k */
++#define FALCON_EVQ_ORDER EVQ_SIZE_4K
++#define FALCON_EVQ_SIZE 4096
++#define FALCON_EVQ_MASK (FALCON_EVQ_SIZE - 1)
+
-+static struct xenbus_device_id blkfront_ids[] = {
-+ { "vbd" },
-+ { "" }
-+};
-+MODULE_ALIAS("xen:vbd");
++/* Max number of internal errors. After this resets will not be performed */
++#define FALCON_MAX_INT_ERRORS 4
+
-+static struct xenbus_driver blkfront = {
-+ .name = "vbd",
-+ .owner = THIS_MODULE,
-+ .ids = blkfront_ids,
-+ .probe = blkfront_probe,
-+ .remove = blkfront_remove,
-+ .resume = blkfront_resume,
-+ .otherend_changed = backend_changed,
-+ .is_ready = blkfront_is_ready,
-+};
++/* Maximum period that we wait for flush events. If the flush event
++ * doesn't arrive in this period of time then we check if the queue
++ * was disabled anyway. */
++#define FALCON_FLUSH_TIMEOUT 10 /* 10ms */
+
++/**************************************************************************
++ *
++ * Falcon constants
++ *
++ **************************************************************************
++ */
+
-+static int __init xlblk_init(void)
-+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++/* DMA address mask (up to 46-bit, avoiding compiler warnings)
++ *
++ * Note that it is possible to have a platform with 64-bit longs and
++ * 32-bit DMA addresses, or vice versa. EFX_DMA_MASK takes care of the
++ * platform DMA mask.
++ */
++#if BITS_PER_LONG == 64
++#define FALCON_DMA_MASK EFX_DMA_MASK(0x00003fffffffffffUL)
++#else
++#define FALCON_DMA_MASK EFX_DMA_MASK(0x00003fffffffffffULL)
++#endif
+
-+ return xenbus_register_frontend(&blkfront);
-+}
-+module_init(xlblk_init);
++/* TX DMA length mask (13-bit) */
++#define FALCON_TX_DMA_MASK (8192 - 1)
+
++/* Alignment of special buffers (4KB) */
++#define FALCON_BUF_ALIGN 4096
+
-+static void __exit xlblk_exit(void)
++/* Dummy SRAM size code */
++#define SRM_NB_BSZ_ONCHIP_ONLY (-1)
++
++/* Be nice if these (or equiv.) were in linux/pci_regs.h, but they're not. */
++#define PCI_EXP_DEVCAP_PWR_VAL_LBN (18)
++/* This field takes up bits 26 and 27. */
++#define PCI_EXP_DEVCAP_PWR_SCL_LBN (26)
++#define PCI_EXP_LNKSTA_LNK_WID (0x3f0)
++#define PCI_EXP_LNKSTA_LNK_WID_LBN (4)
++
++
++/**************************************************************************
++ *
++ * Falcon hardware access
++ *
++ **************************************************************************/
++
++/* Read the current event from the event queue */
++static inline efx_qword_t *falcon_event(struct efx_channel *channel,
++ unsigned int index)
+{
-+ return xenbus_unregister_driver(&blkfront);
++ return (((efx_qword_t *) (channel->eventq.addr)) + index);
+}
-+module_exit(xlblk_exit);
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkfront/block.h linux-2.6.18-xen.hg/drivers/xen/blkfront/block.h
---- linux-2.6.18/drivers/xen/blkfront/block.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkfront/block.h 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,143 @@
-+/******************************************************************************
-+ * block.h
-+ *
-+ * Shared definitions between all levels of XenLinux Virtual block devices.
-+ *
-+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
-+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
-+ * Copyright (c) 2004-2005, Christian Limpach
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/* See if an event is present
++ *
++ * We check both the high and low dword of the event for all ones. We
++ * wrote all ones when we cleared the event, and no valid event can
++ * have all ones in either its high or low dwords. This approach is
++ * robust against reordering.
++ *
++ * Note that using a single 64-bit comparison is incorrect; even
++ * though the CPU read will be atomic, the DMA write may not be.
+ */
++static inline int falcon_event_present(efx_qword_t *event)
++{
++ return (!(EFX_DWORD_IS_ALL_ONES(event->dword[0]) |
++ EFX_DWORD_IS_ALL_ONES(event->dword[1])));
++}
+
-+#ifndef __XEN_DRIVERS_BLOCK_H__
-+#define __XEN_DRIVERS_BLOCK_H__
-+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/slab.h>
-+#include <linux/string.h>
-+#include <linux/errno.h>
-+#include <linux/fs.h>
-+#include <linux/hdreg.h>
-+#include <linux/blkdev.h>
-+#include <linux/major.h>
-+#include <asm/hypervisor.h>
-+#include <xen/xenbus.h>
-+#include <xen/gnttab.h>
-+#include <xen/interface/xen.h>
-+#include <xen/interface/io/blkif.h>
-+#include <xen/interface/io/ring.h>
-+#include <asm/io.h>
-+#include <asm/atomic.h>
-+#include <asm/uaccess.h>
++/* Read dword from a Falcon PCIE core register */
++static void falcon_pcie_core_read_reg(struct efx_nic *efx, int address,
++ efx_dword_t *result)
++{
++ efx_oword_t temp;
+
-+#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)
++ BUG_ON(FALCON_REV(efx) < FALCON_REV_B0);
++ BUG_ON(address & 3 || address < 0);
+
-+#if 0
-+#define DPRINTK_IOCTL(_f, _a...) printk(KERN_ALERT _f, ## _a)
-+#else
-+#define DPRINTK_IOCTL(_f, _a...) ((void)0)
-+#endif
++ EFX_POPULATE_OWORD_1(temp, PCIE_CORE_ADDR, address);
+
-+struct xlbd_type_info
-+{
-+ int partn_shift;
-+ int disks_per_major;
-+ char *devname;
-+ char *diskname;
-+};
++ falcon_write(efx, &temp, PCIE_CORE_INDIRECT_REG);
++ falcon_read(efx, &temp, PCIE_CORE_INDIRECT_REG);
++ /* Extract PCIE_CORE_VALUE without byte-swapping */
++ BUILD_BUG_ON(PCIE_CORE_VALUE_LBN != 32 ||
++ PCIE_CORE_VALUE_WIDTH != 32);
++ result->u32[0] = temp.u32[1];
++}
+
-+struct xlbd_major_info
++/* Write dword to a Falcon PCIE core register */
++static void falcon_pcie_core_write_reg(struct efx_nic *efx, int address,
++ efx_dword_t value)
+{
-+ int major;
-+ int index;
-+ int usage;
-+ struct xlbd_type_info *type;
-+};
++ efx_oword_t temp;
+
-+struct blk_shadow {
-+ blkif_request_t req;
-+ unsigned long request;
-+ unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-+};
++ BUG_ON(FALCON_REV(efx) < FALCON_REV_B0);
++ BUG_ON(address & 0x3 || address < 0);
+
-+#define BLK_RING_SIZE __RING_SIZE((blkif_sring_t *)0, PAGE_SIZE)
++ EFX_POPULATE_OWORD_2(temp,
++ PCIE_CORE_ADDR, address,
++ PCIE_CORE_RW, 1);
++ /* Fill PCIE_CORE_VALUE without byte-swapping */
++ BUILD_BUG_ON(PCIE_CORE_VALUE_LBN != 32 ||
++ PCIE_CORE_VALUE_WIDTH != 32);
++ temp.u32[1] = value.u32[0];
++ falcon_write(efx, &temp, PCIE_CORE_INDIRECT_REG);
++}
+
-+/*
-+ * We have one of these per vbd, whether ide, scsi or 'other'. They
-+ * hang in private_data off the gendisk structure. We may end up
-+ * putting all kinds of interesting stuff here :-)
++/**************************************************************************
++ *
++ * I2C bus - this is a bit-bashing interface using GPIO pins
++ * Note that it uses the output enables to tristate the outputs
++ * SDA is the data pin and SCL is the clock
++ *
++ **************************************************************************
+ */
-+struct blkfront_info
++static void falcon_setsdascl(struct efx_i2c_interface *i2c)
+{
-+ struct xenbus_device *xbdev;
-+ dev_t dev;
-+ struct gendisk *gd;
-+ int vdevice;
-+ blkif_vdev_t handle;
-+ int connected;
-+ int ring_ref;
-+ blkif_front_ring_t ring;
-+ unsigned int irq;
-+ struct xlbd_major_info *mi;
-+ request_queue_t *rq;
-+ struct work_struct work;
-+ struct gnttab_free_callback callback;
-+ struct blk_shadow shadow[BLK_RING_SIZE];
-+ unsigned long shadow_free;
-+ int feature_barrier;
-+ int is_ready;
++ efx_oword_t reg;
+
-+ /**
-+ * The number of people holding this device open. We won't allow a
-+ * hot-unplug unless this is 0.
-+ */
-+ int users;
-+};
++ falcon_read(i2c->efx, ®, GPIO_CTL_REG_KER);
++ EFX_SET_OWORD_FIELD(reg, GPIO0_OEN, (i2c->scl ? 0 : 1));
++ EFX_SET_OWORD_FIELD(reg, GPIO3_OEN, (i2c->sda ? 0 : 1));
++ falcon_write(i2c->efx, ®, GPIO_CTL_REG_KER);
++}
+
-+extern spinlock_t blkif_io_lock;
++static int falcon_getsda(struct efx_i2c_interface *i2c)
++{
++ efx_oword_t reg;
+
-+extern int blkif_open(struct inode *inode, struct file *filep);
-+extern int blkif_release(struct inode *inode, struct file *filep);
-+extern int blkif_ioctl(struct inode *inode, struct file *filep,
-+ unsigned command, unsigned long argument);
-+extern int blkif_getgeo(struct block_device *, struct hd_geometry *);
-+extern int blkif_check(dev_t dev);
-+extern int blkif_revalidate(dev_t dev);
-+extern void do_blkif_request (request_queue_t *rq);
++ falcon_read(i2c->efx, ®, GPIO_CTL_REG_KER);
++ return EFX_OWORD_FIELD(reg, GPIO3_IN);
++}
+
-+/* Virtual block-device subsystem. */
-+/* Note that xlvbd_add doesn't call add_disk for you: you're expected
-+ to call add_disk on info->gd once the disk is properly connected
-+ up. */
-+int xlvbd_add(blkif_sector_t capacity, int device,
-+ u16 vdisk_info, u16 sector_size, struct blkfront_info *info);
-+void xlvbd_del(struct blkfront_info *info);
-+int xlvbd_barrier(struct blkfront_info *info);
++static int falcon_getscl(struct efx_i2c_interface *i2c)
++{
++ efx_oword_t reg;
+
-+#endif /* __XEN_DRIVERS_BLOCK_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkfront/Makefile linux-2.6.18-xen.hg/drivers/xen/blkfront/Makefile
---- linux-2.6.18/drivers/xen/blkfront/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkfront/Makefile 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,5 @@
++ falcon_read(i2c->efx, ®, GPIO_CTL_REG_KER);
++ return EFX_DWORD_FIELD(reg, GPIO0_IN);
++}
+
-+obj-$(CONFIG_XEN_BLKDEV_FRONTEND) := xenblk.o
++static struct efx_i2c_bit_operations falcon_i2c_bit_operations = {
++ .setsda = falcon_setsdascl,
++ .setscl = falcon_setsdascl,
++ .getsda = falcon_getsda,
++ .getscl = falcon_getscl,
++ .udelay = 100,
++ .mdelay = 10,
++};
+
-+xenblk-objs := blkfront.o vbd.o
++/**************************************************************************
++ *
++ * Falcon special buffer handling
++ * Special buffers are used for event queues and the TX and RX
++ * descriptor rings.
++ *
++ *************************************************************************/
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blkfront/vbd.c linux-2.6.18-xen.hg/drivers/xen/blkfront/vbd.c
---- linux-2.6.18/drivers/xen/blkfront/vbd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blkfront/vbd.c 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,375 @@
-+/******************************************************************************
-+ * vbd.c
-+ *
-+ * XenLinux virtual block-device driver (xvd).
-+ *
-+ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
-+ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
-+ * Copyright (c) 2004-2005, Christian Limpach
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++/* Adds the relevant entries to the full-mode buffer table. */
++static int
++falcon_pin_special_buffer_full(struct efx_nic *efx,
++ struct efx_special_buffer *buffer)
++{
++ efx_qword_t buf_desc;
++ int index;
++ dma_addr_t dma_addr;
++ int i;
+
-+#include "block.h"
-+#include <linux/blkdev.h>
-+#include <linux/list.h>
++ /* Write buffer descriptors to NIC */
++ for (i = 0; i < buffer->entries; i++) {
++ index = buffer->index + i;
++ dma_addr = buffer->dma_addr + (i * 4096);
++ EFX_LOG(efx, "mapping special buffer %d at %llx\n",
++ index, (unsigned long long)dma_addr);
++ EFX_POPULATE_QWORD_4(buf_desc,
++ IP_DAT_BUF_SIZE, IP_DAT_BUF_SIZE_4K,
++ BUF_ADR_REGION, 0,
++ BUF_ADR_FBUF, (dma_addr >> 12),
++ BUF_OWNER_ID_FBUF, 0);
++ falcon_write_sram(efx, &buf_desc, index);
++ }
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
++ return 0;
++}
+
-+#define BLKIF_MAJOR(dev) ((dev)>>8)
-+#define BLKIF_MINOR(dev) ((dev) & 0xff)
++/* Clears the relevant entries from the buffer table */
++static void
++falcon_clear_special_buffer_full(struct efx_nic *efx,
++ struct efx_special_buffer *buffer)
++{
++ efx_oword_t buf_tbl_upd;
++ unsigned int start = buffer->index;
++ unsigned int end = (buffer->index + buffer->entries - 1);
+
-+/*
-+ * For convenience we distinguish between ide, scsi and 'other' (i.e.,
-+ * potentially combinations of the two) in the naming scheme and in a few other
-+ * places.
-+ */
++ EFX_LOG(efx, "unmapping special buffers %d-%d\n",
++ buffer->index, buffer->index + buffer->entries - 1);
+
-+#define NUM_IDE_MAJORS 10
-+#define NUM_SCSI_MAJORS 17
-+#define NUM_VBD_MAJORS 1
++ EFX_POPULATE_OWORD_4(buf_tbl_upd,
++ BUF_UPD_CMD, 0,
++ BUF_CLR_CMD, 1,
++ BUF_CLR_END_ID, end,
++ BUF_CLR_START_ID, start);
++ falcon_write(efx, &buf_tbl_upd, BUF_TBL_UPD_REG_KER);
++}
+
-+static struct xlbd_type_info xlbd_ide_type = {
-+ .partn_shift = 6,
-+ .disks_per_major = 2,
-+ .devname = "ide",
-+ .diskname = "hd",
-+};
++/*
++ * Allocate a new Falcon special buffer
++ *
++ * This allocates memory for a new buffer, clears it and allocates a
++ * new buffer ID range. It does not write into Falcon's buffer table.
++ *
++ * This call will allocate 4kB buffers, since Falcon can't use 8kB
++ * buffers for event queues and descriptor rings. It will always
++ * allocate an even number of 4kB buffers, since when we're in
++ * half-entry mode for the buffer table we can only deal with pairs of
++ * buffers.
++ */
++static int falcon_alloc_special_buffer(struct efx_nic *efx,
++ struct efx_special_buffer *buffer,
++ unsigned int len)
++{
++ struct falcon_nic_data *nic_data = efx->nic_data;
+
-+static struct xlbd_type_info xlbd_scsi_type = {
-+ .partn_shift = 4,
-+ .disks_per_major = 16,
-+ .devname = "sd",
-+ .diskname = "sd",
-+};
++ /* Round size up to an 8kB boundary (i.e. pairs of 4kB buffers) */
++ len = (len + 8192 - 1) & ~(8192 - 1);
+
-+static struct xlbd_type_info xlbd_vbd_type = {
-+ .partn_shift = 4,
-+ .disks_per_major = 16,
-+ .devname = "xvd",
-+ .diskname = "xvd",
-+};
++ /* Allocate buffer as consistent PCI DMA space */
++ buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
++ &buffer->dma_addr);
++ if (!buffer->addr)
++ return -ENOMEM;
++ buffer->len = len;
++ buffer->entries = len / 4096;
++ BUG_ON(buffer->dma_addr & (FALCON_BUF_ALIGN - 1));
+
-+static struct xlbd_major_info *major_info[NUM_IDE_MAJORS + NUM_SCSI_MAJORS +
-+ NUM_VBD_MAJORS];
++ /* All zeros is a potentially valid event so memset to 0xff */
++ memset(buffer->addr, 0xff, len);
+
-+#define XLBD_MAJOR_IDE_START 0
-+#define XLBD_MAJOR_SCSI_START (NUM_IDE_MAJORS)
-+#define XLBD_MAJOR_VBD_START (NUM_IDE_MAJORS + NUM_SCSI_MAJORS)
++ /* Select new buffer ID */
++ buffer->index = nic_data->resources.buffer_table_min;
++ nic_data->resources.buffer_table_min += buffer->entries;
+
-+#define XLBD_MAJOR_IDE_RANGE XLBD_MAJOR_IDE_START ... XLBD_MAJOR_SCSI_START - 1
-+#define XLBD_MAJOR_SCSI_RANGE XLBD_MAJOR_SCSI_START ... XLBD_MAJOR_VBD_START - 1
-+#define XLBD_MAJOR_VBD_RANGE XLBD_MAJOR_VBD_START ... XLBD_MAJOR_VBD_START + NUM_VBD_MAJORS - 1
++ EFX_LOG(efx, "allocating special buffers %d-%d at %llx+%x "
++ "(virt %p phys %lx)\n", buffer->index,
++ buffer->index + buffer->entries - 1,
++ (unsigned long long)buffer->dma_addr, len,
++ buffer->addr, virt_to_phys(buffer->addr));
+
-+/* Information about our VBDs. */
-+#define MAX_VBDS 64
-+static LIST_HEAD(vbds_list);
++ return 0;
++}
+
-+static struct block_device_operations xlvbd_block_fops =
++/*
++ * Initialise a Falcon special buffer
++ *
++ * This will define a buffer (previously allocated via
++ * falcon_alloc_special_buffer()) in Falcon's buffer table, allowing
++ * it to be used for event queues, descriptor rings etc.
++ */
++static int falcon_init_special_buffer(struct efx_nic *efx,
++ struct efx_special_buffer *buffer)
+{
-+ .owner = THIS_MODULE,
-+ .open = blkif_open,
-+ .release = blkif_release,
-+ .ioctl = blkif_ioctl,
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
-+ .getgeo = blkif_getgeo
-+#endif
-+};
++ EFX_BUG_ON_PARANOID(!buffer->addr);
+
-+DEFINE_SPINLOCK(blkif_io_lock);
++ /* Write buffer descriptors to NIC */
++ return falcon_pin_special_buffer_full(efx, buffer);
++}
+
-+static struct xlbd_major_info *
-+xlbd_alloc_major_info(int major, int minor, int index)
++/* Unmaps a buffer from Falcon and clears the buffer table
++ * entries */
++static void falcon_fini_special_buffer(struct efx_nic *efx,
++ struct efx_special_buffer *buffer)
+{
-+ struct xlbd_major_info *ptr;
+
-+ ptr = kzalloc(sizeof(struct xlbd_major_info), GFP_KERNEL);
-+ if (ptr == NULL)
-+ return NULL;
++ if (!buffer->entries)
++ return;
+
-+ ptr->major = major;
++ falcon_clear_special_buffer_full(efx, buffer);
++}
+
-+ switch (index) {
-+ case XLBD_MAJOR_IDE_RANGE:
-+ ptr->type = &xlbd_ide_type;
-+ ptr->index = index - XLBD_MAJOR_IDE_START;
-+ break;
-+ case XLBD_MAJOR_SCSI_RANGE:
-+ ptr->type = &xlbd_scsi_type;
-+ ptr->index = index - XLBD_MAJOR_SCSI_START;
-+ break;
-+ case XLBD_MAJOR_VBD_RANGE:
-+ ptr->type = &xlbd_vbd_type;
-+ ptr->index = index - XLBD_MAJOR_VBD_START;
-+ break;
-+ }
++/* Release the buffer memory. */
++static void falcon_free_special_buffer(struct efx_nic *efx,
++ struct efx_special_buffer *buffer)
++{
++ if (!buffer->addr)
++ return;
+
-+ if (register_blkdev(ptr->major, ptr->type->devname)) {
-+ kfree(ptr);
-+ return NULL;
-+ }
++ EFX_LOG(efx, "deallocating special buffers %d-%d at %llx+%x "
++ "(virt %p phys %lx)\n", buffer->index,
++ buffer->index + buffer->entries - 1,
++ (unsigned long long)buffer->dma_addr, buffer->len,
++ buffer->addr, virt_to_phys(buffer->addr));
+
-+ printk("xen-vbd: registered block device major %i\n", ptr->major);
-+ major_info[index] = ptr;
-+ return ptr;
++ pci_free_consistent(efx->pci_dev, buffer->len, buffer->addr,
++ buffer->dma_addr);
++ buffer->addr = NULL;
++ buffer->entries = 0;
+}
+
-+static struct xlbd_major_info *
-+xlbd_get_major_info(int vdevice)
-+{
-+ struct xlbd_major_info *mi;
-+ int major, minor, index;
++/**************************************************************************
++ *
++ * Falcon generic buffer handling
++ * These buffers are used for interrupt status and MAC stats
++ *
++ **************************************************************************/
+
-+ major = BLKIF_MAJOR(vdevice);
-+ minor = BLKIF_MINOR(vdevice);
++static int falcon_alloc_buffer(struct efx_nic *efx,
++ struct efx_buffer *buffer, unsigned int len)
++{
++ buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
++ &buffer->dma_addr);
++ if (!buffer->addr)
++ return -ENOMEM;
++ buffer->len = len;
++ memset(buffer->addr, 0, len);
++ return 0;
++}
+
-+ switch (major) {
-+ case IDE0_MAJOR: index = 0; break;
-+ case IDE1_MAJOR: index = 1; break;
-+ case IDE2_MAJOR: index = 2; break;
-+ case IDE3_MAJOR: index = 3; break;
-+ case IDE4_MAJOR: index = 4; break;
-+ case IDE5_MAJOR: index = 5; break;
-+ case IDE6_MAJOR: index = 6; break;
-+ case IDE7_MAJOR: index = 7; break;
-+ case IDE8_MAJOR: index = 8; break;
-+ case IDE9_MAJOR: index = 9; break;
-+ case SCSI_DISK0_MAJOR: index = 10; break;
-+ case SCSI_DISK1_MAJOR ... SCSI_DISK7_MAJOR:
-+ index = 11 + major - SCSI_DISK1_MAJOR;
-+ break;
-+ case SCSI_DISK8_MAJOR ... SCSI_DISK15_MAJOR:
-+ index = 18 + major - SCSI_DISK8_MAJOR;
-+ break;
-+ case SCSI_CDROM_MAJOR: index = 26; break;
-+ default: index = 27; break;
++static void falcon_free_buffer(struct efx_nic *efx, struct efx_buffer *buffer)
++{
++ if (buffer->addr) {
++ pci_free_consistent(efx->pci_dev, buffer->len,
++ buffer->addr, buffer->dma_addr);
++ buffer->addr = NULL;
+ }
-+
-+ mi = ((major_info[index] != NULL) ? major_info[index] :
-+ xlbd_alloc_major_info(major, minor, index));
-+ if (mi)
-+ mi->usage++;
-+ return mi;
+}
+
-+static void
-+xlbd_put_major_info(struct xlbd_major_info *mi)
++/**************************************************************************
++ *
++ * Falcon TX path
++ *
++ **************************************************************************/
++
++/* Returns a pointer to the specified transmit descriptor in the TX
++ * descriptor queue belonging to the specified channel.
++ */
++static inline efx_qword_t *falcon_tx_desc(struct efx_tx_queue *tx_queue,
++ unsigned int index)
+{
-+ mi->usage--;
-+ /* XXX: release major if 0 */
++ return (((efx_qword_t *) (tx_queue->txd.addr)) + index);
+}
+
-+static int
-+xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
++/* Update TX descriptor write pointer
++ * This writes to the TX_DESC_WPTR register for the specified
++ * channel's transmit descriptor ring.
++ */
++static inline void falcon_notify_tx_desc(struct efx_tx_queue *tx_queue)
+{
-+ request_queue_t *rq;
++ unsigned write_ptr;
++ efx_dword_t reg;
+
-+ rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
-+ if (rq == NULL)
-+ return -1;
++ write_ptr = tx_queue->write_count & FALCON_TXD_RING_MASK;
++ EFX_POPULATE_DWORD_1(reg, TX_DESC_WPTR_DWORD, write_ptr);
++ falcon_writel_page(tx_queue->efx, ®,
++ TX_DESC_UPD_REG_KER_DWORD, tx_queue->queue);
++}
+
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,10)
-+ elevator_init(rq, "noop");
++
++/* For each entry inserted into the software descriptor ring, create a
++ * descriptor in the hardware TX descriptor ring (in host memory), and
++ * write a doorbell.
++ */
++#if defined(EFX_USE_FASTCALL)
++void fastcall falcon_push_buffers(struct efx_tx_queue *tx_queue)
+#else
-+ elevator_init(rq, &elevator_noop);
++void falcon_push_buffers(struct efx_tx_queue *tx_queue)
+#endif
++{
+
-+ /* Hard sector size and max sectors impersonate the equiv. hardware. */
-+ blk_queue_hardsect_size(rq, sector_size);
-+ blk_queue_max_sectors(rq, 512);
++ struct efx_tx_buffer *buffer;
++ efx_qword_t *txd;
++ unsigned write_ptr;
+
-+ /* Each segment in a request is up to an aligned page in size. */
-+ blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
-+ blk_queue_max_segment_size(rq, PAGE_SIZE);
++ BUG_ON(tx_queue->write_count == tx_queue->insert_count);
+
-+ /* Ensure a merged request will fit in a single I/O ring slot. */
-+ blk_queue_max_phys_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
-+ blk_queue_max_hw_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
++ do {
++ write_ptr = tx_queue->write_count & FALCON_TXD_RING_MASK;
++ buffer = &tx_queue->buffer[write_ptr];
++ txd = falcon_tx_desc(tx_queue, write_ptr);
++ ++tx_queue->write_count;
+
-+ /* Make sure buffer addresses are sector-aligned. */
-+ blk_queue_dma_alignment(rq, 511);
++ /* Create TX descriptor ring entry */
++ EFX_POPULATE_QWORD_5(*txd,
++ TX_KER_PORT, 0,
++ TX_KER_CONT, buffer->continuation,
++ TX_KER_BYTE_CNT, buffer->len,
++ TX_KER_BUF_REGION, 0,
++ TX_KER_BUF_ADR, buffer->dma_addr);
++ } while (tx_queue->write_count != tx_queue->insert_count);
+
-+ /* Make sure we don't use bounce buffers. */
-+ blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY);
++ wmb(); /* Ensure descriptors are written before they are fetched */
++ falcon_notify_tx_desc(tx_queue);
++}
+
-+ gd->queue = rq;
++/* Allocate hardware resources for a TX queue */
++int falcon_probe_tx(struct efx_tx_queue *tx_queue)
++{
++ struct efx_nic *efx = tx_queue->efx;
++ struct falcon_nic_data *nic_data = efx->nic_data;
++ int rc;
++
++ rc = falcon_alloc_special_buffer(efx, &tx_queue->txd,
++ FALCON_TXD_RING_SIZE *
++ sizeof(efx_qword_t));
++ if (rc)
++ return rc;
++
++ nic_data->resources.txq_min = max(nic_data->resources.txq_min,
++ (unsigned)tx_queue->queue + 1);
+
+ return 0;
+}
+
-+static int
-+xlvbd_alloc_gendisk(int minor, blkif_sector_t capacity, int vdevice,
-+ u16 vdisk_info, u16 sector_size,
-+ struct blkfront_info *info)
++/* Prepare channel's TX datapath. */
++int falcon_init_tx(struct efx_tx_queue *tx_queue)
+{
-+ struct gendisk *gd;
-+ struct xlbd_major_info *mi;
-+ int nr_minors = 1;
-+ int err = -ENODEV;
-+ unsigned int offset;
++ efx_oword_t tx_desc_ptr;
++ struct efx_nic *efx = tx_queue->efx;
++ int rc;
+
-+ BUG_ON(info->gd != NULL);
-+ BUG_ON(info->mi != NULL);
-+ BUG_ON(info->rq != NULL);
++ /* Pin TX descriptor ring */
++ rc = falcon_init_special_buffer(efx, &tx_queue->txd);
++ if (rc)
++ return rc;
+
-+ mi = xlbd_get_major_info(vdevice);
-+ if (mi == NULL)
-+ goto out;
-+ info->mi = mi;
++ /* Push TX descriptor ring to card */
++ EFX_POPULATE_OWORD_10(tx_desc_ptr,
++ TX_DESCQ_EN, 1,
++ TX_ISCSI_DDIG_EN, 0,
++ TX_ISCSI_HDIG_EN, 0,
++ TX_DESCQ_BUF_BASE_ID, tx_queue->txd.index,
++ TX_DESCQ_EVQ_ID, tx_queue->channel->evqnum,
++ TX_DESCQ_OWNER_ID, 0,
++ TX_DESCQ_LABEL, tx_queue->queue,
++ TX_DESCQ_SIZE, FALCON_TXD_RING_ORDER,
++ TX_DESCQ_TYPE, 0, /* kernel queue */
++ TX_NON_IP_DROP_DIS_B0, 1);
++
++ if (FALCON_REV(efx) >= FALCON_REV_B0) {
++ int csum = !(efx->net_dev->features & NETIF_F_IP_CSUM);
++ EFX_SET_OWORD_FIELD(tx_desc_ptr, TX_IP_CHKSM_DIS_B0, csum);
++ EFX_SET_OWORD_FIELD(tx_desc_ptr, TX_TCP_CHKSM_DIS_B0, csum);
++ }
++
++ falcon_write_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
++ tx_queue->queue);
++
++ if (FALCON_REV(efx) < FALCON_REV_B0) {
++ efx_oword_t reg;
++
++ /* Only 128 bits in this register */
++ BUG_ON(tx_queue->queue >= 128);
++
++ falcon_read(efx, ®, TX_CHKSM_CFG_REG_KER_A1);
++ if (efx->net_dev->features & NETIF_F_IP_CSUM)
++ clear_bit_le(tx_queue->queue, (void *)®);
++ else
++ set_bit_le(tx_queue->queue, (void *)®);
++ falcon_write(efx, ®, TX_CHKSM_CFG_REG_KER_A1);
++ }
+
-+ if ((minor & ((1 << mi->type->partn_shift) - 1)) == 0)
-+ nr_minors = 1 << mi->type->partn_shift;
++ return 0;
++}
+
-+ gd = alloc_disk(nr_minors);
-+ if (gd == NULL)
-+ goto out;
++static int falcon_flush_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ struct efx_nic *efx = tx_queue->efx;
++ struct efx_channel *channel = &efx->channel[0];
++ efx_oword_t tx_flush_descq;
++ unsigned int read_ptr, i;
+
-+ offset = mi->index * mi->type->disks_per_major +
-+ (minor >> mi->type->partn_shift);
-+ if (nr_minors > 1) {
-+ if (offset < 26) {
-+ sprintf(gd->disk_name, "%s%c",
-+ mi->type->diskname, 'a' + offset );
-+ }
-+ else {
-+ sprintf(gd->disk_name, "%s%c%c",
-+ mi->type->diskname,
-+ 'a' + ((offset/26)-1), 'a' + (offset%26) );
-+ }
-+ }
-+ else {
-+ if (offset < 26) {
-+ sprintf(gd->disk_name, "%s%c%d",
-+ mi->type->diskname,
-+ 'a' + offset,
-+ minor & ((1 << mi->type->partn_shift) - 1));
-+ }
-+ else {
-+ sprintf(gd->disk_name, "%s%c%c%d",
-+ mi->type->diskname,
-+ 'a' + ((offset/26)-1), 'a' + (offset%26),
-+ minor & ((1 << mi->type->partn_shift) - 1));
++ /* Post a flush command */
++ EFX_POPULATE_OWORD_2(tx_flush_descq,
++ TX_FLUSH_DESCQ_CMD, 1,
++ TX_FLUSH_DESCQ, tx_queue->queue);
++ falcon_write(efx, &tx_flush_descq, TX_FLUSH_DESCQ_REG_KER);
++ msleep(FALCON_FLUSH_TIMEOUT);
++
++ /* If the NIC is resetting then don't bother checking */
++ if (EFX_WORKAROUND_7803(efx) || (efx->state == STATE_RESETTING))
++ return 0;
++
++ /* Look for a flush completed event */
++ read_ptr = channel->eventq_read_ptr;
++ for (i = 0; i < FALCON_EVQ_SIZE; ++i) {
++ efx_qword_t *event = falcon_event(channel, read_ptr);
++ int ev_code, ev_sub_code, ev_queue;
++ if (!falcon_event_present(event))
++ break;
++
++ ev_code = EFX_QWORD_FIELD(*event, EV_CODE);
++ ev_sub_code = EFX_QWORD_FIELD(*event, DRIVER_EV_SUB_CODE);
++ ev_queue = EFX_QWORD_FIELD(*event, DRIVER_EV_TX_DESCQ_ID);
++ if ((ev_sub_code == TX_DESCQ_FLS_DONE_EV_DECODE) &&
++ (ev_queue == tx_queue->queue)) {
++ EFX_LOG(efx, "tx queue %d flush command succesful\n",
++ tx_queue->queue);
++ return 0;
+ }
++
++ read_ptr = (read_ptr + 1) & FALCON_EVQ_MASK;
+ }
+
-+ gd->major = mi->major;
-+ gd->first_minor = minor;
-+ gd->fops = &xlvbd_block_fops;
-+ gd->private_data = info;
-+ gd->driverfs_dev = &(info->xbdev->dev);
-+ set_capacity(gd, capacity);
++ if (EFX_WORKAROUND_11557(efx)) {
++ efx_oword_t reg;
++ int enabled;
+
-+ if (xlvbd_init_blk_queue(gd, sector_size)) {
-+ del_gendisk(gd);
-+ goto out;
++ falcon_read_table(efx, ®, efx->type->txd_ptr_tbl_base,
++ tx_queue->queue);
++ enabled = EFX_OWORD_FIELD(reg, TX_DESCQ_EN);
++ if (!enabled) {
++ EFX_LOG(efx, "tx queue %d disabled without a "
++ "flush event seen\n", tx_queue->queue);
++ return 0;
++ }
+ }
+
-+ info->rq = gd->queue;
-+ info->gd = gd;
++ EFX_ERR(efx, "tx queue %d flush command timed out\n", tx_queue->queue);
++ return -ETIMEDOUT;
++}
+
-+ if (info->feature_barrier)
-+ xlvbd_barrier(info);
++void falcon_fini_tx(struct efx_tx_queue *tx_queue)
++{
++ struct efx_nic *efx = tx_queue->efx;
++ efx_oword_t tx_desc_ptr;
+
-+ if (vdisk_info & VDISK_READONLY)
-+ set_disk_ro(gd, 1);
++ /* Stop the hardware using the queue */
++ if (falcon_flush_tx_queue(tx_queue))
++ EFX_ERR(efx, "failed to flush tx queue %d\n", tx_queue->queue);
+
-+ if (vdisk_info & VDISK_REMOVABLE)
-+ gd->flags |= GENHD_FL_REMOVABLE;
++ /* Remove TX descriptor ring from card */
++ EFX_ZERO_OWORD(tx_desc_ptr);
++ falcon_write_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
++ tx_queue->queue);
+
-+ if (vdisk_info & VDISK_CDROM)
-+ gd->flags |= GENHD_FL_CD;
++ /* Unpin TX descriptor ring */
++ falcon_fini_special_buffer(efx, &tx_queue->txd);
++}
+
-+ return 0;
++/* Free buffers backing TX queue */
++void falcon_remove_tx(struct efx_tx_queue *tx_queue)
++{
++ falcon_free_special_buffer(tx_queue->efx, &tx_queue->txd);
++}
+
-+ out:
-+ if (mi)
-+ xlbd_put_major_info(mi);
-+ info->mi = NULL;
-+ return err;
++/**************************************************************************
++ *
++ * Falcon RX path
++ *
++ **************************************************************************/
++
++/* Returns a pointer to the specified transmit descriptor in the RX
++ * descriptor queue.
++ */
++static inline efx_qword_t *falcon_rx_desc(struct efx_rx_queue *rx_queue,
++ unsigned int index)
++{
++ return (((efx_qword_t *) (rx_queue->rxd.addr)) + index);
+}
+
-+int
-+xlvbd_add(blkif_sector_t capacity, int vdevice, u16 vdisk_info,
-+ u16 sector_size, struct blkfront_info *info)
++/* This creates an entry in the RX descriptor queue corresponding to
++ * the receive buffer.
++ */
++static inline void falcon_build_rx_desc(struct efx_rx_queue *rx_queue,
++ unsigned index)
+{
-+ struct block_device *bd;
-+ int err = 0;
++ struct efx_rx_buffer *rx_buf;
++ efx_qword_t *rxd;
+
-+ info->dev = MKDEV(BLKIF_MAJOR(vdevice), BLKIF_MINOR(vdevice));
++ rxd = falcon_rx_desc(rx_queue, index);
++ rx_buf = efx_rx_buffer(rx_queue, index);
++ EFX_POPULATE_QWORD_3(*rxd,
++ RX_KER_BUF_SIZE,
++ rx_buf->len -
++ rx_queue->efx->type->rx_buffer_padding,
++ RX_KER_BUF_REGION, 0,
++ RX_KER_BUF_ADR, rx_buf->dma_addr);
++}
+
-+ bd = bdget(info->dev);
-+ if (bd == NULL)
-+ return -ENODEV;
++/* This writes to the RX_DESC_WPTR register for the specified receive
++ * descriptor ring.
++ */
++#if defined(EFX_USE_FASTCALL)
++void fastcall falcon_notify_rx_desc(struct efx_rx_queue *rx_queue)
++#else
++void falcon_notify_rx_desc(struct efx_rx_queue *rx_queue)
++#endif
++{
++ efx_dword_t reg;
++ unsigned write_ptr;
+
-+ err = xlvbd_alloc_gendisk(BLKIF_MINOR(vdevice), capacity, vdevice,
-+ vdisk_info, sector_size, info);
++ while (rx_queue->notified_count != rx_queue->added_count) {
++ falcon_build_rx_desc(rx_queue,
++ rx_queue->notified_count &
++ FALCON_RXD_RING_MASK);
++ ++rx_queue->notified_count;
++ }
+
-+ bdput(bd);
-+ return err;
++ wmb();
++ write_ptr = rx_queue->added_count & FALCON_RXD_RING_MASK;
++ EFX_POPULATE_DWORD_1(reg, RX_DESC_WPTR_DWORD, write_ptr);
++ falcon_writel_page(rx_queue->efx, ®,
++ RX_DESC_UPD_REG_KER_DWORD, rx_queue->queue);
+}
+
-+void
-+xlvbd_del(struct blkfront_info *info)
++int falcon_probe_rx(struct efx_rx_queue *rx_queue)
+{
-+ if (info->mi == NULL)
-+ return;
++ struct efx_nic *efx = rx_queue->efx;
++ struct falcon_nic_data *nic_data = efx->nic_data;
++ int rc;
+
-+ BUG_ON(info->gd == NULL);
-+ del_gendisk(info->gd);
-+ put_disk(info->gd);
-+ info->gd = NULL;
++ rc = falcon_alloc_special_buffer(efx, &rx_queue->rxd,
++ FALCON_RXD_RING_SIZE *
++ sizeof(efx_qword_t));
++ if (rc)
++ return rc;
+
-+ xlbd_put_major_info(info->mi);
-+ info->mi = NULL;
++ /* Increment the rxq_min counter */
++ nic_data->resources.rxq_min = max(nic_data->resources.rxq_min,
++ (unsigned)rx_queue->queue + 1);
+
-+ BUG_ON(info->rq == NULL);
-+ blk_cleanup_queue(info->rq);
-+ info->rq = NULL;
++ return 0;
+}
+
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
-+int
-+xlvbd_barrier(struct blkfront_info *info)
++int falcon_init_rx(struct efx_rx_queue *rx_queue)
+{
-+ int err;
++ efx_oword_t rx_desc_ptr;
++ struct efx_nic *efx = rx_queue->efx;
++ int rc;
++ int is_b0 = FALCON_REV(efx) >= FALCON_REV_B0;
++ int iscsi_digest_en = is_b0;
+
-+ err = blk_queue_ordered(info->rq,
-+ info->feature_barrier ? QUEUE_ORDERED_DRAIN : QUEUE_ORDERED_NONE, NULL);
-+ if (err)
-+ return err;
-+ printk(KERN_INFO "blkfront: %s: barriers %s\n",
-+ info->gd->disk_name, info->feature_barrier ? "enabled" : "disabled");
++ EFX_LOG(efx, "RX queue %d ring in special buffers %d-%d\n",
++ rx_queue->queue, rx_queue->rxd.index,
++ rx_queue->rxd.index + rx_queue->rxd.entries - 1);
++
++ /* Pin RX descriptor ring */
++ rc = falcon_init_special_buffer(efx, &rx_queue->rxd);
++ if (rc)
++ return rc;
++
++ /* Push RX descriptor ring to card */
++ EFX_POPULATE_OWORD_10(rx_desc_ptr,
++ RX_ISCSI_DDIG_EN, iscsi_digest_en,
++ RX_ISCSI_HDIG_EN, iscsi_digest_en,
++ RX_DESCQ_BUF_BASE_ID, rx_queue->rxd.index,
++ RX_DESCQ_EVQ_ID, rx_queue->channel->evqnum,
++ RX_DESCQ_OWNER_ID, 0,
++ RX_DESCQ_LABEL, rx_queue->queue,
++ RX_DESCQ_SIZE, FALCON_RXD_RING_ORDER,
++ RX_DESCQ_TYPE, 0 /* kernel queue */ ,
++ /* For >=B0 this is scatter so disable */
++ RX_DESCQ_JUMBO, !is_b0,
++ RX_DESCQ_EN, 1);
++ falcon_write_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
++ rx_queue->queue);
+ return 0;
+}
-+#else
-+int
-+xlvbd_barrier(struct blkfront_info *info)
++
++static int falcon_flush_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ printk(KERN_INFO "blkfront: %s: barriers disabled\n", info->gd->disk_name);
-+ return -ENOSYS;
++ struct efx_nic *efx = rx_queue->efx;
++ struct efx_channel *channel = &efx->channel[0];
++ unsigned int read_ptr, i;
++ efx_oword_t rx_flush_descq;
++
++ /* Post a flush command */
++ EFX_POPULATE_OWORD_2(rx_flush_descq,
++ RX_FLUSH_DESCQ_CMD, 1,
++ RX_FLUSH_DESCQ, rx_queue->queue);
++
++ falcon_write(efx, &rx_flush_descq, RX_FLUSH_DESCQ_REG_KER);
++ msleep(FALCON_FLUSH_TIMEOUT);
++
++ /* If the NIC is resetting then don't bother checking */
++ if (EFX_WORKAROUND_7803(efx) || (efx->state == STATE_RESETTING))
++ return 0;
++
++ /* Look for a flush completed event */
++ read_ptr = channel->eventq_read_ptr;
++ for (i = 0; i < FALCON_EVQ_SIZE; ++i) {
++ efx_qword_t *event = falcon_event(channel, read_ptr);
++ int ev_code, ev_sub_code, ev_queue, ev_failed;
++ if (!falcon_event_present(event))
++ break;
++
++ ev_code = EFX_QWORD_FIELD(*event, EV_CODE);
++ ev_sub_code = EFX_QWORD_FIELD(*event, DRIVER_EV_SUB_CODE);
++ ev_queue = EFX_QWORD_FIELD(*event, DRIVER_EV_RX_DESCQ_ID);
++ ev_failed = EFX_QWORD_FIELD(*event, DRIVER_EV_RX_FLUSH_FAIL);
++
++ if ((ev_sub_code == RX_DESCQ_FLS_DONE_EV_DECODE) &&
++ (ev_queue == rx_queue->queue)) {
++ if (ev_failed) {
++ EFX_INFO(efx, "rx queue %d flush command "
++ "failed\n", rx_queue->queue);
++ return -EAGAIN;
++ } else {
++ EFX_LOG(efx, "rx queue %d flush command "
++ "succesful\n", rx_queue->queue);
++ return 0;
++ }
++ }
++
++ read_ptr = (read_ptr + 1) & FALCON_EVQ_MASK;
++ }
++
++ if (EFX_WORKAROUND_11557(efx)) {
++ efx_oword_t reg;
++ int enabled;
++
++ falcon_read_table(efx, ®, efx->type->rxd_ptr_tbl_base,
++ rx_queue->queue);
++ enabled = EFX_OWORD_FIELD(reg, RX_DESCQ_EN);
++ if (!enabled) {
++ EFX_LOG(efx, "rx queue %d disabled without a "
++ "flush event seen\n", rx_queue->queue);
++ return 0;
++ }
++ }
++
++ EFX_ERR(efx, "rx queue %d flush command timed out\n", rx_queue->queue);
++ return -ETIMEDOUT;
+}
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blktap/blktap.c linux-2.6.18-xen.hg/drivers/xen/blktap/blktap.c
---- linux-2.6.18/drivers/xen/blktap/blktap.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blktap/blktap.c 2007-12-23 11:15:33.547908998 +0100
-@@ -0,0 +1,1632 @@
-+/******************************************************************************
-+ * drivers/xen/blktap/blktap.c
-+ *
-+ * Back-end driver for user level virtual block devices. This portion of the
-+ * driver exports a 'unified' block-device interface that can be accessed
-+ * by any operating system that implements a compatible front end. Requests
-+ * are remapped to a user-space memory region.
++
++void falcon_fini_rx(struct efx_rx_queue *rx_queue)
++{
++ efx_oword_t rx_desc_ptr;
++ struct efx_nic *efx = rx_queue->efx;
++ int i, rc;
++
++ /* Try and flush the rx queue. This may need to be repeated */
++ for (i = 0; i < 5; i++) {
++ rc = falcon_flush_rx_queue(rx_queue);
++ if (rc == -EAGAIN)
++ continue;
++ break;
++ }
++ if (rc)
++ EFX_ERR(efx, "failed to flush rx queue %d\n", rx_queue->queue);
++
++ /* Remove RX descriptor ring from card */
++ EFX_ZERO_OWORD(rx_desc_ptr);
++ falcon_write_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
++ rx_queue->queue);
++
++ /* Unpin RX descriptor ring */
++ falcon_fini_special_buffer(efx, &rx_queue->rxd);
++}
++
++/* Free buffers backing RX queue */
++void falcon_remove_rx(struct efx_rx_queue *rx_queue)
++{
++ falcon_free_special_buffer(rx_queue->efx, &rx_queue->rxd);
++}
++
++/**************************************************************************
+ *
-+ * Based on the blkback driver code.
-+ *
-+ * Copyright (c) 2004-2005, Andrew Warfield and Julian Chesterfield
++ * Falcon event queue processing
++ * Event queues are processed by per-channel tasklets.
+ *
-+ * Clean ups and fix ups:
-+ * Copyright (c) 2006, Steven Rostedt - Red Hat, Inc.
++ **************************************************************************/
++
++/* Update a channel's event queue's read pointer (RPTR) register
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * This writes the EVQ_RPTR_REG register for the specified channel's
++ * event queue.
++ *
++ * Note that EVQ_RPTR_REG contains the index of the "last read" event,
++ * whereas channel->eventq_read_ptr contains the index of the "next to
++ * read" event.
+ */
++#if defined(EFX_USE_FASTCALL)
++void fastcall falcon_eventq_read_ack(struct efx_channel *channel)
++#else
++void falcon_eventq_read_ack(struct efx_channel *channel)
++#endif
++{
++ efx_dword_t reg;
++ struct efx_nic *efx = channel->efx;
+
-+#include <linux/spinlock.h>
-+#include <linux/kthread.h>
-+#include <linux/list.h>
-+#include <asm/hypervisor.h>
-+#include "common.h"
-+#include <xen/balloon.h>
-+#include <xen/driver_util.h>
-+#include <linux/kernel.h>
-+#include <linux/fs.h>
-+#include <linux/mm.h>
-+#include <linux/errno.h>
-+#include <linux/major.h>
-+#include <linux/gfp.h>
-+#include <linux/poll.h>
-+#include <asm/tlbflush.h>
++ EFX_POPULATE_DWORD_1(reg, EVQ_RPTR_DWORD, channel->eventq_read_ptr);
++ falcon_writel_table(efx, ®, efx->type->evq_rptr_tbl_base,
++ channel->evqnum);
++}
+
-+#define MAX_TAP_DEV 256 /*the maximum number of tapdisk ring devices */
-+#define MAX_DEV_NAME 100 /*the max tapdisk ring device name e.g. blktap0 */
++/* Use HW to insert a SW defined event */
++void falcon_generate_event(struct efx_channel *channel, efx_qword_t *event)
++{
++ efx_oword_t drv_ev_reg;
+
-+/*
-+ * The maximum number of requests that can be outstanding at any time
-+ * is determined by
-+ *
-+ * [mmap_alloc * MAX_PENDING_REQS * BLKIF_MAX_SEGMENTS_PER_REQUEST]
-+ *
-+ * where mmap_alloc < MAX_DYNAMIC_MEM.
++ EFX_POPULATE_OWORD_2(drv_ev_reg,
++ DRV_EV_QID, channel->evqnum,
++ DRV_EV_DATA,
++ EFX_QWORD_FIELD64(*event, WHOLE_EVENT));
++ falcon_write(channel->efx, &drv_ev_reg, DRV_EV_REG_KER);
++}
++
++/* Handle a transmit completion event
+ *
-+ * TODO:
-+ * mmap_alloc is initialised to 2 and should be adjustable on the fly via
-+ * sysfs.
++ * Falcon batches TX completion events; the message we receive is of
++ * the form "complete all TX events up to this index".
+ */
-+#define BLK_RING_SIZE __RING_SIZE((blkif_sring_t *)0, PAGE_SIZE)
-+#define MAX_DYNAMIC_MEM BLK_RING_SIZE
-+#define MAX_PENDING_REQS BLK_RING_SIZE
-+#define MMAP_PAGES (MAX_PENDING_REQS * BLKIF_MAX_SEGMENTS_PER_REQUEST)
-+#define MMAP_VADDR(_start, _req,_seg) \
-+ (_start + \
-+ ((_req) * BLKIF_MAX_SEGMENTS_PER_REQUEST * PAGE_SIZE) + \
-+ ((_seg) * PAGE_SIZE))
-+static int blkif_reqs = MAX_PENDING_REQS;
-+static int mmap_pages = MMAP_PAGES;
++static inline void falcon_handle_tx_event(struct efx_channel *channel,
++ efx_qword_t *event)
++{
++ unsigned int tx_ev_desc_ptr;
++ unsigned int tx_ev_q_label;
++ struct efx_tx_queue *tx_queue;
++ struct efx_nic *efx = channel->efx;
+
-+#define RING_PAGES 1 /* BLKTAP - immediately before the mmap area, we
-+ * have a bunch of pages reserved for shared
-+ * memory rings.
-+ */
++ if (likely(EFX_QWORD_FIELD(*event, TX_EV_COMP))) {
++ /* Transmit completion */
++ tx_ev_desc_ptr = EFX_QWORD_FIELD(*event, TX_EV_DESC_PTR);
++ tx_ev_q_label = EFX_QWORD_FIELD(*event, TX_EV_Q_LABEL);
++ tx_queue = &efx->tx_queue[tx_ev_q_label];
++ efx_xmit_done(tx_queue, tx_ev_desc_ptr);
++ } else if (EFX_QWORD_FIELD(*event, TX_EV_WQ_FF_FULL)) {
++ /* Rewrite the FIFO write pointer */
++ tx_ev_q_label = EFX_QWORD_FIELD(*event, TX_EV_Q_LABEL);
++ tx_queue = &efx->tx_queue[tx_ev_q_label];
+
-+/*Data struct handed back to userspace for tapdisk device to VBD mapping*/
-+typedef struct domid_translate {
-+ unsigned short domid;
-+ unsigned short busid;
-+} domid_translate_t ;
++ if (efx->net_dev_registered)
++ netif_tx_lock(efx->net_dev);
++ falcon_notify_tx_desc(tx_queue);
++ if (efx->net_dev_registered)
++ netif_tx_unlock(efx->net_dev);
++ } else if (EFX_QWORD_FIELD(*event, TX_EV_PKT_ERR) &&
++ EFX_WORKAROUND_10727(efx)) {
++ efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
++ } else {
++ EFX_ERR(efx, "channel %d unexpected TX event "
++ EFX_QWORD_FMT"\n", channel->channel,
++ EFX_QWORD_VAL(*event));
++ }
++}
+
-+/*Data struct associated with each of the tapdisk devices*/
-+typedef struct tap_blkif {
-+ struct vm_area_struct *vma; /*Shared memory area */
-+ unsigned long rings_vstart; /*Kernel memory mapping */
-+ unsigned long user_vstart; /*User memory mapping */
-+ unsigned long dev_inuse; /*One process opens device at a time. */
-+ unsigned long dev_pending; /*In process of being opened */
-+ unsigned long ring_ok; /*make this ring->state */
-+ blkif_front_ring_t ufe_ring; /*Rings up to user space. */
-+ wait_queue_head_t wait; /*for poll */
-+ unsigned long mode; /*current switching mode */
-+ int minor; /*Minor number for tapdisk device */
-+ pid_t pid; /*tapdisk process id */
-+ enum { RUNNING, CLEANSHUTDOWN } status; /*Detect a clean userspace
-+ shutdown */
-+ unsigned long *idx_map; /*Record the user ring id to kern
-+ [req id, idx] tuple */
-+ blkif_t *blkif; /*Associate blkif with tapdev */
-+ struct domid_translate trans; /*Translation from domid to bus. */
-+} tap_blkif_t;
++/* Check received packet's destination MAC address. */
++static int check_dest_mac(struct efx_rx_queue *rx_queue,
++ const efx_qword_t *event)
++{
++ struct efx_rx_buffer *rx_buf;
++ struct efx_nic *efx = rx_queue->efx;
++ int rx_ev_desc_ptr;
++ struct ethhdr *eh;
+
-+static struct tap_blkif *tapfds[MAX_TAP_DEV];
-+static int blktap_next_minor;
++ if (efx->promiscuous)
++ return 1;
+
-+module_param(blkif_reqs, int, 0);
-+/* Run-time switchable: /sys/module/blktap/parameters/ */
-+static unsigned int log_stats = 0;
-+static unsigned int debug_lvl = 0;
-+module_param(log_stats, int, 0644);
-+module_param(debug_lvl, int, 0644);
++ rx_ev_desc_ptr = EFX_QWORD_FIELD(*event, RX_EV_DESC_PTR);
++ rx_buf = efx_rx_buffer(rx_queue, rx_ev_desc_ptr);
++ eh = (struct ethhdr *)rx_buf->data;
++ if (memcmp(eh->h_dest, efx->net_dev->dev_addr, ETH_ALEN))
++ return 0;
++ return 1;
++}
+
-+/*
-+ * Each outstanding request that we've passed to the lower device layers has a
-+ * 'pending_req' allocated to it. Each buffer_head that completes decrements
-+ * the pendcnt towards zero. When it hits zero, the specified domain has a
-+ * response queued for it, with the saved 'id' passed back.
-+ */
-+typedef struct {
-+ blkif_t *blkif;
-+ u64 id;
-+ unsigned short mem_idx;
-+ int nr_pages;
-+ atomic_t pendcnt;
-+ unsigned short operation;
-+ int status;
-+ struct list_head free_list;
-+ int inuse;
-+} pending_req_t;
++/* Detect errors included in the rx_evt_pkt_ok bit. */
++static void falcon_handle_rx_not_ok(struct efx_rx_queue *rx_queue,
++ const efx_qword_t *event,
++ unsigned *rx_ev_pkt_ok,
++ int *discard, int byte_count)
++{
++ struct efx_nic *efx = rx_queue->efx;
++ unsigned rx_ev_buf_owner_id_err, rx_ev_ip_hdr_chksum_err;
++ unsigned rx_ev_tcp_udp_chksum_err, rx_ev_eth_crc_err;
++ unsigned rx_ev_frm_trunc, rx_ev_drib_nib, rx_ev_tobe_disc;
++ unsigned rx_ev_pkt_type, rx_ev_other_err, rx_ev_pause_frm;
++ unsigned rx_ev_ip_frag_err, rx_ev_hdr_type, rx_ev_mcast_pkt;
++ int snap, non_ip;
++
++ rx_ev_hdr_type = EFX_QWORD_FIELD(*event, RX_EV_HDR_TYPE);
++ rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, RX_EV_MCAST_PKT);
++ rx_ev_tobe_disc = EFX_QWORD_FIELD(*event, RX_EV_TOBE_DISC);
++ rx_ev_pkt_type = EFX_QWORD_FIELD(*event, RX_EV_PKT_TYPE);
++ rx_ev_buf_owner_id_err = EFX_QWORD_FIELD(*event,
++ RX_EV_BUF_OWNER_ID_ERR);
++ rx_ev_ip_frag_err = EFX_QWORD_FIELD(*event, RX_EV_IF_FRAG_ERR);
++ rx_ev_ip_hdr_chksum_err = EFX_QWORD_FIELD(*event,
++ RX_EV_IP_HDR_CHKSUM_ERR);
++ rx_ev_tcp_udp_chksum_err = EFX_QWORD_FIELD(*event,
++ RX_EV_TCP_UDP_CHKSUM_ERR);
++ rx_ev_eth_crc_err = EFX_QWORD_FIELD(*event, RX_EV_ETH_CRC_ERR);
++ rx_ev_frm_trunc = EFX_QWORD_FIELD(*event, RX_EV_FRM_TRUNC);
++ rx_ev_drib_nib = ((FALCON_REV(efx) >= FALCON_REV_B0) ?
++ 0 : EFX_QWORD_FIELD(*event, RX_EV_DRIB_NIB));
++ rx_ev_pause_frm = EFX_QWORD_FIELD(*event, RX_EV_PAUSE_FRM_ERR);
++
++ /* Every error apart from tobe_disc and pause_frm */
++ rx_ev_other_err = (rx_ev_drib_nib | rx_ev_tcp_udp_chksum_err |
++ rx_ev_buf_owner_id_err | rx_ev_eth_crc_err |
++ rx_ev_frm_trunc | rx_ev_ip_hdr_chksum_err);
++
++ snap = (rx_ev_pkt_type == RX_EV_PKT_TYPE_LLC_DECODE) ||
++ (rx_ev_pkt_type == RX_EV_PKT_TYPE_VLAN_LLC_DECODE);
++ non_ip = (rx_ev_hdr_type == RX_EV_HDR_TYPE_NON_IP_DECODE);
++
++ /* SFC bug 5475/8970: The Falcon XMAC incorrectly calculates the
++ * length field of an LLC frame, which sets TOBE_DISC. We could set
++ * PASS_LEN_ERR, but we want the MAC to filter out short frames (to
++ * protect the RX block).
++ *
++ * bug5475 - LLC/SNAP: Falcon identifies SNAP packets.
++ * bug8970 - LLC/noSNAP: Falcon does not provide an LLC flag.
++ * LLC can't encapsulate IP, so by definition
++ * these packets are NON_IP.
++ *
++ * Unicast mismatch will also cause TOBE_DISC, so the driver needs
++ * to check this.
++ */
++ if (EFX_WORKAROUND_5475(efx) && rx_ev_tobe_disc && (snap || non_ip)) {
++ /* If all the other flags are zero then we can state the
++ * entire packet is ok, which will flag to the kernel not
++ * to recalculate checksums.
++ */
++ if (!(non_ip | rx_ev_other_err | rx_ev_pause_frm))
++ *rx_ev_pkt_ok = 1;
+
-+static pending_req_t *pending_reqs[MAX_PENDING_REQS];
-+static struct list_head pending_free;
-+static DEFINE_SPINLOCK(pending_free_lock);
-+static DECLARE_WAIT_QUEUE_HEAD (pending_free_wq);
-+static int alloc_pending_reqs;
++ rx_ev_tobe_disc = 0;
+
-+typedef unsigned int PEND_RING_IDX;
++ /* TOBE_DISC is set for unicast mismatch. But given that
++ * we can't trust TOBE_DISC here, we must validate the dest
++ * MAC address ourselves.
++ */
++ if (!rx_ev_mcast_pkt && !check_dest_mac(rx_queue, event))
++ rx_ev_tobe_disc = 1;
++ }
++
++ /* Count errors that are not in MAC stats. */
++ if (rx_ev_frm_trunc)
++ ++rx_queue->channel->n_rx_frm_trunc;
++ else if (rx_ev_tobe_disc)
++ ++rx_queue->channel->n_rx_tobe_disc;
++ else if (rx_ev_ip_hdr_chksum_err)
++ ++rx_queue->channel->n_rx_ip_hdr_chksum_err;
++ else if (rx_ev_tcp_udp_chksum_err)
++ ++rx_queue->channel->n_rx_tcp_udp_chksum_err;
++ if (rx_ev_ip_frag_err)
++ ++rx_queue->channel->n_rx_ip_frag_err;
++
++ /* The frame must be discarded if any of these are true. */
++ *discard = (rx_ev_eth_crc_err | rx_ev_frm_trunc | rx_ev_drib_nib |
++ rx_ev_tobe_disc | rx_ev_pause_frm);
++
++ /* TOBE_DISC is expected on unicast mismatches; don't print out an
++ * error message. FRM_TRUNC indicates RXDP dropped the packet due
++ * to a FIFO overflow.
++ */
++#ifdef EFX_ENABLE_DEBUG
++ if (rx_ev_other_err) {
++ EFX_INFO_RL(efx, " RX queue %d unexpected RX event "
++ EFX_QWORD_FMT "%s%s%s%s%s%s%s%s%s\n",
++ rx_queue->queue, EFX_QWORD_VAL(*event),
++ rx_ev_buf_owner_id_err ? " [OWNER_ID_ERR]" : "",
++ rx_ev_ip_hdr_chksum_err ?
++ " [IP_HDR_CHKSUM_ERR]" : "",
++ rx_ev_tcp_udp_chksum_err ?
++ " [TCP_UDP_CHKSUM_ERR]" : "",
++ rx_ev_eth_crc_err ? " [ETH_CRC_ERR]" : "",
++ rx_ev_frm_trunc ? " [FRM_TRUNC]" : "",
++ rx_ev_drib_nib ? " [DRIB_NIB]" : "",
++ rx_ev_tobe_disc ? " [TOBE_DISC]" : "",
++ rx_ev_pause_frm ? " [PAUSE]" : "",
++ snap ? " [SNAP/LLC]" : "");
++ }
++#endif
++
++ if (unlikely(rx_ev_eth_crc_err && EFX_WORKAROUND_10750(efx) &&
++ efx->phy_type == PHY_TYPE_10XPRESS))
++ tenxpress_crc_err(efx);
++}
++
++
++/* Handle receive events that are not in-order. */
++static void falcon_handle_rx_bad_index(struct efx_rx_queue *rx_queue,
++ unsigned index)
++{
++ struct efx_nic *efx = rx_queue->efx;
++ unsigned expected, dropped;
++
++ expected = rx_queue->removed_count & FALCON_RXD_RING_MASK;
++ dropped = ((index + FALCON_RXD_RING_SIZE - expected) &
++ FALCON_RXD_RING_MASK);
++ EFX_INFO(efx, "dropped %d events (index=%d expected=%d)\n",
++ dropped, index, expected);
++
++ atomic_inc(&efx->errors.missing_event);
++ efx_schedule_reset(efx, EFX_WORKAROUND_5676(efx) ?
++ RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
++}
++
++
++/* Handle a packet received event
++ *
++ * Falcon silicon gives a "discard" flag if it's a unicast packet with the
++ * wrong destination address
++ * Also "is multicast" and "matches multicast filter" flags can be used to
++ * discard non-matching multicast packets.
++ */
++static inline int falcon_handle_rx_event(struct efx_channel *channel,
++ const efx_qword_t *event)
++{
++ unsigned int rx_ev_q_label, rx_ev_desc_ptr, rx_ev_byte_cnt;
++ unsigned int rx_ev_pkt_ok, rx_ev_hdr_type, rx_ev_mcast_pkt;
++ unsigned expected_ptr;
++ int discard = 0, checksummed;
++ struct efx_rx_queue *rx_queue;
++ struct efx_nic *efx = channel->efx;
++
++ /* Basic packet information */
++ rx_ev_byte_cnt = EFX_QWORD_FIELD(*event, RX_EV_BYTE_CNT);
++ rx_ev_pkt_ok = EFX_QWORD_FIELD(*event, RX_EV_PKT_OK);
++ rx_ev_hdr_type = EFX_QWORD_FIELD(*event, RX_EV_HDR_TYPE);
++ WARN_ON(EFX_QWORD_FIELD(*event, RX_EV_JUMBO_CONT));
++ WARN_ON(EFX_QWORD_FIELD(*event, RX_EV_SOP) != 1);
++
++ rx_ev_q_label = EFX_QWORD_FIELD(*event, RX_EV_Q_LABEL);
++ rx_queue = &efx->rx_queue[rx_ev_q_label];
++
++ rx_ev_desc_ptr = EFX_QWORD_FIELD(*event, RX_EV_DESC_PTR);
++ expected_ptr = rx_queue->removed_count & FALCON_RXD_RING_MASK;
++ if (unlikely(rx_ev_desc_ptr != expected_ptr)) {
++ falcon_handle_rx_bad_index(rx_queue, rx_ev_desc_ptr);
++ return rx_ev_q_label;
++ }
++
++ if (likely(rx_ev_pkt_ok)) {
++ /* If packet is marked as OK and packet type is TCP/IPv4 or
++ * UDP/IPv4, then we can rely on the hardware checksum.
++ */
++ checksummed = RX_EV_HDR_TYPE_HAS_CHECKSUMS(rx_ev_hdr_type);
++ } else {
++ falcon_handle_rx_not_ok(rx_queue, event, &rx_ev_pkt_ok,
++ &discard, rx_ev_byte_cnt);
++ checksummed = 0;
++ }
+
-+static inline int MASK_PEND_IDX(int i) {
-+ return (i & (MAX_PENDING_REQS-1));
-+}
++ /* Detect multicast packets that didn't match the filter */
++ rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, RX_EV_MCAST_PKT);
++ if (rx_ev_mcast_pkt) {
++ unsigned int rx_ev_mcast_hash_match =
++ EFX_QWORD_FIELD(*event, RX_EV_MCAST_HASH_MATCH);
+
-+static inline unsigned int RTN_PEND_IDX(pending_req_t *req, int idx) {
-+ return (req - pending_reqs[idx]);
-+}
++ if (unlikely(!rx_ev_mcast_hash_match))
++ discard = 1;
++ }
+
-+#define NR_PENDING_REQS (MAX_PENDING_REQS - pending_prod + pending_cons)
++ /* Handle received packet */
++ efx_rx_packet(rx_queue, rx_ev_desc_ptr, rx_ev_byte_cnt,
++ checksummed, discard);
+
-+#define BLKBACK_INVALID_HANDLE (~0)
++ return rx_ev_q_label;
++}
+
-+static struct page **foreign_pages[MAX_DYNAMIC_MEM];
-+static inline unsigned long idx_to_kaddr(
-+ unsigned int mmap_idx, unsigned int req_idx, unsigned int sg_idx)
++/* Global events are basically PHY events */
++static void falcon_handle_global_event(struct efx_channel *channel,
++ efx_qword_t *event)
+{
-+ unsigned int arr_idx = req_idx*BLKIF_MAX_SEGMENTS_PER_REQUEST + sg_idx;
-+ unsigned long pfn = page_to_pfn(foreign_pages[mmap_idx][arr_idx]);
-+ return (unsigned long)pfn_to_kaddr(pfn);
++ struct efx_nic *efx = channel->efx;
++ int is_phy_event = 0, handled = 0;
++
++ /* Check for interrupt on either port. Some boards have a
++ * single PHY wired to the interrupt line for port 1. */
++ if (EFX_QWORD_FIELD(*event, G_PHY0_INTR) ||
++ EFX_QWORD_FIELD(*event, G_PHY1_INTR) ||
++ EFX_QWORD_FIELD(*event, XG_PHY_INTR))
++ is_phy_event = 1;
++
++ if ((FALCON_REV(efx) >= FALCON_REV_B0) &&
++ EFX_OWORD_FIELD(*event, XG_MNT_INTR_B0))
++ is_phy_event = 1;
++
++ if (is_phy_event) {
++ efx->phy_op->clear_interrupt(efx);
++ queue_work(efx->workqueue, &efx->reconfigure_work);
++ handled = 1;
++ }
++
++ if (EFX_QWORD_FIELD_VER(efx, *event, RX_RECOVERY)) {
++ EFX_ERR(efx, "channel %d seen global RX_RESET "
++ "event. Resetting.\n", channel->channel);
++
++ atomic_inc(&efx->errors.rx_reset);
++ efx_schedule_reset(efx, EFX_WORKAROUND_6555(efx) ?
++ RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
++ handled = 1;
++ }
++
++ if (!handled)
++ EFX_ERR(efx, "channel %d unknown global event "
++ EFX_QWORD_FMT "\n", channel->channel,
++ EFX_QWORD_VAL(*event));
+}
+
-+static unsigned short mmap_alloc = 0;
-+static unsigned short mmap_lock = 0;
-+static unsigned short mmap_inuse = 0;
++static void falcon_handle_driver_event(struct efx_channel *channel,
++ efx_qword_t *event)
++{
++ struct efx_nic *efx = channel->efx;
++ unsigned int ev_sub_code;
++ unsigned int ev_sub_data;
+
-+/******************************************************************
-+ * GRANT HANDLES
-+ */
++ ev_sub_code = EFX_QWORD_FIELD(*event, DRIVER_EV_SUB_CODE);
++ ev_sub_data = EFX_QWORD_FIELD(*event, DRIVER_EV_SUB_DATA);
+
-+/* When using grant tables to map a frame for device access then the
-+ * handle returned must be used to unmap the frame. This is needed to
-+ * drop the ref count on the frame.
-+ */
-+struct grant_handle_pair
++ switch (ev_sub_code) {
++ case TX_DESCQ_FLS_DONE_EV_DECODE:
++ EFX_TRACE(efx, "channel %d TXQ %d flushed\n",
++ channel->channel, ev_sub_data);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ case RX_DESCQ_FLS_DONE_EV_DECODE:
++ EFX_TRACE(efx, "channel %d RXQ %d flushed\n",
++ channel->channel, ev_sub_data);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ case EVQ_INIT_DONE_EV_DECODE:
++ EFX_LOG(efx, "channel %d EVQ %d initialised\n",
++ channel->channel, ev_sub_data);
++ break;
++ case SRM_UPD_DONE_EV_DECODE:
++ EFX_TRACE(efx, "channel %d SRAM update done\n",
++ channel->channel);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ case WAKE_UP_EV_DECODE:
++ EFX_TRACE(efx, "channel %d RXQ %d wakeup event\n",
++ channel->channel, ev_sub_data);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ case TIMER_EV_DECODE:
++ EFX_TRACE(efx, "channel %d RX queue %d timer expired\n",
++ channel->channel, ev_sub_data);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ case RX_RECOVERY_EV_DECODE:
++ EFX_ERR(efx, "channel %d seen DRIVER RX_RESET event. "
++ "Resetting.\n", channel->channel);
++
++ atomic_inc(&efx->errors.rx_reset);
++ efx_schedule_reset(efx,
++ EFX_WORKAROUND_6555(efx) ?
++ RESET_TYPE_RX_RECOVERY :
++ RESET_TYPE_DISABLE);
++ break;
++ case RX_DSC_ERROR_EV_DECODE:
++ EFX_ERR(efx, "RX DMA Q %d reports descriptor fetch error."
++ " RX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
++ atomic_inc(&efx->errors.rx_desc_fetch);
++ efx_schedule_reset(efx, RESET_TYPE_RX_DESC_FETCH);
++ break;
++ case TX_DSC_ERROR_EV_DECODE:
++ EFX_ERR(efx, "TX DMA Q %d reports descriptor fetch error."
++ " TX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
++ atomic_inc(&efx->errors.tx_desc_fetch);
++ efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
++ break;
++ default:
++ EFX_TRACE(efx, "channel %d unknown driver event code %d "
++ "data %04x\n", channel->channel, ev_sub_code,
++ ev_sub_data);
++ EFX_DL_CALLBACK(efx, event, event);
++ break;
++ }
++}
++
++#if defined(EFX_USE_FASTCALL)
++int fastcall falcon_process_eventq(struct efx_channel *channel, int *rx_quota)
++#else
++int falcon_process_eventq(struct efx_channel *channel, int *rx_quota)
++#endif
+{
-+ grant_handle_t kernel;
-+ grant_handle_t user;
-+};
-+#define INVALID_GRANT_HANDLE 0xFFFF
++ unsigned int read_ptr;
++ efx_qword_t event, *p_event;
++ int ev_code;
++ int rxq;
++ int rxdmaqs = 0;
+
-+static struct grant_handle_pair
-+ pending_grant_handles[MAX_DYNAMIC_MEM][MMAP_PAGES];
-+#define pending_handle(_id, _idx, _i) \
-+ (pending_grant_handles[_id][((_idx) * BLKIF_MAX_SEGMENTS_PER_REQUEST) \
-+ + (_i)])
++ read_ptr = channel->eventq_read_ptr;
+
++ do {
++ p_event = falcon_event(channel, read_ptr);
++ event = *p_event;
+
-+static int blktap_read_ufe_ring(tap_blkif_t *info); /*local prototypes*/
++ if (!falcon_event_present(&event))
++ /* End of events */
++ break;
+
-+#define BLKTAP_MINOR 0 /*/dev/xen/blktap has a dynamic major */
-+#define BLKTAP_DEV_DIR "/dev/xen"
++ EFX_TRACE(channel->efx, "channel %d event is "EFX_QWORD_FMT"\n",
++ channel->channel, EFX_QWORD_VAL(event));
+
-+static int blktap_major;
++ /* Clear this event by marking it all ones */
++ EFX_SET_QWORD(*p_event);
+
-+/* blktap IOCTLs: */
-+#define BLKTAP_IOCTL_KICK_FE 1
-+#define BLKTAP_IOCTL_KICK_BE 2 /* currently unused */
-+#define BLKTAP_IOCTL_SETMODE 3
-+#define BLKTAP_IOCTL_SENDPID 4
-+#define BLKTAP_IOCTL_NEWINTF 5
-+#define BLKTAP_IOCTL_MINOR 6
-+#define BLKTAP_IOCTL_MAJOR 7
-+#define BLKTAP_QUERY_ALLOC_REQS 8
-+#define BLKTAP_IOCTL_FREEINTF 9
-+#define BLKTAP_IOCTL_PRINT_IDXS 100
++ ev_code = EFX_QWORD_FIELD(event, EV_CODE);
+
-+/* blktap switching modes: (Set with BLKTAP_IOCTL_SETMODE) */
-+#define BLKTAP_MODE_PASSTHROUGH 0x00000000 /* default */
-+#define BLKTAP_MODE_INTERCEPT_FE 0x00000001
-+#define BLKTAP_MODE_INTERCEPT_BE 0x00000002 /* unimp. */
++ switch (ev_code) {
++ case RX_IP_EV_DECODE:
++ rxq = falcon_handle_rx_event(channel, &event);
++ rxdmaqs |= (1 << rxq);
++ (*rx_quota)--;
++ break;
++ case TX_IP_EV_DECODE:
++ falcon_handle_tx_event(channel, &event);
++ break;
++ case DRV_GEN_EV_DECODE:
++ channel->eventq_magic
++ = EFX_QWORD_FIELD(event, EVQ_MAGIC);
++ EFX_LOG(channel->efx, "channel %d received generated "
++ "event "EFX_QWORD_FMT"\n", channel->channel,
++ EFX_QWORD_VAL(event));
++ break;
++ case GLOBAL_EV_DECODE:
++ falcon_handle_global_event(channel, &event);
++ break;
++ case DRIVER_EV_DECODE:
++ falcon_handle_driver_event(channel, &event);
++ break;
++ default:
++ EFX_ERR(channel->efx, "channel %d unknown event type %d"
++ " (data " EFX_QWORD_FMT ")\n", channel->channel,
++ ev_code, EFX_QWORD_VAL(event));
++ }
+
-+#define BLKTAP_MODE_INTERPOSE \
-+ (BLKTAP_MODE_INTERCEPT_FE | BLKTAP_MODE_INTERCEPT_BE)
++ /* Increment read pointer */
++ read_ptr = (read_ptr + 1) & FALCON_EVQ_MASK;
+
++ } while (*rx_quota);
+
-+static inline int BLKTAP_MODE_VALID(unsigned long arg)
++ channel->eventq_read_ptr = read_ptr;
++ return rxdmaqs;
++}
++
++void falcon_set_int_moderation(struct efx_channel *channel)
+{
-+ return ((arg == BLKTAP_MODE_PASSTHROUGH ) ||
-+ (arg == BLKTAP_MODE_INTERCEPT_FE) ||
-+ (arg == BLKTAP_MODE_INTERPOSE ));
++ efx_dword_t timer_cmd;
++ struct efx_nic *efx = channel->efx;
++
++ /* Set timer register */
++ if (channel->irq_moderation) {
++ /* Round to resolution supported by hardware. The value we
++ * program is based at 0. So actual interrupt moderation
++ * achieved is ((x + 1) * res).
++ */
++ unsigned int res = 5;
++ channel->irq_moderation -= (channel->irq_moderation % res);
++ if (channel->irq_moderation < res)
++ channel->irq_moderation = res;
++ EFX_POPULATE_DWORD_2(timer_cmd,
++ TIMER_MODE, TIMER_MODE_INT_HLDOFF,
++ TIMER_VAL,
++ (channel->irq_moderation / res) - 1);
++ } else {
++ EFX_POPULATE_DWORD_2(timer_cmd,
++ TIMER_MODE, TIMER_MODE_DIS,
++ TIMER_VAL, 0);
++ }
++ falcon_writel_page_locked(efx, &timer_cmd, TIMER_CMD_REG_KER,
++ channel->evqnum);
++
+}
+
-+/* Requests passing through the tap to userspace are re-assigned an ID.
-+ * We must record a mapping between the BE [IDX,ID] tuple and the userspace
-+ * ring ID.
-+ */
++/* Allocate buffer table entries for event queue */
++int falcon_probe_eventq(struct efx_channel *channel)
++{
++ struct efx_nic *efx = channel->efx;
++ struct falcon_nic_data *nic_data = efx->nic_data;
++ unsigned int evq_size;
++ int rc;
+
-+static inline unsigned long MAKE_ID(domid_t fe_dom, PEND_RING_IDX idx)
++ evq_size = FALCON_EVQ_SIZE * sizeof(efx_qword_t);
++ rc = falcon_alloc_special_buffer(efx, &channel->eventq, evq_size);
++ if (rc)
++ return rc;
++
++ nic_data->resources.evq_int_min = max(nic_data->resources.evq_int_min,
++ (unsigned)channel->evqnum + 1);
++
++ return 0;
++}
++
++int falcon_init_eventq(struct efx_channel *channel)
+{
-+ return ((fe_dom << 16) | MASK_PEND_IDX(idx));
++ efx_oword_t evq_ptr;
++ struct efx_nic *efx = channel->efx;
++ int rc;
++
++ EFX_LOG(efx, "channel %d event queue in special buffers %d-%d\n",
++ channel->channel, channel->eventq.index,
++ channel->eventq.index + channel->eventq.entries - 1);
++
++ /* Pin event queue buffer */
++ rc = falcon_init_special_buffer(efx, &channel->eventq);
++ if (rc)
++ return rc;
++
++ /* Fill event queue with all ones (i.e. empty events) */
++ memset(channel->eventq.addr, 0xff, channel->eventq.len);
++
++ /* Push event queue to card */
++ EFX_POPULATE_OWORD_3(evq_ptr,
++ EVQ_EN, 1,
++ EVQ_SIZE, FALCON_EVQ_ORDER,
++ EVQ_BUF_BASE_ID, channel->eventq.index);
++ falcon_write_table(efx, &evq_ptr, efx->type->evq_ptr_tbl_base,
++ channel->evqnum);
++
++ falcon_set_int_moderation(channel);
++
++ return 0;
+}
+
-+extern inline PEND_RING_IDX ID_TO_IDX(unsigned long id)
++void falcon_fini_eventq(struct efx_channel *channel)
+{
-+ return (PEND_RING_IDX)(id & 0x0000ffff);
++ efx_oword_t eventq_ptr;
++ struct efx_nic *efx = channel->efx;
++
++ /* Remove event queue from card */
++ EFX_ZERO_OWORD(eventq_ptr);
++ falcon_write_table(efx, &eventq_ptr, efx->type->evq_ptr_tbl_base,
++ channel->evqnum);
++
++ /* Unpin event queue */
++ falcon_fini_special_buffer(efx, &channel->eventq);
+}
+
-+extern inline int ID_TO_MIDX(unsigned long id)
++/* Free buffers backing event queue */
++void falcon_remove_eventq(struct efx_channel *channel)
+{
-+ return (int)(id >> 16);
++ falcon_free_special_buffer(channel->efx, &channel->eventq);
+}
+
-+#define INVALID_REQ 0xdead0000
+
-+/*TODO: Convert to a free list*/
-+static inline int GET_NEXT_REQ(unsigned long *idx_map)
++/* Generates a test event on the event queue. A subsequent call to
++ * process_eventq() should pick up the event and place the value of
++ * "magic" into channel->eventq_magic;
++ */
++void falcon_generate_test_event(struct efx_channel *channel, unsigned int magic)
+{
-+ int i;
-+ for (i = 0; i < MAX_PENDING_REQS; i++)
-+ if (idx_map[i] == INVALID_REQ)
-+ return i;
++ efx_qword_t test_event;
+
-+ return INVALID_REQ;
++ EFX_POPULATE_QWORD_2(test_event,
++ EV_CODE, DRV_GEN_EV_DECODE,
++ EVQ_MAGIC, magic);
++ falcon_generate_event(channel, &test_event);
+}
+
-+static inline int OFFSET_TO_USR_IDX(int offset)
++
++/**************************************************************************
++ *
++ * Falcon hardware interrupts
++ * The hardware interrupt handler does very little work; all the event
++ * queue processing is carried out by per-channel tasklets.
++ *
++ **************************************************************************/
++
++/* Enable/disable/generate Falcon interrupts */
++static inline void falcon_interrupts(struct efx_nic *efx, int enabled,
++ int force)
+{
-+ return offset / BLKIF_MAX_SEGMENTS_PER_REQUEST;
++ efx_oword_t int_en_reg_ker;
++
++ EFX_POPULATE_OWORD_2(int_en_reg_ker,
++ KER_INT_KER, force,
++ DRV_INT_EN_KER, enabled);
++ falcon_write(efx, &int_en_reg_ker, INT_EN_REG_KER);
+}
+
-+static inline int OFFSET_TO_SEG(int offset)
++void falcon_enable_interrupts(struct efx_nic *efx)
+{
-+ return offset % BLKIF_MAX_SEGMENTS_PER_REQUEST;
-+}
++ efx_oword_t int_adr_reg_ker;
++ struct efx_channel *channel;
+
++ /* Zero INT_KER */
++ EFX_ZERO_OWORD(*((efx_oword_t *) efx->irq_status.addr));
++ wmb(); /* Ensure interrupt vector is clear before interrupts enabled */
+
-+#define BLKTAP_INVALID_HANDLE(_g) \
-+ (((_g->kernel) == INVALID_GRANT_HANDLE) && \
-+ ((_g->user) == INVALID_GRANT_HANDLE))
++ /* Program INT_ADR_KER_REG */
++ EFX_POPULATE_OWORD_2(int_adr_reg_ker,
++ NORM_INT_VEC_DIS_KER, EFX_INT_MODE_USE_MSI(efx),
++ INT_ADR_KER, efx->irq_status.dma_addr);
++ falcon_write(efx, &int_adr_reg_ker, INT_ADR_REG_KER);
+
-+#define BLKTAP_INVALIDATE_HANDLE(_g) do { \
-+ (_g)->kernel = INVALID_GRANT_HANDLE; (_g)->user = INVALID_GRANT_HANDLE; \
-+ } while(0)
++ /* Enable interrupts */
++ falcon_interrupts(efx, 1, 0);
++
++ /* Force processing of all the channels to get the EVQ RPTRs up to
++ date */
++ efx_for_each_channel_with_interrupt(channel, efx)
++ efx_schedule_channel(channel);
++}
+
++void falcon_disable_interrupts(struct efx_nic *efx)
++{
++ /* Disable interrupts */
++ falcon_interrupts(efx, 0, 0);
++}
+
-+/******************************************************************
-+ * BLKTAP VM OPS
++/* Generate a Falcon test interrupt
++ * Interrupt must already have been enabled, otherwise nasty things
++ * may happen.
+ */
++void falcon_generate_interrupt(struct efx_nic *efx)
++{
++ falcon_interrupts(efx, 1, 1);
++}
+
-+static struct page *blktap_nopage(struct vm_area_struct *vma,
-+ unsigned long address,
-+ int *type)
++/* Acknowledge a legacy interrupt from Falcon
++ *
++ * This acknowledges a legacy (not MSI) interrupt via INT_ACK_KER_REG.
++ *
++ * Due to SFC bug 3706 (silicon revision <=A1) reads can be duplicated in the
++ * BIU. Interrupt acknowledge is read sensitive so must write instead
++ * (then read to ensure the BIU collector is flushed)
++ *
++ * NB most hardware supports MSI interrupts
++ */
++static inline void falcon_irq_ack_a1(struct efx_nic *efx)
+{
-+ /*
-+ * if the page has not been mapped in by the driver then return
-+ * NOPAGE_SIGBUS to the domain.
-+ */
++ efx_dword_t reg;
+
-+ return NOPAGE_SIGBUS;
++ EFX_POPULATE_DWORD_1(reg, INT_ACK_DUMMY_DATA, 0xb7eb7e);
++ falcon_writel(efx, ®, INT_ACK_REG_KER_A1);
++ falcon_readl(efx, ®, WORK_AROUND_BROKEN_PCI_READS_REG_KER_A1);
+}
+
-+static pte_t blktap_clear_pte(struct vm_area_struct *vma,
-+ unsigned long uvaddr,
-+ pte_t *ptep, int is_fullmm)
++/* Process a fatal interrupt
++ * Disable bus mastering ASAP and schedule a reset
++ */
++static irqreturn_t falcon_fatal_interrupt(struct efx_nic *efx)
+{
-+ pte_t copy;
-+ tap_blkif_t *info;
-+ int offset, seg, usr_idx, pending_idx, mmap_idx;
-+ unsigned long uvstart = vma->vm_start + (RING_PAGES << PAGE_SHIFT);
-+ unsigned long kvaddr;
-+ struct page **map;
-+ struct page *pg;
-+ struct grant_handle_pair *khandle;
-+ struct gnttab_unmap_grant_ref unmap[2];
-+ int count = 0;
++ efx_oword_t *int_ker = (efx_oword_t *) efx->irq_status.addr;
++ efx_oword_t fatal_intr;
++ int error, mem_perr;
++ static int n_int_errors;
+
-+ /*
-+ * If the address is before the start of the grant mapped region or
-+ * if vm_file is NULL (meaning mmap failed and we have nothing to do)
-+ */
-+ if (uvaddr < uvstart || vma->vm_file == NULL)
-+ return ptep_get_and_clear_full(vma->vm_mm, uvaddr,
-+ ptep, is_fullmm);
++ falcon_read(efx, &fatal_intr, FATAL_INTR_REG_KER);
++ error = EFX_OWORD_FIELD(fatal_intr, INT_KER_ERROR);
+
-+ info = vma->vm_file->private_data;
-+ map = vma->vm_private_data;
++ EFX_ERR(efx, "SYSTEM ERROR " EFX_OWORD_FMT " status "
++ EFX_OWORD_FMT ": %s\n", EFX_OWORD_VAL(*int_ker),
++ EFX_OWORD_VAL(fatal_intr),
++ error ? "disabling bus mastering" : "no recognised error");
++ if (error == 0)
++ goto out;
+
-+ /* TODO Should these be changed to if statements? */
-+ BUG_ON(!info);
-+ BUG_ON(!info->idx_map);
-+ BUG_ON(!map);
++ /* If this is a memory parity error dump which blocks are offending */
++ mem_perr = EFX_OWORD_FIELD(fatal_intr, MEM_PERR_INT_KER);
++ if (mem_perr) {
++ efx_oword_t reg;
++ falcon_read(efx, ®, MEM_STAT_REG_KER);
++ EFX_ERR(efx, "SYSTEM ERROR: memory parity error "
++ EFX_OWORD_FMT "\n", EFX_OWORD_VAL(reg));
++ }
+
-+ offset = (int) ((uvaddr - uvstart) >> PAGE_SHIFT);
-+ usr_idx = OFFSET_TO_USR_IDX(offset);
-+ seg = OFFSET_TO_SEG(offset);
++ /* Disable DMA bus mastering on both devices */
++ pci_disable_device(efx->pci_dev);
++ if (efx->type->is_dual_func)
++ pci_disable_device(efx->pci_dev2);
+
-+ pending_idx = MASK_PEND_IDX(ID_TO_IDX(info->idx_map[usr_idx]));
-+ mmap_idx = ID_TO_MIDX(info->idx_map[usr_idx]);
++ if (++n_int_errors < FALCON_MAX_INT_ERRORS) {
++ EFX_ERR(efx, "SYSTEM ERROR - reset scheduled\n");
++ efx_schedule_reset(efx, RESET_TYPE_INT_ERROR);
++ } else {
++ EFX_ERR(efx, "SYSTEM ERROR - max number of errors seen."
++ "NIC will be disabled\n");
++ efx_schedule_reset(efx, RESET_TYPE_DISABLE);
++ }
++out:
++ return IRQ_HANDLED;
++}
+
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, seg);
-+ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
-+ ClearPageReserved(pg);
-+ map[offset + RING_PAGES] = NULL;
++/* Handle a legacy interrupt from Falcon
++ * Acknowledges the interrupt and schedule event queue processing.
++ *
++ * This routine must guarantee not to touch the hardware when
++ * interrupts are disabled, to allow for correct semantics of
++ * efx_suspend() and efx_resume().
++ */
++#if !defined(EFX_HAVE_IRQ_HANDLER_REGS)
++static irqreturn_t falcon_legacy_interrupt_b0(int irq, void *dev_id)
++#else
++static irqreturn_t falcon_legacy_interrupt_b0(int irq, void *dev_id,
++ struct pt_regs *regs
++ __attribute__ ((unused)))
++#endif
++{
++ struct efx_nic *efx = (struct efx_nic *)dev_id;
++ efx_oword_t *int_ker = (efx_oword_t *) efx->irq_status.addr;
++ struct efx_channel *channel;
++ efx_dword_t reg;
++ u32 queues;
++ int syserr;
+
-+ khandle = &pending_handle(mmap_idx, pending_idx, seg);
++ /* Read the ISR which also ACKs the interrupts */
++ falcon_readl(efx, ®, INT_ISR0_B0);
++ queues = EFX_EXTRACT_DWORD(reg, 0, 31);
+
-+ if (khandle->kernel != INVALID_GRANT_HANDLE) {
-+ gnttab_set_unmap_op(&unmap[count], kvaddr,
-+ GNTMAP_host_map, khandle->kernel);
-+ count++;
++ /* Check to see if we have a serious error condition */
++ syserr = EFX_OWORD_FIELD(*int_ker, FATAL_INT);
++ if (unlikely(syserr))
++ return falcon_fatal_interrupt(efx);
+
-+ set_phys_to_machine(__pa(kvaddr) >> PAGE_SHIFT,
-+ INVALID_P2M_ENTRY);
++ if (queues == 0)
++ return IRQ_NONE;
++
++ efx->last_irq_cpu = raw_smp_processor_id();
++ EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_DWORD_FMT "\n",
++ irq, raw_smp_processor_id(), EFX_DWORD_VAL(reg));
++
++ /* Schedule processing of any interrupting queues */
++ channel = &efx->channel[0];
++ while (queues) {
++ if (queues & 0x01)
++ efx_schedule_channel(channel);
++ channel++;
++ queues >>= 1;
+ }
+
-+ if (khandle->user != INVALID_GRANT_HANDLE) {
-+ BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
++ return IRQ_HANDLED;
++}
+
-+ copy = *ptep;
-+ gnttab_set_unmap_op(&unmap[count], virt_to_machine(ptep),
-+ GNTMAP_host_map
-+ | GNTMAP_application_map
-+ | GNTMAP_contains_pte,
-+ khandle->user);
-+ count++;
-+ } else {
-+ BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap));
+
-+ /* USING SHADOW PAGE TABLES. */
-+ copy = ptep_get_and_clear_full(vma->vm_mm, uvaddr, ptep,
-+ is_fullmm);
++#if !defined(EFX_HAVE_IRQ_HANDLER_REGS)
++static irqreturn_t falcon_legacy_interrupt_a1(int irq, void *dev_id)
++#else
++static irqreturn_t falcon_legacy_interrupt_a1(int irq, void *dev_id,
++ struct pt_regs *regs
++ __attribute__ ((unused)))
++#endif
++{
++ struct efx_nic *efx = (struct efx_nic *)dev_id;
++ efx_oword_t *int_ker = (efx_oword_t *) efx->irq_status.addr;
++ struct efx_channel *channel;
++ int syserr;
++ int queues;
++
++ /* Check to see if this is our interrupt. If it isn't, we
++ * exit without having touched the hardware.
++ */
++ if (unlikely(EFX_OWORD_IS_ZERO(*int_ker))) {
++ EFX_TRACE(efx, "IRQ %d on CPU %d not for me\n", irq,
++ raw_smp_processor_id());
++ return IRQ_NONE;
+ }
++ efx->last_irq_cpu = raw_smp_processor_id();
++ EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_OWORD_FMT "\n",
++ irq, raw_smp_processor_id(), EFX_OWORD_VAL(*int_ker));
+
-+ if (count) {
-+ BLKTAP_INVALIDATE_HANDLE(khandle);
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref,
-+ unmap, count))
-+ BUG();
++ /* Check to see if we have a serious error condition */
++ syserr = EFX_OWORD_FIELD(*int_ker, FATAL_INT);
++ if (unlikely(syserr))
++ return falcon_fatal_interrupt(efx);
++
++ /* Determine interrupting queues, clear interrupt status
++ * register and acknowledge the device interrupt.
++ */
++ BUILD_BUG_ON(INT_EVQS_WIDTH > EFX_MAX_CHANNELS);
++ queues = EFX_OWORD_FIELD(*int_ker, INT_EVQS);
++ EFX_ZERO_OWORD(*int_ker);
++ wmb(); /* Ensure the vector is cleared before interrupt ack */
++ falcon_irq_ack_a1(efx);
++
++ /* Schedule processing of any interrupting queues */
++ channel = &efx->channel[0];
++ while (queues) {
++ if (queues & 0x01)
++ efx_schedule_channel(channel);
++ channel++;
++ queues >>= 1;
+ }
+
-+ return copy;
++ return IRQ_HANDLED;
+}
+
-+struct vm_operations_struct blktap_vm_ops = {
-+ nopage: blktap_nopage,
-+ zap_pte: blktap_clear_pte,
-+};
-+
-+/******************************************************************
-+ * BLKTAP FILE OPS
++/* Handle an MSI interrupt from Falcon
++ *
++ * Handle an MSI hardware interrupt. This routine schedules event
++ * queue processing. No interrupt acknowledgement cycle is necessary.
++ * Also, we never need to check that the interrupt is for us, since
++ * MSI interrupts cannot be shared.
++ *
++ * This routine must guarantee not to touch the hardware when
++ * interrupts are disabled, to allow for correct semantics of
++ * efx_suspend() and efx_resume().
+ */
-+
-+/*Function Declarations*/
-+static tap_blkif_t *get_next_free_dev(void);
-+static int blktap_open(struct inode *inode, struct file *filp);
-+static int blktap_release(struct inode *inode, struct file *filp);
-+static int blktap_mmap(struct file *filp, struct vm_area_struct *vma);
-+static int blktap_ioctl(struct inode *inode, struct file *filp,
-+ unsigned int cmd, unsigned long arg);
-+static unsigned int blktap_poll(struct file *file, poll_table *wait);
++#if !defined(EFX_HAVE_IRQ_HANDLER_REGS)
++static irqreturn_t falcon_msi_interrupt(int irq, void *dev_id)
++#else
++static irqreturn_t falcon_msi_interrupt(int irq, void *dev_id,
++ struct pt_regs *regs
++ __attribute__ ((unused)))
++#endif
++{
++ struct efx_channel *channel = (struct efx_channel *)dev_id;
++ struct efx_nic *efx = channel->efx;
++ efx_oword_t *int_ker = (efx_oword_t *) efx->irq_status.addr;
++ int syserr;
+
-+static const struct file_operations blktap_fops = {
-+ .owner = THIS_MODULE,
-+ .poll = blktap_poll,
-+ .ioctl = blktap_ioctl,
-+ .open = blktap_open,
-+ .release = blktap_release,
-+ .mmap = blktap_mmap,
-+};
++ efx->last_irq_cpu = raw_smp_processor_id();
++ EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_OWORD_FMT "\n",
++ irq, raw_smp_processor_id(), EFX_OWORD_VAL(*int_ker));
+
++ /* Check to see if we have a serious error condition */
++ syserr = EFX_OWORD_FIELD(*int_ker, FATAL_INT);
++ if (unlikely(syserr))
++ return falcon_fatal_interrupt(efx);
+
-+static tap_blkif_t *get_next_free_dev(void)
-+{
-+ struct class *class;
-+ tap_blkif_t *info;
-+ int minor;
++ /* Schedule processing of the channel */
++ efx_schedule_channel(channel);
+
-+ /*
-+ * This is called only from the ioctl, which
-+ * means we should always have interrupts enabled.
-+ */
-+ BUG_ON(irqs_disabled());
++ return IRQ_HANDLED;
++}
+
-+ spin_lock_irq(&pending_free_lock);
+
-+ /* tapfds[0] is always NULL */
++/* Setup RSS indirection table.
++ * This maps from the hash value of the packet to RXQ
++ */
++static void falcon_setup_rss_indir_table(struct efx_nic *efx)
++{
++ int i = 0;
++ unsigned long offset;
++ unsigned long flags __attribute__ ((unused));
++ efx_dword_t dword;
+
-+ for (minor = 1; minor < blktap_next_minor; minor++) {
-+ info = tapfds[minor];
-+ /* we could have failed a previous attempt. */
-+ if (!info ||
-+ ((info->dev_inuse == 0) &&
-+ (info->dev_pending == 0)) ) {
-+ info->dev_pending = 1;
-+ goto found;
-+ }
++ if (FALCON_REV(efx) < FALCON_REV_B0)
++ return;
++
++ for (offset = RX_RSS_INDIR_TBL_B0;
++ offset < RX_RSS_INDIR_TBL_B0 + 0x800;
++ offset += 0x10) {
++ EFX_POPULATE_DWORD_1(dword, RX_RSS_INDIR_ENT_B0,
++ i % efx->rss_queues);
++ falcon_writel(efx, &dword, offset);
++ i++;
+ }
-+ info = NULL;
-+ minor = -1;
++}
+
-+ /*
-+ * We didn't find free device. If we can still allocate
-+ * more, then we grab the next device minor that is
-+ * available. This is done while we are still under
-+ * the protection of the pending_free_lock.
-+ */
-+ if (blktap_next_minor < MAX_TAP_DEV)
-+ minor = blktap_next_minor++;
-+found:
-+ spin_unlock_irq(&pending_free_lock);
++/* Hook interrupt handler(s)
++ * Try MSI and then legacy interrupts.
++ */
++int falcon_init_interrupt(struct efx_nic *efx)
++{
++ struct efx_channel *channel;
++ int rc;
+
-+ if (!info && minor > 0) {
-+ info = kzalloc(sizeof(*info), GFP_KERNEL);
-+ if (unlikely(!info)) {
-+ /*
-+ * If we failed here, try to put back
-+ * the next minor number. But if one
-+ * was just taken, then we just lose this
-+ * minor. We can try to allocate this
-+ * minor again later.
-+ */
-+ spin_lock_irq(&pending_free_lock);
-+ if (blktap_next_minor == minor+1)
-+ blktap_next_minor--;
-+ spin_unlock_irq(&pending_free_lock);
-+ goto out;
-+ }
++ if (!EFX_INT_MODE_USE_MSI(efx)) {
++ irq_handler_t handler;
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ handler = falcon_legacy_interrupt_b0;
++ else
++ handler = falcon_legacy_interrupt_a1;
+
-+ info->minor = minor;
-+ /*
-+ * Make sure that we have a minor before others can
-+ * see us.
-+ */
-+ wmb();
-+ tapfds[minor] = info;
++ rc = request_irq(efx->legacy_irq, handler, IRQF_SHARED,
++ efx->name, efx);
++ if (rc) {
++ EFX_ERR(efx, "failed to hook legacy IRQ %d\n",
++ efx->pci_dev->irq);
++ goto fail1;
++ }
++ return 0;
++ }
+
-+ if ((class = get_xen_class()) != NULL)
-+ class_device_create(class, NULL,
-+ MKDEV(blktap_major, minor), NULL,
-+ "blktap%d", minor);
++ /* Hook MSI or MSI-X interrupt */
++ efx_for_each_channel_with_interrupt(channel, efx) {
++ rc = request_irq(channel->irq, falcon_msi_interrupt,
++ IRQF_PROBE_SHARED, /* Not shared */
++ efx->name, channel);
++ if (rc) {
++ EFX_ERR(efx, "failed to hook IRQ %d\n", channel->irq);
++ goto fail2;
++ }
+ }
+
-+out:
-+ return info;
++ return 0;
++
++ fail2:
++ efx_for_each_channel_with_interrupt(channel, efx)
++ free_irq(channel->irq, channel);
++ fail1:
++ return rc;
+}
+
-+int dom_to_devid(domid_t domid, int xenbus_id, blkif_t *blkif)
++void falcon_fini_interrupt(struct efx_nic *efx)
+{
-+ tap_blkif_t *info;
-+ int i;
++ struct efx_channel *channel;
++ efx_oword_t reg;
+
-+ for (i = 1; i < blktap_next_minor; i++) {
-+ info = tapfds[i];
-+ if ( info &&
-+ (info->trans.domid == domid) &&
-+ (info->trans.busid == xenbus_id) ) {
-+ info->blkif = blkif;
-+ info->status = RUNNING;
-+ return i;
-+ }
-+ }
-+ return -1;
-+}
++ /* Disable MSI/MSI-X interrupts */
++ efx_for_each_channel_with_interrupt(channel, efx)
++ if (channel->irq)
++ free_irq(channel->irq, channel);
+
-+void signal_tapdisk(int idx)
-+{
-+ tap_blkif_t *info;
-+ struct task_struct *ptask;
++ /* ACK legacy interrupt */
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ falcon_read(efx, ®, INT_ISR0_B0);
++ else
++ falcon_irq_ack_a1(efx);
+
-+ info = tapfds[idx];
-+ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info)
-+ return;
++ /* Disable legacy interrupt */
++ if (efx->legacy_irq)
++ free_irq(efx->legacy_irq, efx);
++}
+
-+ if (info->pid > 0) {
-+ ptask = find_task_by_pid(info->pid);
-+ if (ptask)
-+ info->status = CLEANSHUTDOWN;
-+ }
-+ info->blkif = NULL;
++/**************************************************************************
++ *
++ * EEPROM/flash
++ *
++ **************************************************************************
++ */
+
-+ return;
++/* Wait for SPI command completion */
++static int falcon_spi_wait(struct efx_nic *efx)
++{
++ efx_oword_t reg;
++ int cmd_en, timer_active;
++ int count;
++
++ count = 0;
++ do {
++ falcon_read(efx, ®, EE_SPI_HCMD_REG_KER);
++ cmd_en = EFX_OWORD_FIELD(reg, EE_SPI_HCMD_CMD_EN);
++ timer_active = EFX_OWORD_FIELD(reg, EE_WR_TIMER_ACTIVE);
++ if (!cmd_en && !timer_active)
++ return 0;
++ udelay(10);
++ } while (++count < 10000); /* wait upto 100msec */
++ EFX_ERR(efx, "timed out waiting for SPI\n");
++ return -ETIMEDOUT;
+}
+
-+static int blktap_open(struct inode *inode, struct file *filp)
++static int
++falcon_spi_read(const struct efx_spi_device *spi, struct efx_nic *efx,
++ unsigned int command, int address, void *data, unsigned int len)
+{
-+ blkif_sring_t *sring;
-+ int idx = iminor(inode) - BLKTAP_MINOR;
-+ tap_blkif_t *info;
-+ int i;
-+
-+ /* ctrl device, treat differently */
-+ if (!idx)
-+ return 0;
++ int addressed = (address >= 0);
++ efx_oword_t reg;
++ int rc;
+
-+ info = tapfds[idx];
++ /* Input validation */
++ if (len > FALCON_SPI_MAX_LEN)
++ return -EINVAL;
+
-+ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info) {
-+ WPRINTK("Unable to open device /dev/xen/blktap%d\n",
-+ idx);
-+ return -ENODEV;
-+ }
++ /* Acquire SPI lock */
++ mutex_lock(&efx->spi_lock);
+
-+ DPRINTK("Opening device /dev/xen/blktap%d\n",idx);
-+
-+ /*Only one process can access device at a time*/
-+ if (test_and_set_bit(0, &info->dev_inuse))
-+ return -EBUSY;
++ /* Check SPI not currently being accessed */
++ rc = falcon_spi_wait(efx);
++ if (rc)
++ goto out;
+
-+ info->dev_pending = 0;
-+
-+ /* Allocate the fe ring. */
-+ sring = (blkif_sring_t *)get_zeroed_page(GFP_KERNEL);
-+ if (sring == NULL)
-+ goto fail_nomem;
++ /* Program address register, if we have an address */
++ if (addressed) {
++ EFX_POPULATE_OWORD_1(reg, EE_SPI_HADR_ADR, address);
++ falcon_write(efx, ®, EE_SPI_HADR_REG_KER);
++ }
++
++ /* Issue read command */
++ EFX_POPULATE_OWORD_7(reg,
++ EE_SPI_HCMD_CMD_EN, 1,
++ EE_SPI_HCMD_SF_SEL, spi->device_id,
++ EE_SPI_HCMD_DABCNT, len,
++ EE_SPI_HCMD_READ, EE_SPI_READ,
++ EE_SPI_HCMD_DUBCNT, 0,
++ EE_SPI_HCMD_ADBCNT,
++ (addressed ? spi->addr_len : 0),
++ EE_SPI_HCMD_ENC, command);
++ falcon_write(efx, ®, EE_SPI_HCMD_REG_KER);
++
++ /* Wait for read to complete */
++ rc = falcon_spi_wait(efx);
++ if (rc)
++ goto out;
+
-+ SetPageReserved(virt_to_page(sring));
-+
-+ SHARED_RING_INIT(sring);
-+ FRONT_RING_INIT(&info->ufe_ring, sring, PAGE_SIZE);
-+
-+ filp->private_data = info;
-+ info->vma = NULL;
++ /* Read data */
++ falcon_read(efx, ®, EE_SPI_HDATA_REG_KER);
++ memcpy(data, ®, len);
+
-+ info->idx_map = kmalloc(sizeof(unsigned long) * MAX_PENDING_REQS,
-+ GFP_KERNEL);
-+
-+ if (info->idx_map == NULL)
-+ goto fail_nomem;
++ out:
++ /* Release SPI lock */
++ mutex_unlock(&efx->spi_lock);
+
-+ if (idx > 0) {
-+ init_waitqueue_head(&info->wait);
-+ for (i = 0; i < MAX_PENDING_REQS; i++)
-+ info->idx_map[i] = INVALID_REQ;
++ return rc;
++}
++
++static int
++falcon_spi_write(const struct efx_spi_device *spi, struct efx_nic *efx,
++ unsigned int command, int address, const void *data,
++ unsigned int len)
++{
++ int addressed = (address >= 0);
++ efx_oword_t reg;
++ int rc;
++
++ /* Input validation */
++ if (len > (addressed ? efx_spi_write_limit(spi, address)
++ : FALCON_SPI_MAX_LEN))
++ return -EINVAL;
++
++ /* Acquire SPI lock */
++ mutex_lock(&efx->spi_lock);
++
++ /* Check SPI not currently being accessed */
++ rc = falcon_spi_wait(efx);
++ if (rc)
++ goto out;
++
++ /* Program address register, if we have an address */
++ if (addressed) {
++ EFX_POPULATE_OWORD_1(reg, EE_SPI_HADR_ADR, address);
++ falcon_write(efx, ®, EE_SPI_HADR_REG_KER);
+ }
+
-+ DPRINTK("Tap open: device /dev/xen/blktap%d\n",idx);
-+ return 0;
++ /* Program data register, if we have data */
++ if (data) {
++ memcpy(®, data, len);
++ falcon_write(efx, ®, EE_SPI_HDATA_REG_KER);
++ }
++
++ /* Issue write command */
++ EFX_POPULATE_OWORD_7(reg,
++ EE_SPI_HCMD_CMD_EN, 1,
++ EE_SPI_HCMD_SF_SEL, spi->device_id,
++ EE_SPI_HCMD_DABCNT, len,
++ EE_SPI_HCMD_READ, EE_SPI_WRITE,
++ EE_SPI_HCMD_DUBCNT, 0,
++ EE_SPI_HCMD_ADBCNT,
++ (addressed ? spi->addr_len : 0),
++ EE_SPI_HCMD_ENC, command);
++ falcon_write(efx, ®, EE_SPI_HCMD_REG_KER);
++
++ /* Wait for write to complete */
++ rc = falcon_spi_wait(efx);
++ if (rc)
++ goto out;
+
-+ fail_nomem:
-+ return -ENOMEM;
++ out:
++ /* Release SPI lock */
++ mutex_unlock(&efx->spi_lock);
++
++ return rc;
+}
+
-+static int blktap_release(struct inode *inode, struct file *filp)
++/**************************************************************************
++ *
++ * MAC wrapper
++ *
++ **************************************************************************
++ */
++void falcon_drain_tx_fifo(struct efx_nic *efx)
+{
-+ tap_blkif_t *info = filp->private_data;
-+
-+ /* check for control device */
-+ if (!info)
-+ return 0;
++ efx_oword_t temp;
++ efx_oword_t mcast_reg0;
++ efx_oword_t mcast_reg1;
++ int count;
+
-+ info->dev_inuse = 0;
-+ DPRINTK("Freeing device [/dev/xen/blktap%d]\n",info->minor);
++ if (FALCON_REV(efx) < FALCON_REV_B0)
++ return;
+
-+ /* Free the ring page. */
-+ ClearPageReserved(virt_to_page(info->ufe_ring.sring));
-+ free_page((unsigned long) info->ufe_ring.sring);
++ falcon_read(efx, &temp, MAC0_CTRL_REG_KER);
++ /* There is no point in draining more than once */
++ if (EFX_OWORD_FIELD(temp, TXFIFO_DRAIN_EN_B0))
++ return;
+
-+ /* Clear any active mappings and free foreign map table */
-+ if (info->vma) {
-+ zap_page_range(
-+ info->vma, info->vma->vm_start,
-+ info->vma->vm_end - info->vma->vm_start, NULL);
++ /* MAC stats will fail whilst the TX fifo is draining. Serialise
++ * the drain sequence with the statistics fetch */
++ spin_lock(&efx->stats_lock);
+
-+ kfree(info->vma->vm_private_data);
++ EFX_SET_OWORD_FIELD(temp, TXFIFO_DRAIN_EN_B0, 1);
++ falcon_write(efx, &temp, MAC0_CTRL_REG_KER);
+
-+ info->vma = NULL;
-+ }
++ falcon_read(efx, &mcast_reg0, MAC_MCAST_HASH_REG0_KER);
++ falcon_read(efx, &mcast_reg1, MAC_MCAST_HASH_REG1_KER);
+
-+ if (info->idx_map) {
-+ kfree(info->idx_map);
-+ info->idx_map = NULL;
-+ }
++ /* Reset the MAC and EM block. */
++ falcon_read(efx, &temp, GLB_CTL_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, RST_XGTX, 1);
++ EFX_SET_OWORD_FIELD(temp, RST_XGRX, 1);
++ EFX_SET_OWORD_FIELD(temp, RST_EM, 1);
++ falcon_write(efx, &temp, GLB_CTL_REG_KER);
+
-+ if ( (info->status != CLEANSHUTDOWN) && (info->blkif != NULL) ) {
-+ if (info->blkif->xenblkd != NULL) {
-+ kthread_stop(info->blkif->xenblkd);
-+ info->blkif->xenblkd = NULL;
++ count = 0;
++ while (1) {
++ falcon_read(efx, &temp, GLB_CTL_REG_KER);
++ if (!EFX_OWORD_FIELD(temp, RST_XGTX) &&
++ !EFX_OWORD_FIELD(temp, RST_XGRX) &&
++ !EFX_OWORD_FIELD(temp, RST_EM)) {
++ EFX_LOG(efx, "Completed MAC reset after %d loops\n",
++ count);
++ break;
+ }
-+ info->status = CLEANSHUTDOWN;
++ if (count > 20) {
++ EFX_ERR(efx, "MAC reset failed\n");
++ break;
++ }
++ count++;
++ udelay(10);
+ }
+
-+ return 0;
++ spin_unlock(&efx->stats_lock);
++
++ /* Restore the multicast hash registers. */
++ falcon_write(efx, &mcast_reg0, MAC_MCAST_HASH_REG0_KER);
++ falcon_write(efx, &mcast_reg1, MAC_MCAST_HASH_REG1_KER);
++
++ /* If we've reset the EM block and the link is up, then
++ * we'll have to kick the XAUI link so the PHY can recover */
++ if (efx->link_up && EFX_IS10G(efx) && EFX_WORKAROUND_5147(efx))
++ falcon_reset_xaui(efx);
+}
+
++void falcon_deconfigure_mac_wrapper(struct efx_nic *efx)
++{
++ struct falcon_nic_data *nic_data = efx->nic_data;
++ efx_oword_t temp;
++ int changing_loopback;
+
-+/* Note on mmap:
-+ * We need to map pages to user space in a way that will allow the block
-+ * subsystem set up direct IO to them. This couldn't be done before, because
-+ * there isn't really a sane way to translate a user virtual address down to a
-+ * physical address when the page belongs to another domain.
-+ *
-+ * My first approach was to map the page in to kernel memory, add an entry
-+ * for it in the physical frame list (using alloc_lomem_region as in blkback)
-+ * and then attempt to map that page up to user space. This is disallowed
-+ * by xen though, which realizes that we don't really own the machine frame
-+ * underlying the physical page.
-+ *
-+ * The new approach is to provide explicit support for this in xen linux.
-+ * The VMA now has a flag, VM_FOREIGN, to indicate that it contains pages
-+ * mapped from other vms. vma->vm_private_data is set up as a mapping
-+ * from pages to actual page structs. There is a new clause in get_user_pages
-+ * that does the right thing for this sort of mapping.
-+ */
-+static int blktap_mmap(struct file *filp, struct vm_area_struct *vma)
++ if (FALCON_REV(efx) < FALCON_REV_B0)
++ return;
++
++ /* Isolate the MAC -> RX */
++ falcon_read(efx, &temp, RX_CFG_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, RX_INGR_EN_B0, 0);
++ falcon_write(efx, &temp, RX_CFG_REG_KER);
++
++ /* Synchronise the EM block against any loopback mode changes by
++ * draining the TX fifo and resetting. */
++ changing_loopback = (efx->loopback_mode != nic_data->old_loopback_mode);
++ nic_data->old_loopback_mode = efx->loopback_mode;
++ if (changing_loopback || !efx->link_up)
++ falcon_drain_tx_fifo(efx);
++}
++
++void falcon_reconfigure_mac_wrapper(struct efx_nic *efx)
+{
-+ int size;
-+ struct page **map;
++ efx_oword_t reg;
++ int link_speed;
++ unsigned int tx_fc;
++
++ if (efx->link_options & GM_LPA_10000)
++ link_speed = 0x3;
++ else if (efx->link_options & GM_LPA_1000)
++ link_speed = 0x2;
++ else if (efx->link_options & GM_LPA_100)
++ link_speed = 0x1;
++ else
++ link_speed = 0x0;
++ /* MAC_LINK_STATUS controls MAC backpressure but doesn't work
++ * as advertised. Disable to ensure packets are not
++ * indefinitely held and TX queue can be flushed at any point
++ * while the link is down.
++ */
++ EFX_POPULATE_OWORD_5(reg,
++ MAC_XOFF_VAL, 0xffff /* max pause time */,
++ MAC_BCAD_ACPT, 1,
++ MAC_UC_PROM, efx->promiscuous,
++ MAC_LINK_STATUS, 1, /* always set */
++ MAC_SPEED, link_speed);
++ /* On B0, MAC backpressure can be disabled and packets get
++ * discarded. */
++ if (FALCON_REV(efx) >= FALCON_REV_B0) {
++ EFX_SET_OWORD_FIELD(reg, TXFIFO_DRAIN_EN_B0,
++ !efx->link_up);
++ }
++
++ falcon_write(efx, ®, MAC0_CTRL_REG_KER);
++
++ /*
++ * Transmission of pause frames when RX crosses the threshold is
++ * covered by RX_XOFF_MAC_EN and XM_TX_CFG_REG:XM_FCNTL.
++ *
++ * Action on receipt of pause frames is controller by XM_DIS_FCNTL
++ */
++ tx_fc = (efx->flow_control & EFX_FC_TX) ? 1 : 0;
++ falcon_read(efx, ®, RX_CFG_REG_KER);
++ EFX_SET_OWORD_FIELD_VER(efx, reg, RX_XOFF_MAC_EN, tx_fc);
++
++ /* Unisolate the MAC -> RX */
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ EFX_SET_OWORD_FIELD(reg, RX_INGR_EN_B0, 1);
++ falcon_write(efx, ®, RX_CFG_REG_KER);
++}
++
++int falcon_dma_stats(struct efx_nic *efx, unsigned int done_offset)
++{
++ efx_oword_t reg;
++ u32 *dma_done;
+ int i;
-+ tap_blkif_t *info = filp->private_data;
-+ int ret;
+
-+ if (info == NULL) {
-+ WPRINTK("blktap: mmap, retrieving idx failed\n");
-+ return -ENOMEM;
-+ }
-+
-+ vma->vm_flags |= VM_RESERVED;
-+ vma->vm_ops = &blktap_vm_ops;
++ if (disable_dma_stats)
++ return 0;
+
-+ size = vma->vm_end - vma->vm_start;
-+ if (size != ((mmap_pages + RING_PAGES) << PAGE_SHIFT)) {
-+ WPRINTK("you _must_ map exactly %d pages!\n",
-+ mmap_pages + RING_PAGES);
-+ return -EAGAIN;
++ /* Statistics fetch will fail if the MAC is in TX drain */
++ if (FALCON_REV(efx) >= FALCON_REV_B0) {
++ efx_oword_t temp;
++ falcon_read(efx, &temp, MAC0_CTRL_REG_KER);
++ if (EFX_OWORD_FIELD(temp, TXFIFO_DRAIN_EN_B0))
++ return 0;
+ }
+
-+ size >>= PAGE_SHIFT;
-+ info->rings_vstart = vma->vm_start;
-+ info->user_vstart = info->rings_vstart + (RING_PAGES << PAGE_SHIFT);
-+
-+ /* Map the ring pages to the start of the region and reserve it. */
-+ if (xen_feature(XENFEAT_auto_translated_physmap))
-+ ret = vm_insert_page(vma, vma->vm_start,
-+ virt_to_page(info->ufe_ring.sring));
-+ else
-+ ret = remap_pfn_range(vma, vma->vm_start,
-+ __pa(info->ufe_ring.sring) >> PAGE_SHIFT,
-+ PAGE_SIZE, vma->vm_page_prot);
-+ if (ret) {
-+ WPRINTK("Mapping user ring failed!\n");
-+ goto fail;
++ /* Clear completion pointer */
++ dma_done = (efx->stats_buffer.addr + done_offset);
++ *dma_done = FALCON_STATS_NOT_DONE;
++ wmb(); /* ensure done flag is clear */
++
++ /* Initiate DMA transfer of stats */
++ EFX_POPULATE_OWORD_2(reg,
++ MAC_STAT_DMA_CMD, 1,
++ MAC_STAT_DMA_ADR,
++ efx->stats_buffer.dma_addr);
++ falcon_write(efx, ®, MAC0_STAT_DMA_REG_KER);
++
++ /* Wait for transfer to complete */
++ for (i = 0; i < 400; i++) {
++ if (*(volatile u32 *)dma_done == FALCON_STATS_DONE)
++ return 0;
++ udelay(10);
+ }
+
-+ /* Mark this VM as containing foreign pages, and set up mappings. */
-+ map = kzalloc(((vma->vm_end - vma->vm_start) >> PAGE_SHIFT)
-+ * sizeof(struct page_struct*),
-+ GFP_KERNEL);
-+ if (map == NULL) {
-+ WPRINTK("Couldn't alloc VM_FOREIGN map.\n");
-+ goto fail;
++ if (EFX_WORKAROUND_8419(efx)) {
++ disable_dma_stats = 1;
++ EFX_INFO(efx, "MAC stats DMA disabled\n");
++ } else {
++ EFX_ERR(efx, "timed out waiting for statistics\n");
+ }
+
-+ for (i = 0; i < ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT); i++)
-+ map[i] = NULL;
-+
-+ vma->vm_private_data = map;
-+ vma->vm_flags |= VM_FOREIGN;
-+ vma->vm_flags |= VM_DONTCOPY;
++ return -ETIMEDOUT;
++}
+
-+#ifdef CONFIG_X86
-+ vma->vm_mm->context.has_foreign_mappings = 1;
-+#endif
++/**************************************************************************
++ *
++ * PHY access via GMII
++ *
++ **************************************************************************
++ */
+
-+ info->vma = vma;
-+ info->ring_ok = 1;
-+ return 0;
-+ fail:
-+ /* Clear any active mappings. */
-+ zap_page_range(vma, vma->vm_start,
-+ vma->vm_end - vma->vm_start, NULL);
++/* Use the top bit of the MII PHY id to indicate the PHY type
++ * (1G/10G), with the remaining bits as the actual PHY id.
++ *
++ * This allows us to avoid leaking information from the mii_if_info
++ * structure into other data structures.
++ */
++#define FALCON_PHY_ID_ID_WIDTH EFX_WIDTH(MD_PRT_DEV_ADR)
++#define FALCON_PHY_ID_ID_MASK ((1 << FALCON_PHY_ID_ID_WIDTH) - 1)
++#define FALCON_PHY_ID_WIDTH (FALCON_PHY_ID_ID_WIDTH + 1)
++#define FALCON_PHY_ID_MASK ((1 << FALCON_PHY_ID_WIDTH) - 1)
++#define FALCON_PHY_ID_10G (1 << (FALCON_PHY_ID_WIDTH - 1))
+
-+ return -ENOMEM;
-+}
+
++/* Packing the clause 45 port and device fields into a single value */
++#define MD_PRT_ADR_COMP_LBN (MD_PRT_ADR_LBN - MD_DEV_ADR_LBN)
++#define MD_PRT_ADR_COMP_WIDTH MD_PRT_ADR_WIDTH
++#define MD_DEV_ADR_COMP_LBN 0
++#define MD_DEV_ADR_COMP_WIDTH MD_DEV_ADR_WIDTH
+
-+static int blktap_ioctl(struct inode *inode, struct file *filp,
-+ unsigned int cmd, unsigned long arg)
++
++/* Wait for GMII access to complete */
++static int falcon_gmii_wait(struct efx_nic *efx)
+{
-+ tap_blkif_t *info = filp->private_data;
++ efx_dword_t md_stat;
++ int count;
+
-+ switch(cmd) {
-+ case BLKTAP_IOCTL_KICK_FE:
-+ {
-+ /* There are fe messages to process. */
-+ return blktap_read_ufe_ring(info);
-+ }
-+ case BLKTAP_IOCTL_SETMODE:
-+ {
-+ if (info) {
-+ if (BLKTAP_MODE_VALID(arg)) {
-+ info->mode = arg;
-+ /* XXX: may need to flush rings here. */
-+ DPRINTK("blktap: set mode to %lx\n",
-+ arg);
-+ return 0;
++ for (count = 0; count < 1000; count++) { /* wait upto 10ms */
++ falcon_readl(efx, &md_stat, MD_STAT_REG_KER);
++ if (EFX_DWORD_FIELD(md_stat, MD_BSY) == 0) {
++ if (EFX_DWORD_FIELD(md_stat, MD_LNFL) != 0 ||
++ EFX_DWORD_FIELD(md_stat, MD_BSERR) != 0) {
++ EFX_ERR(efx, "error from GMII access "
++ EFX_DWORD_FMT"\n",
++ EFX_DWORD_VAL(md_stat));
++ return -EIO;
+ }
++ return 0;
+ }
-+ return 0;
-+ }
-+ case BLKTAP_IOCTL_PRINT_IDXS:
-+ {
-+ if (info) {
-+ printk("User Rings: \n-----------\n");
-+ printk("UF: rsp_cons: %2d, req_prod_prv: %2d "
-+ "| req_prod: %2d, rsp_prod: %2d\n",
-+ info->ufe_ring.rsp_cons,
-+ info->ufe_ring.req_prod_pvt,
-+ info->ufe_ring.sring->req_prod,
-+ info->ufe_ring.sring->rsp_prod);
-+ }
-+ return 0;
-+ }
-+ case BLKTAP_IOCTL_SENDPID:
-+ {
-+ if (info) {
-+ info->pid = (pid_t)arg;
-+ DPRINTK("blktap: pid received %d\n",
-+ info->pid);
-+ }
-+ return 0;
++ udelay(10);
+ }
-+ case BLKTAP_IOCTL_NEWINTF:
-+ {
-+ uint64_t val = (uint64_t)arg;
-+ domid_translate_t *tr = (domid_translate_t *)&val;
++ EFX_ERR(efx, "timed out waiting for GMII\n");
++ return -ETIMEDOUT;
++}
+
-+ DPRINTK("NEWINTF Req for domid %d and bus id %d\n",
-+ tr->domid, tr->busid);
-+ info = get_next_free_dev();
-+ if (!info) {
-+ WPRINTK("Error initialising /dev/xen/blktap - "
-+ "No more devices\n");
-+ return -1;
-+ }
-+ info->trans.domid = tr->domid;
-+ info->trans.busid = tr->busid;
-+ return info->minor;
++/* Writes a GMII register of a PHY connected to Falcon using MDIO. */
++static void falcon_mdio_write(struct net_device *net_dev, int phy_id,
++ int addr, int value)
++{
++ struct efx_nic *efx = (struct efx_nic *)net_dev->priv;
++ unsigned int phy_id2 = phy_id & FALCON_PHY_ID_ID_MASK;
++ unsigned int phy_10g = phy_id & FALCON_PHY_ID_10G;
++ efx_oword_t reg;
++
++ /* The 'generic' prt/dev packing in mdio_10g.h is conveniently
++ * chosen so that the only current user, Falcon, can take the
++ * packed value and use them directly.
++ * Fail to build if this assumption is broken.
++ */
++ BUILD_BUG_ON(FALCON_PHY_ID_10G != MDIO45_XPRT_ID_IS10G);
++ BUILD_BUG_ON(FALCON_PHY_ID_ID_WIDTH != MDIO45_PRT_DEV_WIDTH);
++ BUILD_BUG_ON(MD_PRT_ADR_COMP_LBN != MDIO45_PRT_ID_COMP_LBN);
++ BUILD_BUG_ON(MD_DEV_ADR_COMP_LBN != MDIO45_DEV_ID_COMP_LBN);
++
++ if (phy_id2 == PHY_ADDR_INVALID)
++ return;
++
++ /* See falcon_mdio_read for an explanation. */
++ if (EFX_ISCLAUSE45(efx) && !phy_10g) {
++ int mmd = ffs(efx->phy_op->mmds) - 1;
++ EFX_TRACE(efx, "Fixing erroneous clause22 write\n");
++ phy_id2 = mdio_clause45_pack(phy_id2, mmd)
++ & FALCON_PHY_ID_ID_MASK;
++ phy_10g = 1;
+ }
-+ case BLKTAP_IOCTL_FREEINTF:
-+ {
-+ unsigned long dev = arg;
-+ unsigned long flags;
+
-+ info = tapfds[dev];
++ EFX_REGDUMP(efx, "writing GMII %d register %02x with %04x\n", phy_id,
++ addr, value);
+
-+ if ((dev > MAX_TAP_DEV) || !info)
-+ return 0; /* should this be an error? */
++ /* Obtain PHY lock */
++ spin_lock_bh(&efx->phy_lock);
+
-+ spin_lock_irqsave(&pending_free_lock, flags);
-+ if (info->dev_pending)
-+ info->dev_pending = 0;
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
++ /* Check MII not currently being accessed */
++ if (falcon_gmii_wait(efx) != 0)
++ goto out;
+
-+ return 0;
++ /* Write the address/ID register */
++ EFX_POPULATE_OWORD_1(reg, MD_PHY_ADR, addr);
++ falcon_write(efx, ®, MD_PHY_ADR_REG_KER);
++
++ if (phy_10g)
++ EFX_POPULATE_OWORD_1(reg, MD_PRT_DEV_ADR, phy_id2);
++ else
++ /* MDIO clause 22 */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_PRT_ADR, phy_id2,
++ MD_DEV_ADR, addr);
++ falcon_write(efx, ®, MD_ID_REG_KER);
++
++ /* Write data */
++ EFX_POPULATE_OWORD_1(reg, MD_TXD, value);
++ falcon_write(efx, ®, MD_TXD_REG_KER);
++
++ EFX_POPULATE_OWORD_2(reg,
++ MD_WRC, 1,
++ MD_GC, (phy_10g ? 0 : 1));
++ falcon_write(efx, ®, MD_CS_REG_KER);
++
++ /* Wait for data to be written */
++ if (falcon_gmii_wait(efx) != 0) {
++ /* Abort the write operation */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_WRC, 0,
++ MD_GC, 1);
++ falcon_write(efx, ®, MD_CS_REG_KER);
++ udelay(10);
+ }
-+ case BLKTAP_IOCTL_MINOR:
-+ {
-+ unsigned long dev = arg;
+
-+ info = tapfds[dev];
++ out:
++ /* Release PHY lock */
++ spin_unlock_bh(&efx->phy_lock);
++}
+
-+ if ((dev > MAX_TAP_DEV) || !info)
-+ return -EINVAL;
++/* Reads a GMII register from a PHY connected to Falcon. If no value
++ * could be read, -1 will be returned. */
++static int falcon_mdio_read(struct net_device *net_dev, int phy_id, int addr)
++{
++ struct efx_nic *efx = (struct efx_nic *)net_dev->priv;
++ unsigned int phy_addr = phy_id & FALCON_PHY_ID_ID_MASK;
++ unsigned int phy_10g = phy_id & FALCON_PHY_ID_10G;
++ efx_oword_t reg;
++ int value = -1;
++ unsigned long flags __attribute__ ((unused));
+
-+ return info->minor;
-+ }
-+ case BLKTAP_IOCTL_MAJOR:
-+ return blktap_major;
++ if (phy_addr == PHY_ADDR_INVALID)
++ return -1;
+
-+ case BLKTAP_QUERY_ALLOC_REQS:
-+ {
-+ WPRINTK("BLKTAP_QUERY_ALLOC_REQS ioctl: %d/%d\n",
-+ alloc_pending_reqs, blkif_reqs);
-+ return (alloc_pending_reqs/blkif_reqs) * 100;
++ /* Our PHY code knows whether it needs to talk clause 22(1G) or 45(10G)
++ * but the generic Linux code does not make any distinction or have
++ * any state for this.
++ * We spot the case where someone tried to talk 22 to a 45 PHY and
++ * redirect the request to the lowest numbered MMD as a clause45
++ * request. This is enough to allow simple queries like id and link
++ * state to succeed. TODO: We may need to do more in future.
++ */
++ if (EFX_ISCLAUSE45(efx) && !phy_10g) {
++ int mmd = ffs(efx->phy_op->mmds) - 1;
++ EFX_TRACE(efx, "Fixing erroneous clause22 read\n");
++ phy_addr = mdio_clause45_pack(phy_addr, mmd)
++ & FALCON_PHY_ID_ID_MASK;
++ phy_10g = 1;
+ }
++
++ /* Obtain PHY lock */
++ spin_lock_bh(&efx->phy_lock);
++
++ /* Check MII not currently being accessed */
++ if (falcon_gmii_wait(efx) != 0)
++ goto out;
++
++ if (!phy_10g) {
++ /* Write the address registers */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_PRT_ADR, phy_addr,
++ MD_DEV_ADR, addr);
++ falcon_write(efx, ®, MD_ID_REG_KER);
++ /* Request data to be read */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_RIC, 1,
++ MD_GC, 1);
++ } else {
++ EFX_POPULATE_OWORD_1(reg, MD_PHY_ADR, addr);
++ falcon_write(efx, ®, MD_PHY_ADR_REG_KER);
++
++ EFX_POPULATE_OWORD_1(reg, MD_PRT_DEV_ADR, phy_addr);
++ falcon_write(efx, ®, MD_ID_REG_KER);
++
++ /* Request data to be read */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_RDC, 1,
++ MD_GC, 0);
++ }
++ falcon_write(efx, ®, MD_CS_REG_KER);
++
++ /* Wait for data to become available */
++ value = falcon_gmii_wait(efx);
++ if (value == 0) {
++ falcon_read(efx, ®, MD_RXD_REG_KER);
++ value = EFX_OWORD_FIELD(reg, MD_RXD);
++ EFX_REGDUMP(efx, "read from GMII %d register %02x, got %04x\n",
++ phy_id, addr, value);
++ } else {
++ /* Abort the read operation */
++ EFX_POPULATE_OWORD_2(reg,
++ MD_RIC, 0,
++ MD_GC, 1);
++ falcon_write(efx, ®, MD_CS_REG_KER);
++
++ EFX_LOG(efx, "read from GMII 0x%x register %02x, got "
++ "error %d\n", phy_id, addr, value);
+ }
-+ return -ENOIOCTLCMD;
++
++ out:
++ /* Release PHY lock */
++ spin_unlock_bh(&efx->phy_lock);
++
++ return value;
+}
+
-+static unsigned int blktap_poll(struct file *filp, poll_table *wait)
++static void falcon_init_mdio(struct mii_if_info *gmii)
+{
-+ tap_blkif_t *info = filp->private_data;
-+
-+ /* do not work on the control device */
-+ if (!info)
-+ return 0;
++ gmii->mdio_read = falcon_mdio_read;
++ gmii->mdio_write = falcon_mdio_write;
++ gmii->phy_id_mask = FALCON_PHY_ID_MASK;
++ gmii->reg_num_mask = ((1 << EFX_WIDTH(MD_DEV_ADR)) - 1);
++}
+
-+ poll_wait(filp, &info->wait, wait);
-+ if (info->ufe_ring.req_prod_pvt != info->ufe_ring.sring->req_prod) {
-+ RING_PUSH_REQUESTS(&info->ufe_ring);
-+ return POLLIN | POLLRDNORM;
-+ }
++static int falcon_probe_gmac_port(struct efx_nic *efx)
++{
++ struct efx_phy_operations *phy_op = efx->phy_op;
++
++ efx->mac_op = &falcon_gmac_operations;
++ efx->loopback_modes = LOOPBACKS_1G_INTERNAL | phy_op->loopbacks;
++ efx->startup_loopbacks = ((1 << LOOPBACK_MAC) |
++ (1 << phy_op->startup_loopback));
+ return 0;
+}
+
-+void blktap_kick_user(int idx)
++static int falcon_probe_xmac_port(struct efx_nic *efx)
+{
-+ tap_blkif_t *info;
-+
-+ info = tapfds[idx];
++ struct efx_phy_operations *phy_op = efx->phy_op;
+
-+ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info)
-+ return;
++ efx->mac_op = &falcon_xmac_operations;
+
-+ wake_up_interruptible(&info->wait);
++ /* The Falcon B0 FPGA only supports XGMII loopback */
++ if (FALCON_REV(efx) >= FALCON_REV_B0 && !efx->is_asic)
++ efx->loopback_modes = (1 << LOOPBACK_XGMII);
++ else
++ efx->loopback_modes = LOOPBACKS_10G_INTERNAL;
++ efx->loopback_modes |= phy_op->loopbacks;
+
-+ return;
++ efx->startup_loopbacks = ((1 << LOOPBACK_XGMII) |
++ (1 << phy_op->startup_loopback));
++ return 0;
+}
+
-+static int do_block_io_op(blkif_t *blkif);
-+static void dispatch_rw_block_io(blkif_t *blkif,
-+ blkif_request_t *req,
-+ pending_req_t *pending_req);
-+static void make_response(blkif_t *blkif, u64 id,
-+ unsigned short op, int st);
++static int falcon_probe_phy(struct efx_nic *efx)
++{
++ switch (efx->phy_type) {
++ case PHY_TYPE_1G_ALASKA:
++ efx->phy_op = &alaska_phy_operations;
++ break;
++ case PHY_TYPE_10XPRESS:
++ efx->phy_op = &falcon_tenxpress_phy_ops;
++ break;
++ case PHY_TYPE_NONE:
++ efx->phy_op = &falcon_null_phy_ops;
++ break;
++ case PHY_TYPE_XFP:
++ efx->phy_op = &falcon_xfp_phy_ops;
++ break;
++ case PHY_TYPE_CX4_RTMR:
++ efx->phy_op = &falcon_txc_phy_ops;
++ break;
++ case PHY_TYPE_PM8358:
++ efx->phy_op = &falcon_pm8358_phy_ops;
++ break;
++ default:
++ EFX_ERR(efx, "Unknown PHY type %d\n",
++ efx->phy_type);
++ return -1;
++ }
++ return 0;
++}
+
-+/******************************************************************
-+ * misc small helpers
-+ */
-+static int req_increase(void)
++/* This call is responsible for hooking in the MAC and PHY operations */
++int falcon_probe_port(struct efx_nic *efx)
+{
-+ int i, j;
++ int rc;
+
-+ if (mmap_alloc >= MAX_PENDING_REQS || mmap_lock)
-+ return -EINVAL;
++ /* Hook in PHY operations table */
++ rc = falcon_probe_phy(efx);
++ if (rc)
++ return rc;
+
-+ pending_reqs[mmap_alloc] = kzalloc(sizeof(pending_req_t)
-+ * blkif_reqs, GFP_KERNEL);
-+ foreign_pages[mmap_alloc] = alloc_empty_pages_and_pagevec(mmap_pages);
++ /* Hook in MAC operations table */
++ if (EFX_IS10G(efx))
++ rc = falcon_probe_xmac_port(efx);
++ else
++ rc = falcon_probe_gmac_port(efx);
++ if (rc)
++ return rc;
+
-+ if (!pending_reqs[mmap_alloc] || !foreign_pages[mmap_alloc])
-+ goto out_of_memory;
++ EFX_LOG(efx, "created port using %cMAC\n",
++ EFX_IS10G(efx) ? 'X' : 'G');
+
-+ DPRINTK("%s: reqs=%d, pages=%d\n",
-+ __FUNCTION__, blkif_reqs, mmap_pages);
++ /* Set up GMII structure for PHY */
++ efx->mii.supports_gmii = 1;
++ falcon_init_mdio(&efx->mii);
+
-+ for (i = 0; i < MAX_PENDING_REQS; i++) {
-+ list_add_tail(&pending_reqs[mmap_alloc][i].free_list,
-+ &pending_free);
-+ pending_reqs[mmap_alloc][i].mem_idx = mmap_alloc;
-+ for (j = 0; j < BLKIF_MAX_SEGMENTS_PER_REQUEST; j++)
-+ BLKTAP_INVALIDATE_HANDLE(&pending_handle(mmap_alloc,
-+ i, j));
-+ }
++ /* Hardware flow ctrl. FalconA RX FIFO too small for pause generation */
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ efx->flow_control = EFX_FC_RX | EFX_FC_TX;
++ else
++ efx->flow_control = EFX_FC_RX;
+
-+ mmap_alloc++;
-+ DPRINTK("# MMAPs increased to %d\n",mmap_alloc);
-+ return 0;
++ /* Allocate buffer for stats */
++ rc = falcon_alloc_buffer(efx, &efx->stats_buffer,
++ FALCON_MAC_STATS_SIZE);
++ if (rc)
++ return rc;
++ EFX_LOG(efx, "stats buffer at %llx (virt %p phys %lx)\n",
++ (unsigned long long)efx->stats_buffer.dma_addr,
++ efx->stats_buffer.addr,
++ virt_to_phys(efx->stats_buffer.addr));
+
-+ out_of_memory:
-+ free_empty_pages_and_pagevec(foreign_pages[mmap_alloc], mmap_pages);
-+ kfree(pending_reqs[mmap_alloc]);
-+ WPRINTK("%s: out of memory\n", __FUNCTION__);
-+ return -ENOMEM;
++ return 0;
+}
+
-+static void mmap_req_del(int mmap)
++void falcon_remove_port(struct efx_nic *efx)
+{
-+ BUG_ON(!spin_is_locked(&pending_free_lock));
++ /* Free stats buffer */
++ falcon_free_buffer(efx, &efx->stats_buffer);
++}
+
-+ kfree(pending_reqs[mmap]);
-+ pending_reqs[mmap] = NULL;
++/**************************************************************************
++ *
++ * Multicast filtering
++ *
++ **************************************************************************
++ */
+
-+ free_empty_pages_and_pagevec(foreign_pages[mmap_alloc], mmap_pages);
-+ foreign_pages[mmap] = NULL;
++void falcon_set_multicast_hash(struct efx_nic *efx)
++{
++ union efx_multicast_hash falcon_mc_hash;
+
-+ mmap_lock = 0;
-+ DPRINTK("# MMAPs decreased to %d\n",mmap_alloc);
-+ mmap_alloc--;
++ /* Broadcast packets go through the multicast hash filter.
++ * ether_crc_le() of the broadcast address is 0xbe2612ff
++ * so we always add bit 0xff to the mask we are given.
++ */
++ memcpy(&falcon_mc_hash, &efx->multicast_hash, sizeof(falcon_mc_hash));
++ set_bit_le(0xff, (void *)&falcon_mc_hash);
++
++ falcon_write(efx, &falcon_mc_hash.oword[0], MAC_MCAST_HASH_REG0_KER);
++ falcon_write(efx, &falcon_mc_hash.oword[1], MAC_MCAST_HASH_REG1_KER);
+}
+
-+static pending_req_t* alloc_req(void)
++/**************************************************************************
++ *
++ * Device reset
++ *
++ **************************************************************************
++ */
++
++static int falcon_clear_b0_memories(struct efx_nic *efx)
+{
-+ pending_req_t *req = NULL;
-+ unsigned long flags;
++ /* Need to clear memories after a reset. On B0 we can do this
++ * via the net function.
++ */
++ int rc = 0, offset;
++ efx_oword_t blanko;
++ efx_dword_t blankd;
++ unsigned long membase_phys, membase_len;
++ void __iomem *membase_orig;
++ unsigned long flags __attribute__ ((unused));
+
-+ spin_lock_irqsave(&pending_free_lock, flags);
++ EFX_ZERO_OWORD(blanko);
++ EFX_ZERO_DWORD(blankd);
++ membase_orig = efx->membase;
++ membase_phys = pci_resource_start(efx->pci_dev, efx->type->mem_bar);
+
-+ if (!list_empty(&pending_free)) {
-+ req = list_entry(pending_free.next, pending_req_t, free_list);
-+ list_del(&req->free_list);
-+ }
++ for (offset = RX_FILTER_TBL0;
++ offset < RX_RSS_INDIR_TBL_B0;
++ offset += 0x10)
++ falcon_write(efx, &blanko, offset);
+
-+ if (req) {
-+ req->inuse = 1;
-+ alloc_pending_reqs++;
++ /* Clear RSS indirection table */
++ for (offset = RX_RSS_INDIR_TBL_B0;
++ offset < RX_RSS_INDIR_TBL_B0 + 0x800;
++ offset += 0x10)
++ /* Clear 6 bits every 16 bytes */
++ falcon_writel(efx, &blankd, offset);
++
++ /* Need to split this into several mappings so MSI-X table and PBA
++ * never get mapped
++ */
++ membase_phys = membase_phys + 0x2800000;
++ membase_len = 0x3000000 - 0x2800000;
++
++ efx->membase = ioremap_nocache(membase_phys, membase_len);
++ if (efx->membase == NULL) {
++ EFX_ERR(efx, "could not map memory BAR %d at %lx+%lx\n",
++ efx->type->mem_bar, membase_phys, membase_len);
++ rc = -ENOMEM;
++ goto out;
++ }
++ /* Clear the buffer table. The first 7/8 of it is a duplicate
++ * of the mapping at 0x800000 and must be accessed 2 DWORDs at
++ * a time. The final 1/8 must be accessed 4 DWORDs at a time.
++ * We make sure to obey both rules at the same time.
++ */
++ for (offset = 0; offset < membase_len; offset += 0x10) {
++ spin_lock_irqsave(&efx->biu_lock, flags);
++ _falcon_writel(efx, 0, offset + 0x0);
++ wmb();
++ _falcon_writel(efx, 0, offset + 0x4);
++ wmb();
++ _falcon_writel(efx, 0, offset + 0x8);
++ wmb();
++ _falcon_writel(efx, 0, offset + 0xc);
++ mmiowb();
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
+ }
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
+
-+ return req;
++ iounmap(efx->membase);
++
++out:
++ /* Restore */
++ efx->membase = membase_orig;
++
++ return rc;
+}
+
-+static void free_req(pending_req_t *req)
++
++/* Resets NIC to known state. This routine must be called in process
++ * context and is allowed to sleep. */
++int falcon_reset_hw(struct efx_nic *efx, enum reset_type method)
+{
-+ unsigned long flags;
-+ int was_empty;
++ efx_oword_t glb_ctl_reg_ker;
++ int rc;
+
-+ spin_lock_irqsave(&pending_free_lock, flags);
++ EFX_LOG(efx, "performing %s hardware reset\n", RESET_TYPE(method));
+
-+ alloc_pending_reqs--;
-+ req->inuse = 0;
-+ if (mmap_lock && (req->mem_idx == mmap_alloc-1)) {
-+ mmap_inuse--;
-+ if (mmap_inuse == 0) mmap_req_del(mmap_alloc-1);
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
-+ return;
++ /* Initiate device reset */
++ if (method == RESET_TYPE_WORLD) {
++ /* Save PCI config space */
++ rc = pci_save_state(efx->pci_dev);
++ if (rc) {
++ EFX_ERR(efx, "failed to backup PCI state of primary "
++ "function prior to hardware reset\n");
++ goto fail1;
++ }
++ if (efx->type->is_dual_func) {
++ rc = pci_save_state(efx->pci_dev2);
++ if (rc) {
++ EFX_ERR(efx, "failed to backup PCI state of "
++ "secondary function prior to "
++ "hardware reset\n");
++ goto fail2;
++ }
++ }
++
++ EFX_POPULATE_OWORD_2(glb_ctl_reg_ker,
++ EXT_PHY_RST_DUR, 0x7,
++ SWRST, 1);
++ } else {
++ int reset_phy = (method == RESET_TYPE_INVISIBLE ?
++ EXCLUDE_FROM_RESET : 0);
++
++ EFX_POPULATE_OWORD_7(glb_ctl_reg_ker,
++ EXT_PHY_RST_CTL, reset_phy,
++ PCIE_CORE_RST_CTL, EXCLUDE_FROM_RESET,
++ PCIE_NSTCK_RST_CTL, EXCLUDE_FROM_RESET,
++ PCIE_SD_RST_CTL, EXCLUDE_FROM_RESET,
++ EE_RST_CTL, EXCLUDE_FROM_RESET,
++ EXT_PHY_RST_DUR, 0x7 /* 10ms */,
++ SWRST, 1);
++ }
++ falcon_write(efx, &glb_ctl_reg_ker, GLB_CTL_REG_KER);
++
++ /* Wait for 50ms for the chip to come out of reset */
++ EFX_LOG(efx, "waiting for hardware reset\n");
++ schedule_timeout_uninterruptible(HZ / 20);
++
++ /* Restore PCI configuration if needed */
++ if (method == RESET_TYPE_WORLD) {
++ if (efx->type->is_dual_func) {
++ rc = pci_restore_state(efx->pci_dev2);
++ if (rc) {
++ EFX_ERR(efx, "failed to restore PCI config for "
++ "the secondary function\n");
++ goto fail3;
++ }
++ }
++ rc = pci_restore_state(efx->pci_dev);
++ if (rc) {
++ EFX_ERR(efx, "failed to restore PCI config for the "
++ "primary function\n");
++ goto fail4;
++ }
++ EFX_LOG(efx, "successfully restored PCI config\n");
+ }
-+ was_empty = list_empty(&pending_free);
-+ list_add(&req->free_list, &pending_free);
+
-+ spin_unlock_irqrestore(&pending_free_lock, flags);
++ /* Assert that reset complete */
++ falcon_read(efx, &glb_ctl_reg_ker, GLB_CTL_REG_KER);
++ if (EFX_OWORD_FIELD(glb_ctl_reg_ker, SWRST) != 0) {
++ rc = -ETIMEDOUT;
++ EFX_ERR(efx, "timed out waiting for hardware reset\n");
++ goto fail5;
++ }
++ EFX_LOG(efx, "hardware reset complete\n");
+
-+ if (was_empty)
-+ wake_up(&pending_free_wq);
++ if (EFX_WORKAROUND_8202(efx)) {
++ rc = falcon_clear_b0_memories(efx);
++ if (rc)
++ goto fail6;
++ }
++
++ return 0;
++
++ /* pci_save_state() and pci_restore_state() MUST be called in pairs */
++fail2:
++fail3:
++ pci_restore_state(efx->pci_dev);
++ /* fall-thru */
++fail1:
++fail4:
++fail5:
++fail6:
++ return rc;
+}
+
-+static void fast_flush_area(pending_req_t *req, int k_idx, int u_idx,
-+ int tapidx)
++/* Zeroes out the SRAM contents. This routine must be called in
++ * process context and is allowed to sleep.
++ */
++static int falcon_reset_sram(struct efx_nic *efx)
+{
-+ struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST*2];
-+ unsigned int i, invcount = 0;
-+ struct grant_handle_pair *khandle;
-+ uint64_t ptep;
-+ int ret, mmap_idx;
-+ unsigned long kvaddr, uvaddr;
-+ tap_blkif_t *info;
-+
++ efx_oword_t srm_cfg_reg_ker, gpio_cfg_reg_ker;
++ int count, onchip, sram_cfg_val;
+
-+ info = tapfds[tapidx];
++ /* Set the SRAM wake/sleep GPIO appropriately. */
++ onchip = (efx->external_sram_cfg == SRM_NB_BSZ_ONCHIP_ONLY);
++ falcon_read(efx, &gpio_cfg_reg_ker, GPIO_CTL_REG_KER);
++ EFX_SET_OWORD_FIELD(gpio_cfg_reg_ker, GPIO1_OEN, 1);
++ EFX_SET_OWORD_FIELD(gpio_cfg_reg_ker, GPIO1_OUT, onchip ? 1 : 0);
++ falcon_write(efx, &gpio_cfg_reg_ker, GPIO_CTL_REG_KER);
+
-+ if ((tapidx < 0) || (tapidx > MAX_TAP_DEV) || !info) {
-+ WPRINTK("fast_flush: Couldn't get info!\n");
-+ return;
-+ }
++ /* Initiate SRAM reset */
++ sram_cfg_val = (efx->external_sram_cfg == SRM_NB_BSZ_ONCHIP_ONLY) ?
++ 0 : efx->external_sram_cfg;
+
-+ if (info->vma != NULL &&
-+ xen_feature(XENFEAT_auto_translated_physmap)) {
-+ down_write(&info->vma->vm_mm->mmap_sem);
-+ zap_page_range(info->vma,
-+ MMAP_VADDR(info->user_vstart, u_idx, 0),
-+ req->nr_pages << PAGE_SHIFT, NULL);
-+ up_write(&info->vma->vm_mm->mmap_sem);
-+ return;
++ EFX_POPULATE_OWORD_2(srm_cfg_reg_ker,
++ SRAM_OOB_BT_INIT_EN, 1,
++ SRM_NUM_BANKS_AND_BANK_SIZE, sram_cfg_val);
++ falcon_write(efx, &srm_cfg_reg_ker, SRM_CFG_REG_KER);
++
++ /* Wait for SRAM reset to complete */
++ count = 0;
++ do {
++ EFX_LOG(efx, "waiting for SRAM reset (attempt %d)...\n", count);
++
++ /* SRAM reset is slow; expect around 16ms */
++ schedule_timeout_uninterruptible(HZ / 50);
++
++ /* Check for reset complete */
++ falcon_read(efx, &srm_cfg_reg_ker, SRM_CFG_REG_KER);
++ if (!EFX_OWORD_FIELD(srm_cfg_reg_ker, SRAM_OOB_BT_INIT_EN)) {
++ EFX_LOG(efx, "SRAM reset complete\n");
++
++ return 0;
++ }
++ } while (++count < 20); /* wait upto 0.4 sec */
++
++ EFX_ERR(efx, "timed out waiting for SRAM reset\n");
++ return -ETIMEDOUT;
++}
++
++static void falcon_spi_device_init(struct efx_spi_device **spi_device_ret,
++ unsigned int device_id, u32 device_type)
++{
++ struct efx_spi_device *spi_device;
++
++ if (device_type != 0) {
++ spi_device = kmalloc(sizeof(*spi_device), GFP_KERNEL);
++ spi_device->device_id = device_id;
++ spi_device->size =
++ 1 << SPI_DEV_TYPE_FIELD(device_type, SPI_DEV_TYPE_SIZE);
++ spi_device->addr_len =
++ SPI_DEV_TYPE_FIELD(device_type, SPI_DEV_TYPE_ADDR_LEN);
++ spi_device->munge_address = (spi_device->size == 1 << 9 &&
++ spi_device->addr_len == 1);
++ spi_device->erase_command =
++ SPI_DEV_TYPE_FIELD(device_type, SPI_DEV_TYPE_ERASE_CMD);
++ spi_device->erase_size =
++ 1 << SPI_DEV_TYPE_FIELD(device_type,
++ SPI_DEV_TYPE_ERASE_SIZE);
++ spi_device->block_size =
++ 1 << SPI_DEV_TYPE_FIELD(device_type,
++ SPI_DEV_TYPE_BLOCK_SIZE);
++ spi_device->read = falcon_spi_read;
++ spi_device->write = falcon_spi_write;
++ } else {
++ spi_device = NULL;
+ }
+
-+ mmap_idx = req->mem_idx;
++ kfree(*spi_device_ret);
++ *spi_device_ret = spi_device;
++}
+
-+ for (i = 0; i < req->nr_pages; i++) {
-+ kvaddr = idx_to_kaddr(mmap_idx, k_idx, i);
-+ uvaddr = MMAP_VADDR(info->user_vstart, u_idx, i);
++/* Extract non-volatile configuration */
++static int falcon_probe_nvconfig(struct efx_nic *efx)
++{
++ int rc;
++ struct falcon_nvconfig *nvconfig;
++ struct efx_spi_device *spi;
++ size_t offset, len;
++ int magic_num, struct_ver, board_rev, onchip_sram;
++
++ nvconfig = kmalloc(sizeof(*nvconfig), GFP_KERNEL);
++
++ /* Read the whole configuration structure into memory. It's
++ * in Falcon's boot device, which may be either flash or
++ * EEPROM, but if both are present Falcon prefers flash. The
++ * boot device is always too large for 9-bit addressing, so we
++ * don't have to munge commands.
++ */
++ spi = efx->spi_flash ? efx->spi_flash : efx->spi_eeprom;
++ for (offset = 0; offset < sizeof(*nvconfig); offset += len) {
++ len = min(sizeof(*nvconfig) - offset,
++ (size_t) FALCON_SPI_MAX_LEN);
++ rc = falcon_spi_read(spi, efx, SPI_READ,
++ NVCONFIG_BASE + offset,
++ (char *)nvconfig + offset, len);
++ if (rc)
++ goto out;
++ }
+
-+ khandle = &pending_handle(mmap_idx, k_idx, i);
++ /* Read the MAC addresses */
++ memcpy(efx->mac_address, nvconfig->mac_address[0], ETH_ALEN);
+
-+ if (khandle->kernel != INVALID_GRANT_HANDLE) {
-+ gnttab_set_unmap_op(&unmap[invcount],
-+ idx_to_kaddr(mmap_idx, k_idx, i),
-+ GNTMAP_host_map, khandle->kernel);
-+ invcount++;
++ /* Read the board configuration. */
++ magic_num = le16_to_cpu(nvconfig->board_magic_num);
++ struct_ver = le16_to_cpu(nvconfig->board_struct_ver);
+
-+ set_phys_to_machine(
-+ __pa(idx_to_kaddr(mmap_idx, k_idx, i))
-+ >> PAGE_SHIFT, INVALID_P2M_ENTRY);
-+ }
++ if (magic_num != NVCONFIG_BOARD_MAGIC_NUM || struct_ver < 2) {
++ EFX_ERR(efx, "Non volatile memory bad magic=%x ver=%x "
++ "therefore using defaults\n", magic_num, struct_ver);
++ efx->phy_type = PHY_TYPE_NONE;
++ efx->mii.phy_id = PHY_ADDR_INVALID;
++ board_rev = 0;
++ onchip_sram = 1;
+
-+ if (khandle->user != INVALID_GRANT_HANDLE) {
-+ BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
-+ if (create_lookup_pte_addr(
-+ info->vma->vm_mm,
-+ MMAP_VADDR(info->user_vstart, u_idx, i),
-+ &ptep) !=0) {
-+ WPRINTK("Couldn't get a pte addr!\n");
-+ return;
-+ }
++ } else {
++ struct falcon_nvconfig_board_v2 *v2 = &nvconfig->board_v2;
++ struct falcon_nvconfig_board_v3 *v3 = &nvconfig->board_v3;
+
-+ gnttab_set_unmap_op(&unmap[invcount], ptep,
-+ GNTMAP_host_map
-+ | GNTMAP_application_map
-+ | GNTMAP_contains_pte,
-+ khandle->user);
-+ invcount++;
-+ }
++ efx->phy_type = v2->port0_phy_type;
++ efx->mii.phy_id = v2->port0_phy_addr;
++ board_rev = le16_to_cpu(v2->board_revision);
++ onchip_sram = EFX_OWORD_FIELD(nvconfig->nic_stat_reg,
++ ONCHIP_SRAM);
+
-+ BLKTAP_INVALIDATE_HANDLE(khandle);
++ if (struct_ver >= 3) {
++ __le32 fl = v3->spi_device_type[EE_SPI_FLASH];
++ __le32 ee = v3->spi_device_type[EE_SPI_EEPROM];
++ falcon_spi_device_init(&efx->spi_flash, EE_SPI_FLASH,
++ le32_to_cpu(fl));
++ falcon_spi_device_init(&efx->spi_eeprom, EE_SPI_EEPROM,
++ le32_to_cpu(ee));
++ }
+ }
-+ ret = HYPERVISOR_grant_table_op(
-+ GNTTABOP_unmap_grant_ref, unmap, invcount);
-+ BUG_ON(ret);
-+
-+ if (info->vma != NULL && !xen_feature(XENFEAT_auto_translated_physmap))
-+ zap_page_range(info->vma,
-+ MMAP_VADDR(info->user_vstart, u_idx, 0),
-+ req->nr_pages << PAGE_SHIFT, NULL);
-+}
+
-+/******************************************************************
-+ * SCHEDULER FUNCTIONS
-+ */
++ EFX_LOG(efx, "PHY is %s(%d) phy_id %d\n",
++ PHY_TYPE(efx), efx->phy_type,
++ efx->mii.phy_id);
+
-+static void print_stats(blkif_t *blkif)
-+{
-+ printk(KERN_DEBUG "%s: oo %3d | rd %4d | wr %4d\n",
-+ current->comm, blkif->st_oo_req,
-+ blkif->st_rd_req, blkif->st_wr_req);
-+ blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
-+ blkif->st_rd_req = 0;
-+ blkif->st_wr_req = 0;
-+ blkif->st_oo_req = 0;
-+}
++ efx_set_board_info(efx, board_rev);
+
-+int tap_blkif_schedule(void *arg)
-+{
-+ blkif_t *blkif = arg;
++ /* Read the SRAM configuration. The register is initialised
++ * automatically but might may been reset since boot.
++ */
++ if (onchip_sram) {
++ efx->external_sram_cfg = SRM_NB_BSZ_ONCHIP_ONLY;
++ } else {
++ efx->external_sram_cfg =
++ EFX_OWORD_FIELD(nvconfig->srm_cfg_reg,
++ SRM_NUM_BANKS_AND_BANK_SIZE);
++ WARN_ON(efx->external_sram_cfg == SRM_NB_BSZ_RESERVED);
++ /* Replace invalid setting with the smallest defaults */
++ if (efx->external_sram_cfg == SRM_NB_BSZ_DEFAULT)
++ efx->external_sram_cfg = SRM_NB_BSZ_1BANKS_2M;
++ }
++ EFX_LOG(efx, "external_sram_cfg=%d (>=0 is external)\n",
++ efx->external_sram_cfg);
+
-+ blkif_get(blkif);
++ out:
++ kfree(nvconfig);
++ return rc;
++}
+
-+ if (debug_lvl)
-+ printk(KERN_DEBUG "%s: started\n", current->comm);
++/* Looks at available SRAM resources and silicon revision, and works out
++ * how many queues we can support, and where things like descriptor caches
++ * should live. */
++static int falcon_dimension_resources(struct efx_nic *efx)
++{
++ unsigned buffer_entry_bytes, internal_dcs_entries, dcs;
++ struct falcon_nic_data *nic_data = efx->nic_data;
++ struct efx_dl_falcon_resources *res = &nic_data->resources;
++
++ /* Fill out the driverlink resource list */
++ res->hdr.type = EFX_DL_FALCON_RESOURCES;
++ res->biu_lock = &efx->biu_lock;
++ efx->dl_info = &res->hdr;
++
++ /* This is set to 16 for a good reason. In summary, if larger than
++ * 16, the descriptor cache holds more than a default socket
++ * buffer's worth of packets (for UDP we can only have at most one
++ * socket buffer's worth outstanding). This combined with the fact
++ * that we only get 1 TX event per descriptor cache means the NIC
++ * goes idle.
++ * 16 gives us up to 256 TXQs on Falcon B in internal-SRAM mode,
++ * and up to 512 on Falcon A.
++ */
++ nic_data->tx_dc_entries = 16;
++
++ /* Set the RX descriptor cache size. Values 16, 32 and 64 are
++ * supported (8 won't work). Bigger is better, especially on B
++ * silicon.
++ */
++ nic_data->rx_dc_entries = descriptor_cache_size;
++ dcs = ffs(nic_data->rx_dc_entries);
++ if ((dcs < 5) || (dcs > 7) ||
++ ((1 << (dcs - 1)) != nic_data->rx_dc_entries)) {
++ EFX_ERR(efx, "bad descriptor_cache_size=%d (dcs=%d)\n",
++ nic_data->rx_dc_entries, dcs);
++ return -EINVAL;
++ }
+
-+ while (!kthread_should_stop()) {
-+ if (try_to_freeze())
-+ continue;
++ /* NB. The minimum values get increased as this driver initialises
++ * its resources, so this should prevent any overlap.
++ */
++ switch (FALCON_REV(efx)) {
++ case FALCON_REV_A1:
++ res->rxq_min = res->txq_min = 16;
++ res->evq_int_min = res->evq_int_max = 4;
++ res->evq_timer_min = 5;
++ res->evq_timer_max = 4096;
++ internal_dcs_entries = 8192;
++ break;
++ case FALCON_REV_B0:
++ default:
++ res->rxq_min = res->txq_min = res->evq_int_min = 0;
++ res->evq_int_max = 64;
++ res->evq_timer_min = 64;
++ res->evq_timer_max = 4096;
++ internal_dcs_entries = 4096;
++ break;
++ }
+
-+ wait_event_interruptible(
-+ blkif->wq,
-+ blkif->waiting_reqs || kthread_should_stop());
-+ wait_event_interruptible(
-+ pending_free_wq,
-+ !list_empty(&pending_free) || kthread_should_stop());
++ buffer_entry_bytes = 8;
+
-+ blkif->waiting_reqs = 0;
-+ smp_mb(); /* clear flag *before* checking for work */
++ if (efx->external_sram_cfg == SRM_NB_BSZ_ONCHIP_ONLY) {
++ res->rxq_max = internal_dcs_entries / nic_data->rx_dc_entries;
++ res->txq_max = internal_dcs_entries / nic_data->tx_dc_entries;
++ /* Prog model says 8K entries for buffer table in internal
++ * mode. But does this not depend on full/half mode?
++ */
++ res->buffer_table_max = 8192;
++ nic_data->tx_dc_base = 0x130000;
++ nic_data->rx_dc_base = 0x100000;
++ } else {
++ unsigned sram_bytes, vnic_bytes, max_vnics, n_vnics;
+
-+ if (do_block_io_op(blkif))
-+ blkif->waiting_reqs = 1;
++ /* Determine how much SRAM we have to play with. We have
++ * to fit buffer table and descriptor caches in.
++ */
++ switch (efx->external_sram_cfg) {
++ case SRM_NB_BSZ_1BANKS_2M:
++ default:
++ sram_bytes = 2 * 1024 * 1024;
++ break;
++ case SRM_NB_BSZ_1BANKS_4M:
++ case SRM_NB_BSZ_2BANKS_4M:
++ sram_bytes = 4 * 1024 * 1024;
++ break;
++ case SRM_NB_BSZ_1BANKS_8M:
++ case SRM_NB_BSZ_2BANKS_8M:
++ sram_bytes = 8 * 1024 * 1024;
++ break;
++ case SRM_NB_BSZ_2BANKS_16M:
++ sram_bytes = 16 * 1024 * 1024;
++ break;
++ }
++ /* For each VNIC allow at least 512 buffer table entries
++ * and descriptor cache for an rxq and txq. Buffer table
++ * space for evqs and dmaqs is relatively trivial, so not
++ * considered in this calculation.
++ */
++ vnic_bytes = (512 * buffer_entry_bytes
++ + nic_data->rx_dc_entries * 8
++ + nic_data->tx_dc_entries * 8);
++ max_vnics = sram_bytes / vnic_bytes;
++ for (n_vnics = 1; n_vnics < res->evq_timer_min + max_vnics;)
++ n_vnics *= 2;
++ res->rxq_max = n_vnics;
++ res->txq_max = n_vnics;
+
-+ if (log_stats && time_after(jiffies, blkif->st_print))
-+ print_stats(blkif);
++ dcs = n_vnics * nic_data->tx_dc_entries * 8;
++ nic_data->tx_dc_base = sram_bytes - dcs;
++ dcs = n_vnics * nic_data->rx_dc_entries * 8;
++ nic_data->rx_dc_base = nic_data->tx_dc_base - dcs;
++ res->buffer_table_max = nic_data->rx_dc_base / 8;
+ }
+
-+ if (log_stats)
-+ print_stats(blkif);
-+ if (debug_lvl)
-+ printk(KERN_DEBUG "%s: exiting\n", current->comm);
++ if (efx->type->is_dual_func)
++ res->flags |= EFX_DL_FALCON_DUAL_FUNC;
+
-+ blkif->xenblkd = NULL;
-+ blkif_put(blkif);
++ if (EFX_INT_MODE_USE_MSI(efx))
++ res->flags |= EFX_DL_FALCON_USE_MSI;
+
+ return 0;
+}
+
-+/******************************************************************
-+ * COMPLETION CALLBACK -- Called by user level ioctl()
++/* Probe the NIC variant (revision, ASIC vs FPGA, function count, port
++ * count, port speed). Set workaround and feature flags accordingly.
+ */
-+
-+static int blktap_read_ufe_ring(tap_blkif_t *info)
++static int falcon_probe_nic_variant(struct efx_nic *efx)
+{
-+ /* This is called to read responses from the UFE ring. */
-+ RING_IDX i, j, rp;
-+ blkif_response_t *resp;
-+ blkif_t *blkif=NULL;
-+ int pending_idx, usr_idx, mmap_idx;
-+ pending_req_t *pending_req;
-+
-+ if (!info)
-+ return 0;
++ efx_oword_t altera_build;
+
-+ /* We currently only forward packets in INTERCEPT_FE mode. */
-+ if (!(info->mode & BLKTAP_MODE_INTERCEPT_FE))
-+ return 0;
-+
-+ /* for each outstanding message on the UFEring */
-+ rp = info->ufe_ring.sring->rsp_prod;
-+ rmb();
-+
-+ for (i = info->ufe_ring.rsp_cons; i != rp; i++) {
-+ blkif_response_t res;
-+ resp = RING_GET_RESPONSE(&info->ufe_ring, i);
-+ memcpy(&res, resp, sizeof(res));
-+ mb(); /* rsp_cons read by RING_FULL() in do_block_io_op(). */
-+ ++info->ufe_ring.rsp_cons;
-+
-+ /*retrieve [usr_idx] to [mmap_idx,pending_idx] mapping*/
-+ usr_idx = (int)res.id;
-+ pending_idx = MASK_PEND_IDX(ID_TO_IDX(info->idx_map[usr_idx]));
-+ mmap_idx = ID_TO_MIDX(info->idx_map[usr_idx]);
-+
-+ if ( (mmap_idx >= mmap_alloc) ||
-+ (ID_TO_IDX(info->idx_map[usr_idx]) >= MAX_PENDING_REQS) )
-+ WPRINTK("Incorrect req map"
-+ "[%d], internal map [%d,%d (%d)]\n",
-+ usr_idx, mmap_idx,
-+ ID_TO_IDX(info->idx_map[usr_idx]),
-+ MASK_PEND_IDX(
-+ ID_TO_IDX(info->idx_map[usr_idx])));
++ falcon_read(efx, &altera_build, ALTERA_BUILD_REG_KER);
++ efx->is_asic = EFX_OWORD_FIELD(altera_build, VER_ALL) == 0;
+
-+ pending_req = &pending_reqs[mmap_idx][pending_idx];
-+ blkif = pending_req->blkif;
-+
-+ for (j = 0; j < pending_req->nr_pages; j++) {
++#if !defined(EFX_USE_PCI_DEV_REVISION)
++ {
++ int rc;
++ rc = pci_read_config_byte(efx->pci_dev, PCI_CLASS_REVISION,
++ &efx->revision);
++ if (rc)
++ return rc;
++ }
++#endif
++ switch (FALCON_REV(efx)) {
++ case FALCON_REV_A0:
++ case 0xff:
++ EFX_ERR(efx, "Falcon rev A0 not supported\n");
++ return -ENODEV;
+
-+ unsigned long kvaddr, uvaddr;
-+ struct page **map = info->vma->vm_private_data;
-+ struct page *pg;
-+ int offset;
++ case FALCON_REV_A1:{
++ efx_oword_t nic_stat;
+
-+ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, j);
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, j);
++ falcon_read(efx, &nic_stat, NIC_STAT_REG);
+
-+ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
-+ ClearPageReserved(pg);
-+ offset = (uvaddr - info->vma->vm_start)
-+ >> PAGE_SHIFT;
-+ map[offset] = NULL;
++ if (!efx->is_asic) {
++ EFX_ERR(efx, "Falcon rev A1 FPGA not supported\n");
++ return -ENODEV;
+ }
-+ fast_flush_area(pending_req, pending_idx, usr_idx, info->minor);
-+ info->idx_map[usr_idx] = INVALID_REQ;
-+ make_response(blkif, pending_req->id, res.operation,
-+ res.status);
-+ blkif_put(pending_req->blkif);
-+ free_req(pending_req);
++ if (EFX_OWORD_FIELD(nic_stat, STRAP_PCIE) == 0) {
++ EFX_ERR(efx, "Falcon rev A1 PCI-X not supported\n");
++ return -ENODEV;
++ }
++ efx->is_10g = EFX_OWORD_FIELD(nic_stat, STRAP_10G);
++ efx->silicon_rev = "falcon/a1";
++ break;
+ }
-+
-+ return 0;
-+}
+
++ case FALCON_REV_B0:{
++ efx->is_10g = 1;
++ efx->silicon_rev = "falcon/b0";
++ break;
++ }
+
-+/******************************************************************************
-+ * NOTIFICATION FROM GUEST OS.
-+ */
++ default:
++ EFX_ERR(efx, "Unknown Falcon rev %d\n", FALCON_REV(efx));
++ return -ENODEV;
++ }
+
-+static void blkif_notify_work(blkif_t *blkif)
-+{
-+ blkif->waiting_reqs = 1;
-+ wake_up(&blkif->wq);
++ return 0;
+}
+
-+irqreturn_t tap_blkif_be_int(int irq, void *dev_id, struct pt_regs *regs)
++/* Probe all SPI devices on the NIC */
++static void falcon_probe_spi_devices(struct efx_nic *efx)
+{
-+ blkif_notify_work(dev_id);
-+ return IRQ_HANDLED;
-+}
++ efx_oword_t nic_stat, gpio_ctl, ee_vpd_cfg;
++ unsigned int has_flash, has_eeprom, boot_is_external;
+
++ falcon_read(efx, &gpio_ctl, GPIO_CTL_REG_KER);
++ falcon_read(efx, &nic_stat, NIC_STAT_REG);
++ falcon_read(efx, &ee_vpd_cfg, EE_VPD_CFG_REG_KER);
+
++ has_flash = EFX_OWORD_FIELD(nic_stat, SF_PRST);
++ has_eeprom = EFX_OWORD_FIELD(nic_stat, EE_PRST);
++ boot_is_external = EFX_OWORD_FIELD(gpio_ctl, BOOTED_USING_NVDEVICE);
+
-+/******************************************************************
-+ * DOWNWARD CALLS -- These interface with the block-device layer proper.
-+ */
-+static int print_dbug = 1;
-+static int do_block_io_op(blkif_t *blkif)
-+{
-+ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
-+ blkif_request_t req;
-+ pending_req_t *pending_req;
-+ RING_IDX rc, rp;
-+ int more_to_do = 0;
-+ tap_blkif_t *info;
++ if (has_flash) {
++ u32 flash_device_type;
+
-+ rc = blk_rings->common.req_cons;
-+ rp = blk_rings->common.sring->req_prod;
-+ rmb(); /* Ensure we see queued requests up to 'rp'. */
++ if (flash_type == -1) {
++ /* Default flash SPI device: Atmel AT25F1024
++ * 128 KB, 24-bit address, 32 KB erase block,
++ * 256 B write block
++ */
++ flash_device_type =
++ (17 << SPI_DEV_TYPE_SIZE_LBN)
++ | (3 << SPI_DEV_TYPE_ADDR_LEN_LBN)
++ | (0x52 << SPI_DEV_TYPE_ERASE_CMD_LBN)
++ | (15 << SPI_DEV_TYPE_ERASE_SIZE_LBN)
++ | (8 << SPI_DEV_TYPE_BLOCK_SIZE_LBN);
++ } else {
++ flash_device_type = flash_type;
++ }
+
-+ /*Check blkif has corresponding UE ring*/
-+ if (blkif->dev_num < 0) {
-+ /*oops*/
-+ if (print_dbug) {
-+ WPRINTK("Corresponding UE "
-+ "ring does not exist!\n");
-+ print_dbug = 0; /*We only print this message once*/
++ falcon_spi_device_init(&efx->spi_flash, EE_SPI_FLASH,
++ flash_device_type);
++
++ if (!boot_is_external) {
++ /* Disable VPD and set clock dividers to safe
++ * values for initial programming.
++ */
++ EFX_LOG(efx, "Booted from internal ASIC settings;"
++ " setting SPI config\n");
++ EFX_POPULATE_OWORD_3(ee_vpd_cfg, EE_VPD_EN, 0,
++ /* 125 MHz / 7 ~= 20 MHz */
++ EE_SF_CLOCK_DIV, 7,
++ /* 125 MHz / 63 ~= 2 MHz */
++ EE_EE_CLOCK_DIV, 63);
++ falcon_write(efx, &ee_vpd_cfg, EE_VPD_CFG_REG_KER);
+ }
-+ return 0;
+ }
+
-+ info = tapfds[blkif->dev_num];
++ if (has_eeprom) {
++ u32 eeprom_device_type;
+
-+ if (blkif->dev_num > MAX_TAP_DEV || !info || !info->dev_inuse) {
-+ if (print_dbug) {
-+ WPRINTK("Can't get UE info!\n");
-+ print_dbug = 0;
++ /* eeprom_type may be -1 (default) for automatic detection,
++ * 0 or 1 to select the default or large EEPROM, or
++ * some larger number to specify the precise configuration
++ */
++ if (eeprom_type == -1 || eeprom_type <= 1) {
++ /* If it has no flash, it must have a large EEPROM
++ * for chip config; otherwise check whether 9-bit
++ * addressing is used for VPD configuration
++ */
++ if (eeprom_type == 0 ||
++ (eeprom_type == -1 && has_flash &&
++ (!boot_is_external ||
++ EFX_OWORD_FIELD(ee_vpd_cfg,
++ EE_VPD_EN_AD9_MODE)))) {
++ /* Default SPI device: Atmel AT25040 or similar
++ * 512 B, 9-bit address, 8 B write block
++ */
++ eeprom_device_type =
++ (9 << SPI_DEV_TYPE_SIZE_LBN)
++ | (1 << SPI_DEV_TYPE_ADDR_LEN_LBN)
++ | (3 << SPI_DEV_TYPE_BLOCK_SIZE_LBN);
++ } else {
++ /* "Large" SPI device: Atmel AT25640 or similar
++ * 8 KB, 16-bit address, 32 B write block
++ */
++ eeprom_device_type =
++ (13 << SPI_DEV_TYPE_SIZE_LBN)
++ | (2 << SPI_DEV_TYPE_ADDR_LEN_LBN)
++ | (5 << SPI_DEV_TYPE_BLOCK_SIZE_LBN);
++ }
++ } else {
++ eeprom_device_type = eeprom_type;
+ }
-+ return 0;
++
++ falcon_spi_device_init(&efx->spi_eeprom, EE_SPI_EEPROM,
++ eeprom_device_type);
+ }
+
-+ while (rc != rp) {
-+
-+ if (RING_FULL(&info->ufe_ring)) {
-+ WPRINTK("RING_FULL! More to do\n");
-+ more_to_do = 1;
-+ break;
-+ }
++ EFX_LOG(efx, "flash is %s, EEPROM is %s\n",
++ (has_flash ? "present" : "absent"),
++ (has_eeprom ? "present" : "absent"));
++}
+
-+ if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc)) {
-+ WPRINTK("RING_REQUEST_CONS_OVERFLOW!"
-+ " More to do\n");
-+ more_to_do = 1;
-+ break;
-+ }
++static void falcon_remove_spi_devices(struct efx_nic *efx)
++{
++ kfree(efx->spi_eeprom);
++ efx->spi_eeprom = NULL;
++ kfree(efx->spi_flash);
++ efx->spi_flash = NULL;
++}
+
-+ pending_req = alloc_req();
-+ if (NULL == pending_req) {
-+ blkif->st_oo_req++;
-+ more_to_do = 1;
-+ break;
-+ }
++#ifdef CONFIG_SFC_DEBUGFS
+
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc),
-+ sizeof(req));
-+ break;
-+ case BLKIF_PROTOCOL_X86_32:
-+ blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
-+ break;
-+ case BLKIF_PROTOCOL_X86_64:
-+ blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
-+ break;
-+ default:
-+ BUG();
-+ }
-+ blk_rings->common.req_cons = ++rc; /* before make_response() */
++/* Generate a hardware revision string */
++int falcon_debugfs_read_hardware_desc(struct seq_file *file, void *data)
++{
++ struct efx_nic *efx = data;
++ efx_oword_t altera_build;
++ int major, minor, build;
++ int rc, len;
+
-+ switch (req.operation) {
-+ case BLKIF_OP_READ:
-+ blkif->st_rd_req++;
-+ dispatch_rw_block_io(blkif, &req, pending_req);
-+ break;
++ if (efx->is_asic) {
++ rc = seq_puts(file, "Falcon ASIC");
++ } else {
++ falcon_read(efx, &altera_build, ALTERA_BUILD_REG_KER);
+
-+ case BLKIF_OP_WRITE:
-+ blkif->st_wr_req++;
-+ dispatch_rw_block_io(blkif, &req, pending_req);
-+ break;
++ major = EFX_OWORD_FIELD(altera_build, VER_MAJOR);
++ minor = EFX_OWORD_FIELD(altera_build, VER_MINOR);
++ build = EFX_OWORD_FIELD(altera_build, VER_BUILD);
++ rc = seq_printf(file, "Falcon FPGA v%x.%x.%x",
++ major, minor, build);
++ }
++ len = rc;
+
-+ default:
-+ WPRINTK("unknown operation [%d]\n",
-+ req.operation);
-+ make_response(blkif, req.id, req.operation,
-+ BLKIF_RSP_ERROR);
-+ free_req(pending_req);
-+ break;
-+ }
++ switch (FALCON_REV(efx)) {
++ case FALCON_REV_A1:
++ rc = seq_puts(file, " rev A1 ");
++ break;
++ case FALCON_REV_B0:
++ rc = seq_puts(file, " rev B0 ");
++ break;
++ default:
++ rc = seq_puts(file, " rev ?? ");
++ break;
+ }
-+
-+ blktap_kick_user(blkif->dev_num);
++ len += rc;
+
-+ return more_to_do;
++ rc = seq_printf(file, "%s %s\n",
++ efx->is_10g ? "10G" : "1G", PHY_TYPE(efx));
++ len += rc;
++
++ return rc < 0 ? rc : len;
+}
+
-+static void dispatch_rw_block_io(blkif_t *blkif,
-+ blkif_request_t *req,
-+ pending_req_t *pending_req)
++#endif /* CONFIG_SFC_DEBUGFS */
++
++int falcon_probe_nic(struct efx_nic *efx)
+{
-+ extern void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]);
-+ int op, operation = (req->operation == BLKIF_OP_WRITE) ? WRITE : READ;
-+ struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST*2];
-+ unsigned int nseg;
-+ int ret, i, nr_sects = 0;
-+ tap_blkif_t *info;
-+ blkif_request_t *target;
-+ int pending_idx = RTN_PEND_IDX(pending_req,pending_req->mem_idx);
-+ int usr_idx;
-+ uint16_t mmap_idx = pending_req->mem_idx;
++ struct falcon_nic_data *nic_data;
++ int rc;
+
-+ if (blkif->dev_num < 0 || blkif->dev_num > MAX_TAP_DEV)
-+ goto fail_response;
++ /* Initialise I2C interface state */
++ efx->i2c.efx = efx;
++ efx->i2c.op = &falcon_i2c_bit_operations;
++ efx->i2c.sda = 1;
++ efx->i2c.scl = 1;
+
-+ info = tapfds[blkif->dev_num];
-+ if (info == NULL)
-+ goto fail_response;
++ /* Determine number of ports etc. */
++ rc = falcon_probe_nic_variant(efx);
++ if (rc)
++ goto fail1;
+
-+ /* Check we have space on user ring - should never fail. */
-+ usr_idx = GET_NEXT_REQ(info->idx_map);
-+ if (usr_idx == INVALID_REQ) {
-+ BUG();
-+ goto fail_response;
-+ }
++ /* Probe secondary function if expected */
++ if (efx->type->is_dual_func) {
++ struct pci_dev *dev = pci_dev_get(efx->pci_dev);
+
-+ /* Check that number of segments is sane. */
-+ nseg = req->nr_segments;
-+ if ( unlikely(nseg == 0) ||
-+ unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST) ) {
-+ WPRINTK("Bad number of segments in request (%d)\n", nseg);
-+ goto fail_response;
-+ }
-+
-+ /* Make sure userspace is ready. */
-+ if (!info->ring_ok) {
-+ WPRINTK("blktap: ring not ready for requests!\n");
-+ goto fail_response;
++ while ((dev = pci_get_device(EFX_VENDID_SFC, FALCON_A_S_DEVID,
++ dev))) {
++ if (dev->bus == efx->pci_dev->bus &&
++ dev->devfn == efx->pci_dev->devfn + 1) {
++ efx->pci_dev2 = dev;
++ break;
++ }
++ }
++ if (!efx->pci_dev2) {
++ EFX_ERR(efx, "failed to find secondary function\n");
++ rc = -ENODEV;
++ goto fail2;
++ }
+ }
+
-+ if (RING_FULL(&info->ufe_ring)) {
-+ WPRINTK("blktap: fe_ring is full, can't add "
-+ "IO Request will be dropped. %d %d\n",
-+ RING_SIZE(&info->ufe_ring),
-+ RING_SIZE(&blkif->blk_rings.common));
-+ goto fail_response;
++ /* Now we can reset the NIC */
++ rc = falcon_reset_hw(efx, RESET_TYPE_ALL);
++ if (rc) {
++ EFX_ERR(efx, "failed to reset NIC\n");
++ goto fail3;
+ }
+
-+ pending_req->blkif = blkif;
-+ pending_req->id = req->id;
-+ pending_req->operation = operation;
-+ pending_req->status = BLKIF_RSP_OKAY;
-+ pending_req->nr_pages = nseg;
-+ op = 0;
-+ for (i = 0; i < nseg; i++) {
-+ unsigned long uvaddr;
-+ unsigned long kvaddr;
-+ uint64_t ptep;
-+ uint32_t flags;
++ /* Allocate memory for INT_KER */
++ rc = falcon_alloc_buffer(efx, &efx->irq_status, sizeof(efx_oword_t));
++ if (rc)
++ goto fail4;
++ BUG_ON(efx->irq_status.dma_addr & 0x0f);
+
-+ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i);
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
++ EFX_LOG(efx, "INT_KER at %llx (virt %p phys %lx)\n",
++ (unsigned long long)efx->irq_status.dma_addr,
++ efx->irq_status.addr, virt_to_phys(efx->irq_status.addr));
+
-+ flags = GNTMAP_host_map;
-+ if (operation == WRITE)
-+ flags |= GNTMAP_readonly;
-+ gnttab_set_map_op(&map[op], kvaddr, flags,
-+ req->seg[i].gref, blkif->domid);
-+ op++;
-+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* Now map it to user. */
-+ ret = create_lookup_pte_addr(info->vma->vm_mm,
-+ uvaddr, &ptep);
-+ if (ret) {
-+ WPRINTK("Couldn't get a pte addr!\n");
-+ goto fail_flush;
-+ }
++ /* Determine attached SPI devices */
++ falcon_probe_spi_devices(efx);
+
-+ flags = GNTMAP_host_map | GNTMAP_application_map
-+ | GNTMAP_contains_pte;
-+ if (operation == WRITE)
-+ flags |= GNTMAP_readonly;
-+ gnttab_set_map_op(&map[op], ptep, flags,
-+ req->seg[i].gref, blkif->domid);
-+ op++;
-+ }
++ /* Read in the non-volatile configuration */
++ rc = falcon_probe_nvconfig(efx);
++ if (rc)
++ goto fail5;
+
-+ nr_sects += (req->seg[i].last_sect -
-+ req->seg[i].first_sect + 1);
++ if (!efx->is_10g && efx->phy_type != PHY_TYPE_1G_ALASKA) {
++ /* Actually using 1G port, not 10G port */
++ efx->phy_type = PHY_TYPE_1G_ALASKA;
++ efx->mii.phy_id = 2;
+ }
+
-+ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, op);
-+ BUG_ON(ret);
++ /* Decide how many resources we can allocate, to ourselves
++ * and to driverlink clients */
++ nic_data = kzalloc(sizeof(*nic_data), GFP_KERNEL);
++ efx->nic_data = (void *) nic_data;
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ for (i = 0; i < (nseg*2); i+=2) {
-+ unsigned long uvaddr;
-+ unsigned long kvaddr;
-+ unsigned long offset;
-+ struct page *pg;
++ rc = falcon_dimension_resources(efx);
++ if (rc)
++ goto fail6;
+
-+ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i/2);
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i/2);
++ return 0;
+
-+ if (unlikely(map[i].status != 0)) {
-+ WPRINTK("invalid kernel buffer -- "
-+ "could not remap it\n");
-+ ret |= 1;
-+ map[i].handle = INVALID_GRANT_HANDLE;
-+ }
++ fail6:
++ kfree(nic_data);
++ efx->nic_data = efx->dl_info = NULL;
++ fail5:
++ falcon_remove_spi_devices(efx);
++ falcon_free_buffer(efx, &efx->irq_status);
++ fail4:
++ /* fall-thru */
++ fail3:
++ if (efx->pci_dev2) {
++ pci_dev_put(efx->pci_dev2);
++ efx->pci_dev2 = NULL;
++ }
++ fail2:
++ /* fall-thru */
++ fail1:
++ return rc;
++}
+
-+ if (unlikely(map[i+1].status != 0)) {
-+ WPRINTK("invalid user buffer -- "
-+ "could not remap it\n");
-+ ret |= 1;
-+ map[i+1].handle = INVALID_GRANT_HANDLE;
-+ }
++static int falcon_check_power_limit(struct efx_nic *efx)
++{
++ int pciecap_offset = pci_find_capability(efx->pci_dev, PCI_CAP_ID_EXP);
++ u32 pcie_devcap;
++ unsigned val, scale;
++ int rc;
+
-+ pending_handle(mmap_idx, pending_idx, i/2).kernel
-+ = map[i].handle;
-+ pending_handle(mmap_idx, pending_idx, i/2).user
-+ = map[i+1].handle;
++ if (!pciecap_offset)
++ return -EIO;
++ rc = pci_read_config_dword(efx->pci_dev,
++ (pciecap_offset + PCI_EXP_DEVCAP),
++ &pcie_devcap);
++ if (rc)
++ return rc;
+
-+ if (ret)
-+ continue;
++ val = ((pcie_devcap & PCI_EXP_DEVCAP_PWR_VAL) >>
++ PCI_EXP_DEVCAP_PWR_VAL_LBN);
++ scale = ((pcie_devcap & PCI_EXP_DEVCAP_PWR_SCL) >>
++ PCI_EXP_DEVCAP_PWR_SCL_LBN);
+
-+ set_phys_to_machine(__pa(kvaddr) >> PAGE_SHIFT,
-+ FOREIGN_FRAME(map[i].dev_bus_addr
-+ >> PAGE_SHIFT));
-+ offset = (uvaddr - info->vma->vm_start) >> PAGE_SHIFT;
-+ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
-+ ((struct page **)info->vma->vm_private_data)[offset] =
-+ pg;
-+ }
-+ } else {
-+ for (i = 0; i < nseg; i++) {
-+ unsigned long uvaddr;
-+ unsigned long kvaddr;
-+ unsigned long offset;
-+ struct page *pg;
++ /* Re-scale to milliwatts if necessary */
++ while (scale != 3) {
++ val *= 10;
++ scale++;
++ }
+
-+ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i);
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
++ if (val != 0 && efx->board_info.mwatts > val) {
++ EFX_ERR(efx, "board needs %d mW but only %d mW available\n",
++ efx->board_info.mwatts, val);
++ return -EIO;
++ }
+
-+ if (unlikely(map[i].status != 0)) {
-+ WPRINTK("invalid kernel buffer -- "
-+ "could not remap it\n");
-+ ret |= 1;
-+ map[i].handle = INVALID_GRANT_HANDLE;
-+ }
++ return 0;
++}
+
-+ pending_handle(mmap_idx, pending_idx, i).kernel
-+ = map[i].handle;
++static void falcon_init_ack_repl_timer(struct efx_nic *efx, int num_lanes)
++{
++ unsigned tlp_size;
++ efx_dword_t pcie_ack_rpl_reg;
++ efx_dword_t pcie_ack_freq_reg;
++ efx_dword_t pcie_ctrl_stat_reg;
++ u16 pcie_devicectrl;
++ int lut_index, tlp_size_decoded;
++ int current_replay, expected_replay;
++ int current_ack_timer, current_ack_freq;
++
++ static struct efx_tlp_ack_factor {
++ int tlp;
++ int replay[4]; /* 0=1x, 1=2x, 3=4x, 4=8x (see pcie docs) */
++ } tlp_ack_factor_lut[4] = {
++ { 128, { 421, 257, 174, 166 } },
++ { 256, { 689, 391, 241, 225 } },
++ { 512, { 903, 498, 295, 193 } },
++ { 1024, { 1670, 881, 487, 290 } }
++ };
++ struct efx_tlp_ack_factor *tlp_ack_factor;
++
++ /* Get TLP size */
++ falcon_pcie_core_read_reg(efx, PCIE_CORE_ADDR_PCIE_DEVICE_CTRL_STAT,
++ &pcie_ctrl_stat_reg);
++ pcie_devicectrl = (u16) EFX_EXTRACT_DWORD(pcie_ctrl_stat_reg, 0, 15);
++ tlp_size = ((PCI_EXP_DEVCTL_PAYLOAD & pcie_devicectrl) >>
++ ffs(PCI_EXP_DEVCTL_PAYLOAD));
++ EFX_WARN_ON_PARANOID(tlp_size > 3); /* => 1024 bytes */
++ tlp_ack_factor = &tlp_ack_factor_lut[tlp_size & 0x3];
++ tlp_size_decoded = tlp_ack_factor->tlp;
++
++ /* Get actual ack & actual and expected replay settings */
++ falcon_pcie_core_read_reg(efx, PCIE_CORE_ADDR_ACK_RPL_TIMER,
++ &pcie_ack_rpl_reg);
++ current_replay = EFX_DWORD_FIELD(pcie_ack_rpl_reg, PCIE_CORE_RPL_TL);
++ current_ack_timer = EFX_DWORD_FIELD(pcie_ack_rpl_reg,
++ PCIE_CORE_ACK_TL);
++
++ lut_index = ffs(num_lanes) - 1;
++ expected_replay = tlp_ack_factor->replay[lut_index & 0x3];
++
++ falcon_pcie_core_read_reg(efx, PCIE_CORE_ADDR_ACK_FREQ,
++ &pcie_ack_freq_reg);
++ current_ack_freq = EFX_DWORD_FIELD(pcie_ack_freq_reg,
++ PCIE_CORE_ACK_FREQ);
++
++ EFX_LOG(efx, "pcie x%d tlp=%d replay_reg=" EFX_DWORD_FMT " { ack=%d "
++ "current_replay=%d expected_replay=%d } ack_reg="
++ EFX_DWORD_FMT " { current_freq=%d expected_freq=%d }\n",
++ num_lanes, tlp_size_decoded,
++ EFX_DWORD_VAL(pcie_ack_rpl_reg), current_ack_timer,
++ current_replay, expected_replay,
++ EFX_DWORD_VAL(pcie_ack_rpl_reg), current_ack_freq, 0);
++
++ /* If expected replay setting needs to be bigger then set it */
++ if (expected_replay > current_replay) {
++ EFX_SET_DWORD_FIELD(pcie_ack_rpl_reg, PCIE_CORE_RPL_TL,
++ expected_replay);
++
++ falcon_pcie_core_write_reg(efx, PCIE_CORE_ADDR_ACK_RPL_TIMER,
++ pcie_ack_rpl_reg);
++ }
++}
++
++static int falcon_init_pcie_core(struct efx_nic *efx)
++{
++ int pciecap_offset;
++ unsigned num_lanes = 0;
++
++ /* Get num lanes */
++ pciecap_offset = pci_find_capability(efx->pci_dev, PCI_CAP_ID_EXP);
++ if (pciecap_offset) {
++ u16 pcie_linkstat;
++ int rc, link_sta;
++
++ link_sta = pciecap_offset + PCI_EXP_LNKSTA;
++ rc = pci_read_config_word(efx->pci_dev, link_sta,
++ &pcie_linkstat);
++ if (rc)
++ return rc;
+
-+ if (ret)
-+ continue;
++ num_lanes = ((pcie_linkstat & PCI_EXP_LNKSTA_LNK_WID)
++ >> PCI_EXP_LNKSTA_LNK_WID_LBN);
++ EFX_BUG_ON_PARANOID(num_lanes <= 0 || num_lanes > 8);
+
-+ offset = (uvaddr - info->vma->vm_start) >> PAGE_SHIFT;
-+ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
-+ ((struct page **)info->vma->vm_private_data)[offset] =
-+ pg;
-+ }
++ if (num_lanes < 8)
++ EFX_ERR(efx, "WARNING: the Solarflare Network Adapter "
++ "has been plugged into a PCI-Express slot with "
++ "less than 8 lanes (%d detected). This will "
++ "limit the maximum achievable bandwidth! "
++ "Consult your motherboard documentation to "
++ "find a slot that is 8 lanes electrically and "
++ "physically\n", num_lanes);
+ }
+
-+ if (ret)
-+ goto fail_flush;
++ if (FALCON_REV(efx) <= FALCON_REV_A1)
++ return 0;
+
-+ if (xen_feature(XENFEAT_auto_translated_physmap))
-+ down_write(&info->vma->vm_mm->mmap_sem);
-+ /* Mark mapped pages as reserved: */
-+ for (i = 0; i < req->nr_segments; i++) {
-+ unsigned long kvaddr;
-+ struct page *pg;
++ if (EFX_WORKAROUND_6943(efx) && num_lanes > 0)
++ falcon_init_ack_repl_timer(efx, num_lanes);
+
-+ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
-+ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
-+ SetPageReserved(pg);
-+ if (xen_feature(XENFEAT_auto_translated_physmap)) {
-+ ret = vm_insert_page(info->vma,
-+ MMAP_VADDR(info->user_vstart,
-+ usr_idx, i), pg);
-+ if (ret) {
-+ up_write(&info->vma->vm_mm->mmap_sem);
-+ goto fail_flush;
-+ }
-+ }
++ if (EFX_WORKAROUND_9096(efx)) {
++ efx_dword_t pcie_ack_freq_reg;
++
++ /* ensure ack freq timer is 0 = always ack after timeout */
++ falcon_pcie_core_read_reg(efx, PCIE_CORE_ADDR_ACK_FREQ,
++ &pcie_ack_freq_reg);
++ EFX_SET_DWORD_FIELD(pcie_ack_freq_reg, PCIE_CORE_ACK_FREQ, 0);
++ falcon_pcie_core_write_reg(efx, PCIE_CORE_ADDR_ACK_FREQ,
++ pcie_ack_freq_reg);
+ }
-+ if (xen_feature(XENFEAT_auto_translated_physmap))
-+ up_write(&info->vma->vm_mm->mmap_sem);
-+
-+ /*record [mmap_idx,pending_idx] to [usr_idx] mapping*/
-+ info->idx_map[usr_idx] = MAKE_ID(mmap_idx, pending_idx);
+
-+ blkif_get(blkif);
-+ /* Finally, write the request message to the user ring. */
-+ target = RING_GET_REQUEST(&info->ufe_ring,
-+ info->ufe_ring.req_prod_pvt);
-+ memcpy(target, req, sizeof(*req));
-+ target->id = usr_idx;
-+ wmb(); /* blktap_poll() reads req_prod_pvt asynchronously */
-+ info->ufe_ring.req_prod_pvt++;
++ return 0;
++}
+
-+ if (operation == READ)
-+ blkif->st_rd_sect += nr_sects;
-+ else if (operation == WRITE)
-+ blkif->st_wr_sect += nr_sects;
++static void falcon_fini_pcie_core(struct efx_nic *efx)
++{
++ efx_dword_t pcie_ack_freq_reg;
+
-+ return;
++ if (FALCON_REV(efx) <= FALCON_REV_A1)
++ return;
+
-+ fail_flush:
-+ WPRINTK("Reached Fail_flush\n");
-+ fast_flush_area(pending_req, pending_idx, usr_idx, blkif->dev_num);
-+ fail_response:
-+ make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
-+ free_req(pending_req);
-+}
++ if (EFX_WORKAROUND_9096(efx)) {
++ /* Set the ACK frequency timer to 1, so TLP's are acked in
++ * a timely fashion.
++ */
++ falcon_pcie_core_read_reg(efx, PCIE_CORE_ADDR_ACK_FREQ,
++ &pcie_ack_freq_reg);
++ EFX_SET_DWORD_FIELD(pcie_ack_freq_reg, PCIE_CORE_ACK_FREQ, 1);
++ falcon_pcie_core_write_reg(efx, PCIE_CORE_ADDR_ACK_FREQ,
++ pcie_ack_freq_reg);
++ }
++}
+
++/* This call performs hardware-specific global initialisation, such as
++ * defining the descriptor cache sizes and number of RSS channels.
++ * It does not set up any buffers, descriptor rings or event queues.
++ */
++int falcon_init_nic(struct efx_nic *efx)
++{
++ struct falcon_nic_data *data;
++ efx_oword_t temp;
++ unsigned thresh;
++ int rc;
+
++ data = (struct falcon_nic_data *)efx->nic_data;
+
-+/******************************************************************
-+ * MISCELLANEOUS SETUP / TEARDOWN / DEBUGGING
-+ */
++ /* Set up the address region register. This is only needed
++ * for the B0 FPGA, but since we are just pushing in the
++ * reset defaults this may as well be unconditional. */
++ EFX_POPULATE_OWORD_4(temp, ADR_REGION0, 0,
++ ADR_REGION1, (1 << 16),
++ ADR_REGION2, (2 << 16),
++ ADR_REGION3, (3 << 16));
++ falcon_write(efx, &temp, ADR_REGION_REG_KER);
+
++ /* Use on-chip SRAM if needed.
++ */
++ falcon_read(efx, &temp, NIC_STAT_REG);
++ if (efx->external_sram_cfg == SRM_NB_BSZ_ONCHIP_ONLY)
++ EFX_SET_OWORD_FIELD(temp, ONCHIP_SRAM, 1);
++ else
++ EFX_SET_OWORD_FIELD(temp, ONCHIP_SRAM, 0);
++ falcon_write(efx, &temp, NIC_STAT_REG);
+
-+static void make_response(blkif_t *blkif, u64 id,
-+ unsigned short op, int st)
-+{
-+ blkif_response_t resp;
-+ unsigned long flags;
-+ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
-+ int more_to_do = 0;
-+ int notify;
++ /* Check power requirements against PCIe power budgeting */
++ rc = falcon_check_power_limit(efx);
++ if (rc)
++ return rc;
+
-+ resp.id = id;
-+ resp.operation = op;
-+ resp.status = st;
++ /* Warn if <8 lanes of PCIe detected & set pcie timers */
++ rc = falcon_init_pcie_core(efx);
++ if (rc)
++ return rc;
+
-+ spin_lock_irqsave(&blkif->blk_ring_lock, flags);
-+ /* Place on the response ring for the relevant domain. */
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->native,
-+ blk_rings->native.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ case BLKIF_PROTOCOL_X86_32:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->x86_32,
-+ blk_rings->x86_32.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ case BLKIF_PROTOCOL_X86_64:
-+ memcpy(RING_GET_RESPONSE(&blk_rings->x86_64,
-+ blk_rings->x86_64.rsp_prod_pvt),
-+ &resp, sizeof(resp));
-+ break;
-+ default:
-+ BUG();
-+ }
-+ blk_rings->common.rsp_prod_pvt++;
-+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
++ /* Set buffer table mode */
++ EFX_POPULATE_OWORD_1(temp, BUF_TBL_MODE, BUF_TBL_MODE_FULL);
++ falcon_write(efx, &temp, BUF_TBL_CFG_REG_KER);
+
-+ if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) {
-+ /*
-+ * Tail check for pending requests. Allows frontend to avoid
-+ * notifications if requests are already in flight (lower
-+ * overheads and promotes batching).
-+ */
-+ RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
-+ } else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) {
-+ more_to_do = 1;
-+ }
++ rc = falcon_reset_sram(efx);
++ if (rc)
++ return rc;
+
-+ spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
-+ if (more_to_do)
-+ blkif_notify_work(blkif);
-+ if (notify)
-+ notify_remote_via_irq(blkif->irq);
-+}
++ /* Set positions of descriptor caches in SRAM. */
++ EFX_POPULATE_OWORD_1(temp, SRM_TX_DC_BASE_ADR, data->tx_dc_base / 8);
++ falcon_write(efx, &temp, SRM_TX_DC_CFG_REG_KER);
++ EFX_POPULATE_OWORD_1(temp, SRM_RX_DC_BASE_ADR, data->rx_dc_base / 8);
++ falcon_write(efx, &temp, SRM_RX_DC_CFG_REG_KER);
+
-+static int __init blkif_init(void)
-+{
-+ int i, ret;
-+ struct class *class;
++ /* Set TX descriptor cache size. */
++ EFX_POPULATE_OWORD_1(temp, TX_DC_SIZE, ffs(data->tx_dc_entries) - 4);
++ falcon_write(efx, &temp, TX_DC_CFG_REG_KER);
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ /* Set RX descriptor cache size. Set low watermark to size-8, as
++ * this allows most efficient prefetching.
++ */
++ EFX_POPULATE_OWORD_1(temp, RX_DC_SIZE, ffs(data->rx_dc_entries) - 4);
++ falcon_write(efx, &temp, RX_DC_CFG_REG_KER);
++ EFX_POPULATE_OWORD_1(temp, RX_DC_PF_LWM, data->rx_dc_entries - 8);
++ falcon_write(efx, &temp, RX_DC_PF_WM_REG_KER);
+
-+ INIT_LIST_HEAD(&pending_free);
-+ for(i = 0; i < 2; i++) {
-+ ret = req_increase();
-+ if (ret)
-+ break;
++ /* Clear the parity enables on the TX data fifos as
++ * they produce false parity errors because of timing issues
++ */
++ if (EFX_WORKAROUND_5129(efx)) {
++ falcon_read(efx, &temp, SPARE_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, MEM_PERR_EN_TX_DATA, 0);
++ falcon_write(efx, &temp, SPARE_REG_KER);
+ }
-+ if (i == 0)
-+ return ret;
+
-+ tap_blkif_interface_init();
-+
-+ alloc_pending_reqs = 0;
++ /* Enable all the genuinely fatal interrupts. (They are still
++ * masked by the overall interrupt mask, controlled by
++ * falcon_interrupts()).
++ *
++ * Note: All other fatal interrupts are enabled
++ */
++ EFX_POPULATE_OWORD_3(temp,
++ ILL_ADR_INT_KER_EN, 1,
++ RBUF_OWN_INT_KER_EN, 1,
++ TBUF_OWN_INT_KER_EN, 1);
++ EFX_INVERT_OWORD(temp);
++ falcon_write(efx, &temp, FATAL_INTR_REG_KER);
++
++ /* Set number of RSS queues for receive path. */
++ falcon_read(efx, &temp, RX_FILTER_CTL_REG);
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ EFX_SET_OWORD_FIELD(temp, NUM_KER, 0);
++ else
++ EFX_SET_OWORD_FIELD(temp, NUM_KER, efx->rss_queues - 1);
++ if (EFX_WORKAROUND_7244(efx)) {
++ EFX_SET_OWORD_FIELD(temp, UDP_FULL_SRCH_LIMIT, 8);
++ EFX_SET_OWORD_FIELD(temp, UDP_WILD_SRCH_LIMIT, 8);
++ EFX_SET_OWORD_FIELD(temp, TCP_FULL_SRCH_LIMIT, 8);
++ EFX_SET_OWORD_FIELD(temp, TCP_WILD_SRCH_LIMIT, 8);
++ }
++ falcon_write(efx, &temp, RX_FILTER_CTL_REG);
++
++ falcon_setup_rss_indir_table(efx);
++
++ /* Setup RX. Wait for descriptor is broken and must
++ * be disabled. RXDP recovery shouldn't be needed, but is.
++ */
++ falcon_read(efx, &temp, RX_SELF_RST_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, RX_NODESC_WAIT_DIS, 1);
++ EFX_SET_OWORD_FIELD(temp, RX_RECOVERY_EN, 1);
++ if (EFX_WORKAROUND_5583(efx))
++ EFX_SET_OWORD_FIELD(temp, RX_ISCSI_DIS, 1);
++ falcon_write(efx, &temp, RX_SELF_RST_REG_KER);
++
++ /* Disable the ugly timer-based TX DMA backoff and allow TX DMA to be
++ * controlled by the RX FIFO fill level. Set arbitration to one pkt/Q.
++ */
++ falcon_read(efx, &temp, TX_CFG2_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, TX_RX_SPACER, 0xfe);
++ EFX_SET_OWORD_FIELD(temp, TX_RX_SPACER_EN, 1);
++ EFX_SET_OWORD_FIELD(temp, TX_ONE_PKT_PER_Q, 1);
++ EFX_SET_OWORD_FIELD(temp, TX_CSR_PUSH_EN, 0);
++ EFX_SET_OWORD_FIELD(temp, TX_DIS_NON_IP_EV, 1);
++ /* Enable SW_EV to inherit in char driver - assume harmless here */
++ EFX_SET_OWORD_FIELD(temp, TX_SW_EV_EN, 1);
++ /* Prefetch threshold 2 => fetch when descriptor cache half empty */
++ EFX_SET_OWORD_FIELD(temp, TX_PREF_THRESHOLD, 2);
++ if (EFX_WORKAROUND_9008(efx))
++ EFX_SET_OWORD_FIELD(temp, TX_PREF_WD_TMR, (unsigned)0x3fffff);
++ /* Squash TX of packets of 16 bytes or less */
++ if (FALCON_REV(efx) >= FALCON_REV_B0 && EFX_WORKAROUND_9141(efx))
++ EFX_SET_OWORD_FIELD(temp, TX_FLUSH_MIN_LEN_EN_B0, 1);
++ falcon_write(efx, &temp, TX_CFG2_REG_KER);
++
++ /* Do not enable TX_NO_EOP_DISC_EN, since it limits packets to 16
++ * descriptors (which is bad).
++ */
++ falcon_read(efx, &temp, TX_CFG_REG_KER);
++ EFX_SET_OWORD_FIELD(temp, TX_NO_EOP_DISC_EN, 0);
++ falcon_write(efx, &temp, TX_CFG_REG_KER);
++
++ /* RX config */
++ falcon_read(efx, &temp, RX_CFG_REG_KER);
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_DESC_PUSH_EN, 0);
++ if (EFX_WORKAROUND_7575(efx))
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_USR_BUF_SIZE,
++ (3 * 4096) / 32);
++ if (FALCON_REV(efx) >= FALCON_REV_B0)
++ EFX_SET_OWORD_FIELD(temp, RX_INGR_EN_B0, 1);
++
++ /* RX FIFO flow control thresholds */
++ thresh = ((rx_xon_thresh_bytes >= 0) ?
++ rx_xon_thresh_bytes : efx->type->rx_xon_thresh);
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_XON_MAC_TH, thresh / 256);
++ thresh = ((rx_xoff_thresh_bytes >= 0) ?
++ rx_xoff_thresh_bytes : efx->type->rx_xoff_thresh);
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_XOFF_MAC_TH, thresh / 256);
++ /* RX control FIFO thresholds [32 entries] */
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_XON_TX_TH, 25);
++ EFX_SET_OWORD_FIELD_VER(efx, temp, RX_XOFF_TX_TH, 20);
++ falcon_write(efx, &temp, RX_CFG_REG_KER);
++
++ /* Set destination of both TX and RX Flush events */
++ if (FALCON_REV(efx) >= FALCON_REV_B0) {
++ EFX_POPULATE_OWORD_1(temp, FLS_EVQ_ID, 0);
++ falcon_write(efx, &temp, DP_CTRL_REG);
++ }
+
-+ tap_blkif_xenbus_init();
++ return 0;
++}
+
-+ /* Dynamically allocate a major for this device */
-+ ret = register_chrdev(0, "blktap", &blktap_fops);
++void falcon_fini_nic(struct efx_nic *efx)
++{
++ falcon_fini_pcie_core(efx);
++}
+
-+ if (ret < 0) {
-+ WPRINTK("Couldn't register /dev/xen/blktap\n");
-+ return -ENOMEM;
-+ }
-+
-+ blktap_major = ret;
++void falcon_remove_nic(struct efx_nic *efx)
++{
++ /* Tear down the private nic state, and the driverlink nic params */
++ kfree(efx->nic_data);
++ efx->nic_data = efx->dl_info = NULL;
+
-+ /* tapfds[0] is always NULL */
-+ blktap_next_minor++;
++ falcon_remove_spi_devices(efx);
++ falcon_free_buffer(efx, &efx->irq_status);
+
-+ DPRINTK("Created misc_dev [/dev/xen/blktap%d]\n",i);
++ /* Reset the NIC finally */
++ (void) falcon_reset_hw(efx, RESET_TYPE_ALL);
+
-+ /* Make sure the xen class exists */
-+ if ((class = get_xen_class()) != NULL) {
-+ /*
-+ * This will allow udev to create the blktap ctrl device.
-+ * We only want to create blktap0 first. We don't want
-+ * to flood the sysfs system with needless blktap devices.
-+ * We only create the device when a request of a new device is
-+ * made.
-+ */
-+ class_device_create(class, NULL,
-+ MKDEV(blktap_major, 0), NULL,
-+ "blktap0");
-+ } else {
-+ /* this is bad, but not fatal */
-+ WPRINTK("blktap: sysfs xen_class not created\n");
++ /* Release the second function after the reset */
++ if (efx->pci_dev2) {
++ pci_dev_put(efx->pci_dev2);
++ efx->pci_dev2 = NULL;
+ }
++}
+
-+ DPRINTK("Blktap device successfully created\n");
++void falcon_update_nic_stats(struct efx_nic *efx)
++{
++ efx_oword_t cnt;
+
-+ return 0;
++ /* Read the RX drop counter */
++ falcon_read(efx, &cnt, RX_NODESC_DROP_REG_KER);
++ efx->n_rx_nodesc_drop_cnt += EFX_OWORD_FIELD(cnt, RX_NODESC_DROP_CNT);
+}
+
-+module_init(blkif_init);
-+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blktap/common.h linux-2.6.18-xen.hg/drivers/xen/blktap/common.h
---- linux-2.6.18/drivers/xen/blktap/common.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blktap/common.h 2007-12-23 11:15:33.547908998 +0100
-@@ -0,0 +1,121 @@
-+/*
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/**************************************************************************
++ *
++ * Revision-dependent attributes used by efx.c
++ *
++ **************************************************************************
+ */
+
-+#ifndef __BLKIF__BACKEND__COMMON_H__
-+#define __BLKIF__BACKEND__COMMON_H__
++struct efx_nic_type falcon_a_nic_type = {
++ .is_dual_func = 1,
++ .mem_bar = 2,
++ .mem_map_size = 0x20000,
++ .txd_ptr_tbl_base = TX_DESC_PTR_TBL_KER_A1,
++ .rxd_ptr_tbl_base = RX_DESC_PTR_TBL_KER_A1,
++ .buf_tbl_base = BUF_TBL_KER_A1,
++ .evq_ptr_tbl_base = EVQ_PTR_TBL_KER_A1,
++ .evq_rptr_tbl_base = EVQ_RPTR_REG_KER_A1,
++ .txd_ring_mask = FALCON_TXD_RING_MASK,
++ .rxd_ring_mask = FALCON_RXD_RING_MASK,
++ .evq_size = FALCON_EVQ_SIZE,
++ .max_dma_mask = FALCON_DMA_MASK,
++ .tx_dma_mask = FALCON_TX_DMA_MASK,
++ .bug5391_mask = 0xf,
++ .rx_xoff_thresh = 2048,
++ .rx_xon_thresh = 512,
++ .rx_buffer_padding = 0x24,
++ .max_interrupt_mode = EFX_INT_MODE_MSI,
++ .phys_addr_channels = 4,
++};
+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/interrupt.h>
-+#include <linux/slab.h>
-+#include <linux/blkdev.h>
-+#include <linux/vmalloc.h>
-+#include <asm/io.h>
-+#include <asm/setup.h>
-+#include <asm/pgalloc.h>
-+#include <xen/evtchn.h>
-+#include <asm/hypervisor.h>
-+#include <xen/blkif.h>
-+#include <xen/gnttab.h>
-+#include <xen/driver_util.h>
++struct efx_nic_type falcon_b_nic_type = {
++ .is_dual_func = 0,
++ .mem_bar = 2,
++ /* Map everything up to and including the RSS indirection
++ * table. Don't map MSI-X table, MSI-X PBA since Linux
++ * requires that they not be mapped. */
++ .mem_map_size = RX_RSS_INDIR_TBL_B0 + 0x800,
++ .txd_ptr_tbl_base = TX_DESC_PTR_TBL_KER_B0,
++ .rxd_ptr_tbl_base = RX_DESC_PTR_TBL_KER_B0,
++ .buf_tbl_base = BUF_TBL_KER_B0,
++ .evq_ptr_tbl_base = EVQ_PTR_TBL_KER_B0,
++ .evq_rptr_tbl_base = EVQ_RPTR_REG_KER_B0,
++ .txd_ring_mask = FALCON_TXD_RING_MASK,
++ .rxd_ring_mask = FALCON_RXD_RING_MASK,
++ .evq_size = FALCON_EVQ_SIZE,
++ .max_dma_mask = FALCON_DMA_MASK,
++ .tx_dma_mask = FALCON_TX_DMA_MASK,
++ .bug5391_mask = 0,
++ .rx_xoff_thresh = 54272, /* ~80Kb - 3*max MTU */
++ .rx_xon_thresh = 27648, /* ~3*max MTU */
++ .rx_buffer_padding = 0,
++ .max_interrupt_mode = EFX_INT_MODE_MSIX,
++ .phys_addr_channels = 32, /* Hardware limit is 64, but the legacy
++ * interrupt handler only supports 32
++ * channels */
+
-+#define DPRINTK(_f, _a...) pr_debug("(file=%s, line=%d) " _f, \
-+ __FILE__ , __LINE__ , ## _a )
++};
+
-+#define WPRINTK(fmt, args...) printk(KERN_WARNING "blk_tap: " fmt, ##args)
+--- linux-2.6.18.8/drivers/net/sfc/falcon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon.h 2008-05-19 00:33:28.845808833 +0300
+@@ -0,0 +1,177 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+struct backend_info;
++#ifndef EFX_FALCON_H
++#define EFX_FALCON_H
+
-+typedef struct blkif_st {
-+ /* Unique identifier for this interface. */
-+ domid_t domid;
-+ unsigned int handle;
-+ /* Physical parameters of the comms window. */
-+ unsigned int irq;
-+ /* Comms information. */
-+ enum blkif_protocol blk_protocol;
-+ blkif_back_rings_t blk_rings;
-+ struct vm_struct *blk_ring_area;
-+ /* Back pointer to the backend_info. */
-+ struct backend_info *be;
-+ /* Private fields. */
-+ spinlock_t blk_ring_lock;
-+ atomic_t refcnt;
++#include <asm/io.h>
++#include <linux/spinlock.h>
++#include "net_driver.h"
+
-+ wait_queue_head_t wq;
-+ struct task_struct *xenblkd;
-+ unsigned int waiting_reqs;
-+ request_queue_t *plug;
++/*
++ * Falcon hardware control
++ */
+
-+ /* statistics */
-+ unsigned long st_print;
-+ int st_rd_req;
-+ int st_wr_req;
-+ int st_oo_req;
-+ int st_rd_sect;
-+ int st_wr_sect;
++enum falcon_revision {
++ FALCON_REV_A0 = 0,
++ FALCON_REV_A1 = 1,
++ FALCON_REV_B0 = 2,
++};
+
-+ wait_queue_head_t waiting_to_free;
++#if defined(EFX_USE_PCI_DEV_REVISION)
++#define FALCON_REV(efx) ((efx)->pci_dev->revision)
++#else
++#define FALCON_REV(efx) ((efx)->revision)
++#endif
+
-+ grant_handle_t shmem_handle;
-+ grant_ref_t shmem_ref;
-+
-+ int dev_num;
-+ uint64_t sectors;
-+} blkif_t;
++extern struct efx_nic_type falcon_a_nic_type;
++extern struct efx_nic_type falcon_b_nic_type;
+
-+blkif_t *tap_alloc_blkif(domid_t domid);
-+void tap_blkif_free(blkif_t *blkif);
-+int tap_blkif_map(blkif_t *blkif, unsigned long shared_page,
-+ unsigned int evtchn);
-+void tap_blkif_unmap(blkif_t *blkif);
++/**************************************************************************
++ *
++ * Externs
++ *
++ **************************************************************************
++ */
+
-+#define blkif_get(_b) (atomic_inc(&(_b)->refcnt))
-+#define blkif_put(_b) \
-+ do { \
-+ if (atomic_dec_and_test(&(_b)->refcnt)) \
-+ wake_up(&(_b)->waiting_to_free);\
-+ } while (0)
++/* TX data path */
++extern int falcon_probe_tx(struct efx_tx_queue *tx_queue);
++extern int falcon_init_tx(struct efx_tx_queue *tx_queue);
++extern void falcon_fini_tx(struct efx_tx_queue *tx_queue);
++extern void falcon_remove_tx(struct efx_tx_queue *tx_queue);
++#if defined(EFX_USE_FASTCALL)
++extern void fastcall falcon_push_buffers(struct efx_tx_queue *tx_queue);
++#else
++extern void falcon_push_buffers(struct efx_tx_queue *tx_queue);
++#endif
+
++/* RX data path */
++extern int falcon_probe_rx(struct efx_rx_queue *rx_queue);
++extern int falcon_init_rx(struct efx_rx_queue *rx_queue);
++extern void falcon_fini_rx(struct efx_rx_queue *rx_queue);
++extern void falcon_remove_rx(struct efx_rx_queue *rx_queue);
++#if defined(EFX_USE_FASTCALL)
++extern void fastcall falcon_notify_rx_desc(struct efx_rx_queue *rx_queue);
++#else
++extern void falcon_notify_rx_desc(struct efx_rx_queue *rx_queue);
++#endif
+
-+struct phys_req {
-+ unsigned short dev;
-+ unsigned short nr_sects;
-+ struct block_device *bdev;
-+ blkif_sector_t sector_number;
-+};
++/* Event data path */
++extern int falcon_probe_eventq(struct efx_channel *channel);
++extern int falcon_init_eventq(struct efx_channel *channel);
++extern void falcon_fini_eventq(struct efx_channel *channel);
++extern void falcon_remove_eventq(struct efx_channel *channel);
++#if defined(EFX_USE_FASTCALL)
++extern int fastcall falcon_process_eventq(struct efx_channel *channel,
++ int *rx_quota);
++#else
++extern int falcon_process_eventq(struct efx_channel *channel, int *rx_quota);
++#endif
++#if defined(EFX_USE_FASTCALL)
++extern void fastcall falcon_eventq_read_ack(struct efx_channel *channel);
++#else
++extern void falcon_eventq_read_ack(struct efx_channel *channel);
++#endif
++
++/* Ports */
++extern int falcon_probe_port(struct efx_nic *efx);
++extern void falcon_remove_port(struct efx_nic *efx);
++
++/* MAC/PHY */
++extern void falcon_check_xaui_link_up(struct efx_nic *efx);
++extern int falcon_xaui_link_ok(struct efx_nic *efx);
++extern int falcon_dma_stats(struct efx_nic *efx,
++ unsigned int done_offset);
++extern void falcon_drain_tx_fifo(struct efx_nic *efx);
++extern void falcon_deconfigure_mac_wrapper(struct efx_nic *efx);
++extern void falcon_reconfigure_mac_wrapper(struct efx_nic *efx);
++
++/* Interrupts and test events */
++extern int falcon_init_interrupt(struct efx_nic *efx);
++extern void falcon_enable_interrupts(struct efx_nic *efx);
++extern void falcon_generate_test_event(struct efx_channel *channel,
++ unsigned int magic);
++extern void falcon_generate_interrupt(struct efx_nic *efx);
++extern void falcon_set_int_moderation(struct efx_channel *channel);
++extern void falcon_disable_interrupts(struct efx_nic *efx);
++extern void falcon_fini_interrupt(struct efx_nic *efx);
++
++/* Global Resources */
++extern int falcon_probe_nic(struct efx_nic *efx);
++extern int falcon_probe_resources(struct efx_nic *efx);
++extern int falcon_init_nic(struct efx_nic *efx);
++extern int falcon_reset_hw(struct efx_nic *efx, enum reset_type method);
++extern void falcon_fini_nic(struct efx_nic *efx);
++extern void falcon_remove_resources(struct efx_nic *efx);
++extern void falcon_remove_nic(struct efx_nic *efx);
++extern void falcon_update_nic_stats(struct efx_nic *efx);
++extern void falcon_set_multicast_hash(struct efx_nic *efx);
++extern int falcon_reset_xaui(struct efx_nic *efx);
++
++/**************************************************************************
++ *
++ * Falcon MAC stats
++ *
++ **************************************************************************
++ */
++
++#define FALCON_STAT_OFFSET(falcon_stat) EFX_VAL(falcon_stat, offset)
++#define FALCON_STAT_WIDTH(falcon_stat) EFX_VAL(falcon_stat, WIDTH)
++
++/* Retrieve statistic from statistics block */
++#define FALCON_STAT(efx, falcon_stat, efx_stat) do { \
++ if (FALCON_STAT_WIDTH(falcon_stat) == 16) \
++ (efx)->mac_stats.efx_stat += le16_to_cpu( \
++ *((__force __le16 *) \
++ (efx->stats_buffer.addr + \
++ FALCON_STAT_OFFSET(falcon_stat)))); \
++ else if (FALCON_STAT_WIDTH(falcon_stat) == 32) \
++ (efx)->mac_stats.efx_stat += le32_to_cpu( \
++ *((__force __le32 *) \
++ (efx->stats_buffer.addr + \
++ FALCON_STAT_OFFSET(falcon_stat)))); \
++ else \
++ (efx)->mac_stats.efx_stat += le64_to_cpu( \
++ *((__force __le64 *) \
++ (efx->stats_buffer.addr + \
++ FALCON_STAT_OFFSET(falcon_stat)))); \
++ } while (0)
+
-+void tap_blkif_interface_init(void);
++#define FALCON_MAC_STATS_SIZE 0x100
+
-+void tap_blkif_xenbus_init(void);
++#define MAC_DATA_LBN 0
++#define MAC_DATA_WIDTH 32
+
-+irqreturn_t tap_blkif_be_int(int irq, void *dev_id, struct pt_regs *regs);
-+int tap_blkif_schedule(void *arg);
++extern void falcon_generate_event(struct efx_channel *channel,
++ efx_qword_t *event);
+
-+int dom_to_devid(domid_t domid, int xenbus_id, blkif_t *blkif);
-+void signal_tapdisk(int idx);
++#ifdef CONFIG_SFC_DEBUGFS
++struct seq_file;
++extern int falcon_debugfs_read_hardware_desc(struct seq_file *file, void *data);
++#endif
+
-+#endif /* __BLKIF__BACKEND__COMMON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blktap/interface.c linux-2.6.18-xen.hg/drivers/xen/blktap/interface.c
---- linux-2.6.18/drivers/xen/blktap/interface.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blktap/interface.c 2007-12-23 11:15:33.547908998 +0100
-@@ -0,0 +1,174 @@
-+/******************************************************************************
-+ * drivers/xen/blktap/interface.c
-+ *
-+ * Block-device interface management.
-+ *
-+ * Copyright (c) 2004, Keir Fraser
++#endif /* EFX_FALCON_H */
+--- linux-2.6.18.8/drivers/net/sfc/falcon_gmac.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon_gmac.c 2008-05-19 00:33:28.845808833 +0300
+@@ -0,0 +1,320 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include "common.h"
-+#include <xen/evtchn.h>
++#include <linux/delay.h>
++#include "net_driver.h"
++#include "efx.h"
++#include "falcon.h"
++#include "mac.h"
++#include "falcon_hwdefs.h"
++#include "falcon_io.h"
++#include "gmii.h"
+
-+static kmem_cache_t *blkif_cachep;
++/**************************************************************************
++ *
++ * MAC register access
++ *
++ **************************************************************************/
+
-+blkif_t *tap_alloc_blkif(domid_t domid)
++/* Offset of a GMAC register within Falcon */
++#define FALCON_GMAC_REG(mac_reg) \
++ (FALCON_GMAC_REGBANK + ((mac_reg) * FALCON_GMAC_REG_SIZE))
++
++static void falcon_gmac_writel(struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg)
+{
-+ blkif_t *blkif;
++ efx_oword_t temp;
+
-+ blkif = kmem_cache_alloc(blkif_cachep, GFP_KERNEL);
-+ if (!blkif)
-+ return ERR_PTR(-ENOMEM);
++ EFX_POPULATE_OWORD_1(temp, MAC_DATA, EFX_DWORD_FIELD(*value, MAC_DATA));
++ falcon_write(efx, &temp, FALCON_GMAC_REG(mac_reg));
++}
+
-+ memset(blkif, 0, sizeof(*blkif));
-+ blkif->domid = domid;
-+ spin_lock_init(&blkif->blk_ring_lock);
-+ atomic_set(&blkif->refcnt, 1);
-+ init_waitqueue_head(&blkif->wq);
-+ blkif->st_print = jiffies;
-+ init_waitqueue_head(&blkif->waiting_to_free);
++static void falcon_gmac_readl(struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg)
++{
++ efx_oword_t temp;
+
-+ return blkif;
++ falcon_read(efx, &temp, FALCON_GMAC_REG(mac_reg));
++ EFX_POPULATE_DWORD_1(*value, MAC_DATA, EFX_OWORD_FIELD(temp, MAC_DATA));
+}
+
-+static int map_frontend_page(blkif_t *blkif, unsigned long shared_page)
++/**************************************************************************
++ *
++ * MAC operations
++ *
++ *************************************************************************/
++
++static int falcon_init_gmac(struct efx_nic *efx)
+{
-+ struct gnttab_map_grant_ref op;
++ int rc;
+
-+ gnttab_set_map_op(&op, (unsigned long)blkif->blk_ring_area->addr,
-+ GNTMAP_host_map, shared_page, blkif->domid);
++ /* Reset the MAC */
++ mentormac_reset(efx);
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
-+ BUG();
++ /* Initialise PHY */
++ rc = efx->phy_op->init(efx);
++ if (rc)
++ return rc;
+
-+ if (op.status) {
-+ DPRINTK(" Grant table operation failure !\n");
-+ return op.status;
-+ }
++ return 0;
++}
+
-+ blkif->shmem_ref = shared_page;
-+ blkif->shmem_handle = op.handle;
++static void falcon_reconfigure_gmac(struct efx_nic *efx)
++{
++ /* Reconfigure PHY and pick up PHY parameters. This updates
++ * the link status. */
++ efx->phy_op->reconfigure(efx);
+
-+ return 0;
++ /* Isolate the MAC. */
++ falcon_deconfigure_mac_wrapper(efx);
++
++ /* Reconfigure MAC */
++ mentormac_reconfigure(efx);
++
++ /* Reconfigure MAC wrapper */
++ falcon_reconfigure_mac_wrapper(efx);
+}
+
-+static void unmap_frontend_page(blkif_t *blkif)
++static void falcon_fini_gmac(struct efx_nic *efx)
+{
-+ struct gnttab_unmap_grant_ref op;
++ /* Isolate the MAC - PHY */
++ falcon_deconfigure_mac_wrapper(efx);
+
-+ gnttab_set_unmap_op(&op, (unsigned long)blkif->blk_ring_area->addr,
-+ GNTMAP_host_map, blkif->shmem_handle);
++ /* Shut down PHY */
++ efx->phy_op->fini(efx);
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
-+ BUG();
++ /* Reset MAC */
++ mentormac_reset(efx);
+}
+
-+int tap_blkif_map(blkif_t *blkif, unsigned long shared_page,
-+ unsigned int evtchn)
++static void falcon_update_stats_gmac(struct efx_nic *efx)
+{
-+ int err;
++ struct efx_mac_stats *mac_stats = &efx->mac_stats;
++ unsigned long old_rx_pause, old_tx_pause;
++ unsigned long new_rx_pause, new_tx_pause;
++ int rc;
+
-+ /* Already connected through? */
-+ if (blkif->irq)
-+ return 0;
++ rc = falcon_dma_stats(efx, GDmaDone_offset);
++ if (rc)
++ return;
+
-+ if ( (blkif->blk_ring_area = alloc_vm_area(PAGE_SIZE)) == NULL )
-+ return -ENOMEM;
++ /* Pause frames are erroneously counted as errors (SFC bug 3269) */
++ old_rx_pause = mac_stats->rx_pause;
++ old_tx_pause = mac_stats->tx_pause;
++
++ /* Update MAC stats from DMAed values */
++ FALCON_STAT(efx, GRxGoodOct, rx_good_bytes);
++ FALCON_STAT(efx, GRxBadOct, rx_bad_bytes);
++ FALCON_STAT(efx, GRxMissPkt, rx_missed);
++ FALCON_STAT(efx, GRxFalseCRS, rx_false_carrier);
++ FALCON_STAT(efx, GRxPausePkt, rx_pause);
++ FALCON_STAT(efx, GRxBadPkt, rx_bad);
++ FALCON_STAT(efx, GRxUcastPkt, rx_unicast);
++ FALCON_STAT(efx, GRxMcastPkt, rx_multicast);
++ FALCON_STAT(efx, GRxBcastPkt, rx_broadcast);
++ FALCON_STAT(efx, GRxGoodLt64Pkt, rx_good_lt64);
++ FALCON_STAT(efx, GRxBadLt64Pkt, rx_bad_lt64);
++ FALCON_STAT(efx, GRx64Pkt, rx_64);
++ FALCON_STAT(efx, GRx65to127Pkt, rx_65_to_127);
++ FALCON_STAT(efx, GRx128to255Pkt, rx_128_to_255);
++ FALCON_STAT(efx, GRx256to511Pkt, rx_256_to_511);
++ FALCON_STAT(efx, GRx512to1023Pkt, rx_512_to_1023);
++ FALCON_STAT(efx, GRx1024to15xxPkt, rx_1024_to_15xx);
++ FALCON_STAT(efx, GRx15xxtoJumboPkt, rx_15xx_to_jumbo);
++ FALCON_STAT(efx, GRxGtJumboPkt, rx_gtjumbo);
++ FALCON_STAT(efx, GRxFcsErr64to15xxPkt, rx_bad_64_to_15xx);
++ FALCON_STAT(efx, GRxFcsErr15xxtoJumboPkt, rx_bad_15xx_to_jumbo);
++ FALCON_STAT(efx, GRxFcsErrGtJumboPkt, rx_bad_gtjumbo);
++ FALCON_STAT(efx, GTxGoodBadOct, tx_bytes);
++ FALCON_STAT(efx, GTxGoodOct, tx_good_bytes);
++ FALCON_STAT(efx, GTxSglColPkt, tx_single_collision);
++ FALCON_STAT(efx, GTxMultColPkt, tx_multiple_collision);
++ FALCON_STAT(efx, GTxExColPkt, tx_excessive_collision);
++ FALCON_STAT(efx, GTxDefPkt, tx_deferred);
++ FALCON_STAT(efx, GTxLateCol, tx_late_collision);
++ FALCON_STAT(efx, GTxExDefPkt, tx_excessive_deferred);
++ FALCON_STAT(efx, GTxPausePkt, tx_pause);
++ FALCON_STAT(efx, GTxBadPkt, tx_bad);
++ FALCON_STAT(efx, GTxUcastPkt, tx_unicast);
++ FALCON_STAT(efx, GTxMcastPkt, tx_multicast);
++ FALCON_STAT(efx, GTxBcastPkt, tx_broadcast);
++ FALCON_STAT(efx, GTxLt64Pkt, tx_lt64);
++ FALCON_STAT(efx, GTx64Pkt, tx_64);
++ FALCON_STAT(efx, GTx65to127Pkt, tx_65_to_127);
++ FALCON_STAT(efx, GTx128to255Pkt, tx_128_to_255);
++ FALCON_STAT(efx, GTx256to511Pkt, tx_256_to_511);
++ FALCON_STAT(efx, GTx512to1023Pkt, tx_512_to_1023);
++ FALCON_STAT(efx, GTx1024to15xxPkt, tx_1024_to_15xx);
++ FALCON_STAT(efx, GTx15xxtoJumboPkt, tx_15xx_to_jumbo);
++ FALCON_STAT(efx, GTxGtJumboPkt, tx_gtjumbo);
++ FALCON_STAT(efx, GTxNonTcpUdpPkt, tx_non_tcpudp);
++ FALCON_STAT(efx, GTxMacSrcErrPkt, tx_mac_src_error);
++ FALCON_STAT(efx, GTxIpSrcErrPkt, tx_ip_src_error);
++
++ /* Pause frames are erroneously counted as errors (SFC bug 3269) */
++ new_rx_pause = mac_stats->rx_pause;
++ new_tx_pause = mac_stats->tx_pause;
++ mac_stats->rx_bad -= (new_rx_pause - old_rx_pause);
++ mac_stats->tx_bad -= (new_tx_pause - old_tx_pause);
++
++ /* Derive stats that the MAC doesn't provide directly */
++ mac_stats->tx_bad_bytes =
++ mac_stats->tx_bytes - mac_stats->tx_good_bytes;
++ mac_stats->tx_packets =
++ mac_stats->tx_lt64 + mac_stats->tx_64 +
++ mac_stats->tx_65_to_127 + mac_stats->tx_128_to_255 +
++ mac_stats->tx_256_to_511 + mac_stats->tx_512_to_1023 +
++ mac_stats->tx_1024_to_15xx + mac_stats->tx_15xx_to_jumbo +
++ mac_stats->tx_gtjumbo;
++ mac_stats->tx_collision =
++ mac_stats->tx_single_collision +
++ mac_stats->tx_multiple_collision +
++ mac_stats->tx_excessive_collision +
++ mac_stats->tx_late_collision;
++ mac_stats->rx_bytes =
++ mac_stats->rx_good_bytes + mac_stats->rx_bad_bytes;
++ mac_stats->rx_packets =
++ mac_stats->rx_good_lt64 + mac_stats->rx_bad_lt64 +
++ mac_stats->rx_64 + mac_stats->rx_65_to_127 +
++ mac_stats->rx_128_to_255 + mac_stats->rx_256_to_511 +
++ mac_stats->rx_512_to_1023 + mac_stats->rx_1024_to_15xx +
++ mac_stats->rx_15xx_to_jumbo + mac_stats->rx_gtjumbo;
++ mac_stats->rx_good = mac_stats->rx_packets - mac_stats->rx_bad;
++ mac_stats->rx_lt64 = mac_stats->rx_good_lt64 + mac_stats->rx_bad_lt64;
++}
++
++static int falcon_check_gmac(struct efx_nic *efx)
++{
++ /* Nothing to do */
++ return 0;
++}
+
-+ err = map_frontend_page(blkif, shared_page);
-+ if (err) {
-+ free_vm_area(blkif->blk_ring_area);
-+ return err;
-+ }
++static void falcon_gmac_sim_phy_event(struct efx_nic *efx)
++{
++ efx_qword_t phy_event;
+
-+ switch (blkif->blk_protocol) {
-+ case BLKIF_PROTOCOL_NATIVE:
-+ {
-+ blkif_sring_t *sring;
-+ sring = (blkif_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.native, sring, PAGE_SIZE);
-+ break;
-+ }
-+ case BLKIF_PROTOCOL_X86_32:
-+ {
-+ blkif_x86_32_sring_t *sring_x86_32;
-+ sring_x86_32 = (blkif_x86_32_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.x86_32, sring_x86_32, PAGE_SIZE);
-+ break;
-+ }
-+ case BLKIF_PROTOCOL_X86_64:
-+ {
-+ blkif_x86_64_sring_t *sring_x86_64;
-+ sring_x86_64 = (blkif_x86_64_sring_t *)blkif->blk_ring_area->addr;
-+ BACK_RING_INIT(&blkif->blk_rings.x86_64, sring_x86_64, PAGE_SIZE);
-+ break;
-+ }
-+ default:
-+ BUG();
-+ }
++ EFX_POPULATE_QWORD_2(phy_event,
++ EV_CODE, GLOBAL_EV_DECODE,
++ G_PHY0_INTR, 1);
++ falcon_generate_event(&efx->channel[0], &phy_event);
++}
+
-+ err = bind_interdomain_evtchn_to_irqhandler(
-+ blkif->domid, evtchn, tap_blkif_be_int,
-+ 0, "blkif-backend", blkif);
-+ if (err < 0) {
-+ unmap_frontend_page(blkif);
-+ free_vm_area(blkif->blk_ring_area);
-+ blkif->blk_rings.common.sring = NULL;
-+ return err;
++static void falcon_gmac_reset_phy(struct efx_nic *efx)
++{
++ struct mii_if_info *gmii = &efx->mii;
++ int bmcr, i;
++
++ /* Perform software reset to make new settings take effect */
++ bmcr = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_BMCR);
++ bmcr |= BMCR_RESET;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, MII_BMCR, bmcr);
++
++ /* Wait for the reset to deassert */
++ for (i = 20; i; --i) {
++ udelay(10);
++ if ((gmii->mdio_read(gmii->dev, gmii->phy_id, MII_BMCR) &
++ BMCR_RESET) == 0)
++ return;
+ }
-+ blkif->irq = err;
+
-+ return 0;
++ EFX_ERR(efx, "wait for PHY reset timed out\n");
+}
+
-+void tap_blkif_unmap(blkif_t *blkif)
++
++static int falcon_gmac_get_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
+{
-+ if (blkif->irq) {
-+ unbind_from_irqhandler(blkif->irq, blkif);
-+ blkif->irq = 0;
-+ }
-+ if (blkif->blk_rings.common.sring) {
-+ unmap_frontend_page(blkif);
-+ free_vm_area(blkif->blk_ring_area);
-+ blkif->blk_rings.common.sring = NULL;
-+ }
++ struct mii_if_info *gmii = &efx->mii;
++ int rc;
++
++ rc = mii_ethtool_gset(gmii, ecmd);
++ ecmd->supported &= ~(SUPPORTED_1000baseT_Half);
++ return rc;
+}
+
-+void tap_blkif_free(blkif_t *blkif)
++static int falcon_gmac_set_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
+{
-+ atomic_dec(&blkif->refcnt);
-+ wait_event(blkif->waiting_to_free, atomic_read(&blkif->refcnt) == 0);
++ struct mii_if_info *gmii = &efx->mii;
++ int rc;
+
-+ tap_blkif_unmap(blkif);
-+ kmem_cache_free(blkif_cachep, blkif);
++ /* 1000Mbps half-duplex is technically legal, but none of our
++ * current hardware supports it, so just disallow it. */
++ if (ecmd->speed == SPEED_1000 && ecmd->duplex != DUPLEX_FULL) {
++ EFX_LOG(efx, "rejecting unsupported 1000Mbps HD"
++ " setting\n");
++ return -EINVAL;
++ }
++
++ /* Use MII to set all other settings */
++ rc = mii_ethtool_sset(gmii, ecmd);
++ if (rc)
++ return rc;
++
++ /* Reset the PHY */
++ falcon_gmac_reset_phy(efx);
++
++ return 0;
+}
+
-+void __init tap_blkif_interface_init(void)
++static int falcon_gmac_set_pause(struct efx_nic *efx,
++ enum efx_fc_type flow_control)
+{
-+ blkif_cachep = kmem_cache_create("blktapif_cache", sizeof(blkif_t),
-+ 0, 0, NULL, NULL);
++ struct mii_if_info *gmii = &efx->mii;
++ int adv;
++
++ /* GMAC has tiny MAC FIFO, so TX flow control won't work */
++ if (flow_control & EFX_FC_TX)
++ return -EINVAL;
++
++ efx->flow_control = flow_control;
++
++ /* Push autonegotiation to PHY */
++ adv = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_ADVERTISE);
++ adv &= ~GM_ADVERTISE_PAUSE_CAP;
++ adv |= (flow_control & EFX_FC_AUTO) ? GM_ADVERTISE_PAUSE_CAP : 0;
++ gmii->mdio_write(gmii->dev, gmii->phy_id, MII_ADVERTISE, adv);
++
++ falcon_gmac_reset_phy(efx);
++
++ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blktap/Makefile linux-2.6.18-xen.hg/drivers/xen/blktap/Makefile
---- linux-2.6.18/drivers/xen/blktap/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blktap/Makefile 2007-12-23 11:15:33.544575492 +0100
-@@ -0,0 +1,5 @@
-+LINUXINCLUDE += -I../xen/include/public/io
+
-+obj-$(CONFIG_XEN_BLKDEV_TAP) := xenblktap.o
+
-+xenblktap-y := xenbus.o interface.o blktap.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/blktap/xenbus.c linux-2.6.18-xen.hg/drivers/xen/blktap/xenbus.c
---- linux-2.6.18/drivers/xen/blktap/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/blktap/xenbus.c 2007-12-23 11:15:33.547908998 +0100
-@@ -0,0 +1,477 @@
-+/* drivers/xen/blktap/xenbus.c
++struct efx_mac_operations falcon_gmac_operations = {
++ .mac_writel = falcon_gmac_writel,
++ .mac_readl = falcon_gmac_readl,
++ .init = falcon_init_gmac,
++ .reconfigure = falcon_reconfigure_gmac,
++ .update_stats = falcon_update_stats_gmac,
++ .fini = falcon_fini_gmac,
++ .check_hw = falcon_check_gmac,
++ .fake_phy_event = falcon_gmac_sim_phy_event,
++ .get_settings = falcon_gmac_get_settings,
++ .set_settings = falcon_gmac_set_settings,
++ .set_pause = falcon_gmac_set_pause,
++};
+--- linux-2.6.18.8/drivers/net/sfc/falcon_hwdefs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon_hwdefs.h 2008-05-19 00:33:28.849809063 +0300
+@@ -0,0 +1,1620 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * Xenbus code for blktap
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * Copyright (c) 2004-2005, Andrew Warfield and Julian Chesterfield
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
+ *
-+ * Based on the blkback xenbus code:
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
-+ * Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
-+ * Copyright (C) 2005 XenSource Ltd
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_FALCON_HWDEFS_H
++#define EFX_FALCON_HWDEFS_H
++
++/*
++ * Falcon hardware value definitions.
++ * Falcon is the internal codename for the SFC4000 controller that is
++ * present in SFE400X evaluation boards
++ */
++
++/**************************************************************************
++ *
++ * Falcon registers
++ *
++ **************************************************************************
++ */
++
++/* Address region register */
++#define ADR_REGION_REG_KER 0x00
++#define ADR_REGION0_LBN 0
++#define ADR_REGION0_WIDTH 18
++#define ADR_REGION1_LBN 32
++#define ADR_REGION1_WIDTH 18
++#define ADR_REGION2_LBN 64
++#define ADR_REGION2_WIDTH 18
++#define ADR_REGION3_LBN 96
++#define ADR_REGION3_WIDTH 18
++
++/* Interrupt enable register */
++#define INT_EN_REG_KER 0x0010
++#define INT_LEVEL_SEL_LBN 8
++#define INT_LEVEL_SEL_WIDTH 6
++#define MEM_PERR_INT_EN_KER_LBN 5
++#define MEM_PERR_INT_EN_KER_WIDTH 1
++#define KER_INT_CHAR_LBN 4
++#define KER_INT_CHAR_WIDTH 1
++#define KER_INT_KER_LBN 3
++#define KER_INT_KER_WIDTH 1
++#define ILL_ADR_ERR_INT_EN_KER_LBN 2
++#define ILL_ADR_ERR_INT_EN_KER_WIDTH 1
++#define SRM_PERR_INT_EN_KER_LBN 1
++#define SRM_PERR_INT_EN_KER_WIDTH 1
++#define DRV_INT_EN_KER_LBN 0
++#define DRV_INT_EN_KER_WIDTH 1
++
++/* Interrupt status address register */
++#define INT_ADR_REG_KER 0x0030
++#define NORM_INT_VEC_DIS_KER_LBN 64
++#define NORM_INT_VEC_DIS_KER_WIDTH 1
++#define INT_ADR_KER_LBN 0
++#define INT_ADR_KER_WIDTH EFX_DMA_TYPE_WIDTH(64) /* not 46 for this one */
++
++/* Interrupt status register (B0 only) */
++#define INT_ISR0_B0 0x90
++#define INT_ISR1_B0 0xA0
++
++/* Interrupt acknowledge register (A0/A1 only) */
++#define INT_ACK_REG_KER_A1 0x0050
++#define INT_ACK_DUMMY_DATA_LBN 0
++#define INT_ACK_DUMMY_DATA_WIDTH 32
++
++/* Interrupt acknowledge work-around register (A0/A1 only )*/
++#define WORK_AROUND_BROKEN_PCI_READS_REG_KER_A1 0x0070
++
++/* Hardware initialisation register */
++#define HW_INIT_REG_KER 0x00c0
++#define BCSR_TARGET_MASK_LBN 101
++#define BCSR_TARGET_MASK_WIDTH 4
++#define PCIE_TIMEOUT_DIS_LBN 123
++#define PCIE_TIMEOUT_DIS_WIDTH 1
++#define B2B_REQ_EN_B0_LBN 45
++#define B2B_REQ_EN_B0_WIDTH 1
++#define FC_BLOCKING_EN_B0_LBN 44
++#define FC_BLOCKING_EN_B0_WIDTH 1
++
++/* SPI host command register */
++#define EE_SPI_HCMD_REG_KER 0x0100
++#define EE_SPI_HCMD_CMD_EN_LBN 31
++#define EE_SPI_HCMD_CMD_EN_WIDTH 1
++#define EE_WR_TIMER_ACTIVE_LBN 28
++#define EE_WR_TIMER_ACTIVE_WIDTH 1
++#define EE_SPI_HCMD_SF_SEL_LBN 24
++#define EE_SPI_HCMD_SF_SEL_WIDTH 1
++#define EE_SPI_EEPROM 0
++#define EE_SPI_FLASH 1
++#define EE_SPI_HCMD_DABCNT_LBN 16
++#define EE_SPI_HCMD_DABCNT_WIDTH 5
++#define EE_SPI_HCMD_READ_LBN 15
++#define EE_SPI_HCMD_READ_WIDTH 1
++#define EE_SPI_READ 1
++#define EE_SPI_WRITE 0
++#define EE_SPI_HCMD_DUBCNT_LBN 12
++#define EE_SPI_HCMD_DUBCNT_WIDTH 2
++#define EE_SPI_HCMD_ADBCNT_LBN 8
++#define EE_SPI_HCMD_ADBCNT_WIDTH 2
++#define EE_SPI_HCMD_ENC_LBN 0
++#define EE_SPI_HCMD_ENC_WIDTH 8
++
++/* SPI host address register */
++#define EE_SPI_HADR_REG_KER 0x0110
++#define EE_SPI_HADR_DUBYTE_LBN 24
++#define EE_SPI_HADR_DUBYTE_WIDTH 8
++#define EE_SPI_HADR_ADR_LBN 0
++#define EE_SPI_HADR_ADR_WIDTH 24
++
++/* SPI host data register */
++#define EE_SPI_HDATA_REG_KER 0x0120
++#define EE_SPI_HDATA3_LBN 96
++#define EE_SPI_HDATA3_WIDTH 32
++#define EE_SPI_HDATA2_LBN 64
++#define EE_SPI_HDATA2_WIDTH 32
++#define EE_SPI_HDATA1_LBN 32
++#define EE_SPI_HDATA1_WIDTH 32
++#define EE_SPI_HDATA0_LBN 0
++#define EE_SPI_HDATA0_WIDTH 32
++
++/* SPI/VPD config register */
++#define EE_VPD_CFG_REG_KER 0x0140
++#define EE_VPD_EN_LBN 0
++#define EE_VPD_EN_WIDTH 1
++#define EE_VPD_EN_AD9_MODE_LBN 1
++#define EE_VPD_EN_AD9_MODE_WIDTH 1
++#define EE_EE_CLOCK_DIV_LBN 112
++#define EE_EE_CLOCK_DIV_WIDTH 7
++#define EE_SF_CLOCK_DIV_LBN 120
++#define EE_SF_CLOCK_DIV_WIDTH 7
++
++/* PCIE CORE ACCESS REG */
++#define PCIE_CORE_INDIRECT_REG 0x01f0
++#define PCIE_CORE_ADDR_LBN 0
++#define PCIE_CORE_ADDR_WIDTH 12
++#define PCIE_CORE_RW_LBN 15
++#define PCIE_CORE_RW_WIDTH 1
++#define PCIE_CORE_VALUE_LBN 32
++#define PCIE_CORE_VALUE_WIDTH 32
++
++#define PCIE_CORE_ADDR_PCIE_DEVICE_CTRL_STAT 0x68
++#define PCIE_CORE_ADDR_PCIE_LINK_CTRL_STAT 0x70
++#define PCIE_CORE_ADDR_ACK_RPL_TIMER 0x700
++#define PCIE_CORE_ACK_TL_LBN 0
++#define PCIE_CORE_ACK_TL_WIDTH 16
++#define PCIE_CORE_RPL_TL_LBN 16
++#define PCIE_CORE_RPL_TL_WIDTH 16
++
++#define PCIE_CORE_ADDR_ACK_FREQ 0x70C
++#define PCIE_CORE_ACK_FREQ_LBN 0
++#define PCIE_CORE_ACK_FREQ_WIDTH 7
++
++
++/* NIC status register */
++#define NIC_STAT_REG 0x0200
++#define EE_STRAP_EN_LBN 31
++#define EE_STRAP_EN_WIDTH 1
++#define EE_STRAP_OVR_LBN 24
++#define EE_STRAP_OVR_WIDTH 4
++#define ONCHIP_SRAM_LBN 16
++#define ONCHIP_SRAM_WIDTH 1
++#define SF_PRST_LBN 9
++#define SF_PRST_WIDTH 1
++#define EE_PRST_LBN 8
++#define EE_PRST_WIDTH 1
++#define EE_STRAP_LBN 7
++#define EE_STRAP_WIDTH 1
++/* See pic_mode_t for decoding of this field */
++#define STRAP_ISCSI_EN_LBN 3
++#define STRAP_ISCSI_EN_WIDTH 1
++#define STRAP_PINS_LBN 0
++#define STRAP_PINS_WIDTH 3
++/* These bit definitions are extrapolated from the list of numerical
++ * values for STRAP_PINS.
++ */
++#define STRAP_10G_LBN 2
++#define STRAP_10G_WIDTH 1
++#define STRAP_DUAL_PORT_LBN 1
++#define STRAP_DUAL_PORT_WIDTH 1
++#define STRAP_PCIE_LBN 0
++#define STRAP_PCIE_WIDTH 1
++
++#define FLASH_PRESENT_LBN 7
++#define FLASH_PRESENT_WIDTH 1
++#define EEPROM_PRESENT_LBN 6
++#define EEPROM_PRESENT_WIDTH 1
++#define BOOTED_USING_NVDEVICE_LBN 3
++#define BOOTED_USING_NVDEVICE_WIDTH 1
++
++/* GPIO control register */
++
++#define GPIO_CTL_REG_KER 0x0210
++
++#define GPIO_USE_NIC_CLK_LBN (30)
++#define GPIO_USE_NIC_CLK_WIDTH (1)
++
++#define GPIO_OUTPUTS_LBN (16)
++#define GPIO_OUTPUTS_WIDTH (4)
++
++#define GPIO_INPUTS_LBN (8)
++#define GPIO_INPUT_WIDTH (4)
++
++#define GPIO_DIRECTION_LBN (24)
++#define GPIO_DIRECTION_WIDTH (4)
++#define GPIO_DIRECTION_OUT (1)
++#define GPIO_SRAM_SLEEP (1 << 1)
++
++#define GPIO3_OEN_LBN (GPIO_DIRECTION_LBN + 3)
++#define GPIO3_OEN_WIDTH 1
++#define GPIO2_OEN_LBN (GPIO_DIRECTION_LBN + 2)
++#define GPIO2_OEN_WIDTH 1
++#define GPIO1_OEN_LBN (GPIO_DIRECTION_LBN + 1)
++#define GPIO1_OEN_WIDTH 1
++#define GPIO0_OEN_LBN (GPIO_DIRECTION_LBN + 0)
++#define GPIO0_OEN_WIDTH 1
++
++#define GPIO3_OUT_LBN (GPIO_OUTPUTS_LBN + 3)
++#define GPIO3_OUT_WIDTH 1
++#define GPIO2_OUT_LBN (GPIO_OUTPUTS_LBN + 2)
++#define GPIO2_OUT_WIDTH 1
++#define GPIO1_OUT_LBN (GPIO_OUTPUTS_LBN + 1)
++#define GPIO1_OUT_WIDTH 1
++#define GPIO0_OUT_LBN (GPIO_OUTPUTS_LBN + 0)
++#define GPIO0_OUT_WIDTH 1
++
++#define GPIO3_IN_LBN (GPIO_INPUTS_LBN + 3)
++#define GPIO3_IN_WIDTH 1
++#define GPIO2_IN_LBN (GPIO_INPUTS_LBN + 2)
++#define GPIO2_IN_WIDTH 1
++#define GPIO1_IN_LBN (GPIO_INPUTS_LBN + 1)
++#define GPIO1_IN_WIDTH 1
++#define GPIO0_IN_LBN (GPIO_INPUTS_LBN + 0)
++#define GPIO0_IN_WIDTH 1
++
++/* Global control register */
++#define GLB_CTL_REG_KER 0x0220
++#define EXT_PHY_RST_CTL_LBN 63
++#define EXT_PHY_RST_CTL_WIDTH 1
++#define PCIE_SD_RST_CTL_LBN 61
++#define PCIE_SD_RST_CTL_WIDTH 1
++#define PCIX_RST_CTL_LBN 60
++#define PCIX_RST_CTL_WIDTH 1
++#define PCIE_STCK_RST_CTL_LBN 59
++
++#define PCIE_STCK_RST_CTL_WIDTH 1
++#define PCIE_NSTCK_RST_CTL_LBN 58
++#define PCIE_NSTCK_RST_CTL_WIDTH 1
++#define PCIE_CORE_RST_CTL_LBN 57
++#define PCIE_CORE_RST_CTL_WIDTH 1
++#define EE_RST_CTL_LBN 49
++#define EE_RST_CTL_WIDTH 1
++#define CS_RST_CTL_LBN 48
++#define CS_RST_CTL_WIDTH 1
++#define RST_EXT_PHY_LBN 31
++#define RST_EXT_PHY_WIDTH 1
++#define RST_XGRX_LBN 24
++#define RST_XGRX_WIDTH 1
++#define RST_XGTX_LBN 23
++#define RST_XGTX_WIDTH 1
++#define RST_EM_LBN 22
++#define RST_EM_WIDTH 1
++#define INT_RST_DUR_LBN 4
++#define INT_RST_DUR_WIDTH 3
++#define EXT_PHY_RST_DUR_LBN 1
++#define EXT_PHY_RST_DUR_WIDTH 3
++#define SWRST_LBN 0
++#define SWRST_WIDTH 1
++#define INCLUDE_IN_RESET 0
++#define EXCLUDE_FROM_RESET 1
++
++/* Fatal interrupt register */
++#define FATAL_INTR_REG_KER 0x0230
++#define PCI_BUSERR_INT_KER_EN_LBN 43
++#define PCI_BUSERR_INT_KER_EN_WIDTH 1
++#define SRAM_OOB_INT_KER_EN_LBN 42
++#define SRAM_OOB_INT_KER_EN_WIDTH 1
++#define BUFID_OOB_INT_KER_EN_LBN 41
++#define BUFID_OOB_INT_KER_EN_WIDTH 1
++#define MEM_PERR_INT_KER_EN_LBN 40
++#define MEM_PERR_INT_KER_EN_WIDTH 1
++#define RBUF_OWN_INT_KER_EN_LBN 39
++#define RBUF_OWN_INT_KER_EN_WIDTH 1
++#define TBUF_OWN_INT_KER_EN_LBN 38
++#define TBUF_OWN_INT_KER_EN_WIDTH 1
++#define RDESCQ_OWN_INT_KER_EN_LBN 37
++#define RDESCQ_OWN_INT_KER_EN_WIDTH 1
++#define TDESCQ_OWN_INT_KER_EN_LBN 36
++#define TDESCQ_OWN_INT_KER_EN_WIDTH 1
++#define EVQ_OWN_INT_KER_EN_LBN 35
++#define EVQ_OWN_INT_KER_EN_WIDTH 1
++#define EVFF_OFLO_INT_KER_EN_LBN 34
++#define EVFF_OFLO_INT_KER_EN_WIDTH 1
++#define ILL_ADR_INT_KER_EN_LBN 33
++#define ILL_ADR_INT_KER_EN_WIDTH 1
++#define SRM_PERR_INT_KER_EN_LBN 32
++#define SRM_PERR_INT_KER_EN_WIDTH 1
++#define MEM_PERR_INT_KER_LBN 8
++#define MEM_PERR_INT_KER_WIDTH 1
++#define INT_KER_ERROR_LBN 0
++#define INT_KER_ERROR_WIDTH 12
++
++#define DP_CTRL_REG 0x250
++#define FLS_EVQ_ID_LBN 0
++#define FLS_EVQ_ID_WIDTH 11
++
++#define MEM_STAT_REG_KER 0x260
++
++/* Debug probe register */
++#define DEBUG_REG_KER 0x0270
++#define DEBUG_BLK_SEL2_LBN 47
++#define DEBUG_BLK_SEL2_WIDTH 3
++#define DEBUG_BLK_SEL1_LBN 44
++#define DEBUG_BLK_SEL1_WIDTH 3
++#define DEBUG_BLK_SEL0_LBN 41
++#define DEBUG_BLK_SEL0_WIDTH 3
++#define DEBUG_BLK_SEL_MISC 7
++#define DEBUG_BLK_SEL_SERDES 6
++#define DEBUG_BLK_SEL_EM 5
++#define DEBUG_BLK_SEL_SR 4
++#define DEBUG_BLK_SEL_EV 3
++#define DEBUG_BLK_SEL_RX 2
++#define DEBUG_BLK_SEL_TX 1
++#define DEBUG_BLK_SEL_BIU 0
++#define MISC_DEBUG_ADDR_LBN 36
++#define MISC_DEBUG_ADDR_WIDTH 5
++#define SERDES_DEBUG_ADDR_LBN 31
++#define SERDES_DEBUG_ADDR_WIDTH 5
++#define EM_DEBUG_ADDR_LBN 26
++#define EM_DEBUG_ADDR_WIDTH 5
++#define SR_DEBUG_ADDR_LBN 21
++#define SR_DEBUG_ADDR_WIDTH 5
++#define EV_DEBUG_ADDR_LBN 16
++#define EV_DEBUG_ADDR_WIDTH 5
++#define RX_DEBUG_ADDR_LBN 11
++#define RX_DEBUG_ADDR_WIDTH 5
++#define TX_DEBUG_ADDR_LBN 5
++#define TX_DEBUG_ADDR_WIDTH 5
++#define BIU_DEBUG_ADDR_LBN 1
++#define BIU_DEBUG_ADDR_WIDTH 5
++#define DEBUG_DIS_LBN 0
++#define DEBUG_DIS_WIDTH 1
++
++/* Scratch register 0 */
++#define DRIVER_REG0_KER 0x0280
++#define DRIVER_DW0_LBN 0
++#define DRIVER_DW0_WIDTH 32
++
++/* Scratch register 1 */
++#define DRIVER_REG1_KER 0x0290
++#define DRIVER_DW1_LBN 0
++#define DRIVER_DW1_WIDTH 32
++
++/* Scratch register 2 */
++#define DRIVER_REG2_KER 0x02A0
++#define DRIVER_DW2_LBN 0
++#define DRIVER_DW2_WIDTH 32
++
++/* Scratch register 3 */
++#define DRIVER_REG3_KER 0x02B0
++#define DRIVER_DW3_LBN 0
++#define DRIVER_DW3_WIDTH 32
++
++/* Scratch register 4 */
++#define DRIVER_REG4_KER 0x02C0
++#define DRIVER_DW3_LBN 0
++#define DRIVER_DW3_WIDTH 32
++
++/* Scratch register 5 */
++#define DRIVER_REG5_KER 0x02D0
++#define DRIVER_DW3_LBN 0
++#define DRIVER_DW3_WIDTH 32
++
++/* Scratch register 6 */
++#define DRIVER_REG6_KER 0x02E0
++#define DRIVER_DW3_LBN 0
++#define DRIVER_DW3_WIDTH 32
++
++/* Scratch register 7 */
++#define DRIVER_REG7_KER 0x02F0
++#define DRIVER_DW3_LBN 0
++#define DRIVER_DW3_WIDTH 32
++
++/* FPGA build version */
++#define ALTERA_BUILD_REG_KER 0x0300
++#define VER_MAJOR_LBN 24
++#define VER_MAJOR_WIDTH 8
++#define VER_MINOR_LBN 16
++#define VER_MINOR_WIDTH 8
++#define VER_BUILD_LBN 0
++#define VER_BUILD_WIDTH 16
++#define VER_ALL_LBN 0
++#define VER_ALL_WIDTH 32
++
++/* Spare EEPROM bits register (flash 0x390) */
++#define SPARE_REG_KER 0x310
++#define MEM_PERR_EN_LBN 64
++#define MEM_PERR_EN_WIDTH 38
++#define MEM_PERR_EN_TX_DATA_LBN 72
++#define MEM_PERR_EN_TX_DATA_WIDTH 2
++#define SPARE_EE_BITS_LBN 1
++#define SPARE_EE_BITS_WIDTH 31
++#define PCIE_LEGACY_ENDPOINT_LBN 0
++#define PCIE_LEGACY_ENDPOINT_WIDTH 1
++
++/* Page mapped view of bottom 1024 EVQ RPTRS */
++#define EVQ_RPTR_REG_P0 0x400
++/* Bit definitions are as for the densely mapped
++ * RPTR registers. */
++
++/* Timer table for kernel access */
++#define TIMER_CMD_REG_KER 0x420
++#define TIMER_MODE_LBN 12
++#define TIMER_MODE_WIDTH 2
++#define TIMER_MODE_DIS 0
++#define TIMER_MODE_INT_HLDOFF 2
++#define TIMER_VAL_LBN 0
++#define TIMER_VAL_WIDTH 12
++
++/* Driver generated event register */
++#define DRV_EV_REG_KER 0x440
++#define DRV_EV_QID_LBN 64
++#define DRV_EV_QID_WIDTH 12
++#define DRV_EV_DATA_LBN 0
++#define DRV_EV_DATA_WIDTH 64
++
++/* Event Queue control register */
++#define EVQ_CTL_REG_KER 0x450
++#define EVQ_FIFO_NOTAF_TH_LBN 0
++#define EVQ_FIFO_NOTAF_TH_WIDTH 6
++#define EVQ_FIFO_AF_TH_LBN 8
++#define EVQ_FIFO_AF_TH_WIDTH 6
++
++/* Buffer table configuration register */
++#define BUF_TBL_CFG_REG_KER 0x600
++#define BUF_TBL_MODE_LBN 3
++#define BUF_TBL_MODE_WIDTH 1
++#define BUF_TBL_MODE_HALF 0
++#define BUF_TBL_MODE_FULL 1
++
++/* SRAM receive descriptor cache configuration register */
++#define SRM_RX_DC_CFG_REG_KER 0x610
++#define SRM_RX_DC_BASE_ADR_LBN 0
++#define SRM_RX_DC_BASE_ADR_WIDTH 21
++
++/* SRAM transmit descriptor cache configuration register */
++#define SRM_TX_DC_CFG_REG_KER 0x620
++#define SRM_TX_DC_BASE_ADR_LBN 0
++#define SRM_TX_DC_BASE_ADR_WIDTH 21
++
++/* SRAM configuration register */
++#define SRM_CFG_REG_KER 0x630
++#define SRAM_OOB_ADR_INTEN_LBN 5
++#define SRAM_OOB_ADR_INTEN_WIDTH 1
++#define SRAM_OOB_BUF_INTEN_LBN 4
++#define SRAM_OOB_BUF_INTEN_WIDTH 1
++#define SRAM_OOB_BT_INIT_EN_LBN 3
++#define SRAM_OOB_BT_INIT_EN_WIDTH 1
++#define SRM_NUM_BANK_LBN 2
++#define SRM_NUM_BANK_WIDTH 1
++#define SRM_BANK_SIZE_LBN 0
++#define SRM_BANK_SIZE_WIDTH 2
++#define SRM_NUM_BANKS_AND_BANK_SIZE_LBN 0
++#define SRM_NUM_BANKS_AND_BANK_SIZE_WIDTH 3
++#define SRM_NB_BSZ_1BANKS_2M 0
++#define SRM_NB_BSZ_1BANKS_4M 1
++#define SRM_NB_BSZ_1BANKS_8M 2
++#define SRM_NB_BSZ_DEFAULT 3 /* char driver will set the default */
++#define SRM_NB_BSZ_2BANKS_4M 4
++#define SRM_NB_BSZ_2BANKS_8M 5
++#define SRM_NB_BSZ_2BANKS_16M 6
++#define SRM_NB_BSZ_RESERVED 7
++
++/* Special buffer table update register */
++#define BUF_TBL_UPD_REG_KER 0x0650
++#define BUF_UPD_CMD_LBN 63
++#define BUF_UPD_CMD_WIDTH 1
++#define BUF_CLR_CMD_LBN 62
++#define BUF_CLR_CMD_WIDTH 1
++#define BUF_CLR_END_ID_LBN 32
++#define BUF_CLR_END_ID_WIDTH 20
++#define BUF_CLR_START_ID_LBN 0
++#define BUF_CLR_START_ID_WIDTH 20
++
++/* Receive configuration register */
++#define RX_CFG_REG_KER 0x800
++
++/* B0 */
++#define RX_TOEP_TCP_SUPPRESS_B0_LBN 48
++#define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++#define RX_INGR_EN_B0_LBN 47
++#define RX_INGR_EN_B0_WIDTH 1
++#define RX_TOEP_IPV4_B0_LBN 46
++#define RX_TOEP_IPV4_B0_WIDTH 1
++#define RX_HASH_ALG_B0_LBN 45
++#define RX_HASH_ALG_B0_WIDTH 1
++#define RX_HASH_INSERT_HDR_B0_LBN 44
++#define RX_HASH_INSERT_HDR_B0_WIDTH 1
++#define RX_DESC_PUSH_EN_B0_LBN 43
++#define RX_DESC_PUSH_EN_B0_WIDTH 1
++#define RX_RDW_PATCH_EN_LBN 42 /* Non head of line blocking */
++#define RX_RDW_PATCH_EN_WIDTH 1
++#define RX_PCI_BURST_SIZE_B0_LBN 39
++#define RX_PCI_BURST_SIZE_B0_WIDTH 3
++#define RX_OWNERR_CTL_B0_LBN 38
++#define RX_OWNERR_CTL_B0_WIDTH 1
++#define RX_XON_TX_TH_B0_LBN 33
++#define RX_XON_TX_TH_B0_WIDTH 5
++#define RX_XOFF_TX_TH_B0_LBN 28
++#define RX_XOFF_TX_TH_B0_WIDTH 5
++#define RX_USR_BUF_SIZE_B0_LBN 19
++#define RX_USR_BUF_SIZE_B0_WIDTH 9
++#define RX_XON_MAC_TH_B0_LBN 10
++#define RX_XON_MAC_TH_B0_WIDTH 9
++#define RX_XOFF_MAC_TH_B0_LBN 1
++#define RX_XOFF_MAC_TH_B0_WIDTH 9
++#define RX_XOFF_MAC_EN_B0_LBN 0
++#define RX_XOFF_MAC_EN_B0_WIDTH 1
++
++/* A1 */
++#define RX_DESC_PUSH_EN_A1_LBN 35
++#define RX_DESC_PUSH_EN_A1_WIDTH 1
++#define RX_PCI_BURST_SIZE_A1_LBN 31
++#define RX_PCI_BURST_SIZE_A1_WIDTH 3
++#define RX_OWNERR_CTL_A1_LBN 30
++#define RX_OWNERR_CTL_A1_WIDTH 1
++#define RX_XON_TX_TH_A1_LBN 25
++#define RX_XON_TX_TH_A1_WIDTH 5
++#define RX_XOFF_TX_TH_A1_LBN 20
++#define RX_XOFF_TX_TH_A1_WIDTH 5
++#define RX_USR_BUF_SIZE_A1_LBN 11
++#define RX_USR_BUF_SIZE_A1_WIDTH 9
++#define RX_XON_MAC_TH_A1_LBN 6
++#define RX_XON_MAC_TH_A1_WIDTH 5
++#define RX_XOFF_MAC_TH_A1_LBN 1
++#define RX_XOFF_MAC_TH_A1_WIDTH 5
++#define RX_XOFF_MAC_EN_A1_LBN 0
++#define RX_XOFF_MAC_EN_A1_WIDTH 1
++
++/* Receive filter control register */
++#define RX_FILTER_CTL_REG 0x810
++#define SCATTER_ENBL_NO_MATCH_Q_B0_LBN 40
++#define SCATTER_ENBL_NO_MATCH_Q_B0_WIDTH 1
++#define UDP_FULL_SRCH_LIMIT_LBN 32
++#define UDP_FULL_SRCH_LIMIT_WIDTH 8
++#define NUM_KER_LBN 24
++#define NUM_KER_WIDTH 2
++#define UDP_WILD_SRCH_LIMIT_LBN 16
++#define UDP_WILD_SRCH_LIMIT_WIDTH 8
++#define TCP_WILD_SRCH_LIMIT_LBN 8
++#define TCP_WILD_SRCH_LIMIT_WIDTH 8
++#define TCP_FULL_SRCH_LIMIT_LBN 0
++#define TCP_FULL_SRCH_LIMIT_WIDTH 8
++
++/* RX queue flush register */
++#define RX_FLUSH_DESCQ_REG_KER 0x0820
++#define RX_FLUSH_DESCQ_CMD_LBN 24
++#define RX_FLUSH_DESCQ_CMD_WIDTH 1
++#define RX_FLUSH_DESCQ_LBN 0
++#define RX_FLUSH_DESCQ_WIDTH 12
++
++/* Receive descriptor update register */
++#define RX_DESC_UPD_REG_KER 0x0830
++#define RX_DESC_WPTR_LBN 96
++#define RX_DESC_WPTR_WIDTH 12
++#define RX_DESC_UPD_REG_KER_DWORD (RX_DESC_UPD_REG_KER + 12)
++#define RX_DESC_WPTR_DWORD_LBN 0
++#define RX_DESC_WPTR_DWORD_WIDTH 12
++
++/* Receive descriptor cache configuration register */
++#define RX_DC_CFG_REG_KER 0x840
++#define RX_DC_SIZE_LBN 0
++#define RX_DC_SIZE_WIDTH 2
++
++#define RX_DC_PF_WM_REG_KER 0x850
++#define RX_DC_PF_LWM_LBN 0
++#define RX_DC_PF_LWM_WIDTH 6
++
++/* RX no descriptor drop counter */
++#define RX_NODESC_DROP_REG_KER 0x880
++#define RX_NODESC_DROP_CNT_LBN 0
++#define RX_NODESC_DROP_CNT_WIDTH 16
++
++/* RX black magic register */
++#define RX_SELF_RST_REG_KER 0x890
++#define RX_ISCSI_DIS_LBN 17
++#define RX_ISCSI_DIS_WIDTH 1
++#define RX_PREFETCH_TIMEOUT_EN_LBN 10
++#define RX_PREFETCH_TIMEOUT_EN_WIDTH 1
++#define RX_NODESC_WAIT_DIS_LBN 9
++#define RX_NODESC_WAIT_DIS_WIDTH 1
++#define RX_RECOVERY_EN_LBN 8
++#define RX_RECOVERY_EN_WIDTH 1
++#define RX_SHUTDOWN_REASON_LBN 0
++#define RX_SHUTDOWN_REASON_WIDTH (3)
++
++/* TX queue flush register */
++#define TX_FLUSH_DESCQ_REG_KER 0x0a00
++#define TX_FLUSH_DESCQ_CMD_LBN 12
++#define TX_FLUSH_DESCQ_CMD_WIDTH 1
++#define TX_FLUSH_DESCQ_LBN 0
++#define TX_FLUSH_DESCQ_WIDTH 12
++
++/* Transmit descriptor update register */
++#define TX_DESC_UPD_REG_KER 0x0a10
++#define TX_DESC_WPTR_LBN 96
++#define TX_DESC_WPTR_WIDTH 12
++#define TX_DESC_UPD_REG_KER_DWORD (TX_DESC_UPD_REG_KER + 12)
++#define TX_DESC_WPTR_DWORD_LBN 0
++#define TX_DESC_WPTR_DWORD_WIDTH 12
++
++/* Transmit descriptor cache configuration register */
++#define TX_DC_CFG_REG_KER 0xa20
++#define TX_DC_SIZE_LBN 0
++#define TX_DC_SIZE_WIDTH 2
++
++/* Transmit checksum configuration register (A0/A1 only) */
++#define TX_CHKSM_CFG_REG_KER_A1 0xa30
++
++/* Transmit configuration register */
++#define TX_CFG_REG_KER 0xa50
++#define TX_NO_EOP_DISC_EN_LBN 5
++#define TX_NO_EOP_DISC_EN_WIDTH 1
++
++/* Transmit configuration register 2 */
++#define TX_CFG2_REG_KER 0xa80
++#define TX_CSR_PUSH_EN_LBN 89
++#define TX_CSR_PUSH_EN_WIDTH 1
++#define TX_RX_SPACER_LBN 64
++#define TX_RX_SPACER_WIDTH 8
++#define TX_SW_EV_EN_LBN 59
++#define TX_SW_EV_EN_WIDTH 1
++#define TX_RX_SPACER_EN_LBN 57
++#define TX_RX_SPACER_EN_WIDTH 1
++#define TX_PREF_WD_TMR_LBN 22
++#define TX_PREF_WD_TMR_WIDTH 22
++#define TX_PREF_THRESHOLD_LBN 19
++#define TX_PREF_THRESHOLD_WIDTH 2
++#define TX_ONE_PKT_PER_Q_LBN 18
++#define TX_ONE_PKT_PER_Q_WIDTH 1
++#define TX_DIS_NON_IP_EV_LBN 17
++#define TX_DIS_NON_IP_EV_WIDTH 1
++#define TX_DMA_SPACER_LBN 8
++#define TX_DMA_SPACER_WIDTH 8
++#define TX_FLUSH_MIN_LEN_EN_B0_LBN 7
++#define TX_FLUSH_MIN_LEN_EN_B0_WIDTH 1
++#define TX_TCP_DIS_A1_LBN 7
++#define TX_TCP_DIS_A1_WIDTH 1
++#define TX_IP_DIS_A1_LBN 6
++#define TX_IP_DIS_A1_WIDTH 1
++#define TX_MAX_CPL_LBN 2
++#define TX_MAX_CPL_WIDTH 2
++#define TX_MAX_PREF_LBN 0
++#define TX_MAX_PREF_WIDTH 2
++
++/* Transmit VLAN filter control register */
++#define TX_VLAN_REG 0xae0
++
++/* PHY management transmit data register */
++#define MD_TXD_REG_KER 0xc00
++#define MD_TXD_LBN 0
++#define MD_TXD_WIDTH 16
++
++/* PHY management receive data register */
++#define MD_RXD_REG_KER 0xc10
++#define MD_RXD_LBN 0
++#define MD_RXD_WIDTH 16
++
++/* PHY management configuration & status register */
++#define MD_CS_REG_KER 0xc20
++#define MD_PT_LBN 7
++#define MD_PT_WIDTH 3
++#define MD_PL_LBN 6
++#define MD_PL_WIDTH 1
++#define MD_INT_CLR_LBN 5
++#define MD_INT_CLR_WIDTH 1
++#define MD_GC_LBN 4
++#define MD_GC_WIDTH 1
++#define MD_PRSP_LBN 3
++#define MD_PRSP_WIDTH 1
++#define MD_RIC_LBN 2
++#define MD_RIC_WIDTH 1
++#define MD_RDC_LBN 1
++#define MD_RDC_WIDTH 1
++#define MD_WRC_LBN 0
++#define MD_WRC_WIDTH 1
++
++/* PHY management PHY address register */
++#define MD_PHY_ADR_REG_KER 0xc30
++#define MD_PHY_ADR_LBN 0
++#define MD_PHY_ADR_WIDTH 16
++
++/* PHY management ID register */
++#define MD_ID_REG_KER 0xc40
++#define MD_PRT_ADR_LBN 11
++#define MD_PRT_ADR_WIDTH 5
++#define MD_DEV_ADR_LBN 6
++#define MD_DEV_ADR_WIDTH 5
++/* Used for writing both at once */
++#define MD_PRT_DEV_ADR_LBN 6
++#define MD_PRT_DEV_ADR_WIDTH 10
++
++/* PHY management status & mask register (DWORD read only) */
++#define MD_STAT_REG_KER 0xc50
++#define MD_PINT_LBN 4
++#define MD_PINT_WIDTH 1
++#define MD_DONE_LBN 3
++#define MD_DONE_WIDTH 1
++#define MD_BSERR_LBN 2
++#define MD_BSERR_WIDTH 1
++#define MD_LNFL_LBN 1
++#define MD_LNFL_WIDTH 1
++#define MD_BSY_LBN 0
++#define MD_BSY_WIDTH 1
++
++/* Port 0 and 1 MAC stats registers */
++#define MAC0_STAT_DMA_REG_KER 0xc60
++#define MAC1_STAT_DMA_REG_KER 0xc70
++#define MAC_STAT_DMA_CMD_LBN 48
++#define MAC_STAT_DMA_CMD_WIDTH 1
++#define MAC_STAT_DMA_REGION_LBN 46
++#define MAC_STAT_DMA_REGION_WIDTH 2
++#define MAC_STAT_DMA_ADR_LBN 0
++#define MAC_STAT_DMA_ADR_WIDTH EFX_DMA_TYPE_WIDTH(46)
++
++/* Port 0 and 1 MAC control registers */
++#define MAC0_CTRL_REG_KER 0xc80
++#define MAC1_CTRL_REG_KER 0xc90
++#define MAC_XOFF_VAL_LBN 16
++#define MAC_XOFF_VAL_WIDTH 16
++#define TXFIFO_DRAIN_EN_B0_LBN 7
++#define TXFIFO_DRAIN_EN_B0_WIDTH 1
++#define MAC_XG_DISTXCRC_LBN 5
++#define MAC_XG_DISTXCRC_WIDTH 1
++#define MAC_BCAD_ACPT_LBN 4
++#define MAC_BCAD_ACPT_WIDTH 1
++#define MAC_UC_PROM_LBN 3
++#define MAC_UC_PROM_WIDTH 1
++#define MAC_LINK_STATUS_LBN 2
++#define MAC_LINK_STATUS_WIDTH 1
++#define MAC_SPEED_LBN 0
++#define MAC_SPEED_WIDTH 2
++
++/* External interrupt control (replaces MAC1_CTRL in B0) */
++#define GEN_MODE_REG_KER 0xc90
++#define XFP_PHY_INT_POL_SEL_LBN 3
++#define XFP_PHY_INT_POL_SEL_WIDTH 1
++#define XG_PHY_INT_POL_SEL_LBN 2
++#define XG_PHY_INT_POL_SEL_WIDTH 1
++#define XFP_PHY_INT_MASK_LBN 1
++#define XFP_PHY_INT_MASK_WIDTH 1
++#define XG_PHY_INT_MASK_LBN 0
++#define XG_PHY_INT_MASK_WIDTH 1
++
++
++/* 10G XAUI XGXS default values */
++#define XX_TXDRV_DEQ_DEFAULT 0xe /* deq=.6 */
++#define XX_TXDRV_DTX_DEFAULT 0x5 /* 1.25 */
++#define XX_SD_CTL_DRV_DEFAULT 0 /* 20mA */
++
++/* Multicast address hash table */
++#define MAC_MCAST_HASH_REG0_KER 0xca0
++#define MAC_MCAST_HASH_REG1_KER 0xcb0
++
++/* MAC test register. */
++#define MAC_TEST_REG_KER 0xcc0
++#define MAC_PTLOOP_EN_LBN 0
++#define MAC_PTLOOP_EN_WIDTH 1
++
++/* GMAC registers */
++#define FALCON_GMAC_REGBANK 0xe00
++#define FALCON_GMAC_REGBANK_SIZE 0x200
++#define FALCON_GMAC_REG_SIZE 0x10
++
++/* Source MAC filter control register */
++#define TX_SRC_MAC_CTRL_REG 0x1100
++
++/* XMAC registers */
++#define FALCON_XMAC_REGBANK 0x1200
++#define FALCON_XMAC_REGBANK_SIZE 0x200
++#define FALCON_XMAC_REG_SIZE 0x10
++
++/* XGMAC address register low */
++#define XM_ADR_LO_REG_MAC 0x00
++#define XM_ADR_3_LBN 24
++#define XM_ADR_3_WIDTH 8
++#define XM_ADR_2_LBN 16
++#define XM_ADR_2_WIDTH 8
++#define XM_ADR_1_LBN 8
++#define XM_ADR_1_WIDTH 8
++#define XM_ADR_0_LBN 0
++#define XM_ADR_0_WIDTH 8
++
++/* XGMAC address register high */
++#define XM_ADR_HI_REG_MAC 0x01
++#define XM_ADR_5_LBN 8
++#define XM_ADR_5_WIDTH 8
++#define XM_ADR_4_LBN 0
++#define XM_ADR_4_WIDTH 8
++
++/* XGMAC global configuration */
++#define XM_GLB_CFG_REG_MAC 0x02
++#define XM_LINE_LB_DEEP_RSVD_LBN 28
++#define XM_LINE_LB_DEEP_RSVD_WIDTH 1
++#define XM_RMTFLT_GEN_LBN 17
++#define XM_RMTFLT_GEN_WIDTH 1
++#define XM_DEBUG_MODE_LBN 16
++#define XM_DEBUG_MODE_WIDTH 1
++#define XM_RX_STAT_EN_LBN 11
++#define XM_RX_STAT_EN_WIDTH 1
++#define XM_TX_STAT_EN_LBN 10
++#define XM_TX_STAT_EN_WIDTH 1
++#define XM_RX_JUMBO_MODE_LBN 6
++#define XM_RX_JUMBO_MODE_WIDTH 1
++#define XM_WAN_MODE_LBN 5
++#define XM_WAN_MODE_WIDTH 1
++#define XM_AUTOCLR_MODE_LBN 4
++#define XM_AUTOCLR_MODE_WIDTH 1
++#define XM_INTCLR_MODE_LBN 3
++#define XM_INTCLR_MODE_WIDTH 1
++#define XM_CORE_RST_LBN 0
++#define XM_CORE_RST_WIDTH 1
++
++/* XGMAC transmit configuration */
++#define XM_TX_CFG_REG_MAC 0x03
++#define XM_TX_PROG_LBN 24
++#define XM_TX_PROG_WIDTH 1
++#define XM_IPG_LBN 16
++#define XM_IPG_WIDTH 4
++#define XM_FCNTL_LBN 10
++#define XM_FCNTL_WIDTH 1
++#define XM_TXCRC_LBN 8
++#define XM_TXCRC_WIDTH 1
++#define XM_EDRC_LBN 6
++#define XM_EDRC_WIDTH 1
++#define XM_AUTO_PAD_LBN 5
++#define XM_AUTO_PAD_WIDTH 1
++#define XM_TX_PRMBL_LBN 2
++#define XM_TX_PRMBL_WIDTH 1
++#define XM_TXEN_LBN 1
++#define XM_TXEN_WIDTH 1
++#define XM_TX_RST_LBN 0
++#define XM_TX_RST_WIDTH 1
++
++/* XGMAC receive configuration */
++#define XM_RX_CFG_REG_MAC 0x04
++#define XM_PASS_LENERR_LBN 26
++#define XM_PASS_LENERR_WIDTH 1
++#define XM_PASS_CRC_ERR_LBN 25
++#define XM_PASS_CRC_ERR_WIDTH 1
++#define XM_PASS_PRMBLE_ERR_LBN 24
++#define XM_PASS_PRMBLE_ERR_WIDTH 1
++#define XM_ACPT_ALL_MCAST_LBN 11
++#define XM_ACPT_ALL_MCAST_WIDTH 1
++#define XM_ACPT_ALL_UCAST_LBN 9
++#define XM_ACPT_ALL_UCAST_WIDTH 1
++#define XM_AUTO_DEPAD_LBN 8
++#define XM_AUTO_DEPAD_WIDTH 1
++#define XM_RXCRC_LBN 3
++#define XM_RXCRC_WIDTH 1
++#define XM_RX_PRMBL_LBN 2
++#define XM_RX_PRMBL_WIDTH 1
++#define XM_RXEN_LBN 1
++#define XM_RXEN_WIDTH 1
++#define XM_RX_RST_LBN 0
++#define XM_RX_RST_WIDTH 1
++
++/* XGMAC flow control register */
++#define XM_FC_REG_MAC 0x7
++#define XM_PAUSE_TIME_LBN 16
++#define XM_PAUSE_TIME_WIDTH 16
++#define XM_DIS_FCNTL_LBN 0
++#define XM_DIS_FCNTL_WIDTH 1
++
++/* XGMAC pause time count register */
++#define XM_PAUSE_TIME_REG_MAC 0x9
++#define XM_TX_PAUSE_CNT_LBN 16
++#define XM_TX_PAUSE_CNT_WIDTH 16
++#define XM_RX_PAUSE_CNT_LBN 0
++#define XM_RX_PAUSE_CNT_WIDTH 16
++
++/* XGMAC transmit parameter register */
++#define XM_TX_PARAM_REG_MAC 0x0d
++#define XM_TX_JUMBO_MODE_LBN 31
++#define XM_TX_JUMBO_MODE_WIDTH 1
++#define XM_MAX_TX_FRM_SIZE_LBN 16
++#define XM_MAX_TX_FRM_SIZE_WIDTH 14
++#define XM_PAD_CHAR_LBN 0
++#define XM_PAD_CHAR_WIDTH 8
++
++/* XGMAC receive parameter register */
++#define XM_RX_PARAM_REG_MAC 0x0e
++#define XM_MAX_RX_FRM_SIZE_LBN 0
++#define XM_MAX_RX_FRM_SIZE_WIDTH 14
++
++/* XGXS/XAUI powerdown/reset register */
++#define XX_PWR_RST_REG_MAC 0x10
++
++#define XX_PWRDND_EN_LBN 15
++#define XX_PWRDND_EN_WIDTH 1
++#define XX_PWRDNC_EN_LBN 14
++#define XX_PWRDNC_EN_WIDTH 1
++#define XX_PWRDNB_EN_LBN 13
++#define XX_PWRDNB_EN_WIDTH 1
++#define XX_PWRDNA_EN_LBN 12
++#define XX_PWRDNA_EN_WIDTH 1
++#define XX_RSTPLLCD_EN_LBN 9
++#define XX_RSTPLLCD_EN_WIDTH 1
++#define XX_RSTPLLAB_EN_LBN 8
++#define XX_RSTPLLAB_EN_WIDTH 1
++#define XX_RESETD_EN_LBN 7
++#define XX_RESETD_EN_WIDTH 1
++#define XX_RESETC_EN_LBN 6
++#define XX_RESETC_EN_WIDTH 1
++#define XX_RESETB_EN_LBN 5
++#define XX_RESETB_EN_WIDTH 1
++#define XX_RESETA_EN_LBN 4
++#define XX_RESETA_EN_WIDTH 1
++#define XX_RSTXGXSRX_EN_LBN 2
++#define XX_RSTXGXSRX_EN_WIDTH 1
++#define XX_RSTXGXSTX_EN_LBN 1
++#define XX_RSTXGXSTX_EN_WIDTH 1
++#define XX_RST_XX_EN_LBN 0
++#define XX_RST_XX_EN_WIDTH 1
++
++/* XGXS/XAUI powerdown/reset control register */
++#define XX_SD_CTL_REG_MAC 0x11
++#define XX_TERMADJ1_LBN 17
++#define XX_TERMADJ1_WIDTH 1
++#define XX_TERMADJ0_LBN 16
++#define XX_TERMADJ0_WIDTH 1
++#define XX_HIDRVD_LBN 15
++#define XX_HIDRVD_WIDTH 1
++#define XX_LODRVD_LBN 14
++#define XX_LODRVD_WIDTH 1
++#define XX_HIDRVC_LBN 13
++#define XX_HIDRVC_WIDTH 1
++#define XX_LODRVC_LBN 12
++#define XX_LODRVC_WIDTH 1
++#define XX_HIDRVB_LBN 11
++#define XX_HIDRVB_WIDTH 1
++#define XX_LODRVB_LBN 10
++#define XX_LODRVB_WIDTH 1
++#define XX_HIDRVA_LBN 9
++#define XX_HIDRVA_WIDTH 1
++#define XX_LODRVA_LBN 8
++#define XX_LODRVA_WIDTH 1
++#define XX_LPBKD_LBN 3
++#define XX_LPBKD_WIDTH 1
++#define XX_LPBKC_LBN 2
++#define XX_LPBKC_WIDTH 1
++#define XX_LPBKB_LBN 1
++#define XX_LPBKB_WIDTH 1
++#define XX_LPBKA_LBN 0
++#define XX_LPBKA_WIDTH 1
++
++#define XX_TXDRV_CTL_REG_MAC 0x12
++#define XX_DEQD_LBN 28
++#define XX_DEQD_WIDTH 4
++#define XX_DEQC_LBN 24
++#define XX_DEQC_WIDTH 4
++#define XX_DEQB_LBN 20
++#define XX_DEQB_WIDTH 4
++#define XX_DEQA_LBN 16
++#define XX_DEQA_WIDTH 4
++#define XX_DTXD_LBN 12
++#define XX_DTXD_WIDTH 4
++#define XX_DTXC_LBN 8
++#define XX_DTXC_WIDTH 4
++#define XX_DTXB_LBN 4
++#define XX_DTXB_WIDTH 4
++#define XX_DTXA_LBN 0
++#define XX_DTXA_WIDTH 4
++
++/* XAUI XGXS core status register */
++#define XX_FORCE_SIG_LBN 24
++#define XX_FORCE_SIG_WIDTH 8
++#define XX_FORCE_SIG_DECODE_FORCED 0xff
++#define XX_XGXS_LB_EN_LBN 23
++#define XX_XGXS_LB_EN_WIDTH 1
++#define XX_XGMII_LB_EN_LBN 22
++#define XX_XGMII_LB_EN_WIDTH 1
++#define XX_CORE_STAT_REG_MAC 0x16
++#define XX_ALIGN_DONE_LBN 20
++#define XX_ALIGN_DONE_WIDTH 1
++#define XX_SYNC_STAT_LBN 16
++#define XX_SYNC_STAT_WIDTH 4
++#define XX_SYNC_STAT_DECODE_SYNCED 0xf
++#define XX_COMMA_DET_LBN 12
++#define XX_COMMA_DET_WIDTH 4
++#define XX_COMMA_DET_DECODE_DETECTED 0xf
++#define XX_COMMA_DET_RESET 0xf
++#define XX_CHARERR_LBN 4
++#define XX_CHARERR_WIDTH 4
++#define XX_CHARERR_RESET 0xf
++#define XX_DISPERR_LBN 0
++#define XX_DISPERR_WIDTH 4
++#define XX_DISPERR_RESET 0xf
++
++/* Receive filter table */
++#define RX_FILTER_TBL0 0xF00000
++
++/* Receive descriptor pointer table */
++#define RX_DESC_PTR_TBL_KER_A1 0x11800
++#define RX_DESC_PTR_TBL_KER_B0 0xF40000
++#define RX_DESC_PTR_TBL_KER_P0 0x900
++#define RX_ISCSI_DDIG_EN_LBN 88
++#define RX_ISCSI_DDIG_EN_WIDTH 1
++#define RX_ISCSI_HDIG_EN_LBN 87
++#define RX_ISCSI_HDIG_EN_WIDTH 1
++#define RX_DESC_PREF_ACT_LBN 86
++#define RX_DESC_PREF_ACT_WIDTH 1
++#define RX_DC_HW_RPTR_LBN 80
++#define RX_DC_HW_RPTR_WIDTH 6
++#define RX_DESCQ_HW_RPTR_LBN 68
++#define RX_DESCQ_HW_RPTR_WIDTH 12
++#define RX_DESCQ_SW_WPTR_LBN 56
++#define RX_DESCQ_SW_WPTR_WIDTH 12
++#define RX_DESCQ_BUF_BASE_ID_LBN 36
++#define RX_DESCQ_BUF_BASE_ID_WIDTH 20
++#define RX_DESCQ_EVQ_ID_LBN 24
++#define RX_DESCQ_EVQ_ID_WIDTH 12
++#define RX_DESCQ_OWNER_ID_LBN 10
++#define RX_DESCQ_OWNER_ID_WIDTH 14
++#define RX_DESCQ_LABEL_LBN 5
++#define RX_DESCQ_LABEL_WIDTH 5
++#define RX_DESCQ_SIZE_LBN 3
++#define RX_DESCQ_SIZE_WIDTH 2
++#define RX_DESCQ_SIZE_4K 3
++#define RX_DESCQ_SIZE_2K 2
++#define RX_DESCQ_SIZE_1K 1
++#define RX_DESCQ_SIZE_512 0
++#define RX_DESCQ_TYPE_LBN 2
++#define RX_DESCQ_TYPE_WIDTH 1
++#define RX_DESCQ_JUMBO_LBN 1
++#define RX_DESCQ_JUMBO_WIDTH 1
++#define RX_DESCQ_EN_LBN 0
++#define RX_DESCQ_EN_WIDTH 1
++
++/* Transmit descriptor pointer table */
++#define TX_DESC_PTR_TBL_KER_A1 0x11900
++#define TX_DESC_PTR_TBL_KER_B0 0xF50000
++#define TX_DESC_PTR_TBL_KER_P0 0xa40
++#define TX_NON_IP_DROP_DIS_B0_LBN 91
++#define TX_NON_IP_DROP_DIS_B0_WIDTH 1
++#define TX_IP_CHKSM_DIS_B0_LBN 90
++#define TX_IP_CHKSM_DIS_B0_WIDTH 1
++#define TX_TCP_CHKSM_DIS_B0_LBN 89
++#define TX_TCP_CHKSM_DIS_B0_WIDTH 1
++#define TX_DESCQ_EN_LBN 88
++#define TX_DESCQ_EN_WIDTH 1
++#define TX_ISCSI_DDIG_EN_LBN 87
++#define TX_ISCSI_DDIG_EN_WIDTH 1
++#define TX_ISCSI_HDIG_EN_LBN 86
++#define TX_ISCSI_HDIG_EN_WIDTH 1
++#define TX_DC_HW_RPTR_LBN 80
++#define TX_DC_HW_RPTR_WIDTH 6
++#define TX_DESCQ_HW_RPTR_LBN 68
++#define TX_DESCQ_HW_RPTR_WIDTH 12
++#define TX_DESCQ_SW_WPTR_LBN 56
++#define TX_DESCQ_SW_WPTR_WIDTH 12
++#define TX_DESCQ_BUF_BASE_ID_LBN 36
++#define TX_DESCQ_BUF_BASE_ID_WIDTH 20
++#define TX_DESCQ_EVQ_ID_LBN 24
++#define TX_DESCQ_EVQ_ID_WIDTH 12
++#define TX_DESCQ_OWNER_ID_LBN 10
++#define TX_DESCQ_OWNER_ID_WIDTH 14
++#define TX_DESCQ_LABEL_LBN 5
++#define TX_DESCQ_LABEL_WIDTH 5
++#define TX_DESCQ_SIZE_LBN 3
++#define TX_DESCQ_SIZE_WIDTH 2
++#define TX_DESCQ_SIZE_4K 3
++#define TX_DESCQ_SIZE_2K 2
++#define TX_DESCQ_SIZE_1K 1
++#define TX_DESCQ_SIZE_512 0
++#define TX_DESCQ_TYPE_LBN 1
++#define TX_DESCQ_TYPE_WIDTH 2
++#define TX_DESCQ_FLUSH_LBN 0
++#define TX_DESCQ_FLUSH_WIDTH 1
++
++/* Event queue pointer */
++#define EVQ_PTR_TBL_KER_A1 0x11a00
++#define EVQ_PTR_TBL_KER_B0 0xf60000
++#define EVQ_PTR_TBL_KER_P0 0x500
++#define EVQ_WKUP_OR_INT_EN_LBN 39
++#define EVQ_WKUP_OR_INT_EN_WIDTH 1
++#define EVQ_NXT_WPTR_LBN 24
++#define EVQ_NXT_WPTR_WIDTH 15
++#define EVQ_EN_LBN 23
++#define EVQ_EN_WIDTH 1
++#define EVQ_SIZE_LBN 20
++#define EVQ_SIZE_WIDTH 3
++#define EVQ_SIZE_32K 6
++#define EVQ_SIZE_16K 5
++#define EVQ_SIZE_8K 4
++#define EVQ_SIZE_4K 3
++#define EVQ_SIZE_2K 2
++#define EVQ_SIZE_1K 1
++#define EVQ_SIZE_512 0
++#define EVQ_BUF_BASE_ID_LBN 0
++#define EVQ_BUF_BASE_ID_WIDTH 20
++
++/* Event queue read pointer */
++#define EVQ_RPTR_REG_KER_A1 0x11b00
++#define EVQ_RPTR_REG_KER_B0 0xfa0000
++#define EVQ_RPTR_LBN 0
++#define EVQ_RPTR_WIDTH 14
++#define EVQ_RPTR_REG_KER_DWORD (EVQ_RPTR_REG_KER + 0)
++#define EVQ_RPTR_DWORD_LBN 0
++#define EVQ_RPTR_DWORD_WIDTH 14
++
++/* RSS indirection table */
++#define RX_RSS_INDIR_TBL_B0 0xFB0000
++#define RX_RSS_INDIR_ENT_B0_LBN 0
++#define RX_RSS_INDIR_ENT_B0_WIDTH 6
++
++/* Special buffer descriptors (full-mode) */
++#define BUF_FULL_TBL_KER_A1 0x8000
++#define BUF_FULL_TBL_KER_B0 0x800000
++#define IP_DAT_BUF_SIZE_LBN 50
++#define IP_DAT_BUF_SIZE_WIDTH 1
++#define IP_DAT_BUF_SIZE_8K 1
++#define IP_DAT_BUF_SIZE_4K 0
++#define BUF_ADR_REGION_LBN 48
++#define BUF_ADR_REGION_WIDTH 2
++#define BUF_ADR_FBUF_LBN 14
++#define BUF_ADR_FBUF_WIDTH 34
++#define BUF_OWNER_ID_FBUF_LBN 0
++#define BUF_OWNER_ID_FBUF_WIDTH 14
++
++/* Special buffer descriptors (half-mode) */
++#define BUF_HALF_TBL_KER_A1 0x8000
++#define BUF_HALF_TBL_KER_B0 0x800000
++#define BUF_ADR_HBUF_ODD_LBN 44
++#define BUF_ADR_HBUF_ODD_WIDTH 20
++#define BUF_OWNER_ID_HBUF_ODD_LBN 32
++#define BUF_OWNER_ID_HBUF_ODD_WIDTH 12
++#define BUF_ADR_HBUF_EVEN_LBN 12
++#define BUF_ADR_HBUF_EVEN_WIDTH 20
++#define BUF_OWNER_ID_HBUF_EVEN_LBN 0
++#define BUF_OWNER_ID_HBUF_EVEN_WIDTH 12
++
++#define SRM_DBG_REG_B0 0x3000000
++
++/* Transmit descriptor */
++#define TX_KER_PORT_LBN 63
++#define TX_KER_PORT_WIDTH 1
++#define TX_KER_CONT_LBN 62
++#define TX_KER_CONT_WIDTH 1
++#define TX_KER_BYTE_CNT_LBN 48
++#define TX_KER_BYTE_CNT_WIDTH 14
++#define TX_KER_BUF_REGION_LBN 46
++#define TX_KER_BUF_REGION_WIDTH 2
++#define TX_KER_BUF_REGION0_DECODE 0
++#define TX_KER_BUF_REGION1_DECODE 1
++#define TX_KER_BUF_REGION2_DECODE 2
++#define TX_KER_BUF_REGION3_DECODE 3
++#define TX_KER_BUF_ADR_LBN 0
++#define TX_KER_BUF_ADR_WIDTH EFX_DMA_TYPE_WIDTH(46)
++
++/* Receive descriptor */
++#define RX_KER_BUF_SIZE_LBN 48
++#define RX_KER_BUF_SIZE_WIDTH 14
++#define RX_KER_BUF_REGION_LBN 46
++#define RX_KER_BUF_REGION_WIDTH 2
++#define RX_KER_BUF_REGION0_DECODE 0
++#define RX_KER_BUF_REGION1_DECODE 1
++#define RX_KER_BUF_REGION2_DECODE 2
++#define RX_KER_BUF_REGION3_DECODE 3
++#define RX_KER_BUF_ADR_LBN 0
++#define RX_KER_BUF_ADR_WIDTH EFX_DMA_TYPE_WIDTH(46)
++
++/**************************************************************************
++ *
++ * Falcon events
++ *
++ **************************************************************************
++ */
++
++/* Event queue entries */
++#define EV_CODE_LBN 60
++#define EV_CODE_WIDTH 4
++#define RX_IP_EV_DECODE 0
++#define TX_IP_EV_DECODE 2
++#define DRIVER_EV_DECODE 5
++#define GLOBAL_EV_DECODE 6
++#define DRV_GEN_EV_DECODE 7
++#define WHOLE_EVENT_LBN 0
++#define WHOLE_EVENT_WIDTH 64
++
++/* Receive events */
++#define RX_EV_PKT_OK_LBN 56
++#define RX_EV_PKT_OK_WIDTH 1
++#define RX_EV_PAUSE_FRM_ERR_LBN 55
++#define RX_EV_PAUSE_FRM_ERR_WIDTH 1
++#define RX_EV_BUF_OWNER_ID_ERR_LBN 54
++#define RX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++#define RX_EV_IF_FRAG_ERR_LBN 53
++#define RX_EV_IF_FRAG_ERR_WIDTH 1
++#define RX_EV_IP_HDR_CHKSUM_ERR_LBN 52
++#define RX_EV_IP_HDR_CHKSUM_ERR_WIDTH 1
++#define RX_EV_TCP_UDP_CHKSUM_ERR_LBN 51
++#define RX_EV_TCP_UDP_CHKSUM_ERR_WIDTH 1
++#define RX_EV_ETH_CRC_ERR_LBN 50
++#define RX_EV_ETH_CRC_ERR_WIDTH 1
++#define RX_EV_FRM_TRUNC_LBN 49
++#define RX_EV_FRM_TRUNC_WIDTH 1
++#define RX_EV_DRIB_NIB_LBN 48
++#define RX_EV_DRIB_NIB_WIDTH 1
++#define RX_EV_TOBE_DISC_LBN 47
++#define RX_EV_TOBE_DISC_WIDTH 1
++#define RX_EV_PKT_TYPE_LBN 44
++#define RX_EV_PKT_TYPE_WIDTH 3
++#define RX_EV_PKT_TYPE_ETH_DECODE 0
++#define RX_EV_PKT_TYPE_LLC_DECODE 1
++#define RX_EV_PKT_TYPE_JUMBO_DECODE 2
++#define RX_EV_PKT_TYPE_VLAN_DECODE 3
++#define RX_EV_PKT_TYPE_VLAN_LLC_DECODE 4
++#define RX_EV_PKT_TYPE_VLAN_JUMBO_DECODE 5
++#define RX_EV_HDR_TYPE_LBN 42
++#define RX_EV_HDR_TYPE_WIDTH 2
++#define RX_EV_HDR_TYPE_TCP_IPV4_DECODE 0
++#define RX_EV_HDR_TYPE_UDP_IPV4_DECODE 1
++#define RX_EV_HDR_TYPE_OTHER_IP_DECODE 2
++#define RX_EV_HDR_TYPE_NON_IP_DECODE 3
++#define RX_EV_HDR_TYPE_HAS_CHECKSUMS(hdr_type) \
++ ((hdr_type) <= RX_EV_HDR_TYPE_UDP_IPV4_DECODE)
++#define RX_EV_DESC_Q_EMPTY_LBN 41
++#define RX_EV_DESC_Q_EMPTY_WIDTH 1
++#define RX_EV_MCAST_HASH_MATCH_LBN 40
++#define RX_EV_MCAST_HASH_MATCH_WIDTH 1
++#define RX_EV_MCAST_PKT_LBN 39
++#define RX_EV_MCAST_PKT_WIDTH 1
++#define RX_EV_RECOVERY_FLAG_LBN 37
++#define RX_EV_RECOVERY_FLAG_WIDTH 1
++#define RX_EV_Q_LABEL_LBN 32
++#define RX_EV_Q_LABEL_WIDTH 5
++#define RX_EV_JUMBO_CONT_LBN 31
++#define RX_EV_JUMBO_CONT_WIDTH 1
++#define RX_EV_PORT_LBN 30
++#define RX_EV_PORT_WIDTH 1
++#define RX_EV_BYTE_CNT_LBN 16
++#define RX_EV_BYTE_CNT_WIDTH 14
++#define RX_EV_SOP_LBN 15
++#define RX_EV_SOP_WIDTH 1
++#define RX_ISCSI_DDIG_ERR_LBN 13
++#define RX_ISCSI_DDIG_ERR_WIDTH 1
++#define RX_ISCSI_HDIG_ERR_LBN 12
++#define RX_ISCSI_HDIG_ERR_WIDTH 1
++#define RX_EV_DESC_PTR_LBN 0
++#define RX_EV_DESC_PTR_WIDTH 12
++
++/* Transmit events */
++#define TX_EV_PKT_ERR_LBN 38
++#define TX_EV_PKT_ERR_WIDTH 1
++#define TX_EV_PKT_TOO_BIG_LBN 37
++#define TX_EV_PKT_TOO_BIG_WIDTH 1
++#define TX_EV_Q_LABEL_LBN 32
++#define TX_EV_Q_LABEL_WIDTH 5
++#define TX_EV_PORT_LBN 16
++#define TX_EV_PORT_WIDTH 1
++#define TX_EV_WQ_FF_FULL_LBN 15
++#define TX_EV_WQ_FF_FULL_WIDTH 1
++#define TX_EV_BUF_OWNER_ID_ERR_LBN 14
++#define TX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++#define TX_EV_COMP_LBN 12
++#define TX_EV_COMP_WIDTH 1
++#define TX_EV_DESC_PTR_LBN 0
++#define TX_EV_DESC_PTR_WIDTH 12
++
++/* Driver events */
++#define DRIVER_EV_SUB_CODE_LBN 56
++#define DRIVER_EV_SUB_CODE_WIDTH 4
++#define DRIVER_EV_SUB_DATA_LBN 0
++#define DRIVER_EV_SUB_DATA_WIDTH 14
++#define TX_DESCQ_FLS_DONE_EV_DECODE 0
++#define RX_DESCQ_FLS_DONE_EV_DECODE 1
++#define EVQ_INIT_DONE_EV_DECODE 2
++#define EVQ_NOT_EN_EV_DECODE 3
++#define RX_DESCQ_FLSFF_OVFL_EV_DECODE 4
++#define SRM_UPD_DONE_EV_DECODE 5
++#define WAKE_UP_EV_DECODE 6
++#define TX_PKT_NON_TCP_UDP_DECODE 9
++#define TIMER_EV_DECODE 10
++#define RX_RECOVERY_EV_DECODE 11
++#define RX_DSC_ERROR_EV_DECODE 14
++#define TX_DSC_ERROR_EV_DECODE 15
++#define DRIVER_EV_TX_DESCQ_ID_LBN 0
++#define DRIVER_EV_TX_DESCQ_ID_WIDTH 12
++#define DRIVER_EV_RX_FLUSH_FAIL_LBN 12
++#define DRIVER_EV_RX_FLUSH_FAIL_WIDTH 1
++#define DRIVER_EV_RX_DESCQ_ID_LBN 0
++#define DRIVER_EV_RX_DESCQ_ID_WIDTH 12
++#define DRIVER_EV_EVQ_ID_LBN 0
++#define DRIVER_EV_EVQ_ID_WIDTH 12
++#define DRIVER_EV_SRM_UPD_LBN 0
++#define DRIVER_EV_SRM_UPD_WIDTH 2
++#define SRM_CLR_EV_DECODE 0
++#define SRM_UPD_EV_DECODE 1
++#define SRM_ILLCLR_EV_DECODE 2
++
++/* Global events */
++#define RX_RECOVERY_B0_LBN 12
++#define RX_RECOVERY_B0_WIDTH 1
++#define XG_MNT_INTR_B0_LBN 11
++#define XG_MNT_INTR_B0_WIDTH 1
++
++#define RX_RECOVERY_A1_LBN 11
++#define RX_RECOVERY_A1_WIDTH 1
++
++#define XFP_PHY_INTR_LBN 10
++#define XFP_PHY_INTR_WIDTH 1
++#define XG_PHY_INTR_LBN 9
++#define XG_PHY_INTR_WIDTH 1
++#define G_PHY1_INTR_LBN 8
++#define G_PHY1_INTR_WIDTH 1
++#define G_PHY0_INTR_LBN 7
++#define G_PHY0_INTR_WIDTH 1
++
++/* Driver-generated test events */
++#define DRV_GEN_EV_CODE_LBN 60
++#define DRV_GEN_EV_CODE_WIDTH 4
++#define DRV_GEN_EV_DATA_LBN 0
++#define DRV_GEN_EV_DATA_WIDTH 60
++#define EVQ_MAGIC_LBN 0
++#define EVQ_MAGIC_WIDTH 32
++
++/**************************************************************************
++ *
++ * Falcon MAC stats
++ *
++ **************************************************************************
++ *
++ */
++
++#define GRxGoodOct_offset 0x0
++#define GRxGoodOct_WIDTH 48
++#define GRxBadOct_offset 0x8
++#define GRxBadOct_WIDTH 48
++#define GRxMissPkt_offset 0x10
++#define GRxMissPkt_WIDTH 32
++#define GRxFalseCRS_offset 0x14
++#define GRxFalseCRS_WIDTH 32
++#define GRxPausePkt_offset 0x18
++#define GRxPausePkt_WIDTH 32
++#define GRxBadPkt_offset 0x1C
++#define GRxBadPkt_WIDTH 32
++#define GRxUcastPkt_offset 0x20
++#define GRxUcastPkt_WIDTH 32
++#define GRxMcastPkt_offset 0x24
++#define GRxMcastPkt_WIDTH 32
++#define GRxBcastPkt_offset 0x28
++#define GRxBcastPkt_WIDTH 32
++#define GRxGoodLt64Pkt_offset 0x2C
++#define GRxGoodLt64Pkt_WIDTH 32
++#define GRxBadLt64Pkt_offset 0x30
++#define GRxBadLt64Pkt_WIDTH 32
++#define GRx64Pkt_offset 0x34
++#define GRx64Pkt_WIDTH 32
++#define GRx65to127Pkt_offset 0x38
++#define GRx65to127Pkt_WIDTH 32
++#define GRx128to255Pkt_offset 0x3C
++#define GRx128to255Pkt_WIDTH 32
++#define GRx256to511Pkt_offset 0x40
++#define GRx256to511Pkt_WIDTH 32
++#define GRx512to1023Pkt_offset 0x44
++#define GRx512to1023Pkt_WIDTH 32
++#define GRx1024to15xxPkt_offset 0x48
++#define GRx1024to15xxPkt_WIDTH 32
++#define GRx15xxtoJumboPkt_offset 0x4C
++#define GRx15xxtoJumboPkt_WIDTH 32
++#define GRxGtJumboPkt_offset 0x50
++#define GRxGtJumboPkt_WIDTH 32
++#define GRxFcsErr64to15xxPkt_offset 0x54
++#define GRxFcsErr64to15xxPkt_WIDTH 32
++#define GRxFcsErr15xxtoJumboPkt_offset 0x58
++#define GRxFcsErr15xxtoJumboPkt_WIDTH 32
++#define GRxFcsErrGtJumboPkt_offset 0x5C
++#define GRxFcsErrGtJumboPkt_WIDTH 32
++#define GTxGoodBadOct_offset 0x80
++#define GTxGoodBadOct_WIDTH 48
++#define GTxGoodOct_offset 0x88
++#define GTxGoodOct_WIDTH 48
++#define GTxSglColPkt_offset 0x90
++#define GTxSglColPkt_WIDTH 32
++#define GTxMultColPkt_offset 0x94
++#define GTxMultColPkt_WIDTH 32
++#define GTxExColPkt_offset 0x98
++#define GTxExColPkt_WIDTH 32
++#define GTxDefPkt_offset 0x9C
++#define GTxDefPkt_WIDTH 32
++#define GTxLateCol_offset 0xA0
++#define GTxLateCol_WIDTH 32
++#define GTxExDefPkt_offset 0xA4
++#define GTxExDefPkt_WIDTH 32
++#define GTxPausePkt_offset 0xA8
++#define GTxPausePkt_WIDTH 32
++#define GTxBadPkt_offset 0xAC
++#define GTxBadPkt_WIDTH 32
++#define GTxUcastPkt_offset 0xB0
++#define GTxUcastPkt_WIDTH 32
++#define GTxMcastPkt_offset 0xB4
++#define GTxMcastPkt_WIDTH 32
++#define GTxBcastPkt_offset 0xB8
++#define GTxBcastPkt_WIDTH 32
++#define GTxLt64Pkt_offset 0xBC
++#define GTxLt64Pkt_WIDTH 32
++#define GTx64Pkt_offset 0xC0
++#define GTx64Pkt_WIDTH 32
++#define GTx65to127Pkt_offset 0xC4
++#define GTx65to127Pkt_WIDTH 32
++#define GTx128to255Pkt_offset 0xC8
++#define GTx128to255Pkt_WIDTH 32
++#define GTx256to511Pkt_offset 0xCC
++#define GTx256to511Pkt_WIDTH 32
++#define GTx512to1023Pkt_offset 0xD0
++#define GTx512to1023Pkt_WIDTH 32
++#define GTx1024to15xxPkt_offset 0xD4
++#define GTx1024to15xxPkt_WIDTH 32
++#define GTx15xxtoJumboPkt_offset 0xD8
++#define GTx15xxtoJumboPkt_WIDTH 32
++#define GTxGtJumboPkt_offset 0xDC
++#define GTxGtJumboPkt_WIDTH 32
++#define GTxNonTcpUdpPkt_offset 0xE0
++#define GTxNonTcpUdpPkt_WIDTH 16
++#define GTxMacSrcErrPkt_offset 0xE4
++#define GTxMacSrcErrPkt_WIDTH 16
++#define GTxIpSrcErrPkt_offset 0xE8
++#define GTxIpSrcErrPkt_WIDTH 16
++#define GDmaDone_offset 0xEC
++#define GDmaDone_WIDTH 32
++
++#define XgRxOctets_offset 0x0
++#define XgRxOctets_WIDTH 48
++#define XgRxOctetsOK_offset 0x8
++#define XgRxOctetsOK_WIDTH 48
++#define XgRxPkts_offset 0x10
++#define XgRxPkts_WIDTH 32
++#define XgRxPktsOK_offset 0x14
++#define XgRxPktsOK_WIDTH 32
++#define XgRxBroadcastPkts_offset 0x18
++#define XgRxBroadcastPkts_WIDTH 32
++#define XgRxMulticastPkts_offset 0x1C
++#define XgRxMulticastPkts_WIDTH 32
++#define XgRxUnicastPkts_offset 0x20
++#define XgRxUnicastPkts_WIDTH 32
++#define XgRxUndersizePkts_offset 0x24
++#define XgRxUndersizePkts_WIDTH 32
++#define XgRxOversizePkts_offset 0x28
++#define XgRxOversizePkts_WIDTH 32
++#define XgRxJabberPkts_offset 0x2C
++#define XgRxJabberPkts_WIDTH 32
++#define XgRxUndersizeFCSerrorPkts_offset 0x30
++#define XgRxUndersizeFCSerrorPkts_WIDTH 32
++#define XgRxDropEvents_offset 0x34
++#define XgRxDropEvents_WIDTH 32
++#define XgRxFCSerrorPkts_offset 0x38
++#define XgRxFCSerrorPkts_WIDTH 32
++#define XgRxAlignError_offset 0x3C
++#define XgRxAlignError_WIDTH 32
++#define XgRxSymbolError_offset 0x40
++#define XgRxSymbolError_WIDTH 32
++#define XgRxInternalMACError_offset 0x44
++#define XgRxInternalMACError_WIDTH 32
++#define XgRxControlPkts_offset 0x48
++#define XgRxControlPkts_WIDTH 32
++#define XgRxPausePkts_offset 0x4C
++#define XgRxPausePkts_WIDTH 32
++#define XgRxPkts64Octets_offset 0x50
++#define XgRxPkts64Octets_WIDTH 32
++#define XgRxPkts65to127Octets_offset 0x54
++#define XgRxPkts65to127Octets_WIDTH 32
++#define XgRxPkts128to255Octets_offset 0x58
++#define XgRxPkts128to255Octets_WIDTH 32
++#define XgRxPkts256to511Octets_offset 0x5C
++#define XgRxPkts256to511Octets_WIDTH 32
++#define XgRxPkts512to1023Octets_offset 0x60
++#define XgRxPkts512to1023Octets_WIDTH 32
++#define XgRxPkts1024to15xxOctets_offset 0x64
++#define XgRxPkts1024to15xxOctets_WIDTH 32
++#define XgRxPkts15xxtoMaxOctets_offset 0x68
++#define XgRxPkts15xxtoMaxOctets_WIDTH 32
++#define XgRxLengthError_offset 0x6C
++#define XgRxLengthError_WIDTH 32
++#define XgTxPkts_offset 0x80
++#define XgTxPkts_WIDTH 32
++#define XgTxOctets_offset 0x88
++#define XgTxOctets_WIDTH 48
++#define XgTxMulticastPkts_offset 0x90
++#define XgTxMulticastPkts_WIDTH 32
++#define XgTxBroadcastPkts_offset 0x94
++#define XgTxBroadcastPkts_WIDTH 32
++#define XgTxUnicastPkts_offset 0x98
++#define XgTxUnicastPkts_WIDTH 32
++#define XgTxControlPkts_offset 0x9C
++#define XgTxControlPkts_WIDTH 32
++#define XgTxPausePkts_offset 0xA0
++#define XgTxPausePkts_WIDTH 32
++#define XgTxPkts64Octets_offset 0xA4
++#define XgTxPkts64Octets_WIDTH 32
++#define XgTxPkts65to127Octets_offset 0xA8
++#define XgTxPkts65to127Octets_WIDTH 32
++#define XgTxPkts128to255Octets_offset 0xAC
++#define XgTxPkts128to255Octets_WIDTH 32
++#define XgTxPkts256to511Octets_offset 0xB0
++#define XgTxPkts256to511Octets_WIDTH 32
++#define XgTxPkts512to1023Octets_offset 0xB4
++#define XgTxPkts512to1023Octets_WIDTH 32
++#define XgTxPkts1024to15xxOctets_offset 0xB8
++#define XgTxPkts1024to15xxOctets_WIDTH 32
++#define XgTxPkts1519toMaxOctets_offset 0xBC
++#define XgTxPkts1519toMaxOctets_WIDTH 32
++#define XgTxUndersizePkts_offset 0xC0
++#define XgTxUndersizePkts_WIDTH 32
++#define XgTxOversizePkts_offset 0xC4
++#define XgTxOversizePkts_WIDTH 32
++#define XgTxNonTcpUdpPkt_offset 0xC8
++#define XgTxNonTcpUdpPkt_WIDTH 16
++#define XgTxMacSrcErrPkt_offset 0xCC
++#define XgTxMacSrcErrPkt_WIDTH 16
++#define XgTxIpSrcErrPkt_offset 0xD0
++#define XgTxIpSrcErrPkt_WIDTH 16
++#define XgDmaDone_offset 0xD4
++#define XgDmaDone_WIDTH 32
++
++#define FALCON_STATS_NOT_DONE 0x00000000
++#define FALCON_STATS_DONE 0xffffffff
++
++/* Interrupt status register bits */
++#define FATAL_INT_LBN 64
++#define FATAL_INT_WIDTH 1
++#define INT_EVQS_LBN 40
++#define INT_EVQS_WIDTH 4
++#define INT_FLAG_LBN 32
++#define INT_FLAG_WIDTH 1
++#define EVQ_FIFO_HF_LBN 1
++#define EVQ_FIFO_HF_WIDTH 1
++#define EVQ_FIFO_AF_LBN 0
++#define EVQ_FIFO_AF_WIDTH 1
++
++/**************************************************************************
++ *
++ * Falcon non-volatile configuration
++ *
++ **************************************************************************
++ */
++
++/* Board configuration v2 (v1 is obsolete; later versions are compatible) */
++struct falcon_nvconfig_board_v2 {
++ __le16 nports;
++ u8 port0_phy_addr;
++ u8 port0_phy_type;
++ u8 port1_phy_addr;
++ u8 port1_phy_type;
++ __le16 asic_sub_revision;
++ __le16 board_revision;
++} __attribute__ ((packed));
++
++/* Board configuration v3 extra information */
++struct falcon_nvconfig_board_v3 {
++ __le32 spi_device_type[2];
++} __attribute__ ((packed));
++
++/* Bit numbers for spi_device_type */
++#define SPI_DEV_TYPE_SIZE_LBN 0
++#define SPI_DEV_TYPE_SIZE_WIDTH 5
++#define SPI_DEV_TYPE_ADDR_LEN_LBN 6
++#define SPI_DEV_TYPE_ADDR_LEN_WIDTH 2
++#define SPI_DEV_TYPE_ERASE_CMD_LBN 8
++#define SPI_DEV_TYPE_ERASE_CMD_WIDTH 8
++#define SPI_DEV_TYPE_ERASE_SIZE_LBN 16
++#define SPI_DEV_TYPE_ERASE_SIZE_WIDTH 5
++#define SPI_DEV_TYPE_BLOCK_SIZE_LBN 24
++#define SPI_DEV_TYPE_BLOCK_SIZE_WIDTH 5
++#define SPI_DEV_TYPE_FIELD(type, field) \
++ (((type) >> EFX_LOW_BIT(field)) & EFX_MASK32(field))
++
++#define NVCONFIG_BASE 0x300
++#define NVCONFIG_BOARD_MAGIC_NUM 0xFA1C
++struct falcon_nvconfig {
++ efx_oword_t ee_vpd_cfg_reg; /* 0x300 */
++ u8 mac_address[2][8]; /* 0x310 */
++ efx_oword_t pcie_sd_ctl0123_reg; /* 0x320 */
++ efx_oword_t pcie_sd_ctl45_reg; /* 0x330 */
++ efx_oword_t pcie_pcs_ctl_stat_reg; /* 0x340 */
++ efx_oword_t hw_init_reg; /* 0x350 */
++ efx_oword_t nic_stat_reg; /* 0x360 */
++ efx_oword_t glb_ctl_reg; /* 0x370 */
++ efx_oword_t srm_cfg_reg; /* 0x380 */
++ efx_oword_t spare_reg; /* 0x390 */
++ __le16 board_magic_num; /* 0x3A0 */
++ __le16 board_struct_ver;
++ __le16 board_checksum;
++ struct falcon_nvconfig_board_v2 board_v2;
++ efx_oword_t ee_base_page_reg; /* 0x3B0 */
++ struct falcon_nvconfig_board_v3 board_v3; /* 0x3C0 */
++} __attribute__ ((packed));
++
++#endif /* EFX_FALCON_HWDEFS_H */
+--- linux-2.6.18.8/drivers/net/sfc/falcon_io.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon_io.h 2008-05-19 00:33:28.849809063 +0300
+@@ -0,0 +1,259 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <stdarg.h>
-+#include <linux/module.h>
-+#include <linux/kthread.h>
-+#include <xen/xenbus.h>
-+#include "common.h"
++#ifndef EFX_FALCON_IO_H
++#define EFX_FALCON_IO_H
+
++#include "net_driver.h"
++#include "falcon.h"
+
-+struct backend_info
-+{
-+ struct xenbus_device *dev;
-+ blkif_t *blkif;
-+ struct xenbus_watch backend_watch;
-+ int xenbus_id;
-+ int group_added;
-+};
++/**************************************************************************
++ *
++ * Falcon hardware access
++ *
++ **************************************************************************
++ *
++ * Notes on locking strategy:
++ *
++ * Most Falcon registers require 16-byte (or 8-byte, for SRAM
++ * registers) atomic writes which necessitates locking.
++ * Under normal operation few writes to the Falcon BAR are made and these
++ * registers (EVQ_RPTR_REG, RX_DESC_UPD_REG and TX_DESC_UPD_REG) are special
++ * cased to allow 4-byte (hence lockless) accesses.
++ *
++ * It *is* safe to write to these 4-byte registers in the middle of an
++ * access to an 8-byte or 16-byte register. We therefore use a
++ * spinlock to protect accesses to the larger registers, but no locks
++ * for the 4-byte registers.
++ *
++ * A write barrier is needed to ensure that DW3 is written after DW0/1/2
++ * due to the way the 16byte registers are "collected" in the Falcon BIU
++ *
++ * We also lock when carrying out reads, to ensure consistency of the
++ * data (made possible since the BIU reads all 128 bits into a cache).
++ * Reads are very rare, so this isn't a significant performance
++ * impact. (Most data transferred from NIC to host is DMAed directly
++ * into host memory).
++ *
++ * I/O BAR access uses locks for both reads and writes (but is only provided
++ * for testing purposes).
++ */
+
++/* Special buffer descriptors (Falcon SRAM) */
++#define BUF_TBL_KER_A1 0x18000
++#define BUF_TBL_KER_B0 0x800000
+
-+static void connect(struct backend_info *);
-+static int connect_ring(struct backend_info *);
-+static int blktap_remove(struct xenbus_device *dev);
-+static int blktap_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id);
-+static void tap_backend_changed(struct xenbus_watch *, const char **,
-+ unsigned int);
-+static void tap_frontend_changed(struct xenbus_device *dev,
-+ enum xenbus_state frontend_state);
+
-+static int strsep_len(const char *str, char c, unsigned int len)
-+{
-+ unsigned int i;
++#if BITS_PER_LONG == 64
++#define FALCON_USE_QWORD_IO 1
++#endif
+
-+ for (i = 0; str[i]; i++)
-+ if (str[i] == c) {
-+ if (len == 0)
-+ return i;
-+ len--;
-+ }
-+ return (len == 0) ? i : -ERANGE;
-+}
++#define _falcon_writeq(efx, value, reg) \
++ __raw_writeq((__force u64) (value), (efx)->membase + (reg))
++#define _falcon_writel(efx, value, reg) \
++ __raw_writel((__force u32) (value), (efx)->membase + (reg))
++#define _falcon_readq(efx, reg) \
++ ((__force __le64) __raw_readq((efx)->membase + (reg)))
++#define _falcon_readl(efx, reg) \
++ ((__force __le32) __raw_readl((efx)->membase + (reg)))
+
-+static long get_id(const char *str)
++/* Writes to a normal 16-byte Falcon register, locking as appropriate. */
++static inline void falcon_write(struct efx_nic *efx, efx_oword_t *value,
++ unsigned int reg)
+{
-+ int len,end;
-+ const char *ptr;
-+ char *tptr, num[10];
-+
-+ len = strsep_len(str, '/', 2);
-+ end = strlen(str);
-+ if ( (len < 0) || (end < 0) ) return -1;
-+
-+ ptr = str + len + 1;
-+ strncpy(num,ptr,end - len);
-+ tptr = num + (end - (len + 1));
-+ *tptr = '\0';
-+ DPRINTK("Get_id called for %s (%s)\n",str,num);
-+
-+ return simple_strtol(num, NULL, 10);
-+}
++ unsigned long flags __attribute__ ((unused));
+
-+static int blktap_name(blkif_t *blkif, char *buf)
++ EFX_REGDUMP(efx, "writing register %x with " EFX_OWORD_FMT "\n", reg,
++ EFX_OWORD_VAL(*value));
++
++ spin_lock_irqsave(&efx->biu_lock, flags);
++#ifdef FALCON_USE_QWORD_IO
++ _falcon_writeq(efx, value->u64[0], reg + 0);
++ wmb();
++ _falcon_writeq(efx, value->u64[1], reg + 8);
++#else
++ _falcon_writel(efx, value->u32[0], reg + 0);
++ _falcon_writel(efx, value->u32[1], reg + 4);
++ _falcon_writel(efx, value->u32[2], reg + 8);
++ wmb();
++ _falcon_writel(efx, value->u32[3], reg + 12);
++#endif
++ mmiowb();
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
++}
++
++/* Writes to an 8-byte Falcon SRAM register, locking as appropriate. */
++static inline void falcon_write_sram(struct efx_nic *efx, efx_qword_t *value,
++ unsigned int index)
+{
-+ char *devpath, *devname;
-+ struct xenbus_device *dev = blkif->be->dev;
++ unsigned int reg = efx->type->buf_tbl_base + (index * sizeof(*value));
++ unsigned long flags __attribute__ ((unused));
+
-+ devpath = xenbus_read(XBT_NIL, dev->nodename, "dev", NULL);
-+ if (IS_ERR(devpath))
-+ return PTR_ERR(devpath);
-+
-+ if ((devname = strstr(devpath, "/dev/")) != NULL)
-+ devname += strlen("/dev/");
-+ else
-+ devname = devpath;
++ EFX_REGDUMP(efx, "writing SRAM register %x with " EFX_QWORD_FMT "\n",
++ reg, EFX_QWORD_VAL(*value));
+
-+ snprintf(buf, TASK_COMM_LEN, "blktap.%d.%s", blkif->domid, devname);
-+ kfree(devpath);
-+
-+ return 0;
++ spin_lock_irqsave(&efx->biu_lock, flags);
++#ifdef FALCON_USE_QWORD_IO
++ _falcon_writeq(efx, value->u64[0], reg + 0);
++#else
++ _falcon_writel(efx, value->u32[0], reg + 0);
++ wmb();
++ _falcon_writel(efx, value->u32[1], reg + 4);
++#endif
++ mmiowb();
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
+}
+
-+/****************************************************************
-+ * sysfs interface for I/O requests of blktap device
++/* Write dword to Falcon register that allows partial writes
++ *
++ * Some Falcon registers (EVQ_RPTR_REG, RX_DESC_UPD_REG and
++ * TX_DESC_UPD_REG) can be written to as a single dword. This allows
++ * for lockless writes.
+ */
++static inline void falcon_writel(struct efx_nic *efx, efx_dword_t *value,
++ unsigned int reg)
++{
++ EFX_REGDUMP(efx, "writing partial register %x with "EFX_DWORD_FMT"\n",
++ reg, EFX_DWORD_VAL(*value));
+
-+#define VBD_SHOW(name, format, args...) \
-+ static ssize_t show_##name(struct device *_dev, \
-+ struct device_attribute *attr, \
-+ char *buf) \
-+ { \
-+ struct xenbus_device *dev = to_xenbus_device(_dev); \
-+ struct backend_info *be = dev->dev.driver_data; \
-+ \
-+ return sprintf(buf, format, ##args); \
-+ } \
-+ static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL)
++ /* No lock required */
++ _falcon_writel(efx, value->u32[0], reg);
++}
+
-+VBD_SHOW(oo_req, "%d\n", be->blkif->st_oo_req);
-+VBD_SHOW(rd_req, "%d\n", be->blkif->st_rd_req);
-+VBD_SHOW(wr_req, "%d\n", be->blkif->st_wr_req);
-+VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect);
-+VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect);
++/* Read from a Falcon register
++ *
++ * This reads an entire 16-byte Falcon register in one go, locking as
++ * appropriate. It is essential to read the first dword first, as this
++ * prompts Falcon to load the current value into the shadow register.
++ */
++static inline void falcon_read(struct efx_nic *efx, efx_oword_t *value,
++ unsigned int reg)
++{
++ unsigned long flags __attribute__ ((unused));
+
-+static struct attribute *tapstat_attrs[] = {
-+ &dev_attr_oo_req.attr,
-+ &dev_attr_rd_req.attr,
-+ &dev_attr_wr_req.attr,
-+ &dev_attr_rd_sect.attr,
-+ &dev_attr_wr_sect.attr,
-+ NULL
-+};
++ spin_lock_irqsave(&efx->biu_lock, flags);
++ value->u32[0] = _falcon_readl(efx, reg + 0);
++ rmb();
++ value->u32[1] = _falcon_readl(efx, reg + 4);
++ value->u32[2] = _falcon_readl(efx, reg + 8);
++ value->u32[3] = _falcon_readl(efx, reg + 12);
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
+
-+static struct attribute_group tapstat_group = {
-+ .name = "statistics",
-+ .attrs = tapstat_attrs,
-+};
++ EFX_REGDUMP(efx, "read from register %x, got " EFX_OWORD_FMT "\n", reg,
++ EFX_OWORD_VAL(*value));
++}
+
-+int xentap_sysfs_addif(struct xenbus_device *dev)
++/* This reads an 8-byte Falcon SRAM entry in one go. */
++static inline void falcon_read_sram(struct efx_nic *efx, efx_qword_t *value,
++ unsigned int index)
+{
-+ int err;
-+ struct backend_info *be = dev->dev.driver_data;
-+ err = sysfs_create_group(&dev->dev.kobj, &tapstat_group);
-+ if (!err)
-+ be->group_added = 1;
-+ return err;
++ unsigned int reg = efx->type->buf_tbl_base + (index * sizeof(*value));
++ unsigned long flags __attribute__ ((unused));
++
++ spin_lock_irqsave(&efx->biu_lock, flags);
++#ifdef FALCON_USE_QWORD_IO
++ value->u64[0] = _falcon_readq(efx, reg + 0);
++#else
++ value->u32[0] = _falcon_readl(efx, reg + 0);
++ rmb();
++ value->u32[1] = _falcon_readl(efx, reg + 4);
++#endif
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
++
++ EFX_REGDUMP(efx, "read from SRAM register %x, got "EFX_QWORD_FMT"\n",
++ reg, EFX_QWORD_VAL(*value));
+}
+
-+void xentap_sysfs_delif(struct xenbus_device *dev)
++/* Read dword from Falcon register that allows partial writes (sic) */
++static inline void falcon_readl(struct efx_nic *efx, efx_dword_t *value,
++ unsigned int reg)
+{
-+ struct backend_info *be = dev->dev.driver_data;
-+ sysfs_remove_group(&dev->dev.kobj, &tapstat_group);
-+ be->group_added = 0;
++ value->u32[0] = _falcon_readl(efx, reg);
++ EFX_REGDUMP(efx, "read from register %x, got "EFX_DWORD_FMT"\n",
++ reg, EFX_DWORD_VAL(*value));
+}
+
-+static int blktap_remove(struct xenbus_device *dev)
++/* Write to a register forming part of a table */
++static inline void falcon_write_table(struct efx_nic *efx, efx_oword_t *value,
++ unsigned int reg, unsigned int index)
+{
-+ struct backend_info *be = dev->dev.driver_data;
++ falcon_write(efx, value, reg + index * sizeof(efx_oword_t));
++}
+
-+ if (be->group_added)
-+ xentap_sysfs_delif(be->dev);
-+ if (be->backend_watch.node) {
-+ unregister_xenbus_watch(&be->backend_watch);
-+ kfree(be->backend_watch.node);
-+ be->backend_watch.node = NULL;
-+ }
-+ if (be->blkif) {
-+ if (be->blkif->xenblkd)
-+ kthread_stop(be->blkif->xenblkd);
-+ signal_tapdisk(be->blkif->dev_num);
-+ tap_blkif_free(be->blkif);
-+ be->blkif = NULL;
-+ }
-+ kfree(be);
-+ dev->dev.driver_data = NULL;
-+ return 0;
++/* Read to a register forming part of a table */
++static inline void falcon_read_table(struct efx_nic *efx, efx_oword_t *value,
++ unsigned int reg, unsigned int index)
++{
++ falcon_read(efx, value, reg + index * sizeof(efx_oword_t));
+}
+
-+static void tap_update_blkif_status(blkif_t *blkif)
-+{
-+ int err;
-+ char name[TASK_COMM_LEN];
++/* Write to a dword register forming part of a table */
++static inline void falcon_writel_table(struct efx_nic *efx, efx_dword_t *value,
++ unsigned int reg, unsigned int index)
++{
++ falcon_writel(efx, value, reg + index * sizeof(efx_oword_t));
++}
+
-+ /* Not ready to connect? */
-+ if(!blkif->irq || !blkif->sectors) {
-+ return;
-+ }
++/* Page-mapped register block size */
++#define FALCON_PAGE_BLOCK_SIZE 0x2000
+
-+ /* Already connected? */
-+ if (blkif->be->dev->state == XenbusStateConnected)
-+ return;
++/* Calculate offset to page-mapped register block */
++#define FALCON_PAGED_REG(page, reg) \
++ ((page) * FALCON_PAGE_BLOCK_SIZE + (reg))
+
-+ /* Attempt to connect: exit if we fail to. */
-+ connect(blkif->be);
-+ if (blkif->be->dev->state != XenbusStateConnected)
-+ return;
++/* As for falcon_write(), but for a page-mapped register. */
++static inline void falcon_write_page(struct efx_nic *efx, efx_oword_t *value,
++ unsigned int reg, unsigned int page)
++{
++ falcon_write(efx, value, FALCON_PAGED_REG(page, reg));
++}
+
-+ err = blktap_name(blkif, name);
-+ if (err) {
-+ xenbus_dev_error(blkif->be->dev, err, "get blktap dev name");
-+ return;
-+ }
++/* As for falcon_writel(), but for a page-mapped register. */
++static inline void falcon_writel_page(struct efx_nic *efx, efx_dword_t *value,
++ unsigned int reg, unsigned int page)
++{
++ falcon_writel(efx, value, FALCON_PAGED_REG(page, reg));
++}
+
-+ if (!blkif->be->group_added) {
-+ err = xentap_sysfs_addif(blkif->be->dev);
-+ if (err) {
-+ xenbus_dev_fatal(blkif->be->dev, err,
-+ "creating sysfs entries");
-+ return;
-+ }
-+ }
++/* Write dword to Falcon page-mapped register with an extra lock.
++ *
++ * As for falcon_writel_page(), but for a register that suffers from
++ * SFC bug 3181. Take out a lock so the BIU collector cannot be
++ * confused. */
++static inline void falcon_writel_page_locked(struct efx_nic *efx,
++ efx_dword_t *value,
++ unsigned int reg,
++ unsigned int page)
++{
++ unsigned long flags __attribute__ ((unused));
+
-+ blkif->xenblkd = kthread_run(tap_blkif_schedule, blkif, name);
-+ if (IS_ERR(blkif->xenblkd)) {
-+ err = PTR_ERR(blkif->xenblkd);
-+ blkif->xenblkd = NULL;
-+ xenbus_dev_fatal(blkif->be->dev, err, "start xenblkd");
-+ WPRINTK("Error starting thread\n");
-+ }
++ spin_lock_irqsave(&efx->biu_lock, flags);
++ falcon_writel(efx, value, FALCON_PAGED_REG(page, reg));
++ spin_unlock_irqrestore(&efx->biu_lock, flags);
+}
+
-+/**
-+ * Entry point to this code when a new device is created. Allocate
-+ * the basic structures, and watch the store waiting for the
-+ * user-space program to tell us the physical device info. Switch to
-+ * InitWait.
++#endif /* EFX_FALCON_IO_H */
+--- linux-2.6.18.8/drivers/net/sfc/falcon_xmac.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/falcon_xmac.c 2008-05-19 00:33:28.849809063 +0300
+@@ -0,0 +1,691 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static int blktap_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
-+{
-+ int err;
-+ struct backend_info *be = kzalloc(sizeof(struct backend_info),
-+ GFP_KERNEL);
-+ if (!be) {
-+ xenbus_dev_fatal(dev, -ENOMEM,
-+ "allocating backend structure");
-+ return -ENOMEM;
-+ }
+
-+ be->dev = dev;
-+ dev->dev.driver_data = be;
-+ be->xenbus_id = get_id(dev->nodename);
++#include <linux/delay.h>
++#include "net_driver.h"
++#include "efx.h"
++#include "falcon.h"
++#include "falcon_hwdefs.h"
++#include "falcon_io.h"
++#include "mac.h"
++#include "gmii.h"
++#include "mdio_10g.h"
++#include "phy.h"
++#include "boards.h"
++#include "workarounds.h"
+
-+ be->blkif = tap_alloc_blkif(dev->otherend_id);
-+ if (IS_ERR(be->blkif)) {
-+ err = PTR_ERR(be->blkif);
-+ be->blkif = NULL;
-+ xenbus_dev_fatal(dev, err, "creating block interface");
-+ goto fail;
-+ }
++/**************************************************************************
++ *
++ * MAC register access
++ *
++ **************************************************************************/
+
-+ /* setup back pointer */
-+ be->blkif->be = be;
-+ be->blkif->sectors = 0;
++/* Offset of an XMAC register within Falcon */
++#define FALCON_XMAC_REG(mac_reg) \
++ (FALCON_XMAC_REGBANK + ((mac_reg) * FALCON_XMAC_REG_SIZE))
+
-+ /* set a watch on disk info, waiting for userspace to update details*/
-+ err = xenbus_watch_path2(dev, dev->nodename, "info",
-+ &be->backend_watch, tap_backend_changed);
-+ if (err)
-+ goto fail;
-+
-+ err = xenbus_switch_state(dev, XenbusStateInitWait);
-+ if (err)
-+ goto fail;
-+ return 0;
++static void falcon_xmac_writel(struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg)
++{
++ efx_oword_t temp;
+
-+fail:
-+ DPRINTK("blktap probe failed\n");
-+ blktap_remove(dev);
-+ return err;
++ EFX_POPULATE_OWORD_1(temp, MAC_DATA, EFX_DWORD_FIELD(*value, MAC_DATA));
++ falcon_write(efx, &temp, FALCON_XMAC_REG(mac_reg));
+}
+
-+
-+/**
-+ * Callback received when the user space code has placed the device
-+ * information in xenstore.
-+ */
-+static void tap_backend_changed(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
++static void falcon_xmac_readl(struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg)
+{
-+ int err;
-+ unsigned long info;
-+ struct backend_info *be
-+ = container_of(watch, struct backend_info, backend_watch);
-+ struct xenbus_device *dev = be->dev;
-+
-+ /**
-+ * Check to see whether userspace code has opened the image
-+ * and written sector
-+ * and disk info to xenstore
-+ */
-+ err = xenbus_gather(XBT_NIL, dev->nodename, "info", "%lu", &info,
-+ NULL);
-+ if (XENBUS_EXIST_ERR(err))
-+ return;
-+ if (err) {
-+ xenbus_dev_error(dev, err, "getting info");
-+ return;
-+ }
++ efx_oword_t temp;
+
-+ DPRINTK("Userspace update on disk info, %lu\n",info);
++ falcon_read(efx, &temp, FALCON_XMAC_REG(mac_reg));
++ EFX_POPULATE_DWORD_1(*value, MAC_DATA, EFX_OWORD_FIELD(temp, MAC_DATA));
++}
+
-+ err = xenbus_gather(XBT_NIL, dev->nodename, "sectors", "%llu",
-+ &be->blkif->sectors, NULL);
++/**************************************************************************
++ *
++ * MAC operations
++ *
++ *************************************************************************/
++static int falcon_reset_xmac(struct efx_nic *efx)
++{
++ efx_dword_t reg;
++ int count;
+
-+ /* Associate tap dev with domid*/
-+ be->blkif->dev_num = dom_to_devid(be->blkif->domid, be->xenbus_id,
-+ be->blkif);
-+ DPRINTK("Thread started for domid [%d], connecting disk\n",
-+ be->blkif->dev_num);
++ EFX_POPULATE_DWORD_1(reg, XM_CORE_RST, 1);
++ efx->mac_op->mac_writel(efx, ®, XM_GLB_CFG_REG_MAC);
+
-+ tap_update_blkif_status(be->blkif);
++ for (count = 0; count < 10000; count++) { /* wait upto 100ms */
++ efx->mac_op->mac_readl(efx, ®, XM_GLB_CFG_REG_MAC);
++ if (EFX_DWORD_FIELD(reg, XM_CORE_RST) == 0)
++ return 0;
++ udelay(10);
++ }
++
++ EFX_ERR(efx, "timed out waiting for XMAC core reset\n");
++ return -ETIMEDOUT;
+}
+
-+/**
-+ * Callback received when the frontend's state changes.
-+ */
-+static void tap_frontend_changed(struct xenbus_device *dev,
-+ enum xenbus_state frontend_state)
++/* Configure the XAUI driver that is an output from Falcon */
++static void falcon_setup_xaui(struct efx_nic *efx)
+{
-+ struct backend_info *be = dev->dev.driver_data;
-+ int err;
++ efx_dword_t sdctl, txdrv;
+
-+ DPRINTK("\n");
++ /* Move the XAUI into low power, unless there is no PHY, in
++ * which case the XAUI will have to drive a cable. */
++ if (efx->phy_type == PHY_TYPE_NONE)
++ return;
+
-+ switch (frontend_state) {
-+ case XenbusStateInitialising:
-+ if (dev->state == XenbusStateClosed) {
-+ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
-+ __FUNCTION__, dev->nodename);
-+ xenbus_switch_state(dev, XenbusStateInitWait);
-+ }
-+ break;
++ efx->mac_op->mac_readl(efx, &sdctl, XX_SD_CTL_REG_MAC);
++ EFX_SET_DWORD_FIELD(sdctl, XX_HIDRVD, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_LODRVD, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_HIDRVC, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_LODRVC, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_HIDRVB, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_LODRVB, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_HIDRVA, XX_SD_CTL_DRV_DEFAULT);
++ EFX_SET_DWORD_FIELD(sdctl, XX_LODRVA, XX_SD_CTL_DRV_DEFAULT);
++ efx->mac_op->mac_writel(efx, &sdctl, XX_SD_CTL_REG_MAC);
++
++ EFX_POPULATE_DWORD_8(txdrv,
++ XX_DEQD, XX_TXDRV_DEQ_DEFAULT,
++ XX_DEQC, XX_TXDRV_DEQ_DEFAULT,
++ XX_DEQB, XX_TXDRV_DEQ_DEFAULT,
++ XX_DEQA, XX_TXDRV_DEQ_DEFAULT,
++ XX_DTXD, XX_TXDRV_DTX_DEFAULT,
++ XX_DTXC, XX_TXDRV_DTX_DEFAULT,
++ XX_DTXB, XX_TXDRV_DTX_DEFAULT,
++ XX_DTXA, XX_TXDRV_DTX_DEFAULT);
++ efx->mac_op->mac_writel(efx, &txdrv, XX_TXDRV_CTL_REG_MAC);
++}
++
++static void falcon_hold_xaui_in_rst(struct efx_nic *efx)
++{
++ efx_dword_t reg;
++
++ EFX_ZERO_DWORD(reg);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNA_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNB_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNC_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDND_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTPLLAB_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTPLLCD_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETA_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETB_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETC_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETD_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSRX_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSTX_EN, 1);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
++}
++
++static int _falcon_reset_xaui_a(struct efx_nic *efx)
++{
++ efx_dword_t reg;
++
++ if (!efx->is_asic)
++ return 0;
+
-+ case XenbusStateInitialised:
-+ case XenbusStateConnected:
-+ /* Ensure we connect even when two watches fire in
-+ close successsion and we miss the intermediate value
-+ of frontend_state. */
-+ if (dev->state == XenbusStateConnected)
-+ break;
++ falcon_hold_xaui_in_rst(efx);
++ efx->mac_op->mac_readl(efx, ®, XX_PWR_RST_REG_MAC);
+
-+ err = connect_ring(be);
-+ if (err)
-+ break;
-+ tap_update_blkif_status(be->blkif);
-+ break;
++ /* Follow the RAMBUS XAUI data reset sequencing
++ * Channels A and B first: power down, reset PLL, reset, clear
++ */
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNA_EN, 0);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNB_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+ case XenbusStateClosing:
-+ if (be->blkif->xenblkd) {
-+ kthread_stop(be->blkif->xenblkd);
-+ be->blkif->xenblkd = NULL;
-+ }
-+ xenbus_switch_state(dev, XenbusStateClosing);
-+ break;
++ EFX_SET_DWORD_FIELD(reg, XX_RSTPLLAB_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+ case XenbusStateClosed:
-+ xenbus_switch_state(dev, XenbusStateClosed);
-+ if (xenbus_dev_is_online(dev))
-+ break;
-+ /* fall through if not online */
-+ case XenbusStateUnknown:
-+ device_unregister(&dev->dev);
-+ break;
++ EFX_SET_DWORD_FIELD(reg, XX_RESETA_EN, 0);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETB_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+ default:
-+ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
-+ frontend_state);
-+ break;
-+ }
-+}
++ /* Channels C and D: power down, reset PLL, reset, clear */
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDNC_EN, 0);
++ EFX_SET_DWORD_FIELD(reg, XX_PWRDND_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
++ EFX_SET_DWORD_FIELD(reg, XX_RSTPLLCD_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+/**
-+ * Switch to Connected state.
-+ */
-+static void connect(struct backend_info *be)
-+{
-+ int err;
++ EFX_SET_DWORD_FIELD(reg, XX_RESETC_EN, 0);
++ EFX_SET_DWORD_FIELD(reg, XX_RESETD_EN, 0);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+ struct xenbus_device *dev = be->dev;
++ /* Setup XAUI */
++ falcon_setup_xaui(efx);
++ udelay(10);
+
-+ err = xenbus_switch_state(dev, XenbusStateConnected);
-+ if (err)
-+ xenbus_dev_fatal(dev, err, "switching to Connected state",
-+ dev->nodename);
++ /* Take XGXS out of reset */
++ EFX_ZERO_DWORD(reg);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
+
-+ return;
++ return 0;
+}
+
-+
-+static int connect_ring(struct backend_info *be)
++static int _falcon_reset_xaui_b(struct efx_nic *efx)
+{
-+ struct xenbus_device *dev = be->dev;
-+ unsigned long ring_ref;
-+ unsigned int evtchn;
-+ char protocol[64];
-+ int err;
++ efx_dword_t reg;
++ int count;
+
-+ DPRINTK("%s\n", dev->otherend);
++ if (!efx->is_asic)
++ return 0;
+
-+ err = xenbus_gather(XBT_NIL, dev->otherend, "ring-ref", "%lu",
-+ &ring_ref, "event-channel", "%u", &evtchn, NULL);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err,
-+ "reading %s/ring-ref and event-channel",
-+ dev->otherend);
-+ return err;
-+ }
++ EFX_POPULATE_DWORD_1(reg, XX_RST_XX_EN, 1);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
+
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
-+ err = xenbus_gather(XBT_NIL, dev->otherend, "protocol",
-+ "%63s", protocol, NULL);
-+ if (err)
-+ strcpy(protocol, "unspecified, assuming native");
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_NATIVE))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_32))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_32;
-+ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_64))
-+ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_64;
-+ else {
-+ xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol);
-+ return -1;
++ /* Give some time for the link to establish */
++ for (count = 0; count < 1000; count++) { /* wait upto 10ms */
++ efx->mac_op->mac_readl(efx, ®, XX_PWR_RST_REG_MAC);
++ if (EFX_DWORD_FIELD(reg, XX_RST_XX_EN) == 0) {
++ falcon_setup_xaui(efx);
++ return 0;
++ }
++ udelay(10);
+ }
-+ printk(KERN_INFO
-+ "blktap: ring-ref %ld, event-channel %d, protocol %d (%s)\n",
-+ ring_ref, evtchn, be->blkif->blk_protocol, protocol);
++ EFX_ERR(efx, "timed out waiting for XAUI/XGXS reset\n");
++ return -ETIMEDOUT;
++}
+
-+ /* Map the shared frame, irq etc. */
-+ err = tap_blkif_map(be->blkif, ring_ref, evtchn);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "mapping ring-ref %lu port %u",
-+ ring_ref, evtchn);
-+ return err;
-+ }
++int falcon_reset_xaui(struct efx_nic *efx)
++{
++ int rc;
+
-+ return 0;
++ if (EFX_WORKAROUND_9388(efx)) {
++ falcon_hold_xaui_in_rst(efx);
++ efx->phy_op->reset_xaui(efx);
++ rc = _falcon_reset_xaui_a(efx);
++ } else {
++ rc = _falcon_reset_xaui_b(efx);
++ }
++ return rc;
+}
+
++static int falcon_init_xmac(struct efx_nic *efx)
++{
++ int rc;
+
-+/* ** Driver Registration ** */
-+
++ /* Initialize the PHY first so the clock is around */
++ rc = efx->phy_op->init(efx);
++ if (rc)
++ goto fail1;
+
-+static struct xenbus_device_id blktap_ids[] = {
-+ { "tap" },
-+ { "" }
-+};
++ rc = falcon_reset_xaui(efx);
++ if (rc)
++ goto fail2;
+
++ /* Wait again. Give the PHY and MAC time to faff */
++ schedule_timeout_uninterruptible(HZ / 10);
+
-+static struct xenbus_driver blktap = {
-+ .name = "tap",
-+ .owner = THIS_MODULE,
-+ .ids = blktap_ids,
-+ .probe = blktap_probe,
-+ .remove = blktap_remove,
-+ .otherend_changed = tap_frontend_changed
-+};
++ /* Reset the MAC */
++ rc = falcon_reset_xmac(efx);
++ if (rc)
++ goto fail2;
+
++ return 0;
+
-+void tap_blkif_xenbus_init(void)
-+{
-+ xenbus_register_backend(&blktap);
++ fail2:
++ efx->phy_op->fini(efx);
++ fail1:
++ return rc;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/char/Makefile linux-2.6.18-xen.hg/drivers/xen/char/Makefile
---- linux-2.6.18/drivers/xen/char/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/char/Makefile 2007-12-23 11:15:33.547908998 +0100
-@@ -0,0 +1 @@
-+obj-$(CONFIG_XEN_DEVMEM) := mem.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/char/mem.c linux-2.6.18-xen.hg/drivers/xen/char/mem.c
---- linux-2.6.18/drivers/xen/char/mem.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/char/mem.c 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,190 @@
-+/*
-+ * Originally from linux/drivers/char/mem.c
-+ *
-+ * Copyright (C) 1991, 1992 Linus Torvalds
-+ *
-+ * Added devfs support.
-+ * Jan-11-1998, C. Scott Ananian <cananian@alumni.princeton.edu>
-+ * Shared /dev/zero mmaping support, Feb 2000, Kanoj Sarcar <kanoj@sgi.com>
-+ */
-+
-+#include <linux/mm.h>
-+#include <linux/miscdevice.h>
-+#include <linux/slab.h>
-+#include <linux/vmalloc.h>
-+#include <linux/mman.h>
-+#include <linux/random.h>
-+#include <linux/init.h>
-+#include <linux/raw.h>
-+#include <linux/tty.h>
-+#include <linux/capability.h>
-+#include <linux/smp_lock.h>
-+#include <linux/ptrace.h>
-+#include <linux/device.h>
-+#include <asm/pgalloc.h>
-+#include <asm/uaccess.h>
-+#include <asm/io.h>
-+#include <asm/hypervisor.h>
+
-+static inline int uncached_access(struct file *file)
++/* Get status of XAUI link */
++int falcon_xaui_link_ok(struct efx_nic *efx)
+{
-+ if (file->f_flags & O_SYNC)
++ efx_dword_t reg;
++ int align_done;
++ int sync_status;
++ int link_ok = 0;
++
++ /* If we're in internal loopback, then the link is up.
++ * The A1 FPGA/4G has RX and TX XAUI wired together, so the link is up.
++ * The B0 FPGA has XAUI offchip, so it is always up.
++ */
++ if (!efx->is_asic || LOOPBACK_INTERNAL(efx))
+ return 1;
-+ /* Xen sets correct MTRR type on non-RAM for us. */
-+ return 0;
-+}
+
-+/*
-+ * This funcion reads the *physical* memory. The f_pos points directly to the
-+ * memory location.
-+ */
-+static ssize_t read_mem(struct file * file, char __user * buf,
-+ size_t count, loff_t *ppos)
-+{
-+ unsigned long p = *ppos, ignored;
-+ ssize_t read = 0, sz;
-+ void __iomem *v;
++ /* Read link status */
++ efx->mac_op->mac_readl(efx, ®, XX_CORE_STAT_REG_MAC);
++
++ align_done = EFX_DWORD_FIELD(reg, XX_ALIGN_DONE);
++ sync_status = EFX_DWORD_FIELD(reg, XX_SYNC_STAT);
++ if (align_done && (sync_status == XX_SYNC_STAT_DECODE_SYNCED))
++ link_ok = 1;
++
++ /* Clear link status ready for next read */
++ EFX_SET_DWORD_FIELD(reg, XX_COMMA_DET, XX_COMMA_DET_RESET);
++ EFX_SET_DWORD_FIELD(reg, XX_CHARERR, XX_CHARERR_RESET);
++ EFX_SET_DWORD_FIELD(reg, XX_DISPERR, XX_DISPERR_RESET);
++
++ efx->mac_op->mac_writel(efx, ®, XX_CORE_STAT_REG_MAC);
++
++ return link_ok;
++}
++
++/* Do most of the heavy lifting of falcon_reconfigure_xmac */
++static void falcon_reconfigure_xmac_core(struct efx_nic *efx)
++{
++ unsigned int max_frame_len;
++ efx_dword_t reg;
++ efx_oword_t mac_test_reg;
++ int rx_fc = (efx->flow_control & EFX_FC_RX) ? 1 : 0;
++
++ if (FALCON_REV(efx) <= FALCON_REV_A1 && !efx->is_asic) {
++ /* 10G FPGA's have the XAUI TX and RX wired together. Fake
++ * the link status and configure the link options before
++ * the MAC wrapper is configured */
++ efx->link_options = GM_LPA_10000FULL;
++ efx->link_up = 1;
++ }
++
++ /* Configure MAC - cut-thru mode is hard wired on */
++ EFX_POPULATE_DWORD_3(reg,
++ XM_RX_JUMBO_MODE, 1,
++ XM_TX_STAT_EN, 1,
++ XM_RX_STAT_EN, 1);
++ efx->mac_op->mac_writel(efx, ®, XM_GLB_CFG_REG_MAC);
++
++ /* Configure TX */
++ EFX_POPULATE_DWORD_6(reg,
++ XM_TXEN, 1,
++ XM_TX_PRMBL, 1,
++ XM_AUTO_PAD, 1,
++ XM_TXCRC, 1,
++ XM_FCNTL, 1,
++ XM_IPG, 0x3);
++ efx->mac_op->mac_writel(efx, ®, XM_TX_CFG_REG_MAC);
++
++ /* Configure RX */
++ EFX_POPULATE_DWORD_5(reg,
++ XM_RXEN, 1,
++ XM_AUTO_DEPAD, 0,
++ XM_ACPT_ALL_MCAST, 1,
++ XM_ACPT_ALL_UCAST, efx->promiscuous,
++ XM_PASS_CRC_ERR, 1);
++ efx->mac_op->mac_writel(efx, ®, XM_RX_CFG_REG_MAC);
++
++ /* Set frame length */
++ max_frame_len = EFX_MAX_FRAME_LEN(efx->net_dev->mtu);
++ EFX_POPULATE_DWORD_1(reg, XM_MAX_RX_FRM_SIZE, max_frame_len);
++ efx->mac_op->mac_writel(efx, ®, XM_RX_PARAM_REG_MAC);
++ EFX_POPULATE_DWORD_2(reg,
++ XM_MAX_TX_FRM_SIZE, max_frame_len,
++ XM_TX_JUMBO_MODE, 1);
++ efx->mac_op->mac_writel(efx, ®, XM_TX_PARAM_REG_MAC);
++
++ EFX_POPULATE_DWORD_2(reg,
++ XM_PAUSE_TIME, 0xfffe, /* MAX PAUSE TIME */
++ XM_DIS_FCNTL, rx_fc ? 0 : 1);
++ efx->mac_op->mac_writel(efx, ®, XM_FC_REG_MAC);
++
++ /* Set MAC address */
++ EFX_POPULATE_DWORD_4(reg,
++ XM_ADR_0, efx->net_dev->dev_addr[0],
++ XM_ADR_1, efx->net_dev->dev_addr[1],
++ XM_ADR_2, efx->net_dev->dev_addr[2],
++ XM_ADR_3, efx->net_dev->dev_addr[3]);
++ efx->mac_op->mac_writel(efx, ®, XM_ADR_LO_REG_MAC);
++ EFX_POPULATE_DWORD_2(reg,
++ XM_ADR_4, efx->net_dev->dev_addr[4],
++ XM_ADR_5, efx->net_dev->dev_addr[5]);
++ efx->mac_op->mac_writel(efx, ®, XM_ADR_HI_REG_MAC);
++
++ /* Handle B0 FPGA loopback where RAMBUS XGXS block not present */
++ if (FALCON_REV(efx) == FALCON_REV_B0 && !efx->is_asic) {
++ int xgmii_loopback =
++ (efx->loopback_mode == LOOPBACK_XGMII) ? 1 : 0;
++
++ /* Set the MAC loopback bit. */
++ EFX_POPULATE_OWORD_1(mac_test_reg,
++ MAC_PTLOOP_EN, xgmii_loopback);
++ falcon_write(efx, &mac_test_reg, MAC_TEST_REG_KER);
++ }
++}
++
++/* Do most of the heavy lifting of falcon_reconfigure_xmac */
++static void falcon_reconfigure_xgxs_core(struct efx_nic *efx)
++{
++ efx_dword_t reg;
++ int xgxs_loopback = (efx->loopback_mode == LOOPBACK_XGXS) ? 1 : 0;
++ int xaui_loopback = (efx->loopback_mode == LOOPBACK_XAUI) ? 1 : 0;
++ int xgmii_loopback =
++ (efx->loopback_mode == LOOPBACK_XGMII) ? 1 : 0;
++
++ if (FALCON_REV(efx) == FALCON_REV_B0 && !efx->is_asic)
++ /* RAMBUS XGXS block is not present */
++ return;
+
-+ while (count > 0) {
-+ /*
-+ * Handle first page in case it's not aligned
-+ */
-+ if (-p & (PAGE_SIZE - 1))
-+ sz = -p & (PAGE_SIZE - 1);
-+ else
-+ sz = PAGE_SIZE;
++ /* XGXS block is flaky and will need to be reset if moving
++ * into our out of XGMII, XGXS or XAUI loopbacks. */
++ if (EFX_WORKAROUND_5147(efx)) {
++ int old_xgmii_loopback, old_xgxs_loopback, old_xaui_loopback;
++ int reset_xgxs;
++
++ efx->mac_op->mac_readl(efx, ®,
++ XX_CORE_STAT_REG_MAC);
++ old_xgxs_loopback = EFX_DWORD_FIELD(reg, XX_XGXS_LB_EN);
++ old_xgmii_loopback = EFX_DWORD_FIELD(reg, XX_XGMII_LB_EN);
++
++ efx->mac_op->mac_readl(efx, ®, XX_SD_CTL_REG_MAC);
++ old_xaui_loopback = EFX_DWORD_FIELD(reg, XX_LPBKA);
++
++ /* The PHY driver may have turned XAUI off */
++ reset_xgxs = ((xgxs_loopback != old_xgxs_loopback) ||
++ (xaui_loopback != old_xaui_loopback) ||
++ (xgmii_loopback != old_xgmii_loopback));
++ if (reset_xgxs) {
++ efx->mac_op->mac_readl(efx, ®,
++ XX_PWR_RST_REG_MAC);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSTX_EN, 1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSRX_EN, 1);
++ efx->mac_op->mac_writel(efx, ®,
++ XX_PWR_RST_REG_MAC);
++ udelay(1);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSTX_EN, 0);
++ EFX_SET_DWORD_FIELD(reg, XX_RSTXGXSRX_EN, 0);
++ efx->mac_op->mac_writel(efx, ®,
++ XX_PWR_RST_REG_MAC);
++ udelay(1);
++ }
++ }
++
++ efx->mac_op->mac_readl(efx, ®, XX_CORE_STAT_REG_MAC);
++ EFX_SET_DWORD_FIELD(reg, XX_FORCE_SIG,
++ (xgxs_loopback || xaui_loopback) ?
++ XX_FORCE_SIG_DECODE_FORCED : 0);
++ EFX_SET_DWORD_FIELD(reg, XX_XGXS_LB_EN, xgxs_loopback);
++ EFX_SET_DWORD_FIELD(reg, XX_XGMII_LB_EN, xgmii_loopback);
++ efx->mac_op->mac_writel(efx, ®, XX_CORE_STAT_REG_MAC);
++
++ efx->mac_op->mac_readl(efx, ®, XX_SD_CTL_REG_MAC);
++ EFX_SET_DWORD_FIELD(reg, XX_LPBKD, xaui_loopback);
++ EFX_SET_DWORD_FIELD(reg, XX_LPBKC, xaui_loopback);
++ EFX_SET_DWORD_FIELD(reg, XX_LPBKB, xaui_loopback);
++ EFX_SET_DWORD_FIELD(reg, XX_LPBKA, xaui_loopback);
++ efx->mac_op->mac_writel(efx, ®, XX_SD_CTL_REG_MAC);
++}
++
++
++/* Sometimes the XAUI link between Falcon and XFP fails to come up. The state
++ * of the link is checked during phy_reconfigure(). After XAIU is reset then
++ * the MAC must be reconfigured.
++ */
++#define MAX_XAUI_TRIES (5) /* It's never been seen to take more than 2 */
++
++void falcon_check_xaui_link_up(struct efx_nic *efx)
++{
++ int max_tries, tries;
++ tries = EFX_WORKAROUND_5147(efx) ? MAX_XAUI_TRIES : 1;
++ max_tries = tries;
++
++ if ((efx->loopback_mode == LOOPBACK_NETWORK) ||
++ (efx->phy_type == PHY_TYPE_NONE) ||
++ !efx->phy_powered)
++ return;
+
-+ sz = min_t(unsigned long, sz, count);
++ while (tries) {
++ if (falcon_xaui_link_ok(efx))
++ return;
+
-+ v = ioremap(p, sz);
-+ if (IS_ERR(v) || v == NULL) {
-+ /*
-+ * Some programs (e.g., dmidecode) groove off into
-+ * weird RAM areas where no tables can possibly exist
-+ * (because Xen will have stomped on them!). These
-+ * programs get rather upset if we let them know that
-+ * Xen failed their access, so we fake out a read of
-+ * all zeroes.
-+ */
-+ if (clear_user(buf, count))
-+ return -EFAULT;
-+ read += count;
-+ break;
-+ }
++ EFX_LOG(efx, "%s Clobbering XAUI (%d tries left).\n",
++ __func__, tries);
++ (void) falcon_reset_xaui(efx);
++ /* Cannot use full reconfigure. Need to avoid recursion */
+
-+ ignored = copy_to_user(buf, v, sz);
-+ iounmap(v);
-+ if (ignored)
-+ return -EFAULT;
-+ buf += sz;
-+ p += sz;
-+ count -= sz;
-+ read += sz;
++ /* Give the poor thing time to sort itself out: if we retry
++ * too fast it will never train. */
++ udelay(200);
++
++ falcon_reconfigure_xgxs_core(efx);
++
++ tries--;
+ }
+
-+ *ppos += read;
-+ return read;
++ EFX_ERR(efx, "Failed to bring XAUI link back up in %d tries!\n",
++ max_tries);
+}
+
-+static ssize_t write_mem(struct file * file, const char __user * buf,
-+ size_t count, loff_t *ppos)
++static void falcon_reconfigure_xmac(struct efx_nic *efx)
+{
-+ unsigned long p = *ppos, ignored;
-+ ssize_t written = 0, sz;
-+ void __iomem *v;
++ falcon_deconfigure_mac_wrapper(efx);
+
-+ while (count > 0) {
-+ /*
-+ * Handle first page in case it's not aligned
-+ */
-+ if (-p & (PAGE_SIZE - 1))
-+ sz = -p & (PAGE_SIZE - 1);
-+ else
-+ sz = PAGE_SIZE;
++ /* In internal loopback modes disable transmit */
++ efx->tx_disabled = LOOPBACK_INTERNAL(efx);
+
-+ sz = min_t(unsigned long, sz, count);
++ efx->phy_op->reconfigure(efx);
+
-+ v = ioremap(p, sz);
-+ if (v == NULL)
-+ break;
-+ if (IS_ERR(v)) {
-+ if (written == 0)
-+ return PTR_ERR(v);
-+ break;
-+ }
++ falcon_reconfigure_xgxs_core(efx);
++ falcon_reconfigure_xmac_core(efx);
+
-+ ignored = copy_from_user(v, buf, sz);
-+ iounmap(v);
-+ if (ignored) {
-+ written += sz - ignored;
-+ if (written)
-+ break;
-+ return -EFAULT;
-+ }
-+ buf += sz;
-+ p += sz;
-+ count -= sz;
-+ written += sz;
-+ }
++ /* Reconfigure MAC wrapper */
++ falcon_reconfigure_mac_wrapper(efx);
+
-+ *ppos += written;
-+ return written;
++ /* Ensure XAUI link is up - might repeat reconfigure_xmac_core */
++ falcon_check_xaui_link_up(efx);
+}
+
-+#ifndef ARCH_HAS_DEV_MEM_MMAP_MEM
-+static int xen_mmap_mem(struct file * file, struct vm_area_struct * vma)
++static void falcon_fini_xmac(struct efx_nic *efx)
+{
-+ size_t size = vma->vm_end - vma->vm_start;
-+
-+ if (uncached_access(file))
-+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
++ /* Isolate the MAC - PHY */
++ falcon_deconfigure_mac_wrapper(efx);
+
-+ /* We want to return the real error code, not EAGAIN. */
-+ return direct_remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
-+ size, vma->vm_page_prot, DOMID_IO);
++ /* Potentially power down the PHY */
++ efx->phy_op->fini(efx);
+}
-+#endif
+
-+/*
-+ * The memory devices use the full 32/64 bits of the offset, and so we cannot
-+ * check against negative addresses: they are ok. The return value is weird,
-+ * though, in that case (0).
-+ *
-+ * also note that seeking relative to the "end of file" isn't supported:
-+ * it has no meaning, so it returns -EINVAL.
-+ */
-+static loff_t memory_lseek(struct file * file, loff_t offset, int orig)
++static void falcon_update_stats_xmac(struct efx_nic *efx)
+{
-+ loff_t ret;
++ struct efx_mac_stats *mac_stats = &efx->mac_stats;
++ int rc;
+
-+ mutex_lock(&file->f_dentry->d_inode->i_mutex);
-+ switch (orig) {
-+ case 0:
-+ file->f_pos = offset;
-+ ret = file->f_pos;
-+ force_successful_syscall_return();
-+ break;
-+ case 1:
-+ file->f_pos += offset;
-+ ret = file->f_pos;
-+ force_successful_syscall_return();
-+ break;
-+ default:
-+ ret = -EINVAL;
++ rc = falcon_dma_stats(efx, XgDmaDone_offset);
++ if (rc)
++ return;
++
++ /* Update MAC stats from DMAed values */
++ FALCON_STAT(efx, XgRxOctets, rx_bytes);
++ FALCON_STAT(efx, XgRxOctetsOK, rx_good_bytes);
++ FALCON_STAT(efx, XgRxPkts, rx_packets);
++ FALCON_STAT(efx, XgRxPktsOK, rx_good);
++ FALCON_STAT(efx, XgRxBroadcastPkts, rx_broadcast);
++ FALCON_STAT(efx, XgRxMulticastPkts, rx_multicast);
++ FALCON_STAT(efx, XgRxUnicastPkts, rx_unicast);
++ FALCON_STAT(efx, XgRxUndersizePkts, rx_lt64);
++ FALCON_STAT(efx, XgRxOversizePkts, rx_gtjumbo);
++ FALCON_STAT(efx, XgRxJabberPkts, rx_bad_gtjumbo);
++ FALCON_STAT(efx, XgRxUndersizeFCSerrorPkts, rx_bad_lt64);
++ FALCON_STAT(efx, XgRxDropEvents, rx_overflow);
++ FALCON_STAT(efx, XgRxFCSerrorPkts, rx_bad);
++ FALCON_STAT(efx, XgRxAlignError, rx_align_error);
++ FALCON_STAT(efx, XgRxSymbolError, rx_symbol_error);
++ FALCON_STAT(efx, XgRxInternalMACError, rx_internal_error);
++ FALCON_STAT(efx, XgRxControlPkts, rx_control);
++ FALCON_STAT(efx, XgRxPausePkts, rx_pause);
++ FALCON_STAT(efx, XgRxPkts64Octets, rx_64);
++ FALCON_STAT(efx, XgRxPkts65to127Octets, rx_65_to_127);
++ FALCON_STAT(efx, XgRxPkts128to255Octets, rx_128_to_255);
++ FALCON_STAT(efx, XgRxPkts256to511Octets, rx_256_to_511);
++ FALCON_STAT(efx, XgRxPkts512to1023Octets, rx_512_to_1023);
++ FALCON_STAT(efx, XgRxPkts1024to15xxOctets, rx_1024_to_15xx);
++ FALCON_STAT(efx, XgRxPkts15xxtoMaxOctets, rx_15xx_to_jumbo);
++ FALCON_STAT(efx, XgRxLengthError, rx_length_error);
++ FALCON_STAT(efx, XgTxPkts, tx_packets);
++ FALCON_STAT(efx, XgTxOctets, tx_bytes);
++ FALCON_STAT(efx, XgTxMulticastPkts, tx_multicast);
++ FALCON_STAT(efx, XgTxBroadcastPkts, tx_broadcast);
++ FALCON_STAT(efx, XgTxUnicastPkts, tx_unicast);
++ FALCON_STAT(efx, XgTxControlPkts, tx_control);
++ FALCON_STAT(efx, XgTxPausePkts, tx_pause);
++ FALCON_STAT(efx, XgTxPkts64Octets, tx_64);
++ FALCON_STAT(efx, XgTxPkts65to127Octets, tx_65_to_127);
++ FALCON_STAT(efx, XgTxPkts128to255Octets, tx_128_to_255);
++ FALCON_STAT(efx, XgTxPkts256to511Octets, tx_256_to_511);
++ FALCON_STAT(efx, XgTxPkts512to1023Octets, tx_512_to_1023);
++ FALCON_STAT(efx, XgTxPkts1024to15xxOctets, tx_1024_to_15xx);
++ FALCON_STAT(efx, XgTxPkts1519toMaxOctets, tx_15xx_to_jumbo);
++ FALCON_STAT(efx, XgTxUndersizePkts, tx_lt64);
++ FALCON_STAT(efx, XgTxOversizePkts, tx_gtjumbo);
++ FALCON_STAT(efx, XgTxNonTcpUdpPkt, tx_non_tcpudp);
++ FALCON_STAT(efx, XgTxMacSrcErrPkt, tx_mac_src_error);
++ FALCON_STAT(efx, XgTxIpSrcErrPkt, tx_ip_src_error);
++
++ /* Update derived statistics */
++ mac_stats->tx_good_bytes =
++ (mac_stats->tx_bytes - mac_stats->tx_bad_bytes);
++ mac_stats->rx_bad_bytes =
++ (mac_stats->rx_bytes - mac_stats->rx_good_bytes);
++}
++
++#define EFX_XAUI_RETRAIN_MAX 8
++
++static int falcon_check_xmac(struct efx_nic *efx)
++{
++ unsigned link_ok, phyxs_ok = 1;
++ unsigned has_phyxs = efx->phy_op->mmds & (1 << MDIO_MMD_PHYXS);
++
++ /* Check the remote XAUI link status */
++ link_ok = falcon_xaui_link_ok(efx);
++
++ if ((efx->loopback_mode == LOOPBACK_NETWORK) ||
++ !efx->phy_powered)
++ return 0;
++
++ if (link_ok && has_phyxs && !LOOPBACK_INTERNAL(efx)) {
++ /* Does the PHYXS think we have lane sync? */
++ phyxs_ok = mdio_clause45_phyxgxs_lane_sync(efx);
+ }
-+ mutex_unlock(&file->f_dentry->d_inode->i_mutex);
-+ return ret;
++
++ if (EFX_WORKAROUND_5147(efx) && (!link_ok || !phyxs_ok)) {
++ (void) falcon_reset_xaui(efx);
++ falcon_reconfigure_xgxs_core(efx);
++ }
++
++ /* Call the PHY check_hw routine */
++ efx->phy_op->check_hw(efx);
++ return 0;
+}
+
-+static int open_mem(struct inode * inode, struct file * filp)
++/* Simulate a PHY event */
++static void falcon_xmac_sim_phy_event(struct efx_nic *efx)
+{
-+ return capable(CAP_SYS_RAWIO) ? 0 : -EPERM;
-+}
++ efx_qword_t phy_event;
+
-+const struct file_operations mem_fops = {
-+ .llseek = memory_lseek,
-+ .read = read_mem,
-+ .write = write_mem,
-+ .mmap = xen_mmap_mem,
-+ .open = open_mem,
-+};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/console/console.c linux-2.6.18-xen.hg/drivers/xen/console/console.c
---- linux-2.6.18/drivers/xen/console/console.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/console/console.c 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,731 @@
-+/******************************************************************************
-+ * console.c
-+ *
-+ * Virtual console driver.
-+ *
-+ * Copyright (c) 2002-2004, K A Fraser.
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++ EFX_POPULATE_QWORD_2(phy_event,
++ EV_CODE, GLOBAL_EV_DECODE,
++ XG_PHY_INTR, 1);
++ falcon_generate_event(&efx->channel[0], &phy_event);
++}
+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/errno.h>
-+#include <linux/signal.h>
-+#include <linux/sched.h>
-+#include <linux/interrupt.h>
-+#include <linux/tty.h>
-+#include <linux/tty_flip.h>
-+#include <linux/serial.h>
-+#include <linux/major.h>
-+#include <linux/ptrace.h>
-+#include <linux/ioport.h>
-+#include <linux/mm.h>
-+#include <linux/slab.h>
-+#include <linux/init.h>
-+#include <linux/console.h>
-+#include <linux/bootmem.h>
-+#include <linux/sysrq.h>
-+#include <linux/screen_info.h>
-+#include <linux/vt.h>
-+#include <asm/io.h>
-+#include <asm/irq.h>
-+#include <asm/uaccess.h>
-+#include <xen/interface/xen.h>
-+#include <xen/interface/event_channel.h>
-+#include <asm/hypervisor.h>
-+#include <xen/evtchn.h>
-+#include <xen/xenbus.h>
-+#include <xen/xencons.h>
++static int falcon_xmac_get_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
++{
++ mdio_clause45_get_settings(efx, ecmd);
++ ecmd->transceiver = XCVR_INTERNAL;
++ ecmd->phy_address = efx->mii.phy_id;
++ ecmd->autoneg = AUTONEG_DISABLE;
++ ecmd->duplex = DUPLEX_FULL;
++ return 0;
++}
+
-+/*
-+ * Modes:
-+ * 'xencons=off' [XC_OFF]: Console is disabled.
-+ * 'xencons=tty' [XC_TTY]: Console attached to '/dev/tty[0-9]+'.
-+ * 'xencons=ttyS' [XC_SERIAL]: Console attached to '/dev/ttyS[0-9]+'.
-+ * 'xencons=xvc' [XC_XVC]: Console attached to '/dev/xvc0'.
-+ * default: XC_XVC
-+ *
-+ * NB. In mode XC_TTY, we create dummy consoles for tty2-63. This suppresses
-+ * warnings from standard distro startup scripts.
-+ */
-+static enum {
-+ XC_OFF, XC_TTY, XC_SERIAL, XC_XVC
-+} xc_mode = XC_XVC;
-+static int xc_num = -1;
++static int falcon_xmac_set_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
++{
++ if (ecmd->transceiver != XCVR_INTERNAL)
++ return -EINVAL;
++ if (ecmd->autoneg != AUTONEG_DISABLE)
++ return -EINVAL;
++ if (ecmd->duplex != DUPLEX_FULL)
++ return -EINVAL;
+
-+/* /dev/xvc0 device number allocated by lanana.org. */
-+#define XEN_XVC_MAJOR 204
-+#define XEN_XVC_MINOR 191
++ return mdio_clause45_set_settings(efx, ecmd);
++}
+
-+#ifdef CONFIG_MAGIC_SYSRQ
-+static unsigned long sysrq_requested;
-+extern int sysrq_enabled;
-+#endif
+
-+static int __init xencons_setup(char *str)
++static int falcon_xmac_set_pause(struct efx_nic *efx,
++ enum efx_fc_type flow_control)
+{
-+ char *q;
-+ int n;
-+ extern int console_use_vt;
++ int reset;
+
-+ console_use_vt = 1;
-+ if (!strncmp(str, "ttyS", 4)) {
-+ xc_mode = XC_SERIAL;
-+ str += 4;
-+ } else if (!strncmp(str, "tty", 3)) {
-+ xc_mode = XC_TTY;
-+ str += 3;
-+ console_use_vt = 0;
-+ } else if (!strncmp(str, "xvc", 3)) {
-+ xc_mode = XC_XVC;
-+ str += 3;
-+ } else if (!strncmp(str, "off", 3)) {
-+ xc_mode = XC_OFF;
-+ str += 3;
++ if (flow_control & EFX_FC_AUTO) {
++ EFX_LOG(efx, "10G does not support flow control "
++ "autonegotiation\n");
++ return -EINVAL;
+ }
+
-+ n = simple_strtol(str, &q, 10);
-+ if (q != str)
-+ xc_num = n;
++ if ((flow_control & EFX_FC_TX) && !(flow_control & EFX_FC_RX))
++ return -EINVAL;
+
-+ return 1;
++ /* TX flow control may automatically turn itself off if the
++ * link partner (intermittently) stops responding to pause
++ * frames. There isn't any indication that this has happened,
++ * so the best we do is leave it up to the user to spot this
++ * and fix it be cycling transmit flow control on this end. */
++ reset = ((flow_control & EFX_FC_TX) &&
++ !(efx->flow_control & EFX_FC_TX));
++ if (EFX_WORKAROUND_11482(efx) && reset) {
++ if (FALCON_REV(efx) == FALCON_REV_B0) {
++ /* Recover by resetting the EM block */
++ mutex_lock(&efx->mac_lock);
++ if (efx->link_up)
++ falcon_drain_tx_fifo(efx);
++ mutex_unlock(&efx->mac_lock);
++ } else {
++ /* Schedule a reset to recover */
++ efx_schedule_reset(efx, RESET_TYPE_INVISIBLE);
++ }
++ }
++
++ efx->flow_control = flow_control;
++
++ return 0;
+}
-+__setup("xencons=", xencons_setup);
+
-+/* The kernel and user-land drivers share a common transmit buffer. */
-+static unsigned int wbuf_size = 4096;
-+#define WBUF_MASK(_i) ((_i)&(wbuf_size-1))
-+static char *wbuf;
-+static unsigned int wc, wp; /* write_cons, write_prod */
++struct efx_mac_operations falcon_xmac_operations = {
++ .mac_writel = falcon_xmac_writel,
++ .mac_readl = falcon_xmac_readl,
++ .init = falcon_init_xmac,
++ .reconfigure = falcon_reconfigure_xmac,
++ .update_stats = falcon_update_stats_xmac,
++ .fini = falcon_fini_xmac,
++ .check_hw = falcon_check_xmac,
++ .fake_phy_event = falcon_xmac_sim_phy_event,
++ .get_settings = falcon_xmac_get_settings,
++ .set_settings = falcon_xmac_set_settings,
++ .set_pause = falcon_xmac_set_pause,
++};
+--- linux-2.6.18.8/drivers/net/sfc/gmii.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/gmii.h 2008-05-19 00:33:28.849809063 +0300
+@@ -0,0 +1,212 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_GMII_H
++#define EFX_GMII_H
++
++/*
++ * GMII interface
++ */
++
++#include <linux/mii.h>
++
++/* GMII registers, excluding registers already defined as MII
++ * registers in mii.h
++ */
++#define GMII_IER 0x12 /* Interrupt enable register */
++#define GMII_ISR 0x13 /* Interrupt status register */
++
++/* Interrupt enable register */
++#define IER_ANEG_ERR 0x8000 /* Bit 15 - autonegotiation error */
++#define IER_SPEED_CHG 0x4000 /* Bit 14 - speed changed */
++#define IER_DUPLEX_CHG 0x2000 /* Bit 13 - duplex changed */
++#define IER_PAGE_RCVD 0x1000 /* Bit 12 - page received */
++#define IER_ANEG_DONE 0x0800 /* Bit 11 - autonegotiation complete */
++#define IER_LINK_CHG 0x0400 /* Bit 10 - link status changed */
++#define IER_SYM_ERR 0x0200 /* Bit 9 - symbol error */
++#define IER_FALSE_CARRIER 0x0100 /* Bit 8 - false carrier */
++#define IER_FIFO_ERR 0x0080 /* Bit 7 - FIFO over/underflow */
++#define IER_MDIX_CHG 0x0040 /* Bit 6 - MDI crossover changed */
++#define IER_DOWNSHIFT 0x0020 /* Bit 5 - downshift */
++#define IER_ENERGY 0x0010 /* Bit 4 - energy detect */
++#define IER_DTE_POWER 0x0004 /* Bit 2 - DTE power detect */
++#define IER_POLARITY_CHG 0x0002 /* Bit 1 - polarity changed */
++#define IER_JABBER 0x0001 /* Bit 0 - jabber */
++
++/* Interrupt status register */
++#define ISR_ANEG_ERR 0x8000 /* Bit 15 - autonegotiation error */
++#define ISR_SPEED_CHG 0x4000 /* Bit 14 - speed changed */
++#define ISR_DUPLEX_CHG 0x2000 /* Bit 13 - duplex changed */
++#define ISR_PAGE_RCVD 0x1000 /* Bit 12 - page received */
++#define ISR_ANEG_DONE 0x0800 /* Bit 11 - autonegotiation complete */
++#define ISR_LINK_CHG 0x0400 /* Bit 10 - link status changed */
++#define ISR_SYM_ERR 0x0200 /* Bit 9 - symbol error */
++#define ISR_FALSE_CARRIER 0x0100 /* Bit 8 - false carrier */
++#define ISR_FIFO_ERR 0x0080 /* Bit 7 - FIFO over/underflow */
++#define ISR_MDIX_CHG 0x0040 /* Bit 6 - MDI crossover changed */
++#define ISR_DOWNSHIFT 0x0020 /* Bit 5 - downshift */
++#define ISR_ENERGY 0x0010 /* Bit 4 - energy detect */
++#define ISR_DTE_POWER 0x0004 /* Bit 2 - DTE power detect */
++#define ISR_POLARITY_CHG 0x0002 /* Bit 1 - polarity changed */
++#define ISR_JABBER 0x0001 /* Bit 0 - jabber */
++
++/* Logically extended advertisement register */
++#define GM_ADVERTISE_SLCT ADVERTISE_SLCT
++#define GM_ADVERTISE_CSMA ADVERTISE_CSMA
++#define GM_ADVERTISE_10HALF ADVERTISE_10HALF
++#define GM_ADVERTISE_1000XFULL ADVERTISE_1000XFULL
++#define GM_ADVERTISE_10FULL ADVERTISE_10FULL
++#define GM_ADVERTISE_1000XHALF ADVERTISE_1000XHALF
++#define GM_ADVERTISE_100HALF ADVERTISE_100HALF
++#define GM_ADVERTISE_1000XPAUSE ADVERTISE_1000XPAUSE
++#define GM_ADVERTISE_100FULL ADVERTISE_100FULL
++#define GM_ADVERTISE_1000XPSE_ASYM ADVERTISE_1000XPSE_ASYM
++#define GM_ADVERTISE_100BASE4 ADVERTISE_100BASE4
++#define GM_ADVERTISE_PAUSE_CAP ADVERTISE_PAUSE_CAP
++#define GM_ADVERTISE_PAUSE_ASYM ADVERTISE_PAUSE_ASYM
++#define GM_ADVERTISE_RESV ADVERTISE_RESV
++#define GM_ADVERTISE_RFAULT ADVERTISE_RFAULT
++#define GM_ADVERTISE_LPACK ADVERTISE_LPACK
++#define GM_ADVERTISE_NPAGE ADVERTISE_NPAGE
++#define GM_ADVERTISE_1000FULL (ADVERTISE_1000FULL << 8)
++#define GM_ADVERTISE_1000HALF (ADVERTISE_1000HALF << 8)
++#define GM_ADVERTISE_1000 (GM_ADVERTISE_1000FULL | \
++ GM_ADVERTISE_1000HALF)
++#define GM_ADVERTISE_FULL (GM_ADVERTISE_1000FULL | \
++ ADVERTISE_FULL)
++#define GM_ADVERTISE_ALL (GM_ADVERTISE_1000FULL | \
++ GM_ADVERTISE_1000HALF | \
++ ADVERTISE_ALL)
++
++/* Logically extended link partner ability register */
++#define GM_LPA_SLCT LPA_SLCT
++#define GM_LPA_10HALF LPA_10HALF
++#define GM_LPA_1000XFULL LPA_1000XFULL
++#define GM_LPA_10FULL LPA_10FULL
++#define GM_LPA_1000XHALF LPA_1000XHALF
++#define GM_LPA_100HALF LPA_100HALF
++#define GM_LPA_1000XPAUSE LPA_1000XPAUSE
++#define GM_LPA_100FULL LPA_100FULL
++#define GM_LPA_1000XPAUSE_ASYM LPA_1000XPAUSE_ASYM
++#define GM_LPA_100BASE4 LPA_100BASE4
++#define GM_LPA_PAUSE_CAP LPA_PAUSE_CAP
++#define GM_LPA_PAUSE_ASYM LPA_PAUSE_ASYM
++#define GM_LPA_RESV LPA_RESV
++#define GM_LPA_RFAULT LPA_RFAULT
++#define GM_LPA_LPACK LPA_LPACK
++#define GM_LPA_NPAGE LPA_NPAGE
++#define GM_LPA_1000FULL (LPA_1000FULL << 6)
++#define GM_LPA_1000HALF (LPA_1000HALF << 6)
++#define GM_LPA_10000FULL 0x00040000
++#define GM_LPA_10000HALF 0x00080000
++#define GM_LPA_DUPLEX (GM_LPA_1000FULL | GM_LPA_10000FULL \
++ | LPA_DUPLEX)
++#define GM_LPA_10 (LPA_10FULL | LPA_10HALF)
++#define GM_LPA_100 LPA_100
++#define GM_LPA_1000 (GM_LPA_1000FULL | GM_LPA_1000HALF)
++#define GM_LPA_10000 (GM_LPA_10000FULL | GM_LPA_10000HALF)
++
++/* Retrieve GMII autonegotiation advertised abilities
++ *
++ * The MII advertisment register (MII_ADVERTISE) is logically extended
++ * to include advertisement bits ADVERTISE_1000FULL and
++ * ADVERTISE_1000HALF from MII_CTRL1000. The result can be tested
++ * against the GM_ADVERTISE_xxx constants.
++ */
++static inline unsigned int gmii_advertised(struct mii_if_info *gmii)
++{
++ unsigned int advertise;
++ unsigned int ctrl1000;
++
++ advertise = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_ADVERTISE);
++ ctrl1000 = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_CTRL1000);
++ return (((ctrl1000 << 8) & GM_ADVERTISE_1000) | advertise);
++}
++
++/* Retrieve GMII autonegotiation link partner abilities
++ *
++ * The MII link partner ability register (MII_LPA) is logically
++ * extended by adding bits LPA_1000HALF and LPA_1000FULL from
++ * MII_STAT1000. The result can be tested against the GM_LPA_xxx
++ * constants.
++ */
++static inline unsigned int gmii_lpa(struct mii_if_info *gmii)
++{
++ unsigned int lpa;
++ unsigned int stat1000;
++
++ lpa = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_LPA);
++ stat1000 = gmii->mdio_read(gmii->dev, gmii->phy_id, MII_STAT1000);
++ return (((stat1000 << 6) & GM_LPA_1000) | lpa);
++}
++
++/* Calculate GMII autonegotiated link technology
++ *
++ * "negotiated" should be the result of gmii_advertised() logically
++ * ANDed with the result of gmii_lpa().
++ *
++ * "tech" will be negotiated with the unused bits masked out. For
++ * example, if both ends of the link are capable of both
++ * GM_LPA_1000FULL and GM_LPA_100FULL, GM_LPA_100FULL will be masked
++ * out.
++ */
++static inline unsigned int gmii_nway_result(unsigned int negotiated)
++{
++ unsigned int other_bits;
++
++ /* Mask out the speed and duplexity bits */
++ other_bits = negotiated & ~(GM_LPA_10 | GM_LPA_100 | GM_LPA_1000);
++
++ if (negotiated & GM_LPA_1000FULL)
++ return (other_bits | GM_LPA_1000FULL);
++ else if (negotiated & GM_LPA_1000HALF)
++ return (other_bits | GM_LPA_1000HALF);
++ else
++ return (other_bits | mii_nway_result(negotiated));
++}
+
-+static int __init xencons_bufsz_setup(char *str)
++/* Calculate GMII non-autonegotiated link technology
++ *
++ * This provides an equivalent to gmii_nway_result for the case when
++ * autonegotiation is disabled.
++ */
++static inline unsigned int gmii_forced_result(unsigned int bmcr)
+{
-+ unsigned int goal;
-+ goal = simple_strtoul(str, NULL, 0);
-+ if (goal) {
-+ goal = roundup_pow_of_two(goal);
-+ if (wbuf_size < goal)
-+ wbuf_size = goal;
-+ }
-+ return 1;
++ unsigned int result;
++ int full_duplex;
++
++ full_duplex = bmcr & BMCR_FULLDPLX;
++ if (bmcr & BMCR_SPEED1000)
++ result = full_duplex ? GM_LPA_1000FULL : GM_LPA_1000HALF;
++ else if (bmcr & BMCR_SPEED100)
++ result = full_duplex ? GM_LPA_100FULL : GM_LPA_100HALF;
++ else
++ result = full_duplex ? GM_LPA_10FULL : GM_LPA_10HALF;
++ return result;
+}
-+__setup("xencons_bufsz=", xencons_bufsz_setup);
+
-+/* This lock protects accesses to the common transmit buffer. */
-+static DEFINE_SPINLOCK(xencons_lock);
++#endif /* EFX_GMII_H */
+--- linux-2.6.18.8/drivers/net/sfc/i2c-direct.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/i2c-direct.c 2008-05-19 00:33:28.853809294 +0300
+@@ -0,0 +1,398 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+/* Common transmit-kick routine. */
-+static void __xencons_tx_flush(void);
++#include <asm/io.h>
++#include <linux/delay.h>
++#include "net_driver.h"
++#include "i2c-direct.h"
+
-+static struct tty_driver *xencons_driver;
++/* EEPROM access via I2C
++ * data (SDA) and clock (SCL) line read/writes
++ */
+
-+/******************** Kernel console driver ********************************/
++static inline void setsda(struct efx_i2c_interface *i2c, int state)
++{
++ udelay(i2c->op->udelay);
++ i2c->sda = state;
++ i2c->op->setsda(i2c);
++ udelay(i2c->op->udelay);
++}
+
-+static void kcons_write(struct console *c, const char *s, unsigned int count)
++static inline void setscl(struct efx_i2c_interface *i2c, int state)
+{
-+ int i = 0;
-+ unsigned long flags;
++ udelay(i2c->op->udelay);
++ i2c->scl = state;
++ i2c->op->setscl(i2c);
++ udelay(i2c->op->udelay);
++}
+
-+ spin_lock_irqsave(&xencons_lock, flags);
++static inline int getsda(struct efx_i2c_interface *i2c)
++{
++ int sda;
+
-+ while (i < count) {
-+ for (; i < count; i++) {
-+ if ((wp - wc) >= (wbuf_size - 1))
-+ break;
-+ if ((wbuf[WBUF_MASK(wp++)] = s[i]) == '\n')
-+ wbuf[WBUF_MASK(wp++)] = '\r';
-+ }
++ udelay(i2c->op->udelay);
++ sda = i2c->op->getsda(i2c);
++ udelay(i2c->op->udelay);
++ return sda;
++}
+
-+ __xencons_tx_flush();
-+ }
++static inline int getscl(struct efx_i2c_interface *i2c)
++{
++ int scl;
+
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++ udelay(i2c->op->udelay);
++ scl = i2c->op->getscl(i2c);
++ udelay(i2c->op->udelay);
++ return scl;
+}
+
-+static void kcons_write_dom0(struct console *c, const char *s, unsigned int count)
-+{
++/*
++ * I2C low-level protocol operations
++ *
++ */
+
-+ while (count > 0) {
-+ int rc;
-+ rc = HYPERVISOR_console_io( CONSOLEIO_write, count, (char *)s);
-+ if (rc <= 0)
-+ break;
-+ count -= rc;
-+ s += rc;
-+ }
++static inline void i2c_release(struct efx_i2c_interface *i2c)
++{
++ EFX_WARN_ON_PARANOID(!i2c->scl);
++ EFX_WARN_ON_PARANOID(!i2c->sda);
++ /* Just in case */
++ setscl(i2c, 1);
++ setsda(i2c, 1);
++ EFX_BUG_ON_PARANOID(getsda(i2c) != 1);
++ EFX_BUG_ON_PARANOID(getscl(i2c) != 1);
+}
+
-+static struct tty_driver *kcons_device(struct console *c, int *index)
++static inline void i2c_start(struct efx_i2c_interface *i2c)
+{
-+ *index = 0;
-+ return xencons_driver;
++ /* We may be restarting immediately after a {send,recv}_bit,
++ * so SCL will not necessarily already be high.
++ */
++ EFX_WARN_ON_PARANOID(!i2c->sda);
++ setscl(i2c, 1);
++ setsda(i2c, 0);
++ setscl(i2c, 0);
++ setsda(i2c, 1);
+}
+
-+static struct console kcons_info = {
-+ .device = kcons_device,
-+ .flags = CON_PRINTBUFFER | CON_ENABLED,
-+ .index = -1,
-+};
++static inline void i2c_send_bit(struct efx_i2c_interface *i2c, int bit)
++{
++ EFX_WARN_ON_PARANOID(i2c->scl != 0);
++ setsda(i2c, bit);
++ setscl(i2c, 1);
++ setscl(i2c, 0);
++ setsda(i2c, 1);
++}
+
-+static int __init xen_console_init(void)
++static inline int i2c_recv_bit(struct efx_i2c_interface *i2c)
+{
-+ if (!is_running_on_xen())
-+ goto out;
++ int bit;
+
-+ if (is_initial_xendomain()) {
-+ kcons_info.write = kcons_write_dom0;
-+ } else {
-+ if (!xen_start_info->console.domU.evtchn)
-+ goto out;
-+ kcons_info.write = kcons_write;
-+ }
++ EFX_WARN_ON_PARANOID(i2c->scl != 0);
++ EFX_WARN_ON_PARANOID(!i2c->sda);
++ setscl(i2c, 1);
++ bit = getsda(i2c);
++ setscl(i2c, 0);
++ return bit;
++}
+
-+ switch (xc_mode) {
-+ case XC_XVC:
-+ strcpy(kcons_info.name, "xvc");
-+ if (xc_num == -1)
-+ xc_num = 0;
-+ break;
++static inline void i2c_stop(struct efx_i2c_interface *i2c)
++{
++ EFX_WARN_ON_PARANOID(i2c->scl != 0);
++ setsda(i2c, 0);
++ setscl(i2c, 1);
++ setsda(i2c, 1);
++}
+
-+ case XC_SERIAL:
-+ strcpy(kcons_info.name, "ttyS");
-+ if (xc_num == -1)
-+ xc_num = 0;
-+ break;
++/*
++ * I2C mid-level protocol operations
++ *
++ */
+
-+ case XC_TTY:
-+ strcpy(kcons_info.name, "tty");
-+ if (xc_num == -1)
-+ xc_num = 1;
-+ break;
++/* Sends a byte via the I2C bus and checks for an acknowledgement from
++ * the slave device.
++ */
++static int i2c_send_byte(struct efx_i2c_interface *i2c, u8 byte)
++{
++ int i;
+
-+ default:
-+ goto out;
++ /* Send byte */
++ for (i = 0; i < 8; i++) {
++ i2c_send_bit(i2c, !!(byte & 0x80));
++ byte <<= 1;
+ }
+
-+ wbuf = alloc_bootmem(wbuf_size);
-+
-+ register_console(&kcons_info);
-+
-+ out:
-+ return 0;
++ /* Check for acknowledgement from slave */
++ return (i2c_recv_bit(i2c) == 0 ? 0 : -EIO);
+}
-+console_initcall(xen_console_init);
+
-+/*** Useful function for console debugging -- goes straight to Xen. ***/
-+asmlinkage int xprintk(const char *fmt, ...)
++/* Receives a byte via the I2C bus and sends ACK/NACK to the slave device. */
++static u8 i2c_recv_byte(struct efx_i2c_interface *i2c, int ack)
+{
-+ va_list args;
-+ int printk_len;
-+ static char printk_buf[1024];
++ u8 value = 0;
++ int i;
+
-+ /* Emit the output into the temporary buffer */
-+ va_start(args, fmt);
-+ printk_len = vsnprintf(printk_buf, sizeof(printk_buf), fmt, args);
-+ va_end(args);
++ /* Receive byte */
++ for (i = 0; i < 8; i++)
++ value = (value << 1) | i2c_recv_bit(i2c);
+
-+ /* Send the processed output directly to Xen. */
-+ kcons_write_dom0(NULL, printk_buf, printk_len);
++ /* Send ACK/NACK */
++ i2c_send_bit(i2c, (ack ? 0 : 1));
+
-+ return 0;
++ return value;
+}
+
-+/*** Forcibly flush console data before dying. ***/
-+void xencons_force_flush(void)
++/* Calculate command byte for a read operation */
++static inline u8 i2c_read_cmd(u8 device_id)
+{
-+ int sz;
-+
-+ /* Emergency console is synchronous, so there's nothing to flush. */
-+ if (!is_running_on_xen() ||
-+ is_initial_xendomain() ||
-+ !xen_start_info->console.domU.evtchn)
-+ return;
-+
-+ /* Spin until console data is flushed through to the daemon. */
-+ while (wc != wp) {
-+ int sent = 0;
-+ if ((sz = wp - wc) == 0)
-+ continue;
-+ sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
-+ if (sent > 0)
-+ wc += sent;
-+ }
++ return ((device_id << 1) | 1);
+}
+
-+
-+void __init dom0_init_screen_info(const struct dom0_vga_console_info *info, size_t size)
++/* Calculate command byte for a write operation */
++static inline u8 i2c_write_cmd(u8 device_id)
+{
-+ /* This is drawn from a dump from vgacon:startup in
-+ * standard Linux. */
-+ screen_info.orig_video_mode = 3;
-+ screen_info.orig_video_isVGA = 1;
-+ screen_info.orig_video_lines = 25;
-+ screen_info.orig_video_cols = 80;
-+ screen_info.orig_video_ega_bx = 3;
-+ screen_info.orig_video_points = 16;
-+ screen_info.orig_y = screen_info.orig_video_lines - 1;
++ return ((device_id << 1) | 0);
++}
+
-+ switch (info->video_type) {
-+ case XEN_VGATYPE_TEXT_MODE_3:
-+ if (size < offsetof(struct dom0_vga_console_info, u.text_mode_3)
-+ + sizeof(info->u.text_mode_3))
-+ break;
-+ screen_info.orig_video_lines = info->u.text_mode_3.rows;
-+ screen_info.orig_video_cols = info->u.text_mode_3.columns;
-+ screen_info.orig_x = info->u.text_mode_3.cursor_x;
-+ screen_info.orig_y = info->u.text_mode_3.cursor_y;
-+ screen_info.orig_video_points =
-+ info->u.text_mode_3.font_height;
-+ break;
++int efx_i2c_check_presence(struct efx_i2c_interface *i2c, u8 device_id)
++{
++ int rc;
+
-+ case XEN_VGATYPE_VESA_LFB:
-+ if (size < offsetof(struct dom0_vga_console_info,
-+ u.vesa_lfb.gbl_caps))
-+ break;
-+ screen_info.orig_video_isVGA = VIDEO_TYPE_VLFB;
-+ screen_info.lfb_width = info->u.vesa_lfb.width;
-+ screen_info.lfb_height = info->u.vesa_lfb.height;
-+ screen_info.lfb_depth = info->u.vesa_lfb.bits_per_pixel;
-+ screen_info.lfb_base = info->u.vesa_lfb.lfb_base;
-+ screen_info.lfb_size = info->u.vesa_lfb.lfb_size;
-+ screen_info.lfb_linelength = info->u.vesa_lfb.bytes_per_line;
-+ screen_info.red_size = info->u.vesa_lfb.red_size;
-+ screen_info.red_pos = info->u.vesa_lfb.red_pos;
-+ screen_info.green_size = info->u.vesa_lfb.green_size;
-+ screen_info.green_pos = info->u.vesa_lfb.green_pos;
-+ screen_info.blue_size = info->u.vesa_lfb.blue_size;
-+ screen_info.blue_pos = info->u.vesa_lfb.blue_pos;
-+ screen_info.rsvd_size = info->u.vesa_lfb.rsvd_size;
-+ screen_info.rsvd_pos = info->u.vesa_lfb.rsvd_pos;
-+ if (size >= offsetof(struct dom0_vga_console_info,
-+ u.vesa_lfb.gbl_caps)
-+ + sizeof(info->u.vesa_lfb.gbl_caps))
-+ screen_info.capabilities = info->u.vesa_lfb.gbl_caps;
-+ if (size >= offsetof(struct dom0_vga_console_info,
-+ u.vesa_lfb.mode_attrs)
-+ + sizeof(info->u.vesa_lfb.mode_attrs))
-+ screen_info.vesa_attributes = info->u.vesa_lfb.mode_attrs;
-+ break;
++ /* If someone is driving the bus low we just give up. */
++ if (getsda(i2c) == 0 || getscl(i2c) == 0) {
++ EFX_ERR(i2c->efx, "%s someone is holding the I2C bus low."
++ " Giving up.\n", __func__);
++ return -EFAULT;
+ }
++
++ /* Pretend to initiate a device write */
++ i2c_start(i2c);
++ rc = i2c_send_byte(i2c, i2c_write_cmd(device_id));
++ if (rc)
++ goto out;
++
++ out:
++ i2c_stop(i2c);
++ i2c_release(i2c);
++
++ return rc;
+}
+
++/* This performs a fast read of one or more consecutive bytes from an
++ * I2C device. Not all devices support consecutive reads of more than
++ * one byte; for these devices use efx_i2c_read() instead.
++ */
++int efx_i2c_fast_read(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, u8 *data, unsigned int len)
++{
++ int i;
++ int rc;
+
-+/******************** User-space console driver (/dev/console) ************/
++ EFX_WARN_ON_PARANOID(getsda(i2c) != 1);
++ EFX_WARN_ON_PARANOID(getscl(i2c) != 1);
++ EFX_WARN_ON_PARANOID(data == NULL);
++ EFX_WARN_ON_PARANOID(len < 1);
+
-+#define DRV(_d) (_d)
-+#define DUMMY_TTY(_tty) ((xc_mode == XC_TTY) && \
-+ ((_tty)->index != (xc_num - 1)))
++ /* Select device and starting offset */
++ i2c_start(i2c);
++ rc = i2c_send_byte(i2c, i2c_write_cmd(device_id));
++ if (rc)
++ goto out;
++ rc = i2c_send_byte(i2c, offset);
++ if (rc)
++ goto out;
+
-+static struct termios *xencons_termios[MAX_NR_CONSOLES];
-+static struct termios *xencons_termios_locked[MAX_NR_CONSOLES];
-+static struct tty_struct *xencons_tty;
-+static int xencons_priv_irq;
-+static char x_char;
++ /* Read data from device */
++ i2c_start(i2c);
++ rc = i2c_send_byte(i2c, i2c_read_cmd(device_id));
++ if (rc)
++ goto out;
++ for (i = 0; i < (len - 1); i++)
++ /* Read and acknowledge all but the last byte */
++ data[i] = i2c_recv_byte(i2c, 1);
++ /* Read last byte with no acknowledgement */
++ data[i] = i2c_recv_byte(i2c, 0);
+
-+void xencons_rx(char *buf, unsigned len, struct pt_regs *regs)
++ out:
++ i2c_stop(i2c);
++ i2c_release(i2c);
++
++ return rc;
++}
++
++/* This performs a fast write of one or more consecutive bytes to an
++ * I2C device. Not all devices support consecutive writes of more
++ * than one byte; for these devices use efx_i2c_write() instead.
++ */
++int efx_i2c_fast_write(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset,
++ const u8 *data, unsigned int len)
+{
-+ int i;
-+ unsigned long flags;
++ int i;
++ int rc;
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ if (xencons_tty == NULL)
++ EFX_WARN_ON_PARANOID(getsda(i2c) != 1);
++ EFX_WARN_ON_PARANOID(getscl(i2c) != 1);
++ EFX_WARN_ON_PARANOID(len < 1);
++
++ /* Select device and starting offset */
++ i2c_start(i2c);
++ rc = i2c_send_byte(i2c, i2c_write_cmd(device_id));
++ if (rc)
++ goto out;
++ rc = i2c_send_byte(i2c, offset);
++ if (rc)
+ goto out;
+
++ /* Write data to device */
+ for (i = 0; i < len; i++) {
-+#ifdef CONFIG_MAGIC_SYSRQ
-+ if (sysrq_enabled) {
-+ if (buf[i] == '\x0f') { /* ^O */
-+ if (!sysrq_requested) {
-+ sysrq_requested = jiffies;
-+ continue; /* don't print sysrq key */
-+ }
-+ sysrq_requested = 0;
-+ } else if (sysrq_requested) {
-+ unsigned long sysrq_timeout =
-+ sysrq_requested + HZ*2;
-+ sysrq_requested = 0;
-+ if (time_before(jiffies, sysrq_timeout)) {
-+ spin_unlock_irqrestore(
-+ &xencons_lock, flags);
-+ handle_sysrq(
-+ buf[i], regs, xencons_tty);
-+ spin_lock_irqsave(
-+ &xencons_lock, flags);
-+ continue;
-+ }
-+ }
-+ }
-+#endif
-+ tty_insert_flip_char(xencons_tty, buf[i], 0);
++ rc = i2c_send_byte(i2c, data[i]);
++ if (rc)
++ goto out;
+ }
-+ tty_flip_buffer_push(xencons_tty);
+
+ out:
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++ i2c_stop(i2c);
++ i2c_release(i2c);
++
++ return rc;
+}
+
-+static void __xencons_tx_flush(void)
++/* I2C byte-by-byte read */
++int efx_i2c_read(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, u8 *data, unsigned int len)
+{
-+ int sent, sz, work_done = 0;
++ int rc;
+
-+ if (x_char) {
-+ if (is_initial_xendomain())
-+ kcons_write_dom0(NULL, &x_char, 1);
-+ else
-+ while (x_char)
-+ if (xencons_ring_send(&x_char, 1) == 1)
-+ break;
-+ x_char = 0;
-+ work_done = 1;
++ /* i2c_fast_read with length 1 is a single byte read */
++ for (; len > 0; offset++, data++, len--) {
++ rc = efx_i2c_fast_read(i2c, device_id, offset, data, 1);
++ if (rc)
++ return rc;
+ }
+
-+ while (wc != wp) {
-+ sz = wp - wc;
-+ if (sz > (wbuf_size - WBUF_MASK(wc)))
-+ sz = wbuf_size - WBUF_MASK(wc);
-+ if (is_initial_xendomain()) {
-+ kcons_write_dom0(NULL, &wbuf[WBUF_MASK(wc)], sz);
-+ wc += sz;
-+ } else {
-+ sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
-+ if (sent == 0)
-+ break;
-+ wc += sent;
-+ }
-+ work_done = 1;
-+ }
++ return 0;
++}
+
-+ if (work_done && (xencons_tty != NULL)) {
-+ wake_up_interruptible(&xencons_tty->write_wait);
-+ if ((xencons_tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
-+ (xencons_tty->ldisc.write_wakeup != NULL))
-+ (xencons_tty->ldisc.write_wakeup)(xencons_tty);
++/* I2C byte-by-byte write */
++int efx_i2c_write(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, const u8 *data, unsigned int len)
++{
++ int rc;
++
++ /* i2c_fast_write with length 1 is a single byte write */
++ for (; len > 0; offset++, data++, len--) {
++ rc = efx_i2c_fast_write(i2c, device_id, offset, data, 1);
++ if (rc)
++ return rc;
++ mdelay(i2c->op->mdelay);
+ }
++
++ return 0;
+}
+
-+void xencons_tx(void)
-+{
-+ unsigned long flags;
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ __xencons_tx_flush();
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++/* This is just a slightly neater wrapper round efx_i2c_fast_write
++ * in the case where the target doesn't take an offset
++ */
++int efx_i2c_send_bytes(struct efx_i2c_interface *i2c,
++ u8 device_id, const u8 *data, unsigned int len)
++{
++ return efx_i2c_fast_write(i2c, device_id, data[0], data + 1, len - 1);
+}
+
-+/* Privileged receive callback and transmit kicker. */
-+static irqreturn_t xencons_priv_interrupt(int irq, void *dev_id,
-+ struct pt_regs *regs)
++/* I2C receiving of bytes - does not send an offset byte */
++int efx_i2c_recv_bytes(struct efx_i2c_interface *i2c, u8 device_id,
++ u8 *bytes, unsigned int len)
+{
-+ static char rbuf[16];
-+ int l;
++ int i;
++ int rc;
+
-+ while ((l = HYPERVISOR_console_io(CONSOLEIO_read, 16, rbuf)) > 0)
-+ xencons_rx(rbuf, l, regs);
++ EFX_WARN_ON_PARANOID(getsda(i2c) != 1);
++ EFX_WARN_ON_PARANOID(getscl(i2c) != 1);
++ EFX_WARN_ON_PARANOID(len < 1);
+
-+ xencons_tx();
++ /* Select device */
++ i2c_start(i2c);
+
-+ return IRQ_HANDLED;
++ /* Read data from device */
++ rc = i2c_send_byte(i2c, i2c_read_cmd(device_id));
++ if (rc)
++ goto out;
++
++ for (i = 0; i < (len - 1); i++)
++ /* Read and acknowledge all but the last byte */
++ bytes[i] = i2c_recv_byte(i2c, 1);
++ /* Read last byte with no acknowledgement */
++ bytes[i] = i2c_recv_byte(i2c, 0);
++
++ out:
++ i2c_stop(i2c);
++ i2c_release(i2c);
++
++ return rc;
+}
+
-+static int xencons_write_room(struct tty_struct *tty)
++/* SMBus and some I2C devices will time out if the I2C clock is
++ * held low for too long. This is most likely to happen in virtualised
++ * systems (when the entire domain is descheduled) but could in
++ * principle happen due to preemption on any busy system (and given the
++ * potential length of an I2C operation turning preemption off is not
++ * a sensible option). The following functions deal with the failure by
++ * retrying up to a fixed number of times.
++ */
++
++#define I2C_MAX_RETRIES (10)
++
++/* The timeout problem will result in -EIO. If the wrapped function
++ * returns any other error, pass this up and do not retry. */
++#define RETRY_WRAPPER(_f) \
++ int retries = I2C_MAX_RETRIES; \
++ int rc; \
++ while (retries) { \
++ rc = _f; \
++ if (rc != -EIO) \
++ return rc; \
++ retries--; \
++ } \
++ return rc; \
++
++int efx_i2c_check_presence_retry(struct efx_i2c_interface *i2c, u8 device_id)
+{
-+ return wbuf_size - (wp - wc);
++ RETRY_WRAPPER(efx_i2c_check_presence(i2c, device_id))
+}
+
-+static int xencons_chars_in_buffer(struct tty_struct *tty)
++int efx_i2c_read_retry(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, u8 *data, unsigned int len)
+{
-+ return wp - wc;
++ RETRY_WRAPPER(efx_i2c_read(i2c, device_id, offset, data, len))
+}
+
-+static void xencons_send_xchar(struct tty_struct *tty, char ch)
++int efx_i2c_write_retry(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, const u8 *data, unsigned int len)
+{
-+ unsigned long flags;
++ RETRY_WRAPPER(efx_i2c_write(i2c, device_id, offset, data, len))
++}
+--- linux-2.6.18.8/drivers/net/sfc/i2c-direct.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/i2c-direct.h 2008-05-19 00:33:28.853809294 +0300
+@@ -0,0 +1,108 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (DUMMY_TTY(tty))
-+ return;
++#ifndef EFX_I2C_DIRECT_H
++#define EFX_I2C_DIRECT_H
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ x_char = ch;
-+ __xencons_tx_flush();
-+ spin_unlock_irqrestore(&xencons_lock, flags);
-+}
++#include "net_driver.h"
+
-+static void xencons_throttle(struct tty_struct *tty)
-+{
-+ if (DUMMY_TTY(tty))
-+ return;
++/*
++ * Direct control of an I2C bus
++ */
+
-+ if (I_IXOFF(tty))
-+ xencons_send_xchar(tty, STOP_CHAR(tty));
-+}
++struct efx_i2c_interface;
+
-+static void xencons_unthrottle(struct tty_struct *tty)
-+{
-+ if (DUMMY_TTY(tty))
-+ return;
++/**
++ * struct efx_i2c_bit_operations - I2C bus direct control methods
++ *
++ * I2C bus direct control methods.
++ *
++ * @setsda: Set state of SDA line
++ * @setscl: Set state of SCL line
++ * @getsda: Get state of SDA line
++ * @getscl: Get state of SCL line
++ * @udelay: Delay between each bit operation
++ * @mdelay: Delay between each byte write
++ */
++struct efx_i2c_bit_operations {
++ void (*setsda) (struct efx_i2c_interface *i2c);
++ void (*setscl) (struct efx_i2c_interface *i2c);
++ int (*getsda) (struct efx_i2c_interface *i2c);
++ int (*getscl) (struct efx_i2c_interface *i2c);
++ unsigned int udelay;
++ unsigned int mdelay;
++};
+
-+ if (I_IXOFF(tty)) {
-+ if (x_char != 0)
-+ x_char = 0;
-+ else
-+ xencons_send_xchar(tty, START_CHAR(tty));
-+ }
-+}
++/**
++ * struct efx_i2c_interface - an I2C interface
++ *
++ * An I2C interface.
++ *
++ * @efx: Attached Efx NIC
++ * @op: I2C bus control methods
++ * @sda: Current output state of SDA line
++ * @scl: Current output state of SCL line
++ */
++struct efx_i2c_interface {
++ struct efx_nic *efx;
++ struct efx_i2c_bit_operations *op;
++ unsigned int sda:1;
++ unsigned int scl:1;
++};
+
-+static void xencons_flush_buffer(struct tty_struct *tty)
-+{
-+ unsigned long flags;
++extern int efx_i2c_check_presence(struct efx_i2c_interface *i2c, u8 device_id);
++extern int efx_i2c_fast_read(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset,
++ u8 *data, unsigned int len);
++extern int efx_i2c_fast_write(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset,
++ const u8 *data, unsigned int len);
++extern int efx_i2c_read(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, u8 *data, unsigned int len);
++extern int efx_i2c_write(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset,
++ const u8 *data, unsigned int len);
+
-+ if (DUMMY_TTY(tty))
-+ return;
++extern int efx_i2c_send_bytes(struct efx_i2c_interface *i2c, u8 device_id,
++ const u8 *bytes, unsigned int len);
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ wc = wp = 0;
-+ spin_unlock_irqrestore(&xencons_lock, flags);
-+}
++extern int efx_i2c_recv_bytes(struct efx_i2c_interface *i2c, u8 device_id,
++ u8 *bytes, unsigned int len);
+
-+static inline int __xencons_put_char(int ch)
-+{
-+ char _ch = (char)ch;
-+ if ((wp - wc) == wbuf_size)
-+ return 0;
-+ wbuf[WBUF_MASK(wp++)] = _ch;
-+ return 1;
-+}
+
-+static int xencons_write(
-+ struct tty_struct *tty,
-+ const unsigned char *buf,
-+ int count)
-+{
-+ int i;
-+ unsigned long flags;
++/* Versions of the API that retry on failure. */
++extern int efx_i2c_check_presence_retry(struct efx_i2c_interface *i2c,
++ u8 device_id);
+
-+ if (DUMMY_TTY(tty))
-+ return count;
++extern int efx_i2c_read_retry(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset, u8 *data, unsigned int len);
+
-+ spin_lock_irqsave(&xencons_lock, flags);
++extern int efx_i2c_write_retry(struct efx_i2c_interface *i2c,
++ u8 device_id, u8 offset,
++ const u8 *data, unsigned int len);
+
-+ for (i = 0; i < count; i++)
-+ if (!__xencons_put_char(buf[i]))
-+ break;
++#endif /* EFX_I2C_DIRECT_H */
+--- linux-2.6.18.8/drivers/net/sfc/kernel_compat.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/kernel_compat.c 2008-05-19 00:33:28.865809986 +0300
+@@ -0,0 +1,654 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (i != 0)
-+ __xencons_tx_flush();
++#define EFX_IN_KCOMPAT_C 1
+
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++#include "net_driver.h"
++#include <linux/mii.h>
++#include <linux/ethtool.h>
++#include <linux/random.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include <linux/rtnetlink.h>
++#include <linux/bootmem.h>
++#include <asm/uaccess.h>
++#include "gmii.h"
++#include "ethtool.h"
+
-+ return i;
-+}
++/*
++ * Kernel backwards compatibility
++ *
++ * This file provides functionality missing from earlier kernels.
++ */
+
-+static void xencons_put_char(struct tty_struct *tty, u_char ch)
++/**************************************************************************
++ *
++ * GMII-friendly versions of mii_ethtool_[gs]set
++ *
++ **************************************************************************
++ *
++ * Kernels prior to 2.6.12 don't support GMII PHYs via
++ * mii_ethtool_gset and mii_ethtool_sset. These are those functions
++ * taken from a 2.6.12 kernel tree, with the tests for
++ * mii->supports_gmii removed (since that field doesn't exist in older
++ * kernels).
++ *
++ */
++
++#ifdef EFX_NEED_MII_ETHTOOL_FIX
++int efx_mii_ethtool_gset(struct mii_if_info *mii, struct ethtool_cmd *ecmd)
+{
-+ unsigned long flags;
++ struct net_device *dev = mii->dev;
++ u32 advert, bmcr, lpa, nego;
++ u32 advert2 = 0, bmcr2 = 0, lpa2 = 0;
+
-+ if (DUMMY_TTY(tty))
-+ return;
++ ecmd->supported =
++ (SUPPORTED_10baseT_Half | SUPPORTED_10baseT_Full |
++ SUPPORTED_100baseT_Half | SUPPORTED_100baseT_Full |
++ SUPPORTED_Autoneg | SUPPORTED_TP | SUPPORTED_MII);
++ ecmd->supported |= SUPPORTED_1000baseT_Half | SUPPORTED_1000baseT_Full;
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ (void)__xencons_put_char(ch);
-+ spin_unlock_irqrestore(&xencons_lock, flags);
-+}
++ /* only supports twisted-pair */
++ ecmd->port = PORT_MII;
+
-+static void xencons_flush_chars(struct tty_struct *tty)
-+{
-+ unsigned long flags;
++ /* only supports internal transceiver */
++ ecmd->transceiver = XCVR_INTERNAL;
+
-+ if (DUMMY_TTY(tty))
-+ return;
++ /* this isn't fully supported at higher layers */
++ ecmd->phy_address = mii->phy_id;
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ __xencons_tx_flush();
-+ spin_unlock_irqrestore(&xencons_lock, flags);
-+}
++ ecmd->advertising = ADVERTISED_TP | ADVERTISED_MII;
++ advert = mii->mdio_read(dev, mii->phy_id, MII_ADVERTISE);
++ advert2 = mii->mdio_read(dev, mii->phy_id, MII_CTRL1000);
+
-+static void xencons_wait_until_sent(struct tty_struct *tty, int timeout)
-+{
-+ unsigned long orig_jiffies = jiffies;
++ if (advert & ADVERTISE_10HALF)
++ ecmd->advertising |= ADVERTISED_10baseT_Half;
++ if (advert & ADVERTISE_10FULL)
++ ecmd->advertising |= ADVERTISED_10baseT_Full;
++ if (advert & ADVERTISE_100HALF)
++ ecmd->advertising |= ADVERTISED_100baseT_Half;
++ if (advert & ADVERTISE_100FULL)
++ ecmd->advertising |= ADVERTISED_100baseT_Full;
++ if (advert2 & ADVERTISE_1000HALF)
++ ecmd->advertising |= ADVERTISED_1000baseT_Half;
++ if (advert2 & ADVERTISE_1000FULL)
++ ecmd->advertising |= ADVERTISED_1000baseT_Full;
+
-+ if (DUMMY_TTY(tty))
-+ return;
++ bmcr = mii->mdio_read(dev, mii->phy_id, MII_BMCR);
++ lpa = mii->mdio_read(dev, mii->phy_id, MII_LPA);
++ bmcr2 = mii->mdio_read(dev, mii->phy_id, MII_CTRL1000);
++ lpa2 = mii->mdio_read(dev, mii->phy_id, MII_STAT1000);
++ if (bmcr & BMCR_ANENABLE) {
++ ecmd->advertising |= ADVERTISED_Autoneg;
++ ecmd->autoneg = AUTONEG_ENABLE;
+
-+ while (DRV(tty->driver)->chars_in_buffer(tty)) {
-+ set_current_state(TASK_INTERRUPTIBLE);
-+ schedule_timeout(1);
-+ if (signal_pending(current))
-+ break;
-+ if (timeout && time_after(jiffies, orig_jiffies + timeout))
-+ break;
++ nego = mii_nway_result(advert & lpa);
++ if ((bmcr2 & (ADVERTISE_1000HALF | ADVERTISE_1000FULL)) &
++ (lpa2 >> 2))
++ ecmd->speed = SPEED_1000;
++ else if (nego == LPA_100FULL || nego == LPA_100HALF)
++ ecmd->speed = SPEED_100;
++ else
++ ecmd->speed = SPEED_10;
++ if ((lpa2 & LPA_1000FULL) || nego == LPA_100FULL ||
++ nego == LPA_10FULL) {
++ ecmd->duplex = DUPLEX_FULL;
++ mii->full_duplex = 1;
++ } else {
++ ecmd->duplex = DUPLEX_HALF;
++ mii->full_duplex = 0;
++ }
++ } else {
++ ecmd->autoneg = AUTONEG_DISABLE;
++
++ ecmd->speed = ((bmcr & BMCR_SPEED1000 &&
++ (bmcr & BMCR_SPEED100) == 0) ? SPEED_1000 :
++ (bmcr & BMCR_SPEED100) ? SPEED_100 : SPEED_10);
++ ecmd->duplex =
++ (bmcr & BMCR_FULLDPLX) ? DUPLEX_FULL : DUPLEX_HALF;
+ }
+
-+ set_current_state(TASK_RUNNING);
++ /* ignore maxtxpkt, maxrxpkt for now */
++
++ return 0;
+}
+
-+static int xencons_open(struct tty_struct *tty, struct file *filp)
++int efx_mii_ethtool_sset(struct mii_if_info *mii, struct ethtool_cmd *ecmd)
+{
-+ unsigned long flags;
++ struct net_device *dev = mii->dev;
+
-+ if (DUMMY_TTY(tty))
-+ return 0;
++ if (ecmd->speed != SPEED_10 &&
++ ecmd->speed != SPEED_100 &&
++ ecmd->speed != SPEED_1000)
++ return -EINVAL;
++ if (ecmd->duplex != DUPLEX_HALF && ecmd->duplex != DUPLEX_FULL)
++ return -EINVAL;
++ if (ecmd->port != PORT_MII)
++ return -EINVAL;
++ if (ecmd->transceiver != XCVR_INTERNAL)
++ return -EINVAL;
++ if (ecmd->phy_address != mii->phy_id)
++ return -EINVAL;
++ if (ecmd->autoneg != AUTONEG_DISABLE && ecmd->autoneg != AUTONEG_ENABLE)
++ return -EINVAL;
+
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ tty->driver_data = NULL;
-+ if (xencons_tty == NULL)
-+ xencons_tty = tty;
-+ __xencons_tx_flush();
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++ /* ignore supported, maxtxpkt, maxrxpkt */
++
++ if (ecmd->autoneg == AUTONEG_ENABLE) {
++ u32 bmcr, advert, tmp;
++ u32 advert2 = 0, tmp2 = 0;
++
++ if ((ecmd->advertising & (ADVERTISED_10baseT_Half |
++ ADVERTISED_10baseT_Full |
++ ADVERTISED_100baseT_Half |
++ ADVERTISED_100baseT_Full |
++ ADVERTISED_1000baseT_Half |
++ ADVERTISED_1000baseT_Full)) == 0)
++ return -EINVAL;
++
++ /* advertise only what has been requested */
++ advert = mii->mdio_read(dev, mii->phy_id, MII_ADVERTISE);
++ tmp = advert & ~(ADVERTISE_ALL | ADVERTISE_100BASE4);
++ advert2 = mii->mdio_read(dev, mii->phy_id, MII_CTRL1000);
++ tmp2 = advert2 & ~(ADVERTISE_1000HALF | ADVERTISE_1000FULL);
++ if (ecmd->advertising & ADVERTISED_10baseT_Half)
++ tmp |= ADVERTISE_10HALF;
++ if (ecmd->advertising & ADVERTISED_10baseT_Full)
++ tmp |= ADVERTISE_10FULL;
++ if (ecmd->advertising & ADVERTISED_100baseT_Half)
++ tmp |= ADVERTISE_100HALF;
++ if (ecmd->advertising & ADVERTISED_100baseT_Full)
++ tmp |= ADVERTISE_100FULL;
++ if (ecmd->advertising & ADVERTISED_1000baseT_Half)
++ tmp2 |= ADVERTISE_1000HALF;
++ if (ecmd->advertising & ADVERTISED_1000baseT_Full)
++ tmp2 |= ADVERTISE_1000FULL;
++ if (advert != tmp) {
++ mii->mdio_write(dev, mii->phy_id, MII_ADVERTISE, tmp);
++ mii->advertising = tmp;
++ }
++ if (advert2 != tmp2)
++ mii->mdio_write(dev, mii->phy_id, MII_CTRL1000, tmp2);
++
++ /* turn on autonegotiation, and force a renegotiate */
++ bmcr = mii->mdio_read(dev, mii->phy_id, MII_BMCR);
++ bmcr |= (BMCR_ANENABLE | BMCR_ANRESTART);
++ mii->mdio_write(dev, mii->phy_id, MII_BMCR, bmcr);
++
++ mii->force_media = 0;
++ } else {
++ u32 bmcr, tmp;
++
++ /* turn off auto negotiation, set speed and duplexity */
++ bmcr = mii->mdio_read(dev, mii->phy_id, MII_BMCR);
++ tmp = bmcr & ~(BMCR_ANENABLE | BMCR_SPEED100 |
++ BMCR_SPEED1000 | BMCR_FULLDPLX);
++ if (ecmd->speed == SPEED_1000)
++ tmp |= BMCR_SPEED1000;
++ else if (ecmd->speed == SPEED_100)
++ tmp |= BMCR_SPEED100;
++ if (ecmd->duplex == DUPLEX_FULL) {
++ tmp |= BMCR_FULLDPLX;
++ mii->full_duplex = 1;
++ } else {
++ mii->full_duplex = 0;
++ }
++ if (bmcr != tmp)
++ mii->mdio_write(dev, mii->phy_id, MII_BMCR, tmp);
+
++ mii->force_media = 1;
++ }
+ return 0;
+}
++#endif /* NEED_EFX_MII_ETHTOOL_GSET */
+
-+static void xencons_close(struct tty_struct *tty, struct file *filp)
++/**************************************************************************
++ *
++ * unregister_netdevice_notifier : Has a race before 2.6.17
++ *
++ **************************************************************************
++ *
++ */
++
++#ifdef EFX_NEED_UNREGISTER_NETDEVICE_NOTIFIER_FIX
++/**
++ * efx_unregister_netdevice_notifier - fixed unregister_netdevice_notifier
++ * @nb: notifier to unregister
++ *
++ * unregister_netdevice_notifier() does not wait for the notifier
++ * to be unused before 2.6.17. This wrapper fixes that.
++ */
++int efx_unregister_netdevice_notifier(struct notifier_block *nb)
+{
-+ unsigned long flags;
++ int res;
+
-+ if (DUMMY_TTY(tty))
-+ return;
++ res = unregister_netdevice_notifier(nb);
++ /* Ensure any outstanding calls complete. */
++ rtnl_lock();
++ rtnl_unlock();
++ return res;
++}
++#endif /* NEED_EFX_UNREGISTER_NETDEVICE_NOTIFIER */
+
-+ mutex_lock(&tty_mutex);
++/**************************************************************************
++ *
++ * IOMMU-locking versions of pci_[un]map_single and
++ * pci_{alloc,free}_consistent. See SFC bug 4560.
++ *
++ **************************************************************************
++ *
++ */
++#ifdef EFX_NEED_IOMMU_LOCK
+
-+ if (tty->count != 1) {
-+ mutex_unlock(&tty_mutex);
-+ return;
-+ }
++/*
++ * efx_use_iommu_lock - IOMMU lock use control
++ *
++ * If set to 1, the driver will attempt to mitigate the race condition
++ * bug around IOMMU accesses in some 2.6 kernels. If set to 2, the
++ * driver will use the lock even if it thinks it doesn't need to.
++ * Note that this is only a best-effort attempt; in particular, we
++ * cannot do anything about other drivers touching the IOMMU.
++ */
++static unsigned int efx_use_iommu_lock = 1;
++EXPORT_SYMBOL(efx_use_iommu_lock);
+
-+ /* Prevent other threads from re-opening this tty. */
-+ set_bit(TTY_CLOSING, &tty->flags);
-+ mutex_unlock(&tty_mutex);
++/*
++ * efx_iommu_lock - lock around IOMMU accesses
++ *
++ * This spinlock should be held while calling functions that access
++ * the IOMMU if efx_use_iommu_lock is >= 2. The efx_pci_*()
++ * functions do this where possible.
++ */
++static spinlock_t efx_iommu_lock = SPIN_LOCK_UNLOCKED;
++EXPORT_SYMBOL(efx_iommu_lock);
+
-+ tty->closing = 1;
-+ tty_wait_until_sent(tty, 0);
-+ if (DRV(tty->driver)->flush_buffer != NULL)
-+ DRV(tty->driver)->flush_buffer(tty);
-+ if (tty->ldisc.flush_buffer != NULL)
-+ tty->ldisc.flush_buffer(tty);
-+ tty->closing = 0;
-+ spin_lock_irqsave(&xencons_lock, flags);
-+ xencons_tty = NULL;
-+ spin_unlock_irqrestore(&xencons_lock, flags);
++/* Don't use the IOMMU lock if the device can access the whole of memory */
++#define EFX_DMA_CONSISTENT(_efx) \
++ (((_efx)->dma_mask >> PAGE_SHIFT) >= max_pfn)
++/**
++ * efx_pci_map_single - map buffer for DMA, under IOMMU lock
++ * @pci: PCI device
++ * @ptr: Buffer
++ * @size: Buffer length
++ * @direction: DMA direction
++ *
++ * Wrapper for pci_map_single that uses efx_iommu_lock if necessary.
++ */
++dma_addr_t efx_pci_map_single(struct pci_dev *pci, void *ptr, size_t size,
++ int direction)
++{
++ struct efx_nic *efx = pci_get_drvdata(pci);
++ unsigned long flags __attribute__ ((unused));
++ dma_addr_t dma_addr;
++
++ if (unlikely((efx_use_iommu_lock &&
++ (!EFX_NO_IOMMU) &&
++ (!EFX_DMA_CONSISTENT(efx))) ||
++ efx_use_iommu_lock >= 2)) {
++ spin_lock_irqsave(&efx_iommu_lock, flags);
++ dma_addr = pci_map_single(pci, ptr, size, direction);
++ spin_unlock_irqrestore(&efx_iommu_lock, flags);
++ } else {
++ dma_addr = pci_map_single(pci, ptr, size, direction);
++ }
++ return dma_addr;
+}
+
-+static struct tty_operations xencons_ops = {
-+ .open = xencons_open,
-+ .close = xencons_close,
-+ .write = xencons_write,
-+ .write_room = xencons_write_room,
-+ .put_char = xencons_put_char,
-+ .flush_chars = xencons_flush_chars,
-+ .chars_in_buffer = xencons_chars_in_buffer,
-+ .send_xchar = xencons_send_xchar,
-+ .flush_buffer = xencons_flush_buffer,
-+ .throttle = xencons_throttle,
-+ .unthrottle = xencons_unthrottle,
-+ .wait_until_sent = xencons_wait_until_sent,
-+};
++/**
++ * efx_pci_unmap_single - unmap buffer for DMA, under IOMMU lock
++ * @pci: PCI device
++ * @dma_addr: DMA address
++ * @size: Buffer length
++ * @direction: DMA direction
++ *
++ * Wrapper for pci_unmap_single that uses efx_iommu_lock if necessary.
++ */
++void efx_pci_unmap_single(struct pci_dev *pci, dma_addr_t dma_addr,
++ size_t size, int direction)
++{
++ struct efx_nic *efx = pci_get_drvdata(pci);
++ unsigned long flags __attribute__ ((unused));
++
++ if (unlikely((efx_use_iommu_lock &&
++ (!EFX_NO_IOMMU) &&
++ (!EFX_DMA_CONSISTENT(efx))) ||
++ efx_use_iommu_lock >= 2)) {
++ spin_lock_irqsave(&efx_iommu_lock, flags);
++ pci_unmap_single(pci, dma_addr, size, direction);
++ spin_unlock_irqrestore(&efx_iommu_lock, flags);
++ } else {
++ pci_unmap_single(pci, dma_addr, size, direction);
++ }
++}
+
-+static int __init xencons_init(void)
++/**
++ * efx_pci_alloc_consistent - allocate DMA-consistent buffer, under IOMMU lock
++ * @pci: PCI device
++ * @size: Buffer length
++ * @dma_addr: DMA address
++ *
++ * Wrapper for pci_alloc_consistent that uses efx_iommu_lock if necessary.
++ *
++ * Bugs: Currently this can't use the spinlock because
++ * pci_alloc_consistent may block.
++ */
++void *efx_pci_alloc_consistent(struct pci_dev *pci, size_t size,
++ dma_addr_t *dma_addr)
+{
-+ int rc;
++ return pci_alloc_consistent(pci, size, dma_addr);
++}
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++/**
++ * efx_pci_free_consistent - free DMA-consistent buffer, under IOMMU lock
++ * @pci: PCI device
++ * @size: Buffer length
++ * @ptr: Buffer
++ * @dma_addr: DMA address
++ *
++ * Wrapper for pci_free_consistent that uses efx_iommu_lock if necessary.
++ */
++void efx_pci_free_consistent(struct pci_dev *pci, size_t size, void *ptr,
++ dma_addr_t dma_addr)
++{
++ struct efx_nic *efx = pci_get_drvdata(pci);
++ unsigned long flags __attribute__ ((unused));
++
++ if (unlikely((efx_use_iommu_lock &&
++ (!EFX_NO_IOMMU) &&
++ (!EFX_DMA_CONSISTENT(efx))) ||
++ efx_use_iommu_lock >= 2)) {
++ spin_lock_irqsave(&efx_iommu_lock, flags);
++ pci_free_consistent(pci, size, ptr, dma_addr);
++ spin_unlock_irqrestore(&efx_iommu_lock, flags);
++ } else {
++ pci_free_consistent(pci, size, ptr, dma_addr);
++ }
++}
+
-+ if (xc_mode == XC_OFF)
-+ return 0;
++module_param(efx_use_iommu_lock, uint, 0644);
++MODULE_PARM_DESC(efx_use_iommu_lock, "Enable lock for bug in free_iommu");
+
-+ if (!is_initial_xendomain()) {
-+ rc = xencons_ring_init();
-+ if (rc)
-+ return rc;
-+ }
++#endif
+
-+ xencons_driver = alloc_tty_driver((xc_mode == XC_TTY) ?
-+ MAX_NR_CONSOLES : 1);
-+ if (xencons_driver == NULL)
-+ return -ENOMEM;
-+
-+ DRV(xencons_driver)->name = "xencons";
-+ DRV(xencons_driver)->major = TTY_MAJOR;
-+ DRV(xencons_driver)->type = TTY_DRIVER_TYPE_SERIAL;
-+ DRV(xencons_driver)->subtype = SERIAL_TYPE_NORMAL;
-+ DRV(xencons_driver)->init_termios = tty_std_termios;
-+ DRV(xencons_driver)->flags =
-+ TTY_DRIVER_REAL_RAW |
-+ TTY_DRIVER_RESET_TERMIOS;
-+ DRV(xencons_driver)->termios = xencons_termios;
-+ DRV(xencons_driver)->termios_locked = xencons_termios_locked;
-+
-+ switch (xc_mode) {
-+ case XC_XVC:
-+ DRV(xencons_driver)->name = "xvc";
-+ DRV(xencons_driver)->major = XEN_XVC_MAJOR;
-+ DRV(xencons_driver)->minor_start = XEN_XVC_MINOR;
-+ DRV(xencons_driver)->name_base = xc_num;
-+ break;
-+ case XC_SERIAL:
-+ DRV(xencons_driver)->name = "ttyS";
-+ DRV(xencons_driver)->minor_start = 64 + xc_num;
-+ DRV(xencons_driver)->name_base = xc_num;
-+ break;
-+ default:
-+ DRV(xencons_driver)->name = "tty";
-+ DRV(xencons_driver)->minor_start = 1;
-+ DRV(xencons_driver)->name_base = 1;
-+ break;
-+ }
-+
-+ tty_set_operations(xencons_driver, &xencons_ops);
-+
-+ if ((rc = tty_register_driver(DRV(xencons_driver))) != 0) {
-+ printk("WARNING: Failed to register Xen virtual "
-+ "console driver as '%s%d'\n",
-+ DRV(xencons_driver)->name,
-+ DRV(xencons_driver)->name_base);
-+ put_tty_driver(xencons_driver);
-+ xencons_driver = NULL;
-+ return rc;
-+ }
-+
-+ if (is_initial_xendomain()) {
-+ xencons_priv_irq = bind_virq_to_irqhandler(
-+ VIRQ_CONSOLE,
-+ 0,
-+ xencons_priv_interrupt,
-+ 0,
-+ "console",
-+ NULL);
-+ BUG_ON(xencons_priv_irq < 0);
-+ }
++#ifdef EFX_NEED_COMPOUND_PAGE_FIX
+
-+ printk("Xen virtual console successfully installed as %s%d\n",
-+ DRV(xencons_driver)->name, xc_num);
++void efx_compound_page_destructor(struct page *page)
++{
++ /* Fake up page state to keep __free_pages happy */
++ set_page_count(page, 1);
++ page[1].mapping = NULL;
+
-+ return 0;
++ __free_pages(page, (unsigned long)page[1].index);
+}
+
-+module_init(xencons_init);
-+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/console/Makefile linux-2.6.18-xen.hg/drivers/xen/console/Makefile
---- linux-2.6.18/drivers/xen/console/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/console/Makefile 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,2 @@
++#endif /* NEED_COMPOUND_PAGE_FIX */
+
-+obj-y := console.o xencons_ring.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/console/xencons_ring.c linux-2.6.18-xen.hg/drivers/xen/console/xencons_ring.c
---- linux-2.6.18/drivers/xen/console/xencons_ring.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/console/xencons_ring.c 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,143 @@
-+/*
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/**************************************************************************
++ *
++ * print_hex_dump, taken from lib/hexdump.c.
++ *
++ **************************************************************************
++ *
+ */
++#ifdef EFX_NEED_HEX_DUMP
+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/errno.h>
-+#include <linux/signal.h>
-+#include <linux/sched.h>
-+#include <linux/interrupt.h>
-+#include <linux/tty.h>
-+#include <linux/tty_flip.h>
-+#include <linux/serial.h>
-+#include <linux/major.h>
-+#include <linux/ptrace.h>
-+#include <linux/ioport.h>
-+#include <linux/mm.h>
-+#include <linux/slab.h>
-+
-+#include <asm/hypervisor.h>
-+#include <xen/evtchn.h>
-+#include <xen/xencons.h>
-+#include <linux/wait.h>
-+#include <linux/interrupt.h>
-+#include <linux/sched.h>
-+#include <linux/err.h>
-+#include <xen/interface/io/console.h>
-+
-+static int xencons_irq;
-+
-+static inline struct xencons_interface *xencons_interface(void)
-+{
-+ return mfn_to_virt(xen_start_info->console.domU.mfn);
-+}
-+
-+static inline void notify_daemon(void)
-+{
-+ /* Use evtchn: this is called early, before irq is set up. */
-+ notify_remote_via_evtchn(xen_start_info->console.domU.evtchn);
-+}
++#define hex_asc(x) "0123456789abcdef"[x]
++#define isascii(c) (((unsigned char)(c))<=0x7f)
+
-+int xencons_ring_send(const char *data, unsigned len)
++static void hex_dump_to_buffer(const void *buf, size_t len, int rowsize,
++ int groupsize, char *linebuf, size_t linebuflen,
++ int ascii)
+{
-+ int sent = 0;
-+ struct xencons_interface *intf = xencons_interface();
-+ XENCONS_RING_IDX cons, prod;
-+
-+ cons = intf->out_cons;
-+ prod = intf->out_prod;
-+ mb();
-+ BUG_ON((prod - cons) > sizeof(intf->out));
++ const u8 *ptr = buf;
++ u8 ch;
++ int j, lx = 0;
++ int ascii_column;
+
-+ while ((sent < len) && ((prod - cons) < sizeof(intf->out)))
-+ intf->out[MASK_XENCONS_IDX(prod++, intf->out)] = data[sent++];
++ if (rowsize != 16 && rowsize != 32)
++ rowsize = 16;
+
-+ wmb();
-+ intf->out_prod = prod;
++ if (!len)
++ goto nil;
++ if (len > rowsize) /* limit to one line at a time */
++ len = rowsize;
++ if ((len % groupsize) != 0) /* no mixed size output */
++ groupsize = 1;
+
-+ notify_daemon();
++ switch (groupsize) {
++ case 8: {
++ const u64 *ptr8 = buf;
++ int ngroups = len / groupsize;
+
-+ return sent;
-+}
++ for (j = 0; j < ngroups; j++)
++ lx += scnprintf(linebuf + lx, linebuflen - lx,
++ "%16.16llx ", (unsigned long long)*(ptr8 + j));
++ ascii_column = 17 * ngroups + 2;
++ break;
++ }
+
-+static irqreturn_t handle_input(int irq, void *unused, struct pt_regs *regs)
-+{
-+ struct xencons_interface *intf = xencons_interface();
-+ XENCONS_RING_IDX cons, prod;
++ case 4: {
++ const u32 *ptr4 = buf;
++ int ngroups = len / groupsize;
+
-+ cons = intf->in_cons;
-+ prod = intf->in_prod;
-+ mb();
-+ BUG_ON((prod - cons) > sizeof(intf->in));
++ for (j = 0; j < ngroups; j++)
++ lx += scnprintf(linebuf + lx, linebuflen - lx,
++ "%8.8x ", *(ptr4 + j));
++ ascii_column = 9 * ngroups + 2;
++ break;
++ }
+
-+ while (cons != prod) {
-+ xencons_rx(intf->in+MASK_XENCONS_IDX(cons,intf->in), 1, regs);
-+ cons++;
-+ }
++ case 2: {
++ const u16 *ptr2 = buf;
++ int ngroups = len / groupsize;
+
-+ mb();
-+ intf->in_cons = cons;
++ for (j = 0; j < ngroups; j++)
++ lx += scnprintf(linebuf + lx, linebuflen - lx,
++ "%4.4x ", *(ptr2 + j));
++ ascii_column = 5 * ngroups + 2;
++ break;
++ }
+
-+ notify_daemon();
++ default:
++ for (j = 0; (j < rowsize) && (j < len) && (lx + 4) < linebuflen;
++ j++) {
++ ch = ptr[j];
++ linebuf[lx++] = hex_asc(ch >> 4);
++ linebuf[lx++] = hex_asc(ch & 0x0f);
++ linebuf[lx++] = ' ';
++ }
++ ascii_column = 3 * rowsize + 2;
++ break;
++ }
++ if (!ascii)
++ goto nil;
++
++ while (lx < (linebuflen - 1) && lx < (ascii_column - 1))
++ linebuf[lx++] = ' ';
++ /* Removed is_print() check */
++ for (j = 0; (j < rowsize) && (j < len) && (lx + 2) < linebuflen; j++)
++ linebuf[lx++] = isascii(ptr[j]) ? ptr[j] : '.';
++nil:
++ linebuf[lx++] = '\0';
++}
++
++void print_hex_dump(const char *level, const char *prefix_str, int prefix_type,
++ int rowsize, int groupsize,
++ const void *buf, size_t len, int ascii)
++{
++ const u8 *ptr = buf;
++ int i, linelen, remaining = len;
++ char linebuf[200];
++
++ if (rowsize != 16 && rowsize != 32)
++ rowsize = 16;
++
++ for (i = 0; i < len; i += rowsize) {
++ linelen = min(remaining, rowsize);
++ remaining -= rowsize;
++ hex_dump_to_buffer(ptr + i, linelen, rowsize, groupsize,
++ linebuf, sizeof(linebuf), ascii);
++
++ switch (prefix_type) {
++ case DUMP_PREFIX_ADDRESS:
++ printk("%s%s%*p: %s\n", level, prefix_str,
++ (int)(2 * sizeof(void *)), ptr + i, linebuf);
++ break;
++ case DUMP_PREFIX_OFFSET:
++ printk("%s%s%.8x: %s\n", level, prefix_str, i, linebuf);
++ break;
++ default:
++ printk("%s%s%s\n", level, prefix_str, linebuf);
++ break;
++ }
++ }
++}
+
-+ xencons_tx();
++#endif /* EFX_NEED_HEX_DUMP */
+
-+ return IRQ_HANDLED;
++/**************************************************************************
++ *
++ * print_mac, from net/ethernet/eth.c in v2.6.24
++ *
++ **************************************************************************
++ *
++ */
++#ifdef EFX_NEED_PRINT_MAC
++char *print_mac(char *buf, const u8 *addr)
++{
++ sprintf(buf, "%02x:%02x:%02x:%02x:%02x:%02x",
++ addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
++ return buf;
+}
++#endif /* EFX_NEED_PRINT_MAC */
+
-+int xencons_ring_init(void)
++#ifdef EFX_NEED_CSUM_TCPUDP_NOFOLD
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++__wsum
++csum_tcpudp_nofold (__be32 saddr, __be32 daddr, unsigned short len,
++ unsigned short proto, __wsum sum)
++#else
++__wsum
++csum_tcpudp_nofold (unsigned long saddr, unsigned long daddr,
++ unsigned short len, unsigned short proto, __wsum sum)
++#endif
+{
-+ int irq;
-+
-+ if (xencons_irq)
-+ unbind_from_irqhandler(xencons_irq, NULL);
-+ xencons_irq = 0;
-+
-+ if (!is_running_on_xen() ||
-+ is_initial_xendomain() ||
-+ !xen_start_info->console.domU.evtchn)
-+ return -ENODEV;
-+
-+ irq = bind_caller_port_to_irqhandler(
-+ xen_start_info->console.domU.evtchn,
-+ handle_input, 0, "xencons", NULL);
-+ if (irq < 0) {
-+ printk(KERN_ERR "XEN console request irq failed %i\n", irq);
-+ return irq;
-+ }
++ unsigned long result;
+
-+ xencons_irq = irq;
++ result = (__force u64)saddr + (__force u64)daddr +
++ (__force u64)sum + ((len + proto) << 8);
+
-+ /* In case we have in-flight data after save/restore... */
-+ notify_daemon();
++ /* Fold down to 32-bits so we don't lose in the typedef-less network stack. */
++ /* 64 to 33 */
++ result = (result & 0xffffffff) + (result >> 32);
++ /* 33 to 32 */
++ result = (result & 0xffffffff) + (result >> 32);
++ return (__force __wsum)result;
+
-+ return 0;
+}
++#endif /* EFX_NEED_CSUM_TCPUDP_NOFOLD */
+
-+void xencons_resume(void)
-+{
-+ (void)xencons_ring_init();
++#ifdef EFX_NEED_RANDOM_ETHER_ADDR
++/* Generate random MAC address */
++void efx_random_ether_addr(uint8_t *addr) {
++ get_random_bytes (addr, ETH_ALEN);
++ addr [0] &= 0xfe; /* clear multicast bit */
++ addr [0] |= 0x02; /* set local assignment bit (IEEE802) */
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/cpu_hotplug.c linux-2.6.18-xen.hg/drivers/xen/core/cpu_hotplug.c
---- linux-2.6.18/drivers/xen/core/cpu_hotplug.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/cpu_hotplug.c 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,172 @@
-+#include <linux/init.h>
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/notifier.h>
-+#include <linux/cpu.h>
-+#include <xen/cpu_hotplug.h>
-+#include <xen/xenbus.h>
++#endif /* EFX_NEED_RANDOM_ETHER_ADDR */
+
++#ifdef EFX_NEED_MSECS_TO_JIFFIES
+/*
-+ * Set of CPUs that remote admin software will allow us to bring online.
-+ * Notified to us via xenbus.
++ * When we convert to jiffies then we interpret incoming values
++ * the following way:
++ *
++ * - negative values mean 'infinite timeout' (MAX_JIFFY_OFFSET)
++ *
++ * - 'too large' values [that would result in larger than
++ * MAX_JIFFY_OFFSET values] mean 'infinite timeout' too.
++ *
++ * - all other values are converted to jiffies by either multiplying
++ * the input value by a factor or dividing it with a factor
++ *
++ * We must also be careful about 32-bit overflows.
+ */
-+static cpumask_t xenbus_allowed_cpumask;
-+
-+/* Set of CPUs that local admin will allow us to bring online. */
-+static cpumask_t local_allowed_cpumask = CPU_MASK_ALL;
-+
-+static int local_cpu_hotplug_request(void)
++#ifndef MSEC_PER_SEC
++#define MSEC_PER_SEC 1000L
++#endif
++unsigned long msecs_to_jiffies(const unsigned int m)
+{
+ /*
-+ * We assume a CPU hotplug request comes from local admin if it is made
-+ * via a userspace process (i.e., one with a real mm_struct).
++ * Negative value, means infinite timeout:
+ */
-+ return (current->mm != NULL);
-+}
-+
-+static void vcpu_hotplug(unsigned int cpu)
-+{
-+ int err;
-+ char dir[32], state[32];
-+
-+ if ((cpu >= NR_CPUS) || !cpu_possible(cpu))
-+ return;
-+
-+ sprintf(dir, "cpu/%d", cpu);
-+ err = xenbus_scanf(XBT_NIL, dir, "availability", "%s", state);
-+ if (err != 1) {
-+ printk(KERN_ERR "XENBUS: Unable to read cpu state\n");
-+ return;
-+ }
++ if ((int)m < 0)
++ return MAX_JIFFY_OFFSET;
+
-+ if (strcmp(state, "online") == 0) {
-+ cpu_set(cpu, xenbus_allowed_cpumask);
-+ (void)cpu_up(cpu);
-+ } else if (strcmp(state, "offline") == 0) {
-+ cpu_clear(cpu, xenbus_allowed_cpumask);
-+ (void)cpu_down(cpu);
-+ } else {
-+ printk(KERN_ERR "XENBUS: unknown state(%s) on CPU%d\n",
-+ state, cpu);
-+ }
-+}
++#if HZ <= MSEC_PER_SEC && !(MSEC_PER_SEC % HZ)
++ /*
++ * HZ is equal to or smaller than 1000, and 1000 is a nice
++ * round multiple of HZ, divide with the factor between them,
++ * but round upwards:
++ */
++ return (m + (MSEC_PER_SEC / HZ) - 1) / (MSEC_PER_SEC / HZ);
++#elif HZ > MSEC_PER_SEC && !(HZ % MSEC_PER_SEC)
++ /*
++ * HZ is larger than 1000, and HZ is a nice round multiple of
++ * 1000 - simply multiply with the factor between them.
++ *
++ * But first make sure the multiplication result cannot
++ * overflow:
++ */
++ if (m > jiffies_to_msecs(MAX_JIFFY_OFFSET))
++ return MAX_JIFFY_OFFSET;
+
-+static void handle_vcpu_hotplug_event(
-+ struct xenbus_watch *watch, const char **vec, unsigned int len)
-+{
-+ int cpu;
-+ char *cpustr;
-+ const char *node = vec[XS_WATCH_PATH];
++ return m * (HZ / MSEC_PER_SEC);
++#else
++ /*
++ * Generic case - multiply, round and divide. But first
++ * check that if we are doing a net multiplication, that
++ * we wouldnt overflow:
++ */
++ if (HZ > MSEC_PER_SEC && m > jiffies_to_msecs(MAX_JIFFY_OFFSET))
++ return MAX_JIFFY_OFFSET;
+
-+ if ((cpustr = strstr(node, "cpu/")) != NULL) {
-+ sscanf(cpustr, "cpu/%d", &cpu);
-+ vcpu_hotplug(cpu);
-+ }
++ return (m * HZ + MSEC_PER_SEC - 1) / MSEC_PER_SEC;
++#endif
+}
++#endif /* EFX_NEED_MSECS_TO_JIFFIES */
+
-+static int smpboot_cpu_notify(struct notifier_block *notifier,
-+ unsigned long action, void *hcpu)
++#ifdef EFX_NEED_MSLEEP
++/**
++ * msleep - sleep safely even with waitqueue interruptions
++ * @msecs: Time in milliseconds to sleep for
++ */
++void msleep(unsigned int msecs)
+{
-+ int cpu = (long)hcpu;
-+
-+ /*
-+ * We do this in a callback notifier rather than __cpu_disable()
-+ * because local_cpu_hotplug_request() does not work in the latter
-+ * as it's always executed from within a stopmachine kthread.
-+ */
-+ if ((action == CPU_DOWN_PREPARE) && local_cpu_hotplug_request())
-+ cpu_clear(cpu, local_allowed_cpumask);
++ unsigned long timeout = msecs_to_jiffies(msecs) + 1;
+
-+ return NOTIFY_OK;
++ while (timeout)
++ timeout = schedule_timeout_uninterruptible(timeout);
+}
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/kernel_compat.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/kernel_compat.h 2008-05-19 00:33:28.865809986 +0300
+@@ -0,0 +1,925 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static int setup_cpu_watcher(struct notifier_block *notifier,
-+ unsigned long event, void *data)
-+{
-+ int i;
++#ifndef EFX_KERNEL_COMPAT_H
++#define EFX_KERNEL_COMPAT_H
+
-+ static struct xenbus_watch cpu_watch = {
-+ .node = "cpu",
-+ .callback = handle_vcpu_hotplug_event,
-+ .flags = XBWF_new_thread };
-+ (void)register_xenbus_watch(&cpu_watch);
++#include <linux/version.h>
++#include <linux/sched.h>
++#include <linux/errno.h>
++#include <linux/pci.h>
++#include <linux/workqueue.h>
++#include <linux/moduleparam.h>
++#include <linux/interrupt.h>
++#include <linux/skbuff.h>
++#include <linux/netdevice.h>
+
-+ if (!is_initial_xendomain()) {
-+ for_each_possible_cpu(i)
-+ vcpu_hotplug(i);
-+ printk(KERN_INFO "Brought up %ld CPUs\n",
-+ (long)num_online_cpus());
-+ }
++#include "extraversion.h"
+
-+ return NOTIFY_DONE;
-+}
++/*
++ * Kernel backwards compatibility
++ *
++ * This file provides macros that enable the driver to be compiled on
++ * any kernel from 2.6.9 onward (plus SLES 9 2.6.5), without requiring
++ * explicit version tests scattered throughout the code.
++ */
+
-+static int __init setup_vcpu_hotplug_event(void)
-+{
-+ static struct notifier_block hotplug_cpu = {
-+ .notifier_call = smpboot_cpu_notify };
-+ static struct notifier_block xsn_cpu = {
-+ .notifier_call = setup_cpu_watcher };
++/**************************************************************************
++ *
++ * Version/config/architecture tests to set feature flags
++ *
++ **************************************************************************
++ *
++ * NOTE: For simplicity, these initial version tests cover kernel.org
++ * releases only. Backported features in "enterprise" kernels are
++ * handled further down.
++ */
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,9) && \
++ !(defined(EFX_DIST_SUSE) && \
++ LINUX_VERSION_CODE == KERNEL_VERSION(2,6,5) && \
++ EFX_DIST_KVER_LEVEL_1 == 7)
++ #error "This kernel version is now unsupported"
++#endif
+
-+ register_cpu_notifier(&hotplug_cpu);
-+ register_xenstore_notifier(&xsn_cpu);
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,6)
++ #define EFX_NEED_RANDOM_ETHER_ADDR yes
++#endif
+
-+ return 0;
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,7)
++ #define EFX_NEED_I2C_CLASS_HWMON yes
++ #define EFX_NEED_IF_MII yes
++ #define EFX_NEED_MSLEEP yes
++ #define EFX_NEED_MSECS_TO_JIFFIES yes
++#endif
+
-+arch_initcall(setup_vcpu_hotplug_event);
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,8)
++ #define EFX_USE_MTD_ERASE_FAIL_ADDR yes
++#else
++ #define EFX_NEED_MTD_ERASE_CALLBACK yes
++ #define EFX_NEED_DUMMY_PCI_DISABLE_MSI yes
++ #define EFX_NEED_DUMMY_MSIX yes
++#endif
+
-+int smp_suspend(void)
-+{
-+ int cpu, err;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,9)
++ #define EFX_NEED_BYTEORDER_TYPES yes
++#endif
+
-+ for_each_online_cpu(cpu) {
-+ if (cpu == 0)
-+ continue;
-+ err = cpu_down(cpu);
-+ if (err) {
-+ printk(KERN_CRIT "Failed to take all CPUs "
-+ "down: %d.\n", err);
-+ for_each_possible_cpu(cpu)
-+ vcpu_hotplug(cpu);
-+ return err;
-+ }
-+ }
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,10)
++ #define EFX_NEED_MMIOWB yes
++ #define EFX_NEED_PCI_SAVE_RESTORE_WRAPPERS yes
++#endif
+
-+ return 0;
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,12)
++ #define EFX_NEED_DUMMY_SUPPORTS_GMII yes
++ #define EFX_NEED_MII_CONSTANTS yes
++ #define EFX_NEED_MII_ETHTOOL_FIX yes
++ #define EFX_HAVE_MSIX_TABLE_RESERVED yes
++#endif
+
-+void smp_resume(void)
-+{
-+ int cpu;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,14)
++ #define EFX_NEED_SCHEDULE_TIMEOUT_INTERRUPTIBLE yes
++ #define EFX_NEED_SCHEDULE_TIMEOUT_UNINTERRUPTIBLE yes
++ #define EFX_NEED_GFP_T yes
++ #define EFX_NEED_KZALLOC yes
++#endif
+
-+ for_each_possible_cpu(cpu)
-+ vcpu_hotplug(cpu);
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,15)
++ #define EFX_NEED_SETUP_TIMER yes
++ #ifdef CONFIG_HUGETLB_PAGE
++ #define EFX_USE_COMPOUND_PAGES yes
++ #endif
++#else
++ #define EFX_USE_COMPOUND_PAGES yes
++#endif
+
-+int cpu_up_check(unsigned int cpu)
-+{
-+ int rc = 0;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16)
++ #define EFX_NEED_MUTEX yes
++ #define EFX_NEED_SAFE_LISTS yes
++ #ifdef EFX_USE_COMPOUND_PAGES
++ #define EFX_NEED_COMPOUND_PAGE_FIX yes
++ #endif
++#endif
+
-+ if (local_cpu_hotplug_request()) {
-+ cpu_set(cpu, local_allowed_cpumask);
-+ if (!cpu_isset(cpu, xenbus_allowed_cpumask)) {
-+ printk("%s: attempt to bring up CPU %u disallowed by "
-+ "remote admin.\n", __FUNCTION__, cpu);
-+ rc = -EBUSY;
-+ }
-+ } else if (!cpu_isset(cpu, local_allowed_cpumask) ||
-+ !cpu_isset(cpu, xenbus_allowed_cpumask)) {
-+ rc = -EBUSY;
-+ }
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
++ #define EFX_NEED_UNREGISTER_NETDEVICE_NOTIFIER_FIX yes
++ #define EFX_NEED_DEV_NOTICE yes
++#endif
+
-+ return rc;
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18)
++ #define EFX_NEED_IRQF_FLAGS yes
++ #define EFX_NEED_NETDEV_ALLOC_SKB yes
++ /* Fedora backported 2.6.18 netdevice.h changes */
++ #ifndef NETIF_F_GSO
++ #define EFX_NEED_NETIF_TX_LOCK yes
++ #endif
++#else
++ #define EFX_USE_MTD_WRITESIZE yes
++#endif
+
-+void init_xenbus_allowed_cpumask(void)
-+{
-+ xenbus_allowed_cpumask = cpu_present_map;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/evtchn.c linux-2.6.18-xen.hg/drivers/xen/core/evtchn.c
---- linux-2.6.18/drivers/xen/core/evtchn.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/evtchn.c 2007-12-23 11:15:33.554576016 +0100
-@@ -0,0 +1,1060 @@
-+/******************************************************************************
-+ * evtchn.c
-+ *
-+ * Communication via Xen event channels.
-+ *
-+ * Copyright (c) 2002-2005, K A Fraser
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
++ #define EFX_NEED_IRQ_HANDLER_T yes
++ #define EFX_HAVE_IRQ_HANDLER_REGS yes
++#endif
+
-+#include <linux/module.h>
-+#include <linux/irq.h>
-+#include <linux/interrupt.h>
-+#include <linux/sched.h>
-+#include <linux/kernel_stat.h>
-+#include <linux/version.h>
-+#include <asm/atomic.h>
-+#include <asm/system.h>
-+#include <asm/ptrace.h>
-+#include <asm/synch_bitops.h>
-+#include <xen/evtchn.h>
-+#include <xen/interface/event_channel.h>
-+#include <xen/interface/physdev.h>
-+#include <asm/hypervisor.h>
-+#include <linux/mc146818rtc.h> /* RTC_IRQ */
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)
++ #define EFX_NEED_WORK_API_WRAPPERS yes
++ #define EFX_USE_FASTCALL yes
++ #define EFX_NEED_CSUM_UNFOLDED yes
++#endif
+
-+/*
-+ * This lock protects updates to the following mapping and reference-count
-+ * arrays. The lock does not need to be acquired to read the mapping tables.
-+ */
-+static DEFINE_SPINLOCK(irq_mapping_update_lock);
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21)
++ /*
++ * debugfs was introduced earlier, but only supports sym-links
++ * from 2.6.21
++ */
++ #ifdef CONFIG_DEBUG_FS
++ #define EFX_USE_DEBUGFS yes
++ #endif
++#endif
+
-+/* IRQ <-> event-channel mappings. */
-+static int evtchn_to_irq[NR_EVENT_CHANNELS] = {
-+ [0 ... NR_EVENT_CHANNELS-1] = -1 };
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,22)
++ #define EFX_NEED_SKB_HEADER_MACROS yes
++ #define EFX_NEED_HEX_DUMP yes
++#else
++ #define EFX_USE_CANCEL_WORK_SYNC yes
++#endif
+
-+/* Packed IRQ information: binding type, sub-type index, and event channel. */
-+static u32 irq_info[NR_IRQS];
++#if LINUX_VERSION_CODE == KERNEL_VERSION(2,6,22)
++ #define EFX_NEED_HEX_DUMP_CONST_FIX yes
++#endif
+
-+/* Binding types. */
-+enum {
-+ IRQT_UNBOUND,
-+ IRQT_PIRQ,
-+ IRQT_VIRQ,
-+ IRQT_IPI,
-+ IRQT_LOCAL_PORT,
-+ IRQT_CALLER_PORT
-+};
++#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,14)) && \
++ (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23))
++ #define EFX_USE_ETHTOOL_GET_PERM_ADDR yes
++#endif
+
-+/* Constructor for packed IRQ information. */
-+static inline u32 mk_irq_info(u32 type, u32 index, u32 evtchn)
-+{
-+ return ((type << 24) | (index << 16) | evtchn);
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
++ #ifdef __ia64__
++ /* csum_tcpudp_nofold() is extern but not exported */
++ #define EFX_NEED_CSUM_TCPUDP_NOFOLD yes
++ #endif
++#else
++ #define EFX_USE_PCI_DEV_REVISION yes
++ #define EFX_USE_CANCEL_DELAYED_WORK_SYNC yes
++#endif
+
-+/* Convenient shorthand for packed representation of an unbound IRQ. */
-+#define IRQ_UNBOUND mk_irq_info(IRQT_UNBOUND, 0, 0)
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
++ #define EFX_HAVE_OLD_NAPI yes
++ #define EFX_NEED_GENERIC_LRO yes
++ #define EFX_NEED_PRINT_MAC yes
++#else
++ #define EFX_USE_ETHTOOL_FLAGS yes
++#endif
+
+/*
-+ * Accessors for packed IRQ information.
++ * SFC Bug 4560: Some kernels leak IOMMU entries under heavy load. Use a
++ * spinlock to serialise access where possible to alleviate the
++ * problem.
++ *
++ * NB. The following definition is duplicated in
++ * the char driver. Please keep in sync.
+ */
++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) && \
++ defined(__x86_64__) && defined(CONFIG_SMP))
++ #define EFX_NEED_IOMMU_LOCK yes
++ #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,15)
++ #if defined(CONFIG_GART_IOMMU)
++ #define EFX_NO_IOMMU no_iommu
++ #else
++ #define EFX_NO_IOMMU 1
++ #endif
++ #else
++ #define EFX_NO_IOMMU 0
++ #endif
++#endif
+
-+static inline unsigned int evtchn_from_irq(int irq)
-+{
-+ return (u16)(irq_info[irq]);
-+}
++#ifdef CONFIG_PPC64
++ /* __raw_writel and friends are broken on ppc64 */
++ #define EFX_NEED_RAW_READ_AND_WRITE_FIX yes
++#endif
+
-+static inline unsigned int index_from_irq(int irq)
-+{
-+ return (u8)(irq_info[irq] >> 16);
-+}
++/**************************************************************************
++ *
++ * Exceptions for backported features
++ *
++ **************************************************************************
++ */
+
-+static inline unsigned int type_from_irq(int irq)
-+{
-+ return (u8)(irq_info[irq] >> 24);
-+}
++/* RHEL4 */
++#if defined(EFX_DIST_RHEL) && LINUX_VERSION_CODE == KERNEL_VERSION(2,6,9)
++ #if EFX_DIST_KVER_LEVEL_1 >= 22
++ /* linux-2.6.9-mmiowb.patch */
++ #undef EFX_NEED_MMIOWB
++ #endif
++ #if EFX_DIST_KVER_LEVEL_1 >= 34
++ /* linux-2.6.9-net-mii-update.patch */
++ #undef EFX_NEED_DUMMY_SUPPORTS_GMII
++ #undef EFX_NEED_MII_CONSTANTS
++ #undef EFX_NEED_MII_ETHTOOL_FIX
++ /* linux-2.6.9-gfp_t-typedef.patch */
++ #undef EFX_NEED_GFP_T
++ /* linux-2.6.9-slab-update.patch */
++ #undef EFX_NEED_KZALLOC
++ #endif
++ #if EFX_DIST_KVER_LEVEL_1 >= 55
++ /* linux-2.6.18-sata-update.patch (ported from 2.6.18->2.6.9) */
++ #undef EFX_NEED_SCHEDULE_TIMEOUT_UNINTERRUPTIBLE
++ #undef EFX_NEED_IRQ_HANDLER_T
++ #endif
++#endif
+
-+/* IRQ <-> VIRQ mapping. */
-+DEFINE_PER_CPU(int, virq_to_irq[NR_VIRQS]) = {[0 ... NR_VIRQS-1] = -1};
++/* RHEL5 */
++#if defined(EFX_DIST_RHEL) && LINUX_VERSION_CODE == KERNEL_VERSION(2,6,18)
++ #if EFX_DIST_KVER_LEVEL_1 >= 53
++ /* linux-2.6.18-sata-update.patch */
++ #undef EFX_NEED_SCHEDULE_TIMEOUT_UNINTERRUPTIBLE
++ #undef EFX_NEED_IRQ_HANDLER_T
++ #endif
++#endif
+
-+/* IRQ <-> IPI mapping. */
-+#ifndef NR_IPIS
-+#define NR_IPIS 1
++#if defined(EFX_DIST_RHEL)
++ #if (LINUX_VERSION_CODE != KERNEL_VERSION(2,6,9)) && \
++ (LINUX_VERSION_CODE != KERNEL_VERSION(2,6,18))
++ #error "Unknown Red Hat Enterprise kernel version"
++ #endif
+#endif
-+DEFINE_PER_CPU(int, ipi_to_irq[NR_IPIS]) = {[0 ... NR_IPIS-1] = -1};
+
-+/* Reference counts for bindings to IRQs. */
-+static int irq_bindcount[NR_IRQS];
++/* SLES9 */
++#if defined(EFX_DIST_SUSE) && LINUX_VERSION_CODE == KERNEL_VERSION(2,6,5) && \
++ EFX_DIST_KVER_LEVEL_1 == 7
++ #if EFX_DIST_KVER_LEVEL_2 >= 139
++ #undef EFX_NEED_MMIOWB
++ #endif
++ #if EFX_DIST_KVER_LEVEL_2 >= 191
++ #undef EFX_NEED_MSLEEP
++ #undef EFX_NEED_MSECS_TO_JIFFIES
++ #endif
++ #if EFX_DIST_KVER_LEVEL_2 >= 244
++ #undef EFX_NEED_BYTEORDER_TYPES
++ #endif
++ #if EFX_DIST_KVER_LEVEL_2 >= 252
++ #undef EFX_NEED_KZALLOC
++ #endif
++#endif
+
-+/* Bitmap indicating which PIRQs require Xen to be notified on unmask. */
-+static DECLARE_BITMAP(pirq_needs_eoi, NR_PIRQS);
++/**************************************************************************
++ *
++ * Definitions of missing constants, types, functions and macros
++ *
++ **************************************************************************
++ *
++ */
+
-+#ifdef CONFIG_SMP
++#ifndef DMA_40BIT_MASK
++ #define DMA_40BIT_MASK 0x000000ffffffffffULL
++#endif
+
-+static u8 cpu_evtchn[NR_EVENT_CHANNELS];
-+static unsigned long cpu_evtchn_mask[NR_CPUS][NR_EVENT_CHANNELS/BITS_PER_LONG];
++#ifndef spin_trylock_irqsave
++ #define spin_trylock_irqsave(lock, flags) \
++ ({ \
++ local_irq_save(flags); \
++ spin_trylock(lock) ? \
++ 1 : ({local_irq_restore(flags); 0;}); \
++ })
++#endif
+
-+static inline unsigned long active_evtchns(unsigned int cpu, shared_info_t *sh,
-+ unsigned int idx)
-+{
-+ return (sh->evtchn_pending[idx] &
-+ cpu_evtchn_mask[cpu][idx] &
-+ ~sh->evtchn_mask[idx]);
-+}
++#ifndef raw_smp_processor_id
++ #define raw_smp_processor_id() (current_thread_info()->cpu)
++#endif
+
-+static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
-+{
-+ shared_info_t *s = HYPERVISOR_shared_info;
-+ int irq = evtchn_to_irq[chn];
++#ifndef NETIF_F_LRO
++ #define NETIF_F_LRO 0
++#endif
+
-+ BUG_ON(!test_bit(chn, s->evtchn_mask));
++/* Cope with small changes in PCI constants between minor kernel revisions */
++#if PCI_X_STATUS != 4
++ #undef PCI_X_STATUS
++ #define PCI_X_STATUS 4
++ #undef PCI_X_STATUS_MAX_SPLIT
++ #define PCI_X_STATUS_MAX_SPLIT 0x03800000
++#endif
+
-+ if (irq != -1)
-+ set_native_irq_info(irq, cpumask_of_cpu(cpu));
++#ifndef PCI_EXP_LNKSTA
++ #define PCI_EXP_LNKSTA 18 /* Link Status */
++#endif
+
-+ clear_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu_evtchn[chn]]);
-+ set_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu]);
-+ cpu_evtchn[chn] = cpu;
-+}
++/* Used for struct pt_regs */
++#ifndef regs_return_value
++ #if defined(__x86_64__)
++ #define regs_return_value(regs) ((regs)->rax)
++ #elif defined(__i386__)
++ #define regs_return_value(regs) ((regs)->eax)
++ #elif defined(__ia64__)
++ #define regs_return_value(regs) ((regs)->r8)
++ #else
++ #error "Need definition for regs_return_value()"
++ #endif
++#endif
+
-+static void init_evtchn_cpu_bindings(void)
-+{
-+ int i;
++#ifndef __GFP_COMP
++ #define __GFP_COMP 0
++#endif
+
-+ /* By default all event channels notify CPU#0. */
-+ for (i = 0; i < NR_IRQS; i++)
-+ set_native_irq_info(i, cpumask_of_cpu(0));
++#ifndef __iomem
++ #define __iomem
++#endif
+
-+ memset(cpu_evtchn, 0, sizeof(cpu_evtchn));
-+ memset(cpu_evtchn_mask[0], ~0, sizeof(cpu_evtchn_mask[0]));
-+}
++#ifndef NET_IP_ALIGN
++ #define NET_IP_ALIGN 2
++#endif
+
-+static inline unsigned int cpu_from_evtchn(unsigned int evtchn)
-+{
-+ return cpu_evtchn[evtchn];
-+}
++#ifndef PCI_CAP_ID_EXP
++#define PCI_CAP_ID_EXP 0x10 /* PCI Express */
++#endif
+
-+#else
++#ifndef PCI_EXP_FLAGS
++#define PCI_EXP_FLAGS 2 /* Capabilities register */
++#define PCI_EXP_FLAGS_TYPE 0x00f0 /* Capability version */
++#define PCI_EXP_TYPE_ENDPOINT 0x0 /* Express Endpoint */
++#define PCI_EXP_TYPE_LEG_END 0x1 /* Legacy Endpoint */
++#define PCI_EXP_TYPE_ROOT_PORT 0x4 /* Root Port */
++#endif
+
-+static inline unsigned long active_evtchns(unsigned int cpu, shared_info_t *sh,
-+ unsigned int idx)
-+{
-+ return (sh->evtchn_pending[idx] & ~sh->evtchn_mask[idx]);
-+}
++#ifndef PCI_EXP_DEVCAP
++#define PCI_EXP_DEVCAP 4 /* Device capabilities */
++#define PCI_EXP_DEVCAP_PAYLOAD 0x07 /* Max_Payload_Size */
++#define PCI_EXP_DEVCAP_PWR_VAL 0x3fc0000 /* Slot Power Limit Value */
++#define PCI_EXP_DEVCAP_PWR_SCL 0xc000000 /* Slot Power Limit Scale */
++#endif
+
-+static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
-+{
-+}
++#ifndef PCI_EXP_DEVCTL
++#define PCI_EXP_DEVCTL 8 /* Device Control */
++#define PCI_EXP_DEVCTL_PAYLOAD 0x00e0 /* Max_Payload_Size */
++#define PCI_EXP_DEVCTL_READRQ 0x7000 /* Max_Read_Request_Size */
++#endif
+
-+static void init_evtchn_cpu_bindings(void)
-+{
-+}
++#ifndef PCI_EXP_LNKSTA
++#define PCI_EXP_LNKSTA 18 /* Link Status */
++#endif
+
-+static inline unsigned int cpu_from_evtchn(unsigned int evtchn)
-+{
-+ return 0;
-+}
++#ifndef NETDEV_TX_OK
++ #define NETDEV_TX_OK 0
++#endif
+
++#ifndef NETDEV_TX_BUSY
++ #define NETDEV_TX_BUSY 1
+#endif
+
-+/* Upcall to generic IRQ layer. */
-+#ifdef CONFIG_X86
-+extern fastcall unsigned int do_IRQ(struct pt_regs *regs);
-+void __init xen_init_IRQ(void);
-+void __init init_IRQ(void)
-+{
-+ irq_ctx_init(0);
-+ xen_init_IRQ();
-+}
-+#if defined (__i386__)
-+static inline void exit_idle(void) {}
-+#define IRQ_REG orig_eax
-+#elif defined (__x86_64__)
-+#include <asm/idle.h>
-+#define IRQ_REG orig_rax
++#ifndef __force
++ #define __force
+#endif
-+#define do_IRQ(irq, regs) do { \
-+ (regs)->IRQ_REG = ~(irq); \
-+ do_IRQ((regs)); \
-+} while (0)
-+#elif defined (__powerpc__)
-+#define do_IRQ(irq, regs) __do_IRQ(irq, regs)
-+static inline void exit_idle(void) {}
++
++#if ! defined(for_each_cpu_mask) && ! defined(CONFIG_SMP)
++ #define for_each_cpu_mask(cpu, mask) \
++ for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask)
+#endif
+
-+/* Xen will never allocate port zero for any purpose. */
-+#define VALID_EVTCHN(chn) ((chn) != 0)
++/**************************************************************************/
+
-+/*
-+ * Force a proper event-channel callback from Xen after clearing the
-+ * callback mask. We do this in a very simple manner, by making a call
-+ * down into Xen. The pending flag will be checked by Xen on return.
-+ */
-+void force_evtchn_callback(void)
-+{
-+ (void)HYPERVISOR_xen_version(0, NULL);
-+}
-+/* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */
-+EXPORT_SYMBOL(force_evtchn_callback);
++#ifdef EFX_NEED_IRQ_HANDLER_T
++ typedef irqreturn_t (*irq_handler_t)(int, void *, struct pt_regs *);
++#endif
+
-+static DEFINE_PER_CPU(unsigned int, upcall_count) = { 0 };
-+static DEFINE_PER_CPU(unsigned int, last_processed_l1i) = { BITS_PER_LONG - 1 };
-+static DEFINE_PER_CPU(unsigned int, last_processed_l2i) = { BITS_PER_LONG - 1 };
++#ifdef EFX_NEED_I2C_CLASS_HWMON
++ #define I2C_CLASS_HWMON (1<<0)
++#endif
+
-+/* NB. Interrupts are disabled on entry. */
-+asmlinkage void evtchn_do_upcall(struct pt_regs *regs)
-+{
-+ unsigned long l1, l2;
-+ unsigned long masked_l1, masked_l2;
-+ unsigned int l1i, l2i, port, count;
-+ int irq, cpu = smp_processor_id();
-+ shared_info_t *s = HYPERVISOR_shared_info;
-+ vcpu_info_t *vcpu_info = &s->vcpu_info[cpu];
++#ifdef EFX_NEED_MII_CONSTANTS
++ #define MII_CTRL1000 0x09
++ #define MII_STAT1000 0x0a
++ #define BMCR_SPEED1000 0x0040
++ #define ADVERTISE_PAUSE_ASYM 0x0800
++ #define ADVERTISE_PAUSE_CAP 0x0400
++ #define LPA_PAUSE_ASYM 0x0800
++ #define LPA_PAUSE_CAP 0x0400
++ #define ADVERTISE_1000FULL 0x0200
++ #define ADVERTISE_1000HALF 0x0100
++ #define LPA_1000FULL 0x0800
++ #define LPA_1000HALF 0x0400
++#endif
+
++#ifdef EFX_NEED_DUMMY_SUPPORTS_GMII
++ #include <linux/mii.h>
++ /* Ugly; redirect nonexistent new field to an old unused field */
++ #undef supports_gmii
++ #define supports_gmii full_duplex
++#endif
+
-+ do {
-+ /* Avoid a callback storm when we reenable delivery. */
-+ vcpu_info->evtchn_upcall_pending = 0;
++#ifdef EFX_NEED_SKB_HEADER_MACROS
++ #define skb_mac_header(skb) (skb)->mac.raw
++ #define skb_network_header(skb) (skb)->nh.raw
++ #define tcp_hdr(skb) (skb)->h.th
++ #define ip_hdr(skb) (skb)->nh.iph
++ #define skb_tail_pointer(skb) (skb)->tail
++#endif
+
-+ /* Nested invocations bail immediately. */
-+ if (unlikely(per_cpu(upcall_count, cpu)++))
-+ return;
++#ifdef EFX_NEED_RAW_READ_AND_WRITE_FIX
++ #include <asm/io.h>
++ static inline void
++ efx_raw_writeb(u8 value, volatile void __iomem *addr)
++ {
++ writeb(value, addr);
++ }
++ static inline void
++ efx_raw_writew(u16 value, volatile void __iomem *addr)
++ {
++ writew(le16_to_cpu(value), addr);
++ }
++ static inline void
++ efx_raw_writel(u32 value, volatile void __iomem *addr)
++ {
++ writel(le32_to_cpu(value), addr);
++ }
++ static inline void
++ efx_raw_writeq(u64 value, volatile void __iomem *addr)
++ {
++ writeq(le64_to_cpu(value), addr);
++ }
++ static inline u8
++ efx_raw_readb(const volatile void __iomem *addr)
++ {
++ return readb(addr);
++ }
++ static inline u16
++ efx_raw_readw(const volatile void __iomem *addr)
++ {
++ return cpu_to_le16(readw(addr));
++ }
++ static inline u32
++ efx_raw_readl(const volatile void __iomem *addr)
++ {
++ return cpu_to_le32(readl(addr));
++ }
++ static inline u64
++ efx_raw_readq(const volatile void __iomem *addr)
++ {
++ return cpu_to_le64(readq(addr));
++ }
++
++ #undef __raw_writeb
++ #undef __raw_writew
++ #undef __raw_writel
++ #undef __raw_writeq
++ #undef __raw_readb
++ #undef __raw_readw
++ #undef __raw_readl
++ #undef __raw_readq
++ #define __raw_writeb efx_raw_writeb
++ #define __raw_writew efx_raw_writew
++ #define __raw_writel efx_raw_writel
++ #define __raw_writeq efx_raw_writeq
++ #define __raw_readb efx_raw_readb
++ #define __raw_readw efx_raw_readw
++ #define __raw_readl efx_raw_readl
++ #define __raw_readq efx_raw_readq
++#endif
++
++#ifdef EFX_NEED_SCHEDULE_TIMEOUT_INTERRUPTIBLE
++ static inline signed long
++ schedule_timeout_interruptible(signed long timeout)
++ {
++ set_current_state(TASK_INTERRUPTIBLE);
++ return schedule_timeout(timeout);
++ }
++#endif
+
-+#ifndef CONFIG_X86 /* No need for a barrier -- XCHG is a barrier on x86. */
-+ /* Clear master flag /before/ clearing selector flag. */
-+ rmb();
++#ifdef EFX_NEED_SCHEDULE_TIMEOUT_UNINTERRUPTIBLE
++ static inline signed long
++ schedule_timeout_uninterruptible(signed long timeout)
++ {
++ set_current_state(TASK_UNINTERRUPTIBLE);
++ return schedule_timeout(timeout);
++ }
+#endif
-+ l1 = xchg(&vcpu_info->evtchn_pending_sel, 0);
+
-+ l1i = per_cpu(last_processed_l1i, cpu);
-+ l2i = per_cpu(last_processed_l2i, cpu);
++#ifdef EFX_NEED_MMIOWB
++ #if defined(__i386__) || defined(__x86_64__)
++ #define mmiowb()
++ #elif defined(__ia64__)
++ #ifndef ia64_mfa
++ #define ia64_mfa() asm volatile ("mf.a" ::: "memory")
++ #endif
++ #define mmiowb ia64_mfa
++ #else
++ #error "Need definition for mmiowb()"
++ #endif
++#endif
+
-+ while (l1 != 0) {
++#ifdef EFX_NEED_KZALLOC
++ static inline void *kzalloc(size_t size, int flags)
++ {
++ void *buf = kmalloc(size, flags);
++ if (buf)
++ memset(buf, 0,size);
++ return buf;
++ }
++#endif
+
-+ l1i = (l1i + 1) % BITS_PER_LONG;
-+ masked_l1 = l1 & ((~0UL) << l1i);
++#ifdef EFX_NEED_SETUP_TIMER
++ static inline void setup_timer(struct timer_list * timer,
++ void (*function)(unsigned long),
++ unsigned long data)
++ {
++ timer->function = function;
++ timer->data = data;
++ init_timer(timer);
++ }
++#endif
+
-+ if (masked_l1 == 0) { /* if we masked out all events, wrap around to the beginning */
-+ l1i = BITS_PER_LONG - 1;
-+ l2i = BITS_PER_LONG - 1;
-+ continue;
-+ }
-+ l1i = __ffs(masked_l1);
++#ifdef EFX_NEED_MUTEX
++ #define EFX_DEFINE_MUTEX(x) DECLARE_MUTEX(x)
++ #undef DEFINE_MUTEX
++ #define DEFINE_MUTEX EFX_DEFINE_MUTEX
+
-+ do {
-+ l2 = active_evtchns(cpu, s, l1i);
++ #define efx_mutex semaphore
++ #undef mutex
++ #define mutex efx_mutex
+
-+ l2i = (l2i + 1) % BITS_PER_LONG;
-+ masked_l2 = l2 & ((~0UL) << l2i);
++ #define efx_mutex_init(x) init_MUTEX(x)
++ #undef mutex_init
++ #define mutex_init efx_mutex_init
+
-+ if (masked_l2 == 0) { /* if we masked out all events, move on */
-+ l2i = BITS_PER_LONG - 1;
-+ break;
-+ }
++ #define efx_mutex_destroy(x) do { } while(0)
++ #undef mutex_destroy
++ #define mutex_destroy efx_mutex_destroy
+
-+ l2i = __ffs(masked_l2);
++ #define efx_mutex_lock(x) down(x)
++ #undef mutex_lock
++ #define mutex_lock efx_mutex_lock
+
-+ /* process port */
-+ port = (l1i * BITS_PER_LONG) + l2i;
-+ if ((irq = evtchn_to_irq[port]) != -1)
-+ do_IRQ(irq, regs);
-+ else {
-+ exit_idle();
-+ evtchn_device_upcall(port);
-+ }
++ #define efx_mutex_lock_interruptible(x) down_interruptible(x)
++ #undef mutex_lock_interruptible
++ #define mutex_lock_interruptible efx_mutex_lock_interruptible
+
-+ /* if this is the final port processed, we'll pick up here+1 next time */
-+ per_cpu(last_processed_l1i, cpu) = l1i;
-+ per_cpu(last_processed_l2i, cpu) = l2i;
++ #define efx_mutex_unlock(x) up(x)
++ #undef mutex_unlock
++ #define mutex_unlock efx_mutex_unlock
+
-+ } while (l2i != BITS_PER_LONG - 1);
++ #define efx_mutex_trylock(x) (!down_trylock(x))
++ #undef mutex_trylock
++ #define mutex_trylock efx_mutex_trylock
+
-+ l2 = active_evtchns(cpu, s, l1i);
-+ if (l2 == 0) /* we handled all ports, so we can clear the selector bit */
-+ l1 &= ~(1UL << l1i);
++ static inline int efx_mutex_is_locked(struct efx_mutex *m)
++ {
++ /* NB. This is quite inefficient, but it's the best we
++ * can do with the semaphore API. */
++ if ( down_trylock(m) )
++ return 1;
++ /* Undo the effect of down_trylock. */
++ up(m);
++ return 0;
++ }
++ #undef mutex_is_locked
++ #define mutex_is_locked efx_mutex_is_locked
++#endif
+
-+ }
++#ifndef NETIF_F_GSO
++ #define efx_gso_size tso_size
++ #undef gso_size
++ #define gso_size efx_gso_size
++ #define efx_gso_segs tso_segs
++ #undef gso_segs
++ #define gso_segs efx_gso_segs
++#endif
+
-+ /* If there were nested callbacks then we have more to do. */
-+ count = per_cpu(upcall_count, cpu);
-+ per_cpu(upcall_count, cpu) = 0;
-+ } while (unlikely(count != 1));
-+}
++#ifdef EFX_NEED_IRQF_FLAGS
++ #ifdef SA_PROBEIRQ
++ #define IRQF_PROBE_SHARED SA_PROBEIRQ
++ #else
++ #define IRQF_PROBE_SHARED 0
++ #endif
++ #define IRQF_SHARED SA_SHIRQ
++#endif
+
-+static int find_unbound_irq(void)
-+{
-+ static int warned;
-+ int dynirq, irq;
++#ifdef EFX_NEED_NETDEV_ALLOC_SKB
++ #ifndef NET_SKB_PAD
++ #define NET_SKB_PAD 16
++ #endif
+
-+ for (dynirq = 0; dynirq < NR_DYNIRQS; dynirq++) {
-+ irq = dynirq_to_irq(dynirq);
-+ if (irq_bindcount[irq] == 0)
-+ return irq;
++ static inline
++ struct sk_buff *netdev_alloc_skb(struct net_device *dev,
++ unsigned int length)
++ {
++ struct sk_buff *skb = alloc_skb(length + NET_SKB_PAD,
++ GFP_ATOMIC | __GFP_COLD);
++ if (likely(skb)) {
++ skb_reserve(skb, NET_SKB_PAD);
++ skb->dev = dev;
++ }
++ return skb;
+ }
++#endif
+
-+ if (!warned) {
-+ warned = 1;
-+ printk(KERN_WARNING "No available IRQ to bind to: "
-+ "increase NR_DYNIRQS.\n");
++#ifdef EFX_NEED_NETIF_TX_LOCK
++ static inline void netif_tx_lock(struct net_device *dev)
++ {
++ spin_lock(&dev->xmit_lock);
++ dev->xmit_lock_owner = smp_processor_id();
++ }
++ static inline void netif_tx_lock_bh(struct net_device *dev)
++ {
++ spin_lock_bh(&dev->xmit_lock);
++ dev->xmit_lock_owner = smp_processor_id();
++ }
++ static inline void netif_tx_unlock_bh(struct net_device *dev)
++ {
++ dev->xmit_lock_owner = -1;
++ spin_unlock_bh(&dev->xmit_lock);
++ }
++ static inline void netif_tx_unlock(struct net_device *dev)
++ {
++ dev->xmit_lock_owner = -1;
++ spin_unlock(&dev->xmit_lock);
+ }
++#endif
+
-+ return -ENOSPC;
-+}
++#ifdef EFX_NEED_CSUM_UNFOLDED
++ typedef u32 __wsum;
++ #define csum_unfold(x) ((__force __wsum) x)
++#endif
+
-+static int bind_caller_port_to_irq(unsigned int caller_port)
-+{
-+ int irq;
++#ifdef EFX_NEED_HEX_DUMP
++ enum {
++ DUMP_PREFIX_NONE,
++ DUMP_PREFIX_ADDRESS,
++ DUMP_PREFIX_OFFSET
++ };
++#endif
+
-+ spin_lock(&irq_mapping_update_lock);
++#ifdef EFX_NEED_PRINT_MAC
++ #define DECLARE_MAC_BUF(var) char var[18] __attribute__((unused))
++#endif
+
-+ if ((irq = evtchn_to_irq[caller_port]) == -1) {
-+ if ((irq = find_unbound_irq()) < 0)
-+ goto out;
++#ifdef EFX_NEED_GFP_T
++ typedef unsigned int gfp_t;
++#endif
+
-+ evtchn_to_irq[caller_port] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_CALLER_PORT, 0, caller_port);
++#ifdef EFX_NEED_SAFE_LISTS
++ #define list_for_each_entry_safe_reverse(pos, n, head, member) \
++ for (pos = list_entry((head)->prev, typeof(*pos), member), \
++ n = list_entry(pos->member.prev, typeof(*pos), member); \
++ &pos->member != (head); \
++ pos = n, \
++ n = list_entry(n->member.prev, typeof(*n), member))
++#endif
++
++#ifdef EFX_NEED_DEV_NOTICE
++ #define dev_notice dev_warn
++#endif
++
++#ifdef EFX_NEED_IF_MII
++ #include <linux/mii.h>
++ static inline struct mii_ioctl_data *efx_if_mii ( struct ifreq *rq ) {
++ return ( struct mii_ioctl_data * ) &rq->ifr_ifru;
+ }
++ #undef if_mii
++ #define if_mii efx_if_mii
++#endif
+
-+ irq_bindcount[irq]++;
++#ifdef EFX_NEED_MTD_ERASE_CALLBACK
++ #include <linux/mtd/mtd.h>
++ static inline void efx_mtd_erase_callback(struct erase_info *instr) {
++ if ( instr->callback )
++ instr->callback ( instr );
++ }
++ #undef mtd_erase_callback
++ #define mtd_erase_callback efx_mtd_erase_callback
++#endif
+
-+ out:
-+ spin_unlock(&irq_mapping_update_lock);
-+ return irq;
-+}
++#ifdef EFX_NEED_DUMMY_PCI_DISABLE_MSI
++ #include <linux/pci.h>
++ static inline void dummy_pci_disable_msi ( struct pci_dev *dev ) {
++ /* Do nothing */
++ }
++ #undef pci_disable_msi
++ #define pci_disable_msi dummy_pci_disable_msi
++#endif
+
-+static int bind_local_port_to_irq(unsigned int local_port)
-+{
-+ int irq;
++#ifdef EFX_NEED_DUMMY_MSIX
++ struct msix_entry {
++ u16 vector; /* kernel uses to write allocated vector */
++ u16 entry; /* driver uses to specify entry, OS writes */
++ };
++ static inline int pci_enable_msix(struct pci_dev* dev,
++ struct msix_entry *entries, int nvec)
++ {return -1;}
++ static inline void pci_disable_msix(struct pci_dev *dev) { /* Do nothing */}
++#endif
+
-+ spin_lock(&irq_mapping_update_lock);
++#ifdef EFX_NEED_BYTEORDER_TYPES
++ typedef __u16 __be16;
++ typedef __u32 __be32;
++ typedef __u64 __be64;
++ typedef __u16 __le16;
++ typedef __u32 __le32;
++ typedef __u64 __le64;
++#endif
+
-+ BUG_ON(evtchn_to_irq[local_port] != -1);
++/**************************************************************************
++ *
++ * Missing functions provided by kernel_compat.c
++ *
++ **************************************************************************
++ *
++ */
++#ifdef EFX_NEED_RANDOM_ETHER_ADDR
++ extern void efx_random_ether_addr(uint8_t *addr);
++ #ifndef EFX_IN_KCOMPAT_C
++ #undef random_ether_addr
++ #define random_ether_addr efx_random_ether_addr
++ #endif
++#endif
+
-+ if ((irq = find_unbound_irq()) < 0) {
-+ struct evtchn_close close = { .port = local_port };
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close))
-+ BUG();
-+ goto out;
-+ }
++#ifdef EFX_NEED_MII_ETHTOOL_FIX
++ extern int efx_mii_ethtool_gset(struct mii_if_info *mii,
++ struct ethtool_cmd *ecmd);
++ extern int efx_mii_ethtool_sset(struct mii_if_info *mii,
++ struct ethtool_cmd *ecmd);
++ #ifndef EFX_IN_KCOMPAT_C
++ #undef mii_ethtool_gset
++ #define mii_ethtool_gset efx_mii_ethtool_gset
++ #undef mii_ethtool_sset
++ #define mii_ethtool_sset efx_mii_ethtool_sset
++ #endif
++#endif
+
-+ evtchn_to_irq[local_port] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_LOCAL_PORT, 0, local_port);
-+ irq_bindcount[irq]++;
++#ifdef EFX_NEED_UNREGISTER_NETDEVICE_NOTIFIER_FIX
++ extern int efx_unregister_netdevice_notifier(struct notifier_block *nb);
++ #ifndef EFX_IN_KCOMPAT_C
++ #undef unregister_netdevice_notifier
++ #define unregister_netdevice_notifier \
++ efx_unregister_netdevice_notifier
++ #endif
++#endif
+
-+ out:
-+ spin_unlock(&irq_mapping_update_lock);
-+ return irq;
-+}
++#ifdef EFX_NEED_IOMMU_LOCK
++ extern dma_addr_t efx_pci_map_single(struct pci_dev *pci, void *ptr,
++ size_t size, int direction);
++ extern void efx_pci_unmap_single(struct pci_dev *pci,
++ dma_addr_t dma_addr, size_t size,
++ int direction);
++ extern void * efx_pci_alloc_consistent(struct pci_dev *pci,
++ size_t size,
++ dma_addr_t *dma_addr);
++ extern void efx_pci_free_consistent(struct pci_dev *pci,
++ size_t size, void *ptr,
++ dma_addr_t dma_addr);
++ #ifndef EFX_IN_KCOMPAT_C
++ #undef pci_map_single
++ #undef pci_unmap_single
++ #undef pci_alloc_consistent
++ #undef pci_free_consistent
++ #define pci_map_single efx_pci_map_single
++ #define pci_unmap_single efx_pci_unmap_single
++ #define pci_alloc_consistent efx_pci_alloc_consistent
++ #define pci_free_consistent efx_pci_free_consistent
++ #endif
++#endif
+
-+static int bind_listening_port_to_irq(unsigned int remote_domain)
-+{
-+ struct evtchn_alloc_unbound alloc_unbound;
-+ int err;
++#ifdef EFX_NEED_PRINT_MAC
++ extern char *print_mac(char *buf, const u8 *addr);
++#endif
+
-+ alloc_unbound.dom = DOMID_SELF;
-+ alloc_unbound.remote_dom = remote_domain;
++#ifdef EFX_NEED_COMPOUND_PAGE_FIX
++ extern void efx_compound_page_destructor(struct page *page);
++#endif
+
-+ err = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
-+ &alloc_unbound);
++#ifdef EFX_NEED_HEX_DUMP
++ extern void
++ print_hex_dump(const char *level, const char *prefix_str,
++ int prefix_type, int rowsize, int groupsize,
++ const void *buf, size_t len, int ascii);
++#endif
+
-+ return err ? : bind_local_port_to_irq(alloc_unbound.port);
-+}
++#ifdef EFX_NEED_MSECS_TO_JIFFIES
++ extern unsigned long msecs_to_jiffies(const unsigned int m);
++#endif
+
-+static int bind_interdomain_evtchn_to_irq(unsigned int remote_domain,
-+ unsigned int remote_port)
-+{
-+ struct evtchn_bind_interdomain bind_interdomain;
-+ int err;
++#ifdef EFX_NEED_MSLEEP
++ extern void msleep(unsigned int msecs);
++#endif
+
-+ bind_interdomain.remote_dom = remote_domain;
-+ bind_interdomain.remote_port = remote_port;
++/**************************************************************************
++ *
++ * Wrappers to fix bugs and parameter changes
++ *
++ **************************************************************************
++ *
++ */
+
-+ err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
-+ &bind_interdomain);
++#ifdef EFX_NEED_PCI_SAVE_RESTORE_WRAPPERS
++ #define pci_save_state(_dev) \
++ pci_save_state(_dev, (_dev)->saved_config_space)
+
-+ return err ? : bind_local_port_to_irq(bind_interdomain.local_port);
-+}
++ #define pci_restore_state(_dev) \
++ pci_restore_state(_dev, (_dev)->saved_config_space)
++#endif
+
-+static int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
-+{
-+ struct evtchn_bind_virq bind_virq;
-+ int evtchn, irq;
++#ifdef EFX_NEED_WORK_API_WRAPPERS
++ /**
++ * queue_delayed_work in pre 2.6.20 can't rearm from inside
++ * the work member. So instead do a rather hacky sleep
++ */
++ #define delayed_work work_struct
++ #define INIT_DELAYED_WORK INIT_WORK
+
-+ spin_lock(&irq_mapping_update_lock);
++ static int inline efx_queue_delayed_work(struct workqueue_struct *wq,
++ struct work_struct *work,
++ unsigned long delay)
++ {
++ if (unlikely(delay > 0))
++ schedule_timeout_uninterruptible(delay);
++ return queue_work(wq, work);
++ }
++ #define queue_delayed_work efx_queue_delayed_work
+
-+ if ((irq = per_cpu(virq_to_irq, cpu)[virq]) == -1) {
-+ if ((irq = find_unbound_irq()) < 0)
-+ goto out;
++ /**
++ * The old and new work-function prototypes just differ
++ * in the type of the pointer returned, so it's safe
++ * to cast between the prototypes.
++ */
++ typedef void (*efx_old_work_func_t)(void *p);
++
++ #undef INIT_WORK
++ #define INIT_WORK(_work, _func) \
++ do { \
++ INIT_LIST_HEAD(&(_work)->entry); \
++ (_work)->pending = 0; \
++ PREPARE_WORK((_work), \
++ (efx_old_work_func_t) (_func), \
++ (_work)); \
++ } while (0)
++#endif
++
++#ifdef EFX_HAVE_OLD_NAPI
++ #define napi_str napi_dev[0]
++
++ static inline void netif_napi_add(struct net_device *dev,
++ struct net_device *dummy,
++ int (*poll) (struct net_device *,
++ int *),
++ int weight)
++ {
++ dev->weight = weight;
++ dev->poll = poll;
++ }
+
-+ bind_virq.virq = virq;
-+ bind_virq.vcpu = cpu;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
-+ &bind_virq) != 0)
-+ BUG();
-+ evtchn = bind_virq.port;
++ #define napi_enable netif_poll_enable
++ #define napi_disable netif_poll_disable
+
-+ evtchn_to_irq[evtchn] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_VIRQ, virq, evtchn);
++ #define netif_rx_complete(dev, dummy) netif_rx_complete(dev)
++#endif
+
-+ per_cpu(virq_to_irq, cpu)[virq] = irq;
++#ifdef EFX_NEED_COMPOUND_PAGE_FIX
++ static inline
++ struct page *efx_alloc_pages(gfp_t flags, unsigned int order)
++ {
++ struct page *p = alloc_pages(flags, order);
++ if ((flags & __GFP_COMP) && (p != NULL) && (order > 0))
++ p[1].mapping = (void *)efx_compound_page_destructor;
++ return p;
++ }
++ #undef alloc_pages
++ #define alloc_pages efx_alloc_pages
+
-+ bind_evtchn_to_cpu(evtchn, cpu);
++ static inline
++ void efx_free_pages(struct page *p, unsigned int order)
++ {
++ if ((order > 0) && (page_count(p) == 1))
++ p[1].mapping = NULL;
++ __free_pages(p, order);
+ }
++ #define __free_pages efx_free_pages
++#endif
+
-+ irq_bindcount[irq]++;
++#ifdef EFX_NEED_HEX_DUMP_CONST_FIX
++ #define print_hex_dump(v,s,t,r,g,b,l,a) \
++ print_hex_dump((v),(s),(t),(r),(g),(void*)(b),(l),(a))
++#endif
+
-+ out:
-+ spin_unlock(&irq_mapping_update_lock);
-+ return irq;
-+}
++#endif /* EFX_KERNEL_COMPAT_H */
+--- linux-2.6.18.8/drivers/net/sfc/lm87_support.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/lm87_support.c 2008-05-19 00:33:29.309835579 +0300
+@@ -0,0 +1,295 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************/
+
-+static int bind_ipi_to_irq(unsigned int ipi, unsigned int cpu)
-+{
-+ struct evtchn_bind_ipi bind_ipi;
-+ int evtchn, irq;
++#include "net_driver.h"
++#include "lm87_support.h"
++#include "workarounds.h"
+
-+ spin_lock(&irq_mapping_update_lock);
++/* Setting this to 1 will cause efx_check_lm87 to dump the status when it
++ * detects an alarm. This will result in the canonical name (i.e. that in
++ * the LM87 data book) being printed for each set status bit, along with
++ * the reading for that sensor value, if applicable. If set to 0 only the
++ * raw status1 and status2 register values are printed. */
++#define LM87_VERBOSE_ALARMS 1
+
-+ if ((irq = per_cpu(ipi_to_irq, cpu)[ipi]) == -1) {
-+ if ((irq = find_unbound_irq()) < 0)
-+ goto out;
++/**************************************************************************
++ *
++ * Onboard LM87 temperature and voltage monitor
++ *
++ **************************************************************************
++ */
+
-+ bind_ipi.vcpu = cpu;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
-+ &bind_ipi) != 0)
-+ BUG();
-+ evtchn = bind_ipi.port;
++/* LM87 channel mode: all current boards either do not use AIN1/FAN1 and 2
++ * or use them as AIN. */
++#define LM87_CHANNEL_MODE 0x16
++#define LM87_CHANNEL_AIN1 1
++#define LM87_CHANNEL_AIN2 2
++#define LM87_CHANNEL_INIT (LM87_CHANNEL_AIN2 | LM87_CHANNEL_AIN1)
+
-+ evtchn_to_irq[evtchn] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_IPI, ipi, evtchn);
++/* LM87 configuration register 1 */
++#define LM87_CONFIG_1 0x40
++#define LM87_START 0x01
++#define LM87_INTEN 0x02
++#define LM87_INITIALIZATION 0x80
+
-+ per_cpu(ipi_to_irq, cpu)[ipi] = irq;
++/* LM87 interrupt status register 1 */
++#define LM87_INT_STATUS_1 0x41
+
-+ bind_evtchn_to_cpu(evtchn, cpu);
-+ }
++/* LM87 interrupt status register 2 */
++#define LM87_INT_STATUS_2 0x42
+
-+ irq_bindcount[irq]++;
++/* LM87 interrupt mask register 1 */
++#define LM87_INT_MASK_1 0x43
+
-+ out:
-+ spin_unlock(&irq_mapping_update_lock);
-+ return irq;
-+}
++/* LM87 interrupt mask register 2 */
++#define LM87_INT_MASK_2 0x44
+
-+static void unbind_from_irq(unsigned int irq)
-+{
-+ struct evtchn_close close;
-+ int cpu, evtchn = evtchn_from_irq(irq);
++/* LM87 monitoring limits */
++#define LM87_LIMITS 0x2b
+
-+ spin_lock(&irq_mapping_update_lock);
+
-+ if ((--irq_bindcount[irq] == 0) && VALID_EVTCHN(evtchn)) {
-+ close.port = evtchn;
-+ if ((type_from_irq(irq) != IRQT_CALLER_PORT) &&
-+ HYPERVISOR_event_channel_op(EVTCHNOP_close, &close))
-+ BUG();
++int efx_probe_lm87(struct efx_nic *efx, int addr,
++ const u8 *limits, int nlimits, const u16 irqmask)
++{
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ u8 byte;
++ int rc;
+
-+ switch (type_from_irq(irq)) {
-+ case IRQT_VIRQ:
-+ per_cpu(virq_to_irq, cpu_from_evtchn(evtchn))
-+ [index_from_irq(irq)] = -1;
-+ break;
-+ case IRQT_IPI:
-+ per_cpu(ipi_to_irq, cpu_from_evtchn(evtchn))
-+ [index_from_irq(irq)] = -1;
-+ break;
-+ default:
-+ break;
++ /* Check for onboard LM87 */
++ rc = efx_i2c_check_presence_retry(i2c, addr);
++ if (rc) {
++ /* Not an error to lack an LM87, but failure to probe the
++ * bus is worrying. */
++ if (rc == -EFAULT) {
++ EFX_ERR(efx, "Failed to probe I2C bus for LM87!\n");
++ return rc;
++ } else {
++ EFX_LOG(efx, "has no onboard LM87 chip\n");
++ return 0;
+ }
++ }
++ efx->board_info.lm87_addr = addr;
++ EFX_LOG(efx, "detected onboard LM87 chip at 0x%2x\n", addr);
+
-+ /* Closed ports are implicitly re-bound to VCPU0. */
-+ bind_evtchn_to_cpu(evtchn, 0);
++ /* Reset chip */
++ byte = LM87_INITIALIZATION;
++ rc = efx_i2c_write_retry(i2c, addr, LM87_CONFIG_1, &byte, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not reset LM87\n");
++ return rc;
++ }
+
-+ evtchn_to_irq[evtchn] = -1;
-+ irq_info[irq] = IRQ_UNBOUND;
++ /* Configure channel mode: currently hardwire to make pins 5 and 6
++ * AIN1 and AIN2 rather than FAN1, FAN2. */
++ byte = LM87_CHANNEL_INIT;
++ rc = efx_i2c_write_retry(i2c, addr, LM87_CHANNEL_MODE, &byte, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not program LM87 chan. mode\n");
++ return rc;
++ }
+
-+ /* Zap stats across IRQ changes of use. */
-+ for_each_possible_cpu(cpu)
-+ kstat_cpu(cpu).irqs[irq] = 0;
++ /* Configure limits */
++ rc = efx_i2c_write_retry(i2c, addr, LM87_LIMITS, limits, nlimits);
++ if (rc) {
++ EFX_ERR(efx, "could not program LM87 limits\n");
++ return rc;
+ }
+
-+ spin_unlock(&irq_mapping_update_lock);
-+}
++ /* Mask off unwanted interrupts */
++ byte = (irqmask & 0xff);
++ rc = efx_i2c_write_retry(i2c, addr, LM87_INT_MASK_1, &byte, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not mask LM87 interrupts\n");
++ return rc;
++ }
+
-+int bind_caller_port_to_irqhandler(
-+ unsigned int caller_port,
-+ irqreturn_t (*handler)(int, void *, struct pt_regs *),
-+ unsigned long irqflags,
-+ const char *devname,
-+ void *dev_id)
-+{
-+ int irq, retval;
++ byte = (irqmask >> 8);
++ rc = efx_i2c_write_retry(i2c, addr, LM87_INT_MASK_2, &byte, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not mask LM87 interrupts\n");
++ return rc;
++ }
+
-+ irq = bind_caller_port_to_irq(caller_port);
-+ if (irq < 0)
-+ return irq;
++ /* Start monitoring */
++ byte = LM87_START;
++ if (irqmask != EFX_LM87_NO_INTS)
++ byte |= LM87_INTEN;
+
-+ retval = request_irq(irq, handler, irqflags, devname, dev_id);
-+ if (retval != 0) {
-+ unbind_from_irq(irq);
-+ return retval;
++ rc = efx_i2c_write_retry(i2c, addr, LM87_CONFIG_1, &byte, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not start LM87\n");
++ return rc;
+ }
+
-+ return irq;
++ return rc;
+}
-+EXPORT_SYMBOL_GPL(bind_caller_port_to_irqhandler);
+
-+int bind_listening_port_to_irqhandler(
-+ unsigned int remote_domain,
-+ irqreturn_t (*handler)(int, void *, struct pt_regs *),
-+ unsigned long irqflags,
-+ const char *devname,
-+ void *dev_id)
++void efx_remove_lm87(struct efx_nic *efx)
+{
-+ int irq, retval;
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ u8 byte;
+
-+ irq = bind_listening_port_to_irq(remote_domain);
-+ if (irq < 0)
-+ return irq;
++ if (!efx->board_info.lm87_addr)
++ return;
+
-+ retval = request_irq(irq, handler, irqflags, devname, dev_id);
-+ if (retval != 0) {
-+ unbind_from_irq(irq);
-+ return retval;
++ /* Reset chip */
++ byte = LM87_INITIALIZATION;
++ if (efx_i2c_write_retry(i2c, efx->board_info.lm87_addr,
++ LM87_CONFIG_1, &byte, 1) != 0)
++ EFX_ERR(efx, "could not reset LM87 on exit\n");
++}
++
++#if LM87_VERBOSE_ALARMS
++/* Bit number to name mapping for status1 */
++static const char *lm_stat_names[] = {
++/* Status 1 contents */
++ "+2.5Vin",
++ "Vccp1",
++ "Vcc",
++ "+5Vin",
++ "Int. Temp.",
++ "Ext. Temp.",
++ "FAN1/AIN1",
++ "FAN2/AIN2",
++/* Status 2 contents */
++ "+12Vin",
++ "Vccp2",
++ "Reserved",
++ "Reserved",
++ "CI",
++ "THERM#",
++ "D1 Fault",
++ "D2 Fault"
++};
++
++/* Where to read the value corresponding to an alarm bit. */
++static const int lm_stat_regs[] = {
++ 0x20, 0x21, 0x22, 0x23, 0x27, 0x26, 0x28, 0x29,
++ 0x24, 0x25, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
++};
++
++/* The positions of the alarm bits do not correspond exactly to the
++ * order of the limit values. Convert so the user only needs to maintain
++ * one array */
++static int lm_bit_to_lim[] = {
++ 0, /* 2.5V */
++ 1, /* Vccp1 */
++ 2, /* Vcc */
++ 3, /* 5V */
++ 7, /* Int temp. */
++ 6, /* Ext temp. */
++ 8, /* AIN1 */
++ 9, /* AIN2 */
++ 4, /* 12V */
++ 5 /* Vccp2 */
++};
++
++/* These are bit numbers. I feel justified in hardwiring the max. */
++static const int lm_stat_max = 16;
++
++static void lm87_show_alarm(struct efx_nic *efx, int bit)
++{
++ char valbuf[8];
++ u8 val;
++
++ if (lm_stat_regs[bit] != 0xff) {
++ efx_i2c_read_retry(&efx->i2c, efx->board_info.lm87_addr,
++ lm_stat_regs[bit], &val, 1);
++ sprintf(valbuf, "0x%02x ", val);
++ } else {
++ strcpy(valbuf, "---- ");
+ }
-+
-+ return irq;
++ /* If the board code knows what this sensor is wired to, let it tell
++ * us, else just print the LM87 datasheet name of the input, and the
++ * value. */
++ if (efx->board_info.interpret_sensor == NULL ||
++ (bit < ARRAY_SIZE(lm_bit_to_lim) &&
++ efx->board_info.interpret_sensor(efx, lm_bit_to_lim[bit], val)
++ == 0))
++ EFX_ERR(efx, ": %10s %4s\n",
++ STRING_TABLE_LOOKUP(bit, lm_stat), valbuf);
+}
-+EXPORT_SYMBOL_GPL(bind_listening_port_to_irqhandler);
+
-+int bind_interdomain_evtchn_to_irqhandler(
-+ unsigned int remote_domain,
-+ unsigned int remote_port,
-+ irqreturn_t (*handler)(int, void *, struct pt_regs *),
-+ unsigned long irqflags,
-+ const char *devname,
-+ void *dev_id)
++static void lm87_dump_alarms(struct efx_nic *efx, int stat1, int stat2)
+{
-+ int irq, retval;
-+
-+ irq = bind_interdomain_evtchn_to_irq(remote_domain, remote_port);
-+ if (irq < 0)
-+ return irq;
-+
-+ retval = request_irq(irq, handler, irqflags, devname, dev_id);
-+ if (retval != 0) {
-+ unbind_from_irq(irq);
-+ return retval;
++ int i;
++ EFX_ERR(efx, " NAME value\n");
++ for (i = 0; i < 8; i++) {
++ if (stat1 & (1 << i))
++ lm87_show_alarm(efx, i);
++ if (stat2 & (1 << i))
++ lm87_show_alarm(efx, i + 8);
+ }
-+
-+ return irq;
+}
-+EXPORT_SYMBOL_GPL(bind_interdomain_evtchn_to_irqhandler);
-+
-+int bind_virq_to_irqhandler(
-+ unsigned int virq,
-+ unsigned int cpu,
-+ irqreturn_t (*handler)(int, void *, struct pt_regs *),
-+ unsigned long irqflags,
-+ const char *devname,
-+ void *dev_id)
-+{
-+ int irq, retval;
-+
-+ irq = bind_virq_to_irq(virq, cpu);
-+ if (irq < 0)
-+ return irq;
-+
-+ retval = request_irq(irq, handler, irqflags, devname, dev_id);
-+ if (retval != 0) {
-+ unbind_from_irq(irq);
-+ return retval;
-+ }
+
-+ return irq;
-+}
-+EXPORT_SYMBOL_GPL(bind_virq_to_irqhandler);
++#else
++#define lm87_dump_alarms(_name, _stat1, _stat2) do {} while (0)
++#endif
+
-+int bind_ipi_to_irqhandler(
-+ unsigned int ipi,
-+ unsigned int cpu,
-+ irqreturn_t (*handler)(int, void *, struct pt_regs *),
-+ unsigned long irqflags,
-+ const char *devname,
-+ void *dev_id)
++/* Read onboard LM87 (if present)
++ * Return error code if lm87 could not be read (-EIO)
++ * _or_ is raising an alarm (-ERANGE). 0 if AOK.
++ */
++int efx_check_lm87(struct efx_nic *efx, unsigned mask)
+{
-+ int irq, retval;
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ u8 int_status_1, int_status_2;
++ unsigned ints;
++ int rc = 0;
+
-+ irq = bind_ipi_to_irq(ipi, cpu);
-+ if (irq < 0)
-+ return irq;
++ /* If link is up then do not monitor temperature */
++ if (EFX_WORKAROUND_7884(efx) && efx->link_up)
++ return 0;
+
-+ retval = request_irq(irq, handler, irqflags, devname, dev_id);
-+ if (retval != 0) {
-+ unbind_from_irq(irq);
-+ return retval;
++ if (!efx->board_info.lm87_addr)
++ return 0;
++
++ /* Read interrupt status registers */
++ rc = efx_i2c_read_retry(i2c, efx->board_info.lm87_addr,
++ LM87_INT_STATUS_1, &int_status_1, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not read LM87 INT status 1\n");
++ return rc;
++ }
++ rc = efx_i2c_read_retry(i2c, efx->board_info.lm87_addr,
++ LM87_INT_STATUS_2, &int_status_2, 1);
++ if (rc) {
++ EFX_ERR(efx, "could not read LM87 INT status 2\n");
++ return rc;
+ }
+
-+ return irq;
-+}
-+EXPORT_SYMBOL_GPL(bind_ipi_to_irqhandler);
++ int_status_1 &= mask;
++ int_status_2 &= (mask >> 8);
++ ints = ((int_status_2 << 8) | int_status_1);
+
-+void unbind_from_irqhandler(unsigned int irq, void *dev_id)
-+{
-+ free_irq(irq, dev_id);
-+ unbind_from_irq(irq);
-+}
-+EXPORT_SYMBOL_GPL(unbind_from_irqhandler);
++ /* Check interrupt status */
++ if (ints == 0)
++ return 0;
+
-+#ifdef CONFIG_SMP
-+void rebind_evtchn_to_cpu(int port, unsigned int cpu)
-+{
-+ struct evtchn_bind_vcpu ebv = { .port = port, .vcpu = cpu };
-+ int masked;
++ EFX_ERR(efx, "LM87 detected a hardware failure (status %02x:%02x)\n",
++ int_status_1, int_status_2);
++ lm87_dump_alarms(efx, int_status_1, int_status_2);
+
-+ masked = test_and_set_evtchn_mask(port);
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_vcpu, &ebv) == 0)
-+ bind_evtchn_to_cpu(port, cpu);
-+ if (!masked)
-+ unmask_evtchn(port);
++ return -ERANGE;
+}
+--- linux-2.6.18.8/drivers/net/sfc/lm87_support.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/lm87_support.h 2008-05-19 00:33:29.309835579 +0300
+@@ -0,0 +1,58 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************/
++
++#ifndef EFX_LM87_SUPPORT_H
++#define EFX_LM87_SUPPORT_H
++
++/* The interrupt bit masks. These are the same in the interrupt status and
++ * interrupt mask registers. */
++/* Register 1 bits */
++#define EFX_LM87_2_5V_INT (1)
++#define EFX_LM87_VCCP1_INT (2)
++#define EFX_LM87_VCC_INT (4)
++#define EFX_LM87_5_V_INT (8)
++#define EFX_LM87_ITMP_INT (0x10)
++#define EFX_LM87_ETMP_INT (0x20)
++#define EFX_LM87_FAN1_INT (0x40)
++#define EFX_LM87_FAN2_INT (0x80)
++/* Register 2 bits */
++#define EFX_LM87_12V_INT (0x100)
++#define EFX_LM87_VCCP2_INT (0x200)
++/* Bits 2 and 3 are reserved. */
++#define EFX_LM87_CI_INT (0x1000)
++#define EFX_LM87_THERM_INT (0x2000)
++#define EFX_LM87_D1_INT (0x4000)
++#define EFX_LM87_D2_INT (0x8000)
++
++#define EFX_LM87_NO_INTS ((u16)-1)
+
-+static void rebind_irq_to_cpu(unsigned int irq, unsigned int tcpu)
-+{
-+ int evtchn = evtchn_from_irq(irq);
++extern
++int efx_probe_lm87(struct efx_nic *efx, int addr, const u8 *limits,
++ int nlimits, const u16 irqmask);
+
-+ if (VALID_EVTCHN(evtchn))
-+ rebind_evtchn_to_cpu(evtchn, tcpu);
-+}
++extern void efx_remove_lm87(struct efx_nic *efx);
+
-+static void set_affinity_irq(unsigned int irq, cpumask_t dest)
-+{
-+ unsigned tcpu = first_cpu(dest);
-+ rebind_irq_to_cpu(irq, tcpu);
-+}
-+#endif
++extern int efx_check_lm87(struct efx_nic *efx, unsigned mask);
+
-+int resend_irq_on_evtchn(unsigned int irq)
-+{
-+ int masked, evtchn = evtchn_from_irq(irq);
-+ shared_info_t *s = HYPERVISOR_shared_info;
++#endif /* EFX_LM87_SUPPORT_H */
+--- linux-2.6.18.8/drivers/net/sfc/mac.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/mac.h 2008-05-19 00:33:29.309835579 +0300
+@@ -0,0 +1,38 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (!VALID_EVTCHN(evtchn))
-+ return 1;
++#ifndef EFX_MAC_H
++#define EFX_MAC_H
+
-+ masked = test_and_set_evtchn_mask(evtchn);
-+ synch_set_bit(evtchn, s->evtchn_pending);
-+ if (!masked)
-+ unmask_evtchn(evtchn);
++#include "net_driver.h"
+
-+ return 1;
-+}
++extern void mentormac_reset(struct efx_nic *efx);
++extern void mentormac_reconfigure(struct efx_nic *efx);
++extern struct efx_mac_operations falcon_gmac_operations;
++extern struct efx_mac_operations falcon_xmac_operations;
+
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/mdio_10g.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/mdio_10g.c 2008-05-19 00:33:29.313835810 +0300
+@@ -0,0 +1,441 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+/*
-+ * Interface to generic handling in irq.c
++ * Useful functions for working with MDIO clause 45 PHYs
+ */
++#include <linux/types.h>
++#include <linux/ethtool.h>
++#include <linux/delay.h>
++#include "net_driver.h"
++#include "mdio_10g.h"
++#include "boards.h"
+
-+static unsigned int startup_dynirq(unsigned int irq)
-+{
-+ int evtchn = evtchn_from_irq(irq);
-+
-+ if (VALID_EVTCHN(evtchn))
-+ unmask_evtchn(evtchn);
-+ return 0;
-+}
++static const char *mmd_block_names[] = {
++ [0] = "(illegal)",
++ [MDIO_MMD_PMAPMD] = "PMA/PMD",
++ [MDIO_MMD_WIS] = "WIS",
++ [MDIO_MMD_PCS] = "PCS",
++ [MDIO_MMD_PHYXS] = "PHY XS",
++ [MDIO_MMD_DTEXS] = "DTE XS",
++ [MDIO_MMD_TC] = "TC",
++ [MDIO_MMD_AN] = "AN",
++};
++static const int mmd_block_max = ARRAY_SIZE(mmd_block_names);
+
-+static void shutdown_dynirq(unsigned int irq)
++const char *mdio_clause45_mmd_name(int mmd)
+{
-+ int evtchn = evtchn_from_irq(irq);
-+
-+ if (VALID_EVTCHN(evtchn))
-+ mask_evtchn(evtchn);
++ return STRING_TABLE_LOOKUP(mmd, mmd_block);
+}
+
-+static void enable_dynirq(unsigned int irq)
++int mdio_clause45_reset_mmd(struct efx_nic *port, int mmd,
++ int spins, int spintime)
+{
-+ int evtchn = evtchn_from_irq(irq);
++ u32 ctrl;
++ int phy_id = port->mii.phy_id;
+
-+ if (VALID_EVTCHN(evtchn))
-+ unmask_evtchn(evtchn);
-+}
++ /* Catch callers passing values in the wrong units (or just silly) */
++ EFX_BUG_ON_PARANOID(spins * spintime >= 5000);
+
-+static void disable_dynirq(unsigned int irq)
-+{
-+ int evtchn = evtchn_from_irq(irq);
++ mdio_clause45_write(port, phy_id, mmd, MDIO_MMDREG_CTRL1,
++ (1 << MDIO_MMDREG_CTRL1_RESET_LBN));
++ /* Wait for the reset bit to clear. */
++ do {
++ msleep(spintime);
++ ctrl = mdio_clause45_read(port, phy_id, mmd, MDIO_MMDREG_CTRL1);
++ spins--;
+
-+ if (VALID_EVTCHN(evtchn))
-+ mask_evtchn(evtchn);
++ } while (spins && (ctrl & (1 << MDIO_MMDREG_CTRL1_RESET_LBN)));
++
++ return spins ? spins : -ETIMEDOUT;
+}
+
-+static void ack_dynirq(unsigned int irq)
++static int mdio_clause45_check_mmd(struct efx_nic *efx, int mmd,
++ int fault_fatal)
+{
-+ int evtchn = evtchn_from_irq(irq);
++ int status;
++ int phy_id = efx->mii.phy_id;
++ const char *mmdname = STRING_TABLE_LOOKUP(mmd, mmd_block);
+
-+ move_native_irq(irq);
++ if (LOOPBACK_INTERNAL(efx))
++ return 0;
+
-+ if (VALID_EVTCHN(evtchn)) {
-+ mask_evtchn(evtchn);
-+ clear_evtchn(evtchn);
++ /* Read MMD STATUS2 to check it is responding. */
++ status = mdio_clause45_read(efx, phy_id, mmd, MDIO_MMDREG_STAT2);
++ if (((status >> MDIO_MMDREG_STAT2_PRESENT_LBN) &
++ ((1 << MDIO_MMDREG_STAT2_PRESENT_WIDTH) - 1)) !=
++ MDIO_MMDREG_STAT2_PRESENT_VAL) {
++ EFX_ERR(efx, "PHY MMD %s not responding.\n", mmdname);
++ return -EIO;
+ }
-+}
-+
-+static void end_dynirq(unsigned int irq)
-+{
-+ int evtchn = evtchn_from_irq(irq);
+
-+ if (VALID_EVTCHN(evtchn) && !(irq_desc[irq].status & IRQ_DISABLED))
-+ unmask_evtchn(evtchn);
++ /* Read MMD STATUS 1 to check for fault. */
++ status = mdio_clause45_read(efx, phy_id, mmd, MDIO_MMDREG_STAT1);
++ if ((status & (1 << MDIO_MMDREG_STAT1_FAULT_LBN)) != 0) {
++ if (fault_fatal) {
++ EFX_ERR(efx, "PHY MMD %s reporting fatal"
++ " fault: status %x\n", mmdname, status);
++ return -EIO;
++ } else {
++ EFX_LOG(efx, "PHY MMD %s reporting status"
++ " %x (expected)\n", mmdname, status);
++ }
++ }
++ return 0;
+}
+
-+static struct hw_interrupt_type dynirq_type = {
-+ .typename = "Dynamic-irq",
-+ .startup = startup_dynirq,
-+ .shutdown = shutdown_dynirq,
-+ .enable = enable_dynirq,
-+ .disable = disable_dynirq,
-+ .ack = ack_dynirq,
-+ .end = end_dynirq,
-+#ifdef CONFIG_SMP
-+ .set_affinity = set_affinity_irq,
-+#endif
-+ .retrigger = resend_irq_on_evtchn,
-+};
-+
-+static inline void pirq_unmask_notify(int pirq)
-+{
-+ struct physdev_eoi eoi = { .irq = pirq };
-+ if (unlikely(test_bit(pirq, pirq_needs_eoi)))
-+ (void)HYPERVISOR_physdev_op(PHYSDEVOP_eoi, &eoi);
-+}
++/* This ought to be ridiculous overkill. We expect it to fail rarely */
++#define MDIO45_RESET_TIME HZ
++#define MDIO45_RESET_ITERS (100)
+
-+static inline void pirq_query_unmask(int pirq)
++int mdio_clause45_wait_reset_mmds(struct efx_nic *efx,
++ unsigned int mmd_mask)
+{
-+ struct physdev_irq_status_query irq_status;
-+ irq_status.irq = pirq;
-+ (void)HYPERVISOR_physdev_op(PHYSDEVOP_irq_status_query, &irq_status);
-+ clear_bit(pirq, pirq_needs_eoi);
-+ if (irq_status.flags & XENIRQSTAT_needs_eoi)
-+ set_bit(pirq, pirq_needs_eoi);
++ const int spintime = MDIO45_RESET_TIME / MDIO45_RESET_ITERS;
++ int tries = MDIO45_RESET_ITERS;
++ int rc = 0;
++ int in_reset;
++
++ while (tries) {
++ int mask = mmd_mask;
++ int mmd = 0;
++ int stat;
++ in_reset = 0;
++ while (mask) {
++ if (mask & 1) {
++ stat = mdio_clause45_read(efx,
++ efx->mii.phy_id,
++ mmd,
++ MDIO_MMDREG_CTRL1);
++ if (stat < 0) {
++ EFX_ERR(efx, "failed to read"
++ " status of MMD %s\n",
++ STRING_TABLE_LOOKUP(mmd,
++ mmd_block));
++ return -EIO;
++ }
++ if (stat & (1 << MDIO_MMDREG_CTRL1_RESET_LBN))
++ in_reset |= (1 << mmd);
++ }
++ mask = mask >> 1;
++ mmd++;
++ }
++ if (!in_reset)
++ break;
++ tries--;
++ msleep(spintime);
++ }
++ if (in_reset != 0) {
++ EFX_ERR(efx, "not all MMDs came out of reset in time."
++ " MMDs still in reset: %x\n", in_reset);
++ rc = -ETIMEDOUT;
++ }
++ return rc;
+}
+
-+/*
-+ * On startup, if there is no action associated with the IRQ then we are
-+ * probing. In this case we should not share with others as it will confuse us.
-+ */
-+#define probing_irq(_irq) (irq_desc[(_irq)].action == NULL)
-+
-+static unsigned int startup_pirq(unsigned int irq)
++int mdio_clause45_check_mmds(struct efx_nic *efx,
++ unsigned int mmd_mask, unsigned int fatal_mask)
+{
-+ struct evtchn_bind_pirq bind_pirq;
-+ int evtchn = evtchn_from_irq(irq);
++ int devices, mmd = 0;
++ int probe_mmd;
+
-+ if (VALID_EVTCHN(evtchn))
-+ goto out;
++ /* Historically we have probed the PHYXS to find out what devices are
++ * present,but that doesn't work so well if the PHYXS isn't expected
++ * to exist, if so just find the first item in the list supplied. */
++ probe_mmd = (mmd_mask & MDIO_MMDREG_DEVS0_PHYXS) ? MDIO_MMD_PHYXS :
++ __ffs(mmd_mask);
++ devices = mdio_clause45_read(efx, efx->mii.phy_id,
++ probe_mmd, MDIO_MMDREG_DEVS0);
+
-+ bind_pirq.pirq = irq;
-+ /* NB. We are happy to share unless we are probing. */
-+ bind_pirq.flags = probing_irq(irq) ? 0 : BIND_PIRQ__WILL_SHARE;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_pirq, &bind_pirq) != 0) {
-+ if (!probing_irq(irq))
-+ printk(KERN_INFO "Failed to obtain physical IRQ %d\n",
-+ irq);
-+ return 0;
++ /* Check all the expected MMDs are present */
++ if (devices < 0) {
++ EFX_ERR(efx, "failed to read devices present\n");
++ return -EIO;
+ }
-+ evtchn = bind_pirq.port;
++ if ((devices & mmd_mask) != mmd_mask) {
++ EFX_ERR(efx, "required MMDs not present: got %x, "
++ "wanted %x\n", devices, mmd_mask);
++ return -ENODEV;
++ }
++ EFX_TRACE(efx, "Devices present: %x\n", devices);
+
-+ pirq_query_unmask(irq_to_pirq(irq));
++ /* Check all required MMDs are responding and happy. */
++ while (mmd_mask) {
++ if (mmd_mask & 1) {
++ int fault_fatal = fatal_mask & 1;
++ if (mdio_clause45_check_mmd(efx, mmd, fault_fatal))
++ return -EIO;
++ }
++ mmd_mask = mmd_mask >> 1;
++ fatal_mask = fatal_mask >> 1;
++ mmd++;
++ }
+
-+ evtchn_to_irq[evtchn] = irq;
-+ bind_evtchn_to_cpu(evtchn, 0);
-+ irq_info[irq] = mk_irq_info(IRQT_PIRQ, irq, evtchn);
++ return 0;
++}
+
-+ out:
-+ unmask_evtchn(evtchn);
-+ pirq_unmask_notify(irq_to_pirq(irq));
++int mdio_clause45_links_ok(struct efx_nic *efx, unsigned int mmd_mask)
++{
++ int phy_id = efx->mii.phy_id;
++ int status;
++ int ok = 1;
++ int mmd = 0;
++ int good;
+
-+ return 0;
++ /* If the port is in loopback, then we should only consider a subset
++ * of mmd's */
++ if (LOOPBACK_INTERNAL(efx))
++ return 1;
++ else if (efx->loopback_mode == LOOPBACK_NETWORK)
++ return 0;
++ else if (!efx->phy_powered)
++ return 0;
++ else if (efx->loopback_mode == LOOPBACK_PHYXS)
++ mmd_mask &= ~(MDIO_MMDREG_DEVS0_PHYXS |
++ MDIO_MMDREG_DEVS0_PCS |
++ MDIO_MMDREG_DEVS0_PMAPMD);
++ else if (efx->loopback_mode == LOOPBACK_PCS)
++ mmd_mask &= ~(MDIO_MMDREG_DEVS0_PCS |
++ MDIO_MMDREG_DEVS0_PMAPMD);
++ else if (efx->loopback_mode == LOOPBACK_PMAPMD)
++ mmd_mask &= ~MDIO_MMDREG_DEVS0_PMAPMD;
++
++ while (mmd_mask) {
++ if (mmd_mask & 1) {
++ /* Double reads because link state is latched, and a
++ * read moves the current state into the register */
++ status = mdio_clause45_read(efx, phy_id,
++ mmd, MDIO_MMDREG_STAT1);
++ status = mdio_clause45_read(efx, phy_id,
++ mmd, MDIO_MMDREG_STAT1);
++
++ good = status & (1 << MDIO_MMDREG_STAT1_LINK_LBN);
++ ok = ok && good;
++ }
++ mmd_mask = (mmd_mask >> 1);
++ mmd++;
++ }
++ return ok;
++}
++
++void mdio_clause45_transmit_disable(struct efx_nic *efx, int disable)
++{
++ int phy_id = efx->mii.phy_id;
++ int ctrl1, ctrl2;
++
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_MMDREG_TXDIS);
++ if (disable)
++ ctrl2 |= (1 << MDIO_MMDREG_TXDIS_GLOBAL_LBN);
++ else
++ ctrl1 &= ~(1 << MDIO_MMDREG_TXDIS_GLOBAL_LBN);
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_MMDREG_TXDIS, ctrl2);
+}
+
-+static void shutdown_pirq(unsigned int irq)
++void mdio_clause45_phy_reconfigure(struct efx_nic *efx)
+{
-+ struct evtchn_close close;
-+ int evtchn = evtchn_from_irq(irq);
++ int phy_id = efx->mii.phy_id;
++ int ctrl1, ctrl2;
+
-+ if (!VALID_EVTCHN(evtchn))
-+ return;
++ /* Handle (with debouncing) PMA/PMD loopback */
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_MMDREG_CTRL1);
+
-+ mask_evtchn(evtchn);
++ if (efx->loopback_mode == LOOPBACK_PMAPMD)
++ ctrl2 |= (1 << MDIO_PMAPMD_CTRL1_LBACK_LBN);
++ else
++ ctrl2 &= ~(1 << MDIO_PMAPMD_CTRL1_LBACK_LBN);
+
-+ close.port = evtchn;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
-+ BUG();
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_MMDREG_CTRL1, ctrl2);
+
-+ bind_evtchn_to_cpu(evtchn, 0);
-+ evtchn_to_irq[evtchn] = -1;
-+ irq_info[irq] = IRQ_UNBOUND;
++ /* Handle (with debouncing) PCS loopback */
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy_id, MDIO_MMD_PCS,
++ MDIO_MMDREG_CTRL1);
++ if (efx->loopback_mode == LOOPBACK_PCS)
++ ctrl2 |= (1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++ else
++ ctrl2 &= ~(1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PCS,
++ MDIO_MMDREG_CTRL1, ctrl2);
++
++ /* Handle (with debouncing) PHYXS network loopback */
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy_id, MDIO_MMD_PHYXS,
++ MDIO_MMDREG_CTRL1);
++ if (efx->loopback_mode == LOOPBACK_NETWORK)
++ ctrl2 |= (1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++ else
++ ctrl2 &= ~(1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PHYXS,
++ MDIO_MMDREG_CTRL1, ctrl2);
+}
+
-+static void enable_pirq(unsigned int irq)
++static void mdio_clause45_set_mmd_lpower(struct efx_nic *efx,
++ int lpower, int mmd)
+{
-+ startup_pirq(irq);
++ const char *mmdname = STRING_TABLE_LOOKUP(mmd, mmd_block);
++ int phy = efx->mii.phy_id;
++ int stat = mdio_clause45_read(efx, phy, mmd, MDIO_MMDREG_STAT1);
++ int ctrl1, ctrl2;
++
++ EFX_TRACE(efx, "Setting low power mode for MMD %s to %d\n",
++ mmdname, lpower);
++
++ if (stat & (1 << MDIO_MMDREG_STAT1_LPABLE_LBN)) {
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy,
++ mmd, MDIO_MMDREG_CTRL1);
++ if (lpower)
++ ctrl2 |= (1 << MDIO_MMDREG_CTRL1_LPOWER_LBN);
++ else
++ ctrl2 &= ~(1 << MDIO_MMDREG_CTRL1_LPOWER_LBN);
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy, mmd,
++ MDIO_MMDREG_CTRL1, ctrl2);
++ } else {
++ /* If we ever want a completely generic PHY driver
++ * that which just does clause 45, we may consider not
++ * complaining, but for now expect the driver to know
++ * which MMDs to apply this to. */
++ EFX_ERR(efx, "Attempt change power setting of MMD %s"
++ " which doesn't support it.\n", mmdname);
++ }
+}
+
-+static void disable_pirq(unsigned int irq)
++void mdio_clause45_set_mmds_lpower(struct efx_nic *efx,
++ int low_power, unsigned int mmd_mask)
+{
++ int mmd = 0;
++ while (mmd_mask) {
++ if (mmd_mask & 1)
++ mdio_clause45_set_mmd_lpower(efx, low_power, mmd);
++ mmd_mask = (mmd_mask >> 1);
++ mmd++;
++ }
+}
+
-+static void ack_pirq(unsigned int irq)
++/**
++ * mdio_clause45_get_settings - Read (some of) the PHY settings over MDIO.
++ * @efx: Efx NIC
++ * @ecmd: Buffer for settings
++ *
++ * On return the 'port', 'speed', 'supported' and 'advertising' fields of
++ * ecmd have been filled out based on the PMA type.
++ */
++void mdio_clause45_get_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
+{
-+ int evtchn = evtchn_from_irq(irq);
-+
-+ move_native_irq(irq);
++ int pma_type;
+
-+ if (VALID_EVTCHN(evtchn)) {
-+ mask_evtchn(evtchn);
-+ clear_evtchn(evtchn);
++ /* If no PMA is present we are presumably talking something XAUI-ish
++ * like CX4. Which we report as FIBRE (see below) */
++ if ((efx->phy_op->mmds & DEV_PRESENT_BIT(MDIO_MMD_PMAPMD)) == 0) {
++ ecmd->speed = SPEED_10000;
++ ecmd->port = PORT_FIBRE;
++ ecmd->supported = SUPPORTED_FIBRE;
++ ecmd->advertising = ADVERTISED_FIBRE;
++ return;
+ }
-+}
+
-+static void end_pirq(unsigned int irq)
-+{
-+ int evtchn = evtchn_from_irq(irq);
++ pma_type = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, MDIO_MMDREG_CTRL2);
++ pma_type &= MDIO_PMAPMD_CTRL2_TYPE_MASK;
+
-+ if ((irq_desc[irq].status & (IRQ_DISABLED|IRQ_PENDING)) ==
-+ (IRQ_DISABLED|IRQ_PENDING)) {
-+ shutdown_pirq(irq);
-+ } else if (VALID_EVTCHN(evtchn)) {
-+ unmask_evtchn(evtchn);
-+ pirq_unmask_notify(irq_to_pirq(irq));
++ switch (pma_type) {
++ /* We represent CX4 as fibre in the absence of anything
++ better. */
++ case MDIO_PMAPMD_CTRL2_10G_CX4:
++ ecmd->speed = SPEED_10000;
++ ecmd->port = PORT_FIBRE;
++ ecmd->supported = SUPPORTED_FIBRE;
++ ecmd->advertising = ADVERTISED_FIBRE;
++ break;
++ /* 10G Base-T */
++ case MDIO_PMAPMD_CTRL2_10G_BT:
++ ecmd->speed = SPEED_10000;
++ ecmd->port = PORT_TP;
++ ecmd->supported = SUPPORTED_TP | SUPPORTED_10000baseT_Full;
++ ecmd->advertising = (ADVERTISED_FIBRE
++ | ADVERTISED_10000baseT_Full);
++ break;
++ case MDIO_PMAPMD_CTRL2_1G_BT:
++ ecmd->speed = SPEED_1000;
++ ecmd->port = PORT_TP;
++ ecmd->supported = SUPPORTED_TP | SUPPORTED_1000baseT_Full;
++ ecmd->advertising = (ADVERTISED_FIBRE
++ | ADVERTISED_1000baseT_Full);
++ break;
++ case MDIO_PMAPMD_CTRL2_100_BT:
++ ecmd->speed = SPEED_100;
++ ecmd->port = PORT_TP;
++ ecmd->supported = SUPPORTED_TP | SUPPORTED_100baseT_Full;
++ ecmd->advertising = (ADVERTISED_FIBRE
++ | ADVERTISED_100baseT_Full);
++ break;
++ case MDIO_PMAPMD_CTRL2_10_BT:
++ ecmd->speed = SPEED_10;
++ ecmd->port = PORT_TP;
++ ecmd->supported = SUPPORTED_TP | SUPPORTED_10baseT_Full;
++ ecmd->advertising = ADVERTISED_FIBRE | ADVERTISED_10baseT_Full;
++ break;
++ /* All the other defined modes are flavours of
++ * 10G optical */
++ default:
++ ecmd->speed = SPEED_10000;
++ ecmd->port = PORT_FIBRE;
++ ecmd->supported = SUPPORTED_FIBRE;
++ ecmd->advertising = ADVERTISED_FIBRE;
++ break;
+ }
+}
+
-+static struct hw_interrupt_type pirq_type = {
-+ .typename = "Phys-irq",
-+ .startup = startup_pirq,
-+ .shutdown = shutdown_pirq,
-+ .enable = enable_pirq,
-+ .disable = disable_pirq,
-+ .ack = ack_pirq,
-+ .end = end_pirq,
-+#ifdef CONFIG_SMP
-+ .set_affinity = set_affinity_irq,
-+#endif
-+ .retrigger = resend_irq_on_evtchn,
-+};
++/**
++ * mdio_clause45_set_settings - Set (some of) the PHY settings over MDIO.
++ * @efx: Efx NIC
++ * @ecmd: New settings
++ *
++ * Currently this just enforces that we are _not_ changing the
++ * 'port', 'speed', 'supported' or 'advertising' settings as these
++ * cannot be changed on any currently supported PHY.
++ */
++int mdio_clause45_set_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd)
++{
++ struct ethtool_cmd tmpcmd;
++ mdio_clause45_get_settings(efx, &tmpcmd);
++ /* None of the current PHYs support more than one mode
++ * of operation (and only 10GBT ever will), so keep things
++ * simple for now */
++ if ((ecmd->speed == tmpcmd.speed) && (ecmd->port == tmpcmd.port) &&
++ (ecmd->supported == tmpcmd.supported) &&
++ (ecmd->advertising == tmpcmd.advertising))
++ return 0;
++ return -EOPNOTSUPP;
++}
+--- linux-2.6.18.8/drivers/net/sfc/mdio_10g.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/mdio_10g.h 2008-05-19 00:33:29.313835810 +0300
+@@ -0,0 +1,295 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_MDIO_10G_H
++#define EFX_MDIO_10G_H
++
++/*
++ * Definitions needed for doing 10G MDIO as specified in clause 45
++ * MDIO, which do not appear in Linux yet. Also some helper functions.
++ */
++
++#include "efx.h"
++#include "boards.h"
++
++/* Numbering of the MDIO Manageable Devices (MMDs) */
++/* Physical Medium Attachment/ Physical Medium Dependent sublayer */
++#define MDIO_MMD_PMAPMD (1)
++/* WAN Interface Sublayer */
++#define MDIO_MMD_WIS (2)
++/* Physical Coding Sublayer */
++#define MDIO_MMD_PCS (3)
++/* PHY Extender Sublayer */
++#define MDIO_MMD_PHYXS (4)
++/* Extender Sublayer */
++#define MDIO_MMD_DTEXS (5)
++/* Transmission convergence */
++#define MDIO_MMD_TC (6)
++/* Auto negotiation */
++#define MDIO_MMD_AN (7)
++
++/* Generic register locations */
++#define MDIO_MMDREG_CTRL1 (0)
++#define MDIO_MMDREG_STAT1 (1)
++#define MDIO_MMDREG_IDHI (2)
++#define MDIO_MMDREG_IDLOW (3)
++#define MDIO_MMDREG_SPEED (4)
++#define MDIO_MMDREG_DEVS0 (5)
++#define MDIO_MMDREG_DEVS1 (6)
++#define MDIO_MMDREG_CTRL2 (7)
++#define MDIO_MMDREG_STAT2 (8)
++#define MDIO_MMDREG_TXDIS (9)
++
++/* Bits in MMDREG_CTRL1 */
++/* Reset */
++#define MDIO_MMDREG_CTRL1_RESET_LBN (15)
++#define MDIO_MMDREG_CTRL1_RESET_WIDTH (1)
++/* Loopback */
++/* Note that while WIS, PCS, PHYXS and DTEXS have the loopback bit here,
++ * the PMA/PMD does not. */
++#define MDIO_MMDREG_CTRL1_LBACK_LBN (14)
++#define MDIO_MMDREG_CTRL1_LBACK_WIDTH (1)
++/* Low power */
++#define MDIO_MMDREG_CTRL1_LPOWER_LBN (11)
++#define MDIO_MMDREG_CTRL1_LPOWER_WIDTH (1)
++
++/* Bits in MMDREG_STAT1 */
++#define MDIO_MMDREG_STAT1_FAULT_LBN (7)
++#define MDIO_MMDREG_STAT1_FAULT_WIDTH (1)
++/* Link state */
++#define MDIO_MMDREG_STAT1_LINK_LBN (2)
++#define MDIO_MMDREG_STAT1_LINK_WIDTH (1)
++/* Low power ability */
++#define MDIO_MMDREG_STAT1_LPABLE_LBN (1)
++#define MDIO_MMDREG_STAT1_LPABLE_WIDTH (1)
++
++/* Bits in ID reg */
++#define MDIO_ID_REV(_id32) (_id32 & 0xf)
++#define MDIO_ID_MODEL(_id32) ((_id32 >> 4) & 0x3f)
++#define MDIO_ID_OUI(_id32) (_id32 >> 10)
++
++/* Bits in MMDREG_DEVS0. Someone thoughtfully layed things out
++ * so the 'bit present' bit number of an MMD is the number of
++ * that MMD */
++#define DEV_PRESENT_BIT(_b) (1 << _b)
++
++#define MDIO_MMDREG_DEVS0_DTEXS DEV_PRESENT_BIT(MDIO_MMD_DTEXS)
++#define MDIO_MMDREG_DEVS0_PHYXS DEV_PRESENT_BIT(MDIO_MMD_PHYXS)
++#define MDIO_MMDREG_DEVS0_PCS DEV_PRESENT_BIT(MDIO_MMD_PCS)
++#define MDIO_MMDREG_DEVS0_WIS DEV_PRESENT_BIT(MDIO_MMD_WIS)
++#define MDIO_MMDREG_DEVS0_PMAPMD DEV_PRESENT_BIT(MDIO_MMD_PMAPMD)
++#define MDIO_MMDREG_DEVS0_AN DEV_PRESENT_BIT(MDIO_MMD_AN)
++
++
++/* Bits in MMDREG_STAT2 */
++#define MDIO_MMDREG_STAT2_PRESENT_VAL (2)
++#define MDIO_MMDREG_STAT2_PRESENT_LBN (14)
++#define MDIO_MMDREG_STAT2_PRESENT_WIDTH (2)
++
++/* Bits in MMDREG_TXDIS */
++#define MDIO_MMDREG_TXDIS_GLOBAL_LBN (0)
++#define MDIO_MMDREG_TXDIS_GLOBAL_WIDTH (1)
++
++/* MMD-specific bits, ordered by MMD, then register */
++#define MDIO_PMAPMD_CTRL1_LBACK_LBN (0)
++#define MDIO_PMAPMD_CTRL1_LBACK_WIDTH (1)
++
++/* PMA type (4 bits) */
++#define MDIO_PMAPMD_CTRL2_10G_CX4 (0x0)
++#define MDIO_PMAPMD_CTRL2_10G_EW (0x1)
++#define MDIO_PMAPMD_CTRL2_10G_LW (0x2)
++#define MDIO_PMAPMD_CTRL2_10G_SW (0x3)
++#define MDIO_PMAPMD_CTRL2_10G_LX4 (0x4)
++#define MDIO_PMAPMD_CTRL2_10G_ER (0x5)
++#define MDIO_PMAPMD_CTRL2_10G_LR (0x6)
++#define MDIO_PMAPMD_CTRL2_10G_SR (0x7)
++/* Reserved */
++#define MDIO_PMAPMD_CTRL2_10G_BT (0x9)
++/* Reserved */
++/* Reserved */
++#define MDIO_PMAPMD_CTRL2_1G_BT (0xc)
++/* Reserved */
++#define MDIO_PMAPMD_CTRL2_100_BT (0xe)
++#define MDIO_PMAPMD_CTRL2_10_BT (0xf)
++#define MDIO_PMAPMD_CTRL2_TYPE_MASK (0xf)
++
++/* PCS 10GBT registers */
++#define MDIO_PCS_10GBT_STATUS (32)
++#define MDIO_PCS_10GBT_STATUS2 (33)
++#define MDIO_PCS_10GBT_STATUS2_BER_LBN (8)
++#define MDIO_PCS_10GBT_STATUS2_BER_WIDTH (6)
++#define MDIO_PCS_10GBT_STATUS2_ERR_LBN (0)
++#define MDIO_PCS_10GBT_STATUS2_ERR_WIDTH (8)
++
++/* PHY XGXS lane state */
++#define MDIO_PHYXS_LANE_STATE (0x18)
++#define MDIO_PHYXS_LANE_ALIGNED_LBN (12)
++#define MDIO_PHYXS_LANE_SYNC0_LBN (0)
++#define MDIO_PHYXS_LANE_SYNC1_LBN (1)
++#define MDIO_PHYXS_LANE_SYNC2_LBN (2)
++#define MDIO_PHYXS_LANE_SYNC3_LBN (3)
++
++/* AN registers */
++#define MDIO_AN_STATUS (1)
++#define MDIO_AN_STATUS_XNP_LBN (7)
++#define MDIO_AN_STATUS_PAGE_LBN (6)
++#define MDIO_AN_STATUS_AN_DONE_LBN (5)
++#define MDIO_AN_STATUS_LP_AN_CAP_LBN (0)
++
++#define MDIO_AN_10GBT_STATUS (33)
++#define MDIO_AN_10GBT_STATUS_MS_FLT_LBN (15) /* MASTER/SLAVE config fault */
++#define MDIO_AN_10GBT_STATUS_MS_LBN (14) /* MASTER/SLAVE config */
++#define MDIO_AN_10GBT_STATUS_LOC_OK_LBN (13) /* Local OK */
++#define MDIO_AN_10GBT_STATUS_REM_OK_LBN (12) /* Remote OK */
++#define MDIO_AN_10GBT_STATUS_LP_10G_LBN (11) /* Link partner is 10GBT capable */
++#define MDIO_AN_10GBT_STATUS_LP_LTA_LBN (10) /* LP loop timing ability */
++#define MDIO_AN_10GBT_STATUS_LP_TRR_LBN (9) /* LP Training Reset Request */
++
++
++/* Packing of the prt and dev arguments of clause 45 style MDIO into a
++ * single int so they can be passed into the mdio_read/write functions
++ * that currently exist. Note that as Falcon is the only current user,
++ * the packed form is chosen to match what Falcon needs to write into
++ * a register. This is checked at compile-time so do not change it. If
++ * your target chip needs things layed out differently you will need
++ * to unpack the arguments in your chip-specific mdio functions.
++ */
++ /* These are defined by the standard. */
++#define MDIO45_PRT_ID_WIDTH (5)
++#define MDIO45_DEV_ID_WIDTH (5)
+
-+int irq_ignore_unhandled(unsigned int irq)
-+{
-+ struct physdev_irq_status_query irq_status = { .irq = irq };
++/* The prt ID is just packed in immediately to the left of the dev ID */
++#define MDIO45_PRT_DEV_WIDTH (MDIO45_PRT_ID_WIDTH + MDIO45_DEV_ID_WIDTH)
+
-+ if (!is_running_on_xen())
-+ return 0;
++#define MDIO45_PRT_ID_MASK ((1 << MDIO45_PRT_DEV_WIDTH) - 1)
++/* This is the prt + dev extended by 1 bit to hold the 'is clause 45' flag. */
++#define MDIO45_XPRT_ID_WIDTH (MDIO45_PRT_DEV_WIDTH + 1)
++#define MDIO45_XPRT_ID_MASK ((1 << MDIO45_XPRT_ID_WIDTH) - 1)
++#define MDIO45_XPRT_ID_IS10G (1 << (MDIO45_XPRT_ID_WIDTH - 1))
+
-+ (void)HYPERVISOR_physdev_op(PHYSDEVOP_irq_status_query, &irq_status);
-+ return !!(irq_status.flags & XENIRQSTAT_shared);
-+}
+
-+void notify_remote_via_irq(int irq)
-+{
-+ int evtchn = evtchn_from_irq(irq);
++#define MDIO45_PRT_ID_COMP_LBN MDIO45_DEV_ID_WIDTH
++#define MDIO45_PRT_ID_COMP_WIDTH MDIO45_PRT_ID_WIDTH
++#define MDIO45_DEV_ID_COMP_LBN 0
++#define MDIO45_DEV_ID_COMP_WIDTH MDIO45_DEV_ID_WIDTH
+
-+ if (VALID_EVTCHN(evtchn))
-+ notify_remote_via_evtchn(evtchn);
++/* Compose port and device into a phy_id */
++static inline int mdio_clause45_pack(u8 prt, u8 dev)
++{
++ efx_dword_t phy_id;
++ EFX_POPULATE_DWORD_2(phy_id, MDIO45_PRT_ID_COMP, prt,
++ MDIO45_DEV_ID_COMP, dev);
++ return MDIO45_XPRT_ID_IS10G | EFX_DWORD_VAL(phy_id);
+}
-+EXPORT_SYMBOL_GPL(notify_remote_via_irq);
+
-+int irq_to_evtchn_port(int irq)
++static inline void mdio_clause45_unpack(u32 val, u8 *prt, u8 *dev)
+{
-+ return evtchn_from_irq(irq);
++ efx_dword_t phy_id;
++ EFX_POPULATE_DWORD_1(phy_id, EFX_DWORD_0, val);
++ *prt = EFX_DWORD_FIELD(phy_id, MDIO45_PRT_ID_COMP);
++ *dev = EFX_DWORD_FIELD(phy_id, MDIO45_DEV_ID_COMP);
+}
-+EXPORT_SYMBOL_GPL(irq_to_evtchn_port);
+
-+void mask_evtchn(int port)
++static inline int mdio_clause45_read(struct efx_nic *efx,
++ u8 prt, u8 dev, u16 addr)
+{
-+ shared_info_t *s = HYPERVISOR_shared_info;
-+ synch_set_bit(port, s->evtchn_mask);
++ return efx->mii.mdio_read(efx->net_dev,
++ mdio_clause45_pack(prt, dev), addr);
+}
-+EXPORT_SYMBOL_GPL(mask_evtchn);
+
-+void unmask_evtchn(int port)
++static inline void mdio_clause45_write(struct efx_nic *efx,
++ u8 prt, u8 dev, u16 addr, int value)
+{
-+ shared_info_t *s = HYPERVISOR_shared_info;
-+ unsigned int cpu = smp_processor_id();
-+ vcpu_info_t *vcpu_info = &s->vcpu_info[cpu];
-+
-+ BUG_ON(!irqs_disabled());
-+
-+ /* Slow path (hypercall) if this is a non-local port. */
-+ if (unlikely(cpu != cpu_from_evtchn(port))) {
-+ struct evtchn_unmask unmask = { .port = port };
-+ (void)HYPERVISOR_event_channel_op(EVTCHNOP_unmask, &unmask);
-+ return;
-+ }
++ efx->mii.mdio_write(efx->net_dev,
++ mdio_clause45_pack(prt, dev), addr, value);
++}
+
-+ synch_clear_bit(port, s->evtchn_mask);
+
-+ /* Did we miss an interrupt 'edge'? Re-fire if so. */
-+ if (synch_test_bit(port, s->evtchn_pending) &&
-+ !synch_test_and_set_bit(port / BITS_PER_LONG,
-+ &vcpu_info->evtchn_pending_sel))
-+ vcpu_info->evtchn_upcall_pending = 1;
++static inline u32 mdio_clause45_read_id(struct efx_nic *efx, int mmd)
++{
++ int phy_id = efx->mii.phy_id;
++ u16 id_low = mdio_clause45_read(efx, phy_id, mmd, MDIO_MMDREG_IDLOW);
++ u16 id_hi = mdio_clause45_read(efx, phy_id, mmd, MDIO_MMDREG_IDHI);
++ return (id_hi << 16) | (id_low);
+}
-+EXPORT_SYMBOL_GPL(unmask_evtchn);
+
-+void disable_all_local_evtchn(void)
++static inline int mdio_clause45_phyxgxs_lane_sync(struct efx_nic *efx)
+{
-+ unsigned i, cpu = smp_processor_id();
-+ shared_info_t *s = HYPERVISOR_shared_info;
++ int i, sync, lane_status;
+
-+ for (i = 0; i < NR_EVENT_CHANNELS; ++i)
-+ if (cpu_from_evtchn(i) == cpu)
-+ synch_set_bit(i, &s->evtchn_mask[0]);
++ for (i = 0; i < 2; ++i)
++ lane_status = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS,
++ MDIO_PHYXS_LANE_STATE);
++
++ sync = (lane_status & (1 << MDIO_PHYXS_LANE_ALIGNED_LBN)) != 0;
++ if (!sync)
++ EFX_INFO(efx, "XGXS lane status: %x\n", lane_status);
++ return sync;
+}
+
-+static void restore_cpu_virqs(int cpu)
-+{
-+ struct evtchn_bind_virq bind_virq;
-+ int virq, irq, evtchn;
++extern const char *mdio_clause45_mmd_name(int mmd);
+
-+ for (virq = 0; virq < NR_VIRQS; virq++) {
-+ if ((irq = per_cpu(virq_to_irq, cpu)[virq]) == -1)
-+ continue;
++/*
++ * Reset a specific MMD and wait for reset to clear.
++ * Return number of spins left (>0) on success, -%ETIMEDOUT on failure.
++ *
++ * This function will sleep
++ */
++extern int mdio_clause45_reset_mmd(struct efx_nic *efx, int mmd,
++ int spins, int spintime);
+
-+ BUG_ON(irq_info[irq] != mk_irq_info(IRQT_VIRQ, virq, 0));
++/* As mdio_clause45_check_mmd but for multiple MMDs */
++int mdio_clause45_check_mmds(struct efx_nic *efx,
++ unsigned int mmd_mask, unsigned int fatal_mask);
+
-+ /* Get a new binding from Xen. */
-+ bind_virq.virq = virq;
-+ bind_virq.vcpu = cpu;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
-+ &bind_virq) != 0)
-+ BUG();
-+ evtchn = bind_virq.port;
++/* Check the link status of specified mmds in bit mask */
++extern int mdio_clause45_links_ok(struct efx_nic *efx,
++ unsigned int mmd_mask);
+
-+ /* Record the new mapping. */
-+ evtchn_to_irq[evtchn] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_VIRQ, virq, evtchn);
-+ bind_evtchn_to_cpu(evtchn, cpu);
++/* Generic transmit disable support though PMAPMD */
++extern void mdio_clause45_transmit_disable(struct efx_nic *efx,
++ int disable);
+
-+ /* Ready for use. */
-+ unmask_evtchn(evtchn);
-+ }
-+}
++/* Generic part of reconfigure: set/clear loopback bits */
++extern void mdio_clause45_phy_reconfigure(struct efx_nic *efx);
+
-+static void restore_cpu_ipis(int cpu)
-+{
-+ struct evtchn_bind_ipi bind_ipi;
-+ int ipi, irq, evtchn;
++/* Set the power state of the specified MMDs */
++extern void mdio_clause45_set_mmds_lpower(struct efx_nic *efx,
++ int low_power, unsigned int mmd_mask);
+
-+ for (ipi = 0; ipi < NR_IPIS; ipi++) {
-+ if ((irq = per_cpu(ipi_to_irq, cpu)[ipi]) == -1)
-+ continue;
++/* Read (some of) the PHY settings over MDIO */
++extern void mdio_clause45_get_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd);
+
-+ BUG_ON(irq_info[irq] != mk_irq_info(IRQT_IPI, ipi, 0));
++/* Set (some of) the PHY settings over MDIO */
++extern int mdio_clause45_set_settings(struct efx_nic *efx,
++ struct ethtool_cmd *ecmd);
+
-+ /* Get a new binding from Xen. */
-+ bind_ipi.vcpu = cpu;
-+ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
-+ &bind_ipi) != 0)
-+ BUG();
-+ evtchn = bind_ipi.port;
++/* Wait for specified MMDs to exit reset within a timeout */
++extern int mdio_clause45_wait_reset_mmds(struct efx_nic *efx,
++ unsigned int mmd_mask);
+
-+ /* Record the new mapping. */
-+ evtchn_to_irq[evtchn] = irq;
-+ irq_info[irq] = mk_irq_info(IRQT_IPI, ipi, evtchn);
-+ bind_evtchn_to_cpu(evtchn, cpu);
++#endif /* EFX_MDIO_10G_H */
+--- linux-2.6.18.8/drivers/net/sfc/mentormac.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/mentormac.c 2008-05-19 00:33:29.313835810 +0300
+@@ -0,0 +1,506 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* Ready for use. */
-+ unmask_evtchn(evtchn);
++#include <linux/delay.h>
++#include "net_driver.h"
++#include "gmii.h"
++#include "mac.h"
++
++/*
++ * Mentor MAC control
++ */
++
++/**************************************************************************
++ *
++ * Mentor MAC registers
++ *
++ **************************************************************************
++ *
++ * Register addresses are Mentor MAC register numbers. Falcon maps these
++ * registers in at 16-byte intervals. The mac_writel() and mac_readl()
++ * methods take care of abstracting away this difference.
++ */
++
++/* GMAC configuration register 1 */
++#define GM_CFG1_REG_MAC 0x00
++#define GM_SW_RST_LBN 31
++#define GM_SW_RST_WIDTH 1
++#define GM_SIM_RST_LBN 30
++#define GM_SIM_RST_WIDTH 1
++#define GM_RST_RX_MAC_CTL_LBN 19
++#define GM_RST_RX_MAC_CTL_WIDTH 1
++#define GM_RST_TX_MAC_CTL_LBN 18
++#define GM_RST_TX_MAC_CTL_WIDTH 1
++#define GM_RST_RX_FUNC_LBN 17
++#define GM_RST_RX_FUNC_WIDTH 1
++#define GM_RST_TX_FUNC_LBN 16
++#define GM_RST_TX_FUNC_WIDTH 1
++#define GM_LOOP_LBN 8
++#define GM_LOOP_WIDTH 1
++#define GM_RX_FC_EN_LBN 5
++#define GM_RX_FC_EN_WIDTH 1
++#define GM_TX_FC_EN_LBN 4
++#define GM_TX_FC_EN_WIDTH 1
++#define GM_SYNC_RXEN_LBN 3
++#define GM_SYNC_RXEN_WIDTH 1
++#define GM_RX_EN_LBN 2
++#define GM_RX_EN_WIDTH 1
++#define GM_SYNC_TXEN_LBN 1
++#define GM_SYNC_TXEN_WIDTH 1
++#define GM_TX_EN_LBN 0
++#define GM_TX_EN_WIDTH 1
++
++/* GMAC configuration register 2 */
++#define GM_CFG2_REG_MAC 0x01
++#define GM_PAMBL_LEN_LBN 12
++#define GM_PAMBL_LEN_WIDTH 4
++#define GM_IF_MODE_LBN 8
++#define GM_IF_MODE_WIDTH 2
++#define GM_HUGE_FRM_EN_LBN 5
++#define GM_HUGE_FRM_EN_WIDTH 1
++#define GM_LEN_CHK_LBN 4
++#define GM_LEN_CHK_WIDTH 1
++#define GM_PAD_CRC_EN_LBN 2
++#define GM_PAD_CRC_EN_WIDTH 1
++#define GM_CRC_EN_LBN 1
++#define GM_CRC_EN_WIDTH 1
++#define GM_FD_LBN 0
++#define GM_FD_WIDTH 1
++
++/* GMAC maximum frame length register */
++#define GM_MAX_FLEN_REG_MAC 0x04
++#define GM_MAX_FLEN_LBN 0
++#define GM_MAX_FLEN_WIDTH 16
++
++/* GMAC MII management configuration register */
++#define GM_MII_MGMT_CFG_REG_MAC 0x08
++#define GM_RST_MII_MGMT_LBN 31
++#define GM_RST_MII_MGMT_WIDTH 1
++#define GM_MGMT_SCAN_AUTO_INC_LBN 5
++#define GM_MGMT_SCAN_AUTO_INC_WIDTH 1
++#define GM_MGMT_PREM_SUPR_LBN 4
++#define GM_MGMT_PREM_SUPR_WIDTH 1
++#define GM_MGMT_CLK_SEL_LBN 0
++#define GM_MGMT_CLK_SEL_WIDTH 3
++
++/* GMAC MII management command register */
++#define GM_MII_MGMT_CMD_REG_MAC 0x09
++#define GM_MGMT_SCAN_CYC_LBN 1
++#define GM_MGMT_SCAN_CYC_WIDTH 1
++#define GM_MGMT_RD_CYC_LBN 0
++#define GM_MGMT_RD_CYC_WIDTH 1
++
++/* GMAC MII management address register */
++#define GM_MII_MGMT_ADR_REG_MAC 0x0a
++#define GM_MGMT_PHY_ADDR_LBN 8
++#define GM_MGMT_PHY_ADDR_WIDTH 5
++#define GM_MGMT_REG_ADDR_LBN 0
++#define GM_MGMT_REG_ADDR_WIDTH 5
++
++/* GMAC MII management control register */
++#define GM_MII_MGMT_CTL_REG_MAC 0x0b
++#define GM_MGMT_CTL_LBN 0
++#define GM_MGMT_CTL_WIDTH 16
++
++/* GMAC MII management status register */
++#define GM_MII_MGMT_STAT_REG_MAC 0x0c
++#define GM_MGMT_STAT_LBN 0
++#define GM_MGMT_STAT_WIDTH 16
++
++/* GMAC MII management indicators register */
++#define GM_MII_MGMT_IND_REG_MAC 0x0d
++#define GM_MGMT_NOT_VLD_LBN 2
++#define GM_MGMT_NOT_VLD_WIDTH 1
++#define GM_MGMT_SCANNING_LBN 1
++#define GM_MGMT_SCANNING_WIDTH 1
++#define GM_MGMT_BUSY_LBN 0
++#define GM_MGMT_BUSY_WIDTH 1
++
++/* GMAC station address register 1 */
++#define GM_ADR1_REG_MAC 0x10
++#define GM_HWADDR_5_LBN 24
++#define GM_HWADDR_5_WIDTH 8
++#define GM_HWADDR_4_LBN 16
++#define GM_HWADDR_4_WIDTH 8
++#define GM_HWADDR_3_LBN 8
++#define GM_HWADDR_3_WIDTH 8
++#define GM_HWADDR_2_LBN 0
++#define GM_HWADDR_2_WIDTH 8
++
++/* GMAC station address register 2 */
++#define GM_ADR2_REG_MAC 0x11
++#define GM_HWADDR_1_LBN 24
++#define GM_HWADDR_1_WIDTH 8
++#define GM_HWADDR_0_LBN 16
++#define GM_HWADDR_0_WIDTH 8
++
++/* GMAC FIFO configuration register 0 */
++#define GMF_CFG0_REG_MAC 0x12
++#define GMF_FTFENRPLY_LBN 20
++#define GMF_FTFENRPLY_WIDTH 1
++#define GMF_STFENRPLY_LBN 19
++#define GMF_STFENRPLY_WIDTH 1
++#define GMF_FRFENRPLY_LBN 18
++#define GMF_FRFENRPLY_WIDTH 1
++#define GMF_SRFENRPLY_LBN 17
++#define GMF_SRFENRPLY_WIDTH 1
++#define GMF_WTMENRPLY_LBN 16
++#define GMF_WTMENRPLY_WIDTH 1
++#define GMF_FTFENREQ_LBN 12
++#define GMF_FTFENREQ_WIDTH 1
++#define GMF_STFENREQ_LBN 11
++#define GMF_STFENREQ_WIDTH 1
++#define GMF_FRFENREQ_LBN 10
++#define GMF_FRFENREQ_WIDTH 1
++#define GMF_SRFENREQ_LBN 9
++#define GMF_SRFENREQ_WIDTH 1
++#define GMF_WTMENREQ_LBN 8
++#define GMF_WTMENREQ_WIDTH 1
++#define GMF_HSTRSTFT_LBN 4
++#define GMF_HSTRSTFT_WIDTH 1
++#define GMF_HSTRSTST_LBN 3
++#define GMF_HSTRSTST_WIDTH 1
++#define GMF_HSTRSTFR_LBN 2
++#define GMF_HSTRSTFR_WIDTH 1
++#define GMF_HSTRSTSR_LBN 1
++#define GMF_HSTRSTSR_WIDTH 1
++#define GMF_HSTRSTWT_LBN 0
++#define GMF_HSTRSTWT_WIDTH 1
++
++/* GMAC FIFO configuration register 1 */
++#define GMF_CFG1_REG_MAC 0x13
++#define GMF_CFGFRTH_LBN 16
++#define GMF_CFGFRTH_WIDTH 5
++#define GMF_CFGXOFFRTX_LBN 0
++#define GMF_CFGXOFFRTX_WIDTH 16
++
++/* GMAC FIFO configuration register 2 */
++#define GMF_CFG2_REG_MAC 0x14
++#define GMF_CFGHWM_LBN 16
++#define GMF_CFGHWM_WIDTH 6
++#define GMF_CFGLWM_LBN 0
++#define GMF_CFGLWM_WIDTH 6
++
++/* GMAC FIFO configuration register 3 */
++#define GMF_CFG3_REG_MAC 0x15
++#define GMF_CFGHWMFT_LBN 16
++#define GMF_CFGHWMFT_WIDTH 6
++#define GMF_CFGFTTH_LBN 0
++#define GMF_CFGFTTH_WIDTH 6
++
++/* GMAC FIFO configuration register 4 */
++#define GMF_CFG4_REG_MAC 0x16
++#define GMF_HSTFLTRFRM_LBN 0
++#define GMF_HSTFLTRFRM_WIDTH 18
++#define GMF_HSTFLTRFRM_PAUSE_LBN 12
++#define GMF_HSTFLTRFRM_PAUSE_WIDTH 12
++
++/* GMAC FIFO configuration register 5 */
++#define GMF_CFG5_REG_MAC 0x17
++#define GMF_CFGHDPLX_LBN 22
++#define GMF_CFGHDPLX_WIDTH 1
++#define GMF_SRFULL_LBN 21
++#define GMF_SRFULL_WIDTH 1
++#define GMF_HSTSRFULLCLR_LBN 20
++#define GMF_HSTSRFULLCLR_WIDTH 1
++#define GMF_CFGBYTMODE_LBN 19
++#define GMF_CFGBYTMODE_WIDTH 1
++#define GMF_HSTDRPLT64_LBN 18
++#define GMF_HSTDRPLT64_WIDTH 1
++#define GMF_HSTFLTRFRMDC_LBN 0
++#define GMF_HSTFLTRFRMDC_WIDTH 18
++#define GMF_HSTFLTRFRMDC_PAUSE_LBN 12
++#define GMF_HSTFLTRFRMDC_PAUSE_WIDTH 1
++
++/* TX total octet count */
++#define GM_TX_OCT_CNT_REG_MAC 0x40
++#define GM_STAT_LBN 0
++#define GM_STAT_WIDTH 32
++
++/* TX good octet count */
++#define GM_TX_GOOD_OCT_CNT_REG_MAC 0x41
++
++/* TX single collision packet count */
++#define GM_TX_SGLCOL_PKT_CNT_REG_MAC 0x42
++
++/* TX multiple collision packet count */
++#define GM_TX_MULTCOL_PKT_CNT_REG_MAC 0x43
++
++/* TX excessive collision packet count */
++#define GM_TX_EXCOL_PKT_CNT_REG_MAC 0x44
++
++/* TX deferred packet count */
++#define GM_TX_DEF_PKT_CNT_REG_MAC 0x45
++
++/* TX late packet count */
++#define GM_TX_LATECOL_PKT_CNT_REG_MAC 0x46
++
++/* TX excessive deferral packet count */
++#define GM_TX_EXDEF_PKT_CNT_REG_MAC 0x47
++
++/* TX pause packet count */
++#define GM_TX_PAUSE_PKT_CNT_REG_MAC 0x48
++
++/* TX bad packet count */
++#define GM_TX_BAD_PKT_CNT_REG_MAC 0x49
++
++/* TX unicast packet count */
++#define GM_TX_UCAST_PKT_CNT_REG_MAC 0x4a
+
-+ }
-+}
++/* TX multicast packet count */
++#define GM_TX_MCAST_PKT_CNT_REG_MAC 0x4b
+
-+void irq_resume(void)
-+{
-+ int cpu, pirq, irq, evtchn;
++/* TX broadcast packet count */
++#define GM_TX_BCAST_PKT_CNT_REG_MAC 0x4c
+
-+ init_evtchn_cpu_bindings();
++/* TX <64-byte packet count */
++#define GM_TX_LT64_PKT_CNT_REG_MAC 0x4d
+
-+ /* New event-channel space is not 'live' yet. */
-+ for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
-+ mask_evtchn(evtchn);
++/* TX 64-byte packet count */
++#define GM_TX_64_PKT_CNT_REG_MAC 0x4e
+
-+ /* Check that no PIRQs are still bound. */
-+ for (pirq = 0; pirq < NR_PIRQS; pirq++)
-+ BUG_ON(irq_info[pirq_to_irq(pirq)] != IRQ_UNBOUND);
++/* TX 65-byte to 127-byte packet count */
++#define GM_TX_65_TO_127_PKT_CNT_REG_MAC 0x4f
+
-+ /* No IRQ <-> event-channel mappings. */
-+ for (irq = 0; irq < NR_IRQS; irq++)
-+ irq_info[irq] &= ~0xFFFF; /* zap event-channel binding */
-+ for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
-+ evtchn_to_irq[evtchn] = -1;
++/* TX 128-byte to 255-byte packet count */
++#define GM_TX_128_TO_255_PKT_CNT_REG_MAC 0x50
+
-+ for_each_possible_cpu(cpu) {
-+ restore_cpu_virqs(cpu);
-+ restore_cpu_ipis(cpu);
-+ }
++/* TX 256-byte to 511-byte packet count */
++#define GM_TX_256_TO_511_PKT_CNT_REG_MAC 0x51
+
-+}
++/* TX 512-byte to 1023-byte packet count */
++#define GM_TX_512_TO_1023_PKT_CNT_REG_MAC 0x52
+
-+void __init xen_init_IRQ(void)
-+{
-+ int i;
++/* TX 1024-byte to 15xx-byte packet count */
++#define GM_TX_1024_TO_15XX_PKT_CNT_REG_MAC 0x53
+
-+ init_evtchn_cpu_bindings();
++/* TX 15xx-byte to jumbo packet count */
++#define GM_TX_15XX_TO_JUMBO_PKT_CNT_REG_MAC 0x54
+
-+ /* No event channels are 'live' right now. */
-+ for (i = 0; i < NR_EVENT_CHANNELS; i++)
-+ mask_evtchn(i);
++/* TX >jumbo packet count */
++#define GM_TX_GTJUMBO_PKT_CNT_REG_MAC 0x55
+
-+ /* No IRQ -> event-channel mappings. */
-+ for (i = 0; i < NR_IRQS; i++)
-+ irq_info[i] = IRQ_UNBOUND;
++/* RX good octet count */
++#define GM_RX_GOOD_OCT_CNT_REG_MAC 0x60
+
-+ /* Dynamic IRQ space is currently unbound. Zero the refcnts. */
-+ for (i = 0; i < NR_DYNIRQS; i++) {
-+ irq_bindcount[dynirq_to_irq(i)] = 0;
++/* RX bad octet count */
++#define GM_RX_BAD_OCT_CNT_REG_MAC 0x61
+
-+ irq_desc[dynirq_to_irq(i)].status = IRQ_DISABLED;
-+ irq_desc[dynirq_to_irq(i)].action = NULL;
-+ irq_desc[dynirq_to_irq(i)].depth = 1;
-+ irq_desc[dynirq_to_irq(i)].chip = &dynirq_type;
-+ }
++/* RX missed packet count */
++#define GM_RX_MISS_PKT_CNT_REG_MAC 0x62
+
-+ /* Phys IRQ space is statically bound (1:1 mapping). Nail refcnts. */
-+ for (i = 0; i < NR_PIRQS; i++) {
-+ irq_bindcount[pirq_to_irq(i)] = 1;
++/* RX false carrier count */
++#define GM_RX_FALSE_CRS_CNT_REG_MAC 0x63
+
-+#ifdef RTC_IRQ
-+ /* If not domain 0, force our RTC driver to fail its probe. */
-+ if ((i == RTC_IRQ) && !is_initial_xendomain())
-+ continue;
-+#endif
++/* RX pause packet count */
++#define GM_RX_PAUSE_PKT_CNT_REG_MAC 0x64
+
-+ irq_desc[pirq_to_irq(i)].status = IRQ_DISABLED;
-+ irq_desc[pirq_to_irq(i)].action = NULL;
-+ irq_desc[pirq_to_irq(i)].depth = 1;
-+ irq_desc[pirq_to_irq(i)].chip = &pirq_type;
-+ }
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/features.c linux-2.6.18-xen.hg/drivers/xen/core/features.c
---- linux-2.6.18/drivers/xen/core/features.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/features.c 2007-12-23 11:15:33.554576016 +0100
-@@ -0,0 +1,34 @@
-+/******************************************************************************
-+ * features.c
-+ *
-+ * Xen feature flags.
-+ *
-+ * Copyright (c) 2006, Ian Campbell, XenSource Inc.
-+ */
-+#include <linux/types.h>
-+#include <linux/cache.h>
-+#include <linux/module.h>
-+#include <asm/hypervisor.h>
-+#include <xen/features.h>
++/* RX bad packet count */
++#define GM_RX_BAD_PKT_CNT_REG_MAC 0x65
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
++/* RX unicast packet count */
++#define GM_RX_UCAST_PKT_CNT_REG_MAC 0x66
+
-+u8 xen_features[XENFEAT_NR_SUBMAPS * 32] __read_mostly;
-+/* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */
-+EXPORT_SYMBOL(xen_features);
++/* RX multicast packet count */
++#define GM_RX_MCAST_PKT_CNT_REG_MAC 0x67
+
-+void setup_xen_features(void)
-+{
-+ xen_feature_info_t fi;
-+ int i, j;
++/* RX broadcast packet count */
++#define GM_RX_BCAST_PKT_CNT_REG_MAC 0x68
+
-+ for (i = 0; i < XENFEAT_NR_SUBMAPS; i++) {
-+ fi.submap_idx = i;
-+ if (HYPERVISOR_xen_version(XENVER_get_features, &fi) < 0)
-+ break;
-+ for (j=0; j<32; j++)
-+ xen_features[i*32+j] = !!(fi.submap & 1<<j);
-+ }
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/firmware.c linux-2.6.18-xen.hg/drivers/xen/core/firmware.c
---- linux-2.6.18/drivers/xen/core/firmware.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/firmware.c 2007-12-23 11:15:33.554576016 +0100
-@@ -0,0 +1,74 @@
-+#include <linux/kernel.h>
-+#include <linux/errno.h>
-+#include <linux/init.h>
-+#include <linux/edd.h>
-+#include <video/edid.h>
-+#include <xen/interface/platform.h>
-+#include <asm/hypervisor.h>
++/* RX <64-byte good packet count */
++#define GM_RX_GOOD_LT64_PKT_CNT_REG_MAC 0x69
+
-+#if defined(CONFIG_EDD) || defined(CONFIG_EDD_MODULE)
-+void __init copy_edd(void)
-+{
-+ int ret;
-+ struct xen_platform_op op;
++/* RX <64-byte bad packet count */
++#define GM_RX_BAD_LT64_PKT_CNT_REG_MAC 0x6a
+
-+ if (!is_initial_xendomain())
-+ return;
++/* RX 64-byte packet count */
++#define GM_RX_64_PKT_CNT_REG_MAC 0x6b
+
-+ op.cmd = XENPF_firmware_info;
++/* RX 65-byte to 127-byte packet count */
++#define GM_RX_65_TO_127_PKT_CNT_REG_MAC 0x6c
+
-+ op.u.firmware_info.type = XEN_FW_DISK_INFO;
-+ for (op.u.firmware_info.index = 0;
-+ edd.edd_info_nr < EDDMAXNR;
-+ op.u.firmware_info.index++) {
-+ struct edd_info *info = edd.edd_info + edd.edd_info_nr;
++/* RX 128-byte to 255-byte packet count*/
++#define GM_RX_128_TO_255_PKT_CNT_REG_MAC 0x6d
+
-+ info->params.length = sizeof(info->params);
-+ set_xen_guest_handle(op.u.firmware_info.u.disk_info.edd_params,
-+ &info->params);
-+ ret = HYPERVISOR_platform_op(&op);
-+ if (ret)
-+ break;
++/* RX 256-byte to 511-byte packet count */
++#define GM_RX_256_TO_511_PKT_CNT_REG_MAC 0x6e
+
-+#define C(x) info->x = op.u.firmware_info.u.disk_info.x
-+ C(device);
-+ C(version);
-+ C(interface_support);
-+ C(legacy_max_cylinder);
-+ C(legacy_max_head);
-+ C(legacy_sectors_per_track);
-+#undef C
++/* RX 512-byte to 1023-byte packet count */
++#define GM_RX_512_TO_1023_PKT_CNT_REG_MAC 0x6f
+
-+ edd.edd_info_nr++;
-+ }
++/* RX 1024-byte to 15xx-byte packet count */
++#define GM_RX_1024_TO_15XX_PKT_CNT_REG_MAC 0x70
+
-+ op.u.firmware_info.type = XEN_FW_DISK_MBR_SIGNATURE;
-+ for (op.u.firmware_info.index = 0;
-+ edd.mbr_signature_nr < EDD_MBR_SIG_MAX;
-+ op.u.firmware_info.index++) {
-+ ret = HYPERVISOR_platform_op(&op);
-+ if (ret)
-+ break;
-+ edd.mbr_signature[edd.mbr_signature_nr++] =
-+ op.u.firmware_info.u.disk_mbr_signature.mbr_signature;
-+ }
-+}
-+#endif
++/* RX 15xx-byte to jumbo packet count */
++#define GM_RX_15XX_TO_JUMBO_PKT_CNT_REG_MAC 0x71
+
-+void __init copy_edid(void)
-+{
-+#if defined(CONFIG_FIRMWARE_EDID) && defined(CONFIG_X86)
-+ struct xen_platform_op op;
++/* RX >jumbo packet count */
++#define GM_RX_GTJUMBO_PKT_CNT_REG_MAC 0x72
+
-+ if (!is_initial_xendomain())
-+ return;
++/* RX 64-byte to 15xx-byte bad crc packet count */
++#define GM_RX_BAD_64_TO_15XX_PKT_CNT_REG_MAC 0x73
+
-+ op.cmd = XENPF_firmware_info;
-+ op.u.firmware_info.index = 0;
-+ op.u.firmware_info.type = XEN_FW_VBEDDC_INFO;
-+ set_xen_guest_handle(op.u.firmware_info.u.vbeddc_info.edid,
-+ edid_info.dummy);
-+ if (HYPERVISOR_platform_op(&op) != 0)
-+ memset(edid_info.dummy, 0x13, sizeof(edid_info.dummy));
-+#endif
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/gnttab.c linux-2.6.18-xen.hg/drivers/xen/core/gnttab.c
---- linux-2.6.18/drivers/xen/core/gnttab.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/gnttab.c 2007-12-23 11:15:33.661248285 +0100
-@@ -0,0 +1,766 @@
-+/******************************************************************************
-+ * gnttab.c
-+ *
-+ * Granting foreign access to our memory reservation.
-+ *
-+ * Copyright (c) 2005-2006, Christopher Clark
-+ * Copyright (c) 2004-2005, K A Fraser
++/* RX 15xx-byte to jumbo bad crc packet count */
++#define GM_RX_BAD_15XX_TO_JUMBO_PKT_CNT_REG_MAC 0x74
++
++/* RX >jumbo bad crc packet count */
++#define GM_RX_BAD_GTJUMBO_PKT_CNT_REG_MAC 0x75
++
++/**************************************************************************
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
++ * GMII access to PHY
+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
++ **************************************************************************
++ */
++
++/* This does not reset the PHY, only the MAC. However, TX and RX will
++ * both be disabled on the MAC after this, so the state of the PHY is
++ * somewhat irrelevant until the MAC is reinitialised.
++ */
++void mentormac_reset(struct efx_nic *efx)
++{
++ efx_dword_t reg;
++
++ EFX_POPULATE_DWORD_1(reg, GM_SW_RST, 1);
++ efx->mac_op->mac_writel(efx, ®, GM_CFG1_REG_MAC);
++ udelay(1000);
++
++ EFX_POPULATE_DWORD_1(reg, GM_SW_RST, 0);
++ efx->mac_op->mac_writel(efx, ®, GM_CFG1_REG_MAC);
++ udelay(1000);
++
++ /* Configure GMII interface so PHY is accessible */
++ EFX_POPULATE_DWORD_1(reg, GM_MGMT_CLK_SEL, 0x4);
++ efx->mac_op->mac_writel(efx, ®,
++ GM_MII_MGMT_CFG_REG_MAC);
++ udelay(10);
++}
++
++void mentormac_reconfigure(struct efx_nic *efx)
++{
++ int loopback, tx_fc, rx_fc, if_mode, full_duplex, bytemode, half_duplex;
++ unsigned int max_frame_len;
++ efx_dword_t reg;
++
++ /* Configuration register 1 */
++ tx_fc = (efx->flow_control & EFX_FC_TX) ? 1 : 0;
++ rx_fc = (efx->flow_control & EFX_FC_RX) ? 1 : 0;
++ loopback = (efx->loopback_mode == LOOPBACK_MAC) ? 1 : 0;
++ bytemode = (efx->link_options & GM_LPA_1000) ? 1 : 0;
++
++ if (efx->loopback_mode != LOOPBACK_NONE)
++ bytemode = 1;
++ if (!(efx->link_options & GM_LPA_DUPLEX))
++ /* Half-duplex operation requires TX flow control */
++ tx_fc = 1;
++ EFX_POPULATE_DWORD_5(reg,
++ GM_LOOP, loopback,
++ GM_TX_EN, 1,
++ GM_TX_FC_EN, tx_fc,
++ GM_RX_EN, 1,
++ GM_RX_FC_EN, rx_fc);
++ efx->mac_op->mac_writel(efx, ®, GM_CFG1_REG_MAC);
++ udelay(10);
++
++ /* Configuration register 2 */
++ if_mode = (bytemode) ? 2 : 1;
++ full_duplex = (efx->link_options & GM_LPA_DUPLEX) ? 1 : 0;
++ EFX_POPULATE_DWORD_4(reg,
++ GM_IF_MODE, if_mode,
++ GM_PAD_CRC_EN, 1,
++ GM_FD, full_duplex,
++ GM_PAMBL_LEN, 0x7/*datasheet recommended */);
++
++ efx->mac_op->mac_writel(efx, ®, GM_CFG2_REG_MAC);
++ udelay(10);
++
++ /* Max frame len register */
++ max_frame_len = EFX_MAX_FRAME_LEN(efx->net_dev->mtu);
++ EFX_POPULATE_DWORD_1(reg, GM_MAX_FLEN, max_frame_len);
++ efx->mac_op->mac_writel(efx, ®, GM_MAX_FLEN_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 0 */
++ EFX_POPULATE_DWORD_5(reg,
++ GMF_FTFENREQ, 1,
++ GMF_STFENREQ, 1,
++ GMF_FRFENREQ, 1,
++ GMF_SRFENREQ, 1,
++ GMF_WTMENREQ, 1);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG0_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 1 */
++ EFX_POPULATE_DWORD_2(reg,
++ GMF_CFGFRTH, 0x12,
++ GMF_CFGXOFFRTX, 0xffff);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG1_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 2 */
++ EFX_POPULATE_DWORD_2(reg,
++ GMF_CFGHWM, 0x3f,
++ GMF_CFGLWM, 0xa);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG2_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 3 */
++ EFX_POPULATE_DWORD_2(reg,
++ GMF_CFGHWMFT, 0x1c,
++ GMF_CFGFTTH, 0x08);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG3_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 4 */
++ EFX_POPULATE_DWORD_1(reg, GMF_HSTFLTRFRM_PAUSE, 1);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG4_REG_MAC);
++ udelay(10);
++
++ /* FIFO configuration register 5 */
++ half_duplex = (efx->link_options & GM_LPA_DUPLEX) ? 0 : 1;
++ efx->mac_op->mac_readl(efx, ®, GMF_CFG5_REG_MAC);
++ EFX_SET_DWORD_FIELD(reg, GMF_CFGBYTMODE, bytemode);
++ EFX_SET_DWORD_FIELD(reg, GMF_CFGHDPLX, half_duplex);
++ EFX_SET_DWORD_FIELD(reg, GMF_HSTDRPLT64, half_duplex);
++ EFX_SET_DWORD_FIELD(reg, GMF_HSTFLTRFRMDC_PAUSE, 0);
++ efx->mac_op->mac_writel(efx, ®, GMF_CFG5_REG_MAC);
++ udelay(10);
++
++ /* MAC address */
++ EFX_POPULATE_DWORD_4(reg,
++ GM_HWADDR_5, efx->net_dev->dev_addr[5],
++ GM_HWADDR_4, efx->net_dev->dev_addr[4],
++ GM_HWADDR_3, efx->net_dev->dev_addr[3],
++ GM_HWADDR_2, efx->net_dev->dev_addr[2]);
++ efx->mac_op->mac_writel(efx, ®, GM_ADR1_REG_MAC);
++ udelay(10);
++ EFX_POPULATE_DWORD_2(reg,
++ GM_HWADDR_1, efx->net_dev->dev_addr[1],
++ GM_HWADDR_0, efx->net_dev->dev_addr[0]);
++ efx->mac_op->mac_writel(efx, ®, GM_ADR2_REG_MAC);
++ udelay(10);
++}
+--- linux-2.6.18.8/drivers/net/sfc/mtd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/mtd.c 2008-05-19 00:33:29.313835810 +0300
+@@ -0,0 +1,602 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
+#include <linux/module.h>
-+#include <linux/sched.h>
-+#include <linux/mm.h>
-+#include <linux/seqlock.h>
-+#include <xen/interface/xen.h>
-+#include <xen/gnttab.h>
-+#include <asm/pgtable.h>
-+#include <asm/uaccess.h>
-+#include <asm/synch_bitops.h>
-+#include <asm/io.h>
-+#include <xen/interface/memory.h>
-+#include <xen/driver_util.h>
-+#include <asm/gnttab_dma.h>
-+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
++#include <linux/mtd/mtd.h>
++#include <linux/mtd/partitions.h>
++#include <linux/delay.h>
+
-+/* External tools reserve first few grant table entries. */
-+#define NR_RESERVED_ENTRIES 8
-+#define GNTTAB_LIST_END 0xffffffff
-+#define GREFS_PER_GRANT_FRAME (PAGE_SIZE / sizeof(grant_entry_t))
++#define EFX_DRIVER_NAME "sfc_mtd"
++#include "driverlink_api.h"
++#include "net_driver.h"
++#include "spi.h"
+
-+static grant_ref_t **gnttab_list;
-+static unsigned int nr_grant_frames;
-+static unsigned int boot_max_nr_grant_frames;
-+static int gnttab_free_count;
-+static grant_ref_t gnttab_free_head;
-+static DEFINE_SPINLOCK(gnttab_list_lock);
++/*
++ * Flash and EEPROM (MTD) device driver
++ *
++ * This file provides a separate kernel module (sfc_mtd) which
++ * exposes the flash and EEPROM devices present on Solarflare NICs as
++ * MTD devices, enabling you to reflash the boot ROM code (or use the
++ * remaining space on the flash as a jffs2 filesystem, should you want
++ * to do so).
++ */
+
-+static struct grant_entry *shared;
++#define EFX_MTD_VERIFY_BUF_LEN 16
++#define EFX_MAX_PARTITIONS 2
++#define EFX_FLASH_BOOTROM_OFFSET 0x8000U
+
-+static struct gnttab_free_callback *gnttab_free_callback_list;
++/* Write enable for EEPROM/flash configuration area
++ *
++ * Normally, writes to parts of non-volatile storage which contain
++ * critical configuration are disabled to prevent accidents. This
++ * parameter allows enabling of such writes.
++ */
++static unsigned int efx_allow_nvconfig_writes;
+
-+static int gnttab_expand(unsigned int req_entries);
++struct efx_mtd {
++ struct mtd_info mtd;
++ struct mtd_partition part[EFX_MAX_PARTITIONS];
++ char part_name[EFX_MAX_PARTITIONS][32];
++ char name[32];
++ struct efx_dl_device *efx_dev;
++ struct efx_nic *efx;
++ /* This must be held when using *spi; it guards against races
++ * with device reset and between sequences of dependent
++ * commands. */
++ struct semaphore access_lock;
++ struct efx_spi_device *spi;
++};
+
-+#define RPP (PAGE_SIZE / sizeof(grant_ref_t))
-+#define gnttab_entry(entry) (gnttab_list[(entry) / RPP][(entry) % RPP])
++/* SPI utilities */
+
-+static int get_free_entries(int count)
++static int efx_spi_fast_wait(struct efx_mtd *efx_mtd)
+{
-+ unsigned long flags;
-+ int ref, rc;
-+ grant_ref_t head;
++ struct efx_spi_device *spi = efx_mtd->spi;
++ u8 status;
++ int i, rc;
+
-+ spin_lock_irqsave(&gnttab_list_lock, flags);
++ /* Wait up to 1000us for flash/EEPROM to finish a fast operation. */
++ for (i = 0; i < 50; i++) {
++ udelay(20);
+
-+ if ((gnttab_free_count < count) &&
-+ ((rc = gnttab_expand(count - gnttab_free_count)) < 0)) {
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
-+ return rc;
++ rc = spi->read(spi, efx_mtd->efx, SPI_RDSR, -1,
++ &status, sizeof(status));
++ if (rc)
++ return rc;
++ if (!(status & SPI_STATUS_NRDY))
++ return 0;
+ }
-+
-+ ref = head = gnttab_free_head;
-+ gnttab_free_count -= count;
-+ while (count-- > 1)
-+ head = gnttab_entry(head);
-+ gnttab_free_head = gnttab_entry(head);
-+ gnttab_entry(head) = GNTTAB_LIST_END;
-+
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
-+
-+ return ref;
++ EFX_ERR(efx_mtd->efx, "timed out waiting for %s last status=0x%02x\n",
++ efx_mtd->name, status);
++ return -ETIMEDOUT;
+}
+
-+#define get_free_entry() get_free_entries(1)
-+
-+static void do_free_callbacks(void)
++static int efx_spi_slow_wait(struct efx_mtd *efx_mtd, int uninterruptible)
+{
-+ struct gnttab_free_callback *callback, *next;
-+
-+ callback = gnttab_free_callback_list;
-+ gnttab_free_callback_list = NULL;
++ struct efx_spi_device *spi = efx_mtd->spi;
++ u8 status;
++ int rc, i;
+
-+ while (callback != NULL) {
-+ next = callback->next;
-+ if (gnttab_free_count >= callback->count) {
-+ callback->next = NULL;
-+ callback->fn(callback->arg);
-+ } else {
-+ callback->next = gnttab_free_callback_list;
-+ gnttab_free_callback_list = callback;
-+ }
-+ callback = next;
++ /* Wait up to 4s for flash/EEPROM to finish a slow operation. */
++ for (i = 0; i < 40; i++) {
++ __set_current_state(uninterruptible ?
++ TASK_UNINTERRUPTIBLE : TASK_INTERRUPTIBLE);
++ schedule_timeout(HZ / 10);
++ rc = spi->read(spi, efx_mtd->efx, SPI_RDSR, -1,
++ &status, sizeof(status));
++ if (rc)
++ return rc;
++ if (!(status & SPI_STATUS_NRDY))
++ return 0;
++ if (signal_pending(current))
++ return -EINTR;
+ }
++ EFX_ERR(efx_mtd->efx, "timed out waiting for %s\n", efx_mtd->name);
++ return -ETIMEDOUT;
+}
+
-+static inline void check_free_callbacks(void)
++static int
++efx_spi_write_enable(struct efx_mtd *efx_mtd)
+{
-+ if (unlikely(gnttab_free_callback_list))
-+ do_free_callbacks();
++ struct efx_spi_device *spi = efx_mtd->spi;
++
++ return spi->write(spi, efx_mtd->efx, SPI_WREN, -1, NULL, 0);
+}
+
-+static void put_free_entry(grant_ref_t ref)
++static int efx_spi_unlock(struct efx_mtd *efx_mtd)
+{
-+ unsigned long flags;
-+ spin_lock_irqsave(&gnttab_list_lock, flags);
-+ gnttab_entry(ref) = gnttab_free_head;
-+ gnttab_free_head = ref;
-+ gnttab_free_count++;
-+ check_free_callbacks();
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
-+}
++ struct efx_spi_device *spi = efx_mtd->spi;
++ const u8 unlock_mask = (SPI_STATUS_BP2 | SPI_STATUS_BP1 |
++ SPI_STATUS_BP0);
++ u8 status;
++ int rc;
+
-+/*
-+ * Public grant-issuing interface functions
-+ */
++ rc = spi->read(spi, efx_mtd->efx, SPI_RDSR, -1, &status,
++ sizeof(status));
++ if (rc)
++ return rc;
+
-+int gnttab_grant_foreign_access(domid_t domid, unsigned long frame,
-+ int flags)
-+{
-+ int ref;
++ if (!(status & unlock_mask))
++ return 0; /* already unlocked */
+
-+ if (unlikely((ref = get_free_entry()) < 0))
-+ return -ENOSPC;
++ rc = efx_spi_write_enable(efx_mtd);
++ if (rc)
++ return rc;
++ rc = spi->write(spi, efx_mtd->efx, SPI_SST_EWSR, -1, NULL, 0);
++ if (rc)
++ return rc;
+
-+ shared[ref].frame = frame;
-+ shared[ref].domid = domid;
-+ wmb();
-+ BUG_ON(flags & (GTF_accept_transfer | GTF_reading | GTF_writing));
-+ shared[ref].flags = GTF_permit_access | flags;
++ status &= ~unlock_mask;
++ rc = spi->write(spi, efx_mtd->efx, SPI_WRSR, -1, &status,
++ sizeof(status));
++ if (rc)
++ return rc;
++ rc = efx_spi_fast_wait(efx_mtd);
++ if (rc)
++ return rc;
+
-+ return ref;
++ return 0;
+}
-+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access);
+
-+void gnttab_grant_foreign_access_ref(grant_ref_t ref, domid_t domid,
-+ unsigned long frame, int flags)
++/* Dummy device used in case of a failed reset */
++
++static int efx_spi_dummy_read(const struct efx_spi_device *spi,
++ struct efx_nic *efx, unsigned int command,
++ int address, void *data, unsigned int len)
+{
-+ shared[ref].frame = frame;
-+ shared[ref].domid = domid;
-+ wmb();
-+ BUG_ON(flags & (GTF_accept_transfer | GTF_reading | GTF_writing));
-+ shared[ref].flags = GTF_permit_access | flags;
++ return -EIO;
+}
-+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access_ref);
-+
+
-+int gnttab_query_foreign_access(grant_ref_t ref)
++static int efx_spi_dummy_write(const struct efx_spi_device *spi,
++ struct efx_nic *efx, unsigned int command,
++ int address, const void *data, unsigned int len)
+{
-+ u16 nflags;
++ return -EIO;
++}
+
-+ nflags = shared[ref].flags;
++static struct efx_spi_device efx_spi_dummy_device = {
++ .block_size = 1,
++ .erase_command = 0xff,
++ .read = efx_spi_dummy_read,
++ .write = efx_spi_dummy_write,
++};
+
-+ return (nflags & (GTF_reading|GTF_writing));
-+}
-+EXPORT_SYMBOL_GPL(gnttab_query_foreign_access);
++/* MTD interface */
+
-+int gnttab_end_foreign_access_ref(grant_ref_t ref)
++static int efx_mtd_read(struct mtd_info *mtd, loff_t start, size_t len,
++ size_t *retlen, u8 *buffer)
+{
-+ u16 flags, nflags;
++ struct efx_mtd *efx_mtd = mtd->priv;
++ struct efx_spi_device *spi;
++ unsigned int command;
++ unsigned int block_len;
++ unsigned int pos = 0;
++ int rc;
+
-+ nflags = shared[ref].flags;
-+ do {
-+ if ((flags = nflags) & (GTF_reading|GTF_writing)) {
-+ printk(KERN_DEBUG "WARNING: g.e. still in use!\n");
-+ return 0;
++ rc = down_interruptible(&efx_mtd->access_lock);
++ if (rc)
++ goto out;
++ spi = efx_mtd->spi;
++
++ while (pos < len) {
++ block_len = min((unsigned int)len - pos,
++ efx_spi_read_limit(spi, start + pos));
++ command = efx_spi_munge_command(spi, SPI_READ, start + pos);
++ rc = spi->read(spi, efx_mtd->efx, command, start + pos,
++ buffer + pos, block_len);
++ if (rc)
++ break;
++ pos += block_len;
++
++ /* Avoid locking up the system */
++ cond_resched();
++ if (signal_pending(current)) {
++ rc = -EINTR;
++ break;
+ }
-+ } while ((nflags = synch_cmpxchg_subword(&shared[ref].flags, flags, 0)) !=
-+ flags);
++ }
+
-+ return 1;
++ up(&efx_mtd->access_lock);
++out:
++ *retlen = pos;
++ return rc;
+}
-+EXPORT_SYMBOL_GPL(gnttab_end_foreign_access_ref);
+
-+void gnttab_end_foreign_access(grant_ref_t ref, unsigned long page)
++/* Check that device contents match buffer. If repeat is true, buffer
++ * contains a pattern of length EFX_MTD_VERIFY_BUF_LEN which the
++ * device contents should match repeatedly.
++ */
++static int efx_mtd_verify(struct mtd_info *mtd, loff_t start,
++ size_t len, const u8 *buffer, int repeat)
+{
-+ if (gnttab_end_foreign_access_ref(ref)) {
-+ put_free_entry(ref);
-+ if (page != 0)
-+ free_page(page);
-+ } else {
-+ /* XXX This needs to be fixed so that the ref and page are
-+ placed on a list to be freed up later. */
-+ printk(KERN_DEBUG
-+ "WARNING: leaking g.e. and page still in use!\n");
++ u8 verify_buffer[EFX_MTD_VERIFY_BUF_LEN];
++ unsigned int block_len;
++ size_t read_len;
++ unsigned int pos = 0;
++ int rc = 0;
++
++ while (pos < len) {
++ block_len = min(len - pos, sizeof(verify_buffer));
++ rc = efx_mtd_read(mtd, start + pos, block_len, &read_len,
++ verify_buffer);
++ if (rc)
++ return rc;
++ if (memcmp(repeat ? buffer : buffer + pos, verify_buffer,
++ block_len))
++ return -EIO;
++ pos += block_len;
+ }
++
++ return 0;
+}
-+EXPORT_SYMBOL_GPL(gnttab_end_foreign_access);
+
-+int gnttab_grant_foreign_transfer(domid_t domid, unsigned long pfn)
++static int efx_mtd_erase(struct mtd_info *mtd, struct erase_info *erase)
+{
-+ int ref;
++ struct efx_mtd *efx_mtd = mtd->priv;
++ struct efx_spi_device *spi;
++ u8 empty[EFX_MTD_VERIFY_BUF_LEN];
++ int rc;
+
-+ if (unlikely((ref = get_free_entry()) < 0))
-+ return -ENOSPC;
-+ gnttab_grant_foreign_transfer_ref(ref, domid, pfn);
++ if (erase->len != mtd->erasesize) {
++ rc = -EINVAL;
++ goto out;
++ }
+
-+ return ref;
-+}
-+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer);
++ rc = down_interruptible(&efx_mtd->access_lock);
++ if (rc)
++ goto out;
++ spi = efx_mtd->spi;
++ if (spi->erase_command == 0) {
++ rc = -EOPNOTSUPP;
++ goto out_up;
++ }
+
-+void gnttab_grant_foreign_transfer_ref(grant_ref_t ref, domid_t domid,
-+ unsigned long pfn)
-+{
-+ shared[ref].frame = pfn;
-+ shared[ref].domid = domid;
-+ wmb();
-+ shared[ref].flags = GTF_accept_transfer;
++ rc = efx_spi_unlock(efx_mtd);
++ if (rc)
++ goto out_up;
++ rc = efx_spi_write_enable(efx_mtd);
++ if (rc)
++ goto out_up;
++ rc = spi->write(spi, efx_mtd->efx, spi->erase_command, erase->addr,
++ NULL, 0);
++ if (rc)
++ goto out_up;
++ rc = efx_spi_slow_wait(efx_mtd, 0);
++
++out_up:
++ up(&efx_mtd->access_lock);
++ if (rc)
++ goto out;
++
++ memset(empty, 0xff, sizeof(empty));
++ rc = efx_mtd_verify(mtd, erase->addr, erase->len, empty, 1);
++
++out:
++ if (rc == 0) {
++ erase->state = MTD_ERASE_DONE;
++ } else {
++ erase->state = MTD_ERASE_FAILED;
++#if defined(EFX_USE_MTD_ERASE_FAIL_ADDR)
++ erase->fail_addr = 0xffffffff;
++#endif
++ }
++ mtd_erase_callback(erase);
++ return rc;
+}
-+EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer_ref);
+
-+unsigned long gnttab_end_foreign_transfer_ref(grant_ref_t ref)
++static int efx_mtd_write(struct mtd_info *mtd, loff_t start,
++ size_t len, size_t *retlen, const u8 *buffer)
+{
-+ unsigned long frame;
-+ u16 flags;
++ struct efx_mtd *efx_mtd = mtd->priv;
++ struct efx_spi_device *spi;
++ unsigned int command;
++ unsigned int block_len;
++ unsigned int pos = 0;
++ int rc;
+
-+ /*
-+ * If a transfer is not even yet started, try to reclaim the grant
-+ * reference and return failure (== 0).
-+ */
-+ while (!((flags = shared[ref].flags) & GTF_transfer_committed)) {
-+ if (synch_cmpxchg_subword(&shared[ref].flags, flags, 0) == flags)
-+ return 0;
-+ cpu_relax();
-+ }
++ rc = down_interruptible(&efx_mtd->access_lock);
++ if (rc)
++ goto out;
++ spi = efx_mtd->spi;
+
-+ /* If a transfer is in progress then wait until it is completed. */
-+ while (!(flags & GTF_transfer_completed)) {
-+ flags = shared[ref].flags;
-+ cpu_relax();
-+ }
++ rc = efx_spi_unlock(efx_mtd);
++ if (rc)
++ goto out_up;
+
-+ /* Read the frame number /after/ reading completion status. */
-+ rmb();
-+ frame = shared[ref].frame;
-+ BUG_ON(frame == 0);
++ while (pos < len) {
++ rc = efx_spi_write_enable(efx_mtd);
++ if (rc)
++ break;
+
-+ return frame;
-+}
-+EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer_ref);
++ block_len = min((unsigned int)len - pos,
++ efx_spi_write_limit(spi, start + pos));
++ command = efx_spi_munge_command(spi, SPI_WRITE, start + pos);
++ rc = spi->write(spi, efx_mtd->efx, command, start + pos,
++ buffer + pos, block_len);
++ if (rc)
++ break;
++ pos += block_len;
+
-+unsigned long gnttab_end_foreign_transfer(grant_ref_t ref)
-+{
-+ unsigned long frame = gnttab_end_foreign_transfer_ref(ref);
-+ put_free_entry(ref);
-+ return frame;
++ rc = efx_spi_fast_wait(efx_mtd);
++ if (rc)
++ break;
++
++ /* Avoid locking up the system */
++ cond_resched();
++ if (signal_pending(current)) {
++ rc = -EINTR;
++ break;
++ }
++ }
++
++out_up:
++ up(&efx_mtd->access_lock);
++ if (rc == 0)
++ rc = efx_mtd_verify(mtd, start, len, buffer, 0);
++out:
++ *retlen = pos;
++ return rc;
+}
-+EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer);
+
-+void gnttab_free_grant_reference(grant_ref_t ref)
++static void efx_mtd_sync(struct mtd_info *mtd)
+{
-+ put_free_entry(ref);
++ struct efx_mtd *efx_mtd = mtd->priv;
++ int rc;
++
++ down(&efx_mtd->access_lock);
++ rc = efx_spi_slow_wait(efx_mtd, 1);
++ if (rc)
++ EFX_ERR(efx_mtd->efx, "%s sync failed (%d)\n",
++ efx_mtd->name, rc);
++ up(&efx_mtd->access_lock);
+}
-+EXPORT_SYMBOL_GPL(gnttab_free_grant_reference);
+
-+void gnttab_free_grant_references(grant_ref_t head)
++/* Driverlink interface */
++
++static void efx_mtd_reset_suspend(struct efx_dl_device *efx_dev)
+{
-+ grant_ref_t ref;
-+ unsigned long flags;
-+ int count = 1;
-+ if (head == GNTTAB_LIST_END)
++ struct efx_mtd *efx_mtd = efx_dev->priv;
++
++ if (!efx_mtd)
+ return;
-+ spin_lock_irqsave(&gnttab_list_lock, flags);
-+ ref = head;
-+ while (gnttab_entry(ref) != GNTTAB_LIST_END) {
-+ ref = gnttab_entry(ref);
-+ count++;
-+ }
-+ gnttab_entry(ref) = gnttab_free_head;
-+ gnttab_free_head = head;
-+ gnttab_free_count += count;
-+ check_free_callbacks();
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++
++ /* Acquire lock to ensure that any in-progress operations have
++ * completed, and no new ones can start.
++ */
++ down(&efx_mtd->access_lock);
+}
-+EXPORT_SYMBOL_GPL(gnttab_free_grant_references);
+
-+int gnttab_alloc_grant_references(u16 count, grant_ref_t *head)
++static void efx_mtd_reset_resume(struct efx_dl_device *efx_dev, int ok)
+{
-+ int h = get_free_entries(count);
++ struct efx_mtd *efx_mtd = efx_dev->priv;
+
-+ if (h < 0)
-+ return -ENOSPC;
++ if (!efx_mtd)
++ return;
+
-+ *head = h;
++ /* If device reset failed already, or SPI device doesn't
++ * become ready, disable device.
++ */
++ if (!ok || efx_spi_slow_wait(efx_mtd, 1) != 0) {
++ efx_mtd->spi = &efx_spi_dummy_device;
++ EFX_ERR(efx_mtd->efx, "%s disabled after failed reset\n",
++ efx_mtd->name);
++ }
+
-+ return 0;
++ up(&efx_mtd->access_lock);
+}
-+EXPORT_SYMBOL_GPL(gnttab_alloc_grant_references);
+
-+int gnttab_empty_grant_references(const grant_ref_t *private_head)
++static void efx_mtd_remove(struct efx_dl_device *efx_dev)
+{
-+ return (*private_head == GNTTAB_LIST_END);
-+}
-+EXPORT_SYMBOL_GPL(gnttab_empty_grant_references);
++ struct efx_mtd *efx_mtd = efx_dev->priv;
+
-+int gnttab_claim_grant_reference(grant_ref_t *private_head)
-+{
-+ grant_ref_t g = *private_head;
-+ if (unlikely(g == GNTTAB_LIST_END))
-+ return -ENOSPC;
-+ *private_head = gnttab_entry(g);
-+ return g;
++ del_mtd_partitions(&efx_mtd->mtd);
++ kfree(efx_mtd);
++ efx_dev->priv = NULL;
+}
-+EXPORT_SYMBOL_GPL(gnttab_claim_grant_reference);
+
-+void gnttab_release_grant_reference(grant_ref_t *private_head,
-+ grant_ref_t release)
++static __devinit int efx_mtd_register(struct efx_mtd *efx_mtd,
++ struct efx_dl_device *efx_dev,
++ struct efx_nic *efx,
++ struct efx_spi_device *spi,
++ const char *type_name,
++ const char **part_type_name,
++ unsigned int num_parts)
+{
-+ gnttab_entry(release) = *private_head;
-+ *private_head = release;
-+}
-+EXPORT_SYMBOL_GPL(gnttab_release_grant_reference);
++ int i;
+
-+void gnttab_request_free_callback(struct gnttab_free_callback *callback,
-+ void (*fn)(void *), void *arg, u16 count)
-+{
-+ unsigned long flags;
-+ spin_lock_irqsave(&gnttab_list_lock, flags);
-+ if (callback->next)
-+ goto out;
-+ callback->fn = fn;
-+ callback->arg = arg;
-+ callback->count = count;
-+ callback->next = gnttab_free_callback_list;
-+ gnttab_free_callback_list = callback;
-+ check_free_callbacks();
-+out:
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
-+}
-+EXPORT_SYMBOL_GPL(gnttab_request_free_callback);
++ efx_dev->priv = efx_mtd;
+
-+void gnttab_cancel_free_callback(struct gnttab_free_callback *callback)
-+{
-+ struct gnttab_free_callback **pcb;
-+ unsigned long flags;
++ efx_mtd->efx_dev = efx_dev;
++ efx_mtd->efx = efx;
++ efx_mtd->spi = spi;
++ sema_init(&efx_mtd->access_lock, 1);
+
-+ spin_lock_irqsave(&gnttab_list_lock, flags);
-+ for (pcb = &gnttab_free_callback_list; *pcb; pcb = &(*pcb)->next) {
-+ if (*pcb == callback) {
-+ *pcb = callback->next;
-+ break;
-+ }
-+ }
-+ spin_unlock_irqrestore(&gnttab_list_lock, flags);
-+}
-+EXPORT_SYMBOL_GPL(gnttab_cancel_free_callback);
++ efx_mtd->mtd.size = spi->size;
++ efx_mtd->mtd.erasesize = spi->erase_size;
++#if defined(EFX_USE_MTD_WRITESIZE)
++ efx_mtd->mtd.writesize = 1;
++#endif
++ if (snprintf(efx_mtd->name, sizeof(efx_mtd->name),
++ "%s %s", efx->name, type_name) >=
++ sizeof(efx_mtd->name))
++ return -ENAMETOOLONG;
+
-+static int grow_gnttab_list(unsigned int more_frames)
-+{
-+ unsigned int new_nr_grant_frames, extra_entries, i;
++ efx_mtd->mtd.priv = efx_mtd;
++ efx_mtd->mtd.name = efx_mtd->name;
++ efx_mtd->mtd.erase = efx_mtd_erase;
++ efx_mtd->mtd.read = efx_mtd_read;
++ efx_mtd->mtd.write = efx_mtd_write;
++ efx_mtd->mtd.sync = efx_mtd_sync;
+
-+ new_nr_grant_frames = nr_grant_frames + more_frames;
-+ extra_entries = more_frames * GREFS_PER_GRANT_FRAME;
++ for (i = 0; i < num_parts; i++) {
++ efx_mtd->part[i].name = efx_mtd->part_name[i];
++ if (snprintf(efx_mtd->part_name[i],
++ sizeof(efx_mtd->part_name[i]),
++ "%s %s", efx->name, part_type_name[i]) >=
++ sizeof(efx_mtd->part_name[i]))
++ return -ENAMETOOLONG;
+
-+ for (i = nr_grant_frames; i < new_nr_grant_frames; i++)
-+ {
-+ gnttab_list[i] = (grant_ref_t *)__get_free_page(GFP_ATOMIC);
-+ if (!gnttab_list[i])
-+ goto grow_nomem;
++ if (efx_allow_nvconfig_writes)
++ efx_mtd->part[i].mask_flags &= ~MTD_WRITEABLE;
+ }
+
++ return add_mtd_partitions(&efx_mtd->mtd, efx_mtd->part, num_parts);
++}
+
-+ for (i = GREFS_PER_GRANT_FRAME * nr_grant_frames;
-+ i < GREFS_PER_GRANT_FRAME * new_nr_grant_frames - 1; i++)
-+ gnttab_entry(i) = i + 1;
++static int __devinit
++efx_flash_probe(struct efx_dl_device *efx_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char *silicon_rev)
++{
++ struct efx_nic *efx = efx_dl_get_nic(efx_dev);
++ struct efx_mtd *efx_mtd;
++ const char *part_type_name[2];
++ unsigned int num_parts;
++ int rc;
+
-+ gnttab_entry(i) = gnttab_free_head;
-+ gnttab_free_head = GREFS_PER_GRANT_FRAME * nr_grant_frames;
-+ gnttab_free_count += extra_entries;
++ if (!efx->spi_flash)
++ return -ENODEV;
+
-+ nr_grant_frames = new_nr_grant_frames;
++ efx_mtd = kzalloc(sizeof(*efx_mtd), GFP_KERNEL);
++ if (!efx_mtd)
++ return -ENOMEM;
+
-+ check_free_callbacks();
++ efx_mtd->mtd.type = MTD_NORFLASH;
++ efx_mtd->mtd.flags = MTD_CAP_NORFLASH;
+
-+ return 0;
-+
-+grow_nomem:
-+ for ( ; i >= nr_grant_frames; i--)
-+ free_page((unsigned long) gnttab_list[i]);
-+ return -ENOMEM;
++ part_type_name[0] = "sfc_flash_config";
++ efx_mtd->part[0].offset = 0;
++ efx_mtd->part[0].size = min(efx->spi_flash->size,
++ EFX_FLASH_BOOTROM_OFFSET);
++ efx_mtd->part[0].mask_flags = MTD_WRITEABLE;
++
++ if (efx->spi_flash->size <= EFX_FLASH_BOOTROM_OFFSET) {
++ num_parts = 1;
++ } else {
++ part_type_name[1] = "sfc_flash_bootrom";
++ efx_mtd->part[1].offset = EFX_FLASH_BOOTROM_OFFSET;
++ efx_mtd->part[1].size = (efx->spi_flash->size
++ - EFX_FLASH_BOOTROM_OFFSET);
++ num_parts = 2;
++ }
++
++ rc = efx_mtd_register(efx_mtd, efx_dev, efx, efx->spi_flash,
++ "sfc_flash", part_type_name, num_parts);
++ if (rc)
++ kfree(efx_mtd);
++ return rc;
+}
+
-+static unsigned int __max_nr_grant_frames(void)
-+{
-+ struct gnttab_query_size query;
++static struct efx_dl_driver efx_flash_driver = {
++ .name = "sfc_flash",
++ .probe = efx_flash_probe,
++ .remove = efx_mtd_remove,
++ .reset_suspend = efx_mtd_reset_suspend,
++ .reset_resume = efx_mtd_reset_resume,
++};
++
++static int __devinit
++efx_eeprom_probe(struct efx_dl_device *efx_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char *silicon_rev)
++{
++ struct efx_nic *efx = efx_dl_get_nic(efx_dev);
++ struct efx_mtd *efx_mtd;
++ const char *type_name;
++ const char *part_type_name[1];
+ int rc;
+
-+ query.dom = DOMID_SELF;
++ if (!efx->spi_eeprom)
++ return -ENODEV;
+
-+ rc = HYPERVISOR_grant_table_op(GNTTABOP_query_size, &query, 1);
-+ if ((rc < 0) || (query.status != GNTST_okay))
-+ return 4; /* Legacy max supported number of frames */
++ efx_mtd = kzalloc(sizeof(*efx_mtd), GFP_KERNEL);
++ if (!efx_mtd)
++ return -ENOMEM;
+
-+ return query.max_nr_frames;
-+}
++ efx_mtd->mtd.type = MTD_RAM;
++ efx_mtd->mtd.flags = MTD_CAP_RAM;
+
-+static inline unsigned int max_nr_grant_frames(void)
-+{
-+ unsigned int xen_max = __max_nr_grant_frames();
++ efx_mtd->part[0].offset = 0;
++ efx_mtd->part[0].size = efx->spi_eeprom->size;
++ efx_mtd->part[0].mask_flags = MTD_WRITEABLE;
+
-+ if (xen_max > boot_max_nr_grant_frames)
-+ return boot_max_nr_grant_frames;
-+ return xen_max;
++ if (efx->spi_eeprom->size <= 0x200) {
++ type_name = "sfc_small_eeprom";
++ part_type_name[0] = "sfc_small_config";
++ } else {
++ type_name = "sfc_large_eeprom";
++ part_type_name[0] = "sfc_large_config";
++ }
++
++ rc = efx_mtd_register(efx_mtd, efx_dev, efx, efx->spi_eeprom,
++ type_name, part_type_name, 1);
++ if (rc)
++ kfree(efx_mtd);
++ return rc;
+}
+
-+#ifdef CONFIG_XEN
++static struct efx_dl_driver efx_eeprom_driver = {
++ .name = "sfc_eeprom",
++ .probe = efx_eeprom_probe,
++ .remove = efx_mtd_remove,
++ .reset_suspend = efx_mtd_reset_suspend,
++ .reset_resume = efx_mtd_reset_resume,
++};
+
-+static DEFINE_SEQLOCK(gnttab_dma_lock);
++/* Kernel module interface */
+
-+#ifdef CONFIG_X86
-+static int map_pte_fn(pte_t *pte, struct page *pmd_page,
-+ unsigned long addr, void *data)
++static int __init efx_mtd_init_module(void)
+{
-+ unsigned long **frames = (unsigned long **)data;
-+
-+ set_pte_at(&init_mm, addr, pte, pfn_pte_ma((*frames)[0], PAGE_KERNEL));
-+ (*frames)++;
-+ return 0;
-+}
++ int rc;
+
-+static int unmap_pte_fn(pte_t *pte, struct page *pmd_page,
-+ unsigned long addr, void *data)
-+{
++ rc = efx_dl_register_driver(&efx_flash_driver);
++ if (rc)
++ return rc;
++ rc = efx_dl_register_driver(&efx_eeprom_driver);
++ if (rc) {
++ efx_dl_unregister_driver(&efx_flash_driver);
++ return rc;
++ }
+
-+ set_pte_at(&init_mm, addr, pte, __pte(0));
+ return 0;
+}
+
-+void *arch_gnttab_alloc_shared(unsigned long *frames)
++static void __exit efx_mtd_exit_module(void)
+{
-+ struct vm_struct *area;
-+ area = alloc_vm_area(PAGE_SIZE * max_nr_grant_frames());
-+ BUG_ON(area == NULL);
-+ return area->addr;
++ efx_dl_unregister_driver(&efx_eeprom_driver);
++ efx_dl_unregister_driver(&efx_flash_driver);
+}
-+#endif /* CONFIG_X86 */
-+
-+static int gnttab_map(unsigned int start_idx, unsigned int end_idx)
-+{
-+ struct gnttab_setup_table setup;
-+ unsigned long *frames;
-+ unsigned int nr_gframes = end_idx + 1;
-+ int rc;
+
-+ frames = kmalloc(nr_gframes * sizeof(unsigned long), GFP_ATOMIC);
-+ if (!frames)
-+ return -ENOMEM;
++module_init(efx_mtd_init_module);
++module_exit(efx_mtd_exit_module);
+
-+ setup.dom = DOMID_SELF;
-+ setup.nr_frames = nr_gframes;
-+ set_xen_guest_handle(setup.frame_list, frames);
++MODULE_AUTHOR("Michael Brown <mbrown@fensystems.co.uk> and "
++ "Solarflare Communications");
++MODULE_DESCRIPTION("SFC MTD driver");
++MODULE_LICENSE("GPL");
+--- linux-2.6.18.8/drivers/net/sfc/net_driver.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/net_driver.h 2008-05-19 00:33:29.317836041 +0300
+@@ -0,0 +1,1096 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ rc = HYPERVISOR_grant_table_op(GNTTABOP_setup_table, &setup, 1);
-+ if (rc == -ENOSYS) {
-+ kfree(frames);
-+ return -ENOSYS;
-+ }
++/* Common definitions for all Efx net driver code */
+
-+ BUG_ON(rc || setup.status);
++#ifndef EFX_NET_DRIVER_H
++#define EFX_NET_DRIVER_H
+
-+ if (shared == NULL)
-+ shared = arch_gnttab_alloc_shared(frames);
++#include <linux/version.h>
++#include <linux/netdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/ethtool.h>
++#include <linux/if_vlan.h>
++#include <linux/timer.h>
++#include <linux/mii.h>
++#include <linux/list.h>
++#include <linux/pci.h>
++#include <linux/device.h>
+
-+#ifdef CONFIG_X86
-+ rc = apply_to_page_range(&init_mm, (unsigned long)shared,
-+ PAGE_SIZE * nr_gframes,
-+ map_pte_fn, &frames);
-+ BUG_ON(rc);
-+ frames -= nr_gframes; /* adjust after map_pte_fn() */
-+#endif /* CONFIG_X86 */
++/* Must come before other headers */
++#include "kernel_compat.h"
+
-+ kfree(frames);
++#include "enum.h"
++#include "bitfield.h"
++#include "driverlink_api.h"
++#include "driverlink.h"
++#include "i2c-direct.h"
+
-+ return 0;
-+}
++ #ifndef EFX_USE_DEBUGFS
++ /* Sick, but we have no other use for dentry */
++ #define dentry proc_dir_entry
++ #endif
+
-+static void gnttab_page_free(struct page *page)
-+{
-+ ClearPageForeign(page);
-+ gnttab_reset_grant_page(page);
-+ put_page(page);
-+}
++#define EFX_MAX_LRO_DESCRIPTORS 8
++#define EFX_MAX_LRO_AGGR MAX_SKB_FRAGS
+
-+/*
-+ * Must not be called with IRQs off. This should only be used on the
-+ * slow path.
++/**************************************************************************
+ *
-+ * Copy a foreign granted page to local memory.
-+ */
-+int gnttab_copy_grant_page(grant_ref_t ref, struct page **pagep)
-+{
-+ struct gnttab_unmap_and_replace unmap;
-+ mmu_update_t mmu;
-+ struct page *page;
-+ struct page *new_page;
-+ void *new_addr;
-+ void *addr;
-+ paddr_t pfn;
-+ maddr_t mfn;
-+ maddr_t new_mfn;
-+ int err;
++ * Build definitions
++ *
++ **************************************************************************/
++#ifndef EFX_DRIVER_NAME
++#define EFX_DRIVER_NAME "sfc"
++#endif
++#define EFX_DRIVER_VERSION "2.2.0101"
+
-+ page = *pagep;
-+ if (!get_page_unless_zero(page))
-+ return -ENOENT;
++#ifdef EFX_ENABLE_DEBUG
++#define EFX_BUG_ON_PARANOID(x) BUG_ON(x)
++#define EFX_WARN_ON_PARANOID(x) WARN_ON(x)
++#else
++#define EFX_BUG_ON_PARANOID(x) do {} while (0)
++#define EFX_WARN_ON_PARANOID(x) do {} while (0)
++#endif
+
-+ err = -ENOMEM;
-+ new_page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
-+ if (!new_page)
-+ goto out;
++/* Include net device name in log messages if it has been registered.
++ * Use efx->name not efx->net_dev->name so that races with (un)registration
++ * are harmless.
++ */
++#define NET_DEV_NAME(efx) ((efx)->net_dev_registered ? (efx)->name : "")
+
-+ new_addr = page_address(new_page);
-+ addr = page_address(page);
-+ memcpy(new_addr, addr, PAGE_SIZE);
++/* Un-rate-limited logging */
++#define EFX_ERR(efx, fmt, args...) \
++dev_err(&((efx)->pci_dev->dev), "ERR: %s " fmt, NET_DEV_NAME(efx), ##args)
+
-+ pfn = page_to_pfn(page);
-+ mfn = pfn_to_mfn(pfn);
-+ new_mfn = virt_to_mfn(new_addr);
++#define EFX_INFO(efx, fmt, args...) \
++dev_info(&((efx)->pci_dev->dev), "INFO: %s " fmt, NET_DEV_NAME(efx), ##args)
+
-+ write_seqlock(&gnttab_dma_lock);
++#ifdef EFX_ENABLE_DEBUG
++#define EFX_LOG(efx, fmt, args...) \
++dev_info(&((efx)->pci_dev->dev), "DBG: %s " fmt, NET_DEV_NAME(efx), ##args)
++#else
++#define EFX_LOG(efx, fmt, args...) \
++dev_dbg(&((efx)->pci_dev->dev), "DBG: %s " fmt, NET_DEV_NAME(efx), ##args)
++#endif
+
-+ /* Make seq visible before checking page_mapped. */
-+ smp_mb();
++#define EFX_TRACE(efx, fmt, args...) do {} while (0)
+
-+ /* Has the page been DMA-mapped? */
-+ if (unlikely(page_mapped(page))) {
-+ write_sequnlock(&gnttab_dma_lock);
-+ put_page(new_page);
-+ err = -EBUSY;
-+ goto out;
-+ }
++#define EFX_REGDUMP(efx, fmt, args...) do {} while (0)
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap))
-+ set_phys_to_machine(pfn, new_mfn);
++/* Rate-limited logging */
++#define EFX_ERR_RL(efx, fmt, args...) \
++do {if (net_ratelimit()) EFX_ERR(efx, fmt, ##args); } while (0)
+
-+ gnttab_set_replace_op(&unmap, (unsigned long)addr,
-+ (unsigned long)new_addr, ref);
++#define EFX_INFO_RL(efx, fmt, args...) \
++do {if (net_ratelimit()) EFX_INFO(efx, fmt, ##args); } while (0)
+
-+ err = HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_replace,
-+ &unmap, 1);
-+ BUG_ON(err);
-+ BUG_ON(unmap.status);
++#define EFX_LOG_RL(efx, fmt, args...) \
++do {if (net_ratelimit()) EFX_LOG(efx, fmt, ##args); } while (0)
+
-+ write_sequnlock(&gnttab_dma_lock);
++/* Kernel headers may redefine inline anyway */
++#ifndef inline
++#define inline inline __attribute__ ((always_inline))
++#endif
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ set_phys_to_machine(page_to_pfn(new_page), INVALID_P2M_ENTRY);
++/**************************************************************************
++ *
++ * Efx data structures
++ *
++ **************************************************************************/
+
-+ mmu.ptr = (new_mfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE;
-+ mmu.val = pfn;
-+ err = HYPERVISOR_mmu_update(&mmu, 1, NULL, DOMID_SELF);
-+ BUG_ON(err);
-+ }
++#define EFX_MAX_CHANNELS 32
++#define EFX_MAX_TX_QUEUES 1
++#define EFX_MAX_RX_QUEUES EFX_MAX_CHANNELS
+
-+ new_page->mapping = page->mapping;
-+ new_page->index = page->index;
-+ set_bit(PG_foreign, &new_page->flags);
-+ *pagep = new_page;
++/**
++ * struct efx_special_buffer - An Efx special buffer
++ * @addr: CPU base address of the buffer
++ * @dma_addr: DMA base address of the buffer
++ * @len: Buffer length, in bytes
++ * @index: Buffer index within controller;s buffer table
++ * @entries: Number of buffer table entries
++ *
++ * Special buffers are used for the event queues and the TX and RX
++ * descriptor queues for each channel. They are *not* used for the
++ * actual transmit and receive buffers.
++ *
++ * Note that for Falcon, TX and RX descriptor queues live in host memory.
++ * Allocation and freeing procedures must take this into account.
++ */
++struct efx_special_buffer {
++ void *addr;
++ dma_addr_t dma_addr;
++ unsigned int len;
++ int index;
++ int entries;
++};
+
-+ SetPageForeign(page, gnttab_page_free);
-+ page->mapping = NULL;
++/**
++ * struct efx_tx_buffer - An Efx TX buffer
++ * @skb: The associated socket buffer.
++ * Set only on the final fragment of a packet; %NULL for all other
++ * fragments. When this fragment completes, then we can free this
++ * skb.
++ * @dma_addr: DMA address of the fragment.
++ * @len: Length of this fragment.
++ * This field is zero when the queue slot is empty.
++ * @continuation: True if this fragment is not the end of a packet.
++ * @unmap_single: True if pci_unmap_single should be used.
++ * @unmap_addr: DMA address to unmap
++ * @unmap_len: Length of this fragment to unmap
++ */
++struct efx_tx_buffer {
++ const struct sk_buff *skb;
++ dma_addr_t dma_addr;
++ unsigned short len;
++ unsigned char continuation;
++ unsigned char unmap_single;
++ dma_addr_t unmap_addr;
++ unsigned short unmap_len;
++};
+
-+out:
-+ put_page(page);
-+ return err;
-+}
-+EXPORT_SYMBOL_GPL(gnttab_copy_grant_page);
++/**
++ * struct efx_tx_queue - An Efx TX queue
++ *
++ * This is a ring buffer of TX fragments.
++ * Since the TX completion path always executes on the same
++ * CPU and the xmit path can operate on different CPUs,
++ * performance is increased by ensuring that the completion
++ * path and the xmit path operate on different cache lines.
++ * This is particularly important if the xmit path is always
++ * executing on one CPU which is different from the completion
++ * path. There is also a cache line for members which are
++ * read but not written on the fast path.
++ *
++ * @efx: The associated Efx NIC
++ * @queue: DMA queue number
++ * @used: Queue is used by net driver
++ * @channel: The associated channel
++ * @buffer: The software buffer ring
++ * @txd: The hardware descriptor ring
++ * @debug_dir: debugfs directory
++ * @read_count: Current read pointer.
++ * This is the number of buffers that have been removed from both rings.
++ * @stopped: Stopped flag.
++ * Set if this TX queue is currently stopping its port.
++ * @insert_count: Current insert pointer
++ * This is the number of buffers that have been added to the
++ * software ring.
++ * @write_count: Current write pointer
++ * This is the number of buffers that have been added to the
++ * hardware ring.
++ * @old_read_count: The value of read_count when last checked.
++ * This is here for performance reasons. The xmit path will
++ * only get the up-to-date value of read_count if this
++ * variable indicates that the queue is full. This is to
++ * avoid cache-line ping-pong between the xmit path and the
++ * completion path.
++ */
++struct efx_tx_queue {
++ /* Members which don't change on the fast path */
++ struct efx_nic *efx ____cacheline_aligned_in_smp;
++ int queue;
++ int used;
++ struct efx_channel *channel;
++ struct efx_nic *nic;
++ struct efx_tx_buffer *buffer;
++ struct efx_special_buffer txd;
++#ifdef CONFIG_SFC_DEBUGFS
++ struct dentry *debug_dir;
++#endif
++
++ /* Members used mainly on the completion path */
++ unsigned int read_count ____cacheline_aligned_in_smp;
++ int stopped;
++
++ /* Members used only on the xmit path */
++ unsigned int insert_count ____cacheline_aligned_in_smp;
++ unsigned int write_count;
++ unsigned int old_read_count;
++};
+
-+void gnttab_reset_grant_page(struct page *page)
-+{
-+ init_page_count(page);
-+ reset_page_mapcount(page);
-+}
-+EXPORT_SYMBOL_GPL(gnttab_reset_grant_page);
++/**
++ * struct efx_rx_buffer - An Efx RX data buffer
++ * @dma_addr: DMA base address of the buffer
++ * @skb: The associated socket buffer, if any.
++ * If both this and page are %NULL, the buffer slot is currently free.
++ * @page: The associated page buffer, if any.
++ * If both this and skb are %NULL, the buffer slot is currently free.
++ * @data: Pointer to ethernet header
++ * @len: Buffer length, in bytes.
++ * @unmap_addr: DMA address to unmap
++ */
++struct efx_rx_buffer {
++ dma_addr_t dma_addr;
++ struct sk_buff *skb;
++ struct page *page;
++ char *data;
++ unsigned int len;
++ dma_addr_t unmap_addr;
++};
+
-+/*
-+ * Keep track of foreign pages marked as PageForeign so that we don't
-+ * return them to the remote domain prematurely.
-+ *
-+ * PageForeign pages are pinned down by increasing their mapcount.
++/**
++ * struct efx_rx_queue - An Efx RX queue
++ * @efx: The associated Efx NIC
++ * @queue: DMA queue number
++ * @used: Queue is used by net driver
++ * @channel: The associated channel
++ * @buffer: The software buffer ring
++ * @rxd: The hardware descriptor ring
++ * @added_count: Number of buffers added to the receive queue.
++ * @notified_count: Number of buffers given to NIC (<= @added_count).
++ * @removed_count: Number of buffers removed from the receive queue.
++ * @add_lock: Receive queue descriptor add spin lock.
++ * This lock must be held in order to add buffers to the RX
++ * descriptor ring (rxd and buffer) and to update added_count (but
++ * not removed_count).
++ * @max_fill: RX descriptor maximum fill level (<= ring size)
++ * @fast_fill_trigger: RX descriptor fill level that will trigger a fast fill
++ * (<= @max_fill)
++ * @fast_fill_limit: The level to which a fast fill will fill
++ * (@fast_fill_trigger <= @fast_fill_limit <= @max_fill)
++ * @min_fill: RX descriptor minimum non-zero fill level.
++ * This records the minimum fill level observed when a ring
++ * refill was triggered.
++ * @min_overfill: RX descriptor minimum overflow fill level.
++ * This records the minimum fill level at which RX queue
++ * overflow was observed. It should never be set.
++ * @alloc_page_count: RX allocation strategy counter.
++ * @alloc_skb_count: RX allocation strategy counter.
++ * @work: Descriptor push work thread
++ * @buf_page: Page for next RX buffer.
++ * We can use a single page for multiple RX buffers. This tracks
++ * the remaining space in the allocation.
++ * @buf_dma_addr: Page's DMA address.
++ * @buf_data: Page's host address.
++ * @debug_dir: debugfs directory
++ */
++struct efx_rx_queue {
++ struct efx_nic *efx;
++ int queue;
++ int used;
++ struct efx_channel *channel;
++ struct efx_rx_buffer *buffer;
++ struct efx_special_buffer rxd;
++
++ int added_count;
++ int notified_count;
++ int removed_count;
++ spinlock_t add_lock;
++ unsigned int max_fill;
++ unsigned int fast_fill_trigger;
++ unsigned int fast_fill_limit;
++ unsigned int min_fill;
++ unsigned int min_overfill;
++ unsigned int alloc_page_count;
++ unsigned int alloc_skb_count;
++ struct delayed_work work;
++ unsigned int slow_fill_count;
++
++ struct page *buf_page;
++ dma_addr_t buf_dma_addr;
++ char *buf_data;
++
++#ifdef CONFIG_SFC_DEBUGFS
++ struct dentry *debug_dir;
++#endif
++};
++
++/**
++ * struct efx_buffer - An Efx general-purpose buffer
++ * @addr: host base address of the buffer
++ * @dma_addr: DMA base address of the buffer
++ * @len: Buffer length, in bytes
+ *
-+ * All other pages are simply returned as is.
++ * Falcon uses these buffers for its interrupt status registers and
++ * MAC stats dumps.
+ */
-+void __gnttab_dma_map_page(struct page *page)
-+{
-+ unsigned int seq;
++struct efx_buffer {
++ void *addr;
++ dma_addr_t dma_addr;
++ unsigned int len;
++};
+
-+ if (!is_running_on_xen() || !PageForeign(page))
-+ return;
+
-+ do {
-+ seq = read_seqbegin(&gnttab_dma_lock);
++/* Flags for channel->used_flags */
++#define EFX_USED_BY_RX 1
++#define EFX_USED_BY_TX 2
++#define EFX_USED_BY_RX_TX (EFX_USED_BY_RX | EFX_USED_BY_TX)
+
-+ if (gnttab_dma_local_pfn(page))
-+ break;
++enum efx_rx_alloc_method {
++ RX_ALLOC_METHOD_AUTO = 0,
++ RX_ALLOC_METHOD_SKB = 1,
++ RX_ALLOC_METHOD_PAGE = 2,
++};
+
-+ atomic_set(&page->_mapcount, 0);
++/**
++ * struct efx_channel - An Efx channel
++ *
++ * A channel comprises an event queue, at least one TX queue, at least
++ * one RX queue, and an associated tasklet for processing the event
++ * queue.
++ *
++ * @efx: Associated Efx NIC
++ * @evqnum: Event queue number
++ * @channel: Channel instance number
++ * @used_flags: Channel is used by net driver
++ * @enabled: Channel enabled indicator
++ * @irq: IRQ number (MSI and MSI-X only)
++ * @has_interrupt: Channel has an interrupt
++ * @irq_moderation: IRQ moderation value (in us)
++ * @napi_dev: Net device used with NAPI
++ * @napi_str: NAPI control structure
++ * @reset_work: Scheduled reset work thread
++ * @work_pending: Is work pending via NAPI?
++ * @eventq: Event queue buffer
++ * @eventq_read_ptr: Event queue read pointer
++ * @last_eventq_read_ptr: Last event queue read pointer value.
++ * @eventq_magic: Event queue magic value for driver-generated test events
++ * @debug_dir: debugfs directory
++ * @rx_alloc_level: Watermark based heuristic counter for pushing descriptors
++ * and diagnostic counters
++ * @rx_alloc_push_pages: RX allocation method currently in use for pushing
++ * descriptors
++ * @rx_alloc_pop_pages: RX allocation method currently in use for popping
++ * descriptors
++ * @n_rx_tobe_disc: Count of RX_TOBE_DISC errors
++ * @n_rx_ip_frag_err: Count of RX IP fragment errors
++ * @n_rx_ip_hdr_chksum_err: Count of RX IP header checksum errors
++ * @n_rx_tcp_udp_chksum_err: Count of RX TCP and UDP checksum errors
++ * @n_rx_frm_trunc: Count of RX_FRM_TRUNC errors
++ * @n_rx_overlength: Count of RX_OVERLENGTH errors
++ * @n_skbuff_leaks: Count of skbuffs leaked due to RX overrun
++ */
++struct efx_channel {
++ struct efx_nic *efx;
++ int evqnum;
++ int channel;
++ int used_flags;
++ int enabled;
++ int irq;
++ unsigned int has_interrupt;
++ unsigned int irq_moderation;
++ struct net_device *napi_dev;
++#if !defined(EFX_HAVE_OLD_NAPI)
++ struct napi_struct napi_str;
++#endif
++ struct work_struct reset_work;
++ int work_pending;
++ struct efx_special_buffer eventq;
++ unsigned int eventq_read_ptr;
++ unsigned int last_eventq_read_ptr;
++ unsigned int eventq_magic;
+
-+ /* Make _mapcount visible before read_seqretry. */
-+ smp_mb();
-+ } while (unlikely(read_seqretry(&gnttab_dma_lock, seq)));
-+}
++#ifdef CONFIG_SFC_DEBUGFS
++ struct dentry *debug_dir;
++#endif
+
-+int gnttab_resume(void)
-+{
-+ if (max_nr_grant_frames() < nr_grant_frames)
-+ return -ENOSYS;
-+ return gnttab_map(0, nr_grant_frames - 1);
-+}
++ int rx_alloc_level;
++ int rx_alloc_push_pages;
++ int rx_alloc_pop_pages;
+
-+int gnttab_suspend(void)
-+{
-+#ifdef CONFIG_X86
-+ apply_to_page_range(&init_mm, (unsigned long)shared,
-+ PAGE_SIZE * nr_grant_frames,
-+ unmap_pte_fn, NULL);
-+#endif
-+ return 0;
-+}
++ unsigned n_rx_tobe_disc;
++ unsigned n_rx_ip_frag_err;
++ unsigned n_rx_ip_hdr_chksum_err;
++ unsigned n_rx_tcp_udp_chksum_err;
++ unsigned n_rx_frm_trunc;
++ unsigned n_rx_overlength;
++ unsigned n_skbuff_leaks;
+
-+#else /* !CONFIG_XEN */
++ /* Used to pipeline received packets in order to optimise memory
++ * access with prefetches.
++ */
++ struct efx_rx_buffer *rx_pkt;
++ int rx_pkt_csummed;
+
-+#include <platform-pci.h>
++};
+
-+static unsigned long resume_frames;
++/**
++ * struct efx_blinker - S/W LED blinking context
++ * @led_num: LED ID (board-specific meaning)
++ * @state: Current state - on or off
++ * @resubmit: Timer resubmission flag
++ * @timer: Control timer for blinking
++ */
++struct efx_blinker {
++ int led_num;
++ int state;
++ int resubmit;
++ struct timer_list timer;
++};
+
-+static int gnttab_map(unsigned int start_idx, unsigned int end_idx)
-+{
-+ struct xen_add_to_physmap xatp;
-+ unsigned int i = end_idx;
+
-+ /* Loop backwards, so that the first hypercall has the largest index,
-+ * ensuring that the table will grow only once.
-+ */
-+ do {
-+ xatp.domid = DOMID_SELF;
-+ xatp.idx = i;
-+ xatp.space = XENMAPSPACE_grant_table;
-+ xatp.gpfn = (resume_frames >> PAGE_SHIFT) + i;
-+ if (HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp))
-+ BUG();
-+ } while (i-- > start_idx);
++/**
++ * struct efx_board - board information
++ * @type: Board model type
++ * @major: Major rev. ('A', 'B' ...)
++ * @minor: Minor rev. (0, 1, ...)
++ * @init: Initialisation function
++ * @init_leds: Sets up board LEDs
++ * @set_fault_led: Turns the fault LED on or off
++ * @blink: Starts/stops blinking
++ * @monitor: Board-specific health check function
++ * @fini: Cleanup function
++ * @interpret_sensor: Function to interpret LM87 sensor meanings.
++ * Returns %FALSE if no board-specific meaning.
++ * @mwatts: Power requirements (mW)
++ * @blinker: used to blink LEDs in software
++ * @lm87_addr: I2C address of LM87 monitoring chip (0 if absent)
++ */
++struct efx_board {
++ int type;
++ int major;
++ int minor;
++ int (*init) (struct efx_nic *nic);
++ /* As the LEDs are typically attached to the PHY, LEDs
++ * have a separate init callback that happens later than
++ * board init. */
++ int (*init_leds)(struct efx_nic *efx);
++ int (*monitor) (struct efx_nic *nic);
++ void (*set_fault_led) (struct efx_nic *efx, int state);
++ void (*blink) (struct efx_nic *efx, int start);
++ void (*fini) (struct efx_nic *nic);
++ int (*interpret_sensor) (struct efx_nic *nic, int num, unsigned val);
++ unsigned mwatts;
++ struct efx_blinker blinker;
++ unsigned int lm87_addr;
++};
++
++#define STRING_TABLE_LOOKUP(val, member) \
++ ((val) < member ## _max) ? member ## _names[val] : "(invalid)"
++
++extern const char *efx_loopback_mode_names[];
++extern const unsigned int efx_loopback_mode_max;
++#define LOOPBACK_MODE(efx) \
++ STRING_TABLE_LOOKUP((efx)->loopback_mode, efx_loopback_mode)
++
++extern const char *efx_interrupt_mode_names[];
++extern const unsigned int efx_interrupt_mode_max;
++#define INT_MODE(efx) \
++ STRING_TABLE_LOOKUP(efx->interrupt_mode, efx_interrupt_mode)
++
++extern const char *efx_reset_type_names[];
++extern const unsigned int efx_reset_type_max;
++#define RESET_TYPE(type) \
++ STRING_TABLE_LOOKUP(type, efx_reset_type)
++
++extern const char *efx_phy_type_names[];
++extern const unsigned int efx_phy_type_max;
++#define PHY_TYPE(efx) \
++ STRING_TABLE_LOOKUP((efx)->phy_type, efx_phy_type)
++
++extern const char *efx_nic_state_names[];
++extern const unsigned int efx_nic_state_max;
++#define STATE_NAME(efx) \
++ STRING_TABLE_LOOKUP((efx)->state, efx_nic_state)
++
++enum efx_int_mode {
++ /* Be careful if altering to correct macro below */
++ EFX_INT_MODE_MSIX = 0,
++ EFX_INT_MODE_MSI = 1,
++ EFX_INT_MODE_LEGACY = 2,
++ EFX_INT_MODE_MAX /* Insert any new items before this */
++};
++#define EFX_INT_MODE_USE_MSI(x) (((x)->interrupt_mode) <= EFX_INT_MODE_MSI)
++
++enum phy_type {
++ PHY_TYPE_NONE = 0,
++ PHY_TYPE_CX4_RTMR = 1,
++ PHY_TYPE_1G_ALASKA = 2,
++ PHY_TYPE_10XPRESS = 3,
++ PHY_TYPE_XFP = 4,
++ PHY_TYPE_PM8358 = 6,
++ PHY_TYPE_MAX /* Insert any new items before this */
++};
++
++#define PHY_ADDR_INVALID 0xff
++
++#define EFX_IS10G(efx) ((efx)->is_10g)
++#define EFX_ISCLAUSE45(efx) ((efx)->phy_type != PHY_TYPE_1G_ALASKA)
++
++enum nic_state {
++ STATE_INIT = 0, /* suspend_lock always held */
++ STATE_RUNNING = 1,
++ STATE_FINI = 2,
++ STATE_RESETTING = 3, /* suspend_lock always held */
++ STATE_DISABLED = 4,
++ STATE_MAX,
++};
++
++/*
++ * Alignment of page-allocated RX buffers
++ *
++ * Controls the number of bytes inserted at the start of an RX buffer.
++ * This is the equivalent of NET_IP_ALIGN [which controls the alignment
++ * of the skb->head for hardware DMA].
++ */
++#ifdef __ia64__
++#define EFX_PAGE_IP_ALIGN 2
++#else
++#define EFX_PAGE_IP_ALIGN 0
++#endif
+
-+ return 0;
-+}
++/*
++ * Alignment of the skb->head which wraps a page-allocated RX buffer
++ *
++ * The skb allocated to wrap an rx_buffer can have this alignment. Since
++ * the data is memcpy'd from the rx_buf, it does not need to be equal to
++ * EFX_PAGE_IP_ALIGN.
++ */
++#define EFX_PAGE_SKB_ALIGN 2
+
-+int gnttab_resume(void)
-+{
-+ unsigned int max_nr_gframes, nr_gframes;
++/* Forward declaration */
++struct efx_nic;
+
-+ nr_gframes = nr_grant_frames;
-+ max_nr_gframes = max_nr_grant_frames();
-+ if (max_nr_gframes < nr_gframes)
-+ return -ENOSYS;
++/* Pseudo bit-mask flow control field */
++enum efx_fc_type {
++ EFX_FC_RX = 1,
++ EFX_FC_TX = 2,
++ EFX_FC_AUTO = 4,
++};
+
-+ if (!resume_frames) {
-+ resume_frames = alloc_xen_mmio(PAGE_SIZE * max_nr_gframes);
-+ shared = ioremap(resume_frames, PAGE_SIZE * max_nr_gframes);
-+ if (shared == NULL) {
-+ printk("error to ioremap gnttab share frames\n");
-+ return -1;
-+ }
-+ }
++/**
++ * struct efx_mac_operations - Efx MAC operations table
++ * @mac_writel: Write dword to MAC register
++ * @mac_readl: Read dword from a MAC register
++ * @init: Initialise MAC and PHY
++ * @reconfigure: Reconfigure MAC and PHY (e.g. for new link parameters)
++ * @update_stats: Update statistics
++ * @fini: Shut down MAC and PHY
++ * @check_hw: Check hardware
++ * @fake_phy_event: Simulate a PHY event on a port
++ * @get_settings: Get ethtool settings
++ * @set_settings: Set ethtool settings
++ * @set_pause: Set pause parameters
++ */
++struct efx_mac_operations {
++ void (*mac_writel) (struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg);
++ void (*mac_readl) (struct efx_nic *efx,
++ efx_dword_t *value, unsigned int mac_reg);
++ int (*init) (struct efx_nic *efx);
++ void (*reconfigure) (struct efx_nic *efx);
++ void (*update_stats) (struct efx_nic *efx);
++ void (*fini) (struct efx_nic *efx);
++ int (*check_hw) (struct efx_nic *efx);
++ void (*fake_phy_event) (struct efx_nic *efx);
++
++ int (*get_settings) (struct efx_nic *efx,
++ struct ethtool_cmd *ecmd);
++ int (*set_settings) (struct efx_nic *efx,
++ struct ethtool_cmd *ecmd);
++ int (*set_pause) (struct efx_nic *efx,
++ enum efx_fc_type pause_params);
++};
+
-+ gnttab_map(0, nr_gframes - 1);
++/**
++ * struct efx_phy_operations - Efx PHY operations table
++ * @init: Initialise PHY
++ * @fini: Shut down PHY
++ * @reconfigure: Reconfigure PHY (e.g. for new link parameters)
++ * @clear_interrupt: Clear down interrupt
++ * @blink: Blink LEDs
++ * @check_hw: Check hardware
++ * @reset_xaui: Reset XAUI side of PHY for (software sequenced reset)
++ * @mmds: MMD presence mask
++ * @loopbacks: Supported loopback modes mask
++ * @startup_loopback: Loopback mode for start-of-day self-test
++ */
++struct efx_phy_operations {
++ int (*init) (struct efx_nic *efx);
++ void (*fini) (struct efx_nic *efx);
++ void (*reconfigure) (struct efx_nic *efx);
++ void (*clear_interrupt) (struct efx_nic *efx);
++ int (*check_hw) (struct efx_nic *efx);
++ void (*reset_xaui) (struct efx_nic *efx);
++ int mmds;
++ unsigned loopbacks;
++ unsigned startup_loopback;
++};
++
++/*
++ * Efx extended statistics
++ *
++ * Not all statistics are provided by all supported MACs. The purpose
++ * is this structure is to contain the raw statistics provided by each
++ * MAC.
++ */
++struct efx_mac_stats {
++ u64 tx_bytes;
++ u64 tx_good_bytes;
++ u64 tx_bad_bytes;
++ unsigned long tx_packets;
++ unsigned long tx_bad;
++ unsigned long tx_pause;
++ unsigned long tx_control;
++ unsigned long tx_unicast;
++ unsigned long tx_multicast;
++ unsigned long tx_broadcast;
++ unsigned long tx_lt64;
++ unsigned long tx_64;
++ unsigned long tx_65_to_127;
++ unsigned long tx_128_to_255;
++ unsigned long tx_256_to_511;
++ unsigned long tx_512_to_1023;
++ unsigned long tx_1024_to_15xx;
++ unsigned long tx_15xx_to_jumbo;
++ unsigned long tx_gtjumbo;
++ unsigned long tx_collision;
++ unsigned long tx_single_collision;
++ unsigned long tx_multiple_collision;
++ unsigned long tx_excessive_collision;
++ unsigned long tx_deferred;
++ unsigned long tx_late_collision;
++ unsigned long tx_excessive_deferred;
++ unsigned long tx_non_tcpudp;
++ unsigned long tx_mac_src_error;
++ unsigned long tx_ip_src_error;
++ u64 rx_bytes;
++ u64 rx_good_bytes;
++ u64 rx_bad_bytes;
++ unsigned long rx_packets;
++ unsigned long rx_good;
++ unsigned long rx_bad;
++ unsigned long rx_pause;
++ unsigned long rx_control;
++ unsigned long rx_unicast;
++ unsigned long rx_multicast;
++ unsigned long rx_broadcast;
++ unsigned long rx_lt64;
++ unsigned long rx_64;
++ unsigned long rx_65_to_127;
++ unsigned long rx_128_to_255;
++ unsigned long rx_256_to_511;
++ unsigned long rx_512_to_1023;
++ unsigned long rx_1024_to_15xx;
++ unsigned long rx_15xx_to_jumbo;
++ unsigned long rx_gtjumbo;
++ unsigned long rx_bad_lt64;
++ unsigned long rx_bad_64_to_15xx;
++ unsigned long rx_bad_15xx_to_jumbo;
++ unsigned long rx_bad_gtjumbo;
++ unsigned long rx_overflow;
++ unsigned long rx_missed;
++ unsigned long rx_false_carrier;
++ unsigned long rx_symbol_error;
++ unsigned long rx_align_error;
++ unsigned long rx_length_error;
++ unsigned long rx_internal_error;
++ unsigned long rx_good_lt64;
++};
++
++/* Number of bits used in a multicast filter hash address */
++#define EFX_MCAST_HASH_BITS 8
++
++/* Number of (single-bit) entries in a multicast filter hash */
++#define EFX_MCAST_HASH_ENTRIES (1 << EFX_MCAST_HASH_BITS)
++
++/* An Efx multicast filter hash */
++union efx_multicast_hash {
++ u8 byte[EFX_MCAST_HASH_ENTRIES / sizeof(u8)];
++ efx_oword_t oword[EFX_MCAST_HASH_ENTRIES / sizeof(efx_oword_t)];
++};
++
++/* Efx Error condition statistics */
++struct efx_nic_errors {
++ atomic_t missing_event;
++ atomic_t rx_reset;
++ atomic_t rx_desc_fetch;
++ atomic_t tx_desc_fetch;
++ atomic_t spurious_tx;
++
++#ifdef CONFIG_SFC_DEBUGFS
++ struct dentry *debug_dir;
++#endif
++};
+
-+ return 0;
-+}
++/**
++ * struct efx_nic - an Efx NIC
++ * @name: Device name (net device name or bus id before net device registered)
++ * @pci_dev: The PCI device
++ * @pci_dev2: The secondary PCI device if present
++ * @type: Controller type attributes
++ * @dma_mask: DMA mask
++ * @legacy_irq: IRQ number
++ * @workqueue: Workqueue for resets, port reconfigures and the HW monitor
++ * @refill_workqueue: RX refill workqueue
++ * @reset_work: Scheduled reset workitem
++ * @monitor_work: Hardware monitor workitem
++ * @membase_phys: Memory BAR value as physical address
++ * @membase: Memory BAR value
++ * @biu_lock: BIU (bus interface unit) lock
++ * @interrupt_mode: Interrupt mode
++ * @is_asic: Is ASIC (else FPGA)
++ * @is_10g: Is set to 10G (else 1G)
++ * @external_sram_cfg: Size and number of banks of external SRAM
++ * @i2c: I2C interface
++ * @board_info: Board-level information
++ * @state: Device state flag. Can only be manipulated when both
++ * suspend_lock and rtnl_lock are held. Can be read when
++ * either is held.
++ * @reset_pending: Pending reset method (normally RESET_TYPE_NONE)
++ * @suspend_lock: Device suspend lock. This must not be acquired with
++ * rtnl_lock held.
++ * @tx_queue: TX DMA queues
++ * @rx_queue: RX DMA queues
++ * @channel: Channels
++ * @rss_queues: Number of RSS queues
++ * @rx_buffer_len: RX buffer length
++ * @rx_buffer_order: Order (log2) of number of pages for each RX buffer
++ * @errors: Error condition stats
++ * @irq_status: Interrupt status buffer
++ * @last_irq_cpu: Last CPU to handle interrupt.
++ * This register is written with the SMP processor ID whenever an
++ * interrupt is handled. It is used by falcon_test_interrupt()
++ * to verify that an interrupt has occurred.
++ * @spi_flash: SPI flash device
++ * This field will be %NULL if no flash device is present.
++ * @spi_eeprom: SPI EEPROM device
++ * This field will be %NULL if no EEPROM device is present.
++ * @spi_lock: SPI bus lock
++ * @n_rx_nodesc_drop_cnt: RX no descriptor drop count
++ * @nic_data: Hardware dependant state
++ * @mac_lock: MAC access lock. Protects efx->port_enabled/net_dev_registered
++ * and efx_reconfigure_port()
++ * @port_enabled: Port enabled indicator.
++ * Serialises efx_stop_all and efx_start_all with kernel interfaces.
++ * Safe to read under the rtnl_lock, mac_lock, or netif_tx_lock, but
++ * all three must be held to modify it.
++ * @net_dev_registered: Port is registered with operating system.
++ * @port_initialized: Port initialized?
++ * @net_dev: Operating system network device. Consider holding the rtnl lock
++ * @rx_checksum_enabled: RX checksumming enabled
++ * @netif_stop_count: Port stop count
++ * @netif_stop_lock: Port stop lock
++ * @mac_stats: MAC statistics
++ * @stats: Net device statistics.
++ * Hardware-specific code fills in @mac_stats, which provides a
++ * detailed breakdown. Generic code aggregates these statistics
++ * into a standard &struct net_device_stats.
++ * @stats_buffer: DMA buffer for statistics
++ * @stats_lock: Statistics update lock
++ * @mac_op: MAC interface
++ * @mac_address: Permanent MAC address
++ * @phy_type: PHY type
++ * @phy_lock: PHY access lock
++ * @phy_op: PHY interface
++ * @phy_data: PHY private data (including PHY-specific stats)
++ * @mii: PHY interface
++ * @phy_powered: PHY power state
++ * @tx_disabled: PHY transmitter turned off
++ * @link_up: Link status
++ * @link_options: Link options (MII/GMII format)
++ * @n_link_state_changes: Number of times the link has changed state
++ * @promiscuous: Promiscuous flag. Protected by netif_tx_lock.
++ * @multicast_hash: Multicast hash table
++ * @flow_control: Flow control flags - separate RX/TX so can't use link_options
++ * @reconfigure_work: work item for dealing with PHY events
++ * @loopback_mode: Loopback status
++ * @loopback_modes: Supported loopback mode bitmask
++ * @loopback_selftest: Offline self-test private state
++ * @silicon_rev: Silicon revision description for driverlink
++ * @dl_info: Linked list of hardware parameters exposed through driverlink
++ * @dl_node: Driverlink port list
++ * @dl_device_list: Driverlink device list
++ * @dl_cb: Driverlink callbacks table
++ * @dl_cb_dev: Driverlink callback owner devices
++ * @debug_dir: NIC debugfs directory
++ * @debug_symlink: NIC debugfs sym-link (nic_eth\%d)
++ * @debug_port_dir: Port debugfs directory
++ * @debug_port_symlink: Port debugfs sym-link (if_eth\%d)
++ *
++ * The @priv field of the corresponding &struct net_device points to
++ * this.
++ */
++struct efx_nic {
++ char name[IFNAMSIZ];
++ struct pci_dev *pci_dev;
++ struct pci_dev *pci_dev2;
++#if !defined(EFX_USE_PCI_DEV_REVISION)
++ u8 revision;
++#endif
++ const struct efx_nic_type *type;
++ dma_addr_t dma_mask;
++ int legacy_irq;
++ struct workqueue_struct *workqueue;
++#if !defined(EFX_USE_CANCEL_DELAYED_WORK_SYNC)
++ /* Since we can't use cancel_delayed_work_sync efx_reset() has to
++ * flush efx->workqueue to serialise against efx_reconfigure_port
++ * and efx_monitor. So it can't also run on workqueue */
++ struct workqueue_struct *reset_workqueue;
++#endif
++ struct workqueue_struct *refill_workqueue;
++ struct work_struct reset_work;
++ struct delayed_work monitor_work;
++ unsigned long membase_phys;
++ void __iomem *membase;
++ spinlock_t biu_lock;
++ enum efx_int_mode interrupt_mode;
++ unsigned int is_asic:1;
++ unsigned int is_10g:1;
++ int external_sram_cfg;
++
++ struct efx_i2c_interface i2c;
++ struct efx_board board_info;
++
++ enum nic_state state;
++ enum reset_type reset_pending;
++
++ struct semaphore suspend_lock;
++
++ struct efx_tx_queue tx_queue[EFX_MAX_TX_QUEUES];
++ struct efx_rx_queue rx_queue[EFX_MAX_RX_QUEUES];
++ struct efx_channel channel[EFX_MAX_CHANNELS];
++
++ int rss_queues;
++ unsigned int rx_buffer_len;
++ unsigned int rx_buffer_order;
++
++ struct efx_nic_errors errors;
++
++ struct efx_buffer irq_status;
++ volatile signed int last_irq_cpu;
++
++ struct efx_spi_device *spi_flash;
++ struct efx_spi_device *spi_eeprom;
++ struct mutex spi_lock;
++
++ unsigned n_rx_nodesc_drop_cnt;
++
++ void *nic_data;
++
++ struct mutex mac_lock;
++ int port_enabled;
++ int net_dev_registered;
++
++ int port_initialized;
++ struct net_device *net_dev;
++ int rx_checksum_enabled;
++
++ atomic_t netif_stop_count;
++ spinlock_t netif_stop_lock;
++
++ struct efx_mac_stats mac_stats;
++ struct net_device_stats stats;
++ struct efx_buffer stats_buffer;
++ spinlock_t stats_lock;
+
-+#endif /* !CONFIG_XEN */
++ struct efx_mac_operations *mac_op;
++ unsigned char mac_address[ETH_ALEN];
+
-+static int gnttab_expand(unsigned int req_entries)
-+{
-+ int rc;
-+ unsigned int cur, extra;
++ enum phy_type phy_type;
++ spinlock_t phy_lock;
++ struct efx_phy_operations *phy_op;
++ void *phy_data;
++ struct mii_if_info mii;
++ unsigned phy_powered;
++ unsigned tx_disabled;
+
-+ cur = nr_grant_frames;
-+ extra = ((req_entries + (GREFS_PER_GRANT_FRAME-1)) /
-+ GREFS_PER_GRANT_FRAME);
-+ if (cur + extra > max_nr_grant_frames())
-+ return -ENOSPC;
++ int link_up;
++ unsigned int link_options;
++ unsigned int n_link_state_changes;
+
-+ if ((rc = gnttab_map(cur, cur + extra - 1)) == 0)
-+ rc = grow_gnttab_list(extra);
++ int promiscuous;
++ union efx_multicast_hash multicast_hash;
++ enum efx_fc_type flow_control;
++ struct work_struct reconfigure_work;
+
-+ return rc;
-+}
++ enum efx_loopback_mode loopback_mode;
++ unsigned int loopback_modes;
++ unsigned int startup_loopbacks;
+
-+int __devinit gnttab_init(void)
-+{
-+ int i;
-+ unsigned int max_nr_glist_frames;
-+ unsigned int nr_init_grefs;
++ void *loopback_selftest;
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ const char *silicon_rev;
++ struct efx_dl_device_info *dl_info;
++ struct list_head dl_node;
++ struct list_head dl_device_list;
++ struct efx_dl_callbacks dl_cb;
++ struct efx_dl_cb_devices dl_cb_dev;
+
-+ nr_grant_frames = 1;
-+ boot_max_nr_grant_frames = __max_nr_grant_frames();
++#ifdef CONFIG_SFC_DEBUGFS
++ struct dentry *debug_dir;
++ struct dentry *debug_symlink;
++ struct dentry *debug_port_dir;
++ struct dentry *debug_port_symlink;
++#endif
++};
+
-+ /* Determine the maximum number of frames required for the
-+ * grant reference free list on the current hypervisor.
-+ */
-+ max_nr_glist_frames = (boot_max_nr_grant_frames *
-+ GREFS_PER_GRANT_FRAME /
-+ (PAGE_SIZE / sizeof(grant_ref_t)));
++/**
++ * struct efx_nic_type - Efx device type definition
++ * @is_dual_func: Is dual-function (else single-function)
++ * @mem_bar: Memory BAR number
++ * @mem_map_size: Memory BAR mapped size
++ * @txd_ptr_tbl_base: TX descriptor ring base address
++ * @rxd_ptr_tbl_base: RX descriptor ring base address
++ * @buf_tbl_base: Buffer table base address
++ * @evq_ptr_tbl_base: Event queue pointer table base address
++ * @evq_rptr_tbl_base: Event queue read-pointer table base address
++ * @txd_ring_mask: TX descriptor ring size - 1 (must be a power of two - 1)
++ * @rxd_ring_mask: RX descriptor ring size - 1 (must be a power of two - 1)
++ * @evq_size: Event queue size (must be a power of two)
++ * @max_dma_mask: Maximum possible DMA mask
++ * @tx_dma_mask: TX DMA mask
++ * @bug5391_mask: Address mask for bug 5391 workaround
++ * @rx_xoff_thresh: RX FIFO XOFF watermark (bytes)
++ * @rx_xon_thresh: RX FIFO XON watermark (bytes)
++ * @rx_buffer_padding: Padding added to each RX buffer
++ * @max_interrupt_mode: Highest capability interrupt mode supported
++ * from &enum efx_init_mode.
++ * @phys_addr_channels: Number of channels with physically addressed
++ * descriptors
++ */
++struct efx_nic_type {
++ unsigned int is_dual_func;
++ unsigned int mem_bar;
++ unsigned int mem_map_size;
++ unsigned int txd_ptr_tbl_base;
++ unsigned int rxd_ptr_tbl_base;
++ unsigned int buf_tbl_base;
++ unsigned int evq_ptr_tbl_base;
++ unsigned int evq_rptr_tbl_base;
++
++ unsigned int txd_ring_mask;
++ unsigned int rxd_ring_mask;
++ unsigned int evq_size;
++ dma_addr_t max_dma_mask;
++ unsigned int tx_dma_mask;
++ unsigned bug5391_mask;
++
++ int rx_xoff_thresh;
++ int rx_xon_thresh;
++ unsigned int rx_buffer_padding;
++ unsigned int max_interrupt_mode;
++ unsigned int phys_addr_channels;
++};
++
++/**************************************************************************
++ *
++ * Prototypes and inline functions
++ *
++ *************************************************************************/
++
++/* Iterate over all used channels */
++#define efx_for_each_channel(_channel, _efx) \
++ for (_channel = &_efx->channel[0]; \
++ _channel < &_efx->channel[EFX_MAX_CHANNELS]; \
++ _channel++) \
++ if (!_channel->used_flags) \
++ continue; \
++ else
+
-+ gnttab_list = kmalloc(max_nr_glist_frames * sizeof(grant_ref_t *),
-+ GFP_KERNEL);
-+ if (gnttab_list == NULL)
-+ return -ENOMEM;
++/* Iterate over all used channels with interrupts */
++#define efx_for_each_channel_with_interrupt(_channel, _efx) \
++ for (_channel = &_efx->channel[0]; \
++ _channel < &_efx->channel[EFX_MAX_CHANNELS]; \
++ _channel++) \
++ if (!(_channel->used_flags && _channel->has_interrupt)) \
++ continue; \
++ else
+
-+ for (i = 0; i < nr_grant_frames; i++) {
-+ gnttab_list[i] = (grant_ref_t *)__get_free_page(GFP_KERNEL);
-+ if (gnttab_list[i] == NULL)
-+ goto ini_nomem;
-+ }
++/* Iterate over all used TX queues */
++#define efx_for_each_tx_queue(_tx_queue, _efx) \
++ for (_tx_queue = &_efx->tx_queue[0]; \
++ _tx_queue < &_efx->tx_queue[EFX_MAX_TX_QUEUES]; \
++ _tx_queue++) \
++ if (!_tx_queue->used) \
++ continue; \
++ else
+
-+ if (gnttab_resume() < 0)
-+ return -ENODEV;
++/* Iterate over all TX queues belonging to a channel */
++#define efx_for_each_channel_tx_queue(_tx_queue, _channel) \
++ for (_tx_queue = &_channel->efx->tx_queue[0]; \
++ _tx_queue < &_channel->efx->tx_queue[EFX_MAX_TX_QUEUES]; \
++ _tx_queue++) \
++ if ((!_tx_queue->used) || \
++ (_tx_queue->channel != _channel)) \
++ continue; \
++ else
+
-+ nr_init_grefs = nr_grant_frames * GREFS_PER_GRANT_FRAME;
++/* Iterate over all used RX queues */
++#define efx_for_each_rx_queue(_rx_queue, _efx) \
++ for (_rx_queue = &_efx->rx_queue[0]; \
++ _rx_queue < &_efx->rx_queue[EFX_MAX_RX_QUEUES]; \
++ _rx_queue++) \
++ if (!_rx_queue->used) \
++ continue; \
++ else
+
-+ for (i = NR_RESERVED_ENTRIES; i < nr_init_grefs - 1; i++)
-+ gnttab_entry(i) = i + 1;
++/* Iterate over all RX queues belonging to a channel */
++#define efx_for_each_channel_rx_queue(_rx_queue, _channel) \
++ for (_rx_queue = &_channel->efx->rx_queue[0]; \
++ _rx_queue < &_channel->efx->rx_queue[EFX_MAX_RX_QUEUES]; \
++ _rx_queue++) \
++ if ((!_rx_queue->used) || \
++ (_rx_queue->channel != _channel)) \
++ continue; \
++ else
+
-+ gnttab_entry(nr_init_grefs - 1) = GNTTAB_LIST_END;
-+ gnttab_free_count = nr_init_grefs - NR_RESERVED_ENTRIES;
-+ gnttab_free_head = NR_RESERVED_ENTRIES;
++/* Name formats */
++#define EFX_CHANNEL_NAME(_channel) "channel%d", _channel->channel
++#define EFX_TX_QUEUE_NAME(_tx_queue) "txq%d", _tx_queue->queue
++#define EFX_RX_QUEUE_NAME(_rx_queue) "rxq%d", _rx_queue->queue
+
-+ return 0;
++/* Returns a pointer to the specified receive buffer in the RX
++ * descriptor queue.
++ */
++static inline struct efx_rx_buffer *efx_rx_buffer(struct efx_rx_queue *rx_queue,
++ unsigned int index)
++{
++ return (&rx_queue->buffer[index]);
++}
+
-+ ini_nomem:
-+ for (i--; i >= 0; i--)
-+ free_page((unsigned long)gnttab_list[i]);
-+ kfree(gnttab_list);
-+ return -ENOMEM;
++/* Set bit in a little-endian bitfield */
++static inline void set_bit_le(int nr, unsigned char *addr)
++{
++ addr[nr / 8] |= (1 << (nr % 8));
+}
+
-+#ifdef CONFIG_XEN
-+core_initcall(gnttab_init);
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/hypervisor_sysfs.c linux-2.6.18-xen.hg/drivers/xen/core/hypervisor_sysfs.c
---- linux-2.6.18/drivers/xen/core/hypervisor_sysfs.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/hypervisor_sysfs.c 2007-12-23 11:15:33.661248285 +0100
-@@ -0,0 +1,57 @@
-+/*
-+ * copyright (c) 2006 IBM Corporation
-+ * Authored by: Mike D. Day <ncmike@us.ibm.com>
++/* Clear bit in a little-endian bitfield */
++static inline void clear_bit_le(int nr, unsigned char *addr)
++{
++ addr[nr / 8] &= ~(1 << (nr % 8));
++}
++
++
++/**
++ * EFX_MAX_FRAME_LEN - calculate maximum frame length
+ *
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License version 2 as
-+ * published by the Free Software Foundation.
++ * This calculates the maximum frame length that will be used for a
++ * given MTU. The frame length will be equal to the MTU plus a
++ * constant amount of header space and padding. This is the quantity
++ * that the net driver will program into the MAC as the maximum frame
++ * length.
++ *
++ * The 10G MAC used in Falcon requires 8-byte alignment on the frame
++ * length, so we round up to the nearest 8.
+ */
++#define EFX_MAX_FRAME_LEN(mtu) \
++ ((((mtu) + ETH_HLEN + VLAN_HLEN + 4/* FCS */) + 7) & ~7)
+
-+#include <linux/kernel.h>
-+#include <linux/module.h>
-+#include <linux/kobject.h>
-+#include <xen/hypervisor_sysfs.h>
-+#include <asm/hypervisor.h>
+
-+static ssize_t hyp_sysfs_show(struct kobject *kobj,
-+ struct attribute *attr,
-+ char *buffer)
++#endif /* EFX_NET_DRIVER_H */
+--- linux-2.6.18.8/drivers/net/sfc/null_phy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/null_phy.c 2008-05-19 00:33:29.317836041 +0300
+@@ -0,0 +1,62 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "efx.h"
++#include "falcon.h"
++#include "gmii.h"
++#include "phy.h"
++
++static int falcon_null_phy_check_hw(struct efx_nic *efx)
+{
-+ struct hyp_sysfs_attr *hyp_attr;
-+ hyp_attr = container_of(attr, struct hyp_sysfs_attr, attr);
-+ if (hyp_attr->show)
-+ return hyp_attr->show(hyp_attr, buffer);
++ int link_ok = falcon_xaui_link_ok(efx);
++
++ /* Generate PHY event that a PHY would have generated */
++ if (link_ok != efx->link_up) {
++ efx->link_up = link_ok;
++ efx->mac_op->fake_phy_event(efx);
++ }
++
+ return 0;
+}
+
-+static ssize_t hyp_sysfs_store(struct kobject *kobj,
-+ struct attribute *attr,
-+ const char *buffer,
-+ size_t len)
++static void falcon_null_phy_reconfigure(struct efx_nic *efx)
+{
-+ struct hyp_sysfs_attr *hyp_attr;
-+ hyp_attr = container_of(attr, struct hyp_sysfs_attr, attr);
-+ if (hyp_attr->store)
-+ return hyp_attr->store(hyp_attr, buffer, len);
-+ return 0;
++ /* CX4 is always 10000FD only */
++ efx->link_options = GM_LPA_10000FULL;
++
++ falcon_null_phy_check_hw(efx);
+}
+
-+static struct sysfs_ops hyp_sysfs_ops = {
-+ .show = hyp_sysfs_show,
-+ .store = hyp_sysfs_store,
++struct efx_phy_operations falcon_null_phy_ops = {
++ .reconfigure = falcon_null_phy_reconfigure,
++ .check_hw = falcon_null_phy_check_hw,
++ .fini = efx_port_dummy_op_void,
++ .clear_interrupt = efx_port_dummy_op_void,
++ .init = efx_port_dummy_op_int,
++ .reset_xaui = efx_port_dummy_op_void,
++ .mmds = 0,
++ .loopbacks = 0,
++ .startup_loopback = 0,
+};
+--- linux-2.6.18.8/drivers/net/sfc/phy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/phy.c 2008-05-19 00:33:29.317836041 +0300
+@@ -0,0 +1,28 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static struct kobj_type hyp_sysfs_kobj_type = {
-+ .sysfs_ops = &hyp_sysfs_ops,
-+};
++#include "net_driver.h"
++#include "phy.h"
+
-+static int __init hypervisor_subsys_init(void)
-+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
+--- linux-2.6.18.8/drivers/net/sfc/phy.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/phy.h 2008-05-19 00:33:29.317836041 +0300
+@@ -0,0 +1,90 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ hypervisor_subsys.kset.kobj.ktype = &hyp_sysfs_kobj_type;
-+ return 0;
-+}
++#ifndef EFX_PHY_H
++#define EFX_PHY_H
+
-+device_initcall(hypervisor_subsys_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/machine_kexec.c linux-2.6.18-xen.hg/drivers/xen/core/machine_kexec.c
---- linux-2.6.18/drivers/xen/core/machine_kexec.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/machine_kexec.c 2007-12-23 11:15:33.664581791 +0100
-@@ -0,0 +1,189 @@
-+/*
-+ * drivers/xen/core/machine_kexec.c
-+ * handle transition of Linux booting another kernel
++/****************************************************************************
++ * 10Xpress (SFX7101) PHY
+ */
++extern struct efx_phy_operations falcon_tenxpress_phy_ops;
+
-+#include <linux/kexec.h>
-+#include <xen/interface/kexec.h>
-+#include <linux/mm.h>
-+#include <linux/bootmem.h>
++enum tenxpress_state {
++ TENXPRESS_STATUS_OFF = 0,
++ TENXPRESS_STATUS_OTEMP = 1,
++ TENXPRESS_STATUS_NORMAL = 2,
++};
+
-+extern void machine_kexec_setup_load_arg(xen_kexec_image_t *xki,
-+ struct kimage *image);
++extern void tenxpress_set_state(struct efx_nic *efx,
++ enum tenxpress_state state);
++extern void tenxpress_phy_blink(struct efx_nic *efx, int blink);
++extern void tenxpress_crc_err(struct efx_nic *efx);
+
-+static int __initdata xen_max_nr_phys_cpus;
-+static struct resource xen_hypervisor_res;
-+static struct resource *xen_phys_cpus;
++/****************************************************************************
++ * Marvell 88E1111 "Alaska" PHY control
++ */
++extern struct efx_phy_operations alaska_phy_operations;
+
-+void __init xen_machine_kexec_setup_resources(void)
-+{
-+ xen_kexec_range_t range;
-+ struct resource *res;
-+ int k = 0;
++/****************************************************************************
++* Exported functions from the driver for Transwitch CX4 retimer
++*/
++extern struct efx_phy_operations falcon_txc_phy_ops;
+
-+ if (!is_initial_xendomain())
-+ return;
++#define TXC_GPIO_DIR_INPUT (0)
++#define TXC_GPIO_DIR_OUTPUT (1)
+
-+ /* determine maximum number of physical cpus */
++extern void txc_set_gpio_dir(struct efx_nic *p, int pin, int dir);
++extern void txc_set_gpio_val(struct efx_nic *p, int pin, int val);
+
-+ while (1) {
-+ memset(&range, 0, sizeof(range));
-+ range.range = KEXEC_RANGE_MA_CPU;
-+ range.nr = k;
++/****************************************************************************
++ * Exported functions from the driver for PMC PM8358 PHY
++ */
++extern struct efx_phy_operations falcon_pm8358_phy_ops;
+
-+ if(HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
-+ break;
++/****************************************************************************
++ * Exported functions from the driver for XFP optical PHYs
++ */
++extern struct efx_phy_operations falcon_xfp_phy_ops;
+
-+ k++;
-+ }
++/* The QUAKE XFP PHY provides various H/W control states for LEDs */
++#define QUAKE_LED_LINK_INVAL (0)
++#define QUAKE_LED_LINK_STAT (1)
++#define QUAKE_LED_LINK_ACT (2)
++#define QUAKE_LED_LINK_ACTSTAT (3)
++#define QUAKE_LED_OFF (4)
++#define QUAKE_LED_ON (5)
++#define QUAKE_LED_LINK_INPUT (6) /* Pin is an input. */
++/* What link the LED tracks */
++#define QUAKE_LED_TXLINK (0)
++#define QUAKE_LED_RXLINK (8)
+
-+ if (k == 0)
-+ return;
++extern void xfp_set_led(struct efx_nic *p, int led, int state);
+
-+ xen_max_nr_phys_cpus = k;
++/****************************************************************************
++ * NULL PHY ops
++ */
++extern struct efx_phy_operations falcon_null_phy_ops;
+
-+ /* allocate xen_phys_cpus */
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/pm8358_phy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/pm8358_phy.c 2008-05-19 00:33:29.317836041 +0300
+@@ -0,0 +1,206 @@
++/* Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++/*
++ * Driver for PMC-Sierra PM8358 XAUI PHY
++ */
+
-+ xen_phys_cpus = alloc_bootmem_low(k * sizeof(struct resource));
-+ BUG_ON(xen_phys_cpus == NULL);
++#include <linux/delay.h>
++#include "efx.h"
++#include "gmii.h"
++#include "mdio_10g.h"
++#include "phy.h"
++
++#define PM8358_REQUIRED_DEVS (MDIO_MMDREG_DEVS0_DTEXS)
++#define PM8358_LOOPBACKS (1 << LOOPBACK_PHY)
++
++/* PHY-specific definitions */
++/* Master ID and Global Performance Monitor Update */
++#define PMC_MASTER_REG (0xd000)
++/* Analog TX/RX settings under software control */
++#define PMC_MASTER_ANLG_CTRL (1 << 11)
++
++#define PMC_MCONF2_REG (0xd002)
++/* Drive TX off centre of data eye (1) vs. clock edge (0) */
++#define PMC_MCONF2_TEDGE (1 << 2)
++/* Drive RX off centre of data eye (1) vs. clock edge (0) */
++#define PMC_MCONF2_REDGE (1 << 3)
++
++/* Analog RX settings */
++#define PMC_ANALOG_RX_CFG0 (0xd025)
++#define PMC_ANALOG_RX_CFG1 (0xd02d)
++#define PMC_ANALOG_RX_CFG2 (0xd035)
++#define PMC_ANALOG_RX_CFG3 (0xd03d)
++#define PMC_ANALOG_RX_TERM (1 << 15) /* Bit 15 of RX CFG: 0 for 100 ohms
++ float, 1 for 50 to 1.2V */
++#define PMC_ANALOG_RX_EQ_MASK (3 << 8)
++#define PMC_ANALOG_RX_EQ_NONE (0 << 8)
++#define PMC_ANALOG_RX_EQ_HALF (1 << 8)
++#define PMC_ANALOG_RX_EQ_FULL (2 << 8)
++#define PMC_ANALOG_RX_EQ_RSVD (3 << 8)
++
++/* Reset the DTE XS MMD. */
++#define PMC_MAX_RESET_TIME 500
++#define PMC_RESET_WAIT 10
++
++static int pmc_reset_phy(struct efx_nic *efx)
++{
++ int rc = mdio_clause45_reset_mmd(efx, MDIO_MMD_DTEXS,
++ PMC_MAX_RESET_TIME / PMC_RESET_WAIT,
++ PMC_RESET_WAIT);
++ if (rc >= 0) {
++ EFX_TRACE(efx, "PMC8358: came out of reset with "
++ "%d0 ms left\n", rc);
++ rc = 0;
++ } else {
++ EFX_ERR(efx, "PMC8358: reset timed out!\n");
++ }
++ return rc;
++}
+
-+ /* fill in xen_phys_cpus with per-cpu crash note information */
+
-+ for (k = 0; k < xen_max_nr_phys_cpus; k++) {
-+ memset(&range, 0, sizeof(range));
-+ range.range = KEXEC_RANGE_MA_CPU;
-+ range.nr = k;
++static void pmc_full_rx_eq(struct efx_nic *efx)
++{
++ int i, reg;
+
-+ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
-+ goto err;
++ /* Enable software control of analog settings */
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, PMC_MASTER_REG);
++ reg |= PMC_MASTER_ANLG_CTRL;
+
-+ res = xen_phys_cpus + k;
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, PMC_MASTER_REG, reg);
+
-+ memset(res, 0, sizeof(*res));
-+ res->name = "Crash note";
-+ res->start = range.start;
-+ res->end = range.start + range.size - 1;
-+ res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
++ /* Turn RX eq on full for all channels. */
++ for (i = 0; i < 3; i++) {
++ /* The analog CFG registers are evenly spaced 8 apart */
++ u16 addr = PMC_ANALOG_RX_CFG0 + 8 * i;
++
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, addr);
++ reg = (reg & ~PMC_ANALOG_RX_EQ_MASK) | PMC_ANALOG_RX_EQ_FULL;
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, addr, reg);
+ }
++}
+
-+ /* fill in xen_hypervisor_res with hypervisor machine address range */
++static void pmc_set_data_edges(struct efx_nic *efx)
++{
++ int reg;
++ /* Set TEDGE, clear REDGE */
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, PMC_MCONF2_REG);
++ reg &= ~PMC_MCONF2_REDGE;
++ reg |= PMC_MCONF2_TEDGE;
+
-+ memset(&range, 0, sizeof(range));
-+ range.range = KEXEC_RANGE_MA_XEN;
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_DTEXS, PMC_MCONF2_REG, reg);
++}
+
-+ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
-+ goto err;
++static int pm8358_phy_init(struct efx_nic *efx)
++{
++ u32 devid;
++ int rc;
+
-+ xen_hypervisor_res.name = "Hypervisor code and data";
-+ xen_hypervisor_res.start = range.start;
-+ xen_hypervisor_res.end = range.start + range.size - 1;
-+ xen_hypervisor_res.flags = IORESOURCE_BUSY | IORESOURCE_MEM;
++ /* The GLB_CTL reset line has been pulled before this is called,
++ * and it may take up to 5ms for the PLL's to synchronise after
++ * this is done. Best to wait 10ms here */
++ schedule_timeout_uninterruptible(HZ / 100);
+
-+ /* fill in crashk_res if range is reserved by hypervisor */
++ rc = pmc_reset_phy(efx);
++ if (rc < 0)
++ return rc;
+
-+ memset(&range, 0, sizeof(range));
-+ range.range = KEXEC_RANGE_MA_CRASH;
++ /* Check that all the MMDs we expect are present and responding. We
++ * expect faults on some if the link is down, but not on the PHY XS */
++ rc = mdio_clause45_check_mmds(efx, PM8358_REQUIRED_DEVS, 0);
++ if (rc < 0)
++ return rc;
+
-+ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
-+ return;
++ devid = mdio_clause45_read_id(efx, MDIO_MMD_DTEXS);
++ EFX_LOG(efx, "PM8358: PHY ID reg %x (OUI %x model %x revision"
++ " %x)\n", devid, MDIO_ID_OUI(devid), MDIO_ID_MODEL(devid),
++ MDIO_ID_REV(devid));
+
-+ if (range.size) {
-+ crashk_res.start = range.start;
-+ crashk_res.end = range.start + range.size - 1;
-+ }
++ /* Turn on full RX equalisation */
++ pmc_full_rx_eq(efx);
+
-+ return;
++ /* Adjust RX and TX data edge position */
++ pmc_set_data_edges(efx);
+
-+ err:
-+ /*
-+ * It isn't possible to free xen_phys_cpus this early in the
-+ * boot. Failure at this stage is unexpected and the amount of
-+ * memory is small therefore we tolerate the potential leak.
-+ */
-+ xen_max_nr_phys_cpus = 0;
-+ return;
++ EFX_LOG(efx, "PM8358: PHY init successful.\n");
++ return rc;
+}
+
-+void __init xen_machine_kexec_register_resources(struct resource *res)
++static int pm8358_link_ok(struct efx_nic *efx)
+{
-+ int k;
-+
-+ request_resource(res, &xen_hypervisor_res);
++ return mdio_clause45_links_ok(efx, PM8358_REQUIRED_DEVS);
++}
+
-+ for (k = 0; k < xen_max_nr_phys_cpus; k++)
-+ request_resource(&xen_hypervisor_res, xen_phys_cpus + k);
++static int pm8358_phy_check_hw(struct efx_nic *efx)
++{
++ int rc = 0;
++ int link_up = pm8358_link_ok(efx);
++ /* Simulate a PHY event if link state has changed */
++ if (link_up != efx->link_up) {
++ efx->link_up = link_up;
++ efx->mac_op->fake_phy_event(efx);
++ }
+
++ return rc;
+}
+
-+static void setup_load_arg(xen_kexec_image_t *xki, struct kimage *image)
++static void pm8358_phy_reconfigure(struct efx_nic *efx)
+{
-+ machine_kexec_setup_load_arg(xki, image);
++ int phy_id = efx->mii.phy_id;
++ int ctrl;
++ /* Handle DTE loopback */
++ ctrl = mdio_clause45_read(efx, phy_id, MDIO_MMD_DTEXS,
++ MDIO_MMDREG_CTRL1);
++ if (efx->loopback_mode == LOOPBACK_PHY) {
++ EFX_TRACE(efx, "PM8358: setting DTE loopback\n");
++ ctrl |= (1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++ } else {
++ if (ctrl & (1 << MDIO_MMDREG_CTRL1_LBACK_LBN))
++ EFX_TRACE(efx,
++ "PM8358: clearing DTE loopback\n");
++ ctrl &= ~(1 << MDIO_MMDREG_CTRL1_LBACK_LBN);
++ }
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_DTEXS,
++ MDIO_MMDREG_CTRL1, ctrl);
++
++ efx->link_up = pm8358_link_ok(efx);
++ efx->link_options = GM_LPA_10000FULL;
++}
++
++struct efx_phy_operations falcon_pm8358_phy_ops = {
++ .init = pm8358_phy_init,
++ .reconfigure = pm8358_phy_reconfigure,
++ .check_hw = pm8358_phy_check_hw,
++ .fini = efx_port_dummy_op_void,
++ .clear_interrupt = efx_port_dummy_op_void,
++ .reset_xaui = efx_port_dummy_op_void,
++ .mmds = PM8358_REQUIRED_DEVS,
++ .loopbacks = PM8358_LOOPBACKS,
++ .startup_loopback = LOOPBACK_PHY,
++};
+--- linux-2.6.18.8/drivers/net/sfc/rx.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/rx.c 2008-05-19 00:33:29.321836271 +0300
+@@ -0,0 +1,798 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ xki->indirection_page = image->head;
-+ xki->start_address = image->start;
-+}
++#include <linux/socket.h>
++#include <linux/in.h>
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/udp.h>
++#include <net/ip.h>
++#include <net/checksum.h>
++#include "net_driver.h"
++#include "rx.h"
++#include "efx.h"
++#include "falcon.h"
++#include "selftest.h"
++#include "workarounds.h"
++
++
++/* Number of RX descriptors pushed at once. */
++#define EFX_RX_BATCH 8
++
++/* Size of buffer allocated for skb header area. */
++#define EFX_SKB_HEADERS 64u
+
+/*
-+ * Load the image into xen so xen can kdump itself
-+ * This might have been done in prepare, but prepare
-+ * is currently called too early. It might make sense
-+ * to move prepare, but for now, just add an extra hook.
++ * rx_alloc_method - RX buffer allocation method
++ *
++ * This driver supports two methods for allocating and using RX buffers:
++ * each RX buffer may be backed by an skb or by an order-n page.
++ *
++ * When LRO is in use then the second method has a lower overhead,
++ * since we don't have to allocate then free skbs on reassembled frames.
++ *
++ * Values:
++ * - RX_ALLOC_METHOD_AUTO = 0
++ * - RX_ALLOC_METHOD_SKB = 1
++ * - RX_ALLOC_METHOD_PAGE = 2
++ *
++ * The heuristic for %RX_ALLOC_METHOD_AUTO is a simple hysteresis count
++ * controlled by the parameters below.
++ *
++ * - Since pushing and popping descriptors are separated by the rx_queue
++ * size, so the watermarks should be ~rxd_size.
++ * - The performance win by using page-based allocation for LRO is less
++ * than the performance hit of using page-based allocation of non-LRO,
++ * so the watermarks should reflect this.
++ *
++ * Per channel we maintain a single variable, updated by each channel:
++ *
++ * rx_alloc_level += (lro_performed ? RX_ALLOC_FACTOR_LRO :
++ * RX_ALLOC_FACTOR_SKB)
++ * Per NAPI poll interval, we constrain rx_alloc_level to 0..MAX (which
++ * limits the hysteresis), and update the allocation strategy:
++ *
++ * rx_alloc_method = (rx_alloc_level > RX_ALLOC_LEVEL_LRO ?
++ * RX_ALLOC_METHOD_PAGE : RX_ALLOC_METHOD_SKB)
+ */
-+int xen_machine_kexec_load(struct kimage *image)
-+{
-+ xen_kexec_load_t xkl;
++static int rx_alloc_method = RX_ALLOC_METHOD_PAGE;
+
-+ memset(&xkl, 0, sizeof(xkl));
-+ xkl.type = image->type;
-+ setup_load_arg(&xkl.image, image);
-+ return HYPERVISOR_kexec_op(KEXEC_CMD_kexec_load, &xkl);
-+}
++#define RX_ALLOC_LEVEL_LRO 0x2000
++#define RX_ALLOC_LEVEL_MAX 0x3000
++#define RX_ALLOC_FACTOR_LRO 1
++#define RX_ALLOC_FACTOR_SKB -2
+
-+/*
-+ * Unload the image that was stored by machine_kexec_load()
-+ * This might have been done in machine_kexec_cleanup() but it
-+ * is called too late, and its possible xen could try and kdump
-+ * using resources that have been freed.
++/* This is the percentage fill level below which new RX descriptors
++ * will be added to the RX descriptor ring.
+ */
-+void xen_machine_kexec_unload(struct kimage *image)
-+{
-+ xen_kexec_load_t xkl;
++static unsigned int rx_refill_threshold = 90;
+
-+ memset(&xkl, 0, sizeof(xkl));
-+ xkl.type = image->type;
-+ HYPERVISOR_kexec_op(KEXEC_CMD_kexec_unload, &xkl);
-+}
++/* This is the percentage fill level to which an RX queue will be refilled
++ * when the "RX refill threshold" is reached.
++ */
++static unsigned int rx_refill_limit = 95;
+
+/*
-+ * Do not allocate memory (or fail in any way) in machine_kexec().
-+ * We are past the point of no return, committed to rebooting now.
++ * RX maximum head room required.
+ *
-+ * This has the hypervisor move to the prefered reboot CPU,
-+ * stop all CPUs and kexec. That is it combines machine_shutdown()
-+ * and machine_kexec() in Linux kexec terms.
++ * This must be at least 1 to prevent overflow and at least 2 to allow
++ * pipelined receives.
+ */
-+NORET_TYPE void machine_kexec(struct kimage *image)
++#define EFX_RXD_HEAD_ROOM 2
++
++/* Macros for zero-order pages (potentially) containing multiple RX buffers */
++#define RX_DATA_OFFSET(_data) \
++ (((unsigned long) (_data)) & (PAGE_SIZE-1))
++#define RX_BUF_OFFSET(_rx_buf) \
++ RX_DATA_OFFSET((_rx_buf)->data)
++
++#define RX_PAGE_SIZE(_efx) \
++ (PAGE_SIZE * (1u << (_efx)->rx_buffer_order))
++
++
++/**
++ * efx_init_rx_buffer_skb - create new RX buffer using skb-based allocation
++ *
++ * @rx_queue: Efx RX queue
++ * @rx_buf: RX buffer structure to populate
++ *
++ * This allocates memory for a new receive buffer, maps it for DMA,
++ * and populates a struct efx_rx_buffer with the relevant
++ * information. Return a negative error code or 0 on success.
++ */
++static inline int efx_init_rx_buffer_skb(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *rx_buf)
+{
-+ xen_kexec_exec_t xke;
++ struct efx_nic *efx = rx_queue->efx;
++ struct net_device *net_dev = efx->net_dev;
++ int skb_len = efx->rx_buffer_len;
+
-+ memset(&xke, 0, sizeof(xke));
-+ xke.type = image->type;
-+ HYPERVISOR_kexec_op(KEXEC_CMD_kexec, &xke);
-+ panic("KEXEC_CMD_kexec hypercall should not return\n");
++ rx_buf->skb = netdev_alloc_skb(net_dev, skb_len);
++ if (unlikely(!rx_buf->skb))
++ return -ENOMEM;
++
++ /* Adjust the SKB for padding and checksum */
++ skb_reserve(rx_buf->skb, NET_IP_ALIGN);
++ rx_buf->len = skb_len - NET_IP_ALIGN;
++ rx_buf->data = (char *)rx_buf->skb->data;
++
++ rx_buf->skb->ip_summed = CHECKSUM_UNNECESSARY;
++
++ /* Map for DMA */
++ rx_buf->dma_addr = pci_map_single(efx->pci_dev,
++ rx_buf->data, rx_buf->len,
++ PCI_DMA_FROMDEVICE);
++
++ if (unlikely(pci_dma_mapping_error(rx_buf->dma_addr))) {
++ /* Free the SKB */
++ dev_kfree_skb_any(rx_buf->skb);
++ rx_buf->skb = NULL;
++ return -EIO;
++ }
++
++ return 0;
+}
+
-+void machine_shutdown(void)
++/**
++ * efx_init_rx_buffer_page - create new RX buffer using page-based allocation
++ *
++ * @rx_queue: Efx RX queue
++ * @rx_buf: RX buffer structure to populate
++ *
++ * This allocates memory for a new receive buffer, maps it for DMA,
++ * and populates a struct efx_rx_buffer with the relevant
++ * information. Return a negative error code or 0 on success.
++ */
++static inline int efx_init_rx_buffer_page(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *rx_buf)
+{
-+ /* do nothing */
-+}
++ struct efx_nic *efx = rx_queue->efx;
++ int bytes, space, offset;
+
++ bytes = efx->rx_buffer_len - EFX_PAGE_IP_ALIGN;
+
-+/*
-+ * Local variables:
-+ * c-file-style: "linux"
-+ * indent-tabs-mode: t
-+ * c-indent-level: 8
-+ * c-basic-offset: 8
-+ * tab-width: 8
-+ * End:
-+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/machine_reboot.c linux-2.6.18-xen.hg/drivers/xen/core/machine_reboot.c
---- linux-2.6.18/drivers/xen/core/machine_reboot.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/machine_reboot.c 2007-12-23 11:15:33.664581791 +0100
-@@ -0,0 +1,240 @@
-+#include <linux/version.h>
-+#include <linux/kernel.h>
-+#include <linux/mm.h>
-+#include <linux/unistd.h>
-+#include <linux/module.h>
-+#include <linux/reboot.h>
-+#include <linux/sysrq.h>
-+#include <linux/stringify.h>
-+#include <linux/stop_machine.h>
-+#include <asm/irq.h>
-+#include <asm/mmu_context.h>
-+#include <xen/evtchn.h>
-+#include <asm/hypervisor.h>
-+#include <xen/xenbus.h>
-+#include <linux/cpu.h>
-+#include <xen/gnttab.h>
-+#include <xen/xencons.h>
-+#include <xen/cpu_hotplug.h>
-+#include <xen/interface/vcpu.h>
++ /* If there is space left in the previously allocated page,
++ * then use it. Otherwise allocate a new one */
++ rx_buf->page = rx_queue->buf_page;
++ if (rx_buf->page == NULL) {
++ dma_addr_t dma_addr;
+
-+#if defined(__i386__) || defined(__x86_64__)
++ rx_buf->page = alloc_pages(__GFP_COLD | __GFP_COMP | GFP_ATOMIC,
++ efx->rx_buffer_order);
++ if (unlikely(rx_buf->page == NULL))
++ return -ENOMEM;
+
-+/*
-+ * Power off function, if any
-+ */
-+void (*pm_power_off)(void);
-+EXPORT_SYMBOL(pm_power_off);
++ /* Map the entire page for DMA */
++ dma_addr = pci_map_page(efx->pci_dev, rx_buf->page,
++ 0, RX_PAGE_SIZE(efx),
++ PCI_DMA_FROMDEVICE);
+
-+void machine_emergency_restart(void)
-+{
-+ /* We really want to get pending console data out before we die. */
-+ xencons_force_flush();
-+ HYPERVISOR_shutdown(SHUTDOWN_reboot);
++ if (unlikely(pci_dma_mapping_error(dma_addr))) {
++ __free_pages(rx_buf->page, efx->rx_buffer_order);
++ rx_buf->page = NULL;
++ return -EIO;
++ }
++
++ rx_queue->buf_page = rx_buf->page;
++ rx_queue->buf_dma_addr = dma_addr;
++ rx_queue->buf_data = ((char *) page_address(rx_buf->page) +
++ EFX_PAGE_IP_ALIGN);
++ }
++
++ offset = RX_DATA_OFFSET(rx_queue->buf_data);
++ rx_buf->len = bytes;
++ rx_buf->dma_addr = rx_queue->buf_dma_addr + offset;
++ rx_buf->data = rx_queue->buf_data;
++
++ /* Try to pack multiple buffers per page */
++ if (efx->rx_buffer_order == 0) {
++ /* The next buffer starts on the next 512 byte boundary */
++ rx_queue->buf_data += ((bytes + 0x1ff) & ~0x1ff);
++ offset += ((bytes + 0x1ff) & ~0x1ff);
++
++ space = RX_PAGE_SIZE(efx) - offset;
++ if (space >= bytes) {
++ /* Refs dropped on kernel releasing each skb */
++ get_page(rx_queue->buf_page);
++ goto out;
++ }
++ }
++
++ /* This is the final RX buffer for this page, so mark it for
++ * unmapping */
++ rx_queue->buf_page = NULL;
++ rx_buf->unmap_addr = rx_queue->buf_dma_addr;
++
++ out:
++ return 0;
+}
+
-+void machine_restart(char * __unused)
++/* This allocates memory for a new receive buffer, maps it for DMA,
++ * and populates a struct efx_rx_buffer with the relevant
++ * information.
++ */
++static inline int efx_init_rx_buffer(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *new_rx_buf)
+{
-+ machine_emergency_restart();
++ int rc = 0;
++
++ if (rx_queue->channel->rx_alloc_push_pages) {
++ new_rx_buf->skb = NULL;
++ rc = efx_init_rx_buffer_page(rx_queue, new_rx_buf);
++ rx_queue->alloc_page_count++;
++ } else {
++ new_rx_buf->page = NULL;
++ rc = efx_init_rx_buffer_skb(rx_queue, new_rx_buf);
++ rx_queue->alloc_skb_count++;
++ }
++
++ if (unlikely(rc < 0))
++ EFX_LOG_RL(rx_queue->efx, "%s RXQ[%d] =%d\n", __func__,
++ rx_queue->queue, rc);
++ return rc;
+}
+
-+void machine_halt(void)
++static inline void efx_unmap_rx_buffer(struct efx_nic *efx,
++ struct efx_rx_buffer *rx_buf)
+{
-+ machine_power_off();
++ if (rx_buf->page) {
++ EFX_BUG_ON_PARANOID(rx_buf->skb);
++ if (rx_buf->unmap_addr) {
++ pci_unmap_page(efx->pci_dev, rx_buf->unmap_addr,
++ RX_PAGE_SIZE(efx), PCI_DMA_FROMDEVICE);
++ rx_buf->unmap_addr = 0;
++ }
++ } else if (likely(rx_buf->skb)) {
++ pci_unmap_single(efx->pci_dev, rx_buf->dma_addr,
++ rx_buf->len, PCI_DMA_FROMDEVICE);
++ }
+}
+
-+void machine_power_off(void)
++static inline void efx_free_rx_buffer(struct efx_nic *efx,
++ struct efx_rx_buffer *rx_buf)
+{
-+ /* We really want to get pending console data out before we die. */
-+ xencons_force_flush();
-+ if (pm_power_off)
-+ pm_power_off();
-+ HYPERVISOR_shutdown(SHUTDOWN_poweroff);
++ if (rx_buf->page) {
++ __free_pages(rx_buf->page, efx->rx_buffer_order);
++ rx_buf->page = NULL;
++ } else if (likely(rx_buf->skb)) {
++ dev_kfree_skb_any(rx_buf->skb);
++ rx_buf->skb = NULL;
++ }
+}
+
-+int reboot_thru_bios = 0; /* for dmi_scan.c */
-+EXPORT_SYMBOL(machine_restart);
-+EXPORT_SYMBOL(machine_halt);
-+EXPORT_SYMBOL(machine_power_off);
-+
-+static void pre_suspend(void)
++inline void efx_fini_rx_buffer(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *rx_buf)
+{
-+ HYPERVISOR_shared_info = (shared_info_t *)empty_zero_page;
-+ HYPERVISOR_update_va_mapping(fix_to_virt(FIX_SHARED_INFO),
-+ __pte_ma(0), 0);
++ /* Unmap for DMA */
++ efx_unmap_rx_buffer(rx_queue->efx, rx_buf);
+
-+ xen_start_info->store_mfn = mfn_to_pfn(xen_start_info->store_mfn);
-+ xen_start_info->console.domU.mfn =
-+ mfn_to_pfn(xen_start_info->console.domU.mfn);
++ /* Free the skb/page */
++ efx_free_rx_buffer(rx_queue->efx, rx_buf);
+}
+
-+static void post_suspend(int suspend_cancelled)
-+{
-+ int i, j, k, fpp;
-+ unsigned long shinfo_mfn;
-+ extern unsigned long max_pfn;
-+ extern unsigned long *pfn_to_mfn_frame_list_list;
-+ extern unsigned long *pfn_to_mfn_frame_list[];
++/**
++ * efx_fast_push_rx_descriptors - push new RX descriptors quickly
++ * @rx_queue: RX descriptor queue
++ * @retry: Recheck the fill level
++ * This will aim to fill the RX descriptor queue up to
++ * @rx_queue->@fast_fill_limit. If there is insufficient atomic
++ * memory to do so, the caller should retry.
++ */
++static int __efx_fast_push_rx_descriptors(struct efx_rx_queue *rx_queue,
++ int retry)
++{
++ struct efx_rx_buffer *rx_buf;
++ unsigned fill_level, index;
++ int i, space, rc = 0;
++
++ /* Calculate current fill level. Do this outside the lock,
++ * because most of the time we'll end up not wanting to do the
++ * fill anyway.
++ */
++ fill_level = (rx_queue->added_count - rx_queue->removed_count);
++ EFX_BUG_ON_PARANOID(fill_level >
++ rx_queue->efx->type->rxd_ring_mask + 1);
++
++ /* Don't fill if we don't need to */
++ if (fill_level >= rx_queue->fast_fill_trigger)
++ return 0;
+
-+ if (suspend_cancelled) {
-+ xen_start_info->store_mfn =
-+ pfn_to_mfn(xen_start_info->store_mfn);
-+ xen_start_info->console.domU.mfn =
-+ pfn_to_mfn(xen_start_info->console.domU.mfn);
-+ } else {
-+#ifdef CONFIG_SMP
-+ cpu_initialized_map = cpu_online_map;
-+#endif
-+ }
++ /* Record minimum fill level */
++ if (unlikely(fill_level < rx_queue->min_fill))
++ if (fill_level)
++ rx_queue->min_fill = fill_level;
+
-+ shinfo_mfn = xen_start_info->shared_info >> PAGE_SHIFT;
-+ HYPERVISOR_update_va_mapping(fix_to_virt(FIX_SHARED_INFO),
-+ pfn_pte_ma(shinfo_mfn, PAGE_KERNEL), 0);
-+ HYPERVISOR_shared_info = (shared_info_t *)fix_to_virt(FIX_SHARED_INFO);
++ /* Acquire RX add lock. If this lock is contended, then a fast
++ * fill must already be in progress (e.g. in the refill
++ * tasklet), so we don't need to do anything
++ */
++ if (!spin_trylock_bh(&rx_queue->add_lock))
++ return -1;
+
-+ memset(empty_zero_page, 0, PAGE_SIZE);
++ retry:
++ /* Recalculate current fill level now that we have the lock */
++ fill_level = (rx_queue->added_count - rx_queue->removed_count);
++ EFX_BUG_ON_PARANOID(fill_level >
++ rx_queue->efx->type->rxd_ring_mask + 1);
++ space = rx_queue->fast_fill_limit - fill_level;
++ if (space < EFX_RX_BATCH)
++ goto out_unlock;
+
-+ fpp = PAGE_SIZE/sizeof(unsigned long);
-+ for (i = 0, j = 0, k = -1; i < max_pfn; i += fpp, j++) {
-+ if ((j % fpp) == 0) {
-+ k++;
-+ pfn_to_mfn_frame_list_list[k] =
-+ virt_to_mfn(pfn_to_mfn_frame_list[k]);
-+ j = 0;
++ EFX_TRACE(rx_queue->efx, "RX queue %d fast-filling descriptor ring from"
++ " level %d to level %d using %s allocation\n",
++ rx_queue->queue, fill_level, rx_queue->fast_fill_limit,
++ rx_queue->channel->rx_alloc_push_pages ? "page" : "skb");
++
++ do {
++ for (i = 0; i < EFX_RX_BATCH; ++i) {
++ index = (rx_queue->added_count &
++ rx_queue->efx->type->rxd_ring_mask);
++ rx_buf = efx_rx_buffer(rx_queue, index);
++ rc = efx_init_rx_buffer(rx_queue, rx_buf);
++ if (unlikely(rc))
++ goto out;
++ ++rx_queue->added_count;
+ }
-+ pfn_to_mfn_frame_list[k][j] =
-+ virt_to_mfn(&phys_to_machine_mapping[i]);
-+ }
-+ HYPERVISOR_shared_info->arch.max_pfn = max_pfn;
-+ HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list =
-+ virt_to_mfn(pfn_to_mfn_frame_list_list);
-+}
++ } while ((space -= EFX_RX_BATCH) >= EFX_RX_BATCH);
+
-+#else /* !(defined(__i386__) || defined(__x86_64__)) */
++ EFX_TRACE(rx_queue->efx, "RX queue %d fast-filled descriptor ring "
++ "to level %d\n", rx_queue->queue,
++ rx_queue->added_count - rx_queue->removed_count);
+
-+#ifndef HAVE_XEN_PRE_SUSPEND
-+#define xen_pre_suspend() ((void)0)
-+#endif
++ out:
++ /* Send write pointer to card. */
++ falcon_notify_rx_desc(rx_queue);
+
-+#ifndef HAVE_XEN_POST_SUSPEND
-+#define xen_post_suspend(x) ((void)0)
-+#endif
++ /* If the fast fill is running inside from the refill tasklet, then
++ * for SMP systems it may be running on a different CPU to
++ * RX event processing, which means that the fill level may now be
++ * out of date. */
++ if (unlikely(retry && (rc == 0)))
++ goto retry;
+
-+#define switch_idle_mm() ((void)0)
-+#define mm_pin_all() ((void)0)
-+#define pre_suspend() xen_pre_suspend()
-+#define post_suspend(x) xen_post_suspend(x)
++ out_unlock:
++ /* Release RX add lock */
++ spin_unlock_bh(&rx_queue->add_lock);
+
-+#endif
++ return rc;
++}
+
-+static int take_machine_down(void *p_fast_suspend)
++/**
++ * efx_fast_push_rx_descriptors - push new RX descriptors quickly
++ * @rx_queue: RX descriptor queue
++ *
++ * This will aim to fill the RX descriptor queue up to
++ * @rx_queue->@fast_fill_limit. If there is insufficient memory to do so,
++ * it will schedule a work item to immediately continue the fast fill
++ */
++void efx_fast_push_rx_descriptors(struct efx_rx_queue *rx_queue)
+{
-+ int fast_suspend = *(int *)p_fast_suspend;
-+ int suspend_cancelled, err;
-+ extern void time_resume(void);
++ int rc;
+
-+ if (fast_suspend) {
-+ BUG_ON(!irqs_disabled());
-+ } else {
-+ BUG_ON(irqs_disabled());
++ rc = __efx_fast_push_rx_descriptors(rx_queue, 0);
++ if (unlikely(rc)) {
++ /* Schedule the work item to run immediately. The hope is
++ * that work is immediately pending to free some memory
++ * (e.g. an RX event or TX completion)
++ */
++ queue_delayed_work(rx_queue->efx->refill_workqueue,
++ &rx_queue->work, 0);
++ }
++}
+
-+ for (;;) {
-+ err = smp_suspend();
-+ if (err)
-+ return err;
++void efx_rx_work(struct work_struct *data)
++{
++ struct efx_rx_queue *rx_queue;
++ int rc;
+
-+ xenbus_suspend();
-+ preempt_disable();
++#if !defined(EFX_NEED_WORK_API_WRAPPERS)
++ rx_queue = container_of(data, struct efx_rx_queue, work.work);
++#else
++ rx_queue = container_of(data, struct efx_rx_queue, work);
++#endif
+
-+ if (num_online_cpus() == 1)
-+ break;
++ if (unlikely(!rx_queue->channel->enabled))
++ return;
+
-+ preempt_enable();
-+ xenbus_suspend_cancel();
-+ }
++ EFX_TRACE(rx_queue->efx, "RX queue %d worker thread executing on CPU "
++ "%d\n", rx_queue->queue, raw_smp_processor_id());
+
-+ local_irq_disable();
++ ++rx_queue->slow_fill_count;
++ /* Push new RX descriptors, allowing at least 1 jiffy for
++ * the kernel to free some more memory. */
++ rc = __efx_fast_push_rx_descriptors(rx_queue, 1);
++ if (rc) {
++ queue_delayed_work(rx_queue->efx->refill_workqueue,
++ &rx_queue->work, 1);
+ }
++}
+
-+ mm_pin_all();
-+ gnttab_suspend();
-+ pre_suspend();
++static inline void efx_rx_packet__check_len(struct efx_rx_queue *rx_queue,
++ struct efx_rx_buffer *rx_buf,
++ int len, int *discard,
++ int *leak_packet)
++{
++ struct efx_nic *efx = rx_queue->efx;
++ unsigned max_len = rx_buf->len - efx->type->rx_buffer_padding;
+
-+ /*
-+ * This hypercall returns 1 if suspend was cancelled or the domain was
-+ * merely checkpointed, and 0 if it is resuming in a new domain.
++ if (likely(len <= max_len))
++ return;
++
++ /* The packet must be discarded, but this is only a fatal error
++ * if the caller indicated it was
+ */
-+ suspend_cancelled = HYPERVISOR_suspend(virt_to_mfn(xen_start_info));
++ *discard = 1;
+
-+ post_suspend(suspend_cancelled);
-+ gnttab_resume();
-+ if (!suspend_cancelled) {
-+ irq_resume();
-+#ifdef __x86_64__
-+ /*
-+ * Older versions of Xen do not save/restore the user %cr3.
-+ * We do it here just in case, but there's no need if we are
-+ * in fast-suspend mode as that implies a new enough Xen.
++ if ((len > rx_buf->len) && EFX_WORKAROUND_8071(efx)) {
++ EFX_ERR_RL(efx, " RX queue %d seriously overlength "
++ "RX event (0x%x > 0x%x+0x%x). Leaking\n",
++ rx_queue->queue, len, max_len,
++ efx->type->rx_buffer_padding);
++ /* If this buffer was skb-allocated, then the meta
++ * data at the end of the skb will be trashed. So
++ * we have no choice but to leak the fragment.
+ */
-+ if (!fast_suspend) {
-+ struct mmuext_op op;
-+ op.cmd = MMUEXT_NEW_USER_BASEPTR;
-+ op.arg1.mfn = pfn_to_mfn(__pa(__user_pgd(
-+ current->active_mm->pgd)) >> PAGE_SHIFT);
-+ if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
-+ BUG();
-+ }
-+#endif
++ *leak_packet = (rx_buf->skb != NULL);
++ efx_schedule_reset(efx, RESET_TYPE_RX_RECOVERY);
++ } else {
++ EFX_ERR_RL(efx, " RX queue %d overlength RX event "
++ "(0x%x > 0x%x)\n", rx_queue->queue, len, max_len);
+ }
-+ time_resume();
-+
-+ if (!fast_suspend)
-+ local_irq_enable();
+
-+ return suspend_cancelled;
++ rx_queue->channel->n_rx_overlength++;
+}
+
-+int __xen_suspend(int fast_suspend)
++/* Allocate and construct an SKB around a struct page.*/
++static inline struct sk_buff *efx_rx_mk_skb(struct efx_rx_buffer *rx_buf,
++ struct efx_nic *efx,
++ int hdr_len)
+{
-+ int err, suspend_cancelled;
-+
-+ BUG_ON(smp_processor_id() != 0);
-+ BUG_ON(in_interrupt());
++ struct sk_buff *skb;
+
-+#if defined(__i386__) || defined(__x86_64__)
-+ if (xen_feature(XENFEAT_auto_translated_physmap)) {
-+ printk(KERN_WARNING "Cannot suspend in "
-+ "auto_translated_physmap mode.\n");
-+ return -EOPNOTSUPP;
++ /* Allocate an SKB to store the headers */
++ skb = netdev_alloc_skb(efx->net_dev, hdr_len + EFX_PAGE_SKB_ALIGN);
++ if (unlikely(skb == NULL)) {
++ EFX_ERR_RL(efx, "RX out of memory for skb\n");
++ return NULL;
+ }
-+#endif
+
-+ /* If we are definitely UP then 'slow mode' is actually faster. */
-+ if (num_possible_cpus() == 1)
-+ fast_suspend = 0;
++ EFX_BUG_ON_PARANOID(skb_shinfo(skb)->nr_frags);
++ EFX_BUG_ON_PARANOID(rx_buf->len < hdr_len);
+
-+ if (fast_suspend) {
-+ xenbus_suspend();
-+ err = stop_machine_run(take_machine_down, &fast_suspend, 0);
-+ if (err < 0)
-+ xenbus_suspend_cancel();
-+ } else {
-+ err = take_machine_down(&fast_suspend);
-+ }
++ skb->ip_summed = CHECKSUM_UNNECESSARY;
++ skb_reserve(skb, EFX_PAGE_SKB_ALIGN);
+
-+ if (err < 0)
-+ return err;
++ skb->len = rx_buf->len;
++ skb->truesize = rx_buf->len + sizeof(struct sk_buff);
++ memcpy(skb->data, rx_buf->data, hdr_len);
++ skb->tail += hdr_len;
+
-+ suspend_cancelled = err;
-+ if (!suspend_cancelled) {
-+ xencons_resume();
-+ xenbus_resume();
++ /* Append the remaining page onto the frag list */
++ if (unlikely(rx_buf->len > hdr_len)) {
++ struct skb_frag_struct *frag = skb_shinfo(skb)->frags;
++ frag->page = rx_buf->page;
++ frag->page_offset = RX_BUF_OFFSET(rx_buf) + hdr_len;
++ frag->size = skb->len - hdr_len;
++ skb_shinfo(skb)->nr_frags = 1;
++ skb->data_len = frag->size;
+ } else {
-+ xenbus_suspend_cancel();
++ __free_pages(rx_buf->page, efx->rx_buffer_order);
++ skb->data_len = 0;
+ }
+
-+ if (!fast_suspend)
-+ smp_resume();
++ /* Ownership has transferred from the rx_buf to skb */
++ rx_buf->page = NULL;
+
-+ return 0;
++ /* Move past the ethernet header */
++ skb->protocol = eth_type_trans(skb, efx->net_dev);
++
++ return skb;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/Makefile linux-2.6.18-xen.hg/drivers/xen/core/Makefile
---- linux-2.6.18/drivers/xen/core/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/Makefile 2007-12-23 11:15:33.551242507 +0100
-@@ -0,0 +1,13 @@
-+#
-+# Makefile for the linux kernel.
-+#
+
-+obj-y := evtchn.o gnttab.o features.o reboot.o machine_reboot.o firmware.o
++#if defined(EFX_USE_FASTCALL)
++void fastcall efx_rx_packet(struct efx_rx_queue *rx_queue,
++ unsigned int index, unsigned int len,
++ int checksummed, int discard)
++#else
++void efx_rx_packet(struct efx_rx_queue *rx_queue, unsigned int index,
++ unsigned int len, int checksummed, int discard)
++#endif
++{
++ struct efx_nic *efx = rx_queue->efx;
++ struct efx_rx_buffer *rx_buf;
++ int leak_packet = 0;
+
-+obj-$(CONFIG_PROC_FS) += xen_proc.o
-+obj-$(CONFIG_SYS_HYPERVISOR) += hypervisor_sysfs.o
-+obj-$(CONFIG_HOTPLUG_CPU) += cpu_hotplug.o
-+obj-$(CONFIG_XEN_SYSFS) += xen_sysfs.o
-+obj-$(CONFIG_XEN_SMPBOOT) += smpboot.o
-+obj-$(CONFIG_KEXEC) += machine_kexec.o
-+obj-$(CONFIG_XEN_XENCOMM) += xencomm.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/reboot.c linux-2.6.18-xen.hg/drivers/xen/core/reboot.c
---- linux-2.6.18/drivers/xen/core/reboot.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/reboot.c 2007-12-23 11:15:33.664581791 +0100
-@@ -0,0 +1,246 @@
-+#define __KERNEL_SYSCALLS__
-+#include <linux/version.h>
-+#include <linux/kernel.h>
-+#include <linux/unistd.h>
-+#include <linux/module.h>
-+#include <linux/reboot.h>
-+#include <linux/sysrq.h>
-+#include <asm/hypervisor.h>
-+#include <xen/xenbus.h>
-+#include <linux/kmod.h>
-+#include <linux/slab.h>
-+#include <linux/workqueue.h>
++ rx_buf = efx_rx_buffer(rx_queue, index);
++ EFX_BUG_ON_PARANOID(!rx_buf->data);
++ EFX_BUG_ON_PARANOID(rx_buf->skb && rx_buf->page);
++ EFX_BUG_ON_PARANOID(!(rx_buf->skb || rx_buf->page));
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
++ /* This allows the refill path to post another buffer.
++ * EFX_RXD_HEAD_ROOM ensures that the slot we are using
++ * isn't overwritten yet.
++ */
++ rx_queue->removed_count++;
+
-+MODULE_LICENSE("Dual BSD/GPL");
++ /* Validate the length encoded in the event vs the descriptor pushed */
++ efx_rx_packet__check_len(rx_queue, rx_buf, len,
++ &discard, &leak_packet);
+
-+#define SHUTDOWN_INVALID -1
-+#define SHUTDOWN_POWEROFF 0
-+#define SHUTDOWN_SUSPEND 2
-+/* Code 3 is SHUTDOWN_CRASH, which we don't use because the domain can only
-+ * report a crash, not be instructed to crash!
-+ * HALT is the same as POWEROFF, as far as we're concerned. The tools use
-+ * the distinction when we return the reason code to them.
-+ */
-+#define SHUTDOWN_HALT 4
++ EFX_TRACE(efx, "RX queue %d received id %x at %llx+%x %s%s\n",
++ rx_queue->queue, index,
++ (unsigned long long)rx_buf->dma_addr, len,
++ (checksummed ? " [SUMMED]" : ""),
++ (discard ? " [DISCARD]" : ""));
+
-+/* Ignore multiple shutdown requests. */
-+static int shutting_down = SHUTDOWN_INVALID;
++ /* Discard packet, if instructed to do so */
++ if (unlikely(discard)) {
++ if (unlikely(leak_packet))
++ rx_queue->channel->n_skbuff_leaks++;
++ else
++ /* We haven't called efx_unmap_rx_buffer yet,
++ * so fini the entire rx_buffer here */
++ efx_fini_rx_buffer(rx_queue, rx_buf);
++ return;
++ }
+
-+/* Can we leave APs online when we suspend? */
-+static int fast_suspend;
++ /* Release card resources - assumes all RX buffers consumed in-order
++ * per RX queue
++ */
++ efx_unmap_rx_buffer(efx, rx_buf);
+
-+static void __shutdown_handler(void *unused);
-+static DECLARE_WORK(shutdown_work, __shutdown_handler, NULL);
++ /* Prefetch nice and early so data will (hopefully) be in cache by
++ * the time we look at it.
++ */
++ prefetch(rx_buf->data);
+
-+int __xen_suspend(int fast_suspend);
++ /* Pipeline receives so that we give time for packet headers to be
++ * prefetched into cache.
++ */
++ rx_buf->len = len;
++ if (rx_queue->channel->rx_pkt)
++ __efx_rx_packet(rx_queue->channel,
++ rx_queue->channel->rx_pkt,
++ rx_queue->channel->rx_pkt_csummed);
++ rx_queue->channel->rx_pkt = rx_buf;
++ rx_queue->channel->rx_pkt_csummed = checksummed;
++}
+
-+static int shutdown_process(void *__unused)
++/* Handle a received packet. Second half: Touches packet payload. */
++void __efx_rx_packet(struct efx_channel *channel,
++ struct efx_rx_buffer *rx_buf, int checksummed)
+{
-+ static char *envp[] = { "HOME=/", "TERM=linux",
-+ "PATH=/sbin:/usr/sbin:/bin:/usr/bin", NULL };
-+ static char *poweroff_argv[] = { "/sbin/poweroff", NULL };
++ struct efx_nic *efx = channel->efx;
++ enum efx_veto veto;
++ struct sk_buff *skb;
+
-+ extern asmlinkage long sys_reboot(int magic1, int magic2,
-+ unsigned int cmd, void *arg);
++ /* If we're in loopback test, then pass the packet directly to the
++ * loopback layer, and free the rx_buf here
++ */
++ if (unlikely(efx->loopback_selftest)) {
++ efx_loopback_rx_packet(efx, rx_buf->data, rx_buf->len);
++ efx_free_rx_buffer(efx, rx_buf);
++ goto done;
++ }
+
-+ if ((shutting_down == SHUTDOWN_POWEROFF) ||
-+ (shutting_down == SHUTDOWN_HALT)) {
-+ if (call_usermodehelper("/sbin/poweroff", poweroff_argv,
-+ envp, 0) < 0) {
-+#ifdef CONFIG_XEN
-+ sys_reboot(LINUX_REBOOT_MAGIC1,
-+ LINUX_REBOOT_MAGIC2,
-+ LINUX_REBOOT_CMD_POWER_OFF,
-+ NULL);
-+#endif /* CONFIG_XEN */
-+ }
-+ }
++ if (rx_buf->skb) {
++ /* Prefetch more information */
++ prefetch(skb_shinfo(rx_buf->skb));
+
-+ shutting_down = SHUTDOWN_INVALID; /* could try again */
++ /* Reserve space for the data */
++ skb_put(rx_buf->skb, rx_buf->len);
+
-+ return 0;
-+}
++ /* Move past the ethernet header. rx_buf->data still points
++ * at the ethernet header */
++ rx_buf->skb->protocol = eth_type_trans(rx_buf->skb,
++ efx->net_dev);
++ }
+
-+static int xen_suspend(void *__unused)
-+{
-+ int err;
++ /* Both our generic-LRO and SFC-SSR support skb and page based
++ * allocation, but neither support switching from one to the
++ * other on the fly. If we spot that the allocation mode has
++ * changed, then flush the LRO state.
++ */
++ if (unlikely(channel->rx_alloc_pop_pages != (rx_buf->page != NULL))) {
++ channel->rx_alloc_pop_pages = (rx_buf->page != NULL);
++ }
+
-+ daemonize("suspend");
-+ err = set_cpus_allowed(current, cpumask_of_cpu(0));
-+ if (err) {
-+ printk(KERN_ERR "Xen suspend can't run on CPU0 (%d)\n", err);
-+ goto out;
++ /* Allow callback to veto the packet */
++ veto = EFX_DL_CALLBACK(efx, rx_packet, rx_buf->data, rx_buf->len);
++ if (unlikely(veto)) {
++ EFX_LOG(efx, "RX vetoed by driverlink %s driver\n",
++ efx->dl_cb_dev.rx_packet->driver->name);
++ /* Free the buffer now */
++ efx_free_rx_buffer(efx, rx_buf);
++ goto done;
+ }
+
-+ err = __xen_suspend(fast_suspend);
-+ if (err)
-+ printk(KERN_ERR "Xen suspend failed (%d)\n", err);
++ /* Form an skb if required */
++ if (rx_buf->page) {
++ int hdr_len = min(rx_buf->len, EFX_SKB_HEADERS);
++ skb = efx_rx_mk_skb(rx_buf, efx, hdr_len);
++ if (unlikely(skb == NULL)) {
++ efx_free_rx_buffer(efx, rx_buf);
++ goto done;
++ }
++ } else {
++ /* We now own the SKB */
++ skb = rx_buf->skb;
++ rx_buf->skb = NULL;
++ }
+
-+ out:
-+ shutting_down = SHUTDOWN_INVALID;
-+ return 0;
++ EFX_BUG_ON_PARANOID(rx_buf->page);
++ EFX_BUG_ON_PARANOID(rx_buf->skb);
++ EFX_BUG_ON_PARANOID(!skb);
++
++ /* Set the SKB flags */
++ if (unlikely(!checksummed || !efx->rx_checksum_enabled))
++ skb->ip_summed = CHECKSUM_NONE;
++
++ /* Pass the packet up */
++ netif_receive_skb(skb);
++
++ /* Update allocation strategy method */
++ channel->rx_alloc_level += RX_ALLOC_FACTOR_SKB;
++
++ /* fall-thru */
++done:
++ /* Update statistics */
++ efx->net_dev->last_rx = jiffies;
+}
+
-+static void __shutdown_handler(void *unused)
++void efx_rx_strategy(struct efx_channel *channel)
+{
-+ int err;
++ enum efx_rx_alloc_method method = rx_alloc_method;
+
-+ err = kernel_thread((shutting_down == SHUTDOWN_SUSPEND) ?
-+ xen_suspend : shutdown_process,
-+ NULL, CLONE_FS | CLONE_FILES);
++ /* Only makes sense to use page based allocation if LRO is enabled */
++ if (!(channel->efx->net_dev->features & NETIF_F_LRO)) {
++ method = RX_ALLOC_METHOD_SKB;
++ } else if (method == RX_ALLOC_METHOD_AUTO) {
++ /* Constrain the rx_alloc_level */
++ if (channel->rx_alloc_level < 0)
++ channel->rx_alloc_level = 0;
++ else if (channel->rx_alloc_level > RX_ALLOC_LEVEL_MAX)
++ channel->rx_alloc_level = RX_ALLOC_LEVEL_MAX;
+
-+ if (err < 0) {
-+ printk(KERN_WARNING "Error creating shutdown process (%d): "
-+ "retrying...\n", -err);
-+ schedule_delayed_work(&shutdown_work, HZ/2);
++ /* Decide on the allocation method */
++ method = ((channel->rx_alloc_level > RX_ALLOC_LEVEL_LRO) ?
++ RX_ALLOC_METHOD_PAGE : RX_ALLOC_METHOD_SKB);
+ }
++
++ /* Push the option */
++ channel->rx_alloc_push_pages = (method == RX_ALLOC_METHOD_PAGE);
+}
+
-+static void shutdown_handler(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
++int efx_probe_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ extern void ctrl_alt_del(void);
-+ char *str;
-+ struct xenbus_transaction xbt;
-+ int err;
-+
-+ if (shutting_down != SHUTDOWN_INVALID)
-+ return;
++ struct efx_nic *efx = rx_queue->efx;
++ unsigned int rxq_size;
++ int rc;
+
-+ again:
-+ err = xenbus_transaction_start(&xbt);
-+ if (err)
-+ return;
++ EFX_LOG(efx, "creating RX queue %d\n", rx_queue->queue);
+
-+ str = (char *)xenbus_read(xbt, "control", "shutdown", NULL);
-+ /* Ignore read errors and empty reads. */
-+ if (XENBUS_IS_ERR_READ(str)) {
-+ xenbus_transaction_end(xbt, 1);
-+ return;
++ /* Allocate RX buffers */
++ rxq_size = (efx->type->rxd_ring_mask + 1) * sizeof(*rx_queue->buffer);
++ rx_queue->buffer = kzalloc(rxq_size, GFP_KERNEL);
++ if (!rx_queue->buffer) {
++ rc = -ENOMEM;
++ goto fail1;
+ }
+
-+ xenbus_write(xbt, "control", "shutdown", "");
-+
-+ err = xenbus_transaction_end(xbt, 0);
-+ if (err == -EAGAIN) {
-+ kfree(str);
-+ goto again;
-+ }
++ rc = falcon_probe_rx(rx_queue);
++ if (rc)
++ goto fail2;
+
-+ if (strcmp(str, "poweroff") == 0)
-+ shutting_down = SHUTDOWN_POWEROFF;
-+ else if (strcmp(str, "reboot") == 0)
-+ ctrl_alt_del();
-+ else if (strcmp(str, "suspend") == 0)
-+ shutting_down = SHUTDOWN_SUSPEND;
-+ else if (strcmp(str, "halt") == 0)
-+ shutting_down = SHUTDOWN_HALT;
-+ else {
-+ printk("Ignoring shutdown request: %s\n", str);
-+ shutting_down = SHUTDOWN_INVALID;
-+ }
++ return 0;
+
-+ if (shutting_down != SHUTDOWN_INVALID)
-+ schedule_work(&shutdown_work);
++ fail2:
++ kfree(rx_queue->buffer);
++ rx_queue->buffer = NULL;
++ fail1:
++ /* Mark queue as unused */
++ rx_queue->used = 0;
+
-+ kfree(str);
++ return rc;
+}
+
-+static void sysrq_handler(struct xenbus_watch *watch, const char **vec,
-+ unsigned int len)
++int efx_init_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ char sysrq_key = '\0';
-+ struct xenbus_transaction xbt;
-+ int err;
++ struct efx_nic *efx = rx_queue->efx;
++ unsigned int max_fill, trigger, limit;
+
-+ again:
-+ err = xenbus_transaction_start(&xbt);
-+ if (err)
-+ return;
-+ if (!xenbus_scanf(xbt, "control", "sysrq", "%c", &sysrq_key)) {
-+ printk(KERN_ERR "Unable to read sysrq code in "
-+ "control/sysrq\n");
-+ xenbus_transaction_end(xbt, 1);
-+ return;
-+ }
++ EFX_LOG(rx_queue->efx, "initialising RX queue %d\n", rx_queue->queue);
+
-+ if (sysrq_key != '\0')
-+ xenbus_printf(xbt, "control", "sysrq", "%c", '\0');
++ ASSERT_RTNL();
+
-+ err = xenbus_transaction_end(xbt, 0);
-+ if (err == -EAGAIN)
-+ goto again;
++ /* Initialise ptr fields */
++ rx_queue->added_count = 0;
++ rx_queue->notified_count = 0;
++ rx_queue->removed_count = 0;
++ rx_queue->min_fill = -1U;
++ rx_queue->min_overfill = -1U;
+
-+#ifdef CONFIG_MAGIC_SYSRQ
-+ if (sysrq_key != '\0')
-+ handle_sysrq(sysrq_key, NULL, NULL);
-+#endif
-+}
++ /* Initialise limit fields */
++ max_fill = efx->type->rxd_ring_mask + 1 - EFX_RXD_HEAD_ROOM;
++ trigger = max_fill * min(rx_refill_threshold, 100U) / 100U;
++ limit = max_fill * min(rx_refill_limit, 100U) / 100U;
+
-+static struct xenbus_watch shutdown_watch = {
-+ .node = "control/shutdown",
-+ .callback = shutdown_handler
-+};
++ rx_queue->max_fill = max_fill;
++ rx_queue->fast_fill_trigger = trigger;
++ rx_queue->fast_fill_limit = limit;
+
-+static struct xenbus_watch sysrq_watch = {
-+ .node = "control/sysrq",
-+ .callback = sysrq_handler
-+};
++ /* Set up RX descriptor ring */
++ return falcon_init_rx(rx_queue);
++}
+
-+static int setup_shutdown_watcher(void)
++void efx_fini_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ int err;
++ int i;
++ struct efx_rx_buffer *rx_buf;
+
-+ xenbus_scanf(XBT_NIL, "control",
-+ "platform-feature-multiprocessor-suspend",
-+ "%d", &fast_suspend);
++ EFX_LOG(rx_queue->efx, "shutting down RX queue %d\n", rx_queue->queue);
+
-+ err = register_xenbus_watch(&shutdown_watch);
-+ if (err) {
-+ printk(KERN_ERR "Failed to set shutdown watcher\n");
-+ return err;
-+ }
++ ASSERT_RTNL();
+
-+ err = register_xenbus_watch(&sysrq_watch);
-+ if (err) {
-+ printk(KERN_ERR "Failed to set sysrq watcher\n");
-+ return err;
++ /* Flush RX queue and remove descriptor ring */
++ falcon_fini_rx(rx_queue);
++
++ /* Release RX buffers NB start at index 0 not current HW ptr */
++ if (rx_queue->buffer) {
++ for (i = 0; i <= rx_queue->efx->type->rxd_ring_mask; i++) {
++ rx_buf = efx_rx_buffer(rx_queue, i);
++ efx_fini_rx_buffer(rx_queue, rx_buf);
++ }
+ }
+
-+ return 0;
++ /* For a page that is part-way through splitting into RX buffers */
++ if (rx_queue->buf_page != NULL) {
++ pci_unmap_page(rx_queue->efx->pci_dev, rx_queue->buf_dma_addr,
++ RX_PAGE_SIZE(rx_queue->efx), PCI_DMA_FROMDEVICE);
++ __free_pages(rx_queue->buf_page,
++ rx_queue->efx->rx_buffer_order);
++ rx_queue->buf_page = NULL;
++ }
+}
+
-+#ifdef CONFIG_XEN
-+
-+static int shutdown_event(struct notifier_block *notifier,
-+ unsigned long event,
-+ void *data)
++void efx_remove_rx_queue(struct efx_rx_queue *rx_queue)
+{
-+ setup_shutdown_watcher();
-+ return NOTIFY_DONE;
-+}
++ EFX_LOG(rx_queue->efx, "destroying RX queue %d\n", rx_queue->queue);
+
-+static int __init setup_shutdown_event(void)
-+{
-+ static struct notifier_block xenstore_notifier = {
-+ .notifier_call = shutdown_event
-+ };
-+ register_xenstore_notifier(&xenstore_notifier);
++ falcon_remove_rx(rx_queue);
+
-+ return 0;
++ kfree(rx_queue->buffer);
++ rx_queue->buffer = NULL;
++ rx_queue->used = 0;
+}
+
-+subsys_initcall(setup_shutdown_event);
+
-+#else /* !defined(CONFIG_XEN) */
++module_param(rx_alloc_method, int, 0644);
++MODULE_PARM_DESC(rx_alloc_method, "Allocation method used for RX buffers");
+
-+int xen_reboot_init(void)
-+{
-+ return setup_shutdown_watcher();
-+}
++module_param(rx_refill_threshold, uint, 0444);
++MODULE_PARM_DESC(rx_refill_threshold,
++ "RX descriptor ring fast/slow fill threshold (%)");
+
-+#endif /* !defined(CONFIG_XEN) */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/smpboot.c linux-2.6.18-xen.hg/drivers/xen/core/smpboot.c
---- linux-2.6.18/drivers/xen/core/smpboot.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/smpboot.c 2007-12-23 11:15:33.674582320 +0100
-@@ -0,0 +1,435 @@
-+/*
-+ * Xen SMP booting functions
+--- linux-2.6.18.8/drivers/net/sfc/rx.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/rx.h 2008-05-19 00:33:29.321836271 +0300
+@@ -0,0 +1,44 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * See arch/i386/kernel/smpboot.c for copyright and credits for derived
-+ * portions of this file.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_RX_H
++#define EFX_RX_H
++
++#include "net_driver.h"
++
++
++int efx_probe_rx_queue(struct efx_rx_queue *rx_queue);
++void efx_remove_rx_queue(struct efx_rx_queue *rx_queue);
++int efx_init_rx_queue(struct efx_rx_queue *rx_queue);
++void efx_fini_rx_queue(struct efx_rx_queue *rx_queue);
++
++void efx_rx_strategy(struct efx_channel *channel);
++void efx_fast_push_rx_descriptors(struct efx_rx_queue *rx_queue);
++void efx_rx_work(struct work_struct *data);
++void __efx_rx_packet(struct efx_channel *channel,
++ struct efx_rx_buffer *rx_buf, int checksummed);
++
++
++#endif /* EFX_RX_H */
+--- linux-2.6.18.8/drivers/net/sfc/selftest.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/selftest.c 2008-05-19 00:33:29.321836271 +0300
+@@ -0,0 +1,815 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
++#include <linux/netdevice.h>
+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <linux/kernel.h>
-+#include <linux/mm.h>
-+#include <linux/sched.h>
++#include <linux/delay.h>
+#include <linux/kernel_stat.h>
-+#include <linux/smp_lock.h>
-+#include <linux/irq.h>
-+#include <linux/bootmem.h>
-+#include <linux/notifier.h>
-+#include <linux/cpu.h>
-+#include <linux/percpu.h>
-+#include <asm/desc.h>
-+#include <asm/arch_hooks.h>
-+#include <asm/pgalloc.h>
-+#include <xen/evtchn.h>
-+#include <xen/interface/vcpu.h>
-+#include <xen/cpu_hotplug.h>
-+#include <xen/xenbus.h>
++#include <linux/pci.h>
++#include <linux/ethtool.h>
++#include <linux/ip.h>
++#include <linux/in.h>
++#include <linux/udp.h>
++#include <linux/rtnetlink.h>
++#include <asm/io.h>
++#include "net_driver.h"
++#include "ethtool.h"
++#include "efx.h"
++#include "falcon.h"
++#include "selftest.h"
++#include "boards.h"
++#include "workarounds.h"
+
-+extern irqreturn_t smp_reschedule_interrupt(int, void *, struct pt_regs *);
-+extern irqreturn_t smp_call_function_interrupt(int, void *, struct pt_regs *);
++/* Self tests */
+
-+extern int local_setup_timer(unsigned int cpu);
-+extern void local_teardown_timer(unsigned int cpu);
++/*
++ * Loopback test packet structure
++ *
++ * The self-test should stress every RSS vector, and unfortunately
++ * Falcon only performs RSS on TCP/UDP packets.
++ */
++struct efx_loopback_payload {
++ struct ethhdr header;
++ struct iphdr ip;
++ struct udphdr udp;
++ __be16 iteration;
++ const char msg[64];
++} __attribute__ ((packed));
+
-+extern void hypervisor_callback(void);
-+extern void failsafe_callback(void);
-+extern void system_call(void);
-+extern void smp_trap_init(trap_info_t *);
++/* Loopback test source MAC address */
++static const unsigned char payload_source[ETH_ALEN] = {
++ 0x00, 0x0f, 0x53, 0x1b, 0x1b, 0x1b,
++};
+
-+/* Number of siblings per CPU package */
-+int smp_num_siblings = 1;
++static const char *payload_msg =
++ "Hello world! This is an Efx loopback test in progress!";
+
-+cpumask_t cpu_online_map;
-+EXPORT_SYMBOL(cpu_online_map);
-+cpumask_t cpu_possible_map;
-+EXPORT_SYMBOL(cpu_possible_map);
-+cpumask_t cpu_initialized_map;
++struct efx_selftest_state {
++ /* Drop all packets in efx_loopback_rx_packet */
++ int flush;
+
-+struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;
-+EXPORT_SYMBOL(cpu_data);
++ /* Number of packets being used in this test */
++ int packet_count;
+
-+#ifdef CONFIG_HOTPLUG_CPU
-+DEFINE_PER_CPU(int, cpu_state) = { 0 };
-+#endif
++ /* RX good packet count */
++ atomic_t rx_good;
+
-+static DEFINE_PER_CPU(int, resched_irq);
-+static DEFINE_PER_CPU(int, callfunc_irq);
-+static char resched_name[NR_CPUS][15];
-+static char callfunc_name[NR_CPUS][15];
++ /* RX bad packet count */
++ atomic_t rx_bad;
+
-+u8 cpu_2_logical_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
++ /* Payload used in tests */
++ struct efx_loopback_payload payload;
++};
+
-+cpumask_t cpu_sibling_map[NR_CPUS] __cacheline_aligned;
-+cpumask_t cpu_core_map[NR_CPUS] __cacheline_aligned;
-+EXPORT_SYMBOL(cpu_core_map);
++/**************************************************************************
++ *
++ * Configurable values
++ *
++ **************************************************************************/
+
-+#if defined(__i386__)
-+u8 x86_cpu_to_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = 0xff };
-+EXPORT_SYMBOL(x86_cpu_to_apicid);
-+#elif !defined(CONFIG_X86_IO_APIC)
-+unsigned int maxcpus = NR_CPUS;
-+#endif
++/* Level of loopback testing
++ *
++ * The maximum packet burst length is 16**(n-1), i.e.
++ *
++ * - Level 0 : no packets
++ * - Level 1 : 1 packet
++ * - Level 2 : 17 packets (1 * 1 packet, 1 * 16 packets)
++ * - Level 3 : 273 packets (1 * 1 packet, 1 * 16 packet, 1 * 256 packets)
++ *
++ */
++static unsigned int loopback_test_level = 3;
+
-+void __init prefill_possible_map(void)
++/**************************************************************************
++ *
++ * Interrupt and event queue testing
++ *
++ **************************************************************************/
++
++/*
++ * Test interrupts
++ *
++ * This generates a test interrupt and waits for it to be received by
++ * a CPU. This routine must be called from process context and will
++ * sleep.
++ */
++static int efx_test_interrupts(struct efx_nic *efx,
++ struct efx_self_tests *tests)
+{
-+ int i, rc;
++ struct efx_channel *channel;
++ unsigned long j_start;
+
-+ for_each_possible_cpu(i)
-+ if (i != smp_processor_id())
-+ return;
++ EFX_LOG(efx, "testing interrupts\n");
++ tests->interrupt = -1;
+
-+ for (i = 0; i < NR_CPUS; i++) {
-+ rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
-+ if (rc >= 0)
-+ cpu_set(i, cpu_possible_map);
++ /* Reset interrupt flag */
++ efx->last_irq_cpu = -1;
++ smp_wmb();
++
++ /* To guarantee that an interrupt is received make sure that the
++ * channels are ack'd at least once, reenabling interrupts.
++ * We will then receive an interrupt, either by explicitly
++ * requesting one, or receiving traffic.
++ */
++ efx_for_each_channel_with_interrupt(channel, efx) {
++ /* Process the eventq synchronously */
++ if (channel->work_pending)
++ efx_process_channel_now(channel);
++ /* Check if we haven't already received an interrupt */
++ if (efx->last_irq_cpu >= 0)
++ goto success;
+ }
-+}
+
-+void __init smp_alloc_memory(void)
-+{
-+}
++ j_start = jiffies;
+
-+static inline void
-+set_cpu_sibling_map(int cpu)
-+{
-+ cpu_data[cpu].phys_proc_id = cpu;
-+ cpu_data[cpu].cpu_core_id = 0;
++ /* Generate test interrupt. */
++ falcon_generate_interrupt(efx);
+
-+ cpu_sibling_map[cpu] = cpumask_of_cpu(cpu);
-+ cpu_core_map[cpu] = cpumask_of_cpu(cpu);
++ /* Wait for arrival of test interrupt. */
++ EFX_LOG(efx, "waiting for test interrupt\n");
++ schedule_timeout_uninterruptible(HZ / 10);
++ if (efx->last_irq_cpu >= 0)
++ goto success;
+
-+ cpu_data[cpu].booted_cores = 1;
++ EFX_ERR(efx, "timed out in %ld jiffies waiting for interrupt\n",
++ jiffies - j_start);
++ return -ETIMEDOUT;
++
++ success:
++ EFX_LOG(efx, "%s test interrupt seen on CPU%d\n", INT_MODE(efx),
++ efx->last_irq_cpu);
++ tests->interrupt = 1;
++ return 0;
+}
+
-+static void
-+remove_siblinginfo(int cpu)
++/*
++ * Test capability of events to receive events
++ * This generates a test event and waits for it to be processed
++ * This routine must be called from process context and will sleep.
++ */
++static int efx_test_eventq(struct efx_channel *channel,
++ struct efx_self_tests *tests)
+{
-+ cpu_data[cpu].phys_proc_id = BAD_APICID;
-+ cpu_data[cpu].cpu_core_id = BAD_APICID;
++ unsigned int magic;
+
-+ cpus_clear(cpu_sibling_map[cpu]);
-+ cpus_clear(cpu_core_map[cpu]);
++ /* Channel specific code, limited to 20 bits */
++ magic = (0x00010150 + channel->channel);
++ EFX_LOG(channel->efx, "channel %d testing event queue with code %x\n",
++ channel->channel, magic);
+
-+ cpu_data[cpu].booted_cores = 0;
++ tests->eventq_dma[channel->channel] = -1;
++ tests->eventq_int[channel->channel] = 1; /* fake pass */
++ tests->eventq_poll[channel->channel] = 1; /* fake pass */
++
++ /* Reset flag and zero magic word */
++ channel->efx->last_irq_cpu = -1;
++ channel->eventq_magic = 0;
++ smp_wmb();
++
++ /* Generate test event */
++ falcon_generate_test_event(channel, magic);
++ udelay(1);
++
++ efx_process_channel_now(channel);
++ if (channel->eventq_magic != magic) {
++ EFX_ERR(channel->efx, "channel %d failed to see test event\n",
++ channel->channel);
++ return -ETIMEDOUT;
++ } else {
++ tests->eventq_dma[channel->channel] = 1;
++ }
++
++ return 0;
+}
+
-+static int __cpuinit xen_smp_intr_init(unsigned int cpu)
++
++/*
++ * Test capability of events to generate interrupts
++ * This generates a test event and waits for it to be processed by an
++ * ISR. This routine must be called from process context and will
++ * sleep.
++ */
++static int efx_test_eventq_irq(struct efx_channel *channel,
++ struct efx_self_tests *tests)
+{
-+ int rc;
++ unsigned int magic, count;
++ unsigned long j_start = jiffies;
+
-+ per_cpu(resched_irq, cpu) = per_cpu(callfunc_irq, cpu) = -1;
++ /* Channel specific code, limited to 20 bits */
++ magic = (0x00010150 + channel->channel);
++ EFX_LOG(channel->efx, "channel %d testing event queue with code %x\n",
++ channel->channel, magic);
+
-+ sprintf(resched_name[cpu], "resched%d", cpu);
-+ rc = bind_ipi_to_irqhandler(RESCHEDULE_VECTOR,
-+ cpu,
-+ smp_reschedule_interrupt,
-+ SA_INTERRUPT,
-+ resched_name[cpu],
-+ NULL);
-+ if (rc < 0)
-+ goto fail;
-+ per_cpu(resched_irq, cpu) = rc;
++ tests->eventq_dma[channel->channel] = -1;
++ tests->eventq_int[channel->channel] = -1;
++ tests->eventq_poll[channel->channel] = -1;
+
-+ sprintf(callfunc_name[cpu], "callfunc%d", cpu);
-+ rc = bind_ipi_to_irqhandler(CALL_FUNCTION_VECTOR,
-+ cpu,
-+ smp_call_function_interrupt,
-+ SA_INTERRUPT,
-+ callfunc_name[cpu],
-+ NULL);
-+ if (rc < 0)
-+ goto fail;
-+ per_cpu(callfunc_irq, cpu) = rc;
++ /* Reset flag and zero magic word */
++ channel->efx->last_irq_cpu = -1;
++ channel->eventq_magic = 0;
++ smp_wmb();
+
-+ if ((cpu != 0) && ((rc = local_setup_timer(cpu)) != 0))
-+ goto fail;
++ /* Generate test event */
++ falcon_generate_test_event(channel, magic);
+
-+ return 0;
++ /* Wait for arrival of interrupt */
++ count = 0;
++ do {
++ schedule_timeout_uninterruptible(HZ / 100);
+
-+ fail:
-+ if (per_cpu(resched_irq, cpu) >= 0)
-+ unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
-+ if (per_cpu(callfunc_irq, cpu) >= 0)
-+ unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
-+ return rc;
-+}
++ if (channel->work_pending)
++ efx_process_channel_now(channel);
+
-+#ifdef CONFIG_HOTPLUG_CPU
-+static void xen_smp_intr_exit(unsigned int cpu)
-+{
-+ if (cpu != 0)
-+ local_teardown_timer(cpu);
++ if (channel->eventq_magic == magic)
++ goto eventq_ok;
++ }
++ while (++count < 2);
+
-+ unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
-+ unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
-+}
-+#endif
++ EFX_ERR(channel->efx, "channel %d timed out in %ld jiffies waiting for"
++ " event queue\n", channel->channel, jiffies - j_start);
+
-+void __cpuinit cpu_bringup(void)
-+{
-+ cpu_init();
-+ identify_cpu(cpu_data + smp_processor_id());
-+ touch_softlockup_watchdog();
-+ preempt_disable();
-+ local_irq_enable();
++ /* See if interrupt arrived */
++ if (channel->efx->last_irq_cpu >= 0) {
++ EFX_ERR(channel->efx, "channel %d saw interrupt on CPU%d "
++ "during event queue test\n", channel->channel,
++ raw_smp_processor_id());
++ tests->eventq_int[channel->channel] = 1;
++ }
++
++ /* Check to see if event was received even if interrupt wasn't */
++ efx_process_channel_now(channel);
++ if (channel->eventq_magic == magic) {
++ EFX_ERR(channel->efx, "channel %d event was generated, but "
++ "failed to trigger an interrupt\n", channel->channel);
++ tests->eventq_dma[channel->channel] = 1;
++ }
++
++ return -ETIMEDOUT;
++ eventq_ok:
++ EFX_LOG(channel->efx, "channel %d event queue passed\n",
++ channel->channel);
++ tests->eventq_dma[channel->channel] = 1;
++ tests->eventq_int[channel->channel] = 1;
++ tests->eventq_poll[channel->channel] = 1;
++ return 0;
+}
+
-+static void __cpuinit cpu_bringup_and_idle(void)
++/**************************************************************************
++ *
++ * PHY testing
++ *
++ **************************************************************************/
++
++/*
++ * Check PHY presence
++ * This reads the PHY ID registers via GMII and checks that neither
++ * are all-zeroes or all-ones (indicating a nonexistent or
++ * uncommunicative PHY).
++ */
++static int efx_test_phy(struct efx_nic *efx,
++ struct efx_self_tests *tests)
+{
-+ cpu_bringup();
-+ cpu_idle();
++ u16 physid1, physid2;
++ struct mii_if_info *mii = &efx->mii;
++ struct net_device *net_dev = efx->net_dev;
++
++ if (efx->phy_type == PHY_TYPE_NONE)
++ return 0;
++
++ EFX_LOG(efx, "testing PHY presence\n");
++ tests->phy_ok = -1;
++
++ physid1 = mii->mdio_read(net_dev, mii->phy_id, MII_PHYSID1);
++ physid2 = mii->mdio_read(net_dev, mii->phy_id, MII_PHYSID2);
++
++ if ((physid1 != 0x0000) && (physid1 != 0xffff) &&
++ (physid2 != 0x0000) && (physid2 != 0xffff)) {
++ EFX_LOG(efx, "found MII PHY %d ID 0x%x:%x\n",
++ mii->phy_id, physid1, physid2);
++ tests->phy_ok = 1;
++ return 0;
++ }
++
++ EFX_ERR(efx, "no MII PHY present with ID %d\n", mii->phy_id);
++ return -ENODEV;
+}
+
-+static void __cpuinit cpu_initialize_context(unsigned int cpu)
++/**************************************************************************
++ *
++ * Loopback testing
++ * NB Only one loopback test can be executing concurrently.
++ *
++ **************************************************************************/
++
++/* Loopback test RX callback
++ * This is called for each received packet during loopback testing.
++ */
++void efx_loopback_rx_packet(struct efx_nic *efx,
++ const char *buf_ptr, int pkt_len)
+{
-+ vcpu_guest_context_t ctxt;
-+ struct task_struct *idle = idle_task(cpu);
-+#ifdef __x86_64__
-+ struct desc_ptr *gdt_descr = &cpu_gdt_descr[cpu];
-+#else
-+ struct Xgt_desc_struct *gdt_descr = &per_cpu(cpu_gdt_descr, cpu);
-+#endif
++ struct efx_selftest_state *state = efx->loopback_selftest;
++ struct efx_loopback_payload *received;
++ struct efx_loopback_payload *payload;
+
-+ if (cpu_test_and_set(cpu, cpu_initialized_map))
++ BUG_ON(!buf_ptr);
++
++ /* If we are just flushing, then drop the packet */
++ if ((state == NULL) || state->flush)
+ return;
+
-+ memset(&ctxt, 0, sizeof(ctxt));
++ payload = &state->payload;
++
++ /* The packet should have been passed up to us before any LRO/SSR, so
++ * we should be able to compare the data directly. Since the packet
++ * is going to be thrown away by the caller, modify it in place.
++ * efx_test_loopback guarantees to not touch state->payload during
++ * the test */
++ received = (struct efx_loopback_payload *)(char *)buf_ptr;
++ received->ip.saddr = payload->ip.saddr;
++ received->ip.check = payload->ip.check;
++
++ /* Check that header exists */
++ if (pkt_len < sizeof(received->header)) {
++ EFX_ERR(efx, "saw runt RX packet (length %d) "
++ "in %s loopback test\n", pkt_len,
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+ ctxt.flags = VGCF_IN_KERNEL;
-+ ctxt.user_regs.ds = __USER_DS;
-+ ctxt.user_regs.es = __USER_DS;
-+ ctxt.user_regs.fs = 0;
-+ ctxt.user_regs.gs = 0;
-+ ctxt.user_regs.ss = __KERNEL_DS;
-+ ctxt.user_regs.eip = (unsigned long)cpu_bringup_and_idle;
-+ ctxt.user_regs.eflags = X86_EFLAGS_IF | 0x1000; /* IOPL_RING1 */
++ /* Check that header matches */
++ if (memcmp(&received->header, &payload->header, ETH_HLEN) != 0) {
++ EFX_ERR(efx, "saw non-loopback RX packet in"
++ " %s loopback test\n",
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+ memset(&ctxt.fpu_ctxt, 0, sizeof(ctxt.fpu_ctxt));
++ /* Check packet length */
++ if (pkt_len != sizeof(*payload)) {
++ EFX_ERR(efx, "saw incorrect RX packet length"
++ " %d (wanted %d) in %s loopback test\n",
++ pkt_len, (int)sizeof(*payload),
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+ smp_trap_init(ctxt.trap_ctxt);
++ /* Check that IP header matches */
++ if (memcmp(&received->ip, &payload->ip, sizeof(payload->ip)) != 0) {
++ EFX_ERR(efx, "saw corrupted IP header in %s "
++ "loopback test\n",
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+ ctxt.ldt_ents = 0;
++ /* Check that msg and padding matches */
++ if (memcmp(&received->msg, &payload->msg, sizeof(received->msg)) != 0) {
++ EFX_ERR(efx, "saw corrupted RX packet in %s "
++ "loopback test\n",
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+ ctxt.gdt_frames[0] = virt_to_mfn(gdt_descr->address);
-+ ctxt.gdt_ents = gdt_descr->size / 8;
++ /* Check that iteration matches */
++ if (received->iteration != payload->iteration) {
++ EFX_ERR(efx, "saw RX packet from iteration %d"
++ " (wanted %d) in %s loopback test\n",
++ ntohs(received->iteration), ntohs(payload->iteration),
++ LOOPBACK_MODE(efx));
++ goto err;
++ }
+
-+#ifdef __i386__
-+ ctxt.user_regs.cs = __KERNEL_CS;
-+ ctxt.user_regs.esp = idle->thread.esp0 - sizeof(struct pt_regs);
++ /* Increase correct RX count */
++ EFX_TRACE(efx, "got loopback RX in %s loopback test\n",
++ LOOPBACK_MODE(efx));
+
-+ ctxt.kernel_ss = __KERNEL_DS;
-+ ctxt.kernel_sp = idle->thread.esp0;
++ atomic_inc(&state->rx_good);
++ return;
+
-+ ctxt.event_callback_cs = __KERNEL_CS;
-+ ctxt.event_callback_eip = (unsigned long)hypervisor_callback;
-+ ctxt.failsafe_callback_cs = __KERNEL_CS;
-+ ctxt.failsafe_callback_eip = (unsigned long)failsafe_callback;
++ err:
++#ifdef EFX_ENABLE_DEBUG
++ if (atomic_read(&state->rx_bad) == 0) {
++ EFX_ERR(efx, "received packet:\n");
++ print_hex_dump(KERN_ERR, "", DUMP_PREFIX_OFFSET, 0x10, 1,
++ buf_ptr, pkt_len, 0);
++ EFX_ERR(efx, "expected packet:\n");
++ print_hex_dump(KERN_ERR, "", DUMP_PREFIX_OFFSET, 0x10, 1,
++ &state->payload, sizeof(state->payload), 0);
++ }
++#endif
++ atomic_inc(&state->rx_bad);
++}
++
++/* Initialise an efx_selftest_state for a new iteration */
++static void efx_iterate_state(struct efx_nic *efx)
++{
++ struct efx_selftest_state *state = efx->loopback_selftest;
++ struct net_device *net_dev = efx->net_dev;
++ struct efx_loopback_payload *payload = &state->payload;
++
++ /* Initialise the layerII header */
++ memcpy(&payload->header.h_dest, net_dev->dev_addr, ETH_ALEN);
++ memcpy(&payload->header.h_source, &payload_source, ETH_ALEN);
++ payload->header.h_proto = htons(ETH_P_IP);
++
++ /* saddr set later and used as incrementing count */
++ payload->ip.daddr = htonl(INADDR_LOOPBACK);
++ payload->ip.ihl = 5;
++ payload->ip.check = 0; /* offloaded */
++ payload->ip.tot_len = htons(sizeof(*payload) - sizeof(struct ethhdr));
++ payload->ip.version = IPVERSION;
++ payload->ip.protocol = IPPROTO_UDP;
++
++ /* Initialise udp header */
++ payload->udp.source = 0;
++ payload->udp.len = htons(sizeof(*payload) - sizeof(struct ethhdr) -
++ sizeof(struct iphdr));
++ payload->udp.check = 0; /* checksum ignored */
++
++ /* Fill out payload */
++ payload->iteration = htons(ntohs(payload->iteration) + 1);
++ memcpy(&payload->msg, payload_msg, sizeof(payload_msg));
++
++ /* Fill out remaining state members */
++ atomic_set(&state->rx_good, 0);
++ atomic_set(&state->rx_bad, 0);
++ smp_wmb();
++}
+
-+ ctxt.ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
-+#else /* __x86_64__ */
-+ ctxt.user_regs.cs = __KERNEL_CS;
-+ ctxt.user_regs.esp = idle->thread.rsp0 - sizeof(struct pt_regs);
++/*
++ * Perform loopback test with N packets
++ *
++ * This will transmit "num_packets" copies of a test packet, and check
++ * that they were both transmitted (i.e. a TX completion event was
++ * received) and received (i.e. the data arrived intact via loopback).
++ * The port must have already been placed into the desired loopback
++ * mode.
++ */
++static int efx_test_loopback(struct efx_nic *efx,
++ struct efx_tx_queue *tx_queue,
++ struct efx_loopback_self_tests *lb_tests)
++{
++#if !defined(EFX_HAVE_OLD_NAPI)
++ struct efx_channel *channel;
++#endif
++ struct efx_selftest_state *state = efx->loopback_selftest;
++ struct efx_loopback_payload *payload;
++ struct sk_buff *skb;
++ int rc = 0, i, tx_done, rx_good, rx_bad;
+
-+ ctxt.kernel_ss = __KERNEL_DS;
-+ ctxt.kernel_sp = idle->thread.rsp0;
++ /* Fill out the packet contents */
++ efx_iterate_state(efx);
+
-+ ctxt.event_callback_eip = (unsigned long)hypervisor_callback;
-+ ctxt.failsafe_callback_eip = (unsigned long)failsafe_callback;
-+ ctxt.syscall_callback_eip = (unsigned long)system_call;
++ /* Create and fill skb */
++ skb = alloc_skb(sizeof(state->payload), GFP_KERNEL);
++ if (!skb) {
++ rc = -ENOMEM;
++ goto out1;
++ }
++ payload = ((struct efx_loopback_payload *)
++ skb_put(skb, sizeof(state->payload)));
++ memcpy(payload, &state->payload, sizeof(state->payload));
+
-+ ctxt.ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(init_level4_pgt));
++ /* Transmit N copies of buffer */
++ for (i = 0; i < state->packet_count; i++) {
++ /* Set the source address in the copy of the packet.
++ * Incrementing the source address on a per-packet basis
++ * should ensure that we stress all RSS vectors */
++ payload->ip.saddr = htonl(INADDR_LOOPBACK | (i << 2));
++ skb_get(skb);
+
-+ ctxt.gs_base_kernel = (unsigned long)(cpu_pda(cpu));
-+#endif
++ if (efx_xmit(efx, tx_queue, skb) != NETDEV_TX_OK) {
++ EFX_ERR(efx, "TX queue %d could not transmit "
++ "packet %d of %d in %s loopback test\n",
++ tx_queue->queue, i + 1,
++ state->packet_count, LOOPBACK_MODE(efx));
++ rc = -EPIPE;
++ goto out2;
++ }
+
-+ BUG_ON(HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, &ctxt));
-+}
++ /* Avoid hogging the PCI bus */
++ udelay(10);
++ }
+
-+void __init smp_prepare_cpus(unsigned int max_cpus)
-+{
-+ int cpu;
-+ struct task_struct *idle;
-+#ifdef __x86_64__
-+ struct desc_ptr *gdt_descr;
++#if !defined(EFX_HAVE_OLD_NAPI)
++ /* NAPI polling is not enabled, so process channels synchronously */
++ schedule_timeout_uninterruptible(HZ / 50);
++ efx_for_each_channel_with_interrupt(channel, efx) {
++ if (channel->work_pending)
++ efx_process_channel_now(channel);
++ }
+#else
-+ struct Xgt_desc_struct *gdt_descr;
++ /* Allow time for processing */
++ schedule_timeout_uninterruptible(HZ / 10);
+#endif
+
-+ boot_cpu_data.apicid = 0;
-+ cpu_data[0] = boot_cpu_data;
++ if (state->flush)
++ goto out3;
+
-+ cpu_2_logical_apicid[0] = 0;
-+ x86_cpu_to_apicid[0] = 0;
++ /* Check TX completion and received packet counts */
++ tx_done = state->packet_count - (atomic_read(&skb->users) - 1);
++ rx_good = atomic_read(&state->rx_good);
++ rx_bad = atomic_read(&state->rx_bad);
++ if (tx_done != state->packet_count) {
++ /* Don't free the skbs; they will be picked up on TX
++ * overflow or channel teardown.
++ */
++ EFX_ERR(efx, "TX queue %d saw only %d out of an "
++ "expected %d TX completion events in %s loopback "
++ "test\n", tx_queue->queue, tx_done,
++ state->packet_count, LOOPBACK_MODE(efx));
++ rc = -ETIMEDOUT;
++ /* Allow to fall through so we see the RX errors as well */
++ }
++
++ /* We may always be up to a flush away from our desired packet total */
++ if (rx_good != state->packet_count) {
++ EFX_LOG(efx, "TX queue %d saw only %d out of an "
++ "expected %d received packets in %s loopback "
++ "test\n", tx_queue->queue, rx_good,
++ state->packet_count, LOOPBACK_MODE(efx));
++ rc = -ETIMEDOUT;
++ /* Fall through */
++ }
++
++ /* Update loopback test structure */
++ lb_tests->tx_sent[tx_queue->queue] += state->packet_count;
++ lb_tests->tx_done[tx_queue->queue] += tx_done;
++ lb_tests->rx_good += rx_good;
++ lb_tests->rx_bad += rx_bad;
++
++ out3:
++ out2:
++ /* Free skb */
++ if (skb) {
++ /* If the selftest failed then the skb needs its reference
++ * count decreasing */
++ while (skb_shared(skb))
++ kfree_skb(skb);
++ dev_kfree_skb_any(skb);
++ }
+
-+ current_thread_info()->cpu = 0;
++ out1:
++ return rc;
++}
+
-+ for (cpu = 0; cpu < NR_CPUS; cpu++) {
-+ cpus_clear(cpu_sibling_map[cpu]);
-+ cpus_clear(cpu_core_map[cpu]);
++/* Perform loopback test safely
++ *
++ * This performs a safe loopback test by starting with a single packet
++ * and only increasing the number of packets while the tests are
++ * passing. This avoids flooding a network with garbage packets if
++ * e.g. setting the loopback mode fails.
++ */
++static int
++efx_test_loopback_safely(struct efx_nic *efx,
++ struct efx_tx_queue *tx_queue,
++ struct efx_loopback_self_tests *lb_tests)
++{
++ struct efx_selftest_state *state = efx->loopback_selftest;
++ int i, rc = 0;
++
++ for (i = 0; i < loopback_test_level; i++) {
++ /* Determine how many packets to send */
++ state->packet_count = (efx->type->txd_ring_mask + 1) / 3;
++ state->packet_count = min(1 << (i << 2), state->packet_count);
++ state->flush = 0;
++
++ EFX_LOG(efx, "TX queue %d testing %s loopback"
++ " with %d packets\n", tx_queue->queue,
++ LOOPBACK_MODE(efx), state->packet_count);
++
++ rc = efx_test_loopback(efx, tx_queue, lb_tests);
++ if (rc) {
++ /* Wait a while to ensure there are no packets
++ * floating around after a failure.
++ */
++ schedule_timeout_uninterruptible(HZ / 5);
++ return rc;
++ }
+ }
+
-+ set_cpu_sibling_map(0);
++ EFX_LOG(efx, "TX queue %d passed %s loopback test "
++ "with a burst length of %d packets\n",
++ tx_queue->queue, LOOPBACK_MODE(efx), state->packet_count);
+
-+ if (xen_smp_intr_init(0))
-+ BUG();
++ return rc;
++}
+
-+ cpu_initialized_map = cpumask_of_cpu(0);
++static int efx_test_loopbacks(struct efx_nic *efx,
++ struct efx_self_tests *tests,
++ unsigned int loopback_modes)
++{
++ struct efx_selftest_state *state = efx->loopback_selftest;
++ struct ethtool_cmd ecmd, ecmd_loopback;
++ struct efx_tx_queue *tx_queue;
++ enum efx_loopback_mode old_mode, mode;
++ int old_powered, count, rc = 0;
++ int retry = EFX_WORKAROUND_8909(efx);
+
-+ /* Restrict the possible_map according to max_cpus. */
-+ while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
-+ for (cpu = NR_CPUS-1; !cpu_isset(cpu, cpu_possible_map); cpu--)
-+ continue;
-+ cpu_clear(cpu, cpu_possible_map);
++ /* Get current PHY settings */
++ rc = efx_ethtool_get_settings(efx->net_dev, &ecmd);
++ if (rc) {
++ EFX_ERR(efx, "could not get GMII settings\n");
++ return rc;
+ }
++ old_mode = efx->loopback_mode;
++ old_powered = efx->phy_powered;
+
-+ for_each_possible_cpu (cpu) {
-+ if (cpu == 0)
-+ continue;
++ /* Disable autonegotiation for the purposes of loopback */
++ memcpy(&ecmd_loopback, &ecmd, sizeof(ecmd_loopback));
++ if (ecmd_loopback.autoneg == AUTONEG_ENABLE) {
++ ecmd_loopback.autoneg = AUTONEG_DISABLE;
++ ecmd_loopback.duplex = DUPLEX_FULL;
++ ecmd_loopback.speed = EFX_IS10G(efx) ?
++ SPEED_10000 : SPEED_1000;
++ }
+
-+#ifdef __x86_64__
-+ gdt_descr = &cpu_gdt_descr[cpu];
-+#else
-+ gdt_descr = &per_cpu(cpu_gdt_descr, cpu);
-+#endif
-+ gdt_descr->address = get_zeroed_page(GFP_KERNEL);
-+ if (unlikely(!gdt_descr->address)) {
-+ printk(KERN_CRIT "CPU%d failed to allocate GDT\n",
-+ cpu);
++ rc = efx_ethtool_set_settings(efx->net_dev, &ecmd_loopback);
++ if (rc) {
++ EFX_ERR(efx, "could not disable autonegotiation\n");
++ goto out;
++ }
++ tests->loopback_speed = ecmd_loopback.speed;
++ tests->loopback_full_duplex = ecmd_loopback.duplex;
++
++ /* Test all supported loopback modes */
++ for (mode = LOOPBACK_NONE; mode < LOOPBACK_TEST_MAX; mode++) {
++ if (!(loopback_modes & (1 << mode)))
+ continue;
++
++ /* Move the port into the specified loopback mode. */
++ state->flush = 1;
++ efx->phy_powered = 1;
++ efx->loopback_mode = mode;
++ efx_reconfigure_port(efx, 0);
++
++ /* Wait for the PHY to signal the link is up */
++ count = 0;
++ do {
++ struct efx_channel *channel = &efx->channel[0];
++
++ (void) efx->mac_op->check_hw(efx);
++ schedule_timeout_uninterruptible(HZ / 10);
++ if (channel->work_pending)
++ efx_process_channel_now(channel);
++ /* Wait for PHY events to be processed */
++ flush_workqueue(efx->workqueue);
++ rmb();
++ } while ((++count < 20) && !efx->link_up);
++
++ /* The link should now be up. If it isn't, there is no point
++ * in attempting a loopback test */
++ if (!efx->link_up) {
++ EFX_ERR(efx, "loopback %s never came up\n",
++ LOOPBACK_MODE(efx));
++ rc = -EIO;
++ goto out;
+ }
-+ gdt_descr->size = GDT_SIZE;
-+ memcpy((void *)gdt_descr->address, cpu_gdt_table, GDT_SIZE);
-+ make_page_readonly(
-+ (void *)gdt_descr->address,
-+ XENFEAT_writable_descriptor_tables);
+
-+ cpu_data[cpu] = boot_cpu_data;
-+ cpu_data[cpu].apicid = cpu;
++ EFX_LOG(efx, "link came up in %s loopback in %d iterations\n",
++ LOOPBACK_MODE(efx), count);
+
-+ cpu_2_logical_apicid[cpu] = cpu;
-+ x86_cpu_to_apicid[cpu] = cpu;
++ /* Test every TX queue */
++ efx_for_each_tx_queue(tx_queue, efx) {
++ rc |= efx_test_loopback_safely(efx, tx_queue,
++ &tests->loopback[mode]);
++ if (rc)
++ goto fail;
++ }
+
-+ idle = fork_idle(cpu);
-+ if (IS_ERR(idle))
-+ panic("failed fork for CPU %d", cpu);
++ continue;
+
-+#ifdef __x86_64__
-+ cpu_pda(cpu)->pcurrent = idle;
-+ cpu_pda(cpu)->cpunumber = cpu;
-+ clear_ti_thread_flag(idle->thread_info, TIF_FORK);
-+#endif
++fail:
++ if (retry) {
++ /* Give the PHY a kick by pretending to move into
++ * a Falcon internal loopback mode, then back out */
++ int first = ffs(efx->loopback_modes) - 1;
+
-+ irq_ctx_init(cpu);
++ EFX_INFO(efx, "retrying %s loopback\n",
++ LOOPBACK_MODE(efx));
+
-+#ifdef CONFIG_HOTPLUG_CPU
-+ if (is_initial_xendomain())
-+ cpu_set(cpu, cpu_present_map);
-+#else
-+ cpu_set(cpu, cpu_present_map);
-+#endif
-+ }
++ state->flush = 1;
++ efx->loopback_mode = first;
++ efx_reconfigure_port(efx, 0);
+
-+ init_xenbus_allowed_cpumask();
++ retry = rc = 0;
++ --mode;
++ continue;
++ }
++ break;
++ }
+
-+#ifdef CONFIG_X86_IO_APIC
-+ /*
-+ * Here we can be sure that there is an IO-APIC in the system. Let's
-+ * go and set it up:
-+ */
-+ if (!skip_ioapic_setup && nr_ioapics)
-+ setup_IO_APIC();
-+#endif
-+}
++ out:
++ /* Take out of loopback and restore PHY settings */
++ state->flush = 1;
++ efx->loopback_mode = old_mode;
++ efx->phy_powered = old_powered;
++ /* Push the loopback change, and restore any other
++ * settings we may have trodden on */
++ (void) efx_ethtool_set_settings(efx->net_dev, &ecmd);
+
-+void __devinit smp_prepare_boot_cpu(void)
-+{
-+ prefill_possible_map();
++ return rc;
+}
+
-+#ifdef CONFIG_HOTPLUG_CPU
++/**************************************************************************
++ *
++ * Entry points
++ *
++ *************************************************************************/
+
-+/*
-+ * Initialize cpu_present_map late to skip SMP boot code in init/main.c.
-+ * But do it early enough to catch critical for_each_present_cpu() loops
-+ * in i386-specific code.
++/* Online (i.e. non-disruptive) testing
++ *
++ * This checks interrupt generation, event delivery and PHY presence.
++ * The caller should hold the suspend lock
++ *
+ */
-+static int __init initialize_cpu_present_map(void)
++int efx_online_test(struct efx_nic *efx, struct efx_self_tests *tests)
+{
-+ cpu_present_map = cpu_possible_map;
-+ return 0;
-+}
-+core_initcall(initialize_cpu_present_map);
++ struct efx_channel *channel;
++ int rc = 0;
+
-+int __cpu_disable(void)
-+{
-+ cpumask_t map = cpu_online_map;
-+ int cpu = smp_processor_id();
++ ASSERT_RTNL();
+
-+ if (cpu == 0)
-+ return -EBUSY;
++ EFX_LOG(efx, "performing online self-tests\n");
+
-+ remove_siblinginfo(cpu);
++ rc |= efx_test_interrupts(efx, tests);
++ efx_for_each_channel(channel, efx) {
++ if (channel->has_interrupt)
++ rc |= efx_test_eventq_irq(channel, tests);
++ else
++ rc |= efx_test_eventq(channel, tests);
++ }
++ rc |= efx_test_phy(efx, tests);
+
-+ cpu_clear(cpu, map);
-+ fixup_irqs(map);
-+ cpu_clear(cpu, cpu_online_map);
++ if (rc)
++ EFX_ERR(efx, "failed online self-tests\n");
+
-+ return 0;
++ return rc;
+}
+
-+void __cpu_die(unsigned int cpu)
++/* Offline (i.e. disruptive) testing
++ * This checks MAC and PHY loopback on the specified port. The caller
++ * should hold the rtnl lock
++ */
++int efx_offline_test(struct efx_nic *efx,
++ struct efx_self_tests *tests, unsigned int loopback_modes)
+{
-+ while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
-+ current->state = TASK_UNINTERRUPTIBLE;
-+ schedule_timeout(HZ/10);
++ struct efx_selftest_state *state;
++ int rc = 0;
++
++ ASSERT_RTNL();
++
++ EFX_LOG(efx, "performing offline self-tests\n");
++
++ /* Create a selftest_state structure to hold state for the test */
++ state = kzalloc(sizeof(*state), GFP_KERNEL);
++ if (state == NULL) {
++ rc = -ENOMEM;
++ goto out;
+ }
+
-+ xen_smp_intr_exit(cpu);
++ /* Set the port loopback_selftest member. From this point on
++ * all received packets will be dropped. Mark the state as
++ * "flushing" so all inflight packets are dropped */
++ BUG_ON(efx->loopback_selftest);
++ state->flush = 1;
++ efx->loopback_selftest = (void *)state;
++ wmb();
+
-+ if (num_online_cpus() == 1)
-+ alternatives_smp_switch(0);
++ /* Test all loopback modes */
++ rc = efx_test_loopbacks(efx, tests, loopback_modes);
++
++ /* Tidy up the port test state */
++ efx->loopback_selftest = NULL;
++ wmb();
++ kfree(state);
++
++ out:
++ if (rc)
++ EFX_ERR(efx, "failed offline self-tests\n");
++
++ return rc;
+}
+
-+#endif /* CONFIG_HOTPLUG_CPU */
+--- linux-2.6.18.8/drivers/net/sfc/selftest.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/selftest.h 2008-05-19 00:33:29.321836271 +0300
+@@ -0,0 +1,67 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+int __cpuinit __cpu_up(unsigned int cpu)
-+{
-+ int rc;
++#ifndef EFX_SELFTEST_H
++#define EFX_SELFTEST_H
+
-+ rc = cpu_up_check(cpu);
-+ if (rc)
-+ return rc;
++#include "net_driver.h"
+
-+ cpu_initialize_context(cpu);
++/*
++ * Self tests
++ */
+
-+ if (num_online_cpus() == 1)
-+ alternatives_smp_switch(1);
++struct efx_loopback_self_tests {
++ int tx_sent[EFX_MAX_TX_QUEUES];
++ int tx_done[EFX_MAX_TX_QUEUES];
++ int rx_good;
++ int rx_bad;
++};
+
-+ /* This must be done before setting cpu_online_map */
-+ set_cpu_sibling_map(cpu);
-+ wmb();
++/* Efx self test results
++ * For fields which are not counters, 1 indicates success and -1
++ * indicates failure.
++ */
++struct efx_self_tests {
++ int interrupt;
++ int eventq_dma[EFX_MAX_CHANNELS];
++ int eventq_int[EFX_MAX_CHANNELS];
++ int eventq_poll[EFX_MAX_CHANNELS];
++ int phy_ok;
++ int loopback_speed;
++ int loopback_full_duplex;
++ struct efx_loopback_self_tests loopback[LOOPBACK_TEST_MAX];
++};
+
-+ rc = xen_smp_intr_init(cpu);
-+ if (rc) {
-+ remove_siblinginfo(cpu);
-+ return rc;
-+ }
++extern void efx_loopback_rx_packet(struct efx_nic *efx,
++ const char *buf_ptr, int pkt_len);
++extern int efx_online_test(struct efx_nic *efx,
++ struct efx_self_tests *tests);
++extern int efx_offline_test(struct efx_nic *efx,
++ struct efx_self_tests *tests,
++ unsigned int loopback_modes);
+
-+ cpu_set(cpu, cpu_online_map);
++#endif /* EFX_SELFTEST_H */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/Makefile 2008-05-19 00:33:29.325836502 +0300
+@@ -0,0 +1,15 @@
++obj-$(CONFIG_SFC_RESOURCE) := sfc_resource.o
+
-+ rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
-+ BUG_ON(rc);
++EXTRA_CFLAGS += -D__CI_HARDWARE_CONFIG_FALCON__
++EXTRA_CFLAGS += -D__ci_driver__
++EXTRA_CFLAGS += -Werror
++EXTRA_CFLAGS += -Idrivers/net/sfc -Idrivers/net/sfc/sfc_resource
+
-+ return 0;
-+}
++sfc_resource-objs := resource_driver.o iopage.o efx_vi_shm.o \
++ driverlink_new.o kernel_proc.o kfifo.o \
++ nic.o eventq.o falcon.o falcon_mac.o falcon_hash.o \
++ assert_valid.o buddy.o buffer_table.o filter_resource.o \
++ iobufset_resource.o resource_manager.o resources.o \
++ vi_resource_alloc.o vi_resource_event.o vi_resource_flush.o \
++ vi_resource_manager.o driver_object.o kernel_compat.o
+
-+void __init smp_cpus_done(unsigned int max_cpus)
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/assert_valid.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/assert_valid.c 2008-05-19 00:33:29.325836502 +0300
+@@ -0,0 +1,95 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains functions to assert validness of resources and
++ * resource manager in DEBUG build of the resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/sysdep.h>
++
++#ifndef NDEBUG
++#include <ci/efrm/resource.h>
++#include <ci/efrm/driver_private.h>
++#include <ci/efrm/debug.h>
++
++void
++efrm_resource_manager_assert_valid(struct efrm_resource_manager *rm,
++ const char *file, int line)
+{
++ _EFRM_ASSERT(rm, file, line);
++ _EFRM_ASSERT(rm->rm_name, file, line);
++ _EFRM_ASSERT(rm->rm_type < EFRM_RESOURCE_NUM, file, line);
++ _EFRM_ASSERT(rm->rm_table, file, line);
++ _EFRM_ASSERT(rm->rm_table_size > 0, file, line);
++ _EFRM_ASSERT(rm->rm_dtor, file, line);
+}
++EXPORT_SYMBOL(efrm_resource_manager_assert_valid);
+
-+#ifndef CONFIG_X86_LOCAL_APIC
-+int setup_profiling_timer(unsigned int multiplier)
++/*
++ * \param rs resource to validate
++ * \param ref_count_is_zero One of 3 values
++ * > 0 - check ref count is zero
++ * = 0 - check ref count is non-zero
++ * < 0 - ref count could be any value
++ */
++void
++efrm_resource_assert_valid(struct efrm_resource *rs, int ref_count_is_zero,
++ const char *file, int line)
+{
-+ return -EINVAL;
++ struct efrm_resource_manager *rm;
++
++ _EFRM_ASSERT(rs, file, line);
++
++ if (ref_count_is_zero >= 0) {
++ if (!(ref_count_is_zero || atomic_read(&rs->rs_ref_count) > 0)
++ || !(!ref_count_is_zero
++ || atomic_read(&rs->rs_ref_count) == 0))
++ EFRM_WARN("%s: check %szero ref=%d " EFRM_RESOURCE_FMT,
++ __FUNCTION__,
++ ref_count_is_zero == 0 ? "non-" : "",
++ atomic_read(&rs->rs_ref_count),
++ EFRM_RESOURCE_PRI_ARG(rs->rs_handle));
++
++ _EFRM_ASSERT(!(ref_count_is_zero == 0) ||
++ atomic_read(&rs->rs_ref_count) != 0, file, line);
++ _EFRM_ASSERT(!(ref_count_is_zero > 0) ||
++ atomic_read(&rs->rs_ref_count) == 0, file, line);
++ }
++
++ rm = efrm_rm_table[EFRM_RESOURCE_TYPE(rs->rs_handle)];
++ efrm_resource_manager_assert_valid(rm, file, line);
+}
++EXPORT_SYMBOL(efrm_resource_assert_valid);
++
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/xencomm.c linux-2.6.18-xen.hg/drivers/xen/core/xencomm.c
---- linux-2.6.18/drivers/xen/core/xencomm.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/xencomm.c 2007-12-23 11:15:33.867925803 +0100
-@@ -0,0 +1,229 @@
-+/*
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License as published by
-+ * the Free Software Foundation; either version 2 of the License, or
-+ * (at your option) any later version.
-+ *
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/buddy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/buddy.c 2008-05-19 00:33:29.325836502 +0300
+@@ -0,0 +1,307 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains implementation of a buddy allocator.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
-+ *
++ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+ *
-+ * Copyright (C) IBM Corp. 2006
-+ *
-+ * Authors: Hollis Blanchard <hollisb@us.ibm.com>
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/gfp.h>
-+#include <linux/mm.h>
-+#include <asm/page.h>
-+#include <xen/xencomm.h>
-+#include <xen/interface/xen.h>
-+#ifdef __ia64__
-+#include <asm/xen/xencomm.h> /* for is_kern_addr() */
-+#endif
++#include <ci/efhw/common.h> /* get uintXX types on win32 */
++#include <ci/efrm/sysdep.h>
++#include <ci/efrm/buddy.h>
++#include <ci/efrm/debug.h>
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
++#if 1
++#define DEBUG_ALLOC(x)
++#else
++#define DEBUG_ALLOC(x) x
++
++static inline void efrm_buddy_dump(struct efrm_buddy_allocator *b)
++{
++ unsigned o;
++
++ EFRM_NOTICE("%s: dump allocator with order %u",
++ __FUNCTION__, b->order);
++ for (o = 0; o <= b->order; o++) {
++ struct list_head *l = &b->free_lists[o];
++ while (l->next != &b->free_lists[o]) {
++ l = l->next;
++ EFRM_NOTICE("%s: order %x: %zx", __FUNCTION__, o,
++ l - b->links);
++ }
++ }
++}
+#endif
+
-+static int xencomm_init(struct xencomm_desc *desc,
-+ void *buffer, unsigned long bytes)
++/*
++ * The purpose of the following inline functions is to give the
++ * understandable names to the simple actions.
++ */
++static inline void
++efrm_buddy_free_list_add(struct efrm_buddy_allocator *b,
++ unsigned order, unsigned addr)
+{
-+ unsigned long recorded = 0;
-+ int i = 0;
++ list_add(&b->links[addr], &b->free_lists[order]);
++ b->orders[addr] = (uint8_t) b->order;
++}
++static inline void
++efrm_buddy_free_list_del(struct efrm_buddy_allocator *b, unsigned addr)
++{
++ list_del(&b->links[addr]);
++ b->links[addr].next = NULL;
++}
++static inline int
++efrm_buddy_free_list_empty(struct efrm_buddy_allocator *b, unsigned order)
++{
++ return list_empty(&b->free_lists[order]);
++}
++static inline unsigned
++efrm_buddy_free_list_pop(struct efrm_buddy_allocator *b, unsigned order)
++{
++ struct list_head *l = list_pop(&b->free_lists[order]);
++ l->next = NULL;
++ return (unsigned)(l - b->links);
++}
++static inline int
++efrm_buddy_addr_in_free_list(struct efrm_buddy_allocator *b, unsigned addr)
++{
++ return b->links[addr].next != NULL;
++}
++static inline unsigned
++efrm_buddy_free_list_first(struct efrm_buddy_allocator *b, unsigned order)
++{
++ return (unsigned)(b->free_lists[order].next - b->links);
++}
+
-+ while ((recorded < bytes) && (i < desc->nr_addrs)) {
-+ unsigned long vaddr = (unsigned long)buffer + recorded;
-+ unsigned long paddr;
-+ int offset;
-+ int chunksz;
++int efrm_buddy_ctor(struct efrm_buddy_allocator *b, unsigned order)
++{
++ unsigned o;
++ unsigned size = 1 << order;
+
-+ offset = vaddr % PAGE_SIZE; /* handle partial pages */
-+ chunksz = min(PAGE_SIZE - offset, bytes - recorded);
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u)", __FUNCTION__, order));
++ EFRM_ASSERT(b);
++ EFRM_ASSERT(order <= sizeof(unsigned) * 8 - 1);
+
-+ paddr = xencomm_vtop(vaddr);
-+ if (paddr == ~0UL) {
-+ printk("%s: couldn't translate vaddr %lx\n",
-+ __func__, vaddr);
-+ return -EINVAL;
-+ }
++ b->order = order;
++ b->free_lists = vmalloc((order + 1) * sizeof(struct list_head));
++ if (b->free_lists == NULL)
++ goto fail1;
+
-+ desc->address[i++] = paddr;
-+ recorded += chunksz;
-+ }
++ b->links = vmalloc(size * sizeof(struct list_head));
++ if (b->links == NULL)
++ goto fail2;
+
-+ if (recorded < bytes) {
-+ printk("%s: could only translate %ld of %ld bytes\n",
-+ __func__, recorded, bytes);
-+ return -ENOSPC;
-+ }
++ b->orders = vmalloc(size);
++ if (b->orders == NULL)
++ goto fail3;
+
-+ /* mark remaining addresses invalid (just for safety) */
-+ while (i < desc->nr_addrs)
-+ desc->address[i++] = XENCOMM_INVALID;
++ memset(b->links, 0, size * sizeof(struct list_head));
+
-+ desc->magic = XENCOMM_MAGIC;
++ for (o = 0; o <= b->order; ++o)
++ INIT_LIST_HEAD(b->free_lists + o);
++
++ efrm_buddy_free_list_add(b, b->order, 0);
+
+ return 0;
++
++fail3:
++ vfree(b->links);
++fail2:
++ vfree(b->free_lists);
++fail1:
++ return -ENOMEM;
+}
+
-+static struct xencomm_desc *xencomm_alloc(gfp_t gfp_mask,
-+ void *buffer, unsigned long bytes)
++void efrm_buddy_dtor(struct efrm_buddy_allocator *b)
+{
-+ struct xencomm_desc *desc;
-+ unsigned long buffer_ulong = (unsigned long)buffer;
-+ unsigned long start = buffer_ulong & PAGE_MASK;
-+ unsigned long end = (buffer_ulong + bytes) | ~PAGE_MASK;
-+ unsigned long nr_addrs = (end - start + 1) >> PAGE_SHIFT;
-+ unsigned long size = sizeof(*desc) +
-+ sizeof(desc->address[0]) * nr_addrs;
++ EFRM_ASSERT(b);
+
-+ /*
-+ * slab allocator returns at least sizeof(void*) aligned pointer.
-+ * When sizeof(*desc) > sizeof(void*), struct xencomm_desc might
-+ * cross page boundary.
++ vfree(b->free_lists);
++ vfree(b->links);
++ vfree(b->orders);
++}
++
++int efrm_buddy_alloc(struct efrm_buddy_allocator *b, unsigned order)
++{
++ unsigned smallest;
++ unsigned addr;
++
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u)", __FUNCTION__, order));
++ EFRM_ASSERT(b);
++
++ /* Find smallest chunk that is big enough. ?? Can optimise this by
++ ** keeping array of pointers to smallest chunk for each order.
+ */
-+ if (sizeof(*desc) > sizeof(void*)) {
-+ unsigned long order = get_order(size);
-+ desc = (struct xencomm_desc *)__get_free_pages(gfp_mask,
-+ order);
-+ if (desc == NULL)
-+ return NULL;
++ smallest = order;
++ while (smallest <= b->order &&
++ efrm_buddy_free_list_empty(b, smallest))
++ ++smallest;
+
-+ desc->nr_addrs =
-+ ((PAGE_SIZE << order) - sizeof(struct xencomm_desc)) /
-+ sizeof(*desc->address);
-+ } else {
-+ desc = kmalloc(size, gfp_mask);
-+ if (desc == NULL)
-+ return NULL;
++ if (smallest > b->order) {
++ DEBUG_ALLOC(EFRM_NOTICE
++ ("buddy - alloc order %d failed - max order %d",
++ order, b->order););
++ return -ENOMEM;
++ }
+
-+ desc->nr_addrs = nr_addrs;
++ /* Split blocks until we get one of the correct size. */
++ addr = efrm_buddy_free_list_pop(b, smallest);
++
++ DEBUG_ALLOC(EFRM_NOTICE("buddy - alloc %x order %d cut from order %d",
++ addr, order, smallest););
++ while (smallest-- > order)
++ efrm_buddy_free_list_add(b, smallest, addr + (1 << smallest));
++
++ EFRM_DO_DEBUG(b->orders[addr] = (uint8_t) order);
++
++ EFRM_ASSERT(addr < 1u << b->order);
++ return addr;
++}
++
++void
++efrm_buddy_free(struct efrm_buddy_allocator *b, unsigned addr,
++ unsigned order)
++{
++ unsigned buddy_addr;
++
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u, %u)", __FUNCTION__, addr, order));
++ EFRM_ASSERT(b);
++ EFRM_ASSERT(order <= b->order);
++ EFRM_ASSERT((unsigned long)addr + ((unsigned long)1 << order) <=
++ (unsigned long)1 << b->order);
++ EFRM_ASSERT(!efrm_buddy_addr_in_free_list(b, addr));
++ EFRM_ASSERT(b->orders[addr] == order);
++
++ /* merge free blocks */
++ while (order < b->order) {
++ buddy_addr = addr ^ (1 << order);
++ if (!efrm_buddy_addr_in_free_list(b, buddy_addr) ||
++ b->orders[buddy_addr] != order)
++ break;
++ efrm_buddy_free_list_del(b, addr);
++ if (buddy_addr < addr)
++ addr = buddy_addr;
++ ++order;
+ }
-+ return desc;
++
++ DEBUG_ALLOC(EFRM_NOTICE
++ ("buddy - free %x merged into order %d", addr, order););
++ efrm_buddy_free_list_add(b, order, addr);
+}
+
-+void xencomm_free(struct xencomm_handle *desc)
++void efrm_buddy_reserve_at_start(struct efrm_buddy_allocator *b, unsigned n)
+{
-+ if (desc && !((ulong)desc & XENCOMM_INLINE_FLAG)) {
-+ struct xencomm_desc *desc__ = (struct xencomm_desc*)desc;
-+ if (sizeof(*desc__) > sizeof(void*)) {
-+ unsigned long size = sizeof(*desc__) +
-+ sizeof(desc__->address[0]) * desc__->nr_addrs;
-+ unsigned long order = get_order(size);
-+ free_pages((unsigned long)__va(desc), order);
-+ } else
-+ kfree(__va(desc));
++ int addr;
++ unsigned o;
++ EFRM_DO_DEBUG(int n_save = n);
++
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u)", __FUNCTION__, n));
++ EFRM_ASSERT(b);
++ EFRM_ASSERT(n <= 1u << b->order && n > 0);
++ /* Whole space must be free. */
++ EFRM_ASSERT(!efrm_buddy_free_list_empty(b, b->order));
++
++ o = fls(n);
++
++ while (n) {
++ while (((unsigned)1 << o) > n)
++ --o;
++ EFRM_ASSERT(((unsigned)1 << o) <= n);
++ addr = efrm_buddy_alloc(b, o);
++ EFRM_ASSERT(addr + (1 << o) <= n_save);
++ n -= 1 << o;
+ }
+}
+
-+static int xencomm_create(void *buffer, unsigned long bytes, struct xencomm_desc **ret, gfp_t gfp_mask)
++static int
++__efrm_buddy_reserve_at_end(struct efrm_buddy_allocator *b, unsigned order,
++ int threshold)
+{
-+ struct xencomm_desc *desc;
-+ int rc;
++ unsigned o, addr;
+
-+ pr_debug("%s: %p[%ld]\n", __func__, buffer, bytes);
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u, %d)", __FUNCTION__, order, threshold));
++ EFRM_ASSERT(b);
+
-+ if (bytes == 0) {
-+ /* don't create a descriptor; Xen recognizes NULL. */
-+ BUG_ON(buffer != NULL);
-+ *ret = NULL;
-+ return 0;
++ /* Find largest block; there must be one big enough (or caller has
++ ** goofed).
++ */
++ for (o = b->order;; --o) {
++ if (efrm_buddy_free_list_empty(b, o))
++ continue;
++ addr = efrm_buddy_free_list_first(b, o);
++ if (addr + (1 << o) <= (unsigned)threshold)
++ continue;
++ break;
+ }
++ EFRM_ASSERT(o >= order);
+
-+ BUG_ON(buffer == NULL); /* 'bytes' is non-zero */
++ /* Split down (keeping second half) until we reach
++ * the requested size. */
++ addr = efrm_buddy_free_list_pop(b, o);
+
-+ desc = xencomm_alloc(gfp_mask, buffer, bytes);
-+ if (!desc) {
-+ printk("%s failure\n", "xencomm_alloc");
-+ return -ENOMEM;
++ while (o-- > order) {
++ efrm_buddy_free_list_add(b, o, addr);
++ addr += 1 << o;
+ }
+
-+ rc = xencomm_init(desc, buffer, bytes);
-+ if (rc) {
-+ printk("%s failure: %d\n", "xencomm_init", rc);
-+ xencomm_free((struct xencomm_handle *)__pa(desc));
-+ return rc;
-+ }
++ EFRM_DO_DEBUG(b->orders[addr] = (uint8_t) order);
+
-+ *ret = desc;
-+ return 0;
++ return addr;
+}
+
-+/* check if memory address is within VMALLOC region */
-+static int is_phys_contiguous(unsigned long addr)
++void efrm_buddy_reserve_at_end(struct efrm_buddy_allocator *b, unsigned n)
+{
-+ if (!is_kernel_addr(addr))
-+ return 0;
++ int addr, threshold;
++ unsigned o;
++ EFRM_DO_DEBUG(int n_save = n);
+
-+ return (addr < VMALLOC_START) || (addr >= VMALLOC_END);
-+}
++ DEBUG_ALLOC(EFRM_NOTICE("%s(%u)", __FUNCTION__, n));
++ DEBUG_ALLOC(efrm_buddy_dump(b));
++ EFRM_ASSERT(b);
++ EFRM_ASSERT(n <= 1u << b->order);
+
-+static struct xencomm_handle *xencomm_create_inline(void *ptr)
-+{
-+ unsigned long paddr;
++ if (!n)
++ return;
+
-+ BUG_ON(!is_phys_contiguous((unsigned long)ptr));
++ threshold = (1 << b->order) - n;
++ o = fls(n);
+
-+ paddr = (unsigned long)xencomm_pa(ptr);
-+ BUG_ON(paddr & XENCOMM_INLINE_FLAG);
-+ return (struct xencomm_handle *)(paddr | XENCOMM_INLINE_FLAG);
++ while (n) {
++ while (((unsigned)1 << o) > n)
++ --o;
++ EFRM_ASSERT(((unsigned)1 << o) <= n);
++ addr = __efrm_buddy_reserve_at_end(b, o, threshold);
++ EFRM_ASSERT(addr >= (1 << b->order) - n_save);
++ n -= 1 << o;
++ }
++ DEBUG_ALLOC(efrm_buddy_dump(b));
+}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/buffer_table.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/buffer_table.c 2008-05-19 00:33:29.325836502 +0300
+@@ -0,0 +1,210 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains abstraction of the buffer table on the NIC.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+/* "mini" routine, for stack-based communications: */
-+static int xencomm_create_mini(void *buffer,
-+ unsigned long bytes, struct xencomm_mini *xc_desc,
-+ struct xencomm_desc **ret)
-+{
-+ int rc = 0;
-+ struct xencomm_desc *desc;
-+ BUG_ON(((unsigned long)xc_desc) % sizeof(*xc_desc) != 0);
-+
-+ desc = (void *)xc_desc;
++/*
++** Might be worth keeping a bitmap of which entries are clear. Then we
++** wouldn't need to clear them all again when we free an allocation.
++*/
+
-+ desc->nr_addrs = XENCOMM_MINI_ADDRS;
++#include <ci/efrm/debug.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efrm/nic_table.h>
++#include <ci/efrm/buffer_table.h>
++#include <ci/efrm/buddy.h>
+
-+ if (!(rc = xencomm_init(desc, buffer, bytes)))
-+ *ret = desc;
++/*! Comment? */
++struct efrm_buffer_table {
++ spinlock_t lock;
++ struct efrm_buddy_allocator buddy;
++};
+
-+ return rc;
-+}
++/* Efab buffer state. */
++static struct efrm_buffer_table efrm_buffers;
+
-+struct xencomm_handle *xencomm_map(void *ptr, unsigned long bytes)
++int efrm_buffer_table_ctor(unsigned low, unsigned high)
+{
-+ int rc;
-+ struct xencomm_desc *desc;
++ int log2_n_entries, rc;
+
-+ if (is_phys_contiguous((unsigned long)ptr))
-+ return xencomm_create_inline(ptr);
++ EFRM_ASSERT(high > 0);
++ EFRM_ASSERT(low < high);
+
-+ rc = xencomm_create(ptr, bytes, &desc, GFP_KERNEL);
++ EFRM_TRACE("efrm_buffer_table_ctor: low=%u high=%u", low, high);
++ EFRM_NOTICE("efrm_buffer_table_ctor: low=%u high=%u", low, high);
+
-+ if (rc || desc == NULL)
-+ return NULL;
++ log2_n_entries = fls(high - 1);
+
-+ return xencomm_pa(desc);
-+}
++ rc = efrm_buddy_ctor(&efrm_buffers.buddy, log2_n_entries);
++ if (rc < 0) {
++ EFRM_ERR("efrm_buffer_table_ctor: efrm_buddy_ctor(%d) "
++ "failed (%d)", log2_n_entries, rc);
++ return rc;
++ }
+
-+struct xencomm_handle *__xencomm_map_no_alloc(void *ptr, unsigned long bytes,
-+ struct xencomm_mini *xc_desc)
-+{
-+ int rc;
-+ struct xencomm_desc *desc = NULL;
++ spin_lock_init(&efrm_buffers.lock);
+
-+ if (is_phys_contiguous((unsigned long)ptr))
-+ return xencomm_create_inline(ptr);
++ efrm_buddy_reserve_at_start(&efrm_buffers.buddy, low);
++ efrm_buddy_reserve_at_end(&efrm_buffers.buddy,
++ (1 << log2_n_entries) - high);
+
-+ rc = xencomm_create_mini(ptr, bytes, xc_desc,
-+ &desc);
++ EFRM_TRACE("efrm_buffer_table_ctor: done");
+
-+ if (rc)
-+ return NULL;
-+
-+ return xencomm_pa(desc);
++ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/xen_proc.c linux-2.6.18-xen.hg/drivers/xen/core/xen_proc.c
---- linux-2.6.18/drivers/xen/core/xen_proc.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/xen_proc.c 2007-12-23 11:15:33.674582320 +0100
-@@ -0,0 +1,23 @@
+
-+#include <linux/module.h>
-+#include <linux/proc_fs.h>
-+#include <xen/xen_proc.h>
++void efrm_buffer_table_dtor(void)
++{
++ /* ?? debug check that all allocations have been freed? */
+
-+static struct proc_dir_entry *xen_base;
++ spin_lock_destroy(&efrm_buffers.lock);
++ efrm_buddy_dtor(&efrm_buffers.buddy);
+
-+struct proc_dir_entry *create_xen_proc_entry(const char *name, mode_t mode)
-+{
-+ if ( xen_base == NULL )
-+ if ( (xen_base = proc_mkdir("xen", &proc_root)) == NULL )
-+ panic("Couldn't create /proc/xen");
-+ return create_proc_entry(name, mode, xen_base);
++ EFRM_TRACE("efrm_buffer_table_dtor: done");
+}
+
-+EXPORT_SYMBOL_GPL(create_xen_proc_entry);
++/**********************************************************************/
+
-+void remove_xen_proc_entry(const char *name)
++int
++efrm_buffer_table_alloc(unsigned order,
++ struct efhw_buffer_table_allocation *a)
+{
-+ remove_proc_entry(name, xen_base);
-+}
++ irq_flags_t lock_flags;
++ int rc;
+
-+EXPORT_SYMBOL_GPL(remove_xen_proc_entry);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/core/xen_sysfs.c linux-2.6.18-xen.hg/drivers/xen/core/xen_sysfs.c
---- linux-2.6.18/drivers/xen/core/xen_sysfs.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/core/xen_sysfs.c 2007-12-23 11:15:33.867925803 +0100
-@@ -0,0 +1,378 @@
-+/*
-+ * copyright (c) 2006 IBM Corporation
-+ * Authored by: Mike D. Day <ncmike@us.ibm.com>
-+ *
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License version 2 as
-+ * published by the Free Software Foundation.
-+ */
++ EFRM_ASSERT(&efrm_buffers.buddy);
++ EFRM_ASSERT(a);
+
-+#include <linux/err.h>
-+#include <linux/kernel.h>
-+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <asm/hypervisor.h>
-+#include <xen/features.h>
-+#include <xen/hypervisor_sysfs.h>
-+#include <xen/xenbus.h>
++ /* Round up to multiple of two, as the buffer clear logic works in
++ * pairs when not in "full" mode. */
++ order = max_t(unsigned, order, 1);
+
-+MODULE_LICENSE("GPL");
-+MODULE_AUTHOR("Mike D. Day <ncmike@us.ibm.com>");
++ spin_lock_irqsave(&efrm_buffers.lock, lock_flags);
++ rc = efrm_buddy_alloc(&efrm_buffers.buddy, order);
++ spin_unlock_irqrestore(&efrm_buffers.lock, lock_flags);
+
-+static ssize_t type_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ return sprintf(buffer, "xen\n");
-+}
++ if (rc < 0) {
++ EFRM_ERR("efrm_buffer_table_alloc: failed (n=%ld) rc %d",
++ 1ul << order, rc);
++ return rc;
++ }
+
-+HYPERVISOR_ATTR_RO(type);
++ EFRM_TRACE("efrm_buffer_table_alloc: base=%d n=%ld",
++ rc, 1ul << order);
++ a->order = order;
++ a->base = (unsigned)rc;
++ return 0;
++}
+
-+static int __init xen_sysfs_type_init(void)
++void efrm_buffer_table_free(struct efhw_buffer_table_allocation *a)
+{
-+ return sysfs_create_file(&hypervisor_subsys.kset.kobj, &type_attr.attr);
++ irq_flags_t lock_flags;
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_ASSERT(&efrm_buffers.buddy);
++ EFRM_ASSERT(a);
++ EFRM_ASSERT(a->base != -1);
++ EFRM_ASSERT((unsigned long)a->base + (1ul << a->order) <=
++ efrm_buddy_size(&efrm_buffers.buddy));
++
++ EFRM_TRACE("efrm_buffer_table_free: base=%d n=%ld",
++ a->base, (1ul << a->order));
++
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ efhw_nic_buffer_table_clear(nic, a->base, 1ul << a->order);
++
++ spin_lock_irqsave(&efrm_buffers.lock, lock_flags);
++ efrm_buddy_free(&efrm_buffers.buddy, a->base, a->order);
++ spin_unlock_irqrestore(&efrm_buffers.lock, lock_flags);
++
++ EFRM_DO_DEBUG(a->base = a->order = -1);
+}
+
-+static void xen_sysfs_type_destroy(void)
++/**********************************************************************/
++
++void
++efrm_buffer_table_set(struct efhw_buffer_table_allocation *a,
++ unsigned i, dma_addr_t dma_addr, int owner)
+{
-+ sysfs_remove_file(&hypervisor_subsys.kset.kobj, &type_attr.attr);
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_ASSERT(a);
++ EFRM_ASSERT(i < (unsigned)1 << a->order);
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ efhw_nic_buffer_table_set(nic, dma_addr, EFHW_NIC_PAGE_SIZE,
++ 0, owner, a->base + i);
++ /* NB. No commit Caller should call efrm_buffer_table_commit. There
++ are underlying hardware constraints regarding the number of
++ buffer table entries which can be pushed before commiting. */
+}
+
-+/* xen version attributes */
-+static ssize_t major_show(struct hyp_sysfs_attr *attr, char *buffer)
++unsigned long efrm_buffer_table_size(void)
+{
-+ int version = HYPERVISOR_xen_version(XENVER_version, NULL);
-+ if (version)
-+ return sprintf(buffer, "%d\n", version >> 16);
-+ return -ENODEV;
++ return efrm_buddy_size(&efrm_buffers.buddy);
+}
+
-+HYPERVISOR_ATTR_RO(major);
++/**********************************************************************/
+
-+static ssize_t minor_show(struct hyp_sysfs_attr *attr, char *buffer)
++int
++efrm_page_register(dma_addr_t dma_addr, int owner,
++ efhw_buffer_addr_t *buf_addr_out)
+{
-+ int version = HYPERVISOR_xen_version(XENVER_version, NULL);
-+ if (version)
-+ return sprintf(buffer, "%d\n", version & 0xff);
-+ return -ENODEV;
++ struct efhw_buffer_table_allocation alloc;
++ int rc;
++
++ rc = efrm_buffer_table_alloc(0, &alloc);
++ if (rc == 0) {
++ efrm_buffer_table_set(&alloc, 0, dma_addr, owner);
++ efrm_buffer_table_commit();
++ *buf_addr_out = EFHW_BUFFER_ADDR(alloc.base, 0);
++ }
++ return rc;
+}
++EXPORT_SYMBOL(efrm_page_register);
+
-+HYPERVISOR_ATTR_RO(minor);
-+
-+static ssize_t extra_show(struct hyp_sysfs_attr *attr, char *buffer)
++void efrm_page_unregister(efhw_buffer_addr_t buf_addr)
+{
-+ int ret = -ENOMEM;
-+ char *extra;
++ struct efhw_buffer_table_allocation alloc;
+
-+ extra = kmalloc(XEN_EXTRAVERSION_LEN, GFP_KERNEL);
-+ if (extra) {
-+ ret = HYPERVISOR_xen_version(XENVER_extraversion, extra);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", extra);
-+ kfree(extra);
-+ }
++ alloc.order = 0;
++ alloc.base = EFHW_BUFFER_PAGE(buf_addr);
++ efrm_buffer_table_free(&alloc);
++}
++EXPORT_SYMBOL(efrm_page_unregister);
+
-+ return ret;
++void efrm_buffer_table_commit(void)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ efhw_nic_buffer_table_commit(nic);
+}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/common.h 2008-05-19 00:33:29.329836732 +0300
+@@ -0,0 +1,68 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC hardware interface common
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_DRIVER_EFAB_HARDWARE_COMMON_H__
++#define __CI_DRIVER_EFAB_HARDWARE_COMMON_H__
++
++/*----------------------------------------------------------------------------
++ *
++ * EtherFabric constants
++ *
++ *---------------------------------------------------------------------------*/
++
++#define EFHW_1K 0x00000400u
++#define EFHW_2K 0x00000800u
++#define EFHW_4K 0x00001000u
++#define EFHW_8K 0x00002000u
++#define EFHW_16K 0x00004000u
++#define EFHW_32K 0x00008000u
++#define EFHW_64K 0x00010000u
++#define EFHW_128K 0x00020000u
++#define EFHW_256K 0x00040000u
++#define EFHW_512K 0x00080000u
++#define EFHW_1M 0x00100000u
++#define EFHW_2M 0x00200000u
++#define EFHW_4M 0x00400000u
++#define EFHW_8M 0x00800000u
++#define EFHW_16M 0x01000000u
++#define EFHW_32M 0x02000000u
++#define EFHW_48M 0x03000000u
++#define EFHW_64M 0x04000000u
++#define EFHW_128M 0x08000000u
++#define EFHW_256M 0x10000000u
++#define EFHW_512M 0x20000000u
++#define EFHW_1G 0x40000000u
++#define EFHW_2G 0x80000000u
++#define EFHW_4G 0x100000000ULL
++#define EFHW_8G 0x200000000ULL
++
++#endif /* __CI_DRIVER_EFAB_HARDWARE_COMMON_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_core.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_core.h 2008-05-19 00:33:29.329836732 +0300
+@@ -0,0 +1,1149 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) core register
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#define FALCON_EXTENDED_P_BAR 1
++
++/*************---- Bus Interface Unit Registers C Header ----*************/
++#define IOM_IND_ADR_REG_OFST 0x0 /* IO-mapped indirect access address
++ register */
++ #define IOM_AUTO_ADR_INC_EN_LBN 16
++ #define IOM_AUTO_ADR_INC_EN_WIDTH 1
++ #define IOM_IND_ADR_LBN 0
++ #define IOM_IND_ADR_WIDTH 16
++#define IOM_IND_DAT_REG_OFST 0x4 /* IO-mapped indirect access data register */
++ #define IOM_IND_DAT_LBN 0
++ #define IOM_IND_DAT_WIDTH 32
++#define ADR_REGION_REG_KER_OFST 0x0 /* Address region register */
++#define ADR_REGION_REG_OFST 0x0 /* Address region register */
++ #define ADR_REGION3_LBN 96
++ #define ADR_REGION3_WIDTH 18
++ #define ADR_REGION2_LBN 64
++ #define ADR_REGION2_WIDTH 18
++ #define ADR_REGION1_LBN 32
++ #define ADR_REGION1_WIDTH 18
++ #define ADR_REGION0_LBN 0
++ #define ADR_REGION0_WIDTH 18
++#define INT_EN_REG_KER_OFST 0x10 /* Kernel driver Interrupt enable register */
++ #define KER_INT_CHAR_LBN 4
++ #define KER_INT_CHAR_WIDTH 1
++ #define KER_INT_KER_LBN 3
++ #define KER_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_EN_KER_LBN 2
++ #define ILL_ADR_ERR_INT_EN_KER_WIDTH 1
++ #define SRM_PERR_INT_EN_KER_LBN 1
++ #define SRM_PERR_INT_EN_KER_WIDTH 1
++ #define DRV_INT_EN_KER_LBN 0
++ #define DRV_INT_EN_KER_WIDTH 1
++#define INT_EN_REG_CHAR_OFST 0x20 /* Char Driver interrupt enable register */
++ #define CHAR_INT_CHAR_LBN 4
++ #define CHAR_INT_CHAR_WIDTH 1
++ #define CHAR_INT_KER_LBN 3
++ #define CHAR_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_EN_CHAR_LBN 2
++ #define ILL_ADR_ERR_INT_EN_CHAR_WIDTH 1
++ #define SRM_PERR_INT_EN_CHAR_LBN 1
++ #define SRM_PERR_INT_EN_CHAR_WIDTH 1
++ #define DRV_INT_EN_CHAR_LBN 0
++ #define DRV_INT_EN_CHAR_WIDTH 1
++#define INT_ADR_REG_KER_OFST 0x30 /* Interrupt host address for Kernel driver */
++ #define INT_ADR_KER_LBN 0
++ #define INT_ADR_KER_WIDTH 64
++ #define DRV_INT_KER_LBN 32
++ #define DRV_INT_KER_WIDTH 1
++ #define EV_FF_HALF_INT_KER_LBN 3
++ #define EV_FF_HALF_INT_KER_WIDTH 1
++ #define EV_FF_FULL_INT_KER_LBN 2
++ #define EV_FF_FULL_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_KER_LBN 1
++ #define ILL_ADR_ERR_INT_KER_WIDTH 1
++ #define SRAM_PERR_INT_KER_LBN 0
++ #define SRAM_PERR_INT_KER_WIDTH 1
++#define INT_ADR_REG_CHAR_OFST 0x40 /* Interrupt host address for Char driver */
++ #define INT_ADR_CHAR_LBN 0
++ #define INT_ADR_CHAR_WIDTH 64
++ #define DRV_INT_CHAR_LBN 32
++ #define DRV_INT_CHAR_WIDTH 1
++ #define EV_FF_HALF_INT_CHAR_LBN 3
++ #define EV_FF_HALF_INT_CHAR_WIDTH 1
++ #define EV_FF_FULL_INT_CHAR_LBN 2
++ #define EV_FF_FULL_INT_CHAR_WIDTH 1
++ #define ILL_ADR_ERR_INT_CHAR_LBN 1
++ #define ILL_ADR_ERR_INT_CHAR_WIDTH 1
++ #define SRAM_PERR_INT_CHAR_LBN 0
++ #define SRAM_PERR_INT_CHAR_WIDTH 1
++#define INT_ISR0_B0_OFST 0x90 /* B0 only */
++#define INT_ISR1_B0_OFST 0xA0
++#define INT_ACK_REG_KER_A1_OFST 0x50 /* Kernel interrupt acknowledge register */
++ #define RESERVED_LBN 0
++ #define RESERVED_WIDTH 32
++#define INT_ACK_REG_CHAR_A1_OFST 0x60 /* CHAR interrupt acknowledge register */
++ #define RESERVED_LBN 0
++ #define RESERVED_WIDTH 32
++/*************---- Global CSR Registers C Header ----*************/
++#define STRAP_REG_KER_OFST 0x200 /* ASIC strap status register */
++#define STRAP_REG_OFST 0x200 /* ASIC strap status register */
++ #define ONCHIP_SRAM_LBN 16
++ #define ONCHIP_SRAM_WIDTH 0
++ #define STRAP_ISCSI_EN_LBN 3
++ #define STRAP_ISCSI_EN_WIDTH 1
++ #define STRAP_PINS_LBN 0
++ #define STRAP_PINS_WIDTH 3
++#define GPIO_CTL_REG_KER_OFST 0x210 /* GPIO control register */
++#define GPIO_CTL_REG_OFST 0x210 /* GPIO control register */
++ #define GPIO_OEN_LBN 24
++ #define GPIO_OEN_WIDTH 4
++ #define GPIO_OUT_LBN 16
++ #define GPIO_OUT_WIDTH 4
++ #define GPIO_IN_LBN 8
++ #define GPIO_IN_WIDTH 4
++ #define GPIO_PWRUP_VALUE_LBN 0
++ #define GPIO_PWRUP_VALUE_WIDTH 4
++#define GLB_CTL_REG_KER_OFST 0x220 /* Global control register */
++#define GLB_CTL_REG_OFST 0x220 /* Global control register */
++ #define SWRST_LBN 0
++ #define SWRST_WIDTH 1
++#define FATAL_INTR_REG_KER_OFST 0x230 /* Fatal interrupt register for Kernel */
++ #define PCI_BUSERR_INT_KER_EN_LBN 43
++ #define PCI_BUSERR_INT_KER_EN_WIDTH 1
++ #define SRAM_OOB_INT_KER_EN_LBN 42
++ #define SRAM_OOB_INT_KER_EN_WIDTH 1
++ #define BUFID_OOB_INT_KER_EN_LBN 41
++ #define BUFID_OOB_INT_KER_EN_WIDTH 1
++ #define MEM_PERR_INT_KER_EN_LBN 40
++ #define MEM_PERR_INT_KER_EN_WIDTH 1
++ #define RBUF_OWN_INT_KER_EN_LBN 39
++ #define RBUF_OWN_INT_KER_EN_WIDTH 1
++ #define TBUF_OWN_INT_KER_EN_LBN 38
++ #define TBUF_OWN_INT_KER_EN_WIDTH 1
++ #define RDESCQ_OWN_INT_KER_EN_LBN 37
++ #define RDESCQ_OWN_INT_KER_EN_WIDTH 1
++ #define TDESCQ_OWN_INT_KER_EN_LBN 36
++ #define TDESCQ_OWN_INT_KER_EN_WIDTH 1
++ #define EVQ_OWN_INT_KER_EN_LBN 35
++ #define EVQ_OWN_INT_KER_EN_WIDTH 1
++ #define EVFF_OFLO_INT_KER_EN_LBN 34
++ #define EVFF_OFLO_INT_KER_EN_WIDTH 1
++ #define ILL_ADR_INT_KER_EN_LBN 33
++ #define ILL_ADR_INT_KER_EN_WIDTH 1
++ #define SRM_PERR_INT_KER_EN_LBN 32
++ #define SRM_PERR_INT_KER_EN_WIDTH 1
++ #define PCI_BUSERR_INT_KER_LBN 11
++ #define PCI_BUSERR_INT_KER_WIDTH 1
++ #define SRAM_OOB_INT_KER_LBN 10
++ #define SRAM_OOB_INT_KER_WIDTH 1
++ #define BUFID_OOB_INT_KER_LBN 9
++ #define BUFID_OOB_INT_KER_WIDTH 1
++ #define MEM_PERR_INT_KER_LBN 8
++ #define MEM_PERR_INT_KER_WIDTH 1
++ #define RBUF_OWN_INT_KER_LBN 7
++ #define RBUF_OWN_INT_KER_WIDTH 1
++ #define TBUF_OWN_INT_KER_LBN 6
++ #define TBUF_OWN_INT_KER_WIDTH 1
++ #define RDESCQ_OWN_INT_KER_LBN 5
++ #define RDESCQ_OWN_INT_KER_WIDTH 1
++ #define TDESCQ_OWN_INT_KER_LBN 4
++ #define TDESCQ_OWN_INT_KER_WIDTH 1
++ #define EVQ_OWN_INT_KER_LBN 3
++ #define EVQ_OWN_INT_KER_WIDTH 1
++ #define EVFF_OFLO_INT_KER_LBN 2
++ #define EVFF_OFLO_INT_KER_WIDTH 1
++ #define ILL_ADR_INT_KER_LBN 1
++ #define ILL_ADR_INT_KER_WIDTH 1
++ #define SRM_PERR_INT_KER_LBN 0
++ #define SRM_PERR_INT_KER_WIDTH 1
++#define FATAL_INTR_REG_OFST 0x240 /* Fatal interrupt register for Char */
++ #define PCI_BUSERR_INT_CHAR_EN_LBN 43
++ #define PCI_BUSERR_INT_CHAR_EN_WIDTH 1
++ #define SRAM_OOB_INT_CHAR_EN_LBN 42
++ #define SRAM_OOB_INT_CHAR_EN_WIDTH 1
++ #define BUFID_OOB_INT_CHAR_EN_LBN 41
++ #define BUFID_OOB_INT_CHAR_EN_WIDTH 1
++ #define MEM_PERR_INT_CHAR_EN_LBN 40
++ #define MEM_PERR_INT_CHAR_EN_WIDTH 1
++ #define RBUF_OWN_INT_CHAR_EN_LBN 39
++ #define RBUF_OWN_INT_CHAR_EN_WIDTH 1
++ #define TBUF_OWN_INT_CHAR_EN_LBN 38
++ #define TBUF_OWN_INT_CHAR_EN_WIDTH 1
++ #define RDESCQ_OWN_INT_CHAR_EN_LBN 37
++ #define RDESCQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define TDESCQ_OWN_INT_CHAR_EN_LBN 36
++ #define TDESCQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define EVQ_OWN_INT_CHAR_EN_LBN 35
++ #define EVQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define EVFF_OFLO_INT_CHAR_EN_LBN 34
++ #define EVFF_OFLO_INT_CHAR_EN_WIDTH 1
++ #define ILL_ADR_INT_CHAR_EN_LBN 33
++ #define ILL_ADR_INT_CHAR_EN_WIDTH 1
++ #define SRM_PERR_INT_CHAR_EN_LBN 32
++ #define SRM_PERR_INT_CHAR_EN_WIDTH 1
++ #define FATAL_INTR_REG_EN_BITS 0xffffffffffffffffULL
++ #define PCI_BUSERR_INT_CHAR_LBN 11
++ #define PCI_BUSERR_INT_CHAR_WIDTH 1
++ #define SRAM_OOB_INT_CHAR_LBN 10
++ #define SRAM_OOB_INT_CHAR_WIDTH 1
++ #define BUFID_OOB_INT_CHAR_LBN 9
++ #define BUFID_OOB_INT_CHAR_WIDTH 1
++ #define MEM_PERR_INT_CHAR_LBN 8
++ #define MEM_PERR_INT_CHAR_WIDTH 1
++ #define RBUF_OWN_INT_CHAR_LBN 7
++ #define RBUF_OWN_INT_CHAR_WIDTH 1
++ #define TBUF_OWN_INT_CHAR_LBN 6
++ #define TBUF_OWN_INT_CHAR_WIDTH 1
++ #define RDESCQ_OWN_INT_CHAR_LBN 5
++ #define RDESCQ_OWN_INT_CHAR_WIDTH 1
++ #define TDESCQ_OWN_INT_CHAR_LBN 4
++ #define TDESCQ_OWN_INT_CHAR_WIDTH 1
++ #define EVQ_OWN_INT_CHAR_LBN 3
++ #define EVQ_OWN_INT_CHAR_WIDTH 1
++ #define EVFF_OFLO_INT_CHAR_LBN 2
++ #define EVFF_OFLO_INT_CHAR_WIDTH 1
++ #define ILL_ADR_INT_CHAR_LBN 1
++ #define ILL_ADR_INT_CHAR_WIDTH 1
++ #define SRM_PERR_INT_CHAR_LBN 0
++ #define SRM_PERR_INT_CHAR_WIDTH 1
++#define DP_CTRL_REG_OFST 0x250 /* Datapath control register */
++ #define FLS_EVQ_ID_LBN 0
++ #define FLS_EVQ_ID_WIDTH 12
++#define MEM_STAT_REG_KER_OFST 0x260 /* Memory status register */
++#define MEM_STAT_REG_OFST 0x260 /* Memory status register */
++ #define MEM_PERR_VEC_LBN 53
++ #define MEM_PERR_VEC_WIDTH 38
++ #define MBIST_CORR_LBN 38
++ #define MBIST_CORR_WIDTH 15
++ #define MBIST_ERR_LBN 0
++ #define MBIST_ERR_WIDTH 38
++#define DEBUG_REG_KER_OFST 0x270 /* Debug register */
++#define DEBUG_REG_OFST 0x270 /* Debug register */
++ #define DEBUG_BLK_SEL2_LBN 47
++ #define DEBUG_BLK_SEL2_WIDTH 3
++ #define DEBUG_BLK_SEL1_LBN 44
++ #define DEBUG_BLK_SEL1_WIDTH 3
++ #define DEBUG_BLK_SEL0_LBN 41
++ #define DEBUG_BLK_SEL0_WIDTH 3
++ #define MISC_DEBUG_ADDR_LBN 36
++ #define MISC_DEBUG_ADDR_WIDTH 5
++ #define SERDES_DEBUG_ADDR_LBN 31
++ #define SERDES_DEBUG_ADDR_WIDTH 5
++ #define EM_DEBUG_ADDR_LBN 26
++ #define EM_DEBUG_ADDR_WIDTH 5
++ #define SR_DEBUG_ADDR_LBN 21
++ #define SR_DEBUG_ADDR_WIDTH 5
++ #define EV_DEBUG_ADDR_LBN 16
++ #define EV_DEBUG_ADDR_WIDTH 5
++ #define RX_DEBUG_ADDR_LBN 11
++ #define RX_DEBUG_ADDR_WIDTH 5
++ #define TX_DEBUG_ADDR_LBN 6
++ #define TX_DEBUG_ADDR_WIDTH 5
++ #define BIU_DEBUG_ADDR_LBN 1
++ #define BIU_DEBUG_ADDR_WIDTH 5
++ #define DEBUG_EN_LBN 0
++ #define DEBUG_EN_WIDTH 1
++#define DRIVER_REG0_KER_OFST 0x280 /* Driver scratch register 0 */
++#define DRIVER_REG0_OFST 0x280 /* Driver scratch register 0 */
++ #define DRIVER_DW0_LBN 0
++ #define DRIVER_DW0_WIDTH 32
++#define DRIVER_REG1_KER_OFST 0x290 /* Driver scratch register 1 */
++#define DRIVER_REG1_OFST 0x290 /* Driver scratch register 1 */
++ #define DRIVER_DW1_LBN 0
++ #define DRIVER_DW1_WIDTH 32
++#define DRIVER_REG2_KER_OFST 0x2A0 /* Driver scratch register 2 */
++#define DRIVER_REG2_OFST 0x2A0 /* Driver scratch register 2 */
++ #define DRIVER_DW2_LBN 0
++ #define DRIVER_DW2_WIDTH 32
++#define DRIVER_REG3_KER_OFST 0x2B0 /* Driver scratch register 3 */
++#define DRIVER_REG3_OFST 0x2B0 /* Driver scratch register 3 */
++ #define DRIVER_DW3_LBN 0
++ #define DRIVER_DW3_WIDTH 32
++#define DRIVER_REG4_KER_OFST 0x2C0 /* Driver scratch register 4 */
++#define DRIVER_REG4_OFST 0x2C0 /* Driver scratch register 4 */
++ #define DRIVER_DW4_LBN 0
++ #define DRIVER_DW4_WIDTH 32
++#define DRIVER_REG5_KER_OFST 0x2D0 /* Driver scratch register 5 */
++#define DRIVER_REG5_OFST 0x2D0 /* Driver scratch register 5 */
++ #define DRIVER_DW5_LBN 0
++ #define DRIVER_DW5_WIDTH 32
++#define DRIVER_REG6_KER_OFST 0x2E0 /* Driver scratch register 6 */
++#define DRIVER_REG6_OFST 0x2E0 /* Driver scratch register 6 */
++ #define DRIVER_DW6_LBN 0
++ #define DRIVER_DW6_WIDTH 32
++#define DRIVER_REG7_KER_OFST 0x2F0 /* Driver scratch register 7 */
++#define DRIVER_REG7_OFST 0x2F0 /* Driver scratch register 7 */
++ #define DRIVER_DW7_LBN 0
++ #define DRIVER_DW7_WIDTH 32
++#define ALTERA_BUILD_REG_OFST 0x300 /* Altera build register */
++#define ALTERA_BUILD_REG_OFST 0x300 /* Altera build register */
++ #define ALTERA_BUILD_VER_LBN 0
++ #define ALTERA_BUILD_VER_WIDTH 32
++
++/* so called CSR spare register
++ - contains separate parity enable bits for the various internal memory
++ blocks */
++#define MEM_PARITY_ERR_EN_REG_KER 0x310
++#define MEM_PARITY_ALL_BLOCKS_EN_LBN 64
++#define MEM_PARITY_ALL_BLOCKS_EN_WIDTH 38
++#define MEM_PARITY_TX_DATA_EN_LBN 72
++#define MEM_PARITY_TX_DATA_EN_WIDTH 2
++
++/*************---- Event & Timer Module Registers C Header ----*************/
++
++#if FALCON_EXTENDED_P_BAR
++#define EVQ_RPTR_REG_KER_OFST 0x11B00 /* Event queue read pointer register */
++#else
++#define EVQ_RPTR_REG_KER_OFST 0x1B00 /* Event queue read pointer register */
++#endif
+
-+HYPERVISOR_ATTR_RO(extra);
++#define EVQ_RPTR_REG_OFST 0xFA0000 /* Event queue read pointer register
++ array. */
++ #define EVQ_RPTR_LBN 0
++ #define EVQ_RPTR_WIDTH 15
+
-+static struct attribute *version_attrs[] = {
-+ &major_attr.attr,
-+ &minor_attr.attr,
-+ &extra_attr.attr,
-+ NULL
-+};
++#if FALCON_EXTENDED_P_BAR
++#define EVQ_PTR_TBL_KER_OFST 0x11A00 /* Event queue pointer table for kernel
++ access */
++#else
++#define EVQ_PTR_TBL_KER_OFST 0x1A00 /* Event queue pointer table for kernel
++ access */
++#endif
++
++#define EVQ_PTR_TBL_CHAR_OFST 0xF60000 /* Event queue pointer table for char
++ direct access */
++ #define EVQ_WKUP_OR_INT_EN_LBN 39
++ #define EVQ_WKUP_OR_INT_EN_WIDTH 1
++ #define EVQ_NXT_WPTR_LBN 24
++ #define EVQ_NXT_WPTR_WIDTH 15
++ #define EVQ_EN_LBN 23
++ #define EVQ_EN_WIDTH 1
++ #define EVQ_SIZE_LBN 20
++ #define EVQ_SIZE_WIDTH 3
++ #define EVQ_BUF_BASE_ID_LBN 0
++ #define EVQ_BUF_BASE_ID_WIDTH 20
++#define TIMER_CMD_REG_KER_OFST 0x420 /* Timer table for kernel access.
++ Page-mapped */
++#define TIMER_CMD_REG_PAGE4_OFST 0x8420 /* Timer table for user-level access.
++ Page-mapped. For lowest 1K queues.
++ */
++#define TIMER_CMD_REG_PAGE123K_OFST 0x1000420 /* Timer table for user-level
++ access. Page-mapped.
++ For upper 3K queues. */
++#define TIMER_TBL_OFST 0xF70000 /* Timer table for char driver direct access */
++ #define TIMER_MODE_LBN 12
++ #define TIMER_MODE_WIDTH 2
++ #define TIMER_VAL_LBN 0
++ #define TIMER_VAL_WIDTH 12
++ #define TIMER_MODE_INT_HLDOFF 2
++ #define EVQ_BUF_SIZE_LBN 0
++ #define EVQ_BUF_SIZE_WIDTH 1
++#define DRV_EV_REG_KER_OFST 0x440 /* Driver generated event register */
++#define DRV_EV_REG_OFST 0x440 /* Driver generated event register */
++ #define DRV_EV_QID_LBN 64
++ #define DRV_EV_QID_WIDTH 12
++ #define DRV_EV_DATA_LBN 0
++ #define DRV_EV_DATA_WIDTH 64
++#define EVQ_CTL_REG_KER_OFST 0x450 /* Event queue control register */
++#define EVQ_CTL_REG_OFST 0x450 /* Event queue control register */
++ #define RX_EVQ_WAKEUP_MASK_B0_LBN 15
++ #define RX_EVQ_WAKEUP_MASK_B0_WIDTH 6
++ #define EVQ_OWNERR_CTL_LBN 14
++ #define EVQ_OWNERR_CTL_WIDTH 1
++ #define EVQ_FIFO_AF_TH_LBN 8
++ #define EVQ_FIFO_AF_TH_WIDTH 6
++ #define EVQ_FIFO_NOTAF_TH_LBN 0
++ #define EVQ_FIFO_NOTAF_TH_WIDTH 6
++/*************---- SRAM Module Registers C Header ----*************/
++#define BUF_TBL_CFG_REG_KER_OFST 0x600 /* Buffer table configuration register */
++#define BUF_TBL_CFG_REG_OFST 0x600 /* Buffer table configuration register */
++ #define BUF_TBL_MODE_LBN 3
++ #define BUF_TBL_MODE_WIDTH 1
++#define SRM_RX_DC_CFG_REG_KER_OFST 0x610 /* SRAM receive descriptor cache
++ configuration register */
++#define SRM_RX_DC_CFG_REG_OFST 0x610 /* SRAM receive descriptor cache
++ configuration register */
++ #define SRM_RX_DC_BASE_ADR_LBN 0
++ #define SRM_RX_DC_BASE_ADR_WIDTH 21
++#define SRM_TX_DC_CFG_REG_KER_OFST 0x620 /* SRAM transmit descriptor cache
++ configuration register */
++#define SRM_TX_DC_CFG_REG_OFST 0x620 /* SRAM transmit descriptor cache
++ configuration register */
++ #define SRM_TX_DC_BASE_ADR_LBN 0
++ #define SRM_TX_DC_BASE_ADR_WIDTH 21
++#define SRM_CFG_REG_KER_OFST 0x630 /* SRAM configuration register */
++#define SRM_CFG_REG_OFST 0x630 /* SRAM configuration register */
++ #define SRAM_OOB_ADR_INTEN_LBN 5
++ #define SRAM_OOB_ADR_INTEN_WIDTH 1
++ #define SRAM_OOB_BUF_INTEN_LBN 4
++ #define SRAM_OOB_BUF_INTEN_WIDTH 1
++ #define SRAM_BT_INIT_EN_LBN 3
++ #define SRAM_BT_INIT_EN_WIDTH 1
++ #define SRM_NUM_BANK_LBN 2
++ #define SRM_NUM_BANK_WIDTH 1
++ #define SRM_BANK_SIZE_LBN 0
++ #define SRM_BANK_SIZE_WIDTH 2
++#define BUF_TBL_UPD_REG_KER_OFST 0x650 /* Buffer table update register */
++#define BUF_TBL_UPD_REG_OFST 0x650 /* Buffer table update register */
++ #define BUF_UPD_CMD_LBN 63
++ #define BUF_UPD_CMD_WIDTH 1
++ #define BUF_CLR_CMD_LBN 62
++ #define BUF_CLR_CMD_WIDTH 1
++ #define BUF_CLR_END_ID_LBN 32
++ #define BUF_CLR_END_ID_WIDTH 20
++ #define BUF_CLR_START_ID_LBN 0
++ #define BUF_CLR_START_ID_WIDTH 20
++#define SRM_UPD_EVQ_REG_KER_OFST 0x660 /* Buffer table update register */
++#define SRM_UPD_EVQ_REG_OFST 0x660 /* Buffer table update register */
++ #define SRM_UPD_EVQ_ID_LBN 0
++ #define SRM_UPD_EVQ_ID_WIDTH 12
++#define SRAM_PARITY_REG_KER_OFST 0x670 /* SRAM parity register. */
++#define SRAM_PARITY_REG_OFST 0x670 /* SRAM parity register. */
++ #define FORCE_SRAM_PERR_LBN 0
++ #define FORCE_SRAM_PERR_WIDTH 1
++
++#if FALCON_EXTENDED_P_BAR
++#define BUF_HALF_TBL_KER_OFST 0x18000 /* Buffer table in half buffer table
++ mode direct access by kernel driver */
++#else
++#define BUF_HALF_TBL_KER_OFST 0x8000 /* Buffer table in half buffer table
++ mode direct access by kernel driver */
++#endif
+
-+static struct attribute_group version_group = {
-+ .name = "version",
-+ .attrs = version_attrs,
-+};
+
-+static int __init xen_sysfs_version_init(void)
-+{
-+ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
-+ &version_group);
-+}
++#define BUF_HALF_TBL_OFST 0x800000 /* Buffer table in half buffer table mode
++ direct access by char driver */
++ #define BUF_ADR_HBUF_ODD_LBN 44
++ #define BUF_ADR_HBUF_ODD_WIDTH 20
++ #define BUF_OWNER_ID_HBUF_ODD_LBN 32
++ #define BUF_OWNER_ID_HBUF_ODD_WIDTH 12
++ #define BUF_ADR_HBUF_EVEN_LBN 12
++ #define BUF_ADR_HBUF_EVEN_WIDTH 20
++ #define BUF_OWNER_ID_HBUF_EVEN_LBN 0
++ #define BUF_OWNER_ID_HBUF_EVEN_WIDTH 12
+
-+static void xen_sysfs_version_destroy(void)
-+{
-+ sysfs_remove_group(&hypervisor_subsys.kset.kobj, &version_group);
-+}
+
-+/* UUID */
++#if FALCON_EXTENDED_P_BAR
++#define BUF_FULL_TBL_KER_OFST 0x18000 /* Buffer table in full buffer table
++ mode direct access by kernel driver */
++#else
++#define BUF_FULL_TBL_KER_OFST 0x8000 /* Buffer table in full buffer table mode
++ direct access by kernel driver */
++#endif
+
-+static ssize_t uuid_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ char *vm, *val;
-+ int ret;
+
-+ vm = xenbus_read(XBT_NIL, "vm", "", NULL);
-+ if (IS_ERR(vm))
-+ return PTR_ERR(vm);
-+ val = xenbus_read(XBT_NIL, vm, "uuid", NULL);
-+ kfree(vm);
-+ if (IS_ERR(val))
-+ return PTR_ERR(val);
-+ ret = sprintf(buffer, "%s\n", val);
-+ kfree(val);
-+ return ret;
-+}
+
-+HYPERVISOR_ATTR_RO(uuid);
+
-+static int __init xen_sysfs_uuid_init(void)
-+{
-+ return sysfs_create_file(&hypervisor_subsys.kset.kobj, &uuid_attr.attr);
-+}
++#define BUF_FULL_TBL_OFST 0x800000 /* Buffer table in full buffer table mode
++ direct access by char driver */
++ #define IP_DAT_BUF_SIZE_LBN 50
++ #define IP_DAT_BUF_SIZE_WIDTH 1
++ #define BUF_ADR_REGION_LBN 48
++ #define BUF_ADR_REGION_WIDTH 2
++ #define BUF_ADR_FBUF_LBN 14
++ #define BUF_ADR_FBUF_WIDTH 34
++ #define BUF_OWNER_ID_FBUF_LBN 0
++ #define BUF_OWNER_ID_FBUF_WIDTH 14
++#define SRM_DBG_REG_OFST 0x3000000 /* SRAM debug access */
++ #define SRM_DBG_LBN 0
++ #define SRM_DBG_WIDTH 64
++/*************---- RX Datapath Registers C Header ----*************/
+
-+static void xen_sysfs_uuid_destroy(void)
-+{
-+ sysfs_remove_file(&hypervisor_subsys.kset.kobj, &uuid_attr.attr);
-+}
++#define RX_CFG_REG_KER_OFST 0x800 /* Receive configuration register */
++#define RX_CFG_REG_OFST 0x800 /* Receive configuration register */
+
-+/* xen compilation attributes */
++#if !defined(FALCON_64K_RXFIFO) && !defined(FALCON_PRE_02020029)
++# if !defined(FALCON_128K_RXFIFO)
++# define FALCON_128K_RXFIFO
++# endif
++#endif
+
-+static ssize_t compiler_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ int ret = -ENOMEM;
-+ struct xen_compile_info *info;
++#if defined(FALCON_128K_RXFIFO)
++
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 48
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 47
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 46
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 45
++ #define RX_HASH_ALG_B0_WIDTH 1
++ #define RX_HASH_INSERT_HDR_B0_LBN 44
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 43
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_RDW_PATCH_EN_LBN 42 /* Non head of line blocking */
++ #define RX_RDW_PATCH_EN_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 39
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 3
++ #define RX_OWNERR_CTL_B0_LBN 38
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 33
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 28
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 19
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 10
++ #define RX_XON_MAC_TH_B0_WIDTH 9
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 9
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
++
++#elif !defined(FALCON_PRE_02020029)
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 46
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 45
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 44
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 43
++ #define RX_HASH_ALG_B0_WIDTH 41
++ #define RX_HASH_INSERT_HDR_B0_LBN 42
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 41
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 37
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 3
++ #define RX_OWNERR_CTL_B0_LBN 36
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 31
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 26
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 17
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 9
++ #define RX_XON_MAC_TH_B0_WIDTH 8
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 8
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
+
-+ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
-+ if (info) {
-+ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", info->compiler);
-+ kfree(info);
-+ }
++#else
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 44
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 43
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 42
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 41
++ #define RX_HASH_ALG_B0_WIDTH 41
++ #define RX_HASH_INSERT_HDR_B0_LBN 40
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 35
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 35
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 2
++ #define RX_OWNERR_CTL_B0_LBN 34
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 29
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 24
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 15
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 8
++ #define RX_XON_MAC_TH_B0_WIDTH 7
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 7
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
++
++#endif
++
++/* A0/A1 */
++ #define RX_PUSH_EN_A1_LBN 35
++ #define RX_PUSH_EN_A1_WIDTH 1
++ #define RX_PCI_BURST_SIZE_A1_LBN 31
++ #define RX_PCI_BURST_SIZE_A1_WIDTH 3
++ #define RX_OWNERR_CTL_A1_LBN 30
++ #define RX_OWNERR_CTL_A1_WIDTH 1
++ #define RX_XON_TX_TH_A1_LBN 25
++ #define RX_XON_TX_TH_A1_WIDTH 5
++ #define RX_XOFF_TX_TH_A1_LBN 20
++ #define RX_XOFF_TX_TH_A1_WIDTH 5
++ #define RX_USR_BUF_SIZE_A1_LBN 11
++ #define RX_USR_BUF_SIZE_A1_WIDTH 9
++ #define RX_XON_MAC_TH_A1_LBN 6
++ #define RX_XON_MAC_TH_A1_WIDTH 5
++ #define RX_XOFF_MAC_TH_A1_LBN 1
++ #define RX_XOFF_MAC_TH_A1_WIDTH 5
++ #define RX_XOFF_MAC_EN_A1_LBN 0
++ #define RX_XOFF_MAC_EN_A1_WIDTH 1
++
++#define RX_FILTER_CTL_REG_OFST 0x810 /* Receive filter control registers */
++ #define SCATTER_ENBL_NO_MATCH_Q_B0_LBN 40
++ #define SCATTER_ENBL_NO_MATCH_Q_B0_WIDTH 1
++ #define UDP_FULL_SRCH_LIMIT_LBN 32
++ #define UDP_FULL_SRCH_LIMIT_WIDTH 8
++ #define NUM_KER_LBN 24
++ #define NUM_KER_WIDTH 2
++ #define UDP_WILD_SRCH_LIMIT_LBN 16
++ #define UDP_WILD_SRCH_LIMIT_WIDTH 8
++ #define TCP_WILD_SRCH_LIMIT_LBN 8
++ #define TCP_WILD_SRCH_LIMIT_WIDTH 8
++ #define TCP_FULL_SRCH_LIMIT_LBN 0
++ #define TCP_FULL_SRCH_LIMIT_WIDTH 8
++#define RX_FLUSH_DESCQ_REG_KER_OFST 0x820 /* Receive flush descriptor queue
++ register */
++#define RX_FLUSH_DESCQ_REG_OFST 0x820 /* Receive flush descriptor queue
++ register */
++ #define RX_FLUSH_DESCQ_CMD_LBN 24
++ #define RX_FLUSH_DESCQ_CMD_WIDTH 1
++ #define RX_FLUSH_EVQ_ID_LBN 12
++ #define RX_FLUSH_EVQ_ID_WIDTH 12
++ #define RX_FLUSH_DESCQ_LBN 0
++ #define RX_FLUSH_DESCQ_WIDTH 12
++#define RX_DESC_UPD_REG_KER_OFST 0x830 /* Kernel receive descriptor update
++ register. Page-mapped */
++#define RX_DESC_UPD_REG_PAGE4_OFST 0x8830 /* Char & user receive descriptor
++ update register. Page-mapped.
++ For lowest 1K queues. */
++#define RX_DESC_UPD_REG_PAGE123K_OFST 0x1000830 /* Char & user receive
++ descriptor update register.
++ Page-mapped. For upper
++ 3K queues. */
++ #define RX_DESC_WPTR_LBN 96
++ #define RX_DESC_WPTR_WIDTH 12
++ #define RX_DESC_PUSH_CMD_LBN 95
++ #define RX_DESC_PUSH_CMD_WIDTH 1
++ #define RX_DESC_LBN 0
++ #define RX_DESC_WIDTH 64
++ #define RX_KER_DESC_LBN 0
++ #define RX_KER_DESC_WIDTH 64
++ #define RX_USR_DESC_LBN 0
++ #define RX_USR_DESC_WIDTH 32
++#define RX_DC_CFG_REG_KER_OFST 0x840 /* Receive descriptor cache
++ configuration register */
++#define RX_DC_CFG_REG_OFST 0x840 /* Receive descriptor cache
++ configuration register */
++ #define RX_DC_SIZE_LBN 0
++ #define RX_DC_SIZE_WIDTH 2
++#define RX_DC_PF_WM_REG_KER_OFST 0x850 /* Receive descriptor cache pre-fetch
++ watermark register */
++#define RX_DC_PF_WM_REG_OFST 0x850 /* Receive descriptor cache pre-fetch
++ watermark register */
++ #define RX_DC_PF_LWM_LO_LBN 0
++ #define RX_DC_PF_LWM_LO_WIDTH 6
++
++#define RX_RSS_TKEY_B0_OFST 0x860 /* RSS Toeplitz hash key (B0 only) */
++
++#define RX_NODESC_DROP_REG 0x880
++ #define RX_NODESC_DROP_CNT_LBN 0
++ #define RX_NODESC_DROP_CNT_WIDTH 16
++
++#define XM_TX_CFG_REG_OFST 0x1230
++ #define XM_AUTO_PAD_LBN 5
++ #define XM_AUTO_PAD_WIDTH 1
++
++#define RX_FILTER_TBL0_OFST 0xF00000 /* Receive filter table - even entries */
++ #define RSS_EN_0_B0_LBN 110
++ #define RSS_EN_0_B0_WIDTH 1
++ #define SCATTER_EN_0_B0_LBN 109
++ #define SCATTER_EN_0_B0_WIDTH 1
++ #define TCP_UDP_0_LBN 108
++ #define TCP_UDP_0_WIDTH 1
++ #define RXQ_ID_0_LBN 96
++ #define RXQ_ID_0_WIDTH 12
++ #define DEST_IP_0_LBN 64
++ #define DEST_IP_0_WIDTH 32
++ #define DEST_PORT_TCP_0_LBN 48
++ #define DEST_PORT_TCP_0_WIDTH 16
++ #define SRC_IP_0_LBN 16
++ #define SRC_IP_0_WIDTH 32
++ #define SRC_TCP_DEST_UDP_0_LBN 0
++ #define SRC_TCP_DEST_UDP_0_WIDTH 16
++#define RX_FILTER_TBL1_OFST 0xF00010 /* Receive filter table - odd entries */
++ #define RSS_EN_1_B0_LBN 110
++ #define RSS_EN_1_B0_WIDTH 1
++ #define SCATTER_EN_1_B0_LBN 109
++ #define SCATTER_EN_1_B0_WIDTH 1
++ #define TCP_UDP_1_LBN 108
++ #define TCP_UDP_1_WIDTH 1
++ #define RXQ_ID_1_LBN 96
++ #define RXQ_ID_1_WIDTH 12
++ #define DEST_IP_1_LBN 64
++ #define DEST_IP_1_WIDTH 32
++ #define DEST_PORT_TCP_1_LBN 48
++ #define DEST_PORT_TCP_1_WIDTH 16
++ #define SRC_IP_1_LBN 16
++ #define SRC_IP_1_WIDTH 32
++ #define SRC_TCP_DEST_UDP_1_LBN 0
++ #define SRC_TCP_DEST_UDP_1_WIDTH 16
++
++#if FALCON_EXTENDED_P_BAR
++#define RX_DESC_PTR_TBL_KER_OFST 0x11800 /* Receive descriptor pointer
++ kernel access */
++#else
++#define RX_DESC_PTR_TBL_KER_OFST 0x1800 /* Receive descriptor pointer
++ kernel access */
++#endif
++
++
++#define RX_DESC_PTR_TBL_OFST 0xF40000 /* Receive descriptor pointer table */
++ #define RX_ISCSI_DDIG_EN_LBN 88
++ #define RX_ISCSI_DDIG_EN_WIDTH 1
++ #define RX_ISCSI_HDIG_EN_LBN 87
++ #define RX_ISCSI_HDIG_EN_WIDTH 1
++ #define RX_DESC_PREF_ACT_LBN 86
++ #define RX_DESC_PREF_ACT_WIDTH 1
++ #define RX_DC_HW_RPTR_LBN 80
++ #define RX_DC_HW_RPTR_WIDTH 6
++ #define RX_DESCQ_HW_RPTR_LBN 68
++ #define RX_DESCQ_HW_RPTR_WIDTH 12
++ #define RX_DESCQ_SW_WPTR_LBN 56
++ #define RX_DESCQ_SW_WPTR_WIDTH 12
++ #define RX_DESCQ_BUF_BASE_ID_LBN 36
++ #define RX_DESCQ_BUF_BASE_ID_WIDTH 20
++ #define RX_DESCQ_EVQ_ID_LBN 24
++ #define RX_DESCQ_EVQ_ID_WIDTH 12
++ #define RX_DESCQ_OWNER_ID_LBN 10
++ #define RX_DESCQ_OWNER_ID_WIDTH 14
++ #define RX_DESCQ_LABEL_LBN 5
++ #define RX_DESCQ_LABEL_WIDTH 5
++ #define RX_DESCQ_SIZE_LBN 3
++ #define RX_DESCQ_SIZE_WIDTH 2
++ #define RX_DESCQ_TYPE_LBN 2
++ #define RX_DESCQ_TYPE_WIDTH 1
++ #define RX_DESCQ_JUMBO_LBN 1
++ #define RX_DESCQ_JUMBO_WIDTH 1
++ #define RX_DESCQ_EN_LBN 0
++ #define RX_DESCQ_EN_WIDTH 1
++
++
++#define RX_RSS_INDIR_TBL_B0_OFST 0xFB0000 /* RSS indirection table (B0 only) */
++ #define RX_RSS_INDIR_ENT_B0_LBN 0
++ #define RX_RSS_INDIR_ENT_B0_WIDTH 6
++
++/*************---- TX Datapath Registers C Header ----*************/
++#define TX_FLUSH_DESCQ_REG_KER_OFST 0xA00 /* Transmit flush descriptor
++ queue register */
++#define TX_FLUSH_DESCQ_REG_OFST 0xA00 /* Transmit flush descriptor queue
++ register */
++ #define TX_FLUSH_DESCQ_CMD_LBN 12
++ #define TX_FLUSH_DESCQ_CMD_WIDTH 1
++ #define TX_FLUSH_DESCQ_LBN 0
++ #define TX_FLUSH_DESCQ_WIDTH 12
++#define TX_DESC_UPD_REG_KER_OFST 0xA10 /* Kernel transmit descriptor update
++ register. Page-mapped */
++#define TX_DESC_UPD_REG_PAGE4_OFST 0x8A10 /* Char & user transmit descriptor
++ update register. Page-mapped */
++#define TX_DESC_UPD_REG_PAGE123K_OFST 0x1000A10 /* Char & user transmit
++ descriptor update register.
++ Page-mapped */
++ #define TX_DESC_WPTR_LBN 96
++ #define TX_DESC_WPTR_WIDTH 12
++ #define TX_DESC_PUSH_CMD_LBN 95
++ #define TX_DESC_PUSH_CMD_WIDTH 1
++ #define TX_DESC_LBN 0
++ #define TX_DESC_WIDTH 95
++ #define TX_KER_DESC_LBN 0
++ #define TX_KER_DESC_WIDTH 64
++ #define TX_USR_DESC_LBN 0
++ #define TX_USR_DESC_WIDTH 64
++#define TX_DC_CFG_REG_KER_OFST 0xA20 /* Transmit descriptor cache
++ configuration register */
++#define TX_DC_CFG_REG_OFST 0xA20 /* Transmit descriptor cache configuration
++ register */
++ #define TX_DC_SIZE_LBN 0
++ #define TX_DC_SIZE_WIDTH 2
++
++#if FALCON_EXTENDED_P_BAR
++#define TX_DESC_PTR_TBL_KER_OFST 0x11900 /* Transmit descriptor pointer. */
++#else
++#define TX_DESC_PTR_TBL_KER_OFST 0x1900 /* Transmit descriptor pointer. */
++#endif
++
++
++#define TX_DESC_PTR_TBL_OFST 0xF50000 /* Transmit descriptor pointer */
++ #define TX_NON_IP_DROP_DIS_B0_LBN 91
++ #define TX_NON_IP_DROP_DIS_B0_WIDTH 1
++ #define TX_IP_CHKSM_DIS_B0_LBN 90
++ #define TX_IP_CHKSM_DIS_B0_WIDTH 1
++ #define TX_TCP_CHKSM_DIS_B0_LBN 89
++ #define TX_TCP_CHKSM_DIS_B0_WIDTH 1
++ #define TX_DESCQ_EN_LBN 88
++ #define TX_DESCQ_EN_WIDTH 1
++ #define TX_ISCSI_DDIG_EN_LBN 87
++ #define TX_ISCSI_DDIG_EN_WIDTH 1
++ #define TX_ISCSI_HDIG_EN_LBN 86
++ #define TX_ISCSI_HDIG_EN_WIDTH 1
++ #define TX_DC_HW_RPTR_LBN 80
++ #define TX_DC_HW_RPTR_WIDTH 6
++ #define TX_DESCQ_HW_RPTR_LBN 68
++ #define TX_DESCQ_HW_RPTR_WIDTH 12
++ #define TX_DESCQ_SW_WPTR_LBN 56
++ #define TX_DESCQ_SW_WPTR_WIDTH 12
++ #define TX_DESCQ_BUF_BASE_ID_LBN 36
++ #define TX_DESCQ_BUF_BASE_ID_WIDTH 20
++ #define TX_DESCQ_EVQ_ID_LBN 24
++ #define TX_DESCQ_EVQ_ID_WIDTH 12
++ #define TX_DESCQ_OWNER_ID_LBN 10
++ #define TX_DESCQ_OWNER_ID_WIDTH 14
++ #define TX_DESCQ_LABEL_LBN 5
++ #define TX_DESCQ_LABEL_WIDTH 5
++ #define TX_DESCQ_SIZE_LBN 3
++ #define TX_DESCQ_SIZE_WIDTH 2
++ #define TX_DESCQ_TYPE_LBN 1
++ #define TX_DESCQ_TYPE_WIDTH 2
++ #define TX_DESCQ_FLUSH_LBN 0
++ #define TX_DESCQ_FLUSH_WIDTH 1
++#define TX_CFG_REG_KER_OFST 0xA50 /* Transmit configuration register */
++#define TX_CFG_REG_OFST 0xA50 /* Transmit configuration register */
++ #define TX_IP_ID_P1_OFS_LBN 32
++ #define TX_IP_ID_P1_OFS_WIDTH 15
++ #define TX_IP_ID_P0_OFS_LBN 16
++ #define TX_IP_ID_P0_OFS_WIDTH 15
++ #define TX_TURBO_EN_LBN 3
++ #define TX_TURBO_EN_WIDTH 1
++ #define TX_OWNERR_CTL_LBN 2
++ #define TX_OWNERR_CTL_WIDTH 2
++ #define TX_NON_IP_DROP_DIS_LBN 1
++ #define TX_NON_IP_DROP_DIS_WIDTH 1
++ #define TX_IP_ID_REP_EN_LBN 0
++ #define TX_IP_ID_REP_EN_WIDTH 1
++#define TX_RESERVED_REG_KER_OFST 0xA80 /* Transmit configuration register */
++#define TX_RESERVED_REG_OFST 0xA80 /* Transmit configuration register */
++ #define TX_CSR_PUSH_EN_LBN 89
++ #define TX_CSR_PUSH_EN_WIDTH 1
++ #define TX_RX_SPACER_LBN 64
++ #define TX_RX_SPACER_WIDTH 8
++ #define TX_SW_EV_EN_LBN 59
++ #define TX_SW_EV_EN_WIDTH 1
++ #define TX_RX_SPACER_EN_LBN 57
++ #define TX_RX_SPACER_EN_WIDTH 1
++ #define TX_CSR_PREF_WD_TMR_LBN 24
++ #define TX_CSR_PREF_WD_TMR_WIDTH 16
++ #define TX_CSR_ONLY1TAG_LBN 21
++ #define TX_CSR_ONLY1TAG_WIDTH 1
++ #define TX_PREF_THRESHOLD_LBN 19
++ #define TX_PREF_THRESHOLD_WIDTH 2
++ #define TX_ONE_PKT_PER_Q_LBN 18
++ #define TX_ONE_PKT_PER_Q_WIDTH 1
++ #define TX_DIS_NON_IP_EV_LBN 17
++ #define TX_DIS_NON_IP_EV_WIDTH 1
++ #define TX_DMA_SPACER_LBN 8
++ #define TX_DMA_SPACER_WIDTH 8
++ #define TX_FLUSH_MIN_LEN_EN_B0_LBN 7
++ #define TX_FLUSH_MIN_LEN_EN_B0_WIDTH 1
++ #define TX_TCP_DIS_A1_LBN 7
++ #define TX_TCP_DIS_A1_WIDTH 1
++ #define TX_IP_DIS_A1_LBN 6
++ #define TX_IP_DIS_A1_WIDTH 1
++ #define TX_MAX_CPL_LBN 2
++ #define TX_MAX_CPL_WIDTH 2
++ #define TX_MAX_PREF_LBN 0
++ #define TX_MAX_PREF_WIDTH 2
++#define TX_VLAN_REG_OFST 0xAE0 /* Transmit VLAN tag register */
++ #define TX_VLAN_EN_LBN 127
++ #define TX_VLAN_EN_WIDTH 1
++ #define TX_VLAN7_PORT1_EN_LBN 125
++ #define TX_VLAN7_PORT1_EN_WIDTH 1
++ #define TX_VLAN7_PORT0_EN_LBN 124
++ #define TX_VLAN7_PORT0_EN_WIDTH 1
++ #define TX_VLAN7_LBN 112
++ #define TX_VLAN7_WIDTH 12
++ #define TX_VLAN6_PORT1_EN_LBN 109
++ #define TX_VLAN6_PORT1_EN_WIDTH 1
++ #define TX_VLAN6_PORT0_EN_LBN 108
++ #define TX_VLAN6_PORT0_EN_WIDTH 1
++ #define TX_VLAN6_LBN 96
++ #define TX_VLAN6_WIDTH 12
++ #define TX_VLAN5_PORT1_EN_LBN 93
++ #define TX_VLAN5_PORT1_EN_WIDTH 1
++ #define TX_VLAN5_PORT0_EN_LBN 92
++ #define TX_VLAN5_PORT0_EN_WIDTH 1
++ #define TX_VLAN5_LBN 80
++ #define TX_VLAN5_WIDTH 12
++ #define TX_VLAN4_PORT1_EN_LBN 77
++ #define TX_VLAN4_PORT1_EN_WIDTH 1
++ #define TX_VLAN4_PORT0_EN_LBN 76
++ #define TX_VLAN4_PORT0_EN_WIDTH 1
++ #define TX_VLAN4_LBN 64
++ #define TX_VLAN4_WIDTH 12
++ #define TX_VLAN3_PORT1_EN_LBN 61
++ #define TX_VLAN3_PORT1_EN_WIDTH 1
++ #define TX_VLAN3_PORT0_EN_LBN 60
++ #define TX_VLAN3_PORT0_EN_WIDTH 1
++ #define TX_VLAN3_LBN 48
++ #define TX_VLAN3_WIDTH 12
++ #define TX_VLAN2_PORT1_EN_LBN 45
++ #define TX_VLAN2_PORT1_EN_WIDTH 1
++ #define TX_VLAN2_PORT0_EN_LBN 44
++ #define TX_VLAN2_PORT0_EN_WIDTH 1
++ #define TX_VLAN2_LBN 32
++ #define TX_VLAN2_WIDTH 12
++ #define TX_VLAN1_PORT1_EN_LBN 29
++ #define TX_VLAN1_PORT1_EN_WIDTH 1
++ #define TX_VLAN1_PORT0_EN_LBN 28
++ #define TX_VLAN1_PORT0_EN_WIDTH 1
++ #define TX_VLAN1_LBN 16
++ #define TX_VLAN1_WIDTH 12
++ #define TX_VLAN0_PORT1_EN_LBN 13
++ #define TX_VLAN0_PORT1_EN_WIDTH 1
++ #define TX_VLAN0_PORT0_EN_LBN 12
++ #define TX_VLAN0_PORT0_EN_WIDTH 1
++ #define TX_VLAN0_LBN 0
++ #define TX_VLAN0_WIDTH 12
++#define TX_FIL_CTL_REG_OFST 0xAF0 /* Transmit filter control register */
++ #define TX_MADR1_FIL_EN_LBN 65
++ #define TX_MADR1_FIL_EN_WIDTH 1
++ #define TX_MADR0_FIL_EN_LBN 64
++ #define TX_MADR0_FIL_EN_WIDTH 1
++ #define TX_IPFIL31_PORT1_EN_LBN 63
++ #define TX_IPFIL31_PORT1_EN_WIDTH 1
++ #define TX_IPFIL31_PORT0_EN_LBN 62
++ #define TX_IPFIL31_PORT0_EN_WIDTH 1
++ #define TX_IPFIL30_PORT1_EN_LBN 61
++ #define TX_IPFIL30_PORT1_EN_WIDTH 1
++ #define TX_IPFIL30_PORT0_EN_LBN 60
++ #define TX_IPFIL30_PORT0_EN_WIDTH 1
++ #define TX_IPFIL29_PORT1_EN_LBN 59
++ #define TX_IPFIL29_PORT1_EN_WIDTH 1
++ #define TX_IPFIL29_PORT0_EN_LBN 58
++ #define TX_IPFIL29_PORT0_EN_WIDTH 1
++ #define TX_IPFIL28_PORT1_EN_LBN 57
++ #define TX_IPFIL28_PORT1_EN_WIDTH 1
++ #define TX_IPFIL28_PORT0_EN_LBN 56
++ #define TX_IPFIL28_PORT0_EN_WIDTH 1
++ #define TX_IPFIL27_PORT1_EN_LBN 55
++ #define TX_IPFIL27_PORT1_EN_WIDTH 1
++ #define TX_IPFIL27_PORT0_EN_LBN 54
++ #define TX_IPFIL27_PORT0_EN_WIDTH 1
++ #define TX_IPFIL26_PORT1_EN_LBN 53
++ #define TX_IPFIL26_PORT1_EN_WIDTH 1
++ #define TX_IPFIL26_PORT0_EN_LBN 52
++ #define TX_IPFIL26_PORT0_EN_WIDTH 1
++ #define TX_IPFIL25_PORT1_EN_LBN 51
++ #define TX_IPFIL25_PORT1_EN_WIDTH 1
++ #define TX_IPFIL25_PORT0_EN_LBN 50
++ #define TX_IPFIL25_PORT0_EN_WIDTH 1
++ #define TX_IPFIL24_PORT1_EN_LBN 49
++ #define TX_IPFIL24_PORT1_EN_WIDTH 1
++ #define TX_IPFIL24_PORT0_EN_LBN 48
++ #define TX_IPFIL24_PORT0_EN_WIDTH 1
++ #define TX_IPFIL23_PORT1_EN_LBN 47
++ #define TX_IPFIL23_PORT1_EN_WIDTH 1
++ #define TX_IPFIL23_PORT0_EN_LBN 46
++ #define TX_IPFIL23_PORT0_EN_WIDTH 1
++ #define TX_IPFIL22_PORT1_EN_LBN 45
++ #define TX_IPFIL22_PORT1_EN_WIDTH 1
++ #define TX_IPFIL22_PORT0_EN_LBN 44
++ #define TX_IPFIL22_PORT0_EN_WIDTH 1
++ #define TX_IPFIL21_PORT1_EN_LBN 43
++ #define TX_IPFIL21_PORT1_EN_WIDTH 1
++ #define TX_IPFIL21_PORT0_EN_LBN 42
++ #define TX_IPFIL21_PORT0_EN_WIDTH 1
++ #define TX_IPFIL20_PORT1_EN_LBN 41
++ #define TX_IPFIL20_PORT1_EN_WIDTH 1
++ #define TX_IPFIL20_PORT0_EN_LBN 40
++ #define TX_IPFIL20_PORT0_EN_WIDTH 1
++ #define TX_IPFIL19_PORT1_EN_LBN 39
++ #define TX_IPFIL19_PORT1_EN_WIDTH 1
++ #define TX_IPFIL19_PORT0_EN_LBN 38
++ #define TX_IPFIL19_PORT0_EN_WIDTH 1
++ #define TX_IPFIL18_PORT1_EN_LBN 37
++ #define TX_IPFIL18_PORT1_EN_WIDTH 1
++ #define TX_IPFIL18_PORT0_EN_LBN 36
++ #define TX_IPFIL18_PORT0_EN_WIDTH 1
++ #define TX_IPFIL17_PORT1_EN_LBN 35
++ #define TX_IPFIL17_PORT1_EN_WIDTH 1
++ #define TX_IPFIL17_PORT0_EN_LBN 34
++ #define TX_IPFIL17_PORT0_EN_WIDTH 1
++ #define TX_IPFIL16_PORT1_EN_LBN 33
++ #define TX_IPFIL16_PORT1_EN_WIDTH 1
++ #define TX_IPFIL16_PORT0_EN_LBN 32
++ #define TX_IPFIL16_PORT0_EN_WIDTH 1
++ #define TX_IPFIL15_PORT1_EN_LBN 31
++ #define TX_IPFIL15_PORT1_EN_WIDTH 1
++ #define TX_IPFIL15_PORT0_EN_LBN 30
++ #define TX_IPFIL15_PORT0_EN_WIDTH 1
++ #define TX_IPFIL14_PORT1_EN_LBN 29
++ #define TX_IPFIL14_PORT1_EN_WIDTH 1
++ #define TX_IPFIL14_PORT0_EN_LBN 28
++ #define TX_IPFIL14_PORT0_EN_WIDTH 1
++ #define TX_IPFIL13_PORT1_EN_LBN 27
++ #define TX_IPFIL13_PORT1_EN_WIDTH 1
++ #define TX_IPFIL13_PORT0_EN_LBN 26
++ #define TX_IPFIL13_PORT0_EN_WIDTH 1
++ #define TX_IPFIL12_PORT1_EN_LBN 25
++ #define TX_IPFIL12_PORT1_EN_WIDTH 1
++ #define TX_IPFIL12_PORT0_EN_LBN 24
++ #define TX_IPFIL12_PORT0_EN_WIDTH 1
++ #define TX_IPFIL11_PORT1_EN_LBN 23
++ #define TX_IPFIL11_PORT1_EN_WIDTH 1
++ #define TX_IPFIL11_PORT0_EN_LBN 22
++ #define TX_IPFIL11_PORT0_EN_WIDTH 1
++ #define TX_IPFIL10_PORT1_EN_LBN 21
++ #define TX_IPFIL10_PORT1_EN_WIDTH 1
++ #define TX_IPFIL10_PORT0_EN_LBN 20
++ #define TX_IPFIL10_PORT0_EN_WIDTH 1
++ #define TX_IPFIL9_PORT1_EN_LBN 19
++ #define TX_IPFIL9_PORT1_EN_WIDTH 1
++ #define TX_IPFIL9_PORT0_EN_LBN 18
++ #define TX_IPFIL9_PORT0_EN_WIDTH 1
++ #define TX_IPFIL8_PORT1_EN_LBN 17
++ #define TX_IPFIL8_PORT1_EN_WIDTH 1
++ #define TX_IPFIL8_PORT0_EN_LBN 16
++ #define TX_IPFIL8_PORT0_EN_WIDTH 1
++ #define TX_IPFIL7_PORT1_EN_LBN 15
++ #define TX_IPFIL7_PORT1_EN_WIDTH 1
++ #define TX_IPFIL7_PORT0_EN_LBN 14
++ #define TX_IPFIL7_PORT0_EN_WIDTH 1
++ #define TX_IPFIL6_PORT1_EN_LBN 13
++ #define TX_IPFIL6_PORT1_EN_WIDTH 1
++ #define TX_IPFIL6_PORT0_EN_LBN 12
++ #define TX_IPFIL6_PORT0_EN_WIDTH 1
++ #define TX_IPFIL5_PORT1_EN_LBN 11
++ #define TX_IPFIL5_PORT1_EN_WIDTH 1
++ #define TX_IPFIL5_PORT0_EN_LBN 10
++ #define TX_IPFIL5_PORT0_EN_WIDTH 1
++ #define TX_IPFIL4_PORT1_EN_LBN 9
++ #define TX_IPFIL4_PORT1_EN_WIDTH 1
++ #define TX_IPFIL4_PORT0_EN_LBN 8
++ #define TX_IPFIL4_PORT0_EN_WIDTH 1
++ #define TX_IPFIL3_PORT1_EN_LBN 7
++ #define TX_IPFIL3_PORT1_EN_WIDTH 1
++ #define TX_IPFIL3_PORT0_EN_LBN 6
++ #define TX_IPFIL3_PORT0_EN_WIDTH 1
++ #define TX_IPFIL2_PORT1_EN_LBN 5
++ #define TX_IPFIL2_PORT1_EN_WIDTH 1
++ #define TX_IPFIL2_PORT0_EN_LBN 4
++ #define TX_IPFIL2_PORT0_EN_WIDTH 1
++ #define TX_IPFIL1_PORT1_EN_LBN 3
++ #define TX_IPFIL1_PORT1_EN_WIDTH 1
++ #define TX_IPFIL1_PORT0_EN_LBN 2
++ #define TX_IPFIL1_PORT0_EN_WIDTH 1
++ #define TX_IPFIL0_PORT1_EN_LBN 1
++ #define TX_IPFIL0_PORT1_EN_WIDTH 1
++ #define TX_IPFIL0_PORT0_EN_LBN 0
++ #define TX_IPFIL0_PORT0_EN_WIDTH 1
++#define TX_IPFIL_TBL_OFST 0xB00 /* Transmit IP source address filter table */
++ #define TX_IPFIL_MASK_LBN 32
++ #define TX_IPFIL_MASK_WIDTH 32
++ #define TX_IP_SRC_ADR_LBN 0
++ #define TX_IP_SRC_ADR_WIDTH 32
++#define TX_PACE_REG_A1_OFST 0xF80000 /* Transmit pace control register */
++#define TX_PACE_REG_B0_OFST 0xA90 /* Transmit pace control register */
++ #define TX_PACE_SB_AF_LBN 19
++ #define TX_PACE_SB_AF_WIDTH 10
++ #define TX_PACE_SB_NOTAF_LBN 9
++ #define TX_PACE_SB_NOTAF_WIDTH 10
++ #define TX_PACE_FB_BASE_LBN 5
++ #define TX_PACE_FB_BASE_WIDTH 4
++ #define TX_PACE_BIN_TH_LBN 0
++ #define TX_PACE_BIN_TH_WIDTH 5
++#define TX_PACE_TBL_A1_OFST 0xF80040 /* Transmit pacing table */
++#define TX_PACE_TBL_FIRST_QUEUE_A1 4
++#define TX_PACE_TBL_B0_OFST 0xF80000 /* Transmit pacing table */
++#define TX_PACE_TBL_FIRST_QUEUE_B0 0
++ #define TX_PACE_LBN 0
++ #define TX_PACE_WIDTH 5
++
++/*************---- EE/Flash Registers C Header ----*************/
++#define EE_SPI_HCMD_REG_KER_OFST 0x100 /* SPI host command register */
++#define EE_SPI_HCMD_REG_OFST 0x100 /* SPI host command register */
++ #define EE_SPI_HCMD_CMD_EN_LBN 31
++ #define EE_SPI_HCMD_CMD_EN_WIDTH 1
++ #define EE_WR_TIMER_ACTIVE_LBN 28
++ #define EE_WR_TIMER_ACTIVE_WIDTH 1
++ #define EE_SPI_HCMD_SF_SEL_LBN 24
++ #define EE_SPI_HCMD_SF_SEL_WIDTH 1
++ #define EE_SPI_HCMD_DABCNT_LBN 16
++ #define EE_SPI_HCMD_DABCNT_WIDTH 5
++ #define EE_SPI_HCMD_READ_LBN 15
++ #define EE_SPI_HCMD_READ_WIDTH 1
++ #define EE_SPI_HCMD_DUBCNT_LBN 12
++ #define EE_SPI_HCMD_DUBCNT_WIDTH 2
++ #define EE_SPI_HCMD_ADBCNT_LBN 8
++ #define EE_SPI_HCMD_ADBCNT_WIDTH 2
++ #define EE_SPI_HCMD_ENC_LBN 0
++ #define EE_SPI_HCMD_ENC_WIDTH 8
++#define EE_SPI_HADR_REG_KER_OFST 0X110 /* SPI host address register */
++#define EE_SPI_HADR_REG_OFST 0X110 /* SPI host address register */
++ #define EE_SPI_HADR_DUBYTE_LBN 24
++ #define EE_SPI_HADR_DUBYTE_WIDTH 8
++ #define EE_SPI_HADR_ADR_LBN 0
++ #define EE_SPI_HADR_ADR_WIDTH 24
++#define EE_SPI_HDATA_REG_KER_OFST 0x120 /* SPI host data register */
++#define EE_SPI_HDATA_REG_OFST 0x120 /* SPI host data register */
++ #define EE_SPI_HDATA3_LBN 96
++ #define EE_SPI_HDATA3_WIDTH 32
++ #define EE_SPI_HDATA2_LBN 64
++ #define EE_SPI_HDATA2_WIDTH 32
++ #define EE_SPI_HDATA1_LBN 32
++ #define EE_SPI_HDATA1_WIDTH 32
++ #define EE_SPI_HDATA0_LBN 0
++ #define EE_SPI_HDATA0_WIDTH 32
++#define EE_BASE_PAGE_REG_KER_OFST 0x130 /* Expansion ROM base mirror register */
++#define EE_BASE_PAGE_REG_OFST 0x130 /* Expansion ROM base mirror register */
++ #define EE_EXP_ROM_WINDOW_BASE_LBN 16
++ #define EE_EXP_ROM_WINDOW_BASE_WIDTH 13
++ #define EE_EXPROM_MASK_LBN 0
++ #define EE_EXPROM_MASK_WIDTH 13
++#define EE_VPD_CFG0_REG_KER_OFST 0X140 /* SPI/VPD configuration register */
++#define EE_VPD_CFG0_REG_OFST 0X140 /* SPI/VPD configuration register */
++ #define EE_SF_FASTRD_EN_LBN 127
++ #define EE_SF_FASTRD_EN_WIDTH 1
++ #define EE_SF_CLOCK_DIV_LBN 120
++ #define EE_SF_CLOCK_DIV_WIDTH 7
++ #define EE_VPD_WIP_POLL_LBN 119
++ #define EE_VPD_WIP_POLL_WIDTH 1
++ #define EE_VPDW_LENGTH_LBN 80
++ #define EE_VPDW_LENGTH_WIDTH 15
++ #define EE_VPDW_BASE_LBN 64
++ #define EE_VPDW_BASE_WIDTH 15
++ #define EE_VPD_WR_CMD_EN_LBN 56
++ #define EE_VPD_WR_CMD_EN_WIDTH 8
++ #define EE_VPD_BASE_LBN 32
++ #define EE_VPD_BASE_WIDTH 24
++ #define EE_VPD_LENGTH_LBN 16
++ #define EE_VPD_LENGTH_WIDTH 13
++ #define EE_VPD_AD_SIZE_LBN 8
++ #define EE_VPD_AD_SIZE_WIDTH 5
++ #define EE_VPD_ACCESS_ON_LBN 5
++ #define EE_VPD_ACCESS_ON_WIDTH 1
++#define EE_VPD_SW_CNTL_REG_KER_OFST 0X150 /* VPD access SW control register */
++#define EE_VPD_SW_CNTL_REG_OFST 0X150 /* VPD access SW control register */
++ #define EE_VPD_CYCLE_PENDING_LBN 31
++ #define EE_VPD_CYCLE_PENDING_WIDTH 1
++ #define EE_VPD_CYC_WRITE_LBN 28
++ #define EE_VPD_CYC_WRITE_WIDTH 1
++ #define EE_VPD_CYC_ADR_LBN 0
++ #define EE_VPD_CYC_ADR_WIDTH 15
++#define EE_VPD_SW_DATA_REG_KER_OFST 0x160 /* VPD access SW data register */
++#define EE_VPD_SW_DATA_REG_OFST 0x160 /* VPD access SW data register */
++ #define EE_VPD_CYC_DAT_LBN 0
++ #define EE_VPD_CYC_DAT_WIDTH 32
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_desc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_desc.h 2008-05-19 00:33:29.333836963 +0300
+@@ -0,0 +1,75 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) descriptor
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*************---- Descriptors C Headers ----*************/
++/* Receive Kernel IP Descriptor */
++ #define RX_KER_BUF_SIZE_LBN 48
++ #define RX_KER_BUF_SIZE_WIDTH 14
++ #define RX_KER_BUF_REGION_LBN 46
++ #define RX_KER_BUF_REGION_WIDTH 2
++ #define RX_KER_BUF_REGION0_DECODE 0
++ #define RX_KER_BUF_REGION1_DECODE 1
++ #define RX_KER_BUF_REGION2_DECODE 2
++ #define RX_KER_BUF_REGION3_DECODE 3
++ #define RX_KER_BUF_ADR_LBN 0
++ #define RX_KER_BUF_ADR_WIDTH 46
++/* Receive User IP Descriptor */
++ #define RX_USR_2BYTE_OFS_LBN 20
++ #define RX_USR_2BYTE_OFS_WIDTH 12
++ #define RX_USR_BUF_ID_LBN 0
++ #define RX_USR_BUF_ID_WIDTH 20
++/* Transmit Kernel IP Descriptor */
++ #define TX_KER_PORT_LBN 63
++ #define TX_KER_PORT_WIDTH 1
++ #define TX_KER_CONT_LBN 62
++ #define TX_KER_CONT_WIDTH 1
++ #define TX_KER_BYTE_CNT_LBN 48
++ #define TX_KER_BYTE_CNT_WIDTH 14
++ #define TX_KER_BUF_REGION_LBN 46
++ #define TX_KER_BUF_REGION_WIDTH 2
++ #define TX_KER_BUF_REGION0_DECODE 0
++ #define TX_KER_BUF_REGION1_DECODE 1
++ #define TX_KER_BUF_REGION2_DECODE 2
++ #define TX_KER_BUF_REGION3_DECODE 3
++ #define TX_KER_BUF_ADR_LBN 0
++ #define TX_KER_BUF_ADR_WIDTH 46
++/* Transmit User IP Descriptor */
++ #define TX_USR_PORT_LBN 47
++ #define TX_USR_PORT_WIDTH 1
++ #define TX_USR_CONT_LBN 46
++ #define TX_USR_CONT_WIDTH 1
++ #define TX_USR_BYTE_CNT_LBN 33
++ #define TX_USR_BYTE_CNT_WIDTH 13
++ #define TX_USR_BUF_ID_LBN 13
++ #define TX_USR_BUF_ID_WIDTH 20
++ #define TX_USR_BYTE_OFS_LBN 0
++ #define TX_USR_BYTE_OFS_WIDTH 13
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_event.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_event.h 2008-05-19 00:33:29.333836963 +0300
+@@ -0,0 +1,155 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) event
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*************---- Events Format C Header ----*************/
++/*************---- Event entry ----*************/
++ #define EV_CODE_LBN 60
++ #define EV_CODE_WIDTH 4
++ #define RX_IP_EV_DECODE 0
++ #define TX_IP_EV_DECODE 2
++ #define DRIVER_EV_DECODE 5
++ #define GLOBAL_EV_DECODE 6
++ #define DRV_GEN_EV_DECODE 7
++ #define EV_DATA_LBN 0
++ #define EV_DATA_WIDTH 60
++/******---- Receive IP events for both Kernel & User event queues ----******/
++ #define RX_EV_PKT_OK_LBN 56
++ #define RX_EV_PKT_OK_WIDTH 1
++ #define RX_EV_BUF_OWNER_ID_ERR_LBN 54
++ #define RX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++ #define RX_EV_IP_HDR_CHKSUM_ERR_LBN 52
++ #define RX_EV_IP_HDR_CHKSUM_ERR_WIDTH 1
++ #define RX_EV_TCP_UDP_CHKSUM_ERR_LBN 51
++ #define RX_EV_TCP_UDP_CHKSUM_ERR_WIDTH 1
++ #define RX_EV_ETH_CRC_ERR_LBN 50
++ #define RX_EV_ETH_CRC_ERR_WIDTH 1
++ #define RX_EV_FRM_TRUNC_LBN 49
++ #define RX_EV_FRM_TRUNC_WIDTH 1
++ #define RX_EV_DRIB_NIB_LBN 48
++ #define RX_EV_DRIB_NIB_WIDTH 1
++ #define RX_EV_TOBE_DISC_LBN 47
++ #define RX_EV_TOBE_DISC_WIDTH 1
++ #define RX_EV_PKT_TYPE_LBN 44
++ #define RX_EV_PKT_TYPE_WIDTH 3
++ #define RX_EV_PKT_TYPE_ETH_DECODE 0
++ #define RX_EV_PKT_TYPE_LLC_DECODE 1
++ #define RX_EV_PKT_TYPE_JUMBO_DECODE 2
++ #define RX_EV_PKT_TYPE_VLAN_DECODE 3
++ #define RX_EV_PKT_TYPE_VLAN_LLC_DECODE 4
++ #define RX_EV_PKT_TYPE_VLAN_JUMBO_DECODE 5
++ #define RX_EV_HDR_TYPE_LBN 42
++ #define RX_EV_HDR_TYPE_WIDTH 2
++ #define RX_EV_HDR_TYPE_TCP_IPV4_DECODE 0
++ #define RX_EV_HDR_TYPE_UDP_IPV4_DECODE 1
++ #define RX_EV_HDR_TYPE_OTHER_IP_DECODE 2
++ #define RX_EV_HDR_TYPE_NON_IP_DECODE 3
++ #define RX_EV_DESC_Q_EMPTY_LBN 41
++ #define RX_EV_DESC_Q_EMPTY_WIDTH 1
++ #define RX_EV_MCAST_HASH_MATCH_LBN 40
++ #define RX_EV_MCAST_HASH_MATCH_WIDTH 1
++ #define RX_EV_MCAST_PKT_LBN 39
++ #define RX_EV_MCAST_PKT_WIDTH 1
++ #define RX_EV_Q_LABEL_LBN 32
++ #define RX_EV_Q_LABEL_WIDTH 5
++ #define RX_JUMBO_CONT_LBN 31
++ #define RX_JUMBO_CONT_WIDTH 1
++ #define RX_SOP_LBN 15
++ #define RX_SOP_WIDTH 1
++ #define RX_PORT_LBN 30
++ #define RX_PORT_WIDTH 1
++ #define RX_EV_BYTE_CNT_LBN 16
++ #define RX_EV_BYTE_CNT_WIDTH 14
++ #define RX_iSCSI_PKT_OK_LBN 14
++ #define RX_iSCSI_PKT_OK_WIDTH 1
++ #define RX_ISCSI_DDIG_ERR_LBN 13
++ #define RX_ISCSI_DDIG_ERR_WIDTH 1
++ #define RX_ISCSI_HDIG_ERR_LBN 12
++ #define RX_ISCSI_HDIG_ERR_WIDTH 1
++ #define RX_EV_DESC_PTR_LBN 0
++ #define RX_EV_DESC_PTR_WIDTH 12
++/******---- Transmit IP events for both Kernel & User event queues ----******/
++ #define TX_EV_PKT_ERR_LBN 38
++ #define TX_EV_PKT_ERR_WIDTH 1
++ #define TX_EV_PKT_TOO_BIG_LBN 37
++ #define TX_EV_PKT_TOO_BIG_WIDTH 1
++ #define TX_EV_Q_LABEL_LBN 32
++ #define TX_EV_Q_LABEL_WIDTH 5
++ #define TX_EV_PORT_LBN 16
++ #define TX_EV_PORT_WIDTH 1
++ #define TX_EV_WQ_FF_FULL_LBN 15
++ #define TX_EV_WQ_FF_FULL_WIDTH 1
++ #define TX_EV_BUF_OWNER_ID_ERR_LBN 14
++ #define TX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++ #define TX_EV_COMP_LBN 12
++ #define TX_EV_COMP_WIDTH 1
++ #define TX_EV_DESC_PTR_LBN 0
++ #define TX_EV_DESC_PTR_WIDTH 12
++/*************---- Char or Kernel driver events ----*************/
++ #define DRIVER_EV_SUB_CODE_LBN 56
++ #define DRIVER_EV_SUB_CODE_WIDTH 4
++ #define TX_DESCQ_FLS_DONE_EV_DECODE 0x0
++ #define RX_DESCQ_FLS_DONE_EV_DECODE 0x1
++ #define EVQ_INIT_DONE_EV_DECODE 0x2
++ #define EVQ_NOT_EN_EV_DECODE 0x3
++ #define RX_DESCQ_FLSFF_OVFL_EV_DECODE 0x4
++ #define SRM_UPD_DONE_EV_DECODE 0x5
++ #define WAKE_UP_EV_DECODE 0x6
++ #define TX_PKT_NON_TCP_UDP_DECODE 0x9
++ #define TIMER_EV_DECODE 0xA
++ #define RX_DSC_ERROR_EV_DECODE 0xE
++ #define DRIVER_EV_TX_DESCQ_ID_LBN 0
++ #define DRIVER_EV_TX_DESCQ_ID_WIDTH 12
++ #define DRIVER_EV_RX_DESCQ_ID_LBN 0
++ #define DRIVER_EV_RX_DESCQ_ID_WIDTH 12
++ #define DRIVER_EV_EVQ_ID_LBN 0
++ #define DRIVER_EV_EVQ_ID_WIDTH 12
++ #define DRIVER_TMR_ID_LBN 0
++ #define DRIVER_TMR_ID_WIDTH 12
++ #define DRIVER_EV_SRM_UPD_LBN 0
++ #define DRIVER_EV_SRM_UPD_WIDTH 2
++ #define SRM_CLR_EV_DECODE 0
++ #define SRM_UPD_EV_DECODE 1
++ #define SRM_ILLCLR_EV_DECODE 2
++/********---- Global events. Sent to both event queue 0 and 4. ----********/
++ #define XFP_PHY_INTR_LBN 10
++ #define XFP_PHY_INTR_WIDTH 1
++ #define XG_PHY_INTR_LBN 9
++ #define XG_PHY_INTR_WIDTH 1
++ #define G_PHY1_INTR_LBN 8
++ #define G_PHY1_INTR_WIDTH 1
++ #define G_PHY0_INTR_LBN 7
++ #define G_PHY0_INTR_WIDTH 1
++/*************---- Driver generated events ----*************/
++ #define DRV_GEN_EV_CODE_LBN 60
++ #define DRV_GEN_EV_CODE_WIDTH 4
++ #define DRV_GEN_EV_DATA_LBN 0
++ #define DRV_GEN_EV_DATA_WIDTH 60
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_grmon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_grmon.h 2008-05-19 00:33:29.333836963 +0300
+@@ -0,0 +1,129 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) 1G MAC
++ * counters.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*************---- 1G MAC Statistical Counters C Header ----*************/
++#define GRxGoodOct_offset 0x0
++ #define GRxGoodOct_WIDTH 48
++#define GRxBadOct_offset 0x8
++ #define GRxBadOct_WIDTH 48
++#define GRxMissPkt_offset 0x10
++ #define GRxMissPkt_WIDTH 32
++#define GRxFalseCRS_offset 0x14
++ #define GRxFalseCRS_WIDTH 32
++#define GRxPausePkt_offset 0x18
++ #define GRxPausePkt_WIDTH 32
++#define GRxBadPkt_offset 0x1C
++ #define GRxBadPkt_WIDTH 32
++#define GRxUcastPkt_offset 0x20
++ #define GRxUcastPkt_WIDTH 32
++#define GRxMcastPkt_offset 0x24
++ #define GRxMcastPkt_WIDTH 32
++#define GRxBcastPkt_offset 0x28
++ #define GRxBcastPkt_WIDTH 32
++#define GRxGoodLt64Pkt_offset 0x2C
++ #define GRxGoodLt64Pkt_WIDTH 32
++#define GRxBadLt64Pkt_offset 0x30
++ #define GRxBadLt64Pkt_WIDTH 32
++#define GRx64Pkt_offset 0x34
++ #define GRx64Pkt_WIDTH 32
++#define GRx65to127Pkt_offset 0x38
++ #define GRx65to127Pkt_WIDTH 32
++#define GRx128to255Pkt_offset 0x3C
++ #define GRx128to255Pkt_WIDTH 32
++#define GRx256to511Pkt_offset 0x40
++ #define GRx256to511Pkt_WIDTH 32
++#define GRx512to1023Pkt_offset 0x44
++ #define GRx512to1023Pkt_WIDTH 32
++#define GRx1024to15xxPkt_offset 0x48
++ #define GRx1024to15xxPkt_WIDTH 32
++#define GRx15xxtoJumboPkt_offset 0x4C
++ #define GRx15xxtoJumboPkt_WIDTH 32
++#define GRxGtJumboPkt_offset 0x50
++ #define GRxGtJumboPkt_WIDTH 32
++#define GRxFcsErr64to15xxPkt_offset 0x54
++ #define GRxFcsErr64to15xxPkt_WIDTH 32
++#define GRxFcsErr15xxtoJumboPkt_offset 0x58
++ #define GRxFcsErr15xxtoJumboPkt_WIDTH 32
++#define GRxFcsErrGtJumboPkt_offset 0x5C
++ #define GRxFcsErrGtJumboPkt_WIDTH 32
++#define GTxGoodBadOct_offset 0x80
++ #define GTxGoodBadOct_WIDTH 48
++#define GTxGoodOct_offset 0x88
++ #define GTxGoodOct_WIDTH 48
++#define GTxSglColPkt_offset 0x90
++ #define GTxSglColPkt_WIDTH 32
++#define GTxMultColPkt_offset 0x94
++ #define GTxMultColPkt_WIDTH 32
++#define GTxExColPkt_offset 0x98
++ #define GTxExColPkt_WIDTH 32
++#define GTxDefPkt_offset 0x9C
++ #define GTxDefPkt_WIDTH 32
++#define GTxLateCol_offset 0xA0
++ #define GTxLateCol_WIDTH 32
++#define GTxExDefPkt_offset 0xA4
++ #define GTxExDefPkt_WIDTH 32
++#define GTxPausePkt_offset 0xA8
++ #define GTxPausePkt_WIDTH 32
++#define GTxBadPkt_offset 0xAC
++ #define GTxBadPkt_WIDTH 32
++#define GTxUcastPkt_offset 0xB0
++ #define GTxUcastPkt_WIDTH 32
++#define GTxMcastPkt_offset 0xB4
++ #define GTxMcastPkt_WIDTH 32
++#define GTxBcastPkt_offset 0xB8
++ #define GTxBcastPkt_WIDTH 32
++#define GTxLt64Pkt_offset 0xBC
++ #define GTxLt64Pkt_WIDTH 32
++#define GTx64Pkt_offset 0xC0
++ #define GTx64Pkt_WIDTH 32
++#define GTx65to127Pkt_offset 0xC4
++ #define GTx65to127Pkt_WIDTH 32
++#define GTx128to255Pkt_offset 0xC8
++ #define GTx128to255Pkt_WIDTH 32
++#define GTx256to511Pkt_offset 0xCC
++ #define GTx256to511Pkt_WIDTH 32
++#define GTx512to1023Pkt_offset 0xD0
++ #define GTx512to1023Pkt_WIDTH 32
++#define GTx1024to15xxPkt_offset 0xD4
++ #define GTx1024to15xxPkt_WIDTH 32
++#define GTx15xxtoJumboPkt_offset 0xD8
++ #define GTx15xxtoJumboPkt_WIDTH 32
++#define GTxGtJumboPkt_offset 0xDC
++ #define GTxGtJumboPkt_WIDTH 32
++#define GTxNonTcpUdpPkt_offset 0xE0
++ #define GTxNonTcpUdpPkt_WIDTH 16
++#define GTxMacSrcErrPkt_offset 0xE4
++ #define GTxMacSrcErrPkt_WIDTH 16
++#define GTxIpSrcErrPkt_offset 0xE8
++ #define GTxIpSrcErrPkt_WIDTH 16
++#define GDmaDone_offset 0xEC
++ #define GDmaDone_WIDTH 32
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_intr_vec.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_intr_vec.h 2008-05-19 00:33:29.333836963 +0300
+@@ -0,0 +1,44 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) interrupt
++ * vector definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*************---- Interrupt Vector Format C Header ----*************/
++#define DW0_OFST 0x0 /* Double-word 0: Event queue FIFO interrupts */
++ #define EVQ_FIFO_HF_LBN 1
++ #define EVQ_FIFO_HF_WIDTH 1
++ #define EVQ_FIFO_AF_LBN 0
++ #define EVQ_FIFO_AF_WIDTH 1
++#define DW1_OFST 0x4 /* Double-word 1: Interrupt indicator */
++ #define INT_FLAG_LBN 0
++ #define INT_FLAG_WIDTH 1
++#define DW2_OFST 0x8 /* Double-word 2: Fatal interrupts */
++ #define FATAL_INT_LBN 0
++ #define FATAL_INT_WIDTH 1
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_mac.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_mac.h 2008-05-19 00:33:29.337837194 +0300
+@@ -0,0 +1,711 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) MAC register
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*********---- 1G/10G Ethernet MAC Wrapper Registers C Header ----*********/
++#define MD_TXD_REG_KER_OFST 0xC00 /* PHY management transmit data register */
++#define MD_TXD_REG_OFST 0xC00 /* PHY management transmit data register */
++ #define MD_TXD_LBN 0
++ #define MD_TXD_WIDTH 16
++#define MD_RXD_REG_KER_OFST 0xC10 /* PHY management receive data register */
++#define MD_RXD_REG_OFST 0xC10 /* PHY management receive data register */
++ #define MD_RXD_LBN 0
++ #define MD_RXD_WIDTH 16
++#define MD_CS_REG_KER_OFST 0xC20 /* PHY management configuration &
++ status register */
++#define MD_CS_REG_OFST 0xC20 /* PHY management configuration &
++ status register */
++ #define MD_PT_LBN 7
++ #define MD_PT_WIDTH 3
++ #define MD_PL_LBN 6
++ #define MD_PL_WIDTH 1
++ #define MD_INT_CLR_LBN 5
++ #define MD_INT_CLR_WIDTH 1
++ #define MD_GC_LBN 4
++ #define MD_GC_WIDTH 1
++ #define MD_PRSP_LBN 3
++ #define MD_PRSP_WIDTH 1
++ #define MD_RIC_LBN 2
++ #define MD_RIC_WIDTH 1
++ #define MD_RDC_LBN 1
++ #define MD_RDC_WIDTH 1
++ #define MD_WRC_LBN 0
++ #define MD_WRC_WIDTH 1
++#define MD_PHY_ADR_REG_KER_OFST 0xC30 /* PHY management PHY address register */
++#define MD_PHY_ADR_REG_OFST 0xC30 /* PHY management PHY address register */
++ #define MD_PHY_ADR_LBN 0
++ #define MD_PHY_ADR_WIDTH 16
++#define MD_ID_REG_KER_OFST 0xC40 /* PHY management ID register */
++#define MD_ID_REG_OFST 0xC40 /* PHY management ID register */
++ #define MD_PRT_ADR_LBN 11
++ #define MD_PRT_ADR_WIDTH 5
++ #define MD_DEV_ADR_LBN 6
++ #define MD_DEV_ADR_WIDTH 5
++#define MD_STAT_REG_KER_OFST 0xC50 /* PHY management status & mask register */
++#define MD_STAT_REG_OFST 0xC50 /* PHY management status & mask register */
++ #define MD_PINT_LBN 4
++ #define MD_PINT_WIDTH 1
++ #define MD_DONE_LBN 3
++ #define MD_DONE_WIDTH 1
++ #define MD_BSERR_LBN 2
++ #define MD_BSERR_WIDTH 1
++ #define MD_LNFL_LBN 1
++ #define MD_LNFL_WIDTH 1
++ #define MD_BSY_LBN 0
++ #define MD_BSY_WIDTH 1
++#define MAC0_STAT_DMA_REG_KER_OFST 0xC60 /* Port 0 MAC statistical counter
++ DMA register */
++#define MAC0_STAT_DMA_REG_OFST 0xC60 /* Port 0 MAC statistical counter
++ DMA register */
++ #define MAC0_STAT_DMA_CMD_LBN 48
++ #define MAC0_STAT_DMA_CMD_WIDTH 1
++ #define MAC0_STAT_DMA_ADR_LBN 0
++ #define MAC0_STAT_DMA_ADR_WIDTH 48
++#define MAC1_STAT_DMA_REG_KER_OFST 0xC70 /* Port 1 MAC statistical counter
++ DMA register */
++#define MAC1_STAT_DMA_REG_OFST 0xC70 /* Port 1 MAC statistical counter
++ DMA register */
++ #define MAC1_STAT_DMA_CMD_LBN 48
++ #define MAC1_STAT_DMA_CMD_WIDTH 1
++ #define MAC1_STAT_DMA_ADR_LBN 0
++ #define MAC1_STAT_DMA_ADR_WIDTH 48
++#define MAC0_CTRL_REG_KER_OFST 0xC80 /* Port 0 MAC control register */
++#define MAC0_CTRL_REG_OFST 0xC80 /* Port 0 MAC control register */
++ #define MAC0_XOFF_VAL_LBN 16
++ #define MAC0_XOFF_VAL_WIDTH 16
++ #define MAC0_BCAD_ACPT_LBN 4
++ #define MAC0_BCAD_ACPT_WIDTH 1
++ #define MAC0_UC_PROM_LBN 3
++ #define MAC0_UC_PROM_WIDTH 1
++ #define MAC0_LINK_STATUS_LBN 2
++ #define MAC0_LINK_STATUS_WIDTH 1
++ #define MAC0_SPEED_LBN 0
++ #define MAC0_SPEED_WIDTH 2
++#define MAC1_CTRL_REG_KER_OFST 0xC90 /* Port 1 MAC control register */
++#define MAC1_CTRL_REG_OFST 0xC90 /* Port 1 MAC control register */
++ #define MAC1_XOFF_VAL_LBN 16
++ #define MAC1_XOFF_VAL_WIDTH 16
++ #define MAC1_BCAD_ACPT_LBN 4
++ #define MAC1_BCAD_ACPT_WIDTH 1
++ #define MAC1_UC_PROM_LBN 3
++ #define MAC1_UC_PROM_WIDTH 1
++ #define MAC1_LINK_STATUS_LBN 2
++ #define MAC1_LINK_STATUS_WIDTH 1
++ #define MAC1_SPEED_LBN 0
++ #define MAC1_SPEED_WIDTH 2
++#define MAC_MC_HASH_REG0_KER_OFST 0xCA0 /* Multicast address hash table */
++#define MAC_MC_HASH_REG0_OFST 0xCA0 /* Multicast address hash table */
++ #define MAC_MCAST_HASH0_LBN 0
++ #define MAC_MCAST_HASH0_WIDTH 128
++#define MAC_MC_HASH_REG1_KER_OFST 0xCB0 /* Multicast address hash table */
++#define MAC_MC_HASH_REG1_OFST 0xCB0 /* Multicast address hash table */
++ #define MAC_MCAST_HASH1_LBN 0
++ #define MAC_MCAST_HASH1_WIDTH 128
++/*************---- 1G MAC Port 0 Registers C Header ----*************/
++#define GM_P0_BASE 0xE00
++#define GM_P1_BASE 0x1000
++#define GM_CFG1_REG_KER_OFST 0x00 /* GMAC configuration register 1 */
++#define GM_CFG1_REG_OFST 0x00 /* GMAC configuration register 1 */
++ #define GM_SW_RST_LBN 31
++ #define GM_SW_RST_WIDTH 1
++ #define GM_SIM_RST_LBN 30
++ #define GM_SIM_RST_WIDTH 1
++ #define GM_RST_RX_MAC_CTL_LBN 19
++ #define GM_RST_RX_MAC_CTL_WIDTH 1
++ #define GM_RST_TX_MAC_CTL_LBN 18
++ #define GM_RST_TX_MAC_CTL_WIDTH 1
++ #define GM_RST_RX_FUNC_LBN 17
++ #define GM_RST_RX_FUNC_WIDTH 1
++ #define GM_RST_TX_FUNC_LBN 16
++ #define GM_RST_TX_FUNC_WIDTH 1
++ #define GM_LOOP_LBN 8
++ #define GM_LOOP_WIDTH 1
++ #define GM_RX_FC_EN_LBN 5
++ #define GM_RX_FC_EN_WIDTH 1
++ #define GM_TX_FC_EN_LBN 4
++ #define GM_TX_FC_EN_WIDTH 1
++ #define GM_SYNC_RXEN_LBN 3
++ #define GM_SYNC_RXEN_WIDTH 1
++ #define GM_RX_EN_LBN 2
++ #define GM_RX_EN_WIDTH 1
++ #define GM_SYNC_TXEN_LBN 1
++ #define GM_SYNC_TXEN_WIDTH 1
++ #define GM_TX_EN_LBN 0
++ #define GM_TX_EN_WIDTH 1
++#define GM_CFG2_REG_KER_OFST 0x10 /* GMAC configuration register 2 */
++#define GM_CFG2_REG_OFST 0x10 /* GMAC configuration register 2 */
++ #define GM_PAMBL_LEN_LBN 12
++ #define GM_PAMBL_LEN_WIDTH 4
++ #define GM_IF_MODE_LBN 8
++ #define GM_IF_MODE_WIDTH 2
++ #define GM_HUGE_FRM_EN_LBN 5
++ #define GM_HUGE_FRM_EN_WIDTH 1
++ #define GM_LEN_CHK_LBN 4
++ #define GM_LEN_CHK_WIDTH 1
++ #define GM_PAD_CRC_EN_LBN 2
++ #define GM_PAD_CRC_EN_WIDTH 1
++ #define GM_CRC_EN_LBN 1
++ #define GM_CRC_EN_WIDTH 1
++ #define GM_FD_LBN 0
++ #define GM_FD_WIDTH 1
++#define GM_IPG_REG_KER_OFST 0x20 /* GMAC IPG register */
++#define GM_IPG_REG_OFST 0x20 /* GMAC IPG register */
++ #define GM_NONB2B_IPG1_LBN 24
++ #define GM_NONB2B_IPG1_WIDTH 7
++ #define GM_NONB2B_IPG2_LBN 16
++ #define GM_NONB2B_IPG2_WIDTH 7
++ #define GM_MIN_IPG_ENF_LBN 8
++ #define GM_MIN_IPG_ENF_WIDTH 8
++ #define GM_B2B_IPG_LBN 0
++ #define GM_B2B_IPG_WIDTH 7
++#define GM_HD_REG_KER_OFST 0x30 /* GMAC half duplex register */
++#define GM_HD_REG_OFST 0x30 /* GMAC half duplex register */
++ #define GM_ALT_BOFF_VAL_LBN 20
++ #define GM_ALT_BOFF_VAL_WIDTH 4
++ #define GM_ALT_BOFF_EN_LBN 19
++ #define GM_ALT_BOFF_EN_WIDTH 1
++ #define GM_BP_NO_BOFF_LBN 18
++ #define GM_BP_NO_BOFF_WIDTH 1
++ #define GM_DIS_BOFF_LBN 17
++ #define GM_DIS_BOFF_WIDTH 1
++ #define GM_EXDEF_TX_EN_LBN 16
++ #define GM_EXDEF_TX_EN_WIDTH 1
++ #define GM_RTRY_LIMIT_LBN 12
++ #define GM_RTRY_LIMIT_WIDTH 4
++ #define GM_COL_WIN_LBN 0
++ #define GM_COL_WIN_WIDTH 10
++#define GM_MAX_FLEN_REG_KER_OFST 0x40 /* GMAC maximum frame length register */
++#define GM_MAX_FLEN_REG_OFST 0x40 /* GMAC maximum frame length register */
++ #define GM_MAX_FLEN_LBN 0
++ #define GM_MAX_FLEN_WIDTH 16
++#define GM_TEST_REG_KER_OFST 0x70 /* GMAC test register */
++#define GM_TEST_REG_OFST 0x70 /* GMAC test register */
++ #define GM_MAX_BOFF_LBN 3
++ #define GM_MAX_BOFF_WIDTH 1
++ #define GM_REG_TX_FLOW_EN_LBN 2
++ #define GM_REG_TX_FLOW_EN_WIDTH 1
++ #define GM_TEST_PAUSE_LBN 1
++ #define GM_TEST_PAUSE_WIDTH 1
++ #define GM_SHORT_SLOT_LBN 0
++ #define GM_SHORT_SLOT_WIDTH 1
++#define GM_ADR1_REG_KER_OFST 0x100 /* GMAC station address register 1 */
++#define GM_ADR1_REG_OFST 0x100 /* GMAC station address register 1 */
++ #define GM_ADR1_LBN 0
++ #define GM_ADR1_WIDTH 32
++#define GM_ADR2_REG_KER_OFST 0x110 /* GMAC station address register 2 */
++#define GM_ADR2_REG_OFST 0x110 /* GMAC station address register 2 */
++ #define GM_ADR2_LBN 16
++ #define GM_ADR2_WIDTH 16
++#define GMF_CFG0_REG_KER_OFST 0x120 /* GMAC FIFO configuration register 0 */
++#define GMF_CFG0_REG_OFST 0x120 /* GMAC FIFO configuration register 0 */
++ #define GMF_FTFENRPLY_LBN 20
++ #define GMF_FTFENRPLY_WIDTH 1
++ #define GMF_STFENRPLY_LBN 19
++ #define GMF_STFENRPLY_WIDTH 1
++ #define GMF_FRFENRPLY_LBN 18
++ #define GMF_FRFENRPLY_WIDTH 1
++ #define GMF_SRFENRPLY_LBN 17
++ #define GMF_SRFENRPLY_WIDTH 1
++ #define GMF_WTMENRPLY_LBN 16
++ #define GMF_WTMENRPLY_WIDTH 1
++ #define GMF_FTFENREQ_LBN 12
++ #define GMF_FTFENREQ_WIDTH 1
++ #define GMF_STFENREQ_LBN 11
++ #define GMF_STFENREQ_WIDTH 1
++ #define GMF_FRFENREQ_LBN 10
++ #define GMF_FRFENREQ_WIDTH 1
++ #define GMF_SRFENREQ_LBN 9
++ #define GMF_SRFENREQ_WIDTH 1
++ #define GMF_WTMENREQ_LBN 8
++ #define GMF_WTMENREQ_WIDTH 1
++ #define GMF_HSTRSTFT_LBN 4
++ #define GMF_HSTRSTFT_WIDTH 1
++ #define GMF_HSTRSTST_LBN 3
++ #define GMF_HSTRSTST_WIDTH 1
++ #define GMF_HSTRSTFR_LBN 2
++ #define GMF_HSTRSTFR_WIDTH 1
++ #define GMF_HSTRSTSR_LBN 1
++ #define GMF_HSTRSTSR_WIDTH 1
++ #define GMF_HSTRSTWT_LBN 0
++ #define GMF_HSTRSTWT_WIDTH 1
++#define GMF_CFG1_REG_KER_OFST 0x130 /* GMAC FIFO configuration register 1 */
++#define GMF_CFG1_REG_OFST 0x130 /* GMAC FIFO configuration register 1 */
++ #define GMF_CFGFRTH_LBN 16
++ #define GMF_CFGFRTH_WIDTH 5
++ #define GMF_CFGXOFFRTX_LBN 0
++ #define GMF_CFGXOFFRTX_WIDTH 16
++#define GMF_CFG2_REG_KER_OFST 0x140 /* GMAC FIFO configuration register 2 */
++#define GMF_CFG2_REG_OFST 0x140 /* GMAC FIFO configuration register 2 */
++ #define GMF_CFGHWM_LBN 16
++ #define GMF_CFGHWM_WIDTH 6
++ #define GMF_CFGLWM_LBN 0
++ #define GMF_CFGLWM_WIDTH 6
++#define GMF_CFG3_REG_KER_OFST 0x150 /* GMAC FIFO configuration register 3 */
++#define GMF_CFG3_REG_OFST 0x150 /* GMAC FIFO configuration register 3 */
++ #define GMF_CFGHWMFT_LBN 16
++ #define GMF_CFGHWMFT_WIDTH 6
++ #define GMF_CFGFTTH_LBN 0
++ #define GMF_CFGFTTH_WIDTH 6
++#define GMF_CFG4_REG_KER_OFST 0x160 /* GMAC FIFO configuration register 4 */
++#define GMF_CFG4_REG_OFST 0x160 /* GMAC FIFO configuration register 4 */
++ #define GMF_HSTFLTRFRM_LBN 0
++ #define GMF_HSTFLTRFRM_WIDTH 18
++#define GMF_CFG5_REG_KER_OFST 0x170 /* GMAC FIFO configuration register 5 */
++#define GMF_CFG5_REG_OFST 0x170 /* GMAC FIFO configuration register 5 */
++ #define GMF_CFGHDPLX_LBN 22
++ #define GMF_CFGHDPLX_WIDTH 1
++ #define GMF_SRFULL_LBN 21
++ #define GMF_SRFULL_WIDTH 1
++ #define GMF_HSTSRFULLCLR_LBN 20
++ #define GMF_HSTSRFULLCLR_WIDTH 1
++ #define GMF_CFGBYTMODE_LBN 19
++ #define GMF_CFGBYTMODE_WIDTH 1
++ #define GMF_HSTDRPLT64_LBN 18
++ #define GMF_HSTDRPLT64_WIDTH 1
++ #define GMF_HSTFLTRFRMDC_LBN 0
++ #define GMF_HSTFLTRFRMDC_WIDTH 18
++/*************---- 10G MAC Registers C Header ----*************/
++#define XM_ADR_LO_REG_KER_P0_OFST 0x1200 /* XGMAC address register low -
++ port 0 */
++#define XM_ADR_LO_REG_P0_OFST 0x1200 /* XGMAC address register low -
++ port 0 */
++ #define XM_ADR_LO_LBN 0
++ #define XM_ADR_LO_WIDTH 32
++#define XM_ADR_HI_REG_KER_P0_OFST 0x1210 /* XGMAC address register high -
++ port 0 */
++#define XM_ADR_HI_REG_P0_OFST 0x1210 /* XGMAC address register high -
++ port 0 */
++ #define XM_ADR_HI_LBN 0
++ #define XM_ADR_HI_WIDTH 16
++#define XM_GLB_CFG_REG_KER_P0_OFST 0x1220 /* XGMAC global configuration -
++ port 0 */
++#define XM_GLB_CFG_REG_P0_OFST 0x1220 /* XGMAC global configuration -
++ port 0 */
++ #define XM_LINE_LB_DEEP_RSVD_LBN 28
++ #define XM_LINE_LB_DEEP_RSVD_WIDTH 1
++ #define XM_RMTFLT_GEN_LBN 17
++ #define XM_RMTFLT_GEN_WIDTH 1
++ #define XM_DEBUG_MODE_LBN 16
++ #define XM_DEBUG_MODE_WIDTH 1
++ #define XM_RX_STAT_EN_LBN 11
++ #define XM_RX_STAT_EN_WIDTH 1
++ #define XM_TX_STAT_EN_LBN 10
++ #define XM_TX_STAT_EN_WIDTH 1
++ #define XM_CUT_THRU_MODE_LBN 7
++ #define XM_CUT_THRU_MODE_WIDTH 1
++ #define XM_RX_JUMBO_MODE_LBN 6
++ #define XM_RX_JUMBO_MODE_WIDTH 1
++ #define XM_WAN_MODE_LBN 5
++ #define XM_WAN_MODE_WIDTH 1
++ #define XM_AUTOCLR_MODE_LBN 4
++ #define XM_AUTOCLR_MODE_WIDTH 1
++ #define XM_INTCLR_MODE_LBN 3
++ #define XM_INTCLR_MODE_WIDTH 1
++ #define XM_CORE_RST_LBN 0
++ #define XM_CORE_RST_WIDTH 1
++#define XM_TX_CFG_REG_KER_P0_OFST 0x1230 /* XGMAC transmit configuration -
++ port 0 */
++#define XM_TX_CFG_REG_P0_OFST 0x1230 /* XGMAC transmit configuration -
++ port 0 */
++ #define XM_TX_PROG_LBN 24
++ #define XM_TX_PROG_WIDTH 1
++ #define XM_IPG_LBN 16
++ #define XM_IPG_WIDTH 4
++ #define XM_FCNTL_LBN 10
++ #define XM_FCNTL_WIDTH 1
++ #define XM_TXCRC_LBN 8
++ #define XM_TXCRC_WIDTH 1
++ #define XM_EDRC_LBN 6
++ #define XM_EDRC_WIDTH 1
++ #define XM_AUTO_PAD_LBN 5
++ #define XM_AUTO_PAD_WIDTH 1
++ #define XM_TX_PRMBL_LBN 2
++ #define XM_TX_PRMBL_WIDTH 1
++ #define XM_TXEN_LBN 1
++ #define XM_TXEN_WIDTH 1
++ #define XM_TX_RST_LBN 0
++ #define XM_TX_RST_WIDTH 1
++#define XM_RX_CFG_REG_KER_P0_OFST 0x1240 /* XGMAC receive configuration -
++ port 0 */
++#define XM_RX_CFG_REG_P0_OFST 0x1240 /* XGMAC receive configuration -
++ port 0 */
++ #define XM_PASS_LENERR_LBN 26
++ #define XM_PASS_LENERR_WIDTH 1
++ #define XM_PASS_CRC_ERR_LBN 25
++ #define XM_PASS_CRC_ERR_WIDTH 1
++ #define XM_PASS_PRMBLE_ERR_LBN 24
++ #define XM_PASS_PRMBLE_ERR_WIDTH 1
++ #define XM_REJ_UCAST_LBN 18
++ #define XM_REJ_UCAST_WIDTH 1
++ #define XM_BSC_EN_LBN 17
++ #define XM_BSC_EN_WIDTH 1
++ #define XM_ACPT_ALL_MCAST_LBN 11
++ #define XM_ACPT_ALL_MCAST_WIDTH 1
++ #define XM_PASS_SAP_LBN 10
++ #define XM_PASS_SAP_WIDTH 1
++ #define XM_ACPT_ALL_UCAST_LBN 9
++ #define XM_ACPT_ALL_UCAST_WIDTH 1
++ #define XM_AUTO_DEPAD_LBN 8
++ #define XM_AUTO_DEPAD_WIDTH 1
++ #define XM_RXCRC_LBN 3
++ #define XM_RXCRC_WIDTH 1
++ #define XM_RX_PRMBL_LBN 2
++ #define XM_RX_PRMBL_WIDTH 1
++ #define XM_RXEN_LBN 1
++ #define XM_RXEN_WIDTH 1
++ #define XM_RX_RST_LBN 0
++ #define XM_RX_RST_WIDTH 1
++#define XM_FC_REG_KER_P0_OFST 0x1270 /* XGMAC flow control register -
++ port 0 */
++#define XM_FC_REG_P0_OFST 0x1270 /* XGMAC flow control register -
++ port 0 */
++ #define XM_PAUSE_TIME_LBN 16
++ #define XM_PAUSE_TIME_WIDTH 16
++ #define XM_RX_MAC_STAT_LBN 11
++ #define XM_RX_MAC_STAT_WIDTH 1
++ #define XM_TX_MAC_STAT_LBN 10
++ #define XM_TX_MAC_STAT_WIDTH 1
++ #define XM_MCNTL_PASS_LBN 8
++ #define XM_MCNTL_PASS_WIDTH 2
++ #define XM_REJ_CNTL_UCAST_LBN 6
++ #define XM_REJ_CNTL_UCAST_WIDTH 1
++ #define XM_REJ_CNTL_MCAST_LBN 5
++ #define XM_REJ_CNTL_MCAST_WIDTH 1
++ #define XM_AUTO_XMIT_ZPAUSE_LBN 4
++ #define XM_AUTO_XMIT_ZPAUSE_WIDTH 1
++ #define XM_AUTO_XMIT_PAUSE_LBN 3
++ #define XM_AUTO_XMIT_PAUSE_WIDTH 1
++ #define XM_ZPAUSE_LBN 2
++ #define XM_ZPAUSE_WIDTH 1
++ #define XM_XMIT_PAUSE_LBN 1
++ #define XM_XMIT_PAUSE_WIDTH 1
++ #define XM_DIS_FCNTL_LBN 0
++ #define XM_DIS_FCNTL_WIDTH 1
++#define XM_PAUSE_TIME_REG_KER_P0_OFST 0x1290 /* XGMAC pause time register -
++ port 0 */
++#define XM_PAUSE_TIME_REG_P0_OFST 0x1290 /* XGMAC pause time register -
++ port 0 */
++ #define XM_TX_PAUSE_CNT_LBN 16
++ #define XM_TX_PAUSE_CNT_WIDTH 16
++ #define XM_RX_PAUSE_CNT_LBN 0
++ #define XM_RX_PAUSE_CNT_WIDTH 16
++#define XM_TX_PARAM_REG_KER_P0_OFST 0x12D0 /* XGMAC transmit parameter
++ register - port 0 */
++#define XM_TX_PARAM_REG_P0_OFST 0x12D0 /* XGMAC transmit parameter register -
++ port 0 */
++ #define XM_TX_JUMBO_MODE_LBN 31
++ #define XM_TX_JUMBO_MODE_WIDTH 1
++ #define XM_MAX_TX_FRM_SIZE_LBN 16
++ #define XM_MAX_TX_FRM_SIZE_WIDTH 14
++ #define XM_PAD_CHAR_LBN 0
++ #define XM_PAD_CHAR_WIDTH 8
++#define XM_RX_PARAM_REG_KER_P0_OFST 0x12E0 /* XGMAC receive parameter
++ register - port 0 */
++#define XM_RX_PARAM_REG_P0_OFST 0x12E0 /* XGMAC receive parameter register -
++ port 0 */
++ #define XM_MAX_RX_FRM_SIZE_LBN 0
++ #define XM_MAX_RX_FRM_SIZE_WIDTH 14
++#define XX_PWR_RST_REG_KER_P0_OFST 0x1300 /* XGXS/XAUI powerdown/reset
++ register */
++#define XX_PWR_RST_REG_P0_OFST 0x1300 /* XGXS/XAUI powerdown/reset register */
++ #define XX_PWRDND_SIG_LBN 31
++ #define XX_PWRDND_SIG_WIDTH 1
++ #define XX_PWRDNC_SIG_LBN 30
++ #define XX_PWRDNC_SIG_WIDTH 1
++ #define XX_PWRDNB_SIG_LBN 29
++ #define XX_PWRDNB_SIG_WIDTH 1
++ #define XX_PWRDNA_SIG_LBN 28
++ #define XX_PWRDNA_SIG_WIDTH 1
++ #define XX_SIM_MODE_LBN 27
++ #define XX_SIM_MODE_WIDTH 1
++ #define XX_RSTPLLCD_SIG_LBN 25
++ #define XX_RSTPLLCD_SIG_WIDTH 1
++ #define XX_RSTPLLAB_SIG_LBN 24
++ #define XX_RSTPLLAB_SIG_WIDTH 1
++ #define XX_RESETD_SIG_LBN 23
++ #define XX_RESETD_SIG_WIDTH 1
++ #define XX_RESETC_SIG_LBN 22
++ #define XX_RESETC_SIG_WIDTH 1
++ #define XX_RESETB_SIG_LBN 21
++ #define XX_RESETB_SIG_WIDTH 1
++ #define XX_RESETA_SIG_LBN 20
++ #define XX_RESETA_SIG_WIDTH 1
++ #define XX_RSTXGXSTX_SIG_LBN 18
++ #define XX_RSTXGXSTX_SIG_WIDTH 1
++ #define XX_RSTXGXSRX_SIG_LBN 17
++ #define XX_RSTXGXSRX_SIG_WIDTH 1
++ #define XX_SD_RST_ACT_LBN 16
++ #define XX_SD_RST_ACT_WIDTH 1
++ #define XX_PWRDND_EN_LBN 15
++ #define XX_PWRDND_EN_WIDTH 1
++ #define XX_PWRDNC_EN_LBN 14
++ #define XX_PWRDNC_EN_WIDTH 1
++ #define XX_PWRDNB_EN_LBN 13
++ #define XX_PWRDNB_EN_WIDTH 1
++ #define XX_PWRDNA_EN_LBN 12
++ #define XX_PWRDNA_EN_WIDTH 1
++ #define XX_RSTPLLCD_EN_LBN 9
++ #define XX_RSTPLLCD_EN_WIDTH 1
++ #define XX_RSTPLLAB_EN_LBN 8
++ #define XX_RSTPLLAB_EN_WIDTH 1
++ #define XX_RESETD_EN_LBN 7
++ #define XX_RESETD_EN_WIDTH 1
++ #define XX_RESETC_EN_LBN 6
++ #define XX_RESETC_EN_WIDTH 1
++ #define XX_RESETB_EN_LBN 5
++ #define XX_RESETB_EN_WIDTH 1
++ #define XX_RESETA_EN_LBN 4
++ #define XX_RESETA_EN_WIDTH 1
++ #define XX_RSTXGXSTX_EN_LBN 2
++ #define XX_RSTXGXSTX_EN_WIDTH 1
++ #define XX_RSTXGXSRX_EN_LBN 1
++ #define XX_RSTXGXSRX_EN_WIDTH 1
++ #define XX_RST_XX_EN_LBN 0
++ #define XX_RST_XX_EN_WIDTH 1
++#define XX_SD_CTL_REG_KER_P0_OFST 0x1310 /* XGXS/XAUI powerdown/reset control
++ register */
++#define XX_SD_CTL_REG_P0_OFST 0x1310 /* XGXS/XAUI powerdown/reset control
++ register */
++ #define XX_TERMADJ1_LBN 17
++ #define XX_TERMADJ1_WIDTH 1
++ #define XX_TERMADJ0_LBN 16
++ #define XX_TERMADJ0_WIDTH 1
++ #define XX_HIDRVD_LBN 15
++ #define XX_HIDRVD_WIDTH 1
++ #define XX_LODRVD_LBN 14
++ #define XX_LODRVD_WIDTH 1
++ #define XX_HIDRVC_LBN 13
++ #define XX_HIDRVC_WIDTH 1
++ #define XX_LODRVC_LBN 12
++ #define XX_LODRVC_WIDTH 1
++ #define XX_HIDRVB_LBN 11
++ #define XX_HIDRVB_WIDTH 1
++ #define XX_LODRVB_LBN 10
++ #define XX_LODRVB_WIDTH 1
++ #define XX_HIDRVA_LBN 9
++ #define XX_HIDRVA_WIDTH 1
++ #define XX_LODRVA_LBN 8
++ #define XX_LODRVA_WIDTH 1
++ #define XX_LPBKD_LBN 3
++ #define XX_LPBKD_WIDTH 1
++ #define XX_LPBKC_LBN 2
++ #define XX_LPBKC_WIDTH 1
++ #define XX_LPBKB_LBN 1
++ #define XX_LPBKB_WIDTH 1
++ #define XX_LPBKA_LBN 0
++ #define XX_LPBKA_WIDTH 1
++#define XX_TXDRV_CTL_REG_KER_P0_OFST 0x1320 /* XAUI SerDes transmit drive
++ control register */
++#define XX_TXDRV_CTL_REG_P0_OFST 0x1320 /* XAUI SerDes transmit drive
++ control register */
++ #define XX_DEQD_LBN 28
++ #define XX_DEQD_WIDTH 4
++ #define XX_DEQC_LBN 24
++ #define XX_DEQC_WIDTH 4
++ #define XX_DEQB_LBN 20
++ #define XX_DEQB_WIDTH 4
++ #define XX_DEQA_LBN 16
++ #define XX_DEQA_WIDTH 4
++ #define XX_DTXD_LBN 12
++ #define XX_DTXD_WIDTH 4
++ #define XX_DTXC_LBN 8
++ #define XX_DTXC_WIDTH 4
++ #define XX_DTXB_LBN 4
++ #define XX_DTXB_WIDTH 4
++ #define XX_DTXA_LBN 0
++ #define XX_DTXA_WIDTH 4
++#define XX_PRBS_CTL_REG_KER_P0_OFST 0x1330 /* XAUI PRBS control register */
++#define XX_PRBS_CTL_REG_P0_OFST 0x1330 /* XAUI PRBS control register */
++ #define XX_CH3_RX_PRBS_SEL_LBN 30
++ #define XX_CH3_RX_PRBS_SEL_WIDTH 2
++ #define XX_CH3_RX_PRBS_INV_LBN 29
++ #define XX_CH3_RX_PRBS_INV_WIDTH 1
++ #define XX_CH3_RX_PRBS_CHKEN_LBN 28
++ #define XX_CH3_RX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH2_RX_PRBS_SEL_LBN 26
++ #define XX_CH2_RX_PRBS_SEL_WIDTH 2
++ #define XX_CH2_RX_PRBS_INV_LBN 25
++ #define XX_CH2_RX_PRBS_INV_WIDTH 1
++ #define XX_CH2_RX_PRBS_CHKEN_LBN 24
++ #define XX_CH2_RX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH1_RX_PRBS_SEL_LBN 22
++ #define XX_CH1_RX_PRBS_SEL_WIDTH 2
++ #define XX_CH1_RX_PRBS_INV_LBN 21
++ #define XX_CH1_RX_PRBS_INV_WIDTH 1
++ #define XX_CH1_RX_PRBS_CHKEN_LBN 20
++ #define XX_CH1_RX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH0_RX_PRBS_SEL_LBN 18
++ #define XX_CH0_RX_PRBS_SEL_WIDTH 2
++ #define XX_CH0_RX_PRBS_INV_LBN 17
++ #define XX_CH0_RX_PRBS_INV_WIDTH 1
++ #define XX_CH0_RX_PRBS_CHKEN_LBN 16
++ #define XX_CH0_RX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH3_TX_PRBS_SEL_LBN 14
++ #define XX_CH3_TX_PRBS_SEL_WIDTH 2
++ #define XX_CH3_TX_PRBS_INV_LBN 13
++ #define XX_CH3_TX_PRBS_INV_WIDTH 1
++ #define XX_CH3_TX_PRBS_CHKEN_LBN 12
++ #define XX_CH3_TX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH2_TX_PRBS_SEL_LBN 10
++ #define XX_CH2_TX_PRBS_SEL_WIDTH 2
++ #define XX_CH2_TX_PRBS_INV_LBN 9
++ #define XX_CH2_TX_PRBS_INV_WIDTH 1
++ #define XX_CH2_TX_PRBS_CHKEN_LBN 8
++ #define XX_CH2_TX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH1_TX_PRBS_SEL_LBN 6
++ #define XX_CH1_TX_PRBS_SEL_WIDTH 2
++ #define XX_CH1_TX_PRBS_INV_LBN 5
++ #define XX_CH1_TX_PRBS_INV_WIDTH 1
++ #define XX_CH1_TX_PRBS_CHKEN_LBN 4
++ #define XX_CH1_TX_PRBS_CHKEN_WIDTH 1
++ #define XX_CH0_TX_PRBS_SEL_LBN 2
++ #define XX_CH0_TX_PRBS_SEL_WIDTH 2
++ #define XX_CH0_TX_PRBS_INV_LBN 1
++ #define XX_CH0_TX_PRBS_INV_WIDTH 1
++ #define XX_CH0_TX_PRBS_CHKEN_LBN 0
++ #define XX_CH0_TX_PRBS_CHKEN_WIDTH 1
++#define XX_PRBS_CHK_REG_KER_P0_OFST 0x1340 /* XAUI PRBS checker control
++ register */
++#define XX_PRBS_CHK_REG_P0_OFST 0x1340 /* XAUI PRBS checker control
++ register */
++ #define XX_REV_LB_EN_LBN 16
++ #define XX_REV_LB_EN_WIDTH 1
++ #define XX_CH3_DEG_DET_LBN 15
++ #define XX_CH3_DEG_DET_WIDTH 1
++ #define XX_CH3_LFSR_LOCK_IND_LBN 14
++ #define XX_CH3_LFSR_LOCK_IND_WIDTH 1
++ #define XX_CH3_PRBS_FRUN_LBN 13
++ #define XX_CH3_PRBS_FRUN_WIDTH 1
++ #define XX_CH3_ERR_CHK_LBN 12
++ #define XX_CH3_ERR_CHK_WIDTH 1
++ #define XX_CH2_DEG_DET_LBN 11
++ #define XX_CH2_DEG_DET_WIDTH 1
++ #define XX_CH2_LFSR_LOCK_IND_LBN 10
++ #define XX_CH2_LFSR_LOCK_IND_WIDTH 1
++ #define XX_CH2_PRBS_FRUN_LBN 9
++ #define XX_CH2_PRBS_FRUN_WIDTH 1
++ #define XX_CH2_ERR_CHK_LBN 8
++ #define XX_CH2_ERR_CHK_WIDTH 1
++ #define XX_CH1_DEG_DET_LBN 7
++ #define XX_CH1_DEG_DET_WIDTH 1
++ #define XX_CH1_LFSR_LOCK_IND_LBN 6
++ #define XX_CH1_LFSR_LOCK_IND_WIDTH 1
++ #define XX_CH1_PRBS_FRUN_LBN 5
++ #define XX_CH1_PRBS_FRUN_WIDTH 1
++ #define XX_CH1_ERR_CHK_LBN 4
++ #define XX_CH1_ERR_CHK_WIDTH 1
++ #define XX_CH0_DEG_DET_LBN 3
++ #define XX_CH0_DEG_DET_WIDTH 1
++ #define XX_CH0_LFSR_LOCK_IND_LBN 2
++ #define XX_CH0_LFSR_LOCK_IND_WIDTH 1
++ #define XX_CH0_PRBS_FRUN_LBN 1
++ #define XX_CH0_PRBS_FRUN_WIDTH 1
++ #define XX_CH0_ERR_CHK_LBN 0
++ #define XX_CH0_ERR_CHK_WIDTH 1
++#define XX_PRBS_ERR_REG_KER_P0_OFST 0x1350 /* XAUI PRBS checker error
++ count register */
++#define XX_PRBS_ERR_REG_P0_OFST 0x1350 /* XAUI PRBS checker error count
++ register */
++ #define XX_CH3_PRBS_ERR_CNT_LBN 24
++ #define XX_CH3_PRBS_ERR_CNT_WIDTH 8
++ #define XX_CH2_PRBS_ERR_CNT_LBN 16
++ #define XX_CH2_PRBS_ERR_CNT_WIDTH 8
++ #define XX_CH1_PRBS_ERR_CNT_LBN 8
++ #define XX_CH1_PRBS_ERR_CNT_WIDTH 8
++ #define XX_CH0_PRBS_ERR_CNT_LBN 0
++ #define XX_CH0_PRBS_ERR_CNT_WIDTH 8
++#define XX_CORE_STAT_REG_KER_P0_OFST 0x1360 /* XAUI XGXS core status
++ register */
++#define XX_CORE_STAT_REG_P0_OFST 0x1360 /* XAUI XGXS core status register */
++ #define XX_FORCE_SIG3_LBN 31
++ #define XX_FORCE_SIG3_WIDTH 1
++ #define XX_FORCE_SIG3_VAL_LBN 30
++ #define XX_FORCE_SIG3_VAL_WIDTH 1
++ #define XX_FORCE_SIG2_LBN 29
++ #define XX_FORCE_SIG2_WIDTH 1
++ #define XX_FORCE_SIG2_VAL_LBN 28
++ #define XX_FORCE_SIG2_VAL_WIDTH 1
++ #define XX_FORCE_SIG1_LBN 27
++ #define XX_FORCE_SIG1_WIDTH 1
++ #define XX_FORCE_SIG1_VAL_LBN 26
++ #define XX_FORCE_SIG1_VAL_WIDTH 1
++ #define XX_FORCE_SIG0_LBN 25
++ #define XX_FORCE_SIG0_WIDTH 1
++ #define XX_FORCE_SIG0_VAL_LBN 24
++ #define XX_FORCE_SIG0_VAL_WIDTH 1
++ #define XX_XGXS_LB_EN_LBN 23
++ #define XX_XGXS_LB_EN_WIDTH 1
++ #define XX_XGMII_LB_EN_LBN 22
++ #define XX_XGMII_LB_EN_WIDTH 1
++ #define XX_MATCH_FAULT_LBN 21
++ #define XX_MATCH_FAULT_WIDTH 1
++ #define XX_ALIGN_DONE_LBN 20
++ #define XX_ALIGN_DONE_WIDTH 1
++ #define XX_SYNC_STAT3_LBN 19
++ #define XX_SYNC_STAT3_WIDTH 1
++ #define XX_SYNC_STAT2_LBN 18
++ #define XX_SYNC_STAT2_WIDTH 1
++ #define XX_SYNC_STAT1_LBN 17
++ #define XX_SYNC_STAT1_WIDTH 1
++ #define XX_SYNC_STAT0_LBN 16
++ #define XX_SYNC_STAT0_WIDTH 1
++ #define XX_COMMA_DET_CH3_LBN 15
++ #define XX_COMMA_DET_CH3_WIDTH 1
++ #define XX_COMMA_DET_CH2_LBN 14
++ #define XX_COMMA_DET_CH2_WIDTH 1
++ #define XX_COMMA_DET_CH1_LBN 13
++ #define XX_COMMA_DET_CH1_WIDTH 1
++ #define XX_COMMA_DET_CH0_LBN 12
++ #define XX_COMMA_DET_CH0_WIDTH 1
++ #define XX_CGRP_ALIGN_CH3_LBN 11
++ #define XX_CGRP_ALIGN_CH3_WIDTH 1
++ #define XX_CGRP_ALIGN_CH2_LBN 10
++ #define XX_CGRP_ALIGN_CH2_WIDTH 1
++ #define XX_CGRP_ALIGN_CH1_LBN 9
++ #define XX_CGRP_ALIGN_CH1_WIDTH 1
++ #define XX_CGRP_ALIGN_CH0_LBN 8
++ #define XX_CGRP_ALIGN_CH0_WIDTH 1
++ #define XX_CHAR_ERR_CH3_LBN 7
++ #define XX_CHAR_ERR_CH3_WIDTH 1
++ #define XX_CHAR_ERR_CH2_LBN 6
++ #define XX_CHAR_ERR_CH2_WIDTH 1
++ #define XX_CHAR_ERR_CH1_LBN 5
++ #define XX_CHAR_ERR_CH1_WIDTH 1
++ #define XX_CHAR_ERR_CH0_LBN 4
++ #define XX_CHAR_ERR_CH0_WIDTH 1
++ #define XX_DISPERR_CH3_LBN 3
++ #define XX_DISPERR_CH3_WIDTH 1
++ #define XX_DISPERR_CH2_LBN 2
++ #define XX_DISPERR_CH2_WIDTH 1
++ #define XX_DISPERR_CH1_LBN 1
++ #define XX_DISPERR_CH1_WIDTH 1
++ #define XX_DISPERR_CH0_LBN 0
++ #define XX_DISPERR_CH0_WIDTH 1
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_xgrmon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon/falcon_xgrmon.h 2008-05-19 00:33:29.341837424 +0300
+@@ -0,0 +1,125 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) 10G MAC
++ * statistics register definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*************---- 10G MAC Statistical Counters C Header ----*************/
++#define XgRxOctets_offset 0x0
++ #define XgRxOctets_WIDTH 48
++#define XgRxOctetsOK_offset 0x8
++ #define XgRxOctetsOK_WIDTH 48
++#define XgRxPkts_offset 0x10
++ #define XgRxPkts_WIDTH 32
++#define XgRxPktsOK_offset 0x14
++ #define XgRxPktsOK_WIDTH 32
++#define XgRxBroadcastPkts_offset 0x18
++ #define XgRxBroadcastPkts_WIDTH 32
++#define XgRxMulticastPkts_offset 0x1C
++ #define XgRxMulticastPkts_WIDTH 32
++#define XgRxUnicastPkts_offset 0x20
++ #define XgRxUnicastPkts_WIDTH 32
++#define XgRxUndersizePkts_offset 0x24
++ #define XgRxUndersizePkts_WIDTH 32
++#define XgRxOversizePkts_offset 0x28
++ #define XgRxOversizePkts_WIDTH 32
++#define XgRxJabberPkts_offset 0x2C
++ #define XgRxJabberPkts_WIDTH 32
++#define XgRxUndersizeFCSerrorPkts_offset 0x30
++ #define XgRxUndersizeFCSerrorPkts_WIDTH 32
++#define XgRxDropEvents_offset 0x34
++ #define XgRxDropEvents_WIDTH 32
++#define XgRxFCSerrorPkts_offset 0x38
++ #define XgRxFCSerrorPkts_WIDTH 32
++#define XgRxAlignError_offset 0x3C
++ #define XgRxAlignError_WIDTH 32
++#define XgRxSymbolError_offset 0x40
++ #define XgRxSymbolError_WIDTH 32
++#define XgRxInternalMACError_offset 0x44
++ #define XgRxInternalMACError_WIDTH 32
++#define XgRxControlPkts_offset 0x48
++ #define XgRxControlPkts_WIDTH 32
++#define XgRxPausePkts_offset 0x4C
++ #define XgRxPausePkts_WIDTH 32
++#define XgRxPkts64Octets_offset 0x50
++ #define XgRxPkts64Octets_WIDTH 32
++#define XgRxPkts65to127Octets_offset 0x54
++ #define XgRxPkts65to127Octets_WIDTH 32
++#define XgRxPkts128to255Octets_offset 0x58
++ #define XgRxPkts128to255Octets_WIDTH 32
++#define XgRxPkts256to511Octets_offset 0x5C
++ #define XgRxPkts256to511Octets_WIDTH 32
++#define XgRxPkts512to1023Octets_offset 0x60
++ #define XgRxPkts512to1023Octets_WIDTH 32
++#define XgRxPkts1024to15xxOctets_offset 0x64
++ #define XgRxPkts1024to15xxOctets_WIDTH 32
++#define XgRxPkts15xxtoMaxOctets_offset 0x68
++ #define XgRxPkts15xxtoMaxOctets_WIDTH 32
++#define XgRxLengthError_offset 0x6C
++ #define XgRxLengthError_WIDTH 32
++#define XgTxPkts_offset 0x80
++ #define XgTxPkts_WIDTH 32
++#define XgTxOctets_offset 0x88
++ #define XgTxOctets_WIDTH 48
++#define XgTxMulticastPkts_offset 0x90
++ #define XgTxMulticastPkts_WIDTH 32
++#define XgTxBroadcastPkts_offset 0x94
++ #define XgTxBroadcastPkts_WIDTH 32
++#define XgTxUnicastPkts_offset 0x98
++ #define XgTxUnicastPkts_WIDTH 32
++#define XgTxControlPkts_offset 0x9C
++ #define XgTxControlPkts_WIDTH 32
++#define XgTxPausePkts_offset 0xA0
++ #define XgTxPausePkts_WIDTH 32
++#define XgTxPkts64Octets_offset 0xA4
++ #define XgTxPkts64Octets_WIDTH 32
++#define XgTxPkts65to127Octets_offset 0xA8
++ #define XgTxPkts65to127Octets_WIDTH 32
++#define XgTxPkts128to255Octets_offset 0xAC
++ #define XgTxPkts128to255Octets_WIDTH 32
++#define XgTxPkts256to511Octets_offset 0xB0
++ #define XgTxPkts256to511Octets_WIDTH 32
++#define XgTxPkts512to1023Octets_offset 0xB4
++ #define XgTxPkts512to1023Octets_WIDTH 32
++#define XgTxPkts1024to15xxOctets_offset 0xB8
++ #define XgTxPkts1024to15xxOctets_WIDTH 32
++#define XgTxPkts1519toMaxOctets_offset 0xBC
++ #define XgTxPkts1519toMaxOctets_WIDTH 32
++#define XgTxUndersizePkts_offset 0xC0
++ #define XgTxUndersizePkts_WIDTH 32
++#define XgTxOversizePkts_offset 0xC4
++ #define XgTxOversizePkts_WIDTH 32
++#define xGTxNonTcpUdpPkt_offset 0xC8
++ #define xGTxNonTcpUdpPkt_WIDTH 16
++#define xGTxMacSrcErrPkt_offset 0xCC
++ #define xGTxMacSrcErrPkt_WIDTH 16
++#define xGTxIpSrcErrPkt_offset 0xD0
++ #define xGTxIpSrcErrPkt_WIDTH 16
++#define XgDmaDone_offset 0xD4
++ #define XgDmaDone_WIDTH 32
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/falcon.h 2008-05-19 00:33:29.329836732 +0300
+@@ -0,0 +1,420 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC - EFXXXX (aka Falcon) specific
++ * definitions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_DRIVER_EFAB_HARDWARE_FALCON_H__
++#define __CI_DRIVER_EFAB_HARDWARE_FALCON_H__
++
++/*----------------------------------------------------------------------------
++ * Compile options
++ *---------------------------------------------------------------------------*/
++
++/* Falcon has an 8K maximum page size. */
++#define FALCON_MAX_PAGE_SIZE EFHW_8K
++
++/* include the register definitions */
++#include <ci/driver/efab/hardware/falcon/falcon_core.h>
++#include <ci/driver/efab/hardware/falcon/falcon_desc.h>
++#include <ci/driver/efab/hardware/falcon/falcon_event.h>
++#include <ci/driver/efab/hardware/falcon/falcon_mac.h>
++#include <ci/driver/efab/hardware/falcon/falcon_grmon.h>
++#include <ci/driver/efab/hardware/falcon/falcon_xgrmon.h>
++#include <ci/driver/efab/hardware/falcon/falcon_intr_vec.h>
++
++#define FALCON_DMA_TX_DESC_BYTES 8
++#define FALCON_DMA_RX_PHYS_DESC_BYTES 8
++#define FALCON_DMA_RX_BUF_DESC_BYTES 4
++
++
++/* ---- efhw_event_t helpers --- */
++
++#ifndef EFHW_IS_LITTLE_ENDIAN
++#error This needs lots of cpu_to_le64s() in
++#endif
++
++/*!\ TODO look at whether there is an efficiency gain to be had by
++ treating the event codes to 32bit masks as is done for EF1
++
++ These masks apply to the full 64 bits of the event to extract the
++ event code - followed by the common event codes to expect
++ */
++#define __FALCON_OPEN_MASK(WIDTH) ((((uint64_t)1) << (WIDTH)) - 1)
++#define FALCON_EVENT_CODE_MASK \
++ (__FALCON_OPEN_MASK(EV_CODE_WIDTH) << EV_CODE_LBN)
++#define FALCON_EVENT_EV_Q_ID_MASK \
++ (__FALCON_OPEN_MASK(DRIVER_EV_EVQ_ID_WIDTH) << DRIVER_EV_EVQ_ID_LBN)
++#define FALCON_EVENT_TX_FLUSH_Q_ID_MASK \
++ (__FALCON_OPEN_MASK(DRIVER_EV_TX_DESCQ_ID_WIDTH) << \
++ DRIVER_EV_TX_DESCQ_ID_LBN)
++#define FALCON_EVENT_RX_FLUSH_Q_ID_MASK \
++ (__FALCON_OPEN_MASK(DRIVER_EV_RX_DESCQ_ID_WIDTH) << \
++ DRIVER_EV_RX_DESCQ_ID_LBN)
++#define FALCON_EVENT_DRV_SUBCODE_MASK \
++ (__FALCON_OPEN_MASK(DRIVER_EV_SUB_CODE_WIDTH) << \
++ DRIVER_EV_SUB_CODE_LBN)
++
++#define FALCON_EVENT_FMT "[ev:%x:%08x:%08x]"
++#define FALCON_EVENT_PRI_ARG(e) \
++ ((unsigned)(((e).u64 & FALCON_EVENT_CODE_MASK) >> EV_CODE_LBN)), \
++ ((unsigned)((e).u64 >> 32)), ((unsigned)((e).u64 & 0xFFFFFFFF))
++
++#define FALCON_EVENT_CODE(evp) ((evp)->u64 & FALCON_EVENT_CODE_MASK)
++#define FALCON_EVENT_WAKE_EVQ_ID(evp) \
++ (((evp)->u64 & FALCON_EVENT_EV_Q_ID_MASK) >> DRIVER_EV_EVQ_ID_LBN)
++#define FALCON_EVENT_TX_FLUSH_Q_ID(evp) \
++ (((evp)->u64 & FALCON_EVENT_TX_FLUSH_Q_ID_MASK) >> \
++ DRIVER_EV_TX_DESCQ_ID_LBN)
++#define FALCON_EVENT_RX_FLUSH_Q_ID(evp) \
++ (((evp)->u64 & FALCON_EVENT_RX_FLUSH_Q_ID_MASK) >> \
++ DRIVER_EV_RX_DESCQ_ID_LBN)
++#define FALCON_EVENT_DRIVER_SUBCODE(evp) \
++ (((evp)->u64 & FALCON_EVENT_DRV_SUBCODE_MASK) >> \
++ DRIVER_EV_SUB_CODE_LBN)
++
++#define FALCON_EVENT_CODE_CHAR ((uint64_t)DRIVER_EV_DECODE << EV_CODE_LBN)
++#define FALCON_EVENT_CODE_SW ((uint64_t)DRV_GEN_EV_DECODE << EV_CODE_LBN)
++
++
++/* so this is the size in bytes of an awful lot of things */
++#define FALCON_REGISTER128 (16)
++
++/* we define some unique dummy values as a debug aid */
++#ifdef _WIN32
++#define FALCON_ATOMIC_BASE 0xdeadbeef00000000ui64
++#else
++#define FALCON_ATOMIC_BASE 0xdeadbeef00000000ULL
++#endif
++#define FALCON_ATOMIC_UPD_REG (FALCON_ATOMIC_BASE | 0x1)
++#define FALCON_ATOMIC_PTR_TBL_REG (FALCON_ATOMIC_BASE | 0x2)
++#define FALCON_ATOMIC_SRPM_UDP_EVQ_REG (FALCON_ATOMIC_BASE | 0x3)
++#define FALCON_ATOMIC_RX_FLUSH_DESCQ (FALCON_ATOMIC_BASE | 0x4)
++#define FALCON_ATOMIC_TX_FLUSH_DESCQ (FALCON_ATOMIC_BASE | 0x5)
++#define FALCON_ATOMIC_INT_EN_REG (FALCON_ATOMIC_BASE | 0x6)
++#define FALCON_ATOMIC_TIMER_CMD_REG (FALCON_ATOMIC_BASE | 0x7)
++#define FALCON_ATOMIC_PACE_REG (FALCON_ATOMIC_BASE | 0x8)
++#define FALCON_ATOMIC_INT_ACK_REG (FALCON_ATOMIC_BASE | 0x9)
++/* XXX It crashed with odd value in FALCON_ATOMIC_INT_ADR_REG */
++#define FALCON_ATOMIC_INT_ADR_REG (FALCON_ATOMIC_BASE | 0xa)
+
-+ return ret;
-+}
++/*----------------------------------------------------------------------------
++ *
++ * PCI control blocks for Falcon -
++ * (P) primary is for NET
++ * (S) secondary is for CHAR
++ *
++ *---------------------------------------------------------------------------*/
+
-+HYPERVISOR_ATTR_RO(compiler);
++#define FALCON_P_CTR_AP_BAR 2
++#define FALCON_S_CTR_AP_BAR 0
++#define FALCON_S_DEVID 0x6703
+
-+static ssize_t compiled_by_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ int ret = -ENOMEM;
-+ struct xen_compile_info *info;
+
-+ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
-+ if (info) {
-+ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", info->compile_by);
-+ kfree(info);
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * Falcon constants
++ *
++ *---------------------------------------------------------------------------*/
+
-+ return ret;
-+}
++#define FALCON_DMAQ_NUM (EFHW_4K)
++#define FALCON_EVQ_TBL_NUM (EFHW_4K)
++#define FALCON_TIMERS_NUM (EFHW_4K)
+
-+HYPERVISOR_ATTR_RO(compiled_by);
++/* This value is an upper limit on the total number of filter table
++ * entries, including odd and even banks. The actual size of filter table
++ * is determined at runtime, as it can vary.
++ */
++#define FALCON_FILTER_TBL_NUM (EFHW_16K)
+
-+static ssize_t compile_date_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ int ret = -ENOMEM;
-+ struct xen_compile_info *info;
++/* max number of buffers which can be pushed before commiting */
++#define FALCON_BUFFER_UPD_MAX (128)
+
-+ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
-+ if (info) {
-+ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", info->compile_date);
-+ kfree(info);
-+ }
++/* We can tell falcon to write its RX buffers in 32 byte quantums,
++ and since we pad packets 2 bytes to the right we can't use
++ a full page (not unless we use jumbo mode for all queues)
+
-+ return ret;
-+}
++ NOTE: tests/nic/dma.c assumes that the value here is the real NIC
++ value, so we explicitly round it down to the nearest 32 bytes */
+
-+HYPERVISOR_ATTR_RO(compile_date);
++/* #define FALCON_RX_USR_BUF_SIZE round_down(4096-2,32) */
++#define FALCON_RX_USR_BUF_SIZE 4064
+
-+static struct attribute *xen_compile_attrs[] = {
-+ &compiler_attr.attr,
-+ &compiled_by_attr.attr,
-+ &compile_date_attr.attr,
-+ NULL
-+};
++#define FALCON_EVQ_RPTR_REG_P0 0x400
+
-+static struct attribute_group xen_compilation_group = {
-+ .name = "compilation",
-+ .attrs = xen_compile_attrs,
++/*----------------------------------------------------------------------------
++ *
++ * Falcon requires user-space descriptor pushes to be:
++ * dword[0-2]; wiob(); dword[3]
++ *
++ * Driver register access must be locked against other threads from
++ * the same driver but can be in any order: i.e dword[0-3]; wiob()
++ *
++ * The following helpers ensure that valid dword orderings are exercised
++ *
++ *---------------------------------------------------------------------------*/
++
++/* A union to allow writting 64bit values as 32bit values, without
++ * hitting the compilers aliasing rules. We hope the compiler optimises
++ * away the copy's anyway */
++union __u64to32 {
++ uint64_t u64;
++ struct {
++#ifdef EFHW_IS_LITTLE_ENDIAN
++ uint32_t a;
++ uint32_t b;
++#else
++ uint32_t b;
++ uint32_t a;
++#endif
++ } s;
+};
+
-+int __init static xen_compilation_init(void)
++static inline void
++falcon_write_ddd_d(efhw_ioaddr_t kva,
++ uint32_t d0, uint32_t d1, uint32_t d2, uint32_t d3)
+{
-+ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
-+ &xen_compilation_group);
++ writel(d0, kva + 0);
++ writel(d1, kva + 4);
++ writel(d2, kva + 8);
++ mmiowb();
++ writel(d3, kva + 12);
+}
+
-+static void xen_compilation_destroy(void)
++static inline void falcon_write_q(efhw_ioaddr_t kva, uint64_t q)
+{
-+ sysfs_remove_group(&hypervisor_subsys.kset.kobj,
-+ &xen_compilation_group);
-+}
++ union __u64to32 u;
++ u.u64 = q;
+
-+/* xen properties info */
++ writel(u.s.a, kva);
++ mmiowb();
++ writel(u.s.b, kva + 4);
++}
+
-+static ssize_t capabilities_show(struct hyp_sysfs_attr *attr, char *buffer)
++static inline void falcon_read_q(efhw_ioaddr_t addr, uint64_t *q0)
+{
-+ int ret = -ENOMEM;
-+ char *caps;
-+
-+ caps = kmalloc(XEN_CAPABILITIES_INFO_LEN, GFP_KERNEL);
-+ if (caps) {
-+ ret = HYPERVISOR_xen_version(XENVER_capabilities, caps);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", caps);
-+ kfree(caps);
-+ }
++ /* It is essential that we read dword0 first, so that
++ * the shadow register is updated with the latest value
++ * and we get a self consistent value.
++ */
++ union __u64to32 u;
++ u.s.a = readl(addr);
++ rmb();
++ u.s.b = readl(addr + 4);
+
-+ return ret;
++ *q0 = u.u64;
+}
+
-+HYPERVISOR_ATTR_RO(capabilities);
++static inline void
++falcon_write_qq(efhw_ioaddr_t kva, uint64_t q0, uint64_t q1)
++{
++ writeq(q0, kva + 0);
++ falcon_write_q(kva + 8, q1);
++}
+
-+static ssize_t changeset_show(struct hyp_sysfs_attr *attr, char *buffer)
++static inline void
++falcon_read_qq(efhw_ioaddr_t addr, uint64_t *q0, uint64_t *q1)
+{
-+ int ret = -ENOMEM;
-+ char *cset;
++ falcon_read_q(addr, q0);
++ *q1 = readq(addr + 8);
++}
+
-+ cset = kmalloc(XEN_CHANGESET_INFO_LEN, GFP_KERNEL);
-+ if (cset) {
-+ ret = HYPERVISOR_xen_version(XENVER_changeset, cset);
-+ if (!ret)
-+ ret = sprintf(buffer, "%s\n", cset);
-+ kfree(cset);
-+ }
+
-+ return ret;
-+}
+
-+HYPERVISOR_ATTR_RO(changeset);
++/*----------------------------------------------------------------------------
++ *
++ * Buffer virtual addresses (4K buffers)
++ *
++ *---------------------------------------------------------------------------*/
+
-+static ssize_t virtual_start_show(struct hyp_sysfs_attr *attr, char *buffer)
-+{
-+ int ret = -ENOMEM;
-+ struct xen_platform_parameters *parms;
++/* Form a buffer virtual address from buffer ID and offset. If the offset
++** is larger than the buffer size, then the buffer indexed will be
++** calculated appropriately. It is the responsibility of the caller to
++** ensure that they have valid buffers programmed at that address.
++*/
++#define FALCON_VADDR_8K_S (13)
++#define FALCON_VADDR_4K_S (12)
++#define FALCON_VADDR_M 0xfffff /* post shift mask */
+
-+ parms = kmalloc(sizeof(struct xen_platform_parameters), GFP_KERNEL);
-+ if (parms) {
-+ ret = HYPERVISOR_xen_version(XENVER_platform_parameters,
-+ parms);
-+ if (!ret)
-+ ret = sprintf(buffer, "%lx\n", parms->virt_start);
-+ kfree(parms);
-+ }
++#define FALCON_BUFFER_8K_ADDR(id, off) (((id) << FALCON_VADDR_8K_S) + (off))
++#define FALCON_BUFFER_8K_PAGE(vaddr) \
++ (((vaddr) >> FALCON_VADDR_8K_S) & FALCON_VADDR_M)
++#define FALCON_BUFFER_8K_OFF(vaddr) \
++ ((vaddr) & __FALCON_MASK32(FALCON_VADDR_8K_S))
+
-+ return ret;
-+}
++#define FALCON_BUFFER_4K_ADDR(id, off) (((id) << FALCON_VADDR_4K_S) + (off))
++#define FALCON_BUFFER_4K_PAGE(vaddr) \
++ (((vaddr) >> FALCON_VADDR_4K_S) & FALCON_VADDR_M)
++#define FALCON_BUFFER_4K_OFF(vaddr) \
++ ((vaddr) & __FALCON_MASK32(FALCON_VADDR_4K_S))
+
-+HYPERVISOR_ATTR_RO(virtual_start);
++/*----------------------------------------------------------------------------
++ *
++ * Timer helpers
++ *
++ *---------------------------------------------------------------------------*/
+
-+static ssize_t pagesize_show(struct hyp_sysfs_attr *attr, char *buffer)
++static inline int falcon_timer_page_addr(uint idx)
+{
-+ int ret;
+
-+ ret = HYPERVISOR_xen_version(XENVER_pagesize, NULL);
-+ if (ret > 0)
-+ ret = sprintf(buffer, "%x\n", ret);
++ EFHW_ASSERT(TIMER_CMD_REG_KER_OFST ==
++ (TIMER_CMD_REG_PAGE4_OFST - 4 * EFHW_8K));
+
-+ return ret;
++ EFHW_ASSERT(idx < FALCON_TIMERS_NUM);
++
++ if (idx < 4)
++ return TIMER_CMD_REG_KER_OFST + (idx * EFHW_8K);
++ else if (idx < 1024)
++ return TIMER_CMD_REG_PAGE4_OFST + ((idx - 4) * EFHW_8K);
++ else
++ return TIMER_CMD_REG_PAGE123K_OFST + ((idx - 1024) * EFHW_8K);
+}
+
-+HYPERVISOR_ATTR_RO(pagesize);
++#define FALCON_TIMER_PAGE_MASK (EFHW_8K-1)
+
-+/* eventually there will be several more features to export */
-+static ssize_t xen_feature_show(int index, char *buffer)
++static inline int falcon_timer_page_offset(uint idx)
+{
-+ int ret = -ENOMEM;
-+ struct xen_feature_info *info;
++ return falcon_timer_page_addr(idx) & FALCON_TIMER_PAGE_MASK;
++}
+
-+ info = kmalloc(sizeof(struct xen_feature_info), GFP_KERNEL);
-+ if (info) {
-+ info->submap_idx = index;
-+ ret = HYPERVISOR_xen_version(XENVER_get_features, info);
-+ if (!ret)
-+ ret = sprintf(buffer, "%d\n", info->submap);
-+ kfree(info);
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * DMA Queue helpers
++ *
++ *---------------------------------------------------------------------------*/
+
-+ return ret;
-+}
++/* iSCSI queue for A1; see bug 5427 for more details. */
++#define FALCON_A1_ISCSI_DMAQ 4
+
-+static ssize_t writable_pt_show(struct hyp_sysfs_attr *attr, char *buffer)
++/*! returns an address within a bar of the TX DMA doorbell */
++static inline uint falcon_tx_dma_page_addr(uint dmaq_idx)
+{
-+ return xen_feature_show(XENFEAT_writable_page_tables, buffer);
-+}
++ uint page;
+
-+HYPERVISOR_ATTR_RO(writable_pt);
++ EFHW_ASSERT((((TX_DESC_UPD_REG_PAGE123K_OFST) & (EFHW_8K - 1)) ==
++ (((TX_DESC_UPD_REG_PAGE4_OFST) & (EFHW_8K - 1)))));
+
-+static struct attribute *xen_properties_attrs[] = {
-+ &capabilities_attr.attr,
-+ &changeset_attr.attr,
-+ &virtual_start_attr.attr,
-+ &pagesize_attr.attr,
-+ &writable_pt_attr.attr,
-+ NULL
-+};
++ EFHW_ASSERT(dmaq_idx < FALCON_DMAQ_NUM);
+
-+static struct attribute_group xen_properties_group = {
-+ .name = "properties",
-+ .attrs = xen_properties_attrs,
-+};
++ if (dmaq_idx < 1024)
++ page = TX_DESC_UPD_REG_PAGE4_OFST + ((dmaq_idx - 4) * EFHW_8K);
++ else
++ page =
++ TX_DESC_UPD_REG_PAGE123K_OFST +
++ ((dmaq_idx - 1024) * EFHW_8K);
+
-+static int __init xen_properties_init(void)
-+{
-+ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
-+ &xen_properties_group);
++ return page;
+}
+
-+static void xen_properties_destroy(void)
++/*! returns an address within a bar of the RX DMA doorbell */
++static inline uint falcon_rx_dma_page_addr(uint dmaq_idx)
+{
-+ sysfs_remove_group(&hypervisor_subsys.kset.kobj,
-+ &xen_properties_group);
-+}
++ uint page;
+
-+static int __init hyper_sysfs_init(void)
-+{
-+ int ret;
++ EFHW_ASSERT((((RX_DESC_UPD_REG_PAGE123K_OFST) & (EFHW_8K - 1)) ==
++ ((RX_DESC_UPD_REG_PAGE4_OFST) & (EFHW_8K - 1))));
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ EFHW_ASSERT(dmaq_idx < FALCON_DMAQ_NUM);
+
-+ ret = xen_sysfs_type_init();
-+ if (ret)
-+ goto out;
-+ ret = xen_sysfs_version_init();
-+ if (ret)
-+ goto version_out;
-+ ret = xen_compilation_init();
-+ if (ret)
-+ goto comp_out;
-+ ret = xen_sysfs_uuid_init();
-+ if (ret)
-+ goto uuid_out;
-+ ret = xen_properties_init();
-+ if (!ret)
-+ goto out;
++ if (dmaq_idx < 1024)
++ page = RX_DESC_UPD_REG_PAGE4_OFST + ((dmaq_idx - 4) * EFHW_8K);
++ else
++ page =
++ RX_DESC_UPD_REG_PAGE123K_OFST +
++ ((dmaq_idx - 1024) * EFHW_8K);
+
-+ xen_sysfs_uuid_destroy();
-+uuid_out:
-+ xen_compilation_destroy();
-+comp_out:
-+ xen_sysfs_version_destroy();
-+version_out:
-+ xen_sysfs_type_destroy();
-+out:
-+ return ret;
++ return page;
+}
+
-+static void __exit hyper_sysfs_exit(void)
-+{
-+ xen_properties_destroy();
-+ xen_compilation_destroy();
-+ xen_sysfs_uuid_destroy();
-+ xen_sysfs_version_destroy();
-+ xen_sysfs_type_destroy();
++/*! "page"=NIC-dependent register set size */
++#define FALCON_DMA_PAGE_MASK (EFHW_8K-1)
+
++/*! returns an address within a bar of the start of the "page"
++ containing the TX DMA doorbell */
++static inline int falcon_tx_dma_page_base(uint dma_idx)
++{
++ return falcon_tx_dma_page_addr(dma_idx) & ~FALCON_DMA_PAGE_MASK;
+}
+
-+module_init(hyper_sysfs_init);
-+module_exit(hyper_sysfs_exit);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/evtchn/evtchn.c linux-2.6.18-xen.hg/drivers/xen/evtchn/evtchn.c
---- linux-2.6.18/drivers/xen/evtchn/evtchn.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/evtchn/evtchn.c 2007-12-23 11:15:33.867925803 +0100
-@@ -0,0 +1,556 @@
-+/******************************************************************************
-+ * evtchn.c
-+ *
-+ * Driver for receiving and demuxing event-channel signals.
-+ *
-+ * Copyright (c) 2004-2005, K A Fraser
-+ * Multi-process extensions Copyright (c) 2004, Steven Smith
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++/*! returns an address within a bar of the start of the "page"
++ containing the RX DMA doorbell */
++static inline int falcon_rx_dma_page_base(uint dma_idx)
++{
++ return falcon_rx_dma_page_addr(dma_idx) & ~FALCON_DMA_PAGE_MASK;
++}
+
-+#include <linux/module.h>
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/slab.h>
-+#include <linux/string.h>
-+#include <linux/errno.h>
-+#include <linux/fs.h>
-+#include <linux/errno.h>
-+#include <linux/miscdevice.h>
-+#include <linux/major.h>
-+#include <linux/proc_fs.h>
-+#include <linux/stat.h>
-+#include <linux/poll.h>
-+#include <linux/irq.h>
-+#include <linux/init.h>
-+#include <linux/gfp.h>
-+#include <linux/mutex.h>
-+#include <linux/cpu.h>
-+#include <xen/evtchn.h>
-+#include <xen/public/evtchn.h>
++/*! returns an offset within a "page" of the TX DMA doorbell */
++static inline int falcon_tx_dma_page_offset(uint dma_idx)
++{
++ return falcon_tx_dma_page_addr(dma_idx) & FALCON_DMA_PAGE_MASK;
++}
+
-+struct per_user_data {
-+ /* Notification ring, accessed via /dev/xen/evtchn. */
-+#define EVTCHN_RING_SIZE (PAGE_SIZE / sizeof(evtchn_port_t))
-+#define EVTCHN_RING_MASK(_i) ((_i)&(EVTCHN_RING_SIZE-1))
-+ evtchn_port_t *ring;
-+ unsigned int ring_cons, ring_prod, ring_overflow;
-+ struct mutex ring_cons_mutex; /* protect against concurrent readers */
++/*! returns an offset within a "page" of the RX DMA doorbell */
++static inline int falcon_rx_dma_page_offset(uint dma_idx)
++{
++ return falcon_rx_dma_page_addr(dma_idx) & FALCON_DMA_PAGE_MASK;
++}
+
-+ /* Processes wait on this queue when ring is empty. */
-+ wait_queue_head_t evtchn_wait;
-+ struct fasync_struct *evtchn_async_queue;
++/*----------------------------------------------------------------------------
++ *
++ * Events
++ *
++ *---------------------------------------------------------------------------*/
+
-+ int bind_cpu;
-+ int nr_event_wrong_delivery;
-+};
++/* Falcon nails down the event queue mappings */
++#define FALCON_EVQ_KERNEL0 (0) /* hardwired for net driver */
++#define FALCON_EVQ_CHAR (4) /* char driver's event queue */
++#define FALCON_EVQ_NONIRQ (5) /* char driver's non interrupting
++ queue. Subsequent queues are
++ available for user apps */
+
-+/* Who's bound to each port? */
-+static struct per_user_data *port_user[NR_EVENT_CHANNELS];
-+static spinlock_t port_user_lock;
++/* reserved by the drivers */
++#define FALCON_EVQ_TBL_RESERVED (8)
+
-+void evtchn_device_upcall(int port)
-+{
-+ struct per_user_data *u;
++/* default DMA-Q sizes */
++#define FALCON_DMA_Q_DEFAULT_TX_SIZE 512
+
-+ spin_lock(&port_user_lock);
++#define FALCON_DMA_Q_DEFAULT_RX_SIZE 512
+
-+ mask_evtchn(port);
-+ clear_evtchn(port);
++#define FALCON_DMA_Q_DEFAULT_MMAP \
++ (FALCON_DMA_Q_DEFAULT_TX_SIZE * (FALCON_DMA_TX_DESC_BYTES * 2))
+
-+ if ((u = port_user[port]) != NULL) {
-+ if ((u->ring_prod - u->ring_cons) < EVTCHN_RING_SIZE) {
-+ u->ring[EVTCHN_RING_MASK(u->ring_prod)] = port;
-+ if (u->ring_cons == u->ring_prod++) {
-+ wake_up_interruptible(&u->evtchn_wait);
-+ kill_fasync(&u->evtchn_async_queue,
-+ SIGIO, POLL_IN);
-+ }
-+ } else {
-+ u->ring_overflow = 1;
-+ }
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * DEBUG - Analyser trigger
++ *
++ *---------------------------------------------------------------------------*/
+
-+ spin_unlock(&port_user_lock);
++static inline void falcon_deadbeef(efhw_ioaddr_t efhw_kva, unsigned what)
++{
++ writel(what, efhw_kva + 0x300);
++ mmiowb();
+}
++#endif /* __CI_DRIVER_EFAB_HARDWARE_FALCON_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/workarounds.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware/workarounds.h 2008-05-19 00:33:29.341837424 +0300
+@@ -0,0 +1,75 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides workaround settings for EtherFabric NICs.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static void evtchn_check_wrong_delivery(struct per_user_data *u)
-+{
-+ evtchn_port_t port;
-+ unsigned int current_cpu = smp_processor_id();
++#ifndef __CI_DRIVER_EFAB_WORKAROUNDS_H__
++#define __CI_DRIVER_EFAB_WORKAROUNDS_H__
+
-+ /* Delivered to correct CPU? All is good. */
-+ if (u->bind_cpu == current_cpu) {
-+ u->nr_event_wrong_delivery = 0;
-+ return;
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * Hardware workarounds which have global scope
++ *
++ *---------------------------------------------------------------------------*/
+
-+ /* Tolerate up to 100 consecutive misdeliveries. */
-+ if (++u->nr_event_wrong_delivery < 100)
-+ return;
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
+
-+ spin_lock_irq(&port_user_lock);
++#if defined(__CI_HARDWARE_CONFIG_FALCON_B0__)
++/*------------------------------- B0 ---------------------------------------*/
+
-+ for (port = 0; port < NR_EVENT_CHANNELS; port++)
-+ if (port_user[port] == u)
-+ rebind_evtchn_to_cpu(port, current_cpu);
++#define BUG2175_WORKAROUND 0 /* TX event batching for dual port operation.
++ This removes the effect (dup TX events)
++ of the fix
++ (TX event per packet + batch events) */
++#define BUG5302_WORKAROUND 0 /* unstick TX DMAQ after out-of-range wr ptr */
++#define BUG5475_WORKAROUND 1 /* 10G SNAP encapsulation broken */
++#define BUG5762_WORKAROUND 0 /* Set all queues to jumbo mode */
++#define BUG5391_WORKAROUND 0 /* Misaligned TX can't span 512-byte boundary */
++#define BUG7916_WORKAROUND 0 /* RX flush gets lost */
+
-+ u->bind_cpu = current_cpu;
-+ u->nr_event_wrong_delivery = 0;
++#else
++/*------------------------------- A0/A1 ------------------------------------*/
+
-+ spin_unlock_irq(&port_user_lock);
-+}
++#define BUG2175_WORKAROUND 1 /* TX event batching for dual port operation.
++ This removes the effect (dup TX events)
++ of the fix
++ (TX event per packet + batch events) */
++#define BUG5302_WORKAROUND 1 /* unstick TX DMAQ after out-of-range wr ptr */
++#define BUG5475_WORKAROUND 1 /* 10G SNAP encapsulation broken */
++#define BUG5762_WORKAROUND 1 /* Set all queues to jumbo mode */
++#define BUG5391_WORKAROUND 1 /* Misaligned TX can't span 512-byte boundary */
++#define BUG7916_WORKAROUND 1 /* RX flush gets lost */
+
-+static ssize_t evtchn_read(struct file *file, char __user *buf,
-+ size_t count, loff_t *ppos)
-+{
-+ int rc;
-+ unsigned int c, p, bytes1 = 0, bytes2 = 0;
-+ struct per_user_data *u = file->private_data;
++#endif /* B0/A01 */
+
-+ /* Whole number of ports. */
-+ count &= ~(sizeof(evtchn_port_t)-1);
++#else
++# error Need hw support.
++#endif
+
-+ if (count == 0)
-+ return 0;
++#endif /* __CI_DRIVER_EFAB_WORKAROUNDS_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/efab/hardware.h 2008-05-19 00:33:29.329836732 +0300
+@@ -0,0 +1,199 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides EtherFabric NIC hardware interface.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (count > PAGE_SIZE)
-+ count = PAGE_SIZE;
++#ifndef __CI_DRIVER_EFAB_HARDWARE_H__
++#define __CI_DRIVER_EFAB_HARDWARE_H__
+
-+ for (;;) {
-+ mutex_lock(&u->ring_cons_mutex);
++#include "ci/driver/efab/hardware/workarounds.h"
++#include <ci/efhw/hardware_sysdep.h>
+
-+ rc = -EFBIG;
-+ if (u->ring_overflow)
-+ goto unlock_out;
+
-+ if ((c = u->ring_cons) != (p = u->ring_prod))
-+ break;
++/*----------------------------------------------------------------------------
++ *
++ * Common EtherFabric definitions
++ *
++ *---------------------------------------------------------------------------*/
+
-+ mutex_unlock(&u->ring_cons_mutex);
++#include <ci/efhw/debug.h>
++#include <ci/efhw/common.h>
++#include <ci/driver/efab/hardware/common.h>
+
-+ if (file->f_flags & O_NONBLOCK)
-+ return -EAGAIN;
++/*----------------------------------------------------------------------------
++ *
++ * EtherFabric varients
++ *
++ *---------------------------------------------------------------------------*/
+
-+ rc = wait_event_interruptible(
-+ u->evtchn_wait, u->ring_cons != u->ring_prod);
-+ if (rc)
-+ return rc;
-+ }
++#include <ci/driver/efab/hardware/falcon.h>
+
-+ /* Byte lengths of two chunks. Chunk split (if any) is at ring wrap. */
-+ if (((c ^ p) & EVTCHN_RING_SIZE) != 0) {
-+ bytes1 = (EVTCHN_RING_SIZE - EVTCHN_RING_MASK(c)) *
-+ sizeof(evtchn_port_t);
-+ bytes2 = EVTCHN_RING_MASK(p) * sizeof(evtchn_port_t);
-+ } else {
-+ bytes1 = (p - c) * sizeof(evtchn_port_t);
-+ bytes2 = 0;
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * EtherFabric Portable Hardware Layer defines
++ *
++ *---------------------------------------------------------------------------*/
+
-+ /* Truncate chunks according to caller's maximum byte count. */
-+ if (bytes1 > count) {
-+ bytes1 = count;
-+ bytes2 = 0;
-+ } else if ((bytes1 + bytes2) > count) {
-+ bytes2 = count - bytes1;
-+ }
++ /*-------------- Initialisation ------------ */
++#define efhw_nic_close_hardware(nic) \
++ ((nic)->efhw_func->close_hardware(nic))
+
-+ rc = -EFAULT;
-+ if (copy_to_user(buf, &u->ring[EVTCHN_RING_MASK(c)], bytes1) ||
-+ ((bytes2 != 0) &&
-+ copy_to_user(&buf[bytes1], &u->ring[0], bytes2)))
-+ goto unlock_out;
-+
-+ evtchn_check_wrong_delivery(u);
++#define efhw_nic_init_hardware(nic, ev_handlers, mac_addr) \
++ ((nic)->efhw_func->init_hardware((nic), (ev_handlers), (mac_addr)))
+
-+ u->ring_cons += (bytes1 + bytes2) / sizeof(evtchn_port_t);
-+ rc = bytes1 + bytes2;
++/*-------------- Interrupt support ------------ */
++/** Handle interrupt. Return 0 if not handled, 1 if handled. */
++#define efhw_nic_interrupt(nic) \
++ ((nic)->efhw_func->interrupt(nic))
+
-+ unlock_out:
-+ mutex_unlock(&u->ring_cons_mutex);
-+ return rc;
-+}
++#define efhw_nic_interrupt_enable(nic, index) \
++ ((nic)->efhw_func->interrupt_enable(nic, index))
+
-+static ssize_t evtchn_write(struct file *file, const char __user *buf,
-+ size_t count, loff_t *ppos)
-+{
-+ int rc, i;
-+ evtchn_port_t *kbuf = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
-+ struct per_user_data *u = file->private_data;
++#define efhw_nic_interrupt_disable(nic, index) \
++ ((nic)->efhw_func->interrupt_disable(nic, index))
+
-+ if (kbuf == NULL)
-+ return -ENOMEM;
++#define efhw_nic_set_interrupt_moderation(nic, index, val) \
++ ((nic)->efhw_func->set_interrupt_moderation(nic, index, val))
+
-+ /* Whole number of ports. */
-+ count &= ~(sizeof(evtchn_port_t)-1);
++/*-------------- Event support ------------ */
+
-+ rc = 0;
-+ if (count == 0)
-+ goto out;
++#define efhw_nic_event_queue_enable(nic, evq, size, q_base, buf_base) \
++ ((nic)->efhw_func->event_queue_enable(nic, evq, size, q_base, \
++ buf_base))
+
-+ if (count > PAGE_SIZE)
-+ count = PAGE_SIZE;
++#define efhw_nic_event_queue_disable(nic, evq, timer_only) \
++ ((nic)->efhw_func->event_queue_disable(nic, evq, timer_only))
+
-+ rc = -EFAULT;
-+ if (copy_from_user(kbuf, buf, count) != 0)
-+ goto out;
++#define efhw_nic_wakeup_request(nic, q_base, index, evq) \
++ ((nic)->efhw_func->wakeup_request(nic, q_base, index, evq))
+
-+ spin_lock_irq(&port_user_lock);
-+ for (i = 0; i < (count/sizeof(evtchn_port_t)); i++)
-+ if ((kbuf[i] < NR_EVENT_CHANNELS) && (port_user[kbuf[i]] == u))
-+ unmask_evtchn(kbuf[i]);
-+ spin_unlock_irq(&port_user_lock);
++#define efhw_nic_sw_event(nic, data, ev) \
++ ((nic)->efhw_func->sw_event(nic, data, ev))
+
-+ rc = count;
++/*-------------- Filter support ------------ */
++#define efhw_nic_ipfilter_set(nic, type, index, dmaq, \
++ saddr, sport, daddr, dport) \
++ ((nic)->efhw_func->ipfilter_set(nic, type, index, dmaq, \
++ saddr, sport, daddr, dport))
+
-+ out:
-+ free_page((unsigned long)kbuf);
-+ return rc;
-+}
++#define efhw_nic_ipfilter_attach(nic, index, dmaq) \
++ ((nic)->efhw_func->ipfilter_attach(nic, index, dmaq))
+
-+static unsigned int next_bind_cpu(cpumask_t map)
-+{
-+ static unsigned int bind_cpu;
-+ bind_cpu = next_cpu(bind_cpu, map);
-+ if (bind_cpu >= NR_CPUS)
-+ bind_cpu = first_cpu(map);
-+ return bind_cpu;
-+}
++#define efhw_nic_ipfilter_detach(nic, index) \
++ ((nic)->efhw_func->ipfilter_detach(nic, index))
+
-+static void evtchn_bind_to_user(struct per_user_data *u, int port)
-+{
-+ spin_lock_irq(&port_user_lock);
++#define efhw_nic_ipfilter_clear(nic, index) \
++ ((nic)->efhw_func->ipfilter_clear(nic, index))
+
-+ BUG_ON(port_user[port] != NULL);
-+ port_user[port] = u;
++/*-------------- DMA support ------------ */
++#define efhw_nic_dmaq_tx_q_init(nic, dmaq, evq, owner, tag, \
++ dmaq_size, index, flags) \
++ ((nic)->efhw_func->dmaq_tx_q_init(nic, dmaq, evq, owner, tag, \
++ dmaq_size, index, flags))
+
-+ if (u->bind_cpu == -1)
-+ u->bind_cpu = next_bind_cpu(cpu_online_map);
++#define efhw_nic_dmaq_rx_q_init(nic, dmaq, evq, owner, tag, \
++ dmaq_size, index, flags) \
++ ((nic)->efhw_func->dmaq_rx_q_init(nic, dmaq, evq, owner, tag, \
++ dmaq_size, index, flags))
+
-+ rebind_evtchn_to_cpu(port, u->bind_cpu);
++#define efhw_nic_dmaq_tx_q_disable(nic, dmaq) \
++ ((nic)->efhw_func->dmaq_tx_q_disable(nic, dmaq))
+
-+ unmask_evtchn(port);
++#define efhw_nic_dmaq_rx_q_disable(nic, dmaq) \
++ ((nic)->efhw_func->dmaq_rx_q_disable(nic, dmaq))
+
-+ spin_unlock_irq(&port_user_lock);
-+}
++#define efhw_nic_flush_tx_dma_channel(nic, dmaq) \
++ ((nic)->efhw_func->flush_tx_dma_channel(nic, dmaq))
+
-+static long evtchn_ioctl(struct file *file,
-+ unsigned int cmd, unsigned long arg)
-+{
-+ int rc;
-+ struct per_user_data *u = file->private_data;
-+ void __user *uarg = (void __user *) arg;
++#define efhw_nic_flush_rx_dma_channel(nic, dmaq) \
++ ((nic)->efhw_func->flush_rx_dma_channel(nic, dmaq))
+
-+ switch (cmd) {
-+ case IOCTL_EVTCHN_BIND_VIRQ: {
-+ struct ioctl_evtchn_bind_virq bind;
-+ struct evtchn_bind_virq bind_virq;
++/*-------------- MAC Low level interface ---- */
++#define efhw_gmac_get_mac_addr(nic) \
++ ((nic)->gmac->get_mac_addr((nic)->gmac))
+
-+ rc = -EFAULT;
-+ if (copy_from_user(&bind, uarg, sizeof(bind)))
-+ break;
++/*-------------- Buffer table -------------- */
++#define efhw_nic_buffer_table_set(nic, addr, bufsz, region, \
++ own_id, buf_id) \
++ ((nic)->efhw_func->buffer_table_set(nic, addr, bufsz, region, \
++ own_id, buf_id))
+
-+ bind_virq.virq = bind.virq;
-+ bind_virq.vcpu = 0;
-+ rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
-+ &bind_virq);
-+ if (rc != 0)
-+ break;
++#define efhw_nic_buffer_table_set_n(nic, buf_id, addr, bufsz, \
++ region, n_pages, own_id) \
++ ((nic)->efhw_func->buffer_table_set_n(nic, buf_id, addr, bufsz, \
++ region, n_pages, own_id))
+
-+ rc = bind_virq.port;
-+ evtchn_bind_to_user(u, rc);
-+ break;
-+ }
++#define efhw_nic_buffer_table_clear(nic, id, num) \
++ ((nic)->efhw_func->buffer_table_clear(nic, id, num))
+
-+ case IOCTL_EVTCHN_BIND_INTERDOMAIN: {
-+ struct ioctl_evtchn_bind_interdomain bind;
-+ struct evtchn_bind_interdomain bind_interdomain;
++#define efhw_nic_buffer_table_commit(nic) \
++ ((nic)->efhw_func->buffer_table_commit(nic))
+
-+ rc = -EFAULT;
-+ if (copy_from_user(&bind, uarg, sizeof(bind)))
-+ break;
++/*----------------------------------------------------------------------------
++ * Hardware specific portability macros for performance critical code.
++ *
++ * Warning: and driver code which is using these defines is not
++ * capable of supporting multiple NIC varients and should be built and
++ * marked appropriately
++ *
++ *---------------------------------------------------------------------------*/
+
-+ bind_interdomain.remote_dom = bind.remote_domain;
-+ bind_interdomain.remote_port = bind.remote_port;
-+ rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
-+ &bind_interdomain);
-+ if (rc != 0)
-+ break;
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
+
-+ rc = bind_interdomain.local_port;
-+ evtchn_bind_to_user(u, rc);
-+ break;
-+ }
++/* --- DMA --- */
++#define EFHW_DMA_ADDRMASK (0xffffffffffffffffULL)
+
-+ case IOCTL_EVTCHN_BIND_UNBOUND_PORT: {
-+ struct ioctl_evtchn_bind_unbound_port bind;
-+ struct evtchn_alloc_unbound alloc_unbound;
++/* --- Buffers --- */
++#define EFHW_BUFFER_ADDR FALCON_BUFFER_4K_ADDR
++#define EFHW_BUFFER_PAGE FALCON_BUFFER_4K_PAGE
++#define EFHW_BUFFER_OFF FALCON_BUFFER_4K_OFF
+
-+ rc = -EFAULT;
-+ if (copy_from_user(&bind, uarg, sizeof(bind)))
-+ break;
++/* --- Filters --- */
++#define EFHW_IP_FILTER_NUM FALCON_FILTER_TBL_NUM
+
-+ alloc_unbound.dom = DOMID_SELF;
-+ alloc_unbound.remote_dom = bind.remote_domain;
-+ rc = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
-+ &alloc_unbound);
-+ if (rc != 0)
-+ break;
++#define EFHW_MAX_PAGE_SIZE FALCON_MAX_PAGE_SIZE
+
-+ rc = alloc_unbound.port;
-+ evtchn_bind_to_user(u, rc);
-+ break;
-+ }
++#else
++# error no hardware definition found
++#endif
+
-+ case IOCTL_EVTCHN_UNBIND: {
-+ struct ioctl_evtchn_unbind unbind;
-+ struct evtchn_close close;
-+ int ret;
++#if PAGE_SIZE <= EFHW_MAX_PAGE_SIZE
++#define EFHW_NIC_PAGE_SIZE PAGE_SIZE
++#else
++#define EFHW_NIC_PAGE_SIZE EFHW_MAX_PAGE_SIZE
++#endif
++#define EFHW_NIC_PAGE_MASK (~(EFHW_NIC_PAGE_SIZE-1))
+
-+ rc = -EFAULT;
-+ if (copy_from_user(&unbind, uarg, sizeof(unbind)))
-+ break;
++#endif /* __CI_DRIVER_EFAB_HARDWARE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/resource/efx_vi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/resource/efx_vi.h 2008-05-19 00:33:29.341837424 +0300
+@@ -0,0 +1,276 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains public EFX VI API to Solarflare resource manager.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ rc = -EINVAL;
-+ if (unbind.port >= NR_EVENT_CHANNELS)
-+ break;
++#ifndef __CI_DRIVER_RESOURCE_EFX_VI_H__
++#define __CI_DRIVER_RESOURCE_EFX_VI_H__
+
-+ spin_lock_irq(&port_user_lock);
-+
-+ rc = -ENOTCONN;
-+ if (port_user[unbind.port] != u) {
-+ spin_unlock_irq(&port_user_lock);
-+ break;
-+ }
++/* Default size of event queue in the efx_vi resource. Copied from
++ * CI_CFG_NETIF_EVENTQ_SIZE */
++#define EFX_VI_EVENTQ_SIZE_DEFAULT 1024
+
-+ port_user[unbind.port] = NULL;
-+ mask_evtchn(unbind.port);
++extern int efx_vi_eventq_size;
+
-+ spin_unlock_irq(&port_user_lock);
++/**************************************************************************
++ * efx_vi_state types, allocation and free
++ **************************************************************************/
+
-+ close.port = unbind.port;
-+ ret = HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
-+ BUG_ON(ret);
++/*! Handle for refering to a efx_vi */
++struct efx_vi_state;
+
-+ rc = 0;
-+ break;
-+ }
++/*!
++ * Allocate an efx_vi, including event queue and pt_endpoint
++ *
++ * \param vih_out Pointer to a handle that is set on success
++ * \param nic_index Index of NIC to apply this resource to
++ * \return Zero on success (and vih_out set), non-zero on failure.
++ */
++extern int
++efx_vi_alloc(struct efx_vi_state **vih_out, int nic_index);
+
-+ case IOCTL_EVTCHN_NOTIFY: {
-+ struct ioctl_evtchn_notify notify;
++/*!
++ * Free a previously allocated efx_vi
++ *
++ * \param vih The handle of the efx_vi to free
++ */
++extern void
++efx_vi_free(struct efx_vi_state *vih);
+
-+ rc = -EFAULT;
-+ if (copy_from_user(¬ify, uarg, sizeof(notify)))
-+ break;
++/*!
++ * Reset a previously allocated efx_vi
++ *
++ * \param vih The handle of the efx_vi to reset
++ */
++extern void
++efx_vi_reset(struct efx_vi_state *vih);
+
-+ if (notify.port >= NR_EVENT_CHANNELS) {
-+ rc = -EINVAL;
-+ } else if (port_user[notify.port] != u) {
-+ rc = -ENOTCONN;
-+ } else {
-+ notify_remote_via_evtchn(notify.port);
-+ rc = 0;
-+ }
-+ break;
-+ }
++/**************************************************************************
++ * efx_vi_eventq types and functions
++ **************************************************************************/
+
-+ case IOCTL_EVTCHN_RESET: {
-+ /* Initialise the ring to empty. Clear errors. */
-+ mutex_lock(&u->ring_cons_mutex);
-+ spin_lock_irq(&port_user_lock);
-+ u->ring_cons = u->ring_prod = u->ring_overflow = 0;
-+ spin_unlock_irq(&port_user_lock);
-+ mutex_unlock(&u->ring_cons_mutex);
-+ rc = 0;
-+ break;
-+ }
++/*!
++ * Register a function to receive callbacks when event queue timeouts
++ * or wakeups occur. Only one function per efx_vi can be registered
++ * at once.
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param callback The function to callback
++ * \param context An argument to pass to the callback function
++ * \return Zero on success, non-zero on failure.
++ */
++extern int
++efx_vi_eventq_register_callback(struct efx_vi_state *vih,
++ void (*callback)(void *context, int is_timeout),
++ void *context);
+
-+ default:
-+ rc = -ENOSYS;
-+ break;
-+ }
++/*!
++ * Remove the current eventq timeout or wakeup callback function
++ *
++ * \param vih The handle to identify the efx_vi
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_eventq_kill_callback(struct efx_vi_state *vih);
+
-+ return rc;
-+}
++/**************************************************************************
++ * efx_vi_dma_map types and functions
++ **************************************************************************/
+
-+static unsigned int evtchn_poll(struct file *file, poll_table *wait)
-+{
-+ unsigned int mask = POLLOUT | POLLWRNORM;
-+ struct per_user_data *u = file->private_data;
++/*!
++ * Handle for refering to a efx_vi
++ */
++struct efx_vi_dma_map_state;
+
-+ poll_wait(file, &u->evtchn_wait, wait);
-+ if (u->ring_cons != u->ring_prod)
-+ mask |= POLLIN | POLLRDNORM;
-+ if (u->ring_overflow)
-+ mask = POLLERR;
-+ return mask;
-+}
++/*!
++ * Map a list of buffer pages so they are registered with the hardware
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param addrs An array of page pointers to map
++ * \param n_addrs Length of the page pointer array. Must be a power of two.
++ * \param dmh_out Set on success to a handle used to refer to this mapping
++ * \return Zero on success, non-zero on failure.
++ */
++extern int
++efx_vi_dma_map_pages(struct efx_vi_state *vih, struct page **pages,
++ int n_pages, struct efx_vi_dma_map_state **dmh_out);
++extern int
++efx_vi_dma_map_addrs(struct efx_vi_state *vih,
++ unsigned long long *dev_bus_addrs, int n_pages,
++ struct efx_vi_dma_map_state **dmh_out);
+
-+static int evtchn_fasync(int fd, struct file *filp, int on)
-+{
-+ struct per_user_data *u = filp->private_data;
-+ return fasync_helper(fd, filp, on, &u->evtchn_async_queue);
-+}
++/*!
++ * Unmap a previously mapped set of pages so they are no longer registered
++ * with the hardware.
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param dmh The handle to identify the dma mapping
++ */
++extern void
++efx_vi_dma_unmap_pages(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
++extern void
++efx_vi_dma_unmap_addrs(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
+
-+static int evtchn_open(struct inode *inode, struct file *filp)
-+{
-+ struct per_user_data *u;
++/*!
++ * Retrieve the buffer address of the mapping
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param dmh The handle to identify the buffer mapping
++ * \return The buffer address on success, or zero on failure
++ */
++extern unsigned
++efx_vi_dma_get_map_addr(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
+
-+ if ((u = kmalloc(sizeof(*u), GFP_KERNEL)) == NULL)
-+ return -ENOMEM;
++/**************************************************************************
++ * efx_vi filter functions
++ **************************************************************************/
+
-+ memset(u, 0, sizeof(*u));
-+ init_waitqueue_head(&u->evtchn_wait);
++#define EFX_VI_STATIC_FILTERS 32
+
-+ u->ring = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
-+ if (u->ring == NULL) {
-+ kfree(u);
-+ return -ENOMEM;
-+ }
++/*! Handle to refer to a filter instance */
++struct filter_resource_t;
+
-+ mutex_init(&u->ring_cons_mutex);
++/*!
++ * Allocate and add a filter
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param protocol The protocol of the new filter: UDP or TCP
++ * \param ip_addr_be32 The local ip address of the filter
++ * \param port_le16 The local port of the filter
++ * \param fh_out Set on success to be a handle to refer to this filter
++ * \return Zero on success, non-zero on failure.
++ */
++extern int
++efx_vi_filter(struct efx_vi_state *vih, int protocol, unsigned ip_addr_be32,
++ int port_le16, struct filter_resource_t **fh_out);
+
-+ filp->private_data = u;
++/*!
++ * Remove a filter and free resources associated with it
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param fh The handle to identify the filter
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_filter_stop(struct efx_vi_state *vih, struct filter_resource_t *fh);
++
++/**************************************************************************
++ * efx_vi hw resources types and functions
++ **************************************************************************/
++
++/*! Constants for the type field in efx_vi_hw_resource */
++#define EFX_VI_HW_RESOURCE_TXDMAQ 0x0 /* PFN of TX DMA Q */
++#define EFX_VI_HW_RESOURCE_RXDMAQ 0x1 /* PFN of RX DMA Q */
++#define EFX_VI_HW_RESOURCE_TXBELL 0x2 /* PFN of TX Doorbell (EF1) */
++#define EFX_VI_HW_RESOURCE_RXBELL 0x3 /* PFN of RX Doorbell (EF1) */
++#define EFX_VI_HW_RESOURCE_EVQTIMER 0x4 /* Address of event q timer */
++
++/* Address of event q pointer (EF1) */
++#define EFX_VI_HW_RESOURCE_EVQPTR 0x5
++/* Address of register pointer (Falcon A) */
++#define EFX_VI_HW_RESOURCE_EVQRPTR 0x6
++/* Offset of register pointer (Falcon B) */
++#define EFX_VI_HW_RESOURCE_EVQRPTR_OFFSET 0x7
++/* Address of mem KVA */
++#define EFX_VI_HW_RESOURCE_EVQMEMKVA 0x8
++/* PFN of doorbell page (Falcon) */
++#define EFX_VI_HW_RESOURCE_BELLPAGE 0x9
++
++/*! How large an array to allocate for the get_() functions - smaller
++ than the total number of constants as some are mutually exclusive */
++#define EFX_VI_HW_RESOURCE_MAXSIZE 0x7
++
++/*! Constants for the mem_type field in efx_vi_hw_resource */
++#define EFX_VI_HW_RESOURCE_IOBUFFER 0 /* Host memory */
++#define EFX_VI_HW_RESOURCE_PERIPHERAL 1 /* Card memory/registers */
++
++/*!
++ * Data structure providing information on a hardware resource mapping
++ */
++struct efx_vi_hw_resource {
++ u8 type; /*!< What this resource represents */
++ u8 mem_type; /*!< What type of memory is it in, eg,
++ * host or iomem */
++ u8 more_to_follow; /*!< Is this part of a multi-region resource */
++ u32 length; /*!< Length of the resource in bytes */
++ unsigned long address; /*!< Address of this resource */
++};
++
++/*!
++ * Metadata concerning the list of hardware resource mappings
++ */
++struct efx_vi_hw_resource_metadata {
++ int version;
++ int evq_order;
++ int evq_offs;
++ int evq_capacity;
++ int instance;
++ unsigned rx_capacity;
++ unsigned tx_capacity;
++ int nic_arch;
++ int nic_revision;
++ char nic_variant;
++};
++
++/*!
++ * Obtain a list of hardware resource mappings, using virtual addresses
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param mdata Pointer to a structure to receive the metadata
++ * \param hw_res_array An array to receive the list of hardware resources
++ * \param length The length of hw_res_array. Updated on success to contain
++ * the number of entries in the supplied array that were used.
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_hw_resource_get_virt(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length);
+
-+ u->bind_cpu = -1;
++/*!
++ * Obtain a list of hardware resource mappings, using physical addresses
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param mdata Pointer to a structure to receive the metadata
++ * \param hw_res_array An array to receive the list of hardware resources
++ * \param length The length of hw_res_array. Updated on success to contain
++ * the number of entries in the supplied array that were used.
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_hw_resource_get_phys(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length);
++
++#endif /* __CI_DRIVER_RESOURCE_EFX_VI_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/driver/resource/linux_efhw_nic.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/driver/resource/linux_efhw_nic.h 2008-05-19 00:33:29.341837424 +0300
+@@ -0,0 +1,76 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains definition of the public type struct linux_efhw_nic.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ return 0;
-+}
++#ifndef __CI_DRIVER_RESOURCE_LINUX_RESOURCE__
++#define __CI_DRIVER_RESOURCE_LINUX_RESOURCE__
+
-+static int evtchn_release(struct inode *inode, struct file *filp)
-+{
-+ int i;
-+ struct per_user_data *u = filp->private_data;
-+ struct evtchn_close close;
++#ifndef __linux__
++# error Silly
++#endif
++#ifndef __KERNEL__
++# error Silly
++#endif
+
-+ spin_lock_irq(&port_user_lock);
++#include <ci/efhw/efhw_types.h>
++#include <linux/interrupt.h>
+
-+ free_page((unsigned long)u->ring);
+
-+ for (i = 0; i < NR_EVENT_CHANNELS; i++) {
-+ int ret;
-+ if (port_user[i] != u)
-+ continue;
++/************************************************************************
++ * Per-nic structure in the resource driver *
++ ************************************************************************/
+
-+ port_user[i] = NULL;
-+ mask_evtchn(i);
++struct linux_efhw_nic {
++ struct efhw_nic nic;
+
-+ close.port = i;
-+ ret = HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
-+ BUG_ON(ret);
-+ }
++ struct pci_dev *pci_dev; /*!< pci descriptor */
++ struct tasklet_struct tasklet; /*!< for interrupt bottom half */
+
-+ spin_unlock_irq(&port_user_lock);
++ /* Physical addresses of the control aperture bar. */
++ unsigned long ctr_ap_pci_addr;
+
-+ kfree(u);
++ /*! Callbacks for driverlink, when needed. */
++ struct efx_dl_callbacks *dl_callbacks;
+
-+ return 0;
-+}
++ /*! Event handlers. */
++ struct efhw_ev_handler *ev_handlers;
+
-+static const struct file_operations evtchn_fops = {
-+ .owner = THIS_MODULE,
-+ .read = evtchn_read,
-+ .write = evtchn_write,
-+ .unlocked_ioctl = evtchn_ioctl,
-+ .poll = evtchn_poll,
-+ .fasync = evtchn_fasync,
-+ .open = evtchn_open,
-+ .release = evtchn_release,
+};
+
-+static struct miscdevice evtchn_miscdev = {
-+ .minor = MISC_DYNAMIC_MINOR,
-+ .name = "evtchn",
-+ .fops = &evtchn_fops,
-+};
++#define linux_efhw_nic(efhw_nic) \
++ container_of(efhw_nic, struct linux_efhw_nic, nic)
+
-+static int __cpuinit evtchn_cpu_notify(struct notifier_block *nfb,
-+ unsigned long action, void *hcpu)
-+{
-+ int hotcpu = (unsigned long)hcpu;
-+ cpumask_t map = cpu_online_map;
-+ int port, newcpu;
-+ struct per_user_data *u;
++#endif /* __CI_DRIVER_RESOURCE_LINUX_RESOURCE__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/checks.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/checks.h 2008-05-19 00:33:29.345837655 +0300
+@@ -0,0 +1,118 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides helpers to turn bit shifts into dword shifts and
++ * check that the bit fields haven't overflown the dword etc.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_CHECK_H__
++#define __CI_EFHW_CHECK_H__
++
++/*----------------------------------------------------------------------------
++ *
++ * Helpers to turn bit shifts into dword shifts and check that the bit fields
++ * haven't overflown the dword etc. Aim is to preserve consistency with the
++ * autogenerated headers - once stable we could hard code.
++ *
++ *---------------------------------------------------------------------------*/
++
++/* mask constructors */
++#define __FALCON_MASK(WIDTH, T) ((((T)1) << (WIDTH)) - 1)
++#define __FALCON_MASK32(WIDTH) __FALCON_MASK((WIDTH), uint32_t)
++#define __FALCON_MASK64(WIDTH) __FALCON_MASK((WIDTH), uint64_t)
++
++#define __FALCON_MASKFIELD32(LBN, WIDTH) \
++ ((uint32_t)(__FALCON_MASK32(WIDTH) << (LBN)))
++
++/* constructors for fields which span the first and second dwords */
++#define __LW(LBN) (32 - LBN)
++#define __LOW(v, LBN, WIDTH) \
++ ((uint32_t)(((v) & __FALCON_MASK64(__LW((LBN)))) << (LBN)))
++#define __HIGH(v, LBN, WIDTH) \
++ ((uint32_t)(((v) >> __LW((LBN))) & \
++ __FALCON_MASK64((WIDTH - __LW((LBN))))))
++/* constructors for fields within the second dword */
++#define __DW2(LBN) ((LBN) - 32)
++
++/* constructors for fields which span the second and third dwords */
++#define __LW2(LBN) (64 - LBN)
++#define __LOW2(v, LBN, WIDTH) \
++ ((uint32_t)(((v) & __FALCON_MASK64(__LW2((LBN)))) << ((LBN) - 32)))
++#define __HIGH2(v, LBN, WIDTH) \
++ ((uint32_t)(((v) >> __LW2((LBN))) & \
++ __FALCON_MASK64((WIDTH - __LW2((LBN))))))
++
++/* constructors for fields within the third dword */
++#define __DW3(LBN) ((LBN) - 64)
++
++/* constructors for fields which span the third and fourth dwords */
++#define __LW3(LBN) (96 - LBN)
++#define __LOW3(v, LBN, WIDTH) \
++ ((uint32_t)(((v) & __FALCON_MASK64(__LW3((LBN)))) << ((LBN) - 64)))
++#define __HIGH3(v, LBN, WIDTH) \
++ ((ci_unit32)(((v) >> __LW3((LBN))) & \
++ __FALCON_MASK64((WIDTH - __LW3((LBN))))))
++
++/* constructors for fields within the fourth dword */
++#define __DW4(LBN) ((LBN) - 96)
++
++/* checks that the autogenerated headers are consistent with our model */
++#define __WIDTHCHCK(a, b) EFHW_ASSERT((a) == (b))
++#define __RANGECHCK(v, WIDTH) \
++ EFHW_ASSERT(((uint64_t)(v) & ~(__FALCON_MASK64((WIDTH)))) == 0)
++
++/* fields within the first dword */
++#define __DWCHCK(LBN, WIDTH) \
++ EFHW_ASSERT(((LBN) >= 0) && (((LBN)+(WIDTH)) <= 32))
++
++/* fields which span the first and second dwords */
++#define __LWCHK(LBN, WIDTH) EFHW_ASSERT(WIDTH >= __LW(LBN))
++
++/* fields within the second dword */
++#define __DW2CHCK(LBN, WIDTH) \
++ EFHW_ASSERT(((LBN) >= 32) && (((LBN)+(WIDTH)) <= 64))
++
++/* fields which span the second and third dwords */
++#define __LW2CHK(LBN, WIDTH) EFHW_ASSERT(WIDTH >= __LW2(LBN))
++
++/* fields within the third dword */
++#define __DW3CHCK(LBN, WIDTH) \
++ EFHW_ASSERT(((LBN) >= 64) && (((LBN)+(WIDTH)) <= 96))
++
++/* fields which span the third and fourth dwords */
++#define __LW3CHK(LBN, WIDTH) EFHW_ASSERT(WIDTH >= __LW3(LBN))
++
++/* fields within the fourth dword */
++#define __DW4CHCK(LBN, WIDTH) \
++ EFHW_ASSERT(((LBN) >= 96) && (((LBN)+(WIDTH)) <= 128))
++
++/* fields in the first qword */
++#define __QWCHCK(LBN, WIDTH) \
++ EFHW_ASSERT(((LBN) >= 0) && (((LBN)+(WIDTH)) <= 64))
++
++#endif /* __CI_EFHW_CHECK_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/common.h 2008-05-19 00:33:29.345837655 +0300
+@@ -0,0 +1,102 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides API of the efhw library which may be used both from
++ * the kernel and from the user-space code.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ switch (action) {
-+ case CPU_DOWN_PREPARE:
-+ cpu_clear(hotcpu, map);
-+ spin_lock_irq(&port_user_lock);
-+ for (port = 0; port < NR_EVENT_CHANNELS; port++) {
-+ if ((u = port_user[port]) != NULL &&
-+ u->bind_cpu == hotcpu &&
-+ (newcpu = next_bind_cpu(map)) < NR_CPUS) {
-+ rebind_evtchn_to_cpu(port, newcpu);
-+ u->bind_cpu = newcpu;
-+ }
-+ }
-+ spin_unlock_irq(&port_user_lock);
-+ break;
-+ default:
-+ return NOTIFY_DONE;
-+ }
-+ return NOTIFY_OK;
-+}
++#ifndef __CI_EFHW_COMMON_H__
++#define __CI_EFHW_COMMON_H__
+
-+static struct notifier_block __cpuinitdata evtchn_cpu_nfb = {
-+ .notifier_call = evtchn_cpu_notify
-+};
++#include <ci/efhw/common_sysdep.h>
+
-+static int __init evtchn_init(void)
-+{
-+ int err;
++enum efhw_arch {
++ EFHW_ARCH_FALCON,
++ EFHW_ARCH_SIENA,
++};
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++typedef uint32_t efhw_buffer_addr_t;
++#define EFHW_BUFFER_ADDR_FMT "[ba:%"PRIx32"]"
+
-+ spin_lock_init(&port_user_lock);
-+ memset(port_user, 0, sizeof(port_user));
++/*! Comment? */
++typedef union {
++ uint64_t u64;
++ struct {
++ uint32_t a;
++ uint32_t b;
++ } opaque;
++ struct {
++ uint32_t code;
++ uint32_t status;
++ } ev1002;
++} efhw_event_t;
++
++/* Flags for TX/RX queues */
++#define EFHW_VI_JUMBO_EN 0x01 /*! scatter RX over multiple desc */
++#define EFHW_VI_ISCSI_RX_HDIG_EN 0x02 /*! iscsi rx header digest */
++#define EFHW_VI_ISCSI_TX_HDIG_EN 0x04 /*! iscsi tx header digest */
++#define EFHW_VI_ISCSI_RX_DDIG_EN 0x08 /*! iscsi rx data digest */
++#define EFHW_VI_ISCSI_TX_DDIG_EN 0x10 /*! iscsi tx data digest */
++#define EFHW_VI_TX_PHYS_ADDR_EN 0x20 /*! TX physical address mode */
++#define EFHW_VI_RX_PHYS_ADDR_EN 0x40 /*! RX physical address mode */
++#define EFHW_VI_RM_WITH_INTERRUPT 0x80 /*! VI with an interrupt */
++#define EFHW_VI_TX_IP_CSUM_DIS 0x100 /*! enable ip checksum generation */
++#define EFHW_VI_TX_TCPUDP_CSUM_DIS 0x200 /*! enable tcp/udp checksum
++ generation */
++#define EFHW_VI_TX_TCPUDP_ONLY 0x400 /*! drop non-tcp/udp packets */
++
++/* Types of hardware filter */
++/* Each of these values implicitly selects scatter filters on B0 - or in
++ EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK if a non-scatter filter is required */
++#define EFHW_IP_FILTER_TYPE_UDP_WILDCARD (0) /* dest host only */
++#define EFHW_IP_FILTER_TYPE_UDP_FULL (1) /* dest host and port */
++#define EFHW_IP_FILTER_TYPE_TCP_WILDCARD (2) /* dest based filter */
++#define EFHW_IP_FILTER_TYPE_TCP_FULL (3) /* src filter */
++/* Same again, but with RSS (for B0 only) */
++#define EFHW_IP_FILTER_TYPE_UDP_WILDCARD_RSS_B0 (4)
++#define EFHW_IP_FILTER_TYPE_UDP_FULL_RSS_B0 (5)
++#define EFHW_IP_FILTER_TYPE_TCP_WILDCARD_RSS_B0 (6)
++#define EFHW_IP_FILTER_TYPE_TCP_FULL_RSS_B0 (7)
++
++#define EFHW_IP_FILTER_TYPE_FULL_MASK (0x1) /* Mask for full / wildcard */
++#define EFHW_IP_FILTER_TYPE_TCP_MASK (0x2) /* Mask for TCP type */
++#define EFHW_IP_FILTER_TYPE_RSS_B0_MASK (0x4) /* Mask for B0 RSS enable */
++#define EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK (0x8) /* Mask for B0 SCATTER dsbl */
++
++#define EFHW_IP_FILTER_TYPE_MASK (0xffff) /* Mask of types above */
++
++#define EFHW_IP_FILTER_BROADCAST (0x10000) /* driverlink filter
++ support */
++
++#endif /* __CI_EFHW_COMMON_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/common_sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/common_sysdep.h 2008-05-19 00:33:29.365838808 +0300
+@@ -0,0 +1,71 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for
++ * userland-to-kernel interfaces.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* Create '/dev/misc/evtchn'. */
-+ err = misc_register(&evtchn_miscdev);
-+ if (err != 0) {
-+ printk(KERN_ALERT "Could not register /dev/misc/evtchn\n");
-+ return err;
-+ }
++#ifndef __CI_EFHW_COMMON_LINUX_H__
++#define __CI_EFHW_COMMON_LINUX_H__
+
-+ register_cpu_notifier(&evtchn_cpu_nfb);
++#include <linux/types.h>
++#include <linux/version.h>
+
-+ printk("Event-channel device installed.\n");
++/* Dirty hack, but Linux kernel does not provide DMA_ADDR_T_FMT */
++#if BITS_PER_LONG == 64 || defined(CONFIG_HIGHMEM64G)
++#define DMA_ADDR_T_FMT "%llx"
++#else
++#define DMA_ADDR_T_FMT "%x"
++#endif
+
-+ return 0;
-+}
++/* Linux kernel also does not provide PRIx32... Sigh. */
++#define PRIx32 "x"
+
-+static void __exit evtchn_cleanup(void)
-+{
-+ misc_deregister(&evtchn_miscdev);
-+ unregister_cpu_notifier(&evtchn_cpu_nfb);
-+}
++#ifdef __ia64__
++# define PRIx64 "lx"
++#else
++# define PRIx64 "llx"
++#endif
+
-+module_init(evtchn_init);
-+module_exit(evtchn_cleanup);
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
++enum {
++ false = 0,
++ true = 1
++};
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/evtchn/Makefile linux-2.6.18-xen.hg/drivers/xen/evtchn/Makefile
---- linux-2.6.18/drivers/xen/evtchn/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/evtchn/Makefile 2007-12-23 11:15:33.867925803 +0100
-@@ -0,0 +1,2 @@
++typedef _Bool bool;
++#endif /* LINUX_VERSION_CODE < 2.6.19 */
+
-+obj-y := evtchn.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/fbfront/Makefile linux-2.6.18-xen.hg/drivers/xen/fbfront/Makefile
---- linux-2.6.18/drivers/xen/fbfront/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/fbfront/Makefile 2007-12-23 11:15:33.871259312 +0100
-@@ -0,0 +1,2 @@
-+obj-$(CONFIG_XEN_FRAMEBUFFER) := xenfb.o
-+obj-$(CONFIG_XEN_KEYBOARD) += xenkbd.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/fbfront/xenfb.c linux-2.6.18-xen.hg/drivers/xen/fbfront/xenfb.c
---- linux-2.6.18/drivers/xen/fbfront/xenfb.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/fbfront/xenfb.c 2007-12-23 11:15:33.871259312 +0100
-@@ -0,0 +1,758 @@
-+/*
-+ * linux/drivers/video/xenfb.c -- Xen para-virtual frame buffer device
++#endif /* __CI_EFHW_COMMON_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/debug.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/debug.h 2008-05-19 00:33:29.365838808 +0300
+@@ -0,0 +1,84 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * Copyright (C) 2005-2006 Anthony Liguori <aliguori@us.ibm.com>
-+ * Copyright (C) 2006 Red Hat, Inc., Markus Armbruster <armbru@redhat.com>
++ * This file provides debug-related API for efhw library using Linux kernel
++ * primitives.
+ *
-+ * Based on linux/drivers/video/q40fb.c
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * This file is subject to the terms and conditions of the GNU General Public
-+ * License. See the file COPYING in the main directory of this archive for
-+ * more details.
-+ */
-+
-+/*
-+ * TODO:
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
+ *
-+ * Switch to grant tables when they become capable of dealing with the
-+ * frame buffer.
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/kernel.h>
-+#include <linux/errno.h>
-+#include <linux/fb.h>
-+#include <linux/module.h>
-+#include <linux/vmalloc.h>
-+#include <linux/mm.h>
-+#include <linux/mutex.h>
-+#include <asm/hypervisor.h>
-+#include <xen/evtchn.h>
-+#include <xen/interface/io/fbif.h>
-+#include <xen/interface/io/protocols.h>
-+#include <xen/xenbus.h>
-+#include <linux/kthread.h>
++#ifndef __CI_EFHW_DEBUG_LINUX_H__
++#define __CI_EFHW_DEBUG_LINUX_H__
+
-+struct xenfb_mapping
-+{
-+ struct list_head link;
-+ struct vm_area_struct *vma;
-+ atomic_t map_refs;
-+ int faults;
-+ struct xenfb_info *info;
-+};
++#define EFHW_PRINTK_PREFIX "[sfc efhw] "
+
-+struct xenfb_info
-+{
-+ struct task_struct *kthread;
-+ wait_queue_head_t wq;
++#define EFHW_PRINTK(level, fmt, ...) \
++ printk(level EFHW_PRINTK_PREFIX fmt "\n", __VA_ARGS__)
+
-+ unsigned char *fb;
-+ struct fb_info *fb_info;
-+ struct timer_list refresh;
-+ int dirty;
-+ int x1, y1, x2, y2; /* dirty rectangle,
-+ protected by dirty_lock */
-+ spinlock_t dirty_lock;
-+ struct mutex mm_lock;
-+ int nr_pages;
-+ struct page **pages;
-+ struct list_head mappings; /* protected by mm_lock */
++/* Following macros should be used with non-zero format parameters
++ * due to __VA_ARGS__ limitations. Use "%s" with __FUNCTION__ if you can't
++ * find better parameters. */
++#define EFHW_ERR(fmt, ...) EFHW_PRINTK(KERN_ERR, fmt, __VA_ARGS__)
++#define EFHW_WARN(fmt, ...) EFHW_PRINTK(KERN_WARNING, fmt, __VA_ARGS__)
++#define EFHW_NOTICE(fmt, ...) EFHW_PRINTK(KERN_NOTICE, fmt, __VA_ARGS__)
++#if 0 && !defined(NDEBUG)
++#define EFHW_TRACE(fmt, ...) EFHW_PRINTK(KERN_DEBUG, fmt, __VA_ARGS__)
++#else
++#define EFHW_TRACE(fmt, ...)
++#endif
+
-+ int irq;
-+ struct xenfb_page *page;
-+ unsigned long *mfns;
-+ int update_wanted; /* XENFB_TYPE_UPDATE wanted */
++#ifndef NDEBUG
++#define EFHW_ASSERT(cond) BUG_ON((cond) == 0)
++#define EFHW_DO_DEBUG(expr) expr
++#else
++#define EFHW_ASSERT(cond)
++#define EFHW_DO_DEBUG(expr)
++#endif
+
-+ struct xenbus_device *xbdev;
-+};
++#define EFHW_TEST(expr) \
++ do { \
++ if (unlikely(!(expr))) \
++ BUG(); \
++ } while (0)
+
-+/*
-+ * How the locks work together
++/* Build time asserts. We paste the line number into the type name
++ * so that the macro can be used more than once per file even if the
++ * compiler objects to multiple identical typedefs. Collisions
++ * between use in different header files is still possible. */
++#ifndef EFHW_BUILD_ASSERT
++#define __EFHW_BUILD_ASSERT_NAME(_x) __EFHW_BUILD_ASSERT_ILOATHECPP(_x)
++#define __EFHW_BUILD_ASSERT_ILOATHECPP(_x) __EFHW_BUILD_ASSERT__ ##_x
++#define EFHW_BUILD_ASSERT(e) \
++ typedef char __EFHW_BUILD_ASSERT_NAME(__LINE__)[(e) ? 1 : -1]
++#endif
++
++#endif /* __CI_EFHW_DEBUG_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/efhw_config.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/efhw_config.h 2008-05-19 00:33:29.365838808 +0300
+@@ -0,0 +1,43 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * There are two locks: spinlock dirty_lock protecting the dirty
-+ * rectangle, and mutex mm_lock protecting mappings.
++ * This file provides some limits used in both kernel and userland code.
+ *
-+ * The problem is that dirty rectangle and mappings aren't
-+ * independent: the dirty rectangle must cover all faulted pages in
-+ * mappings. We need to prove that our locking maintains this
-+ * invariant.
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * There are several kinds of critical regions:
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
+ *
-+ * 1. Holding only dirty_lock: xenfb_refresh(). May run in
-+ * interrupts. Extends the dirty rectangle. Trivially preserves
-+ * invariant.
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
+ *
-+ * 2. Holding only mm_lock: xenfb_mmap() and xenfb_vm_close(). Touch
-+ * only mappings. The former creates unfaulted pages. Preserves
-+ * invariant. The latter removes pages. Preserves invariant.
+ *
-+ * 3. Holding both locks: xenfb_vm_nopage(). Extends the dirty
-+ * rectangle and updates mappings consistently. Preserves
-+ * invariant.
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
-+ * 4. The ugliest one: xenfb_update_screen(). Clear the dirty
-+ * rectangle and update mappings consistently.
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
+ *
-+ * We can't simply hold both locks, because zap_page_range() cannot
-+ * be called with a spinlock held.
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_EFAB_CONFIG_H__
++#define __CI_EFHW_EFAB_CONFIG_H__
++
++#define EFHW_MAX_NR_DEVS 5 /* max number of efhw devices supported */
++
++#endif /* __CI_EFHW_EFAB_CONFIG_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/efhw_types.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/efhw_types.h 2008-05-19 00:33:29.365838808 +0300
+@@ -0,0 +1,342 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * Therefore, we first clear the dirty rectangle with both locks
-+ * held. Then we unlock dirty_lock and update the mappings.
-+ * Critical regions that hold only dirty_lock may interfere with
-+ * that. This can only be region 1: xenfb_refresh(). But that
-+ * just extends the dirty rectangle, which can't harm the
-+ * invariant.
++ * This file provides struct efhw_nic and some related types.
+ *
-+ * But FIXME: the invariant is too weak. It misses that the fault
-+ * record in mappings must be consistent with the mapping of pages in
-+ * the associated address space! do_no_page() updates the PTE after
-+ * xenfb_vm_nopage() returns, i.e. outside the critical region. This
-+ * allows the following race:
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * X writes to some address in the Xen frame buffer
-+ * Fault - call do_no_page()
-+ * call xenfb_vm_nopage()
-+ * grab mm_lock
-+ * map->faults++;
-+ * release mm_lock
-+ * return back to do_no_page()
-+ * (preempted, or SMP)
-+ * Xen worker thread runs.
-+ * grab mm_lock
-+ * look at mappings
-+ * find this mapping, zaps its pages (but page not in pte yet)
-+ * clear map->faults
-+ * releases mm_lock
-+ * (back to X process)
-+ * put page in X's pte
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
+ *
-+ * Oh well, we wont be updating the writes to this page anytime soon.
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+static int xenfb_fps = 20;
-+static unsigned long xenfb_mem_len = XENFB_WIDTH * XENFB_HEIGHT * XENFB_DEPTH / 8;
++#ifndef __CI_EFHW_EFAB_TYPES_H__
++#define __CI_EFHW_EFAB_TYPES_H__
+
-+static int xenfb_remove(struct xenbus_device *);
-+static void xenfb_init_shared_page(struct xenfb_info *);
-+static int xenfb_connect_backend(struct xenbus_device *, struct xenfb_info *);
-+static void xenfb_disconnect_backend(struct xenfb_info *);
++#include <ci/efhw/efhw_config.h>
++#include <ci/efhw/hardware_sysdep.h>
++#include <ci/efhw/iopage_types.h>
++#include <ci/efhw/sysdep.h>
+
-+static void xenfb_do_update(struct xenfb_info *info,
-+ int x, int y, int w, int h)
-+{
-+ union xenfb_out_event event;
-+ __u32 prod;
++/*--------------------------------------------------------------------
++ *
++ * hardware limits used in the types
++ *
++ *--------------------------------------------------------------------*/
+
-+ event.type = XENFB_TYPE_UPDATE;
-+ event.update.x = x;
-+ event.update.y = y;
-+ event.update.width = w;
-+ event.update.height = h;
++#define EFHW_KEVENTQ_MAX 8
+
-+ prod = info->page->out_prod;
-+ /* caller ensures !xenfb_queue_full() */
-+ mb(); /* ensure ring space available */
-+ XENFB_OUT_RING_REF(info->page, prod) = event;
-+ wmb(); /* ensure ring contents visible */
-+ info->page->out_prod = prod + 1;
++/*--------------------------------------------------------------------
++ *
++ * forward type declarations
++ *
++ *--------------------------------------------------------------------*/
+
-+ notify_remote_via_irq(info->irq);
-+}
++struct efhw_nic;
+
-+static int xenfb_queue_full(struct xenfb_info *info)
-+{
-+ __u32 cons, prod;
++/*--------------------------------------------------------------------
++ *
++ * Managed interface
++ *
++ *--------------------------------------------------------------------*/
+
-+ prod = info->page->out_prod;
-+ cons = info->page->out_cons;
-+ return prod - cons == XENFB_OUT_RING_LEN;
-+}
++struct efhw_buffer_table_allocation{
++ unsigned base;
++ unsigned order;
++};
+
-+static void xenfb_update_screen(struct xenfb_info *info)
-+{
-+ unsigned long flags;
-+ int y1, y2, x1, x2;
-+ struct xenfb_mapping *map;
++struct eventq_resource_hardware {
++ /*!iobuffer allocated for eventq - can be larger than eventq */
++ efhw_iopages_t iobuff;
++ unsigned iobuff_off;
++ struct efhw_buffer_table_allocation buf_tbl_alloc;
++ int capacity; /*!< capacity of event queue */
++};
+
-+ if (!info->update_wanted)
-+ return;
-+ if (xenfb_queue_full(info))
-+ return;
++/*--------------------------------------------------------------------
++ *
++ * event queues and event driven callbacks
++ *
++ *--------------------------------------------------------------------*/
+
-+ mutex_lock(&info->mm_lock);
++struct efhw_keventq {
++ volatile int lock;
++ caddr_t evq_base;
++ int32_t evq_ptr;
++ uint32_t evq_mask;
++ unsigned instance;
++ struct eventq_resource_hardware hw;
++ struct efhw_ev_handler *ev_handlers;
++};
+
-+ spin_lock_irqsave(&info->dirty_lock, flags);
-+ y1 = info->y1;
-+ y2 = info->y2;
-+ x1 = info->x1;
-+ x2 = info->x2;
-+ info->x1 = info->y1 = INT_MAX;
-+ info->x2 = info->y2 = 0;
-+ spin_unlock_irqrestore(&info->dirty_lock, flags);
++/**********************************************************************
++ * Portable HW interface. ***************************************
++ **********************************************************************/
+
-+ list_for_each_entry(map, &info->mappings, link) {
-+ if (!map->faults)
-+ continue;
-+ zap_page_range(map->vma, map->vma->vm_start,
-+ map->vma->vm_end - map->vma->vm_start, NULL);
-+ map->faults = 0;
-+ }
++/*--------------------------------------------------------------------
++ *
++ * EtherFabric Functional units - configuration and control
++ *
++ *--------------------------------------------------------------------*/
+
-+ mutex_unlock(&info->mm_lock);
++struct efhw_func_ops {
+
-+ if (x2 < x1 || y2 < y1) {
-+ printk("xenfb_update_screen bogus rect %d %d %d %d\n",
-+ x1, x2, y1, y2);
-+ WARN_ON(1);
-+ }
-+ xenfb_do_update(info, x1, y1, x2 - x1, y2 - y1);
-+}
++ /*-------------- Initialisation ------------ */
+
-+static int xenfb_thread(void *data)
-+{
-+ struct xenfb_info *info = data;
++ /*! close down all hardware functional units - leaves NIC in a safe
++ state for driver unload */
++ void (*close_hardware) (struct efhw_nic *nic);
+
-+ while (!kthread_should_stop()) {
-+ if (info->dirty) {
-+ info->dirty = 0;
-+ xenfb_update_screen(info);
-+ }
-+ wait_event_interruptible(info->wq,
-+ kthread_should_stop() || info->dirty);
-+ try_to_freeze();
-+ }
-+ return 0;
-+}
++ /*! initialise all hardware functional units */
++ int (*init_hardware) (struct efhw_nic *nic,
++ struct efhw_ev_handler *,
++ const uint8_t *mac_addr);
+
-+static int xenfb_setcolreg(unsigned regno, unsigned red, unsigned green,
-+ unsigned blue, unsigned transp,
-+ struct fb_info *info)
-+{
-+ u32 v;
++ /*-------------- Interrupt support ------------ */
+
-+ if (regno > info->cmap.len)
-+ return 1;
++ /*! Main interrupt routine
++ ** This function returns,
++ ** - zero, if the IRQ was not generated by EF1
++ ** - non-zero, if EF1 was the source of the IRQ
++ **
++ **
++ ** opaque is an OS provided pointer for use by the OS callbacks
++ ** e.g in Windows used to indicate DPC scheduled
++ */
++ int (*interrupt) (struct efhw_nic *nic);
+
-+ red >>= (16 - info->var.red.length);
-+ green >>= (16 - info->var.green.length);
-+ blue >>= (16 - info->var.blue.length);
++ /*! Enable given interrupt mask for the given IRQ unit */
++ void (*interrupt_enable) (struct efhw_nic *nic, uint idx);
+
-+ v = (red << info->var.red.offset) |
-+ (green << info->var.green.offset) |
-+ (blue << info->var.blue.offset);
++ /*! Disable given interrupt mask for the given IRQ unit */
++ void (*interrupt_disable) (struct efhw_nic *nic, uint idx);
+
-+ /* FIXME is this sane? check against xxxfb_setcolreg()! */
-+ switch (info->var.bits_per_pixel) {
-+ case 16:
-+ case 24:
-+ case 32:
-+ ((u32 *)info->pseudo_palette)[regno] = v;
-+ break;
-+ }
-+
-+ return 0;
-+}
++ /*! Set interrupt moderation strategy for the given IRQ unit
++ ** val is in usec
++ */
++ void (*set_interrupt_moderation)(struct efhw_nic *nic,
++ uint idx, uint val);
+
-+static void xenfb_timer(unsigned long data)
-+{
-+ struct xenfb_info *info = (struct xenfb_info *)data;
-+ wake_up(&info->wq);
-+}
++ /*-------------- Event support ------------ */
+
-+static void __xenfb_refresh(struct xenfb_info *info,
-+ int x1, int y1, int w, int h)
-+{
-+ int y2, x2;
++ /*! Enable the given event queue
++ depending on the underlying implementation (EF1 or Falcon) then
++ either a q_base_addr in host memory, or a buffer base id should
++ be proivded
++ */
++ void (*event_queue_enable) (struct efhw_nic *nic,
++ uint evq, /* evnt queue index */
++ uint evq_size, /* units of #entries */
++ dma_addr_t q_base_addr, uint buf_base_id);
+
-+ y2 = y1 + h;
-+ x2 = x1 + w;
++ /*! Disable the given event queue (and any associated timer) */
++ void (*event_queue_disable) (struct efhw_nic *nic, uint evq,
++ int timer_only);
+
-+ if (info->y1 > y1)
-+ info->y1 = y1;
-+ if (info->y2 < y2)
-+ info->y2 = y2;
-+ if (info->x1 > x1)
-+ info->x1 = x1;
-+ if (info->x2 < x2)
-+ info->x2 = x2;
-+ info->dirty = 1;
++ /*! request wakeup from the NIC on a given event Q */
++ void (*wakeup_request) (struct efhw_nic *nic, dma_addr_t q_base_addr,
++ int next_i, int evq);
+
-+ if (timer_pending(&info->refresh))
-+ return;
++ /*! Push a SW event on a given eventQ */
++ void (*sw_event) (struct efhw_nic *nic, int data, int evq);
+
-+ mod_timer(&info->refresh, jiffies + HZ/xenfb_fps);
-+}
++ /*-------------- Filter support ------------ */
+
-+static void xenfb_refresh(struct xenfb_info *info,
-+ int x1, int y1, int w, int h)
-+{
-+ unsigned long flags;
++ /*! Setup a given filter - The software can request a filter_i,
++ * but some EtherFabric implementations will override with
++ * a more suitable index
++ */
++ int (*ipfilter_set) (struct efhw_nic *nic, int type,
++ int *filter_i, int dmaq,
++ unsigned saddr_be32, unsigned sport_be16,
++ unsigned daddr_be32, unsigned dport_be16);
+
-+ spin_lock_irqsave(&info->dirty_lock, flags);
-+ __xenfb_refresh(info, x1, y1, w, h);
-+ spin_unlock_irqrestore(&info->dirty_lock, flags);
-+}
++ /*! Attach a given filter to a DMAQ */
++ void (*ipfilter_attach) (struct efhw_nic *nic, int filter_idx,
++ int dmaq_idx);
+
-+static void xenfb_fillrect(struct fb_info *p, const struct fb_fillrect *rect)
-+{
-+ struct xenfb_info *info = p->par;
++ /*! Detach a filter from its DMAQ */
++ void (*ipfilter_detach) (struct efhw_nic *nic, int filter_idx);
+
-+ cfb_fillrect(p, rect);
-+ xenfb_refresh(info, rect->dx, rect->dy, rect->width, rect->height);
-+}
++ /*! Clear down a given filter */
++ void (*ipfilter_clear) (struct efhw_nic *nic, int filter_idx);
+
-+static void xenfb_imageblit(struct fb_info *p, const struct fb_image *image)
-+{
-+ struct xenfb_info *info = p->par;
++ /*-------------- DMA support ------------ */
+
-+ cfb_imageblit(p, image);
-+ xenfb_refresh(info, image->dx, image->dy, image->width, image->height);
-+}
++ /*! Initialise NIC state for a given TX DMAQ */
++ void (*dmaq_tx_q_init) (struct efhw_nic *nic,
++ uint dmaq, uint evq, uint owner, uint tag,
++ uint dmaq_size, uint buf_idx, uint flags);
+
-+static void xenfb_copyarea(struct fb_info *p, const struct fb_copyarea *area)
-+{
-+ struct xenfb_info *info = p->par;
++ /*! Initialise NIC state for a given RX DMAQ */
++ void (*dmaq_rx_q_init) (struct efhw_nic *nic,
++ uint dmaq, uint evq, uint owner, uint tag,
++ uint dmaq_size, uint buf_idx, uint flags);
+
-+ cfb_copyarea(p, area);
-+ xenfb_refresh(info, area->dx, area->dy, area->width, area->height);
-+}
++ /*! Disable a given TX DMAQ */
++ void (*dmaq_tx_q_disable) (struct efhw_nic *nic, uint dmaq);
+
-+static void xenfb_vm_open(struct vm_area_struct *vma)
-+{
-+ struct xenfb_mapping *map = vma->vm_private_data;
-+ atomic_inc(&map->map_refs);
-+}
++ /*! Disable a given RX DMAQ */
++ void (*dmaq_rx_q_disable) (struct efhw_nic *nic, uint dmaq);
+
-+static void xenfb_vm_close(struct vm_area_struct *vma)
-+{
-+ struct xenfb_mapping *map = vma->vm_private_data;
-+ struct xenfb_info *info = map->info;
++ /*! Flush a given TX DMA channel */
++ int (*flush_tx_dma_channel) (struct efhw_nic *nic, uint dmaq);
+
-+ mutex_lock(&info->mm_lock);
-+ if (atomic_dec_and_test(&map->map_refs)) {
-+ list_del(&map->link);
-+ kfree(map);
-+ }
-+ mutex_unlock(&info->mm_lock);
-+}
++ /*! Flush a given RX DMA channel */
++ int (*flush_rx_dma_channel) (struct efhw_nic *nic, uint dmaq);
+
-+static struct page *xenfb_vm_nopage(struct vm_area_struct *vma,
-+ unsigned long vaddr, int *type)
-+{
-+ struct xenfb_mapping *map = vma->vm_private_data;
-+ struct xenfb_info *info = map->info;
-+ int pgnr = (vaddr - vma->vm_start) >> PAGE_SHIFT;
-+ unsigned long flags;
-+ struct page *page;
-+ int y1, y2;
++ /*-------------- Buffer table Support ------------ */
+
-+ if (pgnr >= info->nr_pages)
-+ return NOPAGE_SIGBUS;
++ /*! Initialise a buffer table page */
++ void (*buffer_table_set) (struct efhw_nic *nic,
++ dma_addr_t dma_addr,
++ uint bufsz, uint region,
++ int own_id, int buffer_id);
+
-+ mutex_lock(&info->mm_lock);
-+ spin_lock_irqsave(&info->dirty_lock, flags);
-+ page = info->pages[pgnr];
-+ get_page(page);
-+ map->faults++;
++ /*! Initialise a block of buffer table pages */
++ void (*buffer_table_set_n) (struct efhw_nic *nic, int buffer_id,
++ dma_addr_t dma_addr,
++ uint bufsz, uint region,
++ int n_pages, int own_id);
+
-+ y1 = pgnr * PAGE_SIZE / info->fb_info->fix.line_length;
-+ y2 = (pgnr * PAGE_SIZE + PAGE_SIZE - 1) / info->fb_info->fix.line_length;
-+ if (y2 > info->fb_info->var.yres)
-+ y2 = info->fb_info->var.yres;
-+ __xenfb_refresh(info, 0, y1, info->fb_info->var.xres, y2 - y1);
-+ spin_unlock_irqrestore(&info->dirty_lock, flags);
-+ mutex_unlock(&info->mm_lock);
++ /*! Clear a block of buffer table pages */
++ void (*buffer_table_clear) (struct efhw_nic *nic, int buffer_id,
++ int num);
+
-+ if (type)
-+ *type = VM_FAULT_MINOR;
++ /*! Commit a buffer table update */
++ void (*buffer_table_commit) (struct efhw_nic *nic);
+
-+ return page;
-+}
++};
+
-+static struct vm_operations_struct xenfb_vm_ops = {
-+ .open = xenfb_vm_open,
-+ .close = xenfb_vm_close,
-+ .nopage = xenfb_vm_nopage,
++
++/*----------------------------------------------------------------------------
++ *
++ * NIC type
++ *
++ *---------------------------------------------------------------------------*/
++
++struct efhw_device_type {
++ int arch; /* enum efhw_arch */
++ char variant; /* 'A', 'B', ... */
++ int revision; /* 0, 1, ... */
+};
+
-+static int xenfb_mmap(struct fb_info *fb_info, struct vm_area_struct *vma)
-+{
-+ struct xenfb_info *info = fb_info->par;
-+ struct xenfb_mapping *map;
-+ int map_pages;
+
-+ if (!(vma->vm_flags & VM_WRITE))
-+ return -EINVAL;
-+ if (!(vma->vm_flags & VM_SHARED))
-+ return -EINVAL;
-+ if (vma->vm_pgoff != 0)
-+ return -EINVAL;
++/*----------------------------------------------------------------------------
++ *
++ * EtherFabric NIC instance - nic.c for HW independent functions
++ *
++ *---------------------------------------------------------------------------*/
+
-+ map_pages = (vma->vm_end - vma->vm_start + PAGE_SIZE-1) >> PAGE_SHIFT;
-+ if (map_pages > info->nr_pages)
-+ return -EINVAL;
++/*! */
++struct efhw_nic {
++ /*! zero base index in efrm_nic_table.nic array */
++ volatile int index;
++ int ifindex; /*!< OS level nic index */
++#ifdef HAS_NET_NAMESPACE
++ struct net *nd_net;
++#endif
+
-+ map = kzalloc(sizeof(*map), GFP_KERNEL);
-+ if (map == NULL)
-+ return -ENOMEM;
++ struct efhw_device_type devtype;
+
-+ map->vma = vma;
-+ map->faults = 0;
-+ map->info = info;
-+ atomic_set(&map->map_refs, 1);
++ /*! Options that can be set by user. */
++ unsigned options;
++# define NIC_OPT_EFTEST 0x1 /* owner is an eftest app */
+
-+ mutex_lock(&info->mm_lock);
-+ list_add(&map->link, &info->mappings);
-+ mutex_unlock(&info->mm_lock);
++# define NIC_OPT_DEFAULT 0
+
-+ vma->vm_ops = &xenfb_vm_ops;
-+ vma->vm_flags |= (VM_DONTEXPAND | VM_RESERVED);
-+ vma->vm_private_data = map;
++ /*! Internal flags that indicate hardware properties at runtime. */
++ unsigned flags;
++# define NIC_FLAG_NO_INTERRUPT 0x01 /* to be set at init time only */
++# define NIC_FLAG_TRY_MSI 0x02
++# define NIC_FLAG_MSI 0x04
++# define NIC_FLAG_OS_IRQ_EN 0x08
++# define NIC_FLAG_10G 0x10
+
-+ return 0;
-+}
++ unsigned mtu; /*!< MAC MTU (includes MAC hdr) */
+
-+static struct fb_ops xenfb_fb_ops = {
-+ .owner = THIS_MODULE,
-+ .fb_setcolreg = xenfb_setcolreg,
-+ .fb_fillrect = xenfb_fillrect,
-+ .fb_copyarea = xenfb_copyarea,
-+ .fb_imageblit = xenfb_imageblit,
-+ .fb_mmap = xenfb_mmap,
-+};
++ /* hardware resources */
+
-+static irqreturn_t xenfb_event_handler(int rq, void *dev_id,
-+ struct pt_regs *regs)
-+{
-+ /*
-+ * No in events recognized, simply ignore them all.
-+ * If you need to recognize some, see xenbkd's input_handler()
-+ * for how to do that.
-+ */
-+ struct xenfb_info *info = dev_id;
-+ struct xenfb_page *page = info->page;
++ /*! I/O address of the start of the bar */
++ efhw_ioaddr_t bar_ioaddr;
+
-+ if (page->in_cons != page->in_prod) {
-+ info->page->in_cons = info->page->in_prod;
-+ notify_remote_via_irq(info->irq);
-+ }
-+ return IRQ_HANDLED;
-+}
++ /*! Bar number of control aperture. */
++ unsigned ctr_ap_bar;
++ /*! Length of control aperture in bytes. */
++ unsigned ctr_ap_bytes;
+
-+static unsigned long vmalloc_to_mfn(void *address)
-+{
-+ return pfn_to_mfn(vmalloc_to_pfn(address));
-+}
++ uint8_t mac_addr[ETH_ALEN]; /*!< mac address */
+
-+static int __devinit xenfb_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
-+{
-+ struct xenfb_info *info;
-+ struct fb_info *fb_info;
-+ int ret;
++ /*! EtherFabric Functional Units -- functions */
++ const struct efhw_func_ops *efhw_func;
+
-+ info = kzalloc(sizeof(*info), GFP_KERNEL);
-+ if (info == NULL) {
-+ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
-+ return -ENOMEM;
-+ }
-+ dev->dev.driver_data = info;
-+ info->xbdev = dev;
-+ info->irq = -1;
-+ info->x1 = info->y1 = INT_MAX;
-+ spin_lock_init(&info->dirty_lock);
-+ mutex_init(&info->mm_lock);
-+ init_waitqueue_head(&info->wq);
-+ init_timer(&info->refresh);
-+ info->refresh.function = xenfb_timer;
-+ info->refresh.data = (unsigned long)info;
-+ INIT_LIST_HEAD(&info->mappings);
++ /* Value read from FPGA version register. Zero for asic. */
++ unsigned fpga_version;
+
-+ info->fb = vmalloc(xenfb_mem_len);
-+ if (info->fb == NULL)
-+ goto error_nomem;
-+ memset(info->fb, 0, xenfb_mem_len);
++ /*! This lock protects a number of misc NIC resources. It should
++ * only be used for things that can be at the bottom of the lock
++ * order. ie. You mustn't attempt to grab any other lock while
++ * holding this one.
++ */
++ spinlock_t *reg_lock;
++ spinlock_t the_reg_lock;
+
-+ info->nr_pages = (xenfb_mem_len + PAGE_SIZE - 1) >> PAGE_SHIFT;
++ int buf_commit_outstanding; /*!< outstanding buffer commits */
+
-+ info->pages = kmalloc(sizeof(struct page *) * info->nr_pages,
-+ GFP_KERNEL);
-+ if (info->pages == NULL)
-+ goto error_nomem;
++ /*! interrupt callbacks (hard-irq) */
++ void (*irq_handler) (struct efhw_nic *, int unit);
+
-+ info->mfns = vmalloc(sizeof(unsigned long) * info->nr_pages);
-+ if (!info->mfns)
-+ goto error_nomem;
++ /*! event queues per driver */
++ struct efhw_keventq evq[EFHW_KEVENTQ_MAX];
+
-+ /* set up shared page */
-+ info->page = (void *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
-+ if (!info->page)
-+ goto error_nomem;
++/* for marking when we are not using an IRQ unit
++ - 0 is a valid offset to an IRQ unit on EF1! */
++#define EFHW_IRQ_UNIT_UNUSED 0xffff
++ /*! interrupt unit in use */
++ unsigned int irq_unit[EFHW_KEVENTQ_MAX];
++ efhw_iopage_t irq_iobuff; /*!< Falcon SYSERR interrupt */
+
-+ xenfb_init_shared_page(info);
++ /* The new driverlink infrastructure. */
++ struct efx_dl_device *net_driver_dev;
++ struct efx_dlfilt_cb_s *dlfilter_cb;
+
-+ fb_info = framebuffer_alloc(sizeof(u32) * 256, NULL);
-+ /* see fishy hackery below */
-+ if (fb_info == NULL)
-+ goto error_nomem;
++ /*! Bit masks of the sizes of event queues and dma queues supported
++ * by the nic. */
++ unsigned evq_sizes;
++ unsigned rxq_sizes;
++ unsigned txq_sizes;
+
-+ /* FIXME fishy hackery */
-+ fb_info->pseudo_palette = fb_info->par;
-+ fb_info->par = info;
-+ /* /FIXME */
-+ fb_info->screen_base = info->fb;
++ /* Size of filter table (including odd and even banks). */
++ unsigned filter_tbl_size;
++};
+
-+ fb_info->fbops = &xenfb_fb_ops;
-+ fb_info->var.xres_virtual = fb_info->var.xres = info->page->width;
-+ fb_info->var.yres_virtual = fb_info->var.yres = info->page->height;
-+ fb_info->var.bits_per_pixel = info->page->depth;
+
-+ fb_info->var.red = (struct fb_bitfield){16, 8, 0};
-+ fb_info->var.green = (struct fb_bitfield){8, 8, 0};
-+ fb_info->var.blue = (struct fb_bitfield){0, 8, 0};
++#define EFHW_KVA(nic) ((nic)->bar_ioaddr)
+
-+ fb_info->var.activate = FB_ACTIVATE_NOW;
-+ fb_info->var.height = -1;
-+ fb_info->var.width = -1;
-+ fb_info->var.vmode = FB_VMODE_NONINTERLACED;
+
-+ fb_info->fix.visual = FB_VISUAL_TRUECOLOR;
-+ fb_info->fix.line_length = info->page->line_length;
-+ fb_info->fix.smem_start = 0;
-+ fb_info->fix.smem_len = xenfb_mem_len;
-+ strcpy(fb_info->fix.id, "xen");
-+ fb_info->fix.type = FB_TYPE_PACKED_PIXELS;
-+ fb_info->fix.accel = FB_ACCEL_NONE;
++#endif /* __CI_EFHW_EFHW_TYPES_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/eventq.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/eventq.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,73 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains API provided by efhw/eventq.c file. This file is not
++ * designed for use outside of the SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_EVENTQ_H__
++#define __CI_EFHW_EVENTQ_H__
++
++#include <ci/efhw/efhw_types.h>
++#include <ci/efhw/eventq_macros.h>
++
++/*! Poll the event queue. */
++extern int efhw_keventq_poll(struct efhw_nic *, struct efhw_keventq *);
++
++/*! Callbacks for handling events. */
++struct efhw_ev_handler {
++ void (*wakeup_fn)(struct efhw_nic *nic, efhw_event_t *ev);
++ void (*timeout_fn)(struct efhw_nic *nic, efhw_event_t *ev);
++ void (*sw_fn)(struct efhw_nic *nic, efhw_event_t *ev);
++ void (*dmaq_flushed_fn) (struct efhw_nic *, int, int);
++};
++
++extern int efhw_keventq_ctor(struct efhw_nic *, int instance,
++ struct efhw_keventq *, struct efhw_ev_handler *);
++extern void efhw_keventq_dtor(struct efhw_nic *, struct efhw_keventq *);
++
++extern void efhw_handle_txdmaq_flushed(struct efhw_nic *,
++ struct efhw_ev_handler *,
++ efhw_event_t *);
++extern void efhw_handle_rxdmaq_flushed(struct efhw_nic *,
++ struct efhw_ev_handler *,
++ efhw_event_t *);
++extern void efhw_handle_wakeup_event(struct efhw_nic *,
++ struct efhw_ev_handler *,
++ efhw_event_t *);
++extern void efhw_handle_timeout_event(struct efhw_nic *,
++ struct efhw_ev_handler *,
++ efhw_event_t *);
++
++#endif /* __CI_EFHW_EVENTQ_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/eventq_macros.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/eventq_macros.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,81 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides some event-related macros. This file is designed for
++ * use from kernel and from the userland contexts.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ fb_info->flags = FBINFO_FLAG_DEFAULT;
++#ifndef __CI_EFHW_EVENTQ_MACROS_H__
++#define __CI_EFHW_EVENTQ_MACROS_H__
+
-+ ret = fb_alloc_cmap(&fb_info->cmap, 256, 0);
-+ if (ret < 0) {
-+ framebuffer_release(fb_info);
-+ xenbus_dev_fatal(dev, ret, "fb_alloc_cmap");
-+ goto error;
-+ }
++#include <ci/efhw/common.h>
+
-+ ret = register_framebuffer(fb_info);
-+ if (ret) {
-+ fb_dealloc_cmap(&info->fb_info->cmap);
-+ framebuffer_release(fb_info);
-+ xenbus_dev_fatal(dev, ret, "register_framebuffer");
-+ goto error;
-+ }
-+ info->fb_info = fb_info;
++/*--------------------------------------------------------------------
++ *
++ * Event Queue manipulation
++ *
++ *--------------------------------------------------------------------*/
+
-+ /* FIXME should this be delayed until backend XenbusStateConnected? */
-+ info->kthread = kthread_run(xenfb_thread, info, "xenfb thread");
-+ if (IS_ERR(info->kthread)) {
-+ ret = PTR_ERR(info->kthread);
-+ info->kthread = NULL;
-+ xenbus_dev_fatal(dev, ret, "register_framebuffer");
-+ goto error;
-+ }
++#define EFHW_EVENT_OFFSET(q, s, i) \
++ (((s)->evq_ptr - (i) * (int32_t)sizeof(efhw_event_t)) \
++ & (q)->evq_mask)
+
-+ ret = xenfb_connect_backend(dev, info);
-+ if (ret < 0)
-+ goto error;
++#define EFHW_EVENT_PTR(q, s, i) \
++ ((efhw_event_t *)((q)->evq_base + EFHW_EVENT_OFFSET(q, s, i)))
+
-+ return 0;
++#define EFHW_EVENTQ_NEXT(s) \
++ do { ((s)->evq_ptr += sizeof(efhw_event_t)); } while (0)
+
-+ error_nomem:
-+ ret = -ENOMEM;
-+ xenbus_dev_fatal(dev, ret, "allocating device memory");
-+ error:
-+ xenfb_remove(dev);
-+ return ret;
-+}
++#define EFHW_EVENTQ_PREV(s) \
++ do { ((s)->evq_ptr -= sizeof(efhw_event_t)); } while (0)
+
-+static int xenfb_resume(struct xenbus_device *dev)
-+{
-+ struct xenfb_info *info = dev->dev.driver_data;
++/* Be worried about this on byteswapped machines */
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ /* Due to crazy chipsets, we see the event words being written in
++ ** arbitrary order (bug4539). So test for presence of event must ensure
++ ** that both halves have changed from the null.
++ */
++ #define EFHW_IS_EVENT(evp) \
++ (((evp)->opaque.a != (uint32_t)-1) && \
++ ((evp)->opaque.b != (uint32_t)-1))
++ #define EFHW_CLEAR_EVENT(evp) ((evp)->u64 = (uint64_t)-1)
++ #define EFHW_CLEAR_EVENT_VALUE 0xff
++#else
++ #error Fixme - unknown hardware configuration
++#endif
+
-+ xenfb_disconnect_backend(info);
-+ xenfb_init_shared_page(info);
-+ return xenfb_connect_backend(dev, info);
-+}
++#define EFHW_EVENT_OVERFLOW(evq, s) \
++ (EFHW_IS_EVENT(EFHW_EVENT_PTR(evq, s, 1)))
+
-+static int xenfb_remove(struct xenbus_device *dev)
-+{
-+ struct xenfb_info *info = dev->dev.driver_data;
++#endif /* __CI_EFHW_EVENTQ_MACROS_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/falcon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/falcon.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,93 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains API provided by efhw/falcon.c file. This file is not
++ * designed for use outside of the SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ del_timer(&info->refresh);
-+ if (info->kthread)
-+ kthread_stop(info->kthread);
-+ xenfb_disconnect_backend(info);
-+ if (info->fb_info) {
-+ unregister_framebuffer(info->fb_info);
-+ fb_dealloc_cmap(&info->fb_info->cmap);
-+ framebuffer_release(info->fb_info);
-+ }
-+ free_page((unsigned long)info->page);
-+ vfree(info->mfns);
-+ kfree(info->pages);
-+ vfree(info->fb);
-+ kfree(info);
++#ifndef __CI_EFHW_FALCON_H__
++#define __CI_EFHW_FALCON_H__
+
-+ return 0;
-+}
++#include <ci/efhw/efhw_types.h>
++#include <ci/efhw/common.h>
+
-+static void xenfb_init_shared_page(struct xenfb_info *info)
-+{
-+ int i;
++/*----------------------------------------------------------------------------
++ *
++ * Locks - unfortunately required
++ *
++ *---------------------------------------------------------------------------*/
+
-+ for (i = 0; i < info->nr_pages; i++)
-+ info->pages[i] = vmalloc_to_page(info->fb + i * PAGE_SIZE);
++#define FALCON_LOCK_DECL irq_flags_t lock_state
++#define FALCON_LOCK_LOCK(nic) \
++ spin_lock_irqsave((nic)->reg_lock, lock_state)
++#define FALCON_LOCK_UNLOCK(nic) \
++ spin_unlock_irqrestore((nic)->reg_lock, lock_state)
+
-+ for (i = 0; i < info->nr_pages; i++)
-+ info->mfns[i] = vmalloc_to_mfn(info->fb + i * PAGE_SIZE);
++extern struct efhw_func_ops falcon_char_functional_units;
+
-+ info->page->pd[0] = vmalloc_to_mfn(info->mfns);
-+ info->page->pd[1] = 0;
-+ info->page->width = XENFB_WIDTH;
-+ info->page->height = XENFB_HEIGHT;
-+ info->page->depth = XENFB_DEPTH;
-+ info->page->line_length = (info->page->depth / 8) * info->page->width;
-+ info->page->mem_length = xenfb_mem_len;
-+ info->page->in_cons = info->page->in_prod = 0;
-+ info->page->out_cons = info->page->out_prod = 0;
-+}
++/*! specify a pace value for a TX DMA Queue */
++extern void falcon_nic_pace(struct efhw_nic *nic, uint dmaq, uint pace);
+
-+static int xenfb_connect_backend(struct xenbus_device *dev,
-+ struct xenfb_info *info)
-+{
-+ int ret;
-+ struct xenbus_transaction xbt;
++/*! confirm buffer table updates - should be used for items where
++ loss of data would be unacceptable. E.g for the buffers that back
++ an event or DMA queue */
++extern void falcon_nic_buffer_table_confirm(struct efhw_nic *nic);
+
-+ ret = bind_listening_port_to_irqhandler(
-+ dev->otherend_id, xenfb_event_handler, 0, "xenfb", info);
-+ if (ret < 0) {
-+ xenbus_dev_fatal(dev, ret,
-+ "bind_listening_port_to_irqhandler");
-+ return ret;
-+ }
-+ info->irq = ret;
++/*! Reset the all the TX DMA queue pointers. */
++extern void falcon_clobber_tx_dma_ptrs(struct efhw_nic *nic, uint dmaq);
+
-+ again:
-+ ret = xenbus_transaction_start(&xbt);
-+ if (ret) {
-+ xenbus_dev_fatal(dev, ret, "starting transaction");
-+ return ret;
-+ }
-+ ret = xenbus_printf(xbt, dev->nodename, "page-ref", "%lu",
-+ virt_to_mfn(info->page));
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
-+ irq_to_evtchn_port(info->irq));
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
-+ XEN_IO_PROTO_ABI_NATIVE);
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_printf(xbt, dev->nodename, "feature-update", "1");
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_transaction_end(xbt, 0);
-+ if (ret) {
-+ if (ret == -EAGAIN)
-+ goto again;
-+ xenbus_dev_fatal(dev, ret, "completing transaction");
-+ return ret;
-+ }
++extern int
++falcon_handle_char_event(struct efhw_nic *nic,
++ struct efhw_ev_handler *h, efhw_event_t *evp);
+
-+ xenbus_switch_state(dev, XenbusStateInitialised);
-+ return 0;
++/*! map event queue instance space (0,1,2,..) onto event queue
++ number. This function takes into account the allocation rules for
++ the underlying driver model */
++extern int falcon_idx_to_evq(struct efhw_nic *nic, uint idx);
+
-+ error_xenbus:
-+ xenbus_transaction_end(xbt, 1);
-+ xenbus_dev_fatal(dev, ret, "writing xenstore");
-+ return ret;
-+}
++/*! Acknowledge to HW that processing is complete on a given event queue */
++extern void falcon_nic_evq_ack(struct efhw_nic *nic, uint evq, /* evq id */
++ uint rptr, /* new read pointer update */
++ bool wakeup /* request a wakeup event if
++ ptr's != */
++ );
+
-+static void xenfb_disconnect_backend(struct xenfb_info *info)
-+{
-+ if (info->irq >= 0)
-+ unbind_from_irqhandler(info->irq, info);
-+ info->irq = -1;
-+}
++extern void
++falcon_nic_buffer_table_set_n(struct efhw_nic *nic, int buffer_id,
++ dma_addr_t dma_addr, uint bufsz, uint region,
++ int n_pages, int own_id);
+
-+static void xenfb_backend_changed(struct xenbus_device *dev,
-+ enum xenbus_state backend_state)
-+{
-+ struct xenfb_info *info = dev->dev.driver_data;
-+ int val;
++extern void falcon_nic_ipfilter_ctor(struct efhw_nic *nic);
+
-+ switch (backend_state) {
-+ case XenbusStateInitialising:
-+ case XenbusStateInitialised:
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ break;
++#endif /* __CI_EFHW_FALCON_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/falcon_hash.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/falcon_hash.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,58 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains API provided by efhw/falcon_hash.c file.
++ * Function declared in this file are not exported from the Linux
++ * sfc_resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ case XenbusStateInitWait:
-+ InitWait:
-+ xenbus_switch_state(dev, XenbusStateConnected);
-+ break;
++#ifndef __CI_EFHW_FALCON_HASH_H__
++#define __CI_EFHW_FALCON_HASH_H__
+
-+ case XenbusStateConnected:
-+ /*
-+ * Work around xenbus race condition: If backend goes
-+ * through InitWait to Connected fast enough, we can
-+ * get Connected twice here.
-+ */
-+ if (dev->state != XenbusStateConnected)
-+ goto InitWait; /* no InitWait seen yet, fudge it */
++/* All LE parameters */
++extern unsigned int
++falcon_hash_get_key(unsigned int src_ip, unsigned int src_port,
++ unsigned int dest_ip, unsigned int dest_port,
++ int tcp, int full);
+
-+ if (xenbus_scanf(XBT_NIL, info->xbdev->otherend,
-+ "request-update", "%d", &val) < 0)
-+ val = 0;
-+ if (val)
-+ info->update_wanted = 1;
-+ break;
++unsigned int falcon_hash_function1(unsigned int key, unsigned int nfilters);
+
-+ case XenbusStateClosing:
-+ // FIXME is this safe in any dev->state?
-+ xenbus_frontend_closed(dev);
-+ break;
-+ }
-+}
++extern unsigned int
++falcon_hash_function2(unsigned int key, unsigned int nfitlers);
+
-+static struct xenbus_device_id xenfb_ids[] = {
-+ { "vfb" },
-+ { "" }
-+};
-+MODULE_ALIAS("xen:vfb");
++extern unsigned int
++falcon_hash_iterator(unsigned int hash1, unsigned int hash2,
++ unsigned int n_search, unsigned int nfilters);
+
-+static struct xenbus_driver xenfb = {
-+ .name = "vfb",
-+ .owner = THIS_MODULE,
-+ .ids = xenfb_ids,
-+ .probe = xenfb_probe,
-+ .remove = xenfb_remove,
-+ .resume = xenfb_resume,
-+ .otherend_changed = xenfb_backend_changed,
-+};
++#endif /* __CI_EFHW_FALCON_HASH_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/hardware_sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/hardware_sysdep.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,84 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for header files
++ * with hardware-related definitions (in ci/driver/efab/hardware*).
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static int __init xenfb_init(void)
-+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++#ifndef __CI_EFHW_HARDWARE_LINUX_H__
++#define __CI_EFHW_HARDWARE_LINUX_H__
+
-+ /* Nothing to do if running in dom0. */
-+ if (is_initial_xendomain())
-+ return -ENODEV;
++#include <asm/io.h>
+
-+ return xenbus_register_frontend(&xenfb);
-+}
++#ifdef __LITTLE_ENDIAN
++#define EFHW_IS_LITTLE_ENDIAN
++#elif __BIG_ENDIAN
++#define EFHW_IS_BIG_ENDIAN
++#else
++#error Unknown endianness
++#endif
++
++#ifndef mmiowb
++ #if defined(__i386__) || defined(__x86_64__)
++ #define mmiowb()
++ #elif defined(__ia64__)
++ #ifndef ia64_mfa
++ #define ia64_mfa() asm volatile ("mf.a" ::: "memory")
++ #endif
++ #define mmiowb ia64_mfa
++ #else
++ #error "Need definition for mmiowb()"
++ #endif
++#endif
+
-+static void __exit xenfb_cleanup(void)
++typedef char *efhw_ioaddr_t;
++
++#ifndef readq
++static inline uint64_t __readq(void __iomem *addr)
+{
-+ return xenbus_unregister_driver(&xenfb);
++ return *(volatile uint64_t *)addr;
+}
++#define readq(x) __readq(x)
++#endif
+
-+module_init(xenfb_init);
-+module_exit(xenfb_cleanup);
++#ifndef writeq
++static inline void __writeq(uint64_t v, void __iomem *addr)
++{
++ *(volatile uint64_t *)addr = v;
++}
++#define writeq(val, addr) __writeq((val), (addr))
++#endif
+
-+MODULE_LICENSE("GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/fbfront/xenkbd.c linux-2.6.18-xen.hg/drivers/xen/fbfront/xenkbd.c
---- linux-2.6.18/drivers/xen/fbfront/xenkbd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/fbfront/xenkbd.c 2007-12-23 11:15:33.871259312 +0100
-@@ -0,0 +1,344 @@
-+/*
-+ * linux/drivers/input/keyboard/xenkbd.c -- Xen para-virtual input device
++#endif /* __CI_EFHW_HARDWARE_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/iopage.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/iopage.h 2008-05-19 00:33:29.369839038 +0300
+@@ -0,0 +1,58 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
+ *
-+ * Copyright (C) 2005 Anthony Liguori <aliguori@us.ibm.com>
-+ * Copyright (C) 2006 Red Hat, Inc., Markus Armbruster <armbru@redhat.com>
++ * This file contains OS-independent API for allocating iopage types.
++ * The implementation of these functions is highly OS-dependent.
++ * This file is not designed for use outside of the SFC resource driver.
+ *
-+ * Based on linux/drivers/input/mouse/sermouse.c
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * This file is subject to the terms and conditions of the GNU General Public
-+ * License. See the file COPYING in the main directory of this archive for
-+ * more details.
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+/*
-+ * TODO:
++#ifndef __CI_DRIVER_RESOURCE_IOPAGE_H__
++#define __CI_DRIVER_RESOURCE_IOPAGE_H__
++
++#include <ci/efhw/efhw_types.h>
++
++/*--------------------------------------------------------------------
+ *
-+ * Switch to grant tables together with xenfb.c.
++ * memory allocation
++ *
++ *--------------------------------------------------------------------*/
++
++extern int efhw_iopage_alloc(struct efhw_nic *, efhw_iopage_t *p);
++extern void efhw_iopage_free(struct efhw_nic *, efhw_iopage_t *p);
++
++extern int efhw_iopages_alloc(struct efhw_nic *, efhw_iopages_t *p,
++ unsigned order);
++extern void efhw_iopages_free(struct efhw_nic *, efhw_iopages_t *p);
++
++#endif /* __CI_DRIVER_RESOURCE_IOPAGE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/iopage_types.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/iopage_types.h 2008-05-19 00:33:29.373839269 +0300
+@@ -0,0 +1,188 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides efhw_page_t and efhw_iopage_t for Linux kernel.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/kernel.h>
-+#include <linux/errno.h>
-+#include <linux/module.h>
-+#include <linux/input.h>
-+#include <asm/hypervisor.h>
-+#include <xen/evtchn.h>
-+#include <xen/interface/io/fbif.h>
-+#include <xen/interface/io/kbdif.h>
-+#include <xen/xenbus.h>
++#ifndef __CI_EFHW_IOPAGE_LINUX_H__
++#define __CI_EFHW_IOPAGE_LINUX_H__
+
-+struct xenkbd_info
-+{
-+ struct input_dev *kbd;
-+ struct input_dev *ptr;
-+ struct xenkbd_page *page;
-+ int irq;
-+ struct xenbus_device *xbdev;
-+ char phys[32];
-+};
++#include <linux/gfp.h>
++#include <linux/hardirq.h>
++#include <ci/efhw/debug.h>
+
-+static int xenkbd_remove(struct xenbus_device *);
-+static int xenkbd_connect_backend(struct xenbus_device *, struct xenkbd_info *);
-+static void xenkbd_disconnect_backend(struct xenkbd_info *);
++/*--------------------------------------------------------------------
++ *
++ * efhw_page_t: A single page of memory. Directly mapped in the driver,
++ * and can be mapped to userlevel.
++ *
++ *--------------------------------------------------------------------*/
+
-+/*
-+ * Note: if you need to send out events, see xenfb_do_update() for how
-+ * to do that.
-+ */
++typedef struct {
++ unsigned long kva;
++} efhw_page_t;
+
-+static irqreturn_t input_handler(int rq, void *dev_id, struct pt_regs *regs)
++static inline int efhw_page_alloc(efhw_page_t *p)
+{
-+ struct xenkbd_info *info = dev_id;
-+ struct xenkbd_page *page = info->page;
-+ __u32 cons, prod;
++ p->kva = __get_free_page(in_interrupt()? GFP_ATOMIC : GFP_KERNEL);
++ return p->kva ? 0 : -ENOMEM;
++}
+
-+ prod = page->in_prod;
-+ if (prod == page->out_cons)
-+ return IRQ_HANDLED;
-+ rmb(); /* ensure we see ring contents up to prod */
-+ for (cons = page->in_cons; cons != prod; cons++) {
-+ union xenkbd_in_event *event;
-+ struct input_dev *dev;
-+ event = &XENKBD_IN_RING_REF(page, cons);
++static inline int efhw_page_alloc_zeroed(efhw_page_t *p)
++{
++ p->kva = get_zeroed_page(in_interrupt()? GFP_ATOMIC : GFP_KERNEL);
++ return p->kva ? 0 : -ENOMEM;
++}
+
-+ dev = info->ptr;
-+ switch (event->type) {
-+ case XENKBD_TYPE_MOTION:
-+ if ( event->motion.rel_z == 1 || event->motion.rel_z == -1 ) {
-+ input_report_rel(dev, REL_WHEEL, 0 - event->motion.rel_z);
-+ }
-+ else {
-+ input_report_rel(dev, REL_X, event->motion.rel_x);
-+ input_report_rel(dev, REL_Y, event->motion.rel_y);
-+ }
-+ break;
-+ case XENKBD_TYPE_KEY:
-+ dev = NULL;
-+ if (test_bit(event->key.keycode, info->kbd->keybit))
-+ dev = info->kbd;
-+ if (test_bit(event->key.keycode, info->ptr->keybit))
-+ dev = info->ptr;
-+ if (dev)
-+ input_report_key(dev, event->key.keycode,
-+ event->key.pressed);
-+ else
-+ printk("xenkbd: unhandled keycode 0x%x\n",
-+ event->key.keycode);
-+ break;
-+ case XENKBD_TYPE_POS:
-+ if ( event->pos.abs_z == 1 || event->pos.abs_z == -1 ) {
-+ input_report_rel(dev, REL_WHEEL, 0 - event->pos.abs_z);
-+ }
-+ else {
-+ input_report_abs(dev, ABS_X, event->pos.abs_x);
-+ input_report_abs(dev, ABS_Y, event->pos.abs_y);
-+ }
-+ break;
-+ }
-+ if (dev)
-+ input_sync(dev);
-+ }
-+ mb(); /* ensure we got ring contents */
-+ page->in_cons = cons;
-+ notify_remote_via_irq(info->irq);
++static inline void efhw_page_free(efhw_page_t *p)
++{
++ free_page(p->kva);
++ EFHW_DO_DEBUG(memset(p, 0, sizeof(*p)));
++}
+
-+ return IRQ_HANDLED;
++static inline char *efhw_page_ptr(efhw_page_t *p)
++{
++ return (char *)p->kva;
+}
+
-+int __devinit xenkbd_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
++static inline unsigned efhw_page_pfn(efhw_page_t *p)
+{
-+ int ret, i;
-+ struct xenkbd_info *info;
-+ struct input_dev *kbd, *ptr;
++ return (unsigned)(__pa(p->kva) >> PAGE_SHIFT);
++}
+
-+ info = kzalloc(sizeof(*info), GFP_KERNEL);
-+ if (!info) {
-+ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
-+ return -ENOMEM;
-+ }
-+ dev->dev.driver_data = info;
-+ info->xbdev = dev;
-+ snprintf(info->phys, sizeof(info->phys), "xenbus/%s", dev->nodename);
++static inline void efhw_page_mark_invalid(efhw_page_t *p)
++{
++ p->kva = 0;
++}
+
-+ info->page = (void *)__get_free_page(GFP_KERNEL);
-+ if (!info->page)
-+ goto error_nomem;
-+ info->page->in_cons = info->page->in_prod = 0;
-+ info->page->out_cons = info->page->out_prod = 0;
++static inline int efhw_page_is_valid(efhw_page_t *p)
++{
++ return p->kva != 0;
++}
+
-+ /* keyboard */
-+ kbd = input_allocate_device();
-+ if (!kbd)
-+ goto error_nomem;
-+ kbd->name = "Xen Virtual Keyboard";
-+ kbd->phys = info->phys;
-+ kbd->id.bustype = BUS_PCI;
-+ kbd->id.vendor = 0x5853;
-+ kbd->id.product = 0xffff;
-+ kbd->evbit[0] = BIT(EV_KEY);
-+ for (i = KEY_ESC; i < KEY_UNKNOWN; i++)
-+ set_bit(i, kbd->keybit);
-+ for (i = KEY_OK; i < KEY_MAX; i++)
-+ set_bit(i, kbd->keybit);
++static inline void efhw_page_init_from_va(efhw_page_t *p, void *va)
++{
++ p->kva = (unsigned long)va;
++}
+
-+ ret = input_register_device(kbd);
-+ if (ret) {
-+ input_free_device(kbd);
-+ xenbus_dev_fatal(dev, ret, "input_register_device(kbd)");
-+ goto error;
-+ }
-+ info->kbd = kbd;
++/*--------------------------------------------------------------------
++ *
++ * efhw_iopage_t: A single page of memory. Directly mapped in the driver,
++ * and can be mapped to userlevel. Can also be accessed by the NIC.
++ *
++ *--------------------------------------------------------------------*/
+
-+ /* pointing device */
-+ ptr = input_allocate_device();
-+ if (!ptr)
-+ goto error_nomem;
-+ ptr->name = "Xen Virtual Pointer";
-+ ptr->phys = info->phys;
-+ ptr->id.bustype = BUS_PCI;
-+ ptr->id.vendor = 0x5853;
-+ ptr->id.product = 0xfffe;
-+ ptr->evbit[0] = BIT(EV_KEY) | BIT(EV_REL) | BIT(EV_ABS);
-+ for (i = BTN_LEFT; i <= BTN_TASK; i++)
-+ set_bit(i, ptr->keybit);
-+ ptr->relbit[0] = BIT(REL_X) | BIT(REL_Y) | BIT(REL_WHEEL);
-+ input_set_abs_params(ptr, ABS_X, 0, XENFB_WIDTH, 0, 0);
-+ input_set_abs_params(ptr, ABS_Y, 0, XENFB_HEIGHT, 0, 0);
++typedef struct {
++ efhw_page_t p;
++ dma_addr_t dma_addr;
++} efhw_iopage_t;
+
-+ ret = input_register_device(ptr);
-+ if (ret) {
-+ input_free_device(ptr);
-+ xenbus_dev_fatal(dev, ret, "input_register_device(ptr)");
-+ goto error;
-+ }
-+ info->ptr = ptr;
++static inline dma_addr_t efhw_iopage_dma_addr(efhw_iopage_t *p)
++{
++ return p->dma_addr;
++}
+
-+ ret = xenkbd_connect_backend(dev, info);
-+ if (ret < 0)
-+ goto error;
++#define efhw_iopage_ptr(iop) efhw_page_ptr(&(iop)->p)
++#define efhw_iopage_pfn(iop) efhw_page_pfn(&(iop)->p)
++#define efhw_iopage_mark_invalid(iop) efhw_page_mark_invalid(&(iop)->p)
++#define efhw_iopage_is_valid(iop) efhw_page_is_valid(&(iop)->p)
+
-+ return 0;
++/*--------------------------------------------------------------------
++ *
++ * efhw_iopages_t: A set of pages that are contiguous in physical memory.
++ * Directly mapped in the driver, and can be mapped to userlevel. Can also
++ * be accessed by the NIC.
++ *
++ * NB. The O/S may be unwilling to allocate many, or even any of these. So
++ * only use this type where the NIC really needs a physically contiguous
++ * buffer.
++ *
++ *--------------------------------------------------------------------*/
+
-+ error_nomem:
-+ ret = -ENOMEM;
-+ xenbus_dev_fatal(dev, ret, "allocating device memory");
-+ error:
-+ xenkbd_remove(dev);
-+ return ret;
++typedef struct {
++ caddr_t kva;
++ unsigned order;
++ dma_addr_t dma_addr;
++} efhw_iopages_t;
++
++static inline caddr_t efhw_iopages_ptr(efhw_iopages_t *p)
++{
++ return p->kva;
+}
+
-+static int xenkbd_resume(struct xenbus_device *dev)
++static inline unsigned efhw_iopages_pfn(efhw_iopages_t *p)
+{
-+ struct xenkbd_info *info = dev->dev.driver_data;
++ return (unsigned)(__pa(p->kva) >> PAGE_SHIFT);
++}
+
-+ xenkbd_disconnect_backend(info);
-+ return xenkbd_connect_backend(dev, info);
++static inline dma_addr_t efhw_iopages_dma_addr(efhw_iopages_t *p)
++{
++ return p->dma_addr;
+}
+
-+static int xenkbd_remove(struct xenbus_device *dev)
++static inline unsigned efhw_iopages_size(efhw_iopages_t *p)
+{
-+ struct xenkbd_info *info = dev->dev.driver_data;
++ return 1u << (p->order + PAGE_SHIFT);
++}
+
-+ xenkbd_disconnect_backend(info);
-+ input_unregister_device(info->kbd);
-+ input_unregister_device(info->ptr);
-+ free_page((unsigned long)info->page);
-+ kfree(info);
-+ return 0;
++/* efhw_iopage_t <-> efhw_iopages_t conversions for handling physically
++ * contiguous allocations in iobufsets for iSCSI. This allows the
++ * essential information about contiguous allocations from
++ * efhw_iopages_alloc() to be saved away in the efhw_iopage_t array in an
++ * iobufset. (Changing the iobufset resource to use a union type would
++ * involve a lot of code changes, and make the iobufset's metadata larger
++ * which could be bad as it's supposed to fit into a single page on some
++ * platforms.)
++ */
++static inline void
++efhw_iopage_init_from_iopages(efhw_iopage_t *iopage,
++ efhw_iopages_t *iopages, unsigned pageno)
++{
++ iopage->p.kva = ((unsigned long)efhw_iopages_ptr(iopages))
++ + (pageno * PAGE_SIZE);
++ iopage->dma_addr = efhw_iopages_dma_addr(iopages) +
++ (pageno * PAGE_SIZE);
+}
+
-+static int xenkbd_connect_backend(struct xenbus_device *dev,
-+ struct xenkbd_info *info)
++static inline void
++efhw_iopages_init_from_iopage(efhw_iopages_t *iopages,
++ efhw_iopage_t *iopage, unsigned order)
+{
-+ int ret;
-+ struct xenbus_transaction xbt;
++ iopages->kva = (caddr_t) efhw_iopage_ptr(iopage);
++ EFHW_ASSERT(iopages->kva);
++ iopages->order = order;
++ iopages->dma_addr = efhw_iopage_dma_addr(iopage);
++}
+
-+ ret = bind_listening_port_to_irqhandler(
-+ dev->otherend_id, input_handler, 0, "xenkbd", info);
-+ if (ret < 0) {
-+ xenbus_dev_fatal(dev, ret,
-+ "bind_listening_port_to_irqhandler");
-+ return ret;
-+ }
-+ info->irq = ret;
++#endif /* __CI_EFHW_IOPAGE_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/nic.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/nic.h 2008-05-19 00:33:29.373839269 +0300
+@@ -0,0 +1,62 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains API provided by efhw/nic.c file. This file is not
++ * designed for use outside of the SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ again:
-+ ret = xenbus_transaction_start(&xbt);
-+ if (ret) {
-+ xenbus_dev_fatal(dev, ret, "starting transaction");
++#ifndef __CI_EFHW_NIC_H__
++#define __CI_EFHW_NIC_H__
++
++#include <ci/efhw/efhw_types.h>
++#include <ci/efhw/public.h>
++
++
++/* Convert PCI info to device type. Returns false when device is not
++ * recognised.
++ */
++extern int efhw_device_type_init(struct efhw_device_type *dt,
++ int vendor_id, int device_id, int revision);
++
++/* Initialise fields that do not involve touching hardware. */
++extern void efhw_nic_init(struct efhw_nic *nic, unsigned flags,
++ unsigned options, struct efhw_device_type dev_type);
++
++/*! Destruct NIC resources */
++extern void efhw_nic_dtor(struct efhw_nic *nic);
++
++/*! Shutdown interrupts */
++extern void efhw_nic_close_interrupts(struct efhw_nic *nic);
++
++#endif /* __CI_EFHW_NIC_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/public.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/public.h 2008-05-19 00:33:29.373839269 +0300
+@@ -0,0 +1,83 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API of efhw library exported from the SFC
++ * resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_PUBLIC_H__
++#define __CI_EFHW_PUBLIC_H__
++
++#include <ci/efhw/common.h>
++#include <ci/efhw/efhw_types.h>
++
++/*! Returns true if we have some EtherFabric functional units -
++ whether configured or not */
++static inline int efhw_nic_have_functional_units(struct efhw_nic *nic)
++{
++ return nic->efhw_func != 0;
++}
++
++/*! Returns true if the EtherFabric functional units have been configured */
++static inline int efhw_nic_have_hw(struct efhw_nic *nic)
++{
++ return efhw_nic_have_functional_units(nic) && (EFHW_KVA(nic) != 0);
++}
++
++/*! Helper function to allocate the iobuffer needed by an eventq
++ * - it ensures the eventq has the correct alignment for the NIC
++ *
++ * \param rm Event-queue resource manager
++ * \param instance Event-queue instance (index)
++ * \param buf_bytes Requested size of eventq
++ * \return < 0 if iobuffer allocation fails
++ */
++int efhw_nic_event_queue_alloc_iobuffer(struct efhw_nic *nic,
++ struct eventq_resource_hardware *h,
++ int evq_instance, unsigned buf_bytes);
++
++extern void falcon_nic_set_rx_usr_buf_size(struct efhw_nic *,
++ int rx_usr_buf_size);
++
++extern void
++falcon_nic_rx_filter_ctl_set(struct efhw_nic *nic, uint32_t tcp_full,
++ uint32_t tcp_wild,
++ uint32_t udp_full, uint32_t udp_wild);
++
++extern void
++falcon_nic_rx_filter_ctl_get(struct efhw_nic *nic, uint32_t *tcp_full,
++ uint32_t *tcp_wild,
++ uint32_t *udp_full, uint32_t *udp_wild);
++
++#endif /* __CI_EFHW_PUBLIC_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efhw/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efhw/sysdep.h 2008-05-19 00:33:29.405841113 +0300
+@@ -0,0 +1,72 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for efhw library.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_SYSDEP_LINUX_H__
++#define __CI_EFHW_SYSDEP_LINUX_H__
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/spinlock.h>
++#include <linux/delay.h>
++#include <linux/if_ether.h>
++
++#include <linux/netdevice.h> /* necessary for etherdevice.h on some kernels */
++#include <linux/etherdevice.h>
++
++#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,21)
++static inline int is_local_ether_addr(const u8 *addr)
++{
++ return (0x02 & addr[0]);
++}
++#endif
++
++typedef unsigned long irq_flags_t;
++
++#define spin_lock_destroy(l_) do {} while (0)
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24)
++#define HAS_NET_NAMESPACE
++#endif
++
++/* Funny, but linux has round_up for x86 only, defined in
++ * x86-specific header */
++#ifndef round_up
++#define round_up(x, y) (((x) + (y) - 1) & ~((y)-1))
++#endif
++
++#endif /* __CI_EFHW_SYSDEP_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/buddy.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/buddy.h 2008-05-19 00:33:29.405841113 +0300
+@@ -0,0 +1,69 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides private API for buddy allocator. This API is not
++ * designed for use outside of SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_BUDDY_H__
++#define __CI_EFRM_BUDDY_H__
++
++#include <ci/efrm/sysdep.h>
++
++/*! Comment? */
++struct efrm_buddy_allocator {
++ struct list_head *free_lists; /* array[order+1] */
++ struct list_head *links; /* array[1<<order] */
++ uint8_t *orders; /* array[1<<order] */
++ unsigned order; /*!< total size == (1 << order) */
++ /* ?? Consider recording largest available order + for each order the
++ ** smallest available order that is big enough.
++ */
++};
++
++ /*! Returns total size of managed space. */
++static inline unsigned long efrm_buddy_size(struct efrm_buddy_allocator *b)
++{
++ return 1ul << b->order;
++}
++
++int efrm_buddy_ctor(struct efrm_buddy_allocator *b, unsigned order);
++void efrm_buddy_dtor(struct efrm_buddy_allocator *b);
++int efrm_buddy_alloc(struct efrm_buddy_allocator *b, unsigned order);
++void efrm_buddy_free(struct efrm_buddy_allocator *b, unsigned addr,
++ unsigned order);
++void efrm_buddy_reserve_at_start(struct efrm_buddy_allocator *b, unsigned n);
++void efrm_buddy_reserve_at_end(struct efrm_buddy_allocator *b, unsigned n);
++
++#endif /* __CI_EFRM_BUDDY_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/buffer_table.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/buffer_table.h 2008-05-19 00:33:29.405841113 +0300
+@@ -0,0 +1,86 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides private buffer table API. This API is not designed
++ * for use outside of SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_BUFFER_TABLE_H__
++#define __CI_EFRM_BUFFER_TABLE_H__
++
++#include <ci/efhw/efhw_types.h>
++
++/*--------------------------------------------------------------------
++ *
++ * NIC's buffer table.
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Managed interface. */
++
++/*! construct a managed buffer table object, allocated over a region of
++ * the NICs buffer table space
++ */
++extern int efrm_buffer_table_ctor(unsigned low, unsigned high);
++/*! destructor for above */
++extern void efrm_buffer_table_dtor(void);
++
++/*! allocate a contiguous region of buffer table space */
++extern int efrm_buffer_table_alloc(unsigned order,
++ struct efhw_buffer_table_allocation *a);
++
++/*! current size of the buffer table.
++ * FIXME This function should be inline, but it is never used from
++ * the fast path, so let it as-is. */
++unsigned long efrm_buffer_table_size(void);
++
++/*--------------------------------------------------------------------
++ *
++ * buffer table operations through the HW independent API
++ *
++ *--------------------------------------------------------------------*/
++
++/*! free a previously allocated region of buffer table space */
++extern void efrm_buffer_table_free(struct efhw_buffer_table_allocation *a);
++
++/*! commit the update of a buffer table entry to every NIC */
++void efrm_buffer_table_commit(void);
++
++/*! set a given buffer table entry. [pa] should be the physical
++ address of pinned down memory. This function can only be called from
++ the char driver */
++void efrm_buffer_table_set(struct efhw_buffer_table_allocation *a,
++ unsigned i, dma_addr_t dma_addr, int owner);
++
++#endif /* __CI_EFRM_BUFFER_TABLE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/debug.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/debug.h 2008-05-19 00:33:29.409841344 +0300
+@@ -0,0 +1,78 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides debug-related API for efrm library using Linux kernel
++ * primitives.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_DEBUG_LINUX_H__
++#define __CI_EFRM_DEBUG_LINUX_H__
++
++#define EFRM_PRINTK_PREFIX "[sfc efrm] "
++
++#define EFRM_PRINTK(level, fmt, ...) \
++ printk(level EFRM_PRINTK_PREFIX fmt "\n", __VA_ARGS__)
++
++/* Following macros should be used with non-zero format parameters
++ * due to __VA_ARGS__ limitations. Use "%s" with __FUNCTION__ if you can't
++ * find better parameters. */
++#define EFRM_ERR(fmt, ...) EFRM_PRINTK(KERN_ERR, fmt, __VA_ARGS__)
++#define EFRM_WARN(fmt, ...) EFRM_PRINTK(KERN_WARNING, fmt, __VA_ARGS__)
++#define EFRM_NOTICE(fmt, ...) EFRM_PRINTK(KERN_NOTICE, fmt, __VA_ARGS__)
++#if 0 && !defined(NDEBUG)
++#define EFRM_TRACE(fmt, ...) EFRM_PRINTK(KERN_DEBUG, fmt, __VA_ARGS__)
++#else
++#define EFRM_TRACE(fmt, ...)
++#endif
++
++#ifndef NDEBUG
++#define EFRM_ASSERT(cond) BUG_ON((cond) == 0)
++#define _EFRM_ASSERT(cond, file, line) \
++ do { \
++ if (unlikely(!(cond))) { \
++ EFRM_ERR("assertion \"%s\" failed at %s %d", \
++ #cond, file, line); \
++ BUG(); \
++ } \
++ } while (0)
++
++#define EFRM_DO_DEBUG(expr) expr
++#define EFRM_VERIFY_EQ(expr, val) EFRM_ASSERT((expr) == (val))
++#else
++#define EFRM_ASSERT(cond)
++#define EFRM_DO_DEBUG(expr)
++#define EFRM_VERIFY_EQ(expr, val) expr
++#endif
++
++#endif /* __CI_EFRM_DEBUG_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/driver_private.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/driver_private.h 2008-05-19 00:33:29.409841344 +0300
+@@ -0,0 +1,86 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides private API of efrm library to be used from the SFC
++ * resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_DRIVER_PRIVATE_H__
++#define __CI_EFRM_DRIVER_PRIVATE_H__
++
++#include <ci/efrm/resource.h>
++#include <ci/efrm/sysdep.h>
++
++/*--------------------------------------------------------------------
++ *
++ * global variables
++ *
++ *--------------------------------------------------------------------*/
++
++/* Internal structure for resource driver */
++extern struct efrm_resource_manager *efrm_rm_table[];
++
++/*--------------------------------------------------------------------
++ *
++ * efrm_nic_table handling
++ *
++ *--------------------------------------------------------------------*/
++
++extern int efrm_driver_ctor(void);
++extern int efrm_driver_dtor(void);
++extern int efrm_driver_register_nic(struct efhw_nic *, int nic_index);
++extern int efrm_driver_unregister_nic(struct efhw_nic *);
++
++/*--------------------------------------------------------------------
++ *
++ * create/destroy resource managers
++ *
++ *--------------------------------------------------------------------*/
++
++struct vi_resource_dimensions {
++ unsigned evq_int_min, evq_int_max;
++ unsigned evq_timer_min, evq_timer_max;
++ unsigned rxq_min, rxq_max;
++ unsigned txq_min, txq_max;
++};
++
++/*! Initialise resources */
++extern int
++efrm_resources_init(const struct vi_resource_dimensions *,
++ int buffer_table_min, int buffer_table_max);
++
++/*! Tear down resources */
++extern void efrm_resources_fini(void);
++
++#endif /* __CI_EFRM_DRIVER_PRIVATE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/filter.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/filter.h 2008-05-19 00:33:29.409841344 +0300
+@@ -0,0 +1,147 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API for filter resource.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_FILTER_H__
++#define __CI_EFRM_FILTER_H__
++
++#include <ci/efrm/resource.h>
++#include <ci/efrm/vi_resource.h>
++#include <ci/efrm/nic_set.h>
++#include <ci/efhw/common.h>
++
++/*! Comment? */
++struct filter_resource {
++ struct efrm_resource rs;
++ struct vi_resource *pt;
++ int filter_idx;
++ efrm_nic_set_t nic_set;
++};
++
++#define filter_resource(rs1) container_of((rs1), struct filter_resource, rs)
++
++/*!
++ * Allocate filter resource.
++ *
++ * \param vi_parent VI resource to use as parent. The function takes
++ * reference to the VI resource on success.
++ * \param frs_out pointer to return the new filter resource
++ *
++ * \return status code; if non-zero, frs_out is unchanged
++ */
++extern int
++efrm_filter_resource_alloc(struct vi_resource *vi_parent,
++ struct filter_resource **frs_out);
++
++/* efrm_filter_resource_free should be called only if
++ * __efrm_resource_ref_count_zero() returned true.
++ * The easiest way is to call efrm_filter_resource_release() */
++void efrm_filter_resource_free(struct filter_resource *frs);
++static inline void efrm_filter_resource_release(struct filter_resource *frs)
++{
++ unsigned id;
++
++ EFRM_RESOURCE_ASSERT_VALID(&frs->rs, 0);
++ id = EFRM_RESOURCE_INSTANCE(frs->rs.rs_handle);
++
++ if (atomic_dec_and_test(&frs->rs.rs_ref_count)) {
++ if (__efrm_resource_ref_count_zero(EFRM_RESOURCE_FILTER, id)) {
++ EFRM_ASSERT(EFRM_RESOURCE_INSTANCE(frs->rs.rs_handle) ==
++ id);
++ efrm_filter_resource_free(frs);
++ }
++ }
++}
++
++/*--------------------------------------------------------------------
++ *!
++ * Called to set/change the PT endpoint of a filter
++ *
++ * Example of use is TCP helper when it finds a wildcard IP filter
++ * needs to change which application it delivers traffic to
++ *
++ * \param frs filter resource
++ * \param pt_handle handle of new PT endpoint
++ *
++ * \return standard error codes
++ *
++ *--------------------------------------------------------------------*/
++extern int
++efrm_filter_resource_set_ptresource(struct filter_resource *frs,
++ struct vi_resource *virs);
++
++extern int efrm_filter_resource_clear(struct filter_resource *frs);
++
++extern int __efrm_filter_resource_set(struct filter_resource *frs, int type,
++ unsigned saddr_be32, uint16_t sport_be16,
++ unsigned daddr_be32, uint16_t dport_be16);
++
++static inline int
++efrm_filter_resource_tcp_set(struct filter_resource *frs,
++ unsigned saddr, uint16_t sport,
++ unsigned daddr, uint16_t dport)
++{
++ int type;
++
++ EFRM_ASSERT((saddr && sport) || (!saddr && !sport));
++
++ type =
++ saddr ? EFHW_IP_FILTER_TYPE_TCP_FULL :
++ EFHW_IP_FILTER_TYPE_TCP_WILDCARD;
++
++ return __efrm_filter_resource_set(frs, type,
++ saddr, sport, daddr, dport);
++}
++
++static inline int
++efrm_filter_resource_udp_set(struct filter_resource *frs,
++ unsigned saddr, uint16_t sport,
++ unsigned daddr, uint16_t dport)
++{
++ int type;
++
++ EFRM_ASSERT((saddr && sport) || (!saddr && !sport));
++
++ type =
++ saddr ? EFHW_IP_FILTER_TYPE_UDP_FULL :
++ EFHW_IP_FILTER_TYPE_UDP_WILDCARD;
++
++ return __efrm_filter_resource_set(frs,
++ type, saddr, sport, daddr, dport);
++}
++
++#endif /* __CI_EFRM_FILTER_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/iobufset.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/iobufset.h 2008-05-19 00:33:29.409841344 +0300
+@@ -0,0 +1,123 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API for iobufset resource.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_IOBUFSET_H__
++#define __CI_EFRM_IOBUFSET_H__
++
++#include <ci/efrm/vi_resource.h>
++
++/*! Iobufset resource structture.
++ * Users should not access the structure fields directly, but use the API
++ * below.
++ * However, this structure should not be moved out of public headers,
++ * because part of API (ex. efrm_iobufset_dma_addr function) is inline and
++ * is used in the fast-path code.
++ */
++struct iobufset_resource {
++ struct efrm_resource rs;
++ struct vi_resource *evq;
++ struct efhw_buffer_table_allocation buf_tbl_alloc;
++ unsigned int faultonaccess;
++ unsigned int n_bufs;
++ unsigned int pages_per_contiguous_chunk;
++ unsigned order;
++ efhw_iopage_t bufs[1];
++ /*!< up to n_bufs can follow this, so this must be the last member */
++};
++
++#define iobufset_resource(rs1) \
++ container_of((rs1), struct iobufset_resource, rs)
++
++/*!
++ * Allocate iobufset resource.
++ *
++ * \param vi_evq VI resource to use. The function takes
++ * reference to the VI resource on success.
++ * \param iobrs_out pointer to return the new filter resource
++ *
++ * \return status code; if non-zero, frs_out is unchanged
++ */
++extern int
++efrm_iobufset_resource_alloc(int32_t n_pages,
++ int32_t pages_per_contiguous_chunk,
++ struct vi_resource *vi_evq,
++ bool phys_addr_mode,
++ uint32_t faultonaccess,
++ struct iobufset_resource **iobrs_out);
++
++/* efrm_iobufset_resource_free should be called only if
++ * __efrm_resource_ref_count_zero() returned true.
++ * The easiest way is to call efrm_iobufset_resource_release() */
++void efrm_iobufset_resource_free(struct iobufset_resource *rs);
++static inline void
++efrm_iobufset_resource_release(struct iobufset_resource *iobrs)
++{
++ unsigned id;
++
++ EFRM_RESOURCE_ASSERT_VALID(&iobrs->rs, 0);
++ id = EFRM_RESOURCE_INSTANCE(iobrs->rs.rs_handle);
++
++ if (atomic_dec_and_test(&iobrs->rs.rs_ref_count)) {
++ if (__efrm_resource_ref_count_zero(EFRM_RESOURCE_IOBUFSET, id))
++ efrm_iobufset_resource_free(iobrs);
++ }
++}
++
++static inline char *
++efrm_iobufset_ptr(struct iobufset_resource *rs, unsigned offs)
++{
++ EFRM_ASSERT(offs < (unsigned)(rs->n_bufs << PAGE_SHIFT));
++ return efhw_iopage_ptr(&rs->bufs[offs >> PAGE_SHIFT])
++ + (offs & (PAGE_SIZE - 1));
++}
++
++static inline char *efrm_iobufset_page_ptr(struct iobufset_resource *rs,
++ unsigned page_i)
++{
++ EFRM_ASSERT(page_i < (unsigned)rs->n_bufs);
++ return efhw_iopage_ptr(&rs->bufs[page_i]);
++}
++
++static inline dma_addr_t
++efrm_iobufset_dma_addr(struct iobufset_resource *rs, unsigned offs)
++{
++ EFRM_ASSERT(offs < (unsigned)(rs->n_bufs << PAGE_SHIFT));
++ return efhw_iopage_dma_addr(&rs->bufs[offs >> PAGE_SHIFT])
++ + (offs & (PAGE_SIZE - 1));
++}
++
++#endif /* __CI_EFRM_IOBUFSET_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/nic_set.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/nic_set.h 2008-05-19 00:33:29.409841344 +0300
+@@ -0,0 +1,104 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API for NIC sets.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_NIC_SET_H__
++#define __CI_EFRM_NIC_SET_H__
++
++#include <ci/efrm/debug.h>
++#include <ci/efhw/common_sysdep.h>
++#include <ci/efhw/efhw_config.h>
++
++/*--------------------------------------------------------------------
++ *
++ * efrm_nic_set_t - tracks which NICs something has been done on
++ *
++ *--------------------------------------------------------------------*/
++
++/* Internal suructure of efrm_nic_set_t should not be referenced outside of
++ * this file. Add a new accessor if you should do it. */
++typedef struct {
++ uint32_t nics;
++} efrm_nic_set_t;
++
++#if EFHW_MAX_NR_DEVS > 32
++#error change efrm_nic_set to handle EFHW_MAX_NR_DEVS number of devices
++#endif
++
++static inline bool
++efrm_nic_set_read(const efrm_nic_set_t *nic_set, unsigned index)
++{
++ EFRM_ASSERT(nic_set);
++ EFRM_ASSERT(index < EFHW_MAX_NR_DEVS && index < 32);
++ return (nic_set->nics & (1 << index)) ? true : false;
++}
++
++static inline void
++efrm_nic_set_write(efrm_nic_set_t *nic_set, unsigned index, bool value)
++{
++ EFRM_ASSERT(nic_set);
++ EFRM_ASSERT(index < EFHW_MAX_NR_DEVS && index < 32);
++ EFRM_ASSERT(value == false || value == true);
++ nic_set->nics = (nic_set->nics & (~(1 << index))) + (value << index);
++}
++
++static inline void efrm_nic_set_clear(efrm_nic_set_t *nic_set)
++{
++ nic_set->nics = 0;
++}
++
++static inline void efrm_nic_set_all(efrm_nic_set_t *nic_set)
++{
++ nic_set->nics = 0xffffffff;
++}
++
++static inline bool efrm_nic_set_is_all_clear(efrm_nic_set_t *nic_set)
++{
++ return nic_set->nics == 0 ? true : false;
++}
++
++#define EFRM_NIC_SET_FMT "%x"
++
++static inline uint32_t efrm_nic_set_pri_arg(efrm_nic_set_t *nic_set)
++{
++ return nic_set->nics;
++}
++
++#define EFRM_FOR_EACH_NIC_INDEX_IN_SET(_set, _nic_i) \
++ for ((_nic_i) = 0; (_nic_i) < EFHW_MAX_NR_DEVS; ++(_nic_i)) \
++ if (efrm_nic_set_read((_set), (_nic_i)))
++
++#endif /* __CI_EFRM_NIC_SET_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/nic_table.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/nic_table.h 2008-05-19 00:33:29.413841574 +0300
+@@ -0,0 +1,98 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API for NIC table.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_NIC_TABLE_H__
++#define __CI_EFRM_NIC_TABLE_H__
++
++#include <ci/efhw/efhw_types.h>
++#include <ci/efrm/sysdep.h>
++
++/*--------------------------------------------------------------------
++ *
++ * struct efrm_nic_table - top level driver object keeping all NICs -
++ * implemented in driver_object.c
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Comment? */
++struct efrm_nic_table {
++ /*! nics attached to this driver */
++ struct efhw_nic *nic[EFHW_MAX_NR_DEVS];
++ /*! pointer to an arbitrary struct efhw_nic if one exists;
++ * for code which does not care which NIC it wants but
++ * still needs one. Note you cannot assume nic[0] exists. */
++ struct efhw_nic *a_nic;
++ uint32_t nic_count; /*!< number of nics attached to this driver */
++ spinlock_t lock; /*!< lock for table modifications */
++ atomic_t ref_count; /*!< refcount for users of nic table */
++};
++
++/* Resource driver structures used by other drivers as well */
++extern struct efrm_nic_table efrm_nic_table;
++
++static inline void efrm_nic_table_hold(void)
++{
++ atomic_inc(&efrm_nic_table.ref_count);
++}
++
++static inline void efrm_nic_table_rele(void)
++{
++ atomic_dec(&efrm_nic_table.ref_count);
++}
++
++static inline int efrm_nic_table_held(void)
++{
++ return (atomic_read(&efrm_nic_table.ref_count) != 0);
++}
++
++/* Run code block _x multiple times with variable nic set to each
++ * registered NIC in turn.
++ * DO NOT "break" out of this loop early. */
++#define EFRM_FOR_EACH_NIC(_nic_i, _nic) \
++ for ((_nic_i) = (efrm_nic_table_hold(), 0); \
++ (_nic_i) < EFHW_MAX_NR_DEVS || (efrm_nic_table_rele(), 0); \
++ (_nic_i)++) \
++ if (((_nic) = efrm_nic_table.nic[_nic_i]))
++
++#define EFRM_FOR_EACH_NIC_IN_SET(_set, _i, _nic) \
++ for ((_i) = (efrm_nic_table_hold(), 0); \
++ (_i) < EFHW_MAX_NR_DEVS || (efrm_nic_table_rele(), 0); \
++ ++(_i)) \
++ if (((_nic) = efrm_nic_table.nic[_i]) && \
++ efrm_nic_set_read((_set), (_i)))
++
++#endif /* __CI_EFRM_NIC_TABLE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/private.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/private.h 2008-05-19 00:33:29.413841574 +0300
+@@ -0,0 +1,141 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides private API of efrm library -- resource handling.
++ * This API is not designed for use outside of SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_PRIVATE_H__
++#define __CI_EFRM_PRIVATE_H__
++
++#include <ci/efrm/resource.h>
++#include <ci/efrm/driver_private.h>
++#include <ci/efrm/sysdep.h>
++#include <ci/efrm/debug.h>
++
++/*--------------------------------------------------------------------
++ *
++ * create resource managers
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Create a resource manager for various types of resources
++ */
++extern int
++efrm_create_iobufset_resource_manager(struct efrm_resource_manager **out);
++
++extern int
++efrm_create_filter_resource_manager(struct efrm_resource_manager **out);
++
++extern int
++efrm_create_vi_resource_manager(struct efrm_resource_manager **out,
++ const struct vi_resource_dimensions *);
++
++/*--------------------------------------------------------------------
++ *
++ * efrm_resource_handle_t handling
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Initialize an area of memory to be used as a resource */
++static inline void efrm_resource_init(struct efrm_resource *rs,
++ int type, int instance)
++{
++ EFRM_ASSERT(instance >= 0);
++ EFRM_ASSERT(type >= 0 && type < EFRM_RESOURCE_NUM);
++ atomic_set(&rs->rs_ref_count, 1);
++ rs->rs_handle.handle = (type << 28u) |
++ (((unsigned)jiffies & 0xfff) << 16) | instance;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Instance pool management
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Allocate instance pool. Use kfifo_vfree to destroy it. */
++static inline int
++efrm_kfifo_id_ctor(struct kfifo **ids_out,
++ unsigned int base, unsigned int limit, spinlock_t *lock)
++{
++ unsigned int i;
++ struct kfifo *ids;
++ unsigned char *buffer;
++ unsigned int size = roundup_pow_of_two((limit - base) * sizeof(int));
++
++ EFRM_ASSERT(base <= limit);
++ buffer = vmalloc(size);
++ ids = kfifo_init(buffer, size, GFP_KERNEL, lock);
++ if (IS_ERR(ids))
++ return PTR_ERR(ids);
++ for (i = base; i < limit; i++)
++ EFRM_VERIFY_EQ(__kfifo_put(ids, (unsigned char *)&i,
++ sizeof(i)), sizeof(i));
++
++ *ids_out = ids;
++ return 0;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Various private functions
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Initialize the fields in the provided resource manager memory area
++ * \param rm The area of memory to be initialized
++ * \param dtor A method to destroy the resource manager
++ * \param name A Textual name for the resource manager
++ * \param type The type of resource managed
++ * \param initial_table_size Initial size of the ID table
++ * \param auto_destroy Destroy resource manager on driver onload iff true
++ *
++ * A default table size is provided if the value 0 is provided.
++ */
++extern int
++efrm_resource_manager_ctor(struct efrm_resource_manager *rm,
++ void (*dtor)(struct efrm_resource_manager *),
++ const char *name, unsigned type,
++ int initial_table_size);
++
++extern void efrm_resource_manager_dtor(struct efrm_resource_manager *rm);
++
++/*! Insert a resource into table in the resource manager.
++ *
++ * Caller should free the resource if this function returns non-zero.
++ */
++extern int efrm_resource_manager_insert(struct efrm_resource *rs);
++
++#endif /* __CI_EFRM_PRIVATE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/resource.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/resource.h 2008-05-19 00:33:29.413841574 +0300
+@@ -0,0 +1,122 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public interface of efrm library -- resource handling.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_RESOURCE_H__
++#define __CI_EFRM_RESOURCE_H__
++
++/*--------------------------------------------------------------------
++ *
++ * headers for type dependencies
++ *
++ *--------------------------------------------------------------------*/
++
++#include <ci/efhw/efhw_types.h>
++#include <ci/efrm/resource_id.h>
++#include <ci/efrm/sysdep.h>
++#include <ci/efhw/common_sysdep.h>
++
++#ifndef __ci_driver__
++#error "Driver-only file"
++#endif
++
++/*--------------------------------------------------------------------
++ *
++ * struct efrm_resource - represents an allocated resource
++ * (eg. pinned pages of memory, or resource on a NIC)
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Representation of an allocated resource */
++struct efrm_resource {
++ atomic_t rs_ref_count; /*!< users count; see
++ * __efrm_resource_ref_count_zero() */
++ efrm_resource_handle_t rs_handle;
++};
++
++/*--------------------------------------------------------------------
++ *
++ * managed resource abstraction
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Factory for resources of a specific type */
++struct efrm_resource_manager {
++ const char *rm_name; /*!< human readable only */
++ spinlock_t rm_lock;
++#ifndef NDEBUG
++ unsigned rm_type;
++#endif
++ int rm_resources;
++ int rm_resources_hiwat;
++ /*! table of allocated resources */
++ struct efrm_resource **rm_table;
++ unsigned rm_table_size;
++ /**
++ * Destructor for the resource manager. Other resource managers
++ * might be already dead, although the system guarantees that
++ * managers are destructed in the order by which they were created
++ */
++ void (*rm_dtor)(struct efrm_resource_manager *);
++};
++
++#ifdef NDEBUG
++# define EFRM_RESOURCE_ASSERT_VALID(rs, rc_mbz)
++# define EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm)
++#else
++/*! Check validity of resource and report on failure */
++extern void efrm_resource_assert_valid(struct efrm_resource *,
++ int rc_may_be_zero,
++ const char *file, int line);
++# define EFRM_RESOURCE_ASSERT_VALID(rs, rc_mbz) \
++ efrm_resource_assert_valid((rs), (rc_mbz), __FILE__, __LINE__)
++
++/*! Check validity of resource manager and report on failure */
++extern void efrm_resource_manager_assert_valid(struct efrm_resource_manager *,
++ const char *file, int line);
++# define EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm) \
++ efrm_resource_manager_assert_valid((rm), __FILE__, __LINE__)
++#endif
++
++/*! Check the reference count on the resource provided and delete its
++ * handle it in its owning resource manager if the
++ * reference count has fallen to zero.
++ *
++ * Returns TRUE if the caller should really free the resource.
++ */
++extern bool __efrm_resource_ref_count_zero(unsigned type, unsigned instance);
++
++#endif /* __CI_EFRM_RESOURCE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/resource_id.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/resource_id.h 2008-05-19 00:33:29.413841574 +0300
+@@ -0,0 +1,104 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public type and definitions resource handle, and the
++ * definitions of resource types.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_DRIVER_EFRM_RESOURCE_ID_H__
++#define __CI_DRIVER_EFRM_RESOURCE_ID_H__
++
++/***********************************************************************
++ * Resource handles
++ *
++ * Resource handles are intended for identifying resources at kernel
++ * level, within the context of a particular NIC. particularly because
++ * for some resource types, the low 16 bites correspond to hardware
++ * IDs. They were historically also used at user level, with a nonce
++ * stored in the bits 16 to 27 (inclusive), but that approach is
++ * deprecated (but sill alive!).
++ *
++ * The handle value 0 is used to mean "no resource".
++ * Identify resources within the context of a file descriptor at user
++ * level.
++ ***********************************************************************/
++
++typedef struct efrm_resource_handle_s {
++ uint32_t handle;
++} efrm_resource_handle_t;
++
++/* You may think these following functions should all have
++ * _HANDLE_ in their names, but really we are providing an abstract set
++ * of methods on a (hypothetical) efrm_resource_t object, with
++ * efrm_resource_handle_t being just the reference one holds to access
++ * the object (aka "this" or "self").
++ */
++
++/* Below I use inline instead of macros where possible in order to get
++ * more type checking help from the compiler; hopefully we'll never
++ * have to rewrite these to use #define as we've found some horrible
++ * compiler on which we cannot make static inline do the Right Thing (tm).
++ *
++ * For consistency and to avoid pointless change I spell these
++ * routines as macro names (CAPTILIZE_UNDERSCORED), which also serves
++ * to remind people they are compact and inlined.
++ */
++
++#define EFRM_RESOURCE_FMT "[rs:%08x]"
++
++static inline unsigned EFRM_RESOURCE_PRI_ARG(efrm_resource_handle_t h)
++{
++ return (h.handle);
++}
++
++static inline unsigned EFRM_RESOURCE_INSTANCE(efrm_resource_handle_t h)
++{
++ return (h.handle & 0x0000ffff);
++}
++
++static inline unsigned EFRM_RESOURCE_TYPE(efrm_resource_handle_t h)
++{
++ return (h.handle & 0xf0000000) >> 28;
++}
++
++/***********************************************************************
++ * Resource type codes
++ ***********************************************************************/
++
++#define EFRM_RESOURCE_IOBUFSET 0x0
++#define EFRM_RESOURCE_VI 0x1
++#define EFRM_RESOURCE_FILTER 0x2
++#define EFRM_RESOURCE_NUM 0x3 /* This isn't a resource! */
++
++#define EFRM_RESOURCE_NAME(type) \
++ ((type) == EFRM_RESOURCE_IOBUFSET? "IOBUFSET" : \
++ (type) == EFRM_RESOURCE_VI? "VI" : \
++ (type) == EFRM_RESOURCE_FILTER? "FILTER" : \
++ "<invalid>")
++
++#endif /* __CI_DRIVER_EFRM_RESOURCE_ID_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/sysdep.h 2008-05-19 00:33:29.413841574 +0300
+@@ -0,0 +1,54 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides Linux-like system-independent API for efrm library.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_SYSDEP_H__
++#define __CI_EFRM_SYSDEP_H__
++
++/* Spinlocks are defined in efhw/sysdep.h */
++#include <ci/efhw/sysdep.h>
++
++#if defined(__linux__) && defined(__KERNEL__)
++
++# include <ci/efrm/sysdep_linux.h>
++
++#else
++
++# include <ci/efrm/sysdep_ci2linux.h>
++
++#endif
++
++#endif /* __CI_EFRM_SYSDEP_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/sysdep_linux.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/sysdep_linux.h 2008-05-19 00:33:29.417841805 +0300
+@@ -0,0 +1,248 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for efrm library.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Kfifo API is partially stolen from linux-2.6.22/include/linux/list.h
++ * Copyright (C) 2004 Stelian Pop <stelian@popies.net>
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_SYSDEP_LINUX_H__
++#define __CI_EFRM_SYSDEP_LINUX_H__
++
++#include <linux/version.h>
++#include <linux/list.h>
++#include <linux/vmalloc.h>
++#include <linux/errno.h>
++#include <linux/string.h>
++#include <linux/workqueue.h>
++#include <linux/gfp.h>
++#include <linux/slab.h>
++#include <linux/hardirq.h>
++#include <linux/kernel.h>
++#include <linux/if_ether.h>
++#include <linux/completion.h>
++#include <linux/in.h>
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++/* get roundup_pow_of_two(), which was in kernel.h in early kernel versions */
++#include <linux/log2.h>
++#endif
++
++/********************************************************************
++ *
++ * List API
++ *
++ ********************************************************************/
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18)
++static inline void
++list_replace_init(struct list_head *old, struct list_head *new)
++{
++ new->next = old->next;
++ new->next->prev = new;
++ new->prev = old->prev;
++ new->prev->next = new;
++ INIT_LIST_HEAD(old);
++}
++#endif
++
++static inline struct list_head *list_pop(struct list_head *list)
++{
++ struct list_head *link = list->next;
++ list_del(link);
++ return link;
++}
++
++static inline struct list_head *list_pop_tail(struct list_head *list)
++{
++ struct list_head *link = list->prev;
++ list_del(link);
++ return link;
++}
++
++/********************************************************************
++ *
++ * Workqueue API
++ *
++ ********************************************************************/
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)
++#define NEED_OLD_WORK_API
++
++/**
++ * The old and new work function prototypes just change
++ * the type of the pointer in the only argument, so it's
++ * safe to cast one function type to the other
++ */
++typedef void (*efrm_old_work_func_t) (void *p);
++
++#undef INIT_WORK
++#define INIT_WORK(_work, _func) \
++ do { \
++ INIT_LIST_HEAD(&(_work)->entry); \
++ (_work)->pending = 0; \
++ PREPARE_WORK((_work), \
++ (efrm_old_work_func_t) (_func), \
++ (_work)); \
++ } while (0)
++
++#endif
++
++/********************************************************************
++ *
++ * Kfifo API
++ *
++ ********************************************************************/
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,10)
++
++#if !defined(RHEL_RELEASE_CODE) || (RHEL_RELEASE_CODE < 1029)
++typedef unsigned gfp_t;
++#endif
++
++#define HAS_NO_KFIFO
++
++struct kfifo {
++ unsigned char *buffer; /* the buffer holding the data */
++ unsigned int size; /* the size of the allocated buffer */
++ unsigned int in; /* data is added at offset (in % size) */
++ unsigned int out; /* data is extracted from off. (out % size) */
++ spinlock_t *lock; /* protects concurrent modifications */
++};
++
++extern struct kfifo *kfifo_init(unsigned char *buffer, unsigned int size,
++ gfp_t gfp_mask, spinlock_t *lock);
++extern struct kfifo *kfifo_alloc(unsigned int size, gfp_t gfp_mask,
++ spinlock_t *lock);
++extern void kfifo_free(struct kfifo *fifo);
++extern unsigned int __kfifo_put(struct kfifo *fifo,
++ unsigned char *buffer, unsigned int len);
++extern unsigned int __kfifo_get(struct kfifo *fifo,
++ unsigned char *buffer, unsigned int len);
++
++/**
++ * kfifo_put - puts some data into the FIFO
++ * @fifo: the fifo to be used.
++ * @buffer: the data to be added.
++ * @len: the length of the data to be added.
++ *
++ * This function copies at most @len bytes from the @buffer into
++ * the FIFO depending on the free space, and returns the number of
++ * bytes copied.
++ */
++static inline unsigned int
++kfifo_put(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
++{
++ unsigned long flags;
++ unsigned int ret;
++
++ spin_lock_irqsave(fifo->lock, flags);
++
++ ret = __kfifo_put(fifo, buffer, len);
++
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
++}
++
++/**
++ * kfifo_get - gets some data from the FIFO
++ * @fifo: the fifo to be used.
++ * @buffer: where the data must be copied.
++ * @len: the size of the destination buffer.
++ *
++ * This function copies at most @len bytes from the FIFO into the
++ * @buffer and returns the number of copied bytes.
++ */
++static inline unsigned int
++kfifo_get(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
++{
++ unsigned long flags;
++ unsigned int ret;
++
++ spin_lock_irqsave(fifo->lock, flags);
++
++ ret = __kfifo_get(fifo, buffer, len);
++
++ /*
++ * optimization: if the FIFO is empty, set the indices to 0
++ * so we don't wrap the next time
++ */
++ if (fifo->in == fifo->out)
++ fifo->in = fifo->out = 0;
++
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
++}
++
++/**
++ * __kfifo_len - returns the number of bytes available in the FIFO, no locking version
++ * @fifo: the fifo to be used.
++ */
++static inline unsigned int __kfifo_len(struct kfifo *fifo)
++{
++ return fifo->in - fifo->out;
++}
++
++/**
++ * kfifo_len - returns the number of bytes available in the FIFO
++ * @fifo: the fifo to be used.
++ */
++static inline unsigned int kfifo_len(struct kfifo *fifo)
++{
++ unsigned long flags;
++ unsigned int ret;
++
++ spin_lock_irqsave(fifo->lock, flags);
++
++ ret = __kfifo_len(fifo);
++
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
++}
++
++#else
++#include <linux/kfifo.h>
++#endif
++
++static inline void kfifo_vfree(struct kfifo *fifo)
++{
++ vfree(fifo->buffer);
++ kfree(fifo);
++}
++
++#endif /* __CI_EFRM_SYSDEP_LINUX_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource.h 2008-05-19 00:33:29.417841805 +0300
+@@ -0,0 +1,171 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains public API for VI resource.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_VI_RESOURCE_H__
++#define __CI_EFRM_VI_RESOURCE_H__
++
++#include <ci/efhw/efhw_types.h>
++#include <ci/efrm/resource.h>
++#include <ci/efrm/debug.h>
++
++struct vi_resource;
++
++/* Make these inline instead of macros for type checking */
++static inline struct vi_resource *
++efrm_to_vi_resource(struct efrm_resource *rs)
++{
++ EFRM_ASSERT(EFRM_RESOURCE_TYPE(rs->rs_handle) == EFRM_RESOURCE_VI);
++ return (struct vi_resource *) rs;
++}
++static inline struct
++efrm_resource *efrm_from_vi_resource(struct vi_resource *rs)
++{
++ return (struct efrm_resource *)rs;
++}
++
++#define EFAB_VI_RESOURCE_INSTANCE(virs) \
++ EFRM_RESOURCE_INSTANCE(efrm_from_vi_resource(virs)->rs_handle)
++
++#define EFAB_VI_RESOURCE_PRI_ARG(virs) \
++ EFRM_RESOURCE_PRI_ARG(efrm_from_vi_resource(virs)->rs_handle)
++
++extern int
++efrm_vi_resource_alloc(struct vi_resource *evq_virs,
++ uint16_t vi_flags, int32_t evq_capacity,
++ int32_t txq_capacity, int32_t rxq_capacity,
++ uint8_t tx_q_tag, uint8_t rx_q_tag,
++ struct vi_resource **virs_in_out,
++ uint32_t *out_io_mmap_bytes,
++ uint32_t *out_mem_mmap_bytes,
++ uint32_t *out_txq_capacity,
++ uint32_t *out_rxq_capacity);
++
++static inline void efrm_vi_resource_ref(struct vi_resource *virs)
++{
++ atomic_inc(&efrm_from_vi_resource(virs)->rs_ref_count);
++}
++
++/* efrm_vi_resource_free should be called only if
++ * __efrm_resource_ref_count_zero() returned true.
++ * The easiest way is to call efrm_vi_resource_release() */
++extern void efrm_vi_resource_free(struct vi_resource *virs);
++static inline void efrm_vi_resource_release(struct vi_resource *virs)
++{
++ unsigned id;
++ struct efrm_resource *rs = efrm_from_vi_resource(virs);
++
++ id = EFRM_RESOURCE_INSTANCE(rs->rs_handle);
++
++ if (atomic_dec_and_test(&rs->rs_ref_count)) {
++ if (__efrm_resource_ref_count_zero(EFRM_RESOURCE_VI, id)) {
++ EFRM_ASSERT(EFRM_RESOURCE_INSTANCE(rs->rs_handle) ==
++ id);
++ efrm_vi_resource_free(virs);
++ }
++ }
++}
++
++/*--------------------------------------------------------------------
++ *
++ * eventq handling
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Reset an event queue and clear any associated timers */
++extern void efrm_eventq_reset(struct vi_resource *virs, int nic_index);
++
++/*! Register a kernel-level handler for the event queue. This function is
++ * called whenever a timer expires, or whenever the event queue is woken
++ * but no thread is blocked on it.
++ *
++ * This function returns -EBUSY if a callback is already installed.
++ *
++ * \param rs Event-queue resource
++ * \param handler Callback-handler
++ * \param arg Argument to pass to callback-handler
++ * \return Status code
++ */
++extern int
++efrm_eventq_register_callback(struct vi_resource *rs,
++ void (*handler)(void *arg, int is_timeout,
++ struct efhw_nic *nic),
++ void *arg);
++
++/*! Kill the kernel-level callback.
++ *
++ * This function stops the timer from running and unregisters the callback
++ * function. It waits for any running timeout handlers to complete before
++ * returning.
++ *
++ * \param rs Event-queue resource
++ * \return Nothing
++ */
++extern void efrm_eventq_kill_callback(struct vi_resource *rs);
++
++/*! Ask the NIC to generate a wakeup when an event is next delivered. */
++extern void efrm_eventq_request_wakeup(struct vi_resource *rs,
++ unsigned current_ptr,
++ unsigned nic_index);
++
++/*! Register a kernel-level handler for flush completions.
++ * \TODO Currently, it is unsafe to install a callback more than once.
++ *
++ * \param rs VI resource being flushed.
++ * \param handler Callback handler function.
++ * \param arg Argument to be passed to handler.
++ */
++extern void
++efrm_vi_register_flush_callback(struct vi_resource *rs,
++ void (*handler)(void *),
++ void *arg);
++
++int efrm_vi_resource_flush_retry(struct vi_resource *virs);
++
++/*! Comment? */
++extern int efrm_pt_flush(struct vi_resource *);
++
++/*! Comment? */
++extern int efrm_pt_pace(struct vi_resource *, unsigned int val);
++
++uint32_t efrm_vi_rm_txq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ );
++uint32_t efrm_vi_rm_rxq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ );
++uint32_t efrm_vi_rm_evq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ );
++
++#endif /* __CI_EFRM_VI_RESOURCE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource_manager.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource_manager.h 2008-05-19 00:33:29.417841805 +0300
+@@ -0,0 +1,182 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains type definitions for VI resource. These types
++ * may be used outside of the SFC resource driver, but such use is not
++ * recommended.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_DRIVER_EFAB_VI_RESOURCE_MANAGER_H__
++#define __CI_DRIVER_EFAB_VI_RESOURCE_MANAGER_H__
++
++#include <ci/efhw/common.h>
++#include <ci/efrm/vi_resource.h>
++#include <ci/efrm/nic_set.h>
++
++#define EFRM_VI_RM_DMA_QUEUE_COUNT 2
++#define EFRM_VI_RM_DMA_QUEUE_TX 0
++#define EFRM_VI_RM_DMA_QUEUE_RX 1
++
++/** Numbers of bits which can be set in the evq_state member of
++ * vi_resource_evq_info. */
++enum {
++ /** This bit is set if a wakeup has been requested on the NIC. */
++ VI_RESOURCE_EVQ_STATE_WAKEUP_PENDING,
++ /** This bit is set if the wakeup is valid for the sleeping
++ * process. */
++ VI_RESOURCE_EVQ_STATE_CALLBACK_REGISTERED,
++ /** This bit is set if a wakeup or timeout event is currently being
++ * processed. */
++ VI_RESOURCE_EVQ_STATE_BUSY,
++};
++#define VI_RESOURCE_EVQ_STATE(X) \
++ (((int32_t)1) << (VI_RESOURCE_EVQ_STATE_##X))
++
++/** Information about an event queue. */
++struct vi_resource_evq_info {
++ /** Flag bits indicating the state of wakeups. */
++ unsigned long evq_state;
++ /** A pointer to the resource instance for this queue. This member
++ * is only valid if evq_state is non-zero or the resource is known
++ * to have a non-zero reference count. */
++ struct vi_resource *evq_virs;
++};
++
++#ifdef __ci_ul_driver__
++#define EFRM_VI_USE_WORKQUEUE 0
++#else
++#define EFRM_VI_USE_WORKQUEUE 1
++#endif
++
++/*! Global information for the VI resource manager. */
++struct vi_resource_manager {
++ struct efrm_resource_manager rm;
++
++ struct kfifo *instances_with_timer;
++ int with_timer_base;
++ int with_timer_limit;
++ struct kfifo *instances_with_interrupt;
++ int with_interrupt_base;
++ int with_interrupt_limit;
++
++ bool iscsi_dmaq_instance_is_free;
++ struct vi_resource_evq_info *evq_infos;
++
++ /* We keep VI resources which need flushing on these lists. The VI
++ * is put on the outstanding list when the flush request is issued
++ * to the hardware and removed when the flush event arrives. The
++ * hardware can only handle a limited number of RX flush requests at
++ * once, so VIs are placed in the waiting list until the flush can
++ * be issued. Flushes can be requested by the client or internally
++ * by the VI resource manager. In the former case, the reference
++ * count must be non-zero for the duration of the flush and in the
++ * later case, the reference count must be zero. */
++ struct list_head rx_flush_waiting_list;
++ struct list_head rx_flush_outstanding_list;
++ struct list_head tx_flush_outstanding_list;
++ int rx_flush_outstanding_count;
++
++ /* once the flush has happened we push the close into the work queue
++ * so its OK on Windows to free the resources (Bug 3469). Resources
++ * on this list have zero reference count.
++ */
++ struct list_head close_pending;
++ struct work_struct work_item;
++#if EFRM_VI_USE_WORKQUEUE
++ struct workqueue_struct *workqueue;
++#endif
++};
++
++struct vi_resource_nic_info {
++ struct eventq_resource_hardware evq_pages;
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ efhw_iopages_t dmaq_pages[EFRM_VI_RM_DMA_QUEUE_COUNT];
++#endif
++};
++
++struct vi_resource {
++ /* Some macros make the assumption that the struct efrm_resource is
++ * the first member of a struct vi_resource. */
++ struct efrm_resource rs;
++ atomic_t evq_refs; /*!< Number of users of the event queue. */
++
++ efrm_nic_set_t nic_set;
++
++ uint32_t bar_mmap_bytes;
++ uint32_t mem_mmap_bytes;
++
++ int32_t evq_capacity;
++ int32_t dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_COUNT];
++
++ uint8_t dmaq_tag[EFRM_VI_RM_DMA_QUEUE_COUNT];
++ uint16_t flags;
++
++ /* we keep PT endpoints that have been destroyed on a list
++ * until we have seen their TX and RX DMAQs flush complete
++ * (see Bug 1217)
++ */
++ struct list_head rx_flush_link;
++ struct list_head tx_flush_link;
++ efrm_nic_set_t rx_flush_nic_set;
++ efrm_nic_set_t rx_flush_outstanding_nic_set;
++ efrm_nic_set_t tx_flush_nic_set;
++ uint64_t flush_time;
++ int flush_count;
++
++ void (*flush_callback_fn)(void *);
++ void *flush_callback_arg;
++
++ void (*evq_callback_fn) (void *arg, int is_timeout,
++ struct efhw_nic *nic);
++ void *evq_callback_arg;
++
++ struct vi_resource *evq_virs; /*!< EVQ for DMA queues */
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ struct efhw_buffer_table_allocation
++ dmaq_buf_tbl_alloc[EFRM_VI_RM_DMA_QUEUE_COUNT];
++#endif
++
++ struct vi_resource_nic_info nic_info[EFHW_MAX_NR_DEVS];
++};
++
++#undef vi_resource
++#define vi_resource(rs1) container_of((rs1), struct vi_resource, rs)
++
++static inline dma_addr_t
++efrm_eventq_dma_addr(struct vi_resource *virs, uint32_t nic_index)
++{
++ struct eventq_resource_hardware *hw;
++ EFRM_ASSERT(efrm_nic_set_read(&virs->nic_set, nic_index));
++
++ hw = &(virs->nic_info[nic_index].evq_pages);
++
++ return efhw_iopages_dma_addr(&(hw->iobuff)) + hw->iobuff_off;
++}
++
++#endif /* __CI_DRIVER_EFAB_VI_RESOURCE_MANAGER_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource_private.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/ci/efrm/vi_resource_private.h 2008-05-19 00:33:29.417841805 +0300
+@@ -0,0 +1,83 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains private API for VI resource. The API is not designed
++ * to be used outside of the SFC resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFRM_VI_RESOURCE_PRIVATE_H__
++#define __CI_EFRM_VI_RESOURCE_PRIVATE_H__
++
++#include <ci/efhw/common.h>
++#include <ci/efrm/vi_resource_manager.h>
++
++extern struct vi_resource_manager *efrm_vi_manager;
++
++/*************************************************************************/
++
++extern void efrm_vi_rm_delayed_free(struct work_struct *data);
++
++extern void efrm_vi_rm_salvage_flushed_vis(void);
++
++void efrm_vi_rm_free_flushed_resource(struct vi_resource *virs);
++
++void efrm_vi_rm_init_dmaq(struct vi_resource *virs, int queue_index,
++ struct efhw_nic *nic);
++
++static inline int
++efrm_eventq_bytes(struct vi_resource *virs, uint32_t nic_index)
++{
++ EFRM_ASSERT(efrm_nic_set_read(&virs->nic_set, nic_index));
++
++ return efrm_vi_rm_evq_bytes(virs);
++}
++
++static inline efhw_event_t *
++efrm_eventq_base(struct vi_resource *virs, uint32_t nic_index)
++{
++ struct eventq_resource_hardware *hw;
++
++ EFRM_ASSERT(efrm_nic_set_read(&virs->nic_set, nic_index));
++
++ hw = &(virs->nic_info[nic_index].evq_pages);
++
++ return (efhw_event_t *) (efhw_iopages_ptr(&(hw->iobuff)) +
++ hw->iobuff_off);
++}
++
++/*! Wakeup handler, see efhw_ev_handler_t for prototype */
++extern void efrm_handle_wakeup_event(struct efhw_nic *nic, efhw_event_t *ev);
++
++/*! Timeout handler, see efhw_ev_handler_t for prototype */
++extern void efrm_handle_timeout_event(struct efhw_nic *nic, efhw_event_t *ev);
++
++/*! DMA flush handler, see efhw_ev_handler_t for prototype */
++extern void efrm_handle_dmaq_flushed(struct efhw_nic *nic, int instance,
++ int rx_flush);
++
++#endif /* __CI_EFRM_VI_RESOURCE_PRIVATE_H__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/driver_object.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/driver_object.c 2008-05-19 00:33:29.417841805 +0300
+@@ -0,0 +1,174 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains support for the global driver variables.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/efrm/resource.h>
++#include <ci/efrm/debug.h>
++
++/* We use #define rather than static inline here so that the Windows
++ * "prefast" compiler can see its own locking primitive when these
++ * two function are used (and then perform extra checking where they
++ * are used)
++ *
++ * Both macros operate on an irq_flags_t
++*/
++
++#define efrm_driver_lock(irqlock_state) \
++ spin_lock_irqsave(&efrm_nic_table.lock, irqlock_state)
++
++#define efrm_driver_unlock(irqlock_state) \
++ spin_unlock_irqrestore(&efrm_nic_table.lock, \
++ irqlock_state);
++
++/* These routines are all methods on the architecturally singleton
++ global variables: efrm_nic_table, efrm_rm_table.
++
++ I hope we never find a driver model that does not allow global
++ structure variables :) (but that would break almost every driver I've
++ ever seen).
++*/
++
++/*! Exported driver state */
++struct efrm_nic_table efrm_nic_table;
++EXPORT_SYMBOL(efrm_nic_table);
++
++/* Internal table with resource managers.
++ * We'd like to not export it, but we are still using efrm_rm_table
++ * in the char driver. So, it is declared in the private header with
++ * a purpose. */
++struct efrm_resource_manager *efrm_rm_table[EFRM_RESOURCE_NUM];
++EXPORT_SYMBOL(efrm_rm_table);
++
++int efrm_driver_ctor(void)
++{
++ memset(&efrm_nic_table, 0, sizeof(efrm_nic_table));
++ memset(&efrm_rm_table, 0, sizeof(efrm_rm_table));
++
++ spin_lock_init(&efrm_nic_table.lock);
++
++ EFRM_TRACE("%s: driver created", __FUNCTION__);
++ return 0;
++}
++
++int efrm_driver_dtor(void)
++{
++ EFRM_ASSERT(!efrm_nic_table_held());
++
++ spin_lock_destroy(&efrm_nic_table.lock);
++ EFRM_TRACE("%s: driver deleted", __FUNCTION__);
++ return 0;
++}
++
++int efrm_driver_register_nic(struct efhw_nic *nic, int nic_index)
++{
++ int rc = 0;
++ irq_flags_t lock_flags;
++
++ EFRM_ASSERT(nic_index >= 0);
++
++ efrm_driver_lock(lock_flags);
++
++ if (efrm_nic_table_held()) {
++ EFRM_WARN("%s: driver object is in use", __FUNCTION__);
++ rc = -EBUSY;
++ goto done;
++ }
++
++ if (efrm_nic_table.nic_count == EFHW_MAX_NR_DEVS) {
++ EFRM_WARN("%s: filled up NIC table size %d", __FUNCTION__,
++ EFHW_MAX_NR_DEVS);
++ rc = -E2BIG;
++ goto done;
++ }
++
++ EFRM_ASSERT(efrm_nic_table.nic[nic_index] == NULL);
++ efrm_nic_table.nic[nic_index] = nic;
++ nic->index = nic_index;
++
++ if (efrm_nic_table.a_nic == NULL)
++ efrm_nic_table.a_nic = nic;
++
++ efrm_nic_table.nic_count++;
++ efrm_driver_unlock(lock_flags);
++ return rc;
++
++done:
++ efrm_driver_unlock(lock_flags);
++ return rc;
++}
++
++int efrm_driver_unregister_nic(struct efhw_nic *nic)
++{
++ int rc = 0;
++ int nic_index = nic->index;
++ irq_flags_t lock_flags;
++
++ EFRM_ASSERT(nic_index >= 0);
++
++ efrm_driver_lock(lock_flags);
++
++ if (efrm_nic_table_held()) {
++ EFRM_WARN("%s: driver object is in use", __FUNCTION__);
++ rc = -EBUSY;
++ goto done;
++ }
++
++ EFRM_ASSERT(efrm_nic_table.nic[nic_index] == nic);
++
++ nic->index = -1;
++ efrm_nic_table.nic[nic_index] = NULL;
++
++ --efrm_nic_table.nic_count;
++
++ if (efrm_nic_table.a_nic == nic) {
++ if (efrm_nic_table.nic_count == 0) {
++ efrm_nic_table.a_nic = NULL;
++ } else {
++ for (nic_index = 0; nic_index < EFHW_MAX_NR_DEVS;
++ nic_index++) {
++ if (efrm_nic_table.nic[nic_index] != NULL)
++ efrm_nic_table.a_nic =
++ efrm_nic_table.nic[nic_index];
++ }
++ EFRM_ASSERT(efrm_nic_table.a_nic);
++ }
++ }
++
++done:
++ efrm_driver_unlock(lock_flags);
++ return rc;
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/driverlink_new.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/driverlink_new.c 2008-05-19 00:33:29.421842036 +0300
+@@ -0,0 +1,290 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains driverlink code which interacts with the sfc network
++ * driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "linux_resource_internal.h"
++#include "driverlink_api.h"
++#include "kernel_compat.h"
++#include <ci/efhw/falcon.h>
++
++#include <linux/rtnetlink.h>
++#include <linux/netdevice.h>
++
++/* The DL driver and associated calls */
++static int efrm_dl_probe(struct efx_dl_device *efrm_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char *silicon_rev);
++
++static void efrm_dl_remove(struct efx_dl_device *efrm_dev);
++
++static void efrm_dl_reset_suspend(struct efx_dl_device *efrm_dev);
++
++static void efrm_dl_reset_resume(struct efx_dl_device *efrm_dev, int ok);
++
++static void efrm_dl_mtu_changed(struct efx_dl_device *, int);
++static void efrm_dl_event_falcon(struct efx_dl_device *efx_dev, void *p_event);
++
++static struct efx_dl_driver efrm_dl_driver = {
++ .name = "resource",
++ .probe = efrm_dl_probe,
++ .remove = efrm_dl_remove,
++ .reset_suspend = efrm_dl_reset_suspend,
++ .reset_resume = efrm_dl_reset_resume
++};
++
++static void
++init_vi_resource_dimensions(struct vi_resource_dimensions *rd,
++ const struct efx_dl_falcon_resources *res)
++{
++ rd->evq_timer_min = res->evq_timer_min;
++ rd->evq_timer_max = res->evq_timer_max;
++ rd->evq_int_min = res->evq_int_min;
++ rd->evq_int_max = res->evq_int_max;
++ rd->rxq_min = res->rxq_min;
++ rd->rxq_max = res->rxq_max;
++ rd->txq_min = res->txq_min;
++ rd->txq_max = res->txq_max;
++ EFRM_TRACE
++ ("Using evq_int(%d-%d) evq_timer(%d-%d) RXQ(%d-%d) TXQ(%d-%d)",
++ res->evq_int_min, res->evq_int_max, res->evq_timer_min,
++ res->evq_timer_max, res->rxq_min, res->rxq_max, res->txq_min,
++ res->txq_max);
++}
++
++#if defined(EFX_NOT_UPSTREAM)
++/* We have a module parameter that can tell us to only load the char driver
++ * for 1 NIC (if there are multiple NICs in the system), and if so which one.
++ * This tells us the PCI bus and slot of the NIC to load for, or -1 to just
++ * load on all NICs (the default).
++ * Value is a hex number in the format
++ * bbbbss
++ * where:
++ * bbbb - PCI bus number
++ * ss - PCI slot number
++ */
++unsigned int only_NIC = -1;
++
++/** @ingroup module_params */
++module_param(only_NIC, uint, 0444);
++MODULE_PARM_DESC(only_NIC,
++ "Initialise sfc_resource driver for one NIC only, "
++ "with specified PCI bus and slot");
++#endif
++
++static int
++efrm_dl_probe(struct efx_dl_device *efrm_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char *silicon_rev)
++{
++ struct vi_resource_dimensions res_dim;
++ struct efx_dl_falcon_resources *res;
++ struct linux_efhw_nic *lnic;
++ struct pci_dev *dev;
++ struct efhw_nic *nic;
++ unsigned probe_flags = 0;
++ int rc;
++
++ efrm_dev->priv = NULL;
++
++ efx_dl_for_each_device_info_matching(dev_info, EFX_DL_FALCON_RESOURCES,
++ struct efx_dl_falcon_resources,
++ hdr, res) {
++ /* break out, leaving res pointing at the falcon resources */
++ break;
++ }
++
++ if (res == NULL) {
++ EFRM_ERR("%s: Unable to find falcon driverlink resources",
++ __FUNCTION__);
++ return -EINVAL;
++ }
++
++ if (res->flags & EFX_DL_FALCON_USE_MSI)
++ probe_flags |= NIC_FLAG_TRY_MSI;
++
++ dev = efrm_dev->pci_dev;
++ if (res->flags & EFX_DL_FALCON_DUAL_FUNC) {
++ unsigned vendor = dev->vendor;
++ EFRM_ASSERT(dev->bus != NULL);
++ dev = NULL;
++
++#if defined(EFX_NOT_UPSTREAM)
++ if (only_NIC != -1 &&
++ (efrm_dev->pci_dev->bus->number !=
++ ((only_NIC >> 8) & 0xFFFF)
++ || PCI_SLOT(efrm_dev->pci_dev->devfn) !=
++ (only_NIC & 0xFF))) {
++ EFRM_NOTICE("Hiding char device %x:%x",
++ efrm_dev->pci_dev->bus->number,
++ PCI_SLOT(efrm_dev->pci_dev->devfn));
++ return -ENODEV;
++ }
++#endif
++
++ while ((dev = pci_get_device(vendor, FALCON_S_DEVID, dev))
++ != NULL) {
++ EFRM_ASSERT(dev->bus != NULL);
++ /* With PCIe (since it's point to point)
++ * the slot ID is usually 0 and
++ * the bus ID changes NIC to NIC, so we really
++ * need to check both. */
++ if (PCI_SLOT(dev->devfn) ==
++ PCI_SLOT(efrm_dev->pci_dev->devfn)
++ && dev->bus->number ==
++ efrm_dev->pci_dev->bus->number)
++ break;
++ }
++ if (dev == NULL) {
++ EFRM_ERR("%s: Unable to find falcon secondary "
++ "PCI device.", __FUNCTION__);
++ return -ENODEV;
++ }
++ pci_dev_put(dev);
++ }
++
++ init_vi_resource_dimensions(&res_dim, res);
++
++ rc = efrm_nic_add(dev, probe_flags, net_dev->dev_addr, &lnic,
++ res->biu_lock,
++ res->buffer_table_min, res->buffer_table_max,
++ &res_dim);
++ if (rc != 0)
++ return rc;
++
++ nic = &lnic->nic;
++ nic->mtu = net_dev->mtu + ETH_HLEN;
++ nic->net_driver_dev = efrm_dev;
++ nic->ifindex = net_dev->ifindex;
++#ifdef HAS_NET_NAMESPACE
++ nic->nd_net = net_dev->nd_net;
++#endif
++ efrm_dev->priv = nic;
++
++ /* Register a callback so we're told when MTU changes.
++ * We dynamically allocate efx_dl_callbacks, because
++ * the callbacks that we want depends on the NIC type.
++ */
++ lnic->dl_callbacks =
++ kmalloc(sizeof(struct efx_dl_callbacks), GFP_KERNEL);
++ if (!lnic->dl_callbacks) {
++ EFRM_ERR("Out of memory (%s)", __FUNCTION__);
++ efrm_nic_del(lnic);
++ return -ENOMEM;
++ }
++ memset(lnic->dl_callbacks, 0, sizeof(*lnic->dl_callbacks));
++ lnic->dl_callbacks->mtu_changed = efrm_dl_mtu_changed;
++
++ if ((res->flags & EFX_DL_FALCON_DUAL_FUNC) == 0) {
++ /* Net driver receives all management events.
++ * Register a callback to receive the ones
++ * we're interested in. */
++ lnic->dl_callbacks->event = efrm_dl_event_falcon;
++ }
++
++ rc = efx_dl_register_callbacks(efrm_dev, lnic->dl_callbacks);
++ if (rc < 0) {
++ EFRM_ERR("%s: efx_dl_register_callbacks failed (%d)",
++ __FUNCTION__, rc);
++ kfree(lnic->dl_callbacks);
++ efrm_nic_del(lnic);
++ return rc;
++ }
++
++ return 0;
++}
++
++/* When we unregister ourselves on module removal, this function will be
++ * called for all the devices we claimed */
++static void efrm_dl_remove(struct efx_dl_device *efrm_dev)
++{
++ struct efhw_nic *nic = efrm_dev->priv;
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++ EFRM_TRACE("%s called", __FUNCTION__);
++ if (lnic->dl_callbacks) {
++ efx_dl_unregister_callbacks(efrm_dev, lnic->dl_callbacks);
++ kfree(lnic->dl_callbacks);
++ }
++ if (efrm_dev->priv)
++ efrm_nic_del(lnic);
++ EFRM_TRACE("%s OK", __FUNCTION__);
++}
++
++static void efrm_dl_reset_suspend(struct efx_dl_device *efrm_dev)
++{
++ EFRM_NOTICE("%s:", __FUNCTION__);
++}
++
++static void efrm_dl_reset_resume(struct efx_dl_device *efrm_dev, int ok)
++{
++ EFRM_NOTICE("%s: ok=%d", __FUNCTION__, ok);
++}
++
++int efrm_driverlink_register(void)
++{
++ EFRM_TRACE("%s:", __FUNCTION__);
++ return efx_dl_register_driver(&efrm_dl_driver);
++}
++
++void efrm_driverlink_unregister(void)
++{
++ EFRM_TRACE("%s:", __FUNCTION__);
++ efx_dl_unregister_driver(&efrm_dl_driver);
++}
++
++static void efrm_dl_mtu_changed(struct efx_dl_device *efx_dev, int mtu)
++{
++ struct efhw_nic *nic = efx_dev->priv;
++
++ ASSERT_RTNL(); /* Since we're looking at efx_dl_device::port_net_dev */
++
++ EFRM_TRACE("%s: old=%d new=%d", __FUNCTION__, nic->mtu, mtu + ETH_HLEN);
++ /* If this happened we must have agreed to it above */
++ nic->mtu = mtu + ETH_HLEN;
++}
++
++static void efrm_dl_event_falcon(struct efx_dl_device *efx_dev, void *p_event)
++{
++ struct efhw_nic *nic = efx_dev->priv;
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++ efhw_event_t *ev = p_event;
++
++ switch (FALCON_EVENT_CODE(ev)) {
++ case FALCON_EVENT_CODE_CHAR:
++ falcon_handle_char_event(nic, lnic->ev_handlers, ev);
++ break;
++ default:
++ EFRM_WARN("%s: unknown event type=%x", __FUNCTION__,
++ (unsigned)FALCON_EVENT_CODE(ev));
++ break;
++ }
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/efx_vi_shm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/efx_vi_shm.c 2008-05-19 00:33:29.421842036 +0300
+@@ -0,0 +1,701 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides implementation of EFX VI API, used from Xen
++ * acceleration driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "linux_resource_internal.h"
++#include <ci/efrm/nic_table.h>
++#include <ci/efrm/vi_resource_manager.h>
++#include <ci/driver/resource/efx_vi.h>
++#include <ci/efrm/filter.h>
++#include <ci/efrm/buffer_table.h>
++#include <linux/pci.h>
++#include "kernel_compat.h"
++
++#if EFX_VI_STATIC_FILTERS
++struct filter_list_t {
++ struct filter_list_t *next;
++ struct filter_resource *fres;
++};
++#endif
++
++struct efx_vi_state {
++ struct vi_resource *vi_res;
++
++ int nic_index;
++
++ void (*callback_fn)(void *arg, int is_timeout);
++ void *callback_arg;
++
++ struct completion flush_completion;
++
++#if EFX_VI_STATIC_FILTERS
++ struct filter_list_t fres[EFX_VI_STATIC_FILTERS];
++ struct filter_list_t *free_fres;
++ struct filter_list_t *used_fres;
++#endif
++};
++
++static void efx_vi_flush_complete(void *state_void)
++{
++ struct efx_vi_state *state = (struct efx_vi_state *)state_void;
++
++ complete(&state->flush_completion);
++}
++
++static inline int alloc_ep(struct efx_vi_state *state)
++{
++ int rc;
++
++ rc = efrm_vi_resource_alloc(NULL, EFHW_VI_JUMBO_EN,
++ efx_vi_eventq_size,
++ FALCON_DMA_Q_DEFAULT_TX_SIZE,
++ FALCON_DMA_Q_DEFAULT_RX_SIZE,
++ 0, 0, &state->vi_res, NULL, NULL, NULL,
++ NULL);
++ if (rc < 0) {
++ EFRM_ERR("%s: ERROR efrm_vi_resource_alloc error %d",
++ __FUNCTION__, rc);
++ return rc;
++ }
++
++ efrm_vi_register_flush_callback(state->vi_res, &efx_vi_flush_complete,
++ (void *)state);
++
++ return 0;
++}
++
++static int free_ep(struct efx_vi_state *efx_state)
++{
++ efrm_vi_resource_release(efx_state->vi_res);
++
++ return 0;
++}
++
++#if EFX_VI_STATIC_FILTERS
++static int efx_vi_alloc_static_filters(struct efx_vi_state *efx_state)
++{
++ int i;
++ int rc;
++
++ efx_state->free_fres = efx_state->used_fres = NULL;
++
++ for (i = 0; i < EFX_VI_STATIC_FILTERS; i++) {
++ rc = efrm_filter_resource_alloc(efx_state->vi_res,
++ &efx_state->fres[i].fres);
++ if (rc < 0) {
++ EFRM_ERR("%s: efrm_filter_resource_alloc failed: %d",
++ __FUNCTION__, rc);
++ while (i > 0) {
++ i--;
++ efrm_filter_resource_release(efx_state->
++ fres[i].fres);
++ }
++ efx_state->free_fres = NULL;
++ return rc;
++ }
++ efx_state->fres[i].next = efx_state->free_fres;
++ efx_state->free_fres = &efx_state->fres[i];
++ }
++
++ return 0;
++}
++#endif
++
++int efx_vi_alloc(struct efx_vi_state **vih_out, int nic_index)
++{
++ struct efx_vi_state *efx_state;
++ int rc;
++
++ BUG_ON(nic_index < 0 || nic_index >= EFHW_MAX_NR_DEVS);
++
++ efx_state = kmalloc(sizeof(struct efx_vi_state), GFP_KERNEL);
++
++ if (!efx_state) {
++ EFRM_ERR("%s: failed to allocate memory for efx_vi_state",
++ __FUNCTION__);
++ rc = -ENOMEM;
++ goto fail;
++ }
++
++ efx_state->nic_index = nic_index;
++ init_completion(&efx_state->flush_completion);
++
++ /* basically allocate_pt_endpoint() */
++ rc = alloc_ep(efx_state);
++ if (rc) {
++ EFRM_ERR("%s: alloc_ep failed: %d", __FUNCTION__, rc);
++ goto fail_no_pt;
++ }
++#if EFX_VI_STATIC_FILTERS
++ /* Statically allocate a set of filter resources - removes the
++ restriction on not being able to use efx_vi_filter() from
++ in_atomic() */
++ rc = efx_vi_alloc_static_filters(efx_state);
++ if (rc)
++ goto fail_no_filters;
++#endif
++
++ *vih_out = efx_state;
++
++ return 0;
++#if EFX_VI_STATIC_FILTERS
++fail_no_filters:
++ free_ep(efx_state);
++#endif
++fail_no_pt:
++ kfree(efx_state);
++fail:
++ return rc;
++}
++EXPORT_SYMBOL(efx_vi_alloc);
++
++void efx_vi_free(struct efx_vi_state *vih)
++{
++ struct efx_vi_state *efx_state = vih;
++
++ /* TODO flush dma channels, init dma queues?. See ef_free_vnic() */
++#if EFX_VI_STATIC_FILTERS
++ int i;
++
++ for (i = 0; i < EFX_VI_STATIC_FILTERS; i++)
++ efrm_filter_resource_release(efx_state->fres[i].fres);
++#endif
++
++ if (efx_state->vi_res)
++ free_ep(efx_state);
++
++ kfree(efx_state);
++}
++EXPORT_SYMBOL(efx_vi_free);
++
++void efx_vi_reset(struct efx_vi_state *vih)
++{
++ struct efx_vi_state *efx_state = vih;
++
++ efrm_pt_flush(efx_state->vi_res);
++
++ while (wait_for_completion_timeout(&efx_state->flush_completion, HZ)
++ == 0)
++ efrm_vi_resource_flush_retry(efx_state->vi_res);
++
++ /* Bosch the eventq */
++ efrm_eventq_reset(efx_state->vi_res, 0);
++ return;
++}
++EXPORT_SYMBOL(efx_vi_reset);
++
++static void
++efx_vi_eventq_callback(void *context, int is_timeout, struct efhw_nic *nic)
++{
++ struct efx_vi_state *efx_state = (struct efx_vi_state *)context;
++
++ EFRM_ASSERT(efx_state->callback_fn);
++
++ return efx_state->callback_fn(efx_state->callback_arg, is_timeout);
++}
++
++int
++efx_vi_eventq_register_callback(struct efx_vi_state *vih,
++ void (*callback)(void *context, int is_timeout),
++ void *context)
++{
++ struct efx_vi_state *efx_state = vih;
++
++ efx_state->callback_fn = callback;
++ efx_state->callback_arg = context;
++
++ /* Register the eventq timeout event callback */
++ efrm_eventq_register_callback(efx_state->vi_res,
++ efx_vi_eventq_callback, efx_state);
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_eventq_register_callback);
++
++int efx_vi_eventq_kill_callback(struct efx_vi_state *vih)
++{
++ struct efx_vi_state *efx_state = vih;
++
++ if (efx_state->vi_res->evq_callback_fn)
++ efrm_eventq_kill_callback(efx_state->vi_res);
++
++ efx_state->callback_fn = NULL;
++ efx_state->callback_arg = NULL;
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_eventq_kill_callback);
++
++struct efx_vi_dma_map_state {
++ struct efhw_buffer_table_allocation bt_handle;
++ int n_pages;
++ dma_addr_t *dma_addrs;
++};
++
++int
++efx_vi_dma_map_pages(struct efx_vi_state *vih, struct page **pages,
++ int n_pages, struct efx_vi_dma_map_state **dmh_out)
++{
++ struct efx_vi_state *efx_state = vih;
++ int order = fls(n_pages - 1), rc, i, evq_id;
++ dma_addr_t dma_addr;
++ struct efx_vi_dma_map_state *dm_state;
++
++ if (n_pages != (1 << order)) {
++ EFRM_WARN("%s: Can only allocate buffers in power of 2 "
++ "sizes (not %d)", __FUNCTION__, n_pages);
++ return -EINVAL;
++ }
++
++ dm_state = kmalloc(sizeof(struct efx_vi_dma_map_state), GFP_KERNEL);
++ if (!dm_state)
++ return -ENOMEM;
++
++ dm_state->dma_addrs = kmalloc(sizeof(dma_addr_t) * n_pages,
++ GFP_KERNEL);
++ if (!dm_state->dma_addrs) {
++ kfree(dm_state);
++ return -ENOMEM;
++ }
++
++ rc = efrm_buffer_table_alloc(order, &dm_state->bt_handle);
++ if (rc < 0) {
++ kfree(dm_state->dma_addrs);
++ kfree(dm_state);
++ return rc;
++ }
++
++ evq_id = EFRM_RESOURCE_INSTANCE(efx_state->vi_res->rs.rs_handle);
++ for (i = 0; i < n_pages; i++) {
++ /* TODO do we need to get_page() here ? */
++
++ dma_addr = pci_map_page
++ (linux_efhw_nic(efrm_nic_table.nic[efx_state->nic_index])->
++ pci_dev, pages[i], 0, PAGE_SIZE, PCI_DMA_TODEVICE);
++
++ efrm_buffer_table_set(&dm_state->bt_handle, i, dma_addr,
++ evq_id);
++
++ dm_state->dma_addrs[i] = dma_addr;
++
++ /* Would be nice to not have to call commit each time, but
++ * comment says there are hardware restrictions on how often
++ * you can go without it, so do this to be safe */
++ efrm_buffer_table_commit();
++ }
++
++ dm_state->n_pages = n_pages;
++
++ *dmh_out = dm_state;
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_dma_map_pages);
++
++/* Function needed as Xen can't get pages for grants in dom0, but can
++ get dma address */
++int
++efx_vi_dma_map_addrs(struct efx_vi_state *vih,
++ unsigned long long *bus_dev_addrs,
++ int n_pages, struct efx_vi_dma_map_state **dmh_out)
++{
++ struct efx_vi_state *efx_state = vih;
++ int order = fls(n_pages - 1), rc, i, evq_id;
++ dma_addr_t dma_addr;
++ struct efx_vi_dma_map_state *dm_state;
++
++ if (n_pages != (1 << order)) {
++ EFRM_WARN("%s: Can only allocate buffers in power of 2 "
++ "sizes (not %d)", __FUNCTION__, n_pages);
++ return -EINVAL;
++ }
++
++ dm_state = kmalloc(sizeof(struct efx_vi_dma_map_state), GFP_KERNEL);
++ if (!dm_state)
++ return -ENOMEM;
++
++ dm_state->dma_addrs = kmalloc(sizeof(dma_addr_t) * n_pages,
++ GFP_KERNEL);
++ if (!dm_state->dma_addrs) {
++ kfree(dm_state);
++ return -ENOMEM;
++ }
++
++ rc = efrm_buffer_table_alloc(order, &dm_state->bt_handle);
++ if (rc < 0) {
++ kfree(dm_state->dma_addrs);
++ kfree(dm_state);
++ return rc;
++ }
++
++ evq_id = EFRM_RESOURCE_INSTANCE(efx_state->vi_res->rs.rs_handle);
++#if 0
++ EFRM_WARN("%s: mapping %d pages to evq %d, bt_ids %d-%d\n",
++ __FUNCTION__, n_pages, evq_id,
++ dm_state->bt_handle.base,
++ dm_state->bt_handle.base + n_pages);
++#endif
++ for (i = 0; i < n_pages; i++) {
++
++ dma_addr = (dma_addr_t)bus_dev_addrs[i];
++
++ efrm_buffer_table_set(&dm_state->bt_handle, i, dma_addr,
++ evq_id);
++
++ dm_state->dma_addrs[i] = dma_addr;
++
++ /* Would be nice to not have to call commit each time, but
++ * comment says there are hardware restrictions on how often
++ * you can go without it, so do this to be safe */
++ efrm_buffer_table_commit();
++ }
++
++ dm_state->n_pages = n_pages;
++
++ *dmh_out = dm_state;
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_dma_map_addrs);
++
++void
++efx_vi_dma_unmap_pages(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh)
++{
++ struct efx_vi_state *efx_state = vih;
++ struct efx_vi_dma_map_state *dm_state =
++ (struct efx_vi_dma_map_state *)dmh;
++ int i;
++
++ efrm_buffer_table_free(&dm_state->bt_handle);
++
++ for (i = 0; i < dm_state->n_pages; ++i)
++ pci_unmap_page(linux_efhw_nic
++ (efrm_nic_table.nic[efx_state->nic_index])->pci_dev,
++ dm_state->dma_addrs[i], PAGE_SIZE, PCI_DMA_TODEVICE);
++
++ kfree(dm_state->dma_addrs);
++ kfree(dm_state);
++
++ return;
++}
++EXPORT_SYMBOL(efx_vi_dma_unmap_pages);
++
++void
++efx_vi_dma_unmap_addrs(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh)
++{
++ struct efx_vi_dma_map_state *dm_state =
++ (struct efx_vi_dma_map_state *)dmh;
++
++ efrm_buffer_table_free(&dm_state->bt_handle);
++
++ kfree(dm_state->dma_addrs);
++ kfree(dm_state);
++
++ return;
++}
++EXPORT_SYMBOL(efx_vi_dma_unmap_addrs);
++
++unsigned
++efx_vi_dma_get_map_addr(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh)
++{
++ struct efx_vi_dma_map_state *dm_state =
++ (struct efx_vi_dma_map_state *)dmh;
++
++ return EFHW_BUFFER_ADDR(dm_state->bt_handle.base, 0);
++}
++EXPORT_SYMBOL(efx_vi_dma_get_map_addr);
++
++#if EFX_VI_STATIC_FILTERS
++static int
++get_filter(struct efx_vi_state *efx_state,
++ efrm_resource_handle_t pthandle, struct filter_resource **fres_out)
++{
++ struct filter_list_t *flist;
++ if (efx_state->free_fres == NULL)
++ return -ENOMEM;
++ else {
++ flist = efx_state->free_fres;
++ efx_state->free_fres = flist->next;
++ flist->next = efx_state->used_fres;
++ efx_state->used_fres = flist;
++ *fres_out = flist->fres;
++ return 0;
++ }
++}
++#endif
++
++static void
++release_filter(struct efx_vi_state *efx_state, struct filter_resource *fres)
++{
++#if EFX_VI_STATIC_FILTERS
++ struct filter_list_t *flist = efx_state->used_fres, *prev = NULL;
++ while (flist) {
++ if (flist->fres == fres) {
++ if (prev)
++ prev->next = flist->next;
++ else
++ efx_state->used_fres = flist->next;
++ flist->next = efx_state->free_fres;
++ efx_state->free_fres = flist;
++ return;
++ }
++ prev = flist;
++ flist = flist->next;
++ }
++ EFRM_ERR("%s: couldn't find filter", __FUNCTION__);
++#else
++ return efrm_filter_resource_release(fres);
++#endif
++}
++
++int
++efx_vi_filter(struct efx_vi_state *vih, int protocol,
++ unsigned ip_addr_be32, int port_le16,
++ struct filter_resource_t **fh_out)
++{
++ struct efx_vi_state *efx_state = vih;
++ struct filter_resource *frs;
++ int rc;
++
++#if EFX_VI_STATIC_FILTERS
++ rc = get_filter(efx_state, efx_state->vi_res->rs.rs_handle, &frs);
++#else
++ rc = efrm_filter_resource_alloc(efx_state->vi_res, &frs);
++#endif
++ if (rc < 0)
++ return rc;
++
++ /* Add the hardware filter. We pass in the source port and address
++ * as 0 (wildcard) to minimise the number of filters needed. */
++ if (protocol == IPPROTO_TCP) {
++ rc = efrm_filter_resource_tcp_set(frs, 0, 0, ip_addr_be32,
++ port_le16);
++ } else {
++ rc = efrm_filter_resource_udp_set(frs, 0, 0, ip_addr_be32,
++ port_le16);
++ }
++
++ *fh_out = (struct filter_resource_t *)frs;
++
++ return rc;
++}
++EXPORT_SYMBOL(efx_vi_filter);
++
++int
++efx_vi_filter_stop(struct efx_vi_state *vih, struct filter_resource_t *fh)
++{
++ struct efx_vi_state *efx_state = vih;
++ struct filter_resource *frs = (struct filter_resource *)fh;
++ int rc;
++
++ rc = efrm_filter_resource_clear(frs);
++ release_filter(efx_state, frs);
++
++ return rc;
++}
++EXPORT_SYMBOL(efx_vi_filter_stop);
++
++int
++efx_vi_hw_resource_get_virt(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length)
++{
++ EFRM_NOTICE("%s: TODO!", __FUNCTION__);
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_hw_resource_get_virt);
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++int
++efx_vi_hw_resource_get_phys(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length)
++{
++ struct efx_vi_state *efx_state = vih;
++ int i, ni = efx_state->nic_index;
++ struct linux_efhw_nic *lnic = linux_efhw_nic(efrm_nic_table.nic[ni]);
++ unsigned long phys = lnic->ctr_ap_pci_addr;
++ struct efrm_resource *ep_res = &efx_state->vi_res->rs;
++ unsigned ep_mmap_bytes;
++
++ if (*length < EFX_VI_HW_RESOURCE_MAXSIZE)
++ return -EINVAL;
++
++ mdata->version = 0;
++
++ mdata->nic_arch = efrm_nic_table.nic[ni]->devtype.arch;
++ mdata->nic_variant = efrm_nic_table.nic[ni]->devtype.variant;
++ mdata->nic_revision = efrm_nic_table.nic[ni]->devtype.revision;
++
++ mdata->evq_order =
++ efx_state->vi_res->nic_info[ni].evq_pages.iobuff.order;
++ mdata->evq_offs = efx_state->vi_res->nic_info[ni].evq_pages.iobuff_off;
++ mdata->evq_capacity = efx_vi_eventq_size;
++ mdata->instance = EFRM_RESOURCE_INSTANCE(ep_res->rs_handle);
++ mdata->rx_capacity = FALCON_DMA_Q_DEFAULT_RX_SIZE;
++ mdata->tx_capacity = FALCON_DMA_Q_DEFAULT_TX_SIZE;
++
++ ep_mmap_bytes = FALCON_DMA_Q_DEFAULT_MMAP;
++ EFRM_ASSERT(ep_mmap_bytes == PAGE_SIZE * 2);
++
++#ifndef NDEBUG
++ {
++ /* Sanity about doorbells */
++ unsigned long tx_dma_page_addr, rx_dma_page_addr;
++
++ /* get rx doorbell address */
++ rx_dma_page_addr =
++ phys + falcon_rx_dma_page_addr(mdata->instance);
++ /* get tx doorbell address */
++ tx_dma_page_addr =
++ phys + falcon_tx_dma_page_addr(mdata->instance);
++
++ /* Check the lower bits of the TX doorbell will be
++ * consistent. */
++ EFRM_ASSERT((TX_DESC_UPD_REG_PAGE4_OFST &
++ FALCON_DMA_PAGE_MASK) ==
++ (TX_DESC_UPD_REG_PAGE123K_OFST &
++ FALCON_DMA_PAGE_MASK));
++
++ /* Check the lower bits of the RX doorbell will be
++ * consistent. */
++ EFRM_ASSERT((RX_DESC_UPD_REG_PAGE4_OFST &
++ FALCON_DMA_PAGE_MASK) ==
++ (RX_DESC_UPD_REG_PAGE123K_OFST &
++ FALCON_DMA_PAGE_MASK));
++
++ /* Check that the doorbells will be in the same page. */
++ EFRM_ASSERT((TX_DESC_UPD_REG_PAGE4_OFST & PAGE_MASK) ==
++ (RX_DESC_UPD_REG_PAGE4_OFST & PAGE_MASK));
++
++ /* Check that the doorbells are in the same page. */
++ EFRM_ASSERT((tx_dma_page_addr & PAGE_MASK) ==
++ (rx_dma_page_addr & PAGE_MASK));
++
++ /* Check that the TX doorbell offset is correct. */
++ EFRM_ASSERT((TX_DESC_UPD_REG_PAGE4_OFST & ~PAGE_MASK) ==
++ (tx_dma_page_addr & ~PAGE_MASK));
++
++ /* Check that the RX doorbell offset is correct. */
++ EFRM_ASSERT((RX_DESC_UPD_REG_PAGE4_OFST & ~PAGE_MASK) ==
++ (rx_dma_page_addr & ~PAGE_MASK));
++ }
++#endif
++
++ i = 0;
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_TXDMAQ;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address =
++ (unsigned long)efx_state->vi_res->nic_info[ni].
++ dmaq_pages[EFRM_VI_RM_DMA_QUEUE_TX].kva;
++
++ i++;
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_RXDMAQ;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address =
++ (unsigned long)efx_state->vi_res->nic_info[ni].
++ dmaq_pages[EFRM_VI_RM_DMA_QUEUE_RX].kva;
++
++ /* NB EFX_VI_HW_RESOURCE_TXBELL not used on Falcon */
++ /* NB EFX_VI_HW_RESOURCE_RXBELL not used on Falcon */
++
++ i++;
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_EVQTIMER;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address =
++ (unsigned long)phys + falcon_timer_page_addr(mdata->instance);
++
++ /* NB EFX_VI_HW_RESOURCE_EVQPTR not used on Falcon */
++
++ i++;
++ switch (efrm_nic_table.nic[ni]->devtype.variant) {
++ case 'A':
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_EVQRPTR;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address = (unsigned long)phys +
++ EVQ_RPTR_REG_OFST +
++ (FALCON_REGISTER128 * mdata->instance);
++ break;
++ case 'B':
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_EVQRPTR_OFFSET;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address =
++ (unsigned long)FALCON_EVQ_RPTR_REG_P0;
++ break;
++ default:
++ EFRM_ASSERT(0);
++ break;
++ }
++
++ i++;
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_EVQMEMKVA;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_IOBUFFER;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address = (unsigned long)efx_state->vi_res->
++ nic_info[ni].evq_pages.iobuff.kva;
++
++ i++;
++ hw_res_array[i].type = EFX_VI_HW_RESOURCE_BELLPAGE;
++ hw_res_array[i].mem_type = EFX_VI_HW_RESOURCE_PERIPHERAL;
++ hw_res_array[i].more_to_follow = 0;
++ hw_res_array[i].length = PAGE_SIZE;
++ hw_res_array[i].address =
++ (unsigned long)(phys +
++ falcon_tx_dma_page_addr(mdata->instance))
++ >> PAGE_SHIFT;
++
++ i++;
++
++ EFRM_ASSERT(i <= *length);
++
++ *length = i;
++
++ return 0;
++}
++EXPORT_SYMBOL(efx_vi_hw_resource_get_phys);
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/eventq.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/eventq.c 2008-05-19 00:33:29.421842036 +0300
+@@ -0,0 +1,320 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains event queue support.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efhw/debug.h>
++#include <ci/efhw/iopage.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/eventq.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efhw/nic.h>
++
++#define KEVENTQ_MAGIC 0x07111974
++
++/*! Helper function to allocate the iobuffer needed by an eventq
++ * - it ensures the eventq has the correct alignment for the NIC
++ *
++ * \param rm Event-queue resource manager
++ * \param instance Event-queue instance (index)
++ * \param buf_bytes Requested size of eventq
++ * \return < 0 if iobuffer allocation fails
++ */
++int
++efhw_nic_event_queue_alloc_iobuffer(struct efhw_nic *nic,
++ struct eventq_resource_hardware *h,
++ int evq_instance, unsigned buf_bytes)
++{
++ unsigned int page_order;
++ int rc;
++
++ /* Allocate an iobuffer. */
++ page_order = get_order(buf_bytes);
++
++ h->iobuff_off = 0;
++
++ EFHW_TRACE("allocating eventq size %x",
++ 1u << (page_order + PAGE_SHIFT));
++ rc = efhw_iopages_alloc(nic, &h->iobuff, page_order);
++ if (rc < 0) {
++ EFHW_WARN("%s: failed to allocate %u pages",
++ __FUNCTION__, 1u << page_order);
++ return rc;
++ }
++
++ /* Set the eventq pages to match EFHW_CLEAR_EVENT() */
++ if (EFHW_CLEAR_EVENT_VALUE)
++ memset(efhw_iopages_ptr(&h->iobuff) + h->iobuff_off,
++ EFHW_CLEAR_EVENT_VALUE, (1u << page_order) * PAGE_SIZE);
++
++ EFHW_TRACE("%s: allocated %u pages", __FUNCTION__, 1u << (page_order));
++
++ /* For Falcon the NIC is programmed with the base buffer address of a
++ * contiguous region of buffer space. This means that larger than a
++ * PAGE event queues can be expected to allocate even when the host's
++ * physical memory is fragmented */
++ EFHW_ASSERT(efhw_nic_have_hw(nic));
++ EFHW_ASSERT(page_order <= h->buf_tbl_alloc.order);
++
++ /* Initialise the buffer table entries. */
++ falcon_nic_buffer_table_set_n(nic, h->buf_tbl_alloc.base,
++ efhw_iopages_dma_addr(&h->iobuff) +
++ h->iobuff_off, EFHW_NIC_PAGE_SIZE, 0,
++ 1 << page_order, 0);
++
++ if (evq_instance >= FALCON_EVQ_TBL_RESERVED)
++ falcon_nic_buffer_table_confirm(nic);
++ return 0;
++}
++
++/**********************************************************************
++ * Kernel event queue management.
++ */
++
++/* Values for [struct efhw_keventq::lock] field. */
++#define KEVQ_UNLOCKED 0
++#define KEVQ_LOCKED 1
++#define KEVQ_RECHECK 2
++
++int
++efhw_keventq_ctor(struct efhw_nic *nic, int instance,
++ struct efhw_keventq *evq,
++ struct efhw_ev_handler *ev_handlers)
++{
++ int rc;
++ unsigned buf_bytes = evq->hw.capacity * sizeof(efhw_event_t);
++
++ evq->instance = instance;
++ evq->ev_handlers = ev_handlers;
++
++ /* allocate an IObuffer for the eventq */
++ rc = efhw_nic_event_queue_alloc_iobuffer(nic, &evq->hw, evq->instance,
++ buf_bytes);
++ if (rc < 0)
++ return rc;
++
++ /* Zero the timer-value for this queue.
++ AND Tell the nic about the event queue. */
++ efhw_nic_event_queue_enable(nic, evq->instance, evq->hw.capacity,
++ efhw_iopages_dma_addr(&evq->hw.iobuff) +
++ evq->hw.iobuff_off,
++ evq->hw.buf_tbl_alloc.base);
++
++ evq->lock = KEVQ_UNLOCKED;
++ evq->evq_base = efhw_iopages_ptr(&evq->hw.iobuff) + evq->hw.iobuff_off;
++ evq->evq_ptr = 0;
++ evq->evq_mask = (evq->hw.capacity * sizeof(efhw_event_t)) - 1u;
++
++ EFHW_TRACE("%s: [%d] base=%p end=%p", __FUNCTION__, evq->instance,
++ evq->evq_base, evq->evq_base + buf_bytes);
++
++ return 0;
++}
++
++void efhw_keventq_dtor(struct efhw_nic *nic, struct efhw_keventq *evq)
++{
++ EFHW_ASSERT(evq);
++
++ EFHW_TRACE("%s: [%d]", __FUNCTION__, evq->instance);
++
++ /* Zero the timer-value for this queue.
++ And Tell NIC to stop using this event queue. */
++ efhw_nic_event_queue_disable(nic, evq->instance, 0);
++
++ /* free the pages used by the eventq itself */
++ efhw_iopages_free(nic, &evq->hw.iobuff);
++}
++
++void
++efhw_handle_txdmaq_flushed(struct efhw_nic *nic, struct efhw_ev_handler *h,
++ efhw_event_t *evp)
++{
++ int instance = (int)FALCON_EVENT_TX_FLUSH_Q_ID(evp);
++ EFHW_TRACE("%s: instance=%d", __FUNCTION__, instance);
++
++ if (!h->dmaq_flushed_fn) {
++ EFHW_WARN("%s: no handler registered", __FUNCTION__);
++ return;
++ }
++
++ h->dmaq_flushed_fn(nic, instance, false);
++}
++
++void
++efhw_handle_rxdmaq_flushed(struct efhw_nic *nic, struct efhw_ev_handler *h,
++ efhw_event_t *evp)
++{
++ int instance = (int)FALCON_EVENT_RX_FLUSH_Q_ID(evp);
++ EFHW_TRACE("%s: instance=%d", __FUNCTION__, instance);
++
++ if (!h->dmaq_flushed_fn) {
++ EFHW_WARN("%s: no handler registered", __FUNCTION__);
++ return;
++ }
++
++ h->dmaq_flushed_fn(nic, instance, true);
++}
++
++void
++efhw_handle_wakeup_event(struct efhw_nic *nic, struct efhw_ev_handler *h,
++ efhw_event_t *evp)
++{
++ if (!h->wakeup_fn) {
++ EFHW_WARN("%s: no handler registered", __FUNCTION__);
++ return;
++ }
++
++ h->wakeup_fn(nic, evp);
++}
++
++void
++efhw_handle_timeout_event(struct efhw_nic *nic, struct efhw_ev_handler *h,
++ efhw_event_t *evp)
++{
++ if (!h->timeout_fn) {
++ EFHW_WARN("%s: no handler registered", __FUNCTION__);
++ return;
++ }
++
++ h->timeout_fn(nic, evp);
++}
++
++/**********************************************************************
++ * Kernel event queue event handling.
++ */
++
++int efhw_keventq_poll(struct efhw_nic *nic, struct efhw_keventq *q)
++{
++ efhw_event_t *ev;
++ int l, count = 0;
++
++ EFHW_ASSERT(nic);
++ EFHW_ASSERT(q);
++ EFHW_ASSERT(q->ev_handlers);
++
++ /* Acquire the lock, or mark the queue as needing re-checking. */
++ for (;;) {
++ l = q->lock;
++ if (l == KEVQ_UNLOCKED) {
++ if ((int)cmpxchg(&q->lock, l, KEVQ_LOCKED) == l)
++ break;
++ } else if (l == KEVQ_LOCKED) {
++ if ((int)cmpxchg(&q->lock, l, KEVQ_RECHECK) == l)
++ return 0;
++ } else { /* already marked for re-checking */
++ EFHW_ASSERT(l == KEVQ_RECHECK);
++ return 0;
++ }
++ }
++
++ if (unlikely(EFHW_EVENT_OVERFLOW(q, q)))
++ goto overflow;
++
++ ev = EFHW_EVENT_PTR(q, q, 0);
++
++#ifndef NDEBUG
++ if (!EFHW_IS_EVENT(ev))
++ EFHW_TRACE("%s: %d NO EVENTS!", __FUNCTION__, q->instance);
++#endif
++
++ for (;;) {
++ /* Convention for return codes for handlers is:
++ ** 0 - no error, event consumed
++ ** 1 - no error, event not consumed
++ ** -ve - error, event not consumed
++ */
++ if (likely(EFHW_IS_EVENT(ev))) {
++ count++;
++
++ switch (FALCON_EVENT_CODE(ev)) {
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ case FALCON_EVENT_CODE_CHAR:
++ falcon_handle_char_event(nic, q->ev_handlers,
++ ev);
++ break;
++#endif
++
++ default:
++ EFHW_ERR("efhw_keventq_poll: [%d] UNEXPECTED "
++ "EVENT:"FALCON_EVENT_FMT,
++ q->instance,
++ FALCON_EVENT_PRI_ARG(*ev));
++ }
++
++ EFHW_CLEAR_EVENT(ev);
++ EFHW_EVENTQ_NEXT(q);
++
++ ev = EFHW_EVENT_PTR(q, q, 0);
++ } else {
++ /* No events left. Release the lock (checking if we
++ * need to re-poll to avoid race). */
++ l = q->lock;
++ if (l == KEVQ_LOCKED) {
++ if ((int)cmpxchg(&q->lock, l, KEVQ_UNLOCKED)
++ == l) {
++ EFHW_TRACE
++ ("efhw_keventq_poll: %d clean exit",
++ q->instance);
++ goto clean_exit;
++ }
++ }
++
++ /* Potentially more work to do. */
++ l = q->lock;
++ EFHW_ASSERT(l == KEVQ_RECHECK);
++ EFHW_TEST((int)cmpxchg(&q->lock, l, KEVQ_LOCKED) == l);
++ EFHW_TRACE("efhw_keventq_poll: %d re-poll required",
++ q->instance);
++ }
++ }
++
++ /* shouldn't get here */
++ EFHW_ASSERT(0);
++
++overflow:
++ /* ?? Oh dear. Should we poll everything that could have possibly
++ ** happened? Or merely cry out in anguish...
++ */
++ EFHW_WARN("efhw_keventq_poll: %d ***** OVERFLOW nic %d *****",
++ q->instance, nic->index);
++
++ q->lock = KEVQ_UNLOCKED;
++ return count;
++
++clean_exit:
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ /* Ack the processed events so that this event queue can potentially
++ raise interrupts again */
++ falcon_nic_evq_ack(nic, q->instance,
++ (EFHW_EVENT_OFFSET(q, q, 0) / sizeof(efhw_event_t)),
++ false);
++#endif
++ return count;
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/falcon.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/falcon.c 2008-05-19 00:33:29.425842266 +0300
+@@ -0,0 +1,2758 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains Falcon hardware support.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/debug.h>
++#include <ci/efhw/iopage.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efhw/falcon_hash.h>
++#include <ci/efhw/nic.h>
++#include <ci/efhw/eventq.h>
++#include <ci/efhw/checks.h>
++
++
++/*----------------------------------------------------------------------------
++ *
++ * Workarounds and options
++ *
++ *---------------------------------------------------------------------------*/
++
++/* on for debug builds */
++#ifndef NDEBUG
++# define FALCON_FULL_FILTER_CACHE 1 /* complete SW shadow of filter tbl */
++# define FALCON_VERIFY_FILTERS 0
++#else /* Also adds duplicate filter check */
++# define FALCON_FULL_FILTER_CACHE 1 /* keep this on for some security */
++# define FALCON_VERIFY_FILTERS 0
++#endif
++
++/* options */
++#define RX_FILTER_CTL_SRCH_LIMIT_TCP_FULL 8 /* default search limit */
++#define RX_FILTER_CTL_SRCH_LIMIT_TCP_WILD 8 /* default search limit */
++#define RX_FILTER_CTL_SRCH_LIMIT_UDP_FULL 8 /* default search limit */
++#define RX_FILTER_CTL_SRCH_LIMIT_UDP_WILD 8 /* default search limit */
++#define RX_FILTER_CTL_SRCH_FUDGE_WILD 3 /* increase the search limit */
++#define RX_FILTER_CTL_SRCH_FUDGE_FULL 1 /* increase the search limit */
++
++#define FALCON_MAC_SET_TYPE_BY_SPEED 1
++
++/* FIXME: We should detect mode at runtime. */
++#define FALCON_BUFFER_TABLE_FULL_MODE 1
++
++/*----------------------------------------------------------------------------
++ *
++ * Debug Macros
++ *
++ *---------------------------------------------------------------------------*/
++
++#ifndef __KERNEL__
++#define _DEBUG_SYM_ extern
++#else
++#define _DEBUG_SYM_ static inline
++#endif
++
++ /*----------------------------------------------------------------------------
++ *
++ * Macros and forward declarations
++ *
++ *--------------------------------------------------------------------------*/
++
++#define FALCON_REGION_NUM 4 /* number of supported memory regions */
++
++#define FALCON_BUFFER_TBL_HALF_BYTES 4
++#define FALCON_BUFFER_TBL_FULL_BYTES 8
++
++/* Shadow buffer table - hack for testing only */
++#if FALCON_BUFFER_TABLE_FULL_MODE == 0
++# define FALCON_USE_SHADOW_BUFFER_TABLE 1
++#else
++# define FALCON_USE_SHADOW_BUFFER_TABLE 0
++#endif
++
++#if FALCON_USE_SHADOW_BUFFER_TABLE
++static uint64_t _falcon_buffer_table[FALCON_BUFFER_TBL_NUM];
++#endif
++
++/*----------------------------------------------------------------------------
++ *
++ * Header assertion checks
++ *
++ *---------------------------------------------------------------------------*/
++
++#define FALCON_ASSERT_VALID() /* nothing yet */
++
++/* Falcon has a 128bit register model but most registers have useful
++ defaults or only implement a small number of bits. Some registers
++ can be programmed 32bits UNLOCKED all others should be interlocked
++ against other threads within the same protection domain.
++
++ Aim is for software to perform the minimum number of writes and
++ also to minimise the read-modify-write activity (which generally
++ indicates a lack of clarity in the use model).
++
++ Registers which are programmed in this module are listed below
++ together with the method of access. Care must be taken to ensure
++ remain adequate if the register spec changes.
++
++ All 128bits programmed
++ FALCON_BUFFER_TBL_HALF
++ RX_FILTER_TBL
++ TX_DESC_PTR_TBL
++ RX_DESC_PTR_TBL
++ DRV_EV_REG
++
++ All 64bits programmed
++ FALCON_BUFFER_TBL_FULL
++
++ 32 bits are programmed (UNLOCKED)
++ EVQ_RPTR_REG
++
++ Low 64bits programmed remainder are written with a random number
++ RX_DC_CFG_REG
++ TX_DC_CFG_REG
++ SRM_RX_DC_CFG_REG
++ SRM_TX_DC_CFG_REG
++ BUF_TBL_CFG_REG
++ BUF_TBL_UPD_REG
++ SRM_UPD_EVQ_REG
++ EVQ_PTR_TBL
++ TIMER_CMD_REG
++ TX_PACE_TBL
++ FATAL_INTR_REG
++ INT_EN_REG (When enabling interrupts)
++ TX_FLUSH_DESCQ_REG
++ RX_FLUSH_DESCQ
++
++ Read Modify Write on low 32bits remainder are written with a random number
++ INT_EN_REG (When sending a driver interrupt)
++ DRIVER_REGX
++
++ Read Modify Write on low 64bits remainder are written with a random number
++ SRM_CFG_REG_OFST
++ RX_CFG_REG_OFST
++ RX_FILTER_CTL_REG
++
++ Read Modify Write on full 128bits
++ TXDP_RESERVED_REG (aka TXDP_UNDOCUMENTED)
++ TX_CFG_REG
++
++*/
++
++/*----------------------------------------------------------------------------
++ *
++ * Filters static data
++ *
++ *---------------------------------------------------------------------------*/
++
++/* Defaults are set here to support dma.c */
++static unsigned tcp_full_srch_limit = RX_FILTER_CTL_SRCH_LIMIT_TCP_FULL;
++static unsigned tcp_wild_srch_limit = RX_FILTER_CTL_SRCH_LIMIT_TCP_WILD;
++static unsigned udp_full_srch_limit = RX_FILTER_CTL_SRCH_LIMIT_UDP_FULL;
++static unsigned udp_wild_srch_limit = RX_FILTER_CTL_SRCH_LIMIT_UDP_WILD;
++
++#if FALCON_VERIFY_FILTERS
++static void _falcon_nic_ipfilter_sanity(struct efhw_nic *nic);
++#endif
++
++/*----------------------------------------------------------------------------
++ *
++ * Filters low-level register interface
++ *
++ *---------------------------------------------------------------------------*/
++
++/* Build the filter entry */
++static void
++_falcon_nic_ipfilter_build(struct efhw_nic *nic,
++ int tcp, int full, int rss_b0, int scat_b0,
++ uint filter_i, uint dmaq_id,
++ unsigned saddr_le32, unsigned sport_le16,
++ unsigned daddr_le32, unsigned dport_le16,
++ uint64_t *q0, uint64_t *q1)
++{
++ uint64_t v1, v2, v3, v4;
++ int type = tcp << 4 | full;
++
++ v4 = (((!tcp) << __DW4(TCP_UDP_1_LBN)) |
++ (dmaq_id << __DW4(RXQ_ID_1_LBN)));
++
++ switch (nic->devtype.variant) {
++ case 'A':
++ EFHW_ASSERT(!rss_b0);
++ break;
++ case 'B':
++ v4 |= scat_b0 << __DW4(SCATTER_EN_1_B0_LBN);
++ v4 |= rss_b0 << __DW4(RSS_EN_1_B0_LBN);
++ break;
++ default:
++ EFHW_ASSERT(0);
++ break;
++ }
++
++ v3 = daddr_le32;
++
++ switch (type) {
++
++ case 0x11: /* TCP_FULL */
++ case 0x01: /* UDP_FULL */
++ v2 = ((dport_le16 << __DW2(DEST_PORT_TCP_1_LBN)) |
++ (__HIGH(saddr_le32, SRC_IP_1_LBN, SRC_IP_1_WIDTH)));
++ v1 = ((__LOW(saddr_le32, SRC_IP_1_LBN, SRC_IP_1_WIDTH)) |
++ (sport_le16 << SRC_TCP_DEST_UDP_1_LBN));
++ break;
++
++ case 0x10: /* TCP_WILD */
++ v2 = ((uint64_t) dport_le16 << __DW2(DEST_PORT_TCP_1_LBN));
++ v1 = 0;
++ break;
++
++ case 0x00: /* UDP_WILD */
++ v2 = 0;
++ v1 = ((uint64_t) dport_le16 << SRC_TCP_DEST_UDP_0_LBN);
++ break;
++
++ default:
++ EFHW_ASSERT(0);
++ v2 = 0;
++ v1 = 0;
++ }
++
++ *q0 = (v2 << 32) | v1;
++ *q1 = (v4 << 32) | v3;
++}
++
++static void
++_falcon_nic_ipfilter_set(struct efhw_nic *nic, int tcp,
++ int full, int rss_b0, int scat_b0,
++ uint filter_i, uint dmaq_id,
++ unsigned saddr_le32, unsigned sport_le16,
++ unsigned daddr_le32, unsigned dport_le16)
++{
++ uint64_t q0, q1;
++
++ /* wish you wouldn't do this */
++ EFHW_BUILD_ASSERT(RX_FILTER_TBL1_OFST ==
++ RX_FILTER_TBL0_OFST + FALCON_REGISTER128);
++ EFHW_BUILD_ASSERT(TCP_UDP_1_LBN == TCP_UDP_0_LBN);
++ EFHW_BUILD_ASSERT(RXQ_ID_1_LBN == RXQ_ID_0_LBN);
++ EFHW_BUILD_ASSERT(DEST_IP_1_LBN == DEST_IP_0_LBN);
++ EFHW_BUILD_ASSERT(DEST_PORT_TCP_1_LBN == DEST_PORT_TCP_0_LBN);
++ EFHW_BUILD_ASSERT(SRC_IP_1_LBN == SRC_IP_0_LBN);
++ EFHW_BUILD_ASSERT(SRC_TCP_DEST_UDP_1_LBN == SRC_TCP_DEST_UDP_0_LBN);
++ EFHW_BUILD_ASSERT(SCATTER_EN_1_B0_LBN == SCATTER_EN_0_B0_LBN);
++ EFHW_BUILD_ASSERT(RSS_EN_1_B0_LBN == RSS_EN_0_B0_LBN);
++
++ EFHW_BUILD_ASSERT(TCP_UDP_1_WIDTH == TCP_UDP_0_WIDTH);
++ EFHW_BUILD_ASSERT(RXQ_ID_1_WIDTH == RXQ_ID_0_WIDTH);
++ EFHW_BUILD_ASSERT(DEST_IP_1_WIDTH == DEST_IP_0_WIDTH);
++ EFHW_BUILD_ASSERT(DEST_PORT_TCP_1_WIDTH == DEST_PORT_TCP_0_WIDTH);
++ EFHW_BUILD_ASSERT(SRC_IP_1_WIDTH == SRC_IP_0_WIDTH);
++ EFHW_BUILD_ASSERT(SRC_TCP_DEST_UDP_1_WIDTH == SRC_TCP_DEST_UDP_0_WIDTH);
++ EFHW_BUILD_ASSERT(SCATTER_EN_1_B0_WIDTH == SCATTER_EN_0_B0_WIDTH);
++ EFHW_BUILD_ASSERT(RSS_EN_1_B0_WIDTH == RSS_EN_0_B0_WIDTH);
++
++ /* TODO: Use filter table 1 as well */
++ ulong offset = RX_FILTER_TBL0_OFST + filter_i * 2 * FALCON_REGISTER128;
++
++ EFHW_TRACE("%s[%x]: offset=%lx", __FUNCTION__, filter_i, offset);
++
++ EFHW_TRACE("%s[%x]: filter %d tcp %d full %d src=%x:%x dest=%x:%x%s%s",
++ __FUNCTION__, filter_i, tcp, full, dmaq_id,
++ saddr_le32, sport_le16, daddr_le32, dport_le16,
++ rss_b0 ? " RSS" : "", scat_b0 ? " SCAT" : "");
++
++ EFHW_ASSERT(filter_i < nic->filter_tbl_size);
++
++ /* dword 4 */
++ __DW4CHCK(TCP_UDP_1_LBN, TCP_UDP_1_WIDTH);
++ __DW4CHCK(RXQ_ID_1_LBN, RXQ_ID_1_WIDTH);
++
++ __RANGECHCK(tcp, TCP_UDP_1_WIDTH);
++ __RANGECHCK(dmaq_id, RXQ_ID_1_WIDTH);
++
++ /* dword 3 */
++ __DW3CHCK(DEST_IP_1_LBN, DEST_IP_1_WIDTH);
++ __RANGECHCK(daddr_le32, DEST_IP_1_WIDTH);
++
++ /* dword 2 */
++ __DW2CHCK(DEST_PORT_TCP_1_LBN, DEST_PORT_TCP_1_WIDTH);
++ __LWCHK(SRC_IP_1_LBN, SRC_IP_1_WIDTH);
++ __RANGECHCK(saddr_le32, SRC_IP_1_WIDTH);
++
++ /* dword 1 */
++ __DWCHCK(SRC_TCP_DEST_UDP_1_LBN, SRC_TCP_DEST_UDP_1_WIDTH);
++ __RANGECHCK(sport_le16, SRC_TCP_DEST_UDP_1_WIDTH);
++ __RANGECHCK(dport_le16, SRC_TCP_DEST_UDP_1_WIDTH);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ _falcon_nic_ipfilter_build(nic, tcp, full, rss_b0, scat_b0,
++ filter_i, dmaq_id, saddr_le32, sport_le16,
++ daddr_le32, dport_le16, &q0, &q1);
++
++ EFHW_TRACE("%s[%x]@%p+%lx: %" PRIx64 " %" PRIx64, __FUNCTION__,
++ filter_i, EFHW_KVA(nic), offset, q0, q1);
++
++ falcon_write_qq(EFHW_KVA(nic) + offset, q0, q1);
++ mmiowb();
++
++#if FALCON_VERIFY_FILTERS
++ {
++ uint64_t q0read, q1read;
++
++ /* Read a different entry first - entry BIU flushed shadow */
++ falcon_read_qq(EFHW_KVA(nic) + offset+0x10, &q0read, &q1read);
++ falcon_read_qq(EFHW_KVA(nic) + offset, &q0read, &q1read);
++ EFHW_ASSERT(q0read == q0);
++ EFHW_ASSERT(q1read == q1);
++
++ _falcon_nic_ipfilter_sanity(nic);
++ }
++#endif
++}
++
++static void _falcon_nic_ipfilter_clear(struct efhw_nic *nic, uint filter_i)
++{
++ /* TODO: Use filter table 1 as well */
++ ulong offset = RX_FILTER_TBL0_OFST + filter_i * 2 * FALCON_REGISTER128;
++
++ EFHW_ASSERT(filter_i < nic->filter_tbl_size);
++
++ EFHW_TRACE("%s[%x]", __FUNCTION__, filter_i);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ falcon_write_qq(EFHW_KVA(nic) + offset, 0, 0);
++ mmiowb();
++#if FALCON_VERIFY_FILTERS
++ {
++ uint64_t q0read, q1read;
++
++ /* Read a different entry first - entry BIU flushed shadow */
++ falcon_read_qq(EFHW_KVA(nic) + offset+0x10, &q0read, &q1read);
++ falcon_read_qq(EFHW_KVA(nic) + offset, &q0read, &q1read);
++ EFHW_ASSERT(q0read == 0);
++ EFHW_ASSERT(q1read == 0);
++
++ _falcon_nic_ipfilter_sanity(nic);
++ }
++#endif
++}
++
++/*----------------------------------------------------------------------------
++ *
++ * DMAQ low-level register interface
++ *
++ *---------------------------------------------------------------------------*/
++
++static unsigned dmaq_sizes[] = {
++ 512,
++ EFHW_1K,
++ EFHW_2K,
++ EFHW_4K,
++};
++
++#define N_DMAQ_SIZES (sizeof(dmaq_sizes) / sizeof(dmaq_sizes[0]))
++
++static inline ulong falcon_dma_tx_q_offset(struct efhw_nic *nic, unsigned dmaq)
++{
++ EFHW_ASSERT(dmaq < FALCON_DMAQ_NUM);
++ return TX_DESC_PTR_TBL_OFST + dmaq * FALCON_REGISTER128;
++}
++
++static inline uint falcon_dma_tx_q_size_index(uint dmaq_size)
++{
++ uint i;
++
++ /* size must be one of the various options, otherwise we assert */
++ for (i = 0; i < N_DMAQ_SIZES; i++) {
++ if (dmaq_size == dmaq_sizes[i])
++ break;
++ }
++ EFHW_ASSERT(i < N_DMAQ_SIZES);
++ return i;
++}
++
++static void
++falcon_dmaq_tx_q_init(struct efhw_nic *nic,
++ uint dmaq, uint evq_id, uint own_id,
++ uint tag, uint dmaq_size, uint buf_idx, uint flags)
++{
++ FALCON_LOCK_DECL;
++ uint index, desc_type;
++ uint64_t val1, val2, val3;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* Q attributes */
++ int iscsi_hdig_en = ((flags & EFHW_VI_ISCSI_TX_HDIG_EN) != 0);
++ int iscsi_ddig_en = ((flags & EFHW_VI_ISCSI_TX_DDIG_EN) != 0);
++ int csum_ip_dis = ((flags & EFHW_VI_TX_IP_CSUM_DIS) != 0);
++ int csum_tcp_dis = ((flags & EFHW_VI_TX_TCPUDP_CSUM_DIS) != 0);
++ int non_ip_drop_dis = ((flags & EFHW_VI_TX_TCPUDP_ONLY) == 0);
++
++ /* initialise the TX descriptor queue pointer table */
++
++ /* NB physical vs buffer addressing is determined by the Queue ID. */
++
++ offset = falcon_dma_tx_q_offset(nic, dmaq);
++ index = falcon_dma_tx_q_size_index(dmaq_size);
++
++ /* allow VI flag to override this queue's descriptor type */
++ desc_type = (flags & EFHW_VI_TX_PHYS_ADDR_EN) ? 0 : 1;
++
++ /* bug9403: It is dangerous to allow buffer-addressed queues to
++ * have owner_id=0. */
++ EFHW_ASSERT((own_id > 0) || desc_type == 0);
++
++ /* dword 1 */
++ __DWCHCK(TX_DESCQ_FLUSH_LBN, TX_DESCQ_FLUSH_WIDTH);
++ __DWCHCK(TX_DESCQ_TYPE_LBN, TX_DESCQ_TYPE_WIDTH);
++ __DWCHCK(TX_DESCQ_SIZE_LBN, TX_DESCQ_SIZE_WIDTH);
++ __DWCHCK(TX_DESCQ_LABEL_LBN, TX_DESCQ_LABEL_WIDTH);
++ __DWCHCK(TX_DESCQ_OWNER_ID_LBN, TX_DESCQ_OWNER_ID_WIDTH);
++
++ __LWCHK(TX_DESCQ_EVQ_ID_LBN, TX_DESCQ_EVQ_ID_WIDTH);
++
++ __RANGECHCK(1, TX_DESCQ_FLUSH_WIDTH);
++ __RANGECHCK(desc_type, TX_DESCQ_TYPE_WIDTH);
++ __RANGECHCK(index, TX_DESCQ_SIZE_WIDTH);
++ __RANGECHCK(tag, TX_DESCQ_LABEL_WIDTH);
++ __RANGECHCK(own_id, TX_DESCQ_OWNER_ID_WIDTH);
++ __RANGECHCK(evq_id, TX_DESCQ_EVQ_ID_WIDTH);
++
++ val1 = ((desc_type << TX_DESCQ_TYPE_LBN) |
++ (index << TX_DESCQ_SIZE_LBN) |
++ (tag << TX_DESCQ_LABEL_LBN) |
++ (own_id << TX_DESCQ_OWNER_ID_LBN) |
++ (__LOW(evq_id, TX_DESCQ_EVQ_ID_LBN, TX_DESCQ_EVQ_ID_WIDTH)));
++
++ /* dword 2 */
++ __DW2CHCK(TX_DESCQ_BUF_BASE_ID_LBN, TX_DESCQ_BUF_BASE_ID_WIDTH);
++ __RANGECHCK(buf_idx, TX_DESCQ_BUF_BASE_ID_WIDTH);
++
++ val2 = ((__HIGH(evq_id, TX_DESCQ_EVQ_ID_LBN, TX_DESCQ_EVQ_ID_WIDTH)) |
++ (buf_idx << __DW2(TX_DESCQ_BUF_BASE_ID_LBN)));
++
++ /* dword 3 */
++ __DW3CHCK(TX_ISCSI_HDIG_EN_LBN, TX_ISCSI_HDIG_EN_WIDTH);
++ __DW3CHCK(TX_ISCSI_DDIG_EN_LBN, TX_ISCSI_DDIG_EN_WIDTH);
++ __RANGECHCK(iscsi_hdig_en, TX_ISCSI_HDIG_EN_WIDTH);
++ __RANGECHCK(iscsi_ddig_en, TX_ISCSI_DDIG_EN_WIDTH);
++
++ val3 = ((iscsi_hdig_en << __DW3(TX_ISCSI_HDIG_EN_LBN)) |
++ (iscsi_ddig_en << __DW3(TX_ISCSI_DDIG_EN_LBN)) |
++ (1 << __DW3(TX_DESCQ_EN_LBN))); /* queue enable bit */
++
++ switch (nic->devtype.variant) {
++ case 'B':
++ __DW3CHCK(TX_NON_IP_DROP_DIS_B0_LBN,
++ TX_NON_IP_DROP_DIS_B0_WIDTH);
++ __DW3CHCK(TX_IP_CHKSM_DIS_B0_LBN, TX_IP_CHKSM_DIS_B0_WIDTH);
++ __DW3CHCK(TX_TCP_CHKSM_DIS_B0_LBN, TX_TCP_CHKSM_DIS_B0_WIDTH);
++
++ val3 |= ((non_ip_drop_dis << __DW3(TX_NON_IP_DROP_DIS_B0_LBN))|
++ (csum_ip_dis << __DW3(TX_IP_CHKSM_DIS_B0_LBN)) |
++ (csum_tcp_dis << __DW3(TX_TCP_CHKSM_DIS_B0_LBN)));
++ break;
++ case 'A':
++ if (csum_ip_dis || csum_tcp_dis || !non_ip_drop_dis)
++ EFHW_WARN
++ ("%s: bad settings for A1 csum_ip_dis=%d "
++ "csum_tcp_dis=%d non_ip_drop_dis=%d",
++ __FUNCTION__, csum_ip_dis,
++ csum_tcp_dis, non_ip_drop_dis);
++ break;
++ default:
++ EFHW_ASSERT(0);
++ break;
++ }
++
++ EFHW_TRACE("%s: txq %x evq %u tag %x id %x buf %x "
++ "%x:%x:%x->%" PRIx64 ":%" PRIx64 ":%" PRIx64,
++ __FUNCTION__,
++ dmaq, evq_id, tag, own_id, buf_idx, dmaq_size,
++ iscsi_hdig_en, iscsi_ddig_en, val1, val2, val3);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, ((val2 << 32) | val1), val3);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++static inline ulong
++falcon_dma_rx_q_offset(struct efhw_nic *nic, unsigned dmaq)
++{
++ EFHW_ASSERT(dmaq < FALCON_DMAQ_NUM);
++ return RX_DESC_PTR_TBL_OFST + dmaq * FALCON_REGISTER128;
++}
++
++static void
++falcon_dmaq_rx_q_init(struct efhw_nic *nic,
++ uint dmaq, uint evq_id, uint own_id,
++ uint tag, uint dmaq_size, uint buf_idx, uint flags)
++{
++ FALCON_LOCK_DECL;
++ uint i, desc_type = 1;
++ uint64_t val1, val2, val3;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* Q attributes */
++#if BUG5762_WORKAROUND
++ int jumbo = 1; /* Queues must not have mixed types */
++#else
++ int jumbo = ((flags & EFHW_VI_JUMBO_EN) != 0);
++#endif
++ int iscsi_hdig_en = ((flags & EFHW_VI_ISCSI_RX_HDIG_EN) != 0);
++ int iscsi_ddig_en = ((flags & EFHW_VI_ISCSI_RX_DDIG_EN) != 0);
++
++ /* initialise the TX descriptor queue pointer table */
++ offset = falcon_dma_rx_q_offset(nic, dmaq);
++
++ /* size must be one of the various options, otherwise we assert */
++ for (i = 0; i < N_DMAQ_SIZES; i++) {
++ if (dmaq_size == dmaq_sizes[i])
++ break;
++ }
++ EFHW_ASSERT(i < N_DMAQ_SIZES);
++
++ /* allow VI flag to override this queue's descriptor type */
++ desc_type = (flags & EFHW_VI_RX_PHYS_ADDR_EN) ? 0 : 1;
++
++ /* bug9403: It is dangerous to allow buffer-addressed queues to have
++ * owner_id=0 */
++ EFHW_ASSERT((own_id > 0) || desc_type == 0);
++
++ /* dword 1 */
++ __DWCHCK(RX_DESCQ_EN_LBN, RX_DESCQ_EN_WIDTH);
++ __DWCHCK(RX_DESCQ_JUMBO_LBN, RX_DESCQ_JUMBO_WIDTH);
++ __DWCHCK(RX_DESCQ_TYPE_LBN, RX_DESCQ_TYPE_WIDTH);
++ __DWCHCK(RX_DESCQ_SIZE_LBN, RX_DESCQ_SIZE_WIDTH);
++ __DWCHCK(RX_DESCQ_LABEL_LBN, RX_DESCQ_LABEL_WIDTH);
++ __DWCHCK(RX_DESCQ_OWNER_ID_LBN, RX_DESCQ_OWNER_ID_WIDTH);
++
++ __LWCHK(RX_DESCQ_EVQ_ID_LBN, RX_DESCQ_EVQ_ID_WIDTH);
++
++ __RANGECHCK(1, RX_DESCQ_EN_WIDTH);
++ __RANGECHCK(jumbo, RX_DESCQ_JUMBO_WIDTH);
++ __RANGECHCK(desc_type, RX_DESCQ_TYPE_WIDTH);
++ __RANGECHCK(i, RX_DESCQ_SIZE_WIDTH);
++ __RANGECHCK(tag, RX_DESCQ_LABEL_WIDTH);
++ __RANGECHCK(own_id, RX_DESCQ_OWNER_ID_WIDTH);
++ __RANGECHCK(evq_id, RX_DESCQ_EVQ_ID_WIDTH);
++
++ val1 = ((1 << RX_DESCQ_EN_LBN) |
++ (jumbo << RX_DESCQ_JUMBO_LBN) |
++ (desc_type << RX_DESCQ_TYPE_LBN) |
++ (i << RX_DESCQ_SIZE_LBN) |
++ (tag << RX_DESCQ_LABEL_LBN) |
++ (own_id << RX_DESCQ_OWNER_ID_LBN) |
++ (__LOW(evq_id, RX_DESCQ_EVQ_ID_LBN, RX_DESCQ_EVQ_ID_WIDTH)));
++
++ /* dword 2 */
++ __DW2CHCK(RX_DESCQ_BUF_BASE_ID_LBN, RX_DESCQ_BUF_BASE_ID_WIDTH);
++ __RANGECHCK(buf_idx, RX_DESCQ_BUF_BASE_ID_WIDTH);
++
++ val2 = ((__HIGH(evq_id, RX_DESCQ_EVQ_ID_LBN, RX_DESCQ_EVQ_ID_WIDTH)) |
++ (buf_idx << __DW2(RX_DESCQ_BUF_BASE_ID_LBN)));
++
++ /* dword 3 */
++ __DW3CHCK(RX_ISCSI_HDIG_EN_LBN, RX_ISCSI_HDIG_EN_WIDTH);
++ __DW3CHCK(RX_ISCSI_DDIG_EN_LBN, RX_ISCSI_DDIG_EN_WIDTH);
++ __RANGECHCK(iscsi_hdig_en, RX_ISCSI_HDIG_EN_WIDTH);
++ __RANGECHCK(iscsi_ddig_en, RX_ISCSI_DDIG_EN_WIDTH);
++
++ val3 = (iscsi_hdig_en << __DW3(RX_ISCSI_HDIG_EN_LBN)) |
++ (iscsi_ddig_en << __DW3(RX_ISCSI_DDIG_EN_LBN));
++
++ EFHW_TRACE("%s: rxq %x evq %u tag %x id %x buf %x %s "
++ "%x:%x:%x -> %" PRIx64 ":%" PRIx64 ":%" PRIx64,
++ __FUNCTION__,
++ dmaq, evq_id, tag, own_id, buf_idx,
++ jumbo ? "jumbo" : "normal", dmaq_size,
++ iscsi_hdig_en, iscsi_ddig_en, val1, val2, val3);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, ((val2 << 32) | val1), val3);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++static void falcon_dmaq_tx_q_disable(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val1, val2, val3;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* initialise the TX descriptor queue pointer table */
++
++ offset = falcon_dma_tx_q_offset(nic, dmaq);
++
++ /* dword 1 */
++ __DWCHCK(TX_DESCQ_TYPE_LBN, TX_DESCQ_TYPE_WIDTH);
++
++ val1 = ((uint64_t) 1 << TX_DESCQ_TYPE_LBN);
++
++ /* dword 2 */
++ val2 = 0;
++
++ /* dword 3 */
++ val3 = (0 << __DW3(TX_DESCQ_EN_LBN)); /* queue enable bit */
++
++ EFHW_TRACE("%s: %x->%" PRIx64 ":%" PRIx64 ":%" PRIx64,
++ __FUNCTION__, dmaq, val1, val2, val3);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, ((val2 << 32) | val1), val3);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++static void falcon_dmaq_rx_q_disable(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val1, val2, val3;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* initialise the TX descriptor queue pointer table */
++ offset = falcon_dma_rx_q_offset(nic, dmaq);
++
++ /* dword 1 */
++ __DWCHCK(RX_DESCQ_EN_LBN, RX_DESCQ_EN_WIDTH);
++ __DWCHCK(RX_DESCQ_TYPE_LBN, RX_DESCQ_TYPE_WIDTH);
++
++ val1 = ((0 << RX_DESCQ_EN_LBN) | (1 << RX_DESCQ_TYPE_LBN));
++
++ /* dword 2 */
++ val2 = 0;
++
++ /* dword 3 */
++ val3 = 0;
++
++ EFHW_TRACE("falcon_dmaq_rx_q_disable: %x->%"
++ PRIx64 ":%" PRIx64 ":%" PRIx64,
++ dmaq, val1, val2, val3);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, ((val2 << 32) | val1), val3);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++
++/*----------------------------------------------------------------------------
++ *
++ * Buffer Table low-level register interface
++ *
++ *---------------------------------------------------------------------------*/
++
++/*! Convert a (potentially) 64-bit physical address to 32-bits. Every use
++** of this function is a place where we're not 64-bit clean.
++*/
++static inline uint32_t dma_addr_to_u32(dma_addr_t addr)
++{
++ /* Top bits had better be zero! */
++ EFHW_ASSERT(addr == (addr & 0xffffffff));
++ return (uint32_t) addr;
++}
++
++static inline uint32_t
++falcon_nic_buffer_table_entry32_mk(dma_addr_t dma_addr, int own_id)
++{
++ uint32_t dma_addr32 = FALCON_BUFFER_4K_PAGE(dma_addr_to_u32(dma_addr));
++
++ /* don't do this to me */
++ EFHW_BUILD_ASSERT(BUF_ADR_HBUF_ODD_LBN == BUF_ADR_HBUF_EVEN_LBN + 32);
++ EFHW_BUILD_ASSERT(BUF_OWNER_ID_HBUF_ODD_LBN ==
++ BUF_OWNER_ID_HBUF_EVEN_LBN + 32);
++
++ EFHW_BUILD_ASSERT(BUF_OWNER_ID_HBUF_ODD_WIDTH ==
++ BUF_OWNER_ID_HBUF_EVEN_WIDTH);
++ EFHW_BUILD_ASSERT(BUF_ADR_HBUF_ODD_WIDTH == BUF_ADR_HBUF_EVEN_WIDTH);
++
++ __DWCHCK(BUF_ADR_HBUF_EVEN_LBN, BUF_ADR_HBUF_EVEN_WIDTH);
++ __DWCHCK(BUF_OWNER_ID_HBUF_EVEN_LBN, BUF_OWNER_ID_HBUF_EVEN_WIDTH);
++
++ __RANGECHCK(dma_addr32, BUF_ADR_HBUF_EVEN_WIDTH);
++ __RANGECHCK(own_id, BUF_OWNER_ID_HBUF_EVEN_WIDTH);
++
++ return ((dma_addr32 << BUF_ADR_HBUF_EVEN_LBN) |
++ (own_id << BUF_OWNER_ID_HBUF_EVEN_LBN));
++}
++
++static inline uint64_t
++falcon_nic_buffer_table_entry64_mk(dma_addr_t dma_addr,
++ int bufsz, /* bytes */
++ int region, int own_id)
++{
++ __DW2CHCK(IP_DAT_BUF_SIZE_LBN, IP_DAT_BUF_SIZE_WIDTH);
++ __DW2CHCK(BUF_ADR_REGION_LBN, BUF_ADR_REGION_WIDTH);
++ __LWCHK(BUF_ADR_FBUF_LBN, BUF_ADR_FBUF_WIDTH);
++ __DWCHCK(BUF_OWNER_ID_FBUF_LBN, BUF_OWNER_ID_FBUF_WIDTH);
++
++ EFHW_ASSERT((bufsz == EFHW_4K) || (bufsz == EFHW_8K));
++
++ dma_addr = (dma_addr >> 12) & __FALCON_MASK64(BUF_ADR_FBUF_WIDTH);
++
++ __RANGECHCK(dma_addr, BUF_ADR_FBUF_WIDTH);
++ __RANGECHCK(1, IP_DAT_BUF_SIZE_WIDTH);
++ __RANGECHCK(region, BUF_ADR_REGION_WIDTH);
++ __RANGECHCK(own_id, BUF_OWNER_ID_FBUF_WIDTH);
++
++ return (((uint64_t) (bufsz == EFHW_8K) << IP_DAT_BUF_SIZE_LBN) |
++ ((uint64_t) region << BUF_ADR_REGION_LBN) |
++ ((uint64_t) dma_addr << BUF_ADR_FBUF_LBN) |
++ ((uint64_t) own_id << BUF_OWNER_ID_FBUF_LBN));
++}
++
++static inline void
++_falcon_nic_buffer_table_set32(struct efhw_nic *nic,
++ dma_addr_t dma_addr, uint bufsz,
++ uint region, /* not used */
++ int own_id, int buffer_id)
++{
++ /* programming the half table needs to be done in pairs. */
++ uint64_t entry, val, shift;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ efhw_ioaddr_t offset;
++
++ EFHW_BUILD_ASSERT(BUF_ADR_HBUF_ODD_LBN == BUF_ADR_HBUF_EVEN_LBN + 32);
++ EFHW_BUILD_ASSERT(BUF_OWNER_ID_HBUF_ODD_LBN ==
++ BUF_OWNER_ID_HBUF_EVEN_LBN + 32);
++
++ shift = (buffer_id & 1) ? 32 : 0;
++
++ offset = (efhw_kva + BUF_HALF_TBL_OFST +
++ ((buffer_id & ~1) * FALCON_BUFFER_TBL_HALF_BYTES));
++
++ entry = falcon_nic_buffer_table_entry32_mk(dma_addr_to_u32(dma_addr),
++ own_id);
++
++#if FALCON_USE_SHADOW_BUFFER_TABLE
++ val = _falcon_buffer_table[buffer_id & ~1];
++#else
++ /* This will not work unless we've completed
++ * the buffer table updates */
++ falcon_read_q(offset, &val);
++#endif
++ val &= ~(((uint64_t) 0xffffffff) << shift);
++ val |= (entry << shift);
++
++ EFHW_TRACE("%s[%x]: " ci_dma_addr_fmt ":%x:%" PRIx64 "->%x = %"
++ PRIx64, __FUNCTION__, buffer_id, dma_addr, own_id, entry,
++ (unsigned)(offset - efhw_kva), val);
++
++ /* Falcon requires that access to this register is serialised */
++ falcon_write_q(offset, val);
++
++ /* NB. No mmiowb(). Caller should do that e.g by calling commit */
++
++#if FALCON_USE_SHADOW_BUFFER_TABLE
++ _falcon_buffer_table[buffer_id & ~1] = val;
++#endif
++
++ /* Confirm the entry if the event queues haven't been set up. */
++ if (!nic->irq_handler) {
++ uint64_t new_val;
++ int count = 0;
++ while (1) {
++ mmiowb();
++ falcon_read_q(offset, &new_val);
++ if (new_val == val)
++ break;
++ count++;
++ if (count > 1000) {
++ EFHW_WARN("%s: poll Timeout", __FUNCTION__);
++ break;
++ }
++ udelay(1);
++ }
++ }
++}
++
++static inline void
++_falcon_nic_buffer_table_set64(struct efhw_nic *nic,
++ dma_addr_t dma_addr, uint bufsz,
++ uint region, int own_id, int buffer_id)
++{
++ efhw_ioaddr_t offset;
++ uint64_t entry;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_ASSERT(region < FALCON_REGION_NUM);
++
++ EFHW_ASSERT((bufsz == EFHW_4K) ||
++ (bufsz == EFHW_8K && FALCON_BUFFER_TABLE_FULL_MODE));
++
++ offset = (efhw_kva + BUF_FULL_TBL_OFST +
++ (buffer_id * FALCON_BUFFER_TBL_FULL_BYTES));
++
++ entry = falcon_nic_buffer_table_entry64_mk(dma_addr, bufsz, region,
++ own_id);
++
++ EFHW_TRACE("%s[%x]: " ci_dma_addr_fmt
++ ":bufsz=%x:region=%x:ownid=%x",
++ __FUNCTION__, buffer_id, dma_addr, bufsz, region, own_id);
++
++ EFHW_TRACE("%s: BUF[%x]:NIC[%x]->%" PRIx64,
++ __FUNCTION__, buffer_id,
++ (unsigned int)(offset - efhw_kva), entry);
++
++ /* Falcon requires that access to this register is serialised */
++ falcon_write_q(offset, entry);
++
++ /* NB. No mmiowb(). Caller should do that e.g by calling commit */
++
++ /* Confirm the entry if the event queues haven't been set up. */
++ if (!nic->irq_handler) {
++ uint64_t new_entry;
++ int count = 0;
++ while (1) {
++ mmiowb();
++ falcon_read_q(offset, &new_entry);
++ if (new_entry == entry)
++ return;
++ count++;
++ if (count > 1000) {
++ EFHW_WARN("%s: poll Timeout waiting for "
++ "value %"PRIx64
++ " (last was %"PRIx64")",
++ __FUNCTION__, entry, new_entry);
++ break;
++ }
++ udelay(1);
++ }
++ }
++}
++
++#if FALCON_BUFFER_TABLE_FULL_MODE
++#define _falcon_nic_buffer_table_set _falcon_nic_buffer_table_set64
++#else
++#define _falcon_nic_buffer_table_set _falcon_nic_buffer_table_set32
++#endif
++
++static inline void _falcon_nic_buffer_table_commit(struct efhw_nic *nic)
++{
++ /* MUST be called holding the FALCON_LOCK */
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint64_t cmd;
++
++ EFHW_BUILD_ASSERT(BUF_TBL_UPD_REG_KER_OFST == BUF_TBL_UPD_REG_OFST);
++
++ __DW2CHCK(BUF_UPD_CMD_LBN, BUF_UPD_CMD_WIDTH);
++ __RANGECHCK(1, BUF_UPD_CMD_WIDTH);
++
++ cmd = ((uint64_t) 1 << BUF_UPD_CMD_LBN);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ falcon_write_qq(efhw_kva + BUF_TBL_UPD_REG_OFST,
++ cmd, FALCON_ATOMIC_UPD_REG);
++ mmiowb();
++
++ nic->buf_commit_outstanding++;
++ EFHW_TRACE("COMMIT REQ out=%d", nic->buf_commit_outstanding);
++}
++
++static void falcon_nic_buffer_table_commit(struct efhw_nic *nic)
++{
++ /* nothing to do */
++}
++
++static inline void
++_falcon_nic_buffer_table_clear(struct efhw_nic *nic, int buffer_id, int num)
++{
++ uint64_t cmd;
++ uint64_t start_id = buffer_id;
++ uint64_t end_id = buffer_id + num - 1;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ efhw_ioaddr_t offset = (efhw_kva + BUF_TBL_UPD_REG_OFST);
++
++ EFHW_BUILD_ASSERT(BUF_TBL_UPD_REG_KER_OFST == BUF_TBL_UPD_REG_OFST);
++
++#if !FALCON_BUFFER_TABLE_FULL_MODE
++ /* buffer_ids in half buffer mode reference pairs of buffers */
++ EFHW_ASSERT(buffer_id % 1 == 0);
++ EFHW_ASSERT(num % 1 == 0);
++ start_id = start_id >> 1;
++ end_id = end_id >> 1;
++#endif
++
++ EFHW_ASSERT(num >= 1);
++
++ __DWCHCK(BUF_CLR_START_ID_LBN, BUF_CLR_START_ID_WIDTH);
++ __DW2CHCK(BUF_CLR_END_ID_LBN, BUF_CLR_END_ID_WIDTH);
++
++ __DW2CHCK(BUF_CLR_CMD_LBN, BUF_CLR_CMD_WIDTH);
++ __RANGECHCK(1, BUF_CLR_CMD_WIDTH);
++
++ __RANGECHCK(start_id, BUF_CLR_START_ID_WIDTH);
++ __RANGECHCK(end_id, BUF_CLR_END_ID_WIDTH);
++
++ cmd = (((uint64_t) 1 << BUF_CLR_CMD_LBN) |
++ (start_id << BUF_CLR_START_ID_LBN) |
++ (end_id << BUF_CLR_END_ID_LBN));
++
++ /* Falcon requires 128 bit atomic access for this register */
++ falcon_write_qq(offset, cmd, FALCON_ATOMIC_UPD_REG);
++ mmiowb();
++
++ nic->buf_commit_outstanding++;
++ EFHW_TRACE("COMMIT CLEAR out=%d", nic->buf_commit_outstanding);
++}
++
++/*----------------------------------------------------------------------------
++ *
++ * Events low-level register interface
++ *
++ *---------------------------------------------------------------------------*/
++
++static unsigned eventq_sizes[] = {
++ 512,
++ EFHW_1K,
++ EFHW_2K,
++ EFHW_4K,
++ EFHW_8K,
++ EFHW_16K,
++ EFHW_32K
++};
++
++#define N_EVENTQ_SIZES (sizeof(eventq_sizes) / sizeof(eventq_sizes[0]))
++
++static inline void falcon_nic_srm_upd_evq(struct efhw_nic *nic, int evq)
++{
++ /* set up the eventq which will receive events from the SRAM module.
++ * i.e buffer table updates and clears, TX and RX aperture table
++ * updates */
++
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_BUILD_ASSERT(SRM_UPD_EVQ_REG_OFST == SRM_UPD_EVQ_REG_KER_OFST);
++
++ EFHW_ASSERT((evq == FALCON_EVQ_KERNEL0) || (evq == FALCON_EVQ_CHAR) ||
++ (evq == FALCON_EVQ_NONIRQ));
++
++ __DWCHCK(SRM_UPD_EVQ_ID_LBN, SRM_UPD_EVQ_ID_WIDTH);
++ __RANGECHCK(evq, SRM_UPD_EVQ_ID_WIDTH);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + SRM_UPD_EVQ_REG_OFST,
++ ((uint64_t) evq << SRM_UPD_EVQ_ID_LBN),
++ FALCON_ATOMIC_SRPM_UDP_EVQ_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++static inline void
++falcon_nic_evq_ptr_tbl(struct efhw_nic *nic,
++ uint evq, /* evq id */
++ uint enable, /* 1 to enable, 0 to disable */
++ uint buf_base_id,/* Buffer table base for EVQ */
++ uint evq_size /* Number of events */ )
++{
++ FALCON_LOCK_DECL;
++ uint i, val;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* size must be one of the various options, otherwise we assert */
++ for (i = 0; i < N_EVENTQ_SIZES; i++) {
++ if (evq_size <= eventq_sizes[i])
++ break;
++ }
++ EFHW_ASSERT(i < N_EVENTQ_SIZES);
++
++ __DWCHCK(EVQ_BUF_BASE_ID_LBN, EVQ_BUF_BASE_ID_WIDTH);
++ __DWCHCK(EVQ_SIZE_LBN, EVQ_SIZE_WIDTH);
++ __DWCHCK(EVQ_EN_LBN, EVQ_EN_WIDTH);
++
++ __RANGECHCK(i, EVQ_SIZE_WIDTH);
++ __RANGECHCK(buf_base_id, EVQ_BUF_BASE_ID_WIDTH);
++ __RANGECHCK(1, EVQ_EN_WIDTH);
++
++ /* if !enable then only evq needs to be correct, although valid
++ * values need to be passed in for other arguments to prevent
++ * assertions */
++
++ val = ((i << EVQ_SIZE_LBN) | (buf_base_id << EVQ_BUF_BASE_ID_LBN) |
++ (enable ? (1 << EVQ_EN_LBN) : 0));
++
++ EFHW_ASSERT(evq < FALCON_EVQ_TBL_NUM);
++
++ offset = EVQ_PTR_TBL_CHAR_OFST;
++ offset += evq * FALCON_REGISTER128;
++
++ EFHW_TRACE("%s: evq %u en=%x:buf=%x:size=%x->%x at %lx",
++ __FUNCTION__, evq, enable, buf_base_id, evq_size, val,
++ offset);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, val, FALCON_ATOMIC_PTR_TBL_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++
++ /* caller must wait for an update done event before writing any more
++ table entries */
++
++ return;
++}
++
++void
++falcon_nic_evq_ack(struct efhw_nic *nic,
++ uint evq, /* evq id */
++ uint rptr, /* new read pointer update */
++ bool wakeup /* request a wakeup event if ptr's != */
++ )
++{
++ uint val;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_BUILD_ASSERT(FALCON_EVQ_CHAR == 4);
++
++ __DWCHCK(EVQ_RPTR_LBN, EVQ_RPTR_WIDTH);
++ __RANGECHCK(rptr, EVQ_RPTR_WIDTH);
++
++ val = (rptr << EVQ_RPTR_LBN);
++
++ EFHW_ASSERT(evq < FALCON_EVQ_TBL_NUM);
++
++ if (evq < FALCON_EVQ_CHAR) {
++ offset = EVQ_RPTR_REG_KER_OFST;
++ offset += evq * FALCON_REGISTER128;
++
++ EFHW_ASSERT(!wakeup); /* don't try this at home */
++ } else {
++ offset = EVQ_RPTR_REG_OFST + (FALCON_EVQ_CHAR *
++ FALCON_REGISTER128);
++ offset += (evq - FALCON_EVQ_CHAR) * FALCON_REGISTER128;
++
++ /* nothing to do for interruptless event queues which do
++ * not want a wakeup */
++ if (evq != FALCON_EVQ_CHAR && !wakeup)
++ return;
++ }
++
++ EFHW_TRACE("%s: %x %x %x->%x", __FUNCTION__, evq, rptr, wakeup, val);
++
++ writel(val, efhw_kva + offset);
++ mmiowb();
++}
++
++/*----------------------------------------------------------------------------
++ *
++ * Helper for evq mapping
++ *
++ * idx = 0 && char => hw eventq[4]
++ * idx = 0 && net => hw eventq[0]
++ * 0 < idx < 5 => hw eventq[idx] (5 is non-interrupting)
++ *
++ *
++ *---------------------------------------------------------------------------*/
++
++int falcon_idx_to_evq(struct efhw_nic *nic, uint idx)
++{
++ EFHW_BUILD_ASSERT(FALCON_EVQ_CHAR == 4);
++ EFHW_ASSERT(idx <= FALCON_EVQ_NONIRQ);
++ return (idx > 0) ? idx : FALCON_EVQ_CHAR;
++}
++
++static inline int falcon_evq_is_interrupting(struct efhw_nic *nic, uint idx)
++{
++ EFHW_BUILD_ASSERT(FALCON_EVQ_CHAR == 4);
++ EFHW_ASSERT(idx <= FALCON_EVQ_NONIRQ);
++
++ /* only the first CHAR driver event queue is interrupting */
++ return (idx == FALCON_EVQ_CHAR);
++}
++
++static inline void
++falcon_drv_ev(struct efhw_nic *nic, uint64_t data, uint qid)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ /* send an event from one driver to the other */
++ EFHW_BUILD_ASSERT(DRV_EV_REG_KER_OFST == DRV_EV_REG_OFST);
++ EFHW_BUILD_ASSERT(DRV_EV_DATA_LBN == 0);
++ EFHW_BUILD_ASSERT(DRV_EV_DATA_WIDTH == 64);
++ EFHW_BUILD_ASSERT(DRV_EV_QID_LBN == 64);
++ EFHW_BUILD_ASSERT(DRV_EV_QID_WIDTH == 12);
++
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + DRV_EV_REG_OFST, data, qid);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++_DEBUG_SYM_ void
++falcon_timer_cmd(struct efhw_nic *nic,
++ uint evq, /* timer id */
++ uint mode, /* mode bits */
++ uint countdown /* counting value to set */ )
++{
++ FALCON_LOCK_DECL;
++ uint val;
++ ulong offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_BUILD_ASSERT(TIMER_VAL_LBN == 0);
++
++ __DWCHCK(TIMER_MODE_LBN, TIMER_MODE_WIDTH);
++ __DWCHCK(TIMER_VAL_LBN, TIMER_VAL_WIDTH);
++
++ __RANGECHCK(mode, TIMER_MODE_WIDTH);
++ __RANGECHCK(countdown, TIMER_VAL_WIDTH);
++
++ val = ((mode << TIMER_MODE_LBN) | (countdown << TIMER_VAL_LBN));
++
++ if (evq < FALCON_EVQ_CHAR) {
++ offset = TIMER_CMD_REG_KER_OFST;
++ offset += evq * EFHW_8K; /* PAGE mapped register */
++ } else {
++ offset = TIMER_TBL_OFST;
++ offset += evq * FALCON_REGISTER128;
++ }
++ EFHW_ASSERT(evq < FALCON_EVQ_TBL_NUM);
++
++ EFHW_TRACE("%s: evq %u mode %x (%s) time %x -> %08x",
++ __FUNCTION__, evq, mode,
++ mode == 0 ? "DISABLE" :
++ mode == 1 ? "IMMED" :
++ mode == 2 ? (evq < 5 ? "HOLDOFF" : "RX_TRIG") :
++ "<BAD>", countdown, val);
++
++ /* Falcon requires 128 bit atomic access for this register when
++ * accessed from the driver. User access to timers is paged mapped
++ */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, val, FALCON_ATOMIC_TIMER_CMD_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Rate pacing - Low level interface
++ *
++ *--------------------------------------------------------------------*/
++void falcon_nic_pace(struct efhw_nic *nic, uint dmaq, uint pace)
++{
++ /* Pace specified in 2^(units of microseconds). This is the minimum
++ additional delay imposed over and above the IPG.
++
++ Pacing only available on the virtual interfaces
++ */
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ ulong offset;
++
++ if (pace > 20)
++ pace = 20; /* maxm supported value */
++
++ __DWCHCK(TX_PACE_LBN, TX_PACE_WIDTH);
++ __RANGECHCK(pace, TX_PACE_WIDTH);
++
++ switch (nic->devtype.variant) {
++ case 'A':
++ EFHW_ASSERT(dmaq >= TX_PACE_TBL_FIRST_QUEUE_A1);
++ offset = TX_PACE_TBL_A1_OFST;
++ offset += (dmaq - TX_PACE_TBL_FIRST_QUEUE_A1) * 16;
++ break;
++ case 'B':
++ /* Would be nice to assert this, but as dmaq is unsigned and
++ * TX_PACE_TBL_FIRST_QUEUE_B0 is 0, it makes no sense
++ * EFHW_ASSERT(dmaq >= TX_PACE_TBL_FIRST_QUEUE_B0);
++ */
++ offset = TX_PACE_TBL_B0_OFST;
++ offset += (dmaq - TX_PACE_TBL_FIRST_QUEUE_B0) * 16;
++ break;
++ default:
++ EFHW_ASSERT(0);
++ offset = 0;
++ break;
++ }
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + offset, pace, FALCON_ATOMIC_PACE_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++
++ EFHW_TRACE("%s: txq %d offset=%lx pace=2^%x",
++ __FUNCTION__, dmaq, offset, pace);
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Interrupt - Low level interface
++ *
++ *--------------------------------------------------------------------*/
++
++static void falcon_nic_handle_fatal_int(struct efhw_nic *nic)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint64_t val;
++
++ offset = (efhw_kva + FATAL_INTR_REG_OFST);
++
++ /* Falcon requires 32 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ val = readl(offset);
++ FALCON_LOCK_UNLOCK(nic);
++
++ /* ?? BUG3249 - need to disable illegal address interrupt */
++ /* ?? BUG3114 - need to backport interrupt storm protection code */
++ EFHW_ERR("fatal interrupt: %s%s%s%s%s%s%s%s%s%s%s%s[%" PRIx64 "]",
++ val & (1 << PCI_BUSERR_INT_CHAR_LBN) ? "PCI-bus-error " : "",
++ val & (1 << SRAM_OOB_INT_CHAR_LBN) ? "SRAM-oob " : "",
++ val & (1 << BUFID_OOB_INT_CHAR_LBN) ? "bufid-oob " : "",
++ val & (1 << MEM_PERR_INT_CHAR_LBN) ? "int-parity " : "",
++ val & (1 << RBUF_OWN_INT_CHAR_LBN) ? "rx-bufid-own " : "",
++ val & (1 << TBUF_OWN_INT_CHAR_LBN) ? "tx-bufid-own " : "",
++ val & (1 << RDESCQ_OWN_INT_CHAR_LBN) ? "rx-desc-own " : "",
++ val & (1 << TDESCQ_OWN_INT_CHAR_LBN) ? "tx-desc-own " : "",
++ val & (1 << EVQ_OWN_INT_CHAR_LBN) ? "evq-own " : "",
++ val & (1 << EVFF_OFLO_INT_CHAR_LBN) ? "evq-fifo " : "",
++ val & (1 << ILL_ADR_INT_CHAR_LBN) ? "ill-addr " : "",
++ val & (1 << SRM_PERR_INT_CHAR_LBN) ? "sram-parity " : "", val);
++}
++
++static void falcon_nic_interrupt_hw_enable(struct efhw_nic *nic)
++{
++ FALCON_LOCK_DECL;
++ uint val;
++ efhw_ioaddr_t offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_BUILD_ASSERT(DRV_INT_EN_CHAR_WIDTH == 1);
++
++ if (nic->flags & NIC_FLAG_NO_INTERRUPT)
++ return;
++
++ offset = (efhw_kva + INT_EN_REG_CHAR_OFST);
++ val = 1 << DRV_INT_EN_CHAR_LBN;
++
++ EFHW_NOTICE("%s: %x -> %x", __FUNCTION__, (int)(offset - efhw_kva),
++ val);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(offset, val, FALCON_ATOMIC_INT_EN_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++static void falcon_nic_interrupt_hw_disable(struct efhw_nic *nic)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ EFHW_BUILD_ASSERT(SRAM_PERR_INT_KER_WIDTH == 1);
++ EFHW_BUILD_ASSERT(DRV_INT_EN_KER_LBN == 0);
++ EFHW_BUILD_ASSERT(SRAM_PERR_INT_CHAR_WIDTH == 1);
++ EFHW_BUILD_ASSERT(DRV_INT_EN_CHAR_LBN == 0);
++ EFHW_BUILD_ASSERT(SRAM_PERR_INT_KER_LBN == SRAM_PERR_INT_CHAR_LBN);
++ EFHW_BUILD_ASSERT(DRV_INT_EN_KER_LBN == DRV_INT_EN_CHAR_LBN);
++
++ if (nic->flags & NIC_FLAG_NO_INTERRUPT)
++ return;
++
++ offset = (efhw_kva + INT_EN_REG_CHAR_OFST);
++
++ EFHW_NOTICE("%s: %x -> 0", __FUNCTION__, (int)(offset - efhw_kva));
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(offset, 0, FALCON_ATOMIC_INT_EN_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++#ifndef __ci_ul_driver__
++
++static void falcon_nic_irq_addr_set(struct efhw_nic *nic, dma_addr_t dma_addr)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t offset;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++
++ offset = (efhw_kva + INT_ADR_REG_CHAR_OFST);
++
++ EFHW_NOTICE("%s: %x -> " DMA_ADDR_T_FMT, __FUNCTION__,
++ (int)(offset - efhw_kva), dma_addr);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(offset, dma_addr, FALCON_ATOMIC_INT_ADR_REG);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++#endif
++
++
++/*--------------------------------------------------------------------
++ *
++ * RXDP - low level interface
++ *
++ *--------------------------------------------------------------------*/
++
++void
++falcon_nic_set_rx_usr_buf_size(struct efhw_nic *nic, int usr_buf_bytes)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint64_t val, val2, usr_buf_size = usr_buf_bytes / 32;
++ int rubs_lbn, rubs_width, roec_lbn;
++
++ EFHW_BUILD_ASSERT(RX_CFG_REG_OFST == RX_CFG_REG_KER_OFST);
++
++ switch (nic->devtype.variant) {
++ default:
++ EFHW_ASSERT(0);
++ /* Fall-through to avoid compiler warnings. */
++ case 'A':
++ rubs_lbn = RX_USR_BUF_SIZE_A1_LBN;
++ rubs_width = RX_USR_BUF_SIZE_A1_WIDTH;
++ roec_lbn = RX_OWNERR_CTL_A1_LBN;
++ break;
++ case 'B':
++ rubs_lbn = RX_USR_BUF_SIZE_B0_LBN;
++ rubs_width = RX_USR_BUF_SIZE_B0_WIDTH;
++ roec_lbn = RX_OWNERR_CTL_B0_LBN;
++ break;
++ }
++
++ __DWCHCK(rubs_lbn, rubs_width);
++ __QWCHCK(roec_lbn, 1);
++ __RANGECHCK(usr_buf_size, rubs_width);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + RX_CFG_REG_OFST, &val, &val2);
++
++ val &= ~((__FALCON_MASK64(rubs_width)) << rubs_lbn);
++ val |= (usr_buf_size << rubs_lbn);
++
++ /* shouldn't be needed for a production driver */
++ val |= ((uint64_t) 1 << roec_lbn);
++
++ falcon_write_qq(efhw_kva + RX_CFG_REG_OFST, val, val2);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++EXPORT_SYMBOL(falcon_nic_set_rx_usr_buf_size);
++
++void
++falcon_nic_rx_filter_ctl_get(struct efhw_nic *nic, uint32_t *tcp_full,
++ uint32_t *tcp_wild,
++ uint32_t *udp_full, uint32_t *udp_wild)
++{
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ FALCON_LOCK_DECL;
++ uint64_t val;
++
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_q(efhw_kva + RX_FILTER_CTL_REG_OFST, &val);
++ FALCON_LOCK_UNLOCK(nic);
++
++ *tcp_full = (uint32_t)((val >> TCP_FULL_SRCH_LIMIT_LBN) &
++ (__FALCON_MASK64(TCP_FULL_SRCH_LIMIT_WIDTH)));
++
++ *tcp_wild = (uint32_t)((val >> TCP_WILD_SRCH_LIMIT_LBN) &
++ (__FALCON_MASK64(TCP_WILD_SRCH_LIMIT_WIDTH)));
++
++ *udp_full = (uint32_t)((val >> UDP_FULL_SRCH_LIMIT_LBN) &
++ (__FALCON_MASK64(UDP_FULL_SRCH_LIMIT_WIDTH)));
++
++ *udp_wild = (uint32_t)((val >> UDP_WILD_SRCH_LIMIT_LBN) &
++ (__FALCON_MASK64(UDP_WILD_SRCH_LIMIT_WIDTH)));
++}
++EXPORT_SYMBOL(falcon_nic_rx_filter_ctl_get);
++
++void
++falcon_nic_rx_filter_ctl_set(struct efhw_nic *nic, uint32_t tcp_full,
++ uint32_t tcp_wild,
++ uint32_t udp_full, uint32_t udp_wild)
++{
++ uint64_t val, val2;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ FALCON_LOCK_DECL;
++
++ EFHW_ASSERT(tcp_full < nic->filter_tbl_size);
++ EFHW_ASSERT(tcp_wild < nic->filter_tbl_size);
++ EFHW_ASSERT(udp_full < nic->filter_tbl_size);
++ EFHW_ASSERT(udp_wild < nic->filter_tbl_size);
++
++ /* until we implement a dynamic scaling of search limits we wish to
++ * maintain the same limits set up by default in the net driver
++ * when we initialize the char driver */
++ tcp_full_srch_limit = tcp_full;
++ tcp_wild_srch_limit = tcp_wild;
++ udp_full_srch_limit = udp_full;
++ udp_wild_srch_limit = udp_wild;
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + RX_FILTER_CTL_REG_OFST, &val, &val2);
++
++ /* Search limits */
++ val &= ~((__FALCON_MASK64(TCP_FULL_SRCH_LIMIT_WIDTH))
++ << TCP_FULL_SRCH_LIMIT_LBN);
++
++ val |= ((uint64_t)tcp_full + RX_FILTER_CTL_SRCH_FUDGE_FULL)
++ << TCP_FULL_SRCH_LIMIT_LBN;
++
++ val &= ~((__FALCON_MASK64(TCP_WILD_SRCH_LIMIT_WIDTH))
++ << TCP_WILD_SRCH_LIMIT_LBN);
++
++ val |= ((uint64_t)tcp_wild + RX_FILTER_CTL_SRCH_FUDGE_WILD)
++ << TCP_WILD_SRCH_LIMIT_LBN;
++
++ val &= ~((__FALCON_MASK64(UDP_FULL_SRCH_LIMIT_WIDTH))
++ << UDP_FULL_SRCH_LIMIT_LBN);
++
++ val |= ((uint64_t)udp_full + RX_FILTER_CTL_SRCH_FUDGE_FULL)
++ << UDP_FULL_SRCH_LIMIT_LBN;
++
++ val &= ~((__FALCON_MASK64(UDP_WILD_SRCH_LIMIT_WIDTH))
++ << UDP_WILD_SRCH_LIMIT_LBN);
++
++ val |= ((uint64_t)udp_wild + RX_FILTER_CTL_SRCH_FUDGE_WILD)
++ << UDP_WILD_SRCH_LIMIT_LBN;
++
++ falcon_write_qq(efhw_kva + RX_FILTER_CTL_REG_OFST, val, val2);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++EXPORT_SYMBOL(falcon_nic_rx_filter_ctl_set);
++
++/*--------------------------------------------------------------------
++ *
++ * TXDP - low level interface
++ *
++ *--------------------------------------------------------------------*/
++
++_DEBUG_SYM_ void falcon_nic_tx_cfg(struct efhw_nic *nic, int unlocked)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint64_t val1, val2;
++
++ EFHW_BUILD_ASSERT(TX_CFG_REG_OFST == TX_CFG_REG_KER_OFST);
++ __DWCHCK(TX_OWNERR_CTL_LBN, TX_OWNERR_CTL_WIDTH);
++ __DWCHCK(TX_NON_IP_DROP_DIS_LBN, TX_NON_IP_DROP_DIS_WIDTH);
++
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + TX_CFG_REG_OFST, &val1, &val2);
++
++ /* Will flag fatal interrupts on owner id errors. This should not be
++ on for production code because there is otherwise a denial of
++ serivce attack possible */
++ val1 |= (1 << TX_OWNERR_CTL_LBN);
++
++ /* Setup user queue TCP/UDP only packet security */
++ if (unlocked)
++ val1 |= (1 << TX_NON_IP_DROP_DIS_LBN);
++ else
++ val1 &= ~(1 << TX_NON_IP_DROP_DIS_LBN);
++
++ falcon_write_qq(efhw_kva + TX_CFG_REG_OFST, val1, val2);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Random thresholds - Low level interface (Would like these to be op
++ * defaults wherever possible)
++ *
++ *--------------------------------------------------------------------*/
++
++static void falcon_nic_pace_cfg(struct efhw_nic *nic)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ unsigned offset = 0;
++ uint64_t val;
++
++ val = 0xa81682; /* !!!! */
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ switch (nic->devtype.variant) {
++ case 'A': offset = TX_PACE_REG_A1_OFST; break;
++ case 'B': offset = TX_PACE_REG_B0_OFST; break;
++ default: EFHW_ASSERT(0); break;
++ }
++ falcon_write_qq(efhw_kva + offset, val, 0);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++/**********************************************************************
++ * Supporting modules. ************************************************
++ **********************************************************************/
++
++/*--------------------------------------------------------------------
++ *
++ * Filter support
++ *
++ *--------------------------------------------------------------------*/
++
++/*! \TODO this table should be per nic */
++struct falcon_cached_ipfilter {
++#if FALCON_FULL_FILTER_CACHE
++ unsigned dmaq;
++ unsigned saddr_le32;
++ unsigned daddr_le32;
++ unsigned sport_le16;
++ unsigned dport_le16;
++ unsigned tcp:1;
++ unsigned full:1;
++ unsigned rss_b0:1;
++ unsigned scat_b0:1;
++#endif
++ unsigned addr_valid:1;
++
++};
++
++
++/* TODO: Dynamically allocate this and store in struct efhw_nic. */
++static struct falcon_cached_ipfilter
++ host_ipfilter_cache[EFHW_MAX_NR_DEVS][FALCON_FILTER_TBL_NUM];
++
++
++static inline void host_ipfilter_cache_init(struct efhw_nic *nic)
++{
++ memset(host_ipfilter_cache[nic->index], 0,
++ sizeof(host_ipfilter_cache[0][0]) * nic->filter_tbl_size);
++}
++
++static inline int host_ipfilter_cache_active(struct efhw_nic *nic, uint idx)
++{
++ EFHW_ASSERT(nic->index < EFHW_MAX_NR_DEVS);
++ EFHW_ASSERT(idx < nic->filter_tbl_size);
++
++ return (host_ipfilter_cache[nic->index][idx].addr_valid);
++
++}
++
++static inline void host_ipfilter_cache_flush(struct efhw_nic *nic, uint idx)
++{
++ EFHW_ASSERT(nic->index < EFHW_MAX_NR_DEVS);
++ EFHW_ASSERT(idx < nic->filter_tbl_size);
++
++ memset(&host_ipfilter_cache[nic->index][idx], 0,
++ sizeof(struct falcon_cached_ipfilter));
++ mmiowb();
++}
++
++static inline void
++host_ipfilter_cache_set_addr(struct efhw_nic *nic, uint idx, uint dmaq,
++ unsigned tcp, unsigned full,
++ unsigned rss_b0, unsigned scat_b0,
++ unsigned saddr_le32, unsigned sport_le16,
++ unsigned daddr_le32, unsigned dport_le16)
++{
++ unsigned nic_i = nic->index;
++
++ EFHW_ASSERT(nic_i < EFHW_MAX_NR_DEVS);
++ EFHW_ASSERT(idx < nic->filter_tbl_size);
++ EFHW_ASSERT(!host_ipfilter_cache[nic_i][idx].addr_valid);
++
++ __RANGECHCK(sport_le16, SRC_TCP_DEST_UDP_1_WIDTH);
++ __RANGECHCK(dport_le16, SRC_TCP_DEST_UDP_1_WIDTH);
++
++#if FALCON_FULL_FILTER_CACHE
++ host_ipfilter_cache[nic_i][idx].dmaq = dmaq;
++ host_ipfilter_cache[nic_i][idx].saddr_le32 = saddr_le32;
++ host_ipfilter_cache[nic_i][idx].daddr_le32 = daddr_le32;
++ host_ipfilter_cache[nic_i][idx].sport_le16 = sport_le16;
++ host_ipfilter_cache[nic_i][idx].dport_le16 = dport_le16;
++ host_ipfilter_cache[nic_i][idx].tcp = tcp;
++ host_ipfilter_cache[nic_i][idx].full = full;
++ host_ipfilter_cache[nic_i][idx].rss_b0 = rss_b0;
++ host_ipfilter_cache[nic_i][idx].scat_b0 = scat_b0;
++#endif
++ host_ipfilter_cache[nic_i][idx].addr_valid = 1;
++ mmiowb();
++}
++
++#if FALCON_VERIFY_FILTERS
++/* Check that all active filters still exist by reading from H/W */
++static void _falcon_nic_ipfilter_sanity(struct efhw_nic *nic)
++{
++ unsigned i;
++ struct falcon_cached_ipfilter *f;
++ uint64_t q0_expect, q1_expect, q0_got, q1_got;
++
++ for (i = 0; i < nic->filter_tbl_size; i++) {
++ f = host_ipfilter_cache[nic->index] + i;
++ if (!f->addr_valid)
++ continue;
++
++ _falcon_nic_ipfilter_build(nic, f->tcp, f->full,
++ f->rss_b0, f->scat_b0, i, f->dmaq,
++ f->saddr_le32, f->sport_le16,
++ f->daddr_le32, f->dport_le16,
++ &q0_expect, &q1_expect);
++
++ falcon_read_qq(EFHW_KVA(nic) + RX_FILTER_TBL0_OFST +
++ i * 2 * FALCON_REGISTER128,
++ &q0_got, &q1_got);
++
++ if ((q0_got != q0_expect) || (q1_got != q1_expect)) {
++ falcon_write_qq(EFHW_KVA(nic) + 0x300,
++ q0_got, q1_got);
++ EFHW_ERR("ERROR: RX-filter[%d][%d] was "
++ "%"PRIx64":%" PRIx64" expected "
++ "%"PRIx64":%"PRIx64,
++ nic->index, i, q0_got, q1_got,
++ q0_expect, q1_expect);
++ }
++ }
++}
++#endif /* FALCON_VERIFY_FILTERS */
++
++#if FALCON_FULL_FILTER_CACHE
++static inline int
++host_ipfilter_cache_check_not(uint nic, uint idx, int tcp, int full,
++ unsigned saddr_le32, unsigned sport_le16,
++ unsigned daddr_le32, unsigned dport_le16)
++{
++ return ((host_ipfilter_cache[nic][idx].saddr_le32 != saddr_le32) ||
++ (host_ipfilter_cache[nic][idx].daddr_le32 != daddr_le32) ||
++ (host_ipfilter_cache[nic][idx].sport_le16 != sport_le16) ||
++ (host_ipfilter_cache[nic][idx].dport_le16 != dport_le16) ||
++ (host_ipfilter_cache[nic][idx].tcp != tcp) ||
++ (host_ipfilter_cache[nic][idx].full != full));
++}
++#endif
++
++#define host_ipfilter_cache_saddr_le32(nic, idx) \
++ host_ipfilter_cache[nic][idx].saddr_le32
++#define host_ipfilter_cache_daddr_le32(nic, idx) \
++ host_ipfilter_cache[nic][idx].daddr_le32
++#define host_ipfilter_cache_sport_le16(nic, idx) \
++ host_ipfilter_cache[nic][idx].sport_le16
++#define host_ipfilter_cache_dport_le16(nic, idx) \
++ host_ipfilter_cache[nic][idx].dport_le16
++#define host_ipfilter_cache_tcp(nic, idx) \
++ host_ipfilter_cache[nic][idx].tcp
++#define host_ipfilter_cache_full(nic, idx) \
++ host_ipfilter_cache[nic][idx].full
++
++/**********************************************************************
++ * Implementation of the HAL. ********************************************
++ **********************************************************************/
++
++/*----------------------------------------------------------------------------
++ *
++ * Initialisation and configuration discovery
++ *
++ *---------------------------------------------------------------------------*/
++
++#ifdef __ci_ul_driver__
++
++static int falcon_nic_init_irq_channel(struct efhw_nic *nic, int enable)
++{
++ EFHW_ERR("%s: not implemented for ul driver", __FUNCTION__);
++ return -EOPNOTSUPP;
++}
++
++#else
++
++static int falcon_nic_init_irq_channel(struct efhw_nic *nic, int enable)
++{
++ /* create a buffer for the irq channel */
++ int rc;
++
++ if (enable) {
++ rc = efhw_iopage_alloc(nic, &nic->irq_iobuff);
++ if (rc < 0)
++ return rc;
++
++ falcon_nic_irq_addr_set(nic,
++ efhw_iopage_dma_addr(&nic->irq_iobuff));
++ } else {
++ if (efhw_iopage_is_valid(&nic->irq_iobuff))
++ efhw_iopage_free(nic, &nic->irq_iobuff);
++
++ efhw_iopage_mark_invalid(&nic->irq_iobuff);
++ falcon_nic_irq_addr_set(nic, 0);
++ }
++
++ EFHW_TRACE("%s: " ci_dma_addr_fmt " %sable", __FUNCTION__,
++ efhw_iopage_dma_addr(&nic->irq_iobuff), enable ?
++ "en" : "dis");
++
++ return 0;
++}
++
++#endif
++
++static void falcon_nic_close_hardware(struct efhw_nic *nic)
++{
++ /* check we are in possession of some hardware */
++ if (!efhw_nic_have_hw(nic))
++ return;
++
++ falcon_nic_init_irq_channel(nic, 0);
++
++ EFHW_NOTICE("%s:", __FUNCTION__);
++}
++
++#ifdef __ci_ul_driver__
++extern
++#else
++static
++#endif
++int falcon_nic_get_mac_config(struct efhw_nic *nic)
++{
++ efhw_ioaddr_t efhw_kva = nic->bar_ioaddr;
++ int is_mac_type_1g;
++ uint32_t strap, altera;
++ uint64_t rx_cfg, r;
++
++ altera = readl(efhw_kva + ALTERA_BUILD_REG_OFST);
++ strap = readl(efhw_kva + STRAP_REG_KER_OFST) & 0x7;
++
++ switch (nic->devtype.variant) {
++ case 'A':
++ if ((altera & 0x0fff0000) == 0x1130000) {
++ strap = 2; /* FPGA - PCI-X 2G */
++ } else if ((altera & 0x00ff0000) == 0x140000) {
++ /* should be 114 */
++ strap = 4; /* FPGA - PCI-X 4G */
++ } else if (strap < 2 || strap > 5) {
++ EFHW_ERR("Invalid strap option %d altera_buid_ver=%x",
++ strap, altera);
++ return -EINVAL;
++ }
++ is_mac_type_1g = (0 != (strap & 2));
++ break;
++ case 'B':
++ /* Runtime check that the hardware and software agree about
++ * the size of the RXFIFO. Write binary 11 across the left
++ * most bit, and assert we get 1 back.
++ */
++ r = 1LL << RX_TOEP_TCP_SUPPRESS_B0_LBN;
++ r |= (r << 1);
++
++ /* Save the original value */
++ falcon_read_q(efhw_kva + RX_CFG_REG_OFST, &rx_cfg);
++
++ /* Write and ready the dummy value */
++ falcon_write_qq(efhw_kva + RX_CFG_REG_OFST, r, 0);
++ falcon_read_q(efhw_kva + RX_CFG_REG_OFST, &r);
++
++ /* Restore the original value */
++ falcon_write_qq(efhw_kva + RX_CFG_REG_OFST, rx_cfg, 0);
++
++ if (r != (1LL << RX_TOEP_TCP_SUPPRESS_B0_LBN)) {
++ EFHW_ERR("The FPGA build (%x) RXFIFO size does not "
++ "match the software", altera);
++ return -EINVAL;
++ }
++ is_mac_type_1g = (0 != (strap & 2));
++#if FALCON_MAC_SET_TYPE_BY_SPEED
++ /* Check the selected strap pins against the MAC speed -
++ * and adjust if necessary.
++ */
++ {
++ int speed;
++ speed = readl(efhw_kva + MAC0_CTRL_REG_OFST) & 0x3;
++ is_mac_type_1g = (speed <= 2);
++ }
++#endif
++ break;
++ default:
++ EFHW_ASSERT(0);
++ is_mac_type_1g = 0;
++ break;
++ }
++
++ nic->fpga_version = altera;
++
++ /* We can now set the MAC type correctly based on the strap pins. */
++ if (is_mac_type_1g) {
++ nic->flags &= ~NIC_FLAG_10G;
++ } else {
++ /* strap & 4 must be set according to checks above */
++ nic->flags |= NIC_FLAG_10G;
++ }
++ EFHW_NOTICE("Board has %s MAC: strap=%d",
++ 0 != (nic->flags & NIC_FLAG_10G) ? "10G" : "1G", strap);
++ return 0;
++}
++
++static int
++falcon_nic_init_hardware(struct efhw_nic *nic,
++ struct efhw_ev_handler *ev_handlers,
++ const uint8_t *mac_addr)
++{
++ int rc;
++
++ /* header sanity checks */
++ FALCON_ASSERT_VALID();
++
++ rc = falcon_nic_get_mac_config(nic);
++ if (rc < 0)
++ return rc;
++
++ /* Initialise supporting modules */
++ falcon_nic_ipfilter_ctor(nic);
++
++#if FALCON_USE_SHADOW_BUFFER_TABLE
++ CI_ZERO_ARRAY(_falcon_buffer_table, FALCON_BUFFER_TBL_NUM);
++#endif
++
++ /* Initialise the top level hardware blocks */
++ memcpy(nic->mac_addr, mac_addr, ETH_ALEN);
++
++ EFHW_TRACE("%s:", __FUNCTION__);
++
++ /* nic.c:efhw_nic_init marks all the interrupt units as unused.
++
++ ?? TODO we should be able to request the non-interrupting event
++ queue and the net driver's (for a net driver that is using libefhw)
++ additional RSS queues here.
++
++ Result would be that that net driver could call
++ nic.c:efhw_nic_allocate_common_hardware_resources() and that the
++ IFDEF FALCON's can be removed from
++ nic.c:efhw_nic_allocate_common_hardware_resources()
++ */
++ nic->irq_unit[0] = INT_EN_REG_CHAR_OFST;
++
++ /*****************************************************************
++ * The rest of this function deals with initialization of the NICs
++ * hardware (as opposed to the initialization of the
++ * struct efhw_nic data structure */
++
++ /* char driver grabs SRM events onto the non interrupting
++ * event queue */
++ falcon_nic_srm_upd_evq(nic, FALCON_EVQ_NONIRQ);
++
++ /* RXDP tweaks */
++
++ /* ?? bug2396 rx_cfg should be ok so long as the net driver
++ * always pushes buffers big enough for the link MTU */
++
++ /* set the RX buffer cutoff size to be the same as PAGE_SIZE.
++ * Use this value when we think that there will be a lot of
++ * jumbo frames.
++ *
++ * The default value 1600 is useful when packets are small,
++ * but would means that jumbo frame RX queues would need more
++ * descriptors pushing */
++ falcon_nic_set_rx_usr_buf_size(nic, FALCON_RX_USR_BUF_SIZE);
++
++ /* TXDP tweaks */
++ /* ?? bug2396 looks ok */
++ falcon_nic_tx_cfg(nic, /*unlocked(for non-UDP/TCP)= */ 0);
++ falcon_nic_pace_cfg(nic);
++
++ /* ?? bug2396
++ * netdriver must load first or else must RMW this register */
++ falcon_nic_rx_filter_ctl_set(nic, RX_FILTER_CTL_SRCH_LIMIT_TCP_FULL,
++ RX_FILTER_CTL_SRCH_LIMIT_TCP_WILD,
++ RX_FILTER_CTL_SRCH_LIMIT_UDP_FULL,
++ RX_FILTER_CTL_SRCH_LIMIT_UDP_WILD);
++
++ if (!(nic->flags & NIC_FLAG_NO_INTERRUPT)) {
++ rc = efhw_keventq_ctor(nic, FALCON_EVQ_CHAR, &nic->evq[0],
++ ev_handlers);
++ if (rc < 0) {
++ EFHW_ERR("%s: efhw_keventq_ctor() failed (%d) evq=%d",
++ __FUNCTION__, rc, FALCON_EVQ_CHAR);
++ return rc;
++ }
++ }
++ rc = efhw_keventq_ctor(nic, FALCON_EVQ_NONIRQ,
++ &nic->evq[FALCON_EVQ_NONIRQ], NULL);
++ if (rc < 0) {
++ EFHW_ERR("%s: efhw_keventq_ctor() failed (%d) evq=%d",
++ __FUNCTION__, rc, FALCON_EVQ_NONIRQ);
++ return rc;
++ }
++
++ /* allocate IRQ channel */
++ rc = falcon_nic_init_irq_channel(nic, 1);
++ /* ignore failure at user-level for eftest */
++ if ((rc < 0) && !(nic->options & NIC_OPT_EFTEST))
++ return rc;
++
++ return 0;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Interrupt
++ *
++ *--------------------------------------------------------------------*/
++
++static void
++falcon_nic_interrupt_enable(struct efhw_nic *nic, unsigned idx)
++{
++ int evq;
++
++ if (idx || (nic->flags & NIC_FLAG_NO_INTERRUPT))
++ return;
++
++ /* Enable driver interrupts */
++ EFHW_NOTICE("%s: enable master interrupt", __FUNCTION__);
++ falcon_nic_interrupt_hw_enable(nic);
++
++ /* An interrupting eventq must start of day ack its read pointer */
++ evq = falcon_idx_to_evq(nic, idx);
++
++ if (falcon_evq_is_interrupting(nic, evq)) {
++ struct efhw_keventq *q = &nic->evq[idx];
++ unsigned rdptr =
++ EFHW_EVENT_OFFSET(q, q, 1) / sizeof(efhw_event_t);
++ falcon_nic_evq_ack(nic, evq, rdptr, false);
++ EFHW_NOTICE("%s: ACK evq[%d]:%x", __FUNCTION__, evq, rdptr);
++ }
++}
++
++static void falcon_nic_interrupt_disable(struct efhw_nic *nic, uint idx)
++{
++ /* NB. No need to check for NIC_FLAG_NO_INTERRUPT, as
++ ** falcon_nic_interrupt_hw_disable() will do it. */
++ if (idx)
++ return;
++ falcon_nic_interrupt_hw_disable(nic);
++}
++
++static void
++falcon_nic_set_interrupt_moderation(struct efhw_nic *nic, uint idx,
++ uint32_t val)
++{
++ falcon_timer_cmd(nic, falcon_idx_to_evq(nic, idx),
++ TIMER_MODE_INT_HLDOFF, val / 5);
++}
++
++static inline void legacy_irq_ack(struct efhw_nic *nic)
++{
++ EFHW_ASSERT(!(nic->flags & NIC_FLAG_NO_INTERRUPT));
++
++ if (!(nic->flags & NIC_FLAG_MSI)) {
++ writel(1, EFHW_KVA(nic) + INT_ACK_REG_CHAR_A1_OFST);
++ mmiowb();
++ /* ?? FIXME: We should be doing a read here to ensure IRQ is
++ * thoroughly acked before we return from ISR. */
++ }
++}
++
++static int falcon_nic_interrupt(struct efhw_nic *nic)
++{
++ volatile uint32_t *syserr_ptr =
++ (uint32_t *) efhw_iopage_ptr(&nic->irq_iobuff);
++ int handled = 0;
++ int done_ack = 0;
++
++ EFHW_ASSERT(!(nic->flags & NIC_FLAG_NO_INTERRUPT));
++ EFHW_ASSERT(syserr_ptr);
++
++ /* FIFO fill level interrupt - just log it. */
++ if (unlikely(*(syserr_ptr + (DW0_OFST / 4)))) {
++ EFHW_WARN("%s: *** FIFO *** %x", __FUNCTION__,
++ *(syserr_ptr + (DW0_OFST / 4)));
++ *(syserr_ptr + (DW0_OFST / 4)) = 0;
++ handled++;
++ }
++
++ /* Fatal interrupts. */
++ if (unlikely(*(syserr_ptr + (DW2_OFST / 4)))) {
++ *(syserr_ptr + (DW2_OFST / 4)) = 0;
++ falcon_nic_handle_fatal_int(nic);
++ handled++;
++ }
++
++ /* Event queue interrupt. For legacy interrupts we have to check
++ * that the interrupt is for us, because it could be shared. */
++ if (*(syserr_ptr + (DW1_OFST / 4))) {
++ *(syserr_ptr + (DW1_OFST / 4)) = 0;
++ /* ACK must come before callback to handler fn. */
++ legacy_irq_ack(nic);
++ done_ack = 1;
++ handled++;
++ if (nic->irq_handler)
++ nic->irq_handler(nic, 0);
++ }
++
++ if (unlikely(!done_ack)) {
++ if (!handled)
++ /* Shared interrupt line (hopefully). */
++ return 0;
++ legacy_irq_ack(nic);
++ }
++
++ EFHW_TRACE("%s: handled %d", __FUNCTION__, handled);
++ return 1;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Event Management - and SW event posting
++ *
++ *--------------------------------------------------------------------*/
++
++static void
++falcon_nic_event_queue_enable(struct efhw_nic *nic, uint evq, uint evq_size,
++ dma_addr_t q_base_addr, /* not used */
++ uint buf_base_id)
++{
++ EFHW_ASSERT(nic);
++
++ /*!\ TODO we can be more efficient if we know whether or not there
++ * is a timer attached */
++ falcon_timer_cmd(nic, evq, 0 /* disable */ , 0);
++
++ falcon_nic_evq_ptr_tbl(nic, evq, 1, buf_base_id, evq_size);
++ EFHW_TRACE("%s: enable evq %u size %u", __FUNCTION__, evq, evq_size);
++}
++
++static void
++falcon_nic_event_queue_disable(struct efhw_nic *nic, uint evq, int timer_only)
++{
++ EFHW_ASSERT(nic);
++
++ /*!\ TODO we can be more efficient if we know whether or not there
++ * is a timer attached */
++ falcon_timer_cmd(nic, evq, 0 /* disable */ , 0);
++
++ if (!timer_only)
++ falcon_nic_evq_ptr_tbl(nic, evq, 0, 0, 0);
++ EFHW_TRACE("%s: disenable evq %u", __FUNCTION__, evq);
++}
++
++static void
++falcon_nic_wakeup_request(struct efhw_nic *nic, dma_addr_t q_base_addr,
++ int next_i, int evq)
++{
++ EFHW_ASSERT(evq > FALCON_EVQ_CHAR);
++ falcon_nic_evq_ack(nic, evq, next_i, true);
++ EFHW_TRACE("%s: evq %d next_i %d", __FUNCTION__, evq, next_i);
++}
++
++static void falcon_nic_sw_event(struct efhw_nic *nic, int data, int evq)
++{
++ uint64_t ev_data = data;
++
++ ev_data &= ~FALCON_EVENT_CODE_MASK;
++ ev_data |= FALCON_EVENT_CODE_SW;
++
++ falcon_drv_ev(nic, ev_data, evq);
++ EFHW_NOTICE("%s: evq[%d]->%x", __FUNCTION__, evq, data);
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Filter support - TODO vary the depth of the search
++ *
++ *--------------------------------------------------------------------*/
++
++void
++falcon_nic_ipfilter_ctor(struct efhw_nic *nic)
++{
++ if (nic->devtype.variant == 'B' && nic->fpga_version)
++ nic->filter_tbl_size = 8 * 1024;
++ else
++ nic->filter_tbl_size = 16 * 1024;
++
++ host_ipfilter_cache_init(nic);
++}
++
++
++static int
++falcon_nic_ipfilter_set(struct efhw_nic *nic, int type, int *_filter_idx,
++ int dmaq,
++ unsigned saddr_be32, unsigned sport_be16,
++ unsigned daddr_be32, unsigned dport_be16)
++{
++ FALCON_LOCK_DECL;
++ int tcp;
++ int full;
++ int rss_b0;
++ int scat_b0;
++ int key, hash1, hash2, idx = -1;
++ int k;
++ int rc = 0;
++ unsigned max_srch = -1;
++
++ /* oh joy of joys .. maybe one day we'll optimise */
++ unsigned int saddr = ntohl(saddr_be32);
++ unsigned int daddr = ntohl(daddr_be32);
++ unsigned int sport = ntohs(sport_be16);
++ unsigned int dport = ntohs(dport_be16);
++
++ __RANGECHCK(sport, SRC_TCP_DEST_UDP_1_WIDTH);
++ __RANGECHCK(dport, SRC_TCP_DEST_UDP_1_WIDTH);
++
++ tcp = ((type & EFHW_IP_FILTER_TYPE_TCP_MASK) != 0) ? 1 : 0;
++ full = ((type & EFHW_IP_FILTER_TYPE_FULL_MASK) != 0) ? 1 : 0;
++ rss_b0 = ((type & EFHW_IP_FILTER_TYPE_RSS_B0_MASK) != 0) ? 1 : 0;
++ scat_b0 = ((type & EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK) != 0) ? 0 : 1;
++ if (tcp && full)
++ max_srch = tcp_full_srch_limit;
++ else if (tcp && !full)
++ max_srch = tcp_wild_srch_limit;
++ else if (!tcp && full)
++ max_srch = udp_full_srch_limit;
++ else if (!tcp && !full)
++ max_srch = udp_wild_srch_limit;
++
++ EFHW_TRACE("%s: %x tcp %d full %d max_srch=%d",
++ __FUNCTION__, type, tcp, full, max_srch);
++
++ /* The second hash function is simply
++ * h2(key) = 13 LSB of (key * 2 - 1)
++ * And the index(k), or the filter table address for kth search is
++ * index(k) = 13 LSB of (h1(key) + k * h2(key))
++ */
++ key = falcon_hash_get_key(saddr, sport, daddr, dport, tcp, full);
++ hash1 = falcon_hash_function1(key, nic->filter_tbl_size);
++ hash2 = falcon_hash_function2(key, nic->filter_tbl_size);
++
++ /* Avoid race to claim a filter entry */
++ FALCON_LOCK_LOCK(nic);
++
++ for (k = 0; (unsigned)k < max_srch; k++) {
++ idx = falcon_hash_iterator(hash1, hash2, k,
++ nic->filter_tbl_size);
++
++ EFHW_TRACE("ipfilter_set[%d:%d:%d]: src=%x:%d dest=%x:%d %s",
++ *_filter_idx, idx, k,
++ saddr, sport, daddr, dport,
++ host_ipfilter_cache_active(nic, idx) ?
++ "Active" : "Clear");
++
++ if (!host_ipfilter_cache_active(nic, idx))
++ break;
++
++#if FALCON_FULL_FILTER_CACHE
++ /* Check that we are not duplicating the filter */
++ if (!host_ipfilter_cache_check_not(nic->index, idx, tcp, full,
++ saddr, sport, daddr,
++ dport)) {
++ EFHW_WARN("%s: ERROR: duplicate filter (disabling "
++ "interrupts)", __FUNCTION__);
++ FALCON_LOCK_UNLOCK(nic);
++ falcon_nic_interrupt_hw_disable(nic);
++ return -EINVAL;
++ }
++#endif
++
++ }
++ if (k == max_srch) {
++ rc = -EADDRINUSE;
++ idx = -1;
++ goto fail1;
++ }
++
++ EFHW_ASSERT(idx < (int)nic->filter_tbl_size);
++
++ host_ipfilter_cache_set_addr(nic, idx, dmaq, tcp, full, rss_b0,
++ scat_b0, saddr, sport, daddr, dport);
++
++ _falcon_nic_ipfilter_set(nic, tcp, full, rss_b0,
++ scat_b0, idx, dmaq,
++ saddr, sport, daddr, dport);
++
++ *_filter_idx = idx;
++
++ EFHW_TRACE("%s: filter %x rxq %d src " NIPQUAD_FMT
++ ":%d dest " NIPQUAD_FMT ":%d set in %d",
++ __FUNCTION__, idx, dmaq,
++ NIPQUAD(&saddr), sport, NIPQUAD(&daddr), dport, k);
++
++fail1:
++ FALCON_LOCK_UNLOCK(nic);
++ return rc;
++}
++
++static void
++falcon_nic_ipfilter_attach(struct efhw_nic *nic, int filter_idx, int dmaq_idx)
++{
++ /* Intentionally empty - Falcon attaches and sets the filter
++ * in filter_set */
++ EFHW_TRACE("%s: attach filter %x with rxq %d - ignored",
++ __FUNCTION__, filter_idx, dmaq_idx);
++}
++
++static void falcon_nic_ipfilter_detach(struct efhw_nic *nic, int filter_idx)
++{
++ /* Intentionally empty - Falcon attaches and sets the filter
++ * in filter_clear */
++ EFHW_TRACE("%s: detach filter %x from rxq - ignored",
++ __FUNCTION__, filter_idx);
++}
++
++static void falcon_nic_ipfilter_clear(struct efhw_nic *nic, int filter_idx)
++{
++ FALCON_LOCK_DECL;
++
++ EFHW_TRACE("%s: filter %x", __FUNCTION__, filter_idx);
++
++ /* In case the filter has already been freed */
++ if (filter_idx == -1)
++ return;
++
++ FALCON_LOCK_LOCK(nic);
++
++ /* if we flush a chained hash then all we need to do is zero it out */
++ host_ipfilter_cache_flush(nic, filter_idx);
++ _falcon_nic_ipfilter_clear(nic, filter_idx);
++
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Buffer table - helpers
++ *
++ *--------------------------------------------------------------------*/
++
++#define FALCON_LAZY_COMMIT_HWM (FALCON_BUFFER_UPD_MAX - 16)
++
++/* Note re.:
++ * falcon_nic_buffer_table_lazy_commit(struct efhw_nic *nic)
++ * falcon_nic_buffer_table_update_poll(struct efhw_nic *nic)
++ * falcon_nic_buffer_table_confirm(struct efhw_nic *nic)
++ * -- these are no-ops in the user-level driver because it would need to
++ * coordinate with the real driver on the number of outstanding commits.
++ *
++ * An exception is made for eftest apps, which manage the hardware without
++ * using the char driver.
++ */
++
++static inline void falcon_nic_buffer_table_lazy_commit(struct efhw_nic *nic)
++{
++#if defined(__ci_ul_driver__)
++ if (!(nic->options & NIC_OPT_EFTEST))
++ return;
++#endif
++
++ /* Do nothing if operating in synchronous mode. */
++ if (!nic->irq_handler)
++ return;
++}
++
++static inline void falcon_nic_buffer_table_update_poll(struct efhw_nic *nic)
++{
++ FALCON_LOCK_DECL;
++ int count = 0, rc = 0;
++
++#if defined(__ci_ul_driver__)
++ if (!(nic->options & NIC_OPT_EFTEST))
++ return;
++#endif
++
++ /* We can be called here early days */
++ if (!nic->irq_handler)
++ return;
++
++ /* If we need to gather buffer update events then poll the
++ non-interrupting event queue */
++
++ /* For each _buffer_table_commit there will be an update done
++ event. We don't keep track of how many buffers each commit has
++ committed, just make sure that all the expected events have been
++ gathered */
++ FALCON_LOCK_LOCK(nic);
++
++ EFHW_TRACE("%s: %d", __FUNCTION__, nic->buf_commit_outstanding);
++
++ while (nic->buf_commit_outstanding > 0) {
++ /* we're not expecting to handle any events that require
++ * upcalls into the core driver */
++ struct efhw_ev_handler handler;
++ memset(&handler, 0, sizeof(handler));
++ nic->evq[FALCON_EVQ_NONIRQ].ev_handlers = &handler;
++ rc = efhw_keventq_poll(nic, &nic->evq[FALCON_EVQ_NONIRQ]);
++ nic->evq[FALCON_EVQ_NONIRQ].ev_handlers = NULL;
++
++ if (rc < 0) {
++ EFHW_ERR("%s: poll ERROR (%d:%d) ***** ",
++ __FUNCTION__, rc,
++ nic->buf_commit_outstanding);
++ goto out;
++ }
++
++ FALCON_LOCK_UNLOCK(nic);
++
++ if (count++)
++ udelay(1);
++
++ if (count > 1000) {
++ EFHW_WARN("%s: poll Timeout ***** (%d)", __FUNCTION__,
++ nic->buf_commit_outstanding);
++ nic->buf_commit_outstanding = 0;
++ return;
++ }
++ FALCON_LOCK_LOCK(nic);
++ }
++
++out:
++ FALCON_LOCK_UNLOCK(nic);
++ return;
++}
++
++void falcon_nic_buffer_table_confirm(struct efhw_nic *nic)
++{
++ /* confirm buffer table updates - should be used for items where
++ loss of data would be unacceptable. E.g for the buffers that back
++ an event or DMA queue */
++ FALCON_LOCK_DECL;
++
++#if defined(__ci_ul_driver__)
++ if (!(nic->options & NIC_OPT_EFTEST))
++ return;
++#endif
++
++ /* Do nothing if operating in synchronous mode. */
++ if (!nic->irq_handler)
++ return;
++
++ FALCON_LOCK_LOCK(nic);
++
++ _falcon_nic_buffer_table_commit(nic);
++
++ FALCON_LOCK_UNLOCK(nic);
++
++ falcon_nic_buffer_table_update_poll(nic);
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Buffer table - API
++ *
++ *--------------------------------------------------------------------*/
++
++static void
++falcon_nic_buffer_table_clear(struct efhw_nic *nic, int buffer_id, int num)
++{
++ FALCON_LOCK_DECL;
++ FALCON_LOCK_LOCK(nic);
++ _falcon_nic_buffer_table_clear(nic, buffer_id, num);
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++static void
++falcon_nic_buffer_table_set(struct efhw_nic *nic, dma_addr_t dma_addr,
++ uint bufsz, uint region,
++ int own_id, int buffer_id)
++{
++ FALCON_LOCK_DECL;
++
++ EFHW_ASSERT(region < FALCON_REGION_NUM);
++
++ EFHW_ASSERT((bufsz == EFHW_4K) ||
++ (bufsz == EFHW_8K && FALCON_BUFFER_TABLE_FULL_MODE));
++
++ falcon_nic_buffer_table_update_poll(nic);
++
++ FALCON_LOCK_LOCK(nic);
++
++ _falcon_nic_buffer_table_set(nic, dma_addr, bufsz, region, own_id,
++ buffer_id);
++
++ falcon_nic_buffer_table_lazy_commit(nic);
++
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++void
++falcon_nic_buffer_table_set_n(struct efhw_nic *nic, int buffer_id,
++ dma_addr_t dma_addr, uint bufsz, uint region,
++ int n_pages, int own_id)
++{
++ /* used to set up a contiguous range of buffers */
++ FALCON_LOCK_DECL;
++
++ EFHW_ASSERT(region < FALCON_REGION_NUM);
++
++ EFHW_ASSERT((bufsz == EFHW_4K) ||
++ (bufsz == EFHW_8K && FALCON_BUFFER_TABLE_FULL_MODE));
++
++ while (n_pages--) {
++
++ falcon_nic_buffer_table_update_poll(nic);
++
++ FALCON_LOCK_LOCK(nic);
++
++ _falcon_nic_buffer_table_set(nic, dma_addr, bufsz, region,
++ own_id, buffer_id++);
++
++ falcon_nic_buffer_table_lazy_commit(nic);
++
++ FALCON_LOCK_UNLOCK(nic);
++
++ dma_addr += bufsz;
++ }
++}
++
++/*--------------------------------------------------------------------
++ *
++ * DMA Queues - mid level API
++ *
++ *--------------------------------------------------------------------*/
++
++#if BUG5302_WORKAROUND
++
++/* Tx queues can get stuck if the software write pointer is set to an index
++ * beyond the configured size of the queue, such that they will not flush.
++ * This code can be run before attempting a flush; it will detect the bogus
++ * value and reset it. This fixes most instances of this problem, although
++ * sometimes it does not work, or we may not detect it in the first place,
++ * if the out-of-range value was replaced by an in-range value earlier.
++ * (In those cases we have to apply a bigger hammer later, if we see that
++ * the queue is still not flushing.)
++ */
++static void
++falcon_check_for_bogus_tx_dma_wptr(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val_low64, val_high64;
++ uint64_t size, hwptr, swptr, val;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ ulong offset = falcon_dma_tx_q_offset(nic, dmaq);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + offset, &val_low64, &val_high64);
++ FALCON_LOCK_UNLOCK(nic);
++
++ size = (val_low64 >> TX_DESCQ_SIZE_LBN)
++ & __FALCON_MASK64(TX_DESCQ_SIZE_WIDTH);
++ size = (1 << size) * 512;
++ hwptr = (val_high64 >> __DW3(TX_DESCQ_HW_RPTR_LBN))
++ & __FALCON_MASK64(TX_DESCQ_HW_RPTR_WIDTH);
++ swptr = (val_low64 >> TX_DESCQ_SW_WPTR_LBN)
++ & __FALCON_MASK64(__LW2(TX_DESCQ_SW_WPTR_LBN));
++ val = (val_high64)
++ &
++ __FALCON_MASK64(__DW3
++ (TX_DESCQ_SW_WPTR_LBN + TX_DESCQ_SW_WPTR_WIDTH));
++ val = val << __LW2(TX_DESCQ_SW_WPTR_LBN);
++ swptr = swptr | val;
++
++ if (swptr >= size) {
++ EFHW_WARN("Resetting bad write pointer for TXQ[%d]", dmaq);
++ writel((uint32_t) ((hwptr + 0) & (size - 1)),
++ efhw_kva + falcon_tx_dma_page_addr(dmaq) + 12);
++ mmiowb();
++ }
++}
++
++/* Here's that "bigger hammer": we reset all the pointers (hardware read,
++ * hardware descriptor cache read, software write) to zero.
++ */
++void falcon_clobber_tx_dma_ptrs(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val_low64, val_high64;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ ulong offset = falcon_dma_tx_q_offset(nic, dmaq);
++
++ EFHW_WARN("Recovering stuck TXQ[%d]", dmaq);
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + offset, &val_low64, &val_high64);
++ val_high64 &= ~(__FALCON_MASK64(TX_DESCQ_HW_RPTR_WIDTH)
++ << __DW3(TX_DESCQ_HW_RPTR_LBN));
++ val_high64 &= ~(__FALCON_MASK64(TX_DC_HW_RPTR_WIDTH)
++ << __DW3(TX_DC_HW_RPTR_LBN));
++ falcon_write_qq(efhw_kva + offset, val_low64, val_high64);
++ mmiowb();
++ writel(0, efhw_kva + falcon_tx_dma_page_addr(dmaq) + 12);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++}
++
++#endif
++
++static inline int
++__falcon_really_flush_tx_dma_channel(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint val;
++
++ EFHW_BUILD_ASSERT(TX_FLUSH_DESCQ_REG_KER_OFST ==
++ TX_FLUSH_DESCQ_REG_OFST);
++
++ __DWCHCK(TX_FLUSH_DESCQ_CMD_LBN, TX_FLUSH_DESCQ_CMD_WIDTH);
++ __DWCHCK(TX_FLUSH_DESCQ_LBN, TX_FLUSH_DESCQ_WIDTH);
++ __RANGECHCK(dmaq, TX_FLUSH_DESCQ_WIDTH);
++
++ val = ((1 << TX_FLUSH_DESCQ_CMD_LBN) | (dmaq << TX_FLUSH_DESCQ_LBN));
++
++ EFHW_TRACE("TX DMA flush[%d]", dmaq);
++
++#if BUG5302_WORKAROUND
++ falcon_check_for_bogus_tx_dma_wptr(nic, dmaq);
++#endif
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + TX_FLUSH_DESCQ_REG_OFST,
++ val, FALCON_ATOMIC_TX_FLUSH_DESCQ);
++
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return 0;
++}
++
++static inline int
++__falcon_is_tx_dma_channel_flushed(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val_low64, val_high64;
++ uint64_t enable, flush_pending;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ ulong offset = falcon_dma_tx_q_offset(nic, dmaq);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_qq(efhw_kva + offset, &val_low64, &val_high64);
++ FALCON_LOCK_UNLOCK(nic);
++
++ /* should see one of three values for these 2 bits
++ * 1, queue enabled no flush pending
++ * - i.e. first flush request
++ * 2, queue enabled, flush pending
++ * - i.e. request to reflush before flush finished
++ * 3, queue disabled (no flush pending)
++ * - flush complete
++ */
++ __DWCHCK(TX_DESCQ_FLUSH_LBN, TX_DESCQ_FLUSH_WIDTH);
++ __DW3CHCK(TX_DESCQ_EN_LBN, TX_DESCQ_EN_WIDTH);
++ enable = val_high64 & (1 << __DW3(TX_DESCQ_EN_LBN));
++ flush_pending = val_low64 & (1 << TX_DESCQ_FLUSH_LBN);
++
++ if (enable && !flush_pending)
++ return 0;
++
++ EFHW_TRACE("%d, %s: %s, %sflush pending", dmaq, __FUNCTION__,
++ enable ? "enabled" : "disabled",
++ flush_pending ? "" : "NO ");
++ /* still in progress */
++ if (enable && flush_pending)
++ return -EALREADY;
++
++ return -EAGAIN;
++}
++
++static int falcon_flush_tx_dma_channel(struct efhw_nic *nic, uint dmaq)
++{
++ int rc;
++ rc = __falcon_is_tx_dma_channel_flushed(nic, dmaq);
++ if (rc < 0) {
++ EFHW_WARN("%s: failed %d", __FUNCTION__, rc);
++ return rc;
++ }
++ return __falcon_really_flush_tx_dma_channel(nic, dmaq);
++}
++
++static int
++__falcon_really_flush_rx_dma_channel(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ uint val;
++
++ EFHW_BUILD_ASSERT(RX_FLUSH_DESCQ_REG_KER_OFST ==
++ RX_FLUSH_DESCQ_REG_OFST);
++
++ __DWCHCK(RX_FLUSH_DESCQ_CMD_LBN, RX_FLUSH_DESCQ_CMD_WIDTH);
++ __DWCHCK(RX_FLUSH_DESCQ_LBN, RX_FLUSH_DESCQ_WIDTH);
++ __RANGECHCK(dmaq, RX_FLUSH_DESCQ_WIDTH);
++
++ val = ((1 << RX_FLUSH_DESCQ_CMD_LBN) | (dmaq << RX_FLUSH_DESCQ_LBN));
++
++ EFHW_TRACE("RX DMA flush[%d]", dmaq);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_write_qq(efhw_kva + RX_FLUSH_DESCQ_REG_OFST, val,
++ FALCON_ATOMIC_RX_FLUSH_DESCQ);
++ mmiowb();
++ FALCON_LOCK_UNLOCK(nic);
++ return 0;
++}
++
++static inline int
++__falcon_is_rx_dma_channel_flushed(struct efhw_nic *nic, uint dmaq)
++{
++ FALCON_LOCK_DECL;
++ uint64_t val;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ ulong offset = falcon_dma_rx_q_offset(nic, dmaq);
++
++ /* Falcon requires 128 bit atomic access for this register */
++ FALCON_LOCK_LOCK(nic);
++ falcon_read_q(efhw_kva + offset, &val);
++ FALCON_LOCK_UNLOCK(nic);
++
++ __DWCHCK(RX_DESCQ_EN_LBN, RX_DESCQ_EN_WIDTH);
++
++ /* is it enabled? */
++ return (val & (1 << RX_DESCQ_EN_LBN))
++ ? 0 : -EAGAIN;
++}
++
++static int falcon_flush_rx_dma_channel(struct efhw_nic *nic, uint dmaq)
++{
++ int rc;
++ rc = __falcon_is_rx_dma_channel_flushed(nic, dmaq);
++ if (rc < 0) {
++ EFHW_ERR("%s: failed %d", __FUNCTION__, rc);
++ return rc;
++ }
++ return __falcon_really_flush_rx_dma_channel(nic, dmaq);
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Falcon specific event callbacks
++ *
++ *--------------------------------------------------------------------*/
++
++int
++falcon_handle_char_event(struct efhw_nic *nic, struct efhw_ev_handler *h,
++ efhw_event_t *ev)
++{
++ EFHW_TRACE("DRIVER EVENT: "FALCON_EVENT_FMT,
++ FALCON_EVENT_PRI_ARG(*ev));
++
++ switch (FALCON_EVENT_DRIVER_SUBCODE(ev)) {
++
++ case TX_DESCQ_FLS_DONE_EV_DECODE:
++ EFHW_TRACE("TX[%d] flushed",
++ (int)FALCON_EVENT_TX_FLUSH_Q_ID(ev));
++#if !defined(__ci_ul_driver__)
++ efhw_handle_txdmaq_flushed(nic, h, ev);
++#endif
++ break;
++
++ case RX_DESCQ_FLS_DONE_EV_DECODE:
++ EFHW_TRACE("RX[%d] flushed",
++ (int)FALCON_EVENT_TX_FLUSH_Q_ID(ev));
++#if !defined(__ci_ul_driver__)
++ efhw_handle_rxdmaq_flushed(nic, h, ev);
++#endif
++ break;
++
++ case SRM_UPD_DONE_EV_DECODE:
++ nic->buf_commit_outstanding =
++ max(0, nic->buf_commit_outstanding - 1);
++ EFHW_TRACE("COMMIT DONE %d", nic->buf_commit_outstanding);
++ break;
++
++ case EVQ_INIT_DONE_EV_DECODE:
++ EFHW_TRACE("EVQ INIT");
++ break;
++
++ case WAKE_UP_EV_DECODE:
++ EFHW_TRACE("WAKE UP");
++ efhw_handle_wakeup_event(nic, h, ev);
++ break;
++
++ case TIMER_EV_DECODE:
++ EFHW_TRACE("TIMER");
++ efhw_handle_timeout_event(nic, h, ev);
++ break;
++
++ case RX_DESCQ_FLSFF_OVFL_EV_DECODE:
++ /* This shouldn't happen. */
++ EFHW_ERR("%s: RX flush fifo overflowed", __FUNCTION__);
++ return -EINVAL;
++
++ default:
++ EFHW_TRACE("UNKOWN DRIVER EVENT: " FALCON_EVENT_FMT,
++ FALCON_EVENT_PRI_ARG(*ev));
++ break;
++ }
++ return 0;
++}
++
++/*--------------------------------------------------------------------
++ *
++ * Abstraction Layer Hooks
++ *
++ *--------------------------------------------------------------------*/
++
++struct efhw_func_ops falcon_char_functional_units = {
++ falcon_nic_close_hardware,
++ falcon_nic_init_hardware,
++ falcon_nic_interrupt,
++ falcon_nic_interrupt_enable,
++ falcon_nic_interrupt_disable,
++ falcon_nic_set_interrupt_moderation,
++ falcon_nic_event_queue_enable,
++ falcon_nic_event_queue_disable,
++ falcon_nic_wakeup_request,
++ falcon_nic_sw_event,
++ falcon_nic_ipfilter_set,
++ falcon_nic_ipfilter_attach,
++ falcon_nic_ipfilter_detach,
++ falcon_nic_ipfilter_clear,
++ falcon_dmaq_tx_q_init,
++ falcon_dmaq_rx_q_init,
++ falcon_dmaq_tx_q_disable,
++ falcon_dmaq_rx_q_disable,
++ falcon_flush_tx_dma_channel,
++ falcon_flush_rx_dma_channel,
++ falcon_nic_buffer_table_set,
++ falcon_nic_buffer_table_set_n,
++ falcon_nic_buffer_table_clear,
++ falcon_nic_buffer_table_commit,
++};
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/falcon_hash.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/falcon_hash.c 2008-05-19 00:33:29.425842266 +0300
+@@ -0,0 +1,178 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains EtherFabric NIC hash algorithms implementation.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efhw/debug.h>
++#include <ci/driver/efab/hardware.h>
++
++
++/* this mask is per filter bank hence /2 */
++#define FILTER_MASK(n) ((n) / 2u - 1u)
++
++/*
++ * Main Functions related to the Hash Table Generation
++ * Author: Srinivasaih, Nataraj
++ * Created: Thu May 13:32:41 PDT 2004
++ * $Id$
++ */
++/***************************************************************************
++Class Maximum number of Valid address ranges
++ hosts per network
++A 16777214 1.0.0.1 through 9.255.255.254
++ 11.0.0.1 through 126.255.255.254
++B 65534 128.0.0.1 through 172.15.255.254
++ 172.32.0.1 through 191.255.255.254
++C 254 192.0.0.1 through 192.167.255.254
++ 192.169.0.1 through 223.255.255.254
++P 16777214 10.0.0.1 through 10.255.255.254 (10/8)
++ 1048574 172.16.0.1 through 172.31.255.254 (172.16/12)
++ 65534 192.168.0.1 through 192.168.255.254 (192.168/16)
++
++R - 0.0.0.0 through 0.255.255.255
++ (used if host will be assigned a
++ valid address dynamically)
++ 127.0.0.0 through 127.255.255.255
++ (loopback addresses)
++
++P : Private internets only
++R : Reserved
++****************************************************************************/
++
++/* All LE parameters */
++unsigned int
++falcon_hash_get_key(unsigned int src_ip, unsigned int src_port,
++ unsigned int dest_ip, unsigned int dest_port,
++ int tcp, int full)
++{
++
++ unsigned int result = 0;
++ int net_type;
++
++ EFHW_ASSERT(tcp == 0 || tcp == 1);
++ EFHW_ASSERT(full == 0 || full == 1);
++
++ net_type = tcp << 4 | full;
++
++ /* Note that src_ip and src_port ignored if a wildcard filter */
++ switch (net_type) {
++ case 0x01: /* UDP Full */
++ result = ((dest_ip & 0xfffffffe) | (!(dest_ip & 1))) ^
++ (((dest_port << 16) & 0xFFFF0000) |
++ ((src_ip >> 16) & 0x0000FFFF)) ^
++ (((src_ip << 16) & 0xFFFF0000) |
++ ((src_port & 0x0000FFFF)));
++ EFHW_TRACE("falcon_hash_get_key: UDP Full %x", result);
++ break;
++ case 0x00: /* UDP Wild Card */
++ result = ((dest_ip & 0xfffffffe) | (!(dest_ip & 1))) ^
++ (((dest_port << 16) & 0x00000000) |
++ ((src_ip >> 16) & 0x00000000)) ^
++ (((src_ip << 16) & 0x00000000) |
++ ((dest_port & 0x0000FFFF)));
++ EFHW_TRACE("falcon_hash_get_key: UDP Wildcard %x", result);
++ break;
++ case 0x10: /* TCP Wild Card */
++ result = (dest_ip) ^
++ (((dest_port << 16) & 0xFFFF0000) |
++ ((src_ip >> 16) & 0x00000000)) ^
++ (((src_ip << 16) & 0x00000000) |
++ ((src_port & 0x00000000)));
++ EFHW_TRACE("falcon_hash_get_key: TCP Wildcard %x", result);
++ break;
++ case 0x11: /* TCP Full */
++ result = (dest_ip) ^
++ (((dest_port << 16) & 0xFFFF0000) |
++ ((src_ip >> 16) & 0x0000FFFF)) ^
++ (((src_ip << 16) & 0xFFFF0000) |
++ ((src_port & 0x0000FFFF)));
++ EFHW_TRACE("falcon_hash_get_key: TCP Full %x", result);
++ break;
++ default:
++ EFHW_ASSERT(0);
++
++ }
++ return (result);
++}
++
++/* This function generates the First Hash key */
++unsigned int falcon_hash_function1(unsigned int key, unsigned int nfilters)
++{
++
++ unsigned short int lfsr_reg;
++ unsigned int tmp_key;
++ int index;
++
++ unsigned short int lfsr_input;
++ unsigned short int single_bit_key;
++ unsigned short int bit16_lfsr;
++ unsigned short int bit3_lfsr;
++
++ lfsr_reg = 0xFFFF;
++ tmp_key = key;
++
++ /* For Polynomial equation X^16+X^3+1 */
++ for (index = 0; index < 32; index++) {
++ /* Get the bit from key and shift the key */
++ single_bit_key = (tmp_key & 0x80000000) >> 31;
++ tmp_key = tmp_key << 1;
++
++ /* get the Tap bits to XOR operation */
++ bit16_lfsr = (lfsr_reg & 0x8000) >> 15;
++ bit3_lfsr = (lfsr_reg & 0x0004) >> 2;
++
++ /* Get the Input value to the LFSR */
++ lfsr_input = ((bit16_lfsr ^ bit3_lfsr) ^ single_bit_key);
++
++ /* Shift and store out of the two TAPs */
++ lfsr_reg = lfsr_reg << 1;
++ lfsr_reg = lfsr_reg | (lfsr_input & 0x0001);
++
++ }
++
++ lfsr_reg = lfsr_reg & FILTER_MASK(nfilters);
++
++ return lfsr_reg;
++}
++
++/* This function generates the Second Hash */
++unsigned int
++falcon_hash_function2(unsigned int key, unsigned int nfilters)
++{
++ return (unsigned int)(((unsigned long long)key * 2 - 1) &
++ FILTER_MASK(nfilters));
++}
++
++/* This function iterates through the hash table */
++unsigned int
++falcon_hash_iterator(unsigned int hash1, unsigned int hash2,
++ unsigned int n_search, unsigned int nfilters)
++{
++ return ((hash1 + (n_search * hash2)) & FILTER_MASK(nfilters));
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/falcon_mac.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/falcon_mac.c 2008-05-19 00:33:29.425842266 +0300
+@@ -0,0 +1,171 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains MACs (Mentor MAC & GDACT1 ) support for Falcon.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efhw/falcon.h>
++#include <ci/driver/efab/hardware.h>
++
++/********************************************************************
++ * Mentor MAC
++ */
++
++#define _PRE(x) GM##x
++
++/*--------------------------------------------------------------------
++ *
++ * Debug Support
++ *
++ *--------------------------------------------------------------------*/
++
++#define MENTOR_MAC_ASSERT_VALID() \
++ EFHW_ASSERT(nic); \
++ EFHW_ASSERT(EFHW_KVA(nic)); \
++ EFHW_ASSERT(_PRE(_CFG1_REG_OFST) == _PRE(_CFG1_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_CFG2_REG_OFST) == _PRE(_CFG2_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_IPG_REG_OFST) == _PRE(_IPG_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_HD_REG_OFST) == _PRE(_HD_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_MAX_FLEN_REG_OFST) == _PRE(_MAX_FLEN_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_TEST_REG_OFST) == _PRE(_TEST_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_ADR1_REG_OFST) == _PRE(_ADR1_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(_ADR2_REG_OFST) == _PRE(_ADR2_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG0_REG_OFST) == _PRE(F_CFG0_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG1_REG_OFST) == _PRE(F_CFG1_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG2_REG_OFST) == _PRE(F_CFG2_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG3_REG_OFST) == _PRE(F_CFG3_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG4_REG_OFST) == _PRE(F_CFG4_REG_KER_OFST)); \
++ EFHW_ASSERT(_PRE(F_CFG5_REG_OFST) == _PRE(F_CFG5_REG_KER_OFST));
++
++/*! Get MAC current address - i.e not necessarily the one in the EEPROM */
++static inline void mentormac_get_mac_addr(struct efhw_nic *nic)
++{
++ efhw_ioaddr_t mac_kva;
++ uint val1, val2;
++
++ MENTOR_MAC_ASSERT_VALID();
++
++ mac_kva = GM_P0_BASE + EFHW_KVA(nic);
++
++ val1 = readl(mac_kva + _PRE(_ADR1_REG_OFST));
++ val2 = readl(mac_kva + _PRE(_ADR2_REG_OFST));
++
++#if 0
++ nic->mac_addr[0] = (val1 & 0xff000000) >> 24;
++ nic->mac_addr[1] = (val1 & 0x00ff0000) >> 16;
++ nic->mac_addr[2] = (val1 & 0x0000ff00) >> 8;
++ nic->mac_addr[3] = (val1 & 0x000000ff) >> 0;
++ nic->mac_addr[4] = (val2 & 0xff000000) >> 24;
++ nic->mac_addr[5] = (val2 & 0x00ff0000) >> 16;
++#else
++ nic->mac_addr[5] = (val1 & 0xff000000) >> 24;
++ nic->mac_addr[4] = (val1 & 0x00ff0000) >> 16;
++ nic->mac_addr[3] = (val1 & 0x0000ff00) >> 8;
++ nic->mac_addr[2] = (val1 & 0x000000ff) >> 0;
++ nic->mac_addr[1] = (val2 & 0xff000000) >> 24;
++ nic->mac_addr[0] = (val2 & 0x00ff0000) >> 16;
++#endif
++}
++
++
++/********************************************************************
++ * GDACT10 MAC
++ */
++
++/*--------------------------------------------------------------------
++ *
++ * Debug Support
++ *
++ *--------------------------------------------------------------------*/
++
++#define GDACT10_MAC_ASSERT_VALID() \
++ EFHW_ASSERT(nic); \
++ EFHW_ASSERT(EFHW_KVA(nic)); \
++ EFHW_ASSERT(XM_GLB_CFG_REG_P0_OFST == XM_GLB_CFG_REG_KER_P0_OFST); \
++ EFHW_ASSERT(XM_TX_CFG_REG_P0_OFST == XM_TX_CFG_REG_KER_P0_OFST); \
++ EFHW_ASSERT(XM_RX_CFG_REG_P0_OFST == XM_RX_CFG_REG_KER_P0_OFST); \
++ EFHW_ASSERT(MAC0_SPEED_LBN == MAC1_SPEED_LBN); \
++ EFHW_ASSERT(MAC0_SPEED_WIDTH == MAC1_SPEED_WIDTH); \
++ EFHW_ASSERT(MAC0_LINK_STATUS_LBN == MAC1_LINK_STATUS_LBN); \
++ EFHW_ASSERT(MAC0_LINK_STATUS_WIDTH == MAC1_LINK_STATUS_WIDTH); \
++ EFHW_ASSERT(MAC1_BCAD_ACPT_LBN == MAC0_BCAD_ACPT_LBN); \
++ EFHW_ASSERT(MAC1_UC_PROM_LBN == MAC0_UC_PROM_LBN); \
++ EFHW_ASSERT(MAC0_CTRL_REG_KER_OFST == MAC0_CTRL_REG_OFST); \
++ EFHW_ASSERT(MAC1_CTRL_REG_KER_OFST == MAC1_CTRL_REG_OFST); \
++ EFHW_ASSERT(XM_ADR_LO_REG_KER_P0_OFST == XM_ADR_LO_REG_P0_OFST); \
++ EFHW_ASSERT(XM_ADR_HI_REG_KER_P0_OFST == XM_ADR_HI_REG_P0_OFST); \
++ EFHW_ASSERT(XM_RX_PARAM_REG_KER_P0_OFST == XM_RX_PARAM_REG_P0_OFST);
++
++/*--------------------------------------------------------------------
++ *
++ * Information gathering
++ *
++ *--------------------------------------------------------------------*/
++
++/*! Get MAC current address - i.e not necessarily the one in the EEPROM */
++static inline void GDACT10mac_get_mac_addr(struct efhw_nic *nic)
++{
++ uint val1, val2;
++ efhw_ioaddr_t efhw_kva = EFHW_KVA(nic);
++ FALCON_LOCK_DECL;
++
++ GDACT10_MAC_ASSERT_VALID();
++
++ EFHW_ASSERT(XM_ADR_LO_LBN == 0);
++ EFHW_ASSERT(XM_ADR_LO_WIDTH == 32);
++ EFHW_ASSERT(XM_ADR_HI_LBN == 0);
++ EFHW_ASSERT(XM_ADR_HI_WIDTH == 16);
++
++ FALCON_LOCK_LOCK(nic);
++
++ val1 = readl(efhw_kva + XM_ADR_LO_REG_P0_OFST);
++ val2 = readl(efhw_kva + XM_ADR_HI_REG_P0_OFST);
++
++ FALCON_LOCK_UNLOCK(nic);
++
++ /* The HW scores no points for consistency */
++ nic->mac_addr[5] = (val2 & 0x0000ff00) >> 8;
++ nic->mac_addr[4] = (val2 & 0x000000ff) >> 0;
++ nic->mac_addr[3] = (val1 & 0xff000000) >> 24;
++ nic->mac_addr[2] = (val1 & 0x00ff0000) >> 16;
++ nic->mac_addr[1] = (val1 & 0x0000ff00) >> 8;
++ nic->mac_addr[0] = (val1 & 0x000000ff) >> 0;
++}
++
++
++/********************************************************************
++ * Call one or another function
++ */
++
++void falcon_get_mac_addr(struct efhw_nic *nic)
++{
++ if (nic->flags & NIC_FLAG_10G)
++ GDACT10mac_get_mac_addr(nic);
++ else
++ mentormac_get_mac_addr(nic);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/filter_resource.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/filter_resource.c 2008-05-19 00:33:29.425842266 +0300
+@@ -0,0 +1,317 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains filters support.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efrm/vi_resource_manager.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/filter.h>
++#include <ci/efrm/buffer_table.h>
++
++struct filter_resource_manager {
++ struct efrm_resource_manager rm;
++ struct kfifo *free_ids;
++};
++
++static struct filter_resource_manager *efrm_filter_manager;
++
++void efrm_filter_resource_free(struct filter_resource *frs)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++ int id;
++
++ EFRM_RESOURCE_ASSERT_VALID(&frs->rs, 1);
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(frs->rs.rs_handle));
++
++ /* if we have a PT endpoint */
++ if (NULL != frs->pt) {
++ /* Detach the filter */
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ efhw_nic_ipfilter_detach(nic, frs->filter_idx);
++
++ /* Release our ref to the PT resource. */
++ EFRM_TRACE("%s: releasing PT resource reference",
++ __FUNCTION__);
++ efrm_vi_resource_release(frs->pt);
++ }
++
++ /* Disable the filter. */
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ efhw_nic_ipfilter_clear(nic, frs->filter_idx);
++
++ /* Free this filter. */
++ id = EFRM_RESOURCE_INSTANCE(frs->rs.rs_handle);
++ EFRM_VERIFY_EQ(kfifo_put(efrm_filter_manager->free_ids,
++ (unsigned char *)&id, sizeof(id)),
++ sizeof(id));
++
++ EFRM_DO_DEBUG(memset(frs, 0, sizeof(*frs)));
++ kfree(frs);
++}
++EXPORT_SYMBOL(efrm_filter_resource_free);
++
++static void filter_rm_dtor(struct efrm_resource_manager *rm)
++{
++ EFRM_TRACE("filter_rm_dtor");
++
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(&efrm_filter_manager->rm);
++ EFRM_ASSERT(&efrm_filter_manager->rm == rm);
++
++ kfifo_vfree(efrm_filter_manager->free_ids);
++ EFRM_TRACE("filter_rm_dtor: done");
++}
++
++/**********************************************************************/
++/**********************************************************************/
++/**********************************************************************/
++
++int efrm_create_filter_resource_manager(struct efrm_resource_manager **rm_out)
++{
++ int rc;
++
++ EFRM_ASSERT(rm_out);
++
++ efrm_filter_manager =
++ kmalloc(sizeof(struct filter_resource_manager), GFP_KERNEL);
++ if (efrm_filter_manager == 0)
++ return -ENOMEM;
++ memset(efrm_filter_manager, 0, sizeof(*efrm_filter_manager));
++
++ rc = efrm_resource_manager_ctor(&efrm_filter_manager->rm,
++ filter_rm_dtor, "FILTER",
++ EFRM_RESOURCE_FILTER, 0);
++ if (rc < 0)
++ goto fail1;
++
++ /* Create a pool of free instances */
++ rc = efrm_kfifo_id_ctor(&efrm_filter_manager->free_ids,
++ 0, EFHW_IP_FILTER_NUM,
++ &efrm_filter_manager->rm.rm_lock);
++ if (rc != 0)
++ goto fail2;
++
++ *rm_out = &efrm_filter_manager->rm;
++ EFRM_TRACE("%s: filter resources created - %d IDs",
++ __FUNCTION__, kfifo_len(efrm_filter_manager->free_ids));
++ return 0;
++
++fail2:
++ efrm_resource_manager_dtor(&efrm_filter_manager->rm);
++fail1:
++ memset(efrm_filter_manager, 0, sizeof(*efrm_filter_manager));
++ kfree(efrm_filter_manager);
++ return rc;
++
++}
++
++/*--------------------------------------------------------------------
++ *!
++ * Called to set/change the PT endpoint of a filter
++ *
++ * Example of use is TCP helper when it finds a wildcard IP filter
++ * needs to change which application it delivers traffic to
++ *
++ * \param frs filter resource
++ * \param pt_handle handle of new PT endpoint
++ *
++ * \return standard error codes
++ *
++ *--------------------------------------------------------------------*/
++int
++efrm_filter_resource_set_ptresource(struct filter_resource *frs,
++ struct vi_resource *ptrs)
++{
++ int rc, pti, nic_i;
++ struct efhw_nic *nic;
++
++ EFRM_ASSERT(frs);
++
++ /* if filter is attached to a valid PT endpoint */
++ if (NULL != frs->pt) {
++
++ EFRM_TRACE("%s: detaching PT resource " EFRM_RESOURCE_FMT
++ " from filter ",
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(frs->rs.rs_handle));
++ /* Detach the filter */
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ efhw_nic_ipfilter_detach(nic, frs->filter_idx);
++
++ /* release reference */
++ efrm_vi_resource_release(frs->pt);
++ frs->pt = NULL;
++ }
++
++ if (ptrs != NULL) {
++ /* get PT endpoint index */
++ EFRM_RESOURCE_ASSERT_VALID(&ptrs->rs, 0);
++ EFRM_ASSERT(EFRM_RESOURCE_TYPE(ptrs->rs.rs_handle) ==
++ EFRM_RESOURCE_VI);
++ pti = EFRM_RESOURCE_INSTANCE(ptrs->rs.rs_handle);
++ if (pti == 0) {
++ EFRM_ERR("%s: cannot filter for channel 0",
++ __FUNCTION__);
++ rc = -EINVAL;
++ goto fail2;
++ }
++ frs->pt = ptrs;
++ EFRM_TRACE("%s: attaching PT resource " EFRM_RESOURCE_FMT
++ " to filter",
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(frs->pt->rs.rs_handle));
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ efhw_nic_ipfilter_attach(nic, frs->filter_idx, pti);
++ efrm_vi_resource_ref(frs->pt);
++ }
++ return 0;
++
++fail2:
++ efrm_vi_resource_release(frs->pt);
++ return rc;
++}
++EXPORT_SYMBOL(efrm_filter_resource_set_ptresource);
++
++int efrm_filter_resource_clear(struct filter_resource *frs)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_ASSERT(frs);
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ efhw_nic_ipfilter_clear(nic, frs->filter_idx);
++
++ return 0;
++}
++EXPORT_SYMBOL(efrm_filter_resource_clear);
++
++int
++__efrm_filter_resource_set(struct filter_resource *frs, int type,
++ unsigned saddr, uint16_t sport,
++ unsigned daddr, uint16_t dport)
++{
++ struct efhw_nic *nic;
++ int nic_i, rc = 0;
++ unsigned instance = EFRM_RESOURCE_INSTANCE(frs->pt->rs.rs_handle);
++
++ EFRM_ASSERT(frs);
++ EFRM_ASSERT(frs->pt);
++
++ if (efrm_nic_table.a_nic->devtype.variant >= 'B') {
++ /* Scatter setting must match the setting for
++ * the corresponding RX queue */
++ if (!(frs->pt->flags & EFHW_VI_JUMBO_EN))
++ type |= EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK;
++ }
++
++ EFRM_FOR_EACH_NIC_IN_SET(&frs->nic_set, nic_i, nic)
++ if (rc >= 0)
++ rc = efhw_nic_ipfilter_set(nic, type, &frs->filter_idx,
++ instance,
++ saddr, sport, daddr, dport);
++
++ return rc;
++}
++EXPORT_SYMBOL(__efrm_filter_resource_set);;
++
++int
++efrm_filter_resource_alloc(struct vi_resource *vi_parent,
++ struct filter_resource **frs_out)
++{
++ struct efhw_nic *nic;
++ int nic_i, rc, instance;
++ struct filter_resource *frs;
++
++ EFRM_ASSERT(frs_out);
++ EFRM_ASSERT(efrm_filter_manager);
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(&efrm_filter_manager->rm);
++ EFRM_ASSERT(vi_parent == NULL ||
++ EFRM_RESOURCE_TYPE(vi_parent->rs.rs_handle) ==
++ EFRM_RESOURCE_VI);
++
++ /* Allocate resource data structure. */
++ frs = kmalloc(sizeof(struct filter_resource), GFP_KERNEL);
++ if (!frs)
++ return -ENOMEM;
++ efrm_nic_set_clear(&frs->nic_set);
++
++ /* Allocate an instance. */
++ rc = kfifo_get(efrm_filter_manager->free_ids,
++ (unsigned char *)&instance, sizeof(instance));
++ if (rc != sizeof(instance)) {
++ EFRM_TRACE("%s: out of instances", __FUNCTION__);
++ EFRM_ASSERT(rc == 0);
++ rc = -EBUSY;
++ goto fail1;
++ }
++
++ /* Initialise the resource DS. */
++ efrm_resource_init(&frs->rs, EFRM_RESOURCE_FILTER, instance);
++ frs->pt = vi_parent;
++ if (frs->pt)
++ efrm_vi_resource_ref(frs->pt);
++ frs->filter_idx = -1;
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ efrm_nic_set_write(&frs->nic_set, nic_i, true);
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " Q %d idx %x",
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(frs->rs.rs_handle),
++ vi_parent == NULL ? -1 :
++ EFRM_RESOURCE_INSTANCE(vi_parent->rs.rs_handle),
++ frs->filter_idx);
++
++ /* Put it into the resource manager's table. */
++ rc = efrm_resource_manager_insert(&frs->rs);
++ if (rc != 0) {
++ if (atomic_dec_and_test(&frs->rs.rs_ref_count))
++ efrm_filter_resource_free(frs);
++ return rc;
++ }
++
++ *frs_out = frs;
++ return 0;
++
++fail1:
++ memset(frs, 0, sizeof(*frs));
++ kfree(frs);
++ return rc;
++}
++EXPORT_SYMBOL(efrm_filter_resource_alloc);
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/iobufset_resource.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/iobufset_resource.c 2008-05-19 00:33:29.429842497 +0300
+@@ -0,0 +1,373 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains non-contiguous I/O buffers support.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/efhw/iopage.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/iobufset.h>
++#include <ci/efrm/vi_resource_manager.h>
++#include <ci/efrm/buffer_table.h>
++
++#define EFRM_IOBUFSET_MAX_NUM_INSTANCES 0x00010000
++
++struct iobufset_resource_manager {
++ struct efrm_resource_manager rm;
++ struct kfifo *free_ids;
++};
++
++struct iobufset_resource_manager *efrm_iobufset_manager;
++
++#define iobsrs(rs1) iobufset_resource(rs1)
++
++/* Returns size of iobufset resource data structure. */
++static inline size_t iobsrs_size(int no_pages)
++{
++ return offsetof(struct iobufset_resource, bufs) +
++ no_pages * sizeof(efhw_iopage_t);
++}
++
++void efrm_iobufset_resource_free(struct iobufset_resource *rs)
++{
++ unsigned int no_pages;
++ unsigned int i;
++ int id;
++
++ EFRM_RESOURCE_ASSERT_VALID(&rs->rs, 1);
++ no_pages = rs->n_bufs;
++
++ if (rs->buf_tbl_alloc.base != (unsigned)-1)
++ efrm_buffer_table_free(&rs->buf_tbl_alloc);
++
++ /* see comment on call to efhw_iopage_alloc in the alloc routine above
++ for discussion on use of efrm_nic_table.a_nic here */
++ EFRM_ASSERT(efrm_nic_table.a_nic);
++ if (rs->order == 0) {
++ for (i = 0; i < rs->n_bufs; ++i)
++ efhw_iopage_free(efrm_nic_table.a_nic, &rs->bufs[i]);
++ } else {
++ /* it is important that this is executed in increasing page
++ * order because some implementations of
++ * efhw_iopages_init_from_iopage() assume this */
++ for (i = 0; i < rs->n_bufs;
++ i += rs->pages_per_contiguous_chunk) {
++ efhw_iopages_t iopages;
++ efhw_iopages_init_from_iopage(&iopages, &rs->bufs[i],
++ rs->order);
++ efhw_iopages_free(efrm_nic_table.a_nic, &iopages);
++ }
++ }
++
++ /* free the instance number */
++ id = EFRM_RESOURCE_INSTANCE(rs->rs.rs_handle);
++ EFRM_VERIFY_EQ(kfifo_put(efrm_iobufset_manager->free_ids,
++ (unsigned char *)&id, sizeof(id)), sizeof(id));
++
++ efrm_vi_resource_release(rs->evq);
++
++ EFRM_DO_DEBUG(memset(rs, 0, sizeof(*rs)));
++ if (iobsrs_size(no_pages) < PAGE_SIZE) {
++ kfree(rs);
++ } else {
++ vfree(rs);
++ }
++}
++EXPORT_SYMBOL(efrm_iobufset_resource_free);
++
++int
++efrm_iobufset_resource_alloc(int32_t n_pages,
++ int32_t pages_per_contiguous_chunk,
++ struct vi_resource *vi_evq,
++ bool phys_addr_mode,
++ uint32_t faultonaccess,
++ struct iobufset_resource **iobrs_out)
++{
++ struct iobufset_resource *iobrs;
++ int rc, instance, object_size;
++ unsigned int i;
++
++ EFRM_ASSERT(iobrs_out);
++ EFRM_ASSERT(efrm_iobufset_manager);
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(&efrm_iobufset_manager->rm);
++ EFRM_RESOURCE_ASSERT_VALID(&vi_evq->rs, 0);
++ EFRM_ASSERT(EFRM_RESOURCE_TYPE(vi_evq->rs.rs_handle) ==
++ EFRM_RESOURCE_VI);
++ EFRM_ASSERT(efrm_nic_table.a_nic);
++
++ /* allocate the resource data structure. */
++ object_size = iobsrs_size(n_pages);
++ if (object_size < PAGE_SIZE) {
++ /* this should be OK from a tasklet */
++ /* Necessary to do atomic alloc() as this
++ can be called from a weird-ass iSCSI context that is
++ !in_interrupt but is in_atomic - See BUG3163 */
++ iobrs = kmalloc(object_size, GFP_ATOMIC);
++ } else { /* can't do this within a tasklet */
++#ifndef NDEBUG
++ if (in_interrupt() || in_atomic()) {
++ EFRM_ERR("%s(): alloc->u.iobufset.in_n_pages=%d",
++ __FUNCTION__, n_pages);
++ EFRM_ASSERT(!in_interrupt());
++ EFRM_ASSERT(!in_atomic());
++ }
++#endif
++ iobrs = (struct iobufset_resource *) vmalloc(object_size);
++ }
++ if (iobrs == 0) {
++ rc = -ENOMEM;
++ goto fail1;
++ }
++
++ /* Allocate an instance number. */
++ rc = kfifo_get(efrm_iobufset_manager->free_ids,
++ (unsigned char *)&instance, sizeof(instance));
++ if (rc != sizeof(instance)) {
++ EFRM_TRACE("%s: out of instances", __FUNCTION__);
++ EFRM_ASSERT(rc == 0);
++ rc = -EBUSY;
++ goto fail3;
++ }
++
++ efrm_resource_init(&iobrs->rs, EFRM_RESOURCE_IOBUFSET, instance);
++
++ iobrs->evq = vi_evq;
++ efrm_vi_resource_ref(iobrs->evq);
++
++ iobrs->n_bufs = n_pages;
++ iobrs->pages_per_contiguous_chunk = pages_per_contiguous_chunk;
++ iobrs->order = fls(iobrs->pages_per_contiguous_chunk - 1);
++ iobrs->faultonaccess = faultonaccess;
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " %u pages", __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(iobrs->rs.rs_handle), iobrs->n_bufs);
++
++ /* Allocate the iobuffers. */
++ if (iobrs->order == 0) {
++ /* make sure iobufs are in a known state in case we don't
++ * finish our allocation */
++ for (i = 0; i < iobrs->n_bufs; ++i)
++ memset(&iobrs->bufs[i], 0, sizeof(iobrs->bufs[i]));
++
++ for (i = 0; i < iobrs->n_bufs; ++i) {
++ /* due to bug2426 we have to specifiy a NIC when
++ * allocating a DMAable page, which is a bit messy.
++ * For now we assume that if the page is suitable
++ * (e.g. DMAable) by one nic (efrm_nic_table.a_nic),
++ * it is suitable for all NICs.
++ * XXX I bet that breaks in Solaris.
++ */
++ rc = efhw_iopage_alloc(efrm_nic_table.a_nic,
++ &iobrs->bufs[i]);
++ if (rc < 0) {
++ EFRM_ERR("%s: failed (rc %d) to allocate "
++ "page (i=%u)", __FUNCTION__, rc, i);
++ goto fail4;
++ }
++ }
++ } else {
++ efhw_iopages_t iopages;
++ unsigned j;
++
++ /* make sure iobufs are in a known state in case we don't
++ * finish our allocation */
++ for (i = 0; i < iobrs->n_bufs; ++i)
++ memset(&iobrs->bufs[i], 0, sizeof(iobrs->bufs[i]));
++
++ for (i = 0; i < iobrs->n_bufs;
++ i += iobrs->pages_per_contiguous_chunk) {
++ rc = efhw_iopages_alloc(efrm_nic_table.a_nic,
++ &iopages, iobrs->order);
++ if (rc < 0) {
++ EFRM_ERR("%s: failed (rc %d) to allocate "
++ "pages (i=%u order %d)",
++ __FUNCTION__, rc, i, iobrs->order);
++ goto fail4;
++ }
++ for (j = 0; j < iobrs->pages_per_contiguous_chunk;
++ j++) {
++ /* some implementation of
++ * efhw_iopage_init_from_iopages() rely on
++ * this function being called for
++ * _all_ pages in the chunk */
++ efhw_iopage_init_from_iopages(
++ &iobrs->bufs[i + j],
++ &iopages, j);
++ }
++ }
++ }
++
++ iobrs->buf_tbl_alloc.base = (unsigned)-1;
++
++ if (!phys_addr_mode) {
++ unsigned instance = EFAB_VI_RESOURCE_INSTANCE(iobrs->evq);
++ /* Allocate space in the NIC's buffer table. */
++ rc = efrm_buffer_table_alloc(fls(iobrs->n_bufs - 1),
++ &iobrs->buf_tbl_alloc);
++ if (rc < 0) {
++ EFRM_ERR("%s: failed (%d) to alloc %d buffer table "
++ "entries", __FUNCTION__, rc, iobrs->n_bufs);
++ goto fail5;
++ }
++ EFRM_ASSERT(((unsigned)1 << iobrs->buf_tbl_alloc.order) >=
++ (unsigned)iobrs->n_bufs);
++
++ /* Initialise the buffer table entries. */
++ for (i = 0; i < iobrs->n_bufs; ++i) {
++ /*\ ?? \TODO burst them! */
++ efrm_buffer_table_set(&iobrs->buf_tbl_alloc, i,
++ efhw_iopage_dma_addr(&iobrs->
++ bufs[i]),
++ instance);
++ }
++ efrm_buffer_table_commit();
++ }
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " %d pages @ "
++ EFHW_BUFFER_ADDR_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(iobrs->rs.rs_handle),
++ iobrs->n_bufs, EFHW_BUFFER_ADDR(iobrs->buf_tbl_alloc.base,
++ 0));
++
++ /* Put it into the resource manager's table. */
++ rc = efrm_resource_manager_insert(&iobrs->rs);
++ if (rc != 0) {
++ if (atomic_dec_and_test(&iobrs->rs.rs_ref_count))
++ efrm_iobufset_resource_free(iobrs);
++ return rc;
++ }
++
++ *iobrs_out = iobrs;
++ return 0;
++
++fail5:
++ i = iobrs->n_bufs;
++fail4:
++ /* see comment on call to efhw_iopage_alloc above for a discussion
++ * on use of efrm_nic_table.a_nic here */
++ if (iobrs->order == 0) {
++ while (i--) {
++ efhw_iopage_t *page = &iobrs->bufs[i];
++ efhw_iopage_free(efrm_nic_table.a_nic, page);
++ }
++ } else {
++ unsigned int j;
++ for (j = 0; j < i; j += iobrs->pages_per_contiguous_chunk) {
++ efhw_iopages_t iopages;
++
++ EFRM_ASSERT(j % iobrs->pages_per_contiguous_chunk
++ == 0);
++ /* it is important that this is executed in increasing
++ * page order because some implementations of
++ * efhw_iopages_init_from_iopage() assume this */
++ efhw_iopages_init_from_iopage(&iopages,
++ &iobrs->bufs[j],
++ iobrs->order);
++ efhw_iopages_free(efrm_nic_table.a_nic, &iopages);
++ }
++ }
++ efrm_vi_resource_release(iobrs->evq);
++fail3:
++ if (object_size < PAGE_SIZE) {
++ kfree(iobrs);
++ } else {
++ vfree(iobrs);
++ }
++fail1:
++ return rc;
++}
++EXPORT_SYMBOL(efrm_iobufset_resource_alloc);
++
++static void iobufset_rm_dtor(struct efrm_resource_manager *rm)
++{
++ EFRM_ASSERT(&efrm_iobufset_manager->rm == rm);
++ kfifo_vfree(efrm_iobufset_manager->free_ids);
++}
++
++int
++efrm_create_iobufset_resource_manager(struct efrm_resource_manager **rm_out)
++{
++ int rc, max;
++
++ EFRM_ASSERT(rm_out);
++
++ efrm_iobufset_manager =
++ kmalloc(sizeof(*efrm_iobufset_manager), GFP_KERNEL);
++ if (efrm_iobufset_manager == 0)
++ return -ENOMEM;
++ memset(efrm_iobufset_manager, 0, sizeof(*efrm_iobufset_manager));
++
++ /*
++ * Bug 1145, 1370: We need to set initial size of both the resource
++ * table and instance id table so they never need to grow as we
++ * want to be allocate new iobufset at tasklet time. Lets make
++ * a pessimistic guess at maximum number of iobufsets possible.
++ * Could be less because
++ * - jumbo frames have same no of packets per iobufset BUT more
++ * pages per buffer
++ * - buffer table entries used independently of iobufsets by
++ * sendfile
++ *
++ * Based on TCP/IP stack setting of PKTS_PER_SET_S=5 ...
++ * - can't use this define here as it breaks the layering.
++ */
++#define MIN_PAGES_PER_IOBUFSET (1 << 4)
++
++ max = efrm_buffer_table_size() / MIN_PAGES_PER_IOBUFSET;
++ max = min_t(int, max, EFRM_IOBUFSET_MAX_NUM_INSTANCES);
++
++ rc = efrm_kfifo_id_ctor(&efrm_iobufset_manager->free_ids,
++ 0, max, &efrm_iobufset_manager->rm.rm_lock);
++ if (rc != 0)
++ goto fail1;
++
++ rc = efrm_resource_manager_ctor(&efrm_iobufset_manager->rm,
++ iobufset_rm_dtor, "IOBUFSET",
++ EFRM_RESOURCE_IOBUFSET, max);
++ if (rc < 0)
++ goto fail2;
++
++ *rm_out = &efrm_iobufset_manager->rm;
++ return 0;
++
++fail2:
++ kfifo_vfree(efrm_iobufset_manager->free_ids);
++fail1:
++ EFRM_DO_DEBUG(memset(efrm_iobufset_manager, 0,
++ sizeof(*efrm_iobufset_manager)));
++ kfree(efrm_iobufset_manager);
++ return rc;
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/iopage.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/iopage.c 2008-05-19 00:33:29.429842497 +0300
+@@ -0,0 +1,101 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides Linux-specific implementation for iopage API used
++ * from efhw library.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/driver/resource/linux_efhw_nic.h>
++#include "kernel_compat.h"
++#include <ci/efhw/common_sysdep.h> /* for dma_addr_t */
++
++int efhw_iopage_alloc(struct efhw_nic *nic, efhw_iopage_t *p)
++{
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++ dma_addr_t handle;
++ void *kva;
++
++ kva = efrm_pci_alloc_consistent(lnic->pci_dev, PAGE_SIZE,
++ &handle);
++ if (kva == 0)
++ return -ENOMEM;
++
++ EFHW_ASSERT((handle & ~PAGE_MASK) == 0);
++
++ memset((void *)kva, 0, PAGE_SIZE);
++ efhw_page_init_from_va(&p->p, kva);
++
++ p->dma_addr = handle;
++
++ return 0;
++}
++
++void efhw_iopage_free(struct efhw_nic *nic, efhw_iopage_t *p)
++{
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++ EFHW_ASSERT(efhw_page_is_valid(&p->p));
++
++ efrm_pci_free_consistent(lnic->pci_dev, PAGE_SIZE,
++ efhw_iopage_ptr(p), p->dma_addr);
++}
++
++int efhw_iopages_alloc(struct efhw_nic *nic, efhw_iopages_t *p, unsigned order)
++{
++ unsigned bytes = 1u << (order + PAGE_SHIFT);
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++ dma_addr_t handle;
++ caddr_t addr;
++ int gfp_flag;
++
++ /* Set __GFP_COMP if available to make reference counting work.
++ * This is recommended here:
++ * http://www.forbiddenweb.org/viewtopic.php?id=83167&page=4#348331
++ */
++ gfp_flag = ((in_atomic() ? GFP_ATOMIC : GFP_KERNEL) | __GFP_COMP);
++ addr = efrm_dma_alloc_coherent(&lnic->pci_dev->dev, bytes, &handle,
++ gfp_flag);
++ if (addr == NULL)
++ return -ENOMEM;
++
++ EFHW_ASSERT((handle & ~PAGE_MASK) == 0);
++
++ p->order = order;
++ p->dma_addr = handle;
++ p->kva = addr;
++
++ return 0;
++}
++
++void efhw_iopages_free(struct efhw_nic *nic, efhw_iopages_t *p)
++{
++ unsigned bytes = 1u << (p->order + PAGE_SHIFT);
++ struct linux_efhw_nic *lnic = linux_efhw_nic(nic);
++
++ efrm_dma_free_coherent(&lnic->pci_dev->dev, bytes,
++ (void *)p->kva, p->dma_addr);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/kernel_compat.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/kernel_compat.c 2008-05-19 00:33:29.429842497 +0300
+@@ -0,0 +1,584 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides compatibility layer for various Linux kernel versions
++ * (starting from 2.6.9 RHEL kernel).
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#define IN_KERNEL_COMPAT_C
++#include <linux/types.h>
++#include <ci/efrm/debug.h>
++#include "kernel_compat.h"
++
++/* Set this to 1 to enable very basic counting of iopage(s) allocations, then
++ * call dump_iopage_counts() to show the number of current allocations of
++ * orders 0-7.
++ */
++#define EFRM_IOPAGE_COUNTS_ENABLED 0
++
++
++
++/* I admit that it's a bit ugly going straight to the field, but it
++ * seems easiest given that get_page followed by put_page on a page
++ * with PG_reserved set will increment the ref count on 2.6.14 and
++ * below, but not 2.6.15. Also, RedHat have hidden put_page_testzero
++ * in a header file which produces warnings when compiled. This
++ * doesn't agree with our use of -Werror.
++ */
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,5)
++# define page_count_field(pg) ((pg)->count)
++#else
++# define page_count_field(pg) ((pg)->_count)
++#endif
++
++#define inc_page_count(page) atomic_inc(&page_count_field(page))
++#define dec_page_count(page) atomic_dec(&page_count_field(page))
++
++/* Bug 5531: set_page_count doesn't work if the new page count is an
++ * expression. */
++#define ci_set_page_count(page, n) set_page_count(page, (n))
++
++ /* Bug 3965: Gak! Reference counts just don't work on memory
++ * allocated through pci_alloc_consistent. Different versions and
++ * architectures do different things. There are several interacting
++ * bugs/features which have been described below and then summarised
++ * in a table for each kernel version. For each feature, there is a
++ * question, a short description, a hint at how to examine the
++ * kernel code for this feature and a description of the keys in the
++ * table.
++ *
++ * A. Is PG_compound set on multi-page allocations?
++ *
++ * When a multi-page allocation succeeds, the kernel sets the
++ * reference count of the first page to one and the count of the
++ * remaining pages to zero. This is an immediate problem because
++ * if these pages are mapped into user space, the VM will do
++ * get_page followed by put_page, at which point the reference
++ * count will return to zero and the page will be freed.
++ * PG_compound was introduced in 2.6.0 and back-ported to rhel3
++ * kernels. When it is set, all the pages have a pointer to the
++ * first page so that they can share the reference count. If
++ * PG_compound is set, calling get_page(pg+1) can change
++ * page_count(pg). It was originally set on all multi-page
++ * allocations, but later only set if the __GFP_COMP flag was
++ * provided to the allocator.
++ *
++ * See mm/page_alloc.c
++ * Does prep_compound_page get called when __GFP_COMP not set?
++ *
++ * Keys:
++ * NotDef - prep_compound_page and PG_compound are not defined.
++ * Comp - prep_compound_page is called for any multi-page allocation.
++ * Opt - prep_compound_page is only called if __GFP_COMP is set.
++ * OptInv - prep_compound_page is only called if __GFP_NO_COMP is not set.
++ *
++ * B. Are bounce buffers ever used to satisfy pci_alloc_consistent?
++ * (x86_64 only)
++ *
++ * 2.6 kernels introduced bounce buffers on x86_64 machines to access
++ * memory above 4G when using the DMA mapping API. At some point,
++ * code was added to allow pci_alloc_consistent/dma_alloc_coherent to
++ * allocate memory from the bounce buffers if the general purpose
++ * allocator produced memory which wasn't suitable. Such memory can
++ * be recognised by the PG_reserved bit being set. At a later point,
++ * the __GFP_DMA32 flag was added and used to restrict the allocator
++ * to below 4G. The effect of this later change was that 4G capable
++ * cards would no longer get memory from the bounce buffers, although
++ * a card which can address less than 4G might get memory from the
++ * bounce buffers.
++ *
++ * See dma_alloc_coherent or pci_alloc_consistent in
++ * arch/x86_64/kernel/pci-gart.c or arch/x86/kernel/pci-dma_64.c
++ * Is (gfp |= GFP_DMA32) before dma_alloc_pages?
++ * Is swiotlb_alloc_coherent called?
++ *
++ * Keys:
++ * NU - bounce buffers are Never Used
++ * Used - bounce buffers are sometimes used
++ *
++ * C. Does munmap decrement the reference count of a PG_reserved page?
++ *
++ * Originally, the munmap code would not decrement the reference count
++ * of a page which had PG_reserved set. At some point in the 2.6
++ * series, VM_PFNMAP was introduced and could be set on a vma to
++ * indicate that no pages in that vma should have the reference count
++ * decremented (unless they are copy-on-write copies). At that point,
++ * the check for PG_reserved pages in the munmap code path was
++ * removed. Some hackery in vm_normal_page means that a VM_PFNMAP vma
++ * must map contiguous physical pages. As a result, such pages should
++ * be mapped during mmap using remap_pfn_range (for an example, see
++ * drivers/char/mem.c).
++ *
++ * In 2.6 kernels: See release_pages in mm/swap.c
++ * Does PageReserved get tested?
++ * In 2.6 kernels: See mm/memory.c
++ * Is VM_PFNMAP used?
++ * In 2.4 kernels: See __free_pte in mm/memory.c
++ * Does PageReserved get tested?
++ * In 2.4 kernels: See __free_pages in mm/page_alloc.c
++ * Does PageReserved get tested?
++ *
++ * Keys:
++ * resv - The reference count is not touched for PG_reserved pages.
++ * pfnmap - The VM_PFNMAP flag is checked instead of PG_reserved.
++ *
++ * D. Does munmap honour the PG_compound bit?
++ *
++ * When PG_compound was originally introduced, the munmap code path
++ * didn't check it before decrementing the reference count on the
++ * page. As a result, the wrong reference count would be updated if a
++ * PG_compound page was ever mapped into user space.
++ *
++ * In 2.6 kernels: See release_pages in mm/swap.c
++ * Does PageCompound get tested?
++ * In 2.4 kernels: See __free_pages in mm/page_alloc.c
++ * Does PageCompound get tested?
++ *
++ * Keys:
++ * NotHon - The PG_compound bit isn't honoured by munmap.
++ * Hon - The PG_compound bit is honoured by munmap.
++ *
++ * OS A B C D
++ * 2.4.18 NotDef NU resv NotHon
++ * 2.4.29 NotDef NU resv NotHon
++ * 2.4.20-31.9 rhl9 NotDef NU resv NotHon
++ *
++ * 2.4.21-4.EL rhel3 Comp NU resv Hon
++ * 2.4.21-15.EL rhel3 Comp NU resv Hon
++ * 2.4.21-32.EL rhel3 Comp NU resv Hon
++ * 2.4.21-40.EL rhel3 Comp NU resv Hon
++ *
++ * 2.6.0 Comp NU resv NotHon
++ *
++ * 2.6.5-7.97 sles9 OptInv NU resv NotHon
++ * 2.6.9 rhel4 Opt NU resv NotHon
++ *
++ * 2.6.11 fc4 ? ? ? ?
++ * 2.6.12 fc4 Opt Used resv NotHon
++ * 2.6.13 Opt Used resv NotHon
++ *
++ * 2.6.15 Opt NU pfnmap NotHon
++ *
++ * 2.6.16 Opt NU pfnmap Hon
++ * 2.6.16.9 Opt NU pfnmap Hon
++ * 2.6.17.2 Opt NU pfnmap Hon
++ * 2.6.24-rc7 k.org Opt NU pfnmap Hon
++ *
++ * This LKML thread gives some low down on mapping pages into user
++ * space and using DMA.
++ * http://www.forbiddenweb.org/viewtopic.php?id=83167&page=1
++ *
++ * There is no problem with single page allocations (until some
++ * kernel hands us a PG_reserved page and expects us to use
++ * VM_PFNMAP on the vma).
++ *
++ * Bug 5450: Most kernels set the reference count to one on the
++ * first sub-page of a high-order page allocation and zero on
++ * subsequent sub-pages. Some kernels, however, set the page count
++ * to one on all the sub-pages. The SLES 9 range are affected, as
++ * are kernels built without CONFIG_MMU defined.
++ *
++ * Possible strategies for multi-page allocations:
++ *
++ * EFRM_MMAP_USE_COMPOUND
++ * 1. Allocate a compound page. Reference counting should then work
++ * on the whole allocation. This is a good theory, but is broken
++ * by bug/feature D (above).
++ *
++ * EFRM_MMAP_USE_SPLIT
++ * 2. Convert the multi-page allocation to many single page
++ * allocations. This involves incrementing the reference counts
++ * and clearing PG_compound on all the pages (including the
++ * first). The references should be released _after_ calling
++ * pci_free_consistent so that that call doesn't release the
++ * memory.
++ *
++ * EFRM_MMAP_USE_INCREMENT
++ * 3. Increment the reference count on all the pages after
++ * allocating and decrement them again before freeing. This gets
++ * round the zero reference count problem. It doesn't handle the
++ * case where someone else is holding a reference to one of our
++ * pages when we free the pages, but we think VM_IO stops this
++ * from happening.
++ */
++
++/* Should we use strategy 1? This can be forced on us by the OS. */
++#if defined(PG_compound)
++#define EFRM_MMAP_USE_COMPOUND 1
++#else
++#define EFRM_MMAP_USE_COMPOUND 0
++#endif
++
++/* Should we use strategy 2? This can be used even if strategy 1 is
++ * used. */
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16)
++#define EFRM_MMAP_USE_SPLIT 1
++#else
++#define EFRM_MMAP_USE_SPLIT 0
++#endif
++
++/* Should we use strategy 3? There's no point doing this if either
++ * strategy 1 or strategy 2 is used. */
++#if !EFRM_MMAP_USE_COMPOUND && !EFRM_MMAP_USE_SPLIT
++#error "We shouldn't have to use this strategy."
++#define EFRM_MMAP_USE_INCREMENT 1
++#else
++#define EFRM_MMAP_USE_INCREMENT 0
++#endif
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
++#define EFRM_MMAP_RESET_REFCNT 1
++#else
++#define EFRM_MMAP_RESET_REFCNT 0
++#endif
++
++/* NB. 2.6.17 has renamed SetPageCompound to __SetPageCompound and
++ * ClearPageCompound to __ClearPageCompound. */
++#if ((defined(PageCompound) != defined(PG_compound)) || \
++ (defined(SetPageCompound) != defined(PG_compound) && \
++ defined(__SetPageCompound) != defined(PG_compound)) || \
++ (defined(ClearPageCompound) != defined(PG_compound) && \
++ defined(__ClearPageCompound) != defined(PG_compound)) || \
++ (defined(__GFP_COMP) && !defined(PG_compound)))
++#error Mismatch of defined page-flags.
++#endif
++
++extern int use_pci_alloc; /* Use pci_alloc_consistent to alloc iopages */
++
++/****************************************************************************
++ *
++ * allocate a buffer suitable for DMA to/from the NIC
++ *
++ ****************************************************************************/
++
++static inline void pci_mmap_pages_hack_after_alloc(caddr_t kva, unsigned order)
++{
++ unsigned pfn = __pa(kva) >> PAGE_SHIFT;
++ struct page *start_pg = pfn_to_page(pfn);
++#if !defined(NDEBUG) || EFRM_MMAP_USE_SPLIT
++ struct page *end_pg = start_pg + (1 << order);
++ struct page *pg;
++#endif
++
++ /* Compound pages don't get created for order 0 pages and there's no
++ * fixing up needs to be done. */
++ if (order == 0)
++ return;
++
++ /* If we've been given a reserved page then it must have come from
++ * the bounce buffer pool. */
++ if (PageReserved(start_pg)) {
++#if defined(VM_PFNMAP) || !defined(__x86_64__)
++ /* Kernel allocated reserved pages when not expected */
++ BUG();
++#endif
++ return;
++ }
++
++ /* Check the page count and PG_compound bit. */
++#ifndef NDEBUG
++# if defined(PG_compound)
++ EFRM_ASSERT(PageCompound(start_pg) == EFRM_MMAP_USE_COMPOUND);
++# endif
++ EFRM_ASSERT(page_count(start_pg) == 1);
++
++ {
++ /* Some kernels have the page count field hold (ref_count-1)
++ * rather than (ref_count). This is so that decrementing the
++ * reference count to "zero" causes the internal value to change
++ * from 0 to -1 which sets the carry flag. Other kernels store
++ * the real reference count value in the obvious way. We handle
++ * this here by reading the reference count field of the first
++ * page, which is always 1. */
++ int pg_count_zero;
++ pg_count_zero = atomic_read(&page_count_field(start_pg)) - 1;
++ for (pg = start_pg + 1; pg < end_pg; pg++) {
++ int pg_count;
++# if defined(PG_compound)
++ EFRM_ASSERT(PageCompound(pg) == EFRM_MMAP_USE_COMPOUND);
++# endif
++
++ /* Bug 5450: Some kernels initialise the page count
++ * to one for pages other than the first and some
++ * leave it at zero. We allow either behaviour
++ * here, but disallow anything strange. Newer
++ * kernels only define set_page_count in an
++ * internal header file, so we have to make do with
++ * incrementing and decrementing the reference
++ * count. Fortunately, those kernels don't set the
++ * reference count to one on all the pages. */
++ pg_count = atomic_read(&page_count_field(pg));
++# if EFRM_MMAP_RESET_REFCNT
++ if (pg_count != pg_count_zero)
++ EFRM_ASSERT(pg_count == pg_count_zero + 1);
++# else
++ EFRM_ASSERT(pg_count == pg_count_zero);
++# endif
++ }
++ }
++#endif
++
++ /* Split the multi-page allocation if necessary. */
++#if EFRM_MMAP_USE_SPLIT
++ for (pg = start_pg; pg < end_pg; pg++) {
++
++ /* This is no longer a compound page. */
++# if EFRM_MMAP_USE_COMPOUND
++ ClearPageCompound(pg);
++ EFRM_ASSERT(PageCompound(pg) == 0);
++# endif
++
++# ifndef NDEBUG
++ {
++ int pg_count = page_count(pg);
++ /* Bug 5450: The page count can be zero or one here. */
++ if (pg == start_pg) {
++ EFRM_ASSERT(pg_count == 1);
++ } else {
++# if EFRM_MMAP_RESET_REFCNT
++ if (pg_count != 0)
++ EFRM_ASSERT(pg_count == 1);
++# else
++ EFRM_ASSERT(pg_count == 0);
++# endif
++ }
++ }
++# endif
++
++ /* Get a reference which will be released after the pages have
++ * been passed back to pci_free_consistent. */
++# if EFRM_MMAP_RESET_REFCNT
++ /* Bug 5450: Reset the reference count since the count might
++ * already be 1. */
++ ci_set_page_count(pg, (pg == start_pg) ? 2 : 1);
++# else
++ get_page(pg);
++# endif
++ }
++#endif
++
++ /* Fudge the reference count if necessary. */
++#if EFRM_MMAP_USE_INCREMENT
++ for (pg = start_pg; pg < end_pg; pg++)
++ inc_page_count(pg);
++#endif
++}
++
++static inline void pci_mmap_pages_hack_before_free(caddr_t kva, unsigned order)
++{
++#if EFRM_MMAP_USE_INCREMENT || !defined(NDEBUG)
++ /* Drop the references taken in pci_mmap_pages_hack_after_alloc */
++ unsigned pfn = __pa(kva) >> PAGE_SHIFT;
++ struct page *start_pg = pfn_to_page(pfn);
++ struct page *end_pg = start_pg + (1 << order);
++ struct page *pg;
++
++ /* Compound pages don't get created for order 0 pages and there's no
++ * fixing up needs to be done. */
++ if (order == 0)
++ return;
++
++ if (PageReserved(start_pg))
++ return;
++
++# if EFRM_MMAP_USE_INCREMENT
++ for (pg = start_pg; pg < end_pg; pg++)
++ dec_page_count(pg);
++# endif
++
++#if !defined(NDEBUG)
++ EFRM_ASSERT(page_count(start_pg) == 1+EFRM_MMAP_USE_SPLIT);
++
++# if EFRM_MMAP_USE_COMPOUND && !EFRM_MMAP_USE_SPLIT
++ for (pg = start_pg; pg < end_pg; pg++)
++ EFRM_ASSERT(PageCompound(pg));
++# else
++ for (pg = start_pg+1; pg < end_pg; pg++) {
++ unsigned exp_pg_count = EFRM_MMAP_USE_SPLIT;
++ /* NB. If this assertion fires, either we've messed up the
++ * page counting or someone is holding on to a reference.
++ */
++ EFRM_ASSERT(page_count(pg) == exp_pg_count);
++ }
++# endif
++#endif
++
++#endif
++}
++
++static inline void pci_mmap_pages_hack_after_free(caddr_t kva, unsigned order)
++{
++#if EFRM_MMAP_USE_SPLIT
++ /* Drop the references taken in pci_mmap_pages_hack_after_alloc */
++ unsigned pfn = __pa(kva) >> PAGE_SHIFT;
++ struct page *start_pg = pfn_to_page(pfn);
++ struct page *end_pg = start_pg + (1 << order);
++ struct page *pg;
++
++ /* Compound pages don't get created for order 0 pages and there's no
++ * fixing up needs to be done. */
++ if (order == 0)
++ return;
++
++ if (PageReserved(start_pg))
++ return;
++
++ for (pg = start_pg; pg < end_pg; pg++) {
++ EFRM_ASSERT(page_count(pg) == 1);
++ put_page(pg);
++ }
++#endif
++}
++
++
++#if EFRM_IOPAGE_COUNTS_ENABLED
++
++static int iopage_counts[8];
++
++void dump_iopage_counts(void)
++{
++ EFRM_NOTICE("iopage counts: %d %d %d %d %d %d %d %d", iopage_counts[0],
++ iopage_counts[1], iopage_counts[2], iopage_counts[3],
++ iopage_counts[4], iopage_counts[5], iopage_counts[6],
++ iopage_counts[7]);
++}
++
++#endif
++
++
++
++/*********** pci_alloc_consistent / pci_free_consistent ***********/
++
++void *efrm_dma_alloc_coherent(struct device *dev, size_t size,
++ dma_addr_t *dma_addr, int flag)
++{
++ struct pci_dev *pci_dev;
++ void *ptr;
++ unsigned order;
++ EFRM_IOMMU_DECL;
++
++ order = __ffs(size/PAGE_SIZE);
++ EFRM_ASSERT(size == (PAGE_SIZE<<order));
++
++ /* NB. The caller may well set __GFP_COMP. However we can't
++ * rely on this working on older kernels. 2.6.9 only acts on
++ * __GFP_COMP if CONFIG_HUGETLB_PAGE is defined. If the flag
++ * did have an effect then PG_compound will be set on the
++ * pages. */
++
++ if (use_pci_alloc) {
++ /* Can't take a spinlock here since the allocation can
++ * block. */
++ ptr = dma_alloc_coherent(dev, size, dma_addr, flag);
++ if (ptr == NULL)
++ return ptr;
++ } else {
++#ifdef CONFIG_SWIOTLB /* BUG1340 */
++ if (swiotlb) {
++ EFRM_ERR("%s: This kernel is using DMA bounce "
++ "buffers. Please upgrade kernel to "
++ "linux2.6 or reduce the amount of RAM "
++ "with mem=XXX.", __FUNCTION__);
++ return NULL;
++ }
++#endif
++ ptr = (void *)__get_free_pages(flag, order);
++
++ if (ptr == NULL)
++ return NULL;
++
++ EFRM_IOMMU_LOCK();
++ pci_dev = container_of(dev, struct pci_dev, dev);
++ *dma_addr = pci_map_single(pci_dev, ptr, size,
++ PCI_DMA_BIDIRECTIONAL);
++ EFRM_IOMMU_UNLOCK();
++ if (pci_dma_mapping_error(*dma_addr)) {
++ free_pages((unsigned long)ptr, order);
++ return NULL;
++ }
++ }
++
++#ifndef CONFIG_IA64
++ pci_mmap_pages_hack_after_alloc(ptr, order);
++#endif
++
++#if EFRM_IOPAGE_COUNTS_ENABLED
++ if (order < 8)
++ iopage_counts[order]++;
++ else
++ EFRM_ERR("Huge iopages alloc (order=%d) ??? (not counted)",
++ order);
++#endif
++
++ return ptr;
++}
++
++void efrm_dma_free_coherent(struct device *dev, size_t size,
++ void *ptr, dma_addr_t dma_addr)
++{
++ struct pci_dev *pci_dev;
++ unsigned order;
++ EFRM_IOMMU_DECL;
++
++ order = __ffs(size/PAGE_SIZE);
++ EFRM_ASSERT(size == (PAGE_SIZE<<order));
++
++#if EFRM_IOPAGE_COUNTS_ENABLED
++ if (order < 8)
++ --iopage_counts[order];
++ else
++ EFRM_ERR("Huge iopages free (order=%d) ??? (not counted)",
++ order);
++#endif
++#ifndef CONFIG_IA64
++ pci_mmap_pages_hack_before_free(ptr, order);
++#endif
++ if (use_pci_alloc) {
++ EFRM_IOMMU_LOCK();
++ dma_free_coherent(dev, size, ptr, dma_addr);
++ EFRM_IOMMU_UNLOCK();
++ } else {
++ pci_dev = container_of(dev, struct pci_dev, dev);
++ EFRM_IOMMU_LOCK();
++ efrm_pci_unmap_single(pci_dev, dma_addr, size,
++ PCI_DMA_BIDIRECTIONAL);
++ EFRM_IOMMU_UNLOCK();
++
++ free_pages((unsigned long)ptr, order);
++ }
++
++#ifndef CONFIG_IA64
++ pci_mmap_pages_hack_after_free(ptr, order);
++#endif
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/kernel_compat.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/kernel_compat.h 2008-05-19 00:33:29.429842497 +0300
+@@ -0,0 +1,239 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides compatibility layer for various Linux kernel versions
++ * (starting from 2.6.9 RHEL kernel).
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef DRIVER_LINUX_RESOURCE_KERNEL_COMPAT_H
++#define DRIVER_LINUX_RESOURCE_KERNEL_COMPAT_H
++
++#include <linux/version.h>
++
++/********* wait_for_completion_timeout() ********************/
++#include <linux/sched.h>
++
++/* RHEL_RELEASE_CODE from linux/version.h is only defined for 2.6.9-55EL
++ * UTS_RELEASE is unfortunately unusable
++ * Really only need this fix for <2.6.9-34EL
++ */
++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,11)) && \
++ !defined(RHEL_RELEASE_CODE)
++
++static inline unsigned long fastcall __sched
++efrm_wait_for_completion_timeout(struct completion *x, unsigned long timeout)
++{
++ might_sleep();
++
++ spin_lock_irq(&x->wait.lock);
++ if (!x->done) {
++ DECLARE_WAITQUEUE(wait, current);
++
++ wait.flags |= WQ_FLAG_EXCLUSIVE;
++ __add_wait_queue_tail(&x->wait, &wait);
++ do {
++ __set_current_state(TASK_UNINTERRUPTIBLE);
++ spin_unlock_irq(&x->wait.lock);
++ timeout = schedule_timeout(timeout);
++ spin_lock_irq(&x->wait.lock);
++ if (!timeout) {
++ __remove_wait_queue(&x->wait, &wait);
++ goto out;
++ }
++ } while (!x->done);
++ __remove_wait_queue(&x->wait, &wait);
++ }
++ x->done--;
++out:
++ spin_unlock_irq(&x->wait.lock);
++ return timeout;
++}
++
++# ifdef wait_for_completion_timeout
++# undef wait_for_completion_timeout
++# endif
++# define wait_for_completion_timeout efrm_wait_for_completion_timeout
++
++#endif
++
++/********* pci_map_*() ********************/
++
++#include <linux/pci.h>
++
++/* Bug 4560: Some kernels leak IOMMU entries under heavy load. Use a
++ * spinlock to serialise access where possible to alleviate the
++ * problem.
++ *
++ * NB. This is duplicated in the net driver. Please keep in sync. */
++#if ((LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && \
++ (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)) && \
++ defined(__x86_64__) && defined(CONFIG_SMP))
++
++#define EFRM_HAVE_IOMMU_LOCK 1
++
++#if ((LINUX_VERSION_CODE == KERNEL_VERSION(2,6,5)) && \
++ defined(CONFIG_SUSE_KERNEL))
++#define EFRM_NEED_ALTERNATE_MAX_PFN 1
++#endif
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,15)
++#if defined(CONFIG_GART_IOMMU)
++#define EFRM_NO_IOMMU no_iommu
++#else
++#define EFRM_NO_IOMMU 1
++#endif
++#else
++#define EFRM_NO_IOMMU 0
++#endif
++
++/* Set to 0 if we should never use the lock. Set to 1 if we should
++ * automatically determine if we should use the lock. Set to 2 if we
++ * should always use the lock. */
++extern unsigned int efx_use_iommu_lock;
++/* Defined in the net driver. */
++extern spinlock_t efx_iommu_lock;
++/* Non-zero if there is a card which needs the lock. */
++extern int efrm_need_iommu_lock;
++
++/* The IRQ state is needed if the lock is being used. The flag is
++ * cached to ensure that every lock is followed by an unlock, even
++ * if the global flag changes in the middle of the operation. */
++
++#define EFRM_IOMMU_DECL \
++ unsigned long efx_iommu_irq_state = 0; \
++ int efx_iommu_using_lock;
++#define EFRM_IOMMU_LOCK() \
++ do { \
++ efx_iommu_using_lock = (efx_use_iommu_lock && \
++ (efrm_need_iommu_lock || \
++ efx_use_iommu_lock >= 2)); \
++ if (efx_iommu_using_lock) \
++ spin_lock_irqsave(&efx_iommu_lock, efx_iommu_irq_state);\
++ } while (0)
++#define EFRM_IOMMU_UNLOCK() \
++ do { \
++ if (efx_iommu_using_lock) \
++ spin_unlock_irqrestore(&efx_iommu_lock, \
++ efx_iommu_irq_state); \
++ } while (0)
++
++#else /* defined(__x86_64__) && defined(CONFIG_SMP) */
++
++#define EFRM_HAVE_IOMMU_LOCK 0
++#define EFRM_IOMMU_DECL
++#define EFRM_IOMMU_LOCK() do {} while (0)
++#define EFRM_IOMMU_UNLOCK() do {} while (0)
++
++#endif
++
++static inline dma_addr_t efrm_pci_map_single(struct pci_dev *hwdev, void *ptr,
++ size_t size, int direction)
++{
++ dma_addr_t dma_addr;
++ EFRM_IOMMU_DECL;
++
++ EFRM_IOMMU_LOCK();
++ dma_addr = pci_map_single(hwdev, ptr, size, direction);
++ EFRM_IOMMU_UNLOCK();
++
++ return dma_addr;
++}
++
++static inline void efrm_pci_unmap_single(struct pci_dev *hwdev,
++ dma_addr_t dma_addr, size_t size,
++ int direction)
++{
++ EFRM_IOMMU_DECL;
++
++ EFRM_IOMMU_LOCK();
++ pci_unmap_single(hwdev, dma_addr, size, direction);
++ EFRM_IOMMU_UNLOCK();
++}
++
++static inline dma_addr_t efrm_pci_map_page(struct pci_dev *hwdev,
++ struct page *page,
++ unsigned long offset, size_t size,
++ int direction)
++{
++ dma_addr_t dma_addr;
++ EFRM_IOMMU_DECL;
++
++ EFRM_IOMMU_LOCK();
++ dma_addr = pci_map_page(hwdev, page, offset, size, direction);
++ EFRM_IOMMU_UNLOCK();
++
++ return dma_addr;
++}
++
++static inline void efrm_pci_unmap_page(struct pci_dev *hwdev,
++ dma_addr_t dma_addr, size_t size,
++ int direction)
++{
++ EFRM_IOMMU_DECL;
++
++ EFRM_IOMMU_LOCK();
++ pci_unmap_page(hwdev, dma_addr, size, direction);
++ EFRM_IOMMU_UNLOCK();
++}
++
++#ifndef IN_KERNEL_COMPAT_C
++# ifndef __GFP_COMP
++# define __GFP_COMP 0
++# endif
++# ifndef __GFP_ZERO
++# define __GFP_ZERO 0
++# endif
++#endif
++
++extern void *efrm_dma_alloc_coherent(struct device *dev, size_t size,
++ dma_addr_t *dma_addr, int flag);
++
++extern void efrm_dma_free_coherent(struct device *dev, size_t size,
++ void *ptr, dma_addr_t dma_addr);
++
++static inline void *efrm_pci_alloc_consistent(struct pci_dev *hwdev,
++ size_t size,
++ dma_addr_t *dma_addr)
++{
++ return efrm_dma_alloc_coherent(&hwdev->dev, size, dma_addr,
++ GFP_ATOMIC);
++}
++
++static inline void efrm_pci_free_consistent(struct pci_dev *hwdev, size_t size,
++ void *ptr, dma_addr_t dma_addr)
++{
++ efrm_dma_free_coherent(&hwdev->dev, size, ptr, dma_addr);
++}
++
++#endif /* DRIVER_LINUX_RESOURCE_KERNEL_COMPAT_H */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/kernel_proc.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/kernel_proc.c 2008-05-19 00:33:29.429842497 +0300
+@@ -0,0 +1,111 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains /proc/driver/sfc_resource/ implementation.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/debug.h>
++#include <ci/efrm/nic_table.h>
++#include <ci/efrm/driver_private.h>
++#include <linux/proc_fs.h>
++
++/** Top level directory for sfc specific stats **/
++static struct proc_dir_entry *efrm_proc_root; /* = NULL */
++
++static int
++efrm_resource_read_proc(char *buf, char **start, off_t offset, int count,
++ int *eof, void *data);
++
++int efrm_install_proc_entries(void)
++{
++ /* create the top-level directory for etherfabric specific stuff */
++ efrm_proc_root = proc_mkdir("sfc_resource", proc_root_driver);
++ if (!efrm_proc_root)
++ return -ENOMEM;
++ EFRM_ASSERT(efrm_proc_root);
++
++ if (create_proc_read_entry("resources", 0, efrm_proc_root,
++ efrm_resource_read_proc, 0) == NULL) {
++ EFRM_WARN("%s: Unable to create /proc/drivers/sfc_resource/"
++ "resources", __FUNCTION__);
++ }
++ return 0;
++}
++
++void efrm_uninstall_proc_entries(void)
++{
++ EFRM_ASSERT(efrm_proc_root);
++ remove_proc_entry("resources", efrm_proc_root);
++ remove_proc_entry("sfc_resource", proc_root_driver);
++ efrm_proc_root = NULL;
++}
++
++/****************************************************************************
++ *
++ * /proc/drivers/sfc/resources
++ *
++ ****************************************************************************/
++
++#define EFRM_PROC_PRINTF(buf, len, fmt, ...) \
++ do { \
++ if (count - len > 0) \
++ len += snprintf(buf+len, count-len, (fmt), \
++ __VA_ARGS__); \
++ } while (0)
++
++static int
++efrm_resource_read_proc(char *buf, char **start, off_t offset, int count,
++ int *eof, void *data)
++{
++ irq_flags_t lock_flags;
++ int len = 0;
++ int type;
++ struct efrm_resource_manager *rm;
++
++ for (type = 0; type < EFRM_RESOURCE_NUM; type++) {
++ rm = efrm_rm_table[type];
++ if (rm == NULL)
++ continue;
++
++ EFRM_PROC_PRINTF(buf, len, "*** %s ***\n", rm->rm_name);
++
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++ EFRM_PROC_PRINTF(buf, len, "current = %u\n", rm->rm_resources);
++ EFRM_PROC_PRINTF(buf, len, " max = %u\n\n",
++ rm->rm_resources_hiwat);
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ }
++
++ return count ? strlen(buf) : 0;
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/kfifo.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/kfifo.c 2008-05-19 00:33:29.433842727 +0300
+@@ -0,0 +1,212 @@
++/*
++ * A simple kernel FIFO implementation.
++ *
++ * Copyright (C) 2004 Stelian Pop <stelian@popies.net>
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
++ *
++ */
++
++/*
++ * This file is stolen from the Linux kernel sources
++ * (linux-2.6.22/kernel/kfifo.c) into sfc_resource driver.
++ * It should be used for old kernels without kfifo implementation.
++ * Most part of linux/kfifo.h is incorporated into
++ * ci/efrm/sysdep_linux.h.
++ */
++#include <ci/efrm/sysdep_linux.h>
++#ifdef HAS_NO_KFIFO
++
++#include <linux/kernel.h>
++#include <linux/module.h>
++#include <linux/slab.h>
++#include <linux/err.h>
++/*#include <linux/kfifo.h>*/
++
++/**
++ * kfifo_init - allocates a new FIFO using a preallocated buffer
++ * @buffer: the preallocated buffer to be used.
++ * @size: the size of the internal buffer, this have to be a power of 2.
++ * @gfp_mask: get_free_pages mask, passed to kmalloc()
++ * @lock: the lock to be used to protect the fifo buffer
++ *
++ * Do NOT pass the kfifo to kfifo_free() after use! Simply free the
++ * &struct kfifo with kfree().
++ */
++struct kfifo *kfifo_init(unsigned char *buffer, unsigned int size,
++ gfp_t gfp_mask, spinlock_t * lock)
++{
++ struct kfifo *fifo;
++
++ /* size must be a power of 2 */
++ BUG_ON(size & (size - 1));
++
++ fifo = kmalloc(sizeof(struct kfifo), gfp_mask);
++ if (!fifo)
++ return ERR_PTR(-ENOMEM);
++
++ fifo->buffer = buffer;
++ fifo->size = size;
++ fifo->in = fifo->out = 0;
++ fifo->lock = lock;
++
++ return fifo;
++}
++
++EXPORT_SYMBOL(kfifo_init);
++
++/**
++ * kfifo_alloc - allocates a new FIFO and its internal buffer
++ * @size: the size of the internal buffer to be allocated.
++ * @gfp_mask: get_free_pages mask, passed to kmalloc()
++ * @lock: the lock to be used to protect the fifo buffer
++ *
++ * The size will be rounded-up to a power of 2.
++ */
++struct kfifo *kfifo_alloc(unsigned int size, gfp_t gfp_mask, spinlock_t * lock)
++{
++ unsigned char *buffer;
++ struct kfifo *ret;
++
++ /*
++ * round up to the next power of 2, since our 'let the indices
++ * wrap' tachnique works only in this case.
++ */
++ if (size & (size - 1)) {
++ BUG_ON(size > 0x80000000);
++ size = roundup_pow_of_two(size);
++ }
++
++ buffer = kmalloc(size, gfp_mask);
++ if (!buffer)
++ return ERR_PTR(-ENOMEM);
++
++ ret = kfifo_init(buffer, size, gfp_mask, lock);
++
++ if (IS_ERR(ret))
++ kfree(buffer);
++
++ return ret;
++}
++
++EXPORT_SYMBOL(kfifo_alloc);
++
++/**
++ * kfifo_free - frees the FIFO
++ * @fifo: the fifo to be freed.
++ */
++void kfifo_free(struct kfifo *fifo)
++{
++ kfree(fifo->buffer);
++ kfree(fifo);
++}
++
++EXPORT_SYMBOL(kfifo_free);
++
++/**
++ * __kfifo_put - puts some data into the FIFO, no locking version
++ * @fifo: the fifo to be used.
++ * @buffer: the data to be added.
++ * @len: the length of the data to be added.
++ *
++ * This function copies at most @len bytes from the @buffer into
++ * the FIFO depending on the free space, and returns the number of
++ * bytes copied.
++ *
++ * Note that with only one concurrent reader and one concurrent
++ * writer, you don't need extra locking to use these functions.
++ */
++unsigned int
++__kfifo_put(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
++{
++ unsigned int l;
++
++ len = min(len, fifo->size - fifo->in + fifo->out);
++
++ /*
++ * Ensure that we sample the fifo->out index -before- we
++ * start putting bytes into the kfifo.
++ */
++
++ smp_mb();
++
++ /* first put the data starting from fifo->in to buffer end */
++ l = min(len, fifo->size - (fifo->in & (fifo->size - 1)));
++ memcpy(fifo->buffer + (fifo->in & (fifo->size - 1)), buffer, l);
++
++ /* then put the rest (if any) at the beginning of the buffer */
++ memcpy(fifo->buffer, buffer + l, len - l);
++
++ /*
++ * Ensure that we add the bytes to the kfifo -before-
++ * we update the fifo->in index.
++ */
++
++ smp_wmb();
++
++ fifo->in += len;
++
++ return len;
++}
++
++EXPORT_SYMBOL(__kfifo_put);
++
++/**
++ * __kfifo_get - gets some data from the FIFO, no locking version
++ * @fifo: the fifo to be used.
++ * @buffer: where the data must be copied.
++ * @len: the size of the destination buffer.
++ *
++ * This function copies at most @len bytes from the FIFO into the
++ * @buffer and returns the number of copied bytes.
++ *
++ * Note that with only one concurrent reader and one concurrent
++ * writer, you don't need extra locking to use these functions.
++ */
++unsigned int
++__kfifo_get(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
++{
++ unsigned int l;
++
++ len = min(len, fifo->in - fifo->out);
++
++ /*
++ * Ensure that we sample the fifo->in index -before- we
++ * start removing bytes from the kfifo.
++ */
++
++ smp_rmb();
++
++ /* first get the data from fifo->out until the end of the buffer */
++ l = min(len, fifo->size - (fifo->out & (fifo->size - 1)));
++ memcpy(buffer, fifo->buffer + (fifo->out & (fifo->size - 1)), l);
++
++ /* then get the rest (if any) from the beginning of the buffer */
++ memcpy(buffer + l, fifo->buffer, len - l);
++
++ /*
++ * Ensure that we remove the bytes from the kfifo -before-
++ * we update the fifo->out index.
++ */
++
++ smp_mb();
++
++ fifo->out += len;
++
++ return len;
++}
++
++EXPORT_SYMBOL(__kfifo_get);
++#endif
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/linux_resource_internal.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/linux_resource_internal.h 2008-05-19 00:33:29.433842727 +0300
+@@ -0,0 +1,75 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains Linux-specific API internal for the resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __LINUX_RESOURCE_INTERNAL__
++#define __LINUX_RESOURCE_INTERNAL__
++
++#include <ci/driver/resource/linux_efhw_nic.h>
++#include <ci/efrm/debug.h>
++#include <ci/efrm/driver_private.h>
++#include <ci/driver/efab/hardware.h>
++
++
++/*! Linux specific EtherFabric initialisation */
++extern int
++linux_efrm_nic_ctor(struct linux_efhw_nic *, struct pci_dev *,
++ spinlock_t *reg_lock,
++ unsigned nic_flags, unsigned nic_options);
++
++/*! Linux specific EtherFabric initialisation */
++extern void linux_efrm_nic_dtor(struct linux_efhw_nic *);
++
++/*! Linux specific EtherFabric initialisation -- interrupt registration */
++extern int linux_efrm_irq_ctor(struct linux_efhw_nic *);
++
++/*! Linux specific EtherFabric initialisation -- interrupt deregistration */
++extern void linux_efrm_irq_dtor(struct linux_efhw_nic *);
++
++extern int efrm_driverlink_register(void);
++extern void efrm_driverlink_unregister(void);
++
++extern int
++efrm_nic_add(struct pci_dev *dev, unsigned int opts, const uint8_t *mac_addr,
++ struct linux_efhw_nic **lnic_out, spinlock_t *reg_lock,
++ int bt_min, int bt_max, const struct vi_resource_dimensions *);
++extern void efrm_nic_del(struct linux_efhw_nic *);
++
++
++extern int efrm_install_proc_entries(void);
++extern void efrm_uninstall_proc_entries(void);
++
++#endif /* __LINUX_RESOURCE_INTERNAL__ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/nic.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/nic.c 2008-05-19 00:33:29.433842727 +0300
+@@ -0,0 +1,190 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains EtherFabric Generic NIC instance (init, interrupts,
++ * etc)
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efhw/debug.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efhw/nic.h>
++#include <ci/efhw/eventq.h>
++
++
++int efhw_device_type_init(struct efhw_device_type *dt,
++ int vendor_id, int device_id,
++ int class_revision)
++{
++ if (vendor_id != 0x1924)
++ return 0;
++
++ switch (device_id) {
++ case 0x0703:
++ case 0x6703:
++ dt->arch = EFHW_ARCH_FALCON;
++ dt->variant = 'A';
++ switch (class_revision) {
++ case 0:
++ dt->revision = 0;
++ break;
++ case 1:
++ dt->revision = 1;
++ break;
++ default:
++ return 0;
++ }
++ break;
++ case 0x0710:
++ dt->arch = EFHW_ARCH_FALCON;
++ dt->variant = 'B';
++ switch (class_revision) {
++ case 2:
++ dt->revision = 0;
++ break;
++ default:
++ return 0;
++ }
++ break;
++ default:
++ return 0;
++ }
++
++ return 1;
++}
++
++
++/*--------------------------------------------------------------------
++ *
++ * NIC Initialisation
++ *
++ *--------------------------------------------------------------------*/
++
++/* make this separate from initialising data structure
++** to allow this to be called at a later time once we can access PCI
++** config space to find out what hardware we have
++*/
++void efhw_nic_init(struct efhw_nic *nic, unsigned flags, unsigned options,
++ struct efhw_device_type dev_type)
++{
++ int i;
++
++ nic->devtype = dev_type;
++ nic->flags = flags;
++ nic->options = options;
++ nic->bar_ioaddr = 0;
++ spin_lock_init(&nic->the_reg_lock);
++ nic->reg_lock = &nic->the_reg_lock;
++ nic->mtu = 1500 + ETH_HLEN;
++
++ for (i = 0; i < EFHW_KEVENTQ_MAX; i++)
++ nic->irq_unit[i] = EFHW_IRQ_UNIT_UNUSED;
++
++ switch (nic->devtype.arch) {
++ case EFHW_ARCH_FALCON:
++ nic->evq_sizes = 512 | 1024 | 2048 | 4096 | 8192 |
++ 16384 | 32768;
++ nic->txq_sizes = 512 | 1024 | 2048 | 4096;
++ nic->rxq_sizes = 512 | 1024 | 2048 | 4096;
++ nic->efhw_func = &falcon_char_functional_units;
++ nic->ctr_ap_bytes = EFHW_64M;
++ switch (nic->devtype.variant) {
++ case 'A':
++ nic->ctr_ap_bar = FALCON_S_CTR_AP_BAR;
++ break;
++ case 'B':
++ nic->flags |= NIC_FLAG_NO_INTERRUPT;
++ nic->ctr_ap_bar = FALCON_P_CTR_AP_BAR;
++ break;
++ default:
++ EFHW_ASSERT(0);
++ break;
++ }
++ break;
++ default:
++ EFHW_ASSERT(0);
++ break;
++ }
++}
++
++
++void efhw_nic_close_interrupts(struct efhw_nic *nic)
++{
++ int i;
++
++ EFHW_ASSERT(nic);
++ if (!efhw_nic_have_hw(nic))
++ return;
++
++ EFHW_ASSERT(efhw_nic_have_hw(nic));
++
++ for (i = 0; i < EFHW_KEVENTQ_MAX; i++) {
++ if (nic->irq_unit[i] != EFHW_IRQ_UNIT_UNUSED)
++ efhw_nic_interrupt_disable(nic, i);
++ }
++}
++
++void efhw_nic_dtor(struct efhw_nic *nic)
++{
++ EFHW_ASSERT(nic);
++
++ /* Check that we have functional units because the software only
++ * driver doesn't initialise anything hardware related any more */
++
++#ifndef __ci_ul_driver__
++ /* close interrupts is called first because the act of deregistering
++ the driver could cause this driver to change from master to slave
++ and hence the implicit interrupt mappings would be wrong */
++
++ EFHW_TRACE("%s: functional units ... ", __FUNCTION__);
++
++ if (efhw_nic_have_functional_units(nic)) {
++ efhw_nic_close_interrupts(nic);
++ efhw_nic_close_hardware(nic);
++ }
++ EFHW_TRACE("%s: functional units ... done", __FUNCTION__);
++#endif
++
++ /* destroy event queues */
++ EFHW_TRACE("%s: event queues ... ", __FUNCTION__);
++
++#ifndef __ci_ul_driver__
++ {
++ int i;
++ for (i = 0; i < EFHW_KEVENTQ_MAX; ++i)
++ if (nic->evq[i].evq_mask)
++ efhw_keventq_dtor(nic, &nic->evq[i]);
++ }
++#endif
++
++ EFHW_TRACE("%s: event queues ... done", __FUNCTION__);
++
++ spin_lock_destroy(&nic->the_reg_lock);
++
++ EFHW_TRACE("%s: DONE", __FUNCTION__);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/resource_driver.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/resource_driver.c 2008-05-19 00:33:29.433842727 +0300
+@@ -0,0 +1,640 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains main driver entry points.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "linux_resource_internal.h"
++#include "kernel_compat.h"
++#include <ci/efrm/nic_table.h>
++#include <ci/driver/resource/efx_vi.h>
++#include <ci/efhw/eventq.h>
++#include <ci/efhw/nic.h>
++#include <ci/efrm/buffer_table.h>
++#include <ci/efrm/vi_resource_private.h>
++#include <ci/efrm/driver_private.h>
++
++#if EFRM_HAVE_IOMMU_LOCK
++#ifdef EFRM_NEED_ALTERNATE_MAX_PFN
++extern unsigned long blk_max_pfn;
++#define max_pfn blk_max_pfn
++#else
++#include <linux/bootmem.h>
++#endif
++#endif
++
++MODULE_AUTHOR("Solarflare Communications");
++MODULE_LICENSE("GPL");
++
++static struct efhw_ev_handler ev_handler = {
++ .wakeup_fn = efrm_handle_wakeup_event,
++ .timeout_fn = efrm_handle_timeout_event,
++ .dmaq_flushed_fn = efrm_handle_dmaq_flushed,
++};
++
++#if EFRM_HAVE_IOMMU_LOCK
++int efrm_need_iommu_lock;
++EXPORT_SYMBOL(efrm_need_iommu_lock);
++#endif
++
++const int max_hardware_init_repeats = 10;
++
++/*--------------------------------------------------------------------
++ *
++ * Module load time variables
++ *
++ *--------------------------------------------------------------------*/
++/* See docs/notes/pci_alloc_consistent */
++int use_pci_alloc = 1; /* Use pci_alloc_consistent to alloc iopages */
++static int do_irq = 1; /* enable interrupts */
++
++#if defined(CONFIG_X86_XEN)
++static int irq_moderation = 60; /* interrupt moderation (60 usec) */
++#else
++static int irq_moderation = 20; /* interrupt moderation (20 usec) */
++#endif
++static int nic_options = NIC_OPT_DEFAULT;
++int efx_vi_eventq_size = EFX_VI_EVENTQ_SIZE_DEFAULT;
++
++module_param(do_irq, int, S_IRUGO);
++MODULE_PARM_DESC(do_irq, "Enable interrupts. "
++ "Do not turn it off unless you know what are you doing.");
++module_param(irq_moderation, int, S_IRUGO);
++MODULE_PARM_DESC(irq_moderation, "IRQ moderation in usec");
++module_param(nic_options, int, S_IRUGO);
++MODULE_PARM_DESC(nic_options, "Nic options -- see efhw_types.h");
++module_param(use_pci_alloc, int, S_IRUGO);
++MODULE_PARM_DESC(use_pci_alloc, "Use pci_alloc_consistent to alloc iopages "
++ "(autodetected by kernel version)");
++module_param(efx_vi_eventq_size, int, S_IRUGO);
++MODULE_PARM_DESC(efx_vi_eventq_size,
++ "Size of event queue allocated by efx_vi library");
++
++/*--------------------------------------------------------------------
++ *
++ * Linux specific NIC initialisation
++ *
++ *--------------------------------------------------------------------*/
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
++# define IRQ_PT_REGS_ARG , struct pt_regs *regs __attribute__ ((unused))
++#else
++# define IRQ_PT_REGS_ARG
++#endif
++
++#ifndef IRQF_SHARED
++# define IRQF_SHARED SA_SHIRQ
++#endif
++
++static inline irqreturn_t
++linux_efrm_interrupt(int irr, void *dev_id IRQ_PT_REGS_ARG)
++{
++ return efhw_nic_interrupt((struct efhw_nic *)dev_id);
++}
++
++int linux_efrm_irq_ctor(struct linux_efhw_nic *lnic)
++{
++ struct efhw_nic *nic = &lnic->nic;
++
++ nic->flags &= ~NIC_FLAG_MSI;
++ if (nic->flags & NIC_FLAG_TRY_MSI) {
++ int rc = pci_enable_msi(lnic->pci_dev);
++ if (rc < 0) {
++ EFRM_WARN("%s: Could not enable MSI (%d)",
++ __FUNCTION__, rc);
++ EFRM_WARN("%s: Continuing with legacy interrupt mode",
++ __FUNCTION__);
++ } else {
++ EFRM_NOTICE("%s: MSI enabled", __FUNCTION__);
++ nic->flags |= NIC_FLAG_MSI;
++ }
++ }
++
++ if (request_irq(lnic->pci_dev->irq, linux_efrm_interrupt,
++ IRQF_SHARED, "sfc_resource", nic)) {
++ EFRM_ERR("Request for interrupt #%d failed",
++ lnic->pci_dev->irq);
++ nic->flags &= ~NIC_FLAG_OS_IRQ_EN;
++ return -EBUSY;
++ }
++ nic->flags |= NIC_FLAG_OS_IRQ_EN;
++
++ return 0;
++}
++
++void linux_efrm_irq_dtor(struct linux_efhw_nic *lnic)
++{
++ EFRM_TRACE("linux_efrm_irq_dtor: start");
++
++ if (lnic->nic.flags & NIC_FLAG_OS_IRQ_EN) {
++ free_irq(lnic->pci_dev->irq, &lnic->nic);
++ lnic->nic.flags &= ~NIC_FLAG_OS_IRQ_EN;
++ }
++
++ if (lnic->nic.flags & NIC_FLAG_MSI) {
++ pci_disable_msi(lnic->pci_dev);
++ lnic->nic.flags &= ~NIC_FLAG_MSI;
++ }
++
++ EFRM_TRACE("linux_efrm_irq_dtor: done");
++}
++
++/* Allocate buffer table entries for a particular NIC.
++ */
++static int efrm_nic_buffer_table_alloc(struct efhw_nic *nic)
++{
++ int capacity;
++ int page_order;
++ int i;
++ int rc;
++
++ /* Choose queue size. */
++ for (capacity = 8192; capacity <= nic->evq_sizes; capacity <<= 1) {
++ if (capacity > nic->evq_sizes) {
++ EFRM_ERR
++ ("%s: Unable to choose EVQ size (supported=%x)",
++ __FUNCTION__, nic->evq_sizes);
++ return -E2BIG;
++ } else if (capacity & nic->evq_sizes)
++ break;
++ }
++ for (i = 0; i < EFHW_KEVENTQ_MAX; ++i) {
++ nic->evq[i].hw.capacity = capacity;
++ nic->evq[i].hw.buf_tbl_alloc.base = (unsigned)-1;
++ }
++
++ /* allocate buffer table entries to map onto the iobuffer */
++ page_order = get_order(capacity * sizeof(efhw_event_t));
++ if (!(nic->flags & NIC_FLAG_NO_INTERRUPT)) {
++ rc = efrm_buffer_table_alloc(page_order,
++ &nic->evq[0].hw.buf_tbl_alloc);
++ if (rc < 0) {
++ EFRM_WARN
++ ("%s: failed (%d) to alloc %d buffer table entries",
++ __FUNCTION__, rc, page_order);
++ return rc;
++ }
++ }
++ rc = efrm_buffer_table_alloc(page_order,
++ &nic->evq[FALCON_EVQ_NONIRQ].hw.
++ buf_tbl_alloc);
++ if (rc < 0) {
++ EFRM_WARN
++ ("%s: failed (%d) to alloc %d buffer table entries",
++ __FUNCTION__, rc, page_order);
++ return rc;
++ }
++
++ return 0;
++}
++
++/* Free buffer table entries allocated for a particular NIC.
++ */
++static void efrm_nic_buffer_table_free(struct efhw_nic *nic)
++{
++ int i;
++ for (i = 0; i <= FALCON_EVQ_NONIRQ; i++)
++ if (nic->evq[i].hw.buf_tbl_alloc.base != (unsigned)-1)
++ efrm_buffer_table_free(&nic->evq[i].hw.buf_tbl_alloc);
++
++}
++
++static int iomap_bar(struct linux_efhw_nic *lnic, size_t len)
++{
++ efhw_ioaddr_t ioaddr;
++
++ ioaddr = ioremap_nocache(lnic->ctr_ap_pci_addr, len);
++ if (ioaddr == 0)
++ return -ENOMEM;
++
++ lnic->nic.bar_ioaddr = ioaddr;
++ return 0;
++}
++
++static int linux_efhw_nic_map_ctr_ap(struct linux_efhw_nic *lnic)
++{
++ struct efhw_nic *nic = &lnic->nic;
++ int rc;
++
++ rc = iomap_bar(lnic, nic->ctr_ap_bytes);
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ /* Bug 5195: workaround for now. */
++ if (rc != 0 && nic->ctr_ap_bytes > 16 * 1024 * 1024) {
++ /* Try half the size for now. */
++ nic->ctr_ap_bytes /= 2;
++ EFRM_WARN("Bug 5195 WORKAROUND: retrying iomap of %d bytes",
++ nic->ctr_ap_bytes);
++ rc = iomap_bar(lnic, nic->ctr_ap_bytes);
++ }
++#endif
++
++ if (rc < 0) {
++ EFRM_ERR("Failed (%d) to map bar (%d bytes)",
++ rc, nic->ctr_ap_bytes);
++ return rc;
++ }
++
++ return rc;
++}
++
++int
++linux_efrm_nic_ctor(struct linux_efhw_nic *lnic, struct pci_dev *dev,
++ spinlock_t *reg_lock,
++ unsigned nic_flags, unsigned nic_options)
++{
++ struct efhw_device_type dev_type;
++ struct efhw_nic *nic = &lnic->nic;
++ u8 class_revision;
++ int rc;
++
++ rc = pci_read_config_byte(dev, PCI_CLASS_REVISION, &class_revision);
++ if (rc != 0) {
++ EFRM_ERR("%s: pci_read_config_byte failed (%d)",
++ __FUNCTION__, rc);
++ return rc;
++ }
++
++ if (!efhw_device_type_init(&dev_type, dev->vendor, dev->device,
++ class_revision)) {
++ EFRM_ERR("%s: efhw_device_type_init failed %04x:%04x(%d)",
++ __FUNCTION__, (unsigned) dev->vendor,
++ (unsigned) dev->device, (int) class_revision);
++ return -ENODEV;
++ }
++
++ EFRM_NOTICE("attaching device type %04x:%04x %d:%c%d",
++ (unsigned) dev->vendor, (unsigned) dev->device,
++ dev_type.arch, dev_type.variant, dev_type.revision);
++
++ /* Initialise the adapter-structure. */
++ efhw_nic_init(nic, nic_flags, nic_options, dev_type);
++ lnic->pci_dev = dev;
++
++ rc = pci_enable_device(dev);
++ if (rc < 0) {
++ EFRM_ERR("%s: pci_enable_device failed (%d)",
++ __FUNCTION__, rc);
++ return rc;
++ }
++
++ lnic->ctr_ap_pci_addr = pci_resource_start(dev, nic->ctr_ap_bar);
++
++ if (!pci_dma_supported(dev, (dma_addr_t)EFHW_DMA_ADDRMASK)) {
++ EFRM_ERR("%s: pci_dma_supported(%lx) failed", __FUNCTION__,
++ (unsigned long)EFHW_DMA_ADDRMASK);
++ return -ENODEV;
++ }
++
++ if (pci_set_dma_mask(dev, (dma_addr_t)EFHW_DMA_ADDRMASK)) {
++ EFRM_ERR("%s: pci_set_dma_mask(%lx) failed", __FUNCTION__,
++ (unsigned long)EFHW_DMA_ADDRMASK);
++ return -ENODEV;
++ }
++
++ if (pci_set_consistent_dma_mask(dev, (dma_addr_t)EFHW_DMA_ADDRMASK)) {
++ EFRM_ERR("%s: pci_set_consistent_dma_mask(%lx) failed",
++ __FUNCTION__, (unsigned long)EFHW_DMA_ADDRMASK);
++ return -ENODEV;
++ }
++
++ rc = linux_efhw_nic_map_ctr_ap(lnic);
++ if (rc < 0)
++ return rc;
++
++ /* By default struct efhw_nic contains its own lock for protecting
++ * access to nic registers. We override it with a pointer to the
++ * lock in the net driver. This is needed when resource and net
++ * drivers share a single PCI function (falcon B series).
++ */
++ nic->reg_lock = reg_lock;
++ return 0;
++}
++
++void linux_efrm_nic_dtor(struct linux_efhw_nic *lnic)
++{
++ struct efhw_nic *nic = &lnic->nic;
++ efhw_ioaddr_t bar_ioaddr = nic->bar_ioaddr;
++
++ efhw_nic_dtor(nic);
++
++ efrm_nic_buffer_table_free(nic);
++
++ /* Unmap the bar. */
++ EFRM_ASSERT(bar_ioaddr);
++ iounmap(bar_ioaddr);
++ nic->bar_ioaddr = 0;
++}
++
++/****************************************************************************
++ *
++ * efrm_tasklet - used to poll the eventq which may result in further callbacks
++ *
++ ****************************************************************************/
++
++static void efrm_tasklet(unsigned long pdev)
++{
++ struct efhw_nic *nic = (struct efhw_nic *)pdev;
++
++ EFRM_ASSERT(!(nic->flags & NIC_FLAG_NO_INTERRUPT));
++
++ efhw_keventq_poll(nic, &nic->evq[0]);
++ EFRM_TRACE("tasklet complete");
++}
++
++/****************************************************************************
++ *
++ * char driver specific interrupt callbacks -- run at hard IRQL
++ *
++ ****************************************************************************/
++static void efrm_handle_eventq_irq(struct efhw_nic *nic, int evq)
++{
++ /* NB. The interrupt must have already been acked (for legacy mode). */
++
++ EFRM_TRACE("%s: starting tasklet", __FUNCTION__);
++ EFRM_ASSERT(!(nic->flags & NIC_FLAG_NO_INTERRUPT));
++
++ tasklet_schedule(&linux_efhw_nic(nic)->tasklet);
++}
++
++/* A count of how many NICs this driver knows about. */
++static int n_nics_probed;
++
++/****************************************************************************
++ *
++ * efrm_nic_add: add the NIC to the resource driver
++ *
++ * NOTE: the flow of control through this routine is quite subtle
++ * because of the number of operations that can fail. We therefore
++ * take the apporaching of keeping the return code (rc) variable
++ * accurate, and only do operations while it is non-negative. Tear down
++ * is done at the end if rc is negative, depending on what has been set up
++ * by that point.
++ *
++ * So basically just make sure that any code you add checks rc>=0 before
++ * doing any work and you'll be fine.
++ *
++ ****************************************************************************/
++int
++efrm_nic_add(struct pci_dev *dev, unsigned flags, const uint8_t *mac_addr,
++ struct linux_efhw_nic **lnic_out, spinlock_t *reg_lock,
++ int bt_min, int bt_max,
++ const struct vi_resource_dimensions *res_dim)
++{
++ struct linux_efhw_nic *lnic = NULL;
++ struct efhw_nic *nic = NULL;
++ int count = 0, rc = 0, resources_init = 0;
++ int constructed = 0;
++ int registered_nic = 0;
++ int buffers_allocated = 0;
++ static unsigned nic_index; /* = 0; */
++
++ EFRM_TRACE("%s: device detected (Slot '%s', IRQ %d)", __FUNCTION__,
++ pci_name(dev) ? pci_name(dev) : "?", dev->irq);
++
++ /* Ensure that we have room for the new adapter-structure. */
++ if (efrm_nic_table.nic_count == EFHW_MAX_NR_DEVS) {
++ EFRM_WARN("%s: WARNING: too many devices", __FUNCTION__);
++ rc = -ENOMEM;
++ goto failed;
++ }
++
++ if (n_nics_probed == 0) {
++ rc = efrm_resources_init(res_dim, bt_min, bt_max);
++ if (rc != 0)
++ goto failed;
++ resources_init = 1;
++ }
++
++ /* Allocate memory for the new adapter-structure. */
++ lnic = kmalloc(sizeof(*lnic), GFP_KERNEL);
++ if (lnic == NULL) {
++ EFRM_ERR("%s: ERROR: failed to allocate memory", __FUNCTION__);
++ rc = -ENOMEM;
++ goto failed;
++ }
++ memset(lnic, 0, sizeof(*lnic));
++ nic = &lnic->nic;
++
++ lnic->ev_handlers = &ev_handler;
++
++ /* OS specific hardware mappings */
++ rc = linux_efrm_nic_ctor(lnic, dev, reg_lock, flags, nic_options);
++ if (rc < 0) {
++ EFRM_ERR("%s: ERROR: initialisation failed", __FUNCTION__);
++ goto failed;
++ }
++
++ constructed = 1;
++
++ /* Tell the driver about the NIC - this needs to be done before the
++ resources managers get created below. Note we haven't initialised
++ the hardware yet, and I don't like doing this before the perhaps
++ unreliable hardware initialisation. However, there's quite a lot
++ of code to review if we wanted to hardware init before bringing
++ up the resource managers. */
++ rc = efrm_driver_register_nic(nic, nic_index++);
++ if (rc < 0) {
++ EFRM_ERR("%s: cannot register nic %d with nic error code %d",
++ __FUNCTION__, efrm_nic_table.nic_count, rc);
++ goto failed;
++ }
++ registered_nic = 1;
++
++ rc = efrm_nic_buffer_table_alloc(nic);
++ if (rc < 0)
++ goto failed;
++ buffers_allocated = 1;
++
++ /****************************************************/
++ /* hardware bringup */
++ /****************************************************/
++ /* Detecting hardware can be a slightly unreliable process;
++ we want to make sure that we maximise our chances, so we
++ loop a few times until all is good. */
++ for (count = 0; count < max_hardware_init_repeats; count++) {
++ rc = efhw_nic_init_hardware(nic, &ev_handler, mac_addr);
++ if (rc >= 0)
++ break;
++
++ /* pain */
++ EFRM_ERR
++ ("error - hardware initialisation failed code %d, "
++ "attempt %d of %d", rc, count + 1,
++ max_hardware_init_repeats);
++ }
++ if (rc < 0)
++ goto failed;
++
++ tasklet_init(&lnic->tasklet, efrm_tasklet, (ulong)nic);
++
++ /* set up interrupt handlers (hard-irq) */
++ nic->irq_handler = &efrm_handle_eventq_irq;
++
++ /* this device can now take management interrupts */
++ if (do_irq && !(nic->flags & NIC_FLAG_NO_INTERRUPT)) {
++ rc = linux_efrm_irq_ctor(lnic);
++ if (rc < 0) {
++ EFRM_ERR("Interrupt initialisation failed (%d)", rc);
++ goto failed;
++ }
++ efhw_nic_set_interrupt_moderation(nic, 0, irq_moderation);
++ efhw_nic_interrupt_enable(nic, 0);
++ }
++ EFRM_TRACE("interrupts are %sregistered", do_irq ? "" : "not ");
++
++#if EFRM_HAVE_IOMMU_LOCK
++ /* Bug 4560: We need the lock if there is memory which cannot be
++ * accessed by the card and there is an IOMMU to access it. In that
++ * case, the kernel will use the IOMMU to access the high memory. */
++ if ((dev->dma_mask >> PAGE_SHIFT) < max_pfn && !EFRM_NO_IOMMU)
++ efrm_need_iommu_lock = 1;
++#endif
++
++ *lnic_out = lnic;
++ EFRM_ASSERT(rc == 0);
++ ++n_nics_probed;
++ return 0;
++
++failed:
++ if (buffers_allocated)
++ efrm_nic_buffer_table_free(nic);
++ if (registered_nic)
++ efrm_driver_unregister_nic(nic);
++ if (constructed)
++ linux_efrm_nic_dtor(lnic);
++ kfree(lnic); /* safe in any case */
++ if (resources_init)
++ efrm_resources_fini();
++ return rc;
++}
++
++/****************************************************************************
++ *
++ * efrm_nic_del: Remove the nic from the resource driver structures
++ *
++ ****************************************************************************/
++void efrm_nic_del(struct linux_efhw_nic *lnic)
++{
++ struct efhw_nic *nic = &lnic->nic;
++
++ EFRM_TRACE("%s:", __FUNCTION__);
++ EFRM_ASSERT(nic);
++
++ efrm_driver_unregister_nic(nic);
++
++ /*
++ * Synchronise here with any running ISR.
++ * Remove the OS handler. There should be no IRQs being generated
++ * by our NIC at this point.
++ */
++ if (efhw_nic_have_functional_units(nic)) {
++ efhw_nic_close_interrupts(nic);
++ linux_efrm_irq_dtor(lnic);
++ tasklet_kill(&lnic->tasklet);
++ }
++
++ /* Close down hardware and free resources. */
++ linux_efrm_nic_dtor(lnic);
++ kfree(lnic);
++
++ if (--n_nics_probed == 0)
++ efrm_resources_fini();
++
++ EFRM_TRACE("NIC teardown: Done");
++}
++
++/****************************************************************************
++ *
++ * init_module: register as a PCI driver.
++ *
++ ****************************************************************************/
++static int init_sfc_resource(void)
++{
++ int rc = 0;
++
++ EFRM_TRACE("%s: RESOURCE driver starting", __FUNCTION__);
++
++ rc = efrm_driver_ctor();
++ if (rc < 0) {
++ EFRM_ERR("%s: efrm_driver_ctor: error %d", __FUNCTION__, rc);
++ goto fail_driver_ctor;
++ }
++
++ /* Register the driver so that our 'probe' function is called for
++ * each EtherFabric device in the system.
++ */
++ rc = efrm_driverlink_register();
++ if (rc == -ENODEV)
++ EFRM_ERR("%s: no devices found", __FUNCTION__);
++ if (rc < 0)
++ goto failed_driverlink;
++
++ if (efrm_install_proc_entries() != 0) {
++ /* Do not fail, but print a warning */
++ EFRM_WARN("%s: WARNING: failed to install /proc entries",
++ __FUNCTION__);
++ }
++
++ return 0;
++
++failed_driverlink:
++ /* No need to release resource managers here since they register
++ * destructors with the driver. */
++ efrm_driver_dtor();
++fail_driver_ctor:
++ EFRM_ASSERT(rc != 0);
++ return rc;
++}
++
++/****************************************************************************
++ *
++ * cleanup_module: module-removal entry-point
++ *
++ ****************************************************************************/
++static void cleanup_sfc_resource(void)
++{
++ efrm_uninstall_proc_entries();
++
++ efrm_driverlink_unregister();
++
++ /* Clean up char-driver specific initialisation.
++ - driver dtor can use both work queue and buffer table entries */
++ efrm_driver_dtor();
++
++ EFRM_TRACE("unloaded");
++}
++
++module_init(init_sfc_resource);
++module_exit(cleanup_sfc_resource);
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/resource_manager.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/resource_manager.c 2008-05-19 00:33:29.445843419 +0300
+@@ -0,0 +1,263 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains generic code for resources and resource managers.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/debug.h>
++#include <ci/efrm/nic_table.h>
++#include <ci/efhw/iopage.h>
++#include <ci/efrm/driver_private.h>
++
++/**********************************************************************
++ * Internal stuff.
++ */
++
++#define EFRM_RM_TABLE_SIZE_INIT 256
++
++static int grow_table(struct efrm_resource_manager *rm, unsigned min_size)
++{
++ irq_flags_t lock_flags;
++ struct efrm_resource **table, **old_table;
++ unsigned new_size;
++
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm);
++
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++
++ /* Check whether the size of the table increased whilst the lock was
++ * dropped. */
++ if (min_size <= rm->rm_table_size) {
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ return 0;
++ }
++
++ new_size = rm->rm_table_size << 1;
++ if (new_size < min_size)
++ new_size = min_size;
++
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ if (in_atomic()) {
++ EFRM_WARN("%s: in_atomic in grow_table()", __FUNCTION__);
++ EFRM_WARN("%s: allocating %u bytes", __FUNCTION__,
++ (unsigned)(new_size *
++ sizeof(struct efrm_resource *)));
++ return -ENOMEM;
++ }
++
++ table =
++ (struct efrm_resource **)vmalloc(new_size *
++ sizeof(struct efrm_resource *));
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++
++ if (table == 0) {
++ EFRM_ERR("%s: out of memory in grow_table()", __FUNCTION__);
++ EFRM_ERR("%s: allocating %u bytes", __FUNCTION__,
++ (unsigned)(new_size *
++ sizeof(struct efrm_resource *)));
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ return -ENOMEM;
++ }
++
++ /* Could have got bigger while we dropped the lock... */
++ if (new_size <= rm->rm_table_size) {
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ vfree(table);
++ return 0;
++ }
++
++ memcpy(table, rm->rm_table, rm->rm_table_size * sizeof(*table));
++ memset(table + rm->rm_table_size, 0,
++ sizeof(*table) * (new_size - rm->rm_table_size));
++ /* remember old table so we can free the
++ memory after we drop the lock (bug 1040) */
++ old_table = rm->rm_table;
++ rm->rm_table = table;
++ rm->rm_table_size = new_size;
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ vfree(old_table);
++
++ return 0;
++}
++
++/**********************************************************************
++ * struct efrm_resource_manager
++ */
++
++void efrm_resource_manager_dtor(struct efrm_resource_manager *rm)
++{
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm);
++
++ /* call destructor */
++ EFRM_DO_DEBUG(if (rm->rm_resources)
++ EFRM_ERR("%s: %s leaked %d resources",
++ __FUNCTION__, rm->rm_name, rm->rm_resources));
++ EFRM_ASSERT(rm->rm_resources == 0);
++
++ rm->rm_dtor(rm);
++
++ /* clear out things built by efrm_resource_manager_ctor */
++ spin_lock_destroy(&rm->rm_lock);
++ vfree(rm->rm_table);
++
++ /* and the free the memory */
++ EFRM_DO_DEBUG(memset(rm, 0, sizeof(*rm)));
++ kfree(rm);
++}
++
++/* Construct a resource manager. Resource managers are singletons. */
++int
++efrm_resource_manager_ctor(struct efrm_resource_manager *rm,
++ void (*dtor)(struct efrm_resource_manager *),
++ const char *name, unsigned type,
++ int initial_table_size)
++{
++ EFRM_ASSERT(rm);
++ EFRM_ASSERT(dtor);
++
++ rm->rm_name = name;
++ EFRM_DO_DEBUG(rm->rm_type = type);
++ rm->rm_dtor = dtor;
++ spin_lock_init(&rm->rm_lock);
++ rm->rm_resources = 0;
++ rm->rm_resources_hiwat = 0;
++
++ /* if not set then pick a number */
++ rm->rm_table_size = (initial_table_size) ?
++ initial_table_size : EFRM_RM_TABLE_SIZE_INIT;
++
++ rm->rm_table = vmalloc(rm->rm_table_size *
++ sizeof(struct efrm_resource *));
++
++ if (rm->rm_table == 0) {
++ spin_lock_destroy(&rm->rm_lock);
++ return -ENOMEM;
++ }
++ memset(rm->rm_table, 0, sizeof(*rm->rm_table) * rm->rm_table_size);
++
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm);
++ return 0;
++}
++
++int efrm_resource_manager_insert(struct efrm_resource *rs)
++{
++ irq_flags_t lock_flags;
++ struct efrm_resource_manager *rm;
++ int instance = EFRM_RESOURCE_INSTANCE(rs->rs_handle);
++
++ EFRM_ASSERT(EFRM_RESOURCE_TYPE(rs->rs_handle) < EFRM_RESOURCE_NUM);
++ rm = efrm_rm_table[EFRM_RESOURCE_TYPE(rs->rs_handle)];
++ EFRM_ASSERT(EFRM_RESOURCE_TYPE(rs->rs_handle) == rm->rm_type);
++ EFRM_RESOURCE_ASSERT_VALID(rs, 0);
++
++ /* Put an entry in the resource table. */
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++ if ((unsigned)instance >= rm->rm_table_size) {
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++ if (grow_table(rm, instance + 1) < 0)
++ return -ENOMEM;
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++ }
++ EFRM_ASSERT(rm->rm_table_size > (unsigned)instance);
++ EFRM_ASSERT(rm->rm_table[instance] == NULL);
++ rm->rm_table[instance] = rs;
++ rm->rm_resources++;
++ if (rm->rm_resources > rm->rm_resources_hiwat)
++ rm->rm_resources_hiwat = rm->rm_resources;
++
++ /* Put the resource in the linked list. */
++ /* ?? broken list_add(&rm->rm_resources, &rs->rs_link); */
++ /* DJR wrote that it causes problem on driver unload, and DR tried
++ * it and saw (probably) this cause an assertion failure due to a
++ * bad link structure in
++ * /runbench/results/2005/09/22/0_DupTester_15-16-46 */
++
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++
++ return 0;
++}
++
++bool __efrm_resource_ref_count_zero(unsigned type, unsigned instance)
++{
++ /* This is rather nasty because when a resource's ref count goes to
++ * zero there is still a pointer to it in the [rm_table]. Thus
++ * arriving here does not guarantee that we have exclusive access
++ * to the resource and can free it. In fact the resource may
++ * already have been freed by another thread (after we dropped our
++ * ref, but before arriving here).
++ *
++ * At this point the only pointers to this resource should be [rs]
++ * and the one in [rm_table]. EXCEPT: Someone could have got in
++ * and looked-up the resource in the table before we got the lock.
++ * In this case the ref will have been hiked again.
++ *
++ * Therefore, if ref count is non-zero here, we shouldn't do
++ * anything, as someone else holds a ref to the resource, and will
++ * eventually release it.
++ *
++ * Otherwise, we zero-out the table entry. Therefore we have the
++ * only pointer to the resource, and can kill it safely.
++ */
++ struct efrm_resource_manager *rm = efrm_rm_table[type];
++ irq_flags_t lock_flags;
++ struct efrm_resource *rs;
++ bool do_free = false;
++
++ EFRM_TRACE("efrm_resource_ref_count_zero: type=%d instance=%d",
++ rm->rm_type, instance);
++
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(rm);
++ EFRM_ASSERT(rm->rm_table_size > instance);
++
++ spin_lock_irqsave(&rm->rm_lock, lock_flags);
++
++ rs = rm->rm_table[instance];
++ if (rs != NULL) {
++ do_free = atomic_read(&rs->rs_ref_count) == 0;
++ if (do_free) {
++ EFRM_ASSERT(rm->rm_resources > 0);
++ --rm->rm_resources;
++ rm->rm_table[instance] = 0;
++ }
++ }
++
++ spin_unlock_irqrestore(&rm->rm_lock, lock_flags);
++
++ return do_free;
++}
++EXPORT_SYMBOL(__efrm_resource_ref_count_zero);
++
++/*
++ * vi: sw=8:ai:aw
++ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/resources.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/resources.c 2008-05-19 00:33:29.445843419 +0300
+@@ -0,0 +1,94 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains resource managers initialisation functions.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/private.h>
++#include <ci/efrm/buffer_table.h>
++
++int
++efrm_resources_init(const struct vi_resource_dimensions *vi_res_dim,
++ int buffer_table_min, int buffer_table_max)
++{
++ int i, rc;
++
++ rc = efrm_buffer_table_ctor(buffer_table_min, buffer_table_max);
++ if (rc != 0)
++ return rc;
++
++ /* Create resources in the correct order */
++ for (i = 0; i < EFRM_RESOURCE_NUM; ++i) {
++ struct efrm_resource_manager **rmp = &efrm_rm_table[i];
++
++ EFRM_ASSERT(*rmp == NULL);
++ switch (i) {
++ case EFRM_RESOURCE_VI:
++ rc = efrm_create_vi_resource_manager(rmp,
++ vi_res_dim);
++ break;
++ case EFRM_RESOURCE_FILTER:
++ rc = efrm_create_filter_resource_manager(rmp);
++ break;
++ case EFRM_RESOURCE_IOBUFSET:
++ rc = efrm_create_iobufset_resource_manager(rmp);
++ break;
++ default:
++ rc = 0;
++ break;
++ }
++
++ if (rc < 0) {
++ EFRM_ERR("%s: failed type=%d (%d)",
++ __FUNCTION__, i, rc);
++ efrm_buffer_table_dtor();
++ return rc;
++ }
++ }
++
++ return 0;
++}
++
++void efrm_resources_fini(void)
++{
++ int i;
++
++ for (i = EFRM_RESOURCE_NUM - 1; i >= 0; --i)
++ if (efrm_rm_table[i]) {
++ efrm_resource_manager_dtor(efrm_rm_table[i]);
++ efrm_rm_table[i] = NULL;
++ }
++
++ efrm_buffer_table_dtor();
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/vi_resource_alloc.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/vi_resource_alloc.c 2008-05-19 00:33:29.445843419 +0300
+@@ -0,0 +1,876 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains allocation of VI resources.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/efhw/iopage.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/public.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/buffer_table.h>
++#include <ci/efrm/vi_resource_private.h>
++
++/*** Data definitions ****************************************************/
++
++static const char *dmaq_names[] = { "TX", "RX" };
++
++struct vi_resource_manager *efrm_vi_manager;
++
++/*** Forward references **************************************************/
++
++static int
++efrm_vi_resource_alloc_or_free(int alloc, struct vi_resource *evq_virs,
++ uint16_t vi_flags, int32_t evq_capacity,
++ int32_t txq_capacity, int32_t rxq_capacity,
++ uint8_t tx_q_tag, uint8_t rx_q_tag,
++ struct vi_resource **virs_in_out);
++
++/*** Reference count handling ********************************************/
++
++static inline void efrm_vi_rm_get_ref(struct vi_resource *virs)
++{
++ atomic_inc(&virs->evq_refs);
++}
++
++static inline void efrm_vi_rm_drop_ref(struct vi_resource *virs)
++{
++ EFRM_ASSERT(atomic_read(&virs->evq_refs) != 0);
++ if (atomic_dec_and_test(&virs->evq_refs))
++ efrm_vi_resource_alloc_or_free(false, NULL, 0, 0, 0, 0, 0, 0,
++ &virs);
++}
++
++/*** Instance numbers ****************************************************/
++
++static inline int efrm_vi_rm_alloc_id(uint16_t vi_flags, int32_t evq_capacity)
++{
++ irq_flags_t lock_flags;
++ int instance;
++ int rc;
++
++ if (efrm_nic_table.a_nic == NULL) /* ?? FIXME: surely not right */
++ return -ENODEV;
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ /* Falcon A1 RX phys addr wierdness. */
++ if (efrm_nic_table.a_nic->devtype.variant == 'A' &&
++ (vi_flags & EFHW_VI_RX_PHYS_ADDR_EN)) {
++ if (vi_flags & EFHW_VI_JUMBO_EN) {
++ /* Falcon-A cannot do phys + scatter. */
++ EFRM_WARN
++ ("%s: falcon-A does not support phys+scatter mode",
++ __FUNCTION__);
++ instance = -1;
++ } else if (efrm_vi_manager->iscsi_dmaq_instance_is_free
++ && evq_capacity == 0) {
++ /* Falcon-A has a single RXQ that gives the correct
++ * semantics for physical addressing. However, it
++ * happens to have the same instance number as the
++ * 'char' event queue, so we cannot also hand out
++ * the event queue. */
++ efrm_vi_manager->iscsi_dmaq_instance_is_free = false;
++ instance = FALCON_A1_ISCSI_DMAQ;
++ } else {
++ EFRM_WARN("%s: iSCSI receive queue not free",
++ __FUNCTION__);
++ instance = -1;
++ }
++ goto unlock_out;
++ }
++
++ if (vi_flags & EFHW_VI_RM_WITH_INTERRUPT) {
++ rc = __kfifo_get(efrm_vi_manager->instances_with_interrupt,
++ (unsigned char *)&instance, sizeof(instance));
++ if (rc != sizeof(instance)) {
++ EFRM_ASSERT(rc == 0);
++ instance = -1;
++ }
++ goto unlock_out;
++ }
++
++ /* Otherwise a normal run-of-the-mill VI. */
++ rc = __kfifo_get(efrm_vi_manager->instances_with_timer,
++ (unsigned char *)&instance, sizeof(instance));
++ if (rc != sizeof(instance)) {
++ EFRM_ASSERT(rc == 0);
++ instance = -1;
++ }
++
++unlock_out:
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++ return instance;
++}
++
++static void efrm_vi_rm_free_id(int instance)
++{
++ irq_flags_t lock_flags;
++ struct kfifo *instances;
++
++ if (efrm_nic_table.a_nic == NULL) /* ?? FIXME: surely not right */
++ return;
++
++ if (efrm_nic_table.a_nic->devtype.variant == 'A' &&
++ instance == FALCON_A1_ISCSI_DMAQ) {
++ EFRM_ASSERT(efrm_vi_manager->iscsi_dmaq_instance_is_free ==
++ false);
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++ efrm_vi_manager->iscsi_dmaq_instance_is_free = true;
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock,
++ lock_flags);
++ } else {
++ if (instance >= efrm_vi_manager->with_timer_base &&
++ instance < efrm_vi_manager->with_timer_limit) {
++ instances = efrm_vi_manager->instances_with_timer;
++ } else {
++ EFRM_ASSERT(instance >=
++ efrm_vi_manager->with_interrupt_base);
++ EFRM_ASSERT(instance <
++ efrm_vi_manager->with_interrupt_limit);
++ instances = efrm_vi_manager->instances_with_interrupt;
++ }
++
++ EFRM_VERIFY_EQ(kfifo_put(instances, (unsigned char *)&instance,
++ sizeof(instance)), sizeof(instance));
++ }
++}
++
++/*** Queue sizes *********************************************************/
++
++/* NB. This should really take a nic as an argument, but that makes
++ * the buffer table allocation difficult. */
++uint32_t efrm_vi_rm_evq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ )
++{
++ return virs->evq_capacity * sizeof(efhw_event_t);
++}
++EXPORT_SYMBOL(efrm_vi_rm_evq_bytes);
++
++/* NB. This should really take a nic as an argument, but that makes
++ * the buffer table allocation difficult. */
++uint32_t efrm_vi_rm_txq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ )
++{
++ return virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX] *
++ FALCON_DMA_TX_DESC_BYTES;
++}
++EXPORT_SYMBOL(efrm_vi_rm_txq_bytes);
++
++/* NB. This should really take a nic as an argument, but that makes
++ * the buffer table allocation difficult. */
++uint32_t efrm_vi_rm_rxq_bytes(struct vi_resource *virs
++ /*,struct efhw_nic *nic */ )
++{
++ uint32_t bytes_per_desc = ((virs->flags & EFHW_VI_RX_PHYS_ADDR_EN)
++ ? FALCON_DMA_RX_PHYS_DESC_BYTES
++ : FALCON_DMA_RX_BUF_DESC_BYTES);
++ return virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX] * bytes_per_desc;
++}
++EXPORT_SYMBOL(efrm_vi_rm_rxq_bytes);
++
++static int choose_size(int size_rq, unsigned sizes)
++{
++ int size;
++
++ /* size_rq < 0 means default, but we interpret this as 'minimum'. */
++
++ for (size = 256;; size <<= 1)
++ if ((sizes & ~((size - 1) | size)) == 0)
++ return -1;
++ else if ((size & sizes) && size >= size_rq)
++ return size;
++}
++
++static int
++efrm_vi_rm_adjust_alloc_request(struct vi_resource *virs, struct efhw_nic *nic)
++{
++ int capacity;
++
++ EFRM_ASSERT(nic->efhw_func);
++
++ if (virs->evq_capacity) {
++ capacity = choose_size(virs->evq_capacity, nic->evq_sizes);
++ if (capacity < 0) {
++ EFRM_ERR("vi_resource: bad evq size %d (supported=%x)",
++ virs->evq_capacity, nic->evq_sizes);
++ return -E2BIG;
++ }
++ virs->evq_capacity = capacity;
++ }
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX]) {
++ capacity =
++ choose_size(virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX],
++ nic->txq_sizes);
++ if (capacity < 0) {
++ EFRM_ERR("vi_resource: bad txq size %d (supported=%x)",
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX],
++ nic->txq_sizes);
++ return -E2BIG;
++ }
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX] = capacity;
++ }
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX]) {
++ capacity =
++ choose_size(virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX],
++ nic->rxq_sizes);
++ if (capacity < 0) {
++ EFRM_ERR("vi_resource: bad rxq size %d (supported=%x)",
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX],
++ nic->rxq_sizes);
++ return -E2BIG;
++ }
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX] = capacity;
++ }
++
++ return 0;
++}
++
++/* remove the reference to the event queue in this VI resource and decrement
++ the event queue's use count */
++static inline void efrm_vi_rm_detach_evq(struct vi_resource *virs)
++{
++ struct vi_resource *evq_virs;
++
++ EFRM_ASSERT(virs != NULL);
++
++ evq_virs = virs->evq_virs;
++
++ if (evq_virs != NULL) {
++ virs->evq_virs = NULL;
++ if (evq_virs == virs) {
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT
++ " had internal event queue ", __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++ } else {
++ efrm_vi_rm_drop_ref(evq_virs);
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " had event queue "
++ EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle),
++ EFRM_RESOURCE_PRI_ARG(evq_virs->rs.
++ rs_handle));
++ }
++ } else {
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT
++ " had no event queue (nothing to do)",
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++ }
++}
++
++/*** Buffer Table allocations ********************************************/
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++static int
++efrm_vi_rm_alloc_or_free_buffer_table(struct vi_resource *virs, bool is_alloc)
++{
++ uint32_t bytes;
++ int page_order;
++ int rc;
++
++ if (!is_alloc)
++ goto destroy;
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX]) {
++ bytes = efrm_vi_rm_txq_bytes(virs);
++ page_order = get_order(bytes);
++ rc = efrm_buffer_table_alloc(page_order,
++ (virs->dmaq_buf_tbl_alloc +
++ EFRM_VI_RM_DMA_QUEUE_TX));
++ if (rc != 0) {
++ EFRM_TRACE
++ ("%s: Error %d allocating TX buffer table entry",
++ __FUNCTION__, rc);
++ goto fail_txq_alloc;
++ }
++ }
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX]) {
++ bytes = efrm_vi_rm_rxq_bytes(virs);
++ page_order = get_order(bytes);
++ rc = efrm_buffer_table_alloc(page_order,
++ (virs->dmaq_buf_tbl_alloc +
++ EFRM_VI_RM_DMA_QUEUE_RX));
++ if (rc != 0) {
++ EFRM_TRACE
++ ("%s: Error %d allocating RX buffer table entry",
++ __FUNCTION__, rc);
++ goto fail_rxq_alloc;
++ }
++ }
++ return 0;
++
++destroy:
++ rc = 0;
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX]) {
++ efrm_buffer_table_free(&virs->
++ dmaq_buf_tbl_alloc
++ [EFRM_VI_RM_DMA_QUEUE_RX]);
++ }
++fail_rxq_alloc:
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX]) {
++ efrm_buffer_table_free(&virs->
++ dmaq_buf_tbl_alloc
++ [EFRM_VI_RM_DMA_QUEUE_TX]);
++ }
++fail_txq_alloc:
++
++ return rc;
++}
++
++#endif /* defined(__CI_HARDWARE_CONFIG_FALCON__) */
++
++/*** Per-NIC allocations *************************************************/
++
++static inline int
++efrm_vi_rm_init_evq(struct vi_resource *virs, int nic_index)
++{
++ int rc;
++ struct efhw_nic *nic = efrm_nic_table.nic[nic_index];
++ int instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++ struct eventq_resource_hardware *evq_hw =
++ &virs->nic_info[nic_index].evq_pages;
++ uint32_t buf_bytes = efrm_vi_rm_evq_bytes(virs);
++
++ if (virs->evq_capacity == 0)
++ return 0;
++ evq_hw->capacity = virs->evq_capacity;
++
++ /* Allocate buffer table entries to map onto the iobuffer. This
++ * currently allocates its own buffer table entries on Falcon which is
++ * a bit wasteful on a multi-NIC system. */
++ evq_hw->buf_tbl_alloc.base = (unsigned)-1;
++ rc = efrm_buffer_table_alloc(get_order(buf_bytes),
++ &evq_hw->buf_tbl_alloc);
++ if (rc < 0) {
++ EFHW_WARN("%s: failed (%d) to alloc %d buffer table entries",
++ __FUNCTION__, rc, get_order(buf_bytes));
++ return rc;
++ }
++
++ /* Allocate the event queue memory. */
++ rc = efhw_nic_event_queue_alloc_iobuffer(nic, evq_hw, instance,
++ buf_bytes);
++ if (rc != 0) {
++ EFRM_ERR("%s: Error allocating iobuffer: %d", __FUNCTION__, rc);
++ efrm_buffer_table_free(&evq_hw->buf_tbl_alloc);
++ return rc;
++ }
++
++ /* Initialise the event queue hardware */
++ efhw_nic_event_queue_enable(nic, instance, virs->evq_capacity,
++ efhw_iopages_dma_addr(&evq_hw->iobuff) +
++ evq_hw->iobuff_off,
++ evq_hw->buf_tbl_alloc.base);
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " capacity=%u", __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle),
++ virs->evq_capacity);
++
++#if defined(__ia64__)
++ /* Page size may be large, so for now just increase the
++ * size of the requested evq up to a round number of
++ * pages
++ */
++ buf_bytes = CI_ROUND_UP(buf_bytes, PAGE_SIZE);
++#endif
++ EFRM_ASSERT(buf_bytes % PAGE_SIZE == 0);
++
++ virs->mem_mmap_bytes += buf_bytes;
++
++ return 0;
++}
++
++static inline void
++efrm_vi_rm_fini_evq(struct vi_resource *virs, int nic_index)
++{
++ struct efhw_nic *nic = efrm_nic_table.nic[nic_index];
++ int instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++ struct vi_resource_nic_info *nic_info = &virs->nic_info[nic_index];
++
++ if (virs->evq_capacity == 0)
++ return;
++
++ /* Zero the timer-value for this queue.
++ And Tell NIC to stop using this event queue. */
++ efhw_nic_event_queue_disable(nic, instance, 0);
++
++ if (nic_info->evq_pages.buf_tbl_alloc.base != (unsigned)-1)
++ efrm_buffer_table_free(&nic_info->evq_pages.buf_tbl_alloc);
++
++ efhw_iopages_free(nic, &nic_info->evq_pages.iobuff);
++}
++
++/*! FIXME: we should make sure this number is never zero (=> unprotected) */
++/*! FIXME: put this definition in a relevant header (e.g. as (evqid)+1) */
++#define EFAB_EVQ_OWNER_ID(evqid) ((evqid))
++
++void
++efrm_vi_rm_init_dmaq(struct vi_resource *virs, int queue_type,
++ struct efhw_nic *nic)
++{
++ int instance;
++ struct vi_resource *evq_virs;
++ int evq_instance;
++ efhw_buffer_addr_t buf_addr;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++ evq_virs = virs->evq_virs;
++ evq_instance = EFRM_RESOURCE_INSTANCE(evq_virs->rs.rs_handle);
++
++ buf_addr = virs->dmaq_buf_tbl_alloc[queue_type].base;
++
++ if (queue_type == EFRM_VI_RM_DMA_QUEUE_TX) {
++ efhw_nic_dmaq_tx_q_init(nic,
++ instance, /* dmaq */
++ evq_instance, /* evq */
++ EFAB_EVQ_OWNER_ID(evq_instance), /* owner */
++ virs->dmaq_tag[queue_type], /* tag */
++ virs->dmaq_capacity[queue_type], /* size of queue */
++ buf_addr, /* buffer index */
++ virs->flags); /* user specified Q attrs */
++ } else {
++ efhw_nic_dmaq_rx_q_init(nic,
++ instance, /* dmaq */
++ evq_instance, /* evq */
++ EFAB_EVQ_OWNER_ID(evq_instance), /* owner */
++ virs->dmaq_tag[queue_type], /* tag */
++ virs->dmaq_capacity[queue_type], /* size of queue */
++ buf_addr, /* buffer index */
++ virs->flags); /* user specified Q attrs */
++ }
++}
++
++static int
++efrm_vi_rm_init_or_fini_dmaq(struct vi_resource *virs,
++ int queue_type, int init, int nic_index)
++{
++ int rc;
++ struct efhw_nic *nic = efrm_nic_table.nic[nic_index];
++ int instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++ uint32_t buf_bytes;
++ struct vi_resource *evq_virs;
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ struct vi_resource_nic_info *nic_info = &virs->nic_info[nic_index];
++ int page_order;
++ uint32_t num_pages;
++ efhw_iopages_t *iobuff;
++#endif
++
++ if (!init)
++ goto destroy;
++
++ /* Ignore disabled queues. */
++ if (virs->dmaq_capacity[queue_type] == 0) {
++ if (queue_type == EFRM_VI_RM_DMA_QUEUE_TX)
++ efhw_nic_dmaq_tx_q_disable(nic, instance);
++ else
++ efhw_nic_dmaq_rx_q_disable(nic, instance);
++ return 0;
++ }
++
++ buf_bytes = (queue_type == EFRM_VI_RM_DMA_QUEUE_TX
++ ? efrm_vi_rm_txq_bytes(virs)
++ : efrm_vi_rm_rxq_bytes(virs));
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ page_order = get_order(buf_bytes);
++
++ rc = efhw_iopages_alloc(nic, &nic_info->dmaq_pages[queue_type],
++ page_order);
++ if (rc != 0) {
++ EFRM_ERR("%s: Failed to allocate %s DMA buffer.", __FUNCTION__,
++ dmaq_names[queue_type]);
++ goto fail_iopages;
++ }
++
++ num_pages = 1 << page_order;
++ iobuff = &nic_info->dmaq_pages[queue_type];
++ efhw_nic_buffer_table_set_n(nic,
++ virs->dmaq_buf_tbl_alloc[queue_type].base,
++ efhw_iopages_dma_addr(iobuff),
++ EFHW_NIC_PAGE_SIZE, 0, num_pages, 0);
++
++ falcon_nic_buffer_table_confirm(nic);
++
++ virs->mem_mmap_bytes += round_up(buf_bytes, PAGE_SIZE);
++#endif /* __CI_HARDWARE_CONFIG_FALCON__ */
++
++ evq_virs = virs->evq_virs;
++ EFRM_ASSERT(evq_virs);
++
++ /* Make sure there is an event queue. */
++ if (evq_virs->evq_capacity <= 0) {
++ EFRM_ERR("%s: Cannot use empty event queue for %s DMA",
++ __FUNCTION__, dmaq_names[queue_type]);
++ rc = -EINVAL;
++ goto fail_evq;
++ }
++
++ efrm_vi_rm_init_dmaq(virs, queue_type, nic);
++
++ return 0;
++
++destroy:
++ rc = 0;
++
++ /* Ignore disabled queues. */
++ if (virs->dmaq_capacity[queue_type] == 0)
++ return 0;
++
++ /* No need to disable the queue here. Nobody is using it anyway. */
++
++fail_evq:
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ efhw_iopages_free(nic, &nic_info->dmaq_pages[queue_type]);
++fail_iopages:
++#endif
++
++ return rc;
++}
++
++static int
++efrm_vi_rm_init_or_fini_nic(struct vi_resource *virs, int init, int nic_index)
++{
++ struct vi_resource *evq_virs;
++ int rc;
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++#ifndef NDEBUG
++ int instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++#endif
++#endif
++
++ if (!init)
++ goto destroy;
++
++ evq_virs = virs->evq_virs;
++ if (evq_virs != virs) {
++ if (!efrm_nic_set_read(&evq_virs->nic_set, nic_index)) {
++ /* Ignore this NIC. It's not supported by the event
++ * queue. */
++ return 0;
++ }
++ }
++
++ rc = efrm_vi_rm_init_evq(virs, nic_index);
++ if (rc != 0)
++ goto fail_evq;
++
++ rc = efrm_vi_rm_init_or_fini_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_TX,
++ init, nic_index);
++ if (rc != 0)
++ goto fail_txq;
++
++ rc = efrm_vi_rm_init_or_fini_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_RX,
++ init, nic_index);
++ if (rc != 0)
++ goto fail_rxq;
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ /* Allocate space for the control page. */
++ EFRM_ASSERT(falcon_tx_dma_page_offset(instance) < PAGE_SIZE);
++ EFRM_ASSERT(falcon_rx_dma_page_offset(instance) < PAGE_SIZE);
++ EFRM_ASSERT(falcon_timer_page_offset(instance) < PAGE_SIZE);
++ virs->bar_mmap_bytes += PAGE_SIZE;
++#endif
++
++ /* Mark the NIC as having been initialised. */
++ efrm_nic_set_write(&virs->nic_set, nic_index, true);
++
++ return 0;
++
++destroy:
++ rc = 0;
++
++ efrm_vi_rm_init_or_fini_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_RX,
++ false, nic_index);
++fail_rxq:
++
++ efrm_vi_rm_init_or_fini_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_TX,
++ false, nic_index);
++fail_txq:
++
++ efrm_vi_rm_fini_evq(virs, nic_index);
++fail_evq:
++
++ /* Mark the NIC as having been finalised. */
++ efrm_nic_set_write(&virs->nic_set, nic_index, false);
++ EFRM_ASSERT(rc != 0 || !init);
++
++ return rc;
++}
++
++static int
++efrm_vi_resource_alloc_or_free(int alloc, struct vi_resource *evq_virs,
++ uint16_t vi_flags, int32_t evq_capacity,
++ int32_t txq_capacity, int32_t rxq_capacity,
++ uint8_t tx_q_tag, uint8_t rx_q_tag,
++ struct vi_resource **virs_in_out)
++{
++ struct vi_resource *virs;
++ int rc;
++ int instance;
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_ASSERT(virs_in_out);
++ EFRM_ASSERT(efrm_vi_manager);
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(&efrm_vi_manager->rm);
++
++ if (!alloc)
++ goto destroy;
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ rx_q_tag &= (1 << TX_DESCQ_LABEL_WIDTH) - 1;
++ tx_q_tag &= (1 << RX_DESCQ_LABEL_WIDTH) - 1;
++#endif
++
++ virs = kmalloc(sizeof(*virs), GFP_KERNEL);
++ if (virs == NULL) {
++ EFRM_ERR("%s: Error allocating VI resource object",
++ __FUNCTION__);
++ rc = -ENOMEM;
++ goto fail_alloc;
++ }
++ memset(virs, 0, sizeof(*virs));
++
++ /* Some macros make the assumption that the struct efrm_resource is
++ * the first member of a struct vi_resource. */
++ EFRM_ASSERT(&virs->rs == (struct efrm_resource *) (virs));
++
++ instance = efrm_vi_rm_alloc_id(vi_flags, evq_capacity);
++ if (instance < 0) {
++ /* Clear out the close list... */
++ efrm_vi_rm_salvage_flushed_vis();
++ instance = efrm_vi_rm_alloc_id(vi_flags, evq_capacity);
++ if (instance >= 0)
++ EFRM_TRACE("%s: Salvaged a closed VI.", __FUNCTION__);
++ }
++
++ if (instance < 0) {
++ /* Could flush resources and try again here. */
++ EFRM_ERR("%s: Out of appropriate VI resources", __FUNCTION__);
++ rc = -EBUSY;
++ goto fail_alloc_id;
++ }
++
++ EFRM_TRACE("%s: new VI ID %d", __FUNCTION__, instance);
++ efrm_resource_init(&virs->rs, EFRM_RESOURCE_VI, instance);
++
++ /* Start with one reference. Any external VIs using the EVQ of this
++ * resource will increment this reference rather than the resource
++ * reference to avoid DMAQ flushes from waiting for other DMAQ
++ * flushes to complete. When the resource reference goes to zero,
++ * the DMAQ flush happens. When the flush completes, this reference
++ * is decremented. When this reference reaches zero, the instance
++ * is freed. */
++ atomic_set(&virs->evq_refs, 1);
++
++ efrm_nic_set_clear(&virs->nic_set);
++
++ virs->bar_mmap_bytes = 0;
++ virs->mem_mmap_bytes = 0;
++ virs->evq_capacity = evq_capacity;
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX] = txq_capacity;
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX] = rxq_capacity;
++ virs->dmaq_tag[EFRM_VI_RM_DMA_QUEUE_TX] = tx_q_tag;
++ virs->dmaq_tag[EFRM_VI_RM_DMA_QUEUE_RX] = rx_q_tag;
++ virs->flags = vi_flags;
++
++ INIT_LIST_HEAD(&virs->tx_flush_link);
++ INIT_LIST_HEAD(&virs->rx_flush_link);
++ efrm_nic_set_clear(&virs->tx_flush_nic_set);
++ efrm_nic_set_clear(&virs->rx_flush_nic_set);
++
++ memset(&efrm_vi_manager->evq_infos[instance], 0,
++ sizeof(struct vi_resource_evq_info));
++ efrm_vi_manager->evq_infos[instance].evq_virs = virs;
++
++ /* Adjust the queue sizes. */
++ rc = 0;
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ if (rc == 0)
++ rc = efrm_vi_rm_adjust_alloc_request(virs, nic);
++ if (rc != 0)
++ goto fail_adjust_request;
++
++ /* Attach the EVQ early so that we can ensure that the NIC sets
++ * match. */
++ if (evq_virs == NULL) {
++ evq_virs = virs;
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT
++ " has no external event queue", __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++ } else {
++ /* Make sure the resource managers are the same. */
++ if (EFRM_RESOURCE_TYPE(evq_virs->rs.rs_handle) !=
++ EFRM_RESOURCE_VI) {
++ EFRM_ERR("%s: Mismatched owner for event queue VI "
++ EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(evq_virs->rs.rs_handle));
++ return -EINVAL;
++ }
++ EFRM_ASSERT(atomic_read(&evq_virs->evq_refs) != 0);
++ efrm_vi_rm_get_ref(evq_virs);
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " uses event queue "
++ EFRM_RESOURCE_FMT,
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle),
++ EFRM_RESOURCE_PRI_ARG(evq_virs->rs.rs_handle));
++ }
++ virs->evq_virs = evq_virs;
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ rc = efrm_vi_rm_alloc_or_free_buffer_table(virs, true);
++ if (rc != 0)
++ goto fail_buffer_table;
++#endif
++
++ rc = 0;
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ if (rc == 0)
++ /* This updates virs->nic_set for the NICs which need
++ * finalising. */
++ rc = efrm_vi_rm_init_or_fini_nic(virs, true, nic_i);
++ if (rc != 0)
++ goto fail_init_nic;
++
++ /* Put it into the resource manager's table. */
++ rc = efrm_resource_manager_insert(&virs->rs);
++ if (rc != 0) {
++ if (atomic_dec_and_test(&virs->rs.rs_ref_count))
++ efrm_vi_resource_free(virs);
++ return rc;
++ }
++
++ *virs_in_out = virs;
++ EFRM_TRACE("%s: Allocated " EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++ return 0;
++
++destroy:
++ virs = *virs_in_out;
++ EFRM_RESOURCE_ASSERT_VALID(&virs->rs, 1);
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ EFRM_TRACE("%s: Freeing %d", __FUNCTION__,
++ EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle));
++
++ /* Destroying the VI. The reference count must be zero. */
++ EFRM_ASSERT(atomic_read(&virs->evq_refs) == 0);
++
++ /* The EVQ should have gone (and DMA disabled) so that this
++ * function can't be re-entered to destroy the EVQ VI. */
++ EFRM_ASSERT(virs->evq_virs == NULL);
++ rc = 0;
++
++fail_init_nic:
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic)
++ efrm_vi_rm_init_or_fini_nic(virs, false, nic_i);
++
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ efrm_vi_rm_alloc_or_free_buffer_table(virs, false);
++fail_buffer_table:
++#endif
++
++ efrm_vi_rm_detach_evq(virs);
++
++fail_adjust_request:
++
++ EFRM_ASSERT(virs->evq_callback_fn == NULL);
++ memset(&efrm_vi_manager->evq_infos[instance], 0,
++ sizeof(struct vi_resource_evq_info));
++ EFRM_TRACE("%s: delete VI ID %d", __FUNCTION__, instance);
++ efrm_vi_rm_free_id(instance);
++fail_alloc_id:
++
++ EFRM_DO_DEBUG(memset(virs, 0, sizeof(*virs)));
++ kfree(virs);
++fail_alloc:
++ *virs_in_out = NULL;
++
++ return rc;
++}
++
++/*** Resource object ****************************************************/
++
++int
++efrm_vi_resource_alloc(struct vi_resource *evq_virs,
++ uint16_t vi_flags, int32_t evq_capacity,
++ int32_t txq_capacity, int32_t rxq_capacity,
++ uint8_t tx_q_tag, uint8_t rx_q_tag,
++ struct vi_resource **virs_out,
++ uint32_t *out_io_mmap_bytes,
++ uint32_t *out_mem_mmap_bytes,
++ uint32_t *out_txq_capacity, uint32_t *out_rxq_capacity)
++{
++ int rc;
++ rc = efrm_vi_resource_alloc_or_free(true, evq_virs, vi_flags,
++ evq_capacity, txq_capacity,
++ rxq_capacity, tx_q_tag, rx_q_tag,
++ virs_out);
++ if (rc == 0) {
++ if (out_io_mmap_bytes != NULL)
++ *out_io_mmap_bytes = (*virs_out)->bar_mmap_bytes;
++ if (out_mem_mmap_bytes != NULL)
++ *out_mem_mmap_bytes = (*virs_out)->mem_mmap_bytes;
++ if (out_txq_capacity != NULL)
++ *out_txq_capacity =
++ (*virs_out)->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX];
++ if (out_rxq_capacity != NULL)
++ *out_rxq_capacity =
++ (*virs_out)->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX];
++ }
++
++ return rc;
++}
++EXPORT_SYMBOL(efrm_vi_resource_alloc);
++
++void efrm_vi_rm_free_flushed_resource(struct vi_resource *virs)
++{
++ EFRM_ASSERT(virs != NULL);
++ EFRM_ASSERT(atomic_read(&virs->rs.rs_ref_count) == 0);
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++ /* release the associated event queue then drop our own reference
++ * count */
++ efrm_vi_rm_detach_evq(virs);
++ efrm_vi_rm_drop_ref(virs);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/vi_resource_event.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/vi_resource_event.c 2008-05-19 00:33:29.445843419 +0300
+@@ -0,0 +1,232 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains event handling for VI resource.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/eventq.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/vi_resource_private.h>
++
++void
++efrm_eventq_request_wakeup(struct vi_resource *virs, unsigned current_ptr,
++ unsigned nic_index)
++{
++ struct efhw_nic *nic;
++ int next_i;
++ EFRM_ASSERT(efrm_nic_set_read(&virs->nic_set, nic_index));
++ nic = efrm_nic_table.nic[nic_index];
++ EFRM_ASSERT(nic);
++ next_i = ((current_ptr / sizeof(efhw_event_t)) &
++ (virs->evq_capacity - 1));
++
++ efhw_nic_wakeup_request(nic, efrm_eventq_dma_addr(virs, nic_index),
++ next_i,
++ EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle));
++}
++EXPORT_SYMBOL(efrm_eventq_request_wakeup);
++
++void efrm_eventq_reset(struct vi_resource *virs, int nic_index)
++{
++ struct efhw_nic *nic = efrm_nic_table.nic[nic_index];
++ int instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ EFRM_ASSERT(virs->evq_capacity != 0);
++ EFRM_ASSERT(efrm_nic_set_read(&virs->nic_set, nic_index));
++
++ /* FIXME: Protect against concurrent resets. */
++
++ efhw_nic_event_queue_disable(nic, instance, 0);
++
++ memset(efrm_eventq_base(virs, nic_index), EFHW_CLEAR_EVENT_VALUE,
++ efrm_eventq_bytes(virs, nic_index));
++ efhw_nic_event_queue_enable(nic, instance, virs->evq_capacity,
++ efrm_eventq_dma_addr(virs, nic_index),
++ virs->nic_info[nic_index].evq_pages.
++ buf_tbl_alloc.base);
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT, __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle));
++}
++EXPORT_SYMBOL(efrm_eventq_reset);
++
++int
++efrm_eventq_register_callback(struct vi_resource *virs,
++ void (*handler) (void *, int,
++ struct efhw_nic *nic),
++ void *arg)
++{
++ int instance;
++ int bit;
++
++ EFRM_RESOURCE_ASSERT_VALID(&virs->rs, 0);
++ EFRM_ASSERT(virs->evq_capacity != 0);
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ /* The handler can be set only once. */
++ bit = test_and_set_bit(VI_RESOURCE_EVQ_STATE_CALLBACK_REGISTERED,
++ &efrm_vi_manager->evq_infos[instance].evq_state);
++ if (bit)
++ return -EBUSY;
++
++ /* Store the details. The order is important here. */
++ virs->evq_callback_arg = arg;
++ virs->evq_callback_fn = handler;
++
++ return 0;
++}
++EXPORT_SYMBOL(efrm_eventq_register_callback);
++
++void efrm_eventq_kill_callback(struct vi_resource *virs)
++{
++ int nic_i, instance;
++ struct efhw_nic *nic;
++ struct vi_resource_evq_info *evq_info;
++ int32_t evq_state;
++ int bit;
++
++ EFRM_RESOURCE_ASSERT_VALID(&virs->rs, 0);
++ EFRM_ASSERT(virs->evq_capacity != 0);
++
++ /* Clean out the callback so a new one can be installed. */
++ virs->evq_callback_fn = NULL;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++ evq_info = &efrm_vi_manager->evq_infos[instance];
++
++ /* Disable the event queue. */
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic)
++ efhw_nic_event_queue_disable(nic, instance, /*timer_only */ 1);
++
++ /* Disable the callback. */
++ bit = test_and_clear_bit(VI_RESOURCE_EVQ_STATE_CALLBACK_REGISTERED,
++ &evq_info->evq_state);
++ EFRM_ASSERT(bit); /* do not call me twice! */
++
++ /* Spin until the callback is complete. */
++ do {
++ rmb();
++
++ udelay(1);
++ evq_state = evq_info->evq_state;
++ } while ((evq_state & VI_RESOURCE_EVQ_STATE(BUSY)));
++}
++EXPORT_SYMBOL(efrm_eventq_kill_callback);
++
++static void
++efrm_eventq_do_callback(struct efhw_nic *nic, unsigned instance,
++ bool is_timeout)
++{
++ void (*handler) (void *, int is_timeout, struct efhw_nic *nic);
++ void *arg;
++ struct vi_resource_evq_info *evq_info;
++ int32_t evq_state;
++ int32_t new_evq_state;
++ struct vi_resource *virs;
++ int bit;
++
++ EFRM_TRACE("%s: q=%d %s", __FUNCTION__, instance,
++ is_timeout ? "timeout" : "wakeup");
++ EFRM_ASSERT(efrm_vi_manager);
++
++ evq_info = &efrm_vi_manager->evq_infos[instance];
++
++ /* Set the BUSY bit and clear WAKEUP_PENDING. Do this
++ * before waking up the sleeper to avoid races. */
++ while (1) {
++ evq_state = evq_info->evq_state;
++ new_evq_state = evq_state;
++
++ if ((evq_state & VI_RESOURCE_EVQ_STATE(BUSY)) != 0) {
++ EFRM_ERR("%s:%d: evq_state[%d] corrupted!",
++ __FUNCTION__, __LINE__, instance);
++ return;
++ }
++
++ if (!is_timeout)
++ new_evq_state &= ~VI_RESOURCE_EVQ_STATE(WAKEUP_PENDING);
++
++ if (evq_state & VI_RESOURCE_EVQ_STATE(CALLBACK_REGISTERED)) {
++ new_evq_state |= VI_RESOURCE_EVQ_STATE(BUSY);
++ if (cmpxchg(&evq_info->evq_state, evq_state,
++ new_evq_state) == evq_state) {
++ virs = evq_info->evq_virs;
++ break;
++ }
++
++ } else {
++ /* Just update the state if necessary. */
++ if (new_evq_state == evq_state ||
++ cmpxchg(&evq_info->evq_state, evq_state,
++ new_evq_state) == evq_state)
++ return;
++ }
++
++ udelay(1);
++ }
++
++ /* Call the callback if any. */
++ if (evq_state & VI_RESOURCE_EVQ_STATE(CALLBACK_REGISTERED)) {
++ /* Retrieve the callback fn. */
++ handler = virs->evq_callback_fn;
++ arg = virs->evq_callback_arg;
++ if (handler != NULL) /* avoid races */
++ handler(arg, is_timeout, nic);
++ }
++
++ /* Clear the BUSY bit. */
++ bit =
++ test_and_clear_bit(VI_RESOURCE_EVQ_STATE_BUSY,
++ &evq_info->evq_state);
++ if (!bit) {
++ EFRM_ERR("%s:%d: evq_state corrupted!",
++ __FUNCTION__, __LINE__);
++ }
++}
++
++void efrm_handle_wakeup_event(struct efhw_nic *nic, efhw_event_t *ev)
++{
++ efrm_eventq_do_callback(nic,
++ (unsigned int)FALCON_EVENT_WAKE_EVQ_ID(ev),
++ false);
++}
++
++void efrm_handle_timeout_event(struct efhw_nic *nic, efhw_event_t *ev)
++{
++ efrm_eventq_do_callback(nic,
++ (unsigned int)FALCON_EVENT_WAKE_EVQ_ID(ev),
++ true);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/vi_resource_flush.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/vi_resource_flush.c 2008-05-19 00:33:29.449843650 +0300
+@@ -0,0 +1,506 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains DMA queue flushing of VI resources.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/sysdep.h>
++#include <ci/efrm/buffer_table.h>
++#include <ci/efrm/vi_resource_private.h>
++
++#if EFRM_VI_USE_WORKQUEUE
++ /* can fail as workitem can already be scheuled -- ignore failure */
++#define EFRM_VI_RM_DELAYED_FREE(manager) \
++ queue_work(manager->workqueue, &manager->work_item)
++#else
++#define EFRM_VI_RM_DELAYED_FREE(manager) \
++ efrm_vi_rm_delayed_free(&manager->work_item)
++#endif
++
++static const int flush_fifo_hwm = 8 /* TODO should be a HW specific const */ ;
++
++static void
++efrm_vi_resource_rx_flush_done(struct vi_resource *virs, int nic_i,
++ bool *completed)
++{
++ /* We should only get a flush event if there is a flush
++ * outstanding. */
++ EFRM_ASSERT(efrm_nic_set_read
++ (&virs->rx_flush_outstanding_nic_set, nic_i));
++
++ efrm_nic_set_write(&virs->rx_flush_outstanding_nic_set, nic_i, false);
++ efrm_nic_set_write(&virs->rx_flush_nic_set, nic_i, false);
++
++ if (efrm_nic_set_is_all_clear(&virs->rx_flush_outstanding_nic_set)) {
++ list_del(&virs->rx_flush_link);
++ efrm_vi_manager->rx_flush_outstanding_count--;
++
++ if (efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set)) {
++ list_add_tail(&virs->rx_flush_link,
++ &efrm_vi_manager->close_pending);
++ *completed = 1;
++ }
++ }
++}
++
++static void
++efrm_vi_resource_tx_flush_done(struct vi_resource *virs, int nic_i,
++ bool *completed)
++{
++ /* We should only get a flush event if there is a flush
++ * outstanding. */
++ EFRM_ASSERT(efrm_nic_set_read(&virs->tx_flush_nic_set, nic_i));
++
++ efrm_nic_set_write(&virs->tx_flush_nic_set, nic_i, false);
++
++ if (efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set)) {
++ list_del(&virs->tx_flush_link);
++
++ if (efrm_nic_set_is_all_clear(&virs->rx_flush_nic_set)) {
++ list_add_tail(&virs->rx_flush_link,
++ &efrm_vi_manager->close_pending);
++ *completed = 1;
++ }
++ }
++}
++
++static void
++efrm_vi_resource_issue_rx_flush(struct vi_resource *virs, bool *completed)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++ int instance;
++ int rc;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ list_add_tail(&virs->rx_flush_link,
++ &efrm_vi_manager->rx_flush_outstanding_list);
++ virs->rx_flush_outstanding_nic_set = virs->rx_flush_nic_set;
++ efrm_vi_manager->rx_flush_outstanding_count++;
++
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic) {
++ EFRM_TRACE("%s: rx queue %d flush requested for nic %d",
++ __FUNCTION__, instance, nic->index);
++ rc = efhw_nic_flush_rx_dma_channel(nic, instance);
++ if (rc == -EAGAIN)
++ efrm_vi_resource_rx_flush_done(virs, nic_i, completed);
++ }
++}
++
++static void
++efrm_vi_resource_issue_tx_flush(struct vi_resource *virs, bool *completed)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++ int instance;
++ int rc;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ list_add_tail(&virs->tx_flush_link,
++ &efrm_vi_manager->tx_flush_outstanding_list);
++
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic) {
++ EFRM_TRACE("%s: tx queue %d flush requested for nic %d",
++ __FUNCTION__, instance, nic->index);
++ rc = efhw_nic_flush_tx_dma_channel(nic, instance);
++ if (rc == -EAGAIN)
++ efrm_vi_resource_tx_flush_done(virs, nic_i, completed);
++ }
++}
++
++static void efrm_vi_resource_process_waiting_flushes(bool *completed)
++{
++ struct vi_resource *virs;
++
++ while (efrm_vi_manager->rx_flush_outstanding_count < flush_fifo_hwm &&
++ !list_empty(&efrm_vi_manager->rx_flush_waiting_list)) {
++ virs =
++ list_entry(list_pop
++ (&efrm_vi_manager->rx_flush_waiting_list),
++ struct vi_resource, rx_flush_link);
++ efrm_vi_resource_issue_rx_flush(virs, completed);
++ }
++}
++
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++static void
++efrm_vi_resource_flush_retry_vi(struct vi_resource *virs,
++ int64_t time_now, bool *completed)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++ int instance;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ virs->flush_count++;
++ virs->flush_time = time_now;
++
++#if BUG7916_WORKAROUND
++ if (!efrm_nic_set_is_all_clear(&virs->rx_flush_outstanding_nic_set)) {
++ EFRM_TRACE("%s: Retrying RX flush on instance %d",
++ __FUNCTION__, instance);
++
++ list_del(&virs->rx_flush_link);
++ efrm_vi_manager->rx_flush_outstanding_count--;
++ efrm_vi_resource_issue_rx_flush(virs, completed);
++ efrm_vi_resource_process_waiting_flushes(completed);
++ }
++#endif
++
++#if BUG5302_WORKAROUND
++ if (!efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set)) {
++ if (virs->flush_count > 5) {
++ EFRM_TRACE("%s: VI resource stuck flush pending "
++ "(instance=%d, count=%d)",
++ __FUNCTION__, instance, virs->flush_count);
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->tx_flush_nic_set,
++ nic_i, nic) {
++ falcon_clobber_tx_dma_ptrs(nic, instance);
++ }
++ } else {
++ EFRM_TRACE("%s: Retrying TX flush on instance %d",
++ __FUNCTION__, instance);
++ }
++
++ list_del(&virs->tx_flush_link);
++ efrm_vi_resource_issue_tx_flush(virs, completed);
++ }
++#endif
++}
++#endif
++
++int efrm_vi_resource_flush_retry(struct vi_resource *virs)
++{
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++ irq_flags_t lock_flags;
++ bool completed = false;
++
++ if (efrm_nic_set_is_all_clear(&virs->rx_flush_nic_set) &&
++ efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set))
++ return -EALREADY;
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++ efrm_vi_resource_flush_retry_vi(virs, get_jiffies_64(), &completed);
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ if (completed)
++ EFRM_VI_RM_DELAYED_FREE(efrm_vi_manager);
++#endif
++
++ return 0;
++}
++EXPORT_SYMBOL(efrm_vi_resource_flush_retry);
++
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++/* resource manager lock should be taken before this call */
++static void efrm_vi_handle_flush_loss(bool *completed)
++{
++ struct list_head *pos, *temp;
++ struct vi_resource *virs;
++ int64_t time_now, time_pending;
++
++ /* It's possible we miss flushes - the list is sorted in order we
++ * generate flushes, see if any are very old. It's also possible
++ * that we decide an endpoint is flushed even though we've not
++ * received all the flush events. We *should * mark as
++ * completed, reclaim and loop again. ??
++ * THIS NEEDS BACKPORTING FROM THE FALCON branch
++ */
++ time_now = get_jiffies_64();
++
++#if BUG7916_WORKAROUND
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->rx_flush_outstanding_list) {
++ virs = container_of(pos, struct vi_resource, rx_flush_link);
++
++ time_pending = time_now - virs->flush_time;
++
++ /* List entries are held in reverse chronological order. Only
++ * process the old ones. */
++ if (time_pending <= 0x100000000LL)
++ break;
++
++ efrm_vi_resource_flush_retry_vi(virs, time_now, completed);
++ }
++#endif
++
++#if BUG5302_WORKAROUND
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->tx_flush_outstanding_list) {
++ virs = container_of(pos, struct vi_resource, tx_flush_link);
++
++ time_pending = time_now - virs->flush_time;
++
++ /* List entries are held in reverse chronological order.
++ * Only process the old ones. */
++ if (time_pending <= 0x100000000LL)
++ break;
++
++ efrm_vi_resource_flush_retry_vi(virs, time_now, completed);
++ }
++#endif
++}
++#endif
++
++void
++efrm_vi_register_flush_callback(struct vi_resource *virs,
++ void (*handler)(void *), void *arg)
++{
++ if (handler == NULL) {
++ virs->flush_callback_fn = handler;
++ wmb();
++ virs->flush_callback_arg = arg;
++ } else {
++ virs->flush_callback_arg = arg;
++ wmb();
++ virs->flush_callback_fn = handler;
++ }
++}
++EXPORT_SYMBOL(efrm_vi_register_flush_callback);
++
++int efrm_pt_flush(struct vi_resource *virs)
++{
++ int instance;
++ irq_flags_t lock_flags;
++ bool completed = false;
++
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ EFRM_ASSERT(efrm_nic_set_is_all_clear(&virs->rx_flush_nic_set));
++ EFRM_ASSERT(efrm_nic_set_is_all_clear
++ (&virs->rx_flush_outstanding_nic_set));
++ EFRM_ASSERT(efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set));
++
++ EFRM_TRACE("%s: " EFRM_RESOURCE_FMT " EVQ=%d TXQ=%d RXQ=%d",
++ __FUNCTION__, EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle),
++ virs->evq_capacity,
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX],
++ virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX]);
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX] != 0)
++ virs->rx_flush_nic_set = virs->nic_set;
++
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX] != 0)
++ virs->tx_flush_nic_set = virs->nic_set;
++
++ /* Clean up immediately if there are no flushes. */
++ if (efrm_nic_set_is_all_clear(&virs->rx_flush_nic_set) &&
++ efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set)) {
++ list_add_tail(&virs->rx_flush_link,
++ &efrm_vi_manager->close_pending);
++ completed = true;
++ }
++
++ /* Issue the RX flush if possible or queue it for later. */
++ if (!efrm_nic_set_is_all_clear(&virs->rx_flush_nic_set)) {
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++ if (efrm_vi_manager->rx_flush_outstanding_count >=
++ flush_fifo_hwm)
++ efrm_vi_handle_flush_loss(&completed);
++#endif
++ if (efrm_vi_manager->rx_flush_outstanding_count >=
++ flush_fifo_hwm) {
++ list_add_tail(&virs->rx_flush_link,
++ &efrm_vi_manager->rx_flush_waiting_list);
++ } else {
++ efrm_vi_resource_issue_rx_flush(virs, &completed);
++ }
++ }
++
++ /* Issue the TX flush. There's no limit to the number of
++ * outstanding TX flushes. */
++ if (!efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set))
++ efrm_vi_resource_issue_tx_flush(virs, &completed);
++
++ virs->flush_time = get_jiffies_64();
++
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ if (completed)
++ EFRM_VI_RM_DELAYED_FREE(efrm_vi_manager);
++
++ return 0;
++}
++EXPORT_SYMBOL(efrm_pt_flush);
++
++static void
++efrm_handle_rx_dmaq_flushed(struct efhw_nic *flush_nic, int instance,
++ bool *completed)
++{
++ struct list_head *pos, *temp;
++ struct vi_resource *virs;
++
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->rx_flush_outstanding_list) {
++ virs = container_of(pos, struct vi_resource, rx_flush_link);
++
++ if (instance == EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle)) {
++ efrm_vi_resource_rx_flush_done(virs,
++ flush_nic->index,
++ completed);
++ efrm_vi_resource_process_waiting_flushes(completed);
++ return;
++ }
++ }
++ EFRM_TRACE("%s: Unhandled rx flush event, nic %d, instance %d",
++ __FUNCTION__, flush_nic->index, instance);
++}
++
++static void
++efrm_handle_tx_dmaq_flushed(struct efhw_nic *flush_nic, int instance,
++ bool *completed)
++{
++ struct list_head *pos, *temp;
++ struct vi_resource *virs;
++
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->tx_flush_outstanding_list) {
++ virs = container_of(pos, struct vi_resource, tx_flush_link);
++
++ if (instance == EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle)) {
++ efrm_vi_resource_tx_flush_done(virs,
++ flush_nic->index,
++ completed);
++ return;
++ }
++ }
++ EFRM_TRACE("%s: Unhandled tx flush event, nic %d, instance %d",
++ __FUNCTION__, flush_nic->index, instance);
++}
++
++void
++efrm_handle_dmaq_flushed(struct efhw_nic *flush_nic, int instance,
++ int rx_flush)
++{
++ irq_flags_t lock_flags;
++ bool completed = false;
++
++ EFRM_TRACE("%s: nic_i=%d instance=%d rx_flush=%d", __FUNCTION__,
++ flush_nic->index, instance, rx_flush);
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ if (rx_flush)
++ efrm_handle_rx_dmaq_flushed(flush_nic, instance, &completed);
++ else
++ efrm_handle_tx_dmaq_flushed(flush_nic, instance, &completed);
++
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++ efrm_vi_handle_flush_loss(&completed);
++#endif
++
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ if (completed)
++ EFRM_VI_RM_DELAYED_FREE(efrm_vi_manager);
++}
++
++static void
++efrm_vi_rm_reinit_dmaqs(struct vi_resource *virs)
++{
++ struct efhw_nic *nic;
++ int nic_i;
++
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic) {
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_TX] != 0)
++ efrm_vi_rm_init_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_TX,
++ nic);
++ if (virs->dmaq_capacity[EFRM_VI_RM_DMA_QUEUE_RX])
++ efrm_vi_rm_init_dmaq(virs, EFRM_VI_RM_DMA_QUEUE_RX,
++ nic);
++ }
++}
++
++/* free any PT endpoints whose flush has now complete */
++void efrm_vi_rm_delayed_free(struct work_struct *data)
++{
++ irq_flags_t lock_flags;
++ struct list_head close_pending;
++ struct vi_resource *virs;
++
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(&efrm_vi_manager->rm);
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++ list_replace_init(&efrm_vi_manager->close_pending, &close_pending);
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ EFRM_TRACE("%s: %p", __FUNCTION__, efrm_vi_manager);
++ while (!list_empty(&close_pending)) {
++ virs =
++ list_entry(list_pop(&close_pending), struct vi_resource,
++ rx_flush_link);
++ EFRM_TRACE("%s: flushed VI instance=%d", __FUNCTION__,
++ EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle));
++
++ if (virs->flush_callback_fn != NULL) {
++ efrm_vi_rm_reinit_dmaqs(virs);
++ virs->flush_callback_fn(virs->flush_callback_arg);
++ } else
++ efrm_vi_rm_free_flushed_resource(virs);
++ }
++}
++
++void efrm_vi_rm_salvage_flushed_vis(void)
++{
++#if BUG7916_WORKAROUND || BUG5302_WORKAROUND
++ irq_flags_t lock_flags;
++ bool completed;
++
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++ efrm_vi_handle_flush_loss(&completed);
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++#endif
++
++ efrm_vi_rm_delayed_free(&efrm_vi_manager->work_item);
++}
++
++void efrm_vi_resource_free(struct vi_resource *virs)
++{
++ efrm_vi_register_flush_callback(virs, NULL, NULL);
++ efrm_pt_flush(virs);
++}
++EXPORT_SYMBOL(efrm_vi_resource_free);
++
++/*
++ * vi: sw=8:ai:aw
++ */
+--- linux-2.6.18.8/drivers/net/sfc/sfc_resource/vi_resource_manager.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfc_resource/vi_resource_manager.c 2008-05-19 00:33:29.449843650 +0300
+@@ -0,0 +1,259 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains the VI resource manager.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <ci/efrm/nic_table.h>
++#include <ci/driver/efab/hardware.h>
++#include <ci/efhw/falcon.h>
++#include <ci/efrm/private.h>
++#include <ci/efrm/vi_resource_private.h>
++
++int efrm_pt_pace(struct vi_resource *virs, unsigned int val)
++{
++#if defined(__CI_HARDWARE_CONFIG_FALCON__)
++ int instance, nic_i;
++ struct efhw_nic *nic;
++
++ EFRM_RESOURCE_ASSERT_VALID(&virs->rs, 0);
++ instance = EFRM_RESOURCE_INSTANCE(virs->rs.rs_handle);
++
++ EFRM_FOR_EACH_NIC_IN_SET(&virs->nic_set, nic_i, nic)
++ falcon_nic_pace(nic, instance, val);
++
++ EFRM_TRACE("%s[%d]=%d DONE", __FUNCTION__, instance, val);
++ return 0;
++#else
++ return -EOPNOTSUPP;
++#endif
++}
++EXPORT_SYMBOL(efrm_pt_pace);
++
++/*** Resource manager creation/destruction *******************************/
++
++static void efrm_vi_rm_dtor(struct efrm_resource_manager *rm);
++
++static int
++efrm_create_or_destroy_vi_resource_manager(
++ struct efrm_resource_manager **rm_in_out,
++ const struct vi_resource_dimensions *dims,
++ bool destroy)
++{
++ struct vi_resource *virs;
++ struct list_head *pos, *temp;
++ struct list_head flush_pending;
++ irq_flags_t lock_flags;
++ int rc, i, n_evqs;
++ unsigned dmaq_min, dmaq_max;
++
++ EFRM_ASSERT(rm_in_out);
++
++ if (destroy)
++ goto destroy;
++
++ EFRM_ASSERT(dims);
++ EFRM_NOTICE("vi_resource_manager: evq_int=%u-%u evq_timer=%u-%u",
++ dims->evq_int_min, dims->evq_int_max,
++ dims->evq_timer_min, dims->evq_timer_max);
++ EFRM_NOTICE("vi_resource_manager: rxq=%u-%u txq=%u-%u",
++ dims->rxq_min, dims->rxq_max,
++ dims->txq_min, dims->txq_max);
++
++ efrm_vi_manager = kmalloc(sizeof(*efrm_vi_manager), GFP_KERNEL);
++ if (efrm_vi_manager == NULL) {
++ rc = -ENOMEM;
++ goto fail_alloc;
++ }
++
++ memset(efrm_vi_manager, 0, sizeof(*efrm_vi_manager));
++
++ efrm_vi_manager->iscsi_dmaq_instance_is_free = true;
++
++ dmaq_min = max(dims->rxq_min, dims->txq_min);
++ dmaq_max = min(dims->rxq_max, dims->txq_max);
++
++ efrm_vi_manager->with_timer_base =
++ max(dmaq_min, dims->evq_timer_min);
++ efrm_vi_manager->with_timer_limit =
++ min(dmaq_max, dims->evq_timer_max);
++ rc = efrm_kfifo_id_ctor(&efrm_vi_manager->instances_with_timer,
++ efrm_vi_manager->with_timer_base,
++ efrm_vi_manager->with_timer_limit,
++ &efrm_vi_manager->rm.rm_lock);
++ if (rc < 0)
++ goto fail_with_timer_id_pool;
++
++ efrm_vi_manager->with_interrupt_base =
++ max(dmaq_min, dims->evq_int_min);
++ efrm_vi_manager->with_interrupt_limit =
++ min(dmaq_max, dims->evq_int_max);
++ efrm_vi_manager->with_interrupt_limit =
++ max(efrm_vi_manager->with_interrupt_limit,
++ efrm_vi_manager->with_interrupt_base);
++ rc = efrm_kfifo_id_ctor(&efrm_vi_manager->instances_with_interrupt,
++ efrm_vi_manager->with_interrupt_base,
++ efrm_vi_manager->with_interrupt_limit,
++ &efrm_vi_manager->rm.rm_lock);
++ if (rc < 0)
++ goto fail_with_int_id_pool;
++
++ n_evqs = max(efrm_vi_manager->with_timer_limit,
++ efrm_vi_manager->with_interrupt_limit);
++ rc = -ENOMEM;
++ efrm_vi_manager->evq_infos =
++ vmalloc(n_evqs * sizeof(struct vi_resource_evq_info));
++ if (efrm_vi_manager->evq_infos == NULL)
++ goto fail_alloc_evq_infos;
++
++ for (i = 0; i < n_evqs; ++i) {
++ efrm_vi_manager->evq_infos[i].evq_state = 0;
++ efrm_vi_manager->evq_infos[i].evq_virs = NULL;
++ }
++
++ INIT_LIST_HEAD(&efrm_vi_manager->rx_flush_waiting_list);
++ INIT_LIST_HEAD(&efrm_vi_manager->rx_flush_outstanding_list);
++ INIT_LIST_HEAD(&efrm_vi_manager->tx_flush_outstanding_list);
++ efrm_vi_manager->rx_flush_outstanding_count = 0;
++
++ INIT_LIST_HEAD(&efrm_vi_manager->close_pending);
++#if EFRM_VI_USE_WORKQUEUE
++ efrm_vi_manager->workqueue = create_workqueue("sfc_vi");
++ if (efrm_vi_manager->workqueue == NULL)
++ goto fail_create_workqueue;
++#endif
++ INIT_WORK(&efrm_vi_manager->work_item, efrm_vi_rm_delayed_free);
++
++ /* NB. This must be the last step to avoid things getting tangled.
++ * efrm_resource_manager_dtor calls the vi_rm_dtor which ends up in
++ * this function. */
++ rc = efrm_resource_manager_ctor(&efrm_vi_manager->rm, efrm_vi_rm_dtor,
++ "VI", EFRM_RESOURCE_VI, 0);
++ if (rc < 0)
++ goto fail_rm_ctor;
++
++ *rm_in_out = &efrm_vi_manager->rm;
++ return 0;
++
++destroy:
++ rc = 0;
++ EFRM_RESOURCE_MANAGER_ASSERT_VALID(*rm_in_out);
++
++ /* Abort outstanding flushes. Note, a VI resource can be on more
++ * than one of these lists. We handle this by starting with the TX
++ * list and then append VIs to this list if they aren't on the TX
++ * list already. A VI is on the TX flush list if tx_flush_nic_set
++ * is not empty. */
++ spin_lock_irqsave(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ list_replace_init(&efrm_vi_manager->tx_flush_outstanding_list,
++ &flush_pending);
++
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->rx_flush_waiting_list) {
++ virs = container_of(pos, struct vi_resource, rx_flush_link);
++
++ list_del(&virs->rx_flush_link);
++ if (efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set))
++ list_add_tail(&virs->tx_flush_link, &flush_pending);
++ }
++
++ list_for_each_safe(pos, temp,
++ &efrm_vi_manager->rx_flush_outstanding_list) {
++ virs = container_of(pos, struct vi_resource, rx_flush_link);
++
++ list_del(&virs->rx_flush_link);
++ if (efrm_nic_set_is_all_clear(&virs->tx_flush_nic_set))
++ list_add_tail(&virs->tx_flush_link, &flush_pending);
++ }
++
++ spin_unlock_irqrestore(&efrm_vi_manager->rm.rm_lock, lock_flags);
++
++ while (!list_empty(&flush_pending)) {
++ virs =
++ list_entry(list_pop(&flush_pending), struct vi_resource,
++ tx_flush_link);
++ EFRM_TRACE("%s: found PT endpoint " EFRM_RESOURCE_FMT
++ " with flush pending [Tx=0x%x, Rx=0x%x, RxO=0x%x]",
++ __FUNCTION__,
++ EFRM_RESOURCE_PRI_ARG(virs->rs.rs_handle),
++ virs->tx_flush_nic_set.nics,
++ virs->rx_flush_nic_set.nics,
++ virs->rx_flush_outstanding_nic_set.nics);
++ efrm_vi_rm_free_flushed_resource(virs);
++ }
++
++fail_rm_ctor:
++
++ /* Complete outstanding closes. */
++#if EFRM_VI_USE_WORKQUEUE
++ destroy_workqueue(efrm_vi_manager->workqueue);
++fail_create_workqueue:
++#endif
++ EFRM_ASSERT(list_empty(&efrm_vi_manager->close_pending));
++
++ n_evqs = max(efrm_vi_manager->with_timer_limit,
++ efrm_vi_manager->with_interrupt_limit);
++ vfree(efrm_vi_manager->evq_infos);
++fail_alloc_evq_infos:
++
++ kfifo_vfree(efrm_vi_manager->instances_with_interrupt);
++fail_with_int_id_pool:
++
++ kfifo_vfree(efrm_vi_manager->instances_with_timer);
++fail_with_timer_id_pool:
++
++ if (destroy)
++ return 0;
++
++ EFRM_DO_DEBUG(memset(efrm_vi_manager, 0, sizeof(*efrm_vi_manager)));
++ kfree(efrm_vi_manager);
++fail_alloc:
++
++ *rm_in_out = NULL;
++ EFRM_ERR("%s: failed rc=%d", __FUNCTION__, rc);
++ return rc;
++}
++
++int
++efrm_create_vi_resource_manager(struct efrm_resource_manager **rm_out,
++ const struct vi_resource_dimensions *dims)
++{
++ return efrm_create_or_destroy_vi_resource_manager(rm_out, dims, false);
++}
++
++static void efrm_vi_rm_dtor(struct efrm_resource_manager *rm)
++{
++ efrm_create_or_destroy_vi_resource_manager(&rm, NULL, true);
++}
+--- linux-2.6.18.8/drivers/net/sfc/sfe4001.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/sfe4001.c 2008-05-19 00:33:29.449843650 +0300
+@@ -0,0 +1,315 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************/
++
++/*****************************************************************************
++ * Support for the SFE4001 NIC: driver code for the PCA9539 I/O expander that
++ * controls the PHY power rails, and for the MAX6647 temp. sensor used to check
++ * the PHY
++ */
++#include <linux/delay.h>
++#include "efx.h"
++#include "phy.h"
++#include "boards.h"
++#include "falcon.h"
++#include "falcon_hwdefs.h"
++
++/**************************************************************************
++ *
++ * I2C IO Expander device
++ *
++ **************************************************************************/
++#define PCA9539 0x74
++
++#define P0_IN 0x00
++#define P0_OUT 0x02
++#define P0_INVERT 0x04
++#define P0_CONFIG 0x06
++
++#define P0_EN_1V0X_LBN 0
++#define P0_EN_1V0X_WIDTH 1
++#define P0_EN_1V2_LBN 1
++#define P0_EN_1V2_WIDTH 1
++#define P0_EN_2V5_LBN 2
++#define P0_EN_2V5_WIDTH 1
++#define P0_EN_3V3X_LBN 3
++#define P0_EN_3V3X_WIDTH 1
++#define P0_EN_5V_LBN 4
++#define P0_EN_5V_WIDTH 1
++#define P0_SHORTEN_JTAG_LBN 5
++#define P0_SHORTEN_JTAG_WIDTH 1
++#define P0_X_TRST_LBN 6
++#define P0_X_TRST_WIDTH 1
++#define P0_DSP_RESET_LBN 7
++#define P0_DSP_RESET_WIDTH 1
++
++#define P1_IN 0x01
++#define P1_OUT 0x03
++#define P1_INVERT 0x05
++#define P1_CONFIG 0x07
++
++#define P1_AFE_PWD_LBN 0
++#define P1_AFE_PWD_WIDTH 1
++#define P1_DSP_PWD25_LBN 1
++#define P1_DSP_PWD25_WIDTH 1
++#define P1_RESERVED_LBN 2
++#define P1_RESERVED_WIDTH 2
++#define P1_SPARE_LBN 4
++#define P1_SPARE_WIDTH 4
++
++
++/**************************************************************************
++ *
++ * Temperature Sensor
++ *
++ **************************************************************************/
++#define MAX6647 0x4e
++
++#define RLTS 0x00
++#define RLTE 0x01
++#define RSL 0x02
++#define RCL 0x03
++#define RCRA 0x04
++#define RLHN 0x05
++#define RLLI 0x06
++#define RRHI 0x07
++#define RRLS 0x08
++#define WCRW 0x0a
++#define WLHO 0x0b
++#define WRHA 0x0c
++#define WRLN 0x0e
++#define OSHT 0x0f
++#define REET 0x10
++#define RIET 0x11
++#define RWOE 0x19
++#define RWOI 0x20
++#define HYS 0x21
++#define QUEUE 0x22
++#define MFID 0xfe
++#define REVID 0xff
++
++/* Status bits */
++#define MAX6647_BUSY (1 << 7) /* ADC is converting */
++#define MAX6647_LHIGH (1 << 6) /* Local high temp. alarm */
++#define MAX6647_LLOW (1 << 5) /* Local low temp. alarm */
++#define MAX6647_RHIGH (1 << 4) /* Remote high temp. alarm */
++#define MAX6647_RLOW (1 << 3) /* Remote low temp. alarm */
++#define MAX6647_FAULT (1 << 2) /* DXN/DXP short/open circuit */
++#define MAX6647_EOT (1 << 1) /* Remote junction overtemp. */
++#define MAX6647_IOT (1 << 0) /* Local junction overtemp. */
++
++static const u8 xgphy_max_temperature = 90;
++
++void sfe4001_poweroff(struct efx_nic *efx)
++{
++ struct efx_i2c_interface *i2c = &efx->i2c;
++
++ u8 cfg, out, in;
++
++ EFX_INFO(efx, "%s\n", __func__);
++
++ /* Turn off all power rails */
++ out = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P0_OUT, &out, EFX_BYTE);
++
++ /* Disable port 1 outputs on IO expander */
++ cfg = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P1_CONFIG, &cfg, EFX_BYTE);
++
++ /* Disable port 0 outputs on IO expander */
++ cfg = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P0_CONFIG, &cfg, EFX_BYTE);
++
++ /* Clear any over-temperature alert */
++ (void) efx_i2c_read(i2c, MAX6647, RSL, &in, EFX_BYTE);
++}
++
++static int sfe4001_check_hw(struct efx_nic *efx)
++{
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ int rc;
++ u8 status;
++
++ /* Check the powered status of the PHY. Lack of power implies that
++ * the MAX6647 has shut down power to it, probably due to a temp.
++ * alarm. Reading the power status rather than the MAX6647 status
++ * directly because the later is read-to-clear and would thus
++ * start to power up the PHY again when polled, causing us to blip
++ * the power undesirably */
++
++ /* If XAUI link is down, check power status. Reading
++ * power requires a I2C byte read, which is too slow
++ * to poll (see SFC bug 7884). */
++ if (falcon_xaui_link_ok(efx))
++ return 0;
++
++ rc = efx_i2c_read(i2c, PCA9539, P1_IN, &status, EFX_BYTE);
++ status &= ((1 << P1_AFE_PWD_LBN) | (1 << P1_DSP_PWD25_LBN));
++
++ /* We know we can read from the IO expander because we did
++ * it during power-on. Assume failure now is bad news. */
++ if (rc != 0 || status == 0) {
++ sfe4001_poweroff(efx);
++
++ /* Note that the PHY is pining for the cooling fans */
++ tenxpress_set_state(efx, TENXPRESS_STATUS_OTEMP);
++
++ /* Log the info */
++ if (status == 0) {
++ EFX_ERR(efx, "%s: Temperature sensor reports "
++ "alarm! (0x%x) Shutting down PHY.\n",
++ __func__, status);
++ rc = -EIO;
++ } else {
++ EFX_ERR(efx, "%s: Failed to read PHY status!"
++ " Shutting down PHY.\n",
++ __func__);
++ }
++ }
++
++ return rc;
++}
++
++/* This board uses an I2C expander to provider power to the PHY, which needs to
++ * be turned on before the PHY can be used.
++ * Context: Process context, rtnl lock held
++ */
++int sfe4001_poweron(struct efx_nic *efx)
++{
++ struct efx_i2c_interface *i2c = &efx->i2c;
++ unsigned int count;
++ int rc;
++ u8 out, in, cfg;
++ efx_dword_t reg;
++
++ /* 10Xpress has fixed-function LED pins, so there is no board-specific
++ * blink code. */
++ efx->board_info.blink = tenxpress_phy_blink;
++
++ /* Ensure that XGXS and XAUI SerDes are held in reset */
++ EFX_POPULATE_DWORD_7(reg, XX_PWRDNA_EN, 1,
++ XX_PWRDNB_EN, 1,
++ XX_RSTPLLAB_EN, 1,
++ XX_RESETA_EN, 1,
++ XX_RESETB_EN, 1,
++ XX_RSTXGXSRX_EN, 1,
++ XX_RSTXGXSTX_EN, 1);
++ efx->mac_op->mac_writel(efx, ®, XX_PWR_RST_REG_MAC);
++ udelay(10);
++
++ efx->board_info.monitor = sfe4001_check_hw;
++ efx->board_info.fini = sfe4001_poweroff;
++
++ /* Set DSP over-temperature alert threshold */
++ EFX_INFO(efx, "DSP cut-out at %dC\n", xgphy_max_temperature);
++ rc = efx_i2c_write(i2c, MAX6647, WLHO,
++ &xgphy_max_temperature, EFX_BYTE);
++ if (rc)
++ goto fail1;
++
++ /* Read it back and verify */
++ rc = efx_i2c_read(i2c, MAX6647, RLHN, &in, EFX_BYTE);
++ if (rc)
++ goto fail1;
++ if (in != xgphy_max_temperature) {
++ rc = -EFAULT;
++ goto fail1;
++ }
++
++ /* Clear any previous over-temperature alert */
++ rc = efx_i2c_read(i2c, MAX6647, RSL, &in, EFX_BYTE);
++ if (rc)
++ goto fail1;
++
++ /* Enable port 0 and port 1 outputs on IO expander */
++ cfg = 0x00;
++ rc = efx_i2c_write(i2c, PCA9539, P0_CONFIG, &cfg, EFX_BYTE);
++ if (rc)
++ goto fail1;
++ cfg = 0xff & ~(1 << P1_SPARE_LBN);
++ rc = efx_i2c_write(i2c, PCA9539, P1_CONFIG, &cfg, EFX_BYTE);
++ if (rc)
++ goto fail2;
++
++ /* Turn all power off then wait 1 sec. This ensures PHY is reset */
++ out = 0xff & ~((0 << P0_EN_1V2_LBN) | (0 << P0_EN_2V5_LBN) |
++ (0 << P0_EN_3V3X_LBN) | (0 << P0_EN_5V_LBN) |
++ (0 << P0_EN_1V0X_LBN));
++ rc = efx_i2c_write(i2c, PCA9539, P0_OUT, &out, EFX_BYTE);
++ if (rc)
++ goto fail3;
++
++ schedule_timeout_uninterruptible(HZ);
++ count = 0;
++ do {
++ /* Turn on 1.2V, 2.5V, 3.3V and 5V power rails */
++ out = 0xff & ~((1 << P0_EN_1V2_LBN) | (1 << P0_EN_2V5_LBN) |
++ (1 << P0_EN_3V3X_LBN) | (1 << P0_EN_5V_LBN) |
++ (1 << P0_X_TRST_LBN));
++
++ rc = efx_i2c_write(i2c, PCA9539, P0_OUT, &out, EFX_BYTE);
++ if (rc)
++ goto fail3;
++ msleep(10);
++
++ /* Turn on 1V power rail */
++ out &= ~(1 << P0_EN_1V0X_LBN);
++ rc = efx_i2c_write(i2c, PCA9539, P0_OUT, &out, EFX_BYTE);
++ if (rc)
++ goto fail3;
++
++ EFX_INFO(efx, "waiting for power (attempt %d)...\n", count);
++
++ schedule_timeout_uninterruptible(HZ);
++
++ /* Check DSP is powered */
++ rc = efx_i2c_read(i2c, PCA9539, P1_IN, &in, EFX_BYTE);
++ if (rc)
++ goto fail3;
++ if (in & (1 << P1_AFE_PWD_LBN))
++ goto done;
++
++ } while (++count < 20);
++
++ EFX_INFO(efx, "timed out waiting for power\n");
++ rc = -ETIMEDOUT;
++ goto fail3;
++
++done:
++ EFX_INFO(efx, "PHY is powered on\n");
++ return 0;
++
++fail3:
++ /* Turn off all power rails */
++ out = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P0_OUT, &out, EFX_BYTE);
++ /* Disable port 1 outputs on IO expander */
++ out = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P1_CONFIG, &out, EFX_BYTE);
++fail2:
++ /* Disable port 0 outputs on IO expander */
++ out = 0xff;
++ (void) efx_i2c_write(i2c, PCA9539, P1_CONFIG, &out, EFX_BYTE);
++fail1:
++ return rc;
++}
+--- linux-2.6.18.8/drivers/net/sfc/spi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/spi.h 2008-05-19 00:33:29.449843650 +0300
+@@ -0,0 +1,186 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005: Fen Systems Ltd.
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_SPI_H
++#define EFX_SPI_H
++
++#include "net_driver.h"
++
++/**************************************************************************
++ *
++ * Basic SPI command set and bit definitions
++ *
++ *************************************************************************/
++
++/*
++ * Commands common to all known devices.
++ *
++ */
++
++/* Write status register */
++#define SPI_WRSR 0x01
++
++/* Write data to memory array */
++#define SPI_WRITE 0x02
++
++/* Read data from memory array */
++#define SPI_READ 0x03
++
++/* Reset write enable latch */
++#define SPI_WRDI 0x04
++
++/* Read status register */
++#define SPI_RDSR 0x05
++
++/* Set write enable latch */
++#define SPI_WREN 0x06
++
++/* SST: Enable write to status register */
++#define SPI_SST_EWSR 0x50
++
++/*
++ * Status register bits. Not all bits are supported on all devices.
++ *
++ */
++
++/* Write-protect pin enabled */
++#define SPI_STATUS_WPEN 0x80
++
++/* Block protection bit 2 */
++#define SPI_STATUS_BP2 0x10
++
++/* Block protection bit 1 */
++#define SPI_STATUS_BP1 0x08
++
++/* Block protection bit 0 */
++#define SPI_STATUS_BP0 0x04
++
++/* State of the write enable latch */
++#define SPI_STATUS_WEN 0x02
++
++/* Device busy flag */
++#define SPI_STATUS_NRDY 0x01
++
++/**************************************************************************
++ *
++ * Efx SPI devices
++ *
++ **************************************************************************
++ */
++
++/**
++ * struct efx_spi_device - an Efx SPI (Serial Peripheral Interface) device
++ * @device_id: Controller's id for the device
++ * @size: Size (in bytes)
++ * @addr_len: Number of address bytes in read/write commands
++ * @munge_address: Flag whether addresses should be munged.
++ * Some devices with 9-bit addresses (e.g. AT25040A EEPROM)
++ * use bit 3 of the command byte as address bit A8, rather
++ * than having a two-byte address. If this flag is set, then
++ * commands should be munged in this way.
++ * @erase_command: Erase command (or 0 if sector erase not needed).
++ * @erase_size: Erase sector size (in bytes)
++ * Erase commands affect sectors with this size and alignment.
++ * This must be a power of two.
++ * @block_size: Write block size (in bytes).
++ * Write commands are limited to blocks with this size and alignment.
++ * @read: Read function for the device
++ * @write: Write function for the device
++ */
++struct efx_spi_device {
++ int device_id;
++ unsigned int size;
++ unsigned int addr_len;
++ unsigned int munge_address:1;
++ u8 erase_command;
++ unsigned int erase_size;
++ unsigned int block_size;
++ int (*read) (const struct efx_spi_device *spi,
++ struct efx_nic *efx, unsigned int command,
++ int address, void *data, unsigned int len);
++ int (*write) (const struct efx_spi_device *spi,
++ struct efx_nic *efx, unsigned int command,
++ int address, const void *data, unsigned int len);
++};
++
++/* Maximum length for SPI read or write through Falcon */
++#define FALCON_SPI_MAX_LEN 16U
++
++/**
++ * efx_spi_write_limit - calculate maximum permitted length for write
++ * @spi: SPI device description
++ * @start: Starting address
++ *
++ * Return the maximum length for a write starting at the given address
++ * in the device.
++ *
++ * SPI writes must not cross block boundaries. Devices tend
++ * to wrap addresses at block boundaries; e.g. trying to write 5 bytes
++ * starting at offset 14 with a block size of 16 might write
++ * {14,15,0,1,2} rather than {14,15,16,17,18}.
++ */
++static inline unsigned int
++efx_spi_write_limit(const struct efx_spi_device *spi, unsigned int start)
++{
++ return min(FALCON_SPI_MAX_LEN,
++ (spi->block_size - (start & (spi->block_size - 1))));
++}
++
++/**
++ * efx_spi_read_limit - calculate maximum permitted length for read
++ * @spi: SPI device description
++ * @start: Starting address
++ *
++ * Return the maximum length for a read starting at the given address
++ * in the device.
++ */
++static inline unsigned int
++efx_spi_read_limit(const struct efx_spi_device *spi __attribute__ ((unused)),
++ unsigned int start __attribute__ ((unused)))
++{
++ return FALCON_SPI_MAX_LEN;
++}
++
++/**
++ * efx_spi_munge_command - adjust command as necessary for given address
++ * @spi: SPI device description
++ * @command: Normal SPI command
++ * @address: Address for command
++ *
++ * Some devices with 9-bit addresses (e.g. AT25040A EEPROM) use bit 3
++ * of the command byte as address bit A8, rather than having a
++ * two-byte address. This function calculates the appropriate command
++ * byte for the device, taking this munging into account.
++ */
++static inline u8 efx_spi_munge_command(const struct efx_spi_device *spi,
++ const u8 command,
++ const unsigned int address)
++{
++ return (command | (((address >> 8) & spi->munge_address) << 3));
++}
++
++#endif /* EFX_SPI_H */
+--- linux-2.6.18.8/drivers/net/sfc/tenxpress.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/tenxpress.c 2008-05-19 00:33:29.449843650 +0300
+@@ -0,0 +1,697 @@
++/****************************************************************************
++ * Driver for Solarflare 802.3an compliant PHY
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************/
++
++#include <linux/delay.h>
++#include <linux/seq_file.h>
++#include "efx.h"
++#include "debugfs.h"
++#include "gmii.h"
++#include "mdio_10g.h"
++#include "falcon.h"
++#include "phy.h"
++#include "falcon_hwdefs.h"
++#include "boards.h"
++
++/* We expect these MMDs to be in the package */
++/* AN not here as mdio_check_mmds() requires STAT2 support */
++#define TENXPRESS_REQUIRED_DEVS (MDIO_MMDREG_DEVS0_PMAPMD | \
++ MDIO_MMDREG_DEVS0_PCS | \
++ MDIO_MMDREG_DEVS0_PHYXS)
++
++#define TENXPRESS_LOOPBACKS ((1 << LOOPBACK_PHYXS) | \
++ (1 << LOOPBACK_PCS) | \
++ (1 << LOOPBACK_PMAPMD) | \
++ (1 << LOOPBACK_NETWORK))
++
++/* We complain if we fail to see the link partner as 10G capable this many
++ * times in a row (must be > 1 as sampling the autoneg. registers is racy)
++ */
++#define MAX_BAD_LP_TRIES (5)
++
++/* SNR operating margin register */
++#define PMA_PMD_SNR_MARGIN_0 (133)
++#define PMA_PMD_SNR_MARGIN_1 (134)
++#define PMA_PMD_SNR_MARGIN_2 (135)
++#define PMA_PMD_SNR_MARGIN_3 (136)
++
++/* Extended control register */
++#define PMA_PMD_XCONTROL_REG 0xc000
++#define PMA_PMD_LNPGA_POWERDOWN_LBN 8
++#define PMA_PMD_LNPGA_POWERDOWN_WIDTH 1
++#define PMA_PMD_AFE_POWERDOWN_LBN 9
++#define PMA_PMD_AFE_POWERDOWN_WIDTH 1
++#define PMA_PMD_DSP_POWERDOWN_LBN 10
++#define PMA_PMD_DSP_POWERDOWN_WIDTH 1
++#define PMA_PMD_PHY_POWERDOWN_LBN 11
++#define PMA_PMD_PHY_POWERDOWN_WI
++
++/* extended status register */
++#define PMA_PMD_XSTATUS_REG 0xc001
++#define PMA_PMD_XSTAT_FLP_LBN (12)
++
++
++/* LED control register */
++#define PMA_PMD_LED_CTRL_REG (0xc007)
++#define PMA_PMA_LED_ACTIVITY_LBN (3)
++
++/* LED function override register */
++#define PMA_PMD_LED_OVERR_REG (0xc009)
++/* Bit positions for different LEDs (there are more but not wired on SFE4001)*/
++#define PMA_PMD_LED_LINK_LBN (0)
++#define PMA_PMD_LED_SPEED_LBN (2)
++#define PMA_PMD_LED_TX_LBN (4)
++#define PMA_PMD_LED_RX_LBN (6)
++/* Override settings */
++#define PMA_PMD_LED_AUTO (0) /* H/W control */
++#define PMA_PMD_LED_ON (1)
++#define PMA_PMD_LED_OFF (2)
++#define PMA_PMD_LED_FLASH (3)
++/* All LEDs under hardware control */
++#define PMA_PMD_LED_FULL_AUTO (0)
++/* Green and Amber under hardware control, Red off */
++#define PMA_PMD_LED_DEFAULT (PMA_PMD_LED_OFF << PMA_PMD_LED_RX_LBN)
++
++
++/* Self test (BIST) control register */
++#define PMA_PMD_BIST_CTRL_REG (0xc014)
++#define PMA_PMD_BIST_BER_LBN (2) /* Run BER test */
++#define PMA_PMD_BIST_CONT_LBN (1) /* Run continuous BIST until cleared */
++#define PMA_PMD_BIST_SINGLE_LBN (0) /* Run 1 BIST iteration (self clears) */
++/* Self test status register */
++#define PMA_PMD_BIST_STAT_REG (0xc015)
++#define PMA_PMD_BIST_ENX_LBN (3)
++#define PMA_PMD_BIST_PMA_LBN (2)
++#define PMA_PMD_BIST_RXD_LBN (1)
++#define PMA_PMD_BIST_AFE_LBN (0)
++
++/* Special Software reset register */
++#define PMA_PMD_EXT_CTRL_REG 49152
++#define PMA_PMD_EXT_SSR_LBN 15
++
++#define BIST_MAX_DELAY (1000)
++#define BIST_POLL_DELAY (10)
++
++static const char *bist_names[] = {
++ [PMA_PMD_BIST_AFE_LBN] = "AFE communication",
++ [PMA_PMD_BIST_RXD_LBN] = "RX data path",
++ [PMA_PMD_BIST_PMA_LBN] = "PMA loopback",
++ [PMA_PMD_BIST_ENX_LBN] = "ENX"
++};
++
++/* Identifier registers: each identifier has 4 part number and 2 revision
++ * registers starting at one of these addresses */
++#define PMA_PMD_AFE_ID_REG 49174
++#define PMA_PMD_DSP_ID_REG 49180
++#define PMA_PMD_FIRMWARE_ID_REG 49186
++
++/* Misc register defines */
++#define PCS_CLOCK_CTRL_REG 0xd801
++#define PLL312_RST_N_LBN 2
++
++#define PCS_SOFT_RST2_REG 0xd806
++#define SERDES_RST_N_LBN 13
++#define XGXS_RST_N_LBN 12
++
++#define PCS_TEST_SELECT_REG 0xd807 /* PRM 10.5.8 */
++#define CLK312_EN_LBN 3
++
++/* PHYXS registers */
++#define PHYXS_TEST1 (49162)
++#define LOOPBACK_NEAR_LBN (8)
++#define LOOPBACK_NEAR_WIDTH (1)
++
++/* Boot status register */
++#define PCS_BOOT_STATUS_REG (0xd000)
++#define PCS_BOOT_FATAL_ERR_LBN (0)
++#define PCS_BOOT_PROGRESS_LBN (1)
++#define PCS_BOOT_PROGRESS_WIDTH (2)
++#define PCS_BOOT_COMPLETE_LBN (3)
++
++#define PCS_BOOT_MAX_DELAY (100)
++#define PCS_BOOT_POLL_DELAY (10)
++
++#define TENXPRESS_ID_PN_LEN (8)
++#define TENXPRESS_ID_REV_LEN (4)
++#define TENXPRESS_ID_LEN (TENXPRESS_ID_PN_LEN+1+TENXPRESS_ID_REV_LEN)
++
++static const int bist_max = ARRAY_SIZE(bist_names);
++
++/* Time to wait between powering down the LNPGA and turning off the power
++ * rails */
++#define LNPGA_PDOWN_WAIT (HZ / 5)
++
++
++static int crc_error_reset_threshold = 100;
++module_param(crc_error_reset_threshold, int, 0644);
++MODULE_PARM_DESC(crc_error_reset_threshold,
++ "Max number of CRC errors before XAUI reset");
++
++struct tenxpress_phy_data {
++#ifdef CONFIG_SFC_DEBUGFS
++ char phy_snr[4];
++ char phy_afe_id[TENXPRESS_ID_LEN + 1];
++ char phy_dsp_id[TENXPRESS_ID_LEN + 1];
++ char phy_firmware_id[TENXPRESS_ID_LEN + 1];
++ struct efx_nic *efx;
++#endif
++ enum tenxpress_state state;
++ enum efx_loopback_mode loopback_mode;
++ atomic_t bad_crc_count;
++ int phy_powered;
++ int tx_disabled;
++ int bad_lp_tries;
++};
++
++static int tenxpress_state_is(struct efx_nic *efx, int state)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ return (phy_data != NULL) && (state == phy_data->state);
++}
++
++void tenxpress_set_state(struct efx_nic *efx,
++ enum tenxpress_state state)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ if (phy_data != NULL)
++ phy_data->state = state;
++}
++
++void tenxpress_crc_err(struct efx_nic *efx)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ if (phy_data != NULL)
++ atomic_inc(&phy_data->bad_crc_count);
++}
++
++#ifdef CONFIG_SFC_DEBUGFS
++
++/* debugfs entries for this PHY */
++static int tenxpress_ber_read(struct seq_file *file, void *data)
++{
++ struct efx_nic *efx = *(struct efx_nic **)data;
++ int reg, ber;
++
++ reg = mdio_clause45_read(efx, efx->mii.phy_id, MDIO_MMD_PCS,
++ MDIO_PCS_10GBT_STATUS2);
++
++ /* Extract the BER */
++ ber = (reg >> MDIO_PCS_10GBT_STATUS2_BER_LBN) &
++ ((1 << MDIO_PCS_10GBT_STATUS2_BER_WIDTH) - 1);
++
++ return seq_printf(file, "%d", ber);
++}
++
++
++static int tenxpress_snr_read(struct seq_file *file, void *data)
++{
++ struct tenxpress_phy_data *phy_data = NULL;
++ struct efx_nic *efx;
++ int lane = *(char *) data;
++ int reg, snr;
++
++ EFX_BUG_ON_PARANOID(lane < 0 || lane >= 4);
++ phy_data = container_of(data, struct tenxpress_phy_data, phy_snr[lane]);
++ efx = phy_data->efx;
++
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, PMA_PMD_SNR_MARGIN_0 + lane);
++
++ /* Convert from SNR margin to SNR to match phychk output */
++ snr = (reg - 0x8000 + 238);
++
++ return seq_printf(file, "%d.%d", snr / 10, snr % 10);
++}
++
++
++static struct efx_debugfs_parameter debug_entries[] = {
++ EFX_PER_LANE_PARAMETER("phy_lane", "_snr",
++ struct tenxpress_phy_data, phy_snr, char,
++ tenxpress_snr_read),
++ EFX_NAMED_PARAMETER(phy_ber, struct tenxpress_phy_data, efx,
++ struct efx_nic *, tenxpress_ber_read),
++ EFX_STRING_PARAMETER(struct tenxpress_phy_data, phy_afe_id),
++ EFX_STRING_PARAMETER(struct tenxpress_phy_data, phy_dsp_id),
++ EFX_STRING_PARAMETER(struct tenxpress_phy_data, phy_firmware_id),
++ {NULL}
++};
++
++static void tenxpress_phy_get_id(struct efx_nic *efx,
++ char *id_buf, int id_addr)
++{
++ int i, reg;
++ char ch;
++
++ for (i = TENXPRESS_ID_PN_LEN / 2 - 1; i >= 0; --i) {
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, id_addr + i);
++ ch = reg & 0xFF;
++ *id_buf++ = ch ? ch : ' ';
++ ch = (reg & 0xFF00) >> 8;
++ *id_buf++ = ch ? ch : ' ';
++ }
++ *id_buf++ = ' ';
++ for (i = TENXPRESS_ID_REV_LEN / 2 - 1; i >= 0; --i) {
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD,
++ id_addr + TENXPRESS_ID_PN_LEN / 2 + i);
++ ch = reg & 0xFF;
++ *id_buf++ = ch ? ch : ' ';
++ ch = (reg & 0xFF00) >> 8;
++ *id_buf++ = ch ? ch : ' ';
++ }
++}
++
++static int tenxpress_debugfs_init(struct efx_nic *efx)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ int lane, rc;
++
++ for (lane = 0; lane < 4; lane++)
++ phy_data->phy_snr[lane] = lane;
++
++ phy_data->efx = efx;
++ rc = efx_extend_debugfs_port(efx, efx->phy_data,
++ debug_entries);
++ if (rc < 0)
++ return rc;
++
++ tenxpress_phy_get_id(efx, phy_data->phy_afe_id,
++ PMA_PMD_AFE_ID_REG);
++ tenxpress_phy_get_id(efx, phy_data->phy_dsp_id,
++ PMA_PMD_DSP_ID_REG);
++ tenxpress_phy_get_id(efx, phy_data->phy_firmware_id,
++ PMA_PMD_FIRMWARE_ID_REG);
++
++ return 0;
++}
++
++#endif /* CONFIG_SFC_DEBUGFS */
++
++/* Check that the C166 has booted successfully */
++static int tenxpress_phy_check(struct efx_nic *efx)
++{
++ int phy_id = efx->mii.phy_id;
++ int count = PCS_BOOT_MAX_DELAY / PCS_BOOT_POLL_DELAY;
++ int boot_stat;
++
++ /* Wait for the boot to complete (or not) */
++ while (count) {
++ boot_stat = mdio_clause45_read(efx, phy_id,
++ MDIO_MMD_PCS,
++ PCS_BOOT_STATUS_REG);
++ if (boot_stat & (1 << PCS_BOOT_COMPLETE_LBN))
++ break;
++ count--;
++ udelay(PCS_BOOT_POLL_DELAY);
++ }
++
++ if (!count) {
++ EFX_ERR(efx, "%s: PHY boot timed out. Last status "
++ "%x\n", __func__,
++ (boot_stat >> PCS_BOOT_PROGRESS_LBN) &
++ ((1 << PCS_BOOT_PROGRESS_WIDTH) - 1));
++ return -ETIMEDOUT;
++ }
++
++ return 0;
++}
++
++static void tenxpress_reset_xaui(struct efx_nic *efx);
++
++/* Initialise the part post power on reset or software special reset */
++static int tenxpress_init(struct efx_nic *efx)
++{
++ int rc, reg;
++
++ /* Turn on the clock */
++ reg = (1 << CLK312_EN_LBN);
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_PCS, PCS_TEST_SELECT_REG, reg);
++
++ rc = tenxpress_phy_check(efx);
++ if (rc < 0)
++ return rc;
++
++ /* Set the LEDs up as: Green = Link, Amber = Link/Act, Red = Off */
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, PMA_PMD_LED_CTRL_REG);
++ reg |= (1 << PMA_PMA_LED_ACTIVITY_LBN);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_LED_CTRL_REG, reg);
++
++ reg = PMA_PMD_LED_DEFAULT;
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_LED_OVERR_REG, reg);
++
++ return rc;
++}
++
++static int tenxpress_phy_init(struct efx_nic *efx)
++{
++ struct tenxpress_phy_data *phy_data;
++ int rc = 0;
++
++ phy_data = kzalloc(sizeof(*phy_data), GFP_KERNEL);
++ efx->phy_data = phy_data;
++ phy_data->phy_powered = efx->phy_powered;
++
++ tenxpress_set_state(efx, TENXPRESS_STATUS_NORMAL);
++
++ rc = mdio_clause45_wait_reset_mmds(efx,
++ TENXPRESS_REQUIRED_DEVS);
++ if (rc < 0)
++ goto fail;
++
++ rc = mdio_clause45_check_mmds(efx, TENXPRESS_REQUIRED_DEVS, 0);
++ if (rc < 0)
++ goto fail;
++
++ rc = tenxpress_init(efx);
++ if (rc < 0)
++ goto fail;
++
++#ifdef CONFIG_SFC_DEBUGFS
++ rc = tenxpress_debugfs_init(efx);
++ if (rc < 0)
++ goto fail;
++#endif
++
++ schedule_timeout_uninterruptible(HZ / 5); /* 200ms */
++
++ /* Let XGXS and SerDes out of reset and resets 10XPress */
++ falcon_reset_xaui(efx);
++
++ return 0;
++
++ fail:
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++ return rc;
++}
++
++static int tenxpress_special_reset(struct efx_nic *efx)
++{
++ int rc, reg;
++
++ EFX_TRACE(efx, "%s\n", __func__);
++
++ /* Initiate reset */
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, PMA_PMD_EXT_CTRL_REG);
++ reg |= (1 << PMA_PMD_EXT_SSR_LBN);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_EXT_CTRL_REG, reg);
++
++ msleep(200);
++
++ /* Wait for the blocks to come out of reset */
++ rc = mdio_clause45_wait_reset_mmds(efx,
++ TENXPRESS_REQUIRED_DEVS);
++ if (rc < 0)
++ return rc;
++
++ /* Try and reconfigure the device */
++ rc = tenxpress_init(efx);
++ if (rc < 0)
++ return rc;
++
++ return 0;
++}
++
++static void tenxpress_set_bad_lp(struct efx_nic *efx, int bad_lp)
++{
++ struct tenxpress_phy_data *pd = efx->phy_data;
++ int reg;
++
++ /* Nothing to do if all is well and was previously so. */
++ if (!(bad_lp || pd->bad_lp_tries))
++ return;
++
++ reg = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, PMA_PMD_LED_OVERR_REG);
++
++ if (bad_lp)
++ pd->bad_lp_tries++;
++ else
++ pd->bad_lp_tries = 0;
++
++ if (pd->bad_lp_tries == MAX_BAD_LP_TRIES) {
++ pd->bad_lp_tries = 0; /* Restart count */
++ reg &= ~(PMA_PMD_LED_FLASH << PMA_PMD_LED_RX_LBN);
++ reg |= (PMA_PMD_LED_FLASH << PMA_PMD_LED_RX_LBN);
++ EFX_ERR(efx, "This NIC appears to be plugged into"
++ " a port that is not 10GBASE-T capable.\n"
++ " This PHY is 10GBASE-T ONLY, so no link can"
++ " be established.\n");
++ } else {
++ reg |= (PMA_PMD_LED_OFF << PMA_PMD_LED_RX_LBN);
++ }
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_LED_OVERR_REG, reg);
++}
++
++/* Check link status and return a boolean OK value. If the link is NOT
++ * OK we have a quick rummage round to see if we appear to be plugged
++ * into a non-10GBT port and if so warn the user that they won't get
++ * link any time soon as we are 10GBT only, unless caller specified
++ * not to do this check (it isn't useful in loopback) */
++static int tenxpress_link_ok(struct efx_nic *efx, int check_lp)
++{
++ int ok = mdio_clause45_links_ok(efx, TENXPRESS_REQUIRED_DEVS);
++
++ if (ok) {
++ tenxpress_set_bad_lp(efx, 0);
++ } else if (check_lp) {
++ /* Are we plugged into the wrong sort of link? */
++ int bad_lp = 0;
++ int phy_id = efx->mii.phy_id;
++ int an_stat = mdio_clause45_read(efx, phy_id, MDIO_MMD_AN,
++ MDIO_AN_STATUS);
++ int xphy_stat = mdio_clause45_read(efx, phy_id,
++ MDIO_MMD_PMAPMD,
++ PMA_PMD_XSTATUS_REG);
++ /* Are we plugged into anything that sends FLPs? If
++ * not we can't distinguish between not being plugged
++ * in and being plugged into a non-AN antique. The FLP
++ * bit has the advantage of not clearing when autoneg
++ * restarts. */
++ if (!(xphy_stat & (1 << PMA_PMD_XSTAT_FLP_LBN))) {
++ tenxpress_set_bad_lp(efx, 0);
++ return ok;
++ }
++
++ /* If it can do 10GBT it must be XNP capable */
++ bad_lp = !(an_stat & (1 << MDIO_AN_STATUS_XNP_LBN));
++ if (!bad_lp && (an_stat & (1 << MDIO_AN_STATUS_PAGE_LBN))) {
++ bad_lp = !(mdio_clause45_read(efx, phy_id,
++ MDIO_MMD_AN, MDIO_AN_10GBT_STATUS) &
++ (1 << MDIO_AN_10GBT_STATUS_LP_10G_LBN));
++ }
++ tenxpress_set_bad_lp(efx, bad_lp);
++ }
++ return ok;
++}
++
++static void tenxpress_phyxs_loopback(struct efx_nic *efx)
++{
++ int phy_id = efx->mii.phy_id;
++ int ctrl1, ctrl2;
++
++ ctrl1 = ctrl2 = mdio_clause45_read(efx, phy_id, MDIO_MMD_PHYXS,
++ PHYXS_TEST1);
++ if (efx->loopback_mode == LOOPBACK_PHYXS)
++ ctrl2 |= (1 << LOOPBACK_NEAR_LBN);
++ else
++ ctrl2 &= ~(1 << LOOPBACK_NEAR_LBN);
++ if (ctrl1 != ctrl2)
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PHYXS,
++ PHYXS_TEST1, ctrl2);
++}
++
++static void tenxpress_phy_reconfigure(struct efx_nic *efx)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ int loop_change = LOOPBACK_OUT_OF(phy_data, efx,
++ TENXPRESS_LOOPBACKS);
++
++ if (!tenxpress_state_is(efx, TENXPRESS_STATUS_NORMAL))
++ return;
++
++ /* When coming out of transmit disable, coming out of low power
++ * mode, or moving out of any PHY internal loopback mode,
++ * perform a special software reset */
++ if (((efx->phy_powered && !efx->tx_disabled) &&
++ (!phy_data->phy_powered || phy_data->tx_disabled)) ||
++ loop_change) {
++ (void) tenxpress_special_reset(efx);
++ falcon_reset_xaui(efx);
++ }
++
++ mdio_clause45_transmit_disable(efx, efx->tx_disabled);
++ mdio_clause45_phy_reconfigure(efx);
++ tenxpress_phyxs_loopback(efx);
++
++ phy_data->tx_disabled = efx->tx_disabled;
++ phy_data->loopback_mode = efx->loopback_mode;
++ phy_data->phy_powered = efx->phy_powered;
++ efx->link_up = tenxpress_link_ok(efx, 0);
++ efx->link_options = GM_LPA_10000FULL;
++}
++
++static void tenxpress_phy_clear_interrupt(struct efx_nic *efx)
++{
++ /* Nothing done here - LASI interrupts aren't reliable so poll */
++}
++
++
++/* Poll PHY for interrupt */
++static int tenxpress_phy_check_hw(struct efx_nic *efx)
++{
++ struct tenxpress_phy_data *phy_data = efx->phy_data;
++ int phy_up = tenxpress_state_is(efx, TENXPRESS_STATUS_NORMAL);
++ int link_ok, rc = 0;
++
++ link_ok = phy_up && tenxpress_link_ok(efx, 1);
++
++ if (link_ok != efx->link_up) {
++ efx->link_up = link_ok;
++ efx->mac_op->fake_phy_event(efx);
++ }
++
++ /* Nothing to check if we've already shut down the PHY */
++ if (!phy_up)
++ return 0;
++
++ if (atomic_read(&phy_data->bad_crc_count) > crc_error_reset_threshold) {
++ EFX_ERR(efx, "Resetting XAUI due to too many CRC errors\n");
++ falcon_reset_xaui(efx);
++ atomic_set(&phy_data->bad_crc_count, 0);
++ }
++
++ rc = efx->board_info.monitor(efx);
++ if (rc)
++ efx->link_up = 0;
++
++ return rc;
++}
++
++static void tenxpress_phy_fini(struct efx_nic *efx)
++{
++ int reg;
++
++#ifdef CONFIG_SFC_DEBUGFS
++ efx_trim_debugfs_port(efx, debug_entries);
++#endif
++ /* Power down the LNPGA */
++ reg = (1 << PMA_PMD_LNPGA_POWERDOWN_LBN);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_XCONTROL_REG, reg);
++
++ /* Waiting here ensures that the board fini, which can turn off the
++ * power to the PHY, won't get run until the LNPGA powerdown has been
++ * given long enough to complete. */
++ schedule_timeout_uninterruptible(LNPGA_PDOWN_WAIT); /* 200 ms */
++
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++}
++
++
++/* Set the RX and TX LEDs and Link LED flashing. The other LEDs
++ * (which probably aren't wired anyway) are left in AUTO mode */
++void tenxpress_phy_blink(struct efx_nic *efx, int blink)
++{
++ int reg;
++
++ if (blink)
++ reg = (PMA_PMD_LED_FLASH << PMA_PMD_LED_TX_LBN) |
++ (PMA_PMD_LED_FLASH << PMA_PMD_LED_RX_LBN) |
++ (PMA_PMD_LED_FLASH << PMA_PMD_LED_LINK_LBN);
++ else
++ reg = PMA_PMD_LED_DEFAULT;
++
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ PMA_PMD_LED_OVERR_REG, reg);
++}
++
++static void tenxpress_reset_xaui(struct efx_nic *efx)
++{
++ int phy = efx->mii.phy_id;
++ int clk_ctrl, test_select, soft_rst2;
++
++ /* Real work is done on clock_ctrl other resets are thought to be
++ * optional but make the reset more reliable
++ */
++
++ /* Read */
++ clk_ctrl = mdio_clause45_read(efx, phy, MDIO_MMD_PCS,
++ PCS_CLOCK_CTRL_REG);
++ test_select = mdio_clause45_read(efx, phy, MDIO_MMD_PCS,
++ PCS_TEST_SELECT_REG);
++ soft_rst2 = mdio_clause45_read(efx, phy, MDIO_MMD_PCS,
++ PCS_SOFT_RST2_REG);
++
++ /* Modify => put in reset */
++ test_select &= ~(1 << CLK312_EN_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_TEST_SELECT_REG, test_select);
++
++ soft_rst2 &= ~((1 << XGXS_RST_N_LBN) | (1 << SERDES_RST_N_LBN));
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_SOFT_RST2_REG, soft_rst2);
++
++ clk_ctrl &= ~(1 << PLL312_RST_N_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_CLOCK_CTRL_REG, clk_ctrl);
++ udelay(10);
++
++ /* Modify => remove reset */
++ clk_ctrl |= (1 << PLL312_RST_N_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_CLOCK_CTRL_REG, clk_ctrl);
++ udelay(10);
++
++ soft_rst2 |= ((1 << XGXS_RST_N_LBN) | (1 << SERDES_RST_N_LBN));
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_SOFT_RST2_REG, soft_rst2);
++ udelay(10);
++
++ test_select |= (1 << CLK312_EN_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ PCS_TEST_SELECT_REG, test_select);
++ udelay(10);
++}
++
++
++struct efx_phy_operations falcon_tenxpress_phy_ops = {
++ .init = tenxpress_phy_init,
++ .reconfigure = tenxpress_phy_reconfigure,
++ .check_hw = tenxpress_phy_check_hw,
++ .fini = tenxpress_phy_fini,
++ .clear_interrupt = tenxpress_phy_clear_interrupt,
++ .reset_xaui = tenxpress_reset_xaui,
++ .mmds = TENXPRESS_REQUIRED_DEVS,
++ .loopbacks = TENXPRESS_LOOPBACKS,
++ .startup_loopback = LOOPBACK_PCS,
++};
+--- linux-2.6.18.8/drivers/net/sfc/tx.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/tx.c 2008-05-19 00:33:29.453843880 +0300
+@@ -0,0 +1,522 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2005-2006: Fen Systems Ltd.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include <linux/pci.h>
++#include <linux/tcp.h>
++#include <linux/ip.h>
++#include <linux/in.h>
++#include <linux/if_ether.h>
++#include <linux/version.h>
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++#include <linux/highmem.h>
++#endif
++#include "net_driver.h"
++#include "tx.h"
++#include "efx.h"
++#include "falcon.h"
++#include "workarounds.h"
++
++
++/*
++ * TX descriptor ring full threshold
++ *
++ * The tx_queue descriptor ring fill-level must fall below this value
++ * before we restart the netif queue
++ */
++#define EFX_NETDEV_TX_THRESHOLD(_tx_queue) \
++ (_tx_queue->efx->type->txd_ring_mask / 2u)
++
++
++
++/* We want to be able to nest calls to netif_stop_queue(), since each
++ * channel can have an individual stop on the queue.
++ */
++void efx_stop_queue(struct efx_nic *efx)
++{
++ spin_lock_bh(&efx->netif_stop_lock);
++ EFX_TRACE(efx, "stop TX queue\n");
++
++ atomic_inc(&efx->netif_stop_count);
++ if (likely(efx->net_dev_registered))
++ netif_stop_queue(efx->net_dev);
++
++ spin_unlock_bh(&efx->netif_stop_lock);
++}
++
++/* Wake netif's TX queue
++ * We want to be able to nest calls to netif_stop_queue(), since each
++ * channel can have an individual stop on the queue.
++ */
++inline void efx_wake_queue(struct efx_nic *efx)
++{
++ local_bh_disable();
++ if (atomic_dec_and_lock(&efx->netif_stop_count,
++ &efx->netif_stop_lock)) {
++ EFX_TRACE(efx, "waking TX queue\n");
++ if (likely(efx->net_dev_registered))
++ netif_wake_queue(efx->net_dev);
++ spin_unlock(&efx->netif_stop_lock);
++ }
++ local_bh_enable();
++}
++
++/*
++ * Add a socket buffer to a TX queue
++ *
++ * This maps all fragments of a socket buffer for DMA and adds them to
++ * the TX queue. The queue's insert pointer will be incremented by
++ * the number of fragments in the socket buffer.
++ *
++ * If any DMA mapping fails, any mapped fragments will be unmapped,
++ * the queue's insert pointer will be restored to its original value.
++ *
++ * Returns NETDEV_TX_OK or NETDEV_TX_BUSY
++ * You must hold netif_tx_lock() to call this function.
++ */
++static inline int efx_enqueue_skb(struct efx_tx_queue *tx_queue,
++ const struct sk_buff *skb)
++{
++ struct efx_nic *efx = tx_queue->efx;
++ struct pci_dev *pci_dev = efx->pci_dev;
++ struct efx_tx_buffer *buffer;
++ skb_frag_t *fragment;
++ struct page *page;
++ int page_offset;
++ unsigned int len, unmap_len = 0, fill_level, insert_ptr, misalign;
++ dma_addr_t dma_addr, unmap_addr = 0;
++ unsigned int dma_len;
++ unsigned unmap_single;
++ int q_space, i = 0;
++ int rc = NETDEV_TX_OK;
++
++ EFX_BUG_ON_PARANOID(tx_queue->write_count != tx_queue->insert_count);
++
++ /* Get size of the initial fragment */
++ len = skb_headlen(skb);
++
++ fill_level = tx_queue->insert_count - tx_queue->old_read_count;
++ q_space = efx->type->txd_ring_mask - 1 - fill_level;
++
++ /* Map for DMA. Use pci_map_single rather than pci_map_page
++ * since this is more efficient on machines with sparse
++ * memory.
++ */
++ unmap_single = 1;
++ dma_addr = pci_map_single(pci_dev, skb->data, len, PCI_DMA_TODEVICE);
++
++ /* Process all fragments */
++ while (1) {
++ if (unlikely(pci_dma_mapping_error(dma_addr)))
++ goto pci_err;
++
++ /* Store fields for marking in the per-fragment final
++ * descriptor */
++ unmap_len = len;
++ unmap_addr = dma_addr;
++
++ /* Add to TX queue, splitting across DMA boundaries */
++ do {
++ if (unlikely(q_space-- <= 0)) {
++ /* It might be that completions have
++ * happened since the xmit path last
++ * checked. Update the xmit path's
++ * copy of read_count.
++ */
++ ++tx_queue->stopped;
++ /* This memory barrier protects the
++ * change of stopped from the access
++ * of read_count. */
++ smp_mb();
++ tx_queue->old_read_count =
++ *(volatile unsigned *)
++ &tx_queue->read_count;
++ fill_level = (tx_queue->insert_count
++ - tx_queue->old_read_count);
++ q_space = (efx->type->txd_ring_mask - 1 -
++ fill_level);
++ if (unlikely(q_space-- <= 0))
++ goto stop;
++ smp_mb();
++ --tx_queue->stopped;
++ }
++
++ insert_ptr = (tx_queue->insert_count &
++ efx->type->txd_ring_mask);
++ buffer = &tx_queue->buffer[insert_ptr];
++ EFX_BUG_ON_PARANOID(buffer->skb);
++ EFX_BUG_ON_PARANOID(buffer->len);
++ EFX_BUG_ON_PARANOID(buffer->continuation != 1);
++ EFX_BUG_ON_PARANOID(buffer->unmap_len);
++
++ dma_len = (((~dma_addr) & efx->type->tx_dma_mask) + 1);
++ if (likely(dma_len > len))
++ dma_len = len;
++
++ misalign = (unsigned)dma_addr & efx->type->bug5391_mask;
++ if (misalign && dma_len + misalign > 512)
++ dma_len = 512 - misalign;
++
++ /* Fill out per descriptor fields */
++ buffer->len = dma_len;
++ buffer->dma_addr = dma_addr;
++ len -= dma_len;
++ dma_addr += dma_len;
++ ++tx_queue->insert_count;
++ } while (len);
++
++ /* Transfer ownership of the unmapping to the final buffer */
++ buffer->unmap_addr = unmap_addr;
++ buffer->unmap_single = unmap_single;
++ buffer->unmap_len = unmap_len;
++ unmap_len = 0;
++
++ /* Get address and size of next fragment */
++ if (i >= skb_shinfo(skb)->nr_frags)
++ break;
++ fragment = &skb_shinfo(skb)->frags[i];
++ len = fragment->size;
++ page = fragment->page;
++ page_offset = fragment->page_offset;
++ i++;
++ /* Map for DMA */
++ unmap_single = 0;
++ dma_addr = pci_map_page(pci_dev, page, page_offset, len,
++ PCI_DMA_TODEVICE);
++ }
++
++ /* Transfer ownership of the skb to the final buffer */
++ buffer->skb = skb;
++ buffer->continuation = 0;
++
++ /* Pass off to hardware */
++ falcon_push_buffers(tx_queue);
++
++ return NETDEV_TX_OK;
++
++ pci_err:
++ EFX_ERR_RL(efx, " TX queue %d could not map skb with %d bytes %d "
++ "fragments for DMA\n", tx_queue->queue, skb->len,
++ skb_shinfo(skb)->nr_frags + 1);
++
++ /* Mark the packet as transmitted, and free the SKB ourselves */
++ dev_kfree_skb_any((struct sk_buff *)skb);
++ goto unwind;
++
++ stop:
++ rc = NETDEV_TX_BUSY;
++
++ /* Stop the queue if it wasn't stopped before. */
++ if (tx_queue->stopped == 1)
++ efx_stop_queue(efx);
++
++ unwind:
++ /* Work backwards until we hit the original insert pointer value */
++ while (tx_queue->insert_count != tx_queue->write_count) {
++ --tx_queue->insert_count;
++ insert_ptr = tx_queue->insert_count & efx->type->txd_ring_mask;
++ buffer = &tx_queue->buffer[insert_ptr];
++ if (buffer->unmap_len) {
++ if (buffer->unmap_single)
++ pci_unmap_single(pci_dev, buffer->unmap_addr,
++ buffer->unmap_len,
++ PCI_DMA_TODEVICE);
++ else
++ pci_unmap_page(pci_dev, buffer->unmap_addr,
++ buffer->unmap_len,
++ PCI_DMA_TODEVICE);
++ }
++ buffer->unmap_len = 0;
++ buffer->len = 0;
++ }
++
++ /* Free the fragment we were mid-way through pushing */
++ if (unmap_len)
++ pci_unmap_page(pci_dev, unmap_addr, unmap_len,
++ PCI_DMA_TODEVICE);
++
++ return rc;
++}
++
++/* Remove packets from the TX queue
++ *
++ * This removes packets from the TX queue, up to and including the
++ * specified index.
++ */
++static inline void efx_dequeue_buffers(struct efx_tx_queue *tx_queue,
++ unsigned int index)
++{
++ struct pci_dev *pci_dev = tx_queue->efx->pci_dev;
++ struct efx_tx_buffer *buffer;
++ unsigned int stop_index, read_ptr;
++
++ /* Calculate the stopping point. Doing the check this way
++ * avoids wrongly completing every buffer in the ring if we
++ * get called twice with the same index. (Hardware should
++ * never do this, since it can't complete that many buffers in
++ * one go.)
++ */
++ stop_index = (index + 1) & tx_queue->efx->type->txd_ring_mask;
++ read_ptr = tx_queue->read_count & tx_queue->efx->type->txd_ring_mask;
++
++ while (read_ptr != stop_index) {
++ buffer = &tx_queue->buffer[read_ptr];
++ if (unlikely(buffer->len == 0)) {
++ EFX_ERR(tx_queue->efx, "TX queue %d spurious TX "
++ "completion id %x\n", tx_queue->queue,
++ read_ptr);
++ atomic_inc(&tx_queue->efx->errors.spurious_tx);
++ /* Don't reset */
++ } else {
++ if (buffer->unmap_len) {
++ if (buffer->unmap_single)
++ pci_unmap_single(pci_dev,
++ buffer->unmap_addr,
++ buffer->unmap_len,
++ PCI_DMA_TODEVICE);
++ else
++ pci_unmap_page(pci_dev,
++ buffer->unmap_addr,
++ buffer->unmap_len,
++ PCI_DMA_TODEVICE);
++ buffer->unmap_single = 0;
++ buffer->unmap_len = 0;
++ }
++ if (buffer->skb) {
++ dev_kfree_skb_any((struct sk_buff *)
++ buffer->skb);
++ buffer->skb = NULL;
++ EFX_TRACE(tx_queue->efx, "TX queue %d "
++ "transmission id %x complete\n",
++ tx_queue->queue, read_ptr);
++ }
++ buffer->continuation = 1;
++ buffer->len = 0;
++ }
++ ++tx_queue->read_count;
++ read_ptr = (tx_queue->read_count &
++ tx_queue->efx->type->txd_ring_mask);
++ }
++}
++
++/* Initiate a packet transmission on the specified TX queue.
++ * Note that returning anything other than NETDEV_TX_OK will cause the
++ * OS to free the skb.
++ *
++ * This function is split out from efx_hard_start_xmit to allow the
++ * loopback test to direct packets via specific TX queues. It is
++ * therefore a non-static inline, so as not to penalise performance
++ * for non-loopback transmissions.
++ *
++ * Context: netif_tx_lock held
++ */
++inline int efx_xmit(struct efx_nic *efx,
++ struct efx_tx_queue *tx_queue, struct sk_buff *skb)
++{
++ int rc;
++
++ /* Map fragments for DMA and add to TX queue */
++ rc = efx_enqueue_skb(tx_queue, skb);
++ if (unlikely(rc != NETDEV_TX_OK))
++ goto out;
++
++ /* Update last TX timer */
++ efx->net_dev->trans_start = jiffies;
++
++ out:
++ return rc;
++}
++
++/* Initiate a packet transmission. We use one channel per CPU
++ * (sharing when we have more CPUs than channels). On Falcon, the TX
++ * completion events will be directed back to the CPU that transmitted
++ * the packet, which should be cache-efficient.
++ *
++ * Context: non-blocking.
++ * Note that returning anything other than NETDEV_TX_OK will cause the
++ * OS to free the skb.
++ */
++int efx_hard_start_xmit(struct sk_buff *skb, struct net_device *net_dev)
++{
++ struct efx_nic *efx = net_dev->priv;
++ struct efx_tx_queue *tx_queue;
++ enum efx_veto veto;
++ int rc = NETDEV_TX_OK;
++
++ /* We have one TX queue. */
++ tx_queue = &efx->tx_queue[0];
++
++ /* See if driverlink wants to veto the packet. */
++ veto = EFX_DL_CALLBACK(efx, tx_packet, skb);
++ if (unlikely(veto)) {
++ EFX_LOG(efx, "TX queue %d packet vetoed by "
++ "driverlink %s driver\n", tx_queue->queue,
++ efx->dl_cb_dev.tx_packet->driver->name);
++ /* Free the skb; nothing else will do it */
++ dev_kfree_skb_any((struct sk_buff *)skb);
++ goto out;
++ }
++
++ rc = efx_xmit(efx, tx_queue, skb);
++out:
++ return rc;
++}
++
++#if defined(EFX_USE_FASTCALL)
++void fastcall efx_xmit_done(struct efx_tx_queue *tx_queue, unsigned int index)
++#else
++void efx_xmit_done(struct efx_tx_queue *tx_queue, unsigned int index)
++#endif
++{
++ unsigned long flags __attribute__ ((unused));
++ unsigned fill_level;
++ struct efx_nic *efx = tx_queue->efx;
++
++ EFX_BUG_ON_PARANOID(index > efx->type->txd_ring_mask);
++
++ /* Remove buffers from TX queue */
++ efx_dequeue_buffers(tx_queue, index);
++
++ /* See if we need to restart the netif queue. This barrier
++ * separates the update of read_count from the test of
++ * stopped. */
++ smp_mb();
++ if (unlikely(tx_queue->stopped)) {
++ fill_level = tx_queue->insert_count - tx_queue->read_count;
++ if (fill_level < EFX_NETDEV_TX_THRESHOLD(tx_queue)) {
++ /* If the port is stopped and the net_dev isn't
++ * registered, then the caller must be performing
++ * flow control manually */
++ if (unlikely(!efx->net_dev_registered))
++ return;
++
++ /* Do this under netif_tx_lock(), to avoid racing
++ * with efx_xmit(). */
++ netif_tx_lock(efx->net_dev);
++ if (tx_queue->stopped) {
++ tx_queue->stopped = 0;
++ efx_wake_queue(efx);
++ }
++ netif_tx_unlock(efx->net_dev);
++ }
++ }
++}
++
++int efx_probe_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ struct efx_nic *efx = tx_queue->efx;
++ unsigned int txq_size;
++ int i, rc;
++
++ EFX_LOG(efx, "creating TX queue %d\n", tx_queue->queue);
++
++ /* Allocate software ring */
++ txq_size = (efx->type->txd_ring_mask + 1) * sizeof(*tx_queue->buffer);
++ tx_queue->buffer = kzalloc(txq_size, GFP_KERNEL);
++ if (!tx_queue->buffer) {
++ rc = -ENOMEM;
++ goto fail1;
++ }
++ for (i = 0; i <= efx->type->txd_ring_mask; ++i)
++ tx_queue->buffer[i].continuation = 1;
++
++ /* Allocate hardware ring */
++ rc = falcon_probe_tx(tx_queue);
++ if (rc)
++ goto fail2;
++
++ return 0;
++
++ fail2:
++ kfree(tx_queue->buffer);
++ tx_queue->buffer = NULL;
++ fail1:
++ /* Mark queue as unused */
++ tx_queue->used = 0;
++
++ return rc;
++}
++
++int efx_init_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ EFX_LOG(tx_queue->efx, "initialising TX queue %d\n", tx_queue->queue);
++
++ ASSERT_RTNL();
++
++ /* Initialise fields */
++ tx_queue->insert_count = 0;
++ tx_queue->write_count = 0;
++ tx_queue->read_count = 0;
++ tx_queue->old_read_count = 0;
++ BUG_ON(tx_queue->stopped);
++
++ /* Set up TX descriptor ring */
++ return falcon_init_tx(tx_queue);
++}
++
++void efx_release_tx_buffers(struct efx_tx_queue *tx_queue)
++{
++ unsigned int last_index, mask;
++ if (tx_queue->buffer) {
++ /* Free any buffers left in the ring */
++ mask = tx_queue->efx->type->txd_ring_mask;
++ last_index = (tx_queue->insert_count - 1) & mask;
++ EFX_LOG(tx_queue->efx, "Will dequeue up to 0x%x from 0x%x\n",
++ last_index, tx_queue->read_count & mask);
++ efx_dequeue_buffers(tx_queue, last_index);
++ }
++}
++
++void efx_fini_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ EFX_LOG(tx_queue->efx, "shutting down TX queue %d\n", tx_queue->queue);
++
++ ASSERT_RTNL();
++
++ /* Flush TX queue, remove descriptor ring */
++ falcon_fini_tx(tx_queue);
++
++ /* Release TX buffers */
++ efx_release_tx_buffers(tx_queue);
++
++ /* Release queue's stop on port, if any */
++ if (tx_queue->stopped) {
++ tx_queue->stopped = 0;
++ efx_wake_queue(tx_queue->efx);
++ }
++}
++
++void efx_remove_tx_queue(struct efx_tx_queue *tx_queue)
++{
++ EFX_LOG(tx_queue->efx, "destroying TX queue %d\n", tx_queue->queue);
++ falcon_remove_tx(tx_queue);
++
++ kfree(tx_queue->buffer);
++ tx_queue->buffer = NULL;
++ tx_queue->used = 0;
++}
++
++
+--- linux-2.6.18.8/drivers/net/sfc/tx.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/tx.h 2008-05-19 00:33:29.453843880 +0300
+@@ -0,0 +1,41 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006: Fen Systems Ltd.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Initially developed by Michael Brown <mbrown@fensystems.co.uk>
++ * Maintained by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_TX_H
++#define EFX_TX_H
++
++#include "net_driver.h"
++
++int efx_probe_tx_queue(struct efx_tx_queue *tx_queue);
++void efx_remove_tx_queue(struct efx_tx_queue *tx_queue);
++int efx_init_tx_queue(struct efx_tx_queue *tx_queue);
++void efx_fini_tx_queue(struct efx_tx_queue *tx_queue);
++
++int efx_hard_start_xmit(struct sk_buff *skb, struct net_device *net_dev);
++void efx_release_tx_buffers(struct efx_tx_queue *tx_queue);
++
++#endif /* EFX_TX_H */
+--- linux-2.6.18.8/drivers/net/sfc/txc43128_phy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/txc43128_phy.c 2008-05-19 00:33:29.453843880 +0300
+@@ -0,0 +1,725 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++/*
++ * Driver for Transwitch/Mysticom CX4 retimer
++ * see www.transwitch.com, part is TXC-43128
++ */
++
++#include <linux/delay.h>
++#include <linux/seq_file.h>
++#include "efx.h"
++#include "debugfs.h"
++#include "gmii.h"
++#include "mdio_10g.h"
++#include "xenpack.h"
++#include "phy.h"
++#include "lm87_support.h"
++#include "falcon.h"
++#include "workarounds.h"
++
++/* We expect these MMDs to be in the package */
++#define TXC_REQUIRED_DEVS (MDIO_MMDREG_DEVS0_PCS | \
++ MDIO_MMDREG_DEVS0_PMAPMD | \
++ MDIO_MMDREG_DEVS0_PHYXS)
++
++#define TXC_LOOPBACKS ((1 << LOOPBACK_PCS) | \
++ (1 << LOOPBACK_PMAPMD) | \
++ (1 << LOOPBACK_NETWORK))
++
++/**************************************************************************
++ *
++ * Compile-time config
++ *
++ **************************************************************************
++ */
++#define TXCNAME "TXC43128"
++/* Total length of time we'll wait for the PHY to come out of reset */
++#define TXC_MAX_RESET_TIME 500
++/* Interval between checks */
++#define TXC_RESET_WAIT 10
++/* How long to run BIST: At 10Gbps 50 microseconds should be plenty to get
++ * some stats */
++#define TXC_BIST_DURATION (50)
++
++#define BER_INTERVAL (10 * efx_monitor_interval)
++
++/**************************************************************************
++ *
++ * Register definitions
++ *
++ **************************************************************************
++ */
++#define XAUI_NUM_LANES (4)
++
++/*** Global register bank */
++/* Silicon ID register */
++#define TXC_GLRGS_SLID (0xc000)
++#define TXC_GLRGS_SLID_MASK (0x1f)
++
++/* Command register */
++#define TXC_GLRGS_GLCMD (0xc004)
++/* Useful bits in command register */
++/* Lane power-down */
++#define TXC_GLCMD_L01PD_LBN (5)
++#define TXC_GLCMD_L23PD_LBN (6)
++/* Limited SW reset: preserves configuration but
++ * initiates a logic reset. Self-clearing */
++#define TXC_GLCMD_LMTSWRST_LBN (14)
++
++/* Signal Quality Control */
++#define TXC_GLRGS_GSGQLCTL (0xc01a)
++/* Enable bit */
++#define TXC_GSGQLCT_SGQLEN_LBN (15)
++/* Lane selection */
++#define TXC_GSGQLCT_LNSL_LBN (13)
++#define TXC_GSGQLCT_LNSL_WIDTH (2)
++
++/* Signal Quality Input */
++#define TXC_GLRGS_GSGQLIN (0xc01b)
++/* Signal Quality Grade */
++#define TXC_GLRGS_GSGQLGRD (0xc01c)
++/* Drift sign */
++#define TXC_GSGQLGRD_DRFTSGN_LBN (15)
++/* Grade valid flag */
++#define TXC_GSGQLGRD_GRDVAL_LBN (14)
++/* Remaining bits are the actual grade */
++#define TXC_GSGQLGRD_GRADE_LBN (0)
++#define TXC_GSGQLGRD_GRADE_WIDTH (14)
++
++/* Signal Quality Drift: 16-bit drift value */
++#define TXC_GLRGS_GSGQLDRFT (0xc01d)
++
++/**** Analog register bank */
++#define TXC_ALRGS_ATXCTL (0xc040)
++/* Lane power-down */
++#define TXC_ATXCTL_TXPD3_LBN (15)
++#define TXC_ATXCTL_TXPD2_LBN (14)
++#define TXC_ATXCTL_TXPD1_LBN (13)
++#define TXC_ATXCTL_TXPD0_LBN (12)
++
++/* Amplitude on lanes 0, 1 */
++#define TXC_ALRGS_ATXAMP0 (0xc041)
++/* Amplitude on lanes 2, 3 */
++#define TXC_ALRGS_ATXAMP1 (0xc042)
++/* Bit position of value for lane 0 (or 2) */
++#define TXC_ATXAMP_LANE02_LBN (3)
++/* Bit position of value for lane 1 (or 3) */
++#define TXC_ATXAMP_LANE13_LBN (11)
++
++#define TXC_ATXAMP_1280_mV (0)
++#define TXC_ATXAMP_1200_mV (8)
++#define TXC_ATXAMP_1120_mV (12)
++#define TXC_ATXAMP_1060_mV (14)
++#define TXC_ATXAMP_0820_mV (25)
++#define TXC_ATXAMP_0720_mV (26)
++#define TXC_ATXAMP_0580_mV (27)
++#define TXC_ATXAMP_0440_mV (28)
++
++#define TXC_ATXAMP_0820_BOTH \
++ ((TXC_ATXAMP_0820_mV << TXC_ATXAMP_LANE02_LBN) \
++ | (TXC_ATXAMP_0820_mV << TXC_ATXAMP_LANE13_LBN))
++
++#define TXC_ATXAMP_DEFAULT (0x6060) /* From databook */
++
++/* Preemphasis on lanes 0, 1 */
++#define TXC_ALRGS_ATXPRE0 (0xc043)
++/* Preemphasis on lanes 2, 3 */
++#define TXC_ALRGS_ATXPRE1 (0xc044)
++
++#define TXC_ATXPRE_NONE (0)
++#define TXC_ATXPRE_DEFAULT (0x1010) /* From databook */
++
++#define TXC_ALRGS_ARXCTL (0xc045)
++/* Lane power-down */
++#define TXC_ARXCTL_RXPD3_LBN (15)
++#define TXC_ARXCTL_RXPD2_LBN (14)
++#define TXC_ARXCTL_RXPD1_LBN (13)
++#define TXC_ARXCTL_RXPD0_LBN (12)
++
++/*** receiver control registers: Bit Error Rate measurement */
++/* Per lane BER timers */
++#define TXC_RXCTL_BERTMR0 (0xc0d4)
++#define TXC_RXCTL_BERTMR1 (0xc154)
++#define TXC_RXCTL_BERTMR2 (0xc1d4)
++#define TXC_RXCTL_BERTMR3 (0xc254)
++/* Per lane BER counters */
++#define TXC_RXCTL_BERCNT0 (0xc0d5)
++#define TXC_RXCTL_BERCNT1 (0xc155)
++#define TXC_RXCTL_BERCNT2 (0xc1d5)
++#define TXC_RXCTL_BERCNT3 (0xc255)
++
++#define BER_REG_SPACING (TXC_RXCTL_BERTMR1 - TXC_RXCTL_BERTMR0)
++
++/*** Main user-defined register set */
++/* Main control */
++#define TXC_MRGS_CTL (0xc340)
++/* Bits in main control */
++#define TXC_MCTL_RESET_LBN (15) /* Self clear */
++#define TXC_MCTL_TXLED_LBN (14) /* 1 to show align status */
++#define TXC_MCTL_RXLED_LBN (13) /* 1 to show align status */
++
++/* GPIO output */
++#define TXC_GPIO_OUTPUT (0xc346)
++#define TXC_GPIO_DIR (0xc348)
++
++/*** Vendor-specific BIST registers */
++#define TXC_BIST_CTL (0xc280)
++#define TXC_BIST_TXFRMCNT (0xc281)
++#define TXC_BIST_RX0FRMCNT (0xc282)
++#define TXC_BIST_RX1FRMCNT (0xc283)
++#define TXC_BIST_RX2FRMCNT (0xc284)
++#define TXC_BIST_RX3FRMCNT (0xc285)
++#define TXC_BIST_RX0ERRCNT (0xc286)
++#define TXC_BIST_RX1ERRCNT (0xc287)
++#define TXC_BIST_RX2ERRCNT (0xc288)
++#define TXC_BIST_RX3ERRCNT (0xc289)
++
++/*** BIST control bits */
++/* BIST type (controls bit patter in test) */
++#define TXC_BIST_CTRL_TYPE_LBN (10)
++#define TXC_BIST_CTRL_TYPE_TSD (0) /* TranSwitch Deterministic */
++#define TXC_BIST_CTRL_TYPE_CRP (1) /* CRPAT standard */
++#define TXC_BIST_CTRL_TYPE_CJP (2) /* CJPAT standard */
++#define TXC_BIST_CTRL_TYPE_TSR (3) /* TranSwitch pseudo-random */
++/* Set this to 1 for 10 bit and 0 for 8 bit */
++#define TXC_BIST_CTRL_B10EN_LBN (12)
++/* Enable BIST (write 0 to disable) */
++#define TXC_BIST_CTRL_ENAB_LBN (13)
++/*Stop BIST (self-clears when stop complete) */
++#define TXC_BIST_CTRL_STOP_LBN (14)
++/* Start BIST (cleared by writing 1 to STOP) */
++#define TXC_BIST_CTRL_STRT_LBN (15)
++
++/* Mt. Diablo test configuration */
++#define TXC_MTDIABLO_CTRL (0xc34f)
++#define TXC_MTDIABLO_CTRL_PMA_LOOP_LBN (10)
++
++struct txc43128_data {
++#ifdef CONFIG_SFC_DEBUGFS
++ /* BER stats update from check_hw. Note that this is in errors/second,
++ * converting it to errors/bit is left as an exercise for user-space.
++ */
++ unsigned phy_ber_pcs[4];
++ unsigned phy_ber_phyxs[4];
++#endif
++ unsigned bug10934_timer;
++ int phy_powered;
++ int tx_disabled;
++ enum efx_loopback_mode loopback_mode;
++};
++
++/* Perform the bug 10934 workaround every 5s */
++#define BUG10934_RESET_INTERVAL (5 * HZ)
++
++
++/* Perform a reset that doesn't clear configuration changes */
++static void txc_reset_logic(struct efx_nic *efx);
++
++/* Set the output value of a gpio */
++void txc_set_gpio_val(struct efx_nic *efx, int pin, int on)
++{
++ int outputs;
++
++ outputs = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_GPIO_OUTPUT);
++
++ outputs = (outputs & ~(1 << pin)) | (on << pin);
++
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_GPIO_OUTPUT,
++ outputs);
++}
++
++/* Set up the GPIO direction register */
++void txc_set_gpio_dir(struct efx_nic *efx, int pin, int dir)
++{
++ int dirs;
++
++ if (efx->board_info.minor < 3 &&
++ efx->board_info.major == 0)
++ return;
++
++ dirs = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_GPIO_DIR);
++ dirs = (dir & ~(1 << pin)) | (dir << pin);
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_GPIO_DIR, dirs);
++
++}
++
++/* Reset the PMA/PMD MMD. The documentation is explicit that this does a
++ * global reset (it's less clear what reset of other MMDs does).*/
++static int txc_reset_phy(struct efx_nic *efx)
++{
++ int rc = mdio_clause45_reset_mmd(efx, MDIO_MMD_PMAPMD,
++ TXC_MAX_RESET_TIME / TXC_RESET_WAIT,
++ TXC_RESET_WAIT);
++ if (rc < 0)
++ goto fail;
++
++ /* Check that all the MMDs we expect are present and responding. We
++ * expect faults on some if the link is down, but not on the PHY XS */
++ rc = mdio_clause45_check_mmds(efx, TXC_REQUIRED_DEVS, 0);
++ if (rc < 0)
++ goto fail;
++
++ return 0;
++
++ fail:
++ EFX_ERR(efx, TXCNAME ": reset timed out!\n");
++ return rc;
++}
++
++/* Run a single BIST on one MMD*/
++static int txc_bist_one(struct efx_nic *efx, int mmd, int test)
++{
++ int phy = efx->mii.phy_id;
++ int ctrl, bctl;
++ int lane;
++ int rc = 0;
++
++ EFX_INFO(efx, "" TXCNAME ": running BIST on %s MMD\n",
++ mdio_clause45_mmd_name(mmd));
++
++ /* Set PMA to test into loopback using Mt Diablo reg as per app note */
++ ctrl = mdio_clause45_read(efx, phy, MDIO_MMD_PCS,
++ TXC_MTDIABLO_CTRL);
++ ctrl |= (1 << TXC_MTDIABLO_CTRL_PMA_LOOP_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ TXC_MTDIABLO_CTRL, ctrl);
++
++
++ /* The BIST app. note lists these as 3 distinct steps. */
++ /* Set the BIST type */
++ bctl = (test << TXC_BIST_CTRL_TYPE_LBN);
++ mdio_clause45_write(efx, phy, mmd, TXC_BIST_CTL, bctl);
++
++ /* Set the BSTEN bit in the BIST Control register to enable */
++ bctl |= (1 << TXC_BIST_CTRL_ENAB_LBN);
++ mdio_clause45_write(efx, phy, mmd, TXC_BIST_CTL, bctl);
++
++ /* Set the BSTRT bit in the BIST Control register */
++ mdio_clause45_write(efx, phy, mmd, TXC_BIST_CTL, bctl |
++ (1 << TXC_BIST_CTRL_STRT_LBN));
++
++ /* Wait. */
++ udelay(TXC_BIST_DURATION);
++
++ /* Set the BSTOP bit in the BIST Control register */
++ bctl |= (1 << TXC_BIST_CTRL_STOP_LBN);
++ mdio_clause45_write(efx, phy, mmd, TXC_BIST_CTL, bctl);
++
++ /* The STOP bit should go off when things have stopped */
++ while (bctl & (1 << TXC_BIST_CTRL_STOP_LBN))
++ bctl = mdio_clause45_read(efx, phy, mmd, TXC_BIST_CTL);
++
++ /* Check all the error counts are 0 and all the frame counts are
++ non-zero */
++ for (lane = 0; lane < 4; lane++) {
++ int count = mdio_clause45_read(efx, phy, mmd,
++ TXC_BIST_RX0ERRCNT + lane);
++ if (count != 0) {
++ EFX_ERR(efx, ""TXCNAME": BIST error. "
++ "Lane %d had %d errs\n", lane, count);
++ rc = -EIO;
++ }
++ count = mdio_clause45_read(efx, phy, mmd,
++ TXC_BIST_RX0FRMCNT + lane);
++ if (count == 0) {
++ EFX_ERR(efx, ""TXCNAME": BIST error. "
++ "Lane %d got 0 frames\n", lane);
++ rc = -EIO;
++ }
++ }
++
++ if (rc == 0)
++ EFX_INFO(efx, ""TXCNAME": BIST pass\n");
++
++ /* Disable BIST */
++ mdio_clause45_write(efx, phy, mmd, TXC_BIST_CTL, 0);
++
++ /* Turn off loopback */
++ ctrl &= ~(1 << TXC_MTDIABLO_CTRL_PMA_LOOP_LBN);
++ mdio_clause45_write(efx, phy, MDIO_MMD_PCS,
++ TXC_MTDIABLO_CTRL, ctrl);
++
++ return rc;
++}
++
++/* Run all the desired BIST tests for the PHY */
++static int txc_bist(struct efx_nic *efx)
++{
++ int rc;
++ /*!\todo: experiment with running more of the BIST patterns to
++ * see if it actually shows up more problems. */
++ rc = txc_bist_one(efx, MDIO_MMD_PCS, TXC_BIST_CTRL_TYPE_TSD);
++ return rc;
++}
++
++#ifdef CONFIG_SFC_DEBUGFS
++
++/* debugfs entries for this PHY */
++static struct efx_debugfs_parameter debug_entries[] = {
++ EFX_PER_LANE_PARAMETER("phy_ber_lane", "_pcs",
++ struct txc43128_data, phy_ber_pcs,
++ unsigned, efx_debugfs_read_uint),
++ EFX_PER_LANE_PARAMETER("phy_ber_lane", "_phyxs",
++ struct txc43128_data, phy_ber_phyxs,
++ unsigned, efx_debugfs_read_uint),
++ EFX_INT_PARAMETER(struct txc43128_data, phy_powered),
++ {NULL}
++};
++
++#endif /* CONFIG_SFC_DEBUGFS */
++
++/* Push the non-configurable defaults into the PHY. This must be
++ * done after every full reset */
++static void txc_apply_defaults(struct efx_nic *efx)
++{
++ int mctrl;
++
++ /* Turn amplitude down and preemphasis off on the host side
++ * (PHY<->MAC) as this is believed less likely to upset Falcon
++ * and no adverse effects have been noted. It probably also
++ * saves a picowatt or two */
++
++ /* Turn off preemphasis */
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PHYXS,
++ TXC_ALRGS_ATXPRE0, TXC_ATXPRE_NONE);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PHYXS,
++ TXC_ALRGS_ATXPRE1, TXC_ATXPRE_NONE);
++
++ /* Turn down the amplitude */
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PHYXS,
++ TXC_ALRGS_ATXAMP0, TXC_ATXAMP_0820_BOTH);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PHYXS,
++ TXC_ALRGS_ATXAMP1, TXC_ATXAMP_0820_BOTH);
++
++ /* Set the line side amplitude and preemphasis to the databook
++ * defaults as an erratum causes them to be 0 on at least some
++ * PHY rev.s */
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ TXC_ALRGS_ATXPRE0, TXC_ATXPRE_DEFAULT);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ TXC_ALRGS_ATXPRE1, TXC_ATXPRE_DEFAULT);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ TXC_ALRGS_ATXAMP0, TXC_ATXAMP_DEFAULT);
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ TXC_ALRGS_ATXAMP1, TXC_ATXAMP_DEFAULT);
++
++ /* Set up the LEDs */
++ mctrl = mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_MRGS_CTL);
++
++ /* Set the Green and Red LEDs to their default modes */
++ mctrl &= ~((1 << TXC_MCTL_TXLED_LBN) | (1 << TXC_MCTL_RXLED_LBN));
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_MRGS_CTL, mctrl);
++
++ /* Databook recommends doing this after configuration changes */
++ txc_reset_logic(efx);
++
++ efx->board_info.init_leds(efx);
++}
++
++/* Initialisation entry point for this PHY driver */
++static int txc43128_phy_init(struct efx_nic *efx)
++{
++ u32 devid;
++ int rc = 0;
++ struct txc43128_data *phy_data;
++
++ devid = mdio_clause45_read_id(efx, MDIO_MMD_PHYXS);
++
++ phy_data = kzalloc(sizeof(*phy_data), GFP_KERNEL);
++ efx->phy_data = phy_data;
++
++ /* This is the default after reset */
++ phy_data->phy_powered = efx->phy_powered;
++ phy_data->tx_disabled = efx->tx_disabled;
++
++#ifdef CONFIG_SFC_DEBUGFS
++ rc = efx_extend_debugfs_port(efx, phy_data, debug_entries);
++ if (rc < 0)
++ goto fail1;
++#endif
++ EFX_INFO(efx, ""TXCNAME ": PHY ID reg %x (OUI %x model %x "
++ "revision %x)\n", devid, MDIO_ID_OUI(devid),
++ MDIO_ID_MODEL(devid), MDIO_ID_REV(devid));
++
++ EFX_INFO(efx, ""TXCNAME ": Silicon ID %x\n",
++ mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PHYXS, TXC_GLRGS_SLID) &
++ TXC_GLRGS_SLID_MASK);
++
++ rc = txc_reset_phy(efx);
++ if (rc < 0)
++ goto fail2;
++
++ rc = txc_bist(efx);
++ if (rc < 0)
++ goto fail2;
++
++ txc_apply_defaults(efx);
++
++ return 0;
++
++ fail2:
++#ifdef CONFIG_SFC_DEBUGFS
++ efx_trim_debugfs_port(efx, debug_entries);
++ /* fall-thru */
++ fail1:
++#endif
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++ return rc;
++}
++
++/* Set the lane power down state in the global registers */
++static void txc_glrgs_lane_power(struct efx_nic *efx, int mmd)
++{
++ int pd = (1 << TXC_GLCMD_L01PD_LBN) | (1 << TXC_GLCMD_L23PD_LBN);
++ int ctl = mdio_clause45_read(efx, efx->mii.phy_id,
++ mmd, TXC_GLRGS_GLCMD);
++
++ if (efx->phy_powered)
++ ctl &= ~pd;
++ else
++ ctl |= pd;
++
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ mmd, TXC_GLRGS_GLCMD, ctl);
++}
++
++/* Set the lane power down state in the analog control registers */
++static void txc_analog_lane_power(struct efx_nic *efx, int mmd)
++{
++ int txpd = (1 << TXC_ATXCTL_TXPD3_LBN) | (1 << TXC_ATXCTL_TXPD2_LBN)
++ | (1 << TXC_ATXCTL_TXPD1_LBN) | (1 << TXC_ATXCTL_TXPD0_LBN);
++
++ int rxpd = (1 << TXC_ATXCTL_TXPD3_LBN) | (1 << TXC_ATXCTL_TXPD2_LBN)
++ | (1 << TXC_ATXCTL_TXPD1_LBN) | (1 << TXC_ATXCTL_TXPD0_LBN);
++
++ int txctl = mdio_clause45_read(efx, efx->mii.phy_id,
++ mmd, TXC_ALRGS_ATXCTL);
++ int rxctl = mdio_clause45_read(efx, efx->mii.phy_id,
++ mmd, TXC_ALRGS_ARXCTL);
++
++ if (efx->phy_powered) {
++ txctl &= ~txpd;
++ rxctl &= ~rxpd;
++ } else {
++ txctl |= txpd;
++ rxctl |= rxpd;
++ }
++
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ mmd, TXC_ALRGS_ATXCTL, txctl);
++ mdio_clause45_write(efx, efx->mii.phy_id,
++ mmd, TXC_ALRGS_ARXCTL, rxctl);
++}
++
++static void txc_set_power(struct efx_nic *efx)
++{
++ /* According to the data book, all the MMDs can do low power */
++ mdio_clause45_set_mmds_lpower(efx, !efx->phy_powered,
++ TXC_REQUIRED_DEVS);
++
++ /* Global register bank is in PCS, PHY XS. These control the host
++ * side and line side settings respectively. */
++ txc_glrgs_lane_power(efx, MDIO_MMD_PCS);
++ txc_glrgs_lane_power(efx, MDIO_MMD_PHYXS);
++
++ /* Analog register bank in PMA/PMD, PHY XS */
++ txc_analog_lane_power(efx, MDIO_MMD_PMAPMD);
++ txc_analog_lane_power(efx, MDIO_MMD_PHYXS);
++}
++
++
++static void txc_reset_logic_mmd(struct efx_nic *efx, int mmd)
++{
++ int portid = efx->mii.phy_id;
++ int val = mdio_clause45_read(efx, portid, mmd, TXC_GLRGS_GLCMD);
++ int tries = 50;
++ val |= (1 << TXC_GLCMD_LMTSWRST_LBN);
++ mdio_clause45_write(efx, portid, mmd, TXC_GLRGS_GLCMD, val);
++ while (tries--) {
++ val = mdio_clause45_read(efx, portid, mmd,
++ TXC_GLRGS_GLCMD);
++ if (!(val & (1 << TXC_GLCMD_LMTSWRST_LBN)))
++ break;
++ udelay(1);
++ }
++ if (!tries)
++ EFX_INFO(efx, TXCNAME " Logic reset timed out!\n");
++}
++
++
++/* Perform a logic reset. This preserves the configuration registers
++ * and is needed for some configuration changes to take effect */
++static void txc_reset_logic(struct efx_nic *efx)
++{
++ /* The data sheet claims we can do the logic reset on either the
++ * PCS or the PHYXS and the result is a reset of both host- and
++ * line-side logic. */
++ txc_reset_logic_mmd(efx, MDIO_MMD_PCS);
++}
++
++static int txc43128_phy_read_link(struct efx_nic *efx)
++{
++ return mdio_clause45_links_ok(efx, TXC_REQUIRED_DEVS);
++}
++
++static void txc43128_phy_reconfigure(struct efx_nic *efx)
++{
++ struct txc43128_data *phy_data = efx->phy_data;
++ int power_change = (efx->phy_powered != phy_data->phy_powered);
++ int loop_change = LOOPBACK_CHANGED(phy_data, efx, TXC_LOOPBACKS);
++ int disable_change = (efx->tx_disabled != phy_data->tx_disabled);
++
++ if (!phy_data->tx_disabled && efx->tx_disabled) {
++ txc_reset_phy(efx);
++ txc_apply_defaults(efx);
++ falcon_reset_xaui(efx);
++ disable_change = 0;
++ }
++
++ mdio_clause45_transmit_disable(efx, efx->tx_disabled);
++ mdio_clause45_phy_reconfigure(efx);
++ if (power_change)
++ txc_set_power(efx);
++
++ /* The data sheet claims this is required after every reconfiguration
++ * (note at end of 7.1), but we mustn't do it when nothing changes as
++ * it glitches the link, and reconfigure gets called on link change,
++ * so we get an IRQ storm on link up. */
++ if (loop_change || power_change || disable_change)
++ txc_reset_logic(efx);
++
++ phy_data->phy_powered = efx->phy_powered;
++ phy_data->loopback_mode = efx->loopback_mode;
++ phy_data->tx_disabled = efx->tx_disabled;
++ efx->link_up = txc43128_phy_read_link(efx);
++ efx->link_options = GM_LPA_10000FULL;
++}
++
++static void txc43128_phy_fini(struct efx_nic *efx)
++{
++ efx->board_info.blink(efx, 0);
++
++ /* Disable link events */
++ xenpack_disable_lasi_irqs(efx);
++
++#ifdef CONFIG_SFC_DEBUGFS
++ /* Remove the extra debug entries and free data */
++ efx_trim_debugfs_port(efx, debug_entries);
++#endif
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++}
++
++/* Periodic callback: this exists mainly to poll link status as we currently
++ * don't use LASI interrupts. Also update the BER counters and poll the lm87 */
++static int txc43128_phy_check_hw(struct efx_nic *efx)
++{
++ struct txc43128_data *data = efx->phy_data;
++#ifdef CONFIG_SFC_DEBUGFS
++ int phy = efx->mii.phy_id;
++ int timer, count, i, mmd;
++#endif
++ int rc = 0;
++ int link_up = txc43128_phy_read_link(efx);
++
++ /* Simulate a PHY event if link state has changed */
++ if (link_up != efx->link_up) {
++ efx->link_up = link_up;
++ efx->mac_op->fake_phy_event(efx);
++ } else if (EFX_WORKAROUND_10934(efx)) {
++ if (link_up || (efx->loopback_mode != LOOPBACK_NONE))
++ data->bug10934_timer = jiffies;
++ else {
++ int delta = jiffies - data->bug10934_timer;
++ if (delta >= BUG10934_RESET_INTERVAL) {
++ data->bug10934_timer = jiffies;
++ txc_reset_logic(efx);
++ }
++ }
++ }
++
++ rc = efx->board_info.monitor(efx);
++ if (rc) {
++ EFX_ERR(efx, "" TXCNAME
++ ": sensor alert! Putting PHY into low power.\n");
++ efx->phy_powered = 0;
++ txc_set_power(efx);
++ }
++
++#ifdef CONFIG_SFC_DEBUGFS
++ /* There are 2 MMDs with RX BER counters: PCS and PHY XS,
++ * which happen to be consecutively numbered */
++ for (mmd = MDIO_MMD_PCS; mmd <= MDIO_MMD_PHYXS; mmd++) {
++ for (i = 0; i < XAUI_NUM_LANES; i++) {
++ timer = mdio_clause45_read(efx, phy, mmd,
++ TXC_RXCTL_BERTMR0 +
++ i * BER_REG_SPACING);
++ count = mdio_clause45_read(efx, phy, mmd,
++ TXC_RXCTL_BERCNT0 +
++ i * BER_REG_SPACING);
++ /* The BER timer counts down in seconds. If it would
++ * expire before the next check_hw, update the stats &
++ * restart the timer (clears the count) */
++ if (timer * HZ < efx_monitor_interval) {
++ /* Record count, allowing for the fact that the
++ * timer may not have reached zero */
++ unsigned ber = (count * BER_INTERVAL) /
++ (BER_INTERVAL - timer * HZ);
++ if (mmd == MDIO_MMD_PCS)
++ data->phy_ber_pcs[i] = ber;
++ else
++ data->phy_ber_phyxs[i] = ber;
++ /* Reprogram the timer */
++ mdio_clause45_write(efx, phy, mmd,
++ TXC_RXCTL_BERTMR0 +
++ i * BER_REG_SPACING,
++ BER_INTERVAL / HZ);
++ }
++ }
++ mmd = (mmd == MDIO_MMD_PCS) ? MDIO_MMD_PHYXS : 0;
++ }
++#endif /* CONFIG_SFC_DEBUGFS */
++ return rc;
++}
++
++struct efx_phy_operations falcon_txc_phy_ops = {
++ .init = txc43128_phy_init,
++ .reconfigure = txc43128_phy_reconfigure,
++ .check_hw = txc43128_phy_check_hw,
++ .fini = txc43128_phy_fini,
++ .clear_interrupt = efx_port_dummy_op_void,
++ .reset_xaui = efx_port_dummy_op_void,
++ .mmds = TXC_REQUIRED_DEVS,
++ .loopbacks = TXC_LOOPBACKS,
++ .startup_loopback = LOOPBACK_PMAPMD,
++};
+--- linux-2.6.18.8/drivers/net/sfc/workarounds.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/workarounds.h 2008-05-19 00:33:29.453843880 +0300
+@@ -0,0 +1,97 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_WORKAROUNDS_H
++#define EFX_WORKAROUNDS_H
++
++/*
++ * Hardware workarounds.
++ * Bug numbers are from Solarflare's Bugzilla.
++ */
++
++#define EFX_WORKAROUND_ALWAYS(efx) 1
++#define EFX_WORKAROUND_FALCON_A(efx) (FALCON_REV(efx) <= FALCON_REV_A1)
++#define EFX_WORKAROUND_FALCON_B0FPGA(efx) \
++ (FALCON_REV(efx) == FALCON_REV_B0 && !(efx)->is_asic)
++
++/* XAUI resets if link not detected */
++#define EFX_WORKAROUND_5147 EFX_WORKAROUND_ALWAYS
++/* SNAP frames have TOBE_DISC set */
++#define EFX_WORKAROUND_5475 EFX_WORKAROUND_ALWAYS
++/* PHY interrupts can go to the wrong port */
++#define EFX_WORKAROUND_6263 EFX_WORKAROUND_ALWAYS
++/* Reprog PCIe ACK timer to workaround issue in PCIe IP block */
++#define EFX_WORKAROUND_6943 EFX_WORKAROUND_ALWAYS
++/* RX PCIe double split performance issue */
++#define EFX_WORKAROUND_7575 EFX_WORKAROUND_ALWAYS
++/* Bit-bashed I2C reads cause performance drop */
++#define EFX_WORKAROUND_7884 EFX_WORKAROUND_ALWAYS
++/* Selftests need to be retried */
++#define EFX_WORKAROUND_8909 EFX_WORKAROUND_ALWAYS
++/* Queued ACKs aren't flushed before L1 entry */
++#define EFX_WORKAROUND_9096 EFX_WORKAROUND_ALWAYS
++/* TX pkt parser problem with <= 16 byte TXes */
++#define EFX_WORKAROUND_9141 EFX_WORKAROUND_ALWAYS
++/* XGXS and XAUI reset sequencing in SW */
++#define EFX_WORKAROUND_9388 EFX_WORKAROUND_ALWAYS
++/* Low rate CRC errors require XAUI reset */
++#define EFX_WORKAROUND_10750 EFX_WORKAROUND_ALWAYS
++/* TX_EV_PKT_ERR can be caused by a dangling TX descriptor
++ * or a PCIe error (bug 11028) */
++#define EFX_WORKAROUND_10727 EFX_WORKAROUND_ALWAYS
++/* CX4 retimer fails to bring link up after reset */
++#define EFX_WORKAROUND_10934 EFX_WORKAROUND_ALWAYS
++/* Transmit flow control may get disabled */
++#define EFX_WORKAROUND_11482 EFX_WORKAROUND_ALWAYS
++/* Flush events can take a very long time to appear */
++#define EFX_WORKAROUND_11557 EFX_WORKAROUND_ALWAYS
++
++/* Spurious parity errors in TSORT buffers */
++#define EFX_WORKAROUND_5129 EFX_WORKAROUND_FALCON_A
++/* No unaligned TX over 512 byte boundaries */
++#define EFX_WORKAROUND_5391 EFX_WORKAROUND_FALCON_A
++/* iSCSI parsing errors */
++#define EFX_WORKAROUND_5583 EFX_WORKAROUND_FALCON_A
++/* RX events go missing */
++#define EFX_WORKAROUND_5676 EFX_WORKAROUND_FALCON_A
++/* RX_RESET on A1 */
++#define EFX_WORKAROUND_6555 EFX_WORKAROUND_FALCON_A
++/* Spurious duplicate RX events */
++#define EFX_WORKAROUND_7062 EFX_WORKAROUND_FALCON_A
++/* Increase filter depth to avoid RX_RESET */
++#define EFX_WORKAROUND_7244 EFX_WORKAROUND_FALCON_A
++/* Flushes may never complete */
++#define EFX_WORKAROUND_7803 EFX_WORKAROUND_FALCON_A
++/* Leak overlength packets rather than free */
++#define EFX_WORKAROUND_8071 EFX_WORKAROUND_FALCON_A
++
++/* Memory needs clearing at start-of-day */
++#define EFX_WORKAROUND_8202 EFX_WORKAROUND_FALCON_B0FPGA
++/* MAC statistics are transient */
++#define EFX_WORKAROUND_8419 EFX_WORKAROUND_FALCON_B0FPGA
++/* Prefetch watchdog timer may trigger erroneously on busy systems */
++#define EFX_WORKAROUND_9008 EFX_WORKAROUND_FALCON_B0FPGA
++
++#endif /* EFX_WORKAROUNDS_H */
+--- linux-2.6.18.8/drivers/net/sfc/xenpack.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/xenpack.h 2008-05-19 00:33:29.457844111 +0300
+@@ -0,0 +1,80 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef EFX_XENPACK_H
++#define EFX_XENPACK_H
++
++/* Exported functions from Xenpack standard PHY control */
++
++#include "mdio_10g.h"
++
++/****************************************************************************/
++/* XENPACK MDIO register extensions */
++#define MDIO_XP_LASI_RX_CTRL (0x9000)
++#define MDIO_XP_LASI_TX_CTRL (0x9001)
++#define MDIO_XP_LASI_CTRL (0x9002)
++#define MDIO_XP_LASI_RX_STAT (0x9003)
++#define MDIO_XP_LASI_TX_STAT (0x9004)
++#define MDIO_XP_LASI_STAT (0x9005)
++
++/* Control/Status bits */
++#define XP_LASI_LS_ALARM (1 << 0)
++#define XP_LASI_TX_ALARM (1 << 1)
++#define XP_LASI_RX_ALARM (1 << 2)
++/* These two are Quake vendor extensions to the standard XENPACK defines */
++#define XP_LASI_LS_INTB (1 << 3)
++#define XP_LASI_TEST (1 << 7)
++
++/* Enable LASI interrupts for PHY */
++static inline void xenpack_enable_lasi_irqs(struct efx_nic *efx)
++{
++ int reg;
++ int phy_id = efx->mii.phy_id;
++ /* Read to clear LASI status register */
++ reg = mdio_clause45_read(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_XP_LASI_STAT);
++
++ /* Enable LASI interrupts from PMA/PMD */
++ mdio_clause45_write(efx, phy_id, MDIO_MMD_PMAPMD,
++ MDIO_XP_LASI_CTRL, XP_LASI_LS_ALARM);
++}
++
++/* Read the LASI interrupt status to clear the interrupt. */
++static inline int xenpack_clear_lasi_irqs(struct efx_nic *efx)
++{
++ /* Read to clear link status alarm */
++ return mdio_clause45_read(efx, efx->mii.phy_id,
++ MDIO_MMD_PMAPMD, MDIO_XP_LASI_STAT);
++}
++
++/* Turn off LASI interrupts */
++static inline void xenpack_disable_lasi_irqs(struct efx_nic *efx)
++{
++ /* Turn LASI interrupts off */
++ mdio_clause45_write(efx, efx->mii.phy_id, MDIO_MMD_PMAPMD,
++ MDIO_XP_LASI_CTRL, 0);
++}
++
++#endif /* EFX_XENPACK_H */
+--- linux-2.6.18.8/drivers/net/sfc/xfp_phy.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/net/sfc/xfp_phy.c 2008-05-19 00:33:29.457844111 +0300
+@@ -0,0 +1,206 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed by Solarflare Communications <linux-net-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++/*
++ * Driver for XFP optical PHYs (plus some support specific to the Quake 2032)
++ * See www.amcc.com for details (search for qt2032)
++ */
++
++#include <linux/timer.h>
++#include <linux/delay.h>
++#include "efx.h"
++#include "gmii.h"
++#include "mdio_10g.h"
++#include "xenpack.h"
++#include "phy.h"
++
++#define XFP_REQUIRED_DEVS (MDIO_MMDREG_DEVS0_PCS | \
++ MDIO_MMDREG_DEVS0_PMAPMD | \
++ MDIO_MMDREG_DEVS0_PHYXS)
++
++#define XFP_LOOPBACKS ((1 << LOOPBACK_PCS) | \
++ (1 << LOOPBACK_PMAPMD) | \
++ (1 << LOOPBACK_NETWORK))
++
++/****************************************************************************/
++/* Quake-specific MDIO registers */
++#define MDIO_QUAKE_LED0_REG (0xD006)
++
++
++void xfp_set_led(struct efx_nic *p, int led, int mode)
++{
++ int addr = MDIO_QUAKE_LED0_REG + led;
++ mdio_clause45_write(p, p->mii.phy_id, MDIO_MMD_PMAPMD, addr,
++ mode);
++}
++
++struct xfp_phy_data {
++ int phy_powered;
++ int tx_disabled;
++};
++
++
++#define XFP_MAX_RESET_TIME 500
++#define XFP_RESET_WAIT 10
++
++/* Reset the PHYXS MMD. This is documented (for the Quake PHY) as doing
++ * a complete soft reset.
++ */
++static int xfp_reset_phy(struct efx_nic *efx)
++{
++ int rc;
++
++ rc = mdio_clause45_reset_mmd(efx, MDIO_MMD_PHYXS,
++ XFP_MAX_RESET_TIME / XFP_RESET_WAIT,
++ XFP_RESET_WAIT);
++ if (rc < 0)
++ goto fail;
++
++ /* Wait 250ms for the PHY to complete bootup */
++ msleep(250);
++
++ /* Check that all the MMDs we expect are present and responding. We
++ * expect faults on some if the link is down, but not on the PHY XS */
++ rc = mdio_clause45_check_mmds(efx, XFP_REQUIRED_DEVS,
++ MDIO_MMDREG_DEVS0_PHYXS);
++ if (rc < 0)
++ goto fail;
++
++ efx->board_info.init_leds(efx);
++
++ return rc;
++
++ fail:
++ EFX_ERR(efx, "XFP: reset timed out!\n");
++ return rc;
++}
++
++
++static int xfp_phy_init(struct efx_nic *efx)
++{
++ struct xfp_phy_data *phy_data;
++ u32 devid = mdio_clause45_read_id(efx, MDIO_MMD_PHYXS);
++ int rc;
++
++ phy_data = kzalloc(sizeof(struct xfp_phy_data), GFP_KERNEL);
++ efx->phy_data = (void *) phy_data;
++
++ EFX_INFO(efx, "XFP: PHY ID reg %x (OUI %x model %x revision"
++ " %x)\n", devid, MDIO_ID_OUI(devid), MDIO_ID_MODEL(devid),
++ MDIO_ID_REV(devid));
++
++ phy_data->phy_powered = efx->phy_powered;
++ phy_data->tx_disabled = efx->tx_disabled;
++
++ rc = xfp_reset_phy(efx);
++ if (rc < 0)
++ goto fail;
++
++ EFX_INFO(efx, "XFP: PHY init %s.\n",
++ rc ? "failed" : "successful");
++ return 0;
++
++ fail:
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++ return rc;
++}
++
++static void xfp_phy_clear_interrupt(struct efx_nic *efx)
++{
++ xenpack_clear_lasi_irqs(efx);
++}
++
++static int xfp_link_ok(struct efx_nic *efx)
++{
++ return mdio_clause45_links_ok(efx, XFP_REQUIRED_DEVS);
++}
++
++static int xfp_phy_check_hw(struct efx_nic *efx)
++{
++ int rc = 0;
++ int link_up = xfp_link_ok(efx);
++ /* Simulate a PHY event if link state has changed */
++ if (link_up != efx->link_up) {
++ efx->link_up = link_up;
++ efx->mac_op->fake_phy_event(efx);
++ }
++
++ rc = efx->board_info.monitor(efx);
++ if (rc) {
++ EFX_ERR(efx, ": XFP sensor alert! Putting PHY into "
++ "low power.\n");
++ efx->phy_powered = 0;
++
++ mdio_clause45_set_mmds_lpower(efx, 1, XFP_REQUIRED_DEVS);
++ }
++
++ return rc;
++}
++
++static void xfp_phy_reconfigure(struct efx_nic *efx)
++{
++ struct xfp_phy_data *phy_data = efx->phy_data;
++
++ /* Reset the PHY when moving from transmitter off or powered off,
++ * to transmitter on and powered on */
++ if ((efx->phy_powered && !efx->tx_disabled) &&
++ (!phy_data->phy_powered || phy_data->tx_disabled))
++ xfp_reset_phy(efx);
++
++ mdio_clause45_transmit_disable(efx, efx->tx_disabled);
++ mdio_clause45_set_mmds_lpower(efx, !efx->phy_powered,
++ XFP_REQUIRED_DEVS);
++ mdio_clause45_phy_reconfigure(efx);
++
++ phy_data->tx_disabled = efx->tx_disabled;
++ phy_data->phy_powered = efx->phy_powered;
++ efx->link_up = xfp_link_ok(efx);
++ efx->link_options = GM_LPA_10000FULL;
++}
++
++
++static void xfp_phy_fini(struct efx_nic *efx)
++{
++ /* Clobber the LED if it was blinking */
++ efx->board_info.blink(efx, 0);
++
++ /* Free the context block */
++ kfree(efx->phy_data);
++ efx->phy_data = NULL;
++}
++
++struct efx_phy_operations falcon_xfp_phy_ops = {
++ .init = xfp_phy_init,
++ .reconfigure = xfp_phy_reconfigure,
++ .check_hw = xfp_phy_check_hw,
++ .fini = xfp_phy_fini,
++ .clear_interrupt = xfp_phy_clear_interrupt,
++ .reset_xaui = efx_port_dummy_op_void,
++ .mmds = XFP_REQUIRED_DEVS,
++ .loopbacks = XFP_LOOPBACKS,
++ /* No loopback appears to be reliable enough for self-test
++ * operation. So don't do it. */
++ .startup_loopback = LOOPBACK_PCS,
++};
+--- linux-2.6.18.8/drivers/oprofile/buffer_sync.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/buffer_sync.c 2008-05-19 00:33:32.550022346 +0300
+@@ -6,6 +6,10 @@
+ *
+ * @author John Levon <levon@movementarian.org>
+ *
++ * Modified by Aravind Menon for Xen
++ * These modifications are:
++ * Copyright (C) 2005 Hewlett-Packard Co.
++ *
+ * This is the core of the buffer management. Each
+ * CPU buffer is processed and entered into the
+ * global event buffer. Such processing is necessary
+@@ -38,6 +42,7 @@
+ static DEFINE_SPINLOCK(task_mortuary);
+ static void process_task_mortuary(void);
+
++static int cpu_current_domain[NR_CPUS];
+
+ /* Take ownership of the task struct and place it on the
+ * list for processing. Only after two full buffer syncs
+@@ -146,6 +151,11 @@
+ int sync_start(void)
+ {
+ int err;
++ int i;
++
++ for (i = 0; i < NR_CPUS; i++) {
++ cpu_current_domain[i] = COORDINATOR_DOMAIN;
++ }
+
+ start_cpu_work();
+
+@@ -275,13 +285,29 @@
+ last_cookie = INVALID_COOKIE;
+ }
+
+-static void add_kernel_ctx_switch(unsigned int in_kernel)
++static void add_cpu_mode_switch(unsigned int cpu_mode)
+ {
+ add_event_entry(ESCAPE_CODE);
+- if (in_kernel)
++ switch (cpu_mode) {
++ case CPU_MODE_USER:
++ add_event_entry(USER_ENTER_SWITCH_CODE);
++ break;
++ case CPU_MODE_KERNEL:
+ add_event_entry(KERNEL_ENTER_SWITCH_CODE);
+- else
+- add_event_entry(KERNEL_EXIT_SWITCH_CODE);
++ break;
++ case CPU_MODE_XEN:
++ add_event_entry(XEN_ENTER_SWITCH_CODE);
++ break;
++ default:
++ break;
++ }
++}
++
++static void add_domain_switch(unsigned long domain_id)
++{
++ add_event_entry(ESCAPE_CODE);
++ add_event_entry(DOMAIN_SWITCH_CODE);
++ add_event_entry(domain_id);
+ }
+
+ static void
+@@ -348,9 +374,9 @@
+ * for later lookup from userspace.
+ */
+ static int
+-add_sample(struct mm_struct * mm, struct op_sample * s, int in_kernel)
++add_sample(struct mm_struct * mm, struct op_sample * s, int cpu_mode)
+ {
+- if (in_kernel) {
++ if (cpu_mode >= CPU_MODE_KERNEL) {
+ add_sample_entry(s->eip, s->event);
+ return 1;
+ } else if (mm) {
+@@ -496,15 +522,21 @@
+ struct mm_struct *mm = NULL;
+ struct task_struct * new;
+ unsigned long cookie = 0;
+- int in_kernel = 1;
++ int cpu_mode = 1;
+ unsigned int i;
+ sync_buffer_state state = sb_buffer_start;
+ unsigned long available;
++ int domain_switch = 0;
+
+ mutex_lock(&buffer_mutex);
+
+ add_cpu_switch(cpu);
+
++ /* We need to assign the first samples in this CPU buffer to the
++ same domain that we were processing at the last sync_buffer */
++ if (cpu_current_domain[cpu] != COORDINATOR_DOMAIN) {
++ add_domain_switch(cpu_current_domain[cpu]);
++ }
+ /* Remember, only we can modify tail_pos */
+
+ available = get_slots(cpu_buf);
+@@ -512,16 +544,18 @@
+ for (i = 0; i < available; ++i) {
+ struct op_sample * s = &cpu_buf->buffer[cpu_buf->tail_pos];
+
+- if (is_code(s->eip)) {
+- if (s->event <= CPU_IS_KERNEL) {
+- /* kernel/userspace switch */
+- in_kernel = s->event;
++ if (is_code(s->eip) && !domain_switch) {
++ if (s->event <= CPU_MODE_XEN) {
++ /* xen/kernel/userspace switch */
++ cpu_mode = s->event;
+ if (state == sb_buffer_start)
+ state = sb_sample_start;
+- add_kernel_ctx_switch(s->event);
++ add_cpu_mode_switch(s->event);
+ } else if (s->event == CPU_TRACE_BEGIN) {
+ state = sb_bt_start;
+ add_trace_begin();
++ } else if (s->event == CPU_DOMAIN_SWITCH) {
++ domain_switch = 1;
+ } else {
+ struct mm_struct * oldmm = mm;
+
+@@ -535,19 +569,34 @@
+ add_user_ctx_switch(new, cookie);
+ }
+ } else {
+- if (state >= sb_bt_start &&
+- !add_sample(mm, s, in_kernel)) {
++ if (domain_switch) {
++ cpu_current_domain[cpu] = s->eip;
++ add_domain_switch(s->eip);
++ domain_switch = 0;
++ } else {
++ if (cpu_current_domain[cpu] !=
++ COORDINATOR_DOMAIN) {
++ add_sample_entry(s->eip, s->event);
++ }
++ else if (state >= sb_bt_start &&
++ !add_sample(mm, s, cpu_mode)) {
+ if (state == sb_bt_start) {
+ state = sb_bt_ignore;
+ atomic_inc(&oprofile_stats.bt_lost_no_mapping);
+ }
+ }
+ }
++ }
+
+ increment_tail(cpu_buf);
+ }
+ release_mm(mm);
+
++ /* We reset domain to COORDINATOR at each CPU switch */
++ if (cpu_current_domain[cpu] != COORDINATOR_DOMAIN) {
++ add_domain_switch(COORDINATOR_DOMAIN);
++ }
++
+ mark_done(cpu);
+
+ mutex_unlock(&buffer_mutex);
+--- linux-2.6.18.8/drivers/oprofile/cpu_buffer.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.c 2008-05-19 00:33:32.566023268 +0300
+@@ -6,6 +6,10 @@
+ *
+ * @author John Levon <levon@movementarian.org>
+ *
++ * Modified by Aravind Menon for Xen
++ * These modifications are:
++ * Copyright (C) 2005 Hewlett-Packard Co.
++ *
+ * Each CPU has a local buffer that stores PC value/event
+ * pairs. We also log context switches when we notice them.
+ * Eventually each CPU's buffer is processed into the global
+@@ -34,6 +38,8 @@
+ #define DEFAULT_TIMER_EXPIRE (HZ / 10)
+ static int work_enabled;
+
++static int32_t current_domain = COORDINATOR_DOMAIN;
++
+ void free_cpu_buffers(void)
+ {
+ int i;
+@@ -57,7 +63,7 @@
+ goto fail;
+
+ b->last_task = NULL;
+- b->last_is_kernel = -1;
++ b->last_cpu_mode = -1;
+ b->tracing = 0;
+ b->buffer_size = buffer_size;
+ b->tail_pos = 0;
+@@ -113,7 +119,7 @@
+ * collected will populate the buffer with proper
+ * values to initialize the buffer
+ */
+- cpu_buf->last_is_kernel = -1;
++ cpu_buf->last_cpu_mode = -1;
+ cpu_buf->last_task = NULL;
+ }
+
+@@ -163,13 +169,13 @@
+ * because of the head/tail separation of the writer and reader
+ * of the CPU buffer.
+ *
+- * is_kernel is needed because on some architectures you cannot
++ * cpu_mode is needed because on some architectures you cannot
+ * tell if you are in kernel or user space simply by looking at
+- * pc. We tag this in the buffer by generating kernel enter/exit
+- * events whenever is_kernel changes
++ * pc. We tag this in the buffer by generating kernel/user (and xen)
++ * enter events whenever cpu_mode changes
+ */
+ static int log_sample(struct oprofile_cpu_buffer * cpu_buf, unsigned long pc,
+- int is_kernel, unsigned long event)
++ int cpu_mode, unsigned long event)
+ {
+ struct task_struct * task;
+
+@@ -180,18 +186,18 @@
+ return 0;
+ }
+
+- is_kernel = !!is_kernel;
+-
+ task = current;
+
+ /* notice a switch from user->kernel or vice versa */
+- if (cpu_buf->last_is_kernel != is_kernel) {
+- cpu_buf->last_is_kernel = is_kernel;
+- add_code(cpu_buf, is_kernel);
++ if (cpu_buf->last_cpu_mode != cpu_mode) {
++ cpu_buf->last_cpu_mode = cpu_mode;
++ add_code(cpu_buf, cpu_mode);
+ }
+
+ /* notice a task switch */
+- if (cpu_buf->last_task != task) {
++ /* if not processing other domain samples */
++ if ((cpu_buf->last_task != task) &&
++ (current_domain == COORDINATOR_DOMAIN)) {
+ cpu_buf->last_task = task;
+ add_code(cpu_buf, (unsigned long)task);
+ }
+@@ -275,6 +281,25 @@
+ add_sample(cpu_buf, pc, 0);
+ }
+
++int oprofile_add_domain_switch(int32_t domain_id)
++{
++ struct oprofile_cpu_buffer * cpu_buf = &cpu_buffer[smp_processor_id()];
++
++ /* should have space for switching into and out of domain
++ (2 slots each) plus one sample and one cpu mode switch */
++ if (((nr_available_slots(cpu_buf) < 6) &&
++ (domain_id != COORDINATOR_DOMAIN)) ||
++ (nr_available_slots(cpu_buf) < 2))
++ return 0;
++
++ add_code(cpu_buf, CPU_DOMAIN_SWITCH);
++ add_sample(cpu_buf, domain_id, 0);
++
++ current_domain = domain_id;
++
++ return 1;
++}
++
+ /*
+ * This serves to avoid cpu buffer overflow, and makes sure
+ * the task mortuary progresses
+--- linux-2.6.18.8/drivers/oprofile/cpu_buffer.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/cpu_buffer.h 2008-05-19 00:33:32.594024882 +0300
+@@ -36,7 +36,7 @@
+ volatile unsigned long tail_pos;
+ unsigned long buffer_size;
+ struct task_struct * last_task;
+- int last_is_kernel;
++ int last_cpu_mode;
+ int tracing;
+ struct op_sample * buffer;
+ unsigned long sample_received;
+@@ -51,7 +51,10 @@
+ void cpu_buffer_reset(struct oprofile_cpu_buffer * cpu_buf);
+
+ /* transient events for the CPU buffer -> event buffer */
+-#define CPU_IS_KERNEL 1
+-#define CPU_TRACE_BEGIN 2
++#define CPU_MODE_USER 0
++#define CPU_MODE_KERNEL 1
++#define CPU_MODE_XEN 2
++#define CPU_TRACE_BEGIN 3
++#define CPU_DOMAIN_SWITCH 4
+
+ #endif /* OPROFILE_CPU_BUFFER_H */
+--- linux-2.6.18.8/drivers/oprofile/event_buffer.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/event_buffer.h 2008-05-19 00:33:32.598025113 +0300
+@@ -29,15 +29,20 @@
+ #define CPU_SWITCH_CODE 2
+ #define COOKIE_SWITCH_CODE 3
+ #define KERNEL_ENTER_SWITCH_CODE 4
+-#define KERNEL_EXIT_SWITCH_CODE 5
++#define USER_ENTER_SWITCH_CODE 5
+ #define MODULE_LOADED_CODE 6
+ #define CTX_TGID_CODE 7
+ #define TRACE_BEGIN_CODE 8
+ #define TRACE_END_CODE 9
++#define XEN_ENTER_SWITCH_CODE 10
++#define DOMAIN_SWITCH_CODE 11
+
+ #define INVALID_COOKIE ~0UL
+ #define NO_COOKIE 0UL
+
++/* Constant used to refer to coordinator domain (Xen) */
++#define COORDINATOR_DOMAIN -1
++
+ /* add data to the event buffer */
+ void add_event_entry(unsigned long data);
+
+--- linux-2.6.18.8/drivers/oprofile/oprof.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/oprof.c 2008-05-19 00:33:32.598025113 +0300
+@@ -5,6 +5,10 @@
+ * @remark Read the file COPYING
+ *
+ * @author John Levon <levon@movementarian.org>
++ *
++ * Modified by Aravind Menon for Xen
++ * These modifications are:
++ * Copyright (C) 2005 Hewlett-Packard Co.
+ */
+
+ #include <linux/kernel.h>
+@@ -33,6 +37,32 @@
+ */
+ static int timer = 0;
+
++int oprofile_set_active(int active_domains[], unsigned int adomains)
++{
++ int err;
++
++ if (!oprofile_ops.set_active)
++ return -EINVAL;
++
++ mutex_lock(&start_mutex);
++ err = oprofile_ops.set_active(active_domains, adomains);
++ mutex_unlock(&start_mutex);
++ return err;
++}
++
++int oprofile_set_passive(int passive_domains[], unsigned int pdomains)
++{
++ int err;
++
++ if (!oprofile_ops.set_passive)
++ return -EINVAL;
++
++ mutex_lock(&start_mutex);
++ err = oprofile_ops.set_passive(passive_domains, pdomains);
++ mutex_unlock(&start_mutex);
++ return err;
++}
++
+ int oprofile_setup(void)
+ {
+ int err;
+--- linux-2.6.18.8/drivers/oprofile/oprof.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/oprof.h 2008-05-19 00:33:32.598025113 +0300
+@@ -36,4 +36,7 @@
+
+ int oprofile_set_backtrace(unsigned long depth);
+
++int oprofile_set_active(int active_domains[], unsigned int adomains);
++int oprofile_set_passive(int passive_domains[], unsigned int pdomains);
++
+ #endif /* OPROF_H */
+--- linux-2.6.18.8/drivers/oprofile/oprofile_files.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/oprofile/oprofile_files.c 2008-05-19 00:33:32.598025113 +0300
+@@ -5,10 +5,16 @@
+ * @remark Read the file COPYING
+ *
+ * @author John Levon <levon@movementarian.org>
++ *
++ * Modified by Aravind Menon for Xen
++ * These modifications are:
++ * Copyright (C) 2005 Hewlett-Packard Co.
+ */
+
+ #include <linux/fs.h>
+ #include <linux/oprofile.h>
++#include <asm/uaccess.h>
++#include <linux/ctype.h>
+
+ #include "event_buffer.h"
+ #include "oprofile_stats.h"
+@@ -118,10 +124,201 @@
+ .write = dump_write,
+ };
+
++#define TMPBUFSIZE 512
++
++static unsigned int adomains = 0;
++static int active_domains[MAX_OPROF_DOMAINS + 1];
++static DEFINE_MUTEX(adom_mutex);
++
++static ssize_t adomain_write(struct file * file, char const __user * buf,
++ size_t count, loff_t * offset)
++{
++ char *tmpbuf;
++ char *startp, *endp;
++ int i;
++ unsigned long val;
++ ssize_t retval = count;
++
++ if (*offset)
++ return -EINVAL;
++ if (count > TMPBUFSIZE - 1)
++ return -EINVAL;
++
++ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
++ return -ENOMEM;
++
++ if (copy_from_user(tmpbuf, buf, count)) {
++ kfree(tmpbuf);
++ return -EFAULT;
++ }
++ tmpbuf[count] = 0;
++
++ mutex_lock(&adom_mutex);
++
++ startp = tmpbuf;
++ /* Parse one more than MAX_OPROF_DOMAINS, for easy error checking */
++ for (i = 0; i <= MAX_OPROF_DOMAINS; i++) {
++ val = simple_strtoul(startp, &endp, 0);
++ if (endp == startp)
++ break;
++ while (ispunct(*endp) || isspace(*endp))
++ endp++;
++ active_domains[i] = val;
++ if (active_domains[i] != val)
++ /* Overflow, force error below */
++ i = MAX_OPROF_DOMAINS + 1;
++ startp = endp;
++ }
++ /* Force error on trailing junk */
++ adomains = *startp ? MAX_OPROF_DOMAINS + 1 : i;
++
++ kfree(tmpbuf);
++
++ if (adomains > MAX_OPROF_DOMAINS
++ || oprofile_set_active(active_domains, adomains)) {
++ adomains = 0;
++ retval = -EINVAL;
++ }
++
++ mutex_unlock(&adom_mutex);
++ return retval;
++}
++
++static ssize_t adomain_read(struct file * file, char __user * buf,
++ size_t count, loff_t * offset)
++{
++ char * tmpbuf;
++ size_t len;
++ int i;
++ ssize_t retval;
++
++ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
++ return -ENOMEM;
++
++ mutex_lock(&adom_mutex);
++
++ len = 0;
++ for (i = 0; i < adomains; i++)
++ len += snprintf(tmpbuf + len,
++ len < TMPBUFSIZE ? TMPBUFSIZE - len : 0,
++ "%u ", active_domains[i]);
++ WARN_ON(len > TMPBUFSIZE);
++ if (len != 0 && len <= TMPBUFSIZE)
++ tmpbuf[len-1] = '\n';
++
++ mutex_unlock(&adom_mutex);
++
++ retval = simple_read_from_buffer(buf, count, offset, tmpbuf, len);
++
++ kfree(tmpbuf);
++ return retval;
++}
++
++
++static struct file_operations active_domain_ops = {
++ .read = adomain_read,
++ .write = adomain_write,
++};
++
++static unsigned int pdomains = 0;
++static int passive_domains[MAX_OPROF_DOMAINS];
++static DEFINE_MUTEX(pdom_mutex);
++
++static ssize_t pdomain_write(struct file * file, char const __user * buf,
++ size_t count, loff_t * offset)
++{
++ char *tmpbuf;
++ char *startp, *endp;
++ int i;
++ unsigned long val;
++ ssize_t retval = count;
++
++ if (*offset)
++ return -EINVAL;
++ if (count > TMPBUFSIZE - 1)
++ return -EINVAL;
++
++ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
++ return -ENOMEM;
++
++ if (copy_from_user(tmpbuf, buf, count)) {
++ kfree(tmpbuf);
++ return -EFAULT;
++ }
++ tmpbuf[count] = 0;
++
++ mutex_lock(&pdom_mutex);
++
++ startp = tmpbuf;
++ /* Parse one more than MAX_OPROF_DOMAINS, for easy error checking */
++ for (i = 0; i <= MAX_OPROF_DOMAINS; i++) {
++ val = simple_strtoul(startp, &endp, 0);
++ if (endp == startp)
++ break;
++ while (ispunct(*endp) || isspace(*endp))
++ endp++;
++ passive_domains[i] = val;
++ if (passive_domains[i] != val)
++ /* Overflow, force error below */
++ i = MAX_OPROF_DOMAINS + 1;
++ startp = endp;
++ }
++ /* Force error on trailing junk */
++ pdomains = *startp ? MAX_OPROF_DOMAINS + 1 : i;
++
++ kfree(tmpbuf);
++
++ if (pdomains > MAX_OPROF_DOMAINS
++ || oprofile_set_passive(passive_domains, pdomains)) {
++ pdomains = 0;
++ retval = -EINVAL;
++ }
++
++ mutex_unlock(&pdom_mutex);
++ return retval;
++}
++
++static ssize_t pdomain_read(struct file * file, char __user * buf,
++ size_t count, loff_t * offset)
++{
++ char * tmpbuf;
++ size_t len;
++ int i;
++ ssize_t retval;
++
++ if (!(tmpbuf = kmalloc(TMPBUFSIZE, GFP_KERNEL)))
++ return -ENOMEM;
++
++ mutex_lock(&pdom_mutex);
++
++ len = 0;
++ for (i = 0; i < pdomains; i++)
++ len += snprintf(tmpbuf + len,
++ len < TMPBUFSIZE ? TMPBUFSIZE - len : 0,
++ "%u ", passive_domains[i]);
++ WARN_ON(len > TMPBUFSIZE);
++ if (len != 0 && len <= TMPBUFSIZE)
++ tmpbuf[len-1] = '\n';
++
++ mutex_unlock(&pdom_mutex);
++
++ retval = simple_read_from_buffer(buf, count, offset, tmpbuf, len);
++
++ kfree(tmpbuf);
++ return retval;
++}
++
++static struct file_operations passive_domain_ops = {
++ .read = pdomain_read,
++ .write = pdomain_write,
++};
++
+ void oprofile_create_files(struct super_block * sb, struct dentry * root)
+ {
+ oprofilefs_create_file(sb, root, "enable", &enable_fops);
+ oprofilefs_create_file_perm(sb, root, "dump", &dump_fops, 0666);
++ oprofilefs_create_file(sb, root, "active_domains", &active_domain_ops);
++ oprofilefs_create_file(sb, root, "passive_domains", &passive_domain_ops);
+ oprofilefs_create_file(sb, root, "buffer", &event_buffer_fops);
+ oprofilefs_create_ulong(sb, root, "buffer_size", &fs_buffer_size);
+ oprofilefs_create_ulong(sb, root, "buffer_watershed", &fs_buffer_watershed);
+--- linux-2.6.18.8/drivers/pci/bus.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/pci/bus.c 2008-05-19 00:33:32.938044711 +0300
+@@ -17,6 +17,8 @@
+
+ #include "pci.h"
+
++extern int pci_mem_align;
++
+ /**
+ * pci_bus_alloc_resource - allocate a resource from a parent bus
+ * @bus: PCI bus
+@@ -44,6 +46,11 @@
+
+ type_mask |= IORESOURCE_IO | IORESOURCE_MEM;
+
++ /* If the boot parameter 'pci-mem-align' was specified then we need to
++ align the memory addresses, at page size alignment. */
++ if (pci_mem_align && (align < (PAGE_SIZE-1)))
++ align = PAGE_SIZE - 1;
++
+ for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
+ struct resource *r = bus->resource[i];
+ if (!r)
+--- linux-2.6.18.8/drivers/pci/msi-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/pci/msi-xen.c 2008-05-19 00:33:33.654085984 +0300
+@@ -0,0 +1,721 @@
++/*
++ * File: msi.c
++ * Purpose: PCI Message Signaled Interrupt (MSI)
++ *
++ * Copyright (C) 2003-2004 Intel
++ * Copyright (C) Tom Long Nguyen (tom.l.nguyen@intel.com)
++ */
++
++#include <linux/mm.h>
++#include <linux/irq.h>
++#include <linux/interrupt.h>
++#include <linux/init.h>
++#include <linux/ioport.h>
++#include <linux/smp_lock.h>
++#include <linux/pci.h>
++#include <linux/proc_fs.h>
++
++#include <asm/errno.h>
++#include <asm/io.h>
++#include <asm/smp.h>
++
++#include "pci.h"
++#include "msi.h"
++
++static int pci_msi_enable = 1;
++
++static struct msi_ops *msi_ops;
++
++int msi_register(struct msi_ops *ops)
++{
++ msi_ops = ops;
++ return 0;
++}
++
++static LIST_HEAD(msi_dev_head);
++DEFINE_SPINLOCK(msi_dev_lock);
++
++struct msi_dev_list {
++ struct pci_dev *dev;
++ struct list_head list;
++ spinlock_t pirq_list_lock;
++ struct list_head pirq_list_head;
++};
++
++struct msi_pirq_entry {
++ struct list_head list;
++ int pirq;
++ int entry_nr;
++};
++
++static struct msi_dev_list *get_msi_dev_pirq_list(struct pci_dev *dev)
++{
++ struct msi_dev_list *msi_dev_list, *ret = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&msi_dev_lock, flags);
++
++ list_for_each_entry(msi_dev_list, &msi_dev_head, list)
++ if ( msi_dev_list->dev == dev )
++ ret = msi_dev_list;
++
++ if ( ret ) {
++ spin_unlock_irqrestore(&msi_dev_lock, flags);
++ return ret;
++ }
++
++ /* Has not allocate msi_dev until now. */
++ ret = kmalloc(sizeof(struct msi_dev_list), GFP_ATOMIC);
++
++ /* Failed to allocate msi_dev structure */
++ if ( !ret ) {
++ spin_unlock_irqrestore(&msi_dev_lock, flags);
++ return NULL;
++ }
++
++ spin_lock_init(&ret->pirq_list_lock);
++ INIT_LIST_HEAD(&ret->pirq_list_head);
++ list_add_tail(&ret->list, &msi_dev_head);
++ spin_unlock_irqrestore(&msi_dev_lock, flags);
++ return ret;
++}
++
++static int attach_pirq_entry(int pirq, int entry_nr,
++ struct msi_dev_list *msi_dev_entry)
++{
++ struct msi_pirq_entry *entry = kmalloc(sizeof(*entry), GFP_ATOMIC);
++ unsigned long flags;
++
++ if (!entry)
++ return -ENOMEM;
++ entry->pirq = pirq;
++ entry->entry_nr = entry_nr;
++ spin_lock_irqsave(&msi_dev_entry->pirq_list_lock, flags);
++ list_add_tail(&entry->list, &msi_dev_entry->pirq_list_head);
++ spin_unlock_irqrestore(&msi_dev_entry->pirq_list_lock, flags);
++ return 0;
++}
++
++static void detach_pirq_entry(int entry_nr,
++ struct msi_dev_list *msi_dev_entry)
++{
++ unsigned long flags;
++ struct msi_pirq_entry *pirq_entry;
++
++ list_for_each_entry(pirq_entry, &msi_dev_entry->pirq_list_head, list) {
++ if (pirq_entry->entry_nr == entry_nr) {
++ spin_lock_irqsave(&msi_dev_entry->pirq_list_lock, flags);
++ list_del(&pirq_entry->list);
++ spin_unlock_irqrestore(&msi_dev_entry->pirq_list_lock, flags);
++ kfree(pirq_entry);
++ return;
++ }
++ }
++}
++
++/*
++ * pciback will provide device's owner
++ */
++int (*get_owner)(struct pci_dev *dev);
++
++int register_msi_get_owner(int (*func)(struct pci_dev *dev))
++{
++ if (get_owner) {
++ printk(KERN_WARNING "register msi_get_owner again\n");
++ return -EEXIST;
++ }
++ get_owner = func;
++ return 0;
++}
++
++int unregister_msi_get_owner(int (*func)(struct pci_dev *dev))
++{
++ if (get_owner == func)
++ get_owner = NULL;
++ return 0;
++}
++
++static int msi_get_dev_owner(struct pci_dev *dev)
++{
++ int owner = DOMID_SELF;
++
++ BUG_ON(!is_initial_xendomain());
++ if (get_owner && (owner = get_owner(dev)) >=0 ) {
++ printk(KERN_INFO "get owner for dev %x get %x \n",
++ dev->devfn, owner);
++ return owner;
++ }
++ else
++ return DOMID_SELF;
++}
++
++static int msi_unmap_pirq(struct pci_dev *dev, int pirq)
++{
++ struct physdev_unmap_pirq unmap;
++ int rc;
++ domid_t domid = DOMID_SELF;
++
++ domid = msi_get_dev_owner(dev);
++ unmap.domid = domid;
++ unmap.pirq = pirq;
++
++ if ((rc = HYPERVISOR_physdev_op(PHYSDEVOP_unmap_pirq, &unmap)))
++ printk(KERN_WARNING "unmap irq %x failed\n", pirq);
++
++ if (rc < 0)
++ return rc;
++ return 0;
++}
++
++/*
++ * Protected by msi_lock
++ */
++static int msi_map_pirq_to_vector(struct pci_dev *dev, int pirq,
++ int entry_nr, int msi)
++{
++ struct physdev_map_pirq map_irq;
++ int rc;
++ domid_t domid = DOMID_SELF;
++
++ domid = msi_get_dev_owner(dev);
++
++ map_irq.domid = domid;
++ map_irq.type = MAP_PIRQ_TYPE_MSI;
++ map_irq.index = -1;
++ map_irq.pirq = pirq;
++ map_irq.msi_info.bus = dev->bus->number;
++ map_irq.msi_info.devfn = dev->devfn;
++ map_irq.msi_info.entry_nr = entry_nr;
++ map_irq.msi_info.msi = msi;
++
++ if ((rc = HYPERVISOR_physdev_op(PHYSDEVOP_map_pirq, &map_irq)))
++ printk(KERN_WARNING "map irq failed\n");
++
++ if (rc < 0)
++ return rc;
++
++ return map_irq.pirq;
++}
++
++static int msi_map_vector(struct pci_dev *dev, int entry_nr, int msi)
++{
++ return msi_map_pirq_to_vector(dev, -1, entry_nr, msi);
++}
++
++static int msi_init(void)
++{
++ static int status = 0;
++
++ if (pci_msi_quirk) {
++ pci_msi_enable = 0;
++ printk(KERN_WARNING "PCI: MSI quirk detected. MSI disabled.\n");
++ status = -EINVAL;
++ }
++
++ return status;
++}
++
++void pci_scan_msi_device(struct pci_dev *dev) { }
++
++void disable_msi_mode(struct pci_dev *dev, int pos, int type)
++{
++ u16 control;
++
++ pci_read_config_word(dev, msi_control_reg(pos), &control);
++ if (type == PCI_CAP_ID_MSI) {
++ /* Set enabled bits to single MSI & enable MSI_enable bit */
++ msi_disable(control);
++ pci_write_config_word(dev, msi_control_reg(pos), control);
++ dev->msi_enabled = 0;
++ } else {
++ msix_disable(control);
++ pci_write_config_word(dev, msi_control_reg(pos), control);
++ dev->msix_enabled = 0;
++ }
++ if (pci_find_capability(dev, PCI_CAP_ID_EXP)) {
++ /* PCI Express Endpoint device detected */
++ pci_intx(dev, 1); /* enable intx */
++ }
++}
++
++static void enable_msi_mode(struct pci_dev *dev, int pos, int type)
++{
++ u16 control;
++
++ pci_read_config_word(dev, msi_control_reg(pos), &control);
++ if (type == PCI_CAP_ID_MSI) {
++ /* Set enabled bits to single MSI & enable MSI_enable bit */
++ msi_enable(control, 1);
++ pci_write_config_word(dev, msi_control_reg(pos), control);
++ dev->msi_enabled = 1;
++ } else {
++ msix_enable(control);
++ pci_write_config_word(dev, msi_control_reg(pos), control);
++ dev->msix_enabled = 1;
++ }
++ if (pci_find_capability(dev, PCI_CAP_ID_EXP)) {
++ /* PCI Express Endpoint device detected */
++ pci_intx(dev, 0); /* disable intx */
++ }
++}
++
++#ifdef CONFIG_PM
++int pci_save_msi_state(struct pci_dev *dev)
++{
++ int pos;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
++ if (pos <= 0 || dev->no_msi)
++ return 0;
++
++ if (!dev->msi_enabled)
++ return 0;
++
++ /* Restore dev->irq to its default pin-assertion vector */
++ msi_unmap_pirq(dev, dev->irq);
++ /* Disable MSI mode */
++ disable_msi_mode(dev, pos, PCI_CAP_ID_MSI);
++ /* Set the flags for use of restore */
++ dev->msi_enabled = 1;
++ return 0;
++}
++
++void pci_restore_msi_state(struct pci_dev *dev)
++{
++ int pos, pirq;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
++ if (pos <= 0)
++ return;
++
++ if (!dev->msi_enabled)
++ return;
++
++ pirq = msi_map_pirq_to_vector(dev, dev->irq, 0, 1);
++ if (pirq < 0)
++ return;
++ enable_msi_mode(dev, pos, PCI_CAP_ID_MSI);
++}
++
++int pci_save_msix_state(struct pci_dev *dev)
++{
++ int pos;
++ unsigned long flags;
++ struct msi_dev_list *msi_dev_entry;
++ struct msi_pirq_entry *pirq_entry, *tmp;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
++ if (pos <= 0 || dev->no_msi)
++ return 0;
++
++ /* save the capability */
++ if (!dev->msix_enabled)
++ return 0;
++
++ msi_dev_entry = get_msi_dev_pirq_list(dev);
++
++ spin_lock_irqsave(&msi_dev_entry->pirq_list_lock, flags);
++ list_for_each_entry_safe(pirq_entry, tmp,
++ &msi_dev_entry->pirq_list_head, list)
++ msi_unmap_pirq(dev, pirq_entry->pirq);
++ spin_unlock_irqrestore(&msi_dev_entry->pirq_list_lock, flags);
++
++ disable_msi_mode(dev, pos, PCI_CAP_ID_MSIX);
++ /* Set the flags for use of restore */
++ dev->msix_enabled = 1;
++
++ return 0;
++}
++
++void pci_restore_msix_state(struct pci_dev *dev)
++{
++ int pos;
++ unsigned long flags;
++ struct msi_dev_list *msi_dev_entry;
++ struct msi_pirq_entry *pirq_entry, *tmp;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
++ if (pos <= 0)
++ return;
++
++ if (!dev->msix_enabled)
++ return;
++
++ msi_dev_entry = get_msi_dev_pirq_list(dev);
++
++ spin_lock_irqsave(&msi_dev_entry->pirq_list_lock, flags);
++ list_for_each_entry_safe(pirq_entry, tmp,
++ &msi_dev_entry->pirq_list_head, list)
++ msi_map_pirq_to_vector(dev, pirq_entry->pirq, pirq_entry->entry_nr, 0);
++ spin_unlock_irqrestore(&msi_dev_entry->pirq_list_lock, flags);
++
++ enable_msi_mode(dev, pos, PCI_CAP_ID_MSIX);
++}
++#endif
++
++/**
++ * msi_capability_init - configure device's MSI capability structure
++ * @dev: pointer to the pci_dev data structure of MSI device function
++ *
++ * Setup the MSI capability structure of device function with a single
++ * MSI vector, regardless of device function is capable of handling
++ * multiple messages. A return of zero indicates the successful setup
++ * of an entry zero with the new MSI vector or non-zero for otherwise.
++ **/
++static int msi_capability_init(struct pci_dev *dev)
++{
++ int pos, pirq;
++ u16 control;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
++ pci_read_config_word(dev, msi_control_reg(pos), &control);
++
++ pirq = msi_map_vector(dev, 0, 1);
++ if (pirq < 0)
++ return -EBUSY;
++
++ dev->irq = pirq;
++ /* Set MSI enabled bits */
++ enable_msi_mode(dev, pos, PCI_CAP_ID_MSI);
++ dev->msi_enabled = 1;
++
++ return 0;
++}
++
++/**
++ * msix_capability_init - configure device's MSI-X capability
++ * @dev: pointer to the pci_dev data structure of MSI-X device function
++ * @entries: pointer to an array of struct msix_entry entries
++ * @nvec: number of @entries
++ *
++ * Setup the MSI-X capability structure of device function with a
++ * single MSI-X vector. A return of zero indicates the successful setup of
++ * requested MSI-X entries with allocated vectors or non-zero for otherwise.
++ **/
++static int msix_capability_init(struct pci_dev *dev,
++ struct msix_entry *entries, int nvec)
++{
++ int pirq, i, j, mapped, pos;
++ struct msi_dev_list *msi_dev_entry = get_msi_dev_pirq_list(dev);
++ struct msi_pirq_entry *pirq_entry;
++
++ if (!msi_dev_entry)
++ return -ENOMEM;
++
++ /* MSI-X Table Initialization */
++ for (i = 0; i < nvec; i++) {
++ mapped = 0;
++ list_for_each_entry(pirq_entry, &msi_dev_entry->pirq_list_head, list) {
++ if (pirq_entry->entry_nr == entries[i].entry) {
++ printk(KERN_WARNING "msix entry %d for dev %02x:%02x:%01x are \
++ not freed before acquire again.\n", entries[i].entry,
++ dev->bus->number, PCI_SLOT(dev->devfn),
++ PCI_FUNC(dev->devfn));
++ (entries + i)->vector = pirq_entry->pirq;
++ mapped = 1;
++ break;
++ }
++ }
++ if (mapped)
++ continue;
++ pirq = msi_map_vector(dev, entries[i].entry, 0);
++ if (pirq < 0)
++ break;
++ attach_pirq_entry(pirq, entries[i].entry, msi_dev_entry);
++ (entries + i)->vector = pirq;
++ }
++
++ if (i != nvec) {
++ for (j = --i; j >= 0; j--) {
++ msi_unmap_pirq(dev, entries[j].vector);
++ detach_pirq_entry(entries[j].entry, msi_dev_entry);
++ entries[j].vector = 0;
++ }
++ return -EBUSY;
++ }
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
++ enable_msi_mode(dev, pos, PCI_CAP_ID_MSIX);
++ dev->msix_enabled = 1;
++
++ return 0;
++}
++
++/**
++ * pci_enable_msi - configure device's MSI capability structure
++ * @dev: pointer to the pci_dev data structure of MSI device function
++ *
++ * Setup the MSI capability structure of device function with
++ * a single MSI vector upon its software driver call to request for
++ * MSI mode enabled on its hardware device function. A return of zero
++ * indicates the successful setup of an entry zero with the new MSI
++ * vector or non-zero for otherwise.
++ **/
++extern int pci_frontend_enable_msi(struct pci_dev *dev);
++int pci_enable_msi(struct pci_dev* dev)
++{
++ struct pci_bus *bus;
++ int pos, temp, status = -EINVAL;
++
++ if (!pci_msi_enable || !dev)
++ return status;
++
++ if (dev->no_msi)
++ return status;
++
++ for (bus = dev->bus; bus; bus = bus->parent)
++ if (bus->bus_flags & PCI_BUS_FLAGS_NO_MSI)
++ return -EINVAL;
++
++ status = msi_init();
++ if (status < 0)
++ return status;
++
++#ifdef CONFIG_XEN_PCIDEV_FRONTEND
++ if (!is_initial_xendomain())
++ {
++ int ret;
++
++ temp = dev->irq;
++ ret = pci_frontend_enable_msi(dev);
++ if (ret)
++ return ret;
++
++ dev->irq_old = temp;
++
++ return ret;
++ }
++#endif
++
++ temp = dev->irq;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
++ if (!pos)
++ return -EINVAL;
++
++ /* Check whether driver already requested for MSI-X vectors */
++ if (dev->msix_enabled) {
++ printk(KERN_INFO "PCI: %s: Can't enable MSI. "
++ "Device already has MSI-X vectors assigned\n",
++ pci_name(dev));
++ dev->irq = temp;
++ return -EINVAL;
++ }
++
++ status = msi_capability_init(dev);
++ if ( !status )
++ dev->irq_old = temp;
++ else
++ dev->irq = temp;
++
++ return status;
++}
++
++extern void pci_frontend_disable_msi(struct pci_dev* dev);
++void pci_disable_msi(struct pci_dev* dev)
++{
++ int pos;
++ int pirq;
++
++ if (!pci_msi_enable)
++ return;
++ if (!dev)
++ return;
++
++#ifdef CONFIG_XEN_PCIDEV_FRONTEND
++ if (!is_initial_xendomain()) {
++ pci_frontend_disable_msi(dev);
++ dev->irq = dev->irq_old;
++ return;
++ }
++#endif
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
++ if (!pos)
++ return;
++
++ pirq = dev->irq;
++ /* Restore dev->irq to its default pin-assertion vector */
++ dev->irq = dev->irq_old;
++ msi_unmap_pirq(dev, pirq);
++
++ /* Disable MSI mode */
++ disable_msi_mode(dev, pos, PCI_CAP_ID_MSI);
++}
++
++/**
++ * pci_enable_msix - configure device's MSI-X capability structure
++ * @dev: pointer to the pci_dev data structure of MSI-X device function
++ * @entries: pointer to an array of MSI-X entries
++ * @nvec: number of MSI-X vectors requested for allocation by device driver
++ *
++ * Setup the MSI-X capability structure of device function with the number
++ * of requested vectors upon its software driver call to request for
++ * MSI-X mode enabled on its hardware device function. A return of zero
++ * indicates the successful configuration of MSI-X capability structure
++ * with new allocated MSI-X vectors. A return of < 0 indicates a failure.
++ * Or a return of > 0 indicates that driver request is exceeding the number
++ * of vectors available. Driver should use the returned value to re-send
++ * its request.
++ **/
++extern int pci_frontend_enable_msix(struct pci_dev *dev,
++ struct msix_entry *entries, int nvec);
++int pci_enable_msix(struct pci_dev* dev, struct msix_entry *entries, int nvec)
++{
++ struct pci_bus *bus;
++ int status, pos, nr_entries;
++ int i, j, temp;
++ u16 control;
++
++ if (!pci_msi_enable || !dev || !entries)
++ return -EINVAL;
++
++ if (dev->no_msi)
++ return -EINVAL;
++
++ for (bus = dev->bus; bus; bus = bus->parent)
++ if (bus->bus_flags & PCI_BUS_FLAGS_NO_MSI)
++ return -EINVAL;
++
++#ifdef CONFIG_XEN_PCIDEV_FRONTEND
++ if (!is_initial_xendomain()) {
++ int ret;
++
++ ret = pci_frontend_enable_msix(dev, entries, nvec);
++ if (ret) {
++ printk("get %x from pci_frontend_enable_msix\n", ret);
++ return ret;
++ }
++
++ return 0;
++ }
++#endif
++
++ status = msi_init();
++ if (status < 0)
++ return status;
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
++ if (!pos)
++ return -EINVAL;
++
++ pci_read_config_word(dev, msi_control_reg(pos), &control);
++ nr_entries = multi_msix_capable(control);
++ if (nvec > nr_entries)
++ return -EINVAL;
++
++ /* Check for any invalid entries */
++ for (i = 0; i < nvec; i++) {
++ if (entries[i].entry >= nr_entries)
++ return -EINVAL; /* invalid entry */
++ for (j = i + 1; j < nvec; j++) {
++ if (entries[i].entry == entries[j].entry)
++ return -EINVAL; /* duplicate entry */
++ }
++ }
++
++ temp = dev->irq;
++ /* Check whether driver already requested for MSI vector */
++ if (dev->msi_enabled) {
++ printk(KERN_INFO "PCI: %s: Can't enable MSI-X. "
++ "Device already has an MSI vector assigned\n",
++ pci_name(dev));
++ dev->irq = temp;
++ return -EINVAL;
++ }
++
++ status = msix_capability_init(dev, entries, nvec);
++
++ if ( !status )
++ dev->irq_old = temp;
++ else
++ dev->irq = temp;
++
++ return status;
++}
++
++extern void pci_frontend_disable_msix(struct pci_dev* dev);
++void pci_disable_msix(struct pci_dev* dev)
++{
++ int pos;
++ u16 control;
++
++
++ if (!pci_msi_enable)
++ return;
++ if (!dev)
++ return;
++
++#ifdef CONFIG_XEN_PCIDEV_FRONTEND
++ if (!is_initial_xendomain()) {
++ pci_frontend_disable_msix(dev);
++ dev->irq = dev->irq_old;
++ return;
++ }
++#endif
++
++ pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
++ if (!pos)
++ return;
++
++ pci_read_config_word(dev, msi_control_reg(pos), &control);
++ if (!(control & PCI_MSIX_FLAGS_ENABLE))
++ return;
++
++ msi_remove_pci_irq_vectors(dev);
++
++ /* Disable MSI mode */
++ disable_msi_mode(dev, pos, PCI_CAP_ID_MSIX);
++}
++
++/**
++ * msi_remove_pci_irq_vectors - reclaim MSI(X) vectors to unused state
++ * @dev: pointer to the pci_dev data structure of MSI(X) device function
++ *
++ * Being called during hotplug remove, from which the device function
++ * is hot-removed. All previous assigned MSI/MSI-X vectors, if
++ * allocated for this device function, are reclaimed to unused state,
++ * which may be used later on.
++ **/
++void msi_remove_pci_irq_vectors(struct pci_dev* dev)
++{
++ unsigned long flags;
++ struct msi_dev_list *msi_dev_entry;
++ struct msi_pirq_entry *pirq_entry, *tmp;
++
++ if (!pci_msi_enable || !dev)
++ return;
++
++ msi_dev_entry = get_msi_dev_pirq_list(dev);
++
++ spin_lock_irqsave(&msi_dev_entry->pirq_list_lock, flags);
++ if (!list_empty(&msi_dev_entry->pirq_list_head))
++ {
++ printk(KERN_WARNING "msix pirqs for dev %02x:%02x:%01x are not freed \
++ before acquire again.\n", dev->bus->number, PCI_SLOT(dev->devfn),
++ PCI_FUNC(dev->devfn));
++ list_for_each_entry_safe(pirq_entry, tmp,
++ &msi_dev_entry->pirq_list_head, list) {
++ msi_unmap_pirq(dev, pirq_entry->pirq);
++ list_del(&pirq_entry->list);
++ kfree(pirq_entry);
++ }
++ }
++ spin_unlock_irqrestore(&msi_dev_entry->pirq_list_lock, flags);
++ dev->irq = dev->irq_old;
++}
++
++void pci_no_msi(void)
++{
++ pci_msi_enable = 0;
++}
++
++EXPORT_SYMBOL(pci_enable_msi);
++EXPORT_SYMBOL(pci_disable_msi);
++EXPORT_SYMBOL(pci_enable_msix);
++EXPORT_SYMBOL(pci_disable_msix);
++#ifdef CONFIG_XEN
++EXPORT_SYMBOL(register_msi_get_owner);
++EXPORT_SYMBOL(unregister_msi_get_owner);
++#endif
++
+--- linux-2.6.18.8/drivers/pci/msi.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/pci/msi.h 2008-05-19 00:33:33.654085984 +0300
+@@ -84,6 +84,11 @@
+ extern void (*interrupt[NR_IRQS])(void);
+ extern int pci_vector_resources(int last, int nr_released);
+
++#ifdef CONFIG_XEN
++extern int unregister_msi_get_owner(int (*func)(struct pci_dev *dev));
++extern int register_msi_get_owner(int (*func)(struct pci_dev *dev));
++#endif
++
+ /*
+ * MSI-X Address Register
+ */
+--- linux-2.6.18.8/drivers/pci/quirks.c 2008-05-19 00:42:34.057238961 +0300
++++ linux-2.6.18-xen.hg/drivers/pci/quirks.c 2008-05-19 00:33:33.770092671 +0300
+@@ -23,6 +23,40 @@
+ #include <linux/acpi.h>
+ #include "pci.h"
+
++/* A global flag which signals if we should page-align PCI mem windows. */
++int pci_mem_align = 0;
++
++static int __init set_pci_mem_align(char *str)
++{
++ pci_mem_align = 1;
++ return 1;
++}
++__setup("pci-mem-align", set_pci_mem_align);
++
++/* This quirk function enables us to force all memory resources which are
++ * assigned to PCI devices, to be page-aligned.
++ */
++static void __devinit quirk_align_mem_resources(struct pci_dev *dev)
++{
++ int i;
++ struct resource *r;
++ resource_size_t old_start;
++
++ if (!pci_mem_align)
++ return;
++
++ for (i=0; i < DEVICE_COUNT_RESOURCE; i++) {
++ r = &dev->resource[i];
++ if ((r == NULL) || !(r->flags & IORESOURCE_MEM))
++ continue;
++
++ old_start = r->start;
++ r->start = (r->start + PAGE_SIZE - 1) & ~(PAGE_SIZE - 1);
++ r->end = r->end - (old_start - r->start);
++ }
++}
++DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, quirk_align_mem_resources);
++
+ /* The Mellanox Tavor device gives false positive parity errors
+ * Mark this device with a broken_parity_status, to allow
+ * PCI scanning code to "skip" this now blacklisted device.
+@@ -1494,10 +1528,11 @@
+
+ static void __devinit quirk_e100_interrupt(struct pci_dev *dev)
+ {
+- u16 command;
++ u16 command, pmcsr;
+ u32 bar;
+ u8 __iomem *csr;
+ u8 cmd_hi;
++ int pm;
+
+ switch (dev->device) {
+ /* PCI IDs taken from drivers/net/e100.c */
+@@ -1532,6 +1567,17 @@
+ if (!(command & PCI_COMMAND_MEMORY) || !bar)
+ return;
+
++ /*
++ * Check that the device is in the D0 power state. If it's not,
++ * there is no point to look any further.
++ */
++ pm = pci_find_capability(dev, PCI_CAP_ID_PM);
++ if (pm) {
++ pci_read_config_word(dev, pm + PCI_PM_CTRL, &pmcsr);
++ if ((pmcsr & PCI_PM_CTRL_STATE_MASK) != PCI_D0)
++ return;
++ }
++
+ csr = ioremap(bar, 8);
+ if (!csr) {
+ printk(KERN_WARNING "PCI: Can't map %s e100 registers\n",
+--- linux-2.6.18.8/drivers/pnp/manager.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/pnp/manager.c 2008-05-19 00:33:34.394128641 +0300
+@@ -168,7 +168,7 @@
+ return 0;
+ }
+
+-static int pnp_assign_dma(struct pnp_dev *dev, struct pnp_dma *rule, int idx)
++static void pnp_assign_dma(struct pnp_dev *dev, struct pnp_dma *rule, int idx)
+ {
+ resource_size_t *start, *end;
+ unsigned long *flags;
+@@ -179,18 +179,14 @@
+ 1, 3, 5, 6, 7, 0, 2, 4
+ };
+
+- if (!dev || !rule)
+- return -EINVAL;
+-
+ if (idx >= PNP_MAX_DMA) {
+ pnp_err("More than 2 dmas is incompatible with pnp specifications.");
+- /* pretend we were successful so at least the manager won't try again */
+- return 1;
++ return;
+ }
+
+ /* check if this resource has been manually set, if so skip */
+ if (!(dev->res.dma_resource[idx].flags & IORESOURCE_AUTO))
+- return 1;
++ return;
+
+ start = &dev->res.dma_resource[idx].start;
+ end = &dev->res.dma_resource[idx].end;
+@@ -200,19 +196,17 @@
+ *flags |= rule->flags | IORESOURCE_DMA;
+ *flags &= ~IORESOURCE_UNSET;
+
+- if (!rule->map) {
+- *flags |= IORESOURCE_DISABLED;
+- return 1; /* skip disabled resource requests */
+- }
+-
+ for (i = 0; i < 8; i++) {
+ if(rule->map & (1<<xtab[i])) {
+ *start = *end = xtab[i];
+ if(pnp_check_dma(dev, idx))
+- return 1;
++ return;
+ }
+ }
+- return 0;
++#ifdef MAX_DMA_CHANNELS
++ *start = *end = MAX_DMA_CHANNELS;
++#endif
++ *flags |= IORESOURCE_UNSET | IORESOURCE_DISABLED;
+ }
+
+ /**
+@@ -331,8 +325,7 @@
+ irq = irq->next;
+ }
+ while (dma) {
+- if (!pnp_assign_dma(dev, dma, ndma))
+- goto fail;
++ pnp_assign_dma(dev, dma, ndma);
+ ndma++;
+ dma = dma->next;
+ }
+@@ -367,8 +360,7 @@
+ irq = irq->next;
+ }
+ while (dma) {
+- if (!pnp_assign_dma(dev, dma, ndma))
+- goto fail;
++ pnp_assign_dma(dev, dma, ndma);
+ ndma++;
+ dma = dma->next;
+ }
+--- linux-2.6.18.8/drivers/scsi/ahci.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/scsi/ahci.c 2008-05-19 00:33:35.798209573 +0300
+@@ -317,6 +317,34 @@
+ board_ahci }, /* ICH8M */
+ { PCI_VENDOR_ID_INTEL, 0x282a, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+ board_ahci }, /* ICH8M */
++ { PCI_VENDOR_ID_INTEL, 0x2922, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x2923, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x2924, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x2925, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x2927, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x2929, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9M */
++ { PCI_VENDOR_ID_INTEL, 0x292a, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9M */
++ { PCI_VENDOR_ID_INTEL, 0x292b, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9M */
++ { PCI_VENDOR_ID_INTEL, 0x292f, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9M */
++ { PCI_VENDOR_ID_INTEL, 0x294d, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9 */
++ { PCI_VENDOR_ID_INTEL, 0x294e, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH9M */
++ { PCI_VENDOR_ID_INTEL, 0x3a02, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH10 */
++ { PCI_VENDOR_ID_INTEL, 0x3a05, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH10 */
++ { PCI_VENDOR_ID_INTEL, 0x3a25, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
++ board_ahci }, /* ICH10 */
+
+ /* JMicron */
+ { 0x197b, 0x2360, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+--- linux-2.6.18.8/drivers/scsi/ata_piix.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/scsi/ata_piix.c 2008-05-19 00:33:37.286295348 +0300
+@@ -125,6 +125,8 @@
+ ich6m_sata_ahci = 6,
+ ich7m_sata_ahci = 7,
+ ich8_sata_ahci = 8,
++ ich9_sata_ahci = 9,
++ ich8_2port_sata = 10,
+
+ /* constants for mapping table */
+ P0 = 0, /* port 0 */
+@@ -192,12 +194,32 @@
+ { 0x8086, 0x27c4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich7m_sata_ahci },
+ /* Enterprise Southbridge 2 (where's the datasheet?) */
+ { 0x8086, 0x2680, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich6_sata_ahci },
+- /* SATA Controller 1 IDE (ICH8, no datasheet yet) */
++ /* SATA Controller 1 IDE (ICH8) */
+ { 0x8086, 0x2820, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
+- /* SATA Controller 2 IDE (ICH8, ditto) */
+- { 0x8086, 0x2825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
++ /* SATA Controller 2 IDE (ICH8) */
++ { 0x8086, 0x2825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_2port_sata },
+ /* Mobile SATA Controller IDE (ICH8M, ditto) */
+ { 0x8086, 0x2828, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
++ /* SATA Controller 1 IDE (ICH9) */
++ { 0x8086, 0x2920, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* SATA Controller 1 IDE (ICH9) */
++ { 0x8086, 0x2921, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* SATA Controller 2 IDE (ICH9) */
++ { 0x8086, 0x2926, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* Mobile SATA Controller 1 IDE (ICH9M) */
++ { 0x8086, 0x2928, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* Mobile SATA Controller 2 IDE (ICH9M) */
++ { 0x8086, 0x292d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* Mobile SATA Controller 2 IDE (ICH9M) */
++ { 0x8086, 0x292e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich9_sata_ahci },
++ /* SATA Controller IDE (ICH10) */
++ { 0x8086, 0x3a00, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
++ /* SATA Controller IDE (ICH10) */
++ { 0x8086, 0x3a06, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_2port_sata },
++ /* SATA Controller IDE (ICH10) */
++ { 0x8086, 0x3a20, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_sata_ahci },
++ /* SATA Controller IDE (ICH10) */
++ { 0x8086, 0x3a26, PCI_ANY_ID, PCI_ANY_ID, 0, 0, ich8_2port_sata },
+
+ { } /* terminate list */
+ };
+@@ -361,9 +383,34 @@
+ .present_shift = 8,
+ .map = {
+ /* PM PS SM SS MAP */
+- { P0, NA, P1, NA }, /* 00b (hardwired) */
++ { P0, P2, P1, P3 }, /* 00b (hardwired when in AHCI) */
+ { RV, RV, RV, RV },
+- { RV, RV, RV, RV }, /* 10b (never) */
++ { IDE, IDE, NA, NA }, /* 10b (IDE mode) */
++ { RV, RV, RV, RV },
++ },
++};
++
++static const struct piix_map_db ich9_map_db = {
++ .mask = 0x3,
++ .port_enable = 0x3,
++ .present_shift = 8,
++ .map = {
++ /* PM PS SM SS MAP */
++ { P0, P2, P1, P3 }, /* 00b (hardwired when in AHCI) */
++ { RV, RV, RV, RV },
++ { IDE, IDE, NA, NA }, /* 10b (IDE mode) */
++ { RV, RV, RV, RV },
++ },
++};
++
++static const struct piix_map_db ich8_2port_map_db = {
++ .mask = 0x3,
++ .port_enable = 0x3,
++ .map = {
++ /* PM PS SM SS MAP */
++ { P0, NA, P1, NA }, /* 00b */
++ { RV, RV, RV, RV }, /* 01b */
++ { RV, RV, RV, RV }, /* 10b */
+ { RV, RV, RV, RV },
+ },
+ };
+@@ -376,6 +423,8 @@
+ [ich6m_sata_ahci] = &ich6m_map_db,
+ [ich7m_sata_ahci] = &ich7m_map_db,
+ [ich8_sata_ahci] = &ich8_map_db,
++ [ich9_sata_ahci] = &ich9_map_db,
++ [ich8_2port_sata] = &ich8_2port_map_db,
+ };
+
+ static struct ata_port_info piix_port_info[] = {
+@@ -487,6 +536,30 @@
+ .udma_mask = 0x7f, /* udma0-6 */
+ .port_ops = &piix_sata_ops,
+ },
++
++ /* ich9_sata_ahci */
++ {
++ .sht = &piix_sht,
++ .host_flags = ATA_FLAG_SATA |
++ PIIX_FLAG_CHECKINTR | PIIX_FLAG_SCR |
++ PIIX_FLAG_AHCI,
++ .pio_mask = 0x1f, /* pio0-4 */
++ .mwdma_mask = 0x07, /* mwdma0-2 */
++ .udma_mask = 0x7f, /* udma0-6 */
++ .port_ops = &piix_sata_ops,
++ },
++
++ /* ich8_2port_sata: 11: */
++ {
++ .sht = &piix_sht,
++ .host_flags = ATA_FLAG_SATA |
++ PIIX_FLAG_CHECKINTR | PIIX_FLAG_SCR |
++ PIIX_FLAG_AHCI,
++ .pio_mask = 0x1f, /* pio0-4 */
++ .mwdma_mask = 0x07, /* mwdma0-2 */
++ .udma_mask = ATA_UDMA6,
++ .port_ops = &piix_pata_ops,
++ },
+ };
+
+ static struct pci_bits piix_enable_bits[] = {
+--- linux-2.6.18.8/drivers/serial/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/serial/Kconfig 2008-05-19 00:33:39.506423318 +0300
+@@ -11,6 +11,7 @@
+ config SERIAL_8250
+ tristate "8250/16550 and compatible serial support"
+ depends on (BROKEN || !SPARC)
++ depends on !XEN_DISABLE_SERIAL
+ select SERIAL_CORE
+ ---help---
+ This selects whether you want to include the driver for the standard
+--- linux-2.6.18.8/drivers/video/console/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/video/console/Kconfig 2008-05-19 00:33:44.130689864 +0300
+@@ -53,6 +53,7 @@
+ config VIDEO_SELECT
+ bool "Video mode selection support"
+ depends on X86 && VGA_CONSOLE
++ depends on !XEN
+ ---help---
+ This enables support for text mode selection on kernel startup. If
+ you want to take advantage of some high-resolution text mode your
+--- linux-2.6.18.8/drivers/xen/Kconfig 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/Kconfig 2008-05-19 00:33:45.658777945 +0300
+@@ -0,0 +1,298 @@
++#
++# This Kconfig describe xen options
++#
++
++mainmenu "Xen Configuration"
++
++config XEN
++ bool
++
++if XEN
++config XEN_INTERFACE_VERSION
++ hex
++ default 0x00030207
++
++menu "XEN"
++
++config XEN_PRIVILEGED_GUEST
++ bool "Privileged Guest (domain 0)"
++ help
++ Support for privileged operation (domain 0)
++
++config XEN_UNPRIVILEGED_GUEST
++ def_bool !XEN_PRIVILEGED_GUEST
++
++config XEN_PRIVCMD
++ def_bool y
++ depends on PROC_FS
++
++config XEN_XENBUS_DEV
++ def_bool y
++ depends on PROC_FS
++
++config XEN_NETDEV_ACCEL_SFC_UTIL
++ depends on X86
++ tristate
++
++config XEN_BACKEND
++ tristate "Backend driver support"
++ default XEN_PRIVILEGED_GUEST
++ help
++ Support for backend device drivers that provide I/O services
++ to other virtual machines.
++
++config XEN_BLKDEV_BACKEND
++ tristate "Block-device backend driver"
++ depends on XEN_BACKEND
++ default XEN_BACKEND
++ help
++ The block-device backend driver allows the kernel to export its
++ block devices to other guests via a high-performance shared-memory
++ interface.
++
++config XEN_BLKDEV_TAP
++ tristate "Block-device tap backend driver"
++ depends on XEN_BACKEND
++ default XEN_BACKEND
++ help
++ The block tap driver is an alternative to the block back driver
++ and allows VM block requests to be redirected to userspace through
++ a device interface. The tap allows user-space development of
++ high-performance block backends, where disk images may be implemented
++ as files, in memory, or on other hosts across the network. This
++ driver can safely coexist with the existing blockback driver.
++
++config XEN_NETDEV_BACKEND
++ tristate "Network-device backend driver"
++ depends on XEN_BACKEND && NET
++ default XEN_BACKEND
++ help
++ The network-device backend driver allows the kernel to export its
++ network devices to other guests via a high-performance shared-memory
++ interface.
++
++config XEN_NETDEV_PIPELINED_TRANSMITTER
++ bool "Pipelined transmitter (DANGEROUS)"
++ depends on XEN_NETDEV_BACKEND
++ help
++ If the net backend is a dumb domain, such as a transparent Ethernet
++ bridge with no local IP interface, it is safe to say Y here to get
++ slightly lower network overhead.
++ If the backend has a local IP interface; or may be doing smart things
++ like reassembling packets to perform firewall filtering; or if you
++ are unsure; or if you experience network hangs when this option is
++ enabled; then you must say N here.
++
++config XEN_NETDEV_ACCEL_SFC_BACKEND
++ tristate "Network-device backend driver acceleration for Solarflare NICs"
++ depends on XEN_NETDEV_BACKEND && SFC && SFC_RESOURCE && X86
++ select XEN_NETDEV_ACCEL_SFC_UTIL
++ default m
++
++config XEN_NETDEV_LOOPBACK
++ tristate "Network-device loopback driver"
++ depends on XEN_NETDEV_BACKEND
++ help
++ A two-interface loopback device to emulate a local netfront-netback
++ connection. If unsure, it is probably safe to say N here.
++
++config XEN_PCIDEV_BACKEND
++ tristate "PCI-device backend driver"
++ depends on PCI && XEN_BACKEND
++ default XEN_BACKEND
++ help
++ The PCI device backend driver allows the kernel to export arbitrary
++ PCI devices to other guests. If you select this to be a module, you
++ will need to make sure no other driver has bound to the device(s)
++ you want to make visible to other guests.
++
++choice
++ prompt "PCI Backend Mode"
++ depends on XEN_PCIDEV_BACKEND
++ default XEN_PCIDEV_BACKEND_VPCI if !IA64
++ default XEN_PCIDEV_BACKEND_CONTROLLER if IA64
++
++config XEN_PCIDEV_BACKEND_VPCI
++ bool "Virtual PCI"
++ ---help---
++ This PCI Backend hides the true PCI topology and makes the frontend
++ think there is a single PCI bus with only the exported devices on it.
++ For example, a device at 03:05.0 will be re-assigned to 00:00.0. A
++ second device at 02:1a.1 will be re-assigned to 00:01.1.
++
++config XEN_PCIDEV_BACKEND_PASS
++ bool "Passthrough"
++ ---help---
++ This PCI Backend provides a real view of the PCI topology to the
++ frontend (for example, a device at 06:01.b will still appear at
++ 06:01.b to the frontend). This is similar to how Xen 2.0.x exposed
++ PCI devices to its driver domains. This may be required for drivers
++ which depend on finding their hardward in certain bus/slot
++ locations.
++
++config XEN_PCIDEV_BACKEND_SLOT
++ bool "Slot"
++ ---help---
++ This PCI Backend hides the true PCI topology and makes the frontend
++ think there is a single PCI bus with only the exported devices on it.
++ Contrary to the virtual PCI backend, a function becomes a new slot.
++ For example, a device at 03:05.2 will be re-assigned to 00:00.0. A
++ second device at 02:1a.1 will be re-assigned to 00:01.0.
++
++config XEN_PCIDEV_BACKEND_CONTROLLER
++ bool "Controller"
++ depends on IA64
++ ---help---
++ This PCI backend virtualizes the PCI bus topology by providing a
++ virtual bus per PCI root device. Devices which are physically under
++ the same root bus will appear on the same virtual bus. For systems
++ with complex I/O addressing, this is the only backend which supports
++ extended I/O port spaces and MMIO translation offsets. This backend
++ also supports slot virtualization. For example, a device at
++ 0000:01:02.1 will be re-assigned to 0000:00:00.0. A second device
++ at 0000:02:05.0 (behind a P2P bridge on bus 0000:01) will be
++ re-assigned to 0000:00:01.0. A third device at 0000:16:05.0 (under
++ a different PCI root bus) will be re-assigned to 0000:01:00.0.
++
++endchoice
++
++config XEN_PCIDEV_BE_DEBUG
++ bool "PCI Backend Debugging"
++ depends on XEN_PCIDEV_BACKEND
++
++config XEN_TPMDEV_BACKEND
++ tristate "TPM-device backend driver"
++ depends on XEN_BACKEND
++ help
++ The TPM-device backend driver
++
++config XEN_BLKDEV_FRONTEND
++ tristate "Block-device frontend driver"
++ default y
++ help
++ The block-device frontend driver allows the kernel to access block
++ devices mounted within another guest OS. Unless you are building a
++ dedicated device-driver domain, or your master control domain
++ (domain 0), then you almost certainly want to say Y here.
++
++config XEN_NETDEV_FRONTEND
++ tristate "Network-device frontend driver"
++ depends on NET
++ default y
++ help
++ The network-device frontend driver allows the kernel to access
++ network interfaces within another guest OS. Unless you are building a
++ dedicated device-driver domain, or your master control domain
++ (domain 0), then you almost certainly want to say Y here.
++
++config XEN_NETDEV_ACCEL_SFC_FRONTEND
++ tristate "Network-device frontend driver acceleration for Solarflare NICs"
++ depends on XEN_NETDEV_FRONTEND && X86
++ select XEN_NETDEV_ACCEL_SFC_UTIL
++ default m
++
++config XEN_GRANT_DEV
++ tristate "User-space granted page access driver"
++ default XEN_PRIVILEGED_GUEST
++ help
++ Device for accessing (in user-space) pages that have been granted
++ by other domains.
++
++config XEN_FRAMEBUFFER
++ tristate "Framebuffer-device frontend driver"
++ depends on FB
++ select FB_CFB_FILLRECT
++ select FB_CFB_COPYAREA
++ select FB_CFB_IMAGEBLIT
++ default y
++ help
++ The framebuffer-device frontend drivers allows the kernel to create a
++ virtual framebuffer. This framebuffer can be viewed in another
++ domain. Unless this domain has access to a real video card, you
++ probably want to say Y here.
++
++config XEN_KEYBOARD
++ tristate "Keyboard-device frontend driver"
++ depends on XEN_FRAMEBUFFER && INPUT
++ default y
++ help
++ The keyboard-device frontend driver allows the kernel to create a
++ virtual keyboard. This keyboard can then be driven by another
++ domain. If you've said Y to CONFIG_XEN_FRAMEBUFFER, you probably
++ want to say Y here.
++
++config XEN_SCRUB_PAGES
++ bool "Scrub memory before freeing it to Xen"
++ default y
++ help
++ Erase memory contents before freeing it back to Xen's global
++ pool. This ensures that any secrets contained within that
++ memory (e.g., private keys) cannot be found by other guests that
++ may be running on the machine. Most people will want to say Y here.
++ If security is not a concern then you may increase performance by
++ saying N.
++
++config XEN_DISABLE_SERIAL
++ bool "Disable serial port drivers"
++ default y
++ help
++ Disable serial port drivers, allowing the Xen console driver
++ to provide a serial console at ttyS0.
++
++config XEN_SYSFS
++ tristate "Export Xen attributes in sysfs"
++ depends on SYSFS
++ select SYS_HYPERVISOR
++ default y
++ help
++ Xen hypervisor attributes will show up under /sys/hypervisor/.
++
++choice
++ prompt "Xen version compatibility"
++ default XEN_COMPAT_030002_AND_LATER
++
++ config XEN_COMPAT_030002_AND_LATER
++ bool "3.0.2 and later"
++
++ config XEN_COMPAT_030004_AND_LATER
++ bool "3.0.4 and later"
++
++ config XEN_COMPAT_030100_AND_LATER
++ bool "3.1.0 and later"
++
++ config XEN_COMPAT_LATEST_ONLY
++ bool "no compatibility code"
++
++endchoice
++
++config XEN_COMPAT
++ hex
++ default 0xffffff if XEN_COMPAT_LATEST_ONLY
++ default 0x030100 if XEN_COMPAT_030100_AND_LATER
++ default 0x030004 if XEN_COMPAT_030004_AND_LATER
++ default 0x030002 if XEN_COMPAT_030002_AND_LATER
++ default 0
++
++endmenu
++
++config HAVE_IRQ_IGNORE_UNHANDLED
++ def_bool y
++
++config NO_IDLE_HZ
++ def_bool y
++
++config XEN_SMPBOOT
++ def_bool y
++ depends on SMP && !PPC_XEN
++
++config XEN_BALLOON
++ def_bool y
++ depends on !PPC_XEN
++
++config XEN_XENCOMM
++ bool
++
++config XEN_DEVMEM
++ def_bool y
++
++endif
+--- linux-2.6.18.8/drivers/xen/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/Makefile 2008-05-19 00:33:45.662778175 +0300
+@@ -0,0 +1,23 @@
++obj-y += core/
++obj-y += console/
++obj-y += evtchn/
++obj-y += xenbus/
++obj-y += char/
++
++obj-y += util.o
++obj-$(CONFIG_XEN_BALLOON) += balloon/
++obj-$(CONFIG_XEN_BLKDEV_BACKEND) += blkback/
++obj-$(CONFIG_XEN_BLKDEV_TAP) += blktap/
++obj-$(CONFIG_XEN_NETDEV_BACKEND) += netback/
++obj-$(CONFIG_XEN_TPMDEV_BACKEND) += tpmback/
++obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += blkfront/
++obj-$(CONFIG_XEN_NETDEV_FRONTEND) += netfront/
++obj-$(CONFIG_XEN_PCIDEV_BACKEND) += pciback/
++obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += pcifront/
++obj-$(CONFIG_XEN_FRAMEBUFFER) += fbfront/
++obj-$(CONFIG_XEN_KEYBOARD) += fbfront/
++obj-$(CONFIG_XEN_PRIVCMD) += privcmd/
++obj-$(CONFIG_XEN_GRANT_DEV) += gntdev/
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL) += sfc_netutil/
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND) += sfc_netfront/
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND) += sfc_netback/
+--- linux-2.6.18.8/drivers/xen/balloon/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/balloon/Makefile 2008-05-19 00:33:45.674778867 +0300
+@@ -0,0 +1,2 @@
++
++obj-y := balloon.o sysfs.o
+--- linux-2.6.18.8/drivers/xen/balloon/balloon.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/balloon/balloon.c 2008-05-19 00:33:45.674778867 +0300
+@@ -0,0 +1,724 @@
++/******************************************************************************
++ * balloon.c
++ *
++ * Xen balloon driver - enables returning/claiming memory to/from Xen.
++ *
++ * Copyright (c) 2003, B Dragovic
++ * Copyright (c) 2003-2004, M Williamson, K Fraser
++ * Copyright (c) 2005 Dan M. Smith, IBM Corporation
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/kernel.h>
++#include <linux/module.h>
++#include <linux/sched.h>
++#include <linux/errno.h>
++#include <linux/mm.h>
++#include <linux/mman.h>
++#include <linux/smp_lock.h>
++#include <linux/pagemap.h>
++#include <linux/bootmem.h>
++#include <linux/highmem.h>
++#include <linux/vmalloc.h>
++#include <linux/mutex.h>
++#include <xen/xen_proc.h>
++#include <asm/hypervisor.h>
++#include <xen/balloon.h>
++#include <xen/interface/memory.h>
++#include <asm/maddr.h>
++#include <asm/page.h>
++#include <asm/pgalloc.h>
++#include <asm/pgtable.h>
++#include <asm/uaccess.h>
++#include <asm/tlb.h>
++#include <linux/highmem.h>
++#include <linux/list.h>
++#include <xen/xenbus.h>
++#include "common.h"
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++#ifdef CONFIG_PROC_FS
++static struct proc_dir_entry *balloon_pde;
++#endif
++
++static DEFINE_MUTEX(balloon_mutex);
++
++/*
++ * Protects atomic reservation decrease/increase against concurrent increases.
++ * Also protects non-atomic updates of current_pages and driver_pages, and
++ * balloon lists.
++ */
++DEFINE_SPINLOCK(balloon_lock);
++
++struct balloon_stats balloon_stats;
++
++/* We increase/decrease in batches which fit in a page */
++static unsigned long frame_list[PAGE_SIZE / sizeof(unsigned long)];
++
++/* VM /proc information for memory */
++extern unsigned long totalram_pages;
++
++#ifndef MODULE
++extern unsigned long totalhigh_pages;
++#define inc_totalhigh_pages() (totalhigh_pages++)
++#define dec_totalhigh_pages() (totalhigh_pages--)
++#else
++#define inc_totalhigh_pages() ((void)0)
++#define dec_totalhigh_pages() ((void)0)
++#endif
++
++/* List of ballooned pages, threaded through the mem_map array. */
++static LIST_HEAD(ballooned_pages);
++
++/* Main work function, always executed in process context. */
++static void balloon_process(void *unused);
++static DECLARE_WORK(balloon_worker, balloon_process, NULL);
++static struct timer_list balloon_timer;
++
++/* When ballooning out (allocating memory to return to Xen) we don't really
++ want the kernel to try too hard since that can trigger the oom killer. */
++#define GFP_BALLOON \
++ (GFP_HIGHUSER|__GFP_NOWARN|__GFP_NORETRY|__GFP_NOMEMALLOC|__GFP_COLD)
++
++#define PAGE_TO_LIST(p) (&(p)->lru)
++#define LIST_TO_PAGE(l) list_entry((l), struct page, lru)
++#define UNLIST_PAGE(p) \
++ do { \
++ list_del(PAGE_TO_LIST(p)); \
++ PAGE_TO_LIST(p)->next = NULL; \
++ PAGE_TO_LIST(p)->prev = NULL; \
++ } while(0)
++
++#define IPRINTK(fmt, args...) \
++ printk(KERN_INFO "xen_mem: " fmt, ##args)
++#define WPRINTK(fmt, args...) \
++ printk(KERN_WARNING "xen_mem: " fmt, ##args)
++
++/* balloon_append: add the given page to the balloon. */
++static void balloon_append(struct page *page)
++{
++ /* Lowmem is re-populated first, so highmem pages go at list tail. */
++ if (PageHighMem(page)) {
++ list_add_tail(PAGE_TO_LIST(page), &ballooned_pages);
++ bs.balloon_high++;
++ dec_totalhigh_pages();
++ } else {
++ list_add(PAGE_TO_LIST(page), &ballooned_pages);
++ bs.balloon_low++;
++ }
++}
++
++/* balloon_retrieve: rescue a page from the balloon, if it is not empty. */
++static struct page *balloon_retrieve(void)
++{
++ struct page *page;
++
++ if (list_empty(&ballooned_pages))
++ return NULL;
++
++ page = LIST_TO_PAGE(ballooned_pages.next);
++ UNLIST_PAGE(page);
++
++ if (PageHighMem(page)) {
++ bs.balloon_high--;
++ inc_totalhigh_pages();
++ }
++ else
++ bs.balloon_low--;
++
++ return page;
++}
++
++static struct page *balloon_first_page(void)
++{
++ if (list_empty(&ballooned_pages))
++ return NULL;
++ return LIST_TO_PAGE(ballooned_pages.next);
++}
++
++static struct page *balloon_next_page(struct page *page)
++{
++ struct list_head *next = PAGE_TO_LIST(page)->next;
++ if (next == &ballooned_pages)
++ return NULL;
++ return LIST_TO_PAGE(next);
++}
++
++static inline void balloon_free_page(struct page *page)
++{
++#ifndef MODULE
++ if (put_page_testzero(page))
++ free_cold_page(page);
++#else
++ /* free_cold_page() is not being exported. */
++ __free_page(page);
++#endif
++}
++
++static void balloon_alarm(unsigned long unused)
++{
++ schedule_work(&balloon_worker);
++}
++
++static unsigned long current_target(void)
++{
++ unsigned long target = min(bs.target_pages, bs.hard_limit);
++ if (target > (bs.current_pages + bs.balloon_low + bs.balloon_high))
++ target = bs.current_pages + bs.balloon_low + bs.balloon_high;
++ return target;
++}
++
++static unsigned long minimum_target(void)
++{
++#ifndef CONFIG_XEN
++#define max_pfn num_physpages
++#endif
++ unsigned long min_pages, curr_pages = current_target();
++
++#define MB2PAGES(mb) ((mb) << (20 - PAGE_SHIFT))
++ /* Simple continuous piecewiese linear function:
++ * max MiB -> min MiB gradient
++ * 0 0
++ * 16 16
++ * 32 24
++ * 128 72 (1/2)
++ * 512 168 (1/4)
++ * 2048 360 (1/8)
++ * 8192 552 (1/32)
++ * 32768 1320
++ * 131072 4392
++ */
++ if (max_pfn < MB2PAGES(128))
++ min_pages = MB2PAGES(8) + (max_pfn >> 1);
++ else if (max_pfn < MB2PAGES(512))
++ min_pages = MB2PAGES(40) + (max_pfn >> 2);
++ else if (max_pfn < MB2PAGES(2048))
++ min_pages = MB2PAGES(104) + (max_pfn >> 3);
++ else
++ min_pages = MB2PAGES(296) + (max_pfn >> 5);
++#undef MB2PAGES
++
++ /* Don't enforce growth */
++ return min(min_pages, curr_pages);
++#ifndef CONFIG_XEN
++#undef max_pfn
++#endif
++}
++
++static int increase_reservation(unsigned long nr_pages)
++{
++ unsigned long pfn, i, flags;
++ struct page *page;
++ long rc;
++ struct xen_memory_reservation reservation = {
++ .address_bits = 0,
++ .extent_order = 0,
++ .domid = DOMID_SELF
++ };
++
++ if (nr_pages > ARRAY_SIZE(frame_list))
++ nr_pages = ARRAY_SIZE(frame_list);
++
++ balloon_lock(flags);
++
++ page = balloon_first_page();
++ for (i = 0; i < nr_pages; i++) {
++ BUG_ON(page == NULL);
++ frame_list[i] = page_to_pfn(page);;
++ page = balloon_next_page(page);
++ }
++
++ set_xen_guest_handle(reservation.extent_start, frame_list);
++ reservation.nr_extents = nr_pages;
++ rc = HYPERVISOR_memory_op(
++ XENMEM_populate_physmap, &reservation);
++ if (rc < nr_pages) {
++ if (rc > 0) {
++ int ret;
++
++ /* We hit the Xen hard limit: reprobe. */
++ reservation.nr_extents = rc;
++ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
++ &reservation);
++ BUG_ON(ret != rc);
++ }
++ if (rc >= 0)
++ bs.hard_limit = (bs.current_pages + rc -
++ bs.driver_pages);
++ goto out;
++ }
++
++ for (i = 0; i < nr_pages; i++) {
++ page = balloon_retrieve();
++ BUG_ON(page == NULL);
++
++ pfn = page_to_pfn(page);
++ BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap) &&
++ phys_to_machine_mapping_valid(pfn));
++
++ set_phys_to_machine(pfn, frame_list[i]);
++
++#ifdef CONFIG_XEN
++ /* Link back into the page tables if not highmem. */
++ if (pfn < max_low_pfn) {
++ int ret;
++ ret = HYPERVISOR_update_va_mapping(
++ (unsigned long)__va(pfn << PAGE_SHIFT),
++ pfn_pte_ma(frame_list[i], PAGE_KERNEL),
++ 0);
++ BUG_ON(ret);
++ }
++#endif
++
++ /* Relinquish the page back to the allocator. */
++ ClearPageReserved(page);
++ init_page_count(page);
++ balloon_free_page(page);
++ }
++
++ bs.current_pages += nr_pages;
++ totalram_pages = bs.current_pages;
++
++ out:
++ balloon_unlock(flags);
++
++ return 0;
++}
++
++static int decrease_reservation(unsigned long nr_pages)
++{
++ unsigned long pfn, i, flags;
++ struct page *page;
++ void *v;
++ int need_sleep = 0;
++ int ret;
++ struct xen_memory_reservation reservation = {
++ .address_bits = 0,
++ .extent_order = 0,
++ .domid = DOMID_SELF
++ };
++
++ if (nr_pages > ARRAY_SIZE(frame_list))
++ nr_pages = ARRAY_SIZE(frame_list);
++
++ for (i = 0; i < nr_pages; i++) {
++ if ((page = alloc_page(GFP_BALLOON)) == NULL) {
++ nr_pages = i;
++ need_sleep = 1;
++ break;
++ }
++
++ pfn = page_to_pfn(page);
++ frame_list[i] = pfn_to_mfn(pfn);
++
++ if (!PageHighMem(page)) {
++ v = phys_to_virt(pfn << PAGE_SHIFT);
++ scrub_pages(v, 1);
++#ifdef CONFIG_XEN
++ ret = HYPERVISOR_update_va_mapping(
++ (unsigned long)v, __pte_ma(0), 0);
++ BUG_ON(ret);
++#endif
++ }
++#ifdef CONFIG_XEN_SCRUB_PAGES
++ else {
++ v = kmap(page);
++ scrub_pages(v, 1);
++ kunmap(page);
++ }
++#endif
++ }
++
++#ifdef CONFIG_XEN
++ /* Ensure that ballooned highmem pages don't have kmaps. */
++ kmap_flush_unused();
++ flush_tlb_all();
++#endif
++
++ balloon_lock(flags);
++
++ /* No more mappings: invalidate P2M and add to balloon. */
++ for (i = 0; i < nr_pages; i++) {
++ pfn = mfn_to_pfn(frame_list[i]);
++ set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
++ balloon_append(pfn_to_page(pfn));
++ }
++
++ set_xen_guest_handle(reservation.extent_start, frame_list);
++ reservation.nr_extents = nr_pages;
++ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
++ BUG_ON(ret != nr_pages);
++
++ bs.current_pages -= nr_pages;
++ totalram_pages = bs.current_pages;
++
++ balloon_unlock(flags);
++
++ return need_sleep;
++}
++
++/*
++ * We avoid multiple worker processes conflicting via the balloon mutex.
++ * We may of course race updates of the target counts (which are protected
++ * by the balloon lock), or with changes to the Xen hard limit, but we will
++ * recover from these in time.
++ */
++static void balloon_process(void *unused)
++{
++ int need_sleep = 0;
++ long credit;
++
++ mutex_lock(&balloon_mutex);
++
++ do {
++ credit = current_target() - bs.current_pages;
++ if (credit > 0)
++ need_sleep = (increase_reservation(credit) != 0);
++ if (credit < 0)
++ need_sleep = (decrease_reservation(-credit) != 0);
++
++#ifndef CONFIG_PREEMPT
++ if (need_resched())
++ schedule();
++#endif
++ } while ((credit != 0) && !need_sleep);
++
++ /* Schedule more work if there is some still to be done. */
++ if (current_target() != bs.current_pages)
++ mod_timer(&balloon_timer, jiffies + HZ);
++
++ mutex_unlock(&balloon_mutex);
++}
++
++/* Resets the Xen limit, sets new target, and kicks off processing. */
++void balloon_set_new_target(unsigned long target)
++{
++ /* No need for lock. Not read-modify-write updates. */
++ bs.hard_limit = ~0UL;
++ bs.target_pages = max(target, minimum_target());
++ schedule_work(&balloon_worker);
++}
++
++static struct xenbus_watch target_watch =
++{
++ .node = "memory/target"
++};
++
++/* React to a change in the target key */
++static void watch_target(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ unsigned long long new_target;
++ int err;
++
++ err = xenbus_scanf(XBT_NIL, "memory", "target", "%llu", &new_target);
++ if (err != 1) {
++ /* This is ok (for domain0 at least) - so just return */
++ return;
++ }
++
++ /* The given memory/target value is in KiB, so it needs converting to
++ * pages. PAGE_SHIFT converts bytes to pages, hence PAGE_SHIFT - 10.
++ */
++ balloon_set_new_target(new_target >> (PAGE_SHIFT - 10));
++}
++
++static int balloon_init_watcher(struct notifier_block *notifier,
++ unsigned long event,
++ void *data)
++{
++ int err;
++
++ err = register_xenbus_watch(&target_watch);
++ if (err)
++ printk(KERN_ERR "Failed to set balloon watcher\n");
++
++ return NOTIFY_DONE;
++}
++
++#ifdef CONFIG_PROC_FS
++static int balloon_write(struct file *file, const char __user *buffer,
++ unsigned long count, void *data)
++{
++ char memstring[64], *endchar;
++ unsigned long long target_bytes;
++
++ if (!capable(CAP_SYS_ADMIN))
++ return -EPERM;
++
++ if (count <= 1)
++ return -EBADMSG; /* runt */
++ if (count > sizeof(memstring))
++ return -EFBIG; /* too long */
++
++ if (copy_from_user(memstring, buffer, count))
++ return -EFAULT;
++ memstring[sizeof(memstring)-1] = '\0';
++
++ target_bytes = memparse(memstring, &endchar);
++ balloon_set_new_target(target_bytes >> PAGE_SHIFT);
++
++ return count;
++}
++
++static int balloon_read(char *page, char **start, off_t off,
++ int count, int *eof, void *data)
++{
++ int len;
++
++ len = sprintf(
++ page,
++ "Current allocation: %8lu kB\n"
++ "Requested target: %8lu kB\n"
++ "Low-mem balloon: %8lu kB\n"
++ "High-mem balloon: %8lu kB\n"
++ "Driver pages: %8lu kB\n"
++ "Xen hard limit: ",
++ PAGES2KB(bs.current_pages), PAGES2KB(bs.target_pages),
++ PAGES2KB(bs.balloon_low), PAGES2KB(bs.balloon_high),
++ PAGES2KB(bs.driver_pages));
++
++ if (bs.hard_limit != ~0UL)
++ len += sprintf(page + len, "%8lu kB\n",
++ PAGES2KB(bs.hard_limit));
++ else
++ len += sprintf(page + len, " ??? kB\n");
++
++ *eof = 1;
++ return len;
++}
++#endif
++
++static struct notifier_block xenstore_notifier;
++
++static int __init balloon_init(void)
++{
++#if defined(CONFIG_X86) && defined(CONFIG_XEN)
++ unsigned long pfn;
++ struct page *page;
++#endif
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ IPRINTK("Initialising balloon driver.\n");
++
++#ifdef CONFIG_XEN
++ bs.current_pages = min(xen_start_info->nr_pages, max_pfn);
++ totalram_pages = bs.current_pages;
++#else
++ bs.current_pages = totalram_pages;
++#endif
++ bs.target_pages = bs.current_pages;
++ bs.balloon_low = 0;
++ bs.balloon_high = 0;
++ bs.driver_pages = 0UL;
++ bs.hard_limit = ~0UL;
++
++ init_timer(&balloon_timer);
++ balloon_timer.data = 0;
++ balloon_timer.function = balloon_alarm;
++
++#ifdef CONFIG_PROC_FS
++ if ((balloon_pde = create_xen_proc_entry("balloon", 0644)) == NULL) {
++ WPRINTK("Unable to create /proc/xen/balloon.\n");
++ return -1;
++ }
++
++ balloon_pde->read_proc = balloon_read;
++ balloon_pde->write_proc = balloon_write;
++#endif
++ balloon_sysfs_init();
++
++#if defined(CONFIG_X86) && defined(CONFIG_XEN)
++ /* Initialise the balloon with excess memory space. */
++ for (pfn = xen_start_info->nr_pages; pfn < max_pfn; pfn++) {
++ page = pfn_to_page(pfn);
++ if (!PageReserved(page))
++ balloon_append(page);
++ }
++#endif
++
++ target_watch.callback = watch_target;
++ xenstore_notifier.notifier_call = balloon_init_watcher;
++
++ register_xenstore_notifier(&xenstore_notifier);
++
++ return 0;
++}
++
++subsys_initcall(balloon_init);
++
++static void __exit balloon_exit(void)
++{
++ /* XXX - release balloon here */
++ return;
++}
++
++module_exit(balloon_exit);
++
++void balloon_update_driver_allowance(long delta)
++{
++ unsigned long flags;
++
++ balloon_lock(flags);
++ bs.driver_pages += delta;
++ balloon_unlock(flags);
++}
++
++#ifdef CONFIG_XEN
++static int dealloc_pte_fn(
++ pte_t *pte, struct page *pmd_page, unsigned long addr, void *data)
++{
++ unsigned long mfn = pte_mfn(*pte);
++ int ret;
++ struct xen_memory_reservation reservation = {
++ .nr_extents = 1,
++ .extent_order = 0,
++ .domid = DOMID_SELF
++ };
++ set_xen_guest_handle(reservation.extent_start, &mfn);
++ set_pte_at(&init_mm, addr, pte, __pte_ma(0));
++ set_phys_to_machine(__pa(addr) >> PAGE_SHIFT, INVALID_P2M_ENTRY);
++ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
++ BUG_ON(ret != 1);
++ return 0;
++}
++#endif
++
++struct page **alloc_empty_pages_and_pagevec(int nr_pages)
++{
++ unsigned long flags;
++ void *v;
++ struct page *page, **pagevec;
++ int i, ret;
++
++ pagevec = kmalloc(sizeof(page) * nr_pages, GFP_KERNEL);
++ if (pagevec == NULL)
++ return NULL;
++
++ for (i = 0; i < nr_pages; i++) {
++ page = pagevec[i] = alloc_page(GFP_KERNEL|__GFP_COLD);
++ if (page == NULL)
++ goto err;
++
++ v = page_address(page);
++ scrub_pages(v, 1);
++
++ balloon_lock(flags);
++
++ if (xen_feature(XENFEAT_auto_translated_physmap)) {
++ unsigned long gmfn = page_to_pfn(page);
++ struct xen_memory_reservation reservation = {
++ .nr_extents = 1,
++ .extent_order = 0,
++ .domid = DOMID_SELF
++ };
++ set_xen_guest_handle(reservation.extent_start, &gmfn);
++ ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
++ &reservation);
++ if (ret == 1)
++ ret = 0; /* success */
++ } else {
++#ifdef CONFIG_XEN
++ ret = apply_to_page_range(&init_mm, (unsigned long)v,
++ PAGE_SIZE, dealloc_pte_fn,
++ NULL);
++#else
++ /* Cannot handle non-auto translate mode. */
++ ret = 1;
++#endif
++ }
++
++ if (ret != 0) {
++ balloon_unlock(flags);
++ balloon_free_page(page);
++ goto err;
++ }
++
++ totalram_pages = --bs.current_pages;
++
++ balloon_unlock(flags);
++ }
++
++ out:
++ schedule_work(&balloon_worker);
++#ifdef CONFIG_XEN
++ flush_tlb_all();
++#endif
++ return pagevec;
++
++ err:
++ balloon_lock(flags);
++ while (--i >= 0)
++ balloon_append(pagevec[i]);
++ balloon_unlock(flags);
++ kfree(pagevec);
++ pagevec = NULL;
++ goto out;
++}
++
++void free_empty_pages_and_pagevec(struct page **pagevec, int nr_pages)
++{
++ unsigned long flags;
++ int i;
++
++ if (pagevec == NULL)
++ return;
++
++ balloon_lock(flags);
++ for (i = 0; i < nr_pages; i++) {
++ BUG_ON(page_count(pagevec[i]) != 1);
++ balloon_append(pagevec[i]);
++ }
++ balloon_unlock(flags);
++
++ kfree(pagevec);
++
++ schedule_work(&balloon_worker);
++}
++
++void balloon_release_driver_page(struct page *page)
++{
++ unsigned long flags;
++
++ balloon_lock(flags);
++ balloon_append(page);
++ bs.driver_pages--;
++ balloon_unlock(flags);
++
++ schedule_work(&balloon_worker);
++}
++
++EXPORT_SYMBOL_GPL(balloon_update_driver_allowance);
++EXPORT_SYMBOL_GPL(alloc_empty_pages_and_pagevec);
++EXPORT_SYMBOL_GPL(free_empty_pages_and_pagevec);
++EXPORT_SYMBOL_GPL(balloon_release_driver_page);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/balloon/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/balloon/common.h 2008-05-19 00:33:45.674778867 +0300
+@@ -0,0 +1,58 @@
++/******************************************************************************
++ * balloon/common.h
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef __XEN_BALLOON_COMMON_H__
++#define __XEN_BALLOON_COMMON_H__
++
++#define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
++
++struct balloon_stats {
++ /* We aim for 'current allocation' == 'target allocation'. */
++ unsigned long current_pages;
++ unsigned long target_pages;
++ /* We may hit the hard limit in Xen. If we do then we remember it. */
++ unsigned long hard_limit;
++ /*
++ * Drivers may alter the memory reservation independently, but they
++ * must inform the balloon driver so we avoid hitting the hard limit.
++ */
++ unsigned long driver_pages;
++ /* Number of pages in high- and low-memory balloons. */
++ unsigned long balloon_low;
++ unsigned long balloon_high;
++};
++
++extern struct balloon_stats balloon_stats;
++#define bs balloon_stats
++
++int balloon_sysfs_init(void);
++void balloon_sysfs_exit(void);
++
++void balloon_set_new_target(unsigned long target);
++
++#endif /* __XEN_BALLOON_COMMON_H__ */
+--- linux-2.6.18.8/drivers/xen/balloon/sysfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/balloon/sysfs.c 2008-05-19 00:33:45.678779098 +0300
+@@ -0,0 +1,170 @@
++/******************************************************************************
++ * balloon/sysfs.c
++ *
++ * Xen balloon driver - sysfs interfaces.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/capability.h>
++#include <linux/errno.h>
++#include <linux/stat.h>
++#include <linux/string.h>
++#include <linux/sysdev.h>
++#include "common.h"
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++#define BALLOON_CLASS_NAME "xen_memory"
++
++#define BALLOON_SHOW(name, format, args...) \
++ static ssize_t show_##name(struct sys_device *dev, \
++ char *buf) \
++ { \
++ return sprintf(buf, format, ##args); \
++ } \
++ static SYSDEV_ATTR(name, S_IRUGO, show_##name, NULL)
++
++BALLOON_SHOW(current_kb, "%lu\n", PAGES2KB(bs.current_pages));
++BALLOON_SHOW(low_kb, "%lu\n", PAGES2KB(bs.balloon_low));
++BALLOON_SHOW(high_kb, "%lu\n", PAGES2KB(bs.balloon_high));
++BALLOON_SHOW(hard_limit_kb,
++ (bs.hard_limit!=~0UL) ? "%lu\n" : "???\n",
++ (bs.hard_limit!=~0UL) ? PAGES2KB(bs.hard_limit) : 0);
++BALLOON_SHOW(driver_kb, "%lu\n", PAGES2KB(bs.driver_pages));
++
++static ssize_t show_target_kb(struct sys_device *dev, char *buf)
++{
++ return sprintf(buf, "%lu\n", PAGES2KB(bs.target_pages));
++}
++
++static ssize_t store_target_kb(struct sys_device *dev,
++ const char *buf,
++ size_t count)
++{
++ char memstring[64], *endchar;
++ unsigned long long target_bytes;
++
++ if (!capable(CAP_SYS_ADMIN))
++ return -EPERM;
++
++ if (count <= 1)
++ return -EBADMSG; /* runt */
++ if (count > sizeof(memstring))
++ return -EFBIG; /* too long */
++ strcpy(memstring, buf);
++
++ target_bytes = memparse(memstring, &endchar);
++ balloon_set_new_target(target_bytes >> PAGE_SHIFT);
++
++ return count;
++}
++
++static SYSDEV_ATTR(target_kb, S_IRUGO | S_IWUSR,
++ show_target_kb, store_target_kb);
++
++static struct sysdev_attribute *balloon_attrs[] = {
++ &attr_target_kb,
++};
++
++static struct attribute *balloon_info_attrs[] = {
++ &attr_current_kb.attr,
++ &attr_low_kb.attr,
++ &attr_high_kb.attr,
++ &attr_hard_limit_kb.attr,
++ &attr_driver_kb.attr,
++ NULL
++};
++
++static struct attribute_group balloon_info_group = {
++ .name = "info",
++ .attrs = balloon_info_attrs,
++};
++
++static struct sysdev_class balloon_sysdev_class = {
++ set_kset_name(BALLOON_CLASS_NAME),
++};
++
++static struct sys_device balloon_sysdev;
++
++static int register_balloon(struct sys_device *sysdev)
++{
++ int i, error;
++
++ error = sysdev_class_register(&balloon_sysdev_class);
++ if (error)
++ return error;
++
++ sysdev->id = 0;
++ sysdev->cls = &balloon_sysdev_class;
++
++ error = sysdev_register(sysdev);
++ if (error) {
++ sysdev_class_unregister(&balloon_sysdev_class);
++ return error;
++ }
++
++ for (i = 0; i < ARRAY_SIZE(balloon_attrs); i++) {
++ error = sysdev_create_file(sysdev, balloon_attrs[i]);
++ if (error)
++ goto fail;
++ }
++
++ error = sysfs_create_group(&sysdev->kobj, &balloon_info_group);
++ if (error)
++ goto fail;
++
++ return 0;
++
++ fail:
++ while (--i >= 0)
++ sysdev_remove_file(sysdev, balloon_attrs[i]);
++ sysdev_unregister(sysdev);
++ sysdev_class_unregister(&balloon_sysdev_class);
++ return error;
++}
++
++static void unregister_balloon(struct sys_device *sysdev)
++{
++ int i;
++
++ sysfs_remove_group(&sysdev->kobj, &balloon_info_group);
++ for (i = 0; i < ARRAY_SIZE(balloon_attrs); i++)
++ sysdev_remove_file(sysdev, balloon_attrs[i]);
++ sysdev_unregister(sysdev);
++ sysdev_class_unregister(&balloon_sysdev_class);
++}
++
++int balloon_sysfs_init(void)
++{
++ return register_balloon(&balloon_sysdev);
++}
++
++void balloon_sysfs_exit(void)
++{
++ unregister_balloon(&balloon_sysdev);
++}
+--- linux-2.6.18.8/drivers/xen/blkback/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/Makefile 2008-05-19 00:33:45.678779098 +0300
+@@ -0,0 +1,3 @@
++obj-$(CONFIG_XEN_BLKDEV_BACKEND) := blkbk.o
++
++blkbk-y := blkback.o xenbus.o interface.o vbd.o
+--- linux-2.6.18.8/drivers/xen/blkback/blkback.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/blkback.c 2008-05-19 00:33:45.678779098 +0300
+@@ -0,0 +1,646 @@
++/******************************************************************************
++ * arch/xen/drivers/blkif/backend/main.c
++ *
++ * Back-end of the driver for virtual block devices. This portion of the
++ * driver exports a 'unified' block-device interface that can be accessed
++ * by any operating system that implements a compatible front end. A
++ * reference front-end implementation can be found in:
++ * arch/xen/drivers/blkif/frontend
++ *
++ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
++ * Copyright (c) 2005, Christopher Clark
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/spinlock.h>
++#include <linux/kthread.h>
++#include <linux/list.h>
++#include <linux/delay.h>
++#include <xen/balloon.h>
++#include <asm/hypervisor.h>
++#include "common.h"
++
++/*
++ * These are rather arbitrary. They are fairly large because adjacent requests
++ * pulled from a communication ring are quite likely to end up being part of
++ * the same scatter/gather request at the disc.
++ *
++ * ** TRY INCREASING 'blkif_reqs' IF WRITE SPEEDS SEEM TOO LOW **
++ *
++ * This will increase the chances of being able to write whole tracks.
++ * 64 should be enough to keep us competitive with Linux.
++ */
++static int blkif_reqs = 64;
++module_param_named(reqs, blkif_reqs, int, 0);
++MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate");
++
++/* Run-time switchable: /sys/module/blkback/parameters/ */
++static unsigned int log_stats = 0;
++static unsigned int debug_lvl = 0;
++module_param(log_stats, int, 0644);
++module_param(debug_lvl, int, 0644);
++
++/*
++ * Each outstanding request that we've passed to the lower device layers has a
++ * 'pending_req' allocated to it. Each buffer_head that completes decrements
++ * the pendcnt towards zero. When it hits zero, the specified domain has a
++ * response queued for it, with the saved 'id' passed back.
++ */
++typedef struct {
++ blkif_t *blkif;
++ u64 id;
++ int nr_pages;
++ atomic_t pendcnt;
++ unsigned short operation;
++ int status;
++ struct list_head free_list;
++} pending_req_t;
++
++static pending_req_t *pending_reqs;
++static struct list_head pending_free;
++static DEFINE_SPINLOCK(pending_free_lock);
++static DECLARE_WAIT_QUEUE_HEAD(pending_free_wq);
++
++#define BLKBACK_INVALID_HANDLE (~0)
++
++static struct page **pending_pages;
++static grant_handle_t *pending_grant_handles;
++
++static inline int vaddr_pagenr(pending_req_t *req, int seg)
++{
++ return (req - pending_reqs) * BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
++}
++
++static inline unsigned long vaddr(pending_req_t *req, int seg)
++{
++ unsigned long pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, seg)]);
++ return (unsigned long)pfn_to_kaddr(pfn);
++}
++
++#define pending_handle(_req, _seg) \
++ (pending_grant_handles[vaddr_pagenr(_req, _seg)])
++
++
++static int do_block_io_op(blkif_t *blkif);
++static void dispatch_rw_block_io(blkif_t *blkif,
++ blkif_request_t *req,
++ pending_req_t *pending_req);
++static void make_response(blkif_t *blkif, u64 id,
++ unsigned short op, int st);
++
++/******************************************************************
++ * misc small helpers
++ */
++static pending_req_t* alloc_req(void)
++{
++ pending_req_t *req = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pending_free_lock, flags);
++ if (!list_empty(&pending_free)) {
++ req = list_entry(pending_free.next, pending_req_t, free_list);
++ list_del(&req->free_list);
++ }
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++ return req;
++}
++
++static void free_req(pending_req_t *req)
++{
++ unsigned long flags;
++ int was_empty;
++
++ spin_lock_irqsave(&pending_free_lock, flags);
++ was_empty = list_empty(&pending_free);
++ list_add(&req->free_list, &pending_free);
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++ if (was_empty)
++ wake_up(&pending_free_wq);
++}
++
++static void unplug_queue(blkif_t *blkif)
++{
++ if (blkif->plug == NULL)
++ return;
++ if (blkif->plug->unplug_fn)
++ blkif->plug->unplug_fn(blkif->plug);
++ blk_put_queue(blkif->plug);
++ blkif->plug = NULL;
++}
++
++static void plug_queue(blkif_t *blkif, struct bio *bio)
++{
++ request_queue_t *q = bdev_get_queue(bio->bi_bdev);
++
++ if (q == blkif->plug)
++ return;
++ unplug_queue(blkif);
++ blk_get_queue(q);
++ blkif->plug = q;
++}
++
++static void fast_flush_area(pending_req_t *req)
++{
++ struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST];
++ unsigned int i, invcount = 0;
++ grant_handle_t handle;
++ int ret;
++
++ for (i = 0; i < req->nr_pages; i++) {
++ handle = pending_handle(req, i);
++ if (handle == BLKBACK_INVALID_HANDLE)
++ continue;
++ gnttab_set_unmap_op(&unmap[invcount], vaddr(req, i),
++ GNTMAP_host_map, handle);
++ pending_handle(req, i) = BLKBACK_INVALID_HANDLE;
++ invcount++;
++ }
++
++ ret = HYPERVISOR_grant_table_op(
++ GNTTABOP_unmap_grant_ref, unmap, invcount);
++ BUG_ON(ret);
++}
++
++/******************************************************************
++ * SCHEDULER FUNCTIONS
++ */
++
++static void print_stats(blkif_t *blkif)
++{
++ printk(KERN_DEBUG "%s: oo %3d | rd %4d | wr %4d | br %4d\n",
++ current->comm, blkif->st_oo_req,
++ blkif->st_rd_req, blkif->st_wr_req, blkif->st_br_req);
++ blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
++ blkif->st_rd_req = 0;
++ blkif->st_wr_req = 0;
++ blkif->st_oo_req = 0;
++}
++
++int blkif_schedule(void *arg)
++{
++ blkif_t *blkif = arg;
++
++ blkif_get(blkif);
++
++ if (debug_lvl)
++ printk(KERN_DEBUG "%s: started\n", current->comm);
++
++ while (!kthread_should_stop()) {
++ if (try_to_freeze())
++ continue;
++
++ wait_event_interruptible(
++ blkif->wq,
++ blkif->waiting_reqs || kthread_should_stop());
++ wait_event_interruptible(
++ pending_free_wq,
++ !list_empty(&pending_free) || kthread_should_stop());
++
++ blkif->waiting_reqs = 0;
++ smp_mb(); /* clear flag *before* checking for work */
++
++ if (do_block_io_op(blkif))
++ blkif->waiting_reqs = 1;
++ unplug_queue(blkif);
++
++ if (log_stats && time_after(jiffies, blkif->st_print))
++ print_stats(blkif);
++ }
++
++ if (log_stats)
++ print_stats(blkif);
++ if (debug_lvl)
++ printk(KERN_DEBUG "%s: exiting\n", current->comm);
++
++ blkif->xenblkd = NULL;
++ blkif_put(blkif);
++
++ return 0;
++}
++
++/******************************************************************
++ * COMPLETION CALLBACK -- Called as bh->b_end_io()
++ */
++
++static void __end_block_io_op(pending_req_t *pending_req, int error)
++{
++ /* An error fails the entire request. */
++ if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) &&
++ (error == -EOPNOTSUPP)) {
++ DPRINTK("blkback: write barrier op failed, not supported\n");
++ blkback_barrier(XBT_NIL, pending_req->blkif->be, 0);
++ pending_req->status = BLKIF_RSP_EOPNOTSUPP;
++ } else if (error) {
++ DPRINTK("Buffer not up-to-date at end of operation, "
++ "error=%d\n", error);
++ pending_req->status = BLKIF_RSP_ERROR;
++ }
++
++ if (atomic_dec_and_test(&pending_req->pendcnt)) {
++ fast_flush_area(pending_req);
++ make_response(pending_req->blkif, pending_req->id,
++ pending_req->operation, pending_req->status);
++ blkif_put(pending_req->blkif);
++ free_req(pending_req);
++ }
++}
++
++static int end_block_io_op(struct bio *bio, unsigned int done, int error)
++{
++ if (bio->bi_size != 0)
++ return 1;
++ __end_block_io_op(bio->bi_private, error);
++ bio_put(bio);
++ return error;
++}
++
++
++/******************************************************************************
++ * NOTIFICATION FROM GUEST OS.
++ */
++
++static void blkif_notify_work(blkif_t *blkif)
++{
++ blkif->waiting_reqs = 1;
++ wake_up(&blkif->wq);
++}
++
++irqreturn_t blkif_be_int(int irq, void *dev_id, struct pt_regs *regs)
++{
++ blkif_notify_work(dev_id);
++ return IRQ_HANDLED;
++}
++
++
++
++/******************************************************************
++ * DOWNWARD CALLS -- These interface with the block-device layer proper.
++ */
++
++static int do_block_io_op(blkif_t *blkif)
++{
++ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
++ blkif_request_t req;
++ pending_req_t *pending_req;
++ RING_IDX rc, rp;
++ int more_to_do = 0;
++
++ rc = blk_rings->common.req_cons;
++ rp = blk_rings->common.sring->req_prod;
++ rmb(); /* Ensure we see queued requests up to 'rp'. */
++
++ while (rc != rp) {
++
++ if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc))
++ break;
++
++ pending_req = alloc_req();
++ if (NULL == pending_req) {
++ blkif->st_oo_req++;
++ more_to_do = 1;
++ break;
++ }
++
++ if (kthread_should_stop()) {
++ more_to_do = 1;
++ break;
++ }
++
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req));
++ break;
++ case BLKIF_PROTOCOL_X86_32:
++ blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
++ break;
++ case BLKIF_PROTOCOL_X86_64:
++ blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
++ break;
++ default:
++ BUG();
++ }
++ blk_rings->common.req_cons = ++rc; /* before make_response() */
++
++ /* Apply all sanity checks to /private copy/ of request. */
++ barrier();
++
++ switch (req.operation) {
++ case BLKIF_OP_READ:
++ blkif->st_rd_req++;
++ dispatch_rw_block_io(blkif, &req, pending_req);
++ break;
++ case BLKIF_OP_WRITE_BARRIER:
++ blkif->st_br_req++;
++ /* fall through */
++ case BLKIF_OP_WRITE:
++ blkif->st_wr_req++;
++ dispatch_rw_block_io(blkif, &req, pending_req);
++ break;
++ default:
++ /* A good sign something is wrong: sleep for a while to
++ * avoid excessive CPU consumption by a bad guest. */
++ msleep(1);
++ DPRINTK("error: unknown block io operation [%d]\n",
++ req.operation);
++ make_response(blkif, req.id, req.operation,
++ BLKIF_RSP_ERROR);
++ free_req(pending_req);
++ break;
++ }
++
++ /* Yield point for this unbounded loop. */
++ cond_resched();
++ }
++
++ return more_to_do;
++}
++
++static void dispatch_rw_block_io(blkif_t *blkif,
++ blkif_request_t *req,
++ pending_req_t *pending_req)
++{
++ extern void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]);
++ struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
++ struct phys_req preq;
++ struct {
++ unsigned long buf; unsigned int nsec;
++ } seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
++ unsigned int nseg;
++ struct bio *bio = NULL, *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST];
++ int ret, i, nbio = 0;
++ int operation;
++
++ switch (req->operation) {
++ case BLKIF_OP_READ:
++ operation = READ;
++ break;
++ case BLKIF_OP_WRITE:
++ operation = WRITE;
++ break;
++ case BLKIF_OP_WRITE_BARRIER:
++ operation = WRITE_BARRIER;
++ break;
++ default:
++ operation = 0; /* make gcc happy */
++ BUG();
++ }
++
++ /* Check that number of segments is sane. */
++ nseg = req->nr_segments;
++ if (unlikely(nseg == 0 && operation != WRITE_BARRIER) ||
++ unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
++ DPRINTK("Bad number of segments in request (%d)\n", nseg);
++ goto fail_response;
++ }
++
++ preq.dev = req->handle;
++ preq.sector_number = req->sector_number;
++ preq.nr_sects = 0;
++
++ pending_req->blkif = blkif;
++ pending_req->id = req->id;
++ pending_req->operation = req->operation;
++ pending_req->status = BLKIF_RSP_OKAY;
++ pending_req->nr_pages = nseg;
++
++ for (i = 0; i < nseg; i++) {
++ uint32_t flags;
++
++ seg[i].nsec = req->seg[i].last_sect -
++ req->seg[i].first_sect + 1;
++
++ if ((req->seg[i].last_sect >= (PAGE_SIZE >> 9)) ||
++ (req->seg[i].last_sect < req->seg[i].first_sect))
++ goto fail_response;
++ preq.nr_sects += seg[i].nsec;
++
++ flags = GNTMAP_host_map;
++ if (operation != READ)
++ flags |= GNTMAP_readonly;
++ gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
++ req->seg[i].gref, blkif->domid);
++ }
++
++ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg);
++ BUG_ON(ret);
++
++ for (i = 0; i < nseg; i++) {
++ if (unlikely(map[i].status != 0)) {
++ DPRINTK("invalid buffer -- could not remap it\n");
++ map[i].handle = BLKBACK_INVALID_HANDLE;
++ ret |= 1;
++ }
++
++ pending_handle(pending_req, i) = map[i].handle;
++
++ if (ret)
++ continue;
++
++ set_phys_to_machine(__pa(vaddr(
++ pending_req, i)) >> PAGE_SHIFT,
++ FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
++ seg[i].buf = map[i].dev_bus_addr |
++ (req->seg[i].first_sect << 9);
++ }
++
++ if (ret)
++ goto fail_flush;
++
++ if (vbd_translate(&preq, blkif, operation) != 0) {
++ DPRINTK("access denied: %s of [%llu,%llu] on dev=%04x\n",
++ operation == READ ? "read" : "write",
++ preq.sector_number,
++ preq.sector_number + preq.nr_sects, preq.dev);
++ goto fail_flush;
++ }
++
++ for (i = 0; i < nseg; i++) {
++ if (((int)preq.sector_number|(int)seg[i].nsec) &
++ ((bdev_hardsect_size(preq.bdev) >> 9) - 1)) {
++ DPRINTK("Misaligned I/O request from domain %d",
++ blkif->domid);
++ goto fail_put_bio;
++ }
++
++ while ((bio == NULL) ||
++ (bio_add_page(bio,
++ virt_to_page(vaddr(pending_req, i)),
++ seg[i].nsec << 9,
++ seg[i].buf & ~PAGE_MASK) == 0)) {
++ bio = biolist[nbio++] = bio_alloc(GFP_KERNEL, nseg-i);
++ if (unlikely(bio == NULL))
++ goto fail_put_bio;
++
++ bio->bi_bdev = preq.bdev;
++ bio->bi_private = pending_req;
++ bio->bi_end_io = end_block_io_op;
++ bio->bi_sector = preq.sector_number;
++ }
++
++ preq.sector_number += seg[i].nsec;
++ }
++
++ if (!bio) {
++ BUG_ON(operation != WRITE_BARRIER);
++ bio = biolist[nbio++] = bio_alloc(GFP_KERNEL, 0);
++ if (unlikely(bio == NULL))
++ goto fail_put_bio;
++
++ bio->bi_bdev = preq.bdev;
++ bio->bi_private = pending_req;
++ bio->bi_end_io = end_block_io_op;
++ bio->bi_sector = -1;
++ }
++
++ plug_queue(blkif, bio);
++ atomic_set(&pending_req->pendcnt, nbio);
++ blkif_get(blkif);
++
++ for (i = 0; i < nbio; i++)
++ submit_bio(operation, biolist[i]);
++
++ if (operation == READ)
++ blkif->st_rd_sect += preq.nr_sects;
++ else if (operation == WRITE || operation == WRITE_BARRIER)
++ blkif->st_wr_sect += preq.nr_sects;
++
++ return;
++
++ fail_put_bio:
++ for (i = 0; i < (nbio-1); i++)
++ bio_put(biolist[i]);
++ fail_flush:
++ fast_flush_area(pending_req);
++ fail_response:
++ make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
++ free_req(pending_req);
++ msleep(1); /* back off a bit */
++}
++
++
++
++/******************************************************************
++ * MISCELLANEOUS SETUP / TEARDOWN / DEBUGGING
++ */
++
++
++static void make_response(blkif_t *blkif, u64 id,
++ unsigned short op, int st)
++{
++ blkif_response_t resp;
++ unsigned long flags;
++ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
++ int more_to_do = 0;
++ int notify;
++
++ resp.id = id;
++ resp.operation = op;
++ resp.status = st;
++
++ spin_lock_irqsave(&blkif->blk_ring_lock, flags);
++ /* Place on the response ring for the relevant domain. */
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ case BLKIF_PROTOCOL_X86_32:
++ memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ case BLKIF_PROTOCOL_X86_64:
++ memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ default:
++ BUG();
++ }
++ blk_rings->common.rsp_prod_pvt++;
++ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
++ if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) {
++ /*
++ * Tail check for pending requests. Allows frontend to avoid
++ * notifications if requests are already in flight (lower
++ * overheads and promotes batching).
++ */
++ RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
++
++ } else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) {
++ more_to_do = 1;
++ }
++
++ spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
++
++ if (more_to_do)
++ blkif_notify_work(blkif);
++ if (notify)
++ notify_remote_via_irq(blkif->irq);
++}
++
++static int __init blkif_init(void)
++{
++ int i, mmap_pages;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ mmap_pages = blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
++
++ pending_reqs = kmalloc(sizeof(pending_reqs[0]) *
++ blkif_reqs, GFP_KERNEL);
++ pending_grant_handles = kmalloc(sizeof(pending_grant_handles[0]) *
++ mmap_pages, GFP_KERNEL);
++ pending_pages = alloc_empty_pages_and_pagevec(mmap_pages);
++
++ if (!pending_reqs || !pending_grant_handles || !pending_pages)
++ goto out_of_memory;
++
++ for (i = 0; i < mmap_pages; i++)
++ pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
++
++ blkif_interface_init();
++
++ memset(pending_reqs, 0, sizeof(pending_reqs));
++ INIT_LIST_HEAD(&pending_free);
++
++ for (i = 0; i < blkif_reqs; i++)
++ list_add_tail(&pending_reqs[i].free_list, &pending_free);
++
++ blkif_xenbus_init();
++
++ return 0;
++
++ out_of_memory:
++ kfree(pending_reqs);
++ kfree(pending_grant_handles);
++ free_empty_pages_and_pagevec(pending_pages, mmap_pages);
++ printk("%s: out of memory\n", __FUNCTION__);
++ return -ENOMEM;
++}
++
++module_init(blkif_init);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/blkback/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/common.h 2008-05-19 00:33:45.678779098 +0300
+@@ -0,0 +1,139 @@
++/*
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef __BLKIF__BACKEND__COMMON_H__
++#define __BLKIF__BACKEND__COMMON_H__
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/interrupt.h>
++#include <linux/slab.h>
++#include <linux/blkdev.h>
++#include <linux/vmalloc.h>
++#include <linux/wait.h>
++#include <asm/io.h>
++#include <asm/setup.h>
++#include <asm/pgalloc.h>
++#include <xen/evtchn.h>
++#include <asm/hypervisor.h>
++#include <xen/blkif.h>
++#include <xen/gnttab.h>
++#include <xen/driver_util.h>
++#include <xen/xenbus.h>
++
++#define DPRINTK(_f, _a...) \
++ pr_debug("(file=%s, line=%d) " _f, \
++ __FILE__ , __LINE__ , ## _a )
++
++struct vbd {
++ blkif_vdev_t handle; /* what the domain refers to this vbd as */
++ unsigned char readonly; /* Non-zero -> read-only */
++ unsigned char type; /* VDISK_xxx */
++ u32 pdevice; /* phys device that this vbd maps to */
++ struct block_device *bdev;
++};
++
++struct backend_info;
++
++typedef struct blkif_st {
++ /* Unique identifier for this interface. */
++ domid_t domid;
++ unsigned int handle;
++ /* Physical parameters of the comms window. */
++ unsigned int irq;
++ /* Comms information. */
++ enum blkif_protocol blk_protocol;
++ blkif_back_rings_t blk_rings;
++ struct vm_struct *blk_ring_area;
++ /* The VBD attached to this interface. */
++ struct vbd vbd;
++ /* Back pointer to the backend_info. */
++ struct backend_info *be;
++ /* Private fields. */
++ spinlock_t blk_ring_lock;
++ atomic_t refcnt;
++
++ wait_queue_head_t wq;
++ struct task_struct *xenblkd;
++ unsigned int waiting_reqs;
++ request_queue_t *plug;
++
++ /* statistics */
++ unsigned long st_print;
++ int st_rd_req;
++ int st_wr_req;
++ int st_oo_req;
++ int st_br_req;
++ int st_rd_sect;
++ int st_wr_sect;
++
++ wait_queue_head_t waiting_to_free;
++
++ grant_handle_t shmem_handle;
++ grant_ref_t shmem_ref;
++} blkif_t;
++
++blkif_t *blkif_alloc(domid_t domid);
++void blkif_disconnect(blkif_t *blkif);
++void blkif_free(blkif_t *blkif);
++int blkif_map(blkif_t *blkif, unsigned long shared_page, unsigned int evtchn);
++
++#define blkif_get(_b) (atomic_inc(&(_b)->refcnt))
++#define blkif_put(_b) \
++ do { \
++ if (atomic_dec_and_test(&(_b)->refcnt)) \
++ wake_up(&(_b)->waiting_to_free);\
++ } while (0)
++
++/* Create a vbd. */
++int vbd_create(blkif_t *blkif, blkif_vdev_t vdevice, unsigned major,
++ unsigned minor, int readonly, int cdrom);
++void vbd_free(struct vbd *vbd);
++
++unsigned long long vbd_size(struct vbd *vbd);
++unsigned int vbd_info(struct vbd *vbd);
++unsigned long vbd_secsize(struct vbd *vbd);
++
++struct phys_req {
++ unsigned short dev;
++ unsigned short nr_sects;
++ struct block_device *bdev;
++ blkif_sector_t sector_number;
++};
++
++int vbd_translate(struct phys_req *req, blkif_t *blkif, int operation);
++
++void blkif_interface_init(void);
++
++void blkif_xenbus_init(void);
++
++irqreturn_t blkif_be_int(int irq, void *dev_id, struct pt_regs *regs);
++int blkif_schedule(void *arg);
++
++int blkback_barrier(struct xenbus_transaction xbt,
++ struct backend_info *be, int state);
++
++#endif /* __BLKIF__BACKEND__COMMON_H__ */
+--- linux-2.6.18.8/drivers/xen/blkback/interface.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/interface.c 2008-05-19 00:33:45.682779328 +0300
+@@ -0,0 +1,181 @@
++/******************************************************************************
++ * arch/xen/drivers/blkif/backend/interface.c
++ *
++ * Block-device interface management.
++ *
++ * Copyright (c) 2004, Keir Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include "common.h"
++#include <xen/evtchn.h>
++#include <linux/kthread.h>
++
++static kmem_cache_t *blkif_cachep;
++
++blkif_t *blkif_alloc(domid_t domid)
++{
++ blkif_t *blkif;
++
++ blkif = kmem_cache_alloc(blkif_cachep, GFP_KERNEL);
++ if (!blkif)
++ return ERR_PTR(-ENOMEM);
++
++ memset(blkif, 0, sizeof(*blkif));
++ blkif->domid = domid;
++ spin_lock_init(&blkif->blk_ring_lock);
++ atomic_set(&blkif->refcnt, 1);
++ init_waitqueue_head(&blkif->wq);
++ blkif->st_print = jiffies;
++ init_waitqueue_head(&blkif->waiting_to_free);
++
++ return blkif;
++}
++
++static int map_frontend_page(blkif_t *blkif, unsigned long shared_page)
++{
++ struct gnttab_map_grant_ref op;
++
++ gnttab_set_map_op(&op, (unsigned long)blkif->blk_ring_area->addr,
++ GNTMAP_host_map, shared_page, blkif->domid);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
++ BUG();
++
++ if (op.status) {
++ DPRINTK(" Grant table operation failure !\n");
++ return op.status;
++ }
++
++ blkif->shmem_ref = shared_page;
++ blkif->shmem_handle = op.handle;
++
++ return 0;
++}
++
++static void unmap_frontend_page(blkif_t *blkif)
++{
++ struct gnttab_unmap_grant_ref op;
++
++ gnttab_set_unmap_op(&op, (unsigned long)blkif->blk_ring_area->addr,
++ GNTMAP_host_map, blkif->shmem_handle);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
++ BUG();
++}
++
++int blkif_map(blkif_t *blkif, unsigned long shared_page, unsigned int evtchn)
++{
++ int err;
++
++ /* Already connected through? */
++ if (blkif->irq)
++ return 0;
++
++ if ( (blkif->blk_ring_area = alloc_vm_area(PAGE_SIZE)) == NULL )
++ return -ENOMEM;
++
++ err = map_frontend_page(blkif, shared_page);
++ if (err) {
++ free_vm_area(blkif->blk_ring_area);
++ return err;
++ }
++
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ {
++ blkif_sring_t *sring;
++ sring = (blkif_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.native, sring, PAGE_SIZE);
++ break;
++ }
++ case BLKIF_PROTOCOL_X86_32:
++ {
++ blkif_x86_32_sring_t *sring_x86_32;
++ sring_x86_32 = (blkif_x86_32_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.x86_32, sring_x86_32, PAGE_SIZE);
++ break;
++ }
++ case BLKIF_PROTOCOL_X86_64:
++ {
++ blkif_x86_64_sring_t *sring_x86_64;
++ sring_x86_64 = (blkif_x86_64_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.x86_64, sring_x86_64, PAGE_SIZE);
++ break;
++ }
++ default:
++ BUG();
++ }
++
++ err = bind_interdomain_evtchn_to_irqhandler(
++ blkif->domid, evtchn, blkif_be_int, 0, "blkif-backend", blkif);
++ if (err < 0)
++ {
++ unmap_frontend_page(blkif);
++ free_vm_area(blkif->blk_ring_area);
++ blkif->blk_rings.common.sring = NULL;
++ return err;
++ }
++ blkif->irq = err;
++
++ return 0;
++}
++
++void blkif_disconnect(blkif_t *blkif)
++{
++ if (blkif->xenblkd) {
++ kthread_stop(blkif->xenblkd);
++ blkif->xenblkd = NULL;
++ }
++
++ atomic_dec(&blkif->refcnt);
++ wait_event(blkif->waiting_to_free, atomic_read(&blkif->refcnt) == 0);
++ atomic_inc(&blkif->refcnt);
++
++ if (blkif->irq) {
++ unbind_from_irqhandler(blkif->irq, blkif);
++ blkif->irq = 0;
++ }
++
++ if (blkif->blk_rings.common.sring) {
++ unmap_frontend_page(blkif);
++ free_vm_area(blkif->blk_ring_area);
++ blkif->blk_rings.common.sring = NULL;
++ }
++}
++
++void blkif_free(blkif_t *blkif)
++{
++ if (!atomic_dec_and_test(&blkif->refcnt))
++ BUG();
++ kmem_cache_free(blkif_cachep, blkif);
++}
++
++void __init blkif_interface_init(void)
++{
++ blkif_cachep = kmem_cache_create("blkif_cache", sizeof(blkif_t),
++ 0, 0, NULL, NULL);
++}
+--- linux-2.6.18.8/drivers/xen/blkback/vbd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/vbd.c 2008-05-19 00:33:45.682779328 +0300
+@@ -0,0 +1,118 @@
++/******************************************************************************
++ * blkback/vbd.c
++ *
++ * Routines for managing virtual block devices (VBDs).
++ *
++ * Copyright (c) 2003-2005, Keir Fraser & Steve Hand
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include "common.h"
++
++#define vbd_sz(_v) ((_v)->bdev->bd_part ? \
++ (_v)->bdev->bd_part->nr_sects : (_v)->bdev->bd_disk->capacity)
++
++unsigned long long vbd_size(struct vbd *vbd)
++{
++ return vbd_sz(vbd);
++}
++
++unsigned int vbd_info(struct vbd *vbd)
++{
++ return vbd->type | (vbd->readonly?VDISK_READONLY:0);
++}
++
++unsigned long vbd_secsize(struct vbd *vbd)
++{
++ return bdev_hardsect_size(vbd->bdev);
++}
++
++int vbd_create(blkif_t *blkif, blkif_vdev_t handle, unsigned major,
++ unsigned minor, int readonly, int cdrom)
++{
++ struct vbd *vbd;
++ struct block_device *bdev;
++
++ vbd = &blkif->vbd;
++ vbd->handle = handle;
++ vbd->readonly = readonly;
++ vbd->type = 0;
++
++ vbd->pdevice = MKDEV(major, minor);
++
++ bdev = open_by_devnum(vbd->pdevice,
++ vbd->readonly ? FMODE_READ : FMODE_WRITE);
++
++ if (IS_ERR(bdev)) {
++ DPRINTK("vbd_creat: device %08x could not be opened.\n",
++ vbd->pdevice);
++ return -ENOENT;
++ }
++
++ vbd->bdev = bdev;
++
++ if (vbd->bdev->bd_disk == NULL) {
++ DPRINTK("vbd_creat: device %08x doesn't exist.\n",
++ vbd->pdevice);
++ vbd_free(vbd);
++ return -ENOENT;
++ }
++
++ if (vbd->bdev->bd_disk->flags & GENHD_FL_CD || cdrom)
++ vbd->type |= VDISK_CDROM;
++ if (vbd->bdev->bd_disk->flags & GENHD_FL_REMOVABLE)
++ vbd->type |= VDISK_REMOVABLE;
++
++ DPRINTK("Successful creation of handle=%04x (dom=%u)\n",
++ handle, blkif->domid);
++ return 0;
++}
++
++void vbd_free(struct vbd *vbd)
++{
++ if (vbd->bdev)
++ blkdev_put(vbd->bdev);
++ vbd->bdev = NULL;
++}
++
++int vbd_translate(struct phys_req *req, blkif_t *blkif, int operation)
++{
++ struct vbd *vbd = &blkif->vbd;
++ int rc = -EACCES;
++
++ if ((operation != READ) && vbd->readonly)
++ goto out;
++
++ if (unlikely((req->sector_number + req->nr_sects) > vbd_sz(vbd)))
++ goto out;
++
++ req->dev = vbd->pdevice;
++ req->bdev = vbd->bdev;
++ rc = 0;
++
++ out:
++ return rc;
++}
+--- linux-2.6.18.8/drivers/xen/blkback/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkback/xenbus.c 2008-05-19 00:33:45.682779328 +0300
+@@ -0,0 +1,541 @@
++/* Xenbus code for blkif backend
++ Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
++ Copyright (C) 2005 XenSource Ltd
++
++ This program is free software; you can redistribute it and/or modify
++ it under the terms of the GNU General Public License as published by
++ the Free Software Foundation; either version 2 of the License, or
++ (at your option) any later version.
++
++ This program is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ GNU General Public License for more details.
++
++ You should have received a copy of the GNU General Public License
++ along with this program; if not, write to the Free Software
++ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++*/
++
++#include <stdarg.h>
++#include <linux/module.h>
++#include <linux/kthread.h>
++#include "common.h"
++
++#undef DPRINTK
++#define DPRINTK(fmt, args...) \
++ pr_debug("blkback/xenbus (%s:%d) " fmt ".\n", \
++ __FUNCTION__, __LINE__, ##args)
++
++struct backend_info
++{
++ struct xenbus_device *dev;
++ blkif_t *blkif;
++ struct xenbus_watch backend_watch;
++ unsigned major;
++ unsigned minor;
++ char *mode;
++};
++
++static void connect(struct backend_info *);
++static int connect_ring(struct backend_info *);
++static void backend_changed(struct xenbus_watch *, const char **,
++ unsigned int);
++
++static int blkback_name(blkif_t *blkif, char *buf)
++{
++ char *devpath, *devname;
++ struct xenbus_device *dev = blkif->be->dev;
++
++ devpath = xenbus_read(XBT_NIL, dev->nodename, "dev", NULL);
++ if (IS_ERR(devpath))
++ return PTR_ERR(devpath);
++
++ if ((devname = strstr(devpath, "/dev/")) != NULL)
++ devname += strlen("/dev/");
++ else
++ devname = devpath;
++
++ snprintf(buf, TASK_COMM_LEN, "blkback.%d.%s", blkif->domid, devname);
++ kfree(devpath);
++
++ return 0;
++}
++
++static void update_blkif_status(blkif_t *blkif)
++{
++ int err;
++ char name[TASK_COMM_LEN];
++
++ /* Not ready to connect? */
++ if (!blkif->irq || !blkif->vbd.bdev)
++ return;
++
++ /* Already connected? */
++ if (blkif->be->dev->state == XenbusStateConnected)
++ return;
++
++ /* Attempt to connect: exit if we fail to. */
++ connect(blkif->be);
++ if (blkif->be->dev->state != XenbusStateConnected)
++ return;
++
++ err = blkback_name(blkif, name);
++ if (err) {
++ xenbus_dev_error(blkif->be->dev, err, "get blkback dev name");
++ return;
++ }
++
++ blkif->xenblkd = kthread_run(blkif_schedule, blkif, name);
++ if (IS_ERR(blkif->xenblkd)) {
++ err = PTR_ERR(blkif->xenblkd);
++ blkif->xenblkd = NULL;
++ xenbus_dev_error(blkif->be->dev, err, "start xenblkd");
++ }
++}
++
++
++/****************************************************************
++ * sysfs interface for VBD I/O requests
++ */
++
++#define VBD_SHOW(name, format, args...) \
++ static ssize_t show_##name(struct device *_dev, \
++ struct device_attribute *attr, \
++ char *buf) \
++ { \
++ struct xenbus_device *dev = to_xenbus_device(_dev); \
++ struct backend_info *be = dev->dev.driver_data; \
++ \
++ return sprintf(buf, format, ##args); \
++ } \
++ static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL)
++
++VBD_SHOW(oo_req, "%d\n", be->blkif->st_oo_req);
++VBD_SHOW(rd_req, "%d\n", be->blkif->st_rd_req);
++VBD_SHOW(wr_req, "%d\n", be->blkif->st_wr_req);
++VBD_SHOW(br_req, "%d\n", be->blkif->st_br_req);
++VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect);
++VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect);
++
++static struct attribute *vbdstat_attrs[] = {
++ &dev_attr_oo_req.attr,
++ &dev_attr_rd_req.attr,
++ &dev_attr_wr_req.attr,
++ &dev_attr_br_req.attr,
++ &dev_attr_rd_sect.attr,
++ &dev_attr_wr_sect.attr,
++ NULL
++};
++
++static struct attribute_group vbdstat_group = {
++ .name = "statistics",
++ .attrs = vbdstat_attrs,
++};
++
++VBD_SHOW(physical_device, "%x:%x\n", be->major, be->minor);
++VBD_SHOW(mode, "%s\n", be->mode);
++
++int xenvbd_sysfs_addif(struct xenbus_device *dev)
++{
++ int error;
++
++ error = device_create_file(&dev->dev, &dev_attr_physical_device);
++ if (error)
++ goto fail1;
++
++ error = device_create_file(&dev->dev, &dev_attr_mode);
++ if (error)
++ goto fail2;
++
++ error = sysfs_create_group(&dev->dev.kobj, &vbdstat_group);
++ if (error)
++ goto fail3;
++
++ return 0;
++
++fail3: sysfs_remove_group(&dev->dev.kobj, &vbdstat_group);
++fail2: device_remove_file(&dev->dev, &dev_attr_mode);
++fail1: device_remove_file(&dev->dev, &dev_attr_physical_device);
++ return error;
++}
++
++void xenvbd_sysfs_delif(struct xenbus_device *dev)
++{
++ sysfs_remove_group(&dev->dev.kobj, &vbdstat_group);
++ device_remove_file(&dev->dev, &dev_attr_mode);
++ device_remove_file(&dev->dev, &dev_attr_physical_device);
++}
++
++static int blkback_remove(struct xenbus_device *dev)
++{
++ struct backend_info *be = dev->dev.driver_data;
++
++ DPRINTK("");
++
++ if (be->major || be->minor)
++ xenvbd_sysfs_delif(dev);
++
++ if (be->backend_watch.node) {
++ unregister_xenbus_watch(&be->backend_watch);
++ kfree(be->backend_watch.node);
++ be->backend_watch.node = NULL;
++ }
++
++ if (be->blkif) {
++ blkif_disconnect(be->blkif);
++ vbd_free(&be->blkif->vbd);
++ blkif_free(be->blkif);
++ be->blkif = NULL;
++ }
++
++ kfree(be);
++ dev->dev.driver_data = NULL;
++ return 0;
++}
++
++int blkback_barrier(struct xenbus_transaction xbt,
++ struct backend_info *be, int state)
++{
++ struct xenbus_device *dev = be->dev;
++ int err;
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-barrier",
++ "%d", state);
++ if (err)
++ xenbus_dev_fatal(dev, err, "writing feature-barrier");
++
++ return err;
++}
++
++/**
++ * Entry point to this code when a new device is created. Allocate the basic
++ * structures, and watch the store waiting for the hotplug scripts to tell us
++ * the device's physical major and minor numbers. Switch to InitWait.
++ */
++static int blkback_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int err;
++ struct backend_info *be = kzalloc(sizeof(struct backend_info),
++ GFP_KERNEL);
++ if (!be) {
++ xenbus_dev_fatal(dev, -ENOMEM,
++ "allocating backend structure");
++ return -ENOMEM;
++ }
++ be->dev = dev;
++ dev->dev.driver_data = be;
++
++ be->blkif = blkif_alloc(dev->otherend_id);
++ if (IS_ERR(be->blkif)) {
++ err = PTR_ERR(be->blkif);
++ be->blkif = NULL;
++ xenbus_dev_fatal(dev, err, "creating block interface");
++ goto fail;
++ }
++
++ /* setup back pointer */
++ be->blkif->be = be;
++
++ err = xenbus_watch_path2(dev, dev->nodename, "physical-device",
++ &be->backend_watch, backend_changed);
++ if (err)
++ goto fail;
++
++ err = xenbus_switch_state(dev, XenbusStateInitWait);
++ if (err)
++ goto fail;
++
++ return 0;
++
++fail:
++ DPRINTK("failed");
++ blkback_remove(dev);
++ return err;
++}
++
++
++/**
++ * Callback received when the hotplug scripts have placed the physical-device
++ * node. Read it and the mode node, and create a vbd. If the frontend is
++ * ready, connect.
++ */
++static void backend_changed(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ int err;
++ unsigned major;
++ unsigned minor;
++ struct backend_info *be
++ = container_of(watch, struct backend_info, backend_watch);
++ struct xenbus_device *dev = be->dev;
++ int cdrom = 0;
++ char *device_type;
++
++ DPRINTK("");
++
++ err = xenbus_scanf(XBT_NIL, dev->nodename, "physical-device", "%x:%x",
++ &major, &minor);
++ if (XENBUS_EXIST_ERR(err)) {
++ /* Since this watch will fire once immediately after it is
++ registered, we expect this. Ignore it, and wait for the
++ hotplug scripts. */
++ return;
++ }
++ if (err != 2) {
++ xenbus_dev_fatal(dev, err, "reading physical-device");
++ return;
++ }
++
++ if ((be->major || be->minor) &&
++ ((be->major != major) || (be->minor != minor))) {
++ printk(KERN_WARNING
++ "blkback: changing physical device (from %x:%x to "
++ "%x:%x) not supported.\n", be->major, be->minor,
++ major, minor);
++ return;
++ }
++
++ be->mode = xenbus_read(XBT_NIL, dev->nodename, "mode", NULL);
++ if (IS_ERR(be->mode)) {
++ err = PTR_ERR(be->mode);
++ be->mode = NULL;
++ xenbus_dev_fatal(dev, err, "reading mode");
++ return;
++ }
++
++ device_type = xenbus_read(XBT_NIL, dev->otherend, "device-type", NULL);
++ if (!IS_ERR(device_type)) {
++ cdrom = strcmp(device_type, "cdrom") == 0;
++ kfree(device_type);
++ }
++
++ if (be->major == 0 && be->minor == 0) {
++ /* Front end dir is a number, which is used as the handle. */
++
++ char *p = strrchr(dev->otherend, '/') + 1;
++ long handle = simple_strtoul(p, NULL, 0);
++
++ be->major = major;
++ be->minor = minor;
++
++ err = vbd_create(be->blkif, handle, major, minor,
++ (NULL == strchr(be->mode, 'w')), cdrom);
++ if (err) {
++ be->major = be->minor = 0;
++ xenbus_dev_fatal(dev, err, "creating vbd structure");
++ return;
++ }
++
++ err = xenvbd_sysfs_addif(dev);
++ if (err) {
++ vbd_free(&be->blkif->vbd);
++ be->major = be->minor = 0;
++ xenbus_dev_fatal(dev, err, "creating sysfs entries");
++ return;
++ }
++
++ /* We're potentially connected now */
++ update_blkif_status(be->blkif);
++ }
++}
++
++
++/**
++ * Callback received when the frontend's state changes.
++ */
++static void frontend_changed(struct xenbus_device *dev,
++ enum xenbus_state frontend_state)
++{
++ struct backend_info *be = dev->dev.driver_data;
++ int err;
++
++ DPRINTK("%s", xenbus_strstate(frontend_state));
++
++ switch (frontend_state) {
++ case XenbusStateInitialising:
++ if (dev->state == XenbusStateClosed) {
++ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
++ __FUNCTION__, dev->nodename);
++ xenbus_switch_state(dev, XenbusStateInitWait);
++ }
++ break;
++
++ case XenbusStateInitialised:
++ case XenbusStateConnected:
++ /* Ensure we connect even when two watches fire in
++ close successsion and we miss the intermediate value
++ of frontend_state. */
++ if (dev->state == XenbusStateConnected)
++ break;
++
++ err = connect_ring(be);
++ if (err)
++ break;
++ update_blkif_status(be->blkif);
++ break;
++
++ case XenbusStateClosing:
++ blkif_disconnect(be->blkif);
++ xenbus_switch_state(dev, XenbusStateClosing);
++ break;
++
++ case XenbusStateClosed:
++ xenbus_switch_state(dev, XenbusStateClosed);
++ if (xenbus_dev_is_online(dev))
++ break;
++ /* fall through if not online */
++ case XenbusStateUnknown:
++ device_unregister(&dev->dev);
++ break;
++
++ default:
++ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
++ frontend_state);
++ break;
++ }
++}
++
++
++/* ** Connection ** */
++
++
++/**
++ * Write the physical details regarding the block device to the store, and
++ * switch to Connected state.
++ */
++static void connect(struct backend_info *be)
++{
++ struct xenbus_transaction xbt;
++ int err;
++ struct xenbus_device *dev = be->dev;
++
++ DPRINTK("%s", dev->otherend);
++
++ /* Supply the information about the device the frontend needs */
++again:
++ err = xenbus_transaction_start(&xbt);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "starting transaction");
++ return;
++ }
++
++ err = blkback_barrier(xbt, be, 1);
++ if (err)
++ goto abort;
++
++ err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
++ vbd_size(&be->blkif->vbd));
++ if (err) {
++ xenbus_dev_fatal(dev, err, "writing %s/sectors",
++ dev->nodename);
++ goto abort;
++ }
++
++ /* FIXME: use a typename instead */
++ err = xenbus_printf(xbt, dev->nodename, "info", "%u",
++ vbd_info(&be->blkif->vbd));
++ if (err) {
++ xenbus_dev_fatal(dev, err, "writing %s/info",
++ dev->nodename);
++ goto abort;
++ }
++ err = xenbus_printf(xbt, dev->nodename, "sector-size", "%lu",
++ vbd_secsize(&be->blkif->vbd));
++ if (err) {
++ xenbus_dev_fatal(dev, err, "writing %s/sector-size",
++ dev->nodename);
++ goto abort;
++ }
++
++ err = xenbus_transaction_end(xbt, 0);
++ if (err == -EAGAIN)
++ goto again;
++ if (err)
++ xenbus_dev_fatal(dev, err, "ending transaction");
++
++ err = xenbus_switch_state(dev, XenbusStateConnected);
++ if (err)
++ xenbus_dev_fatal(dev, err, "switching to Connected state",
++ dev->nodename);
++
++ return;
++ abort:
++ xenbus_transaction_end(xbt, 1);
++}
++
++
++static int connect_ring(struct backend_info *be)
++{
++ struct xenbus_device *dev = be->dev;
++ unsigned long ring_ref;
++ unsigned int evtchn;
++ char protocol[64] = "";
++ int err;
++
++ DPRINTK("%s", dev->otherend);
++
++ err = xenbus_gather(XBT_NIL, dev->otherend, "ring-ref", "%lu", &ring_ref,
++ "event-channel", "%u", &evtchn, NULL);
++ if (err) {
++ xenbus_dev_fatal(dev, err,
++ "reading %s/ring-ref and event-channel",
++ dev->otherend);
++ return err;
++ }
++
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
++ err = xenbus_gather(XBT_NIL, dev->otherend, "protocol",
++ "%63s", protocol, NULL);
++ if (err)
++ strcpy(protocol, "unspecified, assuming native");
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_NATIVE))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_32))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_32;
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_64))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_64;
++ else {
++ xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol);
++ return -1;
++ }
++ printk(KERN_INFO
++ "blkback: ring-ref %ld, event-channel %d, protocol %d (%s)\n",
++ ring_ref, evtchn, be->blkif->blk_protocol, protocol);
++
++ /* Map the shared frame, irq etc. */
++ err = blkif_map(be->blkif, ring_ref, evtchn);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "mapping ring-ref %lu port %u",
++ ring_ref, evtchn);
++ return err;
++ }
++
++ return 0;
++}
++
++
++/* ** Driver Registration ** */
++
++
++static const struct xenbus_device_id blkback_ids[] = {
++ { "vbd" },
++ { "" }
++};
++
++
++static struct xenbus_driver blkback = {
++ .name = "vbd",
++ .owner = THIS_MODULE,
++ .ids = blkback_ids,
++ .probe = blkback_probe,
++ .remove = blkback_remove,
++ .otherend_changed = frontend_changed
++};
++
++
++void blkif_xenbus_init(void)
++{
++ xenbus_register_backend(&blkback);
++}
+--- linux-2.6.18.8/drivers/xen/blkfront/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkfront/Makefile 2008-05-19 00:33:45.686779559 +0300
+@@ -0,0 +1,5 @@
++
++obj-$(CONFIG_XEN_BLKDEV_FRONTEND) := xenblk.o
++
++xenblk-objs := blkfront.o vbd.o
++
+--- linux-2.6.18.8/drivers/xen/blkfront/blkfront.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkfront/blkfront.c 2008-05-19 00:33:45.686779559 +0300
+@@ -0,0 +1,922 @@
++/******************************************************************************
++ * blkfront.c
++ *
++ * XenLinux virtual block-device driver.
++ *
++ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
++ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
++ * Copyright (c) 2004, Christian Limpach
++ * Copyright (c) 2004, Andrew Warfield
++ * Copyright (c) 2005, Christopher Clark
++ * Copyright (c) 2005, XenSource Ltd
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/version.h>
++#include "block.h"
++#include <linux/cdrom.h>
++#include <linux/sched.h>
++#include <linux/interrupt.h>
++#include <scsi/scsi.h>
++#include <xen/evtchn.h>
++#include <xen/xenbus.h>
++#include <xen/interface/grant_table.h>
++#include <xen/interface/io/protocols.h>
++#include <xen/gnttab.h>
++#include <asm/hypervisor.h>
++#include <asm/maddr.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++#define BLKIF_STATE_DISCONNECTED 0
++#define BLKIF_STATE_CONNECTED 1
++#define BLKIF_STATE_SUSPENDED 2
++
++#define MAXIMUM_OUTSTANDING_BLOCK_REQS \
++ (BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
++#define GRANT_INVALID_REF 0
++
++static void connect(struct blkfront_info *);
++static void blkfront_closing(struct xenbus_device *);
++static int blkfront_remove(struct xenbus_device *);
++static int talk_to_backend(struct xenbus_device *, struct blkfront_info *);
++static int setup_blkring(struct xenbus_device *, struct blkfront_info *);
++
++static void kick_pending_request_queues(struct blkfront_info *);
++
++static irqreturn_t blkif_int(int irq, void *dev_id, struct pt_regs *ptregs);
++static void blkif_restart_queue(void *arg);
++static void blkif_recover(struct blkfront_info *);
++static void blkif_completion(struct blk_shadow *);
++static void blkif_free(struct blkfront_info *, int);
++
++
++/**
++ * Entry point to this code when a new device is created. Allocate the basic
++ * structures and the ring buffer for communication with the backend, and
++ * inform the backend of the appropriate details for those. Switch to
++ * Initialised state.
++ */
++static int blkfront_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int err, vdevice, i;
++ struct blkfront_info *info;
++
++ /* FIXME: Use dynamic device id if this is not set. */
++ err = xenbus_scanf(XBT_NIL, dev->nodename,
++ "virtual-device", "%i", &vdevice);
++ if (err != 1) {
++ xenbus_dev_fatal(dev, err, "reading virtual-device");
++ return err;
++ }
++
++ info = kzalloc(sizeof(*info), GFP_KERNEL);
++ if (!info) {
++ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
++ return -ENOMEM;
++ }
++
++ info->xbdev = dev;
++ info->vdevice = vdevice;
++ info->connected = BLKIF_STATE_DISCONNECTED;
++ INIT_WORK(&info->work, blkif_restart_queue, (void *)info);
++
++ for (i = 0; i < BLK_RING_SIZE; i++)
++ info->shadow[i].req.id = i+1;
++ info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
++
++ /* Front end dir is a number, which is used as the id. */
++ info->handle = simple_strtoul(strrchr(dev->nodename,'/')+1, NULL, 0);
++ dev->dev.driver_data = info;
++
++ err = talk_to_backend(dev, info);
++ if (err) {
++ kfree(info);
++ dev->dev.driver_data = NULL;
++ return err;
++ }
++
++ return 0;
++}
++
++
++/**
++ * We are reconnecting to the backend, due to a suspend/resume, or a backend
++ * driver restart. We tear down our blkif structure and recreate it, but
++ * leave the device-layer structures intact so that this is transparent to the
++ * rest of the kernel.
++ */
++static int blkfront_resume(struct xenbus_device *dev)
++{
++ struct blkfront_info *info = dev->dev.driver_data;
++ int err;
++
++ DPRINTK("blkfront_resume: %s\n", dev->nodename);
++
++ blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
++
++ err = talk_to_backend(dev, info);
++ if (info->connected == BLKIF_STATE_SUSPENDED && !err)
++ blkif_recover(info);
++
++ return err;
++}
++
++
++/* Common code used when first setting up, and when resuming. */
++static int talk_to_backend(struct xenbus_device *dev,
++ struct blkfront_info *info)
++{
++ const char *message = NULL;
++ struct xenbus_transaction xbt;
++ int err;
++
++ /* Create shared ring, alloc event channel. */
++ err = setup_blkring(dev, info);
++ if (err)
++ goto out;
++
++again:
++ err = xenbus_transaction_start(&xbt);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "starting transaction");
++ goto destroy_blkring;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename,
++ "ring-ref","%u", info->ring_ref);
++ if (err) {
++ message = "writing ring-ref";
++ goto abort_transaction;
++ }
++ err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
++ irq_to_evtchn_port(info->irq));
++ if (err) {
++ message = "writing event-channel";
++ goto abort_transaction;
++ }
++ err = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
++ XEN_IO_PROTO_ABI_NATIVE);
++ if (err) {
++ message = "writing protocol";
++ goto abort_transaction;
++ }
++
++ err = xenbus_transaction_end(xbt, 0);
++ if (err) {
++ if (err == -EAGAIN)
++ goto again;
++ xenbus_dev_fatal(dev, err, "completing transaction");
++ goto destroy_blkring;
++ }
++
++ xenbus_switch_state(dev, XenbusStateInitialised);
++
++ return 0;
++
++ abort_transaction:
++ xenbus_transaction_end(xbt, 1);
++ if (message)
++ xenbus_dev_fatal(dev, err, "%s", message);
++ destroy_blkring:
++ blkif_free(info, 0);
++ out:
++ return err;
++}
++
++
++static int setup_blkring(struct xenbus_device *dev,
++ struct blkfront_info *info)
++{
++ blkif_sring_t *sring;
++ int err;
++
++ info->ring_ref = GRANT_INVALID_REF;
++
++ sring = (blkif_sring_t *)__get_free_page(GFP_KERNEL|__GFP_HIGH);
++ if (!sring) {
++ xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
++ return -ENOMEM;
++ }
++ SHARED_RING_INIT(sring);
++ FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
++
++ err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
++ if (err < 0) {
++ free_page((unsigned long)sring);
++ info->ring.sring = NULL;
++ goto fail;
++ }
++ info->ring_ref = err;
++
++ err = bind_listening_port_to_irqhandler(
++ dev->otherend_id, blkif_int, SA_SAMPLE_RANDOM, "blkif", info);
++ if (err <= 0) {
++ xenbus_dev_fatal(dev, err,
++ "bind_listening_port_to_irqhandler");
++ goto fail;
++ }
++ info->irq = err;
++
++ return 0;
++fail:
++ blkif_free(info, 0);
++ return err;
++}
++
++
++/**
++ * Callback received when the backend's state changes.
++ */
++static void backend_changed(struct xenbus_device *dev,
++ enum xenbus_state backend_state)
++{
++ struct blkfront_info *info = dev->dev.driver_data;
++ struct block_device *bd;
++
++ DPRINTK("blkfront:backend_changed.\n");
++
++ switch (backend_state) {
++ case XenbusStateInitialising:
++ case XenbusStateInitWait:
++ case XenbusStateInitialised:
++ case XenbusStateReconfiguring:
++ case XenbusStateReconfigured:
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ break;
++
++ case XenbusStateConnected:
++ connect(info);
++ break;
++
++ case XenbusStateClosing:
++ bd = bdget(info->dev);
++ if (bd == NULL)
++ xenbus_dev_fatal(dev, -ENODEV, "bdget failed");
++
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
++ down(&bd->bd_sem);
++#else
++ mutex_lock(&bd->bd_mutex);
++#endif
++ if (info->users > 0)
++ xenbus_dev_error(dev, -EBUSY,
++ "Device in use; refusing to close");
++ else
++ blkfront_closing(dev);
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17)
++ up(&bd->bd_sem);
++#else
++ mutex_unlock(&bd->bd_mutex);
++#endif
++ bdput(bd);
++ break;
++ }
++}
++
++
++/* ** Connection ** */
++
++
++/*
++ * Invoked when the backend is finally 'ready' (and has told produced
++ * the details about the physical device - #sectors, size, etc).
++ */
++static void connect(struct blkfront_info *info)
++{
++ unsigned long long sectors;
++ unsigned long sector_size;
++ unsigned int binfo;
++ int err;
++
++ if ((info->connected == BLKIF_STATE_CONNECTED) ||
++ (info->connected == BLKIF_STATE_SUSPENDED) )
++ return;
++
++ DPRINTK("blkfront.c:connect:%s.\n", info->xbdev->otherend);
++
++ err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
++ "sectors", "%Lu", §ors,
++ "info", "%u", &binfo,
++ "sector-size", "%lu", §or_size,
++ NULL);
++ if (err) {
++ xenbus_dev_fatal(info->xbdev, err,
++ "reading backend fields at %s",
++ info->xbdev->otherend);
++ return;
++ }
++
++ err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
++ "feature-barrier", "%lu", &info->feature_barrier,
++ NULL);
++ if (err)
++ info->feature_barrier = 0;
++
++ err = xlvbd_add(sectors, info->vdevice, binfo, sector_size, info);
++ if (err) {
++ xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
++ info->xbdev->otherend);
++ return;
++ }
++
++ (void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
++
++ /* Kick pending requests. */
++ spin_lock_irq(&blkif_io_lock);
++ info->connected = BLKIF_STATE_CONNECTED;
++ kick_pending_request_queues(info);
++ spin_unlock_irq(&blkif_io_lock);
++
++ add_disk(info->gd);
++
++ info->is_ready = 1;
++}
++
++/**
++ * Handle the change of state of the backend to Closing. We must delete our
++ * device-layer structures now, to ensure that writes are flushed through to
++ * the backend. Once is this done, we can switch to Closed in
++ * acknowledgement.
++ */
++static void blkfront_closing(struct xenbus_device *dev)
++{
++ struct blkfront_info *info = dev->dev.driver_data;
++ unsigned long flags;
++
++ DPRINTK("blkfront_closing: %s removed\n", dev->nodename);
++
++ if (info->rq == NULL)
++ goto out;
++
++ spin_lock_irqsave(&blkif_io_lock, flags);
++ /* No more blkif_request(). */
++ blk_stop_queue(info->rq);
++ /* No more gnttab callback work. */
++ gnttab_cancel_free_callback(&info->callback);
++ spin_unlock_irqrestore(&blkif_io_lock, flags);
++
++ /* Flush gnttab callback work. Must be done with no locks held. */
++ flush_scheduled_work();
++
++ xlvbd_del(info);
++
++ out:
++ xenbus_frontend_closed(dev);
++}
++
++
++static int blkfront_remove(struct xenbus_device *dev)
++{
++ struct blkfront_info *info = dev->dev.driver_data;
++
++ DPRINTK("blkfront_remove: %s removed\n", dev->nodename);
++
++ blkif_free(info, 0);
++
++ kfree(info);
++
++ return 0;
++}
++
++
++static inline int GET_ID_FROM_FREELIST(
++ struct blkfront_info *info)
++{
++ unsigned long free = info->shadow_free;
++ BUG_ON(free > BLK_RING_SIZE);
++ info->shadow_free = info->shadow[free].req.id;
++ info->shadow[free].req.id = 0x0fffffee; /* debug */
++ return free;
++}
++
++static inline void ADD_ID_TO_FREELIST(
++ struct blkfront_info *info, unsigned long id)
++{
++ info->shadow[id].req.id = info->shadow_free;
++ info->shadow[id].request = 0;
++ info->shadow_free = id;
++}
++
++static inline void flush_requests(struct blkfront_info *info)
++{
++ int notify;
++
++ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&info->ring, notify);
++
++ if (notify)
++ notify_remote_via_irq(info->irq);
++}
++
++static void kick_pending_request_queues(struct blkfront_info *info)
++{
++ if (!RING_FULL(&info->ring)) {
++ /* Re-enable calldowns. */
++ blk_start_queue(info->rq);
++ /* Kick things off immediately. */
++ do_blkif_request(info->rq);
++ }
++}
++
++static void blkif_restart_queue(void *arg)
++{
++ struct blkfront_info *info = (struct blkfront_info *)arg;
++ spin_lock_irq(&blkif_io_lock);
++ if (info->connected == BLKIF_STATE_CONNECTED)
++ kick_pending_request_queues(info);
++ spin_unlock_irq(&blkif_io_lock);
++}
++
++static void blkif_restart_queue_callback(void *arg)
++{
++ struct blkfront_info *info = (struct blkfront_info *)arg;
++ schedule_work(&info->work);
++}
++
++int blkif_open(struct inode *inode, struct file *filep)
++{
++ struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
++ info->users++;
++ return 0;
++}
++
++
++int blkif_release(struct inode *inode, struct file *filep)
++{
++ struct blkfront_info *info = inode->i_bdev->bd_disk->private_data;
++ info->users--;
++ if (info->users == 0) {
++ /* Check whether we have been instructed to close. We will
++ have ignored this request initially, as the device was
++ still mounted. */
++ struct xenbus_device * dev = info->xbdev;
++ enum xenbus_state state = xenbus_read_driver_state(dev->otherend);
++
++ if (state == XenbusStateClosing && info->is_ready)
++ blkfront_closing(dev);
++ }
++ return 0;
++}
++
++
++int blkif_ioctl(struct inode *inode, struct file *filep,
++ unsigned command, unsigned long argument)
++{
++ int i;
++
++ DPRINTK_IOCTL("command: 0x%x, argument: 0x%lx, dev: 0x%04x\n",
++ command, (long)argument, inode->i_rdev);
++
++ switch (command) {
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16)
++ case HDIO_GETGEO: {
++ struct block_device *bd = inode->i_bdev;
++ struct hd_geometry geo;
++ int ret;
++
++ if (!argument)
++ return -EINVAL;
++
++ geo.start = get_start_sect(bd);
++ ret = blkif_getgeo(bd, &geo);
++ if (ret)
++ return ret;
++
++ if (copy_to_user((struct hd_geometry __user *)argument, &geo,
++ sizeof(geo)))
++ return -EFAULT;
++
++ return 0;
++ }
++#endif
++ case CDROMMULTISESSION:
++ DPRINTK("FIXME: support multisession CDs later\n");
++ for (i = 0; i < sizeof(struct cdrom_multisession); i++)
++ if (put_user(0, (char __user *)(argument + i)))
++ return -EFAULT;
++ return 0;
++
++ case CDROM_GET_CAPABILITY: {
++ struct blkfront_info *info =
++ inode->i_bdev->bd_disk->private_data;
++ struct gendisk *gd = info->gd;
++ if (gd->flags & GENHD_FL_CD)
++ return 0;
++ return -EINVAL;
++ }
++ default:
++ /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
++ command);*/
++ return -EINVAL; /* same return as native Linux */
++ }
++
++ return 0;
++}
++
++
++int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
++{
++ /* We don't have real geometry info, but let's at least return
++ values consistent with the size of the device */
++ sector_t nsect = get_capacity(bd->bd_disk);
++ sector_t cylinders = nsect;
++
++ hg->heads = 0xff;
++ hg->sectors = 0x3f;
++ sector_div(cylinders, hg->heads * hg->sectors);
++ hg->cylinders = cylinders;
++ if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
++ hg->cylinders = 0xffff;
++ return 0;
++}
++
++
++/*
++ * blkif_queue_request
++ *
++ * request block io
++ *
++ * id: for guest use only.
++ * operation: BLKIF_OP_{READ,WRITE,PROBE}
++ * buffer: buffer to read/write into. this should be a
++ * virtual address in the guest os.
++ */
++static int blkif_queue_request(struct request *req)
++{
++ struct blkfront_info *info = req->rq_disk->private_data;
++ unsigned long buffer_mfn;
++ blkif_request_t *ring_req;
++ struct bio *bio;
++ struct bio_vec *bvec;
++ int idx;
++ unsigned long id;
++ unsigned int fsect, lsect;
++ int ref;
++ grant_ref_t gref_head;
++
++ if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
++ return 1;
++
++ if (gnttab_alloc_grant_references(
++ BLKIF_MAX_SEGMENTS_PER_REQUEST, &gref_head) < 0) {
++ gnttab_request_free_callback(
++ &info->callback,
++ blkif_restart_queue_callback,
++ info,
++ BLKIF_MAX_SEGMENTS_PER_REQUEST);
++ return 1;
++ }
++
++ /* Fill out a communications ring structure. */
++ ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
++ id = GET_ID_FROM_FREELIST(info);
++ info->shadow[id].request = (unsigned long)req;
++
++ ring_req->id = id;
++ ring_req->sector_number = (blkif_sector_t)req->sector;
++ ring_req->handle = info->handle;
++
++ ring_req->operation = rq_data_dir(req) ?
++ BLKIF_OP_WRITE : BLKIF_OP_READ;
++ if (blk_barrier_rq(req))
++ ring_req->operation = BLKIF_OP_WRITE_BARRIER;
++
++ ring_req->nr_segments = 0;
++ rq_for_each_bio (bio, req) {
++ bio_for_each_segment (bvec, bio, idx) {
++ BUG_ON(ring_req->nr_segments
++ == BLKIF_MAX_SEGMENTS_PER_REQUEST);
++ buffer_mfn = page_to_phys(bvec->bv_page) >> PAGE_SHIFT;
++ fsect = bvec->bv_offset >> 9;
++ lsect = fsect + (bvec->bv_len >> 9) - 1;
++ /* install a grant reference. */
++ ref = gnttab_claim_grant_reference(&gref_head);
++ BUG_ON(ref == -ENOSPC);
++
++ gnttab_grant_foreign_access_ref(
++ ref,
++ info->xbdev->otherend_id,
++ buffer_mfn,
++ rq_data_dir(req) ? GTF_readonly : 0 );
++
++ info->shadow[id].frame[ring_req->nr_segments] =
++ mfn_to_pfn(buffer_mfn);
++
++ ring_req->seg[ring_req->nr_segments] =
++ (struct blkif_request_segment) {
++ .gref = ref,
++ .first_sect = fsect,
++ .last_sect = lsect };
++
++ ring_req->nr_segments++;
++ }
++ }
++
++ info->ring.req_prod_pvt++;
++
++ /* Keep a private copy so we can reissue requests when recovering. */
++ info->shadow[id].req = *ring_req;
++
++ gnttab_free_grant_references(gref_head);
++
++ return 0;
++}
++
++/*
++ * do_blkif_request
++ * read a block; request is in a request queue
++ */
++void do_blkif_request(request_queue_t *rq)
++{
++ struct blkfront_info *info = NULL;
++ struct request *req;
++ int queued;
++
++ DPRINTK("Entered do_blkif_request\n");
++
++ queued = 0;
++
++ while ((req = elv_next_request(rq)) != NULL) {
++ info = req->rq_disk->private_data;
++ if (!blk_fs_request(req)) {
++ end_request(req, 0);
++ continue;
++ }
++
++ if (RING_FULL(&info->ring))
++ goto wait;
++
++ DPRINTK("do_blk_req %p: cmd %p, sec %llx, "
++ "(%u/%li) buffer:%p [%s]\n",
++ req, req->cmd, (long long)req->sector,
++ req->current_nr_sectors,
++ req->nr_sectors, req->buffer,
++ rq_data_dir(req) ? "write" : "read");
++
++
++ blkdev_dequeue_request(req);
++ if (blkif_queue_request(req)) {
++ blk_requeue_request(rq, req);
++ wait:
++ /* Avoid pointless unplugs. */
++ blk_stop_queue(rq);
++ break;
++ }
++
++ queued++;
++ }
++
++ if (queued != 0)
++ flush_requests(info);
++}
++
++
++static irqreturn_t blkif_int(int irq, void *dev_id, struct pt_regs *ptregs)
++{
++ struct request *req;
++ blkif_response_t *bret;
++ RING_IDX i, rp;
++ unsigned long flags;
++ struct blkfront_info *info = (struct blkfront_info *)dev_id;
++ int uptodate;
++
++ spin_lock_irqsave(&blkif_io_lock, flags);
++
++ if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) {
++ spin_unlock_irqrestore(&blkif_io_lock, flags);
++ return IRQ_HANDLED;
++ }
++
++ again:
++ rp = info->ring.sring->rsp_prod;
++ rmb(); /* Ensure we see queued responses up to 'rp'. */
++
++ for (i = info->ring.rsp_cons; i != rp; i++) {
++ unsigned long id;
++ int ret;
++
++ bret = RING_GET_RESPONSE(&info->ring, i);
++ id = bret->id;
++ req = (struct request *)info->shadow[id].request;
++
++ blkif_completion(&info->shadow[id]);
++
++ ADD_ID_TO_FREELIST(info, id);
++
++ uptodate = (bret->status == BLKIF_RSP_OKAY);
++ switch (bret->operation) {
++ case BLKIF_OP_WRITE_BARRIER:
++ if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
++ printk("blkfront: %s: write barrier op failed\n",
++ info->gd->disk_name);
++ uptodate = -EOPNOTSUPP;
++ info->feature_barrier = 0;
++ xlvbd_barrier(info);
++ }
++ /* fall through */
++ case BLKIF_OP_READ:
++ case BLKIF_OP_WRITE:
++ if (unlikely(bret->status != BLKIF_RSP_OKAY))
++ DPRINTK("Bad return from blkdev data "
++ "request: %x\n", bret->status);
++
++ ret = end_that_request_first(req, uptodate,
++ req->hard_nr_sectors);
++ BUG_ON(ret);
++ end_that_request_last(req, uptodate);
++ break;
++ default:
++ BUG();
++ }
++ }
++
++ info->ring.rsp_cons = i;
++
++ if (i != info->ring.req_prod_pvt) {
++ int more_to_do;
++ RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
++ if (more_to_do)
++ goto again;
++ } else
++ info->ring.sring->rsp_event = i + 1;
++
++ kick_pending_request_queues(info);
++
++ spin_unlock_irqrestore(&blkif_io_lock, flags);
++
++ return IRQ_HANDLED;
++}
++
++static void blkif_free(struct blkfront_info *info, int suspend)
++{
++ /* Prevent new requests being issued until we fix things up. */
++ spin_lock_irq(&blkif_io_lock);
++ info->connected = suspend ?
++ BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
++ /* No more blkif_request(). */
++ if (info->rq)
++ blk_stop_queue(info->rq);
++ /* No more gnttab callback work. */
++ gnttab_cancel_free_callback(&info->callback);
++ spin_unlock_irq(&blkif_io_lock);
++
++ /* Flush gnttab callback work. Must be done with no locks held. */
++ flush_scheduled_work();
++
++ /* Free resources associated with old device channel. */
++ if (info->ring_ref != GRANT_INVALID_REF) {
++ gnttab_end_foreign_access(info->ring_ref,
++ (unsigned long)info->ring.sring);
++ info->ring_ref = GRANT_INVALID_REF;
++ info->ring.sring = NULL;
++ }
++ if (info->irq)
++ unbind_from_irqhandler(info->irq, info);
++ info->irq = 0;
++}
++
++static void blkif_completion(struct blk_shadow *s)
++{
++ int i;
++ for (i = 0; i < s->req.nr_segments; i++)
++ gnttab_end_foreign_access(s->req.seg[i].gref, 0UL);
++}
++
++static void blkif_recover(struct blkfront_info *info)
++{
++ int i;
++ blkif_request_t *req;
++ struct blk_shadow *copy;
++ int j;
++
++ /* Stage 1: Make a safe copy of the shadow state. */
++ copy = kmalloc(sizeof(info->shadow), GFP_KERNEL | __GFP_NOFAIL | __GFP_HIGH);
++ memcpy(copy, info->shadow, sizeof(info->shadow));
++
++ /* Stage 2: Set up free list. */
++ memset(&info->shadow, 0, sizeof(info->shadow));
++ for (i = 0; i < BLK_RING_SIZE; i++)
++ info->shadow[i].req.id = i+1;
++ info->shadow_free = info->ring.req_prod_pvt;
++ info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
++
++ /* Stage 3: Find pending requests and requeue them. */
++ for (i = 0; i < BLK_RING_SIZE; i++) {
++ /* Not in use? */
++ if (copy[i].request == 0)
++ continue;
++
++ /* Grab a request slot and copy shadow state into it. */
++ req = RING_GET_REQUEST(
++ &info->ring, info->ring.req_prod_pvt);
++ *req = copy[i].req;
++
++ /* We get a new request id, and must reset the shadow state. */
++ req->id = GET_ID_FROM_FREELIST(info);
++ memcpy(&info->shadow[req->id], ©[i], sizeof(copy[i]));
++
++ /* Rewrite any grant references invalidated by susp/resume. */
++ for (j = 0; j < req->nr_segments; j++)
++ gnttab_grant_foreign_access_ref(
++ req->seg[j].gref,
++ info->xbdev->otherend_id,
++ pfn_to_mfn(info->shadow[req->id].frame[j]),
++ rq_data_dir((struct request *)
++ info->shadow[req->id].request) ?
++ GTF_readonly : 0);
++ info->shadow[req->id].req = *req;
++
++ info->ring.req_prod_pvt++;
++ }
++
++ kfree(copy);
++
++ (void)xenbus_switch_state(info->xbdev, XenbusStateConnected);
++
++ spin_lock_irq(&blkif_io_lock);
++
++ /* Now safe for us to use the shared ring */
++ info->connected = BLKIF_STATE_CONNECTED;
++
++ /* Send off requeued requests */
++ flush_requests(info);
++
++ /* Kick any other new requests queued since we resumed */
++ kick_pending_request_queues(info);
++
++ spin_unlock_irq(&blkif_io_lock);
++}
++
++int blkfront_is_ready(struct xenbus_device *dev)
++{
++ struct blkfront_info *info = dev->dev.driver_data;
++
++ return info->is_ready;
++}
++
++
++/* ** Driver Registration ** */
++
++
++static const struct xenbus_device_id blkfront_ids[] = {
++ { "vbd" },
++ { "" }
++};
++MODULE_ALIAS("xen:vbd");
++
++static struct xenbus_driver blkfront = {
++ .name = "vbd",
++ .owner = THIS_MODULE,
++ .ids = blkfront_ids,
++ .probe = blkfront_probe,
++ .remove = blkfront_remove,
++ .resume = blkfront_resume,
++ .otherend_changed = backend_changed,
++ .is_ready = blkfront_is_ready,
++};
++
++
++static int __init xlblk_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ return xenbus_register_frontend(&blkfront);
++}
++module_init(xlblk_init);
++
++
++static void __exit xlblk_exit(void)
++{
++ return xenbus_unregister_driver(&blkfront);
++}
++module_exit(xlblk_exit);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/blkfront/block.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkfront/block.h 2008-05-19 00:33:45.686779559 +0300
+@@ -0,0 +1,143 @@
++/******************************************************************************
++ * block.h
++ *
++ * Shared definitions between all levels of XenLinux Virtual block devices.
++ *
++ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
++ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
++ * Copyright (c) 2004-2005, Christian Limpach
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef __XEN_DRIVERS_BLOCK_H__
++#define __XEN_DRIVERS_BLOCK_H__
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/slab.h>
++#include <linux/string.h>
++#include <linux/errno.h>
++#include <linux/fs.h>
++#include <linux/hdreg.h>
++#include <linux/blkdev.h>
++#include <linux/major.h>
++#include <asm/hypervisor.h>
++#include <xen/xenbus.h>
++#include <xen/gnttab.h>
++#include <xen/interface/xen.h>
++#include <xen/interface/io/blkif.h>
++#include <xen/interface/io/ring.h>
++#include <asm/io.h>
++#include <asm/atomic.h>
++#include <asm/uaccess.h>
++
++#define DPRINTK(_f, _a...) pr_debug(_f, ## _a)
++
++#if 0
++#define DPRINTK_IOCTL(_f, _a...) printk(KERN_ALERT _f, ## _a)
++#else
++#define DPRINTK_IOCTL(_f, _a...) ((void)0)
++#endif
++
++struct xlbd_type_info
++{
++ int partn_shift;
++ int disks_per_major;
++ char *devname;
++ char *diskname;
++};
++
++struct xlbd_major_info
++{
++ int major;
++ int index;
++ int usage;
++ struct xlbd_type_info *type;
++};
++
++struct blk_shadow {
++ blkif_request_t req;
++ unsigned long request;
++ unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST];
++};
++
++#define BLK_RING_SIZE __RING_SIZE((blkif_sring_t *)0, PAGE_SIZE)
++
++/*
++ * We have one of these per vbd, whether ide, scsi or 'other'. They
++ * hang in private_data off the gendisk structure. We may end up
++ * putting all kinds of interesting stuff here :-)
++ */
++struct blkfront_info
++{
++ struct xenbus_device *xbdev;
++ dev_t dev;
++ struct gendisk *gd;
++ int vdevice;
++ blkif_vdev_t handle;
++ int connected;
++ int ring_ref;
++ blkif_front_ring_t ring;
++ unsigned int irq;
++ struct xlbd_major_info *mi;
++ request_queue_t *rq;
++ struct work_struct work;
++ struct gnttab_free_callback callback;
++ struct blk_shadow shadow[BLK_RING_SIZE];
++ unsigned long shadow_free;
++ int feature_barrier;
++ int is_ready;
++
++ /**
++ * The number of people holding this device open. We won't allow a
++ * hot-unplug unless this is 0.
++ */
++ int users;
++};
++
++extern spinlock_t blkif_io_lock;
++
++extern int blkif_open(struct inode *inode, struct file *filep);
++extern int blkif_release(struct inode *inode, struct file *filep);
++extern int blkif_ioctl(struct inode *inode, struct file *filep,
++ unsigned command, unsigned long argument);
++extern int blkif_getgeo(struct block_device *, struct hd_geometry *);
++extern int blkif_check(dev_t dev);
++extern int blkif_revalidate(dev_t dev);
++extern void do_blkif_request (request_queue_t *rq);
++
++/* Virtual block-device subsystem. */
++/* Note that xlvbd_add doesn't call add_disk for you: you're expected
++ to call add_disk on info->gd once the disk is properly connected
++ up. */
++int xlvbd_add(blkif_sector_t capacity, int device,
++ u16 vdisk_info, u16 sector_size, struct blkfront_info *info);
++void xlvbd_del(struct blkfront_info *info);
++int xlvbd_barrier(struct blkfront_info *info);
++
++#endif /* __XEN_DRIVERS_BLOCK_H__ */
+--- linux-2.6.18.8/drivers/xen/blkfront/vbd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blkfront/vbd.c 2008-05-19 00:33:45.686779559 +0300
+@@ -0,0 +1,375 @@
++/******************************************************************************
++ * vbd.c
++ *
++ * XenLinux virtual block-device driver (xvd).
++ *
++ * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
++ * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
++ * Copyright (c) 2004-2005, Christian Limpach
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include "block.h"
++#include <linux/blkdev.h>
++#include <linux/list.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++#define BLKIF_MAJOR(dev) ((dev)>>8)
++#define BLKIF_MINOR(dev) ((dev) & 0xff)
++
++/*
++ * For convenience we distinguish between ide, scsi and 'other' (i.e.,
++ * potentially combinations of the two) in the naming scheme and in a few other
++ * places.
++ */
++
++#define NUM_IDE_MAJORS 10
++#define NUM_SCSI_MAJORS 17
++#define NUM_VBD_MAJORS 1
++
++static struct xlbd_type_info xlbd_ide_type = {
++ .partn_shift = 6,
++ .disks_per_major = 2,
++ .devname = "ide",
++ .diskname = "hd",
++};
++
++static struct xlbd_type_info xlbd_scsi_type = {
++ .partn_shift = 4,
++ .disks_per_major = 16,
++ .devname = "sd",
++ .diskname = "sd",
++};
++
++static struct xlbd_type_info xlbd_vbd_type = {
++ .partn_shift = 4,
++ .disks_per_major = 16,
++ .devname = "xvd",
++ .diskname = "xvd",
++};
++
++static struct xlbd_major_info *major_info[NUM_IDE_MAJORS + NUM_SCSI_MAJORS +
++ NUM_VBD_MAJORS];
++
++#define XLBD_MAJOR_IDE_START 0
++#define XLBD_MAJOR_SCSI_START (NUM_IDE_MAJORS)
++#define XLBD_MAJOR_VBD_START (NUM_IDE_MAJORS + NUM_SCSI_MAJORS)
++
++#define XLBD_MAJOR_IDE_RANGE XLBD_MAJOR_IDE_START ... XLBD_MAJOR_SCSI_START - 1
++#define XLBD_MAJOR_SCSI_RANGE XLBD_MAJOR_SCSI_START ... XLBD_MAJOR_VBD_START - 1
++#define XLBD_MAJOR_VBD_RANGE XLBD_MAJOR_VBD_START ... XLBD_MAJOR_VBD_START + NUM_VBD_MAJORS - 1
++
++/* Information about our VBDs. */
++#define MAX_VBDS 64
++static LIST_HEAD(vbds_list);
++
++static struct block_device_operations xlvbd_block_fops =
++{
++ .owner = THIS_MODULE,
++ .open = blkif_open,
++ .release = blkif_release,
++ .ioctl = blkif_ioctl,
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
++ .getgeo = blkif_getgeo
++#endif
++};
++
++DEFINE_SPINLOCK(blkif_io_lock);
++
++static struct xlbd_major_info *
++xlbd_alloc_major_info(int major, int minor, int index)
++{
++ struct xlbd_major_info *ptr;
++
++ ptr = kzalloc(sizeof(struct xlbd_major_info), GFP_KERNEL);
++ if (ptr == NULL)
++ return NULL;
++
++ ptr->major = major;
++
++ switch (index) {
++ case XLBD_MAJOR_IDE_RANGE:
++ ptr->type = &xlbd_ide_type;
++ ptr->index = index - XLBD_MAJOR_IDE_START;
++ break;
++ case XLBD_MAJOR_SCSI_RANGE:
++ ptr->type = &xlbd_scsi_type;
++ ptr->index = index - XLBD_MAJOR_SCSI_START;
++ break;
++ case XLBD_MAJOR_VBD_RANGE:
++ ptr->type = &xlbd_vbd_type;
++ ptr->index = index - XLBD_MAJOR_VBD_START;
++ break;
++ }
++
++ if (register_blkdev(ptr->major, ptr->type->devname)) {
++ kfree(ptr);
++ return NULL;
++ }
++
++ printk("xen-vbd: registered block device major %i\n", ptr->major);
++ major_info[index] = ptr;
++ return ptr;
++}
++
++static struct xlbd_major_info *
++xlbd_get_major_info(int vdevice)
++{
++ struct xlbd_major_info *mi;
++ int major, minor, index;
++
++ major = BLKIF_MAJOR(vdevice);
++ minor = BLKIF_MINOR(vdevice);
++
++ switch (major) {
++ case IDE0_MAJOR: index = 0; break;
++ case IDE1_MAJOR: index = 1; break;
++ case IDE2_MAJOR: index = 2; break;
++ case IDE3_MAJOR: index = 3; break;
++ case IDE4_MAJOR: index = 4; break;
++ case IDE5_MAJOR: index = 5; break;
++ case IDE6_MAJOR: index = 6; break;
++ case IDE7_MAJOR: index = 7; break;
++ case IDE8_MAJOR: index = 8; break;
++ case IDE9_MAJOR: index = 9; break;
++ case SCSI_DISK0_MAJOR: index = 10; break;
++ case SCSI_DISK1_MAJOR ... SCSI_DISK7_MAJOR:
++ index = 11 + major - SCSI_DISK1_MAJOR;
++ break;
++ case SCSI_DISK8_MAJOR ... SCSI_DISK15_MAJOR:
++ index = 18 + major - SCSI_DISK8_MAJOR;
++ break;
++ case SCSI_CDROM_MAJOR: index = 26; break;
++ default: index = 27; break;
++ }
++
++ mi = ((major_info[index] != NULL) ? major_info[index] :
++ xlbd_alloc_major_info(major, minor, index));
++ if (mi)
++ mi->usage++;
++ return mi;
++}
++
++static void
++xlbd_put_major_info(struct xlbd_major_info *mi)
++{
++ mi->usage--;
++ /* XXX: release major if 0 */
++}
++
++static int
++xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
++{
++ request_queue_t *rq;
++
++ rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
++ if (rq == NULL)
++ return -1;
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,10)
++ elevator_init(rq, "noop");
++#else
++ elevator_init(rq, &elevator_noop);
++#endif
++
++ /* Hard sector size and max sectors impersonate the equiv. hardware. */
++ blk_queue_hardsect_size(rq, sector_size);
++ blk_queue_max_sectors(rq, 512);
++
++ /* Each segment in a request is up to an aligned page in size. */
++ blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
++ blk_queue_max_segment_size(rq, PAGE_SIZE);
++
++ /* Ensure a merged request will fit in a single I/O ring slot. */
++ blk_queue_max_phys_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
++ blk_queue_max_hw_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
++
++ /* Make sure buffer addresses are sector-aligned. */
++ blk_queue_dma_alignment(rq, 511);
++
++ /* Make sure we don't use bounce buffers. */
++ blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY);
++
++ gd->queue = rq;
++
++ return 0;
++}
++
++static int
++xlvbd_alloc_gendisk(int minor, blkif_sector_t capacity, int vdevice,
++ u16 vdisk_info, u16 sector_size,
++ struct blkfront_info *info)
++{
++ struct gendisk *gd;
++ struct xlbd_major_info *mi;
++ int nr_minors = 1;
++ int err = -ENODEV;
++ unsigned int offset;
++
++ BUG_ON(info->gd != NULL);
++ BUG_ON(info->mi != NULL);
++ BUG_ON(info->rq != NULL);
++
++ mi = xlbd_get_major_info(vdevice);
++ if (mi == NULL)
++ goto out;
++ info->mi = mi;
++
++ if ((minor & ((1 << mi->type->partn_shift) - 1)) == 0)
++ nr_minors = 1 << mi->type->partn_shift;
++
++ gd = alloc_disk(nr_minors);
++ if (gd == NULL)
++ goto out;
++
++ offset = mi->index * mi->type->disks_per_major +
++ (minor >> mi->type->partn_shift);
++ if (nr_minors > 1) {
++ if (offset < 26) {
++ sprintf(gd->disk_name, "%s%c",
++ mi->type->diskname, 'a' + offset );
++ }
++ else {
++ sprintf(gd->disk_name, "%s%c%c",
++ mi->type->diskname,
++ 'a' + ((offset/26)-1), 'a' + (offset%26) );
++ }
++ }
++ else {
++ if (offset < 26) {
++ sprintf(gd->disk_name, "%s%c%d",
++ mi->type->diskname,
++ 'a' + offset,
++ minor & ((1 << mi->type->partn_shift) - 1));
++ }
++ else {
++ sprintf(gd->disk_name, "%s%c%c%d",
++ mi->type->diskname,
++ 'a' + ((offset/26)-1), 'a' + (offset%26),
++ minor & ((1 << mi->type->partn_shift) - 1));
++ }
++ }
++
++ gd->major = mi->major;
++ gd->first_minor = minor;
++ gd->fops = &xlvbd_block_fops;
++ gd->private_data = info;
++ gd->driverfs_dev = &(info->xbdev->dev);
++ set_capacity(gd, capacity);
++
++ if (xlvbd_init_blk_queue(gd, sector_size)) {
++ del_gendisk(gd);
++ goto out;
++ }
++
++ info->rq = gd->queue;
++ info->gd = gd;
++
++ if (info->feature_barrier)
++ xlvbd_barrier(info);
++
++ if (vdisk_info & VDISK_READONLY)
++ set_disk_ro(gd, 1);
++
++ if (vdisk_info & VDISK_REMOVABLE)
++ gd->flags |= GENHD_FL_REMOVABLE;
++
++ if (vdisk_info & VDISK_CDROM)
++ gd->flags |= GENHD_FL_CD;
++
++ return 0;
++
++ out:
++ if (mi)
++ xlbd_put_major_info(mi);
++ info->mi = NULL;
++ return err;
++}
++
++int
++xlvbd_add(blkif_sector_t capacity, int vdevice, u16 vdisk_info,
++ u16 sector_size, struct blkfront_info *info)
++{
++ struct block_device *bd;
++ int err = 0;
++
++ info->dev = MKDEV(BLKIF_MAJOR(vdevice), BLKIF_MINOR(vdevice));
++
++ bd = bdget(info->dev);
++ if (bd == NULL)
++ return -ENODEV;
++
++ err = xlvbd_alloc_gendisk(BLKIF_MINOR(vdevice), capacity, vdevice,
++ vdisk_info, sector_size, info);
++
++ bdput(bd);
++ return err;
++}
++
++void
++xlvbd_del(struct blkfront_info *info)
++{
++ if (info->mi == NULL)
++ return;
++
++ BUG_ON(info->gd == NULL);
++ del_gendisk(info->gd);
++ put_disk(info->gd);
++ info->gd = NULL;
++
++ xlbd_put_major_info(info->mi);
++ info->mi = NULL;
++
++ BUG_ON(info->rq == NULL);
++ blk_cleanup_queue(info->rq);
++ info->rq = NULL;
++}
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
++int
++xlvbd_barrier(struct blkfront_info *info)
++{
++ int err;
++
++ err = blk_queue_ordered(info->rq,
++ info->feature_barrier ? QUEUE_ORDERED_DRAIN : QUEUE_ORDERED_NONE, NULL);
++ if (err)
++ return err;
++ printk(KERN_INFO "blkfront: %s: barriers %s\n",
++ info->gd->disk_name, info->feature_barrier ? "enabled" : "disabled");
++ return 0;
++}
++#else
++int
++xlvbd_barrier(struct blkfront_info *info)
++{
++ printk(KERN_INFO "blkfront: %s: barriers disabled\n", info->gd->disk_name);
++ return -ENOSYS;
++}
++#endif
+--- linux-2.6.18.8/drivers/xen/blktap/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blktap/Makefile 2008-05-19 00:33:45.698780251 +0300
+@@ -0,0 +1,5 @@
++LINUXINCLUDE += -I../xen/include/public/io
++
++obj-$(CONFIG_XEN_BLKDEV_TAP) := xenblktap.o
++
++xenblktap-y := xenbus.o interface.o blktap.o
+--- linux-2.6.18.8/drivers/xen/blktap/blktap.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blktap/blktap.c 2008-05-19 00:33:45.698780251 +0300
+@@ -0,0 +1,1648 @@
++/******************************************************************************
++ * drivers/xen/blktap/blktap.c
++ *
++ * Back-end driver for user level virtual block devices. This portion of the
++ * driver exports a 'unified' block-device interface that can be accessed
++ * by any operating system that implements a compatible front end. Requests
++ * are remapped to a user-space memory region.
++ *
++ * Based on the blkback driver code.
++ *
++ * Copyright (c) 2004-2005, Andrew Warfield and Julian Chesterfield
++ *
++ * Clean ups and fix ups:
++ * Copyright (c) 2006, Steven Rostedt - Red Hat, Inc.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/spinlock.h>
++#include <linux/kthread.h>
++#include <linux/list.h>
++#include <asm/hypervisor.h>
++#include "common.h"
++#include <xen/balloon.h>
++#include <xen/driver_util.h>
++#include <linux/kernel.h>
++#include <linux/fs.h>
++#include <linux/mm.h>
++#include <linux/errno.h>
++#include <linux/major.h>
++#include <linux/gfp.h>
++#include <linux/poll.h>
++#include <linux/delay.h>
++#include <asm/tlbflush.h>
++
++#define MAX_TAP_DEV 256 /*the maximum number of tapdisk ring devices */
++#define MAX_DEV_NAME 100 /*the max tapdisk ring device name e.g. blktap0 */
++
++/*
++ * The maximum number of requests that can be outstanding at any time
++ * is determined by
++ *
++ * [mmap_alloc * MAX_PENDING_REQS * BLKIF_MAX_SEGMENTS_PER_REQUEST]
++ *
++ * where mmap_alloc < MAX_DYNAMIC_MEM.
++ *
++ * TODO:
++ * mmap_alloc is initialised to 2 and should be adjustable on the fly via
++ * sysfs.
++ */
++#define BLK_RING_SIZE __RING_SIZE((blkif_sring_t *)0, PAGE_SIZE)
++#define MAX_DYNAMIC_MEM BLK_RING_SIZE
++#define MAX_PENDING_REQS BLK_RING_SIZE
++#define MMAP_PAGES (MAX_PENDING_REQS * BLKIF_MAX_SEGMENTS_PER_REQUEST)
++#define MMAP_VADDR(_start, _req,_seg) \
++ (_start + \
++ ((_req) * BLKIF_MAX_SEGMENTS_PER_REQUEST * PAGE_SIZE) + \
++ ((_seg) * PAGE_SIZE))
++static int blkif_reqs = MAX_PENDING_REQS;
++static int mmap_pages = MMAP_PAGES;
++
++#define RING_PAGES 1 /* BLKTAP - immediately before the mmap area, we
++ * have a bunch of pages reserved for shared
++ * memory rings.
++ */
++
++/*Data struct handed back to userspace for tapdisk device to VBD mapping*/
++typedef struct domid_translate {
++ unsigned short domid;
++ unsigned short busid;
++} domid_translate_t ;
++
++/*Data struct associated with each of the tapdisk devices*/
++typedef struct tap_blkif {
++ struct vm_area_struct *vma; /*Shared memory area */
++ unsigned long rings_vstart; /*Kernel memory mapping */
++ unsigned long user_vstart; /*User memory mapping */
++ unsigned long dev_inuse; /*One process opens device at a time. */
++ unsigned long dev_pending; /*In process of being opened */
++ unsigned long ring_ok; /*make this ring->state */
++ blkif_front_ring_t ufe_ring; /*Rings up to user space. */
++ wait_queue_head_t wait; /*for poll */
++ unsigned long mode; /*current switching mode */
++ int minor; /*Minor number for tapdisk device */
++ pid_t pid; /*tapdisk process id */
++ enum { RUNNING, CLEANSHUTDOWN } status; /*Detect a clean userspace
++ shutdown */
++ unsigned long *idx_map; /*Record the user ring id to kern
++ [req id, idx] tuple */
++ blkif_t *blkif; /*Associate blkif with tapdev */
++ struct domid_translate trans; /*Translation from domid to bus. */
++} tap_blkif_t;
++
++static struct tap_blkif *tapfds[MAX_TAP_DEV];
++static int blktap_next_minor;
++
++module_param(blkif_reqs, int, 0);
++/* Run-time switchable: /sys/module/blktap/parameters/ */
++static unsigned int log_stats = 0;
++static unsigned int debug_lvl = 0;
++module_param(log_stats, int, 0644);
++module_param(debug_lvl, int, 0644);
++
++/*
++ * Each outstanding request that we've passed to the lower device layers has a
++ * 'pending_req' allocated to it. Each buffer_head that completes decrements
++ * the pendcnt towards zero. When it hits zero, the specified domain has a
++ * response queued for it, with the saved 'id' passed back.
++ */
++typedef struct {
++ blkif_t *blkif;
++ u64 id;
++ unsigned short mem_idx;
++ int nr_pages;
++ atomic_t pendcnt;
++ unsigned short operation;
++ int status;
++ struct list_head free_list;
++ int inuse;
++} pending_req_t;
++
++static pending_req_t *pending_reqs[MAX_PENDING_REQS];
++static struct list_head pending_free;
++static DEFINE_SPINLOCK(pending_free_lock);
++static DECLARE_WAIT_QUEUE_HEAD (pending_free_wq);
++static int alloc_pending_reqs;
++
++typedef unsigned int PEND_RING_IDX;
++
++static inline int MASK_PEND_IDX(int i) {
++ return (i & (MAX_PENDING_REQS-1));
++}
++
++static inline unsigned int RTN_PEND_IDX(pending_req_t *req, int idx) {
++ return (req - pending_reqs[idx]);
++}
++
++#define NR_PENDING_REQS (MAX_PENDING_REQS - pending_prod + pending_cons)
++
++#define BLKBACK_INVALID_HANDLE (~0)
++
++static struct page **foreign_pages[MAX_DYNAMIC_MEM];
++static inline unsigned long idx_to_kaddr(
++ unsigned int mmap_idx, unsigned int req_idx, unsigned int sg_idx)
++{
++ unsigned int arr_idx = req_idx*BLKIF_MAX_SEGMENTS_PER_REQUEST + sg_idx;
++ unsigned long pfn = page_to_pfn(foreign_pages[mmap_idx][arr_idx]);
++ return (unsigned long)pfn_to_kaddr(pfn);
++}
++
++static unsigned short mmap_alloc = 0;
++static unsigned short mmap_lock = 0;
++static unsigned short mmap_inuse = 0;
++
++/******************************************************************
++ * GRANT HANDLES
++ */
++
++/* When using grant tables to map a frame for device access then the
++ * handle returned must be used to unmap the frame. This is needed to
++ * drop the ref count on the frame.
++ */
++struct grant_handle_pair
++{
++ grant_handle_t kernel;
++ grant_handle_t user;
++};
++#define INVALID_GRANT_HANDLE 0xFFFF
++
++static struct grant_handle_pair
++ pending_grant_handles[MAX_DYNAMIC_MEM][MMAP_PAGES];
++#define pending_handle(_id, _idx, _i) \
++ (pending_grant_handles[_id][((_idx) * BLKIF_MAX_SEGMENTS_PER_REQUEST) \
++ + (_i)])
++
++
++static int blktap_read_ufe_ring(tap_blkif_t *info); /*local prototypes*/
++
++#define BLKTAP_MINOR 0 /*/dev/xen/blktap has a dynamic major */
++#define BLKTAP_DEV_DIR "/dev/xen"
++
++static int blktap_major;
++
++/* blktap IOCTLs: */
++#define BLKTAP_IOCTL_KICK_FE 1
++#define BLKTAP_IOCTL_KICK_BE 2 /* currently unused */
++#define BLKTAP_IOCTL_SETMODE 3
++#define BLKTAP_IOCTL_SENDPID 4
++#define BLKTAP_IOCTL_NEWINTF 5
++#define BLKTAP_IOCTL_MINOR 6
++#define BLKTAP_IOCTL_MAJOR 7
++#define BLKTAP_QUERY_ALLOC_REQS 8
++#define BLKTAP_IOCTL_FREEINTF 9
++#define BLKTAP_IOCTL_PRINT_IDXS 100
++
++/* blktap switching modes: (Set with BLKTAP_IOCTL_SETMODE) */
++#define BLKTAP_MODE_PASSTHROUGH 0x00000000 /* default */
++#define BLKTAP_MODE_INTERCEPT_FE 0x00000001
++#define BLKTAP_MODE_INTERCEPT_BE 0x00000002 /* unimp. */
++
++#define BLKTAP_MODE_INTERPOSE \
++ (BLKTAP_MODE_INTERCEPT_FE | BLKTAP_MODE_INTERCEPT_BE)
++
++
++static inline int BLKTAP_MODE_VALID(unsigned long arg)
++{
++ return ((arg == BLKTAP_MODE_PASSTHROUGH ) ||
++ (arg == BLKTAP_MODE_INTERCEPT_FE) ||
++ (arg == BLKTAP_MODE_INTERPOSE ));
++}
++
++/* Requests passing through the tap to userspace are re-assigned an ID.
++ * We must record a mapping between the BE [IDX,ID] tuple and the userspace
++ * ring ID.
++ */
++
++static inline unsigned long MAKE_ID(domid_t fe_dom, PEND_RING_IDX idx)
++{
++ return ((fe_dom << 16) | MASK_PEND_IDX(idx));
++}
++
++extern inline PEND_RING_IDX ID_TO_IDX(unsigned long id)
++{
++ return (PEND_RING_IDX)(id & 0x0000ffff);
++}
++
++extern inline int ID_TO_MIDX(unsigned long id)
++{
++ return (int)(id >> 16);
++}
++
++#define INVALID_REQ 0xdead0000
++
++/*TODO: Convert to a free list*/
++static inline int GET_NEXT_REQ(unsigned long *idx_map)
++{
++ int i;
++ for (i = 0; i < MAX_PENDING_REQS; i++)
++ if (idx_map[i] == INVALID_REQ)
++ return i;
++
++ return INVALID_REQ;
++}
++
++static inline int OFFSET_TO_USR_IDX(int offset)
++{
++ return offset / BLKIF_MAX_SEGMENTS_PER_REQUEST;
++}
++
++static inline int OFFSET_TO_SEG(int offset)
++{
++ return offset % BLKIF_MAX_SEGMENTS_PER_REQUEST;
++}
++
++
++#define BLKTAP_INVALID_HANDLE(_g) \
++ (((_g->kernel) == INVALID_GRANT_HANDLE) && \
++ ((_g->user) == INVALID_GRANT_HANDLE))
++
++#define BLKTAP_INVALIDATE_HANDLE(_g) do { \
++ (_g)->kernel = INVALID_GRANT_HANDLE; (_g)->user = INVALID_GRANT_HANDLE; \
++ } while(0)
++
++
++/******************************************************************
++ * BLKTAP VM OPS
++ */
++
++static struct page *blktap_nopage(struct vm_area_struct *vma,
++ unsigned long address,
++ int *type)
++{
++ /*
++ * if the page has not been mapped in by the driver then return
++ * NOPAGE_SIGBUS to the domain.
++ */
++
++ return NOPAGE_SIGBUS;
++}
++
++static pte_t blktap_clear_pte(struct vm_area_struct *vma,
++ unsigned long uvaddr,
++ pte_t *ptep, int is_fullmm)
++{
++ pte_t copy;
++ tap_blkif_t *info;
++ int offset, seg, usr_idx, pending_idx, mmap_idx;
++ unsigned long uvstart = vma->vm_start + (RING_PAGES << PAGE_SHIFT);
++ unsigned long kvaddr;
++ struct page **map;
++ struct page *pg;
++ struct grant_handle_pair *khandle;
++ struct gnttab_unmap_grant_ref unmap[2];
++ int count = 0;
++
++ /*
++ * If the address is before the start of the grant mapped region or
++ * if vm_file is NULL (meaning mmap failed and we have nothing to do)
++ */
++ if (uvaddr < uvstart || vma->vm_file == NULL)
++ return ptep_get_and_clear_full(vma->vm_mm, uvaddr,
++ ptep, is_fullmm);
++
++ info = vma->vm_file->private_data;
++ map = vma->vm_private_data;
++
++ /* TODO Should these be changed to if statements? */
++ BUG_ON(!info);
++ BUG_ON(!info->idx_map);
++ BUG_ON(!map);
++
++ offset = (int) ((uvaddr - uvstart) >> PAGE_SHIFT);
++ usr_idx = OFFSET_TO_USR_IDX(offset);
++ seg = OFFSET_TO_SEG(offset);
++
++ pending_idx = MASK_PEND_IDX(ID_TO_IDX(info->idx_map[usr_idx]));
++ mmap_idx = ID_TO_MIDX(info->idx_map[usr_idx]);
++
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, seg);
++ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
++ ClearPageReserved(pg);
++ map[offset + RING_PAGES] = NULL;
++
++ khandle = &pending_handle(mmap_idx, pending_idx, seg);
++
++ if (khandle->kernel != INVALID_GRANT_HANDLE) {
++ gnttab_set_unmap_op(&unmap[count], kvaddr,
++ GNTMAP_host_map, khandle->kernel);
++ count++;
++
++ set_phys_to_machine(__pa(kvaddr) >> PAGE_SHIFT,
++ INVALID_P2M_ENTRY);
++ }
++
++ if (khandle->user != INVALID_GRANT_HANDLE) {
++ BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
++
++ copy = *ptep;
++ gnttab_set_unmap_op(&unmap[count], virt_to_machine(ptep),
++ GNTMAP_host_map
++ | GNTMAP_application_map
++ | GNTMAP_contains_pte,
++ khandle->user);
++ count++;
++ } else {
++ BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap));
++
++ /* USING SHADOW PAGE TABLES. */
++ copy = ptep_get_and_clear_full(vma->vm_mm, uvaddr, ptep,
++ is_fullmm);
++ }
++
++ if (count) {
++ BLKTAP_INVALIDATE_HANDLE(khandle);
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref,
++ unmap, count))
++ BUG();
++ }
++
++ return copy;
++}
++
++struct vm_operations_struct blktap_vm_ops = {
++ nopage: blktap_nopage,
++ zap_pte: blktap_clear_pte,
++};
++
++/******************************************************************
++ * BLKTAP FILE OPS
++ */
++
++/*Function Declarations*/
++static tap_blkif_t *get_next_free_dev(void);
++static int blktap_open(struct inode *inode, struct file *filp);
++static int blktap_release(struct inode *inode, struct file *filp);
++static int blktap_mmap(struct file *filp, struct vm_area_struct *vma);
++static int blktap_ioctl(struct inode *inode, struct file *filp,
++ unsigned int cmd, unsigned long arg);
++static unsigned int blktap_poll(struct file *file, poll_table *wait);
++
++static const struct file_operations blktap_fops = {
++ .owner = THIS_MODULE,
++ .poll = blktap_poll,
++ .ioctl = blktap_ioctl,
++ .open = blktap_open,
++ .release = blktap_release,
++ .mmap = blktap_mmap,
++};
++
++
++static tap_blkif_t *get_next_free_dev(void)
++{
++ struct class *class;
++ tap_blkif_t *info;
++ int minor;
++
++ /*
++ * This is called only from the ioctl, which
++ * means we should always have interrupts enabled.
++ */
++ BUG_ON(irqs_disabled());
++
++ spin_lock_irq(&pending_free_lock);
++
++ /* tapfds[0] is always NULL */
++
++ for (minor = 1; minor < blktap_next_minor; minor++) {
++ info = tapfds[minor];
++ /* we could have failed a previous attempt. */
++ if (!info ||
++ ((info->dev_inuse == 0) &&
++ (info->dev_pending == 0)) ) {
++ info->dev_pending = 1;
++ goto found;
++ }
++ }
++ info = NULL;
++ minor = -1;
++
++ /*
++ * We didn't find free device. If we can still allocate
++ * more, then we grab the next device minor that is
++ * available. This is done while we are still under
++ * the protection of the pending_free_lock.
++ */
++ if (blktap_next_minor < MAX_TAP_DEV)
++ minor = blktap_next_minor++;
++found:
++ spin_unlock_irq(&pending_free_lock);
++
++ if (!info && minor > 0) {
++ info = kzalloc(sizeof(*info), GFP_KERNEL);
++ if (unlikely(!info)) {
++ /*
++ * If we failed here, try to put back
++ * the next minor number. But if one
++ * was just taken, then we just lose this
++ * minor. We can try to allocate this
++ * minor again later.
++ */
++ spin_lock_irq(&pending_free_lock);
++ if (blktap_next_minor == minor+1)
++ blktap_next_minor--;
++ spin_unlock_irq(&pending_free_lock);
++ goto out;
++ }
++
++ info->minor = minor;
++ /*
++ * Make sure that we have a minor before others can
++ * see us.
++ */
++ wmb();
++ tapfds[minor] = info;
++
++ if ((class = get_xen_class()) != NULL)
++ class_device_create(class, NULL,
++ MKDEV(blktap_major, minor), NULL,
++ "blktap%d", minor);
++ }
++
++out:
++ return info;
++}
++
++int dom_to_devid(domid_t domid, int xenbus_id, blkif_t *blkif)
++{
++ tap_blkif_t *info;
++ int i;
++
++ for (i = 1; i < blktap_next_minor; i++) {
++ info = tapfds[i];
++ if ( info &&
++ (info->trans.domid == domid) &&
++ (info->trans.busid == xenbus_id) ) {
++ info->blkif = blkif;
++ info->status = RUNNING;
++ return i;
++ }
++ }
++ return -1;
++}
++
++void signal_tapdisk(int idx)
++{
++ tap_blkif_t *info;
++ struct task_struct *ptask;
++
++ info = tapfds[idx];
++ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info)
++ return;
++
++ if (info->pid > 0) {
++ ptask = find_task_by_pid(info->pid);
++ if (ptask)
++ info->status = CLEANSHUTDOWN;
++ }
++ info->blkif = NULL;
++
++ return;
++}
++
++static int blktap_open(struct inode *inode, struct file *filp)
++{
++ blkif_sring_t *sring;
++ int idx = iminor(inode) - BLKTAP_MINOR;
++ tap_blkif_t *info;
++ int i;
++
++ /* ctrl device, treat differently */
++ if (!idx)
++ return 0;
++
++ info = tapfds[idx];
++
++ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info) {
++ WPRINTK("Unable to open device /dev/xen/blktap%d\n",
++ idx);
++ return -ENODEV;
++ }
++
++ DPRINTK("Opening device /dev/xen/blktap%d\n",idx);
++
++ /*Only one process can access device at a time*/
++ if (test_and_set_bit(0, &info->dev_inuse))
++ return -EBUSY;
++
++ info->dev_pending = 0;
++
++ /* Allocate the fe ring. */
++ sring = (blkif_sring_t *)get_zeroed_page(GFP_KERNEL);
++ if (sring == NULL)
++ goto fail_nomem;
++
++ SetPageReserved(virt_to_page(sring));
++
++ SHARED_RING_INIT(sring);
++ FRONT_RING_INIT(&info->ufe_ring, sring, PAGE_SIZE);
++
++ filp->private_data = info;
++ info->vma = NULL;
++
++ info->idx_map = kmalloc(sizeof(unsigned long) * MAX_PENDING_REQS,
++ GFP_KERNEL);
++
++ if (info->idx_map == NULL)
++ goto fail_nomem;
++
++ if (idx > 0) {
++ init_waitqueue_head(&info->wait);
++ for (i = 0; i < MAX_PENDING_REQS; i++)
++ info->idx_map[i] = INVALID_REQ;
++ }
++
++ DPRINTK("Tap open: device /dev/xen/blktap%d\n",idx);
++ return 0;
++
++ fail_nomem:
++ return -ENOMEM;
++}
++
++static int blktap_release(struct inode *inode, struct file *filp)
++{
++ tap_blkif_t *info = filp->private_data;
++
++ /* check for control device */
++ if (!info)
++ return 0;
++
++ info->dev_inuse = 0;
++ DPRINTK("Freeing device [/dev/xen/blktap%d]\n",info->minor);
++
++ /* Free the ring page. */
++ ClearPageReserved(virt_to_page(info->ufe_ring.sring));
++ free_page((unsigned long) info->ufe_ring.sring);
++
++ /* Clear any active mappings and free foreign map table */
++ if (info->vma) {
++ zap_page_range(
++ info->vma, info->vma->vm_start,
++ info->vma->vm_end - info->vma->vm_start, NULL);
++
++ kfree(info->vma->vm_private_data);
++
++ info->vma = NULL;
++ }
++
++ if (info->idx_map) {
++ kfree(info->idx_map);
++ info->idx_map = NULL;
++ }
++
++ if ( (info->status != CLEANSHUTDOWN) && (info->blkif != NULL) ) {
++ if (info->blkif->xenblkd != NULL) {
++ kthread_stop(info->blkif->xenblkd);
++ info->blkif->xenblkd = NULL;
++ }
++ info->status = CLEANSHUTDOWN;
++ }
++
++ return 0;
++}
++
++
++/* Note on mmap:
++ * We need to map pages to user space in a way that will allow the block
++ * subsystem set up direct IO to them. This couldn't be done before, because
++ * there isn't really a sane way to translate a user virtual address down to a
++ * physical address when the page belongs to another domain.
++ *
++ * My first approach was to map the page in to kernel memory, add an entry
++ * for it in the physical frame list (using alloc_lomem_region as in blkback)
++ * and then attempt to map that page up to user space. This is disallowed
++ * by xen though, which realizes that we don't really own the machine frame
++ * underlying the physical page.
++ *
++ * The new approach is to provide explicit support for this in xen linux.
++ * The VMA now has a flag, VM_FOREIGN, to indicate that it contains pages
++ * mapped from other vms. vma->vm_private_data is set up as a mapping
++ * from pages to actual page structs. There is a new clause in get_user_pages
++ * that does the right thing for this sort of mapping.
++ */
++static int blktap_mmap(struct file *filp, struct vm_area_struct *vma)
++{
++ int size;
++ struct page **map;
++ int i;
++ tap_blkif_t *info = filp->private_data;
++ int ret;
++
++ if (info == NULL) {
++ WPRINTK("blktap: mmap, retrieving idx failed\n");
++ return -ENOMEM;
++ }
++
++ vma->vm_flags |= VM_RESERVED;
++ vma->vm_ops = &blktap_vm_ops;
++
++ size = vma->vm_end - vma->vm_start;
++ if (size != ((mmap_pages + RING_PAGES) << PAGE_SHIFT)) {
++ WPRINTK("you _must_ map exactly %d pages!\n",
++ mmap_pages + RING_PAGES);
++ return -EAGAIN;
++ }
++
++ size >>= PAGE_SHIFT;
++ info->rings_vstart = vma->vm_start;
++ info->user_vstart = info->rings_vstart + (RING_PAGES << PAGE_SHIFT);
++
++ /* Map the ring pages to the start of the region and reserve it. */
++ if (xen_feature(XENFEAT_auto_translated_physmap))
++ ret = vm_insert_page(vma, vma->vm_start,
++ virt_to_page(info->ufe_ring.sring));
++ else
++ ret = remap_pfn_range(vma, vma->vm_start,
++ __pa(info->ufe_ring.sring) >> PAGE_SHIFT,
++ PAGE_SIZE, vma->vm_page_prot);
++ if (ret) {
++ WPRINTK("Mapping user ring failed!\n");
++ goto fail;
++ }
++
++ /* Mark this VM as containing foreign pages, and set up mappings. */
++ map = kzalloc(((vma->vm_end - vma->vm_start) >> PAGE_SHIFT)
++ * sizeof(struct page *),
++ GFP_KERNEL);
++ if (map == NULL) {
++ WPRINTK("Couldn't alloc VM_FOREIGN map.\n");
++ goto fail;
++ }
++
++ for (i = 0; i < ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT); i++)
++ map[i] = NULL;
++
++ vma->vm_private_data = map;
++ vma->vm_flags |= VM_FOREIGN;
++ vma->vm_flags |= VM_DONTCOPY;
++
++#ifdef CONFIG_X86
++ vma->vm_mm->context.has_foreign_mappings = 1;
++#endif
++
++ info->vma = vma;
++ info->ring_ok = 1;
++ return 0;
++ fail:
++ /* Clear any active mappings. */
++ zap_page_range(vma, vma->vm_start,
++ vma->vm_end - vma->vm_start, NULL);
++
++ return -ENOMEM;
++}
++
++
++static int blktap_ioctl(struct inode *inode, struct file *filp,
++ unsigned int cmd, unsigned long arg)
++{
++ tap_blkif_t *info = filp->private_data;
++
++ switch(cmd) {
++ case BLKTAP_IOCTL_KICK_FE:
++ {
++ /* There are fe messages to process. */
++ return blktap_read_ufe_ring(info);
++ }
++ case BLKTAP_IOCTL_SETMODE:
++ {
++ if (info) {
++ if (BLKTAP_MODE_VALID(arg)) {
++ info->mode = arg;
++ /* XXX: may need to flush rings here. */
++ DPRINTK("blktap: set mode to %lx\n",
++ arg);
++ return 0;
++ }
++ }
++ return 0;
++ }
++ case BLKTAP_IOCTL_PRINT_IDXS:
++ {
++ if (info) {
++ printk("User Rings: \n-----------\n");
++ printk("UF: rsp_cons: %2d, req_prod_prv: %2d "
++ "| req_prod: %2d, rsp_prod: %2d\n",
++ info->ufe_ring.rsp_cons,
++ info->ufe_ring.req_prod_pvt,
++ info->ufe_ring.sring->req_prod,
++ info->ufe_ring.sring->rsp_prod);
++ }
++ return 0;
++ }
++ case BLKTAP_IOCTL_SENDPID:
++ {
++ if (info) {
++ info->pid = (pid_t)arg;
++ DPRINTK("blktap: pid received %d\n",
++ info->pid);
++ }
++ return 0;
++ }
++ case BLKTAP_IOCTL_NEWINTF:
++ {
++ uint64_t val = (uint64_t)arg;
++ domid_translate_t *tr = (domid_translate_t *)&val;
++
++ DPRINTK("NEWINTF Req for domid %d and bus id %d\n",
++ tr->domid, tr->busid);
++ info = get_next_free_dev();
++ if (!info) {
++ WPRINTK("Error initialising /dev/xen/blktap - "
++ "No more devices\n");
++ return -1;
++ }
++ info->trans.domid = tr->domid;
++ info->trans.busid = tr->busid;
++ return info->minor;
++ }
++ case BLKTAP_IOCTL_FREEINTF:
++ {
++ unsigned long dev = arg;
++ unsigned long flags;
++
++ info = tapfds[dev];
++
++ if ((dev > MAX_TAP_DEV) || !info)
++ return 0; /* should this be an error? */
++
++ spin_lock_irqsave(&pending_free_lock, flags);
++ if (info->dev_pending)
++ info->dev_pending = 0;
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++
++ return 0;
++ }
++ case BLKTAP_IOCTL_MINOR:
++ {
++ unsigned long dev = arg;
++
++ info = tapfds[dev];
++
++ if ((dev > MAX_TAP_DEV) || !info)
++ return -EINVAL;
++
++ return info->minor;
++ }
++ case BLKTAP_IOCTL_MAJOR:
++ return blktap_major;
++
++ case BLKTAP_QUERY_ALLOC_REQS:
++ {
++ WPRINTK("BLKTAP_QUERY_ALLOC_REQS ioctl: %d/%d\n",
++ alloc_pending_reqs, blkif_reqs);
++ return (alloc_pending_reqs/blkif_reqs) * 100;
++ }
++ }
++ return -ENOIOCTLCMD;
++}
++
++static unsigned int blktap_poll(struct file *filp, poll_table *wait)
++{
++ tap_blkif_t *info = filp->private_data;
++
++ /* do not work on the control device */
++ if (!info)
++ return 0;
++
++ poll_wait(filp, &info->wait, wait);
++ if (info->ufe_ring.req_prod_pvt != info->ufe_ring.sring->req_prod) {
++ RING_PUSH_REQUESTS(&info->ufe_ring);
++ return POLLIN | POLLRDNORM;
++ }
++ return 0;
++}
++
++void blktap_kick_user(int idx)
++{
++ tap_blkif_t *info;
++
++ info = tapfds[idx];
++
++ if ((idx < 0) || (idx > MAX_TAP_DEV) || !info)
++ return;
++
++ wake_up_interruptible(&info->wait);
++
++ return;
++}
++
++static int do_block_io_op(blkif_t *blkif);
++static void dispatch_rw_block_io(blkif_t *blkif,
++ blkif_request_t *req,
++ pending_req_t *pending_req);
++static void make_response(blkif_t *blkif, u64 id,
++ unsigned short op, int st);
++
++/******************************************************************
++ * misc small helpers
++ */
++static int req_increase(void)
++{
++ int i, j;
++
++ if (mmap_alloc >= MAX_PENDING_REQS || mmap_lock)
++ return -EINVAL;
++
++ pending_reqs[mmap_alloc] = kzalloc(sizeof(pending_req_t)
++ * blkif_reqs, GFP_KERNEL);
++ foreign_pages[mmap_alloc] = alloc_empty_pages_and_pagevec(mmap_pages);
++
++ if (!pending_reqs[mmap_alloc] || !foreign_pages[mmap_alloc])
++ goto out_of_memory;
++
++ DPRINTK("%s: reqs=%d, pages=%d\n",
++ __FUNCTION__, blkif_reqs, mmap_pages);
++
++ for (i = 0; i < MAX_PENDING_REQS; i++) {
++ list_add_tail(&pending_reqs[mmap_alloc][i].free_list,
++ &pending_free);
++ pending_reqs[mmap_alloc][i].mem_idx = mmap_alloc;
++ for (j = 0; j < BLKIF_MAX_SEGMENTS_PER_REQUEST; j++)
++ BLKTAP_INVALIDATE_HANDLE(&pending_handle(mmap_alloc,
++ i, j));
++ }
++
++ mmap_alloc++;
++ DPRINTK("# MMAPs increased to %d\n",mmap_alloc);
++ return 0;
++
++ out_of_memory:
++ free_empty_pages_and_pagevec(foreign_pages[mmap_alloc], mmap_pages);
++ kfree(pending_reqs[mmap_alloc]);
++ WPRINTK("%s: out of memory\n", __FUNCTION__);
++ return -ENOMEM;
++}
++
++static void mmap_req_del(int mmap)
++{
++ BUG_ON(!spin_is_locked(&pending_free_lock));
++
++ kfree(pending_reqs[mmap]);
++ pending_reqs[mmap] = NULL;
++
++ free_empty_pages_and_pagevec(foreign_pages[mmap_alloc], mmap_pages);
++ foreign_pages[mmap] = NULL;
++
++ mmap_lock = 0;
++ DPRINTK("# MMAPs decreased to %d\n",mmap_alloc);
++ mmap_alloc--;
++}
++
++static pending_req_t* alloc_req(void)
++{
++ pending_req_t *req = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pending_free_lock, flags);
++
++ if (!list_empty(&pending_free)) {
++ req = list_entry(pending_free.next, pending_req_t, free_list);
++ list_del(&req->free_list);
++ }
++
++ if (req) {
++ req->inuse = 1;
++ alloc_pending_reqs++;
++ }
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++
++ return req;
++}
++
++static void free_req(pending_req_t *req)
++{
++ unsigned long flags;
++ int was_empty;
++
++ spin_lock_irqsave(&pending_free_lock, flags);
++
++ alloc_pending_reqs--;
++ req->inuse = 0;
++ if (mmap_lock && (req->mem_idx == mmap_alloc-1)) {
++ mmap_inuse--;
++ if (mmap_inuse == 0) mmap_req_del(mmap_alloc-1);
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++ return;
++ }
++ was_empty = list_empty(&pending_free);
++ list_add(&req->free_list, &pending_free);
++
++ spin_unlock_irqrestore(&pending_free_lock, flags);
++
++ if (was_empty)
++ wake_up(&pending_free_wq);
++}
++
++static void fast_flush_area(pending_req_t *req, int k_idx, int u_idx,
++ int tapidx)
++{
++ struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST*2];
++ unsigned int i, invcount = 0;
++ struct grant_handle_pair *khandle;
++ uint64_t ptep;
++ int ret, mmap_idx;
++ unsigned long kvaddr, uvaddr;
++ tap_blkif_t *info;
++
++
++ info = tapfds[tapidx];
++
++ if ((tapidx < 0) || (tapidx > MAX_TAP_DEV) || !info) {
++ WPRINTK("fast_flush: Couldn't get info!\n");
++ return;
++ }
++
++ if (info->vma != NULL &&
++ xen_feature(XENFEAT_auto_translated_physmap)) {
++ down_write(&info->vma->vm_mm->mmap_sem);
++ zap_page_range(info->vma,
++ MMAP_VADDR(info->user_vstart, u_idx, 0),
++ req->nr_pages << PAGE_SHIFT, NULL);
++ up_write(&info->vma->vm_mm->mmap_sem);
++ return;
++ }
++
++ mmap_idx = req->mem_idx;
++
++ for (i = 0; i < req->nr_pages; i++) {
++ kvaddr = idx_to_kaddr(mmap_idx, k_idx, i);
++ uvaddr = MMAP_VADDR(info->user_vstart, u_idx, i);
++
++ khandle = &pending_handle(mmap_idx, k_idx, i);
++
++ if (khandle->kernel != INVALID_GRANT_HANDLE) {
++ gnttab_set_unmap_op(&unmap[invcount],
++ idx_to_kaddr(mmap_idx, k_idx, i),
++ GNTMAP_host_map, khandle->kernel);
++ invcount++;
++
++ set_phys_to_machine(
++ __pa(idx_to_kaddr(mmap_idx, k_idx, i))
++ >> PAGE_SHIFT, INVALID_P2M_ENTRY);
++ }
++
++ if (khandle->user != INVALID_GRANT_HANDLE) {
++ BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
++ if (create_lookup_pte_addr(
++ info->vma->vm_mm,
++ MMAP_VADDR(info->user_vstart, u_idx, i),
++ &ptep) !=0) {
++ WPRINTK("Couldn't get a pte addr!\n");
++ return;
++ }
++
++ gnttab_set_unmap_op(&unmap[invcount], ptep,
++ GNTMAP_host_map
++ | GNTMAP_application_map
++ | GNTMAP_contains_pte,
++ khandle->user);
++ invcount++;
++ }
++
++ BLKTAP_INVALIDATE_HANDLE(khandle);
++ }
++ ret = HYPERVISOR_grant_table_op(
++ GNTTABOP_unmap_grant_ref, unmap, invcount);
++ BUG_ON(ret);
++
++ if (info->vma != NULL && !xen_feature(XENFEAT_auto_translated_physmap))
++ zap_page_range(info->vma,
++ MMAP_VADDR(info->user_vstart, u_idx, 0),
++ req->nr_pages << PAGE_SHIFT, NULL);
++}
++
++/******************************************************************
++ * SCHEDULER FUNCTIONS
++ */
++
++static void print_stats(blkif_t *blkif)
++{
++ printk(KERN_DEBUG "%s: oo %3d | rd %4d | wr %4d\n",
++ current->comm, blkif->st_oo_req,
++ blkif->st_rd_req, blkif->st_wr_req);
++ blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
++ blkif->st_rd_req = 0;
++ blkif->st_wr_req = 0;
++ blkif->st_oo_req = 0;
++}
++
++int tap_blkif_schedule(void *arg)
++{
++ blkif_t *blkif = arg;
++
++ blkif_get(blkif);
++
++ if (debug_lvl)
++ printk(KERN_DEBUG "%s: started\n", current->comm);
++
++ while (!kthread_should_stop()) {
++ if (try_to_freeze())
++ continue;
++
++ wait_event_interruptible(
++ blkif->wq,
++ blkif->waiting_reqs || kthread_should_stop());
++ wait_event_interruptible(
++ pending_free_wq,
++ !list_empty(&pending_free) || kthread_should_stop());
++
++ blkif->waiting_reqs = 0;
++ smp_mb(); /* clear flag *before* checking for work */
++
++ if (do_block_io_op(blkif))
++ blkif->waiting_reqs = 1;
++
++ if (log_stats && time_after(jiffies, blkif->st_print))
++ print_stats(blkif);
++ }
++
++ if (log_stats)
++ print_stats(blkif);
++ if (debug_lvl)
++ printk(KERN_DEBUG "%s: exiting\n", current->comm);
++
++ blkif->xenblkd = NULL;
++ blkif_put(blkif);
++
++ return 0;
++}
++
++/******************************************************************
++ * COMPLETION CALLBACK -- Called by user level ioctl()
++ */
++
++static int blktap_read_ufe_ring(tap_blkif_t *info)
++{
++ /* This is called to read responses from the UFE ring. */
++ RING_IDX i, j, rp;
++ blkif_response_t *resp;
++ blkif_t *blkif=NULL;
++ int pending_idx, usr_idx, mmap_idx;
++ pending_req_t *pending_req;
++
++ if (!info)
++ return 0;
++
++ /* We currently only forward packets in INTERCEPT_FE mode. */
++ if (!(info->mode & BLKTAP_MODE_INTERCEPT_FE))
++ return 0;
++
++ /* for each outstanding message on the UFEring */
++ rp = info->ufe_ring.sring->rsp_prod;
++ rmb();
++
++ for (i = info->ufe_ring.rsp_cons; i != rp; i++) {
++ blkif_response_t res;
++ resp = RING_GET_RESPONSE(&info->ufe_ring, i);
++ memcpy(&res, resp, sizeof(res));
++ mb(); /* rsp_cons read by RING_FULL() in do_block_io_op(). */
++ ++info->ufe_ring.rsp_cons;
++
++ /*retrieve [usr_idx] to [mmap_idx,pending_idx] mapping*/
++ usr_idx = (int)res.id;
++ pending_idx = MASK_PEND_IDX(ID_TO_IDX(info->idx_map[usr_idx]));
++ mmap_idx = ID_TO_MIDX(info->idx_map[usr_idx]);
++
++ if ( (mmap_idx >= mmap_alloc) ||
++ (ID_TO_IDX(info->idx_map[usr_idx]) >= MAX_PENDING_REQS) )
++ WPRINTK("Incorrect req map"
++ "[%d], internal map [%d,%d (%d)]\n",
++ usr_idx, mmap_idx,
++ ID_TO_IDX(info->idx_map[usr_idx]),
++ MASK_PEND_IDX(
++ ID_TO_IDX(info->idx_map[usr_idx])));
++
++ pending_req = &pending_reqs[mmap_idx][pending_idx];
++ blkif = pending_req->blkif;
++
++ for (j = 0; j < pending_req->nr_pages; j++) {
++
++ unsigned long kvaddr, uvaddr;
++ struct page **map = info->vma->vm_private_data;
++ struct page *pg;
++ int offset;
++
++ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, j);
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, j);
++
++ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
++ ClearPageReserved(pg);
++ offset = (uvaddr - info->vma->vm_start)
++ >> PAGE_SHIFT;
++ map[offset] = NULL;
++ }
++ fast_flush_area(pending_req, pending_idx, usr_idx, info->minor);
++ info->idx_map[usr_idx] = INVALID_REQ;
++ make_response(blkif, pending_req->id, res.operation,
++ res.status);
++ blkif_put(pending_req->blkif);
++ free_req(pending_req);
++ }
++
++ return 0;
++}
++
++
++/******************************************************************************
++ * NOTIFICATION FROM GUEST OS.
++ */
++
++static void blkif_notify_work(blkif_t *blkif)
++{
++ blkif->waiting_reqs = 1;
++ wake_up(&blkif->wq);
++}
++
++irqreturn_t tap_blkif_be_int(int irq, void *dev_id, struct pt_regs *regs)
++{
++ blkif_notify_work(dev_id);
++ return IRQ_HANDLED;
++}
++
++
++
++/******************************************************************
++ * DOWNWARD CALLS -- These interface with the block-device layer proper.
++ */
++static int print_dbug = 1;
++static int do_block_io_op(blkif_t *blkif)
++{
++ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
++ blkif_request_t req;
++ pending_req_t *pending_req;
++ RING_IDX rc, rp;
++ int more_to_do = 0;
++ tap_blkif_t *info;
++
++ rc = blk_rings->common.req_cons;
++ rp = blk_rings->common.sring->req_prod;
++ rmb(); /* Ensure we see queued requests up to 'rp'. */
++
++ /*Check blkif has corresponding UE ring*/
++ if (blkif->dev_num < 0) {
++ /*oops*/
++ if (print_dbug) {
++ WPRINTK("Corresponding UE "
++ "ring does not exist!\n");
++ print_dbug = 0; /*We only print this message once*/
++ }
++ return 0;
++ }
++
++ info = tapfds[blkif->dev_num];
++
++ if (blkif->dev_num > MAX_TAP_DEV || !info || !info->dev_inuse) {
++ if (print_dbug) {
++ WPRINTK("Can't get UE info!\n");
++ print_dbug = 0;
++ }
++ return 0;
++ }
++
++ while (rc != rp) {
++
++ if (RING_FULL(&info->ufe_ring)) {
++ WPRINTK("RING_FULL! More to do\n");
++ more_to_do = 1;
++ break;
++ }
++
++ if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc)) {
++ WPRINTK("RING_REQUEST_CONS_OVERFLOW!"
++ " More to do\n");
++ more_to_do = 1;
++ break;
++ }
++
++ pending_req = alloc_req();
++ if (NULL == pending_req) {
++ blkif->st_oo_req++;
++ more_to_do = 1;
++ break;
++ }
++
++ if (kthread_should_stop()) {
++ more_to_do = 1;
++ break;
++ }
++
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc),
++ sizeof(req));
++ break;
++ case BLKIF_PROTOCOL_X86_32:
++ blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
++ break;
++ case BLKIF_PROTOCOL_X86_64:
++ blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
++ break;
++ default:
++ BUG();
++ }
++ blk_rings->common.req_cons = ++rc; /* before make_response() */
++
++ /* Apply all sanity checks to /private copy/ of request. */
++ barrier();
++
++ switch (req.operation) {
++ case BLKIF_OP_READ:
++ blkif->st_rd_req++;
++ dispatch_rw_block_io(blkif, &req, pending_req);
++ break;
++
++ case BLKIF_OP_WRITE:
++ blkif->st_wr_req++;
++ dispatch_rw_block_io(blkif, &req, pending_req);
++ break;
++
++ default:
++ /* A good sign something is wrong: sleep for a while to
++ * avoid excessive CPU consumption by a bad guest. */
++ msleep(1);
++ WPRINTK("unknown operation [%d]\n",
++ req.operation);
++ make_response(blkif, req.id, req.operation,
++ BLKIF_RSP_ERROR);
++ free_req(pending_req);
++ break;
++ }
++
++ /* Yield point for this unbounded loop. */
++ cond_resched();
++ }
++
++ blktap_kick_user(blkif->dev_num);
++
++ return more_to_do;
++}
++
++static void dispatch_rw_block_io(blkif_t *blkif,
++ blkif_request_t *req,
++ pending_req_t *pending_req)
++{
++ extern void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]);
++ int op, operation = (req->operation == BLKIF_OP_WRITE) ? WRITE : READ;
++ struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST*2];
++ unsigned int nseg;
++ int ret, i, nr_sects = 0;
++ tap_blkif_t *info;
++ blkif_request_t *target;
++ int pending_idx = RTN_PEND_IDX(pending_req,pending_req->mem_idx);
++ int usr_idx;
++ uint16_t mmap_idx = pending_req->mem_idx;
++
++ if (blkif->dev_num < 0 || blkif->dev_num > MAX_TAP_DEV)
++ goto fail_response;
++
++ info = tapfds[blkif->dev_num];
++ if (info == NULL)
++ goto fail_response;
++
++ /* Check we have space on user ring - should never fail. */
++ usr_idx = GET_NEXT_REQ(info->idx_map);
++ if (usr_idx == INVALID_REQ) {
++ BUG();
++ goto fail_response;
++ }
++
++ /* Check that number of segments is sane. */
++ nseg = req->nr_segments;
++ if ( unlikely(nseg == 0) ||
++ unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST) ) {
++ WPRINTK("Bad number of segments in request (%d)\n", nseg);
++ goto fail_response;
++ }
++
++ /* Make sure userspace is ready. */
++ if (!info->ring_ok) {
++ WPRINTK("blktap: ring not ready for requests!\n");
++ goto fail_response;
++ }
++
++ if (RING_FULL(&info->ufe_ring)) {
++ WPRINTK("blktap: fe_ring is full, can't add "
++ "IO Request will be dropped. %d %d\n",
++ RING_SIZE(&info->ufe_ring),
++ RING_SIZE(&blkif->blk_rings.common));
++ goto fail_response;
++ }
++
++ pending_req->blkif = blkif;
++ pending_req->id = req->id;
++ pending_req->operation = operation;
++ pending_req->status = BLKIF_RSP_OKAY;
++ pending_req->nr_pages = nseg;
++ op = 0;
++ for (i = 0; i < nseg; i++) {
++ unsigned long uvaddr;
++ unsigned long kvaddr;
++ uint64_t ptep;
++ uint32_t flags;
++
++ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i);
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
++
++ flags = GNTMAP_host_map;
++ if (operation == WRITE)
++ flags |= GNTMAP_readonly;
++ gnttab_set_map_op(&map[op], kvaddr, flags,
++ req->seg[i].gref, blkif->domid);
++ op++;
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* Now map it to user. */
++ ret = create_lookup_pte_addr(info->vma->vm_mm,
++ uvaddr, &ptep);
++ if (ret) {
++ WPRINTK("Couldn't get a pte addr!\n");
++ goto fail_flush;
++ }
++
++ flags = GNTMAP_host_map | GNTMAP_application_map
++ | GNTMAP_contains_pte;
++ if (operation == WRITE)
++ flags |= GNTMAP_readonly;
++ gnttab_set_map_op(&map[op], ptep, flags,
++ req->seg[i].gref, blkif->domid);
++ op++;
++ }
++
++ nr_sects += (req->seg[i].last_sect -
++ req->seg[i].first_sect + 1);
++ }
++
++ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, op);
++ BUG_ON(ret);
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ for (i = 0; i < (nseg*2); i+=2) {
++ unsigned long uvaddr;
++ unsigned long kvaddr;
++ unsigned long offset;
++ struct page *pg;
++
++ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i/2);
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i/2);
++
++ if (unlikely(map[i].status != 0)) {
++ WPRINTK("invalid kernel buffer -- "
++ "could not remap it\n");
++ ret |= 1;
++ map[i].handle = INVALID_GRANT_HANDLE;
++ }
++
++ if (unlikely(map[i+1].status != 0)) {
++ WPRINTK("invalid user buffer -- "
++ "could not remap it\n");
++ ret |= 1;
++ map[i+1].handle = INVALID_GRANT_HANDLE;
++ }
++
++ pending_handle(mmap_idx, pending_idx, i/2).kernel
++ = map[i].handle;
++ pending_handle(mmap_idx, pending_idx, i/2).user
++ = map[i+1].handle;
++
++ if (ret)
++ continue;
++
++ set_phys_to_machine(__pa(kvaddr) >> PAGE_SHIFT,
++ FOREIGN_FRAME(map[i].dev_bus_addr
++ >> PAGE_SHIFT));
++ offset = (uvaddr - info->vma->vm_start) >> PAGE_SHIFT;
++ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
++ ((struct page **)info->vma->vm_private_data)[offset] =
++ pg;
++ }
++ } else {
++ for (i = 0; i < nseg; i++) {
++ unsigned long uvaddr;
++ unsigned long kvaddr;
++ unsigned long offset;
++ struct page *pg;
++
++ uvaddr = MMAP_VADDR(info->user_vstart, usr_idx, i);
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
++
++ if (unlikely(map[i].status != 0)) {
++ WPRINTK("invalid kernel buffer -- "
++ "could not remap it\n");
++ ret |= 1;
++ map[i].handle = INVALID_GRANT_HANDLE;
++ }
++
++ pending_handle(mmap_idx, pending_idx, i).kernel
++ = map[i].handle;
++
++ if (ret)
++ continue;
++
++ offset = (uvaddr - info->vma->vm_start) >> PAGE_SHIFT;
++ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
++ ((struct page **)info->vma->vm_private_data)[offset] =
++ pg;
++ }
++ }
++
++ if (ret)
++ goto fail_flush;
++
++ if (xen_feature(XENFEAT_auto_translated_physmap))
++ down_write(&info->vma->vm_mm->mmap_sem);
++ /* Mark mapped pages as reserved: */
++ for (i = 0; i < req->nr_segments; i++) {
++ unsigned long kvaddr;
++ struct page *pg;
++
++ kvaddr = idx_to_kaddr(mmap_idx, pending_idx, i);
++ pg = pfn_to_page(__pa(kvaddr) >> PAGE_SHIFT);
++ SetPageReserved(pg);
++ if (xen_feature(XENFEAT_auto_translated_physmap)) {
++ ret = vm_insert_page(info->vma,
++ MMAP_VADDR(info->user_vstart,
++ usr_idx, i), pg);
++ if (ret) {
++ up_write(&info->vma->vm_mm->mmap_sem);
++ goto fail_flush;
++ }
++ }
++ }
++ if (xen_feature(XENFEAT_auto_translated_physmap))
++ up_write(&info->vma->vm_mm->mmap_sem);
++
++ /*record [mmap_idx,pending_idx] to [usr_idx] mapping*/
++ info->idx_map[usr_idx] = MAKE_ID(mmap_idx, pending_idx);
++
++ blkif_get(blkif);
++ /* Finally, write the request message to the user ring. */
++ target = RING_GET_REQUEST(&info->ufe_ring,
++ info->ufe_ring.req_prod_pvt);
++ memcpy(target, req, sizeof(*req));
++ target->id = usr_idx;
++ wmb(); /* blktap_poll() reads req_prod_pvt asynchronously */
++ info->ufe_ring.req_prod_pvt++;
++
++ if (operation == READ)
++ blkif->st_rd_sect += nr_sects;
++ else if (operation == WRITE)
++ blkif->st_wr_sect += nr_sects;
++
++ return;
++
++ fail_flush:
++ WPRINTK("Reached Fail_flush\n");
++ fast_flush_area(pending_req, pending_idx, usr_idx, blkif->dev_num);
++ fail_response:
++ make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
++ free_req(pending_req);
++ msleep(1); /* back off a bit */
++}
++
++
++
++/******************************************************************
++ * MISCELLANEOUS SETUP / TEARDOWN / DEBUGGING
++ */
++
++
++static void make_response(blkif_t *blkif, u64 id,
++ unsigned short op, int st)
++{
++ blkif_response_t resp;
++ unsigned long flags;
++ blkif_back_rings_t *blk_rings = &blkif->blk_rings;
++ int more_to_do = 0;
++ int notify;
++
++ resp.id = id;
++ resp.operation = op;
++ resp.status = st;
++
++ spin_lock_irqsave(&blkif->blk_ring_lock, flags);
++ /* Place on the response ring for the relevant domain. */
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ memcpy(RING_GET_RESPONSE(&blk_rings->native,
++ blk_rings->native.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ case BLKIF_PROTOCOL_X86_32:
++ memcpy(RING_GET_RESPONSE(&blk_rings->x86_32,
++ blk_rings->x86_32.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ case BLKIF_PROTOCOL_X86_64:
++ memcpy(RING_GET_RESPONSE(&blk_rings->x86_64,
++ blk_rings->x86_64.rsp_prod_pvt),
++ &resp, sizeof(resp));
++ break;
++ default:
++ BUG();
++ }
++ blk_rings->common.rsp_prod_pvt++;
++ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
++
++ if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) {
++ /*
++ * Tail check for pending requests. Allows frontend to avoid
++ * notifications if requests are already in flight (lower
++ * overheads and promotes batching).
++ */
++ RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
++ } else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) {
++ more_to_do = 1;
++ }
++
++ spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
++ if (more_to_do)
++ blkif_notify_work(blkif);
++ if (notify)
++ notify_remote_via_irq(blkif->irq);
++}
++
++static int __init blkif_init(void)
++{
++ int i, ret;
++ struct class *class;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ INIT_LIST_HEAD(&pending_free);
++ for(i = 0; i < 2; i++) {
++ ret = req_increase();
++ if (ret)
++ break;
++ }
++ if (i == 0)
++ return ret;
++
++ tap_blkif_interface_init();
++
++ alloc_pending_reqs = 0;
++
++ tap_blkif_xenbus_init();
++
++ /* Dynamically allocate a major for this device */
++ ret = register_chrdev(0, "blktap", &blktap_fops);
++
++ if (ret < 0) {
++ WPRINTK("Couldn't register /dev/xen/blktap\n");
++ return -ENOMEM;
++ }
++
++ blktap_major = ret;
++
++ /* tapfds[0] is always NULL */
++ blktap_next_minor++;
++
++ DPRINTK("Created misc_dev [/dev/xen/blktap%d]\n",i);
++
++ /* Make sure the xen class exists */
++ if ((class = get_xen_class()) != NULL) {
++ /*
++ * This will allow udev to create the blktap ctrl device.
++ * We only want to create blktap0 first. We don't want
++ * to flood the sysfs system with needless blktap devices.
++ * We only create the device when a request of a new device is
++ * made.
++ */
++ class_device_create(class, NULL,
++ MKDEV(blktap_major, 0), NULL,
++ "blktap0");
++ } else {
++ /* this is bad, but not fatal */
++ WPRINTK("blktap: sysfs xen_class not created\n");
++ }
++
++ DPRINTK("Blktap device successfully created\n");
++
++ return 0;
++}
++
++module_init(blkif_init);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/blktap/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blktap/common.h 2008-05-19 00:33:45.714781173 +0300
+@@ -0,0 +1,121 @@
++/*
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef __BLKIF__BACKEND__COMMON_H__
++#define __BLKIF__BACKEND__COMMON_H__
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/interrupt.h>
++#include <linux/slab.h>
++#include <linux/blkdev.h>
++#include <linux/vmalloc.h>
++#include <asm/io.h>
++#include <asm/setup.h>
++#include <asm/pgalloc.h>
++#include <xen/evtchn.h>
++#include <asm/hypervisor.h>
++#include <xen/blkif.h>
++#include <xen/gnttab.h>
++#include <xen/driver_util.h>
++
++#define DPRINTK(_f, _a...) pr_debug("(file=%s, line=%d) " _f, \
++ __FILE__ , __LINE__ , ## _a )
++
++#define WPRINTK(fmt, args...) printk(KERN_WARNING "blk_tap: " fmt, ##args)
++
++struct backend_info;
++
++typedef struct blkif_st {
++ /* Unique identifier for this interface. */
++ domid_t domid;
++ unsigned int handle;
++ /* Physical parameters of the comms window. */
++ unsigned int irq;
++ /* Comms information. */
++ enum blkif_protocol blk_protocol;
++ blkif_back_rings_t blk_rings;
++ struct vm_struct *blk_ring_area;
++ /* Back pointer to the backend_info. */
++ struct backend_info *be;
++ /* Private fields. */
++ spinlock_t blk_ring_lock;
++ atomic_t refcnt;
++
++ wait_queue_head_t wq;
++ struct task_struct *xenblkd;
++ unsigned int waiting_reqs;
++ request_queue_t *plug;
++
++ /* statistics */
++ unsigned long st_print;
++ int st_rd_req;
++ int st_wr_req;
++ int st_oo_req;
++ int st_rd_sect;
++ int st_wr_sect;
++
++ wait_queue_head_t waiting_to_free;
++
++ grant_handle_t shmem_handle;
++ grant_ref_t shmem_ref;
++
++ int dev_num;
++ uint64_t sectors;
++} blkif_t;
++
++blkif_t *tap_alloc_blkif(domid_t domid);
++void tap_blkif_free(blkif_t *blkif);
++int tap_blkif_map(blkif_t *blkif, unsigned long shared_page,
++ unsigned int evtchn);
++void tap_blkif_unmap(blkif_t *blkif);
++
++#define blkif_get(_b) (atomic_inc(&(_b)->refcnt))
++#define blkif_put(_b) \
++ do { \
++ if (atomic_dec_and_test(&(_b)->refcnt)) \
++ wake_up(&(_b)->waiting_to_free);\
++ } while (0)
++
++
++struct phys_req {
++ unsigned short dev;
++ unsigned short nr_sects;
++ struct block_device *bdev;
++ blkif_sector_t sector_number;
++};
++
++void tap_blkif_interface_init(void);
++
++void tap_blkif_xenbus_init(void);
++
++irqreturn_t tap_blkif_be_int(int irq, void *dev_id, struct pt_regs *regs);
++int tap_blkif_schedule(void *arg);
++
++int dom_to_devid(domid_t domid, int xenbus_id, blkif_t *blkif);
++void signal_tapdisk(int idx);
++
++#endif /* __BLKIF__BACKEND__COMMON_H__ */
+--- linux-2.6.18.8/drivers/xen/blktap/interface.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blktap/interface.c 2008-05-19 00:33:45.714781173 +0300
+@@ -0,0 +1,174 @@
++/******************************************************************************
++ * drivers/xen/blktap/interface.c
++ *
++ * Block-device interface management.
++ *
++ * Copyright (c) 2004, Keir Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++
++ */
++
++#include "common.h"
++#include <xen/evtchn.h>
++
++static kmem_cache_t *blkif_cachep;
++
++blkif_t *tap_alloc_blkif(domid_t domid)
++{
++ blkif_t *blkif;
++
++ blkif = kmem_cache_alloc(blkif_cachep, GFP_KERNEL);
++ if (!blkif)
++ return ERR_PTR(-ENOMEM);
++
++ memset(blkif, 0, sizeof(*blkif));
++ blkif->domid = domid;
++ spin_lock_init(&blkif->blk_ring_lock);
++ atomic_set(&blkif->refcnt, 1);
++ init_waitqueue_head(&blkif->wq);
++ blkif->st_print = jiffies;
++ init_waitqueue_head(&blkif->waiting_to_free);
++
++ return blkif;
++}
++
++static int map_frontend_page(blkif_t *blkif, unsigned long shared_page)
++{
++ struct gnttab_map_grant_ref op;
++
++ gnttab_set_map_op(&op, (unsigned long)blkif->blk_ring_area->addr,
++ GNTMAP_host_map, shared_page, blkif->domid);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
++ BUG();
++
++ if (op.status) {
++ DPRINTK(" Grant table operation failure !\n");
++ return op.status;
++ }
++
++ blkif->shmem_ref = shared_page;
++ blkif->shmem_handle = op.handle;
++
++ return 0;
++}
++
++static void unmap_frontend_page(blkif_t *blkif)
++{
++ struct gnttab_unmap_grant_ref op;
++
++ gnttab_set_unmap_op(&op, (unsigned long)blkif->blk_ring_area->addr,
++ GNTMAP_host_map, blkif->shmem_handle);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
++ BUG();
++}
++
++int tap_blkif_map(blkif_t *blkif, unsigned long shared_page,
++ unsigned int evtchn)
++{
++ int err;
++
++ /* Already connected through? */
++ if (blkif->irq)
++ return 0;
++
++ if ( (blkif->blk_ring_area = alloc_vm_area(PAGE_SIZE)) == NULL )
++ return -ENOMEM;
++
++ err = map_frontend_page(blkif, shared_page);
++ if (err) {
++ free_vm_area(blkif->blk_ring_area);
++ return err;
++ }
++
++ switch (blkif->blk_protocol) {
++ case BLKIF_PROTOCOL_NATIVE:
++ {
++ blkif_sring_t *sring;
++ sring = (blkif_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.native, sring, PAGE_SIZE);
++ break;
++ }
++ case BLKIF_PROTOCOL_X86_32:
++ {
++ blkif_x86_32_sring_t *sring_x86_32;
++ sring_x86_32 = (blkif_x86_32_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.x86_32, sring_x86_32, PAGE_SIZE);
++ break;
++ }
++ case BLKIF_PROTOCOL_X86_64:
++ {
++ blkif_x86_64_sring_t *sring_x86_64;
++ sring_x86_64 = (blkif_x86_64_sring_t *)blkif->blk_ring_area->addr;
++ BACK_RING_INIT(&blkif->blk_rings.x86_64, sring_x86_64, PAGE_SIZE);
++ break;
++ }
++ default:
++ BUG();
++ }
++
++ err = bind_interdomain_evtchn_to_irqhandler(
++ blkif->domid, evtchn, tap_blkif_be_int,
++ 0, "blkif-backend", blkif);
++ if (err < 0) {
++ unmap_frontend_page(blkif);
++ free_vm_area(blkif->blk_ring_area);
++ blkif->blk_rings.common.sring = NULL;
++ return err;
++ }
++ blkif->irq = err;
++
++ return 0;
++}
++
++void tap_blkif_unmap(blkif_t *blkif)
++{
++ if (blkif->irq) {
++ unbind_from_irqhandler(blkif->irq, blkif);
++ blkif->irq = 0;
++ }
++ if (blkif->blk_rings.common.sring) {
++ unmap_frontend_page(blkif);
++ free_vm_area(blkif->blk_ring_area);
++ blkif->blk_rings.common.sring = NULL;
++ }
++}
++
++void tap_blkif_free(blkif_t *blkif)
++{
++ atomic_dec(&blkif->refcnt);
++ wait_event(blkif->waiting_to_free, atomic_read(&blkif->refcnt) == 0);
++
++ tap_blkif_unmap(blkif);
++ kmem_cache_free(blkif_cachep, blkif);
++}
++
++void __init tap_blkif_interface_init(void)
++{
++ blkif_cachep = kmem_cache_create("blktapif_cache", sizeof(blkif_t),
++ 0, 0, NULL, NULL);
++}
+--- linux-2.6.18.8/drivers/xen/blktap/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/blktap/xenbus.c 2008-05-19 00:33:45.714781173 +0300
+@@ -0,0 +1,477 @@
++/* drivers/xen/blktap/xenbus.c
++ *
++ * Xenbus code for blktap
++ *
++ * Copyright (c) 2004-2005, Andrew Warfield and Julian Chesterfield
++ *
++ * Based on the blkback xenbus code:
++ *
++ * Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
++ * Copyright (C) 2005 XenSource Ltd
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <stdarg.h>
++#include <linux/module.h>
++#include <linux/kthread.h>
++#include <xen/xenbus.h>
++#include "common.h"
++
++
++struct backend_info
++{
++ struct xenbus_device *dev;
++ blkif_t *blkif;
++ struct xenbus_watch backend_watch;
++ int xenbus_id;
++ int group_added;
++};
++
++
++static void connect(struct backend_info *);
++static int connect_ring(struct backend_info *);
++static int blktap_remove(struct xenbus_device *dev);
++static int blktap_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id);
++static void tap_backend_changed(struct xenbus_watch *, const char **,
++ unsigned int);
++static void tap_frontend_changed(struct xenbus_device *dev,
++ enum xenbus_state frontend_state);
++
++static int strsep_len(const char *str, char c, unsigned int len)
++{
++ unsigned int i;
++
++ for (i = 0; str[i]; i++)
++ if (str[i] == c) {
++ if (len == 0)
++ return i;
++ len--;
++ }
++ return (len == 0) ? i : -ERANGE;
++}
++
++static long get_id(const char *str)
++{
++ int len,end;
++ const char *ptr;
++ char *tptr, num[10];
++
++ len = strsep_len(str, '/', 2);
++ end = strlen(str);
++ if ( (len < 0) || (end < 0) ) return -1;
++
++ ptr = str + len + 1;
++ strncpy(num,ptr,end - len);
++ tptr = num + (end - (len + 1));
++ *tptr = '\0';
++ DPRINTK("Get_id called for %s (%s)\n",str,num);
++
++ return simple_strtol(num, NULL, 10);
++}
++
++static int blktap_name(blkif_t *blkif, char *buf)
++{
++ char *devpath, *devname;
++ struct xenbus_device *dev = blkif->be->dev;
++
++ devpath = xenbus_read(XBT_NIL, dev->nodename, "dev", NULL);
++ if (IS_ERR(devpath))
++ return PTR_ERR(devpath);
++
++ if ((devname = strstr(devpath, "/dev/")) != NULL)
++ devname += strlen("/dev/");
++ else
++ devname = devpath;
++
++ snprintf(buf, TASK_COMM_LEN, "blktap.%d.%s", blkif->domid, devname);
++ kfree(devpath);
++
++ return 0;
++}
++
++/****************************************************************
++ * sysfs interface for I/O requests of blktap device
++ */
++
++#define VBD_SHOW(name, format, args...) \
++ static ssize_t show_##name(struct device *_dev, \
++ struct device_attribute *attr, \
++ char *buf) \
++ { \
++ struct xenbus_device *dev = to_xenbus_device(_dev); \
++ struct backend_info *be = dev->dev.driver_data; \
++ \
++ return sprintf(buf, format, ##args); \
++ } \
++ static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL)
++
++VBD_SHOW(oo_req, "%d\n", be->blkif->st_oo_req);
++VBD_SHOW(rd_req, "%d\n", be->blkif->st_rd_req);
++VBD_SHOW(wr_req, "%d\n", be->blkif->st_wr_req);
++VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect);
++VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect);
++
++static struct attribute *tapstat_attrs[] = {
++ &dev_attr_oo_req.attr,
++ &dev_attr_rd_req.attr,
++ &dev_attr_wr_req.attr,
++ &dev_attr_rd_sect.attr,
++ &dev_attr_wr_sect.attr,
++ NULL
++};
++
++static struct attribute_group tapstat_group = {
++ .name = "statistics",
++ .attrs = tapstat_attrs,
++};
++
++int xentap_sysfs_addif(struct xenbus_device *dev)
++{
++ int err;
++ struct backend_info *be = dev->dev.driver_data;
++ err = sysfs_create_group(&dev->dev.kobj, &tapstat_group);
++ if (!err)
++ be->group_added = 1;
++ return err;
++}
++
++void xentap_sysfs_delif(struct xenbus_device *dev)
++{
++ struct backend_info *be = dev->dev.driver_data;
++ sysfs_remove_group(&dev->dev.kobj, &tapstat_group);
++ be->group_added = 0;
++}
++
++static int blktap_remove(struct xenbus_device *dev)
++{
++ struct backend_info *be = dev->dev.driver_data;
++
++ if (be->group_added)
++ xentap_sysfs_delif(be->dev);
++ if (be->backend_watch.node) {
++ unregister_xenbus_watch(&be->backend_watch);
++ kfree(be->backend_watch.node);
++ be->backend_watch.node = NULL;
++ }
++ if (be->blkif) {
++ if (be->blkif->xenblkd)
++ kthread_stop(be->blkif->xenblkd);
++ signal_tapdisk(be->blkif->dev_num);
++ tap_blkif_free(be->blkif);
++ be->blkif = NULL;
++ }
++ kfree(be);
++ dev->dev.driver_data = NULL;
++ return 0;
++}
++
++static void tap_update_blkif_status(blkif_t *blkif)
++{
++ int err;
++ char name[TASK_COMM_LEN];
++
++ /* Not ready to connect? */
++ if(!blkif->irq || !blkif->sectors) {
++ return;
++ }
++
++ /* Already connected? */
++ if (blkif->be->dev->state == XenbusStateConnected)
++ return;
++
++ /* Attempt to connect: exit if we fail to. */
++ connect(blkif->be);
++ if (blkif->be->dev->state != XenbusStateConnected)
++ return;
++
++ err = blktap_name(blkif, name);
++ if (err) {
++ xenbus_dev_error(blkif->be->dev, err, "get blktap dev name");
++ return;
++ }
++
++ if (!blkif->be->group_added) {
++ err = xentap_sysfs_addif(blkif->be->dev);
++ if (err) {
++ xenbus_dev_fatal(blkif->be->dev, err,
++ "creating sysfs entries");
++ return;
++ }
++ }
++
++ blkif->xenblkd = kthread_run(tap_blkif_schedule, blkif, name);
++ if (IS_ERR(blkif->xenblkd)) {
++ err = PTR_ERR(blkif->xenblkd);
++ blkif->xenblkd = NULL;
++ xenbus_dev_fatal(blkif->be->dev, err, "start xenblkd");
++ WPRINTK("Error starting thread\n");
++ }
++}
++
++/**
++ * Entry point to this code when a new device is created. Allocate
++ * the basic structures, and watch the store waiting for the
++ * user-space program to tell us the physical device info. Switch to
++ * InitWait.
++ */
++static int blktap_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int err;
++ struct backend_info *be = kzalloc(sizeof(struct backend_info),
++ GFP_KERNEL);
++ if (!be) {
++ xenbus_dev_fatal(dev, -ENOMEM,
++ "allocating backend structure");
++ return -ENOMEM;
++ }
++
++ be->dev = dev;
++ dev->dev.driver_data = be;
++ be->xenbus_id = get_id(dev->nodename);
++
++ be->blkif = tap_alloc_blkif(dev->otherend_id);
++ if (IS_ERR(be->blkif)) {
++ err = PTR_ERR(be->blkif);
++ be->blkif = NULL;
++ xenbus_dev_fatal(dev, err, "creating block interface");
++ goto fail;
++ }
++
++ /* setup back pointer */
++ be->blkif->be = be;
++ be->blkif->sectors = 0;
++
++ /* set a watch on disk info, waiting for userspace to update details*/
++ err = xenbus_watch_path2(dev, dev->nodename, "info",
++ &be->backend_watch, tap_backend_changed);
++ if (err)
++ goto fail;
++
++ err = xenbus_switch_state(dev, XenbusStateInitWait);
++ if (err)
++ goto fail;
++ return 0;
++
++fail:
++ DPRINTK("blktap probe failed\n");
++ blktap_remove(dev);
++ return err;
++}
++
++
++/**
++ * Callback received when the user space code has placed the device
++ * information in xenstore.
++ */
++static void tap_backend_changed(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ int err;
++ unsigned long info;
++ struct backend_info *be
++ = container_of(watch, struct backend_info, backend_watch);
++ struct xenbus_device *dev = be->dev;
++
++ /**
++ * Check to see whether userspace code has opened the image
++ * and written sector
++ * and disk info to xenstore
++ */
++ err = xenbus_gather(XBT_NIL, dev->nodename, "info", "%lu", &info,
++ NULL);
++ if (XENBUS_EXIST_ERR(err))
++ return;
++ if (err) {
++ xenbus_dev_error(dev, err, "getting info");
++ return;
++ }
++
++ DPRINTK("Userspace update on disk info, %lu\n",info);
++
++ err = xenbus_gather(XBT_NIL, dev->nodename, "sectors", "%llu",
++ &be->blkif->sectors, NULL);
++
++ /* Associate tap dev with domid*/
++ be->blkif->dev_num = dom_to_devid(be->blkif->domid, be->xenbus_id,
++ be->blkif);
++ DPRINTK("Thread started for domid [%d], connecting disk\n",
++ be->blkif->dev_num);
++
++ tap_update_blkif_status(be->blkif);
++}
++
++/**
++ * Callback received when the frontend's state changes.
++ */
++static void tap_frontend_changed(struct xenbus_device *dev,
++ enum xenbus_state frontend_state)
++{
++ struct backend_info *be = dev->dev.driver_data;
++ int err;
++
++ DPRINTK("\n");
++
++ switch (frontend_state) {
++ case XenbusStateInitialising:
++ if (dev->state == XenbusStateClosed) {
++ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
++ __FUNCTION__, dev->nodename);
++ xenbus_switch_state(dev, XenbusStateInitWait);
++ }
++ break;
++
++ case XenbusStateInitialised:
++ case XenbusStateConnected:
++ /* Ensure we connect even when two watches fire in
++ close successsion and we miss the intermediate value
++ of frontend_state. */
++ if (dev->state == XenbusStateConnected)
++ break;
++
++ err = connect_ring(be);
++ if (err)
++ break;
++ tap_update_blkif_status(be->blkif);
++ break;
++
++ case XenbusStateClosing:
++ if (be->blkif->xenblkd) {
++ kthread_stop(be->blkif->xenblkd);
++ be->blkif->xenblkd = NULL;
++ }
++ xenbus_switch_state(dev, XenbusStateClosing);
++ break;
++
++ case XenbusStateClosed:
++ xenbus_switch_state(dev, XenbusStateClosed);
++ if (xenbus_dev_is_online(dev))
++ break;
++ /* fall through if not online */
++ case XenbusStateUnknown:
++ device_unregister(&dev->dev);
++ break;
++
++ default:
++ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
++ frontend_state);
++ break;
++ }
++}
++
++
++/**
++ * Switch to Connected state.
++ */
++static void connect(struct backend_info *be)
++{
++ int err;
++
++ struct xenbus_device *dev = be->dev;
++
++ err = xenbus_switch_state(dev, XenbusStateConnected);
++ if (err)
++ xenbus_dev_fatal(dev, err, "switching to Connected state",
++ dev->nodename);
++
++ return;
++}
++
++
++static int connect_ring(struct backend_info *be)
++{
++ struct xenbus_device *dev = be->dev;
++ unsigned long ring_ref;
++ unsigned int evtchn;
++ char protocol[64];
++ int err;
++
++ DPRINTK("%s\n", dev->otherend);
++
++ err = xenbus_gather(XBT_NIL, dev->otherend, "ring-ref", "%lu",
++ &ring_ref, "event-channel", "%u", &evtchn, NULL);
++ if (err) {
++ xenbus_dev_fatal(dev, err,
++ "reading %s/ring-ref and event-channel",
++ dev->otherend);
++ return err;
++ }
++
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
++ err = xenbus_gather(XBT_NIL, dev->otherend, "protocol",
++ "%63s", protocol, NULL);
++ if (err)
++ strcpy(protocol, "unspecified, assuming native");
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_NATIVE))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE;
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_32))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_32;
++ else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_64))
++ be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_64;
++ else {
++ xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol);
++ return -1;
++ }
++ printk(KERN_INFO
++ "blktap: ring-ref %ld, event-channel %d, protocol %d (%s)\n",
++ ring_ref, evtchn, be->blkif->blk_protocol, protocol);
++
++ /* Map the shared frame, irq etc. */
++ err = tap_blkif_map(be->blkif, ring_ref, evtchn);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "mapping ring-ref %lu port %u",
++ ring_ref, evtchn);
++ return err;
++ }
++
++ return 0;
++}
++
++
++/* ** Driver Registration ** */
++
++
++static const struct xenbus_device_id blktap_ids[] = {
++ { "tap" },
++ { "" }
++};
++
++
++static struct xenbus_driver blktap = {
++ .name = "tap",
++ .owner = THIS_MODULE,
++ .ids = blktap_ids,
++ .probe = blktap_probe,
++ .remove = blktap_remove,
++ .otherend_changed = tap_frontend_changed
++};
++
++
++void tap_blkif_xenbus_init(void)
++{
++ xenbus_register_backend(&blktap);
++}
+--- linux-2.6.18.8/drivers/xen/char/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/char/Makefile 2008-05-19 00:33:45.718781403 +0300
+@@ -0,0 +1 @@
++obj-$(CONFIG_XEN_DEVMEM) := mem.o
+--- linux-2.6.18.8/drivers/xen/char/mem.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/char/mem.c 2008-05-19 00:33:45.718781403 +0300
+@@ -0,0 +1,190 @@
++/*
++ * Originally from linux/drivers/char/mem.c
++ *
++ * Copyright (C) 1991, 1992 Linus Torvalds
++ *
++ * Added devfs support.
++ * Jan-11-1998, C. Scott Ananian <cananian@alumni.princeton.edu>
++ * Shared /dev/zero mmaping support, Feb 2000, Kanoj Sarcar <kanoj@sgi.com>
++ */
++
++#include <linux/mm.h>
++#include <linux/miscdevice.h>
++#include <linux/slab.h>
++#include <linux/vmalloc.h>
++#include <linux/mman.h>
++#include <linux/random.h>
++#include <linux/init.h>
++#include <linux/raw.h>
++#include <linux/tty.h>
++#include <linux/capability.h>
++#include <linux/smp_lock.h>
++#include <linux/ptrace.h>
++#include <linux/device.h>
++#include <asm/pgalloc.h>
++#include <asm/uaccess.h>
++#include <asm/io.h>
++#include <asm/hypervisor.h>
++
++static inline int uncached_access(struct file *file)
++{
++ if (file->f_flags & O_SYNC)
++ return 1;
++ /* Xen sets correct MTRR type on non-RAM for us. */
++ return 0;
++}
++
++/*
++ * This funcion reads the *physical* memory. The f_pos points directly to the
++ * memory location.
++ */
++static ssize_t read_mem(struct file * file, char __user * buf,
++ size_t count, loff_t *ppos)
++{
++ unsigned long p = *ppos, ignored;
++ ssize_t read = 0, sz;
++ void __iomem *v;
++
++ while (count > 0) {
++ /*
++ * Handle first page in case it's not aligned
++ */
++ if (-p & (PAGE_SIZE - 1))
++ sz = -p & (PAGE_SIZE - 1);
++ else
++ sz = PAGE_SIZE;
++
++ sz = min_t(unsigned long, sz, count);
++
++ v = ioremap(p, sz);
++ if (IS_ERR(v) || v == NULL) {
++ /*
++ * Some programs (e.g., dmidecode) groove off into
++ * weird RAM areas where no tables can possibly exist
++ * (because Xen will have stomped on them!). These
++ * programs get rather upset if we let them know that
++ * Xen failed their access, so we fake out a read of
++ * all zeroes.
++ */
++ if (clear_user(buf, count))
++ return -EFAULT;
++ read += count;
++ break;
++ }
++
++ ignored = copy_to_user(buf, v, sz);
++ iounmap(v);
++ if (ignored)
++ return -EFAULT;
++ buf += sz;
++ p += sz;
++ count -= sz;
++ read += sz;
++ }
++
++ *ppos += read;
++ return read;
++}
++
++static ssize_t write_mem(struct file * file, const char __user * buf,
++ size_t count, loff_t *ppos)
++{
++ unsigned long p = *ppos, ignored;
++ ssize_t written = 0, sz;
++ void __iomem *v;
++
++ while (count > 0) {
++ /*
++ * Handle first page in case it's not aligned
++ */
++ if (-p & (PAGE_SIZE - 1))
++ sz = -p & (PAGE_SIZE - 1);
++ else
++ sz = PAGE_SIZE;
++
++ sz = min_t(unsigned long, sz, count);
++
++ v = ioremap(p, sz);
++ if (v == NULL)
++ break;
++ if (IS_ERR(v)) {
++ if (written == 0)
++ return PTR_ERR(v);
++ break;
++ }
++
++ ignored = copy_from_user(v, buf, sz);
++ iounmap(v);
++ if (ignored) {
++ written += sz - ignored;
++ if (written)
++ break;
++ return -EFAULT;
++ }
++ buf += sz;
++ p += sz;
++ count -= sz;
++ written += sz;
++ }
++
++ *ppos += written;
++ return written;
++}
++
++#ifndef ARCH_HAS_DEV_MEM_MMAP_MEM
++static int xen_mmap_mem(struct file * file, struct vm_area_struct * vma)
++{
++ size_t size = vma->vm_end - vma->vm_start;
++
++ if (uncached_access(file))
++ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
++
++ /* We want to return the real error code, not EAGAIN. */
++ return direct_remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
++ size, vma->vm_page_prot, DOMID_IO);
++}
++#endif
++
++/*
++ * The memory devices use the full 32/64 bits of the offset, and so we cannot
++ * check against negative addresses: they are ok. The return value is weird,
++ * though, in that case (0).
++ *
++ * also note that seeking relative to the "end of file" isn't supported:
++ * it has no meaning, so it returns -EINVAL.
++ */
++static loff_t memory_lseek(struct file * file, loff_t offset, int orig)
++{
++ loff_t ret;
++
++ mutex_lock(&file->f_dentry->d_inode->i_mutex);
++ switch (orig) {
++ case 0:
++ file->f_pos = offset;
++ ret = file->f_pos;
++ force_successful_syscall_return();
++ break;
++ case 1:
++ file->f_pos += offset;
++ ret = file->f_pos;
++ force_successful_syscall_return();
++ break;
++ default:
++ ret = -EINVAL;
++ }
++ mutex_unlock(&file->f_dentry->d_inode->i_mutex);
++ return ret;
++}
++
++static int open_mem(struct inode * inode, struct file * filp)
++{
++ return capable(CAP_SYS_RAWIO) ? 0 : -EPERM;
++}
++
++const struct file_operations mem_fops = {
++ .llseek = memory_lseek,
++ .read = read_mem,
++ .write = write_mem,
++ .mmap = xen_mmap_mem,
++ .open = open_mem,
++};
+--- linux-2.6.18.8/drivers/xen/console/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/console/Makefile 2008-05-19 00:33:45.762783940 +0300
+@@ -0,0 +1,2 @@
++
++obj-y := console.o xencons_ring.o
+--- linux-2.6.18.8/drivers/xen/console/console.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/console/console.c 2008-05-19 00:33:45.762783940 +0300
+@@ -0,0 +1,731 @@
++/******************************************************************************
++ * console.c
++ *
++ * Virtual console driver.
++ *
++ * Copyright (c) 2002-2004, K A Fraser.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/errno.h>
++#include <linux/signal.h>
++#include <linux/sched.h>
++#include <linux/interrupt.h>
++#include <linux/tty.h>
++#include <linux/tty_flip.h>
++#include <linux/serial.h>
++#include <linux/major.h>
++#include <linux/ptrace.h>
++#include <linux/ioport.h>
++#include <linux/mm.h>
++#include <linux/slab.h>
++#include <linux/init.h>
++#include <linux/console.h>
++#include <linux/bootmem.h>
++#include <linux/sysrq.h>
++#include <linux/screen_info.h>
++#include <linux/vt.h>
++#include <asm/io.h>
++#include <asm/irq.h>
++#include <asm/uaccess.h>
++#include <xen/interface/xen.h>
++#include <xen/interface/event_channel.h>
++#include <asm/hypervisor.h>
++#include <xen/evtchn.h>
++#include <xen/xenbus.h>
++#include <xen/xencons.h>
++
++/*
++ * Modes:
++ * 'xencons=off' [XC_OFF]: Console is disabled.
++ * 'xencons=tty' [XC_TTY]: Console attached to '/dev/tty[0-9]+'.
++ * 'xencons=ttyS' [XC_SERIAL]: Console attached to '/dev/ttyS[0-9]+'.
++ * 'xencons=xvc' [XC_XVC]: Console attached to '/dev/xvc0'.
++ * default: XC_XVC
++ *
++ * NB. In mode XC_TTY, we create dummy consoles for tty2-63. This suppresses
++ * warnings from standard distro startup scripts.
++ */
++static enum {
++ XC_OFF, XC_TTY, XC_SERIAL, XC_XVC
++} xc_mode = XC_XVC;
++static int xc_num = -1;
++
++/* /dev/xvc0 device number allocated by lanana.org. */
++#define XEN_XVC_MAJOR 204
++#define XEN_XVC_MINOR 191
++
++#ifdef CONFIG_MAGIC_SYSRQ
++static unsigned long sysrq_requested;
++extern int sysrq_enabled;
++#endif
++
++static int __init xencons_setup(char *str)
++{
++ char *q;
++ int n;
++ extern int console_use_vt;
++
++ console_use_vt = 1;
++ if (!strncmp(str, "ttyS", 4)) {
++ xc_mode = XC_SERIAL;
++ str += 4;
++ } else if (!strncmp(str, "tty", 3)) {
++ xc_mode = XC_TTY;
++ str += 3;
++ console_use_vt = 0;
++ } else if (!strncmp(str, "xvc", 3)) {
++ xc_mode = XC_XVC;
++ str += 3;
++ } else if (!strncmp(str, "off", 3)) {
++ xc_mode = XC_OFF;
++ str += 3;
++ }
++
++ n = simple_strtol(str, &q, 10);
++ if (q != str)
++ xc_num = n;
++
++ return 1;
++}
++__setup("xencons=", xencons_setup);
++
++/* The kernel and user-land drivers share a common transmit buffer. */
++static unsigned int wbuf_size = 4096;
++#define WBUF_MASK(_i) ((_i)&(wbuf_size-1))
++static char *wbuf;
++static unsigned int wc, wp; /* write_cons, write_prod */
++
++static int __init xencons_bufsz_setup(char *str)
++{
++ unsigned int goal;
++ goal = simple_strtoul(str, NULL, 0);
++ if (goal) {
++ goal = roundup_pow_of_two(goal);
++ if (wbuf_size < goal)
++ wbuf_size = goal;
++ }
++ return 1;
++}
++__setup("xencons_bufsz=", xencons_bufsz_setup);
++
++/* This lock protects accesses to the common transmit buffer. */
++static DEFINE_SPINLOCK(xencons_lock);
++
++/* Common transmit-kick routine. */
++static void __xencons_tx_flush(void);
++
++static struct tty_driver *xencons_driver;
++
++/******************** Kernel console driver ********************************/
++
++static void kcons_write(struct console *c, const char *s, unsigned int count)
++{
++ int i = 0;
++ unsigned long flags;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++
++ while (i < count) {
++ for (; i < count; i++) {
++ if ((wp - wc) >= (wbuf_size - 1))
++ break;
++ if ((wbuf[WBUF_MASK(wp++)] = s[i]) == '\n')
++ wbuf[WBUF_MASK(wp++)] = '\r';
++ }
++
++ __xencons_tx_flush();
++ }
++
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static void kcons_write_dom0(struct console *c, const char *s, unsigned int count)
++{
++
++ while (count > 0) {
++ int rc;
++ rc = HYPERVISOR_console_io( CONSOLEIO_write, count, (char *)s);
++ if (rc <= 0)
++ break;
++ count -= rc;
++ s += rc;
++ }
++}
++
++static struct tty_driver *kcons_device(struct console *c, int *index)
++{
++ *index = 0;
++ return xencons_driver;
++}
++
++static struct console kcons_info = {
++ .device = kcons_device,
++ .flags = CON_PRINTBUFFER | CON_ENABLED,
++ .index = -1,
++};
++
++static int __init xen_console_init(void)
++{
++ if (!is_running_on_xen())
++ goto out;
++
++ if (is_initial_xendomain()) {
++ kcons_info.write = kcons_write_dom0;
++ } else {
++ if (!xen_start_info->console.domU.evtchn)
++ goto out;
++ kcons_info.write = kcons_write;
++ }
++
++ switch (xc_mode) {
++ case XC_XVC:
++ strcpy(kcons_info.name, "xvc");
++ if (xc_num == -1)
++ xc_num = 0;
++ break;
++
++ case XC_SERIAL:
++ strcpy(kcons_info.name, "ttyS");
++ if (xc_num == -1)
++ xc_num = 0;
++ break;
++
++ case XC_TTY:
++ strcpy(kcons_info.name, "tty");
++ if (xc_num == -1)
++ xc_num = 1;
++ break;
++
++ default:
++ goto out;
++ }
++
++ wbuf = alloc_bootmem(wbuf_size);
++
++ register_console(&kcons_info);
++
++ out:
++ return 0;
++}
++console_initcall(xen_console_init);
++
++/*** Useful function for console debugging -- goes straight to Xen. ***/
++asmlinkage int xprintk(const char *fmt, ...)
++{
++ va_list args;
++ int printk_len;
++ static char printk_buf[1024];
++
++ /* Emit the output into the temporary buffer */
++ va_start(args, fmt);
++ printk_len = vsnprintf(printk_buf, sizeof(printk_buf), fmt, args);
++ va_end(args);
++
++ /* Send the processed output directly to Xen. */
++ kcons_write_dom0(NULL, printk_buf, printk_len);
++
++ return 0;
++}
++
++/*** Forcibly flush console data before dying. ***/
++void xencons_force_flush(void)
++{
++ int sz;
++
++ /* Emergency console is synchronous, so there's nothing to flush. */
++ if (!is_running_on_xen() ||
++ is_initial_xendomain() ||
++ !xen_start_info->console.domU.evtchn)
++ return;
++
++ /* Spin until console data is flushed through to the daemon. */
++ while (wc != wp) {
++ int sent = 0;
++ if ((sz = wp - wc) == 0)
++ continue;
++ sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
++ if (sent > 0)
++ wc += sent;
++ }
++}
++
++
++void __init dom0_init_screen_info(const struct dom0_vga_console_info *info, size_t size)
++{
++ /* This is drawn from a dump from vgacon:startup in
++ * standard Linux. */
++ screen_info.orig_video_mode = 3;
++ screen_info.orig_video_isVGA = 1;
++ screen_info.orig_video_lines = 25;
++ screen_info.orig_video_cols = 80;
++ screen_info.orig_video_ega_bx = 3;
++ screen_info.orig_video_points = 16;
++ screen_info.orig_y = screen_info.orig_video_lines - 1;
++
++ switch (info->video_type) {
++ case XEN_VGATYPE_TEXT_MODE_3:
++ if (size < offsetof(struct dom0_vga_console_info, u.text_mode_3)
++ + sizeof(info->u.text_mode_3))
++ break;
++ screen_info.orig_video_lines = info->u.text_mode_3.rows;
++ screen_info.orig_video_cols = info->u.text_mode_3.columns;
++ screen_info.orig_x = info->u.text_mode_3.cursor_x;
++ screen_info.orig_y = info->u.text_mode_3.cursor_y;
++ screen_info.orig_video_points =
++ info->u.text_mode_3.font_height;
++ break;
++
++ case XEN_VGATYPE_VESA_LFB:
++ if (size < offsetof(struct dom0_vga_console_info,
++ u.vesa_lfb.gbl_caps))
++ break;
++ screen_info.orig_video_isVGA = VIDEO_TYPE_VLFB;
++ screen_info.lfb_width = info->u.vesa_lfb.width;
++ screen_info.lfb_height = info->u.vesa_lfb.height;
++ screen_info.lfb_depth = info->u.vesa_lfb.bits_per_pixel;
++ screen_info.lfb_base = info->u.vesa_lfb.lfb_base;
++ screen_info.lfb_size = info->u.vesa_lfb.lfb_size;
++ screen_info.lfb_linelength = info->u.vesa_lfb.bytes_per_line;
++ screen_info.red_size = info->u.vesa_lfb.red_size;
++ screen_info.red_pos = info->u.vesa_lfb.red_pos;
++ screen_info.green_size = info->u.vesa_lfb.green_size;
++ screen_info.green_pos = info->u.vesa_lfb.green_pos;
++ screen_info.blue_size = info->u.vesa_lfb.blue_size;
++ screen_info.blue_pos = info->u.vesa_lfb.blue_pos;
++ screen_info.rsvd_size = info->u.vesa_lfb.rsvd_size;
++ screen_info.rsvd_pos = info->u.vesa_lfb.rsvd_pos;
++ if (size >= offsetof(struct dom0_vga_console_info,
++ u.vesa_lfb.gbl_caps)
++ + sizeof(info->u.vesa_lfb.gbl_caps))
++ screen_info.capabilities = info->u.vesa_lfb.gbl_caps;
++ if (size >= offsetof(struct dom0_vga_console_info,
++ u.vesa_lfb.mode_attrs)
++ + sizeof(info->u.vesa_lfb.mode_attrs))
++ screen_info.vesa_attributes = info->u.vesa_lfb.mode_attrs;
++ break;
++ }
++}
++
++
++/******************** User-space console driver (/dev/console) ************/
++
++#define DRV(_d) (_d)
++#define DUMMY_TTY(_tty) ((xc_mode == XC_TTY) && \
++ ((_tty)->index != (xc_num - 1)))
++
++static struct termios *xencons_termios[MAX_NR_CONSOLES];
++static struct termios *xencons_termios_locked[MAX_NR_CONSOLES];
++static struct tty_struct *xencons_tty;
++static int xencons_priv_irq;
++static char x_char;
++
++void xencons_rx(char *buf, unsigned len, struct pt_regs *regs)
++{
++ int i;
++ unsigned long flags;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ if (xencons_tty == NULL)
++ goto out;
++
++ for (i = 0; i < len; i++) {
++#ifdef CONFIG_MAGIC_SYSRQ
++ if (sysrq_enabled) {
++ if (buf[i] == '\x0f') { /* ^O */
++ if (!sysrq_requested) {
++ sysrq_requested = jiffies;
++ continue; /* don't print sysrq key */
++ }
++ sysrq_requested = 0;
++ } else if (sysrq_requested) {
++ unsigned long sysrq_timeout =
++ sysrq_requested + HZ*2;
++ sysrq_requested = 0;
++ if (time_before(jiffies, sysrq_timeout)) {
++ spin_unlock_irqrestore(
++ &xencons_lock, flags);
++ handle_sysrq(
++ buf[i], regs, xencons_tty);
++ spin_lock_irqsave(
++ &xencons_lock, flags);
++ continue;
++ }
++ }
++ }
++#endif
++ tty_insert_flip_char(xencons_tty, buf[i], 0);
++ }
++ tty_flip_buffer_push(xencons_tty);
++
++ out:
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static void __xencons_tx_flush(void)
++{
++ int sent, sz, work_done = 0;
++
++ if (x_char) {
++ if (is_initial_xendomain())
++ kcons_write_dom0(NULL, &x_char, 1);
++ else
++ while (x_char)
++ if (xencons_ring_send(&x_char, 1) == 1)
++ break;
++ x_char = 0;
++ work_done = 1;
++ }
++
++ while (wc != wp) {
++ sz = wp - wc;
++ if (sz > (wbuf_size - WBUF_MASK(wc)))
++ sz = wbuf_size - WBUF_MASK(wc);
++ if (is_initial_xendomain()) {
++ kcons_write_dom0(NULL, &wbuf[WBUF_MASK(wc)], sz);
++ wc += sz;
++ } else {
++ sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz);
++ if (sent == 0)
++ break;
++ wc += sent;
++ }
++ work_done = 1;
++ }
++
++ if (work_done && (xencons_tty != NULL)) {
++ wake_up_interruptible(&xencons_tty->write_wait);
++ if ((xencons_tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
++ (xencons_tty->ldisc.write_wakeup != NULL))
++ (xencons_tty->ldisc.write_wakeup)(xencons_tty);
++ }
++}
++
++void xencons_tx(void)
++{
++ unsigned long flags;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ __xencons_tx_flush();
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++/* Privileged receive callback and transmit kicker. */
++static irqreturn_t xencons_priv_interrupt(int irq, void *dev_id,
++ struct pt_regs *regs)
++{
++ static char rbuf[16];
++ int l;
++
++ while ((l = HYPERVISOR_console_io(CONSOLEIO_read, 16, rbuf)) > 0)
++ xencons_rx(rbuf, l, regs);
++
++ xencons_tx();
++
++ return IRQ_HANDLED;
++}
++
++static int xencons_write_room(struct tty_struct *tty)
++{
++ return wbuf_size - (wp - wc);
++}
++
++static int xencons_chars_in_buffer(struct tty_struct *tty)
++{
++ return wp - wc;
++}
++
++static void xencons_send_xchar(struct tty_struct *tty, char ch)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ x_char = ch;
++ __xencons_tx_flush();
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static void xencons_throttle(struct tty_struct *tty)
++{
++ if (DUMMY_TTY(tty))
++ return;
++
++ if (I_IXOFF(tty))
++ xencons_send_xchar(tty, STOP_CHAR(tty));
++}
++
++static void xencons_unthrottle(struct tty_struct *tty)
++{
++ if (DUMMY_TTY(tty))
++ return;
++
++ if (I_IXOFF(tty)) {
++ if (x_char != 0)
++ x_char = 0;
++ else
++ xencons_send_xchar(tty, START_CHAR(tty));
++ }
++}
++
++static void xencons_flush_buffer(struct tty_struct *tty)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ wc = wp = 0;
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static inline int __xencons_put_char(int ch)
++{
++ char _ch = (char)ch;
++ if ((wp - wc) == wbuf_size)
++ return 0;
++ wbuf[WBUF_MASK(wp++)] = _ch;
++ return 1;
++}
++
++static int xencons_write(
++ struct tty_struct *tty,
++ const unsigned char *buf,
++ int count)
++{
++ int i;
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return count;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++
++ for (i = 0; i < count; i++)
++ if (!__xencons_put_char(buf[i]))
++ break;
++
++ if (i != 0)
++ __xencons_tx_flush();
++
++ spin_unlock_irqrestore(&xencons_lock, flags);
++
++ return i;
++}
++
++static void xencons_put_char(struct tty_struct *tty, u_char ch)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ (void)__xencons_put_char(ch);
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static void xencons_flush_chars(struct tty_struct *tty)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ __xencons_tx_flush();
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static void xencons_wait_until_sent(struct tty_struct *tty, int timeout)
++{
++ unsigned long orig_jiffies = jiffies;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ while (DRV(tty->driver)->chars_in_buffer(tty)) {
++ set_current_state(TASK_INTERRUPTIBLE);
++ schedule_timeout(1);
++ if (signal_pending(current))
++ break;
++ if (timeout && time_after(jiffies, orig_jiffies + timeout))
++ break;
++ }
++
++ set_current_state(TASK_RUNNING);
++}
++
++static int xencons_open(struct tty_struct *tty, struct file *filp)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return 0;
++
++ spin_lock_irqsave(&xencons_lock, flags);
++ tty->driver_data = NULL;
++ if (xencons_tty == NULL)
++ xencons_tty = tty;
++ __xencons_tx_flush();
++ spin_unlock_irqrestore(&xencons_lock, flags);
++
++ return 0;
++}
++
++static void xencons_close(struct tty_struct *tty, struct file *filp)
++{
++ unsigned long flags;
++
++ if (DUMMY_TTY(tty))
++ return;
++
++ mutex_lock(&tty_mutex);
++
++ if (tty->count != 1) {
++ mutex_unlock(&tty_mutex);
++ return;
++ }
++
++ /* Prevent other threads from re-opening this tty. */
++ set_bit(TTY_CLOSING, &tty->flags);
++ mutex_unlock(&tty_mutex);
++
++ tty->closing = 1;
++ tty_wait_until_sent(tty, 0);
++ if (DRV(tty->driver)->flush_buffer != NULL)
++ DRV(tty->driver)->flush_buffer(tty);
++ if (tty->ldisc.flush_buffer != NULL)
++ tty->ldisc.flush_buffer(tty);
++ tty->closing = 0;
++ spin_lock_irqsave(&xencons_lock, flags);
++ xencons_tty = NULL;
++ spin_unlock_irqrestore(&xencons_lock, flags);
++}
++
++static struct tty_operations xencons_ops = {
++ .open = xencons_open,
++ .close = xencons_close,
++ .write = xencons_write,
++ .write_room = xencons_write_room,
++ .put_char = xencons_put_char,
++ .flush_chars = xencons_flush_chars,
++ .chars_in_buffer = xencons_chars_in_buffer,
++ .send_xchar = xencons_send_xchar,
++ .flush_buffer = xencons_flush_buffer,
++ .throttle = xencons_throttle,
++ .unthrottle = xencons_unthrottle,
++ .wait_until_sent = xencons_wait_until_sent,
++};
++
++static int __init xencons_init(void)
++{
++ int rc;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ if (xc_mode == XC_OFF)
++ return 0;
++
++ if (!is_initial_xendomain()) {
++ rc = xencons_ring_init();
++ if (rc)
++ return rc;
++ }
++
++ xencons_driver = alloc_tty_driver((xc_mode == XC_TTY) ?
++ MAX_NR_CONSOLES : 1);
++ if (xencons_driver == NULL)
++ return -ENOMEM;
++
++ DRV(xencons_driver)->name = "xencons";
++ DRV(xencons_driver)->major = TTY_MAJOR;
++ DRV(xencons_driver)->type = TTY_DRIVER_TYPE_SERIAL;
++ DRV(xencons_driver)->subtype = SERIAL_TYPE_NORMAL;
++ DRV(xencons_driver)->init_termios = tty_std_termios;
++ DRV(xencons_driver)->flags =
++ TTY_DRIVER_REAL_RAW |
++ TTY_DRIVER_RESET_TERMIOS;
++ DRV(xencons_driver)->termios = xencons_termios;
++ DRV(xencons_driver)->termios_locked = xencons_termios_locked;
++
++ switch (xc_mode) {
++ case XC_XVC:
++ DRV(xencons_driver)->name = "xvc";
++ DRV(xencons_driver)->major = XEN_XVC_MAJOR;
++ DRV(xencons_driver)->minor_start = XEN_XVC_MINOR;
++ DRV(xencons_driver)->name_base = xc_num;
++ break;
++ case XC_SERIAL:
++ DRV(xencons_driver)->name = "ttyS";
++ DRV(xencons_driver)->minor_start = 64 + xc_num;
++ DRV(xencons_driver)->name_base = xc_num;
++ break;
++ default:
++ DRV(xencons_driver)->name = "tty";
++ DRV(xencons_driver)->minor_start = 1;
++ DRV(xencons_driver)->name_base = 1;
++ break;
++ }
++
++ tty_set_operations(xencons_driver, &xencons_ops);
++
++ if ((rc = tty_register_driver(DRV(xencons_driver))) != 0) {
++ printk("WARNING: Failed to register Xen virtual "
++ "console driver as '%s%d'\n",
++ DRV(xencons_driver)->name,
++ DRV(xencons_driver)->name_base);
++ put_tty_driver(xencons_driver);
++ xencons_driver = NULL;
++ return rc;
++ }
++
++ if (is_initial_xendomain()) {
++ xencons_priv_irq = bind_virq_to_irqhandler(
++ VIRQ_CONSOLE,
++ 0,
++ xencons_priv_interrupt,
++ 0,
++ "console",
++ NULL);
++ BUG_ON(xencons_priv_irq < 0);
++ }
++
++ printk("Xen virtual console successfully installed as %s%d\n",
++ DRV(xencons_driver)->name, xc_num);
++
++ return 0;
++}
++
++module_init(xencons_init);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/console/xencons_ring.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/console/xencons_ring.c 2008-05-19 00:33:45.766784170 +0300
+@@ -0,0 +1,143 @@
++/*
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/errno.h>
++#include <linux/signal.h>
++#include <linux/sched.h>
++#include <linux/interrupt.h>
++#include <linux/tty.h>
++#include <linux/tty_flip.h>
++#include <linux/serial.h>
++#include <linux/major.h>
++#include <linux/ptrace.h>
++#include <linux/ioport.h>
++#include <linux/mm.h>
++#include <linux/slab.h>
++
++#include <asm/hypervisor.h>
++#include <xen/evtchn.h>
++#include <xen/xencons.h>
++#include <linux/wait.h>
++#include <linux/interrupt.h>
++#include <linux/sched.h>
++#include <linux/err.h>
++#include <xen/interface/io/console.h>
++
++static int xencons_irq;
++
++static inline struct xencons_interface *xencons_interface(void)
++{
++ return mfn_to_virt(xen_start_info->console.domU.mfn);
++}
++
++static inline void notify_daemon(void)
++{
++ /* Use evtchn: this is called early, before irq is set up. */
++ notify_remote_via_evtchn(xen_start_info->console.domU.evtchn);
++}
++
++int xencons_ring_send(const char *data, unsigned len)
++{
++ int sent = 0;
++ struct xencons_interface *intf = xencons_interface();
++ XENCONS_RING_IDX cons, prod;
++
++ cons = intf->out_cons;
++ prod = intf->out_prod;
++ mb();
++ BUG_ON((prod - cons) > sizeof(intf->out));
++
++ while ((sent < len) && ((prod - cons) < sizeof(intf->out)))
++ intf->out[MASK_XENCONS_IDX(prod++, intf->out)] = data[sent++];
++
++ wmb();
++ intf->out_prod = prod;
++
++ notify_daemon();
++
++ return sent;
++}
++
++static irqreturn_t handle_input(int irq, void *unused, struct pt_regs *regs)
++{
++ struct xencons_interface *intf = xencons_interface();
++ XENCONS_RING_IDX cons, prod;
++
++ cons = intf->in_cons;
++ prod = intf->in_prod;
++ mb();
++ BUG_ON((prod - cons) > sizeof(intf->in));
++
++ while (cons != prod) {
++ xencons_rx(intf->in+MASK_XENCONS_IDX(cons,intf->in), 1, regs);
++ cons++;
++ }
++
++ mb();
++ intf->in_cons = cons;
++
++ notify_daemon();
++
++ xencons_tx();
++
++ return IRQ_HANDLED;
++}
++
++int xencons_ring_init(void)
++{
++ int irq;
++
++ if (xencons_irq)
++ unbind_from_irqhandler(xencons_irq, NULL);
++ xencons_irq = 0;
++
++ if (!is_running_on_xen() ||
++ is_initial_xendomain() ||
++ !xen_start_info->console.domU.evtchn)
++ return -ENODEV;
++
++ irq = bind_caller_port_to_irqhandler(
++ xen_start_info->console.domU.evtchn,
++ handle_input, 0, "xencons", NULL);
++ if (irq < 0) {
++ printk(KERN_ERR "XEN console request irq failed %i\n", irq);
++ return irq;
++ }
++
++ xencons_irq = irq;
++
++ /* In case we have in-flight data after save/restore... */
++ notify_daemon();
++
++ return 0;
++}
++
++void xencons_resume(void)
++{
++ (void)xencons_ring_init();
++}
+--- linux-2.6.18.8/drivers/xen/core/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/Makefile 2008-05-19 00:33:45.766784170 +0300
+@@ -0,0 +1,13 @@
++#
++# Makefile for the linux kernel.
++#
++
++obj-y := evtchn.o gnttab.o features.o reboot.o machine_reboot.o firmware.o
++
++obj-$(CONFIG_PROC_FS) += xen_proc.o
++obj-$(CONFIG_SYS_HYPERVISOR) += hypervisor_sysfs.o
++obj-$(CONFIG_HOTPLUG_CPU) += cpu_hotplug.o
++obj-$(CONFIG_XEN_SYSFS) += xen_sysfs.o
++obj-$(CONFIG_XEN_SMPBOOT) += smpboot.o
++obj-$(CONFIG_KEXEC) += machine_kexec.o
++obj-$(CONFIG_XEN_XENCOMM) += xencomm.o
+--- linux-2.6.18.8/drivers/xen/core/cpu_hotplug.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/cpu_hotplug.c 2008-05-19 00:33:45.766784170 +0300
+@@ -0,0 +1,173 @@
++#include <linux/init.h>
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/notifier.h>
++#include <linux/cpu.h>
++#include <xen/cpu_hotplug.h>
++#include <xen/xenbus.h>
++
++/*
++ * Set of CPUs that remote admin software will allow us to bring online.
++ * Notified to us via xenbus.
++ */
++static cpumask_t xenbus_allowed_cpumask;
++
++/* Set of CPUs that local admin will allow us to bring online. */
++static cpumask_t local_allowed_cpumask = CPU_MASK_ALL;
++
++static int local_cpu_hotplug_request(void)
++{
++ /*
++ * We assume a CPU hotplug request comes from local admin if it is made
++ * via a userspace process (i.e., one with a real mm_struct).
++ */
++ return (current->mm != NULL);
++}
++
++static void vcpu_hotplug(unsigned int cpu)
++{
++ int err;
++ char dir[32], state[32];
++
++ if ((cpu >= NR_CPUS) || !cpu_possible(cpu))
++ return;
++
++ sprintf(dir, "cpu/%u", cpu);
++ err = xenbus_scanf(XBT_NIL, dir, "availability", "%s", state);
++ if (err != 1) {
++ printk(KERN_ERR "XENBUS: Unable to read cpu state\n");
++ return;
++ }
++
++ if (strcmp(state, "online") == 0) {
++ cpu_set(cpu, xenbus_allowed_cpumask);
++ (void)cpu_up(cpu);
++ } else if (strcmp(state, "offline") == 0) {
++ cpu_clear(cpu, xenbus_allowed_cpumask);
++ (void)cpu_down(cpu);
++ } else {
++ printk(KERN_ERR "XENBUS: unknown state(%s) on CPU%d\n",
++ state, cpu);
++ }
++}
++
++static void handle_vcpu_hotplug_event(
++ struct xenbus_watch *watch, const char **vec, unsigned int len)
++{
++ unsigned int cpu;
++ char *cpustr;
++ const char *node = vec[XS_WATCH_PATH];
++
++ if ((cpustr = strstr(node, "cpu/")) != NULL) {
++ sscanf(cpustr, "cpu/%u", &cpu);
++ vcpu_hotplug(cpu);
++ }
++}
++
++static int smpboot_cpu_notify(struct notifier_block *notifier,
++ unsigned long action, void *hcpu)
++{
++ unsigned int cpu = (long)hcpu;
++
++ /*
++ * We do this in a callback notifier rather than __cpu_disable()
++ * because local_cpu_hotplug_request() does not work in the latter
++ * as it's always executed from within a stopmachine kthread.
++ */
++ if ((action == CPU_DOWN_PREPARE) && local_cpu_hotplug_request())
++ cpu_clear(cpu, local_allowed_cpumask);
++
++ return NOTIFY_OK;
++}
++
++static int setup_cpu_watcher(struct notifier_block *notifier,
++ unsigned long event, void *data)
++{
++ unsigned int i;
++
++ static struct xenbus_watch cpu_watch = {
++ .node = "cpu",
++ .callback = handle_vcpu_hotplug_event,
++ .flags = XBWF_new_thread };
++ (void)register_xenbus_watch(&cpu_watch);
++
++ if (!is_initial_xendomain()) {
++ for_each_possible_cpu(i)
++ vcpu_hotplug(i);
++ printk(KERN_INFO "Brought up %ld CPUs\n",
++ (long)num_online_cpus());
++ }
++
++ return NOTIFY_DONE;
++}
++
++static int __init setup_vcpu_hotplug_event(void)
++{
++ static struct notifier_block hotplug_cpu = {
++ .notifier_call = smpboot_cpu_notify };
++ static struct notifier_block xsn_cpu = {
++ .notifier_call = setup_cpu_watcher };
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ register_cpu_notifier(&hotplug_cpu);
++ register_xenstore_notifier(&xsn_cpu);
++
++ return 0;
++}
++
++arch_initcall(setup_vcpu_hotplug_event);
++
++int smp_suspend(void)
++{
++ unsigned int cpu;
++ int err;
++
++ for_each_online_cpu(cpu) {
++ if (cpu == 0)
++ continue;
++ err = cpu_down(cpu);
++ if (err) {
++ printk(KERN_CRIT "Failed to take all CPUs "
++ "down: %d.\n", err);
++ for_each_possible_cpu(cpu)
++ vcpu_hotplug(cpu);
++ return err;
++ }
++ }
++
++ return 0;
++}
++
++void smp_resume(void)
++{
++ unsigned int cpu;
++
++ for_each_possible_cpu(cpu)
++ vcpu_hotplug(cpu);
++}
++
++int cpu_up_check(unsigned int cpu)
++{
++ int rc = 0;
++
++ if (local_cpu_hotplug_request()) {
++ cpu_set(cpu, local_allowed_cpumask);
++ if (!cpu_isset(cpu, xenbus_allowed_cpumask)) {
++ printk("%s: attempt to bring up CPU %u disallowed by "
++ "remote admin.\n", __FUNCTION__, cpu);
++ rc = -EBUSY;
++ }
++ } else if (!cpu_isset(cpu, local_allowed_cpumask) ||
++ !cpu_isset(cpu, xenbus_allowed_cpumask)) {
++ rc = -EBUSY;
++ }
++
++ return rc;
++}
++
++void init_xenbus_allowed_cpumask(void)
++{
++ xenbus_allowed_cpumask = cpu_present_map;
++}
+--- linux-2.6.18.8/drivers/xen/core/evtchn.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/evtchn.c 2008-05-19 00:33:45.770784401 +0300
+@@ -0,0 +1,1061 @@
++/******************************************************************************
++ * evtchn.c
++ *
++ * Communication via Xen event channels.
++ *
++ * Copyright (c) 2002-2005, K A Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/module.h>
++#include <linux/irq.h>
++#include <linux/interrupt.h>
++#include <linux/sched.h>
++#include <linux/kernel_stat.h>
++#include <linux/version.h>
++#include <asm/atomic.h>
++#include <asm/system.h>
++#include <asm/ptrace.h>
++#include <asm/synch_bitops.h>
++#include <xen/evtchn.h>
++#include <xen/interface/event_channel.h>
++#include <xen/interface/physdev.h>
++#include <asm/hypervisor.h>
++#include <linux/mc146818rtc.h> /* RTC_IRQ */
++
++/*
++ * This lock protects updates to the following mapping and reference-count
++ * arrays. The lock does not need to be acquired to read the mapping tables.
++ */
++static DEFINE_SPINLOCK(irq_mapping_update_lock);
++
++/* IRQ <-> event-channel mappings. */
++static int evtchn_to_irq[NR_EVENT_CHANNELS] = {
++ [0 ... NR_EVENT_CHANNELS-1] = -1 };
++
++/* Packed IRQ information: binding type, sub-type index, and event channel. */
++static u32 irq_info[NR_IRQS];
++
++/* Binding types. */
++enum {
++ IRQT_UNBOUND,
++ IRQT_PIRQ,
++ IRQT_VIRQ,
++ IRQT_IPI,
++ IRQT_LOCAL_PORT,
++ IRQT_CALLER_PORT
++};
++
++/* Constructor for packed IRQ information. */
++static inline u32 mk_irq_info(u32 type, u32 index, u32 evtchn)
++{
++ return ((type << 24) | (index << 16) | evtchn);
++}
++
++/* Convenient shorthand for packed representation of an unbound IRQ. */
++#define IRQ_UNBOUND mk_irq_info(IRQT_UNBOUND, 0, 0)
++
++/*
++ * Accessors for packed IRQ information.
++ */
++
++static inline unsigned int evtchn_from_irq(int irq)
++{
++ return (u16)(irq_info[irq]);
++}
++
++static inline unsigned int index_from_irq(int irq)
++{
++ return (u8)(irq_info[irq] >> 16);
++}
++
++static inline unsigned int type_from_irq(int irq)
++{
++ return (u8)(irq_info[irq] >> 24);
++}
++
++/* IRQ <-> VIRQ mapping. */
++DEFINE_PER_CPU(int, virq_to_irq[NR_VIRQS]) = {[0 ... NR_VIRQS-1] = -1};
++
++/* IRQ <-> IPI mapping. */
++#ifndef NR_IPIS
++#define NR_IPIS 1
++#endif
++DEFINE_PER_CPU(int, ipi_to_irq[NR_IPIS]) = {[0 ... NR_IPIS-1] = -1};
++
++/* Reference counts for bindings to IRQs. */
++static int irq_bindcount[NR_IRQS];
++
++/* Bitmap indicating which PIRQs require Xen to be notified on unmask. */
++static DECLARE_BITMAP(pirq_needs_eoi, NR_PIRQS);
++
++#ifdef CONFIG_SMP
++
++static u8 cpu_evtchn[NR_EVENT_CHANNELS];
++static unsigned long cpu_evtchn_mask[NR_CPUS][NR_EVENT_CHANNELS/BITS_PER_LONG];
++
++static inline unsigned long active_evtchns(unsigned int cpu, shared_info_t *sh,
++ unsigned int idx)
++{
++ return (sh->evtchn_pending[idx] &
++ cpu_evtchn_mask[cpu][idx] &
++ ~sh->evtchn_mask[idx]);
++}
++
++static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
++{
++ shared_info_t *s = HYPERVISOR_shared_info;
++ int irq = evtchn_to_irq[chn];
++
++ BUG_ON(!test_bit(chn, s->evtchn_mask));
++
++ if (irq != -1)
++ set_native_irq_info(irq, cpumask_of_cpu(cpu));
++
++ clear_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu_evtchn[chn]]);
++ set_bit(chn, (unsigned long *)cpu_evtchn_mask[cpu]);
++ cpu_evtchn[chn] = cpu;
++}
++
++static void init_evtchn_cpu_bindings(void)
++{
++ int i;
++
++ /* By default all event channels notify CPU#0. */
++ for (i = 0; i < NR_IRQS; i++)
++ set_native_irq_info(i, cpumask_of_cpu(0));
++
++ memset(cpu_evtchn, 0, sizeof(cpu_evtchn));
++ memset(cpu_evtchn_mask[0], ~0, sizeof(cpu_evtchn_mask[0]));
++}
++
++static inline unsigned int cpu_from_evtchn(unsigned int evtchn)
++{
++ return cpu_evtchn[evtchn];
++}
++
++#else
++
++static inline unsigned long active_evtchns(unsigned int cpu, shared_info_t *sh,
++ unsigned int idx)
++{
++ return (sh->evtchn_pending[idx] & ~sh->evtchn_mask[idx]);
++}
++
++static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu)
++{
++}
++
++static void init_evtchn_cpu_bindings(void)
++{
++}
++
++static inline unsigned int cpu_from_evtchn(unsigned int evtchn)
++{
++ return 0;
++}
++
++#endif
++
++/* Upcall to generic IRQ layer. */
++#ifdef CONFIG_X86
++extern fastcall unsigned int do_IRQ(struct pt_regs *regs);
++void __init xen_init_IRQ(void);
++void __init init_IRQ(void)
++{
++ irq_ctx_init(0);
++ xen_init_IRQ();
++}
++#if defined (__i386__)
++static inline void exit_idle(void) {}
++#define IRQ_REG orig_eax
++#elif defined (__x86_64__)
++#include <asm/idle.h>
++#define IRQ_REG orig_rax
++#endif
++#define do_IRQ(irq, regs) do { \
++ (regs)->IRQ_REG = ~(irq); \
++ do_IRQ((regs)); \
++} while (0)
++#endif
++
++/* Xen will never allocate port zero for any purpose. */
++#define VALID_EVTCHN(chn) ((chn) != 0)
++
++/*
++ * Force a proper event-channel callback from Xen after clearing the
++ * callback mask. We do this in a very simple manner, by making a call
++ * down into Xen. The pending flag will be checked by Xen on return.
++ */
++void force_evtchn_callback(void)
++{
++ VOID(HYPERVISOR_xen_version(0, NULL));
++}
++/* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */
++EXPORT_SYMBOL(force_evtchn_callback);
++
++static DEFINE_PER_CPU(unsigned int, upcall_count) = { 0 };
++static DEFINE_PER_CPU(unsigned int, last_processed_l1i) = { BITS_PER_LONG - 1 };
++static DEFINE_PER_CPU(unsigned int, last_processed_l2i) = { BITS_PER_LONG - 1 };
++
++/* NB. Interrupts are disabled on entry. */
++asmlinkage void evtchn_do_upcall(struct pt_regs *regs)
++{
++ unsigned long l1, l2;
++ unsigned long masked_l1, masked_l2;
++ unsigned int l1i, l2i, port, count;
++ int irq;
++ unsigned int cpu = smp_processor_id();
++ shared_info_t *s = HYPERVISOR_shared_info;
++ vcpu_info_t *vcpu_info = &s->vcpu_info[cpu];
++
++
++ do {
++ /* Avoid a callback storm when we reenable delivery. */
++ vcpu_info->evtchn_upcall_pending = 0;
++
++ /* Nested invocations bail immediately. */
++ if (unlikely(per_cpu(upcall_count, cpu)++))
++ return;
++
++#ifndef CONFIG_X86 /* No need for a barrier -- XCHG is a barrier on x86. */
++ /* Clear master flag /before/ clearing selector flag. */
++ wmb();
++#endif
++ l1 = xchg(&vcpu_info->evtchn_pending_sel, 0);
++
++ l1i = per_cpu(last_processed_l1i, cpu);
++ l2i = per_cpu(last_processed_l2i, cpu);
++
++ while (l1 != 0) {
++
++ l1i = (l1i + 1) % BITS_PER_LONG;
++ masked_l1 = l1 & ((~0UL) << l1i);
++
++ if (masked_l1 == 0) { /* if we masked out all events, wrap around to the beginning */
++ l1i = BITS_PER_LONG - 1;
++ l2i = BITS_PER_LONG - 1;
++ continue;
++ }
++ l1i = __ffs(masked_l1);
++
++ do {
++ l2 = active_evtchns(cpu, s, l1i);
++
++ l2i = (l2i + 1) % BITS_PER_LONG;
++ masked_l2 = l2 & ((~0UL) << l2i);
++
++ if (masked_l2 == 0) { /* if we masked out all events, move on */
++ l2i = BITS_PER_LONG - 1;
++ break;
++ }
++
++ l2i = __ffs(masked_l2);
++
++ /* process port */
++ port = (l1i * BITS_PER_LONG) + l2i;
++ if ((irq = evtchn_to_irq[port]) != -1)
++ do_IRQ(irq, regs);
++ else {
++ exit_idle();
++ evtchn_device_upcall(port);
++ }
++
++ /* if this is the final port processed, we'll pick up here+1 next time */
++ per_cpu(last_processed_l1i, cpu) = l1i;
++ per_cpu(last_processed_l2i, cpu) = l2i;
++
++ } while (l2i != BITS_PER_LONG - 1);
++
++ l2 = active_evtchns(cpu, s, l1i);
++ if (l2 == 0) /* we handled all ports, so we can clear the selector bit */
++ l1 &= ~(1UL << l1i);
++
++ }
++
++ /* If there were nested callbacks then we have more to do. */
++ count = per_cpu(upcall_count, cpu);
++ per_cpu(upcall_count, cpu) = 0;
++ } while (unlikely(count != 1));
++}
++
++static int find_unbound_irq(void)
++{
++ static int warned;
++ int dynirq, irq;
++
++ for (dynirq = 0; dynirq < NR_DYNIRQS; dynirq++) {
++ irq = dynirq_to_irq(dynirq);
++ if (irq_bindcount[irq] == 0)
++ return irq;
++ }
++
++ if (!warned) {
++ warned = 1;
++ printk(KERN_WARNING "No available IRQ to bind to: "
++ "increase NR_DYNIRQS.\n");
++ }
++
++ return -ENOSPC;
++}
++
++static int bind_caller_port_to_irq(unsigned int caller_port)
++{
++ int irq;
++
++ spin_lock(&irq_mapping_update_lock);
++
++ if ((irq = evtchn_to_irq[caller_port]) == -1) {
++ if ((irq = find_unbound_irq()) < 0)
++ goto out;
++
++ evtchn_to_irq[caller_port] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_CALLER_PORT, 0, caller_port);
++ }
++
++ irq_bindcount[irq]++;
++
++ out:
++ spin_unlock(&irq_mapping_update_lock);
++ return irq;
++}
++
++static int bind_local_port_to_irq(unsigned int local_port)
++{
++ int irq;
++
++ spin_lock(&irq_mapping_update_lock);
++
++ BUG_ON(evtchn_to_irq[local_port] != -1);
++
++ if ((irq = find_unbound_irq()) < 0) {
++ struct evtchn_close close = { .port = local_port };
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close))
++ BUG();
++ goto out;
++ }
++
++ evtchn_to_irq[local_port] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_LOCAL_PORT, 0, local_port);
++ irq_bindcount[irq]++;
++
++ out:
++ spin_unlock(&irq_mapping_update_lock);
++ return irq;
++}
++
++static int bind_listening_port_to_irq(unsigned int remote_domain)
++{
++ struct evtchn_alloc_unbound alloc_unbound;
++ int err;
++
++ alloc_unbound.dom = DOMID_SELF;
++ alloc_unbound.remote_dom = remote_domain;
++
++ err = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
++ &alloc_unbound);
++
++ return err ? : bind_local_port_to_irq(alloc_unbound.port);
++}
++
++static int bind_interdomain_evtchn_to_irq(unsigned int remote_domain,
++ unsigned int remote_port)
++{
++ struct evtchn_bind_interdomain bind_interdomain;
++ int err;
++
++ bind_interdomain.remote_dom = remote_domain;
++ bind_interdomain.remote_port = remote_port;
++
++ err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
++ &bind_interdomain);
++
++ return err ? : bind_local_port_to_irq(bind_interdomain.local_port);
++}
++
++static int bind_virq_to_irq(unsigned int virq, unsigned int cpu)
++{
++ struct evtchn_bind_virq bind_virq;
++ int evtchn, irq;
++
++ spin_lock(&irq_mapping_update_lock);
++
++ if ((irq = per_cpu(virq_to_irq, cpu)[virq]) == -1) {
++ if ((irq = find_unbound_irq()) < 0)
++ goto out;
++
++ bind_virq.virq = virq;
++ bind_virq.vcpu = cpu;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
++ &bind_virq) != 0)
++ BUG();
++ evtchn = bind_virq.port;
++
++ evtchn_to_irq[evtchn] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_VIRQ, virq, evtchn);
++
++ per_cpu(virq_to_irq, cpu)[virq] = irq;
++
++ bind_evtchn_to_cpu(evtchn, cpu);
++ }
++
++ irq_bindcount[irq]++;
++
++ out:
++ spin_unlock(&irq_mapping_update_lock);
++ return irq;
++}
++
++static int bind_ipi_to_irq(unsigned int ipi, unsigned int cpu)
++{
++ struct evtchn_bind_ipi bind_ipi;
++ int evtchn, irq;
++
++ spin_lock(&irq_mapping_update_lock);
++
++ if ((irq = per_cpu(ipi_to_irq, cpu)[ipi]) == -1) {
++ if ((irq = find_unbound_irq()) < 0)
++ goto out;
++
++ bind_ipi.vcpu = cpu;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
++ &bind_ipi) != 0)
++ BUG();
++ evtchn = bind_ipi.port;
++
++ evtchn_to_irq[evtchn] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_IPI, ipi, evtchn);
++
++ per_cpu(ipi_to_irq, cpu)[ipi] = irq;
++
++ bind_evtchn_to_cpu(evtchn, cpu);
++ }
++
++ irq_bindcount[irq]++;
++
++ out:
++ spin_unlock(&irq_mapping_update_lock);
++ return irq;
++}
++
++static void unbind_from_irq(unsigned int irq)
++{
++ struct evtchn_close close;
++ unsigned int cpu;
++ int evtchn = evtchn_from_irq(irq);
++
++ spin_lock(&irq_mapping_update_lock);
++
++ if ((--irq_bindcount[irq] == 0) && VALID_EVTCHN(evtchn)) {
++ close.port = evtchn;
++ if ((type_from_irq(irq) != IRQT_CALLER_PORT) &&
++ HYPERVISOR_event_channel_op(EVTCHNOP_close, &close))
++ BUG();
++
++ switch (type_from_irq(irq)) {
++ case IRQT_VIRQ:
++ per_cpu(virq_to_irq, cpu_from_evtchn(evtchn))
++ [index_from_irq(irq)] = -1;
++ break;
++ case IRQT_IPI:
++ per_cpu(ipi_to_irq, cpu_from_evtchn(evtchn))
++ [index_from_irq(irq)] = -1;
++ break;
++ default:
++ break;
++ }
++
++ /* Closed ports are implicitly re-bound to VCPU0. */
++ bind_evtchn_to_cpu(evtchn, 0);
++
++ evtchn_to_irq[evtchn] = -1;
++ irq_info[irq] = IRQ_UNBOUND;
++
++ /* Zap stats across IRQ changes of use. */
++ for_each_possible_cpu(cpu)
++ kstat_cpu(cpu).irqs[irq] = 0;
++ }
++
++ spin_unlock(&irq_mapping_update_lock);
++}
++
++int bind_caller_port_to_irqhandler(
++ unsigned int caller_port,
++ irqreturn_t (*handler)(int, void *, struct pt_regs *),
++ unsigned long irqflags,
++ const char *devname,
++ void *dev_id)
++{
++ int irq, retval;
++
++ irq = bind_caller_port_to_irq(caller_port);
++ if (irq < 0)
++ return irq;
++
++ retval = request_irq(irq, handler, irqflags, devname, dev_id);
++ if (retval != 0) {
++ unbind_from_irq(irq);
++ return retval;
++ }
++
++ return irq;
++}
++EXPORT_SYMBOL_GPL(bind_caller_port_to_irqhandler);
++
++int bind_listening_port_to_irqhandler(
++ unsigned int remote_domain,
++ irqreturn_t (*handler)(int, void *, struct pt_regs *),
++ unsigned long irqflags,
++ const char *devname,
++ void *dev_id)
++{
++ int irq, retval;
++
++ irq = bind_listening_port_to_irq(remote_domain);
++ if (irq < 0)
++ return irq;
++
++ retval = request_irq(irq, handler, irqflags, devname, dev_id);
++ if (retval != 0) {
++ unbind_from_irq(irq);
++ return retval;
++ }
++
++ return irq;
++}
++EXPORT_SYMBOL_GPL(bind_listening_port_to_irqhandler);
++
++int bind_interdomain_evtchn_to_irqhandler(
++ unsigned int remote_domain,
++ unsigned int remote_port,
++ irqreturn_t (*handler)(int, void *, struct pt_regs *),
++ unsigned long irqflags,
++ const char *devname,
++ void *dev_id)
++{
++ int irq, retval;
++
++ irq = bind_interdomain_evtchn_to_irq(remote_domain, remote_port);
++ if (irq < 0)
++ return irq;
++
++ retval = request_irq(irq, handler, irqflags, devname, dev_id);
++ if (retval != 0) {
++ unbind_from_irq(irq);
++ return retval;
++ }
++
++ return irq;
++}
++EXPORT_SYMBOL_GPL(bind_interdomain_evtchn_to_irqhandler);
++
++int bind_virq_to_irqhandler(
++ unsigned int virq,
++ unsigned int cpu,
++ irqreturn_t (*handler)(int, void *, struct pt_regs *),
++ unsigned long irqflags,
++ const char *devname,
++ void *dev_id)
++{
++ int irq, retval;
++
++ irq = bind_virq_to_irq(virq, cpu);
++ if (irq < 0)
++ return irq;
++
++ retval = request_irq(irq, handler, irqflags, devname, dev_id);
++ if (retval != 0) {
++ unbind_from_irq(irq);
++ return retval;
++ }
++
++ return irq;
++}
++EXPORT_SYMBOL_GPL(bind_virq_to_irqhandler);
++
++int bind_ipi_to_irqhandler(
++ unsigned int ipi,
++ unsigned int cpu,
++ irqreturn_t (*handler)(int, void *, struct pt_regs *),
++ unsigned long irqflags,
++ const char *devname,
++ void *dev_id)
++{
++ int irq, retval;
++
++ irq = bind_ipi_to_irq(ipi, cpu);
++ if (irq < 0)
++ return irq;
++
++ retval = request_irq(irq, handler, irqflags, devname, dev_id);
++ if (retval != 0) {
++ unbind_from_irq(irq);
++ return retval;
++ }
++
++ return irq;
++}
++EXPORT_SYMBOL_GPL(bind_ipi_to_irqhandler);
++
++void unbind_from_irqhandler(unsigned int irq, void *dev_id)
++{
++ free_irq(irq, dev_id);
++ unbind_from_irq(irq);
++}
++EXPORT_SYMBOL_GPL(unbind_from_irqhandler);
++
++#ifdef CONFIG_SMP
++void rebind_evtchn_to_cpu(int port, unsigned int cpu)
++{
++ struct evtchn_bind_vcpu ebv = { .port = port, .vcpu = cpu };
++ int masked;
++
++ masked = test_and_set_evtchn_mask(port);
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_vcpu, &ebv) == 0)
++ bind_evtchn_to_cpu(port, cpu);
++ if (!masked)
++ unmask_evtchn(port);
++}
++
++static void rebind_irq_to_cpu(unsigned int irq, unsigned int tcpu)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ rebind_evtchn_to_cpu(evtchn, tcpu);
++}
++
++static void set_affinity_irq(unsigned int irq, cpumask_t dest)
++{
++ unsigned tcpu = first_cpu(dest);
++ rebind_irq_to_cpu(irq, tcpu);
++}
++#endif
++
++int resend_irq_on_evtchn(unsigned int irq)
++{
++ int masked, evtchn = evtchn_from_irq(irq);
++ shared_info_t *s = HYPERVISOR_shared_info;
++
++ if (!VALID_EVTCHN(evtchn))
++ return 1;
++
++ masked = test_and_set_evtchn_mask(evtchn);
++ synch_set_bit(evtchn, s->evtchn_pending);
++ if (!masked)
++ unmask_evtchn(evtchn);
++
++ return 1;
++}
++
++/*
++ * Interface to generic handling in irq.c
++ */
++
++static unsigned int startup_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ unmask_evtchn(evtchn);
++ return 0;
++}
++
++static void shutdown_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ mask_evtchn(evtchn);
++}
++
++static void enable_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ unmask_evtchn(evtchn);
++}
++
++static void disable_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ mask_evtchn(evtchn);
++}
++
++static void ack_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ move_native_irq(irq);
++
++ if (VALID_EVTCHN(evtchn)) {
++ mask_evtchn(evtchn);
++ clear_evtchn(evtchn);
++ }
++}
++
++static void end_dynirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn) && !(irq_desc[irq].status & IRQ_DISABLED))
++ unmask_evtchn(evtchn);
++}
++
++static struct hw_interrupt_type dynirq_type = {
++ .typename = "Dynamic-irq",
++ .startup = startup_dynirq,
++ .shutdown = shutdown_dynirq,
++ .enable = enable_dynirq,
++ .disable = disable_dynirq,
++ .ack = ack_dynirq,
++ .end = end_dynirq,
++#ifdef CONFIG_SMP
++ .set_affinity = set_affinity_irq,
++#endif
++ .retrigger = resend_irq_on_evtchn,
++};
++
++static inline void pirq_unmask_notify(int pirq)
++{
++ struct physdev_eoi eoi = { .irq = pirq };
++ if (unlikely(test_bit(pirq, pirq_needs_eoi)))
++ VOID(HYPERVISOR_physdev_op(PHYSDEVOP_eoi, &eoi));
++}
++
++static inline void pirq_query_unmask(int pirq)
++{
++ struct physdev_irq_status_query irq_status;
++ irq_status.irq = pirq;
++ if (HYPERVISOR_physdev_op(PHYSDEVOP_irq_status_query, &irq_status))
++ irq_status.flags = 0;
++ clear_bit(pirq, pirq_needs_eoi);
++ if (irq_status.flags & XENIRQSTAT_needs_eoi)
++ set_bit(pirq, pirq_needs_eoi);
++}
++
++/*
++ * On startup, if there is no action associated with the IRQ then we are
++ * probing. In this case we should not share with others as it will confuse us.
++ */
++#define probing_irq(_irq) (irq_desc[(_irq)].action == NULL)
++
++static unsigned int startup_pirq(unsigned int irq)
++{
++ struct evtchn_bind_pirq bind_pirq;
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ goto out;
++
++ bind_pirq.pirq = irq;
++ /* NB. We are happy to share unless we are probing. */
++ bind_pirq.flags = probing_irq(irq) ? 0 : BIND_PIRQ__WILL_SHARE;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_pirq, &bind_pirq) != 0) {
++ if (!probing_irq(irq))
++ printk(KERN_INFO "Failed to obtain physical IRQ %d\n",
++ irq);
++ return 0;
++ }
++ evtchn = bind_pirq.port;
++
++ pirq_query_unmask(irq_to_pirq(irq));
++
++ evtchn_to_irq[evtchn] = irq;
++ bind_evtchn_to_cpu(evtchn, 0);
++ irq_info[irq] = mk_irq_info(IRQT_PIRQ, irq, evtchn);
++
++ out:
++ unmask_evtchn(evtchn);
++ pirq_unmask_notify(irq_to_pirq(irq));
++
++ return 0;
++}
++
++static void shutdown_pirq(unsigned int irq)
++{
++ struct evtchn_close close;
++ int evtchn = evtchn_from_irq(irq);
++
++ if (!VALID_EVTCHN(evtchn))
++ return;
++
++ mask_evtchn(evtchn);
++
++ close.port = evtchn;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
++ BUG();
++
++ bind_evtchn_to_cpu(evtchn, 0);
++ evtchn_to_irq[evtchn] = -1;
++ irq_info[irq] = IRQ_UNBOUND;
++}
++
++static void enable_pirq(unsigned int irq)
++{
++ startup_pirq(irq);
++}
++
++static void disable_pirq(unsigned int irq)
++{
++}
++
++static void ack_pirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ move_native_irq(irq);
++
++ if (VALID_EVTCHN(evtchn)) {
++ mask_evtchn(evtchn);
++ clear_evtchn(evtchn);
++ }
++}
++
++static void end_pirq(unsigned int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if ((irq_desc[irq].status & (IRQ_DISABLED|IRQ_PENDING)) ==
++ (IRQ_DISABLED|IRQ_PENDING)) {
++ shutdown_pirq(irq);
++ } else if (VALID_EVTCHN(evtchn)) {
++ unmask_evtchn(evtchn);
++ pirq_unmask_notify(irq_to_pirq(irq));
++ }
++}
++
++static struct hw_interrupt_type pirq_type = {
++ .typename = "Phys-irq",
++ .startup = startup_pirq,
++ .shutdown = shutdown_pirq,
++ .enable = enable_pirq,
++ .disable = disable_pirq,
++ .ack = ack_pirq,
++ .end = end_pirq,
++#ifdef CONFIG_SMP
++ .set_affinity = set_affinity_irq,
++#endif
++ .retrigger = resend_irq_on_evtchn,
++};
++
++int irq_ignore_unhandled(unsigned int irq)
++{
++ struct physdev_irq_status_query irq_status = { .irq = irq };
++
++ if (!is_running_on_xen())
++ return 0;
++
++ if (HYPERVISOR_physdev_op(PHYSDEVOP_irq_status_query, &irq_status))
++ return 0;
++ return !!(irq_status.flags & XENIRQSTAT_shared);
++}
++
++void notify_remote_via_irq(int irq)
++{
++ int evtchn = evtchn_from_irq(irq);
++
++ if (VALID_EVTCHN(evtchn))
++ notify_remote_via_evtchn(evtchn);
++}
++EXPORT_SYMBOL_GPL(notify_remote_via_irq);
++
++int irq_to_evtchn_port(int irq)
++{
++ return evtchn_from_irq(irq);
++}
++EXPORT_SYMBOL_GPL(irq_to_evtchn_port);
++
++void mask_evtchn(int port)
++{
++ shared_info_t *s = HYPERVISOR_shared_info;
++ synch_set_bit(port, s->evtchn_mask);
++}
++EXPORT_SYMBOL_GPL(mask_evtchn);
++
++void unmask_evtchn(int port)
++{
++ shared_info_t *s = HYPERVISOR_shared_info;
++ unsigned int cpu = smp_processor_id();
++ vcpu_info_t *vcpu_info = &s->vcpu_info[cpu];
++
++ BUG_ON(!irqs_disabled());
++
++ /* Slow path (hypercall) if this is a non-local port. */
++ if (unlikely(cpu != cpu_from_evtchn(port))) {
++ struct evtchn_unmask unmask = { .port = port };
++ VOID(HYPERVISOR_event_channel_op(EVTCHNOP_unmask, &unmask));
++ return;
++ }
++
++ synch_clear_bit(port, s->evtchn_mask);
++
++ /* Did we miss an interrupt 'edge'? Re-fire if so. */
++ if (synch_test_bit(port, s->evtchn_pending) &&
++ !synch_test_and_set_bit(port / BITS_PER_LONG,
++ &vcpu_info->evtchn_pending_sel))
++ vcpu_info->evtchn_upcall_pending = 1;
++}
++EXPORT_SYMBOL_GPL(unmask_evtchn);
++
++void disable_all_local_evtchn(void)
++{
++ unsigned i, cpu = smp_processor_id();
++ shared_info_t *s = HYPERVISOR_shared_info;
++
++ for (i = 0; i < NR_EVENT_CHANNELS; ++i)
++ if (cpu_from_evtchn(i) == cpu)
++ synch_set_bit(i, &s->evtchn_mask[0]);
++}
++
++static void restore_cpu_virqs(unsigned int cpu)
++{
++ struct evtchn_bind_virq bind_virq;
++ int virq, irq, evtchn;
++
++ for (virq = 0; virq < NR_VIRQS; virq++) {
++ if ((irq = per_cpu(virq_to_irq, cpu)[virq]) == -1)
++ continue;
++
++ BUG_ON(irq_info[irq] != mk_irq_info(IRQT_VIRQ, virq, 0));
++
++ /* Get a new binding from Xen. */
++ bind_virq.virq = virq;
++ bind_virq.vcpu = cpu;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
++ &bind_virq) != 0)
++ BUG();
++ evtchn = bind_virq.port;
++
++ /* Record the new mapping. */
++ evtchn_to_irq[evtchn] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_VIRQ, virq, evtchn);
++ bind_evtchn_to_cpu(evtchn, cpu);
++
++ /* Ready for use. */
++ unmask_evtchn(evtchn);
++ }
++}
++
++static void restore_cpu_ipis(unsigned int cpu)
++{
++ struct evtchn_bind_ipi bind_ipi;
++ int ipi, irq, evtchn;
++
++ for (ipi = 0; ipi < NR_IPIS; ipi++) {
++ if ((irq = per_cpu(ipi_to_irq, cpu)[ipi]) == -1)
++ continue;
++
++ BUG_ON(irq_info[irq] != mk_irq_info(IRQT_IPI, ipi, 0));
++
++ /* Get a new binding from Xen. */
++ bind_ipi.vcpu = cpu;
++ if (HYPERVISOR_event_channel_op(EVTCHNOP_bind_ipi,
++ &bind_ipi) != 0)
++ BUG();
++ evtchn = bind_ipi.port;
++
++ /* Record the new mapping. */
++ evtchn_to_irq[evtchn] = irq;
++ irq_info[irq] = mk_irq_info(IRQT_IPI, ipi, evtchn);
++ bind_evtchn_to_cpu(evtchn, cpu);
++
++ /* Ready for use. */
++ unmask_evtchn(evtchn);
++
++ }
++}
++
++void irq_resume(void)
++{
++ unsigned int cpu, pirq, irq, evtchn;
++
++ init_evtchn_cpu_bindings();
++
++ /* New event-channel space is not 'live' yet. */
++ for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
++ mask_evtchn(evtchn);
++
++ /* Check that no PIRQs are still bound. */
++ for (pirq = 0; pirq < NR_PIRQS; pirq++)
++ BUG_ON(irq_info[pirq_to_irq(pirq)] != IRQ_UNBOUND);
++
++ /* No IRQ <-> event-channel mappings. */
++ for (irq = 0; irq < NR_IRQS; irq++)
++ irq_info[irq] &= ~0xFFFF; /* zap event-channel binding */
++ for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
++ evtchn_to_irq[evtchn] = -1;
++
++ for_each_possible_cpu(cpu) {
++ restore_cpu_virqs(cpu);
++ restore_cpu_ipis(cpu);
++ }
++
++}
++
++void __init xen_init_IRQ(void)
++{
++ unsigned int i;
++
++ init_evtchn_cpu_bindings();
++
++ /* No event channels are 'live' right now. */
++ for (i = 0; i < NR_EVENT_CHANNELS; i++)
++ mask_evtchn(i);
++
++ /* No IRQ -> event-channel mappings. */
++ for (i = 0; i < NR_IRQS; i++)
++ irq_info[i] = IRQ_UNBOUND;
++
++ /* Dynamic IRQ space is currently unbound. Zero the refcnts. */
++ for (i = 0; i < NR_DYNIRQS; i++) {
++ irq_bindcount[dynirq_to_irq(i)] = 0;
++
++ irq_desc[dynirq_to_irq(i)].status = IRQ_DISABLED;
++ irq_desc[dynirq_to_irq(i)].action = NULL;
++ irq_desc[dynirq_to_irq(i)].depth = 1;
++ irq_desc[dynirq_to_irq(i)].chip = &dynirq_type;
++ }
++
++ /* Phys IRQ space is statically bound (1:1 mapping). Nail refcnts. */
++ for (i = 0; i < NR_PIRQS; i++) {
++ irq_bindcount[pirq_to_irq(i)] = 1;
++
++#ifdef RTC_IRQ
++ /* If not domain 0, force our RTC driver to fail its probe. */
++ if ((i == RTC_IRQ) && !is_initial_xendomain())
++ continue;
++#endif
++
++ irq_desc[pirq_to_irq(i)].status = IRQ_DISABLED;
++ irq_desc[pirq_to_irq(i)].action = NULL;
++ irq_desc[pirq_to_irq(i)].depth = 1;
++ irq_desc[pirq_to_irq(i)].chip = &pirq_type;
++ }
++}
+--- linux-2.6.18.8/drivers/xen/core/features.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/features.c 2008-05-19 00:33:45.770784401 +0300
+@@ -0,0 +1,34 @@
++/******************************************************************************
++ * features.c
++ *
++ * Xen feature flags.
++ *
++ * Copyright (c) 2006, Ian Campbell, XenSource Inc.
++ */
++#include <linux/types.h>
++#include <linux/cache.h>
++#include <linux/module.h>
++#include <asm/hypervisor.h>
++#include <xen/features.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++u8 xen_features[XENFEAT_NR_SUBMAPS * 32] __read_mostly;
++/* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */
++EXPORT_SYMBOL(xen_features);
++
++void setup_xen_features(void)
++{
++ xen_feature_info_t fi;
++ int i, j;
++
++ for (i = 0; i < XENFEAT_NR_SUBMAPS; i++) {
++ fi.submap_idx = i;
++ if (HYPERVISOR_xen_version(XENVER_get_features, &fi) < 0)
++ break;
++ for (j=0; j<32; j++)
++ xen_features[i*32+j] = !!(fi.submap & 1<<j);
++ }
++}
+--- linux-2.6.18.8/drivers/xen/core/firmware.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/firmware.c 2008-05-19 00:33:45.770784401 +0300
+@@ -0,0 +1,74 @@
++#include <linux/kernel.h>
++#include <linux/errno.h>
++#include <linux/init.h>
++#include <linux/edd.h>
++#include <video/edid.h>
++#include <xen/interface/platform.h>
++#include <asm/hypervisor.h>
++
++#if defined(CONFIG_EDD) || defined(CONFIG_EDD_MODULE)
++void __init copy_edd(void)
++{
++ int ret;
++ struct xen_platform_op op;
++
++ if (!is_initial_xendomain())
++ return;
++
++ op.cmd = XENPF_firmware_info;
++
++ op.u.firmware_info.type = XEN_FW_DISK_INFO;
++ for (op.u.firmware_info.index = 0;
++ edd.edd_info_nr < EDDMAXNR;
++ op.u.firmware_info.index++) {
++ struct edd_info *info = edd.edd_info + edd.edd_info_nr;
++
++ info->params.length = sizeof(info->params);
++ set_xen_guest_handle(op.u.firmware_info.u.disk_info.edd_params,
++ &info->params);
++ ret = HYPERVISOR_platform_op(&op);
++ if (ret)
++ break;
++
++#define C(x) info->x = op.u.firmware_info.u.disk_info.x
++ C(device);
++ C(version);
++ C(interface_support);
++ C(legacy_max_cylinder);
++ C(legacy_max_head);
++ C(legacy_sectors_per_track);
++#undef C
++
++ edd.edd_info_nr++;
++ }
++
++ op.u.firmware_info.type = XEN_FW_DISK_MBR_SIGNATURE;
++ for (op.u.firmware_info.index = 0;
++ edd.mbr_signature_nr < EDD_MBR_SIG_MAX;
++ op.u.firmware_info.index++) {
++ ret = HYPERVISOR_platform_op(&op);
++ if (ret)
++ break;
++ edd.mbr_signature[edd.mbr_signature_nr++] =
++ op.u.firmware_info.u.disk_mbr_signature.mbr_signature;
++ }
++}
++#endif
++
++void __init copy_edid(void)
++{
++#if defined(CONFIG_FIRMWARE_EDID) && defined(CONFIG_X86)
++ struct xen_platform_op op;
++
++ if (!is_initial_xendomain())
++ return;
++
++ op.cmd = XENPF_firmware_info;
++ op.u.firmware_info.index = 0;
++ op.u.firmware_info.type = XEN_FW_VBEDDC_INFO;
++ set_xen_guest_handle(op.u.firmware_info.u.vbeddc_info.edid,
++ edid_info.dummy);
++ if (HYPERVISOR_platform_op(&op) != 0)
++ memset(edid_info.dummy, 0x13, sizeof(edid_info.dummy));
++#endif
++}
+--- linux-2.6.18.8/drivers/xen/core/gnttab.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/gnttab.c 2008-05-19 00:33:45.818787168 +0300
+@@ -0,0 +1,769 @@
++/******************************************************************************
++ * gnttab.c
++ *
++ * Granting foreign access to our memory reservation.
++ *
++ * Copyright (c) 2005-2006, Christopher Clark
++ * Copyright (c) 2004-2005, K A Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/module.h>
++#include <linux/sched.h>
++#include <linux/mm.h>
++#include <linux/seqlock.h>
++#include <xen/interface/xen.h>
++#include <xen/gnttab.h>
++#include <asm/pgtable.h>
++#include <asm/uaccess.h>
++#include <asm/synch_bitops.h>
++#include <asm/io.h>
++#include <xen/interface/memory.h>
++#include <xen/driver_util.h>
++#include <asm/gnttab_dma.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++/* External tools reserve first few grant table entries. */
++#define NR_RESERVED_ENTRIES 8
++#define GNTTAB_LIST_END 0xffffffff
++#define ENTRIES_PER_GRANT_FRAME (PAGE_SIZE / sizeof(grant_entry_t))
++
++static grant_ref_t **gnttab_list;
++static unsigned int nr_grant_frames;
++static unsigned int boot_max_nr_grant_frames;
++static int gnttab_free_count;
++static grant_ref_t gnttab_free_head;
++static DEFINE_SPINLOCK(gnttab_list_lock);
++
++static struct grant_entry *shared;
++
++static struct gnttab_free_callback *gnttab_free_callback_list;
++
++static int gnttab_expand(unsigned int req_entries);
++
++#define RPP (PAGE_SIZE / sizeof(grant_ref_t))
++#define gnttab_entry(entry) (gnttab_list[(entry) / RPP][(entry) % RPP])
++
++#define nr_freelist_frames(grant_frames) \
++ (((grant_frames) * ENTRIES_PER_GRANT_FRAME + RPP - 1) / RPP)
++
++static int get_free_entries(int count)
++{
++ unsigned long flags;
++ int ref, rc;
++ grant_ref_t head;
++
++ spin_lock_irqsave(&gnttab_list_lock, flags);
++
++ if ((gnttab_free_count < count) &&
++ ((rc = gnttab_expand(count - gnttab_free_count)) < 0)) {
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++ return rc;
++ }
++
++ ref = head = gnttab_free_head;
++ gnttab_free_count -= count;
++ while (count-- > 1)
++ head = gnttab_entry(head);
++ gnttab_free_head = gnttab_entry(head);
++ gnttab_entry(head) = GNTTAB_LIST_END;
++
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++
++ return ref;
++}
++
++#define get_free_entry() get_free_entries(1)
++
++static void do_free_callbacks(void)
++{
++ struct gnttab_free_callback *callback, *next;
++
++ callback = gnttab_free_callback_list;
++ gnttab_free_callback_list = NULL;
++
++ while (callback != NULL) {
++ next = callback->next;
++ if (gnttab_free_count >= callback->count) {
++ callback->next = NULL;
++ callback->fn(callback->arg);
++ } else {
++ callback->next = gnttab_free_callback_list;
++ gnttab_free_callback_list = callback;
++ }
++ callback = next;
++ }
++}
++
++static inline void check_free_callbacks(void)
++{
++ if (unlikely(gnttab_free_callback_list))
++ do_free_callbacks();
++}
++
++static void put_free_entry(grant_ref_t ref)
++{
++ unsigned long flags;
++ spin_lock_irqsave(&gnttab_list_lock, flags);
++ gnttab_entry(ref) = gnttab_free_head;
++ gnttab_free_head = ref;
++ gnttab_free_count++;
++ check_free_callbacks();
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++}
++
++/*
++ * Public grant-issuing interface functions
++ */
++
++int gnttab_grant_foreign_access(domid_t domid, unsigned long frame,
++ int flags)
++{
++ int ref;
++
++ if (unlikely((ref = get_free_entry()) < 0))
++ return -ENOSPC;
++
++ shared[ref].frame = frame;
++ shared[ref].domid = domid;
++ wmb();
++ BUG_ON(flags & (GTF_accept_transfer | GTF_reading | GTF_writing));
++ shared[ref].flags = GTF_permit_access | flags;
++
++ return ref;
++}
++EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access);
++
++void gnttab_grant_foreign_access_ref(grant_ref_t ref, domid_t domid,
++ unsigned long frame, int flags)
++{
++ shared[ref].frame = frame;
++ shared[ref].domid = domid;
++ wmb();
++ BUG_ON(flags & (GTF_accept_transfer | GTF_reading | GTF_writing));
++ shared[ref].flags = GTF_permit_access | flags;
++}
++EXPORT_SYMBOL_GPL(gnttab_grant_foreign_access_ref);
++
++
++int gnttab_query_foreign_access(grant_ref_t ref)
++{
++ u16 nflags;
++
++ nflags = shared[ref].flags;
++
++ return (nflags & (GTF_reading|GTF_writing));
++}
++EXPORT_SYMBOL_GPL(gnttab_query_foreign_access);
++
++int gnttab_end_foreign_access_ref(grant_ref_t ref)
++{
++ u16 flags, nflags;
++
++ nflags = shared[ref].flags;
++ do {
++ if ((flags = nflags) & (GTF_reading|GTF_writing)) {
++ printk(KERN_DEBUG "WARNING: g.e. still in use!\n");
++ return 0;
++ }
++ } while ((nflags = synch_cmpxchg_subword(&shared[ref].flags, flags, 0)) !=
++ flags);
++
++ return 1;
++}
++EXPORT_SYMBOL_GPL(gnttab_end_foreign_access_ref);
++
++void gnttab_end_foreign_access(grant_ref_t ref, unsigned long page)
++{
++ if (gnttab_end_foreign_access_ref(ref)) {
++ put_free_entry(ref);
++ if (page != 0)
++ free_page(page);
++ } else {
++ /* XXX This needs to be fixed so that the ref and page are
++ placed on a list to be freed up later. */
++ printk(KERN_DEBUG
++ "WARNING: leaking g.e. and page still in use!\n");
++ }
++}
++EXPORT_SYMBOL_GPL(gnttab_end_foreign_access);
++
++int gnttab_grant_foreign_transfer(domid_t domid, unsigned long pfn)
++{
++ int ref;
++
++ if (unlikely((ref = get_free_entry()) < 0))
++ return -ENOSPC;
++ gnttab_grant_foreign_transfer_ref(ref, domid, pfn);
++
++ return ref;
++}
++EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer);
++
++void gnttab_grant_foreign_transfer_ref(grant_ref_t ref, domid_t domid,
++ unsigned long pfn)
++{
++ shared[ref].frame = pfn;
++ shared[ref].domid = domid;
++ wmb();
++ shared[ref].flags = GTF_accept_transfer;
++}
++EXPORT_SYMBOL_GPL(gnttab_grant_foreign_transfer_ref);
++
++unsigned long gnttab_end_foreign_transfer_ref(grant_ref_t ref)
++{
++ unsigned long frame;
++ u16 flags;
++
++ /*
++ * If a transfer is not even yet started, try to reclaim the grant
++ * reference and return failure (== 0).
++ */
++ while (!((flags = shared[ref].flags) & GTF_transfer_committed)) {
++ if (synch_cmpxchg_subword(&shared[ref].flags, flags, 0) == flags)
++ return 0;
++ cpu_relax();
++ }
++
++ /* If a transfer is in progress then wait until it is completed. */
++ while (!(flags & GTF_transfer_completed)) {
++ flags = shared[ref].flags;
++ cpu_relax();
++ }
++
++ /* Read the frame number /after/ reading completion status. */
++ rmb();
++ frame = shared[ref].frame;
++ BUG_ON(frame == 0);
++
++ return frame;
++}
++EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer_ref);
++
++unsigned long gnttab_end_foreign_transfer(grant_ref_t ref)
++{
++ unsigned long frame = gnttab_end_foreign_transfer_ref(ref);
++ put_free_entry(ref);
++ return frame;
++}
++EXPORT_SYMBOL_GPL(gnttab_end_foreign_transfer);
++
++void gnttab_free_grant_reference(grant_ref_t ref)
++{
++ put_free_entry(ref);
++}
++EXPORT_SYMBOL_GPL(gnttab_free_grant_reference);
++
++void gnttab_free_grant_references(grant_ref_t head)
++{
++ grant_ref_t ref;
++ unsigned long flags;
++ int count = 1;
++ if (head == GNTTAB_LIST_END)
++ return;
++ spin_lock_irqsave(&gnttab_list_lock, flags);
++ ref = head;
++ while (gnttab_entry(ref) != GNTTAB_LIST_END) {
++ ref = gnttab_entry(ref);
++ count++;
++ }
++ gnttab_entry(ref) = gnttab_free_head;
++ gnttab_free_head = head;
++ gnttab_free_count += count;
++ check_free_callbacks();
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++}
++EXPORT_SYMBOL_GPL(gnttab_free_grant_references);
++
++int gnttab_alloc_grant_references(u16 count, grant_ref_t *head)
++{
++ int h = get_free_entries(count);
++
++ if (h < 0)
++ return -ENOSPC;
++
++ *head = h;
++
++ return 0;
++}
++EXPORT_SYMBOL_GPL(gnttab_alloc_grant_references);
++
++int gnttab_empty_grant_references(const grant_ref_t *private_head)
++{
++ return (*private_head == GNTTAB_LIST_END);
++}
++EXPORT_SYMBOL_GPL(gnttab_empty_grant_references);
++
++int gnttab_claim_grant_reference(grant_ref_t *private_head)
++{
++ grant_ref_t g = *private_head;
++ if (unlikely(g == GNTTAB_LIST_END))
++ return -ENOSPC;
++ *private_head = gnttab_entry(g);
++ return g;
++}
++EXPORT_SYMBOL_GPL(gnttab_claim_grant_reference);
++
++void gnttab_release_grant_reference(grant_ref_t *private_head,
++ grant_ref_t release)
++{
++ gnttab_entry(release) = *private_head;
++ *private_head = release;
++}
++EXPORT_SYMBOL_GPL(gnttab_release_grant_reference);
++
++void gnttab_request_free_callback(struct gnttab_free_callback *callback,
++ void (*fn)(void *), void *arg, u16 count)
++{
++ unsigned long flags;
++ spin_lock_irqsave(&gnttab_list_lock, flags);
++ if (callback->next)
++ goto out;
++ callback->fn = fn;
++ callback->arg = arg;
++ callback->count = count;
++ callback->next = gnttab_free_callback_list;
++ gnttab_free_callback_list = callback;
++ check_free_callbacks();
++out:
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++}
++EXPORT_SYMBOL_GPL(gnttab_request_free_callback);
++
++void gnttab_cancel_free_callback(struct gnttab_free_callback *callback)
++{
++ struct gnttab_free_callback **pcb;
++ unsigned long flags;
++
++ spin_lock_irqsave(&gnttab_list_lock, flags);
++ for (pcb = &gnttab_free_callback_list; *pcb; pcb = &(*pcb)->next) {
++ if (*pcb == callback) {
++ *pcb = callback->next;
++ break;
++ }
++ }
++ spin_unlock_irqrestore(&gnttab_list_lock, flags);
++}
++EXPORT_SYMBOL_GPL(gnttab_cancel_free_callback);
++
++static int grow_gnttab_list(unsigned int more_frames)
++{
++ unsigned int new_nr_grant_frames, extra_entries, i;
++ unsigned int nr_glist_frames, new_nr_glist_frames;
++
++ new_nr_grant_frames = nr_grant_frames + more_frames;
++ extra_entries = more_frames * ENTRIES_PER_GRANT_FRAME;
++
++ nr_glist_frames = nr_freelist_frames(nr_grant_frames);
++ new_nr_glist_frames = nr_freelist_frames(new_nr_grant_frames);
++ for (i = nr_glist_frames; i < new_nr_glist_frames; i++) {
++ gnttab_list[i] = (grant_ref_t *)__get_free_page(GFP_ATOMIC);
++ if (!gnttab_list[i])
++ goto grow_nomem;
++ }
++
++ for (i = ENTRIES_PER_GRANT_FRAME * nr_grant_frames;
++ i < ENTRIES_PER_GRANT_FRAME * new_nr_grant_frames - 1; i++)
++ gnttab_entry(i) = i + 1;
++
++ gnttab_entry(i) = gnttab_free_head;
++ gnttab_free_head = ENTRIES_PER_GRANT_FRAME * nr_grant_frames;
++ gnttab_free_count += extra_entries;
++
++ nr_grant_frames = new_nr_grant_frames;
++
++ check_free_callbacks();
++
++ return 0;
++
++grow_nomem:
++ for ( ; i >= nr_glist_frames; i--)
++ free_page((unsigned long) gnttab_list[i]);
++ return -ENOMEM;
++}
++
++static unsigned int __max_nr_grant_frames(void)
++{
++ struct gnttab_query_size query;
++ int rc;
++
++ query.dom = DOMID_SELF;
++
++ rc = HYPERVISOR_grant_table_op(GNTTABOP_query_size, &query, 1);
++ if ((rc < 0) || (query.status != GNTST_okay))
++ return 4; /* Legacy max supported number of frames */
++
++ return query.max_nr_frames;
++}
++
++static inline unsigned int max_nr_grant_frames(void)
++{
++ unsigned int xen_max = __max_nr_grant_frames();
++
++ if (xen_max > boot_max_nr_grant_frames)
++ return boot_max_nr_grant_frames;
++ return xen_max;
++}
++
++#ifdef CONFIG_XEN
++
++static DEFINE_SEQLOCK(gnttab_dma_lock);
++
++#ifdef CONFIG_X86
++static int map_pte_fn(pte_t *pte, struct page *pmd_page,
++ unsigned long addr, void *data)
++{
++ unsigned long **frames = (unsigned long **)data;
++
++ set_pte_at(&init_mm, addr, pte, pfn_pte_ma((*frames)[0], PAGE_KERNEL));
++ (*frames)++;
++ return 0;
++}
++
++static int unmap_pte_fn(pte_t *pte, struct page *pmd_page,
++ unsigned long addr, void *data)
++{
++
++ set_pte_at(&init_mm, addr, pte, __pte(0));
++ return 0;
++}
++
++void *arch_gnttab_alloc_shared(unsigned long *frames)
++{
++ struct vm_struct *area;
++ area = alloc_vm_area(PAGE_SIZE * max_nr_grant_frames());
++ BUG_ON(area == NULL);
++ return area->addr;
++}
++#endif /* CONFIG_X86 */
++
++static int gnttab_map(unsigned int start_idx, unsigned int end_idx)
++{
++ struct gnttab_setup_table setup;
++ unsigned long *frames;
++ unsigned int nr_gframes = end_idx + 1;
++ int rc;
++
++ frames = kmalloc(nr_gframes * sizeof(unsigned long), GFP_ATOMIC);
++ if (!frames)
++ return -ENOMEM;
++
++ setup.dom = DOMID_SELF;
++ setup.nr_frames = nr_gframes;
++ set_xen_guest_handle(setup.frame_list, frames);
++
++ rc = HYPERVISOR_grant_table_op(GNTTABOP_setup_table, &setup, 1);
++ if (rc == -ENOSYS) {
++ kfree(frames);
++ return -ENOSYS;
++ }
++
++ BUG_ON(rc || setup.status);
++
++ if (shared == NULL)
++ shared = arch_gnttab_alloc_shared(frames);
++
++#ifdef CONFIG_X86
++ rc = apply_to_page_range(&init_mm, (unsigned long)shared,
++ PAGE_SIZE * nr_gframes,
++ map_pte_fn, &frames);
++ BUG_ON(rc);
++ frames -= nr_gframes; /* adjust after map_pte_fn() */
++#endif /* CONFIG_X86 */
++
++ kfree(frames);
++
++ return 0;
++}
++
++static void gnttab_page_free(struct page *page)
++{
++ ClearPageForeign(page);
++ gnttab_reset_grant_page(page);
++ put_page(page);
++}
++
++/*
++ * Must not be called with IRQs off. This should only be used on the
++ * slow path.
++ *
++ * Copy a foreign granted page to local memory.
++ */
++int gnttab_copy_grant_page(grant_ref_t ref, struct page **pagep)
++{
++ struct gnttab_unmap_and_replace unmap;
++ mmu_update_t mmu;
++ struct page *page;
++ struct page *new_page;
++ void *new_addr;
++ void *addr;
++ paddr_t pfn;
++ maddr_t mfn;
++ maddr_t new_mfn;
++ int err;
++
++ page = *pagep;
++ if (!get_page_unless_zero(page))
++ return -ENOENT;
++
++ err = -ENOMEM;
++ new_page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
++ if (!new_page)
++ goto out;
++
++ new_addr = page_address(new_page);
++ addr = page_address(page);
++ memcpy(new_addr, addr, PAGE_SIZE);
++
++ pfn = page_to_pfn(page);
++ mfn = pfn_to_mfn(pfn);
++ new_mfn = virt_to_mfn(new_addr);
++
++ write_seqlock(&gnttab_dma_lock);
++
++ /* Make seq visible before checking page_mapped. */
++ smp_mb();
++
++ /* Has the page been DMA-mapped? */
++ if (unlikely(page_mapped(page))) {
++ write_sequnlock(&gnttab_dma_lock);
++ put_page(new_page);
++ err = -EBUSY;
++ goto out;
++ }
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap))
++ set_phys_to_machine(pfn, new_mfn);
++
++ gnttab_set_replace_op(&unmap, (unsigned long)addr,
++ (unsigned long)new_addr, ref);
++
++ err = HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_replace,
++ &unmap, 1);
++ BUG_ON(err);
++ BUG_ON(unmap.status);
++
++ write_sequnlock(&gnttab_dma_lock);
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ set_phys_to_machine(page_to_pfn(new_page), INVALID_P2M_ENTRY);
++
++ mmu.ptr = (new_mfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE;
++ mmu.val = pfn;
++ err = HYPERVISOR_mmu_update(&mmu, 1, NULL, DOMID_SELF);
++ BUG_ON(err);
++ }
++
++ new_page->mapping = page->mapping;
++ new_page->index = page->index;
++ set_bit(PG_foreign, &new_page->flags);
++ *pagep = new_page;
++
++ SetPageForeign(page, gnttab_page_free);
++ page->mapping = NULL;
++
++out:
++ put_page(page);
++ return err;
++}
++EXPORT_SYMBOL_GPL(gnttab_copy_grant_page);
++
++void gnttab_reset_grant_page(struct page *page)
++{
++ init_page_count(page);
++ reset_page_mapcount(page);
++}
++EXPORT_SYMBOL_GPL(gnttab_reset_grant_page);
++
++/*
++ * Keep track of foreign pages marked as PageForeign so that we don't
++ * return them to the remote domain prematurely.
++ *
++ * PageForeign pages are pinned down by increasing their mapcount.
++ *
++ * All other pages are simply returned as is.
++ */
++void __gnttab_dma_map_page(struct page *page)
++{
++ unsigned int seq;
++
++ if (!is_running_on_xen() || !PageForeign(page))
++ return;
++
++ do {
++ seq = read_seqbegin(&gnttab_dma_lock);
++
++ if (gnttab_dma_local_pfn(page))
++ break;
++
++ atomic_set(&page->_mapcount, 0);
++
++ /* Make _mapcount visible before read_seqretry. */
++ smp_mb();
++ } while (unlikely(read_seqretry(&gnttab_dma_lock, seq)));
++}
++
++int gnttab_resume(void)
++{
++ if (max_nr_grant_frames() < nr_grant_frames)
++ return -ENOSYS;
++ return gnttab_map(0, nr_grant_frames - 1);
++}
++
++int gnttab_suspend(void)
++{
++#ifdef CONFIG_X86
++ apply_to_page_range(&init_mm, (unsigned long)shared,
++ PAGE_SIZE * nr_grant_frames,
++ unmap_pte_fn, NULL);
++#endif
++ return 0;
++}
++
++#else /* !CONFIG_XEN */
++
++#include <platform-pci.h>
++
++static unsigned long resume_frames;
++
++static int gnttab_map(unsigned int start_idx, unsigned int end_idx)
++{
++ struct xen_add_to_physmap xatp;
++ unsigned int i = end_idx;
++
++ /* Loop backwards, so that the first hypercall has the largest index,
++ * ensuring that the table will grow only once.
++ */
++ do {
++ xatp.domid = DOMID_SELF;
++ xatp.idx = i;
++ xatp.space = XENMAPSPACE_grant_table;
++ xatp.gpfn = (resume_frames >> PAGE_SHIFT) + i;
++ if (HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp))
++ BUG();
++ } while (i-- > start_idx);
++
++ return 0;
++}
++
++int gnttab_resume(void)
++{
++ unsigned int max_nr_gframes, nr_gframes;
++
++ nr_gframes = nr_grant_frames;
++ max_nr_gframes = max_nr_grant_frames();
++ if (max_nr_gframes < nr_gframes)
++ return -ENOSYS;
++
++ if (!resume_frames) {
++ resume_frames = alloc_xen_mmio(PAGE_SIZE * max_nr_gframes);
++ shared = ioremap(resume_frames, PAGE_SIZE * max_nr_gframes);
++ if (shared == NULL) {
++ printk("error to ioremap gnttab share frames\n");
++ return -1;
++ }
++ }
++
++ gnttab_map(0, nr_gframes - 1);
++
++ return 0;
++}
++
++#endif /* !CONFIG_XEN */
++
++static int gnttab_expand(unsigned int req_entries)
++{
++ int rc;
++ unsigned int cur, extra;
++
++ cur = nr_grant_frames;
++ extra = ((req_entries + (ENTRIES_PER_GRANT_FRAME-1)) /
++ ENTRIES_PER_GRANT_FRAME);
++ if (cur + extra > max_nr_grant_frames())
++ return -ENOSPC;
++
++ if ((rc = gnttab_map(cur, cur + extra - 1)) == 0)
++ rc = grow_gnttab_list(extra);
++
++ return rc;
++}
++
++int __devinit gnttab_init(void)
++{
++ int i;
++ unsigned int max_nr_glist_frames, nr_glist_frames;
++ unsigned int nr_init_grefs;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ nr_grant_frames = 1;
++ boot_max_nr_grant_frames = __max_nr_grant_frames();
++
++ /* Determine the maximum number of frames required for the
++ * grant reference free list on the current hypervisor.
++ */
++ max_nr_glist_frames = nr_freelist_frames(boot_max_nr_grant_frames);
++
++ gnttab_list = kmalloc(max_nr_glist_frames * sizeof(grant_ref_t *),
++ GFP_KERNEL);
++ if (gnttab_list == NULL)
++ return -ENOMEM;
++
++ nr_glist_frames = nr_freelist_frames(nr_grant_frames);
++ for (i = 0; i < nr_glist_frames; i++) {
++ gnttab_list[i] = (grant_ref_t *)__get_free_page(GFP_KERNEL);
++ if (gnttab_list[i] == NULL)
++ goto ini_nomem;
++ }
++
++ if (gnttab_resume() < 0)
++ return -ENODEV;
++
++ nr_init_grefs = nr_grant_frames * ENTRIES_PER_GRANT_FRAME;
++
++ for (i = NR_RESERVED_ENTRIES; i < nr_init_grefs - 1; i++)
++ gnttab_entry(i) = i + 1;
++
++ gnttab_entry(nr_init_grefs - 1) = GNTTAB_LIST_END;
++ gnttab_free_count = nr_init_grefs - NR_RESERVED_ENTRIES;
++ gnttab_free_head = NR_RESERVED_ENTRIES;
++
++ return 0;
++
++ ini_nomem:
++ for (i--; i >= 0; i--)
++ free_page((unsigned long)gnttab_list[i]);
++ kfree(gnttab_list);
++ return -ENOMEM;
++}
++
++#ifdef CONFIG_XEN
++core_initcall(gnttab_init);
++#endif
+--- linux-2.6.18.8/drivers/xen/core/hypervisor_sysfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/hypervisor_sysfs.c 2008-05-19 00:33:45.822787398 +0300
+@@ -0,0 +1,57 @@
++/*
++ * copyright (c) 2006 IBM Corporation
++ * Authored by: Mike D. Day <ncmike@us.ibm.com>
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License version 2 as
++ * published by the Free Software Foundation.
++ */
++
++#include <linux/kernel.h>
++#include <linux/module.h>
++#include <linux/kobject.h>
++#include <xen/hypervisor_sysfs.h>
++#include <asm/hypervisor.h>
++
++static ssize_t hyp_sysfs_show(struct kobject *kobj,
++ struct attribute *attr,
++ char *buffer)
++{
++ struct hyp_sysfs_attr *hyp_attr;
++ hyp_attr = container_of(attr, struct hyp_sysfs_attr, attr);
++ if (hyp_attr->show)
++ return hyp_attr->show(hyp_attr, buffer);
++ return 0;
++}
++
++static ssize_t hyp_sysfs_store(struct kobject *kobj,
++ struct attribute *attr,
++ const char *buffer,
++ size_t len)
++{
++ struct hyp_sysfs_attr *hyp_attr;
++ hyp_attr = container_of(attr, struct hyp_sysfs_attr, attr);
++ if (hyp_attr->store)
++ return hyp_attr->store(hyp_attr, buffer, len);
++ return 0;
++}
++
++static struct sysfs_ops hyp_sysfs_ops = {
++ .show = hyp_sysfs_show,
++ .store = hyp_sysfs_store,
++};
++
++static struct kobj_type hyp_sysfs_kobj_type = {
++ .sysfs_ops = &hyp_sysfs_ops,
++};
++
++static int __init hypervisor_subsys_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ hypervisor_subsys.kset.kobj.ktype = &hyp_sysfs_kobj_type;
++ return 0;
++}
++
++device_initcall(hypervisor_subsys_init);
+--- linux-2.6.18.8/drivers/xen/core/machine_kexec.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/machine_kexec.c 2008-05-19 00:33:45.822787398 +0300
+@@ -0,0 +1,192 @@
++/*
++ * drivers/xen/core/machine_kexec.c
++ * handle transition of Linux booting another kernel
++ */
++
++#include <linux/kexec.h>
++#include <xen/interface/kexec.h>
++#include <linux/mm.h>
++#include <linux/bootmem.h>
++
++extern void machine_kexec_setup_load_arg(xen_kexec_image_t *xki,
++ struct kimage *image);
++extern int machine_kexec_setup_resources(struct resource *hypervisor,
++ struct resource *phys_cpus,
++ int nr_phys_cpus);
++extern void machine_kexec_register_resources(struct resource *res);
++
++static int __initdata xen_max_nr_phys_cpus;
++static struct resource xen_hypervisor_res;
++static struct resource *xen_phys_cpus;
++
++void __init xen_machine_kexec_setup_resources(void)
++{
++ xen_kexec_range_t range;
++ struct resource *res;
++ int k = 0;
++
++ if (!is_initial_xendomain())
++ return;
++
++ /* determine maximum number of physical cpus */
++
++ while (1) {
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_CPU;
++ range.nr = k;
++
++ if(HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ break;
++
++ k++;
++ }
++
++ if (k == 0)
++ return;
++
++ xen_max_nr_phys_cpus = k;
++
++ /* allocate xen_phys_cpus */
++
++ xen_phys_cpus = alloc_bootmem_low(k * sizeof(struct resource));
++ BUG_ON(xen_phys_cpus == NULL);
++
++ /* fill in xen_phys_cpus with per-cpu crash note information */
++
++ for (k = 0; k < xen_max_nr_phys_cpus; k++) {
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_CPU;
++ range.nr = k;
++
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ goto err;
++
++ res = xen_phys_cpus + k;
++
++ memset(res, 0, sizeof(*res));
++ res->name = "Crash note";
++ res->start = range.start;
++ res->end = range.start + range.size - 1;
++ res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
++ }
++
++ /* fill in xen_hypervisor_res with hypervisor machine address range */
++
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_XEN;
++
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ goto err;
++
++ xen_hypervisor_res.name = "Hypervisor code and data";
++ xen_hypervisor_res.start = range.start;
++ xen_hypervisor_res.end = range.start + range.size - 1;
++ xen_hypervisor_res.flags = IORESOURCE_BUSY | IORESOURCE_MEM;
++
++ /* fill in crashk_res if range is reserved by hypervisor */
++
++ memset(&range, 0, sizeof(range));
++ range.range = KEXEC_RANGE_MA_CRASH;
++
++ if (HYPERVISOR_kexec_op(KEXEC_CMD_kexec_get_range, &range))
++ goto err;
++
++ if (range.size) {
++ crashk_res.start = range.start;
++ crashk_res.end = range.start + range.size - 1;
++ }
++
++ if (machine_kexec_setup_resources(&xen_hypervisor_res, xen_phys_cpus,
++ xen_max_nr_phys_cpus))
++ goto err;
++
++ return;
++
++ err:
++ /*
++ * It isn't possible to free xen_phys_cpus this early in the
++ * boot. Failure at this stage is unexpected and the amount of
++ * memory is small therefore we tolerate the potential leak.
++ */
++ xen_max_nr_phys_cpus = 0;
++ return;
++}
++
++void __init xen_machine_kexec_register_resources(struct resource *res)
++{
++ request_resource(res, &xen_hypervisor_res);
++ machine_kexec_register_resources(res);
++}
++
++static void setup_load_arg(xen_kexec_image_t *xki, struct kimage *image)
++{
++ machine_kexec_setup_load_arg(xki, image);
++
++ xki->indirection_page = image->head;
++ xki->start_address = image->start;
++}
++
++/*
++ * Load the image into xen so xen can kdump itself
++ * This might have been done in prepare, but prepare
++ * is currently called too early. It might make sense
++ * to move prepare, but for now, just add an extra hook.
++ */
++int xen_machine_kexec_load(struct kimage *image)
++{
++ xen_kexec_load_t xkl;
++
++ memset(&xkl, 0, sizeof(xkl));
++ xkl.type = image->type;
++ setup_load_arg(&xkl.image, image);
++ return HYPERVISOR_kexec_op(KEXEC_CMD_kexec_load, &xkl);
++}
++
++/*
++ * Unload the image that was stored by machine_kexec_load()
++ * This might have been done in machine_kexec_cleanup() but it
++ * is called too late, and its possible xen could try and kdump
++ * using resources that have been freed.
++ */
++void xen_machine_kexec_unload(struct kimage *image)
++{
++ xen_kexec_load_t xkl;
++
++ memset(&xkl, 0, sizeof(xkl));
++ xkl.type = image->type;
++ WARN_ON(HYPERVISOR_kexec_op(KEXEC_CMD_kexec_unload, &xkl));
++}
++
++/*
++ * Do not allocate memory (or fail in any way) in machine_kexec().
++ * We are past the point of no return, committed to rebooting now.
++ *
++ * This has the hypervisor move to the prefered reboot CPU,
++ * stop all CPUs and kexec. That is it combines machine_shutdown()
++ * and machine_kexec() in Linux kexec terms.
++ */
++NORET_TYPE void machine_kexec(struct kimage *image)
++{
++ xen_kexec_exec_t xke;
++
++ memset(&xke, 0, sizeof(xke));
++ xke.type = image->type;
++ VOID(HYPERVISOR_kexec_op(KEXEC_CMD_kexec, &xke));
++ panic("KEXEC_CMD_kexec hypercall should not return\n");
++}
++
++void machine_shutdown(void)
++{
++ /* do nothing */
++}
++
++
++/*
++ * Local variables:
++ * c-file-style: "linux"
++ * indent-tabs-mode: t
++ * c-indent-level: 8
++ * c-basic-offset: 8
++ * tab-width: 8
++ * End:
++ */
+--- linux-2.6.18.8/drivers/xen/core/machine_reboot.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/machine_reboot.c 2008-05-19 00:33:45.822787398 +0300
+@@ -0,0 +1,252 @@
++#include <linux/version.h>
++#include <linux/kernel.h>
++#include <linux/mm.h>
++#include <linux/unistd.h>
++#include <linux/module.h>
++#include <linux/reboot.h>
++#include <linux/sysrq.h>
++#include <linux/stringify.h>
++#include <linux/stop_machine.h>
++#include <asm/irq.h>
++#include <asm/mmu_context.h>
++#include <xen/evtchn.h>
++#include <asm/hypervisor.h>
++#include <xen/xenbus.h>
++#include <linux/cpu.h>
++#include <xen/gnttab.h>
++#include <xen/xencons.h>
++#include <xen/cpu_hotplug.h>
++#include <xen/interface/vcpu.h>
++
++#if defined(__i386__) || defined(__x86_64__)
++
++/*
++ * Power off function, if any
++ */
++void (*pm_power_off)(void);
++EXPORT_SYMBOL(pm_power_off);
++
++void machine_emergency_restart(void)
++{
++ /* We really want to get pending console data out before we die. */
++ xencons_force_flush();
++ HYPERVISOR_shutdown(SHUTDOWN_reboot);
++}
++
++void machine_restart(char * __unused)
++{
++ machine_emergency_restart();
++}
++
++void machine_halt(void)
++{
++ machine_power_off();
++}
++
++void machine_power_off(void)
++{
++ /* We really want to get pending console data out before we die. */
++ xencons_force_flush();
++ if (pm_power_off)
++ pm_power_off();
++ HYPERVISOR_shutdown(SHUTDOWN_poweroff);
++}
++
++int reboot_thru_bios = 0; /* for dmi_scan.c */
++EXPORT_SYMBOL(machine_restart);
++EXPORT_SYMBOL(machine_halt);
++EXPORT_SYMBOL(machine_power_off);
++
++static void pre_suspend(void)
++{
++ HYPERVISOR_shared_info = (shared_info_t *)empty_zero_page;
++ WARN_ON(HYPERVISOR_update_va_mapping(fix_to_virt(FIX_SHARED_INFO),
++ __pte_ma(0), 0));
++
++ xen_start_info->store_mfn = mfn_to_pfn(xen_start_info->store_mfn);
++ xen_start_info->console.domU.mfn =
++ mfn_to_pfn(xen_start_info->console.domU.mfn);
++}
++
++static void post_suspend(int suspend_cancelled)
++{
++ int i, j, k, fpp;
++ unsigned long shinfo_mfn;
++ extern unsigned long max_pfn;
++ extern unsigned long *pfn_to_mfn_frame_list_list;
++ extern unsigned long *pfn_to_mfn_frame_list[];
++
++ if (suspend_cancelled) {
++ xen_start_info->store_mfn =
++ pfn_to_mfn(xen_start_info->store_mfn);
++ xen_start_info->console.domU.mfn =
++ pfn_to_mfn(xen_start_info->console.domU.mfn);
++ } else {
++#ifdef CONFIG_SMP
++ cpu_initialized_map = cpu_online_map;
++#endif
++ }
++
++ shinfo_mfn = xen_start_info->shared_info >> PAGE_SHIFT;
++ if (HYPERVISOR_update_va_mapping(fix_to_virt(FIX_SHARED_INFO),
++ pfn_pte_ma(shinfo_mfn, PAGE_KERNEL),
++ 0))
++ BUG();
++ HYPERVISOR_shared_info = (shared_info_t *)fix_to_virt(FIX_SHARED_INFO);
++
++ memset(empty_zero_page, 0, PAGE_SIZE);
++
++ fpp = PAGE_SIZE/sizeof(unsigned long);
++ for (i = 0, j = 0, k = -1; i < max_pfn; i += fpp, j++) {
++ if ((j % fpp) == 0) {
++ k++;
++ pfn_to_mfn_frame_list_list[k] =
++ virt_to_mfn(pfn_to_mfn_frame_list[k]);
++ j = 0;
++ }
++ pfn_to_mfn_frame_list[k][j] =
++ virt_to_mfn(&phys_to_machine_mapping[i]);
++ }
++ HYPERVISOR_shared_info->arch.max_pfn = max_pfn;
++ HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list =
++ virt_to_mfn(pfn_to_mfn_frame_list_list);
++}
++
++#else /* !(defined(__i386__) || defined(__x86_64__)) */
++
++#ifndef HAVE_XEN_PRE_SUSPEND
++#define xen_pre_suspend() ((void)0)
++#endif
++
++#ifndef HAVE_XEN_POST_SUSPEND
++#define xen_post_suspend(x) ((void)0)
++#endif
++
++#define switch_idle_mm() ((void)0)
++#define mm_pin_all() ((void)0)
++#define pre_suspend() xen_pre_suspend()
++#define post_suspend(x) xen_post_suspend(x)
++
++#endif
++
++struct suspend {
++ int fast_suspend;
++ void (*resume_notifier)(void);
++};
++
++static int take_machine_down(void *_suspend)
++{
++ struct suspend *suspend = _suspend;
++ int suspend_cancelled, err;
++ extern void time_resume(void);
++
++ if (suspend->fast_suspend) {
++ BUG_ON(!irqs_disabled());
++ } else {
++ BUG_ON(irqs_disabled());
++
++ for (;;) {
++ err = smp_suspend();
++ if (err)
++ return err;
++
++ xenbus_suspend();
++ preempt_disable();
++
++ if (num_online_cpus() == 1)
++ break;
++
++ preempt_enable();
++ xenbus_suspend_cancel();
++ }
++
++ local_irq_disable();
++ }
++
++ mm_pin_all();
++ gnttab_suspend();
++ pre_suspend();
++
++ /*
++ * This hypercall returns 1 if suspend was cancelled or the domain was
++ * merely checkpointed, and 0 if it is resuming in a new domain.
++ */
++ suspend_cancelled = HYPERVISOR_suspend(virt_to_mfn(xen_start_info));
++
++ suspend->resume_notifier();
++ post_suspend(suspend_cancelled);
++ gnttab_resume();
++ if (!suspend_cancelled) {
++ irq_resume();
++#ifdef __x86_64__
++ /*
++ * Older versions of Xen do not save/restore the user %cr3.
++ * We do it here just in case, but there's no need if we are
++ * in fast-suspend mode as that implies a new enough Xen.
++ */
++ if (!suspend->fast_suspend) {
++ struct mmuext_op op;
++ op.cmd = MMUEXT_NEW_USER_BASEPTR;
++ op.arg1.mfn = pfn_to_mfn(__pa(__user_pgd(
++ current->active_mm->pgd)) >> PAGE_SHIFT);
++ if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
++ BUG();
++ }
++#endif
++ }
++ time_resume();
++
++ if (!suspend->fast_suspend)
++ local_irq_enable();
++
++ return suspend_cancelled;
++}
++
++int __xen_suspend(int fast_suspend, void (*resume_notifier)(void))
++{
++ int err, suspend_cancelled;
++ struct suspend suspend;
++
++ BUG_ON(smp_processor_id() != 0);
++ BUG_ON(in_interrupt());
++
++#if defined(__i386__) || defined(__x86_64__)
++ if (xen_feature(XENFEAT_auto_translated_physmap)) {
++ printk(KERN_WARNING "Cannot suspend in "
++ "auto_translated_physmap mode.\n");
++ return -EOPNOTSUPP;
++ }
++#endif
++
++ /* If we are definitely UP then 'slow mode' is actually faster. */
++ if (num_possible_cpus() == 1)
++ fast_suspend = 0;
++
++ suspend.fast_suspend = fast_suspend;
++ suspend.resume_notifier = resume_notifier;
++
++ if (fast_suspend) {
++ xenbus_suspend();
++ err = stop_machine_run(take_machine_down, &suspend, 0);
++ if (err < 0)
++ xenbus_suspend_cancel();
++ } else {
++ err = take_machine_down(&suspend);
++ }
++
++ if (err < 0)
++ return err;
++
++ suspend_cancelled = err;
++ if (!suspend_cancelled) {
++ xencons_resume();
++ xenbus_resume();
++ } else {
++ xenbus_suspend_cancel();
++ }
++
++ if (!fast_suspend)
++ smp_resume();
++
++ return 0;
++}
+--- linux-2.6.18.8/drivers/xen/core/reboot.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/reboot.c 2008-05-19 00:33:45.822787398 +0300
+@@ -0,0 +1,272 @@
++#define __KERNEL_SYSCALLS__
++#include <linux/version.h>
++#include <linux/kernel.h>
++#include <linux/unistd.h>
++#include <linux/module.h>
++#include <linux/reboot.h>
++#include <linux/sysrq.h>
++#include <asm/hypervisor.h>
++#include <xen/xenbus.h>
++#include <linux/kmod.h>
++#include <linux/slab.h>
++#include <linux/workqueue.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++MODULE_LICENSE("Dual BSD/GPL");
++
++#define SHUTDOWN_INVALID -1
++#define SHUTDOWN_POWEROFF 0
++#define SHUTDOWN_SUSPEND 2
++#define SHUTDOWN_RESUMING 3
++#define SHUTDOWN_HALT 4
++
++/* Ignore multiple shutdown requests. */
++static int shutting_down = SHUTDOWN_INVALID;
++
++/* Can we leave APs online when we suspend? */
++static int fast_suspend;
++
++static void __shutdown_handler(void *unused);
++static DECLARE_WORK(shutdown_work, __shutdown_handler, NULL);
++
++int __xen_suspend(int fast_suspend, void (*resume_notifier)(void));
++
++static int shutdown_process(void *__unused)
++{
++ static char *envp[] = { "HOME=/", "TERM=linux",
++ "PATH=/sbin:/usr/sbin:/bin:/usr/bin", NULL };
++ static char *poweroff_argv[] = { "/sbin/poweroff", NULL };
++
++ extern asmlinkage long sys_reboot(int magic1, int magic2,
++ unsigned int cmd, void *arg);
++
++ if ((shutting_down == SHUTDOWN_POWEROFF) ||
++ (shutting_down == SHUTDOWN_HALT)) {
++ if (call_usermodehelper("/sbin/poweroff", poweroff_argv,
++ envp, 0) < 0) {
++#ifdef CONFIG_XEN
++ sys_reboot(LINUX_REBOOT_MAGIC1,
++ LINUX_REBOOT_MAGIC2,
++ LINUX_REBOOT_CMD_POWER_OFF,
++ NULL);
++#endif /* CONFIG_XEN */
++ }
++ }
++
++ shutting_down = SHUTDOWN_INVALID; /* could try again */
++
++ return 0;
++}
++
++static void xen_resume_notifier(void)
++{
++ int old_state = xchg(&shutting_down, SHUTDOWN_RESUMING);
++ BUG_ON(old_state != SHUTDOWN_SUSPEND);
++}
++
++static int xen_suspend(void *__unused)
++{
++ int err, old_state;
++
++ daemonize("suspend");
++ err = set_cpus_allowed(current, cpumask_of_cpu(0));
++ if (err) {
++ printk(KERN_ERR "Xen suspend can't run on CPU0 (%d)\n", err);
++ goto fail;
++ }
++
++ do {
++ err = __xen_suspend(fast_suspend, xen_resume_notifier);
++ if (err) {
++ printk(KERN_ERR "Xen suspend failed (%d)\n", err);
++ goto fail;
++ }
++ old_state = cmpxchg(
++ &shutting_down, SHUTDOWN_RESUMING, SHUTDOWN_INVALID);
++ } while (old_state == SHUTDOWN_SUSPEND);
++
++ switch (old_state) {
++ case SHUTDOWN_INVALID:
++ case SHUTDOWN_SUSPEND:
++ BUG();
++ case SHUTDOWN_RESUMING:
++ break;
++ default:
++ schedule_work(&shutdown_work);
++ break;
++ }
++
++ return 0;
++
++ fail:
++ old_state = xchg(&shutting_down, SHUTDOWN_INVALID);
++ BUG_ON(old_state != SHUTDOWN_SUSPEND);
++ return 0;
++}
++
++static void __shutdown_handler(void *unused)
++{
++ int err;
++
++ err = kernel_thread((shutting_down == SHUTDOWN_SUSPEND) ?
++ xen_suspend : shutdown_process,
++ NULL, CLONE_FS | CLONE_FILES);
++
++ if (err < 0) {
++ printk(KERN_WARNING "Error creating shutdown process (%d): "
++ "retrying...\n", -err);
++ schedule_delayed_work(&shutdown_work, HZ/2);
++ }
++}
++
++static void shutdown_handler(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ extern void ctrl_alt_del(void);
++ char *str;
++ struct xenbus_transaction xbt;
++ int err, old_state, new_state = SHUTDOWN_INVALID;
++
++ if ((shutting_down != SHUTDOWN_INVALID) &&
++ (shutting_down != SHUTDOWN_RESUMING))
++ return;
++
++ again:
++ err = xenbus_transaction_start(&xbt);
++ if (err)
++ return;
++
++ str = (char *)xenbus_read(xbt, "control", "shutdown", NULL);
++ /* Ignore read errors and empty reads. */
++ if (XENBUS_IS_ERR_READ(str)) {
++ xenbus_transaction_end(xbt, 1);
++ return;
++ }
++
++ xenbus_write(xbt, "control", "shutdown", "");
++
++ err = xenbus_transaction_end(xbt, 0);
++ if (err == -EAGAIN) {
++ kfree(str);
++ goto again;
++ }
++
++ if (strcmp(str, "poweroff") == 0)
++ new_state = SHUTDOWN_POWEROFF;
++ else if (strcmp(str, "reboot") == 0)
++ ctrl_alt_del();
++ else if (strcmp(str, "suspend") == 0)
++ new_state = SHUTDOWN_SUSPEND;
++ else if (strcmp(str, "halt") == 0)
++ new_state = SHUTDOWN_HALT;
++ else
++ printk("Ignoring shutdown request: %s\n", str);
++
++ if (new_state != SHUTDOWN_INVALID) {
++ old_state = xchg(&shutting_down, new_state);
++ if (old_state == SHUTDOWN_INVALID)
++ schedule_work(&shutdown_work);
++ else
++ BUG_ON(old_state != SHUTDOWN_RESUMING);
++ }
++
++ kfree(str);
++}
++
++static void sysrq_handler(struct xenbus_watch *watch, const char **vec,
++ unsigned int len)
++{
++ char sysrq_key = '\0';
++ struct xenbus_transaction xbt;
++ int err;
++
++ again:
++ err = xenbus_transaction_start(&xbt);
++ if (err)
++ return;
++ if (!xenbus_scanf(xbt, "control", "sysrq", "%c", &sysrq_key)) {
++ printk(KERN_ERR "Unable to read sysrq code in "
++ "control/sysrq\n");
++ xenbus_transaction_end(xbt, 1);
++ return;
++ }
++
++ if (sysrq_key != '\0')
++ xenbus_printf(xbt, "control", "sysrq", "%c", '\0');
++
++ err = xenbus_transaction_end(xbt, 0);
++ if (err == -EAGAIN)
++ goto again;
++
++#ifdef CONFIG_MAGIC_SYSRQ
++ if (sysrq_key != '\0')
++ handle_sysrq(sysrq_key, NULL, NULL);
++#endif
++}
++
++static struct xenbus_watch shutdown_watch = {
++ .node = "control/shutdown",
++ .callback = shutdown_handler
++};
++
++static struct xenbus_watch sysrq_watch = {
++ .node = "control/sysrq",
++ .callback = sysrq_handler
++};
++
++static int setup_shutdown_watcher(void)
++{
++ int err;
++
++ xenbus_scanf(XBT_NIL, "control",
++ "platform-feature-multiprocessor-suspend",
++ "%d", &fast_suspend);
++
++ err = register_xenbus_watch(&shutdown_watch);
++ if (err) {
++ printk(KERN_ERR "Failed to set shutdown watcher\n");
++ return err;
++ }
++
++ err = register_xenbus_watch(&sysrq_watch);
++ if (err) {
++ printk(KERN_ERR "Failed to set sysrq watcher\n");
++ return err;
++ }
++
++ return 0;
++}
++
++#ifdef CONFIG_XEN
++
++static int shutdown_event(struct notifier_block *notifier,
++ unsigned long event,
++ void *data)
++{
++ setup_shutdown_watcher();
++ return NOTIFY_DONE;
++}
++
++static int __init setup_shutdown_event(void)
++{
++ static struct notifier_block xenstore_notifier = {
++ .notifier_call = shutdown_event
++ };
++ register_xenstore_notifier(&xenstore_notifier);
++
++ return 0;
++}
++
++subsys_initcall(setup_shutdown_event);
++
++#else /* !defined(CONFIG_XEN) */
++
++int xen_reboot_init(void)
++{
++ return setup_shutdown_watcher();
++}
++
++#endif /* !defined(CONFIG_XEN) */
+--- linux-2.6.18.8/drivers/xen/core/smpboot.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/smpboot.c 2008-05-19 00:33:45.826787629 +0300
+@@ -0,0 +1,464 @@
++/*
++ * Xen SMP booting functions
++ *
++ * See arch/i386/kernel/smpboot.c for copyright and credits for derived
++ * portions of this file.
++ */
++
++#include <linux/module.h>
++#include <linux/init.h>
++#include <linux/kernel.h>
++#include <linux/mm.h>
++#include <linux/sched.h>
++#include <linux/kernel_stat.h>
++#include <linux/smp_lock.h>
++#include <linux/irq.h>
++#include <linux/bootmem.h>
++#include <linux/notifier.h>
++#include <linux/cpu.h>
++#include <linux/percpu.h>
++#include <asm/desc.h>
++#include <asm/arch_hooks.h>
++#include <asm/pgalloc.h>
++#include <xen/evtchn.h>
++#include <xen/interface/vcpu.h>
++#include <xen/cpu_hotplug.h>
++#include <xen/xenbus.h>
++
++extern irqreturn_t smp_reschedule_interrupt(int, void *, struct pt_regs *);
++extern irqreturn_t smp_call_function_interrupt(int, void *, struct pt_regs *);
++
++extern int local_setup_timer(unsigned int cpu);
++extern void local_teardown_timer(unsigned int cpu);
++
++extern void hypervisor_callback(void);
++extern void failsafe_callback(void);
++extern void system_call(void);
++extern void smp_trap_init(trap_info_t *);
++
++/* Number of siblings per CPU package */
++int smp_num_siblings = 1;
++
++cpumask_t cpu_online_map;
++EXPORT_SYMBOL(cpu_online_map);
++cpumask_t cpu_possible_map;
++EXPORT_SYMBOL(cpu_possible_map);
++cpumask_t cpu_initialized_map;
++
++struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;
++EXPORT_SYMBOL(cpu_data);
++
++#ifdef CONFIG_HOTPLUG_CPU
++DEFINE_PER_CPU(int, cpu_state) = { 0 };
++#endif
++
++static DEFINE_PER_CPU(int, resched_irq);
++static DEFINE_PER_CPU(int, callfunc_irq);
++static char resched_name[NR_CPUS][15];
++static char callfunc_name[NR_CPUS][15];
++
++u8 cpu_2_logical_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
++
++cpumask_t cpu_sibling_map[NR_CPUS] __cacheline_aligned;
++cpumask_t cpu_core_map[NR_CPUS] __cacheline_aligned;
++EXPORT_SYMBOL(cpu_core_map);
++
++#if defined(__i386__)
++u8 x86_cpu_to_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = 0xff };
++EXPORT_SYMBOL(x86_cpu_to_apicid);
++#elif !defined(CONFIG_X86_IO_APIC)
++unsigned int maxcpus = NR_CPUS;
++#endif
++
++void __init prefill_possible_map(void)
++{
++ int i, rc;
++
++ for_each_possible_cpu(i)
++ if (i != smp_processor_id())
++ return;
++
++ for (i = 0; i < NR_CPUS; i++) {
++ rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
++ if (rc >= 0)
++ cpu_set(i, cpu_possible_map);
++ }
++}
++
++void __init smp_alloc_memory(void)
++{
++}
++
++static inline void
++set_cpu_sibling_map(unsigned int cpu)
++{
++ cpu_data[cpu].phys_proc_id = cpu;
++ cpu_data[cpu].cpu_core_id = 0;
++
++ cpu_sibling_map[cpu] = cpumask_of_cpu(cpu);
++ cpu_core_map[cpu] = cpumask_of_cpu(cpu);
++
++ cpu_data[cpu].booted_cores = 1;
++}
++
++static void
++remove_siblinginfo(unsigned int cpu)
++{
++ cpu_data[cpu].phys_proc_id = BAD_APICID;
++ cpu_data[cpu].cpu_core_id = BAD_APICID;
++
++ cpus_clear(cpu_sibling_map[cpu]);
++ cpus_clear(cpu_core_map[cpu]);
++
++ cpu_data[cpu].booted_cores = 0;
++}
++
++static int __cpuinit xen_smp_intr_init(unsigned int cpu)
++{
++ int rc;
++
++ per_cpu(resched_irq, cpu) = per_cpu(callfunc_irq, cpu) = -1;
++
++ sprintf(resched_name[cpu], "resched%u", cpu);
++ rc = bind_ipi_to_irqhandler(RESCHEDULE_VECTOR,
++ cpu,
++ smp_reschedule_interrupt,
++ SA_INTERRUPT,
++ resched_name[cpu],
++ NULL);
++ if (rc < 0)
++ goto fail;
++ per_cpu(resched_irq, cpu) = rc;
++
++ sprintf(callfunc_name[cpu], "callfunc%u", cpu);
++ rc = bind_ipi_to_irqhandler(CALL_FUNCTION_VECTOR,
++ cpu,
++ smp_call_function_interrupt,
++ SA_INTERRUPT,
++ callfunc_name[cpu],
++ NULL);
++ if (rc < 0)
++ goto fail;
++ per_cpu(callfunc_irq, cpu) = rc;
++
++ if ((cpu != 0) && ((rc = local_setup_timer(cpu)) != 0))
++ goto fail;
++
++ return 0;
++
++ fail:
++ if (per_cpu(resched_irq, cpu) >= 0)
++ unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
++ if (per_cpu(callfunc_irq, cpu) >= 0)
++ unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
++ return rc;
++}
++
++#ifdef CONFIG_HOTPLUG_CPU
++static void xen_smp_intr_exit(unsigned int cpu)
++{
++ if (cpu != 0)
++ local_teardown_timer(cpu);
++
++ unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
++ unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
++}
++#endif
++
++void __cpuinit cpu_bringup(void)
++{
++ cpu_init();
++ identify_cpu(cpu_data + smp_processor_id());
++ touch_softlockup_watchdog();
++ preempt_disable();
++ local_irq_enable();
++}
++
++static void __cpuinit cpu_bringup_and_idle(void)
++{
++ cpu_bringup();
++ cpu_idle();
++}
++
++static void __cpuinit cpu_initialize_context(unsigned int cpu)
++{
++ /* vcpu_guest_context_t is too large to allocate on the stack.
++ * Hence we allocate statically and protect it with a lock */
++ static vcpu_guest_context_t ctxt;
++ static DEFINE_SPINLOCK(ctxt_lock);
++
++ struct task_struct *idle = idle_task(cpu);
++#ifdef __x86_64__
++ struct desc_ptr *gdt_descr = &cpu_gdt_descr[cpu];
++#else
++ struct Xgt_desc_struct *gdt_descr = &per_cpu(cpu_gdt_descr, cpu);
++#endif
++
++ if (cpu_test_and_set(cpu, cpu_initialized_map))
++ return;
++
++ spin_lock(&ctxt_lock);
++
++ memset(&ctxt, 0, sizeof(ctxt));
++
++ ctxt.flags = VGCF_IN_KERNEL;
++ ctxt.user_regs.ds = __USER_DS;
++ ctxt.user_regs.es = __USER_DS;
++ ctxt.user_regs.fs = 0;
++ ctxt.user_regs.gs = 0;
++ ctxt.user_regs.ss = __KERNEL_DS;
++ ctxt.user_regs.eip = (unsigned long)cpu_bringup_and_idle;
++ ctxt.user_regs.eflags = X86_EFLAGS_IF | 0x1000; /* IOPL_RING1 */
++
++ memset(&ctxt.fpu_ctxt, 0, sizeof(ctxt.fpu_ctxt));
++
++ smp_trap_init(ctxt.trap_ctxt);
++
++ ctxt.ldt_ents = 0;
++
++ ctxt.gdt_frames[0] = virt_to_mfn(gdt_descr->address);
++ ctxt.gdt_ents = gdt_descr->size / 8;
++
++#ifdef __i386__
++ ctxt.user_regs.cs = __KERNEL_CS;
++ ctxt.user_regs.esp = idle->thread.esp0 - sizeof(struct pt_regs);
++
++ ctxt.kernel_ss = __KERNEL_DS;
++ ctxt.kernel_sp = idle->thread.esp0;
++
++ ctxt.event_callback_cs = __KERNEL_CS;
++ ctxt.event_callback_eip = (unsigned long)hypervisor_callback;
++ ctxt.failsafe_callback_cs = __KERNEL_CS;
++ ctxt.failsafe_callback_eip = (unsigned long)failsafe_callback;
++
++ ctxt.ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
++#else /* __x86_64__ */
++ ctxt.user_regs.cs = __KERNEL_CS;
++ ctxt.user_regs.esp = idle->thread.rsp0 - sizeof(struct pt_regs);
++
++ ctxt.kernel_ss = __KERNEL_DS;
++ ctxt.kernel_sp = idle->thread.rsp0;
++
++ ctxt.event_callback_eip = (unsigned long)hypervisor_callback;
++ ctxt.failsafe_callback_eip = (unsigned long)failsafe_callback;
++ ctxt.syscall_callback_eip = (unsigned long)system_call;
++
++ ctxt.ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(init_level4_pgt));
++
++ ctxt.gs_base_kernel = (unsigned long)(cpu_pda(cpu));
++#endif
++
++ if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, &ctxt))
++ BUG();
++
++ spin_unlock(&ctxt_lock);
++}
++
++void __init smp_prepare_cpus(unsigned int max_cpus)
++{
++ unsigned int cpu;
++ struct task_struct *idle;
++ int apicid, acpiid;
++ struct vcpu_get_physid cpu_id;
++#ifdef __x86_64__
++ struct desc_ptr *gdt_descr;
++#else
++ struct Xgt_desc_struct *gdt_descr;
++#endif
++
++ apicid = 0;
++ if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, 0, &cpu_id) == 0) {
++ apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
++ acpiid = xen_vcpu_physid_to_x86_acpiid(cpu_id.phys_id);
++#ifdef CONFIG_ACPI
++ if (acpiid != 0xff)
++ x86_acpiid_to_apicid[acpiid] = apicid;
++#endif
++ }
++ boot_cpu_data.apicid = apicid;
++ cpu_data[0] = boot_cpu_data;
++
++ cpu_2_logical_apicid[0] = apicid;
++ x86_cpu_to_apicid[0] = apicid;
++
++ current_thread_info()->cpu = 0;
++
++ for (cpu = 0; cpu < NR_CPUS; cpu++) {
++ cpus_clear(cpu_sibling_map[cpu]);
++ cpus_clear(cpu_core_map[cpu]);
++ }
++
++ set_cpu_sibling_map(0);
++
++ if (xen_smp_intr_init(0))
++ BUG();
++
++ cpu_initialized_map = cpumask_of_cpu(0);
++
++ /* Restrict the possible_map according to max_cpus. */
++ while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
++ for (cpu = NR_CPUS-1; !cpu_isset(cpu, cpu_possible_map); cpu--)
++ continue;
++ cpu_clear(cpu, cpu_possible_map);
++ }
++
++ for_each_possible_cpu (cpu) {
++ if (cpu == 0)
++ continue;
++
++#ifdef __x86_64__
++ gdt_descr = &cpu_gdt_descr[cpu];
++#else
++ gdt_descr = &per_cpu(cpu_gdt_descr, cpu);
++#endif
++ gdt_descr->address = get_zeroed_page(GFP_KERNEL);
++ if (unlikely(!gdt_descr->address)) {
++ printk(KERN_CRIT "CPU%d failed to allocate GDT\n",
++ cpu);
++ continue;
++ }
++ gdt_descr->size = GDT_SIZE;
++ memcpy((void *)gdt_descr->address, cpu_gdt_table, GDT_SIZE);
++ make_page_readonly(
++ (void *)gdt_descr->address,
++ XENFEAT_writable_descriptor_tables);
++
++ apicid = cpu;
++ if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, cpu, &cpu_id) == 0) {
++ apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
++ acpiid = xen_vcpu_physid_to_x86_acpiid(cpu_id.phys_id);
++#ifdef CONFIG_ACPI
++ if (acpiid != 0xff)
++ x86_acpiid_to_apicid[acpiid] = apicid;
++#endif
++ }
++ cpu_data[cpu] = boot_cpu_data;
++ cpu_data[cpu].apicid = apicid;
++
++ cpu_2_logical_apicid[cpu] = apicid;
++ x86_cpu_to_apicid[cpu] = apicid;
++
++ idle = fork_idle(cpu);
++ if (IS_ERR(idle))
++ panic("failed fork for CPU %d", cpu);
++
++#ifdef __x86_64__
++ cpu_pda(cpu)->pcurrent = idle;
++ cpu_pda(cpu)->cpunumber = cpu;
++ clear_ti_thread_flag(idle->thread_info, TIF_FORK);
++#endif
++
++ irq_ctx_init(cpu);
++
++#ifdef CONFIG_HOTPLUG_CPU
++ if (is_initial_xendomain())
++ cpu_set(cpu, cpu_present_map);
++#else
++ cpu_set(cpu, cpu_present_map);
++#endif
++ }
++
++ init_xenbus_allowed_cpumask();
++
++#ifdef CONFIG_X86_IO_APIC
++ /*
++ * Here we can be sure that there is an IO-APIC in the system. Let's
++ * go and set it up:
++ */
++ if (!skip_ioapic_setup && nr_ioapics)
++ setup_IO_APIC();
++#endif
++}
++
++void __devinit smp_prepare_boot_cpu(void)
++{
++ prefill_possible_map();
++}
++
++#ifdef CONFIG_HOTPLUG_CPU
++
++/*
++ * Initialize cpu_present_map late to skip SMP boot code in init/main.c.
++ * But do it early enough to catch critical for_each_present_cpu() loops
++ * in i386-specific code.
++ */
++static int __init initialize_cpu_present_map(void)
++{
++ cpu_present_map = cpu_possible_map;
++ return 0;
++}
++core_initcall(initialize_cpu_present_map);
++
++int __cpu_disable(void)
++{
++ cpumask_t map = cpu_online_map;
++ unsigned int cpu = smp_processor_id();
++
++ if (cpu == 0)
++ return -EBUSY;
++
++ remove_siblinginfo(cpu);
++
++ cpu_clear(cpu, map);
++ fixup_irqs(map);
++ cpu_clear(cpu, cpu_online_map);
++
++ return 0;
++}
++
++void __cpu_die(unsigned int cpu)
++{
++ while (HYPERVISOR_vcpu_op(VCPUOP_is_up, cpu, NULL)) {
++ current->state = TASK_UNINTERRUPTIBLE;
++ schedule_timeout(HZ/10);
++ }
++
++ xen_smp_intr_exit(cpu);
++
++ if (num_online_cpus() == 1)
++ alternatives_smp_switch(0);
++}
++
++#endif /* CONFIG_HOTPLUG_CPU */
++
++int __cpuinit __cpu_up(unsigned int cpu)
++{
++ int rc;
++
++ rc = cpu_up_check(cpu);
++ if (rc)
++ return rc;
++
++ cpu_initialize_context(cpu);
++
++ if (num_online_cpus() == 1)
++ alternatives_smp_switch(1);
++
++ /* This must be done before setting cpu_online_map */
++ set_cpu_sibling_map(cpu);
++ wmb();
++
++ rc = xen_smp_intr_init(cpu);
++ if (rc) {
++ remove_siblinginfo(cpu);
++ return rc;
++ }
++
++ cpu_set(cpu, cpu_online_map);
++
++ rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
++ BUG_ON(rc);
++
++ return 0;
++}
++
++void __init smp_cpus_done(unsigned int max_cpus)
++{
++}
++
++#ifndef CONFIG_X86_LOCAL_APIC
++int setup_profiling_timer(unsigned int multiplier)
++{
++ return -EINVAL;
++}
++#endif
+--- linux-2.6.18.8/drivers/xen/core/xen_proc.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/xen_proc.c 2008-05-19 00:33:45.826787629 +0300
+@@ -0,0 +1,23 @@
++
++#include <linux/module.h>
++#include <linux/proc_fs.h>
++#include <xen/xen_proc.h>
++
++static struct proc_dir_entry *xen_base;
++
++struct proc_dir_entry *create_xen_proc_entry(const char *name, mode_t mode)
++{
++ if ( xen_base == NULL )
++ if ( (xen_base = proc_mkdir("xen", &proc_root)) == NULL )
++ panic("Couldn't create /proc/xen");
++ return create_proc_entry(name, mode, xen_base);
++}
++
++EXPORT_SYMBOL_GPL(create_xen_proc_entry);
++
++void remove_xen_proc_entry(const char *name)
++{
++ remove_proc_entry(name, xen_base);
++}
++
++EXPORT_SYMBOL_GPL(remove_xen_proc_entry);
+--- linux-2.6.18.8/drivers/xen/core/xen_sysfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/xen_sysfs.c 2008-05-19 00:33:45.826787629 +0300
+@@ -0,0 +1,382 @@
++/*
++ * copyright (c) 2006 IBM Corporation
++ * Authored by: Mike D. Day <ncmike@us.ibm.com>
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License version 2 as
++ * published by the Free Software Foundation.
++ */
++
++#include <linux/err.h>
++#include <linux/kernel.h>
++#include <linux/module.h>
++#include <linux/init.h>
++#include <asm/hypervisor.h>
++#include <xen/features.h>
++#include <xen/hypervisor_sysfs.h>
++#include <xen/xenbus.h>
++
++MODULE_LICENSE("GPL");
++MODULE_AUTHOR("Mike D. Day <ncmike@us.ibm.com>");
++
++static ssize_t type_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ return sprintf(buffer, "xen\n");
++}
++
++HYPERVISOR_ATTR_RO(type);
++
++static int __init xen_sysfs_type_init(void)
++{
++ return sysfs_create_file(&hypervisor_subsys.kset.kobj, &type_attr.attr);
++}
++
++static void xen_sysfs_type_destroy(void)
++{
++ sysfs_remove_file(&hypervisor_subsys.kset.kobj, &type_attr.attr);
++}
++
++/* xen version attributes */
++static ssize_t major_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int version = HYPERVISOR_xen_version(XENVER_version, NULL);
++ if (version)
++ return sprintf(buffer, "%d\n", version >> 16);
++ return -ENODEV;
++}
++
++HYPERVISOR_ATTR_RO(major);
++
++static ssize_t minor_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int version = HYPERVISOR_xen_version(XENVER_version, NULL);
++ if (version)
++ return sprintf(buffer, "%d\n", version & 0xff);
++ return -ENODEV;
++}
++
++HYPERVISOR_ATTR_RO(minor);
++
++static ssize_t extra_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ char *extra;
++
++ extra = kmalloc(XEN_EXTRAVERSION_LEN, GFP_KERNEL);
++ if (extra) {
++ ret = HYPERVISOR_xen_version(XENVER_extraversion, extra);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", extra);
++ kfree(extra);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(extra);
++
++static struct attribute *version_attrs[] = {
++ &major_attr.attr,
++ &minor_attr.attr,
++ &extra_attr.attr,
++ NULL
++};
++
++static struct attribute_group version_group = {
++ .name = "version",
++ .attrs = version_attrs,
++};
++
++static int __init xen_sysfs_version_init(void)
++{
++ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
++ &version_group);
++}
++
++static void xen_sysfs_version_destroy(void)
++{
++ sysfs_remove_group(&hypervisor_subsys.kset.kobj, &version_group);
++}
++
++/* UUID */
++
++static ssize_t uuid_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ char *vm, *val;
++ int ret;
++ extern int xenstored_ready;
++
++ if (!xenstored_ready)
++ return -EBUSY;
++
++ vm = xenbus_read(XBT_NIL, "vm", "", NULL);
++ if (IS_ERR(vm))
++ return PTR_ERR(vm);
++ val = xenbus_read(XBT_NIL, vm, "uuid", NULL);
++ kfree(vm);
++ if (IS_ERR(val))
++ return PTR_ERR(val);
++ ret = sprintf(buffer, "%s\n", val);
++ kfree(val);
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(uuid);
++
++static int __init xen_sysfs_uuid_init(void)
++{
++ return sysfs_create_file(&hypervisor_subsys.kset.kobj, &uuid_attr.attr);
++}
++
++static void xen_sysfs_uuid_destroy(void)
++{
++ sysfs_remove_file(&hypervisor_subsys.kset.kobj, &uuid_attr.attr);
++}
++
++/* xen compilation attributes */
++
++static ssize_t compiler_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ struct xen_compile_info *info;
++
++ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
++ if (info) {
++ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", info->compiler);
++ kfree(info);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(compiler);
++
++static ssize_t compiled_by_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ struct xen_compile_info *info;
++
++ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
++ if (info) {
++ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", info->compile_by);
++ kfree(info);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(compiled_by);
++
++static ssize_t compile_date_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ struct xen_compile_info *info;
++
++ info = kmalloc(sizeof(struct xen_compile_info), GFP_KERNEL);
++ if (info) {
++ ret = HYPERVISOR_xen_version(XENVER_compile_info, info);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", info->compile_date);
++ kfree(info);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(compile_date);
++
++static struct attribute *xen_compile_attrs[] = {
++ &compiler_attr.attr,
++ &compiled_by_attr.attr,
++ &compile_date_attr.attr,
++ NULL
++};
++
++static struct attribute_group xen_compilation_group = {
++ .name = "compilation",
++ .attrs = xen_compile_attrs,
++};
++
++int __init static xen_compilation_init(void)
++{
++ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
++ &xen_compilation_group);
++}
++
++static void xen_compilation_destroy(void)
++{
++ sysfs_remove_group(&hypervisor_subsys.kset.kobj,
++ &xen_compilation_group);
++}
++
++/* xen properties info */
++
++static ssize_t capabilities_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ char *caps;
++
++ caps = kmalloc(XEN_CAPABILITIES_INFO_LEN, GFP_KERNEL);
++ if (caps) {
++ ret = HYPERVISOR_xen_version(XENVER_capabilities, caps);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", caps);
++ kfree(caps);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(capabilities);
++
++static ssize_t changeset_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ char *cset;
++
++ cset = kmalloc(XEN_CHANGESET_INFO_LEN, GFP_KERNEL);
++ if (cset) {
++ ret = HYPERVISOR_xen_version(XENVER_changeset, cset);
++ if (!ret)
++ ret = sprintf(buffer, "%s\n", cset);
++ kfree(cset);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(changeset);
++
++static ssize_t virtual_start_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret = -ENOMEM;
++ struct xen_platform_parameters *parms;
++
++ parms = kmalloc(sizeof(struct xen_platform_parameters), GFP_KERNEL);
++ if (parms) {
++ ret = HYPERVISOR_xen_version(XENVER_platform_parameters,
++ parms);
++ if (!ret)
++ ret = sprintf(buffer, "%lx\n", parms->virt_start);
++ kfree(parms);
++ }
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(virtual_start);
++
++static ssize_t pagesize_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ int ret;
++
++ ret = HYPERVISOR_xen_version(XENVER_pagesize, NULL);
++ if (ret > 0)
++ ret = sprintf(buffer, "%x\n", ret);
++
++ return ret;
++}
++
++HYPERVISOR_ATTR_RO(pagesize);
++
++/* eventually there will be several more features to export */
++static ssize_t xen_feature_show(int index, char *buffer)
++{
++ int ret = -ENOMEM;
++ struct xen_feature_info *info;
++
++ info = kmalloc(sizeof(struct xen_feature_info), GFP_KERNEL);
++ if (info) {
++ info->submap_idx = index;
++ ret = HYPERVISOR_xen_version(XENVER_get_features, info);
++ if (!ret)
++ ret = sprintf(buffer, "%d\n", info->submap);
++ kfree(info);
++ }
++
++ return ret;
++}
++
++static ssize_t writable_pt_show(struct hyp_sysfs_attr *attr, char *buffer)
++{
++ return xen_feature_show(XENFEAT_writable_page_tables, buffer);
++}
++
++HYPERVISOR_ATTR_RO(writable_pt);
++
++static struct attribute *xen_properties_attrs[] = {
++ &capabilities_attr.attr,
++ &changeset_attr.attr,
++ &virtual_start_attr.attr,
++ &pagesize_attr.attr,
++ &writable_pt_attr.attr,
++ NULL
++};
++
++static struct attribute_group xen_properties_group = {
++ .name = "properties",
++ .attrs = xen_properties_attrs,
++};
++
++static int __init xen_properties_init(void)
++{
++ return sysfs_create_group(&hypervisor_subsys.kset.kobj,
++ &xen_properties_group);
++}
++
++static void xen_properties_destroy(void)
++{
++ sysfs_remove_group(&hypervisor_subsys.kset.kobj,
++ &xen_properties_group);
++}
++
++static int __init hyper_sysfs_init(void)
++{
++ int ret;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ ret = xen_sysfs_type_init();
++ if (ret)
++ goto out;
++ ret = xen_sysfs_version_init();
++ if (ret)
++ goto version_out;
++ ret = xen_compilation_init();
++ if (ret)
++ goto comp_out;
++ ret = xen_sysfs_uuid_init();
++ if (ret)
++ goto uuid_out;
++ ret = xen_properties_init();
++ if (!ret)
++ goto out;
++
++ xen_sysfs_uuid_destroy();
++uuid_out:
++ xen_compilation_destroy();
++comp_out:
++ xen_sysfs_version_destroy();
++version_out:
++ xen_sysfs_type_destroy();
++out:
++ return ret;
++}
++
++static void __exit hyper_sysfs_exit(void)
++{
++ xen_properties_destroy();
++ xen_compilation_destroy();
++ xen_sysfs_uuid_destroy();
++ xen_sysfs_version_destroy();
++ xen_sysfs_type_destroy();
++
++}
++
++module_init(hyper_sysfs_init);
++module_exit(hyper_sysfs_exit);
+--- linux-2.6.18.8/drivers/xen/core/xencomm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/core/xencomm.c 2008-05-19 00:33:45.826787629 +0300
+@@ -0,0 +1,229 @@
++/*
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ * Copyright (C) IBM Corp. 2006
++ *
++ * Authors: Hollis Blanchard <hollisb@us.ibm.com>
++ */
++
++#include <linux/gfp.h>
++#include <linux/mm.h>
++#include <asm/page.h>
++#include <xen/xencomm.h>
++#include <xen/interface/xen.h>
++#ifdef __ia64__
++#include <asm/xen/xencomm.h> /* for is_kern_addr() */
++#endif
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++static int xencomm_init(struct xencomm_desc *desc,
++ void *buffer, unsigned long bytes)
++{
++ unsigned long recorded = 0;
++ int i = 0;
++
++ while ((recorded < bytes) && (i < desc->nr_addrs)) {
++ unsigned long vaddr = (unsigned long)buffer + recorded;
++ unsigned long paddr;
++ int offset;
++ int chunksz;
++
++ offset = vaddr % PAGE_SIZE; /* handle partial pages */
++ chunksz = min(PAGE_SIZE - offset, bytes - recorded);
++
++ paddr = xencomm_vtop(vaddr);
++ if (paddr == ~0UL) {
++ printk("%s: couldn't translate vaddr %lx\n",
++ __func__, vaddr);
++ return -EINVAL;
++ }
++
++ desc->address[i++] = paddr;
++ recorded += chunksz;
++ }
++
++ if (recorded < bytes) {
++ printk("%s: could only translate %ld of %ld bytes\n",
++ __func__, recorded, bytes);
++ return -ENOSPC;
++ }
++
++ /* mark remaining addresses invalid (just for safety) */
++ while (i < desc->nr_addrs)
++ desc->address[i++] = XENCOMM_INVALID;
++
++ desc->magic = XENCOMM_MAGIC;
++
++ return 0;
++}
++
++static struct xencomm_desc *xencomm_alloc(gfp_t gfp_mask,
++ void *buffer, unsigned long bytes)
++{
++ struct xencomm_desc *desc;
++ unsigned long buffer_ulong = (unsigned long)buffer;
++ unsigned long start = buffer_ulong & PAGE_MASK;
++ unsigned long end = (buffer_ulong + bytes) | ~PAGE_MASK;
++ unsigned long nr_addrs = (end - start + 1) >> PAGE_SHIFT;
++ unsigned long size = sizeof(*desc) +
++ sizeof(desc->address[0]) * nr_addrs;
++
++ /*
++ * slab allocator returns at least sizeof(void*) aligned pointer.
++ * When sizeof(*desc) > sizeof(void*), struct xencomm_desc might
++ * cross page boundary.
++ */
++ if (sizeof(*desc) > sizeof(void*)) {
++ unsigned long order = get_order(size);
++ desc = (struct xencomm_desc *)__get_free_pages(gfp_mask,
++ order);
++ if (desc == NULL)
++ return NULL;
++
++ desc->nr_addrs =
++ ((PAGE_SIZE << order) - sizeof(struct xencomm_desc)) /
++ sizeof(*desc->address);
++ } else {
++ desc = kmalloc(size, gfp_mask);
++ if (desc == NULL)
++ return NULL;
++
++ desc->nr_addrs = nr_addrs;
++ }
++ return desc;
++}
++
++void xencomm_free(struct xencomm_handle *desc)
++{
++ if (desc && !((ulong)desc & XENCOMM_INLINE_FLAG)) {
++ struct xencomm_desc *desc__ = (struct xencomm_desc*)desc;
++ if (sizeof(*desc__) > sizeof(void*)) {
++ unsigned long size = sizeof(*desc__) +
++ sizeof(desc__->address[0]) * desc__->nr_addrs;
++ unsigned long order = get_order(size);
++ free_pages((unsigned long)__va(desc), order);
++ } else
++ kfree(__va(desc));
++ }
++}
++
++static int xencomm_create(void *buffer, unsigned long bytes, struct xencomm_desc **ret, gfp_t gfp_mask)
++{
++ struct xencomm_desc *desc;
++ int rc;
++
++ pr_debug("%s: %p[%ld]\n", __func__, buffer, bytes);
++
++ if (bytes == 0) {
++ /* don't create a descriptor; Xen recognizes NULL. */
++ BUG_ON(buffer != NULL);
++ *ret = NULL;
++ return 0;
++ }
++
++ BUG_ON(buffer == NULL); /* 'bytes' is non-zero */
++
++ desc = xencomm_alloc(gfp_mask, buffer, bytes);
++ if (!desc) {
++ printk("%s failure\n", "xencomm_alloc");
++ return -ENOMEM;
++ }
++
++ rc = xencomm_init(desc, buffer, bytes);
++ if (rc) {
++ printk("%s failure: %d\n", "xencomm_init", rc);
++ xencomm_free((struct xencomm_handle *)__pa(desc));
++ return rc;
++ }
++
++ *ret = desc;
++ return 0;
++}
++
++/* check if memory address is within VMALLOC region */
++static int is_phys_contiguous(unsigned long addr)
++{
++ if (!is_kernel_addr(addr))
++ return 0;
++
++ return (addr < VMALLOC_START) || (addr >= VMALLOC_END);
++}
++
++static struct xencomm_handle *xencomm_create_inline(void *ptr)
++{
++ unsigned long paddr;
++
++ BUG_ON(!is_phys_contiguous((unsigned long)ptr));
++
++ paddr = (unsigned long)xencomm_pa(ptr);
++ BUG_ON(paddr & XENCOMM_INLINE_FLAG);
++ return (struct xencomm_handle *)(paddr | XENCOMM_INLINE_FLAG);
++}
++
++/* "mini" routine, for stack-based communications: */
++static int xencomm_create_mini(void *buffer,
++ unsigned long bytes, struct xencomm_mini *xc_desc,
++ struct xencomm_desc **ret)
++{
++ int rc = 0;
++ struct xencomm_desc *desc;
++ BUG_ON(((unsigned long)xc_desc) % sizeof(*xc_desc) != 0);
++
++ desc = (void *)xc_desc;
++
++ desc->nr_addrs = XENCOMM_MINI_ADDRS;
++
++ if (!(rc = xencomm_init(desc, buffer, bytes)))
++ *ret = desc;
++
++ return rc;
++}
++
++struct xencomm_handle *xencomm_map(void *ptr, unsigned long bytes)
++{
++ int rc;
++ struct xencomm_desc *desc;
++
++ if (is_phys_contiguous((unsigned long)ptr))
++ return xencomm_create_inline(ptr);
++
++ rc = xencomm_create(ptr, bytes, &desc, GFP_KERNEL);
++
++ if (rc || desc == NULL)
++ return NULL;
++
++ return xencomm_pa(desc);
++}
++
++struct xencomm_handle *__xencomm_map_no_alloc(void *ptr, unsigned long bytes,
++ struct xencomm_mini *xc_desc)
++{
++ int rc;
++ struct xencomm_desc *desc = NULL;
++
++ if (is_phys_contiguous((unsigned long)ptr))
++ return xencomm_create_inline(ptr);
++
++ rc = xencomm_create_mini(ptr, bytes, xc_desc,
++ &desc);
++
++ if (rc)
++ return NULL;
++
++ return xencomm_pa(desc);
++}
+--- linux-2.6.18.8/drivers/xen/evtchn/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/evtchn/Makefile 2008-05-19 00:33:45.830787860 +0300
+@@ -0,0 +1,2 @@
++
++obj-y := evtchn.o
+--- linux-2.6.18.8/drivers/xen/evtchn/evtchn.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/evtchn/evtchn.c 2008-05-19 00:33:45.830787860 +0300
+@@ -0,0 +1,558 @@
++/******************************************************************************
++ * evtchn.c
++ *
++ * Driver for receiving and demuxing event-channel signals.
++ *
++ * Copyright (c) 2004-2005, K A Fraser
++ * Multi-process extensions Copyright (c) 2004, Steven Smith
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/module.h>
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/slab.h>
++#include <linux/string.h>
++#include <linux/errno.h>
++#include <linux/fs.h>
++#include <linux/errno.h>
++#include <linux/miscdevice.h>
++#include <linux/major.h>
++#include <linux/proc_fs.h>
++#include <linux/stat.h>
++#include <linux/poll.h>
++#include <linux/irq.h>
++#include <linux/init.h>
++#include <linux/gfp.h>
++#include <linux/mutex.h>
++#include <linux/cpu.h>
++#include <xen/evtchn.h>
++#include <xen/public/evtchn.h>
++
++struct per_user_data {
++ /* Notification ring, accessed via /dev/xen/evtchn. */
++#define EVTCHN_RING_SIZE (PAGE_SIZE / sizeof(evtchn_port_t))
++#define EVTCHN_RING_MASK(_i) ((_i)&(EVTCHN_RING_SIZE-1))
++ evtchn_port_t *ring;
++ unsigned int ring_cons, ring_prod, ring_overflow;
++ struct mutex ring_cons_mutex; /* protect against concurrent readers */
++
++ /* Processes wait on this queue when ring is empty. */
++ wait_queue_head_t evtchn_wait;
++ struct fasync_struct *evtchn_async_queue;
++
++ int bind_cpu;
++ int nr_event_wrong_delivery;
++};
++
++/* Who's bound to each port? */
++static struct per_user_data *port_user[NR_EVENT_CHANNELS];
++static spinlock_t port_user_lock;
++
++void evtchn_device_upcall(int port)
++{
++ struct per_user_data *u;
++
++ spin_lock(&port_user_lock);
++
++ mask_evtchn(port);
++ clear_evtchn(port);
++
++ if ((u = port_user[port]) != NULL) {
++ if ((u->ring_prod - u->ring_cons) < EVTCHN_RING_SIZE) {
++ u->ring[EVTCHN_RING_MASK(u->ring_prod)] = port;
++ if (u->ring_cons == u->ring_prod++) {
++ wake_up_interruptible(&u->evtchn_wait);
++ kill_fasync(&u->evtchn_async_queue,
++ SIGIO, POLL_IN);
++ }
++ } else {
++ u->ring_overflow = 1;
++ }
++ }
++
++ spin_unlock(&port_user_lock);
++}
++
++static void evtchn_check_wrong_delivery(struct per_user_data *u)
++{
++ evtchn_port_t port;
++ unsigned int current_cpu = smp_processor_id();
++
++ /* Delivered to correct CPU? All is good. */
++ if (u->bind_cpu == current_cpu) {
++ u->nr_event_wrong_delivery = 0;
++ return;
++ }
++
++ /* Tolerate up to 100 consecutive misdeliveries. */
++ if (++u->nr_event_wrong_delivery < 100)
++ return;
++
++ spin_lock_irq(&port_user_lock);
++
++ for (port = 0; port < NR_EVENT_CHANNELS; port++)
++ if (port_user[port] == u)
++ rebind_evtchn_to_cpu(port, current_cpu);
++
++ u->bind_cpu = current_cpu;
++ u->nr_event_wrong_delivery = 0;
++
++ spin_unlock_irq(&port_user_lock);
++}
++
++static ssize_t evtchn_read(struct file *file, char __user *buf,
++ size_t count, loff_t *ppos)
++{
++ int rc;
++ unsigned int c, p, bytes1 = 0, bytes2 = 0;
++ struct per_user_data *u = file->private_data;
++
++ /* Whole number of ports. */
++ count &= ~(sizeof(evtchn_port_t)-1);
++
++ if (count == 0)
++ return 0;
++
++ if (count > PAGE_SIZE)
++ count = PAGE_SIZE;
++
++ for (;;) {
++ mutex_lock(&u->ring_cons_mutex);
++
++ rc = -EFBIG;
++ if (u->ring_overflow)
++ goto unlock_out;
++
++ if ((c = u->ring_cons) != (p = u->ring_prod))
++ break;
++
++ mutex_unlock(&u->ring_cons_mutex);
++
++ if (file->f_flags & O_NONBLOCK)
++ return -EAGAIN;
++
++ rc = wait_event_interruptible(
++ u->evtchn_wait, u->ring_cons != u->ring_prod);
++ if (rc)
++ return rc;
++ }
++
++ /* Byte lengths of two chunks. Chunk split (if any) is at ring wrap. */
++ if (((c ^ p) & EVTCHN_RING_SIZE) != 0) {
++ bytes1 = (EVTCHN_RING_SIZE - EVTCHN_RING_MASK(c)) *
++ sizeof(evtchn_port_t);
++ bytes2 = EVTCHN_RING_MASK(p) * sizeof(evtchn_port_t);
++ } else {
++ bytes1 = (p - c) * sizeof(evtchn_port_t);
++ bytes2 = 0;
++ }
++
++ /* Truncate chunks according to caller's maximum byte count. */
++ if (bytes1 > count) {
++ bytes1 = count;
++ bytes2 = 0;
++ } else if ((bytes1 + bytes2) > count) {
++ bytes2 = count - bytes1;
++ }
++
++ rc = -EFAULT;
++ if (copy_to_user(buf, &u->ring[EVTCHN_RING_MASK(c)], bytes1) ||
++ ((bytes2 != 0) &&
++ copy_to_user(&buf[bytes1], &u->ring[0], bytes2)))
++ goto unlock_out;
++
++ evtchn_check_wrong_delivery(u);
++
++ u->ring_cons += (bytes1 + bytes2) / sizeof(evtchn_port_t);
++ rc = bytes1 + bytes2;
++
++ unlock_out:
++ mutex_unlock(&u->ring_cons_mutex);
++ return rc;
++}
++
++static ssize_t evtchn_write(struct file *file, const char __user *buf,
++ size_t count, loff_t *ppos)
++{
++ int rc, i;
++ evtchn_port_t *kbuf = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
++ struct per_user_data *u = file->private_data;
++
++ if (kbuf == NULL)
++ return -ENOMEM;
++
++ /* Whole number of ports. */
++ count &= ~(sizeof(evtchn_port_t)-1);
++
++ rc = 0;
++ if (count == 0)
++ goto out;
++
++ if (count > PAGE_SIZE)
++ count = PAGE_SIZE;
++
++ rc = -EFAULT;
++ if (copy_from_user(kbuf, buf, count) != 0)
++ goto out;
++
++ spin_lock_irq(&port_user_lock);
++ for (i = 0; i < (count/sizeof(evtchn_port_t)); i++)
++ if ((kbuf[i] < NR_EVENT_CHANNELS) && (port_user[kbuf[i]] == u))
++ unmask_evtchn(kbuf[i]);
++ spin_unlock_irq(&port_user_lock);
++
++ rc = count;
++
++ out:
++ free_page((unsigned long)kbuf);
++ return rc;
++}
++
++static unsigned int next_bind_cpu(cpumask_t map)
++{
++ static unsigned int bind_cpu;
++ bind_cpu = next_cpu(bind_cpu, map);
++ if (bind_cpu >= NR_CPUS)
++ bind_cpu = first_cpu(map);
++ return bind_cpu;
++}
++
++static void evtchn_bind_to_user(struct per_user_data *u, int port)
++{
++ spin_lock_irq(&port_user_lock);
++
++ BUG_ON(port_user[port] != NULL);
++ port_user[port] = u;
++
++ if (u->bind_cpu == -1)
++ u->bind_cpu = next_bind_cpu(cpu_online_map);
++
++ rebind_evtchn_to_cpu(port, u->bind_cpu);
++
++ unmask_evtchn(port);
++
++ spin_unlock_irq(&port_user_lock);
++}
++
++static long evtchn_ioctl(struct file *file,
++ unsigned int cmd, unsigned long arg)
++{
++ int rc;
++ struct per_user_data *u = file->private_data;
++ void __user *uarg = (void __user *) arg;
++
++ switch (cmd) {
++ case IOCTL_EVTCHN_BIND_VIRQ: {
++ struct ioctl_evtchn_bind_virq bind;
++ struct evtchn_bind_virq bind_virq;
++
++ rc = -EFAULT;
++ if (copy_from_user(&bind, uarg, sizeof(bind)))
++ break;
++
++ bind_virq.virq = bind.virq;
++ bind_virq.vcpu = 0;
++ rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_virq,
++ &bind_virq);
++ if (rc != 0)
++ break;
++
++ rc = bind_virq.port;
++ evtchn_bind_to_user(u, rc);
++ break;
++ }
++
++ case IOCTL_EVTCHN_BIND_INTERDOMAIN: {
++ struct ioctl_evtchn_bind_interdomain bind;
++ struct evtchn_bind_interdomain bind_interdomain;
++
++ rc = -EFAULT;
++ if (copy_from_user(&bind, uarg, sizeof(bind)))
++ break;
++
++ bind_interdomain.remote_dom = bind.remote_domain;
++ bind_interdomain.remote_port = bind.remote_port;
++ rc = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
++ &bind_interdomain);
++ if (rc != 0)
++ break;
++
++ rc = bind_interdomain.local_port;
++ evtchn_bind_to_user(u, rc);
++ break;
++ }
++
++ case IOCTL_EVTCHN_BIND_UNBOUND_PORT: {
++ struct ioctl_evtchn_bind_unbound_port bind;
++ struct evtchn_alloc_unbound alloc_unbound;
++
++ rc = -EFAULT;
++ if (copy_from_user(&bind, uarg, sizeof(bind)))
++ break;
++
++ alloc_unbound.dom = DOMID_SELF;
++ alloc_unbound.remote_dom = bind.remote_domain;
++ rc = HYPERVISOR_event_channel_op(EVTCHNOP_alloc_unbound,
++ &alloc_unbound);
++ if (rc != 0)
++ break;
++
++ rc = alloc_unbound.port;
++ evtchn_bind_to_user(u, rc);
++ break;
++ }
++
++ case IOCTL_EVTCHN_UNBIND: {
++ struct ioctl_evtchn_unbind unbind;
++ struct evtchn_close close;
++ int ret;
++
++ rc = -EFAULT;
++ if (copy_from_user(&unbind, uarg, sizeof(unbind)))
++ break;
++
++ rc = -EINVAL;
++ if (unbind.port >= NR_EVENT_CHANNELS)
++ break;
++
++ spin_lock_irq(&port_user_lock);
++
++ rc = -ENOTCONN;
++ if (port_user[unbind.port] != u) {
++ spin_unlock_irq(&port_user_lock);
++ break;
++ }
++
++ port_user[unbind.port] = NULL;
++ mask_evtchn(unbind.port);
++ rebind_evtchn_to_cpu(unbind.port, 0);
++
++ spin_unlock_irq(&port_user_lock);
++
++ close.port = unbind.port;
++ ret = HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
++ BUG_ON(ret);
++
++ rc = 0;
++ break;
++ }
++
++ case IOCTL_EVTCHN_NOTIFY: {
++ struct ioctl_evtchn_notify notify;
++
++ rc = -EFAULT;
++ if (copy_from_user(¬ify, uarg, sizeof(notify)))
++ break;
++
++ if (notify.port >= NR_EVENT_CHANNELS) {
++ rc = -EINVAL;
++ } else if (port_user[notify.port] != u) {
++ rc = -ENOTCONN;
++ } else {
++ notify_remote_via_evtchn(notify.port);
++ rc = 0;
++ }
++ break;
++ }
++
++ case IOCTL_EVTCHN_RESET: {
++ /* Initialise the ring to empty. Clear errors. */
++ mutex_lock(&u->ring_cons_mutex);
++ spin_lock_irq(&port_user_lock);
++ u->ring_cons = u->ring_prod = u->ring_overflow = 0;
++ spin_unlock_irq(&port_user_lock);
++ mutex_unlock(&u->ring_cons_mutex);
++ rc = 0;
++ break;
++ }
++
++ default:
++ rc = -ENOSYS;
++ break;
++ }
++
++ return rc;
++}
++
++static unsigned int evtchn_poll(struct file *file, poll_table *wait)
++{
++ unsigned int mask = POLLOUT | POLLWRNORM;
++ struct per_user_data *u = file->private_data;
++
++ poll_wait(file, &u->evtchn_wait, wait);
++ if (u->ring_cons != u->ring_prod)
++ mask |= POLLIN | POLLRDNORM;
++ if (u->ring_overflow)
++ mask = POLLERR;
++ return mask;
++}
++
++static int evtchn_fasync(int fd, struct file *filp, int on)
++{
++ struct per_user_data *u = filp->private_data;
++ return fasync_helper(fd, filp, on, &u->evtchn_async_queue);
++}
++
++static int evtchn_open(struct inode *inode, struct file *filp)
++{
++ struct per_user_data *u;
++
++ if ((u = kmalloc(sizeof(*u), GFP_KERNEL)) == NULL)
++ return -ENOMEM;
++
++ memset(u, 0, sizeof(*u));
++ init_waitqueue_head(&u->evtchn_wait);
++
++ u->ring = (evtchn_port_t *)__get_free_page(GFP_KERNEL);
++ if (u->ring == NULL) {
++ kfree(u);
++ return -ENOMEM;
++ }
++
++ mutex_init(&u->ring_cons_mutex);
++
++ filp->private_data = u;
++
++ u->bind_cpu = -1;
++
++ return 0;
++}
++
++static int evtchn_release(struct inode *inode, struct file *filp)
++{
++ int i;
++ struct per_user_data *u = filp->private_data;
++ struct evtchn_close close;
++
++ spin_lock_irq(&port_user_lock);
++
++ free_page((unsigned long)u->ring);
++
++ for (i = 0; i < NR_EVENT_CHANNELS; i++) {
++ int ret;
++ if (port_user[i] != u)
++ continue;
++
++ port_user[i] = NULL;
++ mask_evtchn(i);
++ rebind_evtchn_to_cpu(i, 0);
++
++ close.port = i;
++ ret = HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
++ BUG_ON(ret);
++ }
++
++ spin_unlock_irq(&port_user_lock);
++
++ kfree(u);
++
++ return 0;
++}
++
++static const struct file_operations evtchn_fops = {
++ .owner = THIS_MODULE,
++ .read = evtchn_read,
++ .write = evtchn_write,
++ .unlocked_ioctl = evtchn_ioctl,
++ .poll = evtchn_poll,
++ .fasync = evtchn_fasync,
++ .open = evtchn_open,
++ .release = evtchn_release,
++};
++
++static struct miscdevice evtchn_miscdev = {
++ .minor = MISC_DYNAMIC_MINOR,
++ .name = "evtchn",
++ .fops = &evtchn_fops,
++};
++
++static int __cpuinit evtchn_cpu_notify(struct notifier_block *nfb,
++ unsigned long action, void *hcpu)
++{
++ int hotcpu = (unsigned long)hcpu;
++ cpumask_t map = cpu_online_map;
++ int port, newcpu;
++ struct per_user_data *u;
++
++ switch (action) {
++ case CPU_DOWN_PREPARE:
++ cpu_clear(hotcpu, map);
++ spin_lock_irq(&port_user_lock);
++ for (port = 0; port < NR_EVENT_CHANNELS; port++) {
++ if ((u = port_user[port]) != NULL &&
++ u->bind_cpu == hotcpu &&
++ (newcpu = next_bind_cpu(map)) < NR_CPUS) {
++ rebind_evtchn_to_cpu(port, newcpu);
++ u->bind_cpu = newcpu;
++ }
++ }
++ spin_unlock_irq(&port_user_lock);
++ break;
++ default:
++ return NOTIFY_DONE;
++ }
++ return NOTIFY_OK;
++}
++
++static struct notifier_block __cpuinitdata evtchn_cpu_nfb = {
++ .notifier_call = evtchn_cpu_notify
++};
++
++static int __init evtchn_init(void)
++{
++ int err;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ spin_lock_init(&port_user_lock);
++ memset(port_user, 0, sizeof(port_user));
++
++ /* Create '/dev/misc/evtchn'. */
++ err = misc_register(&evtchn_miscdev);
++ if (err != 0) {
++ printk(KERN_ALERT "Could not register /dev/misc/evtchn\n");
++ return err;
++ }
++
++ register_cpu_notifier(&evtchn_cpu_nfb);
++
++ printk("Event-channel device installed.\n");
++
++ return 0;
++}
++
++static void __exit evtchn_cleanup(void)
++{
++ misc_deregister(&evtchn_miscdev);
++ unregister_cpu_notifier(&evtchn_cpu_nfb);
++}
++
++module_init(evtchn_init);
++module_exit(evtchn_cleanup);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/fbfront/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/fbfront/Makefile 2008-05-19 00:33:45.830787860 +0300
+@@ -0,0 +1,2 @@
++obj-$(CONFIG_XEN_FRAMEBUFFER) := xenfb.o
++obj-$(CONFIG_XEN_KEYBOARD) += xenkbd.o
+--- linux-2.6.18.8/drivers/xen/fbfront/xenfb.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/fbfront/xenfb.c 2008-05-19 00:33:45.834788090 +0300
+@@ -0,0 +1,887 @@
++/*
++ * linux/drivers/video/xenfb.c -- Xen para-virtual frame buffer device
++ *
++ * Copyright (C) 2005-2006 Anthony Liguori <aliguori@us.ibm.com>
++ * Copyright (C) 2006 Red Hat, Inc., Markus Armbruster <armbru@redhat.com>
++ *
++ * Based on linux/drivers/video/q40fb.c
++ *
++ * This file is subject to the terms and conditions of the GNU General Public
++ * License. See the file COPYING in the main directory of this archive for
++ * more details.
++ */
++
++/*
++ * TODO:
++ *
++ * Switch to grant tables when they become capable of dealing with the
++ * frame buffer.
++ */
++
++#include <linux/kernel.h>
++#include <linux/errno.h>
++#include <linux/fb.h>
++#include <linux/module.h>
++#include <linux/vmalloc.h>
++#include <linux/mm.h>
++#include <linux/mutex.h>
++#include <asm/hypervisor.h>
++#include <xen/evtchn.h>
++#include <xen/interface/io/fbif.h>
++#include <xen/interface/io/protocols.h>
++#include <xen/xenbus.h>
++#include <linux/kthread.h>
++
++struct xenfb_mapping
++{
++ struct list_head link;
++ struct vm_area_struct *vma;
++ atomic_t map_refs;
++ int faults;
++ struct xenfb_info *info;
++};
++
++struct xenfb_info
++{
++ struct task_struct *kthread;
++ wait_queue_head_t wq;
++
++ unsigned char *fb;
++ struct fb_info *fb_info;
++ struct timer_list refresh;
++ int dirty;
++ int x1, y1, x2, y2; /* dirty rectangle,
++ protected by dirty_lock */
++ spinlock_t dirty_lock;
++ struct mutex mm_lock;
++ int nr_pages;
++ struct page **pages;
++ struct list_head mappings; /* protected by mm_lock */
++
++ int irq;
++ struct xenfb_page *page;
++ unsigned long *mfns;
++ int update_wanted; /* XENFB_TYPE_UPDATE wanted */
++ int feature_resize; /* Backend has resize feature */
++ struct xenfb_resize resize;
++ int resize_dpy;
++ spinlock_t resize_lock;
++
++ struct xenbus_device *xbdev;
++};
++
++/*
++ * There are three locks:
++ * spinlock resize_lock protecting resize_dpy and resize
++ * spinlock dirty_lock protecting the dirty rectangle
++ * mutex mm_lock protecting mappings.
++ *
++ * How the dirty and mapping locks work together
++ *
++ * The problem is that dirty rectangle and mappings aren't
++ * independent: the dirty rectangle must cover all faulted pages in
++ * mappings. We need to prove that our locking maintains this
++ * invariant.
++ *
++ * There are several kinds of critical regions:
++ *
++ * 1. Holding only dirty_lock: xenfb_refresh(). May run in
++ * interrupts. Extends the dirty rectangle. Trivially preserves
++ * invariant.
++ *
++ * 2. Holding only mm_lock: xenfb_mmap() and xenfb_vm_close(). Touch
++ * only mappings. The former creates unfaulted pages. Preserves
++ * invariant. The latter removes pages. Preserves invariant.
++ *
++ * 3. Holding both locks: xenfb_vm_nopage(). Extends the dirty
++ * rectangle and updates mappings consistently. Preserves
++ * invariant.
++ *
++ * 4. The ugliest one: xenfb_update_screen(). Clear the dirty
++ * rectangle and update mappings consistently.
++ *
++ * We can't simply hold both locks, because zap_page_range() cannot
++ * be called with a spinlock held.
++ *
++ * Therefore, we first clear the dirty rectangle with both locks
++ * held. Then we unlock dirty_lock and update the mappings.
++ * Critical regions that hold only dirty_lock may interfere with
++ * that. This can only be region 1: xenfb_refresh(). But that
++ * just extends the dirty rectangle, which can't harm the
++ * invariant.
++ *
++ * But FIXME: the invariant is too weak. It misses that the fault
++ * record in mappings must be consistent with the mapping of pages in
++ * the associated address space! do_no_page() updates the PTE after
++ * xenfb_vm_nopage() returns, i.e. outside the critical region. This
++ * allows the following race:
++ *
++ * X writes to some address in the Xen frame buffer
++ * Fault - call do_no_page()
++ * call xenfb_vm_nopage()
++ * grab mm_lock
++ * map->faults++;
++ * release mm_lock
++ * return back to do_no_page()
++ * (preempted, or SMP)
++ * Xen worker thread runs.
++ * grab mm_lock
++ * look at mappings
++ * find this mapping, zaps its pages (but page not in pte yet)
++ * clear map->faults
++ * releases mm_lock
++ * (back to X process)
++ * put page in X's pte
++ *
++ * Oh well, we wont be updating the writes to this page anytime soon.
++ */
++#define MB_ (1024*1024)
++#define XENFB_DEFAULT_FB_LEN (XENFB_WIDTH * XENFB_HEIGHT * XENFB_DEPTH / 8)
++
++enum {KPARAM_MEM, KPARAM_WIDTH, KPARAM_HEIGHT, KPARAM_CNT};
++static int video[KPARAM_CNT] = {2, XENFB_WIDTH, XENFB_HEIGHT};
++module_param_array(video, int, NULL, 0);
++MODULE_PARM_DESC(video,
++ "Size of video memory in MB and width,height in pixels, default = (2,800,600)");
++
++static int xenfb_fps = 20;
++
++static int xenfb_remove(struct xenbus_device *);
++static void xenfb_init_shared_page(struct xenfb_info *, struct fb_info *);
++static int xenfb_connect_backend(struct xenbus_device *, struct xenfb_info *);
++static void xenfb_disconnect_backend(struct xenfb_info *);
++
++static void xenfb_send_event(struct xenfb_info *info,
++ union xenfb_out_event *event)
++{
++ __u32 prod;
++
++ prod = info->page->out_prod;
++ /* caller ensures !xenfb_queue_full() */
++ mb(); /* ensure ring space available */
++ XENFB_OUT_RING_REF(info->page, prod) = *event;
++ wmb(); /* ensure ring contents visible */
++ info->page->out_prod = prod + 1;
++
++ notify_remote_via_irq(info->irq);
++}
++
++static void xenfb_do_update(struct xenfb_info *info,
++ int x, int y, int w, int h)
++{
++ union xenfb_out_event event;
++
++ memset(&event, 0, sizeof(event));
++ event.type = XENFB_TYPE_UPDATE;
++ event.update.x = x;
++ event.update.y = y;
++ event.update.width = w;
++ event.update.height = h;
++
++ /* caller ensures !xenfb_queue_full() */
++ xenfb_send_event(info, &event);
++}
++
++static void xenfb_do_resize(struct xenfb_info *info)
++{
++ union xenfb_out_event event;
++
++ memset(&event, 0, sizeof(event));
++ event.resize = info->resize;
++
++ /* caller ensures !xenfb_queue_full() */
++ xenfb_send_event(info, &event);
++}
++
++static int xenfb_queue_full(struct xenfb_info *info)
++{
++ __u32 cons, prod;
++
++ prod = info->page->out_prod;
++ cons = info->page->out_cons;
++ return prod - cons == XENFB_OUT_RING_LEN;
++}
++
++static void xenfb_update_screen(struct xenfb_info *info)
++{
++ unsigned long flags;
++ int y1, y2, x1, x2;
++ struct xenfb_mapping *map;
++
++ if (!info->update_wanted)
++ return;
++ if (xenfb_queue_full(info))
++ return;
++
++ mutex_lock(&info->mm_lock);
++
++ spin_lock_irqsave(&info->dirty_lock, flags);
++ y1 = info->y1;
++ y2 = info->y2;
++ x1 = info->x1;
++ x2 = info->x2;
++ info->x1 = info->y1 = INT_MAX;
++ info->x2 = info->y2 = 0;
++ spin_unlock_irqrestore(&info->dirty_lock, flags);
++
++ list_for_each_entry(map, &info->mappings, link) {
++ if (!map->faults)
++ continue;
++ zap_page_range(map->vma, map->vma->vm_start,
++ map->vma->vm_end - map->vma->vm_start, NULL);
++ map->faults = 0;
++ }
++
++ mutex_unlock(&info->mm_lock);
++
++ if (x2 < x1 || y2 < y1) {
++ printk("xenfb_update_screen bogus rect %d %d %d %d\n",
++ x1, x2, y1, y2);
++ WARN_ON(1);
++ }
++ xenfb_do_update(info, x1, y1, x2 - x1, y2 - y1);
++}
++
++static void xenfb_handle_resize_dpy(struct xenfb_info *info)
++{
++ unsigned long flags;
++
++ spin_lock_irqsave(&info->resize_lock, flags);
++ if (info->resize_dpy) {
++ if (!xenfb_queue_full(info)) {
++ info->resize_dpy = 0;
++ xenfb_do_resize(info);
++ }
++ }
++ spin_unlock_irqrestore(&info->resize_lock, flags);
++}
++
++static int xenfb_thread(void *data)
++{
++ struct xenfb_info *info = data;
++
++ while (!kthread_should_stop()) {
++ xenfb_handle_resize_dpy(info);
++ if (info->dirty) {
++ info->dirty = 0;
++ xenfb_update_screen(info);
++ }
++ wait_event_interruptible(info->wq,
++ kthread_should_stop() || info->dirty);
++ try_to_freeze();
++ }
++ return 0;
++}
++
++static int xenfb_setcolreg(unsigned regno, unsigned red, unsigned green,
++ unsigned blue, unsigned transp,
++ struct fb_info *info)
++{
++ u32 v;
++
++ if (regno > info->cmap.len)
++ return 1;
++
++ red >>= (16 - info->var.red.length);
++ green >>= (16 - info->var.green.length);
++ blue >>= (16 - info->var.blue.length);
++
++ v = (red << info->var.red.offset) |
++ (green << info->var.green.offset) |
++ (blue << info->var.blue.offset);
++
++ /* FIXME is this sane? check against xxxfb_setcolreg()! */
++ switch (info->var.bits_per_pixel) {
++ case 16:
++ case 24:
++ case 32:
++ ((u32 *)info->pseudo_palette)[regno] = v;
++ break;
++ }
++
++ return 0;
++}
++
++static void xenfb_timer(unsigned long data)
++{
++ struct xenfb_info *info = (struct xenfb_info *)data;
++ wake_up(&info->wq);
++}
++
++static void __xenfb_refresh(struct xenfb_info *info,
++ int x1, int y1, int w, int h)
++{
++ int y2, x2;
++
++ y2 = y1 + h;
++ x2 = x1 + w;
++
++ if (info->y1 > y1)
++ info->y1 = y1;
++ if (info->y2 < y2)
++ info->y2 = y2;
++ if (info->x1 > x1)
++ info->x1 = x1;
++ if (info->x2 < x2)
++ info->x2 = x2;
++ info->dirty = 1;
++
++ if (timer_pending(&info->refresh))
++ return;
++
++ mod_timer(&info->refresh, jiffies + HZ/xenfb_fps);
++}
++
++static void xenfb_refresh(struct xenfb_info *info,
++ int x1, int y1, int w, int h)
++{
++ unsigned long flags;
++
++ spin_lock_irqsave(&info->dirty_lock, flags);
++ __xenfb_refresh(info, x1, y1, w, h);
++ spin_unlock_irqrestore(&info->dirty_lock, flags);
++}
++
++static void xenfb_fillrect(struct fb_info *p, const struct fb_fillrect *rect)
++{
++ struct xenfb_info *info = p->par;
++
++ cfb_fillrect(p, rect);
++ xenfb_refresh(info, rect->dx, rect->dy, rect->width, rect->height);
++}
++
++static void xenfb_imageblit(struct fb_info *p, const struct fb_image *image)
++{
++ struct xenfb_info *info = p->par;
++
++ cfb_imageblit(p, image);
++ xenfb_refresh(info, image->dx, image->dy, image->width, image->height);
++}
++
++static void xenfb_copyarea(struct fb_info *p, const struct fb_copyarea *area)
++{
++ struct xenfb_info *info = p->par;
++
++ cfb_copyarea(p, area);
++ xenfb_refresh(info, area->dx, area->dy, area->width, area->height);
++}
++
++static void xenfb_vm_open(struct vm_area_struct *vma)
++{
++ struct xenfb_mapping *map = vma->vm_private_data;
++ atomic_inc(&map->map_refs);
++}
++
++static void xenfb_vm_close(struct vm_area_struct *vma)
++{
++ struct xenfb_mapping *map = vma->vm_private_data;
++ struct xenfb_info *info = map->info;
++
++ mutex_lock(&info->mm_lock);
++ if (atomic_dec_and_test(&map->map_refs)) {
++ list_del(&map->link);
++ kfree(map);
++ }
++ mutex_unlock(&info->mm_lock);
++}
++
++static struct page *xenfb_vm_nopage(struct vm_area_struct *vma,
++ unsigned long vaddr, int *type)
++{
++ struct xenfb_mapping *map = vma->vm_private_data;
++ struct xenfb_info *info = map->info;
++ int pgnr = (vaddr - vma->vm_start) >> PAGE_SHIFT;
++ unsigned long flags;
++ struct page *page;
++ int y1, y2;
++
++ if (pgnr >= info->nr_pages)
++ return NOPAGE_SIGBUS;
++
++ mutex_lock(&info->mm_lock);
++ spin_lock_irqsave(&info->dirty_lock, flags);
++ page = info->pages[pgnr];
++ get_page(page);
++ map->faults++;
++
++ y1 = pgnr * PAGE_SIZE / info->fb_info->fix.line_length;
++ y2 = (pgnr * PAGE_SIZE + PAGE_SIZE - 1) / info->fb_info->fix.line_length;
++ if (y2 > info->fb_info->var.yres)
++ y2 = info->fb_info->var.yres;
++ __xenfb_refresh(info, 0, y1, info->fb_info->var.xres, y2 - y1);
++ spin_unlock_irqrestore(&info->dirty_lock, flags);
++ mutex_unlock(&info->mm_lock);
++
++ if (type)
++ *type = VM_FAULT_MINOR;
++
++ return page;
++}
++
++static struct vm_operations_struct xenfb_vm_ops = {
++ .open = xenfb_vm_open,
++ .close = xenfb_vm_close,
++ .nopage = xenfb_vm_nopage,
++};
++
++static int xenfb_mmap(struct fb_info *fb_info, struct vm_area_struct *vma)
++{
++ struct xenfb_info *info = fb_info->par;
++ struct xenfb_mapping *map;
++ int map_pages;
++
++ if (!(vma->vm_flags & VM_WRITE))
++ return -EINVAL;
++ if (!(vma->vm_flags & VM_SHARED))
++ return -EINVAL;
++ if (vma->vm_pgoff != 0)
++ return -EINVAL;
++
++ map_pages = (vma->vm_end - vma->vm_start + PAGE_SIZE-1) >> PAGE_SHIFT;
++ if (map_pages > info->nr_pages)
++ return -EINVAL;
++
++ map = kzalloc(sizeof(*map), GFP_KERNEL);
++ if (map == NULL)
++ return -ENOMEM;
++
++ map->vma = vma;
++ map->faults = 0;
++ map->info = info;
++ atomic_set(&map->map_refs, 1);
++
++ mutex_lock(&info->mm_lock);
++ list_add(&map->link, &info->mappings);
++ mutex_unlock(&info->mm_lock);
++
++ vma->vm_ops = &xenfb_vm_ops;
++ vma->vm_flags |= (VM_DONTEXPAND | VM_RESERVED);
++ vma->vm_private_data = map;
++
++ return 0;
++}
++
++static int
++xenfb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
++{
++ struct xenfb_info *xenfb_info;
++ int required_mem_len;
++
++ xenfb_info = info->par;
++
++ if (!xenfb_info->feature_resize) {
++ if (var->xres == video[KPARAM_WIDTH] &&
++ var->yres == video[KPARAM_HEIGHT] &&
++ var->bits_per_pixel == xenfb_info->page->depth) {
++ return 0;
++ }
++ return -EINVAL;
++ }
++
++ /* Can't resize past initial width and height */
++ if (var->xres > video[KPARAM_WIDTH] || var->yres > video[KPARAM_HEIGHT])
++ return -EINVAL;
++
++ required_mem_len = var->xres * var->yres * (xenfb_info->page->depth / 8);
++ if (var->bits_per_pixel == xenfb_info->page->depth &&
++ var->xres <= info->fix.line_length / (XENFB_DEPTH / 8) &&
++ required_mem_len <= info->fix.smem_len) {
++ var->xres_virtual = var->xres;
++ var->yres_virtual = var->yres;
++ return 0;
++ }
++ return -EINVAL;
++}
++
++static int xenfb_set_par(struct fb_info *info)
++{
++ struct xenfb_info *xenfb_info;
++ unsigned long flags;
++
++ xenfb_info = info->par;
++
++ spin_lock_irqsave(&xenfb_info->resize_lock, flags);
++ xenfb_info->resize.type = XENFB_TYPE_RESIZE;
++ xenfb_info->resize.width = info->var.xres;
++ xenfb_info->resize.height = info->var.yres;
++ xenfb_info->resize.stride = info->fix.line_length;
++ xenfb_info->resize.depth = info->var.bits_per_pixel;
++ xenfb_info->resize.offset = 0;
++ xenfb_info->resize_dpy = 1;
++ spin_unlock_irqrestore(&xenfb_info->resize_lock, flags);
++ return 0;
++}
++
++static struct fb_ops xenfb_fb_ops = {
++ .owner = THIS_MODULE,
++ .fb_setcolreg = xenfb_setcolreg,
++ .fb_fillrect = xenfb_fillrect,
++ .fb_copyarea = xenfb_copyarea,
++ .fb_imageblit = xenfb_imageblit,
++ .fb_mmap = xenfb_mmap,
++ .fb_check_var = xenfb_check_var,
++ .fb_set_par = xenfb_set_par,
++};
++
++static irqreturn_t xenfb_event_handler(int rq, void *dev_id,
++ struct pt_regs *regs)
++{
++ /*
++ * No in events recognized, simply ignore them all.
++ * If you need to recognize some, see xenbkd's input_handler()
++ * for how to do that.
++ */
++ struct xenfb_info *info = dev_id;
++ struct xenfb_page *page = info->page;
++
++ if (page->in_cons != page->in_prod) {
++ info->page->in_cons = info->page->in_prod;
++ notify_remote_via_irq(info->irq);
++ }
++ return IRQ_HANDLED;
++}
++
++static unsigned long vmalloc_to_mfn(void *address)
++{
++ return pfn_to_mfn(vmalloc_to_pfn(address));
++}
++
++static int __devinit xenfb_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ struct xenfb_info *info;
++ struct fb_info *fb_info;
++ int fb_size;
++ int val;
++ int ret;
++
++ info = kzalloc(sizeof(*info), GFP_KERNEL);
++ if (info == NULL) {
++ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
++ return -ENOMEM;
++ }
++
++ /* Limit kernel param videoram amount to what is in xenstore */
++ if (xenbus_scanf(XBT_NIL, dev->otherend, "videoram", "%d", &val) == 1) {
++ if (val < video[KPARAM_MEM])
++ video[KPARAM_MEM] = val;
++ }
++
++ /* If requested res does not fit in available memory, use default */
++ fb_size = video[KPARAM_MEM] * MB_;
++ if (video[KPARAM_WIDTH] * video[KPARAM_HEIGHT] * XENFB_DEPTH/8 > fb_size) {
++ video[KPARAM_WIDTH] = XENFB_WIDTH;
++ video[KPARAM_HEIGHT] = XENFB_HEIGHT;
++ fb_size = XENFB_DEFAULT_FB_LEN;
++ }
++
++ dev->dev.driver_data = info;
++ info->xbdev = dev;
++ info->irq = -1;
++ info->x1 = info->y1 = INT_MAX;
++ spin_lock_init(&info->dirty_lock);
++ spin_lock_init(&info->resize_lock);
++ mutex_init(&info->mm_lock);
++ init_waitqueue_head(&info->wq);
++ init_timer(&info->refresh);
++ info->refresh.function = xenfb_timer;
++ info->refresh.data = (unsigned long)info;
++ INIT_LIST_HEAD(&info->mappings);
++
++ info->fb = vmalloc(fb_size);
++ if (info->fb == NULL)
++ goto error_nomem;
++ memset(info->fb, 0, fb_size);
++
++ info->nr_pages = (fb_size + PAGE_SIZE - 1) >> PAGE_SHIFT;
++
++ info->pages = kmalloc(sizeof(struct page *) * info->nr_pages,
++ GFP_KERNEL);
++ if (info->pages == NULL)
++ goto error_nomem;
++
++ info->mfns = vmalloc(sizeof(unsigned long) * info->nr_pages);
++ if (!info->mfns)
++ goto error_nomem;
++
++ /* set up shared page */
++ info->page = (void *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
++ if (!info->page)
++ goto error_nomem;
++
++ fb_info = framebuffer_alloc(sizeof(u32) * 256, NULL);
++ /* see fishy hackery below */
++ if (fb_info == NULL)
++ goto error_nomem;
++
++ /* FIXME fishy hackery */
++ fb_info->pseudo_palette = fb_info->par;
++ fb_info->par = info;
++ /* /FIXME */
++ fb_info->screen_base = info->fb;
++
++ fb_info->fbops = &xenfb_fb_ops;
++ fb_info->var.xres_virtual = fb_info->var.xres = video[KPARAM_WIDTH];
++ fb_info->var.yres_virtual = fb_info->var.yres = video[KPARAM_HEIGHT];
++ fb_info->var.bits_per_pixel = XENFB_DEPTH;
++
++ fb_info->var.red = (struct fb_bitfield){16, 8, 0};
++ fb_info->var.green = (struct fb_bitfield){8, 8, 0};
++ fb_info->var.blue = (struct fb_bitfield){0, 8, 0};
++
++ fb_info->var.activate = FB_ACTIVATE_NOW;
++ fb_info->var.height = -1;
++ fb_info->var.width = -1;
++ fb_info->var.vmode = FB_VMODE_NONINTERLACED;
++
++ fb_info->fix.visual = FB_VISUAL_TRUECOLOR;
++ fb_info->fix.line_length = fb_info->var.xres * (XENFB_DEPTH / 8);
++ fb_info->fix.smem_start = 0;
++ fb_info->fix.smem_len = fb_size;
++ strcpy(fb_info->fix.id, "xen");
++ fb_info->fix.type = FB_TYPE_PACKED_PIXELS;
++ fb_info->fix.accel = FB_ACCEL_NONE;
++
++ fb_info->flags = FBINFO_FLAG_DEFAULT;
++
++ ret = fb_alloc_cmap(&fb_info->cmap, 256, 0);
++ if (ret < 0) {
++ framebuffer_release(fb_info);
++ xenbus_dev_fatal(dev, ret, "fb_alloc_cmap");
++ goto error;
++ }
++
++ xenfb_init_shared_page(info, fb_info);
++
++ ret = register_framebuffer(fb_info);
++ if (ret) {
++ fb_dealloc_cmap(&info->fb_info->cmap);
++ framebuffer_release(fb_info);
++ xenbus_dev_fatal(dev, ret, "register_framebuffer");
++ goto error;
++ }
++ info->fb_info = fb_info;
++
++ /* FIXME should this be delayed until backend XenbusStateConnected? */
++ info->kthread = kthread_run(xenfb_thread, info, "xenfb thread");
++ if (IS_ERR(info->kthread)) {
++ ret = PTR_ERR(info->kthread);
++ info->kthread = NULL;
++ xenbus_dev_fatal(dev, ret, "register_framebuffer");
++ goto error;
++ }
++
++ ret = xenfb_connect_backend(dev, info);
++ if (ret < 0)
++ goto error;
++
++ return 0;
++
++ error_nomem:
++ ret = -ENOMEM;
++ xenbus_dev_fatal(dev, ret, "allocating device memory");
++ error:
++ xenfb_remove(dev);
++ return ret;
++}
++
++static int xenfb_resume(struct xenbus_device *dev)
++{
++ struct xenfb_info *info = dev->dev.driver_data;
++
++ xenfb_disconnect_backend(info);
++ xenfb_init_shared_page(info, info->fb_info);
++ return xenfb_connect_backend(dev, info);
++}
++
++static int xenfb_remove(struct xenbus_device *dev)
++{
++ struct xenfb_info *info = dev->dev.driver_data;
++
++ del_timer(&info->refresh);
++ if (info->kthread)
++ kthread_stop(info->kthread);
++ xenfb_disconnect_backend(info);
++ if (info->fb_info) {
++ unregister_framebuffer(info->fb_info);
++ fb_dealloc_cmap(&info->fb_info->cmap);
++ framebuffer_release(info->fb_info);
++ }
++ free_page((unsigned long)info->page);
++ vfree(info->mfns);
++ kfree(info->pages);
++ vfree(info->fb);
++ kfree(info);
++
++ return 0;
++}
++
++static void xenfb_init_shared_page(struct xenfb_info *info,
++ struct fb_info * fb_info)
++{
++ int i;
++ int epd = PAGE_SIZE / sizeof(info->mfns[0]);
++
++ for (i = 0; i < info->nr_pages; i++)
++ info->pages[i] = vmalloc_to_page(info->fb + i * PAGE_SIZE);
++
++ for (i = 0; i < info->nr_pages; i++)
++ info->mfns[i] = vmalloc_to_mfn(info->fb + i * PAGE_SIZE);
++
++ for (i = 0; i * epd < info->nr_pages; i++)
++ info->page->pd[i] = vmalloc_to_mfn(&info->mfns[i * epd]);
++
++ info->page->width = fb_info->var.xres;
++ info->page->height = fb_info->var.yres;
++ info->page->depth = fb_info->var.bits_per_pixel;
++ info->page->line_length = fb_info->fix.line_length;
++ info->page->mem_length = fb_info->fix.smem_len;
++ info->page->in_cons = info->page->in_prod = 0;
++ info->page->out_cons = info->page->out_prod = 0;
++}
++
++static int xenfb_connect_backend(struct xenbus_device *dev,
++ struct xenfb_info *info)
++{
++ int ret;
++ struct xenbus_transaction xbt;
++
++ ret = bind_listening_port_to_irqhandler(
++ dev->otherend_id, xenfb_event_handler, 0, "xenfb", info);
++ if (ret < 0) {
++ xenbus_dev_fatal(dev, ret,
++ "bind_listening_port_to_irqhandler");
++ return ret;
++ }
++ info->irq = ret;
++
++ again:
++ ret = xenbus_transaction_start(&xbt);
++ if (ret) {
++ xenbus_dev_fatal(dev, ret, "starting transaction");
++ return ret;
++ }
++ ret = xenbus_printf(xbt, dev->nodename, "page-ref", "%lu",
++ virt_to_mfn(info->page));
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
++ irq_to_evtchn_port(info->irq));
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
++ XEN_IO_PROTO_ABI_NATIVE);
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_printf(xbt, dev->nodename, "feature-update", "1");
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_transaction_end(xbt, 0);
++ if (ret) {
++ if (ret == -EAGAIN)
++ goto again;
++ xenbus_dev_fatal(dev, ret, "completing transaction");
++ return ret;
++ }
++
++ xenbus_switch_state(dev, XenbusStateInitialised);
++ return 0;
++
++ error_xenbus:
++ xenbus_transaction_end(xbt, 1);
++ xenbus_dev_fatal(dev, ret, "writing xenstore");
++ return ret;
++}
++
++static void xenfb_disconnect_backend(struct xenfb_info *info)
++{
++ if (info->irq >= 0)
++ unbind_from_irqhandler(info->irq, info);
++ info->irq = -1;
++}
++
++static void xenfb_backend_changed(struct xenbus_device *dev,
++ enum xenbus_state backend_state)
++{
++ struct xenfb_info *info = dev->dev.driver_data;
++ int val;
++
++ switch (backend_state) {
++ case XenbusStateInitialising:
++ case XenbusStateInitialised:
++ case XenbusStateReconfiguring:
++ case XenbusStateReconfigured:
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ break;
++
++ case XenbusStateInitWait:
++ InitWait:
++ xenbus_switch_state(dev, XenbusStateConnected);
++ break;
++
++ case XenbusStateConnected:
++ /*
++ * Work around xenbus race condition: If backend goes
++ * through InitWait to Connected fast enough, we can
++ * get Connected twice here.
++ */
++ if (dev->state != XenbusStateConnected)
++ goto InitWait; /* no InitWait seen yet, fudge it */
++
++ if (xenbus_scanf(XBT_NIL, info->xbdev->otherend,
++ "request-update", "%d", &val) < 0)
++ val = 0;
++ if (val)
++ info->update_wanted = 1;
++
++ if (xenbus_scanf(XBT_NIL, dev->otherend,
++ "feature-resize", "%d", &val) < 0)
++ val = 0;
++ info->feature_resize = val;
++ break;
++
++ case XenbusStateClosing:
++ // FIXME is this safe in any dev->state?
++ xenbus_frontend_closed(dev);
++ break;
++ }
++}
++
++static const struct xenbus_device_id xenfb_ids[] = {
++ { "vfb" },
++ { "" }
++};
++MODULE_ALIAS("xen:vfb");
++
++static struct xenbus_driver xenfb_driver = {
++ .name = "vfb",
++ .owner = THIS_MODULE,
++ .ids = xenfb_ids,
++ .probe = xenfb_probe,
++ .remove = xenfb_remove,
++ .resume = xenfb_resume,
++ .otherend_changed = xenfb_backend_changed,
++};
++
++static int __init xenfb_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ /* Nothing to do if running in dom0. */
++ if (is_initial_xendomain())
++ return -ENODEV;
++
++ return xenbus_register_frontend(&xenfb_driver);
++}
++
++static void __exit xenfb_cleanup(void)
++{
++ return xenbus_unregister_driver(&xenfb_driver);
++}
++
++module_init(xenfb_init);
++module_exit(xenfb_cleanup);
++
++MODULE_LICENSE("GPL");
+--- linux-2.6.18.8/drivers/xen/fbfront/xenkbd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/fbfront/xenkbd.c 2008-05-19 00:33:45.834788090 +0300
+@@ -0,0 +1,354 @@
++/*
++ * linux/drivers/input/keyboard/xenkbd.c -- Xen para-virtual input device
++ *
++ * Copyright (C) 2005 Anthony Liguori <aliguori@us.ibm.com>
++ * Copyright (C) 2006 Red Hat, Inc., Markus Armbruster <armbru@redhat.com>
++ *
++ * Based on linux/drivers/input/mouse/sermouse.c
++ *
++ * This file is subject to the terms and conditions of the GNU General Public
++ * License. See the file COPYING in the main directory of this archive for
++ * more details.
++ */
++
++/*
++ * TODO:
++ *
++ * Switch to grant tables together with xenfb.c.
++ */
++
++#include <linux/kernel.h>
++#include <linux/errno.h>
++#include <linux/module.h>
++#include <linux/input.h>
++#include <asm/hypervisor.h>
++#include <xen/evtchn.h>
++#include <xen/interface/io/fbif.h>
++#include <xen/interface/io/kbdif.h>
++#include <xen/xenbus.h>
++
++struct xenkbd_info
++{
++ struct input_dev *kbd;
++ struct input_dev *ptr;
++ struct xenkbd_page *page;
++ int irq;
++ struct xenbus_device *xbdev;
++ char phys[32];
++};
++
++static int xenkbd_remove(struct xenbus_device *);
++static int xenkbd_connect_backend(struct xenbus_device *, struct xenkbd_info *);
++static void xenkbd_disconnect_backend(struct xenkbd_info *);
++
++/*
++ * Note: if you need to send out events, see xenfb_do_update() for how
++ * to do that.
++ */
++
++static irqreturn_t input_handler(int rq, void *dev_id, struct pt_regs *regs)
++{
++ struct xenkbd_info *info = dev_id;
++ struct xenkbd_page *page = info->page;
++ __u32 cons, prod;
++
++ prod = page->in_prod;
++ if (prod == page->in_cons)
++ return IRQ_HANDLED;
++ rmb(); /* ensure we see ring contents up to prod */
++ for (cons = page->in_cons; cons != prod; cons++) {
++ union xenkbd_in_event *event;
++ struct input_dev *dev;
++ event = &XENKBD_IN_RING_REF(page, cons);
++
++ dev = info->ptr;
++ switch (event->type) {
++ case XENKBD_TYPE_MOTION:
++ if (event->motion.rel_z)
++ input_report_rel(dev, REL_WHEEL,
++ -event->motion.rel_z);
++ input_report_rel(dev, REL_X, event->motion.rel_x);
++ input_report_rel(dev, REL_Y, event->motion.rel_y);
++ break;
++ case XENKBD_TYPE_KEY:
++ dev = NULL;
++ if (test_bit(event->key.keycode, info->kbd->keybit))
++ dev = info->kbd;
++ if (test_bit(event->key.keycode, info->ptr->keybit))
++ dev = info->ptr;
++ if (dev)
++ input_report_key(dev, event->key.keycode,
++ event->key.pressed);
++ else
++ printk("xenkbd: unhandled keycode 0x%x\n",
++ event->key.keycode);
++ break;
++ case XENKBD_TYPE_POS:
++ if (event->pos.rel_z)
++ input_report_rel(dev, REL_WHEEL,
++ -event->pos.rel_z);
++ input_report_abs(dev, ABS_X, event->pos.abs_x);
++ input_report_abs(dev, ABS_Y, event->pos.abs_y);
++ break;
++ }
++ if (dev)
++ input_sync(dev);
++ }
++ mb(); /* ensure we got ring contents */
++ page->in_cons = cons;
++ notify_remote_via_irq(info->irq);
++
++ return IRQ_HANDLED;
++}
++
++int __devinit xenkbd_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int ret, i;
++ struct xenkbd_info *info;
++ struct input_dev *kbd, *ptr;
++
++ info = kzalloc(sizeof(*info), GFP_KERNEL);
++ if (!info) {
++ xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
++ return -ENOMEM;
++ }
++ dev->dev.driver_data = info;
++ info->xbdev = dev;
++ snprintf(info->phys, sizeof(info->phys), "xenbus/%s", dev->nodename);
++
++ info->page = (void *)__get_free_page(GFP_KERNEL);
++ if (!info->page)
++ goto error_nomem;
++ info->page->in_cons = info->page->in_prod = 0;
++ info->page->out_cons = info->page->out_prod = 0;
++
++ /* keyboard */
++ kbd = input_allocate_device();
++ if (!kbd)
++ goto error_nomem;
++ kbd->name = "Xen Virtual Keyboard";
++ kbd->phys = info->phys;
++ kbd->id.bustype = BUS_PCI;
++ kbd->id.vendor = 0x5853;
++ kbd->id.product = 0xffff;
++ kbd->evbit[0] = BIT(EV_KEY);
++ for (i = KEY_ESC; i < KEY_UNKNOWN; i++)
++ set_bit(i, kbd->keybit);
++ for (i = KEY_OK; i < KEY_MAX; i++)
++ set_bit(i, kbd->keybit);
++
++ ret = input_register_device(kbd);
++ if (ret) {
++ input_free_device(kbd);
++ xenbus_dev_fatal(dev, ret, "input_register_device(kbd)");
++ goto error;
++ }
++ info->kbd = kbd;
++
++ /* pointing device */
++ ptr = input_allocate_device();
++ if (!ptr)
++ goto error_nomem;
++ ptr->name = "Xen Virtual Pointer";
++ ptr->phys = info->phys;
++ ptr->id.bustype = BUS_PCI;
++ ptr->id.vendor = 0x5853;
++ ptr->id.product = 0xfffe;
++ ptr->evbit[0] = BIT(EV_KEY) | BIT(EV_REL) | BIT(EV_ABS);
++ for (i = BTN_LEFT; i <= BTN_TASK; i++)
++ set_bit(i, ptr->keybit);
++ ptr->relbit[0] = BIT(REL_X) | BIT(REL_Y) | BIT(REL_WHEEL);
++ input_set_abs_params(ptr, ABS_X, 0, XENFB_WIDTH, 0, 0);
++ input_set_abs_params(ptr, ABS_Y, 0, XENFB_HEIGHT, 0, 0);
++
++ ret = input_register_device(ptr);
++ if (ret) {
++ input_free_device(ptr);
++ xenbus_dev_fatal(dev, ret, "input_register_device(ptr)");
++ goto error;
++ }
++ info->ptr = ptr;
++
++ ret = xenkbd_connect_backend(dev, info);
++ if (ret < 0)
++ goto error;
++
++ return 0;
++
++ error_nomem:
++ ret = -ENOMEM;
++ xenbus_dev_fatal(dev, ret, "allocating device memory");
++ error:
++ xenkbd_remove(dev);
++ return ret;
++}
++
++static int xenkbd_resume(struct xenbus_device *dev)
++{
++ struct xenkbd_info *info = dev->dev.driver_data;
++
++ xenkbd_disconnect_backend(info);
++ info->page->in_cons = info->page->in_prod = 0;
++ info->page->out_cons = info->page->out_prod = 0;
++ return xenkbd_connect_backend(dev, info);
++}
++
++static int xenkbd_remove(struct xenbus_device *dev)
++{
++ struct xenkbd_info *info = dev->dev.driver_data;
++
++ xenkbd_disconnect_backend(info);
++ input_unregister_device(info->kbd);
++ input_unregister_device(info->ptr);
++ free_page((unsigned long)info->page);
++ kfree(info);
++ return 0;
++}
++
++static int xenkbd_connect_backend(struct xenbus_device *dev,
++ struct xenkbd_info *info)
++{
++ int ret;
++ struct xenbus_transaction xbt;
++
++ ret = bind_listening_port_to_irqhandler(
++ dev->otherend_id, input_handler, 0, "xenkbd", info);
++ if (ret < 0) {
++ xenbus_dev_fatal(dev, ret,
++ "bind_listening_port_to_irqhandler");
++ return ret;
++ }
++ info->irq = ret;
++
++ again:
++ ret = xenbus_transaction_start(&xbt);
++ if (ret) {
++ xenbus_dev_fatal(dev, ret, "starting transaction");
++ return ret;
++ }
++ ret = xenbus_printf(xbt, dev->nodename, "page-ref", "%lu",
++ virt_to_mfn(info->page));
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
++ irq_to_evtchn_port(info->irq));
++ if (ret)
++ goto error_xenbus;
++ ret = xenbus_transaction_end(xbt, 0);
++ if (ret) {
++ if (ret == -EAGAIN)
++ goto again;
++ xenbus_dev_fatal(dev, ret, "completing transaction");
+ return ret;
+ }
-+ ret = xenbus_printf(xbt, dev->nodename, "page-ref", "%lu",
-+ virt_to_mfn(info->page));
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
-+ irq_to_evtchn_port(info->irq));
-+ if (ret)
-+ goto error_xenbus;
-+ ret = xenbus_transaction_end(xbt, 0);
-+ if (ret) {
-+ if (ret == -EAGAIN)
-+ goto again;
-+ xenbus_dev_fatal(dev, ret, "completing transaction");
-+ return ret;
++
++ xenbus_switch_state(dev, XenbusStateInitialised);
++ return 0;
++
++ error_xenbus:
++ xenbus_transaction_end(xbt, 1);
++ xenbus_dev_fatal(dev, ret, "writing xenstore");
++ return ret;
++}
++
++static void xenkbd_disconnect_backend(struct xenkbd_info *info)
++{
++ if (info->irq >= 0)
++ unbind_from_irqhandler(info->irq, info);
++ info->irq = -1;
++}
++
++static void xenkbd_backend_changed(struct xenbus_device *dev,
++ enum xenbus_state backend_state)
++{
++ struct xenkbd_info *info = dev->dev.driver_data;
++ int ret, val;
++
++ switch (backend_state) {
++ case XenbusStateInitialising:
++ case XenbusStateInitialised:
++ case XenbusStateReconfiguring:
++ case XenbusStateReconfigured:
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ break;
++
++ case XenbusStateInitWait:
++ InitWait:
++ ret = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
++ "feature-abs-pointer", "%d", &val);
++ if (ret < 0)
++ val = 0;
++ if (val) {
++ ret = xenbus_printf(XBT_NIL, info->xbdev->nodename,
++ "request-abs-pointer", "1");
++ if (ret)
++ ; /* FIXME */
++ }
++ xenbus_switch_state(dev, XenbusStateConnected);
++ break;
++
++ case XenbusStateConnected:
++ /*
++ * Work around xenbus race condition: If backend goes
++ * through InitWait to Connected fast enough, we can
++ * get Connected twice here.
++ */
++ if (dev->state != XenbusStateConnected)
++ goto InitWait; /* no InitWait seen yet, fudge it */
++
++ /* Set input abs params to match backend screen res */
++ if (xenbus_scanf(XBT_NIL, info->xbdev->otherend,
++ "width", "%d", &val) > 0 )
++ input_set_abs_params(info->ptr, ABS_X, 0, val, 0, 0);
++
++ if (xenbus_scanf(XBT_NIL, info->xbdev->otherend,
++ "height", "%d", &val) > 0 )
++ input_set_abs_params(info->ptr, ABS_Y, 0, val, 0, 0);
++
++ break;
++
++ case XenbusStateClosing:
++ xenbus_frontend_closed(dev);
++ break;
++ }
++}
++
++static const struct xenbus_device_id xenkbd_ids[] = {
++ { "vkbd" },
++ { "" }
++};
++MODULE_ALIAS("xen:vkbd");
++
++static struct xenbus_driver xenkbd_driver = {
++ .name = "vkbd",
++ .owner = THIS_MODULE,
++ .ids = xenkbd_ids,
++ .probe = xenkbd_probe,
++ .remove = xenkbd_remove,
++ .resume = xenkbd_resume,
++ .otherend_changed = xenkbd_backend_changed,
++};
++
++static int __init xenkbd_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ /* Nothing to do if running in dom0. */
++ if (is_initial_xendomain())
++ return -ENODEV;
++
++ return xenbus_register_frontend(&xenkbd_driver);
++}
++
++static void __exit xenkbd_cleanup(void)
++{
++ return xenbus_unregister_driver(&xenkbd_driver);
++}
++
++module_init(xenkbd_init);
++module_exit(xenkbd_cleanup);
++
++MODULE_LICENSE("GPL");
+--- linux-2.6.18.8/drivers/xen/gntdev/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/gntdev/Makefile 2008-05-19 00:33:45.834788090 +0300
+@@ -0,0 +1 @@
++obj-$(CONFIG_XEN_GRANT_DEV) := gntdev.o
+--- linux-2.6.18.8/drivers/xen/gntdev/gntdev.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/gntdev/gntdev.c 2008-05-19 00:33:45.838788321 +0300
+@@ -0,0 +1,1077 @@
++/******************************************************************************
++ * gntdev.c
++ *
++ * Device for accessing (in user-space) pages that have been granted by other
++ * domains.
++ *
++ * Copyright (c) 2006-2007, D G Murray.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ */
++
++#include <asm/atomic.h>
++#include <linux/module.h>
++#include <linux/kernel.h>
++#include <linux/init.h>
++#include <linux/fs.h>
++#include <linux/device.h>
++#include <linux/mm.h>
++#include <linux/mman.h>
++#include <asm/uaccess.h>
++#include <asm/io.h>
++#include <xen/gnttab.h>
++#include <asm/hypervisor.h>
++#include <xen/balloon.h>
++#include <xen/evtchn.h>
++#include <xen/driver_util.h>
++
++#include <linux/types.h>
++#include <xen/public/gntdev.h>
++
++
++#define DRIVER_AUTHOR "Derek G. Murray <Derek.Murray@cl.cam.ac.uk>"
++#define DRIVER_DESC "User-space granted page access driver"
++
++MODULE_LICENSE("GPL");
++MODULE_AUTHOR(DRIVER_AUTHOR);
++MODULE_DESCRIPTION(DRIVER_DESC);
++
++#define MAX_GRANTS_LIMIT 1024
++#define DEFAULT_MAX_GRANTS 128
++
++/* A slot can be in one of three states:
++ *
++ * 0. GNTDEV_SLOT_INVALID:
++ * This slot is not associated with a grant reference, and is therefore free
++ * to be overwritten by a new grant reference.
++ *
++ * 1. GNTDEV_SLOT_NOT_YET_MAPPED:
++ * This slot is associated with a grant reference (via the
++ * IOCTL_GNTDEV_MAP_GRANT_REF ioctl), but it has not yet been mmap()-ed.
++ *
++ * 2. GNTDEV_SLOT_MAPPED:
++ * This slot is associated with a grant reference, and has been mmap()-ed.
++ */
++typedef enum gntdev_slot_state {
++ GNTDEV_SLOT_INVALID = 0,
++ GNTDEV_SLOT_NOT_YET_MAPPED,
++ GNTDEV_SLOT_MAPPED
++} gntdev_slot_state_t;
++
++#define GNTDEV_INVALID_HANDLE -1
++#define GNTDEV_FREE_LIST_INVALID -1
++/* Each opened instance of gntdev is associated with a list of grants,
++ * represented by an array of elements of the following type,
++ * gntdev_grant_info_t.
++ */
++typedef struct gntdev_grant_info {
++ gntdev_slot_state_t state;
++ union {
++ uint32_t free_list_index;
++ struct {
++ domid_t domid;
++ grant_ref_t ref;
++ grant_handle_t kernel_handle;
++ grant_handle_t user_handle;
++ uint64_t dev_bus_addr;
++ } valid;
++ } u;
++} gntdev_grant_info_t;
++
++/* Private data structure, which is stored in the file pointer for files
++ * associated with this device.
++ */
++typedef struct gntdev_file_private_data {
++
++ /* Array of grant information. */
++ gntdev_grant_info_t *grants;
++ uint32_t grants_size;
++
++ /* Read/write semaphore used to protect the grants array. */
++ struct rw_semaphore grants_sem;
++
++ /* An array of indices of free slots in the grants array.
++ * N.B. An entry in this list may temporarily have the value
++ * GNTDEV_FREE_LIST_INVALID if the corresponding slot has been removed
++ * from the list by the contiguous allocator, but the list has not yet
++ * been compressed. However, this is not visible across invocations of
++ * the device.
++ */
++ int32_t *free_list;
++
++ /* The number of free slots in the grants array. */
++ uint32_t free_list_size;
++
++ /* Read/write semaphore used to protect the free list. */
++ struct rw_semaphore free_list_sem;
++
++ /* Index of the next slot after the most recent contiguous allocation,
++ * for use in a next-fit allocator.
++ */
++ uint32_t next_fit_index;
++
++ /* Used to map grants into the kernel, before mapping them into user
++ * space.
++ */
++ struct page **foreign_pages;
++
++} gntdev_file_private_data_t;
++
++/* Module lifecycle operations. */
++static int __init gntdev_init(void);
++static void __exit gntdev_exit(void);
++
++module_init(gntdev_init);
++module_exit(gntdev_exit);
++
++/* File operations. */
++static int gntdev_open(struct inode *inode, struct file *flip);
++static int gntdev_release(struct inode *inode, struct file *flip);
++static int gntdev_mmap(struct file *flip, struct vm_area_struct *vma);
++static long gntdev_ioctl(struct file *flip,
++ unsigned int cmd, unsigned long arg);
++
++static const struct file_operations gntdev_fops = {
++ .owner = THIS_MODULE,
++ .open = gntdev_open,
++ .release = gntdev_release,
++ .mmap = gntdev_mmap,
++ .unlocked_ioctl = gntdev_ioctl
++};
++
++/* VM operations. */
++static void gntdev_vma_close(struct vm_area_struct *vma);
++static pte_t gntdev_clear_pte(struct vm_area_struct *vma, unsigned long addr,
++ pte_t *ptep, int is_fullmm);
++
++static struct vm_operations_struct gntdev_vmops = {
++ .close = gntdev_vma_close,
++ .zap_pte = gntdev_clear_pte
++};
++
++/* Global variables. */
++
++/* The driver major number, for use when unregistering the driver. */
++static int gntdev_major;
++
++#define GNTDEV_NAME "gntdev"
++
++/* Memory mapping functions
++ * ------------------------
++ *
++ * Every granted page is mapped into both kernel and user space, and the two
++ * following functions return the respective virtual addresses of these pages.
++ *
++ * When shadow paging is disabled, the granted page is mapped directly into
++ * user space; when it is enabled, it is mapped into the kernel and remapped
++ * into user space using vm_insert_page() (see gntdev_mmap(), below).
++ */
++
++/* Returns the virtual address (in user space) of the @page_index'th page
++ * in the given VM area.
++ */
++static inline unsigned long get_user_vaddr (struct vm_area_struct *vma,
++ int page_index)
++{
++ return (unsigned long) vma->vm_start + (page_index << PAGE_SHIFT);
++}
++
++/* Returns the virtual address (in kernel space) of the @slot_index'th page
++ * mapped by the gntdev instance that owns the given private data struct.
++ */
++static inline unsigned long get_kernel_vaddr (gntdev_file_private_data_t *priv,
++ int slot_index)
++{
++ unsigned long pfn;
++ void *kaddr;
++ pfn = page_to_pfn(priv->foreign_pages[slot_index]);
++ kaddr = pfn_to_kaddr(pfn);
++ return (unsigned long) kaddr;
++}
++
++/* Helper functions. */
++
++/* Adds information about a grant reference to the list of grants in the file's
++ * private data structure. Returns non-zero on failure. On success, sets the
++ * value of *offset to the offset that should be mmap()-ed in order to map the
++ * grant reference.
++ */
++static int add_grant_reference(struct file *flip,
++ struct ioctl_gntdev_grant_ref *op,
++ uint64_t *offset)
++{
++ gntdev_file_private_data_t *private_data
++ = (gntdev_file_private_data_t *) flip->private_data;
++
++ uint32_t slot_index;
++
++ if (unlikely(private_data->free_list_size == 0)) {
++ return -ENOMEM;
++ }
++
++ slot_index = private_data->free_list[--private_data->free_list_size];
++ private_data->free_list[private_data->free_list_size]
++ = GNTDEV_FREE_LIST_INVALID;
++
++ /* Copy the grant information into file's private data. */
++ private_data->grants[slot_index].state = GNTDEV_SLOT_NOT_YET_MAPPED;
++ private_data->grants[slot_index].u.valid.domid = op->domid;
++ private_data->grants[slot_index].u.valid.ref = op->ref;
++
++ /* The offset is calculated as the index of the chosen entry in the
++ * file's private data's array of grant information. This is then
++ * shifted to give an offset into the virtual "file address space".
++ */
++ *offset = slot_index << PAGE_SHIFT;
++
++ return 0;
++}
++
++/* Adds the @count grant references to the contiguous range in the slot array
++ * beginning at @first_slot. It is assumed that @first_slot was returned by a
++ * previous invocation of find_contiguous_free_range(), during the same
++ * invocation of the driver.
++ */
++static int add_grant_references(struct file *flip,
++ int count,
++ struct ioctl_gntdev_grant_ref *ops,
++ uint32_t first_slot)
++{
++ gntdev_file_private_data_t *private_data
++ = (gntdev_file_private_data_t *) flip->private_data;
++ int i;
++
++ for (i = 0; i < count; ++i) {
++
++ /* First, mark the slot's entry in the free list as invalid. */
++ int free_list_index =
++ private_data->grants[first_slot+i].u.free_list_index;
++ private_data->free_list[free_list_index] =
++ GNTDEV_FREE_LIST_INVALID;
++
++ /* Now, update the slot. */
++ private_data->grants[first_slot+i].state =
++ GNTDEV_SLOT_NOT_YET_MAPPED;
++ private_data->grants[first_slot+i].u.valid.domid =
++ ops[i].domid;
++ private_data->grants[first_slot+i].u.valid.ref = ops[i].ref;
++ }
++
++ return 0;
++}
++
++/* Scans through the free list for @flip, removing entries that are marked as
++ * GNTDEV_SLOT_INVALID. This will reduce the recorded size of the free list to
++ * the number of valid entries.
++ */
++static void compress_free_list(struct file *flip)
++{
++ gntdev_file_private_data_t *private_data
++ = (gntdev_file_private_data_t *) flip->private_data;
++ int i, j = 0, old_size, slot_index;
++
++ old_size = private_data->free_list_size;
++ for (i = 0; i < old_size; ++i) {
++ if (private_data->free_list[i] != GNTDEV_FREE_LIST_INVALID) {
++ if (i > j) {
++ slot_index = private_data->free_list[i];
++ private_data->free_list[j] = slot_index;
++ private_data->grants[slot_index].u
++ .free_list_index = j;
++ private_data->free_list[i]
++ = GNTDEV_FREE_LIST_INVALID;
++ }
++ ++j;
++ } else {
++ --private_data->free_list_size;
++ }
++ }
++}
++
++/* Searches the grant array in the private data of @flip for a range of
++ * @num_slots contiguous slots in the GNTDEV_SLOT_INVALID state.
++ *
++ * Returns the index of the first slot if a range is found, otherwise -ENOMEM.
++ */
++static int find_contiguous_free_range(struct file *flip,
++ uint32_t num_slots)
++{
++ gntdev_file_private_data_t *private_data
++ = (gntdev_file_private_data_t *) flip->private_data;
++
++ int i;
++ int start_index = private_data->next_fit_index;
++ int range_start = 0, range_length;
++
++ if (private_data->free_list_size < num_slots) {
++ return -ENOMEM;
++ }
++
++ /* First search from the start_index to the end of the array. */
++ range_length = 0;
++ for (i = start_index; i < private_data->grants_size; ++i) {
++ if (private_data->grants[i].state == GNTDEV_SLOT_INVALID) {
++ if (range_length == 0) {
++ range_start = i;
++ }
++ ++range_length;
++ if (range_length == num_slots) {
++ return range_start;
++ }
++ }
++ }
++
++ /* Now search from the start of the array to the start_index. */
++ range_length = 0;
++ for (i = 0; i < start_index; ++i) {
++ if (private_data->grants[i].state == GNTDEV_SLOT_INVALID) {
++ if (range_length == 0) {
++ range_start = i;
++ }
++ ++range_length;
++ if (range_length == num_slots) {
++ return range_start;
++ }
++ }
++ }
++
++ return -ENOMEM;
++}
++
++static int init_private_data(gntdev_file_private_data_t *priv,
++ uint32_t max_grants)
++{
++ int i;
++
++ /* Allocate space for the kernel-mapping of granted pages. */
++ priv->foreign_pages =
++ alloc_empty_pages_and_pagevec(max_grants);
++ if (!priv->foreign_pages)
++ goto nomem_out;
++
++ /* Allocate the grant list and free-list. */
++ priv->grants = kmalloc(max_grants * sizeof(gntdev_grant_info_t),
++ GFP_KERNEL);
++ if (!priv->grants)
++ goto nomem_out2;
++ priv->free_list = kmalloc(max_grants * sizeof(int32_t), GFP_KERNEL);
++ if (!priv->free_list)
++ goto nomem_out3;
++
++ /* Initialise the free-list, which contains all slots at first. */
++ for (i = 0; i < max_grants; ++i) {
++ priv->free_list[max_grants - i - 1] = i;
++ priv->grants[i].state = GNTDEV_SLOT_INVALID;
++ priv->grants[i].u.free_list_index = max_grants - i - 1;
++ }
++ priv->grants_size = max_grants;
++ priv->free_list_size = max_grants;
++ priv->next_fit_index = 0;
++
++ up_write(&priv->grants_sem);
++ up_write(&priv->free_list_sem);
++
++ return 0;
++
++nomem_out3:
++ kfree(priv->grants);
++nomem_out2:
++ free_empty_pages_and_pagevec(priv->foreign_pages, max_grants);
++nomem_out:
++ return -ENOMEM;
++
++}
++
++/* Interface functions. */
++
++/* Initialises the driver. Called when the module is loaded. */
++static int __init gntdev_init(void)
++{
++ struct class *class;
++ struct class_device *device;
++
++ if (!is_running_on_xen()) {
++ printk(KERN_ERR "You must be running Xen to use gntdev\n");
++ return -ENODEV;
++ }
++
++ gntdev_major = register_chrdev(0, GNTDEV_NAME, &gntdev_fops);
++ if (gntdev_major < 0)
++ {
++ printk(KERN_ERR "Could not register gntdev device\n");
++ return -ENOMEM;
++ }
++
++ /* Note that if the sysfs code fails, we will still initialise the
++ * device, and output the major number so that the device can be
++ * created manually using mknod.
++ */
++ if ((class = get_xen_class()) == NULL) {
++ printk(KERN_ERR "Error setting up xen_class\n");
++ printk(KERN_ERR "gntdev created with major number = %d\n",
++ gntdev_major);
++ return 0;
++ }
++
++ device = class_device_create(class, NULL, MKDEV(gntdev_major, 0),
++ NULL, GNTDEV_NAME);
++ if (IS_ERR(device)) {
++ printk(KERN_ERR "Error creating gntdev device in xen_class\n");
++ printk(KERN_ERR "gntdev created with major number = %d\n",
++ gntdev_major);
++ return 0;
++ }
++
++ return 0;
++}
++
++/* Cleans up and unregisters the driver. Called when the driver is unloaded.
++ */
++static void __exit gntdev_exit(void)
++{
++ struct class *class;
++ if ((class = get_xen_class()) != NULL)
++ class_device_destroy(class, MKDEV(gntdev_major, 0));
++ unregister_chrdev(gntdev_major, GNTDEV_NAME);
++}
++
++/* Called when the device is opened. */
++static int gntdev_open(struct inode *inode, struct file *flip)
++{
++ gntdev_file_private_data_t *private_data;
++
++ try_module_get(THIS_MODULE);
++
++ /* Allocate space for the per-instance private data. */
++ private_data = kmalloc(sizeof(*private_data), GFP_KERNEL);
++ if (!private_data)
++ goto nomem_out;
++
++ /* These will be lazily initialised by init_private_data. */
++ private_data->grants = NULL;
++ private_data->free_list = NULL;
++ private_data->foreign_pages = NULL;
++
++ init_rwsem(&private_data->grants_sem);
++ init_rwsem(&private_data->free_list_sem);
++
++ flip->private_data = private_data;
++
++ return 0;
++
++nomem_out:
++ return -ENOMEM;
++}
++
++/* Called when the device is closed.
++ */
++static int gntdev_release(struct inode *inode, struct file *flip)
++{
++ if (flip->private_data) {
++ gntdev_file_private_data_t *private_data =
++ (gntdev_file_private_data_t *) flip->private_data;
++ if (private_data->foreign_pages)
++ free_empty_pages_and_pagevec
++ (private_data->foreign_pages,
++ private_data->grants_size);
++ if (private_data->grants)
++ kfree(private_data->grants);
++ if (private_data->free_list)
++ kfree(private_data->free_list);
++ kfree(private_data);
++ }
++ module_put(THIS_MODULE);
++ return 0;
++}
++
++/* Called when an attempt is made to mmap() the device. The private data from
++ * @flip contains the list of grant references that can be mapped. The vm_pgoff
++ * field of @vma contains the index into that list that refers to the grant
++ * reference that will be mapped. Only mappings that are a multiple of
++ * PAGE_SIZE are handled.
++ */
++static int gntdev_mmap (struct file *flip, struct vm_area_struct *vma)
++{
++ struct gnttab_map_grant_ref op;
++ unsigned long slot_index = vma->vm_pgoff;
++ unsigned long kernel_vaddr, user_vaddr;
++ uint32_t size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
++ uint64_t ptep;
++ int ret;
++ int flags;
++ int i;
++ struct page *page;
++ gntdev_file_private_data_t *private_data = flip->private_data;
++
++ if (unlikely(!private_data)) {
++ printk(KERN_ERR "File's private data is NULL.\n");
++ return -EINVAL;
++ }
++
++ /* Test to make sure that the grants array has been initialised. */
++ down_read(&private_data->grants_sem);
++ if (unlikely(!private_data->grants)) {
++ up_read(&private_data->grants_sem);
++ printk(KERN_ERR "Attempted to mmap before ioctl.\n");
++ return -EINVAL;
++ }
++ up_read(&private_data->grants_sem);
++
++ if (unlikely((size <= 0) ||
++ (size + slot_index) > private_data->grants_size)) {
++ printk(KERN_ERR "Invalid number of pages or offset"
++ "(num_pages = %d, first_slot = %ld).\n",
++ size, slot_index);
++ return -ENXIO;
++ }
++
++ if ((vma->vm_flags & VM_WRITE) && !(vma->vm_flags & VM_SHARED)) {
++ printk(KERN_ERR "Writable mappings must be shared.\n");
++ return -EINVAL;
++ }
++
++ /* Slots must be in the NOT_YET_MAPPED state. */
++ down_write(&private_data->grants_sem);
++ for (i = 0; i < size; ++i) {
++ if (private_data->grants[slot_index + i].state !=
++ GNTDEV_SLOT_NOT_YET_MAPPED) {
++ printk(KERN_ERR "Slot (index = %ld) is in the wrong "
++ "state (%d).\n", slot_index + i,
++ private_data->grants[slot_index + i].state);
++ up_write(&private_data->grants_sem);
++ return -EINVAL;
++ }
++ }
++
++ /* Install the hook for unmapping. */
++ vma->vm_ops = &gntdev_vmops;
++
++ /* The VM area contains pages from another VM. */
++ vma->vm_flags |= VM_FOREIGN;
++ vma->vm_private_data = kzalloc(size * sizeof(struct page *),
++ GFP_KERNEL);
++ if (vma->vm_private_data == NULL) {
++ printk(KERN_ERR "Couldn't allocate mapping structure for VM "
++ "area.\n");
++ return -ENOMEM;
++ }
++
++ /* This flag prevents Bad PTE errors when the memory is unmapped. */
++ vma->vm_flags |= VM_RESERVED;
++
++ /* This flag prevents this VM area being copied on a fork(). A better
++ * behaviour might be to explicitly carry out the appropriate mappings
++ * on fork(), but I don't know if there's a hook for this.
++ */
++ vma->vm_flags |= VM_DONTCOPY;
++
++#ifdef CONFIG_X86
++ /* This flag ensures that the page tables are not unpinned before the
++ * VM area is unmapped. Therefore Xen still recognises the PTE as
++ * belonging to an L1 pagetable, and the grant unmap operation will
++ * succeed, even if the process does not exit cleanly.
++ */
++ vma->vm_mm->context.has_foreign_mappings = 1;
++#endif
++
++ for (i = 0; i < size; ++i) {
++
++ flags = GNTMAP_host_map;
++ if (!(vma->vm_flags & VM_WRITE))
++ flags |= GNTMAP_readonly;
++
++ kernel_vaddr = get_kernel_vaddr(private_data, slot_index + i);
++ user_vaddr = get_user_vaddr(vma, i);
++ page = pfn_to_page(__pa(kernel_vaddr) >> PAGE_SHIFT);
++
++ gnttab_set_map_op(&op, kernel_vaddr, flags,
++ private_data->grants[slot_index+i]
++ .u.valid.ref,
++ private_data->grants[slot_index+i]
++ .u.valid.domid);
++
++ /* Carry out the mapping of the grant reference. */
++ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
++ &op, 1);
++ BUG_ON(ret);
++ if (op.status) {
++ printk(KERN_ERR "Error mapping the grant reference "
++ "into the kernel (%d). domid = %d; ref = %d\n",
++ op.status,
++ private_data->grants[slot_index+i]
++ .u.valid.domid,
++ private_data->grants[slot_index+i]
++ .u.valid.ref);
++ goto undo_map_out;
++ }
++
++ /* Store a reference to the page that will be mapped into user
++ * space.
++ */
++ ((struct page **) vma->vm_private_data)[i] = page;
++
++ /* Mark mapped page as reserved. */
++ SetPageReserved(page);
++
++ /* Record the grant handle, for use in the unmap operation. */
++ private_data->grants[slot_index+i].u.valid.kernel_handle =
++ op.handle;
++ private_data->grants[slot_index+i].u.valid.dev_bus_addr =
++ op.dev_bus_addr;
++
++ private_data->grants[slot_index+i].state = GNTDEV_SLOT_MAPPED;
++ private_data->grants[slot_index+i].u.valid.user_handle =
++ GNTDEV_INVALID_HANDLE;
++
++ /* Now perform the mapping to user space. */
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++
++ /* NOT USING SHADOW PAGE TABLES. */
++ /* In this case, we map the grant(s) straight into user
++ * space.
++ */
++
++ /* Get the machine address of the PTE for the user
++ * page.
++ */
++ if ((ret = create_lookup_pte_addr(vma->vm_mm,
++ vma->vm_start
++ + (i << PAGE_SHIFT),
++ &ptep)))
++ {
++ printk(KERN_ERR "Error obtaining PTE pointer "
++ "(%d).\n", ret);
++ goto undo_map_out;
++ }
++
++ /* Configure the map operation. */
++
++ /* The reference is to be used by host CPUs. */
++ flags = GNTMAP_host_map;
++
++ /* Specifies a user space mapping. */
++ flags |= GNTMAP_application_map;
++
++ /* The map request contains the machine address of the
++ * PTE to update.
++ */
++ flags |= GNTMAP_contains_pte;
++
++ if (!(vma->vm_flags & VM_WRITE))
++ flags |= GNTMAP_readonly;
++
++ gnttab_set_map_op(&op, ptep, flags,
++ private_data->grants[slot_index+i]
++ .u.valid.ref,
++ private_data->grants[slot_index+i]
++ .u.valid.domid);
++
++ /* Carry out the mapping of the grant reference. */
++ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
++ &op, 1);
++ BUG_ON(ret);
++ if (op.status) {
++ printk(KERN_ERR "Error mapping the grant "
++ "reference into user space (%d). domid "
++ "= %d; ref = %d\n", op.status,
++ private_data->grants[slot_index+i].u
++ .valid.domid,
++ private_data->grants[slot_index+i].u
++ .valid.ref);
++ goto undo_map_out;
++ }
++
++ /* Record the grant handle, for use in the unmap
++ * operation.
++ */
++ private_data->grants[slot_index+i].u.
++ valid.user_handle = op.handle;
++
++ /* Update p2m structure with the new mapping. */
++ set_phys_to_machine(__pa(kernel_vaddr) >> PAGE_SHIFT,
++ FOREIGN_FRAME(private_data->
++ grants[slot_index+i]
++ .u.valid.dev_bus_addr
++ >> PAGE_SHIFT));
++ } else {
++ /* USING SHADOW PAGE TABLES. */
++ /* In this case, we simply insert the page into the VM
++ * area. */
++ ret = vm_insert_page(vma, user_vaddr, page);
++ }
++
++ }
++
++ up_write(&private_data->grants_sem);
++ return 0;
++
++undo_map_out:
++ /* If we have a mapping failure, the unmapping will be taken care of
++ * by do_mmap_pgoff(), which will eventually call gntdev_clear_pte().
++ * All we need to do here is free the vma_private_data.
++ */
++ kfree(vma->vm_private_data);
++
++ /* THIS IS VERY UNPLEASANT: do_mmap_pgoff() will set the vma->vm_file
++ * to NULL on failure. However, we need this in gntdev_clear_pte() to
++ * unmap the grants. Therefore, we smuggle a reference to the file's
++ * private data in the VM area's private data pointer.
++ */
++ vma->vm_private_data = private_data;
++
++ up_write(&private_data->grants_sem);
++
++ return -ENOMEM;
++}
++
++static pte_t gntdev_clear_pte(struct vm_area_struct *vma, unsigned long addr,
++ pte_t *ptep, int is_fullmm)
++{
++ int slot_index, ret;
++ pte_t copy;
++ struct gnttab_unmap_grant_ref op;
++ gntdev_file_private_data_t *private_data;
++
++ /* THIS IS VERY UNPLEASANT: do_mmap_pgoff() will set the vma->vm_file
++ * to NULL on failure. However, we need this in gntdev_clear_pte() to
++ * unmap the grants. Therefore, we smuggle a reference to the file's
++ * private data in the VM area's private data pointer.
++ */
++ if (vma->vm_file) {
++ private_data = (gntdev_file_private_data_t *)
++ vma->vm_file->private_data;
++ } else if (vma->vm_private_data) {
++ private_data = (gntdev_file_private_data_t *)
++ vma->vm_private_data;
++ } else {
++ private_data = NULL; /* gcc warning */
++ BUG();
++ }
++
++ /* Copy the existing value of the PTE for returning. */
++ copy = *ptep;
++
++ /* Calculate the grant relating to this PTE. */
++ slot_index = vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT);
++
++ /* Only unmap grants if the slot has been mapped. This could be being
++ * called from a failing mmap().
++ */
++ if (private_data->grants[slot_index].state == GNTDEV_SLOT_MAPPED) {
++
++ /* First, we clear the user space mapping, if it has been made.
++ */
++ if (private_data->grants[slot_index].u.valid.user_handle !=
++ GNTDEV_INVALID_HANDLE &&
++ !xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* NOT USING SHADOW PAGE TABLES. */
++ gnttab_set_unmap_op(&op, virt_to_machine(ptep),
++ GNTMAP_contains_pte,
++ private_data->grants[slot_index]
++ .u.valid.user_handle);
++ ret = HYPERVISOR_grant_table_op(
++ GNTTABOP_unmap_grant_ref, &op, 1);
++ BUG_ON(ret);
++ if (op.status)
++ printk("User unmap grant status = %d\n",
++ op.status);
++ } else {
++ /* USING SHADOW PAGE TABLES. */
++ pte_clear_full(vma->vm_mm, addr, ptep, is_fullmm);
++ }
++
++ /* Finally, we unmap the grant from kernel space. */
++ gnttab_set_unmap_op(&op,
++ get_kernel_vaddr(private_data, slot_index),
++ GNTMAP_host_map,
++ private_data->grants[slot_index].u.valid
++ .kernel_handle);
++ ret = HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref,
++ &op, 1);
++ BUG_ON(ret);
++ if (op.status)
++ printk("Kernel unmap grant status = %d\n", op.status);
++
++
++ /* Return slot to the not-yet-mapped state, so that it may be
++ * mapped again, or removed by a subsequent ioctl.
++ */
++ private_data->grants[slot_index].state =
++ GNTDEV_SLOT_NOT_YET_MAPPED;
++
++ /* Invalidate the physical to machine mapping for this page. */
++ set_phys_to_machine(__pa(get_kernel_vaddr(private_data,
++ slot_index))
++ >> PAGE_SHIFT, INVALID_P2M_ENTRY);
++
++ } else {
++ pte_clear_full(vma->vm_mm, addr, ptep, is_fullmm);
++ }
++
++ return copy;
++}
++
++/* "Destructor" for a VM area.
++ */
++static void gntdev_vma_close(struct vm_area_struct *vma) {
++ if (vma->vm_private_data) {
++ kfree(vma->vm_private_data);
++ }
++}
++
++/* Called when an ioctl is made on the device.
++ */
++static long gntdev_ioctl(struct file *flip,
++ unsigned int cmd, unsigned long arg)
++{
++ int rc = 0;
++ gntdev_file_private_data_t *private_data =
++ (gntdev_file_private_data_t *) flip->private_data;
++
++ /* On the first invocation, we will lazily initialise the grant array
++ * and free-list.
++ */
++ if (unlikely(!private_data->grants)
++ && likely(cmd != IOCTL_GNTDEV_SET_MAX_GRANTS)) {
++ down_write(&private_data->grants_sem);
++
++ if (unlikely(private_data->grants)) {
++ up_write(&private_data->grants_sem);
++ goto private_data_initialised;
++ }
++
++ /* Just use the default. Setting to a non-default is handled
++ * in the ioctl switch.
++ */
++ rc = init_private_data(private_data, DEFAULT_MAX_GRANTS);
++
++ up_write(&private_data->grants_sem);
++
++ if (rc) {
++ printk (KERN_ERR "Initialising gntdev private data "
++ "failed.\n");
++ return rc;
++ }
++ }
++
++private_data_initialised:
++ switch (cmd) {
++ case IOCTL_GNTDEV_MAP_GRANT_REF:
++ {
++ struct ioctl_gntdev_map_grant_ref op;
++ down_write(&private_data->grants_sem);
++ down_write(&private_data->free_list_sem);
++
++ if ((rc = copy_from_user(&op, (void __user *) arg,
++ sizeof(op)))) {
++ rc = -EFAULT;
++ goto map_out;
++ }
++ if (unlikely(op.count <= 0)) {
++ rc = -EINVAL;
++ goto map_out;
++ }
++
++ if (op.count == 1) {
++ if ((rc = add_grant_reference(flip, &op.refs[0],
++ &op.index)) < 0) {
++ printk(KERN_ERR "Adding grant reference "
++ "failed (%d).\n", rc);
++ goto map_out;
++ }
++ } else {
++ struct ioctl_gntdev_grant_ref *refs, *u;
++ refs = kmalloc(op.count * sizeof(*refs), GFP_KERNEL);
++ if (!refs) {
++ rc = -ENOMEM;
++ goto map_out;
++ }
++ u = ((struct ioctl_gntdev_map_grant_ref *)arg)->refs;
++ if ((rc = copy_from_user(refs,
++ (void __user *)u,
++ sizeof(*refs) * op.count))) {
++ printk(KERN_ERR "Copying refs from user failed"
++ " (%d).\n", rc);
++ rc = -EINVAL;
++ goto map_out;
++ }
++ if ((rc = find_contiguous_free_range(flip, op.count))
++ < 0) {
++ printk(KERN_ERR "Finding contiguous range "
++ "failed (%d).\n", rc);
++ kfree(refs);
++ goto map_out;
++ }
++ op.index = rc << PAGE_SHIFT;
++ if ((rc = add_grant_references(flip, op.count,
++ refs, rc))) {
++ printk(KERN_ERR "Adding grant references "
++ "failed (%d).\n", rc);
++ kfree(refs);
++ goto map_out;
++ }
++ compress_free_list(flip);
++ kfree(refs);
++ }
++ if ((rc = copy_to_user((void __user *) arg,
++ &op,
++ sizeof(op)))) {
++ printk(KERN_ERR "Copying result back to user failed "
++ "(%d)\n", rc);
++ rc = -EFAULT;
++ goto map_out;
++ }
++ map_out:
++ up_write(&private_data->grants_sem);
++ up_write(&private_data->free_list_sem);
++ return rc;
++ }
++ case IOCTL_GNTDEV_UNMAP_GRANT_REF:
++ {
++ struct ioctl_gntdev_unmap_grant_ref op;
++ int i, start_index;
++
++ down_write(&private_data->grants_sem);
++ down_write(&private_data->free_list_sem);
++
++ if ((rc = copy_from_user(&op,
++ (void __user *) arg,
++ sizeof(op)))) {
++ rc = -EFAULT;
++ goto unmap_out;
++ }
++
++ start_index = op.index >> PAGE_SHIFT;
++
++ /* First, check that all pages are in the NOT_YET_MAPPED
++ * state.
++ */
++ for (i = 0; i < op.count; ++i) {
++ if (unlikely
++ (private_data->grants[start_index + i].state
++ != GNTDEV_SLOT_NOT_YET_MAPPED)) {
++ if (private_data->grants[start_index + i].state
++ == GNTDEV_SLOT_INVALID) {
++ printk(KERN_ERR
++ "Tried to remove an invalid "
++ "grant at offset 0x%x.",
++ (start_index + i)
++ << PAGE_SHIFT);
++ rc = -EINVAL;
++ } else {
++ printk(KERN_ERR
++ "Tried to remove a grant which "
++ "is currently mmap()-ed at "
++ "offset 0x%x.",
++ (start_index + i)
++ << PAGE_SHIFT);
++ rc = -EBUSY;
++ }
++ goto unmap_out;
++ }
++ }
++
++ /* Unmap pages and add them to the free list.
++ */
++ for (i = 0; i < op.count; ++i) {
++ private_data->grants[start_index+i].state =
++ GNTDEV_SLOT_INVALID;
++ private_data->grants[start_index+i].u.free_list_index =
++ private_data->free_list_size;
++ private_data->free_list[private_data->free_list_size] =
++ start_index + i;
++ ++private_data->free_list_size;
++ }
++
++ unmap_out:
++ up_write(&private_data->grants_sem);
++ up_write(&private_data->free_list_sem);
++ return rc;
++ }
++ case IOCTL_GNTDEV_GET_OFFSET_FOR_VADDR:
++ {
++ struct ioctl_gntdev_get_offset_for_vaddr op;
++ struct vm_area_struct *vma;
++ unsigned long vaddr;
++
++ if ((rc = copy_from_user(&op,
++ (void __user *) arg,
++ sizeof(op)))) {
++ rc = -EFAULT;
++ goto get_offset_out;
++ }
++ vaddr = (unsigned long)op.vaddr;
++
++ down_read(¤t->mm->mmap_sem);
++ vma = find_vma(current->mm, vaddr);
++ if (vma == NULL) {
++ rc = -EFAULT;
++ goto get_offset_unlock_out;
++ }
++ if ((!vma->vm_ops) || (vma->vm_ops != &gntdev_vmops)) {
++ printk(KERN_ERR "The vaddr specified does not belong "
++ "to a gntdev instance: %#lx\n", vaddr);
++ rc = -EFAULT;
++ goto get_offset_unlock_out;
++ }
++ if (vma->vm_start != vaddr) {
++ printk(KERN_ERR "The vaddr specified in an "
++ "IOCTL_GNTDEV_GET_OFFSET_FOR_VADDR must be at "
++ "the start of the VM area. vma->vm_start = "
++ "%#lx; vaddr = %#lx\n",
++ vma->vm_start, vaddr);
++ rc = -EFAULT;
++ goto get_offset_unlock_out;
++ }
++ op.offset = vma->vm_pgoff << PAGE_SHIFT;
++ op.count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
++ up_read(¤t->mm->mmap_sem);
++ if ((rc = copy_to_user((void __user *) arg,
++ &op,
++ sizeof(op)))) {
++ rc = -EFAULT;
++ goto get_offset_out;
++ }
++ goto get_offset_out;
++ get_offset_unlock_out:
++ up_read(¤t->mm->mmap_sem);
++ get_offset_out:
++ return rc;
++ }
++ case IOCTL_GNTDEV_SET_MAX_GRANTS:
++ {
++ struct ioctl_gntdev_set_max_grants op;
++ if ((rc = copy_from_user(&op,
++ (void __user *) arg,
++ sizeof(op)))) {
++ rc = -EFAULT;
++ goto set_max_out;
++ }
++ down_write(&private_data->grants_sem);
++ if (private_data->grants) {
++ rc = -EBUSY;
++ goto set_max_unlock_out;
++ }
++ if (op.count > MAX_GRANTS_LIMIT) {
++ rc = -EINVAL;
++ goto set_max_unlock_out;
++ }
++ rc = init_private_data(private_data, op.count);
++ set_max_unlock_out:
++ up_write(&private_data->grants_sem);
++ set_max_out:
++ return rc;
++ }
++ default:
++ return -ENOIOCTLCMD;
++ }
++
++ return 0;
++}
+--- linux-2.6.18.8/drivers/xen/netback/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/Makefile 2008-05-19 00:33:45.838788321 +0300
+@@ -0,0 +1,5 @@
++obj-$(CONFIG_XEN_NETDEV_BACKEND) := netbk.o
++obj-$(CONFIG_XEN_NETDEV_LOOPBACK) += netloop.o
++
++netbk-y := netback.o xenbus.o interface.o accel.o
++netloop-y := loopback.o
+--- linux-2.6.18.8/drivers/xen/netback/accel.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/accel.c 2008-05-19 00:33:45.838788321 +0300
+@@ -0,0 +1,269 @@
++/******************************************************************************
++ * drivers/xen/netback/accel.c
++ *
++ * Interface between backend virtual network device and accelerated plugin.
++ *
++ * Copyright (C) 2007 Solarflare Communications, Inc
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/list.h>
++#include <asm/atomic.h>
++#include <xen/xenbus.h>
++#include <linux/mutex.h>
++
++#include "common.h"
++
++#if 0
++#undef DPRINTK
++#define DPRINTK(fmt, args...) \
++ printk("netback/accel (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
++#endif
++
++/*
++ * A list of available netback accelerator plugin modules (each list
++ * entry is of type struct netback_accelerator)
++ */
++static struct list_head accelerators_list;
++/* Lock used to protect access to accelerators_list */
++DEFINE_MUTEX(accelerators_mutex);
++
++/*
++ * Compare a backend to an accelerator, and decide if they are
++ * compatible (i.e. if the accelerator should be used by the
++ * backend)
++ */
++static int match_accelerator(struct xenbus_device *xendev,
++ struct backend_info *be,
++ struct netback_accelerator *accelerator)
++{
++ int rc = 0;
++ char *eth_name = xenbus_read(XBT_NIL, xendev->nodename, "accel", NULL);
++
++ if (IS_ERR(eth_name)) {
++ /* Probably means not present */
++ DPRINTK("%s: no match due to xenbus_read accel error %d\n",
++ __FUNCTION__, PTR_ERR(eth_name));
++ return 0;
++ } else {
++ if (!strcmp(eth_name, accelerator->eth_name))
++ rc = 1;
++ kfree(eth_name);
++ return rc;
++ }
++}
++
++
++static void do_probe(struct backend_info *be,
++ struct netback_accelerator *accelerator,
++ struct xenbus_device *xendev)
++{
++ be->accelerator = accelerator;
++ atomic_inc(&be->accelerator->use_count);
++ if (be->accelerator->hooks->probe(xendev) != 0) {
++ atomic_dec(&be->accelerator->use_count);
++ module_put(be->accelerator->hooks->owner);
++ be->accelerator = NULL;
++ }
++}
++
++
++/*
++ * Notify suitable backends that a new accelerator is available and
++ * connected. This will also notify the accelerator plugin module
++ * that it is being used for a device through the probe hook.
++ */
++static int netback_accelerator_probe_backend(struct device *dev, void *arg)
++{
++ struct netback_accelerator *accelerator =
++ (struct netback_accelerator *)arg;
++ struct xenbus_device *xendev = to_xenbus_device(dev);
++
++ if (!strcmp("vif", xendev->devicetype)) {
++ struct backend_info *be = xendev->dev.driver_data;
++
++ if (match_accelerator(xendev, be, accelerator) &&
++ try_module_get(accelerator->hooks->owner)) {
++ do_probe(be, accelerator, xendev);
++ }
++ }
++ return 0;
++}
++
++
++/*
++ * Notify suitable backends that an accelerator is unavailable.
++ */
++static int netback_accelerator_remove_backend(struct device *dev, void *arg)
++{
++ struct xenbus_device *xendev = to_xenbus_device(dev);
++ struct netback_accelerator *accelerator =
++ (struct netback_accelerator *)arg;
++
++ if (!strcmp("vif", xendev->devicetype)) {
++ struct backend_info *be = xendev->dev.driver_data;
++
++ if (be->accelerator == accelerator) {
++ be->accelerator->hooks->remove(xendev);
++ atomic_dec(&be->accelerator->use_count);
++ module_put(be->accelerator->hooks->owner);
++ be->accelerator = NULL;
++ }
++ }
++ return 0;
++}
++
++
++
++/*
++ * Entry point for an netback accelerator plugin module. Called to
++ * advertise its presence, and connect to any suitable backends.
++ */
++int netback_connect_accelerator(unsigned version, int id, const char *eth_name,
++ struct netback_accel_hooks *hooks)
++{
++ struct netback_accelerator *new_accelerator;
++ unsigned eth_name_len;
++
++ if (version != NETBACK_ACCEL_VERSION) {
++ if (version > NETBACK_ACCEL_VERSION) {
++ /* Caller has higher version number, leave it
++ up to them to decide whether to continue.
++ They can recall with a lower number if
++ they're happy to be compatible with us */
++ return NETBACK_ACCEL_VERSION;
++ } else {
++ /* We have a more recent version than caller.
++ Currently reject, but may in future be able
++ to be backwardly compatible */
++ return -EPROTO;
++ }
++ }
++
++ new_accelerator =
++ kmalloc(sizeof(struct netback_accelerator), GFP_KERNEL);
++ if (!new_accelerator) {
++ DPRINTK("%s: failed to allocate memory for accelerator\n",
++ __FUNCTION__);
++ return -ENOMEM;
++ }
++
++ new_accelerator->id = id;
++
++ eth_name_len = strlen(eth_name)+1;
++ new_accelerator->eth_name = kmalloc(eth_name_len, GFP_KERNEL);
++ if (!new_accelerator->eth_name) {
++ DPRINTK("%s: failed to allocate memory for eth_name string\n",
++ __FUNCTION__);
++ kfree(new_accelerator);
++ return -ENOMEM;
++ }
++ strlcpy(new_accelerator->eth_name, eth_name, eth_name_len);
++
++ new_accelerator->hooks = hooks;
++
++ atomic_set(&new_accelerator->use_count, 0);
++
++ mutex_lock(&accelerators_mutex);
++ list_add(&new_accelerator->link, &accelerators_list);
++
++ /* tell existing backends about new plugin */
++ xenbus_for_each_backend(new_accelerator,
++ netback_accelerator_probe_backend);
++
++ mutex_unlock(&accelerators_mutex);
++
++ return 0;
++
++}
++EXPORT_SYMBOL_GPL(netback_connect_accelerator);
++
++
++/*
++ * Disconnect an accelerator plugin module that has previously been
++ * connected.
++ */
++void netback_disconnect_accelerator(int id, const char *eth_name)
++{
++ struct netback_accelerator *accelerator, *next;
++
++ mutex_lock(&accelerators_mutex);
++ list_for_each_entry_safe(accelerator, next, &accelerators_list, link) {
++ if (!strcmp(eth_name, accelerator->eth_name)) {
++ xenbus_for_each_backend
++ (accelerator, netback_accelerator_remove_backend);
++ BUG_ON(atomic_read(&accelerator->use_count) != 0);
++ list_del(&accelerator->link);
++ kfree(accelerator->eth_name);
++ kfree(accelerator);
++ break;
++ }
++ }
++ mutex_unlock(&accelerators_mutex);
++}
++EXPORT_SYMBOL_GPL(netback_disconnect_accelerator);
++
++
++void netback_probe_accelerators(struct backend_info *be,
++ struct xenbus_device *dev)
++{
++ struct netback_accelerator *accelerator;
++
++ /*
++ * Check list of accelerators to see if any is suitable, and
++ * use it if it is.
++ */
++ mutex_lock(&accelerators_mutex);
++ list_for_each_entry(accelerator, &accelerators_list, link) {
++ if (match_accelerator(dev, be, accelerator) &&
++ try_module_get(accelerator->hooks->owner)) {
++ do_probe(be, accelerator, dev);
++ break;
++ }
++ }
++ mutex_unlock(&accelerators_mutex);
++}
++
++
++void netback_remove_accelerators(struct backend_info *be,
++ struct xenbus_device *dev)
++{
++ mutex_lock(&accelerators_mutex);
++ /* Notify the accelerator (if any) of this device's removal */
++ if (be->accelerator != NULL) {
++ be->accelerator->hooks->remove(dev);
++ atomic_dec(&be->accelerator->use_count);
++ module_put(be->accelerator->hooks->owner);
++ be->accelerator = NULL;
++ }
++ mutex_unlock(&accelerators_mutex);
++}
++
++
++void netif_accel_init(void)
++{
++ INIT_LIST_HEAD(&accelerators_list);
++}
+--- linux-2.6.18.8/drivers/xen/netback/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/common.h 2008-05-19 00:33:45.838788321 +0300
+@@ -0,0 +1,217 @@
++/******************************************************************************
++ * arch/xen/drivers/netif/backend/common.h
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef __NETIF__BACKEND__COMMON_H__
++#define __NETIF__BACKEND__COMMON_H__
++
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/interrupt.h>
++#include <linux/slab.h>
++#include <linux/ip.h>
++#include <linux/in.h>
++#include <linux/netdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/wait.h>
++#include <xen/evtchn.h>
++#include <xen/interface/io/netif.h>
++#include <asm/io.h>
++#include <asm/pgalloc.h>
++#include <xen/interface/grant_table.h>
++#include <xen/gnttab.h>
++#include <xen/driver_util.h>
++#include <xen/xenbus.h>
++
++#define DPRINTK(_f, _a...) \
++ pr_debug("(file=%s, line=%d) " _f, \
++ __FILE__ , __LINE__ , ## _a )
++#define IPRINTK(fmt, args...) \
++ printk(KERN_INFO "xen_net: " fmt, ##args)
++#define WPRINTK(fmt, args...) \
++ printk(KERN_WARNING "xen_net: " fmt, ##args)
++
++typedef struct netif_st {
++ /* Unique identifier for this interface. */
++ domid_t domid;
++ unsigned int handle;
++
++ u8 fe_dev_addr[6];
++
++ /* Physical parameters of the comms window. */
++ grant_handle_t tx_shmem_handle;
++ grant_ref_t tx_shmem_ref;
++ grant_handle_t rx_shmem_handle;
++ grant_ref_t rx_shmem_ref;
++ unsigned int irq;
++
++ /* The shared rings and indexes. */
++ netif_tx_back_ring_t tx;
++ netif_rx_back_ring_t rx;
++ struct vm_struct *tx_comms_area;
++ struct vm_struct *rx_comms_area;
++
++ /* Set of features that can be turned on in dev->features. */
++ int features;
++
++ /* Internal feature information. */
++ u8 can_queue:1; /* can queue packets for receiver? */
++ u8 copying_receiver:1; /* copy packets to receiver? */
++
++ /* Allow netif_be_start_xmit() to peek ahead in the rx request ring. */
++ RING_IDX rx_req_cons_peek;
++
++ /* Transmit shaping: allow 'credit_bytes' every 'credit_usec'. */
++ unsigned long credit_bytes;
++ unsigned long credit_usec;
++ unsigned long remaining_credit;
++ struct timer_list credit_timeout;
++
++ /* Enforce draining of the transmit queue. */
++ struct timer_list tx_queue_timeout;
++
++ /* Miscellaneous private stuff. */
++ struct list_head list; /* scheduling list */
++ atomic_t refcnt;
++ struct net_device *dev;
++ struct net_device_stats stats;
++
++ unsigned int carrier;
++
++ wait_queue_head_t waiting_to_free;
++} netif_t;
++
++/*
++ * Implement our own carrier flag: the network stack's version causes delays
++ * when the carrier is re-enabled (in particular, dev_activate() may not
++ * immediately be called, which can cause packet loss; also the etherbridge
++ * can be rather lazy in activating its port).
++ */
++#define netback_carrier_on(netif) ((netif)->carrier = 1)
++#define netback_carrier_off(netif) ((netif)->carrier = 0)
++#define netback_carrier_ok(netif) ((netif)->carrier)
++
++enum {
++ NETBK_DONT_COPY_SKB,
++ NETBK_DELAYED_COPY_SKB,
++ NETBK_ALWAYS_COPY_SKB,
++};
++
++extern int netbk_copy_skb_mode;
++
++/* Function pointers into netback accelerator plugin modules */
++struct netback_accel_hooks {
++ struct module *owner;
++ int (*probe)(struct xenbus_device *dev);
++ int (*remove)(struct xenbus_device *dev);
++};
++
++/* Structure to track the state of a netback accelerator plugin */
++struct netback_accelerator {
++ struct list_head link;
++ int id;
++ char *eth_name;
++ atomic_t use_count;
++ struct netback_accel_hooks *hooks;
++};
++
++struct backend_info {
++ struct xenbus_device *dev;
++ netif_t *netif;
++ enum xenbus_state frontend_state;
++
++ /* State relating to the netback accelerator */
++ void *netback_accel_priv;
++ /* The accelerator that this backend is currently using */
++ struct netback_accelerator *accelerator;
++};
++
++#define NETBACK_ACCEL_VERSION 0x00010001
++
++/*
++ * Connect an accelerator plugin module to netback. Returns zero on
++ * success, < 0 on error, > 0 (with highest version number supported)
++ * if version mismatch.
++ */
++extern int netback_connect_accelerator(unsigned version,
++ int id, const char *eth_name,
++ struct netback_accel_hooks *hooks);
++/* Disconnect a previously connected accelerator plugin module */
++extern void netback_disconnect_accelerator(int id, const char *eth_name);
++
++
++extern
++void netback_probe_accelerators(struct backend_info *be,
++ struct xenbus_device *dev);
++extern
++void netback_remove_accelerators(struct backend_info *be,
++ struct xenbus_device *dev);
++extern
++void netif_accel_init(void);
++
++
++#define NET_TX_RING_SIZE __RING_SIZE((netif_tx_sring_t *)0, PAGE_SIZE)
++#define NET_RX_RING_SIZE __RING_SIZE((netif_rx_sring_t *)0, PAGE_SIZE)
++
++void netif_disconnect(netif_t *netif);
++
++netif_t *netif_alloc(domid_t domid, unsigned int handle);
++int netif_map(netif_t *netif, unsigned long tx_ring_ref,
++ unsigned long rx_ring_ref, unsigned int evtchn);
++
++#define netif_get(_b) (atomic_inc(&(_b)->refcnt))
++#define netif_put(_b) \
++ do { \
++ if ( atomic_dec_and_test(&(_b)->refcnt) ) \
++ wake_up(&(_b)->waiting_to_free); \
++ } while (0)
++
++void netif_xenbus_init(void);
++
++#define netif_schedulable(netif) \
++ (netif_running((netif)->dev) && netback_carrier_ok(netif))
++
++void netif_schedule_work(netif_t *netif);
++void netif_deschedule_work(netif_t *netif);
++
++int netif_be_start_xmit(struct sk_buff *skb, struct net_device *dev);
++struct net_device_stats *netif_be_get_stats(struct net_device *dev);
++irqreturn_t netif_be_int(int irq, void *dev_id, struct pt_regs *regs);
++
++static inline int netbk_can_queue(struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++ return netif->can_queue;
++}
++
++static inline int netbk_can_sg(struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++ return netif->features & NETIF_F_SG;
++}
++
++#endif /* __NETIF__BACKEND__COMMON_H__ */
+--- linux-2.6.18.8/drivers/xen/netback/interface.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/interface.c 2008-05-19 00:33:45.842788551 +0300
+@@ -0,0 +1,336 @@
++/******************************************************************************
++ * arch/xen/drivers/netif/backend/interface.c
++ *
++ * Network-device interface management.
++ *
++ * Copyright (c) 2004-2005, Keir Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include "common.h"
++#include <linux/ethtool.h>
++#include <linux/rtnetlink.h>
++
++/*
++ * Module parameter 'queue_length':
++ *
++ * Enables queuing in the network stack when a client has run out of receive
++ * descriptors. Although this feature can improve receive bandwidth by avoiding
++ * packet loss, it can also result in packets sitting in the 'tx_queue' for
++ * unbounded time. This is bad if those packets hold onto foreign resources.
++ * For example, consider a packet that holds onto resources belonging to the
++ * guest for which it is queued (e.g., packet received on vif1.0, destined for
++ * vif1.1 which is not activated in the guest): in this situation the guest
++ * will never be destroyed, unless vif1.1 is taken down. To avoid this, we
++ * run a timer (tx_queue_timeout) to drain the queue when the interface is
++ * blocked.
++ */
++static unsigned long netbk_queue_length = 32;
++module_param_named(queue_length, netbk_queue_length, ulong, 0);
++
++static void __netif_up(netif_t *netif)
++{
++ enable_irq(netif->irq);
++ netif_schedule_work(netif);
++}
++
++static void __netif_down(netif_t *netif)
++{
++ disable_irq(netif->irq);
++ netif_deschedule_work(netif);
++}
++
++static int net_open(struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++ if (netback_carrier_ok(netif)) {
++ __netif_up(netif);
++ netif_start_queue(dev);
++ }
++ return 0;
++}
++
++static int net_close(struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++ if (netback_carrier_ok(netif))
++ __netif_down(netif);
++ netif_stop_queue(dev);
++ return 0;
++}
++
++static int netbk_change_mtu(struct net_device *dev, int mtu)
++{
++ int max = netbk_can_sg(dev) ? 65535 - ETH_HLEN : ETH_DATA_LEN;
++
++ if (mtu > max)
++ return -EINVAL;
++ dev->mtu = mtu;
++ return 0;
++}
++
++static int netbk_set_sg(struct net_device *dev, u32 data)
++{
++ if (data) {
++ netif_t *netif = netdev_priv(dev);
++
++ if (!(netif->features & NETIF_F_SG))
++ return -ENOSYS;
++ }
++
++ return ethtool_op_set_sg(dev, data);
++}
++
++static int netbk_set_tso(struct net_device *dev, u32 data)
++{
++ if (data) {
++ netif_t *netif = netdev_priv(dev);
++
++ if (!(netif->features & NETIF_F_TSO))
++ return -ENOSYS;
++ }
++
++ return ethtool_op_set_tso(dev, data);
++}
++
++static struct ethtool_ops network_ethtool_ops =
++{
++ .get_tx_csum = ethtool_op_get_tx_csum,
++ .set_tx_csum = ethtool_op_set_tx_csum,
++ .get_sg = ethtool_op_get_sg,
++ .set_sg = netbk_set_sg,
++ .get_tso = ethtool_op_get_tso,
++ .set_tso = netbk_set_tso,
++ .get_link = ethtool_op_get_link,
++};
++
++netif_t *netif_alloc(domid_t domid, unsigned int handle)
++{
++ int err = 0;
++ struct net_device *dev;
++ netif_t *netif;
++ char name[IFNAMSIZ] = {};
++
++ snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle);
++ dev = alloc_netdev(sizeof(netif_t), name, ether_setup);
++ if (dev == NULL) {
++ DPRINTK("Could not create netif: out of memory\n");
++ return ERR_PTR(-ENOMEM);
++ }
++
++ netif = netdev_priv(dev);
++ memset(netif, 0, sizeof(*netif));
++ netif->domid = domid;
++ netif->handle = handle;
++ atomic_set(&netif->refcnt, 1);
++ init_waitqueue_head(&netif->waiting_to_free);
++ netif->dev = dev;
++
++ netback_carrier_off(netif);
++
++ netif->credit_bytes = netif->remaining_credit = ~0UL;
++ netif->credit_usec = 0UL;
++ init_timer(&netif->credit_timeout);
++ /* Initialize 'expires' now: it's used to track the credit window. */
++ netif->credit_timeout.expires = jiffies;
++
++ init_timer(&netif->tx_queue_timeout);
++
++ dev->hard_start_xmit = netif_be_start_xmit;
++ dev->get_stats = netif_be_get_stats;
++ dev->open = net_open;
++ dev->stop = net_close;
++ dev->change_mtu = netbk_change_mtu;
++ dev->features = NETIF_F_IP_CSUM;
++
++ SET_ETHTOOL_OPS(dev, &network_ethtool_ops);
++
++ dev->tx_queue_len = netbk_queue_length;
++
++ /*
++ * Initialise a dummy MAC address. We choose the numerically
++ * largest non-broadcast address to prevent the address getting
++ * stolen by an Ethernet bridge for STP purposes.
++ * (FE:FF:FF:FF:FF:FF)
++ */
++ memset(dev->dev_addr, 0xFF, ETH_ALEN);
++ dev->dev_addr[0] &= ~0x01;
++
++ rtnl_lock();
++ err = register_netdevice(dev);
++ rtnl_unlock();
++ if (err) {
++ DPRINTK("Could not register new net device %s: err=%d\n",
++ dev->name, err);
++ free_netdev(dev);
++ return ERR_PTR(err);
++ }
++
++ DPRINTK("Successfully created netif\n");
++ return netif;
++}
++
++static int map_frontend_pages(
++ netif_t *netif, grant_ref_t tx_ring_ref, grant_ref_t rx_ring_ref)
++{
++ struct gnttab_map_grant_ref op;
++
++ gnttab_set_map_op(&op, (unsigned long)netif->tx_comms_area->addr,
++ GNTMAP_host_map, tx_ring_ref, netif->domid);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
++ BUG();
++
++ if (op.status) {
++ DPRINTK(" Gnttab failure mapping tx_ring_ref!\n");
++ return op.status;
++ }
++
++ netif->tx_shmem_ref = tx_ring_ref;
++ netif->tx_shmem_handle = op.handle;
++
++ gnttab_set_map_op(&op, (unsigned long)netif->rx_comms_area->addr,
++ GNTMAP_host_map, rx_ring_ref, netif->domid);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
++ BUG();
++
++ if (op.status) {
++ DPRINTK(" Gnttab failure mapping rx_ring_ref!\n");
++ return op.status;
++ }
++
++ netif->rx_shmem_ref = rx_ring_ref;
++ netif->rx_shmem_handle = op.handle;
++
++ return 0;
++}
++
++static void unmap_frontend_pages(netif_t *netif)
++{
++ struct gnttab_unmap_grant_ref op;
++
++ gnttab_set_unmap_op(&op, (unsigned long)netif->tx_comms_area->addr,
++ GNTMAP_host_map, netif->tx_shmem_handle);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
++ BUG();
++
++ gnttab_set_unmap_op(&op, (unsigned long)netif->rx_comms_area->addr,
++ GNTMAP_host_map, netif->rx_shmem_handle);
++
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
++ BUG();
++}
++
++int netif_map(netif_t *netif, unsigned long tx_ring_ref,
++ unsigned long rx_ring_ref, unsigned int evtchn)
++{
++ int err = -ENOMEM;
++ netif_tx_sring_t *txs;
++ netif_rx_sring_t *rxs;
++
++ /* Already connected through? */
++ if (netif->irq)
++ return 0;
++
++ netif->tx_comms_area = alloc_vm_area(PAGE_SIZE);
++ if (netif->tx_comms_area == NULL)
++ return -ENOMEM;
++ netif->rx_comms_area = alloc_vm_area(PAGE_SIZE);
++ if (netif->rx_comms_area == NULL)
++ goto err_rx;
++
++ err = map_frontend_pages(netif, tx_ring_ref, rx_ring_ref);
++ if (err)
++ goto err_map;
++
++ err = bind_interdomain_evtchn_to_irqhandler(
++ netif->domid, evtchn, netif_be_int, 0,
++ netif->dev->name, netif);
++ if (err < 0)
++ goto err_hypervisor;
++ netif->irq = err;
++ disable_irq(netif->irq);
++
++ txs = (netif_tx_sring_t *)netif->tx_comms_area->addr;
++ BACK_RING_INIT(&netif->tx, txs, PAGE_SIZE);
++
++ rxs = (netif_rx_sring_t *)
++ ((char *)netif->rx_comms_area->addr);
++ BACK_RING_INIT(&netif->rx, rxs, PAGE_SIZE);
++
++ netif->rx_req_cons_peek = 0;
++
++ netif_get(netif);
++
++ rtnl_lock();
++ netback_carrier_on(netif);
++ if (netif_running(netif->dev))
++ __netif_up(netif);
++ rtnl_unlock();
++
++ return 0;
++err_hypervisor:
++ unmap_frontend_pages(netif);
++err_map:
++ free_vm_area(netif->rx_comms_area);
++err_rx:
++ free_vm_area(netif->tx_comms_area);
++ return err;
++}
++
++void netif_disconnect(netif_t *netif)
++{
++ if (netback_carrier_ok(netif)) {
++ rtnl_lock();
++ netback_carrier_off(netif);
++ netif_carrier_off(netif->dev); /* discard queued packets */
++ if (netif_running(netif->dev))
++ __netif_down(netif);
++ rtnl_unlock();
++ netif_put(netif);
++ }
++
++ atomic_dec(&netif->refcnt);
++ wait_event(netif->waiting_to_free, atomic_read(&netif->refcnt) == 0);
++
++ del_timer_sync(&netif->credit_timeout);
++ del_timer_sync(&netif->tx_queue_timeout);
++
++ if (netif->irq)
++ unbind_from_irqhandler(netif->irq, netif);
++
++ unregister_netdev(netif->dev);
++
++ if (netif->tx.sring) {
++ unmap_frontend_pages(netif);
++ free_vm_area(netif->tx_comms_area);
++ free_vm_area(netif->rx_comms_area);
++ }
++
++ free_netdev(netif->dev);
++}
+--- linux-2.6.18.8/drivers/xen/netback/loopback.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/loopback.c 2008-05-19 00:33:45.842788551 +0300
+@@ -0,0 +1,324 @@
++/******************************************************************************
++ * netback/loopback.c
++ *
++ * A two-interface loopback device to emulate a local netfront-netback
++ * connection. This ensures that local packet delivery looks identical
++ * to inter-domain delivery. Most importantly, packets delivered locally
++ * originating from other domains will get *copied* when they traverse this
++ * driver. This prevents unbounded delays in socket-buffer queues from
++ * causing the netback driver to "seize up".
++ *
++ * This driver creates a symmetric pair of loopback interfaces with names
++ * vif0.0 and veth0. The intention is that 'vif0.0' is bound to an Ethernet
++ * bridge, just like a proper netback interface, while a local IP interface
++ * is configured on 'veth0'.
++ *
++ * As with a real netback interface, vif0.0 is configured with a suitable
++ * dummy MAC address. No default is provided for veth0: a reasonable strategy
++ * is to transfer eth0's MAC address to veth0, and give eth0 a dummy address
++ * (to avoid confusing the Etherbridge).
++ *
++ * Copyright (c) 2005 K A Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/module.h>
++#include <linux/netdevice.h>
++#include <linux/inetdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/skbuff.h>
++#include <linux/ethtool.h>
++#include <net/dst.h>
++#include <net/xfrm.h> /* secpath_reset() */
++#include <asm/hypervisor.h> /* is_initial_xendomain() */
++
++static int nloopbacks = -1;
++module_param(nloopbacks, int, 0);
++MODULE_PARM_DESC(nloopbacks, "Number of netback-loopback devices to create");
++
++struct net_private {
++ struct net_device *loopback_dev;
++ struct net_device_stats stats;
++};
++
++static int loopback_open(struct net_device *dev)
++{
++ struct net_private *np = netdev_priv(dev);
++ memset(&np->stats, 0, sizeof(np->stats));
++ netif_start_queue(dev);
++ return 0;
++}
++
++static int loopback_close(struct net_device *dev)
++{
++ netif_stop_queue(dev);
++ return 0;
++}
++
++#ifdef CONFIG_X86
++static int is_foreign(unsigned long pfn)
++{
++ /* NB. Play it safe for auto-translation mode. */
++ return (xen_feature(XENFEAT_auto_translated_physmap) ||
++ (phys_to_machine_mapping[pfn] & FOREIGN_FRAME_BIT));
++}
++#else
++/* How to detect a foreign mapping? Play it safe. */
++#define is_foreign(pfn) (1)
++#endif
++
++static int skb_remove_foreign_references(struct sk_buff *skb)
++{
++ struct page *page;
++ unsigned long pfn;
++ int i, off;
++ char *vaddr;
++
++ BUG_ON(skb_shinfo(skb)->frag_list);
++
++ if (skb_cloned(skb) &&
++ unlikely(pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
++ return 0;
++
++ for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
++ pfn = page_to_pfn(skb_shinfo(skb)->frags[i].page);
++ if (!is_foreign(pfn))
++ continue;
++
++ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
++ if (unlikely(!page))
++ return 0;
++
++ vaddr = kmap_skb_frag(&skb_shinfo(skb)->frags[i]);
++ off = skb_shinfo(skb)->frags[i].page_offset;
++ memcpy(page_address(page) + off,
++ vaddr + off,
++ skb_shinfo(skb)->frags[i].size);
++ kunmap_skb_frag(vaddr);
++
++ put_page(skb_shinfo(skb)->frags[i].page);
++ skb_shinfo(skb)->frags[i].page = page;
++ }
++
++ return 1;
++}
++
++static int loopback_start_xmit(struct sk_buff *skb, struct net_device *dev)
++{
++ struct net_private *np = netdev_priv(dev);
++
++ if (!skb_remove_foreign_references(skb)) {
++ np->stats.tx_dropped++;
++ dev_kfree_skb(skb);
++ return 0;
++ }
++
++ dst_release(skb->dst);
++ skb->dst = NULL;
++
++ skb_orphan(skb);
++
++ np->stats.tx_bytes += skb->len;
++ np->stats.tx_packets++;
++
++ /* Switch to loopback context. */
++ dev = np->loopback_dev;
++ np = netdev_priv(dev);
++
++ np->stats.rx_bytes += skb->len;
++ np->stats.rx_packets++;
++
++ if (skb->ip_summed == CHECKSUM_HW) {
++ /* Defer checksum calculation. */
++ skb->proto_csum_blank = 1;
++ /* Must be a local packet: assert its integrity. */
++ skb->proto_data_valid = 1;
++ }
++
++ skb->ip_summed = skb->proto_data_valid ?
++ CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
++
++ skb->pkt_type = PACKET_HOST; /* overridden by eth_type_trans() */
++ skb->protocol = eth_type_trans(skb, dev);
++ skb->dev = dev;
++ dev->last_rx = jiffies;
++
++ /* Flush netfilter context: rx'ed skbuffs not expected to have any. */
++ nf_reset(skb);
++ secpath_reset(skb);
++
++ netif_rx(skb);
++
++ return 0;
++}
++
++static struct net_device_stats *loopback_get_stats(struct net_device *dev)
++{
++ struct net_private *np = netdev_priv(dev);
++ return &np->stats;
++}
++
++static struct ethtool_ops network_ethtool_ops =
++{
++ .get_tx_csum = ethtool_op_get_tx_csum,
++ .set_tx_csum = ethtool_op_set_tx_csum,
++ .get_sg = ethtool_op_get_sg,
++ .set_sg = ethtool_op_set_sg,
++ .get_tso = ethtool_op_get_tso,
++ .set_tso = ethtool_op_set_tso,
++ .get_link = ethtool_op_get_link,
++};
++
++/*
++ * Nothing to do here. Virtual interface is point-to-point and the
++ * physical interface is probably promiscuous anyway.
++ */
++static void loopback_set_multicast_list(struct net_device *dev)
++{
++}
++
++static void loopback_construct(struct net_device *dev, struct net_device *lo)
++{
++ struct net_private *np = netdev_priv(dev);
++
++ np->loopback_dev = lo;
++
++ dev->open = loopback_open;
++ dev->stop = loopback_close;
++ dev->hard_start_xmit = loopback_start_xmit;
++ dev->get_stats = loopback_get_stats;
++ dev->set_multicast_list = loopback_set_multicast_list;
++ dev->change_mtu = NULL; /* allow arbitrary mtu */
++
++ dev->tx_queue_len = 0;
++
++ dev->features = (NETIF_F_HIGHDMA |
++ NETIF_F_LLTX |
++ NETIF_F_TSO |
++ NETIF_F_SG |
++ NETIF_F_IP_CSUM);
++
++ SET_ETHTOOL_OPS(dev, &network_ethtool_ops);
++
++ /*
++ * We do not set a jumbo MTU on the interface. Otherwise the network
++ * stack will try to send large packets that will get dropped by the
++ * Ethernet bridge (unless the physical Ethernet interface is
++ * configured to transfer jumbo packets). If a larger MTU is desired
++ * then the system administrator can specify it using the 'ifconfig'
++ * command.
++ */
++ /*dev->mtu = 16*1024;*/
++}
++
++static int __init make_loopback(int i)
++{
++ struct net_device *dev1, *dev2;
++ char dev_name[IFNAMSIZ];
++ int err = -ENOMEM;
++
++ sprintf(dev_name, "vif0.%d", i);
++ dev1 = alloc_netdev(sizeof(struct net_private), dev_name, ether_setup);
++ if (!dev1)
++ return err;
++
++ sprintf(dev_name, "veth%d", i);
++ dev2 = alloc_netdev(sizeof(struct net_private), dev_name, ether_setup);
++ if (!dev2)
++ goto fail_netdev2;
++
++ loopback_construct(dev1, dev2);
++ loopback_construct(dev2, dev1);
++
++ /*
++ * Initialise a dummy MAC address for the 'dummy backend' interface. We
++ * choose the numerically largest non-broadcast address to prevent the
++ * address getting stolen by an Ethernet bridge for STP purposes.
++ */
++ memset(dev1->dev_addr, 0xFF, ETH_ALEN);
++ dev1->dev_addr[0] &= ~0x01;
++
++ if ((err = register_netdev(dev1)) != 0)
++ goto fail;
++
++ if ((err = register_netdev(dev2)) != 0) {
++ unregister_netdev(dev1);
++ goto fail;
++ }
++
++ return 0;
++
++ fail:
++ free_netdev(dev2);
++ fail_netdev2:
++ free_netdev(dev1);
++ return err;
++}
++
++static void __exit clean_loopback(int i)
++{
++ struct net_device *dev1, *dev2;
++ char dev_name[IFNAMSIZ];
++
++ sprintf(dev_name, "vif0.%d", i);
++ dev1 = dev_get_by_name(dev_name);
++ sprintf(dev_name, "veth%d", i);
++ dev2 = dev_get_by_name(dev_name);
++ if (dev1 && dev2) {
++ unregister_netdev(dev2);
++ unregister_netdev(dev1);
++ free_netdev(dev2);
++ free_netdev(dev1);
++ }
++}
++
++static int __init loopback_init(void)
++{
++ int i, err = 0;
++
++ if (nloopbacks == -1)
++ nloopbacks = is_initial_xendomain() ? 4 : 0;
++
++ for (i = 0; i < nloopbacks; i++)
++ if ((err = make_loopback(i)) != 0)
++ break;
++
++ return err;
++}
++
++module_init(loopback_init);
++
++static void __exit loopback_exit(void)
++{
++ int i;
++
++ for (i = nloopbacks; i-- > 0; )
++ clean_loopback(i);
++}
++
++module_exit(loopback_exit);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/netback/netback.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/netback.c 2008-05-19 00:33:45.842788551 +0300
+@@ -0,0 +1,1614 @@
++/******************************************************************************
++ * drivers/xen/netback/netback.c
++ *
++ * Back-end of the driver for virtual network devices. This portion of the
++ * driver exports a 'unified' network-device interface that can be accessed
++ * by any operating system that implements a compatible front end. A
++ * reference front-end implementation can be found in:
++ * drivers/xen/netfront/netfront.c
++ *
++ * Copyright (c) 2002-2005, K A Fraser
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include "common.h"
++#include <xen/balloon.h>
++#include <xen/interface/memory.h>
++
++/*define NETBE_DEBUG_INTERRUPT*/
++
++/* extra field used in struct page */
++#define netif_page_index(pg) (*(long *)&(pg)->mapping)
++
++struct netbk_rx_meta {
++ skb_frag_t frag;
++ int id;
++ u8 copy:1;
++};
++
++struct netbk_tx_pending_inuse {
++ struct list_head list;
++ unsigned long alloc_time;
++};
++
++static void netif_idx_release(u16 pending_idx);
++static void netif_page_release(struct page *page);
++static void make_tx_response(netif_t *netif,
++ netif_tx_request_t *txp,
++ s8 st);
++static netif_rx_response_t *make_rx_response(netif_t *netif,
++ u16 id,
++ s8 st,
++ u16 offset,
++ u16 size,
++ u16 flags);
++
++static void net_tx_action(unsigned long unused);
++static DECLARE_TASKLET(net_tx_tasklet, net_tx_action, 0);
++
++static void net_rx_action(unsigned long unused);
++static DECLARE_TASKLET(net_rx_tasklet, net_rx_action, 0);
++
++static struct timer_list net_timer;
++static struct timer_list netbk_tx_pending_timer;
++
++#define MAX_PENDING_REQS 256
++
++static struct sk_buff_head rx_queue;
++
++static struct page **mmap_pages;
++static inline unsigned long idx_to_pfn(unsigned int idx)
++{
++ return page_to_pfn(mmap_pages[idx]);
++}
++
++static inline unsigned long idx_to_kaddr(unsigned int idx)
++{
++ return (unsigned long)pfn_to_kaddr(idx_to_pfn(idx));
++}
++
++#define PKT_PROT_LEN 64
++
++static struct pending_tx_info {
++ netif_tx_request_t req;
++ netif_t *netif;
++} pending_tx_info[MAX_PENDING_REQS];
++static u16 pending_ring[MAX_PENDING_REQS];
++typedef unsigned int PEND_RING_IDX;
++#define MASK_PEND_IDX(_i) ((_i)&(MAX_PENDING_REQS-1))
++static PEND_RING_IDX pending_prod, pending_cons;
++#define NR_PENDING_REQS (MAX_PENDING_REQS - pending_prod + pending_cons)
++
++/* Freed TX SKBs get batched on this ring before return to pending_ring. */
++static u16 dealloc_ring[MAX_PENDING_REQS];
++static PEND_RING_IDX dealloc_prod, dealloc_cons;
++
++/* Doubly-linked list of in-use pending entries. */
++static struct netbk_tx_pending_inuse pending_inuse[MAX_PENDING_REQS];
++static LIST_HEAD(pending_inuse_head);
++
++static struct sk_buff_head tx_queue;
++
++static grant_handle_t grant_tx_handle[MAX_PENDING_REQS];
++static gnttab_unmap_grant_ref_t tx_unmap_ops[MAX_PENDING_REQS];
++static gnttab_map_grant_ref_t tx_map_ops[MAX_PENDING_REQS];
++
++static struct list_head net_schedule_list;
++static spinlock_t net_schedule_list_lock;
++
++#define MAX_MFN_ALLOC 64
++static unsigned long mfn_list[MAX_MFN_ALLOC];
++static unsigned int alloc_index = 0;
++
++/* Setting this allows the safe use of this driver without netloop. */
++static int MODPARM_copy_skb = 1;
++module_param_named(copy_skb, MODPARM_copy_skb, bool, 0);
++MODULE_PARM_DESC(copy_skb, "Copy data received from netfront without netloop");
++
++int netbk_copy_skb_mode;
++
++static inline unsigned long alloc_mfn(void)
++{
++ BUG_ON(alloc_index == 0);
++ return mfn_list[--alloc_index];
++}
++
++static int check_mfn(int nr)
++{
++ struct xen_memory_reservation reservation = {
++ .extent_order = 0,
++ .domid = DOMID_SELF
++ };
++ int rc;
++
++ if (likely(alloc_index >= nr))
++ return 0;
++
++ set_xen_guest_handle(reservation.extent_start, mfn_list + alloc_index);
++ reservation.nr_extents = MAX_MFN_ALLOC - alloc_index;
++ rc = HYPERVISOR_memory_op(XENMEM_increase_reservation, &reservation);
++ if (likely(rc > 0))
++ alloc_index += rc;
++
++ return alloc_index >= nr ? 0 : -ENOMEM;
++}
++
++static inline void maybe_schedule_tx_action(void)
++{
++ smp_mb();
++ if ((NR_PENDING_REQS < (MAX_PENDING_REQS/2)) &&
++ !list_empty(&net_schedule_list))
++ tasklet_schedule(&net_tx_tasklet);
++}
++
++static struct sk_buff *netbk_copy_skb(struct sk_buff *skb)
++{
++ struct skb_shared_info *ninfo;
++ struct sk_buff *nskb;
++ unsigned long offset;
++ int ret;
++ int len;
++ int headlen;
++
++ BUG_ON(skb_shinfo(skb)->frag_list != NULL);
++
++ nskb = alloc_skb(SKB_MAX_HEAD(0), GFP_ATOMIC | __GFP_NOWARN);
++ if (unlikely(!nskb))
++ goto err;
++
++ skb_reserve(nskb, 16 + NET_IP_ALIGN);
++ headlen = nskb->end - nskb->data;
++ if (headlen > skb_headlen(skb))
++ headlen = skb_headlen(skb);
++ ret = skb_copy_bits(skb, 0, __skb_put(nskb, headlen), headlen);
++ BUG_ON(ret);
++
++ ninfo = skb_shinfo(nskb);
++ ninfo->gso_size = skb_shinfo(skb)->gso_size;
++ ninfo->gso_type = skb_shinfo(skb)->gso_type;
++
++ offset = headlen;
++ len = skb->len - headlen;
++
++ nskb->len = skb->len;
++ nskb->data_len = len;
++ nskb->truesize += len;
++
++ while (len) {
++ struct page *page;
++ int copy;
++ int zero;
++
++ if (unlikely(ninfo->nr_frags >= MAX_SKB_FRAGS)) {
++ dump_stack();
++ goto err_free;
++ }
++
++ copy = len >= PAGE_SIZE ? PAGE_SIZE : len;
++ zero = len >= PAGE_SIZE ? 0 : __GFP_ZERO;
++
++ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN | zero);
++ if (unlikely(!page))
++ goto err_free;
++
++ ret = skb_copy_bits(skb, offset, page_address(page), copy);
++ BUG_ON(ret);
++
++ ninfo->frags[ninfo->nr_frags].page = page;
++ ninfo->frags[ninfo->nr_frags].page_offset = 0;
++ ninfo->frags[ninfo->nr_frags].size = copy;
++ ninfo->nr_frags++;
++
++ offset += copy;
++ len -= copy;
++ }
++
++ offset = nskb->data - skb->data;
++
++ nskb->h.raw = skb->h.raw + offset;
++ nskb->nh.raw = skb->nh.raw + offset;
++ nskb->mac.raw = skb->mac.raw + offset;
++
++ return nskb;
++
++ err_free:
++ kfree_skb(nskb);
++ err:
++ return NULL;
++}
++
++static inline int netbk_max_required_rx_slots(netif_t *netif)
++{
++ if (netif->features & (NETIF_F_SG|NETIF_F_TSO))
++ return MAX_SKB_FRAGS + 2; /* header + extra_info + frags */
++ return 1; /* all in one */
++}
++
++static inline int netbk_queue_full(netif_t *netif)
++{
++ RING_IDX peek = netif->rx_req_cons_peek;
++ RING_IDX needed = netbk_max_required_rx_slots(netif);
++
++ return ((netif->rx.sring->req_prod - peek) < needed) ||
++ ((netif->rx.rsp_prod_pvt + NET_RX_RING_SIZE - peek) < needed);
++}
++
++static void tx_queue_callback(unsigned long data)
++{
++ netif_t *netif = (netif_t *)data;
++ if (netif_schedulable(netif))
++ netif_wake_queue(netif->dev);
++}
++
++int netif_be_start_xmit(struct sk_buff *skb, struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++
++ BUG_ON(skb->dev != dev);
++
++ /* Drop the packet if the target domain has no receive buffers. */
++ if (unlikely(!netif_schedulable(netif) || netbk_queue_full(netif)))
++ goto drop;
++
++ /*
++ * Copy the packet here if it's destined for a flipping interface
++ * but isn't flippable (e.g. extra references to data).
++ * XXX For now we also copy skbuffs whose head crosses a page
++ * boundary, because netbk_gop_skb can't handle them.
++ */
++ if (!netif->copying_receiver ||
++ ((skb_headlen(skb) + offset_in_page(skb->data)) >= PAGE_SIZE)) {
++ struct sk_buff *nskb = netbk_copy_skb(skb);
++ if ( unlikely(nskb == NULL) )
++ goto drop;
++ /* Copy only the header fields we use in this driver. */
++ nskb->dev = skb->dev;
++ nskb->ip_summed = skb->ip_summed;
++ nskb->proto_data_valid = skb->proto_data_valid;
++ dev_kfree_skb(skb);
++ skb = nskb;
++ }
++
++ netif->rx_req_cons_peek += skb_shinfo(skb)->nr_frags + 1 +
++ !!skb_shinfo(skb)->gso_size;
++ netif_get(netif);
++
++ if (netbk_can_queue(dev) && netbk_queue_full(netif)) {
++ netif->rx.sring->req_event = netif->rx_req_cons_peek +
++ netbk_max_required_rx_slots(netif);
++ mb(); /* request notification /then/ check & stop the queue */
++ if (netbk_queue_full(netif)) {
++ netif_stop_queue(dev);
++ /*
++ * Schedule 500ms timeout to restart the queue, thus
++ * ensuring that an inactive queue will be drained.
++ * Packets will be immediately be dropped until more
++ * receive buffers become available (see
++ * netbk_queue_full() check above).
++ */
++ netif->tx_queue_timeout.data = (unsigned long)netif;
++ netif->tx_queue_timeout.function = tx_queue_callback;
++ __mod_timer(&netif->tx_queue_timeout, jiffies + HZ/2);
++ }
++ }
++
++ skb_queue_tail(&rx_queue, skb);
++ tasklet_schedule(&net_rx_tasklet);
++
++ return 0;
++
++ drop:
++ netif->stats.tx_dropped++;
++ dev_kfree_skb(skb);
++ return 0;
++}
++
++#if 0
++static void xen_network_done_notify(void)
++{
++ static struct net_device *eth0_dev = NULL;
++ if (unlikely(eth0_dev == NULL))
++ eth0_dev = __dev_get_by_name("eth0");
++ netif_rx_schedule(eth0_dev);
++}
++/*
++ * Add following to poll() function in NAPI driver (Tigon3 is example):
++ * if ( xen_network_done() )
++ * tg3_enable_ints(tp);
++ */
++int xen_network_done(void)
++{
++ return skb_queue_empty(&rx_queue);
++}
++#endif
++
++struct netrx_pending_operations {
++ unsigned trans_prod, trans_cons;
++ unsigned mmu_prod, mmu_mcl;
++ unsigned mcl_prod, mcl_cons;
++ unsigned copy_prod, copy_cons;
++ unsigned meta_prod, meta_cons;
++ mmu_update_t *mmu;
++ gnttab_transfer_t *trans;
++ gnttab_copy_t *copy;
++ multicall_entry_t *mcl;
++ struct netbk_rx_meta *meta;
++};
++
++/* Set up the grant operations for this fragment. If it's a flipping
++ interface, we also set up the unmap request from here. */
++static u16 netbk_gop_frag(netif_t *netif, struct netbk_rx_meta *meta,
++ int i, struct netrx_pending_operations *npo,
++ struct page *page, unsigned long size,
++ unsigned long offset)
++{
++ mmu_update_t *mmu;
++ gnttab_transfer_t *gop;
++ gnttab_copy_t *copy_gop;
++ multicall_entry_t *mcl;
++ netif_rx_request_t *req;
++ unsigned long old_mfn, new_mfn;
++
++ old_mfn = virt_to_mfn(page_address(page));
++
++ req = RING_GET_REQUEST(&netif->rx, netif->rx.req_cons + i);
++ if (netif->copying_receiver) {
++ /* The fragment needs to be copied rather than
++ flipped. */
++ meta->copy = 1;
++ copy_gop = npo->copy + npo->copy_prod++;
++ copy_gop->flags = GNTCOPY_dest_gref;
++ if (PageForeign(page)) {
++ struct pending_tx_info *src_pend =
++ &pending_tx_info[netif_page_index(page)];
++ copy_gop->source.domid = src_pend->netif->domid;
++ copy_gop->source.u.ref = src_pend->req.gref;
++ copy_gop->flags |= GNTCOPY_source_gref;
++ } else {
++ copy_gop->source.domid = DOMID_SELF;
++ copy_gop->source.u.gmfn = old_mfn;
++ }
++ copy_gop->source.offset = offset;
++ copy_gop->dest.domid = netif->domid;
++ copy_gop->dest.offset = 0;
++ copy_gop->dest.u.ref = req->gref;
++ copy_gop->len = size;
++ } else {
++ meta->copy = 0;
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ new_mfn = alloc_mfn();
++
++ /*
++ * Set the new P2M table entry before
++ * reassigning the old data page. Heed the
++ * comment in pgtable-2level.h:pte_page(). :-)
++ */
++ set_phys_to_machine(page_to_pfn(page), new_mfn);
++
++ mcl = npo->mcl + npo->mcl_prod++;
++ MULTI_update_va_mapping(mcl,
++ (unsigned long)page_address(page),
++ pfn_pte_ma(new_mfn, PAGE_KERNEL),
++ 0);
++
++ mmu = npo->mmu + npo->mmu_prod++;
++ mmu->ptr = ((maddr_t)new_mfn << PAGE_SHIFT) |
++ MMU_MACHPHYS_UPDATE;
++ mmu->val = page_to_pfn(page);
++ }
++
++ gop = npo->trans + npo->trans_prod++;
++ gop->mfn = old_mfn;
++ gop->domid = netif->domid;
++ gop->ref = req->gref;
++ }
++ return req->id;
++}
++
++static void netbk_gop_skb(struct sk_buff *skb,
++ struct netrx_pending_operations *npo)
++{
++ netif_t *netif = netdev_priv(skb->dev);
++ int nr_frags = skb_shinfo(skb)->nr_frags;
++ int i;
++ int extra;
++ struct netbk_rx_meta *head_meta, *meta;
++
++ head_meta = npo->meta + npo->meta_prod++;
++ head_meta->frag.page_offset = skb_shinfo(skb)->gso_type;
++ head_meta->frag.size = skb_shinfo(skb)->gso_size;
++ extra = !!head_meta->frag.size + 1;
++
++ for (i = 0; i < nr_frags; i++) {
++ meta = npo->meta + npo->meta_prod++;
++ meta->frag = skb_shinfo(skb)->frags[i];
++ meta->id = netbk_gop_frag(netif, meta, i + extra, npo,
++ meta->frag.page,
++ meta->frag.size,
++ meta->frag.page_offset);
++ }
++
++ /*
++ * This must occur at the end to ensure that we don't trash skb_shinfo
++ * until we're done. We know that the head doesn't cross a page
++ * boundary because such packets get copied in netif_be_start_xmit.
++ */
++ head_meta->id = netbk_gop_frag(netif, head_meta, 0, npo,
++ virt_to_page(skb->data),
++ skb_headlen(skb),
++ offset_in_page(skb->data));
++
++ netif->rx.req_cons += nr_frags + extra;
++}
++
++static inline void netbk_free_pages(int nr_frags, struct netbk_rx_meta *meta)
++{
++ int i;
++
++ for (i = 0; i < nr_frags; i++)
++ put_page(meta[i].frag.page);
++}
++
++/* This is a twin to netbk_gop_skb. Assume that netbk_gop_skb was
++ used to set up the operations on the top of
++ netrx_pending_operations, which have since been done. Check that
++ they didn't give any errors and advance over them. */
++static int netbk_check_gop(int nr_frags, domid_t domid,
++ struct netrx_pending_operations *npo)
++{
++ multicall_entry_t *mcl;
++ gnttab_transfer_t *gop;
++ gnttab_copy_t *copy_op;
++ int status = NETIF_RSP_OKAY;
++ int i;
++
++ for (i = 0; i <= nr_frags; i++) {
++ if (npo->meta[npo->meta_cons + i].copy) {
++ copy_op = npo->copy + npo->copy_cons++;
++ if (copy_op->status != GNTST_okay) {
++ DPRINTK("Bad status %d from copy to DOM%d.\n",
++ copy_op->status, domid);
++ status = NETIF_RSP_ERROR;
++ }
++ } else {
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ mcl = npo->mcl + npo->mcl_cons++;
++ /* The update_va_mapping() must not fail. */
++ BUG_ON(mcl->result != 0);
++ }
++
++ gop = npo->trans + npo->trans_cons++;
++ /* Check the reassignment error code. */
++ if (gop->status != 0) {
++ DPRINTK("Bad status %d from grant transfer to DOM%u\n",
++ gop->status, domid);
++ /*
++ * Page no longer belongs to us unless
++ * GNTST_bad_page, but that should be
++ * a fatal error anyway.
++ */
++ BUG_ON(gop->status == GNTST_bad_page);
++ status = NETIF_RSP_ERROR;
++ }
++ }
++ }
++
++ return status;
++}
++
++static void netbk_add_frag_responses(netif_t *netif, int status,
++ struct netbk_rx_meta *meta, int nr_frags)
++{
++ int i;
++ unsigned long offset;
++
++ for (i = 0; i < nr_frags; i++) {
++ int id = meta[i].id;
++ int flags = (i == nr_frags - 1) ? 0 : NETRXF_more_data;
++
++ if (meta[i].copy)
++ offset = 0;
++ else
++ offset = meta[i].frag.page_offset;
++ make_rx_response(netif, id, status, offset,
++ meta[i].frag.size, flags);
++ }
++}
++
++static void net_rx_action(unsigned long unused)
++{
++ netif_t *netif = NULL;
++ s8 status;
++ u16 id, irq, flags;
++ netif_rx_response_t *resp;
++ multicall_entry_t *mcl;
++ struct sk_buff_head rxq;
++ struct sk_buff *skb;
++ int notify_nr = 0;
++ int ret;
++ int nr_frags;
++ int count;
++ unsigned long offset;
++
++ /*
++ * Putting hundreds of bytes on the stack is considered rude.
++ * Static works because a tasklet can only be on one CPU at any time.
++ */
++ static multicall_entry_t rx_mcl[NET_RX_RING_SIZE+3];
++ static mmu_update_t rx_mmu[NET_RX_RING_SIZE];
++ static gnttab_transfer_t grant_trans_op[NET_RX_RING_SIZE];
++ static gnttab_copy_t grant_copy_op[NET_RX_RING_SIZE];
++ static unsigned char rx_notify[NR_IRQS];
++ static u16 notify_list[NET_RX_RING_SIZE];
++ static struct netbk_rx_meta meta[NET_RX_RING_SIZE];
++
++ struct netrx_pending_operations npo = {
++ mmu: rx_mmu,
++ trans: grant_trans_op,
++ copy: grant_copy_op,
++ mcl: rx_mcl,
++ meta: meta};
++
++ skb_queue_head_init(&rxq);
++
++ count = 0;
++
++ while ((skb = skb_dequeue(&rx_queue)) != NULL) {
++ nr_frags = skb_shinfo(skb)->nr_frags;
++ *(int *)skb->cb = nr_frags;
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap) &&
++ !((netif_t *)netdev_priv(skb->dev))->copying_receiver &&
++ check_mfn(nr_frags + 1)) {
++ /* Memory squeeze? Back off for an arbitrary while. */
++ if ( net_ratelimit() )
++ WPRINTK("Memory squeeze in netback "
++ "driver.\n");
++ mod_timer(&net_timer, jiffies + HZ);
++ skb_queue_head(&rx_queue, skb);
++ break;
++ }
++
++ netbk_gop_skb(skb, &npo);
++
++ count += nr_frags + 1;
++
++ __skb_queue_tail(&rxq, skb);
++
++ /* Filled the batch queue? */
++ if (count + MAX_SKB_FRAGS >= NET_RX_RING_SIZE)
++ break;
++ }
++
++ BUG_ON(npo.meta_prod > ARRAY_SIZE(meta));
++
++ npo.mmu_mcl = npo.mcl_prod;
++ if (npo.mcl_prod) {
++ BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
++ BUG_ON(npo.mmu_prod > ARRAY_SIZE(rx_mmu));
++ mcl = npo.mcl + npo.mcl_prod++;
++
++ BUG_ON(mcl[-1].op != __HYPERVISOR_update_va_mapping);
++ mcl[-1].args[MULTI_UVMFLAGS_INDEX] = UVMF_TLB_FLUSH|UVMF_ALL;
++
++ mcl->op = __HYPERVISOR_mmu_update;
++ mcl->args[0] = (unsigned long)rx_mmu;
++ mcl->args[1] = npo.mmu_prod;
++ mcl->args[2] = 0;
++ mcl->args[3] = DOMID_SELF;
++ }
++
++ if (npo.trans_prod) {
++ BUG_ON(npo.trans_prod > ARRAY_SIZE(grant_trans_op));
++ mcl = npo.mcl + npo.mcl_prod++;
++ mcl->op = __HYPERVISOR_grant_table_op;
++ mcl->args[0] = GNTTABOP_transfer;
++ mcl->args[1] = (unsigned long)grant_trans_op;
++ mcl->args[2] = npo.trans_prod;
++ }
++
++ if (npo.copy_prod) {
++ BUG_ON(npo.copy_prod > ARRAY_SIZE(grant_copy_op));
++ mcl = npo.mcl + npo.mcl_prod++;
++ mcl->op = __HYPERVISOR_grant_table_op;
++ mcl->args[0] = GNTTABOP_copy;
++ mcl->args[1] = (unsigned long)grant_copy_op;
++ mcl->args[2] = npo.copy_prod;
++ }
++
++ /* Nothing to do? */
++ if (!npo.mcl_prod)
++ return;
++
++ BUG_ON(npo.mcl_prod > ARRAY_SIZE(rx_mcl));
++
++ ret = HYPERVISOR_multicall(npo.mcl, npo.mcl_prod);
++ BUG_ON(ret != 0);
++ /* The mmu_machphys_update() must not fail. */
++ BUG_ON(npo.mmu_mcl && npo.mcl[npo.mmu_mcl].result != 0);
++
++ while ((skb = __skb_dequeue(&rxq)) != NULL) {
++ nr_frags = *(int *)skb->cb;
++
++ netif = netdev_priv(skb->dev);
++ /* We can't rely on skb_release_data to release the
++ pages used by fragments for us, since it tries to
++ touch the pages in the fraglist. If we're in
++ flipping mode, that doesn't work. In copying mode,
++ we still have access to all of the pages, and so
++ it's safe to let release_data deal with it. */
++ /* (Freeing the fragments is safe since we copy
++ non-linear skbs destined for flipping interfaces) */
++ if (!netif->copying_receiver) {
++ atomic_set(&(skb_shinfo(skb)->dataref), 1);
++ skb_shinfo(skb)->frag_list = NULL;
++ skb_shinfo(skb)->nr_frags = 0;
++ netbk_free_pages(nr_frags, meta + npo.meta_cons + 1);
++ }
++
++ netif->stats.tx_bytes += skb->len;
++ netif->stats.tx_packets++;
++
++ status = netbk_check_gop(nr_frags, netif->domid, &npo);
++
++ id = meta[npo.meta_cons].id;
++ flags = nr_frags ? NETRXF_more_data : 0;
++
++ if (skb->ip_summed == CHECKSUM_HW) /* local packet? */
++ flags |= NETRXF_csum_blank | NETRXF_data_validated;
++ else if (skb->proto_data_valid) /* remote but checksummed? */
++ flags |= NETRXF_data_validated;
++
++ if (meta[npo.meta_cons].copy)
++ offset = 0;
++ else
++ offset = offset_in_page(skb->data);
++ resp = make_rx_response(netif, id, status, offset,
++ skb_headlen(skb), flags);
++
++ if (meta[npo.meta_cons].frag.size) {
++ struct netif_extra_info *gso =
++ (struct netif_extra_info *)
++ RING_GET_RESPONSE(&netif->rx,
++ netif->rx.rsp_prod_pvt++);
++
++ resp->flags |= NETRXF_extra_info;
++
++ gso->u.gso.size = meta[npo.meta_cons].frag.size;
++ gso->u.gso.type = XEN_NETIF_GSO_TYPE_TCPV4;
++ gso->u.gso.pad = 0;
++ gso->u.gso.features = 0;
++
++ gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
++ gso->flags = 0;
++ }
++
++ netbk_add_frag_responses(netif, status,
++ meta + npo.meta_cons + 1,
++ nr_frags);
++
++ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&netif->rx, ret);
++ irq = netif->irq;
++ if (ret && !rx_notify[irq]) {
++ rx_notify[irq] = 1;
++ notify_list[notify_nr++] = irq;
++ }
++
++ if (netif_queue_stopped(netif->dev) &&
++ netif_schedulable(netif) &&
++ !netbk_queue_full(netif))
++ netif_wake_queue(netif->dev);
++
++ netif_put(netif);
++ dev_kfree_skb(skb);
++ npo.meta_cons += nr_frags + 1;
++ }
++
++ while (notify_nr != 0) {
++ irq = notify_list[--notify_nr];
++ rx_notify[irq] = 0;
++ notify_remote_via_irq(irq);
++ }
++
++ /* More work to do? */
++ if (!skb_queue_empty(&rx_queue) && !timer_pending(&net_timer))
++ tasklet_schedule(&net_rx_tasklet);
++#if 0
++ else
++ xen_network_done_notify();
++#endif
++}
++
++static void net_alarm(unsigned long unused)
++{
++ tasklet_schedule(&net_rx_tasklet);
++}
++
++static void netbk_tx_pending_timeout(unsigned long unused)
++{
++ tasklet_schedule(&net_tx_tasklet);
++}
++
++struct net_device_stats *netif_be_get_stats(struct net_device *dev)
++{
++ netif_t *netif = netdev_priv(dev);
++ return &netif->stats;
++}
++
++static int __on_net_schedule_list(netif_t *netif)
++{
++ return netif->list.next != NULL;
++}
++
++static void remove_from_net_schedule_list(netif_t *netif)
++{
++ spin_lock_irq(&net_schedule_list_lock);
++ if (likely(__on_net_schedule_list(netif))) {
++ list_del(&netif->list);
++ netif->list.next = NULL;
++ netif_put(netif);
++ }
++ spin_unlock_irq(&net_schedule_list_lock);
++}
++
++static void add_to_net_schedule_list_tail(netif_t *netif)
++{
++ if (__on_net_schedule_list(netif))
++ return;
++
++ spin_lock_irq(&net_schedule_list_lock);
++ if (!__on_net_schedule_list(netif) &&
++ likely(netif_schedulable(netif))) {
++ list_add_tail(&netif->list, &net_schedule_list);
++ netif_get(netif);
++ }
++ spin_unlock_irq(&net_schedule_list_lock);
++}
++
++/*
++ * Note on CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER:
++ * If this driver is pipelining transmit requests then we can be very
++ * aggressive in avoiding new-packet notifications -- frontend only needs to
++ * send a notification if there are no outstanding unreceived responses.
++ * If we may be buffer transmit buffers for any reason then we must be rather
++ * more conservative and treat this as the final check for pending work.
++ */
++void netif_schedule_work(netif_t *netif)
++{
++ int more_to_do;
++
++#ifdef CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER
++ more_to_do = RING_HAS_UNCONSUMED_REQUESTS(&netif->tx);
++#else
++ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, more_to_do);
++#endif
++
++ if (more_to_do) {
++ add_to_net_schedule_list_tail(netif);
++ maybe_schedule_tx_action();
++ }
++}
++
++void netif_deschedule_work(netif_t *netif)
++{
++ remove_from_net_schedule_list(netif);
++}
++
++
++static void tx_add_credit(netif_t *netif)
++{
++ unsigned long max_burst, max_credit;
++
++ /*
++ * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
++ * Otherwise the interface can seize up due to insufficient credit.
++ */
++ max_burst = RING_GET_REQUEST(&netif->tx, netif->tx.req_cons)->size;
++ max_burst = min(max_burst, 131072UL);
++ max_burst = max(max_burst, netif->credit_bytes);
++
++ /* Take care that adding a new chunk of credit doesn't wrap to zero. */
++ max_credit = netif->remaining_credit + netif->credit_bytes;
++ if (max_credit < netif->remaining_credit)
++ max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
++
++ netif->remaining_credit = min(max_credit, max_burst);
++}
++
++static void tx_credit_callback(unsigned long data)
++{
++ netif_t *netif = (netif_t *)data;
++ tx_add_credit(netif);
++ netif_schedule_work(netif);
++}
++
++static inline int copy_pending_req(PEND_RING_IDX pending_idx)
++{
++ return gnttab_copy_grant_page(grant_tx_handle[pending_idx],
++ &mmap_pages[pending_idx]);
++}
++
++inline static void net_tx_action_dealloc(void)
++{
++ struct netbk_tx_pending_inuse *inuse, *n;
++ gnttab_unmap_grant_ref_t *gop;
++ u16 pending_idx;
++ PEND_RING_IDX dc, dp;
++ netif_t *netif;
++ int ret;
++ LIST_HEAD(list);
++
++ dc = dealloc_cons;
++ gop = tx_unmap_ops;
++
++ /*
++ * Free up any grants we have finished using
++ */
++ do {
++ dp = dealloc_prod;
++
++ /* Ensure we see all indices enqueued by netif_idx_release(). */
++ smp_rmb();
++
++ while (dc != dp) {
++ unsigned long pfn;
++
++ pending_idx = dealloc_ring[MASK_PEND_IDX(dc++)];
++ list_move_tail(&pending_inuse[pending_idx].list, &list);
++
++ pfn = idx_to_pfn(pending_idx);
++ /* Already unmapped? */
++ if (!phys_to_machine_mapping_valid(pfn))
++ continue;
++
++ gnttab_set_unmap_op(gop, idx_to_kaddr(pending_idx),
++ GNTMAP_host_map,
++ grant_tx_handle[pending_idx]);
++ gop++;
++ }
++
++ if (netbk_copy_skb_mode != NETBK_DELAYED_COPY_SKB ||
++ list_empty(&pending_inuse_head))
++ break;
++
++ /* Copy any entries that have been pending for too long. */
++ list_for_each_entry_safe(inuse, n, &pending_inuse_head, list) {
++ if (time_after(inuse->alloc_time + HZ / 2, jiffies))
++ break;
++
++ switch (copy_pending_req(inuse - pending_inuse)) {
++ case 0:
++ list_move_tail(&inuse->list, &list);
++ continue;
++ case -EBUSY:
++ list_del_init(&inuse->list);
++ continue;
++ case -ENOENT:
++ continue;
++ }
++
++ break;
++ }
++ } while (dp != dealloc_prod);
++
++ dealloc_cons = dc;
++
++ ret = HYPERVISOR_grant_table_op(
++ GNTTABOP_unmap_grant_ref, tx_unmap_ops, gop - tx_unmap_ops);
++ BUG_ON(ret);
++
++ list_for_each_entry_safe(inuse, n, &list, list) {
++ pending_idx = inuse - pending_inuse;
++
++ netif = pending_tx_info[pending_idx].netif;
++
++ make_tx_response(netif, &pending_tx_info[pending_idx].req,
++ NETIF_RSP_OKAY);
++
++ /* Ready for next use. */
++ gnttab_reset_grant_page(mmap_pages[pending_idx]);
++
++ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
++
++ netif_put(netif);
++
++ list_del_init(&inuse->list);
++ }
++}
++
++static void netbk_tx_err(netif_t *netif, netif_tx_request_t *txp, RING_IDX end)
++{
++ RING_IDX cons = netif->tx.req_cons;
++
++ do {
++ make_tx_response(netif, txp, NETIF_RSP_ERROR);
++ if (cons >= end)
++ break;
++ txp = RING_GET_REQUEST(&netif->tx, cons++);
++ } while (1);
++ netif->tx.req_cons = cons;
++ netif_schedule_work(netif);
++ netif_put(netif);
++}
++
++static int netbk_count_requests(netif_t *netif, netif_tx_request_t *first,
++ netif_tx_request_t *txp, int work_to_do)
++{
++ RING_IDX cons = netif->tx.req_cons;
++ int frags = 0;
++
++ if (!(first->flags & NETTXF_more_data))
++ return 0;
++
++ do {
++ if (frags >= work_to_do) {
++ DPRINTK("Need more frags\n");
++ return -frags;
++ }
++
++ if (unlikely(frags >= MAX_SKB_FRAGS)) {
++ DPRINTK("Too many frags\n");
++ return -frags;
++ }
++
++ memcpy(txp, RING_GET_REQUEST(&netif->tx, cons + frags),
++ sizeof(*txp));
++ if (txp->size > first->size) {
++ DPRINTK("Frags galore\n");
++ return -frags;
++ }
++
++ first->size -= txp->size;
++ frags++;
++
++ if (unlikely((txp->offset + txp->size) > PAGE_SIZE)) {
++ DPRINTK("txp->offset: %x, size: %u\n",
++ txp->offset, txp->size);
++ return -frags;
++ }
++ } while ((txp++)->flags & NETTXF_more_data);
++
++ return frags;
++}
++
++static gnttab_map_grant_ref_t *netbk_get_requests(netif_t *netif,
++ struct sk_buff *skb,
++ netif_tx_request_t *txp,
++ gnttab_map_grant_ref_t *mop)
++{
++ struct skb_shared_info *shinfo = skb_shinfo(skb);
++ skb_frag_t *frags = shinfo->frags;
++ unsigned long pending_idx = *((u16 *)skb->data);
++ int i, start;
++
++ /* Skip first skb fragment if it is on same page as header fragment. */
++ start = ((unsigned long)shinfo->frags[0].page == pending_idx);
++
++ for (i = start; i < shinfo->nr_frags; i++, txp++) {
++ pending_idx = pending_ring[MASK_PEND_IDX(pending_cons++)];
++
++ gnttab_set_map_op(mop++, idx_to_kaddr(pending_idx),
++ GNTMAP_host_map | GNTMAP_readonly,
++ txp->gref, netif->domid);
++
++ memcpy(&pending_tx_info[pending_idx].req, txp, sizeof(*txp));
++ netif_get(netif);
++ pending_tx_info[pending_idx].netif = netif;
++ frags[i].page = (void *)pending_idx;
++ }
++
++ return mop;
++}
++
++static int netbk_tx_check_mop(struct sk_buff *skb,
++ gnttab_map_grant_ref_t **mopp)
++{
++ gnttab_map_grant_ref_t *mop = *mopp;
++ int pending_idx = *((u16 *)skb->data);
++ netif_t *netif = pending_tx_info[pending_idx].netif;
++ netif_tx_request_t *txp;
++ struct skb_shared_info *shinfo = skb_shinfo(skb);
++ int nr_frags = shinfo->nr_frags;
++ int i, err, start;
++
++ /* Check status of header. */
++ err = mop->status;
++ if (unlikely(err)) {
++ txp = &pending_tx_info[pending_idx].req;
++ make_tx_response(netif, txp, NETIF_RSP_ERROR);
++ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
++ netif_put(netif);
++ } else {
++ set_phys_to_machine(
++ __pa(idx_to_kaddr(pending_idx)) >> PAGE_SHIFT,
++ FOREIGN_FRAME(mop->dev_bus_addr >> PAGE_SHIFT));
++ grant_tx_handle[pending_idx] = mop->handle;
++ }
++
++ /* Skip first skb fragment if it is on same page as header fragment. */
++ start = ((unsigned long)shinfo->frags[0].page == pending_idx);
++
++ for (i = start; i < nr_frags; i++) {
++ int j, newerr;
++
++ pending_idx = (unsigned long)shinfo->frags[i].page;
++
++ /* Check error status: if okay then remember grant handle. */
++ newerr = (++mop)->status;
++ if (likely(!newerr)) {
++ set_phys_to_machine(
++ __pa(idx_to_kaddr(pending_idx))>>PAGE_SHIFT,
++ FOREIGN_FRAME(mop->dev_bus_addr>>PAGE_SHIFT));
++ grant_tx_handle[pending_idx] = mop->handle;
++ /* Had a previous error? Invalidate this fragment. */
++ if (unlikely(err))
++ netif_idx_release(pending_idx);
++ continue;
++ }
++
++ /* Error on this fragment: respond to client with an error. */
++ txp = &pending_tx_info[pending_idx].req;
++ make_tx_response(netif, txp, NETIF_RSP_ERROR);
++ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
++ netif_put(netif);
++
++ /* Not the first error? Preceding frags already invalidated. */
++ if (err)
++ continue;
++
++ /* First error: invalidate header and preceding fragments. */
++ pending_idx = *((u16 *)skb->data);
++ netif_idx_release(pending_idx);
++ for (j = start; j < i; j++) {
++ pending_idx = (unsigned long)shinfo->frags[i].page;
++ netif_idx_release(pending_idx);
++ }
++
++ /* Remember the error: invalidate all subsequent fragments. */
++ err = newerr;
++ }
++
++ *mopp = mop + 1;
++ return err;
++}
++
++static void netbk_fill_frags(struct sk_buff *skb)
++{
++ struct skb_shared_info *shinfo = skb_shinfo(skb);
++ int nr_frags = shinfo->nr_frags;
++ int i;
++
++ for (i = 0; i < nr_frags; i++) {
++ skb_frag_t *frag = shinfo->frags + i;
++ netif_tx_request_t *txp;
++ unsigned long pending_idx;
++
++ pending_idx = (unsigned long)frag->page;
++
++ pending_inuse[pending_idx].alloc_time = jiffies;
++ list_add_tail(&pending_inuse[pending_idx].list,
++ &pending_inuse_head);
++
++ txp = &pending_tx_info[pending_idx].req;
++ frag->page = virt_to_page(idx_to_kaddr(pending_idx));
++ frag->size = txp->size;
++ frag->page_offset = txp->offset;
++
++ skb->len += txp->size;
++ skb->data_len += txp->size;
++ skb->truesize += txp->size;
++ }
++}
++
++int netbk_get_extras(netif_t *netif, struct netif_extra_info *extras,
++ int work_to_do)
++{
++ struct netif_extra_info extra;
++ RING_IDX cons = netif->tx.req_cons;
++
++ do {
++ if (unlikely(work_to_do-- <= 0)) {
++ DPRINTK("Missing extra info\n");
++ return -EBADR;
++ }
++
++ memcpy(&extra, RING_GET_REQUEST(&netif->tx, cons),
++ sizeof(extra));
++ if (unlikely(!extra.type ||
++ extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
++ netif->tx.req_cons = ++cons;
++ DPRINTK("Invalid extra type: %d\n", extra.type);
++ return -EINVAL;
++ }
++
++ memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
++ netif->tx.req_cons = ++cons;
++ } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
++
++ return work_to_do;
++}
++
++static int netbk_set_skb_gso(struct sk_buff *skb, struct netif_extra_info *gso)
++{
++ if (!gso->u.gso.size) {
++ DPRINTK("GSO size must not be zero.\n");
++ return -EINVAL;
++ }
++
++ /* Currently only TCPv4 S.O. is supported. */
++ if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4) {
++ DPRINTK("Bad GSO type %d.\n", gso->u.gso.type);
++ return -EINVAL;
++ }
++
++ skb_shinfo(skb)->gso_size = gso->u.gso.size;
++ skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
++
++ /* Header must be checked, and gso_segs computed. */
++ skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
++ skb_shinfo(skb)->gso_segs = 0;
++
++ return 0;
++}
++
++/* Called after netfront has transmitted */
++static void net_tx_action(unsigned long unused)
++{
++ struct list_head *ent;
++ struct sk_buff *skb;
++ netif_t *netif;
++ netif_tx_request_t txreq;
++ netif_tx_request_t txfrags[MAX_SKB_FRAGS];
++ struct netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX - 1];
++ u16 pending_idx;
++ RING_IDX i;
++ gnttab_map_grant_ref_t *mop;
++ unsigned int data_len;
++ int ret, work_to_do;
++
++ if (dealloc_cons != dealloc_prod)
++ net_tx_action_dealloc();
++
++ mop = tx_map_ops;
++ while (((NR_PENDING_REQS + MAX_SKB_FRAGS) < MAX_PENDING_REQS) &&
++ !list_empty(&net_schedule_list)) {
++ /* Get a netif from the list with work to do. */
++ ent = net_schedule_list.next;
++ netif = list_entry(ent, netif_t, list);
++ netif_get(netif);
++ remove_from_net_schedule_list(netif);
++
++ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, work_to_do);
++ if (!work_to_do) {
++ netif_put(netif);
++ continue;
++ }
++
++ i = netif->tx.req_cons;
++ rmb(); /* Ensure that we see the request before we copy it. */
++ memcpy(&txreq, RING_GET_REQUEST(&netif->tx, i), sizeof(txreq));
++
++ /* Credit-based scheduling. */
++ if (txreq.size > netif->remaining_credit) {
++ unsigned long now = jiffies;
++ unsigned long next_credit =
++ netif->credit_timeout.expires +
++ msecs_to_jiffies(netif->credit_usec / 1000);
++
++ /* Timer could already be pending in rare cases. */
++ if (timer_pending(&netif->credit_timeout)) {
++ netif_put(netif);
++ continue;
++ }
++
++ /* Passed the point where we can replenish credit? */
++ if (time_after_eq(now, next_credit)) {
++ netif->credit_timeout.expires = now;
++ tx_add_credit(netif);
++ }
++
++ /* Still too big to send right now? Set a callback. */
++ if (txreq.size > netif->remaining_credit) {
++ netif->credit_timeout.data =
++ (unsigned long)netif;
++ netif->credit_timeout.function =
++ tx_credit_callback;
++ __mod_timer(&netif->credit_timeout,
++ next_credit);
++ netif_put(netif);
++ continue;
++ }
++ }
++ netif->remaining_credit -= txreq.size;
++
++ work_to_do--;
++ netif->tx.req_cons = ++i;
++
++ memset(extras, 0, sizeof(extras));
++ if (txreq.flags & NETTXF_extra_info) {
++ work_to_do = netbk_get_extras(netif, extras,
++ work_to_do);
++ i = netif->tx.req_cons;
++ if (unlikely(work_to_do < 0)) {
++ netbk_tx_err(netif, &txreq, i);
++ continue;
++ }
++ }
++
++ ret = netbk_count_requests(netif, &txreq, txfrags, work_to_do);
++ if (unlikely(ret < 0)) {
++ netbk_tx_err(netif, &txreq, i - ret);
++ continue;
++ }
++ i += ret;
++
++ if (unlikely(txreq.size < ETH_HLEN)) {
++ DPRINTK("Bad packet size: %d\n", txreq.size);
++ netbk_tx_err(netif, &txreq, i);
++ continue;
++ }
++
++ /* No crossing a page as the payload mustn't fragment. */
++ if (unlikely((txreq.offset + txreq.size) > PAGE_SIZE)) {
++ DPRINTK("txreq.offset: %x, size: %u, end: %lu\n",
++ txreq.offset, txreq.size,
++ (txreq.offset &~PAGE_MASK) + txreq.size);
++ netbk_tx_err(netif, &txreq, i);
++ continue;
++ }
++
++ pending_idx = pending_ring[MASK_PEND_IDX(pending_cons)];
++
++ data_len = (txreq.size > PKT_PROT_LEN &&
++ ret < MAX_SKB_FRAGS) ?
++ PKT_PROT_LEN : txreq.size;
++
++ skb = alloc_skb(data_len + 16 + NET_IP_ALIGN,
++ GFP_ATOMIC | __GFP_NOWARN);
++ if (unlikely(skb == NULL)) {
++ DPRINTK("Can't allocate a skb in start_xmit.\n");
++ netbk_tx_err(netif, &txreq, i);
++ break;
++ }
++
++ /* Packets passed to netif_rx() must have some headroom. */
++ skb_reserve(skb, 16 + NET_IP_ALIGN);
++
++ if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
++ struct netif_extra_info *gso;
++ gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
++
++ if (netbk_set_skb_gso(skb, gso)) {
++ kfree_skb(skb);
++ netbk_tx_err(netif, &txreq, i);
++ continue;
++ }
++ }
++
++ gnttab_set_map_op(mop, idx_to_kaddr(pending_idx),
++ GNTMAP_host_map | GNTMAP_readonly,
++ txreq.gref, netif->domid);
++ mop++;
++
++ memcpy(&pending_tx_info[pending_idx].req,
++ &txreq, sizeof(txreq));
++ pending_tx_info[pending_idx].netif = netif;
++ *((u16 *)skb->data) = pending_idx;
++
++ __skb_put(skb, data_len);
++
++ skb_shinfo(skb)->nr_frags = ret;
++ if (data_len < txreq.size) {
++ skb_shinfo(skb)->nr_frags++;
++ skb_shinfo(skb)->frags[0].page =
++ (void *)(unsigned long)pending_idx;
++ } else {
++ /* Discriminate from any valid pending_idx value. */
++ skb_shinfo(skb)->frags[0].page = (void *)~0UL;
++ }
++
++ __skb_queue_tail(&tx_queue, skb);
++
++ pending_cons++;
++
++ mop = netbk_get_requests(netif, skb, txfrags, mop);
++
++ netif->tx.req_cons = i;
++ netif_schedule_work(netif);
++
++ if ((mop - tx_map_ops) >= ARRAY_SIZE(tx_map_ops))
++ break;
++ }
++
++ if (mop == tx_map_ops)
++ return;
++
++ ret = HYPERVISOR_grant_table_op(
++ GNTTABOP_map_grant_ref, tx_map_ops, mop - tx_map_ops);
++ BUG_ON(ret);
++
++ mop = tx_map_ops;
++ while ((skb = __skb_dequeue(&tx_queue)) != NULL) {
++ netif_tx_request_t *txp;
++
++ pending_idx = *((u16 *)skb->data);
++ netif = pending_tx_info[pending_idx].netif;
++ txp = &pending_tx_info[pending_idx].req;
++
++ /* Check the remap error code. */
++ if (unlikely(netbk_tx_check_mop(skb, &mop))) {
++ DPRINTK("netback grant failed.\n");
++ skb_shinfo(skb)->nr_frags = 0;
++ kfree_skb(skb);
++ continue;
++ }
++
++ data_len = skb->len;
++ memcpy(skb->data,
++ (void *)(idx_to_kaddr(pending_idx)|txp->offset),
++ data_len);
++ if (data_len < txp->size) {
++ /* Append the packet payload as a fragment. */
++ txp->offset += data_len;
++ txp->size -= data_len;
++ } else {
++ /* Schedule a response immediately. */
++ netif_idx_release(pending_idx);
++ }
++
++ /*
++ * Old frontends do not assert data_validated but we
++ * can infer it from csum_blank so test both flags.
++ */
++ if (txp->flags & (NETTXF_data_validated|NETTXF_csum_blank)) {
++ skb->ip_summed = CHECKSUM_UNNECESSARY;
++ skb->proto_data_valid = 1;
++ } else {
++ skb->ip_summed = CHECKSUM_NONE;
++ skb->proto_data_valid = 0;
++ }
++ skb->proto_csum_blank = !!(txp->flags & NETTXF_csum_blank);
++
++ netbk_fill_frags(skb);
++
++ skb->dev = netif->dev;
++ skb->protocol = eth_type_trans(skb, skb->dev);
++
++ netif->stats.rx_bytes += skb->len;
++ netif->stats.rx_packets++;
++
++ if (unlikely(netbk_copy_skb_mode == NETBK_ALWAYS_COPY_SKB) &&
++ unlikely(skb_linearize(skb))) {
++ DPRINTK("Can't linearize skb in net_tx_action.\n");
++ kfree_skb(skb);
++ continue;
++ }
++
++ netif_rx(skb);
++ netif->dev->last_rx = jiffies;
++ }
++
++ if (netbk_copy_skb_mode == NETBK_DELAYED_COPY_SKB &&
++ !list_empty(&pending_inuse_head)) {
++ struct netbk_tx_pending_inuse *oldest;
++
++ oldest = list_entry(pending_inuse_head.next,
++ struct netbk_tx_pending_inuse, list);
++ mod_timer(&netbk_tx_pending_timer, oldest->alloc_time + HZ);
++ }
++}
++
++static void netif_idx_release(u16 pending_idx)
++{
++ static DEFINE_SPINLOCK(_lock);
++ unsigned long flags;
++
++ spin_lock_irqsave(&_lock, flags);
++ dealloc_ring[MASK_PEND_IDX(dealloc_prod)] = pending_idx;
++ /* Sync with net_tx_action_dealloc: insert idx /then/ incr producer. */
++ smp_wmb();
++ dealloc_prod++;
++ spin_unlock_irqrestore(&_lock, flags);
++
++ tasklet_schedule(&net_tx_tasklet);
++}
++
++static void netif_page_release(struct page *page)
++{
++ netif_idx_release(netif_page_index(page));
++}
++
++irqreturn_t netif_be_int(int irq, void *dev_id, struct pt_regs *regs)
++{
++ netif_t *netif = dev_id;
++
++ add_to_net_schedule_list_tail(netif);
++ maybe_schedule_tx_action();
++
++ if (netif_schedulable(netif) && !netbk_queue_full(netif))
++ netif_wake_queue(netif->dev);
++
++ return IRQ_HANDLED;
++}
++
++static void make_tx_response(netif_t *netif,
++ netif_tx_request_t *txp,
++ s8 st)
++{
++ RING_IDX i = netif->tx.rsp_prod_pvt;
++ netif_tx_response_t *resp;
++ int notify;
++
++ resp = RING_GET_RESPONSE(&netif->tx, i);
++ resp->id = txp->id;
++ resp->status = st;
++
++ if (txp->flags & NETTXF_extra_info)
++ RING_GET_RESPONSE(&netif->tx, ++i)->status = NETIF_RSP_NULL;
++
++ netif->tx.rsp_prod_pvt = ++i;
++ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&netif->tx, notify);
++ if (notify)
++ notify_remote_via_irq(netif->irq);
++
++#ifdef CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER
++ if (i == netif->tx.req_cons) {
++ int more_to_do;
++ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, more_to_do);
++ if (more_to_do)
++ add_to_net_schedule_list_tail(netif);
++ }
++#endif
++}
++
++static netif_rx_response_t *make_rx_response(netif_t *netif,
++ u16 id,
++ s8 st,
++ u16 offset,
++ u16 size,
++ u16 flags)
++{
++ RING_IDX i = netif->rx.rsp_prod_pvt;
++ netif_rx_response_t *resp;
++
++ resp = RING_GET_RESPONSE(&netif->rx, i);
++ resp->offset = offset;
++ resp->flags = flags;
++ resp->id = id;
++ resp->status = (s16)size;
++ if (st < 0)
++ resp->status = (s16)st;
++
++ netif->rx.rsp_prod_pvt = ++i;
++
++ return resp;
++}
++
++#ifdef NETBE_DEBUG_INTERRUPT
++static irqreturn_t netif_be_dbg(int irq, void *dev_id, struct pt_regs *regs)
++{
++ struct list_head *ent;
++ netif_t *netif;
++ int i = 0;
++
++ printk(KERN_ALERT "netif_schedule_list:\n");
++ spin_lock_irq(&net_schedule_list_lock);
++
++ list_for_each (ent, &net_schedule_list) {
++ netif = list_entry(ent, netif_t, list);
++ printk(KERN_ALERT " %d: private(rx_req_cons=%08x "
++ "rx_resp_prod=%08x\n",
++ i, netif->rx.req_cons, netif->rx.rsp_prod_pvt);
++ printk(KERN_ALERT " tx_req_cons=%08x tx_resp_prod=%08x)\n",
++ netif->tx.req_cons, netif->tx.rsp_prod_pvt);
++ printk(KERN_ALERT " shared(rx_req_prod=%08x "
++ "rx_resp_prod=%08x\n",
++ netif->rx.sring->req_prod, netif->rx.sring->rsp_prod);
++ printk(KERN_ALERT " rx_event=%08x tx_req_prod=%08x\n",
++ netif->rx.sring->rsp_event, netif->tx.sring->req_prod);
++ printk(KERN_ALERT " tx_resp_prod=%08x, tx_event=%08x)\n",
++ netif->tx.sring->rsp_prod, netif->tx.sring->rsp_event);
++ i++;
++ }
++
++ spin_unlock_irq(&net_schedule_list_lock);
++ printk(KERN_ALERT " ** End of netif_schedule_list **\n");
++
++ return IRQ_HANDLED;
++}
++#endif
++
++static int __init netback_init(void)
++{
++ int i;
++ struct page *page;
++
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ /* We can increase reservation by this much in net_rx_action(). */
++ balloon_update_driver_allowance(NET_RX_RING_SIZE);
++
++ skb_queue_head_init(&rx_queue);
++ skb_queue_head_init(&tx_queue);
++
++ init_timer(&net_timer);
++ net_timer.data = 0;
++ net_timer.function = net_alarm;
++
++ init_timer(&netbk_tx_pending_timer);
++ netbk_tx_pending_timer.data = 0;
++ netbk_tx_pending_timer.function = netbk_tx_pending_timeout;
++
++ mmap_pages = alloc_empty_pages_and_pagevec(MAX_PENDING_REQS);
++ if (mmap_pages == NULL) {
++ printk("%s: out of memory\n", __FUNCTION__);
++ return -ENOMEM;
++ }
++
++ for (i = 0; i < MAX_PENDING_REQS; i++) {
++ page = mmap_pages[i];
++ SetPageForeign(page, netif_page_release);
++ netif_page_index(page) = i;
++ INIT_LIST_HEAD(&pending_inuse[i].list);
++ }
++
++ pending_cons = 0;
++ pending_prod = MAX_PENDING_REQS;
++ for (i = 0; i < MAX_PENDING_REQS; i++)
++ pending_ring[i] = i;
++
++ spin_lock_init(&net_schedule_list_lock);
++ INIT_LIST_HEAD(&net_schedule_list);
++
++ netbk_copy_skb_mode = NETBK_DONT_COPY_SKB;
++ if (MODPARM_copy_skb) {
++ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_replace,
++ NULL, 0))
++ netbk_copy_skb_mode = NETBK_ALWAYS_COPY_SKB;
++ else
++ netbk_copy_skb_mode = NETBK_DELAYED_COPY_SKB;
++ }
++
++ netif_accel_init();
++
++ netif_xenbus_init();
++
++#ifdef NETBE_DEBUG_INTERRUPT
++ (void)bind_virq_to_irqhandler(VIRQ_DEBUG,
++ 0,
++ netif_be_dbg,
++ SA_SHIRQ,
++ "net-be-dbg",
++ &netif_be_dbg);
++#endif
++
++ return 0;
++}
++
++module_init(netback_init);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/netback/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netback/xenbus.c 2008-05-19 00:33:45.846788782 +0300
+@@ -0,0 +1,452 @@
++/* Xenbus code for netif backend
++ Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
++ Copyright (C) 2005 XenSource Ltd
++
++ This program is free software; you can redistribute it and/or modify
++ it under the terms of the GNU General Public License as published by
++ the Free Software Foundation; either version 2 of the License, or
++ (at your option) any later version.
++
++ This program is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ GNU General Public License for more details.
++
++ You should have received a copy of the GNU General Public License
++ along with this program; if not, write to the Free Software
++ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++*/
++
++#include <stdarg.h>
++#include <linux/module.h>
++#include <xen/xenbus.h>
++#include "common.h"
++
++#if 0
++#undef DPRINTK
++#define DPRINTK(fmt, args...) \
++ printk("netback/xenbus (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
++#endif
++
++
++static int connect_rings(struct backend_info *);
++static void connect(struct backend_info *);
++static void backend_create_netif(struct backend_info *be);
++
++static int netback_remove(struct xenbus_device *dev)
++{
++ struct backend_info *be = dev->dev.driver_data;
++
++ netback_remove_accelerators(be, dev);
++
++ if (be->netif) {
++ netif_disconnect(be->netif);
++ be->netif = NULL;
++ }
++ kfree(be);
++ dev->dev.driver_data = NULL;
++ return 0;
++}
++
++
++/**
++ * Entry point to this code when a new device is created. Allocate the basic
++ * structures and switch to InitWait.
++ */
++static int netback_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ const char *message;
++ struct xenbus_transaction xbt;
++ int err;
++ int sg;
++ struct backend_info *be = kzalloc(sizeof(struct backend_info),
++ GFP_KERNEL);
++ if (!be) {
++ xenbus_dev_fatal(dev, -ENOMEM,
++ "allocating backend structure");
++ return -ENOMEM;
++ }
++
++ be->dev = dev;
++ dev->dev.driver_data = be;
++
++ sg = 1;
++ if (netbk_copy_skb_mode == NETBK_ALWAYS_COPY_SKB)
++ sg = 0;
++
++ do {
++ err = xenbus_transaction_start(&xbt);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "starting transaction");
++ goto fail;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-sg", "%d", sg);
++ if (err) {
++ message = "writing feature-sg";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-gso-tcpv4",
++ "%d", sg);
++ if (err) {
++ message = "writing feature-gso-tcpv4";
++ goto abort_transaction;
++ }
++
++ /* We support rx-copy path. */
++ err = xenbus_printf(xbt, dev->nodename,
++ "feature-rx-copy", "%d", 1);
++ if (err) {
++ message = "writing feature-rx-copy";
++ goto abort_transaction;
++ }
++
++ /*
++ * We don't support rx-flip path (except old guests who don't
++ * grok this feature flag).
++ */
++ err = xenbus_printf(xbt, dev->nodename,
++ "feature-rx-flip", "%d", 0);
++ if (err) {
++ message = "writing feature-rx-flip";
++ goto abort_transaction;
++ }
++
++ err = xenbus_transaction_end(xbt, 0);
++ } while (err == -EAGAIN);
++
++ if (err) {
++ xenbus_dev_fatal(dev, err, "completing transaction");
++ goto fail;
++ }
++
++ netback_probe_accelerators(be, dev);
++
++ err = xenbus_switch_state(dev, XenbusStateInitWait);
++ if (err)
++ goto fail;
++
++ /* This kicks hotplug scripts, so do it immediately. */
++ backend_create_netif(be);
++
++ return 0;
++
++abort_transaction:
++ xenbus_transaction_end(xbt, 1);
++ xenbus_dev_fatal(dev, err, "%s", message);
++fail:
++ DPRINTK("failed");
++ netback_remove(dev);
++ return err;
++}
++
++
++/**
++ * Handle the creation of the hotplug script environment. We add the script
++ * and vif variables to the environment, for the benefit of the vif-* hotplug
++ * scripts.
++ */
++static int netback_uevent(struct xenbus_device *xdev, char **envp,
++ int num_envp, char *buffer, int buffer_size)
++{
++ struct backend_info *be = xdev->dev.driver_data;
++ netif_t *netif = be->netif;
++ int i = 0, length = 0;
++ char *val;
++
++ DPRINTK("netback_uevent");
++
++ val = xenbus_read(XBT_NIL, xdev->nodename, "script", NULL);
++ if (IS_ERR(val)) {
++ int err = PTR_ERR(val);
++ xenbus_dev_fatal(xdev, err, "reading script");
++ return err;
++ }
++ else {
++ add_uevent_var(envp, num_envp, &i, buffer, buffer_size,
++ &length, "script=%s", val);
++ kfree(val);
++ }
++
++ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
++ "vif=%s", netif->dev->name);
++
++ envp[i] = NULL;
++
++ return 0;
++}
++
++
++static void backend_create_netif(struct backend_info *be)
++{
++ int err;
++ long handle;
++ struct xenbus_device *dev = be->dev;
++
++ if (be->netif != NULL)
++ return;
++
++ err = xenbus_scanf(XBT_NIL, dev->nodename, "handle", "%li", &handle);
++ if (err != 1) {
++ xenbus_dev_fatal(dev, err, "reading handle");
++ return;
++ }
++
++ be->netif = netif_alloc(dev->otherend_id, handle);
++ if (IS_ERR(be->netif)) {
++ err = PTR_ERR(be->netif);
++ be->netif = NULL;
++ xenbus_dev_fatal(dev, err, "creating interface");
++ return;
++ }
++
++ kobject_uevent(&dev->dev.kobj, KOBJ_ONLINE);
++}
++
++
++/**
++ * Callback received when the frontend's state changes.
++ */
++static void frontend_changed(struct xenbus_device *dev,
++ enum xenbus_state frontend_state)
++{
++ struct backend_info *be = dev->dev.driver_data;
++
++ DPRINTK("%s", xenbus_strstate(frontend_state));
++
++ be->frontend_state = frontend_state;
++
++ switch (frontend_state) {
++ case XenbusStateInitialising:
++ if (dev->state == XenbusStateClosed) {
++ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
++ __FUNCTION__, dev->nodename);
++ xenbus_switch_state(dev, XenbusStateInitWait);
++ }
++ break;
++
++ case XenbusStateInitialised:
++ break;
++
++ case XenbusStateConnected:
++ backend_create_netif(be);
++ if (be->netif)
++ connect(be);
++ break;
++
++ case XenbusStateClosing:
++ if (be->netif) {
++ netif_disconnect(be->netif);
++ be->netif = NULL;
++ }
++ xenbus_switch_state(dev, XenbusStateClosing);
++ break;
++
++ case XenbusStateClosed:
++ xenbus_switch_state(dev, XenbusStateClosed);
++ if (xenbus_dev_is_online(dev))
++ break;
++ /* fall through if not online */
++ case XenbusStateUnknown:
++ if (be->netif != NULL)
++ kobject_uevent(&dev->dev.kobj, KOBJ_OFFLINE);
++ device_unregister(&dev->dev);
++ break;
++
++ default:
++ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
++ frontend_state);
++ break;
++ }
++}
++
++
++static void xen_net_read_rate(struct xenbus_device *dev,
++ unsigned long *bytes, unsigned long *usec)
++{
++ char *s, *e;
++ unsigned long b, u;
++ char *ratestr;
++
++ /* Default to unlimited bandwidth. */
++ *bytes = ~0UL;
++ *usec = 0;
++
++ ratestr = xenbus_read(XBT_NIL, dev->nodename, "rate", NULL);
++ if (IS_ERR(ratestr))
++ return;
++
++ s = ratestr;
++ b = simple_strtoul(s, &e, 10);
++ if ((s == e) || (*e != ','))
++ goto fail;
++
++ s = e + 1;
++ u = simple_strtoul(s, &e, 10);
++ if ((s == e) || (*e != '\0'))
++ goto fail;
++
++ *bytes = b;
++ *usec = u;
++
++ kfree(ratestr);
++ return;
++
++ fail:
++ WPRINTK("Failed to parse network rate limit. Traffic unlimited.\n");
++ kfree(ratestr);
++}
++
++static int xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
++{
++ char *s, *e, *macstr;
++ int i;
++
++ macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
++ if (IS_ERR(macstr))
++ return PTR_ERR(macstr);
++
++ for (i = 0; i < ETH_ALEN; i++) {
++ mac[i] = simple_strtoul(s, &e, 16);
++ if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
++ kfree(macstr);
++ return -ENOENT;
++ }
++ s = e+1;
++ }
++
++ kfree(macstr);
++ return 0;
++}
++
++static void connect(struct backend_info *be)
++{
++ int err;
++ struct xenbus_device *dev = be->dev;
++
++ err = connect_rings(be);
++ if (err)
++ return;
++
++ err = xen_net_read_mac(dev, be->netif->fe_dev_addr);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "parsing %s/mac", dev->nodename);
++ return;
++ }
++
++ xen_net_read_rate(dev, &be->netif->credit_bytes,
++ &be->netif->credit_usec);
++ be->netif->remaining_credit = be->netif->credit_bytes;
++
++ xenbus_switch_state(dev, XenbusStateConnected);
++
++ netif_wake_queue(be->netif->dev);
++}
++
++
++static int connect_rings(struct backend_info *be)
++{
++ struct xenbus_device *dev = be->dev;
++ unsigned long tx_ring_ref, rx_ring_ref;
++ unsigned int evtchn, rx_copy;
++ int err;
++ int val;
++
++ DPRINTK("");
++
++ err = xenbus_gather(XBT_NIL, dev->otherend,
++ "tx-ring-ref", "%lu", &tx_ring_ref,
++ "rx-ring-ref", "%lu", &rx_ring_ref,
++ "event-channel", "%u", &evtchn, NULL);
++ if (err) {
++ xenbus_dev_fatal(dev, err,
++ "reading %s/ring-ref and event-channel",
++ dev->otherend);
++ return err;
++ }
++
++ err = xenbus_scanf(XBT_NIL, dev->otherend, "request-rx-copy", "%u",
++ &rx_copy);
++ if (err == -ENOENT) {
++ err = 0;
++ rx_copy = 0;
++ }
++ if (err < 0) {
++ xenbus_dev_fatal(dev, err, "reading %s/request-rx-copy",
++ dev->otherend);
++ return err;
++ }
++ be->netif->copying_receiver = !!rx_copy;
++
++ if (be->netif->dev->tx_queue_len != 0) {
++ if (xenbus_scanf(XBT_NIL, dev->otherend,
++ "feature-rx-notify", "%d", &val) < 0)
++ val = 0;
++ if (val)
++ be->netif->can_queue = 1;
++ else
++ /* Must be non-zero for pfifo_fast to work. */
++ be->netif->dev->tx_queue_len = 1;
++ }
++
++ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-sg", "%d", &val) < 0)
++ val = 0;
++ if (val) {
++ be->netif->features |= NETIF_F_SG;
++ be->netif->dev->features |= NETIF_F_SG;
++ }
++
++ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-gso-tcpv4", "%d",
++ &val) < 0)
++ val = 0;
++ if (val) {
++ be->netif->features |= NETIF_F_TSO;
++ be->netif->dev->features |= NETIF_F_TSO;
++ }
++
++ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-no-csum-offload",
++ "%d", &val) < 0)
++ val = 0;
++ if (val) {
++ be->netif->features &= ~NETIF_F_IP_CSUM;
++ be->netif->dev->features &= ~NETIF_F_IP_CSUM;
++ }
++
++ /* Map the shared frame, irq etc. */
++ err = netif_map(be->netif, tx_ring_ref, rx_ring_ref, evtchn);
++ if (err) {
++ xenbus_dev_fatal(dev, err,
++ "mapping shared-frames %lu/%lu port %u",
++ tx_ring_ref, rx_ring_ref, evtchn);
++ return err;
++ }
++ return 0;
++}
++
++
++/* ** Driver Registration ** */
++
++
++static const struct xenbus_device_id netback_ids[] = {
++ { "vif" },
++ { "" }
++};
++
++
++static struct xenbus_driver netback = {
++ .name = "vif",
++ .owner = THIS_MODULE,
++ .ids = netback_ids,
++ .probe = netback_probe,
++ .remove = netback_remove,
++ .uevent = netback_uevent,
++ .otherend_changed = frontend_changed,
++};
++
++
++void netif_xenbus_init(void)
++{
++ xenbus_register_backend(&netback);
++}
+--- linux-2.6.18.8/drivers/xen/netfront/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netfront/Makefile 2008-05-19 00:33:45.846788782 +0300
+@@ -0,0 +1,4 @@
++
++obj-$(CONFIG_XEN_NETDEV_FRONTEND) := xennet.o
++
++xennet-objs := netfront.o accel.o
+--- linux-2.6.18.8/drivers/xen/netfront/accel.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netfront/accel.c 2008-05-19 00:33:45.850789013 +0300
+@@ -0,0 +1,823 @@
++/******************************************************************************
++ * Virtual network driver for conversing with remote driver backends.
++ *
++ * Copyright (C) 2007 Solarflare Communications, Inc.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/netdevice.h>
++#include <linux/skbuff.h>
++#include <linux/list.h>
++#include <linux/mutex.h>
++#include <asm/hypervisor.h>
++#include <xen/xenbus.h>
++
++#include "netfront.h"
++
++#define DPRINTK(fmt, args...) \
++ pr_debug("netfront/accel (%s:%d) " fmt, \
++ __FUNCTION__, __LINE__, ##args)
++#define IPRINTK(fmt, args...) \
++ printk(KERN_INFO "netfront/accel: " fmt, ##args)
++#define WPRINTK(fmt, args...) \
++ printk(KERN_WARNING "netfront/accel: " fmt, ##args)
++
++static int netfront_remove_accelerator(struct netfront_info *np,
++ struct xenbus_device *dev);
++static int netfront_load_accelerator(struct netfront_info *np,
++ struct xenbus_device *dev,
++ const char *frontend);
++
++/*
++ * List of all netfront accelerator plugin modules available. Each
++ * list entry is of type struct netfront_accelerator.
++ */
++static struct list_head accelerators_list;
++
++/* Lock to protect access to accelerators_list */
++static spinlock_t accelerators_lock;
++
++/* Workqueue to process acceleration configuration changes */
++struct workqueue_struct *accel_watch_workqueue;
++
++/* Mutex to prevent concurrent loads and suspends, etc. */
++DEFINE_MUTEX(accelerator_mutex);
++
++void netif_init_accel(void)
++{
++ INIT_LIST_HEAD(&accelerators_list);
++ spin_lock_init(&accelerators_lock);
++
++ accel_watch_workqueue = create_workqueue("net_accel");
++}
++
++void netif_exit_accel(void)
++{
++ struct netfront_accelerator *accelerator, *tmp;
++ unsigned long flags;
++
++ flush_workqueue(accel_watch_workqueue);
++ destroy_workqueue(accel_watch_workqueue);
++
++ spin_lock_irqsave(&accelerators_lock, flags);
++
++ list_for_each_entry_safe(accelerator, tmp, &accelerators_list, link) {
++ BUG_ON(!list_empty(&accelerator->vif_states));
++
++ list_del(&accelerator->link);
++ kfree(accelerator->frontend);
++ kfree(accelerator);
++ }
++
++ spin_unlock_irqrestore(&accelerators_lock, flags);
++}
++
++
++/*
++ * Watch the configured accelerator and change plugin if it's modified
++ */
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++static void accel_watch_work(struct work_struct *context)
++#else
++static void accel_watch_work(void *context)
++#endif
++{
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ struct netfront_accel_vif_state *vif_state =
++ container_of(context, struct netfront_accel_vif_state,
++ accel_work);
++#else
++ struct netfront_accel_vif_state *vif_state =
++ (struct netfront_accel_vif_state *)context;
++#endif
++ struct netfront_info *np = vif_state->np;
++ char *accel_frontend;
++ int accel_len, rc = -1;
++
++ mutex_lock(&accelerator_mutex);
++
++ accel_frontend = xenbus_read(XBT_NIL, np->xbdev->otherend,
++ "accel-frontend", &accel_len);
++ if (IS_ERR(accel_frontend)) {
++ accel_frontend = NULL;
++ netfront_remove_accelerator(np, np->xbdev);
++ } else {
++ /* If this is the first time, request the accelerator,
++ otherwise only request one if it has changed */
++ if (vif_state->accel_frontend == NULL) {
++ rc = netfront_load_accelerator(np, np->xbdev,
++ accel_frontend);
++ } else {
++ if (strncmp(vif_state->accel_frontend, accel_frontend,
++ accel_len)) {
++ netfront_remove_accelerator(np, np->xbdev);
++ rc = netfront_load_accelerator(np, np->xbdev,
++ accel_frontend);
++ }
++ }
++ }
++
++ /* Get rid of previous state and replace with the new name */
++ if (vif_state->accel_frontend != NULL)
++ kfree(vif_state->accel_frontend);
++ vif_state->accel_frontend = accel_frontend;
++
++ mutex_unlock(&accelerator_mutex);
++
++ if (rc == 0) {
++ DPRINTK("requesting module %s\n", accel_frontend);
++ request_module("%s", accel_frontend);
++ /*
++ * Module should now call netfront_accelerator_loaded() once
++ * it's up and running, and we can continue from there
++ */
++ }
++}
++
++
++static void accel_watch_changed(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ struct netfront_accel_vif_state *vif_state =
++ container_of(watch, struct netfront_accel_vif_state,
++ accel_watch);
++ queue_work(accel_watch_workqueue, &vif_state->accel_work);
++}
++
++
++void netfront_accelerator_add_watch(struct netfront_info *np)
++{
++ int err;
++
++ /* Check we're not trying to overwrite an existing watch */
++ BUG_ON(np->accel_vif_state.accel_watch.node != NULL);
++
++ /* Get a watch on the accelerator plugin */
++ err = xenbus_watch_path2(np->xbdev, np->xbdev->otherend,
++ "accel-frontend",
++ &np->accel_vif_state.accel_watch,
++ accel_watch_changed);
++ if (err) {
++ DPRINTK("%s: Failed to register accel watch: %d\n",
++ __FUNCTION__, err);
++ np->accel_vif_state.accel_watch.node = NULL;
++ }
++}
++
++
++static
++void netfront_accelerator_remove_watch(struct netfront_info *np)
++{
++ struct netfront_accel_vif_state *vif_state = &np->accel_vif_state;
++
++ /* Get rid of watch on accelerator plugin */
++ if (vif_state->accel_watch.node != NULL) {
++ unregister_xenbus_watch(&vif_state->accel_watch);
++ kfree(vif_state->accel_watch.node);
++ vif_state->accel_watch.node = NULL;
++
++ flush_workqueue(accel_watch_workqueue);
++
++ /* Clean up any state left from watch */
++ if (vif_state->accel_frontend != NULL) {
++ kfree(vif_state->accel_frontend);
++ vif_state->accel_frontend = NULL;
++ }
++ }
++}
++
++
++/*
++ * Initialise the accel_vif_state field in the netfront state
++ */
++void init_accelerator_vif(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ np->accelerator = NULL;
++
++ /* It's assumed that these things don't change */
++ np->accel_vif_state.np = np;
++ np->accel_vif_state.dev = dev;
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ INIT_WORK(&np->accel_vif_state.accel_work, accel_watch_work);
++#else
++ INIT_WORK(&np->accel_vif_state.accel_work, accel_watch_work,
++ &np->accel_vif_state);
++#endif
++}
++
++
++/*
++ * Compare a frontend description string against an accelerator to see
++ * if they match. Would ultimately be nice to replace the string with
++ * a unique numeric identifier for each accelerator.
++ */
++static int match_accelerator(const char *frontend,
++ struct netfront_accelerator *accelerator)
++{
++ return strcmp(frontend, accelerator->frontend) == 0;
++}
++
++
++/*
++ * Add a frontend vif to the list of vifs that is using a netfront
++ * accelerator plugin module.
++ */
++static void add_accelerator_vif(struct netfront_accelerator *accelerator,
++ struct netfront_info *np)
++{
++ unsigned long flags;
++
++ /* Need lock to write list */
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++
++ if (np->accelerator == NULL) {
++ np->accelerator = accelerator;
++
++ list_add(&np->accel_vif_state.link, &accelerator->vif_states);
++ } else {
++ /*
++ * May get here legitimately if suspend_cancel is
++ * called, but in that case configuration should not
++ * have changed
++ */
++ BUG_ON(np->accelerator != accelerator);
++ }
++
++ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++}
++
++
++/*
++ * Initialise the state to track an accelerator plugin module.
++ */
++static int init_accelerator(const char *frontend,
++ struct netfront_accelerator **result,
++ struct netfront_accel_hooks *hooks)
++{
++ struct netfront_accelerator *accelerator =
++ kmalloc(sizeof(struct netfront_accelerator), GFP_KERNEL);
++ unsigned long flags;
++ int frontend_len;
++
++ if (!accelerator) {
++ DPRINTK("no memory for accelerator\n");
++ return -ENOMEM;
++ }
++
++ frontend_len = strlen(frontend) + 1;
++ accelerator->frontend = kmalloc(frontend_len, GFP_KERNEL);
++ if (!accelerator->frontend) {
++ DPRINTK("no memory for accelerator\n");
++ kfree(accelerator);
++ return -ENOMEM;
++ }
++ strlcpy(accelerator->frontend, frontend, frontend_len);
++
++ INIT_LIST_HEAD(&accelerator->vif_states);
++ spin_lock_init(&accelerator->vif_states_lock);
++
++ accelerator->hooks = hooks;
++
++ spin_lock_irqsave(&accelerators_lock, flags);
++ list_add(&accelerator->link, &accelerators_list);
++ spin_unlock_irqrestore(&accelerators_lock, flags);
++
++ *result = accelerator;
++
++ return 0;
++}
++
++
++/*
++ * Modify the hooks stored in the per-vif state to match that in the
++ * netfront accelerator's state.
++ */
++static void
++accelerator_set_vif_state_hooks(struct netfront_accel_vif_state *vif_state)
++{
++ /* This function must be called with the vif_states_lock held */
++
++ DPRINTK("%p\n",vif_state);
++
++ /* Make sure there are no data path operations going on */
++ netif_poll_disable(vif_state->np->netdev);
++ netif_tx_lock_bh(vif_state->np->netdev);
++
++ vif_state->hooks = vif_state->np->accelerator->hooks;
++
++ netif_tx_unlock_bh(vif_state->np->netdev);
++ netif_poll_enable(vif_state->np->netdev);
++}
++
++
++static void accelerator_probe_new_vif(struct netfront_info *np,
++ struct xenbus_device *dev,
++ struct netfront_accelerator *accelerator)
++{
++ struct netfront_accel_hooks *hooks;
++ unsigned long flags;
++
++ DPRINTK("\n");
++
++ /* Include this frontend device on the accelerator's list */
++ add_accelerator_vif(accelerator, np);
++
++ hooks = accelerator->hooks;
++
++ if (hooks) {
++ if (hooks->new_device(np->netdev, dev) == 0) {
++ spin_lock_irqsave
++ (&accelerator->vif_states_lock, flags);
++
++ accelerator_set_vif_state_hooks(&np->accel_vif_state);
++
++ spin_unlock_irqrestore
++ (&accelerator->vif_states_lock, flags);
++ }
++ }
++
++ return;
++}
++
++
++/*
++ * Request that a particular netfront accelerator plugin is loaded.
++ * Usually called as a result of the vif configuration specifying
++ * which one to use. Must be called with accelerator_mutex held
++ */
++static int netfront_load_accelerator(struct netfront_info *np,
++ struct xenbus_device *dev,
++ const char *frontend)
++{
++ struct netfront_accelerator *accelerator;
++ int rc = 0;
++
++ DPRINTK(" %s\n", frontend);
++
++ /*
++ * Look at list of loaded accelerators to see if the requested
++ * one is already there
++ */
++ list_for_each_entry(accelerator, &accelerators_list, link) {
++ if (match_accelerator(frontend, accelerator)) {
++ accelerator_probe_new_vif(np, dev, accelerator);
++ return 0;
++ }
++ }
++
++ /* Couldn't find it, so create a new one and load the module */
++ if ((rc = init_accelerator(frontend, &accelerator, NULL)) < 0) {
++ return rc;
++ }
++
++ /* Include this frontend device on the accelerator's list */
++ add_accelerator_vif(accelerator, np);
++
++ return rc;
++}
++
++
++/*
++ * Go through all the netfront vifs and see if they have requested
++ * this accelerator. Notify the accelerator plugin of the relevant
++ * device if so. Called when an accelerator plugin module is first
++ * loaded and connects to netfront.
++ */
++static void
++accelerator_probe_vifs(struct netfront_accelerator *accelerator,
++ struct netfront_accel_hooks *hooks)
++{
++ struct netfront_accel_vif_state *vif_state, *tmp;
++ unsigned long flags;
++
++ DPRINTK("%p\n", accelerator);
++
++ /*
++ * Store the hooks for future calls to probe a new device, and
++ * to wire into the vif_state once the accelerator plugin is
++ * ready to accelerate each vif
++ */
++ BUG_ON(hooks == NULL);
++ accelerator->hooks = hooks;
++
++ /*
++ * currently hold accelerator_mutex, so don't need
++ * vif_states_lock to read the list
++ */
++ list_for_each_entry_safe(vif_state, tmp, &accelerator->vif_states,
++ link) {
++ struct netfront_info *np = vif_state->np;
++
++ if (hooks->new_device(np->netdev, vif_state->dev) == 0) {
++ spin_lock_irqsave
++ (&accelerator->vif_states_lock, flags);
++
++ accelerator_set_vif_state_hooks(vif_state);
++
++ spin_unlock_irqrestore
++ (&accelerator->vif_states_lock, flags);
++ }
++ }
++}
++
++
++/*
++ * Called by the netfront accelerator plugin module when it has loaded
++ */
++int netfront_accelerator_loaded(int version, const char *frontend,
++ struct netfront_accel_hooks *hooks)
++{
++ struct netfront_accelerator *accelerator;
++
++ if (is_initial_xendomain())
++ return -EINVAL;
++
++ if (version != NETFRONT_ACCEL_VERSION) {
++ if (version > NETFRONT_ACCEL_VERSION) {
++ /* Caller has higher version number, leave it
++ up to them to decide whether to continue.
++ They can re-call with a lower number if
++ they're happy to be compatible with us */
++ return NETFRONT_ACCEL_VERSION;
++ } else {
++ /* We have a more recent version than caller.
++ Currently reject, but may in future be able
++ to be backwardly compatible */
++ return -EPROTO;
++ }
++ }
++
++ mutex_lock(&accelerator_mutex);
++
++ /*
++ * Look through list of accelerators to see if it has already
++ * been requested
++ */
++ list_for_each_entry(accelerator, &accelerators_list, link) {
++ if (match_accelerator(frontend, accelerator)) {
++ accelerator_probe_vifs(accelerator, hooks);
++ goto out;
++ }
++ }
++
++ /*
++ * If it wasn't in the list, add it now so that when it is
++ * requested the caller will find it
++ */
++ DPRINTK("Couldn't find matching accelerator (%s)\n",
++ frontend);
++
++ init_accelerator(frontend, &accelerator, hooks);
++
++ out:
++ mutex_unlock(&accelerator_mutex);
++ return 0;
++}
++EXPORT_SYMBOL_GPL(netfront_accelerator_loaded);
++
++
++/*
++ * Remove the hooks from a single vif state.
++ */
++static void
++accelerator_remove_single_hook(struct netfront_accelerator *accelerator,
++ struct netfront_accel_vif_state *vif_state)
++{
++ /* Make sure there are no data path operations going on */
++ netif_poll_disable(vif_state->np->netdev);
++ netif_tx_lock_bh(vif_state->np->netdev);
++
++ /*
++ * Remove the hooks, but leave the vif_state on the
++ * accelerator's list as that signifies this vif is
++ * interested in using that accelerator if it becomes
++ * available again
++ */
++ vif_state->hooks = NULL;
++
++ netif_tx_unlock_bh(vif_state->np->netdev);
++ netif_poll_enable(vif_state->np->netdev);
++}
++
++
++/*
++ * Safely remove the accelerator function hooks from a netfront state.
++ */
++static void accelerator_remove_hooks(struct netfront_accelerator *accelerator)
++{
++ struct netfront_accel_hooks *hooks;
++ struct netfront_accel_vif_state *vif_state, *tmp;
++ unsigned long flags;
++
++ /* Mutex is held so don't need vif_states_lock to iterate list */
++ list_for_each_entry_safe(vif_state, tmp,
++ &accelerator->vif_states,
++ link) {
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++
++ if(vif_state->hooks) {
++ hooks = vif_state->hooks;
++
++ /* Last chance to get statistics from the accelerator */
++ hooks->get_stats(vif_state->np->netdev,
++ &vif_state->np->stats);
++
++ spin_unlock_irqrestore(&accelerator->vif_states_lock,
++ flags);
++
++ accelerator_remove_single_hook(accelerator, vif_state);
++
++ accelerator->hooks->remove(vif_state->dev);
++ } else {
++ spin_unlock_irqrestore(&accelerator->vif_states_lock,
++ flags);
++ }
++ }
++
++ accelerator->hooks = NULL;
++}
++
++
++/*
++ * Called by a netfront accelerator when it is unloaded. This safely
++ * removes the hooks into the plugin and blocks until all devices have
++ * finished using it, so on return it is safe to unload.
++ */
++void netfront_accelerator_stop(const char *frontend)
++{
++ struct netfront_accelerator *accelerator;
++ unsigned long flags;
++
++ mutex_lock(&accelerator_mutex);
++ spin_lock_irqsave(&accelerators_lock, flags);
++
++ list_for_each_entry(accelerator, &accelerators_list, link) {
++ if (match_accelerator(frontend, accelerator)) {
++ spin_unlock_irqrestore(&accelerators_lock, flags);
++
++ accelerator_remove_hooks(accelerator);
++
++ goto out;
++ }
++ }
++ spin_unlock_irqrestore(&accelerators_lock, flags);
++ out:
++ mutex_unlock(&accelerator_mutex);
++}
++EXPORT_SYMBOL_GPL(netfront_accelerator_stop);
++
++
++/* Helper for call_remove and do_suspend */
++static int do_remove(struct netfront_info *np, struct xenbus_device *dev,
++ unsigned long *lock_flags)
++{
++ struct netfront_accelerator *accelerator = np->accelerator;
++ struct netfront_accel_hooks *hooks;
++ int rc = 0;
++
++ if (np->accel_vif_state.hooks) {
++ hooks = np->accel_vif_state.hooks;
++
++ /* Last chance to get statistics from the accelerator */
++ hooks->get_stats(np->netdev, &np->stats);
++
++ spin_unlock_irqrestore(&accelerator->vif_states_lock,
++ *lock_flags);
++
++ /*
++ * Try and do the opposite of accelerator_probe_new_vif
++ * to ensure there's no state pointing back at the
++ * netdev
++ */
++ accelerator_remove_single_hook(accelerator,
++ &np->accel_vif_state);
++
++ rc = accelerator->hooks->remove(dev);
++
++ spin_lock_irqsave(&accelerator->vif_states_lock, *lock_flags);
++ }
++
++ return rc;
++}
++
++
++static int netfront_remove_accelerator(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ struct netfront_accelerator *accelerator;
++ struct netfront_accel_vif_state *tmp_vif_state;
++ unsigned long flags;
++ int rc = 0;
++
++ /* Check that we've got a device that was accelerated */
++ if (np->accelerator == NULL)
++ return rc;
++
++ accelerator = np->accelerator;
++
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++
++ list_for_each_entry(tmp_vif_state, &accelerator->vif_states,
++ link) {
++ if (tmp_vif_state == &np->accel_vif_state) {
++ list_del(&np->accel_vif_state.link);
++ break;
++ }
++ }
++
++ rc = do_remove(np, dev, &flags);
++
++ np->accelerator = NULL;
++
++ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++
++ return rc;
++}
++
++
++int netfront_accelerator_call_remove(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ int rc;
++ netfront_accelerator_remove_watch(np);
++ mutex_lock(&accelerator_mutex);
++ rc = netfront_remove_accelerator(np, dev);
++ mutex_unlock(&accelerator_mutex);
++ return rc;
++}
++
++
++int netfront_accelerator_suspend(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ unsigned long flags;
++ int rc = 0;
++
++ netfront_accelerator_remove_watch(np);
++
++ mutex_lock(&accelerator_mutex);
++
++ /* Check that we've got a device that was accelerated */
++ if (np->accelerator == NULL)
++ goto out;
++
++ /*
++ * Call the remove accelerator hook, but leave the vif_state
++ * on the accelerator's list in case there is a suspend_cancel.
++ */
++ spin_lock_irqsave(&np->accelerator->vif_states_lock, flags);
++
++ rc = do_remove(np, dev, &flags);
++
++ spin_unlock_irqrestore(&np->accelerator->vif_states_lock, flags);
++ out:
++ mutex_unlock(&accelerator_mutex);
++ return rc;
++}
++
++
++int netfront_accelerator_suspend_cancel(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ /*
++ * Setting the watch will cause it to fire and probe the
++ * accelerator, so no need to call accelerator_probe_new_vif()
++ * directly here
++ */
++ netfront_accelerator_add_watch(np);
++ return 0;
++}
++
++
++void netfront_accelerator_resume(struct netfront_info *np,
++ struct xenbus_device *dev)
++{
++ struct netfront_accel_vif_state *accel_vif_state = NULL;
++ spinlock_t *vif_states_lock;
++ unsigned long flags;
++
++ mutex_lock(&accelerator_mutex);
++
++ /* Check that we've got a device that was accelerated */
++ if(np->accelerator == NULL)
++ goto out;
++
++ /* Find the vif_state from the accelerator's list */
++ list_for_each_entry(accel_vif_state, &np->accelerator->vif_states,
++ link) {
++ if (accel_vif_state->dev == dev) {
++ BUG_ON(accel_vif_state != &np->accel_vif_state);
++
++ vif_states_lock = &np->accelerator->vif_states_lock;
++ spin_lock_irqsave(vif_states_lock, flags);
++
++ /*
++ * Remove it from the accelerator's list so
++ * state is consistent for probing new vifs
++ * when they get connected
++ */
++ list_del(&accel_vif_state->link);
++ np->accelerator = NULL;
++
++ spin_unlock_irqrestore(vif_states_lock, flags);
++
++ break;
++ }
++ }
++
++ out:
++ mutex_unlock(&accelerator_mutex);
++ return;
++}
++
++
++int netfront_check_accelerator_queue_ready(struct net_device *dev,
++ struct netfront_info *np)
++{
++ struct netfront_accelerator *accelerator;
++ struct netfront_accel_hooks *hooks;
++ int rc = 1;
++ unsigned long flags;
++
++ accelerator = np->accelerator;
++
++ /* Call the check_ready accelerator hook. */
++ if (np->accel_vif_state.hooks && accelerator) {
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++ hooks = np->accel_vif_state.hooks;
++ if (hooks && np->accelerator == accelerator)
++ rc = np->accel_vif_state.hooks->check_ready(dev);
++ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++ }
++
++ return rc;
++}
++
++
++void netfront_accelerator_call_stop_napi_irq(struct netfront_info *np,
++ struct net_device *dev)
++{
++ struct netfront_accelerator *accelerator;
++ struct netfront_accel_hooks *hooks;
++ unsigned long flags;
++
++ accelerator = np->accelerator;
++
++ /* Call the stop_napi_interrupts accelerator hook. */
++ if (np->accel_vif_state.hooks && accelerator != NULL) {
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++ hooks = np->accel_vif_state.hooks;
++ if (hooks && np->accelerator == accelerator)
++ np->accel_vif_state.hooks->stop_napi_irq(dev);
++ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++ }
++}
++
++
++int netfront_accelerator_call_get_stats(struct netfront_info *np,
++ struct net_device *dev)
++{
++ struct netfront_accelerator *accelerator;
++ struct netfront_accel_hooks *hooks;
++ unsigned long flags;
++ int rc = 0;
++
++ accelerator = np->accelerator;
++
++ /* Call the get_stats accelerator hook. */
++ if (np->accel_vif_state.hooks && accelerator != NULL) {
++ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++ hooks = np->accel_vif_state.hooks;
++ if (hooks && np->accelerator == accelerator)
++ rc = np->accel_vif_state.hooks->get_stats(dev,
++ &np->stats);
++ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++ }
++ return rc;
++}
++
+--- linux-2.6.18.8/drivers/xen/netfront/netfront.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.c 2008-05-19 00:33:45.878790627 +0300
+@@ -0,0 +1,2241 @@
++/******************************************************************************
++ * Virtual network driver for conversing with remote driver backends.
++ *
++ * Copyright (c) 2002-2005, K A Fraser
++ * Copyright (c) 2005, XenSource Ltd
++ * Copyright (C) 2007 Solarflare Communications, Inc.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#include <linux/module.h>
++#include <linux/version.h>
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/slab.h>
++#include <linux/string.h>
++#include <linux/errno.h>
++#include <linux/netdevice.h>
++#include <linux/inetdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/skbuff.h>
++#include <linux/init.h>
++#include <linux/bitops.h>
++#include <linux/ethtool.h>
++#include <linux/in.h>
++#include <linux/if_ether.h>
++#include <linux/io.h>
++#include <linux/moduleparam.h>
++#include <net/sock.h>
++#include <net/pkt_sched.h>
++#include <net/arp.h>
++#include <net/route.h>
++#include <asm/uaccess.h>
++#include <xen/evtchn.h>
++#include <xen/xenbus.h>
++#include <xen/interface/io/netif.h>
++#include <xen/interface/memory.h>
++#include <xen/balloon.h>
++#include <asm/page.h>
++#include <asm/maddr.h>
++#include <asm/uaccess.h>
++#include <xen/interface/grant_table.h>
++#include <xen/gnttab.h>
++
++struct netfront_cb {
++ struct page *page;
++ unsigned offset;
++};
++
++#define NETFRONT_SKB_CB(skb) ((struct netfront_cb *)((skb)->cb))
++
++#include "netfront.h"
++
++/*
++ * Mutually-exclusive module options to select receive data path:
++ * rx_copy : Packets are copied by network backend into local memory
++ * rx_flip : Page containing packet data is transferred to our ownership
++ * For fully-virtualised guests there is no option - copying must be used.
++ * For paravirtualised guests, flipping is the default.
++ */
++#ifdef CONFIG_XEN
++static int MODPARM_rx_copy = 0;
++module_param_named(rx_copy, MODPARM_rx_copy, bool, 0);
++MODULE_PARM_DESC(rx_copy, "Copy packets from network card (rather than flip)");
++static int MODPARM_rx_flip = 0;
++module_param_named(rx_flip, MODPARM_rx_flip, bool, 0);
++MODULE_PARM_DESC(rx_flip, "Flip packets from network card (rather than copy)");
++#else
++static const int MODPARM_rx_copy = 1;
++static const int MODPARM_rx_flip = 0;
++#endif
++
++#define RX_COPY_THRESHOLD 256
++
++/* If we don't have GSO, fake things up so that we never try to use it. */
++#if defined(NETIF_F_GSO)
++#define HAVE_GSO 1
++#define HAVE_TSO 1 /* TSO is a subset of GSO */
++#define HAVE_CSUM_OFFLOAD 1
++static inline void dev_disable_gso_features(struct net_device *dev)
++{
++ /* Turn off all GSO bits except ROBUST. */
++ dev->features &= (1 << NETIF_F_GSO_SHIFT) - 1;
++ dev->features |= NETIF_F_GSO_ROBUST;
++}
++#elif defined(NETIF_F_TSO)
++#define HAVE_GSO 0
++#define HAVE_TSO 1
++
++/* Some older kernels cannot cope with incorrect checksums,
++ * particularly in netfilter. I'm not sure there is 100% correlation
++ * with the presence of NETIF_F_TSO but it appears to be a good first
++ * approximiation.
++ */
++#define HAVE_CSUM_OFFLOAD 0
++
++#define gso_size tso_size
++#define gso_segs tso_segs
++static inline void dev_disable_gso_features(struct net_device *dev)
++{
++ /* Turn off all TSO bits. */
++ dev->features &= ~NETIF_F_TSO;
++}
++static inline int skb_is_gso(const struct sk_buff *skb)
++{
++ return skb_shinfo(skb)->tso_size;
++}
++static inline int skb_gso_ok(struct sk_buff *skb, int features)
++{
++ return (features & NETIF_F_TSO);
++}
++
++static inline int netif_needs_gso(struct net_device *dev, struct sk_buff *skb)
++{
++ return skb_is_gso(skb) &&
++ (!skb_gso_ok(skb, dev->features) ||
++ unlikely(skb->ip_summed != CHECKSUM_HW));
++}
++#else
++#define HAVE_GSO 0
++#define HAVE_TSO 0
++#define HAVE_CSUM_OFFLOAD 0
++#define netif_needs_gso(dev, skb) 0
++#define dev_disable_gso_features(dev) ((void)0)
++#define ethtool_op_set_tso(dev, data) (-ENOSYS)
++#endif
++
++#define GRANT_INVALID_REF 0
++
++struct netfront_rx_info {
++ struct netif_rx_response rx;
++ struct netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX - 1];
++};
++
++/*
++ * Implement our own carrier flag: the network stack's version causes delays
++ * when the carrier is re-enabled (in particular, dev_activate() may not
++ * immediately be called, which can cause packet loss).
++ */
++#define netfront_carrier_on(netif) ((netif)->carrier = 1)
++#define netfront_carrier_off(netif) ((netif)->carrier = 0)
++#define netfront_carrier_ok(netif) ((netif)->carrier)
++
++/*
++ * Access macros for acquiring freeing slots in tx_skbs[].
++ */
++
++static inline void add_id_to_freelist(struct sk_buff **list, unsigned short id)
++{
++ list[id] = list[0];
++ list[0] = (void *)(unsigned long)id;
++}
++
++static inline unsigned short get_id_from_freelist(struct sk_buff **list)
++{
++ unsigned int id = (unsigned int)(unsigned long)list[0];
++ list[0] = list[id];
++ return id;
++}
++
++static inline int xennet_rxidx(RING_IDX idx)
++{
++ return idx & (NET_RX_RING_SIZE - 1);
++}
++
++static inline struct sk_buff *xennet_get_rx_skb(struct netfront_info *np,
++ RING_IDX ri)
++{
++ int i = xennet_rxidx(ri);
++ struct sk_buff *skb = np->rx_skbs[i];
++ np->rx_skbs[i] = NULL;
++ return skb;
++}
++
++static inline grant_ref_t xennet_get_rx_ref(struct netfront_info *np,
++ RING_IDX ri)
++{
++ int i = xennet_rxidx(ri);
++ grant_ref_t ref = np->grant_rx_ref[i];
++ np->grant_rx_ref[i] = GRANT_INVALID_REF;
++ return ref;
++}
++
++#define DPRINTK(fmt, args...) \
++ pr_debug("netfront (%s:%d) " fmt, \
++ __FUNCTION__, __LINE__, ##args)
++#define IPRINTK(fmt, args...) \
++ printk(KERN_INFO "netfront: " fmt, ##args)
++#define WPRINTK(fmt, args...) \
++ printk(KERN_WARNING "netfront: " fmt, ##args)
++
++static int setup_device(struct xenbus_device *, struct netfront_info *);
++static struct net_device *create_netdev(struct xenbus_device *);
++
++static void end_access(int, void *);
++static void netif_disconnect_backend(struct netfront_info *);
++
++static int network_connect(struct net_device *);
++static void network_tx_buf_gc(struct net_device *);
++static void network_alloc_rx_buffers(struct net_device *);
++static void send_fake_arp(struct net_device *);
++
++static irqreturn_t netif_int(int irq, void *dev_id, struct pt_regs *ptregs);
++
++#ifdef CONFIG_SYSFS
++static int xennet_sysfs_addif(struct net_device *netdev);
++static void xennet_sysfs_delif(struct net_device *netdev);
++#else /* !CONFIG_SYSFS */
++#define xennet_sysfs_addif(dev) (0)
++#define xennet_sysfs_delif(dev) do { } while(0)
++#endif
++
++static inline int xennet_can_sg(struct net_device *dev)
++{
++ return dev->features & NETIF_F_SG;
++}
++
++/**
++ * Entry point to this code when a new device is created. Allocate the basic
++ * structures and the ring buffers for communication with the backend, and
++ * inform the backend of the appropriate details for those.
++ */
++static int __devinit netfront_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int err;
++ struct net_device *netdev;
++ struct netfront_info *info;
++
++ netdev = create_netdev(dev);
++ if (IS_ERR(netdev)) {
++ err = PTR_ERR(netdev);
++ xenbus_dev_fatal(dev, err, "creating netdev");
++ return err;
++ }
++
++ info = netdev_priv(netdev);
++ dev->dev.driver_data = info;
++
++ err = register_netdev(info->netdev);
++ if (err) {
++ printk(KERN_WARNING "%s: register_netdev err=%d\n",
++ __FUNCTION__, err);
++ goto fail;
++ }
++
++ err = xennet_sysfs_addif(info->netdev);
++ if (err) {
++ unregister_netdev(info->netdev);
++ printk(KERN_WARNING "%s: add sysfs failed err=%d\n",
++ __FUNCTION__, err);
++ goto fail;
++ }
++
++ return 0;
++
++ fail:
++ free_netdev(netdev);
++ dev->dev.driver_data = NULL;
++ return err;
++}
++
++static int __devexit netfront_remove(struct xenbus_device *dev)
++{
++ struct netfront_info *info = dev->dev.driver_data;
++
++ DPRINTK("%s\n", dev->nodename);
++
++ netfront_accelerator_call_remove(info, dev);
++
++ netif_disconnect_backend(info);
++
++ del_timer_sync(&info->rx_refill_timer);
++
++ xennet_sysfs_delif(info->netdev);
++
++ unregister_netdev(info->netdev);
++
++ free_netdev(info->netdev);
++
++ return 0;
++}
++
++
++static int netfront_suspend(struct xenbus_device *dev)
++{
++ struct netfront_info *info = dev->dev.driver_data;
++ return netfront_accelerator_suspend(info, dev);
++}
++
++
++static int netfront_suspend_cancel(struct xenbus_device *dev)
++{
++ struct netfront_info *info = dev->dev.driver_data;
++ return netfront_accelerator_suspend_cancel(info, dev);
++}
++
++
++/**
++ * We are reconnecting to the backend, due to a suspend/resume, or a backend
++ * driver restart. We tear down our netif structure and recreate it, but
++ * leave the device-layer structures intact so that this is transparent to the
++ * rest of the kernel.
++ */
++static int netfront_resume(struct xenbus_device *dev)
++{
++ struct netfront_info *info = dev->dev.driver_data;
++
++ DPRINTK("%s\n", dev->nodename);
++
++ netfront_accelerator_resume(info, dev);
++
++ netif_disconnect_backend(info);
++ return 0;
++}
++
++static int xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
++{
++ char *s, *e, *macstr;
++ int i;
++
++ macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
++ if (IS_ERR(macstr))
++ return PTR_ERR(macstr);
++
++ for (i = 0; i < ETH_ALEN; i++) {
++ mac[i] = simple_strtoul(s, &e, 16);
++ if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
++ kfree(macstr);
++ return -ENOENT;
++ }
++ s = e+1;
++ }
++
++ kfree(macstr);
++ return 0;
++}
++
++/* Common code used when first setting up, and when resuming. */
++static int talk_to_backend(struct xenbus_device *dev,
++ struct netfront_info *info)
++{
++ const char *message;
++ struct xenbus_transaction xbt;
++ int err;
++
++ /* Read mac only in the first setup. */
++ if (!is_valid_ether_addr(info->mac)) {
++ err = xen_net_read_mac(dev, info->mac);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "parsing %s/mac",
++ dev->nodename);
++ goto out;
++ }
++ }
++
++ /* Create shared ring, alloc event channel. */
++ err = setup_device(dev, info);
++ if (err)
++ goto out;
++
++ /* This will load an accelerator if one is configured when the
++ * watch fires */
++ netfront_accelerator_add_watch(info);
++
++again:
++ err = xenbus_transaction_start(&xbt);
++ if (err) {
++ xenbus_dev_fatal(dev, err, "starting transaction");
++ goto destroy_ring;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "tx-ring-ref","%u",
++ info->tx_ring_ref);
++ if (err) {
++ message = "writing tx ring-ref";
++ goto abort_transaction;
++ }
++ err = xenbus_printf(xbt, dev->nodename, "rx-ring-ref","%u",
++ info->rx_ring_ref);
++ if (err) {
++ message = "writing rx ring-ref";
++ goto abort_transaction;
++ }
++ err = xenbus_printf(xbt, dev->nodename,
++ "event-channel", "%u",
++ irq_to_evtchn_port(info->irq));
++ if (err) {
++ message = "writing event-channel";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "request-rx-copy", "%u",
++ info->copying_receiver);
++ if (err) {
++ message = "writing request-rx-copy";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-rx-notify", "%d", 1);
++ if (err) {
++ message = "writing feature-rx-notify";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-no-csum-offload",
++ "%d", !HAVE_CSUM_OFFLOAD);
++ if (err) {
++ message = "writing feature-no-csum-offload";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-sg", "%d", 1);
++ if (err) {
++ message = "writing feature-sg";
++ goto abort_transaction;
++ }
++
++ err = xenbus_printf(xbt, dev->nodename, "feature-gso-tcpv4", "%d",
++ HAVE_TSO);
++ if (err) {
++ message = "writing feature-gso-tcpv4";
++ goto abort_transaction;
++ }
++
++ err = xenbus_transaction_end(xbt, 0);
++ if (err) {
++ if (err == -EAGAIN)
++ goto again;
++ xenbus_dev_fatal(dev, err, "completing transaction");
++ goto destroy_ring;
++ }
++
++ return 0;
++
++ abort_transaction:
++ xenbus_transaction_end(xbt, 1);
++ xenbus_dev_fatal(dev, err, "%s", message);
++ destroy_ring:
++ netfront_accelerator_call_remove(info, dev);
++ netif_disconnect_backend(info);
++ out:
++ return err;
++}
++
++static int setup_device(struct xenbus_device *dev, struct netfront_info *info)
++{
++ struct netif_tx_sring *txs;
++ struct netif_rx_sring *rxs;
++ int err;
++ struct net_device *netdev = info->netdev;
++
++ info->tx_ring_ref = GRANT_INVALID_REF;
++ info->rx_ring_ref = GRANT_INVALID_REF;
++ info->rx.sring = NULL;
++ info->tx.sring = NULL;
++ info->irq = 0;
++
++ txs = (struct netif_tx_sring *)get_zeroed_page(GFP_KERNEL|__GFP_HIGH);
++ if (!txs) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(dev, err, "allocating tx ring page");
++ goto fail;
++ }
++ SHARED_RING_INIT(txs);
++ FRONT_RING_INIT(&info->tx, txs, PAGE_SIZE);
++
++ err = xenbus_grant_ring(dev, virt_to_mfn(txs));
++ if (err < 0) {
++ free_page((unsigned long)txs);
++ goto fail;
++ }
++ info->tx_ring_ref = err;
++
++ rxs = (struct netif_rx_sring *)get_zeroed_page(GFP_KERNEL|__GFP_HIGH);
++ if (!rxs) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(dev, err, "allocating rx ring page");
++ goto fail;
++ }
++ SHARED_RING_INIT(rxs);
++ FRONT_RING_INIT(&info->rx, rxs, PAGE_SIZE);
++
++ err = xenbus_grant_ring(dev, virt_to_mfn(rxs));
++ if (err < 0) {
++ free_page((unsigned long)rxs);
++ goto fail;
++ }
++ info->rx_ring_ref = err;
++
++ memcpy(netdev->dev_addr, info->mac, ETH_ALEN);
++
++ err = bind_listening_port_to_irqhandler(
++ dev->otherend_id, netif_int, SA_SAMPLE_RANDOM, netdev->name,
++ netdev);
++ if (err < 0)
++ goto fail;
++ info->irq = err;
++
++ return 0;
++
++ fail:
++ return err;
++}
++
++/**
++ * Callback received when the backend's state changes.
++ */
++static void backend_changed(struct xenbus_device *dev,
++ enum xenbus_state backend_state)
++{
++ struct netfront_info *np = dev->dev.driver_data;
++ struct net_device *netdev = np->netdev;
++
++ DPRINTK("%s\n", xenbus_strstate(backend_state));
++
++ switch (backend_state) {
++ case XenbusStateInitialising:
++ case XenbusStateInitialised:
++ case XenbusStateConnected:
++ case XenbusStateReconfiguring:
++ case XenbusStateReconfigured:
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ break;
++
++ case XenbusStateInitWait:
++ if (dev->state != XenbusStateInitialising)
++ break;
++ if (network_connect(netdev) != 0)
++ break;
++ xenbus_switch_state(dev, XenbusStateConnected);
++ send_fake_arp(netdev);
++ break;
++
++ case XenbusStateClosing:
++ xenbus_frontend_closed(dev);
++ break;
++ }
++}
++
++/** Send a packet on a net device to encourage switches to learn the
++ * MAC. We send a fake ARP request.
++ *
++ * @param dev device
++ * @return 0 on success, error code otherwise
++ */
++static void send_fake_arp(struct net_device *dev)
++{
++#ifdef CONFIG_INET
++ struct sk_buff *skb;
++ u32 src_ip, dst_ip;
++
++ dst_ip = INADDR_BROADCAST;
++ src_ip = inet_select_addr(dev, dst_ip, RT_SCOPE_LINK);
++
++ /* No IP? Then nothing to do. */
++ if (src_ip == 0)
++ return;
++
++ skb = arp_create(ARPOP_REPLY, ETH_P_ARP,
++ dst_ip, dev, src_ip,
++ /*dst_hw*/ NULL, /*src_hw*/ NULL,
++ /*target_hw*/ dev->dev_addr);
++ if (skb == NULL)
++ return;
++
++ dev_queue_xmit(skb);
++#endif
++}
++
++static inline int netfront_tx_slot_available(struct netfront_info *np)
++{
++ return ((np->tx.req_prod_pvt - np->tx.rsp_cons) <
++ (TX_MAX_TARGET - MAX_SKB_FRAGS - 2));
++}
++
++
++static inline void network_maybe_wake_tx(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++
++ if (unlikely(netif_queue_stopped(dev)) &&
++ netfront_tx_slot_available(np) &&
++ likely(netif_running(dev)) &&
++ netfront_check_accelerator_queue_ready(dev, np))
++ netif_wake_queue(dev);
++}
++
++
++int netfront_check_queue_ready(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++
++ return unlikely(netif_queue_stopped(dev)) &&
++ netfront_tx_slot_available(np) &&
++ likely(netif_running(dev));
++}
++EXPORT_SYMBOL(netfront_check_queue_ready);
++
++
++static int network_open(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++
++ memset(&np->stats, 0, sizeof(np->stats));
++
++ spin_lock_bh(&np->rx_lock);
++ if (netfront_carrier_ok(np)) {
++ network_alloc_rx_buffers(dev);
++ np->rx.sring->rsp_event = np->rx.rsp_cons + 1;
++ if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)){
++ netfront_accelerator_call_stop_napi_irq(np, dev);
++
++ netif_rx_schedule(dev);
++ }
++ }
++ spin_unlock_bh(&np->rx_lock);
++
++ network_maybe_wake_tx(dev);
++
++ return 0;
++}
++
++static void network_tx_buf_gc(struct net_device *dev)
++{
++ RING_IDX cons, prod;
++ unsigned short id;
++ struct netfront_info *np = netdev_priv(dev);
++ struct sk_buff *skb;
++
++ BUG_ON(!netfront_carrier_ok(np));
++
++ do {
++ prod = np->tx.sring->rsp_prod;
++ rmb(); /* Ensure we see responses up to 'rp'. */
++
++ for (cons = np->tx.rsp_cons; cons != prod; cons++) {
++ struct netif_tx_response *txrsp;
++
++ txrsp = RING_GET_RESPONSE(&np->tx, cons);
++ if (txrsp->status == NETIF_RSP_NULL)
++ continue;
++
++ id = txrsp->id;
++ skb = np->tx_skbs[id];
++ if (unlikely(gnttab_query_foreign_access(
++ np->grant_tx_ref[id]) != 0)) {
++ printk(KERN_ALERT "network_tx_buf_gc: warning "
++ "-- grant still in use by backend "
++ "domain.\n");
++ BUG();
++ }
++ gnttab_end_foreign_access_ref(np->grant_tx_ref[id]);
++ gnttab_release_grant_reference(
++ &np->gref_tx_head, np->grant_tx_ref[id]);
++ np->grant_tx_ref[id] = GRANT_INVALID_REF;
++ add_id_to_freelist(np->tx_skbs, id);
++ dev_kfree_skb_irq(skb);
++ }
++
++ np->tx.rsp_cons = prod;
++
++ /*
++ * Set a new event, then check for race with update of tx_cons.
++ * Note that it is essential to schedule a callback, no matter
++ * how few buffers are pending. Even if there is space in the
++ * transmit ring, higher layers may be blocked because too much
++ * data is outstanding: in such cases notification from Xen is
++ * likely to be the only kick that we'll get.
++ */
++ np->tx.sring->rsp_event =
++ prod + ((np->tx.sring->req_prod - prod) >> 1) + 1;
++ mb();
++ } while ((cons == prod) && (prod != np->tx.sring->rsp_prod));
++
++ network_maybe_wake_tx(dev);
++}
++
++static void rx_refill_timeout(unsigned long data)
++{
++ struct net_device *dev = (struct net_device *)data;
++ struct netfront_info *np = netdev_priv(dev);
++
++ netfront_accelerator_call_stop_napi_irq(np, dev);
++
++ netif_rx_schedule(dev);
++}
++
++static void network_alloc_rx_buffers(struct net_device *dev)
++{
++ unsigned short id;
++ struct netfront_info *np = netdev_priv(dev);
++ struct sk_buff *skb;
++ struct page *page;
++ int i, batch_target, notify;
++ RING_IDX req_prod = np->rx.req_prod_pvt;
++ struct xen_memory_reservation reservation;
++ grant_ref_t ref;
++ unsigned long pfn;
++ void *vaddr;
++ int nr_flips;
++ netif_rx_request_t *req;
++
++ if (unlikely(!netfront_carrier_ok(np)))
++ return;
++
++ /*
++ * Allocate skbuffs greedily, even though we batch updates to the
++ * receive ring. This creates a less bursty demand on the memory
++ * allocator, so should reduce the chance of failed allocation requests
++ * both for ourself and for other kernel subsystems.
++ */
++ batch_target = np->rx_target - (req_prod - np->rx.rsp_cons);
++ for (i = skb_queue_len(&np->rx_batch); i < batch_target; i++) {
++ /*
++ * Allocate an skb and a page. Do not use __dev_alloc_skb as
++ * that will allocate page-sized buffers which is not
++ * necessary here.
++ * 16 bytes added as necessary headroom for netif_receive_skb.
++ */
++ skb = alloc_skb(RX_COPY_THRESHOLD + 16 + NET_IP_ALIGN,
++ GFP_ATOMIC | __GFP_NOWARN);
++ if (unlikely(!skb))
++ goto no_skb;
++
++ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
++ if (!page) {
++ kfree_skb(skb);
++no_skb:
++ /* Any skbuffs queued for refill? Force them out. */
++ if (i != 0)
++ goto refill;
++ /* Could not allocate any skbuffs. Try again later. */
++ mod_timer(&np->rx_refill_timer,
++ jiffies + (HZ/10));
++ break;
++ }
++
++ skb_reserve(skb, 16 + NET_IP_ALIGN); /* mimic dev_alloc_skb() */
++ skb_shinfo(skb)->frags[0].page = page;
++ skb_shinfo(skb)->nr_frags = 1;
++ __skb_queue_tail(&np->rx_batch, skb);
++ }
++
++ /* Is the batch large enough to be worthwhile? */
++ if (i < (np->rx_target/2)) {
++ if (req_prod > np->rx.sring->req_prod)
++ goto push;
++ return;
++ }
++
++ /* Adjust our fill target if we risked running out of buffers. */
++ if (((req_prod - np->rx.sring->rsp_prod) < (np->rx_target / 4)) &&
++ ((np->rx_target *= 2) > np->rx_max_target))
++ np->rx_target = np->rx_max_target;
++
++ refill:
++ for (nr_flips = i = 0; ; i++) {
++ if ((skb = __skb_dequeue(&np->rx_batch)) == NULL)
++ break;
++
++ skb->dev = dev;
++
++ id = xennet_rxidx(req_prod + i);
++
++ BUG_ON(np->rx_skbs[id]);
++ np->rx_skbs[id] = skb;
++
++ ref = gnttab_claim_grant_reference(&np->gref_rx_head);
++ BUG_ON((signed short)ref < 0);
++ np->grant_rx_ref[id] = ref;
++
++ pfn = page_to_pfn(skb_shinfo(skb)->frags[0].page);
++ vaddr = page_address(skb_shinfo(skb)->frags[0].page);
++
++ req = RING_GET_REQUEST(&np->rx, req_prod + i);
++ if (!np->copying_receiver) {
++ gnttab_grant_foreign_transfer_ref(ref,
++ np->xbdev->otherend_id,
++ pfn);
++ np->rx_pfn_array[nr_flips] = pfn_to_mfn(pfn);
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* Remove this page before passing
++ * back to Xen. */
++ set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
++ MULTI_update_va_mapping(np->rx_mcl+i,
++ (unsigned long)vaddr,
++ __pte(0), 0);
++ }
++ nr_flips++;
++ } else {
++ gnttab_grant_foreign_access_ref(ref,
++ np->xbdev->otherend_id,
++ pfn_to_mfn(pfn),
++ 0);
++ }
++
++ req->id = id;
++ req->gref = ref;
++ }
++
++ if ( nr_flips != 0 ) {
++ /* Tell the ballon driver what is going on. */
++ balloon_update_driver_allowance(i);
++
++ set_xen_guest_handle(reservation.extent_start,
++ np->rx_pfn_array);
++ reservation.nr_extents = nr_flips;
++ reservation.extent_order = 0;
++ reservation.address_bits = 0;
++ reservation.domid = DOMID_SELF;
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* After all PTEs have been zapped, flush the TLB. */
++ np->rx_mcl[i-1].args[MULTI_UVMFLAGS_INDEX] =
++ UVMF_TLB_FLUSH|UVMF_ALL;
++
++ /* Give away a batch of pages. */
++ np->rx_mcl[i].op = __HYPERVISOR_memory_op;
++ np->rx_mcl[i].args[0] = XENMEM_decrease_reservation;
++ np->rx_mcl[i].args[1] = (unsigned long)&reservation;
++
++ /* Zap PTEs and give away pages in one big
++ * multicall. */
++ if (unlikely(HYPERVISOR_multicall(np->rx_mcl, i+1)))
++ BUG();
++
++ /* Check return status of HYPERVISOR_memory_op(). */
++ if (unlikely(np->rx_mcl[i].result != i))
++ panic("Unable to reduce memory reservation\n");
++ while (i--)
++ BUG_ON(np->rx_mcl[i].result);
++ } else {
++ if (HYPERVISOR_memory_op(XENMEM_decrease_reservation,
++ &reservation) != i)
++ panic("Unable to reduce memory reservation\n");
++ }
++ } else {
++ wmb();
++ }
++
++ /* Above is a suitable barrier to ensure backend will see requests. */
++ np->rx.req_prod_pvt = req_prod + i;
++ push:
++ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&np->rx, notify);
++ if (notify)
++ notify_remote_via_irq(np->irq);
++}
++
++static void xennet_make_frags(struct sk_buff *skb, struct net_device *dev,
++ struct netif_tx_request *tx)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ char *data = skb->data;
++ unsigned long mfn;
++ RING_IDX prod = np->tx.req_prod_pvt;
++ int frags = skb_shinfo(skb)->nr_frags;
++ unsigned int offset = offset_in_page(data);
++ unsigned int len = skb_headlen(skb);
++ unsigned int id;
++ grant_ref_t ref;
++ int i;
++
++ while (len > PAGE_SIZE - offset) {
++ tx->size = PAGE_SIZE - offset;
++ tx->flags |= NETTXF_more_data;
++ len -= tx->size;
++ data += tx->size;
++ offset = 0;
++
++ id = get_id_from_freelist(np->tx_skbs);
++ np->tx_skbs[id] = skb_get(skb);
++ tx = RING_GET_REQUEST(&np->tx, prod++);
++ tx->id = id;
++ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
++ BUG_ON((signed short)ref < 0);
++
++ mfn = virt_to_mfn(data);
++ gnttab_grant_foreign_access_ref(ref, np->xbdev->otherend_id,
++ mfn, GTF_readonly);
++
++ tx->gref = np->grant_tx_ref[id] = ref;
++ tx->offset = offset;
++ tx->size = len;
++ tx->flags = 0;
++ }
++
++ for (i = 0; i < frags; i++) {
++ skb_frag_t *frag = skb_shinfo(skb)->frags + i;
++
++ tx->flags |= NETTXF_more_data;
++
++ id = get_id_from_freelist(np->tx_skbs);
++ np->tx_skbs[id] = skb_get(skb);
++ tx = RING_GET_REQUEST(&np->tx, prod++);
++ tx->id = id;
++ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
++ BUG_ON((signed short)ref < 0);
++
++ mfn = pfn_to_mfn(page_to_pfn(frag->page));
++ gnttab_grant_foreign_access_ref(ref, np->xbdev->otherend_id,
++ mfn, GTF_readonly);
++
++ tx->gref = np->grant_tx_ref[id] = ref;
++ tx->offset = frag->page_offset;
++ tx->size = frag->size;
++ tx->flags = 0;
++ }
++
++ np->tx.req_prod_pvt = prod;
++}
++
++static int network_start_xmit(struct sk_buff *skb, struct net_device *dev)
++{
++ unsigned short id;
++ struct netfront_info *np = netdev_priv(dev);
++ struct netif_tx_request *tx;
++ struct netif_extra_info *extra;
++ char *data = skb->data;
++ RING_IDX i;
++ grant_ref_t ref;
++ unsigned long mfn;
++ int notify;
++ int frags = skb_shinfo(skb)->nr_frags;
++ unsigned int offset = offset_in_page(data);
++ unsigned int len = skb_headlen(skb);
++
++ /* Check the fast path, if hooks are available */
++ if (np->accel_vif_state.hooks &&
++ np->accel_vif_state.hooks->start_xmit(skb, dev)) {
++ /* Fast path has sent this packet */
++ return 0;
++ }
++
++ frags += (offset + len + PAGE_SIZE - 1) / PAGE_SIZE;
++ if (unlikely(frags > MAX_SKB_FRAGS + 1)) {
++ printk(KERN_ALERT "xennet: skb rides the rocket: %d frags\n",
++ frags);
++ dump_stack();
++ goto drop;
++ }
++
++ spin_lock_irq(&np->tx_lock);
++
++ if (unlikely(!netfront_carrier_ok(np) ||
++ (frags > 1 && !xennet_can_sg(dev)) ||
++ netif_needs_gso(dev, skb))) {
++ spin_unlock_irq(&np->tx_lock);
++ goto drop;
++ }
++
++ i = np->tx.req_prod_pvt;
++
++ id = get_id_from_freelist(np->tx_skbs);
++ np->tx_skbs[id] = skb;
++
++ tx = RING_GET_REQUEST(&np->tx, i);
++
++ tx->id = id;
++ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
++ BUG_ON((signed short)ref < 0);
++ mfn = virt_to_mfn(data);
++ gnttab_grant_foreign_access_ref(
++ ref, np->xbdev->otherend_id, mfn, GTF_readonly);
++ tx->gref = np->grant_tx_ref[id] = ref;
++ tx->offset = offset;
++ tx->size = len;
++
++ tx->flags = 0;
++ extra = NULL;
++
++ if (skb->ip_summed == CHECKSUM_HW) /* local packet? */
++ tx->flags |= NETTXF_csum_blank | NETTXF_data_validated;
++#ifdef CONFIG_XEN
++ if (skb->proto_data_valid) /* remote but checksummed? */
++ tx->flags |= NETTXF_data_validated;
++#endif
++
++#if HAVE_TSO
++ if (skb_shinfo(skb)->gso_size) {
++ struct netif_extra_info *gso = (struct netif_extra_info *)
++ RING_GET_REQUEST(&np->tx, ++i);
++
++ if (extra)
++ extra->flags |= XEN_NETIF_EXTRA_FLAG_MORE;
++ else
++ tx->flags |= NETTXF_extra_info;
++
++ gso->u.gso.size = skb_shinfo(skb)->gso_size;
++ gso->u.gso.type = XEN_NETIF_GSO_TYPE_TCPV4;
++ gso->u.gso.pad = 0;
++ gso->u.gso.features = 0;
++
++ gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
++ gso->flags = 0;
++ extra = gso;
++ }
++#endif
++
++ np->tx.req_prod_pvt = i + 1;
++
++ xennet_make_frags(skb, dev, tx);
++ tx->size = skb->len;
++
++ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&np->tx, notify);
++ if (notify)
++ notify_remote_via_irq(np->irq);
++
++ np->stats.tx_bytes += skb->len;
++ np->stats.tx_packets++;
++ dev->trans_start = jiffies;
++
++ /* Note: It is not safe to access skb after network_tx_buf_gc()! */
++ network_tx_buf_gc(dev);
++
++ if (!netfront_tx_slot_available(np))
++ netif_stop_queue(dev);
++
++ spin_unlock_irq(&np->tx_lock);
++
++ return 0;
++
++ drop:
++ np->stats.tx_dropped++;
++ dev_kfree_skb(skb);
++ return 0;
++}
++
++static irqreturn_t netif_int(int irq, void *dev_id, struct pt_regs *ptregs)
++{
++ struct net_device *dev = dev_id;
++ struct netfront_info *np = netdev_priv(dev);
++ unsigned long flags;
++
++ spin_lock_irqsave(&np->tx_lock, flags);
++
++ if (likely(netfront_carrier_ok(np))) {
++ network_tx_buf_gc(dev);
++ /* Under tx_lock: protects access to rx shared-ring indexes. */
++ if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)) {
++ netfront_accelerator_call_stop_napi_irq(np, dev);
++
++ netif_rx_schedule(dev);
++ dev->last_rx = jiffies;
++ }
++ }
++
++ spin_unlock_irqrestore(&np->tx_lock, flags);
++
++ return IRQ_HANDLED;
++}
++
++static void xennet_move_rx_slot(struct netfront_info *np, struct sk_buff *skb,
++ grant_ref_t ref)
++{
++ int new = xennet_rxidx(np->rx.req_prod_pvt);
++
++ BUG_ON(np->rx_skbs[new]);
++ np->rx_skbs[new] = skb;
++ np->grant_rx_ref[new] = ref;
++ RING_GET_REQUEST(&np->rx, np->rx.req_prod_pvt)->id = new;
++ RING_GET_REQUEST(&np->rx, np->rx.req_prod_pvt)->gref = ref;
++ np->rx.req_prod_pvt++;
++}
++
++int xennet_get_extras(struct netfront_info *np,
++ struct netif_extra_info *extras, RING_IDX rp)
++
++{
++ struct netif_extra_info *extra;
++ RING_IDX cons = np->rx.rsp_cons;
++ int err = 0;
++
++ do {
++ struct sk_buff *skb;
++ grant_ref_t ref;
++
++ if (unlikely(cons + 1 == rp)) {
++ if (net_ratelimit())
++ WPRINTK("Missing extra info\n");
++ err = -EBADR;
++ break;
++ }
++
++ extra = (struct netif_extra_info *)
++ RING_GET_RESPONSE(&np->rx, ++cons);
++
++ if (unlikely(!extra->type ||
++ extra->type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
++ if (net_ratelimit())
++ WPRINTK("Invalid extra type: %d\n",
++ extra->type);
++ err = -EINVAL;
++ } else {
++ memcpy(&extras[extra->type - 1], extra,
++ sizeof(*extra));
++ }
++
++ skb = xennet_get_rx_skb(np, cons);
++ ref = xennet_get_rx_ref(np, cons);
++ xennet_move_rx_slot(np, skb, ref);
++ } while (extra->flags & XEN_NETIF_EXTRA_FLAG_MORE);
++
++ np->rx.rsp_cons = cons;
++ return err;
++}
++
++static int xennet_get_responses(struct netfront_info *np,
++ struct netfront_rx_info *rinfo, RING_IDX rp,
++ struct sk_buff_head *list,
++ int *pages_flipped_p)
++{
++ int pages_flipped = *pages_flipped_p;
++ struct mmu_update *mmu;
++ struct multicall_entry *mcl;
++ struct netif_rx_response *rx = &rinfo->rx;
++ struct netif_extra_info *extras = rinfo->extras;
++ RING_IDX cons = np->rx.rsp_cons;
++ struct sk_buff *skb = xennet_get_rx_skb(np, cons);
++ grant_ref_t ref = xennet_get_rx_ref(np, cons);
++ int max = MAX_SKB_FRAGS + (rx->status <= RX_COPY_THRESHOLD);
++ int frags = 1;
++ int err = 0;
++ unsigned long ret;
++
++ if (rx->flags & NETRXF_extra_info) {
++ err = xennet_get_extras(np, extras, rp);
++ cons = np->rx.rsp_cons;
++ }
++
++ for (;;) {
++ unsigned long mfn;
++
++ if (unlikely(rx->status < 0 ||
++ rx->offset + rx->status > PAGE_SIZE)) {
++ if (net_ratelimit())
++ WPRINTK("rx->offset: %x, size: %u\n",
++ rx->offset, rx->status);
++ xennet_move_rx_slot(np, skb, ref);
++ err = -EINVAL;
++ goto next;
++ }
++
++ /*
++ * This definitely indicates a bug, either in this driver or in
++ * the backend driver. In future this should flag the bad
++ * situation to the system controller to reboot the backed.
++ */
++ if (ref == GRANT_INVALID_REF) {
++ if (net_ratelimit())
++ WPRINTK("Bad rx response id %d.\n", rx->id);
++ err = -EINVAL;
++ goto next;
++ }
++
++ if (!np->copying_receiver) {
++ /* Memory pressure, insufficient buffer
++ * headroom, ... */
++ if (!(mfn = gnttab_end_foreign_transfer_ref(ref))) {
++ if (net_ratelimit())
++ WPRINTK("Unfulfilled rx req "
++ "(id=%d, st=%d).\n",
++ rx->id, rx->status);
++ xennet_move_rx_slot(np, skb, ref);
++ err = -ENOMEM;
++ goto next;
++ }
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* Remap the page. */
++ struct page *page =
++ skb_shinfo(skb)->frags[0].page;
++ unsigned long pfn = page_to_pfn(page);
++ void *vaddr = page_address(page);
++
++ mcl = np->rx_mcl + pages_flipped;
++ mmu = np->rx_mmu + pages_flipped;
++
++ MULTI_update_va_mapping(mcl,
++ (unsigned long)vaddr,
++ pfn_pte_ma(mfn,
++ PAGE_KERNEL),
++ 0);
++ mmu->ptr = ((maddr_t)mfn << PAGE_SHIFT)
++ | MMU_MACHPHYS_UPDATE;
++ mmu->val = pfn;
++
++ set_phys_to_machine(pfn, mfn);
++ }
++ pages_flipped++;
++ } else {
++ ret = gnttab_end_foreign_access_ref(ref);
++ BUG_ON(!ret);
++ }
++
++ gnttab_release_grant_reference(&np->gref_rx_head, ref);
++
++ __skb_queue_tail(list, skb);
++
++next:
++ if (!(rx->flags & NETRXF_more_data))
++ break;
++
++ if (cons + frags == rp) {
++ if (net_ratelimit())
++ WPRINTK("Need more frags\n");
++ err = -ENOENT;
++ break;
++ }
++
++ rx = RING_GET_RESPONSE(&np->rx, cons + frags);
++ skb = xennet_get_rx_skb(np, cons + frags);
++ ref = xennet_get_rx_ref(np, cons + frags);
++ frags++;
++ }
++
++ if (unlikely(frags > max)) {
++ if (net_ratelimit())
++ WPRINTK("Too many frags\n");
++ err = -E2BIG;
++ }
++
++ if (unlikely(err))
++ np->rx.rsp_cons = cons + frags;
++
++ *pages_flipped_p = pages_flipped;
++
++ return err;
++}
++
++static RING_IDX xennet_fill_frags(struct netfront_info *np,
++ struct sk_buff *skb,
++ struct sk_buff_head *list)
++{
++ struct skb_shared_info *shinfo = skb_shinfo(skb);
++ int nr_frags = shinfo->nr_frags;
++ RING_IDX cons = np->rx.rsp_cons;
++ skb_frag_t *frag = shinfo->frags + nr_frags;
++ struct sk_buff *nskb;
++
++ while ((nskb = __skb_dequeue(list))) {
++ struct netif_rx_response *rx =
++ RING_GET_RESPONSE(&np->rx, ++cons);
++
++ frag->page = skb_shinfo(nskb)->frags[0].page;
++ frag->page_offset = rx->offset;
++ frag->size = rx->status;
++
++ skb->data_len += rx->status;
++
++ skb_shinfo(nskb)->nr_frags = 0;
++ kfree_skb(nskb);
++
++ frag++;
++ nr_frags++;
++ }
++
++ shinfo->nr_frags = nr_frags;
++ return cons;
++}
++
++static int xennet_set_skb_gso(struct sk_buff *skb,
++ struct netif_extra_info *gso)
++{
++ if (!gso->u.gso.size) {
++ if (net_ratelimit())
++ WPRINTK("GSO size must not be zero.\n");
++ return -EINVAL;
++ }
++
++ /* Currently only TCPv4 S.O. is supported. */
++ if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4) {
++ if (net_ratelimit())
++ WPRINTK("Bad GSO type %d.\n", gso->u.gso.type);
++ return -EINVAL;
++ }
++
++#if HAVE_TSO
++ skb_shinfo(skb)->gso_size = gso->u.gso.size;
++#if HAVE_GSO
++ skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
++
++ /* Header must be checked, and gso_segs computed. */
++ skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
++#endif
++ skb_shinfo(skb)->gso_segs = 0;
++
++ return 0;
++#else
++ if (net_ratelimit())
++ WPRINTK("GSO unsupported by this kernel.\n");
++ return -EINVAL;
++#endif
++}
++
++static int netif_poll(struct net_device *dev, int *pbudget)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ struct sk_buff *skb;
++ struct netfront_rx_info rinfo;
++ struct netif_rx_response *rx = &rinfo.rx;
++ struct netif_extra_info *extras = rinfo.extras;
++ RING_IDX i, rp;
++ struct multicall_entry *mcl;
++ int work_done, budget, more_to_do = 1, accel_more_to_do = 1;
++ struct sk_buff_head rxq;
++ struct sk_buff_head errq;
++ struct sk_buff_head tmpq;
++ unsigned long flags;
++ unsigned int len;
++ int pages_flipped = 0;
++ int err;
++
++ spin_lock(&np->rx_lock); /* no need for spin_lock_bh() in ->poll() */
++
++ if (unlikely(!netfront_carrier_ok(np))) {
++ spin_unlock(&np->rx_lock);
++ return 0;
++ }
++
++ skb_queue_head_init(&rxq);
++ skb_queue_head_init(&errq);
++ skb_queue_head_init(&tmpq);
++
++ if ((budget = *pbudget) > dev->quota)
++ budget = dev->quota;
++ rp = np->rx.sring->rsp_prod;
++ rmb(); /* Ensure we see queued responses up to 'rp'. */
++
++ i = np->rx.rsp_cons;
++ work_done = 0;
++ while ((i != rp) && (work_done < budget)) {
++ memcpy(rx, RING_GET_RESPONSE(&np->rx, i), sizeof(*rx));
++ memset(extras, 0, sizeof(rinfo.extras));
++
++ err = xennet_get_responses(np, &rinfo, rp, &tmpq,
++ &pages_flipped);
++
++ if (unlikely(err)) {
++err:
++ while ((skb = __skb_dequeue(&tmpq)))
++ __skb_queue_tail(&errq, skb);
++ np->stats.rx_errors++;
++ i = np->rx.rsp_cons;
++ continue;
++ }
++
++ skb = __skb_dequeue(&tmpq);
++
++ if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
++ struct netif_extra_info *gso;
++ gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
++
++ if (unlikely(xennet_set_skb_gso(skb, gso))) {
++ __skb_queue_head(&tmpq, skb);
++ np->rx.rsp_cons += skb_queue_len(&tmpq);
++ goto err;
++ }
++ }
++
++ NETFRONT_SKB_CB(skb)->page = skb_shinfo(skb)->frags[0].page;
++ NETFRONT_SKB_CB(skb)->offset = rx->offset;
++
++ len = rx->status;
++ if (len > RX_COPY_THRESHOLD)
++ len = RX_COPY_THRESHOLD;
++ skb_put(skb, len);
++
++ if (rx->status > len) {
++ skb_shinfo(skb)->frags[0].page_offset =
++ rx->offset + len;
++ skb_shinfo(skb)->frags[0].size = rx->status - len;
++ skb->data_len = rx->status - len;
++ } else {
++ skb_shinfo(skb)->frags[0].page = NULL;
++ skb_shinfo(skb)->nr_frags = 0;
++ }
++
++ i = xennet_fill_frags(np, skb, &tmpq);
++
++ /*
++ * Truesize must approximates the size of true data plus
++ * any supervisor overheads. Adding hypervisor overheads
++ * has been shown to significantly reduce achievable
++ * bandwidth with the default receive buffer size. It is
++ * therefore not wise to account for it here.
++ *
++ * After alloc_skb(RX_COPY_THRESHOLD), truesize is set to
++ * RX_COPY_THRESHOLD + the supervisor overheads. Here, we
++ * add the size of the data pulled in xennet_fill_frags().
++ *
++ * We also adjust for any unused space in the main data
++ * area by subtracting (RX_COPY_THRESHOLD - len). This is
++ * especially important with drivers which split incoming
++ * packets into header and data, using only 66 bytes of
++ * the main data area (see the e1000 driver for example.)
++ * On such systems, without this last adjustement, our
++ * achievable receive throughout using the standard receive
++ * buffer size was cut by 25%(!!!).
++ */
++ skb->truesize += skb->data_len - (RX_COPY_THRESHOLD - len);
++ skb->len += skb->data_len;
++
++ /*
++ * Old backends do not assert data_validated but we
++ * can infer it from csum_blank so test both flags.
++ */
++ if (rx->flags & (NETRXF_data_validated|NETRXF_csum_blank))
++ skb->ip_summed = CHECKSUM_UNNECESSARY;
++ else
++ skb->ip_summed = CHECKSUM_NONE;
++#ifdef CONFIG_XEN
++ skb->proto_data_valid = (skb->ip_summed != CHECKSUM_NONE);
++ skb->proto_csum_blank = !!(rx->flags & NETRXF_csum_blank);
++#endif
++ np->stats.rx_packets++;
++ np->stats.rx_bytes += skb->len;
++
++ __skb_queue_tail(&rxq, skb);
++
++ np->rx.rsp_cons = ++i;
++ work_done++;
++ }
++
++ if (pages_flipped) {
++ /* Some pages are no longer absent... */
++ balloon_update_driver_allowance(-pages_flipped);
++
++ /* Do all the remapping work and M2P updates. */
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ mcl = np->rx_mcl + pages_flipped;
++ mcl->op = __HYPERVISOR_mmu_update;
++ mcl->args[0] = (unsigned long)np->rx_mmu;
++ mcl->args[1] = pages_flipped;
++ mcl->args[2] = 0;
++ mcl->args[3] = DOMID_SELF;
++ err = HYPERVISOR_multicall_check(np->rx_mcl,
++ pages_flipped + 1,
++ NULL);
++ BUG_ON(err);
++ }
++ }
++
++ while ((skb = __skb_dequeue(&errq)))
++ kfree_skb(skb);
++
++ while ((skb = __skb_dequeue(&rxq)) != NULL) {
++ struct page *page = NETFRONT_SKB_CB(skb)->page;
++ void *vaddr = page_address(page);
++ unsigned offset = NETFRONT_SKB_CB(skb)->offset;
++
++ memcpy(skb->data, vaddr + offset, skb_headlen(skb));
++
++ if (page != skb_shinfo(skb)->frags[0].page)
++ __free_page(page);
++
++ /* Ethernet work: Delayed to here as it peeks the header. */
++ skb->protocol = eth_type_trans(skb, dev);
++
++ /* Pass it up. */
++ netif_receive_skb(skb);
++ dev->last_rx = jiffies;
++ }
++
++ /* If we get a callback with very few responses, reduce fill target. */
++ /* NB. Note exponential increase, linear decrease. */
++ if (((np->rx.req_prod_pvt - np->rx.sring->rsp_prod) >
++ ((3*np->rx_target) / 4)) &&
++ (--np->rx_target < np->rx_min_target))
++ np->rx_target = np->rx_min_target;
++
++ network_alloc_rx_buffers(dev);
++
++ if (work_done < budget) {
++ /* there's some spare capacity, try the accelerated path */
++ int accel_budget = budget - work_done;
++ int accel_budget_start = accel_budget;
++
++ if (np->accel_vif_state.hooks) {
++ accel_more_to_do =
++ np->accel_vif_state.hooks->netdev_poll
++ (dev, &accel_budget);
++ work_done += (accel_budget_start - accel_budget);
++ } else
++ accel_more_to_do = 0;
++ }
++
++ *pbudget -= work_done;
++ dev->quota -= work_done;
++
++ if (work_done < budget) {
++ local_irq_save(flags);
++
++ RING_FINAL_CHECK_FOR_RESPONSES(&np->rx, more_to_do);
++
++ if (!more_to_do && !accel_more_to_do &&
++ np->accel_vif_state.hooks) {
++ /*
++ * Slow path has nothing more to do, see if
++ * fast path is likewise
++ */
++ accel_more_to_do =
++ np->accel_vif_state.hooks->start_napi_irq(dev);
++ }
++
++ if (!more_to_do && !accel_more_to_do)
++ __netif_rx_complete(dev);
++
++ local_irq_restore(flags);
++ }
++
++ spin_unlock(&np->rx_lock);
++
++ return more_to_do | accel_more_to_do;
++}
++
++static void netif_release_tx_bufs(struct netfront_info *np)
++{
++ struct sk_buff *skb;
++ int i;
++
++ for (i = 1; i <= NET_TX_RING_SIZE; i++) {
++ if ((unsigned long)np->tx_skbs[i] < PAGE_OFFSET)
++ continue;
++
++ skb = np->tx_skbs[i];
++ gnttab_end_foreign_access_ref(np->grant_tx_ref[i]);
++ gnttab_release_grant_reference(
++ &np->gref_tx_head, np->grant_tx_ref[i]);
++ np->grant_tx_ref[i] = GRANT_INVALID_REF;
++ add_id_to_freelist(np->tx_skbs, i);
++ dev_kfree_skb_irq(skb);
++ }
++}
++
++static void netif_release_rx_bufs_flip(struct netfront_info *np)
++{
++ struct mmu_update *mmu = np->rx_mmu;
++ struct multicall_entry *mcl = np->rx_mcl;
++ struct sk_buff_head free_list;
++ struct sk_buff *skb;
++ unsigned long mfn;
++ int xfer = 0, noxfer = 0, unused = 0;
++ int id, ref, rc;
++
++ skb_queue_head_init(&free_list);
++
++ spin_lock_bh(&np->rx_lock);
++
++ for (id = 0; id < NET_RX_RING_SIZE; id++) {
++ if ((ref = np->grant_rx_ref[id]) == GRANT_INVALID_REF) {
++ unused++;
++ continue;
++ }
++
++ skb = np->rx_skbs[id];
++ mfn = gnttab_end_foreign_transfer_ref(ref);
++ gnttab_release_grant_reference(&np->gref_rx_head, ref);
++ np->grant_rx_ref[id] = GRANT_INVALID_REF;
++ add_id_to_freelist(np->rx_skbs, id);
++
++ if (0 == mfn) {
++ struct page *page = skb_shinfo(skb)->frags[0].page;
++ balloon_release_driver_page(page);
++ skb_shinfo(skb)->nr_frags = 0;
++ dev_kfree_skb(skb);
++ noxfer++;
++ continue;
++ }
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* Remap the page. */
++ struct page *page = skb_shinfo(skb)->frags[0].page;
++ unsigned long pfn = page_to_pfn(page);
++ void *vaddr = page_address(page);
++
++ MULTI_update_va_mapping(mcl, (unsigned long)vaddr,
++ pfn_pte_ma(mfn, PAGE_KERNEL),
++ 0);
++ mcl++;
++ mmu->ptr = ((maddr_t)mfn << PAGE_SHIFT)
++ | MMU_MACHPHYS_UPDATE;
++ mmu->val = pfn;
++ mmu++;
++
++ set_phys_to_machine(pfn, mfn);
++ }
++ __skb_queue_tail(&free_list, skb);
++ xfer++;
++ }
++
++ DPRINTK("%s: %d xfer, %d noxfer, %d unused\n",
++ __FUNCTION__, xfer, noxfer, unused);
++
++ if (xfer) {
++ /* Some pages are no longer absent... */
++ balloon_update_driver_allowance(-xfer);
++
++ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++ /* Do all the remapping work and M2P updates. */
++ mcl->op = __HYPERVISOR_mmu_update;
++ mcl->args[0] = (unsigned long)np->rx_mmu;
++ mcl->args[1] = mmu - np->rx_mmu;
++ mcl->args[2] = 0;
++ mcl->args[3] = DOMID_SELF;
++ mcl++;
++ rc = HYPERVISOR_multicall_check(
++ np->rx_mcl, mcl - np->rx_mcl, NULL);
++ BUG_ON(rc);
++ }
++ }
++
++ while ((skb = __skb_dequeue(&free_list)) != NULL)
++ dev_kfree_skb(skb);
++
++ spin_unlock_bh(&np->rx_lock);
++}
++
++static void netif_release_rx_bufs_copy(struct netfront_info *np)
++{
++ struct sk_buff *skb;
++ int i, ref;
++ int busy = 0, inuse = 0;
++
++ spin_lock_bh(&np->rx_lock);
++
++ for (i = 0; i < NET_RX_RING_SIZE; i++) {
++ ref = np->grant_rx_ref[i];
++
++ if (ref == GRANT_INVALID_REF)
++ continue;
++
++ inuse++;
++
++ skb = np->rx_skbs[i];
++
++ if (!gnttab_end_foreign_access_ref(ref))
++ {
++ busy++;
++ continue;
++ }
++
++ gnttab_release_grant_reference(&np->gref_rx_head, ref);
++ np->grant_rx_ref[i] = GRANT_INVALID_REF;
++ add_id_to_freelist(np->rx_skbs, i);
++
++ skb_shinfo(skb)->nr_frags = 0;
++ dev_kfree_skb(skb);
++ }
++
++ if (busy)
++ DPRINTK("%s: Unable to release %d of %d inuse grant references out of %ld total.\n",
++ __FUNCTION__, busy, inuse, NET_RX_RING_SIZE);
++
++ spin_unlock_bh(&np->rx_lock);
++}
++
++static int network_close(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ netif_stop_queue(np->netdev);
++ return 0;
++}
++
++
++static struct net_device_stats *network_get_stats(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++
++ netfront_accelerator_call_get_stats(np, dev);
++ return &np->stats;
++}
++
++static int xennet_set_mac_address(struct net_device *dev, void *p)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ struct sockaddr *addr = p;
++
++ if (netif_running(dev))
++ return -EBUSY;
++
++ if (!is_valid_ether_addr(addr->sa_data))
++ return -EADDRNOTAVAIL;
++
++ memcpy(dev->dev_addr, addr->sa_data, dev->addr_len);
++ memcpy(np->mac, addr->sa_data, ETH_ALEN);
++
++ return 0;
++}
++
++static int xennet_change_mtu(struct net_device *dev, int mtu)
++{
++ int max = xennet_can_sg(dev) ? 65535 - ETH_HLEN : ETH_DATA_LEN;
++
++ if (mtu > max)
++ return -EINVAL;
++ dev->mtu = mtu;
++ return 0;
++}
++
++static int xennet_set_sg(struct net_device *dev, u32 data)
++{
++ if (data) {
++ struct netfront_info *np = netdev_priv(dev);
++ int val;
++
++ if (xenbus_scanf(XBT_NIL, np->xbdev->otherend, "feature-sg",
++ "%d", &val) < 0)
++ val = 0;
++ if (!val)
++ return -ENOSYS;
++ } else if (dev->mtu > ETH_DATA_LEN)
++ dev->mtu = ETH_DATA_LEN;
++
++ return ethtool_op_set_sg(dev, data);
++}
++
++static int xennet_set_tso(struct net_device *dev, u32 data)
++{
++ if (data) {
++ struct netfront_info *np = netdev_priv(dev);
++ int val;
++
++ if (xenbus_scanf(XBT_NIL, np->xbdev->otherend,
++ "feature-gso-tcpv4", "%d", &val) < 0)
++ val = 0;
++ if (!val)
++ return -ENOSYS;
++ }
++
++ return ethtool_op_set_tso(dev, data);
++}
++
++static void xennet_set_features(struct net_device *dev)
++{
++ dev_disable_gso_features(dev);
++ xennet_set_sg(dev, 0);
++
++ /* We need checksum offload to enable scatter/gather and TSO. */
++ if (!(dev->features & NETIF_F_IP_CSUM))
++ return;
++
++ if (xennet_set_sg(dev, 1))
++ return;
++
++ /* Before 2.6.9 TSO seems to be unreliable so do not enable it
++ * on older kernels.
++ */
++ if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,9))
++ xennet_set_tso(dev, 1);
++}
++
++static int network_connect(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ int i, requeue_idx, err;
++ struct sk_buff *skb;
++ grant_ref_t ref;
++ netif_rx_request_t *req;
++ unsigned int feature_rx_copy, feature_rx_flip;
++
++ err = xenbus_scanf(XBT_NIL, np->xbdev->otherend,
++ "feature-rx-copy", "%u", &feature_rx_copy);
++ if (err != 1)
++ feature_rx_copy = 0;
++ err = xenbus_scanf(XBT_NIL, np->xbdev->otherend,
++ "feature-rx-flip", "%u", &feature_rx_flip);
++ if (err != 1)
++ feature_rx_flip = 1;
++
++ /*
++ * Copy packets on receive path if:
++ * (a) This was requested by user, and the backend supports it; or
++ * (b) Flipping was requested, but this is unsupported by the backend.
++ */
++ np->copying_receiver = ((MODPARM_rx_copy && feature_rx_copy) ||
++ (MODPARM_rx_flip && !feature_rx_flip));
++
++ err = talk_to_backend(np->xbdev, np);
++ if (err)
++ return err;
++
++ xennet_set_features(dev);
++
++ DPRINTK("device %s has %sing receive path.\n",
++ dev->name, np->copying_receiver ? "copy" : "flipp");
++
++ spin_lock_bh(&np->rx_lock);
++ spin_lock_irq(&np->tx_lock);
++
++ /*
++ * Recovery procedure:
++ * NB. Freelist index entries are always going to be less than
++ * PAGE_OFFSET, whereas pointers to skbs will always be equal or
++ * greater than PAGE_OFFSET: we use this property to distinguish
++ * them.
++ */
++
++ /* Step 1: Discard all pending TX packet fragments. */
++ netif_release_tx_bufs(np);
++
++ /* Step 2: Rebuild the RX buffer freelist and the RX ring itself. */
++ for (requeue_idx = 0, i = 0; i < NET_RX_RING_SIZE; i++) {
++ if (!np->rx_skbs[i])
++ continue;
++
++ skb = np->rx_skbs[requeue_idx] = xennet_get_rx_skb(np, i);
++ ref = np->grant_rx_ref[requeue_idx] = xennet_get_rx_ref(np, i);
++ req = RING_GET_REQUEST(&np->rx, requeue_idx);
++
++ if (!np->copying_receiver) {
++ gnttab_grant_foreign_transfer_ref(
++ ref, np->xbdev->otherend_id,
++ page_to_pfn(skb_shinfo(skb)->frags->page));
++ } else {
++ gnttab_grant_foreign_access_ref(
++ ref, np->xbdev->otherend_id,
++ pfn_to_mfn(page_to_pfn(skb_shinfo(skb)->
++ frags->page)),
++ 0);
++ }
++ req->gref = ref;
++ req->id = requeue_idx;
++
++ requeue_idx++;
++ }
++
++ np->rx.req_prod_pvt = requeue_idx;
++
++ /*
++ * Step 3: All public and private state should now be sane. Get
++ * ready to start sending and receiving packets and give the driver
++ * domain a kick because we've probably just requeued some
++ * packets.
++ */
++ netfront_carrier_on(np);
++ notify_remote_via_irq(np->irq);
++ network_tx_buf_gc(dev);
++ network_alloc_rx_buffers(dev);
++
++ spin_unlock_irq(&np->tx_lock);
++ spin_unlock_bh(&np->rx_lock);
++
++ return 0;
++}
++
++static void netif_uninit(struct net_device *dev)
++{
++ struct netfront_info *np = netdev_priv(dev);
++ netif_release_tx_bufs(np);
++ if (np->copying_receiver)
++ netif_release_rx_bufs_copy(np);
++ else
++ netif_release_rx_bufs_flip(np);
++ gnttab_free_grant_references(np->gref_tx_head);
++ gnttab_free_grant_references(np->gref_rx_head);
++}
++
++static struct ethtool_ops network_ethtool_ops =
++{
++ .get_tx_csum = ethtool_op_get_tx_csum,
++ .set_tx_csum = ethtool_op_set_tx_csum,
++ .get_sg = ethtool_op_get_sg,
++ .set_sg = xennet_set_sg,
++#if HAVE_TSO
++ .get_tso = ethtool_op_get_tso,
++ .set_tso = xennet_set_tso,
++#endif
++ .get_link = ethtool_op_get_link,
++};
++
++#ifdef CONFIG_SYSFS
++static ssize_t show_rxbuf_min(struct class_device *cd, char *buf)
++{
++ struct net_device *netdev = container_of(cd, struct net_device,
++ class_dev);
++ struct netfront_info *info = netdev_priv(netdev);
++
++ return sprintf(buf, "%u\n", info->rx_min_target);
++}
++
++static ssize_t store_rxbuf_min(struct class_device *cd,
++ const char *buf, size_t len)
++{
++ struct net_device *netdev = container_of(cd, struct net_device,
++ class_dev);
++ struct netfront_info *np = netdev_priv(netdev);
++ char *endp;
++ unsigned long target;
++
++ if (!capable(CAP_NET_ADMIN))
++ return -EPERM;
++
++ target = simple_strtoul(buf, &endp, 0);
++ if (endp == buf)
++ return -EBADMSG;
++
++ if (target < RX_MIN_TARGET)
++ target = RX_MIN_TARGET;
++ if (target > RX_MAX_TARGET)
++ target = RX_MAX_TARGET;
++
++ spin_lock_bh(&np->rx_lock);
++ if (target > np->rx_max_target)
++ np->rx_max_target = target;
++ np->rx_min_target = target;
++ if (target > np->rx_target)
++ np->rx_target = target;
++
++ network_alloc_rx_buffers(netdev);
++
++ spin_unlock_bh(&np->rx_lock);
++ return len;
++}
++
++static ssize_t show_rxbuf_max(struct class_device *cd, char *buf)
++{
++ struct net_device *netdev = container_of(cd, struct net_device,
++ class_dev);
++ struct netfront_info *info = netdev_priv(netdev);
++
++ return sprintf(buf, "%u\n", info->rx_max_target);
++}
++
++static ssize_t store_rxbuf_max(struct class_device *cd,
++ const char *buf, size_t len)
++{
++ struct net_device *netdev = container_of(cd, struct net_device,
++ class_dev);
++ struct netfront_info *np = netdev_priv(netdev);
++ char *endp;
++ unsigned long target;
++
++ if (!capable(CAP_NET_ADMIN))
++ return -EPERM;
++
++ target = simple_strtoul(buf, &endp, 0);
++ if (endp == buf)
++ return -EBADMSG;
++
++ if (target < RX_MIN_TARGET)
++ target = RX_MIN_TARGET;
++ if (target > RX_MAX_TARGET)
++ target = RX_MAX_TARGET;
++
++ spin_lock_bh(&np->rx_lock);
++ if (target < np->rx_min_target)
++ np->rx_min_target = target;
++ np->rx_max_target = target;
++ if (target < np->rx_target)
++ np->rx_target = target;
++
++ network_alloc_rx_buffers(netdev);
++
++ spin_unlock_bh(&np->rx_lock);
++ return len;
++}
++
++static ssize_t show_rxbuf_cur(struct class_device *cd, char *buf)
++{
++ struct net_device *netdev = container_of(cd, struct net_device,
++ class_dev);
++ struct netfront_info *info = netdev_priv(netdev);
++
++ return sprintf(buf, "%u\n", info->rx_target);
++}
++
++static const struct class_device_attribute xennet_attrs[] = {
++ __ATTR(rxbuf_min, S_IRUGO|S_IWUSR, show_rxbuf_min, store_rxbuf_min),
++ __ATTR(rxbuf_max, S_IRUGO|S_IWUSR, show_rxbuf_max, store_rxbuf_max),
++ __ATTR(rxbuf_cur, S_IRUGO, show_rxbuf_cur, NULL),
++};
++
++static int xennet_sysfs_addif(struct net_device *netdev)
++{
++ int i;
++ int error = 0;
++
++ for (i = 0; i < ARRAY_SIZE(xennet_attrs); i++) {
++ error = class_device_create_file(&netdev->class_dev,
++ &xennet_attrs[i]);
++ if (error)
++ goto fail;
++ }
++ return 0;
++
++ fail:
++ while (--i >= 0)
++ class_device_remove_file(&netdev->class_dev,
++ &xennet_attrs[i]);
++ return error;
++}
++
++static void xennet_sysfs_delif(struct net_device *netdev)
++{
++ int i;
++
++ for (i = 0; i < ARRAY_SIZE(xennet_attrs); i++) {
++ class_device_remove_file(&netdev->class_dev,
++ &xennet_attrs[i]);
++ }
++}
++
++#endif /* CONFIG_SYSFS */
++
++
++/*
++ * Nothing to do here. Virtual interface is point-to-point and the
++ * physical interface is probably promiscuous anyway.
++ */
++static void network_set_multicast_list(struct net_device *dev)
++{
++}
++
++static struct net_device * __devinit create_netdev(struct xenbus_device *dev)
++{
++ int i, err = 0;
++ struct net_device *netdev = NULL;
++ struct netfront_info *np = NULL;
++
++ netdev = alloc_etherdev(sizeof(struct netfront_info));
++ if (!netdev) {
++ printk(KERN_WARNING "%s> alloc_etherdev failed.\n",
++ __FUNCTION__);
++ return ERR_PTR(-ENOMEM);
++ }
++
++ np = netdev_priv(netdev);
++ np->xbdev = dev;
++
++ spin_lock_init(&np->tx_lock);
++ spin_lock_init(&np->rx_lock);
++
++ init_accelerator_vif(np, dev);
++
++ skb_queue_head_init(&np->rx_batch);
++ np->rx_target = RX_DFL_MIN_TARGET;
++ np->rx_min_target = RX_DFL_MIN_TARGET;
++ np->rx_max_target = RX_MAX_TARGET;
++
++ init_timer(&np->rx_refill_timer);
++ np->rx_refill_timer.data = (unsigned long)netdev;
++ np->rx_refill_timer.function = rx_refill_timeout;
++
++ /* Initialise {tx,rx}_skbs as a free chain containing every entry. */
++ for (i = 0; i <= NET_TX_RING_SIZE; i++) {
++ np->tx_skbs[i] = (void *)((unsigned long) i+1);
++ np->grant_tx_ref[i] = GRANT_INVALID_REF;
++ }
++
++ for (i = 0; i < NET_RX_RING_SIZE; i++) {
++ np->rx_skbs[i] = NULL;
++ np->grant_rx_ref[i] = GRANT_INVALID_REF;
++ }
++
++ /* A grant for every tx ring slot */
++ if (gnttab_alloc_grant_references(TX_MAX_TARGET,
++ &np->gref_tx_head) < 0) {
++ printk(KERN_ALERT "#### netfront can't alloc tx grant refs\n");
++ err = -ENOMEM;
++ goto exit;
++ }
++ /* A grant for every rx ring slot */
++ if (gnttab_alloc_grant_references(RX_MAX_TARGET,
++ &np->gref_rx_head) < 0) {
++ printk(KERN_ALERT "#### netfront can't alloc rx grant refs\n");
++ err = -ENOMEM;
++ goto exit_free_tx;
++ }
++
++ netdev->open = network_open;
++ netdev->hard_start_xmit = network_start_xmit;
++ netdev->stop = network_close;
++ netdev->get_stats = network_get_stats;
++ netdev->poll = netif_poll;
++ netdev->set_multicast_list = network_set_multicast_list;
++ netdev->uninit = netif_uninit;
++ netdev->set_mac_address = xennet_set_mac_address;
++ netdev->change_mtu = xennet_change_mtu;
++ netdev->weight = 64;
++ netdev->features = NETIF_F_IP_CSUM;
++
++ SET_ETHTOOL_OPS(netdev, &network_ethtool_ops);
++ SET_MODULE_OWNER(netdev);
++ SET_NETDEV_DEV(netdev, &dev->dev);
++
++ np->netdev = netdev;
++
++ netfront_carrier_off(np);
++
++ return netdev;
++
++ exit_free_tx:
++ gnttab_free_grant_references(np->gref_tx_head);
++ exit:
++ free_netdev(netdev);
++ return ERR_PTR(err);
++}
++
++#ifdef CONFIG_INET
++/*
++ * We use this notifier to send out a fake ARP reply to reset switches and
++ * router ARP caches when an IP interface is brought up on a VIF.
++ */
++static int
++inetdev_notify(struct notifier_block *this, unsigned long event, void *ptr)
++{
++ struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
++ struct net_device *dev = ifa->ifa_dev->dev;
++
++ /* UP event and is it one of our devices? */
++ if (event == NETDEV_UP && dev->open == network_open)
++ send_fake_arp(dev);
++
++ return NOTIFY_DONE;
++}
++
++static struct notifier_block notifier_inetdev = {
++ .notifier_call = inetdev_notify,
++ .next = NULL,
++ .priority = 0
++};
++#endif
++
++
++static void netif_disconnect_backend(struct netfront_info *info)
++{
++ /* Stop old i/f to prevent errors whilst we rebuild the state. */
++ spin_lock_bh(&info->rx_lock);
++ spin_lock_irq(&info->tx_lock);
++ netfront_carrier_off(info);
++ spin_unlock_irq(&info->tx_lock);
++ spin_unlock_bh(&info->rx_lock);
++
++ if (info->irq)
++ unbind_from_irqhandler(info->irq, info->netdev);
++ info->irq = 0;
++
++ end_access(info->tx_ring_ref, info->tx.sring);
++ end_access(info->rx_ring_ref, info->rx.sring);
++ info->tx_ring_ref = GRANT_INVALID_REF;
++ info->rx_ring_ref = GRANT_INVALID_REF;
++ info->tx.sring = NULL;
++ info->rx.sring = NULL;
++}
++
++
++static void end_access(int ref, void *page)
++{
++ if (ref != GRANT_INVALID_REF)
++ gnttab_end_foreign_access(ref, (unsigned long)page);
++}
++
++
++/* ** Driver registration ** */
++
++
++static const struct xenbus_device_id netfront_ids[] = {
++ { "vif" },
++ { "" }
++};
++MODULE_ALIAS("xen:vif");
++
++
++static struct xenbus_driver netfront_driver = {
++ .name = "vif",
++ .owner = THIS_MODULE,
++ .ids = netfront_ids,
++ .probe = netfront_probe,
++ .remove = __devexit_p(netfront_remove),
++ .suspend = netfront_suspend,
++ .suspend_cancel = netfront_suspend_cancel,
++ .resume = netfront_resume,
++ .otherend_changed = backend_changed,
++};
++
++
++static int __init netif_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++#ifdef CONFIG_XEN
++ if (MODPARM_rx_flip && MODPARM_rx_copy) {
++ WPRINTK("Cannot specify both rx_copy and rx_flip.\n");
++ return -EINVAL;
++ }
++
++ if (!MODPARM_rx_flip && !MODPARM_rx_copy)
++ MODPARM_rx_flip = 1; /* Default is to flip. */
++#endif
++
++ netif_init_accel();
++
++ IPRINTK("Initialising virtual ethernet driver.\n");
++
++#ifdef CONFIG_INET
++ (void)register_inetaddr_notifier(¬ifier_inetdev);
++#endif
++
++ return xenbus_register_frontend(&netfront_driver);
++}
++module_init(netif_init);
++
++
++static void __exit netif_exit(void)
++{
++#ifdef CONFIG_INET
++ unregister_inetaddr_notifier(¬ifier_inetdev);
++#endif
++
++ netif_exit_accel();
++
++ return xenbus_unregister_driver(&netfront_driver);
++}
++module_exit(netif_exit);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/netfront/netfront.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.h 2008-05-19 00:33:45.878790627 +0300
+@@ -0,0 +1,274 @@
++/******************************************************************************
++ * Virtual network driver for conversing with remote driver backends.
++ *
++ * Copyright (c) 2002-2005, K A Fraser
++ * Copyright (c) 2005, XenSource Ltd
++ * Copyright (C) 2007 Solarflare Communications, Inc.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#ifndef NETFRONT_H
++#define NETFRONT_H
++
++#include <xen/interface/io/netif.h>
++#include <linux/netdevice.h>
++#include <linux/skbuff.h>
++#include <linux/list.h>
++
++#define NET_TX_RING_SIZE __RING_SIZE((struct netif_tx_sring *)0, PAGE_SIZE)
++#define NET_RX_RING_SIZE __RING_SIZE((struct netif_rx_sring *)0, PAGE_SIZE)
++
++#include <xen/xenbus.h>
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
++#endif
++
++/*
++ * Function pointer table for hooks into a network acceleration
++ * plugin. These are called at appropriate points from the netfront
++ * driver
++ */
++struct netfront_accel_hooks {
++ /*
++ * new_device: Accelerator hook to ask the plugin to support a
++ * new network interface
++ */
++ int (*new_device)(struct net_device *net_dev, struct xenbus_device *dev);
++ /*
++ * remove: Opposite of new_device
++ */
++ int (*remove)(struct xenbus_device *dev);
++ /*
++ * The net_device is being polled, check the accelerated
++ * hardware for any pending packets
++ */
++ int (*netdev_poll)(struct net_device *dev, int *pbudget);
++ /*
++ * start_xmit: Used to give the accelerated plugin the option
++ * of sending a packet. Returns non-zero if has done so, or
++ * zero to decline and force the packet onto normal send
++ * path
++ */
++ int (*start_xmit)(struct sk_buff *skb, struct net_device *dev);
++ /*
++ * start/stop_napi_interrupts Used by netfront to indicate
++ * when napi interrupts should be enabled or disabled
++ */
++ int (*start_napi_irq)(struct net_device *dev);
++ void (*stop_napi_irq)(struct net_device *dev);
++ /*
++ * Called before re-enabling the TX queue to check the fast
++ * path has slots too
++ */
++ int (*check_ready)(struct net_device *dev);
++ /*
++ * Get the fastpath network statistics
++ */
++ int (*get_stats)(struct net_device *dev,
++ struct net_device_stats *stats);
++};
++
++
++/* Version of API/protocol for communication between netfront and
++ acceleration plugin supported */
++#define NETFRONT_ACCEL_VERSION 0x00010003
++
++/*
++ * Per-netfront device state for the accelerator. This is used to
++ * allow efficient per-netfront device access to the accelerator
++ * hooks
++ */
++struct netfront_accel_vif_state {
++ struct list_head link;
++
++ struct xenbus_device *dev;
++ struct netfront_info *np;
++ struct netfront_accel_hooks *hooks;
++
++ /* Watch on the accelerator configuration value */
++ struct xenbus_watch accel_watch;
++ /* Work item to process change in accelerator */
++ struct work_struct accel_work;
++ /* The string from xenbus last time accel_watch fired */
++ char *accel_frontend;
++};
++
++/*
++ * Per-accelerator state stored in netfront. These form a list that
++ * is used to track which devices are accelerated by which plugins,
++ * and what plugins are available/have been requested
++ */
++struct netfront_accelerator {
++ /* Used to make a list */
++ struct list_head link;
++ /* ID of the accelerator */
++ int id;
++ /*
++ * String describing the accelerator. Currently this is the
++ * name of the accelerator module. This is provided by the
++ * backend accelerator through xenstore
++ */
++ char *frontend;
++ /* The hooks into the accelerator plugin module */
++ struct netfront_accel_hooks *hooks;
++
++ /*
++ * List of per-netfront device state (struct
++ * netfront_accel_vif_state) for each netfront device that is
++ * using this accelerator
++ */
++ struct list_head vif_states;
++ spinlock_t vif_states_lock;
++};
++
++struct netfront_info {
++ struct list_head list;
++ struct net_device *netdev;
++
++ struct net_device_stats stats;
++
++ struct netif_tx_front_ring tx;
++ struct netif_rx_front_ring rx;
++
++ spinlock_t tx_lock;
++ spinlock_t rx_lock;
++
++ unsigned int irq;
++ unsigned int copying_receiver;
++ unsigned int carrier;
++
++ /* Receive-ring batched refills. */
++#define RX_MIN_TARGET 8
++#define RX_DFL_MIN_TARGET 64
++#define RX_MAX_TARGET min_t(int, NET_RX_RING_SIZE, 256)
++ unsigned rx_min_target, rx_max_target, rx_target;
++ struct sk_buff_head rx_batch;
++
++ struct timer_list rx_refill_timer;
++
++ /*
++ * {tx,rx}_skbs store outstanding skbuffs. The first entry in tx_skbs
++ * is an index into a chain of free entries.
++ */
++ struct sk_buff *tx_skbs[NET_TX_RING_SIZE+1];
++ struct sk_buff *rx_skbs[NET_RX_RING_SIZE];
++
++#define TX_MAX_TARGET min_t(int, NET_RX_RING_SIZE, 256)
++ grant_ref_t gref_tx_head;
++ grant_ref_t grant_tx_ref[NET_TX_RING_SIZE + 1];
++ grant_ref_t gref_rx_head;
++ grant_ref_t grant_rx_ref[NET_RX_RING_SIZE];
++
++ struct xenbus_device *xbdev;
++ int tx_ring_ref;
++ int rx_ring_ref;
++ u8 mac[ETH_ALEN];
++
++ unsigned long rx_pfn_array[NET_RX_RING_SIZE];
++ struct multicall_entry rx_mcl[NET_RX_RING_SIZE+1];
++ struct mmu_update rx_mmu[NET_RX_RING_SIZE];
++
++ /* Private pointer to state internal to accelerator module */
++ void *accel_priv;
++ /* The accelerator used by this netfront device */
++ struct netfront_accelerator *accelerator;
++ /* The accelerator state for this netfront device */
++ struct netfront_accel_vif_state accel_vif_state;
++};
++
++
++/* Exported Functions */
++
++/*
++ * Called by an accelerator plugin module when it has loaded.
++ *
++ * frontend: the string describing the accelerator, currently the module name
++ * hooks: the hooks for netfront to use to call into the accelerator
++ * version: the version of API between frontend and plugin requested
++ *
++ * return: 0 on success, <0 on error, >0 (with version supported) on
++ * version mismatch
++ */
++extern int netfront_accelerator_loaded(int version, const char *frontend,
++ struct netfront_accel_hooks *hooks);
++
++/*
++ * Called by an accelerator plugin module when it is about to unload.
++ *
++ * frontend: the string describing the accelerator. Must match the
++ * one passed to netfront_accelerator_loaded()
++ */
++extern void netfront_accelerator_stop(const char *frontend);
++
++/*
++ * Called by an accelerator before waking the net device's TX queue to
++ * ensure the slow path has available slots. Returns true if OK to
++ * wake, false if still busy
++ */
++extern int netfront_check_queue_ready(struct net_device *net_dev);
++
++
++/* Internal-to-netfront Functions */
++
++/*
++ * Call into accelerator and check to see if it has tx space before we
++ * wake the net device's TX queue. Returns true if OK to wake, false
++ * if still busy
++ */
++extern
++int netfront_check_accelerator_queue_ready(struct net_device *dev,
++ struct netfront_info *np);
++extern
++int netfront_accelerator_call_remove(struct netfront_info *np,
++ struct xenbus_device *dev);
++extern
++int netfront_accelerator_suspend(struct netfront_info *np,
++ struct xenbus_device *dev);
++extern
++int netfront_accelerator_suspend_cancel(struct netfront_info *np,
++ struct xenbus_device *dev);
++extern
++void netfront_accelerator_resume(struct netfront_info *np,
++ struct xenbus_device *dev);
++extern
++void netfront_accelerator_call_stop_napi_irq(struct netfront_info *np,
++ struct net_device *dev);
++extern
++int netfront_accelerator_call_get_stats(struct netfront_info *np,
++ struct net_device *dev);
++extern
++void netfront_accelerator_add_watch(struct netfront_info *np);
++
++extern
++void netif_init_accel(void);
++extern
++void netif_exit_accel(void);
++
++extern
++void init_accelerator_vif(struct netfront_info *np,
++ struct xenbus_device *dev);
++#endif /* NETFRONT_H */
+--- linux-2.6.18.8/drivers/xen/pciback/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/Makefile 2008-05-19 00:33:46.162806998 +0300
+@@ -0,0 +1,17 @@
++obj-$(CONFIG_XEN_PCIDEV_BACKEND) += pciback.o
++
++pciback-y := pci_stub.o pciback_ops.o xenbus.o
++pciback-y += conf_space.o conf_space_header.o \
++ conf_space_capability.o \
++ conf_space_capability_vpd.o \
++ conf_space_capability_pm.o \
++ conf_space_quirks.o
++pciback-$(CONFIG_PCI_MSI) += conf_space_capability_msi.o
++pciback-$(CONFIG_XEN_PCIDEV_BACKEND_VPCI) += vpci.o
++pciback-$(CONFIG_XEN_PCIDEV_BACKEND_SLOT) += slot.o
++pciback-$(CONFIG_XEN_PCIDEV_BACKEND_PASS) += passthrough.o
++pciback-$(CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER) += controller.o
++
++ifeq ($(CONFIG_XEN_PCIDEV_BE_DEBUG),y)
++EXTRA_CFLAGS += -DDEBUG
++endif
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.c 2008-05-19 00:33:46.182808150 +0300
+@@ -0,0 +1,426 @@
++/*
++ * PCI Backend - Functions for creating a virtual configuration space for
++ * exported PCI Devices.
++ * It's dangerous to allow PCI Driver Domains to change their
++ * device's resources (memory, i/o ports, interrupts). We need to
++ * restrict changes to certain PCI Configuration registers:
++ * BARs, INTERRUPT_PIN, most registers in the header...
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/kernel.h>
++#include <linux/pci.h>
++#include "pciback.h"
++#include "conf_space.h"
++#include "conf_space_quirks.h"
++
++#define DEFINE_PCI_CONFIG(op,size,type) \
++int pciback_##op##_config_##size \
++(struct pci_dev *dev, int offset, type value, void *data) \
++{ \
++ return pci_##op##_config_##size (dev, offset, value); \
++}
++
++DEFINE_PCI_CONFIG(read, byte, u8 *)
++DEFINE_PCI_CONFIG(read, word, u16 *)
++DEFINE_PCI_CONFIG(read, dword, u32 *)
++
++DEFINE_PCI_CONFIG(write, byte, u8)
++DEFINE_PCI_CONFIG(write, word, u16)
++DEFINE_PCI_CONFIG(write, dword, u32)
++
++static int conf_space_read(struct pci_dev *dev,
++ struct config_field_entry *entry, int offset,
++ u32 * value)
++{
++ int ret = 0;
++ struct config_field *field = entry->field;
++
++ *value = 0;
++
++ switch (field->size) {
++ case 1:
++ if (field->u.b.read)
++ ret = field->u.b.read(dev, offset, (u8 *) value,
++ entry->data);
++ break;
++ case 2:
++ if (field->u.w.read)
++ ret = field->u.w.read(dev, offset, (u16 *) value,
++ entry->data);
++ break;
++ case 4:
++ if (field->u.dw.read)
++ ret = field->u.dw.read(dev, offset, value, entry->data);
++ break;
++ }
++ return ret;
++}
++
++static int conf_space_write(struct pci_dev *dev,
++ struct config_field_entry *entry, int offset,
++ u32 value)
++{
++ int ret = 0;
++ struct config_field *field = entry->field;
++
++ switch (field->size) {
++ case 1:
++ if (field->u.b.write)
++ ret = field->u.b.write(dev, offset, (u8) value,
++ entry->data);
++ break;
++ case 2:
++ if (field->u.w.write)
++ ret = field->u.w.write(dev, offset, (u16) value,
++ entry->data);
++ break;
++ case 4:
++ if (field->u.dw.write)
++ ret = field->u.dw.write(dev, offset, value,
++ entry->data);
++ break;
++ }
++ return ret;
++}
++
++static inline u32 get_mask(int size)
++{
++ if (size == 1)
++ return 0xff;
++ else if (size == 2)
++ return 0xffff;
++ else
++ return 0xffffffff;
++}
++
++static inline int valid_request(int offset, int size)
++{
++ /* Validate request (no un-aligned requests) */
++ if ((size == 1 || size == 2 || size == 4) && (offset % size) == 0)
++ return 1;
++ return 0;
++}
++
++static inline u32 merge_value(u32 val, u32 new_val, u32 new_val_mask,
++ int offset)
++{
++ if (offset >= 0) {
++ new_val_mask <<= (offset * 8);
++ new_val <<= (offset * 8);
++ } else {
++ new_val_mask >>= (offset * -8);
++ new_val >>= (offset * -8);
++ }
++ val = (val & ~new_val_mask) | (new_val & new_val_mask);
++
++ return val;
++}
++
++static int pcibios_err_to_errno(int err)
++{
++ switch (err) {
++ case PCIBIOS_SUCCESSFUL:
++ return XEN_PCI_ERR_success;
++ case PCIBIOS_DEVICE_NOT_FOUND:
++ return XEN_PCI_ERR_dev_not_found;
++ case PCIBIOS_BAD_REGISTER_NUMBER:
++ return XEN_PCI_ERR_invalid_offset;
++ case PCIBIOS_FUNC_NOT_SUPPORTED:
++ return XEN_PCI_ERR_not_implemented;
++ case PCIBIOS_SET_FAILED:
++ return XEN_PCI_ERR_access_denied;
++ }
++ return err;
++}
++
++int pciback_config_read(struct pci_dev *dev, int offset, int size,
++ u32 * ret_val)
++{
++ int err = 0;
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry;
++ struct config_field *field;
++ int req_start, req_end, field_start, field_end;
++ /* if read fails for any reason, return 0 (as if device didn't respond) */
++ u32 value = 0, tmp_val;
++
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG "pciback: %s: read %d bytes at 0x%x\n",
++ pci_name(dev), size, offset);
++
++ if (!valid_request(offset, size)) {
++ err = XEN_PCI_ERR_invalid_offset;
++ goto out;
++ }
++
++ /* Get the real value first, then modify as appropriate */
++ switch (size) {
++ case 1:
++ err = pci_read_config_byte(dev, offset, (u8 *) & value);
++ break;
++ case 2:
++ err = pci_read_config_word(dev, offset, (u16 *) & value);
++ break;
++ case 4:
++ err = pci_read_config_dword(dev, offset, &value);
++ break;
++ }
++
++ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
++ field = cfg_entry->field;
++
++ req_start = offset;
++ req_end = offset + size;
++ field_start = OFFSET(cfg_entry);
++ field_end = OFFSET(cfg_entry) + field->size;
++
++ if ((req_start >= field_start && req_start < field_end)
++ || (req_end > field_start && req_end <= field_end)) {
++ err = conf_space_read(dev, cfg_entry, field_start,
++ &tmp_val);
++ if (err)
++ goto out;
++
++ value = merge_value(value, tmp_val,
++ get_mask(field->size),
++ field_start - req_start);
++ }
++ }
++
++ out:
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG "pciback: %s: read %d bytes at 0x%x = %x\n",
++ pci_name(dev), size, offset, value);
++
++ *ret_val = value;
++ return pcibios_err_to_errno(err);
++}
++
++int pciback_config_write(struct pci_dev *dev, int offset, int size, u32 value)
++{
++ int err = 0, handled = 0;
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry;
++ struct config_field *field;
++ u32 tmp_val;
++ int req_start, req_end, field_start, field_end;
++
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG
++ "pciback: %s: write request %d bytes at 0x%x = %x\n",
++ pci_name(dev), size, offset, value);
++
++ if (!valid_request(offset, size))
++ return XEN_PCI_ERR_invalid_offset;
++
++ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
++ field = cfg_entry->field;
++
++ req_start = offset;
++ req_end = offset + size;
++ field_start = OFFSET(cfg_entry);
++ field_end = OFFSET(cfg_entry) + field->size;
++
++ if ((req_start >= field_start && req_start < field_end)
++ || (req_end > field_start && req_end <= field_end)) {
++ tmp_val = 0;
++
++ err = pciback_config_read(dev, field_start,
++ field->size, &tmp_val);
++ if (err)
++ break;
++
++ tmp_val = merge_value(tmp_val, value, get_mask(size),
++ req_start - field_start);
++
++ err = conf_space_write(dev, cfg_entry, field_start,
++ tmp_val);
++
++ /* handled is set true here, but not every byte
++ * may have been written! Properly detecting if
++ * every byte is handled is unnecessary as the
++ * flag is used to detect devices that need
++ * special helpers to work correctly.
++ */
++ handled = 1;
++ }
++ }
++
++ if (!handled && !err) {
++ /* By default, anything not specificially handled above is
++ * read-only. The permissive flag changes this behavior so
++ * that anything not specifically handled above is writable.
++ * This means that some fields may still be read-only because
++ * they have entries in the config_field list that intercept
++ * the write and do nothing. */
++ if (dev_data->permissive) {
++ switch (size) {
++ case 1:
++ err = pci_write_config_byte(dev, offset,
++ (u8) value);
++ break;
++ case 2:
++ err = pci_write_config_word(dev, offset,
++ (u16) value);
++ break;
++ case 4:
++ err = pci_write_config_dword(dev, offset,
++ (u32) value);
++ break;
++ }
++ } else if (!dev_data->warned_on_write) {
++ dev_data->warned_on_write = 1;
++ dev_warn(&dev->dev, "Driver tried to write to a "
++ "read-only configuration space field at offset "
++ "0x%x, size %d. This may be harmless, but if "
++ "you have problems with your device:\n"
++ "1) see permissive attribute in sysfs\n"
++ "2) report problems to the xen-devel "
++ "mailing list along with details of your "
++ "device obtained from lspci.\n", offset, size);
++ }
++ }
++
++ return pcibios_err_to_errno(err);
++}
++
++void pciback_config_free_dyn_fields(struct pci_dev *dev)
++{
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry, *t;
++ struct config_field *field;
++
++ dev_dbg(&dev->dev,
++ "free-ing dynamically allocated virtual configuration space fields\n");
++
++ list_for_each_entry_safe(cfg_entry, t, &dev_data->config_fields, list) {
++ field = cfg_entry->field;
++
++ if (field->clean) {
++ field->clean(field);
++
++ if (cfg_entry->data)
++ kfree(cfg_entry->data);
++
++ list_del(&cfg_entry->list);
++ kfree(cfg_entry);
++ }
++
++ }
++}
++
++void pciback_config_reset_dev(struct pci_dev *dev)
++{
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry;
++ struct config_field *field;
++
++ dev_dbg(&dev->dev, "resetting virtual configuration space\n");
++
++ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
++ field = cfg_entry->field;
++
++ if (field->reset)
++ field->reset(dev, OFFSET(cfg_entry), cfg_entry->data);
++ }
++}
++
++void pciback_config_free_dev(struct pci_dev *dev)
++{
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry, *t;
++ struct config_field *field;
++
++ dev_dbg(&dev->dev, "free-ing virtual configuration space fields\n");
++
++ list_for_each_entry_safe(cfg_entry, t, &dev_data->config_fields, list) {
++ list_del(&cfg_entry->list);
++
++ field = cfg_entry->field;
++
++ if (field->release)
++ field->release(dev, OFFSET(cfg_entry), cfg_entry->data);
++
++ kfree(cfg_entry);
++ }
++}
++
++int pciback_config_add_field_offset(struct pci_dev *dev,
++ struct config_field *field,
++ unsigned int base_offset)
++{
++ int err = 0;
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry;
++ void *tmp;
++
++ cfg_entry = kmalloc(sizeof(*cfg_entry), GFP_KERNEL);
++ if (!cfg_entry) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ cfg_entry->data = NULL;
++ cfg_entry->field = field;
++ cfg_entry->base_offset = base_offset;
++
++ /* silently ignore duplicate fields */
++ err = pciback_field_is_dup(dev,OFFSET(cfg_entry));
++ if (err)
++ goto out;
++
++ if (field->init) {
++ tmp = field->init(dev, OFFSET(cfg_entry));
++
++ if (IS_ERR(tmp)) {
++ err = PTR_ERR(tmp);
++ goto out;
++ }
++
++ cfg_entry->data = tmp;
++ }
++
++ dev_dbg(&dev->dev, "added config field at offset 0x%02x\n",
++ OFFSET(cfg_entry));
++ list_add_tail(&cfg_entry->list, &dev_data->config_fields);
++
++ out:
++ if (err)
++ kfree(cfg_entry);
++
++ return err;
++}
++
++/* This sets up the device's virtual configuration space to keep track of
++ * certain registers (like the base address registers (BARs) so that we can
++ * keep the client from manipulating them directly.
++ */
++int pciback_config_init_dev(struct pci_dev *dev)
++{
++ int err = 0;
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++
++ dev_dbg(&dev->dev, "initializing virtual configuration space\n");
++
++ INIT_LIST_HEAD(&dev_data->config_fields);
++
++ err = pciback_config_header_add_fields(dev);
++ if (err)
++ goto out;
++
++ err = pciback_config_capability_add_fields(dev);
++ if (err)
++ goto out;
++
++ err = pciback_config_quirks_init(dev);
++
++ out:
++ return err;
++}
++
++int pciback_config_init(void)
++{
++ return pciback_config_capability_init();
++}
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.h 2008-05-19 00:33:46.182808150 +0300
+@@ -0,0 +1,126 @@
++/*
++ * PCI Backend - Common data structures for overriding the configuration space
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#ifndef __XEN_PCIBACK_CONF_SPACE_H__
++#define __XEN_PCIBACK_CONF_SPACE_H__
++
++#include <linux/list.h>
++#include <linux/err.h>
++
++/* conf_field_init can return an errno in a ptr with ERR_PTR() */
++typedef void *(*conf_field_init) (struct pci_dev * dev, int offset);
++typedef void (*conf_field_reset) (struct pci_dev * dev, int offset, void *data);
++typedef void (*conf_field_free) (struct pci_dev * dev, int offset, void *data);
++
++typedef int (*conf_dword_write) (struct pci_dev * dev, int offset, u32 value,
++ void *data);
++typedef int (*conf_word_write) (struct pci_dev * dev, int offset, u16 value,
++ void *data);
++typedef int (*conf_byte_write) (struct pci_dev * dev, int offset, u8 value,
++ void *data);
++typedef int (*conf_dword_read) (struct pci_dev * dev, int offset, u32 * value,
++ void *data);
++typedef int (*conf_word_read) (struct pci_dev * dev, int offset, u16 * value,
++ void *data);
++typedef int (*conf_byte_read) (struct pci_dev * dev, int offset, u8 * value,
++ void *data);
++
++/* These are the fields within the configuration space which we
++ * are interested in intercepting reads/writes to and changing their
++ * values.
++ */
++struct config_field {
++ unsigned int offset;
++ unsigned int size;
++ unsigned int mask;
++ conf_field_init init;
++ conf_field_reset reset;
++ conf_field_free release;
++ void (*clean) (struct config_field * field);
++ union {
++ struct {
++ conf_dword_write write;
++ conf_dword_read read;
++ } dw;
++ struct {
++ conf_word_write write;
++ conf_word_read read;
++ } w;
++ struct {
++ conf_byte_write write;
++ conf_byte_read read;
++ } b;
++ } u;
++ struct list_head list;
++};
++
++struct config_field_entry {
++ struct list_head list;
++ struct config_field *field;
++ unsigned int base_offset;
++ void *data;
++};
++
++#define OFFSET(cfg_entry) ((cfg_entry)->base_offset+(cfg_entry)->field->offset)
++
++/* Add fields to a device - the add_fields macro expects to get a pointer to
++ * the first entry in an array (of which the ending is marked by size==0)
++ */
++int pciback_config_add_field_offset(struct pci_dev *dev,
++ struct config_field *field,
++ unsigned int offset);
++
++static inline int pciback_config_add_field(struct pci_dev *dev,
++ struct config_field *field)
++{
++ return pciback_config_add_field_offset(dev, field, 0);
++}
++
++static inline int pciback_config_add_fields(struct pci_dev *dev,
++ struct config_field *field)
++{
++ int i, err = 0;
++ for (i = 0; field[i].size != 0; i++) {
++ err = pciback_config_add_field(dev, &field[i]);
++ if (err)
++ break;
++ }
++ return err;
++}
++
++static inline int pciback_config_add_fields_offset(struct pci_dev *dev,
++ struct config_field *field,
++ unsigned int offset)
++{
++ int i, err = 0;
++ for (i = 0; field[i].size != 0; i++) {
++ err = pciback_config_add_field_offset(dev, &field[i], offset);
++ if (err)
++ break;
++ }
++ return err;
++}
++
++/* Read/Write the real configuration space */
++int pciback_read_config_byte(struct pci_dev *dev, int offset, u8 * value,
++ void *data);
++int pciback_read_config_word(struct pci_dev *dev, int offset, u16 * value,
++ void *data);
++int pciback_read_config_dword(struct pci_dev *dev, int offset, u32 * value,
++ void *data);
++int pciback_write_config_byte(struct pci_dev *dev, int offset, u8 value,
++ void *data);
++int pciback_write_config_word(struct pci_dev *dev, int offset, u16 value,
++ void *data);
++int pciback_write_config_dword(struct pci_dev *dev, int offset, u32 value,
++ void *data);
++
++int pciback_config_capability_init(void);
++
++int pciback_config_header_add_fields(struct pci_dev *dev);
++int pciback_config_capability_add_fields(struct pci_dev *dev);
++
++#endif /* __XEN_PCIBACK_CONF_SPACE_H__ */
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_capability.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.c 2008-05-19 00:33:46.186808381 +0300
+@@ -0,0 +1,71 @@
++/*
++ * PCI Backend - Handles the virtual fields found on the capability lists
++ * in the configuration space.
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/kernel.h>
++#include <linux/pci.h>
++#include "pciback.h"
++#include "conf_space.h"
++#include "conf_space_capability.h"
++
++static LIST_HEAD(capabilities);
++
++static struct config_field caplist_header[] = {
++ {
++ .offset = PCI_CAP_LIST_ID,
++ .size = 2, /* encompass PCI_CAP_LIST_ID & PCI_CAP_LIST_NEXT */
++ .u.w.read = pciback_read_config_word,
++ .u.w.write = NULL,
++ },
++ {
++ .size = 0,
++ },
++};
++
++static inline void register_capability(struct pciback_config_capability *cap)
++{
++ list_add_tail(&cap->cap_list, &capabilities);
++}
++
++int pciback_config_capability_add_fields(struct pci_dev *dev)
++{
++ int err = 0;
++ struct pciback_config_capability *cap;
++ int cap_offset;
++
++ list_for_each_entry(cap, &capabilities, cap_list) {
++ cap_offset = pci_find_capability(dev, cap->capability);
++ if (cap_offset) {
++ dev_dbg(&dev->dev, "Found capability 0x%x at 0x%x\n",
++ cap->capability, cap_offset);
++
++ err = pciback_config_add_fields_offset(dev,
++ caplist_header,
++ cap_offset);
++ if (err)
++ goto out;
++ err = pciback_config_add_fields_offset(dev,
++ cap->fields,
++ cap_offset);
++ if (err)
++ goto out;
++ }
++ }
++
++ out:
++ return err;
++}
++
++extern struct pciback_config_capability pciback_config_capability_vpd;
++extern struct pciback_config_capability pciback_config_capability_pm;
++
++int pciback_config_capability_init(void)
++{
++ register_capability(&pciback_config_capability_vpd);
++ register_capability(&pciback_config_capability_pm);
++
++ return 0;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_capability.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.h 2008-05-19 00:33:46.186808381 +0300
+@@ -0,0 +1,23 @@
++/*
++ * PCI Backend - Data structures for special overlays for structures on
++ * the capability list.
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#ifndef __PCIBACK_CONFIG_CAPABILITY_H__
++#define __PCIBACK_CONFIG_CAPABILITY_H__
++
++#include <linux/pci.h>
++#include <linux/list.h>
++
++struct pciback_config_capability {
++ struct list_head cap_list;
++
++ int capability;
++
++ /* If the device has the capability found above, add these fields */
++ struct config_field *fields;
++};
++
++#endif
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_capability_msi.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_msi.c 2008-05-19 00:33:46.186808381 +0300
+@@ -0,0 +1,57 @@
++/*
++ * PCI Backend -- Configuration overlay for MSI capability
++ */
++#include <linux/pci.h>
++#include "conf_space.h"
++#include "conf_space_capability.h"
++#include <xen/interface/io/pciif.h>
++#include "pciback.h"
++
++int pciback_enable_msi(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op)
++{
++ int otherend = pdev->xdev->otherend_id;
++ int status;
++
++ status = pci_enable_msi(dev);
++
++ if (status) {
++ printk("error enable msi for guest %x status %x\n", otherend, status);
++ op->value = 0;
++ return XEN_PCI_ERR_op_failed;
++ }
++
++ op->value = dev->irq;
++ return 0;
++}
++
++int pciback_disable_msi(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op)
++{
++ pci_disable_msi(dev);
++
++ op->value = dev->irq;
++ return 0;
++}
++
++int pciback_enable_msix(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op)
++{
++ int result;
++
++ result = pci_enable_msix(dev, op->msix_entries, op->value);
++
++ op->value = result;
++ return result;
++}
++
++int pciback_disable_msix(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op)
++{
++
++ pci_disable_msix(dev);
++
++ op->value = dev->irq;
++ return 0;
++}
++
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_capability_pm.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_pm.c 2008-05-19 00:33:46.206809534 +0300
+@@ -0,0 +1,128 @@
++/*
++ * PCI Backend - Configuration space overlay for power management
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/pci.h>
++#include "conf_space.h"
++#include "conf_space_capability.h"
++
++static int pm_caps_read(struct pci_dev *dev, int offset, u16 *value,
++ void *data)
++{
++ int err;
++ u16 real_value;
++
++ err = pci_read_config_word(dev, offset, &real_value);
++ if (err)
++ goto out;
++
++ *value = real_value & ~PCI_PM_CAP_PME_MASK;
++
++ out:
++ return err;
++}
++
++/* PM_OK_BITS specifies the bits that the driver domain is allowed to change.
++ * Can't allow driver domain to enable PMEs - they're shared */
++#define PM_OK_BITS (PCI_PM_CTRL_PME_STATUS|PCI_PM_CTRL_DATA_SEL_MASK)
++
++static int pm_ctrl_write(struct pci_dev *dev, int offset, u16 new_value,
++ void *data)
++{
++ int err;
++ u16 old_value;
++ pci_power_t new_state, old_state;
++
++ err = pci_read_config_word(dev, offset, &old_value);
++ if (err)
++ goto out;
++
++ old_state = (pci_power_t)(old_value & PCI_PM_CTRL_STATE_MASK);
++ new_state = (pci_power_t)(new_value & PCI_PM_CTRL_STATE_MASK);
++
++ new_value &= PM_OK_BITS;
++ if ((old_value & PM_OK_BITS) != new_value) {
++ new_value = (old_value & ~PM_OK_BITS) | new_value;
++ err = pci_write_config_word(dev, offset, new_value);
++ if (err)
++ goto out;
++ }
++
++ /* Let pci core handle the power management change */
++ dev_dbg(&dev->dev, "set power state to %x\n", new_state);
++ err = pci_set_power_state(dev, new_state);
++ if (err) {
++ err = PCIBIOS_SET_FAILED;
++ goto out;
++ }
++
++ /*
++ * Device may lose PCI config info on D3->D0 transition. This
++ * is a problem for some guests which will not reset BARs. Even
++ * those that have a go will be foiled by our BAR-write handler
++ * which will discard the write! Since Linux won't re-init
++ * the config space automatically in all cases, we do it here.
++ * Future: Should we re-initialise all first 64 bytes of config space?
++ */
++ if (new_state == PCI_D0 &&
++ (old_state == PCI_D3hot || old_state == PCI_D3cold) &&
++ !(old_value & PCI_PM_CTRL_NO_SOFT_RESET))
++ pci_restore_bars(dev);
++
++ out:
++ return err;
++}
++
++/* Ensure PMEs are disabled */
++static void *pm_ctrl_init(struct pci_dev *dev, int offset)
++{
++ int err;
++ u16 value;
++
++ err = pci_read_config_word(dev, offset, &value);
++ if (err)
++ goto out;
++
++ if (value & PCI_PM_CTRL_PME_ENABLE) {
++ value &= ~PCI_PM_CTRL_PME_ENABLE;
++ err = pci_write_config_word(dev, offset, value);
++ }
++
++ out:
++ return ERR_PTR(err);
++}
++
++static struct config_field caplist_pm[] = {
++ {
++ .offset = PCI_PM_PMC,
++ .size = 2,
++ .u.w.read = pm_caps_read,
++ },
++ {
++ .offset = PCI_PM_CTRL,
++ .size = 2,
++ .init = pm_ctrl_init,
++ .u.w.read = pciback_read_config_word,
++ .u.w.write = pm_ctrl_write,
++ },
++ {
++ .offset = PCI_PM_PPB_EXTENSIONS,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ },
++ {
++ .offset = PCI_PM_DATA_REGISTER,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ },
++ {
++ .size = 0,
++ },
++};
++
++struct pciback_config_capability pciback_config_capability_pm = {
++ .capability = PCI_CAP_ID_PM,
++ .fields = caplist_pm,
++};
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_capability_vpd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_vpd.c 2008-05-19 00:33:46.226810687 +0300
+@@ -0,0 +1,42 @@
++/*
++ * PCI Backend - Configuration space overlay for Vital Product Data
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/pci.h>
++#include "conf_space.h"
++#include "conf_space_capability.h"
++
++static int vpd_address_write(struct pci_dev *dev, int offset, u16 value,
++ void *data)
++{
++ /* Disallow writes to the vital product data */
++ if (value & PCI_VPD_ADDR_F)
++ return PCIBIOS_SET_FAILED;
++ else
++ return pci_write_config_word(dev, offset, value);
++}
++
++static struct config_field caplist_vpd[] = {
++ {
++ .offset = PCI_VPD_ADDR,
++ .size = 2,
++ .u.w.read = pciback_read_config_word,
++ .u.w.write = vpd_address_write,
++ },
++ {
++ .offset = PCI_VPD_DATA,
++ .size = 4,
++ .u.dw.read = pciback_read_config_dword,
++ .u.dw.write = NULL,
++ },
++ {
++ .size = 0,
++ },
++};
++
++struct pciback_config_capability pciback_config_capability_vpd = {
++ .capability = PCI_CAP_ID_VPD,
++ .fields = caplist_vpd,
++};
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_header.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_header.c 2008-05-19 00:33:46.226810687 +0300
+@@ -0,0 +1,323 @@
++/*
++ * PCI Backend - Handles the virtual fields in the configuration space headers.
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/kernel.h>
++#include <linux/pci.h>
++#include "pciback.h"
++#include "conf_space.h"
++
++struct pci_bar_info {
++ u32 val;
++ u32 len_val;
++ int which;
++};
++
++#define is_enable_cmd(value) ((value)&(PCI_COMMAND_MEMORY|PCI_COMMAND_IO))
++#define is_master_cmd(value) ((value)&PCI_COMMAND_MASTER)
++
++static int command_write(struct pci_dev *dev, int offset, u16 value, void *data)
++{
++ int err;
++
++ if (!dev->is_enabled && is_enable_cmd(value)) {
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG "pciback: %s: enable\n",
++ pci_name(dev));
++ err = pci_enable_device(dev);
++ if (err)
++ return err;
++ } else if (dev->is_enabled && !is_enable_cmd(value)) {
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG "pciback: %s: disable\n",
++ pci_name(dev));
++ pci_disable_device(dev);
++ }
++
++ if (!dev->is_busmaster && is_master_cmd(value)) {
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG "pciback: %s: set bus master\n",
++ pci_name(dev));
++ pci_set_master(dev);
++ }
++
++ if (value & PCI_COMMAND_INVALIDATE) {
++ if (unlikely(verbose_request))
++ printk(KERN_DEBUG
++ "pciback: %s: enable memory-write-invalidate\n",
++ pci_name(dev));
++ err = pci_set_mwi(dev);
++ if (err) {
++ printk(KERN_WARNING
++ "pciback: %s: cannot enable memory-write-invalidate (%d)\n",
++ pci_name(dev), err);
++ value &= ~PCI_COMMAND_INVALIDATE;
++ }
++ }
++
++ return pci_write_config_word(dev, offset, value);
++}
++
++static int rom_write(struct pci_dev *dev, int offset, u32 value, void *data)
++{
++ struct pci_bar_info *bar = data;
++
++ if (unlikely(!bar)) {
++ printk(KERN_WARNING "pciback: driver data not found for %s\n",
++ pci_name(dev));
++ return XEN_PCI_ERR_op_failed;
++ }
++
++ /* A write to obtain the length must happen as a 32-bit write.
++ * This does not (yet) support writing individual bytes
++ */
++ if (value == ~PCI_ROM_ADDRESS_ENABLE)
++ bar->which = 1;
++ else {
++ u32 tmpval;
++ pci_read_config_dword(dev, offset, &tmpval);
++ if (tmpval != bar->val && value == bar->val) {
++ /* Allow restoration of bar value. */
++ pci_write_config_dword(dev, offset, bar->val);
++ }
++ bar->which = 0;
++ }
++
++ /* Do we need to support enabling/disabling the rom address here? */
++
++ return 0;
++}
++
++/* For the BARs, only allow writes which write ~0 or
++ * the correct resource information
++ * (Needed for when the driver probes the resource usage)
++ */
++static int bar_write(struct pci_dev *dev, int offset, u32 value, void *data)
++{
++ struct pci_bar_info *bar = data;
++
++ if (unlikely(!bar)) {
++ printk(KERN_WARNING "pciback: driver data not found for %s\n",
++ pci_name(dev));
++ return XEN_PCI_ERR_op_failed;
++ }
++
++ /* A write to obtain the length must happen as a 32-bit write.
++ * This does not (yet) support writing individual bytes
++ */
++ if (value == ~0)
++ bar->which = 1;
++ else {
++ u32 tmpval;
++ pci_read_config_dword(dev, offset, &tmpval);
++ if (tmpval != bar->val && value == bar->val) {
++ /* Allow restoration of bar value. */
++ pci_write_config_dword(dev, offset, bar->val);
++ }
++ bar->which = 0;
++ }
++
++ return 0;
++}
++
++static int bar_read(struct pci_dev *dev, int offset, u32 * value, void *data)
++{
++ struct pci_bar_info *bar = data;
++
++ if (unlikely(!bar)) {
++ printk(KERN_WARNING "pciback: driver data not found for %s\n",
++ pci_name(dev));
++ return XEN_PCI_ERR_op_failed;
++ }
++
++ *value = bar->which ? bar->len_val : bar->val;
++
++ return 0;
++}
++
++static inline void read_dev_bar(struct pci_dev *dev,
++ struct pci_bar_info *bar_info, int offset,
++ u32 len_mask)
++{
++ pci_read_config_dword(dev, offset, &bar_info->val);
++ pci_write_config_dword(dev, offset, len_mask);
++ pci_read_config_dword(dev, offset, &bar_info->len_val);
++ pci_write_config_dword(dev, offset, bar_info->val);
++}
++
++static void *bar_init(struct pci_dev *dev, int offset)
++{
++ struct pci_bar_info *bar = kmalloc(sizeof(*bar), GFP_KERNEL);
++
++ if (!bar)
++ return ERR_PTR(-ENOMEM);
++
++ read_dev_bar(dev, bar, offset, ~0);
++ bar->which = 0;
++
++ return bar;
++}
++
++static void *rom_init(struct pci_dev *dev, int offset)
++{
++ struct pci_bar_info *bar = kmalloc(sizeof(*bar), GFP_KERNEL);
++
++ if (!bar)
++ return ERR_PTR(-ENOMEM);
++
++ read_dev_bar(dev, bar, offset, ~PCI_ROM_ADDRESS_ENABLE);
++ bar->which = 0;
++
++ return bar;
++}
++
++static void bar_reset(struct pci_dev *dev, int offset, void *data)
++{
++ struct pci_bar_info *bar = data;
++
++ bar->which = 0;
++}
++
++static void bar_release(struct pci_dev *dev, int offset, void *data)
++{
++ kfree(data);
++}
++
++static int interrupt_read(struct pci_dev *dev, int offset, u8 * value,
++ void *data)
++{
++ *value = (u8) dev->irq;
++
++ return 0;
++}
++
++static int bist_write(struct pci_dev *dev, int offset, u8 value, void *data)
++{
++ u8 cur_value;
++ int err;
++
++ err = pci_read_config_byte(dev, offset, &cur_value);
++ if (err)
++ goto out;
++
++ if ((cur_value & ~PCI_BIST_START) == (value & ~PCI_BIST_START)
++ || value == PCI_BIST_START)
++ err = pci_write_config_byte(dev, offset, value);
++
++ out:
++ return err;
++}
++
++static struct config_field header_common[] = {
++ {
++ .offset = PCI_COMMAND,
++ .size = 2,
++ .u.w.read = pciback_read_config_word,
++ .u.w.write = command_write,
++ },
++ {
++ .offset = PCI_INTERRUPT_LINE,
++ .size = 1,
++ .u.b.read = interrupt_read,
++ },
++ {
++ .offset = PCI_INTERRUPT_PIN,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ },
++ {
++ /* Any side effects of letting driver domain control cache line? */
++ .offset = PCI_CACHE_LINE_SIZE,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ .u.b.write = pciback_write_config_byte,
++ },
++ {
++ .offset = PCI_LATENCY_TIMER,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ },
++ {
++ .offset = PCI_BIST,
++ .size = 1,
++ .u.b.read = pciback_read_config_byte,
++ .u.b.write = bist_write,
++ },
++ {
++ .size = 0,
++ },
++};
++
++#define CFG_FIELD_BAR(reg_offset) \
++ { \
++ .offset = reg_offset, \
++ .size = 4, \
++ .init = bar_init, \
++ .reset = bar_reset, \
++ .release = bar_release, \
++ .u.dw.read = bar_read, \
++ .u.dw.write = bar_write, \
++ }
++
++#define CFG_FIELD_ROM(reg_offset) \
++ { \
++ .offset = reg_offset, \
++ .size = 4, \
++ .init = rom_init, \
++ .reset = bar_reset, \
++ .release = bar_release, \
++ .u.dw.read = bar_read, \
++ .u.dw.write = rom_write, \
++ }
++
++static struct config_field header_0[] = {
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_0),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_1),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_2),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_3),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_4),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_5),
++ CFG_FIELD_ROM(PCI_ROM_ADDRESS),
++ {
++ .size = 0,
++ },
++};
++
++static struct config_field header_1[] = {
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_0),
++ CFG_FIELD_BAR(PCI_BASE_ADDRESS_1),
++ CFG_FIELD_ROM(PCI_ROM_ADDRESS1),
++ {
++ .size = 0,
++ },
++};
++
++int pciback_config_header_add_fields(struct pci_dev *dev)
++{
++ int err;
++
++ err = pciback_config_add_fields(dev, header_common);
++ if (err)
++ goto out;
++
++ switch (dev->hdr_type) {
++ case PCI_HEADER_TYPE_NORMAL:
++ err = pciback_config_add_fields(dev, header_0);
++ break;
++
++ case PCI_HEADER_TYPE_BRIDGE:
++ err = pciback_config_add_fields(dev, header_1);
++ break;
++
++ default:
++ err = -EINVAL;
++ printk(KERN_ERR "pciback: %s: Unsupported header type %d!\n",
++ pci_name(dev), dev->hdr_type);
++ break;
++ }
++
++ out:
++ return err;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_quirks.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.c 2008-05-19 00:33:46.258812531 +0300
+@@ -0,0 +1,126 @@
++/*
++ * PCI Backend - Handle special overlays for broken devices.
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ * Author: Chris Bookholt <hap10@epoch.ncsc.mil>
++ */
++
++#include <linux/kernel.h>
++#include <linux/pci.h>
++#include "pciback.h"
++#include "conf_space.h"
++#include "conf_space_quirks.h"
++
++LIST_HEAD(pciback_quirks);
++
++struct pciback_config_quirk *pciback_find_quirk(struct pci_dev *dev)
++{
++ struct pciback_config_quirk *tmp_quirk;
++
++ list_for_each_entry(tmp_quirk, &pciback_quirks, quirks_list)
++ if (pci_match_id(&tmp_quirk->devid, dev))
++ goto out;
++ tmp_quirk = NULL;
++ printk(KERN_DEBUG
++ "quirk didn't match any device pciback knows about\n");
++ out:
++ return tmp_quirk;
++}
++
++static inline void register_quirk(struct pciback_config_quirk *quirk)
++{
++ list_add_tail(&quirk->quirks_list, &pciback_quirks);
++}
++
++int pciback_field_is_dup(struct pci_dev *dev, unsigned int reg)
++{
++ int ret = 0;
++ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ struct config_field_entry *cfg_entry;
++
++ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
++ if ( OFFSET(cfg_entry) == reg) {
++ ret = 1;
++ break;
++ }
++ }
++ return ret;
++}
++
++int pciback_config_quirks_add_field(struct pci_dev *dev, struct config_field
++ *field)
++{
++ int err = 0;
++
++ switch (field->size) {
++ case 1:
++ field->u.b.read = pciback_read_config_byte;
++ field->u.b.write = pciback_write_config_byte;
++ break;
++ case 2:
++ field->u.w.read = pciback_read_config_word;
++ field->u.w.write = pciback_write_config_word;
++ break;
++ case 4:
++ field->u.dw.read = pciback_read_config_dword;
++ field->u.dw.write = pciback_write_config_dword;
++ break;
++ default:
++ err = -EINVAL;
++ goto out;
++ }
++
++ pciback_config_add_field(dev, field);
++
++ out:
++ return err;
++}
++
++int pciback_config_quirks_init(struct pci_dev *dev)
++{
++ struct pciback_config_quirk *quirk;
++ int ret = 0;
++
++ quirk = kzalloc(sizeof(*quirk), GFP_ATOMIC);
++ if (!quirk) {
++ ret = -ENOMEM;
++ goto out;
++ }
++
++ quirk->devid.vendor = dev->vendor;
++ quirk->devid.device = dev->device;
++ quirk->devid.subvendor = dev->subsystem_vendor;
++ quirk->devid.subdevice = dev->subsystem_device;
++ quirk->devid.class = 0;
++ quirk->devid.class_mask = 0;
++ quirk->devid.driver_data = 0UL;
++
++ quirk->pdev = dev;
++
++ register_quirk(quirk);
++ out:
++ return ret;
++}
++
++void pciback_config_field_free(struct config_field *field)
++{
++ kfree(field);
++}
++
++int pciback_config_quirk_release(struct pci_dev *dev)
++{
++ struct pciback_config_quirk *quirk;
++ int ret = 0;
++
++ quirk = pciback_find_quirk(dev);
++ if (!quirk) {
++ ret = -ENXIO;
++ goto out;
++ }
++
++ list_del(&quirk->quirks_list);
++ kfree(quirk);
++
++ out:
++ return ret;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/conf_space_quirks.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.h 2008-05-19 00:33:46.262812762 +0300
+@@ -0,0 +1,35 @@
++/*
++ * PCI Backend - Data structures for special overlays for broken devices.
++ *
++ * Ryan Wilson <hap9@epoch.ncsc.mil>
++ * Chris Bookholt <hap10@epoch.ncsc.mil>
++ */
++
++#ifndef __XEN_PCIBACK_CONF_SPACE_QUIRKS_H__
++#define __XEN_PCIBACK_CONF_SPACE_QUIRKS_H__
++
++#include <linux/pci.h>
++#include <linux/list.h>
++
++struct pciback_config_quirk {
++ struct list_head quirks_list;
++ struct pci_device_id devid;
++ struct pci_dev *pdev;
++};
++
++struct pciback_config_quirk *pciback_find_quirk(struct pci_dev *dev);
++
++int pciback_config_quirks_add_field(struct pci_dev *dev, struct config_field
++ *field);
++
++int pciback_config_quirks_remove_field(struct pci_dev *dev, int reg);
++
++int pciback_config_quirks_init(struct pci_dev *dev);
++
++void pciback_config_field_free(struct config_field *field);
++
++int pciback_config_quirk_release(struct pci_dev *dev);
++
++int pciback_field_is_dup(struct pci_dev *dev, unsigned int reg);
++
++#endif
+--- linux-2.6.18.8/drivers/xen/pciback/controller.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/controller.c 2008-05-19 00:33:46.262812762 +0300
+@@ -0,0 +1,408 @@
++/*
++ * Copyright (C) 2007 Hewlett-Packard Development Company, L.P.
++ * Alex Williamson <alex.williamson@hp.com>
++ *
++ * PCI "Controller" Backend - virtualize PCI bus topology based on PCI
++ * controllers. Devices under the same PCI controller are exposed on the
++ * same virtual domain:bus. Within a bus, device slots are virtualized
++ * to compact the bus.
++ *
++ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
++ */
++
++#include <linux/acpi.h>
++#include <linux/list.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include "pciback.h"
++
++#define PCI_MAX_BUSSES 255
++#define PCI_MAX_SLOTS 32
++
++struct controller_dev_entry {
++ struct list_head list;
++ struct pci_dev *dev;
++ unsigned int devfn;
++};
++
++struct controller_list_entry {
++ struct list_head list;
++ struct pci_controller *controller;
++ unsigned int domain;
++ unsigned int bus;
++ unsigned int next_devfn;
++ struct list_head dev_list;
++};
++
++struct controller_dev_data {
++ struct list_head list;
++ unsigned int next_domain;
++ unsigned int next_bus;
++ spinlock_t lock;
++};
++
++struct walk_info {
++ struct pciback_device *pdev;
++ int resource_count;
++ int root_num;
++};
++
++struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn)
++{
++ struct controller_dev_data *dev_data = pdev->pci_dev_data;
++ struct controller_dev_entry *dev_entry;
++ struct controller_list_entry *cntrl_entry;
++ struct pci_dev *dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++
++ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
++ if (cntrl_entry->domain != domain ||
++ cntrl_entry->bus != bus)
++ continue;
++
++ list_for_each_entry(dev_entry, &cntrl_entry->dev_list, list) {
++ if (devfn == dev_entry->devfn) {
++ dev = dev_entry->dev;
++ goto found;
++ }
++ }
++ }
++found:
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++
++ return dev;
++}
++
++int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev,
++ int devid, publish_pci_dev_cb publish_cb)
++{
++ struct controller_dev_data *dev_data = pdev->pci_dev_data;
++ struct controller_dev_entry *dev_entry;
++ struct controller_list_entry *cntrl_entry;
++ struct pci_controller *dev_controller = PCI_CONTROLLER(dev);
++ unsigned long flags;
++ int ret = 0, found = 0;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++
++ /* Look to see if we already have a domain:bus for this controller */
++ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
++ if (cntrl_entry->controller == dev_controller) {
++ found = 1;
++ break;
++ }
++ }
++
++ if (!found) {
++ cntrl_entry = kmalloc(sizeof(*cntrl_entry), GFP_ATOMIC);
++ if (!cntrl_entry) {
++ ret = -ENOMEM;
++ goto out;
++ }
++
++ cntrl_entry->controller = dev_controller;
++ cntrl_entry->next_devfn = PCI_DEVFN(0, 0);
++
++ cntrl_entry->domain = dev_data->next_domain;
++ cntrl_entry->bus = dev_data->next_bus++;
++ if (dev_data->next_bus > PCI_MAX_BUSSES) {
++ dev_data->next_domain++;
++ dev_data->next_bus = 0;
++ }
++
++ INIT_LIST_HEAD(&cntrl_entry->dev_list);
++
++ list_add_tail(&cntrl_entry->list, &dev_data->list);
++ }
++
++ if (PCI_SLOT(cntrl_entry->next_devfn) > PCI_MAX_SLOTS) {
++ /*
++ * While it seems unlikely, this can actually happen if
++ * a controller has P2P bridges under it.
++ */
++ xenbus_dev_fatal(pdev->xdev, -ENOSPC, "Virtual bus %04x:%02x "
++ "is full, no room to export %04x:%02x:%02x.%x",
++ cntrl_entry->domain, cntrl_entry->bus,
++ pci_domain_nr(dev->bus), dev->bus->number,
++ PCI_SLOT(dev->devfn), PCI_FUNC(dev->devfn));
++ ret = -ENOSPC;
++ goto out;
++ }
++
++ dev_entry = kmalloc(sizeof(*dev_entry), GFP_ATOMIC);
++ if (!dev_entry) {
++ if (list_empty(&cntrl_entry->dev_list)) {
++ list_del(&cntrl_entry->list);
++ kfree(cntrl_entry);
++ }
++ ret = -ENOMEM;
++ goto out;
++ }
++
++ dev_entry->dev = dev;
++ dev_entry->devfn = cntrl_entry->next_devfn;
++
++ list_add_tail(&dev_entry->list, &cntrl_entry->dev_list);
++
++ cntrl_entry->next_devfn += PCI_DEVFN(1, 0);
++
++out:
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++
++ /* TODO: Publish virtual domain:bus:slot.func here. */
++
++ return ret;
++}
++
++void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
++{
++ struct controller_dev_data *dev_data = pdev->pci_dev_data;
++ struct controller_list_entry *cntrl_entry;
++ struct controller_dev_entry *dev_entry = NULL;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++
++ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
++ if (cntrl_entry->controller != PCI_CONTROLLER(dev))
++ continue;
++
++ list_for_each_entry(dev_entry, &cntrl_entry->dev_list, list) {
++ if (dev_entry->dev == dev) {
++ found_dev = dev_entry->dev;
++ break;
++ }
++ }
++ }
++
++ if (!found_dev) {
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++ return;
++ }
++
++ list_del(&dev_entry->list);
++ kfree(dev_entry);
++
++ if (list_empty(&cntrl_entry->dev_list)) {
++ list_del(&cntrl_entry->list);
++ kfree(cntrl_entry);
++ }
++
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++ pcistub_put_pci_dev(found_dev);
++}
++
++int pciback_init_devices(struct pciback_device *pdev)
++{
++ struct controller_dev_data *dev_data;
++
++ dev_data = kmalloc(sizeof(*dev_data), GFP_KERNEL);
++ if (!dev_data)
++ return -ENOMEM;
++
++ spin_lock_init(&dev_data->lock);
++
++ INIT_LIST_HEAD(&dev_data->list);
++
++ /* Starting domain:bus numbers */
++ dev_data->next_domain = 0;
++ dev_data->next_bus = 0;
++
++ pdev->pci_dev_data = dev_data;
++
++ return 0;
++}
++
++static acpi_status write_xenbus_resource(struct acpi_resource *res, void *data)
++{
++ struct walk_info *info = data;
++ struct acpi_resource_address64 addr;
++ acpi_status status;
++ int i, len, err;
++ char str[32], tmp[3];
++ unsigned char *ptr, *buf;
++
++ status = acpi_resource_to_address64(res, &addr);
++
++ /* Do we care about this range? Let's check. */
++ if (!ACPI_SUCCESS(status) ||
++ !(addr.resource_type == ACPI_MEMORY_RANGE ||
++ addr.resource_type == ACPI_IO_RANGE) ||
++ !addr.address_length || addr.producer_consumer != ACPI_PRODUCER)
++ return AE_OK;
++
++ /*
++ * Furthermore, we really only care to tell the guest about
++ * address ranges that require address translation of some sort.
++ */
++ if (!(addr.resource_type == ACPI_MEMORY_RANGE &&
++ addr.info.mem.translation) &&
++ !(addr.resource_type == ACPI_IO_RANGE &&
++ addr.info.io.translation))
++ return AE_OK;
++
++ /* Store the resource in xenbus for the guest */
++ len = snprintf(str, sizeof(str), "root-%d-resource-%d",
++ info->root_num, info->resource_count);
++ if (unlikely(len >= (sizeof(str) - 1)))
++ return AE_OK;
++
++ buf = kzalloc((sizeof(*res) * 2) + 1, GFP_KERNEL);
++ if (!buf)
++ return AE_OK;
++
++ /* Clean out resource_source */
++ res->data.address64.resource_source.index = 0xFF;
++ res->data.address64.resource_source.string_length = 0;
++ res->data.address64.resource_source.string_ptr = NULL;
++
++ ptr = (unsigned char *)res;
++
++ /* Turn the acpi_resource into an ASCII byte stream */
++ for (i = 0; i < sizeof(*res); i++) {
++ snprintf(tmp, sizeof(tmp), "%02x", ptr[i]);
++ strncat(buf, tmp, 2);
++ }
++
++ err = xenbus_printf(XBT_NIL, info->pdev->xdev->nodename,
++ str, "%s", buf);
++
++ if (!err)
++ info->resource_count++;
++
++ kfree(buf);
++
++ return AE_OK;
++}
++
++int pciback_publish_pci_roots(struct pciback_device *pdev,
++ publish_pci_root_cb publish_root_cb)
++{
++ struct controller_dev_data *dev_data = pdev->pci_dev_data;
++ struct controller_list_entry *cntrl_entry;
++ int i, root_num, len, err = 0;
++ unsigned int domain, bus;
++ char str[64];
++ struct walk_info info;
++
++ spin_lock(&dev_data->lock);
++
++ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
++ /* First publish all the domain:bus info */
++ err = publish_root_cb(pdev, cntrl_entry->domain,
++ cntrl_entry->bus);
++ if (err)
++ goto out;
++
++ /*
++ * Now figure out which root-%d this belongs to
++ * so we can associate resources with it.
++ */
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ "root_num", "%d", &root_num);
++
++ if (err != 1)
++ goto out;
++
++ for (i = 0; i < root_num; i++) {
++ len = snprintf(str, sizeof(str), "root-%d", i);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ str, "%x:%x", &domain, &bus);
++ if (err != 2)
++ goto out;
++
++ /* Is this the one we just published? */
++ if (domain == cntrl_entry->domain &&
++ bus == cntrl_entry->bus)
++ break;
++ }
++
++ if (i == root_num)
++ goto out;
++
++ info.pdev = pdev;
++ info.resource_count = 0;
++ info.root_num = i;
++
++ /* Let ACPI do the heavy lifting on decoding resources */
++ acpi_walk_resources(cntrl_entry->controller->acpi_handle,
++ METHOD_NAME__CRS, write_xenbus_resource,
++ &info);
++
++ /* No resouces. OK. On to the next one */
++ if (!info.resource_count)
++ continue;
++
++ /* Store the number of resources we wrote for this root-%d */
++ len = snprintf(str, sizeof(str), "root-%d-resources", i);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
++ "%d", info.resource_count);
++ if (err)
++ goto out;
++ }
++
++ /* Finally, write some magic to synchronize with the guest. */
++ len = snprintf(str, sizeof(str), "root-resource-magic");
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
++ "%lx", (sizeof(struct acpi_resource) * 2) + 1);
++
++out:
++ spin_unlock(&dev_data->lock);
++
++ return err;
++}
++
++void pciback_release_devices(struct pciback_device *pdev)
++{
++ struct controller_dev_data *dev_data = pdev->pci_dev_data;
++ struct controller_list_entry *cntrl_entry, *c;
++ struct controller_dev_entry *dev_entry, *d;
++
++ list_for_each_entry_safe(cntrl_entry, c, &dev_data->list, list) {
++ list_for_each_entry_safe(dev_entry, d,
++ &cntrl_entry->dev_list, list) {
++ list_del(&dev_entry->list);
++ pcistub_put_pci_dev(dev_entry->dev);
++ kfree(dev_entry);
++ }
++ list_del(&cntrl_entry->list);
++ kfree(cntrl_entry);
++ }
++
++ kfree(dev_data);
++ pdev->pci_dev_data = NULL;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/passthrough.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/passthrough.c 2008-05-19 00:33:46.262812762 +0300
+@@ -0,0 +1,166 @@
++/*
++ * PCI Backend - Provides restricted access to the real PCI bus topology
++ * to the frontend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/list.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include "pciback.h"
++
++struct passthrough_dev_data {
++ /* Access to dev_list must be protected by lock */
++ struct list_head dev_list;
++ spinlock_t lock;
++};
++
++struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn)
++{
++ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
++ struct pci_dev_entry *dev_entry;
++ struct pci_dev *dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++
++ list_for_each_entry(dev_entry, &dev_data->dev_list, list) {
++ if (domain == (unsigned int)pci_domain_nr(dev_entry->dev->bus)
++ && bus == (unsigned int)dev_entry->dev->bus->number
++ && devfn == dev_entry->dev->devfn) {
++ dev = dev_entry->dev;
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++
++ return dev;
++}
++
++int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev,
++ int devid, publish_pci_dev_cb publish_cb)
++{
++ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
++ struct pci_dev_entry *dev_entry;
++ unsigned long flags;
++ unsigned int domain, bus, devfn;
++ int err;
++
++ dev_entry = kmalloc(sizeof(*dev_entry), GFP_KERNEL);
++ if (!dev_entry)
++ return -ENOMEM;
++ dev_entry->dev = dev;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++ list_add_tail(&dev_entry->list, &dev_data->dev_list);
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++
++ /* Publish this device. */
++ domain = (unsigned int)pci_domain_nr(dev->bus);
++ bus = (unsigned int)dev->bus->number;
++ devfn = dev->devfn;
++ err = publish_cb(pdev, domain, bus, devfn, devid);
++
++ return err;
++}
++
++void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
++{
++ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
++ struct pci_dev_entry *dev_entry, *t;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&dev_data->lock, flags);
++
++ list_for_each_entry_safe(dev_entry, t, &dev_data->dev_list, list) {
++ if (dev_entry->dev == dev) {
++ list_del(&dev_entry->list);
++ found_dev = dev_entry->dev;
++ kfree(dev_entry);
++ }
++ }
++
++ spin_unlock_irqrestore(&dev_data->lock, flags);
++
++ if (found_dev)
++ pcistub_put_pci_dev(found_dev);
++}
++
++int pciback_init_devices(struct pciback_device *pdev)
++{
++ struct passthrough_dev_data *dev_data;
++
++ dev_data = kmalloc(sizeof(*dev_data), GFP_KERNEL);
++ if (!dev_data)
++ return -ENOMEM;
++
++ spin_lock_init(&dev_data->lock);
++
++ INIT_LIST_HEAD(&dev_data->dev_list);
++
++ pdev->pci_dev_data = dev_data;
++
++ return 0;
++}
++
++int pciback_publish_pci_roots(struct pciback_device *pdev,
++ publish_pci_root_cb publish_root_cb)
++{
++ int err = 0;
++ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
++ struct pci_dev_entry *dev_entry, *e;
++ struct pci_dev *dev;
++ int found;
++ unsigned int domain, bus;
++
++ spin_lock(&dev_data->lock);
++
++ list_for_each_entry(dev_entry, &dev_data->dev_list, list) {
++ /* Only publish this device as a root if none of its
++ * parent bridges are exported
++ */
++ found = 0;
++ dev = dev_entry->dev->bus->self;
++ for (; !found && dev != NULL; dev = dev->bus->self) {
++ list_for_each_entry(e, &dev_data->dev_list, list) {
++ if (dev == e->dev) {
++ found = 1;
++ break;
++ }
++ }
++ }
++
++ domain = (unsigned int)pci_domain_nr(dev_entry->dev->bus);
++ bus = (unsigned int)dev_entry->dev->bus->number;
++
++ if (!found) {
++ err = publish_root_cb(pdev, domain, bus);
++ if (err)
++ break;
++ }
++ }
++
++ spin_unlock(&dev_data->lock);
++
++ return err;
++}
++
++void pciback_release_devices(struct pciback_device *pdev)
++{
++ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
++ struct pci_dev_entry *dev_entry, *t;
++
++ list_for_each_entry_safe(dev_entry, t, &dev_data->dev_list, list) {
++ list_del(&dev_entry->list);
++ pcistub_put_pci_dev(dev_entry->dev);
++ kfree(dev_entry);
++ }
++
++ kfree(dev_data);
++ pdev->pci_dev_data = NULL;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/pci_stub.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/pci_stub.c 2008-05-19 00:33:46.266812993 +0300
+@@ -0,0 +1,953 @@
++/*
++ * PCI Stub Driver - Grabs devices in backend to be exported later
++ *
++ * Ryan Wilson <hap9@epoch.ncsc.mil>
++ * Chris Bookholt <hap10@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <linux/init.h>
++#include <linux/list.h>
++#include <linux/spinlock.h>
++#include <linux/kref.h>
++#include <asm/atomic.h>
++#include "pciback.h"
++#include "conf_space.h"
++#include "conf_space_quirks.h"
++
++static char *pci_devs_to_hide = NULL;
++module_param_named(hide, pci_devs_to_hide, charp, 0444);
++
++struct pcistub_device_id {
++ struct list_head slot_list;
++ int domain;
++ unsigned char bus;
++ unsigned int devfn;
++};
++static LIST_HEAD(pcistub_device_ids);
++static DEFINE_SPINLOCK(device_ids_lock);
++
++struct pcistub_device {
++ struct kref kref;
++ struct list_head dev_list;
++ spinlock_t lock;
++
++ struct pci_dev *dev;
++ struct pciback_device *pdev; /* non-NULL if struct pci_dev is in use */
++};
++
++/* Access to pcistub_devices & seized_devices lists and the initialize_devices
++ * flag must be locked with pcistub_devices_lock
++ */
++static DEFINE_SPINLOCK(pcistub_devices_lock);
++static LIST_HEAD(pcistub_devices);
++
++/* wait for device_initcall before initializing our devices
++ * (see pcistub_init_devices_late)
++ */
++static int initialize_devices = 0;
++static LIST_HEAD(seized_devices);
++
++static struct pcistub_device *pcistub_device_alloc(struct pci_dev *dev)
++{
++ struct pcistub_device *psdev;
++
++ dev_dbg(&dev->dev, "pcistub_device_alloc\n");
++
++ psdev = kzalloc(sizeof(*psdev), GFP_ATOMIC);
++ if (!psdev)
++ return NULL;
++
++ psdev->dev = pci_dev_get(dev);
++ if (!psdev->dev) {
++ kfree(psdev);
++ return NULL;
++ }
++
++ kref_init(&psdev->kref);
++ spin_lock_init(&psdev->lock);
++
++ return psdev;
++}
++
++/* Don't call this directly as it's called by pcistub_device_put */
++static void pcistub_device_release(struct kref *kref)
++{
++ struct pcistub_device *psdev;
++
++ psdev = container_of(kref, struct pcistub_device, kref);
++
++ dev_dbg(&psdev->dev->dev, "pcistub_device_release\n");
++
++ /* Clean-up the device */
++ pciback_reset_device(psdev->dev);
++ pciback_config_free_dyn_fields(psdev->dev);
++ pciback_config_free_dev(psdev->dev);
++ kfree(pci_get_drvdata(psdev->dev));
++ pci_set_drvdata(psdev->dev, NULL);
++
++ pci_dev_put(psdev->dev);
++
++ kfree(psdev);
++}
++
++static inline void pcistub_device_get(struct pcistub_device *psdev)
++{
++ kref_get(&psdev->kref);
++}
++
++static inline void pcistub_device_put(struct pcistub_device *psdev)
++{
++ kref_put(&psdev->kref, pcistub_device_release);
++}
++
++static struct pcistub_device *pcistub_device_find(int domain, int bus,
++ int slot, int func)
++{
++ struct pcistub_device *psdev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (psdev->dev != NULL
++ && domain == pci_domain_nr(psdev->dev->bus)
++ && bus == psdev->dev->bus->number
++ && PCI_DEVFN(slot, func) == psdev->dev->devfn) {
++ pcistub_device_get(psdev);
++ goto out;
++ }
++ }
++
++ /* didn't find it */
++ psdev = NULL;
++
++ out:
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++ return psdev;
++}
++
++static struct pci_dev *pcistub_device_get_pci_dev(struct pciback_device *pdev,
++ struct pcistub_device *psdev)
++{
++ struct pci_dev *pci_dev = NULL;
++ unsigned long flags;
++
++ pcistub_device_get(psdev);
++
++ spin_lock_irqsave(&psdev->lock, flags);
++ if (!psdev->pdev) {
++ psdev->pdev = pdev;
++ pci_dev = psdev->dev;
++ }
++ spin_unlock_irqrestore(&psdev->lock, flags);
++
++ if (!pci_dev)
++ pcistub_device_put(psdev);
++
++ return pci_dev;
++}
++
++struct pci_dev *pcistub_get_pci_dev_by_slot(struct pciback_device *pdev,
++ int domain, int bus,
++ int slot, int func)
++{
++ struct pcistub_device *psdev;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (psdev->dev != NULL
++ && domain == pci_domain_nr(psdev->dev->bus)
++ && bus == psdev->dev->bus->number
++ && PCI_DEVFN(slot, func) == psdev->dev->devfn) {
++ found_dev = pcistub_device_get_pci_dev(pdev, psdev);
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++ return found_dev;
++}
++
++struct pci_dev *pcistub_get_pci_dev(struct pciback_device *pdev,
++ struct pci_dev *dev)
++{
++ struct pcistub_device *psdev;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (psdev->dev == dev) {
++ found_dev = pcistub_device_get_pci_dev(pdev, psdev);
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++ return found_dev;
++}
++
++void pcistub_put_pci_dev(struct pci_dev *dev)
++{
++ struct pcistub_device *psdev, *found_psdev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (psdev->dev == dev) {
++ found_psdev = psdev;
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ /* Cleanup our device
++ * (so it's ready for the next domain)
++ */
++ pciback_reset_device(found_psdev->dev);
++ pciback_config_free_dyn_fields(found_psdev->dev);
++ pciback_config_reset_dev(found_psdev->dev);
++
++ spin_lock_irqsave(&found_psdev->lock, flags);
++ found_psdev->pdev = NULL;
++ spin_unlock_irqrestore(&found_psdev->lock, flags);
++
++ pcistub_device_put(found_psdev);
++}
++
++static int __devinit pcistub_match_one(struct pci_dev *dev,
++ struct pcistub_device_id *pdev_id)
++{
++ /* Match the specified device by domain, bus, slot, func and also if
++ * any of the device's parent bridges match.
++ */
++ for (; dev != NULL; dev = dev->bus->self) {
++ if (pci_domain_nr(dev->bus) == pdev_id->domain
++ && dev->bus->number == pdev_id->bus
++ && dev->devfn == pdev_id->devfn)
++ return 1;
++
++ /* Sometimes topmost bridge links to itself. */
++ if (dev == dev->bus->self)
++ break;
++ }
++
++ return 0;
++}
++
++static int __devinit pcistub_match(struct pci_dev *dev)
++{
++ struct pcistub_device_id *pdev_id;
++ unsigned long flags;
++ int found = 0;
++
++ spin_lock_irqsave(&device_ids_lock, flags);
++ list_for_each_entry(pdev_id, &pcistub_device_ids, slot_list) {
++ if (pcistub_match_one(dev, pdev_id)) {
++ found = 1;
++ break;
++ }
++ }
++ spin_unlock_irqrestore(&device_ids_lock, flags);
++
++ return found;
++}
++
++static int __devinit pcistub_init_device(struct pci_dev *dev)
++{
++ struct pciback_dev_data *dev_data;
++ int err = 0;
++
++ dev_dbg(&dev->dev, "initializing...\n");
++
++ /* The PCI backend is not intended to be a module (or to work with
++ * removable PCI devices (yet). If it were, pciback_config_free()
++ * would need to be called somewhere to free the memory allocated
++ * here and then to call kfree(pci_get_drvdata(psdev->dev)).
++ */
++ dev_data = kzalloc(sizeof(*dev_data), GFP_ATOMIC);
++ if (!dev_data) {
++ err = -ENOMEM;
++ goto out;
++ }
++ pci_set_drvdata(dev, dev_data);
++
++ dev_dbg(&dev->dev, "initializing config\n");
++ err = pciback_config_init_dev(dev);
++ if (err)
++ goto out;
++
++ /* HACK: Force device (& ACPI) to determine what IRQ it's on - we
++ * must do this here because pcibios_enable_device may specify
++ * the pci device's true irq (and possibly its other resources)
++ * if they differ from what's in the configuration space.
++ * This makes the assumption that the device's resources won't
++ * change after this point (otherwise this code may break!)
++ */
++ dev_dbg(&dev->dev, "enabling device\n");
++ err = pci_enable_device(dev);
++ if (err)
++ goto config_release;
++
++ /* Now disable the device (this also ensures some private device
++ * data is setup before we export)
++ */
++ dev_dbg(&dev->dev, "reset device\n");
++ pciback_reset_device(dev);
++
++ return 0;
++
++ config_release:
++ pciback_config_free_dev(dev);
++
++ out:
++ pci_set_drvdata(dev, NULL);
++ kfree(dev_data);
++ return err;
++}
++
++/*
++ * Because some initialization still happens on
++ * devices during fs_initcall, we need to defer
++ * full initialization of our devices until
++ * device_initcall.
++ */
++static int __init pcistub_init_devices_late(void)
++{
++ struct pcistub_device *psdev;
++ unsigned long flags;
++ int err = 0;
++
++ pr_debug("pciback: pcistub_init_devices_late\n");
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ while (!list_empty(&seized_devices)) {
++ psdev = container_of(seized_devices.next,
++ struct pcistub_device, dev_list);
++ list_del(&psdev->dev_list);
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ err = pcistub_init_device(psdev->dev);
++ if (err) {
++ dev_err(&psdev->dev->dev,
++ "error %d initializing device\n", err);
++ kfree(psdev);
++ psdev = NULL;
++ }
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ if (psdev)
++ list_add_tail(&psdev->dev_list, &pcistub_devices);
++ }
++
++ initialize_devices = 1;
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ return 0;
++}
++
++static int __devinit pcistub_seize(struct pci_dev *dev)
++{
++ struct pcistub_device *psdev;
++ unsigned long flags;
++ int err = 0;
++
++ psdev = pcistub_device_alloc(dev);
++ if (!psdev)
++ return -ENOMEM;
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ if (initialize_devices) {
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ /* don't want irqs disabled when calling pcistub_init_device */
++ err = pcistub_init_device(psdev->dev);
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ if (!err)
++ list_add(&psdev->dev_list, &pcistub_devices);
++ } else {
++ dev_dbg(&dev->dev, "deferring initialization\n");
++ list_add(&psdev->dev_list, &seized_devices);
++ }
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ if (err)
++ pcistub_device_put(psdev);
++
++ return err;
++}
++
++static int __devinit pcistub_probe(struct pci_dev *dev,
++ const struct pci_device_id *id)
++{
++ int err = 0;
++
++ dev_dbg(&dev->dev, "probing...\n");
++
++ if (pcistub_match(dev)) {
++
++ if (dev->hdr_type != PCI_HEADER_TYPE_NORMAL
++ && dev->hdr_type != PCI_HEADER_TYPE_BRIDGE) {
++ dev_err(&dev->dev, "can't export pci devices that "
++ "don't have a normal (0) or bridge (1) "
++ "header type!\n");
++ err = -ENODEV;
++ goto out;
++ }
++
++ dev_info(&dev->dev, "seizing device\n");
++ err = pcistub_seize(dev);
++ } else
++ /* Didn't find the device */
++ err = -ENODEV;
++
++ out:
++ return err;
++}
++
++static void pcistub_remove(struct pci_dev *dev)
++{
++ struct pcistub_device *psdev, *found_psdev = NULL;
++ unsigned long flags;
++
++ dev_dbg(&dev->dev, "removing\n");
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++
++ pciback_config_quirk_release(dev);
++
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (psdev->dev == dev) {
++ found_psdev = psdev;
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ if (found_psdev) {
++ dev_dbg(&dev->dev, "found device to remove - in use? %p\n",
++ found_psdev->pdev);
++
++ if (found_psdev->pdev) {
++ printk(KERN_WARNING "pciback: ****** removing device "
++ "%s while still in-use! ******\n",
++ pci_name(found_psdev->dev));
++ printk(KERN_WARNING "pciback: ****** driver domain may "
++ "still access this device's i/o resources!\n");
++ printk(KERN_WARNING "pciback: ****** shutdown driver "
++ "domain before binding device\n");
++ printk(KERN_WARNING "pciback: ****** to other drivers "
++ "or domains\n");
++
++ pciback_release_pci_dev(found_psdev->pdev,
++ found_psdev->dev);
++ }
++
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++ list_del(&found_psdev->dev_list);
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++
++ /* the final put for releasing from the list */
++ pcistub_device_put(found_psdev);
++ }
++}
++
++static struct pci_device_id pcistub_ids[] = {
++ {
++ .vendor = PCI_ANY_ID,
++ .device = PCI_ANY_ID,
++ .subvendor = PCI_ANY_ID,
++ .subdevice = PCI_ANY_ID,
++ },
++ {0,},
++};
++
++/*
++ * Note: There is no MODULE_DEVICE_TABLE entry here because this isn't
++ * for a normal device. I don't want it to be loaded automatically.
++ */
++
++static struct pci_driver pciback_pci_driver = {
++ .name = "pciback",
++ .id_table = pcistub_ids,
++ .probe = pcistub_probe,
++ .remove = pcistub_remove,
++};
++
++static inline int str_to_slot(const char *buf, int *domain, int *bus,
++ int *slot, int *func)
++{
++ int err;
++
++ err = sscanf(buf, " %x:%x:%x.%x", domain, bus, slot, func);
++ if (err == 4)
++ return 0;
++ else if (err < 0)
++ return -EINVAL;
++
++ /* try again without domain */
++ *domain = 0;
++ err = sscanf(buf, " %x:%x.%x", bus, slot, func);
++ if (err == 3)
++ return 0;
++
++ return -EINVAL;
++}
++
++static inline int str_to_quirk(const char *buf, int *domain, int *bus, int
++ *slot, int *func, int *reg, int *size, int *mask)
++{
++ int err;
++
++ err =
++ sscanf(buf, " %04x:%02x:%02x.%1x-%08x:%1x:%08x", domain, bus, slot,
++ func, reg, size, mask);
++ if (err == 7)
++ return 0;
++ return -EINVAL;
++}
++
++static int pcistub_device_id_add(int domain, int bus, int slot, int func)
++{
++ struct pcistub_device_id *pci_dev_id;
++ unsigned long flags;
++
++ pci_dev_id = kmalloc(sizeof(*pci_dev_id), GFP_KERNEL);
++ if (!pci_dev_id)
++ return -ENOMEM;
++
++ pci_dev_id->domain = domain;
++ pci_dev_id->bus = bus;
++ pci_dev_id->devfn = PCI_DEVFN(slot, func);
++
++ pr_debug("pciback: wants to seize %04x:%02x:%02x.%01x\n",
++ domain, bus, slot, func);
++
++ spin_lock_irqsave(&device_ids_lock, flags);
++ list_add_tail(&pci_dev_id->slot_list, &pcistub_device_ids);
++ spin_unlock_irqrestore(&device_ids_lock, flags);
++
++ return 0;
++}
++
++static int pcistub_device_id_remove(int domain, int bus, int slot, int func)
++{
++ struct pcistub_device_id *pci_dev_id, *t;
++ int devfn = PCI_DEVFN(slot, func);
++ int err = -ENOENT;
++ unsigned long flags;
++
++ spin_lock_irqsave(&device_ids_lock, flags);
++ list_for_each_entry_safe(pci_dev_id, t, &pcistub_device_ids, slot_list) {
++
++ if (pci_dev_id->domain == domain
++ && pci_dev_id->bus == bus && pci_dev_id->devfn == devfn) {
++ /* Don't break; here because it's possible the same
++ * slot could be in the list more than once
++ */
++ list_del(&pci_dev_id->slot_list);
++ kfree(pci_dev_id);
++
++ err = 0;
++
++ pr_debug("pciback: removed %04x:%02x:%02x.%01x from "
++ "seize list\n", domain, bus, slot, func);
++ }
++ }
++ spin_unlock_irqrestore(&device_ids_lock, flags);
++
++ return err;
++}
++
++static int pcistub_reg_add(int domain, int bus, int slot, int func, int reg,
++ int size, int mask)
++{
++ int err = 0;
++ struct pcistub_device *psdev;
++ struct pci_dev *dev;
++ struct config_field *field;
++
++ psdev = pcistub_device_find(domain, bus, slot, func);
++ if (!psdev || !psdev->dev) {
++ err = -ENODEV;
++ goto out;
++ }
++ dev = psdev->dev;
++
++ field = kzalloc(sizeof(*field), GFP_ATOMIC);
++ if (!field) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ field->offset = reg;
++ field->size = size;
++ field->mask = mask;
++ field->init = NULL;
++ field->reset = NULL;
++ field->release = NULL;
++ field->clean = pciback_config_field_free;
++
++ err = pciback_config_quirks_add_field(dev, field);
++ if (err)
++ kfree(field);
++ out:
++ return err;
++}
++
++static ssize_t pcistub_slot_add(struct device_driver *drv, const char *buf,
++ size_t count)
++{
++ int domain, bus, slot, func;
++ int err;
++
++ err = str_to_slot(buf, &domain, &bus, &slot, &func);
++ if (err)
++ goto out;
++
++ err = pcistub_device_id_add(domain, bus, slot, func);
++
++ out:
++ if (!err)
++ err = count;
++ return err;
++}
++
++DRIVER_ATTR(new_slot, S_IWUSR, NULL, pcistub_slot_add);
++
++static ssize_t pcistub_slot_remove(struct device_driver *drv, const char *buf,
++ size_t count)
++{
++ int domain, bus, slot, func;
++ int err;
++
++ err = str_to_slot(buf, &domain, &bus, &slot, &func);
++ if (err)
++ goto out;
++
++ err = pcistub_device_id_remove(domain, bus, slot, func);
++
++ out:
++ if (!err)
++ err = count;
++ return err;
++}
++
++DRIVER_ATTR(remove_slot, S_IWUSR, NULL, pcistub_slot_remove);
++
++static ssize_t pcistub_slot_show(struct device_driver *drv, char *buf)
++{
++ struct pcistub_device_id *pci_dev_id;
++ size_t count = 0;
++ unsigned long flags;
++
++ spin_lock_irqsave(&device_ids_lock, flags);
++ list_for_each_entry(pci_dev_id, &pcistub_device_ids, slot_list) {
++ if (count >= PAGE_SIZE)
++ break;
++
++ count += scnprintf(buf + count, PAGE_SIZE - count,
++ "%04x:%02x:%02x.%01x\n",
++ pci_dev_id->domain, pci_dev_id->bus,
++ PCI_SLOT(pci_dev_id->devfn),
++ PCI_FUNC(pci_dev_id->devfn));
++ }
++ spin_unlock_irqrestore(&device_ids_lock, flags);
++
++ return count;
++}
++
++DRIVER_ATTR(slots, S_IRUSR, pcistub_slot_show, NULL);
++
++static ssize_t pcistub_quirk_add(struct device_driver *drv, const char *buf,
++ size_t count)
++{
++ int domain, bus, slot, func, reg, size, mask;
++ int err;
++
++ err = str_to_quirk(buf, &domain, &bus, &slot, &func, ®, &size,
++ &mask);
++ if (err)
++ goto out;
++
++ err = pcistub_reg_add(domain, bus, slot, func, reg, size, mask);
++
++ out:
++ if (!err)
++ err = count;
++ return err;
++}
++
++static ssize_t pcistub_quirk_show(struct device_driver *drv, char *buf)
++{
++ int count = 0;
++ unsigned long flags;
++ extern struct list_head pciback_quirks;
++ struct pciback_config_quirk *quirk;
++ struct pciback_dev_data *dev_data;
++ struct config_field *field;
++ struct config_field_entry *cfg_entry;
++
++ spin_lock_irqsave(&device_ids_lock, flags);
++ list_for_each_entry(quirk, &pciback_quirks, quirks_list) {
++ if (count >= PAGE_SIZE)
++ goto out;
++
++ count += scnprintf(buf + count, PAGE_SIZE - count,
++ "%02x:%02x.%01x\n\t%04x:%04x:%04x:%04x\n",
++ quirk->pdev->bus->number,
++ PCI_SLOT(quirk->pdev->devfn),
++ PCI_FUNC(quirk->pdev->devfn),
++ quirk->devid.vendor, quirk->devid.device,
++ quirk->devid.subvendor,
++ quirk->devid.subdevice);
++
++ dev_data = pci_get_drvdata(quirk->pdev);
++
++ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
++ field = cfg_entry->field;
++ if (count >= PAGE_SIZE)
++ goto out;
++
++ count += scnprintf(buf + count, PAGE_SIZE - count,
++ "\t\t%08x:%01x:%08x\n",
++ cfg_entry->base_offset + field->offset,
++ field->size, field->mask);
++ }
++ }
++
++ out:
++ spin_unlock_irqrestore(&device_ids_lock, flags);
++
++ return count;
++}
++
++DRIVER_ATTR(quirks, S_IRUSR | S_IWUSR, pcistub_quirk_show, pcistub_quirk_add);
++
++static ssize_t permissive_add(struct device_driver *drv, const char *buf,
++ size_t count)
++{
++ int domain, bus, slot, func;
++ int err;
++ struct pcistub_device *psdev;
++ struct pciback_dev_data *dev_data;
++ err = str_to_slot(buf, &domain, &bus, &slot, &func);
++ if (err)
++ goto out;
++ psdev = pcistub_device_find(domain, bus, slot, func);
++ if (!psdev) {
++ err = -ENODEV;
++ goto out;
++ }
++ if (!psdev->dev) {
++ err = -ENODEV;
++ goto release;
++ }
++ dev_data = pci_get_drvdata(psdev->dev);
++ /* the driver data for a device should never be null at this point */
++ if (!dev_data) {
++ err = -ENXIO;
++ goto release;
++ }
++ if (!dev_data->permissive) {
++ dev_data->permissive = 1;
++ /* Let user know that what they're doing could be unsafe */
++ dev_warn(&psdev->dev->dev,
++ "enabling permissive mode configuration space accesses!\n");
++ dev_warn(&psdev->dev->dev,
++ "permissive mode is potentially unsafe!\n");
++ }
++ release:
++ pcistub_device_put(psdev);
++ out:
++ if (!err)
++ err = count;
++ return err;
++}
++
++static ssize_t permissive_show(struct device_driver *drv, char *buf)
++{
++ struct pcistub_device *psdev;
++ struct pciback_dev_data *dev_data;
++ size_t count = 0;
++ unsigned long flags;
++ spin_lock_irqsave(&pcistub_devices_lock, flags);
++ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
++ if (count >= PAGE_SIZE)
++ break;
++ if (!psdev->dev)
++ continue;
++ dev_data = pci_get_drvdata(psdev->dev);
++ if (!dev_data || !dev_data->permissive)
++ continue;
++ count +=
++ scnprintf(buf + count, PAGE_SIZE - count, "%s\n",
++ pci_name(psdev->dev));
++ }
++ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++ return count;
++}
++
++DRIVER_ATTR(permissive, S_IRUSR | S_IWUSR, permissive_show, permissive_add);
++
++#ifdef CONFIG_PCI_MSI
++
++int pciback_get_owner(struct pci_dev *dev)
++{
++ struct pcistub_device *psdev;
++
++ psdev = pcistub_device_find(pci_domain_nr(dev->bus), dev->bus->number,
++ PCI_SLOT(dev->devfn), PCI_FUNC(dev->devfn));
++ /* XXX will other domain has pciback support ??? */
++ if (!psdev || !psdev->pdev) {
++ printk(KERN_WARNING "no ownder\n");
++ return -1;
++ }
++ return psdev->pdev->xdev->otherend_id;
++}
++#endif
++
++static void pcistub_exit(void)
++{
++ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_new_slot);
++ driver_remove_file(&pciback_pci_driver.driver,
++ &driver_attr_remove_slot);
++ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_slots);
++ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_quirks);
++ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_permissive);
++
++ pci_unregister_driver(&pciback_pci_driver);
++#ifdef CONFIG_PCI_MSI
++ unregister_msi_get_owner(pciback_get_owner);
++#endif
++}
++
++static int __init pcistub_init(void)
++{
++ int pos = 0;
++ int err = 0;
++ int domain, bus, slot, func;
++ int parsed;
++
++ if (pci_devs_to_hide && *pci_devs_to_hide) {
++ do {
++ parsed = 0;
++
++ err = sscanf(pci_devs_to_hide + pos,
++ " (%x:%x:%x.%x) %n",
++ &domain, &bus, &slot, &func, &parsed);
++ if (err != 4) {
++ domain = 0;
++ err = sscanf(pci_devs_to_hide + pos,
++ " (%x:%x.%x) %n",
++ &bus, &slot, &func, &parsed);
++ if (err != 3)
++ goto parse_error;
++ }
++
++ err = pcistub_device_id_add(domain, bus, slot, func);
++ if (err)
++ goto out;
++
++ /* if parsed<=0, we've reached the end of the string */
++ pos += parsed;
++ } while (parsed > 0 && pci_devs_to_hide[pos]);
++ }
++
++ /* If we're the first PCI Device Driver to register, we're the
++ * first one to get offered PCI devices as they become
++ * available (and thus we can be the first to grab them)
++ */
++ err = pci_register_driver(&pciback_pci_driver);
++ if (err < 0)
++ goto out;
++
++ err = driver_create_file(&pciback_pci_driver.driver,
++ &driver_attr_new_slot);
++ if (!err)
++ err = driver_create_file(&pciback_pci_driver.driver,
++ &driver_attr_remove_slot);
++ if (!err)
++ err = driver_create_file(&pciback_pci_driver.driver,
++ &driver_attr_slots);
++ if (!err)
++ err = driver_create_file(&pciback_pci_driver.driver,
++ &driver_attr_quirks);
++ if (!err)
++ err = driver_create_file(&pciback_pci_driver.driver,
++ &driver_attr_permissive);
++
++#ifdef CONFIG_PCI_MSI
++ if (!err)
++ err = register_msi_get_owner(pciback_get_owner);
++#endif
++ if (err)
++ pcistub_exit();
++
++ out:
++ return err;
++
++ parse_error:
++ printk(KERN_ERR "pciback: Error parsing pci_devs_to_hide at \"%s\"\n",
++ pci_devs_to_hide + pos);
++ return -EINVAL;
++}
++
++#ifndef MODULE
++/*
++ * fs_initcall happens before device_initcall
++ * so pciback *should* get called first (b/c we
++ * want to suck up any device before other drivers
++ * get a chance by being the first pci device
++ * driver to register)
++ */
++fs_initcall(pcistub_init);
++#endif
++
++static int __init pciback_init(void)
++{
++ int err;
++
++ err = pciback_config_init();
++ if (err)
++ return err;
++
++#ifdef MODULE
++ err = pcistub_init();
++ if (err < 0)
++ return err;
++#endif
++
++ pcistub_init_devices_late();
++ err = pciback_xenbus_register();
++ if (err)
++ pcistub_exit();
++
++ return err;
++}
++
++static void __exit pciback_cleanup(void)
++{
++ pciback_xenbus_unregister();
++ pcistub_exit();
++}
++
++module_init(pciback_init);
++module_exit(pciback_cleanup);
++
++MODULE_LICENSE("Dual BSD/GPL");
+--- linux-2.6.18.8/drivers/xen/pciback/pciback.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/pciback.h 2008-05-19 00:33:46.266812993 +0300
+@@ -0,0 +1,111 @@
++/*
++ * PCI Backend Common Data Structures & Function Declarations
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#ifndef __XEN_PCIBACK_H__
++#define __XEN_PCIBACK_H__
++
++#include <linux/pci.h>
++#include <linux/interrupt.h>
++#include <xen/xenbus.h>
++#include <linux/list.h>
++#include <linux/spinlock.h>
++#include <linux/workqueue.h>
++#include <asm/atomic.h>
++#include <xen/interface/io/pciif.h>
++
++struct pci_dev_entry {
++ struct list_head list;
++ struct pci_dev *dev;
++};
++
++#define _PDEVF_op_active (0)
++#define PDEVF_op_active (1<<(_PDEVF_op_active))
++
++struct pciback_device {
++ void *pci_dev_data;
++ spinlock_t dev_lock;
++
++ struct xenbus_device *xdev;
++
++ struct xenbus_watch be_watch;
++ u8 be_watching;
++
++ int evtchn_irq;
++
++ struct vm_struct *sh_area;
++ struct xen_pci_sharedinfo *sh_info;
++
++ unsigned long flags;
++
++ struct work_struct op_work;
++};
++
++struct pciback_dev_data {
++ struct list_head config_fields;
++ int permissive;
++ int warned_on_write;
++};
++
++/* Get/Put PCI Devices that are hidden from the PCI Backend Domain */
++struct pci_dev *pcistub_get_pci_dev_by_slot(struct pciback_device *pdev,
++ int domain, int bus,
++ int slot, int func);
++struct pci_dev *pcistub_get_pci_dev(struct pciback_device *pdev,
++ struct pci_dev *dev);
++void pcistub_put_pci_dev(struct pci_dev *dev);
++
++/* Ensure a device is turned off or reset */
++void pciback_reset_device(struct pci_dev *pdev);
++
++/* Access a virtual configuration space for a PCI device */
++int pciback_config_init(void);
++int pciback_config_init_dev(struct pci_dev *dev);
++void pciback_config_free_dyn_fields(struct pci_dev *dev);
++void pciback_config_reset_dev(struct pci_dev *dev);
++void pciback_config_free_dev(struct pci_dev *dev);
++int pciback_config_read(struct pci_dev *dev, int offset, int size,
++ u32 * ret_val);
++int pciback_config_write(struct pci_dev *dev, int offset, int size, u32 value);
++
++/* Handle requests for specific devices from the frontend */
++typedef int (*publish_pci_dev_cb) (struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn, unsigned int devid);
++typedef int (*publish_pci_root_cb) (struct pciback_device * pdev,
++ unsigned int domain, unsigned int bus);
++int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev,
++ int devid, publish_pci_dev_cb publish_cb);
++void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev);
++struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn);
++int pciback_init_devices(struct pciback_device *pdev);
++int pciback_publish_pci_roots(struct pciback_device *pdev,
++ publish_pci_root_cb cb);
++void pciback_release_devices(struct pciback_device *pdev);
++
++/* Handles events from front-end */
++irqreturn_t pciback_handle_event(int irq, void *dev_id, struct pt_regs *regs);
++void pciback_do_op(void *data);
++
++int pciback_xenbus_register(void);
++void pciback_xenbus_unregister(void);
++
++#ifdef CONFIG_PCI_MSI
++int pciback_enable_msi(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op);
++
++int pciback_disable_msi(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op);
++
++
++int pciback_enable_msix(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op);
++
++int pciback_disable_msix(struct pciback_device *pdev,
++ struct pci_dev *dev, struct xen_pci_op *op);
++#endif
++extern int verbose_request;
++#endif
+--- linux-2.6.18.8/drivers/xen/pciback/pciback_ops.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/pciback_ops.c 2008-05-19 00:33:46.266812993 +0300
+@@ -0,0 +1,117 @@
++/*
++ * PCI Backend Operations - respond to PCI requests from Frontend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <asm/bitops.h>
++#include <xen/evtchn.h>
++#include "pciback.h"
++
++int verbose_request = 0;
++module_param(verbose_request, int, 0644);
++
++/* Ensure a device is "turned off" and ready to be exported.
++ * (Also see pciback_config_reset to ensure virtual configuration space is
++ * ready to be re-exported)
++ */
++void pciback_reset_device(struct pci_dev *dev)
++{
++ u16 cmd;
++
++ /* Disable devices (but not bridges) */
++ if (dev->hdr_type == PCI_HEADER_TYPE_NORMAL) {
++ pci_disable_device(dev);
++
++ pci_write_config_word(dev, PCI_COMMAND, 0);
++
++ dev->is_enabled = 0;
++ dev->is_busmaster = 0;
++ } else {
++ pci_read_config_word(dev, PCI_COMMAND, &cmd);
++ if (cmd & (PCI_COMMAND_INVALIDATE)) {
++ cmd &= ~(PCI_COMMAND_INVALIDATE);
++ pci_write_config_word(dev, PCI_COMMAND, cmd);
++
++ dev->is_busmaster = 0;
++ }
++ }
++}
++
++static inline void test_and_schedule_op(struct pciback_device *pdev)
++{
++ /* Check that frontend is requesting an operation and that we are not
++ * already processing a request */
++ if (test_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags)
++ && !test_and_set_bit(_PDEVF_op_active, &pdev->flags))
++ schedule_work(&pdev->op_work);
++}
++
++/* Performing the configuration space reads/writes must not be done in atomic
++ * context because some of the pci_* functions can sleep (mostly due to ACPI
++ * use of semaphores). This function is intended to be called from a work
++ * queue in process context taking a struct pciback_device as a parameter */
++void pciback_do_op(void *data)
++{
++ struct pciback_device *pdev = data;
++ struct pci_dev *dev;
++ struct xen_pci_op *op = &pdev->sh_info->op;
++
++ dev = pciback_get_pci_dev(pdev, op->domain, op->bus, op->devfn);
++
++ if (dev == NULL)
++ op->err = XEN_PCI_ERR_dev_not_found;
++ else
++ {
++ switch (op->cmd)
++ {
++ case XEN_PCI_OP_conf_read:
++ op->err = pciback_config_read(dev,
++ op->offset, op->size, &op->value);
++ break;
++ case XEN_PCI_OP_conf_write:
++ op->err = pciback_config_write(dev,
++ op->offset, op->size, op->value);
++ break;
++#ifdef CONFIG_PCI_MSI
++ case XEN_PCI_OP_enable_msi:
++ op->err = pciback_enable_msi(pdev, dev, op);
++ break;
++ case XEN_PCI_OP_disable_msi:
++ op->err = pciback_disable_msi(pdev, dev, op);
++ break;
++ case XEN_PCI_OP_enable_msix:
++ op->err = pciback_enable_msix(pdev, dev, op);
++ break;
++ case XEN_PCI_OP_disable_msix:
++ op->err = pciback_disable_msix(pdev, dev, op);
++ break;
++#endif
++ default:
++ op->err = XEN_PCI_ERR_not_implemented;
++ break;
++ }
++ }
++ /* Tell the driver domain that we're done. */
++ wmb();
++ clear_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags);
++ notify_remote_via_irq(pdev->evtchn_irq);
++
++ /* Mark that we're done. */
++ smp_mb__before_clear_bit(); /* /after/ clearing PCIF_active */
++ clear_bit(_PDEVF_op_active, &pdev->flags);
++ smp_mb__after_clear_bit(); /* /before/ final check for work */
++
++ /* Check to see if the driver domain tried to start another request in
++ * between clearing _XEN_PCIF_active and clearing _PDEVF_op_active. */
++ test_and_schedule_op(pdev);
++}
++
++irqreturn_t pciback_handle_event(int irq, void *dev_id, struct pt_regs *regs)
++{
++ struct pciback_device *pdev = dev_id;
++
++ test_and_schedule_op(pdev);
++
++ return IRQ_HANDLED;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/slot.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/slot.c 2008-05-19 00:33:46.266812993 +0300
+@@ -0,0 +1,157 @@
++/*
++ * PCI Backend - Provides a Virtual PCI bus (with real devices)
++ * to the frontend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil> (vpci.c)
++ * Author: Tristan Gingold <tristan.gingold@bull.net>, from vpci.c
++ */
++
++#include <linux/list.h>
++#include <linux/slab.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include "pciback.h"
++
++/* There are at most 32 slots in a pci bus. */
++#define PCI_SLOT_MAX 32
++
++#define PCI_BUS_NBR 2
++
++struct slot_dev_data {
++ /* Access to dev_list must be protected by lock */
++ struct pci_dev *slots[PCI_BUS_NBR][PCI_SLOT_MAX];
++ spinlock_t lock;
++};
++
++struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn)
++{
++ struct pci_dev *dev = NULL;
++ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
++ unsigned long flags;
++
++ if (domain != 0 || PCI_FUNC(devfn) != 0)
++ return NULL;
++
++ if (PCI_SLOT(devfn) >= PCI_SLOT_MAX || bus >= PCI_BUS_NBR)
++ return NULL;
++
++ spin_lock_irqsave(&slot_dev->lock, flags);
++ dev = slot_dev->slots[bus][PCI_SLOT(devfn)];
++ spin_unlock_irqrestore(&slot_dev->lock, flags);
++
++ return dev;
++}
++
++int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev,
++ int devid, publish_pci_dev_cb publish_cb)
++{
++ int err = 0, slot, bus;
++ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
++ unsigned long flags;
++
++ if ((dev->class >> 24) == PCI_BASE_CLASS_BRIDGE) {
++ err = -EFAULT;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Can't export bridges on the virtual PCI bus");
++ goto out;
++ }
++
++ spin_lock_irqsave(&slot_dev->lock, flags);
++
++ /* Assign to a new slot on the virtual PCI bus */
++ for (bus = 0; bus < PCI_BUS_NBR; bus++)
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ if (slot_dev->slots[bus][slot] == NULL) {
++ printk(KERN_INFO
++ "pciback: slot: %s: assign to virtual slot %d, bus %d\n",
++ pci_name(dev), slot, bus);
++ slot_dev->slots[bus][slot] = dev;
++ goto unlock;
++ }
++ }
++
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "No more space on root virtual PCI bus");
++
++ unlock:
++ spin_unlock_irqrestore(&slot_dev->lock, flags);
++
++ /* Publish this device. */
++ if(!err)
++ err = publish_cb(pdev, 0, 0, PCI_DEVFN(slot, 0), devid);
++
++ out:
++ return err;
++}
++
++void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
++{
++ int slot, bus;
++ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&slot_dev->lock, flags);
++
++ for (bus = 0; bus < PCI_BUS_NBR; bus++)
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ if (slot_dev->slots[bus][slot] == dev) {
++ slot_dev->slots[bus][slot] = NULL;
++ found_dev = dev;
++ goto out;
++ }
++ }
++
++ out:
++ spin_unlock_irqrestore(&slot_dev->lock, flags);
++
++ if (found_dev)
++ pcistub_put_pci_dev(found_dev);
++}
++
++int pciback_init_devices(struct pciback_device *pdev)
++{
++ int slot, bus;
++ struct slot_dev_data *slot_dev;
++
++ slot_dev = kmalloc(sizeof(*slot_dev), GFP_KERNEL);
++ if (!slot_dev)
++ return -ENOMEM;
++
++ spin_lock_init(&slot_dev->lock);
++
++ for (bus = 0; bus < PCI_BUS_NBR; bus++)
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++)
++ slot_dev->slots[bus][slot] = NULL;
++
++ pdev->pci_dev_data = slot_dev;
++
++ return 0;
++}
++
++int pciback_publish_pci_roots(struct pciback_device *pdev,
++ publish_pci_root_cb publish_cb)
++{
++ /* The Virtual PCI bus has only one root */
++ return publish_cb(pdev, 0, 0);
++}
++
++void pciback_release_devices(struct pciback_device *pdev)
++{
++ int slot, bus;
++ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
++ struct pci_dev *dev;
++
++ for (bus = 0; bus < PCI_BUS_NBR; bus++)
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ dev = slot_dev->slots[bus][slot];
++ if (dev != NULL)
++ pcistub_put_pci_dev(dev);
++ }
++
++ kfree(slot_dev);
++ pdev->pci_dev_data = NULL;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/vpci.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/vpci.c 2008-05-19 00:33:46.266812993 +0300
+@@ -0,0 +1,212 @@
++/*
++ * PCI Backend - Provides a Virtual PCI bus (with real devices)
++ * to the frontend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++
++#include <linux/list.h>
++#include <linux/slab.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include "pciback.h"
++
++#define PCI_SLOT_MAX 32
++
++struct vpci_dev_data {
++ /* Access to dev_list must be protected by lock */
++ struct list_head dev_list[PCI_SLOT_MAX];
++ spinlock_t lock;
++};
++
++static inline struct list_head *list_first(struct list_head *head)
++{
++ return head->next;
++}
++
++struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn)
++{
++ struct pci_dev_entry *entry;
++ struct pci_dev *dev = NULL;
++ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
++ unsigned long flags;
++
++ if (domain != 0 || bus != 0)
++ return NULL;
++
++ if (PCI_SLOT(devfn) < PCI_SLOT_MAX) {
++ spin_lock_irqsave(&vpci_dev->lock, flags);
++
++ list_for_each_entry(entry,
++ &vpci_dev->dev_list[PCI_SLOT(devfn)],
++ list) {
++ if (PCI_FUNC(entry->dev->devfn) == PCI_FUNC(devfn)) {
++ dev = entry->dev;
++ break;
++ }
++ }
++
++ spin_unlock_irqrestore(&vpci_dev->lock, flags);
++ }
++ return dev;
++}
++
++static inline int match_slot(struct pci_dev *l, struct pci_dev *r)
++{
++ if (pci_domain_nr(l->bus) == pci_domain_nr(r->bus)
++ && l->bus == r->bus && PCI_SLOT(l->devfn) == PCI_SLOT(r->devfn))
++ return 1;
++
++ return 0;
++}
++
++int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev,
++ int devid, publish_pci_dev_cb publish_cb)
++{
++ int err = 0, slot, func;
++ struct pci_dev_entry *t, *dev_entry;
++ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
++ unsigned long flags;
++
++ if ((dev->class >> 24) == PCI_BASE_CLASS_BRIDGE) {
++ err = -EFAULT;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Can't export bridges on the virtual PCI bus");
++ goto out;
++ }
++
++ dev_entry = kmalloc(sizeof(*dev_entry), GFP_KERNEL);
++ if (!dev_entry) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error adding entry to virtual PCI bus");
++ goto out;
++ }
++
++ dev_entry->dev = dev;
++
++ spin_lock_irqsave(&vpci_dev->lock, flags);
++
++ /* Keep multi-function devices together on the virtual PCI bus */
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ if (!list_empty(&vpci_dev->dev_list[slot])) {
++ t = list_entry(list_first(&vpci_dev->dev_list[slot]),
++ struct pci_dev_entry, list);
++
++ if (match_slot(dev, t->dev)) {
++ pr_info("pciback: vpci: %s: "
++ "assign to virtual slot %d func %d\n",
++ pci_name(dev), slot,
++ PCI_FUNC(dev->devfn));
++ list_add_tail(&dev_entry->list,
++ &vpci_dev->dev_list[slot]);
++ func = PCI_FUNC(dev->devfn);
++ goto unlock;
++ }
++ }
++ }
++
++ /* Assign to a new slot on the virtual PCI bus */
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ if (list_empty(&vpci_dev->dev_list[slot])) {
++ printk(KERN_INFO
++ "pciback: vpci: %s: assign to virtual slot %d\n",
++ pci_name(dev), slot);
++ list_add_tail(&dev_entry->list,
++ &vpci_dev->dev_list[slot]);
++ func = PCI_FUNC(dev->devfn);
++ goto unlock;
++ }
++ }
++
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "No more space on root virtual PCI bus");
++
++ unlock:
++ spin_unlock_irqrestore(&vpci_dev->lock, flags);
++
++ /* Publish this device. */
++ if(!err)
++ err = publish_cb(pdev, 0, 0, PCI_DEVFN(slot, func), devid);
++
++ out:
++ return err;
++}
++
++void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
++{
++ int slot;
++ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
++ struct pci_dev *found_dev = NULL;
++ unsigned long flags;
++
++ spin_lock_irqsave(&vpci_dev->lock, flags);
++
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ struct pci_dev_entry *e, *tmp;
++ list_for_each_entry_safe(e, tmp, &vpci_dev->dev_list[slot],
++ list) {
++ if (e->dev == dev) {
++ list_del(&e->list);
++ found_dev = e->dev;
++ kfree(e);
++ goto out;
++ }
++ }
++ }
++
++ out:
++ spin_unlock_irqrestore(&vpci_dev->lock, flags);
++
++ if (found_dev)
++ pcistub_put_pci_dev(found_dev);
++}
++
++int pciback_init_devices(struct pciback_device *pdev)
++{
++ int slot;
++ struct vpci_dev_data *vpci_dev;
++
++ vpci_dev = kmalloc(sizeof(*vpci_dev), GFP_KERNEL);
++ if (!vpci_dev)
++ return -ENOMEM;
++
++ spin_lock_init(&vpci_dev->lock);
++
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ INIT_LIST_HEAD(&vpci_dev->dev_list[slot]);
++ }
++
++ pdev->pci_dev_data = vpci_dev;
++
++ return 0;
++}
++
++int pciback_publish_pci_roots(struct pciback_device *pdev,
++ publish_pci_root_cb publish_cb)
++{
++ /* The Virtual PCI bus has only one root */
++ return publish_cb(pdev, 0, 0);
++}
++
++void pciback_release_devices(struct pciback_device *pdev)
++{
++ int slot;
++ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
++
++ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
++ struct pci_dev_entry *e, *tmp;
++ list_for_each_entry_safe(e, tmp, &vpci_dev->dev_list[slot],
++ list) {
++ list_del(&e->list);
++ pcistub_put_pci_dev(e->dev);
++ kfree(e);
++ }
++ }
++
++ kfree(vpci_dev);
++ pdev->pci_dev_data = NULL;
++}
+--- linux-2.6.18.8/drivers/xen/pciback/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pciback/xenbus.c 2008-05-19 00:33:46.270813223 +0300
+@@ -0,0 +1,685 @@
++/*
++ * PCI Backend Xenbus Setup - handles setup with frontend and xend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <linux/init.h>
++#include <linux/list.h>
++#include <linux/vmalloc.h>
++#include <xen/xenbus.h>
++#include <xen/evtchn.h>
++#include "pciback.h"
++
++#define INVALID_EVTCHN_IRQ (-1)
++
++static struct pciback_device *alloc_pdev(struct xenbus_device *xdev)
++{
++ struct pciback_device *pdev;
++
++ pdev = kzalloc(sizeof(struct pciback_device), GFP_KERNEL);
++ if (pdev == NULL)
++ goto out;
++ dev_dbg(&xdev->dev, "allocated pdev @ 0x%p\n", pdev);
++
++ pdev->xdev = xdev;
++ xdev->dev.driver_data = pdev;
++
++ spin_lock_init(&pdev->dev_lock);
++
++ pdev->sh_area = NULL;
++ pdev->sh_info = NULL;
++ pdev->evtchn_irq = INVALID_EVTCHN_IRQ;
++ pdev->be_watching = 0;
++
++ INIT_WORK(&pdev->op_work, pciback_do_op, pdev);
++
++ if (pciback_init_devices(pdev)) {
++ kfree(pdev);
++ pdev = NULL;
++ }
++ out:
++ return pdev;
++}
++
++static void free_pdev(struct pciback_device *pdev)
++{
++ if (pdev->be_watching)
++ unregister_xenbus_watch(&pdev->be_watch);
++
++ /* Ensure the guest can't trigger our handler before removing devices */
++ if (pdev->evtchn_irq != INVALID_EVTCHN_IRQ)
++ unbind_from_irqhandler(pdev->evtchn_irq, pdev);
++
++ /* If the driver domain started an op, make sure we complete it or
++ * delete it before releasing the shared memory */
++ cancel_delayed_work(&pdev->op_work);
++ flush_scheduled_work();
++
++ if (pdev->sh_info)
++ xenbus_unmap_ring_vfree(pdev->xdev, pdev->sh_area);
++
++ pciback_release_devices(pdev);
++
++ pdev->xdev->dev.driver_data = NULL;
++ pdev->xdev = NULL;
++
++ kfree(pdev);
++}
++
++static int pciback_do_attach(struct pciback_device *pdev, int gnt_ref,
++ int remote_evtchn)
++{
++ int err = 0;
++ struct vm_struct *area;
++
++ dev_dbg(&pdev->xdev->dev,
++ "Attaching to frontend resources - gnt_ref=%d evtchn=%d\n",
++ gnt_ref, remote_evtchn);
++
++ area = xenbus_map_ring_valloc(pdev->xdev, gnt_ref);
++ if (IS_ERR(area)) {
++ err = PTR_ERR(area);
++ goto out;
++ }
++ pdev->sh_area = area;
++ pdev->sh_info = area->addr;
++
++ err = bind_interdomain_evtchn_to_irqhandler(
++ pdev->xdev->otherend_id, remote_evtchn, pciback_handle_event,
++ SA_SAMPLE_RANDOM, "pciback", pdev);
++ if (err < 0) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error binding event channel to IRQ");
++ goto out;
++ }
++ pdev->evtchn_irq = err;
++ err = 0;
++
++ dev_dbg(&pdev->xdev->dev, "Attached!\n");
++ out:
++ return err;
++}
++
++static int pciback_attach(struct pciback_device *pdev)
++{
++ int err = 0;
++ int gnt_ref, remote_evtchn;
++ char *magic = NULL;
++
++ spin_lock(&pdev->dev_lock);
++
++ /* Make sure we only do this setup once */
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateInitialised)
++ goto out;
++
++ /* Wait for frontend to state that it has published the configuration */
++ if (xenbus_read_driver_state(pdev->xdev->otherend) !=
++ XenbusStateInitialised)
++ goto out;
++
++ dev_dbg(&pdev->xdev->dev, "Reading frontend config\n");
++
++ err = xenbus_gather(XBT_NIL, pdev->xdev->otherend,
++ "pci-op-ref", "%u", &gnt_ref,
++ "event-channel", "%u", &remote_evtchn,
++ "magic", NULL, &magic, NULL);
++ if (err) {
++ /* If configuration didn't get read correctly, wait longer */
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading configuration from frontend");
++ goto out;
++ }
++
++ if (magic == NULL || strcmp(magic, XEN_PCI_MAGIC) != 0) {
++ xenbus_dev_fatal(pdev->xdev, -EFAULT,
++ "version mismatch (%s/%s) with pcifront - "
++ "halting pciback",
++ magic, XEN_PCI_MAGIC);
++ goto out;
++ }
++
++ err = pciback_do_attach(pdev, gnt_ref, remote_evtchn);
++ if (err)
++ goto out;
++
++ dev_dbg(&pdev->xdev->dev, "Connecting...\n");
++
++ err = xenbus_switch_state(pdev->xdev, XenbusStateConnected);
++ if (err)
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error switching to connected state!");
++
++ dev_dbg(&pdev->xdev->dev, "Connected? %d\n", err);
++ out:
++ spin_unlock(&pdev->dev_lock);
++
++ if (magic)
++ kfree(magic);
++
++ return err;
++}
++
++static int pciback_publish_pci_dev(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus,
++ unsigned int devfn, unsigned int devid)
++{
++ int err;
++ int len;
++ char str[64];
++
++ len = snprintf(str, sizeof(str), "vdev-%d", devid);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
++ "%04x:%02x:%02x.%02x", domain, bus,
++ PCI_SLOT(devfn), PCI_FUNC(devfn));
++
++ out:
++ return err;
++}
++
++static int pciback_export_device(struct pciback_device *pdev,
++ int domain, int bus, int slot, int func,
++ int devid)
++{
++ struct pci_dev *dev;
++ int err = 0;
++
++ dev_dbg(&pdev->xdev->dev, "exporting dom %x bus %x slot %x func %x\n",
++ domain, bus, slot, func);
++
++ dev = pcistub_get_pci_dev_by_slot(pdev, domain, bus, slot, func);
++ if (!dev) {
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Couldn't locate PCI device "
++ "(%04x:%02x:%02x.%01x)! "
++ "perhaps already in-use?",
++ domain, bus, slot, func);
++ goto out;
++ }
++
++ err = pciback_add_pci_dev(pdev, dev, devid, pciback_publish_pci_dev);
++ if (err)
++ goto out;
++
++ /* TODO: It'd be nice to export a bridge and have all of its children
++ * get exported with it. This may be best done in xend (which will
++ * have to calculate resource usage anyway) but we probably want to
++ * put something in here to ensure that if a bridge gets given to a
++ * driver domain, that all devices under that bridge are not given
++ * to other driver domains (as he who controls the bridge can disable
++ * it and stop the other devices from working).
++ */
++ out:
++ return err;
++}
++
++static int pciback_remove_device(struct pciback_device *pdev,
++ int domain, int bus, int slot, int func)
++{
++ int err = 0;
++ struct pci_dev *dev;
++
++ dev_dbg(&pdev->xdev->dev, "removing dom %x bus %x slot %x func %x\n",
++ domain, bus, slot, func);
++
++ dev = pciback_get_pci_dev(pdev, domain, bus, PCI_DEVFN(slot, func));
++ if (!dev) {
++ err = -EINVAL;
++ dev_dbg(&pdev->xdev->dev, "Couldn't locate PCI device "
++ "(%04x:%02x:%02x.%01x)! not owned by this domain\n",
++ domain, bus, slot, func);
++ goto out;
++ }
++
++ pciback_release_pci_dev(pdev, dev);
++
++ out:
++ return err;
++}
++
++static int pciback_publish_pci_root(struct pciback_device *pdev,
++ unsigned int domain, unsigned int bus)
++{
++ unsigned int d, b;
++ int i, root_num, len, err;
++ char str[64];
++
++ dev_dbg(&pdev->xdev->dev, "Publishing pci roots\n");
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ "root_num", "%d", &root_num);
++ if (err == 0 || err == -ENOENT)
++ root_num = 0;
++ else if (err < 0)
++ goto out;
++
++ /* Verify that we haven't already published this pci root */
++ for (i = 0; i < root_num; i++) {
++ len = snprintf(str, sizeof(str), "root-%d", i);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ str, "%x:%x", &d, &b);
++ if (err < 0)
++ goto out;
++ if (err != 2) {
++ err = -EINVAL;
++ goto out;
++ }
++
++ if (d == domain && b == bus) {
++ err = 0;
++ goto out;
++ }
++ }
++
++ len = snprintf(str, sizeof(str), "root-%d", root_num);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ dev_dbg(&pdev->xdev->dev, "writing root %d at %04x:%02x\n",
++ root_num, domain, bus);
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
++ "%04x:%02x", domain, bus);
++ if (err)
++ goto out;
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename,
++ "root_num", "%d", (root_num + 1));
++
++ out:
++ return err;
++}
++
++static int pciback_reconfigure(struct pciback_device *pdev)
++{
++ int err = 0;
++ int num_devs;
++ int domain, bus, slot, func;
++ int substate;
++ int i, len;
++ char state_str[64];
++ char dev_str[64];
++
++ spin_lock(&pdev->dev_lock);
++
++ dev_dbg(&pdev->xdev->dev, "Reconfiguring device ...\n");
++
++ /* Make sure we only reconfigure once */
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateReconfiguring)
++ goto out;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, "num_devs", "%d",
++ &num_devs);
++ if (err != 1) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading number of devices");
++ goto out;
++ }
++
++ for (i = 0; i < num_devs; i++) {
++ len = snprintf(state_str, sizeof(state_str), "state-%d", i);
++ if (unlikely(len >= (sizeof(state_str) - 1))) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "String overflow while reading "
++ "configuration");
++ goto out;
++ }
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, state_str,
++ "%d", &substate);
++ if (err != 1)
++ substate = XenbusStateUnknown;
++
++ switch (substate) {
++ case XenbusStateInitialising:
++ dev_dbg(&pdev->xdev->dev, "Attaching dev-%d ...\n", i);
++
++ len = snprintf(dev_str, sizeof(dev_str), "dev-%d", i);
++ if (unlikely(len >= (sizeof(dev_str) - 1))) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "String overflow while "
++ "reading configuration");
++ goto out;
++ }
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ dev_str, "%x:%x:%x.%x",
++ &domain, &bus, &slot, &func);
++ if (err < 0) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading device "
++ "configuration");
++ goto out;
++ }
++ if (err != 4) {
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error parsing pci device "
++ "configuration");
++ goto out;
++ }
++
++ err = pciback_export_device(pdev, domain, bus, slot,
++ func, i);
++ if (err)
++ goto out;
++
++ /* Publish pci roots. */
++ err = pciback_publish_pci_roots(pdev, pciback_publish_pci_root);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error while publish PCI root"
++ "buses for frontend");
++ goto out;
++ }
++
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename,
++ state_str, "%d",
++ XenbusStateInitialised);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error switching substate of "
++ "dev-%d\n", i);
++ goto out;
++ }
++ break;
++
++ case XenbusStateClosing:
++ dev_dbg(&pdev->xdev->dev, "Detaching dev-%d ...\n", i);
++
++ len = snprintf(dev_str, sizeof(dev_str), "vdev-%d", i);
++ if (unlikely(len >= (sizeof(dev_str) - 1))) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "String overflow while "
++ "reading configuration");
++ goto out;
++ }
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
++ dev_str, "%x:%x:%x.%x",
++ &domain, &bus, &slot, &func);
++ if (err < 0) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading device "
++ "configuration");
++ goto out;
++ }
++ if (err != 4) {
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error parsing pci device "
++ "configuration");
++ goto out;
++ }
++
++ err = pciback_remove_device(pdev, domain, bus, slot,
++ func);
++ if(err)
++ goto out;
++
++ /* TODO: If at some point we implement support for pci
++ * root hot-remove on pcifront side, we'll need to
++ * remove unnecessary xenstore nodes of pci roots here.
++ */
++
++ break;
++
++ default:
++ break;
++ }
++ }
++
++ err = xenbus_switch_state(pdev->xdev, XenbusStateReconfigured);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error switching to reconfigured state!");
++ goto out;
++ }
++
++ out:
++ spin_unlock(&pdev->dev_lock);
++
++ return 0;
++}
++
++static void pciback_frontend_changed(struct xenbus_device *xdev,
++ enum xenbus_state fe_state)
++{
++ struct pciback_device *pdev = xdev->dev.driver_data;
++
++ dev_dbg(&xdev->dev, "fe state changed %d\n", fe_state);
++
++ switch (fe_state) {
++ case XenbusStateInitialised:
++ pciback_attach(pdev);
++ break;
++
++ case XenbusStateClosing:
++ xenbus_switch_state(xdev, XenbusStateClosed);
++ break;
++
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ dev_dbg(&xdev->dev, "frontend is gone! unregister device\n");
++ device_unregister(&xdev->dev);
++ break;
++
++ case XenbusStateReconfiguring:
++ pciback_reconfigure(pdev);
++ break;
++
++ case XenbusStateConnected:
++ /* pcifront switched its state from reconfiguring to connected.
++ * Then switch to connected state.
++ */
++ xenbus_switch_state(xdev, XenbusStateConnected);
++ break;
++
++ default:
++ break;
++ }
++}
++
++static int pciback_setup_backend(struct pciback_device *pdev)
++{
++ /* Get configuration from xend (if available now) */
++ int domain, bus, slot, func;
++ int err = 0;
++ int i, num_devs;
++ char dev_str[64];
++ char state_str[64];
++
++ spin_lock(&pdev->dev_lock);
++
++ /* It's possible we could get the call to setup twice, so make sure
++ * we're not already connected.
++ */
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateInitWait)
++ goto out;
++
++ dev_dbg(&pdev->xdev->dev, "getting be setup\n");
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, "num_devs", "%d",
++ &num_devs);
++ if (err != 1) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading number of devices");
++ goto out;
++ }
++
++ for (i = 0; i < num_devs; i++) {
++ int l = snprintf(dev_str, sizeof(dev_str), "dev-%d", i);
++ if (unlikely(l >= (sizeof(dev_str) - 1))) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "String overflow while reading "
++ "configuration");
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, dev_str,
++ "%x:%x:%x.%x", &domain, &bus, &slot, &func);
++ if (err < 0) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading device configuration");
++ goto out;
++ }
++ if (err != 4) {
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error parsing pci device "
++ "configuration");
++ goto out;
++ }
++
++ err = pciback_export_device(pdev, domain, bus, slot, func, i);
++ if (err)
++ goto out;
++
++ /* Switch substate of this device. */
++ l = snprintf(state_str, sizeof(state_str), "state-%d", i);
++ if (unlikely(l >= (sizeof(state_str) - 1))) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "String overflow while reading "
++ "configuration");
++ goto out;
++ }
++ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, state_str,
++ "%d", XenbusStateInitialised);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err, "Error switching "
++ "substate of dev-%d\n", i);
++ goto out;
++ }
++ }
++
++ err = pciback_publish_pci_roots(pdev, pciback_publish_pci_root);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error while publish PCI root buses "
++ "for frontend");
++ goto out;
++ }
++
++ err = xenbus_switch_state(pdev->xdev, XenbusStateInitialised);
++ if (err)
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error switching to initialised state!");
++
++ out:
++ spin_unlock(&pdev->dev_lock);
++
++ if (!err)
++ /* see if pcifront is already configured (if not, we'll wait) */
++ pciback_attach(pdev);
++
++ return err;
++}
++
++static void pciback_be_watch(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ struct pciback_device *pdev =
++ container_of(watch, struct pciback_device, be_watch);
++
++ switch (xenbus_read_driver_state(pdev->xdev->nodename)) {
++ case XenbusStateInitWait:
++ pciback_setup_backend(pdev);
++ break;
++
++ default:
++ break;
++ }
++}
++
++static int pciback_xenbus_probe(struct xenbus_device *dev,
++ const struct xenbus_device_id *id)
++{
++ int err = 0;
++ struct pciback_device *pdev = alloc_pdev(dev);
++
++ if (pdev == NULL) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(dev, err,
++ "Error allocating pciback_device struct");
++ goto out;
++ }
++
++ /* wait for xend to configure us */
++ err = xenbus_switch_state(dev, XenbusStateInitWait);
++ if (err)
++ goto out;
++
++ /* watch the backend node for backend configuration information */
++ err = xenbus_watch_path(dev, dev->nodename, &pdev->be_watch,
++ pciback_be_watch);
++ if (err)
++ goto out;
++ pdev->be_watching = 1;
++
++ /* We need to force a call to our callback here in case
++ * xend already configured us!
++ */
++ pciback_be_watch(&pdev->be_watch, NULL, 0);
++
++ out:
++ return err;
++}
++
++static int pciback_xenbus_remove(struct xenbus_device *dev)
++{
++ struct pciback_device *pdev = dev->dev.driver_data;
++
++ if (pdev != NULL)
++ free_pdev(pdev);
++
++ return 0;
++}
++
++static const struct xenbus_device_id xenpci_ids[] = {
++ {"pci"},
++ {{0}},
++};
++
++static struct xenbus_driver xenbus_pciback_driver = {
++ .name = "pciback",
++ .owner = THIS_MODULE,
++ .ids = xenpci_ids,
++ .probe = pciback_xenbus_probe,
++ .remove = pciback_xenbus_remove,
++ .otherend_changed = pciback_frontend_changed,
++};
++
++int __init pciback_xenbus_register(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ return xenbus_register_backend(&xenbus_pciback_driver);
++}
++
++void __exit pciback_xenbus_unregister(void)
++{
++ xenbus_unregister_driver(&xenbus_pciback_driver);
++}
+--- linux-2.6.18.8/drivers/xen/pcifront/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pcifront/Makefile 2008-05-19 00:33:46.286814145 +0300
+@@ -0,0 +1,7 @@
++obj-y += pcifront.o
++
++pcifront-y := pci_op.o xenbus.o pci.o
++
++ifeq ($(CONFIG_XEN_PCIDEV_FE_DEBUG),y)
++EXTRA_CFLAGS += -DDEBUG
++endif
+--- linux-2.6.18.8/drivers/xen/pcifront/pci.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pci.c 2008-05-19 00:33:46.290814376 +0300
+@@ -0,0 +1,46 @@
++/*
++ * PCI Frontend Operations - ensure only one PCI frontend runs at a time
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <linux/init.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include "pcifront.h"
++
++DEFINE_SPINLOCK(pcifront_dev_lock);
++static struct pcifront_device *pcifront_dev = NULL;
++
++int pcifront_connect(struct pcifront_device *pdev)
++{
++ int err = 0;
++
++ spin_lock(&pcifront_dev_lock);
++
++ if (!pcifront_dev) {
++ dev_info(&pdev->xdev->dev, "Installing PCI frontend\n");
++ pcifront_dev = pdev;
++ }
++ else {
++ dev_err(&pdev->xdev->dev, "PCI frontend already installed!\n");
++ err = -EEXIST;
++ }
++
++ spin_unlock(&pcifront_dev_lock);
++
++ return err;
++}
++
++void pcifront_disconnect(struct pcifront_device *pdev)
++{
++ spin_lock(&pcifront_dev_lock);
++
++ if (pdev == pcifront_dev) {
++ dev_info(&pdev->xdev->dev,
++ "Disconnecting PCI Frontend Buses\n");
++ pcifront_dev = NULL;
++ }
++
++ spin_unlock(&pcifront_dev_lock);
++}
+--- linux-2.6.18.8/drivers/xen/pcifront/pci_op.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pci_op.c 2008-05-19 00:33:46.290814376 +0300
+@@ -0,0 +1,551 @@
++/*
++ * PCI Frontend Operations - Communicates with frontend
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <linux/version.h>
++#include <linux/init.h>
++#include <linux/pci.h>
++#include <linux/spinlock.h>
++#include <linux/time.h>
++#include <xen/evtchn.h>
++#include "pcifront.h"
++
++static int verbose_request = 0;
++module_param(verbose_request, int, 0644);
++
++#ifdef __ia64__
++static void pcifront_init_sd(struct pcifront_sd *sd,
++ unsigned int domain, unsigned int bus,
++ struct pcifront_device *pdev)
++{
++ int err, i, j, k, len, root_num, res_count;
++ struct acpi_resource res;
++ unsigned int d, b, byte;
++ unsigned long magic;
++ char str[64], tmp[3];
++ unsigned char *buf, *bufp;
++ u8 *ptr;
++
++ memset(sd, 0, sizeof(*sd));
++
++ sd->segment = domain;
++ sd->node = -1; /* Revisit for NUMA */
++ sd->platform_data = pdev;
++
++ /* Look for resources for this controller in xenbus. */
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, "root_num",
++ "%d", &root_num);
++ if (err != 1)
++ return;
++
++ for (i = 0; i < root_num; i++) {
++ len = snprintf(str, sizeof(str), "root-%d", i);
++ if (unlikely(len >= (sizeof(str) - 1)))
++ return;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
++ str, "%x:%x", &d, &b);
++ if (err != 2)
++ return;
++
++ if (d == domain && b == bus)
++ break;
++ }
++
++ if (i == root_num)
++ return;
++
++ len = snprintf(str, sizeof(str), "root-resource-magic");
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
++ str, "%lx", &magic);
++
++ if (err != 1)
++ return; /* No resources, nothing to do */
++
++ if (magic != (sizeof(res) * 2) + 1) {
++ printk(KERN_WARNING "pcifront: resource magic mismatch\n");
++ return;
++ }
++
++ len = snprintf(str, sizeof(str), "root-%d-resources", i);
++ if (unlikely(len >= (sizeof(str) - 1)))
++ return;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
++ str, "%d", &res_count);
++
++ if (err != 1)
++ return; /* No resources, nothing to do */
++
++ sd->window = kzalloc(sizeof(*sd->window) * res_count, GFP_KERNEL);
++ if (!sd->window)
++ return;
++
++ /* magic is also the size of the byte stream in xenbus */
++ buf = kmalloc(magic, GFP_KERNEL);
++ if (!buf) {
++ kfree(sd->window);
++ sd->window = NULL;
++ return;
++ }
++
++ /* Read the resources out of xenbus */
++ for (j = 0; j < res_count; j++) {
++ memset(&res, 0, sizeof(res));
++ memset(buf, 0, magic);
++
++ len = snprintf(str, sizeof(str), "root-%d-resource-%d", i, j);
++ if (unlikely(len >= (sizeof(str) - 1)))
++ return;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
++ "%s", buf);
++ if (err != 1) {
++ printk(KERN_WARNING "pcifront: error reading "
++ "resource %d on bus %04x:%02x\n",
++ j, domain, bus);
++ continue;
++ }
++
++ bufp = buf;
++ ptr = (u8 *)&res;
++ memset(tmp, 0, sizeof(tmp));
++
++ /* Copy ASCII byte stream into structure */
++ for (k = 0; k < magic - 1; k += 2) {
++ memcpy(tmp, bufp, 2);
++ bufp += 2;
++
++ sscanf(tmp, "%02x", &byte);
++ *ptr = byte;
++ ptr++;
++ }
++
++ xen_add_resource(sd, domain, bus, &res);
++ sd->windows++;
++ }
++ kfree(buf);
++}
++#endif
++
++static int errno_to_pcibios_err(int errno)
++{
++ switch (errno) {
++ case XEN_PCI_ERR_success:
++ return PCIBIOS_SUCCESSFUL;
++
++ case XEN_PCI_ERR_dev_not_found:
++ return PCIBIOS_DEVICE_NOT_FOUND;
++
++ case XEN_PCI_ERR_invalid_offset:
++ case XEN_PCI_ERR_op_failed:
++ return PCIBIOS_BAD_REGISTER_NUMBER;
++
++ case XEN_PCI_ERR_not_implemented:
++ return PCIBIOS_FUNC_NOT_SUPPORTED;
++
++ case XEN_PCI_ERR_access_denied:
++ return PCIBIOS_SET_FAILED;
++ }
++ return errno;
++}
++
++static int do_pci_op(struct pcifront_device *pdev, struct xen_pci_op *op)
++{
++ int err = 0;
++ struct xen_pci_op *active_op = &pdev->sh_info->op;
++ unsigned long irq_flags;
++ evtchn_port_t port = pdev->evtchn;
++ s64 ns, ns_timeout;
++ struct timeval tv;
++
++ spin_lock_irqsave(&pdev->sh_info_lock, irq_flags);
++
++ memcpy(active_op, op, sizeof(struct xen_pci_op));
++
++ /* Go */
++ wmb();
++ set_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags);
++ notify_remote_via_evtchn(port);
++
++ /*
++ * We set a poll timeout of 3 seconds but give up on return after
++ * 2 seconds. It is better to time out too late rather than too early
++ * (in the latter case we end up continually re-executing poll() with a
++ * timeout in the past). 1s difference gives plenty of slack for error.
++ */
++ do_gettimeofday(&tv);
++ ns_timeout = timeval_to_ns(&tv) + 2 * (s64)NSEC_PER_SEC;
++
++ clear_evtchn(port);
++
++ while (test_bit(_XEN_PCIF_active,
++ (unsigned long *)&pdev->sh_info->flags)) {
++ if (HYPERVISOR_poll(&port, 1, jiffies + 3*HZ))
++ BUG();
++ clear_evtchn(port);
++ do_gettimeofday(&tv);
++ ns = timeval_to_ns(&tv);
++ if (ns > ns_timeout) {
++ dev_err(&pdev->xdev->dev,
++ "pciback not responding!!!\n");
++ clear_bit(_XEN_PCIF_active,
++ (unsigned long *)&pdev->sh_info->flags);
++ err = XEN_PCI_ERR_dev_not_found;
++ goto out;
++ }
++ }
++
++ memcpy(op, active_op, sizeof(struct xen_pci_op));
++
++ err = op->err;
++ out:
++ spin_unlock_irqrestore(&pdev->sh_info_lock, irq_flags);
++ return err;
++}
++
++/* Access to this function is spinlocked in drivers/pci/access.c */
++static int pcifront_bus_read(struct pci_bus *bus, unsigned int devfn,
++ int where, int size, u32 * val)
++{
++ int err = 0;
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_conf_read,
++ .domain = pci_domain_nr(bus),
++ .bus = bus->number,
++ .devfn = devfn,
++ .offset = where,
++ .size = size,
++ };
++ struct pcifront_sd *sd = bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ if (verbose_request)
++ dev_info(&pdev->xdev->dev,
++ "read dev=%04x:%02x:%02x.%01x - offset %x size %d\n",
++ pci_domain_nr(bus), bus->number, PCI_SLOT(devfn),
++ PCI_FUNC(devfn), where, size);
++
++ err = do_pci_op(pdev, &op);
++
++ if (likely(!err)) {
++ if (verbose_request)
++ dev_info(&pdev->xdev->dev, "read got back value %x\n",
++ op.value);
++
++ *val = op.value;
++ } else if (err == -ENODEV) {
++ /* No device here, pretend that it just returned 0 */
++ err = 0;
++ *val = 0;
++ }
++
++ return errno_to_pcibios_err(err);
++}
++
++/* Access to this function is spinlocked in drivers/pci/access.c */
++static int pcifront_bus_write(struct pci_bus *bus, unsigned int devfn,
++ int where, int size, u32 val)
++{
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_conf_write,
++ .domain = pci_domain_nr(bus),
++ .bus = bus->number,
++ .devfn = devfn,
++ .offset = where,
++ .size = size,
++ .value = val,
++ };
++ struct pcifront_sd *sd = bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ if (verbose_request)
++ dev_info(&pdev->xdev->dev,
++ "write dev=%04x:%02x:%02x.%01x - "
++ "offset %x size %d val %x\n",
++ pci_domain_nr(bus), bus->number,
++ PCI_SLOT(devfn), PCI_FUNC(devfn), where, size, val);
++
++ return errno_to_pcibios_err(do_pci_op(pdev, &op));
++}
++
++struct pci_ops pcifront_bus_ops = {
++ .read = pcifront_bus_read,
++ .write = pcifront_bus_write,
++};
++
++#ifdef CONFIG_PCI_MSI
++int pci_frontend_enable_msix(struct pci_dev *dev,
++ struct msix_entry *entries,
++ int nvec)
++{
++ int err;
++ int i;
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_enable_msix,
++ .domain = pci_domain_nr(dev->bus),
++ .bus = dev->bus->number,
++ .devfn = dev->devfn,
++ .value = nvec,
++ };
++ struct pcifront_sd *sd = dev->bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ if (nvec > SH_INFO_MAX_VEC) {
++ printk("too much vector for pci frontend%x\n", nvec);
++ return -EINVAL;
++ }
++
++ for (i = 0; i < nvec; i++) {
++ op.msix_entries[i].entry = entries[i].entry;
++ op.msix_entries[i].vector = entries[i].vector;
++ }
++
++ err = do_pci_op(pdev, &op);
++
++ if (!err) {
++ if (!op.value) {
++ /* we get the result */
++ for ( i = 0; i < nvec; i++)
++ entries[i].vector = op.msix_entries[i].vector;
++ return 0;
++ }
++ else {
++ printk("enable msix get value %x\n", op.value);
++ return op.value;
++ }
++ }
++ else {
++ printk("enable msix get err %x\n", err);
++ return err;
++ }
++}
++
++void pci_frontend_disable_msix(struct pci_dev* dev)
++{
++ int err;
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_disable_msix,
++ .domain = pci_domain_nr(dev->bus),
++ .bus = dev->bus->number,
++ .devfn = dev->devfn,
++ };
++ struct pcifront_sd *sd = dev->bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ err = do_pci_op(pdev, &op);
++
++ /* What should do for error ? */
++ if (err)
++ printk("pci_disable_msix get err %x\n", err);
++}
++
++int pci_frontend_enable_msi(struct pci_dev *dev)
++{
++ int err;
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_enable_msi,
++ .domain = pci_domain_nr(dev->bus),
++ .bus = dev->bus->number,
++ .devfn = dev->devfn,
++ };
++ struct pcifront_sd *sd = dev->bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ err = do_pci_op(pdev, &op);
++ if (likely(!err)) {
++ dev->irq = op.value;
++ }
++ else {
++ printk("pci frontend enable msi failed for dev %x:%x \n",
++ op.bus, op.devfn);
++ err = -EINVAL;
++ }
++ return err;
++}
++
++void pci_frontend_disable_msi(struct pci_dev* dev)
++{
++ int err;
++ struct xen_pci_op op = {
++ .cmd = XEN_PCI_OP_disable_msi,
++ .domain = pci_domain_nr(dev->bus),
++ .bus = dev->bus->number,
++ .devfn = dev->devfn,
++ };
++ struct pcifront_sd *sd = dev->bus->sysdata;
++ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++
++ err = do_pci_op(pdev, &op);
++ if (err == XEN_PCI_ERR_dev_not_found) {
++ /* XXX No response from backend, what shall we do? */
++ printk("get no response from backend for disable MSI\n");
++ return;
++ }
++ if (likely(!err))
++ dev->irq = op.value;
++ else
++ /* how can pciback notify us fail? */
++ printk("get fake response frombackend \n");
++}
++#endif /* CONFIG_PCI_MSI */
++
++/* Claim resources for the PCI frontend as-is, backend won't allow changes */
++static void pcifront_claim_resource(struct pci_dev *dev, void *data)
++{
++ struct pcifront_device *pdev = data;
++ int i;
++ struct resource *r;
++
++ for (i = 0; i < PCI_NUM_RESOURCES; i++) {
++ r = &dev->resource[i];
++
++ if (!r->parent && r->start && r->flags) {
++ dev_dbg(&pdev->xdev->dev, "claiming resource %s/%d\n",
++ pci_name(dev), i);
++ pci_claim_resource(dev, i);
++ }
++ }
++}
++
++int __devinit pcifront_scan_root(struct pcifront_device *pdev,
++ unsigned int domain, unsigned int bus)
++{
++ struct pci_bus *b;
++ struct pcifront_sd *sd = NULL;
++ struct pci_bus_entry *bus_entry = NULL;
++ int err = 0;
++
++#ifndef CONFIG_PCI_DOMAINS
++ if (domain != 0) {
++ dev_err(&pdev->xdev->dev,
++ "PCI Root in non-zero PCI Domain! domain=%d\n", domain);
++ dev_err(&pdev->xdev->dev,
++ "Please compile with CONFIG_PCI_DOMAINS\n");
++ err = -EINVAL;
++ goto err_out;
++ }
++#endif
++
++ dev_info(&pdev->xdev->dev, "Creating PCI Frontend Bus %04x:%02x\n",
++ domain, bus);
++
++ bus_entry = kmalloc(sizeof(*bus_entry), GFP_KERNEL);
++ sd = kmalloc(sizeof(*sd), GFP_KERNEL);
++ if (!bus_entry || !sd) {
++ err = -ENOMEM;
++ goto err_out;
++ }
++ pcifront_init_sd(sd, domain, bus, pdev);
++
++ b = pci_scan_bus_parented(&pdev->xdev->dev, bus,
++ &pcifront_bus_ops, sd);
++ if (!b) {
++ dev_err(&pdev->xdev->dev,
++ "Error creating PCI Frontend Bus!\n");
++ err = -ENOMEM;
++ goto err_out;
++ }
++
++ pcifront_setup_root_resources(b, sd);
++ bus_entry->bus = b;
++
++ list_add(&bus_entry->list, &pdev->root_buses);
++
++ /* Claim resources before going "live" with our devices */
++ pci_walk_bus(b, pcifront_claim_resource, pdev);
++
++ pci_bus_add_devices(b);
++
++ return 0;
++
++ err_out:
++ kfree(bus_entry);
++ kfree(sd);
++
++ return err;
++}
++
++int __devinit pcifront_rescan_root(struct pcifront_device *pdev,
++ unsigned int domain, unsigned int bus)
++{
++ struct pci_bus *b;
++ struct pci_dev *d;
++ unsigned int devfn;
++
++#ifndef CONFIG_PCI_DOMAINS
++ if (domain != 0) {
++ dev_err(&pdev->xdev->dev,
++ "PCI Root in non-zero PCI Domain! domain=%d\n", domain);
++ dev_err(&pdev->xdev->dev,
++ "Please compile with CONFIG_PCI_DOMAINS\n");
++ return -EINVAL;
++ }
++#endif
++
++ dev_info(&pdev->xdev->dev, "Rescanning PCI Frontend Bus %04x:%02x\n",
++ domain, bus);
++
++ b = pci_find_bus(domain, bus);
++ if(!b)
++ /* If the bus is unknown, create it. */
++ return pcifront_scan_root(pdev, domain, bus);
++
++ /* Rescan the bus for newly attached functions and add.
++ * We omit handling of PCI bridge attachment because pciback prevents
++ * bridges from being exported.
++ */
++ for (devfn = 0; devfn < 0x100; devfn++) {
++ d = pci_get_slot(b, devfn);
++ if(d) {
++ /* Device is already known. */
++ pci_dev_put(d);
++ continue;
++ }
++
++ d = pci_scan_single_device(b, devfn);
++ if (d) {
++ dev_info(&pdev->xdev->dev, "New device on "
++ "%04x:%02x:%02x.%02x found.\n", domain, bus,
++ PCI_SLOT(devfn), PCI_FUNC(devfn));
++ pci_bus_add_device(d);
++ }
++ }
++
++ return 0;
++}
++
++static void free_root_bus_devs(struct pci_bus *bus)
++{
++ struct pci_dev *dev;
++
++ while (!list_empty(&bus->devices)) {
++ dev = container_of(bus->devices.next, struct pci_dev,
++ bus_list);
++ dev_dbg(&dev->dev, "removing device\n");
++ pci_remove_bus_device(dev);
++ }
++}
++
++void pcifront_free_roots(struct pcifront_device *pdev)
++{
++ struct pci_bus_entry *bus_entry, *t;
++
++ dev_dbg(&pdev->xdev->dev, "cleaning up root buses\n");
++
++ list_for_each_entry_safe(bus_entry, t, &pdev->root_buses, list) {
++ list_del(&bus_entry->list);
++
++ free_root_bus_devs(bus_entry->bus);
++
++ kfree(bus_entry->bus->sysdata);
++
++ device_unregister(bus_entry->bus->bridge);
++ pci_remove_bus(bus_entry->bus);
++
++ kfree(bus_entry);
++ }
++}
+--- linux-2.6.18.8/drivers/xen/pcifront/pcifront.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pcifront.h 2008-05-19 00:33:46.290814376 +0300
+@@ -0,0 +1,42 @@
++/*
++ * PCI Frontend - Common data structures & function declarations
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#ifndef __XEN_PCIFRONT_H__
++#define __XEN_PCIFRONT_H__
++
++#include <linux/spinlock.h>
++#include <linux/pci.h>
++#include <xen/xenbus.h>
++#include <xen/interface/io/pciif.h>
++#include <xen/pcifront.h>
++
++struct pci_bus_entry {
++ struct list_head list;
++ struct pci_bus *bus;
++};
++
++struct pcifront_device {
++ struct xenbus_device *xdev;
++ struct list_head root_buses;
++ spinlock_t dev_lock;
++
++ int evtchn;
++ int gnt_ref;
++
++ /* Lock this when doing any operations in sh_info */
++ spinlock_t sh_info_lock;
++ struct xen_pci_sharedinfo *sh_info;
++};
++
++int pcifront_connect(struct pcifront_device *pdev);
++void pcifront_disconnect(struct pcifront_device *pdev);
++
++int pcifront_scan_root(struct pcifront_device *pdev,
++ unsigned int domain, unsigned int bus);
++int pcifront_rescan_root(struct pcifront_device *pdev,
++ unsigned int domain, unsigned int bus);
++void pcifront_free_roots(struct pcifront_device *pdev);
++
++#endif /* __XEN_PCIFRONT_H__ */
+--- linux-2.6.18.8/drivers/xen/pcifront/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/pcifront/xenbus.c 2008-05-19 00:33:46.318815990 +0300
+@@ -0,0 +1,454 @@
++/*
++ * PCI Frontend Xenbus Setup - handles setup with backend (imports page/evtchn)
++ *
++ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ */
++#include <linux/module.h>
++#include <linux/init.h>
++#include <linux/mm.h>
++#include <xen/xenbus.h>
++#include <xen/gnttab.h>
++#include "pcifront.h"
++
++#ifndef __init_refok
++#define __init_refok
++#endif
++
++#define INVALID_GRANT_REF (0)
++#define INVALID_EVTCHN (-1)
++
++static struct pcifront_device *alloc_pdev(struct xenbus_device *xdev)
++{
++ struct pcifront_device *pdev;
++
++ pdev = kzalloc(sizeof(struct pcifront_device), GFP_KERNEL);
++ if (pdev == NULL)
++ goto out;
++
++ pdev->sh_info =
++ (struct xen_pci_sharedinfo *)__get_free_page(GFP_KERNEL);
++ if (pdev->sh_info == NULL) {
++ kfree(pdev);
++ pdev = NULL;
++ goto out;
++ }
++ pdev->sh_info->flags = 0;
++
++ xdev->dev.driver_data = pdev;
++ pdev->xdev = xdev;
++
++ INIT_LIST_HEAD(&pdev->root_buses);
++
++ spin_lock_init(&pdev->dev_lock);
++ spin_lock_init(&pdev->sh_info_lock);
++
++ pdev->evtchn = INVALID_EVTCHN;
++ pdev->gnt_ref = INVALID_GRANT_REF;
++
++ dev_dbg(&xdev->dev, "Allocated pdev @ 0x%p pdev->sh_info @ 0x%p\n",
++ pdev, pdev->sh_info);
++ out:
++ return pdev;
++}
++
++static void free_pdev(struct pcifront_device *pdev)
++{
++ dev_dbg(&pdev->xdev->dev, "freeing pdev @ 0x%p\n", pdev);
++
++ pcifront_free_roots(pdev);
++
++ if (pdev->evtchn != INVALID_EVTCHN)
++ xenbus_free_evtchn(pdev->xdev, pdev->evtchn);
++
++ if (pdev->gnt_ref != INVALID_GRANT_REF)
++ gnttab_end_foreign_access(pdev->gnt_ref,
++ (unsigned long)pdev->sh_info);
++
++ pdev->xdev->dev.driver_data = NULL;
++
++ kfree(pdev);
++}
++
++static int pcifront_publish_info(struct pcifront_device *pdev)
++{
++ int err = 0;
++ struct xenbus_transaction trans;
++
++ err = xenbus_grant_ring(pdev->xdev, virt_to_mfn(pdev->sh_info));
++ if (err < 0)
++ goto out;
++
++ pdev->gnt_ref = err;
++
++ err = xenbus_alloc_evtchn(pdev->xdev, &pdev->evtchn);
++ if (err)
++ goto out;
++
++ do_publish:
++ err = xenbus_transaction_start(&trans);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error writing configuration for backend "
++ "(start transaction)");
++ goto out;
++ }
++
++ err = xenbus_printf(trans, pdev->xdev->nodename,
++ "pci-op-ref", "%u", pdev->gnt_ref);
++ if (!err)
++ err = xenbus_printf(trans, pdev->xdev->nodename,
++ "event-channel", "%u", pdev->evtchn);
++ if (!err)
++ err = xenbus_printf(trans, pdev->xdev->nodename,
++ "magic", XEN_PCI_MAGIC);
++
++ if (err) {
++ xenbus_transaction_end(trans, 1);
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error writing configuration for backend");
++ goto out;
++ } else {
++ err = xenbus_transaction_end(trans, 0);
++ if (err == -EAGAIN)
++ goto do_publish;
++ else if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error completing transaction "
++ "for backend");
++ goto out;
++ }
++ }
++
++ xenbus_switch_state(pdev->xdev, XenbusStateInitialised);
++
++ dev_dbg(&pdev->xdev->dev, "publishing successful!\n");
++
++ out:
++ return err;
++}
++
++static int __devinit pcifront_try_connect(struct pcifront_device *pdev)
++{
++ int err = -EFAULT;
++ int i, num_roots, len;
++ char str[64];
++ unsigned int domain, bus;
++
++ spin_lock(&pdev->dev_lock);
++
++ /* Only connect once */
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateInitialised)
++ goto out;
++
++ err = pcifront_connect(pdev);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error connecting PCI Frontend");
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
++ "root_num", "%d", &num_roots);
++ if (err == -ENOENT) {
++ xenbus_dev_error(pdev->xdev, err,
++ "No PCI Roots found, trying 0000:00");
++ err = pcifront_scan_root(pdev, 0, 0);
++ num_roots = 0;
++ } else if (err != 1) {
++ if (err == 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading number of PCI roots");
++ goto out;
++ }
++
++ for (i = 0; i < num_roots; i++) {
++ len = snprintf(str, sizeof(str), "root-%d", i);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
++ "%x:%x", &domain, &bus);
++ if (err != 2) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading PCI root %d", i);
++ goto out;
++ }
++
++ err = pcifront_scan_root(pdev, domain, bus);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error scanning PCI root %04x:%02x",
++ domain, bus);
++ goto out;
++ }
++ }
++
++ err = xenbus_switch_state(pdev->xdev, XenbusStateConnected);
++ if (err)
++ goto out;
++
++ out:
++ spin_unlock(&pdev->dev_lock);
++ return err;
++}
++
++static int pcifront_try_disconnect(struct pcifront_device *pdev)
++{
++ int err = 0;
++ enum xenbus_state prev_state;
++
++ spin_lock(&pdev->dev_lock);
++
++ prev_state = xenbus_read_driver_state(pdev->xdev->nodename);
++
++ if (prev_state < XenbusStateClosing)
++ err = xenbus_switch_state(pdev->xdev, XenbusStateClosing);
++
++ if (!err && prev_state == XenbusStateConnected)
++ pcifront_disconnect(pdev);
++
++ spin_unlock(&pdev->dev_lock);
++
++ return err;
++}
++
++static int __devinit pcifront_attach_devices(struct pcifront_device *pdev)
++{
++ int err = -EFAULT;
++ int i, num_roots, len;
++ unsigned int domain, bus;
++ char str[64];
++
++ spin_lock(&pdev->dev_lock);
++
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateReconfiguring)
++ goto out;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
++ "root_num", "%d", &num_roots);
++ if (err == -ENOENT) {
++ xenbus_dev_error(pdev->xdev, err,
++ "No PCI Roots found, trying 0000:00");
++ err = pcifront_rescan_root(pdev, 0, 0);
++ num_roots = 0;
++ } else if (err != 1) {
++ if (err == 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading number of PCI roots");
++ goto out;
++ }
++
++ for (i = 0; i < num_roots; i++) {
++ len = snprintf(str, sizeof(str), "root-%d", i);
++ if (unlikely(len >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
++ "%x:%x", &domain, &bus);
++ if (err != 2) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading PCI root %d", i);
++ goto out;
++ }
++
++ err = pcifront_rescan_root(pdev, domain, bus);
++ if (err) {
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error scanning PCI root %04x:%02x",
++ domain, bus);
++ goto out;
++ }
++ }
++
++ xenbus_switch_state(pdev->xdev, XenbusStateConnected);
++
++ out:
++ spin_unlock(&pdev->dev_lock);
++ return err;
++}
++
++static int pcifront_detach_devices(struct pcifront_device *pdev)
++{
++ int err = 0;
++ int i, num_devs;
++ unsigned int domain, bus, slot, func;
++ struct pci_bus *pci_bus;
++ struct pci_dev *pci_dev;
++ char str[64];
++
++ spin_lock(&pdev->dev_lock);
++
++ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
++ XenbusStateConnected)
++ goto out;
++
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, "num_devs", "%d",
++ &num_devs);
++ if (err != 1) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading number of PCI devices");
++ goto out;
++ }
++
++ /* Find devices being detached and remove them. */
++ for (i = 0; i < num_devs; i++) {
++ int l, state;
++ l = snprintf(str, sizeof(str), "state-%d", i);
++ if (unlikely(l >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str, "%d",
++ &state);
++ if (err != 1)
++ state = XenbusStateUnknown;
++
++ if (state != XenbusStateClosing)
++ continue;
++
++ /* Remove device. */
++ l = snprintf(str, sizeof(str), "vdev-%d", i);
++ if (unlikely(l >= (sizeof(str) - 1))) {
++ err = -ENOMEM;
++ goto out;
++ }
++ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
++ "%x:%x:%x.%x", &domain, &bus, &slot, &func);
++ if (err != 4) {
++ if (err >= 0)
++ err = -EINVAL;
++ xenbus_dev_fatal(pdev->xdev, err,
++ "Error reading PCI device %d", i);
++ goto out;
++ }
++
++ pci_bus = pci_find_bus(domain, bus);
++ if(!pci_bus) {
++ dev_dbg(&pdev->xdev->dev, "Cannot get bus %04x:%02x\n",
++ domain, bus);
++ continue;
++ }
++ pci_dev = pci_get_slot(pci_bus, PCI_DEVFN(slot, func));
++ if(!pci_dev) {
++ dev_dbg(&pdev->xdev->dev,
++ "Cannot get PCI device %04x:%02x:%02x.%02x\n",
++ domain, bus, slot, func);
++ continue;
++ }
++ pci_remove_bus_device(pci_dev);
++ pci_dev_put(pci_dev);
++
++ dev_dbg(&pdev->xdev->dev,
++ "PCI device %04x:%02x:%02x.%02x removed.\n",
++ domain, bus, slot, func);
++ }
++
++ err = xenbus_switch_state(pdev->xdev, XenbusStateReconfiguring);
++
++ out:
++ spin_unlock(&pdev->dev_lock);
++ return err;
++}
++
++static void __init_refok pcifront_backend_changed(struct xenbus_device *xdev,
++ enum xenbus_state be_state)
++{
++ struct pcifront_device *pdev = xdev->dev.driver_data;
++
++ switch (be_state) {
++ case XenbusStateClosing:
++ dev_warn(&xdev->dev, "backend going away!\n");
++ pcifront_try_disconnect(pdev);
++ break;
++
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ dev_warn(&xdev->dev, "backend went away!\n");
++ pcifront_try_disconnect(pdev);
++
++ device_unregister(&pdev->xdev->dev);
++ break;
++
++ case XenbusStateConnected:
++ pcifront_try_connect(pdev);
++ break;
++
++ case XenbusStateReconfiguring:
++ pcifront_detach_devices(pdev);
++ break;
++
++ case XenbusStateReconfigured:
++ pcifront_attach_devices(pdev);
++ break;
++
++ default:
++ break;
++ }
++}
++
++static int pcifront_xenbus_probe(struct xenbus_device *xdev,
++ const struct xenbus_device_id *id)
++{
++ int err = 0;
++ struct pcifront_device *pdev = alloc_pdev(xdev);
++
++ if (pdev == NULL) {
++ err = -ENOMEM;
++ xenbus_dev_fatal(xdev, err,
++ "Error allocating pcifront_device struct");
++ goto out;
++ }
++
++ err = pcifront_publish_info(pdev);
++
++ out:
++ return err;
++}
++
++static int pcifront_xenbus_remove(struct xenbus_device *xdev)
++{
++ if (xdev->dev.driver_data)
++ free_pdev(xdev->dev.driver_data);
++
++ return 0;
++}
++
++static const struct xenbus_device_id xenpci_ids[] = {
++ {"pci"},
++ {{0}},
++};
++MODULE_ALIAS("xen:pci");
++
++static struct xenbus_driver xenbus_pcifront_driver = {
++ .name = "pcifront",
++ .owner = THIS_MODULE,
++ .ids = xenpci_ids,
++ .probe = pcifront_xenbus_probe,
++ .remove = pcifront_xenbus_remove,
++ .otherend_changed = pcifront_backend_changed,
++};
++
++static int __init pcifront_init(void)
++{
++ if (!is_running_on_xen())
++ return -ENODEV;
++
++ return xenbus_register_frontend(&xenbus_pcifront_driver);
++}
++
++/* Initialize after the Xen PCI Frontend Stub is initialized */
++subsys_initcall(pcifront_init);
+--- linux-2.6.18.8/drivers/xen/privcmd/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/privcmd/Makefile 2008-05-19 00:33:46.322816221 +0300
+@@ -0,0 +1,3 @@
++
++obj-y += privcmd.o
++obj-$(CONFIG_COMPAT) += compat_privcmd.o
+--- linux-2.6.18.8/drivers/xen/privcmd/compat_privcmd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/privcmd/compat_privcmd.c 2008-05-19 00:33:46.346817604 +0300
+@@ -0,0 +1,73 @@
++/*
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ * Copyright (C) IBM Corp. 2006
++ *
++ * Authors: Jimi Xenidis <jimix@watson.ibm.com>
++ */
++
++#include <linux/config.h>
++#include <linux/compat.h>
++#include <linux/ioctl.h>
++#include <linux/syscalls.h>
++#include <asm/hypervisor.h>
++#include <asm/uaccess.h>
++#include <xen/public/privcmd.h>
++#include <xen/compat_ioctl.h>
++
++int privcmd_ioctl_32(int fd, unsigned int cmd, unsigned long arg)
++{
++ int ret;
++
++ switch (cmd) {
++ case IOCTL_PRIVCMD_MMAP_32: {
++ struct privcmd_mmap *p;
++ struct privcmd_mmap_32 *p32;
++ struct privcmd_mmap_32 n32;
++
++ p32 = compat_ptr(arg);
++ p = compat_alloc_user_space(sizeof(*p));
++ if (copy_from_user(&n32, p32, sizeof(n32)) ||
++ put_user(n32.num, &p->num) ||
++ put_user(n32.dom, &p->dom) ||
++ put_user(compat_ptr(n32.entry), &p->entry))
++ return -EFAULT;
++
++ ret = sys_ioctl(fd, IOCTL_PRIVCMD_MMAP, (unsigned long)p);
++ }
++ break;
++ case IOCTL_PRIVCMD_MMAPBATCH_32: {
++ struct privcmd_mmapbatch *p;
++ struct privcmd_mmapbatch_32 *p32;
++ struct privcmd_mmapbatch_32 n32;
++
++ p32 = compat_ptr(arg);
++ p = compat_alloc_user_space(sizeof(*p));
++ if (copy_from_user(&n32, p32, sizeof(n32)) ||
++ put_user(n32.num, &p->num) ||
++ put_user(n32.dom, &p->dom) ||
++ put_user(n32.addr, &p->addr) ||
++ put_user(compat_ptr(n32.arr), &p->arr))
++ return -EFAULT;
++
++ ret = sys_ioctl(fd, IOCTL_PRIVCMD_MMAPBATCH, (unsigned long)p);
++ }
++ break;
++ default:
++ ret = -EINVAL;
++ break;
++ }
++ return ret;
++}
+--- linux-2.6.18.8/drivers/xen/privcmd/privcmd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/privcmd/privcmd.c 2008-05-19 00:33:46.350817835 +0300
+@@ -0,0 +1,284 @@
++/******************************************************************************
++ * privcmd.c
++ *
++ * Interface to privileged domain-0 commands.
++ *
++ * Copyright (c) 2002-2004, K A Fraser, B Dragovic
++ */
++
++#include <linux/kernel.h>
++#include <linux/sched.h>
++#include <linux/slab.h>
++#include <linux/string.h>
++#include <linux/errno.h>
++#include <linux/mm.h>
++#include <linux/mman.h>
++#include <linux/swap.h>
++#include <linux/smp_lock.h>
++#include <linux/highmem.h>
++#include <linux/pagemap.h>
++#include <linux/seq_file.h>
++#include <asm/hypervisor.h>
++
++#include <asm/pgalloc.h>
++#include <asm/pgtable.h>
++#include <asm/uaccess.h>
++#include <asm/tlb.h>
++#include <asm/hypervisor.h>
++#include <xen/public/privcmd.h>
++#include <xen/interface/xen.h>
++#include <xen/xen_proc.h>
++#include <xen/features.h>
++
++static struct proc_dir_entry *privcmd_intf;
++static struct proc_dir_entry *capabilities_intf;
++
++#ifndef HAVE_ARCH_PRIVCMD_MMAP
++static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
++#endif
++
++static long privcmd_ioctl(struct file *file,
++ unsigned int cmd, unsigned long data)
++{
++ int ret = -ENOSYS;
++ void __user *udata = (void __user *) data;
++
++ switch (cmd) {
++ case IOCTL_PRIVCMD_HYPERCALL: {
++ privcmd_hypercall_t hypercall;
++
++ if (copy_from_user(&hypercall, udata, sizeof(hypercall)))
++ return -EFAULT;
++
++#if defined(__i386__)
++ if (hypercall.op >= (PAGE_SIZE >> 5))
++ break;
++ __asm__ __volatile__ (
++ "pushl %%ebx; pushl %%ecx; pushl %%edx; "
++ "pushl %%esi; pushl %%edi; "
++ "movl 8(%%eax),%%ebx ;"
++ "movl 16(%%eax),%%ecx ;"
++ "movl 24(%%eax),%%edx ;"
++ "movl 32(%%eax),%%esi ;"
++ "movl 40(%%eax),%%edi ;"
++ "movl (%%eax),%%eax ;"
++ "shll $5,%%eax ;"
++ "addl $hypercall_page,%%eax ;"
++ "call *%%eax ;"
++ "popl %%edi; popl %%esi; popl %%edx; "
++ "popl %%ecx; popl %%ebx"
++ : "=a" (ret) : "0" (&hypercall) : "memory" );
++#elif defined (__x86_64__)
++ if (hypercall.op < (PAGE_SIZE >> 5)) {
++ long ign1, ign2, ign3;
++ __asm__ __volatile__ (
++ "movq %8,%%r10; movq %9,%%r8;"
++ "shll $5,%%eax ;"
++ "addq $hypercall_page,%%rax ;"
++ "call *%%rax"
++ : "=a" (ret), "=D" (ign1),
++ "=S" (ign2), "=d" (ign3)
++ : "0" ((unsigned int)hypercall.op),
++ "1" (hypercall.arg[0]),
++ "2" (hypercall.arg[1]),
++ "3" (hypercall.arg[2]),
++ "g" (hypercall.arg[3]),
++ "g" (hypercall.arg[4])
++ : "r8", "r10", "memory" );
++ }
++#else
++ ret = privcmd_hypercall(&hypercall);
++#endif
++ }
++ break;
++
++ case IOCTL_PRIVCMD_MMAP: {
++ privcmd_mmap_t mmapcmd;
++ privcmd_mmap_entry_t msg;
++ privcmd_mmap_entry_t __user *p;
++ struct mm_struct *mm = current->mm;
++ struct vm_area_struct *vma;
++ unsigned long va;
++ int i, rc;
++
++ if (!is_initial_xendomain())
++ return -EPERM;
++
++ if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
++ return -EFAULT;
++
++ p = mmapcmd.entry;
++ if (copy_from_user(&msg, p, sizeof(msg)))
++ return -EFAULT;
++
++ down_write(&mm->mmap_sem);
++
++ vma = find_vma(mm, msg.va);
++ rc = -EINVAL;
++ if (!vma || (msg.va != vma->vm_start) ||
++ !privcmd_enforce_singleshot_mapping(vma))
++ goto mmap_out;
++
++ va = vma->vm_start;
++
++ for (i = 0; i < mmapcmd.num; i++) {
++ rc = -EFAULT;
++ if (copy_from_user(&msg, p, sizeof(msg)))
++ goto mmap_out;
++
++ /* Do not allow range to wrap the address space. */
++ rc = -EINVAL;
++ if ((msg.npages > (LONG_MAX >> PAGE_SHIFT)) ||
++ ((unsigned long)(msg.npages << PAGE_SHIFT) >= -va))
++ goto mmap_out;
++
++ /* Range chunks must be contiguous in va space. */
++ if ((msg.va != va) ||
++ ((msg.va+(msg.npages<<PAGE_SHIFT)) > vma->vm_end))
++ goto mmap_out;
++
++ if ((rc = direct_remap_pfn_range(
++ vma,
++ msg.va & PAGE_MASK,
++ msg.mfn,
++ msg.npages << PAGE_SHIFT,
++ vma->vm_page_prot,
++ mmapcmd.dom)) < 0)
++ goto mmap_out;
++
++ p++;
++ va += msg.npages << PAGE_SHIFT;
++ }
++
++ rc = 0;
++
++ mmap_out:
++ up_write(&mm->mmap_sem);
++ ret = rc;
++ }
++ break;
++
++ case IOCTL_PRIVCMD_MMAPBATCH: {
++ privcmd_mmapbatch_t m;
++ struct mm_struct *mm = current->mm;
++ struct vm_area_struct *vma;
++ xen_pfn_t __user *p;
++ unsigned long addr, mfn, nr_pages;
++ int i;
++
++ if (!is_initial_xendomain())
++ return -EPERM;
++
++ if (copy_from_user(&m, udata, sizeof(m)))
++ return -EFAULT;
++
++ nr_pages = m.num;
++ if ((m.num <= 0) || (nr_pages > (LONG_MAX >> PAGE_SHIFT)))
++ return -EINVAL;
++
++ down_write(&mm->mmap_sem);
++
++ vma = find_vma(mm, m.addr);
++ if (!vma ||
++ (m.addr != vma->vm_start) ||
++ ((m.addr + (nr_pages << PAGE_SHIFT)) != vma->vm_end) ||
++ !privcmd_enforce_singleshot_mapping(vma)) {
++ up_write(&mm->mmap_sem);
++ return -EINVAL;
++ }
++
++ p = m.arr;
++ addr = m.addr;
++ for (i = 0; i < nr_pages; i++, addr += PAGE_SIZE, p++) {
++ if (get_user(mfn, p)) {
++ up_write(&mm->mmap_sem);
++ return -EFAULT;
++ }
++
++ ret = direct_remap_pfn_range(vma, addr & PAGE_MASK,
++ mfn, PAGE_SIZE,
++ vma->vm_page_prot, m.dom);
++ if (ret < 0)
++ put_user(0xF0000000 | mfn, p);
++ }
++
++ up_write(&mm->mmap_sem);
++ ret = 0;
++ }
++ break;
++
++ default:
++ ret = -EINVAL;
++ break;
+ }
+
-+ xenbus_switch_state(dev, XenbusStateInitialised);
++ return ret;
++}
++
++#ifndef HAVE_ARCH_PRIVCMD_MMAP
++static struct page *privcmd_nopage(struct vm_area_struct *vma,
++ unsigned long address,
++ int *type)
++{
++ return NOPAGE_SIGBUS;
++}
++
++static struct vm_operations_struct privcmd_vm_ops = {
++ .nopage = privcmd_nopage
++};
++
++static int privcmd_mmap(struct file * file, struct vm_area_struct * vma)
++{
++ /* Unsupported for auto-translate guests. */
++ if (xen_feature(XENFEAT_auto_translated_physmap))
++ return -ENOSYS;
++
++ /* DONTCOPY is essential for Xen as copy_page_range is broken. */
++ vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY;
++ vma->vm_ops = &privcmd_vm_ops;
++ vma->vm_private_data = NULL;
++
+ return 0;
++}
+
-+ error_xenbus:
-+ xenbus_transaction_end(xbt, 1);
-+ xenbus_dev_fatal(dev, ret, "writing xenstore");
-+ return ret;
++static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
++{
++ return (xchg(&vma->vm_private_data, (void *)1) == NULL);
+}
++#endif
+
-+static void xenkbd_disconnect_backend(struct xenkbd_info *info)
++static const struct file_operations privcmd_file_ops = {
++ .unlocked_ioctl = privcmd_ioctl,
++ .mmap = privcmd_mmap,
++};
++
++static int capabilities_read(char *page, char **start, off_t off,
++ int count, int *eof, void *data)
+{
-+ if (info->irq >= 0)
-+ unbind_from_irqhandler(info->irq, info);
-+ info->irq = -1;
++ int len = 0;
++ *page = 0;
++
++ if (is_initial_xendomain())
++ len = sprintf( page, "control_d\n" );
++
++ *eof = 1;
++ return len;
+}
+
-+static void xenkbd_backend_changed(struct xenbus_device *dev,
-+ enum xenbus_state backend_state)
++static int __init privcmd_init(void)
+{
-+ struct xenkbd_info *info = dev->dev.driver_data;
-+ int ret, val;
++ if (!is_running_on_xen())
++ return -ENODEV;
+
-+ switch (backend_state) {
-+ case XenbusStateInitialising:
-+ case XenbusStateInitialised:
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ break;
++ privcmd_intf = create_xen_proc_entry("privcmd", 0400);
++ if (privcmd_intf != NULL)
++ privcmd_intf->proc_fops = &privcmd_file_ops;
+
-+ case XenbusStateInitWait:
-+ InitWait:
-+ ret = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
-+ "feature-abs-pointer", "%d", &val);
-+ if (ret < 0)
-+ val = 0;
-+ if (val) {
-+ ret = xenbus_printf(XBT_NIL, info->xbdev->nodename,
-+ "request-abs-pointer", "1");
-+ if (ret)
-+ ; /* FIXME */
-+ }
-+ xenbus_switch_state(dev, XenbusStateConnected);
-+ break;
++ capabilities_intf = create_xen_proc_entry("capabilities", 0400 );
++ if (capabilities_intf != NULL)
++ capabilities_intf->read_proc = capabilities_read;
+
-+ case XenbusStateConnected:
-+ /*
-+ * Work around xenbus race condition: If backend goes
-+ * through InitWait to Connected fast enough, we can
-+ * get Connected twice here.
-+ */
-+ if (dev->state != XenbusStateConnected)
-+ goto InitWait; /* no InitWait seen yet, fudge it */
-+ break;
++ return 0;
++}
+
-+ case XenbusStateClosing:
-+ xenbus_frontend_closed(dev);
-+ break;
++__initcall(privcmd_init);
+--- linux-2.6.18.8/drivers/xen/sfc_netback/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/Makefile 2008-05-19 00:33:46.354818065 +0300
+@@ -0,0 +1,12 @@
++EXTRA_CFLAGS += -Idrivers/xen/sfc_netback -Idrivers/xen/sfc_netutil -Idrivers/xen/netback -Idrivers/net/sfc
++EXTRA_CFLAGS += -D__ci_driver__
++EXTRA_CFLAGS += -DEFX_USE_KCOMPAT
++EXTRA_CFLAGS += -Werror
++
++ifdef GCOV
++EXTRA_CFLAGS += -fprofile-arcs -ftest-coverage -DEFX_GCOV
++endif
++
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_BACKEND) := sfc_netback.o
++
++sfc_netback-objs := accel.o accel_fwd.o accel_msg.o accel_solarflare.o accel_xenbus.o accel_debugfs.o
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel.c 2008-05-19 00:33:46.414821524 +0300
+@@ -0,0 +1,129 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#include "accel.h"
++#include "accel_msg_iface.h"
++#include "accel_solarflare.h"
++
++#include <linux/notifier.h>
++
++#ifdef EFX_GCOV
++#include "gcov.h"
++#endif
++
++static int netback_accel_netdev_event(struct notifier_block *nb,
++ unsigned long event, void *ptr)
++{
++ struct net_device *net_dev = (struct net_device *)ptr;
++ struct netback_accel *bend;
++
++ if ((event == NETDEV_UP) || (event == NETDEV_DOWN)) {
++ mutex_lock(&bend_list_mutex);
++ bend = bend_list;
++ while (bend != NULL) {
++ mutex_lock(&bend->bend_mutex);
++ /*
++ * This happens when the shared pages have
++ * been unmapped, but the bend not yet removed
++ * from list
++ */
++ if (bend->shared_page == NULL)
++ goto next;
++
++ if (bend->net_dev->ifindex == net_dev->ifindex)
++ netback_accel_set_interface_state
++ (bend, event == NETDEV_UP);
++
++ next:
++ mutex_unlock(&bend->bend_mutex);
++ bend = bend->next_bend;
++ }
++ mutex_unlock(&bend_list_mutex);
+ }
++
++ return NOTIFY_DONE;
+}
+
-+static struct xenbus_device_id xenkbd_ids[] = {
-+ { "vkbd" },
-+ { "" }
-+};
-+MODULE_ALIAS("xen:vkbd");
+
-+static struct xenbus_driver xenkbd = {
-+ .name = "vkbd",
-+ .owner = THIS_MODULE,
-+ .ids = xenkbd_ids,
-+ .probe = xenkbd_probe,
-+ .remove = xenkbd_remove,
-+ .resume = xenkbd_resume,
-+ .otherend_changed = xenkbd_backend_changed,
++static struct notifier_block netback_accel_netdev_notifier = {
++ .notifier_call = netback_accel_netdev_event,
+};
+
-+static int __init xenkbd_init(void)
++
++unsigned sfc_netback_max_pages = NETBACK_ACCEL_DEFAULT_MAX_BUF_PAGES;
++module_param_named(max_pages, sfc_netback_max_pages, uint, 0644);
++MODULE_PARM_DESC(max_pages,
++ "The number of buffer pages to enforce on each guest");
++
++/* Initialise subsystems need for the accelerated fast path */
++static int __init netback_accel_init(void)
+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ int rc = 0;
+
-+ /* Nothing to do if running in dom0. */
-+ if (is_initial_xendomain())
-+ return -ENODEV;
++#ifdef EFX_GCOV
++ gcov_provider_init(THIS_MODULE);
++#endif
+
-+ return xenbus_register_frontend(&xenkbd);
++ rc = netback_accel_init_fwd();
++
++ if (rc == 0)
++ netback_accel_debugfs_init();
++
++ if (rc == 0)
++ rc = netback_accel_sf_init();
++
++ if (rc == 0)
++ rc = register_netdevice_notifier
++ (&netback_accel_netdev_notifier);
++
++ /*
++ * What if no device was found, shouldn't we clean up stuff
++ * we've allocated for acceleration subsystem?
++ */
++
++ return rc;
+}
+
-+static void __exit xenkbd_cleanup(void)
++module_init(netback_accel_init);
++
++static void __exit netback_accel_exit(void)
+{
-+ return xenbus_unregister_driver(&xenkbd);
++ unregister_netdevice_notifier(&netback_accel_netdev_notifier);
++
++ netback_accel_sf_shutdown();
++
++ netback_accel_shutdown_bends();
++
++ netback_accel_debugfs_fini();
++
++ netback_accel_shutdown_fwd();
++
++#ifdef EFX_GCOV
++ gcov_provider_fini(THIS_MODULE);
++#endif
+}
+
-+module_init(xenkbd_init);
-+module_exit(xenkbd_cleanup);
++module_exit(netback_accel_exit);
+
+MODULE_LICENSE("GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/gntdev/gntdev.c linux-2.6.18-xen.hg/drivers/xen/gntdev/gntdev.c
---- linux-2.6.18/drivers/xen/gntdev/gntdev.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/gntdev/gntdev.c 2007-12-23 11:15:34.051268764 +0100
-@@ -0,0 +1,973 @@
-+/******************************************************************************
-+ * gntdev.c
-+ *
-+ * Device for accessing (in user-space) pages that have been granted by other
-+ * domains.
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel.h 2008-05-19 00:33:46.414821524 +0300
+@@ -0,0 +1,393 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
-+ * Copyright (c) 2006-2007, D G Murray.
-+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef NETBACK_ACCEL_H
++#define NETBACK_ACCEL_H
++
++#include <linux/slab.h>
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/udp.h>
++#include <linux/in.h>
++#include <linux/netdevice.h>
++#include <linux/etherdevice.h>
++#include <linux/mutex.h>
++#include <linux/wait.h>
++
++#include <xen/xenbus.h>
++
++#include "accel_shared_fifo.h"
++#include "accel_msg_iface.h"
++#include "accel_util.h"
++
++/**************************************************************************
++ * Datatypes
++ **************************************************************************/
++
++#define NETBACK_ACCEL_DEFAULT_MAX_FILTERS (8)
++#define NETBACK_ACCEL_DEFAULT_MAX_MCASTS (8)
++#define NETBACK_ACCEL_DEFAULT_MAX_BUF_PAGES (384)
++/* Variable to store module parameter for max_buf_pages */
++extern unsigned sfc_netback_max_pages;
++
++#define NETBACK_ACCEL_STATS 1
++
++#if NETBACK_ACCEL_STATS
++#define NETBACK_ACCEL_STATS_OP(x) x
++#else
++#define NETBACK_ACCEL_STATS_OP(x)
++#endif
++
++/*! Statistics for a given backend */
++struct netback_accel_stats {
++ /*! Number of eventq wakeup events */
++ u64 evq_wakeups;
++ /*! Number of eventq timeout events */
++ u64 evq_timeouts;
++ /*! Number of filters used */
++ u32 num_filters;
++ /*! Number of buffer pages registered */
++ u32 num_buffer_pages;
++};
++
++
++/* Debug fs nodes for each of the above stats */
++struct netback_accel_dbfs {
++ struct dentry *evq_wakeups;
++ struct dentry *evq_timeouts;
++ struct dentry *num_filters;
++ struct dentry *num_buffer_pages;
++};
++
++
++/*! Resource limits for a given NIC */
++struct netback_accel_limits {
++ int max_filters; /*!< Max. number of filters to use. */
++ int max_mcasts; /*!< Max. number of mcast subscriptions */
++ int max_buf_pages; /*!< Max. number of pages of NIC buffers */
++};
++
++
++/*! The state for an instance of the back end driver. */
++struct netback_accel {
++ /*! mutex to protect this state */
++ struct mutex bend_mutex;
++
++ /*! Watches on xenstore */
++ struct xenbus_watch domu_accel_watch;
++ struct xenbus_watch config_accel_watch;
++
++ /*! Pointer to whatever device cookie ties us in to the hypervisor */
++ void *hdev_data;
++
++ /*! FIFO indices. Next page is msg FIFOs */
++ struct net_accel_shared_page *shared_page;
++
++ /*! Defer control message processing */
++ struct work_struct handle_msg;
++
++ /*! Identifies other end VM and interface.*/
++ int far_end;
++ int vif_num;
++
++ /*!< To unmap the shared pages */
++ void *sh_pages_unmap;
++
++ /* Resource tracking */
++ /*! Limits on H/W & Dom0 resources */
++ struct netback_accel_limits quotas;
++
++ /* Hardware resources */
++ /*! The H/W type of associated NIC */
++ enum net_accel_hw_type hw_type;
++ /*! State of allocation */
++ int hw_state;
++ /*! Index into ci_driver.nics[] for this interface */
++ int nic_index;
++ /*! How to set up the acceleration for this hardware */
++ int (*accel_setup)(struct netback_accel *);
++ /*! And how to stop it. */
++ void (*accel_shutdown)(struct netback_accel *);
++
++ /*! The physical/real net_dev for this interface */
++ struct net_device *net_dev;
++
++ /*! Magic pointer to locate state in fowarding table */
++ void *fwd_priv;
++
++ /*! Message FIFO */
++ sh_msg_fifo2 to_domU;
++ /*! Message FIFO */
++ sh_msg_fifo2 from_domU;
++
++ /*! General notification channel id */
++ int msg_channel;
++ /*! General notification channel irq */
++ int msg_channel_irq;
++
++ /*! Event channel id dedicated to network packet interrupts. */
++ int net_channel;
++ /*! Event channel irq dedicated to network packets interrupts */
++ int net_channel_irq;
++
++ /*! The MAC address the frontend goes by. */
++ u8 mac[ETH_ALEN];
++ /*! Driver name of associated NIC */
++ char *nicname;
++
++ /*! Array of pointers to buffer pages mapped */
++ grant_handle_t *buffer_maps;
++ u64 *buffer_addrs;
++ /*! Index into buffer_maps */
++ int buffer_maps_index;
++ /*! Max number of pages that domU is allowed/will request to map */
++ int max_pages;
++
++ /*! Pointer to hardware specific private area */
++ void *accel_hw_priv;
++
++ /*! Wait queue for changes in accelstate. */
++ wait_queue_head_t state_wait_queue;
++
++ /*! Current state of the frontend according to the xenbus
++ * watch. */
++ XenbusState frontend_state;
++
++ /*! Current state of this backend. */
++ XenbusState backend_state;
++
++ /*! Non-zero if the backend is being removed. */
++ int removing;
++
++ /*! Non-zero if the setup_vnic has been called. */
++ int vnic_is_setup;
++
++#if NETBACK_ACCEL_STATS
++ struct netback_accel_stats stats;
++#endif
++#if defined(CONFIG_DEBUG_FS)
++ char *dbfs_dir_name;
++ struct dentry *dbfs_dir;
++ struct netback_accel_dbfs dbfs;
++#endif
++
++ /*! List */
++ struct netback_accel *next_bend;
++};
++
++
++/*
++ * Values for netback_accel.hw_state. States of resource allocation
++ * we can go through
++ */
++/*! No hardware has yet been allocated. */
++#define NETBACK_ACCEL_RES_NONE (0)
++/*! Hardware has been allocated. */
++#define NETBACK_ACCEL_RES_ALLOC (1)
++#define NETBACK_ACCEL_RES_FILTER (2)
++#define NETBACK_ACCEL_RES_HWINFO (3)
++
++/*! Filtering specification. This assumes that for VNIC support we
++ * will always want wildcard entries, so only specifies the
++ * destination IP/port
++ */
++struct netback_accel_filter_spec {
++ /*! Internal, used to access efx_vi API */
++ void *filter_handle;
++
++ /*! Destination IP in network order */
++ u32 destip_be;
++ /*! Destination port in network order */
++ u16 destport_be;
++ /*! Mac address */
++ u8 mac[ETH_ALEN];
++ /*! TCP or UDP */
++ u8 proto;
++};
++
++
++/**************************************************************************
++ * From accel.c
++ **************************************************************************/
++
++/*! \brief Start up all the acceleration plugins
++ *
++ * \return 0 on success, an errno on failure
++ */
++extern int netback_accel_init_accel(void);
++
++/*! \brief Shut down all the acceleration plugins
++ */
++extern void netback_accel_shutdown_accel(void);
++
++
++/**************************************************************************
++ * From accel_fwd.c
++ **************************************************************************/
++
++/*! \brief Init the forwarding infrastructure
++ * \return 0 on success, or -ENOMEM if it couldn't get memory for the
++ * forward table
++ */
++extern int netback_accel_init_fwd(void);
++
++/*! \brief Shut down the forwarding and free memory. */
++extern void netback_accel_shutdown_fwd(void);
++
++/*! Initialise each nic port's fowarding table */
++extern void *netback_accel_init_fwd_port(void);
++extern void netback_accel_shutdown_fwd_port(void *fwd_priv);
++
++/*! \brief Add an entry to the forwarding table.
++ * \param mac : MAC address, used as hash key
++ * \param ctxt : value to associate with key (can be NULL, see
++ * netback_accel_fwd_set_context)
++ * \return 0 on success, -ENOMEM if table was full and could no grow it
++ */
++extern int netback_accel_fwd_add(const __u8 *mac, void *context,
++ void *fwd_priv);
++
++/*! \brief Remove an entry from the forwarding table.
++ * \param mac : the MAC address to remove
++ * \return nothing: it is not an error if the mac was not in the table
++ */
++extern void netback_accel_fwd_remove(const __u8 *mac, void *fwd_priv);
++
++/*! \brief Set the context pointer for an existing fwd table entry.
++ * \param mac : key that is already present in the table
++ * \param context : new value to associate with key
++ * \return 0 on success, -ENOENT if mac not present in table.
++ */
++extern int netback_accel_fwd_set_context(const __u8 *mac, void *context,
++ void *fwd_priv);
++
++/**************************************************************************
++ * From accel_msg.c
++ **************************************************************************/
++
++
++/*! \brief Send the start-of-day message that handshakes with the VNIC
++ * and tells it its MAC address.
++ *
++ * \param bend The back end driver data structure
++ * \param version The version of communication to use, e.g. NET_ACCEL_MSG_VERSION
++ */
++extern void netback_accel_msg_tx_hello(struct netback_accel *bend,
++ unsigned version);
++
++/*! \brief Send a "there's a new local mac address" message
++ *
++ * \param bend The back end driver data structure for the vnic to send
++ * the message to
++ * \param mac Pointer to the new mac address
++ */
++extern void netback_accel_msg_tx_new_localmac(struct netback_accel *bend,
++ const void *mac);
++
++/*! \brief Send a "a mac address that was local has gone away" message
++ *
++ * \param bend The back end driver data structure for the vnic to send
++ * the message to
++ * \param mac Pointer to the old mac address
++ */
++extern void netback_accel_msg_tx_old_localmac(struct netback_accel *bend,
++ const void *mac);
++
++extern void netback_accel_set_interface_state(struct netback_accel *bend,
++ int up);
++
++/*! \brief Process the message queue for a bend that has just
++ * interrupted.
++ *
++ * Demultiplexs an interrupt from the front end driver, taking
++ * messages from the fifo and taking appropriate action.
+ *
++ * \param bend The back end driver data structure
++ */
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++extern void netback_accel_msg_rx_handler(struct work_struct *arg);
++#else
++extern void netback_accel_msg_rx_handler(void *bend_void);
++#endif
++
++/**************************************************************************
++ * From accel_xenbus.c
++ **************************************************************************/
++/*! List of all the bends currently in existence. */
++extern struct netback_accel *bend_list;
++extern struct mutex bend_list_mutex;
++
++/*! \brief Probe a new network interface. */
++extern int netback_accel_probe(struct xenbus_device *dev);
++
++/*! \brief Remove a network interface. */
++extern int netback_accel_remove(struct xenbus_device *dev);
++
++/*! \brief Shutdown all accelerator backends */
++extern void netback_accel_shutdown_bends(void);
++
++/*! \brief Initiate the xenbus state teardown handshake */
++extern void netback_accel_set_closing(struct netback_accel *bend);
++
++/**************************************************************************
++ * From accel_debugfs.c
++ **************************************************************************/
++/*! Global statistics */
++struct netback_accel_global_stats {
++ /*! Number of TX packets seen through driverlink */
++ u64 dl_tx_packets;
++ /*! Number of TX packets seen through driverlink we didn't like */
++ u64 dl_tx_bad_packets;
++ /*! Number of RX packets seen through driverlink */
++ u64 dl_rx_packets;
++ /*! Number of mac addresses we are forwarding to */
++ u32 num_fwds;
++};
++
++/*! Debug fs entries for each of the above stats */
++struct netback_accel_global_dbfs {
++ struct dentry *dl_tx_packets;
++ struct dentry *dl_tx_bad_packets;
++ struct dentry *dl_rx_packets;
++ struct dentry *num_fwds;
++};
++
++#if NETBACK_ACCEL_STATS
++extern struct netback_accel_global_stats global_stats;
++#endif
++
++/*! \brief Initialise the debugfs root and populate with global stats */
++extern void netback_accel_debugfs_init(void);
++
++/*! \brief Remove our debugfs root directory */
++extern void netback_accel_debugfs_fini(void);
++
++/*! \brief Add per-bend statistics to debug fs */
++extern int netback_accel_debugfs_create(struct netback_accel *bend);
++/*! \brief Remove per-bend statistics from debug fs */
++extern int netback_accel_debugfs_remove(struct netback_accel *bend);
++
++#endif /* NETBACK_ACCEL_H */
++
++
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_debugfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_debugfs.c 2008-05-19 00:33:46.418821755 +0300
+@@ -0,0 +1,148 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <asm/atomic.h>
-+#include <linux/module.h>
-+#include <linux/kernel.h>
-+#include <linux/init.h>
+#include <linux/fs.h>
-+#include <linux/device.h>
-+#include <linux/mm.h>
-+#include <linux/mman.h>
-+#include <asm/uaccess.h>
-+#include <asm/io.h>
-+#include <xen/gnttab.h>
-+#include <asm/hypervisor.h>
-+#include <xen/balloon.h>
-+#include <xen/evtchn.h>
-+#include <xen/driver_util.h>
++#include <linux/debugfs.h>
+
-+#include <linux/types.h>
-+#include <xen/public/gntdev.h>
++#include "accel.h"
+
++#if defined(CONFIG_DEBUG_FS)
++static struct dentry *sfc_debugfs_root = NULL;
++#endif
+
-+#define DRIVER_AUTHOR "Derek G. Murray <Derek.Murray@cl.cam.ac.uk>"
-+#define DRIVER_DESC "User-space granted page access driver"
++#if NETBACK_ACCEL_STATS
++struct netback_accel_global_stats global_stats;
++#if defined(CONFIG_DEBUG_FS)
++static struct netback_accel_global_dbfs global_dbfs;
++#endif
++#endif
+
-+MODULE_LICENSE("GPL");
-+MODULE_AUTHOR(DRIVER_AUTHOR);
-+MODULE_DESCRIPTION(DRIVER_DESC);
++void netback_accel_debugfs_init(void)
++{
++#if defined(CONFIG_DEBUG_FS)
++ sfc_debugfs_root = debugfs_create_dir("sfc_netback", NULL);
++ if (sfc_debugfs_root == NULL)
++ return;
+
-+#define MAX_GRANTS 128
++ global_dbfs.num_fwds = debugfs_create_u32
++ ("num_fwds", S_IRUSR | S_IRGRP | S_IROTH,
++ sfc_debugfs_root, &global_stats.num_fwds);
++ global_dbfs.dl_tx_packets = debugfs_create_u64
++ ("dl_tx_packets", S_IRUSR | S_IRGRP | S_IROTH,
++ sfc_debugfs_root, &global_stats.dl_tx_packets);
++ global_dbfs.dl_rx_packets = debugfs_create_u64
++ ("dl_rx_packets", S_IRUSR | S_IRGRP | S_IROTH,
++ sfc_debugfs_root, &global_stats.dl_rx_packets);
++ global_dbfs.dl_tx_bad_packets = debugfs_create_u64
++ ("dl_tx_bad_packets", S_IRUSR | S_IRGRP | S_IROTH,
++ sfc_debugfs_root, &global_stats.dl_tx_bad_packets);
++#endif
++}
+
-+/* A slot can be in one of three states:
++
++void netback_accel_debugfs_fini(void)
++{
++#if defined(CONFIG_DEBUG_FS)
++ debugfs_remove(global_dbfs.num_fwds);
++ debugfs_remove(global_dbfs.dl_tx_packets);
++ debugfs_remove(global_dbfs.dl_rx_packets);
++ debugfs_remove(global_dbfs.dl_tx_bad_packets);
++
++ debugfs_remove(sfc_debugfs_root);
++#endif
++}
++
++
++int netback_accel_debugfs_create(struct netback_accel *bend)
++{
++#if defined(CONFIG_DEBUG_FS)
++ /* Smallest length is 7 (vif0.0\n) */
++ int length = 7, temp;
++
++ if (sfc_debugfs_root == NULL)
++ return -ENOENT;
++
++ /* Work out length of string representation of far_end and vif_num */
++ temp = bend->far_end;
++ while (temp > 9) {
++ length++;
++ temp = temp / 10;
++ }
++ temp = bend->vif_num;
++ while (temp > 9) {
++ length++;
++ temp = temp / 10;
++ }
++
++ bend->dbfs_dir_name = kmalloc(length, GFP_KERNEL);
++ if (bend->dbfs_dir_name == NULL)
++ return -ENOMEM;
++ sprintf(bend->dbfs_dir_name, "vif%d.%d", bend->far_end, bend->vif_num);
++
++ bend->dbfs_dir = debugfs_create_dir(bend->dbfs_dir_name,
++ sfc_debugfs_root);
++ if (bend->dbfs_dir == NULL) {
++ kfree(bend->dbfs_dir_name);
++ return -ENOMEM;
++ }
++
++#if NETBACK_ACCEL_STATS
++ bend->dbfs.evq_wakeups = debugfs_create_u64
++ ("evq_wakeups", S_IRUSR | S_IRGRP | S_IROTH,
++ bend->dbfs_dir, &bend->stats.evq_wakeups);
++ bend->dbfs.evq_timeouts = debugfs_create_u64
++ ("evq_timeouts", S_IRUSR | S_IRGRP | S_IROTH,
++ bend->dbfs_dir, &bend->stats.evq_timeouts);
++ bend->dbfs.num_filters = debugfs_create_u32
++ ("num_filters", S_IRUSR | S_IRGRP | S_IROTH,
++ bend->dbfs_dir, &bend->stats.num_filters);
++ bend->dbfs.num_buffer_pages = debugfs_create_u32
++ ("num_buffer_pages", S_IRUSR | S_IRGRP | S_IROTH,
++ bend->dbfs_dir, &bend->stats.num_buffer_pages);
++#endif
++#endif
++ return 0;
++}
++
++
++int netback_accel_debugfs_remove(struct netback_accel *bend)
++{
++#if defined(CONFIG_DEBUG_FS)
++ if (bend->dbfs_dir != NULL) {
++#if NETBACK_ACCEL_STATS
++ debugfs_remove(bend->dbfs.evq_wakeups);
++ debugfs_remove(bend->dbfs.evq_timeouts);
++ debugfs_remove(bend->dbfs.num_filters);
++ debugfs_remove(bend->dbfs.num_buffer_pages);
++#endif
++ debugfs_remove(bend->dbfs_dir);
++ }
++
++ if (bend->dbfs_dir_name)
++ kfree(bend->dbfs_dir_name);
++#endif
++ return 0;
++}
++
++
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_fwd.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_fwd.c 2008-05-19 00:33:46.510827058 +0300
+@@ -0,0 +1,420 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
+ *
-+ * 0. GNTDEV_SLOT_INVALID:
-+ * This slot is not associated with a grant reference, and is therefore free
-+ * to be overwritten by a new grant reference.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * 1. GNTDEV_SLOT_NOT_YET_MAPPED:
-+ * This slot is associated with a grant reference (via the
-+ * IOCTL_GNTDEV_MAP_GRANT_REF ioctl), but it has not yet been mmap()-ed.
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
+ *
-+ * 2. GNTDEV_SLOT_MAPPED:
-+ * This slot is associated with a grant reference, and has been mmap()-ed.
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+typedef enum gntdev_slot_state {
-+ GNTDEV_SLOT_INVALID = 0,
-+ GNTDEV_SLOT_NOT_YET_MAPPED,
-+ GNTDEV_SLOT_MAPPED
-+} gntdev_slot_state_t;
+
-+#define GNTDEV_INVALID_HANDLE -1
-+#define GNTDEV_FREE_LIST_INVALID -1
-+/* Each opened instance of gntdev is associated with a list of grants,
-+ * represented by an array of elements of the following type,
-+ * gntdev_grant_info_t.
-+ */
-+typedef struct gntdev_grant_info {
-+ gntdev_slot_state_t state;
-+ union {
-+ uint32_t free_list_index;
-+ struct {
-+ domid_t domid;
-+ grant_ref_t ref;
-+ grant_handle_t kernel_handle;
-+ grant_handle_t user_handle;
-+ uint64_t dev_bus_addr;
-+ } valid;
-+ } u;
-+} gntdev_grant_info_t;
++#include "accel.h"
++#include "accel_cuckoo_hash.h"
++#include "accel_util.h"
++#include "accel_solarflare.h"
+
-+/* Private data structure, which is stored in the file pointer for files
-+ * associated with this device.
++#include "driverlink_api.h"
++
++#include <linux/if_arp.h>
++#include <linux/skbuff.h>
++#include <linux/list.h>
++
++/* State stored in the forward table */
++struct fwd_struct {
++ struct list_head link; /* Forms list */
++ void * context;
++ __u8 valid;
++ __u8 mac[ETH_ALEN];
++};
++
++/* Max value we support */
++#define NUM_FWDS_BITS 8
++#define NUM_FWDS (1 << NUM_FWDS_BITS)
++#define FWD_MASK (NUM_FWDS - 1)
++
++struct port_fwd {
++ /* Make a list */
++ struct list_head link;
++ /* Hash table to store the fwd_structs */
++ cuckoo_hash_table fwd_hash_table;
++ /* The array of fwd_structs */
++ struct fwd_struct *fwd_array;
++ /* Linked list of entries in use. */
++ struct list_head fwd_list;
++ /* Could do something clever with a reader/writer lock. */
++ spinlock_t fwd_lock;
++ /* Make find_free_entry() a bit faster by caching this */
++ int last_free_index;
++};
++
++/*
++ * This is unlocked as it's only called from dl probe and remove,
++ * which are themselves synchronised. Could get rid of it entirely as
++ * it's never iterated, but useful for debug
+ */
-+typedef struct gntdev_file_private_data {
-+
-+ /* Array of grant information. */
-+ gntdev_grant_info_t grants[MAX_GRANTS];
++static struct list_head port_fwds;
+
-+ /* Read/write semaphore used to protect the grants array. */
-+ struct rw_semaphore grants_sem;
+
-+ /* An array of indices of free slots in the grants array.
-+ * N.B. An entry in this list may temporarily have the value
-+ * GNTDEV_FREE_LIST_INVALID if the corresponding slot has been removed
-+ * from the list by the contiguous allocator, but the list has not yet
-+ * been compressed. However, this is not visible across invocations of
-+ * the device.
-+ */
-+ int32_t free_list[MAX_GRANTS];
++/* Search the fwd_array for an unused entry */
++static int fwd_find_free_entry(struct port_fwd *fwd_set)
++{
++ int index = fwd_set->last_free_index;
++
++ do {
++ if (!fwd_set->fwd_array[index].valid) {
++ fwd_set->last_free_index = index;
++ return index;
++ }
++ index++;
++ if (index >= NUM_FWDS)
++ index = 0;
++ } while (index != fwd_set->last_free_index);
++
++ return -ENOMEM;
++}
++
++
++/* Look up a MAC in the hash table. Caller should hold table lock. */
++static inline struct fwd_struct *fwd_find_entry(const __u8 *mac,
++ struct port_fwd *fwd_set)
++{
++ cuckoo_hash_value value;
++ cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
++
++ if (cuckoo_hash_lookup(&fwd_set->fwd_hash_table,
++ (cuckoo_hash_key *)(&key),
++ &value)) {
++ struct fwd_struct *fwd = &fwd_set->fwd_array[value];
++ DPRINTK_ON(memcmp(fwd->mac, mac, ETH_ALEN) != 0);
++ return fwd;
++ }
++
++ return NULL;
++}
++
++
++/* Initialise each nic port's fowarding table */
++void *netback_accel_init_fwd_port(void)
++{
++ struct port_fwd *fwd_set;
++
++ fwd_set = kzalloc(sizeof(struct port_fwd), GFP_KERNEL);
++ if (fwd_set == NULL) {
++ return NULL;
++ }
++
++ spin_lock_init(&fwd_set->fwd_lock);
+
-+ /* The number of free slots in the grants array. */
-+ uint32_t free_list_size;
++ fwd_set->fwd_array = kzalloc(sizeof (struct fwd_struct) * NUM_FWDS,
++ GFP_KERNEL);
++ if (fwd_set->fwd_array == NULL) {
++ kfree(fwd_set);
++ return NULL;
++ }
++
++ if (cuckoo_hash_init(&fwd_set->fwd_hash_table, NUM_FWDS_BITS, 8) != 0) {
++ kfree(fwd_set->fwd_array);
++ kfree(fwd_set);
++ return NULL;
++ }
++
++ INIT_LIST_HEAD(&fwd_set->fwd_list);
++
++ list_add(&fwd_set->link, &port_fwds);
+
-+ /* Read/write semaphore used to protect the free list. */
-+ struct rw_semaphore free_list_sem;
++ return fwd_set;
++}
++
++
++void netback_accel_shutdown_fwd_port(void *fwd_priv)
++{
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
++
++ BUG_ON(fwd_priv == NULL);
+
-+ /* Index of the next slot after the most recent contiguous allocation,
-+ * for use in a next-fit allocator.
++ BUG_ON(list_empty(&port_fwds));
++ list_del(&fwd_set->link);
++
++ BUG_ON(!list_empty(&fwd_set->fwd_list));
++
++ cuckoo_hash_destroy(&fwd_set->fwd_hash_table);
++ kfree(fwd_set->fwd_array);
++ kfree(fwd_set);
++}
++
++
++int netback_accel_init_fwd()
++{
++ INIT_LIST_HEAD(&port_fwds);
++ return 0;
++}
++
++
++void netback_accel_shutdown_fwd()
++{
++ BUG_ON(!list_empty(&port_fwds));
++}
++
++
++/*
++ * Add an entry to the forwarding table. Returns -ENOMEM if no
++ * space.
++ */
++int netback_accel_fwd_add(const __u8 *mac, void *context, void *fwd_priv)
++{
++ struct fwd_struct *fwd;
++ int rc = 0, index;
++ unsigned long flags;
++ cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
++
++ BUG_ON(fwd_priv == NULL);
++
++ DPRINTK("Adding mac " MAC_FMT "\n", MAC_ARG(mac));
++
++ spin_lock_irqsave(&fwd_set->fwd_lock, flags);
++
++ if ((rc = fwd_find_free_entry(fwd_set)) < 0 ) {
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++ return rc;
++ }
++
++ index = rc;
++
++ /* Shouldn't already be in the table */
++ if (cuckoo_hash_lookup(&fwd_set->fwd_hash_table,
++ (cuckoo_hash_key *)(&key), &rc) != 0) {
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++ EPRINTK("MAC address " MAC_FMT " already accelerated.\n",
++ MAC_ARG(mac));
++ return -EEXIST;
++ }
++
++ if ((rc = cuckoo_hash_add(&fwd_set->fwd_hash_table,
++ (cuckoo_hash_key *)(&key), index, 1)) == 0) {
++ fwd = &fwd_set->fwd_array[index];
++ fwd->valid = 1;
++ fwd->context = context;
++ memcpy(fwd->mac, mac, ETH_ALEN);
++ list_add(&fwd->link, &fwd_set->fwd_list);
++ NETBACK_ACCEL_STATS_OP(global_stats.num_fwds++);
++ }
++
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++
++ /*
++ * No need to tell frontend that this mac address is local -
++ * it should auto-discover through packets on fastpath what is
++ * local and what is not, and just being on same server
++ * doesn't make it local (it could be on a different
++ * bridge)
+ */
-+ uint32_t next_fit_index;
+
-+ /* Used to map grants into the kernel, before mapping them into user
-+ * space.
++ return rc;
++}
++
++
++/* remove an entry from the forwarding tables. */
++void netback_accel_fwd_remove(const __u8 *mac, void *fwd_priv)
++{
++ struct fwd_struct *fwd;
++ unsigned long flags;
++ cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
++
++ DPRINTK("Removing mac " MAC_FMT "\n", MAC_ARG(mac));
++
++ BUG_ON(fwd_priv == NULL);
++
++ spin_lock_irqsave(&fwd_set->fwd_lock, flags);
++
++ fwd = fwd_find_entry(mac, fwd_set);
++ if (fwd != NULL) {
++ BUG_ON(list_empty(&fwd_set->fwd_list));
++ list_del(&fwd->link);
++
++ fwd->valid = 0;
++ cuckoo_hash_remove(&fwd_set->fwd_hash_table,
++ (cuckoo_hash_key *)(&key));
++ NETBACK_ACCEL_STATS_OP(global_stats.num_fwds--);
++ }
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++
++ /*
++ * No need to tell frontend that this is no longer present -
++ * the frontend is currently only interested in remote
++ * addresses and it works these out (mostly) by itself
+ */
-+ struct page **foreign_pages;
++}
+
-+} gntdev_file_private_data_t;
+
-+/* Module lifecycle operations. */
-+static int __init gntdev_init(void);
-+static void __exit gntdev_exit(void);
++/* Set the context pointer for a hash table entry. */
++int netback_accel_fwd_set_context(const __u8 *mac, void *context,
++ void *fwd_priv)
++{
++ struct fwd_struct *fwd;
++ unsigned long flags;
++ int rc = -ENOENT;
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
+
-+module_init(gntdev_init);
-+module_exit(gntdev_exit);
++ BUG_ON(fwd_priv == NULL);
+
-+/* File operations. */
-+static int gntdev_open(struct inode *inode, struct file *flip);
-+static int gntdev_release(struct inode *inode, struct file *flip);
-+static int gntdev_mmap(struct file *flip, struct vm_area_struct *vma);
-+static long gntdev_ioctl(struct file *flip,
-+ unsigned int cmd, unsigned long arg);
++ spin_lock_irqsave(&fwd_set->fwd_lock, flags);
++ fwd = fwd_find_entry(mac, fwd_set);
++ if (fwd != NULL) {
++ fwd->context = context;
++ rc = 0;
++ }
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++ return rc;
++}
+
-+static const struct file_operations gntdev_fops = {
-+ .owner = THIS_MODULE,
-+ .open = gntdev_open,
-+ .release = gntdev_release,
-+ .mmap = gntdev_mmap,
-+ .unlocked_ioctl = gntdev_ioctl
-+};
+
-+/* VM operations. */
-+static void gntdev_vma_close(struct vm_area_struct *vma);
-+static pte_t gntdev_clear_pte(struct vm_area_struct *vma, unsigned long addr,
-+ pte_t *ptep, int is_fullmm);
++/**************************************************************************
++ * Process a received packet
++ **************************************************************************/
+
-+static struct vm_operations_struct gntdev_vmops = {
-+ .close = gntdev_vma_close,
-+ .zap_pte = gntdev_clear_pte
-+};
++/*
++ * Returns whether or not we have a match in our forward table for the
++ * this skb. Must be called with appropriate fwd_lock already held
++ */
++static struct netback_accel *for_a_vnic(struct netback_pkt_buf *skb,
++ struct port_fwd *fwd_set)
++{
++ struct fwd_struct *fwd;
++ struct netback_accel *retval = NULL;
+
-+/* Global variables. */
++ fwd = fwd_find_entry(skb->mac.raw, fwd_set);
++ if (fwd != NULL)
++ retval = fwd->context;
++ return retval;
++}
+
-+/* The driver major number, for use when unregistering the driver. */
-+static int gntdev_major;
+
-+#define GNTDEV_NAME "gntdev"
++static inline int packet_is_arp_reply(struct sk_buff *skb)
++{
++ return skb->protocol == ntohs(ETH_P_ARP)
++ && skb->nh.arph->ar_op == ntohs(ARPOP_REPLY);
++}
+
-+/* Memory mapping functions
-+ * ------------------------
++
++static inline void hdr_to_filt(struct ethhdr *ethhdr, struct iphdr *ip,
++ struct netback_accel_filter_spec *spec)
++{
++ spec->proto = ip->protocol;
++ spec->destip_be = ip->daddr;
++ memcpy(spec->mac, ethhdr->h_source, ETH_ALEN);
++
++ if (ip->protocol == IPPROTO_TCP) {
++ struct tcphdr *tcp = (struct tcphdr *)((char *)ip + 4 * ip->ihl);
++ spec->destport_be = tcp->dest;
++ } else {
++ struct udphdr *udp = (struct udphdr *)((char *)ip + 4 * ip->ihl);
++ EPRINTK_ON(ip->protocol != IPPROTO_UDP);
++ spec->destport_be = udp->dest;
++ }
++}
++
++
++static inline int netback_accel_can_filter(struct netback_pkt_buf *skb)
++{
++ return (skb->protocol == htons(ETH_P_IP) &&
++ ((skb->nh.iph->protocol == IPPROTO_TCP) ||
++ (skb->nh.iph->protocol == IPPROTO_UDP)));
++}
++
++
++static inline void netback_accel_filter_packet(struct netback_accel *bend,
++ struct netback_pkt_buf *skb)
++{
++ struct netback_accel_filter_spec fs;
++ struct ethhdr *eh = (struct ethhdr *)(skb->mac.raw);
++
++ hdr_to_filt(eh, skb->nh.iph, &fs);
++
++ netback_accel_filter_check_add(bend, &fs);
++}
++
++
++/*
++ * Receive a packet and do something appropriate with it. Return true
++ * to take exclusive ownership of the packet. This is verging on
++ * solarflare specific
++ */
++void netback_accel_rx_packet(struct netback_pkt_buf *skb, void *fwd_priv)
++{
++ struct netback_accel *bend;
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
++ unsigned long flags;
++
++ BUG_ON(fwd_priv == NULL);
++
++ /* Checking for bcast is cheaper so do that first */
++ if (is_broadcast_ether_addr(skb->mac.raw)) {
++ /* pass through the slow path by not claiming ownership */
++ return;
++ } else if (is_multicast_ether_addr(skb->mac.raw)) {
++ /* pass through the slow path by not claiming ownership */
++ return;
++ } else {
++ /* It is unicast */
++ spin_lock_irqsave(&fwd_set->fwd_lock, flags);
++ /* We insert filter to pass it off to a VNIC */
++ if ((bend = for_a_vnic(skb, fwd_set)) != NULL)
++ if (netback_accel_can_filter(skb))
++ netback_accel_filter_packet(bend, skb);
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++ }
++ return;
++}
++
++
++void netback_accel_tx_packet(struct sk_buff *skb, void *fwd_priv)
++{
++ __u8 *mac;
++ unsigned long flags;
++ struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
++ struct fwd_struct *fwd;
++
++ BUG_ON(fwd_priv == NULL);
++
++ if (is_broadcast_ether_addr(skb->mac.raw) && packet_is_arp_reply(skb)) {
++ /*
++ * update our fast path forwarding to reflect this
++ * gratuitous ARP
++ */
++ mac = skb->mac.raw+ETH_ALEN;
++
++ DPRINTK("%s: found gratuitous ARP for " MAC_FMT "\n",
++ __FUNCTION__, MAC_ARG(mac));
++
++ spin_lock_irqsave(&fwd_set->fwd_lock, flags);
++ /*
++ * Might not be local, but let's tell them all it is,
++ * and they can restore the fastpath if they continue
++ * to get packets that way
++ */
++ list_for_each_entry(fwd, &fwd_set->fwd_list, link) {
++ struct netback_accel *bend = fwd->context;
++ if (bend != NULL)
++ netback_accel_msg_tx_new_localmac(bend, mac);
++ }
++
++ spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
++ }
++ return;
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_msg.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_msg.c 2008-05-19 00:33:46.602832361 +0300
+@@ -0,0 +1,392 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
+ *
-+ * Every granted page is mapped into both kernel and user space, and the two
-+ * following functions return the respective virtual addresses of these pages.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * When shadow paging is disabled, the granted page is mapped directly into
-+ * user space; when it is enabled, it is mapped into the kernel and remapped
-+ * into user space using vm_insert_page() (see gntdev_mmap(), below).
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+/* Returns the virtual address (in user space) of the @page_index'th page
-+ * in the given VM area.
-+ */
-+static inline unsigned long get_user_vaddr (struct vm_area_struct *vma,
-+ int page_index)
++#include <xen/evtchn.h>
++
++#include "accel.h"
++#include "accel_msg_iface.h"
++#include "accel_util.h"
++#include "accel_solarflare.h"
++
++/* Send a HELLO to front end to start things off */
++void netback_accel_msg_tx_hello(struct netback_accel *bend, unsigned version)
+{
-+ return (unsigned long) vma->vm_start + (page_index << PAGE_SHIFT);
++ unsigned long lock_state;
++ struct net_accel_msg *msg =
++ net_accel_msg_start_send(bend->shared_page,
++ &bend->to_domU, &lock_state);
++ /* The queue _cannot_ be full, we're the first users. */
++ EPRINTK_ON(msg == NULL);
++
++ if (msg != NULL) {
++ net_accel_msg_init(msg, NET_ACCEL_MSG_HELLO);
++ msg->u.hello.version = version;
++ msg->u.hello.max_pages = bend->quotas.max_buf_pages;
++ VPRINTK("Sending hello to channel %d\n", bend->msg_channel);
++ net_accel_msg_complete_send_notify(bend->shared_page,
++ &bend->to_domU,
++ &lock_state,
++ bend->msg_channel_irq);
++ }
+}
+
-+/* Returns the virtual address (in kernel space) of the @slot_index'th page
-+ * mapped by the gntdev instance that owns the given private data struct.
++/* Send a local mac message to vnic */
++static void netback_accel_msg_tx_localmac(struct netback_accel *bend,
++ int type, const void *mac)
++{
++ unsigned long lock_state;
++ struct net_accel_msg *msg;
++
++ BUG_ON(bend == NULL || mac == NULL);
++
++ VPRINTK("Sending local mac message: " MAC_FMT "\n",
++ MAC_ARG((const char *)mac));
++
++ msg = net_accel_msg_start_send(bend->shared_page, &bend->to_domU,
++ &lock_state);
++
++ if (msg != NULL) {
++ net_accel_msg_init(msg, NET_ACCEL_MSG_LOCALMAC);
++ msg->u.localmac.flags = type;
++ memcpy(msg->u.localmac.mac, mac, ETH_ALEN);
++ net_accel_msg_complete_send_notify(bend->shared_page,
++ &bend->to_domU,
++ &lock_state,
++ bend->msg_channel_irq);
++ } else {
++ /*
++ * TODO if this happens we may leave a domU
++ * fastpathing packets when they should be delivered
++ * locally. Solution is get domU to timeout entries
++ * in its fastpath lookup table when it receives no RX
++ * traffic
++ */
++ EPRINTK("%s: saw full queue, may need ARP timer to recover\n",
++ __FUNCTION__);
++ }
++}
++
++/* Send an add local mac message to vnic */
++void netback_accel_msg_tx_new_localmac(struct netback_accel *bend,
++ const void *mac)
++{
++ netback_accel_msg_tx_localmac(bend, NET_ACCEL_MSG_ADD, mac);
++}
++
++
++static int netback_accel_msg_rx_buffer_map(struct netback_accel *bend,
++ struct net_accel_msg *msg)
++{
++ int log2_pages, rc;
++
++ /* Can only allocate in power of two */
++ log2_pages = log2_ge(msg->u.mapbufs.pages, 0);
++ if (msg->u.mapbufs.pages != pow2(log2_pages)) {
++ EPRINTK("%s: Can only alloc bufs in power of 2 sizes (%d)\n",
++ __FUNCTION__, msg->u.mapbufs.pages);
++ rc = -EINVAL;
++ goto err_out;
++ }
++
++ /*
++ * Sanity. Assumes NET_ACCEL_MSG_MAX_PAGE_REQ is same for
++ * both directions/domains
++ */
++ if (msg->u.mapbufs.pages > NET_ACCEL_MSG_MAX_PAGE_REQ) {
++ EPRINTK("%s: too many pages in a single message: %d %d\n",
++ __FUNCTION__, msg->u.mapbufs.pages,
++ NET_ACCEL_MSG_MAX_PAGE_REQ);
++ rc = -EINVAL;
++ goto err_out;
++ }
++
++ if ((rc = netback_accel_add_buffers(bend, msg->u.mapbufs.pages,
++ log2_pages, msg->u.mapbufs.grants,
++ &msg->u.mapbufs.buf)) < 0) {
++ goto err_out;
++ }
++
++ msg->id |= NET_ACCEL_MSG_REPLY;
++
++ return 0;
++
++ err_out:
++ EPRINTK("%s: err_out\n", __FUNCTION__);
++ msg->id |= NET_ACCEL_MSG_ERROR | NET_ACCEL_MSG_REPLY;
++ return rc;
++}
++
++
++/* Hint from frontend that one of our filters is out of date */
++static int netback_accel_process_fastpath(struct netback_accel *bend,
++ struct net_accel_msg *msg)
++{
++ struct netback_accel_filter_spec spec;
++
++ if (msg->u.fastpath.flags & NET_ACCEL_MSG_REMOVE) {
++ /*
++ * Would be nice to BUG() this but would leave us
++ * vulnerable to naughty frontend
++ */
++ EPRINTK_ON(msg->u.fastpath.flags & NET_ACCEL_MSG_ADD);
++
++ memcpy(spec.mac, msg->u.fastpath.mac, ETH_ALEN);
++ spec.destport_be = msg->u.fastpath.port;
++ spec.destip_be = msg->u.fastpath.ip;
++ spec.proto = msg->u.fastpath.proto;
++
++ netback_accel_filter_remove_spec(bend, &spec);
++ }
++
++ return 0;
++}
++
++
++/* Flow control for message queues */
++inline void set_queue_not_full(struct netback_accel *bend)
++{
++ if (!test_and_set_bit(NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL_B,
++ (unsigned long *)&bend->shared_page->aflags))
++ notify_remote_via_irq(bend->msg_channel_irq);
++ else
++ VPRINTK("queue not full bit already set, not signalling\n");
++}
++
++
++/* Flow control for message queues */
++inline void set_queue_full(struct netback_accel *bend)
++{
++ if (!test_and_set_bit(NET_ACCEL_MSG_AFLAGS_QUEUE0FULL_B,
++ (unsigned long *)&bend->shared_page->aflags))
++ notify_remote_via_irq(bend->msg_channel_irq);
++ else
++ VPRINTK("queue full bit already set, not signalling\n");
++}
++
++
++void netback_accel_set_interface_state(struct netback_accel *bend, int up)
++{
++ bend->shared_page->net_dev_up = up;
++ if (!test_and_set_bit(NET_ACCEL_MSG_AFLAGS_NETUPDOWN_B,
++ (unsigned long *)&bend->shared_page->aflags))
++ notify_remote_via_irq(bend->msg_channel_irq);
++ else
++ VPRINTK("interface up/down bit already set, not signalling\n");
++}
++
++
++static int check_rx_hello_version(unsigned version)
++{
++ /* Should only happen if there's been a version mismatch */
++ BUG_ON(version == NET_ACCEL_MSG_VERSION);
++
++ if (version > NET_ACCEL_MSG_VERSION) {
++ /* Newer protocol, we must refuse */
++ return -EPROTO;
++ }
++
++ if (version < NET_ACCEL_MSG_VERSION) {
++ /*
++ * We are newer, so have discretion to accept if we
++ * wish. For now however, just reject
++ */
++ return -EPROTO;
++ }
++
++ return -EINVAL;
++}
++
++
++static int process_rx_msg(struct netback_accel *bend,
++ struct net_accel_msg *msg)
++{
++ int err = 0;
++
++ switch (msg->id) {
++ case NET_ACCEL_MSG_REPLY | NET_ACCEL_MSG_HELLO:
++ /* Reply to a HELLO; mark ourselves as connected */
++ DPRINTK("got Hello reply, version %.8x\n",
++ msg->u.hello.version);
++
++ /*
++ * Check that we've not successfully done this
++ * already. NB no check at the moment that this reply
++ * comes after we've actually sent a HELLO as that's
++ * not possible with the current code structure
++ */
++ if (bend->hw_state != NETBACK_ACCEL_RES_NONE)
++ return -EPROTO;
++
++ /* Store max_pages for accel_setup */
++ if (msg->u.hello.max_pages > bend->quotas.max_buf_pages) {
++ EPRINTK("More pages than quota allows (%d > %d)\n",
++ msg->u.hello.max_pages,
++ bend->quotas.max_buf_pages);
++ /* Force it down to the quota */
++ msg->u.hello.max_pages = bend->quotas.max_buf_pages;
++ }
++ bend->max_pages = msg->u.hello.max_pages;
++
++ /* Set up the hardware visible to the other end */
++ err = bend->accel_setup(bend);
++ if (err) {
++ /* This is fatal */
++ DPRINTK("Hello gave accel_setup error %d\n", err);
++ netback_accel_set_closing(bend);
++ } else {
++ /*
++ * Now add the context so that packet
++ * forwarding will commence
++ */
++ netback_accel_fwd_set_context(bend->mac, bend,
++ bend->fwd_priv);
++ }
++ break;
++ case NET_ACCEL_MSG_REPLY | NET_ACCEL_MSG_HELLO | NET_ACCEL_MSG_ERROR:
++ EPRINTK("got Hello error, versions us:%.8x them:%.8x\n",
++ NET_ACCEL_MSG_VERSION, msg->u.hello.version);
++
++ if (bend->hw_state != NETBACK_ACCEL_RES_NONE)
++ return -EPROTO;
++
++ if (msg->u.hello.version != NET_ACCEL_MSG_VERSION) {
++ /* Error is due to version mismatch */
++ err = check_rx_hello_version(msg->u.hello.version);
++ if (err == 0) {
++ /*
++ * It's OK to be compatible, send
++ * another hello with compatible version
++ */
++ netback_accel_msg_tx_hello
++ (bend, msg->u.hello.version);
++ } else {
++ /*
++ * Tell frontend that we're not going to
++ * send another HELLO by going to Closing.
++ */
++ netback_accel_set_closing(bend);
++ }
++ }
++ break;
++ case NET_ACCEL_MSG_MAPBUF:
++ VPRINTK("Got mapped buffers request %d\n",
++ msg->u.mapbufs.reqid);
++
++ if (bend->hw_state == NETBACK_ACCEL_RES_NONE)
++ return -EPROTO;
++
++ /*
++ * Frontend wants a buffer table entry for the
++ * supplied pages
++ */
++ err = netback_accel_msg_rx_buffer_map(bend, msg);
++ if (net_accel_msg_reply_notify(bend->shared_page,
++ bend->msg_channel_irq,
++ &bend->to_domU, msg)) {
++ /*
++ * This is fatal as we can't tell the frontend
++ * about the problem through the message
++ * queue, and so would otherwise stalemate
++ */
++ netback_accel_set_closing(bend);
++ }
++ break;
++ case NET_ACCEL_MSG_FASTPATH:
++ DPRINTK("Got fastpath request\n");
++
++ if (bend->hw_state == NETBACK_ACCEL_RES_NONE)
++ return -EPROTO;
++
++ err = netback_accel_process_fastpath(bend, msg);
++ break;
++ default:
++ EPRINTK("Huh? Message code is %x\n", msg->id);
++ err = -EPROTO;
++ break;
++ }
++ return err;
++}
++
++
++/* Demultiplex an IRQ from the frontend driver. */
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++void netback_accel_msg_rx_handler(struct work_struct *arg)
++#else
++void netback_accel_msg_rx_handler(void *bend_void)
++#endif
++{
++ struct net_accel_msg msg;
++ int err, queue_was_full = 0;
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ struct netback_accel *bend =
++ container_of(arg, struct netback_accel, handle_msg);
++#else
++ struct netback_accel *bend = (struct netback_accel *)bend_void;
++#endif
++
++ mutex_lock(&bend->bend_mutex);
++
++ /*
++ * This happens when the shared pages have been unmapped, but
++ * the workqueue not flushed yet
++ */
++ if (bend->shared_page == NULL)
++ goto done;
++
++ if ((bend->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_TO_DOM0_MASK) != 0) {
++ if (bend->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL) {
++ /* We've been told there may now be space. */
++ clear_bit(NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL_B,
++ (unsigned long *)&bend->shared_page->aflags);
++ }
++
++ if (bend->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_QUEUEUFULL) {
++ clear_bit(NET_ACCEL_MSG_AFLAGS_QUEUEUFULL_B,
++ (unsigned long *)&bend->shared_page->aflags);
++ queue_was_full = 1;
++ }
++ }
++
++ while ((err = net_accel_msg_recv(bend->shared_page, &bend->from_domU,
++ &msg)) == 0) {
++ err = process_rx_msg(bend, &msg);
++
++ if (err != 0) {
++ EPRINTK("%s: Error %d\n", __FUNCTION__, err);
++ goto err;
++ }
++ }
++
++ err:
++ /* There will be space now if we can make any. */
++ if (queue_was_full)
++ set_queue_not_full(bend);
++ done:
++ mutex_unlock(&bend->bend_mutex);
++
++ return;
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_solarflare.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_solarflare.c 2008-05-19 00:33:46.714838817 +0300
+@@ -0,0 +1,1253 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static inline unsigned long get_kernel_vaddr (gntdev_file_private_data_t *priv,
-+ int slot_index)
++
++#include "common.h"
++
++#include "accel.h"
++#include "accel_solarflare.h"
++#include "accel_msg_iface.h"
++#include "accel_util.h"
++
++#include "accel_cuckoo_hash.h"
++
++#include "ci/driver/resource/efx_vi.h"
++
++#include "ci/efrm/nic_table.h"
++#include "ci/efhw/public.h"
++
++#include <xen/evtchn.h>
++#include <xen/driver_util.h>
++#include <linux/list.h>
++#include <linux/mutex.h>
++
++#include "driverlink_api.h"
++
++#define SF_XEN_RX_USR_BUF_SIZE 2048
++
++struct falcon_bend_accel_priv {
++ struct efx_vi_state *efx_vih;
++
++ /*! Array of pointers to dma_map state, used so VNIC can
++ * request their removal in a single message
++ */
++ struct efx_vi_dma_map_state **dma_maps;
++ /*! Index into dma_maps */
++ int dma_maps_index;
++
++ /*! Serialises access to filters */
++ spinlock_t filter_lock;
++ /*! Bitmap of which filters are free */
++ unsigned long free_filters;
++ /*! Used for index normalisation */
++ u32 filter_idx_mask;
++ struct netback_accel_filter_spec *fspecs;
++ cuckoo_hash_table filter_hash_table;
++
++ u32 txdmaq_gnt;
++ u32 rxdmaq_gnt;
++ u32 doorbell_gnt;
++ u32 evq_rptr_gnt;
++ u32 evq_mem_gnts[EF_HW_FALCON_EVQ_PAGES];
++ u32 evq_npages;
++};
++
++/* Forward declaration */
++static int netback_accel_filter_init(struct netback_accel *);
++static void netback_accel_filter_shutdown(struct netback_accel *);
++
++/**************************************************************************
++ *
++ * Driverlink stuff
++ *
++ **************************************************************************/
++
++struct driverlink_port {
++ struct list_head link;
++ enum net_accel_hw_type type;
++ struct net_device *net_dev;
++ struct efx_dl_device *efx_dl_dev;
++ int nic_index;
++ void *fwd_priv;
++};
++
++static struct list_head dl_ports;
++
++/* This mutex protects global state, such as the dl_ports list */
++DEFINE_MUTEX(accel_mutex);
++
++static int init_done = 0;
++
++/* The DL callbacks */
++
++
++#if defined(EFX_USE_FASTCALL)
++static enum efx_veto fastcall
++#else
++static enum efx_veto
++#endif
++bend_dl_tx_packet(struct efx_dl_device *efx_dl_dev,
++ struct sk_buff *skb)
+{
-+ unsigned long pfn;
-+ void *kaddr;
-+ pfn = page_to_pfn(priv->foreign_pages[slot_index]);
-+ kaddr = pfn_to_kaddr(pfn);
-+ return (unsigned long) kaddr;
++ struct driverlink_port *port = efx_dl_dev->priv;
++
++ BUG_ON(port == NULL);
++
++ NETBACK_ACCEL_STATS_OP(global_stats.dl_tx_packets++);
++ if (skb->mac.raw != NULL)
++ netback_accel_tx_packet(skb, port->fwd_priv);
++ else {
++ DPRINTK("Ignoring packet with missing mac address\n");
++ NETBACK_ACCEL_STATS_OP(global_stats.dl_tx_bad_packets++);
++ }
++ return EFX_ALLOW_PACKET;
+}
+
-+/* Helper functions. */
++/* EFX_USE_FASTCALL */
++#if defined(EFX_USE_FASTCALL)
++static enum efx_veto fastcall
++#else
++static enum efx_veto
++#endif
++bend_dl_rx_packet(struct efx_dl_device *efx_dl_dev,
++ const char *pkt_buf, int pkt_len)
++{
++ struct driverlink_port *port = efx_dl_dev->priv;
++ struct netback_pkt_buf pkt;
++ struct ethhdr *eh;
+
-+/* Adds information about a grant reference to the list of grants in the file's
-+ * private data structure. Returns non-zero on failure. On success, sets the
-+ * value of *offset to the offset that should be mmap()-ed in order to map the
-+ * grant reference.
++ BUG_ON(port == NULL);
++
++ pkt.mac.raw = (char *)pkt_buf;
++ pkt.nh.raw = (char *)pkt_buf + ETH_HLEN;
++ eh = (struct ethhdr *)pkt_buf;
++ pkt.protocol = eh->h_proto;
++
++ NETBACK_ACCEL_STATS_OP(global_stats.dl_rx_packets++);
++ netback_accel_rx_packet(&pkt, port->fwd_priv);
++ return EFX_ALLOW_PACKET;
++}
++
++
++/* Callbacks we'd like to get from the netdriver through driverlink */
++struct efx_dl_callbacks bend_dl_callbacks =
++ {
++ .tx_packet = bend_dl_tx_packet,
++ .rx_packet = bend_dl_rx_packet,
++ };
++
++
++static struct netback_accel_hooks accel_hooks = {
++ THIS_MODULE,
++ &netback_accel_probe,
++ &netback_accel_remove
++};
++
++
++/*
++ * Handy helper which given an efx_dl_device works out which
++ * efab_nic_t index into efrm_nic_table.nics[] it corresponds to
+ */
-+static int add_grant_reference(struct file *flip,
-+ struct ioctl_gntdev_grant_ref *op,
-+ uint64_t *offset)
++static int efx_device_to_efab_nic_index(struct efx_dl_device *efx_dl_dev)
++{
++ int i;
++
++ for (i = 0; i < EFHW_MAX_NR_DEVS; i++) {
++ struct efhw_nic *nic = efrm_nic_table.nic[i];
++
++ /*
++ * It's possible for the nic structure to have not
++ * been initialised if the resource driver failed its
++ * driverlink probe
++ */
++ if (nic == NULL || nic->net_driver_dev == NULL)
++ continue;
++
++ /* Work out if these are talking about the same NIC */
++ if (nic->net_driver_dev->pci_dev == efx_dl_dev->pci_dev)
++ return i;
++ }
++
++ return -1;
++}
++
++
++/* Driver link probe - register our callbacks */
++static int bend_dl_probe(struct efx_dl_device *efx_dl_dev,
++ const struct net_device *net_dev,
++ const struct efx_dl_device_info *dev_info,
++ const char* silicon_rev)
++{
++ int rc;
++ enum net_accel_hw_type type;
++ struct driverlink_port *port;
++
++ DPRINTK("%s: %s\n", __FUNCTION__, silicon_rev);
++
++ if (strcmp(silicon_rev, "falcon/a1") == 0)
++ type = NET_ACCEL_MSG_HWTYPE_FALCON_A;
++ else if (strcmp(silicon_rev, "falcon/b0") == 0)
++ type = NET_ACCEL_MSG_HWTYPE_FALCON_B;
++ else {
++ EPRINTK("%s: unsupported silicon %s\n", __FUNCTION__,
++ silicon_rev);
++ rc = -EINVAL;
++ goto fail1;
++ }
++
++ port = kmalloc(sizeof(struct driverlink_port), GFP_KERNEL);
++ if (port == NULL) {
++ EPRINTK("%s: no memory for dl probe\n", __FUNCTION__);
++ rc = -ENOMEM;
++ goto fail1;
++ }
++
++ port->efx_dl_dev = efx_dl_dev;
++ efx_dl_dev->priv = port;
++
++ port->nic_index = efx_device_to_efab_nic_index(efx_dl_dev);
++ if (port->nic_index < 0) {
++ /*
++ * This can happen in theory if the resource driver
++ * failed to initialise properly
++ */
++ EPRINTK("%s: nic structure not found\n", __FUNCTION__);
++ rc = -EINVAL;
++ goto fail2;
++ }
++
++ port->fwd_priv = netback_accel_init_fwd_port();
++ if (port->fwd_priv == NULL) {
++ EPRINTK("%s: failed to set up forwarding for port\n",
++ __FUNCTION__);
++ rc = -ENOMEM;
++ goto fail2;
++ }
++
++ rc = efx_dl_register_callbacks(efx_dl_dev, &bend_dl_callbacks);
++ if (rc != 0) {
++ EPRINTK("%s: register_callbacks failed\n", __FUNCTION__);
++ goto fail3;
++ }
++
++ port->type = type;
++ port->net_dev = (struct net_device *)net_dev;
++
++ mutex_lock(&accel_mutex);
++ list_add(&port->link, &dl_ports);
++ mutex_unlock(&accel_mutex);
++
++ rc = netback_connect_accelerator(NETBACK_ACCEL_VERSION, 0,
++ port->net_dev->name, &accel_hooks);
++
++ if (rc < 0) {
++ EPRINTK("Xen netback accelerator version mismatch\n");
++ goto fail4;
++ } else if (rc > 0) {
++ /*
++ * In future may want to add backwards compatibility
++ * and accept certain subsets of previous versions
++ */
++ EPRINTK("Xen netback accelerator version mismatch\n");
++ goto fail4;
++ }
++
++ return 0;
++
++ fail4:
++ mutex_lock(&accel_mutex);
++ list_del(&port->link);
++ mutex_unlock(&accel_mutex);
++
++ efx_dl_unregister_callbacks(efx_dl_dev, &bend_dl_callbacks);
++ fail3:
++ netback_accel_shutdown_fwd_port(port->fwd_priv);
++ fail2:
++ efx_dl_dev->priv = NULL;
++ kfree(port);
++ fail1:
++ return rc;
++}
++
++
++static void bend_dl_remove(struct efx_dl_device *efx_dl_dev)
++{
++ struct driverlink_port *port;
++
++ DPRINTK("Unregistering driverlink callbacks.\n");
++
++ mutex_lock(&accel_mutex);
++
++ port = (struct driverlink_port *)efx_dl_dev->priv;
++
++ BUG_ON(list_empty(&dl_ports));
++ BUG_ON(port == NULL);
++ BUG_ON(port->efx_dl_dev != efx_dl_dev);
++
++ netback_disconnect_accelerator(0, port->net_dev->name);
++
++ list_del(&port->link);
++
++ mutex_unlock(&accel_mutex);
++
++ efx_dl_unregister_callbacks(efx_dl_dev, &bend_dl_callbacks);
++ netback_accel_shutdown_fwd_port(port->fwd_priv);
++
++ efx_dl_dev->priv = NULL;
++ kfree(port);
++
++ return;
++}
++
++
++static struct efx_dl_driver bend_dl_driver =
++ {
++ .name = "SFC Xen backend",
++ .probe = bend_dl_probe,
++ .remove = bend_dl_remove,
++ };
++
++
++int netback_accel_sf_init(void)
++{
++ int rc, nic_i;
++ struct efhw_nic *nic;
++
++ INIT_LIST_HEAD(&dl_ports);
++
++ rc = efx_dl_register_driver(&bend_dl_driver);
++ /* If we couldn't find the NET driver, give up */
++ if (rc == -ENOENT)
++ return rc;
++
++ if (rc == 0) {
++ EFRM_FOR_EACH_NIC(nic_i, nic)
++ falcon_nic_set_rx_usr_buf_size(nic,
++ SF_XEN_RX_USR_BUF_SIZE);
++ }
++
++ init_done = (rc == 0);
++ return rc;
++}
++
++
++void netback_accel_sf_shutdown(void)
++{
++ if (!init_done)
++ return;
++ DPRINTK("Unregistering driverlink driver\n");
++
++ /*
++ * This will trigger removal callbacks for all the devices, which
++ * will unregister their callbacks, disconnect from netfront, etc.
++ */
++ efx_dl_unregister_driver(&bend_dl_driver);
++}
++
++
++int netback_accel_sf_hwtype(struct netback_accel *bend)
++{
++ struct driverlink_port *port;
++
++ mutex_lock(&accel_mutex);
++
++ list_for_each_entry(port, &dl_ports, link) {
++ if (strcmp(bend->nicname, port->net_dev->name) == 0) {
++ bend->hw_type = port->type;
++ bend->accel_setup = netback_accel_setup_vnic_hw;
++ bend->accel_shutdown = netback_accel_shutdown_vnic_hw;
++ bend->fwd_priv = port->fwd_priv;
++ /* This is just needed to pass to efx_vi_alloc */
++ bend->nic_index = port->nic_index;
++ bend->net_dev = port->net_dev;
++ mutex_unlock(&accel_mutex);
++ return 0;
++ }
++ }
++
++ mutex_unlock(&accel_mutex);
++
++ EPRINTK("Failed to identify backend device '%s' with a NIC\n",
++ bend->nicname);
++
++ return -ENOENT;
++}
++
++
++/****************************************************************************
++ * Resource management code
++ ***************************************************************************/
++
++static int alloc_page_state(struct netback_accel *bend, int max_pages)
+{
-+ gntdev_file_private_data_t *private_data
-+ = (gntdev_file_private_data_t *) flip->private_data;
++ struct falcon_bend_accel_priv *accel_hw_priv;
+
-+ uint32_t slot_index;
++ if (max_pages < 0 || max_pages > bend->quotas.max_buf_pages) {
++ EPRINTK("%s: invalid max_pages: %d\n", __FUNCTION__, max_pages);
++ return -EINVAL;
++ }
+
-+ if (unlikely(private_data->free_list_size == 0)) {
++ accel_hw_priv = kzalloc(sizeof(struct falcon_bend_accel_priv),
++ GFP_KERNEL);
++ if (accel_hw_priv == NULL) {
++ EPRINTK("%s: no memory for accel_hw_priv\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
-+ slot_index = private_data->free_list[--private_data->free_list_size];
++ accel_hw_priv->dma_maps = kzalloc
++ (sizeof(struct efx_vi_dma_map_state **) *
++ (max_pages / NET_ACCEL_MSG_MAX_PAGE_REQ), GFP_KERNEL);
++ if (accel_hw_priv->dma_maps == NULL) {
++ EPRINTK("%s: no memory for dma_maps\n", __FUNCTION__);
++ kfree(accel_hw_priv);
++ return -ENOMEM;
++ }
+
-+ /* Copy the grant information into file's private data. */
-+ private_data->grants[slot_index].state = GNTDEV_SLOT_NOT_YET_MAPPED;
-+ private_data->grants[slot_index].u.valid.domid = op->domid;
-+ private_data->grants[slot_index].u.valid.ref = op->ref;
++ bend->buffer_maps = kzalloc(sizeof(struct vm_struct *) * max_pages,
++ GFP_KERNEL);
++ if (bend->buffer_maps == NULL) {
++ EPRINTK("%s: no memory for buffer_maps\n", __FUNCTION__);
++ kfree(accel_hw_priv->dma_maps);
++ kfree(accel_hw_priv);
++ return -ENOMEM;
++ }
+
-+ /* The offset is calculated as the index of the chosen entry in the
-+ * file's private data's array of grant information. This is then
-+ * shifted to give an offset into the virtual "file address space".
-+ */
-+ *offset = slot_index << PAGE_SHIFT;
++ bend->buffer_addrs = kzalloc(sizeof(u64) * max_pages, GFP_KERNEL);
++ if (bend->buffer_addrs == NULL) {
++ kfree(bend->buffer_maps);
++ kfree(accel_hw_priv->dma_maps);
++ kfree(accel_hw_priv);
++ return -ENOMEM;
++ }
++
++ bend->accel_hw_priv = accel_hw_priv;
+
+ return 0;
+}
+
-+/* Adds the @count grant references to the contiguous range in the slot array
-+ * beginning at @first_slot. It is assumed that @first_slot was returned by a
-+ * previous invocation of find_contiguous_free_range(), during the same
-+ * invocation of the driver.
-+ */
-+static int add_grant_references(struct file *flip,
-+ int count,
-+ struct ioctl_gntdev_grant_ref *ops,
-+ uint32_t first_slot)
++
++static int free_page_state(struct netback_accel *bend)
+{
-+ gntdev_file_private_data_t *private_data
-+ = (gntdev_file_private_data_t *) flip->private_data;
-+ int i;
-+
-+ for (i = 0; i < count; ++i) {
++ struct falcon_bend_accel_priv *accel_hw_priv;
+
-+ /* First, mark the slot's entry in the free list as invalid. */
-+ int free_list_index =
-+ private_data->grants[first_slot+i].u.free_list_index;
-+ private_data->free_list[free_list_index] =
-+ GNTDEV_FREE_LIST_INVALID;
++ DPRINTK("%s: %p\n", __FUNCTION__, bend);
+
-+ /* Now, update the slot. */
-+ private_data->grants[first_slot+i].state =
-+ GNTDEV_SLOT_NOT_YET_MAPPED;
-+ private_data->grants[first_slot+i].u.valid.domid =
-+ ops[i].domid;
-+ private_data->grants[first_slot+i].u.valid.ref = ops[i].ref;
++ accel_hw_priv = bend->accel_hw_priv;
++
++ if (accel_hw_priv) {
++ kfree(accel_hw_priv->dma_maps);
++ kfree(bend->buffer_maps);
++ kfree(bend->buffer_addrs);
++ kfree(accel_hw_priv);
++ bend->accel_hw_priv = NULL;
++ bend->max_pages = 0;
+ }
+
-+ return 0;
++ return 0;
+}
+
-+/* Scans through the free list for @flip, removing entries that are marked as
-+ * GNTDEV_SLOT_INVALID. This will reduce the recorded size of the free list to
-+ * the number of valid entries.
-+ */
-+static void compress_free_list(struct file *flip)
++
++/* The timeout event callback for the event q */
++static void bend_evq_timeout(void *context, int is_timeout)
+{
-+ gntdev_file_private_data_t *private_data
-+ = (gntdev_file_private_data_t *) flip->private_data;
-+ int i, j = 0, old_size;
-+
-+ old_size = private_data->free_list_size;
-+ for (i = 0; i < old_size; ++i) {
-+ if (private_data->free_list[i] != GNTDEV_FREE_LIST_INVALID) {
-+ private_data->free_list[j] =
-+ private_data->free_list[i];
-+ ++j;
-+ } else {
-+ --private_data->free_list_size;
-+ }
++ struct netback_accel *bend = (struct netback_accel *)context;
++ if (is_timeout) {
++ /* Pass event to vnic front end driver */
++ VPRINTK("timeout event to %d\n", bend->net_channel);
++ NETBACK_ACCEL_STATS_OP(bend->stats.evq_timeouts++);
++ notify_remote_via_irq(bend->net_channel_irq);
++ } else {
++ /* It's a wakeup event, used by Falcon */
++ VPRINTK("wakeup to %d\n", bend->net_channel);
++ NETBACK_ACCEL_STATS_OP(bend->stats.evq_wakeups++);
++ notify_remote_via_irq(bend->net_channel_irq);
+ }
+}
+
-+/* Searches the grant array in the private data of @flip for a range of
-+ * @num_slots contiguous slots in the GNTDEV_SLOT_INVALID state.
-+ *
-+ * Returns the index of the first slot if a range is found, otherwise -ENOMEM.
++
++/*
++ * Create the eventq and associated gubbins for communication with the
++ * front end vnic driver
+ */
-+static int find_contiguous_free_range(struct file *flip,
-+ uint32_t num_slots)
++static int ef_get_vnic(struct netback_accel *bend)
+{
-+ gntdev_file_private_data_t *private_data
-+ = (gntdev_file_private_data_t *) flip->private_data;
-+
-+ int i;
-+ int start_index = private_data->next_fit_index;
-+ int range_start = 0, range_length;
++ struct falcon_bend_accel_priv *accel_hw_priv;
++ int rc = 0;
+
-+ if (private_data->free_list_size < num_slots) {
-+ return -ENOMEM;
++ BUG_ON(bend->hw_state != NETBACK_ACCEL_RES_NONE);
++
++ /* Allocate page related state and accel_hw_priv */
++ rc = alloc_page_state(bend, bend->max_pages);
++ if (rc != 0) {
++ EPRINTK("Failed to allocate page state: %d\n", rc);
++ return rc;
+ }
+
-+ /* First search from the start_index to the end of the array. */
-+ range_length = 0;
-+ for (i = start_index; i < MAX_GRANTS; ++i) {
-+ if (private_data->grants[i].state == GNTDEV_SLOT_INVALID) {
-+ if (range_length == 0) {
-+ range_start = i;
-+ }
-+ ++range_length;
-+ if (range_length == num_slots) {
-+ return range_start;
-+ }
-+ }
++ accel_hw_priv = bend->accel_hw_priv;
++
++ rc = efx_vi_alloc(&accel_hw_priv->efx_vih, bend->nic_index);
++ if (rc != 0) {
++ EPRINTK("%s: efx_vi_alloc failed %d\n", __FUNCTION__, rc);
++ free_page_state(bend);
++ return rc;
+ }
-+
-+ /* Now search from the start of the array to the start_index. */
-+ range_length = 0;
-+ for (i = 0; i < start_index; ++i) {
-+ if (private_data->grants[i].state == GNTDEV_SLOT_INVALID) {
-+ if (range_length == 0) {
-+ range_start = i;
-+ }
-+ ++range_length;
-+ if (range_length == num_slots) {
-+ return range_start;
-+ }
-+ }
++
++ rc = efx_vi_eventq_register_callback(accel_hw_priv->efx_vih,
++ bend_evq_timeout,
++ bend);
++ if (rc != 0) {
++ EPRINTK("%s: register_callback failed %d\n", __FUNCTION__, rc);
++ efx_vi_free(accel_hw_priv->efx_vih);
++ free_page_state(bend);
++ return rc;
+ }
++
++ bend->hw_state = NETBACK_ACCEL_RES_ALLOC;
+
-+ return -ENOMEM;
++ return 0;
+}
+
-+/* Interface functions. */
+
-+/* Initialises the driver. Called when the module is loaded. */
-+static int __init gntdev_init(void)
++static void ef_free_vnic(struct netback_accel *bend)
+{
-+ struct class *class;
-+ struct class_device *device;
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
+
-+ if (!is_running_on_xen()) {
-+ printk(KERN_ERR "You must be running Xen to use gntdev\n");
-+ return -ENODEV;
-+ }
++ BUG_ON(bend->hw_state != NETBACK_ACCEL_RES_ALLOC);
+
-+ gntdev_major = register_chrdev(0, GNTDEV_NAME, &gntdev_fops);
-+ if (gntdev_major < 0)
-+ {
-+ printk(KERN_ERR "Could not register gntdev device\n");
-+ return -ENOMEM;
-+ }
++ efx_vi_eventq_kill_callback(accel_hw_priv->efx_vih);
+
-+ /* Note that if the sysfs code fails, we will still initialise the
-+ * device, and output the major number so that the device can be
-+ * created manually using mknod.
-+ */
-+ if ((class = get_xen_class()) == NULL) {
-+ printk(KERN_ERR "Error setting up xen_class\n");
-+ printk(KERN_ERR "gntdev created with major number = %d\n",
-+ gntdev_major);
-+ return 0;
-+ }
++ DPRINTK("Hardware is freeable. Will proceed.\n");
+
-+ device = class_device_create(class, NULL, MKDEV(gntdev_major, 0),
-+ NULL, GNTDEV_NAME);
-+ if (IS_ERR(device)) {
-+ printk(KERN_ERR "Error creating gntdev device in xen_class\n");
-+ printk(KERN_ERR "gntdev created with major number = %d\n",
-+ gntdev_major);
-+ return 0;
-+ }
++ efx_vi_free(accel_hw_priv->efx_vih);
++ accel_hw_priv->efx_vih = NULL;
+
-+ return 0;
++ VPRINTK("Free page state...\n");
++ free_page_state(bend);
++
++ bend->hw_state = NETBACK_ACCEL_RES_NONE;
+}
+
-+/* Cleans up and unregisters the driver. Called when the driver is unloaded.
-+ */
-+static void __exit gntdev_exit(void)
-+{
-+ struct class *class;
-+ if ((class = get_xen_class()) != NULL)
-+ class_device_destroy(class, MKDEV(gntdev_major, 0));
-+ unregister_chrdev(gntdev_major, GNTDEV_NAME);
++
++static inline void ungrant_or_crash(grant_ref_t gntref, int domain) {
++ if (net_accel_ungrant_page(gntref) == -EBUSY)
++ net_accel_shutdown_remote(domain);
+}
+
-+/* Called when the device is opened. */
-+static int gntdev_open(struct inode *inode, struct file *flip)
++
++static void netback_accel_release_hwinfo(struct netback_accel *bend)
+{
-+ gntdev_file_private_data_t *private_data;
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
+ int i;
+
-+ try_module_get(THIS_MODULE);
-+
-+ /* Allocate space for the per-instance private data. */
-+ private_data = kmalloc(sizeof(*private_data), GFP_KERNEL);
-+ if (!private_data)
-+ goto nomem_out;
++ DPRINTK("Remove dma q grants %d %d\n", accel_hw_priv->txdmaq_gnt,
++ accel_hw_priv->rxdmaq_gnt);
++ ungrant_or_crash(accel_hw_priv->txdmaq_gnt, bend->far_end);
++ ungrant_or_crash(accel_hw_priv->rxdmaq_gnt, bend->far_end);
+
-+ /* Allocate space for the kernel-mapping of granted pages. */
-+ private_data->foreign_pages =
-+ alloc_empty_pages_and_pagevec(MAX_GRANTS);
-+ if (!private_data->foreign_pages)
-+ goto nomem_out2;
++ DPRINTK("Remove doorbell grant %d\n", accel_hw_priv->doorbell_gnt);
++ ungrant_or_crash(accel_hw_priv->doorbell_gnt, bend->far_end);
+
-+ /* Initialise the free-list, which contains all slots at first.
-+ */
-+ for (i = 0; i < MAX_GRANTS; ++i) {
-+ private_data->free_list[MAX_GRANTS - i - 1] = i;
-+ private_data->grants[i].state = GNTDEV_SLOT_INVALID;
-+ private_data->grants[i].u.free_list_index = MAX_GRANTS - i - 1;
++ if (bend->hw_type == NET_ACCEL_MSG_HWTYPE_FALCON_A) {
++ DPRINTK("Remove rptr grant %d\n", accel_hw_priv->evq_rptr_gnt);
++ ungrant_or_crash(accel_hw_priv->evq_rptr_gnt, bend->far_end);
+ }
-+ private_data->free_list_size = MAX_GRANTS;
-+ private_data->next_fit_index = 0;
+
-+ init_rwsem(&private_data->grants_sem);
-+ init_rwsem(&private_data->free_list_sem);
-+
-+ flip->private_data = private_data;
++ for (i = 0; i < accel_hw_priv->evq_npages; i++) {
++ DPRINTK("Remove evq grant %d\n", accel_hw_priv->evq_mem_gnts[i]);
++ ungrant_or_crash(accel_hw_priv->evq_mem_gnts[i], bend->far_end);
++ }
+
-+ return 0;
++ bend->hw_state = NETBACK_ACCEL_RES_FILTER;
+
-+nomem_out2:
-+ kfree(private_data);
-+nomem_out:
-+ return -ENOMEM;
++ return;
+}
+
-+/* Called when the device is closed.
-+ */
-+static int gntdev_release(struct inode *inode, struct file *flip)
-+{
-+ if (flip->private_data) {
-+ gntdev_file_private_data_t *private_data =
-+ (gntdev_file_private_data_t *) flip->private_data;
-+ if (private_data->foreign_pages) {
-+ free_empty_pages_and_pagevec
-+ (private_data->foreign_pages, MAX_GRANTS);
-+ }
-+ kfree(private_data);
-+ }
-+ module_put(THIS_MODULE);
-+ return 0;
-+}
+
-+/* Called when an attempt is made to mmap() the device. The private data from
-+ * @flip contains the list of grant references that can be mapped. The vm_pgoff
-+ * field of @vma contains the index into that list that refers to the grant
-+ * reference that will be mapped. Only mappings that are a multiple of
-+ * PAGE_SIZE are handled.
-+ */
-+static int gntdev_mmap (struct file *flip, struct vm_area_struct *vma)
++static int ef_bend_hwinfo_falcon_common(struct netback_accel *bend,
++ struct net_accel_hw_falcon_b *hwinfo)
+{
-+ struct gnttab_map_grant_ref op;
-+ unsigned long slot_index = vma->vm_pgoff;
-+ unsigned long kernel_vaddr, user_vaddr;
-+ uint32_t size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-+ uint64_t ptep;
-+ int ret;
-+ int flags;
-+ int i;
-+ struct page *page;
-+ gntdev_file_private_data_t *private_data = flip->private_data;
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ struct efx_vi_hw_resource_metadata res_mdata;
++ struct efx_vi_hw_resource res_array[EFX_VI_HW_RESOURCE_MAXSIZE];
++ int rc, len = EFX_VI_HW_RESOURCE_MAXSIZE, i, pfn = 0;
++ unsigned long txdmaq_pfn = 0, rxdmaq_pfn = 0;
+
-+ if (unlikely(!private_data)) {
-+ printk(KERN_ERR "File's private data is NULL.\n");
-+ return -EINVAL;
++ rc = efx_vi_hw_resource_get_phys(accel_hw_priv->efx_vih, &res_mdata,
++ res_array, &len);
++ if (rc != 0) {
++ DPRINTK("%s: resource_get_phys returned %d\n",
++ __FUNCTION__, rc);
++ return rc;
+ }
+
-+ if (unlikely((size <= 0) || (size + slot_index) > MAX_GRANTS)) {
-+ printk(KERN_ERR "Invalid number of pages or offset"
-+ "(num_pages = %d, first_slot = %ld).\n",
-+ size, slot_index);
-+ return -ENXIO;
-+ }
++ if (res_mdata.version != 0)
++ return -EPROTO;
+
-+ if ((vma->vm_flags & VM_WRITE) && !(vma->vm_flags & VM_SHARED)) {
-+ printk(KERN_ERR "Writable mappings must be shared.\n");
-+ return -EINVAL;
-+ }
++ hwinfo->nic_arch = res_mdata.nic_arch;
++ hwinfo->nic_variant = res_mdata.nic_variant;
++ hwinfo->nic_revision = res_mdata.nic_revision;
+
-+ /* Slots must be in the NOT_YET_MAPPED state. */
-+ down_write(&private_data->grants_sem);
-+ for (i = 0; i < size; ++i) {
-+ if (private_data->grants[slot_index + i].state !=
-+ GNTDEV_SLOT_NOT_YET_MAPPED) {
-+ printk(KERN_ERR "Slot (index = %ld) is in the wrong "
-+ "state (%d).\n", slot_index + i,
-+ private_data->grants[slot_index + i].state);
-+ up_write(&private_data->grants_sem);
-+ return -EINVAL;
++ hwinfo->evq_order = res_mdata.evq_order;
++ hwinfo->evq_offs = res_mdata.evq_offs;
++ hwinfo->evq_capacity = res_mdata.evq_capacity;
++ hwinfo->instance = res_mdata.instance;
++ hwinfo->rx_capacity = res_mdata.rx_capacity;
++ hwinfo->tx_capacity = res_mdata.tx_capacity;
++
++ VPRINTK("evq_order %d evq_offs %d evq_cap %d inst %d rx_cap %d tx_cap %d\n",
++ hwinfo->evq_order, hwinfo->evq_offs, hwinfo->evq_capacity,
++ hwinfo->instance, hwinfo->rx_capacity, hwinfo->tx_capacity);
++
++ for (i = 0; i < len; i++) {
++ struct efx_vi_hw_resource *res = &(res_array[i]);
++ switch (res->type) {
++ case EFX_VI_HW_RESOURCE_TXDMAQ:
++ txdmaq_pfn = page_to_pfn(virt_to_page(res->address));
++ break;
++ case EFX_VI_HW_RESOURCE_RXDMAQ:
++ rxdmaq_pfn = page_to_pfn(virt_to_page(res->address));
++ break;
++ case EFX_VI_HW_RESOURCE_EVQTIMER:
++ break;
++ case EFX_VI_HW_RESOURCE_EVQRPTR:
++ case EFX_VI_HW_RESOURCE_EVQRPTR_OFFSET:
++ hwinfo->evq_rptr = res->address;
++ break;
++ case EFX_VI_HW_RESOURCE_EVQMEMKVA:
++ accel_hw_priv->evq_npages = 1 << res_mdata.evq_order;
++ pfn = page_to_pfn(virt_to_page(res->address));
++ break;
++ case EFX_VI_HW_RESOURCE_BELLPAGE:
++ hwinfo->doorbell_mfn = res->address;
++ break;
++ default:
++ EPRINTK("%s: Unknown hardware resource type %d\n",
++ __FUNCTION__, res->type);
++ break;
+ }
+ }
+
-+ /* Install the hook for unmapping. */
-+ vma->vm_ops = &gntdev_vmops;
-+
-+ /* The VM area contains pages from another VM. */
-+ vma->vm_flags |= VM_FOREIGN;
-+ vma->vm_private_data = kzalloc(size * sizeof(struct page_struct *),
-+ GFP_KERNEL);
-+ if (vma->vm_private_data == NULL) {
-+ printk(KERN_ERR "Couldn't allocate mapping structure for VM "
-+ "area.\n");
-+ return -ENOMEM;
-+ }
++ VPRINTK("Passing txdmaq page pfn %lx\n", txdmaq_pfn);
++ accel_hw_priv->txdmaq_gnt = hwinfo->txdmaq_gnt =
++ net_accel_grant_page(bend->hdev_data, pfn_to_mfn(txdmaq_pfn),
++ 0);
+
-+ /* This flag prevents Bad PTE errors when the memory is unmapped. */
-+ vma->vm_flags |= VM_RESERVED;
++ VPRINTK("Passing rxdmaq page pfn %lx\n", rxdmaq_pfn);
++ accel_hw_priv->rxdmaq_gnt = hwinfo->rxdmaq_gnt =
++ net_accel_grant_page(bend->hdev_data, pfn_to_mfn(rxdmaq_pfn),
++ 0);
+
-+ /* This flag prevents this VM area being copied on a fork(). A better
-+ * behaviour might be to explicitly carry out the appropriate mappings
-+ * on fork(), but I don't know if there's a hook for this.
-+ */
-+ vma->vm_flags |= VM_DONTCOPY;
++ VPRINTK("Passing doorbell page mfn %x\n", hwinfo->doorbell_mfn);
++ /* Make the relevant H/W pages mappable by the far end */
++ accel_hw_priv->doorbell_gnt = hwinfo->doorbell_gnt =
++ net_accel_grant_page(bend->hdev_data, hwinfo->doorbell_mfn, 1);
++
++ /* Now do the same for the memory pages */
++ /* Convert the page + length we got back for the evq to grants. */
++ for (i = 0; i < accel_hw_priv->evq_npages; i++) {
++ accel_hw_priv->evq_mem_gnts[i] = hwinfo->evq_mem_gnts[i] =
++ net_accel_grant_page(bend->hdev_data, pfn_to_mfn(pfn), 0);
++ VPRINTK("Got grant %u for evq pfn %x\n", hwinfo->evq_mem_gnts[i],
++ pfn);
++ pfn++;
++ }
+
-+#ifdef CONFIG_X86
-+ /* This flag ensures that the page tables are not unpinned before the
-+ * VM area is unmapped. Therefore Xen still recognises the PTE as
-+ * belonging to an L1 pagetable, and the grant unmap operation will
-+ * succeed, even if the process does not exit cleanly.
-+ */
-+ vma->vm_mm->context.has_foreign_mappings = 1;
-+#endif
++ return 0;
++}
+
-+ for (i = 0; i < size; ++i) {
+
-+ flags = GNTMAP_host_map;
-+ if (!(vma->vm_flags & VM_WRITE))
-+ flags |= GNTMAP_readonly;
++static int ef_bend_hwinfo_falcon_a(struct netback_accel *bend,
++ struct net_accel_hw_falcon_a *hwinfo)
++{
++ int rc;
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
+
-+ kernel_vaddr = get_kernel_vaddr(private_data, slot_index + i);
-+ user_vaddr = get_user_vaddr(vma, i);
-+ page = pfn_to_page(__pa(kernel_vaddr) >> PAGE_SHIFT);
++ if ((rc = ef_bend_hwinfo_falcon_common(bend, &hwinfo->common)) != 0)
++ return rc;
+
-+ gnttab_set_map_op(&op, kernel_vaddr, flags,
-+ private_data->grants[slot_index+i]
-+ .u.valid.ref,
-+ private_data->grants[slot_index+i]
-+ .u.valid.domid);
++ /*
++ * Note that unlike the above, where the message field is the
++ * page number, here evq_rptr is the entire address because
++ * it is currently a pointer into the densely mapped timer page.
++ */
++ VPRINTK("Passing evq_rptr pfn %x for rptr %x\n",
++ hwinfo->common.evq_rptr >> PAGE_SHIFT,
++ hwinfo->common.evq_rptr);
++ rc = net_accel_grant_page(bend->hdev_data,
++ hwinfo->common.evq_rptr >> PAGE_SHIFT, 0);
++ if (rc < 0)
++ return rc;
+
-+ /* Carry out the mapping of the grant reference. */
-+ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
-+ &op, 1);
-+ BUG_ON(ret);
-+ if (op.status) {
-+ printk(KERN_ERR "Error mapping the grant reference "
-+ "into the kernel (%d). domid = %d; ref = %d\n",
-+ op.status,
-+ private_data->grants[slot_index+i]
-+ .u.valid.domid,
-+ private_data->grants[slot_index+i]
-+ .u.valid.ref);
-+ goto undo_map_out;
-+ }
++ accel_hw_priv->evq_rptr_gnt = hwinfo->evq_rptr_gnt = rc;
++ VPRINTK("evq_rptr_gnt got %d\n", hwinfo->evq_rptr_gnt);
++
++ return 0;
++}
+
-+ /* Store a reference to the page that will be mapped into user
-+ * space.
-+ */
-+ ((struct page **) vma->vm_private_data)[i] = page;
+
-+ /* Mark mapped page as reserved. */
-+ SetPageReserved(page);
++static int ef_bend_hwinfo_falcon_b(struct netback_accel *bend,
++ struct net_accel_hw_falcon_b *hwinfo)
++{
++ return ef_bend_hwinfo_falcon_common(bend, hwinfo);
++}
+
-+ /* Record the grant handle, for use in the unmap operation. */
-+ private_data->grants[slot_index+i].u.valid.kernel_handle =
-+ op.handle;
-+ private_data->grants[slot_index+i].u.valid.dev_bus_addr =
-+ op.dev_bus_addr;
-+
-+ private_data->grants[slot_index+i].state = GNTDEV_SLOT_MAPPED;
-+ private_data->grants[slot_index+i].u.valid.user_handle =
-+ GNTDEV_INVALID_HANDLE;
+
-+ /* Now perform the mapping to user space. */
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
++/*
++ * Fill in the message with a description of the hardware resources, based on
++ * the H/W type
++ */
++static int netback_accel_hwinfo(struct netback_accel *bend,
++ struct net_accel_msg_hw *msgvi)
++{
++ int rc = 0;
++
++ BUG_ON(bend->hw_state != NETBACK_ACCEL_RES_FILTER);
+
-+ /* NOT USING SHADOW PAGE TABLES. */
-+ /* In this case, we map the grant(s) straight into user
-+ * space.
-+ */
++ msgvi->type = bend->hw_type;
++ switch (bend->hw_type) {
++ case NET_ACCEL_MSG_HWTYPE_FALCON_A:
++ rc = ef_bend_hwinfo_falcon_a(bend, &msgvi->resources.falcon_a);
++ break;
++ case NET_ACCEL_MSG_HWTYPE_FALCON_B:
++ rc = ef_bend_hwinfo_falcon_b(bend, &msgvi->resources.falcon_b);
++ break;
++ case NET_ACCEL_MSG_HWTYPE_NONE:
++ /* Nothing to do. The slow path should just work. */
++ break;
++ }
+
-+ /* Get the machine address of the PTE for the user
-+ * page.
-+ */
-+ if ((ret = create_lookup_pte_addr(vma->vm_mm,
-+ vma->vm_start
-+ + (i << PAGE_SHIFT),
-+ &ptep)))
-+ {
-+ printk(KERN_ERR "Error obtaining PTE pointer "
-+ "(%d).\n", ret);
-+ goto undo_map_out;
-+ }
-+
-+ /* Configure the map operation. */
++ if (rc == 0)
++ bend->hw_state = NETBACK_ACCEL_RES_HWINFO;
+
-+ /* The reference is to be used by host CPUs. */
-+ flags = GNTMAP_host_map;
-+
-+ /* Specifies a user space mapping. */
-+ flags |= GNTMAP_application_map;
-+
-+ /* The map request contains the machine address of the
-+ * PTE to update.
-+ */
-+ flags |= GNTMAP_contains_pte;
-+
-+ if (!(vma->vm_flags & VM_WRITE))
-+ flags |= GNTMAP_readonly;
++ return rc;
++}
+
-+ gnttab_set_map_op(&op, ptep, flags,
-+ private_data->grants[slot_index+i]
-+ .u.valid.ref,
-+ private_data->grants[slot_index+i]
-+ .u.valid.domid);
+
-+ /* Carry out the mapping of the grant reference. */
-+ ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
-+ &op, 1);
-+ BUG_ON(ret);
-+ if (op.status) {
-+ printk(KERN_ERR "Error mapping the grant "
-+ "reference into user space (%d). domid "
-+ "= %d; ref = %d\n", op.status,
-+ private_data->grants[slot_index+i].u
-+ .valid.domid,
-+ private_data->grants[slot_index+i].u
-+ .valid.ref);
-+ goto undo_map_out;
-+ }
-+
-+ /* Record the grant handle, for use in the unmap
-+ * operation.
-+ */
-+ private_data->grants[slot_index+i].u.
-+ valid.user_handle = op.handle;
++/* Allocate hardware resources and make them available to the client domain */
++int netback_accel_setup_vnic_hw(struct netback_accel *bend)
++{
++ struct net_accel_msg msg;
++ int err;
+
-+ /* Update p2m structure with the new mapping. */
-+ set_phys_to_machine(__pa(kernel_vaddr) >> PAGE_SHIFT,
-+ FOREIGN_FRAME(private_data->
-+ grants[slot_index+i]
-+ .u.valid.dev_bus_addr
-+ >> PAGE_SHIFT));
-+ } else {
-+ /* USING SHADOW PAGE TABLES. */
-+ /* In this case, we simply insert the page into the VM
-+ * area. */
-+ ret = vm_insert_page(vma, user_vaddr, page);
-+ }
++ /* Allocate the event queue, VI and so on. */
++ err = ef_get_vnic(bend);
++ if (err) {
++ EPRINTK("Failed to allocate hardware resource for bend:"
++ "error %d\n", err);
++ return err;
++ }
+
++ /* Set up the filter management */
++ err = netback_accel_filter_init(bend);
++ if (err) {
++ EPRINTK("Filter setup failed, error %d", err);
++ ef_free_vnic(bend);
++ return err;
+ }
+
-+ up_write(&private_data->grants_sem);
-+ return 0;
++ net_accel_msg_init(&msg, NET_ACCEL_MSG_SETHW);
+
-+undo_map_out:
-+ /* If we have a mapping failure, the unmapping will be taken care of
-+ * by do_mmap_pgoff(), which will eventually call gntdev_clear_pte().
-+ * All we need to do here is free the vma_private_data.
++ /*
++ * Extract the low-level hardware info we will actually pass to the
++ * other end, and set up the grants/ioremap permissions needed
+ */
-+ kfree(vma->vm_private_data);
++ err = netback_accel_hwinfo(bend, &msg.u.hw);
+
-+ /* THIS IS VERY UNPLEASANT: do_mmap_pgoff() will set the vma->vm_file
-+ * to NULL on failure. However, we need this in gntdev_clear_pte() to
-+ * unmap the grants. Therefore, we smuggle a reference to the file's
-+ * private data in the VM area's private data pointer.
++ if (err != 0) {
++ netback_accel_filter_shutdown(bend);
++ ef_free_vnic(bend);
++ return err;
++ }
++
++ /* Send the message, this is a reply to a hello-reply */
++ err = net_accel_msg_reply_notify(bend->shared_page,
++ bend->msg_channel_irq,
++ &bend->to_domU, &msg);
++
++ /*
++ * The message should succeed as it's logically a reply and we
++ * guarantee space for replies, but a misbehaving frontend
++ * could result in that behaviour, so be tolerant
+ */
-+ vma->vm_private_data = private_data;
-+
-+ up_write(&private_data->grants_sem);
++ if (err != 0) {
++ netback_accel_release_hwinfo(bend);
++ netback_accel_filter_shutdown(bend);
++ ef_free_vnic(bend);
++ }
+
-+ return -ENOMEM;
++ return err;
+}
+
-+static pte_t gntdev_clear_pte(struct vm_area_struct *vma, unsigned long addr,
-+ pte_t *ptep, int is_fullmm)
-+{
-+ int slot_index, ret;
-+ pte_t copy;
-+ struct gnttab_unmap_grant_ref op;
-+ gntdev_file_private_data_t *private_data;
+
-+ /* THIS IS VERY UNPLEASANT: do_mmap_pgoff() will set the vma->vm_file
-+ * to NULL on failure. However, we need this in gntdev_clear_pte() to
-+ * unmap the grants. Therefore, we smuggle a reference to the file's
-+ * private data in the VM area's private data pointer.
-+ */
-+ if (vma->vm_file) {
-+ private_data = (gntdev_file_private_data_t *)
-+ vma->vm_file->private_data;
-+ } else if (vma->vm_private_data) {
-+ private_data = (gntdev_file_private_data_t *)
-+ vma->vm_private_data;
-+ } else {
-+ private_data = NULL; /* gcc warning */
++/* Free hardware resources */
++void netback_accel_shutdown_vnic_hw(struct netback_accel *bend)
++{
++ /*
++ * Only try and release resources if accel_hw_priv was setup,
++ * otherwise there is nothing to do as we're on "null-op"
++ * acceleration
++ */
++ switch (bend->hw_state) {
++ case NETBACK_ACCEL_RES_HWINFO:
++ VPRINTK("Release hardware resources\n");
++ netback_accel_release_hwinfo(bend);
++ /* deliberate drop through */
++ case NETBACK_ACCEL_RES_FILTER:
++ VPRINTK("Free filters...\n");
++ netback_accel_filter_shutdown(bend);
++ /* deliberate drop through */
++ case NETBACK_ACCEL_RES_ALLOC:
++ VPRINTK("Free vnic...\n");
++ ef_free_vnic(bend);
++ /* deliberate drop through */
++ case NETBACK_ACCEL_RES_NONE:
++ break;
++ default:
+ BUG();
+ }
++}
+
-+ /* Copy the existing value of the PTE for returning. */
-+ copy = *ptep;
++/**************************************************************************
++ *
++ * Buffer table stuff
++ *
++ **************************************************************************/
+
-+ /* Calculate the grant relating to this PTE. */
-+ slot_index = vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT);
++/*
++ * Undo any allocation that netback_accel_msg_rx_buffer_map() has made
++ * if it fails half way through
++ */
++static inline void buffer_map_cleanup(struct netback_accel *bend, int i)
++{
++ while (i > 0) {
++ i--;
++ bend->buffer_maps_index--;
++ net_accel_unmap_device_page(bend->hdev_data,
++ bend->buffer_maps[bend->buffer_maps_index],
++ bend->buffer_addrs[bend->buffer_maps_index]);
++ }
++}
+
-+ /* Only unmap grants if the slot has been mapped. This could be being
-+ * called from a failing mmap().
-+ */
-+ if (private_data->grants[slot_index].state == GNTDEV_SLOT_MAPPED) {
+
-+ /* First, we clear the user space mapping, if it has been made.
-+ */
-+ if (private_data->grants[slot_index].u.valid.user_handle !=
-+ GNTDEV_INVALID_HANDLE &&
-+ !xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* NOT USING SHADOW PAGE TABLES. */
-+ gnttab_set_unmap_op(&op, virt_to_machine(ptep),
-+ GNTMAP_contains_pte,
-+ private_data->grants[slot_index]
-+ .u.valid.user_handle);
-+ ret = HYPERVISOR_grant_table_op(
-+ GNTTABOP_unmap_grant_ref, &op, 1);
-+ BUG_ON(ret);
-+ if (op.status)
-+ printk("User unmap grant status = %d\n",
-+ op.status);
-+ } else {
-+ /* USING SHADOW PAGE TABLES. */
-+ pte_clear_full(vma->vm_mm, addr, ptep, is_fullmm);
-+ }
++int netback_accel_add_buffers(struct netback_accel *bend, int pages, int log2_pages,
++ u32 *grants, u32 *buf_addr_out)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ unsigned long long addr_array[NET_ACCEL_MSG_MAX_PAGE_REQ];
++ int rc, i, index;
++ u64 dev_bus_addr;
+
-+ /* Finally, we unmap the grant from kernel space. */
-+ gnttab_set_unmap_op(&op,
-+ get_kernel_vaddr(private_data, slot_index),
-+ GNTMAP_host_map,
-+ private_data->grants[slot_index].u.valid
-+ .kernel_handle);
-+ ret = HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref,
-+ &op, 1);
-+ BUG_ON(ret);
-+ if (op.status)
-+ printk("Kernel unmap grant status = %d\n", op.status);
++ /* Make sure we can't overflow the dma_maps array */
++ if (accel_hw_priv->dma_maps_index >=
++ bend->max_pages / NET_ACCEL_MSG_MAX_PAGE_REQ) {
++ EPRINTK("%s: too many buffer table allocations: %d %d\n",
++ __FUNCTION__, accel_hw_priv->dma_maps_index,
++ bend->max_pages / NET_ACCEL_MSG_MAX_PAGE_REQ);
++ return -EINVAL;
++ }
+
++ /* Make sure we can't overflow the buffer_maps array */
++ if (bend->buffer_maps_index + pages > bend->max_pages) {
++ EPRINTK("%s: too many pages mapped: %d + %d > %d\n",
++ __FUNCTION__, bend->buffer_maps_index,
++ pages, bend->max_pages);
++ return -EINVAL;
++ }
+
-+ /* Return slot to the not-yet-mapped state, so that it may be
-+ * mapped again, or removed by a subsequent ioctl.
-+ */
-+ private_data->grants[slot_index].state =
-+ GNTDEV_SLOT_NOT_YET_MAPPED;
++ for (i = 0; i < pages; i++) {
++ VPRINTK("%s: mapping page %d\n", __FUNCTION__, i);
++ rc = net_accel_map_device_page
++ (bend->hdev_data, grants[i],
++ &bend->buffer_maps[bend->buffer_maps_index],
++ &dev_bus_addr);
++
++ if (rc != 0) {
++ EPRINTK("error in net_accel_map_device_page\n");
++ buffer_map_cleanup(bend, i);
++ return rc;
++ }
++
++ bend->buffer_addrs[bend->buffer_maps_index] = dev_bus_addr;
+
-+ /* Invalidate the physical to machine mapping for this page. */
-+ set_phys_to_machine(__pa(get_kernel_vaddr(private_data,
-+ slot_index))
-+ >> PAGE_SHIFT, INVALID_P2M_ENTRY);
++ bend->buffer_maps_index++;
+
-+ } else {
-+ pte_clear_full(vma->vm_mm, addr, ptep, is_fullmm);
++ addr_array[i] = dev_bus_addr;
+ }
+
-+ return copy;
-+}
++ VPRINTK("%s: mapping dma addresses to vih %p\n", __FUNCTION__,
++ accel_hw_priv->efx_vih);
+
-+/* "Destructor" for a VM area.
-+ */
-+static void gntdev_vma_close(struct vm_area_struct *vma) {
-+ if (vma->vm_private_data) {
-+ kfree(vma->vm_private_data);
++ index = accel_hw_priv->dma_maps_index;
++ if ((rc = efx_vi_dma_map_addrs(accel_hw_priv->efx_vih, addr_array, pages,
++ &(accel_hw_priv->dma_maps[index]))) < 0) {
++ EPRINTK("error in dma_map_pages\n");
++ buffer_map_cleanup(bend, i);
++ return rc;
+ }
++
++ accel_hw_priv->dma_maps_index++;
++ NETBACK_ACCEL_STATS_OP(bend->stats.num_buffer_pages += pages);
++
++ //DPRINTK("%s: getting map address\n", __FUNCTION__);
++
++ *buf_addr_out = efx_vi_dma_get_map_addr(accel_hw_priv->efx_vih,
++ accel_hw_priv->dma_maps[index]);
++
++ //DPRINTK("%s: done\n", __FUNCTION__);
++
++ return 0;
+}
+
-+/* Called when an ioctl is made on the device.
-+ */
-+static long gntdev_ioctl(struct file *flip,
-+ unsigned int cmd, unsigned long arg)
++
++int netback_accel_remove_buffers(struct netback_accel *bend)
+{
-+ int rc = 0;
-+ gntdev_file_private_data_t *private_data =
-+ (gntdev_file_private_data_t *) flip->private_data;
++ /* Only try to free buffers if accel_hw_priv was setup */
++ if (bend->hw_state != NETBACK_ACCEL_RES_NONE) {
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ int i;
+
-+ switch (cmd) {
-+ case IOCTL_GNTDEV_MAP_GRANT_REF:
-+ {
-+ struct ioctl_gntdev_map_grant_ref op;
-+ down_write(&private_data->grants_sem);
-+ down_write(&private_data->free_list_sem);
++ efx_vi_reset(accel_hw_priv->efx_vih);
+
-+ if ((rc = copy_from_user(&op, (void __user *) arg,
-+ sizeof(op)))) {
-+ rc = -EFAULT;
-+ goto map_out;
++ while (accel_hw_priv->dma_maps_index > 0) {
++ accel_hw_priv->dma_maps_index--;
++ i = accel_hw_priv->dma_maps_index;
++ efx_vi_dma_unmap_addrs(accel_hw_priv->efx_vih,
++ accel_hw_priv->dma_maps[i]);
+ }
-+ if (unlikely(op.count <= 0)) {
-+ rc = -EINVAL;
-+ goto map_out;
++
++ while (bend->buffer_maps_index > 0) {
++ VPRINTK("Unmapping granted buffer %d\n",
++ bend->buffer_maps_index);
++ bend->buffer_maps_index--;
++ i = bend->buffer_maps_index;
++ net_accel_unmap_device_page(bend->hdev_data,
++ bend->buffer_maps[i],
++ bend->buffer_addrs[i]);
+ }
+
-+ if (op.count == 1) {
-+ if ((rc = add_grant_reference(flip, &op.refs[0],
-+ &op.index)) < 0) {
-+ printk(KERN_ERR "Adding grant reference "
-+ "failed (%d).\n", rc);
-+ goto map_out;
-+ }
-+ } else {
-+ struct ioctl_gntdev_grant_ref *refs, *u;
-+ refs = kmalloc(op.count * sizeof(*refs), GFP_KERNEL);
-+ if (!refs) {
-+ rc = -ENOMEM;
-+ goto map_out;
-+ }
-+ u = ((struct ioctl_gntdev_map_grant_ref *)arg)->refs;
-+ if ((rc = copy_from_user(refs,
-+ (void __user *)u,
-+ sizeof(*refs) * op.count))) {
-+ printk(KERN_ERR "Copying refs from user failed"
-+ " (%d).\n", rc);
-+ rc = -EINVAL;
-+ goto map_out;
-+ }
-+ if ((rc = find_contiguous_free_range(flip, op.count))
-+ < 0) {
-+ printk(KERN_ERR "Finding contiguous range "
-+ "failed (%d).\n", rc);
-+ kfree(refs);
-+ goto map_out;
-+ }
-+ op.index = rc << PAGE_SHIFT;
-+ if ((rc = add_grant_references(flip, op.count,
-+ refs, rc))) {
-+ printk(KERN_ERR "Adding grant references "
-+ "failed (%d).\n", rc);
-+ kfree(refs);
-+ goto map_out;
-+ }
-+ compress_free_list(flip);
-+ kfree(refs);
-+ }
-+ if ((rc = copy_to_user((void __user *) arg,
-+ &op,
-+ sizeof(op)))) {
-+ printk(KERN_ERR "Copying result back to user failed "
-+ "(%d)\n", rc);
-+ rc = -EFAULT;
-+ goto map_out;
-+ }
-+ map_out:
-+ up_write(&private_data->grants_sem);
-+ up_write(&private_data->free_list_sem);
-+ return rc;
++ NETBACK_ACCEL_STATS_OP(bend->stats.num_buffer_pages = 0);
+ }
-+ case IOCTL_GNTDEV_UNMAP_GRANT_REF:
-+ {
-+ struct ioctl_gntdev_unmap_grant_ref op;
-+ int i, start_index;
+
-+ down_write(&private_data->grants_sem);
-+ down_write(&private_data->free_list_sem);
++ return 0;
++}
+
-+ if ((rc = copy_from_user(&op,
-+ (void __user *) arg,
-+ sizeof(op)))) {
-+ rc = -EFAULT;
-+ goto unmap_out;
-+ }
++/**************************************************************************
++ *
++ * Filter stuff
++ *
++ **************************************************************************/
+
-+ start_index = op.index >> PAGE_SHIFT;
++static int netback_accel_filter_init(struct netback_accel *bend)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ int i, rc;
+
-+ /* First, check that all pages are in the NOT_YET_MAPPED
-+ * state.
-+ */
-+ for (i = 0; i < op.count; ++i) {
-+ if (unlikely
-+ (private_data->grants[start_index + i].state
-+ != GNTDEV_SLOT_NOT_YET_MAPPED)) {
-+ if (private_data->grants[start_index + i].state
-+ == GNTDEV_SLOT_INVALID) {
-+ printk(KERN_ERR
-+ "Tried to remove an invalid "
-+ "grant at offset 0x%x.",
-+ (start_index + i)
-+ << PAGE_SHIFT);
-+ rc = -EINVAL;
-+ } else {
-+ printk(KERN_ERR
-+ "Tried to remove a grant which "
-+ "is currently mmap()-ed at "
-+ "offset 0x%x.",
-+ (start_index + i)
-+ << PAGE_SHIFT);
-+ rc = -EBUSY;
-+ }
-+ goto unmap_out;
-+ }
-+ }
++ BUG_ON(bend->hw_state != NETBACK_ACCEL_RES_ALLOC);
+
-+ /* Unmap pages and add them to the free list.
-+ */
-+ for (i = 0; i < op.count; ++i) {
-+ private_data->grants[start_index+i].state =
-+ GNTDEV_SLOT_INVALID;
-+ private_data->grants[start_index+i].u.free_list_index =
-+ private_data->free_list_size;
-+ private_data->free_list[private_data->free_list_size] =
-+ start_index + i;
-+ ++private_data->free_list_size;
-+ }
-+ compress_free_list(flip);
++ spin_lock_init(&accel_hw_priv->filter_lock);
+
-+ unmap_out:
-+ up_write(&private_data->grants_sem);
-+ up_write(&private_data->free_list_sem);
++ if ((rc = cuckoo_hash_init(&accel_hw_priv->filter_hash_table,
++ 5 /* space for 32 filters */, 8)) != 0) {
++ EPRINTK("Failed to initialise filter hash table\n");
+ return rc;
+ }
-+ case IOCTL_GNTDEV_GET_OFFSET_FOR_VADDR:
-+ {
-+ struct ioctl_gntdev_get_offset_for_vaddr op;
-+ struct vm_area_struct *vma;
-+ unsigned long vaddr;
+
-+ if ((rc = copy_from_user(&op,
-+ (void __user *) arg,
-+ sizeof(op)))) {
-+ rc = -EFAULT;
-+ goto get_offset_out;
-+ }
-+ vaddr = (unsigned long)op.vaddr;
++ accel_hw_priv->fspecs = kzalloc(sizeof(struct netback_accel_filter_spec) *
++ bend->quotas.max_filters,
++ GFP_KERNEL);
+
-+ down_read(¤t->mm->mmap_sem);
-+ vma = find_vma(current->mm, vaddr);
-+ if (vma == NULL) {
-+ rc = -EFAULT;
-+ goto get_offset_unlock_out;
-+ }
-+ if ((!vma->vm_ops) || (vma->vm_ops != &gntdev_vmops)) {
-+ printk(KERN_ERR "The vaddr specified does not belong "
-+ "to a gntdev instance: %#lx\n", vaddr);
-+ rc = -EFAULT;
-+ goto get_offset_unlock_out;
-+ }
-+ if (vma->vm_start != vaddr) {
-+ printk(KERN_ERR "The vaddr specified in an "
-+ "IOCTL_GNTDEV_GET_OFFSET_FOR_VADDR must be at "
-+ "the start of the VM area. vma->vm_start = "
-+ "%#lx; vaddr = %#lx\n",
-+ vma->vm_start, vaddr);
-+ rc = -EFAULT;
-+ goto get_offset_unlock_out;
-+ }
-+ op.offset = vma->vm_pgoff << PAGE_SHIFT;
-+ op.count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-+ up_read(¤t->mm->mmap_sem);
-+ if ((rc = copy_to_user((void __user *) arg,
-+ &op,
-+ sizeof(op)))) {
-+ rc = -EFAULT;
-+ goto get_offset_out;
-+ }
-+ goto get_offset_out;
-+ get_offset_unlock_out:
-+ up_read(¤t->mm->mmap_sem);
-+ get_offset_out:
-+ return rc;
++ if (accel_hw_priv->fspecs == NULL) {
++ EPRINTK("No memory for filter specs.\n");
++ cuckoo_hash_destroy(&accel_hw_priv->filter_hash_table);
++ return -ENOMEM;
+ }
-+ default:
-+ return -ENOIOCTLCMD;
++
++ for (i = 0; i < bend->quotas.max_filters; i++) {
++ accel_hw_priv->free_filters |= (1 << i);
+ }
+
++ /* Base mask on highest set bit in max_filters */
++ accel_hw_priv->filter_idx_mask = (1 << fls(bend->quotas.max_filters)) - 1;
++ VPRINTK("filter setup: max is %x mask is %x\n",
++ bend->quotas.max_filters, accel_hw_priv->filter_idx_mask);
++
++ bend->hw_state = NETBACK_ACCEL_RES_FILTER;
++
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/gntdev/Makefile linux-2.6.18-xen.hg/drivers/xen/gntdev/Makefile
---- linux-2.6.18/drivers/xen/gntdev/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/gntdev/Makefile 2007-12-23 11:15:34.047935255 +0100
-@@ -0,0 +1 @@
-+obj-$(CONFIG_XEN_GRANT_DEV) := gntdev.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/Kconfig linux-2.6.18-xen.hg/drivers/xen/Kconfig
---- linux-2.6.18/drivers/xen/Kconfig 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/Kconfig 2007-12-23 11:15:33.537908474 +0100
-@@ -0,0 +1,286 @@
-+#
-+# This Kconfig describe xen options
-+#
+
-+mainmenu "Xen Configuration"
+
-+config XEN
-+ bool
-+ default y if X86_XEN || X86_64_XEN
-+ help
-+ This is the Linux Xen port.
++static inline void make_filter_key(cuckoo_hash_ip_key *key,
++ struct netback_accel_filter_spec *filt)
+
-+if XEN
-+config XEN_INTERFACE_VERSION
-+ hex
-+ default 0x00030207
++{
++ key->local_ip = filt->destip_be;
++ key->local_port = filt->destport_be;
++ key->proto = filt->proto;
++}
+
-+menu "XEN"
+
-+config XEN_PRIVILEGED_GUEST
-+ bool "Privileged Guest (domain 0)"
-+ default n
-+ help
-+ Support for privileged operation (domain 0)
++static inline
++void netback_accel_free_filter(struct falcon_bend_accel_priv *accel_hw_priv,
++ int filter)
++{
++ cuckoo_hash_ip_key filter_key;
+
-+config XEN_UNPRIVILEGED_GUEST
-+ def_bool !XEN_PRIVILEGED_GUEST
++ if (!(accel_hw_priv->free_filters & (1 << filter))) {
++ efx_vi_filter_stop(accel_hw_priv->efx_vih,
++ accel_hw_priv->fspecs[filter].filter_handle);
++ make_filter_key(&filter_key, &(accel_hw_priv->fspecs[filter]));
++ if (cuckoo_hash_remove(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)&filter_key)) {
++ EPRINTK("%s: Couldn't find filter to remove from table\n",
++ __FUNCTION__);
++ BUG();
++ }
++ }
++}
+
-+config XEN_PRIVCMD
-+ def_bool y
-+ depends on PROC_FS
+
-+config XEN_XENBUS_DEV
-+ def_bool y
-+ depends on PROC_FS
++static void netback_accel_filter_shutdown(struct netback_accel *bend)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ int i;
++ unsigned long flags;
+
-+config XEN_BACKEND
-+ tristate "Backend driver support"
-+ default XEN_PRIVILEGED_GUEST
-+ help
-+ Support for backend device drivers that provide I/O services
-+ to other virtual machines.
++ BUG_ON(bend->hw_state != NETBACK_ACCEL_RES_FILTER);
+
-+config XEN_BLKDEV_BACKEND
-+ tristate "Block-device backend driver"
-+ depends on XEN_BACKEND
-+ default XEN_BACKEND
-+ help
-+ The block-device backend driver allows the kernel to export its
-+ block devices to other guests via a high-performance shared-memory
-+ interface.
++ spin_lock_irqsave(&accel_hw_priv->filter_lock, flags);
+
-+config XEN_BLKDEV_TAP
-+ tristate "Block-device tap backend driver"
-+ depends on XEN_BACKEND
-+ default XEN_BACKEND
-+ help
-+ The block tap driver is an alternative to the block back driver
-+ and allows VM block requests to be redirected to userspace through
-+ a device interface. The tap allows user-space development of
-+ high-performance block backends, where disk images may be implemented
-+ as files, in memory, or on other hosts across the network. This
-+ driver can safely coexist with the existing blockback driver.
++ BUG_ON(accel_hw_priv->fspecs == NULL);
+
-+config XEN_NETDEV_BACKEND
-+ tristate "Network-device backend driver"
-+ depends on XEN_BACKEND && NET
-+ default XEN_BACKEND
-+ help
-+ The network-device backend driver allows the kernel to export its
-+ network devices to other guests via a high-performance shared-memory
-+ interface.
++ for (i = 0; i < bend->quotas.max_filters; i++) {
++ netback_accel_free_filter(accel_hw_priv, i);
++ }
++
++ kfree(accel_hw_priv->fspecs);
++ accel_hw_priv->fspecs = NULL;
++ accel_hw_priv->free_filters = 0;
++
++ cuckoo_hash_destroy(&accel_hw_priv->filter_hash_table);
+
-+config XEN_NETDEV_PIPELINED_TRANSMITTER
-+ bool "Pipelined transmitter (DANGEROUS)"
-+ depends on XEN_NETDEV_BACKEND
-+ help
-+ If the net backend is a dumb domain, such as a transparent Ethernet
-+ bridge with no local IP interface, it is safe to say Y here to get
-+ slightly lower network overhead.
-+ If the backend has a local IP interface; or may be doing smart things
-+ like reassembling packets to perform firewall filtering; or if you
-+ are unsure; or if you experience network hangs when this option is
-+ enabled; then you must say N here.
++ spin_unlock_irqrestore(&accel_hw_priv->filter_lock, flags);
+
-+config XEN_NETDEV_LOOPBACK
-+ tristate "Network-device loopback driver"
-+ depends on XEN_NETDEV_BACKEND
-+ help
-+ A two-interface loopback device to emulate a local netfront-netback
-+ connection. If unsure, it is probably safe to say N here.
++ bend->hw_state = NETBACK_ACCEL_RES_ALLOC;
++}
+
-+config XEN_PCIDEV_BACKEND
-+ tristate "PCI-device backend driver"
-+ depends on PCI && XEN_BACKEND
-+ default XEN_BACKEND
-+ help
-+ The PCI device backend driver allows the kernel to export arbitrary
-+ PCI devices to other guests. If you select this to be a module, you
-+ will need to make sure no other driver has bound to the device(s)
-+ you want to make visible to other guests.
+
-+choice
-+ prompt "PCI Backend Mode"
-+ depends on XEN_PCIDEV_BACKEND
-+ default XEN_PCIDEV_BACKEND_VPCI if !IA64
-+ default XEN_PCIDEV_BACKEND_CONTROLLER if IA64
++/*! Suggest a filter to replace when we want to insert a new one and have
++ * none free.
++ */
++static unsigned get_victim_filter(struct netback_accel *bend)
++{
++ /*
++ * We could attempt to get really clever, and may do at some
++ * point, but random replacement is v. cheap and low on
++ * pathological worst cases.
++ */
++ unsigned index, cycles;
+
-+config XEN_PCIDEV_BACKEND_VPCI
-+ bool "Virtual PCI"
-+ ---help---
-+ This PCI Backend hides the true PCI topology and makes the frontend
-+ think there is a single PCI bus with only the exported devices on it.
-+ For example, a device at 03:05.0 will be re-assigned to 00:00.0. A
-+ second device at 02:1a.1 will be re-assigned to 00:01.1.
++ rdtscl(cycles);
+
-+config XEN_PCIDEV_BACKEND_PASS
-+ bool "Passthrough"
-+ ---help---
-+ This PCI Backend provides a real view of the PCI topology to the
-+ frontend (for example, a device at 06:01.b will still appear at
-+ 06:01.b to the frontend). This is similar to how Xen 2.0.x exposed
-+ PCI devices to its driver domains. This may be required for drivers
-+ which depend on finding their hardward in certain bus/slot
-+ locations.
++ /*
++ * Some doubt about the quality of the bottom few bits, so
++ * throw 'em * away
++ */
++ index = (cycles >> 4) & ((struct falcon_bend_accel_priv *)
++ bend->accel_hw_priv)->filter_idx_mask;
++ /*
++ * We don't enforce that the number of filters is a power of
++ * two, but the masking gets us to within one subtraction of a
++ * valid index
++ */
++ if (index >= bend->quotas.max_filters)
++ index -= bend->quotas.max_filters;
++ DPRINTK("backend %s->%d has no free filters. Filter %d will be evicted\n",
++ bend->nicname, bend->far_end, index);
++ return index;
++}
+
-+config XEN_PCIDEV_BACKEND_SLOT
-+ bool "Slot"
-+ ---help---
-+ This PCI Backend hides the true PCI topology and makes the frontend
-+ think there is a single PCI bus with only the exported devices on it.
-+ Contrary to the virtual PCI backend, a function becomes a new slot.
-+ For example, a device at 03:05.2 will be re-assigned to 00:00.0. A
-+ second device at 02:1a.1 will be re-assigned to 00:01.0.
+
-+config XEN_PCIDEV_BACKEND_CONTROLLER
-+ bool "Controller"
-+ depends on IA64
-+ ---help---
-+ This PCI backend virtualizes the PCI bus topology by providing a
-+ virtual bus per PCI root device. Devices which are physically under
-+ the same root bus will appear on the same virtual bus. For systems
-+ with complex I/O addressing, this is the only backend which supports
-+ extended I/O port spaces and MMIO translation offsets. This backend
-+ also supports slot virtualization. For example, a device at
-+ 0000:01:02.1 will be re-assigned to 0000:00:00.0. A second device
-+ at 0000:02:05.0 (behind a P2P bridge on bus 0000:01) will be
-+ re-assigned to 0000:00:01.0. A third device at 0000:16:05.0 (under
-+ a different PCI root bus) will be re-assigned to 0000:01:00.0.
++/* Add a filter for the specified IP/port to the backend */
++int
++netback_accel_filter_check_add(struct netback_accel *bend,
++ struct netback_accel_filter_spec *filt)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ struct netback_accel_filter_spec *fs;
++ unsigned filter_index;
++ unsigned long flags;
++ int rc, recycling = 0;
++ cuckoo_hash_ip_key filter_key, evict_key;
+
-+endchoice
++ BUG_ON(filt->proto != IPPROTO_TCP && filt->proto != IPPROTO_UDP);
+
-+config XEN_PCIDEV_BE_DEBUG
-+ bool "PCI Backend Debugging"
-+ depends on XEN_PCIDEV_BACKEND
++ DPRINTK("Will add %s filter for dst ip %08x and dst port %d\n",
++ (filt->proto == IPPROTO_TCP) ? "TCP" : "UDP",
++ be32_to_cpu(filt->destip_be), be16_to_cpu(filt->destport_be));
+
-+config XEN_TPMDEV_BACKEND
-+ tristate "TPM-device backend driver"
-+ depends on XEN_BACKEND
-+ help
-+ The TPM-device backend driver
++ spin_lock_irqsave(&accel_hw_priv->filter_lock, flags);
++ /*
++ * Check to see if we're already filtering this IP address and
++ * port. Happens if you insert a filter mid-stream as there
++ * are many packets backed up to be delivered to dom0 already
++ */
++ make_filter_key(&filter_key, filt);
++ if (cuckoo_hash_lookup(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)(&filter_key),
++ &filter_index)) {
++ DPRINTK("Found matching filter %d already in table\n",
++ filter_index);
++ rc = -1;
++ goto out;
++ }
+
-+config XEN_BLKDEV_FRONTEND
-+ tristate "Block-device frontend driver"
-+ default y
-+ help
-+ The block-device frontend driver allows the kernel to access block
-+ devices mounted within another guest OS. Unless you are building a
-+ dedicated device-driver domain, or your master control domain
-+ (domain 0), then you almost certainly want to say Y here.
++ if (accel_hw_priv->free_filters == 0) {
++ filter_index = get_victim_filter(bend);
++ recycling = 1;
++ } else {
++ filter_index = __ffs(accel_hw_priv->free_filters);
++ clear_bit(filter_index, &accel_hw_priv->free_filters);
++ }
+
-+config XEN_NETDEV_FRONTEND
-+ tristate "Network-device frontend driver"
-+ depends on NET
-+ default y
-+ help
-+ The network-device frontend driver allows the kernel to access
-+ network interfaces within another guest OS. Unless you are building a
-+ dedicated device-driver domain, or your master control domain
-+ (domain 0), then you almost certainly want to say Y here.
++ fs = &accel_hw_priv->fspecs[filter_index];
+
-+config XEN_GRANT_DEV
-+ tristate "User-space granted page access driver"
-+ default XEN_PRIVILEGED_GUEST
-+ help
-+ Device for accessing (in user-space) pages that have been granted
-+ by other domains.
++ if (recycling) {
++ DPRINTK("Removing filter index %d handle %p\n", filter_index,
++ fs->filter_handle);
+
-+config XEN_FRAMEBUFFER
-+ tristate "Framebuffer-device frontend driver"
-+ depends on FB
-+ select FB_CFB_FILLRECT
-+ select FB_CFB_COPYAREA
-+ select FB_CFB_IMAGEBLIT
-+ default y
-+ help
-+ The framebuffer-device frontend drivers allows the kernel to create a
-+ virtual framebuffer. This framebuffer can be viewed in another
-+ domain. Unless this domain has access to a real video card, you
-+ probably want to say Y here.
++ if ((rc = efx_vi_filter_stop(accel_hw_priv->efx_vih,
++ fs->filter_handle)) != 0) {
++ EPRINTK("Couldn't clear NIC filter table entry %d\n", rc);
++ }
+
-+config XEN_KEYBOARD
-+ tristate "Keyboard-device frontend driver"
-+ depends on XEN_FRAMEBUFFER && INPUT
-+ default y
-+ help
-+ The keyboard-device frontend driver allows the kernel to create a
-+ virtual keyboard. This keyboard can then be driven by another
-+ domain. If you've said Y to CONFIG_XEN_FRAMEBUFFER, you probably
-+ want to say Y here.
++ make_filter_key(&evict_key, fs);
++ if (cuckoo_hash_remove(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)&evict_key)) {
++ EPRINTK("Couldn't find filter to remove from table\n");
++ BUG();
++ }
++ NETBACK_ACCEL_STATS_OP(bend->stats.num_filters--);
++ }
+
-+config XEN_SCRUB_PAGES
-+ bool "Scrub memory before freeing it to Xen"
-+ default y
-+ help
-+ Erase memory contents before freeing it back to Xen's global
-+ pool. This ensures that any secrets contained within that
-+ memory (e.g., private keys) cannot be found by other guests that
-+ may be running on the machine. Most people will want to say Y here.
-+ If security is not a concern then you may increase performance by
-+ saying N.
++ /* Update the filter spec with new details */
++ *fs = *filt;
+
-+config XEN_DISABLE_SERIAL
-+ bool "Disable serial port drivers"
-+ default y
-+ help
-+ Disable serial port drivers, allowing the Xen console driver
-+ to provide a serial console at ttyS0.
++ if ((rc = cuckoo_hash_add(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)&filter_key, filter_index,
++ 1)) != 0) {
++ EPRINTK("Error (%d) adding filter to table\n", rc);
++ accel_hw_priv->free_filters |= (1 << filter_index);
++ goto out;
++ }
+
-+config XEN_SYSFS
-+ tristate "Export Xen attributes in sysfs"
-+ depends on SYSFS
-+ select SYS_HYPERVISOR
-+ default y
-+ help
-+ Xen hypervisor attributes will show up under /sys/hypervisor/.
++ rc = efx_vi_filter(accel_hw_priv->efx_vih, filt->proto, filt->destip_be,
++ filt->destport_be,
++ (struct filter_resource_t **)&fs->filter_handle);
+
-+choice
-+ prompt "Xen version compatibility"
-+ default XEN_COMPAT_030002_AND_LATER
++ if (rc != 0) {
++ EPRINTK("Hardware filter insertion failed. Error %d\n", rc);
++ accel_hw_priv->free_filters |= (1 << filter_index);
++ cuckoo_hash_remove(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)&filter_key);
++ rc = -1;
++ goto out;
++ }
++
++ NETBACK_ACCEL_STATS_OP(bend->stats.num_filters++);
++
++ VPRINTK("%s: success index %d handle %p\n", __FUNCTION__, filter_index,
++ fs->filter_handle);
++
++ rc = filter_index;
++ out:
++ spin_unlock_irqrestore(&accel_hw_priv->filter_lock, flags);
++ return rc;
++}
++
++
++/* Remove a filter entry for the specific device and IP/port */
++static void netback_accel_filter_remove(struct netback_accel *bend,
++ int filter_index)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++
++ BUG_ON(accel_hw_priv->free_filters & (1 << filter_index));
++ netback_accel_free_filter(accel_hw_priv, filter_index);
++ accel_hw_priv->free_filters |= (1 << filter_index);
++}
++
++
++/* Remove a filter entry for the specific device and IP/port */
++void netback_accel_filter_remove_spec(struct netback_accel *bend,
++ struct netback_accel_filter_spec *filt)
++{
++ struct falcon_bend_accel_priv *accel_hw_priv = bend->accel_hw_priv;
++ unsigned filter_found;
++ unsigned long flags;
++ cuckoo_hash_ip_key filter_key;
++ struct netback_accel_filter_spec *fs;
++
++ if (filt->proto == IPPROTO_TCP) {
++ DPRINTK("Remove TCP filter for dst ip %08x and dst port %d\n",
++ be32_to_cpu(filt->destip_be),
++ be16_to_cpu(filt->destport_be));
++ } else if (filt->proto == IPPROTO_UDP) {
++ DPRINTK("Remove UDP filter for dst ip %08x and dst port %d\n",
++ be32_to_cpu(filt->destip_be),
++ be16_to_cpu(filt->destport_be));
++ } else {
++ /*
++ * This could be provoked by an evil frontend, so can't
++ * BUG(), but harmless as it should fail tests below
++ */
++ DPRINTK("Non-TCP/UDP filter dst ip %08x and dst port %d\n",
++ be32_to_cpu(filt->destip_be),
++ be16_to_cpu(filt->destport_be));
++ }
++
++ spin_lock_irqsave(&accel_hw_priv->filter_lock, flags);
++
++ make_filter_key(&filter_key, filt);
++ if (!cuckoo_hash_lookup(&accel_hw_priv->filter_hash_table,
++ (cuckoo_hash_key *)(&filter_key),
++ &filter_found)) {
++ EPRINTK("Couldn't find matching filter already in table\n");
++ goto out;
++ }
++
++ /* Do a full check to make sure we've not had a hash collision */
++ fs = &accel_hw_priv->fspecs[filter_found];
++ if (fs->destip_be == filt->destip_be &&
++ fs->destport_be == filt->destport_be &&
++ fs->proto == filt->proto &&
++ !memcmp(fs->mac, filt->mac, ETH_ALEN)) {
++ netback_accel_filter_remove(bend, filter_found);
++ } else {
++ EPRINTK("Entry in hash table does not match filter spec\n");
++ goto out;
++ }
++
++ out:
++ spin_unlock_irqrestore(&accel_hw_priv->filter_lock, flags);
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_solarflare.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_solarflare.h 2008-05-19 00:33:46.786842968 +0300
+@@ -0,0 +1,88 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ config XEN_COMPAT_030002_AND_LATER
-+ bool "3.0.2 and later"
++#ifndef NETBACK_ACCEL_SOLARFLARE_H
++#define NETBACK_ACCEL_SOLARFLARE_H
+
-+ config XEN_COMPAT_030004_AND_LATER
-+ bool "3.0.4 and later"
++#include "accel.h"
++#include "accel_msg_iface.h"
+
-+ config XEN_COMPAT_030100_AND_LATER
-+ bool "3.1.0 and later"
++#include "driverlink_api.h"
+
-+ config XEN_COMPAT_LATEST_ONLY
-+ bool "no compatibility code"
++#define MAX_NICS 5
++#define MAX_PORTS 2
+
-+endchoice
+
-+config XEN_COMPAT
-+ hex
-+ default 0xffffff if XEN_COMPAT_LATEST_ONLY
-+ default 0x030100 if XEN_COMPAT_030100_AND_LATER
-+ default 0x030004 if XEN_COMPAT_030004_AND_LATER
-+ default 0x030002 if XEN_COMPAT_030002_AND_LATER
-+ default 0
++extern int netback_accel_sf_init(void);
++extern void netback_accel_sf_shutdown(void);
++extern int netback_accel_sf_hwtype(struct netback_accel *bend);
+
-+endmenu
++extern int netback_accel_sf_char_init(void);
++extern void netback_accel_sf_char_shutdown(void);
+
-+config HAVE_IRQ_IGNORE_UNHANDLED
-+ def_bool y
++extern int netback_accel_setup_vnic_hw(struct netback_accel *bend);
++extern void netback_accel_shutdown_vnic_hw(struct netback_accel *bend);
+
-+config NO_IDLE_HZ
-+ def_bool y
++extern int netback_accel_add_buffers(struct netback_accel *bend, int pages,
++ int log2_pages, u32 *grants,
++ u32 *buf_addr_out);
++extern int netback_accel_remove_buffers(struct netback_accel *bend);
+
-+config XEN_SMPBOOT
-+ def_bool y
-+ depends on SMP && !PPC_XEN
+
-+config XEN_BALLOON
-+ def_bool y
-+ depends on !PPC_XEN
++/* Add a filter for the specified IP/port to the backend */
++extern int
++netback_accel_filter_check_add(struct netback_accel *bend,
++ struct netback_accel_filter_spec *filt);
++/* Remove a filter entry for the specific device and IP/port */
++extern
++void netback_accel_filter_remove_index(struct netback_accel *bend,
++ int filter_index);
++extern
++void netback_accel_filter_remove_spec(struct netback_accel *bend,
++ struct netback_accel_filter_spec *filt);
+
-+config XEN_XENCOMM
-+ bool
++/* This is designed to look a bit like a skb */
++struct netback_pkt_buf {
++ union {
++ unsigned char *raw;
++ } mac;
++ union {
++ struct iphdr *iph;
++ struct arphdr *arph;
++ unsigned char *raw;
++ } nh;
++ int protocol;
++};
+
-+config XEN_DEVMEM
-+ def_bool y
++/*! \brief Handle a received packet: insert fast path filters as necessary
++ * \param skb The packet buffer
++ */
++extern void netback_accel_rx_packet(struct netback_pkt_buf *skb, void *fwd_priv);
+
-+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/Makefile linux-2.6.18-xen.hg/drivers/xen/Makefile
---- linux-2.6.18/drivers/xen/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/Makefile 2007-12-23 11:15:33.537908474 +0100
-@@ -0,0 +1,20 @@
-+obj-y += core/
-+obj-y += console/
-+obj-y += evtchn/
-+obj-y += xenbus/
-+obj-y += char/
++/*! \brief Handle a transmitted packet: update fast path filters as necessary
++ * \param skb The packet buffer
++ */
++extern void netback_accel_tx_packet(struct sk_buff *skb, void *fwd_priv);
+
-+obj-y += util.o
-+obj-$(CONFIG_XEN_BALLOON) += balloon/
-+obj-$(CONFIG_XEN_BLKDEV_BACKEND) += blkback/
-+obj-$(CONFIG_XEN_BLKDEV_TAP) += blktap/
-+obj-$(CONFIG_XEN_NETDEV_BACKEND) += netback/
-+obj-$(CONFIG_XEN_TPMDEV_BACKEND) += tpmback/
-+obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += blkfront/
-+obj-$(CONFIG_XEN_NETDEV_FRONTEND) += netfront/
-+obj-$(CONFIG_XEN_PCIDEV_BACKEND) += pciback/
-+obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += pcifront/
-+obj-$(CONFIG_XEN_FRAMEBUFFER) += fbfront/
-+obj-$(CONFIG_XEN_KEYBOARD) += fbfront/
-+obj-$(CONFIG_XEN_PRIVCMD) += privcmd/
-+obj-$(CONFIG_XEN_GRANT_DEV) += gntdev/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/accel.c linux-2.6.18-xen.hg/drivers/xen/netback/accel.c
---- linux-2.6.18/drivers/xen/netback/accel.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/accel.c 2007-12-23 11:15:34.051268764 +0100
-@@ -0,0 +1,269 @@
-+/******************************************************************************
-+ * drivers/xen/netback/accel.c
++#endif /* NETBACK_ACCEL_SOLARFLARE_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/accel_xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/accel_xenbus.c 2008-05-19 00:33:46.878848271 +0300
+@@ -0,0 +1,831 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
+ *
-+ * Interface between backend virtual network device and accelerated plugin.
-+ *
-+ * Copyright (C) 2007 Solarflare Communications, Inc
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/list.h>
-+#include <asm/atomic.h>
-+#include <xen/xenbus.h>
++#include <xen/evtchn.h>
+#include <linux/mutex.h>
+
++/* drivers/xen/netback/common.h */
+#include "common.h"
+
-+#if 0
-+#undef DPRINTK
-+#define DPRINTK(fmt, args...) \
-+ printk("netback/accel (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
-+#endif
++#include "accel.h"
++#include "accel_solarflare.h"
++#include "accel_util.h"
+
-+/*
-+ * A list of available netback accelerator plugin modules (each list
-+ * entry is of type struct netback_accelerator)
-+ */
-+static struct list_head accelerators_list;
-+/* Lock used to protect access to accelerators_list */
-+DEFINE_MUTEX(accelerators_mutex);
++#define NODENAME_PATH_FMT "backend/vif/%d/%d"
+
-+/*
-+ * Compare a backend to an accelerator, and decide if they are
-+ * compatible (i.e. if the accelerator should be used by the
-+ * backend)
++#define NETBACK_ACCEL_FROM_XENBUS_DEVICE(_dev) (struct netback_accel *) \
++ ((struct backend_info *)(_dev)->dev.driver_data)->netback_accel_priv
++
++/* List of all the bends currently in existence. */
++struct netback_accel *bend_list = NULL;
++DEFINE_MUTEX(bend_list_mutex);
++
++/* Put in bend_list. Must hold bend_list_mutex */
++static void link_bend(struct netback_accel *bend)
++{
++ bend->next_bend = bend_list;
++ bend_list = bend;
++}
++
++/* Remove from bend_list, Must hold bend_list_mutex */
++static void unlink_bend(struct netback_accel *bend)
++{
++ struct netback_accel *tmp = bend_list;
++ struct netback_accel *prev = NULL;
++ while (tmp != NULL) {
++ if (tmp == bend) {
++ if (prev != NULL)
++ prev->next_bend = bend->next_bend;
++ else
++ bend_list = bend->next_bend;
++ return;
++ }
++ prev = tmp;
++ tmp = tmp->next_bend;
++ }
++}
++
++
++/* Demultiplex a message IRQ from the frontend driver. */
++static irqreturn_t msgirq_from_frontend(int irq, void *context,
++ struct pt_regs *unused)
++{
++ struct xenbus_device *dev = context;
++ struct netback_accel *bend = NETBACK_ACCEL_FROM_XENBUS_DEVICE(dev);
++ VPRINTK("irq %d from device %s\n", irq, dev->nodename);
++ schedule_work(&bend->handle_msg);
++ return IRQ_HANDLED;
++}
++
++
++/*
++ * Demultiplex an IRQ from the frontend driver. This is never used
++ * functionally, but we need it to pass to the bind function, and may
++ * get called spuriously
+ */
-+static int match_accelerator(struct xenbus_device *xendev,
-+ struct backend_info *be,
-+ struct netback_accelerator *accelerator)
++static irqreturn_t netirq_from_frontend(int irq, void *context,
++ struct pt_regs *unused)
+{
-+ int rc = 0;
-+ char *eth_name = xenbus_read(XBT_NIL, xendev->nodename, "accel", NULL);
++ VPRINTK("netirq %d from device %s\n", irq,
++ ((struct xenbus_device *)context)->nodename);
+
-+ if (IS_ERR(eth_name)) {
-+ /* Probably means not present */
-+ DPRINTK("%s: no match due to xenbus_read accel error %d\n",
-+ __FUNCTION__, PTR_ERR(eth_name));
-+ return 0;
-+ } else {
-+ if (!strcmp(eth_name, accelerator->eth_name))
-+ rc = 1;
-+ kfree(eth_name);
-+ return rc;
++ return IRQ_HANDLED;
++}
++
++
++/* Read the limits values of the xenbus structure. */
++static
++void cfg_hw_quotas(struct xenbus_device *dev, struct netback_accel *bend)
++{
++ int err = xenbus_gather
++ (XBT_NIL, dev->nodename,
++ "limits/max-filters", "%d", &bend->quotas.max_filters,
++ "limits/max-buf-pages", "%d", &bend->quotas.max_buf_pages,
++ "limits/max-mcasts", "%d", &bend->quotas.max_mcasts,
++ NULL);
++ if (err) {
++ /*
++ * TODO what if they have previously been set by the
++ * user? This will overwrite with defaults. Maybe
++ * not what we want to do, but useful in startup
++ * case
++ */
++ DPRINTK("Failed to read quotas from xenbus, using defaults\n");
++ bend->quotas.max_filters = NETBACK_ACCEL_DEFAULT_MAX_FILTERS;
++ bend->quotas.max_buf_pages = sfc_netback_max_pages;
++ bend->quotas.max_mcasts = NETBACK_ACCEL_DEFAULT_MAX_MCASTS;
+ }
++
++ return;
+}
+
+
-+static void do_probe(struct backend_info *be,
-+ struct netback_accelerator *accelerator,
-+ struct xenbus_device *xendev)
++static void bend_config_accel_change(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
+{
-+ be->accelerator = accelerator;
-+ atomic_inc(&be->accelerator->use_count);
-+ if (be->accelerator->hooks->probe(xendev) != 0) {
-+ atomic_dec(&be->accelerator->use_count);
-+ module_put(be->accelerator->hooks->owner);
-+ be->accelerator = NULL;
++ struct netback_accel *bend;
++
++ bend = container_of(watch, struct netback_accel, config_accel_watch);
++
++ mutex_lock(&bend->bend_mutex);
++ if (bend->config_accel_watch.node != NULL) {
++ struct xenbus_device *dev =
++ (struct xenbus_device *)bend->hdev_data;
++ DPRINTK("Watch matched, got dev %p otherend %p\n",
++ dev, dev->otherend);
++ if(!xenbus_exists(XBT_NIL, watch->node, "")) {
++ DPRINTK("Ignoring watch as otherend seems invalid\n");
++ goto out;
++ }
++
++ cfg_hw_quotas(dev, bend);
+ }
++ out:
++ mutex_unlock(&bend->bend_mutex);
++ return;
+}
+
+
+/*
-+ * Notify suitable backends that a new accelerator is available and
-+ * connected. This will also notify the accelerator plugin module
-+ * that it is being used for a device through the probe hook.
++ * Setup watch on "limits" in the backend vif info to know when
++ * configuration has been set
+ */
-+static int netback_accelerator_probe_backend(struct device *dev, void *arg)
++static int setup_config_accel_watch(struct xenbus_device *dev,
++ struct netback_accel *bend)
+{
-+ struct netback_accelerator *accelerator =
-+ (struct netback_accelerator *)arg;
-+ struct xenbus_device *xendev = to_xenbus_device(dev);
++ int err;
+
-+ if (!strcmp("vif", xendev->devicetype)) {
-+ struct backend_info *be = xendev->dev.driver_data;
++ VPRINTK("Setting watch on %s/%s\n", dev->nodename, "limits");
+
-+ if (match_accelerator(xendev, be, accelerator) &&
-+ try_module_get(accelerator->hooks->owner)) {
-+ do_probe(be, accelerator, xendev);
-+ }
++ err = xenbus_watch_path2(dev, dev->nodename, "limits",
++ &bend->config_accel_watch,
++ bend_config_accel_change);
++
++ if (err) {
++ EPRINTK("%s: Failed to register xenbus watch: %d\n",
++ __FUNCTION__, err);
++ bend->config_accel_watch.node = NULL;
++ return err;
+ }
+ return 0;
+}
+
+
-+/*
-+ * Notify suitable backends that an accelerator is unavailable.
-+ */
-+static int netback_accelerator_remove_backend(struct device *dev, void *arg)
++static int
++cfg_frontend_info(struct xenbus_device *dev, struct netback_accel *bend,
++ int *grants)
+{
-+ struct xenbus_device *xendev = to_xenbus_device(dev);
-+ struct netback_accelerator *accelerator =
-+ (struct netback_accelerator *)arg;
-+
-+ if (!strcmp("vif", xendev->devicetype)) {
-+ struct backend_info *be = xendev->dev.driver_data;
++ /* Get some info from xenbus on the event channel and shmem grant */
++ int err = xenbus_gather(XBT_NIL, dev->otherend,
++ "accel-msg-channel", "%u", &bend->msg_channel,
++ "accel-ctrl-page", "%d", &(grants[0]),
++ "accel-msg-page", "%d", &(grants[1]),
++ "accel-net-channel", "%u", &bend->net_channel,
++ NULL);
++ if (err)
++ EPRINTK("failed to read event channels or shmem grant: %d\n",
++ err);
++ else
++ DPRINTK("got event chan %d and net chan %d from frontend\n",
++ bend->msg_channel, bend->net_channel);
++ return err;
++}
+
-+ if (be->accelerator == accelerator) {
-+ be->accelerator->hooks->remove(xendev);
-+ atomic_dec(&be->accelerator->use_count);
-+ module_put(be->accelerator->hooks->owner);
-+ be->accelerator = NULL;
++
++/* Setup all the comms needed to chat with the front end driver */
++static int setup_vnic(struct xenbus_device *dev)
++{
++ struct netback_accel *bend;
++ int grants[2], err, msgs_per_queue;
++
++ bend = NETBACK_ACCEL_FROM_XENBUS_DEVICE(dev);
++
++ err = cfg_frontend_info(dev, bend, grants);
++ if (err)
++ goto fail1;
++
++ /*
++ * If we get here, both frontend Connected and configuration
++ * options available. All is well.
++ */
++
++ /* Get the hardware quotas for the VNIC in question. */
++ cfg_hw_quotas(dev, bend);
++
++ /* Set up the deferred work handlers */
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ INIT_WORK(&bend->handle_msg,
++ netback_accel_msg_rx_handler);
++#else
++ INIT_WORK(&bend->handle_msg,
++ netback_accel_msg_rx_handler,
++ (void*)bend);
++#endif
++
++ /* Request the frontend mac */
++ err = net_accel_xen_net_read_mac(dev, bend->mac);
++ if (err)
++ goto fail2;
++
++ /* Set up the shared page. */
++ bend->shared_page = net_accel_map_grants_contig(dev, grants, 2,
++ &bend->sh_pages_unmap);
++
++ if (bend->shared_page == NULL) {
++ EPRINTK("failed to map shared page for %s\n", dev->otherend);
++ err = -ENOMEM;
++ goto fail2;
++ }
++
++ /* Initialise the shared page(s) used for comms */
++ net_accel_msg_init_page(bend->shared_page, PAGE_SIZE,
++ bend->net_dev->flags & IFF_UP);
++
++ msgs_per_queue = (PAGE_SIZE/2) / sizeof(struct net_accel_msg);
++
++ net_accel_msg_init_queue
++ (&bend->to_domU, &bend->shared_page->queue0,
++ (struct net_accel_msg *)((__u8*)bend->shared_page + PAGE_SIZE),
++ msgs_per_queue);
++
++ net_accel_msg_init_queue
++ (&bend->from_domU, &bend->shared_page->queue1,
++ (struct net_accel_msg *)((__u8*)bend->shared_page +
++ (3 * PAGE_SIZE / 2)),
++ msgs_per_queue);
++
++ /* Bind the message event channel to a handler
++ *
++ * Note that we will probably get a spurious interrupt when we
++ * do this, so it must not be done until we have set up
++ * everything we need to handle it.
++ */
++ err = bind_interdomain_evtchn_to_irqhandler(dev->otherend_id,
++ bend->msg_channel,
++ msgirq_from_frontend,
++ 0,
++ "netback_accel",
++ dev);
++ if (err < 0) {
++ EPRINTK("failed to bind event channel: %d\n", err);
++ goto fail3;
++ }
++ else
++ bend->msg_channel_irq = err;
++
++ /* TODO: No need to bind this evtchn to an irq. */
++ err = bind_interdomain_evtchn_to_irqhandler(dev->otherend_id,
++ bend->net_channel,
++ netirq_from_frontend,
++ 0,
++ "netback_accel",
++ dev);
++ if (err < 0) {
++ EPRINTK("failed to bind net channel: %d\n", err);
++ goto fail4;
++ }
++ else
++ bend->net_channel_irq = err;
++
++ /*
++ * Grab ourselves an entry in the forwarding hash table. We do
++ * this now so we don't have the embarassmesnt of sorting out
++ * an allocation failure while at IRQ. Because we pass NULL as
++ * the context, the actual hash lookup will succeed for this
++ * NIC, but the check for somewhere to forward to will
++ * fail. This is necessary to prevent forwarding before
++ * hardware resources are set up
++ */
++ err = netback_accel_fwd_add(bend->mac, NULL, bend->fwd_priv);
++ if (err) {
++ EPRINTK("failed to add to fwd hash table\n");
++ goto fail5;
++ }
++
++ /*
++ * Say hello to frontend. Important to do this straight after
++ * obtaining the message queue as otherwise we are vulnerable
++ * to an evil frontend sending a HELLO-REPLY before we've sent
++ * the HELLO and confusing us
++ */
++ netback_accel_msg_tx_hello(bend, NET_ACCEL_MSG_VERSION);
++ return 0;
++
++ fail5:
++ unbind_from_irqhandler(bend->net_channel_irq, dev);
++ fail4:
++ unbind_from_irqhandler(bend->msg_channel_irq, dev);
++ fail3:
++ net_accel_unmap_grants_contig(dev, bend->sh_pages_unmap);
++ bend->shared_page = NULL;
++ bend->sh_pages_unmap = NULL;
++ fail2:
++ fail1:
++ return err;
++}
++
++
++static int read_nicname(struct xenbus_device *dev, struct netback_accel *bend)
++{
++ int len;
++
++ /* nic name used to select interface used for acceleration */
++ bend->nicname = xenbus_read(XBT_NIL, dev->nodename, "accel", &len);
++ if (IS_ERR(bend->nicname))
++ return PTR_ERR(bend->nicname);
++
++ return 0;
++}
++
++static const char *frontend_name = "sfc_netfront";
++
++static int publish_frontend_name(struct xenbus_device *dev)
++{
++ struct xenbus_transaction tr;
++ int err;
++
++ /* Publish the name of the frontend driver */
++ do {
++ err = xenbus_transaction_start(&tr);
++ if (err != 0) {
++ EPRINTK("%s: transaction start failed\n", __FUNCTION__);
++ return err;
++ }
++ err = xenbus_printf(tr, dev->nodename, "accel-frontend",
++ "%s", frontend_name);
++ if (err != 0) {
++ EPRINTK("%s: xenbus_printf failed\n", __FUNCTION__);
++ xenbus_transaction_end(tr, 1);
++ return err;
+ }
++ err = xenbus_transaction_end(tr, 0);
++ } while (err == -EAGAIN);
++
++ if (err != 0) {
++ EPRINTK("failed to end frontend name transaction\n");
++ return err;
+ }
+ return 0;
+}
+
+
++static int unpublish_frontend_name(struct xenbus_device *dev)
++{
++ struct xenbus_transaction tr;
++ int err;
++
++ do {
++ err = xenbus_transaction_start(&tr);
++ if (err != 0)
++ break;
++ err = xenbus_rm(tr, dev->nodename, "accel-frontend");
++ if (err != 0) {
++ xenbus_transaction_end(tr, 1);
++ break;
++ }
++ err = xenbus_transaction_end(tr, 0);
++ } while (err == -EAGAIN);
++
++ return err;
++}
++
++
++static void cleanup_vnic(struct netback_accel *bend)
++{
++ struct xenbus_device *dev;
++
++ dev = (struct xenbus_device *)bend->hdev_data;
++
++ DPRINTK("%s: bend %p dev %p\n", __FUNCTION__, bend, dev);
++
++ DPRINTK("%s: Remove %p's mac from fwd table...\n",
++ __FUNCTION__, bend);
++ netback_accel_fwd_remove(bend->mac, bend->fwd_priv);
++
++ /* Free buffer table allocations */
++ netback_accel_remove_buffers(bend);
++
++ DPRINTK("%s: Release hardware resources...\n", __FUNCTION__);
++ if (bend->accel_shutdown)
++ bend->accel_shutdown(bend);
++
++ if (bend->net_channel_irq) {
++ unbind_from_irqhandler(bend->net_channel_irq, dev);
++ bend->net_channel_irq = 0;
++ }
++
++ if (bend->msg_channel_irq) {
++ unbind_from_irqhandler(bend->msg_channel_irq, dev);
++ bend->msg_channel_irq = 0;
++ }
++
++ if (bend->sh_pages_unmap) {
++ DPRINTK("%s: Unmap grants %p\n", __FUNCTION__,
++ bend->sh_pages_unmap);
++ net_accel_unmap_grants_contig(dev, bend->sh_pages_unmap);
++ bend->sh_pages_unmap = NULL;
++ bend->shared_page = NULL;
++ }
++}
++
++
++/*************************************************************************/
+
+/*
-+ * Entry point for an netback accelerator plugin module. Called to
-+ * advertise its presence, and connect to any suitable backends.
++ * The following code handles accelstate changes between the frontend
++ * and the backend. It calls setup_vnic and cleanup_vnic in matching
++ * pairs in response to transitions.
++ *
++ * Valid state transitions for Dom0 are as follows:
++ *
++ * Closed->Init on probe or in response to Init from domU
++ * Closed->Closing on error/remove
++ *
++ * Init->Connected in response to Connected from domU
++ * Init->Closing on error/remove or in response to Closing from domU
++ *
++ * Connected->Closing on error/remove or in response to Closing from domU
++ *
++ * Closing->Closed in response to Closed from domU
++ *
+ */
-+int netback_connect_accelerator(unsigned version, int id, const char *eth_name,
-+ struct netback_accel_hooks *hooks)
++
++
++static void netback_accel_frontend_changed(struct xenbus_device *dev,
++ XenbusState frontend_state)
+{
-+ struct netback_accelerator *new_accelerator;
-+ unsigned eth_name_len;
++ struct netback_accel *bend = NETBACK_ACCEL_FROM_XENBUS_DEVICE(dev);
++ XenbusState backend_state;
+
-+ if (version != NETBACK_ACCEL_VERSION) {
-+ if (version > NETBACK_ACCEL_VERSION) {
-+ /* Caller has higher version number, leave it
-+ up to them to decide whether to continue.
-+ They can recall with a lower number if
-+ they're happy to be compatible with us */
-+ return NETBACK_ACCEL_VERSION;
-+ } else {
-+ /* We have a more recent version than caller.
-+ Currently reject, but may in future be able
-+ to be backwardly compatible */
-+ return -EPROTO;
++ DPRINTK("%s: changing from %s to %s. nodename %s, otherend %s\n",
++ __FUNCTION__, xenbus_strstate(bend->frontend_state),
++ xenbus_strstate(frontend_state),dev->nodename, dev->otherend);
++
++ /*
++ * Ignore duplicate state changes. This can happen if the
++ * frontend changes state twice in quick succession and the
++ * first watch fires in the backend after the second
++ * transition has completed.
++ */
++ if (bend->frontend_state == frontend_state)
++ return;
++
++ bend->frontend_state = frontend_state;
++ backend_state = bend->backend_state;
++
++ switch (frontend_state) {
++ case XenbusStateInitialising:
++ if (backend_state == XenbusStateClosed &&
++ !bend->removing)
++ backend_state = XenbusStateInitialising;
++ break;
++
++ case XenbusStateConnected:
++ if (backend_state == XenbusStateInitialising) {
++ if (!bend->vnic_is_setup &&
++ setup_vnic(dev) == 0) {
++ bend->vnic_is_setup = 1;
++ backend_state = XenbusStateConnected;
++ } else {
++ backend_state = XenbusStateClosing;
++ }
++ }
++ break;
++
++ case XenbusStateInitWait:
++ case XenbusStateInitialised:
++ default:
++ DPRINTK("Unknown state %s (%d) from frontend.\n",
++ xenbus_strstate(frontend_state), frontend_state);
++ /* Unknown state. Fall through. */
++ case XenbusStateClosing:
++ if (backend_state != XenbusStateClosed)
++ backend_state = XenbusStateClosing;
++
++ /*
++ * The bend will now persist (with watches active) in
++ * case the frontend comes back again, eg. after
++ * frontend module reload or suspend/resume
++ */
++
++ break;
++
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ if (bend->vnic_is_setup) {
++ bend->vnic_is_setup = 0;
++ cleanup_vnic(bend);
+ }
++
++ if (backend_state == XenbusStateClosing)
++ backend_state = XenbusStateClosed;
++ break;
+ }
+
-+ new_accelerator =
-+ kmalloc(sizeof(struct netback_accelerator), GFP_KERNEL);
-+ if (!new_accelerator) {
-+ DPRINTK("%s: failed to allocate memory for accelerator\n",
-+ __FUNCTION__);
-+ return -ENOMEM;
++ if (backend_state != bend->backend_state) {
++ DPRINTK("Switching from state %s (%d) to %s (%d)\n",
++ xenbus_strstate(bend->backend_state),
++ bend->backend_state,
++ xenbus_strstate(backend_state), backend_state);
++ bend->backend_state = backend_state;
++ net_accel_update_state(dev, backend_state);
+ }
+
-+ new_accelerator->id = id;
-+
-+ eth_name_len = strlen(eth_name)+1;
-+ new_accelerator->eth_name = kmalloc(eth_name_len, GFP_KERNEL);
-+ if (!new_accelerator->eth_name) {
-+ DPRINTK("%s: failed to allocate memory for eth_name string\n",
-+ __FUNCTION__);
-+ kfree(new_accelerator);
++ wake_up(&bend->state_wait_queue);
++}
++
++
++/* accelstate on the frontend's xenbus node has changed */
++static void bend_domu_accel_change(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ int state;
++ struct netback_accel *bend;
++
++ bend = container_of(watch, struct netback_accel, domu_accel_watch);
++ if (bend->domu_accel_watch.node != NULL) {
++ struct xenbus_device *dev =
++ (struct xenbus_device *)bend->hdev_data;
++ VPRINTK("Watch matched, got dev %p otherend %p\n",
++ dev, dev->otherend);
++ /*
++ * dev->otherend != NULL check to protect against
++ * watch firing when domain goes away and we haven't
++ * yet cleaned up
++ */
++ if (!dev->otherend ||
++ !xenbus_exists(XBT_NIL, watch->node, "") ||
++ strncmp(dev->otherend, vec[XS_WATCH_PATH],
++ strlen(dev->otherend))) {
++ DPRINTK("Ignoring watch as otherend seems invalid\n");
++ return;
++ }
++
++ mutex_lock(&bend->bend_mutex);
++
++ xenbus_scanf(XBT_NIL, dev->otherend, "accelstate", "%d",
++ &state);
++ netback_accel_frontend_changed(dev, state);
++
++ mutex_unlock(&bend->bend_mutex);
++ }
++}
++
++/* Setup watch on frontend's accelstate */
++static int setup_domu_accel_watch(struct xenbus_device *dev,
++ struct netback_accel *bend)
++{
++ int err;
++
++ VPRINTK("Setting watch on %s/%s\n", dev->otherend, "accelstate");
++
++ err = xenbus_watch_path2(dev, dev->otherend, "accelstate",
++ &bend->domu_accel_watch,
++ bend_domu_accel_change);
++ if (err) {
++ EPRINTK("%s: Failed to register xenbus watch: %d\n",
++ __FUNCTION__, err);
++ goto fail;
++ }
++ return 0;
++ fail:
++ bend->domu_accel_watch.node = NULL;
++ return err;
++}
++
++
++int netback_accel_probe(struct xenbus_device *dev)
++{
++ struct netback_accel *bend;
++ struct backend_info *binfo;
++ int err;
++
++ DPRINTK("%s: passed device %s\n", __FUNCTION__, dev->nodename);
++
++ /* Allocate structure to store all our state... */
++ bend = kzalloc(sizeof(struct netback_accel), GFP_KERNEL);
++ if (bend == NULL) {
++ DPRINTK("%s: no memory for bend\n", __FUNCTION__);
+ return -ENOMEM;
+ }
-+ strlcpy(new_accelerator->eth_name, eth_name, eth_name_len);
+
-+ new_accelerator->hooks = hooks;
++ mutex_init(&bend->bend_mutex);
+
-+ atomic_set(&new_accelerator->use_count, 0);
++ mutex_lock(&bend->bend_mutex);
++
++ /* ...and store it where we can get at it */
++ binfo = (struct backend_info *) dev->dev.driver_data;
++ binfo->netback_accel_priv = bend;
++ /* And vice-versa */
++ bend->hdev_data = dev;
++
++ DPRINTK("%s: Adding bend %p to list\n", __FUNCTION__, bend);
+
-+ mutex_lock(&accelerators_mutex);
-+ list_add(&new_accelerator->link, &accelerators_list);
++ init_waitqueue_head(&bend->state_wait_queue);
++ bend->vnic_is_setup = 0;
++ bend->frontend_state = XenbusStateUnknown;
++ bend->backend_state = XenbusStateClosed;
++ bend->removing = 0;
++
++ sscanf(dev->nodename, NODENAME_PATH_FMT, &bend->far_end,
++ &bend->vif_num);
++
++ err = read_nicname(dev, bend);
++ if (err) {
++ /*
++ * Technically not an error, just means we're not
++ * supposed to accelerate this
++ */
++ DPRINTK("failed to get device name\n");
++ goto fail_nicname;
++ }
++
++ /*
++ * Look up the device name in the list of NICs provided by
++ * driverlink to get the hardware type.
++ */
++ err = netback_accel_sf_hwtype(bend);
++ if (err) {
++ /*
++ * Technically not an error, just means we're not
++ * supposed to accelerate this, probably belongs to
++ * some other backend
++ */
++ DPRINTK("failed to match device name\n");
++ goto fail_init_type;
++ }
++
++ err = publish_frontend_name(dev);
++ if (err)
++ goto fail_publish;
++
++ err = netback_accel_debugfs_create(bend);
++ if (err)
++ goto fail_debugfs;
+
-+ /* tell existing backends about new plugin */
-+ xenbus_for_each_backend(new_accelerator,
-+ netback_accelerator_probe_backend);
++ mutex_unlock(&bend->bend_mutex);
+
-+ mutex_unlock(&accelerators_mutex);
++ err = setup_config_accel_watch(dev, bend);
++ if (err)
++ goto fail_config_watch;
++
++ err = setup_domu_accel_watch(dev, bend);
++ if (err)
++ goto fail_domu_watch;
++
++ /*
++ * Indicate to the other end that we're ready to start unless
++ * the watch has already fired.
++ */
++ mutex_lock(&bend->bend_mutex);
++ if (bend->backend_state == XenbusStateClosed) {
++ bend->backend_state = XenbusStateInitialising;
++ net_accel_update_state(dev, XenbusStateInitialising);
++ }
++ mutex_unlock(&bend->bend_mutex);
++
++ mutex_lock(&bend_list_mutex);
++ link_bend(bend);
++ mutex_unlock(&bend_list_mutex);
+
+ return 0;
+
++fail_domu_watch:
++
++ unregister_xenbus_watch(&bend->config_accel_watch);
++ kfree(bend->config_accel_watch.node);
++fail_config_watch:
++
++ /*
++ * Flush the scheduled work queue before freeing bend to get
++ * rid of any pending netback_accel_msg_rx_handler()
++ */
++ flush_scheduled_work();
++
++ mutex_lock(&bend->bend_mutex);
++ net_accel_update_state(dev, XenbusStateUnknown);
++ netback_accel_debugfs_remove(bend);
++fail_debugfs:
++
++ unpublish_frontend_name(dev);
++fail_publish:
++
++ /* No need to reverse netback_accel_sf_hwtype. */
++fail_init_type:
++
++ kfree(bend->nicname);
++fail_nicname:
++ binfo->netback_accel_priv = NULL;
++ mutex_unlock(&bend->bend_mutex);
++ kfree(bend);
++ return err;
+}
-+EXPORT_SYMBOL_GPL(netback_connect_accelerator);
+
+
-+/*
-+ * Disconnect an accelerator plugin module that has previously been
-+ * connected.
-+ */
-+void netback_disconnect_accelerator(int id, const char *eth_name)
++int netback_accel_remove(struct xenbus_device *dev)
+{
-+ struct netback_accelerator *accelerator, *next;
++ struct backend_info *binfo;
++ struct netback_accel *bend;
++ int frontend_state;
+
-+ mutex_lock(&accelerators_mutex);
-+ list_for_each_entry_safe(accelerator, next, &accelerators_list, link) {
-+ if (!strcmp(eth_name, accelerator->eth_name)) {
-+ xenbus_for_each_backend
-+ (accelerator, netback_accelerator_remove_backend);
-+ BUG_ON(atomic_read(&accelerator->use_count) != 0);
-+ list_del(&accelerator->link);
-+ kfree(accelerator->eth_name);
-+ kfree(accelerator);
-+ break;
-+ }
++ binfo = (struct backend_info *) dev->dev.driver_data;
++ bend = (struct netback_accel *) binfo->netback_accel_priv;
++
++ DPRINTK("%s: dev %p bend %p\n", __FUNCTION__, dev, bend);
++
++ BUG_ON(bend == NULL);
++
++ mutex_lock(&bend_list_mutex);
++ unlink_bend(bend);
++ mutex_unlock(&bend_list_mutex);
++
++ mutex_lock(&bend->bend_mutex);
++
++ /* Reject any requests to connect. */
++ bend->removing = 1;
++
++ /*
++ * Switch to closing to tell the other end that we're going
++ * away.
++ */
++ if (bend->backend_state != XenbusStateClosing) {
++ bend->backend_state = XenbusStateClosing;
++ net_accel_update_state(dev, XenbusStateClosing);
+ }
-+ mutex_unlock(&accelerators_mutex);
-+}
-+EXPORT_SYMBOL_GPL(netback_disconnect_accelerator);
+
++ frontend_state = (int)XenbusStateUnknown;
++ xenbus_scanf(XBT_NIL, dev->otherend, "accelstate", "%d",
++ &frontend_state);
+
-+void netback_probe_accelerators(struct backend_info *be,
-+ struct xenbus_device *dev)
-+{
-+ struct netback_accelerator *accelerator;
++ mutex_unlock(&bend->bend_mutex);
+
-+ /*
-+ * Check list of accelerators to see if any is suitable, and
-+ * use it if it is.
++ /*
++ * Wait until this end goes to the closed state. This happens
++ * in response to the other end going to the closed state.
++ * Don't bother doing this if the other end is already closed
++ * because if it is then there is nothing to do.
+ */
-+ mutex_lock(&accelerators_mutex);
-+ list_for_each_entry(accelerator, &accelerators_list, link) {
-+ if (match_accelerator(dev, be, accelerator) &&
-+ try_module_get(accelerator->hooks->owner)) {
-+ do_probe(be, accelerator, dev);
-+ break;
-+ }
++ if (frontend_state != (int)XenbusStateClosed &&
++ frontend_state != (int)XenbusStateUnknown)
++ wait_event(bend->state_wait_queue,
++ bend->backend_state == XenbusStateClosed);
++
++ unregister_xenbus_watch(&bend->domu_accel_watch);
++ kfree(bend->domu_accel_watch.node);
++
++ unregister_xenbus_watch(&bend->config_accel_watch);
++ kfree(bend->config_accel_watch.node);
++
++ /*
++ * Flush the scheduled work queue before freeing bend to get
++ * rid of any pending netback_accel_msg_rx_handler()
++ */
++ flush_scheduled_work();
++
++ mutex_lock(&bend->bend_mutex);
++
++ /* Tear down the vnic if it was set up. */
++ if (bend->vnic_is_setup) {
++ bend->vnic_is_setup = 0;
++ cleanup_vnic(bend);
+ }
-+ mutex_unlock(&accelerators_mutex);
++
++ bend->backend_state = XenbusStateUnknown;
++ net_accel_update_state(dev, XenbusStateUnknown);
++
++ netback_accel_debugfs_remove(bend);
++
++ unpublish_frontend_name(dev);
++
++ kfree(bend->nicname);
++
++ binfo->netback_accel_priv = NULL;
++
++ mutex_unlock(&bend->bend_mutex);
++
++ kfree(bend);
++
++ return 0;
+}
+
+
-+void netback_remove_accelerators(struct backend_info *be,
-+ struct xenbus_device *dev)
++void netback_accel_shutdown_bends(void)
+{
-+ mutex_lock(&accelerators_mutex);
-+ /* Notify the accelerator (if any) of this device's removal */
-+ if (be->accelerator != NULL) {
-+ be->accelerator->hooks->remove(dev);
-+ atomic_dec(&be->accelerator->use_count);
-+ module_put(be->accelerator->hooks->owner);
-+ be->accelerator = NULL;
-+ }
-+ mutex_unlock(&accelerators_mutex);
++ mutex_lock(&bend_list_mutex);
++ /*
++ * I think we should have had a remove callback for all
++ * interfaces before being allowed to unload the module
++ */
++ BUG_ON(bend_list != NULL);
++ mutex_unlock(&bend_list_mutex);
+}
+
+
-+void netif_accel_init(void)
++void netback_accel_set_closing(struct netback_accel *bend)
+{
-+ INIT_LIST_HEAD(&accelerators_list);
++
++ bend->backend_state = XenbusStateClosing;
++ net_accel_update_state((struct xenbus_device *)bend->hdev_data,
++ XenbusStateClosing);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/common.h linux-2.6.18-xen.hg/drivers/xen/netback/common.h
---- linux-2.6.18/drivers/xen/netback/common.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/common.h 2007-12-23 11:15:34.051268764 +0100
-@@ -0,0 +1,217 @@
-+/******************************************************************************
-+ * arch/xen/drivers/netif/backend/common.h
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/gcc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/gcc.h 2008-05-19 00:33:47.466882166 +0300
+@@ -0,0 +1,158 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#ifndef __NETIF__BACKEND__COMMON_H__
-+#define __NETIF__BACKEND__COMMON_H__
++/*! \cidoxg_include_ci_compat */
+
-+#include <linux/version.h>
-+#include <linux/module.h>
-+#include <linux/interrupt.h>
-+#include <linux/slab.h>
-+#include <linux/ip.h>
-+#include <linux/in.h>
-+#include <linux/netdevice.h>
-+#include <linux/etherdevice.h>
-+#include <linux/wait.h>
-+#include <xen/evtchn.h>
-+#include <xen/interface/io/netif.h>
-+#include <asm/io.h>
-+#include <asm/pgalloc.h>
-+#include <xen/interface/grant_table.h>
-+#include <xen/gnttab.h>
-+#include <xen/driver_util.h>
-+#include <xen/xenbus.h>
++#ifndef __CI_COMPAT_GCC_H__
++#define __CI_COMPAT_GCC_H__
+
-+#define DPRINTK(_f, _a...) \
-+ pr_debug("(file=%s, line=%d) " _f, \
-+ __FILE__ , __LINE__ , ## _a )
-+#define IPRINTK(fmt, args...) \
-+ printk(KERN_INFO "xen_net: " fmt, ##args)
-+#define WPRINTK(fmt, args...) \
-+ printk(KERN_WARNING "xen_net: " fmt, ##args)
+
-+typedef struct netif_st {
-+ /* Unique identifier for this interface. */
-+ domid_t domid;
-+ unsigned int handle;
++#define CI_HAVE_INT64
+
-+ u8 fe_dev_addr[6];
+
-+ /* Physical parameters of the comms window. */
-+ grant_handle_t tx_shmem_handle;
-+ grant_ref_t tx_shmem_ref;
-+ grant_handle_t rx_shmem_handle;
-+ grant_ref_t rx_shmem_ref;
-+ unsigned int irq;
++#if defined(__linux__) && defined(__KERNEL__)
+
-+ /* The shared rings and indexes. */
-+ netif_tx_back_ring_t tx;
-+ netif_rx_back_ring_t rx;
-+ struct vm_struct *tx_comms_area;
-+ struct vm_struct *rx_comms_area;
++# include <linux/types.h>
+
-+ /* Set of features that can be turned on in dev->features. */
-+ int features;
++typedef __u64 ci_uint64;
++typedef __s64 ci_int64;
++# if BITS_PER_LONG == 32
++typedef __s32 ci_ptr_arith_t;
++typedef __u32 ci_uintptr_t;
++# else
++typedef __s64 ci_ptr_arith_t;
++typedef __u64 ci_uintptr_t;
++# endif
+
-+ /* Internal feature information. */
-+ u8 can_queue:1; /* can queue packets for receiver? */
-+ u8 copying_receiver:1; /* copy packets to receiver? */
+
-+ /* Allow netif_be_start_xmit() to peek ahead in the rx request ring. */
-+ RING_IDX rx_req_cons_peek;
++/* it's not obvious to me why the below is wrong for x64_64, but
++ * gcc seems to complain on this platform
++ */
++# if defined(__ia64__)
++# define CI_PRId64 "ld"
++# define CI_PRIi64 "li"
++# define CI_PRIo64 "lo"
++# define CI_PRIu64 "lu"
++# define CI_PRIx64 "lx"
++# define CI_PRIX64 "lX"
++# else
++# define CI_PRId64 "lld"
++# define CI_PRIi64 "lli"
++# define CI_PRIo64 "llo"
++# define CI_PRIu64 "llu"
++# define CI_PRIx64 "llx"
++# define CI_PRIX64 "llX"
++# endif
+
-+ /* Transmit shaping: allow 'credit_bytes' every 'credit_usec'. */
-+ unsigned long credit_bytes;
-+ unsigned long credit_usec;
-+ unsigned long remaining_credit;
-+ struct timer_list credit_timeout;
++# define CI_PRId32 "d"
++# define CI_PRIi32 "i"
++# define CI_PRIo32 "o"
++# define CI_PRIu32 "u"
++# define CI_PRIx32 "x"
++# define CI_PRIX32 "X"
+
-+ /* Enforce draining of the transmit queue. */
-+ struct timer_list tx_queue_timeout;
++#else
+
-+ /* Miscellaneous private stuff. */
-+ struct list_head list; /* scheduling list */
-+ atomic_t refcnt;
-+ struct net_device *dev;
-+ struct net_device_stats stats;
++# include <stdint.h>
++# include <inttypes.h>
+
-+ unsigned int carrier;
++typedef uint64_t ci_uint64;
++typedef int64_t ci_int64;
++typedef intptr_t ci_ptr_arith_t;
++typedef uintptr_t ci_uintptr_t;
+
-+ wait_queue_head_t waiting_to_free;
-+} netif_t;
++# define CI_PRId64 PRId64
++# define CI_PRIi64 PRIi64
++# define CI_PRIo64 PRIo64
++# define CI_PRIu64 PRIu64
++# define CI_PRIx64 PRIx64
++# define CI_PRIX64 PRIX64
++
++# define CI_PRId32 PRId32
++# define CI_PRIi32 PRIi32
++# define CI_PRIo32 PRIo32
++# define CI_PRIu32 PRIu32
++# define CI_PRIx32 PRIx32
++# define CI_PRIX32 PRIX32
++
++#endif
++
++
++typedef ci_uint64 ci_fixed_descriptor_t;
++
++#define from_fixed_descriptor(desc) ((ci_uintptr_t)(desc))
++#define to_fixed_descriptor(desc) ((ci_fixed_descriptor_t)(ci_uintptr_t)(desc))
+
++
++#if __GNUC__ >= 3 && !defined(__cplusplus)
+/*
-+ * Implement our own carrier flag: the network stack's version causes delays
-+ * when the carrier is re-enabled (in particular, dev_activate() may not
-+ * immediately be called, which can cause packet loss; also the etherbridge
-+ * can be rather lazy in activating its port).
++** Checks that [p_mbr] has the same type as [&c_type::mbr_name].
++*/
++# define CI_CONTAINER(c_type, mbr_name, p_mbr) \
++ __builtin_choose_expr( \
++ __builtin_types_compatible_p(__typeof__(&((c_type*)0)->mbr_name), \
++ __typeof__(p_mbr)), \
++ __CI_CONTAINER(c_type, mbr_name, p_mbr), (void)0)
++
++# define ci_restrict __restrict__
++#endif
++
++
++#if !defined(__KERNEL__) || defined(__unix__)
++#define CI_HAVE_NPRINTF 1
++#endif
++
++
++/* At what version was this introduced? */
++#if __GNUC__ >= 3 || (__GNUC__ == 2 && __GNUC_MINOR__ > 91)
++# define CI_LIKELY(t) __builtin_expect((t), 1)
++# define CI_UNLIKELY(t) __builtin_expect((t), 0)
++#endif
++
++/**********************************************************************
++ * Attributes
+ */
-+#define netback_carrier_on(netif) ((netif)->carrier = 1)
-+#define netback_carrier_off(netif) ((netif)->carrier = 0)
-+#define netback_carrier_ok(netif) ((netif)->carrier)
++#if __GNUC__ >= 3 && defined(NDEBUG)
++# define CI_HF __attribute__((visibility("hidden")))
++# define CI_HV __attribute__((visibility("hidden")))
++#else
++# define CI_HF
++# define CI_HV
++#endif
+
-+enum {
-+ NETBK_DONT_COPY_SKB,
-+ NETBK_DELAYED_COPY_SKB,
-+ NETBK_ALWAYS_COPY_SKB,
-+};
++#if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
++# define ci_noinline static __attribute__((__noinline__))
++/* (Linux 2.6 defines its own "noinline", so we use the "__noinline__" form) */
++#else
++# define ci_noinline static
++#endif
+
-+extern int netbk_copy_skb_mode;
++#define CI_ALIGN(x) __attribute__ ((aligned (x)))
+
-+/* Function pointers into netback accelerator plugin modules */
-+struct netback_accel_hooks {
-+ struct module *owner;
-+ int (*probe)(struct xenbus_device *dev);
-+ int (*remove)(struct xenbus_device *dev);
-+};
++#define CI_PRINTF_LIKE(a,b) __attribute__((format(printf,a,b)))
+
-+/* Structure to track the state of a netback accelerator plugin */
-+struct netback_accelerator {
-+ struct list_head link;
-+ int id;
-+ char *eth_name;
-+ atomic_t use_count;
-+ struct netback_accel_hooks *hooks;
-+};
++#endif /* __CI_COMPAT_GCC_H__ */
+
-+struct backend_info {
-+ struct xenbus_device *dev;
-+ netif_t *netif;
-+ enum xenbus_state frontend_state;
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/gcc_x86.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/gcc_x86.h 2008-05-19 00:33:47.538886316 +0300
+@@ -0,0 +1,115 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* State relating to the netback accelerator */
-+ void *netback_accel_priv;
-+ /* The accelerator that this backend is currently using */
-+ struct netback_accelerator *accelerator;
-+};
++/*! \cidoxg_include_ci_compat */
+
-+#define NETBACK_ACCEL_VERSION 0x00010001
++#ifndef __CI_COMPAT_GCC_X86_H__
++#define __CI_COMPAT_GCC_X86_H__
++
++/*
++** The facts:
++**
++** SSE sfence
++** SSE2 lfence, mfence, pause
++*/
+
+/*
-+ * Connect an accelerator plugin module to netback. Returns zero on
-+ * success, < 0 on error, > 0 (with highest version number supported)
-+ * if version mismatch.
++ Barriers to enforce ordering with respect to:
++
++ normal memory use: ci_wmb, ci_rmb, ci_wmb
++ IO bus access use: ci_wiob, ci_riob, ci_iob
++*/
++#if defined(__x86_64__)
++# define ci_x86_mb() __asm__ __volatile__ ("lock; addl $0,0(%%rsp)":::"memory")
++#else
++# define ci_x86_mb() __asm__ __volatile__ ("lock; addl $0,0(%%esp)":::"memory")
++#endif
++
++/* ?? measure the impact of latency of sfence on a modern processor before we
++ take a decision on how to integrate with respect to writecombining */
++
++/* DJR: I don't think we need to add "memory" here. It means the asm does
++** something to memory that GCC doesn't understand. But all this does is
++** commit changes that GCC thinks have already happened. NB. GCC will not
++** reorder across a __volatile__ __asm__ anyway.
++*/
++#define ci_gcc_fence() __asm__ __volatile__ ("")
++
++#if __GNUC__ >= 3 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 96)
++# define ci_x86_sfence() __asm__ __volatile__ ("sfence")
++# define ci_x86_lfence() __asm__ __volatile__ ("lfence")
++# define ci_x86_mfence() __asm__ __volatile__ ("mfence")
++#else
++# define ci_x86_sfence() __asm__ __volatile__ (".byte 0x0F, 0xAE, 0xF8")
++# define ci_x86_lfence() __asm__ __volatile__ (".byte 0x0F, 0xAE, 0xE8")
++# define ci_x86_mfence() __asm__ __volatile__ (".byte 0x0F, 0xAE, 0xF0")
++#endif
++
++
++/* x86 processors to P4 Xeon store in-order unless executing streaming
++ extensions or when using writecombining
++
++ Hence we do not define ci_wmb to use sfence by default. Requirement is that
++ we do not use writecombining to memory and any code which uses SSE
++ extensions must call sfence directly
++
++ We need to track non intel clones which may support out of order store.
++
++*/
++
++#if CI_CPU_OOS
++# if CI_CPU_HAS_SSE
++# define ci_wmb() ci_x86_sfence()
++# else
++# define ci_wmb() ci_x86_mb()
++# endif
++#else
++# define ci_wmb() ci_gcc_fence()
++#endif
++
++#if CI_CPU_HAS_SSE2
++# define ci_rmb() ci_x86_lfence()
++# define ci_mb() ci_x86_mfence()
++# define ci_riob() ci_x86_lfence()
++# define ci_wiob() ci_x86_sfence()
++# define ci_iob() ci_x86_mfence()
++#else
++# if CI_CPU_HAS_SSE
++# define ci_wiob() ci_x86_sfence()
++# else
++# define ci_wiob() ci_x86_mb()
++# endif
++# define ci_rmb() ci_x86_mb()
++# define ci_mb() ci_x86_mb()
++# define ci_riob() ci_x86_mb()
++# define ci_iob() ci_x86_mb()
++#endif
++
++typedef unsigned long ci_phys_addr_t;
++#define ci_phys_addr_fmt "%lx"
++
++#endif /* __CI_COMPAT_GCC_X86_H__ */
++
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/primitive.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/primitive.h 2008-05-19 00:33:47.562887700 +0300
+@@ -0,0 +1,77 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+extern int netback_connect_accelerator(unsigned version,
-+ int id, const char *eth_name,
-+ struct netback_accel_hooks *hooks);
-+/* Disconnect a previously connected accelerator plugin module */
-+extern void netback_disconnect_accelerator(int id, const char *eth_name);
++/*! \cidoxg_include_ci_compat */
+
++#ifndef __CI_COMPAT_PRIMITIVE_H__
++#define __CI_COMPAT_PRIMITIVE_H__
+
-+extern
-+void netback_probe_accelerators(struct backend_info *be,
-+ struct xenbus_device *dev);
-+extern
-+void netback_remove_accelerators(struct backend_info *be,
-+ struct xenbus_device *dev);
-+extern
-+void netif_accel_init(void);
+
++/**********************************************************************
++ * Primitive types.
++ */
+
-+#define NET_TX_RING_SIZE __RING_SIZE((netif_tx_sring_t *)0, PAGE_SIZE)
-+#define NET_RX_RING_SIZE __RING_SIZE((netif_rx_sring_t *)0, PAGE_SIZE)
++typedef unsigned char ci_uint8;
++typedef char ci_int8;
+
-+void netif_disconnect(netif_t *netif);
++typedef unsigned short ci_uint16;
++typedef short ci_int16;
+
-+netif_t *netif_alloc(domid_t domid, unsigned int handle);
-+int netif_map(netif_t *netif, unsigned long tx_ring_ref,
-+ unsigned long rx_ring_ref, unsigned int evtchn);
++typedef unsigned int ci_uint32;
++typedef int ci_int32;
+
-+#define netif_get(_b) (atomic_inc(&(_b)->refcnt))
-+#define netif_put(_b) \
-+ do { \
-+ if ( atomic_dec_and_test(&(_b)->refcnt) ) \
-+ wake_up(&(_b)->waiting_to_free); \
-+ } while (0)
++/* 64-bit support is platform dependent. */
+
-+void netif_xenbus_init(void);
+
-+#define netif_schedulable(netif) \
-+ (netif_running((netif)->dev) && netback_carrier_ok(netif))
++/**********************************************************************
++ * Other fancy types.
++ */
+
-+void netif_schedule_work(netif_t *netif);
-+void netif_deschedule_work(netif_t *netif);
++typedef ci_uint8 ci_octet;
+
-+int netif_be_start_xmit(struct sk_buff *skb, struct net_device *dev);
-+struct net_device_stats *netif_be_get_stats(struct net_device *dev);
-+irqreturn_t netif_be_int(int irq, void *dev_id, struct pt_regs *regs);
++typedef enum {
++ CI_FALSE = 0,
++ CI_TRUE
++} ci_boolean_t;
+
-+static inline int netbk_can_queue(struct net_device *dev)
-+{
-+ netif_t *netif = netdev_priv(dev);
-+ return netif->can_queue;
-+}
+
-+static inline int netbk_can_sg(struct net_device *dev)
-+{
-+ netif_t *netif = netdev_priv(dev);
-+ return netif->features & NETIF_F_SG;
-+}
++/**********************************************************************
++ * Some nice types you'd always assumed were standards.
++ * (Really, they are SYSV "standards".)
++ */
+
-+#endif /* __NETIF__BACKEND__COMMON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/interface.c linux-2.6.18-xen.hg/drivers/xen/netback/interface.c
---- linux-2.6.18/drivers/xen/netback/interface.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/interface.c 2007-12-23 11:15:34.054602273 +0100
-@@ -0,0 +1,336 @@
-+/******************************************************************************
-+ * arch/xen/drivers/netif/backend/interface.c
-+ *
-+ * Network-device interface management.
-+ *
-+ * Copyright (c) 2004-2005, Keir Fraser
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++#ifdef _WIN32
++typedef unsigned long ulong;
++typedef unsigned int uint;
++typedef char* caddr_t;
++#elif defined(__linux__) && defined(__KERNEL__)
++#include <linux/types.h>
++#elif defined(__linux__)
++#include <sys/types.h>
++#endif
++
++
++#endif /* __CI_COMPAT_PRIMITIVE_H__ */
++
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/sysdep.h 2008-05-19 00:33:47.666893695 +0300
+@@ -0,0 +1,166 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include "common.h"
-+#include <linux/ethtool.h>
-+#include <linux/rtnetlink.h>
++/*! \cidoxg_include_ci_compat */
+
-+/*
-+ * Module parameter 'queue_length':
-+ *
-+ * Enables queuing in the network stack when a client has run out of receive
-+ * descriptors. Although this feature can improve receive bandwidth by avoiding
-+ * packet loss, it can also result in packets sitting in the 'tx_queue' for
-+ * unbounded time. This is bad if those packets hold onto foreign resources.
-+ * For example, consider a packet that holds onto resources belonging to the
-+ * guest for which it is queued (e.g., packet received on vif1.0, destined for
-+ * vif1.1 which is not activated in the guest): in this situation the guest
-+ * will never be destroyed, unless vif1.1 is taken down. To avoid this, we
-+ * run a timer (tx_queue_timeout) to drain the queue when the interface is
-+ * blocked.
++#ifndef __CI_COMPAT_SYSDEP_H__
++#define __CI_COMPAT_SYSDEP_H__
++
++
++/**********************************************************************
++ * Platform definition fixups.
+ */
-+static unsigned long netbk_queue_length = 32;
-+module_param_named(queue_length, netbk_queue_length, ulong, 0);
+
-+static void __netif_up(netif_t *netif)
-+{
-+ enable_irq(netif->irq);
-+ netif_schedule_work(netif);
-+}
++#if defined(__ci_ul_driver__) && !defined(__ci_driver__)
++# define __ci_driver__
++#endif
+
-+static void __netif_down(netif_t *netif)
-+{
-+ disable_irq(netif->irq);
-+ netif_deschedule_work(netif);
-+}
++#if defined(__ci_driver__) && !defined(__ci_ul_driver__) && \
++ !defined(__KERNEL__)
++# define __KERNEL__
++#endif
+
-+static int net_open(struct net_device *dev)
-+{
-+ netif_t *netif = netdev_priv(dev);
-+ if (netback_carrier_ok(netif)) {
-+ __netif_up(netif);
-+ netif_start_queue(dev);
-+ }
-+ return 0;
-+}
+
-+static int net_close(struct net_device *dev)
-+{
-+ netif_t *netif = netdev_priv(dev);
-+ if (netback_carrier_ok(netif))
-+ __netif_down(netif);
-+ netif_stop_queue(dev);
-+ return 0;
-+}
++/**********************************************************************
++ * Sanity checks (no cheating!)
++ */
+
-+static int netbk_change_mtu(struct net_device *dev, int mtu)
-+{
-+ int max = netbk_can_sg(dev) ? 65535 - ETH_HLEN : ETH_DATA_LEN;
++#if defined(__KERNEL__) && !defined(__ci_driver__)
++# error Insane.
++#endif
+
-+ if (mtu > max)
-+ return -EINVAL;
-+ dev->mtu = mtu;
-+ return 0;
-+}
++#if defined(__KERNEL__) && defined(__ci_ul_driver__)
++# error Madness.
++#endif
+
-+static int netbk_set_sg(struct net_device *dev, u32 data)
-+{
-+ if (data) {
-+ netif_t *netif = netdev_priv(dev);
++#if defined(__unix__) && defined(_WIN32)
++# error Strange.
++#endif
+
-+ if (!(netif->features & NETIF_F_SG))
-+ return -ENOSYS;
-+ }
++#if defined(__GNUC__) && defined(_MSC_VER)
++# error Crazy.
++#endif
+
-+ return ethtool_op_set_sg(dev, data);
-+}
+
-+static int netbk_set_tso(struct net_device *dev, u32 data)
-+{
-+ if (data) {
-+ netif_t *netif = netdev_priv(dev);
++/**********************************************************************
++ * Compiler and processor dependencies.
++ */
+
-+ if (!(netif->features & NETIF_F_TSO))
-+ return -ENOSYS;
-+ }
++#if defined(__GNUC__)
+
-+ return ethtool_op_set_tso(dev, data);
-+}
++# include <ci/compat/gcc.h>
++
++# if defined(__i386__)
++# include <ci/compat/x86.h>
++# include <ci/compat/gcc_x86.h>
++# elif defined(__x86_64__)
++# include <ci/compat/x86_64.h>
++# include <ci/compat/gcc_x86.h>
++# elif defined(__PPC__)
++# include <ci/compat/ppc.h>
++# include <ci/compat/gcc_ppc.h>
++# elif defined(__ia64__)
++# include <ci/compat/ia64.h>
++# include <ci/compat/gcc_ia64.h>
++# else
++# error Unknown processor - GNU C
++# endif
+
-+static struct ethtool_ops network_ethtool_ops =
-+{
-+ .get_tx_csum = ethtool_op_get_tx_csum,
-+ .set_tx_csum = ethtool_op_set_tx_csum,
-+ .get_sg = ethtool_op_get_sg,
-+ .set_sg = netbk_set_sg,
-+ .get_tso = ethtool_op_get_tso,
-+ .set_tso = netbk_set_tso,
-+ .get_link = ethtool_op_get_link,
-+};
++#elif defined(_MSC_VER)
+
-+netif_t *netif_alloc(domid_t domid, unsigned int handle)
-+{
-+ int err = 0;
-+ struct net_device *dev;
-+ netif_t *netif;
-+ char name[IFNAMSIZ] = {};
++# include <ci/compat/msvc.h>
+
-+ snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle);
-+ dev = alloc_netdev(sizeof(netif_t), name, ether_setup);
-+ if (dev == NULL) {
-+ DPRINTK("Could not create netif: out of memory\n");
-+ return ERR_PTR(-ENOMEM);
-+ }
++# if defined(__i386__)
++# include <ci/compat/x86.h>
++# include <ci/compat/msvc_x86.h>
++# elif defined(__x86_64__)
++# include <ci/compat/x86_64.h>
++# include <ci/compat/msvc_x86_64.h>
++# else
++# error Unknown processor MSC
++# endif
+
-+ netif = netdev_priv(dev);
-+ memset(netif, 0, sizeof(*netif));
-+ netif->domid = domid;
-+ netif->handle = handle;
-+ atomic_set(&netif->refcnt, 1);
-+ init_waitqueue_head(&netif->waiting_to_free);
-+ netif->dev = dev;
++#elif defined(__PGI)
++
++# include <ci/compat/x86.h>
++# include <ci/compat/pg_x86.h>
++
++#elif defined(__INTEL_COMPILER)
++
++/* Intel compilers v7 claim to be very gcc compatible. */
++# if __INTEL_COMPILER >= 700
++# include <ci/compat/gcc.h>
++# include <ci/compat/x86.h>
++# include <ci/compat/gcc_x86.h>
++# else
++# error Old Intel compiler not supported. Yet.
++# endif
++
++#else
++# error Unknown compiler.
++#endif
++
++
++/**********************************************************************
++ * Misc stuff (that probably shouldn't be here).
++ */
++
++#ifdef __sun
++# ifdef __KERNEL__
++# define _KERNEL
++# define _SYSCALL32
++# ifdef _LP64
++# define _SYSCALL32_IMPL
++# endif
++# else
++# define _REENTRANT
++# endif
++#endif
++
++
++/**********************************************************************
++ * Defaults for anything left undefined.
++ */
++
++#ifndef CI_LIKELY
++# define CI_LIKELY(t) (t)
++# define CI_UNLIKELY(t) (t)
++#endif
++
++#ifndef ci_restrict
++# define ci_restrict
++#endif
++
++#ifndef ci_inline
++# define ci_inline static inline
++#endif
++
++#ifndef ci_noinline
++# define ci_noinline static
++#endif
++
++#endif /* __CI_COMPAT_SYSDEP_H__ */
++
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/utils.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/utils.h 2008-05-19 00:33:47.670893925 +0300
+@@ -0,0 +1,269 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*
++ * \author djr
++ * \brief Handy utility macros.
++ * \date 2003/01/17
++ */
++
++/*! \cidoxg_include_ci_compat */
++
++#ifndef __CI_COMPAT_UTILS_H__
++#define __CI_COMPAT_UTILS_H__
+
-+ netback_carrier_off(netif);
+
-+ netif->credit_bytes = netif->remaining_credit = ~0UL;
-+ netif->credit_usec = 0UL;
-+ init_timer(&netif->credit_timeout);
-+ /* Initialize 'expires' now: it's used to track the credit window. */
-+ netif->credit_timeout.expires = jiffies;
++/**********************************************************************
++ * Alignment -- [align] must be a power of 2.
++ **********************************************************************/
+
-+ init_timer(&netif->tx_queue_timeout);
++ /*! Align forward onto next boundary. */
+
-+ dev->hard_start_xmit = netif_be_start_xmit;
-+ dev->get_stats = netif_be_get_stats;
-+ dev->open = net_open;
-+ dev->stop = net_close;
-+ dev->change_mtu = netbk_change_mtu;
-+ dev->features = NETIF_F_IP_CSUM;
++#define CI_ALIGN_FWD(p, align) (((p)+(align)-1u) & ~((align)-1u))
+
-+ SET_ETHTOOL_OPS(dev, &network_ethtool_ops);
+
-+ dev->tx_queue_len = netbk_queue_length;
++ /*! Align back onto prev boundary. */
+
-+ /*
-+ * Initialise a dummy MAC address. We choose the numerically
-+ * largest non-broadcast address to prevent the address getting
-+ * stolen by an Ethernet bridge for STP purposes.
-+ * (FE:FF:FF:FF:FF:FF)
-+ */
-+ memset(dev->dev_addr, 0xFF, ETH_ALEN);
-+ dev->dev_addr[0] &= ~0x01;
++#define CI_ALIGN_BACK(p, align) ((p) & ~((align)-1u))
+
-+ rtnl_lock();
-+ err = register_netdevice(dev);
-+ rtnl_unlock();
-+ if (err) {
-+ DPRINTK("Could not register new net device %s: err=%d\n",
-+ dev->name, err);
-+ free_netdev(dev);
-+ return ERR_PTR(err);
-+ }
+
-+ DPRINTK("Successfully created netif\n");
-+ return netif;
-+}
++ /*! How far to next boundary? */
+
-+static int map_frontend_pages(
-+ netif_t *netif, grant_ref_t tx_ring_ref, grant_ref_t rx_ring_ref)
-+{
-+ struct gnttab_map_grant_ref op;
++#define CI_ALIGN_NEEDED(p, align, signed_t) (-(signed_t)(p) & ((align)-1u))
+
-+ gnttab_set_map_op(&op, (unsigned long)netif->tx_comms_area->addr,
-+ GNTMAP_host_map, tx_ring_ref, netif->domid);
-+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
-+ BUG();
+
-+ if (op.status) {
-+ DPRINTK(" Gnttab failure mapping tx_ring_ref!\n");
-+ return op.status;
-+ }
++ /*! How far beyond prev boundary? */
+
-+ netif->tx_shmem_ref = tx_ring_ref;
-+ netif->tx_shmem_handle = op.handle;
++#define CI_OFFSET(p, align) ((p) & ((align)-1u))
+
-+ gnttab_set_map_op(&op, (unsigned long)netif->rx_comms_area->addr,
-+ GNTMAP_host_map, rx_ring_ref, netif->domid);
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
-+ BUG();
++ /*! Does object fit in gap before next boundary? */
+
-+ if (op.status) {
-+ DPRINTK(" Gnttab failure mapping rx_ring_ref!\n");
-+ return op.status;
-+ }
++#define CI_FITS(p, size, align, signed_t) \
++ (CI_ALIGN_NEEDED((p) + 1, (align), signed_t) + 1 >= (size))
+
-+ netif->rx_shmem_ref = rx_ring_ref;
-+ netif->rx_shmem_handle = op.handle;
+
-+ return 0;
-+}
++ /*! Align forward onto next boundary. */
+
-+static void unmap_frontend_pages(netif_t *netif)
-+{
-+ struct gnttab_unmap_grant_ref op;
++#define CI_PTR_ALIGN_FWD(p, align) \
++ ((char*) CI_ALIGN_FWD(((ci_ptr_arith_t)(p)), ((ci_ptr_arith_t)(align))))
+
-+ gnttab_set_unmap_op(&op, (unsigned long)netif->tx_comms_area->addr,
-+ GNTMAP_host_map, netif->tx_shmem_handle);
++ /*! Align back onto prev boundary. */
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
-+ BUG();
++#define CI_PTR_ALIGN_BACK(p, align) \
++ ((char*) CI_ALIGN_BACK(((ci_ptr_arith_t)(p)), ((ci_ptr_arith_t)(align))))
+
-+ gnttab_set_unmap_op(&op, (unsigned long)netif->rx_comms_area->addr,
-+ GNTMAP_host_map, netif->rx_shmem_handle);
++ /*! How far to next boundary? */
+
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
-+ BUG();
-+}
++#define CI_PTR_ALIGN_NEEDED(p, align) \
++ CI_ALIGN_NEEDED(((ci_ptr_arith_t)(p)), ((ci_ptr_arith_t)(align)), \
++ ci_ptr_arith_t)
+
-+int netif_map(netif_t *netif, unsigned long tx_ring_ref,
-+ unsigned long rx_ring_ref, unsigned int evtchn)
-+{
-+ int err = -ENOMEM;
-+ netif_tx_sring_t *txs;
-+ netif_rx_sring_t *rxs;
++ /*! How far to next boundary? NZ = not zero i.e. give align if on boundary */
+
-+ /* Already connected through? */
-+ if (netif->irq)
-+ return 0;
++#define CI_PTR_ALIGN_NEEDED_NZ(p, align) \
++ ((align) - (((char*)p) - \
++ ((char*) CI_ALIGN_BACK(((ci_ptr_arith_t)(p)), ((ci_ptr_arith_t)(align))))))
+
-+ netif->tx_comms_area = alloc_vm_area(PAGE_SIZE);
-+ if (netif->tx_comms_area == NULL)
-+ return -ENOMEM;
-+ netif->rx_comms_area = alloc_vm_area(PAGE_SIZE);
-+ if (netif->rx_comms_area == NULL)
-+ goto err_rx;
++ /*! How far beyond prev boundary? */
+
-+ err = map_frontend_pages(netif, tx_ring_ref, rx_ring_ref);
-+ if (err)
-+ goto err_map;
++#define CI_PTR_OFFSET(p, align) \
++ CI_OFFSET(((ci_ptr_arith_t)(p)), ((ci_ptr_arith_t)(align)))
+
-+ err = bind_interdomain_evtchn_to_irqhandler(
-+ netif->domid, evtchn, netif_be_int, 0,
-+ netif->dev->name, netif);
-+ if (err < 0)
-+ goto err_hypervisor;
-+ netif->irq = err;
-+ disable_irq(netif->irq);
+
-+ txs = (netif_tx_sring_t *)netif->tx_comms_area->addr;
-+ BACK_RING_INIT(&netif->tx, txs, PAGE_SIZE);
++ /* Same as CI_ALIGN_FWD and CI_ALIGN_BACK. */
+
-+ rxs = (netif_rx_sring_t *)
-+ ((char *)netif->rx_comms_area->addr);
-+ BACK_RING_INIT(&netif->rx, rxs, PAGE_SIZE);
++#define CI_ROUND_UP(i, align) (((i)+(align)-1u) & ~((align)-1u))
+
-+ netif->rx_req_cons_peek = 0;
++#define CI_ROUND_DOWN(i, align) ((i) & ~((align)-1u))
+
-+ netif_get(netif);
+
-+ rtnl_lock();
-+ netback_carrier_on(netif);
-+ if (netif_running(netif->dev))
-+ __netif_up(netif);
-+ rtnl_unlock();
++/**********************************************************************
++ * Byte-order
++ **********************************************************************/
+
-+ return 0;
-+err_hypervisor:
-+ unmap_frontend_pages(netif);
-+err_map:
-+ free_vm_area(netif->rx_comms_area);
-+err_rx:
-+ free_vm_area(netif->tx_comms_area);
-+ return err;
-+}
++/* These are not flags. They are enumeration values for use with
++ * CI_MY_BYTE_ORDER. */
++#define CI_BIG_ENDIAN 1
++#define CI_LITTLE_ENDIAN 0
+
-+void netif_disconnect(netif_t *netif)
-+{
-+ if (netback_carrier_ok(netif)) {
-+ rtnl_lock();
-+ netback_carrier_off(netif);
-+ netif_carrier_off(netif->dev); /* discard queued packets */
-+ if (netif_running(netif->dev))
-+ __netif_down(netif);
-+ rtnl_unlock();
-+ netif_put(netif);
-+ }
++/*
++** Note that these byte-swapping primitives may leave junk in bits above
++** the range they operate on.
++**
++** The CI_BSWAP_nn() routines require that bits above [nn] are zero. Use
++** CI_BSWAPM_nn(x) if this cannot be guaranteed.
++*/
+
-+ atomic_dec(&netif->refcnt);
-+ wait_event(netif->waiting_to_free, atomic_read(&netif->refcnt) == 0);
++/* ?? May be able to improve on some of these with inline assembler on some
++** platforms.
++*/
+
-+ del_timer_sync(&netif->credit_timeout);
-+ del_timer_sync(&netif->tx_queue_timeout);
++#define CI_BSWAP_16(v) ((((v) & 0xff) << 8) | ((v) >> 8))
++#define CI_BSWAPM_16(v) ((((v) & 0xff) << 8) | (((v) & 0xff00) >> 8))
++
++#define CI_BSWAP_32(v) (((v) >> 24) | \
++ (((v) & 0x00ff0000) >> 8) | \
++ (((v) & 0x0000ff00) << 8) | \
++ ((v) << 24))
++#define CI_BSWAPM_32(v) ((((v) & 0xff000000) >> 24) | \
++ (((v) & 0x00ff0000) >> 8) | \
++ (((v) & 0x0000ff00) << 8) | \
++ ((v) << 24))
++
++#define CI_BSWAP_64(v) (((v) >> 56) | \
++ (((v) & 0x00ff000000000000) >> 40) | \
++ (((v) & 0x0000ff0000000000) >> 24) | \
++ (((v) & 0x000000ff00000000) >> 8) | \
++ (((v) & 0x00000000ff000000) << 8) | \
++ (((v) & 0x0000000000ff0000) << 24) | \
++ (((v) & 0x000000000000ff00) << 40) | \
++ ((v) << 56))
++
++# define CI_BSWAPPED_16_IF(c,v) ((c) ? CI_BSWAP_16(v) : (v))
++# define CI_BSWAPPED_32_IF(c,v) ((c) ? CI_BSWAP_32(v) : (v))
++# define CI_BSWAPPED_64_IF(c,v) ((c) ? CI_BSWAP_64(v) : (v))
++# define CI_BSWAP_16_IF(c,v) do{ if((c)) (v) = CI_BSWAP_16(v); }while(0)
++# define CI_BSWAP_32_IF(c,v) do{ if((c)) (v) = CI_BSWAP_32(v); }while(0)
++# define CI_BSWAP_64_IF(c,v) do{ if((c)) (v) = CI_BSWAP_64(v); }while(0)
++
++#if (CI_MY_BYTE_ORDER == CI_LITTLE_ENDIAN)
++# define CI_BSWAP_LE16(v) (v)
++# define CI_BSWAP_LE32(v) (v)
++# define CI_BSWAP_LE64(v) (v)
++# define CI_BSWAP_BE16(v) CI_BSWAP_16(v)
++# define CI_BSWAP_BE32(v) CI_BSWAP_32(v)
++# define CI_BSWAP_BE64(v) CI_BSWAP_64(v)
++# define CI_BSWAPM_LE16(v) (v)
++# define CI_BSWAPM_LE32(v) (v)
++# define CI_BSWAPM_LE64(v) (v)
++# define CI_BSWAPM_BE16(v) CI_BSWAPM_16(v)
++# define CI_BSWAPM_BE32(v) CI_BSWAPM_32(v)
++#elif (CI_MY_BYTE_ORDER == CI_BIG_ENDIAN)
++# define CI_BSWAP_BE16(v) (v)
++# define CI_BSWAP_BE32(v) (v)
++# define CI_BSWAP_BE64(v) (v)
++# define CI_BSWAP_LE16(v) CI_BSWAP_16(v)
++# define CI_BSWAP_LE32(v) CI_BSWAP_32(v)
++# define CI_BSWAP_LE64(v) CI_BSWAP_64(v)
++# define CI_BSWAPM_BE16(v) (v)
++# define CI_BSWAPM_BE32(v) (v)
++# define CI_BSWAPM_BE64(v) (v)
++# define CI_BSWAPM_LE16(v) CI_BSWAPM_16(v)
++# define CI_BSWAPM_LE32(v) CI_BSWAPM_32(v)
++#else
++# error Bad endian.
++#endif
+
-+ if (netif->irq)
-+ unbind_from_irqhandler(netif->irq, netif);
-+
-+ unregister_netdev(netif->dev);
+
-+ if (netif->tx.sring) {
-+ unmap_frontend_pages(netif);
-+ free_vm_area(netif->tx_comms_area);
-+ free_vm_area(netif->rx_comms_area);
-+ }
++/**********************************************************************
++ * Get pointer to struct from pointer to member
++ **********************************************************************/
+
-+ free_netdev(netif->dev);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/loopback.c linux-2.6.18-xen.hg/drivers/xen/netback/loopback.c
---- linux-2.6.18/drivers/xen/netback/loopback.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/loopback.c 2007-12-23 11:15:34.054602273 +0100
-@@ -0,0 +1,324 @@
-+/******************************************************************************
-+ * netback/loopback.c
-+ *
-+ * A two-interface loopback device to emulate a local netfront-netback
-+ * connection. This ensures that local packet delivery looks identical
-+ * to inter-domain delivery. Most importantly, packets delivered locally
-+ * originating from other domains will get *copied* when they traverse this
-+ * driver. This prevents unbounded delays in socket-buffer queues from
-+ * causing the netback driver to "seize up".
-+ *
-+ * This driver creates a symmetric pair of loopback interfaces with names
-+ * vif0.0 and veth0. The intention is that 'vif0.0' is bound to an Ethernet
-+ * bridge, just like a proper netback interface, while a local IP interface
-+ * is configured on 'veth0'.
-+ *
-+ * As with a real netback interface, vif0.0 is configured with a suitable
-+ * dummy MAC address. No default is provided for veth0: a reasonable strategy
-+ * is to transfer eth0's MAC address to veth0, and give eth0 a dummy address
-+ * (to avoid confusing the Etherbridge).
-+ *
-+ * Copyright (c) 2005 K A Fraser
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
++#define CI_MEMBER_OFFSET(c_type, mbr_name) \
++ ((ci_uint32) (ci_uintptr_t)(&((c_type*)0)->mbr_name))
+
-+#include <linux/module.h>
-+#include <linux/netdevice.h>
-+#include <linux/inetdevice.h>
-+#include <linux/etherdevice.h>
-+#include <linux/skbuff.h>
-+#include <linux/ethtool.h>
-+#include <net/dst.h>
-+#include <net/xfrm.h> /* secpath_reset() */
-+#include <asm/hypervisor.h> /* is_initial_xendomain() */
++#define CI_MEMBER_SIZE(c_type, mbr_name) \
++ sizeof(((c_type*)0)->mbr_name)
+
-+static int nloopbacks = -1;
-+module_param(nloopbacks, int, 0);
-+MODULE_PARM_DESC(nloopbacks, "Number of netback-loopback devices to create");
++#define __CI_CONTAINER(c_type, mbr_name, p_mbr) \
++ ( (c_type*) ((char*)(p_mbr) - CI_MEMBER_OFFSET(c_type, mbr_name)) )
+
-+struct net_private {
-+ struct net_device *loopback_dev;
-+ struct net_device_stats stats;
-+};
++#ifndef CI_CONTAINER
++# define CI_CONTAINER(t,m,p) __CI_CONTAINER(t,m,p)
++#endif
+
-+static int loopback_open(struct net_device *dev)
-+{
-+ struct net_private *np = netdev_priv(dev);
-+ memset(&np->stats, 0, sizeof(np->stats));
-+ netif_start_queue(dev);
-+ return 0;
-+}
+
-+static int loopback_close(struct net_device *dev)
-+{
-+ netif_stop_queue(dev);
-+ return 0;
-+}
++/**********************************************************************
++ * Structure member initialiser.
++ **********************************************************************/
+
-+#ifdef CONFIG_X86
-+static int is_foreign(unsigned long pfn)
-+{
-+ /* NB. Play it safe for auto-translation mode. */
-+ return (xen_feature(XENFEAT_auto_translated_physmap) ||
-+ (phys_to_machine_mapping[pfn] & FOREIGN_FRAME_BIT));
-+}
-+#else
-+/* How to detect a foreign mapping? Play it safe. */
-+#define is_foreign(pfn) (1)
++#ifndef CI_STRUCT_MBR
++# define CI_STRUCT_MBR(name, val) .name = val
+#endif
+
-+static int skb_remove_foreign_references(struct sk_buff *skb)
-+{
-+ struct page *page;
-+ unsigned long pfn;
-+ int i, off;
-+ char *vaddr;
+
-+ BUG_ON(skb_shinfo(skb)->frag_list);
++/**********************************************************************
++ * min / max
++ **********************************************************************/
+
-+ if (skb_cloned(skb) &&
-+ unlikely(pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
-+ return 0;
++#define CI_MIN(x,y) (((x) < (y)) ? (x) : (y))
++#define CI_MAX(x,y) (((x) > (y)) ? (x) : (y))
+
-+ for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
-+ pfn = page_to_pfn(skb_shinfo(skb)->frags[i].page);
-+ if (!is_foreign(pfn))
-+ continue;
-+
-+ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
-+ if (unlikely(!page))
-+ return 0;
++/**********************************************************************
++ * abs
++ **********************************************************************/
+
-+ vaddr = kmap_skb_frag(&skb_shinfo(skb)->frags[i]);
-+ off = skb_shinfo(skb)->frags[i].page_offset;
-+ memcpy(page_address(page) + off,
-+ vaddr + off,
-+ skb_shinfo(skb)->frags[i].size);
-+ kunmap_skb_frag(vaddr);
++#define CI_ABS(x) (((x) < 0) ? -(x) : (x))
+
-+ put_page(skb_shinfo(skb)->frags[i].page);
-+ skb_shinfo(skb)->frags[i].page = page;
-+ }
++/**********************************************************************
++ * Conditional debugging
++ **********************************************************************/
+
-+ return 1;
-+}
++#ifdef NDEBUG
++# define CI_DEBUG(x)
++# define CI_NDEBUG(x) x
++# define CI_IF_DEBUG(y,n) (n)
++# define CI_DEBUG_ARG(x)
++#else
++# define CI_DEBUG(x) x
++# define CI_NDEBUG(x)
++# define CI_IF_DEBUG(y,n) (y)
++# define CI_DEBUG_ARG(x) ,x
++#endif
+
-+static int loopback_start_xmit(struct sk_buff *skb, struct net_device *dev)
-+{
-+ struct net_private *np = netdev_priv(dev);
++#ifdef __KERNEL__
++#define CI_KERNEL_ARG(x) ,x
++#else
++#define CI_KERNEL_ARG(x)
++#endif
+
-+ if (!skb_remove_foreign_references(skb)) {
-+ np->stats.tx_dropped++;
-+ dev_kfree_skb(skb);
-+ return 0;
-+ }
++#ifdef _WIN32
++# define CI_KERNEL_ARG_WIN(x) CI_KERNEL_ARG(x)
++# define CI_ARG_WIN(x) ,x
++#else
++# define CI_KERNEL_ARG_WIN(x)
++# define CI_ARG_WIN(x)
++#endif
+
-+ dst_release(skb->dst);
-+ skb->dst = NULL;
++#ifdef __unix__
++# define CI_KERNEL_ARG_UNIX(x) CI_KERNEL_ARG(x)
++# define CI_ARG_UNIX(x) ,x
++#else
++# define CI_KERNEL_ARG_UNIX(x)
++# define CI_ARG_UNIX(x)
++#endif
+
-+ skb_orphan(skb);
++#ifdef __linux__
++# define CI_KERNEL_ARG_LINUX(x) CI_KERNEL_ARG(x)
++# define CI_ARG_LINUX(x) ,x
++#else
++# define CI_KERNEL_ARG_LINUX(x)
++# define CI_ARG_LINUX(x)
++#endif
+
-+ np->stats.tx_bytes += skb->len;
-+ np->stats.tx_packets++;
+
-+ /* Switch to loopback context. */
-+ dev = np->loopback_dev;
-+ np = netdev_priv(dev);
++#endif /* __CI_COMPAT_UTILS_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/x86.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/x86.h 2008-05-19 00:33:47.734897614 +0300
+@@ -0,0 +1,48 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ np->stats.rx_bytes += skb->len;
-+ np->stats.rx_packets++;
++/*! \cidoxg_include_ci_compat */
+
-+ if (skb->ip_summed == CHECKSUM_HW) {
-+ /* Defer checksum calculation. */
-+ skb->proto_csum_blank = 1;
-+ /* Must be a local packet: assert its integrity. */
-+ skb->proto_data_valid = 1;
-+ }
++#ifndef __CI_COMPAT_X86_H__
++#define __CI_COMPAT_X86_H__
+
-+ skb->ip_summed = skb->proto_data_valid ?
-+ CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
+
-+ skb->pkt_type = PACKET_HOST; /* overridden by eth_type_trans() */
-+ skb->protocol = eth_type_trans(skb, dev);
-+ skb->dev = dev;
-+ dev->last_rx = jiffies;
++#define CI_MY_BYTE_ORDER CI_LITTLE_ENDIAN
+
-+ /* Flush netfilter context: rx'ed skbuffs not expected to have any. */
-+ nf_reset(skb);
-+ secpath_reset(skb);
++#define CI_WORD_SIZE 4
++#define CI_PTR_SIZE 4
+
-+ netif_rx(skb);
++#define CI_PAGE_SIZE 4096
++#define CI_PAGE_SHIFT 12
++#define CI_PAGE_MASK (~(CI_PAGE_SIZE - 1))
+
-+ return 0;
-+}
++#define CI_CPU_HAS_SSE 1 /* SSE extensions supported */
++#define CI_CPU_HAS_SSE2 0 /* SSE2 extensions supported */
++#define CI_CPU_OOS 0 /* CPU does out of order stores */
+
-+static struct net_device_stats *loopback_get_stats(struct net_device *dev)
-+{
-+ struct net_private *np = netdev_priv(dev);
-+ return &np->stats;
-+}
+
-+static struct ethtool_ops network_ethtool_ops =
-+{
-+ .get_tx_csum = ethtool_op_get_tx_csum,
-+ .set_tx_csum = ethtool_op_set_tx_csum,
-+ .get_sg = ethtool_op_get_sg,
-+ .set_sg = ethtool_op_set_sg,
-+ .get_tso = ethtool_op_get_tso,
-+ .set_tso = ethtool_op_set_tso,
-+ .get_link = ethtool_op_get_link,
-+};
++#endif /* __CI_COMPAT_X86_H__ */
+
-+/*
-+ * Nothing to do here. Virtual interface is point-to-point and the
-+ * physical interface is probably promiscuous anyway.
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat/x86_64.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat/x86_64.h 2008-05-19 00:33:47.866905224 +0300
+@@ -0,0 +1,54 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static void loopback_set_multicast_list(struct net_device *dev)
-+{
-+}
+
-+static void loopback_construct(struct net_device *dev, struct net_device *lo)
-+{
-+ struct net_private *np = netdev_priv(dev);
++/*
++ * \author djr
++ * \brief Arch stuff for AMD x86_64.
++ * \date 2004/08/17
++ */
+
-+ np->loopback_dev = lo;
++/*! \cidoxg_include_ci_compat */
++#ifndef __CI_COMPAT_X86_64_H__
++#define __CI_COMPAT_X86_64_H__
+
-+ dev->open = loopback_open;
-+ dev->stop = loopback_close;
-+ dev->hard_start_xmit = loopback_start_xmit;
-+ dev->get_stats = loopback_get_stats;
-+ dev->set_multicast_list = loopback_set_multicast_list;
-+ dev->change_mtu = NULL; /* allow arbitrary mtu */
+
-+ dev->tx_queue_len = 0;
++#define CI_MY_BYTE_ORDER CI_LITTLE_ENDIAN
+
-+ dev->features = (NETIF_F_HIGHDMA |
-+ NETIF_F_LLTX |
-+ NETIF_F_TSO |
-+ NETIF_F_SG |
-+ NETIF_F_IP_CSUM);
++#define CI_WORD_SIZE 8
++#define CI_PTR_SIZE 8
+
-+ SET_ETHTOOL_OPS(dev, &network_ethtool_ops);
++#define CI_PAGE_SIZE 4096
++#define CI_PAGE_SHIFT 12
++#define CI_PAGE_MASK (~(CI_PAGE_SIZE - 1))
+
-+ /*
-+ * We do not set a jumbo MTU on the interface. Otherwise the network
-+ * stack will try to send large packets that will get dropped by the
-+ * Ethernet bridge (unless the physical Ethernet interface is
-+ * configured to transfer jumbo packets). If a larger MTU is desired
-+ * then the system administrator can specify it using the 'ifconfig'
-+ * command.
-+ */
-+ /*dev->mtu = 16*1024;*/
-+}
++#define CI_CPU_HAS_SSE 1 /* SSE extensions supported */
+
-+static int __init make_loopback(int i)
-+{
-+ struct net_device *dev1, *dev2;
-+ char dev_name[IFNAMSIZ];
-+ int err = -ENOMEM;
++/* SSE2 disabled while investigating BUG1060 */
++#define CI_CPU_HAS_SSE2 0 /* SSE2 extensions supported */
++#define CI_CPU_OOS 0 /* CPU does out of order stores */
+
-+ sprintf(dev_name, "vif0.%d", i);
-+ dev1 = alloc_netdev(sizeof(struct net_private), dev_name, ether_setup);
-+ if (!dev1)
-+ return err;
+
-+ sprintf(dev_name, "veth%d", i);
-+ dev2 = alloc_netdev(sizeof(struct net_private), dev_name, ether_setup);
-+ if (!dev2)
-+ goto fail_netdev2;
++#endif /* __CI_COMPAT_X86_64_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/compat.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/compat.h 2008-05-19 00:33:47.278871329 +0300
+@@ -0,0 +1,53 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ loopback_construct(dev1, dev2);
-+ loopback_construct(dev2, dev1);
++/*
++ * \author djr
++ * \brief Compatability layer. Provides definitions of fundamental
++ * types and definitions that are used throughout CI source
++ * code. It does not introduce any link time dependencies,
++ * or include any unnecessary system headers.
++ */
++/*! \cidoxg_include_ci */
+
-+ /*
-+ * Initialise a dummy MAC address for the 'dummy backend' interface. We
-+ * choose the numerically largest non-broadcast address to prevent the
-+ * address getting stolen by an Ethernet bridge for STP purposes.
-+ */
-+ memset(dev1->dev_addr, 0xFF, ETH_ALEN);
-+ dev1->dev_addr[0] &= ~0x01;
++#ifndef __CI_COMPAT_H__
++#define __CI_COMPAT_H__
+
-+ if ((err = register_netdev(dev1)) != 0)
-+ goto fail;
++#ifdef __cplusplus
++extern "C" {
++#endif
+
-+ if ((err = register_netdev(dev2)) != 0) {
-+ unregister_netdev(dev1);
-+ goto fail;
-+ }
++#include <ci/compat/primitive.h>
++#include <ci/compat/sysdep.h>
++#include <ci/compat/utils.h>
+
-+ return 0;
+
-+ fail:
-+ free_netdev(dev2);
-+ fail_netdev2:
-+ free_netdev(dev1);
-+ return err;
++#ifdef __cplusplus
+}
++#endif
+
-+static void __exit clean_loopback(int i)
-+{
-+ struct net_device *dev1, *dev2;
-+ char dev_name[IFNAMSIZ];
++#endif /* __CI_COMPAT_H__ */
+
-+ sprintf(dev_name, "vif0.%d", i);
-+ dev1 = dev_get_by_name(dev_name);
-+ sprintf(dev_name, "veth%d", i);
-+ dev2 = dev_get_by_name(dev_name);
-+ if (dev1 && dev2) {
-+ unregister_netdev(dev2);
-+ unregister_netdev(dev1);
-+ free_netdev(dev2);
-+ free_netdev(dev1);
-+ }
-+}
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/driver/resource/efx_vi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/driver/resource/efx_vi.h 2008-05-19 00:33:47.926908682 +0300
+@@ -0,0 +1,276 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file contains public EFX VI API to Solarflare resource manager.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static int __init loopback_init(void)
-+{
-+ int i, err = 0;
++#ifndef __CI_DRIVER_RESOURCE_EFX_VI_H__
++#define __CI_DRIVER_RESOURCE_EFX_VI_H__
+
-+ if (nloopbacks == -1)
-+ nloopbacks = is_initial_xendomain() ? 4 : 0;
++/* Default size of event queue in the efx_vi resource. Copied from
++ * CI_CFG_NETIF_EVENTQ_SIZE */
++#define EFX_VI_EVENTQ_SIZE_DEFAULT 1024
+
-+ for (i = 0; i < nloopbacks; i++)
-+ if ((err = make_loopback(i)) != 0)
-+ break;
++extern int efx_vi_eventq_size;
+
-+ return err;
-+}
++/**************************************************************************
++ * efx_vi_state types, allocation and free
++ **************************************************************************/
+
-+module_init(loopback_init);
++/*! Handle for refering to a efx_vi */
++struct efx_vi_state;
+
-+static void __exit loopback_exit(void)
-+{
-+ int i;
++/*!
++ * Allocate an efx_vi, including event queue and pt_endpoint
++ *
++ * \param vih_out Pointer to a handle that is set on success
++ * \param nic_index Index of NIC to apply this resource to
++ * \return Zero on success (and vih_out set), non-zero on failure.
++ */
++extern int
++efx_vi_alloc(struct efx_vi_state **vih_out, int nic_index);
+
-+ for (i = nloopbacks; i-- > 0; )
-+ clean_loopback(i);
-+}
++/*!
++ * Free a previously allocated efx_vi
++ *
++ * \param vih The handle of the efx_vi to free
++ */
++extern void
++efx_vi_free(struct efx_vi_state *vih);
+
-+module_exit(loopback_exit);
++/*!
++ * Reset a previously allocated efx_vi
++ *
++ * \param vih The handle of the efx_vi to reset
++ */
++extern void
++efx_vi_reset(struct efx_vi_state *vih);
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/Makefile linux-2.6.18-xen.hg/drivers/xen/netback/Makefile
---- linux-2.6.18/drivers/xen/netback/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/Makefile 2007-12-23 11:15:34.051268764 +0100
-@@ -0,0 +1,5 @@
-+obj-$(CONFIG_XEN_NETDEV_BACKEND) := netbk.o
-+obj-$(CONFIG_XEN_NETDEV_LOOPBACK) += netloop.o
++/**************************************************************************
++ * efx_vi_eventq types and functions
++ **************************************************************************/
+
-+netbk-y := netback.o xenbus.o interface.o accel.o
-+netloop-y := loopback.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/netback.c linux-2.6.18-xen.hg/drivers/xen/netback/netback.c
---- linux-2.6.18/drivers/xen/netback/netback.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/netback.c 2007-12-23 11:15:34.054602273 +0100
-@@ -0,0 +1,1608 @@
-+/******************************************************************************
-+ * drivers/xen/netback/netback.c
-+ *
-+ * Back-end of the driver for virtual network devices. This portion of the
-+ * driver exports a 'unified' network-device interface that can be accessed
-+ * by any operating system that implements a compatible front end. A
-+ * reference front-end implementation can be found in:
-+ * drivers/xen/netfront/netfront.c
-+ *
-+ * Copyright (c) 2002-2005, K A Fraser
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++/*!
++ * Register a function to receive callbacks when event queue timeouts
++ * or wakeups occur. Only one function per efx_vi can be registered
++ * at once.
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param callback The function to callback
++ * \param context An argument to pass to the callback function
++ * \return Zero on success, non-zero on failure.
+ */
++extern int
++efx_vi_eventq_register_callback(struct efx_vi_state *vih,
++ void (*callback)(void *context, int is_timeout),
++ void *context);
+
-+#include "common.h"
-+#include <xen/balloon.h>
-+#include <xen/interface/memory.h>
++/*!
++ * Remove the current eventq timeout or wakeup callback function
++ *
++ * \param vih The handle to identify the efx_vi
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_eventq_kill_callback(struct efx_vi_state *vih);
+
-+/*define NETBE_DEBUG_INTERRUPT*/
++/**************************************************************************
++ * efx_vi_dma_map types and functions
++ **************************************************************************/
+
-+/* extra field used in struct page */
-+#define netif_page_index(pg) (*(long *)&(pg)->mapping)
++/*!
++ * Handle for refering to a efx_vi
++ */
++struct efx_vi_dma_map_state;
+
-+struct netbk_rx_meta {
-+ skb_frag_t frag;
-+ int id;
-+ u8 copy:1;
-+};
++/*!
++ * Map a list of buffer pages so they are registered with the hardware
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param addrs An array of page pointers to map
++ * \param n_addrs Length of the page pointer array. Must be a power of two.
++ * \param dmh_out Set on success to a handle used to refer to this mapping
++ * \return Zero on success, non-zero on failure.
++ */
++extern int
++efx_vi_dma_map_pages(struct efx_vi_state *vih, struct page **pages,
++ int n_pages, struct efx_vi_dma_map_state **dmh_out);
++extern int
++efx_vi_dma_map_addrs(struct efx_vi_state *vih,
++ unsigned long long *dev_bus_addrs, int n_pages,
++ struct efx_vi_dma_map_state **dmh_out);
+
-+struct netbk_tx_pending_inuse {
-+ struct list_head list;
-+ unsigned long alloc_time;
-+};
++/*!
++ * Unmap a previously mapped set of pages so they are no longer registered
++ * with the hardware.
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param dmh The handle to identify the dma mapping
++ */
++extern void
++efx_vi_dma_unmap_pages(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
++extern void
++efx_vi_dma_unmap_addrs(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
+
-+static void netif_idx_release(u16 pending_idx);
-+static void netif_page_release(struct page *page);
-+static void make_tx_response(netif_t *netif,
-+ netif_tx_request_t *txp,
-+ s8 st);
-+static netif_rx_response_t *make_rx_response(netif_t *netif,
-+ u16 id,
-+ s8 st,
-+ u16 offset,
-+ u16 size,
-+ u16 flags);
++/*!
++ * Retrieve the buffer address of the mapping
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param dmh The handle to identify the buffer mapping
++ * \return The buffer address on success, or zero on failure
++ */
++extern unsigned
++efx_vi_dma_get_map_addr(struct efx_vi_state *vih,
++ struct efx_vi_dma_map_state *dmh);
+
-+static void net_tx_action(unsigned long unused);
-+static DECLARE_TASKLET(net_tx_tasklet, net_tx_action, 0);
++/**************************************************************************
++ * efx_vi filter functions
++ **************************************************************************/
+
-+static void net_rx_action(unsigned long unused);
-+static DECLARE_TASKLET(net_rx_tasklet, net_rx_action, 0);
++#define EFX_VI_STATIC_FILTERS 32
+
-+static struct timer_list net_timer;
-+static struct timer_list netbk_tx_pending_timer;
++/*! Handle to refer to a filter instance */
++struct filter_resource_t;
+
-+#define MAX_PENDING_REQS 256
++/*!
++ * Allocate and add a filter
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param protocol The protocol of the new filter: UDP or TCP
++ * \param ip_addr_be32 The local ip address of the filter
++ * \param port_le16 The local port of the filter
++ * \param fh_out Set on success to be a handle to refer to this filter
++ * \return Zero on success, non-zero on failure.
++ */
++extern int
++efx_vi_filter(struct efx_vi_state *vih, int protocol, unsigned ip_addr_be32,
++ int port_le16, struct filter_resource_t **fh_out);
+
-+static struct sk_buff_head rx_queue;
++/*!
++ * Remove a filter and free resources associated with it
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param fh The handle to identify the filter
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_filter_stop(struct efx_vi_state *vih, struct filter_resource_t *fh);
++
++/**************************************************************************
++ * efx_vi hw resources types and functions
++ **************************************************************************/
++
++/*! Constants for the type field in efx_vi_hw_resource */
++#define EFX_VI_HW_RESOURCE_TXDMAQ 0x0 /* PFN of TX DMA Q */
++#define EFX_VI_HW_RESOURCE_RXDMAQ 0x1 /* PFN of RX DMA Q */
++#define EFX_VI_HW_RESOURCE_TXBELL 0x2 /* PFN of TX Doorbell (EF1) */
++#define EFX_VI_HW_RESOURCE_RXBELL 0x3 /* PFN of RX Doorbell (EF1) */
++#define EFX_VI_HW_RESOURCE_EVQTIMER 0x4 /* Address of event q timer */
++
++/* Address of event q pointer (EF1) */
++#define EFX_VI_HW_RESOURCE_EVQPTR 0x5
++/* Address of register pointer (Falcon A) */
++#define EFX_VI_HW_RESOURCE_EVQRPTR 0x6
++/* Offset of register pointer (Falcon B) */
++#define EFX_VI_HW_RESOURCE_EVQRPTR_OFFSET 0x7
++/* Address of mem KVA */
++#define EFX_VI_HW_RESOURCE_EVQMEMKVA 0x8
++/* PFN of doorbell page (Falcon) */
++#define EFX_VI_HW_RESOURCE_BELLPAGE 0x9
++
++/*! How large an array to allocate for the get_() functions - smaller
++ than the total number of constants as some are mutually exclusive */
++#define EFX_VI_HW_RESOURCE_MAXSIZE 0x7
++
++/*! Constants for the mem_type field in efx_vi_hw_resource */
++#define EFX_VI_HW_RESOURCE_IOBUFFER 0 /* Host memory */
++#define EFX_VI_HW_RESOURCE_PERIPHERAL 1 /* Card memory/registers */
++
++/*!
++ * Data structure providing information on a hardware resource mapping
++ */
++struct efx_vi_hw_resource {
++ u8 type; /*!< What this resource represents */
++ u8 mem_type; /*!< What type of memory is it in, eg,
++ * host or iomem */
++ u8 more_to_follow; /*!< Is this part of a multi-region resource */
++ u32 length; /*!< Length of the resource in bytes */
++ unsigned long address; /*!< Address of this resource */
++};
++
++/*!
++ * Metadata concerning the list of hardware resource mappings
++ */
++struct efx_vi_hw_resource_metadata {
++ int version;
++ int evq_order;
++ int evq_offs;
++ int evq_capacity;
++ int instance;
++ unsigned rx_capacity;
++ unsigned tx_capacity;
++ int nic_arch;
++ int nic_revision;
++ char nic_variant;
++};
++
++/*!
++ * Obtain a list of hardware resource mappings, using virtual addresses
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param mdata Pointer to a structure to receive the metadata
++ * \param hw_res_array An array to receive the list of hardware resources
++ * \param length The length of hw_res_array. Updated on success to contain
++ * the number of entries in the supplied array that were used.
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_hw_resource_get_virt(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length);
+
-+static struct page **mmap_pages;
-+static inline unsigned long idx_to_pfn(unsigned int idx)
-+{
-+ return page_to_pfn(mmap_pages[idx]);
-+}
++/*!
++ * Obtain a list of hardware resource mappings, using physical addresses
++ *
++ * \param vih The handle to identify the efx_vi
++ * \param mdata Pointer to a structure to receive the metadata
++ * \param hw_res_array An array to receive the list of hardware resources
++ * \param length The length of hw_res_array. Updated on success to contain
++ * the number of entries in the supplied array that were used.
++ * \return Zero on success, non-zero on failure
++ */
++extern int
++efx_vi_hw_resource_get_phys(struct efx_vi_state *vih,
++ struct efx_vi_hw_resource_metadata *mdata,
++ struct efx_vi_hw_resource *hw_res_array,
++ int *length);
++
++#endif /* __CI_DRIVER_RESOURCE_EFX_VI_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/common.h 2008-05-19 00:33:48.038915138 +0300
+@@ -0,0 +1,102 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides API of the efhw library which may be used both from
++ * the kernel and from the user-space code.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static inline unsigned long idx_to_kaddr(unsigned int idx)
-+{
-+ return (unsigned long)pfn_to_kaddr(idx_to_pfn(idx));
-+}
++#ifndef __CI_EFHW_COMMON_H__
++#define __CI_EFHW_COMMON_H__
+
-+#define PKT_PROT_LEN 64
++#include <ci/efhw/common_sysdep.h>
+
-+static struct pending_tx_info {
-+ netif_tx_request_t req;
-+ netif_t *netif;
-+} pending_tx_info[MAX_PENDING_REQS];
-+static u16 pending_ring[MAX_PENDING_REQS];
-+typedef unsigned int PEND_RING_IDX;
-+#define MASK_PEND_IDX(_i) ((_i)&(MAX_PENDING_REQS-1))
-+static PEND_RING_IDX pending_prod, pending_cons;
-+#define NR_PENDING_REQS (MAX_PENDING_REQS - pending_prod + pending_cons)
++enum efhw_arch {
++ EFHW_ARCH_FALCON,
++ EFHW_ARCH_SIENA,
++};
+
-+/* Freed TX SKBs get batched on this ring before return to pending_ring. */
-+static u16 dealloc_ring[MAX_PENDING_REQS];
-+static PEND_RING_IDX dealloc_prod, dealloc_cons;
++typedef uint32_t efhw_buffer_addr_t;
++#define EFHW_BUFFER_ADDR_FMT "[ba:%"PRIx32"]"
+
-+/* Doubly-linked list of in-use pending entries. */
-+static struct netbk_tx_pending_inuse pending_inuse[MAX_PENDING_REQS];
-+static LIST_HEAD(pending_inuse_head);
++/*! Comment? */
++typedef union {
++ uint64_t u64;
++ struct {
++ uint32_t a;
++ uint32_t b;
++ } opaque;
++ struct {
++ uint32_t code;
++ uint32_t status;
++ } ev1002;
++} efhw_event_t;
++
++/* Flags for TX/RX queues */
++#define EFHW_VI_JUMBO_EN 0x01 /*! scatter RX over multiple desc */
++#define EFHW_VI_ISCSI_RX_HDIG_EN 0x02 /*! iscsi rx header digest */
++#define EFHW_VI_ISCSI_TX_HDIG_EN 0x04 /*! iscsi tx header digest */
++#define EFHW_VI_ISCSI_RX_DDIG_EN 0x08 /*! iscsi rx data digest */
++#define EFHW_VI_ISCSI_TX_DDIG_EN 0x10 /*! iscsi tx data digest */
++#define EFHW_VI_TX_PHYS_ADDR_EN 0x20 /*! TX physical address mode */
++#define EFHW_VI_RX_PHYS_ADDR_EN 0x40 /*! RX physical address mode */
++#define EFHW_VI_RM_WITH_INTERRUPT 0x80 /*! VI with an interrupt */
++#define EFHW_VI_TX_IP_CSUM_DIS 0x100 /*! enable ip checksum generation */
++#define EFHW_VI_TX_TCPUDP_CSUM_DIS 0x200 /*! enable tcp/udp checksum
++ generation */
++#define EFHW_VI_TX_TCPUDP_ONLY 0x400 /*! drop non-tcp/udp packets */
++
++/* Types of hardware filter */
++/* Each of these values implicitly selects scatter filters on B0 - or in
++ EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK if a non-scatter filter is required */
++#define EFHW_IP_FILTER_TYPE_UDP_WILDCARD (0) /* dest host only */
++#define EFHW_IP_FILTER_TYPE_UDP_FULL (1) /* dest host and port */
++#define EFHW_IP_FILTER_TYPE_TCP_WILDCARD (2) /* dest based filter */
++#define EFHW_IP_FILTER_TYPE_TCP_FULL (3) /* src filter */
++/* Same again, but with RSS (for B0 only) */
++#define EFHW_IP_FILTER_TYPE_UDP_WILDCARD_RSS_B0 (4)
++#define EFHW_IP_FILTER_TYPE_UDP_FULL_RSS_B0 (5)
++#define EFHW_IP_FILTER_TYPE_TCP_WILDCARD_RSS_B0 (6)
++#define EFHW_IP_FILTER_TYPE_TCP_FULL_RSS_B0 (7)
++
++#define EFHW_IP_FILTER_TYPE_FULL_MASK (0x1) /* Mask for full / wildcard */
++#define EFHW_IP_FILTER_TYPE_TCP_MASK (0x2) /* Mask for TCP type */
++#define EFHW_IP_FILTER_TYPE_RSS_B0_MASK (0x4) /* Mask for B0 RSS enable */
++#define EFHW_IP_FILTER_TYPE_NOSCAT_B0_MASK (0x8) /* Mask for B0 SCATTER dsbl */
++
++#define EFHW_IP_FILTER_TYPE_MASK (0xffff) /* Mask of types above */
++
++#define EFHW_IP_FILTER_BROADCAST (0x10000) /* driverlink filter
++ support */
++
++#endif /* __CI_EFHW_COMMON_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/common_sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/common_sysdep.h 2008-05-19 00:33:48.054916061 +0300
+@@ -0,0 +1,67 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for
++ * userland-to-kernel interfaces.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static struct sk_buff_head tx_queue;
++#ifndef __CI_EFHW_COMMON_LINUX_H__
++#define __CI_EFHW_COMMON_LINUX_H__
+
-+static grant_handle_t grant_tx_handle[MAX_PENDING_REQS];
-+static gnttab_unmap_grant_ref_t tx_unmap_ops[MAX_PENDING_REQS];
-+static gnttab_map_grant_ref_t tx_map_ops[MAX_PENDING_REQS];
++#include <linux/types.h>
++#include <linux/version.h>
+
-+static struct list_head net_schedule_list;
-+static spinlock_t net_schedule_list_lock;
++/* Dirty hack, but Linux kernel does not provide DMA_ADDR_T_FMT */
++#if BITS_PER_LONG == 64 || defined(CONFIG_HIGHMEM64G)
++#define DMA_ADDR_T_FMT "%llx"
++#else
++#define DMA_ADDR_T_FMT "%x"
++#endif
+
-+#define MAX_MFN_ALLOC 64
-+static unsigned long mfn_list[MAX_MFN_ALLOC];
-+static unsigned int alloc_index = 0;
++/* Linux kernel also does not provide PRIx32... Sigh. */
++#define PRIx32 "x"
++#define PRIx64 "llx"
+
-+/* Setting this allows the safe use of this driver without netloop. */
-+static int MODPARM_copy_skb = 1;
-+module_param_named(copy_skb, MODPARM_copy_skb, bool, 0);
-+MODULE_PARM_DESC(copy_skb, "Copy data received from netfront without netloop");
+
-+int netbk_copy_skb_mode;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
++enum {
++ false = 0,
++ true = 1
++};
+
-+static inline unsigned long alloc_mfn(void)
-+{
-+ BUG_ON(alloc_index == 0);
-+ return mfn_list[--alloc_index];
-+}
++typedef _Bool bool;
++#endif /* LINUX_VERSION_CODE < 2.6.19 */
+
-+static int check_mfn(int nr)
-+{
-+ struct xen_memory_reservation reservation = {
-+ .extent_order = 0,
-+ .domid = DOMID_SELF
-+ };
++#endif /* __CI_EFHW_COMMON_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/debug.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/debug.h 2008-05-19 00:33:48.054916061 +0300
+@@ -0,0 +1,84 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides debug-related API for efhw library using Linux kernel
++ * primitives.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (likely(alloc_index >= nr))
-+ return 0;
++#ifndef __CI_EFHW_DEBUG_LINUX_H__
++#define __CI_EFHW_DEBUG_LINUX_H__
+
-+ set_xen_guest_handle(reservation.extent_start, mfn_list + alloc_index);
-+ reservation.nr_extents = MAX_MFN_ALLOC - alloc_index;
-+ alloc_index += HYPERVISOR_memory_op(XENMEM_increase_reservation,
-+ &reservation);
++#define EFHW_PRINTK_PREFIX "[sfc efhw] "
+
-+ return alloc_index >= nr ? 0 : -ENOMEM;
-+}
++#define EFHW_PRINTK(level, fmt, ...) \
++ printk(level EFHW_PRINTK_PREFIX fmt "\n", __VA_ARGS__)
+
-+static inline void maybe_schedule_tx_action(void)
-+{
-+ smp_mb();
-+ if ((NR_PENDING_REQS < (MAX_PENDING_REQS/2)) &&
-+ !list_empty(&net_schedule_list))
-+ tasklet_schedule(&net_tx_tasklet);
-+}
++/* Following macros should be used with non-zero format parameters
++ * due to __VA_ARGS__ limitations. Use "%s" with __FUNCTION__ if you can't
++ * find better parameters. */
++#define EFHW_ERR(fmt, ...) EFHW_PRINTK(KERN_ERR, fmt, __VA_ARGS__)
++#define EFHW_WARN(fmt, ...) EFHW_PRINTK(KERN_WARNING, fmt, __VA_ARGS__)
++#define EFHW_NOTICE(fmt, ...) EFHW_PRINTK(KERN_NOTICE, fmt, __VA_ARGS__)
++#if 0 && !defined(NDEBUG)
++#define EFHW_TRACE(fmt, ...) EFHW_PRINTK(KERN_DEBUG, fmt, __VA_ARGS__)
++#else
++#define EFHW_TRACE(fmt, ...)
++#endif
+
-+static struct sk_buff *netbk_copy_skb(struct sk_buff *skb)
-+{
-+ struct skb_shared_info *ninfo;
-+ struct sk_buff *nskb;
-+ unsigned long offset;
-+ int ret;
-+ int len;
-+ int headlen;
++#ifndef NDEBUG
++#define EFHW_ASSERT(cond) BUG_ON((cond) == 0)
++#define EFHW_DO_DEBUG(expr) expr
++#else
++#define EFHW_ASSERT(cond)
++#define EFHW_DO_DEBUG(expr)
++#endif
+
-+ BUG_ON(skb_shinfo(skb)->frag_list != NULL);
++#define EFHW_TEST(expr) \
++ do { \
++ if (unlikely(!(expr))) \
++ BUG(); \
++ } while (0)
+
-+ nskb = alloc_skb(SKB_MAX_HEAD(0), GFP_ATOMIC | __GFP_NOWARN);
-+ if (unlikely(!nskb))
-+ goto err;
++/* Build time asserts. We paste the line number into the type name
++ * so that the macro can be used more than once per file even if the
++ * compiler objects to multiple identical typedefs. Collisions
++ * between use in different header files is still possible. */
++#ifndef EFHW_BUILD_ASSERT
++#define __EFHW_BUILD_ASSERT_NAME(_x) __EFHW_BUILD_ASSERT_ILOATHECPP(_x)
++#define __EFHW_BUILD_ASSERT_ILOATHECPP(_x) __EFHW_BUILD_ASSERT__ ##_x
++#define EFHW_BUILD_ASSERT(e) \
++ typedef char __EFHW_BUILD_ASSERT_NAME(__LINE__)[(e) ? 1 : -1]
++#endif
+
-+ skb_reserve(nskb, 16 + NET_IP_ALIGN);
-+ headlen = nskb->end - nskb->data;
-+ if (headlen > skb_headlen(skb))
-+ headlen = skb_headlen(skb);
-+ ret = skb_copy_bits(skb, 0, __skb_put(nskb, headlen), headlen);
-+ BUG_ON(ret);
++#endif /* __CI_EFHW_DEBUG_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/efhw_config.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/efhw_config.h 2008-05-19 00:33:48.054916061 +0300
+@@ -0,0 +1,43 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides some limits used in both kernel and userland code.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ ninfo = skb_shinfo(nskb);
-+ ninfo->gso_size = skb_shinfo(skb)->gso_size;
-+ ninfo->gso_type = skb_shinfo(skb)->gso_type;
++#ifndef __CI_EFHW_EFAB_CONFIG_H__
++#define __CI_EFHW_EFAB_CONFIG_H__
+
-+ offset = headlen;
-+ len = skb->len - headlen;
++#define EFHW_MAX_NR_DEVS 5 /* max number of efhw devices supported */
+
-+ nskb->len = skb->len;
-+ nskb->data_len = len;
-+ nskb->truesize += len;
++#endif /* __CI_EFHW_EFAB_CONFIG_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/efhw_types.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/efhw_types.h 2008-05-19 00:33:48.058916291 +0300
+@@ -0,0 +1,342 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides struct efhw_nic and some related types.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ while (len) {
-+ struct page *page;
-+ int copy;
-+ int zero;
++#ifndef __CI_EFHW_EFAB_TYPES_H__
++#define __CI_EFHW_EFAB_TYPES_H__
+
-+ if (unlikely(ninfo->nr_frags >= MAX_SKB_FRAGS)) {
-+ dump_stack();
-+ goto err_free;
-+ }
++#include <ci/efhw/efhw_config.h>
++#include <ci/efhw/hardware_sysdep.h>
++#include <ci/efhw/iopage_types.h>
++#include <ci/efhw/sysdep.h>
+
-+ copy = len >= PAGE_SIZE ? PAGE_SIZE : len;
-+ zero = len >= PAGE_SIZE ? 0 : __GFP_ZERO;
++/*--------------------------------------------------------------------
++ *
++ * hardware limits used in the types
++ *
++ *--------------------------------------------------------------------*/
+
-+ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN | zero);
-+ if (unlikely(!page))
-+ goto err_free;
++#define EFHW_KEVENTQ_MAX 8
+
-+ ret = skb_copy_bits(skb, offset, page_address(page), copy);
-+ BUG_ON(ret);
++/*--------------------------------------------------------------------
++ *
++ * forward type declarations
++ *
++ *--------------------------------------------------------------------*/
+
-+ ninfo->frags[ninfo->nr_frags].page = page;
-+ ninfo->frags[ninfo->nr_frags].page_offset = 0;
-+ ninfo->frags[ninfo->nr_frags].size = copy;
-+ ninfo->nr_frags++;
++struct efhw_nic;
+
-+ offset += copy;
-+ len -= copy;
-+ }
++/*--------------------------------------------------------------------
++ *
++ * Managed interface
++ *
++ *--------------------------------------------------------------------*/
+
-+ offset = nskb->data - skb->data;
++struct efhw_buffer_table_allocation{
++ unsigned base;
++ unsigned order;
++};
+
-+ nskb->h.raw = skb->h.raw + offset;
-+ nskb->nh.raw = skb->nh.raw + offset;
-+ nskb->mac.raw = skb->mac.raw + offset;
++struct eventq_resource_hardware {
++ /*!iobuffer allocated for eventq - can be larger than eventq */
++ efhw_iopages_t iobuff;
++ unsigned iobuff_off;
++ struct efhw_buffer_table_allocation buf_tbl_alloc;
++ int capacity; /*!< capacity of event queue */
++};
+
-+ return nskb;
++/*--------------------------------------------------------------------
++ *
++ * event queues and event driven callbacks
++ *
++ *--------------------------------------------------------------------*/
+
-+ err_free:
-+ kfree_skb(nskb);
-+ err:
-+ return NULL;
-+}
++struct efhw_keventq {
++ volatile int lock;
++ caddr_t evq_base;
++ int32_t evq_ptr;
++ uint32_t evq_mask;
++ unsigned instance;
++ struct eventq_resource_hardware hw;
++ struct efhw_ev_handler *ev_handlers;
++};
+
-+static inline int netbk_max_required_rx_slots(netif_t *netif)
-+{
-+ if (netif->features & (NETIF_F_SG|NETIF_F_TSO))
-+ return MAX_SKB_FRAGS + 2; /* header + extra_info + frags */
-+ return 1; /* all in one */
-+}
++/**********************************************************************
++ * Portable HW interface. ***************************************
++ **********************************************************************/
+
-+static inline int netbk_queue_full(netif_t *netif)
-+{
-+ RING_IDX peek = netif->rx_req_cons_peek;
-+ RING_IDX needed = netbk_max_required_rx_slots(netif);
++/*--------------------------------------------------------------------
++ *
++ * EtherFabric Functional units - configuration and control
++ *
++ *--------------------------------------------------------------------*/
+
-+ return ((netif->rx.sring->req_prod - peek) < needed) ||
-+ ((netif->rx.rsp_prod_pvt + NET_RX_RING_SIZE - peek) < needed);
-+}
++struct efhw_func_ops {
+
-+static void tx_queue_callback(unsigned long data)
-+{
-+ netif_t *netif = (netif_t *)data;
-+ if (netif_schedulable(netif))
-+ netif_wake_queue(netif->dev);
-+}
++ /*-------------- Initialisation ------------ */
+
-+int netif_be_start_xmit(struct sk_buff *skb, struct net_device *dev)
-+{
-+ netif_t *netif = netdev_priv(dev);
++ /*! close down all hardware functional units - leaves NIC in a safe
++ state for driver unload */
++ void (*close_hardware) (struct efhw_nic *nic);
+
-+ BUG_ON(skb->dev != dev);
++ /*! initialise all hardware functional units */
++ int (*init_hardware) (struct efhw_nic *nic,
++ struct efhw_ev_handler *,
++ const uint8_t *mac_addr);
+
-+ /* Drop the packet if the target domain has no receive buffers. */
-+ if (unlikely(!netif_schedulable(netif) || netbk_queue_full(netif)))
-+ goto drop;
++ /*-------------- Interrupt support ------------ */
+
-+ /*
-+ * Copy the packet here if it's destined for a flipping interface
-+ * but isn't flippable (e.g. extra references to data).
-+ * XXX For now we also copy skbuffs whose head crosses a page
-+ * boundary, because netbk_gop_skb can't handle them.
++ /*! Main interrupt routine
++ ** This function returns,
++ ** - zero, if the IRQ was not generated by EF1
++ ** - non-zero, if EF1 was the source of the IRQ
++ **
++ **
++ ** opaque is an OS provided pointer for use by the OS callbacks
++ ** e.g in Windows used to indicate DPC scheduled
+ */
-+ if (!netif->copying_receiver ||
-+ ((skb_headlen(skb) + offset_in_page(skb->data)) >= PAGE_SIZE)) {
-+ struct sk_buff *nskb = netbk_copy_skb(skb);
-+ if ( unlikely(nskb == NULL) )
-+ goto drop;
-+ /* Copy only the header fields we use in this driver. */
-+ nskb->dev = skb->dev;
-+ nskb->ip_summed = skb->ip_summed;
-+ nskb->proto_data_valid = skb->proto_data_valid;
-+ dev_kfree_skb(skb);
-+ skb = nskb;
-+ }
-+
-+ netif->rx_req_cons_peek += skb_shinfo(skb)->nr_frags + 1 +
-+ !!skb_shinfo(skb)->gso_size;
-+ netif_get(netif);
-+
-+ if (netbk_can_queue(dev) && netbk_queue_full(netif)) {
-+ netif->rx.sring->req_event = netif->rx_req_cons_peek +
-+ netbk_max_required_rx_slots(netif);
-+ mb(); /* request notification /then/ check & stop the queue */
-+ if (netbk_queue_full(netif)) {
-+ netif_stop_queue(dev);
-+ /*
-+ * Schedule 500ms timeout to restart the queue, thus
-+ * ensuring that an inactive queue will be drained.
-+ * Packets will be immediately be dropped until more
-+ * receive buffers become available (see
-+ * netbk_queue_full() check above).
-+ */
-+ netif->tx_queue_timeout.data = (unsigned long)netif;
-+ netif->tx_queue_timeout.function = tx_queue_callback;
-+ __mod_timer(&netif->tx_queue_timeout, jiffies + HZ/2);
-+ }
-+ }
-+
-+ skb_queue_tail(&rx_queue, skb);
-+ tasklet_schedule(&net_rx_tasklet);
-+
-+ return 0;
++ int (*interrupt) (struct efhw_nic *nic);
+
-+ drop:
-+ netif->stats.tx_dropped++;
-+ dev_kfree_skb(skb);
-+ return 0;
-+}
++ /*! Enable given interrupt mask for the given IRQ unit */
++ void (*interrupt_enable) (struct efhw_nic *nic, uint idx);
+
-+#if 0
-+static void xen_network_done_notify(void)
-+{
-+ static struct net_device *eth0_dev = NULL;
-+ if (unlikely(eth0_dev == NULL))
-+ eth0_dev = __dev_get_by_name("eth0");
-+ netif_rx_schedule(eth0_dev);
-+}
-+/*
-+ * Add following to poll() function in NAPI driver (Tigon3 is example):
-+ * if ( xen_network_done() )
-+ * tg3_enable_ints(tp);
-+ */
-+int xen_network_done(void)
-+{
-+ return skb_queue_empty(&rx_queue);
-+}
-+#endif
++ /*! Disable given interrupt mask for the given IRQ unit */
++ void (*interrupt_disable) (struct efhw_nic *nic, uint idx);
+
-+struct netrx_pending_operations {
-+ unsigned trans_prod, trans_cons;
-+ unsigned mmu_prod, mmu_cons;
-+ unsigned mcl_prod, mcl_cons;
-+ unsigned copy_prod, copy_cons;
-+ unsigned meta_prod, meta_cons;
-+ mmu_update_t *mmu;
-+ gnttab_transfer_t *trans;
-+ gnttab_copy_t *copy;
-+ multicall_entry_t *mcl;
-+ struct netbk_rx_meta *meta;
-+};
++ /*! Set interrupt moderation strategy for the given IRQ unit
++ ** val is in usec
++ */
++ void (*set_interrupt_moderation)(struct efhw_nic *nic,
++ uint idx, uint val);
+
-+/* Set up the grant operations for this fragment. If it's a flipping
-+ interface, we also set up the unmap request from here. */
-+static u16 netbk_gop_frag(netif_t *netif, struct netbk_rx_meta *meta,
-+ int i, struct netrx_pending_operations *npo,
-+ struct page *page, unsigned long size,
-+ unsigned long offset)
-+{
-+ mmu_update_t *mmu;
-+ gnttab_transfer_t *gop;
-+ gnttab_copy_t *copy_gop;
-+ multicall_entry_t *mcl;
-+ netif_rx_request_t *req;
-+ unsigned long old_mfn, new_mfn;
++ /*-------------- Event support ------------ */
+
-+ old_mfn = virt_to_mfn(page_address(page));
++ /*! Enable the given event queue
++ depending on the underlying implementation (EF1 or Falcon) then
++ either a q_base_addr in host memory, or a buffer base id should
++ be proivded
++ */
++ void (*event_queue_enable) (struct efhw_nic *nic,
++ uint evq, /* evnt queue index */
++ uint evq_size, /* units of #entries */
++ dma_addr_t q_base_addr, uint buf_base_id);
+
-+ req = RING_GET_REQUEST(&netif->rx, netif->rx.req_cons + i);
-+ if (netif->copying_receiver) {
-+ /* The fragment needs to be copied rather than
-+ flipped. */
-+ meta->copy = 1;
-+ copy_gop = npo->copy + npo->copy_prod++;
-+ copy_gop->flags = GNTCOPY_dest_gref;
-+ if (PageForeign(page)) {
-+ struct pending_tx_info *src_pend =
-+ &pending_tx_info[netif_page_index(page)];
-+ copy_gop->source.domid = src_pend->netif->domid;
-+ copy_gop->source.u.ref = src_pend->req.gref;
-+ copy_gop->flags |= GNTCOPY_source_gref;
-+ } else {
-+ copy_gop->source.domid = DOMID_SELF;
-+ copy_gop->source.u.gmfn = old_mfn;
-+ }
-+ copy_gop->source.offset = offset;
-+ copy_gop->dest.domid = netif->domid;
-+ copy_gop->dest.offset = 0;
-+ copy_gop->dest.u.ref = req->gref;
-+ copy_gop->len = size;
-+ } else {
-+ meta->copy = 0;
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ new_mfn = alloc_mfn();
++ /*! Disable the given event queue (and any associated timer) */
++ void (*event_queue_disable) (struct efhw_nic *nic, uint evq,
++ int timer_only);
+
-+ /*
-+ * Set the new P2M table entry before
-+ * reassigning the old data page. Heed the
-+ * comment in pgtable-2level.h:pte_page(). :-)
-+ */
-+ set_phys_to_machine(page_to_pfn(page), new_mfn);
++ /*! request wakeup from the NIC on a given event Q */
++ void (*wakeup_request) (struct efhw_nic *nic, dma_addr_t q_base_addr,
++ int next_i, int evq);
+
-+ mcl = npo->mcl + npo->mcl_prod++;
-+ MULTI_update_va_mapping(mcl,
-+ (unsigned long)page_address(page),
-+ pfn_pte_ma(new_mfn, PAGE_KERNEL),
-+ 0);
++ /*! Push a SW event on a given eventQ */
++ void (*sw_event) (struct efhw_nic *nic, int data, int evq);
+
-+ mmu = npo->mmu + npo->mmu_prod++;
-+ mmu->ptr = ((maddr_t)new_mfn << PAGE_SHIFT) |
-+ MMU_MACHPHYS_UPDATE;
-+ mmu->val = page_to_pfn(page);
-+ }
++ /*-------------- Filter support ------------ */
+
-+ gop = npo->trans + npo->trans_prod++;
-+ gop->mfn = old_mfn;
-+ gop->domid = netif->domid;
-+ gop->ref = req->gref;
-+ }
-+ return req->id;
-+}
++ /*! Setup a given filter - The software can request a filter_i,
++ * but some EtherFabric implementations will override with
++ * a more suitable index
++ */
++ int (*ipfilter_set) (struct efhw_nic *nic, int type,
++ int *filter_i, int dmaq,
++ unsigned saddr_be32, unsigned sport_be16,
++ unsigned daddr_be32, unsigned dport_be16);
+
-+static void netbk_gop_skb(struct sk_buff *skb,
-+ struct netrx_pending_operations *npo)
-+{
-+ netif_t *netif = netdev_priv(skb->dev);
-+ int nr_frags = skb_shinfo(skb)->nr_frags;
-+ int i;
-+ int extra;
-+ struct netbk_rx_meta *head_meta, *meta;
++ /*! Attach a given filter to a DMAQ */
++ void (*ipfilter_attach) (struct efhw_nic *nic, int filter_idx,
++ int dmaq_idx);
+
-+ head_meta = npo->meta + npo->meta_prod++;
-+ head_meta->frag.page_offset = skb_shinfo(skb)->gso_type;
-+ head_meta->frag.size = skb_shinfo(skb)->gso_size;
-+ extra = !!head_meta->frag.size + 1;
++ /*! Detach a filter from its DMAQ */
++ void (*ipfilter_detach) (struct efhw_nic *nic, int filter_idx);
+
-+ for (i = 0; i < nr_frags; i++) {
-+ meta = npo->meta + npo->meta_prod++;
-+ meta->frag = skb_shinfo(skb)->frags[i];
-+ meta->id = netbk_gop_frag(netif, meta, i + extra, npo,
-+ meta->frag.page,
-+ meta->frag.size,
-+ meta->frag.page_offset);
-+ }
++ /*! Clear down a given filter */
++ void (*ipfilter_clear) (struct efhw_nic *nic, int filter_idx);
+
-+ /*
-+ * This must occur at the end to ensure that we don't trash skb_shinfo
-+ * until we're done. We know that the head doesn't cross a page
-+ * boundary because such packets get copied in netif_be_start_xmit.
-+ */
-+ head_meta->id = netbk_gop_frag(netif, head_meta, 0, npo,
-+ virt_to_page(skb->data),
-+ skb_headlen(skb),
-+ offset_in_page(skb->data));
++ /*-------------- DMA support ------------ */
+
-+ netif->rx.req_cons += nr_frags + extra;
-+}
++ /*! Initialise NIC state for a given TX DMAQ */
++ void (*dmaq_tx_q_init) (struct efhw_nic *nic,
++ uint dmaq, uint evq, uint owner, uint tag,
++ uint dmaq_size, uint buf_idx, uint flags);
+
-+static inline void netbk_free_pages(int nr_frags, struct netbk_rx_meta *meta)
-+{
-+ int i;
++ /*! Initialise NIC state for a given RX DMAQ */
++ void (*dmaq_rx_q_init) (struct efhw_nic *nic,
++ uint dmaq, uint evq, uint owner, uint tag,
++ uint dmaq_size, uint buf_idx, uint flags);
+
-+ for (i = 0; i < nr_frags; i++)
-+ put_page(meta[i].frag.page);
-+}
++ /*! Disable a given TX DMAQ */
++ void (*dmaq_tx_q_disable) (struct efhw_nic *nic, uint dmaq);
+
-+/* This is a twin to netbk_gop_skb. Assume that netbk_gop_skb was
-+ used to set up the operations on the top of
-+ netrx_pending_operations, which have since been done. Check that
-+ they didn't give any errors and advance over them. */
-+static int netbk_check_gop(int nr_frags, domid_t domid,
-+ struct netrx_pending_operations *npo)
-+{
-+ multicall_entry_t *mcl;
-+ gnttab_transfer_t *gop;
-+ gnttab_copy_t *copy_op;
-+ int status = NETIF_RSP_OKAY;
-+ int i;
++ /*! Disable a given RX DMAQ */
++ void (*dmaq_rx_q_disable) (struct efhw_nic *nic, uint dmaq);
+
-+ for (i = 0; i <= nr_frags; i++) {
-+ if (npo->meta[npo->meta_cons + i].copy) {
-+ copy_op = npo->copy + npo->copy_cons++;
-+ if (copy_op->status != GNTST_okay) {
-+ DPRINTK("Bad status %d from copy to DOM%d.\n",
-+ copy_op->status, domid);
-+ status = NETIF_RSP_ERROR;
-+ }
-+ } else {
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ mcl = npo->mcl + npo->mcl_cons++;
-+ /* The update_va_mapping() must not fail. */
-+ BUG_ON(mcl->result != 0);
-+ }
++ /*! Flush a given TX DMA channel */
++ int (*flush_tx_dma_channel) (struct efhw_nic *nic, uint dmaq);
+
-+ gop = npo->trans + npo->trans_cons++;
-+ /* Check the reassignment error code. */
-+ if (gop->status != 0) {
-+ DPRINTK("Bad status %d from grant transfer to DOM%u\n",
-+ gop->status, domid);
-+ /*
-+ * Page no longer belongs to us unless
-+ * GNTST_bad_page, but that should be
-+ * a fatal error anyway.
-+ */
-+ BUG_ON(gop->status == GNTST_bad_page);
-+ status = NETIF_RSP_ERROR;
-+ }
-+ }
-+ }
++ /*! Flush a given RX DMA channel */
++ int (*flush_rx_dma_channel) (struct efhw_nic *nic, uint dmaq);
+
-+ return status;
-+}
++ /*-------------- Buffer table Support ------------ */
+
-+static void netbk_add_frag_responses(netif_t *netif, int status,
-+ struct netbk_rx_meta *meta, int nr_frags)
-+{
-+ int i;
-+ unsigned long offset;
++ /*! Initialise a buffer table page */
++ void (*buffer_table_set) (struct efhw_nic *nic,
++ dma_addr_t dma_addr,
++ uint bufsz, uint region,
++ int own_id, int buffer_id);
+
-+ for (i = 0; i < nr_frags; i++) {
-+ int id = meta[i].id;
-+ int flags = (i == nr_frags - 1) ? 0 : NETRXF_more_data;
++ /*! Initialise a block of buffer table pages */
++ void (*buffer_table_set_n) (struct efhw_nic *nic, int buffer_id,
++ dma_addr_t dma_addr,
++ uint bufsz, uint region,
++ int n_pages, int own_id);
+
-+ if (meta[i].copy)
-+ offset = 0;
-+ else
-+ offset = meta[i].frag.page_offset;
-+ make_rx_response(netif, id, status, offset,
-+ meta[i].frag.size, flags);
-+ }
-+}
++ /*! Clear a block of buffer table pages */
++ void (*buffer_table_clear) (struct efhw_nic *nic, int buffer_id,
++ int num);
+
-+static void net_rx_action(unsigned long unused)
-+{
-+ netif_t *netif = NULL;
-+ s8 status;
-+ u16 id, irq, flags;
-+ netif_rx_response_t *resp;
-+ multicall_entry_t *mcl;
-+ struct sk_buff_head rxq;
-+ struct sk_buff *skb;
-+ int notify_nr = 0;
-+ int ret;
-+ int nr_frags;
-+ int count;
-+ unsigned long offset;
++ /*! Commit a buffer table update */
++ void (*buffer_table_commit) (struct efhw_nic *nic);
+
-+ /*
-+ * Putting hundreds of bytes on the stack is considered rude.
-+ * Static works because a tasklet can only be on one CPU at any time.
-+ */
-+ static multicall_entry_t rx_mcl[NET_RX_RING_SIZE+3];
-+ static mmu_update_t rx_mmu[NET_RX_RING_SIZE];
-+ static gnttab_transfer_t grant_trans_op[NET_RX_RING_SIZE];
-+ static gnttab_copy_t grant_copy_op[NET_RX_RING_SIZE];
-+ static unsigned char rx_notify[NR_IRQS];
-+ static u16 notify_list[NET_RX_RING_SIZE];
-+ static struct netbk_rx_meta meta[NET_RX_RING_SIZE];
++};
+
-+ struct netrx_pending_operations npo = {
-+ mmu: rx_mmu,
-+ trans: grant_trans_op,
-+ copy: grant_copy_op,
-+ mcl: rx_mcl,
-+ meta: meta};
+
-+ skb_queue_head_init(&rxq);
++/*----------------------------------------------------------------------------
++ *
++ * NIC type
++ *
++ *---------------------------------------------------------------------------*/
+
-+ count = 0;
++struct efhw_device_type {
++ int arch; /* enum efhw_arch */
++ char variant; /* 'A', 'B', ... */
++ int revision; /* 0, 1, ... */
++};
+
-+ while ((skb = skb_dequeue(&rx_queue)) != NULL) {
-+ nr_frags = skb_shinfo(skb)->nr_frags;
-+ *(int *)skb->cb = nr_frags;
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap) &&
-+ !((netif_t *)netdev_priv(skb->dev))->copying_receiver &&
-+ check_mfn(nr_frags + 1)) {
-+ /* Memory squeeze? Back off for an arbitrary while. */
-+ if ( net_ratelimit() )
-+ WPRINTK("Memory squeeze in netback "
-+ "driver.\n");
-+ mod_timer(&net_timer, jiffies + HZ);
-+ skb_queue_head(&rx_queue, skb);
-+ break;
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * EtherFabric NIC instance - nic.c for HW independent functions
++ *
++ *---------------------------------------------------------------------------*/
+
-+ netbk_gop_skb(skb, &npo);
++/*! */
++struct efhw_nic {
++ /*! zero base index in efrm_nic_table.nic array */
++ volatile int index;
++ int ifindex; /*!< OS level nic index */
++#ifdef HAS_NET_NAMESPACE
++ struct net *nd_net;
++#endif
+
-+ count += nr_frags + 1;
++ struct efhw_device_type devtype;
+
-+ __skb_queue_tail(&rxq, skb);
++ /*! Options that can be set by user. */
++ unsigned options;
++# define NIC_OPT_EFTEST 0x1 /* owner is an eftest app */
+
-+ /* Filled the batch queue? */
-+ if (count + MAX_SKB_FRAGS >= NET_RX_RING_SIZE)
-+ break;
-+ }
++# define NIC_OPT_DEFAULT 0
+
-+ if (npo.mcl_prod &&
-+ !xen_feature(XENFEAT_auto_translated_physmap)) {
-+ mcl = npo.mcl + npo.mcl_prod++;
++ /*! Internal flags that indicate hardware properties at runtime. */
++ unsigned flags;
++# define NIC_FLAG_NO_INTERRUPT 0x01 /* to be set at init time only */
++# define NIC_FLAG_TRY_MSI 0x02
++# define NIC_FLAG_MSI 0x04
++# define NIC_FLAG_OS_IRQ_EN 0x08
++# define NIC_FLAG_10G 0x10
+
-+ BUG_ON(mcl[-1].op != __HYPERVISOR_update_va_mapping);
-+ mcl[-1].args[MULTI_UVMFLAGS_INDEX] = UVMF_TLB_FLUSH|UVMF_ALL;
++ unsigned mtu; /*!< MAC MTU (includes MAC hdr) */
+
-+ mcl->op = __HYPERVISOR_mmu_update;
-+ mcl->args[0] = (unsigned long)rx_mmu;
-+ mcl->args[1] = npo.mmu_prod;
-+ mcl->args[2] = 0;
-+ mcl->args[3] = DOMID_SELF;
-+ }
++ /* hardware resources */
+
-+ if (npo.trans_prod) {
-+ mcl = npo.mcl + npo.mcl_prod++;
-+ mcl->op = __HYPERVISOR_grant_table_op;
-+ mcl->args[0] = GNTTABOP_transfer;
-+ mcl->args[1] = (unsigned long)grant_trans_op;
-+ mcl->args[2] = npo.trans_prod;
-+ }
++ /*! I/O address of the start of the bar */
++ efhw_ioaddr_t bar_ioaddr;
+
-+ if (npo.copy_prod) {
-+ mcl = npo.mcl + npo.mcl_prod++;
-+ mcl->op = __HYPERVISOR_grant_table_op;
-+ mcl->args[0] = GNTTABOP_copy;
-+ mcl->args[1] = (unsigned long)grant_copy_op;
-+ mcl->args[2] = npo.copy_prod;
-+ }
++ /*! Bar number of control aperture. */
++ unsigned ctr_ap_bar;
++ /*! Length of control aperture in bytes. */
++ unsigned ctr_ap_bytes;
+
-+ /* Nothing to do? */
-+ if (!npo.mcl_prod)
-+ return;
++ uint8_t mac_addr[ETH_ALEN]; /*!< mac address */
+
-+ BUG_ON(npo.copy_prod > NET_RX_RING_SIZE);
-+ BUG_ON(npo.mmu_prod > NET_RX_RING_SIZE);
-+ BUG_ON(npo.trans_prod > NET_RX_RING_SIZE);
-+ BUG_ON(npo.mcl_prod > NET_RX_RING_SIZE+3);
-+ BUG_ON(npo.meta_prod > NET_RX_RING_SIZE);
++ /*! EtherFabric Functional Units -- functions */
++ const struct efhw_func_ops *efhw_func;
+
-+ ret = HYPERVISOR_multicall(npo.mcl, npo.mcl_prod);
-+ BUG_ON(ret != 0);
++ /* Value read from FPGA version register. Zero for asic. */
++ unsigned fpga_version;
+
-+ while ((skb = __skb_dequeue(&rxq)) != NULL) {
-+ nr_frags = *(int *)skb->cb;
++ /*! This lock protects a number of misc NIC resources. It should
++ * only be used for things that can be at the bottom of the lock
++ * order. ie. You mustn't attempt to grab any other lock while
++ * holding this one.
++ */
++ spinlock_t *reg_lock;
++ spinlock_t the_reg_lock;
+
-+ netif = netdev_priv(skb->dev);
-+ /* We can't rely on skb_release_data to release the
-+ pages used by fragments for us, since it tries to
-+ touch the pages in the fraglist. If we're in
-+ flipping mode, that doesn't work. In copying mode,
-+ we still have access to all of the pages, and so
-+ it's safe to let release_data deal with it. */
-+ /* (Freeing the fragments is safe since we copy
-+ non-linear skbs destined for flipping interfaces) */
-+ if (!netif->copying_receiver) {
-+ atomic_set(&(skb_shinfo(skb)->dataref), 1);
-+ skb_shinfo(skb)->frag_list = NULL;
-+ skb_shinfo(skb)->nr_frags = 0;
-+ netbk_free_pages(nr_frags, meta + npo.meta_cons + 1);
-+ }
++ int buf_commit_outstanding; /*!< outstanding buffer commits */
+
-+ netif->stats.tx_bytes += skb->len;
-+ netif->stats.tx_packets++;
++ /*! interrupt callbacks (hard-irq) */
++ void (*irq_handler) (struct efhw_nic *, int unit);
+
-+ status = netbk_check_gop(nr_frags, netif->domid, &npo);
++ /*! event queues per driver */
++ struct efhw_keventq evq[EFHW_KEVENTQ_MAX];
+
-+ id = meta[npo.meta_cons].id;
-+ flags = nr_frags ? NETRXF_more_data : 0;
++/* for marking when we are not using an IRQ unit
++ - 0 is a valid offset to an IRQ unit on EF1! */
++#define EFHW_IRQ_UNIT_UNUSED 0xffff
++ /*! interrupt unit in use */
++ unsigned int irq_unit[EFHW_KEVENTQ_MAX];
++ efhw_iopage_t irq_iobuff; /*!< Falcon SYSERR interrupt */
+
-+ if (skb->ip_summed == CHECKSUM_HW) /* local packet? */
-+ flags |= NETRXF_csum_blank | NETRXF_data_validated;
-+ else if (skb->proto_data_valid) /* remote but checksummed? */
-+ flags |= NETRXF_data_validated;
++ /* The new driverlink infrastructure. */
++ struct efx_dl_device *net_driver_dev;
++ struct efx_dlfilt_cb_s *dlfilter_cb;
+
-+ if (meta[npo.meta_cons].copy)
-+ offset = 0;
-+ else
-+ offset = offset_in_page(skb->data);
-+ resp = make_rx_response(netif, id, status, offset,
-+ skb_headlen(skb), flags);
++ /*! Bit masks of the sizes of event queues and dma queues supported
++ * by the nic. */
++ unsigned evq_sizes;
++ unsigned rxq_sizes;
++ unsigned txq_sizes;
+
-+ if (meta[npo.meta_cons].frag.size) {
-+ struct netif_extra_info *gso =
-+ (struct netif_extra_info *)
-+ RING_GET_RESPONSE(&netif->rx,
-+ netif->rx.rsp_prod_pvt++);
++ /* Size of filter table (including odd and even banks). */
++ unsigned filter_tbl_size;
++};
+
-+ resp->flags |= NETRXF_extra_info;
+
-+ gso->u.gso.size = meta[npo.meta_cons].frag.size;
-+ gso->u.gso.type = XEN_NETIF_GSO_TYPE_TCPV4;
-+ gso->u.gso.pad = 0;
-+ gso->u.gso.features = 0;
++#define EFHW_KVA(nic) ((nic)->bar_ioaddr)
+
-+ gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
-+ gso->flags = 0;
-+ }
+
-+ netbk_add_frag_responses(netif, status,
-+ meta + npo.meta_cons + 1,
-+ nr_frags);
++#endif /* __CI_EFHW_EFHW_TYPES_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/hardware_sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/hardware_sysdep.h 2008-05-19 00:33:48.058916291 +0300
+@@ -0,0 +1,84 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for header files
++ * with hardware-related definitions (in ci/driver/efab/hardware*).
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&netif->rx, ret);
-+ irq = netif->irq;
-+ if (ret && !rx_notify[irq]) {
-+ rx_notify[irq] = 1;
-+ notify_list[notify_nr++] = irq;
-+ }
++#ifndef __CI_EFHW_HARDWARE_LINUX_H__
++#define __CI_EFHW_HARDWARE_LINUX_H__
+
-+ if (netif_queue_stopped(netif->dev) &&
-+ netif_schedulable(netif) &&
-+ !netbk_queue_full(netif))
-+ netif_wake_queue(netif->dev);
++#include <asm/io.h>
+
-+ netif_put(netif);
-+ dev_kfree_skb(skb);
-+ npo.meta_cons += nr_frags + 1;
-+ }
++#ifdef __LITTLE_ENDIAN
++#define EFHW_IS_LITTLE_ENDIAN
++#elif __BIG_ENDIAN
++#define EFHW_IS_BIG_ENDIAN
++#else
++#error Unknown endianness
++#endif
++
++#ifndef mmiowb
++ #if defined(__i386__) || defined(__x86_64__)
++ #define mmiowb()
++ #elif defined(__ia64__)
++ #ifndef ia64_mfa
++ #define ia64_mfa() asm volatile ("mf.a" ::: "memory")
++ #endif
++ #define mmiowb ia64_mfa
++ #else
++ #error "Need definition for mmiowb()"
++ #endif
++#endif
+
-+ while (notify_nr != 0) {
-+ irq = notify_list[--notify_nr];
-+ rx_notify[irq] = 0;
-+ notify_remote_via_irq(irq);
-+ }
++typedef char *efhw_ioaddr_t;
+
-+ /* More work to do? */
-+ if (!skb_queue_empty(&rx_queue) && !timer_pending(&net_timer))
-+ tasklet_schedule(&net_rx_tasklet);
-+#if 0
-+ else
-+ xen_network_done_notify();
++#ifndef readq
++static inline uint64_t __readq(void __iomem *addr)
++{
++ return *(volatile uint64_t *)addr;
++}
++#define readq(x) __readq(x)
+#endif
++
++#ifndef writeq
++static inline void __writeq(uint64_t v, void __iomem *addr)
++{
++ *(volatile uint64_t *)addr = v;
+}
++#define writeq(val, addr) __writeq((val), (addr))
++#endif
+
-+static void net_alarm(unsigned long unused)
++#endif /* __CI_EFHW_HARDWARE_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/iopage_types.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/iopage_types.h 2008-05-19 00:33:48.058916291 +0300
+@@ -0,0 +1,188 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides efhw_page_t and efhw_iopage_t for Linux kernel.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_IOPAGE_LINUX_H__
++#define __CI_EFHW_IOPAGE_LINUX_H__
++
++#include <linux/gfp.h>
++#include <linux/hardirq.h>
++#include <ci/efhw/debug.h>
++
++/*--------------------------------------------------------------------
++ *
++ * efhw_page_t: A single page of memory. Directly mapped in the driver,
++ * and can be mapped to userlevel.
++ *
++ *--------------------------------------------------------------------*/
++
++typedef struct {
++ unsigned long kva;
++} efhw_page_t;
++
++static inline int efhw_page_alloc(efhw_page_t *p)
+{
-+ tasklet_schedule(&net_rx_tasklet);
++ p->kva = __get_free_page(in_interrupt()? GFP_ATOMIC : GFP_KERNEL);
++ return p->kva ? 0 : -ENOMEM;
+}
+
-+static void netbk_tx_pending_timeout(unsigned long unused)
++static inline int efhw_page_alloc_zeroed(efhw_page_t *p)
+{
-+ tasklet_schedule(&net_tx_tasklet);
++ p->kva = get_zeroed_page(in_interrupt()? GFP_ATOMIC : GFP_KERNEL);
++ return p->kva ? 0 : -ENOMEM;
+}
+
-+struct net_device_stats *netif_be_get_stats(struct net_device *dev)
++static inline void efhw_page_free(efhw_page_t *p)
+{
-+ netif_t *netif = netdev_priv(dev);
-+ return &netif->stats;
++ free_page(p->kva);
++ EFHW_DO_DEBUG(memset(p, 0, sizeof(*p)));
+}
+
-+static int __on_net_schedule_list(netif_t *netif)
++static inline char *efhw_page_ptr(efhw_page_t *p)
+{
-+ return netif->list.next != NULL;
++ return (char *)p->kva;
+}
+
-+static void remove_from_net_schedule_list(netif_t *netif)
++static inline unsigned efhw_page_pfn(efhw_page_t *p)
+{
-+ spin_lock_irq(&net_schedule_list_lock);
-+ if (likely(__on_net_schedule_list(netif))) {
-+ list_del(&netif->list);
-+ netif->list.next = NULL;
-+ netif_put(netif);
-+ }
-+ spin_unlock_irq(&net_schedule_list_lock);
++ return (unsigned)(__pa(p->kva) >> PAGE_SHIFT);
+}
+
-+static void add_to_net_schedule_list_tail(netif_t *netif)
++static inline void efhw_page_mark_invalid(efhw_page_t *p)
+{
-+ if (__on_net_schedule_list(netif))
-+ return;
++ p->kva = 0;
++}
+
-+ spin_lock_irq(&net_schedule_list_lock);
-+ if (!__on_net_schedule_list(netif) &&
-+ likely(netif_schedulable(netif))) {
-+ list_add_tail(&netif->list, &net_schedule_list);
-+ netif_get(netif);
-+ }
-+ spin_unlock_irq(&net_schedule_list_lock);
++static inline int efhw_page_is_valid(efhw_page_t *p)
++{
++ return p->kva != 0;
+}
+
-+/*
-+ * Note on CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER:
-+ * If this driver is pipelining transmit requests then we can be very
-+ * aggressive in avoiding new-packet notifications -- frontend only needs to
-+ * send a notification if there are no outstanding unreceived responses.
-+ * If we may be buffer transmit buffers for any reason then we must be rather
-+ * more conservative and treat this as the final check for pending work.
-+ */
-+void netif_schedule_work(netif_t *netif)
++static inline void efhw_page_init_from_va(efhw_page_t *p, void *va)
+{
-+ int more_to_do;
++ p->kva = (unsigned long)va;
++}
+
-+#ifdef CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER
-+ more_to_do = RING_HAS_UNCONSUMED_REQUESTS(&netif->tx);
-+#else
-+ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, more_to_do);
-+#endif
++/*--------------------------------------------------------------------
++ *
++ * efhw_iopage_t: A single page of memory. Directly mapped in the driver,
++ * and can be mapped to userlevel. Can also be accessed by the NIC.
++ *
++ *--------------------------------------------------------------------*/
+
-+ if (more_to_do) {
-+ add_to_net_schedule_list_tail(netif);
-+ maybe_schedule_tx_action();
-+ }
-+}
++typedef struct {
++ efhw_page_t p;
++ dma_addr_t dma_addr;
++} efhw_iopage_t;
+
-+void netif_deschedule_work(netif_t *netif)
++static inline dma_addr_t efhw_iopage_dma_addr(efhw_iopage_t *p)
+{
-+ remove_from_net_schedule_list(netif);
++ return p->dma_addr;
+}
+
++#define efhw_iopage_ptr(iop) efhw_page_ptr(&(iop)->p)
++#define efhw_iopage_pfn(iop) efhw_page_pfn(&(iop)->p)
++#define efhw_iopage_mark_invalid(iop) efhw_page_mark_invalid(&(iop)->p)
++#define efhw_iopage_is_valid(iop) efhw_page_is_valid(&(iop)->p)
+
-+static void tx_add_credit(netif_t *netif)
++/*--------------------------------------------------------------------
++ *
++ * efhw_iopages_t: A set of pages that are contiguous in physical memory.
++ * Directly mapped in the driver, and can be mapped to userlevel. Can also
++ * be accessed by the NIC.
++ *
++ * NB. The O/S may be unwilling to allocate many, or even any of these. So
++ * only use this type where the NIC really needs a physically contiguous
++ * buffer.
++ *
++ *--------------------------------------------------------------------*/
++
++typedef struct {
++ caddr_t kva;
++ unsigned order;
++ dma_addr_t dma_addr;
++} efhw_iopages_t;
++
++static inline caddr_t efhw_iopages_ptr(efhw_iopages_t *p)
+{
-+ unsigned long max_burst, max_credit;
++ return p->kva;
++}
+
-+ /*
-+ * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
-+ * Otherwise the interface can seize up due to insufficient credit.
-+ */
-+ max_burst = RING_GET_REQUEST(&netif->tx, netif->tx.req_cons)->size;
-+ max_burst = min(max_burst, 131072UL);
-+ max_burst = max(max_burst, netif->credit_bytes);
++static inline unsigned efhw_iopages_pfn(efhw_iopages_t *p)
++{
++ return (unsigned)(__pa(p->kva) >> PAGE_SHIFT);
++}
+
-+ /* Take care that adding a new chunk of credit doesn't wrap to zero. */
-+ max_credit = netif->remaining_credit + netif->credit_bytes;
-+ if (max_credit < netif->remaining_credit)
-+ max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
++static inline dma_addr_t efhw_iopages_dma_addr(efhw_iopages_t *p)
++{
++ return p->dma_addr;
++}
+
-+ netif->remaining_credit = min(max_credit, max_burst);
++static inline unsigned efhw_iopages_size(efhw_iopages_t *p)
++{
++ return 1u << (p->order + PAGE_SHIFT);
+}
+
-+static void tx_credit_callback(unsigned long data)
++/* efhw_iopage_t <-> efhw_iopages_t conversions for handling physically
++ * contiguous allocations in iobufsets for iSCSI. This allows the
++ * essential information about contiguous allocations from
++ * efhw_iopages_alloc() to be saved away in the efhw_iopage_t array in an
++ * iobufset. (Changing the iobufset resource to use a union type would
++ * involve a lot of code changes, and make the iobufset's metadata larger
++ * which could be bad as it's supposed to fit into a single page on some
++ * platforms.)
++ */
++static inline void
++efhw_iopage_init_from_iopages(efhw_iopage_t *iopage,
++ efhw_iopages_t *iopages, unsigned pageno)
+{
-+ netif_t *netif = (netif_t *)data;
-+ tx_add_credit(netif);
-+ netif_schedule_work(netif);
++ iopage->p.kva = ((unsigned long)efhw_iopages_ptr(iopages))
++ + (pageno * PAGE_SIZE);
++ iopage->dma_addr = efhw_iopages_dma_addr(iopages) +
++ (pageno * PAGE_SIZE);
+}
+
-+static inline int copy_pending_req(PEND_RING_IDX pending_idx)
++static inline void
++efhw_iopages_init_from_iopage(efhw_iopages_t *iopages,
++ efhw_iopage_t *iopage, unsigned order)
+{
-+ return gnttab_copy_grant_page(grant_tx_handle[pending_idx],
-+ &mmap_pages[pending_idx]);
++ iopages->kva = (caddr_t) efhw_iopage_ptr(iopage);
++ EFHW_ASSERT(iopages->kva);
++ iopages->order = order;
++ iopages->dma_addr = efhw_iopage_dma_addr(iopage);
+}
+
-+inline static void net_tx_action_dealloc(void)
++#endif /* __CI_EFHW_IOPAGE_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/public.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/public.h 2008-05-19 00:33:48.058916291 +0300
+@@ -0,0 +1,83 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API of efhw library exported from the SFC
++ * resource driver.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef __CI_EFHW_PUBLIC_H__
++#define __CI_EFHW_PUBLIC_H__
++
++#include <ci/efhw/common.h>
++#include <ci/efhw/efhw_types.h>
++
++/*! Returns true if we have some EtherFabric functional units -
++ whether configured or not */
++static inline int efhw_nic_have_functional_units(struct efhw_nic *nic)
+{
-+ struct netbk_tx_pending_inuse *inuse, *n;
-+ gnttab_unmap_grant_ref_t *gop;
-+ u16 pending_idx;
-+ PEND_RING_IDX dc, dp;
-+ netif_t *netif;
-+ int ret;
-+ LIST_HEAD(list);
++ return nic->efhw_func != 0;
++}
+
-+ dc = dealloc_cons;
-+ gop = tx_unmap_ops;
++/*! Returns true if the EtherFabric functional units have been configured */
++static inline int efhw_nic_have_hw(struct efhw_nic *nic)
++{
++ return efhw_nic_have_functional_units(nic) && (EFHW_KVA(nic) != 0);
++}
+
-+ /*
-+ * Free up any grants we have finished using
-+ */
-+ do {
-+ dp = dealloc_prod;
++/*! Helper function to allocate the iobuffer needed by an eventq
++ * - it ensures the eventq has the correct alignment for the NIC
++ *
++ * \param rm Event-queue resource manager
++ * \param instance Event-queue instance (index)
++ * \param buf_bytes Requested size of eventq
++ * \return < 0 if iobuffer allocation fails
++ */
++int efhw_nic_event_queue_alloc_iobuffer(struct efhw_nic *nic,
++ struct eventq_resource_hardware *h,
++ int evq_instance, unsigned buf_bytes);
+
-+ /* Ensure we see all indices enqueued by netif_idx_release(). */
-+ smp_rmb();
++extern void falcon_nic_set_rx_usr_buf_size(struct efhw_nic *,
++ int rx_usr_buf_size);
+
-+ while (dc != dp) {
-+ unsigned long pfn;
++extern void
++falcon_nic_rx_filter_ctl_set(struct efhw_nic *nic, uint32_t tcp_full,
++ uint32_t tcp_wild,
++ uint32_t udp_full, uint32_t udp_wild);
+
-+ pending_idx = dealloc_ring[MASK_PEND_IDX(dc++)];
-+ list_move_tail(&pending_inuse[pending_idx].list, &list);
++extern void
++falcon_nic_rx_filter_ctl_get(struct efhw_nic *nic, uint32_t *tcp_full,
++ uint32_t *tcp_wild,
++ uint32_t *udp_full, uint32_t *udp_wild);
+
-+ pfn = idx_to_pfn(pending_idx);
-+ /* Already unmapped? */
-+ if (!phys_to_machine_mapping_valid(pfn))
-+ continue;
++#endif /* __CI_EFHW_PUBLIC_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efhw/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efhw/sysdep.h 2008-05-19 00:33:48.062916522 +0300
+@@ -0,0 +1,72 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for efhw library.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ gnttab_set_unmap_op(gop, idx_to_kaddr(pending_idx),
-+ GNTMAP_host_map,
-+ grant_tx_handle[pending_idx]);
-+ gop++;
-+ }
++#ifndef __CI_EFHW_SYSDEP_LINUX_H__
++#define __CI_EFHW_SYSDEP_LINUX_H__
+
-+ if (netbk_copy_skb_mode != NETBK_DELAYED_COPY_SKB ||
-+ list_empty(&pending_inuse_head))
-+ break;
++#include <linux/version.h>
++#include <linux/module.h>
++#include <linux/spinlock.h>
++#include <linux/delay.h>
++#include <linux/if_ether.h>
+
-+ /* Copy any entries that have been pending for too long. */
-+ list_for_each_entry_safe(inuse, n, &pending_inuse_head, list) {
-+ if (time_after(inuse->alloc_time + HZ / 2, jiffies))
-+ break;
++#include <linux/netdevice.h> /* necessary for etherdevice.h on some kernels */
++#include <linux/etherdevice.h>
+
-+ switch (copy_pending_req(inuse - pending_inuse)) {
-+ case 0:
-+ list_move_tail(&inuse->list, &list);
-+ continue;
-+ case -EBUSY:
-+ list_del_init(&inuse->list);
-+ continue;
-+ case -ENOENT:
-+ continue;
-+ }
++#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,21)
++static inline int is_local_ether_addr(const u8 *addr)
++{
++ return (0x02 & addr[0]);
++}
++#endif
+
-+ break;
-+ }
-+ } while (dp != dealloc_prod);
++typedef unsigned long irq_flags_t;
+
-+ dealloc_cons = dc;
++#define spin_lock_destroy(l_) do {} while (0)
+
-+ ret = HYPERVISOR_grant_table_op(
-+ GNTTABOP_unmap_grant_ref, tx_unmap_ops, gop - tx_unmap_ops);
-+ BUG_ON(ret);
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24)
++#define HAS_NET_NAMESPACE
++#endif
+
-+ list_for_each_entry_safe(inuse, n, &list, list) {
-+ pending_idx = inuse - pending_inuse;
++/* Funny, but linux has round_up for x86 only, defined in
++ * x86-specific header */
++#ifndef round_up
++#define round_up(x, y) (((x) + (y) - 1) & ~((y)-1))
++#endif
+
-+ netif = pending_tx_info[pending_idx].netif;
++#endif /* __CI_EFHW_SYSDEP_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efrm/nic_table.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efrm/nic_table.h 2008-05-19 00:33:48.070916983 +0300
+@@ -0,0 +1,98 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides public API for NIC table.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ make_tx_response(netif, &pending_tx_info[pending_idx].req,
-+ NETIF_RSP_OKAY);
++#ifndef __CI_EFRM_NIC_TABLE_H__
++#define __CI_EFRM_NIC_TABLE_H__
+
-+ /* Ready for next use. */
-+ gnttab_reset_grant_page(mmap_pages[pending_idx]);
++#include <ci/efhw/efhw_types.h>
++#include <ci/efrm/sysdep.h>
+
-+ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
++/*--------------------------------------------------------------------
++ *
++ * struct efrm_nic_table - top level driver object keeping all NICs -
++ * implemented in driver_object.c
++ *
++ *--------------------------------------------------------------------*/
+
-+ netif_put(netif);
++/*! Comment? */
++struct efrm_nic_table {
++ /*! nics attached to this driver */
++ struct efhw_nic *nic[EFHW_MAX_NR_DEVS];
++ /*! pointer to an arbitrary struct efhw_nic if one exists;
++ * for code which does not care which NIC it wants but
++ * still needs one. Note you cannot assume nic[0] exists. */
++ struct efhw_nic *a_nic;
++ uint32_t nic_count; /*!< number of nics attached to this driver */
++ spinlock_t lock; /*!< lock for table modifications */
++ atomic_t ref_count; /*!< refcount for users of nic table */
++};
+
-+ list_del_init(&inuse->list);
-+ }
-+}
++/* Resource driver structures used by other drivers as well */
++extern struct efrm_nic_table efrm_nic_table;
+
-+static void netbk_tx_err(netif_t *netif, netif_tx_request_t *txp, RING_IDX end)
++static inline void efrm_nic_table_hold(void)
+{
-+ RING_IDX cons = netif->tx.req_cons;
++ atomic_inc(&efrm_nic_table.ref_count);
++}
+
-+ do {
-+ make_tx_response(netif, txp, NETIF_RSP_ERROR);
-+ if (cons >= end)
-+ break;
-+ txp = RING_GET_REQUEST(&netif->tx, cons++);
-+ } while (1);
-+ netif->tx.req_cons = cons;
-+ netif_schedule_work(netif);
-+ netif_put(netif);
++static inline void efrm_nic_table_rele(void)
++{
++ atomic_dec(&efrm_nic_table.ref_count);
+}
+
-+static int netbk_count_requests(netif_t *netif, netif_tx_request_t *first,
-+ netif_tx_request_t *txp, int work_to_do)
++static inline int efrm_nic_table_held(void)
+{
-+ RING_IDX cons = netif->tx.req_cons;
-+ int frags = 0;
++ return (atomic_read(&efrm_nic_table.ref_count) != 0);
++}
+
-+ if (!(first->flags & NETTXF_more_data))
-+ return 0;
++/* Run code block _x multiple times with variable nic set to each
++ * registered NIC in turn.
++ * DO NOT "break" out of this loop early. */
++#define EFRM_FOR_EACH_NIC(_nic_i, _nic) \
++ for ((_nic_i) = (efrm_nic_table_hold(), 0); \
++ (_nic_i) < EFHW_MAX_NR_DEVS || (efrm_nic_table_rele(), 0); \
++ (_nic_i)++) \
++ if (((_nic) = efrm_nic_table.nic[_nic_i]))
+
-+ do {
-+ if (frags >= work_to_do) {
-+ DPRINTK("Need more frags\n");
-+ return -frags;
-+ }
++#define EFRM_FOR_EACH_NIC_IN_SET(_set, _i, _nic) \
++ for ((_i) = (efrm_nic_table_hold(), 0); \
++ (_i) < EFHW_MAX_NR_DEVS || (efrm_nic_table_rele(), 0); \
++ ++(_i)) \
++ if (((_nic) = efrm_nic_table.nic[_i]) && \
++ efrm_nic_set_read((_set), (_i)))
+
-+ if (unlikely(frags >= MAX_SKB_FRAGS)) {
-+ DPRINTK("Too many frags\n");
-+ return -frags;
-+ }
++#endif /* __CI_EFRM_NIC_TABLE_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efrm/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efrm/sysdep.h 2008-05-19 00:33:48.426937504 +0300
+@@ -0,0 +1,54 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides Linux-like system-independent API for efrm library.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ memcpy(txp, RING_GET_REQUEST(&netif->tx, cons + frags),
-+ sizeof(*txp));
-+ if (txp->size > first->size) {
-+ DPRINTK("Frags galore\n");
-+ return -frags;
-+ }
++#ifndef __CI_EFRM_SYSDEP_H__
++#define __CI_EFRM_SYSDEP_H__
+
-+ first->size -= txp->size;
-+ frags++;
++/* Spinlocks are defined in efhw/sysdep.h */
++#include <ci/efhw/sysdep.h>
+
-+ if (unlikely((txp->offset + txp->size) > PAGE_SIZE)) {
-+ DPRINTK("txp->offset: %x, size: %u\n",
-+ txp->offset, txp->size);
-+ return -frags;
-+ }
-+ } while ((txp++)->flags & NETTXF_more_data);
++#if defined(__linux__) && defined(__KERNEL__)
+
-+ return frags;
-+}
++# include <ci/efrm/sysdep_linux.h>
+
-+static gnttab_map_grant_ref_t *netbk_get_requests(netif_t *netif,
-+ struct sk_buff *skb,
-+ netif_tx_request_t *txp,
-+ gnttab_map_grant_ref_t *mop)
-+{
-+ struct skb_shared_info *shinfo = skb_shinfo(skb);
-+ skb_frag_t *frags = shinfo->frags;
-+ unsigned long pending_idx = *((u16 *)skb->data);
-+ int i, start;
++#else
+
-+ /* Skip first skb fragment if it is on same page as header fragment. */
-+ start = ((unsigned long)shinfo->frags[0].page == pending_idx);
++# include <ci/efrm/sysdep_ci2linux.h>
+
-+ for (i = start; i < shinfo->nr_frags; i++, txp++) {
-+ pending_idx = pending_ring[MASK_PEND_IDX(pending_cons++)];
++#endif
+
-+ gnttab_set_map_op(mop++, idx_to_kaddr(pending_idx),
-+ GNTMAP_host_map | GNTMAP_readonly,
-+ txp->gref, netif->domid);
++#endif /* __CI_EFRM_SYSDEP_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/efrm/sysdep_linux.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/efrm/sysdep_linux.h 2008-05-19 00:33:48.426937504 +0300
+@@ -0,0 +1,248 @@
++/****************************************************************************
++ * Driver for Solarflare network controllers -
++ * resource management for Xen backend, OpenOnload, etc
++ * (including support for SFE4001 10GBT NIC)
++ *
++ * This file provides version-independent Linux kernel API for efrm library.
++ * Only kernels >=2.6.9 are supported.
++ *
++ * Copyright 2005-2007: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Kfifo API is partially stolen from linux-2.6.22/include/linux/list.h
++ * Copyright (C) 2004 Stelian Pop <stelian@popies.net>
++ *
++ * Developed and maintained by Solarflare Communications:
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * Certain parts of the driver were implemented by
++ * Alexandra Kossovsky <Alexandra.Kossovsky@oktetlabs.ru>
++ * OKTET Labs Ltd, Russia,
++ * http://oktetlabs.ru, <info@oktetlabs.ru>
++ * by request of Solarflare Communications
++ *
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ memcpy(&pending_tx_info[pending_idx].req, txp, sizeof(*txp));
-+ netif_get(netif);
-+ pending_tx_info[pending_idx].netif = netif;
-+ frags[i].page = (void *)pending_idx;
-+ }
++#ifndef __CI_EFRM_SYSDEP_LINUX_H__
++#define __CI_EFRM_SYSDEP_LINUX_H__
+
-+ return mop;
++#include <linux/version.h>
++#include <linux/list.h>
++#include <linux/vmalloc.h>
++#include <linux/errno.h>
++#include <linux/string.h>
++#include <linux/workqueue.h>
++#include <linux/gfp.h>
++#include <linux/slab.h>
++#include <linux/hardirq.h>
++#include <linux/kernel.h>
++#include <linux/if_ether.h>
++#include <linux/completion.h>
++#include <linux/in.h>
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++/* get roundup_pow_of_two(), which was in kernel.h in early kernel versions */
++#include <linux/log2.h>
++#endif
++
++/********************************************************************
++ *
++ * List API
++ *
++ ********************************************************************/
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18)
++static inline void
++list_replace_init(struct list_head *old, struct list_head *new)
++{
++ new->next = old->next;
++ new->next->prev = new;
++ new->prev = old->prev;
++ new->prev->next = new;
++ INIT_LIST_HEAD(old);
+}
++#endif
+
-+static int netbk_tx_check_mop(struct sk_buff *skb,
-+ gnttab_map_grant_ref_t **mopp)
++static inline struct list_head *list_pop(struct list_head *list)
+{
-+ gnttab_map_grant_ref_t *mop = *mopp;
-+ int pending_idx = *((u16 *)skb->data);
-+ netif_t *netif = pending_tx_info[pending_idx].netif;
-+ netif_tx_request_t *txp;
-+ struct skb_shared_info *shinfo = skb_shinfo(skb);
-+ int nr_frags = shinfo->nr_frags;
-+ int i, err, start;
++ struct list_head *link = list->next;
++ list_del(link);
++ return link;
++}
+
-+ /* Check status of header. */
-+ err = mop->status;
-+ if (unlikely(err)) {
-+ txp = &pending_tx_info[pending_idx].req;
-+ make_tx_response(netif, txp, NETIF_RSP_ERROR);
-+ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
-+ netif_put(netif);
-+ } else {
-+ set_phys_to_machine(
-+ __pa(idx_to_kaddr(pending_idx)) >> PAGE_SHIFT,
-+ FOREIGN_FRAME(mop->dev_bus_addr >> PAGE_SHIFT));
-+ grant_tx_handle[pending_idx] = mop->handle;
-+ }
++static inline struct list_head *list_pop_tail(struct list_head *list)
++{
++ struct list_head *link = list->prev;
++ list_del(link);
++ return link;
++}
+
-+ /* Skip first skb fragment if it is on same page as header fragment. */
-+ start = ((unsigned long)shinfo->frags[0].page == pending_idx);
++/********************************************************************
++ *
++ * Workqueue API
++ *
++ ********************************************************************/
+
-+ for (i = start; i < nr_frags; i++) {
-+ int j, newerr;
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)
++#define NEED_OLD_WORK_API
+
-+ pending_idx = (unsigned long)shinfo->frags[i].page;
++/**
++ * The old and new work function prototypes just change
++ * the type of the pointer in the only argument, so it's
++ * safe to cast one function type to the other
++ */
++typedef void (*efrm_old_work_func_t) (void *p);
+
-+ /* Check error status: if okay then remember grant handle. */
-+ newerr = (++mop)->status;
-+ if (likely(!newerr)) {
-+ set_phys_to_machine(
-+ __pa(idx_to_kaddr(pending_idx))>>PAGE_SHIFT,
-+ FOREIGN_FRAME(mop->dev_bus_addr>>PAGE_SHIFT));
-+ grant_tx_handle[pending_idx] = mop->handle;
-+ /* Had a previous error? Invalidate this fragment. */
-+ if (unlikely(err))
-+ netif_idx_release(pending_idx);
-+ continue;
-+ }
++#undef INIT_WORK
++#define INIT_WORK(_work, _func) \
++ do { \
++ INIT_LIST_HEAD(&(_work)->entry); \
++ (_work)->pending = 0; \
++ PREPARE_WORK((_work), \
++ (efrm_old_work_func_t) (_func), \
++ (_work)); \
++ } while (0)
+
-+ /* Error on this fragment: respond to client with an error. */
-+ txp = &pending_tx_info[pending_idx].req;
-+ make_tx_response(netif, txp, NETIF_RSP_ERROR);
-+ pending_ring[MASK_PEND_IDX(pending_prod++)] = pending_idx;
-+ netif_put(netif);
++#endif
+
-+ /* Not the first error? Preceding frags already invalidated. */
-+ if (err)
-+ continue;
++/********************************************************************
++ *
++ * Kfifo API
++ *
++ ********************************************************************/
+
-+ /* First error: invalidate header and preceding fragments. */
-+ pending_idx = *((u16 *)skb->data);
-+ netif_idx_release(pending_idx);
-+ for (j = start; j < i; j++) {
-+ pending_idx = (unsigned long)shinfo->frags[i].page;
-+ netif_idx_release(pending_idx);
-+ }
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,10)
+
-+ /* Remember the error: invalidate all subsequent fragments. */
-+ err = newerr;
-+ }
++#if !defined(RHEL_RELEASE_CODE) || (RHEL_RELEASE_CODE < 1029)
++typedef unsigned gfp_t;
++#endif
+
-+ *mopp = mop + 1;
-+ return err;
-+}
++#define HAS_NO_KFIFO
+
-+static void netbk_fill_frags(struct sk_buff *skb)
-+{
-+ struct skb_shared_info *shinfo = skb_shinfo(skb);
-+ int nr_frags = shinfo->nr_frags;
-+ int i;
++struct kfifo {
++ unsigned char *buffer; /* the buffer holding the data */
++ unsigned int size; /* the size of the allocated buffer */
++ unsigned int in; /* data is added at offset (in % size) */
++ unsigned int out; /* data is extracted from off. (out % size) */
++ spinlock_t *lock; /* protects concurrent modifications */
++};
+
-+ for (i = 0; i < nr_frags; i++) {
-+ skb_frag_t *frag = shinfo->frags + i;
-+ netif_tx_request_t *txp;
-+ unsigned long pending_idx;
++extern struct kfifo *kfifo_init(unsigned char *buffer, unsigned int size,
++ gfp_t gfp_mask, spinlock_t *lock);
++extern struct kfifo *kfifo_alloc(unsigned int size, gfp_t gfp_mask,
++ spinlock_t *lock);
++extern void kfifo_free(struct kfifo *fifo);
++extern unsigned int __kfifo_put(struct kfifo *fifo,
++ unsigned char *buffer, unsigned int len);
++extern unsigned int __kfifo_get(struct kfifo *fifo,
++ unsigned char *buffer, unsigned int len);
+
-+ pending_idx = (unsigned long)frag->page;
++/**
++ * kfifo_put - puts some data into the FIFO
++ * @fifo: the fifo to be used.
++ * @buffer: the data to be added.
++ * @len: the length of the data to be added.
++ *
++ * This function copies at most @len bytes from the @buffer into
++ * the FIFO depending on the free space, and returns the number of
++ * bytes copied.
++ */
++static inline unsigned int
++kfifo_put(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
++{
++ unsigned long flags;
++ unsigned int ret;
+
-+ pending_inuse[pending_idx].alloc_time = jiffies;
-+ list_add_tail(&pending_inuse[pending_idx].list,
-+ &pending_inuse_head);
++ spin_lock_irqsave(fifo->lock, flags);
+
-+ txp = &pending_tx_info[pending_idx].req;
-+ frag->page = virt_to_page(idx_to_kaddr(pending_idx));
-+ frag->size = txp->size;
-+ frag->page_offset = txp->offset;
++ ret = __kfifo_put(fifo, buffer, len);
+
-+ skb->len += txp->size;
-+ skb->data_len += txp->size;
-+ skb->truesize += txp->size;
-+ }
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
+}
+
-+int netbk_get_extras(netif_t *netif, struct netif_extra_info *extras,
-+ int work_to_do)
++/**
++ * kfifo_get - gets some data from the FIFO
++ * @fifo: the fifo to be used.
++ * @buffer: where the data must be copied.
++ * @len: the size of the destination buffer.
++ *
++ * This function copies at most @len bytes from the FIFO into the
++ * @buffer and returns the number of copied bytes.
++ */
++static inline unsigned int
++kfifo_get(struct kfifo *fifo, unsigned char *buffer, unsigned int len)
+{
-+ struct netif_extra_info extra;
-+ RING_IDX cons = netif->tx.req_cons;
++ unsigned long flags;
++ unsigned int ret;
+
-+ do {
-+ if (unlikely(work_to_do-- <= 0)) {
-+ DPRINTK("Missing extra info\n");
-+ return -EBADR;
-+ }
++ spin_lock_irqsave(fifo->lock, flags);
+
-+ memcpy(&extra, RING_GET_REQUEST(&netif->tx, cons),
-+ sizeof(extra));
-+ if (unlikely(!extra.type ||
-+ extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
-+ netif->tx.req_cons = ++cons;
-+ DPRINTK("Invalid extra type: %d\n", extra.type);
-+ return -EINVAL;
-+ }
++ ret = __kfifo_get(fifo, buffer, len);
+
-+ memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
-+ netif->tx.req_cons = ++cons;
-+ } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
++ /*
++ * optimization: if the FIFO is empty, set the indices to 0
++ * so we don't wrap the next time
++ */
++ if (fifo->in == fifo->out)
++ fifo->in = fifo->out = 0;
+
-+ return work_to_do;
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
+}
+
-+static int netbk_set_skb_gso(struct sk_buff *skb, struct netif_extra_info *gso)
++/**
++ * __kfifo_len - returns the number of bytes available in the FIFO, no locking version
++ * @fifo: the fifo to be used.
++ */
++static inline unsigned int __kfifo_len(struct kfifo *fifo)
+{
-+ if (!gso->u.gso.size) {
-+ DPRINTK("GSO size must not be zero.\n");
-+ return -EINVAL;
-+ }
++ return fifo->in - fifo->out;
++}
+
-+ /* Currently only TCPv4 S.O. is supported. */
-+ if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4) {
-+ DPRINTK("Bad GSO type %d.\n", gso->u.gso.type);
-+ return -EINVAL;
-+ }
++/**
++ * kfifo_len - returns the number of bytes available in the FIFO
++ * @fifo: the fifo to be used.
++ */
++static inline unsigned int kfifo_len(struct kfifo *fifo)
++{
++ unsigned long flags;
++ unsigned int ret;
+
-+ skb_shinfo(skb)->gso_size = gso->u.gso.size;
-+ skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
++ spin_lock_irqsave(fifo->lock, flags);
+
-+ /* Header must be checked, and gso_segs computed. */
-+ skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
-+ skb_shinfo(skb)->gso_segs = 0;
++ ret = __kfifo_len(fifo);
+
-+ return 0;
++ spin_unlock_irqrestore(fifo->lock, flags);
++
++ return ret;
+}
+
-+/* Called after netfront has transmitted */
-+static void net_tx_action(unsigned long unused)
++#else
++#include <linux/kfifo.h>
++#endif
++
++static inline void kfifo_vfree(struct kfifo *fifo)
+{
-+ struct list_head *ent;
-+ struct sk_buff *skb;
-+ netif_t *netif;
-+ netif_tx_request_t txreq;
-+ netif_tx_request_t txfrags[MAX_SKB_FRAGS];
-+ struct netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX - 1];
-+ u16 pending_idx;
-+ RING_IDX i;
-+ gnttab_map_grant_ref_t *mop;
-+ unsigned int data_len;
-+ int ret, work_to_do;
++ vfree(fifo->buffer);
++ kfree(fifo);
++}
+
-+ if (dealloc_cons != dealloc_prod)
-+ net_tx_action_dealloc();
++#endif /* __CI_EFRM_SYSDEP_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/config.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/config.h 2008-05-19 00:33:48.478940502 +0300
+@@ -0,0 +1,49 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ mop = tx_map_ops;
-+ while (((NR_PENDING_REQS + MAX_SKB_FRAGS) < MAX_PENDING_REQS) &&
-+ !list_empty(&net_schedule_list)) {
-+ /* Get a netif from the list with work to do. */
-+ ent = net_schedule_list.next;
-+ netif = list_entry(ent, netif_t, list);
-+ netif_get(netif);
-+ remove_from_net_schedule_list(netif);
++/*! \cidoxg_include_ci_tools */
+
-+ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, work_to_do);
-+ if (!work_to_do) {
-+ netif_put(netif);
-+ continue;
-+ }
++#ifndef __CI_TOOLS_CONFIG_H__
++#define __CI_TOOLS_CONFIG_H__
+
-+ i = netif->tx.req_cons;
-+ rmb(); /* Ensure that we see the request before we copy it. */
-+ memcpy(&txreq, RING_GET_REQUEST(&netif->tx, i), sizeof(txreq));
+
-+ /* Credit-based scheduling. */
-+ if (txreq.size > netif->remaining_credit) {
-+ unsigned long now = jiffies;
-+ unsigned long next_credit =
-+ netif->credit_timeout.expires +
-+ msecs_to_jiffies(netif->credit_usec / 1000);
++/**********************************************************************
++ * Debugging.
++ */
+
-+ /* Timer could already be pending in rare cases. */
-+ if (timer_pending(&netif->credit_timeout)) {
-+ netif_put(netif);
-+ continue;
-+ }
++#define CI_INCLUDE_ASSERT_VALID 0
+
-+ /* Passed the point where we can replenish credit? */
-+ if (time_after_eq(now, next_credit)) {
-+ netif->credit_timeout.expires = now;
-+ tx_add_credit(netif);
-+ }
++/* Set non-zero to allow info about who has allocated what to appear in
++ * /proc/drivers/level5/mem.
++ * However - Note that doing so can lead to segfault when you unload the
++ * driver, and other weirdness. i.e. I don't think the code for is quite
++ * right (written by Oktet, hacked by gel), but it does work well enough to be
++ * useful.
++ */
++#define CI_MEMLEAK_DEBUG_ALLOC_TABLE 0
+
-+ /* Still too big to send right now? Set a callback. */
-+ if (txreq.size > netif->remaining_credit) {
-+ netif->credit_timeout.data =
-+ (unsigned long)netif;
-+ netif->credit_timeout.function =
-+ tx_credit_callback;
-+ __mod_timer(&netif->credit_timeout,
-+ next_credit);
-+ netif_put(netif);
-+ continue;
-+ }
-+ }
-+ netif->remaining_credit -= txreq.size;
+
-+ work_to_do--;
-+ netif->tx.req_cons = ++i;
++#endif /* __CI_TOOLS_CONFIG_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/debug.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/debug.h 2008-05-19 00:33:48.518942808 +0300
+@@ -0,0 +1,336 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*! \cidoxg_include_ci_tools */
++
++#ifndef __CI_TOOLS_DEBUG_H__
++#define __CI_TOOLS_DEBUG_H__
++
++#define CI_LOG_E(x) x /* errors */
++#define CI_LOG_W(x) x /* warnings */
++#define CI_LOG_I(x) x /* information */
++#define CI_LOG_V(x) x /* verbose */
++
++/* Build time asserts. We paste the line number into the type name
++ * so that the macro can be used more than once per file even if the
++ * compiler objects to multiple identical typedefs. Collisions
++ * between use in different header files is still possible. */
++#ifndef CI_BUILD_ASSERT
++#define __CI_BUILD_ASSERT_NAME(_x) __CI_BUILD_ASSERT_ILOATHECPP(_x)
++#define __CI_BUILD_ASSERT_ILOATHECPP(_x) __CI_BUILD_ASSERT__ ##_x
++#define CI_BUILD_ASSERT(e)\
++ typedef char __CI_BUILD_ASSERT_NAME(__LINE__)[(e)?1:-1]
++#endif
++
++
++#ifdef NDEBUG
++
++# define _ci_check(exp, file, line)
++# define _ci_assert2(e, x, y, file, line)
++# define _ci_assert(exp, file, line)
++# define _ci_assert_equal(exp1, exp2, file, line)
++# define _ci_assert_equiv(exp1, exp2, file, line)
++# define _ci_assert_nequal(exp1, exp2, file, line)
++# define _ci_assert_le(exp1, exp2, file, line)
++# define _ci_assert_lt(exp1, exp2, file, line)
++# define _ci_assert_ge(exp1, exp2, file, line)
++# define _ci_assert_gt(exp1, exp2, file, line)
++# define _ci_assert_impl(exp1, exp2, file, line)
++
++# define _ci_verify(exp, file, line) \
++ do { \
++ (void)(exp); \
++ } while (0)
++
++# define CI_DEBUG_TRY(exp) \
++ do { \
++ (void)(exp); \
++ } while (0)
++
++#define CI_TRACE(exp,fmt)
++#define CI_TRACE_INT(integer)
++#define CI_TRACE_INT32(integer)
++#define CI_TRACE_INT64(integer)
++#define CI_TRACE_UINT(integer)
++#define CI_TRACE_UINT32(integer)
++#define CI_TRACE_UINT64(integer)
++#define CI_TRACE_HEX(integer)
++#define CI_TRACE_HEX32(integer)
++#define CI_TRACE_HEX64(integer)
++#define CI_TRACE_PTR(pointer)
++#define CI_TRACE_STRING(string)
++#define CI_TRACE_MAC(mac)
++#define CI_TRACE_IP(ip_be32)
++#define CI_TRACE_ARP(arp_pkt)
+
-+ memset(extras, 0, sizeof(extras));
-+ if (txreq.flags & NETTXF_extra_info) {
-+ work_to_do = netbk_get_extras(netif, extras,
-+ work_to_do);
-+ i = netif->tx.req_cons;
-+ if (unlikely(work_to_do < 0)) {
-+ netbk_tx_err(netif, &txreq, i);
-+ continue;
-+ }
-+ }
++#else
+
-+ ret = netbk_count_requests(netif, &txreq, txfrags, work_to_do);
-+ if (unlikely(ret < 0)) {
-+ netbk_tx_err(netif, &txreq, i - ret);
-+ continue;
-+ }
-+ i += ret;
++# define _CI_ASSERT_FMT "\nfrom %s:%d"
+
-+ if (unlikely(txreq.size < ETH_HLEN)) {
-+ DPRINTK("Bad packet size: %d\n", txreq.size);
-+ netbk_tx_err(netif, &txreq, i);
-+ continue;
-+ }
++# define _ci_check(exp, file, line) \
++ do { \
++ if (CI_UNLIKELY(!(exp))) \
++ ci_warn(("ci_check(%s)"_CI_ASSERT_FMT, #exp, \
++ (file), (line))); \
++ } while (0)
+
-+ /* No crossing a page as the payload mustn't fragment. */
-+ if (unlikely((txreq.offset + txreq.size) > PAGE_SIZE)) {
-+ DPRINTK("txreq.offset: %x, size: %u, end: %lu\n",
-+ txreq.offset, txreq.size,
-+ (txreq.offset &~PAGE_MASK) + txreq.size);
-+ netbk_tx_err(netif, &txreq, i);
-+ continue;
-+ }
++/*
++ * NOTE: ci_fail() emits the file and line where the assert is actually
++ * coded.
++ */
+
-+ pending_idx = pending_ring[MASK_PEND_IDX(pending_cons)];
++# define _ci_assert(exp, file, line) \
++ do { \
++ if (CI_UNLIKELY(!(exp))) \
++ ci_fail(("ci_assert(%s)"_CI_ASSERT_FMT, #exp, \
++ (file), (line))); \
++ } while (0)
+
-+ data_len = (txreq.size > PKT_PROT_LEN &&
-+ ret < MAX_SKB_FRAGS) ?
-+ PKT_PROT_LEN : txreq.size;
++# define _ci_assert2(e, x, y, file, line) do { \
++ if(CI_UNLIKELY( ! (e) )) \
++ ci_fail(("ci_assert(%s)\nwhere [%s=%"CI_PRIx64"] " \
++ "[%s=%"CI_PRIx64"]\nat %s:%d\nfrom %s:%d", #e \
++ , #x, (ci_uint64)(ci_uintptr_t)(x) \
++ , #y, (ci_uint64)(ci_uintptr_t)(y), \
++ __FILE__, __LINE__, (file), (line))); \
++ } while (0)
+
-+ skb = alloc_skb(data_len + 16 + NET_IP_ALIGN,
-+ GFP_ATOMIC | __GFP_NOWARN);
-+ if (unlikely(skb == NULL)) {
-+ DPRINTK("Can't allocate a skb in start_xmit.\n");
-+ netbk_tx_err(netif, &txreq, i);
-+ break;
-+ }
++# define _ci_verify(exp, file, line) \
++ do { \
++ if (CI_UNLIKELY(!(exp))) \
++ ci_fail(("ci_verify(%s)"_CI_ASSERT_FMT, #exp, \
++ (file), (line))); \
++ } while (0)
+
-+ /* Packets passed to netif_rx() must have some headroom. */
-+ skb_reserve(skb, 16 + NET_IP_ALIGN);
++# define _ci_assert_equal(x, y, f, l) _ci_assert2((x)==(y), x, y, (f), (l))
++# define _ci_assert_nequal(x, y, f, l) _ci_assert2((x)!=(y), x, y, (f), (l))
++# define _ci_assert_le(x, y, f, l) _ci_assert2((x)<=(y), x, y, (f), (l))
++# define _ci_assert_lt(x, y, f, l) _ci_assert2((x)< (y), x, y, (f), (l))
++# define _ci_assert_ge(x, y, f, l) _ci_assert2((x)>=(y), x, y, (f), (l))
++# define _ci_assert_gt(x, y, f, l) _ci_assert2((x)> (y), x, y, (f), (l))
++# define _ci_assert_or(x, y, f, l) _ci_assert2((x)||(y), x, y, (f), (l))
++# define _ci_assert_impl(x, y, f, l) _ci_assert2(!(x) || (y), x, y, (f), (l))
++# define _ci_assert_equiv(x, y, f, l) _ci_assert2(!(x)== !(y), x, y, (f), (l))
+
-+ if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
-+ struct netif_extra_info *gso;
-+ gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
++#define _ci_assert_equal_msg(exp1, exp2, msg, file, line) \
++ do { \
++ if (CI_UNLIKELY((exp1)!=(exp2))) \
++ ci_fail(("ci_assert_equal_msg(%s == %s) were " \
++ "(%"CI_PRIx64":%"CI_PRIx64") with msg[%c%c%c%c]" \
++ _CI_ASSERT_FMT, #exp1, #exp2, \
++ (ci_uint64)(ci_uintptr_t)(exp1), \
++ (ci_uint64)(ci_uintptr_t)(exp2), \
++ (((ci_uint32)msg) >> 24) && 0xff, \
++ (((ci_uint32)msg) >> 16) && 0xff, \
++ (((ci_uint32)msg) >> 8 ) && 0xff, \
++ (((ci_uint32)msg) ) && 0xff, \
++ (file), (line))); \
++ } while (0)
+
-+ if (netbk_set_skb_gso(skb, gso)) {
-+ kfree_skb(skb);
-+ netbk_tx_err(netif, &txreq, i);
-+ continue;
-+ }
-+ }
++# define CI_DEBUG_TRY(exp) CI_TRY(exp)
+
-+ gnttab_set_map_op(mop, idx_to_kaddr(pending_idx),
-+ GNTMAP_host_map | GNTMAP_readonly,
-+ txreq.gref, netif->domid);
-+ mop++;
++#define CI_TRACE(exp,fmt) \
++ ci_log("%s:%d:%s] " #exp "=" fmt, \
++ __FILE__, __LINE__, __FUNCTION__, (exp))
+
-+ memcpy(&pending_tx_info[pending_idx].req,
-+ &txreq, sizeof(txreq));
-+ pending_tx_info[pending_idx].netif = netif;
-+ *((u16 *)skb->data) = pending_idx;
+
-+ __skb_put(skb, data_len);
++#define CI_TRACE_INT(integer) \
++ ci_log("%s:%d:%s] " #integer "=%d", \
++ __FILE__, __LINE__, __FUNCTION__, (integer))
+
-+ skb_shinfo(skb)->nr_frags = ret;
-+ if (data_len < txreq.size) {
-+ skb_shinfo(skb)->nr_frags++;
-+ skb_shinfo(skb)->frags[0].page =
-+ (void *)(unsigned long)pending_idx;
-+ } else {
-+ /* Discriminate from any valid pending_idx value. */
-+ skb_shinfo(skb)->frags[0].page = (void *)~0UL;
-+ }
+
-+ __skb_queue_tail(&tx_queue, skb);
++#define CI_TRACE_INT32(integer) \
++ ci_log("%s:%d:%s] " #integer "=%d", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_int32)integer))
+
-+ pending_cons++;
+
-+ mop = netbk_get_requests(netif, skb, txfrags, mop);
++#define CI_TRACE_INT64(integer) \
++ ci_log("%s:%d:%s] " #integer "=%lld", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_int64)integer))
+
-+ netif->tx.req_cons = i;
-+ netif_schedule_work(netif);
+
-+ if ((mop - tx_map_ops) >= ARRAY_SIZE(tx_map_ops))
-+ break;
-+ }
++#define CI_TRACE_UINT(integer) \
++ ci_log("%s:%d:%s] " #integer "=%ud", \
++ __FILE__, __LINE__, __FUNCTION__, (integer))
+
-+ if (mop == tx_map_ops)
-+ return;
+
-+ ret = HYPERVISOR_grant_table_op(
-+ GNTTABOP_map_grant_ref, tx_map_ops, mop - tx_map_ops);
-+ BUG_ON(ret);
++#define CI_TRACE_UINT32(integer) \
++ ci_log("%s:%d:%s] " #integer "=%ud", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_uint32)integer))
+
-+ mop = tx_map_ops;
-+ while ((skb = __skb_dequeue(&tx_queue)) != NULL) {
-+ netif_tx_request_t *txp;
+
-+ pending_idx = *((u16 *)skb->data);
-+ netif = pending_tx_info[pending_idx].netif;
-+ txp = &pending_tx_info[pending_idx].req;
++#define CI_TRACE_UINT64(integer) \
++ ci_log("%s:%d:%s] " #integer "=%ulld", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_uint64)integer))
+
-+ /* Check the remap error code. */
-+ if (unlikely(netbk_tx_check_mop(skb, &mop))) {
-+ DPRINTK("netback grant failed.\n");
-+ skb_shinfo(skb)->nr_frags = 0;
-+ kfree_skb(skb);
-+ continue;
-+ }
+
-+ data_len = skb->len;
-+ memcpy(skb->data,
-+ (void *)(idx_to_kaddr(pending_idx)|txp->offset),
-+ data_len);
-+ if (data_len < txp->size) {
-+ /* Append the packet payload as a fragment. */
-+ txp->offset += data_len;
-+ txp->size -= data_len;
-+ } else {
-+ /* Schedule a response immediately. */
-+ netif_idx_release(pending_idx);
-+ }
++#define CI_TRACE_HEX(integer) \
++ ci_log("%s:%d:%s] " #integer "=0x%x", \
++ __FILE__, __LINE__, __FUNCTION__, (integer))
+
-+ /*
-+ * Old frontends do not assert data_validated but we
-+ * can infer it from csum_blank so test both flags.
-+ */
-+ if (txp->flags & (NETTXF_data_validated|NETTXF_csum_blank)) {
-+ skb->ip_summed = CHECKSUM_UNNECESSARY;
-+ skb->proto_data_valid = 1;
-+ } else {
-+ skb->ip_summed = CHECKSUM_NONE;
-+ skb->proto_data_valid = 0;
-+ }
-+ skb->proto_csum_blank = !!(txp->flags & NETTXF_csum_blank);
+
-+ netbk_fill_frags(skb);
++#define CI_TRACE_HEX32(integer) \
++ ci_log("%s:%d:%s] " #integer "=0x%x", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_uint32)integer))
+
-+ skb->dev = netif->dev;
-+ skb->protocol = eth_type_trans(skb, skb->dev);
+
-+ netif->stats.rx_bytes += skb->len;
-+ netif->stats.rx_packets++;
++#define CI_TRACE_HEX64(integer) \
++ ci_log("%s:%d:%s] " #integer "=0x%llx", \
++ __FILE__, __LINE__, __FUNCTION__, ((ci_uint64)integer))
+
-+ if (unlikely(netbk_copy_skb_mode == NETBK_ALWAYS_COPY_SKB) &&
-+ unlikely(skb_linearize(skb))) {
-+ DPRINTK("Can't linearize skb in net_tx_action.\n");
-+ kfree_skb(skb);
-+ continue;
-+ }
+
-+ netif_rx(skb);
-+ netif->dev->last_rx = jiffies;
-+ }
++#define CI_TRACE_PTR(pointer) \
++ ci_log("%s:%d:%s] " #pointer "=0x%p", \
++ __FILE__, __LINE__, __FUNCTION__, (pointer))
++
++
++#define CI_TRACE_STRING(string) \
++ ci_log("%s:%d:%s] " #string "=%s", \
++ __FILE__, __LINE__, __FUNCTION__, (string))
++
++
++#define CI_TRACE_MAC(mac) \
++ ci_log("%s:%d:%s] " #mac "=" CI_MAC_PRINTF_FORMAT, \
++ __FILE__, __LINE__, __FUNCTION__, CI_MAC_PRINTF_ARGS(mac))
++
++
++#define CI_TRACE_IP(ip_be32) \
++ ci_log("%s:%d:%s] " #ip_be32 "=" CI_IP_PRINTF_FORMAT, __FILE__, \
++ __LINE__, __FUNCTION__, CI_IP_PRINTF_ARGS(&(ip_be32)))
++
++
++#define CI_TRACE_ARP(arp_pkt) \
++ ci_log("%s:%d:%s]\n"CI_ARP_PRINTF_FORMAT, \
++ __FILE__, __LINE__, __FUNCTION__, CI_ARP_PRINTF_ARGS(arp_pkt))
++
++#endif /* NDEBUG */
++
++#define ci_check(exp) \
++ _ci_check(exp, __FILE__, __LINE__)
++
++#define ci_assert(exp) \
++ _ci_assert(exp, __FILE__, __LINE__)
++
++#define ci_verify(exp) \
++ _ci_verify(exp, __FILE__, __LINE__)
++
++#define ci_assert_equal(exp1, exp2) \
++ _ci_assert_equal(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_equal_msg(exp1, exp2, msg) \
++ _ci_assert_equal_msg(exp1, exp2, msg, __FILE__, __LINE__)
++
++#define ci_assert_nequal(exp1, exp2) \
++ _ci_assert_nequal(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_le(exp1, exp2) \
++ _ci_assert_le(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_lt(exp1, exp2) \
++ _ci_assert_lt(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_ge(exp1, exp2) \
++ _ci_assert_ge(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_gt(exp1, exp2) \
++ _ci_assert_gt(exp1, exp2, __FILE__, __LINE__)
++
++#define ci_assert_impl(exp1, exp2) \
++ _ci_assert_impl(exp1, exp2, __FILE__, __LINE__)
+
-+ if (netbk_copy_skb_mode == NETBK_DELAYED_COPY_SKB &&
-+ !list_empty(&pending_inuse_head)) {
-+ struct netbk_tx_pending_inuse *oldest;
++#define ci_assert_equiv(exp1, exp2) \
++ _ci_assert_equiv(exp1, exp2, __FILE__, __LINE__)
+
-+ oldest = list_entry(pending_inuse_head.next,
-+ struct netbk_tx_pending_inuse, list);
-+ mod_timer(&netbk_tx_pending_timer, oldest->alloc_time + HZ);
-+ }
-+}
+
-+static void netif_idx_release(u16 pending_idx)
-+{
-+ static DEFINE_SPINLOCK(_lock);
-+ unsigned long flags;
++#define CI_TEST(exp) \
++ do{ \
++ if( CI_UNLIKELY(!(exp)) ) \
++ ci_fail(("CI_TEST(%s)", #exp)); \
++ }while(0)
+
-+ spin_lock_irqsave(&_lock, flags);
-+ dealloc_ring[MASK_PEND_IDX(dealloc_prod)] = pending_idx;
-+ /* Sync with net_tx_action_dealloc: insert idx /then/ incr producer. */
-+ smp_wmb();
-+ dealloc_prod++;
-+ spin_unlock_irqrestore(&_lock, flags);
+
-+ tasklet_schedule(&net_tx_tasklet);
-+}
++#define CI_TRY(exp) \
++ do{ \
++ int _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(_trc < 0) ) \
++ ci_sys_fail(#exp, _trc); \
++ }while(0)
+
-+static void netif_page_release(struct page *page)
-+{
-+ netif_idx_release(netif_page_index(page));
-+}
+
-+irqreturn_t netif_be_int(int irq, void *dev_id, struct pt_regs *regs)
-+{
-+ netif_t *netif = dev_id;
++#define CI_TRY_RET(exp) \
++ do{ \
++ int _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(_trc < 0) ) { \
++ ci_log("%s returned %d at %s:%d", #exp, _trc, __FILE__, __LINE__); \
++ return _trc; \
++ } \
++ }while(0)
+
-+ add_to_net_schedule_list_tail(netif);
-+ maybe_schedule_tx_action();
++#define CI_LOGLEVEL_TRY_RET(logfn, exp) \
++ do{ \
++ int _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(_trc < 0) ) { \
++ logfn (ci_log("%s returned %d at %s:%d", #exp, _trc, __FILE__, __LINE__)); \
++ return _trc; \
++ } \
++ }while(0)
+
-+ if (netif_schedulable(netif) && !netbk_queue_full(netif))
-+ netif_wake_queue(netif->dev);
+
-+ return IRQ_HANDLED;
-+}
++#define CI_SOCK_TRY(exp) \
++ do{ \
++ ci_sock_err_t _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(!ci_sock_errok(_trc)) ) \
++ ci_sys_fail(#exp, _trc.val); \
++ }while(0)
+
-+static void make_tx_response(netif_t *netif,
-+ netif_tx_request_t *txp,
-+ s8 st)
-+{
-+ RING_IDX i = netif->tx.rsp_prod_pvt;
-+ netif_tx_response_t *resp;
-+ int notify;
+
-+ resp = RING_GET_RESPONSE(&netif->tx, i);
-+ resp->id = txp->id;
-+ resp->status = st;
++#define CI_SOCK_TRY_RET(exp) \
++ do{ \
++ ci_sock_err_t _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(!ci_sock_errok(_trc)) ) { \
++ ci_log("%s returned %d at %s:%d", #exp, _trc.val, __FILE__, __LINE__); \
++ return ci_sock_errcode(_trc); \
++ } \
++ }while(0)
+
-+ if (txp->flags & NETTXF_extra_info)
-+ RING_GET_RESPONSE(&netif->tx, ++i)->status = NETIF_RSP_NULL;
+
-+ netif->tx.rsp_prod_pvt = ++i;
-+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&netif->tx, notify);
-+ if (notify)
-+ notify_remote_via_irq(netif->irq);
++#define CI_SOCK_TRY_SOCK_RET(exp) \
++ do{ \
++ ci_sock_err_t _trc; \
++ _trc=(exp); \
++ if( CI_UNLIKELY(!ci_sock_errok(_trc)) ) { \
++ ci_log("%s returned %d at %s:%d", #exp, _trc.val, __FILE__, __LINE__); \
++ return _trc; \
++ } \
++ }while(0)
++
++#endif /* __CI_TOOLS_DEBUG_H__ */
++
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/log.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/log.h 2008-05-19 00:33:48.534943730 +0300
+@@ -0,0 +1,262 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+#ifdef CONFIG_XEN_NETDEV_PIPELINED_TRANSMITTER
-+ if (i == netif->tx.req_cons) {
-+ int more_to_do;
-+ RING_FINAL_CHECK_FOR_REQUESTS(&netif->tx, more_to_do);
-+ if (more_to_do)
-+ add_to_net_schedule_list_tail(netif);
-+ }
-+#endif
-+}
++/*
++ * \author djr
++ * \brief Functions for logging and pretty-printing.
++ * \date 2002/08/07
++ */
+
-+static netif_rx_response_t *make_rx_response(netif_t *netif,
-+ u16 id,
-+ s8 st,
-+ u16 offset,
-+ u16 size,
-+ u16 flags)
-+{
-+ RING_IDX i = netif->rx.rsp_prod_pvt;
-+ netif_rx_response_t *resp;
++/*! \cidoxg_include_ci_tools */
+
-+ resp = RING_GET_RESPONSE(&netif->rx, i);
-+ resp->offset = offset;
-+ resp->flags = flags;
-+ resp->id = id;
-+ resp->status = (s16)size;
-+ if (st < 0)
-+ resp->status = (s16)st;
++#ifndef __CI_TOOLS_LOG_H__
++#define __CI_TOOLS_LOG_H__
+
-+ netif->rx.rsp_prod_pvt = ++i;
++#include <stdarg.h>
+
-+ return resp;
-+}
+
-+#ifdef NETBE_DEBUG_INTERRUPT
-+static irqreturn_t netif_be_dbg(int irq, void *dev_id, struct pt_regs *regs)
-+{
-+ struct list_head *ent;
-+ netif_t *netif;
-+ int i = 0;
++/**********************************************************************
++ * Logging.
++ */
+
-+ printk(KERN_ALERT "netif_schedule_list:\n");
-+ spin_lock_irq(&net_schedule_list_lock);
++/* size of internal log buffer */
++#define CI_LOG_MAX_LINE 512
++/* uses of ci_log must ensure that all trace messages are shorter than this */
++#define CI_LOG_MAX_MSG_LENGTH (CI_LOG_MAX_LINE-50)
+
-+ list_for_each (ent, &net_schedule_list) {
-+ netif = list_entry(ent, netif_t, list);
-+ printk(KERN_ALERT " %d: private(rx_req_cons=%08x "
-+ "rx_resp_prod=%08x\n",
-+ i, netif->rx.req_cons, netif->rx.rsp_prod_pvt);
-+ printk(KERN_ALERT " tx_req_cons=%08x tx_resp_prod=%08x)\n",
-+ netif->tx.req_cons, netif->tx.rsp_prod_pvt);
-+ printk(KERN_ALERT " shared(rx_req_prod=%08x "
-+ "rx_resp_prod=%08x\n",
-+ netif->rx.sring->req_prod, netif->rx.sring->rsp_prod);
-+ printk(KERN_ALERT " rx_event=%08x tx_req_prod=%08x\n",
-+ netif->rx.sring->rsp_event, netif->tx.sring->req_prod);
-+ printk(KERN_ALERT " tx_resp_prod=%08x, tx_event=%08x)\n",
-+ netif->tx.sring->rsp_prod, netif->tx.sring->rsp_event);
-+ i++;
-+ }
++extern void ci_vlog(const char* fmt, va_list args) CI_HF;
++extern void ci_log(const char* fmt, ...) CI_PRINTF_LIKE(1,2) CI_HF;
+
-+ spin_unlock_irq(&net_schedule_list_lock);
-+ printk(KERN_ALERT " ** End of netif_schedule_list **\n");
++ /*! Set the prefix for log messages.
++ **
++ ** Uses the storage pointed to by \em prefix. Therefore \em prefix must
++ ** be allocated on the heap, or statically.
++ */
++extern void ci_set_log_prefix(const char* prefix) CI_HF;
++
++typedef void (*ci_log_fn_t)(const char* msg);
++extern ci_log_fn_t ci_log_fn CI_HV;
++
++/* Log functions. */
++extern void ci_log_null(const char* msg) CI_HF;
++extern void ci_log_stderr(const char* msg) CI_HF;
++extern void ci_log_stdout(const char* msg) CI_HF;
++extern void ci_log_syslog(const char* msg) CI_HF;
++
++/*! Call the following to install special logging behaviours. */
++extern void ci_log_buffer_till_fail(void) CI_HF;
++extern void ci_log_buffer_till_exit(void) CI_HF;
++
++extern void __ci_log_unique(const char* msg) CI_HF;
++extern ci_log_fn_t __ci_log_unique_fn CI_HV;
++ci_inline void ci_log_uniquify(void) {
++ if( ci_log_fn != __ci_log_unique ) {
++ __ci_log_unique_fn = ci_log_fn;
++ ci_log_fn = __ci_log_unique;
++ }
++}
+
-+ return IRQ_HANDLED;
++extern void ci_log_file(const char* msg) CI_HF;
++extern int ci_log_file_fd CI_HV;
++
++extern void __ci_log_nth(const char* msg) CI_HF;
++extern ci_log_fn_t __ci_log_nth_fn CI_HV;
++extern int ci_log_nth_n CI_HV; /* default 100 */
++ci_inline void ci_log_nth(void) {
++ if( ci_log_fn != __ci_log_nth ) {
++ __ci_log_nth_fn = ci_log_fn;
++ ci_log_fn = __ci_log_nth;
++ }
+}
++
++extern int ci_log_level CI_HV;
++
++extern int ci_log_options CI_HV;
++#define CI_LOG_PID 0x1
++#define CI_LOG_TID 0x2
++#define CI_LOG_TIME 0x4
++#define CI_LOG_DELTA 0x8
++
++/**********************************************************************
++ * Used to define which mode we are in
++ */
++#if (defined(_WIN32) && !defined(__KERNEL__))
++typedef enum {
++ ci_log_md_NULL=0,
++ ci_log_md_ioctl,
++ ci_log_md_stderr,
++ ci_log_md_stdout,
++ ci_log_md_file,
++ ci_log_md_serial,
++ ci_log_md_syslog,
++ ci_log_md_pidfile
++} ci_log_mode_t;
++extern ci_log_mode_t ci_log_mode;
+#endif
+
-+static int __init netback_init(void)
-+{
-+ int i;
-+ struct page *page;
++/**********************************************************************
++ * Pretty-printing.
++ */
+
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++extern char ci_printable_char(char c) CI_HF;
+
-+ /* We can increase reservation by this much in net_rx_action(). */
-+ balloon_update_driver_allowance(NET_RX_RING_SIZE);
++extern void (*ci_hex_dump_formatter)(char* buf, const ci_octet* s,
++ int i, int off, int len) CI_HV;
++extern void ci_hex_dump_format_octets(char*,const ci_octet*,int,int,int) CI_HF;
++extern void ci_hex_dump_format_dwords(char*,const ci_octet*,int,int,int) CI_HF;
+
-+ skb_queue_head_init(&rx_queue);
-+ skb_queue_head_init(&tx_queue);
++extern void ci_hex_dump_row(char* buf, volatile const void* s, int len,
++ ci_ptr_arith_t address) CI_HF;
++ /*!< A row contains up to 16 bytes. Row starts at [address & 15u], so
++ ** therefore [len + (address & 15u)] must be <= 16.
++ */
+
-+ init_timer(&net_timer);
-+ net_timer.data = 0;
-+ net_timer.function = net_alarm;
++extern void ci_hex_dump(ci_log_fn_t, volatile const void*,
++ int len, ci_ptr_arith_t address) CI_HF;
+
-+ init_timer(&netbk_tx_pending_timer);
-+ netbk_tx_pending_timer.data = 0;
-+ netbk_tx_pending_timer.function = netbk_tx_pending_timeout;
++extern int ci_hex_dump_to_raw(const char* src_hex, void* buf,
++ unsigned* addr_out_opt, int* skip) CI_HF;
++ /*!< Recovers raw data from a single line of a hex dump. [buf] must be at
++ ** least 16 bytes long. Returns the number of bytes written to [buf] (in
++ ** range 1 -> 16), or -1 if [src_hex] doesn't contain hex data. Does not
++ ** cope with missing bytes at the start of a line.
++ */
+
-+ mmap_pages = alloc_empty_pages_and_pagevec(MAX_PENDING_REQS);
-+ if (mmap_pages == NULL) {
-+ printk("%s: out of memory\n", __FUNCTION__);
-+ return -ENOMEM;
-+ }
++extern int ci_format_eth_addr(char* buf, const void* eth_mac_addr,
++ char sep) CI_HF;
++ /*!< This will write 18 characters to <buf> including terminating null.
++ ** Returns number of bytes written excluding null. If [sep] is zero, ':'
++ ** is used.
++ */
+
-+ for (i = 0; i < MAX_PENDING_REQS; i++) {
-+ page = mmap_pages[i];
-+ SetPageForeign(page, netif_page_release);
-+ netif_page_index(page) = i;
-+ INIT_LIST_HEAD(&pending_inuse[i].list);
-+ }
++extern int ci_parse_eth_addr(void* eth_mac_addr,
++ const char* str, char sep) CI_HF;
++ /*!< If [sep] is zero, absolutely any separator is accepted (even
++ ** inconsistent separators). Returns 0 on success, -1 on error.
++ */
+
-+ pending_cons = 0;
-+ pending_prod = MAX_PENDING_REQS;
-+ for (i = 0; i < MAX_PENDING_REQS; i++)
-+ pending_ring[i] = i;
++extern int ci_format_ip4_addr(char* buf, unsigned addr_be32) CI_HF;
++ /*!< Formats the IP address (in network endian) in dotted-quad. Returns
++ ** the number of bytes written (up to 15), excluding the null. [buf]
++ ** must be at least 16 bytes long.
++ */
+
-+ spin_lock_init(&net_schedule_list_lock);
-+ INIT_LIST_HEAD(&net_schedule_list);
+
-+ netbk_copy_skb_mode = NETBK_DONT_COPY_SKB;
-+ if (MODPARM_copy_skb) {
-+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_replace,
-+ NULL, 0))
-+ netbk_copy_skb_mode = NETBK_ALWAYS_COPY_SKB;
-+ else
-+ netbk_copy_skb_mode = NETBK_DELAYED_COPY_SKB;
-+ }
++/**********************************************************************
++ * Error checking.
++ */
+
-+ netif_accel_init();
++extern void (*ci_fail_stop_fn)(void) CI_HV;
+
-+ netif_xenbus_init();
++extern void ci_fail_stop(void) CI_HF;
++extern void ci_fail_hang(void) CI_HF;
++extern void ci_fail_bomb(void) CI_HF;
++extern void ci_backtrace(void) CI_HF;
+
-+#ifdef NETBE_DEBUG_INTERRUPT
-+ (void)bind_virq_to_irqhandler(VIRQ_DEBUG,
-+ 0,
-+ netif_be_dbg,
-+ SA_SHIRQ,
-+ "net-be-dbg",
-+ &netif_be_dbg);
++#if defined __linux__ && !defined __KERNEL__
++extern void ci_fail_abort (void) CI_HF;
+#endif
+
-+ return 0;
-+}
++#ifdef __GNUC__
++extern void
++__ci_fail(const char*, ...) CI_PRINTF_LIKE(1,2) CI_HF;
++#else
++# if _PREFAST_
++ extern void _declspec(noreturn) __ci_fail(const char* fmt, ...);
++# else
++ extern void __ci_fail(const char* fmt, ...);
++# endif
+
-+module_init(netback_init);
++#endif
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netback/xenbus.c linux-2.6.18-xen.hg/drivers/xen/netback/xenbus.c
---- linux-2.6.18/drivers/xen/netback/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netback/xenbus.c 2007-12-23 11:15:34.054602273 +0100
-@@ -0,0 +1,452 @@
-+/* Xenbus code for netif backend
-+ Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au>
-+ Copyright (C) 2005 XenSource Ltd
++#define ci_warn(x) \
++ do{ ci_log("WARN at %s:%d", __FILE__, __LINE__); }while(0)
+
-+ This program is free software; you can redistribute it and/or modify
-+ it under the terms of the GNU General Public License as published by
-+ the Free Software Foundation; either version 2 of the License, or
-+ (at your option) any later version.
++#define ci_fail(x) \
++ do{ ci_log("FAIL at %s:%d", __FILE__, __LINE__); __ci_fail x; }while(0)
+
-+ This program is distributed in the hope that it will be useful,
-+ but WITHOUT ANY WARRANTY; without even the implied warranty of
-+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-+ GNU General Public License for more details.
++extern void __ci_sys_fail(const char* fn, int rc,
++ const char* file, int line) CI_HF;
++#define ci_sys_fail(fn, rc) __ci_sys_fail(fn, rc, __FILE__, __LINE__)
+
-+ You should have received a copy of the GNU General Public License
-+ along with this program; if not, write to the Free Software
-+ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+*/
++/**********************************************************************
++ * Logging to buffer (src/citools/log_buffer.c)
++ */
+
-+#include <stdarg.h>
-+#include <linux/module.h>
-+#include <xen/xenbus.h>
-+#include "common.h"
++/*! Divert ci_log() messages to the log buffer
++ * normally they go to the system console */
++extern void ci_log_buffer_till_fail(void) CI_HF;
+
-+#if 0
-+#undef DPRINTK
-+#define DPRINTK(fmt, args...) \
-+ printk("netback/xenbus (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
-+#endif
++/*! Dump the contents of the log buffer to the system console */
++extern void ci_log_buffer_dump(void) CI_HF;
+
+
-+static int connect_rings(struct backend_info *);
-+static void connect(struct backend_info *);
-+static void backend_create_netif(struct backend_info *be);
++/**********************************************************************
++ * Some useful pretty-printing.
++ */
+
-+static int netback_remove(struct xenbus_device *dev)
-+{
-+ struct backend_info *be = dev->dev.driver_data;
++#ifdef __linux__
++# define CI_SOCKCALL_FLAGS_FMT "%s%s%s%s%s%s%s%s%s%s%s"
+
-+ netback_remove_accelerators(be, dev);
++# define CI_SOCKCALL_FLAGS_PRI_ARG(x) \
++ (((x) & MSG_OOB ) ? "OOB " :""), \
++ (((x) & MSG_PEEK ) ? "PEEK " :""), \
++ (((x) & MSG_DONTROUTE ) ? "DONTROUTE " :""), \
++ (((x) & MSG_EOR ) ? "EOR " :""), \
++ (((x) & MSG_CTRUNC ) ? "CTRUNC " :""), \
++ (((x) & MSG_TRUNC ) ? "TRUNC " :""), \
++ (((x) & MSG_WAITALL ) ? "WAITALL " :""), \
++ (((x) & MSG_DONTWAIT ) ? "DONTWAIT " :""), \
++ (((x) & MSG_NOSIGNAL ) ? "NOSIGNAL " :""), \
++ (((x) & MSG_ERRQUEUE ) ? "ERRQUEUE " :""), \
++ (((x) & MSG_CONFIRM ) ? "CONFIRM " :"")
++#endif
+
-+ if (be->netif) {
-+ netif_disconnect(be->netif);
-+ be->netif = NULL;
-+ }
-+ kfree(be);
-+ dev->dev.driver_data = NULL;
-+ return 0;
-+}
++#ifdef _WIN32
++# define CI_SOCKCALL_FLAGS_FMT "%s%s%s"
+
++# define CI_SOCKCALL_FLAGS_PRI_ARG(x) \
++ (((x) & MSG_OOB ) ? "OOB " :""), \
++ (((x) & MSG_PEEK ) ? "PEEK " :""), \
++ (((x) & MSG_DONTROUTE ) ? "DONTROUTE " :"")
++#endif
+
-+/**
-+ * Entry point to this code when a new device is created. Allocate the basic
-+ * structures and switch to InitWait.
++#ifdef __sun__
++# define CI_SOCKCALL_FLAGS_FMT "%s%s%s%s%s%s%s%s%s"
++
++# define CI_SOCKCALL_FLAGS_PRI_ARG(x) \
++ (((x) & MSG_OOB ) ? "OOB " :""), \
++ (((x) & MSG_PEEK ) ? "PEEK " :""), \
++ (((x) & MSG_DONTROUTE ) ? "DONTROUTE " :""), \
++ (((x) & MSG_EOR ) ? "EOR " :""), \
++ (((x) & MSG_CTRUNC ) ? "CTRUNC " :""), \
++ (((x) & MSG_TRUNC ) ? "TRUNC " :""), \
++ (((x) & MSG_WAITALL ) ? "WAITALL " :""), \
++ (((x) & MSG_DONTWAIT ) ? "DONTWAIT " :""), \
++ (((x) & MSG_NOTIFICATION) ? "NOTIFICATION" :"")
++#endif
++
++#endif /* __CI_TOOLS_LOG_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/platform/gcc_x86.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/platform/gcc_x86.h 2008-05-19 00:33:48.578946266 +0300
+@@ -0,0 +1,361 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static int netback_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
-+{
-+ const char *message;
-+ struct xenbus_transaction xbt;
-+ int err;
-+ int sg;
-+ struct backend_info *be = kzalloc(sizeof(struct backend_info),
-+ GFP_KERNEL);
-+ if (!be) {
-+ xenbus_dev_fatal(dev, -ENOMEM,
-+ "allocating backend structure");
-+ return -ENOMEM;
-+ }
+
-+ be->dev = dev;
-+ dev->dev.driver_data = be;
++/*! \cidoxg_include_ci_tools_platform */
+
-+ sg = 1;
-+ if (netbk_copy_skb_mode == NETBK_ALWAYS_COPY_SKB)
-+ sg = 0;
++#ifndef __CI_TOOLS_GCC_X86_H__
++#define __CI_TOOLS_GCC_X86_H__
+
-+ do {
-+ err = xenbus_transaction_start(&xbt);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "starting transaction");
-+ goto fail;
-+ }
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-sg", "%d", sg);
-+ if (err) {
-+ message = "writing feature-sg";
-+ goto abort_transaction;
-+ }
++/**********************************************************************
++ * Free-running cycle counters.
++ */
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-gso-tcpv4",
-+ "%d", sg);
-+ if (err) {
-+ message = "writing feature-gso-tcpv4";
-+ goto abort_transaction;
-+ }
++#define CI_HAVE_FRC64
++#define CI_HAVE_FRC32
+
-+ /* We support rx-copy path. */
-+ err = xenbus_printf(xbt, dev->nodename,
-+ "feature-rx-copy", "%d", 1);
-+ if (err) {
-+ message = "writing feature-rx-copy";
-+ goto abort_transaction;
-+ }
++#define ci_frc32(pval) __asm__ __volatile__("rdtsc" : "=a" (*pval) : : "edx")
+
-+ /*
-+ * We don't support rx-flip path (except old guests who don't
-+ * grok this feature flag).
-+ */
-+ err = xenbus_printf(xbt, dev->nodename,
-+ "feature-rx-flip", "%d", 0);
-+ if (err) {
-+ message = "writing feature-rx-flip";
-+ goto abort_transaction;
-+ }
++#if defined(__x86_64__)
++ci_inline void ci_frc64(ci_uint64* pval) {
++ /* temp fix until we figure how to get this out in one bite */
++ ci_uint64 low, high;
++ __asm__ __volatile__("rdtsc" : "=a" (low) , "=d" (high));
++ *pval = (high << 32) | low;
++}
+
-+ err = xenbus_transaction_end(xbt, 0);
-+ } while (err == -EAGAIN);
++#else
++#define ci_frc64(pval) __asm__ __volatile__("rdtsc" : "=A" (*pval))
++#endif
+
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "completing transaction");
-+ goto fail;
-+ }
++#define ci_frc_flush() /* ?? Need a pipeline barrier. */
+
-+ netback_probe_accelerators(be, dev);
+
-+ err = xenbus_switch_state(dev, XenbusStateInitWait);
-+ if (err)
-+ goto fail;
++/**********************************************************************
++ * Atomic integer.
++ */
+
-+ /* This kicks hotplug scripts, so do it immediately. */
-+ backend_create_netif(be);
++/*
++** int ci_atomic_read(a) { return a->n; }
++** void ci_atomic_set(a, v) { a->n = v; }
++** void ci_atomic_inc(a) { ++a->n; }
++** void ci_atomic_dec(a) { --a->n; }
++** int ci_atomic_inc_and_test(a) { return ++a->n == 0; }
++** int ci_atomic_dec_and_test(a) { return --a->n == 0; }
++** void ci_atomic_and(a, v) { a->n &= v; }
++** void ci_atomic_or(a, v) { a->n |= v; }
++*/
+
-+ return 0;
++typedef struct { volatile ci_int32 n; } ci_atomic_t;
+
-+abort_transaction:
-+ xenbus_transaction_end(xbt, 1);
-+ xenbus_dev_fatal(dev, err, "%s", message);
-+fail:
-+ DPRINTK("failed");
-+ netback_remove(dev);
-+ return err;
-+}
++#define CI_ATOMIC_INITIALISER(i) {(i)}
+
++static inline ci_int32 ci_atomic_read(const ci_atomic_t* a) { return a->n; }
++static inline void ci_atomic_set(ci_atomic_t* a, int v) { a->n = v; ci_wmb(); }
+
-+/**
-+ * Handle the creation of the hotplug script environment. We add the script
-+ * and vif variables to the environment, for the benefit of the vif-* hotplug
-+ * scripts.
-+ */
-+static int netback_uevent(struct xenbus_device *xdev, char **envp,
-+ int num_envp, char *buffer, int buffer_size)
-+{
-+ struct backend_info *be = xdev->dev.driver_data;
-+ netif_t *netif = be->netif;
-+ int i = 0, length = 0;
-+ char *val;
++static inline void ci_atomic_inc(ci_atomic_t* a)
++{ __asm__ __volatile__("lock; incl %0" : "+m" (a->n)); }
+
-+ DPRINTK("netback_uevent");
++
++static inline void ci_atomic_dec(ci_atomic_t* a)
++{ __asm__ __volatile__("lock; decl %0" : "+m" (a->n)); }
++
++static inline int ci_atomic_inc_and_test(ci_atomic_t* a) {
++ char r;
++ __asm__ __volatile__("lock; incl %0; sete %1"
++ : "+m" (a->n), "=qm" (r));
++ return r;
++}
++
++static inline int ci_atomic_dec_and_test(ci_atomic_t* a) {
++ char r;
++ __asm__ __volatile__("lock; decl %0; sete %1"
++ : "+m" (a->n), "=qm" (r));
++ return r;
++}
++
++ci_inline int
++ci_atomic_xadd (ci_atomic_t *a, int v) {
++ __asm__ ("lock xadd %0, %1" : "=r" (v), "+m" (a->n) : "0" (v));
++ return v;
++}
++ci_inline int
++ci_atomic_xchg (ci_atomic_t *a, int v) {
++ __asm__ ("lock xchg %0, %1" : "=r" (v), "+m" (a->n) : "0" (v));
++ return v;
++}
++
++ci_inline void ci_atomic32_or(volatile ci_uint32* p, ci_uint32 mask)
++{ __asm__ __volatile__("lock; orl %1, %0" : "+m" (*p) : "ir" (mask)); }
++
++ci_inline void ci_atomic32_and(volatile ci_uint32* p, ci_uint32 mask)
++{ __asm__ __volatile__("lock; andl %1, %0" : "+m" (*p) : "ir" (mask)); }
++
++ci_inline void ci_atomic32_add(volatile ci_uint32* p, ci_uint32 v)
++{ __asm__ __volatile__("lock; addl %1, %0" : "+m" (*p) : "ir" (v)); }
++
++#define ci_atomic_or(a, v) ci_atomic32_or ((ci_uint32*) &(a)->n, (v))
++#define ci_atomic_and(a, v) ci_atomic32_and((ci_uint32*) &(a)->n, (v))
++#define ci_atomic_add(a, v) ci_atomic32_add((ci_uint32*) &(a)->n, (v))
++
++extern int ci_glibc_uses_nptl (void) CI_HF;
++extern int ci_glibc_nptl_broken(void) CI_HF;
++extern int ci_glibc_gs_get_is_multihreaded_offset (void) CI_HF;
++extern int ci_glibc_gs_is_multihreaded_offset CI_HV;
++
++#if !defined(__x86_64__)
++#ifdef __GLIBC__
++/* Returns non-zero if the calling process might be mulithreaded, returns 0 if
++ * it definitely isn't (i.e. if reimplementing this function for other
++ * architectures and platforms, you can safely just return 1).
++ */
++static inline int ci_is_multithreaded (void) {
++
++ while (1) {
++ if (ci_glibc_gs_is_multihreaded_offset >= 0) {
++ /* NPTL keeps a variable that tells us this hanging off gs (i.e. in thread-
++ * local storage); just return this
++ */
++ int r;
++ __asm__ __volatile__ ("movl %%gs:(%1), %0"
++ : "=r" (r)
++ : "r" (ci_glibc_gs_is_multihreaded_offset));
++ return r;
++ }
++
++ if (ci_glibc_gs_is_multihreaded_offset == -2) {
++ /* This means we've already determined that the libc version is NOT good
++ * for our funky "is multithreaded" hack
++ */
++ return 1;
++ }
++
++ /* If we get here, it means this is the first time the function has been
++ * called -- detect the libc version and go around again.
++ */
++ ci_glibc_gs_is_multihreaded_offset = ci_glibc_gs_get_is_multihreaded_offset ();
+
-+ val = xenbus_read(XBT_NIL, xdev->nodename, "script", NULL);
-+ if (IS_ERR(val)) {
-+ int err = PTR_ERR(val);
-+ xenbus_dev_fatal(xdev, err, "reading script");
-+ return err;
-+ }
-+ else {
-+ add_uevent_var(envp, num_envp, &i, buffer, buffer_size,
-+ &length, "script=%s", val);
-+ kfree(val);
-+ }
++ /* Go around again. We do the test here rather than at the top so that we go
++ * quicker in the common the case
++ */
++ }
++}
+
-+ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
-+ "vif=%s", netif->dev->name);
++#else /* def __GLIBC__ */
+
-+ envp[i] = NULL;
++#define ci_is_multithreaded() 1 /* ?? Is the the POSIX way of finding out */
++ /* whether the appication is single */
++ /* threaded? */
+
-+ return 0;
++#endif /* def __GLIBC__ */
++
++#else /* defined __x86_64__ */
++
++static inline int ci_is_multithreaded (void) {
++ /* Now easy way to tell on x86_64; so assume we're multithreaded */
++ return 1;
+}
+
++#endif /* defined __x86_64__ */
+
-+static void backend_create_netif(struct backend_info *be)
-+{
-+ int err;
-+ long handle;
-+ struct xenbus_device *dev = be->dev;
+
-+ if (be->netif != NULL)
-+ return;
++/**********************************************************************
++ * Compare and swap.
++ */
+
-+ err = xenbus_scanf(XBT_NIL, dev->nodename, "handle", "%li", &handle);
-+ if (err != 1) {
-+ xenbus_dev_fatal(dev, err, "reading handle");
-+ return;
-+ }
++#define CI_HAVE_COMPARE_AND_SWAP
+
-+ be->netif = netif_alloc(dev->otherend_id, handle);
-+ if (IS_ERR(be->netif)) {
-+ err = PTR_ERR(be->netif);
-+ be->netif = NULL;
-+ xenbus_dev_fatal(dev, err, "creating interface");
-+ return;
-+ }
++ci_inline int ci_cas32_succeed(volatile ci_int32* p, ci_int32 oldval,
++ ci_int32 newval) {
++ char ret;
++ ci_int32 prevval;
++ __asm__ __volatile__("lock; cmpxchgl %3, %1; sete %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
+
-+ kobject_uevent(&dev->dev.kobj, KOBJ_ONLINE);
++ci_inline int ci_cas32_fail(volatile ci_int32* p, ci_int32 oldval,
++ ci_int32 newval) {
++ char ret;
++ ci_int32 prevval;
++ __asm__ __volatile__("lock; cmpxchgl %3, %1; setne %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
+}
+
++#ifdef __x86_64__
++ci_inline int ci_cas64_succeed(volatile ci_int64* p, ci_int64 oldval,
++ ci_int64 newval) {
++ char ret;
++ ci_int64 prevval;
++ __asm__ __volatile__("lock; cmpxchgq %3, %1; sete %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
++
++ci_inline int ci_cas64_fail(volatile ci_int64* p, ci_int64 oldval,
++ ci_int64 newval) {
++ char ret;
++ ci_int64 prevval;
++ __asm__ __volatile__("lock; cmpxchgq %3, %1; setne %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
++#endif
++
++ci_inline int ci_cas32u_succeed(volatile ci_uint32* p, ci_uint32 oldval, ci_uint32 newval) {
++ char ret;
++ ci_uint32 prevval;
++ __asm__ __volatile__("lock; cmpxchgl %3, %1; sete %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
++
++ci_inline int ci_cas32u_fail(volatile ci_uint32* p, ci_uint32 oldval, ci_uint32 newval) {
++ char ret;
++ ci_uint32 prevval;
++ __asm__ __volatile__("lock; cmpxchgl %3, %1; setne %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
++
++ci_inline int ci_cas64u_succeed(volatile ci_uint64* p, ci_uint64 oldval,
++ ci_uint64 newval) {
++ char ret;
++ ci_uint64 prevval;
++ __asm__ __volatile__("lock; cmpxchgq %3, %1; sete %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
++
++ci_inline int ci_cas64u_fail(volatile ci_uint64* p, ci_uint64 oldval,
++ ci_uint64 newval) {
++ char ret;
++ ci_uint64 prevval;
++ __asm__ __volatile__("lock; cmpxchgq %3, %1; setne %0"
++ : "=q"(ret), "+m"(*p), "=a"(prevval)
++ : "r"(newval), "a"(oldval));
++ return ret;
++}
+
-+/**
-+ * Callback received when the frontend's state changes.
-+ */
-+static void frontend_changed(struct xenbus_device *dev,
-+ enum xenbus_state frontend_state)
-+{
-+ struct backend_info *be = dev->dev.driver_data;
++#ifdef __x86_64__
+
-+ DPRINTK("%s", xenbus_strstate(frontend_state));
++# define ci_cas_uintptr_succeed(p,o,n) \
++ ci_cas64u_succeed((volatile ci_uint64*) (p), (o), (n))
++# define ci_cas_uintptr_fail(p,o,n) \
++ ci_cas64u_fail((volatile ci_uint64*) (p), (o), (n))
+
-+ be->frontend_state = frontend_state;
++#else
+
-+ switch (frontend_state) {
-+ case XenbusStateInitialising:
-+ if (dev->state == XenbusStateClosed) {
-+ printk(KERN_INFO "%s: %s: prepare for reconnect\n",
-+ __FUNCTION__, dev->nodename);
-+ if (be->netif) {
-+ netif_disconnect(be->netif);
-+ be->netif = NULL;
-+ }
-+ xenbus_switch_state(dev, XenbusStateInitWait);
-+ }
-+ break;
++# define ci_cas_uintptr_succeed(p,o,n) \
++ ci_cas32u_succeed((volatile ci_uint32*) (p), (o), (n))
++# define ci_cas_uintptr_fail(p,o,n) \
++ ci_cas32u_fail((volatile ci_uint32*) (p), (o), (n))
+
-+ case XenbusStateInitialised:
-+ break;
++#endif
+
-+ case XenbusStateConnected:
-+ backend_create_netif(be);
-+ if (be->netif)
-+ connect(be);
-+ break;
+
-+ case XenbusStateClosing:
-+ xenbus_switch_state(dev, XenbusStateClosing);
-+ break;
++/**********************************************************************
++ * Atomic bit field.
++ */
+
-+ case XenbusStateClosed:
-+ xenbus_switch_state(dev, XenbusStateClosed);
-+ if (xenbus_dev_is_online(dev))
-+ break;
-+ /* fall through if not online */
-+ case XenbusStateUnknown:
-+ if (be->netif != NULL)
-+ kobject_uevent(&dev->dev.kobj, KOBJ_OFFLINE);
-+ device_unregister(&dev->dev);
-+ break;
++typedef ci_uint32 ci_bits;
++#define CI_BITS_N 32u
+
-+ default:
-+ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
-+ frontend_state);
-+ break;
-+ }
++#define CI_BITS_DECLARE(name, n) \
++ ci_bits name[((n) + CI_BITS_N - 1u) / CI_BITS_N]
++
++ci_inline void ci_bits_clear_all(volatile ci_bits* b, int n_bits)
++{ memset((void*) b, 0, (n_bits+CI_BITS_N-1u) / CI_BITS_N * sizeof(ci_bits)); }
++
++ci_inline void ci_bit_set(volatile ci_bits* b, int i) {
++ __asm__ __volatile__("lock; btsl %1, %0"
++ : "=m" (*b)
++ : "Ir" (i));
+}
+
++ci_inline void ci_bit_clear(volatile ci_bits* b, int i) {
++ __asm__ __volatile__("lock; btrl %1, %0"
++ : "=m" (*b)
++ : "Ir" (i));
++}
+
-+static void xen_net_read_rate(struct xenbus_device *dev,
-+ unsigned long *bytes, unsigned long *usec)
-+{
-+ char *s, *e;
-+ unsigned long b, u;
-+ char *ratestr;
++ci_inline int ci_bit_test(volatile ci_bits* b, int i) {
++ char rc;
++ __asm__("btl %2, %1; setc %0"
++ : "=r" (rc)
++ : "m" (*b), "Ir" (i));
++ return rc;
++}
+
-+ /* Default to unlimited bandwidth. */
-+ *bytes = ~0UL;
-+ *usec = 0;
++ci_inline int ci_bit_test_and_set(volatile ci_bits* b, int i) {
++ char rc;
++ __asm__ __volatile__("lock; btsl %2, %1; setc %0"
++ : "=r" (rc), "+m" (*b)
++ : "Ir" (i));
++ return rc;
++}
+
-+ ratestr = xenbus_read(XBT_NIL, dev->nodename, "rate", NULL);
-+ if (IS_ERR(ratestr))
-+ return;
++ci_inline int ci_bit_test_and_clear(volatile ci_bits* b, int i) {
++ char rc;
++ __asm__ __volatile__("lock; btrl %2, %1; setc %0"
++ : "=r" (rc), "+m" (*b)
++ : "Ir" (i));
++ return rc;
++}
+
-+ s = ratestr;
-+ b = simple_strtoul(s, &e, 10);
-+ if ((s == e) || (*e != ','))
-+ goto fail;
++/* These mask ops only work within a single ci_bits word. */
++#define ci_bit_mask_set(b,m) ci_atomic32_or((b), (m))
++#define ci_bit_mask_clear(b,m) ci_atomic32_and((b), ~(m))
+
-+ s = e + 1;
-+ u = simple_strtoul(s, &e, 10);
-+ if ((s == e) || (*e != '\0'))
-+ goto fail;
+
-+ *bytes = b;
-+ *usec = u;
++/**********************************************************************
++ * Misc.
++ */
+
-+ kfree(ratestr);
-+ return;
++#if __GNUC__ >= 3
++# define ci_spinloop_pause() __asm__("pause")
++#else
++# define ci_spinloop_pause() __asm__(".byte 0xf3, 0x90")
++#endif
+
-+ fail:
-+ WPRINTK("Failed to parse network rate limit. Traffic unlimited.\n");
-+ kfree(ratestr);
-+}
+
-+static int xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
-+{
-+ char *s, *e, *macstr;
-+ int i;
++#define CI_HAVE_ADDC32
++#define ci_add_carry32(sum, v) __asm__("addl %1, %0 ;" \
++ "adcl $0, %0 ;" \
++ : "=r" (sum) \
++ : "g" ((ci_uint32) v), "0" (sum))
+
-+ macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
-+ if (IS_ERR(macstr))
-+ return PTR_ERR(macstr);
+
-+ for (i = 0; i < ETH_ALEN; i++) {
-+ mac[i] = simple_strtoul(s, &e, 16);
-+ if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
-+ kfree(macstr);
-+ return -ENOENT;
-+ }
-+ s = e+1;
-+ }
++#endif /* __CI_TOOLS_GCC_X86_H__ */
+
-+ kfree(macstr);
-+ return 0;
-+}
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/platform/linux_kernel.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/platform/linux_kernel.h 2008-05-19 00:33:48.578946266 +0300
+@@ -0,0 +1,362 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static void connect(struct backend_info *be)
-+{
-+ int err;
-+ struct xenbus_device *dev = be->dev;
+
-+ err = connect_rings(be);
-+ if (err)
-+ return;
++/*! \cidoxg_include_ci_tools_platform */
+
-+ err = xen_net_read_mac(dev, be->netif->fe_dev_addr);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "parsing %s/mac", dev->nodename);
-+ return;
-+ }
++#ifndef __CI_TOOLS_LINUX_KERNEL_H__
++#define __CI_TOOLS_LINUX_KERNEL_H__
+
-+ xen_net_read_rate(dev, &be->netif->credit_bytes,
-+ &be->netif->credit_usec);
-+ be->netif->remaining_credit = be->netif->credit_bytes;
++/**********************************************************************
++ * Need to know the kernel version.
++ */
+
-+ xenbus_switch_state(dev, XenbusStateConnected);
++#ifndef LINUX_VERSION_CODE
++# include <linux/version.h>
++# ifndef UTS_RELEASE
++ /* 2.6.18 onwards defines UTS_RELEASE in a separate header */
++# include <linux/utsrelease.h>
++# endif
++#endif
+
-+ netif_wake_queue(be->netif->dev);
-+}
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) || \
++ LINUX_VERSION_CODE >= KERNEL_VERSION(2,7,0)
++# error "Linux 2.6 required"
++#endif
+
+
-+static int connect_rings(struct backend_info *be)
-+{
-+ struct xenbus_device *dev = be->dev;
-+ unsigned long tx_ring_ref, rx_ring_ref;
-+ unsigned int evtchn, rx_copy;
-+ int err;
-+ int val;
++#include <linux/slab.h> /* kmalloc / kfree */
++#include <linux/vmalloc.h> /* vmalloc / vfree */
++#include <linux/interrupt.h>/* in_interrupt() */
++#include <linux/in.h>
++#include <linux/in6.h>
++#include <linux/spinlock.h>
++#include <linux/highmem.h>
++#include <linux/smp_lock.h>
++#include <linux/ctype.h>
++#include <linux/uio.h>
++#include <asm/current.h>
++#include <asm/errno.h>
++#include <asm/kmap_types.h>
++#include <asm/semaphore.h>
+
-+ DPRINTK("");
++#include <ci/tools/config.h>
+
-+ err = xenbus_gather(XBT_NIL, dev->otherend,
-+ "tx-ring-ref", "%lu", &tx_ring_ref,
-+ "rx-ring-ref", "%lu", &rx_ring_ref,
-+ "event-channel", "%u", &evtchn, NULL);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err,
-+ "reading %s/ring-ref and event-channel",
-+ dev->otherend);
-+ return err;
-+ }
++#define ci_in_irq in_irq
++#define ci_in_interrupt in_interrupt
++#define ci_in_atomic in_atomic
+
-+ err = xenbus_scanf(XBT_NIL, dev->otherend, "request-rx-copy", "%u",
-+ &rx_copy);
-+ if (err == -ENOENT) {
-+ err = 0;
-+ rx_copy = 0;
-+ }
-+ if (err < 0) {
-+ xenbus_dev_fatal(dev, err, "reading %s/request-rx-copy",
-+ dev->otherend);
-+ return err;
-+ }
-+ be->netif->copying_receiver = !!rx_copy;
+
-+ if (be->netif->dev->tx_queue_len != 0) {
-+ if (xenbus_scanf(XBT_NIL, dev->otherend,
-+ "feature-rx-notify", "%d", &val) < 0)
-+ val = 0;
-+ if (val)
-+ be->netif->can_queue = 1;
-+ else
-+ /* Must be non-zero for pfifo_fast to work. */
-+ be->netif->dev->tx_queue_len = 1;
-+ }
++/**********************************************************************
++ * Misc stuff.
++ */
+
-+ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-sg", "%d", &val) < 0)
-+ val = 0;
-+ if (val) {
-+ be->netif->features |= NETIF_F_SG;
-+ be->netif->dev->features |= NETIF_F_SG;
-+ }
++#ifdef BUG
++# define CI_BOMB BUG
++#endif
+
-+ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-gso-tcpv4", "%d",
-+ &val) < 0)
-+ val = 0;
-+ if (val) {
-+ be->netif->features |= NETIF_F_TSO;
-+ be->netif->dev->features |= NETIF_F_TSO;
-+ }
++ci_inline void* __ci_alloc(size_t n)
++{ return kmalloc(n, (in_interrupt() ? GFP_ATOMIC : GFP_KERNEL)); }
+
-+ if (xenbus_scanf(XBT_NIL, dev->otherend, "feature-no-csum-offload",
-+ "%d", &val) < 0)
-+ val = 0;
-+ if (val) {
-+ be->netif->features &= ~NETIF_F_IP_CSUM;
-+ be->netif->dev->features &= ~NETIF_F_IP_CSUM;
-+ }
++ci_inline void* __ci_atomic_alloc(size_t n)
++{ return kmalloc(n, GFP_ATOMIC ); }
+
-+ /* Map the shared frame, irq etc. */
-+ err = netif_map(be->netif, tx_ring_ref, rx_ring_ref, evtchn);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err,
-+ "mapping shared-frames %lu/%lu port %u",
-+ tx_ring_ref, rx_ring_ref, evtchn);
-+ return err;
-+ }
-+ return 0;
-+}
++ci_inline void __ci_free(void* p) { return kfree(p); }
++ci_inline void* __ci_vmalloc(size_t n) { return vmalloc(n); }
++ci_inline void __ci_vfree(void* p) { return vfree(p); }
+
+
-+/* ** Driver Registration ** */
++#if CI_MEMLEAK_DEBUG_ALLOC_TABLE
++ #define ci_alloc(s) ci_alloc_memleak_debug (s, __FILE__, __LINE__)
++ #define ci_atomic_alloc(s) ci_atomic_alloc_memleak_debug(s, __FILE__, __LINE__)
++ #define ci_free ci_free_memleak_debug
++ #define ci_vmalloc(s) ci_vmalloc_memleak_debug (s, __FILE__,__LINE__)
++ #define ci_vfree ci_vfree_memleak_debug
++ #define ci_alloc_fn ci_alloc_fn_memleak_debug
++ #define ci_vmalloc_fn ci_vmalloc_fn_memleak_debug
++#else /* !CI_MEMLEAK_DEBUG_ALLOC_TABLE */
++ #define ci_alloc_fn __ci_alloc
++ #define ci_vmalloc_fn __ci_vmalloc
++#endif
+
++#ifndef ci_alloc
++ #define ci_atomic_alloc __ci_atomic_alloc
++ #define ci_alloc __ci_alloc
++ #define ci_free __ci_free
++ #define ci_vmalloc __ci_vmalloc
++ #define ci_vmalloc_fn __ci_vmalloc
++ #define ci_vfree __ci_vfree
++#endif
+
-+static struct xenbus_device_id netback_ids[] = {
-+ { "vif" },
-+ { "" }
-+};
++#define ci_sprintf sprintf
++#define ci_vsprintf vsprintf
++#define ci_snprintf snprintf
++#define ci_vsnprintf vsnprintf
++#define ci_sscanf sscanf
+
+
-+static struct xenbus_driver netback = {
-+ .name = "vif",
-+ .owner = THIS_MODULE,
-+ .ids = netback_ids,
-+ .probe = netback_probe,
-+ .remove = netback_remove,
-+ .uevent = netback_uevent,
-+ .otherend_changed = frontend_changed,
-+};
++#define CI_LOG_FN_DEFAULT ci_log_syslog
+
+
-+void netif_xenbus_init(void)
-+{
-+ xenbus_register_backend(&netback);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netfront/accel.c linux-2.6.18-xen.hg/drivers/xen/netfront/accel.c
---- linux-2.6.18/drivers/xen/netfront/accel.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netfront/accel.c 2007-12-23 11:15:34.057935783 +0100
-@@ -0,0 +1,833 @@
-+/******************************************************************************
-+ * Virtual network driver for conversing with remote driver backends.
++/*--------------------------------------------------------------------
+ *
-+ * Copyright (C) 2007 Solarflare Communications, Inc.
++ * irqs_disabled - needed for kmap helpers on some kernels
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
++ *--------------------------------------------------------------------*/
++#ifdef irqs_disabled
++# define ci_irqs_disabled irqs_disabled
++#else
++# if defined(__i386__) | defined(__x86_64__)
++# define ci_irqs_disabled(x) \
++ ({ \
++ unsigned long flags; \
++ local_save_flags(flags); \
++ !(flags & (1<<9)); \
++ })
++# else
++# error "Need to implement irqs_disabled() for your architecture"
++# endif
++#endif
++
++
++/**********************************************************************
++ * kmap helpers.
+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
++ * Use ci_k(un)map for code paths which are not in an atomic context.
++ * For atomic code you need to use ci_k(un)map_in_atomic. This will grab
++ * one of the per-CPU kmap slots.
+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
++ * NB in_interrupt != in_irq. If you don't know the difference then
++ * don't use kmap_in_atomic
++ *
++ * 2.4 allocates kmap slots by function. We are going to re-use the
++ * skb module's slot - we also use the same interlock
++ *
++ * 2.6 allocates kmap slots by type as well as by function. We are
++ * going to use the currently (2.6.10) unsused SOFTIRQ slot
+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
+ */
+
-+#include <linux/netdevice.h>
-+#include <linux/skbuff.h>
-+#include <linux/list.h>
-+#include <linux/mutex.h>
++ci_inline void* ci_kmap(struct page *page) {
++ CI_DEBUG(if( ci_in_atomic() | ci_in_interrupt() | ci_in_irq() ) BUG());
++ return kmap(page);
++}
+
-+#include <xen/xenbus.h>
++ci_inline void ci_kunmap(struct page *page) {
++ kunmap(page);
++}
+
-+#include "netfront.h"
++#define CI_KM_SLOT KM_SOFTIRQ0
+
-+#define DPRINTK(fmt, args...) \
-+ pr_debug("netfront/accel (%s:%d) " fmt, \
-+ __FUNCTION__, __LINE__, ##args)
-+#define IPRINTK(fmt, args...) \
-+ printk(KERN_INFO "netfront/accel: " fmt, ##args)
-+#define WPRINTK(fmt, args...) \
-+ printk(KERN_WARNING "netfront/accel: " fmt, ##args)
+
-+static int netfront_remove_accelerator(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+static int netfront_load_accelerator(struct netfront_info *np,
-+ struct xenbus_device *dev,
-+ const char *frontend);
++typedef struct semaphore ci_semaphore_t;
+
-+/*
-+ * List of all netfront accelerator plugin modules available. Each
-+ * list entry is of type struct netfront_accelerator.
-+ */
-+static struct list_head accelerators_list;
++ci_inline void
++ci_sem_init (ci_semaphore_t *sem, int val) {
++ sema_init (sem, val);
++}
+
-+/* Lock to protect access to accelerators_list */
-+static spinlock_t accelerators_lock;
++ci_inline void
++ci_sem_down (ci_semaphore_t *sem) {
++ down (sem);
++}
+
-+/* Mutex to prevent concurrent loads and suspends, etc. */
-+DEFINE_MUTEX(accelerator_mutex);
++ci_inline int
++ci_sem_trydown (ci_semaphore_t *sem) {
++ return down_trylock (sem);
++}
+
-+void netif_init_accel(void)
++ci_inline void
++ci_sem_up (ci_semaphore_t *sem) {
++ up (sem);
++}
++
++ci_inline int
++ci_sem_get_count(ci_semaphore_t *sem) {
++ return sem->count.counter;
++}
++
++ci_inline void* ci_kmap_in_atomic(struct page *page)
+{
-+ INIT_LIST_HEAD(&accelerators_list);
-+ spin_lock_init(&accelerators_lock);
++ CI_DEBUG(if( ci_in_irq() ) BUG());
++
++ /* iSCSI can call without in_interrupt() but with irqs_disabled()
++ and in a context that can't sleep, so we need to check that
++ too */
++ if(ci_in_interrupt() || ci_irqs_disabled())
++ return kmap_atomic(page, CI_KM_SLOT);
++ else
++ return kmap(page);
+}
+
-+void netif_exit_accel(void)
++ci_inline void ci_kunmap_in_atomic(struct page *page, void* kaddr)
+{
-+ struct netfront_accelerator *accelerator, *tmp;
-+ unsigned long flags;
++ CI_DEBUG(if( ci_in_irq() ) BUG());
+
-+ spin_lock_irqsave(&accelerators_lock, flags);
++ /* iSCSI can call without in_interrupt() but with irqs_disabled()
++ and in a context that can't sleep, so we need to check that
++ too */
++ if(ci_in_interrupt() || ci_irqs_disabled())
++ kunmap_atomic(kaddr, CI_KM_SLOT);
++ else
++ kunmap(page);
++}
+
-+ list_for_each_entry_safe(accelerator, tmp, &accelerators_list, link) {
-+ BUG_ON(!list_empty(&accelerator->vif_states));
++/**********************************************************************
++ * spinlock implementation: used by <ci/tools/spinlock.h>
++ */
+
-+ list_del(&accelerator->link);
-+ kfree(accelerator->frontend);
-+ kfree(accelerator);
-+ }
++#define CI_HAVE_SPINLOCKS
+
-+ spin_unlock_irqrestore(&accelerators_lock, flags);
-+}
++typedef ci_uintptr_t ci_lock_holder_t;
++#define ci_lock_thisthread (ci_lock_holder_t)current
++#define ci_lock_no_holder (ci_lock_holder_t)NULL
+
++typedef spinlock_t ci_lock_i;
++typedef spinlock_t ci_irqlock_i;
++typedef unsigned long ci_irqlock_state_t;
+
-+/*
-+ * Watch the configured accelerator and change plugin if it's modified
++#define IRQLOCK_CYCLES 500000
++
++#define ci_lock_ctor_i(l) spin_lock_init(l)
++#define ci_lock_dtor_i(l) do{}while(0)
++#define ci_lock_lock_i(l) spin_lock(l)
++#define ci_lock_trylock_i(l) spin_trylock(l)
++#define ci_lock_unlock_i(l) spin_unlock(l)
++
++#define ci_irqlock_ctor_i(l) spin_lock_init(l)
++#define ci_irqlock_dtor_i(l) do{}while(0)
++#define ci_irqlock_lock_i(l,s) spin_lock_irqsave(l,*(s))
++#define ci_irqlock_unlock_i(l,s) spin_unlock_irqrestore(l, *(s))
++
++
++/**********************************************************************
++ * register access
+ */
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
-+static void accel_watch_work(struct work_struct *context)
-+#else
-+static void accel_watch_work(void *context)
-+#endif
-+{
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
-+ struct netfront_accel_vif_state *vif_state =
-+ container_of(context, struct netfront_accel_vif_state,
-+ accel_work);
++
++#include <asm/io.h>
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,9)
++typedef volatile void __iomem* ioaddr_t;
+#else
-+ struct netfront_accel_vif_state *vif_state =
-+ (struct netfront_accel_vif_state *)context;
++typedef unsigned long ioaddr_t;
+#endif
-+ struct netfront_info *np = vif_state->np;
-+ char *accel_frontend;
-+ int accel_len, rc = -1;
+
-+ mutex_lock(&accelerator_mutex);
+
-+ accel_frontend = xenbus_read(XBT_NIL, np->xbdev->otherend,
-+ "accel-frontend", &accel_len);
-+ if (IS_ERR(accel_frontend)) {
-+ accel_frontend = NULL;
-+ netfront_remove_accelerator(np, np->xbdev);
-+ } else {
-+ /* If this is the first time, request the accelerator,
-+ otherwise only request one if it has changed */
-+ if (vif_state->accel_frontend == NULL) {
-+ rc = netfront_load_accelerator(np, np->xbdev,
-+ accel_frontend);
-+ } else {
-+ if (strncmp(vif_state->accel_frontend, accel_frontend,
-+ accel_len)) {
-+ netfront_remove_accelerator(np, np->xbdev);
-+ rc = netfront_load_accelerator(np, np->xbdev,
-+ accel_frontend);
-+ }
-+ }
-+ }
+
-+ /* Get rid of previous state and replace with the new name */
-+ if (vif_state->accel_frontend != NULL)
-+ kfree(vif_state->accel_frontend);
-+ vif_state->accel_frontend = accel_frontend;
++/**********************************************************************
++ * thread implementation -- kernel dependancies probably should be
++ * moved to driver/linux_kernel.h
++ */
+
-+ mutex_unlock(&accelerator_mutex);
++#define ci_linux_daemonize(name) daemonize(name)
+
-+ if (rc == 0) {
-+ DPRINTK("requesting module %s\n", accel_frontend);
-+ request_module("%s", accel_frontend);
-+ /*
-+ * Module should now call netfront_accelerator_loaded() once
-+ * it's up and running, and we can continue from there
-+ */
-+ }
-+}
++#include <linux/workqueue.h>
+
+
-+static void accel_watch_changed(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
-+{
-+ struct netfront_accel_vif_state *vif_state =
-+ container_of(watch, struct netfront_accel_vif_state,
-+ accel_watch);
-+ schedule_work(&vif_state->accel_work);
-+}
++typedef struct {
++ void* (*fn)(void* arg);
++ void* arg;
++ const char* name;
++ int thrd_id;
++ struct completion exit_event;
++ struct work_struct keventd_witem;
++} ci_kernel_thread_t;
+
+
-+void netfront_accelerator_add_watch(struct netfront_info *np)
-+{
-+ int err;
-+
-+ /* Check we're not trying to overwrite an existing watch */
-+ BUG_ON(np->accel_vif_state.accel_watch.node != NULL);
++typedef ci_kernel_thread_t* cithread_t;
+
-+ /* Get a watch on the accelerator plugin */
-+ err = xenbus_watch_path2(np->xbdev, np->xbdev->otherend,
-+ "accel-frontend",
-+ &np->accel_vif_state.accel_watch,
-+ accel_watch_changed);
-+ if (err) {
-+ DPRINTK("%s: Failed to register accel watch: %d\n",
-+ __FUNCTION__, err);
-+ np->accel_vif_state.accel_watch.node = NULL;
-+ }
-+}
+
++extern int cithread_create(cithread_t* tid, void* (*fn)(void*), void* arg,
++ const char* name);
++extern int cithread_detach(cithread_t kt);
++extern int cithread_join(cithread_t kt);
+
-+static
-+void netfront_accelerator_remove_watch(struct netfront_info *np)
-+{
-+ struct netfront_accel_vif_state *vif_state = &np->accel_vif_state;
+
-+ /* Get rid of watch on accelerator plugin */
-+ if (vif_state->accel_watch.node != NULL) {
-+ unregister_xenbus_watch(&vif_state->accel_watch);
-+ kfree(vif_state->accel_watch.node);
-+ vif_state->accel_watch.node = NULL;
++/* Kernel sysctl variables. */
++extern int sysctl_tcp_wmem[3];
++extern int sysctl_tcp_rmem[3];
++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
++#define LINUX_HAS_SYSCTL_MEM_MAX
++extern ci_uint32 sysctl_wmem_max;
++extern ci_uint32 sysctl_rmem_max;
++#endif
+
-+ flush_scheduled_work();
+
-+ /* Clean up any state left from watch */
-+ if (vif_state->accel_frontend != NULL) {
-+ kfree(vif_state->accel_frontend);
-+ vif_state->accel_frontend = NULL;
-+ }
-+ }
-+}
++/*--------------------------------------------------------------------
++ *
++ * ci_bigbuf_t: An abstraction of a large buffer. Needed because in the
++ * Linux kernel, large buffers need to be allocated with vmalloc(), whereas
++ * smaller buffers should use kmalloc(). This abstraction chooses the
++ * appropriate mechansim.
++ *
++ *--------------------------------------------------------------------*/
+
++typedef struct {
++ char* p;
++ int is_vmalloc;
++} ci_bigbuf_t;
+
-+/*
-+ * Initialise the accel_vif_state field in the netfront state
-+ */
-+void init_accelerator_vif(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ np->accelerator = NULL;
+
-+ /* It's assumed that these things don't change */
-+ np->accel_vif_state.np = np;
-+ np->accel_vif_state.dev = dev;
++ci_inline int ci_bigbuf_alloc(ci_bigbuf_t* bb, size_t bytes) {
++ if( bytes >= CI_PAGE_SIZE && ! ci_in_atomic() ) {
++ bb->is_vmalloc = 1;
++ if( (bb->p = vmalloc(bytes)) ) return 0;
++ }
++ bb->is_vmalloc = 0;
++ bb->p = kmalloc(bytes, ci_in_interrupt() ? GFP_ATOMIC : GFP_KERNEL);
++ return bb->p ? 0 : -ENOMEM;
++}
+
-+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
-+ INIT_WORK(&np->accel_vif_state.accel_work, accel_watch_work);
-+#else
-+ INIT_WORK(&np->accel_vif_state.accel_work, accel_watch_work,
-+ &np->accel_vif_state);
-+#endif
++ci_inline void ci_bigbuf_free(ci_bigbuf_t* bb) {
++ if( bb->is_vmalloc ) vfree(bb->p);
++ else kfree(bb->p);
+}
+
++ci_inline char* ci_bigbuf_ptr(ci_bigbuf_t* bb)
++{ return bb->p; }
+
-+/*
-+ * Compare a frontend description string against an accelerator to see
-+ * if they match. Would ultimately be nice to replace the string with
-+ * a unique numeric identifier for each accelerator.
++/**********************************************************************
++ * struct iovec abstraction (for Windows port)
+ */
-+static int match_accelerator(const char *frontend,
-+ struct netfront_accelerator *accelerator)
++
++typedef struct iovec ci_iovec;
++
++/* Accessors for buffer/length */
++#define CI_IOVEC_BASE(i) ((i)->iov_base)
++#define CI_IOVEC_LEN(i) ((i)->iov_len)
++
++/**********************************************************************
++ * Signals
++ */
++
++ci_inline void
++ci_send_sig(int signum)
+{
-+ return strcmp(frontend, accelerator->frontend) == 0;
++ send_sig(signum, current, 0);
+}
+
++#endif /* __CI_TOOLS_LINUX_KERNEL_H__ */
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netback/ci/tools/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netback/ci/tools/sysdep.h 2008-05-19 00:33:48.582946497 +0300
+@@ -0,0 +1,132 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++/*! \cidoxg_include_ci_tools */
+
-+/*
-+ * Add a frontend vif to the list of vifs that is using a netfront
-+ * accelerator plugin module.
++#ifndef __CI_TOOLS_SYSDEP_H__
++#define __CI_TOOLS_SYSDEP_H__
++
++/* Make this header self-sufficient */
++#include <ci/compat.h>
++#include <ci/tools/log.h>
++#include <ci/tools/debug.h>
++
++
++/**********************************************************************
++ * Platform dependencies.
+ */
-+static void add_accelerator_vif(struct netfront_accelerator *accelerator,
-+ struct netfront_info *np)
-+{
-+ unsigned long flags;
+
-+ /* Need lock to write list */
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++#if defined(__KERNEL__)
+
-+ if (np->accelerator == NULL) {
-+ np->accelerator = accelerator;
-+
-+ list_add(&np->accel_vif_state.link, &accelerator->vif_states);
-+ } else {
-+ /*
-+ * May get here legitimately if suspend_cancel is
-+ * called, but in that case configuration should not
-+ * have changed
-+ */
-+ BUG_ON(np->accelerator != accelerator);
-+ }
++# if defined(__linux__)
++# include <ci/tools/platform/linux_kernel.h>
++# elif defined(_WIN32)
++# include <ci/tools/platform/win32_kernel.h>
++# elif defined(__sun__)
++# include <ci/tools/platform/sunos_kernel.h>
++# else
++# error Unknown platform.
++# endif
+
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
-+}
++#elif defined(_WIN32)
+
++# include <ci/tools/platform/win32.h>
+
-+/*
-+ * Initialise the state to track an accelerator plugin module.
-+ */
-+static int init_accelerator(const char *frontend,
-+ struct netfront_accelerator **result,
-+ struct netfront_accel_hooks *hooks)
-+{
-+ struct netfront_accelerator *accelerator =
-+ kmalloc(sizeof(struct netfront_accelerator), GFP_KERNEL);
-+ unsigned long flags;
-+ int frontend_len;
++#elif defined(__unix__)
+
-+ if (!accelerator) {
-+ DPRINTK("no memory for accelerator\n");
-+ return -ENOMEM;
-+ }
++# include <ci/tools/platform/unix.h>
+
-+ frontend_len = strlen(frontend) + 1;
-+ accelerator->frontend = kmalloc(frontend_len, GFP_KERNEL);
-+ if (!accelerator->frontend) {
-+ DPRINTK("no memory for accelerator\n");
-+ kfree(accelerator);
-+ return -ENOMEM;
-+ }
-+ strlcpy(accelerator->frontend, frontend, frontend_len);
-+
-+ INIT_LIST_HEAD(&accelerator->vif_states);
-+ spin_lock_init(&accelerator->vif_states_lock);
++#else
+
-+ accelerator->hooks = hooks;
++# error Unknown platform.
+
-+ spin_lock_irqsave(&accelerators_lock, flags);
-+ list_add(&accelerator->link, &accelerators_list);
-+ spin_unlock_irqrestore(&accelerators_lock, flags);
++#endif
+
-+ *result = accelerator;
++#if defined(__linux__)
++/*! Linux sendfile() support enable/disable. */
++# define CI_HAVE_SENDFILE /* provide sendfile i/f */
+
-+ return 0;
-+}
++# define CI_HAVE_OS_NOPAGE
++#endif
+
++#if defined(__sun__)
++# define CI_HAVE_SENDFILE /* provide sendfile i/f */
++# define CI_HAVE_SENDFILEV /* provide sendfilev i/f */
+
-+/*
-+ * Modify the hooks stored in the per-vif state to match that in the
-+ * netfront accelerator's state.
++# define CI_IOCTL_SENDFILE /* use efrm CI_SENDFILEV ioctl */
++#endif
++
++#if defined(_WIN32)
++typedef ci_uint32 ci_uerr_t; /* range of OS user-mode return codes */
++typedef ci_uint32 ci_kerr_t; /* range of OS kernel-mode return codes */
++#elif defined(__unix__)
++typedef ci_int32 ci_uerr_t; /* range of OS user-mode return codes */
++typedef ci_int32 ci_kerr_t; /* range of OS kernel-mode return codes */
++#endif
++
++
++/**********************************************************************
++ * Compiler and processor dependencies.
+ */
-+static void
-+accelerator_set_vif_state_hooks(struct netfront_accel_vif_state *vif_state)
-+{
-+ /* This function must be called with the vif_states_lock held */
+
-+ DPRINTK("%p\n",vif_state);
++#if defined(__GNUC__)
+
-+ /* Make sure there are no data path operations going on */
-+ netif_poll_disable(vif_state->np->netdev);
-+ netif_tx_lock_bh(vif_state->np->netdev);
++#if defined(__i386__) || defined(__x86_64__)
++# include <ci/tools/platform/gcc_x86.h>
++#elif defined(__PPC__)
++# include <ci/tools/platform/gcc_ppc.h>
++#elif defined(__ia64__)
++# include <ci/tools/platform/gcc_ia64.h>
++#else
++# error Unknown processor.
++#endif
+
-+ vif_state->hooks = vif_state->np->accelerator->hooks;
++#elif defined(_MSC_VER)
+
-+ netif_tx_unlock_bh(vif_state->np->netdev);
-+ netif_poll_enable(vif_state->np->netdev);
-+}
++#if defined(__i386__)
++# include <ci/tools/platform/msvc_x86.h>
++# elif defined(__x86_64__)
++# include <ci/tools/platform/msvc_x86_64.h>
++#else
++# error Unknown processor.
++#endif
+
++#elif defined(__PGI)
+
-+static void accelerator_probe_new_vif(struct netfront_info *np,
-+ struct xenbus_device *dev,
-+ struct netfront_accelerator *accelerator)
-+{
-+ struct netfront_accel_hooks *hooks;
-+ unsigned long flags;
++# include <ci/tools/platform/pg_x86.h>
+
-+ DPRINTK("\n");
++#elif defined(__INTEL_COMPILER)
+
-+ /* Include this frontend device on the accelerator's list */
-+ add_accelerator_vif(accelerator, np);
-+
-+ hooks = accelerator->hooks;
-+
-+ if (hooks) {
-+ if (hooks->new_device(np->netdev, dev) == 0) {
-+ spin_lock_irqsave
-+ (&accelerator->vif_states_lock, flags);
++/* Intel compilers v7 claim to be very gcc compatible. */
++# include <ci/tools/platform/gcc_x86.h>
+
-+ accelerator_set_vif_state_hooks(&np->accel_vif_state);
++#else
++# error Unknown compiler.
++#endif
+
-+ spin_unlock_irqrestore
-+ (&accelerator->vif_states_lock, flags);
-+ }
-+ }
+
-+ return;
-+}
++#endif /* __CI_TOOLS_SYSDEP_H__ */
+
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/Makefile 2008-05-19 00:33:48.582946497 +0300
+@@ -0,0 +1,11 @@
++EXTRA_CFLAGS += -Idrivers/xen/sfc_netfront -Idrivers/xen/sfc_netutil -Idrivers/xen/netfront
++EXTRA_CFLAGS += -D__ci_driver__
++EXTRA_CFLAGS += -Werror
+
-+/*
-+ * Request that a particular netfront accelerator plugin is loaded.
-+ * Usually called as a result of the vif configuration specifying
-+ * which one to use. Must be called with accelerator_mutex held
++ifdef GCOV
++EXTRA_CFLAGS += -fprofile-arcs -ftest-coverage -DEFX_GCOV
++endif
++
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_FRONTEND) := sfc_netfront.o
++
++sfc_netfront-objs := accel_msg.o accel_bufs.o accel_netfront.o accel_vi.o accel_xenbus.o accel_tso.o accel_ssr.o accel_debugfs.o falcon_event.o falcon_vi.o pt_tx.o vi_init.o
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel.h 2008-05-19 00:33:48.598947419 +0300
+@@ -0,0 +1,477 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
-+static int netfront_load_accelerator(struct netfront_info *np,
-+ struct xenbus_device *dev,
-+ const char *frontend)
-+{
-+ struct netfront_accelerator *accelerator;
-+ int rc = 0;
+
-+ DPRINTK(" %s\n", frontend);
++#ifndef NETFRONT_ACCEL_H
++#define NETFRONT_ACCEL_H
+
-+ /*
-+ * Look at list of loaded accelerators to see if the requested
-+ * one is already there
-+ */
-+ list_for_each_entry(accelerator, &accelerators_list, link) {
-+ if (match_accelerator(frontend, accelerator)) {
-+ accelerator_probe_new_vif(np, dev, accelerator);
-+ return 0;
-+ }
-+ }
++#include "accel_msg_iface.h"
++#include "accel_cuckoo_hash.h"
++#include "accel_bufs.h"
++
++#include "etherfabric/ef_vi.h"
++
++#include <xen/xenbus.h>
++#include <xen/evtchn.h>
+
-+ /* Couldn't find it, so create a new one and load the module */
-+ if ((rc = init_accelerator(frontend, &accelerator, NULL)) < 0) {
-+ return rc;
-+ }
++#include <linux/kernel.h>
++#include <linux/list.h>
+
-+ /* Include this frontend device on the accelerator's list */
-+ add_accelerator_vif(accelerator, np);
++enum netfront_accel_post_status {
++ NETFRONT_ACCEL_STATUS_GOOD,
++ NETFRONT_ACCEL_STATUS_BUSY,
++ NETFRONT_ACCEL_STATUS_CANT
++};
+
-+ return rc;
-+}
++#define NETFRONT_ACCEL_STATS 1
++#if NETFRONT_ACCEL_STATS
++#define NETFRONT_ACCEL_STATS_OP(x) x
++#else
++#define NETFRONT_ACCEL_STATS_OP(x)
++#endif
+
+
-+/*
-+ * Go through all the netfront vifs and see if they have requested
-+ * this accelerator. Notify the accelerator plugin of the relevant
-+ * device if so. Called when an accelerator plugin module is first
-+ * loaded and connects to netfront.
-+ */
-+static void
-+accelerator_probe_vifs(struct netfront_accelerator *accelerator,
-+ struct netfront_accel_hooks *hooks)
-+{
-+ struct netfront_accel_vif_state *vif_state, *tmp;
-+ unsigned long flags;
++enum netfront_accel_msg_state {
++ NETFRONT_ACCEL_MSG_NONE = 0,
++ NETFRONT_ACCEL_MSG_HELLO = 1,
++ NETFRONT_ACCEL_MSG_HW = 2
++};
+
-+ DPRINTK("%p\n", accelerator);
+
-+ /*
-+ * Store the hooks for future calls to probe a new device, and
-+ * to wire into the vif_state once the accelerator plugin is
-+ * ready to accelerate each vif
-+ */
-+ BUG_ON(hooks == NULL);
-+ accelerator->hooks = hooks;
++typedef struct {
++ u32 in_progress;
++ u32 total_len;
++ struct sk_buff *skb;
++} netfront_accel_jumbo_state;
+
-+ /*
-+ * currently hold accelerator_mutex, so don't need
-+ * vif_states_lock to read the list
-+ */
-+ list_for_each_entry_safe(vif_state, tmp, &accelerator->vif_states,
-+ link) {
-+ struct netfront_info *np = vif_state->np;
-+
-+ if (hooks->new_device(np->netdev, vif_state->dev) == 0) {
-+ spin_lock_irqsave
-+ (&accelerator->vif_states_lock, flags);
+
-+ accelerator_set_vif_state_hooks(vif_state);
++struct netfront_accel_ssr_state {
++ /** List of tracked connections. */
++ struct list_head conns;
+
-+ spin_unlock_irqrestore
-+ (&accelerator->vif_states_lock, flags);
-+ }
-+ }
-+}
++ /** Free efx_ssr_conn instances. */
++ struct list_head free_conns;
++};
+
+
-+/*
-+ * Called by the netfront accelerator plugin module when it has loaded
-+ */
-+int netfront_accelerator_loaded(int version, const char *frontend,
-+ struct netfront_accel_hooks *hooks)
-+{
-+ struct netfront_accelerator *accelerator;
++struct netfront_accel_netdev_stats {
++ /* Fastpath stats. */
++ u32 fastpath_rx_pkts;
++ u32 fastpath_rx_bytes;
++ u32 fastpath_rx_errors;
++ u32 fastpath_tx_pkts;
++ u32 fastpath_tx_bytes;
++ u32 fastpath_tx_errors;
++};
+
-+ if (version != NETFRONT_ACCEL_VERSION) {
-+ if (version > NETFRONT_ACCEL_VERSION) {
-+ /* Caller has higher version number, leave it
-+ up to them to decide whether to continue.
-+ They can re-call with a lower number if
-+ they're happy to be compatible with us */
-+ return NETFRONT_ACCEL_VERSION;
-+ } else {
-+ /* We have a more recent version than caller.
-+ Currently reject, but may in future be able
-+ to be backwardly compatible */
-+ return -EPROTO;
-+ }
-+ }
+
-+ mutex_lock(&accelerator_mutex);
++struct netfront_accel_netdev_dbfs {
++ struct dentry *fastpath_rx_pkts;
++ struct dentry *fastpath_rx_bytes;
++ struct dentry *fastpath_rx_errors;
++ struct dentry *fastpath_tx_pkts;
++ struct dentry *fastpath_tx_bytes;
++ struct dentry *fastpath_tx_errors;
++};
+
-+ /*
-+ * Look through list of accelerators to see if it has already
-+ * been requested
-+ */
-+ list_for_each_entry(accelerator, &accelerators_list, link) {
-+ if (match_accelerator(frontend, accelerator)) {
-+ accelerator_probe_vifs(accelerator, hooks);
-+ goto out;
-+ }
-+ }
+
-+ /*
-+ * If it wasn't in the list, add it now so that when it is
-+ * requested the caller will find it
-+ */
-+ DPRINTK("Couldn't find matching accelerator (%s)\n",
-+ frontend);
++struct netfront_accel_stats {
++ /** Fast path events */
++ u64 fastpath_tx_busy;
+
-+ init_accelerator(frontend, &accelerator, hooks);
++ /** TX DMA queue status */
++ u64 fastpath_tx_completions;
+
-+ out:
-+ mutex_unlock(&accelerator_mutex);
-+ return 0;
-+}
-+EXPORT_SYMBOL_GPL(netfront_accelerator_loaded);
++ /** The number of events processed. */
++ u64 event_count;
+
++ /** Number of frame trunc events seen on fastpath */
++ u64 fastpath_frm_trunc;
+
-+/*
-+ * Remove the hooks from a single vif state.
-+ */
-+static void
-+accelerator_remove_single_hook(struct netfront_accelerator *accelerator,
-+ struct netfront_accel_vif_state *vif_state)
-+{
-+ /* Make sure there are no data path operations going on */
-+ netif_poll_disable(vif_state->np->netdev);
-+ netif_tx_lock_bh(vif_state->np->netdev);
++ /** Number of no rx descriptor trunc events seen on fastpath */
++ u64 rx_no_desc_trunc;
+
-+ /*
-+ * Remove the hooks, but leave the vif_state on the
-+ * accelerator's list as that signifies this vif is
-+ * interested in using that accelerator if it becomes
-+ * available again
-+ */
-+ vif_state->hooks = NULL;
-+
-+ netif_tx_unlock_bh(vif_state->np->netdev);
-+ netif_poll_enable(vif_state->np->netdev);
-+}
++ /** The number of misc bad events (e.g. RX_DISCARD) processed. */
++ u64 bad_event_count;
+
++ /** Number of events dealt with in poll loop */
++ u32 events_per_poll_max;
++ u32 events_per_poll_tx_max;
++ u32 events_per_poll_rx_max;
+
-+/*
-+ * Safely remove the accelerator function hooks from a netfront state.
-+ */
-+static void accelerator_remove_hooks(struct netfront_accelerator *accelerator)
-+{
-+ struct netfront_accel_hooks *hooks;
-+ struct netfront_accel_vif_state *vif_state, *tmp;
-+ unsigned long flags;
++ /** Largest number of concurrently outstanding tx descriptors */
++ u32 fastpath_tx_pending_max;
+
-+ /* Mutex is held so don't need vif_states_lock to iterate list */
-+ list_for_each_entry_safe(vif_state, tmp,
-+ &accelerator->vif_states,
-+ link) {
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++ /** The number of events since the last interrupts. */
++ u32 event_count_since_irq;
+
-+ if(vif_state->hooks) {
-+ hooks = vif_state->hooks;
-+
-+ /* Last chance to get statistics from the accelerator */
-+ hooks->get_stats(vif_state->np->netdev,
-+ &vif_state->np->stats);
++ /** The max number of events between interrupts. */
++ u32 events_per_irq_max;
+
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock,
-+ flags);
++ /** The number of interrupts. */
++ u64 irq_count;
+
-+ accelerator_remove_single_hook(accelerator, vif_state);
++ /** The number of useless interrupts. */
++ u64 useless_irq_count;
+
-+ accelerator->hooks->remove(vif_state->dev);
-+ } else {
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock,
-+ flags);
-+ }
-+ }
-+
-+ accelerator->hooks = NULL;
-+}
++ /** The number of polls scheduled. */
++ u64 poll_schedule_count;
+
++ /** The number of polls called. */
++ u64 poll_call_count;
+
-+/*
-+ * Called by a netfront accelerator when it is unloaded. This safely
-+ * removes the hooks into the plugin and blocks until all devices have
-+ * finished using it, so on return it is safe to unload.
-+ */
-+void netfront_accelerator_stop(const char *frontend)
-+{
-+ struct netfront_accelerator *accelerator;
-+ unsigned long flags;
++ /** The number of rechecks. */
++ u64 poll_reschedule_count;
+
-+ mutex_lock(&accelerator_mutex);
-+ spin_lock_irqsave(&accelerators_lock, flags);
++ /** Number of times we've called netif_stop_queue/netif_wake_queue */
++ u64 queue_stops;
++ u64 queue_wakes;
+
-+ list_for_each_entry(accelerator, &accelerators_list, link) {
-+ if (match_accelerator(frontend, accelerator)) {
-+ spin_unlock_irqrestore(&accelerators_lock, flags);
++ /** SSR stats */
++ u64 ssr_bursts;
++ u64 ssr_drop_stream;
++ u64 ssr_misorder;
++ u64 ssr_slow_start;
++ u64 ssr_merges;
++ u64 ssr_too_many;
++ u64 ssr_new_stream;
++};
+
-+ accelerator_remove_hooks(accelerator);
+
-+ goto out;
-+ }
-+ }
-+ spin_unlock_irqrestore(&accelerators_lock, flags);
-+ out:
-+ mutex_unlock(&accelerator_mutex);
-+}
-+EXPORT_SYMBOL_GPL(netfront_accelerator_stop);
++struct netfront_accel_dbfs {
++ struct dentry *fastpath_tx_busy;
++ struct dentry *fastpath_tx_completions;
++ struct dentry *fastpath_tx_pending_max;
++ struct dentry *fastpath_frm_trunc;
++ struct dentry *rx_no_desc_trunc;
++ struct dentry *event_count;
++ struct dentry *bad_event_count;
++ struct dentry *events_per_poll_max;
++ struct dentry *events_per_poll_rx_max;
++ struct dentry *events_per_poll_tx_max;
++ struct dentry *event_count_since_irq;
++ struct dentry *events_per_irq_max;
++ struct dentry *irq_count;
++ struct dentry *useless_irq_count;
++ struct dentry *poll_schedule_count;
++ struct dentry *poll_call_count;
++ struct dentry *poll_reschedule_count;
++ struct dentry *queue_stops;
++ struct dentry *queue_wakes;
++ struct dentry *ssr_bursts;
++ struct dentry *ssr_drop_stream;
++ struct dentry *ssr_misorder;
++ struct dentry *ssr_slow_start;
++ struct dentry *ssr_merges;
++ struct dentry *ssr_too_many;
++ struct dentry *ssr_new_stream;
++};
++
+
++typedef struct netfront_accel_vnic {
++ struct netfront_accel_vnic *next;
++
++ struct mutex vnic_mutex;
+
-+/* Helper for call_remove and do_suspend */
-+static int do_remove(struct netfront_info *np, struct xenbus_device *dev,
-+ unsigned long *lock_flags)
-+{
-+ struct netfront_accelerator *accelerator = np->accelerator;
-+ struct netfront_accel_hooks *hooks;
-+ int rc = 0;
-+
-+ if (np->accel_vif_state.hooks) {
-+ hooks = np->accel_vif_state.hooks;
++ spinlock_t tx_lock;
+
-+ /* Last chance to get statistics from the accelerator */
-+ hooks->get_stats(np->netdev, &np->stats);
++ struct netfront_accel_bufpages bufpages;
++ struct netfront_accel_bufinfo *rx_bufs;
++ struct netfront_accel_bufinfo *tx_bufs;
++
++ /** Hardware & VI state */
++ ef_vi vi;
+
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock,
-+ *lock_flags);
++ ef_vi_state *vi_state;
+
-+ /*
-+ * Try and do the opposite of accelerator_probe_new_vif
-+ * to ensure there's no state pointing back at the
-+ * netdev
-+ */
-+ accelerator_remove_single_hook(accelerator,
-+ &np->accel_vif_state);
++ ef_eventq_state evq_state;
+
-+ rc = accelerator->hooks->remove(dev);
++ void *evq_mapping;
+
-+ spin_lock_irqsave(&accelerator->vif_states_lock, *lock_flags);
-+ }
-+
-+ return rc;
-+}
++ /** Hardware dependant state */
++ union {
++ struct {
++ /** Falcon A or B */
++ enum net_accel_hw_type type;
++ u32 *evq_rptr;
++ u32 *doorbell;
++ void *evq_rptr_mapping;
++ void *doorbell_mapping;
++ void *txdmaq_mapping;
++ void *rxdmaq_mapping;
++ } falcon;
++ } hw;
++
++ /** RX DMA queue status */
++ u32 rx_dma_level;
+
++ /** Number of RX descriptors waiting to be pushed to the card. */
++ u32 rx_dma_batched;
++#define NETFRONT_ACCEL_RX_DESC_BATCH 16
+
-+static int netfront_remove_accelerator(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ struct netfront_accelerator *accelerator;
-+ struct netfront_accel_vif_state *tmp_vif_state;
-+ unsigned long flags;
-+ int rc = 0;
++ /**
++ * Hash table of remote mac addresses to decide whether to try
++ * fast path
++ */
++ cuckoo_hash_table fastpath_table;
++ spinlock_t table_lock;
+
-+ /* Check that we've got a device that was accelerated */
-+ if (np->accelerator == NULL)
-+ return rc;
++ /** the local mac address of virtual interface we're accelerating */
++ u8 mac[ETH_ALEN];
+
-+ accelerator = np->accelerator;
++ int rx_pkt_stride;
++ int rx_skb_stride;
+
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
++ /**
++ * Keep track of fragments of jumbo packets as events are
++ * delivered by NIC
++ */
++ netfront_accel_jumbo_state jumbo_state;
+
-+ list_for_each_entry(tmp_vif_state, &accelerator->vif_states,
-+ link) {
-+ if (tmp_vif_state == &np->accel_vif_state) {
-+ list_del(&np->accel_vif_state.link);
-+ break;
-+ }
-+ }
++ struct net_device *net_dev;
+
-+ rc = do_remove(np, dev, &flags);
++ /** These two gate the enabling of fast path operations */
++ int frontend_ready;
++ int backend_netdev_up;
+
-+ np->accelerator = NULL;
++ int irq_enabled;
++ spinlock_t irq_enabled_lock;
+
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
++ int tx_enabled;
+
-+ return rc;
-+}
++ int poll_enabled;
+
++ /** A spare slot for a TX packet. This is treated as an extension
++ * of the DMA queue. */
++ struct sk_buff *tx_skb;
+
-+int netfront_accelerator_call_remove(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ int rc;
-+ netfront_accelerator_remove_watch(np);
-+ mutex_lock(&accelerator_mutex);
-+ rc = netfront_remove_accelerator(np, dev);
-+ mutex_unlock(&accelerator_mutex);
-+ return rc;
-+}
++ /** Keep track of fragments of SSR packets */
++ struct netfront_accel_ssr_state ssr_state;
+
-+
-+int netfront_accelerator_suspend(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ unsigned long flags;
-+ int rc = 0;
-+
-+ netfront_accelerator_remove_watch(np);
++ struct xenbus_device *dev;
+
-+ mutex_lock(&accelerator_mutex);
++ /** Event channel for messages */
++ int msg_channel;
++ int msg_channel_irq;
+
-+ /* Check that we've got a device that was accelerated */
-+ if (np->accelerator == NULL)
-+ goto out;
++ /** Event channel for network interrupts. */
++ int net_channel;
++ int net_channel_irq;
+
-+ /*
-+ * Call the remove accelerator hook, but leave the vif_state
-+ * on the accelerator's list in case there is a suspend_cancel.
-+ */
-+ spin_lock_irqsave(&np->accelerator->vif_states_lock, flags);
-+
-+ rc = do_remove(np, dev, &flags);
++ struct net_accel_shared_page *shared_page;
+
-+ spin_unlock_irqrestore(&np->accelerator->vif_states_lock, flags);
-+ out:
-+ mutex_unlock(&accelerator_mutex);
-+ return rc;
-+}
-+
-+
-+int netfront_accelerator_suspend_cancel(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ struct netfront_accel_vif_state *accel_vif_state = NULL;
-+
-+ mutex_lock(&accelerator_mutex);
++ grant_ref_t ctrl_page_gnt;
++ grant_ref_t msg_page_gnt;
+
-+ /* Check that we've got a device that was accelerated */
-+ if (np->accelerator == NULL)
-+ goto out;
++ /** Message Qs, 1 each way. */
++ sh_msg_fifo2 to_dom0;
++ sh_msg_fifo2 from_dom0;
+
-+ /* Find the vif_state from the accelerator's list */
-+ list_for_each_entry(accel_vif_state, &np->accelerator->vif_states,
-+ link) {
-+ if (accel_vif_state->dev == dev) {
-+ BUG_ON(accel_vif_state != &np->accel_vif_state);
-+
-+ /*
-+ * Kick things off again to restore
-+ * acceleration as it was before suspend
-+ */
-+ accelerator_probe_new_vif(np, dev, np->accelerator);
-+
-+ break;
-+ }
-+ }
-+
-+ out:
-+ mutex_unlock(&accelerator_mutex);
-+ netfront_accelerator_add_watch(np);
-+ return 0;
-+}
-+
-+
-+void netfront_accelerator_resume(struct netfront_info *np,
-+ struct xenbus_device *dev)
-+{
-+ struct netfront_accel_vif_state *accel_vif_state = NULL;
-+ spinlock_t *vif_states_lock;
-+ unsigned long flags;
-+
-+ mutex_lock(&accelerator_mutex);
++ enum netfront_accel_msg_state msg_state;
+
-+ /* Check that we've got a device that was accelerated */
-+ if(np->accelerator == NULL)
-+ goto out;
++ /** Watch on accelstate */
++ struct xenbus_watch backend_accel_watch;
++ /** Watch on frontend's MAC address */
++ struct xenbus_watch mac_address_watch;
+
-+ /* Find the vif_state from the accelerator's list */
-+ list_for_each_entry(accel_vif_state, &np->accelerator->vif_states,
-+ link) {
-+ if (accel_vif_state->dev == dev) {
-+ BUG_ON(accel_vif_state != &np->accel_vif_state);
-+
-+ vif_states_lock = &np->accelerator->vif_states_lock;
-+ spin_lock_irqsave(vif_states_lock, flags);
-+
-+ /*
-+ * Remove it from the accelerator's list so
-+ * state is consistent for probing new vifs
-+ * when they get connected
-+ */
-+ list_del(&accel_vif_state->link);
-+ np->accelerator = NULL;
-+
-+ spin_unlock_irqrestore(vif_states_lock, flags);
-+
-+ break;
-+ }
-+ }
++ /** Work to process received irq/msg */
++ struct work_struct msg_from_bend;
+
-+ out:
-+ mutex_unlock(&accelerator_mutex);
-+ return;
-+}
++ /** Wait queue for changes in accelstate. */
++ wait_queue_head_t state_wait_queue;
+
++ /** The current accelstate of this driver. */
++ XenbusState frontend_state;
+
-+int netfront_check_accelerator_queue_ready(struct net_device *dev,
-+ struct netfront_info *np)
-+{
-+ struct netfront_accelerator *accelerator;
-+ struct netfront_accel_hooks *hooks;
-+ int rc = 1;
-+ unsigned long flags;
++ /** The most recent accelstate seen by the xenbus watch. */
++ XenbusState backend_state;
+
-+ accelerator = np->accelerator;
++ /** Non-zero if we should reject requests to connect. */
++ int removing;
+
-+ /* Call the check_ready accelerator hook. */
-+ if (np->accel_vif_state.hooks && accelerator) {
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
-+ hooks = np->accel_vif_state.hooks;
-+ if (hooks && np->accelerator == accelerator)
-+ rc = np->accel_vif_state.hooks->check_ready(dev);
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
-+ }
++ /** Non-zero if the domU shared state has been initialised. */
++ int domU_state_is_setup;
+
-+ return rc;
-+}
++ /** Non-zero if the dom0 shared state has been initialised. */
++ int dom0_state_is_setup;
+
++ /* Those statistics that are added to the netdev stats */
++ struct netfront_accel_netdev_stats netdev_stats;
++ struct netfront_accel_netdev_stats stats_last_read;
++#ifdef CONFIG_DEBUG_FS
++ struct netfront_accel_netdev_dbfs netdev_dbfs;
++#endif
+
-+void netfront_accelerator_call_stop_napi_irq(struct netfront_info *np,
-+ struct net_device *dev)
-+{
-+ struct netfront_accelerator *accelerator;
-+ struct netfront_accel_hooks *hooks;
-+ unsigned long flags;
++ /* These statistics are internal and optional */
++#if NETFRONT_ACCEL_STATS
++ struct netfront_accel_stats stats;
++#ifdef CONFIG_DEBUG_FS
++ struct netfront_accel_dbfs dbfs;
++#endif
++#endif
+
-+ accelerator = np->accelerator;
++ /** Debufs fs dir for this interface */
++ struct dentry *dbfs_dir;
++} netfront_accel_vnic;
+
-+ /* Call the stop_napi_interrupts accelerator hook. */
-+ if (np->accel_vif_state.hooks && accelerator != NULL) {
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
-+ hooks = np->accel_vif_state.hooks;
-+ if (hooks && np->accelerator == accelerator)
-+ np->accel_vif_state.hooks->stop_napi_irq(dev);
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
-+ }
-+}
+
++/* Module parameters */
++extern unsigned sfc_netfront_max_pages;
++extern unsigned sfc_netfront_buffer_split;
+
-+int netfront_accelerator_call_get_stats(struct netfront_info *np,
-+ struct net_device *dev)
-+{
-+ struct netfront_accelerator *accelerator;
-+ struct netfront_accel_hooks *hooks;
-+ unsigned long flags;
-+ int rc = 0;
++extern const char *frontend_name;
++extern struct netfront_accel_hooks accel_hooks;
++extern struct workqueue_struct *netfront_accel_workqueue;
+
-+ accelerator = np->accelerator;
+
-+ /* Call the get_stats accelerator hook. */
-+ if (np->accel_vif_state.hooks && accelerator != NULL) {
-+ spin_lock_irqsave(&accelerator->vif_states_lock, flags);
-+ hooks = np->accel_vif_state.hooks;
-+ if (hooks && np->accelerator == accelerator)
-+ rc = np->accel_vif_state.hooks->get_stats(dev,
-+ &np->stats);
-+ spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
-+ }
-+ return rc;
-+}
++extern
++void netfront_accel_vi_ctor(netfront_accel_vnic *vnic);
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netfront/Makefile linux-2.6.18-xen.hg/drivers/xen/netfront/Makefile
---- linux-2.6.18/drivers/xen/netfront/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netfront/Makefile 2007-12-23 11:15:34.054602273 +0100
-@@ -0,0 +1,4 @@
++extern
++int netfront_accel_vi_init(netfront_accel_vnic *vnic,
++ struct net_accel_msg_hw *hw_msg);
+
-+obj-$(CONFIG_XEN_NETDEV_FRONTEND) := xennet.o
++extern
++void netfront_accel_vi_dtor(netfront_accel_vnic *vnic);
+
-+xennet-objs := netfront.o accel.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netfront/netfront.c linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.c
---- linux-2.6.18/drivers/xen/netfront/netfront.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.c 2007-12-23 11:15:34.057935783 +0100
-@@ -0,0 +1,2217 @@
-+/******************************************************************************
-+ * Virtual network driver for conversing with remote driver backends.
-+ *
-+ * Copyright (c) 2002-2005, K A Fraser
-+ * Copyright (c) 2005, XenSource Ltd
-+ * Copyright (C) 2007 Solarflare Communications, Inc.
++
++/**
++ * Add new buffers which have been registered with the NIC.
+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
++ * @v vnic The vnic instance to process the response.
+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
++ * The buffers contained in the message are added to the buffer pool.
++ */
++extern
++void netfront_accel_vi_add_bufs(netfront_accel_vnic *vnic, int is_rx);
++
++/**
++ * Put a packet on the tx DMA queue.
+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
++ * @v vnic The vnic instance to accept the packet.
++ * @v skb A sk_buff to send.
+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
++ * Attempt to send a packet. On success, the skb is owned by the DMA
++ * queue and will be released when the completion event arrives.
+ */
++extern enum netfront_accel_post_status
++netfront_accel_vi_tx_post(netfront_accel_vnic *vnic,
++ struct sk_buff *skb);
+
-+#include <linux/module.h>
-+#include <linux/version.h>
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/slab.h>
-+#include <linux/string.h>
-+#include <linux/errno.h>
-+#include <linux/netdevice.h>
-+#include <linux/inetdevice.h>
-+#include <linux/etherdevice.h>
-+#include <linux/skbuff.h>
-+#include <linux/init.h>
-+#include <linux/bitops.h>
-+#include <linux/ethtool.h>
-+#include <linux/in.h>
-+#include <linux/if_ether.h>
-+#include <linux/io.h>
-+#include <linux/moduleparam.h>
-+#include <net/sock.h>
-+#include <net/pkt_sched.h>
-+#include <net/arp.h>
-+#include <net/route.h>
-+#include <asm/uaccess.h>
-+#include <xen/evtchn.h>
-+#include <xen/xenbus.h>
-+#include <xen/interface/io/netif.h>
-+#include <xen/interface/memory.h>
-+#include <xen/balloon.h>
-+#include <asm/page.h>
-+#include <asm/maddr.h>
-+#include <asm/uaccess.h>
-+#include <xen/interface/grant_table.h>
-+#include <xen/gnttab.h>
-+#include <xen/hypercall.h>
-+
-+struct netfront_cb {
-+ struct page *page;
-+ unsigned offset;
-+};
-+
-+#define NETFRONT_SKB_CB(skb) ((struct netfront_cb *)((skb)->cb))
-+
-+#include "netfront.h"
+
-+/*
-+ * Mutually-exclusive module options to select receive data path:
-+ * rx_copy : Packets are copied by network backend into local memory
-+ * rx_flip : Page containing packet data is transferred to our ownership
-+ * For fully-virtualised guests there is no option - copying must be used.
-+ * For paravirtualised guests, flipping is the default.
++/**
++ * Process events in response to an interrupt.
++ *
++ * @v vnic The vnic instance to poll.
++ * @v rx_packets The maximum number of rx packets to process.
++ * @ret rx_done The number of rx packets processed.
++ *
++ * The vnic will process events until there are no more events
++ * remaining or the specified number of rx packets has been processed.
++ * The split from the interrupt call is to allow Linux NAPI
++ * polling.
+ */
-+#ifdef CONFIG_XEN
-+static int MODPARM_rx_copy = 0;
-+module_param_named(rx_copy, MODPARM_rx_copy, bool, 0);
-+MODULE_PARM_DESC(rx_copy, "Copy packets from network card (rather than flip)");
-+static int MODPARM_rx_flip = 0;
-+module_param_named(rx_flip, MODPARM_rx_flip, bool, 0);
-+MODULE_PARM_DESC(rx_flip, "Flip packets from network card (rather than copy)");
-+#else
-+static const int MODPARM_rx_copy = 1;
-+static const int MODPARM_rx_flip = 0;
-+#endif
-+
-+#define RX_COPY_THRESHOLD 256
++extern
++int netfront_accel_vi_poll(netfront_accel_vnic *vnic, int rx_packets);
+
-+/* If we don't have GSO, fake things up so that we never try to use it. */
-+#if defined(NETIF_F_GSO)
-+#define HAVE_GSO 1
-+#define HAVE_TSO 1 /* TSO is a subset of GSO */
-+#define HAVE_CSUM_OFFLOAD 1
-+static inline void dev_disable_gso_features(struct net_device *dev)
-+{
-+ /* Turn off all GSO bits except ROBUST. */
-+ dev->features &= (1 << NETIF_F_GSO_SHIFT) - 1;
-+ dev->features |= NETIF_F_GSO_ROBUST;
-+}
-+#elif defined(NETIF_F_TSO)
-+#define HAVE_GSO 0
-+#define HAVE_TSO 1
+
-+/* Some older kernels cannot cope with incorrect checksums,
-+ * particularly in netfilter. I'm not sure there is 100% correlation
-+ * with the presence of NETIF_F_TSO but it appears to be a good first
-+ * approximiation.
++/**
++ * Iterate over the fragments of a packet buffer.
++ *
++ * @v skb The packet buffer to examine.
++ * @v idx A variable name for the fragment index.
++ * @v data A variable name for the address of the fragment data.
++ * @v length A variable name for the fragment length.
++ * @v code A section of code to execute for each fragment.
++ *
++ * This macro iterates over the fragments in a packet buffer and
++ * executes the code for each of them.
+ */
-+#define HAVE_CSUM_OFFLOAD 0
++#define NETFRONT_ACCEL_PKTBUFF_FOR_EACH_FRAGMENT(skb, frag_idx, \
++ frag_data, frag_len, \
++ code) \
++ do { \
++ int frag_idx; \
++ void *frag_data; \
++ unsigned int frag_len; \
++ \
++ frag_data = skb->data; \
++ frag_len = skb_headlen(skb); \
++ frag_idx = 0; \
++ while (1) { /* For each fragment */ \
++ code; \
++ if (frag_idx >= skb_shinfo(skb)->nr_frags) { \
++ break; \
++ } else { \
++ skb_frag_t *fragment; \
++ fragment = &skb_shinfo(skb)->frags[frag_idx]; \
++ frag_len = fragment->size; \
++ frag_data = ((void*)page_address(fragment->page) \
++ + fragment->page_offset); \
++ }; \
++ frag_idx++; \
++ } \
++ } while(0)
+
-+#define gso_size tso_size
-+#define gso_segs tso_segs
-+static inline void dev_disable_gso_features(struct net_device *dev)
-+{
-+ /* Turn off all TSO bits. */
-+ dev->features &= ~NETIF_F_TSO;
-+}
-+static inline int skb_is_gso(const struct sk_buff *skb)
-+{
-+ return skb_shinfo(skb)->tso_size;
-+}
-+static inline int skb_gso_ok(struct sk_buff *skb, int features)
++static inline
++void netfront_accel_disable_net_interrupts(netfront_accel_vnic *vnic)
+{
-+ return (features & NETIF_F_TSO);
++ mask_evtchn(vnic->net_channel);
+}
+
-+static inline int netif_needs_gso(struct net_device *dev, struct sk_buff *skb)
++static inline
++void netfront_accel_enable_net_interrupts(netfront_accel_vnic *vnic)
+{
-+ return skb_is_gso(skb) &&
-+ (!skb_gso_ok(skb, dev->features) ||
-+ unlikely(skb->ip_summed != CHECKSUM_HW));
++ unmask_evtchn(vnic->net_channel);
+}
++
++void netfront_accel_msg_tx_fastpath(netfront_accel_vnic *vnic, const void *mac,
++ u32 ip, u16 port, u8 protocol);
++
++/* Process an IRQ received from back end driver */
++irqreturn_t netfront_accel_msg_channel_irq_from_bend(int irq, void *context,
++ struct pt_regs *unused);
++irqreturn_t netfront_accel_net_channel_irq_from_bend(int irq, void *context,
++ struct pt_regs *unused);
++
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++extern void netfront_accel_msg_from_bend(struct work_struct *context);
+#else
-+#define HAVE_GSO 0
-+#define HAVE_TSO 0
-+#define HAVE_CSUM_OFFLOAD 0
-+#define netif_needs_gso(dev, skb) 0
-+#define dev_disable_gso_features(dev) ((void)0)
-+#define ethtool_op_set_tso(dev, data) (-ENOSYS)
++extern void netfront_accel_msg_from_bend(void *context);
+#endif
+
-+#define GRANT_INVALID_REF 0
++extern void vnic_stop_fastpath(netfront_accel_vnic *vnic);
+
-+struct netfront_rx_info {
-+ struct netif_rx_response rx;
-+ struct netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX - 1];
-+};
++extern int netfront_accel_probe(struct net_device *net_dev,
++ struct xenbus_device *dev);
++extern int netfront_accel_remove(struct xenbus_device *dev);
++extern void netfront_accel_set_closing(netfront_accel_vnic *vnic);
+
-+/*
-+ * Implement our own carrier flag: the network stack's version causes delays
-+ * when the carrier is re-enabled (in particular, dev_activate() may not
-+ * immediately be called, which can cause packet loss).
-+ */
-+#define netfront_carrier_on(netif) ((netif)->carrier = 1)
-+#define netfront_carrier_off(netif) ((netif)->carrier = 0)
-+#define netfront_carrier_ok(netif) ((netif)->carrier)
++extern int netfront_accel_vi_enable_interrupts(netfront_accel_vnic *vnic);
+
-+/*
-+ * Access macros for acquiring freeing slots in tx_skbs[].
++extern void netfront_accel_debugfs_init(void);
++extern void netfront_accel_debugfs_fini(void);
++extern int netfront_accel_debugfs_create(netfront_accel_vnic *vnic);
++extern int netfront_accel_debugfs_remove(netfront_accel_vnic *vnic);
++
++#endif /* NETFRONT_ACCEL_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_bufs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_bufs.c 2008-05-19 00:33:48.598947419 +0300
+@@ -0,0 +1,393 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+static inline void add_id_to_freelist(struct sk_buff **list, unsigned short id)
-+{
-+ list[id] = list[0];
-+ list[0] = (void *)(unsigned long)id;
-+}
++#include <xen/gnttab.h>
+
-+static inline unsigned short get_id_from_freelist(struct sk_buff **list)
-+{
-+ unsigned int id = (unsigned int)(unsigned long)list[0];
-+ list[0] = list[id];
-+ return id;
-+}
++#include "accel_bufs.h"
++#include "accel_util.h"
+
-+static inline int xennet_rxidx(RING_IDX idx)
-+{
-+ return idx & (NET_RX_RING_SIZE - 1);
-+}
++#include "accel.h"
+
-+static inline struct sk_buff *xennet_get_rx_skb(struct netfront_info *np,
-+ RING_IDX ri)
-+{
-+ int i = xennet_rxidx(ri);
-+ struct sk_buff *skb = np->rx_skbs[i];
-+ np->rx_skbs[i] = NULL;
-+ return skb;
-+}
+
-+static inline grant_ref_t xennet_get_rx_ref(struct netfront_info *np,
-+ RING_IDX ri)
++static int
++netfront_accel_alloc_buf_desc_blocks(struct netfront_accel_bufinfo *manager,
++ int pages)
+{
-+ int i = xennet_rxidx(ri);
-+ grant_ref_t ref = np->grant_rx_ref[i];
-+ np->grant_rx_ref[i] = GRANT_INVALID_REF;
-+ return ref;
++ manager->desc_blocks =
++ kzalloc(sizeof(struct netfront_accel_pkt_desc *) *
++ NETFRONT_ACCEL_BUF_NUM_BLOCKS(pages), GFP_KERNEL);
++ if (manager->desc_blocks == NULL) {
++ return -ENOMEM;
++ }
++
++ return 0;
+}
+
-+#define DPRINTK(fmt, args...) \
-+ pr_debug("netfront (%s:%d) " fmt, \
-+ __FUNCTION__, __LINE__, ##args)
-+#define IPRINTK(fmt, args...) \
-+ printk(KERN_INFO "netfront: " fmt, ##args)
-+#define WPRINTK(fmt, args...) \
-+ printk(KERN_WARNING "netfront: " fmt, ##args)
-+
-+static int setup_device(struct xenbus_device *, struct netfront_info *);
-+static struct net_device *create_netdev(struct xenbus_device *);
-+
-+static void end_access(int, void *);
-+static void netif_disconnect_backend(struct netfront_info *);
++static int
++netfront_accel_alloc_buf_lists(struct netfront_accel_bufpages *bufpages,
++ int pages)
++{
++ bufpages->page_list = kmalloc(pages * sizeof(void *), GFP_KERNEL);
++ if (bufpages->page_list == NULL) {
++ return -ENOMEM;
++ }
+
-+static int network_connect(struct net_device *);
-+static void network_tx_buf_gc(struct net_device *);
-+static void network_alloc_rx_buffers(struct net_device *);
-+static void send_fake_arp(struct net_device *);
++ bufpages->grant_list = kzalloc(pages * sizeof(grant_ref_t), GFP_KERNEL);
++ if (bufpages->grant_list == NULL) {
++ kfree(bufpages->page_list);
++ bufpages->page_list = NULL;
++ return -ENOMEM;
++ }
+
-+static irqreturn_t netif_int(int irq, void *dev_id, struct pt_regs *ptregs);
++ return 0;
++}
+
-+#ifdef CONFIG_SYSFS
-+static int xennet_sysfs_addif(struct net_device *netdev);
-+static void xennet_sysfs_delif(struct net_device *netdev);
-+#else /* !CONFIG_SYSFS */
-+#define xennet_sysfs_addif(dev) (0)
-+#define xennet_sysfs_delif(dev) do { } while(0)
-+#endif
+
-+static inline int xennet_can_sg(struct net_device *dev)
++int netfront_accel_alloc_buffer_mem(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *rx_manager,
++ struct netfront_accel_bufinfo *tx_manager,
++ int pages)
+{
-+ return dev->features & NETIF_F_SG;
-+}
++ int n, rc;
+
-+/**
-+ * Entry point to this code when a new device is created. Allocate the basic
-+ * structures and the ring buffers for communication with the backend, and
-+ * inform the backend of the appropriate details for those.
-+ */
-+static int __devinit netfront_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
-+{
-+ int err;
-+ struct net_device *netdev;
-+ struct netfront_info *info;
++ if ((rc = netfront_accel_alloc_buf_desc_blocks
++ (rx_manager, pages - (pages / sfc_netfront_buffer_split))) < 0) {
++ goto rx_fail;
++ }
+
-+ netdev = create_netdev(dev);
-+ if (IS_ERR(netdev)) {
-+ err = PTR_ERR(netdev);
-+ xenbus_dev_fatal(dev, err, "creating netdev");
-+ return err;
++ if ((rc = netfront_accel_alloc_buf_desc_blocks
++ (tx_manager, pages / sfc_netfront_buffer_split)) < 0) {
++ goto tx_fail;
+ }
+
-+ info = netdev_priv(netdev);
-+ dev->dev.driver_data = info;
++ if ((rc = netfront_accel_alloc_buf_lists(bufpages, pages)) < 0) {
++ goto lists_fail;
++ }
+
-+ err = register_netdev(info->netdev);
-+ if (err) {
-+ printk(KERN_WARNING "%s: register_netdev err=%d\n",
-+ __FUNCTION__, err);
-+ goto fail;
++ for (n = 0; n < pages; n++) {
++ void *tmp = (void*)__get_free_page(GFP_KERNEL);
++ if (tmp == NULL)
++ break;
++
++ bufpages->page_list[n] = tmp;
+ }
+
-+ err = xennet_sysfs_addif(info->netdev);
-+ if (err) {
-+ unregister_netdev(info->netdev);
-+ printk(KERN_WARNING "%s: add sysfs failed err=%d\n",
-+ __FUNCTION__, err);
-+ goto fail;
++ if (n != pages) {
++ EPRINTK("%s: not enough pages: %d != %d\n", __FUNCTION__, n,
++ pages);
++ for (; n >= 0; n--)
++ free_page((unsigned long)(bufpages->page_list[n]));
++ rc = -ENOMEM;
++ goto pages_fail;
+ }
+
++ bufpages->max_pages = pages;
++ bufpages->page_reqs = 0;
++
+ return 0;
+
-+ fail:
-+ free_netdev(netdev);
-+ dev->dev.driver_data = NULL;
-+ return err;
-+}
++ pages_fail:
++ kfree(bufpages->page_list);
++ kfree(bufpages->grant_list);
+
-+static int __devexit netfront_remove(struct xenbus_device *dev)
-+{
-+ struct netfront_info *info = dev->dev.driver_data;
++ bufpages->page_list = NULL;
++ bufpages->grant_list = NULL;
++ lists_fail:
++ kfree(tx_manager->desc_blocks);
++ tx_manager->desc_blocks = NULL;
+
-+ DPRINTK("%s\n", dev->nodename);
++ tx_fail:
++ kfree(rx_manager->desc_blocks);
++ rx_manager->desc_blocks = NULL;
++ rx_fail:
++ return rc;
++}
+
-+ netfront_accelerator_call_remove(info, dev);
+
-+ netif_disconnect_backend(info);
++void netfront_accel_free_buffer_mem(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *rx_manager,
++ struct netfront_accel_bufinfo *tx_manager)
++{
++ int i;
+
-+ del_timer_sync(&info->rx_refill_timer);
++ for (i = 0; i < bufpages->max_pages; i++) {
++ if (bufpages->grant_list[i] != 0)
++ net_accel_ungrant_page(bufpages->grant_list[i]);
++ free_page((unsigned long)(bufpages->page_list[i]));
++ }
+
-+ xennet_sysfs_delif(info->netdev);
++ if (bufpages->max_pages) {
++ kfree(bufpages->page_list);
++ kfree(bufpages->grant_list);
++ kfree(rx_manager->desc_blocks);
++ kfree(tx_manager->desc_blocks);
++ }
++}
+
-+ unregister_netdev(info->netdev);
+
-+ free_netdev(info->netdev);
++/*
++ * Allocate memory for the buffer manager and create a lock. If no
++ * lock is supplied its own is allocated.
++ */
++struct netfront_accel_bufinfo *netfront_accel_init_bufs(spinlock_t *lock)
++{
++ struct netfront_accel_bufinfo *res = kmalloc(sizeof(*res), GFP_KERNEL);
++ if (res != NULL) {
++ res->npages = res->nused = 0;
++ res->first_free = -1;
+
-+ return 0;
++ if (lock == NULL) {
++ res->lock = kmalloc(sizeof(*res->lock), GFP_KERNEL);
++ if (res->lock == NULL) {
++ kfree(res);
++ return NULL;
++ }
++ spin_lock_init(res->lock);
++ res->internally_locked = 1;
++ } else {
++ res->lock = lock;
++ res->internally_locked = 0;
++ }
++
++ res->desc_blocks = NULL;
++ }
++
++ return res;
+}
+
+
-+static int netfront_suspend(struct xenbus_device *dev)
++void netfront_accel_fini_bufs(struct netfront_accel_bufinfo *bufs)
+{
-+ struct netfront_info *info = dev->dev.driver_data;
-+ return netfront_accelerator_suspend(info, dev);
++ if (bufs->internally_locked)
++ kfree(bufs->lock);
++ kfree(bufs);
+}
+
+
-+static int netfront_suspend_cancel(struct xenbus_device *dev)
++int netfront_accel_buf_map_request(struct xenbus_device *dev,
++ struct netfront_accel_bufpages *bufpages,
++ struct net_accel_msg *msg,
++ int pages, int offset)
+{
-+ struct netfront_info *info = dev->dev.driver_data;
-+ return netfront_accelerator_suspend_cancel(info, dev);
-+}
++ int i, mfn;
++ int err;
+
++ net_accel_msg_init(msg, NET_ACCEL_MSG_MAPBUF);
+
-+/**
-+ * We are reconnecting to the backend, due to a suspend/resume, or a backend
-+ * driver restart. We tear down our netif structure and recreate it, but
-+ * leave the device-layer structures intact so that this is transparent to the
-+ * rest of the kernel.
-+ */
-+static int netfront_resume(struct xenbus_device *dev)
-+{
-+ struct netfront_info *info = dev->dev.driver_data;
++ BUG_ON(pages > NET_ACCEL_MSG_MAX_PAGE_REQ);
+
-+ DPRINTK("%s\n", dev->nodename);
++ msg->u.mapbufs.pages = pages;
+
-+ netfront_accelerator_resume(info, dev);
++ for (i = 0; i < msg->u.mapbufs.pages; i++) {
++ /*
++ * This can happen if we tried to send this message
++ * earlier but the queue was full.
++ */
++ if (bufpages->grant_list[offset+i] != 0) {
++ msg->u.mapbufs.grants[i] =
++ bufpages->grant_list[offset+i];
++ continue;
++ }
++
++ mfn = virt_to_mfn(bufpages->page_list[offset+i]);
++ VPRINTK("%s: Granting page %d, mfn %08x\n",
++ __FUNCTION__, i, mfn);
++
++ bufpages->grant_list[offset+i] =
++ net_accel_grant_page(dev, mfn, 0);
++ msg->u.mapbufs.grants[i] = bufpages->grant_list[offset+i];
++
++ if (msg->u.mapbufs.grants[i] < 0) {
++ EPRINTK("%s: Failed to grant buffer: %d\n",
++ __FUNCTION__, msg->u.mapbufs.grants[i]);
++ err = -EIO;
++ goto error;
++ }
++ }
++
++ /* This is interpreted on return as the offset in the the page_list */
++ msg->u.mapbufs.reqid = offset;
+
-+ netif_disconnect_backend(info);
+ return 0;
++
++error:
++ /* Ungrant all the pages we've successfully granted. */
++ for (i--; i >= 0; i--) {
++ net_accel_ungrant_page(bufpages->grant_list[offset+i]);
++ bufpages->grant_list[offset+i] = 0;
++ }
++ return err;
+}
+
-+static int xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
++
++/* Process a response to a buffer request. */
++int netfront_accel_add_bufs(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *manager,
++ struct net_accel_msg *msg)
+{
-+ char *s, *e, *macstr;
-+ int i;
++ int msg_pages, page_offset, i, newtot;
++ int old_block_count, new_block_count;
++ u32 msg_buf;
++ unsigned long flags;
+
-+ macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
-+ if (IS_ERR(macstr))
-+ return PTR_ERR(macstr);
++ VPRINTK("%s: manager %p msg %p\n", __FUNCTION__, manager, msg);
+
-+ for (i = 0; i < ETH_ALEN; i++) {
-+ mac[i] = simple_strtoul(s, &e, 16);
-+ if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
-+ kfree(macstr);
-+ return -ENOENT;
++ BUG_ON(msg->id != (NET_ACCEL_MSG_MAPBUF | NET_ACCEL_MSG_REPLY));
++
++ msg_pages = msg->u.mapbufs.pages;
++ msg_buf = msg->u.mapbufs.buf;
++ page_offset = msg->u.mapbufs.reqid;
++
++ spin_lock_irqsave(manager->lock, flags);
++ newtot = manager->npages + msg_pages;
++ old_block_count =
++ (manager->npages + NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK - 1) >>
++ NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT;
++ new_block_count =
++ (newtot + NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK - 1) >>
++ NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT;
++
++ for (i = old_block_count; i < new_block_count; i++) {
++ struct netfront_accel_pkt_desc *block;
++ if (manager->desc_blocks[i] != NULL) {
++ VPRINTK("Not needed\n");
++ continue;
+ }
-+ s = e+1;
++ block = kzalloc(NETFRONT_ACCEL_BUFS_PER_BLOCK *
++ sizeof(netfront_accel_pkt_desc), GFP_ATOMIC);
++ if (block == NULL) {
++ spin_unlock_irqrestore(manager->lock, flags);
++ return -ENOMEM;
++ }
++ manager->desc_blocks[i] = block;
+ }
++ for (i = manager->npages; i < newtot; i++) {
++ int k, j = i - manager->npages;
++ int block_num;
++ int block_idx;
++ struct netfront_accel_pkt_desc *pkt;
+
-+ kfree(macstr);
++ block_num = i >> NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT;
++ block_idx = (NETFRONT_ACCEL_BUFS_PER_PAGE*i)
++ & (NETFRONT_ACCEL_BUFS_PER_BLOCK-1);
++
++ pkt = manager->desc_blocks[block_num] + block_idx;
++
++ for (k = 0; k < NETFRONT_ACCEL_BUFS_PER_PAGE; k++) {
++ BUG_ON(page_offset + j >= bufpages->max_pages);
++
++ pkt[k].buf_id = NETFRONT_ACCEL_BUFS_PER_PAGE * i + k;
++ pkt[k].pkt_kva = bufpages->page_list[page_offset + j] +
++ (PAGE_SIZE/NETFRONT_ACCEL_BUFS_PER_PAGE) * k;
++ pkt[k].pkt_buff_addr = msg_buf +
++ (PAGE_SIZE/NETFRONT_ACCEL_BUFS_PER_PAGE) *
++ (NETFRONT_ACCEL_BUFS_PER_PAGE * j + k);
++ pkt[k].next_free = manager->first_free;
++ manager->first_free = pkt[k].buf_id;
++ *(int*)(pkt[k].pkt_kva) = pkt[k].buf_id;
++
++ VPRINTK("buf %d desc %p kva %p buffaddr %x\n",
++ pkt[k].buf_id, &(pkt[k]), pkt[k].pkt_kva,
++ pkt[k].pkt_buff_addr);
++ }
++ }
++ manager->npages = newtot;
++ spin_unlock_irqrestore(manager->lock, flags);
++ VPRINTK("Added %d pages. Total is now %d\n", msg_pages,
++ manager->npages);
+ return 0;
+}
+
-+/* Common code used when first setting up, and when resuming. */
-+static int talk_to_backend(struct xenbus_device *dev,
-+ struct netfront_info *info)
-+{
-+ const char *message;
-+ struct xenbus_transaction xbt;
-+ int err;
+
-+ err = xen_net_read_mac(dev, info->mac);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "parsing %s/mac", dev->nodename);
-+ goto out;
-+ }
++netfront_accel_pkt_desc *
++netfront_accel_buf_find(struct netfront_accel_bufinfo *manager, u16 id)
++{
++ netfront_accel_pkt_desc *pkt;
++ int block_num = id >> NETFRONT_ACCEL_BUFS_PER_BLOCK_SHIFT;
++ int block_idx = id & (NETFRONT_ACCEL_BUFS_PER_BLOCK - 1);
++ BUG_ON(id >= manager->npages * NETFRONT_ACCEL_BUFS_PER_PAGE);
++ BUG_ON(block_idx >= NETFRONT_ACCEL_BUFS_PER_BLOCK);
++ pkt = manager->desc_blocks[block_num] + block_idx;
++ return pkt;
++}
+
-+ /* Create shared ring, alloc event channel. */
-+ err = setup_device(dev, info);
-+ if (err)
-+ goto out;
+
-+ /* This will load an accelerator if one is configured when the
-+ * watch fires */
-+ netfront_accelerator_add_watch(info);
++/* Allocate a buffer from the buffer manager */
++netfront_accel_pkt_desc *
++netfront_accel_buf_get(struct netfront_accel_bufinfo *manager)
++{
++ int bufno = -1;
++ netfront_accel_pkt_desc *buf = NULL;
++ unsigned long flags = 0;
+
-+again:
-+ err = xenbus_transaction_start(&xbt);
-+ if (err) {
-+ xenbus_dev_fatal(dev, err, "starting transaction");
-+ goto destroy_ring;
-+ }
++ /* Any spare? */
++ if (manager->first_free == -1)
++ return NULL;
++ /* Take lock */
++ if (manager->internally_locked)
++ spin_lock_irqsave(manager->lock, flags);
++ bufno = manager->first_free;
++ if (bufno != -1) {
++ buf = netfront_accel_buf_find(manager, bufno);
++ manager->first_free = buf->next_free;
++ manager->nused++;
++ }
++ /* Release lock */
++ if (manager->internally_locked)
++ spin_unlock_irqrestore(manager->lock, flags);
++
++ /* Tell the world */
++ VPRINTK("Allocated buffer %i, buffaddr %x\n", bufno,
++ buf->pkt_buff_addr);
+
-+ err = xenbus_printf(xbt, dev->nodename, "tx-ring-ref","%u",
-+ info->tx_ring_ref);
-+ if (err) {
-+ message = "writing tx ring-ref";
-+ goto abort_transaction;
-+ }
-+ err = xenbus_printf(xbt, dev->nodename, "rx-ring-ref","%u",
-+ info->rx_ring_ref);
-+ if (err) {
-+ message = "writing rx ring-ref";
-+ goto abort_transaction;
-+ }
-+ err = xenbus_printf(xbt, dev->nodename,
-+ "event-channel", "%u",
-+ irq_to_evtchn_port(info->irq));
-+ if (err) {
-+ message = "writing event-channel";
-+ goto abort_transaction;
-+ }
++ return buf;
++}
+
-+ err = xenbus_printf(xbt, dev->nodename, "request-rx-copy", "%u",
-+ info->copying_receiver);
-+ if (err) {
-+ message = "writing request-rx-copy";
-+ goto abort_transaction;
-+ }
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-rx-notify", "%d", 1);
-+ if (err) {
-+ message = "writing feature-rx-notify";
-+ goto abort_transaction;
-+ }
++/* Release a buffer back to the buffer manager pool */
++int netfront_accel_buf_put(struct netfront_accel_bufinfo *manager, u16 id)
++{
++ netfront_accel_pkt_desc *buf = netfront_accel_buf_find(manager, id);
++ unsigned long flags = 0;
++ unsigned was_empty = 0;
++ int bufno = id;
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-no-csum-offload",
-+ "%d", !HAVE_CSUM_OFFLOAD);
-+ if (err) {
-+ message = "writing feature-no-csum-offload";
-+ goto abort_transaction;
-+ }
++ VPRINTK("Freeing buffer %i\n", id);
++ BUG_ON(id == (u16)-1);
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-sg", "%d", 1);
-+ if (err) {
-+ message = "writing feature-sg";
-+ goto abort_transaction;
-+ }
++ if (manager->internally_locked)
++ spin_lock_irqsave(manager->lock, flags);
+
-+ err = xenbus_printf(xbt, dev->nodename, "feature-gso-tcpv4", "%d",
-+ HAVE_TSO);
-+ if (err) {
-+ message = "writing feature-gso-tcpv4";
-+ goto abort_transaction;
-+ }
++ if (manager->first_free == -1)
++ was_empty = 1;
+
-+ err = xenbus_transaction_end(xbt, 0);
-+ if (err) {
-+ if (err == -EAGAIN)
-+ goto again;
-+ xenbus_dev_fatal(dev, err, "completing transaction");
-+ goto destroy_ring;
-+ }
++ buf->next_free = manager->first_free;
++ manager->first_free = bufno;
++ manager->nused--;
+
-+ return 0;
++ if (manager->internally_locked)
++ spin_unlock_irqrestore(manager->lock, flags);
+
-+ abort_transaction:
-+ xenbus_transaction_end(xbt, 1);
-+ xenbus_dev_fatal(dev, err, "%s", message);
-+ destroy_ring:
-+ netfront_accelerator_call_remove(info, dev);
-+ netif_disconnect_backend(info);
-+ out:
-+ return err;
++ return was_empty;
+}
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_bufs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_bufs.h 2008-05-19 00:33:48.618948572 +0300
+@@ -0,0 +1,181 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static int setup_device(struct xenbus_device *dev, struct netfront_info *info)
-+{
-+ struct netif_tx_sring *txs;
-+ struct netif_rx_sring *rxs;
-+ int err;
-+ struct net_device *netdev = info->netdev;
++#ifndef NETFRONT_ACCEL_BUFS_H
++#define NETFRONT_ACCEL_BUFS_H
+
-+ info->tx_ring_ref = GRANT_INVALID_REF;
-+ info->rx_ring_ref = GRANT_INVALID_REF;
-+ info->rx.sring = NULL;
-+ info->tx.sring = NULL;
-+ info->irq = 0;
++#include <linux/skbuff.h>
++#include <linux/spinlock.h>
++#include <xen/xenbus.h>
+
-+ txs = (struct netif_tx_sring *)get_zeroed_page(GFP_KERNEL);
-+ if (!txs) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(dev, err, "allocating tx ring page");
-+ goto fail;
-+ }
-+ SHARED_RING_INIT(txs);
-+ FRONT_RING_INIT(&info->tx, txs, PAGE_SIZE);
++#include "accel_msg_iface.h"
+
-+ err = xenbus_grant_ring(dev, virt_to_mfn(txs));
-+ if (err < 0) {
-+ free_page((unsigned long)txs);
-+ goto fail;
-+ }
-+ info->tx_ring_ref = err;
+
-+ rxs = (struct netif_rx_sring *)get_zeroed_page(GFP_KERNEL);
-+ if (!rxs) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(dev, err, "allocating rx ring page");
-+ goto fail;
-+ }
-+ SHARED_RING_INIT(rxs);
-+ FRONT_RING_INIT(&info->rx, rxs, PAGE_SIZE);
++/*! Buffer descriptor structure */
++typedef struct netfront_accel_pkt_desc {
++ int buf_id;
++ u32 pkt_buff_addr;
++ void *pkt_kva;
++ /* This is the socket buffer currently married to this buffer */
++ struct sk_buff *skb;
++ int next_free;
++} netfront_accel_pkt_desc;
++
++
++#define NETFRONT_ACCEL_DEFAULT_BUF_PAGES (384)
++#define NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT (4)
++#define NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK \
++ (1 << (NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT))
++#define NETFRONT_ACCEL_BUFS_PER_PAGE_SHIFT (1)
++#define NETFRONT_ACCEL_BUFS_PER_PAGE \
++ (1 << (NETFRONT_ACCEL_BUFS_PER_PAGE_SHIFT))
++#define NETFRONT_ACCEL_BUFS_PER_BLOCK_SHIFT \
++ (NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK_SHIFT + \
++ NETFRONT_ACCEL_BUFS_PER_PAGE_SHIFT)
++#define NETFRONT_ACCEL_BUFS_PER_BLOCK \
++ (1 << NETFRONT_ACCEL_BUFS_PER_BLOCK_SHIFT)
++#define NETFRONT_ACCEL_BUF_NUM_BLOCKS(max_pages) \
++ (((max_pages)+NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK-1) / \
++ NETFRONT_ACCEL_BUF_PAGES_PER_BLOCK)
++
++/*! Buffer management structure. */
++struct netfront_accel_bufinfo {
++ /* number added to this manager */
++ unsigned npages;
++ /* number currently used from this manager */
++ unsigned nused;
++
++ int first_free;
++
++ int internally_locked;
++ spinlock_t *lock;
+
-+ err = xenbus_grant_ring(dev, virt_to_mfn(rxs));
-+ if (err < 0) {
-+ free_page((unsigned long)rxs);
-+ goto fail;
-+ }
-+ info->rx_ring_ref = err;
++ /*
++ * array of pointers (length NETFRONT_ACCEL_BUF_NUM_BLOCKS) to
++ * pkt descs
++ */
++ struct netfront_accel_pkt_desc **desc_blocks;
++};
+
-+ memcpy(netdev->dev_addr, info->mac, ETH_ALEN);
+
-+ err = bind_listening_port_to_irqhandler(
-+ dev->otherend_id, netif_int, SA_SAMPLE_RANDOM, netdev->name,
-+ netdev);
-+ if (err < 0)
-+ goto fail;
-+ info->irq = err;
++struct netfront_accel_bufpages {
++ /* length of lists of pages/grants */
++ int max_pages;
++ /* list of pages allocated for network buffers */
++ void **page_list;
++ /* list of grants for the above pages */
++ grant_ref_t *grant_list;
++
++ /* number of page requests that have been made */
++ unsigned page_reqs;
++};
+
-+ return 0;
+
-+ fail:
-+ return err;
-+}
++/*! Allocate memory for the buffer manager, set up locks etc.
++ * Optionally takes a lock to use, if not supplied it makes its own.
++ *
++ * \return pointer to netfront_accel_bufinfo structure that represents the
++ * buffer manager
++ */
++extern struct netfront_accel_bufinfo *
++netfront_accel_init_bufs(spinlock_t *lock);
+
-+/**
-+ * Callback received when the backend's state changes.
++/*! Allocate memory for the buffers
+ */
-+static void backend_changed(struct xenbus_device *dev,
-+ enum xenbus_state backend_state)
-+{
-+ struct netfront_info *np = dev->dev.driver_data;
-+ struct net_device *netdev = np->netdev;
++extern int
++netfront_accel_alloc_buffer_mem(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *rx_res,
++ struct netfront_accel_bufinfo *tx_res,
++ int pages);
++extern void
++netfront_accel_free_buffer_mem(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *rx_res,
++ struct netfront_accel_bufinfo *tx_res);
+
-+ DPRINTK("%s\n", xenbus_strstate(backend_state));
++/*! Release memory for the buffer manager, buffers, etc.
++ *
++ * \param manager pointer to netfront_accel_bufinfo structure that
++ * represents the buffer manager
++ */
++extern void netfront_accel_fini_bufs(struct netfront_accel_bufinfo *manager);
+
-+ switch (backend_state) {
-+ case XenbusStateInitialising:
-+ case XenbusStateInitialised:
-+ case XenbusStateConnected:
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ break;
++/*! Release a buffer.
++ *
++ * \param manager The buffer manager which owns the buffer.
++ * \param id The buffer identifier.
++ */
++extern int netfront_accel_buf_put(struct netfront_accel_bufinfo *manager,
++ u16 id);
+
-+ case XenbusStateInitWait:
-+ if (dev->state != XenbusStateInitialising)
-+ break;
-+ if (network_connect(netdev) != 0)
-+ break;
-+ xenbus_switch_state(dev, XenbusStateConnected);
-+ send_fake_arp(netdev);
-+ break;
++/*! Get the packet descriptor associated with a buffer id.
++ *
++ * \param manager The buffer manager which owns the buffer.
++ * \param id The buffer identifier.
++ *
++ * The returned value is the packet descriptor for this buffer.
++ */
++extern netfront_accel_pkt_desc *
++netfront_accel_buf_find(struct netfront_accel_bufinfo *manager, u16 id);
+
-+ case XenbusStateClosing:
-+ xenbus_frontend_closed(dev);
-+ break;
-+ }
-+}
+
-+/** Send a packet on a net device to encourage switches to learn the
-+ * MAC. We send a fake ARP request.
++/*! Fill out a message request for some buffers to be mapped by the
++ * back end driver
++ *
++ * \param manager The buffer manager
++ * \param msg Pointer to an ef_msg to complete.
++ * \return 0 on success
++ */
++extern int
++netfront_accel_buf_map_request(struct xenbus_device *dev,
++ struct netfront_accel_bufpages *bufpages,
++ struct net_accel_msg *msg,
++ int pages, int offset);
++
++/*! Process a response to a buffer request.
++ *
++ * Deal with a received message from the back end in response to our
++ * request for buffers
++ *
++ * \param manager The buffer manager
++ * \param msg The received message from the back end describing new
++ * buffers
++ * \return 0 on success
++ */
++extern int
++netfront_accel_add_bufs(struct netfront_accel_bufpages *bufpages,
++ struct netfront_accel_bufinfo *manager,
++ struct net_accel_msg *msg);
++
++
++/*! Allocate a buffer from the buffer manager
+ *
-+ * @param dev device
-+ * @return 0 on success, error code otherwise
++ * \param manager The buffer manager data structure
++ * \param id On exit, the id of the buffer allocated
++ * \return Pointer to buffer descriptor.
+ */
-+static void send_fake_arp(struct net_device *dev)
-+{
-+#ifdef CONFIG_INET
-+ struct sk_buff *skb;
-+ u32 src_ip, dst_ip;
++struct netfront_accel_pkt_desc *
++netfront_accel_buf_get(struct netfront_accel_bufinfo *manager);
+
-+ dst_ip = INADDR_BROADCAST;
-+ src_ip = inet_select_addr(dev, dst_ip, RT_SCOPE_LINK);
++#endif /* NETFRONT_ACCEL_BUFS_H */
+
-+ /* No IP? Then nothing to do. */
-+ if (src_ip == 0)
-+ return;
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_debugfs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_debugfs.c 2008-05-19 00:33:48.618948572 +0300
+@@ -0,0 +1,211 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ skb = arp_create(ARPOP_REPLY, ETH_P_ARP,
-+ dst_ip, dev, src_ip,
-+ /*dst_hw*/ NULL, /*src_hw*/ NULL,
-+ /*target_hw*/ dev->dev_addr);
-+ if (skb == NULL)
-+ return;
++#include <linux/fs.h>
++#include <linux/debugfs.h>
++
++#include "accel.h"
++
++#if defined(CONFIG_DEBUG_FS)
++static struct dentry *sfc_debugfs_root = NULL;
++#endif
+
-+ dev_queue_xmit(skb);
++void netfront_accel_debugfs_init(void)
++{
++#if defined(CONFIG_DEBUG_FS)
++ sfc_debugfs_root = debugfs_create_dir(frontend_name, NULL);
+#endif
+}
+
-+static inline int netfront_tx_slot_available(struct netfront_info *np)
++
++void netfront_accel_debugfs_fini(void)
+{
-+ return ((np->tx.req_prod_pvt - np->tx.rsp_cons) <
-+ (TX_MAX_TARGET - MAX_SKB_FRAGS - 2));
++#if defined(CONFIG_DEBUG_FS)
++ if (sfc_debugfs_root)
++ debugfs_remove(sfc_debugfs_root);
++#endif
+}
+
+
-+static inline void network_maybe_wake_tx(struct net_device *dev)
++int netfront_accel_debugfs_create(netfront_accel_vnic *vnic)
+{
-+ struct netfront_info *np = netdev_priv(dev);
++#if defined(CONFIG_DEBUG_FS)
++ if (sfc_debugfs_root == NULL)
++ return -ENOENT;
+
-+ if (unlikely(netif_queue_stopped(dev)) &&
-+ netfront_tx_slot_available(np) &&
-+ likely(netif_running(dev)) &&
-+ netfront_check_accelerator_queue_ready(dev, np))
-+ netif_wake_queue(dev);
++ vnic->dbfs_dir = debugfs_create_dir(vnic->net_dev->name,
++ sfc_debugfs_root);
++ if (vnic->dbfs_dir == NULL)
++ return -ENOMEM;
++
++ vnic->netdev_dbfs.fastpath_rx_pkts = debugfs_create_u32
++ ("fastpath_rx_pkts", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_rx_pkts);
++ vnic->netdev_dbfs.fastpath_rx_bytes = debugfs_create_u32
++ ("fastpath_rx_bytes", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_rx_bytes);
++ vnic->netdev_dbfs.fastpath_rx_errors = debugfs_create_u32
++ ("fastpath_rx_errors", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_rx_errors);
++ vnic->netdev_dbfs.fastpath_tx_pkts = debugfs_create_u32
++ ("fastpath_tx_pkts", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_tx_pkts);
++ vnic->netdev_dbfs.fastpath_tx_bytes = debugfs_create_u32
++ ("fastpath_tx_bytes", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_tx_bytes);
++ vnic->netdev_dbfs.fastpath_tx_errors = debugfs_create_u32
++ ("fastpath_tx_errors", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->netdev_stats.fastpath_tx_errors);
++
++#if NETFRONT_ACCEL_STATS
++ vnic->dbfs.irq_count = debugfs_create_u64
++ ("irq_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.irq_count);
++ vnic->dbfs.useless_irq_count = debugfs_create_u64
++ ("useless_irq_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.useless_irq_count);
++ vnic->dbfs.poll_schedule_count = debugfs_create_u64
++ ("poll_schedule_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.poll_schedule_count);
++ vnic->dbfs.poll_call_count = debugfs_create_u64
++ ("poll_call_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.poll_call_count);
++ vnic->dbfs.poll_reschedule_count = debugfs_create_u64
++ ("poll_reschedule_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.poll_reschedule_count);
++ vnic->dbfs.queue_stops = debugfs_create_u64
++ ("queue_stops", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.queue_stops);
++ vnic->dbfs.queue_wakes = debugfs_create_u64
++ ("queue_wakes", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.queue_wakes);
++ vnic->dbfs.ssr_bursts = debugfs_create_u64
++ ("ssr_bursts", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_bursts);
++ vnic->dbfs.ssr_drop_stream = debugfs_create_u64
++ ("ssr_drop_stream", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_drop_stream);
++ vnic->dbfs.ssr_misorder = debugfs_create_u64
++ ("ssr_misorder", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_misorder);
++ vnic->dbfs.ssr_slow_start = debugfs_create_u64
++ ("ssr_slow_start", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_slow_start);
++ vnic->dbfs.ssr_merges = debugfs_create_u64
++ ("ssr_merges", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_merges);
++ vnic->dbfs.ssr_too_many = debugfs_create_u64
++ ("ssr_too_many", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_too_many);
++ vnic->dbfs.ssr_new_stream = debugfs_create_u64
++ ("ssr_new_stream", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.ssr_new_stream);
++
++ vnic->dbfs.fastpath_tx_busy = debugfs_create_u64
++ ("fastpath_tx_busy", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.fastpath_tx_busy);
++ vnic->dbfs.fastpath_tx_completions = debugfs_create_u64
++ ("fastpath_tx_completions", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.fastpath_tx_completions);
++ vnic->dbfs.fastpath_tx_pending_max = debugfs_create_u32
++ ("fastpath_tx_pending_max", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.fastpath_tx_pending_max);
++ vnic->dbfs.event_count = debugfs_create_u64
++ ("event_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.event_count);
++ vnic->dbfs.bad_event_count = debugfs_create_u64
++ ("bad_event_count", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.bad_event_count);
++ vnic->dbfs.event_count_since_irq = debugfs_create_u32
++ ("event_count_since_irq", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.event_count_since_irq);
++ vnic->dbfs.events_per_irq_max = debugfs_create_u32
++ ("events_per_irq_max", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.events_per_irq_max);
++ vnic->dbfs.fastpath_frm_trunc = debugfs_create_u64
++ ("fastpath_frm_trunc", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.fastpath_frm_trunc);
++ vnic->dbfs.rx_no_desc_trunc = debugfs_create_u64
++ ("rx_no_desc_trunc", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.rx_no_desc_trunc);
++ vnic->dbfs.events_per_poll_max = debugfs_create_u32
++ ("events_per_poll_max", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.events_per_poll_max);
++ vnic->dbfs.events_per_poll_rx_max = debugfs_create_u32
++ ("events_per_poll_rx_max", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.events_per_poll_rx_max);
++ vnic->dbfs.events_per_poll_tx_max = debugfs_create_u32
++ ("events_per_poll_tx_max", S_IRUSR | S_IRGRP | S_IROTH,
++ vnic->dbfs_dir, &vnic->stats.events_per_poll_tx_max);
++#endif
++#endif
++ return 0;
+}
+
+
-+int netfront_check_queue_ready(struct net_device *dev)
++int netfront_accel_debugfs_remove(netfront_accel_vnic *vnic)
+{
-+ struct netfront_info *np = netdev_priv(dev);
-+
-+ return unlikely(netif_queue_stopped(dev)) &&
-+ netfront_tx_slot_available(np) &&
-+ likely(netif_running(dev));
++#if defined(CONFIG_DEBUG_FS)
++ if (vnic->dbfs_dir != NULL) {
++ debugfs_remove(vnic->netdev_dbfs.fastpath_rx_pkts);
++ debugfs_remove(vnic->netdev_dbfs.fastpath_rx_bytes);
++ debugfs_remove(vnic->netdev_dbfs.fastpath_rx_errors);
++ debugfs_remove(vnic->netdev_dbfs.fastpath_tx_pkts);
++ debugfs_remove(vnic->netdev_dbfs.fastpath_tx_bytes);
++ debugfs_remove(vnic->netdev_dbfs.fastpath_tx_errors);
++
++#if NETFRONT_ACCEL_STATS
++ debugfs_remove(vnic->dbfs.irq_count);
++ debugfs_remove(vnic->dbfs.useless_irq_count);
++ debugfs_remove(vnic->dbfs.poll_schedule_count);
++ debugfs_remove(vnic->dbfs.poll_call_count);
++ debugfs_remove(vnic->dbfs.poll_reschedule_count);
++ debugfs_remove(vnic->dbfs.queue_stops);
++ debugfs_remove(vnic->dbfs.queue_wakes);
++ debugfs_remove(vnic->dbfs.ssr_bursts);
++ debugfs_remove(vnic->dbfs.ssr_drop_stream);
++ debugfs_remove(vnic->dbfs.ssr_misorder);
++ debugfs_remove(vnic->dbfs.ssr_slow_start);
++ debugfs_remove(vnic->dbfs.ssr_merges);
++ debugfs_remove(vnic->dbfs.ssr_too_many);
++ debugfs_remove(vnic->dbfs.ssr_new_stream);
++
++ debugfs_remove(vnic->dbfs.fastpath_tx_busy);
++ debugfs_remove(vnic->dbfs.fastpath_tx_completions);
++ debugfs_remove(vnic->dbfs.fastpath_tx_pending_max);
++ debugfs_remove(vnic->dbfs.event_count);
++ debugfs_remove(vnic->dbfs.bad_event_count);
++ debugfs_remove(vnic->dbfs.event_count_since_irq);
++ debugfs_remove(vnic->dbfs.events_per_irq_max);
++ debugfs_remove(vnic->dbfs.fastpath_frm_trunc);
++ debugfs_remove(vnic->dbfs.rx_no_desc_trunc);
++ debugfs_remove(vnic->dbfs.events_per_poll_max);
++ debugfs_remove(vnic->dbfs.events_per_poll_rx_max);
++ debugfs_remove(vnic->dbfs.events_per_poll_tx_max);
++#endif
++ debugfs_remove(vnic->dbfs_dir);
++ }
++#endif
++ return 0;
+}
-+EXPORT_SYMBOL(netfront_check_queue_ready);
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_msg.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_msg.c 2008-05-19 00:33:48.618948572 +0300
+@@ -0,0 +1,566 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
++#include <linux/stddef.h>
++#include <linux/errno.h>
+
-+static int network_open(struct net_device *dev)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
++#include <xen/xenbus.h>
+
-+ memset(&np->stats, 0, sizeof(np->stats));
++#include "accel.h"
++#include "accel_msg_iface.h"
++#include "accel_util.h"
++#include "accel_bufs.h"
+
-+ spin_lock_bh(&np->rx_lock);
-+ if (netfront_carrier_ok(np)) {
-+ network_alloc_rx_buffers(dev);
-+ np->rx.sring->rsp_event = np->rx.rsp_cons + 1;
-+ if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)){
-+ netfront_accelerator_call_stop_napi_irq(np, dev);
++#include "netfront.h" /* drivers/xen/netfront/netfront.h */
+
-+ netif_rx_schedule(dev);
-+ }
++static void vnic_start_interrupts(netfront_accel_vnic *vnic)
++{
++ unsigned long flags;
++
++ /* Prime our interrupt */
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
++ if (!netfront_accel_vi_enable_interrupts(vnic)) {
++ /* Cripes, that was quick, better pass it up */
++ netfront_accel_disable_net_interrupts(vnic);
++ vnic->irq_enabled = 0;
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.poll_schedule_count++);
++ netif_rx_schedule(vnic->net_dev);
++ } else {
++ /*
++ * Nothing yet, make sure we get interrupts through
++ * back end
++ */
++ vnic->irq_enabled = 1;
++ netfront_accel_enable_net_interrupts(vnic);
+ }
-+ spin_unlock_bh(&np->rx_lock);
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
++}
+
-+ network_maybe_wake_tx(dev);
+
-+ return 0;
++static void vnic_stop_interrupts(netfront_accel_vnic *vnic)
++{
++ unsigned long flags;
++
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
++ netfront_accel_disable_net_interrupts(vnic);
++ vnic->irq_enabled = 0;
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
+}
+
-+static void network_tx_buf_gc(struct net_device *dev)
++
++static void vnic_start_fastpath(netfront_accel_vnic *vnic)
+{
-+ RING_IDX cons, prod;
-+ unsigned short id;
-+ struct netfront_info *np = netdev_priv(dev);
-+ struct sk_buff *skb;
++ struct net_device *net_dev = vnic->net_dev;
++ unsigned long flags;
+
-+ BUG_ON(!netfront_carrier_ok(np));
++ DPRINTK("%s\n", __FUNCTION__);
+
-+ do {
-+ prod = np->tx.sring->rsp_prod;
-+ rmb(); /* Ensure we see responses up to 'rp'. */
++ spin_lock_irqsave(&vnic->tx_lock, flags);
++ vnic->tx_enabled = 1;
++ spin_unlock_irqrestore(&vnic->tx_lock, flags);
++
++ netif_poll_disable(net_dev);
++ vnic->poll_enabled = 1;
++ netif_poll_enable(net_dev);
++
++ vnic_start_interrupts(vnic);
++}
+
-+ for (cons = np->tx.rsp_cons; cons != prod; cons++) {
-+ struct netif_tx_response *txrsp;
+
-+ txrsp = RING_GET_RESPONSE(&np->tx, cons);
-+ if (txrsp->status == NETIF_RSP_NULL)
-+ continue;
++void vnic_stop_fastpath(netfront_accel_vnic *vnic)
++{
++ struct net_device *net_dev = vnic->net_dev;
++ struct netfront_info *np = (struct netfront_info *)netdev_priv(net_dev);
++ unsigned long flags1, flags2;
+
-+ id = txrsp->id;
-+ skb = np->tx_skbs[id];
-+ if (unlikely(gnttab_query_foreign_access(
-+ np->grant_tx_ref[id]) != 0)) {
-+ printk(KERN_ALERT "network_tx_buf_gc: warning "
-+ "-- grant still in use by backend "
-+ "domain.\n");
-+ BUG();
-+ }
-+ gnttab_end_foreign_access_ref(np->grant_tx_ref[id]);
-+ gnttab_release_grant_reference(
-+ &np->gref_tx_head, np->grant_tx_ref[id]);
-+ np->grant_tx_ref[id] = GRANT_INVALID_REF;
-+ add_id_to_freelist(np->tx_skbs, id);
-+ dev_kfree_skb_irq(skb);
-+ }
++ DPRINTK("%s\n", __FUNCTION__);
+
-+ np->tx.rsp_cons = prod;
++ vnic_stop_interrupts(vnic);
++
++ spin_lock_irqsave(&vnic->tx_lock, flags1);
++ vnic->tx_enabled = 0;
++ spin_lock_irqsave(&np->tx_lock, flags2);
++ if (vnic->tx_skb != NULL) {
++ dev_kfree_skb_any(vnic->tx_skb);
++ vnic->tx_skb = NULL;
++ if (netfront_check_queue_ready(net_dev)) {
++ netif_wake_queue(net_dev);
++ NETFRONT_ACCEL_STATS_OP
++ (vnic->stats.queue_wakes++);
++ }
++ }
++ spin_unlock_irqrestore(&np->tx_lock, flags2);
++ spin_unlock_irqrestore(&vnic->tx_lock, flags1);
++
++ /* Must prevent polls and hold lock to modify poll_enabled */
++ netif_poll_disable(net_dev);
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags1);
++ vnic->poll_enabled = 0;
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags1);
++ netif_poll_enable(net_dev);
++}
+
-+ /*
-+ * Set a new event, then check for race with update of tx_cons.
-+ * Note that it is essential to schedule a callback, no matter
-+ * how few buffers are pending. Even if there is space in the
-+ * transmit ring, higher layers may be blocked because too much
-+ * data is outstanding: in such cases notification from Xen is
-+ * likely to be the only kick that we'll get.
-+ */
-+ np->tx.sring->rsp_event =
-+ prod + ((np->tx.sring->req_prod - prod) >> 1) + 1;
-+ mb();
-+ } while ((cons == prod) && (prod != np->tx.sring->rsp_prod));
+
-+ network_maybe_wake_tx(dev);
++static void netfront_accel_interface_up(netfront_accel_vnic *vnic)
++{
++
++ if (!vnic->backend_netdev_up) {
++ vnic->backend_netdev_up = 1;
++
++ if (vnic->frontend_ready)
++ vnic_start_fastpath(vnic);
++ }
+}
+
-+static void rx_refill_timeout(unsigned long data)
-+{
-+ struct net_device *dev = (struct net_device *)data;
-+ struct netfront_info *np = netdev_priv(dev);
+
-+ netfront_accelerator_call_stop_napi_irq(np, dev);
++static void netfront_accel_interface_down(netfront_accel_vnic *vnic)
++{
+
-+ netif_rx_schedule(dev);
++ if (vnic->backend_netdev_up) {
++ vnic->backend_netdev_up = 0;
++
++ if (vnic->frontend_ready)
++ vnic_stop_fastpath(vnic);
++ }
+}
+
-+static void network_alloc_rx_buffers(struct net_device *dev)
++
++static int vnic_add_bufs(netfront_accel_vnic *vnic,
++ struct net_accel_msg *msg)
+{
-+ unsigned short id;
-+ struct netfront_info *np = netdev_priv(dev);
-+ struct sk_buff *skb;
-+ struct page *page;
-+ int i, batch_target, notify;
-+ RING_IDX req_prod = np->rx.req_prod_pvt;
-+ struct xen_memory_reservation reservation;
-+ grant_ref_t ref;
-+ unsigned long pfn;
-+ void *vaddr;
-+ int nr_flips;
-+ netif_rx_request_t *req;
++ int rc, offset;
++ struct netfront_accel_bufinfo *bufinfo;
++
++ BUG_ON(msg->u.mapbufs.pages > NET_ACCEL_MSG_MAX_PAGE_REQ);
+
-+ if (unlikely(!netfront_carrier_ok(np)))
-+ return;
++ offset = msg->u.mapbufs.reqid;
+
-+ /*
-+ * Allocate skbuffs greedily, even though we batch updates to the
-+ * receive ring. This creates a less bursty demand on the memory
-+ * allocator, so should reduce the chance of failed allocation requests
-+ * both for ourself and for other kernel subsystems.
-+ */
-+ batch_target = np->rx_target - (req_prod - np->rx.rsp_cons);
-+ for (i = skb_queue_len(&np->rx_batch); i < batch_target; i++) {
-+ /*
-+ * Allocate an skb and a page. Do not use __dev_alloc_skb as
-+ * that will allocate page-sized buffers which is not
-+ * necessary here.
-+ * 16 bytes added as necessary headroom for netif_receive_skb.
-+ */
-+ skb = alloc_skb(RX_COPY_THRESHOLD + 16 + NET_IP_ALIGN,
-+ GFP_ATOMIC | __GFP_NOWARN);
-+ if (unlikely(!skb))
-+ goto no_skb;
++ if (offset < vnic->bufpages.max_pages -
++ (vnic->bufpages.max_pages / sfc_netfront_buffer_split)) {
++ bufinfo = vnic->rx_bufs;
++ } else
++ bufinfo = vnic->tx_bufs;
+
-+ page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
-+ if (!page) {
-+ kfree_skb(skb);
-+no_skb:
-+ /* Any skbuffs queued for refill? Force them out. */
-+ if (i != 0)
-+ goto refill;
-+ /* Could not allocate any skbuffs. Try again later. */
-+ mod_timer(&np->rx_refill_timer,
-+ jiffies + (HZ/10));
-+ break;
-+ }
++ /* Queue up some Rx buffers to start things off. */
++ if ((rc = netfront_accel_add_bufs(&vnic->bufpages, bufinfo, msg)) == 0) {
++ netfront_accel_vi_add_bufs(vnic, bufinfo == vnic->rx_bufs);
+
-+ skb_reserve(skb, 16 + NET_IP_ALIGN); /* mimic dev_alloc_skb() */
-+ skb_shinfo(skb)->frags[0].page = page;
-+ skb_shinfo(skb)->nr_frags = 1;
-+ __skb_queue_tail(&np->rx_batch, skb);
++ if (offset + msg->u.mapbufs.pages == vnic->bufpages.max_pages) {
++ VPRINTK("%s: got all buffers back\n", __FUNCTION__);
++ vnic->frontend_ready = 1;
++ if (vnic->backend_netdev_up)
++ vnic_start_fastpath(vnic);
++ } else {
++ VPRINTK("%s: got buffers back %d %d\n", __FUNCTION__,
++ offset, msg->u.mapbufs.pages);
++ }
+ }
+
-+ /* Is the batch large enough to be worthwhile? */
-+ if (i < (np->rx_target/2)) {
-+ if (req_prod > np->rx.sring->req_prod)
-+ goto push;
-+ return;
-+ }
++ return rc;
++}
+
-+ /* Adjust our fill target if we risked running out of buffers. */
-+ if (((req_prod - np->rx.sring->rsp_prod) < (np->rx_target / 4)) &&
-+ ((np->rx_target *= 2) > np->rx_max_target))
-+ np->rx_target = np->rx_max_target;
+
-+ refill:
-+ for (nr_flips = i = 0; ; i++) {
-+ if ((skb = __skb_dequeue(&np->rx_batch)) == NULL)
-+ break;
++/* The largest [o] such that (1u << o) <= n. Requires n > 0. */
+
-+ skb->dev = dev;
++inline unsigned log2_le(unsigned long n) {
++ unsigned order = 1;
++ while ((1ul << order) <= n) ++order;
++ return (order - 1);
++}
+
-+ id = xennet_rxidx(req_prod + i);
++static int vnic_send_buffer_requests(netfront_accel_vnic *vnic,
++ struct netfront_accel_bufpages *bufpages)
++{
++ int pages, offset, rc = 0, sent = 0;
++ struct net_accel_msg msg;
+
-+ BUG_ON(np->rx_skbs[id]);
-+ np->rx_skbs[id] = skb;
++ while (bufpages->page_reqs < bufpages->max_pages) {
++ offset = bufpages->page_reqs;
+
-+ ref = gnttab_claim_grant_reference(&np->gref_rx_head);
-+ BUG_ON((signed short)ref < 0);
-+ np->grant_rx_ref[id] = ref;
++ pages = pow2(log2_le(bufpages->max_pages -
++ bufpages->page_reqs));
++ pages = pages < NET_ACCEL_MSG_MAX_PAGE_REQ ?
++ pages : NET_ACCEL_MSG_MAX_PAGE_REQ;
+
-+ pfn = page_to_pfn(skb_shinfo(skb)->frags[0].page);
-+ vaddr = page_address(skb_shinfo(skb)->frags[0].page);
++ BUG_ON(offset < 0);
++ BUG_ON(pages <= 0);
+
-+ req = RING_GET_REQUEST(&np->rx, req_prod + i);
-+ if (!np->copying_receiver) {
-+ gnttab_grant_foreign_transfer_ref(ref,
-+ np->xbdev->otherend_id,
-+ pfn);
-+ np->rx_pfn_array[nr_flips] = pfn_to_mfn(pfn);
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* Remove this page before passing
-+ * back to Xen. */
-+ set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
-+ MULTI_update_va_mapping(np->rx_mcl+i,
-+ (unsigned long)vaddr,
-+ __pte(0), 0);
++ rc = netfront_accel_buf_map_request(vnic->dev, bufpages,
++ &msg, pages, offset);
++ if (rc == 0) {
++ rc = net_accel_msg_send(vnic->shared_page,
++ &vnic->to_dom0, &msg);
++ if (rc < 0) {
++ VPRINTK("%s: queue full, stopping for now\n",
++ __FUNCTION__);
++ break;
+ }
-+ nr_flips++;
++ sent++;
+ } else {
-+ gnttab_grant_foreign_access_ref(ref,
-+ np->xbdev->otherend_id,
-+ pfn_to_mfn(pfn),
-+ 0);
++ EPRINTK("%s: problem with grant, stopping for now\n",
++ __FUNCTION__);
++ break;
+ }
+
-+ req->id = id;
-+ req->gref = ref;
++ bufpages->page_reqs += pages;
+ }
+
-+ if ( nr_flips != 0 ) {
-+ /* Tell the ballon driver what is going on. */
-+ balloon_update_driver_allowance(i);
-+
-+ set_xen_guest_handle(reservation.extent_start,
-+ np->rx_pfn_array);
-+ reservation.nr_extents = nr_flips;
-+ reservation.extent_order = 0;
-+ reservation.address_bits = 0;
-+ reservation.domid = DOMID_SELF;
-+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* After all PTEs have been zapped, flush the TLB. */
-+ np->rx_mcl[i-1].args[MULTI_UVMFLAGS_INDEX] =
-+ UVMF_TLB_FLUSH|UVMF_ALL;
++ if (sent)
++ net_accel_msg_notify(vnic->msg_channel_irq);
+
-+ /* Give away a batch of pages. */
-+ np->rx_mcl[i].op = __HYPERVISOR_memory_op;
-+ np->rx_mcl[i].args[0] = XENMEM_decrease_reservation;
-+ np->rx_mcl[i].args[1] = (unsigned long)&reservation;
++ return rc;
++}
+
-+ /* Zap PTEs and give away pages in one big
-+ * multicall. */
-+ (void)HYPERVISOR_multicall(np->rx_mcl, i+1);
+
-+ /* Check return status of HYPERVISOR_memory_op(). */
-+ if (unlikely(np->rx_mcl[i].result != i))
-+ panic("Unable to reduce memory reservation\n");
-+ } else {
-+ if (HYPERVISOR_memory_op(XENMEM_decrease_reservation,
-+ &reservation) != i)
-+ panic("Unable to reduce memory reservation\n");
-+ }
-+ } else {
-+ wmb();
-+ }
++/*
++ * In response to dom0 saying "my queue is full", we reply with this
++ * when it is no longer full
++ */
++inline void vnic_set_queue_not_full(netfront_accel_vnic *vnic)
++{
+
-+ /* Above is a suitable barrier to ensure backend will see requests. */
-+ np->rx.req_prod_pvt = req_prod + i;
-+ push:
-+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&np->rx, notify);
-+ if (notify)
-+ notify_remote_via_irq(np->irq);
++ if (test_and_set_bit(NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL_B,
++ (unsigned long *)&vnic->shared_page->aflags))
++ notify_remote_via_irq(vnic->msg_channel_irq);
++ else
++ VPRINTK("queue not full bit already set, not signalling\n");
+}
+
-+static void xennet_make_frags(struct sk_buff *skb, struct net_device *dev,
-+ struct netif_tx_request *tx)
++/*
++ * Notify dom0 that the queue we want to use is full, it should
++ * respond by setting MSG_AFLAGS_QUEUEUNOTFULL in due course
++ */
++inline void vnic_set_queue_full(netfront_accel_vnic *vnic)
+{
-+ struct netfront_info *np = netdev_priv(dev);
-+ char *data = skb->data;
-+ unsigned long mfn;
-+ RING_IDX prod = np->tx.req_prod_pvt;
-+ int frags = skb_shinfo(skb)->nr_frags;
-+ unsigned int offset = offset_in_page(data);
-+ unsigned int len = skb_headlen(skb);
-+ unsigned int id;
-+ grant_ref_t ref;
-+ int i;
-+
-+ while (len > PAGE_SIZE - offset) {
-+ tx->size = PAGE_SIZE - offset;
-+ tx->flags |= NETTXF_more_data;
-+ len -= tx->size;
-+ data += tx->size;
-+ offset = 0;
+
-+ id = get_id_from_freelist(np->tx_skbs);
-+ np->tx_skbs[id] = skb_get(skb);
-+ tx = RING_GET_REQUEST(&np->tx, prod++);
-+ tx->id = id;
-+ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
-+ BUG_ON((signed short)ref < 0);
++ if (!test_and_set_bit(NET_ACCEL_MSG_AFLAGS_QUEUEUFULL_B,
++ (unsigned long *)&vnic->shared_page->aflags))
++ notify_remote_via_irq(vnic->msg_channel_irq);
++ else
++ VPRINTK("queue full bit already set, not signalling\n");
++}
+
-+ mfn = virt_to_mfn(data);
-+ gnttab_grant_foreign_access_ref(ref, np->xbdev->otherend_id,
-+ mfn, GTF_readonly);
+
-+ tx->gref = np->grant_tx_ref[id] = ref;
-+ tx->offset = offset;
-+ tx->size = len;
-+ tx->flags = 0;
++static int vnic_check_hello_version(unsigned version)
++{
++ if (version > NET_ACCEL_MSG_VERSION) {
++ /* Newer protocol, we must refuse */
++ return -EPROTO;
+ }
+
-+ for (i = 0; i < frags; i++) {
-+ skb_frag_t *frag = skb_shinfo(skb)->frags + i;
-+
-+ tx->flags |= NETTXF_more_data;
-+
-+ id = get_id_from_freelist(np->tx_skbs);
-+ np->tx_skbs[id] = skb_get(skb);
-+ tx = RING_GET_REQUEST(&np->tx, prod++);
-+ tx->id = id;
-+ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
-+ BUG_ON((signed short)ref < 0);
-+
-+ mfn = pfn_to_mfn(page_to_pfn(frag->page));
-+ gnttab_grant_foreign_access_ref(ref, np->xbdev->otherend_id,
-+ mfn, GTF_readonly);
-+
-+ tx->gref = np->grant_tx_ref[id] = ref;
-+ tx->offset = frag->page_offset;
-+ tx->size = frag->size;
-+ tx->flags = 0;
++ if (version < NET_ACCEL_MSG_VERSION) {
++ /*
++ * We are newer, so have discretion to accept if we
++ * wish. For now however, just reject
++ */
++ return -EPROTO;
+ }
+
-+ np->tx.req_prod_pvt = prod;
++ BUG_ON(version != NET_ACCEL_MSG_VERSION);
++ return 0;
+}
+
-+static int network_start_xmit(struct sk_buff *skb, struct net_device *dev)
-+{
-+ unsigned short id;
-+ struct netfront_info *np = netdev_priv(dev);
-+ struct netif_tx_request *tx;
-+ struct netif_extra_info *extra;
-+ char *data = skb->data;
-+ RING_IDX i;
-+ grant_ref_t ref;
-+ unsigned long mfn;
-+ int notify;
-+ int frags = skb_shinfo(skb)->nr_frags;
-+ unsigned int offset = offset_in_page(data);
-+ unsigned int len = skb_headlen(skb);
+
-+ /* Check the fast path, if hooks are available */
-+ if (np->accel_vif_state.hooks &&
-+ np->accel_vif_state.hooks->start_xmit(skb, dev)) {
-+ /* Fast path has sent this packet */
-+ return 0;
-+ }
++static int vnic_process_hello_msg(netfront_accel_vnic *vnic,
++ struct net_accel_msg *msg)
++{
++ int err = 0;
++ unsigned pages = sfc_netfront_max_pages;
+
-+ frags += (offset + len + PAGE_SIZE - 1) / PAGE_SIZE;
-+ if (unlikely(frags > MAX_SKB_FRAGS + 1)) {
-+ printk(KERN_ALERT "xennet: skb rides the rocket: %d frags\n",
-+ frags);
-+ dump_stack();
-+ goto drop;
++ if (vnic_check_hello_version(msg->u.hello.version) < 0) {
++ msg->id = NET_ACCEL_MSG_HELLO | NET_ACCEL_MSG_REPLY
++ | NET_ACCEL_MSG_ERROR;
++ msg->u.hello.version = NET_ACCEL_MSG_VERSION;
++ } else {
++ vnic->backend_netdev_up
++ = vnic->shared_page->net_dev_up;
++
++ msg->id = NET_ACCEL_MSG_HELLO | NET_ACCEL_MSG_REPLY;
++ msg->u.hello.version = NET_ACCEL_MSG_VERSION;
++ if (msg->u.hello.max_pages &&
++ msg->u.hello.max_pages < pages)
++ pages = msg->u.hello.max_pages;
++ msg->u.hello.max_pages = pages;
++
++ /* Half of pages for rx, half for tx */
++ err = netfront_accel_alloc_buffer_mem(&vnic->bufpages,
++ vnic->rx_bufs,
++ vnic->tx_bufs,
++ pages);
++ if (err)
++ msg->id |= NET_ACCEL_MSG_ERROR;
+ }
++
++ /* Send reply */
++ net_accel_msg_reply_notify(vnic->shared_page, vnic->msg_channel_irq,
++ &vnic->to_dom0, msg);
++ return err;
++}
+
-+ spin_lock_irq(&np->tx_lock);
+
-+ if (unlikely(!netfront_carrier_ok(np) ||
-+ (frags > 1 && !xennet_can_sg(dev)) ||
-+ netif_needs_gso(dev, skb))) {
-+ spin_unlock_irq(&np->tx_lock);
-+ goto drop;
++static int vnic_process_localmac_msg(netfront_accel_vnic *vnic,
++ struct net_accel_msg *msg)
++{
++ unsigned long flags;
++ cuckoo_hash_mac_key key;
++
++ if (msg->u.localmac.flags & NET_ACCEL_MSG_ADD) {
++ DPRINTK("MAC has moved, could be local: " MAC_FMT "\n",
++ MAC_ARG(msg->u.localmac.mac));
++ key = cuckoo_mac_to_key(msg->u.localmac.mac);
++ spin_lock_irqsave(&vnic->table_lock, flags);
++ /* Try to remove it, not a big deal if not there */
++ cuckoo_hash_remove(&vnic->fastpath_table,
++ (cuckoo_hash_key *)&key);
++ spin_unlock_irqrestore(&vnic->table_lock, flags);
+ }
++
++ return 0;
++}
+
-+ i = np->tx.req_prod_pvt;
+
-+ id = get_id_from_freelist(np->tx_skbs);
-+ np->tx_skbs[id] = skb;
++static
++int vnic_process_rx_msg(netfront_accel_vnic *vnic,
++ struct net_accel_msg *msg)
++{
++ int err;
+
-+ tx = RING_GET_REQUEST(&np->tx, i);
++ switch (msg->id) {
++ case NET_ACCEL_MSG_HELLO:
++ /* Hello, reply with Reply */
++ DPRINTK("got Hello, with version %.8x\n",
++ msg->u.hello.version);
++ BUG_ON(vnic->msg_state != NETFRONT_ACCEL_MSG_NONE);
++ err = vnic_process_hello_msg(vnic, msg);
++ if (err == 0)
++ vnic->msg_state = NETFRONT_ACCEL_MSG_HELLO;
++ break;
++ case NET_ACCEL_MSG_SETHW:
++ /* Hardware info message */
++ DPRINTK("got H/W info\n");
++ BUG_ON(vnic->msg_state != NETFRONT_ACCEL_MSG_HELLO);
++ err = netfront_accel_vi_init(vnic, &msg->u.hw);
++ if (err == 0)
++ vnic->msg_state = NETFRONT_ACCEL_MSG_HW;
++ break;
++ case NET_ACCEL_MSG_MAPBUF | NET_ACCEL_MSG_REPLY:
++ VPRINTK("Got mapped buffers back\n");
++ BUG_ON(vnic->msg_state != NETFRONT_ACCEL_MSG_HW);
++ err = vnic_add_bufs(vnic, msg);
++ break;
++ case NET_ACCEL_MSG_MAPBUF | NET_ACCEL_MSG_REPLY | NET_ACCEL_MSG_ERROR:
++ /* No buffers. Can't use the fast path. */
++ EPRINTK("Got mapped buffers error. Cannot accelerate.\n");
++ BUG_ON(vnic->msg_state != NETFRONT_ACCEL_MSG_HW);
++ err = -EIO;
++ break;
++ case NET_ACCEL_MSG_LOCALMAC:
++ /* Should be add, remove not currently used */
++ EPRINTK_ON(!(msg->u.localmac.flags & NET_ACCEL_MSG_ADD));
++ BUG_ON(vnic->msg_state != NETFRONT_ACCEL_MSG_HW);
++ err = vnic_process_localmac_msg(vnic, msg);
++ break;
++ default:
++ EPRINTK("Huh? Message code is 0x%x\n", msg->id);
++ err = -EPROTO;
++ break;
++ }
+
-+ tx->id = id;
-+ ref = gnttab_claim_grant_reference(&np->gref_tx_head);
-+ BUG_ON((signed short)ref < 0);
-+ mfn = virt_to_mfn(data);
-+ gnttab_grant_foreign_access_ref(
-+ ref, np->xbdev->otherend_id, mfn, GTF_readonly);
-+ tx->gref = np->grant_tx_ref[id] = ref;
-+ tx->offset = offset;
-+ tx->size = len;
++ return err;
++}
+
-+ tx->flags = 0;
-+ extra = NULL;
+
-+ if (skb->ip_summed == CHECKSUM_HW) /* local packet? */
-+ tx->flags |= NETTXF_csum_blank | NETTXF_data_validated;
-+#ifdef CONFIG_XEN
-+ if (skb->proto_data_valid) /* remote but checksummed? */
-+ tx->flags |= NETTXF_data_validated;
++/* Process an IRQ received from back end driver */
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++void netfront_accel_msg_from_bend(struct work_struct *context)
++#else
++void netfront_accel_msg_from_bend(void *context)
++#endif
++{
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ netfront_accel_vnic *vnic =
++ container_of(context, netfront_accel_vnic, msg_from_bend);
++#else
++ netfront_accel_vnic *vnic = (netfront_accel_vnic *)context;
+#endif
++ struct net_accel_msg msg;
++ int err, queue_was_full = 0;
++
++ mutex_lock(&vnic->vnic_mutex);
+
-+#if HAVE_TSO
-+ if (skb_shinfo(skb)->gso_size) {
-+ struct netif_extra_info *gso = (struct netif_extra_info *)
-+ RING_GET_REQUEST(&np->tx, ++i);
++ /*
++ * This happens when the shared pages have been unmapped but
++ * the workqueue has yet to be flushed
++ */
++ if (!vnic->dom0_state_is_setup)
++ goto unlock_out;
+
-+ if (extra)
-+ extra->flags |= XEN_NETIF_EXTRA_FLAG_MORE;
-+ else
-+ tx->flags |= NETTXF_extra_info;
++ while ((vnic->shared_page->aflags & NET_ACCEL_MSG_AFLAGS_TO_DOMU_MASK)
++ != 0) {
++ if (vnic->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL) {
++ /* We've been told there may now be space. */
++ clear_bit(NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL_B,
++ (unsigned long *)&vnic->shared_page->aflags);
++ }
+
-+ gso->u.gso.size = skb_shinfo(skb)->gso_size;
-+ gso->u.gso.type = XEN_NETIF_GSO_TYPE_TCPV4;
-+ gso->u.gso.pad = 0;
-+ gso->u.gso.features = 0;
++ if (vnic->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_QUEUE0FULL) {
++ /*
++ * There will be space at the end of this
++ * function if we can make any.
++ */
++ clear_bit(NET_ACCEL_MSG_AFLAGS_QUEUE0FULL_B,
++ (unsigned long *)&vnic->shared_page->aflags);
++ queue_was_full = 1;
++ }
++
++ if (vnic->shared_page->aflags &
++ NET_ACCEL_MSG_AFLAGS_NETUPDOWN) {
++ DPRINTK("%s: net interface change\n", __FUNCTION__);
++ clear_bit(NET_ACCEL_MSG_AFLAGS_NETUPDOWN_B,
++ (unsigned long *)&vnic->shared_page->aflags);
++ if (vnic->shared_page->net_dev_up)
++ netfront_accel_interface_up(vnic);
++ else
++ netfront_accel_interface_down(vnic);
++ }
++ }
+
-+ gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
-+ gso->flags = 0;
-+ extra = gso;
++ /* Pull msg out of shared memory */
++ while ((err = net_accel_msg_recv(vnic->shared_page, &vnic->from_dom0,
++ &msg)) == 0) {
++ err = vnic_process_rx_msg(vnic, &msg);
++
++ if (err != 0)
++ goto done;
+ }
-+#endif
+
-+ np->tx.req_prod_pvt = i + 1;
++ /*
++ * Send any pending buffer map request messages that we can,
++ * and mark domU->dom0 as full if necessary.
++ */
++ if (vnic->msg_state == NETFRONT_ACCEL_MSG_HW &&
++ vnic->bufpages.page_reqs < vnic->bufpages.max_pages) {
++ if (vnic_send_buffer_requests(vnic, &vnic->bufpages) == -ENOSPC)
++ vnic_set_queue_full(vnic);
++ }
+
-+ xennet_make_frags(skb, dev, tx);
-+ tx->size = skb->len;
++ /*
++ * If there are no messages then this is not an error. It
++ * just means that we've finished processing the queue.
++ */
++ if (err == -ENOENT)
++ err = 0;
++ done:
++ /* We will now have made space in the dom0->domU queue if we can */
++ if (queue_was_full)
++ vnic_set_queue_not_full(vnic);
+
-+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&np->tx, notify);
-+ if (notify)
-+ notify_remote_via_irq(np->irq);
++ if (err != 0) {
++ EPRINTK("%s returned %d\n", __FUNCTION__, err);
++ netfront_accel_set_closing(vnic);
++ }
+
-+ np->stats.tx_bytes += skb->len;
-+ np->stats.tx_packets++;
++ unlock_out:
++ mutex_unlock(&vnic->vnic_mutex);
+
-+ /* Note: It is not safe to access skb after network_tx_buf_gc()! */
-+ network_tx_buf_gc(dev);
++ return;
++}
+
-+ if (!netfront_tx_slot_available(np))
-+ netif_stop_queue(dev);
+
-+ spin_unlock_irq(&np->tx_lock);
++irqreturn_t netfront_accel_msg_channel_irq_from_bend(int irq, void *context,
++ struct pt_regs *unused)
++{
++ netfront_accel_vnic *vnic = (netfront_accel_vnic *)context;
++ VPRINTK("irq %d from device %s\n", irq, vnic->dev->nodename);
+
-+ return 0;
++ queue_work(netfront_accel_workqueue, &vnic->msg_from_bend);
+
-+ drop:
-+ np->stats.tx_dropped++;
-+ dev_kfree_skb(skb);
-+ return 0;
++ return IRQ_HANDLED;
+}
+
-+static irqreturn_t netif_int(int irq, void *dev_id, struct pt_regs *ptregs)
++/* Process an interrupt received from the NIC via backend */
++irqreturn_t netfront_accel_net_channel_irq_from_bend(int irq, void *context,
++ struct pt_regs *unused)
+{
-+ struct net_device *dev = dev_id;
-+ struct netfront_info *np = netdev_priv(dev);
++ netfront_accel_vnic *vnic = (netfront_accel_vnic *)context;
++ struct net_device *net_dev = vnic->net_dev;
+ unsigned long flags;
+
-+ spin_lock_irqsave(&np->tx_lock, flags);
-+
-+ if (likely(netfront_carrier_ok(np))) {
-+ network_tx_buf_gc(dev);
-+ /* Under tx_lock: protects access to rx shared-ring indexes. */
-+ if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)) {
-+ netfront_accelerator_call_stop_napi_irq(np, dev);
++ VPRINTK("net irq %d from device %s\n", irq, vnic->dev->nodename);
++
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.irq_count++);
+
-+ netif_rx_schedule(dev);
-+ }
-+ }
++ BUG_ON(net_dev==NULL);
+
-+ spin_unlock_irqrestore(&np->tx_lock, flags);
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
++ if (vnic->irq_enabled) {
++ netfront_accel_disable_net_interrupts(vnic);
++ vnic->irq_enabled = 0;
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
+
++#if NETFRONT_ACCEL_STATS
++ vnic->stats.poll_schedule_count++;
++ if (vnic->stats.event_count_since_irq >
++ vnic->stats.events_per_irq_max)
++ vnic->stats.events_per_irq_max =
++ vnic->stats.event_count_since_irq;
++ vnic->stats.event_count_since_irq = 0;
++#endif
++ netif_rx_schedule(net_dev);
++ }
++ else {
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.useless_irq_count++);
++ DPRINTK("%s: irq when disabled\n", __FUNCTION__);
++ }
++
+ return IRQ_HANDLED;
+}
+
-+static void xennet_move_rx_slot(struct netfront_info *np, struct sk_buff *skb,
-+ grant_ref_t ref)
++
++void netfront_accel_msg_tx_fastpath(netfront_accel_vnic *vnic, const void *mac,
++ u32 ip, u16 port, u8 protocol)
+{
-+ int new = xennet_rxidx(np->rx.req_prod_pvt);
++ unsigned long lock_state;
++ struct net_accel_msg *msg;
+
-+ BUG_ON(np->rx_skbs[new]);
-+ np->rx_skbs[new] = skb;
-+ np->grant_rx_ref[new] = ref;
-+ RING_GET_REQUEST(&np->rx, np->rx.req_prod_pvt)->id = new;
-+ RING_GET_REQUEST(&np->rx, np->rx.req_prod_pvt)->gref = ref;
-+ np->rx.req_prod_pvt++;
-+}
++ msg = net_accel_msg_start_send(vnic->shared_page, &vnic->to_dom0,
++ &lock_state);
+
-+int xennet_get_extras(struct netfront_info *np,
-+ struct netif_extra_info *extras, RING_IDX rp)
++ if (msg == NULL)
++ return;
+
-+{
-+ struct netif_extra_info *extra;
-+ RING_IDX cons = np->rx.rsp_cons;
-+ int err = 0;
++ net_accel_msg_init(msg, NET_ACCEL_MSG_FASTPATH);
++ msg->u.fastpath.flags = NET_ACCEL_MSG_REMOVE;
++ memcpy(msg->u.fastpath.mac, mac, ETH_ALEN);
+
-+ do {
-+ struct sk_buff *skb;
-+ grant_ref_t ref;
++ msg->u.fastpath.port = port;
++ msg->u.fastpath.ip = ip;
++ msg->u.fastpath.proto = protocol;
+
-+ if (unlikely(cons + 1 == rp)) {
-+ if (net_ratelimit())
-+ WPRINTK("Missing extra info\n");
-+ err = -EBADR;
-+ break;
-+ }
++ net_accel_msg_complete_send_notify(vnic->shared_page, &vnic->to_dom0,
++ &lock_state, vnic->msg_channel_irq);
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_netfront.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_netfront.c 2008-05-19 00:33:48.618948572 +0300
+@@ -0,0 +1,319 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ extra = (struct netif_extra_info *)
-+ RING_GET_RESPONSE(&np->rx, ++cons);
++#include <linux/skbuff.h>
++#include <linux/netdevice.h>
+
-+ if (unlikely(!extra->type ||
-+ extra->type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
-+ if (net_ratelimit())
-+ WPRINTK("Invalid extra type: %d\n",
-+ extra->type);
-+ err = -EINVAL;
-+ } else {
-+ memcpy(&extras[extra->type - 1], extra,
-+ sizeof(*extra));
-+ }
++/* drivers/xen/netfront/netfront.h */
++#include "netfront.h"
+
-+ skb = xennet_get_rx_skb(np, cons);
-+ ref = xennet_get_rx_ref(np, cons);
-+ xennet_move_rx_slot(np, skb, ref);
-+ } while (extra->flags & XEN_NETIF_EXTRA_FLAG_MORE);
++#include "accel.h"
++#include "accel_bufs.h"
++#include "accel_util.h"
++#include "accel_msg_iface.h"
++#include "accel_ssr.h"
++
++#ifdef EFX_GCOV
++#include "gcov.h"
++#endif
+
-+ np->rx.rsp_cons = cons;
-+ return err;
-+}
++#define NETFRONT_ACCEL_VNIC_FROM_NETDEV(_nd) \
++ ((netfront_accel_vnic *)((struct netfront_info *)netdev_priv(net_dev))->accel_priv)
+
-+static int xennet_get_responses(struct netfront_info *np,
-+ struct netfront_rx_info *rinfo, RING_IDX rp,
-+ struct sk_buff_head *list,
-+ int *pages_flipped_p)
++static int netfront_accel_netdev_start_xmit(struct sk_buff *skb,
++ struct net_device *net_dev)
+{
-+ int pages_flipped = *pages_flipped_p;
-+ struct mmu_update *mmu;
-+ struct multicall_entry *mcl;
-+ struct netif_rx_response *rx = &rinfo->rx;
-+ struct netif_extra_info *extras = rinfo->extras;
-+ RING_IDX cons = np->rx.rsp_cons;
-+ struct sk_buff *skb = xennet_get_rx_skb(np, cons);
-+ grant_ref_t ref = xennet_get_rx_ref(np, cons);
-+ int max = MAX_SKB_FRAGS + (rx->status <= RX_COPY_THRESHOLD);
-+ int frags = 1;
-+ int err = 0;
-+ unsigned long ret;
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
++ struct netfront_info *np =
++ (struct netfront_info *)netdev_priv(net_dev);
++ int handled, rc;
++ unsigned long flags1, flags2;
+
-+ if (rx->flags & NETRXF_extra_info) {
-+ err = xennet_get_extras(np, extras, rp);
-+ cons = np->rx.rsp_cons;
++ BUG_ON(vnic == NULL);
++
++ /* Take our tx lock and hold for the duration */
++ spin_lock_irqsave(&vnic->tx_lock, flags1);
++
++ if (!vnic->tx_enabled) {
++ rc = 0;
++ goto unlock_out;
+ }
+
-+ for (;;) {
-+ unsigned long mfn;
++ handled = netfront_accel_vi_tx_post(vnic, skb);
++ if (handled == NETFRONT_ACCEL_STATUS_BUSY) {
++ BUG_ON(vnic->net_dev != net_dev);
++ DPRINTK("%s stopping queue\n", __FUNCTION__);
+
-+ if (unlikely(rx->status < 0 ||
-+ rx->offset + rx->status > PAGE_SIZE)) {
-+ if (net_ratelimit())
-+ WPRINTK("rx->offset: %x, size: %u\n",
-+ rx->offset, rx->status);
-+ xennet_move_rx_slot(np, skb, ref);
-+ err = -EINVAL;
-+ goto next;
-+ }
++ /* Netfront's lock protects tx_skb */
++ spin_lock_irqsave(&np->tx_lock, flags2);
++ BUG_ON(vnic->tx_skb != NULL);
++ vnic->tx_skb = skb;
++ netif_stop_queue(net_dev);
++ spin_unlock_irqrestore(&np->tx_lock, flags2);
+
-+ /*
-+ * This definitely indicates a bug, either in this driver or in
-+ * the backend driver. In future this should flag the bad
-+ * situation to the system controller to reboot the backed.
-+ */
-+ if (ref == GRANT_INVALID_REF) {
-+ if (net_ratelimit())
-+ WPRINTK("Bad rx response id %d.\n", rx->id);
-+ err = -EINVAL;
-+ goto next;
-+ }
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.queue_stops++);
++ }
+
-+ if (!np->copying_receiver) {
-+ /* Memory pressure, insufficient buffer
-+ * headroom, ... */
-+ if (!(mfn = gnttab_end_foreign_transfer_ref(ref))) {
-+ if (net_ratelimit())
-+ WPRINTK("Unfulfilled rx req "
-+ "(id=%d, st=%d).\n",
-+ rx->id, rx->status);
-+ xennet_move_rx_slot(np, skb, ref);
-+ err = -ENOMEM;
-+ goto next;
-+ }
++ if (handled == NETFRONT_ACCEL_STATUS_CANT)
++ rc = 0;
++ else
++ rc = 1;
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* Remap the page. */
-+ struct page *page =
-+ skb_shinfo(skb)->frags[0].page;
-+ unsigned long pfn = page_to_pfn(page);
-+ void *vaddr = page_address(page);
++unlock_out:
++ spin_unlock_irqrestore(&vnic->tx_lock, flags1);
+
-+ mcl = np->rx_mcl + pages_flipped;
-+ mmu = np->rx_mmu + pages_flipped;
++ return rc;
++}
+
-+ MULTI_update_va_mapping(mcl,
-+ (unsigned long)vaddr,
-+ pfn_pte_ma(mfn,
-+ PAGE_KERNEL),
-+ 0);
-+ mmu->ptr = ((maddr_t)mfn << PAGE_SHIFT)
-+ | MMU_MACHPHYS_UPDATE;
-+ mmu->val = pfn;
+
-+ set_phys_to_machine(pfn, mfn);
-+ }
-+ pages_flipped++;
-+ } else {
-+ ret = gnttab_end_foreign_access_ref(ref);
-+ BUG_ON(!ret);
-+ }
++static int netfront_accel_netdev_poll(struct net_device *net_dev, int *budget)
++{
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
++ int rx_allowed = *budget, rx_done;
++
++ BUG_ON(vnic == NULL);
+
-+ gnttab_release_grant_reference(&np->gref_rx_head, ref);
++ /* Can check this without lock as modifier excludes polls */
++ if (!vnic->poll_enabled)
++ return 0;
+
-+ __skb_queue_tail(list, skb);
++ rx_done = netfront_accel_vi_poll(vnic, rx_allowed);
++ *budget -= rx_done;
++
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.poll_call_count++);
+
-+next:
-+ if (!(rx->flags & NETRXF_more_data))
-+ break;
++ VPRINTK("%s: done %d allowed %d\n",
++ __FUNCTION__, rx_done, rx_allowed);
+
-+ if (cons + frags == rp) {
-+ if (net_ratelimit())
-+ WPRINTK("Need more frags\n");
-+ err = -ENOENT;
-+ break;
-+ }
++ netfront_accel_ssr_end_of_burst(vnic, &vnic->ssr_state);
+
-+ rx = RING_GET_RESPONSE(&np->rx, cons + frags);
-+ skb = xennet_get_rx_skb(np, cons + frags);
-+ ref = xennet_get_rx_ref(np, cons + frags);
-+ frags++;
++ if (rx_done < rx_allowed) {
++ return 0; /* Done */
+ }
++
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.poll_reschedule_count++);
+
-+ if (unlikely(frags > max)) {
-+ if (net_ratelimit())
-+ WPRINTK("Too many frags\n");
-+ err = -E2BIG;
-+ }
++ return 1; /* More to do. */
++}
+
-+ if (unlikely(err))
-+ np->rx.rsp_cons = cons + frags;
+
-+ *pages_flipped_p = pages_flipped;
++/*
++ * Process request from netfront to start napi interrupt
++ * mode. (i.e. enable interrupts as it's finished polling)
++ */
++static int netfront_accel_start_napi_interrupts(struct net_device *net_dev)
++{
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
++ unsigned long flags;
+
-+ return err;
-+}
++ BUG_ON(vnic == NULL);
++
++ /*
++ * Can check this without lock as writer excludes poll before
++ * modifying
++ */
++ if (!vnic->poll_enabled)
++ return 0;
+
-+static RING_IDX xennet_fill_frags(struct netfront_info *np,
-+ struct sk_buff *skb,
-+ struct sk_buff_head *list)
-+{
-+ struct skb_shared_info *shinfo = skb_shinfo(skb);
-+ int nr_frags = shinfo->nr_frags;
-+ RING_IDX cons = np->rx.rsp_cons;
-+ skb_frag_t *frag = shinfo->frags + nr_frags;
-+ struct sk_buff *nskb;
++ if (!netfront_accel_vi_enable_interrupts(vnic)) {
++ /*
++ * There was something there, tell caller we had
++ * something to do.
++ */
++ return 1;
++ }
+
-+ while ((nskb = __skb_dequeue(list))) {
-+ struct netif_rx_response *rx =
-+ RING_GET_RESPONSE(&np->rx, ++cons);
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
++ vnic->irq_enabled = 1;
++ netfront_accel_enable_net_interrupts(vnic);
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
+
-+ frag->page = skb_shinfo(nskb)->frags[0].page;
-+ frag->page_offset = rx->offset;
-+ frag->size = rx->status;
++ return 0;
++}
+
-+ skb->data_len += rx->status;
+
-+ skb_shinfo(nskb)->nr_frags = 0;
-+ kfree_skb(nskb);
++/*
++ * Process request from netfront to stop napi interrupt
++ * mode. (i.e. disable interrupts as it's starting to poll
++ */
++static void netfront_accel_stop_napi_interrupts(struct net_device *net_dev)
++{
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
++ unsigned long flags;
+
-+ frag++;
-+ nr_frags++;
++ BUG_ON(vnic == NULL);
++
++ spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
++
++ if (!vnic->poll_enabled) {
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
++ return;
+ }
+
-+ shinfo->nr_frags = nr_frags;
-+ return cons;
++ netfront_accel_disable_net_interrupts(vnic);
++ vnic->irq_enabled = 0;
++ spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
+}
+
-+static int xennet_set_skb_gso(struct sk_buff *skb,
-+ struct netif_extra_info *gso)
++
++static int netfront_accel_check_ready(struct net_device *net_dev)
+{
-+ if (!gso->u.gso.size) {
-+ if (net_ratelimit())
-+ WPRINTK("GSO size must not be zero.\n");
-+ return -EINVAL;
-+ }
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
+
-+ /* Currently only TCPv4 S.O. is supported. */
-+ if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4) {
-+ if (net_ratelimit())
-+ WPRINTK("Bad GSO type %d.\n", gso->u.gso.type);
-+ return -EINVAL;
-+ }
++ BUG_ON(vnic == NULL);
+
-+#if HAVE_TSO
-+ skb_shinfo(skb)->gso_size = gso->u.gso.size;
-+#if HAVE_GSO
-+ skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
++ /* This is protected by netfront's lock */
++ return vnic->tx_skb == NULL;
++}
+
-+ /* Header must be checked, and gso_segs computed. */
-+ skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
-+#endif
-+ skb_shinfo(skb)->gso_segs = 0;
++
++static int netfront_accel_get_stats(struct net_device *net_dev,
++ struct net_device_stats *stats)
++{
++ netfront_accel_vnic *vnic = NETFRONT_ACCEL_VNIC_FROM_NETDEV(net_dev);
++ struct netfront_accel_netdev_stats now;
++
++ BUG_ON(vnic == NULL);
++
++ now.fastpath_rx_pkts = vnic->netdev_stats.fastpath_rx_pkts;
++ now.fastpath_rx_bytes = vnic->netdev_stats.fastpath_rx_bytes;
++ now.fastpath_rx_errors = vnic->netdev_stats.fastpath_rx_errors;
++ now.fastpath_tx_pkts = vnic->netdev_stats.fastpath_tx_pkts;
++ now.fastpath_tx_bytes = vnic->netdev_stats.fastpath_tx_bytes;
++ now.fastpath_tx_errors = vnic->netdev_stats.fastpath_tx_errors;
++
++ stats->rx_packets += (now.fastpath_rx_pkts -
++ vnic->stats_last_read.fastpath_rx_pkts);
++ stats->rx_bytes += (now.fastpath_rx_bytes -
++ vnic->stats_last_read.fastpath_rx_bytes);
++ stats->rx_errors += (now.fastpath_rx_errors -
++ vnic->stats_last_read.fastpath_rx_errors);
++ stats->tx_packets += (now.fastpath_tx_pkts -
++ vnic->stats_last_read.fastpath_tx_pkts);
++ stats->tx_bytes += (now.fastpath_tx_bytes -
++ vnic->stats_last_read.fastpath_tx_bytes);
++ stats->tx_errors += (now.fastpath_tx_errors -
++ vnic->stats_last_read.fastpath_tx_errors);
++
++ vnic->stats_last_read = now;
+
+ return 0;
-+#else
-+ if (net_ratelimit())
-+ WPRINTK("GSO unsupported by this kernel.\n");
-+ return -EINVAL;
-+#endif
+}
+
-+static int netif_poll(struct net_device *dev, int *pbudget)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
-+ struct sk_buff *skb;
-+ struct netfront_rx_info rinfo;
-+ struct netif_rx_response *rx = &rinfo.rx;
-+ struct netif_extra_info *extras = rinfo.extras;
-+ RING_IDX i, rp;
-+ struct multicall_entry *mcl;
-+ int work_done, budget, more_to_do = 1, accel_more_to_do = 1;
-+ struct sk_buff_head rxq;
-+ struct sk_buff_head errq;
-+ struct sk_buff_head tmpq;
-+ unsigned long flags;
-+ unsigned int len;
-+ int pages_flipped = 0;
-+ int err;
+
-+ spin_lock(&np->rx_lock); /* no need for spin_lock_bh() in ->poll() */
++struct netfront_accel_hooks accel_hooks = {
++ .new_device = &netfront_accel_probe,
++ .remove = &netfront_accel_remove,
++ .netdev_poll = &netfront_accel_netdev_poll,
++ .start_xmit = &netfront_accel_netdev_start_xmit,
++ .start_napi_irq = &netfront_accel_start_napi_interrupts,
++ .stop_napi_irq = &netfront_accel_stop_napi_interrupts,
++ .check_ready = &netfront_accel_check_ready,
++ .get_stats = &netfront_accel_get_stats
++};
+
-+ if (unlikely(!netfront_carrier_ok(np))) {
-+ spin_unlock(&np->rx_lock);
-+ return 0;
-+ }
+
-+ skb_queue_head_init(&rxq);
-+ skb_queue_head_init(&errq);
-+ skb_queue_head_init(&tmpq);
++unsigned sfc_netfront_max_pages = NETFRONT_ACCEL_DEFAULT_BUF_PAGES;
++module_param_named (max_pages, sfc_netfront_max_pages, uint, 0644);
++MODULE_PARM_DESC(max_pages, "Number of buffer pages to request");
+
-+ if ((budget = *pbudget) > dev->quota)
-+ budget = dev->quota;
-+ rp = np->rx.sring->rsp_prod;
-+ rmb(); /* Ensure we see queued responses up to 'rp'. */
++unsigned sfc_netfront_buffer_split = 2;
++module_param_named (buffer_split, sfc_netfront_buffer_split, uint, 0644);
++MODULE_PARM_DESC(buffer_split,
++ "Fraction of buffers to use for TX, rest for RX");
+
-+ i = np->rx.rsp_cons;
-+ work_done = 0;
-+ while ((i != rp) && (work_done < budget)) {
-+ memcpy(rx, RING_GET_RESPONSE(&np->rx, i), sizeof(*rx));
-+ memset(extras, 0, sizeof(rinfo.extras));
+
-+ err = xennet_get_responses(np, &rinfo, rp, &tmpq,
-+ &pages_flipped);
++const char *frontend_name = "sfc_netfront";
+
-+ if (unlikely(err)) {
-+err:
-+ while ((skb = __skb_dequeue(&tmpq)))
-+ __skb_queue_tail(&errq, skb);
-+ np->stats.rx_errors++;
-+ i = np->rx.rsp_cons;
-+ continue;
-+ }
++struct workqueue_struct *netfront_accel_workqueue;
+
-+ skb = __skb_dequeue(&tmpq);
++static int __init netfront_accel_init(void)
++{
++ int rc;
++#ifdef EFX_GCOV
++ gcov_provider_init(THIS_MODULE);
++#endif
+
-+ if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
-+ struct netif_extra_info *gso;
-+ gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
++ /*
++ * If we're running on dom0, netfront hasn't initialised
++ * itself, so we need to keep away
++ */
++ if (is_initial_xendomain())
++ return 0;
+
-+ if (unlikely(xennet_set_skb_gso(skb, gso))) {
-+ __skb_queue_head(&tmpq, skb);
-+ np->rx.rsp_cons += skb_queue_len(&tmpq);
-+ goto err;
-+ }
-+ }
++ if (!is_pow2(sizeof(struct net_accel_msg)))
++ EPRINTK("%s: bad structure size\n", __FUNCTION__);
+
-+ NETFRONT_SKB_CB(skb)->page = skb_shinfo(skb)->frags[0].page;
-+ NETFRONT_SKB_CB(skb)->offset = rx->offset;
++ netfront_accel_workqueue = create_workqueue(frontend_name);
+
-+ len = rx->status;
-+ if (len > RX_COPY_THRESHOLD)
-+ len = RX_COPY_THRESHOLD;
-+ skb_put(skb, len);
++ netfront_accel_debugfs_init();
+
-+ if (rx->status > len) {
-+ skb_shinfo(skb)->frags[0].page_offset =
-+ rx->offset + len;
-+ skb_shinfo(skb)->frags[0].size = rx->status - len;
-+ skb->data_len = rx->status - len;
-+ } else {
-+ skb_shinfo(skb)->frags[0].page = NULL;
-+ skb_shinfo(skb)->nr_frags = 0;
-+ }
++ rc = netfront_accelerator_loaded(NETFRONT_ACCEL_VERSION,
++ frontend_name, &accel_hooks);
+
-+ i = xennet_fill_frags(np, skb, &tmpq);
++ if (rc < 0) {
++ EPRINTK("Xen netfront accelerator version mismatch\n");
++ return -EINVAL;
++ }
+
-+ /*
-+ * Truesize must approximates the size of true data plus
-+ * any supervisor overheads. Adding hypervisor overheads
-+ * has been shown to significantly reduce achievable
-+ * bandwidth with the default receive buffer size. It is
-+ * therefore not wise to account for it here.
-+ *
-+ * After alloc_skb(RX_COPY_THRESHOLD), truesize is set to
-+ * RX_COPY_THRESHOLD + the supervisor overheads. Here, we
-+ * add the size of the data pulled in xennet_fill_frags().
-+ *
-+ * We also adjust for any unused space in the main data
-+ * area by subtracting (RX_COPY_THRESHOLD - len). This is
-+ * especially important with drivers which split incoming
-+ * packets into header and data, using only 66 bytes of
-+ * the main data area (see the e1000 driver for example.)
-+ * On such systems, without this last adjustement, our
-+ * achievable receive throughout using the standard receive
-+ * buffer size was cut by 25%(!!!).
++ if (rc > 0) {
++ /*
++ * In future may want to add backwards compatibility
++ * and accept certain subsets of previous versions
+ */
-+ skb->truesize += skb->data_len - (RX_COPY_THRESHOLD - len);
-+ skb->len += skb->data_len;
++ EPRINTK("Xen netfront accelerator version mismatch\n");
++ return -EINVAL;
++ }
+
-+ /*
-+ * Old backends do not assert data_validated but we
-+ * can infer it from csum_blank so test both flags.
-+ */
-+ if (rx->flags & (NETRXF_data_validated|NETRXF_csum_blank))
-+ skb->ip_summed = CHECKSUM_UNNECESSARY;
-+ else
-+ skb->ip_summed = CHECKSUM_NONE;
-+#ifdef CONFIG_XEN
-+ skb->proto_data_valid = (skb->ip_summed != CHECKSUM_NONE);
-+ skb->proto_csum_blank = !!(rx->flags & NETRXF_csum_blank);
-+#endif
-+ np->stats.rx_packets++;
-+ np->stats.rx_bytes += skb->len;
++ return 0;
++}
++module_init(netfront_accel_init);
+
-+ __skb_queue_tail(&rxq, skb);
++static void __exit netfront_accel_exit(void)
++{
++ if (is_initial_xendomain())
++ return;
+
-+ np->rx.rsp_cons = ++i;
-+ work_done++;
-+ }
++ DPRINTK("%s: unhooking\n", __FUNCTION__);
+
-+ if (pages_flipped) {
-+ /* Some pages are no longer absent... */
-+ balloon_update_driver_allowance(-pages_flipped);
++ /* Unhook from normal netfront */
++ netfront_accelerator_stop(frontend_name);
+
-+ /* Do all the remapping work and M2P updates. */
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ mcl = np->rx_mcl + pages_flipped;
-+ mcl->op = __HYPERVISOR_mmu_update;
-+ mcl->args[0] = (unsigned long)np->rx_mmu;
-+ mcl->args[1] = pages_flipped;
-+ mcl->args[2] = 0;
-+ mcl->args[3] = DOMID_SELF;
-+ (void)HYPERVISOR_multicall(np->rx_mcl,
-+ pages_flipped + 1);
-+ }
-+ }
++ DPRINTK("%s: done\n", __FUNCTION__);
+
-+ while ((skb = __skb_dequeue(&errq)))
-+ kfree_skb(skb);
++ netfront_accel_debugfs_fini();
+
-+ while ((skb = __skb_dequeue(&rxq)) != NULL) {
-+ struct page *page = NETFRONT_SKB_CB(skb)->page;
-+ void *vaddr = page_address(page);
-+ unsigned offset = NETFRONT_SKB_CB(skb)->offset;
++ flush_workqueue(netfront_accel_workqueue);
+
-+ memcpy(skb->data, vaddr + offset, skb_headlen(skb));
++ destroy_workqueue(netfront_accel_workqueue);
+
-+ if (page != skb_shinfo(skb)->frags[0].page)
-+ __free_page(page);
++#ifdef EFX_GCOV
++ gcov_provider_fini(THIS_MODULE);
++#endif
++ return;
++}
++module_exit(netfront_accel_exit);
+
-+ /* Ethernet work: Delayed to here as it peeks the header. */
-+ skb->protocol = eth_type_trans(skb, dev);
++MODULE_LICENSE("GPL");
+
-+ /* Pass it up. */
-+ netif_receive_skb(skb);
-+ dev->last_rx = jiffies;
-+ }
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_ssr.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_ssr.c 2008-05-19 00:33:48.622948803 +0300
+@@ -0,0 +1,308 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* If we get a callback with very few responses, reduce fill target. */
-+ /* NB. Note exponential increase, linear decrease. */
-+ if (((np->rx.req_prod_pvt - np->rx.sring->rsp_prod) >
-+ ((3*np->rx_target) / 4)) &&
-+ (--np->rx_target < np->rx_min_target))
-+ np->rx_target = np->rx_min_target;
++#include <linux/socket.h>
++#include <linux/in.h>
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/list.h>
++#include <net/ip.h>
++#include <net/checksum.h>
+
-+ network_alloc_rx_buffers(dev);
++#include "accel.h"
++#include "accel_util.h"
++#include "accel_bufs.h"
+
-+ if (work_done < budget) {
-+ /* there's some spare capacity, try the accelerated path */
-+ int accel_budget = budget - work_done;
-+ int accel_budget_start = accel_budget;
++#include "accel_ssr.h"
+
-+ if (np->accel_vif_state.hooks) {
-+ accel_more_to_do =
-+ np->accel_vif_state.hooks->netdev_poll
-+ (dev, &accel_budget);
-+ work_done += (accel_budget_start - accel_budget);
-+ } else
-+ accel_more_to_do = 0;
++static inline int list_valid(struct list_head *lh) {
++ return(lh->next != NULL);
++}
++
++static void netfront_accel_ssr_deliver (struct netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st,
++ struct netfront_accel_ssr_conn *c);
++
++/** Construct an efx_ssr_state.
++ *
++ * @v st The SSR state (per channel per port)
++ * @v port The port.
++ */
++void netfront_accel_ssr_init(struct netfront_accel_ssr_state *st) {
++ unsigned i;
++
++ INIT_LIST_HEAD(&st->conns);
++ INIT_LIST_HEAD(&st->free_conns);
++ for (i = 0; i < 8; ++i) {
++ struct netfront_accel_ssr_conn *c =
++ kmalloc(sizeof(*c), GFP_KERNEL);
++ if (c == NULL) break;
++ c->n_in_order_pkts = 0;
++ c->skb = NULL;
++ list_add(&c->link, &st->free_conns);
+ }
+
-+ *pbudget -= work_done;
-+ dev->quota -= work_done;
++}
+
-+ if (work_done < budget) {
-+ local_irq_save(flags);
+
-+ RING_FINAL_CHECK_FOR_RESPONSES(&np->rx, more_to_do);
++/** Destructor for an efx_ssr_state.
++ *
++ * @v st The SSR state (per channel per port)
++ */
++void netfront_accel_ssr_fini(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st) {
++ struct netfront_accel_ssr_conn *c;
+
-+ if (!more_to_do && !accel_more_to_do &&
-+ np->accel_vif_state.hooks) {
-+ /*
-+ * Slow path has nothing more to do, see if
-+ * fast path is likewise
-+ */
-+ accel_more_to_do =
-+ np->accel_vif_state.hooks->start_napi_irq(dev);
-+ }
++ /* Return cleanly if efx_ssr_init() not previously called */
++ BUG_ON(list_valid(&st->conns) != list_valid(&st->free_conns));
++ if (! list_valid(&st->conns))
++ return;
+
-+ if (!more_to_do && !accel_more_to_do)
-+ __netif_rx_complete(dev);
++ while ( ! list_empty(&st->free_conns)) {
++ c = list_entry(st->free_conns.prev,
++ struct netfront_accel_ssr_conn, link);
++ list_del(&c->link);
++ BUG_ON(c->skb != NULL);
++ kfree(c);
++ }
++ while ( ! list_empty(&st->conns)) {
++ c = list_entry(st->conns.prev,
++ struct netfront_accel_ssr_conn, link);
++ list_del(&c->link);
++ if (c->skb)
++ netfront_accel_ssr_deliver(vnic, st, c);
++ kfree(c);
++ }
++}
+
-+ local_irq_restore(flags);
++
++/** Calc IP checksum and deliver to the OS
++ *
++ * @v st The SSR state (per channel per port)
++ * @v c The SSR connection state
++ */
++static void netfront_accel_ssr_deliver(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st,
++ struct netfront_accel_ssr_conn *c) {
++ BUG_ON(c->skb == NULL);
++
++ /*
++ * If we've chained packets together, recalculate the IP
++ * checksum.
++ */
++ if (skb_shinfo(c->skb)->frag_list) {
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_bursts);
++ c->iph->check = 0;
++ c->iph->check = ip_fast_csum((unsigned char *) c->iph,
++ c->iph->ihl);
+ }
+
-+ spin_unlock(&np->rx_lock);
-+
-+ return more_to_do | accel_more_to_do;
++ VPRINTK("%s: %d\n", __FUNCTION__, c->skb->len);
++
++ netif_receive_skb(c->skb);
++ c->skb = NULL;
+}
+
-+static void netif_release_tx_bufs(struct netfront_info *np)
-+{
-+ struct sk_buff *skb;
-+ int i;
+
-+ for (i = 1; i <= NET_TX_RING_SIZE; i++) {
-+ if ((unsigned long)np->tx_skbs[i] < PAGE_OFFSET)
-+ continue;
++/** Push held skbs down into network stack.
++ *
++ * @v st SSR state
++ *
++ * Only called if we are tracking one or more connections.
++ */
++void __netfront_accel_ssr_end_of_burst(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st) {
++ struct netfront_accel_ssr_conn *c;
+
-+ skb = np->tx_skbs[i];
-+ gnttab_end_foreign_access_ref(np->grant_tx_ref[i]);
-+ gnttab_release_grant_reference(
-+ &np->gref_tx_head, np->grant_tx_ref[i]);
-+ np->grant_tx_ref[i] = GRANT_INVALID_REF;
-+ add_id_to_freelist(np->tx_skbs, i);
-+ dev_kfree_skb_irq(skb);
++ BUG_ON(list_empty(&st->conns));
++
++ list_for_each_entry(c, &st->conns, link)
++ if (c->skb)
++ netfront_accel_ssr_deliver(vnic, st, c);
++
++ /* Time-out connections that have received no traffic for 20ms. */
++ c = list_entry(st->conns.prev, struct netfront_accel_ssr_conn,
++ link);
++ if (jiffies - c->last_pkt_jiffies > (HZ / 50 + 1)) {
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_drop_stream);
++ list_del(&c->link);
++ list_add(&c->link, &st->free_conns);
+ }
+}
+
-+static void netif_release_rx_bufs_flip(struct netfront_info *np)
-+{
-+ struct mmu_update *mmu = np->rx_mmu;
-+ struct multicall_entry *mcl = np->rx_mcl;
-+ struct sk_buff_head free_list;
-+ struct sk_buff *skb;
-+ unsigned long mfn;
-+ int xfer = 0, noxfer = 0, unused = 0;
-+ int id, ref, rc;
+
-+ skb_queue_head_init(&free_list);
++/** Process SKB and decide whether to dispatch it to the stack now or
++ * later.
++ *
++ * @v st SSR state
++ * @v skb SKB to exmaine
++ * @ret rc 0 => deliver SKB to kernel now, otherwise the SKB belongs
++ * us.
++ */
++int netfront_accel_ssr_skb(struct netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st,
++ struct sk_buff *skb) {
++ int data_length, dont_merge;
++ struct netfront_accel_ssr_conn *c;
++ struct iphdr *iph;
++ struct tcphdr *th;
++ unsigned th_seq;
+
-+ spin_lock_bh(&np->rx_lock);
++ BUG_ON(skb_shinfo(skb)->frag_list != NULL);
++ BUG_ON(skb->next != NULL);
+
-+ for (id = 0; id < NET_RX_RING_SIZE; id++) {
-+ if ((ref = np->grant_rx_ref[id]) == GRANT_INVALID_REF) {
-+ unused++;
++ /* We're not interested if it isn't TCP over IPv4. */
++ iph = (struct iphdr *) skb->data;
++ if (skb->protocol != htons(ETH_P_IP) ||
++ iph->protocol != IPPROTO_TCP) {
++ return 0;
++ }
++
++ /* Ignore segments that fail csum or are fragmented. */
++ if (unlikely((skb->ip_summed - CHECKSUM_UNNECESSARY) |
++ (iph->frag_off & htons(IP_MF | IP_OFFSET)))) {
++ return 0;
++ }
++
++ th = (struct tcphdr*)(skb->data + iph->ihl * 4);
++ data_length = ntohs(iph->tot_len) - iph->ihl * 4 - th->doff * 4;
++ th_seq = ntohl(th->seq);
++ dont_merge = (data_length == 0) | th->urg | th->syn | th->rst;
++
++ list_for_each_entry(c, &st->conns, link) {
++ if ((c->saddr - iph->saddr) |
++ (c->daddr - iph->daddr) |
++ (c->source - th->source) |
++ (c->dest - th->dest ))
+ continue;
++
++ /* Re-insert at head of list to reduce lookup time. */
++ list_del(&c->link);
++ list_add(&c->link, &st->conns);
++ c->last_pkt_jiffies = jiffies;
++
++ if (unlikely(th_seq - c->next_seq)) {
++ /* Out-of-order, so start counting again. */
++ if (c->skb)
++ netfront_accel_ssr_deliver(vnic, st, c);
++ c->n_in_order_pkts = 0;
++ c->next_seq = th_seq + data_length;
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_misorder);
++ return 0;
+ }
++ c->next_seq = th_seq + data_length;
+
-+ skb = np->rx_skbs[id];
-+ mfn = gnttab_end_foreign_transfer_ref(ref);
-+ gnttab_release_grant_reference(&np->gref_rx_head, ref);
-+ np->grant_rx_ref[id] = GRANT_INVALID_REF;
-+ add_id_to_freelist(np->rx_skbs, id);
++ if (++c->n_in_order_pkts < 300) {
++ /* May be in slow-start, so don't merge. */
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_slow_start);
++ return 0;
++ }
+
-+ if (0 == mfn) {
-+ struct page *page = skb_shinfo(skb)->frags[0].page;
-+ balloon_release_driver_page(page);
-+ skb_shinfo(skb)->nr_frags = 0;
-+ dev_kfree_skb(skb);
-+ noxfer++;
-+ continue;
++ if (unlikely(dont_merge)) {
++ if (c->skb)
++ netfront_accel_ssr_deliver(vnic, st, c);
++ return 0;
+ }
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* Remap the page. */
-+ struct page *page = skb_shinfo(skb)->frags[0].page;
-+ unsigned long pfn = page_to_pfn(page);
-+ void *vaddr = page_address(page);
++ if (c->skb) {
++ c->iph->tot_len = ntohs(c->iph->tot_len);
++ c->iph->tot_len += data_length;
++ c->iph->tot_len = htons(c->iph->tot_len);
++ c->th->ack_seq = th->ack_seq;
++ c->th->fin |= th->fin;
++ c->th->psh |= th->psh;
++ c->th->window = th->window;
+
-+ MULTI_update_va_mapping(mcl, (unsigned long)vaddr,
-+ pfn_pte_ma(mfn, PAGE_KERNEL),
-+ 0);
-+ mcl++;
-+ mmu->ptr = ((maddr_t)mfn << PAGE_SHIFT)
-+ | MMU_MACHPHYS_UPDATE;
-+ mmu->val = pfn;
-+ mmu++;
++ /* Remove the headers from this skb. */
++ skb_pull(skb, skb->len - data_length);
+
-+ set_phys_to_machine(pfn, mfn);
-+ }
-+ __skb_queue_tail(&free_list, skb);
-+ xfer++;
-+ }
++ /*
++ * Tack the new skb onto the head skb's frag_list.
++ * This is exactly the format that fragmented IP
++ * datagrams are reassembled into.
++ */
++ BUG_ON(skb->next != 0);
++ if ( ! skb_shinfo(c->skb)->frag_list)
++ skb_shinfo(c->skb)->frag_list = skb;
++ else
++ c->skb_tail->next = skb;
++ c->skb_tail = skb;
++ c->skb->len += skb->len;
++ c->skb->data_len += skb->len;
++ c->skb->truesize += skb->truesize;
+
-+ DPRINTK("%s: %d xfer, %d noxfer, %d unused\n",
-+ __FUNCTION__, xfer, noxfer, unused);
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_merges);
+
-+ if (xfer) {
-+ /* Some pages are no longer absent... */
-+ balloon_update_driver_allowance(-xfer);
++ /*
++ * If the next packet might push this super-packet
++ * over the limit for an IP packet, deliver it now.
++ * This is slightly conservative, but close enough.
++ */
++ if (c->skb->len +
++ (PAGE_SIZE / NETFRONT_ACCEL_BUFS_PER_PAGE)
++ > 16384)
++ netfront_accel_ssr_deliver(vnic, st, c);
+
-+ if (!xen_feature(XENFEAT_auto_translated_physmap)) {
-+ /* Do all the remapping work and M2P updates. */
-+ mcl->op = __HYPERVISOR_mmu_update;
-+ mcl->args[0] = (unsigned long)np->rx_mmu;
-+ mcl->args[1] = mmu - np->rx_mmu;
-+ mcl->args[2] = 0;
-+ mcl->args[3] = DOMID_SELF;
-+ mcl++;
-+ rc = HYPERVISOR_multicall_check(
-+ np->rx_mcl, mcl - np->rx_mcl, NULL);
-+ BUG_ON(rc);
++ return 1;
++ }
++ else {
++ c->iph = iph;
++ c->th = th;
++ c->skb = skb;
++ return 1;
+ }
+ }
+
-+ while ((skb = __skb_dequeue(&free_list)) != NULL)
-+ dev_kfree_skb(skb);
++ /* We're not yet tracking this connection. */
+
-+ spin_unlock_bh(&np->rx_lock);
++ if (dont_merge) {
++ return 0;
++ }
++
++ if (list_empty(&st->free_conns)) {
++ c = list_entry(st->conns.prev,
++ struct netfront_accel_ssr_conn,
++ link);
++ if (c->skb) {
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_too_many);
++ return 0;
++ }
++ }
++ else {
++ c = list_entry(st->free_conns.next,
++ struct netfront_accel_ssr_conn,
++ link);
++ }
++ list_del(&c->link);
++ list_add(&c->link, &st->conns);
++ c->saddr = iph->saddr;
++ c->daddr = iph->daddr;
++ c->source = th->source;
++ c->dest = th->dest;
++ c->next_seq = th_seq + data_length;
++ c->n_in_order_pkts = 0;
++ BUG_ON(c->skb != NULL);
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.ssr_new_stream);
++ return 0;
+}
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_ssr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_ssr.h 2008-05-19 00:33:48.622948803 +0300
+@@ -0,0 +1,88 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static void netif_release_rx_bufs_copy(struct netfront_info *np)
-+{
-+ struct sk_buff *skb;
-+ int i, ref;
-+ int busy = 0, inuse = 0;
++#ifndef NETFRONT_ACCEL_SSR_H
++#define NETFRONT_ACCEL_SSR_H
+
-+ spin_lock_bh(&np->rx_lock);
++#include <linux/skbuff.h>
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/list.h>
+
-+ for (i = 0; i < NET_RX_RING_SIZE; i++) {
-+ ref = np->grant_rx_ref[i];
++#include "accel.h"
+
-+ if (ref == GRANT_INVALID_REF)
-+ continue;
++/** State for Soft Segment Reassembly (SSR). */
+
-+ inuse++;
++struct netfront_accel_ssr_conn {
++ struct list_head link;
+
-+ skb = np->rx_skbs[i];
++ unsigned saddr, daddr;
++ unsigned short source, dest;
+
-+ if (!gnttab_end_foreign_access_ref(ref))
-+ {
-+ busy++;
-+ continue;
-+ }
++ /** Number of in-order packets we've seen with payload. */
++ unsigned n_in_order_pkts;
+
-+ gnttab_release_grant_reference(&np->gref_rx_head, ref);
-+ np->grant_rx_ref[i] = GRANT_INVALID_REF;
-+ add_id_to_freelist(np->rx_skbs, i);
++ /** Next in-order sequence number. */
++ unsigned next_seq;
+
-+ skb_shinfo(skb)->nr_frags = 0;
-+ dev_kfree_skb(skb);
-+ }
++ /** Time we last saw a packet on this connection. */
++ unsigned long last_pkt_jiffies;
+
-+ if (busy)
-+ DPRINTK("%s: Unable to release %d of %d inuse grant references out of %ld total.\n",
-+ __FUNCTION__, busy, inuse, NET_RX_RING_SIZE);
++ /** The SKB we are currently holding. If NULL, then all following
++ * fields are undefined.
++ */
++ struct sk_buff *skb;
+
-+ spin_unlock_bh(&np->rx_lock);
-+}
++ /** The tail of the frag_list of SKBs we're holding. Only valid
++ * after at least one merge.
++ */
++ struct sk_buff *skb_tail;
+
-+static int network_close(struct net_device *dev)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
-+ netif_stop_queue(np->netdev);
-+ return 0;
-+}
++ /** The IP header of the skb we are holding. */
++ struct iphdr *iph;
++
++ /** The TCP header of the skb we are holding. */
++ struct tcphdr *th;
++};
+
++extern void netfront_accel_ssr_init(struct netfront_accel_ssr_state *st);
++extern void netfront_accel_ssr_fini(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st);
+
-+static struct net_device_stats *network_get_stats(struct net_device *dev)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
++extern void
++__netfront_accel_ssr_end_of_burst(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st);
+
-+ netfront_accelerator_call_get_stats(np, dev);
-+ return &np->stats;
++extern int netfront_accel_ssr_skb(netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st,
++ struct sk_buff *skb);
++
++static inline void
++netfront_accel_ssr_end_of_burst (netfront_accel_vnic *vnic,
++ struct netfront_accel_ssr_state *st) {
++ if ( ! list_empty(&st->conns) )
++ __netfront_accel_ssr_end_of_burst(vnic, st);
+}
+
-+static int xennet_change_mtu(struct net_device *dev, int mtu)
-+{
-+ int max = xennet_can_sg(dev) ? 65535 - ETH_HLEN : ETH_DATA_LEN;
++#endif /* NETFRONT_ACCEL_SSR_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_tso.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_tso.c 2008-05-19 00:33:48.622948803 +0300
+@@ -0,0 +1,511 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (mtu > max)
-+ return -EINVAL;
-+ dev->mtu = mtu;
-+ return 0;
-+}
++#include <linux/pci.h>
++#include <linux/tcp.h>
++#include <linux/ip.h>
++#include <linux/in.h>
++#include <linux/if_ether.h>
+
-+static int xennet_set_sg(struct net_device *dev, u32 data)
-+{
-+ if (data) {
-+ struct netfront_info *np = netdev_priv(dev);
-+ int val;
++#include "accel.h"
++#include "accel_util.h"
+
-+ if (xenbus_scanf(XBT_NIL, np->xbdev->otherend, "feature-sg",
-+ "%d", &val) < 0)
-+ val = 0;
-+ if (!val)
-+ return -ENOSYS;
-+ } else if (dev->mtu > ETH_DATA_LEN)
-+ dev->mtu = ETH_DATA_LEN;
++#include "accel_tso.h"
+
-+ return ethtool_op_set_sg(dev, data);
-+}
++#define PTR_DIFF(p1, p2) ((u8*)(p1) - (u8*)(p2))
++#define ETH_HDR_LEN(skb) ((skb)->nh.raw - (skb)->data)
++#define SKB_TCP_OFF(skb) PTR_DIFF ((skb)->h.th, (skb)->data)
++#define SKB_IP_OFF(skb) PTR_DIFF ((skb)->nh.iph, (skb)->data)
+
-+static int xennet_set_tso(struct net_device *dev, u32 data)
-+{
-+ if (data) {
-+ struct netfront_info *np = netdev_priv(dev);
-+ int val;
++/*
++ * Set a maximum number of buffers in each output packet to make life
++ * a little simpler - if this is reached it will just move on to
++ * another packet
++ */
++#define ACCEL_TSO_MAX_BUFFERS (6)
+
-+ if (xenbus_scanf(XBT_NIL, np->xbdev->otherend,
-+ "feature-gso-tcpv4", "%d", &val) < 0)
-+ val = 0;
-+ if (!val)
-+ return -ENOSYS;
-+ }
++/** TSO State.
++ *
++ * The state used during segmentation. It is put into this data structure
++ * just to make it easy to pass into inline functions.
++ */
++struct netfront_accel_tso_state {
++ /** bytes of data we've yet to segment */
++ unsigned remaining_len;
+
-+ return ethtool_op_set_tso(dev, data);
-+}
++ /** current sequence number */
++ unsigned seqnum;
+
-+static void xennet_set_features(struct net_device *dev)
-+{
-+ dev_disable_gso_features(dev);
-+ xennet_set_sg(dev, 0);
++ /** remaining space in current packet */
++ unsigned packet_space;
+
-+ /* We need checksum offload to enable scatter/gather and TSO. */
-+ if (!(dev->features & NETIF_F_IP_CSUM))
-+ return;
++ /** List of packets to be output, containing the buffers and
++ * iovecs to describe each packet
++ */
++ struct netfront_accel_tso_output_packet *output_packets;
+
-+ if (xennet_set_sg(dev, 1))
-+ return;
++ /** Total number of buffers in output_packets */
++ unsigned buffers;
+
-+ /* Before 2.6.9 TSO seems to be unreliable so do not enable it
-+ * on older kernels.
++ /** Total number of packets in output_packets */
++ unsigned packets;
++
++ /** Input Fragment Cursor.
++ *
++ * Where we are in the current fragment of the incoming SKB. These
++ * values get updated in place when we split a fragment over
++ * multiple packets.
+ */
-+ if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,9))
-+ xennet_set_tso(dev, 1);
++ struct {
++ /** address of current position */
++ void *addr;
++ /** remaining length */
++ unsigned int len;
++ } ifc; /* == ifc Input Fragment Cursor */
++
++ /** Parameters.
++ *
++ * These values are set once at the start of the TSO send and do
++ * not get changed as the routine progresses.
++ */
++ struct {
++ /* the number of bytes of header */
++ unsigned int header_length;
++
++ /* The number of bytes to put in each outgoing segment. */
++ int full_packet_size;
++
++ /* Current IP ID, host endian. */
++ unsigned ip_id;
++
++ /* Max size of each output packet payload */
++ int gso_size;
++ } p;
++};
++
++
++/**
++ * Verify that our various assumptions about sk_buffs and the conditions
++ * under which TSO will be attempted hold true.
++ *
++ * @v skb The sk_buff to check.
++ */
++static inline void tso_check_safe(struct sk_buff *skb) {
++ EPRINTK_ON(skb->protocol != htons (ETH_P_IP));
++ EPRINTK_ON(((struct ethhdr*) skb->data)->h_proto != htons (ETH_P_IP));
++ EPRINTK_ON(skb->nh.iph->protocol != IPPROTO_TCP);
++ EPRINTK_ON((SKB_TCP_OFF(skb)
++ + (skb->h.th->doff << 2u)) > skb_headlen(skb));
+}
+
-+static int network_connect(struct net_device *dev)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
-+ int i, requeue_idx, err;
-+ struct sk_buff *skb;
-+ grant_ref_t ref;
-+ netif_rx_request_t *req;
-+ unsigned int feature_rx_copy, feature_rx_flip;
+
-+ err = xenbus_scanf(XBT_NIL, np->xbdev->otherend,
-+ "feature-rx-copy", "%u", &feature_rx_copy);
-+ if (err != 1)
-+ feature_rx_copy = 0;
-+ err = xenbus_scanf(XBT_NIL, np->xbdev->otherend,
-+ "feature-rx-flip", "%u", &feature_rx_flip);
-+ if (err != 1)
-+ feature_rx_flip = 1;
++
++/** Parse the SKB header and initialise state. */
++static inline void tso_start(struct netfront_accel_tso_state *st,
++ struct sk_buff *skb) {
+
+ /*
-+ * Copy packets on receive path if:
-+ * (a) This was requested by user, and the backend supports it; or
-+ * (b) Flipping was requested, but this is unsupported by the backend.
-+ */
-+ np->copying_receiver = ((MODPARM_rx_copy && feature_rx_copy) ||
-+ (MODPARM_rx_flip && !feature_rx_flip));
++ * All ethernet/IP/TCP headers combined size is TCP header size
++ * plus offset of TCP header relative to start of packet.
++ */
++ st->p.header_length = (skb->h.th->doff << 2u) + SKB_TCP_OFF(skb);
++ st->p.full_packet_size = (st->p.header_length
++ + skb_shinfo(skb)->gso_size);
++ st->p.gso_size = skb_shinfo(skb)->gso_size;
+
-+ err = talk_to_backend(np->xbdev, np);
-+ if (err)
-+ return err;
++ st->p.ip_id = htons(skb->nh.iph->id);
++ st->seqnum = ntohl(skb->h.th->seq);
+
-+ xennet_set_features(dev);
++ EPRINTK_ON(skb->h.th->urg);
++ EPRINTK_ON(skb->h.th->syn);
++ EPRINTK_ON(skb->h.th->rst);
+
-+ DPRINTK("device %s has %sing receive path.\n",
-+ dev->name, np->copying_receiver ? "copy" : "flipp");
++ st->remaining_len = skb->len - st->p.header_length;
+
-+ spin_lock_bh(&np->rx_lock);
-+ spin_lock_irq(&np->tx_lock);
++ st->output_packets = NULL;
++ st->buffers = 0;
++ st->packets = 0;
+
-+ /*
-+ * Recovery procedure:
-+ * NB. Freelist index entries are always going to be less than
-+ * PAGE_OFFSET, whereas pointers to skbs will always be equal or
-+ * greater than PAGE_OFFSET: we use this property to distinguish
-+ * them.
-+ */
++ VPRINTK("Starting new TSO: hl %d ps %d gso %d seq %x len %d\n",
++ st->p.header_length, st->p.full_packet_size, st->p.gso_size,
++ st->seqnum, skb->len);
++}
+
-+ /* Step 1: Discard all pending TX packet fragments. */
-+ netif_release_tx_bufs(np);
++/**
++ * Add another NIC mapped buffer onto an output packet
++ */
++static inline int tso_start_new_buffer(netfront_accel_vnic *vnic,
++ struct netfront_accel_tso_state *st,
++ int first)
++{
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct netfront_accel_pkt_desc *buf;
+
-+ /* Step 2: Rebuild the RX buffer freelist and the RX ring itself. */
-+ for (requeue_idx = 0, i = 0; i < NET_RX_RING_SIZE; i++) {
-+ if (!np->rx_skbs[i])
-+ continue;
++ /* Get a mapped packet buffer */
++ buf = netfront_accel_buf_get(vnic->tx_bufs);
++ if (buf == NULL) {
++ DPRINTK("%s: No buffer for TX\n", __FUNCTION__);
++ return -1;
++ }
+
-+ skb = np->rx_skbs[requeue_idx] = xennet_get_rx_skb(np, i);
-+ ref = np->grant_rx_ref[requeue_idx] = xennet_get_rx_ref(np, i);
-+ req = RING_GET_REQUEST(&np->rx, requeue_idx);
++ /* Store a bit of meta-data at the end */
++ tso_buf =(struct netfront_accel_tso_buffer *)
++ (buf->pkt_kva + NETFRONT_ACCEL_TSO_BUF_LENGTH
++ + sizeof(struct netfront_accel_tso_output_packet));
+
-+ if (!np->copying_receiver) {
-+ gnttab_grant_foreign_transfer_ref(
-+ ref, np->xbdev->otherend_id,
-+ page_to_pfn(skb_shinfo(skb)->frags->page));
-+ } else {
-+ gnttab_grant_foreign_access_ref(
-+ ref, np->xbdev->otherend_id,
-+ pfn_to_mfn(page_to_pfn(skb_shinfo(skb)->
-+ frags->page)),
-+ 0);
-+ }
-+ req->gref = ref;
-+ req->id = requeue_idx;
++ tso_buf->buf = buf;
+
-+ requeue_idx++;
++ tso_buf->length = 0;
++
++ if (first) {
++ struct netfront_accel_tso_output_packet *output_packet
++ = (struct netfront_accel_tso_output_packet *)
++ (buf->pkt_kva + NETFRONT_ACCEL_TSO_BUF_LENGTH);
++ output_packet->next = st->output_packets;
++ st->output_packets = output_packet;
++ tso_buf->next = NULL;
++ st->output_packets->tso_bufs = tso_buf;
++ st->output_packets->tso_bufs_len = 1;
++ } else {
++ tso_buf->next = st->output_packets->tso_bufs;
++ st->output_packets->tso_bufs = tso_buf;
++ st->output_packets->tso_bufs_len ++;
+ }
+
-+ np->rx.req_prod_pvt = requeue_idx;
++ BUG_ON(st->output_packets->tso_bufs_len > ACCEL_TSO_MAX_BUFFERS);
++
++ st->buffers ++;
+
+ /*
-+ * Step 3: All public and private state should now be sane. Get
-+ * ready to start sending and receiving packets and give the driver
-+ * domain a kick because we've probably just requeued some
-+ * packets.
++ * Store the context, set to NULL, last packet buffer will get
++ * non-NULL later
+ */
-+ netfront_carrier_on(np);
-+ notify_remote_via_irq(np->irq);
-+ network_tx_buf_gc(dev);
-+ network_alloc_rx_buffers(dev);
-+
-+ spin_unlock_irq(&np->tx_lock);
-+ spin_unlock_bh(&np->rx_lock);
++ tso_buf->buf->skb = NULL;
+
+ return 0;
+}
+
-+static void netif_uninit(struct net_device *dev)
-+{
-+ struct netfront_info *np = netdev_priv(dev);
-+ netif_release_tx_bufs(np);
-+ if (np->copying_receiver)
-+ netif_release_rx_bufs_copy(np);
-+ else
-+ netif_release_rx_bufs_flip(np);
-+ gnttab_free_grant_references(np->gref_tx_head);
-+ gnttab_free_grant_references(np->gref_rx_head);
-+}
+
-+static struct ethtool_ops network_ethtool_ops =
-+{
-+ .get_tx_csum = ethtool_op_get_tx_csum,
-+ .set_tx_csum = ethtool_op_set_tx_csum,
-+ .get_sg = ethtool_op_get_sg,
-+ .set_sg = xennet_set_sg,
-+#if HAVE_TSO
-+ .get_tso = ethtool_op_get_tso,
-+ .set_tso = xennet_set_tso,
-+#endif
-+ .get_link = ethtool_op_get_link,
-+};
++/* Generate a new header, and prepare for the new packet.
++ *
++ * @v vnic VNIC
++ * @v skb Socket buffer
++ * @v st TSO state
++ * @ret rc 0 on success, or -1 if failed to alloc header
++ */
+
-+#ifdef CONFIG_SYSFS
-+static ssize_t show_rxbuf_min(struct class_device *cd, char *buf)
++static inline
++int tso_start_new_packet(netfront_accel_vnic *vnic,
++ struct sk_buff *skb,
++ struct netfront_accel_tso_state *st)
+{
-+ struct net_device *netdev = container_of(cd, struct net_device,
-+ class_dev);
-+ struct netfront_info *info = netdev_priv(netdev);
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct iphdr *tsoh_iph;
++ struct tcphdr *tsoh_th;
++ unsigned ip_length;
+
-+ return sprintf(buf, "%u\n", info->rx_min_target);
-+}
++ if (tso_start_new_buffer(vnic, st, 1) < 0) {
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_busy++);
++ return -1;
++ }
+
-+static ssize_t store_rxbuf_min(struct class_device *cd,
-+ const char *buf, size_t len)
-+{
-+ struct net_device *netdev = container_of(cd, struct net_device,
-+ class_dev);
-+ struct netfront_info *np = netdev_priv(netdev);
-+ char *endp;
-+ unsigned long target;
++ /* This has been set up by tso_start_new_buffer() */
++ tso_buf = st->output_packets->tso_bufs;
+
-+ if (!capable(CAP_NET_ADMIN))
-+ return -EPERM;
++ /* Copy in the header */
++ memcpy(tso_buf->buf->pkt_kva, skb->data, st->p.header_length);
++ tso_buf->length = st->p.header_length;
+
-+ target = simple_strtoul(buf, &endp, 0);
-+ if (endp == buf)
-+ return -EBADMSG;
++ tsoh_th = (struct tcphdr*)
++ (tso_buf->buf->pkt_kva + SKB_TCP_OFF(skb));
++ tsoh_iph = (struct iphdr*)
++ (tso_buf->buf->pkt_kva + SKB_IP_OFF(skb));
+
-+ if (target < RX_MIN_TARGET)
-+ target = RX_MIN_TARGET;
-+ if (target > RX_MAX_TARGET)
-+ target = RX_MAX_TARGET;
++ /* Set to zero to encourage falcon to fill these in */
++ tsoh_th->check = 0;
++ tsoh_iph->check = 0;
+
-+ spin_lock_bh(&np->rx_lock);
-+ if (target > np->rx_max_target)
-+ np->rx_max_target = target;
-+ np->rx_min_target = target;
-+ if (target > np->rx_target)
-+ np->rx_target = target;
++ tsoh_th->seq = htonl(st->seqnum);
++ st->seqnum += st->p.gso_size;
+
-+ network_alloc_rx_buffers(netdev);
++ if (st->remaining_len > st->p.gso_size) {
++ /* This packet will not finish the TSO burst. */
++ ip_length = st->p.full_packet_size - ETH_HDR_LEN(skb);
++ tsoh_th->fin = 0;
++ tsoh_th->psh = 0;
++ } else {
++ /* This packet will be the last in the TSO burst. */
++ ip_length = (st->p.header_length - ETH_HDR_LEN(skb)
++ + st->remaining_len);
++ tsoh_th->fin = skb->h.th->fin;
++ tsoh_th->psh = skb->h.th->psh;
++ }
+
-+ spin_unlock_bh(&np->rx_lock);
-+ return len;
-+}
++ tsoh_iph->tot_len = htons(ip_length);
+
-+static ssize_t show_rxbuf_max(struct class_device *cd, char *buf)
-+{
-+ struct net_device *netdev = container_of(cd, struct net_device,
-+ class_dev);
-+ struct netfront_info *info = netdev_priv(netdev);
++ /* Linux leaves suitable gaps in the IP ID space for us to fill. */
++ tsoh_iph->id = st->p.ip_id++;
++ tsoh_iph->id = htons(tsoh_iph->id);
+
-+ return sprintf(buf, "%u\n", info->rx_max_target);
++ st->packet_space = st->p.gso_size;
++
++ st->packets++;
++
++ return 0;
+}
+
-+static ssize_t store_rxbuf_max(struct class_device *cd,
-+ const char *buf, size_t len)
++
++
++static inline void tso_get_fragment(struct netfront_accel_tso_state *st,
++ int len, void *addr)
+{
-+ struct net_device *netdev = container_of(cd, struct net_device,
-+ class_dev);
-+ struct netfront_info *np = netdev_priv(netdev);
-+ char *endp;
-+ unsigned long target;
++ st->ifc.len = len;
++ st->ifc.addr = addr;
++ return;
++}
+
-+ if (!capable(CAP_NET_ADMIN))
-+ return -EPERM;
+
-+ target = simple_strtoul(buf, &endp, 0);
-+ if (endp == buf)
-+ return -EBADMSG;
++static inline void tso_unwind(netfront_accel_vnic *vnic,
++ struct netfront_accel_tso_state *st)
++{
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct netfront_accel_tso_output_packet *output_packet;
+
-+ if (target < RX_MIN_TARGET)
-+ target = RX_MIN_TARGET;
-+ if (target > RX_MAX_TARGET)
-+ target = RX_MAX_TARGET;
++ DPRINTK("%s\n", __FUNCTION__);
+
-+ spin_lock_bh(&np->rx_lock);
-+ if (target < np->rx_min_target)
-+ np->rx_min_target = target;
-+ np->rx_max_target = target;
-+ if (target < np->rx_target)
-+ np->rx_target = target;
++ while (st->output_packets != NULL) {
++ output_packet = st->output_packets;
++ st->output_packets = output_packet->next;
++ while (output_packet->tso_bufs != NULL) {
++ tso_buf = output_packet->tso_bufs;
++ output_packet->tso_bufs = tso_buf->next;
+
-+ network_alloc_rx_buffers(netdev);
++ st->buffers --;
++ output_packet->tso_bufs_len --;
+
-+ spin_unlock_bh(&np->rx_lock);
-+ return len;
++ netfront_accel_buf_put(vnic->tx_bufs,
++ tso_buf->buf->buf_id);
++ }
++ }
++ BUG_ON(st->buffers != 0);
+}
+
-+static ssize_t show_rxbuf_cur(struct class_device *cd, char *buf)
++
++
++static inline
++void tso_fill_packet_with_fragment(netfront_accel_vnic *vnic,
++ struct netfront_accel_tso_state *st)
+{
-+ struct net_device *netdev = container_of(cd, struct net_device,
-+ class_dev);
-+ struct netfront_info *info = netdev_priv(netdev);
++ struct netfront_accel_tso_buffer *tso_buf;
++ int n, space;
+
-+ return sprintf(buf, "%u\n", info->rx_target);
-+}
++ BUG_ON(st->output_packets == NULL);
++ BUG_ON(st->output_packets->tso_bufs == NULL);
+
-+static const struct class_device_attribute xennet_attrs[] = {
-+ __ATTR(rxbuf_min, S_IRUGO|S_IWUSR, show_rxbuf_min, store_rxbuf_min),
-+ __ATTR(rxbuf_max, S_IRUGO|S_IWUSR, show_rxbuf_max, store_rxbuf_max),
-+ __ATTR(rxbuf_cur, S_IRUGO, show_rxbuf_cur, NULL),
-+};
++ tso_buf = st->output_packets->tso_bufs;
+
-+static int xennet_sysfs_addif(struct net_device *netdev)
-+{
-+ int i;
-+ int error = 0;
++ if (st->ifc.len == 0) return;
++ if (st->packet_space == 0) return;
++ if (tso_buf->length == NETFRONT_ACCEL_TSO_BUF_LENGTH) return;
+
-+ for (i = 0; i < ARRAY_SIZE(xennet_attrs); i++) {
-+ error = class_device_create_file(&netdev->class_dev,
-+ &xennet_attrs[i]);
-+ if (error)
-+ goto fail;
-+ }
-+ return 0;
++ n = min(st->ifc.len, st->packet_space);
+
-+ fail:
-+ while (--i >= 0)
-+ class_device_remove_file(&netdev->class_dev,
-+ &xennet_attrs[i]);
-+ return error;
-+}
++ space = NETFRONT_ACCEL_TSO_BUF_LENGTH - tso_buf->length;
++ n = min(n, space);
+
-+static void xennet_sysfs_delif(struct net_device *netdev)
-+{
-+ int i;
++ st->packet_space -= n;
++ st->remaining_len -= n;
++ st->ifc.len -= n;
+
-+ for (i = 0; i < ARRAY_SIZE(xennet_attrs); i++) {
-+ class_device_remove_file(&netdev->class_dev,
-+ &xennet_attrs[i]);
-+ }
-+}
++ memcpy(tso_buf->buf->pkt_kva + tso_buf->length, st->ifc.addr, n);
+
-+#endif /* CONFIG_SYSFS */
++ tso_buf->length += n;
+
++ BUG_ON(tso_buf->length > NETFRONT_ACCEL_TSO_BUF_LENGTH);
+
-+/*
-+ * Nothing to do here. Virtual interface is point-to-point and the
-+ * physical interface is probably promiscuous anyway.
-+ */
-+static void network_set_multicast_list(struct net_device *dev)
-+{
++ st->ifc.addr += n;
++
++ return;
+}
+
-+static struct net_device * __devinit create_netdev(struct xenbus_device *dev)
++
++int netfront_accel_enqueue_skb_tso(netfront_accel_vnic *vnic,
++ struct sk_buff *skb)
+{
-+ int i, err = 0;
-+ struct net_device *netdev = NULL;
-+ struct netfront_info *np = NULL;
++ struct netfront_accel_tso_state state;
++ struct netfront_accel_tso_buffer *tso_buf = NULL;
++ struct netfront_accel_tso_output_packet *reversed_list = NULL;
++ struct netfront_accel_tso_output_packet *tmp_pkt;
++ ef_iovec iovecs[ACCEL_TSO_MAX_BUFFERS];
++ int frag_i, rc, dma_id;
++ skb_frag_t *f;
+
-+ netdev = alloc_etherdev(sizeof(struct netfront_info));
-+ if (!netdev) {
-+ printk(KERN_WARNING "%s> alloc_etherdev failed.\n",
-+ __FUNCTION__);
-+ return ERR_PTR(-ENOMEM);
++ tso_check_safe(skb);
++
++ if (skb->ip_summed != CHECKSUM_HW)
++ EPRINTK("Trying to TSO send a packet without HW checksum\n");
++
++ tso_start(&state, skb);
++
++ /*
++ * Setup the first payload fragment. If the skb header area
++ * contains exactly the headers and all payload is in the frag
++ * list things are little simpler
++ */
++ if (skb_headlen(skb) == state.p.header_length) {
++ /* Grab the first payload fragment. */
++ BUG_ON(skb_shinfo(skb)->nr_frags < 1);
++ frag_i = 0;
++ f = &skb_shinfo(skb)->frags[frag_i];
++ tso_get_fragment(&state, f->size,
++ page_address(f->page) + f->page_offset);
++ } else {
++ int hl = state.p.header_length;
++ tso_get_fragment(&state, skb_headlen(skb) - hl,
++ skb->data + hl);
++ frag_i = -1;
+ }
+
-+ np = netdev_priv(netdev);
-+ np->xbdev = dev;
++ if (tso_start_new_packet(vnic, skb, &state) < 0) {
++ DPRINTK("%s: out of first start-packet memory\n",
++ __FUNCTION__);
++ goto unwind;
++ }
+
-+ spin_lock_init(&np->tx_lock);
-+ spin_lock_init(&np->rx_lock);
++ while (1) {
++ tso_fill_packet_with_fragment(vnic, &state);
++
++ /* Move onto the next fragment? */
++ if (state.ifc.len == 0) {
++ if (++frag_i >= skb_shinfo(skb)->nr_frags)
++ /* End of payload reached. */
++ break;
++ f = &skb_shinfo(skb)->frags[frag_i];
++ tso_get_fragment(&state, f->size,
++ page_address(f->page) +
++ f->page_offset);
++ }
+
-+ init_accelerator_vif(np, dev);
++ /* Start a new buffer? */
++ if ((state.output_packets->tso_bufs->length ==
++ NETFRONT_ACCEL_TSO_BUF_LENGTH) &&
++ tso_start_new_buffer(vnic, &state, 0)) {
++ DPRINTK("%s: out of start-buffer memory\n",
++ __FUNCTION__);
++ goto unwind;
++ }
+
-+ skb_queue_head_init(&np->rx_batch);
-+ np->rx_target = RX_DFL_MIN_TARGET;
-+ np->rx_min_target = RX_DFL_MIN_TARGET;
-+ np->rx_max_target = RX_MAX_TARGET;
++ /* Start at new packet? */
++ if ((state.packet_space == 0 ||
++ ((state.output_packets->tso_bufs_len >=
++ ACCEL_TSO_MAX_BUFFERS) &&
++ (state.output_packets->tso_bufs->length >=
++ NETFRONT_ACCEL_TSO_BUF_LENGTH))) &&
++ tso_start_new_packet(vnic, skb, &state) < 0) {
++ DPRINTK("%s: out of start-packet memory\n",
++ __FUNCTION__);
++ goto unwind;
++ }
+
-+ init_timer(&np->rx_refill_timer);
-+ np->rx_refill_timer.data = (unsigned long)netdev;
-+ np->rx_refill_timer.function = rx_refill_timeout;
++ }
+
-+ /* Initialise {tx,rx}_skbs as a free chain containing every entry. */
-+ for (i = 0; i <= NET_TX_RING_SIZE; i++) {
-+ np->tx_skbs[i] = (void *)((unsigned long) i+1);
-+ np->grant_tx_ref[i] = GRANT_INVALID_REF;
++ /* Check for space */
++ if (ef_vi_transmit_space(&vnic->vi) < state.buffers) {
++ DPRINTK("%s: Not enough TX space (%d)\n",
++ __FUNCTION__, state.buffers);
++ goto unwind;
+ }
+
-+ for (i = 0; i < NET_RX_RING_SIZE; i++) {
-+ np->rx_skbs[i] = NULL;
-+ np->grant_rx_ref[i] = GRANT_INVALID_REF;
++ /*
++ * Store the skb context in the most recent buffer (i.e. the
++ * last buffer that will be sent)
++ */
++ state.output_packets->tso_bufs->buf->skb = skb;
++
++ /* Reverse the list of packets as we construct it on a stack */
++ while (state.output_packets != NULL) {
++ tmp_pkt = state.output_packets;
++ state.output_packets = tmp_pkt->next;
++ tmp_pkt->next = reversed_list;
++ reversed_list = tmp_pkt;
+ }
+
-+ /* A grant for every tx ring slot */
-+ if (gnttab_alloc_grant_references(TX_MAX_TARGET,
-+ &np->gref_tx_head) < 0) {
-+ printk(KERN_ALERT "#### netfront can't alloc tx grant refs\n");
-+ err = -ENOMEM;
-+ goto exit;
++ /* Pass off to hardware */
++ while (reversed_list != NULL) {
++ tmp_pkt = reversed_list;
++ reversed_list = tmp_pkt->next;
++
++ BUG_ON(tmp_pkt->tso_bufs_len > ACCEL_TSO_MAX_BUFFERS);
++ BUG_ON(tmp_pkt->tso_bufs_len == 0);
++
++ dma_id = tmp_pkt->tso_bufs->buf->buf_id;
++
++ /*
++ * Make an iovec of the buffers in the list, reversing
++ * the buffers as we go as they are constructed on a
++ * stack
++ */
++ tso_buf = tmp_pkt->tso_bufs;
++ for (frag_i = tmp_pkt->tso_bufs_len - 1;
++ frag_i >= 0;
++ frag_i--) {
++ iovecs[frag_i].iov_base = tso_buf->buf->pkt_buff_addr;
++ iovecs[frag_i].iov_len = tso_buf->length;
++ tso_buf = tso_buf->next;
++ }
++
++ rc = ef_vi_transmitv(&vnic->vi, iovecs, tmp_pkt->tso_bufs_len,
++ dma_id);
++ /*
++ * We checked for space already, so it really should
++ * succeed
++ */
++ BUG_ON(rc != 0);
+ }
-+ /* A grant for every rx ring slot */
-+ if (gnttab_alloc_grant_references(RX_MAX_TARGET,
-+ &np->gref_rx_head) < 0) {
-+ printk(KERN_ALERT "#### netfront can't alloc rx grant refs\n");
-+ err = -ENOMEM;
-+ goto exit_free_tx;
++
++ /* Track number of tx fastpath stats */
++ vnic->netdev_stats.fastpath_tx_bytes += skb->len;
++ vnic->netdev_stats.fastpath_tx_pkts += state.packets;
++#if NETFRONT_ACCEL_STATS
++ {
++ unsigned n;
++ n = vnic->netdev_stats.fastpath_tx_pkts -
++ vnic->stats.fastpath_tx_completions;
++ if (n > vnic->stats.fastpath_tx_pending_max)
++ vnic->stats.fastpath_tx_pending_max = n;
+ }
++#endif
+
-+ netdev->open = network_open;
-+ netdev->hard_start_xmit = network_start_xmit;
-+ netdev->stop = network_close;
-+ netdev->get_stats = network_get_stats;
-+ netdev->poll = netif_poll;
-+ netdev->set_multicast_list = network_set_multicast_list;
-+ netdev->uninit = netif_uninit;
-+ netdev->change_mtu = xennet_change_mtu;
-+ netdev->weight = 64;
-+ netdev->features = NETIF_F_IP_CSUM;
++ return NETFRONT_ACCEL_STATUS_GOOD;
++
++ unwind:
++ tso_unwind(vnic, &state);
+
-+ SET_ETHTOOL_OPS(netdev, &network_ethtool_ops);
-+ SET_MODULE_OWNER(netdev);
-+ SET_NETDEV_DEV(netdev, &dev->dev);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_busy++);
+
-+ np->netdev = netdev;
++ return NETFRONT_ACCEL_STATUS_BUSY;
++}
+
-+ netfront_carrier_off(np);
+
-+ return netdev;
+
-+ exit_free_tx:
-+ gnttab_free_grant_references(np->gref_tx_head);
-+ exit:
-+ free_netdev(netdev);
-+ return ERR_PTR(err);
-+}
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_tso.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_tso.h 2008-05-19 00:33:48.622948803 +0300
+@@ -0,0 +1,57 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef NETFRONT_ACCEL_TSO_H
++#define NETFRONT_ACCEL_TSO_H
++
++#include "accel_bufs.h"
++
++/* Track the buffers used in each output packet */
++struct netfront_accel_tso_buffer {
++ struct netfront_accel_tso_buffer *next;
++ struct netfront_accel_pkt_desc *buf;
++ unsigned length;
++};
++
++/* Track the output packets formed from each input packet */
++struct netfront_accel_tso_output_packet {
++ struct netfront_accel_tso_output_packet *next;
++ struct netfront_accel_tso_buffer *tso_bufs;
++ unsigned tso_bufs_len;
++};
++
+
-+#ifdef CONFIG_INET
+/*
-+ * We use this notifier to send out a fake ARP reply to reset switches and
-+ * router ARP caches when an IP interface is brought up on a VIF.
++ * Max available space in a buffer for data once meta-data has taken
++ * its place
+ */
-+static int
-+inetdev_notify(struct notifier_block *this, unsigned long event, void *ptr)
-+{
-+ struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
-+ struct net_device *dev = ifa->ifa_dev->dev;
++#define NETFRONT_ACCEL_TSO_BUF_LENGTH \
++ ((PAGE_SIZE / NETFRONT_ACCEL_BUFS_PER_PAGE) \
++ - sizeof(struct netfront_accel_tso_buffer) \
++ - sizeof(struct netfront_accel_tso_output_packet))
+
-+ /* UP event and is it one of our devices? */
-+ if (event == NETDEV_UP && dev->open == network_open)
-+ send_fake_arp(dev);
++int netfront_accel_enqueue_skb_tso(netfront_accel_vnic *vnic,
++ struct sk_buff *skb);
+
-+ return NOTIFY_DONE;
-+}
++#endif /* NETFRONT_ACCEL_TSO_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_vi.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_vi.c 2008-05-19 00:33:48.626949033 +0300
+@@ -0,0 +1,1194 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+static struct notifier_block notifier_inetdev = {
-+ .notifier_call = inetdev_notify,
-+ .next = NULL,
-+ .priority = 0
-+};
-+#endif
++#include <linux/if_ether.h>
++#include <linux/ip.h>
++#include <net/checksum.h>
++#include <asm/io.h>
+
++#include "accel.h"
++#include "accel_util.h"
++#include "accel_bufs.h"
++#include "accel_tso.h"
++#include "accel_ssr.h"
++#include "netfront.h"
+
-+static void netif_disconnect_backend(struct netfront_info *info)
-+{
-+ /* Stop old i/f to prevent errors whilst we rebuild the state. */
-+ spin_lock_bh(&info->rx_lock);
-+ spin_lock_irq(&info->tx_lock);
-+ netfront_carrier_off(info);
-+ spin_unlock_irq(&info->tx_lock);
-+ spin_unlock_bh(&info->rx_lock);
++#include "etherfabric/ef_vi.h"
+
-+ if (info->irq)
-+ unbind_from_irqhandler(info->irq, info->netdev);
-+ info->irq = 0;
++/*
++ * Max available space in a buffer for data once meta-data has taken
++ * its place
++ */
++#define NETFRONT_ACCEL_TX_BUF_LENGTH \
++ ((PAGE_SIZE / NETFRONT_ACCEL_BUFS_PER_PAGE) \
++ - sizeof(struct netfront_accel_tso_buffer))
+
-+ end_access(info->tx_ring_ref, info->tx.sring);
-+ end_access(info->rx_ring_ref, info->rx.sring);
-+ info->tx_ring_ref = GRANT_INVALID_REF;
-+ info->rx_ring_ref = GRANT_INVALID_REF;
-+ info->tx.sring = NULL;
-+ info->rx.sring = NULL;
-+}
++#define ACCEL_TX_MAX_BUFFERS (6)
++#define ACCEL_VI_POLL_EVENTS (8)
+
++static
++int netfront_accel_vi_init_fini(netfront_accel_vnic *vnic,
++ struct net_accel_msg_hw *hw_msg)
++{
++ struct ef_vi_nic_type nic_type;
++ struct net_accel_hw_falcon_b *hw_info;
++ void *io_kva, *evq_base, *rx_dma_kva, *tx_dma_kva, *doorbell_kva;
++ u32 *evq_gnts;
++ u32 evq_order;
++ int vi_state_size;
++ u8 vi_data[VI_MAPPINGS_SIZE];
++
++ if (hw_msg == NULL)
++ goto fini;
++
++ /* And create the local macs table lock */
++ spin_lock_init(&vnic->table_lock);
++
++ /* Create fastpath table, initial size 8, key length 8 */
++ if (cuckoo_hash_init(&vnic->fastpath_table, 3, 8)) {
++ EPRINTK("failed to allocate fastpath table\n");
++ goto fail_cuckoo;
++ }
+
-+static void end_access(int ref, void *page)
-+{
-+ if (ref != GRANT_INVALID_REF)
-+ gnttab_end_foreign_access(ref, (unsigned long)page);
-+}
++ vnic->hw.falcon.type = hw_msg->type;
+
++ switch (hw_msg->type) {
++ case NET_ACCEL_MSG_HWTYPE_FALCON_A:
++ hw_info = &hw_msg->resources.falcon_a.common;
++ /* Need the extra rptr register page on A1 */
++ io_kva = net_accel_map_iomem_page
++ (vnic->dev, hw_msg->resources.falcon_a.evq_rptr_gnt,
++ &vnic->hw.falcon.evq_rptr_mapping);
++ if (io_kva == NULL) {
++ EPRINTK("%s: evq_rptr permission failed\n", __FUNCTION__);
++ goto evq_rptr_fail;
++ }
+
-+/* ** Driver registration ** */
++ vnic->hw.falcon.evq_rptr = io_kva +
++ (hw_info->evq_rptr & (PAGE_SIZE - 1));
++ break;
++ case NET_ACCEL_MSG_HWTYPE_FALCON_B:
++ hw_info = &hw_msg->resources.falcon_b;
++ break;
++ default:
++ goto bad_type;
++ }
+
++ /**** Event Queue ****/
+
-+static struct xenbus_device_id netfront_ids[] = {
-+ { "vif" },
-+ { "" }
-+};
-+MODULE_ALIAS("xen:vif");
++ /* Map the event queue pages */
++ evq_gnts = hw_info->evq_mem_gnts;
++ evq_order = hw_info->evq_order;
+
++ EPRINTK_ON(hw_info->evq_offs != 0);
+
-+static struct xenbus_driver netfront = {
-+ .name = "vif",
-+ .owner = THIS_MODULE,
-+ .ids = netfront_ids,
-+ .probe = netfront_probe,
-+ .remove = __devexit_p(netfront_remove),
-+ .suspend = netfront_suspend,
-+ .suspend_cancel = netfront_suspend_cancel,
-+ .resume = netfront_resume,
-+ .otherend_changed = backend_changed,
-+};
++ DPRINTK("Will map evq %d pages\n", 1 << evq_order);
+
++ evq_base =
++ net_accel_map_grants_contig(vnic->dev, evq_gnts, 1 << evq_order,
++ &vnic->evq_mapping);
++ if (evq_base == NULL) {
++ EPRINTK("%s: evq_base failed\n", __FUNCTION__);
++ goto evq_fail;
++ }
+
-+static int __init netif_init(void)
-+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ /**** Doorbells ****/
++ /* Set up the doorbell mappings. */
++ doorbell_kva =
++ net_accel_map_iomem_page(vnic->dev, hw_info->doorbell_gnt,
++ &vnic->hw.falcon.doorbell_mapping);
++ if (doorbell_kva == NULL) {
++ EPRINTK("%s: doorbell permission failed\n", __FUNCTION__);
++ goto doorbell_fail;
++ }
++ vnic->hw.falcon.doorbell = doorbell_kva;
+
-+#ifdef CONFIG_XEN
-+ if (MODPARM_rx_flip && MODPARM_rx_copy) {
-+ WPRINTK("Cannot specify both rx_copy and rx_flip.\n");
-+ return -EINVAL;
++ /* On Falcon_B we get the rptr from the doorbell page */
++ if (hw_msg->type == NET_ACCEL_MSG_HWTYPE_FALCON_B) {
++ vnic->hw.falcon.evq_rptr =
++ (u32 *)((char *)vnic->hw.falcon.doorbell
++ + hw_info->evq_rptr);
+ }
+
-+ if (!MODPARM_rx_flip && !MODPARM_rx_copy)
-+ MODPARM_rx_flip = 1; /* Default is to flip. */
-+#endif
++ /**** DMA Queue ****/
+
-+ if (is_initial_xendomain())
-+ return 0;
++ /* Set up the DMA Queues from the message. */
++ tx_dma_kva = net_accel_map_grants_contig
++ (vnic->dev, &(hw_info->txdmaq_gnt), 1,
++ &vnic->hw.falcon.txdmaq_mapping);
++ if (tx_dma_kva == NULL) {
++ EPRINTK("%s: TX dma failed\n", __FUNCTION__);
++ goto tx_dma_fail;
++ }
+
-+ netif_init_accel();
++ rx_dma_kva = net_accel_map_grants_contig
++ (vnic->dev, &(hw_info->rxdmaq_gnt), 1,
++ &vnic->hw.falcon.rxdmaq_mapping);
++ if (rx_dma_kva == NULL) {
++ EPRINTK("%s: RX dma failed\n", __FUNCTION__);
++ goto rx_dma_fail;
++ }
+
-+ IPRINTK("Initialising virtual ethernet driver.\n");
++ /* Full confession */
++ DPRINTK("Mapped H/W"
++ " Tx DMAQ grant %x -> %p\n"
++ " Rx DMAQ grant %x -> %p\n"
++ " EVQ grant %x -> %p\n",
++ hw_info->txdmaq_gnt, tx_dma_kva,
++ hw_info->rxdmaq_gnt, rx_dma_kva,
++ evq_gnts[0], evq_base
++ );
+
-+#ifdef CONFIG_INET
-+ (void)register_inetaddr_notifier(¬ifier_inetdev);
-+#endif
++ memset(vi_data, 0, sizeof(vi_data));
++
++ /* TODO BUG11305: convert efhw_arch to ef_vi_arch
++ * e.g.
++ * arch = ef_vi_arch_from_efhw_arch(hw_info->nic_arch);
++ * assert(arch >= 0);
++ * nic_type.arch = arch;
++ */
++ nic_type.arch = (unsigned char)hw_info->nic_arch;
++ nic_type.variant = (char)hw_info->nic_variant;
++ nic_type.revision = (unsigned char)hw_info->nic_revision;
++
++ ef_vi_init_mapping_evq(vi_data, nic_type, hw_info->instance,
++ 1 << (evq_order + PAGE_SHIFT), evq_base,
++ (void *)0xdeadbeef);
++
++ ef_vi_init_mapping_vi(vi_data, nic_type, hw_info->rx_capacity,
++ hw_info->tx_capacity, hw_info->instance,
++ doorbell_kva, rx_dma_kva, tx_dma_kva, 0);
+
-+ return xenbus_register_frontend(&netfront);
++ vi_state_size = ef_vi_calc_state_bytes(hw_info->rx_capacity,
++ hw_info->tx_capacity);
++ vnic->vi_state = (ef_vi_state *)kmalloc(vi_state_size, GFP_KERNEL);
++ if (vnic->vi_state == NULL) {
++ EPRINTK("%s: kmalloc for VI state failed\n", __FUNCTION__);
++ goto vi_state_fail;
++ }
++ ef_vi_init(&vnic->vi, vi_data, vnic->vi_state, &vnic->evq_state, 0);
++
++ ef_eventq_state_init(&vnic->vi);
++
++ ef_vi_state_init(&vnic->vi);
++
++ return 0;
++
++fini:
++ kfree(vnic->vi_state);
++ vnic->vi_state = NULL;
++vi_state_fail:
++ net_accel_unmap_grants_contig(vnic->dev, vnic->hw.falcon.rxdmaq_mapping);
++rx_dma_fail:
++ net_accel_unmap_grants_contig(vnic->dev, vnic->hw.falcon.txdmaq_mapping);
++tx_dma_fail:
++ net_accel_unmap_iomem_page(vnic->dev, vnic->hw.falcon.doorbell_mapping);
++ vnic->hw.falcon.doorbell = NULL;
++doorbell_fail:
++ net_accel_unmap_grants_contig(vnic->dev, vnic->evq_mapping);
++evq_fail:
++ if (vnic->hw.falcon.type == NET_ACCEL_MSG_HWTYPE_FALCON_A)
++ net_accel_unmap_iomem_page(vnic->dev,
++ vnic->hw.falcon.evq_rptr_mapping);
++ vnic->hw.falcon.evq_rptr = NULL;
++evq_rptr_fail:
++bad_type:
++ cuckoo_hash_destroy(&vnic->fastpath_table);
++fail_cuckoo:
++ return -EIO;
+}
-+module_init(netif_init);
+
+
-+static void __exit netif_exit(void)
++void netfront_accel_vi_ctor(netfront_accel_vnic *vnic)
+{
-+ if (is_initial_xendomain())
-+ return;
-+
-+#ifdef CONFIG_INET
-+ unregister_inetaddr_notifier(¬ifier_inetdev);
-+#endif
++ /* Just mark the VI as uninitialised. */
++ vnic->vi_state = NULL;
++}
+
-+ netif_exit_accel();
+
-+ return xenbus_unregister_driver(&netfront);
++int netfront_accel_vi_init(netfront_accel_vnic *vnic, struct net_accel_msg_hw *hw_msg)
++{
++ BUG_ON(hw_msg == NULL);
++ return netfront_accel_vi_init_fini(vnic, hw_msg);
+}
-+module_exit(netif_exit);
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/netfront/netfront.h linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.h
---- linux-2.6.18/drivers/xen/netfront/netfront.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/netfront/netfront.h 2007-12-23 11:15:34.057935783 +0100
-@@ -0,0 +1,274 @@
-+/******************************************************************************
-+ * Virtual network driver for conversing with remote driver backends.
-+ *
-+ * Copyright (c) 2002-2005, K A Fraser
-+ * Copyright (c) 2005, XenSource Ltd
-+ * Copyright (C) 2007 Solarflare Communications, Inc.
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
+
-+#ifndef NETFRONT_H
-+#define NETFRONT_H
++void netfront_accel_vi_dtor(netfront_accel_vnic *vnic)
++{
++ if (vnic->vi_state != NULL)
++ netfront_accel_vi_init_fini(vnic, NULL);
++}
+
-+#include <xen/interface/io/netif.h>
-+#include <linux/netdevice.h>
-+#include <linux/skbuff.h>
-+#include <linux/list.h>
+
-+#define NET_TX_RING_SIZE __RING_SIZE((struct netif_tx_sring *)0, PAGE_SIZE)
-+#define NET_RX_RING_SIZE __RING_SIZE((struct netif_rx_sring *)0, PAGE_SIZE)
++static
++void netfront_accel_vi_post_rx(netfront_accel_vnic *vnic, u16 id,
++ netfront_accel_pkt_desc *buf)
++{
+
-+#include <xen/xenbus.h>
++ int idx = vnic->rx_dma_batched;
+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
++#if 0
++ VPRINTK("Posting buffer %d (0x%08x) for rx at index %d, space is %d\n",
++ id, buf->pkt_buff_addr, idx, ef_vi_receive_space(&vnic->vi));
+#endif
++ /* Set up a virtual buffer descriptor */
++ ef_vi_receive_init(&vnic->vi, buf->pkt_buff_addr, id,
++ /*rx_bytes=max*/0);
+
-+/*
-+ * Function pointer table for hooks into a network acceleration
-+ * plugin. These are called at appropriate points from the netfront
-+ * driver
-+ */
-+struct netfront_accel_hooks {
-+ /*
-+ * new_device: Accelerator hook to ask the plugin to support a
-+ * new network interface
-+ */
-+ int (*new_device)(struct net_device *net_dev, struct xenbus_device *dev);
-+ /*
-+ * remove: Opposite of new_device
-+ */
-+ int (*remove)(struct xenbus_device *dev);
-+ /*
-+ * The net_device is being polled, check the accelerated
-+ * hardware for any pending packets
-+ */
-+ int (*netdev_poll)(struct net_device *dev, int *pbudget);
-+ /*
-+ * start_xmit: Used to give the accelerated plugin the option
-+ * of sending a packet. Returns non-zero if has done so, or
-+ * zero to decline and force the packet onto normal send
-+ * path
-+ */
-+ int (*start_xmit)(struct sk_buff *skb, struct net_device *dev);
-+ /*
-+ * start/stop_napi_interrupts Used by netfront to indicate
-+ * when napi interrupts should be enabled or disabled
-+ */
-+ int (*start_napi_irq)(struct net_device *dev);
-+ void (*stop_napi_irq)(struct net_device *dev);
++ idx++;
++
++ vnic->rx_dma_level++;
++
+ /*
-+ * Called before re-enabling the TX queue to check the fast
-+ * path has slots too
++ * Only push the descriptor to the card if we've reached the
++ * batch size. Otherwise, the descriptors can sit around for
++ * a while. There will be plenty available.
+ */
-+ int (*check_ready)(struct net_device *dev);
-+ /*
-+ * Get the fastpath network statistics
-+ */
-+ int (*get_stats)(struct net_device *dev,
-+ struct net_device_stats *stats);
-+};
++ if (idx >= NETFRONT_ACCEL_RX_DESC_BATCH ||
++ vnic->rx_dma_level < NETFRONT_ACCEL_RX_DESC_BATCH) {
++#if 0
++ VPRINTK("Flushing %d rx descriptors.\n", idx);
++#endif
+
++ /* Push buffer to hardware */
++ ef_vi_receive_push(&vnic->vi);
++
++ idx = 0;
++ }
++
++ vnic->rx_dma_batched = idx;
++}
+
-+/* Version of API/protocol for communication between netfront and
-+ acceleration plugin supported */
-+#define NETFRONT_ACCEL_VERSION 0x00010003
+
-+/*
-+ * Per-netfront device state for the accelerator. This is used to
-+ * allow efficient per-netfront device access to the accelerator
-+ * hooks
-+ */
-+struct netfront_accel_vif_state {
-+ struct list_head link;
++inline
++void netfront_accel_vi_post_rx_or_free(netfront_accel_vnic *vnic, u16 id,
++ netfront_accel_pkt_desc *buf)
++{
+
-+ struct xenbus_device *dev;
-+ struct netfront_info *np;
-+ struct netfront_accel_hooks *hooks;
++ VPRINTK("%s: %d\n", __FUNCTION__, id);
+
-+ /* Watch on the accelerator configuration value */
-+ struct xenbus_watch accel_watch;
-+ /* Work item to process change in accelerator */
-+ struct work_struct accel_work;
-+ /* The string from xenbus last time accel_watch fired */
-+ char *accel_frontend;
-+};
++ if (ef_vi_receive_space(&vnic->vi) <= vnic->rx_dma_batched) {
++ VPRINTK("RX space is full\n");
++ netfront_accel_buf_put(vnic->rx_bufs, id);
++ return;
++ }
+
-+/*
-+ * Per-accelerator state stored in netfront. These form a list that
-+ * is used to track which devices are accelerated by which plugins,
-+ * and what plugins are available/have been requested
-+ */
-+struct netfront_accelerator {
-+ /* Used to make a list */
-+ struct list_head link;
-+ /* ID of the accelerator */
-+ int id;
++ VPRINTK("Completed buffer %d is reposted\n", id);
++ netfront_accel_vi_post_rx(vnic, id, buf);
++
+ /*
-+ * String describing the accelerator. Currently this is the
-+ * name of the accelerator module. This is provided by the
-+ * backend accelerator through xenstore
++ * Let's see if there's any more to be pushed out to the NIC
++ * while we're here
+ */
-+ char *frontend;
-+ /* The hooks into the accelerator plugin module */
-+ struct netfront_accel_hooks *hooks;
++ while (ef_vi_receive_space(&vnic->vi) > vnic->rx_dma_batched) {
++ /* Try to allocate a buffer. */
++ buf = netfront_accel_buf_get(vnic->rx_bufs);
++ if (buf == NULL)
++ break;
++
++ /* Add it to the rx dma queue. */
++ netfront_accel_vi_post_rx(vnic, buf->buf_id, buf);
++ }
++}
+
-+ /*
-+ * List of per-netfront device state (struct
-+ * netfront_accel_vif_state) for each netfront device that is
-+ * using this accelerator
-+ */
-+ struct list_head vif_states;
-+ spinlock_t vif_states_lock;
-+};
+
-+struct netfront_info {
-+ struct list_head list;
-+ struct net_device *netdev;
++void netfront_accel_vi_add_bufs(netfront_accel_vnic *vnic, int is_rx)
++{
+
-+ struct net_device_stats stats;
++ while (is_rx &&
++ ef_vi_receive_space(&vnic->vi) > vnic->rx_dma_batched) {
++ netfront_accel_pkt_desc *buf;
++
++ VPRINTK("%s: %d\n", __FUNCTION__, vnic->rx_dma_level);
++
++ /* Try to allocate a buffer. */
++ buf = netfront_accel_buf_get(vnic->rx_bufs);
+
-+ struct netif_tx_front_ring tx;
-+ struct netif_rx_front_ring rx;
++ if (buf == NULL)
++ break;
++
++ /* Add it to the rx dma queue. */
++ netfront_accel_vi_post_rx(vnic, buf->buf_id, buf);
++ }
+
-+ spinlock_t tx_lock;
-+ spinlock_t rx_lock;
++ VPRINTK("%s: done\n", __FUNCTION__);
++}
+
-+ unsigned int irq;
-+ unsigned int copying_receiver;
-+ unsigned int carrier;
+
-+ /* Receive-ring batched refills. */
-+#define RX_MIN_TARGET 8
-+#define RX_DFL_MIN_TARGET 64
-+#define RX_MAX_TARGET min_t(int, NET_RX_RING_SIZE, 256)
-+ unsigned rx_min_target, rx_max_target, rx_target;
-+ struct sk_buff_head rx_batch;
++struct netfront_accel_multi_state {
++ unsigned remaining_len;
+
-+ struct timer_list rx_refill_timer;
++ unsigned buffers;
+
-+ /*
-+ * {tx,rx}_skbs store outstanding skbuffs. The first entry in tx_skbs
-+ * is an index into a chain of free entries.
-+ */
-+ struct sk_buff *tx_skbs[NET_TX_RING_SIZE+1];
-+ struct sk_buff *rx_skbs[NET_RX_RING_SIZE];
++ struct netfront_accel_tso_buffer *output_buffers;
+
-+#define TX_MAX_TARGET min_t(int, NET_RX_RING_SIZE, 256)
-+ grant_ref_t gref_tx_head;
-+ grant_ref_t grant_tx_ref[NET_TX_RING_SIZE + 1];
-+ grant_ref_t gref_rx_head;
-+ grant_ref_t grant_rx_ref[NET_RX_RING_SIZE];
++ /* Where we are in the current fragment of the SKB. */
++ struct {
++ /* address of current position */
++ void *addr;
++ /* remaining length */
++ unsigned int len;
++ } ifc; /* == Input Fragment Cursor */
++};
+
-+ struct xenbus_device *xbdev;
-+ int tx_ring_ref;
-+ int rx_ring_ref;
-+ u8 mac[ETH_ALEN];
+
-+ unsigned long rx_pfn_array[NET_RX_RING_SIZE];
-+ struct multicall_entry rx_mcl[NET_RX_RING_SIZE+1];
-+ struct mmu_update rx_mmu[NET_RX_RING_SIZE];
++static inline void multi_post_start(struct netfront_accel_multi_state *st,
++ struct sk_buff *skb)
++{
++ st->remaining_len = skb->len;
++ st->output_buffers = NULL;
++ st->buffers = 0;
++ st->ifc.len = skb_headlen(skb);
++ st->ifc.addr = skb->data;
++}
+
-+ /* Private pointer to state internal to accelerator module */
-+ void *accel_priv;
-+ /* The accelerator used by this netfront device */
-+ struct netfront_accelerator *accelerator;
-+ /* The accelerator state for this netfront device */
-+ struct netfront_accel_vif_state accel_vif_state;
-+};
++static int multi_post_start_new_buffer(netfront_accel_vnic *vnic,
++ struct netfront_accel_multi_state *st)
++{
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct netfront_accel_pkt_desc *buf;
+
++ /* Get a mapped packet buffer */
++ buf = netfront_accel_buf_get(vnic->tx_bufs);
++ if (buf == NULL) {
++ DPRINTK("%s: No buffer for TX\n", __FUNCTION__);
++ return -1;
++ }
+
-+/* Exported Functions */
++ /* Store a bit of meta-data at the end */
++ tso_buf = (struct netfront_accel_tso_buffer *)
++ (buf->pkt_kva + NETFRONT_ACCEL_TX_BUF_LENGTH);
+
-+/*
-+ * Called by an accelerator plugin module when it has loaded.
-+ *
-+ * frontend: the string describing the accelerator, currently the module name
-+ * hooks: the hooks for netfront to use to call into the accelerator
-+ * version: the version of API between frontend and plugin requested
-+ *
-+ * return: 0 on success, <0 on error, >0 (with version supported) on
-+ * version mismatch
-+ */
-+extern int netfront_accelerator_loaded(int version, const char *frontend,
-+ struct netfront_accel_hooks *hooks);
++ tso_buf->buf = buf;
+
-+/*
-+ * Called by an accelerator plugin module when it is about to unload.
-+ *
-+ * frontend: the string describing the accelerator. Must match the
-+ * one passed to netfront_accelerator_loaded()
-+ */
-+extern void netfront_accelerator_stop(const char *frontend);
++ tso_buf->length = 0;
++
++ tso_buf->next = st->output_buffers;
++ st->output_buffers = tso_buf;
++ st->buffers++;
+
-+/*
-+ * Called by an accelerator before waking the net device's TX queue to
-+ * ensure the slow path has available slots. Returns true if OK to
-+ * wake, false if still busy
-+ */
-+extern int netfront_check_queue_ready(struct net_device *net_dev);
++ BUG_ON(st->buffers >= ACCEL_TX_MAX_BUFFERS);
+
++ /*
++ * Store the context, set to NULL, last packet buffer will get
++ * non-NULL later
++ */
++ tso_buf->buf->skb = NULL;
++
++ return 0;
++}
+
-+/* Internal-to-netfront Functions */
+
-+/*
-+ * Call into accelerator and check to see if it has tx space before we
-+ * wake the net device's TX queue. Returns true if OK to wake, false
-+ * if still busy
-+ */
-+extern
-+int netfront_check_accelerator_queue_ready(struct net_device *dev,
-+ struct netfront_info *np);
-+extern
-+int netfront_accelerator_call_remove(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+extern
-+int netfront_accelerator_suspend(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+extern
-+int netfront_accelerator_suspend_cancel(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+extern
-+void netfront_accelerator_resume(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+extern
-+void netfront_accelerator_call_stop_napi_irq(struct netfront_info *np,
-+ struct net_device *dev);
-+extern
-+int netfront_accelerator_call_get_stats(struct netfront_info *np,
-+ struct net_device *dev);
-+extern
-+void netfront_accelerator_add_watch(struct netfront_info *np);
++static void
++multi_post_fill_buffer_with_fragment(netfront_accel_vnic *vnic,
++ struct netfront_accel_multi_state *st)
++{
++ struct netfront_accel_tso_buffer *tso_buf;
++ unsigned n, space;
+
-+extern
-+void netif_init_accel(void);
-+extern
-+void netif_exit_accel(void);
++ BUG_ON(st->output_buffers == NULL);
++ tso_buf = st->output_buffers;
+
-+extern
-+void init_accelerator_vif(struct netfront_info *np,
-+ struct xenbus_device *dev);
-+#endif /* NETFRONT_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.c
---- linux-2.6.18/drivers/xen/pciback/conf_space.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.c 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,426 @@
-+/*
-+ * PCI Backend - Functions for creating a virtual configuration space for
-+ * exported PCI Devices.
-+ * It's dangerous to allow PCI Driver Domains to change their
-+ * device's resources (memory, i/o ports, interrupts). We need to
-+ * restrict changes to certain PCI Configuration registers:
-+ * BARs, INTERRUPT_PIN, most registers in the header...
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
++ if (st->ifc.len == 0) return;
++ if (tso_buf->length == NETFRONT_ACCEL_TX_BUF_LENGTH) return;
+
-+#include <linux/kernel.h>
-+#include <linux/pci.h>
-+#include "pciback.h"
-+#include "conf_space.h"
-+#include "conf_space_quirks.h"
++ BUG_ON(tso_buf->length > NETFRONT_ACCEL_TX_BUF_LENGTH);
+
-+#define DEFINE_PCI_CONFIG(op,size,type) \
-+int pciback_##op##_config_##size \
-+(struct pci_dev *dev, int offset, type value, void *data) \
-+{ \
-+ return pci_##op##_config_##size (dev, offset, value); \
-+}
++ space = NETFRONT_ACCEL_TX_BUF_LENGTH - tso_buf->length;
++ n = min(st->ifc.len, space);
+
-+DEFINE_PCI_CONFIG(read, byte, u8 *)
-+DEFINE_PCI_CONFIG(read, word, u16 *)
-+DEFINE_PCI_CONFIG(read, dword, u32 *)
++ memcpy(tso_buf->buf->pkt_kva + tso_buf->length, st->ifc.addr, n);
+
-+DEFINE_PCI_CONFIG(write, byte, u8)
-+DEFINE_PCI_CONFIG(write, word, u16)
-+DEFINE_PCI_CONFIG(write, dword, u32)
++ st->remaining_len -= n;
++ st->ifc.len -= n;
++ tso_buf->length += n;
++ st->ifc.addr += n;
+
-+static int conf_space_read(struct pci_dev *dev,
-+ struct config_field_entry *entry, int offset,
-+ u32 * value)
++ BUG_ON(tso_buf->length > NETFRONT_ACCEL_TX_BUF_LENGTH);
++
++ return;
++}
++
++
++static inline void multi_post_unwind(netfront_accel_vnic *vnic,
++ struct netfront_accel_multi_state *st)
+{
-+ int ret = 0;
-+ struct config_field *field = entry->field;
++ struct netfront_accel_tso_buffer *tso_buf;
+
-+ *value = 0;
++ DPRINTK("%s\n", __FUNCTION__);
+
-+ switch (field->size) {
-+ case 1:
-+ if (field->u.b.read)
-+ ret = field->u.b.read(dev, offset, (u8 *) value,
-+ entry->data);
-+ break;
-+ case 2:
-+ if (field->u.w.read)
-+ ret = field->u.w.read(dev, offset, (u16 *) value,
-+ entry->data);
-+ break;
-+ case 4:
-+ if (field->u.dw.read)
-+ ret = field->u.dw.read(dev, offset, value, entry->data);
-+ break;
++ while (st->output_buffers != NULL) {
++ tso_buf = st->output_buffers;
++ st->output_buffers = tso_buf->next;
++ st->buffers--;
++ netfront_accel_buf_put(vnic->tx_bufs, tso_buf->buf->buf_id);
+ }
-+ return ret;
++ BUG_ON(st->buffers != 0);
+}
+
-+static int conf_space_write(struct pci_dev *dev,
-+ struct config_field_entry *entry, int offset,
-+ u32 value)
++
++static enum netfront_accel_post_status
++netfront_accel_enqueue_skb_multi(netfront_accel_vnic *vnic, struct sk_buff *skb)
+{
-+ int ret = 0;
-+ struct config_field *field = entry->field;
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct netfront_accel_multi_state state;
++ ef_iovec iovecs[ACCEL_TX_MAX_BUFFERS];
++ skb_frag_t *f;
++ int frag_i, rc, dma_id;
+
-+ switch (field->size) {
-+ case 1:
-+ if (field->u.b.write)
-+ ret = field->u.b.write(dev, offset, (u8) value,
-+ entry->data);
-+ break;
-+ case 2:
-+ if (field->u.w.write)
-+ ret = field->u.w.write(dev, offset, (u16) value,
-+ entry->data);
-+ break;
-+ case 4:
-+ if (field->u.dw.write)
-+ ret = field->u.dw.write(dev, offset, value,
-+ entry->data);
-+ break;
++ multi_post_start(&state, skb);
++
++ frag_i = -1;
++
++ if (skb->ip_summed == CHECKSUM_HW) {
++ /* Set to zero to encourage falcon to work it out for us */
++ *(u16*)(skb->h.raw + skb->csum) = 0;
+ }
-+ return ret;
++
++ if (multi_post_start_new_buffer(vnic, &state)) {
++ DPRINTK("%s: out of buffers\n", __FUNCTION__);
++ goto unwind;
++ }
++
++ while (1) {
++ multi_post_fill_buffer_with_fragment(vnic, &state);
++
++ /* Move onto the next fragment? */
++ if (state.ifc.len == 0) {
++ if (++frag_i >= skb_shinfo(skb)->nr_frags)
++ /* End of payload reached. */
++ break;
++ f = &skb_shinfo(skb)->frags[frag_i];
++ state.ifc.len = f->size;
++ state.ifc.addr = page_address(f->page) + f->page_offset;
++ }
++
++ /* Start a new buffer? */
++ if ((state.output_buffers->length ==
++ NETFRONT_ACCEL_TX_BUF_LENGTH) &&
++ multi_post_start_new_buffer(vnic, &state)) {
++ DPRINTK("%s: out of buffers\n", __FUNCTION__);
++ goto unwind;
++ }
++ }
++
++ /* Check for space */
++ if (ef_vi_transmit_space(&vnic->vi) < state.buffers) {
++ DPRINTK("%s: Not enough TX space (%d)\n", __FUNCTION__, state.buffers);
++ goto unwind;
++ }
++
++ /* Store the skb in what will be the last buffer's context */
++ state.output_buffers->buf->skb = skb;
++ /* Remember dma_id of what will be the last buffer */
++ dma_id = state.output_buffers->buf->buf_id;
++
++ /*
++ * Make an iovec of the buffers in the list, reversing the
++ * buffers as we go as they are constructed on a stack
++ */
++ tso_buf = state.output_buffers;
++ for (frag_i = state.buffers-1; frag_i >= 0; frag_i--) {
++ iovecs[frag_i].iov_base = tso_buf->buf->pkt_buff_addr;
++ iovecs[frag_i].iov_len = tso_buf->length;
++ tso_buf = tso_buf->next;
++ }
++
++ rc = ef_vi_transmitv(&vnic->vi, iovecs, state.buffers, dma_id);
++
++ /* Track number of tx fastpath stats */
++ vnic->netdev_stats.fastpath_tx_bytes += skb->len;
++ vnic->netdev_stats.fastpath_tx_pkts ++;
++#if NETFRONT_ACCEL_STATS
++ {
++ u32 n;
++ n = vnic->netdev_stats.fastpath_tx_pkts -
++ (u32)vnic->stats.fastpath_tx_completions;
++ if (n > vnic->stats.fastpath_tx_pending_max)
++ vnic->stats.fastpath_tx_pending_max = n;
++ }
++#endif
++ return NETFRONT_ACCEL_STATUS_GOOD;
++
++unwind:
++ multi_post_unwind(vnic, &state);
++
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_busy++);
++
++ return NETFRONT_ACCEL_STATUS_BUSY;
+}
+
-+static inline u32 get_mask(int size)
++
++static enum netfront_accel_post_status
++netfront_accel_enqueue_skb_single(netfront_accel_vnic *vnic, struct sk_buff *skb)
+{
-+ if (size == 1)
-+ return 0xff;
-+ else if (size == 2)
-+ return 0xffff;
-+ else
-+ return 0xffffffff;
++ struct netfront_accel_tso_buffer *tso_buf;
++ struct netfront_accel_pkt_desc *buf;
++ u8 *kva;
++ int rc;
++
++ if (ef_vi_transmit_space(&vnic->vi) < 1) {
++ DPRINTK("%s: No TX space\n", __FUNCTION__);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_busy++);
++ return NETFRONT_ACCEL_STATUS_BUSY;
++ }
++
++ buf = netfront_accel_buf_get(vnic->tx_bufs);
++ if (buf == NULL) {
++ DPRINTK("%s: No buffer for TX\n", __FUNCTION__);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_busy++);
++ return NETFRONT_ACCEL_STATUS_BUSY;
++ }
++
++ /* Track number of tx fastpath stats */
++ vnic->netdev_stats.fastpath_tx_pkts++;
++ vnic->netdev_stats.fastpath_tx_bytes += skb->len;
++
++#if NETFRONT_ACCEL_STATS
++ {
++ u32 n;
++ n = vnic->netdev_stats.fastpath_tx_pkts -
++ (u32)vnic->stats.fastpath_tx_completions;
++ if (n > vnic->stats.fastpath_tx_pending_max)
++ vnic->stats.fastpath_tx_pending_max = n;
++ }
++#endif
++
++ /* Store the context */
++ buf->skb = skb;
++
++ kva = buf->pkt_kva;
++
++ if (skb->ip_summed == CHECKSUM_HW) {
++ /* Set to zero to encourage falcon to work it out for us */
++ *(u16*)(skb->h.raw + skb->csum) = 0;
++ }
++ NETFRONT_ACCEL_PKTBUFF_FOR_EACH_FRAGMENT
++ (skb, idx, frag_data, frag_len, {
++ /* Copy in payload */
++ VPRINTK("*** Copying %d bytes to %p\n", frag_len, kva);
++ memcpy(kva, frag_data, frag_len);
++ kva += frag_len;
++ });
++
++ VPRINTK("%s: id %d pkt %p kva %p buff_addr 0x%08x\n", __FUNCTION__,
++ buf->buf_id, buf, buf->pkt_kva, buf->pkt_buff_addr);
++
++
++ /* Set up the TSO meta-data for a single buffer/packet */
++ tso_buf = (struct netfront_accel_tso_buffer *)
++ (buf->pkt_kva + NETFRONT_ACCEL_TX_BUF_LENGTH);
++ tso_buf->next = NULL;
++ tso_buf->buf = buf;
++ tso_buf->length = skb->len;
++
++ rc = ef_vi_transmit(&vnic->vi, buf->pkt_buff_addr, skb->len,
++ buf->buf_id);
++ /* We checked for space already, so it really should succeed */
++ BUG_ON(rc != 0);
++
++ return NETFRONT_ACCEL_STATUS_GOOD;
+}
+
-+static inline int valid_request(int offset, int size)
++
++enum netfront_accel_post_status
++netfront_accel_vi_tx_post(netfront_accel_vnic *vnic, struct sk_buff *skb)
+{
-+ /* Validate request (no un-aligned requests) */
-+ if ((size == 1 || size == 2 || size == 4) && (offset % size) == 0)
-+ return 1;
-+ return 0;
++ struct ethhdr *pkt_eth_hdr;
++ struct iphdr *pkt_ipv4_hdr;
++ int value, try_fastpath;
++
++ /*
++ * This assumes that the data field points to the dest mac
++ * address.
++ */
++ cuckoo_hash_mac_key key = cuckoo_mac_to_key(skb->data);
++
++ /*
++ * NB very important that all things that could return "CANT"
++ * are tested before things that return "BUSY" as if it it
++ * returns "BUSY" it is assumed that it won't return "CANT"
++ * next time it is tried
++ */
++
++ /*
++ * Do a fastpath send if fast path table lookup returns true.
++ * We do this without the table lock and so may get the wrong
++ * answer, but current opinion is that's not a big problem
++ */
++ try_fastpath = cuckoo_hash_lookup(&vnic->fastpath_table,
++ (cuckoo_hash_key *)(&key), &value);
++
++ if (!try_fastpath) {
++ VPRINTK("try fast path false for mac: " MAC_FMT "\n",
++ MAC_ARG(skb->data));
++
++ return NETFRONT_ACCEL_STATUS_CANT;
++ }
++
++ /* Check to see if the packet can be sent. */
++ if (skb_headlen(skb) < sizeof(*pkt_eth_hdr) + sizeof(*pkt_ipv4_hdr)) {
++ EPRINTK("%s: Packet header is too small\n", __FUNCTION__);
++ return NETFRONT_ACCEL_STATUS_CANT;
++ }
++
++ pkt_eth_hdr = (void*)skb->data;
++ pkt_ipv4_hdr = (void*)(pkt_eth_hdr+1);
++
++ if (be16_to_cpu(pkt_eth_hdr->h_proto) != ETH_P_IP) {
++ DPRINTK("%s: Packet is not IPV4 (ether_type=0x%04x)\n", __FUNCTION__,
++ be16_to_cpu(pkt_eth_hdr->h_proto));
++ return NETFRONT_ACCEL_STATUS_CANT;
++ }
++
++ if (pkt_ipv4_hdr->protocol != IPPROTO_TCP &&
++ pkt_ipv4_hdr->protocol != IPPROTO_UDP) {
++ DPRINTK("%s: Packet is not TCP/UDP (ip_protocol=0x%02x)\n",
++ __FUNCTION__, pkt_ipv4_hdr->protocol);
++ return NETFRONT_ACCEL_STATUS_CANT;
++ }
++
++ VPRINTK("%s: %d bytes, gso %d\n", __FUNCTION__, skb->len,
++ skb_shinfo(skb)->gso_size);
++
++ if (skb_shinfo(skb)->gso_size) {
++ return netfront_accel_enqueue_skb_tso(vnic, skb);
++ }
++
++ if (skb->len <= NETFRONT_ACCEL_TX_BUF_LENGTH) {
++ return netfront_accel_enqueue_skb_single(vnic, skb);
++ }
++
++ return netfront_accel_enqueue_skb_multi(vnic, skb);
+}
+
-+static inline u32 merge_value(u32 val, u32 new_val, u32 new_val_mask,
-+ int offset)
++
++/*
++ * Copy the data to required end destination. NB. len is the total new
++ * length of the socket buffer, not the amount of data to copy
++ */
++inline
++int ef_vnic_copy_to_skb(netfront_accel_vnic *vnic, struct sk_buff *skb,
++ struct netfront_accel_pkt_desc *buf, int len)
+{
-+ if (offset >= 0) {
-+ new_val_mask <<= (offset * 8);
-+ new_val <<= (offset * 8);
-+ } else {
-+ new_val_mask >>= (offset * -8);
-+ new_val >>= (offset * -8);
++ int i, extra = len - skb->len;
++ char c;
++ int pkt_stride = vnic->rx_pkt_stride;
++ int skb_stride = vnic->rx_skb_stride;
++ char *skb_start;
++
++ /*
++ * This pulls stuff into the cache - have seen performance
++ * benefit in this, but disabled by default
++ */
++ skb_start = skb->data;
++ if (pkt_stride) {
++ for (i = 0; i < len; i += pkt_stride) {
++ c += ((volatile char*)(buf->pkt_kva))[i];
++ }
++ }
++ if (skb_stride) {
++ for (i = skb->len; i < len ; i += skb_stride) {
++ c += ((volatile char*)(skb_start))[i];
++ }
+ }
-+ val = (val & ~new_val_mask) | (new_val & new_val_mask);
+
-+ return val;
++ if (skb_tailroom(skb) >= extra) {
++ memcpy(skb_put(skb, extra), buf->pkt_kva, extra);
++ return 0;
++ }
++
++ return -ENOSPC;
+}
+
-+static int pcibios_err_to_errno(int err)
++
++static void discard_jumbo_state(netfront_accel_vnic *vnic)
+{
-+ switch (err) {
-+ case PCIBIOS_SUCCESSFUL:
-+ return XEN_PCI_ERR_success;
-+ case PCIBIOS_DEVICE_NOT_FOUND:
-+ return XEN_PCI_ERR_dev_not_found;
-+ case PCIBIOS_BAD_REGISTER_NUMBER:
-+ return XEN_PCI_ERR_invalid_offset;
-+ case PCIBIOS_FUNC_NOT_SUPPORTED:
-+ return XEN_PCI_ERR_not_implemented;
-+ case PCIBIOS_SET_FAILED:
-+ return XEN_PCI_ERR_access_denied;
++
++ if (vnic->jumbo_state.skb != NULL) {
++ dev_kfree_skb_any(vnic->jumbo_state.skb);
++
++ vnic->jumbo_state.skb = NULL;
+ }
-+ return err;
++ vnic->jumbo_state.in_progress = 0;
+}
+
-+int pciback_config_read(struct pci_dev *dev, int offset, int size,
-+ u32 * ret_val)
++
++static void netfront_accel_vi_rx_complete(netfront_accel_vnic *vnic,
++ struct sk_buff *skb)
+{
-+ int err = 0;
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry;
-+ struct config_field *field;
-+ int req_start, req_end, field_start, field_end;
-+ /* if read fails for any reason, return 0 (as if device didn't respond) */
-+ u32 value = 0, tmp_val;
++ cuckoo_hash_mac_key key;
++ unsigned long flags;
++ int value;
++ struct net_device *net_dev;
+
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG "pciback: %s: read %d bytes at 0x%x\n",
-+ pci_name(dev), size, offset);
+
-+ if (!valid_request(offset, size)) {
-+ err = XEN_PCI_ERR_invalid_offset;
-+ goto out;
-+ }
++ key = cuckoo_mac_to_key(skb->data + ETH_ALEN);
+
-+ /* Get the real value first, then modify as appropriate */
-+ switch (size) {
-+ case 1:
-+ err = pci_read_config_byte(dev, offset, (u8 *) & value);
-+ break;
-+ case 2:
-+ err = pci_read_config_word(dev, offset, (u16 *) & value);
-+ break;
-+ case 4:
-+ err = pci_read_config_dword(dev, offset, &value);
-+ break;
++ /*
++ * If this is a MAC address that we want to do fast path TX
++ * to, and we don't already, add it to the fastpath table.
++ * The initial lookup is done without the table lock and so
++ * may get the wrong answer, but current opinion is that's not
++ * a big problem
++ */
++ if (is_valid_ether_addr(skb->data + ETH_ALEN) &&
++ !cuckoo_hash_lookup(&vnic->fastpath_table, (cuckoo_hash_key *)&key,
++ &value)) {
++ spin_lock_irqsave(&vnic->table_lock, flags);
++
++ cuckoo_hash_add_check(&vnic->fastpath_table,
++ (cuckoo_hash_key *)&key,
++ 1, 1);
++
++ spin_unlock_irqrestore(&vnic->table_lock, flags);
+ }
+
-+ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
-+ field = cfg_entry->field;
-+
-+ req_start = offset;
-+ req_end = offset + size;
-+ field_start = OFFSET(cfg_entry);
-+ field_end = OFFSET(cfg_entry) + field->size;
++ if (compare_ether_addr(skb->data, vnic->mac)) {
++ struct iphdr *ip = (struct iphdr *)(skb->data + ETH_HLEN);
++ u16 port;
+
-+ if ((req_start >= field_start && req_start < field_end)
-+ || (req_end > field_start && req_end <= field_end)) {
-+ err = conf_space_read(dev, cfg_entry, field_start,
-+ &tmp_val);
-+ if (err)
-+ goto out;
++ DPRINTK("%s: saw wrong MAC address " MAC_FMT "\n",
++ __FUNCTION__, MAC_ARG(skb->data));
+
-+ value = merge_value(value, tmp_val,
-+ get_mask(field->size),
-+ field_start - req_start);
++ if (ip->protocol == IPPROTO_TCP) {
++ struct tcphdr *tcp = (struct tcphdr *)
++ ((char *)ip + 4 * ip->ihl);
++ port = tcp->dest;
++ } else {
++ struct udphdr *udp = (struct udphdr *)
++ ((char *)ip + 4 * ip->ihl);
++ EPRINTK_ON(ip->protocol != IPPROTO_UDP);
++ port = udp->dest;
+ }
++
++ netfront_accel_msg_tx_fastpath(vnic, skb->data,
++ ip->daddr, port,
++ ip->protocol);
+ }
+
-+ out:
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG "pciback: %s: read %d bytes at 0x%x = %x\n",
-+ pci_name(dev), size, offset, value);
++ net_dev = vnic->net_dev;
++ skb->dev = net_dev;
++ skb->protocol = eth_type_trans(skb, net_dev);
++ /* CHECKSUM_UNNECESSARY as hardware has done it already */
++ skb->ip_summed = CHECKSUM_UNNECESSARY;
+
-+ *ret_val = value;
-+ return pcibios_err_to_errno(err);
++ if (!netfront_accel_ssr_skb(vnic, &vnic->ssr_state, skb))
++ netif_receive_skb(skb);
+}
+
-+int pciback_config_write(struct pci_dev *dev, int offset, int size, u32 value)
++
++static int netfront_accel_vi_poll_process_rx(netfront_accel_vnic *vnic,
++ ef_event *ev)
+{
-+ int err = 0, handled = 0;
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry;
-+ struct config_field *field;
-+ u32 tmp_val;
-+ int req_start, req_end, field_start, field_end;
++ struct netfront_accel_bufinfo *bufinfo = vnic->rx_bufs;
++ struct netfront_accel_pkt_desc *buf = NULL;
++ struct sk_buff *skb;
++ int id, len, sop = 0, cont = 0;
+
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG
-+ "pciback: %s: write request %d bytes at 0x%x = %x\n",
-+ pci_name(dev), size, offset, value);
++ VPRINTK("Rx event.\n");
++ /*
++ * Complete the receive operation, and get the request id of
++ * the buffer
++ */
++ id = ef_vi_receive_done(&vnic->vi, ev);
+
-+ if (!valid_request(offset, size))
-+ return XEN_PCI_ERR_invalid_offset;
++ if (id < 0 || id >= bufinfo->npages*NETFRONT_ACCEL_BUFS_PER_PAGE) {
++ EPRINTK("Rx packet %d is invalid\n", id);
++ /* Carry on round the loop if more events */
++ goto bad_packet;
++ }
++ /* Get our buffer descriptor */
++ buf = netfront_accel_buf_find(bufinfo, id);
+
-+ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
-+ field = cfg_entry->field;
++ len = EF_EVENT_RX_BYTES(*ev);
+
-+ req_start = offset;
-+ req_end = offset + size;
-+ field_start = OFFSET(cfg_entry);
-+ field_end = OFFSET(cfg_entry) + field->size;
++ /* An RX buffer has been removed from the DMA ring. */
++ vnic->rx_dma_level--;
+
-+ if ((req_start >= field_start && req_start < field_end)
-+ || (req_end > field_start && req_end <= field_end)) {
-+ tmp_val = 0;
++ if (EF_EVENT_TYPE(*ev) == EF_EVENT_TYPE_RX) {
++ sop = EF_EVENT_RX_SOP(*ev);
++ cont = EF_EVENT_RX_CONT(*ev);
+
-+ err = pciback_config_read(dev, field_start,
-+ field->size, &tmp_val);
-+ if (err)
-+ break;
++ skb = vnic->jumbo_state.skb;
+
-+ tmp_val = merge_value(tmp_val, value, get_mask(size),
-+ req_start - field_start);
++ VPRINTK("Rx packet %d: %d bytes so far; sop %d; cont %d\n",
++ id, len, sop, cont);
+
-+ err = conf_space_write(dev, cfg_entry, field_start,
-+ tmp_val);
++ if (sop) {
++ if (!vnic->jumbo_state.in_progress) {
++ vnic->jumbo_state.in_progress = 1;
++ BUG_ON(vnic->jumbo_state.skb != NULL);
++ } else {
++ /*
++ * This fragment shows a missing tail in
++ * previous one, but is itself possibly OK
++ */
++ DPRINTK("sop and in_progress => no tail\n");
+
-+ /* handled is set true here, but not every byte
-+ * may have been written! Properly detecting if
-+ * every byte is handled is unnecessary as the
-+ * flag is used to detect devices that need
-+ * special helpers to work correctly.
-+ */
-+ handled = 1;
++ /* Release the socket buffer we already had */
++ discard_jumbo_state(vnic);
++
++ /* Now start processing this fragment */
++ vnic->jumbo_state.in_progress = 1;
++ skb = NULL;
++ }
++ } else if (!vnic->jumbo_state.in_progress) {
++ DPRINTK("!sop and !in_progress => missing head\n");
++ goto missing_head;
+ }
-+ }
+
-+ if (!handled && !err) {
-+ /* By default, anything not specificially handled above is
-+ * read-only. The permissive flag changes this behavior so
-+ * that anything not specifically handled above is writable.
-+ * This means that some fields may still be read-only because
-+ * they have entries in the config_field list that intercept
-+ * the write and do nothing. */
-+ if (dev_data->permissive) {
-+ switch (size) {
-+ case 1:
-+ err = pci_write_config_byte(dev, offset,
-+ (u8) value);
-+ break;
-+ case 2:
-+ err = pci_write_config_word(dev, offset,
-+ (u16) value);
-+ break;
-+ case 4:
-+ err = pci_write_config_dword(dev, offset,
-+ (u32) value);
-+ break;
++ if (!cont) {
++ /* Update state for next time */
++ vnic->jumbo_state.in_progress = 0;
++ vnic->jumbo_state.skb = NULL;
++ } else if (!vnic->jumbo_state.in_progress) {
++ DPRINTK("cont and !in_progress => missing head\n");
++ goto missing_head;
++ }
++
++ if (skb == NULL) {
++ BUG_ON(!sop);
++
++ if (!cont)
++ skb = alloc_skb(len+NET_IP_ALIGN, GFP_ATOMIC);
++ else
++ skb = alloc_skb(vnic->net_dev->mtu+NET_IP_ALIGN,
++ GFP_ATOMIC);
++
++ if (skb == NULL) {
++ DPRINTK("%s: Couldn't get an rx skb.\n",
++ __FUNCTION__);
++ netfront_accel_vi_post_rx_or_free(vnic, (u16)id, buf);
++ /*
++ * Dropping this fragment means we
++ * should discard the rest too
++ */
++ discard_jumbo_state(vnic);
++
++ /* Carry on round the loop if more events */
++ return 0;
+ }
-+ } else if (!dev_data->warned_on_write) {
-+ dev_data->warned_on_write = 1;
-+ dev_warn(&dev->dev, "Driver tried to write to a "
-+ "read-only configuration space field at offset "
-+ "0x%x, size %d. This may be harmless, but if "
-+ "you have problems with your device:\n"
-+ "1) see permissive attribute in sysfs\n"
-+ "2) report problems to the xen-devel "
-+ "mailing list along with details of your "
-+ "device obtained from lspci.\n", offset, size);
++
++ }
++
++ /* Copy the data to required end destination */
++ if (ef_vnic_copy_to_skb(vnic, skb, buf, len) != 0) {
++ /*
++ * No space in the skb - suggests > MTU packet
++ * received
++ */
++ EPRINTK("%s: Rx packet too large (%d)\n",
++ __FUNCTION__, len);
++ netfront_accel_vi_post_rx_or_free(vnic, (u16)id, buf);
++ discard_jumbo_state(vnic);
++ return 0;
++ }
++
++ /* Put the buffer back in the DMA queue. */
++ netfront_accel_vi_post_rx_or_free(vnic, (u16)id, buf);
++
++ if (cont) {
++ vnic->jumbo_state.skb = skb;
++
++ return 0;
++ } else {
++ /* Track number of rx fastpath packets */
++ vnic->netdev_stats.fastpath_rx_pkts++;
++ vnic->netdev_stats.fastpath_rx_bytes += len;
++
++ netfront_accel_vi_rx_complete(vnic, skb);
++
++ return 1;
++ }
++ } else {
++ BUG_ON(EF_EVENT_TYPE(*ev) != EF_EVENT_TYPE_RX_DISCARD);
++
++ if (EF_EVENT_RX_DISCARD_TYPE(*ev)
++ == EF_EVENT_RX_DISCARD_TRUNC) {
++ DPRINTK("%s: " EF_EVENT_FMT
++ " buffer %d FRM_TRUNC q_id %d\n",
++ __FUNCTION__, EF_EVENT_PRI_ARG(*ev), id,
++ EF_EVENT_RX_DISCARD_Q_ID(*ev) );
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.fastpath_frm_trunc);
++ } else if (EF_EVENT_RX_DISCARD_TYPE(*ev)
++ == EF_EVENT_RX_DISCARD_OTHER) {
++ DPRINTK("%s: " EF_EVENT_FMT
++ " buffer %d RX_DISCARD_OTHER q_id %d\n",
++ __FUNCTION__, EF_EVENT_PRI_ARG(*ev), id,
++ EF_EVENT_RX_DISCARD_Q_ID(*ev) );
++ /*
++ * Probably tail of packet for which error has
++ * already been logged, so don't count in
++ * stats
++ */
++ } else {
++ EPRINTK("%s: " EF_EVENT_FMT
++ " buffer %d rx discard type %d q_id %d\n",
++ __FUNCTION__, EF_EVENT_PRI_ARG(*ev), id,
++ EF_EVENT_RX_DISCARD_TYPE(*ev),
++ EF_EVENT_RX_DISCARD_Q_ID(*ev) );
++ NETFRONT_ACCEL_STATS_OP(++vnic->stats.bad_event_count);
+ }
+ }
+
-+ return pcibios_err_to_errno(err);
-+}
++ /* discard type drops through here */
+
-+void pciback_config_free_dyn_fields(struct pci_dev *dev)
-+{
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry, *t;
-+ struct config_field *field;
++bad_packet:
++ /* Release the socket buffer we already had */
++ discard_jumbo_state(vnic);
+
-+ dev_dbg(&dev->dev,
-+ "free-ing dynamically allocated virtual configuration space fields\n");
++missing_head:
++ BUG_ON(vnic->jumbo_state.in_progress != 0);
++ BUG_ON(vnic->jumbo_state.skb != NULL);
+
-+ list_for_each_entry_safe(cfg_entry, t, &dev_data->config_fields, list) {
-+ field = cfg_entry->field;
++ if (id >= 0 && id < bufinfo->npages*NETFRONT_ACCEL_BUFS_PER_PAGE)
++ /* Put the buffer back in the DMA queue. */
++ netfront_accel_vi_post_rx_or_free(vnic, (u16)id, buf);
+
-+ if (field->clean) {
-+ field->clean(field);
++ vnic->netdev_stats.fastpath_rx_errors++;
+
-+ if (cfg_entry->data)
-+ kfree(cfg_entry->data);
++ DPRINTK("%s experienced bad packet/missing fragment error: %d \n",
++ __FUNCTION__, ev->rx.flags);
+
-+ list_del(&cfg_entry->list);
-+ kfree(cfg_entry);
-+ }
++ return 0;
++}
++
++
++static void netfront_accel_vi_not_busy(netfront_accel_vnic *vnic)
++{
++ struct netfront_info *np = ((struct netfront_info *)
++ netdev_priv(vnic->net_dev));
++ struct sk_buff *skb;
++ int handled;
++ unsigned long flags;
++
++ /*
++ * TODO if we could safely check tx_skb == NULL and return
++ * early without taking the lock, that would obviously help
++ * performance
++ */
++
++ /* Take the netfront lock which protects tx_skb. */
++ spin_lock_irqsave(&np->tx_lock, flags);
++ if (vnic->tx_skb != NULL) {
++ DPRINTK("%s trying to send spare buffer\n", __FUNCTION__);
++
++ skb = vnic->tx_skb;
++ vnic->tx_skb = NULL;
++
++ spin_unlock_irqrestore(&np->tx_lock, flags);
+
++ handled = netfront_accel_vi_tx_post(vnic, skb);
++
++ spin_lock_irqsave(&np->tx_lock, flags);
++
++ if (handled != NETFRONT_ACCEL_STATUS_BUSY) {
++ DPRINTK("%s restarting tx\n", __FUNCTION__);
++ if (netfront_check_queue_ready(vnic->net_dev)) {
++ netif_wake_queue(vnic->net_dev);
++ NETFRONT_ACCEL_STATS_OP
++ (vnic->stats.queue_wakes++);
++ }
++ } else {
++ vnic->tx_skb = skb;
++ }
++
++ /*
++ * Should never get a CANT, as it checks that before
++ * deciding it was BUSY first time round
++ */
++ BUG_ON(handled == NETFRONT_ACCEL_STATUS_CANT);
+ }
++ spin_unlock_irqrestore(&np->tx_lock, flags);
+}
+
-+void pciback_config_reset_dev(struct pci_dev *dev)
++
++static void netfront_accel_vi_tx_complete(netfront_accel_vnic *vnic,
++ struct netfront_accel_tso_buffer *tso_buf,
++ int is_last)
+{
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry;
-+ struct config_field *field;
++ struct netfront_accel_tso_buffer *next;
+
-+ dev_dbg(&dev->dev, "resetting virtual configuration space\n");
++ /*
++ * We get a single completion for every call to
++ * ef_vi_transmitv so handle any other buffers which are part
++ * of the same packet
++ */
++ while (tso_buf != NULL) {
++ if (tso_buf->buf->skb != NULL) {
++ dev_kfree_skb_any(tso_buf->buf->skb);
++ tso_buf->buf->skb = NULL;
++ }
+
-+ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
-+ field = cfg_entry->field;
++ next = tso_buf->next;
+
-+ if (field->reset)
-+ field->reset(dev, OFFSET(cfg_entry), cfg_entry->data);
++ netfront_accel_buf_put(vnic->tx_bufs, tso_buf->buf->buf_id);
++
++ tso_buf = next;
+ }
++
++ /*
++ * If this was the last one in the batch, we try and send any
++ * pending tx_skb. There should now be buffers and
++ * descriptors
++ */
++ if (is_last)
++ netfront_accel_vi_not_busy(vnic);
+}
+
-+void pciback_config_free_dev(struct pci_dev *dev)
++
++static void netfront_accel_vi_poll_process_tx(netfront_accel_vnic *vnic,
++ ef_event *ev)
+{
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry, *t;
-+ struct config_field *field;
++ struct netfront_accel_pkt_desc *buf;
++ struct netfront_accel_tso_buffer *tso_buf;
++ ef_request_id ids[EF_VI_TRANSMIT_BATCH];
++ int i, n_ids;
++ unsigned long flags;
+
-+ dev_dbg(&dev->dev, "free-ing virtual configuration space fields\n");
++ /* Get the request ids for this tx completion event. */
++ n_ids = ef_vi_transmit_unbundle(&vnic->vi, ev, ids);
+
-+ list_for_each_entry_safe(cfg_entry, t, &dev_data->config_fields, list) {
-+ list_del(&cfg_entry->list);
++ /* Take the tx buffer spin lock and hold for the duration */
++ spin_lock_irqsave(&vnic->tx_lock, flags);
+
-+ field = cfg_entry->field;
++ for (i = 0; i < n_ids; ++i) {
++ VPRINTK("Tx packet %d complete\n", ids[i]);
++ buf = netfront_accel_buf_find(vnic->tx_bufs, ids[i]);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.fastpath_tx_completions++);
+
-+ if (field->release)
-+ field->release(dev, OFFSET(cfg_entry), cfg_entry->data);
++ tso_buf = (struct netfront_accel_tso_buffer *)
++ (buf->pkt_kva + NETFRONT_ACCEL_TX_BUF_LENGTH);
++ BUG_ON(tso_buf->buf != buf);
+
-+ kfree(cfg_entry);
++ netfront_accel_vi_tx_complete(vnic, tso_buf, i == (n_ids-1));
+ }
++
++ spin_unlock_irqrestore(&vnic->tx_lock, flags);
+}
+
-+int pciback_config_add_field_offset(struct pci_dev *dev,
-+ struct config_field *field,
-+ unsigned int base_offset)
-+{
-+ int err = 0;
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry;
-+ void *tmp;
+
-+ cfg_entry = kmalloc(sizeof(*cfg_entry), GFP_KERNEL);
-+ if (!cfg_entry) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++int netfront_accel_vi_poll(netfront_accel_vnic *vnic, int rx_packets)
++{
++ ef_event ev[ACCEL_VI_POLL_EVENTS];
++ int rx_remain = rx_packets, rc, events, i;
++#if NETFRONT_ACCEL_STATS
++ int n_evs_polled = 0, rx_evs_polled = 0, tx_evs_polled = 0;
++#endif
++ BUG_ON(rx_packets <= 0);
+
-+ cfg_entry->data = NULL;
-+ cfg_entry->field = field;
-+ cfg_entry->base_offset = base_offset;
++ events = ef_eventq_poll(&vnic->vi, ev,
++ min(rx_remain, ACCEL_VI_POLL_EVENTS));
++ i = 0;
++ NETFRONT_ACCEL_STATS_OP(n_evs_polled += events);
+
-+ /* silently ignore duplicate fields */
-+ err = pciback_field_is_dup(dev,OFFSET(cfg_entry));
-+ if (err)
-+ goto out;
++ VPRINTK("%s: %d events\n", __FUNCTION__, events);
+
-+ if (field->init) {
-+ tmp = field->init(dev, OFFSET(cfg_entry));
++ /* Loop over each event */
++ while (events) {
++ VPRINTK("%s: Event "EF_EVENT_FMT", index %lu\n", __FUNCTION__,
++ EF_EVENT_PRI_ARG(ev[i]),
++ (unsigned long)(vnic->vi.evq_state->evq_ptr));
+
-+ if (IS_ERR(tmp)) {
-+ err = PTR_ERR(tmp);
-+ goto out;
++ if ((EF_EVENT_TYPE(ev[i]) == EF_EVENT_TYPE_RX) ||
++ (EF_EVENT_TYPE(ev[i]) == EF_EVENT_TYPE_RX_DISCARD)) {
++ rc = netfront_accel_vi_poll_process_rx(vnic, &ev[i]);
++ rx_remain -= rc;
++ BUG_ON(rx_remain < 0);
++ NETFRONT_ACCEL_STATS_OP(rx_evs_polled++);
++ } else if (EF_EVENT_TYPE(ev[i]) == EF_EVENT_TYPE_TX) {
++ netfront_accel_vi_poll_process_tx(vnic, &ev[i]);
++ NETFRONT_ACCEL_STATS_OP(tx_evs_polled++);
++ } else if (EF_EVENT_TYPE(ev[i]) ==
++ EF_EVENT_TYPE_RX_NO_DESC_TRUNC) {
++ DPRINTK("%s: RX_NO_DESC_TRUNC " EF_EVENT_FMT "\n",
++ __FUNCTION__, EF_EVENT_PRI_ARG(ev[i]));
++ discard_jumbo_state(vnic);
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.rx_no_desc_trunc++);
++ } else {
++ EPRINTK("Unexpected event " EF_EVENT_FMT "\n",
++ EF_EVENT_PRI_ARG(ev[i]));
++ NETFRONT_ACCEL_STATS_OP(vnic->stats.bad_event_count++);
+ }
+
-+ cfg_entry->data = tmp;
-+ }
++ i++;
+
-+ dev_dbg(&dev->dev, "added config field at offset 0x%02x\n",
-+ OFFSET(cfg_entry));
-+ list_add_tail(&cfg_entry->list, &dev_data->config_fields);
++ /* Carry on round the loop if more events and more space */
++ if (i == events) {
++ if (rx_remain == 0)
++ break;
+
-+ out:
-+ if (err)
-+ kfree(cfg_entry);
++ events = ef_eventq_poll(&vnic->vi, ev,
++ min(rx_remain,
++ ACCEL_VI_POLL_EVENTS));
++ i = 0;
++ NETFRONT_ACCEL_STATS_OP(n_evs_polled += events);
++ }
++ }
++
++#if NETFRONT_ACCEL_STATS
++ vnic->stats.event_count += n_evs_polled;
++ vnic->stats.event_count_since_irq += n_evs_polled;
++ if (n_evs_polled > vnic->stats.events_per_poll_max)
++ vnic->stats.events_per_poll_max = n_evs_polled;
++ if (rx_evs_polled > vnic->stats.events_per_poll_rx_max)
++ vnic->stats.events_per_poll_rx_max = rx_evs_polled;
++ if (tx_evs_polled > vnic->stats.events_per_poll_tx_max)
++ vnic->stats.events_per_poll_tx_max = tx_evs_polled;
++#endif
+
-+ return err;
++ return rx_packets - rx_remain;
+}
+
-+/* This sets up the device's virtual configuration space to keep track of
-+ * certain registers (like the base address registers (BARs) so that we can
-+ * keep the client from manipulating them directly.
-+ */
-+int pciback_config_init_dev(struct pci_dev *dev)
++
++int netfront_accel_vi_enable_interrupts(netfront_accel_vnic *vnic)
+{
-+ int err = 0;
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
++ u32 sw_evq_ptr;
+
-+ dev_dbg(&dev->dev, "initializing virtual configuration space\n");
++ VPRINTK("%s: checking for event on %p\n", __FUNCTION__, &vnic->vi.evq_state);
+
-+ INIT_LIST_HEAD(&dev_data->config_fields);
++ BUG_ON(vnic == NULL);
++ BUG_ON(vnic->vi.evq_state == NULL);
+
-+ err = pciback_config_header_add_fields(dev);
-+ if (err)
-+ goto out;
++ /* Do a quick check for an event. */
++ if (ef_eventq_has_event(&vnic->vi)) {
++ VPRINTK("%s: found event\n", __FUNCTION__);
++ return 0;
++ }
+
-+ err = pciback_config_capability_add_fields(dev);
-+ if (err)
-+ goto out;
++ VPRINTK("evq_ptr=0x%08x evq_mask=0x%08x\n",
++ vnic->evq_state.evq_ptr, vnic->vi.evq_mask);
++
++ /* Request a wakeup from the hardware. */
++ sw_evq_ptr = vnic->evq_state.evq_ptr & vnic->vi.evq_mask;
+
-+ err = pciback_config_quirks_init(dev);
++ BUG_ON(vnic->hw.falcon.evq_rptr == NULL);
+
-+ out:
-+ return err;
-+}
++ VPRINTK("Requesting wakeup at 0x%08x, rptr %p\n", sw_evq_ptr,
++ vnic->hw.falcon.evq_rptr);
++ *(volatile u32 *)(vnic->hw.falcon.evq_rptr) = (sw_evq_ptr >> 3);
+
-+int pciback_config_init(void)
-+{
-+ return pciback_config_capability_init();
++ return 1;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_capability.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.c
---- linux-2.6.18/drivers/xen/pciback/conf_space_capability.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.c 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,71 @@
-+/*
-+ * PCI Backend - Handles the virtual fields found on the capability lists
-+ * in the configuration space.
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/accel_xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/accel_xenbus.c 2008-05-19 00:33:48.650950417 +0300
+@@ -0,0 +1,776 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/kernel.h>
-+#include <linux/pci.h>
-+#include "pciback.h"
-+#include "conf_space.h"
-+#include "conf_space_capability.h"
++#include <linux/stddef.h>
++#include <linux/errno.h>
+
-+static LIST_HEAD(capabilities);
++#include <xen/xenbus.h>
++#include <xen/evtchn.h>
++#include <xen/gnttab.h>
+
-+static struct config_field caplist_header[] = {
-+ {
-+ .offset = PCI_CAP_LIST_ID,
-+ .size = 2, /* encompass PCI_CAP_LIST_ID & PCI_CAP_LIST_NEXT */
-+ .u.w.read = pciback_read_config_word,
-+ .u.w.write = NULL,
-+ },
-+ {
-+ .size = 0,
-+ },
-+};
++#include "accel.h"
++#include "accel_util.h"
++#include "accel_msg_iface.h"
++#include "accel_bufs.h"
++#include "accel_ssr.h"
++/* drivers/xen/netfront/netfront.h */
++#include "netfront.h"
+
-+static inline void register_capability(struct pciback_config_capability *cap)
++void netfront_accel_set_closing(netfront_accel_vnic *vnic)
+{
-+ list_add_tail(&cap->cap_list, &capabilities);
++
++ vnic->frontend_state = XenbusStateClosing;
++ net_accel_update_state(vnic->dev, XenbusStateClosing);
+}
++
+
-+int pciback_config_capability_add_fields(struct pci_dev *dev)
++static void mac_address_change(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
+{
-+ int err = 0;
-+ struct pciback_config_capability *cap;
-+ int cap_offset;
++ netfront_accel_vnic *vnic;
++ struct xenbus_device *dev;
++ int rc;
+
-+ list_for_each_entry(cap, &capabilities, cap_list) {
-+ cap_offset = pci_find_capability(dev, cap->capability);
-+ if (cap_offset) {
-+ dev_dbg(&dev->dev, "Found capability 0x%x at 0x%x\n",
-+ cap->capability, cap_offset);
++ DPRINTK("%s\n", __FUNCTION__);
++
++ vnic = container_of(watch, netfront_accel_vnic,
++ mac_address_watch);
++ dev = vnic->dev;
+
-+ err = pciback_config_add_fields_offset(dev,
-+ caplist_header,
-+ cap_offset);
-+ if (err)
-+ goto out;
-+ err = pciback_config_add_fields_offset(dev,
-+ cap->fields,
-+ cap_offset);
-+ if (err)
-+ goto out;
-+ }
-+ }
++ rc = net_accel_xen_net_read_mac(dev, vnic->mac);
+
-+ out:
-+ return err;
++ if (rc != 0)
++ EPRINTK("%s: failed to read mac (%d)\n", __FUNCTION__, rc);
+}
+
-+extern struct pciback_config_capability pciback_config_capability_vpd;
-+extern struct pciback_config_capability pciback_config_capability_pm;
+
-+int pciback_config_capability_init(void)
++static int setup_mac_address_watch(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
+{
-+ register_capability(&pciback_config_capability_vpd);
-+ register_capability(&pciback_config_capability_pm);
++ int err;
++
++ DPRINTK("Setting watch on %s/%s\n", dev->nodename, "mac");
++
++ err = xenbus_watch_path2(dev, dev->nodename, "mac",
++ &vnic->mac_address_watch,
++ mac_address_change);
++ if (err) {
++ EPRINTK("%s: Failed to register xenbus watch: %d\n",
++ __FUNCTION__, err);
++ goto fail;
++ }
+
+ return 0;
++ fail:
++ vnic->mac_address_watch.node = NULL;
++ return err;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_capability.h linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.h
---- linux-2.6.18/drivers/xen/pciback/conf_space_capability.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability.h 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,23 @@
-+/*
-+ * PCI Backend - Data structures for special overlays for structures on
-+ * the capability list.
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+
-+#ifndef __PCIBACK_CONFIG_CAPABILITY_H__
-+#define __PCIBACK_CONFIG_CAPABILITY_H__
+
-+#include <linux/pci.h>
-+#include <linux/list.h>
+
-+struct pciback_config_capability {
-+ struct list_head cap_list;
++/* Grant access to some pages and publish through xenbus */
++static int make_named_grant(struct xenbus_device *dev, void *page,
++ const char *name, grant_ref_t *gnt_ref)
++{
++ struct xenbus_transaction tr;
++ int err;
++ grant_ref_t gnt;
+
-+ int capability;
++ gnt = net_accel_grant_page(dev, virt_to_mfn(page), 0);
++ if (gnt < 0)
++ return gnt;
+
-+ /* If the device has the capability found above, add these fields */
-+ struct config_field *fields;
-+};
++ do {
++ err = xenbus_transaction_start(&tr);
++ if (err != 0) {
++ EPRINTK("%s: transaction start failed %d\n",
++ __FUNCTION__, err);
++ return err;
++ }
++ err = xenbus_printf(tr, dev->nodename, name, "%d", gnt);
++ if (err != 0) {
++ EPRINTK("%s: xenbus_printf failed %d\n", __FUNCTION__,
++ err);
++ xenbus_transaction_end(tr, 1);
++ return err;
++ }
++ err = xenbus_transaction_end(tr, 0);
++ } while (err == -EAGAIN);
++
++ if (err != 0) {
++ EPRINTK("%s: transaction end failed %d\n", __FUNCTION__, err);
++ return err;
++ }
++
++ *gnt_ref = gnt;
+
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_capability_pm.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_pm.c
---- linux-2.6.18/drivers/xen/pciback/conf_space_capability_pm.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_pm.c 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,128 @@
-+/*
-+ * PCI Backend - Configuration space overlay for power management
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
++ return 0;
++}
+
-+#include <linux/pci.h>
-+#include "conf_space.h"
-+#include "conf_space_capability.h"
+
-+static int pm_caps_read(struct pci_dev *dev, int offset, u16 *value,
-+ void *data)
++static int remove_named_grant(struct xenbus_device *dev,
++ const char *name, grant_ref_t gnt_ref)
+{
++ struct xenbus_transaction tr;
+ int err;
-+ u16 real_value;
+
-+ err = pci_read_config_word(dev, offset, &real_value);
-+ if (err)
-+ goto out;
++ net_accel_ungrant_page(gnt_ref);
+
-+ *value = real_value & ~PCI_PM_CAP_PME_MASK;
++ do {
++ err = xenbus_transaction_start(&tr);
++ if (err != 0) {
++ EPRINTK("%s: transaction start failed %d\n",
++ __FUNCTION__, err);
++ return err;
++ }
++ err = xenbus_rm(tr, dev->nodename, name);
++ if (err != 0) {
++ EPRINTK("%s: xenbus_rm failed %d\n", __FUNCTION__,
++ err);
++ xenbus_transaction_end(tr, 1);
++ return err;
++ }
++ err = xenbus_transaction_end(tr, 0);
++ } while (err == -EAGAIN);
++
++ if (err != 0) {
++ EPRINTK("%s: transaction end failed %d\n", __FUNCTION__, err);
++ return err;
++ }
+
-+ out:
-+ return err;
++ return 0;
+}
+
-+/* PM_OK_BITS specifies the bits that the driver domain is allowed to change.
-+ * Can't allow driver domain to enable PMEs - they're shared */
-+#define PM_OK_BITS (PCI_PM_CTRL_PME_STATUS|PCI_PM_CTRL_DATA_SEL_MASK)
+
-+static int pm_ctrl_write(struct pci_dev *dev, int offset, u16 new_value,
-+ void *data)
++static
++netfront_accel_vnic *netfront_accel_vnic_ctor(struct net_device *net_dev,
++ struct xenbus_device *dev)
+{
++ struct netfront_info *np =
++ (struct netfront_info *)netdev_priv(net_dev);
++ netfront_accel_vnic *vnic;
+ int err;
-+ u16 old_value;
-+ pci_power_t new_state, old_state;
-+
-+ err = pci_read_config_word(dev, offset, &old_value);
-+ if (err)
-+ goto out;
+
-+ old_state = (pci_power_t)(old_value & PCI_PM_CTRL_STATE_MASK);
-+ new_state = (pci_power_t)(new_value & PCI_PM_CTRL_STATE_MASK);
++ /*
++ * A bug in earlier versions of Xen accel plugin system meant
++ * you could be probed twice for the same device on suspend
++ * cancel. Be tolerant of that.
++ */
++ if (np->accel_priv != NULL)
++ return ERR_PTR(-EALREADY);
+
-+ new_value &= PM_OK_BITS;
-+ if ((old_value & PM_OK_BITS) != new_value) {
-+ new_value = (old_value & ~PM_OK_BITS) | new_value;
-+ err = pci_write_config_word(dev, offset, new_value);
-+ if (err)
-+ goto out;
++ /* Alloc mem for state */
++ vnic = kzalloc(sizeof(netfront_accel_vnic), GFP_KERNEL);
++ if (vnic == NULL) {
++ EPRINTK("%s: no memory for vnic state\n", __FUNCTION__);
++ return ERR_PTR(-ENOMEM);
+ }
+
-+ /* Let pci core handle the power management change */
-+ dev_dbg(&dev->dev, "set power state to %x\n", new_state);
-+ err = pci_set_power_state(dev, new_state);
-+ if (err) {
-+ err = PCIBIOS_SET_FAILED;
-+ goto out;
-+ }
++ spin_lock_init(&vnic->tx_lock);
+
-+ /*
-+ * Device may lose PCI config info on D3->D0 transition. This
-+ * is a problem for some guests which will not reset BARs. Even
-+ * those that have a go will be foiled by our BAR-write handler
-+ * which will discard the write! Since Linux won't re-init
-+ * the config space automatically in all cases, we do it here.
-+ * Future: Should we re-initialise all first 64 bytes of config space?
-+ */
-+ if (new_state == PCI_D0 &&
-+ (old_state == PCI_D3hot || old_state == PCI_D3cold) &&
-+ !(old_value & PCI_PM_CTRL_NO_SOFT_RESET))
-+ pci_restore_bars(dev);
++ mutex_init(&vnic->vnic_mutex);
++ mutex_lock(&vnic->vnic_mutex);
+
-+ out:
-+ return err;
-+}
++ /* Store so state can be retrieved from device */
++ BUG_ON(np->accel_priv != NULL);
++ np->accel_priv = vnic;
++ vnic->dev = dev;
++ vnic->net_dev = net_dev;
++ spin_lock_init(&vnic->irq_enabled_lock);
++ netfront_accel_ssr_init(&vnic->ssr_state);
+
-+/* Ensure PMEs are disabled */
-+static void *pm_ctrl_init(struct pci_dev *dev, int offset)
-+{
-+ int err;
-+ u16 value;
++ init_waitqueue_head(&vnic->state_wait_queue);
++ vnic->backend_state = XenbusStateUnknown;
++ vnic->frontend_state = XenbusStateClosed;
++ vnic->removing = 0;
++ vnic->domU_state_is_setup = 0;
++ vnic->dom0_state_is_setup = 0;
++ vnic->poll_enabled = 0;
++ vnic->tx_enabled = 0;
++ vnic->tx_skb = NULL;
+
-+ err = pci_read_config_word(dev, offset, &value);
-+ if (err)
-+ goto out;
++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
++ INIT_WORK(&vnic->msg_from_bend, netfront_accel_msg_from_bend);
++#else
++ INIT_WORK(&vnic->msg_from_bend, netfront_accel_msg_from_bend, vnic);
++#endif
+
-+ if (value & PCI_PM_CTRL_PME_ENABLE) {
-+ value &= ~PCI_PM_CTRL_PME_ENABLE;
-+ err = pci_write_config_word(dev, offset, value);
-+ }
++ netfront_accel_debugfs_create(vnic);
+
-+ out:
-+ return ERR_PTR(err);
-+}
++ mutex_unlock(&vnic->vnic_mutex);
+
-+static struct config_field caplist_pm[] = {
-+ {
-+ .offset = PCI_PM_PMC,
-+ .size = 2,
-+ .u.w.read = pm_caps_read,
-+ },
-+ {
-+ .offset = PCI_PM_CTRL,
-+ .size = 2,
-+ .init = pm_ctrl_init,
-+ .u.w.read = pciback_read_config_word,
-+ .u.w.write = pm_ctrl_write,
-+ },
-+ {
-+ .offset = PCI_PM_PPB_EXTENSIONS,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ },
-+ {
-+ .offset = PCI_PM_DATA_REGISTER,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ },
-+ {
-+ .size = 0,
-+ },
-+};
++ err = net_accel_xen_net_read_mac(dev, vnic->mac);
++ if (err)
++ goto fail_mac;
+
-+struct pciback_config_capability pciback_config_capability_pm = {
-+ .capability = PCI_CAP_ID_PM,
-+ .fields = caplist_pm,
-+};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_capability_vpd.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_vpd.c
---- linux-2.6.18/drivers/xen/pciback/conf_space_capability_vpd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_capability_vpd.c 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,42 @@
-+/*
-+ * PCI Backend - Configuration space overlay for Vital Product Data
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
++ /* Setup a watch on the frontend's MAC address */
++ err = setup_mac_address_watch(dev, vnic);
++ if (err)
++ goto fail_mac;
+
-+#include <linux/pci.h>
-+#include "conf_space.h"
-+#include "conf_space_capability.h"
++ return vnic;
+
-+static int vpd_address_write(struct pci_dev *dev, int offset, u16 value,
-+ void *data)
-+{
-+ /* Disallow writes to the vital product data */
-+ if (value & PCI_VPD_ADDR_F)
-+ return PCIBIOS_SET_FAILED;
-+ else
-+ return pci_write_config_word(dev, offset, value);
-+}
++fail_mac:
+
-+static struct config_field caplist_vpd[] = {
-+ {
-+ .offset = PCI_VPD_ADDR,
-+ .size = 2,
-+ .u.w.read = pciback_read_config_word,
-+ .u.w.write = vpd_address_write,
-+ },
-+ {
-+ .offset = PCI_VPD_DATA,
-+ .size = 4,
-+ .u.dw.read = pciback_read_config_dword,
-+ .u.dw.write = NULL,
-+ },
-+ {
-+ .size = 0,
-+ },
-+};
-+
-+struct pciback_config_capability pciback_config_capability_vpd = {
-+ .capability = PCI_CAP_ID_VPD,
-+ .fields = caplist_vpd,
-+};
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space.h linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.h
---- linux-2.6.18/drivers/xen/pciback/conf_space.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space.h 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,126 @@
-+/*
-+ * PCI Backend - Common data structures for overriding the configuration space
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
++ mutex_lock(&vnic->vnic_mutex);
+
-+#ifndef __XEN_PCIBACK_CONF_SPACE_H__
-+#define __XEN_PCIBACK_CONF_SPACE_H__
++ netfront_accel_debugfs_remove(vnic);
+
-+#include <linux/list.h>
-+#include <linux/err.h>
++ netfront_accel_ssr_fini(vnic, &vnic->ssr_state);
+
-+/* conf_field_init can return an errno in a ptr with ERR_PTR() */
-+typedef void *(*conf_field_init) (struct pci_dev * dev, int offset);
-+typedef void (*conf_field_reset) (struct pci_dev * dev, int offset, void *data);
-+typedef void (*conf_field_free) (struct pci_dev * dev, int offset, void *data);
++ EPRINTK_ON(vnic->tx_skb != NULL);
+
-+typedef int (*conf_dword_write) (struct pci_dev * dev, int offset, u32 value,
-+ void *data);
-+typedef int (*conf_word_write) (struct pci_dev * dev, int offset, u16 value,
-+ void *data);
-+typedef int (*conf_byte_write) (struct pci_dev * dev, int offset, u8 value,
-+ void *data);
-+typedef int (*conf_dword_read) (struct pci_dev * dev, int offset, u32 * value,
-+ void *data);
-+typedef int (*conf_word_read) (struct pci_dev * dev, int offset, u16 * value,
-+ void *data);
-+typedef int (*conf_byte_read) (struct pci_dev * dev, int offset, u8 * value,
-+ void *data);
++ vnic->frontend_state = XenbusStateUnknown;
++ net_accel_update_state(dev, XenbusStateUnknown);
+
-+/* These are the fields within the configuration space which we
-+ * are interested in intercepting reads/writes to and changing their
-+ * values.
-+ */
-+struct config_field {
-+ unsigned int offset;
-+ unsigned int size;
-+ unsigned int mask;
-+ conf_field_init init;
-+ conf_field_reset reset;
-+ conf_field_free release;
-+ void (*clean) (struct config_field * field);
-+ union {
-+ struct {
-+ conf_dword_write write;
-+ conf_dword_read read;
-+ } dw;
-+ struct {
-+ conf_word_write write;
-+ conf_word_read read;
-+ } w;
-+ struct {
-+ conf_byte_write write;
-+ conf_byte_read read;
-+ } b;
-+ } u;
-+ struct list_head list;
-+};
++ mutex_unlock(&vnic->vnic_mutex);
+
-+struct config_field_entry {
-+ struct list_head list;
-+ struct config_field *field;
-+ unsigned int base_offset;
-+ void *data;
-+};
++ np->accel_priv = NULL;
++ kfree(vnic);
+
-+#define OFFSET(cfg_entry) ((cfg_entry)->base_offset+(cfg_entry)->field->offset)
++ return ERR_PTR(err);
++}
+
-+/* Add fields to a device - the add_fields macro expects to get a pointer to
-+ * the first entry in an array (of which the ending is marked by size==0)
-+ */
-+int pciback_config_add_field_offset(struct pci_dev *dev,
-+ struct config_field *field,
-+ unsigned int offset);
+
-+static inline int pciback_config_add_field(struct pci_dev *dev,
-+ struct config_field *field)
++static void netfront_accel_vnic_dtor(netfront_accel_vnic *vnic)
+{
-+ return pciback_config_add_field_offset(dev, field, 0);
-+}
++ struct net_device *net_dev = vnic->net_dev;
++ struct netfront_info *np =
++ (struct netfront_info *)netdev_priv(net_dev);
+
-+static inline int pciback_config_add_fields(struct pci_dev *dev,
-+ struct config_field *field)
-+{
-+ int i, err = 0;
-+ for (i = 0; field[i].size != 0; i++) {
-+ err = pciback_config_add_field(dev, &field[i]);
-+ if (err)
-+ break;
-+ }
-+ return err;
-+}
++ /*
++ * Now we don't hold the lock any more it is safe to remove
++ * this watch and synchonrise with the completion of
++ * watches
++ */
++ DPRINTK("%s: unregistering xenbus mac watch\n", __FUNCTION__);
++ unregister_xenbus_watch(&vnic->mac_address_watch);
++ kfree(vnic->mac_address_watch.node);
+
-+static inline int pciback_config_add_fields_offset(struct pci_dev *dev,
-+ struct config_field *field,
-+ unsigned int offset)
-+{
-+ int i, err = 0;
-+ for (i = 0; field[i].size != 0; i++) {
-+ err = pciback_config_add_field_offset(dev, &field[i], offset);
-+ if (err)
-+ break;
-+ }
-+ return err;
-+}
++ flush_workqueue(netfront_accel_workqueue);
+
-+/* Read/Write the real configuration space */
-+int pciback_read_config_byte(struct pci_dev *dev, int offset, u8 * value,
-+ void *data);
-+int pciback_read_config_word(struct pci_dev *dev, int offset, u16 * value,
-+ void *data);
-+int pciback_read_config_dword(struct pci_dev *dev, int offset, u32 * value,
-+ void *data);
-+int pciback_write_config_byte(struct pci_dev *dev, int offset, u8 value,
-+ void *data);
-+int pciback_write_config_word(struct pci_dev *dev, int offset, u16 value,
-+ void *data);
-+int pciback_write_config_dword(struct pci_dev *dev, int offset, u32 value,
-+ void *data);
++ mutex_lock(&vnic->vnic_mutex);
+
-+int pciback_config_capability_init(void);
++ netfront_accel_debugfs_remove(vnic);
+
-+int pciback_config_header_add_fields(struct pci_dev *dev);
-+int pciback_config_capability_add_fields(struct pci_dev *dev);
++ netfront_accel_ssr_fini(vnic, &vnic->ssr_state);
+
-+#endif /* __XEN_PCIBACK_CONF_SPACE_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_header.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_header.c
---- linux-2.6.18/drivers/xen/pciback/conf_space_header.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_header.c 2007-12-23 11:15:34.064602797 +0100
-@@ -0,0 +1,323 @@
-+/*
-+ * PCI Backend - Handles the virtual fields in the configuration space headers.
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
++ EPRINTK_ON(vnic->tx_skb != NULL);
+
-+#include <linux/kernel.h>
-+#include <linux/pci.h>
-+#include "pciback.h"
-+#include "conf_space.h"
++ vnic->frontend_state = XenbusStateUnknown;
++ net_accel_update_state(vnic->dev, XenbusStateUnknown);
+
-+struct pci_bar_info {
-+ u32 val;
-+ u32 len_val;
-+ int which;
-+};
++ mutex_unlock(&vnic->vnic_mutex);
+
-+#define is_enable_cmd(value) ((value)&(PCI_COMMAND_MEMORY|PCI_COMMAND_IO))
-+#define is_master_cmd(value) ((value)&PCI_COMMAND_MASTER)
++ np->accel_priv = NULL;
++ kfree(vnic);
++}
+
-+static int command_write(struct pci_dev *dev, int offset, u16 value, void *data)
++
++static int vnic_setup_domU_shared_state(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
+{
++ struct xenbus_transaction tr;
+ int err;
++ int msgs_per_queue;
+
-+ if (!dev->is_enabled && is_enable_cmd(value)) {
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG "pciback: %s: enable\n",
-+ pci_name(dev));
-+ err = pci_enable_device(dev);
-+ if (err)
-+ return err;
-+ } else if (dev->is_enabled && !is_enable_cmd(value)) {
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG "pciback: %s: disable\n",
-+ pci_name(dev));
-+ pci_disable_device(dev);
-+ }
-+
-+ if (!dev->is_busmaster && is_master_cmd(value)) {
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG "pciback: %s: set bus master\n",
-+ pci_name(dev));
-+ pci_set_master(dev);
-+ }
+
-+ if (value & PCI_COMMAND_INVALIDATE) {
-+ if (unlikely(verbose_request))
-+ printk(KERN_DEBUG
-+ "pciback: %s: enable memory-write-invalidate\n",
-+ pci_name(dev));
-+ err = pci_set_mwi(dev);
-+ if (err) {
-+ printk(KERN_WARNING
-+ "pciback: %s: cannot enable memory-write-invalidate (%d)\n",
-+ pci_name(dev), err);
-+ value &= ~PCI_COMMAND_INVALIDATE;
-+ }
-+ }
++ DPRINTK("Setting up domU shared state.\n");
+
-+ return pci_write_config_word(dev, offset, value);
-+}
++ msgs_per_queue = (PAGE_SIZE/2) / sizeof(struct net_accel_msg);
+
-+static int rom_write(struct pci_dev *dev, int offset, u32 value, void *data)
-+{
-+ struct pci_bar_info *bar = data;
++ /* Allocate buffer state */
++ vnic->tx_bufs = netfront_accel_init_bufs(&vnic->tx_lock);
++ if (vnic->tx_bufs == NULL) {
++ err = -ENOMEM;
++ EPRINTK("%s: Failed to allocate tx buffers\n", __FUNCTION__);
++ goto fail_tx_bufs;
++ }
+
-+ if (unlikely(!bar)) {
-+ printk(KERN_WARNING "pciback: driver data not found for %s\n",
-+ pci_name(dev));
-+ return XEN_PCI_ERR_op_failed;
++ vnic->rx_bufs = netfront_accel_init_bufs(NULL);
++ if (vnic->rx_bufs == NULL) {
++ err = -ENOMEM;
++ EPRINTK("%s: Failed to allocate rx buffers\n", __FUNCTION__);
++ goto fail_rx_bufs;
+ }
+
-+ /* A write to obtain the length must happen as a 32-bit write.
-+ * This does not (yet) support writing individual bytes
++ /*
++ * This allocates two pages, one for the shared page and one
++ * for the message queue.
+ */
-+ if (value == ~PCI_ROM_ADDRESS_ENABLE)
-+ bar->which = 1;
-+ else {
-+ u32 tmpval;
-+ pci_read_config_dword(dev, offset, &tmpval);
-+ if (tmpval != bar->val && value == bar->val) {
-+ /* Allow restoration of bar value. */
-+ pci_write_config_dword(dev, offset, bar->val);
-+ }
-+ bar->which = 0;
++ vnic->shared_page = (struct net_accel_shared_page *)
++ __get_free_pages(GFP_KERNEL, 1);
++ if (vnic->shared_page == NULL) {
++ EPRINTK("%s: no memory for shared pages\n", __FUNCTION__);
++ err = -ENOMEM;
++ goto fail_shared_page;
+ }
+
-+ /* Do we need to support enabling/disabling the rom address here? */
++ net_accel_msg_init_queue
++ (&vnic->from_dom0, &vnic->shared_page->queue0,
++ (struct net_accel_msg *)((u8*)vnic->shared_page + PAGE_SIZE),
++ msgs_per_queue);
+
-+ return 0;
-+}
++ net_accel_msg_init_queue
++ (&vnic->to_dom0, &vnic->shared_page->queue1,
++ (struct net_accel_msg *)((u8*)vnic->shared_page +
++ (3 * PAGE_SIZE / 2)),
++ msgs_per_queue);
++
++ vnic->msg_state = NETFRONT_ACCEL_MSG_NONE;
+
-+/* For the BARs, only allow writes which write ~0 or
-+ * the correct resource information
-+ * (Needed for when the driver probes the resource usage)
-+ */
-+static int bar_write(struct pci_dev *dev, int offset, u32 value, void *data)
-+{
-+ struct pci_bar_info *bar = data;
++ err = make_named_grant(dev, vnic->shared_page, "accel-ctrl-page",
++ &vnic->ctrl_page_gnt);
++ if (err) {
++ EPRINTK("couldn't make ctrl-page named grant\n");
++ goto fail_ctrl_page_grant;
++ }
+
-+ if (unlikely(!bar)) {
-+ printk(KERN_WARNING "pciback: driver data not found for %s\n",
-+ pci_name(dev));
-+ return XEN_PCI_ERR_op_failed;
++ err = make_named_grant(dev, (u8*)vnic->shared_page + PAGE_SIZE,
++ "accel-msg-page", &vnic->msg_page_gnt);
++ if (err) {
++ EPRINTK("couldn't make msg-page named grant\n");
++ goto fail_msg_page_grant;
+ }
+
-+ /* A write to obtain the length must happen as a 32-bit write.
-+ * This does not (yet) support writing individual bytes
-+ */
-+ if (value == ~0)
-+ bar->which = 1;
-+ else {
-+ u32 tmpval;
-+ pci_read_config_dword(dev, offset, &tmpval);
-+ if (tmpval != bar->val && value == bar->val) {
-+ /* Allow restoration of bar value. */
-+ pci_write_config_dword(dev, offset, bar->val);
-+ }
-+ bar->which = 0;
++ /* Create xenbus msg event channel */
++ err = bind_listening_port_to_irqhandler
++ (dev->otherend_id, netfront_accel_msg_channel_irq_from_bend,
++ SA_SAMPLE_RANDOM, "vnicctrl", vnic);
++ if (err < 0) {
++ EPRINTK("Couldn't bind msg event channel\n");
++ goto fail_msg_irq;
++ }
++ vnic->msg_channel_irq = err;
++ vnic->msg_channel = irq_to_evtchn_port(vnic->msg_channel_irq);
++
++ /* Create xenbus net event channel */
++ err = bind_listening_port_to_irqhandler
++ (dev->otherend_id, netfront_accel_net_channel_irq_from_bend,
++ SA_SAMPLE_RANDOM, "vnicfront", vnic);
++ if (err < 0) {
++ EPRINTK("Couldn't bind net event channel\n");
++ goto fail_net_irq;
+ }
++ vnic->net_channel_irq = err;
++ vnic->net_channel = irq_to_evtchn_port(vnic->net_channel_irq);
++ /* Want to ensure we don't get interrupts before we're ready */
++ netfront_accel_disable_net_interrupts(vnic);
+
-+ return 0;
-+}
++ DPRINTK("otherend %d has msg ch %u (%u) and net ch %u (%u)\n",
++ dev->otherend_id, vnic->msg_channel, vnic->msg_channel_irq,
++ vnic->net_channel, vnic->net_channel_irq);
+
-+static int bar_read(struct pci_dev *dev, int offset, u32 * value, void *data)
-+{
-+ struct pci_bar_info *bar = data;
++ do {
++ err = xenbus_transaction_start(&tr);
++ if (err != 0) {
++ EPRINTK("%s: Transaction start failed %d\n",
++ __FUNCTION__, err);
++ goto fail_transaction;
++ }
+
-+ if (unlikely(!bar)) {
-+ printk(KERN_WARNING "pciback: driver data not found for %s\n",
-+ pci_name(dev));
-+ return XEN_PCI_ERR_op_failed;
-+ }
++ err = xenbus_printf(tr, dev->nodename, "accel-msg-channel",
++ "%u", vnic->msg_channel);
++ if (err != 0) {
++ EPRINTK("%s: event channel xenbus write failed %d\n",
++ __FUNCTION__, err);
++ xenbus_transaction_end(tr, 1);
++ goto fail_transaction;
++ }
+
-+ *value = bar->which ? bar->len_val : bar->val;
++ err = xenbus_printf(tr, dev->nodename, "accel-net-channel",
++ "%u", vnic->net_channel);
++ if (err != 0) {
++ EPRINTK("%s: net channel xenbus write failed %d\n",
++ __FUNCTION__, err);
++ xenbus_transaction_end(tr, 1);
++ goto fail_transaction;
++ }
+
-+ return 0;
-+}
++ err = xenbus_transaction_end(tr, 0);
++ } while (err == -EAGAIN);
+
-+static inline void read_dev_bar(struct pci_dev *dev,
-+ struct pci_bar_info *bar_info, int offset,
-+ u32 len_mask)
-+{
-+ pci_read_config_dword(dev, offset, &bar_info->val);
-+ pci_write_config_dword(dev, offset, len_mask);
-+ pci_read_config_dword(dev, offset, &bar_info->len_val);
-+ pci_write_config_dword(dev, offset, bar_info->val);
-+}
++ if (err != 0) {
++ EPRINTK("%s: Transaction end failed %d\n", __FUNCTION__, err);
++ goto fail_transaction;
++ }
+
-+static void *bar_init(struct pci_dev *dev, int offset)
-+{
-+ struct pci_bar_info *bar = kmalloc(sizeof(*bar), GFP_KERNEL);
++ DPRINTK("Completed setting up domU shared state\n");
+
-+ if (!bar)
-+ return ERR_PTR(-ENOMEM);
++ return 0;
+
-+ read_dev_bar(dev, bar, offset, ~0);
-+ bar->which = 0;
++fail_transaction:
+
-+ return bar;
-+}
++ unbind_from_irqhandler(vnic->net_channel_irq, vnic);
++fail_net_irq:
+
-+static void *rom_init(struct pci_dev *dev, int offset)
-+{
-+ struct pci_bar_info *bar = kmalloc(sizeof(*bar), GFP_KERNEL);
++ unbind_from_irqhandler(vnic->msg_channel_irq, vnic);
++fail_msg_irq:
+
-+ if (!bar)
-+ return ERR_PTR(-ENOMEM);
++ remove_named_grant(dev, "accel-ctrl-page", vnic->ctrl_page_gnt);
++fail_msg_page_grant:
+
-+ read_dev_bar(dev, bar, offset, ~PCI_ROM_ADDRESS_ENABLE);
-+ bar->which = 0;
++ remove_named_grant(dev, "accel-msg-page", vnic->msg_page_gnt);
++fail_ctrl_page_grant:
+
-+ return bar;
-+}
++ free_pages((unsigned long)vnic->shared_page, 1);
++ vnic->shared_page = NULL;
++fail_shared_page:
+
-+static void bar_reset(struct pci_dev *dev, int offset, void *data)
-+{
-+ struct pci_bar_info *bar = data;
++ netfront_accel_fini_bufs(vnic->rx_bufs);
++fail_rx_bufs:
+
-+ bar->which = 0;
-+}
++ netfront_accel_fini_bufs(vnic->tx_bufs);
++fail_tx_bufs:
+
-+static void bar_release(struct pci_dev *dev, int offset, void *data)
-+{
-+ kfree(data);
-+}
++ /* Undo the memory allocation created when we got the HELLO */
++ netfront_accel_free_buffer_mem(&vnic->bufpages,
++ vnic->rx_bufs,
++ vnic->tx_bufs);
+
-+static int interrupt_read(struct pci_dev *dev, int offset, u8 * value,
-+ void *data)
-+{
-+ *value = (u8) dev->irq;
++ DPRINTK("Failed to setup domU shared state with code %d\n", err);
+
-+ return 0;
++ return err;
+}
+
-+static int bist_write(struct pci_dev *dev, int offset, u8 value, void *data)
++
++static void vnic_remove_domU_shared_state(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
+{
-+ u8 cur_value;
-+ int err;
++ struct xenbus_transaction tr;
++
++ /*
++ * Don't remove any watches because we currently hold the
++ * mutex and the watches take the mutex.
++ */
+
-+ err = pci_read_config_byte(dev, offset, &cur_value);
-+ if (err)
-+ goto out;
++ DPRINTK("%s: removing event channel irq handlers %d %d\n",
++ __FUNCTION__, vnic->net_channel_irq, vnic->msg_channel_irq);
++ do {
++ if (xenbus_transaction_start(&tr) != 0)
++ break;
++ xenbus_rm(tr, dev->nodename, "accel-msg-channel");
++ xenbus_rm(tr, dev->nodename, "accel-net-channel");
++ } while (xenbus_transaction_end(tr, 0) == -EAGAIN);
+
-+ if ((cur_value & ~PCI_BIST_START) == (value & ~PCI_BIST_START)
-+ || value == PCI_BIST_START)
-+ err = pci_write_config_byte(dev, offset, value);
++ unbind_from_irqhandler(vnic->net_channel_irq, vnic);
++ unbind_from_irqhandler(vnic->msg_channel_irq, vnic);
+
-+ out:
-+ return err;
++ /* ungrant pages for msg channel */
++ remove_named_grant(dev, "accel-ctrl-page", vnic->ctrl_page_gnt);
++ remove_named_grant(dev, "accel-msg-page", vnic->msg_page_gnt);
++ free_pages((unsigned long)vnic->shared_page, 1);
++ vnic->shared_page = NULL;
++
++ /* ungrant pages for buffers, and free buffer memory */
++ netfront_accel_free_buffer_mem(&vnic->bufpages,
++ vnic->rx_bufs,
++ vnic->tx_bufs);
++ netfront_accel_fini_bufs(vnic->rx_bufs);
++ netfront_accel_fini_bufs(vnic->tx_bufs);
+}
+
-+static struct config_field header_common[] = {
-+ {
-+ .offset = PCI_COMMAND,
-+ .size = 2,
-+ .u.w.read = pciback_read_config_word,
-+ .u.w.write = command_write,
-+ },
-+ {
-+ .offset = PCI_INTERRUPT_LINE,
-+ .size = 1,
-+ .u.b.read = interrupt_read,
-+ },
-+ {
-+ .offset = PCI_INTERRUPT_PIN,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ },
-+ {
-+ /* Any side effects of letting driver domain control cache line? */
-+ .offset = PCI_CACHE_LINE_SIZE,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ .u.b.write = pciback_write_config_byte,
-+ },
-+ {
-+ .offset = PCI_LATENCY_TIMER,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ },
-+ {
-+ .offset = PCI_BIST,
-+ .size = 1,
-+ .u.b.read = pciback_read_config_byte,
-+ .u.b.write = bist_write,
-+ },
-+ {
-+ .size = 0,
-+ },
-+};
+
-+#define CFG_FIELD_BAR(reg_offset) \
-+ { \
-+ .offset = reg_offset, \
-+ .size = 4, \
-+ .init = bar_init, \
-+ .reset = bar_reset, \
-+ .release = bar_release, \
-+ .u.dw.read = bar_read, \
-+ .u.dw.write = bar_write, \
-+ }
++static void vnic_setup_dom0_shared_state(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
++{
++ DPRINTK("Setting up dom0 shared state\n");
+
-+#define CFG_FIELD_ROM(reg_offset) \
-+ { \
-+ .offset = reg_offset, \
-+ .size = 4, \
-+ .init = rom_init, \
-+ .reset = bar_reset, \
-+ .release = bar_release, \
-+ .u.dw.read = bar_read, \
-+ .u.dw.write = rom_write, \
-+ }
++ netfront_accel_vi_ctor(vnic);
+
-+static struct config_field header_0[] = {
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_0),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_1),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_2),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_3),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_4),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_5),
-+ CFG_FIELD_ROM(PCI_ROM_ADDRESS),
-+ {
-+ .size = 0,
-+ },
-+};
++ /*
++ * Message processing will be enabled when this function
++ * returns, but we might have missed an interrupt. Schedule a
++ * check just in case.
++ */
++ queue_work(netfront_accel_workqueue, &vnic->msg_from_bend);
++}
+
-+static struct config_field header_1[] = {
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_0),
-+ CFG_FIELD_BAR(PCI_BASE_ADDRESS_1),
-+ CFG_FIELD_ROM(PCI_ROM_ADDRESS1),
-+ {
-+ .size = 0,
-+ },
-+};
+
-+int pciback_config_header_add_fields(struct pci_dev *dev)
++static void vnic_remove_dom0_shared_state(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
+{
-+ int err;
++ DPRINTK("Removing dom0 shared state\n");
+
-+ err = pciback_config_add_fields(dev, header_common);
-+ if (err)
-+ goto out;
++ vnic_stop_fastpath(vnic);
+
-+ switch (dev->hdr_type) {
-+ case PCI_HEADER_TYPE_NORMAL:
-+ err = pciback_config_add_fields(dev, header_0);
-+ break;
++ netfront_accel_vi_dtor(vnic);
++}
+
-+ case PCI_HEADER_TYPE_BRIDGE:
-+ err = pciback_config_add_fields(dev, header_1);
-+ break;
+
-+ default:
-+ err = -EINVAL;
-+ printk(KERN_ERR "pciback: %s: Unsupported header type %d!\n",
-+ pci_name(dev), dev->hdr_type);
-+ break;
-+ }
++/*************************************************************************/
+
-+ out:
-+ return err;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_quirks.c linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.c
---- linux-2.6.18/drivers/xen/pciback/conf_space_quirks.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.c 2007-12-23 11:15:34.064602797 +0100
-@@ -0,0 +1,126 @@
+/*
-+ * PCI Backend - Handle special overlays for broken devices.
++ * The following code handles accelstate changes between the frontend
++ * and the backend. In response to transitions, calls the following
++ * functions in matching pairs:
++ *
++ * vnic_setup_domU_shared_state
++ * vnic_remove_domU_shared_state
++ *
++ * vnic_setup_dom0_shared_state
++ * vnic_remove_dom0_shared_state
++ *
++ * Valid state transitions for DomU are as follows:
++ *
++ * Closed->Init on probe or in response to Init from dom0
++ *
++ * Init->Connected in response to Init from dom0
++ * Init->Closing on error providing dom0 is in Init
++ * Init->Closed on remove or in response to Closing from dom0
++ *
++ * Connected->Closing on error/remove
++ * Connected->Closed in response to Closing from dom0
++ *
++ * Closing->Closed in response to Closing from dom0
+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ * Author: Chris Bookholt <hap10@epoch.ncsc.mil>
+ */
+
-+#include <linux/kernel.h>
-+#include <linux/pci.h>
-+#include "pciback.h"
-+#include "conf_space.h"
-+#include "conf_space_quirks.h"
-+
-+LIST_HEAD(pciback_quirks);
+
-+struct pciback_config_quirk *pciback_find_quirk(struct pci_dev *dev)
++/* Function to deal with Xenbus accel state change in backend */
++static void netfront_accel_backend_accel_changed(netfront_accel_vnic *vnic,
++ XenbusState backend_state)
+{
-+ struct pciback_config_quirk *tmp_quirk;
++ struct xenbus_device *dev = vnic->dev;
++ XenbusState frontend_state;
++ int state;
+
-+ list_for_each_entry(tmp_quirk, &pciback_quirks, quirks_list)
-+ if (pci_match_id(&tmp_quirk->devid, dev))
-+ goto out;
-+ tmp_quirk = NULL;
-+ printk(KERN_DEBUG
-+ "quirk didn't match any device pciback knows about\n");
-+ out:
-+ return tmp_quirk;
-+}
++ DPRINTK("%s: changing from %s to %s. nodename %s, otherend %s\n",
++ __FUNCTION__, xenbus_strstate(vnic->backend_state),
++ xenbus_strstate(backend_state), dev->nodename, dev->otherend);
+
-+static inline void register_quirk(struct pciback_config_quirk *quirk)
-+{
-+ list_add_tail(&quirk->quirks_list, &pciback_quirks);
-+}
++ /*
++ * Ignore duplicate state changes. This can happen if the
++ * backend changes state twice in quick succession and the
++ * first watch fires in the frontend after the second
++ * transition has completed.
++ */
++ if (vnic->backend_state == backend_state)
++ return;
+
-+int pciback_field_is_dup(struct pci_dev *dev, unsigned int reg)
-+{
-+ int ret = 0;
-+ struct pciback_dev_data *dev_data = pci_get_drvdata(dev);
-+ struct config_field_entry *cfg_entry;
++ vnic->backend_state = backend_state;
++ frontend_state = vnic->frontend_state;
+
-+ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
-+ if ( OFFSET(cfg_entry) == reg) {
-+ ret = 1;
-+ break;
++ switch (backend_state) {
++ case XenbusStateInitialising:
++ /*
++ * It's possible for us to miss the closed state from
++ * dom0, so do the work here.
++ */
++ if (vnic->domU_state_is_setup) {
++ vnic_remove_domU_shared_state(dev, vnic);
++ vnic->domU_state_is_setup = 0;
++ }
++
++ if (frontend_state != XenbusStateInitialising) {
++ /* Make sure the backend doesn't go away. */
++ frontend_state = XenbusStateInitialising;
++ net_accel_update_state(dev, frontend_state);
++ xenbus_scanf(XBT_NIL, dev->otherend, "accelstate", "%d", &state);
++ backend_state = (XenbusState)state;
++ if (backend_state != XenbusStateInitialising)
++ break;
+ }
-+ }
-+ return ret;
-+}
+
-+int pciback_config_quirks_add_field(struct pci_dev *dev, struct config_field
-+ *field)
-+{
-+ int err = 0;
-+
-+ switch (field->size) {
-+ case 1:
-+ field->u.b.read = pciback_read_config_byte;
-+ field->u.b.write = pciback_write_config_byte;
-+ break;
-+ case 2:
-+ field->u.w.read = pciback_read_config_word;
-+ field->u.w.write = pciback_write_config_word;
++ /* Start the new connection. */
++ if (!vnic->removing) {
++ BUG_ON(vnic->domU_state_is_setup);
++ if (vnic_setup_domU_shared_state(dev, vnic) == 0) {
++ vnic->domU_state_is_setup = 1;
++ frontend_state = XenbusStateConnected;
++ } else
++ frontend_state = XenbusStateClosing;
++ }
+ break;
-+ case 4:
-+ field->u.dw.read = pciback_read_config_dword;
-+ field->u.dw.write = pciback_write_config_dword;
++ case XenbusStateConnected:
++ if (vnic->domU_state_is_setup &&
++ !vnic->dom0_state_is_setup) {
++ vnic_setup_dom0_shared_state(dev, vnic);
++ vnic->dom0_state_is_setup = 1;
++ }
+ break;
+ default:
-+ err = -EINVAL;
-+ goto out;
++ case XenbusStateClosing:
++ if (vnic->dom0_state_is_setup) {
++ vnic_remove_dom0_shared_state(dev, vnic);
++ vnic->dom0_state_is_setup = 0;
++ }
++ frontend_state = XenbusStateClosed;
++ break;
++ case XenbusStateUnknown:
++ case XenbusStateClosed:
++ if (vnic->domU_state_is_setup) {
++ vnic_remove_domU_shared_state(dev, vnic);
++ vnic->domU_state_is_setup = 0;
++ }
++ break;
+ }
+
-+ pciback_config_add_field(dev, field);
++ if (frontend_state != vnic->frontend_state) {
++ DPRINTK("Switching from state %s (%d) to %s (%d)\n",
++ xenbus_strstate(vnic->frontend_state),
++ vnic->frontend_state,
++ xenbus_strstate(frontend_state), frontend_state);
++ vnic->frontend_state = frontend_state;
++ net_accel_update_state(dev, frontend_state);
++ }
+
-+ out:
-+ return err;
++ wake_up(&vnic->state_wait_queue);
+}
+
-+int pciback_config_quirks_init(struct pci_dev *dev)
++
++static void backend_accel_state_change(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
+{
-+ struct pciback_config_quirk *quirk;
-+ int ret = 0;
++ int state;
++ netfront_accel_vnic *vnic;
++ struct xenbus_device *dev;
+
-+ quirk = kzalloc(sizeof(*quirk), GFP_ATOMIC);
-+ if (!quirk) {
-+ ret = -ENOMEM;
-+ goto out;
-+ }
++ DPRINTK("%s\n", __FUNCTION__);
+
-+ quirk->devid.vendor = dev->vendor;
-+ quirk->devid.device = dev->device;
-+ quirk->devid.subvendor = dev->subsystem_vendor;
-+ quirk->devid.subdevice = dev->subsystem_device;
-+ quirk->devid.class = 0;
-+ quirk->devid.class_mask = 0;
-+ quirk->devid.driver_data = 0UL;
++ vnic = container_of(watch, struct netfront_accel_vnic,
++ backend_accel_watch);
+
-+ quirk->pdev = dev;
++ mutex_lock(&vnic->vnic_mutex);
+
-+ register_quirk(quirk);
-+ out:
-+ return ret;
++ dev = vnic->dev;
++
++ state = (int)XenbusStateUnknown;
++ xenbus_scanf(XBT_NIL, dev->otherend, "accelstate", "%d", &state);
++ netfront_accel_backend_accel_changed(vnic, state);
++
++ mutex_unlock(&vnic->vnic_mutex);
+}
+
-+void pciback_config_field_free(struct config_field *field)
++
++static int setup_dom0_accel_watch(struct xenbus_device *dev,
++ netfront_accel_vnic *vnic)
+{
-+ kfree(field);
++ int err;
++
++ DPRINTK("Setting watch on %s/%s\n", dev->otherend, "accelstate");
++
++ err = xenbus_watch_path2(dev, dev->otherend, "accelstate",
++ &vnic->backend_accel_watch,
++ backend_accel_state_change);
++ if (err) {
++ EPRINTK("%s: Failed to register xenbus watch: %d\n",
++ __FUNCTION__, err);
++ goto fail;
++ }
++ return 0;
++ fail:
++ vnic->backend_accel_watch.node = NULL;
++ return err;
+}
+
-+int pciback_config_quirk_release(struct pci_dev *dev)
++
++int netfront_accel_probe(struct net_device *net_dev, struct xenbus_device *dev)
+{
-+ struct pciback_config_quirk *quirk;
-+ int ret = 0;
++ netfront_accel_vnic *vnic;
++ int err;
+
-+ quirk = pciback_find_quirk(dev);
-+ if (!quirk) {
-+ ret = -ENXIO;
-+ goto out;
++ DPRINTK("Probe passed device %s\n", dev->nodename);
++
++ vnic = netfront_accel_vnic_ctor(net_dev, dev);
++ if (IS_ERR(vnic))
++ return PTR_ERR(vnic);
++
++ /*
++ * Setup a watch on the backend accel state. This sets things
++ * going.
++ */
++ err = setup_dom0_accel_watch(dev, vnic);
++ if (err) {
++ netfront_accel_vnic_dtor(vnic);
++ EPRINTK("%s: probe failed with code %d\n", __FUNCTION__, err);
++ return err;
+ }
+
-+ list_del(&quirk->quirks_list);
-+ kfree(quirk);
++ /*
++ * Indicate to the other end that we're ready to start unless
++ * the watch has already fired.
++ */
++ mutex_lock(&vnic->vnic_mutex);
++ VPRINTK("setup success, updating accelstate\n");
++ if (vnic->frontend_state == XenbusStateClosed) {
++ vnic->frontend_state = XenbusStateInitialising;
++ net_accel_update_state(dev, XenbusStateInitialising);
++ }
++ mutex_unlock(&vnic->vnic_mutex);
+
-+ out:
-+ return ret;
++ DPRINTK("Probe done device %s\n", dev->nodename);
++
++ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/conf_space_quirks.h linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.h
---- linux-2.6.18/drivers/xen/pciback/conf_space_quirks.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/conf_space_quirks.h 2007-12-23 11:15:34.064602797 +0100
-@@ -0,0 +1,35 @@
-+/*
-+ * PCI Backend - Data structures for special overlays for broken devices.
-+ *
-+ * Ryan Wilson <hap9@epoch.ncsc.mil>
-+ * Chris Bookholt <hap10@epoch.ncsc.mil>
-+ */
+
-+#ifndef __XEN_PCIBACK_CONF_SPACE_QUIRKS_H__
-+#define __XEN_PCIBACK_CONF_SPACE_QUIRKS_H__
+
-+#include <linux/pci.h>
-+#include <linux/list.h>
++int netfront_accel_remove(struct xenbus_device *dev)
++{
++ struct netfront_info *np =
++ (struct netfront_info *)dev->dev.driver_data;
++ netfront_accel_vnic *vnic = (netfront_accel_vnic *)np->accel_priv;
+
-+struct pciback_config_quirk {
-+ struct list_head quirks_list;
-+ struct pci_device_id devid;
-+ struct pci_dev *pdev;
-+};
++ DPRINTK("%s %s\n", __FUNCTION__, dev->nodename);
+
-+struct pciback_config_quirk *pciback_find_quirk(struct pci_dev *dev);
++ BUG_ON(vnic == NULL);
+
-+int pciback_config_quirks_add_field(struct pci_dev *dev, struct config_field
-+ *field);
++ mutex_lock(&vnic->vnic_mutex);
+
-+int pciback_config_quirks_remove_field(struct pci_dev *dev, int reg);
++ /* Reject any attempts to connect. */
++ vnic->removing = 1;
+
-+int pciback_config_quirks_init(struct pci_dev *dev);
++ /* Close any existing connection. */
++ if (vnic->frontend_state == XenbusStateConnected) {
++ vnic->frontend_state = XenbusStateClosing;
++ net_accel_update_state(dev, XenbusStateClosing);
++ }
+
-+void pciback_config_field_free(struct config_field *field);
++ mutex_unlock(&vnic->vnic_mutex);
+
-+int pciback_config_quirk_release(struct pci_dev *dev);
++ DPRINTK("%s waiting for release of %s\n", __FUNCTION__, dev->nodename);
+
-+int pciback_field_is_dup(struct pci_dev *dev, unsigned int reg);
++ /*
++ * Wait for the xenbus watch to release the shared resources.
++ * This indicates that dom0 has made the transition
++ * Closing->Closed or that dom0 was in Closed or Init and no
++ * resources were mapped.
++ */
++ wait_event(vnic->state_wait_queue,
++ !vnic->domU_state_is_setup);
+
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/controller.c linux-2.6.18-xen.hg/drivers/xen/pciback/controller.c
---- linux-2.6.18/drivers/xen/pciback/controller.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/controller.c 2007-12-23 11:15:34.064602797 +0100
-@@ -0,0 +1,404 @@
-+/*
-+ * Copyright (C) 2007 Hewlett-Packard Development Company, L.P.
-+ * Alex Williamson <alex.williamson@hp.com>
++ /*
++ * Now we don't need this watch anymore it is safe to remove
++ * it (and so synchronise with it completing if outstanding)
++ */
++ DPRINTK("%s: unregistering xenbus accel watch\n",
++ __FUNCTION__);
++ unregister_xenbus_watch(&vnic->backend_accel_watch);
++ kfree(vnic->backend_accel_watch.node);
++
++ netfront_accel_vnic_dtor(vnic);
++
++ DPRINTK("%s done %s\n", __FUNCTION__, dev->nodename);
++
++ return 0;
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/ef_vi_falcon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/ef_vi_falcon.h 2008-05-19 00:33:48.650950417 +0300
+@@ -0,0 +1,172 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * PCI "Controller" Backend - virtualize PCI bus topology based on PCI
-+ * controllers. Devices under the same PCI controller are exposed on the
-+ * same virtual domain:bus. Within a bus, device slots are virtualized
-+ * to compact the bus.
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
+ *
-+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License as published by
-+ * the Free Software Foundation; either version 2 of the License, or
-+ * (at your option) any later version.
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/acpi.h>
-+#include <linux/list.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include "pciback.h"
-+
-+#define PCI_MAX_BUSSES 255
-+#define PCI_MAX_SLOTS 32
-+
-+struct controller_dev_entry {
-+ struct list_head list;
-+ struct pci_dev *dev;
-+ unsigned int devfn;
-+};
-+
-+struct controller_list_entry {
-+ struct list_head list;
-+ struct pci_controller *controller;
-+ unsigned int domain;
-+ unsigned int bus;
-+ unsigned int next_devfn;
-+ struct list_head dev_list;
-+};
-+
-+struct controller_dev_data {
-+ struct list_head list;
-+ unsigned int next_domain;
-+ unsigned int next_bus;
-+ spinlock_t lock;
-+};
-+
-+struct walk_info {
-+ struct pciback_device *pdev;
-+ int resource_count;
-+ int root_num;
-+};
-+
-+struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus,
-+ unsigned int devfn)
-+{
-+ struct controller_dev_data *dev_data = pdev->pci_dev_data;
-+ struct controller_dev_entry *dev_entry;
-+ struct controller_list_entry *cntrl_entry;
-+ struct pci_dev *dev = NULL;
-+ unsigned long flags;
-+
-+ spin_lock_irqsave(&dev_data->lock, flags);
++/*
++ * \author slp
++ * \brief Falcon specific definitions
++ * \date 2004/08
++ */
+
-+ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
-+ if (cntrl_entry->domain != domain ||
-+ cntrl_entry->bus != bus)
-+ continue;
++#ifndef __EF_VI_FALCON_H__
++#define __EF_VI_FALCON_H__
+
-+ list_for_each_entry(dev_entry, &cntrl_entry->dev_list, list) {
-+ if (devfn == dev_entry->devfn) {
-+ dev = dev_entry->dev;
-+ goto found;
-+ }
-+ }
-+ }
-+found:
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
++#define EFHW_4K 0x00001000u
++#define EFHW_8K 0x00002000u
+
-+ return dev;
-+}
++/* include the autogenerated register definitions */
+
-+int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ struct controller_dev_data *dev_data = pdev->pci_dev_data;
-+ struct controller_dev_entry *dev_entry;
-+ struct controller_list_entry *cntrl_entry;
-+ struct pci_controller *dev_controller = PCI_CONTROLLER(dev);
-+ unsigned long flags;
-+ int ret = 0, found = 0;
++#include "ef_vi_falcon_core.h"
++#include "ef_vi_falcon_desc.h"
++#include "ef_vi_falcon_event.h"
+
-+ spin_lock_irqsave(&dev_data->lock, flags);
+
-+ /* Look to see if we already have a domain:bus for this controller */
-+ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
-+ if (cntrl_entry->controller == dev_controller) {
-+ found = 1;
-+ break;
-+ }
-+ }
++/*----------------------------------------------------------------------------
++ *
++ * Helpers to turn bit shifts into dword shifts and check that the bit fields
++ * haven't overflown the dword etc. Aim is to preserve consistency with the
++ * autogenerated headers - once stable we could hard code.
++ *
++ *---------------------------------------------------------------------------*/
+
-+ if (!found) {
-+ cntrl_entry = kmalloc(sizeof(*cntrl_entry), GFP_ATOMIC);
-+ if (!cntrl_entry) {
-+ ret = -ENOMEM;
-+ goto out;
-+ }
++/* mask constructors */
++#define __FALCON_MASK(WIDTH,T) ((((T)1) << (WIDTH)) - 1)
++#define __EFVI_MASK32(WIDTH) __FALCON_MASK((WIDTH),uint32_t)
++#define __EFVI_MASK64(WIDTH) __FALCON_MASK((WIDTH),uint64_t)
+
-+ cntrl_entry->controller = dev_controller;
-+ cntrl_entry->next_devfn = PCI_DEVFN(0, 0);
++#define __EFVI_FALCON_MASKFIELD32(LBN, WIDTH) ((uint32_t) \
++ (__EFVI_MASK32(WIDTH) << (LBN)))
+
-+ cntrl_entry->domain = dev_data->next_domain;
-+ cntrl_entry->bus = dev_data->next_bus++;
-+ if (dev_data->next_bus > PCI_MAX_BUSSES) {
-+ dev_data->next_domain++;
-+ dev_data->next_bus = 0;
-+ }
++/* constructors for fields which span the first and second dwords */
++#define __LW(LBN) (32 - LBN)
++#define LOW(v, LBN, WIDTH) ((uint32_t) \
++ (((v) & __EFVI_MASK64(__LW((LBN)))) << (LBN)))
++#define HIGH(v, LBN, WIDTH) ((uint32_t)(((v) >> __LW((LBN))) & \
++ __EFVI_MASK64((WIDTH - __LW((LBN))))))
++/* constructors for fields within the second dword */
++#define __DW2(LBN) ((LBN) - 32)
+
-+ INIT_LIST_HEAD(&cntrl_entry->dev_list);
++/* constructors for fields which span the second and third dwords */
++#define __LW2(LBN) (64 - LBN)
++#define LOW2(v, LBN, WIDTH) ((uint32_t) \
++ (((v) & __EFVI_MASK64(__LW2((LBN)))) << ((LBN) - 32)))
++#define HIGH2(v, LBN, WIDTH) ((uint32_t) \
++ (((v) >> __LW2((LBN))) & __EFVI_MASK64((WIDTH - __LW2((LBN))))))
+
-+ list_add_tail(&cntrl_entry->list, &dev_data->list);
-+ }
++/* constructors for fields within the third dword */
++#define __DW3(LBN) ((LBN) - 64)
+
-+ if (PCI_SLOT(cntrl_entry->next_devfn) > PCI_MAX_SLOTS) {
-+ /*
-+ * While it seems unlikely, this can actually happen if
-+ * a controller has P2P bridges under it.
-+ */
-+ xenbus_dev_fatal(pdev->xdev, -ENOSPC, "Virtual bus %04x:%02x "
-+ "is full, no room to export %04x:%02x:%02x.%x",
-+ cntrl_entry->domain, cntrl_entry->bus,
-+ pci_domain_nr(dev->bus), dev->bus->number,
-+ PCI_SLOT(dev->devfn), PCI_FUNC(dev->devfn));
-+ ret = -ENOSPC;
-+ goto out;
-+ }
++
++/* constructors for fields which span the third and fourth dwords */
++#define __LW3(LBN) (96 - LBN)
++#define LOW3(v, LBN, WIDTH) ((uint32_t) \
++ (((v) & __EFVI_MASK64(__LW3((LBN)))) << ((LBN) - 64)))
++#define HIGH3(v, LBN, WIDTH) ((unit32_t) \
++ (((v) >> __LW3((LBN))) & __EFVI_MASK64((WIDTH - __LW3((LBN))))))
+
-+ dev_entry = kmalloc(sizeof(*dev_entry), GFP_ATOMIC);
-+ if (!dev_entry) {
-+ if (list_empty(&cntrl_entry->dev_list)) {
-+ list_del(&cntrl_entry->list);
-+ kfree(cntrl_entry);
-+ }
-+ ret = -ENOMEM;
-+ goto out;
-+ }
++/* constructors for fields within the fourth dword */
++#define __DW4(LBN) ((LBN) - 96)
+
-+ dev_entry->dev = dev;
-+ dev_entry->devfn = cntrl_entry->next_devfn;
++/* checks that the autogenerated headers our consistent with our model */
++#define WIDTHCHCK(a, b) ef_assert((a) == (b))
++#define RANGECHCK(v, WIDTH) \
++ ef_assert(((uint64_t)(v) & ~(__EFVI_MASK64((WIDTH)))) == 0)
+
-+ list_add_tail(&dev_entry->list, &cntrl_entry->dev_list);
++/* fields within the first dword */
++#define DWCHCK(LBN, WIDTH) ef_assert(((LBN) >= 0) &&(((LBN)+(WIDTH)) <= 32))
+
-+ cntrl_entry->next_devfn += PCI_DEVFN(1, 0);
++/* fields which span the first and second dwords */
++#define LWCHK(LBN, WIDTH) ef_assert(WIDTH >= __LW(LBN))
+
-+out:
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
-+ return ret;
-+}
++/*----------------------------------------------------------------------------
++ *
++ * Buffer virtual addresses (4K buffers)
++ *
++ *---------------------------------------------------------------------------*/
+
-+void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ struct controller_dev_data *dev_data = pdev->pci_dev_data;
-+ struct controller_list_entry *cntrl_entry;
-+ struct controller_dev_entry *dev_entry = NULL;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
++/* Form a buffer virtual address from buffer ID and offset. If the offset
++** is larger than the buffer size, then the buffer indexed will be
++** calculated appropriately. It is the responsibility of the caller to
++** ensure that they have valid buffers programmed at that address.
++*/
++#define EFVI_FALCON_VADDR_4K_S (12)
++#define EFVI_FALCON_VADDR_M 0xfffff /* post shift mask */
+
-+ spin_lock_irqsave(&dev_data->lock, flags);
+
-+ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
-+ if (cntrl_entry->controller != PCI_CONTROLLER(dev))
-+ continue;
++#define EFVI_FALCON_BUFFER_4K_ADDR(id,off) \
++ (((id) << EFVI_FALCON_VADDR_4K_S) + (off))
+
-+ list_for_each_entry(dev_entry, &cntrl_entry->dev_list, list) {
-+ if (dev_entry->dev == dev) {
-+ found_dev = dev_entry->dev;
-+ break;
-+ }
-+ }
-+ }
++#define EFVI_FALCON_BUFFER_4K_PAGE(vaddr) \
++ (((vaddr) >> EFVI_FALCON_VADDR_4K_S) & EFVI_FALCON_VADDR_M)
+
-+ if (!found_dev) {
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
-+ return;
-+ }
++#define EFVI_FALCON_BUFFER_4K_OFF(vaddr) \
++ ((vaddr) & __EFVI_MASK32(EFVI_FALCON_VADDR_4K_S))
++
++
++/*----------------------------------------------------------------------------
++ *
++ * Masks
++ *
++ *---------------------------------------------------------------------------*/
++
++#define EFVI_FALCON_CLOCK_ASIC_HZ (125000)
++#define EFVI_FALCON_CLOCK_FPGA_HZ (62500)
++#define EFVI_FALCON_CLOCK_HZ EFVI_FALCON_CLOCK_ASIC_HZ
++
++
++/*----------------------------------------------------------------------------
++ *
++ * Timers
++ *
++ *---------------------------------------------------------------------------*/
++
++/* Event-Queue Timer granularity - measured in us
++ Given by: 4096 * 3 cycle * clock period */
++
++#define EFVI_FALCON_EVQTIMER_PERIOD_US ((4096 * 3 * 1000) / EFVI_FALCON_CLOCK_HZ)
++
++/* mode bits */
++#define EFVI_FALCON_TIMER_MODE_DIS 0 /* disabled */
++#define EFVI_FALCON_TIMER_MODE_RUN 1 /* started counting right away */
++#define EFVI_FALCON_TIMER_MODE_HOLD 2 /* trigger mode (user queues) */
++
++#define EFVI_FALCON_EVQTIMER_HOLD (EFVI_FALCON_TIMER_MODE_HOLD << TIMER_MODE_LBN)
++#define EFVI_FALCON_EVQTIMER_RUN (EFVI_FALCON_TIMER_MODE_RUN << TIMER_MODE_LBN)
++#define EFVI_FALCON_EVQTIMER_DISABLE (EFVI_FALCON_TIMER_MODE_DIS << TIMER_MODE_LBN)
++
++
++/* ---- efhw_event_t helpers --- */
++
++#define EFVI_FALCON_EVENT_CODE(evp) \
++ ((evp)->u64 & EFVI_FALCON_EVENT_CODE_MASK)
++
++#define EFVI_FALCON_EVENT_SW_DATA_MASK 0x0000ffff
++
++#define __EFVI_FALCON_OPEN_MASK(WIDTH) ((((uint64_t)1) << (WIDTH)) - 1)
++
++#define EFVI_FALCON_EVENT_CODE_MASK \
++ (__EFVI_FALCON_OPEN_MASK(EV_CODE_WIDTH) << EV_CODE_LBN)
++
++
++#endif /* __EF_VI_FALCON_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/ef_vi_falcon_core.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/ef_vi_falcon_core.h 2008-05-19 00:33:48.654950647 +0300
+@@ -0,0 +1,1075 @@
++
++#define EFVI_FALCON_EXTENDED_P_BAR 1
++
++//////////////---- Bus Interface Unit Registers C Header ----//////////////
++#define IOM_IND_ADR_REG_OFST 0x0 // IO-mapped indirect access address register
++ #define IOM_AUTO_ADR_INC_EN_LBN 16
++ #define IOM_AUTO_ADR_INC_EN_WIDTH 1
++ #define IOM_IND_ADR_LBN 0
++ #define IOM_IND_ADR_WIDTH 16
++#define IOM_IND_DAT_REG_OFST 0x4 // IO-mapped indirect access data register
++ #define IOM_IND_DAT_LBN 0
++ #define IOM_IND_DAT_WIDTH 32
++#define ADR_REGION_REG_KER_OFST 0x0 // Address region register
++#define ADR_REGION_REG_OFST 0x0 // Address region register
++ #define ADR_REGION3_LBN 96
++ #define ADR_REGION3_WIDTH 18
++ #define ADR_REGION2_LBN 64
++ #define ADR_REGION2_WIDTH 18
++ #define ADR_REGION1_LBN 32
++ #define ADR_REGION1_WIDTH 18
++ #define ADR_REGION0_LBN 0
++ #define ADR_REGION0_WIDTH 18
++#define INT_EN_REG_KER_OFST 0x10 // Kernel driver Interrupt enable register
++ #define KER_INT_CHAR_LBN 4
++ #define KER_INT_CHAR_WIDTH 1
++ #define KER_INT_KER_LBN 3
++ #define KER_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_EN_KER_LBN 2
++ #define ILL_ADR_ERR_INT_EN_KER_WIDTH 1
++ #define SRM_PERR_INT_EN_KER_LBN 1
++ #define SRM_PERR_INT_EN_KER_WIDTH 1
++ #define DRV_INT_EN_KER_LBN 0
++ #define DRV_INT_EN_KER_WIDTH 1
++#define INT_EN_REG_CHAR_OFST 0x20 // Char Driver interrupt enable register
++ #define CHAR_INT_CHAR_LBN 4
++ #define CHAR_INT_CHAR_WIDTH 1
++ #define CHAR_INT_KER_LBN 3
++ #define CHAR_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_EN_CHAR_LBN 2
++ #define ILL_ADR_ERR_INT_EN_CHAR_WIDTH 1
++ #define SRM_PERR_INT_EN_CHAR_LBN 1
++ #define SRM_PERR_INT_EN_CHAR_WIDTH 1
++ #define DRV_INT_EN_CHAR_LBN 0
++ #define DRV_INT_EN_CHAR_WIDTH 1
++#define INT_ADR_REG_KER_OFST 0x30 // Interrupt host address for Kernel driver
++ #define INT_ADR_KER_LBN 0
++ #define INT_ADR_KER_WIDTH 64
++ #define DRV_INT_KER_LBN 32
++ #define DRV_INT_KER_WIDTH 1
++ #define EV_FF_HALF_INT_KER_LBN 3
++ #define EV_FF_HALF_INT_KER_WIDTH 1
++ #define EV_FF_FULL_INT_KER_LBN 2
++ #define EV_FF_FULL_INT_KER_WIDTH 1
++ #define ILL_ADR_ERR_INT_KER_LBN 1
++ #define ILL_ADR_ERR_INT_KER_WIDTH 1
++ #define SRAM_PERR_INT_KER_LBN 0
++ #define SRAM_PERR_INT_KER_WIDTH 1
++#define INT_ADR_REG_CHAR_OFST 0x40 // Interrupt host address for Char driver
++ #define INT_ADR_CHAR_LBN 0
++ #define INT_ADR_CHAR_WIDTH 64
++ #define DRV_INT_CHAR_LBN 32
++ #define DRV_INT_CHAR_WIDTH 1
++ #define EV_FF_HALF_INT_CHAR_LBN 3
++ #define EV_FF_HALF_INT_CHAR_WIDTH 1
++ #define EV_FF_FULL_INT_CHAR_LBN 2
++ #define EV_FF_FULL_INT_CHAR_WIDTH 1
++ #define ILL_ADR_ERR_INT_CHAR_LBN 1
++ #define ILL_ADR_ERR_INT_CHAR_WIDTH 1
++ #define SRAM_PERR_INT_CHAR_LBN 0
++ #define SRAM_PERR_INT_CHAR_WIDTH 1
++#define INT_ISR0_B0_OFST 0x90 // B0 only
++#define INT_ISR1_B0_OFST 0xA0
++#define INT_ACK_REG_KER_A1_OFST 0x50 // Kernel interrupt acknowledge register
++ #define RESERVED_LBN 0
++ #define RESERVED_WIDTH 32
++#define INT_ACK_REG_CHAR_A1_OFST 0x60 // CHAR interrupt acknowledge register
++ #define RESERVED_LBN 0
++ #define RESERVED_WIDTH 32
++//////////////---- Global CSR Registers C Header ----//////////////
++#define STRAP_REG_KER_OFST 0x200 // ASIC strap status register
++#define STRAP_REG_OFST 0x200 // ASIC strap status register
++ #define ONCHIP_SRAM_LBN 16
++ #define ONCHIP_SRAM_WIDTH 0
++ #define STRAP_ISCSI_EN_LBN 3
++ #define STRAP_ISCSI_EN_WIDTH 1
++ #define STRAP_PINS_LBN 0
++ #define STRAP_PINS_WIDTH 3
++#define GPIO_CTL_REG_KER_OFST 0x210 // GPIO control register
++#define GPIO_CTL_REG_OFST 0x210 // GPIO control register
++ #define GPIO_OEN_LBN 24
++ #define GPIO_OEN_WIDTH 4
++ #define GPIO_OUT_LBN 16
++ #define GPIO_OUT_WIDTH 4
++ #define GPIO_IN_LBN 8
++ #define GPIO_IN_WIDTH 4
++ #define GPIO_PWRUP_VALUE_LBN 0
++ #define GPIO_PWRUP_VALUE_WIDTH 4
++#define GLB_CTL_REG_KER_OFST 0x220 // Global control register
++#define GLB_CTL_REG_OFST 0x220 // Global control register
++ #define SWRST_LBN 0
++ #define SWRST_WIDTH 1
++#define FATAL_INTR_REG_KER_OFST 0x230 // Fatal interrupt register for Kernel
++ #define PCI_BUSERR_INT_KER_EN_LBN 43
++ #define PCI_BUSERR_INT_KER_EN_WIDTH 1
++ #define SRAM_OOB_INT_KER_EN_LBN 42
++ #define SRAM_OOB_INT_KER_EN_WIDTH 1
++ #define BUFID_OOB_INT_KER_EN_LBN 41
++ #define BUFID_OOB_INT_KER_EN_WIDTH 1
++ #define MEM_PERR_INT_KER_EN_LBN 40
++ #define MEM_PERR_INT_KER_EN_WIDTH 1
++ #define RBUF_OWN_INT_KER_EN_LBN 39
++ #define RBUF_OWN_INT_KER_EN_WIDTH 1
++ #define TBUF_OWN_INT_KER_EN_LBN 38
++ #define TBUF_OWN_INT_KER_EN_WIDTH 1
++ #define RDESCQ_OWN_INT_KER_EN_LBN 37
++ #define RDESCQ_OWN_INT_KER_EN_WIDTH 1
++ #define TDESCQ_OWN_INT_KER_EN_LBN 36
++ #define TDESCQ_OWN_INT_KER_EN_WIDTH 1
++ #define EVQ_OWN_INT_KER_EN_LBN 35
++ #define EVQ_OWN_INT_KER_EN_WIDTH 1
++ #define EVFF_OFLO_INT_KER_EN_LBN 34
++ #define EVFF_OFLO_INT_KER_EN_WIDTH 1
++ #define ILL_ADR_INT_KER_EN_LBN 33
++ #define ILL_ADR_INT_KER_EN_WIDTH 1
++ #define SRM_PERR_INT_KER_EN_LBN 32
++ #define SRM_PERR_INT_KER_EN_WIDTH 1
++ #define PCI_BUSERR_INT_KER_LBN 11
++ #define PCI_BUSERR_INT_KER_WIDTH 1
++ #define SRAM_OOB_INT_KER_LBN 10
++ #define SRAM_OOB_INT_KER_WIDTH 1
++ #define BUFID_OOB_INT_KER_LBN 9
++ #define BUFID_OOB_INT_KER_WIDTH 1
++ #define MEM_PERR_INT_KER_LBN 8
++ #define MEM_PERR_INT_KER_WIDTH 1
++ #define RBUF_OWN_INT_KER_LBN 7
++ #define RBUF_OWN_INT_KER_WIDTH 1
++ #define TBUF_OWN_INT_KER_LBN 6
++ #define TBUF_OWN_INT_KER_WIDTH 1
++ #define RDESCQ_OWN_INT_KER_LBN 5
++ #define RDESCQ_OWN_INT_KER_WIDTH 1
++ #define TDESCQ_OWN_INT_KER_LBN 4
++ #define TDESCQ_OWN_INT_KER_WIDTH 1
++ #define EVQ_OWN_INT_KER_LBN 3
++ #define EVQ_OWN_INT_KER_WIDTH 1
++ #define EVFF_OFLO_INT_KER_LBN 2
++ #define EVFF_OFLO_INT_KER_WIDTH 1
++ #define ILL_ADR_INT_KER_LBN 1
++ #define ILL_ADR_INT_KER_WIDTH 1
++ #define SRM_PERR_INT_KER_LBN 0
++ #define SRM_PERR_INT_KER_WIDTH 1
++#define FATAL_INTR_REG_OFST 0x240 // Fatal interrupt register for Char
++ #define PCI_BUSERR_INT_CHAR_EN_LBN 43
++ #define PCI_BUSERR_INT_CHAR_EN_WIDTH 1
++ #define SRAM_OOB_INT_CHAR_EN_LBN 42
++ #define SRAM_OOB_INT_CHAR_EN_WIDTH 1
++ #define BUFID_OOB_INT_CHAR_EN_LBN 41
++ #define BUFID_OOB_INT_CHAR_EN_WIDTH 1
++ #define MEM_PERR_INT_CHAR_EN_LBN 40
++ #define MEM_PERR_INT_CHAR_EN_WIDTH 1
++ #define RBUF_OWN_INT_CHAR_EN_LBN 39
++ #define RBUF_OWN_INT_CHAR_EN_WIDTH 1
++ #define TBUF_OWN_INT_CHAR_EN_LBN 38
++ #define TBUF_OWN_INT_CHAR_EN_WIDTH 1
++ #define RDESCQ_OWN_INT_CHAR_EN_LBN 37
++ #define RDESCQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define TDESCQ_OWN_INT_CHAR_EN_LBN 36
++ #define TDESCQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define EVQ_OWN_INT_CHAR_EN_LBN 35
++ #define EVQ_OWN_INT_CHAR_EN_WIDTH 1
++ #define EVFF_OFLO_INT_CHAR_EN_LBN 34
++ #define EVFF_OFLO_INT_CHAR_EN_WIDTH 1
++ #define ILL_ADR_INT_CHAR_EN_LBN 33
++ #define ILL_ADR_INT_CHAR_EN_WIDTH 1
++ #define SRM_PERR_INT_CHAR_EN_LBN 32
++ #define SRM_PERR_INT_CHAR_EN_WIDTH 1
++ #define FATAL_INTR_REG_EN_BITS 0xffffffffffffffffULL
++ #define PCI_BUSERR_INT_CHAR_LBN 11
++ #define PCI_BUSERR_INT_CHAR_WIDTH 1
++ #define SRAM_OOB_INT_CHAR_LBN 10
++ #define SRAM_OOB_INT_CHAR_WIDTH 1
++ #define BUFID_OOB_INT_CHAR_LBN 9
++ #define BUFID_OOB_INT_CHAR_WIDTH 1
++ #define MEM_PERR_INT_CHAR_LBN 8
++ #define MEM_PERR_INT_CHAR_WIDTH 1
++ #define RBUF_OWN_INT_CHAR_LBN 7
++ #define RBUF_OWN_INT_CHAR_WIDTH 1
++ #define TBUF_OWN_INT_CHAR_LBN 6
++ #define TBUF_OWN_INT_CHAR_WIDTH 1
++ #define RDESCQ_OWN_INT_CHAR_LBN 5
++ #define RDESCQ_OWN_INT_CHAR_WIDTH 1
++ #define TDESCQ_OWN_INT_CHAR_LBN 4
++ #define TDESCQ_OWN_INT_CHAR_WIDTH 1
++ #define EVQ_OWN_INT_CHAR_LBN 3
++ #define EVQ_OWN_INT_CHAR_WIDTH 1
++ #define EVFF_OFLO_INT_CHAR_LBN 2
++ #define EVFF_OFLO_INT_CHAR_WIDTH 1
++ #define ILL_ADR_INT_CHAR_LBN 1
++ #define ILL_ADR_INT_CHAR_WIDTH 1
++ #define SRM_PERR_INT_CHAR_LBN 0
++ #define SRM_PERR_INT_CHAR_WIDTH 1
++#define DP_CTRL_REG_OFST 0x250 // Datapath control register
++ #define FLS_EVQ_ID_LBN 0
++ #define FLS_EVQ_ID_WIDTH 12
++#define MEM_STAT_REG_KER_OFST 0x260 // Memory status register
++#define MEM_STAT_REG_OFST 0x260 // Memory status register
++ #define MEM_PERR_VEC_LBN 53
++ #define MEM_PERR_VEC_WIDTH 38
++ #define MBIST_CORR_LBN 38
++ #define MBIST_CORR_WIDTH 15
++ #define MBIST_ERR_LBN 0
++ #define MBIST_ERR_WIDTH 38
++#define DEBUG_REG_KER_OFST 0x270 // Debug register
++#define DEBUG_REG_OFST 0x270 // Debug register
++ #define DEBUG_BLK_SEL2_LBN 47
++ #define DEBUG_BLK_SEL2_WIDTH 3
++ #define DEBUG_BLK_SEL1_LBN 44
++ #define DEBUG_BLK_SEL1_WIDTH 3
++ #define DEBUG_BLK_SEL0_LBN 41
++ #define DEBUG_BLK_SEL0_WIDTH 3
++ #define MISC_DEBUG_ADDR_LBN 36
++ #define MISC_DEBUG_ADDR_WIDTH 5
++ #define SERDES_DEBUG_ADDR_LBN 31
++ #define SERDES_DEBUG_ADDR_WIDTH 5
++ #define EM_DEBUG_ADDR_LBN 26
++ #define EM_DEBUG_ADDR_WIDTH 5
++ #define SR_DEBUG_ADDR_LBN 21
++ #define SR_DEBUG_ADDR_WIDTH 5
++ #define EV_DEBUG_ADDR_LBN 16
++ #define EV_DEBUG_ADDR_WIDTH 5
++ #define RX_DEBUG_ADDR_LBN 11
++ #define RX_DEBUG_ADDR_WIDTH 5
++ #define TX_DEBUG_ADDR_LBN 6
++ #define TX_DEBUG_ADDR_WIDTH 5
++ #define BIU_DEBUG_ADDR_LBN 1
++ #define BIU_DEBUG_ADDR_WIDTH 5
++ #define DEBUG_EN_LBN 0
++ #define DEBUG_EN_WIDTH 1
++#define DRIVER_REG0_KER_OFST 0x280 // Driver scratch register 0
++#define DRIVER_REG0_OFST 0x280 // Driver scratch register 0
++ #define DRIVER_DW0_LBN 0
++ #define DRIVER_DW0_WIDTH 32
++#define DRIVER_REG1_KER_OFST 0x290 // Driver scratch register 1
++#define DRIVER_REG1_OFST 0x290 // Driver scratch register 1
++ #define DRIVER_DW1_LBN 0
++ #define DRIVER_DW1_WIDTH 32
++#define DRIVER_REG2_KER_OFST 0x2A0 // Driver scratch register 2
++#define DRIVER_REG2_OFST 0x2A0 // Driver scratch register 2
++ #define DRIVER_DW2_LBN 0
++ #define DRIVER_DW2_WIDTH 32
++#define DRIVER_REG3_KER_OFST 0x2B0 // Driver scratch register 3
++#define DRIVER_REG3_OFST 0x2B0 // Driver scratch register 3
++ #define DRIVER_DW3_LBN 0
++ #define DRIVER_DW3_WIDTH 32
++#define DRIVER_REG4_KER_OFST 0x2C0 // Driver scratch register 4
++#define DRIVER_REG4_OFST 0x2C0 // Driver scratch register 4
++ #define DRIVER_DW4_LBN 0
++ #define DRIVER_DW4_WIDTH 32
++#define DRIVER_REG5_KER_OFST 0x2D0 // Driver scratch register 5
++#define DRIVER_REG5_OFST 0x2D0 // Driver scratch register 5
++ #define DRIVER_DW5_LBN 0
++ #define DRIVER_DW5_WIDTH 32
++#define DRIVER_REG6_KER_OFST 0x2E0 // Driver scratch register 6
++#define DRIVER_REG6_OFST 0x2E0 // Driver scratch register 6
++ #define DRIVER_DW6_LBN 0
++ #define DRIVER_DW6_WIDTH 32
++#define DRIVER_REG7_KER_OFST 0x2F0 // Driver scratch register 7
++#define DRIVER_REG7_OFST 0x2F0 // Driver scratch register 7
++ #define DRIVER_DW7_LBN 0
++ #define DRIVER_DW7_WIDTH 32
++#define ALTERA_BUILD_REG_OFST 0x300 // Altera build register
++#define ALTERA_BUILD_REG_OFST 0x300 // Altera build register
++ #define ALTERA_BUILD_VER_LBN 0
++ #define ALTERA_BUILD_VER_WIDTH 32
++
++/* so called CSR spare register
++ - contains separate parity enable bits for the various internal memory blocks */
++#define MEM_PARITY_ERR_EN_REG_KER 0x310
++#define MEM_PARITY_ALL_BLOCKS_EN_LBN 64
++#define MEM_PARITY_ALL_BLOCKS_EN_WIDTH 38
++#define MEM_PARITY_TX_DATA_EN_LBN 72
++#define MEM_PARITY_TX_DATA_EN_WIDTH 2
++
++//////////////---- Event & Timer Module Registers C Header ----//////////////
++
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define EVQ_RPTR_REG_KER_OFST 0x11B00 // Event queue read pointer register
++#else
++#define EVQ_RPTR_REG_KER_OFST 0x1B00 // Event queue read pointer register
++#endif
+
-+ list_del(&dev_entry->list);
-+ kfree(dev_entry);
++#define EVQ_RPTR_REG_OFST 0xFA0000 // Event queue read pointer register array.
++ #define EVQ_RPTR_LBN 0
++ #define EVQ_RPTR_WIDTH 15
+
-+ if (list_empty(&cntrl_entry->dev_list)) {
-+ list_del(&cntrl_entry->list);
-+ kfree(cntrl_entry);
-+ }
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define EVQ_PTR_TBL_KER_OFST 0x11A00 // Event queue pointer table for kernel access
++#else
++#define EVQ_PTR_TBL_KER_OFST 0x1A00 // Event queue pointer table for kernel access
++#endif
++
++#define EVQ_PTR_TBL_CHAR_OFST 0xF60000 // Event queue pointer table for char direct access
++ #define EVQ_WKUP_OR_INT_EN_LBN 39
++ #define EVQ_WKUP_OR_INT_EN_WIDTH 1
++ #define EVQ_NXT_WPTR_LBN 24
++ #define EVQ_NXT_WPTR_WIDTH 15
++ #define EVQ_EN_LBN 23
++ #define EVQ_EN_WIDTH 1
++ #define EVQ_SIZE_LBN 20
++ #define EVQ_SIZE_WIDTH 3
++ #define EVQ_BUF_BASE_ID_LBN 0
++ #define EVQ_BUF_BASE_ID_WIDTH 20
++#define TIMER_CMD_REG_KER_OFST 0x420 // Timer table for kernel access. Page-mapped
++#define TIMER_CMD_REG_PAGE4_OFST 0x8420 // Timer table for user-level access. Page-mapped. For lowest 1K queues.
++#define TIMER_CMD_REG_PAGE123K_OFST 0x1000420 // Timer table for user-level access. Page-mapped. For upper 3K queues.
++#define TIMER_TBL_OFST 0xF70000 // Timer table for char driver direct access
++ #define TIMER_MODE_LBN 12
++ #define TIMER_MODE_WIDTH 2
++ #define TIMER_VAL_LBN 0
++ #define TIMER_VAL_WIDTH 12
++ #define TIMER_MODE_INT_HLDOFF 2
++ #define EVQ_BUF_SIZE_LBN 0
++ #define EVQ_BUF_SIZE_WIDTH 1
++#define DRV_EV_REG_KER_OFST 0x440 // Driver generated event register
++#define DRV_EV_REG_OFST 0x440 // Driver generated event register
++ #define DRV_EV_QID_LBN 64
++ #define DRV_EV_QID_WIDTH 12
++ #define DRV_EV_DATA_LBN 0
++ #define DRV_EV_DATA_WIDTH 64
++#define EVQ_CTL_REG_KER_OFST 0x450 // Event queue control register
++#define EVQ_CTL_REG_OFST 0x450 // Event queue control register
++ #define RX_EVQ_WAKEUP_MASK_B0_LBN 15
++ #define RX_EVQ_WAKEUP_MASK_B0_WIDTH 6
++ #define EVQ_OWNERR_CTL_LBN 14
++ #define EVQ_OWNERR_CTL_WIDTH 1
++ #define EVQ_FIFO_AF_TH_LBN 8
++ #define EVQ_FIFO_AF_TH_WIDTH 6
++ #define EVQ_FIFO_NOTAF_TH_LBN 0
++ #define EVQ_FIFO_NOTAF_TH_WIDTH 6
++//////////////---- SRAM Module Registers C Header ----//////////////
++#define BUF_TBL_CFG_REG_KER_OFST 0x600 // Buffer table configuration register
++#define BUF_TBL_CFG_REG_OFST 0x600 // Buffer table configuration register
++ #define BUF_TBL_MODE_LBN 3
++ #define BUF_TBL_MODE_WIDTH 1
++#define SRM_RX_DC_CFG_REG_KER_OFST 0x610 // SRAM receive descriptor cache configuration register
++#define SRM_RX_DC_CFG_REG_OFST 0x610 // SRAM receive descriptor cache configuration register
++ #define SRM_RX_DC_BASE_ADR_LBN 0
++ #define SRM_RX_DC_BASE_ADR_WIDTH 21
++#define SRM_TX_DC_CFG_REG_KER_OFST 0x620 // SRAM transmit descriptor cache configuration register
++#define SRM_TX_DC_CFG_REG_OFST 0x620 // SRAM transmit descriptor cache configuration register
++ #define SRM_TX_DC_BASE_ADR_LBN 0
++ #define SRM_TX_DC_BASE_ADR_WIDTH 21
++#define SRM_CFG_REG_KER_OFST 0x630 // SRAM configuration register
++#define SRM_CFG_REG_OFST 0x630 // SRAM configuration register
++ #define SRAM_OOB_ADR_INTEN_LBN 5
++ #define SRAM_OOB_ADR_INTEN_WIDTH 1
++ #define SRAM_OOB_BUF_INTEN_LBN 4
++ #define SRAM_OOB_BUF_INTEN_WIDTH 1
++ #define SRAM_BT_INIT_EN_LBN 3
++ #define SRAM_BT_INIT_EN_WIDTH 1
++ #define SRM_NUM_BANK_LBN 2
++ #define SRM_NUM_BANK_WIDTH 1
++ #define SRM_BANK_SIZE_LBN 0
++ #define SRM_BANK_SIZE_WIDTH 2
++#define BUF_TBL_UPD_REG_KER_OFST 0x650 // Buffer table update register
++#define BUF_TBL_UPD_REG_OFST 0x650 // Buffer table update register
++ #define BUF_UPD_CMD_LBN 63
++ #define BUF_UPD_CMD_WIDTH 1
++ #define BUF_CLR_CMD_LBN 62
++ #define BUF_CLR_CMD_WIDTH 1
++ #define BUF_CLR_END_ID_LBN 32
++ #define BUF_CLR_END_ID_WIDTH 20
++ #define BUF_CLR_START_ID_LBN 0
++ #define BUF_CLR_START_ID_WIDTH 20
++#define SRM_UPD_EVQ_REG_KER_OFST 0x660 // Buffer table update register
++#define SRM_UPD_EVQ_REG_OFST 0x660 // Buffer table update register
++ #define SRM_UPD_EVQ_ID_LBN 0
++ #define SRM_UPD_EVQ_ID_WIDTH 12
++#define SRAM_PARITY_REG_KER_OFST 0x670 // SRAM parity register.
++#define SRAM_PARITY_REG_OFST 0x670 // SRAM parity register.
++ #define FORCE_SRAM_PERR_LBN 0
++ #define FORCE_SRAM_PERR_WIDTH 1
++
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define BUF_HALF_TBL_KER_OFST 0x18000 // Buffer table in half buffer table mode direct access by kernel driver
++#else
++#define BUF_HALF_TBL_KER_OFST 0x8000 // Buffer table in half buffer table mode direct access by kernel driver
++#endif
+
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
-+ pcistub_put_pci_dev(found_dev);
-+}
+
-+int pciback_init_devices(struct pciback_device *pdev)
-+{
-+ struct controller_dev_data *dev_data;
++#define BUF_HALF_TBL_OFST 0x800000 // Buffer table in half buffer table mode direct access by char driver
++ #define BUF_ADR_HBUF_ODD_LBN 44
++ #define BUF_ADR_HBUF_ODD_WIDTH 20
++ #define BUF_OWNER_ID_HBUF_ODD_LBN 32
++ #define BUF_OWNER_ID_HBUF_ODD_WIDTH 12
++ #define BUF_ADR_HBUF_EVEN_LBN 12
++ #define BUF_ADR_HBUF_EVEN_WIDTH 20
++ #define BUF_OWNER_ID_HBUF_EVEN_LBN 0
++ #define BUF_OWNER_ID_HBUF_EVEN_WIDTH 12
+
-+ dev_data = kmalloc(sizeof(*dev_data), GFP_KERNEL);
-+ if (!dev_data)
-+ return -ENOMEM;
+
-+ spin_lock_init(&dev_data->lock);
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define BUF_FULL_TBL_KER_OFST 0x18000 // Buffer table in full buffer table mode direct access by kernel driver
++#else
++#define BUF_FULL_TBL_KER_OFST 0x8000 // Buffer table in full buffer table mode direct access by kernel driver
++#endif
+
-+ INIT_LIST_HEAD(&dev_data->list);
+
-+ /* Starting domain:bus numbers */
-+ dev_data->next_domain = 0;
-+ dev_data->next_bus = 0;
+
-+ pdev->pci_dev_data = dev_data;
+
-+ return 0;
-+}
++#define BUF_FULL_TBL_OFST 0x800000 // Buffer table in full buffer table mode direct access by char driver
++ #define IP_DAT_BUF_SIZE_LBN 50
++ #define IP_DAT_BUF_SIZE_WIDTH 1
++ #define BUF_ADR_REGION_LBN 48
++ #define BUF_ADR_REGION_WIDTH 2
++ #define BUF_ADR_FBUF_LBN 14
++ #define BUF_ADR_FBUF_WIDTH 34
++ #define BUF_OWNER_ID_FBUF_LBN 0
++ #define BUF_OWNER_ID_FBUF_WIDTH 14
++#define SRM_DBG_REG_OFST 0x3000000 // SRAM debug access
++ #define SRM_DBG_LBN 0
++ #define SRM_DBG_WIDTH 64
++//////////////---- RX Datapath Registers C Header ----//////////////
+
-+static acpi_status write_xenbus_resource(struct acpi_resource *res, void *data)
-+{
-+ struct walk_info *info = data;
-+ struct acpi_resource_address64 addr;
-+ acpi_status status;
-+ int i, len, err;
-+ char str[32], tmp[3];
-+ unsigned char *ptr, *buf;
++#define RX_CFG_REG_KER_OFST 0x800 // Receive configuration register
++#define RX_CFG_REG_OFST 0x800 // Receive configuration register
+
-+ status = acpi_resource_to_address64(res, &addr);
++#if !defined(FALCON_64K_RXFIFO) && !defined(FALCON_PRE_02020029)
++# if !defined(FALCON_128K_RXFIFO)
++# define FALCON_128K_RXFIFO
++# endif
++#endif
+
-+ /* Do we care about this range? Let's check. */
-+ if (!ACPI_SUCCESS(status) ||
-+ !(addr.resource_type == ACPI_MEMORY_RANGE ||
-+ addr.resource_type == ACPI_IO_RANGE) ||
-+ !addr.address_length || addr.producer_consumer != ACPI_PRODUCER)
-+ return AE_OK;
++#if defined(FALCON_128K_RXFIFO)
++
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 48
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 47
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 46
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 45
++ #define RX_HASH_ALG_B0_WIDTH 1
++ #define RX_HASH_INSERT_HDR_B0_LBN 44
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 43
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_RDW_PATCH_EN_LBN 42 /* Non head of line blocking */
++ #define RX_RDW_PATCH_EN_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 39
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 3
++ #define RX_OWNERR_CTL_B0_LBN 38
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 33
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 28
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 19
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 10
++ #define RX_XON_MAC_TH_B0_WIDTH 9
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 9
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
++
++#elif !defined(FALCON_PRE_02020029)
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 46
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 45
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 44
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 43
++ #define RX_HASH_ALG_B0_WIDTH 41
++ #define RX_HASH_INSERT_HDR_B0_LBN 42
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 41
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 37
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 3
++ #define RX_OWNERR_CTL_B0_LBN 36
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 31
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 26
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 17
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 9
++ #define RX_XON_MAC_TH_B0_WIDTH 8
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 8
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
+
-+ /*
-+ * Furthermore, we really only care to tell the guest about
-+ * address ranges that require address translation of some sort.
-+ */
-+ if (!(addr.resource_type == ACPI_MEMORY_RANGE &&
-+ addr.info.mem.translation) &&
-+ !(addr.resource_type == ACPI_IO_RANGE &&
-+ addr.info.io.translation))
-+ return AE_OK;
-+
-+ /* Store the resource in xenbus for the guest */
-+ len = snprintf(str, sizeof(str), "root-%d-resource-%d",
-+ info->root_num, info->resource_count);
-+ if (unlikely(len >= (sizeof(str) - 1)))
-+ return AE_OK;
++#else
++/* new for B0 */
++ #define RX_TOEP_TCP_SUPPRESS_B0_LBN 44
++ #define RX_TOEP_TCP_SUPPRESS_B0_WIDTH 1
++ #define RX_INGR_EN_B0_LBN 43
++ #define RX_INGR_EN_B0_WIDTH 1
++ #define RX_TOEP_IPV4_B0_LBN 42
++ #define RX_TOEP_IPV4_B0_WIDTH 1
++ #define RX_HASH_ALG_B0_LBN 41
++ #define RX_HASH_ALG_B0_WIDTH 41
++ #define RX_HASH_INSERT_HDR_B0_LBN 40
++ #define RX_HASH_INSERT_HDR_B0_WIDTH 1
++/* moved for B0 */
++ #define RX_DESC_PUSH_EN_B0_LBN 35
++ #define RX_DESC_PUSH_EN_B0_WIDTH 1
++ #define RX_PCI_BURST_SIZE_B0_LBN 35
++ #define RX_PCI_BURST_SIZE_B0_WIDTH 2
++ #define RX_OWNERR_CTL_B0_LBN 34
++ #define RX_OWNERR_CTL_B0_WIDTH 1
++ #define RX_XON_TX_TH_B0_LBN 29
++ #define RX_XON_TX_TH_B0_WIDTH 5
++ #define RX_XOFF_TX_TH_B0_LBN 24
++ #define RX_XOFF_TX_TH_B0_WIDTH 5
++ #define RX_USR_BUF_SIZE_B0_LBN 15
++ #define RX_USR_BUF_SIZE_B0_WIDTH 9
++ #define RX_XON_MAC_TH_B0_LBN 8
++ #define RX_XON_MAC_TH_B0_WIDTH 7
++ #define RX_XOFF_MAC_TH_B0_LBN 1
++ #define RX_XOFF_MAC_TH_B0_WIDTH 7
++ #define RX_XOFF_MAC_EN_B0_LBN 0
++ #define RX_XOFF_MAC_EN_B0_WIDTH 1
++
++#endif
++
++/* A0/A1 */
++ #define RX_PUSH_EN_A1_LBN 35
++ #define RX_PUSH_EN_A1_WIDTH 1
++ #define RX_PCI_BURST_SIZE_A1_LBN 31
++ #define RX_PCI_BURST_SIZE_A1_WIDTH 3
++ #define RX_OWNERR_CTL_A1_LBN 30
++ #define RX_OWNERR_CTL_A1_WIDTH 1
++ #define RX_XON_TX_TH_A1_LBN 25
++ #define RX_XON_TX_TH_A1_WIDTH 5
++ #define RX_XOFF_TX_TH_A1_LBN 20
++ #define RX_XOFF_TX_TH_A1_WIDTH 5
++ #define RX_USR_BUF_SIZE_A1_LBN 11
++ #define RX_USR_BUF_SIZE_A1_WIDTH 9
++ #define RX_XON_MAC_TH_A1_LBN 6
++ #define RX_XON_MAC_TH_A1_WIDTH 5
++ #define RX_XOFF_MAC_TH_A1_LBN 1
++ #define RX_XOFF_MAC_TH_A1_WIDTH 5
++ #define RX_XOFF_MAC_EN_A1_LBN 0
++ #define RX_XOFF_MAC_EN_A1_WIDTH 1
++
++#define RX_FILTER_CTL_REG_OFST 0x810 // Receive filter control registers
++ #define SCATTER_ENBL_NO_MATCH_Q_B0_LBN 40
++ #define SCATTER_ENBL_NO_MATCH_Q_B0_WIDTH 1
++ #define UDP_FULL_SRCH_LIMIT_LBN 32
++ #define UDP_FULL_SRCH_LIMIT_WIDTH 8
++ #define NUM_KER_LBN 24
++ #define NUM_KER_WIDTH 2
++ #define UDP_WILD_SRCH_LIMIT_LBN 16
++ #define UDP_WILD_SRCH_LIMIT_WIDTH 8
++ #define TCP_WILD_SRCH_LIMIT_LBN 8
++ #define TCP_WILD_SRCH_LIMIT_WIDTH 8
++ #define TCP_FULL_SRCH_LIMIT_LBN 0
++ #define TCP_FULL_SRCH_LIMIT_WIDTH 8
++#define RX_FLUSH_DESCQ_REG_KER_OFST 0x820 // Receive flush descriptor queue register
++#define RX_FLUSH_DESCQ_REG_OFST 0x820 // Receive flush descriptor queue register
++ #define RX_FLUSH_DESCQ_CMD_LBN 24
++ #define RX_FLUSH_DESCQ_CMD_WIDTH 1
++ #define RX_FLUSH_EVQ_ID_LBN 12
++ #define RX_FLUSH_EVQ_ID_WIDTH 12
++ #define RX_FLUSH_DESCQ_LBN 0
++ #define RX_FLUSH_DESCQ_WIDTH 12
++#define RX_DESC_UPD_REG_KER_OFST 0x830 // Kernel receive descriptor update register. Page-mapped
++#define RX_DESC_UPD_REG_PAGE4_OFST 0x8830 // Char & user receive descriptor update register. Page-mapped. For lowest 1K queues.
++#define RX_DESC_UPD_REG_PAGE123K_OFST 0x1000830 // Char & user receive descriptor update register. Page-mapped. For upper 3K queues.
++ #define RX_DESC_WPTR_LBN 96
++ #define RX_DESC_WPTR_WIDTH 12
++ #define RX_DESC_PUSH_CMD_LBN 95
++ #define RX_DESC_PUSH_CMD_WIDTH 1
++ #define RX_DESC_LBN 0
++ #define RX_DESC_WIDTH 64
++ #define RX_KER_DESC_LBN 0
++ #define RX_KER_DESC_WIDTH 64
++ #define RX_USR_DESC_LBN 0
++ #define RX_USR_DESC_WIDTH 32
++#define RX_DC_CFG_REG_KER_OFST 0x840 // Receive descriptor cache configuration register
++#define RX_DC_CFG_REG_OFST 0x840 // Receive descriptor cache configuration register
++ #define RX_DC_SIZE_LBN 0
++ #define RX_DC_SIZE_WIDTH 2
++#define RX_DC_PF_WM_REG_KER_OFST 0x850 // Receive descriptor cache pre-fetch watermark register
++#define RX_DC_PF_WM_REG_OFST 0x850 // Receive descriptor cache pre-fetch watermark register
++ #define RX_DC_PF_LWM_LO_LBN 0
++ #define RX_DC_PF_LWM_LO_WIDTH 6
++
++#define RX_RSS_TKEY_B0_OFST 0x860 // RSS Toeplitz hash key (B0 only)
++
++#define RX_NODESC_DROP_REG 0x880
++ #define RX_NODESC_DROP_CNT_LBN 0
++ #define RX_NODESC_DROP_CNT_WIDTH 16
++
++#define XM_TX_CFG_REG_OFST 0x1230
++ #define XM_AUTO_PAD_LBN 5
++ #define XM_AUTO_PAD_WIDTH 1
++
++#define RX_FILTER_TBL0_OFST 0xF00000 // Receive filter table - even entries
++ #define RSS_EN_0_B0_LBN 110
++ #define RSS_EN_0_B0_WIDTH 1
++ #define SCATTER_EN_0_B0_LBN 109
++ #define SCATTER_EN_0_B0_WIDTH 1
++ #define TCP_UDP_0_LBN 108
++ #define TCP_UDP_0_WIDTH 1
++ #define RXQ_ID_0_LBN 96
++ #define RXQ_ID_0_WIDTH 12
++ #define DEST_IP_0_LBN 64
++ #define DEST_IP_0_WIDTH 32
++ #define DEST_PORT_TCP_0_LBN 48
++ #define DEST_PORT_TCP_0_WIDTH 16
++ #define SRC_IP_0_LBN 16
++ #define SRC_IP_0_WIDTH 32
++ #define SRC_TCP_DEST_UDP_0_LBN 0
++ #define SRC_TCP_DEST_UDP_0_WIDTH 16
++#define RX_FILTER_TBL1_OFST 0xF00010 // Receive filter table - odd entries
++ #define RSS_EN_1_B0_LBN 110
++ #define RSS_EN_1_B0_WIDTH 1
++ #define SCATTER_EN_1_B0_LBN 109
++ #define SCATTER_EN_1_B0_WIDTH 1
++ #define TCP_UDP_1_LBN 108
++ #define TCP_UDP_1_WIDTH 1
++ #define RXQ_ID_1_LBN 96
++ #define RXQ_ID_1_WIDTH 12
++ #define DEST_IP_1_LBN 64
++ #define DEST_IP_1_WIDTH 32
++ #define DEST_PORT_TCP_1_LBN 48
++ #define DEST_PORT_TCP_1_WIDTH 16
++ #define SRC_IP_1_LBN 16
++ #define SRC_IP_1_WIDTH 32
++ #define SRC_TCP_DEST_UDP_1_LBN 0
++ #define SRC_TCP_DEST_UDP_1_WIDTH 16
++
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define RX_DESC_PTR_TBL_KER_OFST 0x11800 // Receive descriptor pointer kernel access
++#else
++#define RX_DESC_PTR_TBL_KER_OFST 0x1800 // Receive descriptor pointer kernel access
++#endif
++
++
++#define RX_DESC_PTR_TBL_OFST 0xF40000 // Receive descriptor pointer table
++ #define RX_ISCSI_DDIG_EN_LBN 88
++ #define RX_ISCSI_DDIG_EN_WIDTH 1
++ #define RX_ISCSI_HDIG_EN_LBN 87
++ #define RX_ISCSI_HDIG_EN_WIDTH 1
++ #define RX_DESC_PREF_ACT_LBN 86
++ #define RX_DESC_PREF_ACT_WIDTH 1
++ #define RX_DC_HW_RPTR_LBN 80
++ #define RX_DC_HW_RPTR_WIDTH 6
++ #define RX_DESCQ_HW_RPTR_LBN 68
++ #define RX_DESCQ_HW_RPTR_WIDTH 12
++ #define RX_DESCQ_SW_WPTR_LBN 56
++ #define RX_DESCQ_SW_WPTR_WIDTH 12
++ #define RX_DESCQ_BUF_BASE_ID_LBN 36
++ #define RX_DESCQ_BUF_BASE_ID_WIDTH 20
++ #define RX_DESCQ_EVQ_ID_LBN 24
++ #define RX_DESCQ_EVQ_ID_WIDTH 12
++ #define RX_DESCQ_OWNER_ID_LBN 10
++ #define RX_DESCQ_OWNER_ID_WIDTH 14
++ #define RX_DESCQ_LABEL_LBN 5
++ #define RX_DESCQ_LABEL_WIDTH 5
++ #define RX_DESCQ_SIZE_LBN 3
++ #define RX_DESCQ_SIZE_WIDTH 2
++ #define RX_DESCQ_TYPE_LBN 2
++ #define RX_DESCQ_TYPE_WIDTH 1
++ #define RX_DESCQ_JUMBO_LBN 1
++ #define RX_DESCQ_JUMBO_WIDTH 1
++ #define RX_DESCQ_EN_LBN 0
++ #define RX_DESCQ_EN_WIDTH 1
++
++
++#define RX_RSS_INDIR_TBL_B0_OFST 0xFB0000 // RSS indirection table (B0 only)
++ #define RX_RSS_INDIR_ENT_B0_LBN 0
++ #define RX_RSS_INDIR_ENT_B0_WIDTH 6
++
++//////////////---- TX Datapath Registers C Header ----//////////////
++#define TX_FLUSH_DESCQ_REG_KER_OFST 0xA00 // Transmit flush descriptor queue register
++#define TX_FLUSH_DESCQ_REG_OFST 0xA00 // Transmit flush descriptor queue register
++ #define TX_FLUSH_DESCQ_CMD_LBN 12
++ #define TX_FLUSH_DESCQ_CMD_WIDTH 1
++ #define TX_FLUSH_DESCQ_LBN 0
++ #define TX_FLUSH_DESCQ_WIDTH 12
++#define TX_DESC_UPD_REG_KER_OFST 0xA10 // Kernel transmit descriptor update register. Page-mapped
++#define TX_DESC_UPD_REG_PAGE4_OFST 0x8A10 // Char & user transmit descriptor update register. Page-mapped
++#define TX_DESC_UPD_REG_PAGE123K_OFST 0x1000A10 // Char & user transmit descriptor update register. Page-mapped
++ #define TX_DESC_WPTR_LBN 96
++ #define TX_DESC_WPTR_WIDTH 12
++ #define TX_DESC_PUSH_CMD_LBN 95
++ #define TX_DESC_PUSH_CMD_WIDTH 1
++ #define TX_DESC_LBN 0
++ #define TX_DESC_WIDTH 95
++ #define TX_KER_DESC_LBN 0
++ #define TX_KER_DESC_WIDTH 64
++ #define TX_USR_DESC_LBN 0
++ #define TX_USR_DESC_WIDTH 64
++#define TX_DC_CFG_REG_KER_OFST 0xA20 // Transmit descriptor cache configuration register
++#define TX_DC_CFG_REG_OFST 0xA20 // Transmit descriptor cache configuration register
++ #define TX_DC_SIZE_LBN 0
++ #define TX_DC_SIZE_WIDTH 2
++
++#if EFVI_FALCON_EXTENDED_P_BAR
++#define TX_DESC_PTR_TBL_KER_OFST 0x11900 // Transmit descriptor pointer.
++#else
++#define TX_DESC_PTR_TBL_KER_OFST 0x1900 // Transmit descriptor pointer.
++#endif
++
++
++#define TX_DESC_PTR_TBL_OFST 0xF50000 // Transmit descriptor pointer
++ #define TX_NON_IP_DROP_DIS_B0_LBN 91
++ #define TX_NON_IP_DROP_DIS_B0_WIDTH 1
++ #define TX_IP_CHKSM_DIS_B0_LBN 90
++ #define TX_IP_CHKSM_DIS_B0_WIDTH 1
++ #define TX_TCP_CHKSM_DIS_B0_LBN 89
++ #define TX_TCP_CHKSM_DIS_B0_WIDTH 1
++ #define TX_DESCQ_EN_LBN 88
++ #define TX_DESCQ_EN_WIDTH 1
++ #define TX_ISCSI_DDIG_EN_LBN 87
++ #define TX_ISCSI_DDIG_EN_WIDTH 1
++ #define TX_ISCSI_HDIG_EN_LBN 86
++ #define TX_ISCSI_HDIG_EN_WIDTH 1
++ #define TX_DC_HW_RPTR_LBN 80
++ #define TX_DC_HW_RPTR_WIDTH 6
++ #define TX_DESCQ_HW_RPTR_LBN 68
++ #define TX_DESCQ_HW_RPTR_WIDTH 12
++ #define TX_DESCQ_SW_WPTR_LBN 56
++ #define TX_DESCQ_SW_WPTR_WIDTH 12
++ #define TX_DESCQ_BUF_BASE_ID_LBN 36
++ #define TX_DESCQ_BUF_BASE_ID_WIDTH 20
++ #define TX_DESCQ_EVQ_ID_LBN 24
++ #define TX_DESCQ_EVQ_ID_WIDTH 12
++ #define TX_DESCQ_OWNER_ID_LBN 10
++ #define TX_DESCQ_OWNER_ID_WIDTH 14
++ #define TX_DESCQ_LABEL_LBN 5
++ #define TX_DESCQ_LABEL_WIDTH 5
++ #define TX_DESCQ_SIZE_LBN 3
++ #define TX_DESCQ_SIZE_WIDTH 2
++ #define TX_DESCQ_TYPE_LBN 1
++ #define TX_DESCQ_TYPE_WIDTH 2
++ #define TX_DESCQ_FLUSH_LBN 0
++ #define TX_DESCQ_FLUSH_WIDTH 1
++#define TX_CFG_REG_KER_OFST 0xA50 // Transmit configuration register
++#define TX_CFG_REG_OFST 0xA50 // Transmit configuration register
++ #define TX_IP_ID_P1_OFS_LBN 32
++ #define TX_IP_ID_P1_OFS_WIDTH 15
++ #define TX_IP_ID_P0_OFS_LBN 16
++ #define TX_IP_ID_P0_OFS_WIDTH 15
++ #define TX_TURBO_EN_LBN 3
++ #define TX_TURBO_EN_WIDTH 1
++ #define TX_OWNERR_CTL_LBN 2
++ #define TX_OWNERR_CTL_WIDTH 2
++ #define TX_NON_IP_DROP_DIS_LBN 1
++ #define TX_NON_IP_DROP_DIS_WIDTH 1
++ #define TX_IP_ID_REP_EN_LBN 0
++ #define TX_IP_ID_REP_EN_WIDTH 1
++#define TX_RESERVED_REG_KER_OFST 0xA80 // Transmit configuration register
++#define TX_RESERVED_REG_OFST 0xA80 // Transmit configuration register
++ #define TX_CSR_PUSH_EN_LBN 89
++ #define TX_CSR_PUSH_EN_WIDTH 1
++ #define TX_RX_SPACER_LBN 64
++ #define TX_RX_SPACER_WIDTH 8
++ #define TX_SW_EV_EN_LBN 59
++ #define TX_SW_EV_EN_WIDTH 1
++ #define TX_RX_SPACER_EN_LBN 57
++ #define TX_RX_SPACER_EN_WIDTH 1
++ #define TX_CSR_PREF_WD_TMR_LBN 24
++ #define TX_CSR_PREF_WD_TMR_WIDTH 16
++ #define TX_CSR_ONLY1TAG_LBN 21
++ #define TX_CSR_ONLY1TAG_WIDTH 1
++ #define TX_PREF_THRESHOLD_LBN 19
++ #define TX_PREF_THRESHOLD_WIDTH 2
++ #define TX_ONE_PKT_PER_Q_LBN 18
++ #define TX_ONE_PKT_PER_Q_WIDTH 1
++ #define TX_DIS_NON_IP_EV_LBN 17
++ #define TX_DIS_NON_IP_EV_WIDTH 1
++ #define TX_DMA_SPACER_LBN 8
++ #define TX_DMA_SPACER_WIDTH 8
++ #define TX_FLUSH_MIN_LEN_EN_B0_LBN 7
++ #define TX_FLUSH_MIN_LEN_EN_B0_WIDTH 1
++ #define TX_TCP_DIS_A1_LBN 7
++ #define TX_TCP_DIS_A1_WIDTH 1
++ #define TX_IP_DIS_A1_LBN 6
++ #define TX_IP_DIS_A1_WIDTH 1
++ #define TX_MAX_CPL_LBN 2
++ #define TX_MAX_CPL_WIDTH 2
++ #define TX_MAX_PREF_LBN 0
++ #define TX_MAX_PREF_WIDTH 2
++#define TX_VLAN_REG_OFST 0xAE0 // Transmit VLAN tag register
++ #define TX_VLAN_EN_LBN 127
++ #define TX_VLAN_EN_WIDTH 1
++ #define TX_VLAN7_PORT1_EN_LBN 125
++ #define TX_VLAN7_PORT1_EN_WIDTH 1
++ #define TX_VLAN7_PORT0_EN_LBN 124
++ #define TX_VLAN7_PORT0_EN_WIDTH 1
++ #define TX_VLAN7_LBN 112
++ #define TX_VLAN7_WIDTH 12
++ #define TX_VLAN6_PORT1_EN_LBN 109
++ #define TX_VLAN6_PORT1_EN_WIDTH 1
++ #define TX_VLAN6_PORT0_EN_LBN 108
++ #define TX_VLAN6_PORT0_EN_WIDTH 1
++ #define TX_VLAN6_LBN 96
++ #define TX_VLAN6_WIDTH 12
++ #define TX_VLAN5_PORT1_EN_LBN 93
++ #define TX_VLAN5_PORT1_EN_WIDTH 1
++ #define TX_VLAN5_PORT0_EN_LBN 92
++ #define TX_VLAN5_PORT0_EN_WIDTH 1
++ #define TX_VLAN5_LBN 80
++ #define TX_VLAN5_WIDTH 12
++ #define TX_VLAN4_PORT1_EN_LBN 77
++ #define TX_VLAN4_PORT1_EN_WIDTH 1
++ #define TX_VLAN4_PORT0_EN_LBN 76
++ #define TX_VLAN4_PORT0_EN_WIDTH 1
++ #define TX_VLAN4_LBN 64
++ #define TX_VLAN4_WIDTH 12
++ #define TX_VLAN3_PORT1_EN_LBN 61
++ #define TX_VLAN3_PORT1_EN_WIDTH 1
++ #define TX_VLAN3_PORT0_EN_LBN 60
++ #define TX_VLAN3_PORT0_EN_WIDTH 1
++ #define TX_VLAN3_LBN 48
++ #define TX_VLAN3_WIDTH 12
++ #define TX_VLAN2_PORT1_EN_LBN 45
++ #define TX_VLAN2_PORT1_EN_WIDTH 1
++ #define TX_VLAN2_PORT0_EN_LBN 44
++ #define TX_VLAN2_PORT0_EN_WIDTH 1
++ #define TX_VLAN2_LBN 32
++ #define TX_VLAN2_WIDTH 12
++ #define TX_VLAN1_PORT1_EN_LBN 29
++ #define TX_VLAN1_PORT1_EN_WIDTH 1
++ #define TX_VLAN1_PORT0_EN_LBN 28
++ #define TX_VLAN1_PORT0_EN_WIDTH 1
++ #define TX_VLAN1_LBN 16
++ #define TX_VLAN1_WIDTH 12
++ #define TX_VLAN0_PORT1_EN_LBN 13
++ #define TX_VLAN0_PORT1_EN_WIDTH 1
++ #define TX_VLAN0_PORT0_EN_LBN 12
++ #define TX_VLAN0_PORT0_EN_WIDTH 1
++ #define TX_VLAN0_LBN 0
++ #define TX_VLAN0_WIDTH 12
++#define TX_FIL_CTL_REG_OFST 0xAF0 // Transmit filter control register
++ #define TX_MADR1_FIL_EN_LBN 65
++ #define TX_MADR1_FIL_EN_WIDTH 1
++ #define TX_MADR0_FIL_EN_LBN 64
++ #define TX_MADR0_FIL_EN_WIDTH 1
++ #define TX_IPFIL31_PORT1_EN_LBN 63
++ #define TX_IPFIL31_PORT1_EN_WIDTH 1
++ #define TX_IPFIL31_PORT0_EN_LBN 62
++ #define TX_IPFIL31_PORT0_EN_WIDTH 1
++ #define TX_IPFIL30_PORT1_EN_LBN 61
++ #define TX_IPFIL30_PORT1_EN_WIDTH 1
++ #define TX_IPFIL30_PORT0_EN_LBN 60
++ #define TX_IPFIL30_PORT0_EN_WIDTH 1
++ #define TX_IPFIL29_PORT1_EN_LBN 59
++ #define TX_IPFIL29_PORT1_EN_WIDTH 1
++ #define TX_IPFIL29_PORT0_EN_LBN 58
++ #define TX_IPFIL29_PORT0_EN_WIDTH 1
++ #define TX_IPFIL28_PORT1_EN_LBN 57
++ #define TX_IPFIL28_PORT1_EN_WIDTH 1
++ #define TX_IPFIL28_PORT0_EN_LBN 56
++ #define TX_IPFIL28_PORT0_EN_WIDTH 1
++ #define TX_IPFIL27_PORT1_EN_LBN 55
++ #define TX_IPFIL27_PORT1_EN_WIDTH 1
++ #define TX_IPFIL27_PORT0_EN_LBN 54
++ #define TX_IPFIL27_PORT0_EN_WIDTH 1
++ #define TX_IPFIL26_PORT1_EN_LBN 53
++ #define TX_IPFIL26_PORT1_EN_WIDTH 1
++ #define TX_IPFIL26_PORT0_EN_LBN 52
++ #define TX_IPFIL26_PORT0_EN_WIDTH 1
++ #define TX_IPFIL25_PORT1_EN_LBN 51
++ #define TX_IPFIL25_PORT1_EN_WIDTH 1
++ #define TX_IPFIL25_PORT0_EN_LBN 50
++ #define TX_IPFIL25_PORT0_EN_WIDTH 1
++ #define TX_IPFIL24_PORT1_EN_LBN 49
++ #define TX_IPFIL24_PORT1_EN_WIDTH 1
++ #define TX_IPFIL24_PORT0_EN_LBN 48
++ #define TX_IPFIL24_PORT0_EN_WIDTH 1
++ #define TX_IPFIL23_PORT1_EN_LBN 47
++ #define TX_IPFIL23_PORT1_EN_WIDTH 1
++ #define TX_IPFIL23_PORT0_EN_LBN 46
++ #define TX_IPFIL23_PORT0_EN_WIDTH 1
++ #define TX_IPFIL22_PORT1_EN_LBN 45
++ #define TX_IPFIL22_PORT1_EN_WIDTH 1
++ #define TX_IPFIL22_PORT0_EN_LBN 44
++ #define TX_IPFIL22_PORT0_EN_WIDTH 1
++ #define TX_IPFIL21_PORT1_EN_LBN 43
++ #define TX_IPFIL21_PORT1_EN_WIDTH 1
++ #define TX_IPFIL21_PORT0_EN_LBN 42
++ #define TX_IPFIL21_PORT0_EN_WIDTH 1
++ #define TX_IPFIL20_PORT1_EN_LBN 41
++ #define TX_IPFIL20_PORT1_EN_WIDTH 1
++ #define TX_IPFIL20_PORT0_EN_LBN 40
++ #define TX_IPFIL20_PORT0_EN_WIDTH 1
++ #define TX_IPFIL19_PORT1_EN_LBN 39
++ #define TX_IPFIL19_PORT1_EN_WIDTH 1
++ #define TX_IPFIL19_PORT0_EN_LBN 38
++ #define TX_IPFIL19_PORT0_EN_WIDTH 1
++ #define TX_IPFIL18_PORT1_EN_LBN 37
++ #define TX_IPFIL18_PORT1_EN_WIDTH 1
++ #define TX_IPFIL18_PORT0_EN_LBN 36
++ #define TX_IPFIL18_PORT0_EN_WIDTH 1
++ #define TX_IPFIL17_PORT1_EN_LBN 35
++ #define TX_IPFIL17_PORT1_EN_WIDTH 1
++ #define TX_IPFIL17_PORT0_EN_LBN 34
++ #define TX_IPFIL17_PORT0_EN_WIDTH 1
++ #define TX_IPFIL16_PORT1_EN_LBN 33
++ #define TX_IPFIL16_PORT1_EN_WIDTH 1
++ #define TX_IPFIL16_PORT0_EN_LBN 32
++ #define TX_IPFIL16_PORT0_EN_WIDTH 1
++ #define TX_IPFIL15_PORT1_EN_LBN 31
++ #define TX_IPFIL15_PORT1_EN_WIDTH 1
++ #define TX_IPFIL15_PORT0_EN_LBN 30
++ #define TX_IPFIL15_PORT0_EN_WIDTH 1
++ #define TX_IPFIL14_PORT1_EN_LBN 29
++ #define TX_IPFIL14_PORT1_EN_WIDTH 1
++ #define TX_IPFIL14_PORT0_EN_LBN 28
++ #define TX_IPFIL14_PORT0_EN_WIDTH 1
++ #define TX_IPFIL13_PORT1_EN_LBN 27
++ #define TX_IPFIL13_PORT1_EN_WIDTH 1
++ #define TX_IPFIL13_PORT0_EN_LBN 26
++ #define TX_IPFIL13_PORT0_EN_WIDTH 1
++ #define TX_IPFIL12_PORT1_EN_LBN 25
++ #define TX_IPFIL12_PORT1_EN_WIDTH 1
++ #define TX_IPFIL12_PORT0_EN_LBN 24
++ #define TX_IPFIL12_PORT0_EN_WIDTH 1
++ #define TX_IPFIL11_PORT1_EN_LBN 23
++ #define TX_IPFIL11_PORT1_EN_WIDTH 1
++ #define TX_IPFIL11_PORT0_EN_LBN 22
++ #define TX_IPFIL11_PORT0_EN_WIDTH 1
++ #define TX_IPFIL10_PORT1_EN_LBN 21
++ #define TX_IPFIL10_PORT1_EN_WIDTH 1
++ #define TX_IPFIL10_PORT0_EN_LBN 20
++ #define TX_IPFIL10_PORT0_EN_WIDTH 1
++ #define TX_IPFIL9_PORT1_EN_LBN 19
++ #define TX_IPFIL9_PORT1_EN_WIDTH 1
++ #define TX_IPFIL9_PORT0_EN_LBN 18
++ #define TX_IPFIL9_PORT0_EN_WIDTH 1
++ #define TX_IPFIL8_PORT1_EN_LBN 17
++ #define TX_IPFIL8_PORT1_EN_WIDTH 1
++ #define TX_IPFIL8_PORT0_EN_LBN 16
++ #define TX_IPFIL8_PORT0_EN_WIDTH 1
++ #define TX_IPFIL7_PORT1_EN_LBN 15
++ #define TX_IPFIL7_PORT1_EN_WIDTH 1
++ #define TX_IPFIL7_PORT0_EN_LBN 14
++ #define TX_IPFIL7_PORT0_EN_WIDTH 1
++ #define TX_IPFIL6_PORT1_EN_LBN 13
++ #define TX_IPFIL6_PORT1_EN_WIDTH 1
++ #define TX_IPFIL6_PORT0_EN_LBN 12
++ #define TX_IPFIL6_PORT0_EN_WIDTH 1
++ #define TX_IPFIL5_PORT1_EN_LBN 11
++ #define TX_IPFIL5_PORT1_EN_WIDTH 1
++ #define TX_IPFIL5_PORT0_EN_LBN 10
++ #define TX_IPFIL5_PORT0_EN_WIDTH 1
++ #define TX_IPFIL4_PORT1_EN_LBN 9
++ #define TX_IPFIL4_PORT1_EN_WIDTH 1
++ #define TX_IPFIL4_PORT0_EN_LBN 8
++ #define TX_IPFIL4_PORT0_EN_WIDTH 1
++ #define TX_IPFIL3_PORT1_EN_LBN 7
++ #define TX_IPFIL3_PORT1_EN_WIDTH 1
++ #define TX_IPFIL3_PORT0_EN_LBN 6
++ #define TX_IPFIL3_PORT0_EN_WIDTH 1
++ #define TX_IPFIL2_PORT1_EN_LBN 5
++ #define TX_IPFIL2_PORT1_EN_WIDTH 1
++ #define TX_IPFIL2_PORT0_EN_LBN 4
++ #define TX_IPFIL2_PORT0_EN_WIDTH 1
++ #define TX_IPFIL1_PORT1_EN_LBN 3
++ #define TX_IPFIL1_PORT1_EN_WIDTH 1
++ #define TX_IPFIL1_PORT0_EN_LBN 2
++ #define TX_IPFIL1_PORT0_EN_WIDTH 1
++ #define TX_IPFIL0_PORT1_EN_LBN 1
++ #define TX_IPFIL0_PORT1_EN_WIDTH 1
++ #define TX_IPFIL0_PORT0_EN_LBN 0
++ #define TX_IPFIL0_PORT0_EN_WIDTH 1
++#define TX_IPFIL_TBL_OFST 0xB00 // Transmit IP source address filter table
++ #define TX_IPFIL_MASK_LBN 32
++ #define TX_IPFIL_MASK_WIDTH 32
++ #define TX_IP_SRC_ADR_LBN 0
++ #define TX_IP_SRC_ADR_WIDTH 32
++#define TX_PACE_REG_A1_OFST 0xF80000 // Transmit pace control register
++#define TX_PACE_REG_B0_OFST 0xA90 // Transmit pace control register
++ #define TX_PACE_SB_AF_LBN 19
++ #define TX_PACE_SB_AF_WIDTH 10
++ #define TX_PACE_SB_NOTAF_LBN 9
++ #define TX_PACE_SB_NOTAF_WIDTH 10
++ #define TX_PACE_FB_BASE_LBN 5
++ #define TX_PACE_FB_BASE_WIDTH 4
++ #define TX_PACE_BIN_TH_LBN 0
++ #define TX_PACE_BIN_TH_WIDTH 5
++#define TX_PACE_TBL_A1_OFST 0xF80040 // Transmit pacing table
++#define TX_PACE_TBL_FIRST_QUEUE_A1 4
++#define TX_PACE_TBL_B0_OFST 0xF80000 // Transmit pacing table
++#define TX_PACE_TBL_FIRST_QUEUE_B0 0
++ #define TX_PACE_LBN 0
++ #define TX_PACE_WIDTH 5
++
++//////////////---- EE/Flash Registers C Header ----//////////////
++#define EE_SPI_HCMD_REG_KER_OFST 0x100 // SPI host command register
++#define EE_SPI_HCMD_REG_OFST 0x100 // SPI host command register
++ #define EE_SPI_HCMD_CMD_EN_LBN 31
++ #define EE_SPI_HCMD_CMD_EN_WIDTH 1
++ #define EE_WR_TIMER_ACTIVE_LBN 28
++ #define EE_WR_TIMER_ACTIVE_WIDTH 1
++ #define EE_SPI_HCMD_SF_SEL_LBN 24
++ #define EE_SPI_HCMD_SF_SEL_WIDTH 1
++ #define EE_SPI_HCMD_DABCNT_LBN 16
++ #define EE_SPI_HCMD_DABCNT_WIDTH 5
++ #define EE_SPI_HCMD_READ_LBN 15
++ #define EE_SPI_HCMD_READ_WIDTH 1
++ #define EE_SPI_HCMD_DUBCNT_LBN 12
++ #define EE_SPI_HCMD_DUBCNT_WIDTH 2
++ #define EE_SPI_HCMD_ADBCNT_LBN 8
++ #define EE_SPI_HCMD_ADBCNT_WIDTH 2
++ #define EE_SPI_HCMD_ENC_LBN 0
++ #define EE_SPI_HCMD_ENC_WIDTH 8
++#define EE_SPI_HADR_REG_KER_OFST 0X110 // SPI host address register
++#define EE_SPI_HADR_REG_OFST 0X110 // SPI host address register
++ #define EE_SPI_HADR_DUBYTE_LBN 24
++ #define EE_SPI_HADR_DUBYTE_WIDTH 8
++ #define EE_SPI_HADR_ADR_LBN 0
++ #define EE_SPI_HADR_ADR_WIDTH 24
++#define EE_SPI_HDATA_REG_KER_OFST 0x120 // SPI host data register
++#define EE_SPI_HDATA_REG_OFST 0x120 // SPI host data register
++ #define EE_SPI_HDATA3_LBN 96
++ #define EE_SPI_HDATA3_WIDTH 32
++ #define EE_SPI_HDATA2_LBN 64
++ #define EE_SPI_HDATA2_WIDTH 32
++ #define EE_SPI_HDATA1_LBN 32
++ #define EE_SPI_HDATA1_WIDTH 32
++ #define EE_SPI_HDATA0_LBN 0
++ #define EE_SPI_HDATA0_WIDTH 32
++#define EE_BASE_PAGE_REG_KER_OFST 0x130 // Expansion ROM base mirror register
++#define EE_BASE_PAGE_REG_OFST 0x130 // Expansion ROM base mirror register
++ #define EE_EXP_ROM_WINDOW_BASE_LBN 16
++ #define EE_EXP_ROM_WINDOW_BASE_WIDTH 13
++ #define EE_EXPROM_MASK_LBN 0
++ #define EE_EXPROM_MASK_WIDTH 13
++#define EE_VPD_CFG0_REG_KER_OFST 0X140 // SPI/VPD configuration register
++#define EE_VPD_CFG0_REG_OFST 0X140 // SPI/VPD configuration register
++ #define EE_SF_FASTRD_EN_LBN 127
++ #define EE_SF_FASTRD_EN_WIDTH 1
++ #define EE_SF_CLOCK_DIV_LBN 120
++ #define EE_SF_CLOCK_DIV_WIDTH 7
++ #define EE_VPD_WIP_POLL_LBN 119
++ #define EE_VPD_WIP_POLL_WIDTH 1
++ #define EE_VPDW_LENGTH_LBN 80
++ #define EE_VPDW_LENGTH_WIDTH 15
++ #define EE_VPDW_BASE_LBN 64
++ #define EE_VPDW_BASE_WIDTH 15
++ #define EE_VPD_WR_CMD_EN_LBN 56
++ #define EE_VPD_WR_CMD_EN_WIDTH 8
++ #define EE_VPD_BASE_LBN 32
++ #define EE_VPD_BASE_WIDTH 24
++ #define EE_VPD_LENGTH_LBN 16
++ #define EE_VPD_LENGTH_WIDTH 13
++ #define EE_VPD_AD_SIZE_LBN 8
++ #define EE_VPD_AD_SIZE_WIDTH 5
++ #define EE_VPD_ACCESS_ON_LBN 5
++ #define EE_VPD_ACCESS_ON_WIDTH 1
++#define EE_VPD_SW_CNTL_REG_KER_OFST 0X150 // VPD access SW control register
++#define EE_VPD_SW_CNTL_REG_OFST 0X150 // VPD access SW control register
++ #define EE_VPD_CYCLE_PENDING_LBN 31
++ #define EE_VPD_CYCLE_PENDING_WIDTH 1
++ #define EE_VPD_CYC_WRITE_LBN 28
++ #define EE_VPD_CYC_WRITE_WIDTH 1
++ #define EE_VPD_CYC_ADR_LBN 0
++ #define EE_VPD_CYC_ADR_WIDTH 15
++#define EE_VPD_SW_DATA_REG_KER_OFST 0x160 // VPD access SW data register
++#define EE_VPD_SW_DATA_REG_OFST 0x160 // VPD access SW data register
++ #define EE_VPD_CYC_DAT_LBN 0
++ #define EE_VPD_CYC_DAT_WIDTH 32
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/ef_vi_falcon_desc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/ef_vi_falcon_desc.h 2008-05-19 00:33:48.674951800 +0300
+@@ -0,0 +1,43 @@
++//////////////---- Descriptors C Headers ----//////////////
++// Receive Kernel IP Descriptor
++ #define RX_KER_BUF_SIZE_LBN 48
++ #define RX_KER_BUF_SIZE_WIDTH 14
++ #define RX_KER_BUF_REGION_LBN 46
++ #define RX_KER_BUF_REGION_WIDTH 2
++ #define RX_KER_BUF_REGION0_DECODE 0
++ #define RX_KER_BUF_REGION1_DECODE 1
++ #define RX_KER_BUF_REGION2_DECODE 2
++ #define RX_KER_BUF_REGION3_DECODE 3
++ #define RX_KER_BUF_ADR_LBN 0
++ #define RX_KER_BUF_ADR_WIDTH 46
++// Receive User IP Descriptor
++ #define RX_USR_2BYTE_OFS_LBN 20
++ #define RX_USR_2BYTE_OFS_WIDTH 12
++ #define RX_USR_BUF_ID_LBN 0
++ #define RX_USR_BUF_ID_WIDTH 20
++// Transmit Kernel IP Descriptor
++ #define TX_KER_PORT_LBN 63
++ #define TX_KER_PORT_WIDTH 1
++ #define TX_KER_CONT_LBN 62
++ #define TX_KER_CONT_WIDTH 1
++ #define TX_KER_BYTE_CNT_LBN 48
++ #define TX_KER_BYTE_CNT_WIDTH 14
++ #define TX_KER_BUF_REGION_LBN 46
++ #define TX_KER_BUF_REGION_WIDTH 2
++ #define TX_KER_BUF_REGION0_DECODE 0
++ #define TX_KER_BUF_REGION1_DECODE 1
++ #define TX_KER_BUF_REGION2_DECODE 2
++ #define TX_KER_BUF_REGION3_DECODE 3
++ #define TX_KER_BUF_ADR_LBN 0
++ #define TX_KER_BUF_ADR_WIDTH 46
++// Transmit User IP Descriptor
++ #define TX_USR_PORT_LBN 47
++ #define TX_USR_PORT_WIDTH 1
++ #define TX_USR_CONT_LBN 46
++ #define TX_USR_CONT_WIDTH 1
++ #define TX_USR_BYTE_CNT_LBN 33
++ #define TX_USR_BYTE_CNT_WIDTH 13
++ #define TX_USR_BUF_ID_LBN 13
++ #define TX_USR_BUF_ID_WIDTH 20
++ #define TX_USR_BYTE_OFS_LBN 0
++ #define TX_USR_BYTE_OFS_WIDTH 13
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/ef_vi_falcon_event.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/ef_vi_falcon_event.h 2008-05-19 00:33:48.674951800 +0300
+@@ -0,0 +1,123 @@
++//////////////---- Events Format C Header ----//////////////
++//////////////---- Event entry ----//////////////
++ #define EV_CODE_LBN 60
++ #define EV_CODE_WIDTH 4
++ #define RX_IP_EV_DECODE 0
++ #define TX_IP_EV_DECODE 2
++ #define DRIVER_EV_DECODE 5
++ #define GLOBAL_EV_DECODE 6
++ #define DRV_GEN_EV_DECODE 7
++ #define EV_DATA_LBN 0
++ #define EV_DATA_WIDTH 60
++//////////////---- Receive IP events for both Kernel & User event queues ----//////////////
++ #define RX_EV_PKT_OK_LBN 56
++ #define RX_EV_PKT_OK_WIDTH 1
++ #define RX_EV_BUF_OWNER_ID_ERR_LBN 54
++ #define RX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++ #define RX_EV_IP_HDR_CHKSUM_ERR_LBN 52
++ #define RX_EV_IP_HDR_CHKSUM_ERR_WIDTH 1
++ #define RX_EV_TCP_UDP_CHKSUM_ERR_LBN 51
++ #define RX_EV_TCP_UDP_CHKSUM_ERR_WIDTH 1
++ #define RX_EV_ETH_CRC_ERR_LBN 50
++ #define RX_EV_ETH_CRC_ERR_WIDTH 1
++ #define RX_EV_FRM_TRUNC_LBN 49
++ #define RX_EV_FRM_TRUNC_WIDTH 1
++ #define RX_EV_DRIB_NIB_LBN 48
++ #define RX_EV_DRIB_NIB_WIDTH 1
++ #define RX_EV_TOBE_DISC_LBN 47
++ #define RX_EV_TOBE_DISC_WIDTH 1
++ #define RX_EV_PKT_TYPE_LBN 44
++ #define RX_EV_PKT_TYPE_WIDTH 3
++ #define RX_EV_PKT_TYPE_ETH_DECODE 0
++ #define RX_EV_PKT_TYPE_LLC_DECODE 1
++ #define RX_EV_PKT_TYPE_JUMBO_DECODE 2
++ #define RX_EV_PKT_TYPE_VLAN_DECODE 3
++ #define RX_EV_PKT_TYPE_VLAN_LLC_DECODE 4
++ #define RX_EV_PKT_TYPE_VLAN_JUMBO_DECODE 5
++ #define RX_EV_HDR_TYPE_LBN 42
++ #define RX_EV_HDR_TYPE_WIDTH 2
++ #define RX_EV_HDR_TYPE_TCP_IPV4_DECODE 0
++ #define RX_EV_HDR_TYPE_UDP_IPV4_DECODE 1
++ #define RX_EV_HDR_TYPE_OTHER_IP_DECODE 2
++ #define RX_EV_HDR_TYPE_NON_IP_DECODE 3
++ #define RX_EV_DESC_Q_EMPTY_LBN 41
++ #define RX_EV_DESC_Q_EMPTY_WIDTH 1
++ #define RX_EV_MCAST_HASH_MATCH_LBN 40
++ #define RX_EV_MCAST_HASH_MATCH_WIDTH 1
++ #define RX_EV_MCAST_PKT_LBN 39
++ #define RX_EV_MCAST_PKT_WIDTH 1
++ #define RX_EV_Q_LABEL_LBN 32
++ #define RX_EV_Q_LABEL_WIDTH 5
++ #define RX_JUMBO_CONT_LBN 31
++ #define RX_JUMBO_CONT_WIDTH 1
++ #define RX_SOP_LBN 15
++ #define RX_SOP_WIDTH 1
++ #define RX_PORT_LBN 30
++ #define RX_PORT_WIDTH 1
++ #define RX_EV_BYTE_CNT_LBN 16
++ #define RX_EV_BYTE_CNT_WIDTH 14
++ #define RX_iSCSI_PKT_OK_LBN 14
++ #define RX_iSCSI_PKT_OK_WIDTH 1
++ #define RX_ISCSI_DDIG_ERR_LBN 13
++ #define RX_ISCSI_DDIG_ERR_WIDTH 1
++ #define RX_ISCSI_HDIG_ERR_LBN 12
++ #define RX_ISCSI_HDIG_ERR_WIDTH 1
++ #define RX_EV_DESC_PTR_LBN 0
++ #define RX_EV_DESC_PTR_WIDTH 12
++//////////////---- Transmit IP events for both Kernel & User event queues ----//////////////
++ #define TX_EV_PKT_ERR_LBN 38
++ #define TX_EV_PKT_ERR_WIDTH 1
++ #define TX_EV_PKT_TOO_BIG_LBN 37
++ #define TX_EV_PKT_TOO_BIG_WIDTH 1
++ #define TX_EV_Q_LABEL_LBN 32
++ #define TX_EV_Q_LABEL_WIDTH 5
++ #define TX_EV_PORT_LBN 16
++ #define TX_EV_PORT_WIDTH 1
++ #define TX_EV_WQ_FF_FULL_LBN 15
++ #define TX_EV_WQ_FF_FULL_WIDTH 1
++ #define TX_EV_BUF_OWNER_ID_ERR_LBN 14
++ #define TX_EV_BUF_OWNER_ID_ERR_WIDTH 1
++ #define TX_EV_COMP_LBN 12
++ #define TX_EV_COMP_WIDTH 1
++ #define TX_EV_DESC_PTR_LBN 0
++ #define TX_EV_DESC_PTR_WIDTH 12
++//////////////---- Char or Kernel driver events ----//////////////
++ #define DRIVER_EV_SUB_CODE_LBN 56
++ #define DRIVER_EV_SUB_CODE_WIDTH 4
++ #define TX_DESCQ_FLS_DONE_EV_DECODE 0x0
++ #define RX_DESCQ_FLS_DONE_EV_DECODE 0x1
++ #define EVQ_INIT_DONE_EV_DECODE 0x2
++ #define EVQ_NOT_EN_EV_DECODE 0x3
++ #define RX_DESCQ_FLSFF_OVFL_EV_DECODE 0x4
++ #define SRM_UPD_DONE_EV_DECODE 0x5
++ #define WAKE_UP_EV_DECODE 0x6
++ #define TX_PKT_NON_TCP_UDP_DECODE 0x9
++ #define TIMER_EV_DECODE 0xA
++ #define RX_DSC_ERROR_EV_DECODE 0xE
++ #define DRIVER_EV_TX_DESCQ_ID_LBN 0
++ #define DRIVER_EV_TX_DESCQ_ID_WIDTH 12
++ #define DRIVER_EV_RX_DESCQ_ID_LBN 0
++ #define DRIVER_EV_RX_DESCQ_ID_WIDTH 12
++ #define DRIVER_EV_EVQ_ID_LBN 0
++ #define DRIVER_EV_EVQ_ID_WIDTH 12
++ #define DRIVER_TMR_ID_LBN 0
++ #define DRIVER_TMR_ID_WIDTH 12
++ #define DRIVER_EV_SRM_UPD_LBN 0
++ #define DRIVER_EV_SRM_UPD_WIDTH 2
++ #define SRM_CLR_EV_DECODE 0
++ #define SRM_UPD_EV_DECODE 1
++ #define SRM_ILLCLR_EV_DECODE 2
++//////////////---- Global events. Sent to both event queue 0 and 4. ----//////////////
++ #define XFP_PHY_INTR_LBN 10
++ #define XFP_PHY_INTR_WIDTH 1
++ #define XG_PHY_INTR_LBN 9
++ #define XG_PHY_INTR_WIDTH 1
++ #define G_PHY1_INTR_LBN 8
++ #define G_PHY1_INTR_WIDTH 1
++ #define G_PHY0_INTR_LBN 7
++ #define G_PHY0_INTR_WIDTH 1
++//////////////---- Driver generated events ----//////////////
++ #define DRV_GEN_EV_CODE_LBN 60
++ #define DRV_GEN_EV_CODE_WIDTH 4
++ #define DRV_GEN_EV_DATA_LBN 0
++ #define DRV_GEN_EV_DATA_WIDTH 60
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/ef_vi_internal.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/ef_vi_internal.h 2008-05-19 00:33:48.674951800 +0300
+@@ -0,0 +1,256 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ buf = kzalloc((sizeof(*res) * 2) + 1, GFP_KERNEL);
-+ if (!buf)
-+ return AE_OK;
++/*
++ * \author djr
++ * \brief Really-and-truely-honestly internal stuff for libef.
++ * \date 2004/06/13
++ */
+
-+ /* Clean out resource_source */
-+ res->data.address64.resource_source.index = 0xFF;
-+ res->data.address64.resource_source.string_length = 0;
-+ res->data.address64.resource_source.string_ptr = NULL;
++/*! \cidoxg_include_ci_ul */
++#ifndef __CI_EF_VI_INTERNAL_H__
++#define __CI_EF_VI_INTERNAL_H__
+
-+ ptr = (unsigned char *)res;
+
-+ /* Turn the acpi_resource into an ASCII byte stream */
-+ for (i = 0; i < sizeof(*res); i++) {
-+ snprintf(tmp, sizeof(tmp), "%02x", ptr[i]);
-+ strncat(buf, tmp, 2);
-+ }
++/* These flags share space with enum ef_vi_flags. */
++#define EF_VI_BUG5692_WORKAROUND 0x10000
+
-+ err = xenbus_printf(XBT_NIL, info->pdev->xdev->nodename,
-+ str, "%s", buf);
+
-+ if (!err)
-+ info->resource_count++;
++/* ***********************************************************************
++ * COMPILATION CONTROL FLAGS (see ef_vi.h for "workaround" controls)
++ */
+
-+ kfree(buf);
++#define EF_VI_DO_MAGIC_CHECKS 1
+
-+ return AE_OK;
-+}
+
-+int pciback_publish_pci_roots(struct pciback_device *pdev,
-+ publish_pci_root_cb publish_root_cb)
-+{
-+ struct controller_dev_data *dev_data = pdev->pci_dev_data;
-+ struct controller_list_entry *cntrl_entry;
-+ int i, root_num, len, err = 0;
-+ unsigned int domain, bus;
-+ char str[64];
-+ struct walk_info info;
++/**********************************************************************
++ * Headers
++ */
+
-+ spin_lock(&dev_data->lock);
++#include <etherfabric/ef_vi.h>
++#include "sysdep.h"
++#include "ef_vi_falcon.h"
+
-+ list_for_each_entry(cntrl_entry, &dev_data->list, list) {
-+ /* First publish all the domain:bus info */
-+ err = publish_root_cb(pdev, cntrl_entry->domain,
-+ cntrl_entry->bus);
-+ if (err)
-+ goto out;
+
-+ /*
-+ * Now figure out which root-%d this belongs to
-+ * so we can associate resources with it.
-+ */
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
-+ "root_num", "%d", &root_num);
++/**********************************************************************
++ * Debugging.
++ */
+
-+ if (err != 1)
-+ goto out;
++#ifndef NDEBUG
+
-+ for (i = 0; i < root_num; i++) {
-+ len = snprintf(str, sizeof(str), "root-%d", i);
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++# define _ef_assert(exp, file, line) BUG_ON(!(exp));
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
-+ str, "%x:%x", &domain, &bus);
-+ if (err != 2)
-+ goto out;
++# define _ef_assert2(exp, x, y, file, line) do { \
++ if (unlikely(!(exp))) \
++ BUG(); \
++ } while (0)
+
-+ /* Is this the one we just published? */
-+ if (domain == cntrl_entry->domain &&
-+ bus == cntrl_entry->bus)
-+ break;
-+ }
++#else
+
-+ if (i == root_num)
-+ goto out;
++# define _ef_assert(exp, file, line)
++# define _ef_assert2(e, x, y, file, line)
++
++#endif
++
++#define ef_assert(a) do{ _ef_assert((a),__FILE__,__LINE__); } while(0)
++#define ef_assert_equal(a,b) _ef_assert2((a)==(b),(a),(b),__FILE__,__LINE__)
++#define ef_assert_eq ef_assert_equal
++#define ef_assert_lt(a,b) _ef_assert2((a)<(b),(a),(b),__FILE__,__LINE__)
++#define ef_assert_le(a,b) _ef_assert2((a)<=(b),(a),(b),__FILE__,__LINE__)
++#define ef_assert_nequal(a,b) _ef_assert2((a)!=(b),(a),(b),__FILE__,__LINE__)
++#define ef_assert_ne ef_assert_nequal
++#define ef_assert_ge(a,b) _ef_assert2((a)>=(b),(a),(b),__FILE__,__LINE__)
++#define ef_assert_gt(a,b) _ef_assert2((a)>(b),(a),(b),__FILE__,__LINE__)
++
++/**********************************************************************
++ * Debug checks. ******************************************************
++ **********************************************************************/
++
++#ifdef NDEBUG
++# define EF_VI_MAGIC_SET(p, type)
++# define EF_VI_CHECK_VI(p)
++# define EF_VI_CHECK_EVENT_Q(p)
++# define EF_VI_CHECK_IOBUFSET(p)
++# define EF_VI_CHECK_FILTER(p)
++# define EF_VI_CHECK_SHMBUF(p)
++# define EF_VI_CHECK_PT_EP(p)
++#else
++# define EF_VI 0x3
++# define EF_EPLOCK 0x6
++# define EF_IOBUFSET 0x9
++# define EF_FILTER 0xa
++# define EF_SHMBUF 0x11
++
++# define EF_VI_MAGIC(p, type) \
++ (((unsigned)(type) << 28) | \
++ (((unsigned)(intptr_t)(p)) & 0x0fffffffu))
++
++# if !EF_VI_DO_MAGIC_CHECKS
++# define EF_VI_MAGIC_SET(p, type)
++# define EF_VI_MAGIC_CHECK(p, type)
++# else
++# define EF_VI_MAGIC_SET(p, type) \
++ do { \
++ (p)->magic = EF_VI_MAGIC((p), (type)); \
++ } while (0)
+
-+ info.pdev = pdev;
-+ info.resource_count = 0;
-+ info.root_num = i;
++# define EF_VI_MAGIC_OKAY(p, type) \
++ ((p)->magic == EF_VI_MAGIC((p), (type)))
+
-+ /* Let ACPI do the heavy lifting on decoding resources */
-+ acpi_walk_resources(cntrl_entry->controller->acpi_handle,
-+ METHOD_NAME__CRS, write_xenbus_resource,
-+ &info);
++# define EF_VI_MAGIC_CHECK(p, type) \
++ ef_assert(EF_VI_MAGIC_OKAY((p), (type)))
+
-+ /* No resouces. OK. On to the next one */
-+ if (!info.resource_count)
-+ continue;
++#endif /* EF_VI_DO_MAGIC_CHECKS */
+
-+ /* Store the number of resources we wrote for this root-%d */
-+ len = snprintf(str, sizeof(str), "root-%d-resources", i);
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++# define EF_VI_CHECK_VI(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_VI);
+
-+ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
-+ "%d", info.resource_count);
-+ if (err)
-+ goto out;
-+ }
++# define EF_VI_CHECK_EVENT_Q(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_VI); \
++ ef_assert((p)->evq_base); \
++ ef_assert((p)->evq_mask);
+
-+ /* Finally, write some magic to synchronize with the guest. */
-+ len = snprintf(str, sizeof(str), "root-resource-magic");
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++# define EF_VI_CHECK_PT_EP(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_VI); \
++ ef_assert((p)->ep_state);
+
-+ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
-+ "%lx", (sizeof(struct acpi_resource) * 2) + 1);
++# define EF_VI_CHECK_IOBUFSET(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_IOBUFSET)
+
-+out:
-+ spin_unlock(&dev_data->lock);
++# define EF_VI_CHECK_FILTER(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_FILTER);
+
-+ return err;
-+}
++# define EF_VI_CHECK_SHMBUF(p) \
++ ef_assert(p); \
++ EF_VI_MAGIC_CHECK((p), EF_SHMBUF);
+
-+void pciback_release_devices(struct pciback_device *pdev)
-+{
-+ struct controller_dev_data *dev_data = pdev->pci_dev_data;
-+ struct controller_list_entry *cntrl_entry, *c;
-+ struct controller_dev_entry *dev_entry, *d;
++#endif
+
-+ list_for_each_entry_safe(cntrl_entry, c, &dev_data->list, list) {
-+ list_for_each_entry_safe(dev_entry, d,
-+ &cntrl_entry->dev_list, list) {
-+ list_del(&dev_entry->list);
-+ pcistub_put_pci_dev(dev_entry->dev);
-+ kfree(dev_entry);
-+ }
-+ list_del(&cntrl_entry->list);
-+ kfree(cntrl_entry);
-+ }
++#ifndef NDEBUG
++# define EF_DRIVER_MAGIC 0x00f00ba4
++# define EF_ASSERT_THIS_DRIVER_VALID(driver) \
++ do{ ef_assert(driver); \
++ EF_VI_MAGIC_CHECK((driver), EF_DRIVER_MAGIC); \
++ ef_assert((driver)->init); }while(0)
+
-+ kfree(dev_data);
-+ pdev->pci_dev_data = NULL;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/Makefile linux-2.6.18-xen.hg/drivers/xen/pciback/Makefile
---- linux-2.6.18/drivers/xen/pciback/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/Makefile 2007-12-23 11:15:34.061269291 +0100
-@@ -0,0 +1,16 @@
-+obj-$(CONFIG_XEN_PCIDEV_BACKEND) += pciback.o
++# define EF_ASSERT_DRIVER_VALID() EF_ASSERT_THIS_DRIVER_VALID(&ci_driver)
++#else
++# define EF_ASSERT_THIS_DRIVER_VALID(driver)
++# define EF_ASSERT_DRIVER_VALID()
++#endif
+
-+pciback-y := pci_stub.o pciback_ops.o xenbus.o
-+pciback-y += conf_space.o conf_space_header.o \
-+ conf_space_capability.o \
-+ conf_space_capability_vpd.o \
-+ conf_space_capability_pm.o \
-+ conf_space_quirks.o
-+pciback-$(CONFIG_XEN_PCIDEV_BACKEND_VPCI) += vpci.o
-+pciback-$(CONFIG_XEN_PCIDEV_BACKEND_SLOT) += slot.o
-+pciback-$(CONFIG_XEN_PCIDEV_BACKEND_PASS) += passthrough.o
-+pciback-$(CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER) += controller.o
+
-+ifeq ($(CONFIG_XEN_PCIDEV_BE_DEBUG),y)
-+EXTRA_CFLAGS += -DDEBUG
-+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/passthrough.c linux-2.6.18-xen.hg/drivers/xen/pciback/passthrough.c
---- linux-2.6.18/drivers/xen/pciback/passthrough.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/passthrough.c 2007-12-23 11:15:34.064602797 +0100
-@@ -0,0 +1,157 @@
-+/*
-+ * PCI Backend - Provides restricted access to the real PCI bus topology
-+ * to the frontend
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++/* *************************************
++ * Power of 2 FIFO
+ */
+
-+#include <linux/list.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include "pciback.h"
-+
-+struct passthrough_dev_data {
-+ /* Access to dev_list must be protected by lock */
-+ struct list_head dev_list;
-+ spinlock_t lock;
-+};
-+
-+struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus,
-+ unsigned int devfn)
-+{
-+ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
-+ struct pci_dev_entry *dev_entry;
-+ struct pci_dev *dev = NULL;
-+ unsigned long flags;
-+
-+ spin_lock_irqsave(&dev_data->lock, flags);
-+
-+ list_for_each_entry(dev_entry, &dev_data->dev_list, list) {
-+ if (domain == (unsigned int)pci_domain_nr(dev_entry->dev->bus)
-+ && bus == (unsigned int)dev_entry->dev->bus->number
-+ && devfn == dev_entry->dev->devfn) {
-+ dev = dev_entry->dev;
-+ break;
-+ }
-+ }
++#define EF_VI_FIFO2_M(f, x) ((x) & ((f)->fifo_mask))
++#define ef_vi_fifo2_valid(f) ((f) && (f)->fifo && (f)->fifo_mask > 0 && \
++ (f)->fifo_rd_i <= (f)->fifo_mask && \
++ (f)->fifo_wr_i <= (f)->fifo_mask && \
++ EF_VI_IS_POW2((f)->fifo_mask+1u))
+
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
++#define ef_vi_fifo2_init(f, cap) \
++ do{ ef_assert(EF_VI_IS_POW2((cap) + 1)); \
++ (f)->fifo_rd_i = (f)->fifo_wr_i = 0u; \
++ (f)->fifo_mask = (cap); \
++ }while(0)
+
-+ return dev;
-+}
++#define ef_vi_fifo2_is_empty(f) ((f)->fifo_rd_i == (f)->fifo_wr_i)
++#define ef_vi_fifo2_capacity(f) ((f)->fifo_mask)
++#define ef_vi_fifo2_buf_size(f) ((f)->fifo_mask + 1u)
++#define ef_vi_fifo2_end(f) ((f)->fifo + ef_vi_fifo2_buf_size(f))
++#define ef_vi_fifo2_peek(f) ((f)->fifo[(f)->fifo_rd_i])
++#define ef_vi_fifo2_poke(f) ((f)->fifo[(f)->fifo_wr_i])
++#define ef_vi_fifo2_num(f) EF_VI_FIFO2_M((f),(f)->fifo_wr_i-(f)->fifo_rd_i)
+
-+int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
-+ struct pci_dev_entry *dev_entry;
-+ unsigned long flags;
++#define ef_vi_fifo2_wr_prev(f) \
++ do{ (f)->fifo_wr_i = EF_VI_FIFO2_M((f), (f)->fifo_wr_i - 1u); }while(0)
++#define ef_vi_fifo2_wr_next(f) \
++ do{ (f)->fifo_wr_i = EF_VI_FIFO2_M((f), (f)->fifo_wr_i + 1u); }while(0)
++#define ef_vi_fifo2_rd_adv(f, n) \
++ do{ (f)->fifo_rd_i = EF_VI_FIFO2_M((f), (f)->fifo_rd_i + (n)); }while(0)
++#define ef_vi_fifo2_rd_prev(f) \
++ do{ (f)->fifo_rd_i = EF_VI_FIFO2_M((f), (f)->fifo_rd_i - 1u); }while(0)
++#define ef_vi_fifo2_rd_next(f) \
++ do{ (f)->fifo_rd_i = EF_VI_FIFO2_M((f), (f)->fifo_rd_i + 1u); }while(0)
+
-+ dev_entry = kmalloc(sizeof(*dev_entry), GFP_KERNEL);
-+ if (!dev_entry)
-+ return -ENOMEM;
-+ dev_entry->dev = dev;
++#define ef_vi_fifo2_put(f, v) \
++ do{ ef_vi_fifo2_poke(f) = (v); ef_vi_fifo2_wr_next(f); }while(0)
++#define ef_vi_fifo2_get(f, pv) \
++ do{ *(pv) = ef_vi_fifo2_peek(f); ef_vi_fifo2_rd_next(f); }while(0)
+
-+ spin_lock_irqsave(&dev_data->lock, flags);
-+ list_add_tail(&dev_entry->list, &dev_data->dev_list);
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
+
-+ return 0;
-+}
++/* *********************************************************************
++ * Eventq handling
++ */
+
-+void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
-+ struct pci_dev_entry *dev_entry, *t;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
++typedef union {
++ uint64_t u64;
++ struct {
++ uint32_t a;
++ uint32_t b;
++ } opaque;
++} ef_vi_event;
+
-+ spin_lock_irqsave(&dev_data->lock, flags);
+
-+ list_for_each_entry_safe(dev_entry, t, &dev_data->dev_list, list) {
-+ if (dev_entry->dev == dev) {
-+ list_del(&dev_entry->list);
-+ found_dev = dev_entry->dev;
-+ kfree(dev_entry);
-+ }
-+ }
++#define EF_VI_EVENT_OFFSET(q, i) \
++ (((q)->evq_state->evq_ptr - (i) * sizeof(ef_vi_event)) & (q)->evq_mask)
+
-+ spin_unlock_irqrestore(&dev_data->lock, flags);
++#define EF_VI_EVENT_PTR(q, i) \
++ ((ef_vi_event*) ((q)->evq_base + EF_VI_EVENT_OFFSET((q), (i))))
+
-+ if (found_dev)
-+ pcistub_put_pci_dev(found_dev);
-+}
++/* *********************************************************************
++ * Miscellaneous goodies
++ */
++#ifdef NDEBUG
++# define EF_VI_DEBUG(x)
++#else
++# define EF_VI_DEBUG(x) x
++#endif
+
-+int pciback_init_devices(struct pciback_device *pdev)
-+{
-+ struct passthrough_dev_data *dev_data;
++#define EF_VI_ROUND_UP(i, align) (((i)+(align)-1u) & ~((align)-1u))
++#define EF_VI_ALIGN_FWD(p, align) (((p)+(align)-1u) & ~((align)-1u))
++#define EF_VI_ALIGN_BACK(p, align) ((p) & ~((align)-1u))
++#define EF_VI_PTR_ALIGN_BACK(p, align) \
++ ((char*)EF_VI_ALIGN_BACK(((intptr_t)(p)), ((intptr_t)(align))))
++#define EF_VI_IS_POW2(x) ((x) && ! ((x) & ((x) - 1)))
+
-+ dev_data = kmalloc(sizeof(*dev_data), GFP_KERNEL);
-+ if (!dev_data)
-+ return -ENOMEM;
+
-+ spin_lock_init(&dev_data->lock);
++/* ********************************************************************
++ */
+
-+ INIT_LIST_HEAD(&dev_data->dev_list);
++extern void falcon_vi_init(ef_vi*, void* vvis ) EF_VI_HF;
++extern void ef_eventq_state_init(ef_vi* evq) EF_VI_HF;
++extern void __ef_init(void) EF_VI_HF;
+
-+ pdev->pci_dev_data = dev_data;
+
-+ return 0;
-+}
++#endif /* __CI_EF_VI_INTERNAL_H__ */
+
-+int pciback_publish_pci_roots(struct pciback_device *pdev,
-+ publish_pci_root_cb publish_root_cb)
-+{
-+ int err = 0;
-+ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
-+ struct pci_dev_entry *dev_entry, *e;
-+ struct pci_dev *dev;
-+ int found;
-+ unsigned int domain, bus;
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/etherfabric/ef_vi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/etherfabric/ef_vi.h 2008-05-19 00:33:48.678952031 +0300
+@@ -0,0 +1,665 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ spin_lock(&dev_data->lock);
++/*
++ * \brief Virtual Interface
++ * \date 2007/05/16
++ */
+
-+ list_for_each_entry(dev_entry, &dev_data->dev_list, list) {
-+ /* Only publish this device as a root if none of its
-+ * parent bridges are exported
-+ */
-+ found = 0;
-+ dev = dev_entry->dev->bus->self;
-+ for (; !found && dev != NULL; dev = dev->bus->self) {
-+ list_for_each_entry(e, &dev_data->dev_list, list) {
-+ if (dev == e->dev) {
-+ found = 1;
-+ break;
-+ }
-+ }
-+ }
++#ifndef __EFAB_EF_VI_H__
++#define __EFAB_EF_VI_H__
+
-+ domain = (unsigned int)pci_domain_nr(dev_entry->dev->bus);
-+ bus = (unsigned int)dev_entry->dev->bus->number;
+
-+ if (!found) {
-+ err = publish_root_cb(pdev, domain, bus);
-+ if (err)
-+ break;
-+ }
-+ }
++/**********************************************************************
++ * Primitive types ****************************************************
++ **********************************************************************/
+
-+ spin_unlock(&dev_data->lock);
++/* We standardise on the types from stdint.h and synthesise these types
++ * for compilers/platforms that don't provide them */
+
-+ return err;
-+}
++# include <linux/types.h>
++# define EF_VI_ALIGN(x) __attribute__ ((aligned (x)))
++# define ef_vi_inline static inline
+
-+void pciback_release_devices(struct pciback_device *pdev)
-+{
-+ struct passthrough_dev_data *dev_data = pdev->pci_dev_data;
-+ struct pci_dev_entry *dev_entry, *t;
+
-+ list_for_each_entry_safe(dev_entry, t, &dev_data->dev_list, list) {
-+ list_del(&dev_entry->list);
-+ pcistub_put_pci_dev(dev_entry->dev);
-+ kfree(dev_entry);
-+ }
+
-+ kfree(dev_data);
-+ pdev->pci_dev_data = NULL;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/pciback.h linux-2.6.18-xen.hg/drivers/xen/pciback/pciback.h
---- linux-2.6.18/drivers/xen/pciback/pciback.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/pciback.h 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,93 @@
-+/*
-+ * PCI Backend Common Data Structures & Function Declarations
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+#ifndef __XEN_PCIBACK_H__
-+#define __XEN_PCIBACK_H__
++/**********************************************************************
++ * Types **************************************************************
++ **********************************************************************/
+
-+#include <linux/pci.h>
-+#include <linux/interrupt.h>
-+#include <xen/xenbus.h>
-+#include <linux/list.h>
-+#include <linux/spinlock.h>
-+#include <linux/workqueue.h>
-+#include <asm/atomic.h>
-+#include <xen/interface/io/pciif.h>
++typedef uint32_t ef_eventq_ptr;
+
-+struct pci_dev_entry {
-+ struct list_head list;
-+ struct pci_dev *dev;
-+};
++typedef uint64_t ef_addr;
++typedef char* ef_vi_ioaddr_t;
+
-+#define _PDEVF_op_active (0)
-+#define PDEVF_op_active (1<<(_PDEVF_op_active))
++/**********************************************************************
++ * ef_event ***********************************************************
++ **********************************************************************/
+
-+struct pciback_device {
-+ void *pci_dev_data;
-+ spinlock_t dev_lock;
++/*! \i_ef_vi A DMA request identifier.
++**
++** This is an integer token specified by the transport and associated
++** with a DMA request. It is returned to the VI user with DMA completion
++** events. It is typically used to identify the buffer associated with
++** the transfer.
++*/
++typedef int ef_request_id;
+
-+ struct xenbus_device *xdev;
++typedef union {
++ uint64_t u64[1];
++ uint32_t u32[2];
++} ef_vi_qword;
+
-+ struct xenbus_watch be_watch;
-+ u8 be_watching;
++typedef ef_vi_qword ef_hw_event;
+
-+ int evtchn_irq;
++#define EF_REQUEST_ID_BITS 16u
++#define EF_REQUEST_ID_MASK ((1u << EF_REQUEST_ID_BITS) - 1u)
+
-+ struct vm_struct *sh_area;
-+ struct xen_pci_sharedinfo *sh_info;
++/*! \i_ef_event An [ef_event] is a token that identifies something that
++** has happened. Examples include packets received, packets transmitted
++** and errors.
++*/
++typedef union {
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ } generic;
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ /*ef_request_id request_id :EF_REQUEST_ID_BITS;*/
++ unsigned q_id :16;
++ unsigned len :16;
++ unsigned flags :16;
++ } rx;
++ struct { /* This *must* have same layout as [rx]. */
++ ef_hw_event ev;
++ unsigned type :16;
++ /*ef_request_id request_id :EF_REQUEST_ID_BITS;*/
++ unsigned q_id :16;
++ unsigned len :16;
++ unsigned flags :16;
++ unsigned subtype :16;
++ } rx_discard;
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ /*ef_request_id request_id :EF_REQUEST_ID_BITS;*/
++ unsigned q_id :16;
++ } tx;
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ /*ef_request_id request_id :EF_REQUEST_ID_BITS;*/
++ unsigned q_id :16;
++ unsigned subtype :16;
++ } tx_error;
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ unsigned q_id :16;
++ } rx_no_desc_trunc;
++ struct {
++ ef_hw_event ev;
++ unsigned type :16;
++ unsigned data;
++ } sw;
++} ef_event;
+
-+ unsigned long flags;
+
-+ struct work_struct op_work;
++#define EF_EVENT_TYPE(e) ((e).generic.type)
++enum {
++ /** Good data was received. */
++ EF_EVENT_TYPE_RX,
++ /** Packets have been sent. */
++ EF_EVENT_TYPE_TX,
++ /** Data received and buffer consumed, but something is wrong. */
++ EF_EVENT_TYPE_RX_DISCARD,
++ /** Transmit of packet failed. */
++ EF_EVENT_TYPE_TX_ERROR,
++ /** Received packet was truncated due to lack of descriptors. */
++ EF_EVENT_TYPE_RX_NO_DESC_TRUNC,
++ /** Software generated event. */
++ EF_EVENT_TYPE_SW,
++ /** Event queue overflow. */
++ EF_EVENT_TYPE_OFLOW,
++};
++
++#define EF_EVENT_RX_BYTES(e) ((e).rx.len)
++#define EF_EVENT_RX_Q_ID(e) ((e).rx.q_id)
++#define EF_EVENT_RX_CONT(e) ((e).rx.flags & EF_EVENT_FLAG_CONT)
++#define EF_EVENT_RX_SOP(e) ((e).rx.flags & EF_EVENT_FLAG_SOP)
++#define EF_EVENT_RX_ISCSI_OKAY(e) ((e).rx.flags & EF_EVENT_FLAG_ISCSI_OK)
++#define EF_EVENT_FLAG_SOP 0x1
++#define EF_EVENT_FLAG_CONT 0x2
++#define EF_EVENT_FLAG_ISCSI_OK 0x4
++
++#define EF_EVENT_TX_Q_ID(e) ((e).tx.q_id)
++
++#define EF_EVENT_RX_DISCARD_Q_ID(e) ((e).rx_discard.q_id)
++#define EF_EVENT_RX_DISCARD_LEN(e) ((e).rx_discard.len)
++#define EF_EVENT_RX_DISCARD_TYPE(e) ((e).rx_discard.subtype)
++enum {
++ EF_EVENT_RX_DISCARD_CSUM_BAD,
++ EF_EVENT_RX_DISCARD_CRC_BAD,
++ EF_EVENT_RX_DISCARD_TRUNC,
++ EF_EVENT_RX_DISCARD_RIGHTS,
++ EF_EVENT_RX_DISCARD_OTHER,
+};
+
-+struct pciback_dev_data {
-+ struct list_head config_fields;
-+ int permissive;
-+ int warned_on_write;
++#define EF_EVENT_TX_ERROR_Q_ID(e) ((e).tx_error.q_id)
++#define EF_EVENT_TX_ERROR_TYPE(e) ((e).tx_error.subtype)
++enum {
++ EF_EVENT_TX_ERROR_RIGHTS,
++ EF_EVENT_TX_ERROR_OFLOW,
++ EF_EVENT_TX_ERROR_2BIG,
++ EF_EVENT_TX_ERROR_BUS,
+};
+
-+/* Get/Put PCI Devices that are hidden from the PCI Backend Domain */
-+struct pci_dev *pcistub_get_pci_dev_by_slot(struct pciback_device *pdev,
-+ int domain, int bus,
-+ int slot, int func);
-+struct pci_dev *pcistub_get_pci_dev(struct pciback_device *pdev,
-+ struct pci_dev *dev);
-+void pcistub_put_pci_dev(struct pci_dev *dev);
++#define EF_EVENT_RX_NO_DESC_TRUNC_Q_ID(e) ((e).rx_no_desc_trunc.q_id)
+
-+/* Ensure a device is turned off or reset */
-+void pciback_reset_device(struct pci_dev *pdev);
++#define EF_EVENT_SW_DATA_MASK 0xffff
++#define EF_EVENT_SW_DATA(e) ((e).sw.data)
+
-+/* Access a virtual configuration space for a PCI device */
-+int pciback_config_init(void);
-+int pciback_config_init_dev(struct pci_dev *dev);
-+void pciback_config_free_dyn_fields(struct pci_dev *dev);
-+void pciback_config_reset_dev(struct pci_dev *dev);
-+void pciback_config_free_dev(struct pci_dev *dev);
-+int pciback_config_read(struct pci_dev *dev, int offset, int size,
-+ u32 * ret_val);
-+int pciback_config_write(struct pci_dev *dev, int offset, int size, u32 value);
++#define EF_EVENT_FMT "[ev:%x:%08x:%08x]"
++#define EF_EVENT_PRI_ARG(e) (unsigned) (e).generic.type, \
++ (unsigned) (e).generic.ev.u32[1], \
++ (unsigned) (e).generic.ev.u32[0]
+
-+/* Handle requests for specific devices from the frontend */
-+typedef int (*publish_pci_root_cb) (struct pciback_device * pdev,
-+ unsigned int domain, unsigned int bus);
-+int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev);
-+void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev);
-+struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus,
-+ unsigned int devfn);
-+int pciback_init_devices(struct pciback_device *pdev);
-+int pciback_publish_pci_roots(struct pciback_device *pdev,
-+ publish_pci_root_cb cb);
-+void pciback_release_devices(struct pciback_device *pdev);
++#define EF_GET_HW_EV(e) ((e).generic.ev)
++#define EF_GET_HW_EV_PTR(e) (&(e).generic.ev)
++#define EF_GET_HW_EV_U64(e) ((e).generic.ev.u64[0])
+
-+/* Handles events from front-end */
-+irqreturn_t pciback_handle_event(int irq, void *dev_id, struct pt_regs *regs);
-+void pciback_do_op(void *data);
+
-+int pciback_xenbus_register(void);
-+void pciback_xenbus_unregister(void);
++/* ***************** */
+
-+extern int verbose_request;
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/pciback_ops.c linux-2.6.18-xen.hg/drivers/xen/pciback/pciback_ops.c
---- linux-2.6.18/drivers/xen/pciback/pciback_ops.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/pciback_ops.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,95 @@
-+/*
-+ * PCI Backend Operations - respond to PCI requests from Frontend
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+#include <linux/module.h>
-+#include <asm/bitops.h>
-+#include <xen/evtchn.h>
-+#include "pciback.h"
++/*! Used by netif shared state. Must use types of explicit size. */
++typedef struct {
++ uint16_t rx_last_desc_ptr; /* for RX duplicates */
++ uint8_t bad_sop; /* bad SOP detected */
++ uint8_t frag_num; /* next fragment #, 0=>SOP */
++} ef_rx_dup_state_t;
+
-+int verbose_request = 0;
-+module_param(verbose_request, int, 0644);
+
-+/* Ensure a device is "turned off" and ready to be exported.
-+ * (Also see pciback_config_reset to ensure virtual configuration space is
-+ * ready to be re-exported)
-+ */
-+void pciback_reset_device(struct pci_dev *dev)
-+{
-+ u16 cmd;
++/* Max number of ports on any SF NIC. */
++#define EFAB_DMAQS_PER_EVQ_MAX 32
+
-+ /* Disable devices (but not bridges) */
-+ if (dev->hdr_type == PCI_HEADER_TYPE_NORMAL) {
-+ pci_disable_device(dev);
++typedef struct {
++ ef_eventq_ptr evq_ptr;
++ int32_t trashed;
++ ef_rx_dup_state_t rx_dup_state[EFAB_DMAQS_PER_EVQ_MAX];
++} ef_eventq_state;
+
-+ pci_write_config_word(dev, PCI_COMMAND, 0);
+
-+ dev->is_enabled = 0;
-+ dev->is_busmaster = 0;
-+ } else {
-+ pci_read_config_word(dev, PCI_COMMAND, &cmd);
-+ if (cmd & (PCI_COMMAND_INVALIDATE)) {
-+ cmd &= ~(PCI_COMMAND_INVALIDATE);
-+ pci_write_config_word(dev, PCI_COMMAND, cmd);
++/*! \i_ef_base [ef_iovec] is similar the standard [struct iovec]. An
++** array of these is used to designate a scatter/gather list of I/O
++** buffers.
++*/
++typedef struct {
++ ef_addr iov_base EF_VI_ALIGN(8);
++ unsigned iov_len;
++} ef_iovec;
+
-+ dev->is_busmaster = 0;
-+ }
-+ }
-+}
++/* Falcon constants */
++#define TX_EV_DESC_PTR_LBN 0
+
-+static inline void test_and_schedule_op(struct pciback_device *pdev)
-+{
-+ /* Check that frontend is requesting an operation and that we are not
-+ * already processing a request */
-+ if (test_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags)
-+ && !test_and_set_bit(_PDEVF_op_active, &pdev->flags))
-+ schedule_work(&pdev->op_work);
-+}
++/**********************************************************************
++ * ef_iobufset ********************************************************
++ **********************************************************************/
+
-+/* Performing the configuration space reads/writes must not be done in atomic
-+ * context because some of the pci_* functions can sleep (mostly due to ACPI
-+ * use of semaphores). This function is intended to be called from a work
-+ * queue in process context taking a struct pciback_device as a parameter */
-+void pciback_do_op(void *data)
-+{
-+ struct pciback_device *pdev = data;
-+ struct pci_dev *dev;
-+ struct xen_pci_op *op = &pdev->sh_info->op;
++/*! \i_ef_bufs An [ef_iobufset] is a collection of buffers to be used
++** with the NIC.
++*/
++typedef struct ef_iobufset {
++ unsigned magic;
++ unsigned bufs_mmap_bytes;
++ unsigned bufs_handle;
++ int bufs_ptr_off;
++ ef_addr bufs_addr;
++ unsigned bufs_size; /* size rounded to pow2 */
++ int bufs_num;
++ int faultonaccess;
++} ef_iobufset;
++
++
++/**********************************************************************
++ * ef_vi **************************************************************
++ **********************************************************************/
++
++enum ef_vi_flags {
++ EF_VI_RX_SCATTER = 0x1,
++ EF_VI_ISCSI_RX_HDIG = 0x2,
++ EF_VI_ISCSI_TX_HDIG = 0x4,
++ EF_VI_ISCSI_RX_DDIG = 0x8,
++ EF_VI_ISCSI_TX_DDIG = 0x10,
++ EF_VI_TX_PHYS_ADDR = 0x20,
++ EF_VI_RX_PHYS_ADDR = 0x40,
++ EF_VI_TX_IP_CSUM_DIS = 0x80,
++ EF_VI_TX_TCPUDP_CSUM_DIS= 0x100,
++ EF_VI_TX_TCPUDP_ONLY = 0x200,
++ /* Flags in range 0xXXXX0000 are for internal use. */
++};
+
-+ dev = pciback_get_pci_dev(pdev, op->domain, op->bus, op->devfn);
++typedef struct {
++ uint32_t added;
++ uint32_t removed;
++} ef_vi_txq_state;
+
-+ if (dev == NULL)
-+ op->err = XEN_PCI_ERR_dev_not_found;
-+ else if (op->cmd == XEN_PCI_OP_conf_read)
-+ op->err = pciback_config_read(dev, op->offset, op->size,
-+ &op->value);
-+ else if (op->cmd == XEN_PCI_OP_conf_write)
-+ op->err = pciback_config_write(dev, op->offset, op->size,
-+ op->value);
-+ else
-+ op->err = XEN_PCI_ERR_not_implemented;
++typedef struct {
++ uint32_t added;
++ uint32_t removed;
++} ef_vi_rxq_state;
+
-+ /* Tell the driver domain that we're done. */
-+ wmb();
-+ clear_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags);
-+ notify_remote_via_irq(pdev->evtchn_irq);
++typedef struct {
++ uint32_t mask;
++ void* doorbell;
++ void* descriptors;
++ uint16_t* ids;
++ unsigned misalign_mask;
++} ef_vi_txq;
+
-+ /* Mark that we're done. */
-+ smp_mb__before_clear_bit(); /* /after/ clearing PCIF_active */
-+ clear_bit(_PDEVF_op_active, &pdev->flags);
-+ smp_mb__after_clear_bit(); /* /before/ final check for work */
++typedef struct {
++ uint32_t mask;
++ void* doorbell;
++ void* descriptors;
++ uint16_t* ids;
++} ef_vi_rxq;
+
-+ /* Check to see if the driver domain tried to start another request in
-+ * between clearing _XEN_PCIF_active and clearing _PDEVF_op_active. */
-+ test_and_schedule_op(pdev);
-+}
++typedef struct {
++ ef_eventq_state evq;
++ ef_vi_txq_state txq;
++ ef_vi_rxq_state rxq;
++ /* Followed by request id fifos. */
++} ef_vi_state;
++
++/*! \i_ef_vi A virtual interface.
++**
++** An [ef_vi] represents a virtual interface on a specific NIC. A
++** virtual interface is a collection of an event queue and two DMA queues
++** used to pass Ethernet frames between the transport implementation and
++** the network.
++*/
++typedef struct ef_vi {
++ unsigned magic;
+
-+irqreturn_t pciback_handle_event(int irq, void *dev_id, struct pt_regs *regs)
-+{
-+ struct pciback_device *pdev = dev_id;
++ unsigned vi_resource_id;
++ unsigned vi_resource_handle_hack;
++ unsigned vi_i;
+
-+ test_and_schedule_op(pdev);
++ char* vi_mem_mmap_ptr;
++ int vi_mem_mmap_bytes;
++ char* vi_io_mmap_ptr;
++ int vi_io_mmap_bytes;
+
-+ return IRQ_HANDLED;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/pci_stub.c linux-2.6.18-xen.hg/drivers/xen/pciback/pci_stub.c
---- linux-2.6.18/drivers/xen/pciback/pci_stub.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/pci_stub.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,929 @@
-+/*
-+ * PCI Stub Driver - Grabs devices in backend to be exported later
-+ *
-+ * Ryan Wilson <hap9@epoch.ncsc.mil>
-+ * Chris Bookholt <hap10@epoch.ncsc.mil>
-+ */
-+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <linux/list.h>
-+#include <linux/spinlock.h>
-+#include <linux/kref.h>
-+#include <asm/atomic.h>
-+#include "pciback.h"
-+#include "conf_space.h"
-+#include "conf_space_quirks.h"
++ ef_eventq_state* evq_state;
++ char* evq_base;
++ unsigned evq_mask;
++ ef_vi_ioaddr_t evq_timer_reg;
+
-+static char *pci_devs_to_hide = NULL;
-+module_param_named(hide, pci_devs_to_hide, charp, 0444);
++ ef_vi_txq vi_txq;
++ ef_vi_rxq vi_rxq;
++ ef_vi_state* ep_state;
++ enum ef_vi_flags vi_flags;
++} ef_vi;
+
-+struct pcistub_device_id {
-+ struct list_head slot_list;
-+ int domain;
-+ unsigned char bus;
-+ unsigned int devfn;
++
++enum ef_vi_arch {
++ EF_VI_ARCH_FALCON,
+};
-+static LIST_HEAD(pcistub_device_ids);
-+static DEFINE_SPINLOCK(device_ids_lock);
+
-+struct pcistub_device {
-+ struct kref kref;
-+ struct list_head dev_list;
-+ spinlock_t lock;
+
-+ struct pci_dev *dev;
-+ struct pciback_device *pdev; /* non-NULL if struct pci_dev is in use */
++struct ef_vi_nic_type {
++ unsigned char arch;
++ char variant;
++ unsigned char revision;
+};
+
-+/* Access to pcistub_devices & seized_devices lists and the initialize_devices
-+ * flag must be locked with pcistub_devices_lock
-+ */
-+static DEFINE_SPINLOCK(pcistub_devices_lock);
-+static LIST_HEAD(pcistub_devices);
+
-+/* wait for device_initcall before initializing our devices
-+ * (see pcistub_init_devices_late)
++/* This structure is opaque to the client & used to pass mapping data
++ * from the resource manager to the ef_vi lib. for ef_vi_init().
+ */
-+static int initialize_devices = 0;
-+static LIST_HEAD(seized_devices);
++struct vi_mappings {
++ uint32_t signature;
++# define VI_MAPPING_VERSION 0x02 /*Byte: Increment me if struct altered*/
++# define VI_MAPPING_SIGNATURE (0xBA1150 + VI_MAPPING_VERSION)
+
-+static struct pcistub_device *pcistub_device_alloc(struct pci_dev *dev)
-+{
-+ struct pcistub_device *psdev;
++ struct ef_vi_nic_type nic_type;
+
-+ dev_dbg(&dev->dev, "pcistub_device_alloc\n");
++ int vi_instance;
+
-+ psdev = kzalloc(sizeof(*psdev), GFP_ATOMIC);
-+ if (!psdev)
-+ return NULL;
++ unsigned evq_bytes;
++ char* evq_base;
++ ef_vi_ioaddr_t evq_timer_reg;
+
-+ psdev->dev = pci_dev_get(dev);
-+ if (!psdev->dev) {
-+ kfree(psdev);
-+ return NULL;
-+ }
++ unsigned rx_queue_capacity;
++ ef_vi_ioaddr_t rx_dma_ef1;
++ char* rx_dma_falcon;
++ ef_vi_ioaddr_t rx_bell;
+
-+ kref_init(&psdev->kref);
-+ spin_lock_init(&psdev->lock);
++ unsigned tx_queue_capacity;
++ ef_vi_ioaddr_t tx_dma_ef1;
++ char* tx_dma_falcon;
++ ef_vi_ioaddr_t tx_bell;
++};
++/* This is used by clients to allocate a suitably sized buffer for the
++ * resource manager to fill & ef_vi_init() to use. */
++#define VI_MAPPINGS_SIZE (sizeof(struct vi_mappings))
+
-+ return psdev;
-+}
+
-+/* Don't call this directly as it's called by pcistub_device_put */
-+static void pcistub_device_release(struct kref *kref)
-+{
-+ struct pcistub_device *psdev;
++/**********************************************************************
++ * ef_config **********************************************************
++ **********************************************************************/
+
-+ psdev = container_of(kref, struct pcistub_device, kref);
++struct ef_config_t {
++ int log; /* debug logging level */
++};
+
-+ dev_dbg(&psdev->dev->dev, "pcistub_device_release\n");
++extern struct ef_config_t ef_config;
+
-+ /* Clean-up the device */
-+ pciback_reset_device(psdev->dev);
-+ pciback_config_free_dyn_fields(psdev->dev);
-+ pciback_config_free_dev(psdev->dev);
-+ kfree(pci_get_drvdata(psdev->dev));
-+ pci_set_drvdata(psdev->dev, NULL);
+
-+ pci_dev_put(psdev->dev);
++/**********************************************************************
++ * ef_vi **************************************************************
++ **********************************************************************/
+
-+ kfree(psdev);
++/* Initialise [data_area] with information required to initialise an ef_vi.
++ * In the following, an unused param should be set to NULL. Note the case
++ * marked (*) of [iobuf_mmap] for falcon/driver; for normal driver this
++ * must be NULL.
++ *
++ * \param data_area [in,out] required, must ref at least VI_MAPPINGS_SIZE
++ * bytes
++ * \param evq_capacity [in] number of events in event queue. Specify 0 for
++ * no event queue.
++ * \param rxq_capacity [in] number of descriptors in RX DMA queue. Specify
++ * 0 for no RX queue.
++ * \param txq_capacity [in] number of descriptors in TX DMA queue. Specify
++ * 0 for no TX queue.
++ * \param mmap_info [in] mem-map info for resource
++ * \param io_mmap [in] ef1, required
++ * falcon, required
++ * \param iobuf_mmap [in] ef1, UL: unused
++ * falcon, UL: required
++ */
++extern void ef_vi_init_mapping_vi(void* data_area, struct ef_vi_nic_type,
++ unsigned rxq_capacity,
++ unsigned txq_capacity, int instance,
++ void* io_mmap, void* iobuf_mmap_rx,
++ void* iobuf_mmap_tx, enum ef_vi_flags);
++
++
++extern void ef_vi_init_mapping_evq(void* data_area, struct ef_vi_nic_type,
++ int instance, unsigned evq_bytes,
++ void* base, void* timer_reg);
++
++ef_vi_inline unsigned ef_vi_resource_id(ef_vi* vi)
++{
++ return vi->vi_resource_id;
+}
+
-+static inline void pcistub_device_get(struct pcistub_device *psdev)
-+{
-+ kref_get(&psdev->kref);
++ef_vi_inline enum ef_vi_flags ef_vi_flags(ef_vi* vi)
++{
++ return vi->vi_flags;
+}
+
-+static inline void pcistub_device_put(struct pcistub_device *psdev)
++
++/**********************************************************************
++ * Receive interface **************************************************
++ **********************************************************************/
++
++/*! \i_ef_vi Returns the amount of space in the RX descriptor ring.
++**
++** \return the amount of space in the queue.
++*/
++ef_vi_inline int ef_vi_receive_space(ef_vi* vi)
+{
-+ kref_put(&psdev->kref, pcistub_device_release);
++ ef_vi_rxq_state* qs = &vi->ep_state->rxq;
++ return vi->vi_rxq.mask - (qs->added - qs->removed);
+}
+
-+static struct pcistub_device *pcistub_device_find(int domain, int bus,
-+ int slot, int func)
++
++/*! \i_ef_vi Returns the fill level of the RX descriptor ring.
++**
++** \return the fill level of the queue.
++*/
++ef_vi_inline int ef_vi_receive_fill_level(ef_vi* vi)
+{
-+ struct pcistub_device *psdev = NULL;
-+ unsigned long flags;
++ ef_vi_rxq_state* qs = &vi->ep_state->rxq;
++ return qs->added - qs->removed;
++}
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
+
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (psdev->dev != NULL
-+ && domain == pci_domain_nr(psdev->dev->bus)
-+ && bus == psdev->dev->bus->number
-+ && PCI_DEVFN(slot, func) == psdev->dev->devfn) {
-+ pcistub_device_get(psdev);
-+ goto out;
-+ }
-+ }
++ef_vi_inline int ef_vi_receive_capacity(ef_vi* vi)
++{
++ return vi->vi_rxq.mask;
++}
+
-+ /* didn't find it */
-+ psdev = NULL;
++/*! \i_ef_vi Complete a receive operation.
++**
++** When a receive completion event is received, it should be passed to
++** this function. The request-id for the buffer that the packet was
++** delivered to is returned.
++**
++** After this function returns, more space may be available in the
++** receive queue.
++*/
++extern ef_request_id ef_vi_receive_done(const ef_vi*, const ef_event*);
+
-+ out:
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
-+ return psdev;
++/*! \i_ef_vi Return request ID indicated by a receive event
++ */
++ef_vi_inline ef_request_id ef_vi_receive_request_id(const ef_vi* vi,
++ const ef_event* ef_ev)
++{
++ const ef_vi_qword* ev = EF_GET_HW_EV_PTR(*ef_ev);
++ return ev->u32[0] & vi->vi_rxq.mask;
+}
++
+
-+static struct pci_dev *pcistub_device_get_pci_dev(struct pciback_device *pdev,
-+ struct pcistub_device *psdev)
-+{
-+ struct pci_dev *pci_dev = NULL;
-+ unsigned long flags;
++/*! \i_ef_vi Form a receive descriptor.
++**
++** If \c initial_rx_bytes is zero use a reception size at least as large
++** as an MTU.
++*/
++extern int ef_vi_receive_init(ef_vi* vi, ef_addr addr, ef_request_id dma_id,
++ int intial_rx_bytes);
+
-+ pcistub_device_get(psdev);
++/*! \i_ef_vi Submit initialised receive descriptors to the NIC. */
++extern void ef_vi_receive_push(ef_vi* vi);
+
-+ spin_lock_irqsave(&psdev->lock, flags);
-+ if (!psdev->pdev) {
-+ psdev->pdev = pdev;
-+ pci_dev = psdev->dev;
-+ }
-+ spin_unlock_irqrestore(&psdev->lock, flags);
++/*! \i_ef_vi Post a buffer on the receive queue.
++**
++** \return 0 on success, or -EAGAIN if the receive queue is full
++*/
++extern int ef_vi_receive_post(ef_vi*, ef_addr addr,
++ ef_request_id dma_id);
+
-+ if (!pci_dev)
-+ pcistub_device_put(psdev);
++/**********************************************************************
++ * Transmit interface *************************************************
++ **********************************************************************/
+
-+ return pci_dev;
++/*! \i_ef_vi Return the amount of space (in descriptors) in the transmit
++** queue.
++**
++** \return the amount of space in the queue (in descriptors)
++*/
++ef_vi_inline int ef_vi_transmit_space(ef_vi* vi)
++{
++ ef_vi_txq_state* qs = &vi->ep_state->txq;
++ return vi->vi_txq.mask - (qs->added - qs->removed);
+}
+
-+struct pci_dev *pcistub_get_pci_dev_by_slot(struct pciback_device *pdev,
-+ int domain, int bus,
-+ int slot, int func)
-+{
-+ struct pcistub_device *psdev;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
++/*! \i_ef_vi Returns the fill level of the TX descriptor ring.
++**
++** \return the fill level of the queue.
++*/
++ef_vi_inline int ef_vi_transmit_fill_level(ef_vi* vi)
++{
++ ef_vi_txq_state* qs = &vi->ep_state->txq;
++ return qs->added - qs->removed;
++}
+
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (psdev->dev != NULL
-+ && domain == pci_domain_nr(psdev->dev->bus)
-+ && bus == psdev->dev->bus->number
-+ && PCI_DEVFN(slot, func) == psdev->dev->devfn) {
-+ found_dev = pcistub_device_get_pci_dev(pdev, psdev);
-+ break;
-+ }
-+ }
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
-+ return found_dev;
++/*! \i_ef_vi Returns the total capacity of the TX descriptor ring.
++**
++** \return the capacity of the queue.
++*/
++ef_vi_inline int ef_vi_transmit_capacity(ef_vi* vi)
++{
++ return vi->vi_txq.mask;
+}
+
-+struct pci_dev *pcistub_get_pci_dev(struct pciback_device *pdev,
-+ struct pci_dev *dev)
-+{
-+ struct pcistub_device *psdev;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
++/*! \i_ef_vi Transmit a packet.
++**
++** \param bytes must be greater than ETH_ZLEN.
++** \return -EAGAIN if the transmit queue is full, or 0 on success
++*/
++extern int ef_vi_transmit(ef_vi*, ef_addr, int bytes, ef_request_id dma_id);
++
++/*! \i_ef_vi Transmit a packet using a gather list.
++**
++** \param iov_len must be greater than zero
++** \param iov the first must be non-zero in length (but others need not)
++**
++** \return -EAGAIN if the queue is full, or 0 on success
++*/
++extern int ef_vi_transmitv(ef_vi*, const ef_iovec* iov, int iov_len,
++ ef_request_id dma_id);
+
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (psdev->dev == dev) {
-+ found_dev = pcistub_device_get_pci_dev(pdev, psdev);
-+ break;
-+ }
-+ }
++/*! \i_ef_vi Initialise a DMA request.
++**
++** \return -EAGAIN if the queue is full, or 0 on success
++*/
++extern int ef_vi_transmit_init(ef_vi*, ef_addr, int bytes,
++ ef_request_id dma_id);
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
-+ return found_dev;
-+}
++/*! \i_ef_vi Initialise a DMA request.
++**
++** \return -EAGAIN if the queue is full, or 0 on success
++*/
++extern int ef_vi_transmitv_init(ef_vi*, const ef_iovec*, int iov_len,
++ ef_request_id dma_id);
+
-+void pcistub_put_pci_dev(struct pci_dev *dev)
-+{
-+ struct pcistub_device *psdev, *found_psdev = NULL;
-+ unsigned long flags;
++/*! \i_ef_vi Submit DMA requests to the NIC.
++**
++** The DMA requests must have been initialised using
++** ef_vi_transmit_init() or ef_vi_transmitv_init().
++*/
++extern void ef_vi_transmit_push(ef_vi*);
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
+
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (psdev->dev == dev) {
-+ found_psdev = psdev;
-+ break;
-+ }
-+ }
++/*! \i_ef_vi Maximum number of transmit completions per transmit event. */
++#define EF_VI_TRANSMIT_BATCH 64
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++/*! \i_ef_vi Determine the set of [ef_request_id]s for each DMA request
++** which has been completed by a given transmit completion
++** event.
++**
++** \param ids must point to an array of length EF_VI_TRANSMIT_BATCH
++** \return the number of valid [ef_request_id]s (can be zero)
++*/
++extern int ef_vi_transmit_unbundle(ef_vi* ep, const ef_event*,
++ ef_request_id* ids);
+
-+ /* Cleanup our device
-+ * (so it's ready for the next domain)
-+ */
-+ pciback_reset_device(found_psdev->dev);
-+ pciback_config_free_dyn_fields(found_psdev->dev);
-+ pciback_config_reset_dev(found_psdev->dev);
+
-+ spin_lock_irqsave(&found_psdev->lock, flags);
-+ found_psdev->pdev = NULL;
-+ spin_unlock_irqrestore(&found_psdev->lock, flags);
++/*! \i_ef_event Returns true if ef_eventq_poll() will return event(s). */
++extern int ef_eventq_has_event(ef_vi* vi);
+
-+ pcistub_device_put(found_psdev);
-+}
++/*! \i_ef_event Returns true if there are quite a few events in the event
++** queue.
++**
++** This looks ahead in the event queue, so has the property that it will
++** not ping-pong a cache-line when it is called concurrently with events
++** being delivered.
++*/
++extern int ef_eventq_has_many_events(ef_vi* evq, int look_ahead);
+
-+static int __devinit pcistub_match_one(struct pci_dev *dev,
-+ struct pcistub_device_id *pdev_id)
-+{
-+ /* Match the specified device by domain, bus, slot, func and also if
-+ * any of the device's parent bridges match.
-+ */
-+ for (; dev != NULL; dev = dev->bus->self) {
-+ if (pci_domain_nr(dev->bus) == pdev_id->domain
-+ && dev->bus->number == pdev_id->bus
-+ && dev->devfn == pdev_id->devfn)
-+ return 1;
++/*! Type of function to handle unknown events arriving on event queue
++** Return CI_TRUE iff the event has been handled.
++*/
++typedef int/*bool*/ ef_event_handler_fn(void* priv, ef_vi* evq, ef_event* ev);
+
-+ /* Sometimes topmost bridge links to itself. */
-+ if (dev == dev->bus->self)
-+ break;
-+ }
++/*! Standard poll exception routine */
++extern int/*bool*/ ef_eventq_poll_exception(void* priv, ef_vi* evq,
++ ef_event* ev);
+
-+ return 0;
-+}
++/*! \i_ef_event Retrieve events from the event queue, handle RX/TX events
++** and pass any others to an exception handler function
++**
++** \return The number of events retrieved.
++*/
++extern int ef_eventq_poll_evs(ef_vi* evq, ef_event* evs, int evs_len,
++ ef_event_handler_fn *exception, void *expt_priv);
+
-+static int __devinit pcistub_match(struct pci_dev *dev)
++/*! \i_ef_event Retrieve events from the event queue.
++**
++** \return The number of events retrieved.
++*/
++ef_vi_inline int ef_eventq_poll(ef_vi* evq, ef_event* evs, int evs_len)
+{
-+ struct pcistub_device_id *pdev_id;
-+ unsigned long flags;
-+ int found = 0;
-+
-+ spin_lock_irqsave(&device_ids_lock, flags);
-+ list_for_each_entry(pdev_id, &pcistub_device_ids, slot_list) {
-+ if (pcistub_match_one(dev, pdev_id)) {
-+ found = 1;
-+ break;
-+ }
-+ }
-+ spin_unlock_irqrestore(&device_ids_lock, flags);
-+
-+ return found;
++ return ef_eventq_poll_evs(evq, evs, evs_len,
++ &ef_eventq_poll_exception, (void*)0);
+}
+
-+static int __devinit pcistub_init_device(struct pci_dev *dev)
++/*! \i_ef_event Returns the capacity of an event queue. */
++ef_vi_inline int ef_eventq_capacity(ef_vi* vi)
+{
-+ struct pciback_dev_data *dev_data;
-+ int err = 0;
++ return (vi->evq_mask + 1u) / sizeof(ef_hw_event);
++}
+
-+ dev_dbg(&dev->dev, "initializing...\n");
++/* Returns the instance ID of [vi] */
++ef_vi_inline unsigned ef_vi_instance(ef_vi* vi)
++{ return vi->vi_i; }
+
-+ /* The PCI backend is not intended to be a module (or to work with
-+ * removable PCI devices (yet). If it were, pciback_config_free()
-+ * would need to be called somewhere to free the memory allocated
-+ * here and then to call kfree(pci_get_drvdata(psdev->dev)).
-+ */
-+ dev_data = kzalloc(sizeof(*dev_data), GFP_ATOMIC);
-+ if (!dev_data) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
-+ pci_set_drvdata(dev, dev_data);
+
-+ dev_dbg(&dev->dev, "initializing config\n");
-+ err = pciback_config_init_dev(dev);
-+ if (err)
-+ goto out;
++/**********************************************************************
++ * Initialisation *****************************************************
++ **********************************************************************/
+
-+ /* HACK: Force device (& ACPI) to determine what IRQ it's on - we
-+ * must do this here because pcibios_enable_device may specify
-+ * the pci device's true irq (and possibly its other resources)
-+ * if they differ from what's in the configuration space.
-+ * This makes the assumption that the device's resources won't
-+ * change after this point (otherwise this code may break!)
-+ */
-+ dev_dbg(&dev->dev, "enabling device\n");
-+ err = pci_enable_device(dev);
-+ if (err)
-+ goto config_release;
++/*! Return size of state buffer of an initialised VI. */
++extern int ef_vi_state_bytes(ef_vi*);
+
-+ /* Now disable the device (this also ensures some private device
-+ * data is setup before we export)
-+ */
-+ dev_dbg(&dev->dev, "reset device\n");
-+ pciback_reset_device(dev);
++/*! Return size of buffer needed for VI state given sizes of RX and TX
++** DMA queues. Queue sizes must be legal sizes (power of 2), or 0 (no
++** queue).
++*/
++extern int ef_vi_calc_state_bytes(int rxq_size, int txq_size);
+
-+ return 0;
++/*! Initialise [ef_vi] from the provided resources. [vvis] must have been
++** created by ef_make_vi_data() & remains owned by the caller.
++*/
++extern void ef_vi_init(ef_vi*, void* vi_info, ef_vi_state* state,
++ ef_eventq_state* evq_state, enum ef_vi_flags);
+
-+ config_release:
-+ pciback_config_free_dev(dev);
++extern void ef_vi_state_init(ef_vi*);
++extern void ef_eventq_state_init(ef_vi*);
+
-+ out:
-+ pci_set_drvdata(dev, NULL);
-+ kfree(dev_data);
-+ return err;
-+}
++/*! Convert an efhw device arch to ef_vi_arch, or returns -1 if not
++** recognised.
++*/
++extern int ef_vi_arch_from_efhw_arch(int efhw_arch);
++
++
++#endif /* __EFAB_EF_VI_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/falcon_event.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/falcon_event.c 2008-05-19 00:33:48.678952031 +0300
+@@ -0,0 +1,346 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
+/*
-+ * Because some initialization still happens on
-+ * devices during fs_initcall, we need to defer
-+ * full initialization of our devices until
-+ * device_initcall.
++ * \author djr
++ * \brief Routine to poll event queues.
++ * \date 2003/03/04
+ */
-+static int __init pcistub_init_devices_late(void)
-+{
-+ struct pcistub_device *psdev;
-+ unsigned long flags;
-+ int err = 0;
+
-+ pr_debug("pciback: pcistub_init_devices_late\n");
++/*! \cidoxg_lib_ef */
++#include "ef_vi_internal.h"
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
++/* Be worried about this on byteswapped machines */
++/* Due to crazy chipsets, we see the event words being written in
++** arbitrary order (bug4539). So test for presence of event must ensure
++** that both halves have changed from the null.
++*/
++# define EF_VI_IS_EVENT(evp) \
++ ( (((evp)->opaque.a != (uint32_t)-1) && \
++ ((evp)->opaque.b != (uint32_t)-1)) )
+
-+ while (!list_empty(&seized_devices)) {
-+ psdev = container_of(seized_devices.next,
-+ struct pcistub_device, dev_list);
-+ list_del(&psdev->dev_list);
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++#ifdef NDEBUG
++# define IS_DEBUG 0
++#else
++# define IS_DEBUG 1
++#endif
+
-+ err = pcistub_init_device(psdev->dev);
-+ if (err) {
-+ dev_err(&psdev->dev->dev,
-+ "error %d initializing device\n", err);
-+ kfree(psdev);
-+ psdev = NULL;
-+ }
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
++/*! Check for RX events with inconsistent SOP/CONT
++**
++** Returns true if this event should be discarded
++*/
++ef_vi_inline int ef_eventq_is_rx_sop_cont_bad_efab(ef_vi* vi,
++ const ef_vi_qword* ev)
++{
++ ef_rx_dup_state_t* rx_dup_state;
++ uint8_t* bad_sop;
+
-+ if (psdev)
-+ list_add_tail(&psdev->dev_list, &pcistub_devices);
++ unsigned label = QWORD_GET_U(RX_EV_Q_LABEL, *ev);
++ unsigned sop = QWORD_TEST_BIT(RX_SOP, *ev);
++
++ ef_assert(vi);
++ ef_assert_lt(label, EFAB_DMAQS_PER_EVQ_MAX);
++
++ rx_dup_state = &vi->evq_state->rx_dup_state[label];
++ bad_sop = &rx_dup_state->bad_sop;
++
++ if( ! ((vi->vi_flags & EF_VI_BUG5692_WORKAROUND) || IS_DEBUG) ) {
++ *bad_sop = (*bad_sop && !sop);
+ }
++ else {
++ unsigned cont = QWORD_TEST_BIT(RX_JUMBO_CONT, *ev);
++ uint8_t *frag_num = &rx_dup_state->frag_num;
+
-+ initialize_devices = 1;
++ /* bad_sop should latch till the next sop */
++ *bad_sop = (*bad_sop && !sop) || ( !!sop != (*frag_num==0) );
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
++ /* we do not check the number of bytes relative to the
++ * fragment number and size of the user rx buffer here
++ * because we don't know the size of the user rx
++ * buffer - we probably should perform this check in
++ * the nearest code calling this though.
++ */
++ *frag_num = cont ? (*frag_num + 1) : 0;
++ }
+
-+ return 0;
++ return *bad_sop;
+}
+
-+static int __devinit pcistub_seize(struct pci_dev *dev)
++
++ef_vi_inline int falcon_rx_check_dup(ef_vi* evq, ef_event* ev_out,
++ const ef_vi_qword* ev)
+{
-+ struct pcistub_device *psdev;
-+ unsigned long flags;
-+ int err = 0;
++ unsigned q_id = QWORD_GET_U(RX_EV_Q_LABEL, *ev);
++ unsigned desc_ptr = QWORD_GET_U(RX_EV_DESC_PTR, *ev);
++ ef_rx_dup_state_t* rx_dup_state = &evq->evq_state->rx_dup_state[q_id];
+
-+ psdev = pcistub_device_alloc(dev);
-+ if (!psdev)
-+ return -ENOMEM;
++ if(likely( desc_ptr != rx_dup_state->rx_last_desc_ptr )) {
++ rx_dup_state->rx_last_desc_ptr = desc_ptr;
++ return 0;
++ }
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
++ rx_dup_state->rx_last_desc_ptr = desc_ptr;
++ rx_dup_state->bad_sop = 1;
++#ifndef NDEBUG
++ rx_dup_state->frag_num = 0;
++#endif
++ BUG_ON(!QWORD_TEST_BIT(RX_EV_FRM_TRUNC, *ev));
++ BUG_ON( QWORD_TEST_BIT(RX_EV_PKT_OK, *ev));
++ BUG_ON(!QWORD_GET_U(RX_EV_BYTE_CNT, *ev) == 0);
++ ev_out->rx_no_desc_trunc.type = EF_EVENT_TYPE_RX_NO_DESC_TRUNC;
++ ev_out->rx_no_desc_trunc.q_id = q_id;
++ return 1;
++}
+
-+ if (initialize_devices) {
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
+
-+ /* don't want irqs disabled when calling pcistub_init_device */
-+ err = pcistub_init_device(psdev->dev);
++ef_vi_inline void falcon_rx_event(ef_event* ev_out, const ef_vi_qword* ev)
++{
++ if(likely( QWORD_TEST_BIT(RX_EV_PKT_OK, *ev) )) {
++ ev_out->rx.type = EF_EVENT_TYPE_RX;
++ ev_out->rx.q_id = QWORD_GET_U(RX_EV_Q_LABEL, *ev);
++ ev_out->rx.len = QWORD_GET_U(RX_EV_BYTE_CNT, *ev);
++ if( QWORD_TEST_BIT(RX_SOP, *ev) )
++ ev_out->rx.flags = EF_EVENT_FLAG_SOP;
++ else
++ ev_out->rx.flags = 0;
++ if( QWORD_TEST_BIT(RX_JUMBO_CONT, *ev) )
++ ev_out->rx.flags |= EF_EVENT_FLAG_CONT;
++ if( QWORD_TEST_BIT(RX_iSCSI_PKT_OK, *ev) )
++ ev_out->rx.flags |= EF_EVENT_FLAG_ISCSI_OK;
++ }
++ else {
++ ev_out->rx_discard.type = EF_EVENT_TYPE_RX_DISCARD;
++ ev_out->rx_discard.q_id = QWORD_GET_U(RX_EV_Q_LABEL, *ev);
++ ev_out->rx_discard.len = QWORD_GET_U(RX_EV_BYTE_CNT, *ev);
++#if 1 /* hack for ptloop compatability: ?? TODO purge */
++ if( QWORD_TEST_BIT(RX_SOP, *ev) )
++ ev_out->rx_discard.flags = EF_EVENT_FLAG_SOP;
++ else
++ ev_out->rx_discard.flags = 0;
++ if( QWORD_TEST_BIT(RX_JUMBO_CONT, *ev) )
++ ev_out->rx_discard.flags |= EF_EVENT_FLAG_CONT;
++ if( QWORD_TEST_BIT(RX_iSCSI_PKT_OK, *ev) )
++ ev_out->rx_discard.flags |= EF_EVENT_FLAG_ISCSI_OK;
++#endif
++ /* Order matters here: more fundamental errors first. */
++ if( QWORD_TEST_BIT(RX_EV_BUF_OWNER_ID_ERR, *ev) )
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_RIGHTS;
++ else if( QWORD_TEST_BIT(RX_EV_FRM_TRUNC, *ev) )
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_TRUNC;
++ else if( QWORD_TEST_BIT(RX_EV_ETH_CRC_ERR, *ev) )
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_CRC_BAD;
++ else if( QWORD_TEST_BIT(RX_EV_IP_HDR_CHKSUM_ERR, *ev) )
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_CSUM_BAD;
++ else if( QWORD_TEST_BIT(RX_EV_TCP_UDP_CHKSUM_ERR, *ev) )
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_CSUM_BAD;
++ else
++ ev_out->rx_discard.subtype =
++ EF_EVENT_RX_DISCARD_OTHER;
++ }
++}
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
+
-+ if (!err)
-+ list_add(&psdev->dev_list, &pcistub_devices);
-+ } else {
-+ dev_dbg(&dev->dev, "deferring initialization\n");
-+ list_add(&psdev->dev_list, &seized_devices);
++ef_vi_inline void falcon_tx_event(ef_event* ev_out, const ef_vi_qword* ev)
++{
++ /* Danger danger! No matter what we ask for wrt batching, we
++ ** will get a batched event every 16 descriptors, and we also
++ ** get dma-queue-empty events. i.e. Duplicates are expected.
++ **
++ ** In addition, if it's been requested in the descriptor, we
++ ** get an event per descriptor. (We don't currently request
++ ** this).
++ */
++ if(likely( QWORD_TEST_BIT(TX_EV_COMP, *ev) )) {
++ ev_out->tx.type = EF_EVENT_TYPE_TX;
++ ev_out->tx.q_id = QWORD_GET_U(TX_EV_Q_LABEL, *ev);
+ }
++ else {
++ ev_out->tx_error.type = EF_EVENT_TYPE_TX_ERROR;
++ ev_out->tx_error.q_id = QWORD_GET_U(TX_EV_Q_LABEL, *ev);
++ if(likely( QWORD_TEST_BIT(TX_EV_BUF_OWNER_ID_ERR, *ev) ))
++ ev_out->tx_error.subtype = EF_EVENT_TX_ERROR_RIGHTS;
++ else if(likely( QWORD_TEST_BIT(TX_EV_WQ_FF_FULL, *ev) ))
++ ev_out->tx_error.subtype = EF_EVENT_TX_ERROR_OFLOW;
++ else if(likely( QWORD_TEST_BIT(TX_EV_PKT_TOO_BIG, *ev) ))
++ ev_out->tx_error.subtype = EF_EVENT_TX_ERROR_2BIG;
++ else if(likely( QWORD_TEST_BIT(TX_EV_PKT_ERR, *ev) ))
++ ev_out->tx_error.subtype = EF_EVENT_TX_ERROR_BUS;
++ }
++}
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
-+
-+ if (err)
-+ pcistub_device_put(psdev);
+
-+ return err;
++static void mark_bad(ef_event* ev)
++{
++ ev->generic.ev.u64[0] &=~ ((uint64_t) 1u << RX_EV_PKT_OK_LBN);
+}
+
-+static int __devinit pcistub_probe(struct pci_dev *dev,
-+ const struct pci_device_id *id)
++
++int ef_eventq_poll_evs(ef_vi* evq, ef_event* evs, int evs_len,
++ ef_event_handler_fn *exception, void *expt_priv)
+{
-+ int err = 0;
++ int evs_len_orig = evs_len;
+
-+ dev_dbg(&dev->dev, "probing...\n");
++ EF_VI_CHECK_EVENT_Q(evq);
++ ef_assert(evs);
++ ef_assert_gt(evs_len, 0);
+
-+ if (pcistub_match(dev)) {
++ if(unlikely( EF_VI_IS_EVENT(EF_VI_EVENT_PTR(evq, 1)) ))
++ goto overflow;
+
-+ if (dev->hdr_type != PCI_HEADER_TYPE_NORMAL
-+ && dev->hdr_type != PCI_HEADER_TYPE_BRIDGE) {
-+ dev_err(&dev->dev, "can't export pci devices that "
-+ "don't have a normal (0) or bridge (1) "
-+ "header type!\n");
-+ err = -ENODEV;
-+ goto out;
++ do {
++ { /* Read the event out of the ring, then fiddle with
++ * copied version. Reason is that the ring is
++ * likely to get pushed out of cache by another
++ * event being delivered by hardware. */
++ ef_vi_event* ev = EF_VI_EVENT_PTR(evq, 0);
++ if( ! EF_VI_IS_EVENT(ev) )
++ break;
++ evs->generic.ev.u64[0] = cpu_to_le64 (ev->u64);
++ evq->evq_state->evq_ptr += sizeof(ef_vi_event);
++ ev->u64 = (uint64_t)(int64_t) -1;
++ }
++
++ /* Ugly: Exploit the fact that event code lies in top
++ * bits of event. */
++ ef_assert_ge(EV_CODE_LBN, 32u);
++ switch( evs->generic.ev.u32[1] >> (EV_CODE_LBN - 32u) ) {
++ case RX_IP_EV_DECODE:
++ /* Look for duplicate desc_ptr: it signals
++ * that a jumbo frame was truncated because we
++ * ran out of descriptors. */
++ if(unlikely( falcon_rx_check_dup
++ (evq, evs, &evs->generic.ev) )) {
++ --evs_len;
++ ++evs;
++ break;
++ }
++ else {
++ /* Cope with FalconA1 bugs where RX
++ * gives inconsistent RX events Mark
++ * events as bad until SOP becomes
++ * consistent again
++ * ef_eventq_is_rx_sop_cont_bad() has
++ * side effects - order is important
++ */
++ if(unlikely
++ (ef_eventq_is_rx_sop_cont_bad_efab
++ (evq, &evs->generic.ev) )) {
++ mark_bad(evs);
++ }
++ }
++ falcon_rx_event(evs, &evs->generic.ev);
++ --evs_len;
++ ++evs;
++ break;
++
++ case TX_IP_EV_DECODE:
++ falcon_tx_event(evs, &evs->generic.ev);
++ --evs_len;
++ ++evs;
++ break;
++
++ default:
++ break;
+ }
++ } while( evs_len );
+
-+ dev_info(&dev->dev, "seizing device\n");
-+ err = pcistub_seize(dev);
-+ } else
-+ /* Didn't find the device */
-+ err = -ENODEV;
++ return evs_len_orig - evs_len;
+
-+ out:
-+ return err;
++
++ overflow:
++ evs->generic.type = EF_EVENT_TYPE_OFLOW;
++ evs->generic.ev.u64[0] = (uint64_t)((int64_t)-1);
++ return 1;
+}
+
-+static void pcistub_remove(struct pci_dev *dev)
-+{
-+ struct pcistub_device *psdev, *found_psdev = NULL;
-+ unsigned long flags;
+
-+ dev_dbg(&dev->dev, "removing\n");
++int/*bool*/ ef_eventq_poll_exception(void* priv, ef_vi* evq, ef_event* ev)
++{
++ int /*bool*/ handled = 0;
++
++ switch( ev->generic.ev.u32[1] >> (EV_CODE_LBN - 32u) ) {
++ case DRIVER_EV_DECODE:
++ if( QWORD_GET_U(DRIVER_EV_SUB_CODE, ev->generic.ev) ==
++ EVQ_INIT_DONE_EV_DECODE )
++ /* EVQ initialised event: ignore. */
++ handled = 1;
++ break;
++ }
++ return handled;
++}
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
+
-+ pciback_config_quirk_release(dev);
++void ef_eventq_iterate(ef_vi* vi,
++ void (*fn)(void* arg, ef_vi*, int rel_pos,
++ int abs_pos, void* event),
++ void* arg, int stop_at_end)
++{
++ int i, size_evs = (vi->evq_mask + 1) / sizeof(ef_vi_event);
+
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (psdev->dev == dev) {
-+ found_psdev = psdev;
++ for( i = 0; i < size_evs; ++i ) {
++ ef_vi_event* e = EF_VI_EVENT_PTR(vi, -i);
++ if( EF_VI_IS_EVENT(e) )
++ fn(arg, vi, i,
++ EF_VI_EVENT_OFFSET(vi, -i) / sizeof(ef_vi_event),
++ e);
++ else if( stop_at_end )
+ break;
-+ }
+ }
++}
+
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
+
-+ if (found_psdev) {
-+ dev_dbg(&dev->dev, "found device to remove - in use? %p\n",
-+ found_psdev->pdev);
++int ef_eventq_has_event(ef_vi* vi)
++{
++ return EF_VI_IS_EVENT(EF_VI_EVENT_PTR(vi, 0));
++}
+
-+ if (found_psdev->pdev) {
-+ printk(KERN_WARNING "pciback: ****** removing device "
-+ "%s while still in-use! ******\n",
-+ pci_name(found_psdev->dev));
-+ printk(KERN_WARNING "pciback: ****** driver domain may "
-+ "still access this device's i/o resources!\n");
-+ printk(KERN_WARNING "pciback: ****** shutdown driver "
-+ "domain before binding device\n");
-+ printk(KERN_WARNING "pciback: ****** to other drivers "
-+ "or domains\n");
+
-+ pciback_release_pci_dev(found_psdev->pdev,
-+ found_psdev->dev);
-+ }
++int ef_eventq_has_many_events(ef_vi* vi, int look_ahead)
++{
++ ef_assert_ge(look_ahead, 0);
++ return EF_VI_IS_EVENT(EF_VI_EVENT_PTR(vi, -look_ahead));
++}
+
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
-+ list_del(&found_psdev->dev_list);
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
+
-+ /* the final put for releasing from the list */
-+ pcistub_device_put(found_psdev);
++int ef_eventq_has_rx_event(ef_vi* vi)
++{
++ ef_vi_event* ev;
++ int i, n_evs = 0;
++
++ for( i = 0; EF_VI_IS_EVENT(EF_VI_EVENT_PTR(vi, i)); --i ) {
++ ev = EF_VI_EVENT_PTR(vi, i);
++ if( EFVI_FALCON_EVENT_CODE(ev) == EF_EVENT_TYPE_RX ) n_evs++;
+ }
++ return n_evs;
+}
+
-+static struct pci_device_id pcistub_ids[] = {
-+ {
-+ .vendor = PCI_ANY_ID,
-+ .device = PCI_ANY_ID,
-+ .subvendor = PCI_ANY_ID,
-+ .subdevice = PCI_ANY_ID,
-+ },
-+ {0,},
-+};
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/falcon_vi.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/falcon_vi.c 2008-05-19 00:33:48.678952031 +0300
+@@ -0,0 +1,465 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
+/*
-+ * Note: There is no MODULE_DEVICE_TABLE entry here because this isn't
-+ * for a normal device. I don't want it to be loaded automatically.
++ * \author djr, stg
++ * \brief Falcon-specific VI
++ * \date 2006/11/30
+ */
+
-+static struct pci_driver pciback_pci_driver = {
-+ .name = "pciback",
-+ .id_table = pcistub_ids,
-+ .probe = pcistub_probe,
-+ .remove = pcistub_remove,
-+};
-+
-+static inline int str_to_slot(const char *buf, int *domain, int *bus,
-+ int *slot, int *func)
-+{
-+ int err;
-+
-+ err = sscanf(buf, " %x:%x:%x.%x", domain, bus, slot, func);
-+ if (err == 4)
-+ return 0;
-+ else if (err < 0)
-+ return -EINVAL;
++#include "ef_vi_internal.h"
+
-+ /* try again without domain */
-+ *domain = 0;
-+ err = sscanf(buf, " %x:%x.%x", bus, slot, func);
-+ if (err == 3)
-+ return 0;
+
-+ return -EINVAL;
-+}
++#define EFVI_FALCON_DMA_TX_FRAG 1
+
-+static inline int str_to_quirk(const char *buf, int *domain, int *bus, int
-+ *slot, int *func, int *reg, int *size, int *mask)
-+{
-+ int err;
+
-+ err =
-+ sscanf(buf, " %04x:%02x:%02x.%1x-%08x:%1x:%08x", domain, bus, slot,
-+ func, reg, size, mask);
-+ if (err == 7)
-+ return 0;
-+ return -EINVAL;
-+}
++/* TX descriptor for both physical and virtual packet transfers */
++typedef union {
++ uint32_t dword[2];
++} ef_vi_falcon_dma_tx_buf_desc;
++typedef ef_vi_falcon_dma_tx_buf_desc ef_vi_falcon_dma_tx_phys_desc;
+
-+static int pcistub_device_id_add(int domain, int bus, int slot, int func)
-+{
-+ struct pcistub_device_id *pci_dev_id;
-+ unsigned long flags;
+
-+ pci_dev_id = kmalloc(sizeof(*pci_dev_id), GFP_KERNEL);
-+ if (!pci_dev_id)
-+ return -ENOMEM;
++/* RX descriptor for physical addressed transfers */
++typedef union {
++ uint32_t dword[2];
++} ef_vi_falcon_dma_rx_phys_desc;
+
-+ pci_dev_id->domain = domain;
-+ pci_dev_id->bus = bus;
-+ pci_dev_id->devfn = PCI_DEVFN(slot, func);
+
-+ pr_debug("pciback: wants to seize %04x:%02x:%02x.%01x\n",
-+ domain, bus, slot, func);
++/* RX descriptor for virtual packet transfers */
++typedef struct {
++ uint32_t dword[1];
++} ef_vi_falcon_dma_rx_buf_desc;
+
-+ spin_lock_irqsave(&device_ids_lock, flags);
-+ list_add_tail(&pci_dev_id->slot_list, &pcistub_device_ids);
-+ spin_unlock_irqrestore(&device_ids_lock, flags);
++/* Buffer table index */
++typedef uint32_t ef_vi_buffer_addr_t;
+
-+ return 0;
++ef_vi_inline int64_t dma_addr_to_u46(int64_t src_dma_addr)
++{
++ return (src_dma_addr & __FALCON_MASK(46, int64_t));
+}
+
-+static int pcistub_device_id_remove(int domain, int bus, int slot, int func)
++/*! Setup a physical address based descriptor with a specified length */
++ef_vi_inline void
++__falcon_dma_rx_calc_ip_phys(ef_vi_dma_addr_t dest_pa,
++ ef_vi_falcon_dma_rx_phys_desc *desc,
++ int bytes)
+{
-+ struct pcistub_device_id *pci_dev_id, *t;
-+ int devfn = PCI_DEVFN(slot, func);
-+ int err = -ENOENT;
-+ unsigned long flags;
++ int region = 0; /* TODO fixme */
++ int64_t dest = dma_addr_to_u46(dest_pa); /* lower 46 bits */
+
-+ spin_lock_irqsave(&device_ids_lock, flags);
-+ list_for_each_entry_safe(pci_dev_id, t, &pcistub_device_ids, slot_list) {
++ DWCHCK(__DW2(RX_KER_BUF_SIZE_LBN), RX_KER_BUF_SIZE_WIDTH);
++ DWCHCK(__DW2(RX_KER_BUF_REGION_LBN),RX_KER_BUF_REGION_WIDTH);
+
-+ if (pci_dev_id->domain == domain
-+ && pci_dev_id->bus == bus && pci_dev_id->devfn == devfn) {
-+ /* Don't break; here because it's possible the same
-+ * slot could be in the list more than once
-+ */
-+ list_del(&pci_dev_id->slot_list);
-+ kfree(pci_dev_id);
++ LWCHK(RX_KER_BUF_ADR_LBN, RX_KER_BUF_ADR_WIDTH);
+
-+ err = 0;
++ RANGECHCK(bytes, RX_KER_BUF_SIZE_WIDTH);
++ RANGECHCK(region, RX_KER_BUF_REGION_WIDTH);
+
-+ pr_debug("pciback: removed %04x:%02x:%02x.%01x from "
-+ "seize list\n", domain, bus, slot, func);
-+ }
-+ }
-+ spin_unlock_irqrestore(&device_ids_lock, flags);
++ ef_assert(desc);
+
-+ return err;
++ desc->dword[1] = ((bytes << __DW2(RX_KER_BUF_SIZE_LBN)) |
++ (region << __DW2(RX_KER_BUF_REGION_LBN)) |
++ (HIGH(dest,
++ RX_KER_BUF_ADR_LBN,
++ RX_KER_BUF_ADR_WIDTH)));
++
++ desc->dword[0] = LOW(dest,
++ RX_KER_BUF_ADR_LBN,
++ RX_KER_BUF_ADR_WIDTH);
+}
+
-+static int pcistub_reg_add(int domain, int bus, int slot, int func, int reg,
-+ int size, int mask)
++/*! Setup a virtual buffer descriptor for an IPMODE transfer */
++ef_vi_inline void
++__falcon_dma_tx_calc_ip_buf(unsigned buf_id, unsigned buf_ofs, unsigned bytes,
++ int port, int frag,
++ ef_vi_falcon_dma_tx_buf_desc *desc)
+{
-+ int err = 0;
-+ struct pcistub_device *psdev;
-+ struct pci_dev *dev;
-+ struct config_field *field;
++ DWCHCK(__DW2(TX_USR_PORT_LBN), TX_USR_PORT_WIDTH);
++ DWCHCK(__DW2(TX_USR_CONT_LBN), TX_USR_CONT_WIDTH);
++ DWCHCK(__DW2(TX_USR_BYTE_CNT_LBN), TX_USR_BYTE_CNT_WIDTH);
++ LWCHK(RX_KER_BUF_ADR_LBN, RX_KER_BUF_ADR_WIDTH);
++ DWCHCK(TX_USR_BYTE_OFS_LBN, TX_USR_BYTE_OFS_WIDTH);
+
-+ psdev = pcistub_device_find(domain, bus, slot, func);
-+ if (!psdev || !psdev->dev) {
-+ err = -ENODEV;
-+ goto out;
-+ }
-+ dev = psdev->dev;
++ RANGECHCK(bytes, TX_USR_BYTE_CNT_WIDTH);
++ RANGECHCK(port, TX_USR_PORT_WIDTH);
++ RANGECHCK(frag, TX_USR_CONT_WIDTH);
++ RANGECHCK(buf_id, TX_USR_BUF_ID_WIDTH);
++ RANGECHCK(buf_ofs, TX_USR_BYTE_OFS_WIDTH);
+
-+ field = kzalloc(sizeof(*field), GFP_ATOMIC);
-+ if (!field) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++ ef_assert(desc);
+
-+ field->offset = reg;
-+ field->size = size;
-+ field->mask = mask;
-+ field->init = NULL;
-+ field->reset = NULL;
-+ field->release = NULL;
-+ field->clean = pciback_config_field_free;
++ desc->dword[1] = ((port << __DW2(TX_USR_PORT_LBN)) |
++ (frag << __DW2(TX_USR_CONT_LBN)) |
++ (bytes << __DW2(TX_USR_BYTE_CNT_LBN)) |
++ (HIGH(buf_id,
++ TX_USR_BUF_ID_LBN,
++ TX_USR_BUF_ID_WIDTH)));
+
-+ err = pciback_config_quirks_add_field(dev, field);
-+ if (err)
-+ kfree(field);
-+ out:
-+ return err;
++ desc->dword[0] = ((LOW(buf_id,
++ TX_USR_BUF_ID_LBN,
++ (TX_USR_BUF_ID_WIDTH))) |
++ (buf_ofs << TX_USR_BYTE_OFS_LBN));
+}
+
-+static ssize_t pcistub_slot_add(struct device_driver *drv, const char *buf,
-+ size_t count)
++ef_vi_inline void
++falcon_dma_tx_calc_ip_buf_4k(unsigned buf_vaddr, unsigned bytes,
++ int port, int frag,
++ ef_vi_falcon_dma_tx_buf_desc *desc)
+{
-+ int domain, bus, slot, func;
-+ int err;
-+
-+ err = str_to_slot(buf, &domain, &bus, &slot, &func);
-+ if (err)
-+ goto out;
-+
-+ err = pcistub_device_id_add(domain, bus, slot, func);
++ /* TODO FIXME [buf_vaddr] consists of the buffer index in the
++ ** high bits, and an offset in the low bits. Assumptions
++ ** permate the code that these can be rolled into one 32bit
++ ** value, so this is currently preserved for Falcon. But we
++ ** should change to support 8K pages
++ */
++ unsigned buf_id = EFVI_FALCON_BUFFER_4K_PAGE(buf_vaddr);
++ unsigned buf_ofs = EFVI_FALCON_BUFFER_4K_OFF(buf_vaddr);
+
-+ out:
-+ if (!err)
-+ err = count;
-+ return err;
++ __falcon_dma_tx_calc_ip_buf( buf_id, buf_ofs, bytes, port, frag, desc);
+}
+
-+DRIVER_ATTR(new_slot, S_IWUSR, NULL, pcistub_slot_add);
-+
-+static ssize_t pcistub_slot_remove(struct device_driver *drv, const char *buf,
-+ size_t count)
++ef_vi_inline void
++falcon_dma_tx_calc_ip_buf(unsigned buf_vaddr, unsigned bytes, int port,
++ int frag, ef_vi_falcon_dma_tx_buf_desc *desc)
+{
-+ int domain, bus, slot, func;
-+ int err;
-+
-+ err = str_to_slot(buf, &domain, &bus, &slot, &func);
-+ if (err)
-+ goto out;
-+
-+ err = pcistub_device_id_remove(domain, bus, slot, func);
-+
-+ out:
-+ if (!err)
-+ err = count;
-+ return err;
++ falcon_dma_tx_calc_ip_buf_4k(buf_vaddr, bytes, port, frag, desc);
+}
+
-+DRIVER_ATTR(remove_slot, S_IWUSR, NULL, pcistub_slot_remove);
++/*! Setup a virtual buffer based descriptor */
++ef_vi_inline void
++__falcon_dma_rx_calc_ip_buf(unsigned buf_id, unsigned buf_ofs,
++ ef_vi_falcon_dma_rx_buf_desc *desc)
++{
++ /* check alignment of buffer offset and pack */
++ ef_assert((buf_ofs & 0x1) == 0);
+
-+static ssize_t pcistub_slot_show(struct device_driver *drv, char *buf)
-+{
-+ struct pcistub_device_id *pci_dev_id;
-+ size_t count = 0;
-+ unsigned long flags;
++ buf_ofs >>= 1;
+
-+ spin_lock_irqsave(&device_ids_lock, flags);
-+ list_for_each_entry(pci_dev_id, &pcistub_device_ids, slot_list) {
-+ if (count >= PAGE_SIZE)
-+ break;
++ DWCHCK(RX_USR_2BYTE_OFS_LBN, RX_USR_2BYTE_OFS_WIDTH);
++ DWCHCK(RX_USR_BUF_ID_LBN, RX_USR_BUF_ID_WIDTH);
+
-+ count += scnprintf(buf + count, PAGE_SIZE - count,
-+ "%04x:%02x:%02x.%01x\n",
-+ pci_dev_id->domain, pci_dev_id->bus,
-+ PCI_SLOT(pci_dev_id->devfn),
-+ PCI_FUNC(pci_dev_id->devfn));
-+ }
-+ spin_unlock_irqrestore(&device_ids_lock, flags);
++ RANGECHCK(buf_ofs, RX_USR_2BYTE_OFS_WIDTH);
++ RANGECHCK(buf_id, RX_USR_BUF_ID_WIDTH);
+
-+ return count;
++ ef_assert(desc);
++
++ desc->dword[0] = ((buf_ofs << RX_USR_2BYTE_OFS_LBN) |
++ (buf_id << RX_USR_BUF_ID_LBN));
+}
+
-+DRIVER_ATTR(slots, S_IRUSR, pcistub_slot_show, NULL);
++ef_vi_inline void
++falcon_dma_rx_calc_ip_buf_4k(unsigned buf_vaddr,
++ ef_vi_falcon_dma_rx_buf_desc *desc)
++{
++ /* TODO FIXME [buf_vaddr] consists of the buffer index in the
++ ** high bits, and an offset in the low bits. Assumptions
++ ** permeate the code that these can be rolled into one 32bit
++ ** value, so this is currently preserved for Falcon. But we
++ ** should change to support 8K pages
++ */
++ unsigned buf_id = EFVI_FALCON_BUFFER_4K_PAGE(buf_vaddr);
++ unsigned buf_ofs = EFVI_FALCON_BUFFER_4K_OFF(buf_vaddr);
+
-+static ssize_t pcistub_quirk_add(struct device_driver *drv, const char *buf,
-+ size_t count)
-+{
-+ int domain, bus, slot, func, reg, size, mask;
-+ int err;
++ __falcon_dma_rx_calc_ip_buf(buf_id, buf_ofs, desc);
++}
+
-+ err = str_to_quirk(buf, &domain, &bus, &slot, &func, ®, &size,
-+ &mask);
-+ if (err)
-+ goto out;
++ef_vi_inline void
++falcon_dma_rx_calc_ip_buf(unsigned buf_vaddr,
++ ef_vi_falcon_dma_rx_buf_desc *desc)
++{
++ falcon_dma_rx_calc_ip_buf_4k(buf_vaddr, desc);
++}
+
-+ err = pcistub_reg_add(domain, bus, slot, func, reg, size, mask);
+
-+ out:
-+ if (!err)
-+ err = count;
-+ return err;
++ef_vi_inline ef_vi_dma_addr_t ef_physaddr(ef_addr efaddr)
++{
++ return (ef_vi_dma_addr_t) efaddr;
+}
+
-+static ssize_t pcistub_quirk_show(struct device_driver *drv, char *buf)
-+{
-+ int count = 0;
-+ unsigned long flags;
-+ extern struct list_head pciback_quirks;
-+ struct pciback_config_quirk *quirk;
-+ struct pciback_dev_data *dev_data;
-+ struct config_field *field;
-+ struct config_field_entry *cfg_entry;
+
-+ spin_lock_irqsave(&device_ids_lock, flags);
-+ list_for_each_entry(quirk, &pciback_quirks, quirks_list) {
-+ if (count >= PAGE_SIZE)
-+ goto out;
++/*! Convert between an ef_addr and a buffer table index
++** Assert that this was not a physical address
++*/
++ef_vi_inline ef_vi_buffer_addr_t ef_bufaddr(ef_addr efaddr)
++{
++ ef_assert(efaddr < ((uint64_t)1 << 32) );
+
-+ count += scnprintf(buf + count, PAGE_SIZE - count,
-+ "%02x:%02x.%01x\n\t%04x:%04x:%04x:%04x\n",
-+ quirk->pdev->bus->number,
-+ PCI_SLOT(quirk->pdev->devfn),
-+ PCI_FUNC(quirk->pdev->devfn),
-+ quirk->devid.vendor, quirk->devid.device,
-+ quirk->devid.subvendor,
-+ quirk->devid.subdevice);
++ return (ef_vi_buffer_addr_t) efaddr;
++}
+
-+ dev_data = pci_get_drvdata(quirk->pdev);
+
-+ list_for_each_entry(cfg_entry, &dev_data->config_fields, list) {
-+ field = cfg_entry->field;
-+ if (count >= PAGE_SIZE)
-+ goto out;
++/*! Setup an physical address based descriptor for an IPMODE transfer */
++ef_vi_inline void
++falcon_dma_tx_calc_ip_phys(ef_vi_dma_addr_t src_dma_addr, unsigned bytes,
++ int port, int frag,
++ ef_vi_falcon_dma_tx_phys_desc *desc)
++{
+
-+ count += scnprintf(buf + count, PAGE_SIZE - count,
-+ "\t\t%08x:%01x:%08x\n",
-+ cfg_entry->base_offset + field->offset,
-+ field->size, field->mask);
-+ }
-+ }
++ int region = 0; /* FIXME */
++ int64_t src = dma_addr_to_u46(src_dma_addr); /* lower 46 bits */
+
-+ out:
-+ spin_unlock_irqrestore(&device_ids_lock, flags);
++ DWCHCK(__DW2(TX_KER_PORT_LBN), TX_KER_PORT_WIDTH);
++ DWCHCK(__DW2(TX_KER_CONT_LBN), TX_KER_CONT_WIDTH);
++ DWCHCK(__DW2(TX_KER_BYTE_CNT_LBN), TX_KER_BYTE_CNT_WIDTH);
++ DWCHCK(__DW2(TX_KER_BUF_REGION_LBN),TX_KER_BUF_REGION_WIDTH);
+
-+ return count;
-+}
++ LWCHK(TX_KER_BUF_ADR_LBN, TX_KER_BUF_ADR_WIDTH);
+
-+DRIVER_ATTR(quirks, S_IRUSR | S_IWUSR, pcistub_quirk_show, pcistub_quirk_add);
++ RANGECHCK(port, TX_KER_PORT_WIDTH);
++ RANGECHCK(frag, TX_KER_CONT_WIDTH);
++ RANGECHCK(bytes, TX_KER_BYTE_CNT_WIDTH);
++ RANGECHCK(region, TX_KER_BUF_REGION_WIDTH);
+
-+static ssize_t permissive_add(struct device_driver *drv, const char *buf,
-+ size_t count)
-+{
-+ int domain, bus, slot, func;
-+ int err;
-+ struct pcistub_device *psdev;
-+ struct pciback_dev_data *dev_data;
-+ err = str_to_slot(buf, &domain, &bus, &slot, &func);
-+ if (err)
-+ goto out;
-+ psdev = pcistub_device_find(domain, bus, slot, func);
-+ if (!psdev) {
-+ err = -ENODEV;
-+ goto out;
-+ }
-+ if (!psdev->dev) {
-+ err = -ENODEV;
-+ goto release;
-+ }
-+ dev_data = pci_get_drvdata(psdev->dev);
-+ /* the driver data for a device should never be null at this point */
-+ if (!dev_data) {
-+ err = -ENXIO;
-+ goto release;
-+ }
-+ if (!dev_data->permissive) {
-+ dev_data->permissive = 1;
-+ /* Let user know that what they're doing could be unsafe */
-+ dev_warn(&psdev->dev->dev,
-+ "enabling permissive mode configuration space accesses!\n");
-+ dev_warn(&psdev->dev->dev,
-+ "permissive mode is potentially unsafe!\n");
-+ }
-+ release:
-+ pcistub_device_put(psdev);
-+ out:
-+ if (!err)
-+ err = count;
-+ return err;
-+}
++ desc->dword[1] = ((port << __DW2(TX_KER_PORT_LBN)) |
++ (frag << __DW2(TX_KER_CONT_LBN)) |
++ (bytes << __DW2(TX_KER_BYTE_CNT_LBN)) |
++ (region << __DW2(TX_KER_BUF_REGION_LBN)) |
++ (HIGH(src,
++ TX_KER_BUF_ADR_LBN,
++ TX_KER_BUF_ADR_WIDTH)));
+
-+static ssize_t permissive_show(struct device_driver *drv, char *buf)
-+{
-+ struct pcistub_device *psdev;
-+ struct pciback_dev_data *dev_data;
-+ size_t count = 0;
-+ unsigned long flags;
-+ spin_lock_irqsave(&pcistub_devices_lock, flags);
-+ list_for_each_entry(psdev, &pcistub_devices, dev_list) {
-+ if (count >= PAGE_SIZE)
-+ break;
-+ if (!psdev->dev)
-+ continue;
-+ dev_data = pci_get_drvdata(psdev->dev);
-+ if (!dev_data || !dev_data->permissive)
-+ continue;
-+ count +=
-+ scnprintf(buf + count, PAGE_SIZE - count, "%s\n",
-+ pci_name(psdev->dev));
-+ }
-+ spin_unlock_irqrestore(&pcistub_devices_lock, flags);
-+ return count;
++ ef_assert_equal(TX_KER_BUF_ADR_LBN, 0);
++ desc->dword[0] = (uint32_t) src_dma_addr;
+}
+
-+DRIVER_ATTR(permissive, S_IRUSR | S_IWUSR, permissive_show, permissive_add);
+
-+static void pcistub_exit(void)
++void falcon_vi_init(ef_vi* vi, void* vvis)
+{
-+ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_new_slot);
-+ driver_remove_file(&pciback_pci_driver.driver,
-+ &driver_attr_remove_slot);
-+ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_slots);
-+ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_quirks);
-+ driver_remove_file(&pciback_pci_driver.driver, &driver_attr_permissive);
++ struct vi_mappings *vm = (struct vi_mappings*)vvis;
++ uint16_t* ids;
+
-+ pci_unregister_driver(&pciback_pci_driver);
-+}
++ ef_assert(vi);
++ ef_assert(vvis);
++ ef_assert_equal(vm->signature, VI_MAPPING_SIGNATURE);
++ ef_assert_equal(vm->nic_type.arch, EF_VI_ARCH_FALCON);
+
-+static int __init pcistub_init(void)
-+{
-+ int pos = 0;
-+ int err = 0;
-+ int domain, bus, slot, func;
-+ int parsed;
++ /* Initialise masks to zero, so that ef_vi_state_init() will
++ ** not do any harm when we don't have DMA queues. */
++ vi->vi_rxq.mask = vi->vi_txq.mask = 0;
+
-+ if (pci_devs_to_hide && *pci_devs_to_hide) {
-+ do {
-+ parsed = 0;
++ /* Used for BUG5391_WORKAROUND. */
++ vi->vi_txq.misalign_mask = 0;
+
-+ err = sscanf(pci_devs_to_hide + pos,
-+ " (%x:%x:%x.%x) %n",
-+ &domain, &bus, &slot, &func, &parsed);
-+ if (err != 4) {
-+ domain = 0;
-+ err = sscanf(pci_devs_to_hide + pos,
-+ " (%x:%x.%x) %n",
-+ &bus, &slot, &func, &parsed);
-+ if (err != 3)
-+ goto parse_error;
-+ }
++ /* Initialise doorbell addresses to a distinctive small value
++ ** which will cause a segfault, to trap doorbell pushes to VIs
++ ** without DMA queues. */
++ vi->vi_rxq.doorbell = vi->vi_txq.doorbell = (ef_vi_ioaddr_t)0xdb;
+
-+ err = pcistub_device_id_add(domain, bus, slot, func);
-+ if (err)
-+ goto out;
++ ids = (uint16_t*) (vi->ep_state + 1);
+
-+ /* if parsed<=0, we've reached the end of the string */
-+ pos += parsed;
-+ } while (parsed > 0 && pci_devs_to_hide[pos]);
++ if( vm->tx_queue_capacity ) {
++ vi->vi_txq.mask = vm->tx_queue_capacity - 1;
++ vi->vi_txq.doorbell = vm->tx_bell + 12;
++ vi->vi_txq.descriptors = vm->tx_dma_falcon;
++ vi->vi_txq.ids = ids;
++ ids += vi->vi_txq.mask + 1;
++ /* Check that the id fifo fits in the space allocated. */
++ ef_assert_le((char*) (vi->vi_txq.ids + vm->tx_queue_capacity),
++ (char*) vi->ep_state
++ + ef_vi_calc_state_bytes(vm->rx_queue_capacity,
++ vm->tx_queue_capacity));
++ }
++ if( vm->rx_queue_capacity ) {
++ vi->vi_rxq.mask = vm->rx_queue_capacity - 1;
++ vi->vi_rxq.doorbell = vm->rx_bell + 12;
++ vi->vi_rxq.descriptors = vm->rx_dma_falcon;
++ vi->vi_rxq.ids = ids;
++ /* Check that the id fifo fits in the space allocated. */
++ ef_assert_le((char*) (vi->vi_rxq.ids + vm->rx_queue_capacity),
++ (char*) vi->ep_state
++ + ef_vi_calc_state_bytes(vm->rx_queue_capacity,
++ vm->tx_queue_capacity));
+ }
+
-+ /* If we're the first PCI Device Driver to register, we're the
-+ * first one to get offered PCI devices as they become
-+ * available (and thus we can be the first to grab them)
-+ */
-+ err = pci_register_driver(&pciback_pci_driver);
-+ if (err < 0)
-+ goto out;
-+
-+ err = driver_create_file(&pciback_pci_driver.driver,
-+ &driver_attr_new_slot);
-+ if (!err)
-+ err = driver_create_file(&pciback_pci_driver.driver,
-+ &driver_attr_remove_slot);
-+ if (!err)
-+ err = driver_create_file(&pciback_pci_driver.driver,
-+ &driver_attr_slots);
-+ if (!err)
-+ err = driver_create_file(&pciback_pci_driver.driver,
-+ &driver_attr_quirks);
-+ if (!err)
-+ err = driver_create_file(&pciback_pci_driver.driver,
-+ &driver_attr_permissive);
-+
-+ if (err)
-+ pcistub_exit();
-+
-+ out:
-+ return err;
-+
-+ parse_error:
-+ printk(KERN_ERR "pciback: Error parsing pci_devs_to_hide at \"%s\"\n",
-+ pci_devs_to_hide + pos);
-+ return -EINVAL;
++ if( vm->nic_type.variant == 'A' ) {
++ vi->vi_txq.misalign_mask = 15; /* BUG5391_WORKAROUND */
++ vi->vi_flags |= EF_VI_BUG5692_WORKAROUND;
++ }
+}
+
-+#ifndef MODULE
-+/*
-+ * fs_initcall happens before device_initcall
-+ * so pciback *should* get called first (b/c we
-+ * want to suck up any device before other drivers
-+ * get a chance by being the first pci device
-+ * driver to register)
-+ */
-+fs_initcall(pcistub_init);
-+#endif
+
-+static int __init pciback_init(void)
++int ef_vi_transmitv_init(ef_vi* vi, const ef_iovec* iov, int iov_len,
++ ef_request_id dma_id)
+{
-+ int err;
-+
-+ err = pciback_config_init();
-+ if (err)
-+ return err;
++ ef_vi_txq* q = &vi->vi_txq;
++ ef_vi_txq_state* qs = &vi->ep_state->txq;
++ ef_vi_falcon_dma_tx_buf_desc* dp;
++ unsigned len, dma_len, di;
++ unsigned added_save = qs->added;
++ ef_addr dma_addr;
++ unsigned last_len = 0;
+
-+#ifdef MODULE
-+ err = pcistub_init();
-+ if (err < 0)
-+ return err;
-+#endif
++ ef_assert(iov_len > 0);
++ ef_assert(iov);
++ ef_assert_equal((dma_id & EF_REQUEST_ID_MASK), dma_id);
++ ef_assert_nequal(dma_id, 0xffff);
+
-+ pcistub_init_devices_late();
-+ err = pciback_xenbus_register();
-+ if (err)
-+ pcistub_exit();
++ dma_addr = iov->iov_base;
++ len = iov->iov_len;
+
-+ return err;
-+}
++ if( vi->vi_flags & EF_VI_ISCSI_TX_DDIG ) {
++ /* Last 4 bytes of placeholder for digest must be
++ * removed for h/w */
++ ef_assert(len > 4);
++ last_len = iov[iov_len - 1].iov_len;
++ if( last_len <= 4 ) {
++ ef_assert(iov_len > 1);
++ --iov_len;
++ last_len = iov[iov_len - 1].iov_len - (4 - last_len);
++ }
++ else {
++ last_len = iov[iov_len - 1].iov_len - 4;
++ }
++ if( iov_len == 1 )
++ len = last_len;
++ }
+
-+static void __exit pciback_cleanup(void)
-+{
-+ pciback_xenbus_unregister();
-+ pcistub_exit();
-+}
++ while( 1 ) {
++ if( qs->added - qs->removed >= q->mask ) {
++ qs->added = added_save;
++ return -EAGAIN;
++ }
+
-+module_init(pciback_init);
-+module_exit(pciback_cleanup);
++ dma_len = (~((unsigned) dma_addr) & 0xfff) + 1;
++ if( dma_len > len ) dma_len = len;
++ { /* BUG5391_WORKAROUND */
++ unsigned misalign =
++ (unsigned) dma_addr & q->misalign_mask;
++ if( misalign && dma_len + misalign > 512 )
++ dma_len = 512 - misalign;
++ }
+
-+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/slot.c linux-2.6.18-xen.hg/drivers/xen/pciback/slot.c
---- linux-2.6.18/drivers/xen/pciback/slot.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/slot.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,151 @@
-+/*
-+ * PCI Backend - Provides a Virtual PCI bus (with real devices)
-+ * to the frontend
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil> (vpci.c)
-+ * Author: Tristan Gingold <tristan.gingold@bull.net>, from vpci.c
-+ */
++ di = qs->added++ & q->mask;
++ dp = (ef_vi_falcon_dma_tx_buf_desc*) q->descriptors + di;
++ if( vi->vi_flags & EF_VI_TX_PHYS_ADDR )
++ falcon_dma_tx_calc_ip_phys
++ (ef_physaddr(dma_addr), dma_len, /*port*/ 0,
++ (iov_len == 1 && dma_len == len) ? 0 :
++ EFVI_FALCON_DMA_TX_FRAG, dp);
++ else
++ falcon_dma_tx_calc_ip_buf
++ (ef_bufaddr(dma_addr), dma_len, /*port*/ 0,
++ (iov_len == 1 && dma_len == len) ? 0 :
++ EFVI_FALCON_DMA_TX_FRAG, dp);
+
-+#include <linux/list.h>
-+#include <linux/slab.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include "pciback.h"
++ dma_addr += dma_len;
++ len -= dma_len;
+
-+/* There are at most 32 slots in a pci bus. */
-+#define PCI_SLOT_MAX 32
++ if( len == 0 ) {
++ if( --iov_len == 0 ) break;
++ ++iov;
++ dma_addr = iov->iov_base;
++ len = iov->iov_len;
++ if( (vi->vi_flags & EF_VI_ISCSI_TX_DDIG) &&
++ (iov_len == 1) )
++ len = last_len;
++ }
++ }
+
-+#define PCI_BUS_NBR 2
++ q->ids[di] = (uint16_t) dma_id;
++ return 0;
++}
+
-+struct slot_dev_data {
-+ /* Access to dev_list must be protected by lock */
-+ struct pci_dev *slots[PCI_BUS_NBR][PCI_SLOT_MAX];
-+ spinlock_t lock;
-+};
+
-+struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus,
-+ unsigned int devfn)
++void ef_vi_transmit_push(ef_vi* vi)
+{
-+ struct pci_dev *dev = NULL;
-+ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
-+ unsigned long flags;
-+
-+ if (domain != 0 || PCI_FUNC(devfn) != 0)
-+ return NULL;
++ ef_vi_wiob();
++ writel((vi->ep_state->txq.added & vi->vi_txq.mask) <<
++ __DW4(TX_DESC_WPTR_LBN),
++ vi->vi_txq.doorbell);
++}
+
-+ if (PCI_SLOT(devfn) >= PCI_SLOT_MAX || bus >= PCI_BUS_NBR)
-+ return NULL;
+
-+ spin_lock_irqsave(&slot_dev->lock, flags);
-+ dev = slot_dev->slots[bus][PCI_SLOT(devfn)];
-+ spin_unlock_irqrestore(&slot_dev->lock, flags);
++/*! The value of initial_rx_bytes is used to set RX_KER_BUF_SIZE in an initial
++** receive descriptor here if physical addressing is being used. A value of
++** zero represents 16384 bytes. This is okay, because caller must provide a
++** buffer than is > MTU, and mac should filter anything bigger than that.
++*/
++int ef_vi_receive_init(ef_vi* vi, ef_addr addr, ef_request_id dma_id,
++ int initial_rx_bytes)
++{
++ ef_vi_rxq* q = &vi->vi_rxq;
++ ef_vi_rxq_state* qs = &vi->ep_state->rxq;
++ unsigned di;
+
-+ return dev;
-+}
++ if( ef_vi_receive_space(vi) ) {
++ di = qs->added++ & q->mask;
++ ef_assert_equal(q->ids[di], 0xffff);
++ q->ids[di] = (uint16_t) dma_id;
+
-+int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ int err = 0, slot, bus;
-+ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
-+ unsigned long flags;
++ if( ! (vi->vi_flags & EF_VI_RX_PHYS_ADDR) ) {
++ ef_vi_falcon_dma_rx_buf_desc* dp;
++ dp = (ef_vi_falcon_dma_rx_buf_desc*)
++ q->descriptors + di;
++ falcon_dma_rx_calc_ip_buf(ef_bufaddr(addr), dp);
++ }
++ else {
++ ef_vi_falcon_dma_rx_phys_desc* dp;
++ dp = (ef_vi_falcon_dma_rx_phys_desc*)
++ q->descriptors + di;
++ __falcon_dma_rx_calc_ip_phys(addr, dp,
++ initial_rx_bytes);
++ }
+
-+ if ((dev->class >> 24) == PCI_BASE_CLASS_BRIDGE) {
-+ err = -EFAULT;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Can't export bridges on the virtual PCI bus");
-+ goto out;
++ return 0;
+ }
+
-+ spin_lock_irqsave(&slot_dev->lock, flags);
-+
-+ /* Assign to a new slot on the virtual PCI bus */
-+ for (bus = 0; bus < PCI_BUS_NBR; bus++)
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ if (slot_dev->slots[bus][slot] == NULL) {
-+ printk(KERN_INFO
-+ "pciback: slot: %s: assign to virtual slot %d, bus %d\n",
-+ pci_name(dev), slot, bus);
-+ slot_dev->slots[bus][slot] = dev;
-+ goto unlock;
-+ }
-+ }
++ return -EAGAIN;
++}
+
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "No more space on root virtual PCI bus");
+
-+ unlock:
-+ spin_unlock_irqrestore(&slot_dev->lock, flags);
-+ out:
-+ return err;
++void ef_vi_receive_push(ef_vi* vi)
++{
++ ef_vi_wiob();
++ writel ((vi->ep_state->rxq.added & vi->vi_rxq.mask) <<
++ __DW4(RX_DESC_WPTR_LBN),
++ vi->vi_rxq.doorbell);
+}
+
-+void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
++
++ef_request_id ef_vi_receive_done(const ef_vi* vi, const ef_event* ef_ev)
+{
-+ int slot, bus;
-+ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
++ const ef_vi_qword* ev = EF_GET_HW_EV_PTR(*ef_ev);
++ unsigned di = ev->u32[0] & vi->vi_rxq.mask;
++ ef_request_id rq_id;
+
-+ spin_lock_irqsave(&slot_dev->lock, flags);
++ ef_assert(EF_EVENT_TYPE(*ef_ev) == EF_EVENT_TYPE_RX ||
++ EF_EVENT_TYPE(*ef_ev) == EF_EVENT_TYPE_RX_DISCARD);
+
-+ for (bus = 0; bus < PCI_BUS_NBR; bus++)
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ if (slot_dev->slots[bus][slot] == dev) {
-+ slot_dev->slots[bus][slot] = NULL;
-+ found_dev = dev;
-+ goto out;
-+ }
-+ }
++ /* Detect spurious / duplicate RX events. We may need to modify this
++ ** code so that we are robust if they happen. */
++ ef_assert_equal(di, vi->ep_state->rxq.removed & vi->vi_rxq.mask);
+
-+ out:
-+ spin_unlock_irqrestore(&slot_dev->lock, flags);
++ /* We only support 1 port: so events should be in order. */
++ ef_assert(vi->vi_rxq.ids[di] != 0xffff);
+
-+ if (found_dev)
-+ pcistub_put_pci_dev(found_dev);
++ rq_id = vi->vi_rxq.ids[di];
++ vi->vi_rxq.ids[di] = 0xffff;
++ ++vi->ep_state->rxq.removed;
++ return rq_id;
+}
+
-+int pciback_init_devices(struct pciback_device *pdev)
-+{
-+ int slot, bus;
-+ struct slot_dev_data *slot_dev;
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/pt_tx.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/pt_tx.c 2008-05-19 00:33:48.678952031 +0300
+@@ -0,0 +1,91 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ slot_dev = kmalloc(sizeof(*slot_dev), GFP_KERNEL);
-+ if (!slot_dev)
-+ return -ENOMEM;
++/*
++ * \author djr
++ * \brief Packet-mode transmit interface.
++ * \date 2003/04/02
++ */
++
++/*! \cidoxg_lib_ef */
++#include "ef_vi_internal.h"
+
-+ spin_lock_init(&slot_dev->lock);
+
-+ for (bus = 0; bus < PCI_BUS_NBR; bus++)
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++)
-+ slot_dev->slots[bus][slot] = NULL;
++int ef_vi_transmit_init(ef_vi* vi, ef_addr base, int len, ef_request_id dma_id)
++{
++ ef_iovec iov = { base, len };
++ return ef_vi_transmitv_init(vi, &iov, 1, dma_id);
++}
+
-+ pdev->pci_dev_data = slot_dev;
+
-+ return 0;
++int ef_vi_transmit(ef_vi* vi, ef_addr base, int len, ef_request_id dma_id)
++{
++ ef_iovec iov = { base, len };
++ int rc = ef_vi_transmitv_init(vi, &iov, 1, dma_id);
++ if( rc == 0 ) ef_vi_transmit_push(vi);
++ return rc;
+}
+
-+int pciback_publish_pci_roots(struct pciback_device *pdev,
-+ publish_pci_root_cb publish_cb)
++
++int ef_vi_transmitv(ef_vi* vi, const ef_iovec* iov, int iov_len,
++ ef_request_id dma_id)
+{
-+ /* The Virtual PCI bus has only one root */
-+ return publish_cb(pdev, 0, 0);
++ int rc = ef_vi_transmitv_init(vi, iov, iov_len, dma_id);
++ if( rc == 0 ) ef_vi_transmit_push(vi);
++ return rc;
+}
+
-+void pciback_release_devices(struct pciback_device *pdev)
++
++int ef_vi_transmit_unbundle(ef_vi* vi, const ef_event* __ev,
++ ef_request_id* ids)
+{
-+ int slot, bus;
-+ struct slot_dev_data *slot_dev = pdev->pci_dev_data;
-+ struct pci_dev *dev;
++ ef_request_id* ids_in = ids;
++ ef_vi_txq* q = &vi->vi_txq;
++ ef_vi_txq_state* qs = &vi->ep_state->txq;
++ const ef_vi_qword* ev = EF_GET_HW_EV_PTR(*__ev);
++ unsigned i, stop = (ev->u32[0] + 1) & q->mask;
+
-+ for (bus = 0; bus < PCI_BUS_NBR; bus++)
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ dev = slot_dev->slots[bus][slot];
-+ if (dev != NULL)
-+ pcistub_put_pci_dev(dev);
++ ef_assert(EF_EVENT_TYPE(*__ev) == EF_EVENT_TYPE_TX ||
++ EF_EVENT_TYPE(*__ev) == EF_EVENT_TYPE_TX_ERROR);
++
++ /* Shouldn't be batching more than 64 descriptors, and should not go
++ ** backwards. */
++ ef_assert_le((((ev->u32[0] + 1) - qs->removed) & q->mask), 64);
++ /* Should not complete more than we've posted. */
++ ef_assert_le((((ev->u32[0] + 1) - qs->removed) & q->mask),
++ qs->added - qs->removed);
++
++ for( i = qs->removed & q->mask; i != stop; i = ++qs->removed & q->mask )
++ if( q->ids[i] != 0xffff ) {
++ *ids++ = q->ids[i];
++ q->ids[i] = 0xffff;
+ }
+
-+ kfree(slot_dev);
-+ pdev->pci_dev_data = NULL;
++ ef_assert_le(ids - ids_in, EF_VI_TRANSMIT_BATCH);
++
++ return (int) (ids - ids_in);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/vpci.c linux-2.6.18-xen.hg/drivers/xen/pciback/vpci.c
---- linux-2.6.18/drivers/xen/pciback/vpci.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/vpci.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,204 @@
-+/*
-+ * PCI Backend - Provides a Virtual PCI bus (with real devices)
-+ * to the frontend
++
++/*! \cidoxg_end */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/sysdep.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/sysdep.h 2008-05-19 00:33:48.682952261 +0300
+@@ -0,0 +1,184 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/list.h>
-+#include <linux/slab.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include "pciback.h"
++/*
++ * \author stg
++ * \brief System dependent support for ef vi lib
++ * \date 2007/05/10
++ */
+
-+#define PCI_SLOT_MAX 32
++/*! \cidoxg_include_ci_ul */
++#ifndef __CI_CIUL_SYSDEP_LINUX_H__
++#define __CI_CIUL_SYSDEP_LINUX_H__
+
-+struct vpci_dev_data {
-+ /* Access to dev_list must be protected by lock */
-+ struct list_head dev_list[PCI_SLOT_MAX];
-+ spinlock_t lock;
-+};
++/**********************************************************************
++ * Kernel version compatability
++ */
+
-+static inline struct list_head *list_first(struct list_head *head)
-+{
-+ return head->next;
-+}
++#if defined(__GNUC__)
+
-+struct pci_dev *pciback_get_pci_dev(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus,
-+ unsigned int devfn)
-+{
-+ struct pci_dev_entry *entry;
-+ struct pci_dev *dev = NULL;
-+ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
-+ unsigned long flags;
++/* Linux kernel doesn't have stdint.h or [u]intptr_t. */
++# if !defined(LINUX_VERSION_CODE)
++# include <linux/version.h>
++# endif
++# include <asm/io.h>
+
-+ if (domain != 0 || bus != 0)
-+ return NULL;
++/* In Linux 2.6.24, linux/types.h has uintptr_t */
++# if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
++# if BITS_PER_LONG == 32
++ typedef __u32 uintptr_t;
++# else
++ typedef __u64 uintptr_t;
++# endif
++# endif
+
-+ if (PCI_SLOT(devfn) < PCI_SLOT_MAX) {
-+ spin_lock_irqsave(&vpci_dev->lock, flags);
++/* But even 2.6.24 doesn't define intptr_t */
++# if BITS_PER_LONG == 32
++ typedef __s32 intptr_t;
++# else
++ typedef __s64 intptr_t;
++# endif
+
-+ list_for_each_entry(entry,
-+ &vpci_dev->dev_list[PCI_SLOT(devfn)],
-+ list) {
-+ if (PCI_FUNC(entry->dev->devfn) == PCI_FUNC(devfn)) {
-+ dev = entry->dev;
-+ break;
-+ }
-+ }
++# if defined(__ia64__)
++# define EF_VI_PRIx64 "lx"
++# else
++# define EF_VI_PRIx64 "llx"
++# endif
+
-+ spin_unlock_irqrestore(&vpci_dev->lock, flags);
-+ }
-+ return dev;
-+}
++# define EF_VI_HF __attribute__((visibility("hidden")))
++# define EF_VI_HV __attribute__((visibility("hidden")))
+
-+static inline int match_slot(struct pci_dev *l, struct pci_dev *r)
-+{
-+ if (pci_domain_nr(l->bus) == pci_domain_nr(r->bus)
-+ && l->bus == r->bus && PCI_SLOT(l->devfn) == PCI_SLOT(r->devfn))
-+ return 1;
++# if defined(__i386__) || defined(__x86_64__) /* GCC x86/x64 */
++ typedef unsigned long long ef_vi_dma_addr_t;
++# if __GNUC__ >= 3 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 96)
++# define ef_vi_wiob() __asm__ __volatile__ ("sfence")
++# else
++# define ef_vi_wiob() __asm__ __volatile__ (".byte 0x0F, 0xAE, 0xF8")
++# endif
+
-+ return 0;
-+}
++# endif
++#endif
+
-+int pciback_add_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ int err = 0, slot;
-+ struct pci_dev_entry *t, *dev_entry;
-+ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
-+ unsigned long flags;
++#ifdef EFX_NOT_UPSTREAM
++
++/* Stuff for architectures/compilers not officially supported */
++
++#if !defined(__GNUC__)
++# if defined(__PPC__) /* GCC, PPC */
++ typedef unsigned long ef_vi_dma_addr_t;
++# define ef_vi_wiob() wmb()
++
++# ifdef __powerpc64__
++# ifdef CONFIG_SMP
++# define CI_SMP_SYNC "\n eieio \n" /* memory cache sync */
++# define CI_SMP_ISYNC "\n isync \n" /* instr cache sync */
++# else
++# define CI_SMP_SYNC
++# define CI_SMP_ISYNC
++# endif
++# else /* for ppc32 systems */
++# ifdef CONFIG_SMP
++# define CI_SMP_SYNC "\n eieio \n"
++# define CI_SMP_ISYNC "\n sync \n"
++# else
++# define CI_SMP_SYNC
++# define CI_SMP_ISYNC
++# endif
++# endif
+
-+ if ((dev->class >> 24) == PCI_BASE_CLASS_BRIDGE) {
-+ err = -EFAULT;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Can't export bridges on the virtual PCI bus");
-+ goto out;
-+ }
++# elif defined(__ia64__) /* GCC, IA64 */
++ typedef unsigned long ef_vi_dma_addr_t;
++# define ef_vi_wiob() __asm__ __volatile__("mf.a": : :"memory")
+
-+ dev_entry = kmalloc(sizeof(*dev_entry), GFP_KERNEL);
-+ if (!dev_entry) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error adding entry to virtual PCI bus");
-+ goto out;
-+ }
++# else
++# error Unknown processor - GNU C
++# endif
+
-+ dev_entry->dev = dev;
++#elif defined(__PGI)
++# error PGI not supported
+
-+ spin_lock_irqsave(&vpci_dev->lock, flags);
++#elif defined(__INTEL_COMPILER)
+
-+ /* Keep multi-function devices together on the virtual PCI bus */
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ if (!list_empty(&vpci_dev->dev_list[slot])) {
-+ t = list_entry(list_first(&vpci_dev->dev_list[slot]),
-+ struct pci_dev_entry, list);
++/* Intel compilers v7 claim to be very gcc compatible. */
++# if __INTEL_COMPILER >= 700
++# if __GNUC__ >= 3 || (__GNUC__ == 2 && __GNUC_MINOR__ > 91)
++# define EF_VI_LIKELY(t) __builtin_expect((t), 1)
++# define EF_VI_UNLIKELY(t) __builtin_expect((t), 0)
++# endif
+
-+ if (match_slot(dev, t->dev)) {
-+ pr_info("pciback: vpci: %s: "
-+ "assign to virtual slot %d func %d\n",
-+ pci_name(dev), slot,
-+ PCI_FUNC(dev->devfn));
-+ list_add_tail(&dev_entry->list,
-+ &vpci_dev->dev_list[slot]);
-+ goto unlock;
-+ }
-+ }
-+ }
++# if __GNUC__ >= 3 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 96)
++# define ef_vi_wiob() __asm__ __volatile__ ("sfence")
++# else
++# define ef_vi_wiob() __asm__ __volatile__ (".byte 0x0F, 0xAE, 0xF8")
++# endif
+
-+ /* Assign to a new slot on the virtual PCI bus */
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ if (list_empty(&vpci_dev->dev_list[slot])) {
-+ printk(KERN_INFO
-+ "pciback: vpci: %s: assign to virtual slot %d\n",
-+ pci_name(dev), slot);
-+ list_add_tail(&dev_entry->list,
-+ &vpci_dev->dev_list[slot]);
-+ goto unlock;
-+ }
-+ }
++# else
++# error Old Intel compiler not supported.
++# endif
+
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "No more space on root virtual PCI bus");
++#else
++# error Unknown compiler.
++#endif
+
-+ unlock:
-+ spin_unlock_irqrestore(&vpci_dev->lock, flags);
-+ out:
-+ return err;
-+}
++#endif
+
-+void pciback_release_pci_dev(struct pciback_device *pdev, struct pci_dev *dev)
-+{
-+ int slot;
-+ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
-+ struct pci_dev *found_dev = NULL;
-+ unsigned long flags;
+
-+ spin_lock_irqsave(&vpci_dev->lock, flags);
++# include <linux/errno.h>
+
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ struct pci_dev_entry *e, *tmp;
-+ list_for_each_entry_safe(e, tmp, &vpci_dev->dev_list[slot],
-+ list) {
-+ if (e->dev == dev) {
-+ list_del(&e->list);
-+ found_dev = e->dev;
-+ kfree(e);
-+ goto out;
-+ }
-+ }
-+ }
+
-+ out:
-+ spin_unlock_irqrestore(&vpci_dev->lock, flags);
++/**********************************************************************
++ * Extracting bit fields.
++ */
+
-+ if (found_dev)
-+ pcistub_put_pci_dev(found_dev);
-+}
++#define _QWORD_GET_LOW(f, v) \
++ (((v).u32[0] >> (f##_LBN)) & ((1u << f##_WIDTH) - 1u))
++#define _QWORD_GET_HIGH(f, v) \
++ (((v).u32[1] >> (f##_LBN - 32u)) & ((1u << f##_WIDTH) - 1u))
++#define _QWORD_GET_ANY(f, v) \
++ (((v).u64[0] >> f##_LBN) & (((uint64_t) 1u << f##_WIDTH) - 1u))
+
-+int pciback_init_devices(struct pciback_device *pdev)
-+{
-+ int slot;
-+ struct vpci_dev_data *vpci_dev;
++#define QWORD_GET(f, v) \
++ ((f##_LBN + f##_WIDTH) <= 32u \
++ ? _QWORD_GET_LOW(f, (v)) \
++ : ((f##_LBN >= 32u) ? _QWORD_GET_HIGH(f, (v)) : _QWORD_GET_ANY(f, (v))))
+
-+ vpci_dev = kmalloc(sizeof(*vpci_dev), GFP_KERNEL);
-+ if (!vpci_dev)
-+ return -ENOMEM;
++#define QWORD_GET_U(f, v) ((unsigned) QWORD_GET(f, (v)))
+
-+ spin_lock_init(&vpci_dev->lock);
++#define _QWORD_TEST_BIT_LOW(f, v) ((v).u32[0] & (1u << (f##_LBN)))
++#define _QWORD_TEST_BIT_HIGH(f, v) ((v).u32[1] & (1u << (f##_LBN - 32u)))
+
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ INIT_LIST_HEAD(&vpci_dev->dev_list[slot]);
-+ }
++#define QWORD_TEST_BIT(f, v) \
++ (f##_LBN < 32 ? _QWORD_TEST_BIT_LOW(f, (v)) : _QWORD_TEST_BIT_HIGH(f, (v)))
+
-+ pdev->pci_dev_data = vpci_dev;
+
-+ return 0;
-+}
+
-+int pciback_publish_pci_roots(struct pciback_device *pdev,
-+ publish_pci_root_cb publish_cb)
-+{
-+ /* The Virtual PCI bus has only one root */
-+ return publish_cb(pdev, 0, 0);
-+}
+
-+void pciback_release_devices(struct pciback_device *pdev)
-+{
-+ int slot;
-+ struct vpci_dev_data *vpci_dev = pdev->pci_dev_data;
++#ifndef DECLSPEC_NORETURN
++/* normally defined on Windows to expand to a declaration that the
++ function will not return */
++# define DECLSPEC_NORETURN
++#endif
+
-+ for (slot = 0; slot < PCI_SLOT_MAX; slot++) {
-+ struct pci_dev_entry *e, *tmp;
-+ list_for_each_entry_safe(e, tmp, &vpci_dev->dev_list[slot],
-+ list) {
-+ list_del(&e->list);
-+ pcistub_put_pci_dev(e->dev);
-+ kfree(e);
-+ }
-+ }
++#endif /* __CI_CIUL_SYSDEP_LINUX_H__ */
+--- linux-2.6.18.8/drivers/xen/sfc_netfront/vi_init.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netfront/vi_init.c 2008-05-19 00:33:48.682952261 +0300
+@@ -0,0 +1,183 @@
++/****************************************************************************
++ * Copyright 2002-2005: Level 5 Networks Inc.
++ * Copyright 2005-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications
++ * <linux-xen-drivers@solarflare.com>
++ * <onload-dev@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ kfree(vpci_dev);
-+ pdev->pci_dev_data = NULL;
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pciback/xenbus.c linux-2.6.18-xen.hg/drivers/xen/pciback/xenbus.c
---- linux-2.6.18/drivers/xen/pciback/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pciback/xenbus.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,454 @@
+/*
-+ * PCI Backend Xenbus Setup - handles setup with frontend and xend
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ * \author djr
++ * \brief Initialisation of VIs.
++ * \date 2007/06/08
+ */
-+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <linux/list.h>
-+#include <linux/vmalloc.h>
-+#include <xen/xenbus.h>
-+#include <xen/evtchn.h>
-+#include "pciback.h"
+
-+#define INVALID_EVTCHN_IRQ (-1)
++#include "ef_vi_internal.h"
+
-+static struct pciback_device *alloc_pdev(struct xenbus_device *xdev)
++#define EF_VI_STATE_BYTES(rxq_sz, txq_sz) \
++ (sizeof(ef_vi_state) + (rxq_sz) * sizeof(uint16_t) \
++ + (txq_sz) * sizeof(uint16_t))
++
++int ef_vi_calc_state_bytes(int rxq_sz, int txq_sz)
+{
-+ struct pciback_device *pdev;
++ ef_assert(rxq_sz == 0 || EF_VI_IS_POW2(rxq_sz));
++ ef_assert(txq_sz == 0 || EF_VI_IS_POW2(txq_sz));
+
-+ pdev = kzalloc(sizeof(struct pciback_device), GFP_KERNEL);
-+ if (pdev == NULL)
-+ goto out;
-+ dev_dbg(&xdev->dev, "allocated pdev @ 0x%p\n", pdev);
++ return EF_VI_STATE_BYTES(rxq_sz, txq_sz);
++}
+
-+ pdev->xdev = xdev;
-+ xdev->dev.driver_data = pdev;
+
-+ spin_lock_init(&pdev->dev_lock);
++int ef_vi_state_bytes(ef_vi* vi)
++{
++ int rxq_sz = 0, txq_sz = 0;
++ if( ef_vi_receive_capacity(vi) )
++ rxq_sz = ef_vi_receive_capacity(vi) + 1;
++ if( ef_vi_transmit_capacity(vi) )
++ txq_sz = ef_vi_transmit_capacity(vi) + 1;
+
-+ pdev->sh_area = NULL;
-+ pdev->sh_info = NULL;
-+ pdev->evtchn_irq = INVALID_EVTCHN_IRQ;
-+ pdev->be_watching = 0;
++ ef_assert(rxq_sz == 0 || EF_VI_IS_POW2(rxq_sz));
++ ef_assert(txq_sz == 0 || EF_VI_IS_POW2(txq_sz));
+
-+ INIT_WORK(&pdev->op_work, pciback_do_op, pdev);
++ return EF_VI_STATE_BYTES(rxq_sz, txq_sz);
++}
+
-+ if (pciback_init_devices(pdev)) {
-+ kfree(pdev);
-+ pdev = NULL;
++
++void ef_eventq_state_init(ef_vi* evq)
++{
++ int j;
++
++ for (j = 0; j<EFAB_DMAQS_PER_EVQ_MAX; j++) {
++ ef_rx_dup_state_t *rx_dup_state =
++ &evq->evq_state->rx_dup_state[j];
++ rx_dup_state->bad_sop = 0;
++ rx_dup_state->rx_last_desc_ptr = -1;
++ rx_dup_state->frag_num = 0;
+ }
-+ out:
-+ return pdev;
++
++ evq->evq_state->evq_ptr = 0;
+}
+
-+static void free_pdev(struct pciback_device *pdev)
-+{
-+ if (pdev->be_watching)
-+ unregister_xenbus_watch(&pdev->be_watch);
+
-+ /* Ensure the guest can't trigger our handler before removing devices */
-+ if (pdev->evtchn_irq != INVALID_EVTCHN_IRQ)
-+ unbind_from_irqhandler(pdev->evtchn_irq, pdev);
++void ef_vi_state_init(ef_vi* vi)
++{
++ ef_vi_state* state = vi->ep_state;
++ unsigned i;
+
-+ /* If the driver domain started an op, make sure we complete it or
-+ * delete it before releasing the shared memory */
-+ cancel_delayed_work(&pdev->op_work);
-+ flush_scheduled_work();
++ state->txq.added = state->txq.removed = 0;
++ state->rxq.added = state->rxq.removed = 0;
+
-+ if (pdev->sh_info)
-+ xenbus_unmap_ring_vfree(pdev->xdev, pdev->sh_area);
++ if( vi->vi_rxq.mask )
++ for( i = 0; i <= vi->vi_rxq.mask; ++i )
++ vi->vi_rxq.ids[i] = (uint16_t) -1;
++ if( vi->vi_txq.mask )
++ for( i = 0; i <= vi->vi_txq.mask; ++i )
++ vi->vi_txq.ids[i] = (uint16_t) -1;
++}
+
-+ pciback_release_devices(pdev);
+
-+ pdev->xdev->dev.driver_data = NULL;
-+ pdev->xdev = NULL;
++void ef_vi_init_mapping_evq(void* data_area, struct ef_vi_nic_type nic_type,
++ int instance, unsigned evq_bytes, void* base,
++ void* timer_reg)
++{
++ struct vi_mappings* vm = (struct vi_mappings*) data_area;
+
-+ kfree(pdev);
++ vm->signature = VI_MAPPING_SIGNATURE;
++ vm->vi_instance = instance;
++ vm->nic_type = nic_type;
++ vm->evq_bytes = evq_bytes;
++ vm->evq_base = base;
++ vm->evq_timer_reg = timer_reg;
+}
+
-+static int pciback_do_attach(struct pciback_device *pdev, int gnt_ref,
-+ int remote_evtchn)
++
++void ef_vi_init(ef_vi* vi, void* vvis, ef_vi_state* state,
++ ef_eventq_state* evq_state, enum ef_vi_flags vi_flags)
+{
-+ int err = 0;
-+ struct vm_struct *area;
++ struct vi_mappings* vm = (struct vi_mappings*) vvis;
+
-+ dev_dbg(&pdev->xdev->dev,
-+ "Attaching to frontend resources - gnt_ref=%d evtchn=%d\n",
-+ gnt_ref, remote_evtchn);
++ vi->vi_i = vm->vi_instance;
++ vi->ep_state = state;
++ vi->vi_flags = vi_flags;
+
-+ area = xenbus_map_ring_valloc(pdev->xdev, gnt_ref);
-+ if (IS_ERR(area)) {
-+ err = PTR_ERR(area);
-+ goto out;
++ switch( vm->nic_type.arch ) {
++ case EF_VI_ARCH_FALCON:
++ falcon_vi_init(vi, vvis);
++ break;
++ default:
++ /* ?? TODO: We should return an error code. */
++ ef_assert(0);
++ break;
+ }
-+ pdev->sh_area = area;
-+ pdev->sh_info = area->addr;
+
-+ err = bind_interdomain_evtchn_to_irqhandler(
-+ pdev->xdev->otherend_id, remote_evtchn, pciback_handle_event,
-+ SA_SAMPLE_RANDOM, "pciback", pdev);
-+ if (err < 0) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error binding event channel to IRQ");
-+ goto out;
++ if( vm->evq_bytes ) {
++ vi->evq_state = evq_state;
++ vi->evq_mask = vm->evq_bytes - 1u;
++ vi->evq_base = vm->evq_base;
++ vi->evq_timer_reg = vm->evq_timer_reg;
+ }
-+ pdev->evtchn_irq = err;
-+ err = 0;
+
-+ dev_dbg(&pdev->xdev->dev, "Attached!\n");
-+ out:
-+ return err;
++ EF_VI_MAGIC_SET(vi, EF_VI);
+}
+
-+static int pciback_attach(struct pciback_device *pdev)
++
++/* Initialise [data_area] with information required to initialise an ef_vi.
++ * In the following, an unused param should be set to NULL. Note the case
++ * marked (*) of [iobuf_mmap] for falcon/driver; for the normal driver this
++ * must be NULL.
++ *
++ * \param data_area [in,out] required, must ref at least VI_MAPPING_SIZE
++ * bytes
++ * \param io_mmap [in] ef1, required
++ * falcon, required
++ * \param iobuf_mmap [in] ef1, unused
++ * falcon, required
++ */
++void ef_vi_init_mapping_vi(void* data_area, struct ef_vi_nic_type nic_type,
++ unsigned rxq_capacity, unsigned txq_capacity,
++ int instance, void* io_mmap,
++ void* iobuf_mmap_rx, void* iobuf_mmap_tx,
++ enum ef_vi_flags vi_flags)
+{
-+ int err = 0;
-+ int gnt_ref, remote_evtchn;
-+ char *magic = NULL;
++ struct vi_mappings* vm = (struct vi_mappings*) data_area;
++ int rx_desc_bytes, rxq_bytes;
+
-+ spin_lock(&pdev->dev_lock);
++ ef_assert(rxq_capacity > 0 || txq_capacity > 0);
++ ef_assert(vm);
++ ef_assert(io_mmap);
++ ef_assert(iobuf_mmap_rx || iobuf_mmap_tx);
+
-+ /* Make sure we only do this setup once */
-+ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
-+ XenbusStateInitialised)
-+ goto out;
++ vm->signature = VI_MAPPING_SIGNATURE;
++ vm->vi_instance = instance;
++ vm->nic_type = nic_type;
+
-+ /* Wait for frontend to state that it has published the configuration */
-+ if (xenbus_read_driver_state(pdev->xdev->otherend) !=
-+ XenbusStateInitialised)
-+ goto out;
++ rx_desc_bytes = (vi_flags & EF_VI_RX_PHYS_ADDR) ? 8 : 4;
++ rxq_bytes = rxq_capacity * rx_desc_bytes;
++ rxq_bytes = (rxq_bytes + PAGE_SIZE - 1) & ~(PAGE_SIZE - 1);
+
-+ dev_dbg(&pdev->xdev->dev, "Reading frontend config\n");
++ if( iobuf_mmap_rx == iobuf_mmap_tx )
++ iobuf_mmap_tx = (char*) iobuf_mmap_rx + rxq_bytes;
+
-+ err = xenbus_gather(XBT_NIL, pdev->xdev->otherend,
-+ "pci-op-ref", "%u", &gnt_ref,
-+ "event-channel", "%u", &remote_evtchn,
-+ "magic", NULL, &magic, NULL);
-+ if (err) {
-+ /* If configuration didn't get read correctly, wait longer */
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error reading configuration from frontend");
-+ goto out;
-+ }
++ vm->rx_queue_capacity = rxq_capacity;
++ vm->rx_dma_falcon = iobuf_mmap_rx;
++ vm->rx_bell = (char*) io_mmap + (RX_DESC_UPD_REG_KER_OFST & 4095);
++ vm->tx_queue_capacity = txq_capacity;
++ vm->tx_dma_falcon = iobuf_mmap_tx;
++ vm->tx_bell = (char*) io_mmap + (TX_DESC_UPD_REG_KER_OFST & 4095);
++}
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/Makefile 2008-05-19 00:33:48.682952261 +0300
+@@ -0,0 +1,11 @@
++EXTRA_CFLAGS += -Idrivers/xen/sfc_netutil
++EXTRA_CFLAGS += -Werror
+
-+ if (magic == NULL || strcmp(magic, XEN_PCI_MAGIC) != 0) {
-+ xenbus_dev_fatal(pdev->xdev, -EFAULT,
-+ "version mismatch (%s/%s) with pcifront - "
-+ "halting pciback",
-+ magic, XEN_PCI_MAGIC);
-+ goto out;
-+ }
++ifdef GGOV
++EXTRA_CFLAGS += -fprofile-arcs -ftest-coverage -DEFX_GCOV
++endif
+
-+ err = pciback_do_attach(pdev, gnt_ref, remote_evtchn);
-+ if (err)
-+ goto out;
++obj-$(CONFIG_XEN_NETDEV_ACCEL_SFC_UTIL) := sfc_netutil.o
+
-+ dev_dbg(&pdev->xdev->dev, "Connecting...\n");
++sfc_netutil-objs := accel_cuckoo_hash.o accel_msg_iface.o accel_util.o
+
-+ err = xenbus_switch_state(pdev->xdev, XenbusStateConnected);
-+ if (err)
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error switching to connected state!");
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_cuckoo_hash.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_cuckoo_hash.c 2008-05-19 00:33:48.706953645 +0300
+@@ -0,0 +1,651 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ dev_dbg(&pdev->xdev->dev, "Connected? %d\n", err);
-+ out:
-+ spin_unlock(&pdev->dev_lock);
++#include <linux/types.h> /* needed for linux/random.h */
++#include <linux/random.h>
+
-+ if (magic)
-+ kfree(magic);
++#include "accel_cuckoo_hash.h"
++#include "accel_util.h"
+
-+ return err;
++static inline int cuckoo_hash_key_compare(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key1,
++ cuckoo_hash_key *key2)
++{
++ return !memcmp(key1, key2, hashtab->key_length);
+}
+
-+static void pciback_frontend_changed(struct xenbus_device *xdev,
-+ enum xenbus_state fe_state)
++
++static inline void cuckoo_hash_key_set(cuckoo_hash_key *key1,
++ cuckoo_hash_key *key2)
+{
-+ struct pciback_device *pdev = xdev->dev.driver_data;
++ *key1 = *key2;
++}
+
-+ dev_dbg(&xdev->dev, "fe state changed %d\n", fe_state);
+
-+ switch (fe_state) {
-+ case XenbusStateInitialised:
-+ pciback_attach(pdev);
-+ break;
++/*
++ * Sets hash function parameters. Chooses "a" to be odd, 0 < a < 2^w
++ * where w is the length of the key
++ */
++static void set_hash_parameters(cuckoo_hash_table *hashtab)
++{
++ again:
++ hashtab->a0 = hashtab->a1 = 0;
+
-+ case XenbusStateClosing:
-+ xenbus_switch_state(xdev, XenbusStateClosing);
-+ break;
++ /* Make sure random */
++ get_random_bytes(&hashtab->a0, hashtab->key_length);
++ get_random_bytes(&hashtab->a1, hashtab->key_length);
+
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ dev_dbg(&xdev->dev, "frontend is gone! unregister device\n");
-+ device_unregister(&xdev->dev);
-+ break;
++ /* Make sure odd */
++ hashtab->a0 |= 1;
++ hashtab->a1 |= 1;
+
-+ default:
-+ break;
-+ }
++ /* Being different is good */
++ if (hashtab->a0 != hashtab->a1)
++ return;
++
++ goto again;
+}
+
-+static int pciback_publish_pci_root(struct pciback_device *pdev,
-+ unsigned int domain, unsigned int bus)
++int cuckoo_hash_init(cuckoo_hash_table *hashtab, unsigned length_bits,
++ unsigned key_length)
+{
-+ unsigned int d, b;
-+ int i, root_num, len, err;
-+ char str[64];
-+
-+ dev_dbg(&pdev->xdev->dev, "Publishing pci roots\n");
++ char *table_mem;
++ unsigned length = 1 << length_bits;
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
-+ "root_num", "%d", &root_num);
-+ if (err == 0 || err == -ENOENT)
-+ root_num = 0;
-+ else if (err < 0)
-+ goto out;
-+
-+ /* Verify that we haven't already published this pci root */
-+ for (i = 0; i < root_num; i++) {
-+ len = snprintf(str, sizeof(str), "root-%d", i);
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++ BUG_ON(length_bits >= sizeof(unsigned) * 8);
++ BUG_ON(key_length > sizeof(cuckoo_hash_key));
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename,
-+ str, "%x:%x", &d, &b);
-+ if (err < 0)
-+ goto out;
-+ if (err != 2) {
-+ err = -EINVAL;
-+ goto out;
-+ }
++ table_mem = kmalloc(sizeof(cuckoo_hash_entry) * 2 * length, GFP_KERNEL);
+
-+ if (d == domain && b == bus) {
-+ err = 0;
-+ goto out;
-+ }
-+ }
++ if (table_mem == NULL)
++ return -ENOMEM;
+
-+ len = snprintf(str, sizeof(str), "root-%d", root_num);
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++ hashtab->length = length;
++ hashtab->length_bits = length_bits;
++ hashtab->key_length = key_length;
++ hashtab->entries = 0;
+
-+ dev_dbg(&pdev->xdev->dev, "writing root %d at %04x:%02x\n",
-+ root_num, domain, bus);
++ hashtab->table0 = (cuckoo_hash_entry *)table_mem;
++ hashtab->table1 = (cuckoo_hash_entry *)
++ (table_mem + length * sizeof(cuckoo_hash_entry));
+
-+ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename, str,
-+ "%04x:%02x", domain, bus);
-+ if (err)
-+ goto out;
++ set_hash_parameters(hashtab);
+
-+ err = xenbus_printf(XBT_NIL, pdev->xdev->nodename,
-+ "root_num", "%d", (root_num + 1));
++ /* Zero the table */
++ memset(hashtab->table0, 0, length * 2 * sizeof(cuckoo_hash_entry));
+
-+ out:
-+ return err;
++ return 0;
+}
++EXPORT_SYMBOL_GPL(cuckoo_hash_init);
+
-+static int pciback_export_device(struct pciback_device *pdev,
-+ int domain, int bus, int slot, int func)
++void cuckoo_hash_destroy(cuckoo_hash_table *hashtab)
+{
-+ struct pci_dev *dev;
-+ int err = 0;
-+
-+ dev_dbg(&pdev->xdev->dev, "exporting dom %x bus %x slot %x func %x\n",
-+ domain, bus, slot, func);
++ if (hashtab->table0 != NULL)
++ kfree(hashtab->table0);
++}
+
-+ dev = pcistub_get_pci_dev_by_slot(pdev, domain, bus, slot, func);
-+ if (!dev) {
-+ err = -EINVAL;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Couldn't locate PCI device "
-+ "(%04x:%02x:%02x.%01x)! "
-+ "perhaps already in-use?",
-+ domain, bus, slot, func);
-+ goto out;
-+ }
++EXPORT_SYMBOL_GPL(cuckoo_hash_destroy);
+
-+ err = pciback_add_pci_dev(pdev, dev);
-+ if (err)
-+ goto out;
++/*
++ * This computes sizeof(cuckoo_hash) bits of hash, not all will be
++ * necessarily used, but the hash function throws away any that
++ * aren't
++ */
++static inline void cuckoo_compute_hash_helper(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *a,
++ cuckoo_hash_key *x,
++ cuckoo_hash *result)
++{
++ u64 multiply_result = 0, a_temp, x_temp;
++ u32 carry = 0;
++ u32 *a_words;
++ u32 *x_words;
++ int i;
+
-+ /* TODO: It'd be nice to export a bridge and have all of its children
-+ * get exported with it. This may be best done in xend (which will
-+ * have to calculate resource usage anyway) but we probably want to
-+ * put something in here to ensure that if a bridge gets given to a
-+ * driver domain, that all devices under that bridge are not given
-+ * to other driver domains (as he who controls the bridge can disable
-+ * it and stop the other devices from working).
++ /*
++ * As the mod and div operations in the function effectively
++ * reduce and shift the bits of the product down to just the
++ * third word, we need only compute that and return it as a
++ * result.
++ *
++ * Do enough long multiplication to get the word we need
+ */
-+ out:
-+ return err;
++
++ /* This assumes things about the sizes of the key and hash */
++ BUG_ON(hashtab->key_length % sizeof(u32) != 0);
++ BUG_ON(sizeof(cuckoo_hash) != sizeof(u32));
++
++ a_words = (u32 *)a;
++ x_words = (u32 *)x;
++
++ for (i = 0; i < hashtab->key_length / sizeof(u32); i++) {
++ a_temp = a_words[i];
++ x_temp = x_words[i];
++
++ multiply_result = (a_temp * x_temp) + carry;
++ carry = (multiply_result >> 32) & 0xffffffff;
++ }
++
++ *result = multiply_result & 0xffffffff;
+}
+
-+static int pciback_setup_backend(struct pciback_device *pdev)
++
++/*
++ * Want to implement (ax mod 2^w) div 2^(w-q) for odd a, 0 < a < 2^w;
++ * w is the length of the key, q is the length of the hash, I think.
++ * See http://www.it-c.dk/people/pagh/papers/cuckoo-jour.pdf
++ */
++static cuckoo_hash cuckoo_compute_hash(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_key *a)
+{
-+ /* Get configuration from xend (if available now) */
-+ int domain, bus, slot, func;
-+ int err = 0;
-+ int i, num_devs;
-+ char dev_str[64];
++ unsigned q = hashtab->length_bits;
++ unsigned shift = 32 - q;
++ unsigned mask = ((1 << q) - 1) << shift;
++ cuckoo_hash hash;
+
-+ spin_lock(&pdev->dev_lock);
++ cuckoo_compute_hash_helper(hashtab, a, key, &hash);
+
-+ /* It's possible we could get the call to setup twice, so make sure
-+ * we're not already connected.
++ /*
++ * Take the top few bits to get the right length for this
++ * hash table
+ */
-+ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
-+ XenbusStateInitWait)
-+ goto out;
++ hash = (hash & mask) >> shift;
+
-+ dev_dbg(&pdev->xdev->dev, "getting be setup\n");
++ BUG_ON(hash >= hashtab->length);
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, "num_devs", "%d",
-+ &num_devs);
-+ if (err != 1) {
-+ if (err >= 0)
-+ err = -EINVAL;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error reading number of devices");
-+ goto out;
-+ }
++ return hash;
++}
+
-+ for (i = 0; i < num_devs; i++) {
-+ int l = snprintf(dev_str, sizeof(dev_str), "dev-%d", i);
-+ if (unlikely(l >= (sizeof(dev_str) - 1))) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "String overflow while reading "
-+ "configuration");
-+ goto out;
-+ }
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->nodename, dev_str,
-+ "%x:%x:%x.%x", &domain, &bus, &slot, &func);
-+ if (err < 0) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error reading device configuration");
-+ goto out;
-+ }
-+ if (err != 4) {
-+ err = -EINVAL;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error parsing pci device "
-+ "configuration");
-+ goto out;
-+ }
++static int cuckoo_hash_lookup0(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value *value)
++{
++ cuckoo_hash hash = cuckoo_compute_hash(hashtab, key, &hashtab->a0);
+
-+ err = pciback_export_device(pdev, domain, bus, slot, func);
-+ if (err)
-+ goto out;
++ if ((hashtab->table0[hash].state == CUCKOO_HASH_STATE_OCCUPIED)
++ && cuckoo_hash_key_compare(hashtab, &(hashtab->table0[hash].key),
++ key)) {
++ *value = hashtab->table0[hash].value;
++ return 1;
+ }
+
-+ err = pciback_publish_pci_roots(pdev, pciback_publish_pci_root);
-+ if (err) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error while publish PCI root buses "
-+ "for frontend");
-+ goto out;
-+ }
++ return 0;
++}
+
-+ err = xenbus_switch_state(pdev->xdev, XenbusStateInitialised);
-+ if (err)
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error switching to initialised state!");
++static int cuckoo_hash_lookup1(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value *value)
++{
++ cuckoo_hash hash = cuckoo_compute_hash(hashtab, key, &hashtab->a1);
+
-+ out:
-+ spin_unlock(&pdev->dev_lock);
++ if ((hashtab->table1[hash].state == CUCKOO_HASH_STATE_OCCUPIED)
++ && cuckoo_hash_key_compare(hashtab, &(hashtab->table1[hash].key),
++ key)) {
++ *value = hashtab->table1[hash].value;
++ return 1;
++ }
+
-+ if (!err)
-+ /* see if pcifront is already configured (if not, we'll wait) */
-+ pciback_attach(pdev);
++ return 0;
++}
+
-+ return err;
++
++int cuckoo_hash_lookup(cuckoo_hash_table *hashtab, cuckoo_hash_key *key,
++ cuckoo_hash_value *value)
++{
++ return cuckoo_hash_lookup0(hashtab, key, value)
++ || cuckoo_hash_lookup1(hashtab, key, value);
+}
++EXPORT_SYMBOL_GPL(cuckoo_hash_lookup);
+
-+static void pciback_be_watch(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
++
++/* Transfer any active entries from "old_table" into hashtab */
++static int cuckoo_hash_transfer_entries(cuckoo_hash_table *hashtab,
++ cuckoo_hash_entry *old_table,
++ unsigned capacity)
+{
-+ struct pciback_device *pdev =
-+ container_of(watch, struct pciback_device, be_watch);
++ int i, rc;
++ cuckoo_hash_entry *entry;
+
-+ switch (xenbus_read_driver_state(pdev->xdev->nodename)) {
-+ case XenbusStateInitWait:
-+ pciback_setup_backend(pdev);
-+ break;
++ hashtab->entries = 0;
+
-+ default:
-+ break;
++ for (i = 0; i < capacity; i++) {
++ entry = &old_table[i];
++ if (entry->state == CUCKOO_HASH_STATE_OCCUPIED) {
++ rc = cuckoo_hash_add(hashtab, &(entry->key),
++ entry->value, 0);
++ if (rc != 0) {
++ return rc;
++ }
++ }
+ }
++
++ return 0;
+}
+
-+static int pciback_xenbus_probe(struct xenbus_device *dev,
-+ const struct xenbus_device_id *id)
++
++int cuckoo_hash_rehash(cuckoo_hash_table *hashtab)
+{
-+ int err = 0;
-+ struct pciback_device *pdev = alloc_pdev(dev);
++ cuckoo_hash_entry *new_table;
++ cuckoo_hash_table old_hashtab;
++ int resize = 0, rc, rehash_count;
+
-+ if (pdev == NULL) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(dev, err,
-+ "Error allocating pciback_device struct");
-+ goto out;
++ /*
++ * Store old tables so we can access the existing values and
++ * copy across
++ */
++ memcpy(&old_hashtab, hashtab, sizeof(cuckoo_hash_table));
++
++ /* resize if hashtable is more than half full */
++ if (old_hashtab.entries > old_hashtab.length &&
++ old_hashtab.length_bits < 32)
++ resize = 1;
++
++ resize:
++ if (resize) {
++ new_table = kmalloc(sizeof(cuckoo_hash_entry) * 4 * hashtab->length,
++ GFP_ATOMIC);
++ if (new_table == NULL) {
++ rc = -ENOMEM;
++ goto err;
++ }
++
++ hashtab->length = 2 * hashtab->length;
++ hashtab->length_bits++;
++ } else {
++ new_table = kmalloc(sizeof(cuckoo_hash_entry) * 2 * hashtab->length,
++ GFP_ATOMIC);
++ if (new_table == NULL) {
++ rc = -ENOMEM;
++ goto err;
++ }
+ }
++
++ /*
++ * Point hashtab to new memory region so we can try to
++ * construct new table
++ */
++ hashtab->table0 = new_table;
++ hashtab->table1 = (cuckoo_hash_entry *)
++ ((char *)new_table + hashtab->length * sizeof(cuckoo_hash_entry));
++
++ rehash_count = 0;
+
-+ /* wait for xend to configure us */
-+ err = xenbus_switch_state(dev, XenbusStateInitWait);
-+ if (err)
-+ goto out;
++ again:
++ /* Zero the new tables */
++ memset(new_table, 0, hashtab->length * 2 * sizeof(cuckoo_hash_entry));
+
-+ /* watch the backend node for backend configuration information */
-+ err = xenbus_watch_path(dev, dev->nodename, &pdev->be_watch,
-+ pciback_be_watch);
-+ if (err)
-+ goto out;
-+ pdev->be_watching = 1;
++ /* Choose new parameters for the hash functions */
++ set_hash_parameters(hashtab);
+
-+ /* We need to force a call to our callback here in case
-+ * xend already configured us!
-+ */
-+ pciback_be_watch(&pdev->be_watch, NULL, 0);
++ /*
++ * Multiply old_table_length by 2 as the length refers to each
++ * table, and there are two of them. This assumes that they
++ * are arranged sequentially in memory, so assert it
++ */
++ BUG_ON(((char *)old_hashtab.table1) !=
++ ((char *)old_hashtab.table0 + old_hashtab.length
++ * sizeof(cuckoo_hash_entry)));
++ rc = cuckoo_hash_transfer_entries(hashtab, old_hashtab.table0,
++ old_hashtab.length * 2);
++ if (rc < 0) {
++ /* Problem */
++ if (rc == -ENOSPC) {
++ ++rehash_count;
++ if (rehash_count < CUCKOO_HASH_MAX_LOOP) {
++ /*
++ * Wanted to rehash, but rather than
++ * recurse we can just do it here
++ */
++ goto again;
++ } else {
++ /*
++ * Didn't manage to rehash, so let's
++ * go up a size (if we haven't already
++ * and there's space)
++ */
++ if (!resize && hashtab->length_bits < 32) {
++ resize = 1;
++ kfree(new_table);
++ goto resize;
++ }
++ else
++ goto err;
++ }
++ }
++ else
++ goto err;
++ }
+
-+ out:
-+ return err;
++ /* Success, I think. Free up the old table */
++ kfree(old_hashtab.table0);
++
++ /* We should have put all the entries from old table in the new one */
++ BUG_ON(hashtab->entries != old_hashtab.entries);
++
++ return 0;
++ err:
++ EPRINTK("%s: Rehash failed, giving up\n", __FUNCTION__);
++ /* Some other error, give up, at least restore table to how it was */
++ memcpy(hashtab, &old_hashtab, sizeof(cuckoo_hash_table));
++ if (new_table)
++ kfree(new_table);
++ return rc;
+}
++EXPORT_SYMBOL_GPL(cuckoo_hash_rehash);
+
-+static int pciback_xenbus_remove(struct xenbus_device *dev)
++
++static int
++cuckoo_hash_insert_or_displace(cuckoo_hash_entry *table, unsigned hash,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value value,
++ cuckoo_hash_key *displaced_key,
++ cuckoo_hash_value *displaced_value)
+{
-+ struct pciback_device *pdev = dev->dev.driver_data;
++ if (table[hash].state == CUCKOO_HASH_STATE_VACANT) {
++ cuckoo_hash_key_set(&(table[hash].key), key);
++ table[hash].value = value;
++ table[hash].state = CUCKOO_HASH_STATE_OCCUPIED;
+
-+ if (pdev != NULL)
-+ free_pdev(pdev);
++ return 1;
++ } else {
++ cuckoo_hash_key_set(displaced_key, &(table[hash].key));
++ *displaced_value = table[hash].value;
++ cuckoo_hash_key_set(&(table[hash].key), key);
++ table[hash].value = value;
+
-+ return 0;
++ return 0;
++ }
+}
+
-+static struct xenbus_device_id xenpci_ids[] = {
-+ {"pci"},
-+ {{0}},
-+};
-+
-+static struct xenbus_driver xenbus_pciback_driver = {
-+ .name = "pciback",
-+ .owner = THIS_MODULE,
-+ .ids = xenpci_ids,
-+ .probe = pciback_xenbus_probe,
-+ .remove = pciback_xenbus_remove,
-+ .otherend_changed = pciback_frontend_changed,
-+};
+
-+int __init pciback_xenbus_register(void)
++int cuckoo_hash_add(cuckoo_hash_table *hashtab, cuckoo_hash_key *key,
++ cuckoo_hash_value value, int can_rehash)
+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ cuckoo_hash hash0, hash1;
++ int i, rc;
++ cuckoo_hash_key key1, key2;
+
-+ return xenbus_register_backend(&xenbus_pciback_driver);
++ cuckoo_hash_key_set(&key1, key);
++
++ again:
++ i = 0;
++ do {
++ hash0 = cuckoo_compute_hash(hashtab, &key1, &hashtab->a0);
++ if (cuckoo_hash_insert_or_displace(hashtab->table0, hash0,
++ &key1, value, &key2,
++ &value)) {
++ /* Success */
++ hashtab->entries++;
++ return 0;
++ }
++
++ hash1 = cuckoo_compute_hash(hashtab, &key2, &hashtab->a1);
++ if (cuckoo_hash_insert_or_displace(hashtab->table1, hash1,
++ &key2, value, &key1,
++ &value)) {
++ /* Success */
++ hashtab->entries++;
++ return 0;
++ }
++ } while (++i < CUCKOO_HASH_MAX_LOOP);
++
++ if (can_rehash) {
++ if ((rc = cuckoo_hash_rehash(hashtab)) < 0) {
++ /*
++ * Give up - this will drop whichever
++ * key/value pair we have currently displaced
++ * on the floor
++ */
++ return rc;
++ }
++ goto again;
++ }
++
++ EPRINTK("%s: failed hash add\n", __FUNCTION__);
++ /*
++ * Couldn't do it - bad as we've now removed some random thing
++ * from the table, and will just drop it on the floor. Better
++ * would be to somehow revert the table to the state it was in
++ * at the start
++ */
++ return -ENOSPC;
+}
++EXPORT_SYMBOL_GPL(cuckoo_hash_add);
+
-+void __exit pciback_xenbus_unregister(void)
++
++int cuckoo_hash_add_check(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key, cuckoo_hash_value value,
++ int can_rehash)
+{
-+ xenbus_unregister_driver(&xenbus_pciback_driver);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pcifront/Makefile linux-2.6.18-xen.hg/drivers/xen/pcifront/Makefile
---- linux-2.6.18/drivers/xen/pcifront/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pcifront/Makefile 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,7 @@
-+obj-y += pcifront.o
++ int stored_value;
+
-+pcifront-y := pci_op.o xenbus.o pci.o
++ if (cuckoo_hash_lookup(hashtab, key, &stored_value))
++ return -EBUSY;
+
-+ifeq ($(CONFIG_XEN_PCIDEV_FE_DEBUG),y)
-+EXTRA_CFLAGS += -DDEBUG
-+endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pcifront/pci.c linux-2.6.18-xen.hg/drivers/xen/pcifront/pci.c
---- linux-2.6.18/drivers/xen/pcifront/pci.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pci.c 2007-12-23 11:15:34.067936306 +0100
-@@ -0,0 +1,46 @@
-+/*
-+ * PCI Frontend Operations - ensure only one PCI frontend runs at a time
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include "pcifront.h"
++ return cuckoo_hash_add(hashtab, key, value, can_rehash);
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_add_check);
+
-+DEFINE_SPINLOCK(pcifront_dev_lock);
-+static struct pcifront_device *pcifront_dev = NULL;
+
-+int pcifront_connect(struct pcifront_device *pdev)
++int cuckoo_hash_remove(cuckoo_hash_table *hashtab, cuckoo_hash_key *key)
+{
-+ int err = 0;
-+
-+ spin_lock(&pcifront_dev_lock);
++ cuckoo_hash hash;
+
-+ if (!pcifront_dev) {
-+ dev_info(&pdev->xdev->dev, "Installing PCI frontend\n");
-+ pcifront_dev = pdev;
++ hash = cuckoo_compute_hash(hashtab, key, &hashtab->a0);
++ if ((hashtab->table0[hash].state == CUCKOO_HASH_STATE_OCCUPIED) &&
++ cuckoo_hash_key_compare(hashtab, &(hashtab->table0[hash].key),
++ key)) {
++ hashtab->table0[hash].state = CUCKOO_HASH_STATE_VACANT;
++ hashtab->entries--;
++ return 0;
+ }
-+ else {
-+ dev_err(&pdev->xdev->dev, "PCI frontend already installed!\n");
-+ err = -EEXIST;
++
++ hash = cuckoo_compute_hash(hashtab, key, &hashtab->a1);
++ if ((hashtab->table1[hash].state == CUCKOO_HASH_STATE_OCCUPIED) &&
++ cuckoo_hash_key_compare(hashtab, &(hashtab->table1[hash].key),
++ key)) {
++ hashtab->table1[hash].state = CUCKOO_HASH_STATE_VACANT;
++ hashtab->entries--;
++ return 0;
+ }
-+
-+ spin_unlock(&pcifront_dev_lock);
-+
-+ return err;
++
++ return -EINVAL;
+}
++EXPORT_SYMBOL_GPL(cuckoo_hash_remove);
+
-+void pcifront_disconnect(struct pcifront_device *pdev)
++
++int cuckoo_hash_update(cuckoo_hash_table *hashtab, cuckoo_hash_key *key,
++ cuckoo_hash_value value)
+{
-+ spin_lock(&pcifront_dev_lock);
++ cuckoo_hash hash;
+
-+ if (pdev == pcifront_dev) {
-+ dev_info(&pdev->xdev->dev,
-+ "Disconnecting PCI Frontend Buses\n");
-+ pcifront_dev = NULL;
++ hash = cuckoo_compute_hash(hashtab, key, &hashtab->a0);
++ if ((hashtab->table0[hash].state == CUCKOO_HASH_STATE_OCCUPIED) &&
++ cuckoo_hash_key_compare(hashtab, &(hashtab->table0[hash].key),
++ key)) {
++ hashtab->table0[hash].value = value;
++ return 0;
+ }
+
-+ spin_unlock(&pcifront_dev_lock);
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pcifront/pcifront.h linux-2.6.18-xen.hg/drivers/xen/pcifront/pcifront.h
---- linux-2.6.18/drivers/xen/pcifront/pcifront.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pcifront.h 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,40 @@
-+/*
-+ * PCI Frontend - Common data structures & function declarations
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+#ifndef __XEN_PCIFRONT_H__
-+#define __XEN_PCIFRONT_H__
++ hash = cuckoo_compute_hash(hashtab, key, &hashtab->a1);
++ if ((hashtab->table1[hash].state == CUCKOO_HASH_STATE_OCCUPIED) &&
++ cuckoo_hash_key_compare(hashtab, &(hashtab->table1[hash].key),
++ key)) {
++ hashtab->table1[hash].value = value;
++ return 0;
++ }
+
-+#include <linux/spinlock.h>
-+#include <linux/pci.h>
-+#include <xen/xenbus.h>
-+#include <xen/interface/io/pciif.h>
-+#include <xen/pcifront.h>
++ return -EINVAL;
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_update);
+
-+struct pci_bus_entry {
-+ struct list_head list;
-+ struct pci_bus *bus;
-+};
+
-+struct pcifront_device {
-+ struct xenbus_device *xdev;
-+ struct list_head root_buses;
-+ spinlock_t dev_lock;
++void cuckoo_hash_iterate_reset(cuckoo_hash_table *hashtab)
++{
++ hashtab->iterate_index = 0;
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_iterate_reset);
+
-+ int evtchn;
-+ int gnt_ref;
+
-+ /* Lock this when doing any operations in sh_info */
-+ spinlock_t sh_info_lock;
-+ struct xen_pci_sharedinfo *sh_info;
-+};
++int cuckoo_hash_iterate(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key, cuckoo_hash_value *value)
++{
++ unsigned index;
+
-+int pcifront_connect(struct pcifront_device *pdev);
-+void pcifront_disconnect(struct pcifront_device *pdev);
++ while (hashtab->iterate_index < hashtab->length) {
++ index = hashtab->iterate_index;
++ ++hashtab->iterate_index;
++ if (hashtab->table0[index].state == CUCKOO_HASH_STATE_OCCUPIED) {
++ *key = hashtab->table0[index].key;
++ *value = hashtab->table0[index].value;
++ return 0;
++ }
++ }
+
-+int pcifront_scan_root(struct pcifront_device *pdev,
-+ unsigned int domain, unsigned int bus);
-+void pcifront_free_roots(struct pcifront_device *pdev);
++ while (hashtab->iterate_index >= hashtab->length &&
++ hashtab->iterate_index < hashtab->length * 2) {
++ index = hashtab->iterate_index - hashtab->length;
++ ++hashtab->iterate_index;
++ if (hashtab->table1[index].state == CUCKOO_HASH_STATE_OCCUPIED) {
++ *key = hashtab->table1[index].key;
++ *value = hashtab->table1[index].value;
++ return 0;
++ }
++ }
+
-+#endif /* __XEN_PCIFRONT_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pcifront/pci_op.c linux-2.6.18-xen.hg/drivers/xen/pcifront/pci_op.c
---- linux-2.6.18/drivers/xen/pcifront/pci_op.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pcifront/pci_op.c 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,386 @@
-+/*
-+ * PCI Frontend Operations - Communicates with frontend
-+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
-+ */
-+#include <linux/module.h>
-+#include <linux/version.h>
-+#include <linux/init.h>
-+#include <linux/pci.h>
-+#include <linux/spinlock.h>
-+#include <linux/time.h>
-+#include <xen/evtchn.h>
-+#include "pcifront.h"
++ return -ENOSPC;
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_iterate);
+
-+static int verbose_request = 0;
-+module_param(verbose_request, int, 0644);
+
-+#ifdef __ia64__
-+static void pcifront_init_sd(struct pcifront_sd *sd,
-+ unsigned int domain, unsigned int bus,
-+ struct pcifront_device *pdev)
++#if 0
++void cuckoo_hash_valid(cuckoo_hash_table *hashtab)
+{
-+ int err, i, j, k, len, root_num, res_count;
-+ struct acpi_resource res;
-+ unsigned int d, b, byte;
-+ unsigned long magic;
-+ char str[64], tmp[3];
-+ unsigned char *buf, *bufp;
-+ u8 *ptr;
-+
-+ memset(sd, 0, sizeof(*sd));
-+
-+ sd->segment = domain;
-+ sd->node = -1; /* Revisit for NUMA */
-+ sd->platform_data = pdev;
++ int i, entry_count = 0;
+
-+ /* Look for resources for this controller in xenbus. */
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, "root_num",
-+ "%d", &root_num);
-+ if (err != 1)
++ for (i=0; i < hashtab->length; i++) {
++ EPRINTK_ON(hashtab->table0[i].state != CUCKOO_HASH_STATE_VACANT &&
++ hashtab->table0[i].state != CUCKOO_HASH_STATE_OCCUPIED);
++ if (hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ entry_count++;
++ EPRINTK_ON(hashtab->table1[i].state != CUCKOO_HASH_STATE_VACANT &&
++ hashtab->table1[i].state != CUCKOO_HASH_STATE_OCCUPIED);
++ if (hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ entry_count++;
++ }
++
++ if (entry_count != hashtab->entries) {
++ EPRINTK("%s: bad count\n", __FUNCTION__);
++ cuckoo_hash_dump(hashtab);
+ return;
++ }
+
-+ for (i = 0; i < root_num; i++) {
-+ len = snprintf(str, sizeof(str), "root-%d", i);
-+ if (unlikely(len >= (sizeof(str) - 1)))
-+ return;
-+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
-+ str, "%x:%x", &d, &b);
-+ if (err != 2)
-+ return;
++ for (i=0; i< hashtab->length; i++) {
++ if (hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ if (i != cuckoo_compute_hash(hashtab,
++ &hashtab->table0[i].key,
++ &hashtab->a0)) {
++ EPRINTK("%s: Bad key table 0 index %d\n",
++ __FUNCTION__, i);
++ cuckoo_hash_dump(hashtab);
++ return;
++ }
++ if (hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ if (i != cuckoo_compute_hash(hashtab,
++ &hashtab->table1[i].key,
++ &hashtab->a1)) {
++ EPRINTK("%s: Bad key table 1 index %d\n",
++ __FUNCTION__, i);
++ cuckoo_hash_dump(hashtab);
++ return;
++ }
++ }
+
-+ if (d == domain && b == bus)
-+ break;
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_valid);
++
++
++void cuckoo_hash_dump(cuckoo_hash_table *hashtab)
++{
++ int i, entry_count;
++
++ entry_count = 0;
++ for (i=0; i < hashtab->length; i++) {
++ EPRINTK_ON(hashtab->table0[i].state != CUCKOO_HASH_STATE_VACANT &&
++ hashtab->table0[i].state != CUCKOO_HASH_STATE_OCCUPIED);
++ if (hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ entry_count++;
++ EPRINTK_ON(hashtab->table1[i].state != CUCKOO_HASH_STATE_VACANT &&
++ hashtab->table1[i].state != CUCKOO_HASH_STATE_OCCUPIED);
++ if (hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ entry_count++;
++ }
++
++ EPRINTK("======================\n");
++ EPRINTK("Cuckoo hash table dump\n");
++ EPRINTK("======================\n");
++ EPRINTK("length: %d; length_bits: %d; key_length: %d\n", hashtab->length,
++ hashtab->length_bits, hashtab->key_length);
++ EPRINTK("Recorded entries: %d\n", hashtab->entries);
++ EPRINTK("Counted entries: %d\n", entry_count);
++ EPRINTK("a0: %llx; a1: %llx\n", hashtab->a0, hashtab->a1);
++ EPRINTK("-----------------------------------------\n");
++ EPRINTK("Index Occupied Key Value Index0 Index1\n");
++ EPRINTK("-----------------------------------------\n");
++ for (i=0; i< hashtab->length; i++) {
++ if (hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ EPRINTK("%d %d %llx %d %d %d\n", i,
++ hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED,
++ hashtab->table0[i].key, hashtab->table0[i].value,
++ cuckoo_compute_hash(hashtab, &hashtab->table0[i].key,
++ &hashtab->a0),
++ cuckoo_compute_hash(hashtab, &hashtab->table0[i].key,
++ &hashtab->a1));
++ else
++ EPRINTK("%d %d - - - -\n", i,
++ hashtab->table0[i].state == CUCKOO_HASH_STATE_OCCUPIED);
++
+ }
++ EPRINTK("-----------------------------------------\n");
++ EPRINTK("Index Occupied Key Value Index0 Index1\n");
++ EPRINTK("-----------------------------------------\n");
++ for (i=0; i< hashtab->length; i++) {
++ if (hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED)
++ EPRINTK("%d %d %llx %d %d %d\n", i,
++ hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED,
++ hashtab->table1[i].key, hashtab->table1[i].value,
++ cuckoo_compute_hash(hashtab, &hashtab->table1[i].key,
++ &hashtab->a0),
++ cuckoo_compute_hash(hashtab, &hashtab->table1[i].key,
++ &hashtab->a1));
++ else
++ EPRINTK("%d %d - - - -\n", i,
++ hashtab->table1[i].state == CUCKOO_HASH_STATE_OCCUPIED);
++ }
++ EPRINTK("======================\n");
++}
++EXPORT_SYMBOL_GPL(cuckoo_hash_dump);
++#endif
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_cuckoo_hash.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_cuckoo_hash.h 2008-05-19 00:33:48.730955028 +0300
+@@ -0,0 +1,227 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ if (i == root_num)
-+ return;
++/*
++ * A cuckoo hash table consists of two sub tables. Each entry can
++ * hash to a position in each table. If, on entry, its position is
++ * found to be occupied, the existing element is moved to it's other
++ * location. This recurses until success or a loop is found. If a
++ * loop is found the table is rehashed.
++ *
++ * See http://www.it-c.dk/people/pagh/papers/cuckoo-jour.pdf
++ */
+
-+ len = snprintf(str, sizeof(str), "root-resource-magic");
++#ifndef NET_ACCEL_CUCKOO_HASH_H
++#define NET_ACCEL_CUCKOO_HASH_H
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
-+ str, "%lx", &magic);
++/*! Type used for hash table keys of ip pairs */
++typedef struct {
++ u32 local_ip;
++ //u32 remote_ip;
++ u16 local_port;
++ //u16 remote_port;
++ /* Technically only 1 bit, but use 16 to make key a round
++ number size */
++ u16 proto;
++} cuckoo_hash_ip_key;
+
-+ if (err != 1)
-+ return; /* No resources, nothing to do */
++/*! Type used for hash table keys of mac addresses */
++typedef u64 cuckoo_hash_mac_key;
+
-+ if (magic != (sizeof(res) * 2) + 1) {
-+ printk(KERN_WARNING "pcifront: resource magic mismatch\n");
-+ return;
-+ }
++/*! This type is designed to be large enough to hold all supported key
++ * sizes to avoid having to malloc storage for them.
++ */
++typedef u64 cuckoo_hash_key;
+
-+ len = snprintf(str, sizeof(str), "root-%d-resources", i);
-+ if (unlikely(len >= (sizeof(str) - 1)))
-+ return;
++/*! Type used for the values stored in the hash table */
++typedef int cuckoo_hash_value;
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
-+ str, "%d", &res_count);
++/*! Type used for the hash used to index the table */
++typedef u32 cuckoo_hash;
+
-+ if (err != 1)
-+ return; /* No resources, nothing to do */
++/*! How long to spend displacing values when adding before giving up
++ * and rehashing */
++#define CUCKOO_HASH_MAX_LOOP (hashtab->length)
+
-+ sd->window = kzalloc(sizeof(*sd->window) * res_count, GFP_KERNEL);
-+ if (!sd->window)
-+ return;
++/*! State of hash table entry */
++typedef enum {
++ CUCKOO_HASH_STATE_VACANT = 0,
++ CUCKOO_HASH_STATE_OCCUPIED
++} cuckoo_hash_state;
+
-+ /* magic is also the size of the byte stream in xenbus */
-+ buf = kmalloc(magic, GFP_KERNEL);
-+ if (!buf) {
-+ kfree(sd->window);
-+ sd->window = NULL;
-+ return;
-+ }
++/*! An entry in the hash table */
++typedef struct {
++ cuckoo_hash_state state;
++ cuckoo_hash_key key;
++ cuckoo_hash_value value;
++} cuckoo_hash_entry;
+
-+ /* Read the resources out of xenbus */
-+ for (j = 0; j < res_count; j++) {
-+ memset(&res, 0, sizeof(res));
-+ memset(buf, 0, magic);
++/*! A cuckoo hash table */
++typedef struct {
++ /*! The length of each table (NB. there are two tables of this
++ * length) */
++ unsigned length;
++ /*! The length of each table in bits */
++ unsigned length_bits;
++ /*! The length of the key in bytes */
++ unsigned key_length;
++ /*! The number of entries currently stored in the table */
++ unsigned entries;
++ /*! Index into table used by cuckoo_hash_iterate */
++ unsigned iterate_index;
++
++ /* parameter of hash functions */
++ /*! The "a" parameter of the first hash function */
++ cuckoo_hash_key a0;
++ /*! The "a" parameter of the second hash function */
++ cuckoo_hash_key a1;
++
++ /*! The first table */
++ cuckoo_hash_entry *table0;
++ /*! The second table */
++ cuckoo_hash_entry *table1;
++} cuckoo_hash_table;
++
++/*! Initialise the cuckoo has table
++ *
++ * \param hashtab A pointer to an unitialised hash table structure
++ * \param length_bits The number of elements in each table equals
++ * 2**length_bits
++ * \param key_length The length of the key in bytes
++ *
++ * \return 0 on success, -ENOMEM if it couldn't allocate the tables
++ */
++extern
++int cuckoo_hash_init(cuckoo_hash_table *hashtab, unsigned length_bits,
++ unsigned key_length);
+
-+ len = snprintf(str, sizeof(str), "root-%d-resource-%d", i, j);
-+ if (unlikely(len >= (sizeof(str) - 1)))
-+ return;
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
-+ "%s", buf);
-+ if (err != 1) {
-+ printk(KERN_WARNING "pcifront: error reading "
-+ "resource %d on bus %04x:%02x\n",
-+ j, domain, bus);
-+ continue;
-+ }
++/*! Destroy a hash table
++ *
++ * \param hashtab A hash table that has previously been passed to a
++ * successful call of cuckoo_hash_init()
++ */
++extern
++void cuckoo_hash_destroy(cuckoo_hash_table *hashtab);
+
-+ bufp = buf;
-+ ptr = (u8 *)&res;
-+ memset(tmp, 0, sizeof(tmp));
+
-+ /* Copy ASCII byte stream into structure */
-+ for (k = 0; k < magic - 1; k += 2) {
-+ memcpy(tmp, bufp, 2);
-+ bufp += 2;
++/*! Lookup an entry in the hash table
++ *
++ * \param hashtab The hash table in which to look.
++ * \param key Pointer to a mac address to use as the key
++ * \param value On exit set to the value stored if key was present
++ *
++ * \return 0 if not present in the table, non-zero if it is (and value
++ * is set accordingly)
++ */
++extern
++int cuckoo_hash_lookup(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value *value);
+
-+ sscanf(tmp, "%02x", &byte);
-+ *ptr = byte;
-+ ptr++;
-+ }
++/*! Add an entry to the hash table. Key must not be a duplicate of
++ * anything already in the table. If this is a risk, see
++ * cuckoo_hash_add_check
++ *
++ * \param hashtab The hash table to add the entry to
++ * \param key Pointer to a mac address to use as a key
++ * \param value The value to store
++ * \param can_rehash Flag to allow the add function to rehash the
++ * table if necessary
++ *
++ * \return 0 on success, non-zero on failure. -ENOSPC means it just
++ * couldn't find anywhere to put it - this is bad and probably means
++ * an entry has been dropped on the floor (but the entry you just
++ * tried to add may now be included)
++ */
++extern
++int cuckoo_hash_add(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value value,
++ int can_rehash);
+
-+ xen_add_resource(sd, domain, bus, &res);
-+ sd->windows++;
-+ }
-+ kfree(buf);
-+}
-+#endif
++/*! Same as cuckoo_hash_add but first checks to ensure entry is not
++ * already there
++ * \return -EBUSY if already there
++ */
+
-+static int errno_to_pcibios_err(int errno)
-+{
-+ switch (errno) {
-+ case XEN_PCI_ERR_success:
-+ return PCIBIOS_SUCCESSFUL;
++extern
++int cuckoo_hash_add_check(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key,
++ cuckoo_hash_value value,
++ int can_rehash);
++/*! Remove an entry from the table
++ *
++ * \param hashtab The hash table to remove the entry from
++ * \param key The key that was used to previously add the entry
++ *
++ * \return 0 on success, -EINVAL if the entry couldn't be found
++ */
++extern
++int cuckoo_hash_remove(cuckoo_hash_table *hashtab, cuckoo_hash_key *key);
+
-+ case XEN_PCI_ERR_dev_not_found:
-+ return PCIBIOS_DEVICE_NOT_FOUND;
+
-+ case XEN_PCI_ERR_invalid_offset:
-+ case XEN_PCI_ERR_op_failed:
-+ return PCIBIOS_BAD_REGISTER_NUMBER;
++/*! Helper for those using mac addresses to convert to a key for the
++ * hash table
++ */
++static inline cuckoo_hash_mac_key cuckoo_mac_to_key(const u8 *mac)
++{
++ return (cuckoo_hash_mac_key)(mac[0])
++ | (cuckoo_hash_mac_key)(mac[1]) << 8
++ | (cuckoo_hash_mac_key)(mac[2]) << 16
++ | (cuckoo_hash_mac_key)(mac[3]) << 24
++ | (cuckoo_hash_mac_key)(mac[4]) << 32
++ | (cuckoo_hash_mac_key)(mac[5]) << 40;
++}
+
-+ case XEN_PCI_ERR_not_implemented:
-+ return PCIBIOS_FUNC_NOT_SUPPORTED;
+
-+ case XEN_PCI_ERR_access_denied:
-+ return PCIBIOS_SET_FAILED;
-+ }
-+ return errno;
-+}
++/*! Update an entry already in the hash table to take a new value
++ *
++ * \param hashtab The hash table to add the entry to
++ * \param key Pointer to a mac address to use as a key
++ * \param value The value to store
++ *
++ * \return 0 on success, non-zero on failure.
++ */
++int cuckoo_hash_update(cuckoo_hash_table *hashtab, cuckoo_hash_key *key,
++ cuckoo_hash_value value);
+
-+static int do_pci_op(struct pcifront_device *pdev, struct xen_pci_op *op)
-+{
-+ int err = 0;
-+ struct xen_pci_op *active_op = &pdev->sh_info->op;
-+ unsigned long irq_flags;
-+ evtchn_port_t port = pdev->evtchn;
-+ s64 ns, ns_timeout;
-+ struct timeval tv;
+
-+ spin_lock_irqsave(&pdev->sh_info_lock, irq_flags);
++/*! Go through the hash table and return all used entries (one per call)
++ *
++ * \param hashtab The hash table to iterate over
++ * \param key Pointer to a key to take the returned key
++ * \param value Pointer to a value to take the returned value
++ *
++ * \return 0 on success (key, value set), non-zero on failure.
++ */
++int cuckoo_hash_iterate(cuckoo_hash_table *hashtab,
++ cuckoo_hash_key *key, cuckoo_hash_value *value);
++void cuckoo_hash_iterate_reset(cuckoo_hash_table *hashtab);
+
-+ memcpy(active_op, op, sizeof(struct xen_pci_op));
++/* debug, not compiled by default */
++void cuckoo_hash_valid(cuckoo_hash_table *hashtab);
++void cuckoo_hash_dump(cuckoo_hash_table *hashtab);
++
++#endif /* NET_ACCEL_CUCKOO_HASH_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_msg_iface.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_msg_iface.c 2008-05-19 00:33:48.730955028 +0300
+@@ -0,0 +1,301 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ /* Go */
-+ wmb();
-+ set_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags);
-+ notify_remote_via_evtchn(port);
++#include <xen/evtchn.h>
+
-+ /*
-+ * We set a poll timeout of 3 seconds but give up on return after
-+ * 2 seconds. It is better to time out too late rather than too early
-+ * (in the latter case we end up continually re-executing poll() with a
-+ * timeout in the past). 1s difference gives plenty of slack for error.
-+ */
-+ do_gettimeofday(&tv);
-+ ns_timeout = timeval_to_ns(&tv) + 2 * (s64)NSEC_PER_SEC;
++#include "accel_util.h"
++#include "accel_msg_iface.h"
+
-+ clear_evtchn(port);
++#define NET_ACCEL_MSG_Q_SIZE (1024)
++#define NET_ACCEL_MSG_Q_MASK (NET_ACCEL_MSG_Q_SIZE - 1)
+
-+ while (test_bit(_XEN_PCIF_active,
-+ (unsigned long *)&pdev->sh_info->flags)) {
-+ if (HYPERVISOR_poll(&port, 1, jiffies + 3*HZ))
-+ BUG();
-+ clear_evtchn(port);
-+ do_gettimeofday(&tv);
-+ ns = timeval_to_ns(&tv);
-+ if (ns > ns_timeout) {
-+ dev_err(&pdev->xdev->dev,
-+ "pciback not responding!!!\n");
-+ clear_bit(_XEN_PCIF_active,
-+ (unsigned long *)&pdev->sh_info->flags);
-+ err = XEN_PCI_ERR_dev_not_found;
-+ goto out;
-+ }
++#ifdef NDEBUG
++#define NET_ACCEL_CHECK_MAGIC(_p, _errval)
++#define NET_ACCEL_SHOW_QUEUE(_t, _q, _id)
++#else
++#define NET_ACCEL_CHECK_MAGIC(_p, _errval) \
++ if (_p->magic != NET_ACCEL_MSG_MAGIC) { \
++ printk(KERN_ERR "%s: passed invalid shared page %p!\n", \
++ __FUNCTION__, _p); \
++ return _errval; \
+ }
++#define NET_ACCEL_SHOW_QUEUE(_t, _q, _id) \
++ printk(_t ": queue %d write %x read %x base %x limit %x\n", \
++ _id, _q->write, _q->read, _q->base, _q->limit);
++#endif
+
-+ memcpy(op, active_op, sizeof(struct xen_pci_op));
-+
-+ err = op->err;
-+ out:
-+ spin_unlock_irqrestore(&pdev->sh_info_lock, irq_flags);
-+ return err;
-+}
-+
-+/* Access to this function is spinlocked in drivers/pci/access.c */
-+static int pcifront_bus_read(struct pci_bus *bus, unsigned int devfn,
-+ int where, int size, u32 * val)
++/*
++ * We've been passed at least 2 pages. 1 control page and 1 or more
++ * data pages.
++ */
++int net_accel_msg_init_page(void *mem, int len, int up)
+{
-+ int err = 0;
-+ struct xen_pci_op op = {
-+ .cmd = XEN_PCI_OP_conf_read,
-+ .domain = pci_domain_nr(bus),
-+ .bus = bus->number,
-+ .devfn = devfn,
-+ .offset = where,
-+ .size = size,
-+ };
-+ struct pcifront_sd *sd = bus->sysdata;
-+ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++ struct net_accel_shared_page *shared_page =
++ (struct net_accel_shared_page*)mem;
+
-+ if (verbose_request)
-+ dev_info(&pdev->xdev->dev,
-+ "read dev=%04x:%02x:%02x.%01x - offset %x size %d\n",
-+ pci_domain_nr(bus), bus->number, PCI_SLOT(devfn),
-+ PCI_FUNC(devfn), where, size);
++ if ((unsigned long)shared_page & NET_ACCEL_MSG_Q_MASK)
++ return -EINVAL;
+
-+ err = do_pci_op(pdev, &op);
++ shared_page->magic = NET_ACCEL_MSG_MAGIC;
+
-+ if (likely(!err)) {
-+ if (verbose_request)
-+ dev_info(&pdev->xdev->dev, "read got back value %x\n",
-+ op.value);
++ shared_page->aflags = 0;
+
-+ *val = op.value;
-+ } else if (err == -ENODEV) {
-+ /* No device here, pretend that it just returned 0 */
-+ err = 0;
-+ *val = 0;
-+ }
++ shared_page->net_dev_up = up;
+
-+ return errno_to_pcibios_err(err);
++ return 0;
+}
++EXPORT_SYMBOL_GPL(net_accel_msg_init_page);
+
-+/* Access to this function is spinlocked in drivers/pci/access.c */
-+static int pcifront_bus_write(struct pci_bus *bus, unsigned int devfn,
-+ int where, int size, u32 val)
++
++void net_accel_msg_init_queue(sh_msg_fifo2 *queue,
++ struct net_accel_msg_queue *indices,
++ struct net_accel_msg *base, int size)
+{
-+ struct xen_pci_op op = {
-+ .cmd = XEN_PCI_OP_conf_write,
-+ .domain = pci_domain_nr(bus),
-+ .bus = bus->number,
-+ .devfn = devfn,
-+ .offset = where,
-+ .size = size,
-+ .value = val,
-+ };
-+ struct pcifront_sd *sd = bus->sysdata;
-+ struct pcifront_device *pdev = pcifront_get_pdev(sd);
++ queue->fifo = base;
++ spin_lock_init(&queue->lock);
++ sh_fifo2_init(queue, size-1, &indices->read, &indices->write);
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_init_queue);
+
-+ if (verbose_request)
-+ dev_info(&pdev->xdev->dev,
-+ "write dev=%04x:%02x:%02x.%01x - "
-+ "offset %x size %d val %x\n",
-+ pci_domain_nr(bus), bus->number,
-+ PCI_SLOT(devfn), PCI_FUNC(devfn), where, size, val);
+
-+ return errno_to_pcibios_err(do_pci_op(pdev, &op));
++static inline int _net_accel_msg_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ struct net_accel_msg *msg,
++ int is_reply)
++{
++ int rc = 0;
++ NET_ACCEL_CHECK_MAGIC(sp, -EINVAL);
++ rmb();
++ if (is_reply) {
++ EPRINTK_ON(sh_fifo2_is_full(queue));
++ sh_fifo2_put(queue, *msg);
++ } else {
++ if (sh_fifo2_not_half_full(queue)) {
++ sh_fifo2_put(queue, *msg);
++ } else {
++ rc = -ENOSPC;
++ }
++ }
++ wmb();
++ return rc;
+}
+
-+struct pci_ops pcifront_bus_ops = {
-+ .read = pcifront_bus_read,
-+ .write = pcifront_bus_write,
-+};
++/* Notify after a batch of messages have been sent */
++void net_accel_msg_notify(int irq)
++{
++ notify_remote_via_irq(irq);
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_notify);
+
-+/* Claim resources for the PCI frontend as-is, backend won't allow changes */
-+static void pcifront_claim_resource(struct pci_dev *dev, void *data)
++/*
++ * Send a message on the specified FIFO. Returns 0 on success, -errno
++ * on failure. The message in msg is copied to the current slot of the
++ * FIFO.
++ */
++int net_accel_msg_send(struct net_accel_shared_page *sp, sh_msg_fifo2 *q,
++ struct net_accel_msg *msg)
+{
-+ struct pcifront_device *pdev = data;
-+ int i;
-+ struct resource *r;
++ unsigned long flags;
++ int rc;
++ net_accel_msg_lock_queue(q, &flags);
++ rc = _net_accel_msg_send(sp, q, msg, 0);
++ net_accel_msg_unlock_queue(q, &flags);
++ return rc;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_send);
+
-+ for (i = 0; i < PCI_NUM_RESOURCES; i++) {
-+ r = &dev->resource[i];
+
-+ if (!r->parent && r->start && r->flags) {
-+ dev_dbg(&pdev->xdev->dev, "claiming resource %s/%d\n",
-+ pci_name(dev), i);
-+ pci_claim_resource(dev, i);
-+ }
-+ }
++/* As net_accel_msg_send but also posts a notification to the far end. */
++int net_accel_msg_send_notify(struct net_accel_shared_page *sp, int irq,
++ sh_msg_fifo2 *q, struct net_accel_msg *msg)
++{
++ unsigned long flags;
++ int rc;
++ net_accel_msg_lock_queue(q, &flags);
++ rc = _net_accel_msg_send(sp, q, msg, 0);
++ net_accel_msg_unlock_queue(q, &flags);
++ if (rc >= 0)
++ notify_remote_via_irq(irq);
++ return rc;
+}
++EXPORT_SYMBOL_GPL(net_accel_msg_send_notify);
+
-+int pcifront_scan_root(struct pcifront_device *pdev,
-+ unsigned int domain, unsigned int bus)
++
++int net_accel_msg_reply(struct net_accel_shared_page *sp, sh_msg_fifo2 *q,
++ struct net_accel_msg *msg)
+{
-+ struct pci_bus *b;
-+ struct pcifront_sd *sd = NULL;
-+ struct pci_bus_entry *bus_entry = NULL;
-+ int err = 0;
++ unsigned long flags;
++ int rc;
++ net_accel_msg_lock_queue(q, &flags);
++ rc = _net_accel_msg_send(sp, q, msg, 1);
++ net_accel_msg_unlock_queue(q, &flags);
++ return rc;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_reply);
+
-+#ifndef CONFIG_PCI_DOMAINS
-+ if (domain != 0) {
-+ dev_err(&pdev->xdev->dev,
-+ "PCI Root in non-zero PCI Domain! domain=%d\n", domain);
-+ dev_err(&pdev->xdev->dev,
-+ "Please compile with CONFIG_PCI_DOMAINS\n");
-+ err = -EINVAL;
-+ goto err_out;
-+ }
-+#endif
+
-+ dev_info(&pdev->xdev->dev, "Creating PCI Frontend Bus %04x:%02x\n",
-+ domain, bus);
++/* As net_accel_msg_send but also posts a notification to the far end. */
++int net_accel_msg_reply_notify(struct net_accel_shared_page *sp, int irq,
++ sh_msg_fifo2 *q, struct net_accel_msg *msg)
++{
++ unsigned long flags;
++ int rc;
++ net_accel_msg_lock_queue(q, &flags);
++ rc = _net_accel_msg_send(sp, q, msg, 1);
++ net_accel_msg_unlock_queue(q, &flags);
++ if (rc >= 0)
++ notify_remote_via_irq(irq);
++ return rc;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_reply_notify);
+
-+ bus_entry = kmalloc(sizeof(*bus_entry), GFP_KERNEL);
-+ sd = kmalloc(sizeof(*sd), GFP_KERNEL);
-+ if (!bus_entry || !sd) {
-+ err = -ENOMEM;
-+ goto err_out;
-+ }
-+ pcifront_init_sd(sd, domain, bus, pdev);
+
-+ b = pci_scan_bus_parented(&pdev->xdev->dev, bus,
-+ &pcifront_bus_ops, sd);
-+ if (!b) {
-+ dev_err(&pdev->xdev->dev,
-+ "Error creating PCI Frontend Bus!\n");
-+ err = -ENOMEM;
-+ goto err_out;
++/*
++ * Look at a received message, if any, so a decision can be made about
++ * whether to read it now or not. Cookie is a bit of debug which is
++ * set here and checked when passed to net_accel_msg_recv_next()
++ */
++int net_accel_msg_peek(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ struct net_accel_msg *msg, int *cookie)
++{
++ unsigned long flags;
++ int rc = 0;
++ NET_ACCEL_CHECK_MAGIC(sp, -EINVAL);
++ net_accel_msg_lock_queue(queue, &flags);
++ rmb();
++ if (sh_fifo2_is_empty(queue)) {
++ rc = -ENOENT;
++ } else {
++ *msg = sh_fifo2_peek(queue);
++ *cookie = *(queue->fifo_rd_i);
+ }
++ net_accel_msg_unlock_queue(queue, &flags);
++ return rc;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_peek);
+
-+ pcifront_setup_root_resources(b, sd);
-+ bus_entry->bus = b;
-+
-+ list_add(&bus_entry->list, &pdev->root_buses);
+
-+ /* Claim resources before going "live" with our devices */
-+ pci_walk_bus(b, pcifront_claim_resource, pdev);
++/*
++ * Move the queue onto the next element, used after finished with a
++ * peeked msg
++ */
++int net_accel_msg_recv_next(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue, int cookie)
++{
++ unsigned long flags;
++ NET_ACCEL_CHECK_MAGIC(sp, -EINVAL);
++ net_accel_msg_lock_queue(queue, &flags);
++ rmb();
++ /* Mustn't be empty */
++ BUG_ON(sh_fifo2_is_empty(queue));
++ /*
++ * Check cookie matches, i.e. we're advancing over the same message
++ * as was got using peek
++ */
++ BUG_ON(cookie != *(queue->fifo_rd_i));
++ sh_fifo2_rd_next(queue);
++ wmb();
++ net_accel_msg_unlock_queue(queue, &flags);
++ return 0;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_recv_next);
+
-+ pci_bus_add_devices(b);
+
-+ return 0;
++/*
++ * Receive a message on the specified FIFO. Returns 0 on success,
++ * -errno on failure.
++ */
++int net_accel_msg_recv(struct net_accel_shared_page *sp, sh_msg_fifo2 *queue,
++ struct net_accel_msg *msg)
++{
++ unsigned long flags;
++ int rc = 0;
++ NET_ACCEL_CHECK_MAGIC(sp, -EINVAL);
++ net_accel_msg_lock_queue(queue, &flags);
++ rmb();
++ if (sh_fifo2_is_empty(queue)) {
++ rc = -ENOENT;
++ } else {
++ sh_fifo2_get(queue, msg);
++ }
++ wmb();
++ net_accel_msg_unlock_queue(queue, &flags);
++ return rc;
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_recv);
+
-+ err_out:
-+ kfree(bus_entry);
-+ kfree(sd);
+
-+ return err;
++/*
++ * Start sending a message without copying. returns a pointer to a message
++ * that will be filled out in place. The queue is locked until the message
++ * is sent.
++ */
++struct net_accel_msg *net_accel_msg_start_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue, unsigned long *flags)
++{
++ struct net_accel_msg *msg;
++ NET_ACCEL_CHECK_MAGIC(sp, NULL);
++ net_accel_msg_lock_queue(queue, flags);
++ rmb();
++ if (sh_fifo2_not_half_full(queue)) {
++ msg = sh_fifo2_pokep(queue);
++ } else {
++ net_accel_msg_unlock_queue(queue, flags);
++ msg = NULL;
++ }
++ return msg;
+}
++EXPORT_SYMBOL_GPL(net_accel_msg_start_send);
+
-+static void free_root_bus_devs(struct pci_bus *bus)
++
++static inline void _msg_complete(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags)
+{
-+ struct pci_dev *dev;
++ sh_fifo2_wr_next(queue);
++ net_accel_msg_unlock_queue(queue, flags);
++}
+
-+ while (!list_empty(&bus->devices)) {
-+ dev = container_of(bus->devices.next, struct pci_dev,
-+ bus_list);
-+ dev_dbg(&dev->dev, "removing device\n");
-+ pci_remove_bus_device(dev);
-+ }
++/*
++ * Complete the sending of a message started with net_accel_msg_start_send. The
++ * message is implicit since the queue was locked by _start
++ */
++void net_accel_msg_complete_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags)
++{
++ _msg_complete(sp, queue, flags);
+}
++EXPORT_SYMBOL_GPL(net_accel_msg_complete_send);
+
-+void pcifront_free_roots(struct pcifront_device *pdev)
++/* As net_accel_msg_complete_send but does the notify. */
++void net_accel_msg_complete_send_notify(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags, int irq)
+{
-+ struct pci_bus_entry *bus_entry, *t;
++ _msg_complete(sp, queue, flags);
++ notify_remote_via_irq(irq);
++}
++EXPORT_SYMBOL_GPL(net_accel_msg_complete_send_notify);
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_msg_iface.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_msg_iface.h 2008-05-19 00:33:48.742955720 +0300
+@@ -0,0 +1,414 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ dev_dbg(&pdev->xdev->dev, "cleaning up root buses\n");
++#ifndef NET_ACCEL_MSG_IFACE_H
++#define NET_ACCEL_MSG_IFACE_H
+
-+ list_for_each_entry_safe(bus_entry, t, &pdev->root_buses, list) {
-+ list_del(&bus_entry->list);
++#include <linux/ip.h>
++#include <linux/tcp.h>
++#include <linux/udp.h>
++#include <linux/in.h>
++#include <linux/netdevice.h>
++#include <linux/etherdevice.h>
+
-+ free_root_bus_devs(bus_entry->bus);
++#include "accel_shared_fifo.h"
+
-+ kfree(bus_entry->bus->sysdata);
++#define NET_ACCEL_MSG_MAGIC (0x85465479)
++
++/*! We talk version 0.010 of the interdomain protocol */
++#define NET_ACCEL_MSG_VERSION (0x00001000)
++
++/*! Shared memory portion of inter-domain FIFO */
++struct net_accel_msg_queue {
++ u32 read;
++ u32 write;
++};
+
-+ device_unregister(bus_entry->bus->bridge);
-+ pci_remove_bus(bus_entry->bus);
+
-+ kfree(bus_entry);
-+ }
-+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/pcifront/xenbus.c linux-2.6.18-xen.hg/drivers/xen/pcifront/xenbus.c
---- linux-2.6.18/drivers/xen/pcifront/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/pcifront/xenbus.c 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,296 @@
+/*
-+ * PCI Frontend Xenbus Setup - handles setup with backend (imports page/evtchn)
++ * The aflags in the following structure is used as follows:
+ *
-+ * Author: Ryan Wilson <hap9@epoch.ncsc.mil>
++ * - each bit is set when one of the corresponding variables is
++ * changed by either end.
++ *
++ * - the end that has made the change then forwards an IRQ to the
++ * other
++ *
++ * - the IRQ handler deals with these bits either on the fast path, or
++ * for less common changes, by jumping onto the slow path.
++ *
++ * - once it has seen a change, it clears the relevant bit.
++ *
++ * aflags is accessed atomically using clear_bit, test_bit,
++ * test_and_set_bit etc
+ */
-+#include <linux/module.h>
-+#include <linux/init.h>
-+#include <linux/mm.h>
-+#include <xen/xenbus.h>
-+#include <xen/gnttab.h>
-+#include "pcifront.h"
-+
-+#define INVALID_GRANT_REF (0)
-+#define INVALID_EVTCHN (-1)
+
-+static struct pcifront_device *alloc_pdev(struct xenbus_device *xdev)
-+{
-+ struct pcifront_device *pdev;
++/*
++ * The following used to signify to the other domain when the queue
++ * they want to use is full, and when it is no longer full. Could be
++ * compressed to use fewer bits but done this way for simplicity and
++ * clarity
++ */
+
-+ pdev = kmalloc(sizeof(struct pcifront_device), GFP_KERNEL);
-+ if (pdev == NULL)
-+ goto out;
++/* "dom0->domU queue" is full */
++#define NET_ACCEL_MSG_AFLAGS_QUEUE0FULL 0x1
++#define NET_ACCEL_MSG_AFLAGS_QUEUE0FULL_B 0
++/* "dom0->domU queue" is not full */
++#define NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL 0x2
++#define NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL_B 1
++/* "domU->dom0 queue" is full */
++#define NET_ACCEL_MSG_AFLAGS_QUEUEUFULL 0x4
++#define NET_ACCEL_MSG_AFLAGS_QUEUEUFULL_B 2
++/* "domU->dom0 queue" is not full */
++#define NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL 0x8
++#define NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL_B 3
++/* dom0 -> domU net_dev up/down events */
++#define NET_ACCEL_MSG_AFLAGS_NETUPDOWN 0x10
++#define NET_ACCEL_MSG_AFLAGS_NETUPDOWN_B 4
+
-+ pdev->sh_info =
-+ (struct xen_pci_sharedinfo *)__get_free_page(GFP_KERNEL);
-+ if (pdev->sh_info == NULL) {
-+ kfree(pdev);
-+ pdev = NULL;
-+ goto out;
-+ }
-+ pdev->sh_info->flags = 0;
++/*
++ * Masks used to test if there are any messages for domU and dom0
++ * respectively
++ */
++#define NET_ACCEL_MSG_AFLAGS_TO_DOMU_MASK \
++ (NET_ACCEL_MSG_AFLAGS_QUEUE0FULL | \
++ NET_ACCEL_MSG_AFLAGS_QUEUEUNOTFULL | \
++ NET_ACCEL_MSG_AFLAGS_NETUPDOWN)
++#define NET_ACCEL_MSG_AFLAGS_TO_DOM0_MASK \
++ (NET_ACCEL_MSG_AFLAGS_QUEUE0NOTFULL | \
++ NET_ACCEL_MSG_AFLAGS_QUEUEUFULL)
+
-+ xdev->dev.driver_data = pdev;
-+ pdev->xdev = xdev;
++/*! The shared data structure used for inter-VM communication. */
++struct net_accel_shared_page {
++ /*! Sanity check */
++ u32 magic;
++ /*! Used by host/Dom0 */
++ struct net_accel_msg_queue queue0;
++ /*! Used by guest/DomU */
++ struct net_accel_msg_queue queue1;
++ /*! Atomic flags, used to communicate simple state changes */
++ u32 aflags;
++ /*! State of net_dev used for acceleration */
++ u32 net_dev_up;
++};
+
-+ INIT_LIST_HEAD(&pdev->root_buses);
+
-+ spin_lock_init(&pdev->dev_lock);
-+ spin_lock_init(&pdev->sh_info_lock);
++enum net_accel_hw_type {
++ /*! Not a virtualisable NIC: use slow path. */
++ NET_ACCEL_MSG_HWTYPE_NONE = 0,
++ /*! NIC is Falcon-based */
++ NET_ACCEL_MSG_HWTYPE_FALCON_A = 1,
++ NET_ACCEL_MSG_HWTYPE_FALCON_B = 2,
++};
+
-+ pdev->evtchn = INVALID_EVTCHN;
-+ pdev->gnt_ref = INVALID_GRANT_REF;
++/*! The maximum number of pages used by an event queue. */
++#define EF_HW_FALCON_EVQ_PAGES 8
+
-+ dev_dbg(&xdev->dev, "Allocated pdev @ 0x%p pdev->sh_info @ 0x%p\n",
-+ pdev, pdev->sh_info);
-+ out:
-+ return pdev;
-+}
++struct net_accel_hw_falcon_b {
++ /* VI */
++ /*! Grant for Tx DMA Q */
++ u32 txdmaq_gnt;
++ /*! Grant for Rx DMA Q */
++ u32 rxdmaq_gnt;
++ /*! Machine frame number for Tx/Rx doorbell page */
++ u32 doorbell_mfn;
++ /*! Grant for Tx/Rx doorbell page */
++ u32 doorbell_gnt;
+
-+static void free_pdev(struct pcifront_device *pdev)
-+{
-+ dev_dbg(&pdev->xdev->dev, "freeing pdev @ 0x%p\n", pdev);
++ /* Event Q */
++ /*! Grants for the pages of the EVQ */
++ u32 evq_mem_gnts[EF_HW_FALCON_EVQ_PAGES];
++ u32 evq_offs;
++ /*! log2(pages in event Q) */
++ u32 evq_order;
++ /*! Capacity in events */
++ u32 evq_capacity;
++ /*! Eventq pointer register physical address */
++ u32 evq_rptr;
++ /*! Interface instance */
++ u32 instance;
++ /*! Capacity of RX queue */
++ u32 rx_capacity;
++ /*! Capacity of TX queue */
++ u32 tx_capacity;
++
++ /* NIC */
++ s32 nic_arch;
++ s32 nic_revision;
++ u8 nic_variant;
++};
++
++struct net_accel_hw_falcon_a {
++ struct net_accel_hw_falcon_b common;
++ u32 evq_rptr_gnt;
++};
++
++
++/*! Description of the hardware that the DomU is being given. */
++struct net_accel_msg_hw {
++ u32 type; /*!< Hardware type */
++ union {
++ struct net_accel_hw_falcon_a falcon_a;
++ struct net_accel_hw_falcon_b falcon_b;
++ } resources;
++};
+
-+ pcifront_free_roots(pdev);
++/*! Start-of-day handshake message. Dom0 fills in its version and
++ * sends, DomU checks, inserts its version and replies
++ */
++struct net_accel_msg_hello {
++ /*! Sender's version (set by each side in turn) */
++ u32 version;
++ /*! max pages allocated/allowed for buffers */
++ u32 max_pages;
++};
+
-+ if (pdev->evtchn != INVALID_EVTCHN)
-+ xenbus_free_evtchn(pdev->xdev, pdev->evtchn);
++/*! Maximum number of page requests that can fit in a message. */
++#define NET_ACCEL_MSG_MAX_PAGE_REQ (8)
+
-+ if (pdev->gnt_ref != INVALID_GRANT_REF)
-+ gnttab_end_foreign_access(pdev->gnt_ref,
-+ (unsigned long)pdev->sh_info);
++/*! Request for NIC buffers. DomU fils out pages and grants (and
++ * optionally) reqid, dom0 fills out buf and sends reply
++ */
++struct net_accel_msg_map_buffers {
++ u32 reqid; /*!< Optional request ID */
++ u32 pages; /*!< Number of pages to map */
++ u32 grants[NET_ACCEL_MSG_MAX_PAGE_REQ]; /*!< Grant ids to map */
++ u32 buf; /*!< NIC buffer address of pages obtained */
++};
+
-+ pdev->xdev->dev.driver_data = NULL;
++/*! Notification of a change to local mac address, used to filter
++ locally destined packets off the fast path */
++struct net_accel_msg_localmac {
++ u32 flags; /*!< Should this be added or removed? */
++ u8 mac[ETH_ALEN]; /*!< The mac address to filter onto slow path */
++};
+
-+ kfree(pdev);
-+}
++struct net_accel_msg_fastpath {
++ u32 flags; /*!< Should this be added or removed? */
++ u8 mac[ETH_ALEN];/*!< The mac address to filter onto fast path */
++ u16 port; /*!< The port of the connection */
++ u32 ip; /*!< The IP address of the connection */
++ u8 proto; /*!< The protocol of connection (TCP/UDP) */
++};
+
-+static int pcifront_publish_info(struct pcifront_device *pdev)
-+{
-+ int err = 0;
-+ struct xenbus_transaction trans;
++/*! Values for struct ef_msg_localmac/fastpath.flags */
++#define NET_ACCEL_MSG_ADD 0x1
++#define NET_ACCEL_MSG_REMOVE 0x2
+
-+ err = xenbus_grant_ring(pdev->xdev, virt_to_mfn(pdev->sh_info));
-+ if (err < 0)
-+ goto out;
++/*! Overall message structure */
++struct net_accel_msg {
++ /*! ID specifying type of messge */
++ u32 id;
++ union {
++ /*! handshake */
++ struct net_accel_msg_hello hello;
++ /*! hardware description */
++ struct net_accel_msg_hw hw;
++ /*! buffer map request */
++ struct net_accel_msg_map_buffers mapbufs;
++ /*! mac address of a local interface */
++ struct net_accel_msg_localmac localmac;
++ /*! address of a new fastpath connection */
++ struct net_accel_msg_fastpath fastpath;
++ /*! make the message a fixed size */
++ u8 pad[128 - sizeof(u32)];
++ } u;
++};
+
-+ pdev->gnt_ref = err;
+
-+ err = xenbus_alloc_evtchn(pdev->xdev, &pdev->evtchn);
-+ if (err)
-+ goto out;
++#define NET_ACCEL_MSG_HW_TO_MSG(_u) container_of(_u, struct net_accel_msg, u.hw)
+
-+ do_publish:
-+ err = xenbus_transaction_start(&trans);
-+ if (err) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error writing configuration for backend "
-+ "(start transaction)");
-+ goto out;
-+ }
++/*! Inter-domain message FIFO */
++typedef struct {
++ struct net_accel_msg *fifo;
++ u32 fifo_mask;
++ u32 *fifo_rd_i;
++ u32 *fifo_wr_i;
++ spinlock_t lock;
++ u32 is_locked; /* Debug flag */
++} sh_msg_fifo2;
+
-+ err = xenbus_printf(trans, pdev->xdev->nodename,
-+ "pci-op-ref", "%u", pdev->gnt_ref);
-+ if (!err)
-+ err = xenbus_printf(trans, pdev->xdev->nodename,
-+ "event-channel", "%u", pdev->evtchn);
-+ if (!err)
-+ err = xenbus_printf(trans, pdev->xdev->nodename,
-+ "magic", XEN_PCI_MAGIC);
+
-+ if (err) {
-+ xenbus_transaction_end(trans, 1);
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error writing configuration for backend");
-+ goto out;
-+ } else {
-+ err = xenbus_transaction_end(trans, 0);
-+ if (err == -EAGAIN)
-+ goto do_publish;
-+ else if (err) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error completing transaction "
-+ "for backend");
-+ goto out;
-+ }
-+ }
++#define NET_ACCEL_MSG_OFFSET_MASK PAGE_MASK
+
-+ xenbus_switch_state(pdev->xdev, XenbusStateInitialised);
++/* Modifiers */
++#define NET_ACCEL_MSG_REPLY (0x80000000)
++#define NET_ACCEL_MSG_ERROR (0x40000000)
+
-+ dev_dbg(&pdev->xdev->dev, "publishing successful!\n");
++/* Dom0 -> DomU and reply. Handshake/version check. */
++#define NET_ACCEL_MSG_HELLO (0x00000001)
++/* Dom0 -> DomU : hardware setup (VI info.) */
++#define NET_ACCEL_MSG_SETHW (0x00000002)
++/*
++ * Dom0 -> DomU. Notification of a local mac to add/remove from slow
++ * path filter
++ */
++#define NET_ACCEL_MSG_LOCALMAC (0x00000003)
++/*
++ * DomU -> Dom0 and reply. Request for buffer table entries for
++ * preallocated pages.
++ */
++#define NET_ACCEL_MSG_MAPBUF (0x00000004)
++/*
++ * Dom0 -> DomU. Notification of a local mac to add/remove from fast
++ * path filter
++ */
++#define NET_ACCEL_MSG_FASTPATH (0x00000005)
+
-+ out:
-+ return err;
++/*! Initialise a message and set the type
++ * \param message : the message
++ * \param code : the message type
++ */
++static inline void net_accel_msg_init(struct net_accel_msg *msg, int code) {
++ msg->id = (u32)code;
+}
+
-+static int pcifront_try_connect(struct pcifront_device *pdev)
-+{
-+ int err = -EFAULT;
-+ int i, num_roots, len;
-+ char str[64];
-+ unsigned int domain, bus;
++/*! initialise a shared page structure
++ * \param shared_page : mapped memory in which the structure resides
++ * \param len : size of the message FIFO area that follows
++ * \param up : initial up/down state of netdev
++ * \return 0 or an error code
++ */
++extern int net_accel_msg_init_page(void *shared_page, int len, int up);
+
-+ spin_lock(&pdev->dev_lock);
++/*! initialise a message queue
++ * \param queue : the message FIFO to initialise
++ * \param indices : the read and write indices in shared memory
++ * \param base : the start of the memory area for the FIFO
++ * \param size : the size of the FIFO in bytes
++ */
++extern void net_accel_msg_init_queue(sh_msg_fifo2 *queue,
++ struct net_accel_msg_queue *indices,
++ struct net_accel_msg *base, int size);
+
-+ /* Only connect once */
-+ if (xenbus_read_driver_state(pdev->xdev->nodename) !=
-+ XenbusStateInitialised)
-+ goto out;
++/* Notify after a batch of messages have been sent */
++extern void net_accel_msg_notify(int irq);
+
-+ err = pcifront_connect(pdev);
-+ if (err) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error connecting PCI Frontend");
-+ goto out;
-+ }
++/*! Send a message on the specified FIFO. The message is copied to the
++ * current slot of the FIFO.
++ * \param sp : pointer to shared page
++ * \param q : pointer to message FIFO to use
++ * \param msg : pointer to message
++ * \return 0 on success, -errno on
++ */
++extern int net_accel_msg_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *q,
++ struct net_accel_msg *msg);
++extern int net_accel_msg_reply(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *q,
++ struct net_accel_msg *msg);
++
++/*! As net_accel_msg_send but also posts a notification to the far end. */
++extern int net_accel_msg_send_notify(struct net_accel_shared_page *sp,
++ int irq, sh_msg_fifo2 *q,
++ struct net_accel_msg *msg);
++/*! As net_accel_msg_send but also posts a notification to the far end. */
++extern int net_accel_msg_reply_notify(struct net_accel_shared_page *sp,
++ int irq, sh_msg_fifo2 *q,
++ struct net_accel_msg *msg);
++
++/*! Receive a message on the specified FIFO. Returns 0 on success,
++ * -errno on failure.
++ */
++extern int net_accel_msg_recv(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *q,
++ struct net_accel_msg *msg);
++
++/*! Look at a received message, if any, so a decision can be made
++ * about whether to read it now or not. Cookie is a bit of debug
++ * which is set here and checked when passed to
++ * net_accel_msg_recv_next()
++ */
++extern int net_accel_msg_peek(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ struct net_accel_msg *msg, int *cookie);
++/*! Move the queue onto the next element, used after finished with a
++ * peeked msg
++ */
++extern int net_accel_msg_recv_next(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue, int cookie);
++
++/*! Start sending a message without copying. returns a pointer to a
++ * message that will be filled out in place. The queue is locked
++ * until the message is sent.
++ */
++extern
++struct net_accel_msg *net_accel_msg_start_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags);
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend,
-+ "root_num", "%d", &num_roots);
-+ if (err == -ENOENT) {
-+ xenbus_dev_error(pdev->xdev, err,
-+ "No PCI Roots found, trying 0000:00");
-+ err = pcifront_scan_root(pdev, 0, 0);
-+ num_roots = 0;
-+ } else if (err != 1) {
-+ if (err == 0)
-+ err = -EINVAL;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error reading number of PCI roots");
-+ goto out;
-+ }
+
-+ for (i = 0; i < num_roots; i++) {
-+ len = snprintf(str, sizeof(str), "root-%d", i);
-+ if (unlikely(len >= (sizeof(str) - 1))) {
-+ err = -ENOMEM;
-+ goto out;
-+ }
++/*! Complete the sending of a message started with
++ * net_accel_msg_start_send. The message is implicit since the queue
++ * was locked by _start
++ */
++extern void net_accel_msg_complete_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags);
+
-+ err = xenbus_scanf(XBT_NIL, pdev->xdev->otherend, str,
-+ "%x:%x", &domain, &bus);
-+ if (err != 2) {
-+ if (err >= 0)
-+ err = -EINVAL;
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error reading PCI root %d", i);
-+ goto out;
-+ }
++/*! As net_accel_msg_complete_send but does the notify. */
++extern void net_accel_msg_complete_send_notify(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue,
++ unsigned long *flags, int irq);
+
-+ err = pcifront_scan_root(pdev, domain, bus);
-+ if (err) {
-+ xenbus_dev_fatal(pdev->xdev, err,
-+ "Error scanning PCI root %04x:%02x",
-+ domain, bus);
-+ goto out;
-+ }
-+ }
++/*! Lock the queue so that multiple "_locked" functions can be called
++ * without the queue being modified by others
++ */
++static inline
++void net_accel_msg_lock_queue(sh_msg_fifo2 *queue, unsigned long *flags)
++{
++ spin_lock_irqsave(&queue->lock, (*flags));
++ rmb();
++ BUG_ON(queue->is_locked);
++ queue->is_locked = 1;
++}
+
-+ err = xenbus_switch_state(pdev->xdev, XenbusStateConnected);
-+ if (err)
-+ goto out;
++/*! Unlock the queue */
++static inline
++void net_accel_msg_unlock_queue(sh_msg_fifo2 *queue, unsigned long *flags)
++{
++ BUG_ON(!queue->is_locked);
++ queue->is_locked = 0;
++ wmb();
++ spin_unlock_irqrestore(&queue->lock, (*flags));
++}
+
-+ out:
-+ spin_unlock(&pdev->dev_lock);
-+ return err;
++/*! Give up without sending a message that was started with
++ * net_accel_msg_start_send()
++ */
++static inline
++void net_accel_msg_abort_send(struct net_accel_shared_page *sp,
++ sh_msg_fifo2 *queue, unsigned long *flags)
++{
++ net_accel_msg_unlock_queue(queue, flags);
+}
+
-+static int pcifront_try_disconnect(struct pcifront_device *pdev)
++/*! Test the queue to ensure there is sufficient space */
++static inline
++int net_accel_msg_check_space(sh_msg_fifo2 *queue, unsigned space)
+{
-+ int err = 0;
-+ enum xenbus_state prev_state;
++ return sh_fifo2_space(queue) >= space;
++}
+
-+ spin_lock(&pdev->dev_lock);
++#endif /* NET_ACCEL_MSG_IFACE_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_shared_fifo.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_shared_fifo.h 2008-05-19 00:33:48.742955720 +0300
+@@ -0,0 +1,127 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
+
-+ prev_state = xenbus_read_driver_state(pdev->xdev->nodename);
++#ifndef NET_ACCEL_SHARED_FIFO_H
++#define NET_ACCEL_SHARED_FIFO_H
+
-+ if (prev_state < XenbusStateClosing)
-+ err = xenbus_switch_state(pdev->xdev, XenbusStateClosing);
++/*
++ * This is based on fifo.h, but handles sharing between address spaces
++ * that don't trust each other, by splitting out the read and write
++ * indices. This costs at least one pointer indirection more than the
++ * vanilla version per access.
++ */
+
-+ if (!err && prev_state == XenbusStateConnected)
-+ pcifront_disconnect(pdev);
++typedef struct {
++ char* fifo;
++ unsigned fifo_mask;
++ unsigned *fifo_rd_i;
++ unsigned *fifo_wr_i;
++} sh_byte_fifo2;
+
-+ spin_unlock(&pdev->dev_lock);
++#define SH_FIFO2_M(f, x) ((x) & ((f)->fifo_mask))
+
-+ return err;
++static inline unsigned log2_ge(unsigned long n, unsigned min_order) {
++ unsigned order = min_order;
++ while((1ul << order) < n) ++order;
++ return order;
+}
+
-+static void pcifront_backend_changed(struct xenbus_device *xdev,
-+ enum xenbus_state be_state)
-+{
-+ struct pcifront_device *pdev = xdev->dev.driver_data;
-+
-+ switch (be_state) {
-+ case XenbusStateClosing:
-+ dev_warn(&xdev->dev, "backend going away!\n");
-+ pcifront_try_disconnect(pdev);
-+ break;
++static inline unsigned long pow2(unsigned order) {
++ return (1ul << order);
++}
+
-+ case XenbusStateUnknown:
-+ case XenbusStateClosed:
-+ dev_warn(&xdev->dev, "backend went away!\n");
-+ pcifront_try_disconnect(pdev);
++#define is_pow2(x) (pow2(log2_ge((x), 0)) == (x))
+
-+ device_unregister(&pdev->xdev->dev);
-+ break;
++#define sh_fifo2_valid(f) ((f) && (f)->fifo && (f)->fifo_mask > 0 && \
++ is_pow2((f)->fifo_mask+1u))
+
-+ case XenbusStateConnected:
-+ pcifront_try_connect(pdev);
-+ break;
++#define sh_fifo2_init(f, cap, _rptr, _wptr) \
++ do { \
++ BUG_ON(!is_pow2((cap) + 1)); \
++ (f)->fifo_rd_i = _rptr; \
++ (f)->fifo_wr_i = _wptr; \
++ *(f)->fifo_rd_i = *(f)->fifo_wr_i = 0u; \
++ (f)->fifo_mask = (cap); \
++ } while(0)
+
-+ default:
-+ break;
-+ }
-+}
++#define sh_fifo2_num(f) SH_FIFO2_M((f),*(f)->fifo_wr_i - *(f)->fifo_rd_i)
++#define sh_fifo2_space(f) SH_FIFO2_M((f),*(f)->fifo_rd_i - *(f)->fifo_wr_i-1u)
++#define sh_fifo2_is_empty(f) (sh_fifo2_num(f)==0)
++#define sh_fifo2_not_empty(f) (sh_fifo2_num(f)!=0)
++#define sh_fifo2_is_full(f) (sh_fifo2_space(f)==0u)
++#define sh_fifo2_not_full(f) (sh_fifo2_space(f)!=0u)
++#define sh_fifo2_buf_size(f) ((f)->fifo_mask + 1u)
++#define sh_fifo2_capacity(f) ((f)->fifo_mask)
++#define sh_fifo2_end(f) ((f)->fifo + sh_fifo2_buf_size(f))
++#define sh_fifo2_not_half_full(f) (sh_fifo2_space(f) > (sh_fifo2_capacity(f) >> 1))
+
-+static int pcifront_xenbus_probe(struct xenbus_device *xdev,
-+ const struct xenbus_device_id *id)
-+{
-+ int err = 0;
-+ struct pcifront_device *pdev = alloc_pdev(xdev);
++#define sh_fifo2_peek(f) ((f)->fifo[SH_FIFO2_M((f), *(f)->fifo_rd_i)])
++#define sh_fifo2_peekp(f) ((f)->fifo + SH_FIFO2_M((f), *(f)->fifo_rd_i))
++#define sh_fifo2_poke(f) ((f)->fifo[SH_FIFO2_M((f), *(f)->fifo_wr_i)])
++#define sh_fifo2_pokep(f) ((f)->fifo + SH_FIFO2_M((f), *(f)->fifo_wr_i))
++#define sh_fifo2_peek_i(f,i) ((f)->fifo[SH_FIFO2_M((f), *(f)->fifo_rd_i+(i))])
++#define sh_fifo2_poke_i(f,i) ((f)->fifo[SH_FIFO2_M((f), *(f)->fifo_wr_i+(i))])
+
-+ if (pdev == NULL) {
-+ err = -ENOMEM;
-+ xenbus_dev_fatal(xdev, err,
-+ "Error allocating pcifront_device struct");
-+ goto out;
-+ }
++#define sh_fifo2_rd_next(f) \
++ do {*(f)->fifo_rd_i = *(f)->fifo_rd_i + 1u;} while(0)
++#define sh_fifo2_wr_next(f) \
++ do {*(f)->fifo_wr_i = *(f)->fifo_wr_i + 1u;} while(0)
++#define sh_fifo2_rd_adv(f, n) \
++ do {*(f)->fifo_rd_i = *(f)->fifo_rd_i + (n);} while(0)
++#define sh_fifo2_wr_adv(f, n) \
++ do {*(f)->fifo_wr_i = *(f)->fifo_wr_i + (n);} while(0)
+
-+ err = pcifront_publish_info(pdev);
++#define sh_fifo2_put(f, v) \
++ do {sh_fifo2_poke(f) = (v); wmb(); sh_fifo2_wr_next(f);} while(0)
+
-+ out:
-+ return err;
-+}
++#define sh_fifo2_get(f, pv) \
++ do {*(pv) = sh_fifo2_peek(f); mb(); sh_fifo2_rd_next(f);} while(0)
+
-+static int pcifront_xenbus_remove(struct xenbus_device *xdev)
++static inline unsigned sh_fifo2_contig_num(sh_byte_fifo2 *f)
+{
-+ if (xdev->dev.driver_data)
-+ free_pdev(xdev->dev.driver_data);
++ unsigned fifo_wr_i = SH_FIFO2_M(f, *f->fifo_wr_i);
++ unsigned fifo_rd_i = SH_FIFO2_M(f, *f->fifo_rd_i);
+
-+ return 0;
++ return (fifo_wr_i >= fifo_rd_i)
++ ? fifo_wr_i - fifo_rd_i
++ : f->fifo_mask + 1u - *(f)->fifo_rd_i;
+}
+
-+static struct xenbus_device_id xenpci_ids[] = {
-+ {"pci"},
-+ {{0}},
-+};
-+MODULE_ALIAS("xen:pci");
-+
-+static struct xenbus_driver xenbus_pcifront_driver = {
-+ .name = "pcifront",
-+ .owner = THIS_MODULE,
-+ .ids = xenpci_ids,
-+ .probe = pcifront_xenbus_probe,
-+ .remove = pcifront_xenbus_remove,
-+ .otherend_changed = pcifront_backend_changed,
-+};
-+
-+static int __init pcifront_init(void)
++static inline unsigned sh_fifo2_contig_space(sh_byte_fifo2 *f)
+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ unsigned fifo_wr_i = SH_FIFO2_M(f, *f->fifo_wr_i);
++ unsigned fifo_rd_i = SH_FIFO2_M(f, *f->fifo_rd_i);
+
-+ return xenbus_register_frontend(&xenbus_pcifront_driver);
++ return (fifo_rd_i > fifo_wr_i)
++ ? fifo_rd_i - fifo_wr_i - 1
++ : (f->fifo_mask + 1u - fifo_wr_i
++ /*
++ * The last byte can't be used if the read pointer
++ * is at zero.
++ */
++ - (fifo_rd_i==0));
+}
+
-+/* Initialize after the Xen PCI Frontend Stub is initialized */
-+subsys_initcall(pcifront_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/privcmd/compat_privcmd.c linux-2.6.18-xen.hg/drivers/xen/privcmd/compat_privcmd.c
---- linux-2.6.18/drivers/xen/privcmd/compat_privcmd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/privcmd/compat_privcmd.c 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,73 @@
-+/*
-+ * This program is free software; you can redistribute it and/or modify
-+ * it under the terms of the GNU General Public License as published by
-+ * the Free Software Foundation; either version 2 of the License, or
-+ * (at your option) any later version.
++
++#endif /* NET_ACCEL_SHARED_FIFO_H */
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_util.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_util.c 2008-05-19 00:33:48.746955951 +0300
+@@ -0,0 +1,333 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
-+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-+ *
-+ * Copyright (C) IBM Corp. 2006
-+ *
-+ * Authors: Jimi Xenidis <jimix@watson.ibm.com>
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
+ */
+
-+#include <linux/config.h>
-+#include <linux/compat.h>
-+#include <linux/ioctl.h>
-+#include <linux/syscalls.h>
-+#include <asm/hypervisor.h>
-+#include <asm/uaccess.h>
-+#include <xen/public/privcmd.h>
-+#include <xen/compat_ioctl.h>
-+
-+int privcmd_ioctl_32(int fd, unsigned int cmd, unsigned long arg)
-+{
-+ int ret;
++#include <linux/if_ether.h>
++#include <asm/io.h>
++#include <asm/pgtable.h>
++#include <asm/hypercall.h>
++#include <xen/xenbus.h>
++#include <xen/driver_util.h>
++#include <xen/gnttab.h>
+
-+ switch (cmd) {
-+ case IOCTL_PRIVCMD_MMAP_32: {
-+ struct privcmd_mmap *p;
-+ struct privcmd_mmap_32 *p32;
-+ struct privcmd_mmap_32 n32;
++#include "accel_util.h"
+
-+ p32 = compat_ptr(arg);
-+ p = compat_alloc_user_space(sizeof(*p));
-+ if (copy_from_user(&n32, p32, sizeof(n32)) ||
-+ put_user(n32.num, &p->num) ||
-+ put_user(n32.dom, &p->dom) ||
-+ put_user(compat_ptr(n32.entry), &p->entry))
-+ return -EFAULT;
-+
-+ ret = sys_ioctl(fd, IOCTL_PRIVCMD_MMAP, (unsigned long)p);
-+ }
-+ break;
-+ case IOCTL_PRIVCMD_MMAPBATCH_32: {
-+ struct privcmd_mmapbatch *p;
-+ struct privcmd_mmapbatch_32 *p32;
-+ struct privcmd_mmapbatch_32 n32;
++#ifdef EFX_GCOV
++#include "gcov.h"
+
-+ p32 = compat_ptr(arg);
-+ p = compat_alloc_user_space(sizeof(*p));
-+ if (copy_from_user(&n32, p32, sizeof(n32)) ||
-+ put_user(n32.num, &p->num) ||
-+ put_user(n32.dom, &p->dom) ||
-+ put_user(n32.addr, &p->addr) ||
-+ put_user(compat_ptr(n32.arr), &p->arr))
-+ return -EFAULT;
-+
-+ ret = sys_ioctl(fd, IOCTL_PRIVCMD_MMAPBATCH, (unsigned long)p);
-+ }
-+ break;
-+ default:
-+ ret = -EINVAL;
-+ break;
-+ }
-+ return ret;
++static int __init net_accel_init(void)
++{
++ gcov_provider_init(THIS_MODULE);
++ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/privcmd/Makefile linux-2.6.18-xen.hg/drivers/xen/privcmd/Makefile
---- linux-2.6.18/drivers/xen/privcmd/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/privcmd/Makefile 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,3 @@
++module_init(net_accel_init);
+
-+obj-y += privcmd.o
-+obj-$(CONFIG_COMPAT) += compat_privcmd.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/privcmd/privcmd.c linux-2.6.18-xen.hg/drivers/xen/privcmd/privcmd.c
---- linux-2.6.18/drivers/xen/privcmd/privcmd.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/privcmd/privcmd.c 2007-12-23 11:15:34.071269815 +0100
-@@ -0,0 +1,286 @@
-+/******************************************************************************
-+ * privcmd.c
-+ *
-+ * Interface to privileged domain-0 commands.
-+ *
-+ * Copyright (c) 2002-2004, K A Fraser, B Dragovic
-+ */
++static void __exit net_accel_exit(void)
++{
++ gcov_provider_fini(THIS_MODULE);
++}
++module_exit(net_accel_exit);
++#endif
+
-+#include <linux/kernel.h>
-+#include <linux/sched.h>
-+#include <linux/slab.h>
-+#include <linux/string.h>
-+#include <linux/errno.h>
-+#include <linux/mm.h>
-+#include <linux/mman.h>
-+#include <linux/swap.h>
-+#include <linux/smp_lock.h>
-+#include <linux/highmem.h>
-+#include <linux/pagemap.h>
-+#include <linux/seq_file.h>
-+#include <asm/hypervisor.h>
++/* Shutdown remote domain that is misbehaving */
++int net_accel_shutdown_remote(int domain)
++{
++ struct sched_remote_shutdown sched_shutdown = {
++ .domain_id = domain,
++ .reason = SHUTDOWN_crash
++ };
+
-+#include <asm/pgalloc.h>
-+#include <asm/pgtable.h>
-+#include <asm/uaccess.h>
-+#include <asm/tlb.h>
-+#include <asm/hypervisor.h>
-+#include <xen/public/privcmd.h>
-+#include <xen/interface/xen.h>
-+#include <xen/xen_proc.h>
-+#include <xen/features.h>
++ EPRINTK("Crashing domain %d\n", domain);
+
-+static struct proc_dir_entry *privcmd_intf;
-+static struct proc_dir_entry *capabilities_intf;
++ return HYPERVISOR_sched_op(SCHEDOP_remote_shutdown, &sched_shutdown);
++}
++EXPORT_SYMBOL(net_accel_shutdown_remote);
+
-+#ifndef HAVE_ARCH_PRIVCMD_MMAP
-+static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
-+#endif
+
-+static long privcmd_ioctl(struct file *file,
-+ unsigned int cmd, unsigned long data)
++/* Based on xenbus_backend_client.c:xenbus_map_ring() */
++static int net_accel_map_grant(struct xenbus_device *dev, int gnt_ref,
++ grant_handle_t *handle, void *vaddr,
++ u64 *dev_bus_addr, unsigned flags)
+{
-+ int ret = -ENOSYS;
-+ void __user *udata = (void __user *) data;
++ struct gnttab_map_grant_ref op;
++
++ gnttab_set_map_op(&op, (unsigned long)vaddr, flags,
++ gnt_ref, dev->otherend_id);
+
-+ switch (cmd) {
-+ case IOCTL_PRIVCMD_HYPERCALL: {
-+ privcmd_hypercall_t hypercall;
-+
-+ if (copy_from_user(&hypercall, udata, sizeof(hypercall)))
-+ return -EFAULT;
++ BUG_ON(HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1));
+
-+#if defined(__i386__)
-+ if (hypercall.op >= (PAGE_SIZE >> 5))
-+ break;
-+ __asm__ __volatile__ (
-+ "pushl %%ebx; pushl %%ecx; pushl %%edx; "
-+ "pushl %%esi; pushl %%edi; "
-+ "movl 8(%%eax),%%ebx ;"
-+ "movl 16(%%eax),%%ecx ;"
-+ "movl 24(%%eax),%%edx ;"
-+ "movl 32(%%eax),%%esi ;"
-+ "movl 40(%%eax),%%edi ;"
-+ "movl (%%eax),%%eax ;"
-+ "shll $5,%%eax ;"
-+ "addl $hypercall_page,%%eax ;"
-+ "call *%%eax ;"
-+ "popl %%edi; popl %%esi; popl %%edx; "
-+ "popl %%ecx; popl %%ebx"
-+ : "=a" (ret) : "0" (&hypercall) : "memory" );
-+#elif defined (__x86_64__)
-+ if (hypercall.op < (PAGE_SIZE >> 5)) {
-+ long ign1, ign2, ign3;
-+ __asm__ __volatile__ (
-+ "movq %8,%%r10; movq %9,%%r8;"
-+ "shll $5,%%eax ;"
-+ "addq $hypercall_page,%%rax ;"
-+ "call *%%rax"
-+ : "=a" (ret), "=D" (ign1),
-+ "=S" (ign2), "=d" (ign3)
-+ : "0" ((unsigned int)hypercall.op),
-+ "1" (hypercall.arg[0]),
-+ "2" (hypercall.arg[1]),
-+ "3" (hypercall.arg[2]),
-+ "g" (hypercall.arg[3]),
-+ "g" (hypercall.arg[4])
-+ : "r8", "r10", "memory" );
-+ }
-+#else
-+ ret = privcmd_hypercall(&hypercall);
-+#endif
++ if (op.status != GNTST_okay) {
++ xenbus_dev_error
++ (dev, op.status,
++ "failed mapping in shared page %d from domain %d\n",
++ gnt_ref, dev->otherend_id);
++ } else {
++ *handle = op.handle;
++ if (dev_bus_addr)
++ *dev_bus_addr = op.dev_bus_addr;
+ }
-+ break;
+
-+ case IOCTL_PRIVCMD_MMAP: {
-+ privcmd_mmap_t mmapcmd;
-+ privcmd_mmap_entry_t msg;
-+ privcmd_mmap_entry_t __user *p;
-+ struct mm_struct *mm = current->mm;
-+ struct vm_area_struct *vma;
-+ unsigned long va;
-+ int i, rc;
++ return op.status;
++}
+
-+ if (!is_initial_xendomain())
-+ return -EPERM;
+
-+ if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
-+ return -EFAULT;
++/* Based on xenbus_backend_client.c:xenbus_unmap_ring() */
++static int net_accel_unmap_grant(struct xenbus_device *dev,
++ grant_handle_t handle,
++ void *vaddr, u64 dev_bus_addr,
++ unsigned flags)
++{
++ struct gnttab_unmap_grant_ref op;
+
-+ p = mmapcmd.entry;
-+ if (copy_from_user(&msg, p, sizeof(msg)))
-+ return -EFAULT;
++ gnttab_set_unmap_op(&op, (unsigned long)vaddr, flags, handle);
++
++ if (dev_bus_addr)
++ op.dev_bus_addr = dev_bus_addr;
+
-+ down_write(&mm->mmap_sem);
++ BUG_ON(HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1));
+
-+ vma = find_vma(mm, msg.va);
-+ rc = -EINVAL;
-+ if (!vma || (msg.va != vma->vm_start) ||
-+ !privcmd_enforce_singleshot_mapping(vma))
-+ goto mmap_out;
++ if (op.status != GNTST_okay)
++ xenbus_dev_error(dev, op.status,
++ "failed unmapping page at handle %d error %d\n",
++ handle, op.status);
+
-+ va = vma->vm_start;
++ return op.status;
++}
+
-+ for (i = 0; i < mmapcmd.num; i++) {
-+ rc = -EFAULT;
-+ if (copy_from_user(&msg, p, sizeof(msg)))
-+ goto mmap_out;
+
-+ /* Do not allow range to wrap the address space. */
-+ rc = -EINVAL;
-+ if ((msg.npages > (LONG_MAX >> PAGE_SHIFT)) ||
-+ ((unsigned long)(msg.npages << PAGE_SHIFT) >= -va))
-+ goto mmap_out;
++int net_accel_map_device_page(struct xenbus_device *dev,
++ int gnt_ref, grant_handle_t *handle,
++ u64 *dev_bus_addr)
++{
++ return net_accel_map_grant(dev, gnt_ref, handle, 0, dev_bus_addr,
++ GNTMAP_device_map);
++}
++EXPORT_SYMBOL_GPL(net_accel_map_device_page);
+
-+ /* Range chunks must be contiguous in va space. */
-+ if ((msg.va != va) ||
-+ ((msg.va+(msg.npages<<PAGE_SHIFT)) > vma->vm_end))
-+ goto mmap_out;
++
++int net_accel_unmap_device_page(struct xenbus_device *dev,
++ grant_handle_t handle, u64 dev_bus_addr)
++{
++ return net_accel_unmap_grant(dev, handle, 0, dev_bus_addr,
++ GNTMAP_device_map);
++}
++EXPORT_SYMBOL_GPL(net_accel_unmap_device_page);
+
-+ if ((rc = direct_remap_pfn_range(
-+ vma,
-+ msg.va & PAGE_MASK,
-+ msg.mfn,
-+ msg.npages << PAGE_SHIFT,
-+ vma->vm_page_prot,
-+ mmapcmd.dom)) < 0)
-+ goto mmap_out;
+
-+ p++;
-+ va += msg.npages << PAGE_SHIFT;
-+ }
++struct net_accel_valloc_grant_mapping {
++ struct vm_struct *vm;
++ int pages;
++ grant_handle_t grant_handles[0];
++};
+
-+ rc = 0;
++/* Map a series of grants into a contiguous virtual area */
++static void *net_accel_map_grants_valloc(struct xenbus_device *dev,
++ unsigned *grants, int npages,
++ unsigned flags, void **priv)
++{
++ struct net_accel_valloc_grant_mapping *map;
++ struct vm_struct *vm;
++ void *addr;
++ int i, j, rc;
+
-+ mmap_out:
-+ up_write(&mm->mmap_sem);
-+ ret = rc;
++ vm = alloc_vm_area(PAGE_SIZE * npages);
++ if (vm == NULL) {
++ EPRINTK("No memory from alloc_vm_area.\n");
++ return NULL;
+ }
-+ break;
-+
-+ case IOCTL_PRIVCMD_MMAPBATCH: {
-+ privcmd_mmapbatch_t m;
-+ struct mm_struct *mm = current->mm;
-+ struct vm_area_struct *vma;
-+ xen_pfn_t __user *p;
-+ unsigned long addr, mfn, nr_pages;
-+ int i;
-+
-+ if (!is_initial_xendomain())
-+ return -EPERM;
++ /*
++ * Get a structure in which we will record all the info needed
++ * to undo the mapping.
++ */
++ map = kzalloc(sizeof(struct net_accel_valloc_grant_mapping) +
++ npages * sizeof(grant_handle_t), GFP_KERNEL);
++ if (map == NULL) {
++ EPRINTK("No memory for net_accel_valloc_grant_mapping\n");
++ free_vm_area(vm);
++ return NULL;
++ }
++ map->vm = vm;
++ map->pages = npages;
+
-+ if (copy_from_user(&m, udata, sizeof(m)))
-+ return -EFAULT;
++ /* Do the actual mapping */
++ addr = vm->addr;
++ for (i = 0; i < npages; i++) {
++ rc = net_accel_map_grant(dev, grants[i], map->grant_handles + i,
++ addr, NULL, flags);
++ if (rc != 0)
++ goto undo;
++ addr = (void*)((unsigned long)addr + PAGE_SIZE);
++ }
+
-+ nr_pages = m.num;
-+ if ((m.num <= 0) || (nr_pages > (LONG_MAX >> PAGE_SHIFT)))
-+ return -EINVAL;
++ if (priv)
++ *priv = (void *)map;
++ else
++ kfree(map);
+
-+ down_write(&mm->mmap_sem);
++ return vm->addr;
+
-+ vma = find_vma(mm, m.addr);
-+ if (!vma ||
-+ (m.addr != vma->vm_start) ||
-+ ((m.addr + (nr_pages << PAGE_SHIFT)) != vma->vm_end) ||
-+ !privcmd_enforce_singleshot_mapping(vma)) {
-+ up_write(&mm->mmap_sem);
-+ return -EINVAL;
-+ }
++ undo:
++ EPRINTK("Aborting contig map due to single map failure %d (%d of %d)\n",
++ rc, i+1, npages);
++ for (j = 0; j < i; j++) {
++ addr = (void*)((unsigned long)vm->addr + (j * PAGE_SIZE));
++ net_accel_unmap_grant(dev, map->grant_handles[j], addr, 0,
++ flags);
++ }
++ free_vm_area(vm);
++ kfree(map);
++ return NULL;
++}
+
-+ p = m.arr;
-+ addr = m.addr;
-+ for (i = 0; i < nr_pages; i++, addr += PAGE_SIZE, p++) {
-+ if (get_user(mfn, p)) {
-+ up_write(&mm->mmap_sem);
-+ return -EFAULT;
-+ }
++/* Undo the result of the mapping */
++static void net_accel_unmap_grants_vfree(struct xenbus_device *dev,
++ unsigned flags, void *priv)
++{
++ struct net_accel_valloc_grant_mapping *map =
++ (struct net_accel_valloc_grant_mapping *)priv;
+
-+ ret = direct_remap_pfn_range(vma, addr & PAGE_MASK,
-+ mfn, PAGE_SIZE,
-+ vma->vm_page_prot, m.dom);
-+ if (ret < 0)
-+ put_user(0xF0000000 | mfn, p);
-+ }
++ void *addr = map->vm->addr;
++ int npages = map->pages;
++ int i;
+
-+ up_write(&mm->mmap_sem);
-+ ret = 0;
++ for (i = 0; i < npages; i++) {
++ net_accel_unmap_grant(dev, map->grant_handles[i], addr, 0,
++ flags);
++ addr = (void*)((unsigned long)addr + PAGE_SIZE);
+ }
-+ break;
++ free_vm_area(map->vm);
++ kfree(map);
++}
+
-+ default:
-+ ret = -EINVAL;
-+ break;
-+ }
+
-+ return ret;
++void *net_accel_map_grants_contig(struct xenbus_device *dev,
++ unsigned *grants, int npages,
++ void **priv)
++{
++ return net_accel_map_grants_valloc(dev, grants, npages,
++ GNTMAP_host_map, priv);
+}
++EXPORT_SYMBOL(net_accel_map_grants_contig);
+
-+#ifndef HAVE_ARCH_PRIVCMD_MMAP
-+static struct page *privcmd_nopage(struct vm_area_struct *vma,
-+ unsigned long address,
-+ int *type)
++
++void net_accel_unmap_grants_contig(struct xenbus_device *dev,
++ void *priv)
+{
-+ return NOPAGE_SIGBUS;
++ net_accel_unmap_grants_vfree(dev, GNTMAP_host_map, priv);
+}
++EXPORT_SYMBOL(net_accel_unmap_grants_contig);
+
-+static struct vm_operations_struct privcmd_vm_ops = {
-+ .nopage = privcmd_nopage
-+};
+
-+static int privcmd_mmap(struct file * file, struct vm_area_struct * vma)
++void *net_accel_map_iomem_page(struct xenbus_device *dev, int gnt_ref,
++ void **priv)
+{
-+#ifndef __powerpc__ /* PowerPC has a trick to safely do this. */
-+ /* Unsupported for auto-translate guests. */
-+ if (xen_feature(XENFEAT_auto_translated_physmap))
-+ return -ENOSYS;
-+#endif
++ return net_accel_map_grants_valloc(dev, &gnt_ref, 1,
++ GNTMAP_host_map, priv);
++}
++EXPORT_SYMBOL(net_accel_map_iomem_page);
+
-+ /* DONTCOPY is essential for Xen as copy_page_range is broken. */
-+ vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY;
-+ vma->vm_ops = &privcmd_vm_ops;
-+ vma->vm_private_data = NULL;
+
-+ return 0;
++void net_accel_unmap_iomem_page(struct xenbus_device *dev, void *priv)
++{
++ net_accel_unmap_grants_vfree(dev, GNTMAP_host_map, priv);
+}
++EXPORT_SYMBOL(net_accel_unmap_iomem_page);
+
-+static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
++
++int net_accel_grant_page(struct xenbus_device *dev, unsigned long mfn,
++ int is_iomem)
+{
-+ return (xchg(&vma->vm_private_data, (void *)1) == NULL);
++ int err = gnttab_grant_foreign_access(dev->otherend_id, mfn,
++ is_iomem ? GTF_PCD : 0);
++ if (err < 0)
++ xenbus_dev_error(dev, err, "failed granting access to page\n");
++ return err;
+}
-+#endif
++EXPORT_SYMBOL_GPL(net_accel_grant_page);
+
-+static const struct file_operations privcmd_file_ops = {
-+ .unlocked_ioctl = privcmd_ioctl,
-+ .mmap = privcmd_mmap,
-+};
+
-+static int capabilities_read(char *page, char **start, off_t off,
-+ int count, int *eof, void *data)
++int net_accel_ungrant_page(grant_ref_t gntref)
+{
-+ int len = 0;
-+ *page = 0;
-+
-+ if (is_initial_xendomain())
-+ len = sprintf( page, "control_d\n" );
++ if (unlikely(gnttab_query_foreign_access(gntref) != 0)) {
++ EPRINTK("%s: remote domain still using grant %d\n", __FUNCTION__,
++ gntref);
++ return -EBUSY;
++ }
+
-+ *eof = 1;
-+ return len;
++ gnttab_end_foreign_access(gntref, 0);
++ return 0;
+}
++EXPORT_SYMBOL_GPL(net_accel_ungrant_page);
+
-+static int __init privcmd_init(void)
++
++int net_accel_xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
+{
-+ if (!is_running_on_xen())
-+ return -ENODEV;
++ char *s, *e, *macstr;
++ int i;
+
-+ privcmd_intf = create_xen_proc_entry("privcmd", 0400);
-+ if (privcmd_intf != NULL)
-+ privcmd_intf->proc_fops = &privcmd_file_ops;
++ macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
++ if (IS_ERR(macstr))
++ return PTR_ERR(macstr);
+
-+ capabilities_intf = create_xen_proc_entry("capabilities", 0400 );
-+ if (capabilities_intf != NULL)
-+ capabilities_intf->read_proc = capabilities_read;
++ for (i = 0; i < ETH_ALEN; i++) {
++ mac[i] = simple_strtoul(s, &e, 16);
++ if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
++ kfree(macstr);
++ return -ENOENT;
++ }
++ s = e+1;
++ }
+
++ kfree(macstr);
+ return 0;
+}
++EXPORT_SYMBOL_GPL(net_accel_xen_net_read_mac);
+
-+__initcall(privcmd_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/tpmback/common.h linux-2.6.18-xen.hg/drivers/xen/tpmback/common.h
---- linux-2.6.18/drivers/xen/tpmback/common.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/tpmback/common.h 2007-12-23 11:15:34.247945758 +0100
++
++void net_accel_update_state(struct xenbus_device *dev, int state)
++{
++ struct xenbus_transaction tr;
++ int err;
++
++ DPRINTK("%s: setting accelstate to %s\n", __FUNCTION__,
++ xenbus_strstate(state));
++
++ if (xenbus_exists(XBT_NIL, dev->nodename, "")) {
++ VPRINTK("%s: nodename %s\n", __FUNCTION__, dev->nodename);
++ again:
++ err = xenbus_transaction_start(&tr);
++ if (err == 0)
++ err = xenbus_printf(tr, dev->nodename, "accelstate",
++ "%d", state);
++ if (err != 0) {
++ xenbus_transaction_end(tr, 1);
++ } else {
++ err = xenbus_transaction_end(tr, 0);
++ if (err == -EAGAIN)
++ goto again;
++ }
++ }
++}
++EXPORT_SYMBOL_GPL(net_accel_update_state);
++
++MODULE_LICENSE("GPL");
+--- linux-2.6.18.8/drivers/xen/sfc_netutil/accel_util.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/sfc_netutil/accel_util.h 2008-05-19 00:33:48.746955951 +0300
+@@ -0,0 +1,127 @@
++/****************************************************************************
++ * Solarflare driver for Xen network acceleration
++ *
++ * Copyright 2006-2008: Solarflare Communications Inc,
++ * 9501 Jeronimo Road, Suite 250,
++ * Irvine, CA 92618, USA
++ *
++ * Maintained by Solarflare Communications <linux-xen-drivers@solarflare.com>
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License version 2 as published
++ * by the Free Software Foundation, incorporated herein by reference.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
++ ****************************************************************************
++ */
++
++#ifndef NETBACK_ACCEL_UTIL_H
++#define NETBACK_ACCEL_UTIL_H
++
++#ifdef DPRINTK
++#undef DPRINTK
++#endif
++
++#define FILE_LEAF strrchr(__FILE__, '/') ? strrchr(__FILE__, '/') + 1 : __FILE__
++
++#if 1
++#define VPRINTK(_f, _a...)
++#else
++#define VPRINTK(_f, _a...) \
++ printk("(file=%s, line=%d) " _f, \
++ FILE_LEAF , __LINE__ , ## _a )
++#endif
++
++#if 1
++#define DPRINTK(_f, _a...)
++#else
++#define DPRINTK(_f, _a...) \
++ printk("(file=%s, line=%d) " _f, \
++ FILE_LEAF , __LINE__ , ## _a )
++#endif
++
++#define EPRINTK(_f, _a...) \
++ printk("(file=%s, line=%d) " _f, \
++ FILE_LEAF , __LINE__ , ## _a )
++
++#define EPRINTK_ON(exp) \
++ do { \
++ if (exp) \
++ EPRINTK("%s at %s:%d\n", #exp, __FILE__, __LINE__); \
++ } while(0)
++
++#define DPRINTK_ON(exp) \
++ do { \
++ if (exp) \
++ DPRINTK("%s at %s:%d\n", #exp, __FILE__, __LINE__); \
++ } while(0)
++
++#define MAC_FMT "%.2x:%.2x:%.2x:%.2x:%.2x:%.2x"
++#define MAC_ARG(_mac) (_mac)[0], (_mac)[1], (_mac)[2], (_mac)[3], (_mac)[4], (_mac)[5]
++
++#include <xen/xenbus.h>
++
++/*! Map a set of pages from another domain
++ * \param dev The xenbus device context
++ * \param priv The private data returned by the mapping function
++ */
++extern
++void *net_accel_map_grants_contig(struct xenbus_device *dev,
++ unsigned *grants, int npages,
++ void **priv);
++
++/*! Unmap a set of pages mapped using net_accel_map_grants_contig.
++ * \param dev The xenbus device context
++ * \param priv The private data returned by the mapping function
++ */
++extern
++void net_accel_unmap_grants_contig(struct xenbus_device *dev, void *priv);
++
++/*! Read the MAC address of a device from xenstore */
++extern
++int net_accel_xen_net_read_mac(struct xenbus_device *dev, u8 mac[]);
++
++/*! Update the accelstate field for a device in xenstore */
++extern
++void net_accel_update_state(struct xenbus_device *dev, int state);
++
++/* These four map/unmap functions are based on
++ * xenbus_backend_client.c:xenbus_map_ring(). However, they are not
++ * used for ring buffers, instead just to map pages between domains,
++ * or to map a page so that it is accessible by a device
++ */
++extern
++int net_accel_map_device_page(struct xenbus_device *dev,
++ int gnt_ref, grant_handle_t *handle,
++ u64 *dev_bus_addr);
++extern
++int net_accel_unmap_device_page(struct xenbus_device *dev,
++ grant_handle_t handle, u64 dev_bus_addr);
++extern
++void *net_accel_map_iomem_page(struct xenbus_device *dev, int gnt_ref,
++ void **priv);
++extern
++void net_accel_unmap_iomem_page(struct xenbus_device *dev, void *priv);
++
++/*! Grrant a page to remote domain */
++extern
++int net_accel_grant_page(struct xenbus_device *dev, unsigned long mfn,
++ int is_iomem);
++/*! Undo a net_accel_grant_page */
++extern
++int net_accel_ungrant_page(grant_ref_t gntref);
++
++
++/*! Shutdown remote domain that is misbehaving */
++extern
++int net_accel_shutdown_remote(int domain);
++
++
++#endif
+--- linux-2.6.18.8/drivers/xen/tpmback/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/tpmback/Makefile 2008-05-19 00:33:48.746955951 +0300
+@@ -0,0 +1,4 @@
++
++obj-$(CONFIG_XEN_TPMDEV_BACKEND) += tpmbk.o
++
++tpmbk-y += tpmback.o interface.o xenbus.o
+--- linux-2.6.18.8/drivers/xen/tpmback/common.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/tpmback/common.h 2008-05-19 00:33:48.746955951 +0300
@@ -0,0 +1,85 @@
+/******************************************************************************
+ * drivers/xen/tpmback/common.h
+}
+
+#endif /* __TPMIF__BACKEND__COMMON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/tpmback/interface.c linux-2.6.18-xen.hg/drivers/xen/tpmback/interface.c
---- linux-2.6.18/drivers/xen/tpmback/interface.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/tpmback/interface.c 2007-12-23 11:15:34.247945758 +0100
-@@ -0,0 +1,167 @@
+--- linux-2.6.18.8/drivers/xen/tpmback/interface.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/tpmback/interface.c 2008-05-19 00:33:48.746955951 +0300
+@@ -0,0 +1,168 @@
+ /*****************************************************************************
+ * drivers/xen/tpmback/interface.c
+ *
+ }
+
+ tpmif->tx = (tpmif_tx_interface_t *)tpmif->tx_area->addr;
++ memset(tpmif->tx, 0, PAGE_SIZE);
+
+ err = bind_interdomain_evtchn_to_irqhandler(
+ tpmif->domid, evtchn, tpmif_be_int, 0, tpmif->devname, tpmif);
+{
+ kmem_cache_destroy(tpmif_cachep);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/tpmback/Makefile linux-2.6.18-xen.hg/drivers/xen/tpmback/Makefile
---- linux-2.6.18/drivers/xen/tpmback/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/tpmback/Makefile 2007-12-23 11:15:34.247945758 +0100
-@@ -0,0 +1,4 @@
-+
-+obj-$(CONFIG_XEN_TPMDEV_BACKEND) += tpmbk.o
-+
-+tpmbk-y += tpmback.o interface.o xenbus.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/tpmback/tpmback.c linux-2.6.18-xen.hg/drivers/xen/tpmback/tpmback.c
---- linux-2.6.18/drivers/xen/tpmback/tpmback.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/tpmback/tpmback.c 2007-12-23 11:15:34.247945758 +0100
+--- linux-2.6.18.8/drivers/xen/tpmback/tpmback.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/tpmback/tpmback.c 2008-05-19 00:33:48.750956181 +0300
@@ -0,0 +1,944 @@
+/******************************************************************************
+ * drivers/xen/tpmback/tpmback.c
+}
+
+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/tpmback/xenbus.c linux-2.6.18-xen.hg/drivers/xen/tpmback/xenbus.c
---- linux-2.6.18/drivers/xen/tpmback/xenbus.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/tpmback/xenbus.c 2007-12-23 11:15:34.247945758 +0100
+--- linux-2.6.18.8/drivers/xen/tpmback/xenbus.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/tpmback/xenbus.c 2008-05-19 00:33:48.750956181 +0300
@@ -0,0 +1,289 @@
+/* Xenbus code for tpmif backend
+ Copyright (C) 2005 IBM Corporation
+}
+
+
-+static struct xenbus_device_id tpmback_ids[] = {
++static const struct xenbus_device_id tpmback_ids[] = {
+ { "vtpm" },
+ { "" }
+};
+{
+ xenbus_unregister_driver(&tpmback);
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/util.c linux-2.6.18-xen.hg/drivers/xen/util.c
---- linux-2.6.18/drivers/xen/util.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/util.c 2007-12-23 11:15:34.247945758 +0100
+--- linux-2.6.18.8/drivers/xen/util.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/util.c 2008-05-19 00:33:48.750956181 +0300
@@ -0,0 +1,65 @@
+#include <linux/mm.h>
+#include <linux/module.h>
+}
+EXPORT_SYMBOL_GPL(free_vm_area);
+#endif /* CONFIG_X86 */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/Makefile linux-2.6.18-xen.hg/drivers/xen/xenbus/Makefile
---- linux-2.6.18/drivers/xen/xenbus/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/Makefile 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/Makefile 2008-05-19 00:33:48.750956181 +0300
@@ -0,0 +1,9 @@
+obj-y += xenbus_client.o xenbus_comms.o xenbus_xs.o xenbus_probe.o
+obj-$(CONFIG_XEN_BACKEND) += xenbus_be.o
+xenbus-$(CONFIG_XEN_BACKEND) += xenbus_probe_backend.o
+obj-y += $(xenbus-y) $(xenbus-m)
+obj-$(CONFIG_XEN_XENBUS_DEV) += xenbus_dev.o
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_backend_client.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_backend_client.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_backend_client.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_backend_client.c 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_backend_client.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_backend_client.c 2008-05-19 00:33:48.754956412 +0300
@@ -0,0 +1,147 @@
+/******************************************************************************
+ * Backend-client-facing interface for the Xenbus driver. In other words, the
+EXPORT_SYMBOL_GPL(xenbus_dev_is_online);
+
+MODULE_LICENSE("Dual BSD/GPL");
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_client.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_client.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_client.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_client.c 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_client.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_client.c 2008-05-19 00:33:48.754956412 +0300
@@ -0,0 +1,284 @@
+/******************************************************************************
+ * Client-facing interface for the Xenbus driver. In other words, the
+ const char **, unsigned int))
+{
+ int err;
-+ char *state = kasprintf(GFP_KERNEL, "%s/%s", path, path2);
++ char *state = kasprintf(GFP_KERNEL|__GFP_HIGH, "%s/%s", path, path2);
+ if (!state) {
+ xenbus_dev_fatal(dev, -ENOMEM, "allocating path for watch");
+ return -ENOMEM;
+ return result;
+}
+EXPORT_SYMBOL_GPL(xenbus_read_driver_state);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_comms.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_comms.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.c 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_comms.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.c 2008-05-19 00:33:48.754956412 +0300
@@ -0,0 +1,233 @@
+/******************************************************************************
+ * xenbus_comms.c
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_comms.h linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.h
---- linux-2.6.18/drivers/xen/xenbus/xenbus_comms.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.h 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_comms.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_comms.h 2008-05-19 00:33:48.754956412 +0300
@@ -0,0 +1,46 @@
+/*
+ * Private include for xenbus communications.
+extern int xen_store_evtchn;
+
+#endif /* _XENBUS_COMMS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_dev.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_dev.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_dev.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_dev.c 2007-12-23 11:15:34.251279267 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_dev.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_dev.c 2008-05-19 00:33:48.758956642 +0300
@@ -0,0 +1,404 @@
+/*
+ * xenbus_dev.c
+ token++;
+
+ if (msg_type == XS_WATCH) {
-+ watch = kmalloc(sizeof(*watch), GFP_KERNEL);
++ watch = kzalloc(sizeof(*watch), GFP_KERNEL);
+ watch->watch.node = kmalloc(strlen(path)+1,
+ GFP_KERNEL);
+ strcpy((char *)watch->watch.node, path);
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_probe_backend.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe_backend.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_probe_backend.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe_backend.c 2007-12-23 11:15:34.254612776 +0100
-@@ -0,0 +1,293 @@
-+/******************************************************************************
-+ * Talks to Xen Store to figure out what devices we have (backend half).
-+ *
-+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
-+ * Copyright (C) 2005 Mike Wray, Hewlett-Packard
-+ * Copyright (C) 2005, 2006 XenSource Ltd
-+ * Copyright (C) 2007 Solarflare Communications, Inc.
-+ *
-+ * This program is free software; you can redistribute it and/or
-+ * modify it under the terms of the GNU General Public License version 2
-+ * as published by the Free Software Foundation; or, when distributed
-+ * separately from the Linux kernel or incorporated into other
-+ * software packages, subject to the following license:
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this source file (the "Software"), to deal in the Software without
-+ * restriction, including without limitation the rights to use, copy, modify,
-+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
-+ * and to permit persons to whom the Software is furnished to do so, subject to
-+ * the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
-+ * IN THE SOFTWARE.
-+ */
-+
-+#define DPRINTK(fmt, args...) \
-+ pr_debug("xenbus_probe (%s:%d) " fmt ".\n", \
-+ __FUNCTION__, __LINE__, ##args)
-+
-+#include <linux/kernel.h>
-+#include <linux/err.h>
-+#include <linux/string.h>
-+#include <linux/ctype.h>
-+#include <linux/fcntl.h>
-+#include <linux/mm.h>
-+#include <linux/notifier.h>
-+
-+#include <asm/io.h>
-+#include <asm/page.h>
-+#include <asm/maddr.h>
-+#include <asm/pgtable.h>
-+#include <asm/hypervisor.h>
-+#include <xen/xenbus.h>
-+#include <xen/xen_proc.h>
-+#include <xen/evtchn.h>
-+#include <xen/features.h>
-+#include <xen/hvm.h>
-+
-+#include "xenbus_comms.h"
-+#include "xenbus_probe.h"
-+
-+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
-+#include <xen/platform-compat.h>
-+#endif
-+
-+static int xenbus_uevent_backend(struct device *dev, char **envp,
-+ int num_envp, char *buffer, int buffer_size);
-+static int xenbus_probe_backend(const char *type, const char *domid);
-+
-+extern int read_otherend_details(struct xenbus_device *xendev,
-+ char *id_node, char *path_node);
-+
-+static int read_frontend_details(struct xenbus_device *xendev)
-+{
-+ return read_otherend_details(xendev, "frontend-id", "frontend");
-+}
-+
-+/* backend/<type>/<fe-uuid>/<id> => <type>-<fe-domid>-<id> */
-+static int backend_bus_id(char bus_id[BUS_ID_SIZE], const char *nodename)
-+{
-+ int domid, err;
-+ const char *devid, *type, *frontend;
-+ unsigned int typelen;
-+
-+ type = strchr(nodename, '/');
-+ if (!type)
-+ return -EINVAL;
-+ type++;
-+ typelen = strcspn(type, "/");
-+ if (!typelen || type[typelen] != '/')
-+ return -EINVAL;
-+
-+ devid = strrchr(nodename, '/') + 1;
-+
-+ err = xenbus_gather(XBT_NIL, nodename, "frontend-id", "%i", &domid,
-+ "frontend", NULL, &frontend,
-+ NULL);
-+ if (err)
-+ return err;
-+ if (strlen(frontend) == 0)
-+ err = -ERANGE;
-+ if (!err && !xenbus_exists(XBT_NIL, frontend, ""))
-+ err = -ENOENT;
-+ kfree(frontend);
-+
-+ if (err)
-+ return err;
-+
-+ if (snprintf(bus_id, BUS_ID_SIZE,
-+ "%.*s-%i-%s", typelen, type, domid, devid) >= BUS_ID_SIZE)
-+ return -ENOSPC;
-+ return 0;
-+}
-+
-+static struct xen_bus_type xenbus_backend = {
-+ .root = "backend",
-+ .levels = 3, /* backend/type/<frontend>/<id> */
-+ .get_bus_id = backend_bus_id,
-+ .probe = xenbus_probe_backend,
-+ .error = -ENODEV,
-+ .bus = {
-+ .name = "xen-backend",
-+ .match = xenbus_match,
-+ .probe = xenbus_dev_probe,
-+ .remove = xenbus_dev_remove,
-+// .shutdown = xenbus_dev_shutdown,
-+ .uevent = xenbus_uevent_backend,
-+ },
-+ .dev = {
-+ .bus_id = "xen-backend",
-+ },
-+};
-+
-+static int xenbus_uevent_backend(struct device *dev, char **envp,
-+ int num_envp, char *buffer, int buffer_size)
-+{
-+ struct xenbus_device *xdev;
-+ struct xenbus_driver *drv;
-+ int i = 0;
-+ int length = 0;
-+
-+ DPRINTK("");
-+
-+ if (dev == NULL)
-+ return -ENODEV;
-+
-+ xdev = to_xenbus_device(dev);
-+ if (xdev == NULL)
-+ return -ENODEV;
-+
-+ /* stuff we want to pass to /sbin/hotplug */
-+ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
-+ "XENBUS_TYPE=%s", xdev->devicetype);
-+
-+ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
-+ "XENBUS_PATH=%s", xdev->nodename);
-+
-+ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
-+ "XENBUS_BASE_PATH=%s", xenbus_backend.root);
-+
-+ /* terminate, set to next free slot, shrink available space */
-+ envp[i] = NULL;
-+ envp = &envp[i];
-+ num_envp -= i;
-+ buffer = &buffer[length];
-+ buffer_size -= length;
-+
-+ if (dev->driver) {
-+ drv = to_xenbus_driver(dev->driver);
-+ if (drv && drv->uevent)
-+ return drv->uevent(xdev, envp, num_envp, buffer,
-+ buffer_size);
-+ }
-+
-+ return 0;
-+}
-+
-+int xenbus_register_backend(struct xenbus_driver *drv)
-+{
-+ drv->read_otherend_details = read_frontend_details;
-+
-+ return xenbus_register_driver_common(drv, &xenbus_backend);
-+}
-+EXPORT_SYMBOL_GPL(xenbus_register_backend);
-+
-+/* backend/<typename>/<frontend-uuid>/<name> */
-+static int xenbus_probe_backend_unit(const char *dir,
-+ const char *type,
-+ const char *name)
-+{
-+ char *nodename;
-+ int err;
-+
-+ nodename = kasprintf(GFP_KERNEL, "%s/%s", dir, name);
-+ if (!nodename)
-+ return -ENOMEM;
-+
-+ DPRINTK("%s\n", nodename);
-+
-+ err = xenbus_probe_node(&xenbus_backend, type, nodename);
-+ kfree(nodename);
-+ return err;
-+}
-+
-+/* backend/<typename>/<frontend-domid> */
-+static int xenbus_probe_backend(const char *type, const char *domid)
-+{
-+ char *nodename;
-+ int err = 0;
-+ char **dir;
-+ unsigned int i, dir_n = 0;
-+
-+ DPRINTK("");
-+
-+ nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_backend.root, type, domid);
-+ if (!nodename)
-+ return -ENOMEM;
-+
-+ dir = xenbus_directory(XBT_NIL, nodename, "", &dir_n);
-+ if (IS_ERR(dir)) {
-+ kfree(nodename);
-+ return PTR_ERR(dir);
-+ }
-+
-+ for (i = 0; i < dir_n; i++) {
-+ err = xenbus_probe_backend_unit(nodename, type, dir[i]);
-+ if (err)
-+ break;
-+ }
-+ kfree(dir);
-+ kfree(nodename);
-+ return err;
-+}
-+
-+static void backend_changed(struct xenbus_watch *watch,
-+ const char **vec, unsigned int len)
-+{
-+ DPRINTK("");
-+
-+ dev_changed(vec[XS_WATCH_PATH], &xenbus_backend);
-+}
-+
-+static struct xenbus_watch be_watch = {
-+ .node = "backend",
-+ .callback = backend_changed,
-+};
-+
-+void xenbus_backend_suspend(int (*fn)(struct device *, void *))
-+{
-+ DPRINTK("");
-+ if (!xenbus_backend.error)
-+ bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
-+}
-+
-+void xenbus_backend_resume(int (*fn)(struct device *, void *))
-+{
-+ DPRINTK("");
-+ if (!xenbus_backend.error)
-+ bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
-+}
-+
-+void xenbus_backend_probe_and_watch(void)
-+{
-+ xenbus_probe_devices(&xenbus_backend);
-+ register_xenbus_watch(&be_watch);
-+}
-+
-+void xenbus_backend_bus_register(void)
-+{
-+ xenbus_backend.error = bus_register(&xenbus_backend.bus);
-+ if (xenbus_backend.error)
-+ printk(KERN_WARNING
-+ "XENBUS: Error registering backend bus: %i\n",
-+ xenbus_backend.error);
-+}
-+
-+void xenbus_backend_device_register(void)
-+{
-+ if (xenbus_backend.error)
-+ return;
-+
-+ xenbus_backend.error = device_register(&xenbus_backend.dev);
-+ if (xenbus_backend.error) {
-+ bus_unregister(&xenbus_backend.bus);
-+ printk(KERN_WARNING
-+ "XENBUS: Error registering backend device: %i\n",
-+ xenbus_backend.error);
-+ }
-+}
-+
-+int xenbus_for_each_backend(void *arg, int (*fn)(struct device *, void *))
-+{
-+ return bus_for_each_dev(&xenbus_backend.bus, NULL, arg, fn);
-+}
-+EXPORT_SYMBOL_GPL(xenbus_for_each_backend);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_probe.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_probe.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.c 2007-12-23 11:15:34.254612776 +0100
-@@ -0,0 +1,1148 @@
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_probe.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.c 2008-05-19 00:33:48.758956642 +0300
+@@ -0,0 +1,1153 @@
+/******************************************************************************
+ * Talks to Xen Store to figure out what devices we have.
+ *
+#include <xen/xen_proc.h>
+#include <xen/evtchn.h>
+#include <xen/features.h>
++#ifdef MODULE
+#include <xen/hvm.h>
++#endif
+
+#include "xenbus_comms.h"
+#include "xenbus_probe.h"
+
+ DPRINTK("%s", dev->nodename);
+
++ if (is_initial_xendomain())
++ return;
++
+ get_device(&dev->dev);
+ if (dev->state != XenbusStateConnected) {
+ printk("%s: %s: %s != Connected, skipping\n", __FUNCTION__,
+ return bus_for_each_dev(&xenbus_frontend.bus, NULL, arg, fn);
+}
+EXPORT_SYMBOL_GPL(xenbus_for_each_frontend);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_probe.h linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.h
---- linux-2.6.18/drivers/xen/xenbus/xenbus_probe.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.h 2007-12-23 11:15:34.254612776 +0100
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_probe.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe.h 2008-05-19 00:33:48.758956642 +0300
@@ -0,0 +1,75 @@
+/******************************************************************************
+ * xenbus_probe.h
+ * IN THE SOFTWARE.
+ */
+
-+#ifndef _XENBUS_PROBE_H
-+#define _XENBUS_PROBE_H
++#ifndef _XENBUS_PROBE_H
++#define _XENBUS_PROBE_H
++
++#if defined(CONFIG_XEN_BACKEND) || defined(CONFIG_XEN_BACKEND_MODULE)
++extern void xenbus_backend_suspend(int (*fn)(struct device *, void *));
++extern void xenbus_backend_resume(int (*fn)(struct device *, void *));
++extern void xenbus_backend_probe_and_watch(void);
++extern void xenbus_backend_bus_register(void);
++extern void xenbus_backend_device_register(void);
++#else
++static inline void xenbus_backend_suspend(int (*fn)(struct device *, void *)) {}
++static inline void xenbus_backend_resume(int (*fn)(struct device *, void *)) {}
++static inline void xenbus_backend_probe_and_watch(void) {}
++static inline void xenbus_backend_bus_register(void) {}
++static inline void xenbus_backend_device_register(void) {}
++#endif
++
++struct xen_bus_type
++{
++ char *root;
++ int error;
++ unsigned int levels;
++ int (*get_bus_id)(char bus_id[BUS_ID_SIZE], const char *nodename);
++ int (*probe)(const char *type, const char *dir);
++ struct bus_type bus;
++ struct device dev;
++};
++
++extern int xenbus_match(struct device *_dev, struct device_driver *_drv);
++extern int xenbus_dev_probe(struct device *_dev);
++extern int xenbus_dev_remove(struct device *_dev);
++extern int xenbus_register_driver_common(struct xenbus_driver *drv,
++ struct xen_bus_type *bus);
++extern int xenbus_probe_node(struct xen_bus_type *bus,
++ const char *type,
++ const char *nodename);
++extern int xenbus_probe_devices(struct xen_bus_type *bus);
++
++extern void dev_changed(const char *node, struct xen_bus_type *bus);
++
++#endif
++
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_probe_backend.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_probe_backend.c 2008-05-19 00:33:48.762956873 +0300
+@@ -0,0 +1,292 @@
++/******************************************************************************
++ * Talks to Xen Store to figure out what devices we have (backend half).
++ *
++ * Copyright (C) 2005 Rusty Russell, IBM Corporation
++ * Copyright (C) 2005 Mike Wray, Hewlett-Packard
++ * Copyright (C) 2005, 2006 XenSource Ltd
++ * Copyright (C) 2007 Solarflare Communications, Inc.
++ *
++ * This program is free software; you can redistribute it and/or
++ * modify it under the terms of the GNU General Public License version 2
++ * as published by the Free Software Foundation; or, when distributed
++ * separately from the Linux kernel or incorporated into other
++ * software packages, subject to the following license:
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this source file (the "Software"), to deal in the Software without
++ * restriction, including without limitation the rights to use, copy, modify,
++ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
++ * and to permit persons to whom the Software is furnished to do so, subject to
++ * the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
++ * IN THE SOFTWARE.
++ */
++
++#define DPRINTK(fmt, args...) \
++ pr_debug("xenbus_probe (%s:%d) " fmt ".\n", \
++ __FUNCTION__, __LINE__, ##args)
+
-+#if defined(CONFIG_XEN_BACKEND) || defined(CONFIG_XEN_BACKEND_MODULE)
-+extern void xenbus_backend_suspend(int (*fn)(struct device *, void *));
-+extern void xenbus_backend_resume(int (*fn)(struct device *, void *));
-+extern void xenbus_backend_probe_and_watch(void);
-+extern void xenbus_backend_bus_register(void);
-+extern void xenbus_backend_device_register(void);
-+#else
-+static inline void xenbus_backend_suspend(int (*fn)(struct device *, void *)) {}
-+static inline void xenbus_backend_resume(int (*fn)(struct device *, void *)) {}
-+static inline void xenbus_backend_probe_and_watch(void) {}
-+static inline void xenbus_backend_bus_register(void) {}
-+static inline void xenbus_backend_device_register(void) {}
++#include <linux/kernel.h>
++#include <linux/err.h>
++#include <linux/string.h>
++#include <linux/ctype.h>
++#include <linux/fcntl.h>
++#include <linux/mm.h>
++#include <linux/notifier.h>
++
++#include <asm/io.h>
++#include <asm/page.h>
++#include <asm/maddr.h>
++#include <asm/pgtable.h>
++#include <asm/hypervisor.h>
++#include <xen/xenbus.h>
++#include <xen/xen_proc.h>
++#include <xen/evtchn.h>
++#include <xen/features.h>
++
++#include "xenbus_comms.h"
++#include "xenbus_probe.h"
++
++#ifdef HAVE_XEN_PLATFORM_COMPAT_H
++#include <xen/platform-compat.h>
+#endif
+
-+struct xen_bus_type
++static int xenbus_uevent_backend(struct device *dev, char **envp,
++ int num_envp, char *buffer, int buffer_size);
++static int xenbus_probe_backend(const char *type, const char *domid);
++
++extern int read_otherend_details(struct xenbus_device *xendev,
++ char *id_node, char *path_node);
++
++static int read_frontend_details(struct xenbus_device *xendev)
+{
-+ char *root;
-+ int error;
-+ unsigned int levels;
-+ int (*get_bus_id)(char bus_id[BUS_ID_SIZE], const char *nodename);
-+ int (*probe)(const char *type, const char *dir);
-+ struct bus_type bus;
-+ struct device dev;
++ return read_otherend_details(xendev, "frontend-id", "frontend");
++}
++
++/* backend/<type>/<fe-uuid>/<id> => <type>-<fe-domid>-<id> */
++static int backend_bus_id(char bus_id[BUS_ID_SIZE], const char *nodename)
++{
++ int domid, err;
++ const char *devid, *type, *frontend;
++ unsigned int typelen;
++
++ type = strchr(nodename, '/');
++ if (!type)
++ return -EINVAL;
++ type++;
++ typelen = strcspn(type, "/");
++ if (!typelen || type[typelen] != '/')
++ return -EINVAL;
++
++ devid = strrchr(nodename, '/') + 1;
++
++ err = xenbus_gather(XBT_NIL, nodename, "frontend-id", "%i", &domid,
++ "frontend", NULL, &frontend,
++ NULL);
++ if (err)
++ return err;
++ if (strlen(frontend) == 0)
++ err = -ERANGE;
++ if (!err && !xenbus_exists(XBT_NIL, frontend, ""))
++ err = -ENOENT;
++ kfree(frontend);
++
++ if (err)
++ return err;
++
++ if (snprintf(bus_id, BUS_ID_SIZE,
++ "%.*s-%i-%s", typelen, type, domid, devid) >= BUS_ID_SIZE)
++ return -ENOSPC;
++ return 0;
++}
++
++static struct xen_bus_type xenbus_backend = {
++ .root = "backend",
++ .levels = 3, /* backend/type/<frontend>/<id> */
++ .get_bus_id = backend_bus_id,
++ .probe = xenbus_probe_backend,
++ .error = -ENODEV,
++ .bus = {
++ .name = "xen-backend",
++ .match = xenbus_match,
++ .probe = xenbus_dev_probe,
++ .remove = xenbus_dev_remove,
++// .shutdown = xenbus_dev_shutdown,
++ .uevent = xenbus_uevent_backend,
++ },
++ .dev = {
++ .bus_id = "xen-backend",
++ },
+};
+
-+extern int xenbus_match(struct device *_dev, struct device_driver *_drv);
-+extern int xenbus_dev_probe(struct device *_dev);
-+extern int xenbus_dev_remove(struct device *_dev);
-+extern int xenbus_register_driver_common(struct xenbus_driver *drv,
-+ struct xen_bus_type *bus);
-+extern int xenbus_probe_node(struct xen_bus_type *bus,
-+ const char *type,
-+ const char *nodename);
-+extern int xenbus_probe_devices(struct xen_bus_type *bus);
++static int xenbus_uevent_backend(struct device *dev, char **envp,
++ int num_envp, char *buffer, int buffer_size)
++{
++ struct xenbus_device *xdev;
++ struct xenbus_driver *drv;
++ int i = 0;
++ int length = 0;
+
-+extern void dev_changed(const char *node, struct xen_bus_type *bus);
++ DPRINTK("");
+
-+#endif
++ if (dev == NULL)
++ return -ENODEV;
++
++ xdev = to_xenbus_device(dev);
++ if (xdev == NULL)
++ return -ENODEV;
++
++ /* stuff we want to pass to /sbin/hotplug */
++ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
++ "XENBUS_TYPE=%s", xdev->devicetype);
++
++ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
++ "XENBUS_PATH=%s", xdev->nodename);
++
++ add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
++ "XENBUS_BASE_PATH=%s", xenbus_backend.root);
++
++ /* terminate, set to next free slot, shrink available space */
++ envp[i] = NULL;
++ envp = &envp[i];
++ num_envp -= i;
++ buffer = &buffer[length];
++ buffer_size -= length;
++
++ if (dev->driver) {
++ drv = to_xenbus_driver(dev->driver);
++ if (drv && drv->uevent)
++ return drv->uevent(xdev, envp, num_envp, buffer,
++ buffer_size);
++ }
++
++ return 0;
++}
++
++int xenbus_register_backend(struct xenbus_driver *drv)
++{
++ drv->read_otherend_details = read_frontend_details;
++
++ return xenbus_register_driver_common(drv, &xenbus_backend);
++}
++EXPORT_SYMBOL_GPL(xenbus_register_backend);
++
++/* backend/<typename>/<frontend-uuid>/<name> */
++static int xenbus_probe_backend_unit(const char *dir,
++ const char *type,
++ const char *name)
++{
++ char *nodename;
++ int err;
++
++ nodename = kasprintf(GFP_KERNEL, "%s/%s", dir, name);
++ if (!nodename)
++ return -ENOMEM;
++
++ DPRINTK("%s\n", nodename);
++
++ err = xenbus_probe_node(&xenbus_backend, type, nodename);
++ kfree(nodename);
++ return err;
++}
++
++/* backend/<typename>/<frontend-domid> */
++static int xenbus_probe_backend(const char *type, const char *domid)
++{
++ char *nodename;
++ int err = 0;
++ char **dir;
++ unsigned int i, dir_n = 0;
++
++ DPRINTK("");
++
++ nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_backend.root, type, domid);
++ if (!nodename)
++ return -ENOMEM;
++
++ dir = xenbus_directory(XBT_NIL, nodename, "", &dir_n);
++ if (IS_ERR(dir)) {
++ kfree(nodename);
++ return PTR_ERR(dir);
++ }
++
++ for (i = 0; i < dir_n; i++) {
++ err = xenbus_probe_backend_unit(nodename, type, dir[i]);
++ if (err)
++ break;
++ }
++ kfree(dir);
++ kfree(nodename);
++ return err;
++}
++
++static void backend_changed(struct xenbus_watch *watch,
++ const char **vec, unsigned int len)
++{
++ DPRINTK("");
++
++ dev_changed(vec[XS_WATCH_PATH], &xenbus_backend);
++}
++
++static struct xenbus_watch be_watch = {
++ .node = "backend",
++ .callback = backend_changed,
++};
++
++void xenbus_backend_suspend(int (*fn)(struct device *, void *))
++{
++ DPRINTK("");
++ if (!xenbus_backend.error)
++ bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
++}
++
++void xenbus_backend_resume(int (*fn)(struct device *, void *))
++{
++ DPRINTK("");
++ if (!xenbus_backend.error)
++ bus_for_each_dev(&xenbus_backend.bus, NULL, NULL, fn);
++}
++
++void xenbus_backend_probe_and_watch(void)
++{
++ xenbus_probe_devices(&xenbus_backend);
++ register_xenbus_watch(&be_watch);
++}
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenbus/xenbus_xs.c linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_xs.c
---- linux-2.6.18/drivers/xen/xenbus/xenbus_xs.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_xs.c 2007-12-23 11:15:34.254612776 +0100
-@@ -0,0 +1,886 @@
++void xenbus_backend_bus_register(void)
++{
++ xenbus_backend.error = bus_register(&xenbus_backend.bus);
++ if (xenbus_backend.error)
++ printk(KERN_WARNING
++ "XENBUS: Error registering backend bus: %i\n",
++ xenbus_backend.error);
++}
++
++void xenbus_backend_device_register(void)
++{
++ if (xenbus_backend.error)
++ return;
++
++ xenbus_backend.error = device_register(&xenbus_backend.dev);
++ if (xenbus_backend.error) {
++ bus_unregister(&xenbus_backend.bus);
++ printk(KERN_WARNING
++ "XENBUS: Error registering backend device: %i\n",
++ xenbus_backend.error);
++ }
++}
++
++int xenbus_for_each_backend(void *arg, int (*fn)(struct device *, void *))
++{
++ return bus_for_each_dev(&xenbus_backend.bus, NULL, arg, fn);
++}
++EXPORT_SYMBOL_GPL(xenbus_for_each_backend);
+--- linux-2.6.18.8/drivers/xen/xenbus/xenbus_xs.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenbus/xenbus_xs.c 2008-05-19 00:33:48.782958026 +0300
+@@ -0,0 +1,901 @@
+/******************************************************************************
+ * xenbus_xs.c
+ *
+ char *buffer;
+
+ if (strlen(name) == 0)
-+ buffer = kasprintf(GFP_KERNEL, "%s", dir);
++ buffer = kasprintf(GFP_KERNEL|__GFP_HIGH, "%s", dir);
+ else
-+ buffer = kasprintf(GFP_KERNEL, "%s/%s", dir, name);
++ buffer = kasprintf(GFP_KERNEL|__GFP_HIGH, "%s/%s", dir, name);
+ return (!buffer) ? ERR_PTR(-ENOMEM) : buffer;
+}
+
+ *num = count_strings(strings, len);
+
+ /* Transfer to one big alloc for easy freeing. */
-+ ret = kmalloc(*num * sizeof(char *) + len, GFP_KERNEL);
++ ret = kmalloc(*num * sizeof(char *) + len, GFP_KERNEL|__GFP_HIGH);
+ if (!ret) {
+ kfree(strings);
+ return ERR_PTR(-ENOMEM);
+#define PRINTF_BUFFER_SIZE 4096
+ char *printf_buffer;
+
-+ printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL);
++ printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL|__GFP_HIGH);
+ if (printf_buffer == NULL)
+ return -ENOMEM;
+
+ char token[sizeof(watch) * 2 + 1];
+ int err;
+
++ BUG_ON(watch->flags & XBWF_new_thread);
++
+ sprintf(token, "%lX", (long)watch);
+
+ down_read(&xs_state.watch_mutex);
+ list_del(ent);
+ spin_unlock(&watch_events_lock);
+
-+ if (ent != &watch_events) {
-+ msg = list_entry(ent, struct xs_stored_msg, list);
-+ if (msg->u.watch.handle->flags & XBWF_new_thread)
-+ kthread_run(xenwatch_handle_callback,
-+ msg, "xenwatch_cb");
-+ else
-+ xenwatch_handle_callback(msg);
++ if (ent == &watch_events) {
++ mutex_unlock(&xenwatch_mutex);
++ continue;
+ }
+
-+ mutex_unlock(&xenwatch_mutex);
++ msg = list_entry(ent, struct xs_stored_msg, list);
++
++ /*
++ * Unlock the mutex before running an XBWF_new_thread
++ * handler. kthread_run can block which can deadlock
++ * against unregister_xenbus_watch() if we need to
++ * unregister other watches in order to make
++ * progress. This can occur on resume before the swap
++ * device is attached.
++ */
++ if (msg->u.watch.handle->flags & XBWF_new_thread) {
++ mutex_unlock(&xenwatch_mutex);
++ kthread_run(xenwatch_handle_callback,
++ msg, "xenwatch_cb");
++ } else {
++ xenwatch_handle_callback(msg);
++ mutex_unlock(&xenwatch_mutex);
++ }
+ }
+
+ return 0;
+ }
+
+
-+ msg = kmalloc(sizeof(*msg), GFP_KERNEL);
++ msg = kmalloc(sizeof(*msg), GFP_KERNEL|__GFP_HIGH);
+ if (msg == NULL) {
+ err = -ENOMEM;
+ goto out;
+ goto out;
+ }
+
-+ body = kmalloc(msg->hdr.len + 1, GFP_KERNEL);
++ body = kmalloc(msg->hdr.len + 1, GFP_KERNEL|__GFP_HIGH);
+ if (body == NULL) {
+ kfree(msg);
+ err = -ENOMEM;
+
+ return 0;
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/drivers/xen/xenoprof/xenoprofile.c linux-2.6.18-xen.hg/drivers/xen/xenoprof/xenoprofile.c
---- linux-2.6.18/drivers/xen/xenoprof/xenoprofile.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/drivers/xen/xenoprof/xenoprofile.c 2007-12-23 11:15:34.254612776 +0100
-@@ -0,0 +1,544 @@
+--- linux-2.6.18.8/drivers/xen/xenoprof/xenoprofile.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/drivers/xen/xenoprof/xenoprofile.c 2008-05-19 00:33:48.782958026 +0300
+@@ -0,0 +1,546 @@
+/**
+ * @file xenoprofile.c
+ *
+xenoprof_ovf_interrupt(int irq, void * dev_id, struct pt_regs * regs)
+{
+ struct xenoprof_buf * buf;
-+ int cpu;
+ static unsigned long flag;
+
-+ cpu = smp_processor_id();
-+ buf = xenoprof_buf[cpu];
++ buf = xenoprof_buf[smp_processor_id()];
+
+ xenoprof_add_pc(buf, 0);
+
+
+static void unbind_virq(void)
+{
-+ int i;
++ unsigned int i;
+
+ for_each_online_cpu(i) {
+ if (ovf_irq[i] >= 0) {
+
+static int bind_virq(void)
+{
-+ int i, result;
++ unsigned int i;
++ int result;
+
+ for_each_online_cpu(i) {
+ result = bind_virq_to_irqhandler(VIRQ_XENOPROF,
+{
+ xenoprof_enabled = 0;
+
-+ HYPERVISOR_xenoprof_op(XENOPROF_disable_virq, NULL);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_disable_virq, NULL));
+
+ if (xenoprof_is_primary) {
-+ HYPERVISOR_xenoprof_op(XENOPROF_release_counters, NULL);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_release_counters,
++ NULL));
+ active_defined = 0;
+ }
+
+static void xenoprof_stop(void)
+{
+ if (xenoprof_is_primary)
-+ HYPERVISOR_xenoprof_op(XENOPROF_stop, NULL);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_stop, NULL));
+ xenoprof_arch_stop();
+}
+
+
+out:
+ if (ret)
-+ HYPERVISOR_xenoprof_op(XENOPROF_reset_active_list, NULL);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_reset_active_list,
++ NULL));
+ active_defined = !ret;
+ return ret;
+}
+ unsigned int pdoms)
+{
+ int ret;
-+ int i, j;
++ unsigned int i, j;
+ struct xenoprof_buf *buf;
+
+ if (!xenoprof_is_primary)
+int __init xenoprofile_init(struct oprofile_operations * ops)
+{
+ struct xenoprof_init init;
-+ int ret, i;
++ unsigned int i;
++ int ret;
+
+ ret = HYPERVISOR_xenoprof_op(XENOPROF_init, &init);
+ if (!ret) {
+ xenoprof_arch_unmap_shared_buffer(&shared_buffer);
+ if (xenoprof_is_primary) {
+ unmap_passive_list();
-+ HYPERVISOR_xenoprof_op(XENOPROF_shutdown, NULL);
++ WARN_ON(HYPERVISOR_xenoprof_op(XENOPROF_shutdown, NULL));
+ }
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/aio.c linux-2.6.18-xen.hg/fs/aio.c
---- linux-2.6.18/fs/aio.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/aio.c 2007-12-23 11:15:34.304615402 +0100
+--- linux-2.6.18.8/fs/Kconfig 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/Kconfig 2008-05-19 00:33:48.982969555 +0300
+@@ -826,6 +826,11 @@
+ help
+ Exports the dump image of crashed kernel in ELF format.
+
++config PROC_IOMEM_MACHINE
++ bool
++ depends on PROC_FS && EXPERIMENTAL && KEXEC && XEN && IA64
++ default y
++
+ config SYSFS
+ bool "sysfs file system support" if EMBEDDED
+ default y
+@@ -865,6 +870,7 @@
+ config HUGETLBFS
+ bool "HugeTLB file system support"
+ depends X86 || IA64 || PPC64 || SPARC64 || SUPERH || BROKEN
++ depends on !XEN
+ help
+ hugetlbfs is a filesystem backing for HugeTLB pages, based on
+ ramfs. For architectures that support it, say Y here and read
+--- linux-2.6.18.8/fs/aio.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/aio.c 2008-05-19 00:33:49.282986848 +0300
@@ -34,6 +34,11 @@
#include <asm/uaccess.h>
#include <asm/mmu_context.h>
get_ioctx(ioctx); /* io_destroy() expects us to hold a ref */
io_destroy(ioctx);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/compat_ioctl.c linux-2.6.18-xen.hg/fs/compat_ioctl.c
---- linux-2.6.18/fs/compat_ioctl.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/compat_ioctl.c 2007-12-23 11:15:34.384619602 +0100
-@@ -124,6 +124,11 @@
+--- linux-2.6.18.8/fs/binfmt_elf.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/binfmt_elf.c 2008-05-19 00:33:49.506999760 +0300
+@@ -1170,6 +1170,10 @@
+ */
+ static int maydump(struct vm_area_struct *vma)
+ {
++ /* The vma can be set up to tell us the answer directly. */
++ if (vma->vm_flags & VM_ALWAYSDUMP)
++ return 1;
++
+ /* Do not dump I/O mapped devices or special mappings */
+ if (vma->vm_flags & (VM_IO | VM_RESERVED))
+ return 0;
+--- linux-2.6.18.8/fs/bio.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/bio.c 2008-05-19 00:33:49.551002297 +0300
+@@ -112,6 +112,7 @@
+
+ BIO_BUG_ON(pool_idx >= BIOVEC_NR_POOLS);
+
++ if (bio->bi_io_vec)
+ mempool_free(bio->bi_io_vec, bio_set->bvec_pools[pool_idx]);
+ mempool_free(bio, bio_set->bio_pool);
+ }
+--- linux-2.6.18.8/fs/compat_ioctl.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/compat_ioctl.c 2008-05-19 00:33:49.735012903 +0300
+@@ -124,6 +124,13 @@
#include <linux/dvb/video.h>
#include <linux/lp.h>
++#ifdef CONFIG_XEN
+#include <xen/interface/xen.h>
+#include <xen/public/evtchn.h>
+#include <xen/public/privcmd.h>
+#include <xen/compat_ioctl.h>
++#endif
+
/* Aiee. Someone does not find a difference between int and long */
#define EXT2_IOC32_GETFLAGS _IOR('f', 1, int)
#define EXT2_IOC32_SETFLAGS _IOW('f', 2, int)
-@@ -2948,6 +2953,18 @@
+@@ -2948,6 +2955,18 @@
/*LPGETSTATS not implemented, but no kernels seem to compile it in anyways*/
COMPATIBLE_IOCTL(LPGETFLAGS)
HANDLE_IOCTL(LPSETTIMEOUT, lp_timeout_trans)
};
int ioctl_table_size = ARRAY_SIZE(ioctl_start);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/eventpoll.c linux-2.6.18-xen.hg/fs/eventpoll.c
---- linux-2.6.18/fs/eventpoll.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/eventpoll.c 2007-12-23 11:15:34.401287149 +0100
+--- linux-2.6.18.8/fs/debugfs/file.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/debugfs/file.c 2008-05-19 00:33:49.819017745 +0300
+@@ -170,6 +170,51 @@
+ }
+ EXPORT_SYMBOL_GPL(debugfs_create_u32);
+
++
++static void debugfs_u64_set(void *data, u64 val)
++{
++ *(u64 *)data = val;
++}
++
++static u64 debugfs_u64_get(void *data)
++{
++ return *(u64 *)data;
++}
++DEFINE_SIMPLE_ATTRIBUTE(fops_u64, debugfs_u64_get, debugfs_u64_set, "%llu\n");
++
++/**
++ * debugfs_create_u64 - create a file in the debugfs filesystem that is used to read and write an unsigned 64 bit value.
++ *
++ * @name: a pointer to a string containing the name of the file to create.
++ * @mode: the permission that the file should have
++ * @parent: a pointer to the parent dentry for this file. This should be a
++ * directory dentry if set. If this paramater is NULL, then the
++ * file will be created in the root of the debugfs filesystem.
++ * @value: a pointer to the variable that the file should read to and write
++ * from.
++ *
++ * This function creates a file in debugfs with the given name that
++ * contains the value of the variable @value. If the @mode variable is so
++ * set, it can be read from, and written to.
++ *
++ * This function will return a pointer to a dentry if it succeeds. This
++ * pointer must be passed to the debugfs_remove() function when the file is
++ * to be removed (no automatic cleanup happens if your module is unloaded,
++ * you are responsible here.) If an error occurs, NULL will be returned.
++ *
++ * If debugfs is not enabled in the kernel, the value -ENODEV will be
++ * returned. It is not wise to check for this value, but rather, check for
++ * NULL or !NULL instead as to eliminate the need for #ifdef in the calling
++ * code.
++ */
++struct dentry *debugfs_create_u64(const char *name, mode_t mode,
++ struct dentry *parent, u64 *value)
++{
++ return debugfs_create_file(name, mode, parent, value, &fops_u64);
++}
++EXPORT_SYMBOL_GPL(debugfs_create_u64);
++
++
+ static ssize_t read_file_bool(struct file *file, char __user *user_buf,
+ size_t count, loff_t *ppos)
+ {
+--- linux-2.6.18.8/fs/eventpoll.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/eventpoll.c 2008-05-19 00:33:49.903022587 +0300
@@ -236,8 +236,6 @@
static void ep_poll_safewake_init(struct poll_safewake *psw);
/*
* Mark the inode dirty from the very beginning,
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/Kconfig linux-2.6.18-xen.hg/fs/Kconfig
---- linux-2.6.18/fs/Kconfig 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/Kconfig 2007-12-23 11:15:34.274613828 +0100
-@@ -826,6 +826,11 @@
- help
- Exports the dump image of crashed kernel in ELF format.
-
-+config PROC_IOMEM_MACHINE
-+ bool
-+ depends on PROC_FS && EXPERIMENTAL && KEXEC && XEN && IA64
-+ default y
-+
- config SYSFS
- bool "sysfs file system support" if EMBEDDED
- default y
-@@ -865,6 +870,7 @@
- config HUGETLBFS
- bool "HugeTLB file system support"
- depends X86 || IA64 || PPC64 || SPARC64 || SUPERH || BROKEN
-+ depends on !XEN
- help
- hugetlbfs is a filesystem backing for HugeTLB pages, based on
- ramfs. For architectures that support it, say Y here and read
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/lockd/svc.c linux-2.6.18-xen.hg/fs/lockd/svc.c
---- linux-2.6.18/fs/lockd/svc.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/lockd/svc.c 2007-12-23 11:15:35.277999847 +0100
+--- linux-2.6.18.8/fs/lockd/svc.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/lockd/svc.c 2008-05-19 00:33:53.823248553 +0300
@@ -323,9 +323,6 @@
* Sysctl parameters (same as module parameters, different interface).
*/
static ctl_table nlm_sysctls[] = {
{
.ctl_name = CTL_UNNUMBERED,
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/nfs/sysctl.c linux-2.6.18-xen.hg/fs/nfs/sysctl.c
---- linux-2.6.18/fs/nfs/sysctl.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/nfs/sysctl.c 2007-12-23 11:15:35.324668967 +0100
+--- linux-2.6.18.8/fs/nfs/sysctl.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/nfs/sysctl.c 2008-05-19 00:33:54.179269075 +0300
@@ -18,11 +18,6 @@
static const int nfs_set_port_min = 0;
static const int nfs_set_port_max = 65535;
static ctl_table nfs_cb_sysctls[] = {
#ifdef CONFIG_NFS_V4
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/proc/proc_misc.c linux-2.6.18-xen.hg/fs/proc/proc_misc.c
---- linux-2.6.18/fs/proc/proc_misc.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/proc/proc_misc.c 2007-12-23 11:15:36.721408978 +0100
+--- linux-2.6.18.8/fs/proc/proc_misc.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/proc/proc_misc.c 2008-05-19 00:33:55.683355772 +0300
@@ -471,7 +471,7 @@
(unsigned long long)cputime64_to_clock_t(irq),
(unsigned long long)cputime64_to_clock_t(softirq),
/* Copy values here to work around gcc-2.95.3, gcc-2.96 */
user = kstat_cpu(i).cpustat.user;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/proc/vmcore.c linux-2.6.18-xen.hg/fs/proc/vmcore.c
---- linux-2.6.18/fs/proc/vmcore.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/proc/vmcore.c 2007-12-23 11:15:36.724742484 +0100
+--- linux-2.6.18.8/fs/proc/vmcore.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/proc/vmcore.c 2008-05-19 00:33:55.687356003 +0300
@@ -514,7 +514,7 @@
/* Do some basic Verification. */
if (memcmp(ehdr.e_ident, ELFMAG, SELFMAG) != 0 ||
ehdr.e_ident[EI_CLASS] != ELFCLASS64 ||
ehdr.e_ident[EI_VERSION] != EV_CURRENT ||
ehdr.e_version != EV_CURRENT ||
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/fs/xfs/linux-2.6/xfs_buf.c linux-2.6.18-xen.hg/fs/xfs/linux-2.6/xfs_buf.c
---- linux-2.6.18/fs/xfs/linux-2.6/xfs_buf.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/fs/xfs/linux-2.6/xfs_buf.c 2007-12-23 11:15:37.398111178 +0100
+--- linux-2.6.18.8/fs/splice.c 2008-05-19 00:42:34.129243112 +0300
++++ linux-2.6.18-xen.hg/fs/splice.c 2008-05-19 00:33:55.971372374 +0300
+@@ -1141,6 +1141,9 @@
+ if (unlikely(!base))
+ break;
+
++ if (unlikely(!access_ok(VERIFY_READ, base, len)))
++ break;
++
+ /*
+ * Get this base offset and number of pages, then map
+ * in the user pages.
+--- linux-2.6.18.8/fs/xfs/linux-2.6/xfs_buf.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/fs/xfs/linux-2.6/xfs_buf.c 2008-05-19 00:33:58.495517869 +0300
@@ -182,6 +182,19 @@
{
a_list_t *aentry;
aentry = kmalloc(sizeof(a_list_t), GFP_NOWAIT);
if (likely(aentry)) {
spin_lock(&as_lock);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-generic/pci.h linux-2.6.18-xen.hg/include/asm-generic/pci.h
---- linux-2.6.18/include/asm-generic/pci.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-generic/pci.h 2007-12-23 11:15:40.078251913 +0100
+--- linux-2.6.18.8/include/acpi/processor.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/acpi/processor.h 2008-05-19 00:33:59.571579895 +0300
+@@ -21,6 +21,10 @@
+ #define ACPI_PSD_REV0_REVISION 0 /* Support for _PSD as in ACPI 3.0 */
+ #define ACPI_PSD_REV0_ENTRIES 5
+
++#ifdef CONFIG_XEN
++#define NR_ACPI_CPUS 256
++#endif /* CONFIG_XEN */
++
+ /*
+ * Types of coordination defined in ACPI 3.0. Same macros can be used across
+ * P, C and T states
+@@ -33,6 +37,17 @@
+
+ struct acpi_processor_cx;
+
++#ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++struct acpi_csd_package {
++ acpi_integer num_entries;
++ acpi_integer revision;
++ acpi_integer domain;
++ acpi_integer coord_type;
++ acpi_integer num_processors;
++ acpi_integer index;
++} __attribute__ ((packed));
++#endif
++
+ struct acpi_power_register {
+ u8 descriptor;
+ u16 length;
+@@ -63,6 +78,12 @@
+ u32 power;
+ u32 usage;
+ u64 time;
++#ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++ /* Require raw information for external control logic */
++ struct acpi_power_register reg;
++ u32 csd_count;
++ struct acpi_csd_package *domain_info;
++#endif
+ struct acpi_processor_cx_policy promotion;
+ struct acpi_processor_cx_policy demotion;
+ };
+@@ -275,4 +296,80 @@
+ }
+ #endif
+
++#ifdef CONFIG_XEN
++/*
++ * Following are interfaces geared to external processor PM control
++ * logic like a VMM
++ */
++/* Events notified to external control logic */
++#define PROCESSOR_PM_INIT 1
++#define PROCESSOR_PM_CHANGE 2
++#define PROCESSOR_HOTPLUG 3
++
++/* Objects for the PM envents */
++#define PM_TYPE_IDLE 0
++#define PM_TYPE_PERF 1
++#define PM_TYPE_THR 2
++#define PM_TYPE_MAX 3
++
++/* Processor hotplug events */
++#define HOTPLUG_TYPE_ADD 0
++#define HOTPLUG_TYPE_REMOVE 1
++
++#ifdef CONFIG_PROCESSOR_EXTERNAL_CONTROL
++struct processor_extcntl_ops {
++ /* Transfer processor PM events to external control logic */
++ int (*pm_ops[PM_TYPE_MAX])(struct acpi_processor *pr, int event);
++ /* Notify physical processor status to external control logic */
++ int (*hotplug)(struct acpi_processor *pr, int event);
++};
++extern struct processor_extcntl_ops *processor_extcntl_ops;
++
++static inline int processor_cntl_external(void)
++{
++ return (processor_extcntl_ops != NULL);
++}
++
++static inline int processor_pm_external(void)
++{
++ return processor_cntl_external() &&
++ (processor_extcntl_ops->pm_ops[PM_TYPE_IDLE] != NULL);
++}
++
++static inline int processor_pmperf_external(void)
++{
++ return processor_cntl_external() &&
++ (processor_extcntl_ops->pm_ops[PM_TYPE_PERF] != NULL);
++}
++
++static inline int processor_pmthr_external(void)
++{
++ return processor_cntl_external() &&
++ (processor_extcntl_ops->pm_ops[PM_TYPE_THR] != NULL);
++}
++
++extern int processor_notify_external(struct acpi_processor *pr,
++ int event, int type);
++extern int processor_register_extcntl(struct processor_extcntl_ops *ops);
++extern int processor_unregister_extcntl(struct processor_extcntl_ops *ops);
++extern int processor_extcntl_init(struct acpi_processor *pr);
++extern int acpi_processor_get_performance_info(struct acpi_processor *pr);
++extern int acpi_processor_get_psd(struct acpi_processor *pr);
++#else
++static inline int processor_cntl_external(void) {return 0;}
++static inline int processor_pm_external(void) {return 0;}
++static inline int processor_pmperf_external(void) {return 0;}
++static inline int processor_pmthr_external(void) {return 0;}
++static inline int processor_notify_external(struct acpi_processor *pr,
++ int event, int type)
++{
++ return -EINVAL;
++}
++static inline int processor_extcntl_init(struct acpi_processor *pr)
++{
++ return -EINVAL;
++}
++#endif /* CONFIG_PROCESSOR_EXTERNAL_CONTROL */
++#endif /* CONFIG_XEN */
++
+ #endif
+--- linux-2.6.18.8/include/asm-generic/pci.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-generic/pci.h 2008-05-19 00:34:06.095955969 +0300
@@ -43,7 +43,9 @@
return root;
}
#ifndef HAVE_ARCH_PCI_GET_LEGACY_IDE_IRQ
static inline int pci_get_legacy_ide_irq(struct pci_dev *dev, int channel)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-generic/vmlinux.lds.h linux-2.6.18-xen.hg/include/asm-generic/vmlinux.lds.h
---- linux-2.6.18/include/asm-generic/vmlinux.lds.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-generic/vmlinux.lds.h 2007-12-23 11:15:40.084918931 +0100
+--- linux-2.6.18.8/include/asm-generic/pgtable.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-generic/pgtable.h 2008-05-19 00:34:06.499979257 +0300
+@@ -188,6 +188,10 @@
+ })
+ #endif
+
++#ifndef arch_change_pte_range
++#define arch_change_pte_range(mm, pmd, addr, end, newprot) 0
++#endif
++
+ #ifndef __ASSEMBLY__
+ /*
+ * When walking page tables, we usually want to skip any p?d_none entries;
+--- linux-2.6.18.8/include/asm-generic/vmlinux.lds.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-generic/vmlinux.lds.h 2008-05-19 00:34:06.607985483 +0300
@@ -194,3 +194,6 @@
.stab.index 0 : { *(.stab.index) } \
.stab.indexstr 0 : { *(.stab.indexstr) } \
+
+#define NOTES \
+ .notes : { *(.note.*) } :note
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/acpi.h linux-2.6.18-xen.hg/include/asm-i386/acpi.h
---- linux-2.6.18/include/asm-i386/acpi.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/acpi.h 2007-12-23 11:15:40.131588046 +0100
-@@ -177,11 +177,17 @@
- /* early initialization routine */
- extern void acpi_reserve_bootmem(void);
+--- linux-2.6.18.8/include/asm-i386/acpi.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/acpi.h 2008-05-19 00:34:07.420032291 +0300
+@@ -31,6 +31,9 @@
+ #include <acpi/pdc_intel.h>
-+#ifdef CONFIG_ACPI_PV_SLEEP
-+extern int acpi_notify_hypervisor_state(u8 sleep_state,
-+ u32 pm1a_cnt, u32 pm1b_cnt);
-+#endif /* CONFIG_ACPI_PV_SLEEP */
- #endif /*CONFIG_ACPI_SLEEP*/
+ #include <asm/system.h> /* defines cmpxchg */
++#ifdef CONFIG_XEN
++#include <xen/interface/platform.h>
++#endif
+
+ #define COMPILER_DEPENDENT_INT64 long long
+ #define COMPILER_DEPENDENT_UINT64 unsigned long long
+@@ -156,6 +159,27 @@
+ }
+ extern int acpi_irq_balance_set(char *str);
+
++#ifdef CONFIG_XEN
++static inline int acpi_notify_hypervisor_state(u8 sleep_state,
++ u32 pm1a_cnt_val,
++ u32 pm1b_cnt_val)
++{
++ struct xen_platform_op op = {
++ .cmd = XENPF_enter_acpi_sleep,
++ .interface_version = XENPF_INTERFACE_VERSION,
++ .u = {
++ .enter_acpi_sleep = {
++ .pm1a_cnt_val = pm1a_cnt_val,
++ .pm1b_cnt_val = pm1b_cnt_val,
++ .sleep_state = sleep_state,
++ },
++ },
++ };
++
++ return HYPERVISOR_platform_op(&op);
++}
++#endif /* CONFIG_XEN */
++
+ #else /* !CONFIG_ACPI */
+
+ #define acpi_lapic 0
+@@ -181,7 +205,9 @@
extern u8 x86_acpiid_to_apicid[];
#endif /*__KERNEL__*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/agp.h linux-2.6.18-xen.hg/include/asm-i386/agp.h
---- linux-2.6.18/include/asm-i386/agp.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/agp.h 2007-12-23 11:15:40.134921557 +0100
+--- linux-2.6.18.8/include/asm-i386/agp.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/agp.h 2008-05-19 00:34:07.420032291 +0300
@@ -12,8 +12,10 @@
* data corruption on some CPUs.
*/
#define flush_agp_mappings() global_flush_tlb()
/* Could use CLFLUSH here if the cpu supports it. But then it would
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/apic.h linux-2.6.18-xen.hg/include/asm-i386/apic.h
---- linux-2.6.18/include/asm-i386/apic.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/apic.h 2007-12-23 11:15:40.134921557 +0100
+--- linux-2.6.18.8/include/asm-i386/apic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/apic.h 2008-05-19 00:34:07.424032521 +0300
@@ -119,10 +119,12 @@
extern int disable_timer_pin_1;
extern int timer_over_8254;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/e820.h linux-2.6.18-xen.hg/include/asm-i386/e820.h
---- linux-2.6.18/include/asm-i386/e820.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/e820.h 2007-12-23 11:15:40.144922089 +0100
+--- linux-2.6.18.8/include/asm-i386/e820.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/e820.h 2008-05-19 00:34:07.464034827 +0300
@@ -38,6 +38,7 @@
extern int e820_all_mapped(unsigned long start, unsigned long end,
#endif/*!__ASSEMBLY__*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/fixmap.h linux-2.6.18-xen.hg/include/asm-i386/fixmap.h
---- linux-2.6.18/include/asm-i386/fixmap.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/fixmap.h 2007-12-23 11:15:40.148255594 +0100
+--- linux-2.6.18.8/include/asm-i386/elf.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/elf.h 2008-05-19 00:34:07.476035519 +0300
+@@ -169,50 +169,6 @@
+ NEW_AUX_ENT(AT_SYSINFO_EHDR, VDSO_COMPAT_BASE); \
+ } while (0)
+
+-/*
+- * These macros parameterize elf_core_dump in fs/binfmt_elf.c to write out
+- * extra segments containing the vsyscall DSO contents. Dumping its
+- * contents makes post-mortem fully interpretable later without matching up
+- * the same kernel and hardware config to see what PC values meant.
+- * Dumping its extra ELF program headers includes all the other information
+- * a debugger needs to easily find how the vsyscall DSO was being used.
+- */
+-#define ELF_CORE_EXTRA_PHDRS (VDSO_HIGH_EHDR->e_phnum)
+-#define ELF_CORE_WRITE_EXTRA_PHDRS \
+-do { \
+- const struct elf_phdr *const vsyscall_phdrs = \
+- (const struct elf_phdr *) (VDSO_HIGH_BASE \
+- + VDSO_HIGH_EHDR->e_phoff); \
+- int i; \
+- Elf32_Off ofs = 0; \
+- for (i = 0; i < VDSO_HIGH_EHDR->e_phnum; ++i) { \
+- struct elf_phdr phdr = vsyscall_phdrs[i]; \
+- if (phdr.p_type == PT_LOAD) { \
+- BUG_ON(ofs != 0); \
+- ofs = phdr.p_offset = offset; \
+- phdr.p_memsz = PAGE_ALIGN(phdr.p_memsz); \
+- phdr.p_filesz = phdr.p_memsz; \
+- offset += phdr.p_filesz; \
+- } \
+- else \
+- phdr.p_offset += ofs; \
+- phdr.p_paddr = 0; /* match other core phdrs */ \
+- DUMP_WRITE(&phdr, sizeof(phdr)); \
+- } \
+-} while (0)
+-#define ELF_CORE_WRITE_EXTRA_DATA \
+-do { \
+- const struct elf_phdr *const vsyscall_phdrs = \
+- (const struct elf_phdr *) (VDSO_HIGH_BASE \
+- + VDSO_HIGH_EHDR->e_phoff); \
+- int i; \
+- for (i = 0; i < VDSO_HIGH_EHDR->e_phnum; ++i) { \
+- if (vsyscall_phdrs[i].p_type == PT_LOAD) \
+- DUMP_WRITE((void *) vsyscall_phdrs[i].p_vaddr, \
+- PAGE_ALIGN(vsyscall_phdrs[i].p_memsz)); \
+- } \
+-} while (0)
+-
+ #endif
+
+ #endif
+--- linux-2.6.18.8/include/asm-i386/fixmap.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/fixmap.h 2008-05-19 00:34:07.480035749 +0300
@@ -19,7 +19,7 @@
* Leave one empty page between vmalloc'ed areas and
* the start of the fixmap.
#define set_fixmap(idx, phys) \
__set_fixmap(idx, phys, PAGE_KERNEL)
/*
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/kexec.h linux-2.6.18-xen.hg/include/asm-i386/kexec.h
---- linux-2.6.18/include/asm-i386/kexec.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/kexec.h 2007-12-23 11:15:40.194924713 +0100
+--- linux-2.6.18.8/include/asm-i386/io_apic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/io_apic.h 2008-05-19 00:34:07.492036441 +0300
+@@ -12,7 +12,7 @@
+
+ #ifdef CONFIG_X86_IO_APIC
+
+-#ifdef CONFIG_PCI_MSI
++#if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
+ static inline int use_pci_vector(void) {return 1;}
+ static inline void disable_edge_ioapic_vector(unsigned int vector) { }
+ static inline void mask_and_ack_level_ioapic_vector(unsigned int vector) { }
+--- linux-2.6.18.8/include/asm-i386/kexec.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/kexec.h 2008-05-19 00:34:07.512037594 +0300
@@ -1,6 +1,26 @@
#ifndef _I386_KEXEC_H
#define _I386_KEXEC_H
+#endif /* __ASSEMBLY__ */
#endif /* _I386_KEXEC_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-default/mach_traps.h linux-2.6.18-xen.hg/include/asm-i386/mach-default/mach_traps.h
---- linux-2.6.18/include/asm-i386/mach-default/mach_traps.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-default/mach_traps.h 2007-12-23 11:15:40.208258741 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-default/mach_traps.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-default/mach_traps.h 2008-05-19 00:34:07.532038747 +0300
@@ -15,6 +15,18 @@
outb(reason, 0x61);
}
static inline unsigned char get_nmi_reason(void)
{
return inb(0x61);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/agp.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/agp.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/agp.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/agp.h 2007-12-23 11:15:40.228259789 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/agp.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/agp.h 2008-05-19 00:34:07.692047970 +0300
@@ -0,0 +1,44 @@
+#ifndef AGP_H
+#define AGP_H 1
+ dma_free_coherent(NULL,PAGE_SIZE<<(order),(table),virt_to_bus(table))
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/desc.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/desc.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/desc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/desc.h 2007-12-23 11:15:40.228259789 +0100
-@@ -0,0 +1,164 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/desc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/desc.h 2008-05-19 00:34:07.692047970 +0300
+@@ -0,0 +1,166 @@
+#ifndef __ARCH_DESC_H
+#define __ARCH_DESC_H
+
+
+static inline void load_TLS(struct thread_struct *t, unsigned int cpu)
+{
-+#define C(i) HYPERVISOR_update_descriptor(virt_to_machine(&get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + i]), *(u64 *)&t->tls_array[i])
++#define C(i) if (HYPERVISOR_update_descriptor(virt_to_machine(&get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + i]), \
++ *(u64 *)&t->tls_array[i])) \
++ BUG();
+ C(0); C(1); C(2);
+#undef C
+}
+ * it slows down context switching. Noone uses it anyway.
+ */
+ cpu = cpu; /* XXX avoid compiler warning */
-+ xen_set_ldt(0UL, 0);
++ xen_set_ldt(NULL, 0);
+ put_cpu();
+}
+
+ if (likely(!count))
+ segments = NULL;
+
-+ xen_set_ldt((unsigned long)segments, count);
++ xen_set_ldt(segments, count);
+}
+
+static inline void load_LDT(mm_context_t *pc)
+#endif /* !__ASSEMBLY__ */
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/dma-mapping.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/dma-mapping.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/dma-mapping.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/dma-mapping.h 2007-12-23 11:15:40.228259789 +0100
-@@ -0,0 +1,157 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/dma-mapping.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/dma-mapping.h 2008-05-19 00:34:07.692047970 +0300
+@@ -0,0 +1,151 @@
+#ifndef _ASM_I386_DMA_MAPPING_H
+#define _ASM_I386_DMA_MAPPING_H
+
+ return (addr & ~mask) != 0;
+}
+
-+static inline int
-+range_straddles_page_boundary(paddr_t p, size_t size)
-+{
-+ extern unsigned long *contiguous_bitmap;
-+ return ((((p & ~PAGE_MASK) + size) > PAGE_SIZE) &&
-+ !test_bit(p >> PAGE_SHIFT, contiguous_bitmap));
-+}
++extern int range_straddles_page_boundary(paddr_t p, size_t size);
+
+#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)
+#define dma_free_noncoherent(d, s, v, h) dma_free_coherent(d, s, v, h)
+ dma_addr_t device_addr, size_t size);
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/fixmap.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/fixmap.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/fixmap.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/fixmap.h 2007-12-23 11:15:40.228259789 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/fixmap.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/fixmap.h 2008-05-19 00:34:07.764052120 +0300
@@ -0,0 +1,155 @@
+/*
+ * fixmap.h: compile-time virtual memory allocation
+
+#endif /* !__ASSEMBLY__ */
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/gnttab_dma.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/gnttab_dma.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/gnttab_dma.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/gnttab_dma.h 2007-12-23 11:15:40.231593300 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/gnttab_dma.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/gnttab_dma.h 2008-05-19 00:34:07.764052120 +0300
@@ -0,0 +1,41 @@
+/*
+ * Copyright (c) 2007 Herbert Xu <herbert@gondor.apana.org.au>
+}
+
+#endif /* _ASM_I386_GNTTAB_DMA_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/highmem.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/highmem.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/highmem.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/highmem.h 2007-12-23 11:15:40.231593300 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/highmem.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/highmem.h 2008-05-19 00:34:07.812054887 +0300
@@ -0,0 +1,80 @@
+/*
+ * highmem.h: virtual kernel memory mappings for high memory
+#endif /* __KERNEL__ */
+
+#endif /* _ASM_HIGHMEM_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/hw_irq.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hw_irq.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/hw_irq.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hw_irq.h 2007-12-23 11:15:40.231593300 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/hw_irq.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hw_irq.h 2008-05-19 00:34:07.812054887 +0300
@@ -0,0 +1,72 @@
+#ifndef _ASM_HW_IRQ_H
+#define _ASM_HW_IRQ_H
+#define IO_APIC_IRQ(x) (((x) >= 16) || ((1<<(x)) & io_apic_irqs))
+
+#endif /* _ASM_HW_IRQ_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/hypercall.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypercall.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/hypercall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypercall.h 2007-12-23 11:15:40.231593300 +0100
-@@ -0,0 +1,408 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/hypercall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypercall.h 2008-05-19 00:34:07.812054887 +0300
+@@ -0,0 +1,416 @@
+/******************************************************************************
+ * hypercall.h
+ *
+# error "please don't include this file directly"
+#endif
+
-+#define __STR(x) #x
-+#define STR(x) __STR(x)
-+
+#ifdef CONFIG_XEN
+#define HYPERCALL_STR(name) \
-+ "call hypercall_page + ("STR(__HYPERVISOR_##name)" * 32)"
++ "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
+#else
+#define HYPERCALL_STR(name) \
-+ "mov "__stringify(hypercall_stubs)",%%eax; " \
-+ "add $("STR(__HYPERVISOR_##name)" * 32),%%eax; " \
++ "mov hypercall_stubs,%%eax; " \
++ "add $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
+ "call *%%eax"
+#endif
+
+#define _hypercall0(type, name) \
+({ \
-+ long __res; \
++ type __res; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res) \
+ : \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall1(type, name, a1) \
+({ \
-+ long __res, __ign1; \
++ type __res; \
++ long __ign1; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=b" (__ign1) \
+ : "1" ((long)(a1)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall2(type, name, a1, a2) \
+({ \
-+ long __res, __ign1, __ign2; \
++ type __res; \
++ long __ign1, __ign2; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=b" (__ign1), "=c" (__ign2) \
+ : "1" ((long)(a1)), "2" ((long)(a2)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall3(type, name, a1, a2, a3) \
+({ \
-+ long __res, __ign1, __ign2, __ign3; \
++ type __res; \
++ long __ign1, __ign2, __ign3; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
+ : "1" ((long)(a1)), "2" ((long)(a2)), \
+ "3" ((long)(a3)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall4(type, name, a1, a2, a3, a4) \
+({ \
-+ long __res, __ign1, __ign2, __ign3, __ign4; \
++ type __res; \
++ long __ign1, __ign2, __ign3, __ign4; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
+ : "1" ((long)(a1)), "2" ((long)(a2)), \
+ "3" ((long)(a3)), "4" ((long)(a4)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
+({ \
-+ long __res, __ign1, __ign2, __ign3, __ign4, __ign5; \
++ type __res; \
++ long __ign1, __ign2, __ign3, __ign4, __ign5; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
+ "3" ((long)(a3)), "4" ((long)(a4)), \
+ "5" ((long)(a5)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_trap_table(
-+ trap_info_t *table)
++ const trap_info_t *table)
+{
+ return _hypercall1(int, set_trap_table, table);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_mmu_update(
-+ mmu_update_t *req, int count, int *success_count, domid_t domid)
++ mmu_update_t *req, unsigned int count, unsigned int *success_count,
++ domid_t domid)
+{
+ return _hypercall4(int, mmu_update, req, count, success_count, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_mmuext_op(
-+ struct mmuext_op *op, int count, int *success_count, domid_t domid)
++ struct mmuext_op *op, unsigned int count, unsigned int *success_count,
++ domid_t domid)
+{
+ return _hypercall4(int, mmuext_op, op, count, success_count, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_gdt(
-+ unsigned long *frame_list, int entries)
++ unsigned long *frame_list, unsigned int entries)
+{
+ return _hypercall2(int, set_gdt, frame_list, entries);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_stack_switch(
+ unsigned long ss, unsigned long esp)
+{
+ return _hypercall2(int, stack_switch, ss, esp);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_callbacks(
+ unsigned long event_selector, unsigned long event_address,
+ unsigned long failsafe_selector, unsigned long failsafe_address)
+ return _hypercall1(int, fpu_taskswitch, set);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_sched_op_compat(
+ int cmd, unsigned long arg)
+{
+ return _hypercall2(int, sched_op_compat, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_sched_op(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, sched_op, cmd, arg);
+}
+
-+static inline long
++static inline long __must_check
+HYPERVISOR_set_timer_op(
+ u64 timeout)
+{
+ return _hypercall2(long, set_timer_op, timeout_lo, timeout_hi);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_platform_op(
+ struct xen_platform_op *platform_op)
+{
+ return _hypercall1(int, platform_op, platform_op);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_debugreg(
-+ int reg, unsigned long value)
++ unsigned int reg, unsigned long value)
+{
+ return _hypercall2(int, set_debugreg, reg, value);
+}
+
-+static inline unsigned long
++static inline unsigned long __must_check
+HYPERVISOR_get_debugreg(
-+ int reg)
++ unsigned int reg)
+{
+ return _hypercall1(unsigned long, get_debugreg, reg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_descriptor(
+ u64 ma, u64 desc)
+{
+ return _hypercall4(int, update_descriptor, ma, ma>>32, desc, desc>>32);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_memory_op(
+ unsigned int cmd, void *arg)
+{
+ return _hypercall2(int, memory_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_multicall(
-+ multicall_entry_t *call_list, int nr_calls)
++ multicall_entry_t *call_list, unsigned int nr_calls)
+{
+ return _hypercall2(int, multicall, call_list, nr_calls);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_va_mapping(
+ unsigned long va, pte_t new_val, unsigned long flags)
+{
+ new_val.pte_low, pte_hi, flags);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_event_channel_op(
+ int cmd, void *arg)
+{
+ return rc;
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_acm_op(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, acm_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_xen_version(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, xen_version, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_console_io(
-+ int cmd, int count, char *str)
++ int cmd, unsigned int count, char *str)
+{
+ return _hypercall3(int, console_io, cmd, count, str);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_physdev_op(
+ int cmd, void *arg)
+{
+ return rc;
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_grant_table_op(
+ unsigned int cmd, void *uop, unsigned int count)
+{
+ return _hypercall3(int, grant_table_op, cmd, uop, count);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_va_mapping_otherdomain(
+ unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
+{
+ new_val.pte_low, pte_hi, flags, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_vm_assist(
+ unsigned int cmd, unsigned int type)
+{
+ return _hypercall2(int, vm_assist, cmd, type);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_vcpu_op(
-+ int cmd, int vcpuid, void *extra_args)
++ int cmd, unsigned int vcpuid, void *extra_args)
+{
+ return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_suspend(
+ unsigned long srec)
+{
+ return rc;
+}
+
++#if CONFIG_XEN_COMPAT <= 0x030002
+static inline int
+HYPERVISOR_nmi_op(
+ unsigned long op, void *arg)
+{
+ return _hypercall2(int, nmi_op, op, arg);
+}
++#endif
+
-+static inline unsigned long
++#ifndef CONFIG_XEN
++static inline unsigned long __must_check
+HYPERVISOR_hvm_op(
+ int op, void *arg)
+{
+ return _hypercall2(unsigned long, hvm_op, op, arg);
+}
++#endif
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_callback_op(
-+ int cmd, void *arg)
++ int cmd, const void *arg)
+{
+ return _hypercall2(int, callback_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_xenoprof_op(
+ int op, void *arg)
+{
+ return _hypercall2(int, xenoprof_op, op, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_kexec_op(
+ unsigned long op, void *args)
+{
+
+
+#endif /* __HYPERCALL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/hypervisor.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypervisor.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/hypervisor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypervisor.h 2007-12-23 11:15:40.231593300 +0100
-@@ -0,0 +1,255 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/hypervisor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/hypervisor.h 2008-05-19 00:34:07.816055118 +0300
+@@ -0,0 +1,259 @@
+/******************************************************************************
+ * hypervisor.h
+ *
+void xen_pgd_pin(unsigned long ptr);
+void xen_pgd_unpin(unsigned long ptr);
+
-+void xen_set_ldt(unsigned long ptr, unsigned long bytes);
++void xen_set_ldt(const void *ptr, unsigned int ents);
+
+#ifdef CONFIG_SMP
+#include <linux/cpumask.h>
+u64 jiffies_to_st(unsigned long jiffies);
+
+#ifdef CONFIG_XEN_SCRUB_PAGES
-+#define scrub_pages(_p,_n) memset((void *)(_p), 0, (_n) << PAGE_SHIFT)
++void scrub_pages(void *, unsigned int);
+#else
+#define scrub_pages(_p,_n) ((void)0)
+#endif
+#define MULTI_UVMDOMID_INDEX 4
+#endif
+
++#ifdef CONFIG_XEN
+#define is_running_on_xen() 1
++#else
++extern char *hypercall_stubs;
++#define is_running_on_xen() (!!hypercall_stubs)
++#endif
+
+static inline int
+HYPERVISOR_yield(
+ return rc;
+}
+
-+static inline int
++static inline void /*__noreturn*/
+HYPERVISOR_shutdown(
+ unsigned int reason)
+{
+ .reason = reason
+ };
+
-+ int rc = HYPERVISOR_sched_op(SCHEDOP_shutdown, &sched_shutdown);
-+
++ VOID(HYPERVISOR_sched_op(SCHEDOP_shutdown, &sched_shutdown));
+#if CONFIG_XEN_COMPAT <= 0x030002
-+ if (rc == -ENOSYS)
-+ rc = HYPERVISOR_sched_op_compat(SCHEDOP_shutdown, reason);
++ VOID(HYPERVISOR_sched_op_compat(SCHEDOP_shutdown, reason));
+#endif
-+
-+ return rc;
++ /* Don't recurse needlessly. */
++ BUG_ON(reason != SHUTDOWN_crash);
++ for(;;);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_poll(
+ evtchn_port_t *ports, unsigned int nr_ports, u64 timeout)
+{
+#endif
+
+#endif /* __HYPERVISOR_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/io.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/io.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/io.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/io.h 2007-12-23 11:15:40.231593300 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/io.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/io.h 2008-05-19 00:34:07.816055118 +0300
@@ -0,0 +1,389 @@
+#ifndef _ASM_IO_H
+#define _ASM_IO_H
+#define ARCH_HAS_DEV_MEM
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/irqflags.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/irqflags.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/irqflags.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/irqflags.h 2007-12-23 11:15:40.231593300 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/irqflags.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/irqflags.h 2008-05-19 00:34:07.816055118 +0300
@@ -0,0 +1,127 @@
+/*
+ * include/asm-i386/irqflags.h
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/maddr.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/maddr.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/maddr.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/maddr.h 2007-12-23 11:15:40.234926812 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/maddr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/maddr.h 2008-05-19 00:34:07.816055118 +0300
@@ -0,0 +1,193 @@
+#ifndef _I386_MADDR_H
+#define _I386_MADDR_H
+#endif
+
+#ifdef CONFIG_X86_PAE
++#define __pte_ma(x) ((pte_t) { (x), (maddr_t)(x) >> 32 } )
+static inline pte_t pfn_pte_ma(unsigned long page_nr, pgprot_t pgprot)
+{
+ pte_t pte;
+ return pte;
+}
+#else
++#define __pte_ma(x) ((pte_t) { (x) } )
+#define pfn_pte_ma(pfn, prot) __pte_ma(((pfn) << PAGE_SHIFT) | pgprot_val(prot))
+#endif
+
-+#define __pte_ma(x) ((pte_t) { (x) } )
-+
+#else /* !CONFIG_XEN */
+
+#define pfn_to_mfn(pfn) (pfn)
+#define mfn_to_virt(m) (__va(mfn_to_pfn(m) << PAGE_SHIFT))
+
+#endif /* _I386_MADDR_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/mmu_context.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu_context.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/mmu_context.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu_context.h 2007-12-23 11:15:40.234926812 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/mmu.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu.h 2008-05-19 00:34:07.820055349 +0300
+@@ -0,0 +1,29 @@
++#ifndef __i386_MMU_H
++#define __i386_MMU_H
++
++#include <asm/semaphore.h>
++/*
++ * The i386 doesn't have a mmu context, but
++ * we put the segment information here.
++ *
++ * cpu_vm_mask is used to optimize ldt flushing.
++ */
++typedef struct {
++ int size;
++ struct semaphore sem;
++ void *ldt;
++ void *vdso;
++#ifdef CONFIG_XEN
++ int has_foreign_mappings;
++#endif
++} mm_context_t;
++
++/* mm/memory.c:exit_mmap hook */
++extern void _arch_exit_mmap(struct mm_struct *mm);
++#define arch_exit_mmap(_mm) _arch_exit_mmap(_mm)
++
++/* kernel/fork.c:dup_mmap hook */
++extern void _arch_dup_mmap(struct mm_struct *mm);
++#define arch_dup_mmap(mm, oldmm) ((void)(oldmm), _arch_dup_mmap(mm))
++
++#endif
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/mmu_context.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu_context.h 2008-05-19 00:34:07.820055349 +0300
@@ -0,0 +1,108 @@
+#ifndef __I386_SCHED_H
+#define __I386_SCHED_H
+}
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/mmu.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/mmu.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/mmu.h 2007-12-23 11:15:40.234926812 +0100
-@@ -0,0 +1,29 @@
-+#ifndef __i386_MMU_H
-+#define __i386_MMU_H
-+
-+#include <asm/semaphore.h>
-+/*
-+ * The i386 doesn't have a mmu context, but
-+ * we put the segment information here.
-+ *
-+ * cpu_vm_mask is used to optimize ldt flushing.
-+ */
-+typedef struct {
-+ int size;
-+ struct semaphore sem;
-+ void *ldt;
-+ void *vdso;
-+#ifdef CONFIG_XEN
-+ int has_foreign_mappings;
-+#endif
-+} mm_context_t;
-+
-+/* mm/memory.c:exit_mmap hook */
-+extern void _arch_exit_mmap(struct mm_struct *mm);
-+#define arch_exit_mmap(_mm) _arch_exit_mmap(_mm)
-+
-+/* kernel/fork.c:dup_mmap hook */
-+extern void _arch_dup_mmap(struct mm_struct *mm);
-+#define arch_dup_mmap(mm, oldmm) ((void)(oldmm), _arch_dup_mmap(mm))
-+
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/page.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/page.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/page.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/page.h 2007-12-23 11:15:40.234926812 +0100
-@@ -0,0 +1,227 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/page.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/page.h 2008-05-19 00:34:07.820055349 +0300
+@@ -0,0 +1,233 @@
+#ifndef _I386_PAGE_H
+#define _I386_PAGE_H
+
+ * below. The preprocessor will warn if the two definitions aren't identical.
+ */
+#define _PAGE_PRESENT 0x001
++#define _PAGE_IO 0x200
+
+#ifndef __ASSEMBLY__
+
+typedef struct { unsigned long long pgprot; } pgprot_t;
+#define pgprot_val(x) ((x).pgprot)
+#include <asm/maddr.h>
-+#define __pte(x) ({ unsigned long long _x = (x); \
-+ if (_x & _PAGE_PRESENT) _x = pte_phys_to_machine(_x); \
++#define __pte(x) ({ unsigned long long _x = (x); \
++ if ((_x & (_PAGE_PRESENT|_PAGE_IO)) == _PAGE_PRESENT) \
++ _x = pte_phys_to_machine(_x); \
+ ((pte_t) {(unsigned long)(_x), (unsigned long)(_x>>32)}); })
+#define __pgd(x) ({ unsigned long long _x = (x); \
+ (pgd_t) {((_x) & _PAGE_PRESENT) ? pte_phys_to_machine(_x) : (_x)}; })
+static inline unsigned long long pte_val(pte_t x)
+{
+ unsigned long long ret = __pte_val(x);
-+ if (x.pte_low & _PAGE_PRESENT) ret = pte_machine_to_phys(ret);
++ if ((x.pte_low & (_PAGE_PRESENT|_PAGE_IO)) == _PAGE_PRESENT)
++ ret = pte_machine_to_phys(ret);
+ return ret;
+}
+#define __pmd_val(x) ((x).pmd)
+#include <asm/maddr.h>
+#define boot_pte_t pte_t /* or would you rather have a typedef */
+#define __pte_val(x) ((x).pte_low)
-+#define pte_val(x) (__pte_val(x) & _PAGE_PRESENT ? \
-+ machine_to_phys(__pte_val(x)) : \
-+ __pte_val(x))
-+#define __pte(x) ({ unsigned long _x = (x); \
-+ (pte_t) {((_x) & _PAGE_PRESENT) ? phys_to_machine(_x) : (_x)}; })
++#define pte_val(x) ((__pte_val(x) & (_PAGE_PRESENT|_PAGE_IO)) \
++ == _PAGE_PRESENT ? \
++ machine_to_phys(__pte_val(x)) : \
++ __pte_val(x))
++#define __pte(x) ({ unsigned long _x = (x); \
++ if ((_x & (_PAGE_PRESENT|_PAGE_IO)) == _PAGE_PRESENT) \
++ _x = phys_to_machine(_x); \
++ ((pte_t) { _x }); })
+#define __pmd_val(x) __pud_val((x).pud)
+#define __pud_val(x) __pgd_val((x).pgd)
+#define __pgd(x) ({ unsigned long _x = (x); \
+#endif /* __KERNEL__ */
+
+#endif /* _I386_PAGE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pci.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pci.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pci.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pci.h 2007-12-23 11:15:40.234926812 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pci.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pci.h 2008-05-19 00:34:07.820055349 +0300
@@ -0,0 +1,148 @@
+#ifndef __i386_PCI_H
+#define __i386_PCI_H
+#include <asm-generic/pci.h>
+
+#endif /* __i386_PCI_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgalloc.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgalloc.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgalloc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgalloc.h 2007-12-23 11:15:40.234926812 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgalloc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgalloc.h 2008-05-19 00:34:07.820055349 +0300
@@ -0,0 +1,59 @@
+#ifndef _I386_PGALLOC_H
+#define _I386_PGALLOC_H
+#define check_pgt_cache() do { } while (0)
+
+#endif /* _I386_PGALLOC_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h 2007-12-23 11:15:40.238260322 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level-defs.h 2008-05-19 00:34:07.824055579 +0300
@@ -0,0 +1,20 @@
+#ifndef _I386_PGTABLE_2LEVEL_DEFS_H
+#define _I386_PGTABLE_2LEVEL_DEFS_H
+#define PTRS_PER_PTE 1024
+
+#endif /* _I386_PGTABLE_2LEVEL_DEFS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-2level.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-2level.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level.h 2007-12-23 11:15:40.238260322 +0100
-@@ -0,0 +1,119 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgtable-2level.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-2level.h 2008-05-19 00:34:07.824055579 +0300
+@@ -0,0 +1,121 @@
+#ifndef _I386_PGTABLE_2LEVEL_H
+#define _I386_PGTABLE_2LEVEL_H
+
+#define __pte_mfn(_pte) ((_pte).pte_low >> PAGE_SHIFT)
+#define pte_mfn(_pte) ((_pte).pte_low & _PAGE_PRESENT ? \
+ __pte_mfn(_pte) : pfn_to_mfn(__pte_mfn(_pte)))
-+#define pte_pfn(_pte) ((_pte).pte_low & _PAGE_PRESENT ? \
-+ mfn_to_local_pfn(__pte_mfn(_pte)) : __pte_mfn(_pte))
++#define pte_pfn(_pte) ((_pte).pte_low & _PAGE_IO ? max_mapnr : \
++ (_pte).pte_low & _PAGE_PRESENT ? \
++ mfn_to_local_pfn(__pte_mfn(_pte)) : \
++ __pte_mfn(_pte))
+
+#define pte_page(_pte) pfn_to_page(pte_pfn(_pte))
+
+void vmalloc_sync_all(void);
+
+#endif /* _I386_PGTABLE_2LEVEL_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h 2007-12-23 11:15:40.238260322 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level-defs.h 2008-05-19 00:34:07.824055579 +0300
@@ -0,0 +1,24 @@
+#ifndef _I386_PGTABLE_3LEVEL_DEFS_H
+#define _I386_PGTABLE_3LEVEL_DEFS_H
+#define PTRS_PER_PTE 512
+
+#endif /* _I386_PGTABLE_3LEVEL_DEFS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-3level.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable-3level.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level.h 2007-12-23 11:15:40.238260322 +0100
-@@ -0,0 +1,209 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgtable-3level.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable-3level.h 2008-05-19 00:34:07.824055579 +0300
+@@ -0,0 +1,211 @@
+#ifndef _I386_PGTABLE_3LEVEL_H
+#define _I386_PGTABLE_3LEVEL_H
+
+ ((_pte).pte_high << (32-PAGE_SHIFT)))
+#define pte_mfn(_pte) ((_pte).pte_low & _PAGE_PRESENT ? \
+ __pte_mfn(_pte) : pfn_to_mfn(__pte_mfn(_pte)))
-+#define pte_pfn(_pte) ((_pte).pte_low & _PAGE_PRESENT ? \
-+ mfn_to_local_pfn(__pte_mfn(_pte)) : __pte_mfn(_pte))
++#define pte_pfn(_pte) ((_pte).pte_low & _PAGE_IO ? max_mapnr : \
++ (_pte).pte_low & _PAGE_PRESENT ? \
++ mfn_to_local_pfn(__pte_mfn(_pte)) : \
++ __pte_mfn(_pte))
+
+extern unsigned long long __supported_pte_mask;
+
+void vmalloc_sync_all(void);
+
+#endif /* _I386_PGTABLE_3LEVEL_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/pgtable.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable.h 2007-12-23 11:15:40.238260322 +0100
-@@ -0,0 +1,531 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/pgtable.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/pgtable.h 2008-05-19 00:34:07.828055810 +0300
+@@ -0,0 +1,537 @@
+#ifndef _I386_PGTABLE_H
+#define _I386_PGTABLE_H
+
+#define _PAGE_BIT_DIRTY 6
+#define _PAGE_BIT_PSE 7 /* 4 MB (or 2MB) page, Pentium+, if present.. */
+#define _PAGE_BIT_GLOBAL 8 /* Global TLB entry PPro+ */
-+#define _PAGE_BIT_UNUSED1 9 /* available for programmer */
++/*#define _PAGE_BIT_UNUSED1 9*/ /* available for programmer */
+#define _PAGE_BIT_UNUSED2 10
+#define _PAGE_BIT_UNUSED3 11
+#define _PAGE_BIT_NX 63
+#define _PAGE_DIRTY 0x040
+#define _PAGE_PSE 0x080 /* 4 MB (or 2MB) page, Pentium+, if present.. */
+#define _PAGE_GLOBAL 0x100 /* Global TLB entry PPro+ */
-+#define _PAGE_UNUSED1 0x200 /* available for programmer */
++/*#define _PAGE_UNUSED1 0x200*/ /* available for programmer */
+#define _PAGE_UNUSED2 0x400
+#define _PAGE_UNUSED3 0x800
+
+#define _PAGE_NX 0
+#endif
+
++/* Mapped page is I/O or foreign and has no associated page struct. */
++#define _PAGE_IO 0x200
++
+#define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
+#define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
-+#define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
++#define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_IO)
+
+#define PAGE_NONE \
+ __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
+void make_pages_readonly(void *va, unsigned int nr, unsigned int feature);
+void make_pages_writable(void *va, unsigned int nr, unsigned int feature);
+
-+#define virt_to_ptep(__va) \
++#define virt_to_ptep(va) \
+({ \
-+ pgd_t *__pgd = pgd_offset_k((unsigned long)(__va)); \
-+ pud_t *__pud = pud_offset(__pgd, (unsigned long)(__va)); \
-+ pmd_t *__pmd = pmd_offset(__pud, (unsigned long)(__va)); \
-+ pte_offset_kernel(__pmd, (unsigned long)(__va)); \
++ pte_t *__ptep = lookup_address((unsigned long)(va)); \
++ BUG_ON(!__ptep || !pte_present(*__ptep)); \
++ __ptep; \
+})
+
-+#define arbitrary_virt_to_machine(__va) \
-+({ \
-+ maddr_t m = (maddr_t)pte_mfn(*virt_to_ptep(__va)) << PAGE_SHIFT;\
-+ m | ((unsigned long)(__va) & (PAGE_SIZE-1)); \
-+})
++#define arbitrary_virt_to_machine(va) \
++ (((maddr_t)pte_mfn(*virt_to_ptep(va)) << PAGE_SHIFT) \
++ | ((unsigned long)(va) & (PAGE_SIZE - 1)))
+
+#endif /* !__ASSEMBLY__ */
+
+ unsigned long address,
+ unsigned long size);
+
++int xen_change_pte_range(struct mm_struct *mm, pmd_t *pmd,
++ unsigned long addr, unsigned long end, pgprot_t newprot);
++
++#define arch_change_pte_range(mm, pmd, addr, end, newprot) \
++ xen_change_pte_range(mm, pmd, addr, end, newprot)
++
+#define io_remap_pfn_range(vma,from,pfn,size,prot) \
+direct_remap_pfn_range(vma,from,pfn,size,prot,DOMID_IO)
+
+#include <asm-generic/pgtable.h>
+
+#endif /* _I386_PGTABLE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/processor.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/processor.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/processor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/processor.h 2007-12-23 11:15:40.238260322 +0100
-@@ -0,0 +1,741 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/processor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/processor.h 2008-05-19 00:34:07.828055810 +0300
+@@ -0,0 +1,743 @@
+/*
+ * include/asm-i386/processor.h
+ *
+#define load_esp0(tss, thread) \
+ __load_esp0(tss, thread)
+#else
-+#define load_esp0(tss, thread) \
-+ HYPERVISOR_stack_switch(__KERNEL_DS, (thread)->esp0)
++#define load_esp0(tss, thread) do { \
++ if (HYPERVISOR_stack_switch(__KERNEL_DS, (thread)->esp0)) \
++ BUG(); \
++} while (0)
+#endif
+
+#define start_thread(regs, new_eip, new_esp) do { \
+#define get_debugreg(var, register) \
+ (var) = HYPERVISOR_get_debugreg((register))
+#define set_debugreg(value, register) \
-+ HYPERVISOR_set_debugreg((register), (value))
++ WARN_ON(HYPERVISOR_set_debugreg((register), (value)))
+
+/*
+ * Set IOPL bits in EFLAGS from given mask
+
+ /* Force the change at ring 0. */
+ set_iopl.iopl = (mask == 0) ? 1 : (mask >> 12) & 3;
-+ HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
++ WARN_ON(HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl));
+}
+
+/* Forward declaration, a strange C thing */
+extern int sysenter_setup(void);
+
+#endif /* __ASM_I386_PROCESSOR_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/ptrace.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/ptrace.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/ptrace.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/ptrace.h 2007-12-23 11:15:40.238260322 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/ptrace.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/ptrace.h 2008-05-19 00:34:07.828055810 +0300
@@ -0,0 +1,90 @@
+#ifndef _I386_PTRACE_H
+#define _I386_PTRACE_H
+#endif /* __KERNEL__ */
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/scatterlist.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/scatterlist.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/scatterlist.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/scatterlist.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/scatterlist.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/scatterlist.h 2008-05-19 00:34:07.832056040 +0300
@@ -0,0 +1,22 @@
+#ifndef _I386_SCATTERLIST_H
+#define _I386_SCATTERLIST_H
+#define ISA_DMA_THRESHOLD (0x00ffffff)
+
+#endif /* !(_I386_SCATTERLIST_H) */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/segment.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/segment.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/segment.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/segment.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/segment.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/segment.h 2008-05-19 00:34:07.832056040 +0300
@@ -0,0 +1,117 @@
+#ifndef _ASM_SEGMENT_H
+#define _ASM_SEGMENT_H
+#define IDT_ENTRIES 256
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/setup.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/setup.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/setup.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/setup.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/setup.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/setup.h 2008-05-19 00:34:07.832056040 +0300
@@ -0,0 +1,81 @@
+/*
+ * Just a place holder. We don't want to have to test x86 before
+#endif /* __ASSEMBLY__ */
+
+#endif /* _i386_SETUP_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/smp.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/smp.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/smp.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/smp.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/smp.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/smp.h 2008-05-19 00:34:07.832056040 +0300
@@ -0,0 +1,103 @@
+#ifndef __ASM_SMP_H
+#define __ASM_SMP_H
+
+#endif
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/spinlock.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/spinlock.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/spinlock.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/spinlock.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/spinlock.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/spinlock.h 2008-05-19 00:34:07.836056271 +0300
@@ -0,0 +1,202 @@
+#ifndef __ASM_SPINLOCK_H
+#define __ASM_SPINLOCK_H
+}
+
+#endif /* __ASM_SPINLOCK_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/swiotlb.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/swiotlb.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/swiotlb.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/swiotlb.h 2007-12-23 11:15:40.241593831 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/swiotlb.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/swiotlb.h 2008-05-19 00:34:07.836056271 +0300
@@ -0,0 +1,43 @@
+#ifndef _ASM_SWIOTLB_H
+#define _ASM_SWIOTLB_H 1
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/synch_bitops.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/synch_bitops.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/synch_bitops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/synch_bitops.h 2007-12-23 11:15:40.244927335 +0100
-@@ -0,0 +1,145 @@
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/synch_bitops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/synch_bitops.h 2008-05-19 00:34:07.836056271 +0300
+@@ -0,0 +1,126 @@
+#ifndef __XEN_SYNCH_BITOPS_H__
+#define __XEN_SYNCH_BITOPS_H__
+
+ return old;
+}
+
-+static __always_inline int synch_const_test_bit(int nr,
-+ const volatile void * addr)
-+{
-+ return ((1UL << (nr & 31)) &
-+ (((const volatile unsigned int *) addr)[nr >> 5])) != 0;
-+}
-+
-+static __inline__ int synch_var_test_bit(int nr, volatile void * addr)
-+{
-+ int oldbit;
-+ __asm__ __volatile__ (
-+ "btl %2,%1\n\tsbbl %0,%0"
-+ : "=r" (oldbit) : "m" (ADDR), "Ir" (nr) );
-+ return oldbit;
-+}
-+
-+#define synch_test_bit(nr,addr) \
-+(__builtin_constant_p(nr) ? \
-+ synch_const_test_bit((nr),(addr)) : \
-+ synch_var_test_bit((nr),(addr)))
++#define synch_test_bit test_bit
+
+#define synch_cmpxchg_subword synch_cmpxchg
+
+#endif /* __XEN_SYNCH_BITOPS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/system.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/system.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/system.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/system.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/system.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/system.h 2008-05-19 00:34:07.836056271 +0300
@@ -0,0 +1,488 @@
+#ifndef __ASM_SYSTEM_H
+#define __ASM_SYSTEM_H
+void default_idle(void);
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/tlbflush.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/tlbflush.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/tlbflush.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/tlbflush.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/tlbflush.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/tlbflush.h 2008-05-19 00:34:07.840056501 +0300
@@ -0,0 +1,101 @@
+#ifndef _I386_TLBFLUSH_H
+#define _I386_TLBFLUSH_H
+}
+
+#endif /* _I386_TLBFLUSH_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/vga.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/vga.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/vga.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/vga.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/vga.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/vga.h 2008-05-19 00:34:07.840056501 +0300
@@ -0,0 +1,20 @@
+/*
+ * Access to VGA videoram
+#define vga_writeb(x,y) (*(y) = (x))
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/asm/xenoprof.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/xenoprof.h
---- linux-2.6.18/include/asm-i386/mach-xen/asm/xenoprof.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/xenoprof.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/asm/xenoprof.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/asm/xenoprof.h 2008-05-19 00:34:07.840056501 +0300
@@ -0,0 +1,48 @@
+/******************************************************************************
+ * asm-i386/mach-xen/asm/xenoprof.h
+
+#endif /* CONFIG_XEN */
+#endif /* __ASM_XENOPROF_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/irq_vectors.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/irq_vectors.h
---- linux-2.6.18/include/asm-i386/mach-xen/irq_vectors.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/irq_vectors.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/irq_vectors.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/irq_vectors.h 2008-05-19 00:34:07.840056501 +0300
@@ -0,0 +1,125 @@
+/*
+ * This file should contain #defines for all of the interrupt vector
+#define irq_to_dynirq(_x) ((_x) - DYNIRQ_BASE)
+
+#endif /* _ASM_IRQ_VECTORS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/mach_traps.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/mach_traps.h
---- linux-2.6.18/include/asm-i386/mach-xen/mach_traps.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/mach_traps.h 2007-12-23 11:15:40.244927335 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/mach_traps.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/mach_traps.h 2008-05-19 00:34:07.844056732 +0300
@@ -0,0 +1,33 @@
+/*
+ * include/asm-xen/asm-i386/mach-xen/mach_traps.h
+static inline void reassert_nmi(void) {}
+
+#endif /* !_MACH_TRAPS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/mach-xen/setup_arch.h linux-2.6.18-xen.hg/include/asm-i386/mach-xen/setup_arch.h
---- linux-2.6.18/include/asm-i386/mach-xen/setup_arch.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/setup_arch.h 2007-12-23 11:15:40.248260841 +0100
+--- linux-2.6.18.8/include/asm-i386/mach-xen/setup_arch.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/mach-xen/setup_arch.h 2008-05-19 00:34:07.844056732 +0300
@@ -0,0 +1,5 @@
+/* Hook to call BIOS initialisation function */
+
+#define ARCH_SETUP machine_specific_arch_setup();
+
+void __init machine_specific_arch_setup(void);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/page.h linux-2.6.18-xen.hg/include/asm-i386/page.h
---- linux-2.6.18/include/asm-i386/page.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/page.h 2007-12-23 11:15:40.258261372 +0100
+--- linux-2.6.18.8/include/asm-i386/page.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/page.h 2008-05-19 00:34:07.864057885 +0300
@@ -122,7 +122,7 @@
#define PAGE_OFFSET ((unsigned long)__PAGE_OFFSET)
#define __pa(x) ((unsigned long)(x)-PAGE_OFFSET)
#define __va(x) ((void *)((unsigned long)(x)+PAGE_OFFSET))
#define pfn_to_kaddr(pfn) __va((pfn) << PAGE_SHIFT)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/pgtable-2level-defs.h linux-2.6.18-xen.hg/include/asm-i386/pgtable-2level-defs.h
---- linux-2.6.18/include/asm-i386/pgtable-2level-defs.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/pgtable-2level-defs.h 2007-12-23 11:15:40.261594879 +0100
+--- linux-2.6.18.8/include/asm-i386/pgtable-2level-defs.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/pgtable-2level-defs.h 2008-05-19 00:34:07.964063649 +0300
@@ -1,6 +1,8 @@
#ifndef _I386_PGTABLE_2LEVEL_DEFS_H
#define _I386_PGTABLE_2LEVEL_DEFS_H
/*
* traditional i386 two-level paging structure:
*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-i386/pgtable-3level-defs.h linux-2.6.18-xen.hg/include/asm-i386/pgtable-3level-defs.h
---- linux-2.6.18/include/asm-i386/pgtable-3level-defs.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-i386/pgtable-3level-defs.h 2007-12-23 11:15:40.261594879 +0100
+--- linux-2.6.18.8/include/asm-i386/pgtable-3level-defs.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-i386/pgtable-3level-defs.h 2008-05-19 00:34:07.968063880 +0300
@@ -1,6 +1,8 @@
#ifndef _I386_PGTABLE_3LEVEL_DEFS_H
#define _I386_PGTABLE_3LEVEL_DEFS_H
/*
* PGDIR_SHIFT determines what a top-level page table entry can map
*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/agp.h linux-2.6.18-xen.hg/include/asm-ia64/agp.h
---- linux-2.6.18/include/asm-ia64/agp.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/agp.h 2007-12-23 11:15:40.298263470 +0100
+--- linux-2.6.18.8/include/asm-ia64/agp.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/agp.h 2008-05-19 00:34:08.200077254 +0300
@@ -19,13 +19,44 @@
#define flush_agp_cache() mb()
+#endif /* CONFIG_XEN */
#endif /* _ASM_IA64_AGP_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/gcc_intrin.h linux-2.6.18-xen.hg/include/asm-ia64/gcc_intrin.h
---- linux-2.6.18/include/asm-ia64/gcc_intrin.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/gcc_intrin.h 2007-12-23 11:15:40.311597505 +0100
+--- linux-2.6.18.8/include/asm-ia64/gcc_intrin.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/gcc_intrin.h 2008-05-19 00:34:08.652103309 +0300
@@ -26,7 +26,7 @@
register unsigned long ia64_r13 asm ("r13") __attribute_used__;
+#define __ia64_get_psr_i() (__ia64_getreg(_IA64_REG_PSR) & 0x4000UL)
+
#endif /* _ASM_IA64_GCC_INTRIN_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/gnttab_dma.h linux-2.6.18-xen.hg/include/asm-ia64/gnttab_dma.h
---- linux-2.6.18/include/asm-ia64/gnttab_dma.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/gnttab_dma.h 2007-12-23 11:15:40.311597505 +0100
+--- linux-2.6.18.8/include/asm-ia64/gnttab_dma.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/gnttab_dma.h 2008-05-19 00:34:08.652103309 +0300
@@ -0,0 +1,51 @@
+/*
+ * Copyright (c) 2007 Herbert Xu <herbert@gondor.apana.org.au>
+}
+
+#endif /* _ASM_IA64_GNTTAB_DMA_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/hw_irq.h linux-2.6.18-xen.hg/include/asm-ia64/hw_irq.h
---- linux-2.6.18/include/asm-ia64/hw_irq.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/hw_irq.h 2007-12-23 11:15:40.314931012 +0100
+--- linux-2.6.18.8/include/asm-ia64/hw_irq.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/hw_irq.h 2008-05-19 00:34:08.652103309 +0300
@@ -15,7 +15,11 @@
#include <asm/ptrace.h>
#include <asm/smp.h>
platform_send_ipi(smp_processor_id(), vector, IA64_IPI_DM_INT, 0);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/hypercall.h linux-2.6.18-xen.hg/include/asm-ia64/hypercall.h
---- linux-2.6.18/include/asm-ia64/hypercall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/hypercall.h 2007-12-23 11:15:40.314931012 +0100
+--- linux-2.6.18.8/include/asm-ia64/hypercall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/hypercall.h 2008-05-19 00:34:08.656103540 +0300
@@ -0,0 +1,458 @@
+/******************************************************************************
+ * hypercall.h
+#define HYPERVISOR_mmu_update(req, count, success_count, domid) ({BUG();0;})
+
+#endif /* __HYPERCALL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/hypervisor.h linux-2.6.18-xen.hg/include/asm-ia64/hypervisor.h
---- linux-2.6.18/include/asm-ia64/hypervisor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/hypervisor.h 2007-12-23 11:15:40.314931012 +0100
+--- linux-2.6.18.8/include/asm-ia64/hypervisor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/hypervisor.h 2008-05-19 00:34:08.656103540 +0300
@@ -0,0 +1,233 @@
+/******************************************************************************
+ * hypervisor.h
+}
+
+#ifndef CONFIG_VMX_GUEST
-+// for drivers/xen/privcmd/privcmd.c
++/* for drivers/xen/privcmd/privcmd.c */
+#define machine_to_phys_mapping 0
+struct vm_area_struct;
+int direct_remap_pfn_range(struct vm_area_struct *vma,
+int privcmd_mmap(struct file * file, struct vm_area_struct * vma);
+#define HAVE_ARCH_PRIVCMD_MMAP
+
-+// for drivers/xen/balloon/balloon.c
++/* for drivers/xen/balloon/balloon.c */
+#ifdef CONFIG_XEN_SCRUB_PAGES
+#define scrub_pages(_p,_n) memset((void *)(_p), 0, (_n) << PAGE_SHIFT)
+#else
+#define __pte_ma(_x) ((pte_t) {(_x)}) /* unmodified use */
+#define pfn_pte_ma(_x,_y) __pte_ma(0) /* unmodified use */
+
-+// for netfront.c, netback.c
-+#define MULTI_UVMFLAGS_INDEX 0 //XXX any value
++/* for netfront.c, netback.c */
++#define MULTI_UVMFLAGS_INDEX 0 /* XXX any value */
+
+static inline void
+MULTI_update_va_mapping(
+ (-ENOSYS); \
+ })
+
-+// for debug
++/* for debug */
+asmlinkage int xprintk(const char *fmt, ...);
+#define xprintd(fmt, ...) xprintk("%s:%d " fmt, __func__, __LINE__, \
+ ##__VA_ARGS__)
+#endif
+
+#endif /* __HYPERVISOR_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/intel_intrin.h linux-2.6.18-xen.hg/include/asm-ia64/intel_intrin.h
---- linux-2.6.18/include/asm-ia64/intel_intrin.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/intel_intrin.h 2007-12-23 11:15:40.314931012 +0100
+--- linux-2.6.18.8/include/asm-ia64/intel_intrin.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/intel_intrin.h 2008-05-19 00:34:08.660103770 +0300
@@ -16,8 +16,8 @@
* intrinsic
*/
#define __builtin_trap() __break(0);
#endif /* _ASM_IA64_INTEL_INTRIN_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/intrinsics.h linux-2.6.18-xen.hg/include/asm-ia64/intrinsics.h
---- linux-2.6.18/include/asm-ia64/intrinsics.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/intrinsics.h 2007-12-23 11:15:40.318264518 +0100
+--- linux-2.6.18.8/include/asm-ia64/intrinsics.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/intrinsics.h 2008-05-19 00:34:08.660103770 +0300
@@ -18,6 +18,15 @@
# include <asm/gcc_intrin.h>
#endif
#endif
+#include <asm/privop.h>
#endif /* _ASM_IA64_INTRINSICS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/io.h linux-2.6.18-xen.hg/include/asm-ia64/io.h
---- linux-2.6.18/include/asm-ia64/io.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/io.h 2007-12-23 11:15:40.318264518 +0100
+--- linux-2.6.18.8/include/asm-ia64/io.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/io.h 2008-05-19 00:34:08.664104001 +0300
@@ -96,9 +96,39 @@
* The following two macros are deprecated and scheduled for removal.
* Please use the PCI-DMA interface defined in <asm/pci.h> instead.
# endif /* KERNEL */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/iosapic.h linux-2.6.18-xen.hg/include/asm-ia64/iosapic.h
---- linux-2.6.18/include/asm-ia64/iosapic.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/iosapic.h 2007-12-23 11:15:40.318264518 +0100
+--- linux-2.6.18.8/include/asm-ia64/iosapic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/iosapic.h 2008-05-19 00:34:08.664104001 +0300
@@ -53,6 +53,7 @@
#define NR_IOSAPICS 256
static inline void iosapic_eoi(char __iomem *iosapic, u32 vector)
{
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/irq.h linux-2.6.18-xen.hg/include/asm-ia64/irq.h
---- linux-2.6.18/include/asm-ia64/irq.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/irq.h 2007-12-23 11:15:40.318264518 +0100
+--- linux-2.6.18.8/include/asm-ia64/irq.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/irq.h 2008-05-19 00:34:08.668104231 +0300
@@ -11,8 +11,41 @@
* 02/29/00 D.Mosberger moved most things into hw_irq.h
*/
static __inline__ int
irq_canonicalize (int irq)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/kexec.h linux-2.6.18-xen.hg/include/asm-ia64/kexec.h
---- linux-2.6.18/include/asm-ia64/kexec.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/kexec.h 2007-12-23 11:15:40.321598027 +0100
+--- linux-2.6.18.8/include/asm-ia64/kexec.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/kexec.h 2008-05-19 00:34:08.668104231 +0300
@@ -0,0 +1,58 @@
+#ifndef _ASM_IA64_KEXEC_H
+#define _ASM_IA64_KEXEC_H
+#endif
+
+#endif /* _ASM_IA64_KEXEC_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/machvec.h linux-2.6.18-xen.hg/include/asm-ia64/machvec.h
---- linux-2.6.18/include/asm-ia64/machvec.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/machvec.h 2007-12-23 11:15:40.324931536 +0100
+--- linux-2.6.18.8/include/asm-ia64/machvec.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/machvec.h 2008-05-19 00:34:08.676104693 +0300
@@ -35,6 +35,7 @@
typedef int ia64_mv_pci_legacy_write_t (struct pci_bus *, u16 port, u32 val,
u8 size);
#ifndef platform_dma_init
# define platform_dma_init swiotlb_init
#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/machvec_sn2.h linux-2.6.18-xen.hg/include/asm-ia64/machvec_sn2.h
---- linux-2.6.18/include/asm-ia64/machvec_sn2.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/machvec_sn2.h 2007-12-23 11:15:40.324931536 +0100
+--- linux-2.6.18.8/include/asm-ia64/machvec_sn2.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/machvec_sn2.h 2008-05-19 00:34:08.700106076 +0300
@@ -67,6 +67,7 @@
extern ia64_mv_dma_mapping_error sn_dma_mapping_error;
extern ia64_mv_dma_supported sn_dma_supported;
#ifdef CONFIG_PCI_MSI
#define platform_msi_init sn_msi_init
#else
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/machvec_xen.h linux-2.6.18-xen.hg/include/asm-ia64/machvec_xen.h
---- linux-2.6.18/include/asm-ia64/machvec_xen.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/machvec_xen.h 2007-12-23 11:15:40.328265047 +0100
+--- linux-2.6.18.8/include/asm-ia64/machvec_xen.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/machvec_xen.h 2008-05-19 00:34:08.704106307 +0300
@@ -0,0 +1,43 @@
+#ifndef _ASM_IA64_MACHVEC_XEN_h
+#define _ASM_IA64_MACHVEC_XEN_h
+#define platform_dma_mapping_error xen_dma_mapping_error
+
+#endif /* _ASM_IA64_MACHVEC_XEN_h */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/maddr.h linux-2.6.18-xen.hg/include/asm-ia64/maddr.h
---- linux-2.6.18/include/asm-ia64/maddr.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/maddr.h 2007-12-23 11:15:40.328265047 +0100
+--- linux-2.6.18.8/include/asm-ia64/maddr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/maddr.h 2008-05-19 00:34:08.704106307 +0300
@@ -0,0 +1,126 @@
+#ifndef _ASM_IA64_MADDR_H
+#define _ASM_IA64_MADDR_H
+ if (p2m_initialized)
+ return p2m_phystomach(pfn);
+ mfn = HYPERVISOR_phystomach(pfn);
-+ BUG_ON(mfn == 0); // XXX
-+ BUG_ON(mfn == INVALID_P2M_ENTRY); // XXX
++ BUG_ON(mfn == 0); /* XXX */
++ BUG_ON(mfn == INVALID_P2M_ENTRY); /* XXX */
+ BUG_ON(mfn == INVALID_MFN);
+ return mfn;
+}
+ unsigned long pfn;
+ pfn = HYPERVISOR_machtophys(mfn);
+ BUG_ON(pfn == 0);
-+ //BUG_ON(pfn == INVALID_M2P_ENTRY);
++ /* BUG_ON(pfn == INVALID_M2P_ENTRY); */
+ return pfn;
+}
+
+
+#define mfn_to_virt(mfn) (__va((mfn) << PAGE_SHIFT))
+#define virt_to_mfn(virt) (__pa(virt) >> PAGE_SHIFT)
-+#define virt_to_machine(virt) __pa(virt) // for tpmfront.c
++#define virt_to_machine(virt) __pa(virt) /* for tpmfront.c */
+
+#define set_phys_to_machine(pfn, mfn) do { } while (0)
+
-+typedef unsigned long maddr_t; // to compile netback, netfront
++typedef unsigned long maddr_t; /* to compile netback, netfront */
+#ifndef _ASM_IA64_SN_TYPES_H /* paddr_t is defined in asm-ia64/sn/types.h */
+typedef unsigned long paddr_t;
+#endif
+#endif
+
+#endif /* _ASM_IA64_MADDR_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/meminit.h linux-2.6.18-xen.hg/include/asm-ia64/meminit.h
---- linux-2.6.18/include/asm-ia64/meminit.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/meminit.h 2007-12-23 11:15:40.328265047 +0100
+--- linux-2.6.18.8/include/asm-ia64/meminit.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/meminit.h 2008-05-19 00:34:08.708106537 +0300
@@ -15,11 +15,17 @@
* - initrd (optional)
* - command line string
struct rsvd_region {
unsigned long start; /* virtual address of beginning of element */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/mmu_context.h linux-2.6.18-xen.hg/include/asm-ia64/mmu_context.h
---- linux-2.6.18/include/asm-ia64/mmu_context.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/mmu_context.h 2007-12-23 11:15:40.331598554 +0100
+--- linux-2.6.18.8/include/asm-ia64/mmu_context.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/mmu_context.h 2008-05-19 00:34:08.712106768 +0300
@@ -151,11 +151,7 @@
# endif
#endif
ia64_srlz_i(); /* srlz.i implies srlz.d */
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/page.h linux-2.6.18-xen.hg/include/asm-ia64/page.h
---- linux-2.6.18/include/asm-ia64/page.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/page.h 2007-12-23 11:15:40.334932060 +0100
+--- linux-2.6.18.8/include/asm-ia64/page.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/page.h 2008-05-19 00:34:08.720107229 +0300
@@ -119,6 +119,7 @@
#endif
+
# endif /* __KERNEL__ */
#endif /* _ASM_IA64_PAGE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/pgalloc.h linux-2.6.18-xen.hg/include/asm-ia64/pgalloc.h
---- linux-2.6.18/include/asm-ia64/pgalloc.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/pgalloc.h 2007-12-23 11:15:40.338265569 +0100
+--- linux-2.6.18.8/include/asm-ia64/pgalloc.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/pgalloc.h 2008-05-19 00:34:08.740108382 +0300
@@ -125,7 +125,11 @@
static inline void
pmd_populate(struct mm_struct *mm, pmd_t * pmd_entry, struct page *pte)
}
static inline void
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/privop.h linux-2.6.18-xen.hg/include/asm-ia64/privop.h
---- linux-2.6.18/include/asm-ia64/privop.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/privop.h 2007-12-23 11:15:40.341599078 +0100
+--- linux-2.6.18.8/include/asm-ia64/privop.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/privop.h 2008-05-19 00:34:08.748108843 +0300
@@ -0,0 +1,63 @@
+#ifndef _ASM_IA64_PRIVOP_H
+#define _ASM_IA64_PRIVOP_H
+#endif /* !__ASSEMBLY */
+
+#endif /* _ASM_IA64_PRIVOP_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/sal.h linux-2.6.18-xen.hg/include/asm-ia64/sal.h
---- linux-2.6.18/include/asm-ia64/sal.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/sal.h 2007-12-23 11:15:40.344932588 +0100
+--- linux-2.6.18.8/include/asm-ia64/sal.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/sal.h 2008-05-19 00:34:08.760109535 +0300
@@ -42,6 +42,9 @@
#include <asm/pal.h>
#include <asm/system.h>
SAL_CALL_REENTRANT(isrv, SAL_GET_STATE_INFO, sal_info_type, 0,
sal_info, 0, 0, 0, 0);
if (isrv.status)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/sn/sn_sal.h linux-2.6.18-xen.hg/include/asm-ia64/sn/sn_sal.h
---- linux-2.6.18/include/asm-ia64/sn/sn_sal.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/sn/sn_sal.h 2007-12-23 11:15:40.364933638 +0100
+--- linux-2.6.18.8/include/asm-ia64/sn/sn_sal.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/sn/sn_sal.h 2008-05-19 00:34:09.080127981 +0300
@@ -87,6 +87,8 @@
#define SN_SAL_INJECT_ERROR 0x02000067
#define SN_SAL_SET_CPU_NUMBER 0x02000068
+ return rv.status;
+}
#endif /* _ASM_IA64_SN_SN_SAL_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/sn/types.h linux-2.6.18-xen.hg/include/asm-ia64/sn/types.h
---- linux-2.6.18/include/asm-ia64/sn/types.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/sn/types.h 2007-12-23 11:15:40.371600653 +0100
+--- linux-2.6.18.8/include/asm-ia64/sn/types.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/sn/types.h 2008-05-19 00:34:09.096128903 +0300
@@ -20,7 +20,9 @@
typedef unsigned char slabid_t; /* slab (asic) within slot */
typedef u64 nic_t;
typedef short cnodeid_t;
#endif /* _ASM_IA64_SN_TYPES_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/swiotlb.h linux-2.6.18-xen.hg/include/asm-ia64/swiotlb.h
---- linux-2.6.18/include/asm-ia64/swiotlb.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/swiotlb.h 2007-12-23 11:15:40.374934162 +0100
+--- linux-2.6.18.8/include/asm-ia64/swiotlb.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/swiotlb.h 2008-05-19 00:34:09.188134207 +0300
@@ -0,0 +1,41 @@
+#ifndef _ASM_SWIOTLB_H
+#define _ASM_SWIOTLB_H 1
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/synch_bitops.h linux-2.6.18-xen.hg/include/asm-ia64/synch_bitops.h
---- linux-2.6.18/include/asm-ia64/synch_bitops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/synch_bitops.h 2007-12-23 11:15:40.374934162 +0100
+--- linux-2.6.18.8/include/asm-ia64/synch_bitops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/synch_bitops.h 2008-05-19 00:34:09.192134437 +0300
@@ -0,0 +1,61 @@
+#ifndef __XEN_SYNCH_BITOPS_H__
+#define __XEN_SYNCH_BITOPS_H__
+#define synch_cmpxchg_subword synch_cmpxchg
+
+#endif /* __XEN_SYNCH_BITOPS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/system.h linux-2.6.18-xen.hg/include/asm-ia64/system.h
---- linux-2.6.18/include/asm-ia64/system.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-ia64/system.h 2007-12-23 11:15:40.378267672 +0100
+--- linux-2.6.18.8/include/asm-ia64/system.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/system.h 2008-05-19 00:34:09.192134437 +0300
@@ -123,7 +123,7 @@
#define __local_irq_save(x) \
do { \
#define irqs_disabled() \
({ \
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/xen/privop.h linux-2.6.18-xen.hg/include/asm-ia64/xen/privop.h
---- linux-2.6.18/include/asm-ia64/xen/privop.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/xen/privop.h 2007-12-23 11:15:40.384934686 +0100
+--- linux-2.6.18.8/include/asm-ia64/xen/privop.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/xen/privop.h 2008-05-19 00:34:09.236136974 +0300
@@ -0,0 +1,551 @@
+#ifndef _ASM_IA64_XEN_PRIVOP_H
+#define _ASM_IA64_XEN_PRIVOP_H
+#define ia64_pal_call_static xen_pal_call_static
+
+#endif /* _ASM_IA64_XEN_PRIVOP_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/xen/xcom_hcall.h linux-2.6.18-xen.hg/include/asm-ia64/xen/xcom_hcall.h
---- linux-2.6.18/include/asm-ia64/xen/xcom_hcall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/xen/xcom_hcall.h 2007-12-23 11:15:40.388268195 +0100
+--- linux-2.6.18.8/include/asm-ia64/xen/xcom_hcall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/xen/xcom_hcall.h 2008-05-19 00:34:09.236136974 +0300
@@ -0,0 +1,67 @@
+/*
+ * Copyright (C) 2006 Tristan Gingold <tristan.gingold@bull.net>, Bull SAS
+extern int xencomm_hypercall_kexec_op(int cmd, void *arg);
+
+#endif /* _LINUX_XENCOMM_HCALL_H_ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/xen/xencomm.h linux-2.6.18-xen.hg/include/asm-ia64/xen/xencomm.h
---- linux-2.6.18/include/asm-ia64/xen/xencomm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/xen/xencomm.h 2007-12-23 11:15:40.388268195 +0100
+--- linux-2.6.18.8/include/asm-ia64/xen/xencomm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/xen/xencomm.h 2008-05-19 00:34:09.240137204 +0300
@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2006 Hollis Blanchard <hollisb@us.ibm.com>, IBM Corporation
+#include <xen/xencomm.h>
+
+#endif /* _ASM_IA64_XENCOMM_H_ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-ia64/xenoprof.h linux-2.6.18-xen.hg/include/asm-ia64/xenoprof.h
---- linux-2.6.18/include/asm-ia64/xenoprof.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-ia64/xenoprof.h 2007-12-23 11:15:40.388268195 +0100
+--- linux-2.6.18.8/include/asm-ia64/xenoprof.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-ia64/xenoprof.h 2008-05-19 00:34:09.240137204 +0300
@@ -0,0 +1,48 @@
+/******************************************************************************
+ * asm-ia64/xenoprof.h
+};
+
+struct xenoprof_shared_buffer;
-+void xenoprof_arch_unmap_shared_buffer(struct xenoprof_shared_buffer* sbuf);
++void xenoprof_arch_unmap_shared_buffer(struct xenoprof_shared_buffer *sbuf);
+struct xenoprof_get_buffer;
-+int xenoprof_arch_map_shared_buffer(struct xenoprof_get_buffer* get_buffer,
-+ struct xenoprof_shared_buffer* sbuf);
++int xenoprof_arch_map_shared_buffer(struct xenoprof_get_buffer *get_buffer,
++ struct xenoprof_shared_buffer *sbuf);
+struct xenoprof_passive;
-+int xenoprof_arch_set_passive(struct xenoprof_passive* pdomain,
-+ struct xenoprof_shared_buffer* sbuf);
++int xenoprof_arch_set_passive(struct xenoprof_passive *pdomain,
++ struct xenoprof_shared_buffer *sbuf);
+
+#endif /* CONFIG_XEN */
+#endif /* __ASM_XENOPROF_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/mpic.h linux-2.6.18-xen.hg/include/asm-powerpc/mpic.h
---- linux-2.6.18/include/asm-powerpc/mpic.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/mpic.h 2007-12-23 11:15:41.931682576 +0100
+--- linux-2.6.18.8/include/asm-powerpc/mpic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/mpic.h 2008-05-19 00:34:13.772398451 +0300
@@ -305,6 +305,8 @@
#define MPIC_SPV_EOI 0x00000020
/* No passthrough disable */
/* MPIC HW modification ID */
#define MPIC_REGSET_MASK 0xf0000000
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/udbg.h linux-2.6.18-xen.hg/include/asm-powerpc/udbg.h
---- linux-2.6.18/include/asm-powerpc/udbg.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/udbg.h 2007-12-23 11:15:41.998352740 +0100
+--- linux-2.6.18.8/include/asm-powerpc/udbg.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/udbg.h 2008-05-19 00:34:13.988410903 +0300
@@ -42,6 +42,7 @@
extern void __init udbg_init_pmac_realmode(void);
extern void __init udbg_init_maple_realmode(void);
extern void __init udbg_init_rtas_panel(void);
extern void __init udbg_init_rtas_console(void);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/xen/asm/gnttab_dma.h linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/gnttab_dma.h
---- linux-2.6.18/include/asm-powerpc/xen/asm/gnttab_dma.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/gnttab_dma.h 2007-12-23 11:15:42.005019763 +0100
+--- linux-2.6.18.8/include/asm-powerpc/xen/asm/gnttab_dma.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/gnttab_dma.h 2008-05-19 00:34:14.016412517 +0300
@@ -0,0 +1,29 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+}
+
+#endif /* _ASM_PPC_GNTTAB_DMA_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/xen/asm/hypercall.h linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypercall.h
---- linux-2.6.18/include/asm-powerpc/xen/asm/hypercall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypercall.h 2007-12-23 11:15:42.005019763 +0100
+--- linux-2.6.18.8/include/asm-powerpc/xen/asm/hypercall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypercall.h 2008-05-19 00:34:14.020412747 +0300
@@ -0,0 +1,90 @@
+/******************************************************************************
+ * hypercall.h
+extern int privcmd_hypercall(struct privcmd_hypercall *hypercall);
+
+#endif /* __HYPERCALL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/xen/asm/hypervisor.h linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypervisor.h
---- linux-2.6.18/include/asm-powerpc/xen/asm/hypervisor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypervisor.h 2007-12-23 11:15:42.005019763 +0100
+--- linux-2.6.18.8/include/asm-powerpc/xen/asm/hypervisor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/hypervisor.h 2008-05-19 00:34:14.020412747 +0300
@@ -0,0 +1,276 @@
+/******************************************************************************
+ * hypervisor.h
+ })
+
+#endif /* __HYPERVISOR_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/xen/asm/maddr.h linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/maddr.h
---- linux-2.6.18/include/asm-powerpc/xen/asm/maddr.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/maddr.h 2007-12-23 11:15:42.005019763 +0100
+--- linux-2.6.18.8/include/asm-powerpc/xen/asm/maddr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/maddr.h 2008-05-19 00:34:14.020412747 +0300
@@ -0,0 +1,7 @@
+#ifndef _POWERPC_MADDR_H
+#define _POWERPC_MADDR_H
+#include <xen/interface/xen.h>
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-powerpc/xen/asm/synch_bitops.h linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/synch_bitops.h
---- linux-2.6.18/include/asm-powerpc/xen/asm/synch_bitops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/synch_bitops.h 2007-12-23 11:15:42.005019763 +0100
+--- linux-2.6.18.8/include/asm-powerpc/xen/asm/synch_bitops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-powerpc/xen/asm/synch_bitops.h 2008-05-19 00:34:14.020412747 +0300
@@ -0,0 +1,100 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+#endif
+
+#endif /* __SYNCH_BITOPS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/acpi.h linux-2.6.18-xen.hg/include/asm-x86_64/acpi.h
---- linux-2.6.18/include/asm-x86_64/acpi.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/acpi.h 2007-12-23 11:15:43.691775003 +0100
-@@ -153,6 +153,10 @@
+--- linux-2.6.18.8/include/asm-x86_64/acpi.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/acpi.h 2008-05-19 00:34:21.360835862 +0300
+@@ -28,6 +28,9 @@
+
+ #ifdef __KERNEL__
+
++#ifdef CONFIG_XEN
++#include <xen/interface/platform.h>
++#endif
+ #include <acpi/pdc_intel.h>
+
+ #define COMPILER_DEPENDENT_INT64 long long
+@@ -129,6 +132,27 @@
+ }
+ extern int acpi_irq_balance_set(char *str);
+
++#ifdef CONFIG_XEN
++static inline int acpi_notify_hypervisor_state(u8 sleep_state,
++ u32 pm1a_cnt_val,
++ u32 pm1b_cnt_val)
++{
++ struct xen_platform_op op = {
++ .cmd = XENPF_enter_acpi_sleep,
++ .interface_version = XENPF_INTERFACE_VERSION,
++ .u = {
++ .enter_acpi_sleep = {
++ .pm1a_cnt_val = pm1a_cnt_val,
++ .pm1b_cnt_val = pm1b_cnt_val,
++ .sleep_state = sleep_state,
++ },
++ },
++ };
++
++ return HYPERVISOR_platform_op(&op);
++}
++#endif /* CONFIG_XEN */
++
+ #else /* !CONFIG_ACPI */
+
+ #define acpi_lapic 0
+@@ -152,7 +176,6 @@
+
/* early initialization routine */
extern void acpi_reserve_bootmem(void);
-
-+#ifdef CONFIG_ACPI_PV_SLEEP
-+extern int acpi_notify_hypervisor_state(u8 sleep_state,
-+ u32 pm1a_cnt, u32 pm1b_cnt);
-+#endif /* CONFIG_ACPI_PV_SLEEP */
+-
#endif /*CONFIG_ACPI_SLEEP*/
#define boot_cpu_physical_apicid boot_cpu_id
-@@ -162,7 +166,9 @@
+@@ -162,7 +185,9 @@
extern u8 x86_acpiid_to_apicid[];
extern int acpi_skip_timer_override;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/agp.h linux-2.6.18-xen.hg/include/asm-x86_64/agp.h
---- linux-2.6.18/include/asm-x86_64/agp.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/agp.h 2007-12-23 11:15:43.691775003 +0100
+--- linux-2.6.18.8/include/asm-x86_64/agp.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/agp.h 2008-05-19 00:34:21.360835862 +0300
@@ -10,8 +10,10 @@
* with different cachability attributes for the same page.
*/
#define flush_agp_mappings() global_flush_tlb()
/* Could use CLFLUSH here if the cpu supports it. But then it would
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/apic.h linux-2.6.18-xen.hg/include/asm-x86_64/apic.h
---- linux-2.6.18/include/asm-x86_64/apic.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/apic.h 2007-12-23 11:15:43.691775003 +0100
+--- linux-2.6.18.8/include/asm-x86_64/apic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/apic.h 2008-05-19 00:34:21.364836093 +0300
@@ -98,11 +98,13 @@
extern int disable_timer_pin_1;
#endif /* CONFIG_X86_LOCAL_APIC */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/kexec.h linux-2.6.18-xen.hg/include/asm-x86_64/kexec.h
---- linux-2.6.18/include/asm-x86_64/kexec.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/kexec.h 2007-12-23 11:15:43.721776570 +0100
+--- linux-2.6.18.8/include/asm-x86_64/io_apic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/io_apic.h 2008-05-19 00:34:21.700855462 +0300
+@@ -12,7 +12,7 @@
+
+ #ifdef CONFIG_X86_IO_APIC
+
+-#ifdef CONFIG_PCI_MSI
++#if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
+ static inline int use_pci_vector(void) {return 1;}
+ static inline void disable_edge_ioapic_vector(unsigned int vector) { }
+ static inline void mask_and_ack_level_ioapic_vector(unsigned int vector) { }
+--- linux-2.6.18.8/include/asm-x86_64/kexec.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/kexec.h 2008-05-19 00:34:21.708855923 +0300
@@ -1,6 +1,27 @@
#ifndef _X86_64_KEXEC_H
#define _X86_64_KEXEC_H
+#endif /* __ASSEMBLY__ */
+
#endif /* _X86_64_KEXEC_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/agp.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/agp.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/agp.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/agp.h 2007-12-23 11:15:43.725110079 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/agp.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/agp.h 2008-05-19 00:34:21.728857076 +0300
@@ -0,0 +1,40 @@
+#ifndef AGP_H
+#define AGP_H 1
+ dma_free_coherent(NULL,PAGE_SIZE<<(order),(table),virt_to_bus(table))
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/arch_hooks.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/arch_hooks.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/arch_hooks.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/arch_hooks.h 2007-12-23 11:15:43.725110079 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/arch_hooks.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/arch_hooks.h 2008-05-19 00:34:21.728857076 +0300
@@ -0,0 +1,27 @@
+#ifndef _ASM_ARCH_HOOKS_H
+#define _ASM_ARCH_HOOKS_H
+extern void mca_nmi_hook(void);
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/bootsetup.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/bootsetup.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/bootsetup.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/bootsetup.h 2007-12-23 11:15:43.725110079 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/bootsetup.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/bootsetup.h 2008-05-19 00:34:21.732857306 +0300
@@ -0,0 +1,42 @@
+
+#ifndef _X86_64_BOOTSETUP_H
+#define RAMDISK_LOAD_FLAG 0x4000
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/desc.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/desc.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/desc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/desc.h 2007-12-23 11:15:43.725110079 +0100
-@@ -0,0 +1,263 @@
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/desc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/desc.h 2008-05-19 00:34:21.732857306 +0300
+@@ -0,0 +1,265 @@
+/* Written 2000 by Andi Kleen */
+#ifndef __ARCH_DESC_H
+#define __ARCH_DESC_H
+ * it slows down context switching. Noone uses it anyway.
+ */
+ cpu = cpu; /* XXX avoid compiler warning */
-+ xen_set_ldt(0UL, 0);
++ xen_set_ldt(NULL, 0);
+ put_cpu();
+}
+
+ gdt[2] = t->tls_array[2];
+#endif
+#define C(i) \
-+ HYPERVISOR_update_descriptor(virt_to_machine(&cpu_gdt(cpu)[GDT_ENTRY_TLS_MIN + i]), t->tls_array[i])
++ if (HYPERVISOR_update_descriptor(virt_to_machine(&cpu_gdt(cpu)[GDT_ENTRY_TLS_MIN + i]), \
++ t->tls_array[i])) \
++ BUG();
+
+ C(0); C(1); C(2);
+#undef C
+ if (likely(!count))
+ segments = NULL;
+
-+ xen_set_ldt((unsigned long)segments, count);
++ xen_set_ldt(segments, count);
+}
+
+static inline void load_LDT(mm_context_t *pc)
+#endif /* !__ASSEMBLY__ */
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/dma-mapping.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/dma-mapping.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/dma-mapping.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/dma-mapping.h 2007-12-23 11:15:43.725110079 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/dma-mapping.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/dma-mapping.h 2008-05-19 00:34:21.732857306 +0300
@@ -0,0 +1,207 @@
+#ifndef _X8664_DMA_MAPPING_H
+#define _X8664_DMA_MAPPING_H 1
+#endif /* _X8664_DMA_MAPPING_H */
+
+#include <asm-i386/mach-xen/asm/dma-mapping.h>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/e820.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/e820.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/e820.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/e820.h 2007-12-23 11:15:43.725110079 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/e820.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/e820.h 2008-05-19 00:34:21.732857306 +0300
@@ -0,0 +1,66 @@
+/*
+ * structures and definitions for the int 15, ax=e820 memory map
+#endif/*!__ASSEMBLY__*/
+
+#endif/*__E820_HEADER*/
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/fixmap.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/fixmap.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/fixmap.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/fixmap.h 2007-12-23 11:15:43.728443593 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/fixmap.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/fixmap.h 2008-05-19 00:34:21.736857537 +0300
@@ -0,0 +1,112 @@
+/*
+ * fixmap.h: compile-time virtual memory allocation
+}
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/gnttab_dma.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/gnttab_dma.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/gnttab_dma.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/gnttab_dma.h 2007-12-23 11:15:43.728443593 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/gnttab_dma.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/gnttab_dma.h 2008-05-19 00:34:21.736857537 +0300
@@ -0,0 +1 @@
+#include <asm-i386/mach-xen/asm/gnttab_dma.h>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/hw_irq.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hw_irq.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/hw_irq.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hw_irq.h 2007-12-23 11:15:43.728443593 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/hw_irq.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hw_irq.h 2008-05-19 00:34:21.736857537 +0300
@@ -0,0 +1,136 @@
+#ifndef _ASM_HW_IRQ_H
+#define _ASM_HW_IRQ_H
+#endif
+
+#endif /* _ASM_HW_IRQ_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/hypercall.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypercall.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/hypercall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypercall.h 2007-12-23 11:15:43.728443593 +0100
-@@ -0,0 +1,407 @@
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/hypercall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypercall.h 2008-05-19 00:34:21.736857537 +0300
+@@ -0,0 +1,415 @@
+/******************************************************************************
+ * hypercall.h
+ *
+# error "please don't include this file directly"
+#endif
+
-+#define __STR(x) #x
-+#define STR(x) __STR(x)
-+
+#ifdef CONFIG_XEN
+#define HYPERCALL_STR(name) \
-+ "call hypercall_page + ("STR(__HYPERVISOR_##name)" * 32)"
++ "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
+#else
+#define HYPERCALL_STR(name) \
-+ "mov "__stringify(hypercall_stubs)",%%rax; " \
-+ "add $("STR(__HYPERVISOR_##name)" * 32),%%rax; " \
++ "mov $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
++ "add hypercall_stubs(%%rip),%%rax; " \
+ "call *%%rax"
+#endif
+
+#define _hypercall0(type, name) \
+({ \
-+ long __res; \
++ type __res; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res) \
+ : \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall1(type, name, a1) \
+({ \
-+ long __res, __ign1; \
++ type __res; \
++ long __ign1; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=D" (__ign1) \
+ : "1" ((long)(a1)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall2(type, name, a1, a2) \
+({ \
-+ long __res, __ign1, __ign2; \
++ type __res; \
++ long __ign1, __ign2; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=D" (__ign1), "=S" (__ign2) \
+ : "1" ((long)(a1)), "2" ((long)(a2)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall3(type, name, a1, a2, a3) \
+({ \
-+ long __res, __ign1, __ign2, __ign3; \
++ type __res; \
++ long __ign1, __ign2, __ign3; \
+ asm volatile ( \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
+ : "1" ((long)(a1)), "2" ((long)(a2)), \
+ "3" ((long)(a3)) \
+ : "memory" ); \
-+ (type)__res; \
++ __res; \
+})
+
+#define _hypercall4(type, name, a1, a2, a3, a4) \
+({ \
-+ long __res, __ign1, __ign2, __ign3; \
++ type __res; \
++ long __ign1, __ign2, __ign3; \
++ register long __arg4 asm("r10") = (long)(a4); \
+ asm volatile ( \
-+ "movq %7,%%r10; " \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
-+ "=d" (__ign3) \
++ "=d" (__ign3), "+r" (__arg4) \
+ : "1" ((long)(a1)), "2" ((long)(a2)), \
-+ "3" ((long)(a3)), "g" ((long)(a4)) \
-+ : "memory", "r10" ); \
-+ (type)__res; \
++ "3" ((long)(a3)) \
++ : "memory" ); \
++ __res; \
+})
+
+#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
+({ \
-+ long __res, __ign1, __ign2, __ign3; \
++ type __res; \
++ long __ign1, __ign2, __ign3; \
++ register long __arg4 asm("r10") = (long)(a4); \
++ register long __arg5 asm("r8") = (long)(a5); \
+ asm volatile ( \
-+ "movq %7,%%r10; movq %8,%%r8; " \
+ HYPERCALL_STR(name) \
+ : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
-+ "=d" (__ign3) \
++ "=d" (__ign3), "+r" (__arg4), "+r" (__arg5) \
+ : "1" ((long)(a1)), "2" ((long)(a2)), \
-+ "3" ((long)(a3)), "g" ((long)(a4)), \
-+ "g" ((long)(a5)) \
-+ : "memory", "r10", "r8" ); \
-+ (type)__res; \
++ "3" ((long)(a3)) \
++ : "memory" ); \
++ __res; \
+})
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_trap_table(
-+ trap_info_t *table)
++ const trap_info_t *table)
+{
+ return _hypercall1(int, set_trap_table, table);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_mmu_update(
-+ mmu_update_t *req, int count, int *success_count, domid_t domid)
++ mmu_update_t *req, unsigned int count, unsigned int *success_count,
++ domid_t domid)
+{
+ return _hypercall4(int, mmu_update, req, count, success_count, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_mmuext_op(
-+ struct mmuext_op *op, int count, int *success_count, domid_t domid)
++ struct mmuext_op *op, unsigned int count, unsigned int *success_count,
++ domid_t domid)
+{
+ return _hypercall4(int, mmuext_op, op, count, success_count, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_gdt(
-+ unsigned long *frame_list, int entries)
++ unsigned long *frame_list, unsigned int entries)
+{
+ return _hypercall2(int, set_gdt, frame_list, entries);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_stack_switch(
+ unsigned long ss, unsigned long esp)
+{
+ return _hypercall2(int, stack_switch, ss, esp);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_callbacks(
+ unsigned long event_address, unsigned long failsafe_address,
+ unsigned long syscall_address)
+ return _hypercall1(int, fpu_taskswitch, set);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_sched_op_compat(
+ int cmd, unsigned long arg)
+{
+ return _hypercall2(int, sched_op_compat, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_sched_op(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, sched_op, cmd, arg);
+}
+
-+static inline long
++static inline long __must_check
+HYPERVISOR_set_timer_op(
+ u64 timeout)
+{
+ return _hypercall1(long, set_timer_op, timeout);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_platform_op(
+ struct xen_platform_op *platform_op)
+{
+ return _hypercall1(int, platform_op, platform_op);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_debugreg(
-+ int reg, unsigned long value)
++ unsigned int reg, unsigned long value)
+{
+ return _hypercall2(int, set_debugreg, reg, value);
+}
+
-+static inline unsigned long
++static inline unsigned long __must_check
+HYPERVISOR_get_debugreg(
-+ int reg)
++ unsigned int reg)
+{
+ return _hypercall1(unsigned long, get_debugreg, reg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_descriptor(
+ unsigned long ma, unsigned long word)
+{
+ return _hypercall2(int, update_descriptor, ma, word);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_memory_op(
+ unsigned int cmd, void *arg)
+{
+ return _hypercall2(int, memory_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_multicall(
-+ multicall_entry_t *call_list, int nr_calls)
++ multicall_entry_t *call_list, unsigned int nr_calls)
+{
+ return _hypercall2(int, multicall, call_list, nr_calls);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_va_mapping(
+ unsigned long va, pte_t new_val, unsigned long flags)
+{
+ return _hypercall3(int, update_va_mapping, va, new_val.pte, flags);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_event_channel_op(
+ int cmd, void *arg)
+{
+ return rc;
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_acm_op(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, acm_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_xen_version(
+ int cmd, void *arg)
+{
+ return _hypercall2(int, xen_version, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_console_io(
-+ int cmd, int count, char *str)
++ int cmd, unsigned int count, char *str)
+{
+ return _hypercall3(int, console_io, cmd, count, str);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_physdev_op(
+ int cmd, void *arg)
+{
+ return rc;
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_grant_table_op(
+ unsigned int cmd, void *uop, unsigned int count)
+{
+ return _hypercall3(int, grant_table_op, cmd, uop, count);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_update_va_mapping_otherdomain(
+ unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
+{
+ new_val.pte, flags, domid);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_vm_assist(
+ unsigned int cmd, unsigned int type)
+{
+ return _hypercall2(int, vm_assist, cmd, type);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_vcpu_op(
-+ int cmd, int vcpuid, void *extra_args)
++ int cmd, unsigned int vcpuid, void *extra_args)
+{
+ return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_set_segment_base(
+ int reg, unsigned long value)
+{
+ return _hypercall2(int, set_segment_base, reg, value);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_suspend(
+ unsigned long srec)
+{
+ return rc;
+}
+
++#if CONFIG_XEN_COMPAT <= 0x030002
+static inline int
+HYPERVISOR_nmi_op(
+ unsigned long op, void *arg)
+{
+ return _hypercall2(int, nmi_op, op, arg);
+}
++#endif
+
-+static inline unsigned long
++#ifndef CONFIG_XEN
++static inline unsigned long __must_check
+HYPERVISOR_hvm_op(
+ int op, void *arg)
+{
+ return _hypercall2(unsigned long, hvm_op, op, arg);
+}
++#endif
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_callback_op(
-+ int cmd, void *arg)
++ int cmd, const void *arg)
+{
+ return _hypercall2(int, callback_op, cmd, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_xenoprof_op(
+ int op, void *arg)
+{
+ return _hypercall2(int, xenoprof_op, op, arg);
+}
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_kexec_op(
+ unsigned long op, void *args)
+{
+}
+
+#endif /* __HYPERCALL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/hypervisor.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypervisor.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/hypervisor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypervisor.h 2007-12-23 11:15:43.728443593 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/hypervisor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/hypervisor.h 2008-05-19 00:34:21.740857768 +0300
@@ -0,0 +1,2 @@
+
+#include <asm-i386/mach-xen/asm/hypervisor.h>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/io.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/io.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/io.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/io.h 2007-12-23 11:15:43.728443593 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/io.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/io.h 2008-05-19 00:34:21.740857768 +0300
@@ -0,0 +1,329 @@
+#ifndef _ASM_IO_H
+#define _ASM_IO_H
+#define ARCH_HAS_DEV_MEM
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/irqflags.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irqflags.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/irqflags.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irqflags.h 2007-12-23 11:15:43.731777107 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/irq.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irq.h 2008-05-19 00:34:21.740857768 +0300
+@@ -0,0 +1,38 @@
++#ifndef _ASM_IRQ_H
++#define _ASM_IRQ_H
++
++/*
++ * linux/include/asm/irq.h
++ *
++ * (C) 1992, 1993 Linus Torvalds, (C) 1997 Ingo Molnar
++ *
++ * IRQ/IPI changes taken from work by Thomas Radke
++ * <tomsoft@informatik.tu-chemnitz.de>
++ */
++
++#include <linux/sched.h>
++/* include comes from machine specific directory */
++#include "irq_vectors.h"
++#include <asm/thread_info.h>
++
++static __inline__ int irq_canonicalize(int irq)
++{
++ return ((irq == 2) ? 9 : irq);
++}
++
++#ifdef CONFIG_X86_LOCAL_APIC
++#define ARCH_HAS_NMI_WATCHDOG /* See include/linux/nmi.h */
++#endif
++
++#define KDB_VECTOR 0xf9
++
++# define irq_ctx_init(cpu) do { } while (0)
++
++#ifdef CONFIG_HOTPLUG_CPU
++#include <linux/cpumask.h>
++extern void fixup_irqs(cpumask_t map);
++#endif
++
++#define __ARCH_HAS_DO_SOFTIRQ 1
++
++#endif /* _ASM_IRQ_H */
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/irqflags.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irqflags.h 2008-05-19 00:34:21.740857768 +0300
@@ -0,0 +1,139 @@
+/*
+ * include/asm-x86_64/irqflags.h
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/irq.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irq.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/irq.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/irq.h 2007-12-23 11:15:43.728443593 +0100
-@@ -0,0 +1,38 @@
-+#ifndef _ASM_IRQ_H
-+#define _ASM_IRQ_H
-+
-+/*
-+ * linux/include/asm/irq.h
-+ *
-+ * (C) 1992, 1993 Linus Torvalds, (C) 1997 Ingo Molnar
-+ *
-+ * IRQ/IPI changes taken from work by Thomas Radke
-+ * <tomsoft@informatik.tu-chemnitz.de>
-+ */
-+
-+#include <linux/sched.h>
-+/* include comes from machine specific directory */
-+#include "irq_vectors.h"
-+#include <asm/thread_info.h>
-+
-+static __inline__ int irq_canonicalize(int irq)
-+{
-+ return ((irq == 2) ? 9 : irq);
-+}
-+
-+#ifdef CONFIG_X86_LOCAL_APIC
-+#define ARCH_HAS_NMI_WATCHDOG /* See include/linux/nmi.h */
-+#endif
-+
-+#define KDB_VECTOR 0xf9
-+
-+# define irq_ctx_init(cpu) do { } while (0)
-+
-+#ifdef CONFIG_HOTPLUG_CPU
-+#include <linux/cpumask.h>
-+extern void fixup_irqs(cpumask_t map);
-+#endif
-+
-+#define __ARCH_HAS_DO_SOFTIRQ 1
-+
-+#endif /* _ASM_IRQ_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/maddr.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/maddr.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/maddr.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/maddr.h 2007-12-23 11:15:43.731777107 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/maddr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/maddr.h 2008-05-19 00:34:21.744857998 +0300
@@ -0,0 +1,161 @@
+#ifndef _X86_64_MADDR_H
+#define _X86_64_MADDR_H
+
+#endif /* _X86_64_MADDR_H */
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/mmu_context.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu_context.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/mmu_context.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu_context.h 2007-12-23 11:15:43.731777107 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/mmu.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu.h 2008-05-19 00:34:21.744857998 +0300
+@@ -0,0 +1,38 @@
++#ifndef __x86_64_MMU_H
++#define __x86_64_MMU_H
++
++#include <linux/spinlock.h>
++#include <asm/semaphore.h>
++
++/*
++ * The x86_64 doesn't have a mmu context, but
++ * we put the segment information here.
++ *
++ * cpu_vm_mask is used to optimize ldt flushing.
++ */
++typedef struct {
++ void *ldt;
++ rwlock_t ldtlock;
++ int size;
++ struct semaphore sem;
++#ifdef CONFIG_XEN
++ unsigned pinned:1;
++ unsigned has_foreign_mappings:1;
++ struct list_head unpinned;
++#endif
++} mm_context_t;
++
++#ifdef CONFIG_XEN
++extern struct list_head mm_unpinned;
++extern spinlock_t mm_unpinned_lock;
++
++/* mm/memory.c:exit_mmap hook */
++extern void _arch_exit_mmap(struct mm_struct *mm);
++#define arch_exit_mmap(_mm) _arch_exit_mmap(_mm)
++
++/* kernel/fork.c:dup_mmap hook */
++extern void _arch_dup_mmap(struct mm_struct *mm);
++#define arch_dup_mmap(mm, oldmm) ((void)(oldmm), _arch_dup_mmap(mm))
++#endif
++
++#endif
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/mmu_context.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu_context.h 2008-05-19 00:34:21.744857998 +0300
@@ -0,0 +1,136 @@
+#ifndef __X86_64_MMU_CONTEXT_H
+#define __X86_64_MMU_CONTEXT_H
+}
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/mmu.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/mmu.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/mmu.h 2007-12-23 11:15:43.731777107 +0100
-@@ -0,0 +1,38 @@
-+#ifndef __x86_64_MMU_H
-+#define __x86_64_MMU_H
-+
-+#include <linux/spinlock.h>
-+#include <asm/semaphore.h>
-+
-+/*
-+ * The x86_64 doesn't have a mmu context, but
-+ * we put the segment information here.
-+ *
-+ * cpu_vm_mask is used to optimize ldt flushing.
-+ */
-+typedef struct {
-+ void *ldt;
-+ rwlock_t ldtlock;
-+ int size;
-+ struct semaphore sem;
-+#ifdef CONFIG_XEN
-+ unsigned pinned:1;
-+ unsigned has_foreign_mappings:1;
-+ struct list_head unpinned;
-+#endif
-+} mm_context_t;
-+
-+#ifdef CONFIG_XEN
-+extern struct list_head mm_unpinned;
-+extern spinlock_t mm_unpinned_lock;
-+
-+/* mm/memory.c:exit_mmap hook */
-+extern void _arch_exit_mmap(struct mm_struct *mm);
-+#define arch_exit_mmap(_mm) _arch_exit_mmap(_mm)
-+
-+/* kernel/fork.c:dup_mmap hook */
-+extern void _arch_dup_mmap(struct mm_struct *mm);
-+#define arch_dup_mmap(mm, oldmm) ((void)(oldmm), _arch_dup_mmap(mm))
-+#endif
-+
-+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/msr.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/msr.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/msr.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/msr.h 2007-12-23 11:15:43.731777107 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/msr.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/msr.h 2008-05-19 00:34:21.744857998 +0300
@@ -0,0 +1,399 @@
+#ifndef X86_64_MSR_H
+#define X86_64_MSR_H 1
+#define MSR_P4_U2L_ESCR1 0x3b1
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/nmi.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/nmi.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/nmi.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/nmi.h 2007-12-23 11:15:43.731777107 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/nmi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/nmi.h 2008-05-19 00:34:21.744857998 +0300
@@ -0,0 +1,93 @@
+/*
+ * linux/include/asm-i386/nmi.h
+#define NMI_INVALID 3
+
+#endif /* ASM_NMI_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/page.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/page.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/page.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/page.h 2007-12-23 11:15:43.731777107 +0100
-@@ -0,0 +1,209 @@
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/page.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/page.h 2008-05-19 00:34:21.748858229 +0300
+@@ -0,0 +1,212 @@
+#ifndef _X86_64_PAGE_H
+#define _X86_64_PAGE_H
+
+ * below. The preprocessor will warn if the two definitions aren't identical.
+ */
+#define _PAGE_PRESENT 0x001
++#define _PAGE_IO 0x200
+
+/* PAGE_SHIFT determines the page size */
+#define PAGE_SHIFT 12
+typedef struct { unsigned long pgprot; } pgprot_t;
+
+#define __pte_val(x) ((x).pte)
-+#define pte_val(x) ((__pte_val(x) & _PAGE_PRESENT) ? \
-+ pte_machine_to_phys(__pte_val(x)) : \
-+ __pte_val(x))
++#define pte_val(x) ((__pte_val(x) & (_PAGE_PRESENT|_PAGE_IO)) \
++ == _PAGE_PRESENT ? \
++ pte_machine_to_phys(__pte_val(x)) : \
++ __pte_val(x))
+
+#define __pmd_val(x) ((x).pmd)
+static inline unsigned long pmd_val(pmd_t x)
+
+static inline pte_t __pte(unsigned long x)
+{
-+ if (x & _PAGE_PRESENT) x = pte_phys_to_machine(x);
++ if ((x & (_PAGE_PRESENT|_PAGE_IO)) == _PAGE_PRESENT)
++ x = pte_phys_to_machine(x);
+ return ((pte_t) { (x) });
+}
+
+#endif /* __KERNEL__ */
+
+#endif /* _X86_64_PAGE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/pci.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pci.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/pci.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pci.h 2007-12-23 11:15:43.735110615 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/pci.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pci.h 2008-05-19 00:34:21.748858229 +0300
@@ -0,0 +1,168 @@
+#ifndef __x8664_PCI_H
+#define __x8664_PCI_H
+#endif
+
+#endif /* __x8664_PCI_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/pgalloc.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgalloc.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/pgalloc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgalloc.h 2007-12-23 11:15:43.735110615 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/pgalloc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgalloc.h 2008-05-19 00:34:21.748858229 +0300
@@ -0,0 +1,204 @@
+#ifndef _X86_64_PGALLOC_H
+#define _X86_64_PGALLOC_H
+#define __pud_free_tlb(tlb,x) tlb_remove_page((tlb),virt_to_page(x))
+
+#endif /* _X86_64_PGALLOC_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/pgtable.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgtable.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/pgtable.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgtable.h 2007-12-23 11:15:43.735110615 +0100
-@@ -0,0 +1,573 @@
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/pgtable.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/pgtable.h 2008-05-19 00:34:21.748858229 +0300
+@@ -0,0 +1,583 @@
+#ifndef _X86_64_PGTABLE_H
+#define _X86_64_PGTABLE_H
+
+
+extern void xen_init_pt(void);
+
-+#define virt_to_ptep(__va) \
-+({ \
-+ pgd_t *__pgd = pgd_offset_k((unsigned long)(__va)); \
-+ pud_t *__pud = pud_offset(__pgd, (unsigned long)(__va)); \
-+ pmd_t *__pmd = pmd_offset(__pud, (unsigned long)(__va)); \
-+ pte_offset_kernel(__pmd, (unsigned long)(__va)); \
-+})
++extern pte_t *lookup_address(unsigned long address);
+
-+#define arbitrary_virt_to_machine(__va) \
++#define virt_to_ptep(va) \
+({ \
-+ maddr_t m = (maddr_t)pte_mfn(*virt_to_ptep(__va)) << PAGE_SHIFT;\
-+ m | ((unsigned long)(__va) & (PAGE_SIZE-1)); \
++ pte_t *__ptep = lookup_address((unsigned long)(va)); \
++ BUG_ON(!__ptep || !pte_present(*__ptep)); \
++ __ptep; \
+})
++
++#define arbitrary_virt_to_machine(va) \
++ (((maddr_t)pte_mfn(*virt_to_ptep(va)) << PAGE_SHIFT) \
++ | ((unsigned long)(va) & (PAGE_SIZE - 1)))
+#endif
+
+extern pud_t level3_kernel_pgt[512];
+#define _PAGE_PROTNONE 0x080 /* If not present */
+#define _PAGE_NX (1UL<<_PAGE_BIT_NX)
+
++/* Mapped page is I/O or foreign and has no associated page struct. */
++#define _PAGE_IO 0x200
++
+#if CONFIG_XEN_COMPAT <= 0x030002
+extern unsigned int __kernel_page_user;
+#else
+#define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY)
+#define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY | __kernel_page_user)
+
-+#define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
++#define _PAGE_CHG_MASK (PTE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_IO)
+
+#define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)
+#define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_NX)
+#define __pte_mfn(_pte) (((_pte).pte & PTE_MASK) >> PAGE_SHIFT)
+#define pte_mfn(_pte) ((_pte).pte & _PAGE_PRESENT ? \
+ __pte_mfn(_pte) : pfn_to_mfn(__pte_mfn(_pte)))
-+#define pte_pfn(_pte) ((_pte).pte & _PAGE_PRESENT ? \
-+ mfn_to_local_pfn(__pte_mfn(_pte)) : __pte_mfn(_pte))
++#define pte_pfn(_pte) ((_pte).pte & _PAGE_IO ? end_pfn : \
++ (_pte).pte & _PAGE_PRESENT ? \
++ mfn_to_local_pfn(__pte_mfn(_pte)) : \
++ __pte_mfn(_pte))
+
+#define pte_page(x) pfn_to_page(pte_pfn(x))
+
+ unsigned long address,
+ unsigned long size);
+
++int xen_change_pte_range(struct mm_struct *mm, pmd_t *pmd,
++ unsigned long addr, unsigned long end, pgprot_t newprot);
++
++#define arch_change_pte_range(mm, pmd, addr, end, newprot) \
++ xen_change_pte_range(mm, pmd, addr, end, newprot)
++
+#define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \
+ direct_remap_pfn_range(vma,vaddr,pfn,size,prot,DOMID_IO)
+
+#include <asm-generic/pgtable.h>
+
+#endif /* _X86_64_PGTABLE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/processor.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/processor.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/processor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/processor.h 2007-12-23 11:15:43.735110615 +0100
-@@ -0,0 +1,506 @@
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/processor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/processor.h 2008-05-19 00:34:21.752858459 +0300
+@@ -0,0 +1,502 @@
+/*
+ * include/asm-x86_64/processor.h
+ *
+
+
+/*
-+ * Bus types
-+ */
-+#define MCA_bus 0
-+#define MCA_bus__is_a_macro
-+
-+/*
+ * User space process size. 47bits minus one guard page.
+ */
+#define TASK_SIZE64 (0x800000000000UL - 4096)
+
+#define get_debugreg(var, register) \
+ var = HYPERVISOR_get_debugreg(register)
-+#define set_debugreg(value, register) \
-+ HYPERVISOR_set_debugreg(register, value)
++#define set_debugreg(value, register) do { \
++ if (HYPERVISOR_set_debugreg(register, value)) \
++ BUG(); \
++} while (0)
+
+struct task_struct;
+struct mm_struct;
+#define HAVE_ARCH_PICK_MMAP_LAYOUT 1
+
+#endif /* __ASM_X86_64_PROCESSOR_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/ptrace.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/ptrace.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/ptrace.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/ptrace.h 2007-12-23 11:15:43.735110615 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/ptrace.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/ptrace.h 2008-05-19 00:34:21.752858459 +0300
@@ -0,0 +1,127 @@
+#ifndef _X86_64_PTRACE_H
+#define _X86_64_PTRACE_H
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/smp.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/smp.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/smp.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/smp.h 2007-12-23 11:15:43.735110615 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/smp.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/smp.h 2008-05-19 00:34:21.752858459 +0300
@@ -0,0 +1,150 @@
+#ifndef __ASM_SMP_H
+#define __ASM_SMP_H
+
+#endif
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/synch_bitops.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/synch_bitops.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/synch_bitops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/synch_bitops.h 2007-12-23 11:15:43.735110615 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/synch_bitops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/synch_bitops.h 2008-05-19 00:34:21.756858690 +0300
@@ -0,0 +1,2 @@
+
+#include <asm-i386/mach-xen/asm/synch_bitops.h>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/system.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/system.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/system.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/system.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/system.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/system.h 2008-05-19 00:34:21.756858690 +0300
@@ -0,0 +1,256 @@
+#ifndef __ASM_SYSTEM_H
+#define __ASM_SYSTEM_H
+extern void free_init_pages(char *what, unsigned long begin, unsigned long end);
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/timer.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/timer.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/timer.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/timer.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/timer.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/timer.h 2008-05-19 00:34:21.756858690 +0300
@@ -0,0 +1,67 @@
+#ifndef _ASMi386_TIMER_H
+#define _ASMi386_TIMER_H
+extern struct init_timer_opts timer_pmtmr_init;
+#endif
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/tlbflush.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/tlbflush.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/tlbflush.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/tlbflush.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/tlbflush.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/tlbflush.h 2008-05-19 00:34:21.756858690 +0300
@@ -0,0 +1,103 @@
+#ifndef _X8664_TLBFLUSH_H
+#define _X8664_TLBFLUSH_H
+}
+
+#endif /* _X8664_TLBFLUSH_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/vga.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/vga.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/vga.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/vga.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/vga.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/vga.h 2008-05-19 00:34:21.760858920 +0300
@@ -0,0 +1,20 @@
+/*
+ * Access to VGA videoram
+#define vga_writeb(x,y) (*(y) = (x))
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/xenoprof.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xenoprof.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/xenoprof.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xenoprof.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/xenoprof.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xenoprof.h 2008-05-19 00:34:21.760858920 +0300
@@ -0,0 +1 @@
+#include <asm-i386/mach-xen/asm/xenoprof.h>
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/asm/xor.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xor.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/asm/xor.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xor.h 2007-12-23 11:15:43.738444121 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/asm/xor.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/asm/xor.h 2008-05-19 00:34:21.760858920 +0300
@@ -0,0 +1,328 @@
+/*
+ * x86-64 changes / gcc fixes from Andi Kleen.
+ We may also be able to load into the L1 only depending on how the cpu
+ deals with a load to a line that is being prefetched. */
+#define XOR_SELECT_TEMPLATE(FASTEST) (&xor_block_sse)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/irq_vectors.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/irq_vectors.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/irq_vectors.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/irq_vectors.h 2007-12-23 11:15:43.741777625 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/irq_vectors.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/irq_vectors.h 2008-05-19 00:34:21.760858920 +0300
@@ -0,0 +1,123 @@
+/*
+ * This file should contain #defines for all of the interrupt vector
+#define irq_to_dynirq(_x) ((_x) - DYNIRQ_BASE)
+
+#endif /* _ASM_IRQ_VECTORS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/mach_time.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_time.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/mach_time.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_time.h 2007-12-23 11:15:43.741777625 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/mach_time.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_time.h 2008-05-19 00:34:21.764859151 +0300
@@ -0,0 +1,111 @@
+/*
+ * include/asm-i386/mach-default/mach_time.h
+}
+
+#endif /* !_MACH_TIME_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/mach_timer.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_timer.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/mach_timer.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_timer.h 2007-12-23 11:15:43.741777625 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/mach_timer.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/mach_timer.h 2008-05-19 00:34:21.764859151 +0300
@@ -0,0 +1,50 @@
+/*
+ * include/asm-i386/mach-default/mach_timer.h
+}
+
+#endif /* !_MACH_TIMER_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/setup_arch_post.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_post.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/setup_arch_post.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_post.h 2007-12-23 11:15:43.741777625 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/setup_arch_post.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_post.h 2008-05-19 00:34:21.764859151 +0300
@@ -0,0 +1,63 @@
+/**
+ * machine_specific_* - Hooks for machine specific setup.
+#endif
+#endif
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/asm-x86_64/mach-xen/setup_arch_pre.h linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_pre.h
---- linux-2.6.18/include/asm-x86_64/mach-xen/setup_arch_pre.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_pre.h 2007-12-23 11:15:43.741777625 +0100
+--- linux-2.6.18.8/include/asm-x86_64/mach-xen/setup_arch_pre.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/mach-xen/setup_arch_pre.h 2008-05-19 00:34:21.764859151 +0300
@@ -0,0 +1,5 @@
+/* Hook to call BIOS initialisation function */
+
+#define ARCH_SETUP machine_specific_arch_setup();
+
+static void __init machine_specific_arch_setup(void);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/aio.h linux-2.6.18-xen.hg/include/linux/aio.h
---- linux-2.6.18/include/linux/aio.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/aio.h 2007-12-23 11:15:43.865117435 +0100
+--- linux-2.6.18.8/include/asm-x86_64/msi.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/msi.h 2008-05-19 00:34:21.776859843 +0300
+@@ -7,14 +7,21 @@
+ #define ASM_MSI_H
+
+ #include <asm/desc.h>
++#ifndef CONFIG_XEN
+ #include <asm/mach_apic.h>
++#endif
+ #include <asm/smp.h>
+
++#ifndef CONFIG_XEN
+ #define LAST_DEVICE_VECTOR (FIRST_SYSTEM_VECTOR - 1)
++#else
++#define LAST_DYNAMIC_VECTOR 0xdf
++#define LAST_DEVICE_VECTOR (LAST_DYNAMIC_VECTOR)
++#endif
++
+ #define MSI_TARGET_CPU_SHIFT 12
+
+ extern struct msi_ops msi_apic_ops;
+-
+ static inline int msi_arch_init(void)
+ {
+ msi_register(&msi_apic_ops);
+--- linux-2.6.18.8/include/asm-x86_64/signal.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/signal.h 2008-05-19 00:34:21.836863301 +0300
+@@ -24,10 +24,6 @@
+ } sigset_t;
+
+
+-struct pt_regs;
+-asmlinkage int do_signal(struct pt_regs *regs, sigset_t *oldset);
+-
+-
+ #else
+ /* Here we must cater to libcs that poke about in kernel headers. */
+
+--- linux-2.6.18.8/include/asm-x86_64/thread_info.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/thread_info.h 2008-05-19 00:34:21.872865377 +0300
+@@ -114,6 +114,7 @@
+ #define TIF_IRET 5 /* force IRET */
+ #define TIF_SYSCALL_AUDIT 7 /* syscall auditing active */
+ #define TIF_SECCOMP 8 /* secure computing */
++#define TIF_RESTORE_SIGMASK 9 /* restore signal mask in do_signal */
+ /* 16 free */
+ #define TIF_IA32 17 /* 32bit process */
+ #define TIF_FORK 18 /* ret_from_fork */
+@@ -128,6 +129,7 @@
+ #define _TIF_IRET (1<<TIF_IRET)
+ #define _TIF_SYSCALL_AUDIT (1<<TIF_SYSCALL_AUDIT)
+ #define _TIF_SECCOMP (1<<TIF_SECCOMP)
++#define _TIF_RESTORE_SIGMASK (1<<TIF_RESTORE_SIGMASK)
+ #define _TIF_IA32 (1<<TIF_IA32)
+ #define _TIF_FORK (1<<TIF_FORK)
+ #define _TIF_ABI_PENDING (1<<TIF_ABI_PENDING)
+--- linux-2.6.18.8/include/asm-x86_64/unistd.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/asm-x86_64/unistd.h 2008-05-19 00:34:21.880865838 +0300
+@@ -600,9 +600,9 @@
+ #define __NR_faccessat 269
+ __SYSCALL(__NR_faccessat, sys_faccessat)
+ #define __NR_pselect6 270
+-__SYSCALL(__NR_pselect6, sys_ni_syscall) /* for now */
++__SYSCALL(__NR_pselect6, sys_pselect6)
+ #define __NR_ppoll 271
+-__SYSCALL(__NR_ppoll, sys_ni_syscall) /* for now */
++__SYSCALL(__NR_ppoll, sys_ppoll)
+ #define __NR_unshare 272
+ __SYSCALL(__NR_unshare, sys_unshare)
+ #define __NR_set_robust_list 273
+@@ -658,6 +658,7 @@
+ #define __ARCH_WANT_SYS_SIGPENDING
+ #define __ARCH_WANT_SYS_SIGPROCMASK
+ #define __ARCH_WANT_SYS_RT_SIGACTION
++#define __ARCH_WANT_SYS_RT_SIGSUSPEND
+ #define __ARCH_WANT_SYS_TIME
+ #define __ARCH_WANT_COMPAT_SYS_TIME
+
+--- linux-2.6.18.8/include/linux/aio.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/aio.h 2008-05-19 00:34:23.376952075 +0300
@@ -191,6 +191,11 @@
struct aio_ring_info ring_info;
};
/* prototypes */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/crash_dump.h linux-2.6.18-xen.hg/include/linux/crash_dump.h
---- linux-2.6.18/include/linux/crash_dump.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/crash_dump.h 2007-12-23 11:15:43.918453570 +0100
+--- linux-2.6.18.8/include/linux/bio.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/bio.h 2008-05-19 00:34:23.592964527 +0300
+@@ -172,12 +172,27 @@
+ #define bio_offset(bio) bio_iovec((bio))->bv_offset
+ #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
+ #define bio_sectors(bio) ((bio)->bi_size >> 9)
+-#define bio_cur_sectors(bio) (bio_iovec(bio)->bv_len >> 9)
+-#define bio_data(bio) (page_address(bio_page((bio))) + bio_offset((bio)))
+ #define bio_barrier(bio) ((bio)->bi_rw & (1 << BIO_RW_BARRIER))
+ #define bio_sync(bio) ((bio)->bi_rw & (1 << BIO_RW_SYNC))
+ #define bio_failfast(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST))
+ #define bio_rw_ahead(bio) ((bio)->bi_rw & (1 << BIO_RW_AHEAD))
++#define bio_empty_barrier(bio) (bio_barrier(bio) && !(bio)->bi_size)
++
++static inline unsigned int bio_cur_sectors(struct bio *bio)
++{
++ if (bio->bi_vcnt)
++ return bio_iovec(bio)->bv_len >> 9;
++
++ return 0;
++}
++
++static inline void *bio_data(struct bio *bio)
++{
++ if (bio->bi_vcnt)
++ return page_address(bio_page(bio)) + bio_offset(bio);
++
++ return NULL;
++}
+
+ /*
+ * will die
+--- linux-2.6.18.8/include/linux/blkdev.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/blkdev.h 2008-05-19 00:34:23.596964757 +0300
+@@ -506,6 +506,8 @@
+ #define blk_barrier_rq(rq) ((rq)->flags & REQ_HARDBARRIER)
+ #define blk_fua_rq(rq) ((rq)->flags & REQ_FUA)
+
++#define blk_empty_barrier(rq) (blk_barrier_rq(rq) && blk_fs_request(rq) && !(rq)->hard_nr_sectors)
++
+ #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
+
+ #define rq_data_dir(rq) ((rq)->flags & 1)
+--- linux-2.6.18.8/include/linux/crash_dump.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/crash_dump.h 2008-05-19 00:34:24.092993349 +0300
@@ -14,5 +14,13 @@
extern const struct file_operations proc_vmcore_operations;
extern struct proc_dir_entry *proc_vmcore;
+
#endif /* CONFIG_CRASH_DUMP */
#endif /* LINUX_CRASHDUMP_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/elfnote.h linux-2.6.18-xen.hg/include/linux/elfnote.h
---- linux-2.6.18/include/linux/elfnote.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/linux/elfnote.h 2007-12-23 11:15:43.981790229 +0100
+--- linux-2.6.18.8/include/linux/debugfs.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/debugfs.h 2008-05-19 00:34:24.132995655 +0300
+@@ -41,6 +41,8 @@
+ struct dentry *parent, u16 *value);
+ struct dentry *debugfs_create_u32(const char *name, mode_t mode,
+ struct dentry *parent, u32 *value);
++struct dentry *debugfs_create_u64(const char *name, mode_t mode,
++ struct dentry *parent, u64 *value);
+ struct dentry *debugfs_create_bool(const char *name, mode_t mode,
+ struct dentry *parent, u32 *value);
+
+@@ -94,6 +96,13 @@
+ return ERR_PTR(-ENODEV);
+ }
+
++static inline struct dentry *debugfs_create_u64(const char *name, mode_t mode,
++ struct dentry *parent,
++ u64 *value)
++{
++ return ERR_PTR(-ENODEV);
++}
++
+ static inline struct dentry *debugfs_create_bool(const char *name, mode_t mode,
+ struct dentry *parent,
+ u32 *value)
+--- linux-2.6.18.8/include/linux/elfnote.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/elfnote.h 2008-05-19 00:34:24.285004417 +0300
@@ -0,0 +1,104 @@
+#ifndef _LINUX_ELFNOTE_H
+#define _LINUX_ELFNOTE_H
+#endif /* __ASSEMBLER__ */
+
+#endif /* _LINUX_ELFNOTE_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/eventpoll.h linux-2.6.18-xen.hg/include/linux/eventpoll.h
---- linux-2.6.18/include/linux/eventpoll.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/eventpoll.h 2007-12-23 11:15:43.981790229 +0100
+--- linux-2.6.18.8/include/linux/eventpoll.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/eventpoll.h 2008-05-19 00:34:24.289004648 +0300
@@ -90,6 +90,12 @@
eventpoll_release_file(file);
}
#else
static inline void eventpoll_init_file(struct file *file) {}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/highmem.h linux-2.6.18-xen.hg/include/linux/highmem.h
---- linux-2.6.18/include/linux/highmem.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/highmem.h 2007-12-23 11:15:44.051793905 +0100
+--- linux-2.6.18.8/include/linux/highmem.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/highmem.h 2008-05-19 00:34:24.737030473 +0300
@@ -24,10 +24,16 @@
/* declarations for linux/mm/highmem.c */
static inline void *kmap(struct page *page)
{
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/interrupt.h linux-2.6.18-xen.hg/include/linux/interrupt.h
---- linux-2.6.18/include/linux/interrupt.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/interrupt.h 2007-12-23 11:15:44.531819112 +0100
+--- linux-2.6.18.8/include/linux/interrupt.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/interrupt.h 2008-05-19 00:34:25.057048919 +0300
@@ -166,6 +166,12 @@
#endif /* CONFIG_GENERIC_HARDIRQS */
#ifndef __ARCH_SET_SOFTIRQ_PENDING
#define set_softirq_pending(x) (local_softirq_pending() = (x))
#define or_softirq_pending(x) (local_softirq_pending() |= (x))
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/ioport.h linux-2.6.18-xen.hg/include/linux/ioport.h
---- linux-2.6.18/include/linux/ioport.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/ioport.h 2007-12-23 11:15:44.535152621 +0100
+--- linux-2.6.18.8/include/linux/ioport.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/ioport.h 2008-05-19 00:34:25.073049842 +0300
@@ -93,6 +93,9 @@
/* PC/ISA/whatever - the normal PC address spaces: IO and memory */
extern struct resource ioport_resource;
extern struct resource iomem_resource;
-+#ifdef CONFIG_XEN
++#ifdef CONFIG_PROC_IOMEM_MACHINE
+extern struct resource iomem_machine_resource;
+#endif
extern int request_resource(struct resource *root, struct resource *new);
extern struct resource * ____request_resource(struct resource *root, struct resource *new);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/kexec.h linux-2.6.18-xen.hg/include/linux/kexec.h
---- linux-2.6.18/include/linux/kexec.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/kexec.h 2007-12-23 11:15:44.931840116 +0100
+--- linux-2.6.18.8/include/linux/kexec.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/kexec.h 2008-05-19 00:34:25.305063215 +0300
@@ -31,6 +31,13 @@
#error KEXEC_ARCH not defined
#endif
#else /* !CONFIG_KEXEC */
struct pt_regs;
struct task_struct;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/mm.h linux-2.6.18-xen.hg/include/linux/mm.h
---- linux-2.6.18/include/linux/mm.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/mm.h 2007-12-23 11:15:44.965175206 +0100
-@@ -164,6 +164,9 @@
+--- linux-2.6.18.8/include/linux/mm.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/mm.h 2008-05-19 00:34:25.417069672 +0300
+@@ -164,6 +164,10 @@
#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
+#ifdef CONFIG_XEN
+#define VM_FOREIGN 0x04000000 /* Has pages belonging to another VM */
+#endif
++#define VM_ALWAYSDUMP 0x08000000 /* Always include in core dumps */
#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
-@@ -202,6 +205,10 @@
+@@ -202,6 +206,10 @@
/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
#ifdef CONFIG_NUMA
int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
-@@ -1027,6 +1034,13 @@
+@@ -1027,6 +1035,13 @@
#define FOLL_GET 0x04 /* do get_page on page */
#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
#ifdef CONFIG_PROC_FS
void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
#else
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/netfilter_bridge.h linux-2.6.18-xen.hg/include/linux/netfilter_bridge.h
---- linux-2.6.18/include/linux/netfilter_bridge.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/netfilter_bridge.h 2007-12-23 11:15:45.015177827 +0100
+--- linux-2.6.18.8/include/linux/moduleparam.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/moduleparam.h 2008-05-19 00:34:25.449071516 +0300
+@@ -58,13 +58,23 @@
+ void *elem;
+ };
+
++/* On alpha, ia64 and ppc64 relocations to global data cannot go into
++ read-only sections (which is part of respective UNIX ABI on these
++ platforms). So 'const' makes no sense and even causes compile failures
++ with some compilers. */
++#if defined(CONFIG_ALPHA) || defined(CONFIG_IA64) || defined(CONFIG_PPC64)
++#define __moduleparam_const
++#else
++#define __moduleparam_const const
++#endif
++
+ /* This is the fundamental function for registering boot/module
+ parameters. perm sets the visibility in driverfs: 000 means it's
+ not there, read bits mean it's readable, write bits mean it's
+ writable. */
+ #define __module_param_call(prefix, name, set, get, arg, perm) \
+ static char __param_str_##name[] = prefix #name; \
+- static struct kernel_param const __param_##name \
++ static struct kernel_param __moduleparam_const __param_##name \
+ __attribute_used__ \
+ __attribute__ ((unused,__section__ ("__param"),aligned(sizeof(void *)))) \
+ = { __param_str_##name, perm, set, get, arg }
+--- linux-2.6.18.8/include/linux/netfilter_bridge.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/netfilter_bridge.h 2008-05-19 00:34:25.837093883 +0300
@@ -7,6 +7,7 @@
#include <linux/netfilter.h>
#if defined(__KERNEL__) && defined(CONFIG_BRIDGE_NETFILTER)
}
}
return 0;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/oprofile.h linux-2.6.18-xen.hg/include/linux/oprofile.h
---- linux-2.6.18/include/linux/oprofile.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/oprofile.h 2007-12-23 11:15:45.098515532 +0100
+--- linux-2.6.18.8/include/linux/oprofile.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/oprofile.h 2008-05-19 00:34:26.497131929 +0300
@@ -17,6 +17,8 @@
#include <linux/spinlock.h>
#include <asm/atomic.h>
/**
* Create a file of the given name as a child of the given root, with
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/page-flags.h linux-2.6.18-xen.hg/include/linux/page-flags.h
---- linux-2.6.18/include/linux/page-flags.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/page-flags.h 2007-12-23 11:15:45.098515532 +0100
+--- linux-2.6.18.8/include/linux/page-flags.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/page-flags.h 2008-05-19 00:34:26.501132159 +0300
@@ -98,6 +98,8 @@
#define PG_uncached 31 /* Page has been mapped as uncached */
#endif
/*
* Manipulation of page state flags
*/
-@@ -247,6 +249,18 @@
+@@ -247,6 +249,19 @@
#define SetPageUncached(page) set_bit(PG_uncached, &(page)->flags)
#define ClearPageUncached(page) clear_bit(PG_uncached, &(page)->flags)
+#define PageForeign(page) test_bit(PG_foreign, &(page)->flags)
-+#define SetPageForeign(page, dtor) do { \
-+ set_bit(PG_foreign, &(page)->flags); \
-+ (page)->index = (long)(dtor); \
++#define SetPageForeign(_page, dtor) do { \
++ set_bit(PG_foreign, &(_page)->flags); \
++ BUG_ON((dtor) == (void (*)(struct page *))0); \
++ (_page)->index = (long)(dtor); \
+} while (0)
-+#define ClearPageForeign(page) do { \
-+ clear_bit(PG_foreign, &(page)->flags); \
-+ (page)->index = 0; \
++#define ClearPageForeign(page) do { \
++ clear_bit(PG_foreign, &(page)->flags); \
++ (page)->index = 0; \
+} while (0)
-+#define PageForeignDestructor(page) \
-+ ( (void (*) (struct page *)) (page)->index )(page)
++#define PageForeignDestructor(_page) \
++ ((void (*)(struct page *))(_page)->index)(_page)
+
struct page; /* forward declaration */
int test_clear_page_dirty(struct page *page);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/pci_ids.h linux-2.6.18-xen.hg/include/linux/pci_ids.h
---- linux-2.6.18/include/linux/pci_ids.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/pci_ids.h 2007-12-23 11:15:45.105182559 +0100
+--- linux-2.6.18.8/include/linux/pci.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/pci.h 2008-05-19 00:34:26.509132620 +0300
+@@ -152,6 +152,9 @@
+ * directly, use the values stored here. They might be different!
+ */
+ unsigned int irq;
++#ifdef CONFIG_XEN
++ unsigned int irq_old;
++#endif
+ struct resource resource[DEVICE_COUNT_RESOURCE]; /* I/O and memory regions + expansion ROMs */
+
+ /* These fields are used by common fixups */
+--- linux-2.6.18.8/include/linux/pci_ids.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/pci_ids.h 2008-05-19 00:34:26.509132620 +0300
@@ -2188,6 +2188,13 @@
#define PCI_DEVICE_ID_INTEL_ICH8_4 0x2815
#define PCI_DEVICE_ID_INTEL_ICH8_5 0x283e
#define PCI_DEVICE_ID_INTEL_82855PM_HB 0x3340
#define PCI_DEVICE_ID_INTEL_82830_HB 0x3575
#define PCI_DEVICE_ID_INTEL_82830_CGC 0x3577
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/sched.h linux-2.6.18-xen.hg/include/linux/sched.h
---- linux-2.6.18/include/linux/sched.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/sched.h 2007-12-23 11:15:45.168519210 +0100
+@@ -2202,6 +2209,12 @@
+ #define PCI_DEVICE_ID_INTEL_MCH_PC 0x3599
+ #define PCI_DEVICE_ID_INTEL_MCH_PC1 0x359a
+ #define PCI_DEVICE_ID_INTEL_E7525_MCH 0x359e
++#define PCI_DEVICE_ID_INTEL_ICH10_0 0x3a14
++#define PCI_DEVICE_ID_INTEL_ICH10_1 0x3a16
++#define PCI_DEVICE_ID_INTEL_ICH10_2 0x3a18
++#define PCI_DEVICE_ID_INTEL_ICH10_3 0x3a1a
++#define PCI_DEVICE_ID_INTEL_ICH10_4 0x3a30
++#define PCI_DEVICE_ID_INTEL_ICH10_5 0x3a60
+ #define PCI_DEVICE_ID_INTEL_82371SB_0 0x7000
+ #define PCI_DEVICE_ID_INTEL_82371SB_1 0x7010
+ #define PCI_DEVICE_ID_INTEL_82371SB_2 0x7020
+--- linux-2.6.18.8/include/linux/sched.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/sched.h 2008-05-19 00:34:26.605138154 +0300
@@ -211,10 +211,15 @@
extern void scheduler_tick(void);
static inline void softlockup_tick(void)
{
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/skbuff.h linux-2.6.18-xen.hg/include/linux/skbuff.h
---- linux-2.6.18/include/linux/skbuff.h 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/include/linux/skbuff.h 2007-12-23 11:15:45.181853251 +0100
+--- linux-2.6.18.8/include/linux/skbuff.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/skbuff.h 2008-05-19 00:34:26.633139768 +0300
@@ -203,6 +203,8 @@
* @local_df: allow local fragmentation
* @cloned: Head may be cloned (check refcnt to be sure)
__be16 protocol;
void (*destructor)(struct sk_buff *skb);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/linux/sysctl.h linux-2.6.18-xen.hg/include/linux/sysctl.h
---- linux-2.6.18/include/linux/sysctl.h 2007-12-23 11:26:51.286801791 +0100
-+++ linux-2.6.18-xen.hg/include/linux/sysctl.h 2007-12-23 11:15:45.848554922 +0100
+--- linux-2.6.18.8/include/linux/sysctl.h 2008-05-19 00:42:34.397258564 +0300
++++ linux-2.6.18-xen.hg/include/linux/sysctl.h 2008-05-19 00:34:26.689142997 +0300
@@ -6,10 +6,17 @@
****************************************************************
****************************************************************
#endif
enum
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/balloon.h linux-2.6.18-xen.hg/include/xen/balloon.h
---- linux-2.6.18/include/xen/balloon.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/balloon.h 2007-12-23 11:15:46.468587483 +0100
+--- linux-2.6.18.8/include/linux/vermagic.h 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/include/linux/vermagic.h 2008-05-19 00:34:26.765147378 +0300
+@@ -17,6 +17,11 @@
+ #else
+ #define MODULE_VERMAGIC_MODULE_UNLOAD ""
+ #endif
++#ifdef CONFIG_XEN
++#define MODULE_VERMAGIC_XEN "Xen "
++#else
++#define MODULE_VERMAGIC_XEN
++#endif
+ #ifndef MODULE_ARCH_VERMAGIC
+ #define MODULE_ARCH_VERMAGIC ""
+ #endif
+@@ -24,5 +29,6 @@
+ #define VERMAGIC_STRING \
+ UTS_RELEASE " " \
+ MODULE_VERMAGIC_SMP MODULE_VERMAGIC_PREEMPT \
+- MODULE_VERMAGIC_MODULE_UNLOAD MODULE_ARCH_VERMAGIC \
++ MODULE_VERMAGIC_MODULE_UNLOAD MODULE_VERMAGIC_XEN \
++ MODULE_ARCH_VERMAGIC \
+ "gcc-" __stringify(__GNUC__) "." __stringify(__GNUC_MINOR__)
+--- linux-2.6.18.8/include/xen/balloon.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/balloon.h 2008-05-19 00:34:28.317236843 +0300
@@ -0,0 +1,57 @@
+/******************************************************************************
+ * balloon.h
+#define balloon_unlock(__flags) spin_unlock_irqrestore(&balloon_lock, __flags)
+
+#endif /* __ASM_BALLOON_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/blkif.h linux-2.6.18-xen.hg/include/xen/blkif.h
---- linux-2.6.18/include/xen/blkif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/blkif.h 2007-12-23 11:15:46.468587483 +0100
-@@ -0,0 +1,101 @@
+--- linux-2.6.18.8/include/xen/blkif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/blkif.h 2008-05-19 00:34:28.317236843 +0300
+@@ -0,0 +1,123 @@
++/*
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ */
++
+#ifndef __XEN_BLKIF_H__
+#define __XEN_BLKIF_H__
+
+ dst->handle = src->handle;
+ dst->id = src->id;
+ dst->sector_number = src->sector_number;
-+ if (n > src->nr_segments)
-+ n = src->nr_segments;
++ barrier();
++ if (n > dst->nr_segments)
++ n = dst->nr_segments;
+ for (i = 0; i < n; i++)
+ dst->seg[i] = src->seg[i];
+}
+ dst->handle = src->handle;
+ dst->id = src->id;
+ dst->sector_number = src->sector_number;
-+ if (n > src->nr_segments)
-+ n = src->nr_segments;
++ barrier();
++ if (n > dst->nr_segments)
++ n = dst->nr_segments;
+ for (i = 0; i < n; i++)
+ dst->seg[i] = src->seg[i];
+}
+
+#endif /* __XEN_BLKIF_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/compat_ioctl.h linux-2.6.18-xen.hg/include/xen/compat_ioctl.h
---- linux-2.6.18/include/xen/compat_ioctl.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/compat_ioctl.h 2007-12-23 11:15:46.468587483 +0100
+--- linux-2.6.18.8/include/xen/compat_ioctl.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/compat_ioctl.h 2008-05-19 00:34:28.321237074 +0300
@@ -0,0 +1,45 @@
+/*
+ * This program is free software; you can redistribute it and/or
+ _IOC(_IOC_NONE, 'P', 3, sizeof(struct privcmd_mmapbatch_32))
+
+#endif /* __LINUX_XEN_COMPAT_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/cpu_hotplug.h linux-2.6.18-xen.hg/include/xen/cpu_hotplug.h
---- linux-2.6.18/include/xen/cpu_hotplug.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/cpu_hotplug.h 2007-12-23 11:15:46.468587483 +0100
+--- linux-2.6.18.8/include/xen/cpu_hotplug.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/cpu_hotplug.h 2008-05-19 00:34:28.321237074 +0300
@@ -0,0 +1,41 @@
+#ifndef __XEN_CPU_HOTPLUG_H__
+#define __XEN_CPU_HOTPLUG_H__
+#endif /* !defined(CONFIG_HOTPLUG_CPU) */
+
+#endif /* __XEN_CPU_HOTPLUG_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/driver_util.h linux-2.6.18-xen.hg/include/xen/driver_util.h
---- linux-2.6.18/include/xen/driver_util.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/driver_util.h 2007-12-23 11:15:46.468587483 +0100
+--- linux-2.6.18.8/include/xen/driver_util.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/driver_util.h 2008-05-19 00:34:28.321237074 +0300
@@ -0,0 +1,14 @@
+
+#ifndef __ASM_XEN_DRIVER_UTIL_H__
+extern struct class *get_xen_class(void);
+
+#endif /* __ASM_XEN_DRIVER_UTIL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/evtchn.h linux-2.6.18-xen.hg/include/xen/evtchn.h
---- linux-2.6.18/include/xen/evtchn.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/evtchn.h 2007-12-23 11:15:46.468587483 +0100
-@@ -0,0 +1,139 @@
+--- linux-2.6.18.8/include/xen/evtchn.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/evtchn.h 2008-05-19 00:34:28.321237074 +0300
+@@ -0,0 +1,153 @@
+/******************************************************************************
+ * evtchn.h
+ *
+static inline void notify_remote_via_evtchn(int port)
+{
+ struct evtchn_send send = { .port = port };
-+ (void)HYPERVISOR_event_channel_op(EVTCHNOP_send, &send);
++ VOID(HYPERVISOR_event_channel_op(EVTCHNOP_send, &send));
+}
+
+/*
+void notify_remote_via_irq(int irq);
+int irq_to_evtchn_port(int irq);
+
++#define PIRQ_SET_MAPPING 0x0
++#define PIRQ_CLEAR_MAPPING 0x1
++#define PIRQ_GET_MAPPING 0x3
++int pirq_mapstatus(int pirq, int action);
++int set_pirq_hw_action(int pirq, int (*action)(int pirq, int action));
++int clear_pirq_hw_action(int pirq);
++
++#define PIRQ_STARTUP 1
++#define PIRQ_SHUTDOWN 2
++#define PIRQ_ENABLE 3
++#define PIRQ_DISABLE 4
++#define PIRQ_END 5
++#define PIRQ_ACK 6
++
+#endif /* __ASM_EVTCHN_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/features.h linux-2.6.18-xen.hg/include/xen/features.h
---- linux-2.6.18/include/xen/features.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/features.h 2007-12-23 11:15:46.471920991 +0100
+--- linux-2.6.18.8/include/xen/features.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/features.h 2008-05-19 00:34:28.321237074 +0300
@@ -0,0 +1,20 @@
+/******************************************************************************
+ * features.h
+#define xen_feature(flag) (xen_features[flag])
+
+#endif /* __ASM_XEN_FEATURES_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/firmware.h linux-2.6.18-xen.hg/include/xen/firmware.h
---- linux-2.6.18/include/xen/firmware.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/firmware.h 2007-12-23 11:15:46.471920991 +0100
+--- linux-2.6.18.8/include/xen/firmware.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/firmware.h 2008-05-19 00:34:28.325237304 +0300
@@ -0,0 +1,10 @@
+#ifndef __XEN_FIRMWARE_H__
+#define __XEN_FIRMWARE_H__
+void copy_edid(void);
+
+#endif /* __XEN_FIRMWARE_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/gnttab.h linux-2.6.18-xen.hg/include/xen/gnttab.h
---- linux-2.6.18/include/xen/gnttab.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/gnttab.h 2007-12-23 11:15:46.471920991 +0100
+--- linux-2.6.18.8/include/xen/gnttab.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/gnttab.h 2008-05-19 00:34:28.325237304 +0300
@@ -0,0 +1,163 @@
+/******************************************************************************
+ * gnttab.h
+}
+
+#endif /* __ASM_GNTTAB_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/hvm.h linux-2.6.18-xen.hg/include/xen/hvm.h
---- linux-2.6.18/include/xen/hvm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/hvm.h 2007-12-23 11:15:46.471920991 +0100
+--- linux-2.6.18.8/include/xen/hvm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/hvm.h 2008-05-19 00:34:28.325237304 +0300
@@ -0,0 +1,23 @@
+/* Simple wrappers around HVM functions */
+#ifndef XEN_HVM_H__
+}
+
+#endif /* XEN_HVM_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/hypercall.h linux-2.6.18-xen.hg/include/xen/hypercall.h
---- linux-2.6.18/include/xen/hypercall.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/hypercall.h 2007-12-23 11:15:46.471920991 +0100
-@@ -0,0 +1,24 @@
+--- linux-2.6.18.8/include/xen/hypercall.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/hypercall.h 2008-05-19 00:34:28.325237304 +0300
+@@ -0,0 +1,30 @@
+#ifndef __XEN_HYPERCALL_H__
+#define __XEN_HYPERCALL_H__
+
+#include <asm/hypercall.h>
+
-+static inline int
++static inline int __must_check
+HYPERVISOR_multicall_check(
-+ multicall_entry_t *call_list, int nr_calls,
++ multicall_entry_t *call_list, unsigned int nr_calls,
+ const unsigned long *rc_list)
+{
+ int rc = HYPERVISOR_multicall(call_list, nr_calls);
+ if (unlikely(rc < 0))
+ return rc;
+ BUG_ON(rc);
++ BUG_ON((int)nr_calls < 0);
+
+ for ( ; nr_calls > 0; --nr_calls, ++call_list)
+ if (unlikely(call_list->result != (rc_list ? *rc_list++ : 0)))
+ return 0;
+}
+
++/* A construct to ignore the return value of hypercall wrappers in a few
++ * exceptional cases (simply casting the function result to void doesn't
++ * avoid the compiler warning): */
++#define VOID(expr) ((void)((expr)?:0))
++
+#endif /* __XEN_HYPERCALL_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/hypervisor_sysfs.h linux-2.6.18-xen.hg/include/xen/hypervisor_sysfs.h
---- linux-2.6.18/include/xen/hypervisor_sysfs.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/hypervisor_sysfs.h 2007-12-23 11:15:46.471920991 +0100
+--- linux-2.6.18.8/include/xen/hypervisor_sysfs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/hypervisor_sysfs.h 2008-05-19 00:34:28.325237304 +0300
@@ -0,0 +1,30 @@
+/*
+ * copyright (c) 2006 IBM Corporation
+};
+
+#endif /* _HYP_SYSFS_H_ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/acm.h linux-2.6.18-xen.hg/include/xen/interface/acm.h
---- linux-2.6.18/include/xen/interface/acm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/acm.h 2007-12-23 11:15:46.475254497 +0100
+--- linux-2.6.18.8/include/xen/interface/COPYING 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/COPYING 2008-05-19 00:34:28.325237304 +0300
+@@ -0,0 +1,38 @@
++XEN NOTICE
++==========
++
++This copyright applies to all files within this subdirectory and its
++subdirectories:
++ include/public/*.h
++ include/public/hvm/*.h
++ include/public/io/*.h
++
++The intention is that these files can be freely copied into the source
++tree of an operating system when porting that OS to run on Xen. Doing
++so does *not* cause the OS to become subject to the terms of the GPL.
++
++All other files in the Xen source distribution are covered by version
++2 of the GNU General Public License except where explicitly stated
++otherwise within individual source files.
++
++ -- Keir Fraser (on behalf of the Xen team)
++
++=====================================================================
++
++Permission is hereby granted, free of charge, to any person obtaining a copy
++of this software and associated documentation files (the "Software"), to
++deal in the Software without restriction, including without limitation the
++rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++sell copies of the Software, and to permit persons to whom the Software is
++furnished to do so, subject to the following conditions:
++
++The above copyright notice and this permission notice shall be included in
++all copies or substantial portions of the Software.
++
++THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++DEALINGS IN THE SOFTWARE.
+--- linux-2.6.18.8/include/xen/interface/acm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/acm.h 2008-05-19 00:34:28.329237535 +0300
@@ -0,0 +1,228 @@
+/*
+ * acm.h: Xen access control module interface defintions
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/acm_ops.h linux-2.6.18-xen.hg/include/xen/interface/acm_ops.h
---- linux-2.6.18/include/xen/interface/acm_ops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/acm_ops.h 2007-12-23 11:15:46.475254497 +0100
+--- linux-2.6.18.8/include/xen/interface/acm_ops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/acm_ops.h 2008-05-19 00:34:28.329237535 +0300
@@ -0,0 +1,159 @@
+/*
+ * acm_ops.h: Xen access control module hypervisor commands
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-ia64/debug_op.h linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/debug_op.h
---- linux-2.6.18/include/xen/interface/arch-ia64/debug_op.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/debug_op.h 2007-12-23 11:15:46.475254497 +0100
+--- linux-2.6.18.8/include/xen/interface/arch-ia64/debug_op.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/debug_op.h 2008-05-19 00:34:28.333237766 +0300
@@ -0,0 +1,96 @@
+/******************************************************************************
+ * debug_op.h
+DEFINE_XEN_GUEST_HANDLE(xen_ia64_debug_op_t);
+
+#endif /* __XEN_PUBLIC_IA64_DEBUG_OP_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-ia64/hvm/save.h linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/hvm/save.h
---- linux-2.6.18/include/xen/interface/arch-ia64/hvm/save.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/hvm/save.h 2007-12-23 11:15:46.475254497 +0100
+--- linux-2.6.18.8/include/xen/interface/arch-ia64/hvm/memmap.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/hvm/memmap.h 2008-05-19 00:34:28.333237766 +0300
+@@ -0,0 +1,88 @@
++/******************************************************************************
++ * memmap.h
++ *
++ * Copyright (c) 2008 Tristan Gingold <tgingold AT free fr>
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ */
++
++#ifndef __XEN_PUBLIC_HVM_MEMMAP_IA64_H__
++#define __XEN_PUBLIC_HVM_MEMMAP_IA64_H__
++
++#define MEM_G (1UL << 30)
++#define MEM_M (1UL << 20)
++#define MEM_K (1UL << 10)
++
++/* Guest physical address of IO ports space. */
++#define MMIO_START (3 * MEM_G)
++#define MMIO_SIZE (512 * MEM_M)
++
++#define VGA_IO_START 0xA0000UL
++#define VGA_IO_SIZE 0x20000
++
++#define LEGACY_IO_START (MMIO_START + MMIO_SIZE)
++#define LEGACY_IO_SIZE (64 * MEM_M)
++
++#define IO_PAGE_START (LEGACY_IO_START + LEGACY_IO_SIZE)
++#define IO_PAGE_SIZE XEN_PAGE_SIZE
++
++#define STORE_PAGE_START (IO_PAGE_START + IO_PAGE_SIZE)
++#define STORE_PAGE_SIZE XEN_PAGE_SIZE
++
++#define BUFFER_IO_PAGE_START (STORE_PAGE_START + STORE_PAGE_SIZE)
++#define BUFFER_IO_PAGE_SIZE XEN_PAGE_SIZE
++
++#define BUFFER_PIO_PAGE_START (BUFFER_IO_PAGE_START + BUFFER_IO_PAGE_SIZE)
++#define BUFFER_PIO_PAGE_SIZE XEN_PAGE_SIZE
++
++#define IO_SAPIC_START 0xfec00000UL
++#define IO_SAPIC_SIZE 0x100000
++
++#define PIB_START 0xfee00000UL
++#define PIB_SIZE 0x200000
++
++#define GFW_START (4 * MEM_G - 16 * MEM_M)
++#define GFW_SIZE (16 * MEM_M)
++
++/* domVTI */
++#define GPFN_FRAME_BUFFER 0x1 /* VGA framebuffer */
++#define GPFN_LOW_MMIO 0x2 /* Low MMIO range */
++#define GPFN_PIB 0x3 /* PIB base */
++#define GPFN_IOSAPIC 0x4 /* IOSAPIC base */
++#define GPFN_LEGACY_IO 0x5 /* Legacy I/O base */
++#define GPFN_HIGH_MMIO 0x6 /* High MMIO range */
++
++/* Nvram belongs to GFW memory space */
++#define NVRAM_SIZE (MEM_K * 64)
++#define NVRAM_START (GFW_START + 10 * MEM_M)
++
++#define NVRAM_VALID_SIG 0x4650494e45584948 /* "HIXENIPF" */
++struct nvram_save_addr {
++ unsigned long addr;
++ unsigned long signature;
++};
++
++#endif /* __XEN_PUBLIC_HVM_MEMMAP_IA64_H__ */
++
++/*
++ * Local variables:
++ * mode: C
++ * c-set-style: "BSD"
++ * c-basic-offset: 4
++ * tab-width: 4
++ * indent-tabs-mode: nil
++ * End:
++ */
+--- linux-2.6.18.8/include/xen/interface/arch-ia64/hvm/save.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/hvm/save.h 2008-05-19 00:34:28.333237766 +0300
@@ -0,0 +1,201 @@
+/******************************************************************************
+ * save_types.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-ia64.h linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h
---- linux-2.6.18/include/xen/interface/arch-ia64.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h 2007-12-23 11:15:46.475254497 +0100
-@@ -0,0 +1,666 @@
+--- linux-2.6.18.8/include/xen/interface/arch-ia64/sioemu.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64/sioemu.h 2008-05-19 00:34:28.333237766 +0300
+@@ -0,0 +1,89 @@
++/******************************************************************************
++ * sioemu.h
++ *
++ * Copyright (c) 2008 Tristan Gingold <tgingold@free.fr>
++ *
++ * This program is free software; you can redistribute it and/or modify
++ * it under the terms of the GNU General Public License as published by
++ * the Free Software Foundation; either version 2 of the License, or
++ * (at your option) any later version.
++ *
++ * This program is distributed in the hope that it will be useful,
++ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ * GNU General Public License for more details.
++ *
++ * You should have received a copy of the GNU General Public License
++ * along with this program; if not, write to the Free Software
++ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ */
++
++#ifndef __XEN_PUBLIC_IA64_SIOEMU_H__
++#define __XEN_PUBLIC_IA64_SIOEMU_H__
++
++/* SIOEMU specific hypercalls.
++ The numbers are the minor part of FW_HYPERCALL_SIOEMU. */
++
++/* Defines the callback entry point. r8=ip, r9=data.
++ Must be called per-vcpu. */
++#define SIOEMU_HYPERCALL_SET_CALLBACK 0x01
++
++/* Finish sioemu fw initialization and start firmware. r8=ip. */
++#define SIOEMU_HYPERCALL_START_FW 0x02
++
++/* Add IO pages in physmap. */
++#define SIOEMU_HYPERCALL_ADD_IO_PHYSMAP 0x03
++
++/* Get wallclock time. */
++#define SIOEMU_HYPERCALL_GET_TIME 0x04
++
++/* Flush cache. */
++#define SIOEMU_HYPERCALL_FLUSH_CACHE 0x07
++
++/* Get freq base. */
++#define SIOEMU_HYPERCALL_FREQ_BASE 0x08
++
++/* Return from callback. */
++#define SIOEMU_HYPERCALL_CALLBACK_RETURN 0x09
++
++/* Deliver an interrupt. */
++#define SIOEMU_HYPERCALL_DELIVER_INT 0x0a
++
++/* SIOEMU callback reason. */
++
++/* An event (from event channel) has to be delivered. */
++#define SIOEMU_CB_EVENT 0x00
++
++/* Emulate an IO access. */
++#define SIOEMU_CB_IO_EMULATE 0x01
++
++/* An IPI is sent to a dead vcpu. */
++#define SIOEMU_CB_WAKEUP_VCPU 0x02
++
++/* A SAL hypercall is executed. */
++#define SIOEMU_CB_SAL_ASSIST 0x03
++
++#ifndef __ASSEMBLY__
++struct sioemu_callback_info {
++ /* Saved registers. */
++ unsigned long ip;
++ unsigned long psr;
++ unsigned long ifs;
++ unsigned long nats;
++ unsigned long r8;
++ unsigned long r9;
++ unsigned long r10;
++ unsigned long r11;
++
++ /* Callback parameters. */
++ unsigned long cause;
++ unsigned long arg0;
++ unsigned long arg1;
++ unsigned long arg2;
++ unsigned long arg3;
++ unsigned long _pad2[2];
++ unsigned long r2;
++};
++#endif /* __ASSEMBLY__ */
++#endif /* __XEN_PUBLIC_IA64_SIOEMU_H__ */
+--- linux-2.6.18.8/include/xen/interface/arch-ia64.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h 2008-05-19 00:34:28.329237535 +0300
+@@ -0,0 +1,620 @@
+/******************************************************************************
+ * arch-ia64/hypervisor-if.h
+ *
+ *
+ */
+
++#include "xen.h"
++
+#ifndef __HYPERVISOR_IF_IA64_H__
+#define __HYPERVISOR_IF_IA64_H__
+
++#if !defined(__GNUC__) || defined(__STRICT_ANSI__)
++#error "Anonymous structs/unions are a GNU extension."
++#endif
++
+/* Structural guest handles introduced in 0x00030201. */
+#if __XEN_INTERFACE_VERSION__ >= 0x00030201
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
+ typedef struct { type *p; } __guest_handle_ ## name
+#else
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
+ typedef type * __guest_handle_ ## name
+#endif
+
++#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++ ___DEFINE_XEN_GUEST_HANDLE(name, type); \
++ ___DEFINE_XEN_GUEST_HANDLE(const_##name, const type)
++
+#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name)
+#define XEN_GUEST_HANDLE(name) __guest_handle_ ## name
+#define XEN_GUEST_HANDLE_64(name) XEN_GUEST_HANDLE(name)
+#endif
+
+#ifndef __ASSEMBLY__
-+/* Guest handles for primitive C types. */
-+__DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
-+__DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
-+__DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
-+__DEFINE_XEN_GUEST_HANDLE(u64, unsigned long);
-+DEFINE_XEN_GUEST_HANDLE(char);
-+DEFINE_XEN_GUEST_HANDLE(int);
-+DEFINE_XEN_GUEST_HANDLE(long);
-+DEFINE_XEN_GUEST_HANDLE(void);
-+
+typedef unsigned long xen_pfn_t;
-+DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
+#define PRI_xen_pfn "lx"
+#endif
+
+/* WARNING: before changing this, check that shared_info fits on a page */
+#define MAX_VIRT_CPUS 64
+
++/* IO ports location for PV. */
++#define IO_PORTS_PADDR 0x00000ffffc000000UL
++#define IO_PORTS_SIZE 0x0000000004000000UL
++
+#ifndef __ASSEMBLY__
+
+typedef unsigned long xen_ulong_t;
+
+#define INVALID_MFN (~0UL)
+
-+#define MEM_G (1UL << 30)
-+#define MEM_M (1UL << 20)
-+#define MEM_K (1UL << 10)
-+
-+/* Guest physical address of IO ports space. */
-+#define IO_PORTS_PADDR 0x00000ffffc000000UL
-+#define IO_PORTS_SIZE 0x0000000004000000UL
-+
-+#define MMIO_START (3 * MEM_G)
-+#define MMIO_SIZE (512 * MEM_M)
-+
-+#define VGA_IO_START 0xA0000UL
-+#define VGA_IO_SIZE 0x20000
-+
-+#define LEGACY_IO_START (MMIO_START + MMIO_SIZE)
-+#define LEGACY_IO_SIZE (64*MEM_M)
-+
-+#define IO_PAGE_START (LEGACY_IO_START + LEGACY_IO_SIZE)
-+#define IO_PAGE_SIZE XEN_PAGE_SIZE
-+
-+#define STORE_PAGE_START (IO_PAGE_START + IO_PAGE_SIZE)
-+#define STORE_PAGE_SIZE XEN_PAGE_SIZE
-+
-+#define BUFFER_IO_PAGE_START (STORE_PAGE_START + STORE_PAGE_SIZE)
-+#define BUFFER_IO_PAGE_SIZE XEN_PAGE_SIZE
-+
-+#define BUFFER_PIO_PAGE_START (BUFFER_IO_PAGE_START + BUFFER_IO_PAGE_SIZE)
-+#define BUFFER_PIO_PAGE_SIZE XEN_PAGE_SIZE
-+
-+#define IO_SAPIC_START 0xfec00000UL
-+#define IO_SAPIC_SIZE 0x100000
-+
-+#define PIB_START 0xfee00000UL
-+#define PIB_SIZE 0x200000
-+
-+#define GFW_START (4*MEM_G -16*MEM_M)
-+#define GFW_SIZE (16*MEM_M)
-+
-+/* Nvram belongs to GFW memory space */
-+#define NVRAM_SIZE (MEM_K * 64)
-+#define NVRAM_START (GFW_START + 10 * MEM_M)
-+
-+#define NVRAM_VALID_SIG 0x4650494e45584948 // "HIXENIPF"
-+struct nvram_save_addr {
-+ unsigned long addr;
-+ unsigned long signature;
-+};
-+
+struct pt_fpreg {
+ union {
+ unsigned long bits[2];
+/* Internal only: associated with PGC_allocated bit */
+#define _ASSIGN_pgc_allocated 3
+#define ASSIGN_pgc_allocated (1UL << _ASSIGN_pgc_allocated)
++/* Page is an IO page. */
++#define _ASSIGN_io 4
++#define ASSIGN_io (1UL << _ASSIGN_io)
+
+/* This structure has the same layout of struct ia64_boot_param, defined in
+ <asm/system.h>. It is redefined here to ease use. */
+ * This is useful in guests using region 7 for identity mapping
+ * like the linux kernel does.
+ */
-+#define XEN_IA64_OPTF_IDENT_MAP_REG7_BIT 0
-+#define XEN_IA64_OPTF_IDENT_MAP_REG7 \
-+ (1UL << XEN_IA64_OPTF_IDENT_MAP_REG7_BIT)
++#define XEN_IA64_OPTF_IDENT_MAP_REG7 1
+
+/* Identity mapping of region 4 addresses in HVM. */
-+#define XEN_IA64_OPTF_IDENT_MAP_REG4_BIT \
-+ (XEN_IA64_OPTF_IDENT_MAP_REG7_BIT + 1)
-+#define XEN_IA64_OPTF_IDENT_MAP_REG4 \
-+ (1UL << XEN_IA64_OPTF_IDENT_MAP_REG4_BIT)
++#define XEN_IA64_OPTF_IDENT_MAP_REG4 2
+
+/* Identity mapping of region 5 addresses in HVM. */
-+#define XEN_IA64_OPTF_IDENT_MAP_REG5_BIT \
-+ (XEN_IA64_OPTF_IDENT_MAP_REG7_BIT + 2)
-+#define XEN_IA64_OPTF_IDENT_MAP_REG5 \
-+ (1UL << XEN_IA64_OPTF_IDENT_MAP_REG5_BIT)
++#define XEN_IA64_OPTF_IDENT_MAP_REG5 3
+
+#define XEN_IA64_OPTF_IDENT_MAP_NOT_SET (0)
+
+#endif /* __ASSEMBLY__ */
+#endif /* XEN */
+
++#ifndef __ASSEMBLY__
++#include "arch-ia64/hvm/memmap.h"
++#endif
++
+#endif /* __HYPERVISOR_IF_IA64_H__ */
+
+/*
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-powerpc.h linux-2.6.18-xen.hg/include/xen/interface/arch-powerpc.h
---- linux-2.6.18/include/xen/interface/arch-powerpc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-powerpc.h 2007-12-23 11:15:46.478588006 +0100
-@@ -0,0 +1,125 @@
+--- linux-2.6.18.8/include/xen/interface/arch-powerpc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-powerpc.h 2008-05-19 00:34:28.333237766 +0300
+@@ -0,0 +1,120 @@
+/*
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * Authors: Hollis Blanchard <hollisb@us.ibm.com>
+ */
+
++#include "xen.h"
++
+#ifndef __XEN_PUBLIC_ARCH_PPC_64_H__
+#define __XEN_PUBLIC_ARCH_PPC_64_H__
+
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
+ typedef struct { \
+ int __pad[(sizeof (long long) - sizeof (void *)) / sizeof (int)]; \
+ type *p; \
+ } __attribute__((__aligned__(8))) __guest_handle_ ## name
+
++#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++ ___DEFINE_XEN_GUEST_HANDLE(name, type); \
++ ___DEFINE_XEN_GUEST_HANDLE(const_##name, const type)
+#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name)
+#define XEN_GUEST_HANDLE(name) __guest_handle_ ## name
+#define set_xen_guest_handle(hnd, val) \
+#endif
+
+#ifndef __ASSEMBLY__
-+/* Guest handles for primitive C types. */
-+__DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
-+__DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
-+__DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
-+DEFINE_XEN_GUEST_HANDLE(char);
-+DEFINE_XEN_GUEST_HANDLE(int);
-+DEFINE_XEN_GUEST_HANDLE(long);
-+DEFINE_XEN_GUEST_HANDLE(void);
-+
+typedef unsigned long long xen_pfn_t;
-+DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
+#define PRI_xen_pfn "llx"
+#endif
+
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86/hvm/save.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86/hvm/save.h
---- linux-2.6.18/include/xen/interface/arch-x86/hvm/save.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/hvm/save.h 2007-12-23 11:15:46.478588006 +0100
-@@ -0,0 +1,413 @@
+--- linux-2.6.18.8/include/xen/interface/arch-x86/cpuid.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/cpuid.h 2008-05-19 00:34:28.337237996 +0300
+@@ -0,0 +1,68 @@
++/******************************************************************************
++ * arch-x86/cpuid.h
++ *
++ * CPUID interface to Xen.
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ *
++ * Copyright (c) 2007 Citrix Systems, Inc.
++ *
++ * Authors:
++ * Keir Fraser <keir.fraser@citrix.com>
++ */
++
++#ifndef __XEN_PUBLIC_ARCH_X86_CPUID_H__
++#define __XEN_PUBLIC_ARCH_X86_CPUID_H__
++
++/* Xen identification leaves start at 0x40000000. */
++#define XEN_CPUID_FIRST_LEAF 0x40000000
++#define XEN_CPUID_LEAF(i) (XEN_CPUID_FIRST_LEAF + (i))
++
++/*
++ * Leaf 1 (0x40000000)
++ * EAX: Largest Xen-information leaf. All leaves up to an including @EAX
++ * are supported by the Xen host.
++ * EBX-EDX: "XenVMMXenVMM" signature, allowing positive identification
++ * of a Xen host.
++ */
++#define XEN_CPUID_SIGNATURE_EBX 0x566e6558 /* "XenV" */
++#define XEN_CPUID_SIGNATURE_ECX 0x65584d4d /* "MMXe" */
++#define XEN_CPUID_SIGNATURE_EDX 0x4d4d566e /* "nVMM" */
++
++/*
++ * Leaf 2 (0x40000001)
++ * EAX[31:16]: Xen major version.
++ * EAX[15: 0]: Xen minor version.
++ * EBX-EDX: Reserved (currently all zeroes).
++ */
++
++/*
++ * Leaf 3 (0x40000002)
++ * EAX: Number of hypercall transfer pages. This register is always guaranteed
++ * to specify one hypercall page.
++ * EBX: Base address of Xen-specific MSRs.
++ * ECX: Features 1. Unused bits are set to zero.
++ * EDX: Features 2. Unused bits are set to zero.
++ */
++
++/* Does the host support MMU_PT_UPDATE_PRESERVE_AD for this guest? */
++#define _XEN_CPUID_FEAT1_MMU_PT_UPDATE_PRESERVE_AD 0
++#define XEN_CPUID_FEAT1_MMU_PT_UPDATE_PRESERVE_AD (1u<<0)
++
++#endif /* __XEN_PUBLIC_ARCH_X86_CPUID_H__ */
+--- linux-2.6.18.8/include/xen/interface/arch-x86/hvm/save.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/hvm/save.h 2008-05-19 00:34:28.337237996 +0300
+@@ -0,0 +1,429 @@
+/*
+ * Structure definitions for HVM state that is held by Xen and must
+ * be saved along with the domain's memory and device-model state.
+DECLARE_HVM_SAVE_TYPE(LAPIC, 5, struct hvm_hw_lapic);
+
+struct hvm_hw_lapic_regs {
-+ /* A 4k page of register state */
-+ uint8_t data[0x400];
++ uint8_t data[1024];
+};
+
+DECLARE_HVM_SAVE_TYPE(LAPIC_REGS, 6, struct hvm_hw_lapic_regs);
+
+DECLARE_HVM_SAVE_TYPE(PMTIMER, 13, struct hvm_hw_pmtimer);
+
-+/*
-+ * Largest type-code in use
-+ */
-+#define HVM_SAVE_CODE_MAX 13
-+
-+#endif /* __XEN_PUBLIC_HVM_SAVE_X86_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86/xen.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen.h
---- linux-2.6.18/include/xen/interface/arch-x86/xen.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen.h 2007-12-23 11:15:46.478588006 +0100
-@@ -0,0 +1,205 @@
-+/******************************************************************************
-+ * arch-x86/xen.h
-+ *
-+ * Guest OS interface to x86 Xen.
-+ *
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this software and associated documentation files (the "Software"), to
-+ * deal in the Software without restriction, including without limitation the
-+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
-+ * sell copies of the Software, and to permit persons to whom the Software is
-+ * furnished to do so, subject to the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
-+ * DEALINGS IN THE SOFTWARE.
-+ *
-+ * Copyright (c) 2004-2006, K A Fraser
-+ */
-+
-+#ifndef __XEN_PUBLIC_ARCH_X86_XEN_H__
-+#define __XEN_PUBLIC_ARCH_X86_XEN_H__
-+
-+/* Structural guest handles introduced in 0x00030201. */
-+#if __XEN_INTERFACE_VERSION__ >= 0x00030201
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
-+ typedef struct { type *p; } __guest_handle_ ## name
-+#else
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
-+ typedef type * __guest_handle_ ## name
-+#endif
-+
-+#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name)
-+#define __XEN_GUEST_HANDLE(name) __guest_handle_ ## name
-+#define XEN_GUEST_HANDLE(name) __XEN_GUEST_HANDLE(name)
-+#define set_xen_guest_handle(hnd, val) do { (hnd).p = val; } while (0)
-+#ifdef __XEN_TOOLS__
-+#define get_xen_guest_handle(val, hnd) do { val = (hnd).p; } while (0)
-+#endif
-+
-+#if defined(__i386__)
-+#include "xen-x86_32.h"
-+#elif defined(__x86_64__)
-+#include "xen-x86_64.h"
-+#endif
-+
-+#ifndef __ASSEMBLY__
-+/* Guest handles for primitive C types. */
-+__DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
-+__DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
-+__DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
-+DEFINE_XEN_GUEST_HANDLE(char);
-+DEFINE_XEN_GUEST_HANDLE(int);
-+DEFINE_XEN_GUEST_HANDLE(long);
-+DEFINE_XEN_GUEST_HANDLE(void);
-+
-+typedef unsigned long xen_pfn_t;
-+DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
-+#define PRI_xen_pfn "lx"
-+#endif
-+
-+/*
-+ * SEGMENT DESCRIPTOR TABLES
-+ */
-+/*
-+ * A number of GDT entries are reserved by Xen. These are not situated at the
-+ * start of the GDT because some stupid OSes export hard-coded selector values
-+ * in their ABI. These hard-coded values are always near the start of the GDT,
-+ * so Xen places itself out of the way, at the far end of the GDT.
-+ */
-+#define FIRST_RESERVED_GDT_PAGE 14
-+#define FIRST_RESERVED_GDT_BYTE (FIRST_RESERVED_GDT_PAGE * 4096)
-+#define FIRST_RESERVED_GDT_ENTRY (FIRST_RESERVED_GDT_BYTE / 8)
-+
-+/* Maximum number of virtual CPUs in multi-processor guests. */
-+#define MAX_VIRT_CPUS 32
-+
-+#ifndef __ASSEMBLY__
-+
-+typedef unsigned long xen_ulong_t;
-+
+/*
-+ * Send an array of these to HYPERVISOR_set_trap_table().
-+ * The privilege level specifies which modes may enter a trap via a software
-+ * interrupt. On x86/64, since rings 1 and 2 are unavailable, we allocate
-+ * privilege levels as follows:
-+ * Level == 0: Noone may enter
-+ * Level == 1: Kernel may enter
-+ * Level == 2: Kernel may enter
-+ * Level == 3: Everyone may enter
++ * MTRR MSRs
+ */
-+#define TI_GET_DPL(_ti) ((_ti)->flags & 3)
-+#define TI_GET_IF(_ti) ((_ti)->flags & 4)
-+#define TI_SET_DPL(_ti,_dpl) ((_ti)->flags |= (_dpl))
-+#define TI_SET_IF(_ti,_if) ((_ti)->flags |= ((!!(_if))<<2))
-+struct trap_info {
-+ uint8_t vector; /* exception vector */
-+ uint8_t flags; /* 0-3: privilege level; 4: clear event enable? */
-+ uint16_t cs; /* code selector */
-+ unsigned long address; /* code offset */
-+};
-+typedef struct trap_info trap_info_t;
-+DEFINE_XEN_GUEST_HANDLE(trap_info_t);
-+
-+typedef uint64_t tsc_timestamp_t; /* RDTSC timestamp */
-+
-+/*
-+ * The following is all CPU context. Note that the fpu_ctxt block is filled
-+ * in by FXSAVE if the CPU has feature FXSR; otherwise FSAVE is used.
-+ */
-+struct vcpu_guest_context {
-+ /* FPU registers come first so they can be aligned for FXSAVE/FXRSTOR. */
-+ struct { char x[512]; } fpu_ctxt; /* User-level FPU registers */
-+#define VGCF_I387_VALID (1<<0)
-+#define VGCF_IN_KERNEL (1<<2)
-+#define _VGCF_i387_valid 0
-+#define VGCF_i387_valid (1<<_VGCF_i387_valid)
-+#define _VGCF_in_kernel 2
-+#define VGCF_in_kernel (1<<_VGCF_in_kernel)
-+#define _VGCF_failsafe_disables_events 3
-+#define VGCF_failsafe_disables_events (1<<_VGCF_failsafe_disables_events)
-+#define _VGCF_syscall_disables_events 4
-+#define VGCF_syscall_disables_events (1<<_VGCF_syscall_disables_events)
-+#define _VGCF_online 5
-+#define VGCF_online (1<<_VGCF_online)
-+ unsigned long flags; /* VGCF_* flags */
-+ struct cpu_user_regs user_regs; /* User-level CPU registers */
-+ struct trap_info trap_ctxt[256]; /* Virtual IDT */
-+ unsigned long ldt_base, ldt_ents; /* LDT (linear address, # ents) */
-+ unsigned long gdt_frames[16], gdt_ents; /* GDT (machine frames, # ents) */
-+ unsigned long kernel_ss, kernel_sp; /* Virtual TSS (only SS1/SP1) */
-+ /* NB. User pagetable on x86/64 is placed in ctrlreg[1]. */
-+ unsigned long ctrlreg[8]; /* CR0-CR7 (control registers) */
-+ unsigned long debugreg[8]; /* DB0-DB7 (debug registers) */
-+#ifdef __i386__
-+ unsigned long event_callback_cs; /* CS:EIP of event callback */
-+ unsigned long event_callback_eip;
-+ unsigned long failsafe_callback_cs; /* CS:EIP of failsafe callback */
-+ unsigned long failsafe_callback_eip;
-+#else
-+ unsigned long event_callback_eip;
-+ unsigned long failsafe_callback_eip;
-+#ifdef __XEN__
-+ union {
-+ unsigned long syscall_callback_eip;
-+ struct {
-+ unsigned int event_callback_cs; /* compat CS of event cb */
-+ unsigned int failsafe_callback_cs; /* compat CS of failsafe cb */
-+ };
-+ };
-+#else
-+ unsigned long syscall_callback_eip;
-+#endif
-+#endif
-+ unsigned long vm_assist; /* VMASST_TYPE_* bitmap */
-+#ifdef __x86_64__
-+ /* Segment base addresses. */
-+ uint64_t fs_base;
-+ uint64_t gs_base_kernel;
-+ uint64_t gs_base_user;
-+#endif
-+};
-+typedef struct vcpu_guest_context vcpu_guest_context_t;
-+DEFINE_XEN_GUEST_HANDLE(vcpu_guest_context_t);
+
-+struct arch_shared_info {
-+ unsigned long max_pfn; /* max pfn that appears in table */
-+ /* Frame containing list of mfns containing list of mfns containing p2m. */
-+ xen_pfn_t pfn_to_mfn_frame_list_list;
-+ unsigned long nmi_reason;
-+ uint64_t pad[32];
++struct hvm_hw_mtrr {
++#define MTRR_VCNT 8
++#define NUM_FIXED_MSR 11
++ uint64_t msr_pat_cr;
++ /* mtrr physbase & physmask msr pair*/
++ uint64_t msr_mtrr_var[MTRR_VCNT*2];
++ uint64_t msr_mtrr_fixed[NUM_FIXED_MSR];
++ uint64_t msr_mtrr_cap;
++ uint64_t msr_mtrr_def_type;
+};
-+typedef struct arch_shared_info arch_shared_info_t;
+
-+#endif /* !__ASSEMBLY__ */
++DECLARE_HVM_SAVE_TYPE(MTRR, 14, struct hvm_hw_mtrr);
+
-+/*
-+ * Prefix forces emulation of some non-trapping instructions.
-+ * Currently only CPUID.
++/*
++ * Largest type-code in use
+ */
-+#ifdef __ASSEMBLY__
-+#define XEN_EMULATE_PREFIX .byte 0x0f,0x0b,0x78,0x65,0x6e ;
-+#define XEN_CPUID XEN_EMULATE_PREFIX cpuid
-+#else
-+#define XEN_EMULATE_PREFIX ".byte 0x0f,0x0b,0x78,0x65,0x6e ; "
-+#define XEN_CPUID XEN_EMULATE_PREFIX "cpuid"
-+#endif
-+
-+#endif /* __XEN_PUBLIC_ARCH_X86_XEN_H__ */
++#define HVM_SAVE_CODE_MAX 14
+
-+/*
-+ * Local variables:
-+ * mode: C
-+ * c-set-style: "BSD"
-+ * c-basic-offset: 4
-+ * tab-width: 4
-+ * indent-tabs-mode: nil
-+ * End:
-+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86/xen-x86_32.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_32.h
---- linux-2.6.18/include/xen/interface/arch-x86/xen-x86_32.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_32.h 2007-12-23 11:15:46.478588006 +0100
-@@ -0,0 +1,185 @@
++#endif /* __XEN_PUBLIC_HVM_SAVE_X86_H__ */
+--- linux-2.6.18.8/include/xen/interface/arch-x86/xen-x86_32.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_32.h 2008-05-19 00:34:28.337237996 +0300
+@@ -0,0 +1,180 @@
+/******************************************************************************
+ * xen-x86_32.h
+ *
+#define MACH2PHYS_VIRT_END_PAE \
+ mk_unsigned_long(__MACH2PHYS_VIRT_END_PAE)
+
++/* Non-PAE bounds are obsolete. */
+#define __HYPERVISOR_VIRT_START_NONPAE 0xFC000000
+#define __MACH2PHYS_VIRT_START_NONPAE 0xFC000000
+#define __MACH2PHYS_VIRT_END_NONPAE 0xFC400000
+#define MACH2PHYS_VIRT_END_NONPAE \
+ mk_unsigned_long(__MACH2PHYS_VIRT_END_NONPAE)
+
-+#ifdef CONFIG_X86_PAE
+#define __HYPERVISOR_VIRT_START __HYPERVISOR_VIRT_START_PAE
+#define __MACH2PHYS_VIRT_START __MACH2PHYS_VIRT_START_PAE
+#define __MACH2PHYS_VIRT_END __MACH2PHYS_VIRT_END_PAE
-+#else
-+#define __HYPERVISOR_VIRT_START __HYPERVISOR_VIRT_START_NONPAE
-+#define __MACH2PHYS_VIRT_START __MACH2PHYS_VIRT_START_NONPAE
-+#define __MACH2PHYS_VIRT_END __MACH2PHYS_VIRT_END_NONPAE
-+#endif
+
+#ifndef HYPERVISOR_VIRT_START
+#define HYPERVISOR_VIRT_START mk_unsigned_long(__HYPERVISOR_VIRT_START)
+
+/* 32-/64-bit invariability for control interfaces (domctl/sysctl). */
+#if defined(__XEN__) || defined(__XEN_TOOLS__)
-+#undef __DEFINE_XEN_GUEST_HANDLE
-+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++#undef ___DEFINE_XEN_GUEST_HANDLE
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
+ typedef struct { type *p; } \
+ __guest_handle_ ## name; \
+ typedef struct { union { type *p; uint64_aligned_t q; }; } \
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86/xen-x86_64.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_64.h
---- linux-2.6.18/include/xen/interface/arch-x86/xen-x86_64.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_64.h 2007-12-23 11:15:46.478588006 +0100
-@@ -0,0 +1,211 @@
+--- linux-2.6.18.8/include/xen/interface/arch-x86/xen-x86_64.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen-x86_64.h 2008-05-19 00:34:28.337237996 +0300
+@@ -0,0 +1,212 @@
+/******************************************************************************
+ * xen-x86_64.h
+ *
+#define machine_to_phys_mapping ((unsigned long *)HYPERVISOR_VIRT_START)
+#endif
+
-+#ifndef __ASSEMBLY__
-+
+/*
+ * int HYPERVISOR_set_segment_base(unsigned int which, unsigned long base)
+ * @which == SEGBASE_* ; @base == 64-bit base address
+#define _VGCF_in_syscall 8
+#define VGCF_in_syscall (1<<_VGCF_in_syscall)
+#define VGCF_IN_SYSCALL VGCF_in_syscall
++
++#ifndef __ASSEMBLY__
++
+struct iret_context {
+ /* Top of stack (%rsp at point of hypercall). */
+ uint64_t rax, r11, rcx, flags, rip, cs, rflags, rsp, ss;
+ /* Bottom of iret stack frame. */
+};
+
-+#ifdef __GNUC__
++#if defined(__GNUC__) && !defined(__STRICT_ANSI__)
+/* Anonymous union includes both 32- and 64-bit names (e.g., eax/rax). */
+#define __DECL_REG(name) union { \
+ uint64_t r ## name, e ## name; \
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86_32.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86_32.h
---- linux-2.6.18/include/xen/interface/arch-x86_32.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86_32.h 2007-12-23 11:15:46.478588006 +0100
+--- linux-2.6.18.8/include/xen/interface/arch-x86/xen.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86/xen.h 2008-05-19 00:34:28.341238227 +0300
+@@ -0,0 +1,200 @@
++/******************************************************************************
++ * arch-x86/xen.h
++ *
++ * Guest OS interface to x86 Xen.
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ *
++ * Copyright (c) 2004-2006, K A Fraser
++ */
++
++#include "../xen.h"
++
++#ifndef __XEN_PUBLIC_ARCH_X86_XEN_H__
++#define __XEN_PUBLIC_ARCH_X86_XEN_H__
++
++/* Structural guest handles introduced in 0x00030201. */
++#if __XEN_INTERFACE_VERSION__ >= 0x00030201
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
++ typedef struct { type *p; } __guest_handle_ ## name
++#else
++#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
++ typedef type * __guest_handle_ ## name
++#endif
++
++#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
++ ___DEFINE_XEN_GUEST_HANDLE(name, type); \
++ ___DEFINE_XEN_GUEST_HANDLE(const_##name, const type)
++#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name)
++#define __XEN_GUEST_HANDLE(name) __guest_handle_ ## name
++#define XEN_GUEST_HANDLE(name) __XEN_GUEST_HANDLE(name)
++#define set_xen_guest_handle(hnd, val) do { (hnd).p = val; } while (0)
++#ifdef __XEN_TOOLS__
++#define get_xen_guest_handle(val, hnd) do { val = (hnd).p; } while (0)
++#endif
++
++#if defined(__i386__)
++#include "xen-x86_32.h"
++#elif defined(__x86_64__)
++#include "xen-x86_64.h"
++#endif
++
++#ifndef __ASSEMBLY__
++typedef unsigned long xen_pfn_t;
++#define PRI_xen_pfn "lx"
++#endif
++
++/*
++ * SEGMENT DESCRIPTOR TABLES
++ */
++/*
++ * A number of GDT entries are reserved by Xen. These are not situated at the
++ * start of the GDT because some stupid OSes export hard-coded selector values
++ * in their ABI. These hard-coded values are always near the start of the GDT,
++ * so Xen places itself out of the way, at the far end of the GDT.
++ */
++#define FIRST_RESERVED_GDT_PAGE 14
++#define FIRST_RESERVED_GDT_BYTE (FIRST_RESERVED_GDT_PAGE * 4096)
++#define FIRST_RESERVED_GDT_ENTRY (FIRST_RESERVED_GDT_BYTE / 8)
++
++/* Maximum number of virtual CPUs in multi-processor guests. */
++#define MAX_VIRT_CPUS 32
++
++#ifndef __ASSEMBLY__
++
++typedef unsigned long xen_ulong_t;
++
++/*
++ * Send an array of these to HYPERVISOR_set_trap_table().
++ * The privilege level specifies which modes may enter a trap via a software
++ * interrupt. On x86/64, since rings 1 and 2 are unavailable, we allocate
++ * privilege levels as follows:
++ * Level == 0: Noone may enter
++ * Level == 1: Kernel may enter
++ * Level == 2: Kernel may enter
++ * Level == 3: Everyone may enter
++ */
++#define TI_GET_DPL(_ti) ((_ti)->flags & 3)
++#define TI_GET_IF(_ti) ((_ti)->flags & 4)
++#define TI_SET_DPL(_ti,_dpl) ((_ti)->flags |= (_dpl))
++#define TI_SET_IF(_ti,_if) ((_ti)->flags |= ((!!(_if))<<2))
++struct trap_info {
++ uint8_t vector; /* exception vector */
++ uint8_t flags; /* 0-3: privilege level; 4: clear event enable? */
++ uint16_t cs; /* code selector */
++ unsigned long address; /* code offset */
++};
++typedef struct trap_info trap_info_t;
++DEFINE_XEN_GUEST_HANDLE(trap_info_t);
++
++typedef uint64_t tsc_timestamp_t; /* RDTSC timestamp */
++
++/*
++ * The following is all CPU context. Note that the fpu_ctxt block is filled
++ * in by FXSAVE if the CPU has feature FXSR; otherwise FSAVE is used.
++ */
++struct vcpu_guest_context {
++ /* FPU registers come first so they can be aligned for FXSAVE/FXRSTOR. */
++ struct { char x[512]; } fpu_ctxt; /* User-level FPU registers */
++#define VGCF_I387_VALID (1<<0)
++#define VGCF_IN_KERNEL (1<<2)
++#define _VGCF_i387_valid 0
++#define VGCF_i387_valid (1<<_VGCF_i387_valid)
++#define _VGCF_in_kernel 2
++#define VGCF_in_kernel (1<<_VGCF_in_kernel)
++#define _VGCF_failsafe_disables_events 3
++#define VGCF_failsafe_disables_events (1<<_VGCF_failsafe_disables_events)
++#define _VGCF_syscall_disables_events 4
++#define VGCF_syscall_disables_events (1<<_VGCF_syscall_disables_events)
++#define _VGCF_online 5
++#define VGCF_online (1<<_VGCF_online)
++ unsigned long flags; /* VGCF_* flags */
++ struct cpu_user_regs user_regs; /* User-level CPU registers */
++ struct trap_info trap_ctxt[256]; /* Virtual IDT */
++ unsigned long ldt_base, ldt_ents; /* LDT (linear address, # ents) */
++ unsigned long gdt_frames[16], gdt_ents; /* GDT (machine frames, # ents) */
++ unsigned long kernel_ss, kernel_sp; /* Virtual TSS (only SS1/SP1) */
++ /* NB. User pagetable on x86/64 is placed in ctrlreg[1]. */
++ unsigned long ctrlreg[8]; /* CR0-CR7 (control registers) */
++ unsigned long debugreg[8]; /* DB0-DB7 (debug registers) */
++#ifdef __i386__
++ unsigned long event_callback_cs; /* CS:EIP of event callback */
++ unsigned long event_callback_eip;
++ unsigned long failsafe_callback_cs; /* CS:EIP of failsafe callback */
++ unsigned long failsafe_callback_eip;
++#else
++ unsigned long event_callback_eip;
++ unsigned long failsafe_callback_eip;
++#ifdef __XEN__
++ union {
++ unsigned long syscall_callback_eip;
++ struct {
++ unsigned int event_callback_cs; /* compat CS of event cb */
++ unsigned int failsafe_callback_cs; /* compat CS of failsafe cb */
++ };
++ };
++#else
++ unsigned long syscall_callback_eip;
++#endif
++#endif
++ unsigned long vm_assist; /* VMASST_TYPE_* bitmap */
++#ifdef __x86_64__
++ /* Segment base addresses. */
++ uint64_t fs_base;
++ uint64_t gs_base_kernel;
++ uint64_t gs_base_user;
++#endif
++};
++typedef struct vcpu_guest_context vcpu_guest_context_t;
++DEFINE_XEN_GUEST_HANDLE(vcpu_guest_context_t);
++
++struct arch_shared_info {
++ unsigned long max_pfn; /* max pfn that appears in table */
++ /* Frame containing list of mfns containing list of mfns containing p2m. */
++ xen_pfn_t pfn_to_mfn_frame_list_list;
++ unsigned long nmi_reason;
++ uint64_t pad[32];
++};
++typedef struct arch_shared_info arch_shared_info_t;
++
++#endif /* !__ASSEMBLY__ */
++
++/*
++ * Prefix forces emulation of some non-trapping instructions.
++ * Currently only CPUID.
++ */
++#ifdef __ASSEMBLY__
++#define XEN_EMULATE_PREFIX .byte 0x0f,0x0b,0x78,0x65,0x6e ;
++#define XEN_CPUID XEN_EMULATE_PREFIX cpuid
++#else
++#define XEN_EMULATE_PREFIX ".byte 0x0f,0x0b,0x78,0x65,0x6e ; "
++#define XEN_CPUID XEN_EMULATE_PREFIX "cpuid"
++#endif
++
++#endif /* __XEN_PUBLIC_ARCH_X86_XEN_H__ */
++
++/*
++ * Local variables:
++ * mode: C
++ * c-set-style: "BSD"
++ * c-basic-offset: 4
++ * tab-width: 4
++ * indent-tabs-mode: nil
++ * End:
++ */
+--- linux-2.6.18.8/include/xen/interface/arch-x86_32.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86_32.h 2008-05-19 00:34:28.341238227 +0300
@@ -0,0 +1,27 @@
+/******************************************************************************
+ * arch-x86_32.h
+ */
+
+#include "arch-x86/xen.h"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/arch-x86_64.h linux-2.6.18-xen.hg/include/xen/interface/arch-x86_64.h
---- linux-2.6.18/include/xen/interface/arch-x86_64.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86_64.h 2007-12-23 11:15:46.481921515 +0100
+--- linux-2.6.18.8/include/xen/interface/arch-x86_64.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/arch-x86_64.h 2008-05-19 00:34:28.341238227 +0300
@@ -0,0 +1,27 @@
+/******************************************************************************
+ * arch-x86_64.h
+ */
+
+#include "arch-x86/xen.h"
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/callback.h linux-2.6.18-xen.hg/include/xen/interface/callback.h
---- linux-2.6.18/include/xen/interface/callback.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/callback.h 2007-12-23 11:15:46.481921515 +0100
+--- linux-2.6.18.8/include/xen/interface/callback.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/callback.h 2008-05-19 00:34:28.341238227 +0300
@@ -0,0 +1,121 @@
+/******************************************************************************
+ * callback.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/COPYING linux-2.6.18-xen.hg/include/xen/interface/COPYING
---- linux-2.6.18/include/xen/interface/COPYING 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/COPYING 2007-12-23 11:15:46.471920991 +0100
-@@ -0,0 +1,38 @@
-+XEN NOTICE
-+==========
-+
-+This copyright applies to all files within this subdirectory and its
-+subdirectories:
-+ include/public/*.h
-+ include/public/hvm/*.h
-+ include/public/io/*.h
-+
-+The intention is that these files can be freely copied into the source
-+tree of an operating system when porting that OS to run on Xen. Doing
-+so does *not* cause the OS to become subject to the terms of the GPL.
-+
-+All other files in the Xen source distribution are covered by version
-+2 of the GNU General Public License except where explicitly stated
-+otherwise within individual source files.
-+
-+ -- Keir Fraser (on behalf of the Xen team)
-+
-+=====================================================================
-+
-+Permission is hereby granted, free of charge, to any person obtaining a copy
-+of this software and associated documentation files (the "Software"), to
-+deal in the Software without restriction, including without limitation the
-+rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
-+sell copies of the Software, and to permit persons to whom the Software is
-+furnished to do so, subject to the following conditions:
-+
-+The above copyright notice and this permission notice shall be included in
-+all copies or substantial portions of the Software.
-+
-+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
-+DEALINGS IN THE SOFTWARE.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/dom0_ops.h linux-2.6.18-xen.hg/include/xen/interface/dom0_ops.h
---- linux-2.6.18/include/xen/interface/dom0_ops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/dom0_ops.h 2007-12-23 11:15:46.481921515 +0100
+--- linux-2.6.18.8/include/xen/interface/dom0_ops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/dom0_ops.h 2008-05-19 00:34:28.341238227 +0300
@@ -0,0 +1,120 @@
+/******************************************************************************
+ * dom0_ops.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/domctl.h linux-2.6.18-xen.hg/include/xen/interface/domctl.h
---- linux-2.6.18/include/xen/interface/domctl.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/domctl.h 2007-12-23 11:15:46.481921515 +0100
-@@ -0,0 +1,606 @@
+--- linux-2.6.18.8/include/xen/interface/domctl.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/domctl.h 2008-05-19 00:34:28.345238457 +0300
+@@ -0,0 +1,649 @@
+/******************************************************************************
+ * domctl.h
+ *
+#define XEN_DOMCTL_INTERFACE_VERSION 0x00000005
+
+struct xenctl_cpumap {
-+ XEN_GUEST_HANDLE_64(uint8_t) bitmap;
++ XEN_GUEST_HANDLE_64(uint8) bitmap;
+ uint32_t nr_cpus;
+};
+
+ /* Is this an HVM guest (as opposed to a PV guest)? */
+#define _XEN_DOMCTL_CDF_hvm_guest 0
+#define XEN_DOMCTL_CDF_hvm_guest (1U<<_XEN_DOMCTL_CDF_hvm_guest)
++ /* Use hardware-assisted paging if available? */
++#define _XEN_DOMCTL_CDF_hap 1
++#define XEN_DOMCTL_CDF_hap (1U<<_XEN_DOMCTL_CDF_hap)
+ uint32_t flags;
+};
+typedef struct xen_domctl_createdomain xen_domctl_createdomain_t;
+ uint64_aligned_t max_pfns;
+ /* Start index in guest's page list. */
+ uint64_aligned_t start_pfn;
-+ XEN_GUEST_HANDLE_64(uint64_t) buffer;
++ XEN_GUEST_HANDLE_64(uint64) buffer;
+ /* OUT variables. */
+ uint64_aligned_t num_pfns;
+};
+ /* IN variables. */
+ uint64_aligned_t num;
+ /* IN/OUT variables. */
-+ XEN_GUEST_HANDLE_64(uint32_t) array;
++ XEN_GUEST_HANDLE_64(uint32) array;
+};
+typedef struct xen_domctl_getpageframeinfo2 xen_domctl_getpageframeinfo2_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_getpageframeinfo2_t);
+ uint32_t mb; /* Shadow memory allocation in MB */
+
+ /* OP_PEEK / OP_CLEAN */
-+ XEN_GUEST_HANDLE_64(uint8_t) dirty_bitmap;
++ XEN_GUEST_HANDLE_64(uint8) dirty_bitmap;
+ uint64_aligned_t pages; /* Size of buffer. Updated with actual size. */
+ struct xen_domctl_shadow_op_stats stats;
+};
+#define XEN_DOMAINSETUP_hvm_guest (1UL<<_XEN_DOMAINSETUP_hvm_guest)
+#define _XEN_DOMAINSETUP_query 1 /* Get parameters (for save) */
+#define XEN_DOMAINSETUP_query (1UL<<_XEN_DOMAINSETUP_query)
++#define _XEN_DOMAINSETUP_sioemu_guest 2
++#define XEN_DOMAINSETUP_sioemu_guest (1UL<<_XEN_DOMAINSETUP_sioemu_guest)
+typedef struct xen_domctl_arch_setup {
+ uint64_aligned_t flags; /* XEN_DOMAINSETUP_* */
+#ifdef __ia64__
+#define XEN_DOMCTL_sethvmcontext 34
+typedef struct xen_domctl_hvmcontext {
+ uint32_t size; /* IN/OUT: size of buffer / bytes filled */
-+ XEN_GUEST_HANDLE_64(uint8_t) buffer; /* IN/OUT: data, or call
-+ * gethvmcontext with NULL
-+ * buffer to get size
-+ * req'd */
++ XEN_GUEST_HANDLE_64(uint8) buffer; /* IN/OUT: data, or call
++ * gethvmcontext with NULL
++ * buffer to get size req'd */
+} xen_domctl_hvmcontext_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_hvmcontext_t);
+
+
+
+/* Assign PCI device to HVM guest. Sets up IOMMU structures. */
-+#define XEN_DOMCTL_assign_device 37
-+#define DPCI_ADD_MAPPING 1
-+#define DPCI_REMOVE_MAPPING 0
++#define XEN_DOMCTL_assign_device 37
++#define XEN_DOMCTL_test_assign_device 45
++#define XEN_DOMCTL_deassign_device 47
+struct xen_domctl_assign_device {
+ uint32_t machine_bdf; /* machine PCI ID of assigned device */
+};
+
+/* Pass-through interrupts: bind real irq -> hvm devfn. */
+#define XEN_DOMCTL_bind_pt_irq 38
++#define XEN_DOMCTL_unbind_pt_irq 48
+typedef enum pt_irq_type_e {
+ PT_IRQ_TYPE_PCI,
-+ PT_IRQ_TYPE_ISA
++ PT_IRQ_TYPE_ISA,
++ PT_IRQ_TYPE_MSI,
+} pt_irq_type_t;
+struct xen_domctl_bind_pt_irq {
+ uint32_t machine_irq;
+ uint8_t device;
+ uint8_t intx;
+ } pci;
++ struct {
++ uint8_t gvec;
++ uint32_t gflags;
++ } msi;
+ } u;
+};
+typedef struct xen_domctl_bind_pt_irq xen_domctl_bind_pt_irq_t;
+
+/* Bind machine I/O address range -> HVM address range. */
+#define XEN_DOMCTL_memory_mapping 39
++#define DPCI_ADD_MAPPING 1
++#define DPCI_REMOVE_MAPPING 0
+struct xen_domctl_memory_mapping {
+ uint64_aligned_t first_gfn; /* first page (hvm guest phys page) in range */
+ uint64_aligned_t first_mfn; /* first page (machine page) in range */
+ */
+#define XEN_DOMCTL_set_opt_feature 44
+struct xen_domctl_set_opt_feature {
-+#ifdef __ia64__
++#if defined(__ia64__)
+ struct xen_ia64_opt_feature optf;
++#else
++ /* Make struct non-empty: do not depend on this field name! */
++ uint64_t dummy;
+#endif
+};
+typedef struct xen_domctl_set_opt_feature xen_domctl_set_opt_feature_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_opt_feature_t);
+
++/*
++ * Set the target domain for a domain
++ */
++#define XEN_DOMCTL_set_target 46
++struct xen_domctl_set_target {
++ domid_t target;
++};
++typedef struct xen_domctl_set_target xen_domctl_set_target_t;
++DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_target_t);
++
++#if defined(__i386__) || defined(__x86_64__)
++# define XEN_CPUID_INPUT_UNUSED 0xFFFFFFFF
++# define XEN_DOMCTL_set_cpuid 49
++struct xen_domctl_cpuid {
++ unsigned int input[2];
++ unsigned int eax;
++ unsigned int ebx;
++ unsigned int ecx;
++ unsigned int edx;
++};
++typedef struct xen_domctl_cpuid xen_domctl_cpuid_t;
++DEFINE_XEN_GUEST_HANDLE(xen_domctl_cpuid_t);
++#endif
++
+struct xen_domctl {
+ uint32_t cmd;
+ uint32_t interface_version; /* XEN_DOMCTL_INTERFACE_VERSION */
+ struct xen_domctl_pin_mem_cacheattr pin_mem_cacheattr;
+ struct xen_domctl_ext_vcpucontext ext_vcpucontext;
+ struct xen_domctl_set_opt_feature set_opt_feature;
++ struct xen_domctl_set_target set_target;
++#if defined(__i386__) || defined(__x86_64__)
++ struct xen_domctl_cpuid cpuid;
++#endif
+ uint8_t pad[128];
+ } u;
+};
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/elfnote.h linux-2.6.18-xen.hg/include/xen/interface/elfnote.h
---- linux-2.6.18/include/xen/interface/elfnote.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/elfnote.h 2007-12-23 11:15:46.481921515 +0100
+--- linux-2.6.18.8/include/xen/interface/elfnote.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/elfnote.h 2008-05-19 00:34:28.345238457 +0300
@@ -0,0 +1,233 @@
+/******************************************************************************
+ * elfnote.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/elfstructs.h linux-2.6.18-xen.hg/include/xen/interface/elfstructs.h
---- linux-2.6.18/include/xen/interface/elfstructs.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/elfstructs.h 2007-12-23 11:15:46.481921515 +0100
+--- linux-2.6.18.8/include/xen/interface/elfstructs.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/elfstructs.h 2008-05-19 00:34:28.345238457 +0300
@@ -0,0 +1,527 @@
+#ifndef __XEN_PUBLIC_ELFSTRUCTS_H__
+#define __XEN_PUBLIC_ELFSTRUCTS_H__ 1
+#endif
+
+#endif /* __XEN_PUBLIC_ELFSTRUCTS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/event_channel.h linux-2.6.18-xen.hg/include/xen/interface/event_channel.h
---- linux-2.6.18/include/xen/interface/event_channel.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/event_channel.h 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/event_channel.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/event_channel.h 2008-05-19 00:34:28.349238688 +0300
@@ -0,0 +1,264 @@
+/******************************************************************************
+ * event_channel.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/features.h linux-2.6.18-xen.hg/include/xen/interface/features.h
---- linux-2.6.18/include/xen/interface/features.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/features.h 2007-12-23 11:15:46.485255025 +0100
-@@ -0,0 +1,71 @@
+--- linux-2.6.18.8/include/xen/interface/features.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/features.h 2008-05-19 00:34:28.349238688 +0300
+@@ -0,0 +1,74 @@
+/******************************************************************************
+ * features.h
+ *
+ */
+#define XENFEAT_pae_pgdir_above_4gb 4
+
++/* x86: Does this Xen host support the MMU_PT_UPDATE_PRESERVE_AD hypercall? */
++#define XENFEAT_mmu_pt_update_preserve_ad 5
++
+#define XENFEAT_NR_SUBMAPS 1
+
+#endif /* __XEN_PUBLIC_FEATURES_H__ */
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/foreign/Makefile linux-2.6.18-xen.hg/include/xen/interface/foreign/Makefile
---- linux-2.6.18/include/xen/interface/foreign/Makefile 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/foreign/Makefile 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/foreign/Makefile 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/foreign/Makefile 2008-05-19 00:34:28.349238688 +0300
@@ -0,0 +1,37 @@
+XEN_ROOT=../../../..
+include $(XEN_ROOT)/Config.mk
+
+checker.c: $(scripts)
+ python mkchecker.py $(XEN_TARGET_ARCH) $@ $(architectures)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/foreign/mkchecker.py linux-2.6.18-xen.hg/include/xen/interface/foreign/mkchecker.py
---- linux-2.6.18/include/xen/interface/foreign/mkchecker.py 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/foreign/mkchecker.py 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/foreign/mkchecker.py 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/foreign/mkchecker.py 2008-05-19 00:34:28.349238688 +0300
@@ -0,0 +1,58 @@
+#!/usr/bin/python
+
+
+f.close();
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/foreign/mkheader.py linux-2.6.18-xen.hg/include/xen/interface/foreign/mkheader.py
---- linux-2.6.18/include/xen/interface/foreign/mkheader.py 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/foreign/mkheader.py 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/foreign/mkheader.py 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/foreign/mkheader.py 2008-05-19 00:34:28.393241224 +0300
@@ -0,0 +1,167 @@
+#!/usr/bin/python
+
+f.write(output);
+f.close;
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/foreign/reference.size linux-2.6.18-xen.hg/include/xen/interface/foreign/reference.size
---- linux-2.6.18/include/xen/interface/foreign/reference.size 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/foreign/reference.size 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/foreign/reference.size 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/foreign/reference.size 2008-05-19 00:34:28.393241224 +0300
@@ -0,0 +1,18 @@
+
+structs | x86_32 x86_64 ia64
+arch_shared_info | 268 280 272
+shared_info | 2584 3368 4384
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/foreign/structs.py linux-2.6.18-xen.hg/include/xen/interface/foreign/structs.py
---- linux-2.6.18/include/xen/interface/foreign/structs.py 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/foreign/structs.py 2007-12-23 11:15:46.485255025 +0100
+--- linux-2.6.18.8/include/xen/interface/foreign/structs.py 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/foreign/structs.py 2008-05-19 00:34:28.393241224 +0300
@@ -0,0 +1,58 @@
+# configuration: what needs translation
+
+ "MAX_VIRT_CPUS",
+ "MAX_GUEST_CMDLINE" ];
+
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/grant_table.h linux-2.6.18-xen.hg/include/xen/interface/grant_table.h
---- linux-2.6.18/include/xen/interface/grant_table.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/grant_table.h 2007-12-23 11:15:46.488588533 +0100
-@@ -0,0 +1,429 @@
+--- linux-2.6.18.8/include/xen/interface/grant_table.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/grant_table.h 2008-05-19 00:34:28.393241224 +0300
+@@ -0,0 +1,431 @@
+/******************************************************************************
+ * grant_table.h
+ *
+#define GNTST_no_device_space (-7) /* Out of space in I/O MMU. */
+#define GNTST_permission_denied (-8) /* Not enough privilege for operation. */
+#define GNTST_bad_page (-9) /* Specified page was invalid for op. */
-+#define GNTST_bad_copy_arg (-10) /* copy arguments cross page boundary */
++#define GNTST_bad_copy_arg (-10) /* copy arguments cross page boundary. */
++#define GNTST_address_too_big (-11) /* transfer page address too large. */
+
+#define GNTTABOP_error_msgs { \
+ "okay", \
+ "no spare translation slot in the I/O MMU", \
+ "permission denied", \
+ "bad page", \
-+ "copy arguments cross page boundary" \
++ "copy arguments cross page boundary", \
++ "page address size too large" \
+}
+
+#endif /* __XEN_PUBLIC_GRANT_TABLE_H__ */
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/e820.h linux-2.6.18-xen.hg/include/xen/interface/hvm/e820.h
---- linux-2.6.18/include/xen/interface/hvm/e820.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/e820.h 2007-12-23 11:15:46.488588533 +0100
+--- linux-2.6.18.8/include/xen/interface/hvm/e820.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/e820.h 2008-05-19 00:34:28.397241455 +0300
@@ -0,0 +1,34 @@
+
+/*
+#define HVM_BELOW_4G_MMIO_LENGTH ((1ULL << 32) - HVM_BELOW_4G_MMIO_START)
+
+#endif /* __XEN_PUBLIC_HVM_E820_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/hvm_info_table.h linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_info_table.h
---- linux-2.6.18/include/xen/interface/hvm/hvm_info_table.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_info_table.h 2007-12-23 11:15:46.488588533 +0100
+--- linux-2.6.18.8/include/xen/interface/hvm/hvm_info_table.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_info_table.h 2008-05-19 00:34:28.397241455 +0300
@@ -0,0 +1,41 @@
+/******************************************************************************
+ * hvm/hvm_info_table.h
+};
+
+#endif /* __XEN_PUBLIC_HVM_HVM_INFO_TABLE_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/hvm_op.h linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_op.h
---- linux-2.6.18/include/xen/interface/hvm/hvm_op.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_op.h 2007-12-23 11:15:46.488588533 +0100
-@@ -0,0 +1,76 @@
+--- linux-2.6.18.8/include/xen/interface/hvm/hvm_op.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/hvm_op.h 2008-05-19 00:34:28.397241455 +0300
+@@ -0,0 +1,97 @@
+/*
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+/* Flushes all VCPU TLBs: @arg must be NULL. */
+#define HVMOP_flush_tlbs 5
+
++/* Following tools-only interfaces may change in future. */
++#if defined(__XEN__) || defined(__XEN_TOOLS__)
++
++/* Track dirty VRAM. */
++#define HVMOP_track_dirty_vram 6
++struct xen_hvm_track_dirty_vram {
++ /* Domain to be tracked. */
++ domid_t domid;
++ /* First pfn to track. */
++ uint64_aligned_t first_pfn;
++ /* Number of pages to track. */
++ uint64_aligned_t nr;
++ /* OUT variable. */
++ /* Dirty bitmap buffer. */
++ XEN_GUEST_HANDLE_64(uint8) dirty_bitmap;
++};
++typedef struct xen_hvm_track_dirty_vram xen_hvm_track_dirty_vram_t;
++DEFINE_XEN_GUEST_HANDLE(xen_hvm_track_dirty_vram_t);
++
++#endif /* defined(__XEN__) || defined(__XEN_TOOLS__) */
++
+#endif /* __XEN_PUBLIC_HVM_HVM_OP_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/ioreq.h linux-2.6.18-xen.hg/include/xen/interface/hvm/ioreq.h
---- linux-2.6.18/include/xen/interface/hvm/ioreq.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/ioreq.h 2007-12-23 11:15:46.488588533 +0100
-@@ -0,0 +1,122 @@
+--- linux-2.6.18.8/include/xen/interface/hvm/ioreq.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/ioreq.h 2008-05-19 00:34:28.397241455 +0300
+@@ -0,0 +1,127 @@
+/*
+ * ioreq.h: I/O request definitions for device models
+ * Copyright (c) 2004, Intel Corporation.
+
+#define IOREQ_TYPE_PIO 0 /* pio */
+#define IOREQ_TYPE_COPY 1 /* mmio ops */
-+#define IOREQ_TYPE_AND 2
-+#define IOREQ_TYPE_OR 3
-+#define IOREQ_TYPE_XOR 4
-+#define IOREQ_TYPE_XCHG 5
-+#define IOREQ_TYPE_ADD 6
+#define IOREQ_TYPE_TIMEOFFSET 7
+#define IOREQ_TYPE_INVALIDATE 8 /* mapcache */
-+#define IOREQ_TYPE_SUB 9
+
+/*
+ * VMExit dispatcher should cooperate with instruction decoder to
+ * of the real data to use. */
+ uint8_t dir:1; /* 1=read, 0=write */
+ uint8_t df:1;
++ uint8_t pad:1;
+ uint8_t type; /* I/O type */
+ uint8_t _pad0[6];
+ uint64_t io_count; /* How many IO done on a vcpu */
+};
+typedef struct shared_iopage shared_iopage_t;
+
-+#define IOREQ_BUFFER_SLOT_NUM 80
++struct buf_ioreq {
++ uint8_t type; /* I/O type */
++ uint8_t pad:1;
++ uint8_t dir:1; /* 1=read, 0=write */
++ uint8_t size:2; /* 0=>1, 1=>2, 2=>4, 3=>8. If 8, use two buf_ioreqs */
++ uint32_t addr:20;/* physical address */
++ uint32_t data; /* data */
++};
++typedef struct buf_ioreq buf_ioreq_t;
++
++#define IOREQ_BUFFER_SLOT_NUM 511 /* 8 bytes each, plus 2 4-byte indexes */
+struct buffered_iopage {
-+ unsigned int read_pointer;
-+ unsigned int write_pointer;
-+ ioreq_t ioreq[IOREQ_BUFFER_SLOT_NUM];
++ unsigned int read_pointer;
++ unsigned int write_pointer;
++ buf_ioreq_t buf_ioreq[IOREQ_BUFFER_SLOT_NUM];
+}; /* NB. Size of this structure must be no greater than one page. */
+typedef struct buffered_iopage buffered_iopage_t;
+
+};
+#endif /* defined(__ia64__) */
+
-+#if defined(__i386__) || defined(__x86_64__)
+#define ACPI_PM1A_EVT_BLK_ADDRESS 0x0000000000001f40
+#define ACPI_PM1A_CNT_BLK_ADDRESS (ACPI_PM1A_EVT_BLK_ADDRESS + 0x04)
+#define ACPI_PM_TMR_BLK_ADDRESS (ACPI_PM1A_EVT_BLK_ADDRESS + 0x08)
-+#endif /* defined(__i386__) || defined(__x86_64__) */
++#define ACPI_GPE0_BLK_ADDRESS (ACPI_PM_TMR_BLK_ADDRESS + 0x20)
++#define ACPI_GPE0_BLK_LEN 0x08
+
+#endif /* _IOREQ_H_ */
+
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/params.h linux-2.6.18-xen.hg/include/xen/interface/hvm/params.h
---- linux-2.6.18/include/xen/interface/hvm/params.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/params.h 2007-12-23 11:15:46.488588533 +0100
-@@ -0,0 +1,61 @@
+--- linux-2.6.18.8/include/xen/interface/hvm/params.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/params.h 2008-05-19 00:34:28.397241455 +0300
+@@ -0,0 +1,95 @@
+/*
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+#ifdef __ia64__
+#define HVM_PARAM_NVRAM_FD 7
+#define HVM_PARAM_VHPT_SIZE 8
-+#define HVM_NR_PARAMS 9
-+#else
-+#define HVM_NR_PARAMS 7
++#define HVM_PARAM_BUFPIOREQ_PFN 9
+#endif
+
++/*
++ * Set mode for virtual timers (currently x86 only):
++ * delay_for_missed_ticks (default):
++ * Do not advance a vcpu's time beyond the correct delivery time for
++ * interrupts that have been missed due to preemption. Deliver missed
++ * interrupts when the vcpu is rescheduled and advance the vcpu's virtual
++ * time stepwise for each one.
++ * no_delay_for_missed_ticks:
++ * As above, missed interrupts are delivered, but guest time always tracks
++ * wallclock (i.e., real) time while doing so.
++ * no_missed_ticks_pending:
++ * No missed interrupts are held pending. Instead, to ensure ticks are
++ * delivered at some non-zero rate, if we detect missed ticks then the
++ * internal tick alarm is not disabled if the VCPU is preempted during the
++ * next tick period.
++ * one_missed_tick_pending:
++ * Missed interrupts are collapsed together and delivered as one 'late tick'.
++ * Guest time always tracks wallclock (i.e., real) time.
++ */
++#define HVM_PARAM_TIMER_MODE 10
++#define HVMPTM_delay_for_missed_ticks 0
++#define HVMPTM_no_delay_for_missed_ticks 1
++#define HVMPTM_no_missed_ticks_pending 2
++#define HVMPTM_one_missed_tick_pending 3
++
++/* Boolean: Enable virtual HPET (high-precision event timer)? (x86-only) */
++#define HVM_PARAM_HPET_ENABLED 11
++
++/* Identity-map page directory used by Intel EPT when CR0.PG=0. */
++#define HVM_PARAM_IDENT_PT 12
++
++/* Device Model domain, defaults to 0. */
++#define HVM_PARAM_DM_DOMAIN 13
++
++#define HVM_NR_PARAMS 14
++
+#endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/save.h linux-2.6.18-xen.hg/include/xen/interface/hvm/save.h
---- linux-2.6.18/include/xen/interface/hvm/save.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/save.h 2007-12-23 11:15:46.488588533 +0100
-@@ -0,0 +1,84 @@
+--- linux-2.6.18.8/include/xen/interface/hvm/save.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/save.h 2008-05-19 00:34:28.401241685 +0300
+@@ -0,0 +1,88 @@
+/*
+ * hvm/save.h
+ *
+ * Internal mechanisms should be kept in Xen-private headers.
+ */
+
++#if !defined(__GNUC__) || defined(__STRICT_ANSI__)
++#error "Anonymous structs/unions are a GNU extension."
++#endif
++
+/*
+ * Each entry is preceded by a descriptor giving its type and length
+ */
+#endif
+
+#endif /* __XEN_PUBLIC_HVM_SAVE_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/hvm/vmx_assist.h linux-2.6.18-xen.hg/include/xen/interface/hvm/vmx_assist.h
---- linux-2.6.18/include/xen/interface/hvm/vmx_assist.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/hvm/vmx_assist.h 2007-12-23 11:15:46.491922039 +0100
+--- linux-2.6.18.8/include/xen/interface/hvm/vmx_assist.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/hvm/vmx_assist.h 2008-05-19 00:34:28.401241685 +0300
@@ -0,0 +1,122 @@
+/*
+ * vmx_assist.h: Context definitions for the VMXASSIST world switch.
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/blkif.h linux-2.6.18-xen.hg/include/xen/interface/io/blkif.h
---- linux-2.6.18/include/xen/interface/io/blkif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/blkif.h 2007-12-23 11:15:46.491922039 +0100
+--- linux-2.6.18.8/include/xen/interface/io/blkif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/blkif.h 2008-05-19 00:34:28.401241685 +0300
@@ -0,0 +1,141 @@
+/******************************************************************************
+ * blkif.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/console.h linux-2.6.18-xen.hg/include/xen/interface/io/console.h
---- linux-2.6.18/include/xen/interface/io/console.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/console.h 2007-12-23 11:15:46.491922039 +0100
+--- linux-2.6.18.8/include/xen/interface/io/console.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/console.h 2008-05-19 00:34:28.401241685 +0300
@@ -0,0 +1,51 @@
+/******************************************************************************
+ * console.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/fbif.h linux-2.6.18-xen.hg/include/xen/interface/io/fbif.h
---- linux-2.6.18/include/xen/interface/io/fbif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/fbif.h 2007-12-23 11:15:46.491922039 +0100
-@@ -0,0 +1,138 @@
+--- linux-2.6.18.8/include/xen/interface/io/fbif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/fbif.h 2008-05-19 00:34:28.405241916 +0300
+@@ -0,0 +1,157 @@
+/*
+ * fbif.h -- Xen virtual frame buffer device
+ *
+ int32_t height; /* rect height */
+};
+
++/*
++ * Framebuffer resize notification event
++ * Capable backend sets feature-resize in xenstore.
++ */
++#define XENFB_TYPE_RESIZE 3
++
++struct xenfb_resize
++{
++ uint8_t type; /* XENFB_TYPE_RESIZE */
++ int32_t width; /* width in pixels */
++ int32_t height; /* height in pixels */
++ int32_t stride; /* stride in bytes */
++ int32_t depth; /* depth in bits */
++ int32_t offset; /* offset of the framebuffer in bytes */
++};
++
+#define XENFB_OUT_EVENT_SIZE 40
+
+union xenfb_out_event
+{
+ uint8_t type;
+ struct xenfb_update update;
++ struct xenfb_resize resize;
+ char pad[XENFB_OUT_EVENT_SIZE];
+};
+
+ * Each directory page holds PAGE_SIZE / sizeof(*pd)
+ * framebuffer pages, and can thus map up to PAGE_SIZE *
+ * PAGE_SIZE / sizeof(*pd) bytes. With PAGE_SIZE == 4096 and
-+ * sizeof(unsigned long) == 4, that's 4 Megs. Two directory
-+ * pages should be enough for a while.
++ * sizeof(unsigned long) == 4/8, that's 4 Megs 32 bit and 2 Megs
++ * 64 bit. 256 directories give enough room for a 512 Meg
++ * framebuffer with a max resolution of 12,800x10,240. Should
++ * be enough for a while with room leftover for expansion.
+ */
-+ unsigned long pd[2];
++ unsigned long pd[256];
+};
+
+/*
-+ * Wart: xenkbd needs to know resolution. Put it here until a better
-+ * solution is found, but don't leak it to the backend.
++ * Wart: xenkbd needs to know default resolution. Put it here until a
++ * better solution is found, but don't leak it to the backend.
+ */
+#ifdef __KERNEL__
+#define XENFB_WIDTH 800
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/kbdif.h linux-2.6.18-xen.hg/include/xen/interface/io/kbdif.h
---- linux-2.6.18/include/xen/interface/io/kbdif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/kbdif.h 2007-12-23 11:15:46.491922039 +0100
+--- linux-2.6.18.8/include/xen/interface/io/fsif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/fsif.h 2008-05-19 00:34:28.405241916 +0300
+@@ -0,0 +1,181 @@
++/******************************************************************************
++ * fsif.h
++ *
++ * Interface to FS level split device drivers.
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ *
++ * Copyright (c) 2007, Grzegorz Milos, Sun Microsystems, Inc.
++ */
++
++#ifndef __XEN_PUBLIC_IO_FSIF_H__
++#define __XEN_PUBLIC_IO_FSIF_H__
++
++#include "ring.h"
++#include "../grant_table.h"
++
++#define REQ_FILE_OPEN 1
++#define REQ_FILE_CLOSE 2
++#define REQ_FILE_READ 3
++#define REQ_FILE_WRITE 4
++#define REQ_STAT 5
++#define REQ_FILE_TRUNCATE 6
++#define REQ_REMOVE 7
++#define REQ_RENAME 8
++#define REQ_CREATE 9
++#define REQ_DIR_LIST 10
++#define REQ_CHMOD 11
++#define REQ_FS_SPACE 12
++#define REQ_FILE_SYNC 13
++
++struct fsif_open_request {
++ grant_ref_t gref;
++};
++
++struct fsif_close_request {
++ uint32_t fd;
++};
++
++struct fsif_read_request {
++ uint32_t fd;
++ grant_ref_t gref;
++ uint64_t len;
++ uint64_t offset;
++};
++
++struct fsif_write_request {
++ uint32_t fd;
++ grant_ref_t gref;
++ uint64_t len;
++ uint64_t offset;
++};
++
++struct fsif_stat_request {
++ uint32_t fd;
++ grant_ref_t gref;
++};
++
++/* This structure is a copy of some fields from stat structure, writen to the
++ * granted page. */
++struct fsif_stat_response {
++ int32_t stat_mode;
++ uint32_t stat_uid;
++ uint32_t stat_gid;
++ int32_t pad;
++ int64_t stat_size;
++ int64_t stat_atime;
++ int64_t stat_mtime;
++ int64_t stat_ctime;
++};
++
++struct fsif_truncate_request {
++ uint32_t fd;
++ int32_t pad;
++ int64_t length;
++};
++
++struct fsif_remove_request {
++ grant_ref_t gref;
++};
++
++struct fsif_rename_request {
++ uint16_t old_name_offset;
++ uint16_t new_name_offset;
++ grant_ref_t gref;
++};
++
++struct fsif_create_request {
++ int8_t directory;
++ int8_t pad;
++ int16_t pad2;
++ int32_t mode;
++ grant_ref_t gref;
++};
++
++struct fsif_list_request {
++ uint32_t offset;
++ grant_ref_t gref;
++};
++
++#define NR_FILES_SHIFT 0
++#define NR_FILES_SIZE 16 /* 16 bits for the number of files mask */
++#define NR_FILES_MASK (((1ULL << NR_FILES_SIZE) - 1) << NR_FILES_SHIFT)
++#define ERROR_SIZE 32 /* 32 bits for the error mask */
++#define ERROR_SHIFT (NR_FILES_SIZE + NR_FILES_SHIFT)
++#define ERROR_MASK (((1ULL << ERROR_SIZE) - 1) << ERROR_SHIFT)
++#define HAS_MORE_SHIFT (ERROR_SHIFT + ERROR_SIZE)
++#define HAS_MORE_FLAG (1ULL << HAS_MORE_SHIFT)
++
++struct fsif_chmod_request {
++ uint32_t fd;
++ int32_t mode;
++};
++
++struct fsif_space_request {
++ grant_ref_t gref;
++};
++
++struct fsif_sync_request {
++ uint32_t fd;
++};
++
++
++/* FS operation request */
++struct fsif_request {
++ uint8_t type; /* Type of the request */
++ uint8_t pad;
++ uint16_t id; /* Request ID, copied to the response */
++ uint32_t pad2;
++ union {
++ struct fsif_open_request fopen;
++ struct fsif_close_request fclose;
++ struct fsif_read_request fread;
++ struct fsif_write_request fwrite;
++ struct fsif_stat_request fstat;
++ struct fsif_truncate_request ftruncate;
++ struct fsif_remove_request fremove;
++ struct fsif_rename_request frename;
++ struct fsif_create_request fcreate;
++ struct fsif_list_request flist;
++ struct fsif_chmod_request fchmod;
++ struct fsif_space_request fspace;
++ struct fsif_sync_request fsync;
++ } u;
++};
++typedef struct fsif_request fsif_request_t;
++
++/* FS operation response */
++struct fsif_response {
++ uint16_t id;
++ uint16_t pad1;
++ uint32_t pad2;
++ uint64_t ret_val;
++};
++
++typedef struct fsif_response fsif_response_t;
++
++
++DEFINE_RING_TYPES(fsif, struct fsif_request, struct fsif_response);
++
++#define STATE_INITIALISED "init"
++#define STATE_READY "ready"
++
++
++
++#endif
+--- linux-2.6.18.8/include/xen/interface/io/kbdif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/kbdif.h 2008-05-19 00:34:28.405241916 +0300
@@ -0,0 +1,132 @@
+/*
+ * kbdif.h -- Xen virtual keyboard/mouse
+ uint8_t type; /* XENKBD_TYPE_POS */
+ int32_t abs_x; /* absolute X position (in FB pixels) */
+ int32_t abs_y; /* absolute Y position (in FB pixels) */
-+ int32_t abs_z; /* absolute Z position (wheel) */
++ int32_t rel_z; /* relative Z motion (wheel) */
+};
+
+#define XENKBD_IN_EVENT_SIZE 40
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/netif.h linux-2.6.18-xen.hg/include/xen/interface/io/netif.h
---- linux-2.6.18/include/xen/interface/io/netif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/netif.h 2007-12-23 11:15:46.491922039 +0100
+--- linux-2.6.18.8/include/xen/interface/io/netif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/netif.h 2008-05-19 00:34:28.417242608 +0300
@@ -0,0 +1,205 @@
+/******************************************************************************
+ * netif.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/pciif.h linux-2.6.18-xen.hg/include/xen/interface/io/pciif.h
---- linux-2.6.18/include/xen/interface/io/pciif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/pciif.h 2007-12-23 11:15:46.491922039 +0100
-@@ -0,0 +1,83 @@
+--- linux-2.6.18.8/include/xen/interface/io/pciif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/pciif.h 2008-05-19 00:34:28.417242608 +0300
+@@ -0,0 +1,97 @@
+/*
+ * PCI Backend/Frontend Common Data Structures & Macros
+ *
+/* xen_pci_op commands */
+#define XEN_PCI_OP_conf_read (0)
+#define XEN_PCI_OP_conf_write (1)
++#define XEN_PCI_OP_enable_msi (2)
++#define XEN_PCI_OP_disable_msi (3)
++#define XEN_PCI_OP_enable_msix (4)
++#define XEN_PCI_OP_disable_msix (5)
+
+/* xen_pci_op error numbers */
+#define XEN_PCI_ERR_success (0)
+/* XEN_PCI_ERR_op_failed - backend failed to complete the operation */
+#define XEN_PCI_ERR_op_failed (-5)
+
++/*
++ * it should be PAGE_SIZE-sizeof(struct xen_pci_op))/sizeof(struct msix_entry))
++ * Should not exceed 128
++ */
++#define SH_INFO_MAX_VEC 128
++
+struct xen_pci_op {
+ /* IN: what action to perform: XEN_PCI_OP_* */
+ uint32_t cmd;
+
+ /* IN/OUT: Contains the result after a READ or the value to WRITE */
+ uint32_t value;
++ /* IN: Contains extra infor for this operation */
++ uint32_t info;
++ /*IN: param for msi-x */
++ struct msix_entry msix_entries[SH_INFO_MAX_VEC];
+};
+
+struct xen_pci_sharedinfo {
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/protocols.h linux-2.6.18-xen.hg/include/xen/interface/io/protocols.h
---- linux-2.6.18/include/xen/interface/io/protocols.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/protocols.h 2007-12-23 11:15:46.495255548 +0100
-@@ -0,0 +1,21 @@
+--- linux-2.6.18.8/include/xen/interface/io/protocols.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/protocols.h 2008-05-19 00:34:28.417242608 +0300
+@@ -0,0 +1,40 @@
++/******************************************************************************
++ * protocols.h
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ */
++
+#ifndef __XEN_PROTOCOLS_H__
+#define __XEN_PROTOCOLS_H__
+
+#define XEN_IO_PROTO_ABI_X86_32 "x86_32-abi"
+#define XEN_IO_PROTO_ABI_X86_64 "x86_64-abi"
+#define XEN_IO_PROTO_ABI_IA64 "ia64-abi"
-+#define XEN_IO_PROTO_ABI_POWERPC64 "powerpc64-abi"
+
+#if defined(__i386__)
+# define XEN_IO_PROTO_ABI_NATIVE XEN_IO_PROTO_ABI_X86_32
+# define XEN_IO_PROTO_ABI_NATIVE XEN_IO_PROTO_ABI_X86_64
+#elif defined(__ia64__)
+# define XEN_IO_PROTO_ABI_NATIVE XEN_IO_PROTO_ABI_IA64
-+#elif defined(__powerpc64__)
-+# define XEN_IO_PROTO_ABI_NATIVE XEN_IO_PROTO_ABI_POWERPC64
+#else
+# error arch fixup needed here
+#endif
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/ring.h linux-2.6.18-xen.hg/include/xen/interface/io/ring.h
---- linux-2.6.18/include/xen/interface/io/ring.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/ring.h 2007-12-23 11:15:46.495255548 +0100
-@@ -0,0 +1,299 @@
+--- linux-2.6.18.8/include/xen/interface/io/ring.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/ring.h 2008-05-19 00:34:28.417242608 +0300
+@@ -0,0 +1,307 @@
+/******************************************************************************
+ * ring.h
+ *
+#ifndef __XEN_PUBLIC_IO_RING_H__
+#define __XEN_PUBLIC_IO_RING_H__
+
++#include "../xen-compat.h"
++
++#if __XEN_INTERFACE_VERSION__ < 0x00030208
++#define xen_mb() mb()
++#define xen_rmb() rmb()
++#define xen_wmb() wmb()
++#endif
++
+typedef unsigned int RING_IDX;
+
+/* Round a 32-bit unsigned constant down to the nearest power of two. */
+ (((_cons) - (_r)->rsp_prod_pvt) >= RING_SIZE(_r))
+
+#define RING_PUSH_REQUESTS(_r) do { \
-+ wmb(); /* back sees requests /before/ updated producer index */ \
++ xen_wmb(); /* back sees requests /before/ updated producer index */ \
+ (_r)->sring->req_prod = (_r)->req_prod_pvt; \
+} while (0)
+
+#define RING_PUSH_RESPONSES(_r) do { \
-+ wmb(); /* front sees responses /before/ updated producer index */ \
++ xen_wmb(); /* front sees resps /before/ updated producer index */ \
+ (_r)->sring->rsp_prod = (_r)->rsp_prod_pvt; \
+} while (0)
+
+#define RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(_r, _notify) do { \
+ RING_IDX __old = (_r)->sring->req_prod; \
+ RING_IDX __new = (_r)->req_prod_pvt; \
-+ wmb(); /* back sees requests /before/ updated producer index */ \
++ xen_wmb(); /* back sees requests /before/ updated producer index */ \
+ (_r)->sring->req_prod = __new; \
-+ mb(); /* back sees new requests /before/ we check req_event */ \
++ xen_mb(); /* back sees new requests /before/ we check req_event */ \
+ (_notify) = ((RING_IDX)(__new - (_r)->sring->req_event) < \
+ (RING_IDX)(__new - __old)); \
+} while (0)
+#define RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(_r, _notify) do { \
+ RING_IDX __old = (_r)->sring->rsp_prod; \
+ RING_IDX __new = (_r)->rsp_prod_pvt; \
-+ wmb(); /* front sees responses /before/ updated producer index */ \
++ xen_wmb(); /* front sees resps /before/ updated producer index */ \
+ (_r)->sring->rsp_prod = __new; \
-+ mb(); /* front sees new responses /before/ we check rsp_event */ \
++ xen_mb(); /* front sees new resps /before/ we check rsp_event */ \
+ (_notify) = ((RING_IDX)(__new - (_r)->sring->rsp_event) < \
+ (RING_IDX)(__new - __old)); \
+} while (0)
+ (_work_to_do) = RING_HAS_UNCONSUMED_REQUESTS(_r); \
+ if (_work_to_do) break; \
+ (_r)->sring->req_event = (_r)->req_cons + 1; \
-+ mb(); \
++ xen_mb(); \
+ (_work_to_do) = RING_HAS_UNCONSUMED_REQUESTS(_r); \
+} while (0)
+
+ (_work_to_do) = RING_HAS_UNCONSUMED_RESPONSES(_r); \
+ if (_work_to_do) break; \
+ (_r)->sring->rsp_event = (_r)->rsp_cons + 1; \
-+ mb(); \
++ xen_mb(); \
+ (_work_to_do) = RING_HAS_UNCONSUMED_RESPONSES(_r); \
+} while (0)
+
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/tpmif.h linux-2.6.18-xen.hg/include/xen/interface/io/tpmif.h
---- linux-2.6.18/include/xen/interface/io/tpmif.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/tpmif.h 2007-12-23 11:15:46.495255548 +0100
+--- linux-2.6.18.8/include/xen/interface/io/tpmif.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/tpmif.h 2008-05-19 00:34:28.421242838 +0300
@@ -0,0 +1,77 @@
+/******************************************************************************
+ * tpmif.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/xenbus.h linux-2.6.18-xen.hg/include/xen/interface/io/xenbus.h
---- linux-2.6.18/include/xen/interface/io/xenbus.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/xenbus.h 2007-12-23 11:15:46.495255548 +0100
-@@ -0,0 +1,73 @@
+--- linux-2.6.18.8/include/xen/interface/io/xenbus.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/xenbus.h 2008-05-19 00:34:28.469245605 +0300
+@@ -0,0 +1,80 @@
+/*****************************************************************************
+ * xenbus.h
+ *
+ */
+ XenbusStateClosing = 5,
+
-+ XenbusStateClosed = 6
++ XenbusStateClosed = 6,
++
++ /*
++ * Reconfiguring: The device is being reconfigured.
++ */
++ XenbusStateReconfiguring = 7,
++
++ XenbusStateReconfigured = 8
+};
+typedef enum xenbus_state XenbusState;
+
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/io/xs_wire.h linux-2.6.18-xen.hg/include/xen/interface/io/xs_wire.h
---- linux-2.6.18/include/xen/interface/io/xs_wire.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/io/xs_wire.h 2007-12-23 11:15:46.571926242 +0100
-@@ -0,0 +1,121 @@
+--- linux-2.6.18.8/include/xen/interface/io/xs_wire.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/io/xs_wire.h 2008-05-19 00:34:28.469245605 +0300
+@@ -0,0 +1,130 @@
+/*
+ * Details of the "wire" protocol between Xen Store Daemon and client
+ * library or guest kernel.
+ XS_WATCH_EVENT,
+ XS_ERROR,
+ XS_IS_DOMAIN_INTRODUCED,
-+ XS_RESUME
++ XS_RESUME,
++ XS_SET_TARGET
+};
+
+#define XS_WRITE_NONE "NONE"
+ const char *errstring;
+};
+#define XSD_ERROR(x) { x, #x }
++/* LINTED: static unused */
+static struct xsd_errors xsd_errors[]
+#if defined(__GNUC__)
+__attribute__((unused))
+ XENSTORE_RING_IDX rsp_cons, rsp_prod;
+};
+
++/* Violating this is very bad. See docs/misc/xenstore.txt. */
++#define XENSTORE_PAYLOAD_MAX 4096
++
++/* Violating these just gets you an error back */
++#define XENSTORE_ABS_PATH_MAX 3072
++#define XENSTORE_REL_PATH_MAX 2048
++
+#endif /* _XS_WIRE_H */
+
+/*
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/kexec.h linux-2.6.18-xen.hg/include/xen/interface/kexec.h
---- linux-2.6.18/include/xen/interface/kexec.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/kexec.h 2007-12-23 11:15:46.575259748 +0100
-@@ -0,0 +1,140 @@
+--- linux-2.6.18.8/include/xen/interface/kexec.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/kexec.h 2008-05-19 00:34:28.469245605 +0300
+@@ -0,0 +1,167 @@
+/******************************************************************************
+ * kexec.h - Public portion
+ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ *
+ * Xen port written by:
+ * - Simon 'Horms' Horman <horms@verge.net.au>
+ * - Magnus Damm <magnus@valinux.co.jp>
+ xen_kexec_image_t image;
+} xen_kexec_load_t;
+
-+#define KEXEC_RANGE_MA_CRASH 0 /* machine address and size of crash area */
-+#define KEXEC_RANGE_MA_XEN 1 /* machine address and size of Xen itself */
-+#define KEXEC_RANGE_MA_CPU 2 /* machine address and size of a CPU note */
++#define KEXEC_RANGE_MA_CRASH 0 /* machine address and size of crash area */
++#define KEXEC_RANGE_MA_XEN 1 /* machine address and size of Xen itself */
++#define KEXEC_RANGE_MA_CPU 2 /* machine address and size of a CPU note */
++#define KEXEC_RANGE_MA_XENHEAP 3 /* machine address and size of xenheap
++ * Note that although this is adjacent
++ * to Xen it exists in a separate EFI
++ * region on ia64, and thus needs to be
++ * inserted into iomem_machine separately */
++#define KEXEC_RANGE_MA_BOOT_PARAM 4 /* machine address and size of
++ * the ia64_boot_param */
++#define KEXEC_RANGE_MA_EFI_MEMMAP 5 /* machine address and size of
++ * of the EFI Memory Map */
+
+/*
+ * Find the address and size of certain memory areas
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/libelf.h linux-2.6.18-xen.hg/include/xen/interface/libelf.h
---- linux-2.6.18/include/xen/interface/libelf.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/libelf.h 2007-12-23 11:15:46.575259748 +0100
-@@ -0,0 +1,245 @@
+--- linux-2.6.18.8/include/xen/interface/libelf.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/libelf.h 2008-05-19 00:34:28.469245605 +0300
+@@ -0,0 +1,265 @@
++/******************************************************************************
++ * libelf.h
++ *
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ */
++
+#ifndef __XC_LIBELF__
+#define __XC_LIBELF__ 1
+
-+#if defined(__i386__) || defined(__x86_64) || defined(__ia64__)
++#if defined(__i386__) || defined(__x86_64__) || defined(__ia64__)
+#define XEN_ELF_LITTLE_ENDIAN
-+#elif defined(__powerpc__)
-+#define XEN_ELF_BIG_ENDIAN
+#else
+#error define architectural endianness
+#endif
+ struct elf_dom_parms *parms);
+
+#endif /* __XC_LIBELF__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/memory.h linux-2.6.18-xen.hg/include/xen/interface/memory.h
---- linux-2.6.18/include/xen/interface/memory.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/memory.h 2007-12-23 11:15:46.575259748 +0100
+--- linux-2.6.18.8/include/xen/interface/memory.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/memory.h 2008-05-19 00:34:28.473245836 +0300
@@ -0,0 +1,281 @@
+/******************************************************************************
+ * memory.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/nmi.h linux-2.6.18-xen.hg/include/xen/interface/nmi.h
---- linux-2.6.18/include/xen/interface/nmi.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/nmi.h 2007-12-23 11:15:46.575259748 +0100
+--- linux-2.6.18.8/include/xen/interface/nmi.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/nmi.h 2008-05-19 00:34:28.473245836 +0300
@@ -0,0 +1,78 @@
+/******************************************************************************
+ * nmi.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/physdev.h linux-2.6.18-xen.hg/include/xen/interface/physdev.h
---- linux-2.6.18/include/xen/interface/physdev.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/physdev.h 2007-12-23 11:15:46.575259748 +0100
-@@ -0,0 +1,169 @@
+--- linux-2.6.18.8/include/xen/interface/physdev.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/physdev.h 2008-05-19 00:34:28.473245836 +0300
+@@ -0,0 +1,205 @@
+/*
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+#define PHYSDEVOP_set_iobitmap 7
+struct physdev_set_iobitmap {
+ /* IN */
-+ XEN_GUEST_HANDLE_00030205(uint8_t) bitmap;
++#if __XEN_INTERFACE_VERSION__ >= 0x00030205
++ XEN_GUEST_HANDLE(uint8) bitmap;
++#else
++ uint8_t *bitmap;
++#endif
+ uint32_t nr_ports;
+};
+typedef struct physdev_set_iobitmap physdev_set_iobitmap_t;
+};
+typedef struct physdev_irq physdev_irq_t;
+DEFINE_XEN_GUEST_HANDLE(physdev_irq_t);
++
++#define MAP_PIRQ_TYPE_MSI 0x0
++#define MAP_PIRQ_TYPE_GSI 0x1
++#define MAP_PIRQ_TYPE_UNKNOWN 0x2
++
++#define PHYSDEVOP_map_pirq 13
++struct physdev_map_pirq {
++ domid_t domid;
++ /* IN */
++ int type;
++ /* IN */
++ int index;
++ /* IN or OUT */
++ int pirq;
++ /* IN */
++ struct {
++ int bus, devfn, entry_nr;
++ int msi; /* 0 - MSIX 1 - MSI */
++ } msi_info;
++};
++typedef struct physdev_map_pirq physdev_map_pirq_t;
++DEFINE_XEN_GUEST_HANDLE(physdev_map_pirq_t);
++
++#define PHYSDEVOP_unmap_pirq 14
++struct physdev_unmap_pirq {
++ domid_t domid;
++ /* IN */
++ int pirq;
++};
++
++typedef struct physdev_unmap_pirq physdev_unmap_pirq_t;
++DEFINE_XEN_GUEST_HANDLE(physdev_unmap_pirq_t);
+
+/*
+ * Argument to physdev_op_compat() hypercall. Superceded by new physdev_op()
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/platform.h linux-2.6.18-xen.hg/include/xen/interface/platform.h
---- linux-2.6.18/include/xen/interface/platform.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/platform.h 2007-12-23 11:15:46.575259748 +0100
-@@ -0,0 +1,232 @@
+--- linux-2.6.18.8/include/xen/interface/platform.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/platform.h 2008-05-19 00:34:28.473245836 +0300
+@@ -0,0 +1,346 @@
+/******************************************************************************
+ * platform.h
+ *
+ uint8_t capabilities;
+ uint8_t edid_transfer_time;
+ /* must refer to 128-byte buffer */
-+ XEN_GUEST_HANDLE(uint8_t) edid;
++ XEN_GUEST_HANDLE(uint8) edid;
+ } vbeddc_info; /* XEN_FW_VBEDDC_INFO */
+ } u;
+};
+struct xenpf_getidletime {
+ /* IN/OUT variables */
+ /* IN: CPUs to interrogate; OUT: subset of IN which are present */
-+ XEN_GUEST_HANDLE(uint8_t) cpumap_bitmap;
++ XEN_GUEST_HANDLE(uint8) cpumap_bitmap;
+ /* IN variables */
+ /* Size of cpumap bitmap. */
+ uint32_t cpumap_nr_cpus;
+ /* Must be indexable for every cpu in cpumap_bitmap. */
-+ XEN_GUEST_HANDLE(uint64_t) idletime;
++ XEN_GUEST_HANDLE(uint64) idletime;
+ /* OUT variables */
+ /* System time when the idletime snapshots were taken. */
+ uint64_t now;
+typedef struct xenpf_getidletime xenpf_getidletime_t;
+DEFINE_XEN_GUEST_HANDLE(xenpf_getidletime_t);
+
++#define XENPF_set_processor_pminfo 54
++
++/* ability bits */
++#define XEN_PROCESSOR_PM_CX 1
++#define XEN_PROCESSOR_PM_PX 2
++#define XEN_PROCESSOR_PM_TX 4
++
++/* cmd type */
++#define XEN_PM_CX 0
++#define XEN_PM_PX 1
++#define XEN_PM_TX 2
++
++/* Px sub info type */
++#define XEN_PX_PCT 1
++#define XEN_PX_PSS 2
++#define XEN_PX_PPC 4
++#define XEN_PX_PSD 8
++
++struct xen_power_register {
++ uint32_t space_id;
++ uint32_t bit_width;
++ uint32_t bit_offset;
++ uint32_t access_size;
++ uint64_t address;
++};
++
++struct xen_processor_csd {
++ uint32_t domain; /* domain number of one dependent group */
++ uint32_t coord_type; /* coordination type */
++ uint32_t num; /* number of processors in same domain */
++};
++typedef struct xen_processor_csd xen_processor_csd_t;
++DEFINE_XEN_GUEST_HANDLE(xen_processor_csd_t);
++
++struct xen_processor_cx {
++ struct xen_power_register reg; /* GAS for Cx trigger register */
++ uint8_t type; /* cstate value, c0: 0, c1: 1, ... */
++ uint32_t latency; /* worst latency (ms) to enter/exit this cstate */
++ uint32_t power; /* average power consumption(mW) */
++ uint32_t dpcnt; /* number of dependency entries */
++ XEN_GUEST_HANDLE(xen_processor_csd_t) dp; /* NULL if no dependency */
++};
++typedef struct xen_processor_cx xen_processor_cx_t;
++DEFINE_XEN_GUEST_HANDLE(xen_processor_cx_t);
++
++struct xen_processor_flags {
++ uint32_t bm_control:1;
++ uint32_t bm_check:1;
++ uint32_t has_cst:1;
++ uint32_t power_setup_done:1;
++ uint32_t bm_rld_set:1;
++};
++
++struct xen_processor_power {
++ uint32_t count; /* number of C state entries in array below */
++ struct xen_processor_flags flags; /* global flags of this processor */
++ XEN_GUEST_HANDLE(xen_processor_cx_t) states; /* supported c states */
++};
++
++struct xen_pct_register {
++ uint8_t descriptor;
++ uint16_t length;
++ uint8_t space_id;
++ uint8_t bit_width;
++ uint8_t bit_offset;
++ uint8_t reserved;
++ uint64_t address;
++};
++
++struct xen_processor_px {
++ uint64_t core_frequency; /* megahertz */
++ uint64_t power; /* milliWatts */
++ uint64_t transition_latency; /* microseconds */
++ uint64_t bus_master_latency; /* microseconds */
++ uint64_t control; /* control value */
++ uint64_t status; /* success indicator */
++};
++typedef struct xen_processor_px xen_processor_px_t;
++DEFINE_XEN_GUEST_HANDLE(xen_processor_px_t);
++
++struct xen_psd_package {
++ uint64_t num_entries;
++ uint64_t revision;
++ uint64_t domain;
++ uint64_t coord_type;
++ uint64_t num_processors;
++};
++
++struct xen_processor_performance {
++ uint32_t flags; /* flag for Px sub info type */
++ uint32_t ppc; /* Platform limitation on freq usage */
++ struct xen_pct_register control_register;
++ struct xen_pct_register status_register;
++ uint32_t state_count; /* total available performance states */
++ XEN_GUEST_HANDLE(xen_processor_px_t) states;
++ struct xen_psd_package domain_info;
++ uint32_t shared_type; /* coordination type of this processor */
++};
++typedef struct xen_processor_performance xen_processor_performance_t;
++DEFINE_XEN_GUEST_HANDLE(xen_processor_performance_t);
++
++struct xenpf_set_processor_pminfo {
++ /* IN variables */
++ uint32_t id; /* ACPI CPU ID */
++ uint32_t type; /* {XEN_PM_CX, XEN_PM_PX} */
++ union {
++ struct xen_processor_power power;/* Cx: _CST/_CSD */
++ struct xen_processor_performance perf; /* Px: _PPC/_PCT/_PSS/_PSD */
++ };
++};
++typedef struct xenpf_set_processor_pminfo xenpf_set_processor_pminfo_t;
++DEFINE_XEN_GUEST_HANDLE(xenpf_set_processor_pminfo_t);
++
+struct xen_platform_op {
+ uint32_t cmd;
+ uint32_t interface_version; /* XENPF_INTERFACE_VERSION */
+ struct xenpf_enter_acpi_sleep enter_acpi_sleep;
+ struct xenpf_change_freq change_freq;
+ struct xenpf_getidletime getidletime;
++ struct xenpf_set_processor_pminfo set_pminfo;
+ uint8_t pad[128];
+ } u;
+};
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/sched.h linux-2.6.18-xen.hg/include/xen/interface/sched.h
---- linux-2.6.18/include/xen/interface/sched.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/sched.h 2007-12-23 11:15:46.578593257 +0100
+--- linux-2.6.18.8/include/xen/interface/sched.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/sched.h 2008-05-19 00:34:28.477246066 +0300
@@ -0,0 +1,121 @@
+/******************************************************************************
+ * sched.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/sysctl.h linux-2.6.18-xen.hg/include/xen/interface/sysctl.h
---- linux-2.6.18/include/xen/interface/sysctl.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/sysctl.h 2007-12-23 11:15:46.578593257 +0100
-@@ -0,0 +1,238 @@
+--- linux-2.6.18.8/include/xen/interface/sysctl.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/sysctl.h 2008-05-19 00:34:28.477246066 +0300
+@@ -0,0 +1,280 @@
+/******************************************************************************
+ * sysctl.h
+ *
+ * Get physical information about the host machine
+ */
+#define XEN_SYSCTL_physinfo 3
++ /* (x86) The platform supports HVM guests. */
++#define _XEN_SYSCTL_PHYSCAP_hvm 0
++#define XEN_SYSCTL_PHYSCAP_hvm (1u<<_XEN_SYSCTL_PHYSCAP_hvm)
++ /* (x86) The platform supports HVM-guest direct access to I/O devices. */
++#define _XEN_SYSCTL_PHYSCAP_hvm_directio 1
++#define XEN_SYSCTL_PHYSCAP_hvm_directio (1u<<_XEN_SYSCTL_PHYSCAP_hvm_directio)
+struct xen_sysctl_physinfo {
-+ /* IN variables. */
+ uint32_t threads_per_core;
+ uint32_t cores_per_socket;
+ uint32_t nr_cpus;
+ uint64_aligned_t scrub_pages;
+ uint32_t hw_cap[8];
+
-+ /* IN/OUT variables. */
+ /*
+ * IN: maximum addressable entry in the caller-provided cpu_to_node array.
+ * OUT: largest cpu identifier in the system.
+ * If the actual @max_cpu_id is smaller than the array then the trailing
+ * elements of the array will not be written by the sysctl.
+ */
-+ XEN_GUEST_HANDLE_64(uint32_t) cpu_to_node;
++ XEN_GUEST_HANDLE_64(uint32) cpu_to_node;
++
++ /* XEN_SYSCTL_PHYSCAP_??? */
++ uint32_t capabilities;
+};
+typedef struct xen_sysctl_physinfo xen_sysctl_physinfo_t;
+DEFINE_XEN_GUEST_HANDLE(xen_sysctl_physinfo_t);
+};
+typedef struct xen_sysctl_availheap xen_sysctl_availheap_t;
+DEFINE_XEN_GUEST_HANDLE(xen_sysctl_availheap_t);
++
++#define XEN_SYSCTL_get_pmstat 10
++struct pm_px_val {
++ uint64_aligned_t freq; /* Px core frequency */
++ uint64_aligned_t residency; /* Px residency time */
++ uint64_aligned_t count; /* Px transition count */
++};
++typedef struct pm_px_val pm_px_val_t;
++DEFINE_XEN_GUEST_HANDLE(pm_px_val_t);
++
++struct pm_px_stat {
++ uint8_t total; /* total Px states */
++ uint8_t usable; /* usable Px states */
++ uint8_t last; /* last Px state */
++ uint8_t cur; /* current Px state */
++ XEN_GUEST_HANDLE_64(uint64) trans_pt; /* Px transition table */
++ XEN_GUEST_HANDLE_64(pm_px_val_t) pt;
++};
++typedef struct pm_px_stat pm_px_stat_t;
++DEFINE_XEN_GUEST_HANDLE(pm_px_stat_t);
++
++struct xen_sysctl_get_pmstat {
++#define PMSTAT_get_max_px 0x11
++#define PMSTAT_get_pxstat 0x12
++#define PMSTAT_reset_pxstat 0x13
++ uint32_t type;
++ uint32_t cpuid;
++ union {
++ struct pm_px_stat getpx;
++ /* other struct for cx, tx, etc */
++ } u;
++};
++typedef struct xen_sysctl_get_pmstat xen_sysctl_get_pmstat_t;
++DEFINE_XEN_GUEST_HANDLE(xen_sysctl_get_pmstat_t);
+
+struct xen_sysctl {
+ uint32_t cmd;
+ struct xen_sysctl_debug_keys debug_keys;
+ struct xen_sysctl_getcpuinfo getcpuinfo;
+ struct xen_sysctl_availheap availheap;
++ struct xen_sysctl_get_pmstat get_pmstat;
+ uint8_t pad[128];
+ } u;
+};
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/trace.h linux-2.6.18-xen.hg/include/xen/interface/trace.h
---- linux-2.6.18/include/xen/interface/trace.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/trace.h 2007-12-23 11:15:46.578593257 +0100
-@@ -0,0 +1,159 @@
+--- linux-2.6.18.8/include/xen/interface/trace.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/trace.h 2008-05-19 00:34:28.477246066 +0300
+@@ -0,0 +1,167 @@
+/******************************************************************************
+ * include/public/trace.h
+ *
+ * field, indexes into an array of struct t_rec's.
+ */
+struct t_buf {
++ /* Assume the data buffer size is X. X is generally not a power of 2.
++ * CONS and PROD are incremented modulo (2*X):
++ * 0 <= cons < 2*X
++ * 0 <= prod < 2*X
++ * This is done because addition modulo X breaks at 2^32 when X is not a
++ * power of 2:
++ * (((2^32 - 1) % X) + 1) % X != (2^32) % X
++ */
+ uint32_t cons; /* Offset of next item to be consumed by control tools. */
+ uint32_t prod; /* Offset of next item to be produced by Xen. */
+ /* Records follow immediately after the meta-data header. */
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/vcpu.h linux-2.6.18-xen.hg/include/xen/interface/vcpu.h
---- linux-2.6.18/include/xen/interface/vcpu.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/vcpu.h 2007-12-23 11:15:46.578593257 +0100
-@@ -0,0 +1,195 @@
+--- linux-2.6.18.8/include/xen/interface/vcpu.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/vcpu.h 2008-05-19 00:34:28.493246989 +0300
+@@ -0,0 +1,213 @@
+/******************************************************************************
+ * vcpu.h
+ *
+ *
+ * This may be called only once per vcpu.
+ */
-+#define VCPUOP_register_vcpu_info 10 /* arg == struct vcpu_info */
++#define VCPUOP_register_vcpu_info 10 /* arg == vcpu_register_vcpu_info_t */
+struct vcpu_register_vcpu_info {
+ uint64_t mfn; /* mfn of page to place vcpu_info */
+ uint32_t offset; /* offset within page */
+/* Send an NMI to the specified VCPU. @extra_arg == NULL. */
+#define VCPUOP_send_nmi 11
+
++/*
++ * Get the physical ID information for a pinned vcpu's underlying physical
++ * processor. The physical ID informmation is architecture-specific.
++ * On x86: id[31:0]=apic_id, id[63:32]=acpi_id, and all values 0xff and
++ * greater are reserved.
++ * This command returns -EINVAL if it is not a valid operation for this VCPU.
++ */
++#define VCPUOP_get_physid 12 /* arg == vcpu_get_physid_t */
++struct vcpu_get_physid {
++ uint64_t phys_id;
++};
++typedef struct vcpu_get_physid vcpu_get_physid_t;
++DEFINE_XEN_GUEST_HANDLE(vcpu_get_physid_t);
++#define xen_vcpu_physid_to_x86_apicid(physid) \
++ ((((uint32_t)(physid)) >= 0xff) ? 0xff : ((uint8_t)(physid)))
++#define xen_vcpu_physid_to_x86_acpiid(physid) \
++ ((((uint32_t)((physid)>>32)) >= 0xff) ? 0xff : ((uint8_t)((physid)>>32)))
++
+#endif /* __XEN_PUBLIC_VCPU_H__ */
+
+/*
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/version.h linux-2.6.18-xen.hg/include/xen/interface/version.h
---- linux-2.6.18/include/xen/interface/version.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/version.h 2007-12-23 11:15:46.578593257 +0100
+--- linux-2.6.18.8/include/xen/interface/version.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/version.h 2008-05-19 00:34:28.493246989 +0300
@@ -0,0 +1,91 @@
+/******************************************************************************
+ * version.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xencomm.h linux-2.6.18-xen.hg/include/xen/interface/xencomm.h
---- linux-2.6.18/include/xen/interface/xencomm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xencomm.h 2007-12-23 11:15:46.581926766 +0100
-@@ -0,0 +1,41 @@
-+/*
-+ * Permission is hereby granted, free of charge, to any person obtaining a copy
-+ * of this software and associated documentation files (the "Software"), to
-+ * deal in the Software without restriction, including without limitation the
-+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
-+ * sell copies of the Software, and to permit persons to whom the Software is
-+ * furnished to do so, subject to the following conditions:
-+ *
-+ * The above copyright notice and this permission notice shall be included in
-+ * all copies or substantial portions of the Software.
-+ *
-+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
-+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
-+ * DEALINGS IN THE SOFTWARE.
-+ *
-+ * Copyright (C) IBM Corp. 2006
-+ */
-+
-+#ifndef _XEN_XENCOMM_H_
-+#define _XEN_XENCOMM_H_
-+
-+/* A xencomm descriptor is a scatter/gather list containing physical
-+ * addresses corresponding to a virtually contiguous memory area. The
-+ * hypervisor translates these physical addresses to machine addresses to copy
-+ * to and from the virtually contiguous area.
-+ */
-+
-+#define XENCOMM_MAGIC 0x58434F4D /* 'XCOM' */
-+#define XENCOMM_INVALID (~0UL)
-+
-+struct xencomm_desc {
-+ uint32_t magic;
-+ uint32_t nr_addrs; /* the number of entries in address[] */
-+ uint64_t address[0];
-+};
-+
-+#endif /* _XEN_XENCOMM_H_ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xen-compat.h linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h
---- linux-2.6.18/include/xen/interface/xen-compat.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h 2007-12-23 11:15:46.578593257 +0100
-@@ -0,0 +1,51 @@
+--- linux-2.6.18.8/include/xen/interface/xen-compat.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h 2008-05-19 00:34:28.497247219 +0300
+@@ -0,0 +1,44 @@
+/******************************************************************************
+ * xen-compat.h
+ *
+#ifndef __XEN_PUBLIC_XEN_COMPAT_H__
+#define __XEN_PUBLIC_XEN_COMPAT_H__
+
-+#define __XEN_LATEST_INTERFACE_VERSION__ 0x00030207
++#define __XEN_LATEST_INTERFACE_VERSION__ 0x00030208
+
+#if defined(__XEN__) || defined(__XEN_TOOLS__)
+/* Xen is built with matching headers and implements the latest interface. */
+#error "These header files do not support the requested interface version."
+#endif
+
-+/* Fields defined as a Xen guest handle since 0x00030205. */
-+#if __XEN_INTERFACE_VERSION__ >= 0x00030205
-+#define XEN_GUEST_HANDLE_00030205(type) XEN_GUEST_HANDLE(type)
-+#else
-+#define XEN_GUEST_HANDLE_00030205(type) type *
-+#endif
-+
+#endif /* __XEN_PUBLIC_XEN_COMPAT_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xen.h linux-2.6.18-xen.hg/include/xen/interface/xen.h
---- linux-2.6.18/include/xen/interface/xen.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xen.h 2007-12-23 11:15:46.581926766 +0100
-@@ -0,0 +1,618 @@
+--- linux-2.6.18.8/include/xen/interface/xen.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xen.h 2008-05-19 00:34:28.553250447 +0300
+@@ -0,0 +1,638 @@
+/******************************************************************************
+ * xen.h
+ *
+#include "arch-x86/xen.h"
+#elif defined(__ia64__)
+#include "arch-ia64.h"
-+#elif defined(__powerpc__)
-+#include "arch-powerpc.h"
+#else
+#error "Unsupported architecture"
+#endif
+
++#ifndef __ASSEMBLY__
++/* Guest handles for primitive C types. */
++DEFINE_XEN_GUEST_HANDLE(char);
++__DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
++DEFINE_XEN_GUEST_HANDLE(int);
++__DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
++DEFINE_XEN_GUEST_HANDLE(long);
++__DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
++DEFINE_XEN_GUEST_HANDLE(void);
++
++DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
++#endif
++
+/*
+ * HYPERCALLS
+ */
+ * ptr[:2] -- Machine address within the frame whose mapping to modify.
+ * The frame must belong to the FD, if one is specified.
+ * val -- Value to write into the mapping entry.
++ *
++ * ptr[1:0] == MMU_PT_UPDATE_PRESERVE_AD:
++ * As MMU_NORMAL_PT_UPDATE above, but A/D bits currently in the PTE are ORed
++ * with those in @val.
+ */
-+#define MMU_NORMAL_PT_UPDATE 0 /* checked '*ptr = val'. ptr is MA. */
-+#define MMU_MACHPHYS_UPDATE 1 /* ptr = MA of frame to modify entry for */
++#define MMU_NORMAL_PT_UPDATE 0 /* checked '*ptr = val'. ptr is MA. */
++#define MMU_MACHPHYS_UPDATE 1 /* ptr = MA of frame to modify entry for */
++#define MMU_PT_UPDATE_PRESERVE_AD 2 /* atomically: *ptr = val | (*ptr&(A|D)) */
+
+/*
+ * MMU EXTENDED OPERATIONS
+ /* SET_LDT */
+ unsigned int nr_ents;
+ /* TLB_FLUSH_MULTI, INVLPG_MULTI */
-+ XEN_GUEST_HANDLE_00030205(void) vcpumask;
++#if __XEN_INTERFACE_VERSION__ >= 0x00030205
++ XEN_GUEST_HANDLE(void) vcpumask;
++#else
++ void *vcpumask;
++#endif
+ } arg2;
+};
+typedef struct mmuext_op mmuext_op_t;
+#define __mk_unsigned_long(x) x ## UL
+#define mk_unsigned_long(x) __mk_unsigned_long(x)
+
-+DEFINE_XEN_GUEST_HANDLE(uint8_t);
-+DEFINE_XEN_GUEST_HANDLE(uint16_t);
-+DEFINE_XEN_GUEST_HANDLE(uint32_t);
-+DEFINE_XEN_GUEST_HANDLE(uint64_t);
++__DEFINE_XEN_GUEST_HANDLE(uint8, uint8_t);
++__DEFINE_XEN_GUEST_HANDLE(uint16, uint16_t);
++__DEFINE_XEN_GUEST_HANDLE(uint32, uint32_t);
++__DEFINE_XEN_GUEST_HANDLE(uint64, uint64_t);
+
+#else /* __ASSEMBLY__ */
+
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xenoprof.h linux-2.6.18-xen.hg/include/xen/interface/xenoprof.h
---- linux-2.6.18/include/xen/interface/xenoprof.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xenoprof.h 2007-12-23 11:15:46.581926766 +0100
+--- linux-2.6.18.8/include/xen/interface/xencomm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xencomm.h 2008-05-19 00:34:28.553250447 +0300
+@@ -0,0 +1,41 @@
++/*
++ * Permission is hereby granted, free of charge, to any person obtaining a copy
++ * of this software and associated documentation files (the "Software"), to
++ * deal in the Software without restriction, including without limitation the
++ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
++ * sell copies of the Software, and to permit persons to whom the Software is
++ * furnished to do so, subject to the following conditions:
++ *
++ * The above copyright notice and this permission notice shall be included in
++ * all copies or substantial portions of the Software.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
++ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
++ * DEALINGS IN THE SOFTWARE.
++ *
++ * Copyright (C) IBM Corp. 2006
++ */
++
++#ifndef _XEN_XENCOMM_H_
++#define _XEN_XENCOMM_H_
++
++/* A xencomm descriptor is a scatter/gather list containing physical
++ * addresses corresponding to a virtually contiguous memory area. The
++ * hypervisor translates these physical addresses to machine addresses to copy
++ * to and from the virtually contiguous area.
++ */
++
++#define XENCOMM_MAGIC 0x58434F4D /* 'XCOM' */
++#define XENCOMM_INVALID (~0UL)
++
++struct xencomm_desc {
++ uint32_t magic;
++ uint32_t nr_addrs; /* the number of entries in address[] */
++ uint64_t address[0];
++};
++
++#endif /* _XEN_XENCOMM_H_ */
+--- linux-2.6.18.8/include/xen/interface/xenoprof.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xenoprof.h 2008-05-19 00:34:28.553250447 +0300
@@ -0,0 +1,138 @@
+/******************************************************************************
+ * xenoprof.h
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xsm/acm.h linux-2.6.18-xen.hg/include/xen/interface/xsm/acm.h
---- linux-2.6.18/include/xen/interface/xsm/acm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xsm/acm.h 2007-12-23 11:15:46.581926766 +0100
-@@ -0,0 +1,229 @@
+--- linux-2.6.18.8/include/xen/interface/xsm/acm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xsm/acm.h 2008-05-19 00:34:28.557250678 +0300
+@@ -0,0 +1,235 @@
+/*
+ * acm.h: Xen access control module interface defintions
+ *
+ * whenever the interpretation of the related
+ * policy's data structure changes
+ */
-+#define ACM_POLICY_VERSION 3
++#define ACM_POLICY_VERSION 4
+#define ACM_CHWALL_VERSION 1
+#define ACM_STE_VERSION 1
+
+#define ACMHOOK_none 0
+#define ACMHOOK_sharing 1
+#define ACMHOOK_authorization 2
++#define ACMHOOK_conflictset 3
+
+/* -------security policy relevant type definitions-------- */
+
+/* high-16 = version, low-16 = check magic */
+#define ACM_MAGIC 0x0001debc
+
++/* size of the SHA1 hash identifying the XML policy from which the
++ binary policy was created */
++#define ACM_SHA1_HASH_SIZE 20
++
+/* each offset in bytes from start of the struct they
+ * are part of */
+
+ uint32_t secondary_policy_code;
+ uint32_t secondary_buffer_offset;
+ struct acm_policy_version xml_pol_version; /* add in V3 */
++ uint8_t xml_policy_hash[ACM_SHA1_HASH_SIZE]; /* added in V4 */
+};
+
+
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xsm/acm_ops.h linux-2.6.18-xen.hg/include/xen/interface/xsm/acm_ops.h
---- linux-2.6.18/include/xen/interface/xsm/acm_ops.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xsm/acm_ops.h 2007-12-23 11:15:46.581926766 +0100
+--- linux-2.6.18.8/include/xen/interface/xsm/acm_ops.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xsm/acm_ops.h 2008-05-19 00:34:28.557250678 +0300
@@ -0,0 +1,159 @@
+/*
+ * acm_ops.h: Xen access control module hypervisor commands
+ * indent-tabs-mode: nil
+ * End:
+ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/interface/xsm/flask_op.h linux-2.6.18-xen.hg/include/xen/interface/xsm/flask_op.h
---- linux-2.6.18/include/xen/interface/xsm/flask_op.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/interface/xsm/flask_op.h 2007-12-23 11:15:46.581926766 +0100
+--- linux-2.6.18.8/include/xen/interface/xsm/flask_op.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/interface/xsm/flask_op.h 2008-05-19 00:34:28.557250678 +0300
@@ -0,0 +1,43 @@
+/*
+ * This file contains the flask_op hypercall commands and definitions.
+DEFINE_XEN_GUEST_HANDLE(flask_op_t);
+
+#endif
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/pcifront.h linux-2.6.18-xen.hg/include/xen/pcifront.h
---- linux-2.6.18/include/xen/pcifront.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/pcifront.h 2007-12-23 11:15:46.585260277 +0100
+--- linux-2.6.18.8/include/xen/pcifront.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/pcifront.h 2008-05-19 00:34:28.557250678 +0300
@@ -0,0 +1,83 @@
+/*
+ * PCI Frontend - arch-dependendent declarations
+#endif /* __KERNEL__ */
+
+#endif /* __XEN_ASM_PCIFRONT_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/public/evtchn.h linux-2.6.18-xen.hg/include/xen/public/evtchn.h
---- linux-2.6.18/include/xen/public/evtchn.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/public/evtchn.h 2007-12-23 11:15:46.585260277 +0100
+--- linux-2.6.18.8/include/xen/public/evtchn.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/public/evtchn.h 2008-05-19 00:34:28.565251139 +0300
@@ -0,0 +1,88 @@
+/******************************************************************************
+ * evtchn.h
+ _IOC(_IOC_NONE, 'E', 5, 0)
+
+#endif /* __LINUX_PUBLIC_EVTCHN_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/public/gntdev.h linux-2.6.18-xen.hg/include/xen/public/gntdev.h
---- linux-2.6.18/include/xen/public/gntdev.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/public/gntdev.h 2007-12-23 11:15:46.585260277 +0100
-@@ -0,0 +1,105 @@
+--- linux-2.6.18.8/include/xen/public/gntdev.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/public/gntdev.h 2008-05-19 00:34:28.569251370 +0300
+@@ -0,0 +1,119 @@
+/******************************************************************************
+ * gntdev.h
+ *
+ uint32_t pad;
+};
+
++/*
++ * Sets the maximum number of grants that may mapped at once by this gntdev
++ * instance.
++ *
++ * N.B. This must be called before any other ioctl is performed on the device.
++ */
++#define IOCTL_GNTDEV_SET_MAX_GRANTS \
++_IOC(_IOC_NONE, 'G', 3, sizeof(struct ioctl_gntdev_set_max_grants))
++struct ioctl_gntdev_set_max_grants {
++ /* IN parameter */
++ /* The maximum number of grants that may be mapped at once. */
++ uint32_t count;
++};
++
+#endif /* __LINUX_PUBLIC_GNTDEV_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/public/privcmd.h linux-2.6.18-xen.hg/include/xen/public/privcmd.h
---- linux-2.6.18/include/xen/public/privcmd.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/public/privcmd.h 2007-12-23 11:15:46.585260277 +0100
+--- linux-2.6.18.8/include/xen/public/privcmd.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/public/privcmd.h 2008-05-19 00:34:28.569251370 +0300
@@ -0,0 +1,79 @@
+/******************************************************************************
+ * privcmd.h
+ _IOC(_IOC_NONE, 'P', 3, sizeof(privcmd_mmapbatch_t))
+
+#endif /* __LINUX_PUBLIC_PRIVCMD_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/xenbus.h linux-2.6.18-xen.hg/include/xen/xenbus.h
---- linux-2.6.18/include/xen/xenbus.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/xenbus.h 2007-12-23 11:15:46.585260277 +0100
+--- linux-2.6.18.8/include/xen/xen_proc.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/xen_proc.h 2008-05-19 00:34:28.569251370 +0300
+@@ -0,0 +1,12 @@
++
++#ifndef __ASM_XEN_PROC_H__
++#define __ASM_XEN_PROC_H__
++
++#include <linux/proc_fs.h>
++
++extern struct proc_dir_entry *create_xen_proc_entry(
++ const char *name, mode_t mode);
++extern void remove_xen_proc_entry(
++ const char *name);
++
++#endif /* __ASM_XEN_PROC_H__ */
+--- linux-2.6.18.8/include/xen/xenbus.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/xenbus.h 2008-05-19 00:34:28.569251370 +0300
@@ -0,0 +1,307 @@
+/******************************************************************************
+ * xenbus.h
+int xenbus_for_each_frontend(void *arg, int (*fn)(struct device *, void *));
+
+#endif /* _XEN_XENBUS_H */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/xencomm.h linux-2.6.18-xen.hg/include/xen/xencomm.h
---- linux-2.6.18/include/xen/xencomm.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/xencomm.h 2007-12-23 11:15:46.588593784 +0100
+--- linux-2.6.18.8/include/xen/xencomm.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/xencomm.h 2008-05-19 00:34:28.569251370 +0300
@@ -0,0 +1,77 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+#define xen_guest_handle(hnd) ((hnd).p)
+
+#endif /* _LINUX_XENCOMM_H_ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/xencons.h linux-2.6.18-xen.hg/include/xen/xencons.h
---- linux-2.6.18/include/xen/xencons.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/xencons.h 2007-12-23 11:15:46.588593784 +0100
+--- linux-2.6.18.8/include/xen/xencons.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/xencons.h 2008-05-19 00:34:28.573251600 +0300
@@ -0,0 +1,17 @@
+#ifndef __ASM_XENCONS_H__
+#define __ASM_XENCONS_H__
+int xencons_ring_send(const char *data, unsigned len);
+
+#endif /* __ASM_XENCONS_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/xenoprof.h linux-2.6.18-xen.hg/include/xen/xenoprof.h
---- linux-2.6.18/include/xen/xenoprof.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/xenoprof.h 2007-12-23 11:15:46.588593784 +0100
+--- linux-2.6.18.8/include/xen/xenoprof.h 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/include/xen/xenoprof.h 2008-05-19 00:34:28.573251600 +0300
@@ -0,0 +1,42 @@
+/******************************************************************************
+ * xen/xenoprof.h
+
+#endif /* CONFIG_XEN */
+#endif /* __XEN_XENOPROF_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/include/xen/xen_proc.h linux-2.6.18-xen.hg/include/xen/xen_proc.h
---- linux-2.6.18/include/xen/xen_proc.h 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/include/xen/xen_proc.h 2007-12-23 11:15:46.585260277 +0100
-@@ -0,0 +1,12 @@
-+
-+#ifndef __ASM_XEN_PROC_H__
-+#define __ASM_XEN_PROC_H__
-+
-+#include <linux/proc_fs.h>
-+
-+extern struct proc_dir_entry *create_xen_proc_entry(
-+ const char *name, mode_t mode);
-+extern void remove_xen_proc_entry(
-+ const char *name);
-+
-+#endif /* __ASM_XEN_PROC_H__ */
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/cpu.c linux-2.6.18-xen.hg/kernel/cpu.c
---- linux-2.6.18/kernel/cpu.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/cpu.c 2007-12-23 11:15:46.618595361 +0100
+--- linux-2.6.18.8/kernel/Kconfig.preempt 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/Kconfig.preempt 2008-05-19 00:34:29.029277887 +0300
+@@ -35,6 +35,7 @@
+
+ config PREEMPT
+ bool "Preemptible Kernel (Low-Latency Desktop)"
++ depends on !XEN
+ help
+ This option reduces the latency of the kernel by making
+ all kernel code (that is not executing in a critical section)
+--- linux-2.6.18.8/kernel/cpu.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/cpu.c 2008-05-19 00:34:29.037278348 +0300
@@ -48,7 +48,10 @@
void unlock_cpu_hotplug(void)
if (recursive_depth) {
recursive_depth--;
return;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/fork.c linux-2.6.18-xen.hg/kernel/fork.c
---- linux-2.6.18/kernel/fork.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/fork.c 2007-12-23 11:15:46.625262374 +0100
+--- linux-2.6.18.8/kernel/fork.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/fork.c 2008-05-19 00:34:29.049279040 +0300
@@ -276,6 +276,9 @@
if (retval)
goto out;
retval = 0;
out:
up_write(&mm->mmap_sem);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/irq/spurious.c linux-2.6.18-xen.hg/kernel/irq/spurious.c
---- linux-2.6.18/kernel/irq/spurious.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/irq/spurious.c 2007-12-23 11:15:46.631929392 +0100
+--- linux-2.6.18.8/kernel/hrtimer.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/hrtimer.c 2008-05-19 00:34:29.049279040 +0300
+@@ -59,7 +59,7 @@
+ *
+ * returns the time in ktime_t format
+ */
+-static ktime_t ktime_get_real(void)
++ktime_t ktime_get_real(void)
+ {
+ struct timespec now;
+
+--- linux-2.6.18.8/kernel/irq/spurious.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/irq/spurious.c 2008-05-19 00:34:29.085281115 +0300
@@ -139,6 +139,7 @@
irqreturn_t action_ret, struct pt_regs *regs)
{
desc->irqs_unhandled++;
if (unlikely(action_ret != IRQ_NONE))
report_bad_irq(irq, desc, action_ret);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/Kconfig.preempt linux-2.6.18-xen.hg/kernel/Kconfig.preempt
---- linux-2.6.18/kernel/Kconfig.preempt 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/Kconfig.preempt 2007-12-23 11:15:46.611928341 +0100
-@@ -35,6 +35,7 @@
-
- config PREEMPT
- bool "Preemptible Kernel (Low-Latency Desktop)"
-+ depends on !XEN
- help
- This option reduces the latency of the kernel by making
- all kernel code (that is not executing in a critical section)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/kexec.c linux-2.6.18-xen.hg/kernel/kexec.c
---- linux-2.6.18/kernel/kexec.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/kexec.c 2007-12-23 11:15:46.631929392 +0100
+--- linux-2.6.18.8/kernel/kexec.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/kexec.c 2008-05-19 00:34:29.113282729 +0300
@@ -330,13 +330,26 @@
return 0;
}
/* Take the kexec_lock here to prevent sys_kexec_load
* running on one cpu from replacing the crash kernel
* we are using after a panic on a different cpu.
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/resource.c linux-2.6.18-xen.hg/kernel/resource.c
---- linux-2.6.18/kernel/resource.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/resource.c 2007-12-23 11:15:46.655263952 +0100
+--- linux-2.6.18.8/kernel/resource.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/resource.c 2008-05-19 00:34:29.161285496 +0300
@@ -36,6 +36,16 @@
};
EXPORT_SYMBOL(iomem_resource);
return 0;
}
__initcall(ioresources_init);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/softlockup.c linux-2.6.18-xen.hg/kernel/softlockup.c
---- linux-2.6.18/kernel/softlockup.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/softlockup.c 2007-12-23 11:15:46.665264476 +0100
+--- linux-2.6.18.8/kernel/softlockup.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/softlockup.c 2008-05-19 00:34:29.177286418 +0300
@@ -40,6 +40,19 @@
}
EXPORT_SYMBOL(touch_softlockup_watchdog);
/*
* This callback runs from the timer interrupt, and checks
* whether the watchdog thread has hung or not:
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/sysctl.c linux-2.6.18-xen.hg/kernel/sysctl.c
---- linux-2.6.18/kernel/sysctl.c 2007-12-23 11:26:51.296802316 +0100
-+++ linux-2.6.18-xen.hg/kernel/sysctl.c 2007-12-23 11:15:46.668597986 +0100
+--- linux-2.6.18.8/kernel/sysctl.c 2008-05-19 00:42:34.409259256 +0300
++++ linux-2.6.18-xen.hg/kernel/sysctl.c 2008-05-19 00:34:29.185286879 +0300
@@ -661,7 +661,7 @@
.proc_handler = &proc_dointvec,
},
{
.ctl_name = KERN_ACPI_VIDEO_FLAGS,
.procname = "acpi_video_flags",
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/kernel/timer.c linux-2.6.18-xen.hg/kernel/timer.c
---- linux-2.6.18/kernel/timer.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/kernel/timer.c 2007-12-23 11:15:46.671931494 +0100
+--- linux-2.6.18.8/kernel/timer.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/kernel/timer.c 2008-05-19 00:34:29.189287110 +0300
@@ -485,7 +485,9 @@
if (hr_expires < 3)
return hr_expires + jiffies;
base = __get_cpu_var(tvec_bases);
spin_lock(&base->lock);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/lib/Makefile linux-2.6.18-xen.hg/lib/Makefile
---- linux-2.6.18/lib/Makefile 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/lib/Makefile 2007-12-23 11:15:46.678598509 +0100
+--- linux-2.6.18.8/lib/Makefile 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/lib/Makefile 2008-05-19 00:34:29.197287571 +0300
@@ -52,6 +52,9 @@
obj-$(CONFIG_AUDIT_GENERIC) += audit.o
hostprogs-y := gen_crc32table
clean-files := crc32table.h
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/lib/swiotlb-xen.c linux-2.6.18-xen.hg/lib/swiotlb-xen.c
---- linux-2.6.18/lib/swiotlb-xen.c 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/lib/swiotlb-xen.c 2007-12-23 11:15:46.731934648 +0100
+--- linux-2.6.18.8/lib/swiotlb-xen.c 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/lib/swiotlb-xen.c 2008-05-19 00:34:29.249290569 +0300
@@ -0,0 +1,745 @@
+/*
+ * Dynamic DMA mapping support.
+EXPORT_SYMBOL(swiotlb_sync_sg_for_device);
+EXPORT_SYMBOL(swiotlb_dma_mapping_error);
+EXPORT_SYMBOL(swiotlb_dma_supported);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/mm/highmem.c linux-2.6.18-xen.hg/mm/highmem.c
---- linux-2.6.18/mm/highmem.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/mm/highmem.c 2007-12-23 11:15:46.748602187 +0100
+--- linux-2.6.18.8/mm/highmem.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/mm/highmem.c 2008-05-19 00:34:29.281292413 +0300
@@ -142,6 +142,17 @@
return vaddr;
}
void fastcall *kmap_high(struct page *page)
{
unsigned long vaddr;
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/mm/memory.c linux-2.6.18-xen.hg/mm/memory.c
---- linux-2.6.18/mm/memory.c 2007-12-23 11:26:51.296802316 +0100
-+++ linux-2.6.18-xen.hg/mm/memory.c 2007-12-23 11:15:46.751935698 +0100
+@@ -457,6 +468,12 @@
+ mempool_t *pool;
+
+ /*
++ * Data-less bio, nothing to bounce
++ */
++ if (bio_empty_barrier(*bio_orig))
++ return;
++
++ /*
+ * for non-isa bounce case, just check if the bounce pfn is equal
+ * to or bigger than the highest pfn in the system -- in that case,
+ * don't waste time iterating over bio segments
+--- linux-2.6.18.8/mm/memory.c 2008-05-19 00:42:34.409259256 +0300
++++ linux-2.6.18-xen.hg/mm/memory.c 2008-05-19 00:34:29.289292874 +0300
@@ -396,6 +396,12 @@
return NULL;
}
/*
* handle_pte_fault chooses page fault handler according to an entry
* which was read non-atomically. Before making any commitment, on
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/mm/mmap.c linux-2.6.18-xen.hg/mm/mmap.c
---- linux-2.6.18/mm/mmap.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/mm/mmap.c 2007-12-23 11:15:46.755269206 +0100
+--- linux-2.6.18.8/mm/mmap.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/mm/mmap.c 2008-05-19 00:34:29.301293566 +0300
@@ -1963,6 +1963,10 @@
unsigned long nr_accounted = 0;
unsigned long end;
lru_add_drain();
flush_cache_mm(mm);
tlb = tlb_gather_mmu(mm, 1);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/mm/page_alloc.c linux-2.6.18-xen.hg/mm/page_alloc.c
---- linux-2.6.18/mm/page_alloc.c 2007-12-23 11:26:51.300135824 +0100
-+++ linux-2.6.18-xen.hg/mm/page_alloc.c 2007-12-23 11:15:46.761936217 +0100
+--- linux-2.6.18.8/mm/mprotect.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/mm/mprotect.c 2008-05-19 00:34:29.305293797 +0300
+@@ -76,6 +76,8 @@
+ next = pmd_addr_end(addr, end);
+ if (pmd_none_or_clear_bad(pmd))
+ continue;
++ if (arch_change_pte_range(mm, pmd, addr, next, newprot))
++ continue;
+ change_pte_range(mm, pmd, addr, next, newprot);
+ } while (pmd++, addr = next, addr != end);
+ }
+--- linux-2.6.18.8/mm/page_alloc.c 2008-05-19 00:42:34.413259487 +0300
++++ linux-2.6.18-xen.hg/mm/page_alloc.c 2008-05-19 00:34:29.317294489 +0300
@@ -154,7 +154,11 @@
1 << PG_slab |
1 << PG_swapcache |
arch_free_page(page, 0);
if (PageAnon(page))
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/bridge/br_netfilter.c linux-2.6.18-xen.hg/net/bridge/br_netfilter.c
---- linux-2.6.18/net/bridge/br_netfilter.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/net/bridge/br_netfilter.c 2007-12-23 11:15:47.175291252 +0100
+--- linux-2.6.18.8/net/bridge/br_netfilter.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/net/bridge/br_netfilter.c 2008-05-19 00:34:29.813323081 +0300
@@ -127,10 +127,10 @@
static inline void nf_bridge_save_header(struct sk_buff *skb)
memcpy(skb->nf_bridge->data, skb->data - header_size, header_size);
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/core/dev.c linux-2.6.18-xen.hg/net/core/dev.c
---- linux-2.6.18/net/core/dev.c 2007-12-23 11:26:51.306802839 +0100
-+++ linux-2.6.18-xen.hg/net/core/dev.c 2007-12-23 11:15:47.228627387 +0100
+--- linux-2.6.18.8/net/bridge/netfilter/ebtables.c 2008-05-19 00:42:34.441261101 +0300
++++ linux-2.6.18-xen.hg/net/bridge/netfilter/ebtables.c 2008-05-19 00:34:29.889327462 +0300
+@@ -597,7 +597,7 @@
+ struct ebt_entry_target *t;
+ struct ebt_target *target;
+ unsigned int i, j, hook = 0, hookmask = 0;
+- size_t gap = e->next_offset - e->target_offset;
++ size_t gap;
+ int ret;
+
+ /* don't mess with the struct ebt_entries */
+@@ -647,6 +647,7 @@
+ if (ret != 0)
+ goto cleanup_watchers;
+ t = (struct ebt_entry_target *)(((char *)e) + e->target_offset);
++ gap = e->next_offset - e->target_offset;
+ target = find_target_lock(t->u.name, &ret, &ebt_mutex);
+ if (!target)
+ goto cleanup_watchers;
+--- linux-2.6.18.8/net/core/dev.c 2008-05-19 00:42:34.441261101 +0300
++++ linux-2.6.18-xen.hg/net/core/dev.c 2008-05-19 00:34:29.897327923 +0300
@@ -113,11 +113,18 @@
#include <linux/wireless.h>
#include <net/iw_handler.h>
#if defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE)
EXPORT_SYMBOL(br_handle_frame_hook);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/core/skbuff.c linux-2.6.18-xen.hg/net/core/skbuff.c
---- linux-2.6.18/net/core/skbuff.c 2007-12-23 11:26:51.306802839 +0100
-+++ linux-2.6.18-xen.hg/net/core/skbuff.c 2007-12-23 11:15:47.238627923 +0100
+--- linux-2.6.18.8/net/core/neighbour.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/net/core/neighbour.c 2008-05-19 00:34:29.905328384 +0300
+@@ -2679,7 +2679,6 @@
+ EXPORT_SYMBOL(neigh_table_init);
+ EXPORT_SYMBOL(neigh_table_init_no_netlink);
+ EXPORT_SYMBOL(neigh_update);
+-EXPORT_SYMBOL(neigh_update_hhs);
+ EXPORT_SYMBOL(pneigh_enqueue);
+ EXPORT_SYMBOL(pneigh_lookup);
+ EXPORT_SYMBOL(neightbl_dump_info);
+--- linux-2.6.18.8/net/core/skbuff.c 2008-05-19 00:42:34.441261101 +0300
++++ linux-2.6.18-xen.hg/net/core/skbuff.c 2008-05-19 00:34:29.921329306 +0300
@@ -240,6 +240,7 @@
skb_shinfo(skb)->gso_size = 0;
skb_shinfo(skb)->gso_segs = 0;
C(pkt_type);
C(ip_summed);
C(priority);
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/ipv4/netfilter/ip_nat_proto_tcp.c linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_tcp.c
---- linux-2.6.18/net/ipv4/netfilter/ip_nat_proto_tcp.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_tcp.c 2007-12-23 11:15:47.355300712 +0100
+@@ -2071,7 +2076,6 @@
+ EXPORT_SYMBOL(pskb_expand_head);
+ EXPORT_SYMBOL(skb_checksum);
+ EXPORT_SYMBOL(skb_clone);
+-EXPORT_SYMBOL(skb_clone_fraglist);
+ EXPORT_SYMBOL(skb_copy);
+ EXPORT_SYMBOL(skb_copy_and_csum_bits);
+ EXPORT_SYMBOL(skb_copy_and_csum_dev);
+--- linux-2.6.18.8/net/ipv4/netfilter/ip_nat_proto_tcp.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_tcp.c 2008-05-19 00:34:30.301351212 +0300
@@ -129,6 +129,11 @@
if (hdrsize < sizeof(*hdr))
return 1;
hdr->check = ip_nat_cheat_check(~oldip, newip,
ip_nat_cheat_check(oldport ^ 0xFFFF,
newport,
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/ipv4/netfilter/ip_nat_proto_udp.c linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_udp.c
---- linux-2.6.18/net/ipv4/netfilter/ip_nat_proto_udp.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_udp.c 2007-12-23 11:15:47.355300712 +0100
+--- linux-2.6.18.8/net/ipv4/netfilter/ip_nat_proto_udp.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/net/ipv4/netfilter/ip_nat_proto_udp.c 2008-05-19 00:34:30.301351212 +0300
@@ -113,11 +113,17 @@
newport = tuple->dst.u.udp.port;
portptr = &hdr->dest;
*portptr = newport;
return 1;
}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/ipv4/tcp_input.c linux-2.6.18-xen.hg/net/ipv4/tcp_input.c
---- linux-2.6.18/net/ipv4/tcp_input.c 2007-12-23 11:26:51.333470906 +0100
-+++ linux-2.6.18-xen.hg/net/ipv4/tcp_input.c 2007-12-23 11:15:47.401969829 +0100
+--- linux-2.6.18.8/net/ipv4/tcp_input.c 2008-05-19 00:42:34.461262254 +0300
++++ linux-2.6.18-xen.hg/net/ipv4/tcp_input.c 2008-05-19 00:34:30.357354440 +0300
@@ -127,7 +127,7 @@
/* skb->len may jitter because of SACKs, even if peer
* sends good full-sized frames.
if (len >= icsk->icsk_ack.rcv_mss) {
icsk->icsk_ack.rcv_mss = len;
} else {
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/net/ipv4/xfrm4_output.c linux-2.6.18-xen.hg/net/ipv4/xfrm4_output.c
---- linux-2.6.18/net/ipv4/xfrm4_output.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/net/ipv4/xfrm4_output.c 2007-12-23 11:15:47.411970353 +0100
+--- linux-2.6.18.8/net/ipv4/xfrm4_output.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/net/ipv4/xfrm4_output.c 2008-05-19 00:34:30.373355362 +0300
@@ -18,6 +18,8 @@
#include <net/xfrm.h>
#include <net/icmp.h>
if (skb->ip_summed == CHECKSUM_HW) {
err = skb_checksum_help(skb, 0);
if (err)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/scripts/Makefile.build linux-2.6.18-xen.hg/scripts/Makefile.build
---- linux-2.6.18/scripts/Makefile.build 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/scripts/Makefile.build 2007-12-23 11:15:48.838711942 +0100
+--- linux-2.6.18.8/scripts/Makefile.build 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/scripts/Makefile.build 2008-05-19 00:34:31.393414161 +0300
@@ -68,6 +68,20 @@
$(warning kbuild: Makefile.build is included improperly)
endif
# ===========================================================================
ifneq ($(strip $(lib-y) $(lib-m) $(lib-n) $(lib-)),)
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/scripts/Makefile.lib linux-2.6.18-xen.hg/scripts/Makefile.lib
---- linux-2.6.18/scripts/Makefile.lib 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/scripts/Makefile.lib 2007-12-23 11:15:48.842045453 +0100
+--- linux-2.6.18.8/scripts/Makefile.lib 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/scripts/Makefile.lib 2008-05-19 00:34:31.397414391 +0300
@@ -13,6 +13,12 @@
lib-y := $(filter-out $(obj-y), $(sort $(lib-y) $(lib-m)))
# Handle objects in subdirs
# ---------------------------------------------------------------------------
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/scripts/Makefile.xen.awk linux-2.6.18-xen.hg/scripts/Makefile.xen.awk
---- linux-2.6.18/scripts/Makefile.xen.awk 1970-01-01 01:00:00.000000000 +0100
-+++ linux-2.6.18-xen.hg/scripts/Makefile.xen.awk 2007-12-23 11:15:48.842045453 +0100
+--- linux-2.6.18.8/scripts/Makefile.xen.awk 1970-01-01 03:00:00.000000000 +0300
++++ linux-2.6.18-xen.hg/scripts/Makefile.xen.awk 2008-05-19 00:34:31.397414391 +0300
@@ -0,0 +1,34 @@
+BEGIN {
+ is_rule = 0
+ print $0
+ next
+}
-diff -burN --exclude .hg --exclude .hgtags linux-2.6.18/sound/pci/hda/hda_intel.c linux-2.6.18-xen.hg/sound/pci/hda/hda_intel.c
---- linux-2.6.18/sound/pci/hda/hda_intel.c 2006-09-20 05:42:06.000000000 +0200
-+++ linux-2.6.18-xen.hg/sound/pci/hda/hda_intel.c 2007-12-23 11:15:52.298893641 +0100
-@@ -80,6 +80,7 @@
+--- linux-2.6.18.8/sound/pci/hda/hda_intel.c 2006-09-20 06:42:06.000000000 +0300
++++ linux-2.6.18-xen.hg/sound/pci/hda/hda_intel.c 2008-05-19 00:34:34.397587328 +0300
+@@ -80,6 +80,8 @@
"{Intel, ICH7},"
"{Intel, ESB2},"
"{Intel, ICH8},"
+ "{Intel, ICH9},"
++ "{Intel, ICH10},"
"{ATI, SB450},"
"{ATI, SB600},"
"{ATI, RS600},"
-@@ -1634,6 +1635,8 @@
+@@ -1634,6 +1636,10 @@
{ 0x8086, 0x27d8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH7 */
{ 0x8086, 0x269a, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ESB2 */
{ 0x8086, 0x284b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH8 */
+ { 0x8086, 0x293e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH9 */
+ { 0x8086, 0x293f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH9 */
++ { 0x8086, 0x3a3e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH10 */
++ { 0x8086, 0x3a6e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ICH }, /* ICH10 */
{ 0x1002, 0x437b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ATI }, /* ATI SB450 */
{ 0x1002, 0x4383, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ATI }, /* ATI SB600 */
{ 0x1002, 0x793b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, AZX_DRIVER_ATIHDMI }, /* ATI RS600 HDMI */
projects / packages / kernel.git / commitdiff