1 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/Kconfig
2 --- linux-2.6.29.1/arch/alpha/Kconfig 2009-03-24 14:18:07.000000000 +0100
3 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/Kconfig 2009-03-24 14:48:16.000000000 +0100
4 @@ -666,6 +666,8 @@ config DUMMY_CONSOLE
8 +source "kernel/vserver/Kconfig"
10 source "security/Kconfig"
12 source "crypto/Kconfig"
13 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/kernel/entry.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/entry.S
14 --- linux-2.6.29.1/arch/alpha/kernel/entry.S 2009-03-24 14:18:07.000000000 +0100
15 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/entry.S 2009-03-24 14:48:16.000000000 +0100
16 @@ -874,24 +874,15 @@ sys_getxgid:
25 - /* See linux/kernel/timer.c sys_getppid for discussion
27 - ldq $3, TASK_GROUP_LEADER($2)
28 - ldq $4, TASK_REAL_PARENT($3)
29 - ldl $0, TASK_TGID($2)
30 -1: ldl $1, TASK_TGID($4)
34 - ldq $3, TASK_GROUP_LEADER($2)
35 - ldq $4, TASK_REAL_PARENT($3)
48 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/kernel/osf_sys.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/osf_sys.c
49 --- linux-2.6.29.1/arch/alpha/kernel/osf_sys.c 2009-03-24 14:18:07.000000000 +0100
50 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/osf_sys.c 2009-03-24 14:48:16.000000000 +0100
51 @@ -877,7 +877,7 @@ SYSCALL_DEFINE2(osf_gettimeofday, struct
55 - do_gettimeofday(&ktv);
56 + vx_gettimeofday(&ktv);
57 if (put_tv32(tv, &ktv))
60 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/ptrace.c
61 --- linux-2.6.29.1/arch/alpha/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
62 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
64 #include <linux/slab.h>
65 #include <linux/security.h>
66 #include <linux/signal.h>
67 +#include <linux/vs_base.h>
69 #include <asm/uaccess.h>
70 #include <asm/pgtable.h>
71 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/kernel/systbls.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/systbls.S
72 --- linux-2.6.29.1/arch/alpha/kernel/systbls.S 2009-03-24 14:18:08.000000000 +0100
73 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/systbls.S 2009-03-24 14:48:16.000000000 +0100
74 @@ -446,7 +446,7 @@ sys_call_table:
75 .quad sys_stat64 /* 425 */
78 - .quad sys_ni_syscall /* sys_vserver */
79 + .quad sys_vserver /* sys_vserver */
80 .quad sys_ni_syscall /* sys_mbind */
81 .quad sys_ni_syscall /* sys_get_mempolicy */
82 .quad sys_ni_syscall /* sys_set_mempolicy */
83 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/traps.c
84 --- linux-2.6.29.1/arch/alpha/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
85 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
86 @@ -183,7 +183,8 @@ die_if_kernel(char * str, struct pt_regs
88 printk("CPU %d ", hard_smp_processor_id());
90 - printk("%s(%d): %s %ld\n", current->comm, task_pid_nr(current), str, err);
91 + printk("%s(%d[#%u]): %s %ld\n", current->comm,
92 + task_pid_nr(current), current->xid, str, err);
93 dik_show_regs(regs, r9_15);
95 dik_show_trace((unsigned long *)(regs+1));
96 diff -NurpP --minimal linux-2.6.29.1/arch/alpha/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/mm/fault.c
97 --- linux-2.6.29.1/arch/alpha/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
98 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/alpha/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
99 @@ -193,8 +193,8 @@ do_page_fault(unsigned long address, uns
100 down_read(&mm->mmap_sem);
103 - printk(KERN_ALERT "VM: killing process %s(%d)\n",
104 - current->comm, task_pid_nr(current));
105 + printk(KERN_ALERT "VM: killing process %s(%d:#%u)\n",
106 + current->comm, task_pid_nr(current), current->xid);
107 if (!user_mode(regs))
109 do_group_exit(SIGKILL);
110 diff -NurpP --minimal linux-2.6.29.1/arch/arm/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/Kconfig
111 --- linux-2.6.29.1/arch/arm/Kconfig 2009-03-24 14:18:08.000000000 +0100
112 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/Kconfig 2009-03-24 14:48:16.000000000 +0100
113 @@ -1333,6 +1333,8 @@ source "fs/Kconfig"
115 source "arch/arm/Kconfig.debug"
117 +source "kernel/vserver/Kconfig"
119 source "security/Kconfig"
121 source "crypto/Kconfig"
122 diff -NurpP --minimal linux-2.6.29.1/arch/arm/kernel/calls.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/calls.S
123 --- linux-2.6.29.1/arch/arm/kernel/calls.S 2009-03-24 14:18:09.000000000 +0100
124 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/calls.S 2009-02-22 22:54:24.000000000 +0100
126 /* 310 */ CALL(sys_request_key)
128 CALL(ABI(sys_semtimedop, sys_oabi_semtimedop))
129 -/* vserver */ CALL(sys_ni_syscall)
132 /* 315 */ CALL(sys_ioprio_get)
133 CALL(sys_inotify_init)
134 diff -NurpP --minimal linux-2.6.29.1/arch/arm/kernel/process.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/process.c
135 --- linux-2.6.29.1/arch/arm/kernel/process.c 2008-12-25 00:26:37.000000000 +0100
136 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/process.c 2009-02-22 22:54:24.000000000 +0100
137 @@ -262,7 +262,8 @@ void __show_regs(struct pt_regs *regs)
138 void show_regs(struct pt_regs * regs)
141 - printk("Pid: %d, comm: %20s\n", task_pid_nr(current), current->comm);
142 + printk("Pid: %d[#%u], comm: %20s\n",
143 + task_pid_nr(current), current->xid, current->comm);
147 diff -NurpP --minimal linux-2.6.29.1/arch/arm/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/traps.c
148 --- linux-2.6.29.1/arch/arm/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
149 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
150 @@ -214,8 +214,8 @@ static void __die(const char *str, int e
151 str, err, ++die_counter);
154 - printk("Process %s (pid: %d, stack limit = 0x%p)\n",
155 - tsk->comm, task_pid_nr(tsk), thread + 1);
156 + printk("Process %s (pid: %d:#%u, stack limit = 0x%p)\n",
157 + tsk->comm, task_pid_nr(tsk), tsk->xid, thread + 1);
159 if (!user_mode(regs) || in_interrupt()) {
160 dump_mem("Stack: ", regs->ARM_sp,
161 diff -NurpP --minimal linux-2.6.29.1/arch/arm/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/mm/fault.c
162 --- linux-2.6.29.1/arch/arm/mm/fault.c 2009-03-24 14:18:17.000000000 +0100
163 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/arm/mm/fault.c 2009-03-24 14:48:17.000000000 +0100
164 @@ -294,7 +294,8 @@ do_page_fault(unsigned long addr, unsign
165 * happened to us that made us unable to handle
166 * the page fault gracefully.
168 - printk("VM: killing process %s\n", tsk->comm);
169 + printk("VM: killing process %s(%d:#%u)\n",
170 + tsk->comm, task_pid_nr(tsk), tsk->xid);
171 do_group_exit(SIGKILL);
174 diff -NurpP --minimal linux-2.6.29.1/arch/cris/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/cris/Kconfig
175 --- linux-2.6.29.1/arch/cris/Kconfig 2009-03-24 14:18:23.000000000 +0100
176 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/cris/Kconfig 2009-03-24 14:48:19.000000000 +0100
177 @@ -685,6 +685,8 @@ source "drivers/staging/Kconfig"
179 source "arch/cris/Kconfig.debug"
181 +source "kernel/vserver/Kconfig"
183 source "security/Kconfig"
185 source "crypto/Kconfig"
186 diff -NurpP --minimal linux-2.6.29.1/arch/frv/kernel/kernel_thread.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/frv/kernel/kernel_thread.S
187 --- linux-2.6.29.1/arch/frv/kernel/kernel_thread.S 2008-12-25 00:26:37.000000000 +0100
188 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/frv/kernel/kernel_thread.S 2009-02-22 22:54:24.000000000 +0100
189 @@ -37,7 +37,7 @@ kernel_thread:
191 # start by forking the current process, but with shared VM
192 setlos.p #__NR_clone,gr7 ; syscall number
193 - ori gr10,#CLONE_VM,gr8 ; first syscall arg [clone_flags]
194 + ori gr10,#CLONE_KT,gr8 ; first syscall arg [clone_flags]
195 sethi.p #0xe4e4,gr9 ; second syscall arg [newsp]
197 setlos.p #0,gr10 ; third syscall arg [parent_tidptr]
198 diff -NurpP --minimal linux-2.6.29.1/arch/h8300/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/h8300/Kconfig
199 --- linux-2.6.29.1/arch/h8300/Kconfig 2009-03-24 14:18:24.000000000 +0100
200 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/h8300/Kconfig 2009-03-24 14:48:19.000000000 +0100
201 @@ -226,6 +226,8 @@ source "fs/Kconfig"
203 source "arch/h8300/Kconfig.debug"
205 +source "kernel/vserver/Kconfig"
207 source "security/Kconfig"
209 source "crypto/Kconfig"
210 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/ia32/ia32_entry.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/ia32/ia32_entry.S
211 --- linux-2.6.29.1/arch/ia64/ia32/ia32_entry.S 2009-03-24 14:18:24.000000000 +0100
212 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/ia32/ia32_entry.S 2009-02-22 22:54:24.000000000 +0100
213 @@ -451,7 +451,7 @@ ia32_syscall_table:
214 data8 sys_tgkill /* 270 */
215 data8 compat_sys_utimes
216 data8 sys32_fadvise64_64
217 - data8 sys_ni_syscall
218 + data8 sys32_vserver
220 data8 sys_ni_syscall /* 275 */
222 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/Kconfig
223 --- linux-2.6.29.1/arch/ia64/Kconfig 2009-03-24 14:18:24.000000000 +0100
224 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/Kconfig 2009-03-24 14:48:19.000000000 +0100
225 @@ -672,6 +672,8 @@ source "fs/Kconfig"
227 source "arch/ia64/Kconfig.debug"
229 +source "kernel/vserver/Kconfig"
231 source "security/Kconfig"
233 source "crypto/Kconfig"
234 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/kernel/entry.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/entry.S
235 --- linux-2.6.29.1/arch/ia64/kernel/entry.S 2009-03-24 14:18:25.000000000 +0100
236 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/entry.S 2009-02-22 22:54:24.000000000 +0100
237 @@ -1653,7 +1653,7 @@ sys_call_table:
239 data8 sys_mq_getsetattr
241 - data8 sys_ni_syscall // reserved for vserver
243 data8 sys_waitid // 1270
245 data8 sys_request_key
246 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/kernel/perfmon.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/perfmon.c
247 --- linux-2.6.29.1/arch/ia64/kernel/perfmon.c 2009-03-24 14:18:25.000000000 +0100
248 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/perfmon.c 2009-03-24 14:48:19.000000000 +0100
250 #include <linux/rcupdate.h>
251 #include <linux/completion.h>
252 #include <linux/tracehook.h>
253 +#include <linux/vs_memory.h>
255 #include <asm/errno.h>
256 #include <asm/intrinsics.h>
257 @@ -2372,7 +2373,7 @@ pfm_smpl_buffer_alloc(struct task_struct
259 insert_vm_struct(mm, vma);
261 - mm->total_vm += size >> PAGE_SHIFT;
262 + vx_vmpages_add(mm, size >> PAGE_SHIFT);
263 vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
265 up_write(&task->mm->mmap_sem);
266 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/kernel/process.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/process.c
267 --- linux-2.6.29.1/arch/ia64/kernel/process.c 2008-12-25 00:26:37.000000000 +0100
268 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/process.c 2009-02-22 22:54:24.000000000 +0100
269 @@ -110,8 +110,8 @@ show_regs (struct pt_regs *regs)
270 unsigned long ip = regs->cr_iip + ia64_psr(regs)->ri;
273 - printk("\nPid: %d, CPU %d, comm: %20s\n", task_pid_nr(current),
274 - smp_processor_id(), current->comm);
275 + printk("\nPid: %d[#%u], CPU %d, comm: %20s\n", task_pid_nr(current),
276 + current->xid, smp_processor_id(), current->comm);
277 printk("psr : %016lx ifs : %016lx ip : [<%016lx>] %s (%s)\n",
278 regs->cr_ipsr, regs->cr_ifs, ip, print_tainted(),
279 init_utsname()->release);
280 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/ptrace.c
281 --- linux-2.6.29.1/arch/ia64/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
282 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
284 #include <linux/regset.h>
285 #include <linux/elf.h>
286 #include <linux/tracehook.h>
287 +#include <linux/vs_base.h>
289 #include <asm/pgtable.h>
290 #include <asm/processor.h>
291 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/traps.c
292 --- linux-2.6.29.1/arch/ia64/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
293 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
294 @@ -60,8 +60,9 @@ die (const char *str, struct pt_regs *re
297 if (++die.lock_owner_depth < 3) {
298 - printk("%s[%d]: %s %ld [%d]\n",
299 - current->comm, task_pid_nr(current), str, err, ++die_counter);
300 + printk("%s[%d[#%u]]: %s %ld [%d]\n",
301 + current->comm, task_pid_nr(current), current->xid,
302 + str, err, ++die_counter);
303 if (notify_die(DIE_OOPS, str, regs, err, 255, SIGSEGV)
306 @@ -324,8 +325,9 @@ handle_fpu_swa (int fp_fault, struct pt_
307 if ((last.count & 15) < 5 && (ia64_fetchadd(1, &last.count, acq) & 15) < 5) {
308 last.time = current_jiffies + 5 * HZ;
310 - "%s(%d): floating-point assist fault at ip %016lx, isr %016lx\n",
311 - current->comm, task_pid_nr(current), regs->cr_iip + ia64_psr(regs)->ri, isr);
312 + "%s(%d[#%u]): floating-point assist fault at ip %016lx, isr %016lx\n",
313 + current->comm, task_pid_nr(current), current->xid,
314 + regs->cr_iip + ia64_psr(regs)->ri, isr);
318 diff -NurpP --minimal linux-2.6.29.1/arch/ia64/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/mm/fault.c
319 --- linux-2.6.29.1/arch/ia64/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
320 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/ia64/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
322 #include <linux/interrupt.h>
323 #include <linux/kprobes.h>
324 #include <linux/kdebug.h>
325 +#include <linux/vs_memory.h>
327 #include <asm/pgtable.h>
328 #include <asm/processor.h>
329 diff -NurpP --minimal linux-2.6.29.1/arch/m32r/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m32r/kernel/traps.c
330 --- linux-2.6.29.1/arch/m32r/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
331 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m32r/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
332 @@ -196,8 +196,9 @@ static void show_registers(struct pt_reg
334 printk("SPI: %08lx\n", sp);
336 - printk("Process %s (pid: %d, process nr: %d, stackpage=%08lx)",
337 - current->comm, task_pid_nr(current), 0xffff & i, 4096+(unsigned long)current);
338 + printk("Process %s (pid: %d[#%u], process nr: %d, stackpage=%08lx)",
339 + current->comm, task_pid_nr(current), current->xid,
340 + 0xffff & i, 4096+(unsigned long)current);
343 * When in-kernel, we also print out the stack and code at the
344 diff -NurpP --minimal linux-2.6.29.1/arch/m68k/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/Kconfig
345 --- linux-2.6.29.1/arch/m68k/Kconfig 2009-03-24 14:18:26.000000000 +0100
346 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/Kconfig 2009-03-24 14:48:20.000000000 +0100
347 @@ -616,6 +616,8 @@ source "fs/Kconfig"
349 source "arch/m68k/Kconfig.debug"
351 +source "kernel/vserver/Kconfig"
353 source "security/Kconfig"
355 source "crypto/Kconfig"
356 diff -NurpP --minimal linux-2.6.29.1/arch/m68k/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/kernel/ptrace.c
357 --- linux-2.6.29.1/arch/m68k/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
358 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
360 #include <linux/ptrace.h>
361 #include <linux/user.h>
362 #include <linux/signal.h>
363 +#include <linux/vs_base.h>
365 #include <asm/uaccess.h>
366 #include <asm/page.h>
367 @@ -269,6 +270,8 @@ long arch_ptrace(struct task_struct *chi
368 ret = ptrace_request(child, request, addr, data);
371 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
376 diff -NurpP --minimal linux-2.6.29.1/arch/m68k/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/kernel/traps.c
377 --- linux-2.6.29.1/arch/m68k/kernel/traps.c 2009-03-24 14:18:26.000000000 +0100
378 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68k/kernel/traps.c 2009-03-24 14:48:20.000000000 +0100
379 @@ -906,8 +906,8 @@ void show_registers(struct pt_regs *regs
380 printk("d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
381 regs->d4, regs->d5, regs->a0, regs->a1);
383 - printk("Process %s (pid: %d, task=%p)\n",
384 - current->comm, task_pid_nr(current), current);
385 + printk("Process %s (pid: %d[#%u], task=%p)\n",
386 + current->comm, task_pid_nr(current), current->xid, current);
387 addr = (unsigned long)&fp->un;
388 printk("Frame format=%X ", regs->format);
389 switch (regs->format) {
390 diff -NurpP --minimal linux-2.6.29.1/arch/m68knommu/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68knommu/Kconfig
391 --- linux-2.6.29.1/arch/m68knommu/Kconfig 2009-03-24 14:18:27.000000000 +0100
392 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68knommu/Kconfig 2009-03-24 14:48:20.000000000 +0100
393 @@ -720,6 +720,8 @@ source "fs/Kconfig"
395 source "arch/m68knommu/Kconfig.debug"
397 +source "kernel/vserver/Kconfig"
399 source "security/Kconfig"
401 source "crypto/Kconfig"
402 diff -NurpP --minimal linux-2.6.29.1/arch/m68knommu/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68knommu/kernel/traps.c
403 --- linux-2.6.29.1/arch/m68knommu/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
404 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/m68knommu/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
405 @@ -78,8 +78,9 @@ void die_if_kernel(char *str, struct pt_
406 printk(KERN_EMERG "d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
407 fp->d4, fp->d5, fp->a0, fp->a1);
409 - printk(KERN_EMERG "Process %s (pid: %d, stackpage=%08lx)\n",
410 - current->comm, current->pid, PAGE_SIZE+(unsigned long)current);
411 + printk(KERN_EMERG "Process %s (pid: %d[#%u], stackpage=%08lx)\n",
412 + current->comm, task_pid_nr(current), current->xid,
413 + PAGE_SIZE+(unsigned long)current);
414 show_stack(NULL, (unsigned long *)(fp + 1));
415 add_taint(TAINT_DIE);
417 diff -NurpP --minimal linux-2.6.29.1/arch/mips/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/Kconfig
418 --- linux-2.6.29.1/arch/mips/Kconfig 2009-03-24 14:18:29.000000000 +0100
419 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/Kconfig 2009-03-24 14:48:21.000000000 +0100
420 @@ -2142,6 +2142,8 @@ source "fs/Kconfig"
422 source "arch/mips/Kconfig.debug"
424 +source "kernel/vserver/Kconfig"
426 source "security/Kconfig"
428 source "crypto/Kconfig"
429 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/ptrace.c
430 --- linux-2.6.29.1/arch/mips/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
431 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
433 #include <linux/security.h>
434 #include <linux/audit.h>
435 #include <linux/seccomp.h>
436 +#include <linux/vs_base.h>
438 #include <asm/byteorder.h>
440 @@ -259,6 +260,9 @@ long arch_ptrace(struct task_struct *chi
444 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
448 /* when I and D space are separate, these will need to be fixed. */
449 case PTRACE_PEEKTEXT: /* read word at location addr. */
450 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/scall32-o32.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall32-o32.S
451 --- linux-2.6.29.1/arch/mips/kernel/scall32-o32.S 2009-03-24 14:18:31.000000000 +0100
452 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall32-o32.S 2009-03-24 14:48:21.000000000 +0100
453 @@ -597,7 +597,7 @@ einval: li v0, -ENOSYS
454 sys sys_mq_timedreceive 5
455 sys sys_mq_notify 2 /* 4275 */
456 sys sys_mq_getsetattr 3
457 - sys sys_ni_syscall 0 /* sys_vserver */
460 sys sys_ni_syscall 0 /* available, was setaltroot */
461 sys sys_add_key 5 /* 4280 */
462 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/scall64-64.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-64.S
463 --- linux-2.6.29.1/arch/mips/kernel/scall64-64.S 2009-03-24 14:18:31.000000000 +0100
464 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-64.S 2009-03-24 14:48:21.000000000 +0100
465 @@ -434,7 +434,7 @@ sys_call_table:
466 PTR sys_mq_timedreceive
468 PTR sys_mq_getsetattr /* 5235 */
469 - PTR sys_ni_syscall /* sys_vserver */
472 PTR sys_ni_syscall /* available, was setaltroot */
474 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/scall64-n32.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-n32.S
475 --- linux-2.6.29.1/arch/mips/kernel/scall64-n32.S 2009-03-24 14:18:31.000000000 +0100
476 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-n32.S 2009-03-24 14:48:21.000000000 +0100
477 @@ -360,7 +360,7 @@ EXPORT(sysn32_call_table)
478 PTR compat_sys_mq_timedreceive
479 PTR compat_sys_mq_notify
480 PTR compat_sys_mq_getsetattr
481 - PTR sys_ni_syscall /* 6240, sys_vserver */
482 + PTR sys32_vserver /* 6240 */
483 PTR compat_sys_waitid
484 PTR sys_ni_syscall /* available, was setaltroot */
486 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/scall64-o32.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-o32.S
487 --- linux-2.6.29.1/arch/mips/kernel/scall64-o32.S 2009-03-24 14:18:31.000000000 +0100
488 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/scall64-o32.S 2009-03-24 14:59:48.000000000 +0100
489 @@ -480,7 +480,7 @@ sys_call_table:
490 PTR compat_sys_mq_timedreceive
491 PTR compat_sys_mq_notify /* 4275 */
492 PTR compat_sys_mq_getsetattr
493 - PTR sys_ni_syscall /* sys_vserver */
496 PTR sys_ni_syscall /* available, was setaltroot */
497 PTR sys_add_key /* 4280 */
498 diff -NurpP --minimal linux-2.6.29.1/arch/mips/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/traps.c
499 --- linux-2.6.29.1/arch/mips/kernel/traps.c 2009-03-24 14:18:31.000000000 +0100
500 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/mips/kernel/traps.c 2009-03-24 14:48:21.000000000 +0100
501 @@ -335,9 +335,10 @@ void show_registers(const struct pt_regs
505 - printk("Process %s (pid: %d, threadinfo=%p, task=%p, tls=%0*lx)\n",
506 - current->comm, current->pid, current_thread_info(), current,
507 - field, current_thread_info()->tp_value);
508 + printk("Process %s (pid: %d:#%u, threadinfo=%p, task=%p, tls=%0*lx)\n",
509 + current->comm, task_pid_nr(current), current->xid,
510 + current_thread_info(), current,
511 + field, current_thread_info()->tp_value);
512 if (cpu_has_userlocal) {
515 diff -NurpP --minimal linux-2.6.29.1/arch/parisc/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/Kconfig
516 --- linux-2.6.29.1/arch/parisc/Kconfig 2009-03-24 14:18:32.000000000 +0100
517 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/Kconfig 2009-03-24 14:48:21.000000000 +0100
518 @@ -281,6 +281,8 @@ source "fs/Kconfig"
520 source "arch/parisc/Kconfig.debug"
522 +source "kernel/vserver/Kconfig"
524 source "security/Kconfig"
526 source "crypto/Kconfig"
527 diff -NurpP --minimal linux-2.6.29.1/arch/parisc/kernel/syscall_table.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/kernel/syscall_table.S
528 --- linux-2.6.29.1/arch/parisc/kernel/syscall_table.S 2008-12-25 00:26:37.000000000 +0100
529 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/kernel/syscall_table.S 2009-02-22 22:54:24.000000000 +0100
531 ENTRY_COMP(mbind) /* 260 */
532 ENTRY_COMP(get_mempolicy)
533 ENTRY_COMP(set_mempolicy)
534 - ENTRY_SAME(ni_syscall) /* 263: reserved for vserver */
535 + ENTRY_DIFF(vserver)
537 ENTRY_SAME(request_key) /* 265 */
539 diff -NurpP --minimal linux-2.6.29.1/arch/parisc/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/kernel/traps.c
540 --- linux-2.6.29.1/arch/parisc/kernel/traps.c 2009-03-24 14:18:32.000000000 +0100
541 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/kernel/traps.c 2009-03-24 14:48:21.000000000 +0100
542 @@ -236,8 +236,9 @@ void die_if_kernel(char *str, struct pt_
546 - printk(KERN_CRIT "%s (pid %d): %s (code %ld) at " RFMT "\n",
547 - current->comm, task_pid_nr(current), str, err, regs->iaoq[0]);
548 + printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld) at " RFMT "\n",
549 + current->comm, task_pid_nr(current), current->xid,
550 + str, err, regs->iaoq[0]);
551 #ifdef PRINT_USER_FAULTS
552 /* XXX for debugging only */
554 @@ -269,8 +270,8 @@ KERN_CRIT " || |
555 pdc_console_restart();
558 - printk(KERN_CRIT "%s (pid %d): %s (code %ld)\n",
559 - current->comm, task_pid_nr(current), str, err);
560 + printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld)\n",
561 + current->comm, task_pid_nr(current), current->xid, str, err);
563 /* Wot's wrong wif bein' racy? */
564 if (current->thread.flags & PARISC_KERNEL_DEATH) {
565 diff -NurpP --minimal linux-2.6.29.1/arch/parisc/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/mm/fault.c
566 --- linux-2.6.29.1/arch/parisc/mm/fault.c 2009-03-24 14:18:33.000000000 +0100
567 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/parisc/mm/fault.c 2009-03-24 14:48:21.000000000 +0100
568 @@ -238,8 +238,9 @@ bad_area:
570 #ifdef PRINT_USER_FAULTS
571 printk(KERN_DEBUG "\n");
572 - printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n",
573 - task_pid_nr(tsk), tsk->comm, code, address);
574 + printk(KERN_DEBUG "do_page_fault() pid=%d:#%u "
575 + "command='%s' type=%lu address=0x%08lx\n",
576 + task_pid_nr(tsk), tsk->xid, tsk->comm, code, address);
578 printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n",
579 vma->vm_start, vma->vm_end);
580 @@ -265,7 +266,8 @@ no_context:
583 up_read(&mm->mmap_sem);
584 - printk(KERN_CRIT "VM: killing process %s\n", current->comm);
585 + printk(KERN_CRIT "VM: killing process %s(%d:#%u)\n",
586 + current->comm, current->pid, current->xid);
588 do_group_exit(SIGKILL);
590 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/Kconfig
591 --- linux-2.6.29.1/arch/powerpc/Kconfig 2009-03-24 14:18:33.000000000 +0100
592 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/Kconfig 2009-03-24 14:48:22.000000000 +0100
593 @@ -882,6 +882,8 @@ source "lib/Kconfig"
595 source "arch/powerpc/Kconfig.debug"
597 +source "kernel/vserver/Kconfig"
599 source "security/Kconfig"
602 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/kernel/irq.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/irq.c
603 --- linux-2.6.29.1/arch/powerpc/kernel/irq.c 2009-03-24 14:18:35.000000000 +0100
604 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/irq.c 2009-03-24 14:48:22.000000000 +0100
606 #include <linux/bootmem.h>
607 #include <linux/pci.h>
608 #include <linux/debugfs.h>
609 +#include <linux/vs_context.h>
611 #include <asm/uaccess.h>
612 #include <asm/system.h>
613 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/kernel/process.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/process.c
614 --- linux-2.6.29.1/arch/powerpc/kernel/process.c 2009-03-24 14:18:35.000000000 +0100
615 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/process.c 2009-03-24 14:48:22.000000000 +0100
616 @@ -516,8 +516,9 @@ void show_regs(struct pt_regs * regs)
618 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
620 - printk("TASK = %p[%d] '%s' THREAD: %p",
621 - current, task_pid_nr(current), current->comm, task_thread_info(current));
622 + printk("TASK = %p[%d,#%u] '%s' THREAD: %p",
623 + current, task_pid_nr(current), current->xid,
624 + current->comm, task_thread_info(current));
627 printk(" CPU: %d", raw_smp_processor_id());
628 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/kernel/traps.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/traps.c
629 --- linux-2.6.29.1/arch/powerpc/kernel/traps.c 2009-03-24 14:18:35.000000000 +0100
630 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/traps.c 2009-03-24 14:48:22.000000000 +0100
631 @@ -940,8 +940,9 @@ void nonrecoverable_exception(struct pt_
633 void trace_syscall(struct pt_regs *regs)
635 - printk("Task: %p(%d), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
636 - current, task_pid_nr(current), regs->nip, regs->link, regs->gpr[0],
637 + printk("Task: %p(%d[#%u]), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
638 + current, task_pid_nr(current), current->xid,
639 + regs->nip, regs->link, regs->gpr[0],
640 regs->ccr&0x10000000?"Error=":"", regs->gpr[3], print_tainted());
643 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/kernel/vdso.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/vdso.c
644 --- linux-2.6.29.1/arch/powerpc/kernel/vdso.c 2009-03-24 14:18:35.000000000 +0100
645 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/kernel/vdso.c 2009-03-24 14:48:22.000000000 +0100
647 #include <linux/security.h>
648 #include <linux/bootmem.h>
649 #include <linux/lmb.h>
650 +#include <linux/vs_memory.h>
652 #include <asm/pgtable.h>
653 #include <asm/system.h>
654 diff -NurpP --minimal linux-2.6.29.1/arch/powerpc/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/mm/fault.c
655 --- linux-2.6.29.1/arch/powerpc/mm/fault.c 2009-03-24 14:18:36.000000000 +0100
656 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/powerpc/mm/fault.c 2009-03-24 14:48:22.000000000 +0100
657 @@ -362,7 +362,8 @@ out_of_memory:
658 down_read(&mm->mmap_sem);
661 - printk("VM: killing process %s\n", current->comm);
662 + printk("VM: killing process %s(%d:#%u)\n",
663 + current->comm, current->pid, current->xid);
665 do_group_exit(SIGKILL);
667 diff -NurpP --minimal linux-2.6.29.1/arch/s390/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/Kconfig
668 --- linux-2.6.29.1/arch/s390/Kconfig 2009-03-24 14:18:38.000000000 +0100
669 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/Kconfig 2009-03-24 14:48:22.000000000 +0100
670 @@ -586,6 +586,8 @@ source "fs/Kconfig"
672 source "arch/s390/Kconfig.debug"
674 +source "kernel/vserver/Kconfig"
676 source "security/Kconfig"
678 source "crypto/Kconfig"
679 diff -NurpP --minimal linux-2.6.29.1/arch/s390/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/kernel/ptrace.c
680 --- linux-2.6.29.1/arch/s390/kernel/ptrace.c 2009-03-24 14:18:40.000000000 +0100
681 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/kernel/ptrace.c 2009-03-24 14:48:22.000000000 +0100
683 #include <linux/elf.h>
684 #include <linux/regset.h>
685 #include <linux/tracehook.h>
686 +#include <linux/vs_base.h>
688 #include <asm/segment.h>
689 #include <asm/page.h>
690 diff -NurpP --minimal linux-2.6.29.1/arch/s390/kernel/syscalls.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/kernel/syscalls.S
691 --- linux-2.6.29.1/arch/s390/kernel/syscalls.S 2009-03-24 14:18:40.000000000 +0100
692 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/kernel/syscalls.S 2009-03-24 14:48:22.000000000 +0100
693 @@ -271,7 +271,7 @@ SYSCALL(sys_clock_settime,sys_clock_sett
694 SYSCALL(sys_clock_gettime,sys_clock_gettime,sys32_clock_gettime_wrapper) /* 260 */
695 SYSCALL(sys_clock_getres,sys_clock_getres,sys32_clock_getres_wrapper)
696 SYSCALL(sys_clock_nanosleep,sys_clock_nanosleep,sys32_clock_nanosleep_wrapper)
697 -NI_SYSCALL /* reserved for vserver */
698 +SYSCALL(sys_vserver,sys_vserver,sys32_vserver)
699 SYSCALL(sys_s390_fadvise64_64,sys_ni_syscall,sys32_fadvise64_64_wrapper)
700 SYSCALL(sys_statfs64,sys_statfs64,compat_sys_statfs64_wrapper)
701 SYSCALL(sys_fstatfs64,sys_fstatfs64,compat_sys_fstatfs64_wrapper)
702 diff -NurpP --minimal linux-2.6.29.1/arch/s390/mm/fault.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/mm/fault.c
703 --- linux-2.6.29.1/arch/s390/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
704 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/s390/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
705 @@ -216,7 +216,8 @@ static int do_out_of_memory(struct pt_re
706 down_read(&mm->mmap_sem);
709 - printk("VM: killing process %s\n", tsk->comm);
710 + printk("VM: killing process %s(%d:#%u)\n",
711 + tsk->comm, tsk->pid, tsk->xid);
712 if (regs->psw.mask & PSW_MASK_PSTATE)
713 do_group_exit(SIGKILL);
714 do_no_context(regs, error_code, address);
715 diff -NurpP --minimal linux-2.6.29.1/arch/sh/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/Kconfig
716 --- linux-2.6.29.1/arch/sh/Kconfig 2009-03-24 14:18:40.000000000 +0100
717 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/Kconfig 2009-03-24 14:48:22.000000000 +0100
718 @@ -694,6 +694,8 @@ source "fs/Kconfig"
720 source "arch/sh/Kconfig.debug"
722 +source "kernel/vserver/Kconfig"
724 source "security/Kconfig"
726 source "crypto/Kconfig"
727 diff -NurpP --minimal linux-2.6.29.1/arch/sh/kernel/irq.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/kernel/irq.c
728 --- linux-2.6.29.1/arch/sh/kernel/irq.c 2008-12-25 00:26:37.000000000 +0100
729 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/kernel/irq.c 2009-02-22 22:54:24.000000000 +0100
731 #include <linux/module.h>
732 #include <linux/kernel_stat.h>
733 #include <linux/seq_file.h>
734 +#include <linux/vs_context.h>
735 #include <asm/processor.h>
736 #include <asm/machvec.h>
737 #include <asm/uaccess.h>
738 diff -NurpP --minimal linux-2.6.29.1/arch/sh/kernel/vsyscall/vsyscall.c linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/kernel/vsyscall/vsyscall.c
739 --- linux-2.6.29.1/arch/sh/kernel/vsyscall/vsyscall.c 2009-03-24 14:18:42.000000000 +0100
740 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sh/kernel/vsyscall/vsyscall.c 2009-03-24 14:48:22.000000000 +0100
742 #include <linux/elf.h>
743 #include <linux/sched.h>
744 #include <linux/err.h>
745 +#include <linux/vs_memory.h>
748 * Should the kernel map a VDSO page into processes and pass its
749 diff -NurpP --minimal linux-2.6.29.1/arch/sparc/include/asm/tlb_64.h linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/include/asm/tlb_64.h
750 --- linux-2.6.29.1/arch/sparc/include/asm/tlb_64.h 2009-04-15 22:53:11.000000000 +0200
751 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/include/asm/tlb_64.h 2009-04-15 22:58:30.000000000 +0200
754 #include <linux/swap.h>
755 #include <linux/pagemap.h>
756 +#include <linux/vs_memory.h>
757 #include <asm/pgalloc.h>
758 #include <asm/tlbflush.h>
759 #include <asm/mmu_context.h>
760 diff -NurpP --minimal linux-2.6.29.1/arch/sparc/include/asm/unistd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/include/asm/unistd.h
761 --- linux-2.6.29.1/arch/sparc/include/asm/unistd.h 2009-03-24 14:18:44.000000000 +0100
762 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/include/asm/unistd.h 2009-04-08 15:52:53.000000000 +0200
764 #define __NR_timer_getoverrun 264
765 #define __NR_timer_delete 265
766 #define __NR_timer_create 266
767 -/* #define __NR_vserver 267 Reserved for VSERVER */
768 +#define __NR_vserver 267
769 #define __NR_io_setup 268
770 #define __NR_io_destroy 269
771 #define __NR_io_submit 270
772 diff -NurpP --minimal linux-2.6.29.1/arch/sparc/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/Kconfig
773 --- linux-2.6.29.1/arch/sparc/Kconfig 2009-03-24 14:18:43.000000000 +0100
774 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/Kconfig 2009-03-24 14:48:22.000000000 +0100
775 @@ -522,6 +522,8 @@ source "fs/Kconfig"
777 source "arch/sparc/Kconfig.debug"
779 +source "kernel/vserver/Kconfig"
781 source "security/Kconfig"
783 source "crypto/Kconfig"
784 diff -NurpP --minimal linux-2.6.29.1/arch/sparc/kernel/systbls_32.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/kernel/systbls_32.S
785 --- linux-2.6.29.1/arch/sparc/kernel/systbls_32.S 2009-03-24 14:18:45.000000000 +0100
786 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/kernel/systbls_32.S 2009-04-08 15:57:59.000000000 +0200
787 @@ -70,7 +70,7 @@ sys_call_table:
788 /*250*/ .long sparc_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
789 /*255*/ .long sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
790 /*260*/ .long sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
791 -/*265*/ .long sys_timer_delete, sys_timer_create, sys_nis_syscall, sys_io_setup, sys_io_destroy
792 +/*265*/ .long sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
793 /*270*/ .long sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
794 /*275*/ .long sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
795 /*280*/ .long sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
796 diff -NurpP --minimal linux-2.6.29.1/arch/sparc/kernel/systbls_64.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/kernel/systbls_64.S
797 --- linux-2.6.29.1/arch/sparc/kernel/systbls_64.S 2009-03-24 14:18:45.000000000 +0100
798 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/sparc/kernel/systbls_64.S 2009-04-08 15:56:23.000000000 +0200
799 @@ -71,7 +71,7 @@ sys_call_table32:
800 /*250*/ .word sys32_mremap, sys32_sysctl, sys32_getsid, sys_fdatasync, sys32_nfsservctl
801 .word sys32_sync_file_range, compat_sys_clock_settime, compat_sys_clock_gettime, compat_sys_clock_getres, sys32_clock_nanosleep
802 /*260*/ .word compat_sys_sched_getaffinity, compat_sys_sched_setaffinity, sys32_timer_settime, compat_sys_timer_gettime, sys_timer_getoverrun
803 - .word sys_timer_delete, compat_sys_timer_create, sys_ni_syscall, compat_sys_io_setup, sys_io_destroy
804 + .word sys_timer_delete, compat_sys_timer_create, sys32_vserver, compat_sys_io_setup, sys_io_destroy
805 /*270*/ .word sys32_io_submit, sys_io_cancel, compat_sys_io_getevents, sys32_mq_open, sys_mq_unlink
806 .word compat_sys_mq_timedsend, compat_sys_mq_timedreceive, compat_sys_mq_notify, compat_sys_mq_getsetattr, compat_sys_waitid
807 /*280*/ .word sys32_tee, sys_add_key, sys_request_key, sys_keyctl, compat_sys_openat
808 @@ -145,7 +145,7 @@ sys_call_table:
809 /*250*/ .word sys_64_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
810 .word sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
811 /*260*/ .word sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
812 - .word sys_timer_delete, sys_timer_create, sys_ni_syscall, sys_io_setup, sys_io_destroy
813 + .word sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
814 /*270*/ .word sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
815 .word sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
816 /*280*/ .word sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
817 diff -NurpP --minimal linux-2.6.29.1/arch/x86/ia32/ia32entry.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/ia32/ia32entry.S
818 --- linux-2.6.29.1/arch/x86/ia32/ia32entry.S 2009-03-24 14:18:48.000000000 +0100
819 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/ia32/ia32entry.S 2009-03-22 23:29:11.000000000 +0100
820 @@ -768,7 +768,7 @@ ia32_sys_call_table:
821 .quad sys_tgkill /* 270 */
822 .quad compat_sys_utimes
823 .quad sys32_fadvise64_64
824 - .quad quiet_ni_syscall /* sys_vserver */
825 + .quad sys32_vserver
827 .quad compat_sys_get_mempolicy /* 275 */
828 .quad sys_set_mempolicy
829 diff -NurpP --minimal linux-2.6.29.1/arch/x86/include/asm/unistd_64.h linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/include/asm/unistd_64.h
830 --- linux-2.6.29.1/arch/x86/include/asm/unistd_64.h 2008-12-25 00:26:37.000000000 +0100
831 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/include/asm/unistd_64.h 2009-02-22 22:54:24.000000000 +0100
832 @@ -535,7 +535,7 @@ __SYSCALL(__NR_tgkill, sys_tgkill)
833 #define __NR_utimes 235
834 __SYSCALL(__NR_utimes, sys_utimes)
835 #define __NR_vserver 236
836 -__SYSCALL(__NR_vserver, sys_ni_syscall)
837 +__SYSCALL(__NR_vserver, sys_vserver)
838 #define __NR_mbind 237
839 __SYSCALL(__NR_mbind, sys_mbind)
840 #define __NR_set_mempolicy 238
841 diff -NurpP --minimal linux-2.6.29.1/arch/x86/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/Kconfig
842 --- linux-2.6.29.1/arch/x86/Kconfig 2009-03-24 14:18:47.000000000 +0100
843 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/Kconfig 2009-03-24 14:48:23.000000000 +0100
844 @@ -1990,6 +1990,8 @@ source "fs/Kconfig"
846 source "arch/x86/Kconfig.debug"
848 +source "kernel/vserver/Kconfig"
850 source "security/Kconfig"
852 source "crypto/Kconfig"
853 diff -NurpP --minimal linux-2.6.29.1/arch/x86/kernel/syscall_table_32.S linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/kernel/syscall_table_32.S
854 --- linux-2.6.29.1/arch/x86/kernel/syscall_table_32.S 2009-03-24 14:18:51.000000000 +0100
855 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/arch/x86/kernel/syscall_table_32.S 2009-02-22 22:54:24.000000000 +0100
856 @@ -272,7 +272,7 @@ ENTRY(sys_call_table)
857 .long sys_tgkill /* 270 */
859 .long sys_fadvise64_64
860 - .long sys_ni_syscall /* sys_vserver */
863 .long sys_get_mempolicy
864 .long sys_set_mempolicy
865 diff -NurpP --minimal linux-2.6.29.1/Documentation/vserver/debug.txt linux-2.6.29.1-vs2.3.0.36.9-pre5/Documentation/vserver/debug.txt
866 --- linux-2.6.29.1/Documentation/vserver/debug.txt 1970-01-01 01:00:00.000000000 +0100
867 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/Documentation/vserver/debug.txt 2009-02-22 22:54:24.000000000 +0100
872 + 2 4 "vx_map_tgid: %p/%llx: %d -> %d"
873 + "vx_rmap_tgid: %p/%llx: %d -> %d"
877 + 0 1 "ALLOC (%p,#%d)%c inode (%d)"
878 + "FREE (%p,#%d)%c inode"
879 + 1 2 "ALLOC (%p,#%d)%c %lld bytes (%d)"
880 + "FREE (%p,#%d)%c %lld bytes"
881 + 2 4 "ADJUST: %lld,%lld on %ld,%ld [mult=%d]"
882 + 3 8 "ext3_has_free_blocks(%p): %lu<%lu+1, %c, %u!=%u r=%d"
883 + "ext3_has_free_blocks(%p): free=%lu, root=%lu"
884 + "rcu_free_dl_info(%p)"
885 + 4 10 "alloc_dl_info(%p,%d) = %p"
886 + "dealloc_dl_info(%p)"
887 + "get_dl_info(%p[#%d.%d])"
888 + "put_dl_info(%p[#%d.%d])"
889 + 5 20 "alloc_dl_info(%p,%d)*"
890 + 6 40 "__hash_dl_info: %p[#%d]"
891 + "__unhash_dl_info: %p[#%d]"
892 + 7 80 "locate_dl_info(%p,#%d) = %p"
896 + 0 1 "destroy_dqhash: %p [#0x%08x] c=%d"
897 + "new_dqhash: %p [#0x%08x]"
898 + "vroot[%d]_clr_dev: dev=%p[%lu,%d:%d]"
899 + "vroot[%d]_get_real_bdev: dev=%p[%lu,%d:%d]"
900 + "vroot[%d]_set_dev: dev=%p[%lu,%d:%d]"
901 + "vroot_get_real_bdev not set"
902 + 1 2 "cow_break_link(»%s«)"
904 + 2 4 "dentry_open(new): %p"
905 + "dentry_open(old): %p"
906 + "lookup_create(new): %p"
908 + "path_lookup(old): %d"
909 + "vfs_create(new): %d"
912 + 3 8 "fput(new_file=%p[#%d])"
913 + "fput(old_file=%p[#%d])"
914 + 4 10 "vx_info_kill(%p[#%d],%d,%d) = %d"
915 + "vx_info_kill(%p[#%d],%d,%d)*"
916 + 5 20 "vs_reboot(%p[#%d],%d)"
917 + 6 40 "dropping task %p[#%u,%u] for %p[#%u,%u]"
921 + 2 4 "nx_addr_conflict(%p,%p) %d.%d,%d.%d"
922 + 3 8 "inet_bind(%p) %d.%d.%d.%d, %d.%d.%d.%d, %d.%d.%d.%d"
923 + "inet_bind(%p)* %p,%p;%lx %d.%d.%d.%d"
924 + 4 10 "ip_route_connect(%p) %p,%p;%lx"
925 + 5 20 "__addr_in_socket(%p,%d.%d.%d.%d) %p:%d.%d.%d.%d %p;%lx"
926 + 6 40 "sk,egf: %p [#%d] (from %d)"
927 + "sk,egn: %p [#%d] (from %d)"
928 + "sk,req: %p [#%d] (from %d)"
929 + "sk: %p [#%d] (from %d)"
930 + "tw: %p [#%d] (from %d)"
931 + 7 80 "__sock_recvmsg: %p[%p,%p,%p;%d]:%d/%d"
932 + "__sock_sendmsg: %p[%p,%p,%p;%d]:%d/%d"
936 + 0 1 "__lookup_nx_info(#%u): %p[#%u]"
937 + "alloc_nx_info(%d) = %p"
938 + "create_nx_info(%d) (dynamic rejected)"
939 + "create_nx_info(%d) = %p (already there)"
940 + "create_nx_info(%d) = %p (new)"
941 + "dealloc_nx_info(%p)"
942 + 1 2 "alloc_nx_info(%d)*"
943 + "create_nx_info(%d)*"
944 + 2 4 "get_nx_info(%p[#%d.%d])"
945 + "put_nx_info(%p[#%d.%d])"
946 + 3 8 "claim_nx_info(%p[#%d.%d.%d]) %p"
947 + "clr_nx_info(%p[#%d.%d])"
948 + "init_nx_info(%p[#%d.%d])"
949 + "release_nx_info(%p[#%d.%d.%d]) %p"
950 + "set_nx_info(%p[#%d.%d])"
951 + 4 10 "__hash_nx_info: %p[#%d]"
952 + "__nx_dynamic_id: [#%d]"
953 + "__unhash_nx_info: %p[#%d.%d.%d]"
954 + 5 20 "moved task %p into nxi:%p[#%d]"
955 + "nx_migrate_task(%p,%p[#%d.%d.%d])"
956 + "task_get_nx_info(%p)"
957 + 6 40 "nx_clear_persistent(%p[#%d])"
961 + 0 1 "quota_sync_dqh(%p,%d) discard inode %p"
962 + 1 2 "quota_sync_dqh(%p,%d)"
963 + "sync_dquots(%p,%d)"
964 + "sync_dquots_dqh(%p,%d)"
965 + 3 8 "do_quotactl(%p,%d,cmd=%d,id=%d,%p)"
969 + 0 1 "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]"
970 + 1 2 "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]"
971 + 4 10 "%s: (%s %s) returned %s with %d"
975 + 7 80 "dx_parse_tag(»%s«): %d:#%d"
976 + "dx_propagate_tag(%p[#%lu.%d]): %d,%d"
980 + 0 1 "__lookup_vx_info(#%u): %p[#%u]"
981 + "alloc_vx_info(%d) = %p"
982 + "alloc_vx_info(%d)*"
983 + "create_vx_info(%d) (dynamic rejected)"
984 + "create_vx_info(%d) = %p (already there)"
985 + "create_vx_info(%d) = %p (new)"
986 + "dealloc_vx_info(%p)"
987 + "loc_vx_info(%d) = %p (found)"
988 + "loc_vx_info(%d) = %p (new)"
989 + "loc_vx_info(%d) = %p (not available)"
990 + 1 2 "create_vx_info(%d)*"
992 + 2 4 "get_vx_info(%p[#%d.%d])"
993 + "put_vx_info(%p[#%d.%d])"
994 + 3 8 "claim_vx_info(%p[#%d.%d.%d]) %p"
995 + "clr_vx_info(%p[#%d.%d])"
996 + "init_vx_info(%p[#%d.%d])"
997 + "release_vx_info(%p[#%d.%d.%d]) %p"
998 + "set_vx_info(%p[#%d.%d])"
999 + 4 10 "__hash_vx_info: %p[#%d]"
1000 + "__unhash_vx_info: %p[#%d.%d.%d]"
1001 + "__vx_dynamic_id: [#%d]"
1002 + 5 20 "enter_vx_info(%p[#%d],%p) %p[#%d,%p]"
1003 + "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]"
1004 + "moved task %p into vxi:%p[#%d]"
1005 + "task_get_vx_info(%p)"
1006 + "vx_migrate_task(%p,%p[#%d.%d])"
1007 + 6 40 "vx_clear_persistent(%p[#%d])"
1008 + "vx_exit_init(%p[#%d],%p[#%d,%d,%d])"
1009 + "vx_set_init(%p[#%d],%p[#%d,%d,%d])"
1010 + "vx_set_persistent(%p[#%d])"
1011 + "vx_set_reaper(%p[#%d],%p[#%d,%d])"
1012 + 7 80 "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]"
1017 + n 2^n "vx_acc_cres[%5d,%s,%2d]: %5d%s"
1018 + "vx_cres_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1020 + m 2^m "vx_acc_page[%5d,%s,%2d]: %5d%s"
1021 + "vx_acc_pages[%5d,%s,%2d]: %5d += %5d"
1022 + "vx_pages_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1023 diff -NurpP --minimal linux-2.6.29.1/drivers/block/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/Kconfig
1024 --- linux-2.6.29.1/drivers/block/Kconfig 2008-12-25 00:26:37.000000000 +0100
1025 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/Kconfig 2009-02-22 22:54:24.000000000 +0100
1026 @@ -264,6 +264,13 @@ config BLK_DEV_CRYPTOLOOP
1027 instead, which can be configured to be on-disk compatible with the
1030 +config BLK_DEV_VROOT
1031 + tristate "Virtual Root device support"
1032 + depends on QUOTACTL
1034 + Saying Y here will allow you to use quota/fs ioctls on a shared
1035 + partition within a virtual server without compromising security.
1038 tristate "Network block device support"
1040 diff -NurpP --minimal linux-2.6.29.1/drivers/block/loop.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/loop.c
1041 --- linux-2.6.29.1/drivers/block/loop.c 2009-03-24 14:18:56.000000000 +0100
1042 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/loop.c 2009-03-24 15:09:29.000000000 +0100
1044 #include <linux/gfp.h>
1045 #include <linux/kthread.h>
1046 #include <linux/splice.h>
1047 +#include <linux/vs_context.h>
1049 #include <asm/uaccess.h>
1051 @@ -809,6 +810,7 @@ static int loop_set_fd(struct loop_devic
1052 lo->lo_blocksize = lo_blocksize;
1053 lo->lo_device = bdev;
1054 lo->lo_flags = lo_flags;
1055 + lo->lo_xid = vx_current_xid();
1056 lo->lo_backing_file = file;
1057 lo->transfer = transfer_none;
1059 @@ -931,6 +933,7 @@ static int loop_clr_fd(struct loop_devic
1060 lo->lo_encrypt_key_size = 0;
1062 lo->lo_thread = NULL;
1064 memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
1065 memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
1066 memset(lo->lo_file_name, 0, LO_NAME_SIZE);
1067 @@ -958,7 +961,7 @@ loop_set_status(struct loop_device *lo,
1069 if (lo->lo_encrypt_key_size &&
1070 lo->lo_key_owner != uid &&
1071 - !capable(CAP_SYS_ADMIN))
1072 + !vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP))
1074 if (lo->lo_state != Lo_bound)
1076 @@ -1042,7 +1045,8 @@ loop_get_status(struct loop_device *lo,
1077 memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
1078 info->lo_encrypt_type =
1079 lo->lo_encryption ? lo->lo_encryption->number : 0;
1080 - if (lo->lo_encrypt_key_size && capable(CAP_SYS_ADMIN)) {
1081 + if (lo->lo_encrypt_key_size &&
1082 + vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP)) {
1083 info->lo_encrypt_key_size = lo->lo_encrypt_key_size;
1084 memcpy(info->lo_encrypt_key, lo->lo_encrypt_key,
1085 lo->lo_encrypt_key_size);
1086 @@ -1351,6 +1355,9 @@ static int lo_open(struct block_device *
1088 struct loop_device *lo = bdev->bd_disk->private_data;
1090 + if (!vx_check(lo->lo_xid, VS_IDENT|VS_HOSTID|VS_ADMIN_P))
1093 mutex_lock(&lo->lo_ctl_mutex);
1095 mutex_unlock(&lo->lo_ctl_mutex);
1096 diff -NurpP --minimal linux-2.6.29.1/drivers/block/Makefile linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/Makefile
1097 --- linux-2.6.29.1/drivers/block/Makefile 2009-03-24 14:18:55.000000000 +0100
1098 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/Makefile 2009-03-24 14:48:25.000000000 +0100
1099 @@ -31,5 +31,6 @@ obj-$(CONFIG_VIODASD) += viodasd.o
1100 obj-$(CONFIG_BLK_DEV_SX8) += sx8.o
1101 obj-$(CONFIG_BLK_DEV_UB) += ub.o
1102 obj-$(CONFIG_BLK_DEV_HD) += hd.o
1103 +obj-$(CONFIG_BLK_DEV_VROOT) += vroot.o
1105 obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o
1106 diff -NurpP --minimal linux-2.6.29.1/drivers/block/vroot.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/vroot.c
1107 --- linux-2.6.29.1/drivers/block/vroot.c 1970-01-01 01:00:00.000000000 +0100
1108 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/block/vroot.c 2009-02-22 22:54:24.000000000 +0100
1111 + * linux/drivers/block/vroot.c
1113 + * written by Herbert Pötzl, 9/11/2002
1114 + * ported to 2.6.10 by Herbert Pötzl, 30/12/2004
1116 + * based on the loop.c code by Theodore Ts'o.
1118 + * Copyright (C) 2002-2007 by Herbert Pötzl.
1119 + * Redistribution of this file is permitted under the
1120 + * GNU General Public License.
1124 +#include <linux/module.h>
1125 +#include <linux/moduleparam.h>
1126 +#include <linux/file.h>
1127 +#include <linux/major.h>
1128 +#include <linux/blkdev.h>
1130 +#include <linux/vroot.h>
1131 +#include <linux/vs_context.h>
1134 +static int max_vroot = 8;
1136 +static struct vroot_device *vroot_dev;
1137 +static struct gendisk **disks;
1140 +static int vroot_set_dev(
1141 + struct vroot_device *vr,
1142 + struct block_device *bdev,
1145 + struct block_device *real_bdev;
1146 + struct file *file;
1147 + struct inode *inode;
1151 + if (vr->vr_state != Vr_unbound)
1160 + inode = file->f_dentry->d_inode;
1163 + if (S_ISBLK(inode->i_mode)) {
1164 + real_bdev = inode->i_bdev;
1165 + vr->vr_device = real_bdev;
1166 + __iget(real_bdev->bd_inode);
1170 + vxdprintk(VXD_CBIT(misc, 0),
1171 + "vroot[%d]_set_dev: dev=" VXF_DEV,
1172 + vr->vr_number, VXD_DEV(real_bdev));
1174 + vr->vr_state = Vr_bound;
1183 +static int vroot_clr_dev(
1184 + struct vroot_device *vr,
1185 + struct block_device *bdev)
1187 + struct block_device *real_bdev;
1189 + if (vr->vr_state != Vr_bound)
1191 + if (vr->vr_refcnt > 1) /* we needed one fd for the ioctl */
1194 + real_bdev = vr->vr_device;
1196 + vxdprintk(VXD_CBIT(misc, 0),
1197 + "vroot[%d]_clr_dev: dev=" VXF_DEV,
1198 + vr->vr_number, VXD_DEV(real_bdev));
1201 + vr->vr_state = Vr_unbound;
1202 + vr->vr_device = NULL;
1207 +static int vr_ioctl(struct block_device *bdev, fmode_t mode,
1208 + unsigned int cmd, unsigned long arg)
1210 + struct vroot_device *vr = bdev->bd_disk->private_data;
1213 + down(&vr->vr_ctl_mutex);
1215 + case VROOT_SET_DEV:
1216 + err = vroot_set_dev(vr, bdev, arg);
1218 + case VROOT_CLR_DEV:
1219 + err = vroot_clr_dev(vr, bdev);
1225 + up(&vr->vr_ctl_mutex);
1229 +static int vr_open(struct block_device *bdev, fmode_t mode)
1231 + struct vroot_device *vr = bdev->bd_disk->private_data;
1233 + down(&vr->vr_ctl_mutex);
1235 + up(&vr->vr_ctl_mutex);
1239 +static int vr_release(struct gendisk *disk, fmode_t mode)
1241 + struct vroot_device *vr = disk->private_data;
1243 + down(&vr->vr_ctl_mutex);
1245 + up(&vr->vr_ctl_mutex);
1249 +static struct block_device_operations vr_fops = {
1250 + .owner = THIS_MODULE,
1252 + .release = vr_release,
1253 + .ioctl = vr_ioctl,
1256 +struct block_device *__vroot_get_real_bdev(struct block_device *bdev)
1258 + struct inode *inode = bdev->bd_inode;
1259 + struct vroot_device *vr;
1260 + struct block_device *real_bdev;
1261 + int minor = iminor(inode);
1263 + vr = &vroot_dev[minor];
1264 + real_bdev = vr->vr_device;
1266 + vxdprintk(VXD_CBIT(misc, 0),
1267 + "vroot[%d]_get_real_bdev: dev=" VXF_DEV,
1268 + vr->vr_number, VXD_DEV(real_bdev));
1270 + if (vr->vr_state != Vr_bound)
1271 + return ERR_PTR(-ENXIO);
1273 + __iget(real_bdev->bd_inode);
1278 + * And now the modules code and kernel interface.
1281 +module_param(max_vroot, int, 0);
1283 +MODULE_PARM_DESC(max_vroot, "Maximum number of vroot devices (1-256)");
1284 +MODULE_LICENSE("GPL");
1285 +MODULE_ALIAS_BLOCKDEV_MAJOR(VROOT_MAJOR);
1287 +MODULE_AUTHOR ("Herbert Pötzl");
1288 +MODULE_DESCRIPTION ("Virtual Root Device Mapper");
1291 +int __init vroot_init(void)
1295 + if (max_vroot < 1 || max_vroot > 256) {
1296 + max_vroot = MAX_VROOT_DEFAULT;
1297 + printk(KERN_WARNING "vroot: invalid max_vroot "
1298 + "(must be between 1 and 256), "
1299 + "using default (%d)\n", max_vroot);
1302 + if (register_blkdev(VROOT_MAJOR, "vroot"))
1306 + vroot_dev = kmalloc(max_vroot * sizeof(struct vroot_device), GFP_KERNEL);
1309 + memset(vroot_dev, 0, max_vroot * sizeof(struct vroot_device));
1311 + disks = kmalloc(max_vroot * sizeof(struct gendisk *), GFP_KERNEL);
1315 + for (i = 0; i < max_vroot; i++) {
1316 + disks[i] = alloc_disk(1);
1319 + disks[i]->queue = blk_alloc_queue(GFP_KERNEL);
1320 + if (!disks[i]->queue)
1324 + for (i = 0; i < max_vroot; i++) {
1325 + struct vroot_device *vr = &vroot_dev[i];
1326 + struct gendisk *disk = disks[i];
1328 + memset(vr, 0, sizeof(*vr));
1329 + init_MUTEX(&vr->vr_ctl_mutex);
1330 + vr->vr_number = i;
1331 + disk->major = VROOT_MAJOR;
1332 + disk->first_minor = i;
1333 + disk->fops = &vr_fops;
1334 + sprintf(disk->disk_name, "vroot%d", i);
1335 + disk->private_data = vr;
1338 + err = register_vroot_grb(&__vroot_get_real_bdev);
1342 + for (i = 0; i < max_vroot; i++)
1343 + add_disk(disks[i]);
1344 + printk(KERN_INFO "vroot: loaded (max %d devices)\n", max_vroot);
1349 + put_disk(disks[i]);
1354 + unregister_blkdev(VROOT_MAJOR, "vroot");
1355 + printk(KERN_ERR "vroot: ran out of memory\n");
1359 +void vroot_exit(void)
1363 + if (unregister_vroot_grb(&__vroot_get_real_bdev))
1364 + printk(KERN_WARNING "vroot: cannot unregister grb\n");
1366 + for (i = 0; i < max_vroot; i++) {
1367 + del_gendisk(disks[i]);
1368 + put_disk(disks[i]);
1370 + unregister_blkdev(VROOT_MAJOR, "vroot");
1376 +module_init(vroot_init);
1377 +module_exit(vroot_exit);
1381 +static int __init max_vroot_setup(char *str)
1383 + max_vroot = simple_strtol(str, NULL, 0);
1387 +__setup("max_vroot=", max_vroot_setup);
1391 diff -NurpP --minimal linux-2.6.29.1/drivers/char/sysrq.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/char/sysrq.c
1392 --- linux-2.6.29.1/drivers/char/sysrq.c 2009-03-24 14:18:57.000000000 +0100
1393 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/char/sysrq.c 2009-03-24 15:15:27.000000000 +0100
1395 #include <linux/irq.h>
1396 #include <linux/hrtimer.h>
1397 #include <linux/oom.h>
1398 +#include <linux/vserver/debug.h>
1400 #include <asm/ptrace.h>
1401 #include <asm/irq_regs.h>
1402 @@ -369,6 +370,21 @@ static struct sysrq_key_op sysrq_unrt_op
1403 .enable_mask = SYSRQ_ENABLE_RTNICE,
1407 +#ifdef CONFIG_VSERVER_DEBUG
1408 +static void sysrq_handle_vxinfo(int key, struct tty_struct *tty)
1410 + dump_vx_info_inactive((key == 'x')?0:1);
1413 +static struct sysrq_key_op sysrq_showvxinfo_op = {
1414 + .handler = sysrq_handle_vxinfo,
1415 + .help_msg = "conteXt",
1416 + .action_msg = "Show Context Info",
1417 + .enable_mask = SYSRQ_ENABLE_DUMP,
1421 /* Key Operations table and lock */
1422 static DEFINE_SPINLOCK(sysrq_key_table_lock);
1424 @@ -419,7 +435,11 @@ static struct sysrq_key_op *sysrq_key_ta
1426 &sysrq_showstate_blocked_op, /* w */
1427 /* x: May be registered on ppc/powerpc for xmon */
1428 +#ifdef CONFIG_VSERVER_DEBUG
1429 + &sysrq_showvxinfo_op, /* x */
1433 /* y: May be registered on sparc64 for global register dump */
1435 &sysrq_ftrace_dump_op, /* z */
1436 @@ -434,6 +454,8 @@ static int sysrq_key_table_key2index(int
1438 else if ((key >= 'a') && (key <= 'z'))
1439 retval = key + 10 - 'a';
1440 + else if ((key >= 'A') && (key <= 'Z'))
1441 + retval = key + 10 - 'A';
1445 diff -NurpP --minimal linux-2.6.29.1/drivers/char/tty_io.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/char/tty_io.c
1446 --- linux-2.6.29.1/drivers/char/tty_io.c 2009-03-24 14:18:57.000000000 +0100
1447 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/char/tty_io.c 2009-03-24 14:48:25.000000000 +0100
1450 #include <linux/kmod.h>
1451 #include <linux/nsproxy.h>
1452 +#include <linux/vs_pid.h>
1454 #undef TTY_DEBUG_HANGUP
1456 @@ -2303,6 +2304,7 @@ static int tiocspgrp(struct tty_struct *
1458 if (get_user(pgrp_nr, p))
1460 + pgrp_nr = vx_rmap_pid(pgrp_nr);
1464 diff -NurpP --minimal linux-2.6.29.1/drivers/infiniband/hw/ipath/ipath_user_pages.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/infiniband/hw/ipath/ipath_user_pages.c
1465 --- linux-2.6.29.1/drivers/infiniband/hw/ipath/ipath_user_pages.c 2008-12-25 00:26:37.000000000 +0100
1466 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/infiniband/hw/ipath/ipath_user_pages.c 2009-02-22 22:54:24.000000000 +0100
1469 #include <linux/mm.h>
1470 #include <linux/device.h>
1471 +#include <linux/vs_memory.h>
1473 #include "ipath_kernel.h"
1475 @@ -61,7 +62,8 @@ static int __get_user_pages(unsigned lon
1476 lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >>
1479 - if (num_pages > lock_limit) {
1480 + if (num_pages > lock_limit ||
1481 + !vx_vmlocked_avail(current->mm, num_pages)) {
1485 @@ -78,7 +80,7 @@ static int __get_user_pages(unsigned lon
1489 - current->mm->locked_vm += num_pages;
1490 + vx_vmlocked_add(current->mm, num_pages);
1494 @@ -177,7 +179,7 @@ void ipath_release_user_pages(struct pag
1496 __ipath_release_user_pages(p, num_pages, 1);
1498 - current->mm->locked_vm -= num_pages;
1499 + vx_vmlocked_sub(current->mm, num_pages);
1501 up_write(¤t->mm->mmap_sem);
1503 @@ -194,7 +196,7 @@ static void user_pages_account(struct wo
1504 container_of(_work, struct ipath_user_pages_work, work);
1506 down_write(&work->mm->mmap_sem);
1507 - work->mm->locked_vm -= work->num_pages;
1508 + vx_vmlocked_sub(work->mm, work->num_pages);
1509 up_write(&work->mm->mmap_sem);
1512 diff -NurpP --minimal linux-2.6.29.1/drivers/md/dm.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm.c
1513 --- linux-2.6.29.1/drivers/md/dm.c 2009-03-24 14:19:05.000000000 +0100
1514 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm.c 2009-03-24 15:15:57.000000000 +0100
1516 #include <linux/hdreg.h>
1517 #include <linux/blktrace_api.h>
1518 #include <trace/block.h>
1519 +#include <linux/vs_base.h>
1521 #define DM_MSG_PREFIX "core"
1523 @@ -115,6 +116,7 @@ struct mapped_device {
1526 atomic_t open_count;
1529 unsigned long flags;
1531 @@ -295,6 +297,7 @@ static void __exit dm_exit(void)
1532 static int dm_blk_open(struct block_device *bdev, fmode_t mode)
1534 struct mapped_device *md;
1537 spin_lock(&_minor_lock);
1539 @@ -303,18 +306,19 @@ static int dm_blk_open(struct block_devi
1542 if (test_bit(DMF_FREEING, &md->flags) ||
1543 - test_bit(DMF_DELETING, &md->flags)) {
1545 + test_bit(DMF_DELETING, &md->flags))
1549 + if (!vx_check(md->xid, VS_IDENT|VS_HOSTID))
1554 atomic_inc(&md->open_count);
1558 spin_unlock(&_minor_lock);
1560 - return md ? 0 : -ENXIO;
1564 static int dm_blk_close(struct gendisk *disk, fmode_t mode)
1565 @@ -504,6 +508,14 @@ int dm_set_geometry(struct mapped_device
1570 + * Get the xid associated with a dm device
1572 +xid_t dm_get_xid(struct mapped_device *md)
1577 /*-----------------------------------------------------------------
1579 * A more elegant soln is in the works that uses the queue
1580 @@ -1110,6 +1122,7 @@ static struct mapped_device *alloc_dev(i
1581 INIT_LIST_HEAD(&md->uevent_list);
1582 spin_lock_init(&md->uevent_lock);
1584 + md->xid = vx_current_xid();
1585 md->queue = blk_alloc_queue(GFP_KERNEL);
1588 diff -NurpP --minimal linux-2.6.29.1/drivers/md/dm.h linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm.h
1589 --- linux-2.6.29.1/drivers/md/dm.h 2009-03-24 14:19:05.000000000 +0100
1590 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm.h 2009-03-24 14:48:26.000000000 +0100
1591 @@ -54,6 +54,8 @@ int dm_table_any_congested(struct dm_tab
1592 #define dm_target_is_valid(t) ((t)->table)
1593 int dm_table_barrier_ok(struct dm_table *t);
1595 +xid_t dm_get_xid(struct mapped_device *md);
1597 /*-----------------------------------------------------------------
1598 * A registry of target types.
1599 *---------------------------------------------------------------*/
1600 diff -NurpP --minimal linux-2.6.29.1/drivers/md/dm-ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm-ioctl.c
1601 --- linux-2.6.29.1/drivers/md/dm-ioctl.c 2009-03-24 14:19:05.000000000 +0100
1602 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/md/dm-ioctl.c 2009-03-24 14:48:26.000000000 +0100
1604 #include <linux/dm-ioctl.h>
1605 #include <linux/hdreg.h>
1606 #include <linux/compat.h>
1607 +#include <linux/vs_context.h>
1609 #include <asm/uaccess.h>
1611 @@ -101,7 +102,8 @@ static struct hash_cell *__get_name_cell
1612 unsigned int h = hash_str(str);
1614 list_for_each_entry (hc, _name_buckets + h, name_list)
1615 - if (!strcmp(hc->name, str)) {
1616 + if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1617 + !strcmp(hc->name, str)) {
1621 @@ -115,7 +117,8 @@ static struct hash_cell *__get_uuid_cell
1622 unsigned int h = hash_str(str);
1624 list_for_each_entry (hc, _uuid_buckets + h, uuid_list)
1625 - if (!strcmp(hc->uuid, str)) {
1626 + if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1627 + !strcmp(hc->uuid, str)) {
1631 @@ -352,6 +355,9 @@ typedef int (*ioctl_fn)(struct dm_ioctl
1633 static int remove_all(struct dm_ioctl *param, size_t param_size)
1635 + if (!vx_check(0, VS_ADMIN))
1638 dm_hash_remove_all(1);
1639 param->data_size = 0;
1641 @@ -399,6 +405,8 @@ static int list_devices(struct dm_ioctl
1643 for (i = 0; i < NUM_BUCKETS; i++) {
1644 list_for_each_entry (hc, _name_buckets + i, name_list) {
1645 + if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1647 needed += sizeof(struct dm_name_list);
1648 needed += strlen(hc->name) + 1;
1649 needed += ALIGN_MASK;
1650 @@ -422,6 +430,8 @@ static int list_devices(struct dm_ioctl
1652 for (i = 0; i < NUM_BUCKETS; i++) {
1653 list_for_each_entry (hc, _name_buckets + i, name_list) {
1654 + if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1657 old_nl->next = (uint32_t) ((void *) nl -
1659 @@ -612,10 +622,11 @@ static struct hash_cell *__find_device_h
1663 - mdptr = dm_get_mdptr(md);
1664 + if (vx_check(dm_get_xid(md), VS_WATCH_P | VS_IDENT))
1665 + mdptr = dm_get_mdptr(md);
1673 @@ -1405,8 +1416,8 @@ static int ctl_ioctl(uint command, struc
1677 - /* only root can play with this */
1678 - if (!capable(CAP_SYS_ADMIN))
1679 + /* only root and certain contexts can play with this */
1680 + if (!vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_MAPPER))
1683 if (_IOC_TYPE(command) != DM_IOCTL)
1684 diff -NurpP --minimal linux-2.6.29.1/drivers/net/tun.c linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/net/tun.c
1685 --- linux-2.6.29.1/drivers/net/tun.c 2009-03-24 14:19:23.000000000 +0100
1686 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/drivers/net/tun.c 2009-03-25 01:51:59.000000000 +0100
1688 #include <linux/crc32.h>
1689 #include <linux/nsproxy.h>
1690 #include <linux/virtio_net.h>
1691 +#include <linux/vs_network.h>
1692 #include <net/net_namespace.h>
1693 #include <net/netns/generic.h>
1695 @@ -93,6 +94,7 @@ struct tun_struct {
1701 wait_queue_head_t read_wait;
1702 struct sk_buff_head readq;
1703 @@ -697,6 +699,7 @@ static void tun_setup(struct net_device
1707 + tun->nid = current->nid;
1709 dev->ethtool_ops = &tun_ethtool_ops;
1710 dev->destructor = free_netdev;
1711 @@ -727,6 +730,9 @@ static int tun_set_iff(struct net *net,
1712 tn = net_generic(net, tun_net_id);
1713 tun = tun_get_by_name(tn, ifr->ifr_name);
1715 + if (!nx_check(tun->nid, VS_IDENT | VS_HOSTID | VS_ADMIN_P))
1721 @@ -735,7 +741,7 @@ static int tun_set_iff(struct net *net,
1722 cred->euid != tun->owner) ||
1723 (tun->group != -1 &&
1724 cred->egid != tun->group)) &&
1725 - !capable(CAP_NET_ADMIN)) {
1726 + !cap_raised(current_cap(), CAP_NET_ADMIN)) {
1730 @@ -747,7 +753,7 @@ static int tun_set_iff(struct net *net,
1734 - if (!capable(CAP_NET_ADMIN))
1735 + if (!nx_capable(CAP_NET_ADMIN, NXC_TUN_CREATE))
1739 @@ -987,6 +993,16 @@ static int tun_chr_ioctl(struct inode *i
1740 DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
1744 + if (!capable(CAP_CONTEXT))
1747 + /* Set nid owner of the device */
1748 + tun->nid = (nid_t) arg;
1750 + DBG(KERN_INFO "%s: nid owner set to %u\n", tun->dev->name, tun->nid);
1754 /* Only allow setting the type when the interface is down */
1756 diff -NurpP --minimal linux-2.6.29.1/fs/attr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/attr.c
1757 --- linux-2.6.29.1/fs/attr.c 2009-03-24 14:22:24.000000000 +0100
1758 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/attr.c 2009-03-24 14:48:34.000000000 +0100
1760 #include <linux/fcntl.h>
1761 #include <linux/quotaops.h>
1762 #include <linux/security.h>
1763 +#include <linux/proc_fs.h>
1764 +#include <linux/devpts_fs.h>
1765 +#include <linux/vs_tag.h>
1767 /* Taken over from the old code... */
1769 @@ -55,6 +58,10 @@ int inode_change_ok(struct inode *inode,
1770 if (!is_owner_or_cap(inode))
1774 + if (dx_permission(inode, MAY_WRITE))
1780 @@ -78,6 +85,8 @@ int inode_setattr(struct inode * inode,
1781 inode->i_uid = attr->ia_uid;
1782 if (ia_valid & ATTR_GID)
1783 inode->i_gid = attr->ia_gid;
1784 + if ((ia_valid & ATTR_TAG) && IS_TAGGED(inode))
1785 + inode->i_tag = attr->ia_tag;
1786 if (ia_valid & ATTR_ATIME)
1787 inode->i_atime = timespec_trunc(attr->ia_atime,
1788 inode->i_sb->s_time_gran);
1789 @@ -172,7 +181,8 @@ int notify_change(struct dentry * dentry
1790 error = inode_change_ok(inode, attr);
1792 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
1793 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid))
1794 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
1795 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag))
1796 error = DQUOT_TRANSFER(inode, attr) ? -EDQUOT : 0;
1798 error = inode_setattr(inode, attr);
1799 diff -NurpP --minimal linux-2.6.29.1/fs/binfmt_aout.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_aout.c
1800 --- linux-2.6.29.1/fs/binfmt_aout.c 2009-03-24 14:22:24.000000000 +0100
1801 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_aout.c 2009-03-24 14:48:34.000000000 +0100
1803 #include <linux/binfmts.h>
1804 #include <linux/personality.h>
1805 #include <linux/init.h>
1806 +#include <linux/vs_memory.h>
1808 #include <asm/system.h>
1809 #include <asm/uaccess.h>
1810 diff -NurpP --minimal linux-2.6.29.1/fs/binfmt_elf.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_elf.c
1811 --- linux-2.6.29.1/fs/binfmt_elf.c 2009-03-24 14:22:24.000000000 +0100
1812 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_elf.c 2009-03-24 14:48:34.000000000 +0100
1814 #include <linux/random.h>
1815 #include <linux/elf.h>
1816 #include <linux/utsname.h>
1817 +#include <linux/vs_memory.h>
1818 #include <asm/uaccess.h>
1819 #include <asm/param.h>
1820 #include <asm/page.h>
1821 diff -NurpP --minimal linux-2.6.29.1/fs/binfmt_flat.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_flat.c
1822 --- linux-2.6.29.1/fs/binfmt_flat.c 2009-03-24 14:22:24.000000000 +0100
1823 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_flat.c 2009-03-24 14:48:34.000000000 +0100
1825 #include <linux/init.h>
1826 #include <linux/flat.h>
1827 #include <linux/syscalls.h>
1828 +#include <linux/vs_memory.h>
1830 #include <asm/byteorder.h>
1831 #include <asm/system.h>
1832 diff -NurpP --minimal linux-2.6.29.1/fs/binfmt_som.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_som.c
1833 --- linux-2.6.29.1/fs/binfmt_som.c 2009-03-24 14:22:24.000000000 +0100
1834 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/binfmt_som.c 2009-03-24 14:48:34.000000000 +0100
1836 #include <linux/shm.h>
1837 #include <linux/personality.h>
1838 #include <linux/init.h>
1839 +#include <linux/vs_memory.h>
1841 #include <asm/uaccess.h>
1842 #include <asm/pgtable.h>
1843 diff -NurpP --minimal linux-2.6.29.1/fs/block_dev.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/block_dev.c
1844 --- linux-2.6.29.1/fs/block_dev.c 2009-03-24 14:22:24.000000000 +0100
1845 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/block_dev.c 2009-03-24 14:48:34.000000000 +0100
1847 #include <linux/uio.h>
1848 #include <linux/namei.h>
1849 #include <linux/log2.h>
1850 +#include <linux/vs_device.h>
1851 #include <asm/uaccess.h>
1852 #include "internal.h"
1854 @@ -392,6 +393,7 @@ struct block_device *bdget(dev_t dev)
1855 bdev->bd_invalidated = 0;
1856 inode->i_mode = S_IFBLK;
1857 inode->i_rdev = dev;
1858 + inode->i_mdev = dev;
1859 inode->i_bdev = bdev;
1860 inode->i_data.a_ops = &def_blk_aops;
1861 mapping_set_gfp_mask(&inode->i_data, GFP_USER);
1862 @@ -428,6 +430,11 @@ EXPORT_SYMBOL(bdput);
1863 static struct block_device *bd_acquire(struct inode *inode)
1865 struct block_device *bdev;
1868 + if (!vs_map_blkdev(inode->i_rdev, &mdev, DATTR_OPEN))
1870 + inode->i_mdev = mdev;
1872 spin_lock(&bdev_lock);
1873 bdev = inode->i_bdev;
1874 @@ -438,7 +445,7 @@ static struct block_device *bd_acquire(s
1876 spin_unlock(&bdev_lock);
1878 - bdev = bdget(inode->i_rdev);
1879 + bdev = bdget(mdev);
1881 spin_lock(&bdev_lock);
1882 if (!inode->i_bdev) {
1883 diff -NurpP --minimal linux-2.6.29.1/fs/char_dev.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/char_dev.c
1884 --- linux-2.6.29.1/fs/char_dev.c 2009-03-24 14:22:25.000000000 +0100
1885 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/char_dev.c 2009-03-24 14:48:35.000000000 +0100
1887 #include <linux/cdev.h>
1888 #include <linux/mutex.h>
1889 #include <linux/backing-dev.h>
1890 +#include <linux/vs_context.h>
1891 +#include <linux/vs_device.h>
1893 #include "internal.h"
1895 @@ -358,14 +360,21 @@ static int chrdev_open(struct inode *ino
1897 struct cdev *new = NULL;
1901 + if (!vs_map_chrdev(inode->i_rdev, &mdev, DATTR_OPEN))
1903 + inode->i_mdev = mdev;
1905 spin_lock(&cdev_lock);
1908 struct kobject *kobj;
1911 spin_unlock(&cdev_lock);
1912 - kobj = kobj_lookup(cdev_map, inode->i_rdev, &idx);
1914 + kobj = kobj_lookup(cdev_map, mdev, &idx);
1917 new = container_of(kobj, struct cdev, kobj);
1918 diff -NurpP --minimal linux-2.6.29.1/fs/dcache.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/dcache.c
1919 --- linux-2.6.29.1/fs/dcache.c 2009-03-24 14:22:25.000000000 +0100
1920 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/dcache.c 2009-03-24 14:48:35.000000000 +0100
1922 #include <linux/seqlock.h>
1923 #include <linux/swap.h>
1924 #include <linux/bootmem.h>
1925 +#include <linux/vs_limit.h>
1926 #include "internal.h"
1928 int sysctl_vfs_cache_pressure __read_mostly = 100;
1929 @@ -229,6 +230,8 @@ repeat:
1933 + vx_dentry_dec(dentry);
1936 * AV: ->d_delete() is _NOT_ allowed to block now.
1938 @@ -320,6 +323,7 @@ static inline struct dentry * __dget_loc
1940 atomic_inc(&dentry->d_count);
1941 dentry_lru_del_init(dentry);
1942 + vx_dentry_inc(dentry);
1946 @@ -918,6 +922,9 @@ struct dentry *d_alloc(struct dentry * p
1947 struct dentry *dentry;
1950 + if (!vx_dentry_avail(1))
1953 dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL);
1956 @@ -963,6 +970,7 @@ struct dentry *d_alloc(struct dentry * p
1958 list_add(&dentry->d_u.d_child, &parent->d_subdirs);
1959 dentry_stat.nr_dentry++;
1960 + vx_dentry_inc(dentry);
1961 spin_unlock(&dcache_lock);
1964 @@ -1418,6 +1426,7 @@ struct dentry * __d_lookup(struct dentry
1967 atomic_inc(&dentry->d_count);
1968 + vx_dentry_inc(dentry);
1970 spin_unlock(&dentry->d_lock);
1972 diff -NurpP --minimal linux-2.6.29.1/fs/devpts/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/devpts/inode.c
1973 --- linux-2.6.29.1/fs/devpts/inode.c 2009-03-24 14:22:25.000000000 +0100
1974 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/devpts/inode.c 2009-03-24 15:32:47.000000000 +0100
1976 #include <linux/tty.h>
1977 #include <linux/mutex.h>
1978 #include <linux/idr.h>
1979 +#include <linux/magic.h>
1980 #include <linux/devpts_fs.h>
1981 #include <linux/parser.h>
1982 #include <linux/fsnotify.h>
1983 #include <linux/seq_file.h>
1985 -#define DEVPTS_SUPER_MAGIC 0x1cd1
1986 +#include <linux/vs_base.h>
1988 #define DEVPTS_DEFAULT_MODE 0600
1991 #define DEVPTS_DEFAULT_PTMX_MODE 0000
1992 #define PTMX_MINOR 2
1994 +static int devpts_permission(struct inode *inode, int mask)
1996 + int ret = -EACCES;
1998 + /* devpts is xid tagged */
1999 + if (vx_check((xid_t)inode->i_tag, VS_WATCH_P | VS_IDENT))
2000 + ret = generic_permission(inode, mask, NULL);
2004 +static struct inode_operations devpts_file_inode_operations = {
2005 + .permission = devpts_permission,
2008 extern int pty_limit; /* Config limit on Unix98 ptys */
2009 static DEFINE_MUTEX(allocated_ptys_lock);
2011 @@ -254,6 +268,25 @@ static int devpts_show_options(struct se
2015 +static int devpts_filter(struct dentry *de)
2017 + /* devpts is xid tagged */
2018 + return vx_check((xid_t)de->d_inode->i_tag, VS_WATCH_P | VS_IDENT);
2021 +static int devpts_readdir(struct file * filp, void * dirent, filldir_t filldir)
2023 + return dcache_readdir_filter(filp, dirent, filldir, devpts_filter);
2026 +static struct file_operations devpts_dir_operations = {
2027 + .open = dcache_dir_open,
2028 + .release = dcache_dir_close,
2029 + .llseek = dcache_dir_lseek,
2030 + .read = generic_read_dir,
2031 + .readdir = devpts_readdir,
2034 static const struct super_operations devpts_sops = {
2035 .statfs = simple_statfs,
2036 .remount_fs = devpts_remount,
2037 @@ -293,12 +326,15 @@ devpts_fill_super(struct super_block *s,
2038 inode = new_inode(s);
2043 inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2044 inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO | S_IWUSR;
2045 inode->i_op = &simple_dir_inode_operations;
2046 - inode->i_fop = &simple_dir_operations;
2047 + inode->i_fop = &devpts_dir_operations;
2049 + /* devpts is xid tagged */
2050 + inode->i_tag = (tag_t)vx_current_xid();
2052 s->s_root = d_alloc_root(inode);
2054 @@ -479,6 +515,7 @@ static int init_pts_mount(struct file_sy
2059 static int devpts_get_sb(struct file_system_type *fs_type,
2060 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
2062 @@ -590,6 +627,9 @@ int devpts_pty_new(struct inode *ptmx_in
2063 inode->i_gid = opts->setgid ? opts->gid : current_fsgid();
2064 inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2065 init_special_inode(inode, S_IFCHR|opts->mode, device);
2066 + /* devpts is xid tagged */
2067 + inode->i_tag = (tag_t)vx_current_xid();
2068 + inode->i_op = &devpts_file_inode_operations;
2069 inode->i_private = tty;
2070 tty->driver_data = inode;
2072 diff -NurpP --minimal linux-2.6.29.1/fs/exec.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/exec.c
2073 --- linux-2.6.29.1/fs/exec.c 2009-03-24 14:22:25.000000000 +0100
2074 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/exec.c 2009-03-24 15:34:21.000000000 +0100
2075 @@ -257,7 +257,9 @@ static int __bprm_mm_init(struct linux_b
2079 - mm->stack_vm = mm->total_vm = 1;
2081 + vx_vmpages_inc(mm);
2083 up_write(&mm->mmap_sem);
2084 bprm->p = vma->vm_end - sizeof(void *);
2086 @@ -1450,7 +1452,7 @@ static int format_corename(char *corenam
2087 /* UNIX time of coredump */
2090 - do_gettimeofday(&tv);
2091 + vx_gettimeofday(&tv);
2092 rc = snprintf(out_ptr, out_end - out_ptr,
2094 if (rc > out_end - out_ptr)
2095 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/balloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/balloc.c
2096 --- linux-2.6.29.1/fs/ext2/balloc.c 2009-03-24 14:22:25.000000000 +0100
2097 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/balloc.c 2009-03-24 14:48:35.000000000 +0100
2099 #include <linux/sched.h>
2100 #include <linux/buffer_head.h>
2101 #include <linux/capability.h>
2102 +#include <linux/vs_dlimit.h>
2103 +#include <linux/vs_tag.h>
2106 * balloc.c contains the blocks allocation and deallocation routines
2107 @@ -569,6 +571,7 @@ do_more:
2111 + DLIMIT_FREE_BLOCK(inode, freed);
2112 release_blocks(sb, freed);
2113 DQUOT_FREE_BLOCK(inode, freed);
2115 @@ -701,7 +704,6 @@ ext2_try_to_allocate(struct super_block
2117 end = EXT2_BLOCKS_PER_GROUP(sb);
2120 BUG_ON(start > EXT2_BLOCKS_PER_GROUP(sb));
2123 @@ -1251,6 +1253,11 @@ ext2_fsblk_t ext2_new_blocks(struct inod
2127 + if (DLIMIT_ALLOC_BLOCK(inode, num)) {
2129 + DQUOT_FREE_BLOCK(inode, num);
2134 es = EXT2_SB(sb)->s_es;
2135 @@ -1409,6 +1416,7 @@ allocated:
2139 + DLIMIT_FREE_BLOCK(inode, *count-num);
2140 DQUOT_FREE_BLOCK(inode, *count-num);
2143 @@ -1419,8 +1427,10 @@ out:
2145 * Undo the block allocation
2147 - if (!performed_allocation)
2148 + if (!performed_allocation) {
2149 + DLIMIT_FREE_BLOCK(inode, *count);
2150 DQUOT_FREE_BLOCK(inode, *count);
2155 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/ext2.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ext2.h
2156 --- linux-2.6.29.1/fs/ext2/ext2.h 2008-12-25 00:26:37.000000000 +0100
2157 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ext2.h 2009-02-22 22:54:25.000000000 +0100
2158 @@ -170,6 +170,7 @@ extern const struct file_operations ext2
2159 extern const struct address_space_operations ext2_aops;
2160 extern const struct address_space_operations ext2_aops_xip;
2161 extern const struct address_space_operations ext2_nobh_aops;
2162 +extern int ext2_sync_flags(struct inode *inode);
2165 extern const struct inode_operations ext2_dir_inode_operations;
2166 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/file.c
2167 --- linux-2.6.29.1/fs/ext2/file.c 2008-12-25 00:26:37.000000000 +0100
2168 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/file.c 2009-02-22 22:54:25.000000000 +0100
2169 @@ -87,4 +87,5 @@ const struct inode_operations ext2_file_
2170 .setattr = ext2_setattr,
2171 .permission = ext2_permission,
2172 .fiemap = ext2_fiemap,
2173 + .sync_flags = ext2_sync_flags,
2175 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/ialloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ialloc.c
2176 --- linux-2.6.29.1/fs/ext2/ialloc.c 2009-03-24 14:22:25.000000000 +0100
2177 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ialloc.c 2009-03-24 15:36:12.000000000 +0100
2179 #include <linux/backing-dev.h>
2180 #include <linux/buffer_head.h>
2181 #include <linux/random.h>
2182 +#include <linux/vs_dlimit.h>
2183 +#include <linux/vs_tag.h>
2187 @@ -123,6 +125,7 @@ void ext2_free_inode (struct inode * ino
2188 ext2_xattr_delete_inode(inode);
2189 DQUOT_FREE_INODE(inode);
2191 + DLIMIT_FREE_INODE(inode);
2194 es = EXT2_SB(sb)->s_es;
2195 @@ -454,6 +457,11 @@ struct inode *ext2_new_inode(struct inod
2197 return ERR_PTR(-ENOMEM);
2199 + inode->i_tag = dx_current_fstag(sb);
2200 + if (DLIMIT_ALLOC_INODE(inode)) {
2207 @@ -609,6 +617,7 @@ fail_free_drop:
2211 + DLIMIT_FREE_INODE(inode);
2212 inode->i_flags |= S_NOQUOTA;
2214 unlock_new_inode(inode);
2215 @@ -616,6 +625,8 @@ fail_drop:
2216 return ERR_PTR(err);
2219 + DLIMIT_FREE_INODE(inode);
2221 make_bad_inode(inode);
2223 return ERR_PTR(err);
2224 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/inode.c
2225 --- linux-2.6.29.1/fs/ext2/inode.c 2009-03-24 14:22:25.000000000 +0100
2226 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/inode.c 2009-03-24 15:41:38.000000000 +0100
2228 #include <linux/mpage.h>
2229 #include <linux/fiemap.h>
2230 #include <linux/namei.h>
2231 +#include <linux/vs_tag.h>
2235 @@ -1018,7 +1019,7 @@ void ext2_truncate(struct inode *inode)
2237 if (ext2_inode_is_fast_symlink(inode))
2239 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2240 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2243 blocksize = inode->i_sb->s_blocksize;
2244 @@ -1156,38 +1157,72 @@ void ext2_set_inode_flags(struct inode *
2246 unsigned int flags = EXT2_I(inode)->i_flags;
2248 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2249 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2250 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2253 + if (flags & EXT2_IMMUTABLE_FL)
2254 + inode->i_flags |= S_IMMUTABLE;
2255 + if (flags & EXT2_IXUNLINK_FL)
2256 + inode->i_flags |= S_IXUNLINK;
2258 if (flags & EXT2_SYNC_FL)
2259 inode->i_flags |= S_SYNC;
2260 if (flags & EXT2_APPEND_FL)
2261 inode->i_flags |= S_APPEND;
2262 - if (flags & EXT2_IMMUTABLE_FL)
2263 - inode->i_flags |= S_IMMUTABLE;
2264 if (flags & EXT2_NOATIME_FL)
2265 inode->i_flags |= S_NOATIME;
2266 if (flags & EXT2_DIRSYNC_FL)
2267 inode->i_flags |= S_DIRSYNC;
2269 + inode->i_vflags &= ~(V_BARRIER | V_COW);
2271 + if (flags & EXT2_BARRIER_FL)
2272 + inode->i_vflags |= V_BARRIER;
2273 + if (flags & EXT2_COW_FL)
2274 + inode->i_vflags |= V_COW;
2277 /* Propagate flags from i_flags to EXT2_I(inode)->i_flags */
2278 void ext2_get_inode_flags(struct ext2_inode_info *ei)
2280 unsigned int flags = ei->vfs_inode.i_flags;
2281 + unsigned int vflags = ei->vfs_inode.i_vflags;
2283 + ei->i_flags &= ~(EXT2_SYNC_FL | EXT2_APPEND_FL |
2284 + EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL |
2285 + EXT2_NOATIME_FL | EXT2_DIRSYNC_FL |
2286 + EXT2_BARRIER_FL | EXT2_COW_FL);
2288 + if (flags & S_IMMUTABLE)
2289 + ei->i_flags |= EXT2_IMMUTABLE_FL;
2290 + if (flags & S_IXUNLINK)
2291 + ei->i_flags |= EXT2_IXUNLINK_FL;
2293 - ei->i_flags &= ~(EXT2_SYNC_FL|EXT2_APPEND_FL|
2294 - EXT2_IMMUTABLE_FL|EXT2_NOATIME_FL|EXT2_DIRSYNC_FL);
2296 ei->i_flags |= EXT2_SYNC_FL;
2297 if (flags & S_APPEND)
2298 ei->i_flags |= EXT2_APPEND_FL;
2299 - if (flags & S_IMMUTABLE)
2300 - ei->i_flags |= EXT2_IMMUTABLE_FL;
2301 if (flags & S_NOATIME)
2302 ei->i_flags |= EXT2_NOATIME_FL;
2303 if (flags & S_DIRSYNC)
2304 ei->i_flags |= EXT2_DIRSYNC_FL;
2306 + if (vflags & V_BARRIER)
2307 + ei->i_flags |= EXT2_BARRIER_FL;
2308 + if (vflags & V_COW)
2309 + ei->i_flags |= EXT2_COW_FL;
2312 +int ext2_sync_flags(struct inode *inode)
2314 + ext2_get_inode_flags(EXT2_I(inode));
2315 + inode->i_ctime = CURRENT_TIME;
2316 + mark_inode_dirty(inode);
2321 struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
2323 struct ext2_inode_info *ei;
2324 @@ -1195,6 +1230,8 @@ struct inode *ext2_iget (struct super_bl
2325 struct ext2_inode *raw_inode;
2326 struct inode *inode;
2332 inode = iget_locked(sb, ino);
2333 @@ -1217,12 +1254,17 @@ struct inode *ext2_iget (struct super_bl
2336 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2337 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2338 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2339 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2340 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2341 if (!(test_opt (inode->i_sb, NO_UID32))) {
2342 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2343 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2344 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2345 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2347 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2348 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2349 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2350 + le16_to_cpu(raw_inode->i_raw_tag));
2352 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2353 inode->i_size = le32_to_cpu(raw_inode->i_size);
2354 inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2355 @@ -1320,8 +1362,8 @@ static int ext2_update_inode(struct inod
2356 struct ext2_inode_info *ei = EXT2_I(inode);
2357 struct super_block *sb = inode->i_sb;
2358 ino_t ino = inode->i_ino;
2359 - uid_t uid = inode->i_uid;
2360 - gid_t gid = inode->i_gid;
2361 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2362 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2363 struct buffer_head * bh;
2364 struct ext2_inode * raw_inode = ext2_get_inode(sb, ino, &bh);
2366 @@ -1357,6 +1399,9 @@ static int ext2_update_inode(struct inod
2367 raw_inode->i_uid_high = 0;
2368 raw_inode->i_gid_high = 0;
2370 +#ifdef CONFIG_TAGGING_INTERN
2371 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2373 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2374 raw_inode->i_size = cpu_to_le32(inode->i_size);
2375 raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2376 @@ -1443,7 +1488,8 @@ int ext2_setattr(struct dentry *dentry,
2379 if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
2380 - (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
2381 + (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
2382 + (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
2383 error = DQUOT_TRANSFER(inode, iattr) ? -EDQUOT : 0;
2386 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ioctl.c
2387 --- linux-2.6.29.1/fs/ext2/ioctl.c 2009-03-24 14:22:25.000000000 +0100
2388 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/ioctl.c 2009-03-24 14:48:35.000000000 +0100
2390 #include <linux/compat.h>
2391 #include <linux/mount.h>
2392 #include <linux/smp_lock.h>
2393 +#include <linux/mount.h>
2394 #include <asm/current.h>
2395 #include <asm/uaccess.h>
2397 @@ -52,6 +53,11 @@ long ext2_ioctl(struct file *filp, unsig
2399 flags = ext2_mask_flags(inode->i_mode, flags);
2401 + if (IS_BARRIER(inode)) {
2402 + vxwprintk_task(1, "messing with the barrier.");
2406 mutex_lock(&inode->i_mutex);
2407 /* Is it quota file? Do not allow user to mess with it */
2408 if (IS_NOQUOTA(inode)) {
2409 @@ -67,7 +73,9 @@ long ext2_ioctl(struct file *filp, unsig
2411 * This test looks nicer. Thanks to Pauline Middelink
2413 - if ((flags ^ oldflags) & (EXT2_APPEND_FL | EXT2_IMMUTABLE_FL)) {
2414 + if ((oldflags & EXT2_IMMUTABLE_FL) ||
2415 + ((flags ^ oldflags) & (EXT2_APPEND_FL |
2416 + EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL))) {
2417 if (!capable(CAP_LINUX_IMMUTABLE)) {
2418 mutex_unlock(&inode->i_mutex);
2420 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/namei.c
2421 --- linux-2.6.29.1/fs/ext2/namei.c 2009-03-24 14:22:25.000000000 +0100
2422 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/namei.c 2009-03-24 14:48:35.000000000 +0100
2426 #include <linux/pagemap.h>
2427 +#include <linux/vs_tag.h>
2431 @@ -68,6 +69,7 @@ static struct dentry *ext2_lookup(struct
2432 inode = ext2_iget(dir->i_sb, ino);
2434 return ERR_CAST(inode);
2435 + dx_propagate_tag(nd, inode);
2437 return d_splice_alias(inode, dentry);
2439 @@ -388,6 +390,7 @@ const struct inode_operations ext2_dir_i
2441 .setattr = ext2_setattr,
2442 .permission = ext2_permission,
2443 + .sync_flags = ext2_sync_flags,
2446 const struct inode_operations ext2_special_inode_operations = {
2447 @@ -399,4 +402,5 @@ const struct inode_operations ext2_speci
2449 .setattr = ext2_setattr,
2450 .permission = ext2_permission,
2451 + .sync_flags = ext2_sync_flags,
2453 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/super.c
2454 --- linux-2.6.29.1/fs/ext2/super.c 2009-03-24 14:22:25.000000000 +0100
2455 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/super.c 2009-03-24 14:48:35.000000000 +0100
2456 @@ -391,7 +391,8 @@ enum {
2457 Opt_err_ro, Opt_nouid32, Opt_nocheck, Opt_debug,
2458 Opt_oldalloc, Opt_orlov, Opt_nobh, Opt_user_xattr, Opt_nouser_xattr,
2459 Opt_acl, Opt_noacl, Opt_xip, Opt_ignore, Opt_err, Opt_quota,
2460 - Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation
2461 + Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation,
2462 + Opt_tag, Opt_notag, Opt_tagid
2465 static const match_table_t tokens = {
2466 @@ -419,6 +420,9 @@ static const match_table_t tokens = {
2468 {Opt_noacl, "noacl"},
2471 + {Opt_notag, "notag"},
2472 + {Opt_tagid, "tagid=%u"},
2473 {Opt_grpquota, "grpquota"},
2474 {Opt_ignore, "noquota"},
2475 {Opt_quota, "quota"},
2476 @@ -489,6 +493,20 @@ static int parse_options (char * options
2478 set_opt (sbi->s_mount_opt, NO_UID32);
2480 +#ifndef CONFIG_TAGGING_NONE
2482 + set_opt (sbi->s_mount_opt, TAGGED);
2485 + clear_opt (sbi->s_mount_opt, TAGGED);
2488 +#ifdef CONFIG_PROPAGATE
2491 + set_opt (sbi->s_mount_opt, TAGGED);
2495 clear_opt (sbi->s_mount_opt, CHECK);
2497 @@ -838,6 +856,8 @@ static int ext2_fill_super(struct super_
2498 if (!parse_options ((char *) data, sbi))
2501 + if (EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_TAGGED)
2502 + sb->s_flags |= MS_TAGGED;
2503 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
2504 ((EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ?
2506 @@ -1170,6 +1190,13 @@ static int ext2_remount (struct super_bl
2510 + if ((sbi->s_mount_opt & EXT2_MOUNT_TAGGED) &&
2511 + !(sb->s_flags & MS_TAGGED)) {
2512 + printk("EXT2-fs: %s: tagging not permitted on remount.\n",
2517 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
2518 ((sbi->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
2520 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/symlink.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/symlink.c
2521 --- linux-2.6.29.1/fs/ext2/symlink.c 2008-12-25 00:26:37.000000000 +0100
2522 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/symlink.c 2009-02-22 22:54:25.000000000 +0100
2523 @@ -38,6 +38,7 @@ const struct inode_operations ext2_symli
2524 .listxattr = ext2_listxattr,
2525 .removexattr = generic_removexattr,
2527 + .sync_flags = ext2_sync_flags,
2530 const struct inode_operations ext2_fast_symlink_inode_operations = {
2531 @@ -49,4 +50,5 @@ const struct inode_operations ext2_fast_
2532 .listxattr = ext2_listxattr,
2533 .removexattr = generic_removexattr,
2535 + .sync_flags = ext2_sync_flags,
2537 diff -NurpP --minimal linux-2.6.29.1/fs/ext2/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/xattr.c
2538 --- linux-2.6.29.1/fs/ext2/xattr.c 2008-12-25 00:26:37.000000000 +0100
2539 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext2/xattr.c 2009-02-22 22:54:25.000000000 +0100
2541 #include <linux/mbcache.h>
2542 #include <linux/quotaops.h>
2543 #include <linux/rwsem.h>
2544 +#include <linux/vs_dlimit.h>
2548 @@ -641,8 +642,12 @@ ext2_xattr_set2(struct inode *inode, str
2550 ea_bdebug(new_bh, "reusing block");
2553 + if (DLIMIT_ALLOC_BLOCK(inode, 1))
2556 if (DQUOT_ALLOC_BLOCK(inode, 1)) {
2557 + DLIMIT_FREE_BLOCK(inode, 1);
2558 unlock_buffer(new_bh);
2561 @@ -731,6 +736,7 @@ ext2_xattr_set2(struct inode *inode, str
2562 le32_add_cpu(&HDR(old_bh)->h_refcount, -1);
2564 mb_cache_entry_release(ce);
2565 + DLIMIT_FREE_BLOCK(inode, 1);
2566 DQUOT_FREE_BLOCK(inode, 1);
2567 mark_buffer_dirty(old_bh);
2568 ea_bdebug(old_bh, "refcount now=%d",
2569 @@ -794,6 +800,7 @@ ext2_xattr_delete_inode(struct inode *in
2570 mark_buffer_dirty(bh);
2572 sync_dirty_buffer(bh);
2573 + DLIMIT_FREE_BLOCK(inode, 1);
2574 DQUOT_FREE_BLOCK(inode, 1);
2576 EXT2_I(inode)->i_file_acl = 0;
2577 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/balloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/balloc.c
2578 --- linux-2.6.29.1/fs/ext3/balloc.c 2009-03-24 14:22:25.000000000 +0100
2579 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/balloc.c 2009-03-25 00:53:20.000000000 +0100
2581 #include <linux/ext3_jbd.h>
2582 #include <linux/quotaops.h>
2583 #include <linux/buffer_head.h>
2584 +#include <linux/vs_dlimit.h>
2585 +#include <linux/vs_tag.h>
2588 * balloc.c contains the blocks allocation and deallocation routines
2589 @@ -675,8 +677,10 @@ void ext3_free_blocks(handle_t *handle,
2592 ext3_free_blocks_sb(handle, sb, block, count, &dquot_freed_blocks);
2593 - if (dquot_freed_blocks)
2594 + if (dquot_freed_blocks) {
2595 + DLIMIT_FREE_BLOCK(inode, dquot_freed_blocks);
2596 DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
2601 @@ -1415,18 +1419,33 @@ out:
2603 * Check if filesystem has at least 1 free block available for allocation.
2605 -static int ext3_has_free_blocks(struct ext3_sb_info *sbi)
2606 +static int ext3_has_free_blocks(struct super_block *sb)
2608 - ext3_fsblk_t free_blocks, root_blocks;
2609 + struct ext3_sb_info *sbi = EXT3_SB(sb);
2610 + unsigned long long free_blocks, root_blocks;
2613 free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
2614 root_blocks = le32_to_cpu(sbi->s_es->s_r_blocks_count);
2615 - if (free_blocks < root_blocks + 1 && !capable(CAP_SYS_RESOURCE) &&
2617 + vxdprintk(VXD_CBIT(dlim, 3),
2618 + "ext3_has_free_blocks(%p): free=%llu, root=%llu",
2619 + sb, free_blocks, root_blocks);
2621 + DLIMIT_ADJUST_BLOCK(sb, dx_current_tag(), &free_blocks, &root_blocks);
2623 + cond = (free_blocks < root_blocks + 1 &&
2624 + !capable(CAP_SYS_RESOURCE) &&
2625 sbi->s_resuid != current_fsuid() &&
2626 - (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))) {
2630 + (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid)));
2632 + vxdprintk(VXD_CBIT(dlim, 3),
2633 + "ext3_has_free_blocks(%p): %llu<%llu+1, %c, %u!=%u r=%d",
2634 + sb, free_blocks, root_blocks,
2635 + !capable(CAP_SYS_RESOURCE)?'1':'0',
2636 + sbi->s_resuid, current_fsuid(), cond?0:1);
2638 + return (cond ? 0 : 1);
2642 @@ -1443,7 +1462,7 @@ static int ext3_has_free_blocks(struct e
2644 int ext3_should_retry_alloc(struct super_block *sb, int *retries)
2646 - if (!ext3_has_free_blocks(EXT3_SB(sb)) || (*retries)++ > 3)
2647 + if (!ext3_has_free_blocks(sb) || (*retries)++ > 3)
2650 jbd_debug(1, "%s: retrying operation after ENOSPC\n", sb->s_id);
2651 @@ -1506,6 +1525,8 @@ ext3_fsblk_t ext3_new_blocks(handle_t *h
2655 + if (DLIMIT_ALLOC_BLOCK(inode, num))
2659 es = EXT3_SB(sb)->s_es;
2660 @@ -1522,7 +1543,7 @@ ext3_fsblk_t ext3_new_blocks(handle_t *h
2661 if (block_i && ((windowsz = block_i->rsv_window_node.rsv_goal_size) > 0))
2662 my_rsv = &block_i->rsv_window_node;
2664 - if (!ext3_has_free_blocks(sbi)) {
2665 + if (!ext3_has_free_blocks(sb)) {
2669 @@ -1715,12 +1736,16 @@ allocated:
2672 DQUOT_FREE_BLOCK(inode, *count-num);
2673 + DLIMIT_FREE_BLOCK(inode, *count-num);
2680 + if (!performed_allocation)
2681 + DLIMIT_FREE_BLOCK(inode, *count);
2685 ext3_std_error(sb, fatal);
2686 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/file.c
2687 --- linux-2.6.29.1/fs/ext3/file.c 2008-12-25 00:26:37.000000000 +0100
2688 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/file.c 2009-02-22 22:54:25.000000000 +0100
2689 @@ -135,5 +135,6 @@ const struct inode_operations ext3_file_
2691 .permission = ext3_permission,
2692 .fiemap = ext3_fiemap,
2693 + .sync_flags = ext3_sync_flags,
2696 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/ialloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/ialloc.c
2697 --- linux-2.6.29.1/fs/ext3/ialloc.c 2009-03-24 14:22:25.000000000 +0100
2698 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/ialloc.c 2009-03-24 15:44:06.000000000 +0100
2700 #include <linux/buffer_head.h>
2701 #include <linux/random.h>
2702 #include <linux/bitops.h>
2703 +#include <linux/vs_dlimit.h>
2704 +#include <linux/vs_tag.h>
2706 #include <asm/byteorder.h>
2708 @@ -127,6 +129,7 @@ void ext3_free_inode (handle_t *handle,
2709 ext3_xattr_delete_inode(handle, inode);
2710 DQUOT_FREE_INODE(inode);
2712 + DLIMIT_FREE_INODE(inode);
2714 is_directory = S_ISDIR(inode->i_mode);
2716 @@ -440,6 +443,12 @@ struct inode *ext3_new_inode(handle_t *h
2717 inode = new_inode(sb);
2719 return ERR_PTR(-ENOMEM);
2721 + inode->i_tag = dx_current_fstag(sb);
2722 + if (DLIMIT_ALLOC_INODE(inode)) {
2729 @@ -613,6 +622,8 @@ got:
2731 ext3_std_error(sb, err);
2733 + DLIMIT_FREE_INODE(inode);
2738 @@ -624,6 +635,7 @@ fail_free_drop:
2742 + DLIMIT_FREE_INODE(inode);
2743 inode->i_flags |= S_NOQUOTA;
2745 unlock_new_inode(inode);
2746 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/inode.c
2747 --- linux-2.6.29.1/fs/ext3/inode.c 2009-03-24 14:22:25.000000000 +0100
2748 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/inode.c 2009-03-24 15:44:29.000000000 +0100
2750 #include <linux/bio.h>
2751 #include <linux/fiemap.h>
2752 #include <linux/namei.h>
2753 +#include <linux/vs_tag.h>
2757 @@ -2288,7 +2289,7 @@ static void ext3_free_branches(handle_t
2759 int ext3_can_truncate(struct inode *inode)
2761 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2762 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2764 if (S_ISREG(inode->i_mode))
2766 @@ -2662,36 +2663,84 @@ void ext3_set_inode_flags(struct inode *
2768 unsigned int flags = EXT3_I(inode)->i_flags;
2770 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2771 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2772 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2774 + if (flags & EXT3_IMMUTABLE_FL)
2775 + inode->i_flags |= S_IMMUTABLE;
2776 + if (flags & EXT3_IXUNLINK_FL)
2777 + inode->i_flags |= S_IXUNLINK;
2779 if (flags & EXT3_SYNC_FL)
2780 inode->i_flags |= S_SYNC;
2781 if (flags & EXT3_APPEND_FL)
2782 inode->i_flags |= S_APPEND;
2783 - if (flags & EXT3_IMMUTABLE_FL)
2784 - inode->i_flags |= S_IMMUTABLE;
2785 if (flags & EXT3_NOATIME_FL)
2786 inode->i_flags |= S_NOATIME;
2787 if (flags & EXT3_DIRSYNC_FL)
2788 inode->i_flags |= S_DIRSYNC;
2790 + inode->i_vflags &= ~(V_BARRIER | V_COW);
2792 + if (flags & EXT3_BARRIER_FL)
2793 + inode->i_vflags |= V_BARRIER;
2794 + if (flags & EXT3_COW_FL)
2795 + inode->i_vflags |= V_COW;
2798 /* Propagate flags from i_flags to EXT3_I(inode)->i_flags */
2799 void ext3_get_inode_flags(struct ext3_inode_info *ei)
2801 unsigned int flags = ei->vfs_inode.i_flags;
2802 + unsigned int vflags = ei->vfs_inode.i_vflags;
2804 + ei->i_flags &= ~(EXT3_SYNC_FL | EXT3_APPEND_FL |
2805 + EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL |
2806 + EXT3_NOATIME_FL | EXT3_DIRSYNC_FL |
2807 + EXT3_BARRIER_FL | EXT3_COW_FL);
2809 + if (flags & S_IMMUTABLE)
2810 + ei->i_flags |= EXT3_IMMUTABLE_FL;
2811 + if (flags & S_IXUNLINK)
2812 + ei->i_flags |= EXT3_IXUNLINK_FL;
2814 - ei->i_flags &= ~(EXT3_SYNC_FL|EXT3_APPEND_FL|
2815 - EXT3_IMMUTABLE_FL|EXT3_NOATIME_FL|EXT3_DIRSYNC_FL);
2817 ei->i_flags |= EXT3_SYNC_FL;
2818 if (flags & S_APPEND)
2819 ei->i_flags |= EXT3_APPEND_FL;
2820 - if (flags & S_IMMUTABLE)
2821 - ei->i_flags |= EXT3_IMMUTABLE_FL;
2822 if (flags & S_NOATIME)
2823 ei->i_flags |= EXT3_NOATIME_FL;
2824 if (flags & S_DIRSYNC)
2825 ei->i_flags |= EXT3_DIRSYNC_FL;
2827 + if (vflags & V_BARRIER)
2828 + ei->i_flags |= EXT3_BARRIER_FL;
2829 + if (vflags & V_COW)
2830 + ei->i_flags |= EXT3_COW_FL;
2833 +int ext3_sync_flags(struct inode *inode)
2835 + struct ext3_iloc iloc;
2839 + handle = ext3_journal_start(inode, 1);
2840 + if (IS_ERR(handle))
2841 + return PTR_ERR(handle);
2842 + if (IS_SYNC(inode))
2843 + handle->h_sync = 1;
2844 + err = ext3_reserve_inode_write(handle, inode, &iloc);
2848 + ext3_get_inode_flags(EXT3_I(inode));
2849 + inode->i_ctime = CURRENT_TIME;
2851 + err = ext3_mark_iloc_dirty(handle, inode, &iloc);
2853 + ext3_journal_stop(handle);
2857 struct inode *ext3_iget(struct super_block *sb, unsigned long ino)
2858 @@ -2703,6 +2752,8 @@ struct inode *ext3_iget(struct super_blo
2859 struct inode *inode;
2865 inode = iget_locked(sb, ino);
2867 @@ -2723,12 +2774,17 @@ struct inode *ext3_iget(struct super_blo
2869 raw_inode = ext3_raw_inode(&iloc);
2870 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2871 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2872 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2873 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2874 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2875 if(!(test_opt (inode->i_sb, NO_UID32))) {
2876 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2877 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2878 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2879 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2881 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2882 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2883 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2884 + le16_to_cpu(raw_inode->i_raw_tag));
2886 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2887 inode->i_size = le32_to_cpu(raw_inode->i_size);
2888 inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2889 @@ -2859,6 +2915,8 @@ static int ext3_do_update_inode(handle_t
2890 struct ext3_inode *raw_inode = ext3_raw_inode(iloc);
2891 struct ext3_inode_info *ei = EXT3_I(inode);
2892 struct buffer_head *bh = iloc->bh;
2893 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2894 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2895 int err = 0, rc, block;
2897 /* For fields not not tracking in the in-memory inode,
2898 @@ -2869,29 +2927,32 @@ static int ext3_do_update_inode(handle_t
2899 ext3_get_inode_flags(ei);
2900 raw_inode->i_mode = cpu_to_le16(inode->i_mode);
2901 if(!(test_opt(inode->i_sb, NO_UID32))) {
2902 - raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
2903 - raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
2904 + raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
2905 + raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
2907 * Fix up interoperability with old kernels. Otherwise, old inodes get
2908 * re-used with the upper 16 bits of the uid/gid intact
2911 raw_inode->i_uid_high =
2912 - cpu_to_le16(high_16_bits(inode->i_uid));
2913 + cpu_to_le16(high_16_bits(uid));
2914 raw_inode->i_gid_high =
2915 - cpu_to_le16(high_16_bits(inode->i_gid));
2916 + cpu_to_le16(high_16_bits(gid));
2918 raw_inode->i_uid_high = 0;
2919 raw_inode->i_gid_high = 0;
2922 raw_inode->i_uid_low =
2923 - cpu_to_le16(fs_high2lowuid(inode->i_uid));
2924 + cpu_to_le16(fs_high2lowuid(uid));
2925 raw_inode->i_gid_low =
2926 - cpu_to_le16(fs_high2lowgid(inode->i_gid));
2927 + cpu_to_le16(fs_high2lowgid(gid));
2928 raw_inode->i_uid_high = 0;
2929 raw_inode->i_gid_high = 0;
2931 +#ifdef CONFIG_TAGGING_INTERN
2932 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2934 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2935 raw_inode->i_size = cpu_to_le32(ei->i_disksize);
2936 raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2937 @@ -3044,7 +3105,8 @@ int ext3_setattr(struct dentry *dentry,
2940 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
2941 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
2942 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
2943 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
2946 /* (user+group)*(old+new) structure, inode write (sb,
2947 @@ -3066,6 +3128,8 @@ int ext3_setattr(struct dentry *dentry,
2948 inode->i_uid = attr->ia_uid;
2949 if (attr->ia_valid & ATTR_GID)
2950 inode->i_gid = attr->ia_gid;
2951 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
2952 + inode->i_tag = attr->ia_tag;
2953 error = ext3_mark_inode_dirty(handle, inode);
2954 ext3_journal_stop(handle);
2956 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/ioctl.c
2957 --- linux-2.6.29.1/fs/ext3/ioctl.c 2009-03-24 14:22:25.000000000 +0100
2958 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/ioctl.c 2009-03-24 14:48:35.000000000 +0100
2962 #include <linux/fs.h>
2963 +#include <linux/mount.h>
2964 #include <linux/jbd.h>
2965 #include <linux/capability.h>
2966 #include <linux/ext3_fs.h>
2968 #include <linux/time.h>
2969 #include <linux/compat.h>
2970 #include <linux/smp_lock.h>
2971 +#include <linux/vs_tag.h>
2972 #include <asm/uaccess.h>
2974 int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
2975 @@ -55,6 +57,11 @@ int ext3_ioctl (struct inode * inode, st
2977 flags = ext3_mask_flags(inode->i_mode, flags);
2979 + if (IS_BARRIER(inode)) {
2980 + vxwprintk_task(1, "messing with the barrier.");
2984 mutex_lock(&inode->i_mutex);
2985 /* Is it quota file? Do not allow user to mess with it */
2986 if (IS_NOQUOTA(inode)) {
2987 @@ -73,7 +80,9 @@ int ext3_ioctl (struct inode * inode, st
2989 * This test looks nicer. Thanks to Pauline Middelink
2991 - if ((flags ^ oldflags) & (EXT3_APPEND_FL | EXT3_IMMUTABLE_FL)) {
2992 + if ((oldflags & EXT3_IMMUTABLE_FL) ||
2993 + ((flags ^ oldflags) & (EXT3_APPEND_FL |
2994 + EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL))) {
2995 if (!capable(CAP_LINUX_IMMUTABLE)) {
2996 mutex_unlock(&inode->i_mutex);
2998 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/namei.c
2999 --- linux-2.6.29.1/fs/ext3/namei.c 2009-03-24 14:22:25.000000000 +0100
3000 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/namei.c 2009-03-24 14:48:35.000000000 +0100
3002 #include <linux/quotaops.h>
3003 #include <linux/buffer_head.h>
3004 #include <linux/bio.h>
3005 +#include <linux/vs_tag.h>
3009 @@ -912,6 +913,7 @@ restart:
3011 ll_rw_block(READ_META, 1, &bh);
3013 + dx_propagate_tag(nd, inode);
3015 if ((bh = bh_use[ra_ptr++]) == NULL)
3017 @@ -2433,6 +2435,7 @@ const struct inode_operations ext3_dir_i
3018 .removexattr = generic_removexattr,
3020 .permission = ext3_permission,
3021 + .sync_flags = ext3_sync_flags,
3024 const struct inode_operations ext3_special_inode_operations = {
3025 @@ -2444,4 +2447,5 @@ const struct inode_operations ext3_speci
3026 .removexattr = generic_removexattr,
3028 .permission = ext3_permission,
3029 + .sync_flags = ext3_sync_flags,
3031 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/super.c
3032 --- linux-2.6.29.1/fs/ext3/super.c 2009-03-24 14:22:25.000000000 +0100
3033 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/super.c 2009-03-24 14:48:35.000000000 +0100
3034 @@ -790,7 +790,7 @@ enum {
3035 Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
3036 Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
3037 Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
3039 + Opt_grpquota, Opt_tag, Opt_notag, Opt_tagid
3042 static const match_table_t tokens = {
3043 @@ -843,6 +843,9 @@ static const match_table_t tokens = {
3044 {Opt_usrquota, "usrquota"},
3045 {Opt_barrier, "barrier=%u"},
3046 {Opt_resize, "resize"},
3048 + {Opt_notag, "notag"},
3049 + {Opt_tagid, "tagid=%u"},
3053 @@ -935,6 +938,20 @@ static int parse_options (char *options,
3055 set_opt (sbi->s_mount_opt, NO_UID32);
3057 +#ifndef CONFIG_TAGGING_NONE
3059 + set_opt (sbi->s_mount_opt, TAGGED);
3062 + clear_opt (sbi->s_mount_opt, TAGGED);
3065 +#ifdef CONFIG_PROPAGATE
3068 + set_opt (sbi->s_mount_opt, TAGGED);
3072 clear_opt (sbi->s_mount_opt, CHECK);
3074 @@ -1653,6 +1670,9 @@ static int ext3_fill_super (struct super
3078 + if (EXT3_SB(sb)->s_mount_opt & EXT3_MOUNT_TAGGED)
3079 + sb->s_flags |= MS_TAGGED;
3081 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3082 ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3084 @@ -2527,6 +2547,13 @@ static int ext3_remount (struct super_bl
3085 if (sbi->s_mount_opt & EXT3_MOUNT_ABORT)
3086 ext3_abort(sb, __func__, "Abort forced by user");
3088 + if ((sbi->s_mount_opt & EXT3_MOUNT_TAGGED) &&
3089 + !(sb->s_flags & MS_TAGGED)) {
3090 + printk("EXT3-fs: %s: tagging not permitted on remount.\n",
3095 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3096 ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3098 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/symlink.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/symlink.c
3099 --- linux-2.6.29.1/fs/ext3/symlink.c 2008-12-25 00:26:37.000000000 +0100
3100 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/symlink.c 2009-02-22 22:54:25.000000000 +0100
3101 @@ -40,6 +40,7 @@ const struct inode_operations ext3_symli
3102 .listxattr = ext3_listxattr,
3103 .removexattr = generic_removexattr,
3105 + .sync_flags = ext3_sync_flags,
3108 const struct inode_operations ext3_fast_symlink_inode_operations = {
3109 @@ -51,4 +52,5 @@ const struct inode_operations ext3_fast_
3110 .listxattr = ext3_listxattr,
3111 .removexattr = generic_removexattr,
3113 + .sync_flags = ext3_sync_flags,
3115 diff -NurpP --minimal linux-2.6.29.1/fs/ext3/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/xattr.c
3116 --- linux-2.6.29.1/fs/ext3/xattr.c 2008-12-25 00:26:37.000000000 +0100
3117 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext3/xattr.c 2009-02-22 22:54:25.000000000 +0100
3119 #include <linux/mbcache.h>
3120 #include <linux/quotaops.h>
3121 #include <linux/rwsem.h>
3122 +#include <linux/vs_dlimit.h>
3126 @@ -498,6 +499,7 @@ ext3_xattr_release_block(handle_t *handl
3127 error = ext3_journal_dirty_metadata(handle, bh);
3130 + DLIMIT_FREE_BLOCK(inode, 1);
3131 DQUOT_FREE_BLOCK(inode, 1);
3132 ea_bdebug(bh, "refcount now=%d; releasing",
3133 le32_to_cpu(BHDR(bh)->h_refcount));
3134 @@ -771,11 +773,14 @@ inserted:
3135 if (new_bh == bs->bh)
3136 ea_bdebug(new_bh, "keeping");
3139 + if (DLIMIT_ALLOC_BLOCK(inode, 1))
3141 /* The old block is released after updating
3144 if (DQUOT_ALLOC_BLOCK(inode, 1))
3146 + goto cleanup_dlimit;
3147 error = ext3_journal_get_write_access(handle,
3150 @@ -849,6 +854,8 @@ cleanup:
3153 DQUOT_FREE_BLOCK(inode, 1);
3155 + DLIMIT_FREE_BLOCK(inode, 1);
3159 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/balloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/balloc.c
3160 --- linux-2.6.29.1/fs/ext4/balloc.c 2009-03-24 14:22:25.000000000 +0100
3161 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/balloc.c 2009-03-24 14:48:35.000000000 +0100
3163 #include <linux/jbd2.h>
3164 #include <linux/quotaops.h>
3165 #include <linux/buffer_head.h>
3166 +#include <linux/vs_dlimit.h>
3167 +#include <linux/vs_tag.h>
3169 #include "ext4_jbd2.h"
3171 @@ -535,8 +537,10 @@ void ext4_free_blocks(handle_t *handle,
3173 ext4_mb_free_blocks(handle, inode, block, count,
3174 metadata, &dquot_freed_blocks);
3175 - if (dquot_freed_blocks)
3176 + if (dquot_freed_blocks) {
3177 + DLIMIT_FREE_BLOCK(inode, dquot_freed_blocks);
3178 DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
3183 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/ext4.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ext4.h
3184 --- linux-2.6.29.1/fs/ext4/ext4.h 2009-03-24 14:22:25.000000000 +0100
3185 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ext4.h 2009-03-24 15:44:56.000000000 +0100
3186 @@ -243,8 +243,12 @@ struct flex_groups {
3187 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
3188 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
3189 #define EXT4_EXT_MIGRATE 0x00100000 /* Inode is migrating */
3190 +#define EXT4_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
3191 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
3193 +#define EXT4_BARRIER_FL 0x04000000 /* Barrier for chroot() */
3194 +#define EXT4_COW_FL 0x20000000 /* Copy on Write marker */
3196 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
3197 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
3199 @@ -541,6 +545,7 @@ do { \
3200 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
3201 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
3202 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
3203 +#define EXT4_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
3205 /* Compatibility, for having both ext2_fs.h and ext4_fs.h included at once */
3206 #ifndef _LINUX_EXT2_FS_H
3207 @@ -1076,6 +1081,7 @@ struct buffer_head *ext4_bread(handle_t
3208 ext4_lblk_t, int, int *);
3209 int ext4_get_block(struct inode *inode, sector_t iblock,
3210 struct buffer_head *bh_result, int create);
3211 +extern int ext4_sync_flags(struct inode *inode);
3213 extern struct inode *ext4_iget(struct super_block *, unsigned long);
3214 extern int ext4_write_inode(struct inode *, int);
3215 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/file.c
3216 --- linux-2.6.29.1/fs/ext4/file.c 2009-03-24 14:22:25.000000000 +0100
3217 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/file.c 2009-03-24 14:48:35.000000000 +0100
3218 @@ -171,5 +171,6 @@ const struct inode_operations ext4_file_
3219 .permission = ext4_permission,
3220 .fallocate = ext4_fallocate,
3221 .fiemap = ext4_fiemap,
3222 + .sync_flags = ext4_sync_flags,
3225 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/ialloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ialloc.c
3226 --- linux-2.6.29.1/fs/ext4/ialloc.c 2009-03-24 14:22:25.000000000 +0100
3227 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ialloc.c 2009-03-24 14:48:35.000000000 +0100
3229 #include <linux/random.h>
3230 #include <linux/bitops.h>
3231 #include <linux/blkdev.h>
3232 +#include <linux/vs_dlimit.h>
3233 +#include <linux/vs_tag.h>
3234 #include <asm/byteorder.h>
3236 #include "ext4_jbd2.h"
3237 @@ -224,6 +226,7 @@ void ext4_free_inode(handle_t *handle, s
3238 ext4_xattr_delete_inode(handle, inode);
3239 DQUOT_FREE_INODE(inode);
3241 + DLIMIT_FREE_INODE(inode);
3243 is_directory = S_ISDIR(inode->i_mode);
3245 @@ -711,6 +714,12 @@ struct inode *ext4_new_inode(handle_t *h
3246 inode = new_inode(sb);
3248 return ERR_PTR(-ENOMEM);
3250 + inode->i_tag = dx_current_fstag(sb);
3251 + if (DLIMIT_ALLOC_INODE(inode)) {
3258 @@ -889,7 +898,8 @@ got:
3259 * newly created directory and file only if -o extent mount option is
3262 - ei->i_flags = EXT4_I(dir)->i_flags & ~(EXT4_INDEX_FL|EXT4_EXTENTS_FL);
3263 + ei->i_flags = EXT4_I(dir)->i_flags &
3264 + ~(EXT4_INDEX_FL|EXT4_EXTENTS_FL|EXT4_IXUNLINK_FL|EXT4_BARRIER_FL);
3266 ei->i_flags &= ~(EXT4_IMMUTABLE_FL|EXT4_APPEND_FL);
3267 /* dirsync only applies to directories */
3268 @@ -949,6 +959,8 @@ got:
3270 ext4_std_error(sb, err);
3272 + DLIMIT_FREE_INODE(inode);
3277 @@ -960,6 +972,7 @@ fail_free_drop:
3281 + DLIMIT_FREE_INODE(inode);
3282 inode->i_flags |= S_NOQUOTA;
3284 unlock_new_inode(inode);
3285 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/inode.c
3286 --- linux-2.6.29.1/fs/ext4/inode.c 2009-03-24 14:22:25.000000000 +0100
3287 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/inode.c 2009-03-24 14:48:35.000000000 +0100
3289 #include <linux/namei.h>
3290 #include <linux/uio.h>
3291 #include <linux/bio.h>
3292 +#include <linux/vs_tag.h>
3293 #include "ext4_jbd2.h"
3296 @@ -3781,7 +3782,7 @@ static void ext4_free_branches(handle_t
3298 int ext4_can_truncate(struct inode *inode)
3300 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3301 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3303 if (S_ISREG(inode->i_mode))
3305 @@ -4134,37 +4135,86 @@ void ext4_set_inode_flags(struct inode *
3307 unsigned int flags = EXT4_I(inode)->i_flags;
3309 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3310 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3311 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3313 + if (flags & EXT4_IMMUTABLE_FL)
3314 + inode->i_flags |= S_IMMUTABLE;
3315 + if (flags & EXT4_IXUNLINK_FL)
3316 + inode->i_flags |= S_IXUNLINK;
3318 if (flags & EXT4_SYNC_FL)
3319 inode->i_flags |= S_SYNC;
3320 if (flags & EXT4_APPEND_FL)
3321 inode->i_flags |= S_APPEND;
3322 - if (flags & EXT4_IMMUTABLE_FL)
3323 - inode->i_flags |= S_IMMUTABLE;
3324 if (flags & EXT4_NOATIME_FL)
3325 inode->i_flags |= S_NOATIME;
3326 if (flags & EXT4_DIRSYNC_FL)
3327 inode->i_flags |= S_DIRSYNC;
3329 + inode->i_vflags &= ~(V_BARRIER | V_COW);
3331 + if (flags & EXT4_BARRIER_FL)
3332 + inode->i_vflags |= V_BARRIER;
3333 + if (flags & EXT4_COW_FL)
3334 + inode->i_vflags |= V_COW;
3337 /* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
3338 void ext4_get_inode_flags(struct ext4_inode_info *ei)
3340 unsigned int flags = ei->vfs_inode.i_flags;
3341 + unsigned int vflags = ei->vfs_inode.i_vflags;
3343 + ei->i_flags &= ~(EXT4_SYNC_FL | EXT4_APPEND_FL |
3344 + EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL |
3345 + EXT4_NOATIME_FL | EXT4_DIRSYNC_FL |
3346 + EXT4_BARRIER_FL | EXT4_COW_FL);
3348 + if (flags & S_IMMUTABLE)
3349 + ei->i_flags |= EXT4_IMMUTABLE_FL;
3350 + if (flags & S_IXUNLINK)
3351 + ei->i_flags |= EXT4_IXUNLINK_FL;
3353 - ei->i_flags &= ~(EXT4_SYNC_FL|EXT4_APPEND_FL|
3354 - EXT4_IMMUTABLE_FL|EXT4_NOATIME_FL|EXT4_DIRSYNC_FL);
3356 ei->i_flags |= EXT4_SYNC_FL;
3357 if (flags & S_APPEND)
3358 ei->i_flags |= EXT4_APPEND_FL;
3359 - if (flags & S_IMMUTABLE)
3360 - ei->i_flags |= EXT4_IMMUTABLE_FL;
3361 if (flags & S_NOATIME)
3362 ei->i_flags |= EXT4_NOATIME_FL;
3363 if (flags & S_DIRSYNC)
3364 ei->i_flags |= EXT4_DIRSYNC_FL;
3366 + if (vflags & V_BARRIER)
3367 + ei->i_flags |= EXT4_BARRIER_FL;
3368 + if (vflags & V_COW)
3369 + ei->i_flags |= EXT4_COW_FL;
3372 +int ext4_sync_flags(struct inode *inode)
3374 + struct ext4_iloc iloc;
3378 + handle = ext4_journal_start(inode, 1);
3379 + if (IS_ERR(handle))
3380 + return PTR_ERR(handle);
3381 + if (IS_SYNC(inode))
3382 + handle->h_sync = 1;
3383 + err = ext4_reserve_inode_write(handle, inode, &iloc);
3387 + ext4_get_inode_flags(EXT4_I(inode));
3388 + inode->i_ctime = CURRENT_TIME;
3390 + err = ext4_mark_iloc_dirty(handle, inode, &iloc);
3392 + ext4_journal_stop(handle);
3396 static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
3397 struct ext4_inode_info *ei)
3399 @@ -4197,6 +4247,8 @@ struct inode *ext4_iget(struct super_blo
3400 struct inode *inode;
3406 inode = iget_locked(sb, ino);
3408 @@ -4216,12 +4268,16 @@ struct inode *ext4_iget(struct super_blo
3410 raw_inode = ext4_raw_inode(&iloc);
3411 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3412 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3413 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3414 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3415 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3416 if (!(test_opt(inode->i_sb, NO_UID32))) {
3417 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3418 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3419 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3420 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3422 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3423 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3424 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3425 + le16_to_cpu(raw_inode->i_raw_tag));
3426 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3429 @@ -4387,6 +4443,8 @@ static int ext4_do_update_inode(handle_t
3430 struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
3431 struct ext4_inode_info *ei = EXT4_I(inode);
3432 struct buffer_head *bh = iloc->bh;
3433 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3434 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3435 int err = 0, rc, block;
3437 /* For fields not not tracking in the in-memory inode,
3438 @@ -4397,29 +4455,32 @@ static int ext4_do_update_inode(handle_t
3439 ext4_get_inode_flags(ei);
3440 raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3441 if (!(test_opt(inode->i_sb, NO_UID32))) {
3442 - raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3443 - raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3444 + raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3445 + raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3447 * Fix up interoperability with old kernels. Otherwise, old inodes get
3448 * re-used with the upper 16 bits of the uid/gid intact
3451 raw_inode->i_uid_high =
3452 - cpu_to_le16(high_16_bits(inode->i_uid));
3453 + cpu_to_le16(high_16_bits(uid));
3454 raw_inode->i_gid_high =
3455 - cpu_to_le16(high_16_bits(inode->i_gid));
3456 + cpu_to_le16(high_16_bits(gid));
3458 raw_inode->i_uid_high = 0;
3459 raw_inode->i_gid_high = 0;
3462 raw_inode->i_uid_low =
3463 - cpu_to_le16(fs_high2lowuid(inode->i_uid));
3464 + cpu_to_le16(fs_high2lowuid(uid));
3465 raw_inode->i_gid_low =
3466 - cpu_to_le16(fs_high2lowgid(inode->i_gid));
3467 + cpu_to_le16(fs_high2lowgid(gid));
3468 raw_inode->i_uid_high = 0;
3469 raw_inode->i_gid_high = 0;
3471 +#ifdef CONFIG_TAGGING_INTERN
3472 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3474 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3476 EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
3477 @@ -4601,7 +4662,8 @@ int ext4_setattr(struct dentry *dentry,
3480 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3481 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3482 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3483 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3486 /* (user+group)*(old+new) structure, inode write (sb,
3487 @@ -4623,6 +4685,8 @@ int ext4_setattr(struct dentry *dentry,
3488 inode->i_uid = attr->ia_uid;
3489 if (attr->ia_valid & ATTR_GID)
3490 inode->i_gid = attr->ia_gid;
3491 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3492 + inode->i_tag = attr->ia_tag;
3493 error = ext4_mark_inode_dirty(handle, inode);
3494 ext4_journal_stop(handle);
3496 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ioctl.c
3497 --- linux-2.6.29.1/fs/ext4/ioctl.c 2009-03-24 14:22:25.000000000 +0100
3498 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/ioctl.c 2009-03-24 14:48:35.000000000 +0100
3502 #include <linux/fs.h>
3503 +#include <linux/mount.h>
3504 #include <linux/jbd2.h>
3505 #include <linux/capability.h>
3506 #include <linux/time.h>
3507 #include <linux/compat.h>
3508 #include <linux/smp_lock.h>
3509 #include <linux/mount.h>
3510 +#include <linux/vs_tag.h>
3511 #include <asm/uaccess.h>
3512 #include "ext4_jbd2.h"
3514 @@ -51,6 +53,11 @@ long ext4_ioctl(struct file *filp, unsig
3515 if (!S_ISDIR(inode->i_mode))
3516 flags &= ~EXT4_DIRSYNC_FL;
3518 + if (IS_BARRIER(inode)) {
3519 + vxwprintk_task(1, "messing with the barrier.");
3524 mutex_lock(&inode->i_mutex);
3525 /* Is it quota file? Do not allow user to mess with it */
3526 @@ -68,7 +75,9 @@ long ext4_ioctl(struct file *filp, unsig
3528 * This test looks nicer. Thanks to Pauline Middelink
3530 - if ((flags ^ oldflags) & (EXT4_APPEND_FL | EXT4_IMMUTABLE_FL)) {
3531 + if ((oldflags & EXT4_IMMUTABLE_FL) ||
3532 + ((flags ^ oldflags) & (EXT4_APPEND_FL |
3533 + EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL))) {
3534 if (!capable(CAP_LINUX_IMMUTABLE))
3537 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/namei.c
3538 --- linux-2.6.29.1/fs/ext4/namei.c 2009-03-24 14:22:25.000000000 +0100
3539 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/namei.c 2009-03-24 14:48:35.000000000 +0100
3541 #include <linux/quotaops.h>
3542 #include <linux/buffer_head.h>
3543 #include <linux/bio.h>
3544 +#include <linux/vs_tag.h>
3546 #include "ext4_jbd2.h"
3548 @@ -917,6 +918,7 @@ restart:
3550 ll_rw_block(READ_META, 1, &bh);
3552 + dx_propagate_tag(nd, inode);
3554 if ((bh = bh_use[ra_ptr++]) == NULL)
3556 @@ -2481,6 +2483,7 @@ const struct inode_operations ext4_dir_i
3557 .removexattr = generic_removexattr,
3559 .permission = ext4_permission,
3560 + .sync_flags = ext4_sync_flags,
3563 const struct inode_operations ext4_special_inode_operations = {
3564 @@ -2492,4 +2495,5 @@ const struct inode_operations ext4_speci
3565 .removexattr = generic_removexattr,
3567 .permission = ext4_permission,
3568 + .sync_flags = ext4_sync_flags,
3570 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/super.c
3571 --- linux-2.6.29.1/fs/ext4/super.c 2009-03-24 14:22:25.000000000 +0100
3572 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/super.c 2009-03-24 15:46:48.000000000 +0100
3573 @@ -1013,7 +1013,8 @@ enum {
3574 Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
3575 Opt_grpquota, Opt_i_version,
3576 Opt_stripe, Opt_delalloc, Opt_nodelalloc,
3577 - Opt_inode_readahead_blks, Opt_journal_ioprio
3578 + Opt_inode_readahead_blks, Opt_journal_ioprio,
3579 + Opt_tag, Opt_notag, Opt_tagid
3582 static const match_table_t tokens = {
3583 @@ -1073,6 +1074,9 @@ static const match_table_t tokens = {
3584 {Opt_nodelalloc, "nodelalloc"},
3585 {Opt_inode_readahead_blks, "inode_readahead_blks=%u"},
3586 {Opt_journal_ioprio, "journal_ioprio=%u"},
3588 + {Opt_notag, "notag"},
3589 + {Opt_tagid, "tagid=%u"},
3593 @@ -1168,6 +1172,20 @@ static int parse_options(char *options,
3595 set_opt(sbi->s_mount_opt, NO_UID32);
3597 +#ifndef CONFIG_TAGGING_NONE
3599 + set_opt (sbi->s_mount_opt, TAGGED);
3602 + clear_opt (sbi->s_mount_opt, TAGGED);
3605 +#ifdef CONFIG_PROPAGATE
3608 + set_opt (sbi->s_mount_opt, TAGGED);
3612 set_opt(sbi->s_mount_opt, DEBUG);
3614 @@ -2113,6 +2131,9 @@ static int ext4_fill_super(struct super_
3615 &journal_ioprio, NULL, 0))
3618 + if (EXT4_SB(sb)->s_mount_opt & EXT4_MOUNT_TAGGED)
3619 + sb->s_flags |= MS_TAGGED;
3621 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3622 ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3624 @@ -3160,6 +3181,13 @@ static int ext4_remount(struct super_blo
3625 if (sbi->s_mount_opt & EXT4_MOUNT_ABORT)
3626 ext4_abort(sb, __func__, "Abort forced by user");
3628 + if ((sbi->s_mount_opt & EXT4_MOUNT_TAGGED) &&
3629 + !(sb->s_flags & MS_TAGGED)) {
3630 + printk("EXT4-fs: %s: tagging not permitted on remount.\n",
3635 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3636 ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3638 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/symlink.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/symlink.c
3639 --- linux-2.6.29.1/fs/ext4/symlink.c 2008-12-25 00:26:37.000000000 +0100
3640 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/symlink.c 2009-02-22 22:54:25.000000000 +0100
3641 @@ -40,6 +40,7 @@ const struct inode_operations ext4_symli
3642 .listxattr = ext4_listxattr,
3643 .removexattr = generic_removexattr,
3645 + .sync_flags = ext4_sync_flags,
3648 const struct inode_operations ext4_fast_symlink_inode_operations = {
3649 @@ -51,4 +52,5 @@ const struct inode_operations ext4_fast_
3650 .listxattr = ext4_listxattr,
3651 .removexattr = generic_removexattr,
3653 + .sync_flags = ext4_sync_flags,
3655 diff -NurpP --minimal linux-2.6.29.1/fs/ext4/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/xattr.c
3656 --- linux-2.6.29.1/fs/ext4/xattr.c 2009-03-24 14:22:25.000000000 +0100
3657 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ext4/xattr.c 2009-03-24 15:47:44.000000000 +0100
3659 #include <linux/mbcache.h>
3660 #include <linux/quotaops.h>
3661 #include <linux/rwsem.h>
3662 +#include <linux/vs_dlimit.h>
3663 #include "ext4_jbd2.h"
3666 @@ -490,6 +491,7 @@ ext4_xattr_release_block(handle_t *handl
3667 error = ext4_handle_dirty_metadata(handle, inode, bh);
3669 ext4_handle_sync(handle);
3670 + DLIMIT_FREE_BLOCK(inode, 1);
3671 DQUOT_FREE_BLOCK(inode, 1);
3672 ea_bdebug(bh, "refcount now=%d; releasing",
3673 le32_to_cpu(BHDR(bh)->h_refcount));
3674 @@ -781,11 +783,14 @@ inserted:
3675 if (new_bh == bs->bh)
3676 ea_bdebug(new_bh, "keeping");
3679 + if (DLIMIT_ALLOC_BLOCK(inode, 1))
3681 /* The old block is released after updating
3684 if (DQUOT_ALLOC_BLOCK(inode, 1))
3686 + goto cleanup_dlimit;
3687 error = ext4_journal_get_write_access(handle,
3690 @@ -861,6 +866,8 @@ cleanup:
3693 DQUOT_FREE_BLOCK(inode, 1);
3695 + DLIMIT_FREE_BLOCK(inode, 1);
3699 diff -NurpP --minimal linux-2.6.29.1/fs/fcntl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/fcntl.c
3700 --- linux-2.6.29.1/fs/fcntl.c 2009-03-24 14:22:25.000000000 +0100
3701 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/fcntl.c 2009-03-24 14:48:35.000000000 +0100
3703 #include <linux/rcupdate.h>
3704 #include <linux/pid_namespace.h>
3705 #include <linux/smp_lock.h>
3706 +#include <linux/vs_limit.h>
3708 #include <asm/poll.h>
3709 #include <asm/siginfo.h>
3710 @@ -103,6 +104,8 @@ SYSCALL_DEFINE3(dup3, unsigned int, oldf
3713 filp_close(tofree, files);
3715 + vx_openfd_inc(newfd); /* fd was unused */
3719 @@ -343,6 +346,8 @@ SYSCALL_DEFINE3(fcntl, unsigned int, fd,
3723 + if (!vx_files_avail(1))
3726 err = security_file_fcntl(filp, cmd, arg);
3728 diff -NurpP --minimal linux-2.6.29.1/fs/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/file.c
3729 --- linux-2.6.29.1/fs/file.c 2008-12-25 00:26:37.000000000 +0100
3730 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/file.c 2009-02-22 22:54:25.000000000 +0100
3732 #include <linux/spinlock.h>
3733 #include <linux/rcupdate.h>
3734 #include <linux/workqueue.h>
3735 +#include <linux/vs_limit.h>
3737 struct fdtable_defer {
3739 @@ -367,6 +368,8 @@ struct files_struct *dup_fd(struct files
3740 struct file *f = *old_fds++;
3743 + /* TODO: sum it first for check and performance */
3744 + vx_openfd_inc(open_files - i);
3747 * The fd may be claimed in the fd bitmap but not yet
3748 @@ -475,6 +478,7 @@ repeat:
3750 FD_CLR(fd, fdt->close_on_exec);
3752 + vx_openfd_inc(fd);
3755 if (rcu_dereference(fdt->fd[fd]) != NULL) {
3756 diff -NurpP --minimal linux-2.6.29.1/fs/file_table.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/file_table.c
3757 --- linux-2.6.29.1/fs/file_table.c 2009-03-24 14:22:25.000000000 +0100
3758 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/file_table.c 2009-03-24 14:48:35.000000000 +0100
3760 #include <linux/fsnotify.h>
3761 #include <linux/sysctl.h>
3762 #include <linux/percpu_counter.h>
3763 +#include <linux/vs_limit.h>
3764 +#include <linux/vs_context.h>
3766 #include <asm/atomic.h>
3768 @@ -129,6 +131,8 @@ struct file *get_empty_filp(void)
3769 f->f_cred = get_cred(cred);
3770 eventpoll_init_file(f);
3771 /* f->f_version: 0 */
3772 + f->f_xid = vx_current_xid();
3777 @@ -283,6 +287,8 @@ void __fput(struct file *file)
3778 cdev_put(inode->i_cdev);
3779 fops_put(file->f_op);
3780 put_pid(file->f_owner.pid);
3781 + vx_files_dec(file);
3784 if (file->f_mode & FMODE_WRITE)
3785 drop_file_write_access(file);
3786 @@ -350,6 +356,8 @@ void put_filp(struct file *file)
3788 if (atomic_long_dec_and_test(&file->f_count)) {
3789 security_file_free(file);
3790 + vx_files_dec(file);
3795 diff -NurpP --minimal linux-2.6.29.1/fs/hfsplus/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/hfsplus/ioctl.c
3796 --- linux-2.6.29.1/fs/hfsplus/ioctl.c 2008-12-25 00:26:37.000000000 +0100
3797 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/hfsplus/ioctl.c 2009-02-22 22:54:25.000000000 +0100
3799 #include <linux/mount.h>
3800 #include <linux/sched.h>
3801 #include <linux/xattr.h>
3802 +#include <linux/mount.h>
3803 #include <asm/uaccess.h>
3804 #include "hfsplus_fs.h"
3806 diff -NurpP --minimal linux-2.6.29.1/fs/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/inode.c
3807 --- linux-2.6.29.1/fs/inode.c 2009-03-24 14:22:26.000000000 +0100
3808 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/inode.c 2009-03-25 00:51:55.000000000 +0100
3809 @@ -126,6 +126,9 @@ struct inode *inode_init_always(struct s
3810 struct address_space * const mapping = &inode->i_data;
3814 + /* essential because of inode slab reuse */
3816 inode->i_blkbits = sb->s_blocksize_bits;
3818 atomic_set(&inode->i_count, 1);
3819 @@ -146,6 +149,7 @@ struct inode *inode_init_always(struct s
3820 inode->i_bdev = NULL;
3821 inode->i_cdev = NULL;
3823 + inode->i_mdev = 0;
3824 inode->dirtied_when = 0;
3825 if (security_inode_alloc(inode)) {
3826 if (inode->i_sb->s_op->destroy_inode)
3827 @@ -267,6 +271,8 @@ void __iget(struct inode * inode)
3828 inodes_stat.nr_unused--;
3831 +EXPORT_SYMBOL_GPL(__iget);
3834 * clear_inode - clear an inode
3835 * @inode: inode to clear
3836 @@ -1539,9 +1545,11 @@ void init_special_inode(struct inode *in
3837 if (S_ISCHR(mode)) {
3838 inode->i_fop = &def_chr_fops;
3839 inode->i_rdev = rdev;
3840 + inode->i_mdev = rdev;
3841 } else if (S_ISBLK(mode)) {
3842 inode->i_fop = &def_blk_fops;
3843 inode->i_rdev = rdev;
3844 + inode->i_mdev = rdev;
3845 } else if (S_ISFIFO(mode))
3846 inode->i_fop = &def_fifo_fops;
3847 else if (S_ISSOCK(mode))
3848 diff -NurpP --minimal linux-2.6.29.1/fs/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ioctl.c
3849 --- linux-2.6.29.1/fs/ioctl.c 2009-03-24 14:22:26.000000000 +0100
3850 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ioctl.c 2009-03-24 14:48:35.000000000 +0100
3852 #include <linux/uaccess.h>
3853 #include <linux/writeback.h>
3854 #include <linux/buffer_head.h>
3855 +#include <linux/proc_fs.h>
3856 +#include <linux/vserver/inode.h>
3857 +#include <linux/vs_tag.h>
3859 #include <asm/ioctls.h>
3861 diff -NurpP --minimal linux-2.6.29.1/fs/ioprio.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ioprio.c
3862 --- linux-2.6.29.1/fs/ioprio.c 2009-03-24 14:22:26.000000000 +0100
3863 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ioprio.c 2009-03-24 14:48:35.000000000 +0100
3865 #include <linux/syscalls.h>
3866 #include <linux/security.h>
3867 #include <linux/pid_namespace.h>
3868 +#include <linux/vs_base.h>
3870 int set_task_ioprio(struct task_struct *task, int ioprio)
3872 @@ -123,6 +124,8 @@ SYSCALL_DEFINE3(ioprio_set, int, which,
3874 pgrp = find_vpid(who);
3875 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
3876 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
3878 ret = set_task_ioprio(p, ioprio);
3881 @@ -212,6 +215,8 @@ SYSCALL_DEFINE2(ioprio_get, int, which,
3883 pgrp = find_vpid(who);
3884 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
3885 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
3887 tmpio = get_task_ioprio(p);
3890 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/acl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/acl.c
3891 --- linux-2.6.29.1/fs/jfs/acl.c 2008-12-25 00:26:37.000000000 +0100
3892 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/acl.c 2009-02-22 22:54:25.000000000 +0100
3893 @@ -232,7 +232,8 @@ int jfs_setattr(struct dentry *dentry, s
3896 if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
3897 - (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
3898 + (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
3899 + (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
3900 if (DQUOT_TRANSFER(inode, iattr))
3903 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/file.c
3904 --- linux-2.6.29.1/fs/jfs/file.c 2008-12-25 00:26:37.000000000 +0100
3905 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/file.c 2009-02-22 22:54:25.000000000 +0100
3906 @@ -98,6 +98,7 @@ const struct inode_operations jfs_file_i
3907 .setattr = jfs_setattr,
3908 .permission = jfs_permission,
3910 + .sync_flags = jfs_sync_flags,
3913 const struct file_operations jfs_file_operations = {
3914 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/inode.c
3915 --- linux-2.6.29.1/fs/jfs/inode.c 2009-03-24 14:22:26.000000000 +0100
3916 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/inode.c 2009-03-24 14:48:35.000000000 +0100
3918 #include <linux/buffer_head.h>
3919 #include <linux/pagemap.h>
3920 #include <linux/quotaops.h>
3921 +#include <linux/vs_dlimit.h>
3922 #include "jfs_incore.h"
3923 #include "jfs_inode.h"
3924 #include "jfs_filsys.h"
3925 @@ -161,6 +162,7 @@ void jfs_delete_inode(struct inode *inod
3927 DQUOT_FREE_INODE(inode);
3929 + DLIMIT_FREE_INODE(inode);
3933 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/ioctl.c
3934 --- linux-2.6.29.1/fs/jfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
3935 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/ioctl.c 2009-02-22 22:54:25.000000000 +0100
3937 #include <linux/mount.h>
3938 #include <linux/time.h>
3939 #include <linux/sched.h>
3940 +#include <linux/mount.h>
3941 #include <asm/current.h>
3942 #include <asm/uaccess.h>
3944 @@ -85,6 +86,11 @@ long jfs_ioctl(struct file *filp, unsign
3945 if (!S_ISDIR(inode->i_mode))
3946 flags &= ~JFS_DIRSYNC_FL;
3948 + if (IS_BARRIER(inode)) {
3949 + vxwprintk_task(1, "messing with the barrier.");
3953 /* Is it quota file? Do not allow user to mess with it */
3954 if (IS_NOQUOTA(inode)) {
3956 @@ -102,8 +108,8 @@ long jfs_ioctl(struct file *filp, unsign
3957 * the relevant capability.
3959 if ((oldflags & JFS_IMMUTABLE_FL) ||
3960 - ((flags ^ oldflags) &
3961 - (JFS_APPEND_FL | JFS_IMMUTABLE_FL))) {
3962 + ((flags ^ oldflags) & (JFS_APPEND_FL |
3963 + JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL))) {
3964 if (!capable(CAP_LINUX_IMMUTABLE)) {
3965 mutex_unlock(&inode->i_mutex);
3967 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_dinode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_dinode.h
3968 --- linux-2.6.29.1/fs/jfs/jfs_dinode.h 2008-12-25 00:26:37.000000000 +0100
3969 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_dinode.h 2009-02-22 22:54:25.000000000 +0100
3970 @@ -161,9 +161,13 @@ struct dinode {
3972 #define JFS_APPEND_FL 0x01000000 /* writes to file may only append */
3973 #define JFS_IMMUTABLE_FL 0x02000000 /* Immutable file */
3974 +#define JFS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
3976 -#define JFS_FL_USER_VISIBLE 0x03F80000
3977 -#define JFS_FL_USER_MODIFIABLE 0x03F80000
3978 +#define JFS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
3979 +#define JFS_COW_FL 0x20000000 /* Copy on Write marker */
3981 +#define JFS_FL_USER_VISIBLE 0x07F80000
3982 +#define JFS_FL_USER_MODIFIABLE 0x07F80000
3983 #define JFS_FL_INHERIT 0x03C80000
3985 /* These are identical to EXT[23]_IOC_GETFLAGS/SETFLAGS */
3986 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_dtree.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_dtree.c
3987 --- linux-2.6.29.1/fs/jfs/jfs_dtree.c 2008-12-25 00:26:37.000000000 +0100
3988 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_dtree.c 2009-02-22 22:54:25.000000000 +0100
3991 #include <linux/fs.h>
3992 #include <linux/quotaops.h>
3993 +#include <linux/vs_dlimit.h>
3994 #include "jfs_incore.h"
3995 #include "jfs_superblock.h"
3996 #include "jfs_filsys.h"
3997 @@ -383,10 +384,10 @@ static u32 add_index(tid_t tid, struct i
3999 if (DQUOT_ALLOC_BLOCK(ip, sbi->nbperpage))
4001 - if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr)) {
4002 - DQUOT_FREE_BLOCK(ip, sbi->nbperpage);
4005 + if (DLIMIT_ALLOC_BLOCK(ip, sbi->nbperpage))
4006 + goto clean_up_dquot;
4007 + if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr))
4008 + goto clean_up_dlimit;
4011 * Save the table, we're going to overwrite it with the
4012 @@ -480,6 +481,12 @@ static u32 add_index(tid_t tid, struct i
4017 + DLIMIT_FREE_BLOCK(ip, sbi->nbperpage);
4020 + DQUOT_FREE_BLOCK(ip, sbi->nbperpage);
4024 jfs_ip->next_index--;
4025 @@ -951,6 +958,7 @@ static int dtSplitUp(tid_t tid,
4028 int quota_allocation = 0;
4029 + int dlimit_allocation = 0;
4031 /* get split page */
4033 @@ -1033,6 +1041,12 @@ static int dtSplitUp(tid_t tid,
4035 quota_allocation += n;
4037 + if (DLIMIT_ALLOC_BLOCK(ip, n)) {
4041 + dlimit_allocation += n;
4043 if ((rc = dbReAlloc(sbi->ipbmap, xaddr, (s64) xlen,
4046 @@ -1306,6 +1320,9 @@ static int dtSplitUp(tid_t tid,
4050 + /* Rollback dlimit allocation */
4051 + if (rc && dlimit_allocation)
4052 + DLIMIT_FREE_BLOCK(ip, dlimit_allocation);
4053 /* Rollback quota allocation */
4054 if (rc && quota_allocation)
4055 DQUOT_FREE_BLOCK(ip, quota_allocation);
4056 @@ -1373,6 +1390,12 @@ static int dtSplitPage(tid_t tid, struct
4057 release_metapage(rmp);
4060 + /* Allocate blocks to dlimit. */
4061 + if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4062 + DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4063 + release_metapage(rmp);
4067 jfs_info("dtSplitPage: ip:0x%p smp:0x%p rmp:0x%p", ip, smp, rmp);
4069 @@ -1920,6 +1943,12 @@ static int dtSplitRoot(tid_t tid,
4070 release_metapage(rmp);
4073 + /* Allocate blocks to dlimit. */
4074 + if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4075 + DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4076 + release_metapage(rmp);
4080 BT_MARK_DIRTY(rmp, ip);
4082 @@ -2286,6 +2315,8 @@ static int dtDeleteUp(tid_t tid, struct
4084 xlen = lengthPXD(&fp->header.self);
4086 + /* Free dlimit allocation. */
4087 + DLIMIT_FREE_BLOCK(ip, xlen);
4088 /* Free quota allocation. */
4089 DQUOT_FREE_BLOCK(ip, xlen);
4091 @@ -2362,6 +2393,8 @@ static int dtDeleteUp(tid_t tid, struct
4093 xlen = lengthPXD(&p->header.self);
4095 + /* Free dlimit allocation */
4096 + DLIMIT_FREE_BLOCK(ip, xlen);
4097 /* Free quota allocation */
4098 DQUOT_FREE_BLOCK(ip, xlen);
4100 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_extent.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_extent.c
4101 --- linux-2.6.29.1/fs/jfs/jfs_extent.c 2008-12-25 00:26:37.000000000 +0100
4102 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_extent.c 2009-02-22 22:54:25.000000000 +0100
4105 #include <linux/fs.h>
4106 #include <linux/quotaops.h>
4107 +#include <linux/vs_dlimit.h>
4108 #include "jfs_incore.h"
4109 #include "jfs_inode.h"
4110 #include "jfs_superblock.h"
4111 @@ -147,6 +148,14 @@ extAlloc(struct inode *ip, s64 xlen, s64
4115 + /* Allocate blocks to dlimit. */
4116 + if (DLIMIT_ALLOC_BLOCK(ip, nxlen)) {
4117 + DQUOT_FREE_BLOCK(ip, nxlen);
4118 + dbFree(ip, nxaddr, (s64) nxlen);
4119 + mutex_unlock(&JFS_IP(ip)->commit_mutex);
4123 /* determine the value of the extent flag */
4124 xflag = abnr ? XAD_NOTRECORDED : 0;
4126 @@ -164,6 +173,7 @@ extAlloc(struct inode *ip, s64 xlen, s64
4129 dbFree(ip, nxaddr, nxlen);
4130 + DLIMIT_FREE_BLOCK(ip, nxlen);
4131 DQUOT_FREE_BLOCK(ip, nxlen);
4132 mutex_unlock(&JFS_IP(ip)->commit_mutex);
4134 @@ -261,6 +271,13 @@ int extRealloc(struct inode *ip, s64 nxl
4135 mutex_unlock(&JFS_IP(ip)->commit_mutex);
4138 + /* Allocate blocks to dlimit. */
4139 + if (DLIMIT_ALLOC_BLOCK(ip, nxlen)) {
4140 + DQUOT_FREE_BLOCK(ip, nxlen);
4141 + dbFree(ip, nxaddr, (s64) nxlen);
4142 + up(&JFS_IP(ip)->commit_sem);
4146 delta = nxlen - xlen;
4148 @@ -297,6 +314,7 @@ int extRealloc(struct inode *ip, s64 nxl
4149 /* extend the extent */
4150 if ((rc = xtExtend(0, ip, xoff + xlen, (int) nextend, 0))) {
4151 dbFree(ip, xaddr + xlen, delta);
4152 + DLIMIT_FREE_BLOCK(ip, nxlen);
4153 DQUOT_FREE_BLOCK(ip, nxlen);
4156 @@ -308,6 +326,7 @@ int extRealloc(struct inode *ip, s64 nxl
4158 if ((rc = xtTailgate(0, ip, xoff, (int) ntail, nxaddr, 0))) {
4159 dbFree(ip, nxaddr, nxlen);
4160 + DLIMIT_FREE_BLOCK(ip, nxlen);
4161 DQUOT_FREE_BLOCK(ip, nxlen);
4164 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_filsys.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_filsys.h
4165 --- linux-2.6.29.1/fs/jfs/jfs_filsys.h 2008-12-25 00:26:37.000000000 +0100
4166 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_filsys.h 2009-02-22 22:54:25.000000000 +0100
4168 #define JFS_NAME_MAX 255
4169 #define JFS_PATH_MAX BPSIZE
4171 +#define JFS_TAGGED 0x00800000 /* Context Tagging */
4174 * file system state (superblock state)
4175 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_imap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_imap.c
4176 --- linux-2.6.29.1/fs/jfs/jfs_imap.c 2009-03-24 14:22:26.000000000 +0100
4177 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_imap.c 2009-03-24 14:48:35.000000000 +0100
4179 #include <linux/buffer_head.h>
4180 #include <linux/pagemap.h>
4181 #include <linux/quotaops.h>
4182 +#include <linux/vs_tag.h>
4184 #include "jfs_incore.h"
4185 #include "jfs_inode.h"
4186 @@ -3062,6 +3063,8 @@ static int copy_from_dinode(struct dinod
4188 struct jfs_inode_info *jfs_ip = JFS_IP(ip);
4189 struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
4193 jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
4194 jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
4195 @@ -3082,14 +3085,18 @@ static int copy_from_dinode(struct dinod
4197 ip->i_nlink = le32_to_cpu(dip->di_nlink);
4199 - jfs_ip->saved_uid = le32_to_cpu(dip->di_uid);
4200 + uid = le32_to_cpu(dip->di_uid);
4201 + gid = le32_to_cpu(dip->di_gid);
4202 + ip->i_tag = INOTAG_TAG(DX_TAG(ip), uid, gid, 0);
4204 + jfs_ip->saved_uid = INOTAG_UID(DX_TAG(ip), uid, gid);
4206 ip->i_uid = jfs_ip->saved_uid;
4208 ip->i_uid = sbi->uid;
4211 - jfs_ip->saved_gid = le32_to_cpu(dip->di_gid);
4212 + jfs_ip->saved_gid = INOTAG_GID(DX_TAG(ip), uid, gid);
4214 ip->i_gid = jfs_ip->saved_gid;
4216 @@ -3154,14 +3161,12 @@ static void copy_to_dinode(struct dinode
4217 dip->di_size = cpu_to_le64(ip->i_size);
4218 dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
4219 dip->di_nlink = cpu_to_le32(ip->i_nlink);
4220 - if (sbi->uid == -1)
4221 - dip->di_uid = cpu_to_le32(ip->i_uid);
4223 - dip->di_uid = cpu_to_le32(jfs_ip->saved_uid);
4224 - if (sbi->gid == -1)
4225 - dip->di_gid = cpu_to_le32(ip->i_gid);
4227 - dip->di_gid = cpu_to_le32(jfs_ip->saved_gid);
4229 + dip->di_uid = cpu_to_le32(TAGINO_UID(DX_TAG(ip),
4230 + (sbi->uid == -1) ? ip->i_uid : jfs_ip->saved_uid, ip->i_tag));
4231 + dip->di_gid = cpu_to_le32(TAGINO_GID(DX_TAG(ip),
4232 + (sbi->gid == -1) ? ip->i_gid : jfs_ip->saved_gid, ip->i_tag));
4234 jfs_get_inode_flags(jfs_ip);
4236 * mode2 is only needed for storing the higher order bits.
4237 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_inode.c
4238 --- linux-2.6.29.1/fs/jfs/jfs_inode.c 2009-03-24 14:22:26.000000000 +0100
4239 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_inode.c 2009-03-24 15:58:01.000000000 +0100
4242 #include <linux/fs.h>
4243 #include <linux/quotaops.h>
4244 +#include <linux/vs_dlimit.h>
4245 +#include <linux/vs_tag.h>
4246 #include "jfs_incore.h"
4247 #include "jfs_inode.h"
4248 #include "jfs_filsys.h"
4249 @@ -30,29 +32,46 @@ void jfs_set_inode_flags(struct inode *i
4251 unsigned int flags = JFS_IP(inode)->mode2;
4253 - inode->i_flags &= ~(S_IMMUTABLE | S_APPEND |
4254 - S_NOATIME | S_DIRSYNC | S_SYNC);
4255 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
4256 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
4258 if (flags & JFS_IMMUTABLE_FL)
4259 inode->i_flags |= S_IMMUTABLE;
4260 + if (flags & JFS_IXUNLINK_FL)
4261 + inode->i_flags |= S_IXUNLINK;
4263 + if (flags & JFS_SYNC_FL)
4264 + inode->i_flags |= S_SYNC;
4265 if (flags & JFS_APPEND_FL)
4266 inode->i_flags |= S_APPEND;
4267 if (flags & JFS_NOATIME_FL)
4268 inode->i_flags |= S_NOATIME;
4269 if (flags & JFS_DIRSYNC_FL)
4270 inode->i_flags |= S_DIRSYNC;
4271 - if (flags & JFS_SYNC_FL)
4272 - inode->i_flags |= S_SYNC;
4274 + inode->i_vflags &= ~(V_BARRIER | V_COW);
4276 + if (flags & JFS_BARRIER_FL)
4277 + inode->i_vflags |= V_BARRIER;
4278 + if (flags & JFS_COW_FL)
4279 + inode->i_vflags |= V_COW;
4282 void jfs_get_inode_flags(struct jfs_inode_info *jfs_ip)
4284 unsigned int flags = jfs_ip->vfs_inode.i_flags;
4285 + unsigned int vflags = jfs_ip->vfs_inode.i_vflags;
4287 + jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL |
4288 + JFS_APPEND_FL | JFS_NOATIME_FL |
4289 + JFS_DIRSYNC_FL | JFS_SYNC_FL |
4290 + JFS_BARRIER_FL | JFS_COW_FL);
4292 - jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_APPEND_FL | JFS_NOATIME_FL |
4293 - JFS_DIRSYNC_FL | JFS_SYNC_FL);
4294 if (flags & S_IMMUTABLE)
4295 jfs_ip->mode2 |= JFS_IMMUTABLE_FL;
4296 + if (flags & S_IXUNLINK)
4297 + jfs_ip->mode2 |= JFS_IXUNLINK_FL;
4299 if (flags & S_APPEND)
4300 jfs_ip->mode2 |= JFS_APPEND_FL;
4301 if (flags & S_NOATIME)
4302 @@ -61,6 +80,19 @@ void jfs_get_inode_flags(struct jfs_inod
4303 jfs_ip->mode2 |= JFS_DIRSYNC_FL;
4305 jfs_ip->mode2 |= JFS_SYNC_FL;
4307 + if (vflags & V_BARRIER)
4308 + jfs_ip->mode2 |= JFS_BARRIER_FL;
4309 + if (vflags & V_COW)
4310 + jfs_ip->mode2 |= JFS_COW_FL;
4313 +int jfs_sync_flags(struct inode *inode)
4315 + jfs_get_inode_flags(JFS_IP(inode));
4316 + inode->i_ctime = CURRENT_TIME;
4317 + mark_inode_dirty(inode);
4322 @@ -113,6 +145,12 @@ struct inode *ialloc(struct inode *paren
4323 jfs_inode->saved_uid = inode->i_uid;
4324 jfs_inode->saved_gid = inode->i_gid;
4326 + inode->i_tag = dx_current_fstag(sb);
4327 + if (DLIMIT_ALLOC_INODE(inode)) {
4333 * Allocate inode to quota.
4335 @@ -162,6 +200,8 @@ struct inode *ialloc(struct inode *paren
4339 + DLIMIT_FREE_INODE(inode);
4342 inode->i_flags |= S_NOQUOTA;
4344 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_inode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_inode.h
4345 --- linux-2.6.29.1/fs/jfs/jfs_inode.h 2008-12-25 00:26:37.000000000 +0100
4346 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_inode.h 2009-02-22 22:54:25.000000000 +0100
4347 @@ -39,6 +39,7 @@ extern struct dentry *jfs_fh_to_dentry(s
4348 extern struct dentry *jfs_fh_to_parent(struct super_block *sb, struct fid *fid,
4349 int fh_len, int fh_type);
4350 extern void jfs_set_inode_flags(struct inode *);
4351 +extern int jfs_sync_flags(struct inode *);
4352 extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
4354 extern const struct address_space_operations jfs_aops;
4355 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/jfs_xtree.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_xtree.c
4356 --- linux-2.6.29.1/fs/jfs/jfs_xtree.c 2008-12-25 00:26:37.000000000 +0100
4357 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/jfs_xtree.c 2009-02-22 22:54:25.000000000 +0100
4359 #include <linux/module.h>
4360 #include <linux/quotaops.h>
4361 #include <linux/seq_file.h>
4362 +#include <linux/vs_dlimit.h>
4363 #include "jfs_incore.h"
4364 #include "jfs_filsys.h"
4365 #include "jfs_metapage.h"
4366 @@ -848,7 +849,12 @@ int xtInsert(tid_t tid, /* transaction
4368 if ((rc = DQUOT_ALLOC_BLOCK(ip, xlen)))
4370 + if ((rc = DLIMIT_ALLOC_BLOCK(ip, xlen))) {
4371 + DQUOT_FREE_BLOCK(ip, xlen);
4374 if ((rc = dbAlloc(ip, hint, (s64) xlen, &xaddr))) {
4375 + DLIMIT_FREE_BLOCK(ip, xlen);
4376 DQUOT_FREE_BLOCK(ip, xlen);
4379 @@ -878,6 +884,7 @@ int xtInsert(tid_t tid, /* transaction
4380 /* undo data extent allocation */
4382 dbFree(ip, xaddr, (s64) xlen);
4383 + DLIMIT_FREE_BLOCK(ip, xlen);
4384 DQUOT_FREE_BLOCK(ip, xlen);
4387 @@ -1234,6 +1241,7 @@ xtSplitPage(tid_t tid, struct inode *ip,
4389 struct xtlock *sxtlck = NULL, *rxtlck = NULL;
4390 int quota_allocation = 0;
4391 + int dlimit_allocation = 0;
4394 sp = XT_PAGE(ip, smp);
4395 @@ -1253,6 +1261,13 @@ xtSplitPage(tid_t tid, struct inode *ip,
4397 quota_allocation += lengthPXD(pxd);
4399 + /* Allocate blocks to dlimit. */
4400 + if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4404 + dlimit_allocation += lengthPXD(pxd);
4407 * allocate the new right page for the split
4409 @@ -1454,6 +1469,9 @@ xtSplitPage(tid_t tid, struct inode *ip,
4413 + /* Rollback dlimit allocation. */
4414 + if (dlimit_allocation)
4415 + DLIMIT_FREE_BLOCK(ip, dlimit_allocation);
4416 /* Rollback quota allocation. */
4417 if (quota_allocation)
4418 DQUOT_FREE_BLOCK(ip, quota_allocation);
4419 @@ -1517,6 +1535,12 @@ xtSplitRoot(tid_t tid,
4420 release_metapage(rmp);
4423 + /* Allocate blocks to dlimit. */
4424 + if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4425 + DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4426 + release_metapage(rmp);
4430 jfs_info("xtSplitRoot: ip:0x%p rmp:0x%p", ip, rmp);
4432 @@ -3940,6 +3964,8 @@ s64 xtTruncate(tid_t tid, struct inode *
4434 ip->i_size = newsize;
4436 + /* update dlimit allocation to reflect freed blocks */
4437 + DLIMIT_FREE_BLOCK(ip, nfreed);
4438 /* update quota allocation to reflect freed blocks */
4439 DQUOT_FREE_BLOCK(ip, nfreed);
4441 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/namei.c
4442 --- linux-2.6.29.1/fs/jfs/namei.c 2009-03-24 14:22:26.000000000 +0100
4443 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/namei.c 2009-03-24 14:48:35.000000000 +0100
4445 #include <linux/ctype.h>
4446 #include <linux/quotaops.h>
4447 #include <linux/exportfs.h>
4448 +#include <linux/vs_tag.h>
4449 #include "jfs_incore.h"
4450 #include "jfs_superblock.h"
4451 #include "jfs_inode.h"
4452 @@ -1476,6 +1477,7 @@ static struct dentry *jfs_lookup(struct
4453 return ERR_CAST(ip);
4456 + dx_propagate_tag(nd, ip);
4457 dentry = d_splice_alias(ip, dentry);
4459 if (dentry && (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2))
4460 @@ -1545,6 +1547,7 @@ const struct inode_operations jfs_dir_in
4461 .setattr = jfs_setattr,
4462 .permission = jfs_permission,
4464 + .sync_flags = jfs_sync_flags,
4467 const struct file_operations jfs_dir_operations = {
4468 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/super.c
4469 --- linux-2.6.29.1/fs/jfs/super.c 2009-03-24 14:22:26.000000000 +0100
4470 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/super.c 2009-03-24 14:48:35.000000000 +0100
4471 @@ -196,7 +196,8 @@ static void jfs_put_super(struct super_b
4473 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
4474 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
4475 - Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
4476 + Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask,
4477 + Opt_tag, Opt_notag, Opt_tagid
4480 static const match_table_t tokens = {
4481 @@ -206,6 +207,10 @@ static const match_table_t tokens = {
4482 {Opt_resize, "resize=%u"},
4483 {Opt_resize_nosize, "resize"},
4484 {Opt_errors, "errors=%s"},
4486 + {Opt_notag, "notag"},
4487 + {Opt_tagid, "tagid=%u"},
4488 + {Opt_tag, "tagxid"},
4489 {Opt_ignore, "noquota"},
4490 {Opt_ignore, "quota"},
4491 {Opt_usrquota, "usrquota"},
4492 @@ -340,6 +345,20 @@ static int parse_options(char *options,
4496 +#ifndef CONFIG_TAGGING_NONE
4498 + *flag |= JFS_TAGGED;
4501 + *flag &= JFS_TAGGED;
4504 +#ifdef CONFIG_PROPAGATE
4507 + *flag |= JFS_TAGGED;
4511 printk("jfs: Unrecognized mount option \"%s\" "
4512 " or missing value\n", p);
4513 @@ -370,6 +389,13 @@ static int jfs_remount(struct super_bloc
4514 if (!parse_options(data, sb, &newLVSize, &flag)) {
4518 + if ((flag & JFS_TAGGED) && !(sb->s_flags & MS_TAGGED)) {
4519 + printk(KERN_ERR "JFS: %s: tagging not permitted on remount.\n",
4525 if (sb->s_flags & MS_RDONLY) {
4527 @@ -441,6 +467,9 @@ static int jfs_fill_super(struct super_b
4528 #ifdef CONFIG_JFS_POSIX_ACL
4529 sb->s_flags |= MS_POSIXACL;
4531 + /* map mount option tagxid */
4532 + if (sbi->flag & JFS_TAGGED)
4533 + sb->s_flags |= MS_TAGGED;
4536 printk(KERN_ERR "resize option for remount only\n");
4537 diff -NurpP --minimal linux-2.6.29.1/fs/jfs/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/xattr.c
4538 --- linux-2.6.29.1/fs/jfs/xattr.c 2008-12-25 00:26:37.000000000 +0100
4539 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/jfs/xattr.c 2009-02-22 22:54:25.000000000 +0100
4541 #include <linux/posix_acl_xattr.h>
4542 #include <linux/quotaops.h>
4543 #include <linux/security.h>
4544 +#include <linux/vs_dlimit.h>
4545 #include "jfs_incore.h"
4546 #include "jfs_superblock.h"
4547 #include "jfs_dmap.h"
4548 @@ -263,9 +264,16 @@ static int ea_write(struct inode *ip, st
4549 if (DQUOT_ALLOC_BLOCK(ip, nblocks)) {
4552 + /* Allocate new blocks to dlimit. */
4553 + if (DLIMIT_ALLOC_BLOCK(ip, nblocks)) {
4554 + DQUOT_FREE_BLOCK(ip, nblocks);
4558 rc = dbAlloc(ip, INOHINT(ip), nblocks, &blkno);
4560 + /*Rollback dlimit allocation. */
4561 + DLIMIT_FREE_BLOCK(ip, nblocks);
4562 /*Rollback quota allocation. */
4563 DQUOT_FREE_BLOCK(ip, nblocks);
4565 @@ -332,6 +340,8 @@ static int ea_write(struct inode *ip, st
4568 /* Rollback quota allocation. */
4569 + DLIMIT_FREE_BLOCK(ip, nblocks);
4570 + /* Rollback quota allocation. */
4571 DQUOT_FREE_BLOCK(ip, nblocks);
4573 dbFree(ip, blkno, nblocks);
4574 @@ -468,6 +478,7 @@ static int ea_get(struct inode *inode, s
4577 int quota_allocation = 0;
4578 + int dlimit_allocation = 0;
4580 /* When fsck.jfs clears a bad ea, it doesn't clear the size */
4581 if (ji->ea.flag == 0)
4582 @@ -543,6 +554,12 @@ static int ea_get(struct inode *inode, s
4584 quota_allocation = blocks_needed;
4586 + /* Allocate new blocks to dlimit. */
4588 + if (DLIMIT_ALLOC_BLOCK(inode, blocks_needed))
4590 + dlimit_allocation = blocks_needed;
4592 rc = dbAlloc(inode, INOHINT(inode), (s64) blocks_needed,
4595 @@ -600,6 +617,9 @@ static int ea_get(struct inode *inode, s
4599 + /* Rollback dlimit allocation */
4600 + if (dlimit_allocation)
4601 + DLIMIT_FREE_BLOCK(inode, dlimit_allocation);
4602 /* Rollback quota allocation */
4603 if (quota_allocation)
4604 DQUOT_FREE_BLOCK(inode, quota_allocation);
4605 @@ -676,8 +696,10 @@ static int ea_put(tid_t tid, struct inod
4608 /* If old blocks exist, they must be removed from quota allocation. */
4611 + DLIMIT_FREE_BLOCK(inode, old_blocks);
4612 DQUOT_FREE_BLOCK(inode, old_blocks);
4615 inode->i_ctime = CURRENT_TIME;
4617 diff -NurpP --minimal linux-2.6.29.1/fs/libfs.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/libfs.c
4618 --- linux-2.6.29.1/fs/libfs.c 2009-03-24 14:22:26.000000000 +0100
4619 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/libfs.c 2009-03-24 14:48:35.000000000 +0100
4620 @@ -125,7 +125,8 @@ static inline unsigned char dt_type(stru
4621 * both impossible due to the lock on directory.
4624 -int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
4625 +static inline int do_dcache_readdir_filter(struct file *filp,
4626 + void *dirent, filldir_t filldir, int (*filter)(struct dentry *dentry))
4628 struct dentry *dentry = filp->f_path.dentry;
4629 struct dentry *cursor = filp->private_data;
4630 @@ -158,6 +159,8 @@ int dcache_readdir(struct file * filp, v
4631 next = list_entry(p, struct dentry, d_u.d_child);
4632 if (d_unhashed(next) || !next->d_inode)
4634 + if (filter && !filter(next))
4637 spin_unlock(&dcache_lock);
4638 if (filldir(dirent, next->d_name.name,
4639 @@ -176,6 +179,18 @@ int dcache_readdir(struct file * filp, v
4643 +int dcache_readdir(struct file *filp, void *dirent, filldir_t filldir)
4645 + return do_dcache_readdir_filter(filp, dirent, filldir, NULL);
4648 +int dcache_readdir_filter(struct file *filp, void *dirent, filldir_t filldir,
4649 + int (*filter)(struct dentry *))
4651 + return do_dcache_readdir_filter(filp, dirent, filldir, filter);
4655 ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
4658 @@ -796,6 +811,7 @@ EXPORT_SYMBOL(dcache_dir_close);
4659 EXPORT_SYMBOL(dcache_dir_lseek);
4660 EXPORT_SYMBOL(dcache_dir_open);
4661 EXPORT_SYMBOL(dcache_readdir);
4662 +EXPORT_SYMBOL(dcache_readdir_filter);
4663 EXPORT_SYMBOL(generic_read_dir);
4664 EXPORT_SYMBOL(get_sb_pseudo);
4665 EXPORT_SYMBOL(simple_write_begin);
4666 diff -NurpP --minimal linux-2.6.29.1/fs/locks.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/locks.c
4667 --- linux-2.6.29.1/fs/locks.c 2009-03-24 14:22:26.000000000 +0100
4668 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/locks.c 2009-03-24 14:48:35.000000000 +0100
4670 #include <linux/time.h>
4671 #include <linux/rcupdate.h>
4672 #include <linux/pid_namespace.h>
4673 +#include <linux/vs_base.h>
4674 +#include <linux/vs_limit.h>
4676 #include <asm/uaccess.h>
4678 @@ -148,6 +150,8 @@ static struct kmem_cache *filelock_cache
4679 /* Allocate an empty lock structure. */
4680 static struct file_lock *locks_alloc_lock(void)
4682 + if (!vx_locks_avail(1))
4684 return kmem_cache_alloc(filelock_cache, GFP_KERNEL);
4687 @@ -173,6 +177,7 @@ static void locks_free_lock(struct file_
4688 BUG_ON(!list_empty(&fl->fl_block));
4689 BUG_ON(!list_empty(&fl->fl_link));
4692 locks_release_private(fl);
4693 kmem_cache_free(filelock_cache, fl);
4695 @@ -193,6 +198,7 @@ void locks_init_lock(struct file_lock *f
4696 fl->fl_start = fl->fl_end = 0;
4698 fl->fl_lmops = NULL;
4702 EXPORT_SYMBOL(locks_init_lock);
4703 @@ -247,6 +253,7 @@ void locks_copy_lock(struct file_lock *n
4704 new->fl_file = fl->fl_file;
4705 new->fl_ops = fl->fl_ops;
4706 new->fl_lmops = fl->fl_lmops;
4707 + new->fl_xid = fl->fl_xid;
4709 locks_copy_private(new, fl);
4711 @@ -285,6 +292,11 @@ static int flock_make_lock(struct file *
4712 fl->fl_flags = FL_FLOCK;
4714 fl->fl_end = OFFSET_MAX;
4716 + vxd_assert(filp->f_xid == vx_current_xid(),
4717 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4718 + fl->fl_xid = filp->f_xid;
4723 @@ -450,6 +462,7 @@ static int lease_init(struct file *filp,
4725 fl->fl_owner = current->files;
4726 fl->fl_pid = current->tgid;
4727 + fl->fl_xid = vx_current_xid();
4730 fl->fl_flags = FL_LEASE;
4731 @@ -469,6 +482,11 @@ static struct file_lock *lease_alloc(str
4733 return ERR_PTR(error);
4735 + fl->fl_xid = vx_current_xid();
4737 + vxd_assert(filp->f_xid == fl->fl_xid,
4738 + "f_xid(%d) == fl_xid(%d)", filp->f_xid, fl->fl_xid);
4740 error = lease_init(filp, type, fl);
4742 locks_free_lock(fl);
4743 @@ -769,6 +787,7 @@ static int flock_lock_file(struct file *
4747 + new_fl->fl_xid = -1;
4749 for_each_lock(inode, before) {
4750 struct file_lock *fl = *before;
4751 @@ -789,6 +808,7 @@ find_conflict:
4753 locks_copy_lock(new_fl, request);
4754 locks_insert_lock(before, new_fl);
4755 + vx_locks_inc(new_fl);
4759 @@ -799,7 +819,8 @@ out:
4763 -static int __posix_lock_file(struct inode *inode, struct file_lock *request, struct file_lock *conflock)
4764 +static int __posix_lock_file(struct inode *inode, struct file_lock *request,
4765 + struct file_lock *conflock, xid_t xid)
4767 struct file_lock *fl;
4768 struct file_lock *new_fl = NULL;
4769 @@ -809,6 +830,8 @@ static int __posix_lock_file(struct inod
4770 struct file_lock **before;
4771 int error, added = 0;
4773 + vxd_assert(xid == vx_current_xid(),
4774 + "xid(%d) == current(%d)", xid, vx_current_xid());
4776 * We may need two file_lock structures for this operation,
4777 * so we get them in advance to avoid races.
4778 @@ -819,7 +842,11 @@ static int __posix_lock_file(struct inod
4779 (request->fl_type != F_UNLCK ||
4780 request->fl_start != 0 || request->fl_end != OFFSET_MAX)) {
4781 new_fl = locks_alloc_lock();
4782 + new_fl->fl_xid = xid;
4783 + vx_locks_inc(new_fl);
4784 new_fl2 = locks_alloc_lock();
4785 + new_fl2->fl_xid = xid;
4786 + vx_locks_inc(new_fl2);
4790 @@ -1018,7 +1045,8 @@ static int __posix_lock_file(struct inod
4791 int posix_lock_file(struct file *filp, struct file_lock *fl,
4792 struct file_lock *conflock)
4794 - return __posix_lock_file(filp->f_path.dentry->d_inode, fl, conflock);
4795 + return __posix_lock_file(filp->f_path.dentry->d_inode,
4796 + fl, conflock, filp->f_xid);
4798 EXPORT_SYMBOL(posix_lock_file);
4800 @@ -1108,7 +1136,7 @@ int locks_mandatory_area(int read_write,
4801 fl.fl_end = offset + count - 1;
4804 - error = __posix_lock_file(inode, &fl, NULL);
4805 + error = __posix_lock_file(inode, &fl, NULL, filp->f_xid);
4806 if (error != FILE_LOCK_DEFERRED)
4808 error = wait_event_interruptible(fl.fl_wait, !fl.fl_next);
4809 @@ -1423,6 +1451,7 @@ int generic_setlease(struct file *filp,
4811 locks_copy_lock(new_fl, lease);
4812 locks_insert_lock(before, new_fl);
4813 + vx_locks_inc(new_fl);
4817 @@ -1778,6 +1807,11 @@ int fcntl_setlk(unsigned int fd, struct
4818 if (file_lock == NULL)
4821 + vxd_assert(filp->f_xid == vx_current_xid(),
4822 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4823 + file_lock->fl_xid = filp->f_xid;
4824 + vx_locks_inc(file_lock);
4827 * This might block, so we do it before checking the inode.
4829 @@ -1896,6 +1930,11 @@ int fcntl_setlk64(unsigned int fd, struc
4830 if (file_lock == NULL)
4833 + vxd_assert(filp->f_xid == vx_current_xid(),
4834 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4835 + file_lock->fl_xid = filp->f_xid;
4836 + vx_locks_inc(file_lock);
4839 * This might block, so we do it before checking the inode.
4841 @@ -2161,8 +2200,11 @@ static int locks_show(struct seq_file *f
4843 lock_get_status(f, fl, (long)f->private, "");
4845 - list_for_each_entry(bfl, &fl->fl_block, fl_block)
4846 + list_for_each_entry(bfl, &fl->fl_block, fl_block) {
4847 + if (!vx_check(fl->fl_xid, VS_WATCH_P | VS_IDENT))
4849 lock_get_status(f, bfl, (long)f->private, " ->");
4854 diff -NurpP --minimal linux-2.6.29.1/fs/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/namei.c
4855 --- linux-2.6.29.1/fs/namei.c 2009-03-24 14:22:26.000000000 +0100
4856 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/namei.c 2009-03-25 00:50:11.000000000 +0100
4858 #include <linux/file.h>
4859 #include <linux/fcntl.h>
4860 #include <linux/device_cgroup.h>
4861 +#include <linux/proc_fs.h>
4862 +#include <linux/vserver/inode.h>
4863 +#include <linux/vs_base.h>
4864 +#include <linux/vs_tag.h>
4865 +#include <linux/vs_cowbl.h>
4866 +#include <linux/vs_device.h>
4867 +#include <linux/vs_context.h>
4868 +#include <linux/pid_namespace.h>
4869 #include <asm/uaccess.h>
4871 #define ACC_MODE(x) ("\000\004\002\006"[(x)&O_ACCMODE])
4872 @@ -167,6 +175,77 @@ void putname(const char *name)
4873 EXPORT_SYMBOL(putname);
4876 +static inline int dx_barrier(struct inode *inode)
4878 + if (IS_BARRIER(inode) && !vx_check(0, VS_ADMIN | VS_WATCH)) {
4879 + vxwprintk_task(1, "did hit the barrier.");
4885 +static int __dx_permission(struct inode *inode, int mask)
4887 + if (dx_barrier(inode))
4890 + if (inode->i_sb->s_magic == DEVPTS_SUPER_MAGIC) {
4891 + /* devpts is xid tagged */
4892 + if (S_ISDIR(inode->i_mode) ||
4893 + vx_check((xid_t)inode->i_tag, VS_IDENT | VS_WATCH_P))
4896 + else if (inode->i_sb->s_magic == PROC_SUPER_MAGIC) {
4897 + struct proc_dir_entry *de = PDE(inode);
4899 + if (de && !vx_hide_check(0, de->vx_flags))
4902 + if ((mask & (MAY_WRITE | MAY_APPEND))) {
4904 + struct task_struct *tsk;
4906 + if (vx_check(0, VS_ADMIN | VS_WATCH_P) ||
4907 + vx_flags(VXF_STATE_SETUP, 0))
4910 + pid = PROC_I(inode)->pid;
4914 + tsk = pid_task(pid, PIDTYPE_PID);
4915 + vxdprintk(VXD_CBIT(tag, 0), "accessing %p[#%u]",
4916 + tsk, (tsk ? vx_task_xid(tsk) : 0));
4917 + if (tsk && vx_check(vx_task_xid(tsk), VS_IDENT | VS_WATCH_P))
4921 + /* FIXME: Should we block some entries here? */
4926 + if (dx_notagcheck(inode->i_sb) ||
4927 + dx_check(inode->i_tag, DX_HOSTID | DX_ADMIN | DX_WATCH |
4936 +int dx_permission(struct inode *inode, int mask)
4938 + int ret = __dx_permission(inode, mask);
4939 + if (unlikely(ret)) {
4940 + vxwprintk_task(1, "denied %x access to %s:%p[#%d,%lu]",
4941 + mask, inode->i_sb->s_id, inode, inode->i_tag,
4949 * generic_permission - check for access rights on a Posix-like filesystem
4950 @@ -253,10 +332,14 @@ int inode_permission(struct inode *inode
4952 * Nobody gets write access to an immutable file.
4954 - if (IS_IMMUTABLE(inode))
4955 + if (IS_IMMUTABLE(inode) && !IS_COW(inode))
4959 + retval = dx_permission(inode, mask);
4963 if (inode->i_op->permission)
4964 retval = inode->i_op->permission(inode, mask);
4966 @@ -432,6 +515,8 @@ static int exec_permission_lite(struct i
4968 umode_t mode = inode->i_mode;
4970 + if (dx_barrier(inode))
4972 if (inode->i_op->permission)
4975 @@ -749,7 +834,8 @@ static __always_inline void follow_dotdo
4976 if (nd->path.dentry == fs->root.dentry &&
4977 nd->path.mnt == fs->root.mnt) {
4978 read_unlock(&fs->lock);
4980 + /* for sane '/' avoid follow_mount() */
4983 read_unlock(&fs->lock);
4984 spin_lock(&dcache_lock);
4985 @@ -786,16 +872,30 @@ static int do_lookup(struct nameidata *n
4987 struct vfsmount *mnt = nd->path.mnt;
4988 struct dentry *dentry = __d_lookup(nd->path.dentry, name);
4989 + struct inode *inode;
4993 if (dentry->d_op && dentry->d_op->d_revalidate)
4994 goto need_revalidate;
4995 + inode = dentry->d_inode;
4999 + if (__dx_permission(inode, MAY_ACCESS))
5004 path->dentry = dentry;
5005 __follow_mount(path);
5008 + vxwprintk_task(1, "did lookup hidden %s:%p[#%d,%lu] »%s«.",
5009 + inode->i_sb->s_id, inode, inode->i_tag, inode->i_ino,
5010 + vxd_path(&nd->path));
5015 dentry = real_lookup(nd->path.dentry, name, nd);
5016 @@ -1364,7 +1464,7 @@ static int may_delete(struct inode *dir,
5019 if (check_sticky(dir, victim->d_inode)||IS_APPEND(victim->d_inode)||
5020 - IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5021 + IS_IXORUNLINK(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5024 if (!S_ISDIR(victim->d_inode->i_mode))
5025 @@ -1506,6 +1606,14 @@ int may_open(struct path *path, int acc_
5029 +#ifdef CONFIG_VSERVER_COWBL
5030 + if (IS_COW(inode) && (flag & FMODE_WRITE)) {
5031 + if (IS_COW_LINK(inode))
5033 + inode->i_flags &= ~(S_IXUNLINK|S_IMMUTABLE);
5034 + mark_inode_dirty(inode);
5037 error = inode_permission(inode, acc_mode);
5040 @@ -1639,6 +1747,11 @@ struct file *do_filp_open(int dfd, const
5042 int flag = open_to_namei_flags(open_flag);
5044 +#ifdef CONFIG_VSERVER_COWBL
5049 acc_mode = MAY_OPEN | ACC_MODE(flag);
5051 /* O_TRUNC implies we need access checks for write permissions */
5052 @@ -1771,6 +1884,25 @@ ok:
5055 error = may_open(&nd.path, acc_mode, flag);
5056 +#ifdef CONFIG_VSERVER_COWBL
5057 + if (error == -EMLINK) {
5058 + struct dentry *dentry;
5059 + dentry = cow_break_link(pathname);
5060 + if (IS_ERR(dentry)) {
5061 + error = PTR_ERR(dentry);
5066 + mnt_drop_write(nd.path.mnt);
5067 + release_open_intent(&nd);
5068 + path_put(&nd.path);
5077 mnt_drop_write(nd.path.mnt);
5078 @@ -1924,9 +2056,17 @@ int vfs_mknod(struct inode *dir, struct
5082 - if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
5083 + if (!(S_ISCHR(mode) || S_ISBLK(mode)))
5086 + if (!capable(CAP_MKNOD))
5089 + if (S_ISCHR(mode) && !vs_chrdev_perm(dev, DATTR_CREATE))
5091 + if (S_ISBLK(mode) && !vs_blkdev_perm(dev, DATTR_CREATE))
5094 if (!dir->i_op->mknod)
5097 @@ -2393,7 +2533,7 @@ int vfs_link(struct dentry *old_dentry,
5099 * A link to an append-only or immutable file cannot be created.
5101 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
5102 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
5104 if (!dir->i_op->link)
5106 @@ -2766,6 +2906,219 @@ int vfs_follow_link(struct nameidata *nd
5107 return __vfs_follow_link(nd, link);
5111 +#ifdef CONFIG_VSERVER_COWBL
5113 +#include <linux/file.h>
5116 +long do_cow_splice(struct file *in, struct file *out, size_t len)
5120 + return do_splice_direct(in, &ppos, out, len, 0);
5123 +struct dentry *cow_break_link(const char *pathname)
5125 + int ret, mode, pathlen, redo = 0;
5126 + struct nameidata old_nd, dir_nd;
5127 + struct path old_path, new_path;
5128 + struct dentry *dir, *res = NULL;
5129 + struct file *old_file;
5130 + struct file *new_file;
5131 + char *to, *path, pad='\251';
5134 + vxdprintk(VXD_CBIT(misc, 1), "cow_break_link(»%s«)", pathname);
5135 + path = kmalloc(PATH_MAX, GFP_KERNEL);
5140 + /* old_nd will have refs to dentry and mnt */
5141 + ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5142 + vxdprintk(VXD_CBIT(misc, 2), "path_lookup(old): %d", ret);
5144 + goto out_free_path;
5146 + old_path = old_nd.path;
5147 + mode = old_path.dentry->d_inode->i_mode;
5149 + to = d_path(&old_path, path, PATH_MAX-2);
5150 + pathlen = strlen(to);
5151 + vxdprintk(VXD_CBIT(misc, 2), "old path »%s« [»%.*s«:%d]", to,
5152 + old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5153 + old_path.dentry->d_name.len);
5155 + to[pathlen + 1] = 0;
5157 + to[pathlen] = pad--;
5159 + if (pad <= '\240')
5162 + vxdprintk(VXD_CBIT(misc, 1), "temp copy »%s«", to);
5163 + /* dir_nd will have refs to dentry and mnt */
5164 + ret = path_lookup(to,
5165 + LOOKUP_PARENT | LOOKUP_OPEN | LOOKUP_CREATE, &dir_nd);
5166 + vxdprintk(VXD_CBIT(misc, 2),
5167 + "path_lookup(new): %d", ret);
5171 + /* this puppy downs the inode mutex */
5172 + new_path.dentry = lookup_create(&dir_nd, 0);
5173 + if (!new_path.dentry || IS_ERR(new_path.dentry)) {
5174 + vxdprintk(VXD_CBIT(misc, 2),
5175 + "lookup_create(new): %p", new_path.dentry);
5176 + mutex_unlock(&dir_nd.path.dentry->d_inode->i_mutex);
5177 + path_put(&dir_nd.path);
5180 + vxdprintk(VXD_CBIT(misc, 2),
5181 + "lookup_create(new): %p [»%.*s«:%d]", new_path.dentry,
5182 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5183 + new_path.dentry->d_name.len);
5184 + dir = dir_nd.path.dentry;
5186 + ret = vfs_create(dir_nd.path.dentry->d_inode, new_path.dentry, mode, &dir_nd);
5187 + vxdprintk(VXD_CBIT(misc, 2),
5188 + "vfs_create(new): %d", ret);
5189 + if (ret == -EEXIST) {
5190 + mutex_unlock(&dir->d_inode->i_mutex);
5191 + dput(new_path.dentry);
5192 + path_put(&dir_nd.path);
5196 + goto out_unlock_new;
5198 + /* drop out early, ret passes ENOENT */
5200 + if ((redo = d_unhashed(old_path.dentry)))
5201 + goto out_unlock_new;
5203 + new_path.mnt = dir_nd.path.mnt;
5204 + dget(old_path.dentry);
5205 + mntget(old_path.mnt);
5206 + /* this one cleans up the dentry/mnt in case of failure */
5207 + old_file = dentry_open(old_path.dentry, old_path.mnt,
5208 + O_RDONLY, current_cred());
5209 + vxdprintk(VXD_CBIT(misc, 2),
5210 + "dentry_open(old): %p", old_file);
5211 + if (!old_file || IS_ERR(old_file)) {
5212 + res = IS_ERR(old_file) ? (void *) old_file : res;
5213 + goto out_unlock_new;
5216 + dget(new_path.dentry);
5217 + mntget(new_path.mnt);
5218 + /* this one cleans up the dentry/mnt in case of failure */
5219 + new_file = dentry_open(new_path.dentry, new_path.mnt,
5220 + O_WRONLY, current_cred());
5221 + vxdprintk(VXD_CBIT(misc, 2),
5222 + "dentry_open(new): %p", new_file);
5224 + ret = IS_ERR(new_file) ? PTR_ERR(new_file) : -ENOENT;
5225 + if (!new_file || IS_ERR(new_file))
5226 + goto out_fput_old;
5228 + size = i_size_read(old_file->f_dentry->d_inode);
5229 + ret = do_cow_splice(old_file, new_file, size);
5230 + vxdprintk(VXD_CBIT(misc, 2), "do_splice_direct: %d", ret);
5232 + goto out_fput_both;
5233 + } else if (ret < size) {
5235 + goto out_fput_both;
5237 + struct inode *old_inode = old_path.dentry->d_inode;
5238 + struct inode *new_inode = new_path.dentry->d_inode;
5239 + struct iattr attr = {
5240 + .ia_uid = old_inode->i_uid,
5241 + .ia_gid = old_inode->i_gid,
5242 + .ia_valid = ATTR_UID | ATTR_GID
5245 + ret = inode_setattr(new_inode, &attr);
5247 + goto out_fput_both;
5250 + mutex_lock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5252 + /* drop out late */
5254 + if ((redo = d_unhashed(old_path.dentry)))
5257 + vxdprintk(VXD_CBIT(misc, 2),
5258 + "vfs_rename: [»%*s«:%d] -> [»%*s«:%d]",
5259 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5260 + new_path.dentry->d_name.len,
5261 + old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5262 + old_path.dentry->d_name.len);
5263 + ret = vfs_rename(dir_nd.path.dentry->d_inode, new_path.dentry,
5264 + old_nd.path.dentry->d_parent->d_inode, old_path.dentry);
5265 + vxdprintk(VXD_CBIT(misc, 2), "vfs_rename: %d", ret);
5266 + res = new_path.dentry;
5269 + mutex_unlock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5272 + vxdprintk(VXD_CBIT(misc, 3),
5273 + "fput(new_file=%p[#%ld])", new_file,
5274 + atomic_read(&new_file->f_count));
5278 + vxdprintk(VXD_CBIT(misc, 3),
5279 + "fput(old_file=%p[#%ld])", old_file,
5280 + atomic_read(&old_file->f_count));
5284 + mutex_unlock(&dir->d_inode->i_mutex);
5288 + /* error path cleanup */
5289 + vfs_unlink(dir->d_inode, new_path.dentry);
5290 + dput(new_path.dentry);
5294 + goto out_rel_both;
5295 + /* lookup dentry once again */
5296 + path_put(&old_nd.path);
5297 + ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5299 + goto out_rel_both;
5301 + new_path.dentry = old_nd.path.dentry;
5302 + vxdprintk(VXD_CBIT(misc, 2),
5303 + "path_lookup(redo): %p [»%.*s«:%d]", new_path.dentry,
5304 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5305 + new_path.dentry->d_name.len);
5306 + dget(new_path.dentry);
5307 + res = new_path.dentry;
5310 + path_put(&dir_nd.path);
5312 + path_put(&old_nd.path);
5317 + res = ERR_PTR(ret);
5323 /* get the link contents into pagecache */
5324 static char *page_getlink(struct dentry * dentry, struct page **ppage)
5326 diff -NurpP --minimal linux-2.6.29.1/fs/namespace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/namespace.c
5327 --- linux-2.6.29.1/fs/namespace.c 2009-03-24 14:22:26.000000000 +0100
5328 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/namespace.c 2009-04-10 23:15:11.000000000 +0200
5330 #include <linux/ramfs.h>
5331 #include <linux/log2.h>
5332 #include <linux/idr.h>
5333 +#include <linux/vs_base.h>
5334 +#include <linux/vs_context.h>
5335 +#include <linux/vs_tag.h>
5336 +#include <linux/vserver/space.h>
5337 +#include <linux/vserver/global.h>
5338 #include <asm/uaccess.h>
5339 #include <asm/unistd.h>
5341 @@ -573,6 +578,7 @@ static struct vfsmount *clone_mnt(struct
5342 mnt->mnt_root = dget(root);
5343 mnt->mnt_mountpoint = mnt->mnt_root;
5344 mnt->mnt_parent = mnt;
5345 + mnt->mnt_tag = old->mnt_tag;
5347 if (flag & CL_SLAVE) {
5348 list_add(&mnt->mnt_slave, &old->mnt_slave_list);
5349 @@ -687,6 +693,31 @@ static inline void mangle(struct seq_fil
5350 seq_escape(m, s, " \t\n\\");
5353 +static int mnt_is_reachable(struct vfsmount *mnt)
5356 + struct dentry *point;
5359 + if (mnt == mnt->mnt_ns->root)
5362 + spin_lock(&vfsmount_lock);
5363 + root = current->fs->root;
5364 + point = root.dentry;
5366 + while ((mnt != mnt->mnt_parent) && (mnt != root.mnt)) {
5367 + point = mnt->mnt_mountpoint;
5368 + mnt = mnt->mnt_parent;
5371 + ret = (mnt == root.mnt) && is_subdir(point, root.dentry);
5373 + spin_unlock(&vfsmount_lock);
5379 * Simple .show_options callback for filesystems which don't want to
5380 * implement more complex mount option showing.
5381 @@ -759,6 +790,8 @@ static int show_sb_opts(struct seq_file
5382 { MS_SYNCHRONOUS, ",sync" },
5383 { MS_DIRSYNC, ",dirsync" },
5384 { MS_MANDLOCK, ",mand" },
5385 + { MS_TAGGED, ",tag" },
5386 + { MS_NOTAGCHECK, ",notagcheck" },
5389 const struct proc_fs_info *fs_infop;
5390 @@ -805,10 +838,20 @@ static int show_vfsmnt(struct seq_file *
5392 struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5394 - mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5396 - seq_path(m, &mnt_path, " \t\n\\");
5398 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5400 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5403 + if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5404 + mnt == current->fs->root.mnt) {
5405 + seq_puts(m, "/dev/root / ");
5407 + mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5409 + seq_path(m, &mnt_path, " \t\n\\");
5412 show_type(m, mnt->mnt_sb);
5413 seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
5414 err = show_sb_opts(m, mnt->mnt_sb);
5415 @@ -838,6 +881,11 @@ static int show_mountinfo(struct seq_fil
5416 struct path root = p->root;
5419 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5421 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5424 seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
5425 MAJOR(sb->s_dev), MINOR(sb->s_dev));
5426 seq_dentry(m, mnt->mnt_root, " \t\n\\");
5427 @@ -896,17 +944,27 @@ static int show_vfsstat(struct seq_file
5428 struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5432 - if (mnt->mnt_devname) {
5433 - seq_puts(m, "device ");
5434 - mangle(m, mnt->mnt_devname);
5436 - seq_puts(m, "no device");
5437 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5439 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5443 - seq_puts(m, " mounted on ");
5444 - seq_path(m, &mnt_path, " \t\n\\");
5446 + if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5447 + mnt == current->fs->root.mnt) {
5448 + seq_puts(m, "device /dev/root mounted on / ");
5451 + if (mnt->mnt_devname) {
5452 + seq_puts(m, "device ");
5453 + mangle(m, mnt->mnt_devname);
5455 + seq_puts(m, "no device");
5458 + seq_puts(m, " mounted on ");
5459 + seq_path(m, &mnt_path, " \t\n\\");
5463 /* file system type */
5464 seq_puts(m, "with fstype ");
5465 @@ -1145,7 +1203,7 @@ SYSCALL_DEFINE2(umount, char __user *, n
5469 - if (!capable(CAP_SYS_ADMIN))
5470 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5473 retval = do_umount(path.mnt, flags);
5474 @@ -1171,7 +1229,7 @@ SYSCALL_DEFINE1(oldumount, char __user *
5476 static int mount_is_safe(struct path *path)
5478 - if (capable(CAP_SYS_ADMIN))
5479 + if (vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5483 @@ -1462,11 +1520,13 @@ static int do_change_type(struct path *p
5484 * do loopback mount.
5486 static int do_loopback(struct path *path, char *old_name,
5488 + tag_t tag, unsigned long flags, int mnt_flags)
5490 struct path old_path;
5491 struct vfsmount *mnt = NULL;
5492 int err = mount_is_safe(path);
5493 + int recurse = flags & MS_REC;
5497 if (!old_name || !*old_name)
5498 @@ -1500,6 +1560,7 @@ static int do_loopback(struct path *path
5499 spin_unlock(&vfsmount_lock);
5500 release_mounts(&umount_list);
5502 + mnt->mnt_flags = mnt_flags;
5505 up_write(&namespace_sem);
5506 @@ -1530,12 +1591,12 @@ static int change_mount_flags(struct vfs
5507 * on it - tough luck.
5509 static int do_remount(struct path *path, int flags, int mnt_flags,
5511 + void *data, xid_t xid)
5514 struct super_block *sb = path->mnt->mnt_sb;
5516 - if (!capable(CAP_SYS_ADMIN))
5517 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_REMOUNT))
5520 if (!check_mnt(path->mnt))
5521 @@ -1577,7 +1638,7 @@ static int do_move_mount(struct path *pa
5522 struct path old_path, parent_path;
5525 - if (!capable(CAP_SYS_ADMIN))
5526 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5528 if (!old_name || !*old_name)
5530 @@ -1659,7 +1720,7 @@ static int do_new_mount(struct path *pat
5533 /* we need capabilities... */
5534 - if (!capable(CAP_SYS_ADMIN))
5535 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5538 mnt = do_kern_mount(type, flags, name, data);
5539 @@ -1904,6 +1965,7 @@ long do_mount(char *dev_name, char *dir_
5546 if ((flags & MS_MGC_MSK) == MS_MGC_VAL)
5547 @@ -1919,6 +1981,12 @@ long do_mount(char *dev_name, char *dir_
5549 ((char *)data_page)[PAGE_SIZE - 1] = 0;
5551 + if (dx_parse_tag(data_page, &tag, 1, &mnt_flags, &flags)) {
5552 + /* FIXME: bind and re-mounts get the tag flag? */
5553 + if (flags & (MS_BIND|MS_REMOUNT))
5554 + flags |= MS_TAGID;
5557 /* Separate the per-mountpoint flags */
5558 if (flags & MS_NOSUID)
5559 mnt_flags |= MNT_NOSUID;
5560 @@ -1935,6 +2003,8 @@ long do_mount(char *dev_name, char *dir_
5561 if (flags & MS_RDONLY)
5562 mnt_flags |= MNT_READONLY;
5564 + if (!capable(CAP_SYS_ADMIN))
5565 + mnt_flags |= MNT_NODEV;
5566 flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE |
5567 MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT);
5569 @@ -1950,9 +2020,9 @@ long do_mount(char *dev_name, char *dir_
5571 if (flags & MS_REMOUNT)
5572 retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
5575 else if (flags & MS_BIND)
5576 - retval = do_loopback(&path, dev_name, flags & MS_REC);
5577 + retval = do_loopback(&path, dev_name, tag, flags, mnt_flags);
5578 else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
5579 retval = do_change_type(&path, flags);
5580 else if (flags & MS_MOVE)
5581 @@ -2021,6 +2091,7 @@ static struct mnt_namespace *dup_mnt_ns(
5582 q = next_mnt(q, new_ns->root);
5584 up_write(&namespace_sem);
5585 + atomic_inc(&vs_global_mnt_ns);
5589 @@ -2207,9 +2278,10 @@ SYSCALL_DEFINE2(pivot_root, const char _
5590 down_write(&namespace_sem);
5591 mutex_lock(&old.dentry->d_inode->i_mutex);
5593 - if (IS_MNT_SHARED(old.mnt) ||
5594 + if ((IS_MNT_SHARED(old.mnt) ||
5595 IS_MNT_SHARED(new.mnt->mnt_parent) ||
5596 - IS_MNT_SHARED(root.mnt->mnt_parent))
5597 + IS_MNT_SHARED(root.mnt->mnt_parent)) &&
5598 + !vx_flags(VXF_STATE_SETUP, 0))
5600 if (!check_mnt(root.mnt))
5602 @@ -2348,5 +2420,6 @@ void __put_mnt_ns(struct mnt_namespace *
5603 spin_unlock(&vfsmount_lock);
5604 up_write(&namespace_sem);
5605 release_mounts(&umount_list);
5606 + atomic_dec(&vs_global_mnt_ns);
5609 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/client.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/client.c
5610 --- linux-2.6.29.1/fs/nfs/client.c 2009-03-24 14:22:26.000000000 +0100
5611 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/client.c 2009-03-24 14:48:35.000000000 +0100
5612 @@ -696,6 +696,9 @@ static int nfs_init_server_rpcclient(str
5613 if (server->flags & NFS_MOUNT_SOFT)
5614 server->client->cl_softrtry = 1;
5616 + server->client->cl_tag = 0;
5617 + if (server->flags & NFS_MOUNT_TAGGED)
5618 + server->client->cl_tag = 1;
5622 @@ -862,6 +865,10 @@ static void nfs_server_set_fsinfo(struct
5623 server->acdirmin = server->acdirmax = 0;
5626 + /* FIXME: needs fsinfo
5627 + if (server->flags & NFS_MOUNT_TAGGED)
5628 + sb->s_flags |= MS_TAGGED; */
5630 server->maxfilesize = fsinfo->maxfilesize;
5632 /* We're airborne Set socket buffersize */
5633 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/dir.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/dir.c
5634 --- linux-2.6.29.1/fs/nfs/dir.c 2009-03-24 14:22:26.000000000 +0100
5635 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/dir.c 2009-03-24 14:48:35.000000000 +0100
5637 #include <linux/namei.h>
5638 #include <linux/mount.h>
5639 #include <linux/sched.h>
5640 +#include <linux/vs_tag.h>
5642 #include "nfs4_fs.h"
5643 #include "delegation.h"
5644 @@ -950,6 +951,7 @@ static struct dentry *nfs_lookup(struct
5646 goto out_unblock_sillyrename;
5648 + dx_propagate_tag(nd, inode);
5650 res = d_materialise_unique(dentry, inode);
5652 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/inode.c
5653 --- linux-2.6.29.1/fs/nfs/inode.c 2009-03-24 14:22:26.000000000 +0100
5654 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/inode.c 2009-03-24 14:48:35.000000000 +0100
5656 #include <linux/vfs.h>
5657 #include <linux/inet.h>
5658 #include <linux/nfs_xdr.h>
5659 +#include <linux/vs_tag.h>
5661 #include <asm/system.h>
5662 #include <asm/uaccess.h>
5663 @@ -313,8 +314,10 @@ nfs_fhget(struct super_block *sb, struct
5664 nfsi->change_attr = fattr->change_attr;
5665 inode->i_size = nfs_size_to_loff_t(fattr->size);
5666 inode->i_nlink = fattr->nlink;
5667 - inode->i_uid = fattr->uid;
5668 - inode->i_gid = fattr->gid;
5669 + inode->i_uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5670 + inode->i_gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5671 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5672 + /* maybe fattr->xid someday */
5673 if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
5675 * report the blocks in 512byte units
5676 @@ -462,6 +465,8 @@ void nfs_setattr_update_inode(struct ino
5677 inode->i_uid = attr->ia_uid;
5678 if ((attr->ia_valid & ATTR_GID) != 0)
5679 inode->i_gid = attr->ia_gid;
5680 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
5681 + inode->i_tag = attr->ia_tag;
5682 NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5683 spin_unlock(&inode->i_lock);
5685 @@ -850,6 +855,9 @@ static int nfs_check_inode_attributes(st
5686 struct nfs_inode *nfsi = NFS_I(inode);
5687 loff_t cur_size, new_isize;
5688 unsigned long invalid = 0;
5694 /* Has the inode gone and changed behind our back? */
5695 @@ -871,10 +879,15 @@ static int nfs_check_inode_attributes(st
5696 if (cur_size != new_isize && nfsi->npages == 0)
5697 invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
5699 + uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5700 + gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5701 + tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5703 /* Have any file permissions changed? */
5704 if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)
5705 - || inode->i_uid != fattr->uid
5706 - || inode->i_gid != fattr->gid)
5707 + || inode->i_uid != uid
5708 + || inode->i_gid != gid
5709 + || inode->i_tag != tag)
5710 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5712 /* Has the link count changed? */
5713 @@ -1073,6 +1086,9 @@ static int nfs_update_inode(struct inode
5714 loff_t cur_isize, new_isize;
5715 unsigned long invalid = 0;
5716 unsigned long now = jiffies;
5721 dfprintk(VFS, "NFS: %s(%s/%ld ct=%d info=0x%x)\n",
5722 __func__, inode->i_sb->s_id, inode->i_ino,
5723 @@ -1146,9 +1162,14 @@ static int nfs_update_inode(struct inode
5724 memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
5725 nfsi->change_attr = fattr->change_attr;
5727 + uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5728 + gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5729 + tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5731 if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO) ||
5732 - inode->i_uid != fattr->uid ||
5733 - inode->i_gid != fattr->gid)
5734 + inode->i_uid != uid ||
5735 + inode->i_gid != gid ||
5736 + inode->i_tag != tag)
5737 invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5739 if (inode->i_nlink != fattr->nlink)
5740 @@ -1156,8 +1177,9 @@ static int nfs_update_inode(struct inode
5742 inode->i_mode = fattr->mode;
5743 inode->i_nlink = fattr->nlink;
5744 - inode->i_uid = fattr->uid;
5745 - inode->i_gid = fattr->gid;
5746 + inode->i_uid = uid;
5747 + inode->i_gid = gid;
5748 + inode->i_tag = tag;
5750 if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
5752 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/nfs3xdr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/nfs3xdr.c
5753 --- linux-2.6.29.1/fs/nfs/nfs3xdr.c 2009-03-24 14:22:26.000000000 +0100
5754 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/nfs3xdr.c 2009-03-24 14:48:35.000000000 +0100
5756 #include <linux/nfs3.h>
5757 #include <linux/nfs_fs.h>
5758 #include <linux/nfsacl.h>
5759 +#include <linux/vs_tag.h>
5760 #include "internal.h"
5762 #define NFSDBG_FACILITY NFSDBG_XDR
5763 @@ -182,7 +183,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_f
5766 static inline __be32 *
5767 -xdr_encode_sattr(__be32 *p, struct iattr *attr)
5768 +xdr_encode_sattr(__be32 *p, struct iattr *attr, int tag)
5770 if (attr->ia_valid & ATTR_MODE) {
5772 @@ -190,15 +191,17 @@ xdr_encode_sattr(__be32 *p, struct iattr
5776 - if (attr->ia_valid & ATTR_UID) {
5777 + if (attr->ia_valid & ATTR_UID ||
5778 + (tag && (attr->ia_valid & ATTR_TAG))) {
5780 - *p++ = htonl(attr->ia_uid);
5781 + *p++ = htonl(TAGINO_UID(tag, attr->ia_uid, attr->ia_tag));
5785 - if (attr->ia_valid & ATTR_GID) {
5786 + if (attr->ia_valid & ATTR_GID ||
5787 + (tag && (attr->ia_valid & ATTR_TAG))) {
5789 - *p++ = htonl(attr->ia_gid);
5790 + *p++ = htonl(TAGINO_GID(tag, attr->ia_gid, attr->ia_tag));
5794 @@ -283,7 +286,8 @@ static int
5795 nfs3_xdr_sattrargs(struct rpc_rqst *req, __be32 *p, struct nfs3_sattrargs *args)
5797 p = xdr_encode_fhandle(p, args->fh);
5798 - p = xdr_encode_sattr(p, args->sattr);
5799 + p = xdr_encode_sattr(p, args->sattr,
5800 + req->rq_task->tk_client->cl_tag);
5801 *p++ = htonl(args->guard);
5803 p = xdr_encode_time3(p, &args->guardtime);
5804 @@ -388,7 +392,8 @@ nfs3_xdr_createargs(struct rpc_rqst *req
5805 *p++ = args->verifier[0];
5806 *p++ = args->verifier[1];
5808 - p = xdr_encode_sattr(p, args->sattr);
5809 + p = xdr_encode_sattr(p, args->sattr,
5810 + req->rq_task->tk_client->cl_tag);
5812 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5814 @@ -402,7 +407,8 @@ nfs3_xdr_mkdirargs(struct rpc_rqst *req,
5816 p = xdr_encode_fhandle(p, args->fh);
5817 p = xdr_encode_array(p, args->name, args->len);
5818 - p = xdr_encode_sattr(p, args->sattr);
5819 + p = xdr_encode_sattr(p, args->sattr,
5820 + req->rq_task->tk_client->cl_tag);
5821 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5824 @@ -415,7 +421,8 @@ nfs3_xdr_symlinkargs(struct rpc_rqst *re
5826 p = xdr_encode_fhandle(p, args->fromfh);
5827 p = xdr_encode_array(p, args->fromname, args->fromlen);
5828 - p = xdr_encode_sattr(p, args->sattr);
5829 + p = xdr_encode_sattr(p, args->sattr,
5830 + req->rq_task->tk_client->cl_tag);
5831 *p++ = htonl(args->pathlen);
5832 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5834 @@ -433,7 +440,8 @@ nfs3_xdr_mknodargs(struct rpc_rqst *req,
5835 p = xdr_encode_fhandle(p, args->fh);
5836 p = xdr_encode_array(p, args->name, args->len);
5837 *p++ = htonl(args->type);
5838 - p = xdr_encode_sattr(p, args->sattr);
5839 + p = xdr_encode_sattr(p, args->sattr,
5840 + req->rq_task->tk_client->cl_tag);
5841 if (args->type == NF3CHR || args->type == NF3BLK) {
5842 *p++ = htonl(MAJOR(args->rdev));
5843 *p++ = htonl(MINOR(args->rdev));
5844 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/nfsroot.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/nfsroot.c
5845 --- linux-2.6.29.1/fs/nfs/nfsroot.c 2009-03-24 14:22:26.000000000 +0100
5846 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/nfsroot.c 2009-03-24 14:48:35.000000000 +0100
5847 @@ -119,12 +119,12 @@ static int mount_port __initdata = 0; /
5849 /* Options that take integer arguments */
5850 Opt_port, Opt_rsize, Opt_wsize, Opt_timeo, Opt_retrans, Opt_acregmin,
5851 - Opt_acregmax, Opt_acdirmin, Opt_acdirmax,
5852 + Opt_acregmax, Opt_acdirmin, Opt_acdirmax, Opt_tagid,
5853 /* Options that take no arguments */
5854 Opt_soft, Opt_hard, Opt_intr,
5855 Opt_nointr, Opt_posix, Opt_noposix, Opt_cto, Opt_nocto, Opt_ac,
5856 Opt_noac, Opt_lock, Opt_nolock, Opt_v2, Opt_v3, Opt_udp, Opt_tcp,
5857 - Opt_acl, Opt_noacl,
5858 + Opt_acl, Opt_noacl, Opt_tag, Opt_notag,
5862 @@ -161,6 +161,9 @@ static match_table_t __initconst tokens
5865 {Opt_noacl, "noacl"},
5867 + {Opt_notag, "notag"},
5868 + {Opt_tagid, "tagid=%u"},
5872 @@ -272,6 +275,20 @@ static int __init root_nfs_parse(char *n
5874 nfs_data.flags |= NFS_MOUNT_NOACL;
5876 +#ifndef CONFIG_TAGGING_NONE
5878 + nfs_data.flags |= NFS_MOUNT_TAGGED;
5881 + nfs_data.flags &= ~NFS_MOUNT_TAGGED;
5884 +#ifdef CONFIG_PROPAGATE
5887 + nfs_data.flags |= NFS_MOUNT_TAGGED;
5891 printk(KERN_WARNING "Root-NFS: unknown "
5893 diff -NurpP --minimal linux-2.6.29.1/fs/nfs/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/super.c
5894 --- linux-2.6.29.1/fs/nfs/super.c 2009-03-24 14:22:26.000000000 +0100
5895 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfs/super.c 2009-03-24 16:02:06.000000000 +0100
5897 #include <linux/nfs_xdr.h>
5898 #include <linux/magic.h>
5899 #include <linux/parser.h>
5900 +#include <linux/vs_tag.h>
5902 #include <asm/system.h>
5903 #include <asm/uaccess.h>
5904 @@ -517,6 +518,7 @@ static void nfs_show_mount_options(struc
5905 { NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
5906 { NFS_MOUNT_UNSHARED, ",nosharecache", "" },
5907 { NFS_MOUNT_NORESVPORT, ",noresvport", "" },
5908 + { NFS_MOUNT_TAGGED, ",tag", "" },
5911 const struct proc_nfs_info *nfs_infop;
5912 diff -NurpP --minimal linux-2.6.29.1/fs/nfsd/auth.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/auth.c
5913 --- linux-2.6.29.1/fs/nfsd/auth.c 2009-03-24 14:22:26.000000000 +0100
5914 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/auth.c 2009-03-24 16:09:39.000000000 +0100
5916 #include <linux/sunrpc/svcauth.h>
5917 #include <linux/nfsd/nfsd.h>
5918 #include <linux/nfsd/export.h>
5919 +#include <linux/vs_tag.h>
5922 int nfsexp_flags(struct svc_rqst *rqstp, struct svc_export *exp)
5923 @@ -42,6 +43,9 @@ int nfsd_setuser(struct svc_rqst *rqstp,
5925 new->fsuid = rqstp->rq_cred.cr_uid;
5926 new->fsgid = rqstp->rq_cred.cr_gid;
5927 + /* FIXME: this desperately needs a tag :)
5928 + new->xid = (xid_t)INOTAG_TAG(DX_TAG_NFSD, cred.cr_uid, cred.cr_gid, 0);
5931 rqgi = rqstp->rq_cred.cr_group_info;
5933 @@ -69,7 +73,7 @@ int nfsd_setuser(struct svc_rqst *rqstp,
5936 gi = get_group_info(rqgi);
5940 if (new->fsuid == (uid_t) -1)
5941 new->fsuid = exp->ex_anon_uid;
5942 @@ -94,6 +98,6 @@ oom:
5950 diff -NurpP --minimal linux-2.6.29.1/fs/nfsd/nfs3xdr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfs3xdr.c
5951 --- linux-2.6.29.1/fs/nfsd/nfs3xdr.c 2008-12-25 00:26:37.000000000 +0100
5952 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfs3xdr.c 2009-02-22 22:54:25.000000000 +0100
5954 #include <linux/sunrpc/svc.h>
5955 #include <linux/nfsd/nfsd.h>
5956 #include <linux/nfsd/xdr3.h>
5957 +#include <linux/vs_tag.h>
5960 #define NFSDDBG_FACILITY NFSDDBG_XDR
5961 @@ -108,6 +109,8 @@ static __be32 *
5962 decode_sattr3(__be32 *p, struct iattr *iap)
5970 @@ -117,12 +120,15 @@ decode_sattr3(__be32 *p, struct iattr *i
5973 iap->ia_valid |= ATTR_UID;
5974 - iap->ia_uid = ntohl(*p++);
5975 + uid = ntohl(*p++);
5978 iap->ia_valid |= ATTR_GID;
5979 - iap->ia_gid = ntohl(*p++);
5980 + gid = ntohl(*p++);
5982 + iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
5983 + iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
5984 + iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
5988 @@ -178,8 +184,12 @@ encode_fattr3(struct svc_rqst *rqstp, __
5989 *p++ = htonl(nfs3_ftypes[(stat->mode & S_IFMT) >> 12]);
5990 *p++ = htonl((u32) stat->mode);
5991 *p++ = htonl((u32) stat->nlink);
5992 - *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
5993 - *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
5994 + *p++ = htonl((u32) nfsd_ruid(rqstp,
5995 + TAGINO_UID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
5996 + stat->uid, stat->tag)));
5997 + *p++ = htonl((u32) nfsd_rgid(rqstp,
5998 + TAGINO_GID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
5999 + stat->gid, stat->tag)));
6000 if (S_ISLNK(stat->mode) && stat->size > NFS3_MAXPATHLEN) {
6001 p = xdr_encode_hyper(p, (u64) NFS3_MAXPATHLEN);
6003 diff -NurpP --minimal linux-2.6.29.1/fs/nfsd/nfs4xdr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfs4xdr.c
6004 --- linux-2.6.29.1/fs/nfsd/nfs4xdr.c 2009-03-24 14:22:26.000000000 +0100
6005 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfs4xdr.c 2009-03-24 14:48:35.000000000 +0100
6007 #include <linux/nfs4_acl.h>
6008 #include <linux/sunrpc/gss_api.h>
6009 #include <linux/sunrpc/svcauth_gss.h>
6010 +#include <linux/vs_tag.h>
6012 #define NFSDDBG_FACILITY NFSDDBG_XDR
6014 @@ -1714,14 +1715,18 @@ out_acl:
6015 WRITE32(stat.nlink);
6017 if (bmval1 & FATTR4_WORD1_OWNER) {
6018 - status = nfsd4_encode_user(rqstp, stat.uid, &p, &buflen);
6019 + status = nfsd4_encode_user(rqstp,
6020 + TAGINO_UID(DX_TAG(dentry->d_inode),
6021 + stat.uid, stat.tag), &p, &buflen);
6022 if (status == nfserr_resource)
6027 if (bmval1 & FATTR4_WORD1_OWNER_GROUP) {
6028 - status = nfsd4_encode_group(rqstp, stat.gid, &p, &buflen);
6029 + status = nfsd4_encode_group(rqstp,
6030 + TAGINO_GID(DX_TAG(dentry->d_inode),
6031 + stat.gid, stat.tag), &p, &buflen);
6032 if (status == nfserr_resource)
6035 diff -NurpP --minimal linux-2.6.29.1/fs/nfsd/nfsxdr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfsxdr.c
6036 --- linux-2.6.29.1/fs/nfsd/nfsxdr.c 2008-12-25 00:26:37.000000000 +0100
6037 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/nfsd/nfsxdr.c 2009-02-22 22:54:25.000000000 +0100
6039 #include <linux/nfsd/nfsd.h>
6040 #include <linux/nfsd/xdr.h>
6041 #include <linux/mm.h>
6042 +#include <linux/vs_tag.h>
6045 #define NFSDDBG_FACILITY NFSDDBG_XDR
6046 @@ -98,6 +99,8 @@ static __be32 *
6047 decode_sattr(__be32 *p, struct iattr *iap)
6055 @@ -111,12 +114,15 @@ decode_sattr(__be32 *p, struct iattr *ia
6057 if ((tmp = ntohl(*p++)) != (u32)-1) {
6058 iap->ia_valid |= ATTR_UID;
6059 - iap->ia_uid = tmp;
6062 if ((tmp = ntohl(*p++)) != (u32)-1) {
6063 iap->ia_valid |= ATTR_GID;
6064 - iap->ia_gid = tmp;
6067 + iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
6068 + iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
6069 + iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
6070 if ((tmp = ntohl(*p++)) != (u32)-1) {
6071 iap->ia_valid |= ATTR_SIZE;
6073 @@ -161,8 +167,10 @@ encode_fattr(struct svc_rqst *rqstp, __b
6074 *p++ = htonl(nfs_ftypes[type >> 12]);
6075 *p++ = htonl((u32) stat->mode);
6076 *p++ = htonl((u32) stat->nlink);
6077 - *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
6078 - *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
6079 + *p++ = htonl((u32) nfsd_ruid(rqstp,
6080 + TAGINO_UID(DX_TAG(dentry->d_inode), stat->uid, stat->tag)));
6081 + *p++ = htonl((u32) nfsd_rgid(rqstp,
6082 + TAGINO_GID(DX_TAG(dentry->d_inode), stat->gid, stat->tag)));
6084 if (S_ISLNK(type) && stat->size > NFS_MAXPATHLEN) {
6085 *p++ = htonl(NFS_MAXPATHLEN);
6086 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/dlm/dlmfs.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlm/dlmfs.c
6087 --- linux-2.6.29.1/fs/ocfs2/dlm/dlmfs.c 2009-03-24 14:22:27.000000000 +0100
6088 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlm/dlmfs.c 2009-03-24 16:10:48.000000000 +0100
6090 #include <linux/init.h>
6091 #include <linux/string.h>
6092 #include <linux/backing-dev.h>
6093 +#include <linux/vs_tag.h>
6095 #include <asm/uaccess.h>
6097 @@ -341,6 +342,7 @@ static struct inode *dlmfs_get_root_inod
6098 inode->i_mode = mode;
6099 inode->i_uid = current_fsuid();
6100 inode->i_gid = current_fsgid();
6101 + inode->i_tag = dx_current_fstag(sb);
6102 inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
6103 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
6105 @@ -366,6 +368,7 @@ static struct inode *dlmfs_get_inode(str
6106 inode->i_mode = mode;
6107 inode->i_uid = current_fsuid();
6108 inode->i_gid = current_fsgid();
6109 + inode->i_tag = dx_current_fstag(sb);
6110 inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
6111 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
6113 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/dlmglue.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlmglue.c
6114 --- linux-2.6.29.1/fs/ocfs2/dlmglue.c 2009-03-24 14:22:27.000000000 +0100
6115 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlmglue.c 2009-03-24 14:48:35.000000000 +0100
6116 @@ -1885,6 +1885,7 @@ static void __ocfs2_stuff_meta_lvb(struc
6117 lvb->lvb_iclusters = cpu_to_be32(oi->ip_clusters);
6118 lvb->lvb_iuid = cpu_to_be32(inode->i_uid);
6119 lvb->lvb_igid = cpu_to_be32(inode->i_gid);
6120 + lvb->lvb_itag = cpu_to_be16(inode->i_tag);
6121 lvb->lvb_imode = cpu_to_be16(inode->i_mode);
6122 lvb->lvb_inlink = cpu_to_be16(inode->i_nlink);
6123 lvb->lvb_iatime_packed =
6124 @@ -1939,6 +1940,7 @@ static void ocfs2_refresh_inode_from_lvb
6126 inode->i_uid = be32_to_cpu(lvb->lvb_iuid);
6127 inode->i_gid = be32_to_cpu(lvb->lvb_igid);
6128 + inode->i_tag = be16_to_cpu(lvb->lvb_itag);
6129 inode->i_mode = be16_to_cpu(lvb->lvb_imode);
6130 inode->i_nlink = be16_to_cpu(lvb->lvb_inlink);
6131 ocfs2_unpack_timespec(&inode->i_atime,
6132 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/dlmglue.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlmglue.h
6133 --- linux-2.6.29.1/fs/ocfs2/dlmglue.h 2009-03-24 14:22:27.000000000 +0100
6134 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/dlmglue.h 2009-03-24 14:48:35.000000000 +0100
6135 @@ -46,7 +46,8 @@ struct ocfs2_meta_lvb {
6138 __be32 lvb_igeneration;
6139 - __be32 lvb_reserved2;
6141 + __be16 lvb_reserved2;
6144 #define OCFS2_QINFO_LVB_VERSION 1
6145 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/file.c
6146 --- linux-2.6.29.1/fs/ocfs2/file.c 2009-03-24 14:22:27.000000000 +0100
6147 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/file.c 2009-03-24 14:48:35.000000000 +0100
6148 @@ -911,13 +911,15 @@ int ocfs2_setattr(struct dentry *dentry,
6149 mlog(0, "uid change: %d\n", attr->ia_uid);
6150 if (attr->ia_valid & ATTR_GID)
6151 mlog(0, "gid change: %d\n", attr->ia_gid);
6152 + if (attr->ia_valid & ATTR_TAG)
6153 + mlog(0, "tag change: %d\n", attr->ia_tag);
6154 if (attr->ia_valid & ATTR_SIZE)
6155 mlog(0, "size change...\n");
6156 if (attr->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME))
6157 mlog(0, "time change...\n");
6159 #define OCFS2_VALID_ATTRS (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME | ATTR_SIZE \
6160 - | ATTR_GID | ATTR_UID | ATTR_MODE)
6161 + | ATTR_GID | ATTR_UID | ATTR_TAG | ATTR_MODE)
6162 if (!(attr->ia_valid & OCFS2_VALID_ATTRS)) {
6163 mlog(0, "can't handle attrs: 0x%x\n", attr->ia_valid);
6165 @@ -2070,6 +2072,7 @@ const struct inode_operations ocfs2_file
6166 .removexattr = generic_removexattr,
6167 .fallocate = ocfs2_fallocate,
6168 .fiemap = ocfs2_fiemap,
6169 + .sync_flags = ocfs2_sync_flags,
6172 const struct inode_operations ocfs2_special_file_iops = {
6173 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/inode.c
6174 --- linux-2.6.29.1/fs/ocfs2/inode.c 2009-03-24 14:22:27.000000000 +0100
6175 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/inode.c 2009-03-24 16:11:11.000000000 +0100
6177 #include <linux/highmem.h>
6178 #include <linux/pagemap.h>
6179 #include <linux/quotaops.h>
6180 +#include <linux/vs_tag.h>
6182 #include <asm/byteorder.h>
6186 #include "heartbeat.h"
6189 #include "journal.h"
6191 #include "suballoc.h"
6192 @@ -77,11 +79,13 @@ void ocfs2_set_inode_flags(struct inode
6194 unsigned int flags = OCFS2_I(inode)->ip_attr;
6196 - inode->i_flags &= ~(S_IMMUTABLE |
6197 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
6198 S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
6200 if (flags & OCFS2_IMMUTABLE_FL)
6201 inode->i_flags |= S_IMMUTABLE;
6202 + if (flags & OCFS2_IXUNLINK_FL)
6203 + inode->i_flags |= S_IXUNLINK;
6205 if (flags & OCFS2_SYNC_FL)
6206 inode->i_flags |= S_SYNC;
6207 @@ -91,25 +95,85 @@ void ocfs2_set_inode_flags(struct inode
6208 inode->i_flags |= S_NOATIME;
6209 if (flags & OCFS2_DIRSYNC_FL)
6210 inode->i_flags |= S_DIRSYNC;
6212 + inode->i_vflags &= ~(V_BARRIER | V_COW);
6214 + if (flags & OCFS2_BARRIER_FL)
6215 + inode->i_vflags |= V_BARRIER;
6216 + if (flags & OCFS2_COW_FL)
6217 + inode->i_vflags |= V_COW;
6220 /* Propagate flags from i_flags to OCFS2_I(inode)->ip_attr */
6221 void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi)
6223 unsigned int flags = oi->vfs_inode.i_flags;
6224 + unsigned int vflags = oi->vfs_inode.i_vflags;
6226 + oi->ip_attr &= ~(OCFS2_SYNC_FL | OCFS2_APPEND_FL |
6227 + OCFS2_IMMUTABLE_FL | OCFS2_IXUNLINK_FL |
6228 + OCFS2_NOATIME_FL | OCFS2_DIRSYNC_FL |
6229 + OCFS2_BARRIER_FL | OCFS2_COW_FL);
6231 + if (flags & S_IMMUTABLE)
6232 + oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6233 + if (flags & S_IXUNLINK)
6234 + oi->ip_attr |= OCFS2_IXUNLINK_FL;
6236 - oi->ip_attr &= ~(OCFS2_SYNC_FL|OCFS2_APPEND_FL|
6237 - OCFS2_IMMUTABLE_FL|OCFS2_NOATIME_FL|OCFS2_DIRSYNC_FL);
6239 oi->ip_attr |= OCFS2_SYNC_FL;
6240 if (flags & S_APPEND)
6241 oi->ip_attr |= OCFS2_APPEND_FL;
6242 - if (flags & S_IMMUTABLE)
6243 - oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6244 if (flags & S_NOATIME)
6245 oi->ip_attr |= OCFS2_NOATIME_FL;
6246 if (flags & S_DIRSYNC)
6247 oi->ip_attr |= OCFS2_DIRSYNC_FL;
6249 + if (vflags & V_BARRIER)
6250 + oi->ip_attr |= OCFS2_BARRIER_FL;
6251 + if (vflags & V_COW)
6252 + oi->ip_attr |= OCFS2_COW_FL;
6255 +int ocfs2_sync_flags(struct inode *inode)
6257 + struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6258 + struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
6259 + handle_t *handle = NULL;
6260 + struct buffer_head *bh = NULL;
6263 + status = ocfs2_inode_lock(inode, &bh, 1);
6265 + mlog_errno(status);
6270 + if (IS_RDONLY(inode))
6273 + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6274 + if (IS_ERR(handle)) {
6275 + status = PTR_ERR(handle);
6276 + mlog_errno(status);
6280 + ocfs2_get_inode_flags(ocfs2_inode);
6281 + status = ocfs2_mark_inode_dirty(handle, inode, bh);
6283 + mlog_errno(status);
6285 + ocfs2_commit_trans(osb, handle);
6287 + ocfs2_inode_unlock(inode, 1);
6292 + mlog_exit(status);
6296 struct inode *ocfs2_iget(struct ocfs2_super *osb, u64 blkno, unsigned flags,
6297 @@ -222,6 +286,8 @@ void ocfs2_populate_inode(struct inode *
6298 struct super_block *sb;
6299 struct ocfs2_super *osb;
6304 mlog_entry("(0x%p, size:%llu)\n", inode,
6305 (unsigned long long)le64_to_cpu(fe->i_size));
6306 @@ -253,8 +319,12 @@ void ocfs2_populate_inode(struct inode *
6307 inode->i_generation = le32_to_cpu(fe->i_generation);
6308 inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
6309 inode->i_mode = le16_to_cpu(fe->i_mode);
6310 - inode->i_uid = le32_to_cpu(fe->i_uid);
6311 - inode->i_gid = le32_to_cpu(fe->i_gid);
6312 + uid = le32_to_cpu(fe->i_uid);
6313 + gid = le32_to_cpu(fe->i_gid);
6314 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6315 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6316 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6317 + /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6319 /* Fast symlinks will have i_size but no allocated clusters. */
6320 if (S_ISLNK(inode->i_mode) && !fe->i_clusters)
6321 @@ -1206,8 +1276,11 @@ int ocfs2_mark_inode_dirty(handle_t *han
6323 fe->i_size = cpu_to_le64(i_size_read(inode));
6324 fe->i_links_count = cpu_to_le16(inode->i_nlink);
6325 - fe->i_uid = cpu_to_le32(inode->i_uid);
6326 - fe->i_gid = cpu_to_le32(inode->i_gid);
6327 + fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode),
6328 + inode->i_uid, inode->i_tag));
6329 + fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode),
6330 + inode->i_gid, inode->i_tag));
6331 + /* i_tag = = cpu_to_le16(inode->i_tag); */
6332 fe->i_mode = cpu_to_le16(inode->i_mode);
6333 fe->i_atime = cpu_to_le64(inode->i_atime.tv_sec);
6334 fe->i_atime_nsec = cpu_to_le32(inode->i_atime.tv_nsec);
6335 @@ -1235,16 +1308,25 @@ leave:
6336 void ocfs2_refresh_inode(struct inode *inode,
6337 struct ocfs2_dinode *fe)
6342 spin_lock(&OCFS2_I(inode)->ip_lock);
6344 OCFS2_I(inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
6345 OCFS2_I(inode)->ip_attr = le32_to_cpu(fe->i_attr);
6346 + /* OCFS2_I(inode)->ip_flags &= ~OCFS2_FL_MASK;
6347 + OCFS2_I(inode)->ip_flags |= le32_to_cpu(fe->i_flags) & OCFS2_FL_MASK; */
6348 OCFS2_I(inode)->ip_dyn_features = le16_to_cpu(fe->i_dyn_features);
6349 ocfs2_set_inode_flags(inode);
6350 i_size_write(inode, le64_to_cpu(fe->i_size));
6351 inode->i_nlink = le16_to_cpu(fe->i_links_count);
6352 - inode->i_uid = le32_to_cpu(fe->i_uid);
6353 - inode->i_gid = le32_to_cpu(fe->i_gid);
6354 + uid = le32_to_cpu(fe->i_uid);
6355 + gid = le32_to_cpu(fe->i_gid);
6356 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6357 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6358 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6359 + /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6360 inode->i_mode = le16_to_cpu(fe->i_mode);
6361 if (S_ISLNK(inode->i_mode) && le32_to_cpu(fe->i_clusters) == 0)
6362 inode->i_blocks = 0;
6363 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/inode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/inode.h
6364 --- linux-2.6.29.1/fs/ocfs2/inode.h 2009-03-24 14:22:27.000000000 +0100
6365 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/inode.h 2009-03-24 14:48:35.000000000 +0100
6366 @@ -147,6 +147,7 @@ struct buffer_head *ocfs2_bread(struct i
6368 void ocfs2_set_inode_flags(struct inode *inode);
6369 void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi);
6370 +int ocfs2_sync_flags(struct inode *inode);
6372 static inline blkcnt_t ocfs2_inode_sector_count(struct inode *inode)
6374 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ioctl.c
6375 --- linux-2.6.29.1/fs/ocfs2/ioctl.c 2008-12-25 00:26:37.000000000 +0100
6376 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ioctl.c 2009-02-22 22:54:25.000000000 +0100
6377 @@ -42,7 +42,7 @@ static int ocfs2_get_inode_attr(struct i
6381 -static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6382 +int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6385 struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6386 @@ -67,6 +67,11 @@ static int ocfs2_set_inode_attr(struct i
6387 if (!S_ISDIR(inode->i_mode))
6388 flags &= ~OCFS2_DIRSYNC_FL;
6390 + if (IS_BARRIER(inode)) {
6391 + vxwprintk_task(1, "messing with the barrier.");
6395 handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6396 if (IS_ERR(handle)) {
6397 status = PTR_ERR(handle);
6398 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/ioctl.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ioctl.h
6399 --- linux-2.6.29.1/fs/ocfs2/ioctl.h 2008-12-25 00:26:37.000000000 +0100
6400 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ioctl.h 2009-02-22 22:54:25.000000000 +0100
6402 #ifndef OCFS2_IOCTL_H
6403 #define OCFS2_IOCTL_H
6405 +int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6408 long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
6409 long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg);
6411 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/namei.c
6412 --- linux-2.6.29.1/fs/ocfs2/namei.c 2009-03-24 14:22:27.000000000 +0100
6413 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/namei.c 2009-03-25 01:04:31.000000000 +0100
6415 #include <linux/slab.h>
6416 #include <linux/highmem.h>
6417 #include <linux/quotaops.h>
6418 +#include <linux/vs_tag.h>
6420 #define MLOG_MASK_PREFIX ML_NAMEI
6421 #include <cluster/masklog.h>
6422 @@ -462,6 +463,7 @@ static int ocfs2_mknod_locked(struct ocf
6423 struct ocfs2_extent_list *fel;
6428 mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry,
6429 inode->i_mode, (unsigned long)dev, dentry->d_name.len,
6430 @@ -508,8 +510,11 @@ static int ocfs2_mknod_locked(struct ocf
6431 fe->i_blkno = cpu_to_le64(fe_blkno);
6432 fe->i_suballoc_bit = cpu_to_le16(suballoc_bit);
6433 fe->i_suballoc_slot = cpu_to_le16(inode_ac->ac_alloc_slot);
6434 - fe->i_uid = cpu_to_le32(inode->i_uid);
6435 - fe->i_gid = cpu_to_le32(inode->i_gid);
6437 + tag = dx_current_fstag(osb->sb);
6438 + fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode), inode->i_uid, tag));
6439 + fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode), inode->i_gid, tag));
6440 + inode->i_tag = tag;
6441 fe->i_mode = cpu_to_le16(inode->i_mode);
6442 if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode))
6443 fe->id1.dev1.i_rdev = cpu_to_le64(huge_encode_dev(dev));
6444 @@ -2025,6 +2030,7 @@ const struct inode_operations ocfs2_dir_
6445 .rename = ocfs2_rename,
6446 .setattr = ocfs2_setattr,
6447 .getattr = ocfs2_getattr,
6448 + .sync_flags = ocfs2_sync_flags,
6449 .permission = ocfs2_permission,
6450 .setxattr = generic_setxattr,
6451 .getxattr = generic_getxattr,
6452 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/ocfs2_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ocfs2_fs.h
6453 --- linux-2.6.29.1/fs/ocfs2/ocfs2_fs.h 2009-03-24 14:22:27.000000000 +0100
6454 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ocfs2_fs.h 2009-03-24 14:48:35.000000000 +0100
6455 @@ -219,18 +219,23 @@
6456 #define OCFS2_INDEXED_DIR_FL (0x0008)
6458 /* Inode attributes, keep in sync with EXT2 */
6459 -#define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
6460 -#define OCFS2_UNRM_FL (0x00000002) /* Undelete */
6461 -#define OCFS2_COMPR_FL (0x00000004) /* Compress file */
6462 -#define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
6463 -#define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
6464 -#define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
6465 -#define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
6466 -#define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
6467 -#define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
6468 +#define OCFS2_SECRM_FL FS_SECRM_FL /* Secure deletion */
6469 +#define OCFS2_UNRM_FL FS_UNRM_FL /* Undelete */
6470 +#define OCFS2_COMPR_FL FS_COMPR_FL /* Compress file */
6471 +#define OCFS2_SYNC_FL FS_SYNC_FL /* Synchronous updates */
6472 +#define OCFS2_IMMUTABLE_FL FS_IMMUTABLE_FL /* Immutable file */
6473 +#define OCFS2_APPEND_FL FS_APPEND_FL /* writes to file may only append */
6474 +#define OCFS2_NODUMP_FL FS_NODUMP_FL /* do not dump file */
6475 +#define OCFS2_NOATIME_FL FS_NOATIME_FL /* do not update atime */
6477 -#define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
6478 -#define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
6479 +#define OCFS2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
6480 +#define OCFS2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
6482 +#define OCFS2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
6483 +#define OCFS2_COW_FL FS_COW_FL /* Copy on Write marker */
6485 +#define OCFS2_FL_VISIBLE (0x010300FF) /* User visible flags */
6486 +#define OCFS2_FL_MODIFIABLE (0x010300FF) /* User modifiable flags */
6489 * Extent record flags (e_node.leaf.flags)
6490 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/ocfs2.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ocfs2.h
6491 --- linux-2.6.29.1/fs/ocfs2/ocfs2.h 2009-03-24 14:22:27.000000000 +0100
6492 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/ocfs2.h 2009-03-24 16:18:22.000000000 +0100
6493 @@ -199,6 +199,7 @@ enum ocfs2_mount_options
6494 OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* POSIX access control lists */
6495 OCFS2_MOUNT_USRQUOTA = 1 << 9, /* We support user quotas */
6496 OCFS2_MOUNT_GRPQUOTA = 1 << 10, /* We support group quotas */
6497 + OCFS2_MOUNT_TAGGED = 1 << 11, /* use tagging */
6500 #define OCFS2_OSB_SOFT_RO 0x0001
6501 diff -NurpP --minimal linux-2.6.29.1/fs/ocfs2/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/super.c
6502 --- linux-2.6.29.1/fs/ocfs2/super.c 2009-03-24 14:22:27.000000000 +0100
6503 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/ocfs2/super.c 2009-03-24 16:19:44.000000000 +0100
6504 @@ -172,6 +172,7 @@ enum {
6508 + Opt_tag, Opt_notag, Opt_tagid,
6512 @@ -198,6 +199,9 @@ static const match_table_t tokens = {
6513 {Opt_noacl, "noacl"},
6514 {Opt_usrquota, "usrquota"},
6515 {Opt_grpquota, "grpquota"},
6517 + {Opt_notag, "notag"},
6518 + {Opt_tagid, "tagid=%u"},
6522 @@ -436,6 +440,13 @@ static int ocfs2_remount(struct super_bl
6526 + if ((parsed_options.mount_opt & OCFS2_MOUNT_TAGGED) &&
6527 + !(sb->s_flags & MS_TAGGED)) {
6529 + mlog(ML_ERROR, "Cannot change tagging on remount\n");
6533 if ((osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) !=
6534 (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
6536 @@ -950,6 +961,9 @@ static int ocfs2_fill_super(struct super
6538 ocfs2_complete_mount_recovery(osb);
6540 + if (osb->s_mount_opt & OCFS2_MOUNT_TAGGED)
6541 + sb->s_flags |= MS_TAGGED;
6543 if (ocfs2_mount_local(osb))
6544 snprintf(nodestr, sizeof(nodestr), "local");
6546 @@ -1208,6 +1222,20 @@ static int ocfs2_parse_options(struct su
6547 printk(KERN_INFO "ocfs2 (no)acl options not supported\n");
6550 +#ifndef CONFIG_TAGGING_NONE
6552 + mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6555 + mopt->mount_opt &= ~OCFS2_MOUNT_TAGGED;
6558 +#ifdef CONFIG_PROPAGATE
6561 + mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6566 "Unrecognized mount option \"%s\" "
6567 diff -NurpP --minimal linux-2.6.29.1/fs/open.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/open.c
6568 --- linux-2.6.29.1/fs/open.c 2009-03-24 14:22:27.000000000 +0100
6569 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/open.c 2009-03-24 14:48:35.000000000 +0100
6571 #include <linux/rcupdate.h>
6572 #include <linux/audit.h>
6573 #include <linux/falloc.h>
6574 +#include <linux/vs_base.h>
6575 +#include <linux/vs_limit.h>
6576 +#include <linux/vs_dlimit.h>
6577 +#include <linux/vs_tag.h>
6578 +#include <linux/vs_cowbl.h>
6580 int vfs_statfs(struct dentry *dentry, struct kstatfs *buf)
6582 int retval = -ENODEV;
6585 + struct super_block *sb = dentry->d_sb;
6588 - if (dentry->d_sb->s_op->statfs) {
6589 + if (sb->s_op->statfs) {
6590 memset(buf, 0, sizeof(*buf));
6591 retval = security_sb_statfs(dentry);
6594 - retval = dentry->d_sb->s_op->statfs(dentry, buf);
6595 + retval = sb->s_op->statfs(dentry, buf);
6596 if (retval == 0 && buf->f_frsize == 0)
6597 buf->f_frsize = buf->f_bsize;
6599 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
6600 + vx_vsi_statfs(sb, buf);
6604 @@ -638,6 +647,10 @@ SYSCALL_DEFINE3(fchmodat, int, dfd, cons
6605 error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
6609 + error = cow_check_and_break(&path);
6611 + goto dput_and_out;
6612 inode = path.dentry->d_inode;
6614 error = mnt_want_write(path.mnt);
6615 @@ -671,11 +684,11 @@ static int chown_common(struct dentry *
6616 newattrs.ia_valid = ATTR_CTIME;
6617 if (user != (uid_t) -1) {
6618 newattrs.ia_valid |= ATTR_UID;
6619 - newattrs.ia_uid = user;
6620 + newattrs.ia_uid = dx_map_uid(user);
6622 if (group != (gid_t) -1) {
6623 newattrs.ia_valid |= ATTR_GID;
6624 - newattrs.ia_gid = group;
6625 + newattrs.ia_gid = dx_map_gid(group);
6627 if (!S_ISDIR(inode->i_mode))
6628 newattrs.ia_valid |=
6629 @@ -698,7 +711,11 @@ SYSCALL_DEFINE3(chown, const char __user
6630 error = mnt_want_write(path.mnt);
6633 - error = chown_common(path.dentry, user, group);
6634 +#ifdef CONFIG_VSERVER_COWBL
6635 + error = cow_check_and_break(&path);
6638 + error = chown_common(path.dentry, user, group);
6639 mnt_drop_write(path.mnt);
6642 @@ -723,7 +740,11 @@ SYSCALL_DEFINE5(fchownat, int, dfd, cons
6643 error = mnt_want_write(path.mnt);
6646 - error = chown_common(path.dentry, user, group);
6647 +#ifdef CONFIG_VSERVER_COWBL
6648 + error = cow_check_and_break(&path);
6651 + error = chown_common(path.dentry, user, group);
6652 mnt_drop_write(path.mnt);
6655 @@ -742,7 +763,11 @@ SYSCALL_DEFINE3(lchown, const char __use
6656 error = mnt_want_write(path.mnt);
6659 - error = chown_common(path.dentry, user, group);
6660 +#ifdef CONFIG_VSERVER_COWBL
6661 + error = cow_check_and_break(&path);
6664 + error = chown_common(path.dentry, user, group);
6665 mnt_drop_write(path.mnt);
6668 @@ -986,6 +1011,7 @@ static void __put_unused_fd(struct files
6669 __FD_CLR(fd, fdt->open_fds);
6670 if (fd < files->next_fd)
6671 files->next_fd = fd;
6672 + vx_openfd_dec(fd);
6675 void put_unused_fd(unsigned int fd)
6676 diff -NurpP --minimal linux-2.6.29.1/fs/proc/array.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/array.c
6677 --- linux-2.6.29.1/fs/proc/array.c 2009-03-24 14:22:27.000000000 +0100
6678 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/array.c 2009-03-25 01:07:51.000000000 +0100
6680 #include <linux/seq_file.h>
6681 #include <linux/pid_namespace.h>
6682 #include <linux/tracehook.h>
6683 +#include <linux/vs_context.h>
6684 +#include <linux/vs_network.h>
6686 #include <asm/pgtable.h>
6687 #include <asm/processor.h>
6688 @@ -137,8 +139,9 @@ static const char *task_state_array[] =
6689 "D (disk sleep)", /* 2 */
6690 "T (stopped)", /* 4 */
6691 "T (tracing stop)", /* 8 */
6692 - "Z (zombie)", /* 16 */
6693 - "X (dead)" /* 32 */
6694 + "H (on hold)", /* 16 */
6695 + "Z (zombie)", /* 32 */
6696 + "X (dead)", /* 64 */
6699 static inline const char *get_task_state(struct task_struct *tsk)
6700 @@ -165,6 +168,9 @@ static inline void task_state(struct seq
6702 ppid = pid_alive(p) ?
6703 task_tgid_nr_ns(rcu_dereference(p->real_parent), ns) : 0;
6704 + if (unlikely(vx_current_initpid(p->pid)))
6709 struct task_struct *tracer = tracehook_tracer_task(p);
6710 @@ -280,7 +286,7 @@ static inline void task_sig(struct seq_f
6713 static void render_cap_t(struct seq_file *m, const char *header,
6715 + struct vx_info *vxi, kernel_cap_t *a)
6719 @@ -305,10 +311,11 @@ static inline void task_cap(struct seq_f
6720 cap_bset = cred->cap_bset;
6723 - render_cap_t(m, "CapInh:\t", &cap_inheritable);
6724 - render_cap_t(m, "CapPrm:\t", &cap_permitted);
6725 - render_cap_t(m, "CapEff:\t", &cap_effective);
6726 - render_cap_t(m, "CapBnd:\t", &cap_bset);
6727 + /* FIXME: maybe move the p->vx_info masking to __task_cred() ? */
6728 + render_cap_t(m, "CapInh:\t", p->vx_info, &cap_inheritable);
6729 + render_cap_t(m, "CapPrm:\t", p->vx_info, &cap_permitted);
6730 + render_cap_t(m, "CapEff:\t", p->vx_info, &cap_effective);
6731 + render_cap_t(m, "CapBnd:\t", p->vx_info, &cap_bset);
6734 static inline void task_context_switch_counts(struct seq_file *m,
6735 @@ -320,6 +327,42 @@ static inline void task_context_switch_c
6739 +int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
6740 + struct pid *pid, struct task_struct *task)
6742 + seq_printf(m, "Proxy:\t%p(%c)\n"
6750 + (task->nsproxy == init_task.nsproxy ? 'I' : '-'),
6751 + atomic_read(&task->nsproxy->count),
6752 + task->nsproxy->uts_ns,
6753 + (task->nsproxy->uts_ns == init_task.nsproxy->uts_ns ? 'I' : '-'),
6754 + task->nsproxy->ipc_ns,
6755 + (task->nsproxy->ipc_ns == init_task.nsproxy->ipc_ns ? 'I' : '-'),
6756 + task->nsproxy->mnt_ns,
6757 + (task->nsproxy->mnt_ns == init_task.nsproxy->mnt_ns ? 'I' : '-'),
6758 + task->nsproxy->pid_ns,
6759 + (task->nsproxy->pid_ns == init_task.nsproxy->pid_ns ? 'I' : '-'),
6760 + task->nsproxy->net_ns,
6761 + (task->nsproxy->net_ns == init_task.nsproxy->net_ns ? 'I' : '-'));
6765 +void task_vs_id(struct seq_file *m, struct task_struct *task)
6767 + if (task_vx_flags(task, VXF_HIDE_VINFO, 0))
6770 + seq_printf(m, "VxID: %d\n", vx_task_xid(task));
6771 + seq_printf(m, "NxID: %d\n", nx_task_nid(task));
6775 int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
6776 struct pid *pid, struct task_struct *task)
6778 @@ -335,6 +378,7 @@ int proc_pid_status(struct seq_file *m,
6781 cpuset_task_status_allowed(m, task);
6782 + task_vs_id(m, task);
6783 #if defined(CONFIG_S390)
6784 task_show_regs(m, task);
6786 @@ -447,6 +491,17 @@ static int do_task_stat(struct seq_file
6787 /* convert nsec -> ticks */
6788 start_time = nsec_to_clock_t(start_time);
6790 + /* fixup start time for virt uptime */
6791 + if (vx_flags(VXF_VIRT_UPTIME, 0)) {
6792 + unsigned long long bias =
6793 + current->vx_info->cvirt.bias_clock;
6795 + if (start_time > bias)
6796 + start_time -= bias;
6801 seq_printf(m, "%d (%s) %c %d %d %d %d %d %u %lu \
6802 %lu %lu %lu %lu %lu %ld %ld %ld %ld %d 0 %llu %lu %ld %lu %lu %lu %lu %lu \
6803 %lu %lu %lu %lu %lu %lu %lu %lu %d %d %u %u %llu %lu %ld\n",
6804 diff -NurpP --minimal linux-2.6.29.1/fs/proc/base.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/base.c
6805 --- linux-2.6.29.1/fs/proc/base.c 2009-03-24 14:22:27.000000000 +0100
6806 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/base.c 2009-03-24 16:29:35.000000000 +0100
6808 #include <linux/oom.h>
6809 #include <linux/elf.h>
6810 #include <linux/pid_namespace.h>
6811 +#include <linux/vs_context.h>
6812 +#include <linux/vs_network.h>
6813 #include "internal.h"
6816 @@ -1458,6 +1460,8 @@ static struct inode *proc_pid_make_inode
6817 inode->i_gid = cred->egid;
6820 + /* procfs is xid tagged */
6821 + inode->i_tag = (tag_t)vx_task_xid(task);
6822 security_task_to_inode(task, inode);
6825 @@ -2008,6 +2012,13 @@ static struct dentry *proc_pident_lookup
6829 + /* TODO: maybe we can come up with a generic approach? */
6830 + if (task_vx_flags(task, VXF_HIDE_VINFO, 0) &&
6831 + (dentry->d_name.len == 5) &&
6832 + (!memcmp(dentry->d_name.name, "vinfo", 5) ||
6833 + !memcmp(dentry->d_name.name, "ninfo", 5)))
6837 * Yes, it does not scale. And it should not. Don't add
6838 * new entries into /proc/<tgid>/ without very good reasons.
6839 @@ -2393,7 +2404,7 @@ out_iput:
6840 static struct dentry *proc_base_lookup(struct inode *dir, struct dentry *dentry)
6842 struct dentry *error;
6843 - struct task_struct *task = get_proc_task(dir);
6844 + struct task_struct *task = get_proc_task_real(dir);
6845 const struct pid_entry *p, *last;
6847 error = ERR_PTR(-ENOENT);
6848 @@ -2483,6 +2494,9 @@ static int proc_pid_personality(struct s
6849 static const struct file_operations proc_task_operations;
6850 static const struct inode_operations proc_task_inode_operations;
6852 +extern int proc_pid_vx_info(struct task_struct *, char *);
6853 +extern int proc_pid_nx_info(struct task_struct *, char *);
6855 static const struct pid_entry tgid_base_stuff[] = {
6856 DIR("task", S_IRUGO|S_IXUGO, proc_task_inode_operations, proc_task_operations),
6857 DIR("fd", S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
6858 @@ -2541,6 +2555,8 @@ static const struct pid_entry tgid_base_
6859 #ifdef CONFIG_CGROUPS
6860 REG("cgroup", S_IRUGO, proc_cgroup_operations),
6862 + INF("vinfo", S_IRUGO, proc_pid_vx_info),
6863 + INF("ninfo", S_IRUGO, proc_pid_nx_info),
6864 INF("oom_score", S_IRUGO, proc_oom_score),
6865 REG("oom_adj", S_IRUGO|S_IWUSR, proc_oom_adjust_operations),
6866 #ifdef CONFIG_AUDITSYSCALL
6867 @@ -2556,6 +2572,7 @@ static const struct pid_entry tgid_base_
6868 #ifdef CONFIG_TASK_IO_ACCOUNTING
6869 INF("io", S_IRUGO, proc_tgid_io_accounting),
6871 + ONE("nsproxy", S_IRUGO, proc_pid_nsproxy),
6874 static int proc_tgid_base_readdir(struct file * filp,
6875 @@ -2752,7 +2769,7 @@ retry:
6877 pid = find_ge_pid(iter.tgid, ns);
6879 - iter.tgid = pid_nr_ns(pid, ns);
6880 + iter.tgid = pid_unmapped_nr_ns(pid, ns);
6881 iter.task = pid_task(pid, PIDTYPE_PID);
6882 /* What we to know is if the pid we have find is the
6883 * pid of a thread_group_leader. Testing for task
6884 @@ -2782,7 +2799,7 @@ static int proc_pid_fill_cache(struct fi
6885 struct tgid_iter iter)
6887 char name[PROC_NUMBUF];
6888 - int len = snprintf(name, sizeof(name), "%d", iter.tgid);
6889 + int len = snprintf(name, sizeof(name), "%d", vx_map_tgid(iter.tgid));
6890 return proc_fill_cache(filp, dirent, filldir, name, len,
6891 proc_pid_instantiate, iter.task, NULL);
6893 @@ -2791,7 +2808,7 @@ static int proc_pid_fill_cache(struct fi
6894 int proc_pid_readdir(struct file * filp, void * dirent, filldir_t filldir)
6896 unsigned int nr = filp->f_pos - FIRST_PROCESS_ENTRY;
6897 - struct task_struct *reaper = get_proc_task(filp->f_path.dentry->d_inode);
6898 + struct task_struct *reaper = get_proc_task_real(filp->f_path.dentry->d_inode);
6899 struct tgid_iter iter;
6900 struct pid_namespace *ns;
6902 @@ -2811,6 +2828,8 @@ int proc_pid_readdir(struct file * filp,
6904 iter.tgid += 1, iter = next_tgid(ns, iter)) {
6905 filp->f_pos = iter.tgid + TGID_OFFSET;
6906 + if (!vx_proc_task_visible(iter.task))
6908 if (proc_pid_fill_cache(filp, dirent, filldir, iter) < 0) {
6909 put_task_struct(iter.task);
6911 @@ -2957,6 +2976,8 @@ static struct dentry *proc_task_lookup(s
6912 tid = name_to_int(dentry);
6915 + if (vx_current_initpid(tid))
6918 ns = dentry->d_sb->s_fs_info;
6920 diff -NurpP --minimal linux-2.6.29.1/fs/proc/generic.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/generic.c
6921 --- linux-2.6.29.1/fs/proc/generic.c 2009-03-24 14:22:27.000000000 +0100
6922 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/generic.c 2009-03-24 14:48:35.000000000 +0100
6924 #include <linux/bitops.h>
6925 #include <linux/spinlock.h>
6926 #include <linux/completion.h>
6927 +#include <linux/vserver/inode.h>
6928 #include <asm/uaccess.h>
6930 #include "internal.h"
6931 @@ -382,6 +383,8 @@ struct dentry *proc_lookup_de(struct pro
6932 for (de = de->subdir; de ; de = de->next) {
6933 if (de->namelen != dentry->d_name.len)
6935 + if (!vx_hide_check(0, de->vx_flags))
6937 if (!memcmp(dentry->d_name.name, de->name, de->namelen)) {
6940 @@ -390,6 +393,8 @@ struct dentry *proc_lookup_de(struct pro
6941 spin_unlock(&proc_subdir_lock);
6943 inode = proc_get_inode(dir->i_sb, ino, de);
6944 + /* generic proc entries belong to the host */
6949 @@ -467,6 +472,8 @@ int proc_readdir_de(struct proc_dir_entr
6951 /* filldir passes info to user space */
6953 + if (!vx_hide_check(0, de->vx_flags))
6955 spin_unlock(&proc_subdir_lock);
6956 if (filldir(dirent, de->name, de->namelen, filp->f_pos,
6957 de->low_ino, de->mode >> 12) < 0) {
6958 @@ -474,6 +481,7 @@ int proc_readdir_de(struct proc_dir_entr
6961 spin_lock(&proc_subdir_lock);
6966 @@ -588,6 +596,7 @@ static struct proc_dir_entry *__proc_cre
6968 atomic_set(&ent->count, 1);
6970 + ent->vx_flags = IATTR_PROC_DEFAULT;
6971 spin_lock_init(&ent->pde_unload_lock);
6972 ent->pde_unload_completion = NULL;
6973 INIT_LIST_HEAD(&ent->pde_openers);
6974 @@ -611,7 +620,8 @@ struct proc_dir_entry *proc_symlink(cons
6980 + ent->vx_flags = IATTR_PROC_SYMLINK;
6984 diff -NurpP --minimal linux-2.6.29.1/fs/proc/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/inode.c
6985 --- linux-2.6.29.1/fs/proc/inode.c 2009-03-24 14:22:27.000000000 +0100
6986 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/inode.c 2009-03-24 14:48:35.000000000 +0100
6987 @@ -465,6 +465,8 @@ struct inode *proc_get_inode(struct supe
6988 inode->i_uid = de->uid;
6989 inode->i_gid = de->gid;
6992 + PROC_I(inode)->vx_flags = de->vx_flags;
6994 inode->i_size = de->size;
6996 diff -NurpP --minimal linux-2.6.29.1/fs/proc/internal.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/internal.h
6997 --- linux-2.6.29.1/fs/proc/internal.h 2009-03-24 14:22:27.000000000 +0100
6998 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/internal.h 2009-03-24 14:48:35.000000000 +0100
7002 #include <linux/proc_fs.h>
7003 +#include <linux/vs_pid.h>
7005 extern struct proc_dir_entry proc_root;
7006 #ifdef CONFIG_PROC_SYSCTL
7007 @@ -51,6 +52,9 @@ extern int proc_pid_status(struct seq_fi
7008 struct pid *pid, struct task_struct *task);
7009 extern int proc_pid_statm(struct seq_file *m, struct pid_namespace *ns,
7010 struct pid *pid, struct task_struct *task);
7011 +extern int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
7012 + struct pid *pid, struct task_struct *task);
7014 extern loff_t mem_lseek(struct file *file, loff_t offset, int orig);
7016 extern const struct file_operations proc_maps_operations;
7017 @@ -70,11 +74,16 @@ static inline struct pid *proc_pid(struc
7018 return PROC_I(inode)->pid;
7021 -static inline struct task_struct *get_proc_task(struct inode *inode)
7022 +static inline struct task_struct *get_proc_task_real(struct inode *inode)
7024 return get_pid_task(proc_pid(inode), PIDTYPE_PID);
7027 +static inline struct task_struct *get_proc_task(struct inode *inode)
7029 + return vx_get_proc_task(inode, proc_pid(inode));
7032 static inline int proc_fd(struct inode *inode)
7034 return PROC_I(inode)->fd;
7035 diff -NurpP --minimal linux-2.6.29.1/fs/proc/meminfo.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/meminfo.c
7036 --- linux-2.6.29.1/fs/proc/meminfo.c 2009-03-24 14:22:27.000000000 +0100
7037 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/meminfo.c 2009-03-24 14:48:35.000000000 +0100
7038 @@ -41,7 +41,7 @@ static int meminfo_proc_show(struct seq_
7040 cached = global_page_state(NR_FILE_PAGES) -
7041 total_swapcache_pages - i.bufferram;
7043 + if (cached < 0 || vx_flags(VXF_VIRT_MEM, 0))
7046 get_vmalloc_info(&vmi);
7047 diff -NurpP --minimal linux-2.6.29.1/fs/proc/root.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/root.c
7048 --- linux-2.6.29.1/fs/proc/root.c 2009-03-24 14:22:27.000000000 +0100
7049 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/root.c 2009-03-24 14:48:35.000000000 +0100
7051 #include <linux/bitops.h>
7052 #include <linux/mount.h>
7053 #include <linux/pid_namespace.h>
7054 +#include <linux/vserver/inode.h>
7056 #include "internal.h"
7058 +struct proc_dir_entry *proc_virtual;
7060 +extern void proc_vx_init(void);
7062 static int proc_test_super(struct super_block *sb, void *data)
7064 return sb->s_fs_info == data;
7065 @@ -136,6 +141,7 @@ void __init proc_root_init(void)
7067 proc_mkdir("bus", NULL);
7072 static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat
7073 @@ -203,6 +209,7 @@ struct proc_dir_entry proc_root = {
7074 .proc_iops = &proc_root_inode_operations,
7075 .proc_fops = &proc_root_operations,
7076 .parent = &proc_root,
7077 + .vx_flags = IATTR_ADMIN | IATTR_WATCH,
7080 int pid_ns_prepare_proc(struct pid_namespace *ns)
7081 diff -NurpP --minimal linux-2.6.29.1/fs/proc/uptime.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/uptime.c
7082 --- linux-2.6.29.1/fs/proc/uptime.c 2008-12-25 00:26:37.000000000 +0100
7083 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/proc/uptime.c 2009-02-22 22:54:25.000000000 +0100
7085 #include <linux/proc_fs.h>
7086 #include <linux/sched.h>
7087 #include <linux/time.h>
7088 +#include <linux/vserver/cvirt.h>
7089 #include <asm/cputime.h>
7091 static int proc_calc_metrics(char *page, char **start, off_t off,
7092 @@ -29,6 +30,10 @@ static int uptime_read_proc(char *page,
7093 do_posix_clock_monotonic_gettime(&uptime);
7094 monotonic_to_bootbased(&uptime);
7095 cputime_to_timespec(idletime, &idle);
7097 + if (vx_flags(VXF_VIRT_UPTIME, 0))
7098 + vx_vsi_uptime(&uptime, &idle);
7100 len = sprintf(page, "%lu.%02lu %lu.%02lu\n",
7101 (unsigned long) uptime.tv_sec,
7102 (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
7103 diff -NurpP --minimal linux-2.6.29.1/fs/quota.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/quota.c
7104 --- linux-2.6.29.1/fs/quota.c 2009-03-24 14:22:27.000000000 +0100
7105 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/quota.c 2009-03-24 14:48:35.000000000 +0100
7107 #include <linux/capability.h>
7108 #include <linux/quotaops.h>
7109 #include <linux/types.h>
7110 +#include <linux/vs_context.h>
7112 /* Check validity of generic quotactl commands */
7113 static int generic_quotactl_valid(struct super_block *sb, int type, int cmd, qid_t id)
7114 @@ -81,11 +82,11 @@ static int generic_quotactl_valid(struct
7115 if (cmd == Q_GETQUOTA) {
7116 if (((type == USRQUOTA && current_euid() != id) ||
7117 (type == GRPQUOTA && !in_egroup_p(id))) &&
7118 - !capable(CAP_SYS_ADMIN))
7119 + !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7122 else if (cmd != Q_GETFMT && cmd != Q_SYNC && cmd != Q_GETINFO)
7123 - if (!capable(CAP_SYS_ADMIN))
7124 + if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7128 @@ -132,10 +133,10 @@ static int xqm_quotactl_valid(struct sup
7129 if (cmd == Q_XGETQUOTA) {
7130 if (((type == XQM_USRQUOTA && current_euid() != id) ||
7131 (type == XQM_GRPQUOTA && !in_egroup_p(id))) &&
7132 - !capable(CAP_SYS_ADMIN))
7133 + !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7135 } else if (cmd != Q_XGETQSTAT && cmd != Q_XQUOTASYNC) {
7136 - if (!capable(CAP_SYS_ADMIN))
7137 + if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7141 @@ -337,6 +338,46 @@ static int do_quotactl(struct super_bloc
7145 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7147 +#include <linux/vroot.h>
7148 +#include <linux/major.h>
7149 +#include <linux/module.h>
7150 +#include <linux/kallsyms.h>
7151 +#include <linux/vserver/debug.h>
7153 +static vroot_grb_func *vroot_get_real_bdev = NULL;
7155 +static spinlock_t vroot_grb_lock = SPIN_LOCK_UNLOCKED;
7157 +int register_vroot_grb(vroot_grb_func *func) {
7160 + spin_lock(&vroot_grb_lock);
7161 + if (!vroot_get_real_bdev) {
7162 + vroot_get_real_bdev = func;
7165 + spin_unlock(&vroot_grb_lock);
7168 +EXPORT_SYMBOL(register_vroot_grb);
7170 +int unregister_vroot_grb(vroot_grb_func *func) {
7171 + int ret = -EINVAL;
7173 + spin_lock(&vroot_grb_lock);
7174 + if (vroot_get_real_bdev) {
7175 + vroot_get_real_bdev = NULL;
7178 + spin_unlock(&vroot_grb_lock);
7181 +EXPORT_SYMBOL(unregister_vroot_grb);
7186 * look up a superblock on which quota ops will be performed
7187 * - use the name of a block device to find the superblock thereon
7188 @@ -354,6 +395,22 @@ static inline struct super_block *quotac
7191 return ERR_CAST(bdev);
7192 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7193 + if (bdev && bdev->bd_inode &&
7194 + imajor(bdev->bd_inode) == VROOT_MAJOR) {
7195 + struct block_device *bdnew = (void *)-EINVAL;
7197 + if (vroot_get_real_bdev)
7198 + bdnew = vroot_get_real_bdev(bdev);
7200 + vxdprintk(VXD_CBIT(misc, 0),
7201 + "vroot_get_real_bdev not set");
7203 + if (IS_ERR(bdnew))
7204 + return ERR_PTR(PTR_ERR(bdnew));
7208 sb = get_super(bdev);
7211 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/bitmap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/bitmap.c
7212 --- linux-2.6.29.1/fs/reiserfs/bitmap.c 2008-12-25 00:26:37.000000000 +0100
7213 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/bitmap.c 2009-02-22 22:54:25.000000000 +0100
7215 #include <linux/reiserfs_fs_sb.h>
7216 #include <linux/reiserfs_fs_i.h>
7217 #include <linux/quotaops.h>
7218 +#include <linux/vs_dlimit.h>
7220 #define PREALLOCATION_SIZE 9
7222 @@ -429,8 +430,10 @@ static void _reiserfs_free_block(struct
7223 set_sb_free_blocks(rs, sb_free_blocks(rs) + 1);
7225 journal_mark_dirty(th, s, sbh);
7226 - if (for_unformatted)
7227 + if (for_unformatted) {
7228 + DLIMIT_FREE_BLOCK(inode, 1);
7229 DQUOT_FREE_BLOCK_NODIRTY(inode, 1);
7233 void reiserfs_free_block(struct reiserfs_transaction_handle *th,
7234 @@ -1045,6 +1048,7 @@ static inline int blocknrs_and_prealloc_
7235 b_blocknr_t finish = SB_BLOCK_COUNT(s) - 1;
7237 int nr_allocated = 0;
7240 determine_prealloc_size(hint);
7241 if (!hint->formatted_node) {
7242 @@ -1054,19 +1058,30 @@ static inline int blocknrs_and_prealloc_
7243 "reiserquota: allocating %d blocks id=%u",
7244 amount_needed, hint->inode->i_uid);
7247 - DQUOT_ALLOC_BLOCK_NODIRTY(hint->inode, amount_needed);
7248 - if (quota_ret) /* Quota exceeded? */
7249 + quota_ret = DQUOT_ALLOC_BLOCK_NODIRTY(hint->inode,
7252 return QUOTA_EXCEEDED;
7253 + if (DLIMIT_ALLOC_BLOCK(hint->inode, amount_needed)) {
7254 + DQUOT_FREE_BLOCK_NODIRTY(hint->inode,
7256 + return NO_DISK_SPACE;
7259 if (hint->preallocate && hint->prealloc_size) {
7260 #ifdef REISERQUOTA_DEBUG
7261 reiserfs_debug(s, REISERFS_DEBUG_CODE,
7262 "reiserquota: allocating (prealloc) %d blocks id=%u",
7263 hint->prealloc_size, hint->inode->i_uid);
7266 - DQUOT_PREALLOC_BLOCK_NODIRTY(hint->inode,
7267 - hint->prealloc_size);
7268 + quota_ret = DQUOT_PREALLOC_BLOCK_NODIRTY(hint->inode,
7269 + hint->prealloc_size);
7271 + DLIMIT_ALLOC_BLOCK(hint->inode, hint->prealloc_size)) {
7272 + DQUOT_FREE_BLOCK_NODIRTY(hint->inode,
7273 + hint->prealloc_size);
7277 hint->preallocate = hint->prealloc_size = 0;
7279 @@ -1098,7 +1113,10 @@ static inline int blocknrs_and_prealloc_
7281 hint->inode->i_uid);
7283 - DQUOT_FREE_BLOCK_NODIRTY(hint->inode, amount_needed + hint->prealloc_size - nr_allocated); /* Free not allocated blocks */
7284 + /* Free not allocated blocks */
7285 + blocks = amount_needed + hint->prealloc_size - nr_allocated;
7286 + DLIMIT_FREE_BLOCK(hint->inode, blocks);
7287 + DQUOT_FREE_BLOCK_NODIRTY(hint->inode, blocks);
7289 while (nr_allocated--)
7290 reiserfs_free_block(hint->th, hint->inode,
7291 @@ -1129,10 +1147,10 @@ static inline int blocknrs_and_prealloc_
7292 REISERFS_I(hint->inode)->i_prealloc_count,
7293 hint->inode->i_uid);
7295 - DQUOT_FREE_BLOCK_NODIRTY(hint->inode, amount_needed +
7296 - hint->prealloc_size - nr_allocated -
7297 - REISERFS_I(hint->inode)->
7298 - i_prealloc_count);
7299 + blocks = amount_needed + hint->prealloc_size - nr_allocated -
7300 + REISERFS_I(hint->inode)->i_prealloc_count;
7301 + DLIMIT_FREE_BLOCK(hint->inode, blocks);
7302 + DQUOT_FREE_BLOCK_NODIRTY(hint->inode, blocks);
7306 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/file.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/file.c
7307 --- linux-2.6.29.1/fs/reiserfs/file.c 2008-12-25 00:26:37.000000000 +0100
7308 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/file.c 2009-02-22 22:54:25.000000000 +0100
7309 @@ -307,4 +307,5 @@ const struct inode_operations reiserfs_f
7310 .listxattr = reiserfs_listxattr,
7311 .removexattr = reiserfs_removexattr,
7312 .permission = reiserfs_permission,
7313 + .sync_flags = reiserfs_sync_flags,
7315 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/inode.c
7316 --- linux-2.6.29.1/fs/reiserfs/inode.c 2009-03-24 14:22:27.000000000 +0100
7317 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/inode.c 2009-03-24 14:48:35.000000000 +0100
7319 #include <linux/writeback.h>
7320 #include <linux/quotaops.h>
7321 #include <linux/swap.h>
7322 +#include <linux/vs_dlimit.h>
7323 +#include <linux/vs_tag.h>
7325 int reiserfs_commit_write(struct file *f, struct page *page,
7326 unsigned from, unsigned to);
7327 @@ -54,6 +56,7 @@ void reiserfs_delete_inode(struct inode
7328 * stat data deletion */
7330 DQUOT_FREE_INODE(inode);
7331 + DLIMIT_FREE_INODE(inode);
7333 if (journal_end(&th, inode->i_sb, jbegin_count))
7335 @@ -1116,6 +1119,8 @@ static void init_inode(struct inode *ino
7336 struct buffer_head *bh;
7337 struct item_head *ih;
7341 //int version = ITEM_VERSION_1;
7343 bh = PATH_PLAST_BUFFER(path);
7344 @@ -1139,12 +1144,13 @@ static void init_inode(struct inode *ino
7345 (struct stat_data_v1 *)B_I_PITEM(bh, ih);
7346 unsigned long blocks;
7348 + uid = sd_v1_uid(sd);
7349 + gid = sd_v1_gid(sd);
7351 set_inode_item_key_version(inode, KEY_FORMAT_3_5);
7352 set_inode_sd_version(inode, STAT_DATA_V1);
7353 inode->i_mode = sd_v1_mode(sd);
7354 inode->i_nlink = sd_v1_nlink(sd);
7355 - inode->i_uid = sd_v1_uid(sd);
7356 - inode->i_gid = sd_v1_gid(sd);
7357 inode->i_size = sd_v1_size(sd);
7358 inode->i_atime.tv_sec = sd_v1_atime(sd);
7359 inode->i_mtime.tv_sec = sd_v1_mtime(sd);
7360 @@ -1186,11 +1192,12 @@ static void init_inode(struct inode *ino
7361 // (directories and symlinks)
7362 struct stat_data *sd = (struct stat_data *)B_I_PITEM(bh, ih);
7364 + uid = sd_v2_uid(sd);
7365 + gid = sd_v2_gid(sd);
7367 inode->i_mode = sd_v2_mode(sd);
7368 inode->i_nlink = sd_v2_nlink(sd);
7369 - inode->i_uid = sd_v2_uid(sd);
7370 inode->i_size = sd_v2_size(sd);
7371 - inode->i_gid = sd_v2_gid(sd);
7372 inode->i_mtime.tv_sec = sd_v2_mtime(sd);
7373 inode->i_atime.tv_sec = sd_v2_atime(sd);
7374 inode->i_ctime.tv_sec = sd_v2_ctime(sd);
7375 @@ -1220,6 +1227,10 @@ static void init_inode(struct inode *ino
7376 sd_attrs_to_i_attrs(sd_v2_attrs(sd), inode);
7379 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
7380 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
7381 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
7384 if (S_ISREG(inode->i_mode)) {
7385 inode->i_op = &reiserfs_file_inode_operations;
7386 @@ -1242,13 +1253,15 @@ static void init_inode(struct inode *ino
7387 static void inode2sd(void *sd, struct inode *inode, loff_t size)
7389 struct stat_data *sd_v2 = (struct stat_data *)sd;
7390 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
7391 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
7394 + set_sd_v2_uid(sd_v2, uid);
7395 + set_sd_v2_gid(sd_v2, gid);
7396 set_sd_v2_mode(sd_v2, inode->i_mode);
7397 set_sd_v2_nlink(sd_v2, inode->i_nlink);
7398 - set_sd_v2_uid(sd_v2, inode->i_uid);
7399 set_sd_v2_size(sd_v2, size);
7400 - set_sd_v2_gid(sd_v2, inode->i_gid);
7401 set_sd_v2_mtime(sd_v2, inode->i_mtime.tv_sec);
7402 set_sd_v2_atime(sd_v2, inode->i_atime.tv_sec);
7403 set_sd_v2_ctime(sd_v2, inode->i_ctime.tv_sec);
7404 @@ -1763,6 +1776,10 @@ int reiserfs_new_inode(struct reiserfs_t
7406 BUG_ON(!th->t_trans_id);
7408 + if (DLIMIT_ALLOC_INODE(inode)) {
7410 + goto out_bad_dlimit;
7412 if (DQUOT_ALLOC_INODE(inode)) {
7415 @@ -1950,6 +1967,9 @@ int reiserfs_new_inode(struct reiserfs_t
7416 DQUOT_FREE_INODE(inode);
7419 + DLIMIT_FREE_INODE(inode);
7422 journal_end(th, th->t_super, th->t_blocks_allocated);
7423 /* Drop can be outside and it needs more credits so it's better to have it outside */
7425 @@ -2837,14 +2857,19 @@ int reiserfs_commit_write(struct file *f
7426 void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode)
7428 if (reiserfs_attrs(inode->i_sb)) {
7429 - if (sd_attrs & REISERFS_SYNC_FL)
7430 - inode->i_flags |= S_SYNC;
7432 - inode->i_flags &= ~S_SYNC;
7433 if (sd_attrs & REISERFS_IMMUTABLE_FL)
7434 inode->i_flags |= S_IMMUTABLE;
7436 inode->i_flags &= ~S_IMMUTABLE;
7437 + if (sd_attrs & REISERFS_IXUNLINK_FL)
7438 + inode->i_flags |= S_IXUNLINK;
7440 + inode->i_flags &= ~S_IXUNLINK;
7442 + if (sd_attrs & REISERFS_SYNC_FL)
7443 + inode->i_flags |= S_SYNC;
7445 + inode->i_flags &= ~S_SYNC;
7446 if (sd_attrs & REISERFS_APPEND_FL)
7447 inode->i_flags |= S_APPEND;
7449 @@ -2857,6 +2882,15 @@ void sd_attrs_to_i_attrs(__u16 sd_attrs,
7450 REISERFS_I(inode)->i_flags |= i_nopack_mask;
7452 REISERFS_I(inode)->i_flags &= ~i_nopack_mask;
7454 + if (sd_attrs & REISERFS_BARRIER_FL)
7455 + inode->i_vflags |= V_BARRIER;
7457 + inode->i_vflags &= ~V_BARRIER;
7458 + if (sd_attrs & REISERFS_COW_FL)
7459 + inode->i_vflags |= V_COW;
7461 + inode->i_vflags &= ~V_COW;
7465 @@ -2867,6 +2901,11 @@ void i_attrs_to_sd_attrs(struct inode *i
7466 *sd_attrs |= REISERFS_IMMUTABLE_FL;
7468 *sd_attrs &= ~REISERFS_IMMUTABLE_FL;
7469 + if (inode->i_flags & S_IXUNLINK)
7470 + *sd_attrs |= REISERFS_IXUNLINK_FL;
7472 + *sd_attrs &= ~REISERFS_IXUNLINK_FL;
7474 if (inode->i_flags & S_SYNC)
7475 *sd_attrs |= REISERFS_SYNC_FL;
7477 @@ -2879,6 +2918,15 @@ void i_attrs_to_sd_attrs(struct inode *i
7478 *sd_attrs |= REISERFS_NOTAIL_FL;
7480 *sd_attrs &= ~REISERFS_NOTAIL_FL;
7482 + if (inode->i_vflags & V_BARRIER)
7483 + *sd_attrs |= REISERFS_BARRIER_FL;
7485 + *sd_attrs &= ~REISERFS_BARRIER_FL;
7486 + if (inode->i_vflags & V_COW)
7487 + *sd_attrs |= REISERFS_COW_FL;
7489 + *sd_attrs &= ~REISERFS_COW_FL;
7493 @@ -3046,6 +3094,22 @@ static ssize_t reiserfs_direct_IO(int rw
7494 reiserfs_get_blocks_direct_io, NULL);
7497 +int reiserfs_sync_flags(struct inode *inode)
7499 + u16 oldflags, newflags;
7501 + oldflags = REISERFS_I(inode)->i_attrs;
7502 + newflags = oldflags;
7503 + i_attrs_to_sd_attrs(inode, &newflags);
7505 + if (oldflags ^ newflags) {
7506 + REISERFS_I(inode)->i_attrs = newflags;
7507 + inode->i_ctime = CURRENT_TIME_SEC;
7508 + mark_inode_dirty(inode);
7513 int reiserfs_setattr(struct dentry *dentry, struct iattr *attr)
7515 struct inode *inode = dentry->d_inode;
7516 @@ -3099,9 +3163,11 @@ int reiserfs_setattr(struct dentry *dent
7519 error = inode_change_ok(inode, attr);
7522 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
7523 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
7524 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
7525 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
7526 error = reiserfs_chown_xattrs(inode, attr);
7529 @@ -3131,6 +3197,9 @@ int reiserfs_setattr(struct dentry *dent
7530 inode->i_uid = attr->ia_uid;
7531 if (attr->ia_valid & ATTR_GID)
7532 inode->i_gid = attr->ia_gid;
7533 + if ((attr->ia_valid & ATTR_TAG) &&
7535 + inode->i_tag = attr->ia_tag;
7536 mark_inode_dirty(inode);
7538 journal_end(&th, inode->i_sb, jbegin_count);
7539 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/ioctl.c
7540 --- linux-2.6.29.1/fs/reiserfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
7541 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/ioctl.c 2009-02-22 22:54:25.000000000 +0100
7543 #include <linux/mount.h>
7544 #include <linux/reiserfs_fs.h>
7545 #include <linux/time.h>
7546 +#include <linux/mount.h>
7547 #include <asm/uaccess.h>
7548 #include <linux/pagemap.h>
7549 #include <linux/smp_lock.h>
7551 int reiserfs_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
7554 - unsigned int flags;
7555 + unsigned int flags, oldflags;
7559 @@ -43,6 +44,7 @@ int reiserfs_ioctl(struct inode *inode,
7561 flags = REISERFS_I(inode)->i_attrs;
7562 i_attrs_to_sd_attrs(inode, (__u16 *) & flags);
7563 + flags &= REISERFS_FL_USER_VISIBLE;
7564 return put_user(flags, (int __user *)arg);
7565 case REISERFS_IOC_SETFLAGS:{
7566 if (!reiserfs_attrs(inode->i_sb))
7567 @@ -60,6 +62,10 @@ int reiserfs_ioctl(struct inode *inode,
7571 + if (IS_BARRIER(inode)) {
7572 + vxwprintk_task(1, "messing with the barrier.");
7576 * Is it quota file? Do not allow user to mess with it
7578 @@ -84,6 +90,10 @@ int reiserfs_ioctl(struct inode *inode,
7583 + oldflags = REISERFS_I(inode)->i_attrs;
7584 + flags = flags & REISERFS_FL_USER_MODIFIABLE;
7585 + flags |= oldflags & ~REISERFS_FL_USER_MODIFIABLE;
7586 sd_attrs_to_i_attrs(flags, inode);
7587 REISERFS_I(inode)->i_attrs = flags;
7588 inode->i_ctime = CURRENT_TIME_SEC;
7589 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/namei.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/namei.c
7590 --- linux-2.6.29.1/fs/reiserfs/namei.c 2009-03-24 14:22:27.000000000 +0100
7591 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/namei.c 2009-03-24 14:48:35.000000000 +0100
7593 #include <linux/reiserfs_acl.h>
7594 #include <linux/reiserfs_xattr.h>
7595 #include <linux/quotaops.h>
7596 +#include <linux/vs_tag.h>
7598 #define INC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) { inc_nlink(i); if (i->i_nlink >= REISERFS_LINK_MAX) i->i_nlink=1; }
7599 #define DEC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) drop_nlink(i);
7600 @@ -360,6 +361,7 @@ static struct dentry *reiserfs_lookup(st
7601 reiserfs_write_unlock(dir->i_sb);
7602 return ERR_PTR(-EACCES);
7604 + dx_propagate_tag(nd, inode);
7606 /* Propogate the priv_object flag so we know we're in the priv tree */
7607 if (is_reiserfs_priv_object(dir))
7608 @@ -586,6 +588,7 @@ static int new_inode_init(struct inode *
7610 inode->i_gid = current_fsgid();
7612 + inode->i_tag = dx_current_fstag(inode->i_sb);
7616 @@ -1540,6 +1543,7 @@ const struct inode_operations reiserfs_d
7617 .listxattr = reiserfs_listxattr,
7618 .removexattr = reiserfs_removexattr,
7619 .permission = reiserfs_permission,
7620 + .sync_flags = reiserfs_sync_flags,
7624 @@ -1556,6 +1560,7 @@ const struct inode_operations reiserfs_s
7625 .listxattr = reiserfs_listxattr,
7626 .removexattr = reiserfs_removexattr,
7627 .permission = reiserfs_permission,
7628 + .sync_flags = reiserfs_sync_flags,
7632 @@ -1569,5 +1574,6 @@ const struct inode_operations reiserfs_s
7633 .listxattr = reiserfs_listxattr,
7634 .removexattr = reiserfs_removexattr,
7635 .permission = reiserfs_permission,
7636 + .sync_flags = reiserfs_sync_flags,
7639 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/stree.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/stree.c
7640 --- linux-2.6.29.1/fs/reiserfs/stree.c 2008-12-25 00:26:37.000000000 +0100
7641 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/stree.c 2009-02-22 22:54:25.000000000 +0100
7643 #include <linux/reiserfs_fs.h>
7644 #include <linux/buffer_head.h>
7645 #include <linux/quotaops.h>
7646 +#include <linux/vs_dlimit.h>
7648 /* Does the buffer contain a disk block which is in the tree. */
7649 inline int B_IS_IN_TREE(const struct buffer_head *p_s_bh)
7650 @@ -1297,6 +1298,7 @@ int reiserfs_delete_item(struct reiserfs
7651 "reiserquota delete_item(): freeing %u, id=%u type=%c",
7652 quota_cut_bytes, p_s_inode->i_uid, head2type(&s_ih));
7654 + DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
7655 DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
7657 /* Return deleted body length */
7658 @@ -1385,6 +1387,7 @@ void reiserfs_delete_solid_item(struct r
7660 DQUOT_FREE_SPACE_NODIRTY(inode,
7662 + DLIMIT_FREE_SPACE(inode, quota_cut_bytes);
7666 @@ -1734,6 +1737,7 @@ int reiserfs_cut_from_item(struct reiser
7667 "reiserquota cut_from_item(): freeing %u id=%u type=%c",
7668 quota_cut_bytes, p_s_inode->i_uid, '?');
7670 + DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
7671 DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
7674 @@ -1975,6 +1979,11 @@ int reiserfs_paste_into_item(struct reis
7675 pathrelse(p_s_search_path);
7678 + if (DLIMIT_ALLOC_SPACE(inode, n_pasted_size)) {
7679 + DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
7680 + pathrelse(p_s_search_path);
7683 init_tb_struct(th, &s_paste_balance, th->t_super, p_s_search_path,
7685 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
7686 @@ -2027,6 +2036,7 @@ int reiserfs_paste_into_item(struct reis
7687 n_pasted_size, inode->i_uid,
7688 key2type(&(p_s_key->on_disk_key)));
7690 + DLIMIT_FREE_SPACE(inode, n_pasted_size);
7691 DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
7694 @@ -2064,6 +2074,11 @@ int reiserfs_insert_item(struct reiserfs
7695 pathrelse(p_s_path);
7698 + if (DLIMIT_ALLOC_SPACE(inode, quota_bytes)) {
7699 + DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);
7700 + pathrelse(p_s_path);
7704 init_tb_struct(th, &s_ins_balance, th->t_super, p_s_path,
7705 IH_SIZE + ih_item_len(p_s_ih));
7706 @@ -2111,7 +2126,9 @@ int reiserfs_insert_item(struct reiserfs
7707 "reiserquota insert_item(): freeing %u id=%u type=%c",
7708 quota_bytes, inode->i_uid, head2type(p_s_ih));
7712 + DLIMIT_FREE_SPACE(inode, quota_bytes);
7713 DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);
7717 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/super.c
7718 --- linux-2.6.29.1/fs/reiserfs/super.c 2009-03-24 14:22:27.000000000 +0100
7719 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/super.c 2009-03-24 14:48:35.000000000 +0100
7720 @@ -902,6 +902,14 @@ static int reiserfs_parse_options(struct
7721 {"user_xattr",.setmask = 1 << REISERFS_UNSUPPORTED_OPT},
7722 {"nouser_xattr",.clrmask = 1 << REISERFS_UNSUPPORTED_OPT},
7724 +#ifndef CONFIG_TAGGING_NONE
7725 + {"tagxid",.setmask = 1 << REISERFS_TAGGED},
7726 + {"tag",.setmask = 1 << REISERFS_TAGGED},
7727 + {"notag",.clrmask = 1 << REISERFS_TAGGED},
7729 +#ifdef CONFIG_PROPAGATE
7730 + {"tag",.arg_required = 'T',.values = NULL},
7732 #ifdef CONFIG_REISERFS_FS_POSIX_ACL
7733 {"acl",.setmask = 1 << REISERFS_POSIXACL},
7734 {"noacl",.clrmask = 1 << REISERFS_POSIXACL},
7735 @@ -1195,6 +1203,12 @@ static int reiserfs_remount(struct super
7736 handle_quota_files(s, qf_names, &qfmt);
7739 + if ((mount_options & (1 << REISERFS_TAGGED)) &&
7740 + !(s->s_flags & MS_TAGGED)) {
7741 + reiserfs_warning(s, "reiserfs: tagging not permitted on remount.");
7747 /* Add options that are safe here */
7748 @@ -1659,6 +1673,10 @@ static int reiserfs_fill_super(struct su
7752 + /* map mount option tagxid */
7753 + if (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TAGGED))
7754 + s->s_flags |= MS_TAGGED;
7756 rs = SB_DISK_SUPER_BLOCK(s);
7757 /* Let's do basic sanity check to verify that underlying device is not
7758 smaller than the filesystem. If the check fails then abort and scream,
7759 diff -NurpP --minimal linux-2.6.29.1/fs/reiserfs/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/xattr.c
7760 --- linux-2.6.29.1/fs/reiserfs/xattr.c 2008-12-25 00:26:37.000000000 +0100
7761 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/reiserfs/xattr.c 2009-02-22 22:54:25.000000000 +0100
7763 #include <linux/namei.h>
7764 #include <linux/errno.h>
7765 #include <linux/fs.h>
7766 +#include <linux/mount.h>
7767 #include <linux/file.h>
7768 #include <linux/pagemap.h>
7769 #include <linux/xattr.h>
7770 diff -NurpP --minimal linux-2.6.29.1/fs/stat.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/stat.c
7771 --- linux-2.6.29.1/fs/stat.c 2009-03-24 14:22:36.000000000 +0100
7772 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/stat.c 2009-03-24 14:48:35.000000000 +0100
7773 @@ -26,6 +26,7 @@ void generic_fillattr(struct inode *inod
7774 stat->nlink = inode->i_nlink;
7775 stat->uid = inode->i_uid;
7776 stat->gid = inode->i_gid;
7777 + stat->tag = inode->i_tag;
7778 stat->rdev = inode->i_rdev;
7779 stat->atime = inode->i_atime;
7780 stat->mtime = inode->i_mtime;
7781 diff -NurpP --minimal linux-2.6.29.1/fs/super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/super.c
7782 --- linux-2.6.29.1/fs/super.c 2009-03-24 14:22:36.000000000 +0100
7783 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/super.c 2009-03-24 16:32:06.000000000 +0100
7785 #include <linux/mutex.h>
7786 #include <linux/file.h>
7787 #include <linux/async.h>
7788 +#include <linux/devpts_fs.h>
7789 +#include <linux/proc_fs.h>
7790 +#include <linux/vs_context.h>
7791 #include <asm/uaccess.h>
7792 #include "internal.h"
7794 @@ -918,12 +921,18 @@ struct vfsmount *
7795 vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
7797 struct vfsmount *mnt;
7798 + struct super_block *sb;
7799 char *secdata = NULL;
7803 return ERR_PTR(-ENODEV);
7806 + if ((type->fs_flags & FS_BINARY_MOUNTDATA) &&
7807 + !vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT))
7811 mnt = alloc_vfsmnt(name);
7813 @@ -942,9 +951,17 @@ vfs_kern_mount(struct file_system_type *
7814 error = type->get_sb(type, flags, name, data, mnt);
7816 goto out_free_secdata;
7817 - BUG_ON(!mnt->mnt_sb);
7819 - error = security_sb_kern_mount(mnt->mnt_sb, flags, secdata);
7824 + if (!vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT) && !sb->s_bdev &&
7825 + (sb->s_magic != PROC_SUPER_MAGIC) &&
7826 + (sb->s_magic != DEVPTS_SUPER_MAGIC))
7829 + error = security_sb_kern_mount(sb, flags, secdata);
7833 diff -NurpP --minimal linux-2.6.29.1/fs/sysfs/mount.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/sysfs/mount.c
7834 --- linux-2.6.29.1/fs/sysfs/mount.c 2008-12-25 00:26:37.000000000 +0100
7835 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/sysfs/mount.c 2009-02-22 22:54:25.000000000 +0100
7840 -/* Random magic number */
7841 -#define SYSFS_MAGIC 0x62656572
7843 static struct vfsmount *sysfs_mount;
7844 struct super_block * sysfs_sb = NULL;
7845 @@ -47,7 +45,7 @@ static int sysfs_fill_super(struct super
7847 sb->s_blocksize = PAGE_CACHE_SIZE;
7848 sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
7849 - sb->s_magic = SYSFS_MAGIC;
7850 + sb->s_magic = SYSFS_SUPER_MAGIC;
7851 sb->s_op = &sysfs_ops;
7852 sb->s_time_gran = 1;
7854 diff -NurpP --minimal linux-2.6.29.1/fs/utimes.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/utimes.c
7855 --- linux-2.6.29.1/fs/utimes.c 2009-03-24 14:22:37.000000000 +0100
7856 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/utimes.c 2009-02-22 22:54:25.000000000 +0100
7858 #include <linux/stat.h>
7859 #include <linux/utime.h>
7860 #include <linux/syscalls.h>
7861 +#include <linux/mount.h>
7862 +#include <linux/vs_cowbl.h>
7863 #include <asm/uaccess.h>
7864 #include <asm/unistd.h>
7866 diff -NurpP --minimal linux-2.6.29.1/fs/xattr.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xattr.c
7867 --- linux-2.6.29.1/fs/xattr.c 2009-03-24 14:22:37.000000000 +0100
7868 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xattr.c 2009-03-24 14:48:35.000000000 +0100
7870 #include <linux/module.h>
7871 #include <linux/fsnotify.h>
7872 #include <linux/audit.h>
7873 +#include <linux/mount.h>
7874 #include <asm/uaccess.h>
7877 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/linux-2.6/xfs_ioctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_ioctl.c
7878 --- linux-2.6.29.1/fs/xfs/linux-2.6/xfs_ioctl.c 2009-03-24 14:22:37.000000000 +0100
7879 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_ioctl.c 2009-03-24 14:48:35.000000000 +0100
7880 @@ -769,6 +769,10 @@ xfs_merge_ioc_xflags(
7881 xflags |= XFS_XFLAG_IMMUTABLE;
7883 xflags &= ~XFS_XFLAG_IMMUTABLE;
7884 + if (flags & FS_IXUNLINK_FL)
7885 + xflags |= XFS_XFLAG_IXUNLINK;
7887 + xflags &= ~XFS_XFLAG_IXUNLINK;
7888 if (flags & FS_APPEND_FL)
7889 xflags |= XFS_XFLAG_APPEND;
7891 @@ -797,6 +801,8 @@ xfs_di2lxflags(
7893 if (di_flags & XFS_DIFLAG_IMMUTABLE)
7894 flags |= FS_IMMUTABLE_FL;
7895 + if (di_flags & XFS_DIFLAG_IXUNLINK)
7896 + flags |= FS_IXUNLINK_FL;
7897 if (di_flags & XFS_DIFLAG_APPEND)
7898 flags |= FS_APPEND_FL;
7899 if (di_flags & XFS_DIFLAG_SYNC)
7900 @@ -855,6 +861,8 @@ xfs_set_diflags(
7901 di_flags = (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC);
7902 if (xflags & XFS_XFLAG_IMMUTABLE)
7903 di_flags |= XFS_DIFLAG_IMMUTABLE;
7904 + if (xflags & XFS_XFLAG_IXUNLINK)
7905 + di_flags |= XFS_DIFLAG_IXUNLINK;
7906 if (xflags & XFS_XFLAG_APPEND)
7907 di_flags |= XFS_DIFLAG_APPEND;
7908 if (xflags & XFS_XFLAG_SYNC)
7909 @@ -897,6 +905,10 @@ xfs_diflags_to_linux(
7910 inode->i_flags |= S_IMMUTABLE;
7912 inode->i_flags &= ~S_IMMUTABLE;
7913 + if (xflags & XFS_XFLAG_IXUNLINK)
7914 + inode->i_flags |= S_IXUNLINK;
7916 + inode->i_flags &= ~S_IXUNLINK;
7917 if (xflags & XFS_XFLAG_APPEND)
7918 inode->i_flags |= S_APPEND;
7920 @@ -1371,10 +1383,18 @@ xfs_file_ioctl(
7921 case XFS_IOC_FSGETXATTRA:
7922 return xfs_ioc_fsgetxattr(ip, 1, arg);
7923 case XFS_IOC_FSSETXATTR:
7924 + if (IS_BARRIER(inode)) {
7925 + vxwprintk_task(1, "messing with the barrier.");
7926 + return -XFS_ERROR(EACCES);
7928 return xfs_ioc_fssetxattr(ip, filp, arg);
7929 case XFS_IOC_GETXFLAGS:
7930 return xfs_ioc_getxflags(ip, arg);
7931 case XFS_IOC_SETXFLAGS:
7932 + if (IS_BARRIER(inode)) {
7933 + vxwprintk_task(1, "messing with the barrier.");
7934 + return -XFS_ERROR(EACCES);
7936 return xfs_ioc_setxflags(ip, filp, arg);
7938 case XFS_IOC_FSSETDM: {
7939 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/linux-2.6/xfs_iops.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_iops.c
7940 --- linux-2.6.29.1/fs/xfs/linux-2.6/xfs_iops.c 2009-03-24 14:22:37.000000000 +0100
7941 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_iops.c 2009-03-24 16:33:10.000000000 +0100
7943 #include <linux/security.h>
7944 #include <linux/falloc.h>
7945 #include <linux/fiemap.h>
7946 +#include <linux/vs_tag.h>
7949 * Bring the atime in the XFS inode uptodate.
7950 @@ -552,6 +553,7 @@ xfs_vn_getattr(
7951 stat->nlink = ip->i_d.di_nlink;
7952 stat->uid = ip->i_d.di_uid;
7953 stat->gid = ip->i_d.di_gid;
7954 + stat->tag = ip->i_d.di_tag;
7955 stat->ino = ip->i_ino;
7957 stat->ino += mp->m_inoadd;
7958 @@ -591,6 +593,12 @@ xfs_vn_getattr(
7962 +xfs_vn_sync_xflags(struct inode *inode)
7964 + return -xfs_sync_xflags(XFS_I(inode));
7969 struct dentry *dentry,
7970 struct iattr *iattr)
7971 @@ -748,6 +756,7 @@ static const struct inode_operations xfs
7972 .listxattr = xfs_vn_listxattr,
7973 .fallocate = xfs_vn_fallocate,
7974 .fiemap = xfs_vn_fiemap,
7975 + .sync_flags = xfs_vn_sync_xflags,
7978 static const struct inode_operations xfs_dir_inode_operations = {
7979 @@ -773,6 +782,7 @@ static const struct inode_operations xfs
7980 .getxattr = generic_getxattr,
7981 .removexattr = generic_removexattr,
7982 .listxattr = xfs_vn_listxattr,
7983 + .sync_flags = xfs_vn_sync_xflags,
7986 static const struct inode_operations xfs_dir_ci_inode_operations = {
7987 @@ -822,6 +832,10 @@ xfs_diflags_to_iflags(
7988 inode->i_flags |= S_IMMUTABLE;
7990 inode->i_flags &= ~S_IMMUTABLE;
7991 + if (ip->i_d.di_flags & XFS_DIFLAG_IXUNLINK)
7992 + inode->i_flags |= S_IXUNLINK;
7994 + inode->i_flags &= ~S_IXUNLINK;
7995 if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
7996 inode->i_flags |= S_APPEND;
7998 @@ -834,6 +848,15 @@ xfs_diflags_to_iflags(
7999 inode->i_flags |= S_NOATIME;
8001 inode->i_flags &= ~S_NOATIME;
8003 + if (ip->i_d.di_vflags & XFS_DIVFLAG_BARRIER)
8004 + inode->i_vflags |= V_BARRIER;
8006 + inode->i_vflags &= ~V_BARRIER;
8007 + if (ip->i_d.di_vflags & XFS_DIVFLAG_COW)
8008 + inode->i_vflags |= V_COW;
8010 + inode->i_vflags &= ~V_COW;
8014 @@ -862,6 +885,7 @@ xfs_setup_inode(
8015 inode->i_nlink = ip->i_d.di_nlink;
8016 inode->i_uid = ip->i_d.di_uid;
8017 inode->i_gid = ip->i_d.di_gid;
8018 + inode->i_tag = ip->i_d.di_tag;
8020 switch (inode->i_mode & S_IFMT) {
8022 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/linux-2.6/xfs_linux.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_linux.h
8023 --- linux-2.6.29.1/fs/xfs/linux-2.6/xfs_linux.h 2009-03-24 14:22:37.000000000 +0100
8024 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_linux.h 2009-03-25 01:26:04.000000000 +0100
8027 #define current_cpu() (raw_smp_processor_id())
8028 #define current_pid() (current->pid)
8029 +#define current_fstag(cred,vp) (dx_current_fstag((vp)->i_sb))
8030 #define current_test_flags(f) (current->flags & (f))
8031 #define current_set_flags_nested(sp, f) \
8032 (*(sp) = current->flags, current->flags |= (f))
8033 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/linux-2.6/xfs_super.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_super.c
8034 --- linux-2.6.29.1/fs/xfs/linux-2.6/xfs_super.c 2009-03-24 14:22:37.000000000 +0100
8035 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/linux-2.6/xfs_super.c 2009-03-25 01:48:51.000000000 +0100
8036 @@ -120,6 +120,9 @@ mempool_t *xfs_ioend_pool;
8037 #define MNTOPT_DMAPI "dmapi" /* DMI enabled (DMAPI / XDSM) */
8038 #define MNTOPT_XDSM "xdsm" /* DMI enabled (DMAPI / XDSM) */
8039 #define MNTOPT_DMI "dmi" /* DMI enabled (DMAPI / XDSM) */
8040 +#define MNTOPT_TAGXID "tagxid" /* context tagging for inodes */
8041 +#define MNTOPT_TAGGED "tag" /* context tagging for inodes */
8042 +#define MNTOPT_NOTAGTAG "notag" /* do not use context tagging */
8045 * Table driven mount option parser.
8046 @@ -128,10 +131,14 @@ mempool_t *xfs_ioend_pool;
8047 * in the future, too.
8050 + Opt_tag, Opt_notag,
8051 Opt_barrier, Opt_nobarrier, Opt_err
8054 static const match_table_t tokens = {
8055 + {Opt_tag, "tagxid"},
8057 + {Opt_notag, "notag"},
8058 {Opt_barrier, "barrier"},
8059 {Opt_nobarrier, "nobarrier"},
8061 @@ -395,6 +402,19 @@ xfs_parseargs(
8062 } else if (!strcmp(this_char, "irixsgid")) {
8064 "XFS: irixsgid is now a sysctl(2) variable, option is deprecated.");
8065 +#ifndef CONFIG_TAGGING_NONE
8066 + } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
8067 + mp->m_flags |= XFS_MOUNT_TAGGED;
8068 + } else if (!strcmp(this_char, MNTOPT_NOTAGTAG)) {
8069 + mp->m_flags &= ~XFS_MOUNT_TAGGED;
8070 + } else if (!strcmp(this_char, MNTOPT_TAGXID)) {
8071 + mp->m_flags |= XFS_MOUNT_TAGGED;
8073 +#ifdef CONFIG_PROPAGATE
8074 + } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
8076 + mp->m_flags |= XFS_MOUNT_TAGGED;
8080 "XFS: unknown mount option [%s].", this_char);
8081 @@ -1221,6 +1241,16 @@ xfs_fs_remount(
8083 mp->m_flags &= ~XFS_MOUNT_BARRIER;
8086 + if (!(sb->s_flags & MS_TAGGED)) {
8088 + "XFS: %s: tagging not permitted on remount.\n",
8097 * Logically we would return an error here to prevent
8098 @@ -1480,6 +1510,9 @@ xfs_fs_fill_super(
8100 XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, mtpt, mp->m_fsname);
8102 + if (mp->m_flags & XFS_MOUNT_TAGGED)
8103 + sb->s_flags |= MS_TAGGED;
8106 sb->s_magic = XFS_SB_MAGIC;
8107 sb->s_blocksize = mp->m_sb.sb_blocksize;
8108 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/quota/xfs_qm_syscalls.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/quota/xfs_qm_syscalls.c
8109 --- linux-2.6.29.1/fs/xfs/quota/xfs_qm_syscalls.c 2009-03-24 14:22:37.000000000 +0100
8110 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/quota/xfs_qm_syscalls.c 2009-03-24 14:48:35.000000000 +0100
8111 @@ -426,7 +426,7 @@ xfs_qm_scall_quotaon(
8115 - if (!capable(CAP_SYS_ADMIN))
8116 + if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
8117 return XFS_ERROR(EPERM);
8119 flags &= (XFS_ALL_QUOTA_ACCT | XFS_ALL_QUOTA_ENFD);
8120 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_dinode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_dinode.h
8121 --- linux-2.6.29.1/fs/xfs/xfs_dinode.h 2009-03-24 14:22:37.000000000 +0100
8122 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_dinode.h 2009-03-24 16:38:27.000000000 +0100
8123 @@ -50,7 +50,9 @@ typedef struct xfs_dinode {
8124 __be32 di_gid; /* owner's group id */
8125 __be32 di_nlink; /* number of links to file */
8126 __be16 di_projid; /* owner's project id */
8127 - __u8 di_pad[8]; /* unused, zeroed space */
8128 + __be16 di_tag; /* context tagging */
8129 + __be16 di_vflags; /* vserver specific flags */
8130 + __u8 di_pad[4]; /* unused, zeroed space */
8131 __be16 di_flushiter; /* incremented on flush */
8132 xfs_timestamp_t di_atime; /* time last accessed */
8133 xfs_timestamp_t di_mtime; /* time last modified */
8134 @@ -181,6 +183,8 @@ static inline void xfs_dinode_put_rdev(s
8135 #define XFS_DIFLAG_EXTSZINHERIT_BIT 12 /* inherit inode extent size */
8136 #define XFS_DIFLAG_NODEFRAG_BIT 13 /* do not reorganize/defragment */
8137 #define XFS_DIFLAG_FILESTREAM_BIT 14 /* use filestream allocator */
8138 +#define XFS_DIFLAG_IXUNLINK_BIT 15 /* Immutable inver on unlink */
8140 #define XFS_DIFLAG_REALTIME (1 << XFS_DIFLAG_REALTIME_BIT)
8141 #define XFS_DIFLAG_PREALLOC (1 << XFS_DIFLAG_PREALLOC_BIT)
8142 #define XFS_DIFLAG_NEWRTBM (1 << XFS_DIFLAG_NEWRTBM_BIT)
8143 @@ -196,6 +200,7 @@ static inline void xfs_dinode_put_rdev(s
8144 #define XFS_DIFLAG_EXTSZINHERIT (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
8145 #define XFS_DIFLAG_NODEFRAG (1 << XFS_DIFLAG_NODEFRAG_BIT)
8146 #define XFS_DIFLAG_FILESTREAM (1 << XFS_DIFLAG_FILESTREAM_BIT)
8147 +#define XFS_DIFLAG_IXUNLINK (1 << XFS_DIFLAG_IXUNLINK_BIT)
8149 #ifdef CONFIG_XFS_RT
8150 #define XFS_IS_REALTIME_INODE(ip) ((ip)->i_d.di_flags & XFS_DIFLAG_REALTIME)
8151 @@ -208,6 +213,10 @@ static inline void xfs_dinode_put_rdev(s
8152 XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
8153 XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
8154 XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
8155 - XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
8156 + XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM | \
8157 + XFS_DIFLAG_IXUNLINK)
8159 +#define XFS_DIVFLAG_BARRIER 0x01
8160 +#define XFS_DIVFLAG_COW 0x02
8162 #endif /* __XFS_DINODE_H__ */
8163 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_fs.h
8164 --- linux-2.6.29.1/fs/xfs/xfs_fs.h 2009-03-24 14:22:37.000000000 +0100
8165 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_fs.h 2009-03-24 14:48:35.000000000 +0100
8166 @@ -67,6 +67,9 @@ struct fsxattr {
8167 #define XFS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
8168 #define XFS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
8169 #define XFS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
8170 +#define XFS_XFLAG_IXUNLINK 0x00008000 /* immutable invert on unlink */
8171 +#define XFS_XFLAG_BARRIER 0x10000000 /* chroot() barrier */
8172 +#define XFS_XFLAG_COW 0x20000000 /* copy on write mark */
8173 #define XFS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
8176 @@ -289,7 +292,8 @@ typedef struct xfs_bstat {
8177 __s32 bs_extents; /* number of extents */
8178 __u32 bs_gen; /* generation count */
8179 __u16 bs_projid; /* project id */
8180 - unsigned char bs_pad[14]; /* pad space, unused */
8181 + __u16 bs_tag; /* context tagging */
8182 + unsigned char bs_pad[12]; /* pad space, unused */
8183 __u32 bs_dmevmask; /* DMIG event mask */
8184 __u16 bs_dmstate; /* DMIG state info */
8185 __u16 bs_aextents; /* attribute number of extents */
8186 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_ialloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_ialloc.c
8187 --- linux-2.6.29.1/fs/xfs/xfs_ialloc.c 2009-03-24 14:22:37.000000000 +0100
8188 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_ialloc.c 2009-03-24 17:03:17.000000000 +0100
8190 #include "xfs_error.h"
8191 #include "xfs_bmap.h"
8195 * Allocation group level functions.
8197 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_inode.c
8198 --- linux-2.6.29.1/fs/xfs/xfs_inode.c 2009-03-24 14:22:37.000000000 +0100
8199 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_inode.c 2009-03-25 01:42:50.000000000 +0100
8200 @@ -250,6 +250,7 @@ xfs_inotobp(
8204 +#include <linux/vs_tag.h>
8207 * This routine is called to map an inode to the buffer containing
8208 @@ -645,15 +646,25 @@ xfs_iformat_btree(
8210 xfs_dinode_from_disk(
8212 - xfs_dinode_t *from)
8213 + xfs_dinode_t *from,
8216 + uint32_t uid, gid, tag;
8218 to->di_magic = be16_to_cpu(from->di_magic);
8219 to->di_mode = be16_to_cpu(from->di_mode);
8220 to->di_version = from ->di_version;
8221 to->di_format = from->di_format;
8222 to->di_onlink = be16_to_cpu(from->di_onlink);
8223 - to->di_uid = be32_to_cpu(from->di_uid);
8224 - to->di_gid = be32_to_cpu(from->di_gid);
8226 + uid = be32_to_cpu(from->di_uid);
8227 + gid = be32_to_cpu(from->di_gid);
8228 + tag = be16_to_cpu(from->di_tag);
8230 + to->di_uid = INOTAG_UID(tagged, uid, gid);
8231 + to->di_gid = INOTAG_GID(tagged, uid, gid);
8232 + to->di_tag = INOTAG_TAG(tagged, uid, gid, tag);
8234 to->di_nlink = be32_to_cpu(from->di_nlink);
8235 to->di_projid = be16_to_cpu(from->di_projid);
8236 memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8237 @@ -674,21 +685,26 @@ xfs_dinode_from_disk(
8238 to->di_dmevmask = be32_to_cpu(from->di_dmevmask);
8239 to->di_dmstate = be16_to_cpu(from->di_dmstate);
8240 to->di_flags = be16_to_cpu(from->di_flags);
8241 + to->di_vflags = be16_to_cpu(from->di_vflags);
8242 to->di_gen = be32_to_cpu(from->di_gen);
8248 - xfs_icdinode_t *from)
8249 + xfs_icdinode_t *from,
8252 to->di_magic = cpu_to_be16(from->di_magic);
8253 to->di_mode = cpu_to_be16(from->di_mode);
8254 to->di_version = from ->di_version;
8255 to->di_format = from->di_format;
8256 to->di_onlink = cpu_to_be16(from->di_onlink);
8257 - to->di_uid = cpu_to_be32(from->di_uid);
8258 - to->di_gid = cpu_to_be32(from->di_gid);
8260 + to->di_uid = cpu_to_be32(TAGINO_UID(tagged, from->di_uid, from->di_tag));
8261 + to->di_gid = cpu_to_be32(TAGINO_GID(tagged, from->di_gid, from->di_tag));
8262 + to->di_tag = cpu_to_be16(TAGINO_TAG(tagged, from->di_tag));
8264 to->di_nlink = cpu_to_be32(from->di_nlink);
8265 to->di_projid = cpu_to_be16(from->di_projid);
8266 memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8267 @@ -709,12 +725,14 @@ xfs_dinode_to_disk(
8268 to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
8269 to->di_dmstate = cpu_to_be16(from->di_dmstate);
8270 to->di_flags = cpu_to_be16(from->di_flags);
8271 + to->di_vflags = cpu_to_be16(from->di_vflags);
8272 to->di_gen = cpu_to_be32(from->di_gen);
8277 - __uint16_t di_flags)
8278 + __uint16_t di_flags,
8279 + __uint16_t di_vflags)
8283 @@ -725,6 +743,8 @@ _xfs_dic2xflags(
8284 flags |= XFS_XFLAG_PREALLOC;
8285 if (di_flags & XFS_DIFLAG_IMMUTABLE)
8286 flags |= XFS_XFLAG_IMMUTABLE;
8287 + if (di_flags & XFS_DIFLAG_IXUNLINK)
8288 + flags |= XFS_XFLAG_IXUNLINK;
8289 if (di_flags & XFS_DIFLAG_APPEND)
8290 flags |= XFS_XFLAG_APPEND;
8291 if (di_flags & XFS_DIFLAG_SYNC)
8292 @@ -749,6 +769,10 @@ _xfs_dic2xflags(
8293 flags |= XFS_XFLAG_FILESTREAM;
8296 + if (di_vflags & XFS_DIVFLAG_BARRIER)
8297 + flags |= FS_BARRIER_FL;
8298 + if (di_vflags & XFS_DIVFLAG_COW)
8299 + flags |= FS_COW_FL;
8303 @@ -758,7 +782,7 @@ xfs_ip2xflags(
8305 xfs_icdinode_t *dic = &ip->i_d;
8307 - return _xfs_dic2xflags(dic->di_flags) |
8308 + return _xfs_dic2xflags(dic->di_flags, dic->di_vflags) |
8309 (XFS_IFORK_Q(ip) ? XFS_XFLAG_HASATTR : 0);
8312 @@ -766,7 +790,8 @@ uint
8316 - return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
8317 + return _xfs_dic2xflags(be16_to_cpu(dip->di_flags),
8318 + be16_to_cpu(dip->di_vflags)) |
8319 (XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
8322 @@ -802,7 +827,6 @@ xfs_iread(
8325 dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
8328 * If we got something that isn't an inode it means someone
8329 * (nfs or dmi) has a stale handle.
8330 @@ -827,7 +851,8 @@ xfs_iread(
8331 * Otherwise, just get the truly permanent information.
8334 - xfs_dinode_from_disk(&ip->i_d, dip);
8335 + xfs_dinode_from_disk(&ip->i_d, dip,
8336 + mp->m_flags & XFS_MOUNT_TAGGED);
8337 error = xfs_iformat(ip, dip);
8340 @@ -1027,6 +1052,7 @@ xfs_ialloc(
8341 ASSERT(ip->i_d.di_nlink == nlink);
8342 ip->i_d.di_uid = current_fsuid();
8343 ip->i_d.di_gid = current_fsgid();
8344 + ip->i_d.di_tag = current_fstag(cr, &ip->i_vnode);
8345 ip->i_d.di_projid = prid;
8346 memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
8348 @@ -1087,6 +1113,7 @@ xfs_ialloc(
8349 ip->i_d.di_dmevmask = 0;
8350 ip->i_d.di_dmstate = 0;
8351 ip->i_d.di_flags = 0;
8352 + ip->i_d.di_vflags = 0;
8353 flags = XFS_ILOG_CORE;
8354 switch (mode & S_IFMT) {
8356 @@ -2161,6 +2188,7 @@ xfs_ifree(
8358 ip->i_d.di_mode = 0; /* mark incore inode as free */
8359 ip->i_d.di_flags = 0;
8360 + ip->i_d.di_vflags = 0;
8361 ip->i_d.di_dmevmask = 0;
8362 ip->i_d.di_forkoff = 0; /* mark the attr fork not in use */
8363 ip->i_df.if_ext_max =
8364 @@ -3128,7 +3156,8 @@ xfs_iflush_int(
8365 * because if the inode is dirty at all the core must
8368 - xfs_dinode_to_disk(dip, &ip->i_d);
8369 + xfs_dinode_to_disk(dip, &ip->i_d,
8370 + mp->m_flags & XFS_MOUNT_TAGGED);
8372 /* Wrap, we never let the log put out DI_MAX_FLUSH */
8373 if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
8374 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_inode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_inode.h
8375 --- linux-2.6.29.1/fs/xfs/xfs_inode.h 2009-03-24 14:22:37.000000000 +0100
8376 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_inode.h 2009-03-25 01:10:35.000000000 +0100
8377 @@ -134,7 +134,9 @@ typedef struct xfs_icdinode {
8378 __uint32_t di_gid; /* owner's group id */
8379 __uint32_t di_nlink; /* number of links to file */
8380 __uint16_t di_projid; /* owner's project id */
8381 - __uint8_t di_pad[8]; /* unused, zeroed space */
8382 + __uint16_t di_tag; /* context tagging */
8383 + __uint16_t di_vflags; /* vserver specific flags */
8384 + __uint8_t di_pad[4]; /* unused, zeroed space */
8385 __uint16_t di_flushiter; /* incremented on flush */
8386 xfs_ictimestamp_t di_atime; /* time last accessed */
8387 xfs_ictimestamp_t di_mtime; /* time last modified */
8388 @@ -589,9 +591,9 @@ int xfs_itobp(struct xfs_mount *, struc
8389 int xfs_iread(struct xfs_mount *, struct xfs_trans *,
8390 struct xfs_inode *, xfs_daddr_t, uint);
8391 void xfs_dinode_from_disk(struct xfs_icdinode *,
8392 - struct xfs_dinode *);
8393 + struct xfs_dinode *, int);
8394 void xfs_dinode_to_disk(struct xfs_dinode *,
8395 - struct xfs_icdinode *);
8396 + struct xfs_icdinode *, int);
8397 void xfs_idestroy_fork(struct xfs_inode *, int);
8398 void xfs_idata_realloc(struct xfs_inode *, int, int);
8399 void xfs_iroot_realloc(struct xfs_inode *, int, int);
8400 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_itable.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_itable.c
8401 --- linux-2.6.29.1/fs/xfs/xfs_itable.c 2009-03-24 14:22:37.000000000 +0100
8402 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_itable.c 2009-03-24 14:48:35.000000000 +0100
8403 @@ -82,6 +82,7 @@ xfs_bulkstat_one_iget(
8404 buf->bs_mode = dic->di_mode;
8405 buf->bs_uid = dic->di_uid;
8406 buf->bs_gid = dic->di_gid;
8407 + buf->bs_tag = dic->di_tag;
8408 buf->bs_size = dic->di_size;
8409 vn_atime_to_bstime(VFS_I(ip), &buf->bs_atime);
8410 buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
8411 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_log_recover.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_log_recover.c
8412 --- linux-2.6.29.1/fs/xfs/xfs_log_recover.c 2009-03-24 14:22:37.000000000 +0100
8413 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_log_recover.c 2009-03-24 17:17:15.000000000 +0100
8414 @@ -2421,7 +2421,8 @@ xlog_recover_do_inode_trans(
8417 /* The core is in in-core format */
8418 - xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr);
8419 + xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr,
8420 + mp->m_flags & XFS_MOUNT_TAGGED);
8422 /* the rest is in on-disk format */
8423 if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
8424 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_mount.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_mount.h
8425 --- linux-2.6.29.1/fs/xfs/xfs_mount.h 2009-03-24 14:22:37.000000000 +0100
8426 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_mount.h 2009-03-24 14:48:35.000000000 +0100
8427 @@ -371,6 +371,7 @@ typedef struct xfs_mount {
8429 #define XFS_MOUNT_NOATTR2 (1ULL << 25) /* disable use of attr2 format */
8431 +#define XFS_MOUNT_TAGGED (1ULL << 31) /* context tagging */
8434 * Default minimum read and write sizes.
8435 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_vnodeops.c linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_vnodeops.c
8436 --- linux-2.6.29.1/fs/xfs/xfs_vnodeops.c 2009-03-24 14:22:37.000000000 +0100
8437 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_vnodeops.c 2009-03-24 17:18:40.000000000 +0100
8439 #include "xfs_filestream.h"
8440 #include "xfs_vnodeops.h"
8444 +xfs_get_inode_flags(
8447 + struct inode *inode = VFS_I(ip);
8448 + unsigned int flags = inode->i_flags;
8449 + unsigned int vflags = inode->i_vflags;
8451 + if (flags & S_IMMUTABLE)
8452 + ip->i_d.di_flags |= XFS_DIFLAG_IMMUTABLE;
8454 + ip->i_d.di_flags &= ~XFS_DIFLAG_IMMUTABLE;
8455 + if (flags & S_IXUNLINK)
8456 + ip->i_d.di_flags |= XFS_DIFLAG_IXUNLINK;
8458 + ip->i_d.di_flags &= ~XFS_DIFLAG_IXUNLINK;
8460 + if (vflags & V_BARRIER)
8461 + ip->i_d.di_vflags |= XFS_DIVFLAG_BARRIER;
8463 + ip->i_d.di_vflags &= ~XFS_DIVFLAG_BARRIER;
8464 + if (vflags & V_COW)
8465 + ip->i_d.di_vflags |= XFS_DIVFLAG_COW;
8467 + ip->i_d.di_vflags &= ~XFS_DIVFLAG_COW;
8474 + struct xfs_mount *mp = ip->i_mount;
8475 + struct xfs_trans *tp;
8476 + unsigned int lock_flags = 0;
8479 + xfs_itrace_entry(ip);
8481 + if (mp->m_flags & XFS_MOUNT_RDONLY)
8482 + return XFS_ERROR(EROFS);
8485 + * we acquire the inode lock and do an error checking pass.
8487 + tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
8488 + code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
8490 + goto error_return;
8492 + lock_flags = XFS_ILOCK_EXCL;
8493 + xfs_ilock(ip, lock_flags);
8495 + xfs_trans_ijoin(tp, ip, lock_flags);
8496 + xfs_trans_ihold(tp, ip);
8498 + xfs_get_inode_flags(ip);
8499 + // xfs_diflags_to_linux(ip);
8501 + xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
8502 + xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
8504 + XFS_STATS_INC(xs_ig_attrchg);
8507 + * If this is a synchronous mount, make sure that the
8508 + * transaction goes to disk before returning to the user.
8510 + if (mp->m_flags & XFS_MOUNT_WSYNC)
8511 + xfs_trans_set_sync(tp);
8512 + code = xfs_trans_commit(tp, 0);
8513 + xfs_iunlock(ip, lock_flags);
8520 + xfs_trans_cancel(tp, 0);
8522 + xfs_iunlock(ip, lock_flags);
8528 struct xfs_inode *ip,
8529 @@ -68,6 +152,7 @@ xfs_setattr(
8530 uint commit_flags=0;
8531 uid_t uid=0, iuid=0;
8532 gid_t gid=0, igid=0;
8533 + tag_t tag=0, itag=0;
8535 struct xfs_dquot *udqp, *gdqp, *olddquot1, *olddquot2;
8536 int need_iolock = 1;
8537 @@ -164,7 +249,7 @@ xfs_setattr(
8539 * Change file ownership. Must be the owner or privileged.
8541 - if (mask & (ATTR_UID|ATTR_GID)) {
8542 + if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8544 * These IDs could have changed since we last looked at them.
8545 * But, we're assured that if the ownership did change
8546 @@ -173,15 +258,19 @@ xfs_setattr(
8548 iuid = ip->i_d.di_uid;
8549 igid = ip->i_d.di_gid;
8550 + itag = ip->i_d.di_tag;
8551 gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
8552 uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
8553 + tag = (mask & ATTR_TAG) ? iattr->ia_tag : itag;
8556 * Do a quota reservation only if uid/gid is actually
8559 if ((XFS_IS_UQUOTA_ON(mp) && iuid != uid) ||
8560 - (XFS_IS_GQUOTA_ON(mp) && igid != gid)) {
8561 + (XFS_IS_GQUOTA_ON(mp) && igid != gid) ||
8562 + (XFS_IS_GQUOTA_ON(mp) && itag != tag)) {
8563 + /* TODO: handle tagging? */
8565 code = XFS_QM_DQVOPCHOWNRESV(mp, tp, ip, udqp, gdqp,
8566 capable(CAP_FOWNER) ?
8567 @@ -334,7 +423,7 @@ xfs_setattr(
8569 * Change file ownership. Must be the owner or privileged.
8571 - if (mask & (ATTR_UID|ATTR_GID)) {
8572 + if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8574 * CAP_FSETID overrides the following restrictions:
8576 @@ -350,6 +439,10 @@ xfs_setattr(
8577 * Change the ownerships and register quota modifications
8578 * in the transaction.
8580 + if (itag != tag) {
8581 + ip->i_d.di_tag = tag;
8582 + inode->i_tag = tag;
8585 if (XFS_IS_UQUOTA_ON(mp)) {
8586 ASSERT(mask & ATTR_UID);
8587 diff -NurpP --minimal linux-2.6.29.1/fs/xfs/xfs_vnodeops.h linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_vnodeops.h
8588 --- linux-2.6.29.1/fs/xfs/xfs_vnodeops.h 2009-03-24 14:22:37.000000000 +0100
8589 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/fs/xfs/xfs_vnodeops.h 2009-03-24 17:19:30.000000000 +0100
8590 @@ -14,6 +14,7 @@ struct xfs_inode;
8594 +int xfs_sync_xflags(struct xfs_inode *ip);
8595 int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
8596 #define XFS_ATTR_DMI 0x01 /* invocation from a DMI function */
8597 #define XFS_ATTR_NONBLOCK 0x02 /* return EAGAIN if operation would block */
8598 diff -NurpP --minimal linux-2.6.29.1/include/asm-generic/tlb.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/asm-generic/tlb.h
8599 --- linux-2.6.29.1/include/asm-generic/tlb.h 2008-12-25 00:26:37.000000000 +0100
8600 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/asm-generic/tlb.h 2009-04-08 15:50:06.000000000 +0200
8602 #define _ASM_GENERIC__TLB_H
8604 #include <linux/swap.h>
8605 +#include <linux/vs_memory.h>
8606 #include <asm/pgalloc.h>
8607 #include <asm/tlbflush.h>
8609 diff -NurpP --minimal linux-2.6.29.1/include/linux/capability.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/capability.h
8610 --- linux-2.6.29.1/include/linux/capability.h 2009-03-24 14:22:41.000000000 +0100
8611 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/capability.h 2009-03-24 14:48:35.000000000 +0100
8612 @@ -285,6 +285,7 @@ struct cpu_vfs_cap_data {
8613 arbitrary SCSI commands */
8614 /* Allow setting encryption key on loopback filesystem */
8615 /* Allow setting zone reclaim policy */
8616 +/* Allow the selection of a security context */
8618 #define CAP_SYS_ADMIN 21
8620 @@ -357,7 +358,13 @@ struct cpu_vfs_cap_data {
8622 #define CAP_MAC_ADMIN 33
8624 -#define CAP_LAST_CAP CAP_MAC_ADMIN
8625 +/* Allow context manipulations */
8626 +/* Allow changing context info on files */
8628 +#define CAP_CONTEXT 34
8631 +#define CAP_LAST_CAP CAP_CONTEXT
8633 #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
8635 diff -NurpP --minimal linux-2.6.29.1/include/linux/devpts_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/devpts_fs.h
8636 --- linux-2.6.29.1/include/linux/devpts_fs.h 2008-12-25 00:26:37.000000000 +0100
8637 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/devpts_fs.h 2009-02-22 22:54:26.000000000 +0100
8638 @@ -45,5 +45,4 @@ static inline void devpts_pty_kill(struc
8643 #endif /* _LINUX_DEVPTS_FS_H */
8644 diff -NurpP --minimal linux-2.6.29.1/include/linux/ext2_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ext2_fs.h
8645 --- linux-2.6.29.1/include/linux/ext2_fs.h 2009-03-24 14:22:41.000000000 +0100
8646 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ext2_fs.h 2009-03-24 15:39:54.000000000 +0100
8647 @@ -189,8 +189,12 @@ struct ext2_group_desc
8648 #define EXT2_NOTAIL_FL FS_NOTAIL_FL /* file tail should not be merged */
8649 #define EXT2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
8650 #define EXT2_TOPDIR_FL FS_TOPDIR_FL /* Top of directory hierarchies*/
8651 +#define EXT2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
8652 #define EXT2_RESERVED_FL FS_RESERVED_FL /* reserved for ext2 lib */
8654 +#define EXT2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
8655 +#define EXT2_COW_FL FS_COW_FL /* Copy on Write marker */
8657 #define EXT2_FL_USER_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */
8658 #define EXT2_FL_USER_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */
8660 @@ -271,7 +275,7 @@ struct ext2_inode {
8662 __u8 l_i_frag; /* Fragment number */
8663 __u8 l_i_fsize; /* Fragment size */
8665 + __u16 l_i_tag; /* Context Tag */
8666 __le16 l_i_uid_high; /* these 2 fields */
8667 __le16 l_i_gid_high; /* were reserved2[0] */
8668 __u32 l_i_reserved2;
8669 @@ -303,6 +307,7 @@ struct ext2_inode {
8670 #define i_gid_low i_gid
8671 #define i_uid_high osd2.linux2.l_i_uid_high
8672 #define i_gid_high osd2.linux2.l_i_gid_high
8673 +#define i_raw_tag osd2.linux2.l_i_tag
8674 #define i_reserved2 osd2.linux2.l_i_reserved2
8677 @@ -347,6 +352,7 @@ struct ext2_inode {
8678 #define EXT2_MOUNT_USRQUOTA 0x020000 /* user quota */
8679 #define EXT2_MOUNT_GRPQUOTA 0x040000 /* group quota */
8680 #define EXT2_MOUNT_RESERVATION 0x080000 /* Preallocation */
8681 +#define EXT2_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8684 #define clear_opt(o, opt) o &= ~EXT2_MOUNT_##opt
8685 diff -NurpP --minimal linux-2.6.29.1/include/linux/ext3_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ext3_fs.h
8686 --- linux-2.6.29.1/include/linux/ext3_fs.h 2009-03-24 14:22:41.000000000 +0100
8687 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ext3_fs.h 2009-03-24 14:48:36.000000000 +0100
8688 @@ -173,10 +173,14 @@ struct ext3_group_desc
8689 #define EXT3_NOTAIL_FL 0x00008000 /* file tail should not be merged */
8690 #define EXT3_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
8691 #define EXT3_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8692 +#define EXT3_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8693 #define EXT3_RESERVED_FL 0x80000000 /* reserved for ext3 lib */
8695 -#define EXT3_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8696 -#define EXT3_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8697 +#define EXT3_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8698 +#define EXT3_COW_FL 0x20000000 /* Copy on Write marker */
8700 +#define EXT3_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8701 +#define EXT3_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8703 /* Flags that should be inherited by new inodes from their parent. */
8704 #define EXT3_FL_INHERITED (EXT3_SECRM_FL | EXT3_UNRM_FL | EXT3_COMPR_FL |\
8705 @@ -316,7 +320,7 @@ struct ext3_inode {
8707 __u8 l_i_frag; /* Fragment number */
8708 __u8 l_i_fsize; /* Fragment size */
8710 + __u16 l_i_tag; /* Context Tag */
8711 __le16 l_i_uid_high; /* these 2 fields */
8712 __le16 l_i_gid_high; /* were reserved2[0] */
8713 __u32 l_i_reserved2;
8714 @@ -350,6 +354,7 @@ struct ext3_inode {
8715 #define i_gid_low i_gid
8716 #define i_uid_high osd2.linux2.l_i_uid_high
8717 #define i_gid_high osd2.linux2.l_i_gid_high
8718 +#define i_raw_tag osd2.linux2.l_i_tag
8719 #define i_reserved2 osd2.linux2.l_i_reserved2
8721 #elif defined(__GNU__)
8722 @@ -413,6 +418,7 @@ struct ext3_inode {
8723 #define EXT3_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
8724 #define EXT3_MOUNT_DATA_ERR_ABORT 0x400000 /* Abort on file data write
8725 * error in ordered mode */
8726 +#define EXT3_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8728 /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
8729 #ifndef _LINUX_EXT2_FS_H
8730 @@ -874,6 +880,7 @@ struct buffer_head * ext3_bread (handle_
8731 int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
8732 sector_t iblock, unsigned long maxblocks, struct buffer_head *bh_result,
8733 int create, int extend_disksize);
8734 +extern int ext3_sync_flags(struct inode *inode);
8736 extern struct inode *ext3_iget(struct super_block *, unsigned long);
8737 extern int ext3_write_inode (struct inode *, int);
8738 diff -NurpP --minimal linux-2.6.29.1/include/linux/fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/fs.h
8739 --- linux-2.6.29.1/include/linux/fs.h 2009-03-24 14:22:41.000000000 +0100
8740 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/fs.h 2009-03-24 17:21:20.000000000 +0100
8741 @@ -141,6 +141,9 @@ struct inodes_stat_t {
8742 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
8743 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
8744 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
8745 +#define MS_TAGGED (1<<24) /* use generic inode tagging */
8746 +#define MS_TAGID (1<<25) /* use specific tag for this mount */
8747 +#define MS_NOTAGCHECK (1<<26) /* don't check tags */
8748 #define MS_ACTIVE (1<<30)
8749 #define MS_NOUSER (1<<31)
8751 @@ -167,6 +170,14 @@ struct inodes_stat_t {
8752 #define S_PRIVATE 512 /* Inode is fs-internal */
8753 #define S_ATOMIC_COPY 1024 /* Pages mapped with this inode need to be
8754 atomically copied (gem) */
8755 +#define S_IXUNLINK 2048 /* Immutable Invert on unlink */
8757 +/* Linux-VServer related Inode flags */
8761 +#define V_BARRIER 4 /* Barrier for chroot() */
8762 +#define V_COW 8 /* Copy on Write */
8765 * Note that nosuid etc flags are inode-specific: setting some file-system
8766 @@ -189,12 +200,15 @@ struct inodes_stat_t {
8767 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
8768 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
8769 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
8770 -#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8771 -#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8772 +#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8773 +#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8774 +#define IS_TAGGED(inode) __IS_FLG(inode, MS_TAGGED)
8776 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
8777 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
8778 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
8779 +#define IS_IXUNLINK(inode) ((inode)->i_flags & S_IXUNLINK)
8780 +#define IS_IXORUNLINK(inode) ((IS_IXUNLINK(inode) ? S_IMMUTABLE : 0) ^ IS_IMMUTABLE(inode))
8781 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
8783 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
8784 @@ -202,6 +216,16 @@ struct inodes_stat_t {
8785 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
8786 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
8788 +#define IS_BARRIER(inode) (S_ISDIR((inode)->i_mode) && ((inode)->i_vflags & V_BARRIER))
8790 +#ifdef CONFIG_VSERVER_COWBL
8791 +# define IS_COW(inode) (IS_IXUNLINK(inode) && IS_IMMUTABLE(inode))
8792 +# define IS_COW_LINK(inode) (S_ISREG((inode)->i_mode) && ((inode)->i_nlink > 1))
8794 +# define IS_COW(inode) (0)
8795 +# define IS_COW_LINK(inode) (0)
8798 /* the read-only stuff doesn't really belong here, but any other place is
8799 probably as bad and I don't want to create yet another include file. */
8801 @@ -279,11 +303,14 @@ struct inodes_stat_t {
8802 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8803 #define FS_EXTENT_FL 0x00080000 /* Extents */
8804 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
8805 +#define FS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8806 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
8808 -#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8809 -#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8810 +#define FS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8811 +#define FS_COW_FL 0x20000000 /* Copy on Write marker */
8813 +#define FS_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8814 +#define FS_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8816 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
8817 #define SYNC_FILE_RANGE_WRITE 2
8818 @@ -365,6 +392,7 @@ typedef void (dio_iodone_t)(struct kiocb
8819 #define ATTR_KILL_PRIV (1 << 14)
8820 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
8821 #define ATTR_TIMES_SET (1 << 16)
8822 +#define ATTR_TAG (1 << 17)
8825 * This is the Inode Attributes structure, used for notify_change(). It
8826 @@ -380,6 +408,7 @@ struct iattr {
8832 struct timespec ia_atime;
8833 struct timespec ia_mtime;
8834 @@ -393,6 +422,9 @@ struct iattr {
8835 struct file *ia_file;
8838 +#define ATTR_FLAG_BARRIER 512 /* Barrier for chroot() */
8839 +#define ATTR_FLAG_IXUNLINK 1024 /* Immutable invert on unlink */
8842 * Includes for diskquotas.
8844 @@ -656,7 +688,9 @@ struct inode {
8845 unsigned int i_nlink;
8853 #ifdef __NEED_I_SIZE_ORDERED
8854 @@ -704,7 +738,8 @@ struct inode {
8855 unsigned long i_state;
8856 unsigned long dirtied_when; /* jiffies of first dirtying */
8858 - unsigned int i_flags;
8859 + unsigned short i_flags;
8860 + unsigned short i_vflags;
8862 atomic_t i_writecount;
8863 #ifdef CONFIG_SECURITY
8864 @@ -791,12 +826,12 @@ static inline void i_size_write(struct i
8866 static inline unsigned iminor(const struct inode *inode)
8868 - return MINOR(inode->i_rdev);
8869 + return MINOR(inode->i_mdev);
8872 static inline unsigned imajor(const struct inode *inode)
8874 - return MAJOR(inode->i_rdev);
8875 + return MAJOR(inode->i_mdev);
8878 extern struct block_device *I_BDEV(struct inode *inode);
8879 @@ -854,6 +889,7 @@ struct file {
8881 struct fown_struct f_owner;
8882 const struct cred *f_cred;
8884 struct file_ra_state f_ra;
8887 @@ -996,6 +1032,7 @@ struct file_lock {
8888 struct file *fl_file;
8893 struct fasync_struct * fl_fasync; /* for lease break notifications */
8894 unsigned long fl_break_time; /* for nonblocking lease breaks */
8895 @@ -1363,6 +1400,7 @@ struct inode_operations {
8897 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
8899 + int (*sync_flags) (struct inode *);
8903 @@ -1378,6 +1416,7 @@ extern ssize_t vfs_readv(struct file *,
8904 unsigned long, loff_t *);
8905 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
8906 unsigned long, loff_t *);
8907 +ssize_t vfs_sendfile(struct file *, struct file *, loff_t *, size_t, loff_t);
8909 struct super_operations {
8910 struct inode *(*alloc_inode)(struct super_block *sb);
8911 @@ -2096,6 +2135,7 @@ extern int dcache_dir_open(struct inode
8912 extern int dcache_dir_close(struct inode *, struct file *);
8913 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
8914 extern int dcache_readdir(struct file *, void *, filldir_t);
8915 +extern int dcache_readdir_filter(struct file *, void *, filldir_t, int (*)(struct dentry *));
8916 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
8917 extern int simple_statfs(struct dentry *, struct kstatfs *);
8918 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
8919 diff -NurpP --minimal linux-2.6.29.1/include/linux/if_tun.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/if_tun.h
8920 --- linux-2.6.29.1/include/linux/if_tun.h 2008-12-25 00:26:37.000000000 +0100
8921 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/if_tun.h 2009-02-22 22:54:26.000000000 +0100
8923 #define TUNSETOFFLOAD _IOW('T', 208, unsigned int)
8924 #define TUNSETTXFILTER _IOW('T', 209, unsigned int)
8925 #define TUNGETIFF _IOR('T', 210, unsigned int)
8926 +#define TUNSETNID _IOW('T', 215, int)
8928 /* TUNSETIFF ifr flags */
8929 #define IFF_TUN 0x0001
8930 diff -NurpP --minimal linux-2.6.29.1/include/linux/init_task.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/init_task.h
8931 --- linux-2.6.29.1/include/linux/init_task.h 2009-03-24 14:22:41.000000000 +0100
8932 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/init_task.h 2009-03-24 14:48:36.000000000 +0100
8933 @@ -184,6 +184,10 @@ extern struct cred init_cred;
8935 INIT_TRACE_IRQFLAGS \
8938 + .vx_info = NULL, \
8940 + .nx_info = NULL, \
8944 diff -NurpP --minimal linux-2.6.29.1/include/linux/interrupt.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/interrupt.h
8945 --- linux-2.6.29.1/include/linux/interrupt.h 2009-03-24 14:22:41.000000000 +0100
8946 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/interrupt.h 2009-03-24 14:48:36.000000000 +0100
8948 #include <linux/cpumask.h>
8949 #include <linux/irqreturn.h>
8950 #include <linux/irqnr.h>
8951 -#include <linux/hardirq.h>
8952 #include <linux/sched.h>
8953 +#include <linux/hardirq.h>
8954 #include <linux/irqflags.h>
8955 #include <linux/smp.h>
8956 #include <linux/percpu.h>
8957 diff -NurpP --minimal linux-2.6.29.1/include/linux/ipc.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ipc.h
8958 --- linux-2.6.29.1/include/linux/ipc.h 2008-12-25 00:26:37.000000000 +0100
8959 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/ipc.h 2009-02-22 22:54:26.000000000 +0100
8960 @@ -93,6 +93,7 @@ struct kern_ipc_perm
8968 diff -NurpP --minimal linux-2.6.29.1/include/linux/Kbuild linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/Kbuild
8969 --- linux-2.6.29.1/include/linux/Kbuild 2009-03-24 14:22:40.000000000 +0100
8970 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/Kbuild 2009-03-24 17:22:07.000000000 +0100
8971 @@ -373,5 +373,8 @@ unifdef-y += xattr.h
8974 objhdr-y += version.h
8976 +header-y += vserver/
8980 diff -NurpP --minimal linux-2.6.29.1/include/linux/loop.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/loop.h
8981 --- linux-2.6.29.1/include/linux/loop.h 2009-03-24 14:22:42.000000000 +0100
8982 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/loop.h 2009-03-24 14:48:36.000000000 +0100
8983 @@ -45,6 +45,7 @@ struct loop_device {
8984 struct loop_func_table *lo_encryption;
8986 uid_t lo_key_owner; /* Who set the key */
8988 int (*ioctl)(struct loop_device *, int cmd,
8991 diff -NurpP --minimal linux-2.6.29.1/include/linux/magic.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/magic.h
8992 --- linux-2.6.29.1/include/linux/magic.h 2009-03-24 14:22:42.000000000 +0100
8993 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/magic.h 2009-03-24 14:48:36.000000000 +0100
8996 #define ADFS_SUPER_MAGIC 0xadf5
8997 #define AFFS_SUPER_MAGIC 0xadff
8998 -#define AFS_SUPER_MAGIC 0x5346414F
8999 +#define AFS_SUPER_MAGIC 0x5346414F
9000 #define AUTOFS_SUPER_MAGIC 0x0187
9001 #define CODA_SUPER_MAGIC 0x73757245
9002 #define DEBUGFS_MAGIC 0x64626720
9004 #define NFS_SUPER_MAGIC 0x6969
9005 #define OPENPROM_SUPER_MAGIC 0x9fa1
9006 #define PROC_SUPER_MAGIC 0x9fa0
9007 +#define DEVPTS_SUPER_MAGIC 0x1cd1
9008 #define QNX4_SUPER_MAGIC 0x002f /* qnx4 fs detection */
9010 #define REISERFS_SUPER_MAGIC 0x52654973 /* used by gcc */
9011 diff -NurpP --minimal linux-2.6.29.1/include/linux/major.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/major.h
9012 --- linux-2.6.29.1/include/linux/major.h 2008-12-25 00:26:37.000000000 +0100
9013 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/major.h 2009-02-22 22:54:26.000000000 +0100
9015 #define HD_MAJOR IDE0_MAJOR
9016 #define PTY_SLAVE_MAJOR 3
9018 +#define VROOT_MAJOR 4
9019 #define TTYAUX_MAJOR 5
9022 diff -NurpP --minimal linux-2.6.29.1/include/linux/mm_types.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/mm_types.h
9023 --- linux-2.6.29.1/include/linux/mm_types.h 2009-03-24 14:22:42.000000000 +0100
9024 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/mm_types.h 2009-03-24 14:48:36.000000000 +0100
9025 @@ -232,6 +232,7 @@ struct mm_struct {
9027 /* Architecture-specific MM context */
9028 mm_context_t context;
9029 + struct vx_info *mm_vx_info;
9031 /* Swap token stuff */
9033 diff -NurpP --minimal linux-2.6.29.1/include/linux/mount.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/mount.h
9034 --- linux-2.6.29.1/include/linux/mount.h 2008-12-25 00:26:37.000000000 +0100
9035 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/mount.h 2009-02-22 22:54:26.000000000 +0100
9036 @@ -35,6 +35,9 @@ struct mnt_namespace;
9037 #define MNT_UNBINDABLE 0x2000 /* if the vfsmount is a unbindable mount */
9038 #define MNT_PNODE_MASK 0x3000 /* propagation flag mask */
9040 +#define MNT_TAGID 0x10000
9041 +#define MNT_NOTAG 0x20000
9044 struct list_head mnt_hash;
9045 struct vfsmount *mnt_parent; /* fs we are mounted on */
9046 @@ -69,6 +72,7 @@ struct vfsmount {
9047 * are held, and all mnt_writer[]s on this mount have 0 as their ->count
9049 atomic_t __mnt_writers;
9050 + tag_t mnt_tag; /* tagging used for vfsmount */
9053 static inline struct vfsmount *mntget(struct vfsmount *mnt)
9054 diff -NurpP --minimal linux-2.6.29.1/include/linux/net.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/net.h
9055 --- linux-2.6.29.1/include/linux/net.h 2008-12-25 00:26:37.000000000 +0100
9056 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/net.h 2009-02-22 22:54:26.000000000 +0100
9057 @@ -68,6 +68,7 @@ struct net;
9058 #define SOCK_NOSPACE 2
9059 #define SOCK_PASSCRED 3
9060 #define SOCK_PASSSEC 4
9061 +#define SOCK_USER_SOCKET 5
9063 #ifndef ARCH_HAS_SOCKET_TYPES
9065 diff -NurpP --minimal linux-2.6.29.1/include/linux/nfs_mount.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/nfs_mount.h
9066 --- linux-2.6.29.1/include/linux/nfs_mount.h 2009-03-24 14:22:43.000000000 +0100
9067 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/nfs_mount.h 2009-03-24 14:48:36.000000000 +0100
9068 @@ -63,7 +63,8 @@ struct nfs_mount_data {
9069 #define NFS_MOUNT_SECFLAVOUR 0x2000 /* 5 */
9070 #define NFS_MOUNT_NORDIRPLUS 0x4000 /* 5 */
9071 #define NFS_MOUNT_UNSHARED 0x8000 /* 5 */
9072 -#define NFS_MOUNT_FLAGMASK 0xFFFF
9073 +#define NFS_MOUNT_TAGGED 0x10000 /* context tagging */
9074 +#define NFS_MOUNT_FLAGMASK 0x1FFFF
9076 /* The following are for internal use only */
9077 #define NFS_MOUNT_LOOKUP_CACHE_NONEG 0x10000
9078 diff -NurpP --minimal linux-2.6.29.1/include/linux/nsproxy.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/nsproxy.h
9079 --- linux-2.6.29.1/include/linux/nsproxy.h 2009-03-24 14:22:43.000000000 +0100
9080 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/nsproxy.h 2009-03-24 14:48:36.000000000 +0100
9083 #include <linux/spinlock.h>
9084 #include <linux/sched.h>
9085 +#include <linux/vserver/debug.h>
9087 struct mnt_namespace;
9088 struct uts_namespace;
9089 @@ -62,22 +63,33 @@ static inline struct nsproxy *task_nspro
9092 int copy_namespaces(unsigned long flags, struct task_struct *tsk);
9093 +struct nsproxy *copy_nsproxy(struct nsproxy *orig);
9094 void exit_task_namespaces(struct task_struct *tsk);
9095 void switch_task_namespaces(struct task_struct *tsk, struct nsproxy *new);
9096 void free_nsproxy(struct nsproxy *ns);
9097 int unshare_nsproxy_namespaces(unsigned long, struct nsproxy **,
9098 struct fs_struct *);
9100 -static inline void put_nsproxy(struct nsproxy *ns)
9101 +#define get_nsproxy(n) __get_nsproxy(n, __FILE__, __LINE__)
9103 +static inline void __get_nsproxy(struct nsproxy *ns,
9104 + const char *_file, int _line)
9106 - if (atomic_dec_and_test(&ns->count)) {
9109 + vxlprintk(VXD_CBIT(space, 0), "get_nsproxy(%p[%u])",
9110 + ns, atomic_read(&ns->count), _file, _line);
9111 + atomic_inc(&ns->count);
9114 -static inline void get_nsproxy(struct nsproxy *ns)
9115 +#define put_nsproxy(n) __put_nsproxy(n, __FILE__, __LINE__)
9117 +static inline void __put_nsproxy(struct nsproxy *ns,
9118 + const char *_file, int _line)
9120 - atomic_inc(&ns->count);
9121 + vxlprintk(VXD_CBIT(space, 0), "put_nsproxy(%p[%u])",
9122 + ns, atomic_read(&ns->count), _file, _line);
9123 + if (atomic_dec_and_test(&ns->count)) {
9128 #ifdef CONFIG_CGROUP_NS
9129 diff -NurpP --minimal linux-2.6.29.1/include/linux/pid.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/pid.h
9130 --- linux-2.6.29.1/include/linux/pid.h 2009-03-24 14:22:43.000000000 +0100
9131 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/pid.h 2009-03-24 14:48:36.000000000 +0100
9132 @@ -8,7 +8,8 @@ enum pid_type
9142 @@ -160,6 +161,7 @@ static inline pid_t pid_nr(struct pid *p
9145 pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns);
9146 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns);
9147 pid_t pid_vnr(struct pid *pid);
9149 #define do_each_pid_task(pid, type, task) \
9150 diff -NurpP --minimal linux-2.6.29.1/include/linux/proc_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/proc_fs.h
9151 --- linux-2.6.29.1/include/linux/proc_fs.h 2008-12-25 00:26:37.000000000 +0100
9152 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/proc_fs.h 2009-02-22 22:54:26.000000000 +0100
9153 @@ -59,6 +59,7 @@ struct proc_dir_entry {
9159 const struct inode_operations *proc_iops;
9161 @@ -268,12 +269,18 @@ static inline void kclist_add(struct kco
9162 extern void kclist_add(struct kcore_list *, void *, size_t);
9169 int (*proc_get_link)(struct inode *, struct path *);
9170 int (*proc_read)(struct task_struct *task, char *page);
9171 int (*proc_show)(struct seq_file *m,
9172 struct pid_namespace *ns, struct pid *pid,
9173 struct task_struct *task);
9174 + int (*proc_vs_read)(char *page);
9175 + int (*proc_vxi_read)(struct vx_info *vxi, char *page);
9176 + int (*proc_nxi_read)(struct nx_info *nxi, char *page);
9179 struct ctl_table_header;
9180 @@ -281,6 +288,7 @@ struct ctl_table;
9187 struct proc_dir_entry *pde;
9188 diff -NurpP --minimal linux-2.6.29.1/include/linux/reiserfs_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/reiserfs_fs.h
9189 --- linux-2.6.29.1/include/linux/reiserfs_fs.h 2008-12-25 00:26:37.000000000 +0100
9190 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/reiserfs_fs.h 2009-02-22 22:54:26.000000000 +0100
9191 @@ -837,6 +837,11 @@ struct stat_data_v1 {
9192 #define REISERFS_COMPR_FL FS_COMPR_FL
9193 #define REISERFS_NOTAIL_FL FS_NOTAIL_FL
9195 +/* unfortunately reiserfs sdattr is only 16 bit */
9196 +#define REISERFS_IXUNLINK_FL (FS_IXUNLINK_FL >> 16)
9197 +#define REISERFS_BARRIER_FL (FS_BARRIER_FL >> 16)
9198 +#define REISERFS_COW_FL (FS_COW_FL >> 16)
9200 /* persistent flags that file inherits from the parent directory */
9201 #define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL | \
9202 REISERFS_SYNC_FL | \
9203 @@ -846,6 +851,9 @@ struct stat_data_v1 {
9204 REISERFS_COMPR_FL | \
9205 REISERFS_NOTAIL_FL )
9207 +#define REISERFS_FL_USER_VISIBLE 0x80FF
9208 +#define REISERFS_FL_USER_MODIFIABLE 0x80FF
9210 /* Stat Data on disk (reiserfs version of UFS disk inode minus the
9213 @@ -1911,6 +1919,7 @@ static inline void reiserfs_update_sd(st
9214 void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
9215 void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
9216 int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);
9217 +int reiserfs_sync_flags(struct inode *inode);
9220 void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
9221 diff -NurpP --minimal linux-2.6.29.1/include/linux/reiserfs_fs_sb.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/reiserfs_fs_sb.h
9222 --- linux-2.6.29.1/include/linux/reiserfs_fs_sb.h 2008-12-25 00:26:37.000000000 +0100
9223 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/reiserfs_fs_sb.h 2009-02-22 22:54:26.000000000 +0100
9224 @@ -456,6 +456,7 @@ enum reiserfs_mount_options {
9226 REISERFS_BARRIER_NONE,
9227 REISERFS_BARRIER_FLUSH,
9230 /* Actions on error */
9231 REISERFS_ERROR_PANIC,
9232 diff -NurpP --minimal linux-2.6.29.1/include/linux/sched.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sched.h
9233 --- linux-2.6.29.1/include/linux/sched.h 2009-03-24 14:22:43.000000000 +0100
9234 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sched.h 2009-04-08 15:47:01.000000000 +0200
9235 @@ -71,7 +71,6 @@ struct sched_param {
9236 #include <linux/fs_struct.h>
9237 #include <linux/compiler.h>
9238 #include <linux/completion.h>
9239 -#include <linux/pid.h>
9240 #include <linux/percpu.h>
9241 #include <linux/topology.h>
9242 #include <linux/proportions.h>
9243 @@ -88,6 +87,7 @@ struct sched_param {
9244 #include <linux/kobject.h>
9245 #include <linux/latencytop.h>
9246 #include <linux/cred.h>
9247 +#include <linux/pid.h>
9249 #include <asm/processor.h>
9251 @@ -176,12 +176,13 @@ extern unsigned long long time_sync_thre
9252 #define TASK_UNINTERRUPTIBLE 2
9253 #define __TASK_STOPPED 4
9254 #define __TASK_TRACED 8
9255 +#define TASK_ONHOLD 16
9256 /* in tsk->exit_state */
9257 -#define EXIT_ZOMBIE 16
9258 -#define EXIT_DEAD 32
9259 +#define EXIT_ZOMBIE 32
9260 +#define EXIT_DEAD 64
9261 /* in tsk->state again */
9262 -#define TASK_DEAD 64
9263 -#define TASK_WAKEKILL 128
9264 +#define TASK_DEAD 128
9265 +#define TASK_WAKEKILL 256
9267 /* Convenience macros for the sake of set_task_state */
9268 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
9269 @@ -358,25 +359,28 @@ extern void arch_unmap_area_topdown(stru
9270 * The mm counters are not protected by its page_table_lock,
9271 * so must be incremented atomically.
9273 -#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
9274 -#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
9275 -#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
9276 -#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
9277 -#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
9278 +#define __set_mm_counter(mm, member, value) \
9279 + atomic_long_set(&(mm)->_##member, value)
9280 +#define get_mm_counter(mm, member) \
9281 + ((unsigned long)atomic_long_read(&(mm)->_##member))
9283 #else /* !USE_SPLIT_PTLOCKS */
9285 * The mm counters are protected by its page_table_lock,
9286 * so can be incremented directly.
9288 -#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
9289 +#define __set_mm_counter(mm, member, value) (mm)->_##member = (value)
9290 #define get_mm_counter(mm, member) ((mm)->_##member)
9291 -#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
9292 -#define inc_mm_counter(mm, member) (mm)->_##member++
9293 -#define dec_mm_counter(mm, member) (mm)->_##member--
9295 #endif /* !USE_SPLIT_PTLOCKS */
9297 +#define set_mm_counter(mm, member, value) \
9298 + vx_ ## member ## pages_sub((mm), (get_mm_counter(mm, member) - value))
9299 +#define add_mm_counter(mm, member, value) \
9300 + vx_ ## member ## pages_add((mm), (value))
9301 +#define inc_mm_counter(mm, member) vx_ ## member ## pages_inc((mm))
9302 +#define dec_mm_counter(mm, member) vx_ ## member ## pages_dec((mm))
9304 #define get_mm_rss(mm) \
9305 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
9306 #define update_hiwater_rss(mm) do { \
9307 @@ -1131,7 +1135,9 @@ struct task_struct {
9308 const struct sched_class *sched_class;
9309 struct sched_entity se;
9310 struct sched_rt_entity rt;
9312 +#ifdef CONFIG_VSERVER_HARDCPU
9313 + struct list_head hq;
9315 #ifdef CONFIG_PREEMPT_NOTIFIERS
9316 /* list of struct preempt_notifier: */
9317 struct hlist_head preempt_notifiers;
9318 @@ -1287,6 +1293,14 @@ struct task_struct {
9322 +/* vserver context data */
9323 + struct vx_info *vx_info;
9324 + struct nx_info *nx_info;
9330 /* Thread group tracking */
9333 @@ -1499,6 +1513,11 @@ struct pid_namespace;
9334 * see also pid_nr() etc in include/linux/pid.h
9337 +#include <linux/vserver/base.h>
9338 +#include <linux/vserver/context.h>
9339 +#include <linux/vserver/debug.h>
9340 +#include <linux/vserver/pid.h>
9342 static inline pid_t task_pid_nr(struct task_struct *tsk)
9345 @@ -1508,7 +1527,7 @@ pid_t task_pid_nr_ns(struct task_struct
9347 static inline pid_t task_pid_vnr(struct task_struct *tsk)
9349 - return pid_vnr(task_pid(tsk));
9350 + return vx_map_pid(pid_vnr(task_pid(tsk)));
9354 @@ -1521,7 +1540,7 @@ pid_t task_tgid_nr_ns(struct task_struct
9356 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
9358 - return pid_vnr(task_tgid(tsk));
9359 + return vx_map_tgid(pid_vnr(task_tgid(tsk)));
9363 diff -NurpP --minimal linux-2.6.29.1/include/linux/shmem_fs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/shmem_fs.h
9364 --- linux-2.6.29.1/include/linux/shmem_fs.h 2008-12-25 00:26:37.000000000 +0100
9365 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/shmem_fs.h 2009-02-22 22:54:26.000000000 +0100
9368 #define SHMEM_NR_DIRECT 16
9370 +#define TMPFS_SUPER_MAGIC 0x01021994
9373 struct shmem_inode_info {
9375 unsigned long flags;
9376 diff -NurpP --minimal linux-2.6.29.1/include/linux/stat.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/stat.h
9377 --- linux-2.6.29.1/include/linux/stat.h 2008-12-25 00:26:37.000000000 +0100
9378 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/stat.h 2009-02-22 22:54:26.000000000 +0100
9379 @@ -66,6 +66,7 @@ struct kstat {
9386 struct timespec atime;
9387 diff -NurpP --minimal linux-2.6.29.1/include/linux/sunrpc/auth.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sunrpc/auth.h
9388 --- linux-2.6.29.1/include/linux/sunrpc/auth.h 2008-12-25 00:26:37.000000000 +0100
9389 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sunrpc/auth.h 2009-02-22 22:54:26.000000000 +0100
9395 struct group_info *group_info;
9396 unsigned char machine_cred : 1;
9398 diff -NurpP --minimal linux-2.6.29.1/include/linux/sunrpc/clnt.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sunrpc/clnt.h
9399 --- linux-2.6.29.1/include/linux/sunrpc/clnt.h 2009-03-24 14:22:43.000000000 +0100
9400 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sunrpc/clnt.h 2009-03-24 14:48:36.000000000 +0100
9401 @@ -43,7 +43,8 @@ struct rpc_clnt {
9402 unsigned int cl_softrtry : 1,/* soft timeouts */
9403 cl_discrtry : 1,/* disconnect before retry */
9404 cl_autobind : 1,/* use getport() */
9405 - cl_chatty : 1;/* be verbose */
9406 + cl_chatty : 1,/* be verbose */
9407 + cl_tag : 1;/* context tagging */
9409 struct rpc_rtt * cl_rtt; /* RTO estimator data */
9410 const struct rpc_timeout *cl_timeout; /* Timeout strategy */
9411 diff -NurpP --minimal linux-2.6.29.1/include/linux/syscalls.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/syscalls.h
9412 --- linux-2.6.29.1/include/linux/syscalls.h 2009-03-24 14:22:43.000000000 +0100
9413 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/syscalls.h 2009-03-24 14:48:36.000000000 +0100
9414 @@ -369,6 +369,8 @@ asmlinkage long sys_symlink(const char _
9415 asmlinkage long sys_unlink(const char __user *pathname);
9416 asmlinkage long sys_rename(const char __user *oldname,
9417 const char __user *newname);
9418 +asmlinkage long sys_copyfile(const char __user *from, const char __user *to,
9420 asmlinkage long sys_chmod(const char __user *filename, mode_t mode);
9421 asmlinkage long sys_fchmod(unsigned int fd, mode_t mode);
9423 diff -NurpP --minimal linux-2.6.29.1/include/linux/sysctl.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sysctl.h
9424 --- linux-2.6.29.1/include/linux/sysctl.h 2008-12-25 00:26:37.000000000 +0100
9425 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sysctl.h 2009-02-22 22:54:26.000000000 +0100
9426 @@ -70,6 +70,7 @@ enum
9427 CTL_ABI=9, /* Binary emulation */
9428 CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
9429 CTL_ARLAN=254, /* arlan wireless driver */
9430 + CTL_VSERVER=4242, /* Linux-VServer debug */
9431 CTL_S390DBF=5677, /* s390 debug */
9432 CTL_SUNRPC=7249, /* sunrpc debug */
9433 CTL_PM=9899, /* frv power management */
9434 @@ -104,6 +105,7 @@ enum
9436 KERN_PANIC=15, /* int: panic timeout */
9437 KERN_REALROOTDEV=16, /* real root device to mount after initrd */
9438 + KERN_VSHELPER=17, /* string: path to vshelper policy agent */
9440 KERN_SPARC_REBOOT=21, /* reboot command on Sparc */
9441 KERN_CTLALTDEL=22, /* int: allow ctl-alt-del to reboot */
9442 diff -NurpP --minimal linux-2.6.29.1/include/linux/sysfs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sysfs.h
9443 --- linux-2.6.29.1/include/linux/sysfs.h 2008-12-25 00:26:37.000000000 +0100
9444 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/sysfs.h 2009-02-22 22:54:26.000000000 +0100
9446 #include <linux/list.h>
9447 #include <asm/atomic.h>
9449 +#define SYSFS_SUPER_MAGIC 0x62656572
9454 diff -NurpP --minimal linux-2.6.29.1/include/linux/time.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/time.h
9455 --- linux-2.6.29.1/include/linux/time.h 2009-03-24 14:22:43.000000000 +0100
9456 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/time.h 2009-02-22 22:54:26.000000000 +0100
9457 @@ -190,6 +190,9 @@ static __always_inline void timespec_add
9458 a->tv_sec += __iter_div_u64_rem(a->tv_nsec + ns, NSEC_PER_SEC, &ns);
9462 +#include <linux/vs_time.h>
9464 #endif /* __KERNEL__ */
9466 #define NFDBITS __NFDBITS
9467 diff -NurpP --minimal linux-2.6.29.1/include/linux/types.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/types.h
9468 --- linux-2.6.29.1/include/linux/types.h 2009-03-24 14:22:43.000000000 +0100
9469 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/types.h 2009-03-24 14:48:36.000000000 +0100
9470 @@ -36,6 +36,9 @@ typedef __kernel_uid32_t uid_t;
9471 typedef __kernel_gid32_t gid_t;
9472 typedef __kernel_uid16_t uid16_t;
9473 typedef __kernel_gid16_t gid16_t;
9474 +typedef unsigned int xid_t;
9475 +typedef unsigned int nid_t;
9476 +typedef unsigned int tag_t;
9478 typedef unsigned long uintptr_t;
9480 diff -NurpP --minimal linux-2.6.29.1/include/linux/vroot.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vroot.h
9481 --- linux-2.6.29.1/include/linux/vroot.h 1970-01-01 01:00:00.000000000 +0100
9482 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vroot.h 2009-02-22 22:54:26.000000000 +0100
9486 + * include/linux/vroot.h
9488 + * written by Herbert Pötzl, 9/11/2002
9489 + * ported to 2.6 by Herbert Pötzl, 30/12/2004
9491 + * Copyright (C) 2002-2007 by Herbert Pötzl.
9492 + * Redistribution of this file is permitted under the
9493 + * GNU General Public License.
9496 +#ifndef _LINUX_VROOT_H
9497 +#define _LINUX_VROOT_H
9502 +/* Possible states of device */
9508 +struct vroot_device {
9512 + struct semaphore vr_ctl_mutex;
9513 + struct block_device *vr_device;
9518 +typedef struct block_device *(vroot_grb_func)(struct block_device *);
9520 +extern int register_vroot_grb(vroot_grb_func *);
9521 +extern int unregister_vroot_grb(vroot_grb_func *);
9523 +#endif /* __KERNEL__ */
9525 +#define MAX_VROOT_DEFAULT 8
9528 + * IOCTL commands --- we will commandeer 0x56 ('V')
9531 +#define VROOT_SET_DEV 0x5600
9532 +#define VROOT_CLR_DEV 0x5601
9534 +#endif /* _LINUX_VROOT_H */
9535 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_base.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_base.h
9536 --- linux-2.6.29.1/include/linux/vs_base.h 1970-01-01 01:00:00.000000000 +0100
9537 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_base.h 2009-02-22 22:54:26.000000000 +0100
9542 +#include "vserver/base.h"
9543 +#include "vserver/check.h"
9544 +#include "vserver/debug.h"
9547 +#warning duplicate inclusion
9549 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_context.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_context.h
9550 --- linux-2.6.29.1/include/linux/vs_context.h 1970-01-01 01:00:00.000000000 +0100
9551 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_context.h 2009-02-22 22:54:26.000000000 +0100
9553 +#ifndef _VS_CONTEXT_H
9554 +#define _VS_CONTEXT_H
9556 +#include "vserver/base.h"
9557 +#include "vserver/check.h"
9558 +#include "vserver/context.h"
9559 +#include "vserver/history.h"
9560 +#include "vserver/debug.h"
9562 +#include <linux/sched.h>
9565 +#define get_vx_info(i) __get_vx_info(i, __FILE__, __LINE__, __HERE__)
9567 +static inline struct vx_info *__get_vx_info(struct vx_info *vxi,
9568 + const char *_file, int _line, void *_here)
9573 + vxlprintk(VXD_CBIT(xid, 2), "get_vx_info(%p[#%d.%d])",
9574 + vxi, vxi ? vxi->vx_id : 0,
9575 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9577 + __vxh_get_vx_info(vxi, _here);
9579 + atomic_inc(&vxi->vx_usecnt);
9584 +extern void free_vx_info(struct vx_info *);
9586 +#define put_vx_info(i) __put_vx_info(i, __FILE__, __LINE__, __HERE__)
9588 +static inline void __put_vx_info(struct vx_info *vxi,
9589 + const char *_file, int _line, void *_here)
9594 + vxlprintk(VXD_CBIT(xid, 2), "put_vx_info(%p[#%d.%d])",
9595 + vxi, vxi ? vxi->vx_id : 0,
9596 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9598 + __vxh_put_vx_info(vxi, _here);
9600 + if (atomic_dec_and_test(&vxi->vx_usecnt))
9601 + free_vx_info(vxi);
9605 +#define init_vx_info(p, i) \
9606 + __init_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9608 +static inline void __init_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9609 + const char *_file, int _line, void *_here)
9612 + vxlprintk(VXD_CBIT(xid, 3),
9613 + "init_vx_info(%p[#%d.%d])",
9614 + vxi, vxi ? vxi->vx_id : 0,
9615 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9617 + __vxh_init_vx_info(vxi, vxp, _here);
9619 + atomic_inc(&vxi->vx_usecnt);
9625 +#define set_vx_info(p, i) \
9626 + __set_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9628 +static inline void __set_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9629 + const char *_file, int _line, void *_here)
9631 + struct vx_info *vxo;
9636 + vxlprintk(VXD_CBIT(xid, 3), "set_vx_info(%p[#%d.%d])",
9637 + vxi, vxi ? vxi->vx_id : 0,
9638 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9640 + __vxh_set_vx_info(vxi, vxp, _here);
9642 + atomic_inc(&vxi->vx_usecnt);
9643 + vxo = xchg(vxp, vxi);
9648 +#define clr_vx_info(p) __clr_vx_info(p, __FILE__, __LINE__, __HERE__)
9650 +static inline void __clr_vx_info(struct vx_info **vxp,
9651 + const char *_file, int _line, void *_here)
9653 + struct vx_info *vxo;
9655 + vxo = xchg(vxp, NULL);
9659 + vxlprintk(VXD_CBIT(xid, 3), "clr_vx_info(%p[#%d.%d])",
9660 + vxo, vxo ? vxo->vx_id : 0,
9661 + vxo ? atomic_read(&vxo->vx_usecnt) : 0,
9663 + __vxh_clr_vx_info(vxo, vxp, _here);
9665 + if (atomic_dec_and_test(&vxo->vx_usecnt))
9666 + free_vx_info(vxo);
9670 +#define claim_vx_info(v, p) \
9671 + __claim_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9673 +static inline void __claim_vx_info(struct vx_info *vxi,
9674 + struct task_struct *task,
9675 + const char *_file, int _line, void *_here)
9677 + vxlprintk(VXD_CBIT(xid, 3), "claim_vx_info(%p[#%d.%d.%d]) %p",
9678 + vxi, vxi ? vxi->vx_id : 0,
9679 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9680 + vxi ? atomic_read(&vxi->vx_tasks) : 0,
9681 + task, _file, _line);
9682 + __vxh_claim_vx_info(vxi, task, _here);
9684 + atomic_inc(&vxi->vx_tasks);
9688 +extern void unhash_vx_info(struct vx_info *);
9690 +#define release_vx_info(v, p) \
9691 + __release_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9693 +static inline void __release_vx_info(struct vx_info *vxi,
9694 + struct task_struct *task,
9695 + const char *_file, int _line, void *_here)
9697 + vxlprintk(VXD_CBIT(xid, 3), "release_vx_info(%p[#%d.%d.%d]) %p",
9698 + vxi, vxi ? vxi->vx_id : 0,
9699 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9700 + vxi ? atomic_read(&vxi->vx_tasks) : 0,
9701 + task, _file, _line);
9702 + __vxh_release_vx_info(vxi, task, _here);
9706 + if (atomic_dec_and_test(&vxi->vx_tasks))
9707 + unhash_vx_info(vxi);
9711 +#define task_get_vx_info(p) \
9712 + __task_get_vx_info(p, __FILE__, __LINE__, __HERE__)
9714 +static inline struct vx_info *__task_get_vx_info(struct task_struct *p,
9715 + const char *_file, int _line, void *_here)
9717 + struct vx_info *vxi;
9720 + vxlprintk(VXD_CBIT(xid, 5), "task_get_vx_info(%p)",
9722 + vxi = __get_vx_info(p->vx_info, _file, _line, _here);
9728 +static inline void __wakeup_vx_info(struct vx_info *vxi)
9730 + if (waitqueue_active(&vxi->vx_wait))
9731 + wake_up_interruptible(&vxi->vx_wait);
9735 +#define enter_vx_info(v, s) __enter_vx_info(v, s, __FILE__, __LINE__)
9737 +static inline void __enter_vx_info(struct vx_info *vxi,
9738 + struct vx_info_save *vxis, const char *_file, int _line)
9740 + vxlprintk(VXD_CBIT(xid, 5), "enter_vx_info(%p[#%d],%p) %p[#%d,%p]",
9741 + vxi, vxi ? vxi->vx_id : 0, vxis, current,
9742 + current->xid, current->vx_info, _file, _line);
9743 + vxis->vxi = xchg(¤t->vx_info, vxi);
9744 + vxis->xid = current->xid;
9745 + current->xid = vxi ? vxi->vx_id : 0;
9748 +#define leave_vx_info(s) __leave_vx_info(s, __FILE__, __LINE__)
9750 +static inline void __leave_vx_info(struct vx_info_save *vxis,
9751 + const char *_file, int _line)
9753 + vxlprintk(VXD_CBIT(xid, 5), "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]",
9754 + vxis, vxis->xid, vxis->vxi, current,
9755 + current->xid, current->vx_info, _file, _line);
9756 + (void)xchg(¤t->vx_info, vxis->vxi);
9757 + current->xid = vxis->xid;
9761 +static inline void __enter_vx_admin(struct vx_info_save *vxis)
9763 + vxis->vxi = xchg(¤t->vx_info, NULL);
9764 + vxis->xid = xchg(¤t->xid, (xid_t)0);
9767 +static inline void __leave_vx_admin(struct vx_info_save *vxis)
9769 + (void)xchg(¤t->xid, vxis->xid);
9770 + (void)xchg(¤t->vx_info, vxis->vxi);
9773 +extern void exit_vx_info(struct task_struct *, int);
9774 +extern void exit_vx_info_early(struct task_struct *, int);
9778 +#warning duplicate inclusion
9780 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_cowbl.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_cowbl.h
9781 --- linux-2.6.29.1/include/linux/vs_cowbl.h 1970-01-01 01:00:00.000000000 +0100
9782 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_cowbl.h 2009-02-22 22:54:26.000000000 +0100
9784 +#ifndef _VS_COWBL_H
9785 +#define _VS_COWBL_H
9787 +#include <linux/fs.h>
9788 +#include <linux/dcache.h>
9789 +#include <linux/namei.h>
9791 +extern struct dentry *cow_break_link(const char *pathname);
9793 +static inline int cow_check_and_break(struct path *path)
9795 + struct inode *inode = path->dentry->d_inode;
9798 + /* do we need this check? */
9799 + if (IS_RDONLY(inode))
9802 + if (IS_COW(inode)) {
9803 + if (IS_COW_LINK(inode)) {
9804 + struct dentry *new_dentry, *old_dentry = path->dentry;
9807 + buf = kmalloc(PATH_MAX, GFP_KERNEL);
9811 + pp = d_path(path, buf, PATH_MAX);
9812 + new_dentry = cow_break_link(pp);
9814 + if (!IS_ERR(new_dentry)) {
9815 + path->dentry = new_dentry;
9818 + error = PTR_ERR(new_dentry);
9820 + inode->i_flags &= ~(S_IXUNLINK | S_IMMUTABLE);
9821 + inode->i_ctime = CURRENT_TIME;
9822 + mark_inode_dirty(inode);
9829 +#warning duplicate inclusion
9831 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_cvirt.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_cvirt.h
9832 --- linux-2.6.29.1/include/linux/vs_cvirt.h 1970-01-01 01:00:00.000000000 +0100
9833 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_cvirt.h 2009-02-22 22:54:26.000000000 +0100
9835 +#ifndef _VS_CVIRT_H
9836 +#define _VS_CVIRT_H
9838 +#include "vserver/cvirt.h"
9839 +#include "vserver/context.h"
9840 +#include "vserver/base.h"
9841 +#include "vserver/check.h"
9842 +#include "vserver/debug.h"
9845 +static inline void vx_activate_task(struct task_struct *p)
9847 + struct vx_info *vxi;
9849 + if ((vxi = p->vx_info)) {
9850 + vx_update_load(vxi);
9851 + atomic_inc(&vxi->cvirt.nr_running);
9855 +static inline void vx_deactivate_task(struct task_struct *p)
9857 + struct vx_info *vxi;
9859 + if ((vxi = p->vx_info)) {
9860 + vx_update_load(vxi);
9861 + atomic_dec(&vxi->cvirt.nr_running);
9865 +static inline void vx_uninterruptible_inc(struct task_struct *p)
9867 + struct vx_info *vxi;
9869 + if ((vxi = p->vx_info))
9870 + atomic_inc(&vxi->cvirt.nr_uninterruptible);
9873 +static inline void vx_uninterruptible_dec(struct task_struct *p)
9875 + struct vx_info *vxi;
9877 + if ((vxi = p->vx_info))
9878 + atomic_dec(&vxi->cvirt.nr_uninterruptible);
9883 +#warning duplicate inclusion
9885 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_device.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_device.h
9886 --- linux-2.6.29.1/include/linux/vs_device.h 1970-01-01 01:00:00.000000000 +0100
9887 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_device.h 2009-02-22 22:54:26.000000000 +0100
9889 +#ifndef _VS_DEVICE_H
9890 +#define _VS_DEVICE_H
9892 +#include "vserver/base.h"
9893 +#include "vserver/device.h"
9894 +#include "vserver/debug.h"
9897 +#ifdef CONFIG_VSERVER_DEVICE
9899 +int vs_map_device(struct vx_info *, dev_t, dev_t *, umode_t);
9901 +#define vs_device_perm(v, d, m, p) \
9902 + ((vs_map_device(current_vx_info(), d, NULL, m) & (p)) == (p))
9907 +int vs_map_device(struct vx_info *vxi,
9908 + dev_t device, dev_t *target, umode_t mode)
9915 +#define vs_device_perm(v, d, m, p) ((p) == (p))
9920 +#define vs_map_chrdev(d, t, p) \
9921 + ((vs_map_device(current_vx_info(), d, t, S_IFCHR) & (p)) == (p))
9922 +#define vs_map_blkdev(d, t, p) \
9923 + ((vs_map_device(current_vx_info(), d, t, S_IFBLK) & (p)) == (p))
9925 +#define vs_chrdev_perm(d, p) \
9926 + vs_device_perm(current_vx_info(), d, S_IFCHR, p)
9927 +#define vs_blkdev_perm(d, p) \
9928 + vs_device_perm(current_vx_info(), d, S_IFBLK, p)
9932 +#warning duplicate inclusion
9934 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_dlimit.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_dlimit.h
9935 --- linux-2.6.29.1/include/linux/vs_dlimit.h 1970-01-01 01:00:00.000000000 +0100
9936 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_dlimit.h 2009-02-22 22:54:26.000000000 +0100
9938 +#ifndef _VS_DLIMIT_H
9939 +#define _VS_DLIMIT_H
9941 +#include <linux/fs.h>
9943 +#include "vserver/dlimit.h"
9944 +#include "vserver/base.h"
9945 +#include "vserver/debug.h"
9948 +#define get_dl_info(i) __get_dl_info(i, __FILE__, __LINE__)
9950 +static inline struct dl_info *__get_dl_info(struct dl_info *dli,
9951 + const char *_file, int _line)
9955 + vxlprintk(VXD_CBIT(dlim, 4), "get_dl_info(%p[#%d.%d])",
9956 + dli, dli ? dli->dl_tag : 0,
9957 + dli ? atomic_read(&dli->dl_usecnt) : 0,
9959 + atomic_inc(&dli->dl_usecnt);
9964 +#define free_dl_info(i) \
9965 + call_rcu(&(i)->dl_rcu, rcu_free_dl_info)
9967 +#define put_dl_info(i) __put_dl_info(i, __FILE__, __LINE__)
9969 +static inline void __put_dl_info(struct dl_info *dli,
9970 + const char *_file, int _line)
9974 + vxlprintk(VXD_CBIT(dlim, 4), "put_dl_info(%p[#%d.%d])",
9975 + dli, dli ? dli->dl_tag : 0,
9976 + dli ? atomic_read(&dli->dl_usecnt) : 0,
9978 + if (atomic_dec_and_test(&dli->dl_usecnt))
9979 + free_dl_info(dli);
9983 +#define __dlimit_char(d) ((d) ? '*' : ' ')
9985 +static inline int __dl_alloc_space(struct super_block *sb,
9986 + tag_t tag, dlsize_t nr, const char *file, int line)
9988 + struct dl_info *dli = NULL;
9993 + dli = locate_dl_info(sb, tag);
9997 + spin_lock(&dli->dl_lock);
9998 + ret = (dli->dl_space_used + nr > dli->dl_space_total);
10000 + dli->dl_space_used += nr;
10001 + spin_unlock(&dli->dl_lock);
10002 + put_dl_info(dli);
10004 + vxlprintk(VXD_CBIT(dlim, 1),
10005 + "ALLOC (%p,#%d)%c %lld bytes (%d)",
10006 + sb, tag, __dlimit_char(dli), (long long)nr,
10007 + ret, file, line);
10011 +static inline void __dl_free_space(struct super_block *sb,
10012 + tag_t tag, dlsize_t nr, const char *_file, int _line)
10014 + struct dl_info *dli = NULL;
10018 + dli = locate_dl_info(sb, tag);
10022 + spin_lock(&dli->dl_lock);
10023 + if (dli->dl_space_used > nr)
10024 + dli->dl_space_used -= nr;
10026 + dli->dl_space_used = 0;
10027 + spin_unlock(&dli->dl_lock);
10028 + put_dl_info(dli);
10030 + vxlprintk(VXD_CBIT(dlim, 1),
10031 + "FREE (%p,#%d)%c %lld bytes",
10032 + sb, tag, __dlimit_char(dli), (long long)nr,
10036 +static inline int __dl_alloc_inode(struct super_block *sb,
10037 + tag_t tag, const char *_file, int _line)
10039 + struct dl_info *dli;
10042 + dli = locate_dl_info(sb, tag);
10046 + spin_lock(&dli->dl_lock);
10047 + ret = (dli->dl_inodes_used >= dli->dl_inodes_total);
10049 + dli->dl_inodes_used++;
10050 + spin_unlock(&dli->dl_lock);
10051 + put_dl_info(dli);
10053 + vxlprintk(VXD_CBIT(dlim, 0),
10054 + "ALLOC (%p,#%d)%c inode (%d)",
10055 + sb, tag, __dlimit_char(dli), ret, _file, _line);
10059 +static inline void __dl_free_inode(struct super_block *sb,
10060 + tag_t tag, const char *_file, int _line)
10062 + struct dl_info *dli;
10064 + dli = locate_dl_info(sb, tag);
10068 + spin_lock(&dli->dl_lock);
10069 + if (dli->dl_inodes_used > 1)
10070 + dli->dl_inodes_used--;
10072 + dli->dl_inodes_used = 0;
10073 + spin_unlock(&dli->dl_lock);
10074 + put_dl_info(dli);
10076 + vxlprintk(VXD_CBIT(dlim, 0),
10077 + "FREE (%p,#%d)%c inode",
10078 + sb, tag, __dlimit_char(dli), _file, _line);
10081 +static inline void __dl_adjust_block(struct super_block *sb, tag_t tag,
10082 + unsigned long long *free_blocks, unsigned long long *root_blocks,
10083 + const char *_file, int _line)
10085 + struct dl_info *dli;
10086 + uint64_t broot, bfree;
10088 + dli = locate_dl_info(sb, tag);
10092 + spin_lock(&dli->dl_lock);
10093 + broot = (dli->dl_space_total -
10094 + (dli->dl_space_total >> 10) * dli->dl_nrlmult)
10095 + >> sb->s_blocksize_bits;
10096 + bfree = (dli->dl_space_total - dli->dl_space_used)
10097 + >> sb->s_blocksize_bits;
10098 + spin_unlock(&dli->dl_lock);
10100 + vxlprintk(VXD_CBIT(dlim, 2),
10101 + "ADJUST: %lld,%lld on %lld,%lld [mult=%d]",
10102 + (long long)bfree, (long long)broot,
10103 + *free_blocks, *root_blocks, dli->dl_nrlmult,
10105 + if (free_blocks) {
10106 + if (*free_blocks > bfree)
10107 + *free_blocks = bfree;
10109 + if (root_blocks) {
10110 + if (*root_blocks > broot)
10111 + *root_blocks = broot;
10113 + put_dl_info(dli);
10116 +#define DLIMIT_ALLOC_SPACE(in, bytes) \
10117 + __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10118 + __FILE__, __LINE__ )
10120 +#define DLIMIT_FREE_SPACE(in, bytes) \
10121 + __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10122 + __FILE__, __LINE__ )
10124 +#define DLIMIT_ALLOC_BLOCK(in, nr) \
10125 + __dl_alloc_space((in)->i_sb, (in)->i_tag, \
10126 + ((dlsize_t)(nr)) << (in)->i_sb->s_blocksize_bits, \
10127 + __FILE__, __LINE__ )
10129 +#define DLIMIT_FREE_BLOCK(in, nr) \
10130 + __dl_free_space((in)->i_sb, (in)->i_tag, \
10131 + ((dlsize_t)(nr)) << (in)->i_sb->s_blocksize_bits, \
10132 + __FILE__, __LINE__ )
10135 +#define DLIMIT_ALLOC_INODE(in) \
10136 + __dl_alloc_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10138 +#define DLIMIT_FREE_INODE(in) \
10139 + __dl_free_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10142 +#define DLIMIT_ADJUST_BLOCK(sb, tag, fb, rb) \
10143 + __dl_adjust_block(sb, tag, fb, rb, __FILE__, __LINE__ )
10147 +#warning duplicate inclusion
10149 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/base.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/base.h
10150 --- linux-2.6.29.1/include/linux/vserver/base.h 1970-01-01 01:00:00.000000000 +0100
10151 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/base.h 2009-03-25 00:36:15.000000000 +0100
10153 +#ifndef _VX_BASE_H
10154 +#define _VX_BASE_H
10157 +/* context state changes */
10169 +#define vx_task_xid(t) ((t)->xid)
10171 +#define vx_current_xid() vx_task_xid(current)
10173 +#define current_vx_info() (current->vx_info)
10176 +#define nx_task_nid(t) ((t)->nid)
10178 +#define nx_current_nid() nx_task_nid(current)
10180 +#define current_nx_info() (current->nx_info)
10183 +/* generic flag merging */
10185 +#define vs_check_flags(v, m, f) (((v) & (m)) ^ (f))
10187 +#define vs_mask_flags(v, f, m) (((v) & ~(m)) | ((f) & (m)))
10189 +#define vs_mask_mask(v, f, m) (((v) & ~(m)) | ((v) & (f) & (m)))
10191 +#define vs_check_bit(v, n) ((v) & (1LL << (n)))
10194 +/* context flags */
10196 +#define __vx_flags(v) ((v) ? (v)->vx_flags : 0)
10198 +#define vx_current_flags() __vx_flags(current->vx_info)
10200 +#define vx_info_flags(v, m, f) \
10201 + vs_check_flags(__vx_flags(v), m, f)
10203 +#define task_vx_flags(t, m, f) \
10204 + ((t) && vx_info_flags((t)->vx_info, m, f))
10206 +#define vx_flags(m, f) vx_info_flags(current->vx_info, m, f)
10209 +/* context caps */
10211 +#define __vx_ccaps(v) ((v) ? (v)->vx_ccaps : 0)
10213 +#define vx_current_ccaps() __vx_ccaps(current->vx_info)
10215 +#define vx_info_ccaps(v, c) (__vx_ccaps(v) & (c))
10217 +#define vx_ccaps(c) vx_info_ccaps(current->vx_info, (c))
10221 +/* network flags */
10223 +#define __nx_flags(n) ((n) ? (n)->nx_flags : 0)
10225 +#define nx_current_flags() __nx_flags(current->nx_info)
10227 +#define nx_info_flags(n, m, f) \
10228 + vs_check_flags(__nx_flags(n), m, f)
10230 +#define task_nx_flags(t, m, f) \
10231 + ((t) && nx_info_flags((t)->nx_info, m, f))
10233 +#define nx_flags(m, f) nx_info_flags(current->nx_info, m, f)
10236 +/* network caps */
10238 +#define __nx_ncaps(n) ((n) ? (n)->nx_ncaps : 0)
10240 +#define nx_current_ncaps() __nx_ncaps(current->nx_info)
10242 +#define nx_info_ncaps(n, c) (__nx_ncaps(n) & (c))
10244 +#define nx_ncaps(c) nx_info_ncaps(current->nx_info, c)
10247 +/* context mask capabilities */
10249 +#define __vx_mcaps(v) ((v) ? (v)->vx_ccaps >> 32UL : ~0 )
10251 +#define vx_info_mcaps(v, c) (__vx_mcaps(v) & (c))
10253 +#define vx_mcaps(c) vx_info_mcaps(current->vx_info, c)
10256 +/* context bcap mask */
10258 +#define __vx_bcaps(v) ((v)->vx_bcaps)
10260 +#define vx_current_bcaps() __vx_bcaps(current->vx_info)
10263 +/* mask given bcaps */
10265 +#define vx_info_mbcaps(v, c) ((v) ? cap_intersect(__vx_bcaps(v), c) : c)
10267 +#define vx_mbcaps(c) vx_info_mbcaps(current->vx_info, c)
10270 +/* masked cap_bset */
10272 +#define vx_info_cap_bset(v) vx_info_mbcaps(v, current->cap_bset)
10274 +#define vx_current_cap_bset() vx_info_cap_bset(current->vx_info)
10277 +#define vx_info_mbcap(v, b) \
10278 + (!vx_info_flags(v, VXF_STATE_SETUP, 0) ? \
10279 + vx_info_bcaps(v, b) : (b))
10281 +#define task_vx_mbcap(t, b) \
10282 + vx_info_mbcap((t)->vx_info, (t)->b)
10284 +#define vx_mbcap(b) task_vx_mbcap(current, b)
10287 +#define vx_cap_raised(v, c, f) cap_raised(vx_info_mbcaps(v, c), f)
10289 +#define vx_capable(b, c) (capable(b) || \
10290 + (cap_raised(current_cap(), b) && vx_ccaps(c)))
10292 +#define nx_capable(b, c) (capable(b) || \
10293 + (cap_raised(current_cap(), b) && nx_ncaps(c)))
10295 +#define vx_current_initpid(n) \
10296 + (current->vx_info && \
10297 + (current->vx_info->vx_initpid == (n)))
10300 +#define __vx_state(v) ((v) ? ((v)->vx_state) : 0)
10302 +#define vx_info_state(v, m) (__vx_state(v) & (m))
10305 +#define __nx_state(n) ((n) ? ((n)->nx_state) : 0)
10307 +#define nx_info_state(n, m) (__nx_state(n) & (m))
10310 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cacct_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_cmd.h
10311 --- linux-2.6.29.1/include/linux/vserver/cacct_cmd.h 1970-01-01 01:00:00.000000000 +0100
10312 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_cmd.h 2009-02-22 22:54:26.000000000 +0100
10314 +#ifndef _VX_CACCT_CMD_H
10315 +#define _VX_CACCT_CMD_H
10318 +/* virtual host info name commands */
10320 +#define VCMD_sock_stat VC_CMD(VSTAT, 5, 0)
10322 +struct vcmd_sock_stat_v0 {
10324 + uint32_t count[3];
10325 + uint64_t total[3];
10331 +#include <linux/compiler.h>
10333 +extern int vc_sock_stat(struct vx_info *, void __user *);
10335 +#endif /* __KERNEL__ */
10336 +#endif /* _VX_CACCT_CMD_H */
10337 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cacct_def.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_def.h
10338 --- linux-2.6.29.1/include/linux/vserver/cacct_def.h 1970-01-01 01:00:00.000000000 +0100
10339 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_def.h 2009-02-22 22:54:26.000000000 +0100
10341 +#ifndef _VX_CACCT_DEF_H
10342 +#define _VX_CACCT_DEF_H
10344 +#include <asm/atomic.h>
10345 +#include <linux/vserver/cacct.h>
10348 +struct _vx_sock_acc {
10349 + atomic_long_t count;
10350 + atomic_long_t total;
10353 +/* context sub struct */
10355 +struct _vx_cacct {
10356 + struct _vx_sock_acc sock[VXA_SOCK_SIZE][3];
10357 + atomic_t slab[8];
10358 + atomic_t page[6][8];
10361 +#ifdef CONFIG_VSERVER_DEBUG
10363 +static inline void __dump_vx_cacct(struct _vx_cacct *cacct)
10367 + printk("\t_vx_cacct:");
10368 + for (i = 0; i < 6; i++) {
10369 + struct _vx_sock_acc *ptr = cacct->sock[i];
10371 + printk("\t [%d] =", i);
10372 + for (j = 0; j < 3; j++) {
10373 + printk(" [%d] = %8lu, %8lu", j,
10374 + atomic_long_read(&ptr[j].count),
10375 + atomic_long_read(&ptr[j].total));
10383 +#endif /* _VX_CACCT_DEF_H */
10384 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cacct.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct.h
10385 --- linux-2.6.29.1/include/linux/vserver/cacct.h 1970-01-01 01:00:00.000000000 +0100
10386 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct.h 2009-02-22 22:54:26.000000000 +0100
10388 +#ifndef _VX_CACCT_H
10389 +#define _VX_CACCT_H
10392 +enum sock_acc_field {
10393 + VXA_SOCK_UNSPEC = 0,
10399 + VXA_SOCK_SIZE /* array size */
10402 +#endif /* _VX_CACCT_H */
10403 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cacct_int.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_int.h
10404 --- linux-2.6.29.1/include/linux/vserver/cacct_int.h 1970-01-01 01:00:00.000000000 +0100
10405 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cacct_int.h 2009-02-22 22:54:26.000000000 +0100
10407 +#ifndef _VX_CACCT_INT_H
10408 +#define _VX_CACCT_INT_H
10414 +unsigned long vx_sock_count(struct _vx_cacct *cacct, int type, int pos)
10416 + return atomic_long_read(&cacct->sock[type][pos].count);
10421 +unsigned long vx_sock_total(struct _vx_cacct *cacct, int type, int pos)
10423 + return atomic_long_read(&cacct->sock[type][pos].total);
10426 +#endif /* __KERNEL__ */
10427 +#endif /* _VX_CACCT_INT_H */
10428 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/check.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/check.h
10429 --- linux-2.6.29.1/include/linux/vserver/check.h 1970-01-01 01:00:00.000000000 +0100
10430 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/check.h 2009-02-22 22:54:26.000000000 +0100
10432 +#ifndef _VS_CHECK_H
10433 +#define _VS_CHECK_H
10436 +#define MAX_S_CONTEXT 65535 /* Arbitrary limit */
10438 +#ifdef CONFIG_VSERVER_DYNAMIC_IDS
10439 +#define MIN_D_CONTEXT 49152 /* dynamic contexts start here */
10441 +#define MIN_D_CONTEXT 65536
10444 +/* check conditions */
10446 +#define VS_ADMIN 0x0001
10447 +#define VS_WATCH 0x0002
10448 +#define VS_HIDE 0x0004
10449 +#define VS_HOSTID 0x0008
10451 +#define VS_IDENT 0x0010
10452 +#define VS_EQUIV 0x0020
10453 +#define VS_PARENT 0x0040
10454 +#define VS_CHILD 0x0080
10456 +#define VS_ARG_MASK 0x00F0
10458 +#define VS_DYNAMIC 0x0100
10459 +#define VS_STATIC 0x0200
10461 +#define VS_ATR_MASK 0x0F00
10463 +#ifdef CONFIG_VSERVER_PRIVACY
10464 +#define VS_ADMIN_P (0)
10465 +#define VS_WATCH_P (0)
10467 +#define VS_ADMIN_P VS_ADMIN
10468 +#define VS_WATCH_P VS_WATCH
10471 +#define VS_HARDIRQ 0x1000
10472 +#define VS_SOFTIRQ 0x2000
10473 +#define VS_IRQ 0x4000
10475 +#define VS_IRQ_MASK 0xF000
10477 +#include <linux/hardirq.h>
10480 + * check current context for ADMIN/WATCH and
10481 + * optionally against supplied argument
10483 +static inline int __vs_check(int cid, int id, unsigned int mode)
10485 + if (mode & VS_ARG_MASK) {
10486 + if ((mode & VS_IDENT) && (id == cid))
10489 + if (mode & VS_ATR_MASK) {
10490 + if ((mode & VS_DYNAMIC) &&
10491 + (id >= MIN_D_CONTEXT) &&
10492 + (id <= MAX_S_CONTEXT))
10494 + if ((mode & VS_STATIC) &&
10495 + (id > 1) && (id < MIN_D_CONTEXT))
10498 + if (mode & VS_IRQ_MASK) {
10499 + if ((mode & VS_IRQ) && unlikely(in_interrupt()))
10501 + if ((mode & VS_HARDIRQ) && unlikely(in_irq()))
10503 + if ((mode & VS_SOFTIRQ) && unlikely(in_softirq()))
10506 + return (((mode & VS_ADMIN) && (cid == 0)) ||
10507 + ((mode & VS_WATCH) && (cid == 1)) ||
10508 + ((mode & VS_HOSTID) && (id == 0)));
10511 +#define vx_check(c, m) __vs_check(vx_current_xid(), c, (m) | VS_IRQ)
10513 +#define vx_weak_check(c, m) ((m) ? vx_check(c, m) : 1)
10516 +#define nx_check(c, m) __vs_check(nx_current_nid(), c, m)
10518 +#define nx_weak_check(c, m) ((m) ? nx_check(c, m) : 1)
10521 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/context_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/context_cmd.h
10522 --- linux-2.6.29.1/include/linux/vserver/context_cmd.h 1970-01-01 01:00:00.000000000 +0100
10523 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/context_cmd.h 2009-02-22 22:54:26.000000000 +0100
10525 +#ifndef _VX_CONTEXT_CMD_H
10526 +#define _VX_CONTEXT_CMD_H
10529 +/* vinfo commands */
10531 +#define VCMD_task_xid VC_CMD(VINFO, 1, 0)
10534 +extern int vc_task_xid(uint32_t);
10536 +#endif /* __KERNEL__ */
10538 +#define VCMD_vx_info VC_CMD(VINFO, 5, 0)
10540 +struct vcmd_vx_info_v0 {
10542 + uint32_t initpid;
10543 + /* more to come */
10547 +extern int vc_vx_info(struct vx_info *, void __user *);
10549 +#endif /* __KERNEL__ */
10551 +#define VCMD_ctx_stat VC_CMD(VSTAT, 0, 0)
10553 +struct vcmd_ctx_stat_v0 {
10556 + /* more to come */
10560 +extern int vc_ctx_stat(struct vx_info *, void __user *);
10562 +#endif /* __KERNEL__ */
10564 +/* context commands */
10566 +#define VCMD_ctx_create_v0 VC_CMD(VPROC, 1, 0)
10567 +#define VCMD_ctx_create VC_CMD(VPROC, 1, 1)
10569 +struct vcmd_ctx_create {
10570 + uint64_t flagword;
10573 +#define VCMD_ctx_migrate_v0 VC_CMD(PROCMIG, 1, 0)
10574 +#define VCMD_ctx_migrate VC_CMD(PROCMIG, 1, 1)
10576 +struct vcmd_ctx_migrate {
10577 + uint64_t flagword;
10581 +extern int vc_ctx_create(uint32_t, void __user *);
10582 +extern int vc_ctx_migrate(struct vx_info *, void __user *);
10584 +#endif /* __KERNEL__ */
10587 +/* flag commands */
10589 +#define VCMD_get_cflags VC_CMD(FLAGS, 1, 0)
10590 +#define VCMD_set_cflags VC_CMD(FLAGS, 2, 0)
10592 +struct vcmd_ctx_flags_v0 {
10593 + uint64_t flagword;
10598 +extern int vc_get_cflags(struct vx_info *, void __user *);
10599 +extern int vc_set_cflags(struct vx_info *, void __user *);
10601 +#endif /* __KERNEL__ */
10604 +/* context caps commands */
10606 +#define VCMD_get_ccaps VC_CMD(FLAGS, 3, 1)
10607 +#define VCMD_set_ccaps VC_CMD(FLAGS, 4, 1)
10609 +struct vcmd_ctx_caps_v1 {
10615 +extern int vc_get_ccaps(struct vx_info *, void __user *);
10616 +extern int vc_set_ccaps(struct vx_info *, void __user *);
10618 +#endif /* __KERNEL__ */
10621 +/* bcaps commands */
10623 +#define VCMD_get_bcaps VC_CMD(FLAGS, 9, 0)
10624 +#define VCMD_set_bcaps VC_CMD(FLAGS, 10, 0)
10626 +struct vcmd_bcaps {
10632 +extern int vc_get_bcaps(struct vx_info *, void __user *);
10633 +extern int vc_set_bcaps(struct vx_info *, void __user *);
10635 +#endif /* __KERNEL__ */
10640 +#define VCMD_get_badness VC_CMD(MEMCTRL, 5, 0)
10641 +#define VCMD_set_badness VC_CMD(MEMCTRL, 6, 0)
10643 +struct vcmd_badness_v0 {
10648 +extern int vc_get_badness(struct vx_info *, void __user *);
10649 +extern int vc_set_badness(struct vx_info *, void __user *);
10651 +#endif /* __KERNEL__ */
10652 +#endif /* _VX_CONTEXT_CMD_H */
10653 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/context.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/context.h
10654 --- linux-2.6.29.1/include/linux/vserver/context.h 1970-01-01 01:00:00.000000000 +0100
10655 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/context.h 2009-02-22 22:54:26.000000000 +0100
10657 +#ifndef _VX_CONTEXT_H
10658 +#define _VX_CONTEXT_H
10660 +#include <linux/types.h>
10661 +#include <linux/capability.h>
10664 +/* context flags */
10666 +#define VXF_INFO_SCHED 0x00000002
10667 +#define VXF_INFO_NPROC 0x00000004
10668 +#define VXF_INFO_PRIVATE 0x00000008
10670 +#define VXF_INFO_INIT 0x00000010
10671 +#define VXF_INFO_HIDE 0x00000020
10672 +#define VXF_INFO_ULIMIT 0x00000040
10673 +#define VXF_INFO_NSPACE 0x00000080
10675 +#define VXF_SCHED_HARD 0x00000100
10676 +#define VXF_SCHED_PRIO 0x00000200
10677 +#define VXF_SCHED_PAUSE 0x00000400
10679 +#define VXF_VIRT_MEM 0x00010000
10680 +#define VXF_VIRT_UPTIME 0x00020000
10681 +#define VXF_VIRT_CPU 0x00040000
10682 +#define VXF_VIRT_LOAD 0x00080000
10683 +#define VXF_VIRT_TIME 0x00100000
10685 +#define VXF_HIDE_MOUNT 0x01000000
10686 +/* was VXF_HIDE_NETIF 0x02000000 */
10687 +#define VXF_HIDE_VINFO 0x04000000
10689 +#define VXF_STATE_SETUP (1ULL << 32)
10690 +#define VXF_STATE_INIT (1ULL << 33)
10691 +#define VXF_STATE_ADMIN (1ULL << 34)
10693 +#define VXF_SC_HELPER (1ULL << 36)
10694 +#define VXF_REBOOT_KILL (1ULL << 37)
10695 +#define VXF_PERSISTENT (1ULL << 38)
10697 +#define VXF_FORK_RSS (1ULL << 48)
10698 +#define VXF_PROLIFIC (1ULL << 49)
10700 +#define VXF_IGNEG_NICE (1ULL << 52)
10702 +#define VXF_ONE_TIME (0x0007ULL << 32)
10704 +#define VXF_INIT_SET (VXF_STATE_SETUP | VXF_STATE_INIT | VXF_STATE_ADMIN)
10707 +/* context migration */
10709 +#define VXM_SET_INIT 0x00000001
10710 +#define VXM_SET_REAPER 0x00000002
10712 +/* context caps */
10714 +#define VXC_CAP_MASK 0x00000000
10716 +#define VXC_SET_UTSNAME 0x00000001
10717 +#define VXC_SET_RLIMIT 0x00000002
10718 +#define VXC_FS_SECURITY 0x00000004
10720 +/* was VXC_RAW_ICMP 0x00000100 */
10721 +#define VXC_SYSLOG 0x00001000
10723 +#define VXC_SECURE_MOUNT 0x00010000
10724 +#define VXC_SECURE_REMOUNT 0x00020000
10725 +#define VXC_BINARY_MOUNT 0x00040000
10727 +#define VXC_QUOTA_CTL 0x00100000
10728 +#define VXC_ADMIN_MAPPER 0x00200000
10729 +#define VXC_ADMIN_CLOOP 0x00400000
10731 +#define VXC_KTHREAD 0x01000000
10736 +#include <linux/list.h>
10737 +#include <linux/spinlock.h>
10738 +#include <linux/rcupdate.h>
10740 +#include "limit_def.h"
10741 +#include "sched_def.h"
10742 +#include "cvirt_def.h"
10743 +#include "cacct_def.h"
10744 +#include "device_def.h"
10746 +#define VX_SPACES 2
10748 +struct _vx_info_pc {
10749 + struct _vx_sched_pc sched_pc;
10750 + struct _vx_cvirt_pc cvirt_pc;
10754 + struct hlist_node vx_hlist; /* linked list of contexts */
10755 + xid_t vx_id; /* context id */
10756 + atomic_t vx_usecnt; /* usage count */
10757 + atomic_t vx_tasks; /* tasks count */
10758 + struct vx_info *vx_parent; /* parent context */
10759 + int vx_state; /* context state */
10761 + unsigned long vx_nsmask[VX_SPACES]; /* assignment mask */
10762 + struct nsproxy *vx_nsproxy[VX_SPACES]; /* private namespaces */
10763 + struct fs_struct *vx_fs[VX_SPACES]; /* private namespace fs */
10765 + uint64_t vx_flags; /* context flags */
10766 + uint64_t vx_ccaps; /* context caps (vserver) */
10767 + kernel_cap_t vx_bcaps; /* bounding caps (system) */
10768 + // kernel_cap_t vx_cap_bset; /* the guest's bset */
10770 + struct task_struct *vx_reaper; /* guest reaper process */
10771 + pid_t vx_initpid; /* PID of guest init */
10772 + int64_t vx_badness_bias; /* OOM points bias */
10774 + struct _vx_limit limit; /* vserver limits */
10775 + struct _vx_sched sched; /* vserver scheduler */
10776 + struct _vx_cvirt cvirt; /* virtual/bias stuff */
10777 + struct _vx_cacct cacct; /* context accounting */
10779 + struct _vx_device dmap; /* default device map targets */
10781 +#ifndef CONFIG_SMP
10782 + struct _vx_info_pc info_pc; /* per cpu data */
10784 + struct _vx_info_pc *ptr_pc; /* per cpu array */
10787 + wait_queue_head_t vx_wait; /* context exit waitqueue */
10788 + int reboot_cmd; /* last sys_reboot() cmd */
10789 + int exit_code; /* last process exit code */
10791 + char vx_name[65]; /* vserver name */
10794 +#ifndef CONFIG_SMP
10795 +#define vx_ptr_pc(vxi) (&(vxi)->info_pc)
10796 +#define vx_per_cpu(vxi, v, id) vx_ptr_pc(vxi)->v
10798 +#define vx_ptr_pc(vxi) ((vxi)->ptr_pc)
10799 +#define vx_per_cpu(vxi, v, id) per_cpu_ptr(vx_ptr_pc(vxi), id)->v
10802 +#define vx_cpu(vxi, v) vx_per_cpu(vxi, v, smp_processor_id())
10805 +struct vx_info_save {
10806 + struct vx_info *vxi;
10811 +/* status flags */
10813 +#define VXS_HASHED 0x0001
10814 +#define VXS_PAUSED 0x0010
10815 +#define VXS_SHUTDOWN 0x0100
10816 +#define VXS_HELPER 0x1000
10817 +#define VXS_RELEASED 0x8000
10820 +extern void claim_vx_info(struct vx_info *, struct task_struct *);
10821 +extern void release_vx_info(struct vx_info *, struct task_struct *);
10823 +extern struct vx_info *lookup_vx_info(int);
10824 +extern struct vx_info *lookup_or_create_vx_info(int);
10826 +extern int get_xid_list(int, unsigned int *, int);
10827 +extern int xid_is_hashed(xid_t);
10829 +extern int vx_migrate_task(struct task_struct *, struct vx_info *, int);
10831 +extern long vs_state_change(struct vx_info *, unsigned int);
10834 +#endif /* __KERNEL__ */
10835 +#endif /* _VX_CONTEXT_H */
10836 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cvirt_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt_cmd.h
10837 --- linux-2.6.29.1/include/linux/vserver/cvirt_cmd.h 1970-01-01 01:00:00.000000000 +0100
10838 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt_cmd.h 2009-02-22 22:54:26.000000000 +0100
10840 +#ifndef _VX_CVIRT_CMD_H
10841 +#define _VX_CVIRT_CMD_H
10844 +/* virtual host info name commands */
10846 +#define VCMD_set_vhi_name VC_CMD(VHOST, 1, 0)
10847 +#define VCMD_get_vhi_name VC_CMD(VHOST, 2, 0)
10849 +struct vcmd_vhi_name_v0 {
10855 +enum vhi_name_field {
10856 + VHIN_CONTEXT = 0,
10868 +#include <linux/compiler.h>
10870 +extern int vc_set_vhi_name(struct vx_info *, void __user *);
10871 +extern int vc_get_vhi_name(struct vx_info *, void __user *);
10873 +#endif /* __KERNEL__ */
10875 +#define VCMD_virt_stat VC_CMD(VSTAT, 3, 0)
10877 +struct vcmd_virt_stat_v0 {
10880 + uint32_t nr_threads;
10881 + uint32_t nr_running;
10882 + uint32_t nr_uninterruptible;
10883 + uint32_t nr_onhold;
10884 + uint32_t nr_forks;
10885 + uint32_t load[3];
10889 +extern int vc_virt_stat(struct vx_info *, void __user *);
10891 +#endif /* __KERNEL__ */
10892 +#endif /* _VX_CVIRT_CMD_H */
10893 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cvirt_def.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt_def.h
10894 --- linux-2.6.29.1/include/linux/vserver/cvirt_def.h 1970-01-01 01:00:00.000000000 +0100
10895 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt_def.h 2009-02-22 22:54:26.000000000 +0100
10897 +#ifndef _VX_CVIRT_DEF_H
10898 +#define _VX_CVIRT_DEF_H
10900 +#include <linux/jiffies.h>
10901 +#include <linux/spinlock.h>
10902 +#include <linux/wait.h>
10903 +#include <linux/time.h>
10904 +#include <asm/atomic.h>
10907 +struct _vx_usage_stat {
10911 + uint64_t softirq;
10917 +struct _vx_syslog {
10918 + wait_queue_head_t log_wait;
10919 + spinlock_t logbuf_lock; /* lock for the log buffer */
10921 + unsigned long log_start; /* next char to be read by syslog() */
10922 + unsigned long con_start; /* next char to be sent to consoles */
10923 + unsigned long log_end; /* most-recently-written-char + 1 */
10924 + unsigned long logged_chars; /* #chars since last read+clear operation */
10926 + char log_buf[1024];
10930 +/* context sub struct */
10932 +struct _vx_cvirt {
10933 + atomic_t nr_threads; /* number of current threads */
10934 + atomic_t nr_running; /* number of running threads */
10935 + atomic_t nr_uninterruptible; /* number of uninterruptible threads */
10937 + atomic_t nr_onhold; /* processes on hold */
10938 + uint32_t onhold_last; /* jiffies when put on hold */
10940 + struct timeval bias_tv; /* time offset to the host */
10941 + struct timespec bias_idle;
10942 + struct timespec bias_uptime; /* context creation point */
10943 + uint64_t bias_clock; /* offset in clock_t */
10945 + spinlock_t load_lock; /* lock for the load averages */
10946 + atomic_t load_updates; /* nr of load updates done so far */
10947 + uint32_t load_last; /* last time load was calculated */
10948 + uint32_t load[3]; /* load averages 1,5,15 */
10950 + atomic_t total_forks; /* number of forks so far */
10952 + struct _vx_syslog syslog;
10955 +struct _vx_cvirt_pc {
10956 + struct _vx_usage_stat cpustat;
10960 +#ifdef CONFIG_VSERVER_DEBUG
10962 +static inline void __dump_vx_cvirt(struct _vx_cvirt *cvirt)
10964 + printk("\t_vx_cvirt:\n");
10965 + printk("\t threads: %4d, %4d, %4d, %4d\n",
10966 + atomic_read(&cvirt->nr_threads),
10967 + atomic_read(&cvirt->nr_running),
10968 + atomic_read(&cvirt->nr_uninterruptible),
10969 + atomic_read(&cvirt->nr_onhold));
10970 + /* add rest here */
10971 + printk("\t total_forks = %d\n", atomic_read(&cvirt->total_forks));
10976 +#endif /* _VX_CVIRT_DEF_H */
10977 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/cvirt.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt.h
10978 --- linux-2.6.29.1/include/linux/vserver/cvirt.h 1970-01-01 01:00:00.000000000 +0100
10979 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/cvirt.h 2009-02-22 22:54:26.000000000 +0100
10981 +#ifndef _VX_CVIRT_H
10982 +#define _VX_CVIRT_H
10989 +void vx_vsi_uptime(struct timespec *, struct timespec *);
10994 +void vx_update_load(struct vx_info *);
10997 +int vx_do_syslog(int, char __user *, int);
10999 +#endif /* __KERNEL__ */
11000 +#endif /* _VX_CVIRT_H */
11001 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/debug_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/debug_cmd.h
11002 --- linux-2.6.29.1/include/linux/vserver/debug_cmd.h 1970-01-01 01:00:00.000000000 +0100
11003 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/debug_cmd.h 2009-02-22 22:54:26.000000000 +0100
11005 +#ifndef _VX_DEBUG_CMD_H
11006 +#define _VX_DEBUG_CMD_H
11009 +/* debug commands */
11011 +#define VCMD_dump_history VC_CMD(DEBUG, 1, 0)
11013 +#define VCMD_read_history VC_CMD(DEBUG, 5, 0)
11014 +#define VCMD_read_monitor VC_CMD(DEBUG, 6, 0)
11016 +struct vcmd_read_history_v0 {
11019 + char __user *data;
11022 +struct vcmd_read_monitor_v0 {
11025 + char __user *data;
11031 +#ifdef CONFIG_COMPAT
11033 +#include <asm/compat.h>
11035 +struct vcmd_read_history_v0_x32 {
11038 + compat_uptr_t data_ptr;
11041 +struct vcmd_read_monitor_v0_x32 {
11044 + compat_uptr_t data_ptr;
11047 +#endif /* CONFIG_COMPAT */
11049 +extern int vc_dump_history(uint32_t);
11051 +extern int vc_read_history(uint32_t, void __user *);
11052 +extern int vc_read_monitor(uint32_t, void __user *);
11054 +#ifdef CONFIG_COMPAT
11056 +extern int vc_read_history_x32(uint32_t, void __user *);
11057 +extern int vc_read_monitor_x32(uint32_t, void __user *);
11059 +#endif /* CONFIG_COMPAT */
11061 +#endif /* __KERNEL__ */
11062 +#endif /* _VX_DEBUG_CMD_H */
11063 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/debug.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/debug.h
11064 --- linux-2.6.29.1/include/linux/vserver/debug.h 1970-01-01 01:00:00.000000000 +0100
11065 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/debug.h 2009-02-22 22:54:26.000000000 +0100
11067 +#ifndef _VX_DEBUG_H
11068 +#define _VX_DEBUG_H
11071 +#define VXD_CBIT(n, m) (vx_debug_ ## n & (1 << (m)))
11072 +#define VXD_CMIN(n, m) (vx_debug_ ## n > (m))
11073 +#define VXD_MASK(n, m) (vx_debug_ ## n & (m))
11075 +#define VXD_DEV(d) (d), (d)->bd_inode->i_ino, \
11076 + imajor((d)->bd_inode), iminor((d)->bd_inode)
11077 +#define VXF_DEV "%p[%lu,%d:%d]"
11080 +#define vxd_path(p) \
11081 + ({ static char _buffer[PATH_MAX]; \
11082 + d_path(p, _buffer, sizeof(_buffer)); })
11084 +#define vxd_cond_path(n) \
11085 + ((n) ? vxd_path(&(n)->path) : "<null>" )
11088 +#ifdef CONFIG_VSERVER_DEBUG
11090 +extern unsigned int vx_debug_switch;
11091 +extern unsigned int vx_debug_xid;
11092 +extern unsigned int vx_debug_nid;
11093 +extern unsigned int vx_debug_tag;
11094 +extern unsigned int vx_debug_net;
11095 +extern unsigned int vx_debug_limit;
11096 +extern unsigned int vx_debug_cres;
11097 +extern unsigned int vx_debug_dlim;
11098 +extern unsigned int vx_debug_quota;
11099 +extern unsigned int vx_debug_cvirt;
11100 +extern unsigned int vx_debug_space;
11101 +extern unsigned int vx_debug_misc;
11104 +#define VX_LOGLEVEL "vxD: "
11105 +#define VX_PROC_FMT "%p: "
11106 +#define VX_PROCESS current
11108 +#define vxdprintk(c, f, x...) \
11111 + printk(VX_LOGLEVEL VX_PROC_FMT f "\n", \
11112 + VX_PROCESS , ##x); \
11115 +#define vxlprintk(c, f, x...) \
11118 + printk(VX_LOGLEVEL f " @%s:%d\n", x); \
11121 +#define vxfprintk(c, f, x...) \
11124 + printk(VX_LOGLEVEL f " %s@%s:%d\n", x); \
11130 +void dump_vx_info(struct vx_info *, int);
11131 +void dump_vx_info_inactive(int);
11133 +#else /* CONFIG_VSERVER_DEBUG */
11135 +#define vx_debug_switch 0
11136 +#define vx_debug_xid 0
11137 +#define vx_debug_nid 0
11138 +#define vx_debug_tag 0
11139 +#define vx_debug_net 0
11140 +#define vx_debug_limit 0
11141 +#define vx_debug_cres 0
11142 +#define vx_debug_dlim 0
11143 +#define vx_debug_cvirt 0
11145 +#define vxdprintk(x...) do { } while (0)
11146 +#define vxlprintk(x...) do { } while (0)
11147 +#define vxfprintk(x...) do { } while (0)
11149 +#endif /* CONFIG_VSERVER_DEBUG */
11152 +#ifdef CONFIG_VSERVER_WARN
11154 +#define VX_WARNLEVEL KERN_WARNING "vxW: "
11155 +#define VX_WARN_TASK "[»%s«,%u:#%u|%u|%u] "
11156 +#define VX_WARN_XID "[xid #%u] "
11157 +#define VX_WARN_NID "[nid #%u] "
11158 +#define VX_WARN_TAG "[tag #%u] "
11160 +#define vxwprintk(c, f, x...) \
11163 + printk(VX_WARNLEVEL f "\n", ##x); \
11166 +#else /* CONFIG_VSERVER_WARN */
11168 +#define vxwprintk(x...) do { } while (0)
11170 +#endif /* CONFIG_VSERVER_WARN */
11172 +#define vxwprintk_task(c, f, x...) \
11173 + vxwprintk(c, VX_WARN_TASK f, \
11174 + current->comm, current->pid, \
11175 + current->xid, current->nid, current->tag, ##x)
11176 +#define vxwprintk_xid(c, f, x...) \
11177 + vxwprintk(c, VX_WARN_XID f, current->xid, x)
11178 +#define vxwprintk_nid(c, f, x...) \
11179 + vxwprintk(c, VX_WARN_NID f, current->nid, x)
11180 +#define vxwprintk_tag(c, f, x...) \
11181 + vxwprintk(c, VX_WARN_TAG f, current->tag, x)
11183 +#ifdef CONFIG_VSERVER_DEBUG
11184 +#define vxd_assert_lock(l) assert_spin_locked(l)
11185 +#define vxd_assert(c, f, x...) vxlprintk(!(c), \
11186 + "assertion [" f "] failed.", ##x, __FILE__, __LINE__)
11188 +#define vxd_assert_lock(l) do { } while (0)
11189 +#define vxd_assert(c, f, x...) do { } while (0)
11193 +#endif /* _VX_DEBUG_H */
11194 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/device_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device_cmd.h
11195 --- linux-2.6.29.1/include/linux/vserver/device_cmd.h 1970-01-01 01:00:00.000000000 +0100
11196 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device_cmd.h 2009-02-22 22:54:26.000000000 +0100
11198 +#ifndef _VX_DEVICE_CMD_H
11199 +#define _VX_DEVICE_CMD_H
11202 +/* device vserver commands */
11204 +#define VCMD_set_mapping VC_CMD(DEVICE, 1, 0)
11205 +#define VCMD_unset_mapping VC_CMD(DEVICE, 2, 0)
11207 +struct vcmd_set_mapping_v0 {
11208 + const char __user *device;
11209 + const char __user *target;
11216 +#ifdef CONFIG_COMPAT
11218 +#include <asm/compat.h>
11220 +struct vcmd_set_mapping_v0_x32 {
11221 + compat_uptr_t device_ptr;
11222 + compat_uptr_t target_ptr;
11226 +#endif /* CONFIG_COMPAT */
11228 +#include <linux/compiler.h>
11230 +extern int vc_set_mapping(struct vx_info *, void __user *);
11231 +extern int vc_unset_mapping(struct vx_info *, void __user *);
11233 +#ifdef CONFIG_COMPAT
11235 +extern int vc_set_mapping_x32(struct vx_info *, void __user *);
11236 +extern int vc_unset_mapping_x32(struct vx_info *, void __user *);
11238 +#endif /* CONFIG_COMPAT */
11240 +#endif /* __KERNEL__ */
11241 +#endif /* _VX_DEVICE_CMD_H */
11242 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/device_def.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device_def.h
11243 --- linux-2.6.29.1/include/linux/vserver/device_def.h 1970-01-01 01:00:00.000000000 +0100
11244 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device_def.h 2009-02-22 22:54:26.000000000 +0100
11246 +#ifndef _VX_DEVICE_DEF_H
11247 +#define _VX_DEVICE_DEF_H
11249 +#include <linux/types.h>
11251 +struct vx_dmap_target {
11256 +struct _vx_device {
11257 +#ifdef CONFIG_VSERVER_DEVICE
11258 + struct vx_dmap_target targets[2];
11262 +#endif /* _VX_DEVICE_DEF_H */
11263 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/device.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device.h
11264 --- linux-2.6.29.1/include/linux/vserver/device.h 1970-01-01 01:00:00.000000000 +0100
11265 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/device.h 2009-02-22 22:54:26.000000000 +0100
11267 +#ifndef _VX_DEVICE_H
11268 +#define _VX_DEVICE_H
11271 +#define DATTR_CREATE 0x00000001
11272 +#define DATTR_OPEN 0x00000002
11274 +#define DATTR_REMAP 0x00000010
11276 +#define DATTR_MASK 0x00000013
11279 +#else /* _VX_DEVICE_H */
11280 +#warning duplicate inclusion
11281 +#endif /* _VX_DEVICE_H */
11282 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/dlimit_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/dlimit_cmd.h
11283 --- linux-2.6.29.1/include/linux/vserver/dlimit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11284 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/dlimit_cmd.h 2009-02-22 22:54:26.000000000 +0100
11286 +#ifndef _VX_DLIMIT_CMD_H
11287 +#define _VX_DLIMIT_CMD_H
11290 +/* dlimit vserver commands */
11292 +#define VCMD_add_dlimit VC_CMD(DLIMIT, 1, 0)
11293 +#define VCMD_rem_dlimit VC_CMD(DLIMIT, 2, 0)
11295 +#define VCMD_set_dlimit VC_CMD(DLIMIT, 5, 0)
11296 +#define VCMD_get_dlimit VC_CMD(DLIMIT, 6, 0)
11298 +struct vcmd_ctx_dlimit_base_v0 {
11299 + const char __user *name;
11303 +struct vcmd_ctx_dlimit_v0 {
11304 + const char __user *name;
11305 + uint32_t space_used; /* used space in kbytes */
11306 + uint32_t space_total; /* maximum space in kbytes */
11307 + uint32_t inodes_used; /* used inodes */
11308 + uint32_t inodes_total; /* maximum inodes */
11309 + uint32_t reserved; /* reserved for root in % */
11313 +#define CDLIM_UNSET ((uint32_t)0UL)
11314 +#define CDLIM_INFINITY ((uint32_t)~0UL)
11315 +#define CDLIM_KEEP ((uint32_t)~1UL)
11319 +#ifdef CONFIG_COMPAT
11321 +#include <asm/compat.h>
11323 +struct vcmd_ctx_dlimit_base_v0_x32 {
11324 + compat_uptr_t name_ptr;
11328 +struct vcmd_ctx_dlimit_v0_x32 {
11329 + compat_uptr_t name_ptr;
11330 + uint32_t space_used; /* used space in kbytes */
11331 + uint32_t space_total; /* maximum space in kbytes */
11332 + uint32_t inodes_used; /* used inodes */
11333 + uint32_t inodes_total; /* maximum inodes */
11334 + uint32_t reserved; /* reserved for root in % */
11338 +#endif /* CONFIG_COMPAT */
11340 +#include <linux/compiler.h>
11342 +extern int vc_add_dlimit(uint32_t, void __user *);
11343 +extern int vc_rem_dlimit(uint32_t, void __user *);
11345 +extern int vc_set_dlimit(uint32_t, void __user *);
11346 +extern int vc_get_dlimit(uint32_t, void __user *);
11348 +#ifdef CONFIG_COMPAT
11350 +extern int vc_add_dlimit_x32(uint32_t, void __user *);
11351 +extern int vc_rem_dlimit_x32(uint32_t, void __user *);
11353 +extern int vc_set_dlimit_x32(uint32_t, void __user *);
11354 +extern int vc_get_dlimit_x32(uint32_t, void __user *);
11356 +#endif /* CONFIG_COMPAT */
11358 +#endif /* __KERNEL__ */
11359 +#endif /* _VX_DLIMIT_CMD_H */
11360 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/dlimit.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/dlimit.h
11361 --- linux-2.6.29.1/include/linux/vserver/dlimit.h 1970-01-01 01:00:00.000000000 +0100
11362 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/dlimit.h 2009-02-22 22:54:26.000000000 +0100
11364 +#ifndef _VX_DLIMIT_H
11365 +#define _VX_DLIMIT_H
11367 +#include "switch.h"
11372 +/* keep in sync with CDLIM_INFINITY */
11374 +#define DLIM_INFINITY (~0ULL)
11376 +#include <linux/spinlock.h>
11377 +#include <linux/rcupdate.h>
11379 +struct super_block;
11382 + struct hlist_node dl_hlist; /* linked list of contexts */
11383 + struct rcu_head dl_rcu; /* the rcu head */
11384 + tag_t dl_tag; /* context tag */
11385 + atomic_t dl_usecnt; /* usage count */
11386 + atomic_t dl_refcnt; /* reference count */
11388 + struct super_block *dl_sb; /* associated superblock */
11390 + spinlock_t dl_lock; /* protect the values */
11392 + unsigned long long dl_space_used; /* used space in bytes */
11393 + unsigned long long dl_space_total; /* maximum space in bytes */
11394 + unsigned long dl_inodes_used; /* used inodes */
11395 + unsigned long dl_inodes_total; /* maximum inodes */
11397 + unsigned int dl_nrlmult; /* non root limit mult */
11402 +extern void rcu_free_dl_info(struct rcu_head *);
11403 +extern void unhash_dl_info(struct dl_info *);
11405 +extern struct dl_info *locate_dl_info(struct super_block *, tag_t);
11410 +extern void vx_vsi_statfs(struct super_block *, struct kstatfs *);
11412 +typedef uint64_t dlsize_t;
11414 +#endif /* __KERNEL__ */
11415 +#else /* _VX_DLIMIT_H */
11416 +#warning duplicate inclusion
11417 +#endif /* _VX_DLIMIT_H */
11418 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/global.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/global.h
11419 --- linux-2.6.29.1/include/linux/vserver/global.h 1970-01-01 01:00:00.000000000 +0100
11420 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/global.h 2009-02-22 22:54:26.000000000 +0100
11422 +#ifndef _VX_GLOBAL_H
11423 +#define _VX_GLOBAL_H
11426 +extern atomic_t vx_global_ctotal;
11427 +extern atomic_t vx_global_cactive;
11429 +extern atomic_t nx_global_ctotal;
11430 +extern atomic_t nx_global_cactive;
11432 +extern atomic_t vs_global_nsproxy;
11433 +extern atomic_t vs_global_fs;
11434 +extern atomic_t vs_global_mnt_ns;
11435 +extern atomic_t vs_global_uts_ns;
11436 +extern atomic_t vs_global_ipc_ns;
11437 +extern atomic_t vs_global_user_ns;
11438 +extern atomic_t vs_global_pid_ns;
11441 +#endif /* _VX_GLOBAL_H */
11442 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/history.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/history.h
11443 --- linux-2.6.29.1/include/linux/vserver/history.h 1970-01-01 01:00:00.000000000 +0100
11444 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/history.h 2009-02-22 22:54:26.000000000 +0100
11446 +#ifndef _VX_HISTORY_H
11447 +#define _VX_HISTORY_H
11452 + VXH_THROW_OOPS = 1,
11456 + VXH_INIT_VX_INFO,
11459 + VXH_CLAIM_VX_INFO,
11460 + VXH_RELEASE_VX_INFO,
11461 + VXH_ALLOC_VX_INFO,
11462 + VXH_DEALLOC_VX_INFO,
11463 + VXH_HASH_VX_INFO,
11464 + VXH_UNHASH_VX_INFO,
11466 + VXH_LOOKUP_VX_INFO,
11467 + VXH_CREATE_VX_INFO,
11470 +struct _vxhe_vxi {
11471 + struct vx_info *ptr;
11477 +struct _vxhe_set_clr {
11481 +struct _vxhe_loc_lookup {
11485 +struct _vx_hist_entry {
11487 + unsigned short seq;
11488 + unsigned short type;
11489 + struct _vxhe_vxi vxi;
11491 + struct _vxhe_set_clr sc;
11492 + struct _vxhe_loc_lookup ll;
11496 +#ifdef CONFIG_VSERVER_HISTORY
11498 +extern unsigned volatile int vxh_active;
11500 +struct _vx_hist_entry *vxh_advance(void *loc);
11504 +void __vxh_copy_vxi(struct _vx_hist_entry *entry, struct vx_info *vxi)
11506 + entry->vxi.ptr = vxi;
11508 + entry->vxi.usecnt = atomic_read(&vxi->vx_usecnt);
11509 + entry->vxi.tasks = atomic_read(&vxi->vx_tasks);
11510 + entry->vxi.xid = vxi->vx_id;
11515 +#define __HERE__ current_text_addr()
11517 +#define __VXH_BODY(__type, __data, __here) \
11518 + struct _vx_hist_entry *entry; \
11520 + preempt_disable(); \
11521 + entry = vxh_advance(__here); \
11523 + entry->type = __type; \
11524 + preempt_enable();
11527 + /* pass vxi only */
11529 +#define __VXH_SMPL \
11530 + __vxh_copy_vxi(entry, vxi)
11533 +void __vxh_smpl(struct vx_info *vxi, int __type, void *__here)
11535 + __VXH_BODY(__type, __VXH_SMPL, __here)
11538 + /* pass vxi and data (void *) */
11540 +#define __VXH_DATA \
11541 + __vxh_copy_vxi(entry, vxi); \
11542 + entry->sc.data = data
11545 +void __vxh_data(struct vx_info *vxi, void *data,
11546 + int __type, void *__here)
11548 + __VXH_BODY(__type, __VXH_DATA, __here)
11551 + /* pass vxi and arg (long) */
11553 +#define __VXH_LONG \
11554 + __vxh_copy_vxi(entry, vxi); \
11555 + entry->ll.arg = arg
11558 +void __vxh_long(struct vx_info *vxi, long arg,
11559 + int __type, void *__here)
11561 + __VXH_BODY(__type, __VXH_LONG, __here)
11566 +void __vxh_throw_oops(void *__here)
11568 + __VXH_BODY(VXH_THROW_OOPS, {}, __here);
11569 + /* prevent further acquisition */
11574 +#define vxh_throw_oops() __vxh_throw_oops(__HERE__);
11576 +#define __vxh_get_vx_info(v, h) __vxh_smpl(v, VXH_GET_VX_INFO, h);
11577 +#define __vxh_put_vx_info(v, h) __vxh_smpl(v, VXH_PUT_VX_INFO, h);
11579 +#define __vxh_init_vx_info(v, d, h) \
11580 + __vxh_data(v, d, VXH_INIT_VX_INFO, h);
11581 +#define __vxh_set_vx_info(v, d, h) \
11582 + __vxh_data(v, d, VXH_SET_VX_INFO, h);
11583 +#define __vxh_clr_vx_info(v, d, h) \
11584 + __vxh_data(v, d, VXH_CLR_VX_INFO, h);
11586 +#define __vxh_claim_vx_info(v, d, h) \
11587 + __vxh_data(v, d, VXH_CLAIM_VX_INFO, h);
11588 +#define __vxh_release_vx_info(v, d, h) \
11589 + __vxh_data(v, d, VXH_RELEASE_VX_INFO, h);
11591 +#define vxh_alloc_vx_info(v) \
11592 + __vxh_smpl(v, VXH_ALLOC_VX_INFO, __HERE__);
11593 +#define vxh_dealloc_vx_info(v) \
11594 + __vxh_smpl(v, VXH_DEALLOC_VX_INFO, __HERE__);
11596 +#define vxh_hash_vx_info(v) \
11597 + __vxh_smpl(v, VXH_HASH_VX_INFO, __HERE__);
11598 +#define vxh_unhash_vx_info(v) \
11599 + __vxh_smpl(v, VXH_UNHASH_VX_INFO, __HERE__);
11601 +#define vxh_loc_vx_info(v, l) \
11602 + __vxh_long(v, l, VXH_LOC_VX_INFO, __HERE__);
11603 +#define vxh_lookup_vx_info(v, l) \
11604 + __vxh_long(v, l, VXH_LOOKUP_VX_INFO, __HERE__);
11605 +#define vxh_create_vx_info(v, l) \
11606 + __vxh_long(v, l, VXH_CREATE_VX_INFO, __HERE__);
11608 +extern void vxh_dump_history(void);
11611 +#else /* CONFIG_VSERVER_HISTORY */
11613 +#define __HERE__ 0
11615 +#define vxh_throw_oops() do { } while (0)
11617 +#define __vxh_get_vx_info(v, h) do { } while (0)
11618 +#define __vxh_put_vx_info(v, h) do { } while (0)
11620 +#define __vxh_init_vx_info(v, d, h) do { } while (0)
11621 +#define __vxh_set_vx_info(v, d, h) do { } while (0)
11622 +#define __vxh_clr_vx_info(v, d, h) do { } while (0)
11624 +#define __vxh_claim_vx_info(v, d, h) do { } while (0)
11625 +#define __vxh_release_vx_info(v, d, h) do { } while (0)
11627 +#define vxh_alloc_vx_info(v) do { } while (0)
11628 +#define vxh_dealloc_vx_info(v) do { } while (0)
11630 +#define vxh_hash_vx_info(v) do { } while (0)
11631 +#define vxh_unhash_vx_info(v) do { } while (0)
11633 +#define vxh_loc_vx_info(v, l) do { } while (0)
11634 +#define vxh_lookup_vx_info(v, l) do { } while (0)
11635 +#define vxh_create_vx_info(v, l) do { } while (0)
11637 +#define vxh_dump_history() do { } while (0)
11640 +#endif /* CONFIG_VSERVER_HISTORY */
11642 +#endif /* _VX_HISTORY_H */
11643 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/inode_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/inode_cmd.h
11644 --- linux-2.6.29.1/include/linux/vserver/inode_cmd.h 1970-01-01 01:00:00.000000000 +0100
11645 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/inode_cmd.h 2009-02-22 22:54:26.000000000 +0100
11647 +#ifndef _VX_INODE_CMD_H
11648 +#define _VX_INODE_CMD_H
11651 +/* inode vserver commands */
11653 +#define VCMD_get_iattr VC_CMD(INODE, 1, 1)
11654 +#define VCMD_set_iattr VC_CMD(INODE, 2, 1)
11656 +#define VCMD_fget_iattr VC_CMD(INODE, 3, 0)
11657 +#define VCMD_fset_iattr VC_CMD(INODE, 4, 0)
11659 +struct vcmd_ctx_iattr_v1 {
11660 + const char __user *name;
11666 +struct vcmd_ctx_fiattr_v0 {
11676 +#ifdef CONFIG_COMPAT
11678 +#include <asm/compat.h>
11680 +struct vcmd_ctx_iattr_v1_x32 {
11681 + compat_uptr_t name_ptr;
11687 +#endif /* CONFIG_COMPAT */
11689 +#include <linux/compiler.h>
11691 +extern int vc_get_iattr(void __user *);
11692 +extern int vc_set_iattr(void __user *);
11694 +extern int vc_fget_iattr(uint32_t, void __user *);
11695 +extern int vc_fset_iattr(uint32_t, void __user *);
11697 +#ifdef CONFIG_COMPAT
11699 +extern int vc_get_iattr_x32(void __user *);
11700 +extern int vc_set_iattr_x32(void __user *);
11702 +#endif /* CONFIG_COMPAT */
11704 +#endif /* __KERNEL__ */
11705 +#endif /* _VX_INODE_CMD_H */
11706 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/inode.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/inode.h
11707 --- linux-2.6.29.1/include/linux/vserver/inode.h 1970-01-01 01:00:00.000000000 +0100
11708 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/inode.h 2009-02-22 22:54:26.000000000 +0100
11710 +#ifndef _VX_INODE_H
11711 +#define _VX_INODE_H
11714 +#define IATTR_TAG 0x01000000
11716 +#define IATTR_ADMIN 0x00000001
11717 +#define IATTR_WATCH 0x00000002
11718 +#define IATTR_HIDE 0x00000004
11719 +#define IATTR_FLAGS 0x00000007
11721 +#define IATTR_BARRIER 0x00010000
11722 +#define IATTR_IXUNLINK 0x00020000
11723 +#define IATTR_IMMUTABLE 0x00040000
11728 +#ifdef CONFIG_VSERVER_PROC_SECURE
11729 +#define IATTR_PROC_DEFAULT ( IATTR_ADMIN | IATTR_HIDE )
11730 +#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11732 +#define IATTR_PROC_DEFAULT ( IATTR_ADMIN )
11733 +#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11736 +#define vx_hide_check(c, m) (((m) & IATTR_HIDE) ? vx_check(c, m) : 1)
11738 +#endif /* __KERNEL__ */
11740 +/* inode ioctls */
11742 +#define FIOC_GETXFLG _IOR('x', 5, long)
11743 +#define FIOC_SETXFLG _IOW('x', 6, long)
11745 +#else /* _VX_INODE_H */
11746 +#warning duplicate inclusion
11747 +#endif /* _VX_INODE_H */
11748 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/Kbuild linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/Kbuild
11749 --- linux-2.6.29.1/include/linux/vserver/Kbuild 1970-01-01 01:00:00.000000000 +0100
11750 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/Kbuild 2009-02-22 22:54:26.000000000 +0100
11753 +unifdef-y += context_cmd.h network_cmd.h space_cmd.h \
11754 + cacct_cmd.h cvirt_cmd.h limit_cmd.h dlimit_cmd.h \
11755 + inode_cmd.h tag_cmd.h sched_cmd.h signal_cmd.h \
11756 + debug_cmd.h device_cmd.h
11758 +unifdef-y += switch.h network.h monitor.h inode.h device.h
11760 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/limit_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_cmd.h
11761 --- linux-2.6.29.1/include/linux/vserver/limit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11762 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_cmd.h 2009-02-22 22:54:26.000000000 +0100
11764 +#ifndef _VX_LIMIT_CMD_H
11765 +#define _VX_LIMIT_CMD_H
11768 +/* rlimit vserver commands */
11770 +#define VCMD_get_rlimit VC_CMD(RLIMIT, 1, 0)
11771 +#define VCMD_set_rlimit VC_CMD(RLIMIT, 2, 0)
11772 +#define VCMD_get_rlimit_mask VC_CMD(RLIMIT, 3, 0)
11773 +#define VCMD_reset_minmax VC_CMD(RLIMIT, 9, 0)
11775 +struct vcmd_ctx_rlimit_v0 {
11777 + uint64_t minimum;
11778 + uint64_t softlimit;
11779 + uint64_t maximum;
11782 +struct vcmd_ctx_rlimit_mask_v0 {
11783 + uint32_t minimum;
11784 + uint32_t softlimit;
11785 + uint32_t maximum;
11788 +#define VCMD_rlimit_stat VC_CMD(VSTAT, 1, 0)
11790 +struct vcmd_rlimit_stat_v0 {
11794 + uint64_t minimum;
11795 + uint64_t maximum;
11798 +#define CRLIM_UNSET (0ULL)
11799 +#define CRLIM_INFINITY (~0ULL)
11800 +#define CRLIM_KEEP (~1ULL)
11804 +#ifdef CONFIG_IA32_EMULATION
11806 +struct vcmd_ctx_rlimit_v0_x32 {
11808 + uint64_t minimum;
11809 + uint64_t softlimit;
11810 + uint64_t maximum;
11811 +} __attribute__ ((packed));
11813 +#endif /* CONFIG_IA32_EMULATION */
11815 +#include <linux/compiler.h>
11817 +extern int vc_get_rlimit_mask(uint32_t, void __user *);
11818 +extern int vc_get_rlimit(struct vx_info *, void __user *);
11819 +extern int vc_set_rlimit(struct vx_info *, void __user *);
11820 +extern int vc_reset_minmax(struct vx_info *, void __user *);
11822 +extern int vc_rlimit_stat(struct vx_info *, void __user *);
11824 +#ifdef CONFIG_IA32_EMULATION
11826 +extern int vc_get_rlimit_x32(struct vx_info *, void __user *);
11827 +extern int vc_set_rlimit_x32(struct vx_info *, void __user *);
11829 +#endif /* CONFIG_IA32_EMULATION */
11831 +#endif /* __KERNEL__ */
11832 +#endif /* _VX_LIMIT_CMD_H */
11833 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/limit_def.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_def.h
11834 --- linux-2.6.29.1/include/linux/vserver/limit_def.h 1970-01-01 01:00:00.000000000 +0100
11835 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_def.h 2009-02-22 22:54:26.000000000 +0100
11837 +#ifndef _VX_LIMIT_DEF_H
11838 +#define _VX_LIMIT_DEF_H
11840 +#include <asm/atomic.h>
11841 +#include <asm/resource.h>
11843 +#include "limit.h"
11846 +struct _vx_res_limit {
11847 + rlim_t soft; /* Context soft limit */
11848 + rlim_t hard; /* Context hard limit */
11850 + rlim_atomic_t rcur; /* Current value */
11851 + rlim_t rmin; /* Context minimum */
11852 + rlim_t rmax; /* Context maximum */
11854 + atomic_t lhit; /* Limit hits */
11857 +/* context sub struct */
11859 +struct _vx_limit {
11860 + struct _vx_res_limit res[NUM_LIMITS];
11863 +#ifdef CONFIG_VSERVER_DEBUG
11865 +static inline void __dump_vx_limit(struct _vx_limit *limit)
11869 + printk("\t_vx_limit:");
11870 + for (i = 0; i < NUM_LIMITS; i++) {
11871 + printk("\t [%2d] = %8lu %8lu/%8lu, %8ld/%8ld, %8d\n",
11872 + i, (unsigned long)__rlim_get(limit, i),
11873 + (unsigned long)__rlim_rmin(limit, i),
11874 + (unsigned long)__rlim_rmax(limit, i),
11875 + (long)__rlim_soft(limit, i),
11876 + (long)__rlim_hard(limit, i),
11877 + atomic_read(&__rlim_lhit(limit, i)));
11883 +#endif /* _VX_LIMIT_DEF_H */
11884 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/limit.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit.h
11885 --- linux-2.6.29.1/include/linux/vserver/limit.h 1970-01-01 01:00:00.000000000 +0100
11886 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit.h 2009-02-22 22:54:26.000000000 +0100
11888 +#ifndef _VX_LIMIT_H
11889 +#define _VX_LIMIT_H
11891 +#define VLIMIT_NSOCK 16
11892 +#define VLIMIT_OPENFD 17
11893 +#define VLIMIT_ANON 18
11894 +#define VLIMIT_SHMEM 19
11895 +#define VLIMIT_SEMARY 20
11896 +#define VLIMIT_NSEMS 21
11897 +#define VLIMIT_DENTRY 22
11898 +#define VLIMIT_MAPPED 23
11903 +#define VLIM_NOCHECK ((1L << VLIMIT_DENTRY) | (1L << RLIMIT_RSS))
11905 +/* keep in sync with CRLIM_INFINITY */
11907 +#define VLIM_INFINITY (~0ULL)
11909 +#include <asm/atomic.h>
11910 +#include <asm/resource.h>
11912 +#ifndef RLIM_INFINITY
11913 +#warning RLIM_INFINITY is undefined
11916 +#define __rlim_val(l, r, v) ((l)->res[r].v)
11918 +#define __rlim_soft(l, r) __rlim_val(l, r, soft)
11919 +#define __rlim_hard(l, r) __rlim_val(l, r, hard)
11921 +#define __rlim_rcur(l, r) __rlim_val(l, r, rcur)
11922 +#define __rlim_rmin(l, r) __rlim_val(l, r, rmin)
11923 +#define __rlim_rmax(l, r) __rlim_val(l, r, rmax)
11925 +#define __rlim_lhit(l, r) __rlim_val(l, r, lhit)
11926 +#define __rlim_hit(l, r) atomic_inc(&__rlim_lhit(l, r))
11928 +typedef atomic_long_t rlim_atomic_t;
11929 +typedef unsigned long rlim_t;
11931 +#define __rlim_get(l, r) atomic_long_read(&__rlim_rcur(l, r))
11932 +#define __rlim_set(l, r, v) atomic_long_set(&__rlim_rcur(l, r), v)
11933 +#define __rlim_inc(l, r) atomic_long_inc(&__rlim_rcur(l, r))
11934 +#define __rlim_dec(l, r) atomic_long_dec(&__rlim_rcur(l, r))
11935 +#define __rlim_add(l, r, v) atomic_long_add(v, &__rlim_rcur(l, r))
11936 +#define __rlim_sub(l, r, v) atomic_long_sub(v, &__rlim_rcur(l, r))
11939 +#if (RLIM_INFINITY == VLIM_INFINITY)
11940 +#define VX_VLIM(r) ((long long)(long)(r))
11941 +#define VX_RLIM(v) ((rlim_t)(v))
11943 +#define VX_VLIM(r) (((r) == RLIM_INFINITY) \
11944 + ? VLIM_INFINITY : (long long)(r))
11945 +#define VX_RLIM(v) (((v) == VLIM_INFINITY) \
11946 + ? RLIM_INFINITY : (rlim_t)(v))
11951 +void vx_vsi_meminfo(struct sysinfo *);
11952 +void vx_vsi_swapinfo(struct sysinfo *);
11954 +#define NUM_LIMITS 24
11956 +#endif /* __KERNEL__ */
11957 +#endif /* _VX_LIMIT_H */
11958 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/limit_int.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_int.h
11959 --- linux-2.6.29.1/include/linux/vserver/limit_int.h 1970-01-01 01:00:00.000000000 +0100
11960 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/limit_int.h 2009-02-22 22:54:26.000000000 +0100
11962 +#ifndef _VX_LIMIT_INT_H
11963 +#define _VX_LIMIT_INT_H
11965 +#include "context.h"
11969 +#define VXD_RCRES_COND(r) VXD_CBIT(cres, r)
11970 +#define VXD_RLIMIT_COND(r) VXD_CBIT(limit, r)
11972 +extern const char *vlimit_name[NUM_LIMITS];
11974 +static inline void __vx_acc_cres(struct vx_info *vxi,
11975 + int res, int dir, void *_data, char *_file, int _line)
11977 + if (VXD_RCRES_COND(res))
11978 + vxlprintk(1, "vx_acc_cres[%5d,%s,%2d]: %5ld%s (%p)",
11979 + (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
11980 + (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
11981 + (dir > 0) ? "++" : "--", _data, _file, _line);
11986 + __rlim_inc(&vxi->limit, res);
11988 + __rlim_dec(&vxi->limit, res);
11991 +static inline void __vx_add_cres(struct vx_info *vxi,
11992 + int res, int amount, void *_data, char *_file, int _line)
11994 + if (VXD_RCRES_COND(res))
11995 + vxlprintk(1, "vx_add_cres[%5d,%s,%2d]: %5ld += %5d (%p)",
11996 + (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
11997 + (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
11998 + amount, _data, _file, _line);
12003 + __rlim_add(&vxi->limit, res, amount);
12007 +int __vx_cres_adjust_max(struct _vx_limit *limit, int res, rlim_t value)
12009 + int cond = (value > __rlim_rmax(limit, res));
12012 + __rlim_rmax(limit, res) = value;
12017 +int __vx_cres_adjust_min(struct _vx_limit *limit, int res, rlim_t value)
12019 + int cond = (value < __rlim_rmin(limit, res));
12022 + __rlim_rmin(limit, res) = value;
12027 +void __vx_cres_fixup(struct _vx_limit *limit, int res, rlim_t value)
12029 + if (!__vx_cres_adjust_max(limit, res, value))
12030 + __vx_cres_adjust_min(limit, res, value);
12035 + +1 ... no limit hit
12036 + -1 ... over soft limit
12037 + 0 ... over hard limit */
12039 +static inline int __vx_cres_avail(struct vx_info *vxi,
12040 + int res, int num, char *_file, int _line)
12042 + struct _vx_limit *limit;
12045 + if (VXD_RLIMIT_COND(res))
12046 + vxlprintk(1, "vx_cres_avail[%5d,%s,%2d]: %5ld/%5ld > %5ld + %5d",
12047 + (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
12048 + (vxi ? (long)__rlim_soft(&vxi->limit, res) : -1),
12049 + (vxi ? (long)__rlim_hard(&vxi->limit, res) : -1),
12050 + (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
12051 + num, _file, _line);
12055 + limit = &vxi->limit;
12056 + value = __rlim_get(limit, res);
12058 + if (!__vx_cres_adjust_max(limit, res, value))
12059 + __vx_cres_adjust_min(limit, res, value);
12064 + if (__rlim_soft(limit, res) == RLIM_INFINITY)
12066 + if (value + num <= __rlim_soft(limit, res))
12069 + if (__rlim_hard(limit, res) == RLIM_INFINITY)
12071 + if (value + num <= __rlim_hard(limit, res))
12074 + __rlim_hit(limit, res);
12079 +static const int VLA_RSS[] = { RLIMIT_RSS, VLIMIT_ANON, VLIMIT_MAPPED, 0 };
12082 +rlim_t __vx_cres_array_sum(struct _vx_limit *limit, const int *array)
12084 + rlim_t value, sum = 0;
12087 + while ((res = *array++)) {
12088 + value = __rlim_get(limit, res);
12089 + __vx_cres_fixup(limit, res, value);
12096 +rlim_t __vx_cres_array_fixup(struct _vx_limit *limit, const int *array)
12098 + rlim_t value = __vx_cres_array_sum(limit, array + 1);
12099 + int res = *array;
12101 + if (value == __rlim_get(limit, res))
12104 + __rlim_set(limit, res, value);
12105 + /* now adjust min/max */
12106 + if (!__vx_cres_adjust_max(limit, res, value))
12107 + __vx_cres_adjust_min(limit, res, value);
12112 +static inline int __vx_cres_array_avail(struct vx_info *vxi,
12113 + const int *array, int num, char *_file, int _line)
12115 + struct _vx_limit *limit;
12116 + rlim_t value = 0;
12124 + limit = &vxi->limit;
12126 + value = __vx_cres_array_sum(limit, array + 1);
12128 + __rlim_set(limit, res, value);
12129 + __vx_cres_fixup(limit, res, value);
12131 + return __vx_cres_avail(vxi, res, num, _file, _line);
12135 +static inline void vx_limit_fixup(struct _vx_limit *limit, int id)
12140 + /* complex resources first */
12141 + if ((id < 0) || (id == RLIMIT_RSS))
12142 + __vx_cres_array_fixup(limit, VLA_RSS);
12144 + for (res = 0; res < NUM_LIMITS; res++) {
12145 + if ((id > 0) && (res != id))
12148 + value = __rlim_get(limit, res);
12149 + __vx_cres_fixup(limit, res, value);
12151 + /* not supposed to happen, maybe warn? */
12152 + if (__rlim_rmax(limit, res) > __rlim_hard(limit, res))
12153 + __rlim_rmax(limit, res) = __rlim_hard(limit, res);
12158 +#endif /* __KERNEL__ */
12159 +#endif /* _VX_LIMIT_INT_H */
12160 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/monitor.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/monitor.h
12161 --- linux-2.6.29.1/include/linux/vserver/monitor.h 1970-01-01 01:00:00.000000000 +0100
12162 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/monitor.h 2009-02-22 22:54:26.000000000 +0100
12164 +#ifndef _VX_MONITOR_H
12165 +#define _VX_MONITOR_H
12167 +#include <linux/types.h>
12174 + VXM_UPDATE = 0x20,
12178 + VXM_RQINFO_1 = 0x24,
12181 + VXM_ACTIVATE = 0x40,
12188 + VXM_MIGRATE = 0x48,
12191 + /* all other bits are flags */
12192 + VXM_SCHED = 0x80,
12195 +struct _vxm_update_1 {
12196 + uint32_t tokens_max;
12197 + uint32_t fill_rate;
12198 + uint32_t interval;
12201 +struct _vxm_update_2 {
12202 + uint32_t tokens_min;
12203 + uint32_t fill_rate;
12204 + uint32_t interval;
12207 +struct _vxm_rqinfo_1 {
12208 + uint16_t running;
12212 + uint32_t idle_tokens;
12215 +struct _vxm_rqinfo_2 {
12216 + uint32_t norm_time;
12217 + uint32_t idle_time;
12218 + uint32_t idle_skip;
12221 +struct _vxm_sched {
12223 + uint32_t norm_time;
12224 + uint32_t idle_time;
12227 +struct _vxm_task {
12232 +struct _vxm_event {
12241 + struct _vxm_task tsk;
12245 +struct _vx_mon_entry {
12249 + struct _vxm_event ev;
12250 + struct _vxm_sched sd;
12251 + struct _vxm_update_1 u1;
12252 + struct _vxm_update_2 u2;
12253 + struct _vxm_rqinfo_1 q1;
12254 + struct _vxm_rqinfo_2 q2;
12259 +#endif /* _VX_MONITOR_H */
12260 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/network_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/network_cmd.h
12261 --- linux-2.6.29.1/include/linux/vserver/network_cmd.h 1970-01-01 01:00:00.000000000 +0100
12262 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/network_cmd.h 2009-02-22 22:54:26.000000000 +0100
12264 +#ifndef _VX_NETWORK_CMD_H
12265 +#define _VX_NETWORK_CMD_H
12268 +/* vinfo commands */
12270 +#define VCMD_task_nid VC_CMD(VINFO, 2, 0)
12273 +extern int vc_task_nid(uint32_t);
12275 +#endif /* __KERNEL__ */
12277 +#define VCMD_nx_info VC_CMD(VINFO, 6, 0)
12279 +struct vcmd_nx_info_v0 {
12281 + /* more to come */
12285 +extern int vc_nx_info(struct nx_info *, void __user *);
12287 +#endif /* __KERNEL__ */
12289 +#include <linux/in.h>
12290 +#include <linux/in6.h>
12292 +#define VCMD_net_create_v0 VC_CMD(VNET, 1, 0)
12293 +#define VCMD_net_create VC_CMD(VNET, 1, 1)
12295 +struct vcmd_net_create {
12296 + uint64_t flagword;
12299 +#define VCMD_net_migrate VC_CMD(NETMIG, 1, 0)
12301 +#define VCMD_net_add VC_CMD(NETALT, 1, 0)
12302 +#define VCMD_net_remove VC_CMD(NETALT, 2, 0)
12304 +struct vcmd_net_addr_v0 {
12307 + struct in_addr ip[4];
12308 + struct in_addr mask[4];
12311 +#define VCMD_net_add_ipv4 VC_CMD(NETALT, 1, 1)
12312 +#define VCMD_net_remove_ipv4 VC_CMD(NETALT, 2, 1)
12314 +struct vcmd_net_addr_ipv4_v1 {
12317 + struct in_addr ip;
12318 + struct in_addr mask;
12321 +#define VCMD_net_add_ipv6 VC_CMD(NETALT, 3, 1)
12322 +#define VCMD_net_remove_ipv6 VC_CMD(NETALT, 4, 1)
12324 +struct vcmd_net_addr_ipv6_v1 {
12328 + struct in6_addr ip;
12329 + struct in6_addr mask;
12332 +#define VCMD_add_match_ipv4 VC_CMD(NETALT, 5, 0)
12333 +#define VCMD_get_match_ipv4 VC_CMD(NETALT, 6, 0)
12335 +struct vcmd_match_ipv4_v0 {
12340 + struct in_addr ip;
12341 + struct in_addr ip2;
12342 + struct in_addr mask;
12345 +#define VCMD_add_match_ipv6 VC_CMD(NETALT, 7, 0)
12346 +#define VCMD_get_match_ipv6 VC_CMD(NETALT, 8, 0)
12348 +struct vcmd_match_ipv6_v0 {
12353 + struct in6_addr ip;
12354 + struct in6_addr ip2;
12355 + struct in6_addr mask;
12360 +extern int vc_net_create(uint32_t, void __user *);
12361 +extern int vc_net_migrate(struct nx_info *, void __user *);
12363 +extern int vc_net_add(struct nx_info *, void __user *);
12364 +extern int vc_net_remove(struct nx_info *, void __user *);
12366 +extern int vc_net_add_ipv4(struct nx_info *, void __user *);
12367 +extern int vc_net_remove_ipv4(struct nx_info *, void __user *);
12369 +extern int vc_net_add_ipv6(struct nx_info *, void __user *);
12370 +extern int vc_net_remove_ipv6(struct nx_info *, void __user *);
12372 +extern int vc_add_match_ipv4(struct nx_info *, void __user *);
12373 +extern int vc_get_match_ipv4(struct nx_info *, void __user *);
12375 +extern int vc_add_match_ipv6(struct nx_info *, void __user *);
12376 +extern int vc_get_match_ipv6(struct nx_info *, void __user *);
12378 +#endif /* __KERNEL__ */
12381 +/* flag commands */
12383 +#define VCMD_get_nflags VC_CMD(FLAGS, 5, 0)
12384 +#define VCMD_set_nflags VC_CMD(FLAGS, 6, 0)
12386 +struct vcmd_net_flags_v0 {
12387 + uint64_t flagword;
12392 +extern int vc_get_nflags(struct nx_info *, void __user *);
12393 +extern int vc_set_nflags(struct nx_info *, void __user *);
12395 +#endif /* __KERNEL__ */
12398 +/* network caps commands */
12400 +#define VCMD_get_ncaps VC_CMD(FLAGS, 7, 0)
12401 +#define VCMD_set_ncaps VC_CMD(FLAGS, 8, 0)
12403 +struct vcmd_net_caps_v0 {
12409 +extern int vc_get_ncaps(struct nx_info *, void __user *);
12410 +extern int vc_set_ncaps(struct nx_info *, void __user *);
12412 +#endif /* __KERNEL__ */
12413 +#endif /* _VX_CONTEXT_CMD_H */
12414 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/network.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/network.h
12415 --- linux-2.6.29.1/include/linux/vserver/network.h 1970-01-01 01:00:00.000000000 +0100
12416 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/network.h 2009-02-22 22:54:26.000000000 +0100
12418 +#ifndef _VX_NETWORK_H
12419 +#define _VX_NETWORK_H
12421 +#include <linux/types.h>
12424 +#define MAX_N_CONTEXT 65535 /* Arbitrary limit */
12427 +/* network flags */
12429 +#define NXF_INFO_PRIVATE 0x00000008
12431 +#define NXF_SINGLE_IP 0x00000100
12432 +#define NXF_LBACK_REMAP 0x00000200
12433 +#define NXF_LBACK_ALLOW 0x00000400
12435 +#define NXF_HIDE_NETIF 0x02000000
12436 +#define NXF_HIDE_LBACK 0x04000000
12438 +#define NXF_STATE_SETUP (1ULL << 32)
12439 +#define NXF_STATE_ADMIN (1ULL << 34)
12441 +#define NXF_SC_HELPER (1ULL << 36)
12442 +#define NXF_PERSISTENT (1ULL << 38)
12444 +#define NXF_ONE_TIME (0x0005ULL << 32)
12447 +#define NXF_INIT_SET (__nxf_init_set())
12449 +static inline uint64_t __nxf_init_set(void) {
12450 + return NXF_STATE_ADMIN
12451 +#ifdef CONFIG_VSERVER_AUTO_LBACK
12452 + | NXF_LBACK_REMAP
12455 +#ifdef CONFIG_VSERVER_AUTO_SINGLE
12458 + | NXF_HIDE_NETIF;
12462 +/* network caps */
12464 +#define NXC_TUN_CREATE 0x00000001
12466 +#define NXC_RAW_ICMP 0x00000100
12469 +/* address types */
12471 +#define NXA_TYPE_IPV4 0x0001
12472 +#define NXA_TYPE_IPV6 0x0002
12474 +#define NXA_TYPE_NONE 0x0000
12475 +#define NXA_TYPE_ANY 0x00FF
12477 +#define NXA_TYPE_ADDR 0x0010
12478 +#define NXA_TYPE_MASK 0x0020
12479 +#define NXA_TYPE_RANGE 0x0040
12481 +#define NXA_MASK_ALL (NXA_TYPE_ADDR | NXA_TYPE_MASK | NXA_TYPE_RANGE)
12483 +#define NXA_MOD_BCAST 0x0100
12484 +#define NXA_MOD_LBACK 0x0200
12486 +#define NXA_LOOPBACK 0x1000
12488 +#define NXA_MASK_BIND (NXA_MASK_ALL | NXA_MOD_BCAST | NXA_MOD_LBACK)
12489 +#define NXA_MASK_SHOW (NXA_MASK_ALL | NXA_LOOPBACK)
12493 +#include <linux/list.h>
12494 +#include <linux/spinlock.h>
12495 +#include <linux/rcupdate.h>
12496 +#include <linux/in.h>
12497 +#include <linux/in6.h>
12498 +#include <asm/atomic.h>
12500 +struct nx_addr_v4 {
12501 + struct nx_addr_v4 *next;
12502 + struct in_addr ip[2];
12503 + struct in_addr mask;
12508 +struct nx_addr_v6 {
12509 + struct nx_addr_v6 *next;
12510 + struct in6_addr ip;
12511 + struct in6_addr mask;
12518 + struct hlist_node nx_hlist; /* linked list of nxinfos */
12519 + nid_t nx_id; /* vnet id */
12520 + atomic_t nx_usecnt; /* usage count */
12521 + atomic_t nx_tasks; /* tasks count */
12522 + int nx_state; /* context state */
12524 + uint64_t nx_flags; /* network flag word */
12525 + uint64_t nx_ncaps; /* network capabilities */
12527 + struct in_addr v4_lback; /* Loopback address */
12528 + struct in_addr v4_bcast; /* Broadcast address */
12529 + struct nx_addr_v4 v4; /* First/Single ipv4 address */
12530 +#ifdef CONFIG_IPV6
12531 + struct nx_addr_v6 v6; /* First/Single ipv6 address */
12533 + char nx_name[65]; /* network context name */
12537 +/* status flags */
12539 +#define NXS_HASHED 0x0001
12540 +#define NXS_SHUTDOWN 0x0100
12541 +#define NXS_RELEASED 0x8000
12543 +extern struct nx_info *lookup_nx_info(int);
12545 +extern int get_nid_list(int, unsigned int *, int);
12546 +extern int nid_is_hashed(nid_t);
12548 +extern int nx_migrate_task(struct task_struct *, struct nx_info *);
12550 +extern long vs_net_change(struct nx_info *, unsigned int);
12555 +#define NX_IPV4(n) ((n)->v4.type != NXA_TYPE_NONE)
12556 +#ifdef CONFIG_IPV6
12557 +#define NX_IPV6(n) ((n)->v6.type != NXA_TYPE_NONE)
12559 +#define NX_IPV6(n) (0)
12562 +#endif /* __KERNEL__ */
12563 +#endif /* _VX_NETWORK_H */
12564 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/percpu.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/percpu.h
12565 --- linux-2.6.29.1/include/linux/vserver/percpu.h 1970-01-01 01:00:00.000000000 +0100
12566 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/percpu.h 2009-02-22 22:54:26.000000000 +0100
12568 +#ifndef _VX_PERCPU_H
12569 +#define _VX_PERCPU_H
12571 +#include "cvirt_def.h"
12572 +#include "sched_def.h"
12574 +struct _vx_percpu {
12575 + struct _vx_cvirt_pc cvirt;
12576 + struct _vx_sched_pc sched;
12579 +#define PERCPU_PERCTX (sizeof(struct _vx_percpu))
12581 +#endif /* _VX_PERCPU_H */
12582 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/pid.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/pid.h
12583 --- linux-2.6.29.1/include/linux/vserver/pid.h 1970-01-01 01:00:00.000000000 +0100
12584 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/pid.h 2009-02-22 22:54:26.000000000 +0100
12586 +#ifndef _VSERVER_PID_H
12587 +#define _VSERVER_PID_H
12589 +/* pid faking stuff */
12591 +#define vx_info_map_pid(v, p) \
12592 + __vx_info_map_pid((v), (p), __func__, __FILE__, __LINE__)
12593 +#define vx_info_map_tgid(v,p) vx_info_map_pid(v,p)
12594 +#define vx_map_pid(p) vx_info_map_pid(current->vx_info, p)
12595 +#define vx_map_tgid(p) vx_map_pid(p)
12597 +static inline int __vx_info_map_pid(struct vx_info *vxi, int pid,
12598 + const char *func, const char *file, int line)
12600 + if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
12601 + vxfprintk(VXD_CBIT(cvirt, 2),
12602 + "vx_map_tgid: %p/%llx: %d -> %d",
12603 + vxi, (long long)vxi->vx_flags, pid,
12604 + (pid && pid == vxi->vx_initpid) ? 1 : pid,
12605 + func, file, line);
12608 + if (pid == vxi->vx_initpid)
12614 +#define vx_info_rmap_pid(v, p) \
12615 + __vx_info_rmap_pid((v), (p), __func__, __FILE__, __LINE__)
12616 +#define vx_rmap_pid(p) vx_info_rmap_pid(current->vx_info, p)
12617 +#define vx_rmap_tgid(p) vx_rmap_pid(p)
12619 +static inline int __vx_info_rmap_pid(struct vx_info *vxi, int pid,
12620 + const char *func, const char *file, int line)
12622 + if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
12623 + vxfprintk(VXD_CBIT(cvirt, 2),
12624 + "vx_rmap_tgid: %p/%llx: %d -> %d",
12625 + vxi, (long long)vxi->vx_flags, pid,
12626 + (pid == 1) ? vxi->vx_initpid : pid,
12627 + func, file, line);
12628 + if ((pid == 1) && vxi->vx_initpid)
12629 + return vxi->vx_initpid;
12630 + if (pid == vxi->vx_initpid)
12637 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/sched_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched_cmd.h
12638 --- linux-2.6.29.1/include/linux/vserver/sched_cmd.h 1970-01-01 01:00:00.000000000 +0100
12639 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched_cmd.h 2009-02-22 22:54:26.000000000 +0100
12641 +#ifndef _VX_SCHED_CMD_H
12642 +#define _VX_SCHED_CMD_H
12645 +/* sched vserver commands */
12647 +#define VCMD_set_sched_v2 VC_CMD(SCHED, 1, 2)
12648 +#define VCMD_set_sched_v3 VC_CMD(SCHED, 1, 3)
12649 +#define VCMD_set_sched_v4 VC_CMD(SCHED, 1, 4)
12651 +struct vcmd_set_sched_v2 {
12652 + int32_t fill_rate;
12653 + int32_t interval;
12655 + int32_t tokens_min;
12656 + int32_t tokens_max;
12657 + uint64_t cpu_mask;
12660 +struct vcmd_set_sched_v3 {
12661 + uint32_t set_mask;
12662 + int32_t fill_rate;
12663 + int32_t interval;
12665 + int32_t tokens_min;
12666 + int32_t tokens_max;
12667 + int32_t priority_bias;
12670 +struct vcmd_set_sched_v4 {
12671 + uint32_t set_mask;
12672 + int32_t fill_rate;
12673 + int32_t interval;
12675 + int32_t tokens_min;
12676 + int32_t tokens_max;
12677 + int32_t prio_bias;
12679 + int32_t bucket_id;
12682 +#define VCMD_set_sched VC_CMD(SCHED, 1, 5)
12683 +#define VCMD_get_sched VC_CMD(SCHED, 2, 5)
12685 +struct vcmd_sched_v5 {
12688 + int32_t bucket_id;
12689 + int32_t fill_rate[2];
12690 + int32_t interval[2];
12692 + int32_t tokens_min;
12693 + int32_t tokens_max;
12694 + int32_t prio_bias;
12697 +#define VXSM_FILL_RATE 0x0001
12698 +#define VXSM_INTERVAL 0x0002
12699 +#define VXSM_FILL_RATE2 0x0004
12700 +#define VXSM_INTERVAL2 0x0008
12701 +#define VXSM_TOKENS 0x0010
12702 +#define VXSM_TOKENS_MIN 0x0020
12703 +#define VXSM_TOKENS_MAX 0x0040
12704 +#define VXSM_PRIO_BIAS 0x0100
12706 +#define VXSM_IDLE_TIME 0x0200
12707 +#define VXSM_FORCE 0x0400
12709 +#define VXSM_V3_MASK 0x0173
12710 +#define VXSM_SET_MASK 0x01FF
12712 +#define VXSM_CPU_ID 0x1000
12713 +#define VXSM_BUCKET_ID 0x2000
12715 +#define VXSM_MSEC 0x4000
12717 +#define SCHED_KEEP (-2) /* only for v2 */
12721 +#include <linux/compiler.h>
12723 +extern int vc_set_sched_v2(struct vx_info *, void __user *);
12724 +extern int vc_set_sched_v3(struct vx_info *, void __user *);
12725 +extern int vc_set_sched_v4(struct vx_info *, void __user *);
12726 +extern int vc_set_sched(struct vx_info *, void __user *);
12727 +extern int vc_get_sched(struct vx_info *, void __user *);
12729 +#endif /* __KERNEL__ */
12731 +#define VCMD_sched_info VC_CMD(SCHED, 3, 0)
12733 +struct vcmd_sched_info {
12735 + int32_t bucket_id;
12736 + uint64_t user_msec;
12737 + uint64_t sys_msec;
12738 + uint64_t hold_msec;
12739 + uint32_t token_usec;
12740 + int32_t vavavoom;
12745 +extern int vc_sched_info(struct vx_info *, void __user *);
12747 +#endif /* __KERNEL__ */
12748 +#endif /* _VX_SCHED_CMD_H */
12749 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/sched_def.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched_def.h
12750 --- linux-2.6.29.1/include/linux/vserver/sched_def.h 1970-01-01 01:00:00.000000000 +0100
12751 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched_def.h 2009-02-22 22:54:26.000000000 +0100
12753 +#ifndef _VX_SCHED_DEF_H
12754 +#define _VX_SCHED_DEF_H
12756 +#include <linux/spinlock.h>
12757 +#include <linux/jiffies.h>
12758 +#include <linux/cpumask.h>
12759 +#include <asm/atomic.h>
12760 +#include <asm/param.h>
12763 +/* context sub struct */
12765 +struct _vx_sched {
12766 + spinlock_t tokens_lock; /* lock for token bucket */
12768 + int tokens; /* number of CPU tokens */
12769 + int fill_rate[2]; /* Fill rate: add X tokens... */
12770 + int interval[2]; /* Divisor: per Y jiffies */
12771 + int tokens_min; /* Limit: minimum for unhold */
12772 + int tokens_max; /* Limit: no more than N tokens */
12774 + int prio_bias; /* bias offset for priority */
12776 + unsigned update_mask; /* which features should be updated */
12777 + cpumask_t update; /* CPUs which should update */
12780 +struct _vx_sched_pc {
12781 + int tokens; /* number of CPU tokens */
12782 + int flags; /* bucket flags */
12784 + int fill_rate[2]; /* Fill rate: add X tokens... */
12785 + int interval[2]; /* Divisor: per Y jiffies */
12786 + int tokens_min; /* Limit: minimum for unhold */
12787 + int tokens_max; /* Limit: no more than N tokens */
12789 + int prio_bias; /* bias offset for priority */
12790 + int vavavoom; /* last calculated vavavoom */
12792 + unsigned long norm_time; /* last time accounted */
12793 + unsigned long idle_time; /* non linear time for fair sched */
12794 + unsigned long token_time; /* token time for accounting */
12795 + unsigned long onhold; /* jiffies when put on hold */
12797 + uint64_t user_ticks; /* token tick events */
12798 + uint64_t sys_ticks; /* token tick events */
12799 + uint64_t hold_ticks; /* token ticks paused */
12803 +#define VXSF_ONHOLD 0x0001
12804 +#define VXSF_IDLE_TIME 0x0100
12806 +#ifdef CONFIG_VSERVER_DEBUG
12808 +static inline void __dump_vx_sched(struct _vx_sched *sched)
12810 + printk("\t_vx_sched:\n");
12811 + printk("\t tokens: %4d/%4d, %4d/%4d, %4d, %4d\n",
12812 + sched->fill_rate[0], sched->interval[0],
12813 + sched->fill_rate[1], sched->interval[1],
12814 + sched->tokens_min, sched->tokens_max);
12815 + printk("\t priority = %4d\n", sched->prio_bias);
12820 +#endif /* _VX_SCHED_DEF_H */
12821 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/sched.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched.h
12822 --- linux-2.6.29.1/include/linux/vserver/sched.h 1970-01-01 01:00:00.000000000 +0100
12823 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/sched.h 2009-02-22 22:54:26.000000000 +0100
12825 +#ifndef _VX_SCHED_H
12826 +#define _VX_SCHED_H
12833 +void vx_vsi_uptime(struct timespec *, struct timespec *);
12838 +void vx_update_load(struct vx_info *);
12841 +int vx_tokens_recalc(struct _vx_sched_pc *,
12842 + unsigned long *, unsigned long *, int [2]);
12844 +void vx_update_sched_param(struct _vx_sched *sched,
12845 + struct _vx_sched_pc *sched_pc);
12847 +#endif /* __KERNEL__ */
12848 +#else /* _VX_SCHED_H */
12849 +#warning duplicate inclusion
12850 +#endif /* _VX_SCHED_H */
12851 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/signal_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/signal_cmd.h
12852 --- linux-2.6.29.1/include/linux/vserver/signal_cmd.h 1970-01-01 01:00:00.000000000 +0100
12853 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/signal_cmd.h 2009-02-22 22:54:26.000000000 +0100
12855 +#ifndef _VX_SIGNAL_CMD_H
12856 +#define _VX_SIGNAL_CMD_H
12859 +/* signalling vserver commands */
12861 +#define VCMD_ctx_kill VC_CMD(PROCTRL, 1, 0)
12862 +#define VCMD_wait_exit VC_CMD(EVENT, 99, 0)
12864 +struct vcmd_ctx_kill_v0 {
12869 +struct vcmd_wait_exit_v0 {
12870 + int32_t reboot_cmd;
12871 + int32_t exit_code;
12876 +extern int vc_ctx_kill(struct vx_info *, void __user *);
12877 +extern int vc_wait_exit(struct vx_info *, void __user *);
12879 +#endif /* __KERNEL__ */
12881 +/* process alteration commands */
12883 +#define VCMD_get_pflags VC_CMD(PROCALT, 5, 0)
12884 +#define VCMD_set_pflags VC_CMD(PROCALT, 6, 0)
12886 +struct vcmd_pflags_v0 {
12887 + uint32_t flagword;
12893 +extern int vc_get_pflags(uint32_t pid, void __user *);
12894 +extern int vc_set_pflags(uint32_t pid, void __user *);
12896 +#endif /* __KERNEL__ */
12897 +#endif /* _VX_SIGNAL_CMD_H */
12898 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/signal.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/signal.h
12899 --- linux-2.6.29.1/include/linux/vserver/signal.h 1970-01-01 01:00:00.000000000 +0100
12900 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/signal.h 2009-02-22 22:54:26.000000000 +0100
12902 +#ifndef _VX_SIGNAL_H
12903 +#define _VX_SIGNAL_H
12910 +int vx_info_kill(struct vx_info *, int, int);
12912 +#endif /* __KERNEL__ */
12913 +#else /* _VX_SIGNAL_H */
12914 +#warning duplicate inclusion
12915 +#endif /* _VX_SIGNAL_H */
12916 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/space_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/space_cmd.h
12917 --- linux-2.6.29.1/include/linux/vserver/space_cmd.h 1970-01-01 01:00:00.000000000 +0100
12918 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/space_cmd.h 2009-02-22 22:54:26.000000000 +0100
12920 +#ifndef _VX_SPACE_CMD_H
12921 +#define _VX_SPACE_CMD_H
12924 +#define VCMD_enter_space_v0 VC_CMD(PROCALT, 1, 0)
12925 +#define VCMD_enter_space_v1 VC_CMD(PROCALT, 1, 1)
12926 +#define VCMD_enter_space VC_CMD(PROCALT, 1, 2)
12928 +#define VCMD_set_space_v0 VC_CMD(PROCALT, 3, 0)
12929 +#define VCMD_set_space_v1 VC_CMD(PROCALT, 3, 1)
12930 +#define VCMD_set_space VC_CMD(PROCALT, 3, 2)
12932 +#define VCMD_get_space_mask_v0 VC_CMD(PROCALT, 4, 0)
12934 +#define VCMD_get_space_mask VC_CMD(VSPACE, 0, 1)
12935 +#define VCMD_get_space_default VC_CMD(VSPACE, 1, 0)
12938 +struct vcmd_space_mask_v1 {
12942 +struct vcmd_space_mask_v2 {
12950 +extern int vc_enter_space_v1(struct vx_info *, void __user *);
12951 +extern int vc_set_space_v1(struct vx_info *, void __user *);
12952 +extern int vc_enter_space(struct vx_info *, void __user *);
12953 +extern int vc_set_space(struct vx_info *, void __user *);
12954 +extern int vc_get_space_mask(void __user *, int);
12956 +#endif /* __KERNEL__ */
12957 +#endif /* _VX_SPACE_CMD_H */
12958 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/space.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/space.h
12959 --- linux-2.6.29.1/include/linux/vserver/space.h 1970-01-01 01:00:00.000000000 +0100
12960 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/space.h 2009-02-22 22:54:26.000000000 +0100
12962 +#ifndef _VX_SPACE_H
12963 +#define _VX_SPACE_H
12965 +#include <linux/types.h>
12969 +int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index);
12971 +#else /* _VX_SPACE_H */
12972 +#warning duplicate inclusion
12973 +#endif /* _VX_SPACE_H */
12974 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/switch.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/switch.h
12975 --- linux-2.6.29.1/include/linux/vserver/switch.h 1970-01-01 01:00:00.000000000 +0100
12976 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/switch.h 2009-02-22 22:54:26.000000000 +0100
12978 +#ifndef _VX_SWITCH_H
12979 +#define _VX_SWITCH_H
12981 +#include <linux/types.h>
12984 +#define VC_CATEGORY(c) (((c) >> 24) & 0x3F)
12985 +#define VC_COMMAND(c) (((c) >> 16) & 0xFF)
12986 +#define VC_VERSION(c) ((c) & 0xFFF)
12988 +#define VC_CMD(c, i, v) ((((VC_CAT_ ## c) & 0x3F) << 24) \
12989 + | (((i) & 0xFF) << 16) | ((v) & 0xFFF))
12993 + Syscall Matrix V2.8
12995 + |VERSION|CREATE |MODIFY |MIGRATE|CONTROL|EXPERIM| |SPECIAL|SPECIAL|
12996 + |STATS |DESTROY|ALTER |CHANGE |LIMIT |TEST | | | |
12997 + |INFO |SETUP | |MOVE | | | | | |
12998 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
12999 + SYSTEM |VERSION|VSETUP |VHOST | | | | |DEVICE | |
13000 + HOST | 00| 01| 02| 03| 04| 05| | 06| 07|
13001 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13002 + CPU | |VPROC |PROCALT|PROCMIG|PROCTRL| | |SCHED. | |
13003 + PROCESS| 08| 09| 10| 11| 12| 13| | 14| 15|
13004 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13005 + MEMORY | | | | |MEMCTRL| | |SWAP | |
13006 + | 16| 17| 18| 19| 20| 21| | 22| 23|
13007 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13008 + NETWORK| |VNET |NETALT |NETMIG |NETCTL | | |SERIAL | |
13009 + | 24| 25| 26| 27| 28| 29| | 30| 31|
13010 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13011 + DISK | | | |TAGMIG |DLIMIT | | |INODE | |
13012 + VFS | 32| 33| 34| 35| 36| 37| | 38| 39|
13013 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13014 + OTHER |VSTAT | | | | | | |VINFO | |
13015 + | 40| 41| 42| 43| 44| 45| | 46| 47|
13016 + =======+=======+=======+=======+=======+=======+=======+ +=======+=======+
13017 + SPECIAL|EVENT | | | |FLAGS | | |VSPACE | |
13018 + | 48| 49| 50| 51| 52| 53| | 54| 55|
13019 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13020 + SPECIAL|DEBUG | | | |RLIMIT |SYSCALL| | |COMPAT |
13021 + | 56| 57| 58| 59| 60|TEST 61| | 62| 63|
13022 + -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13026 +#define VC_CAT_VERSION 0
13028 +#define VC_CAT_VSETUP 1
13029 +#define VC_CAT_VHOST 2
13031 +#define VC_CAT_DEVICE 6
13033 +#define VC_CAT_VPROC 9
13034 +#define VC_CAT_PROCALT 10
13035 +#define VC_CAT_PROCMIG 11
13036 +#define VC_CAT_PROCTRL 12
13038 +#define VC_CAT_SCHED 14
13039 +#define VC_CAT_MEMCTRL 20
13041 +#define VC_CAT_VNET 25
13042 +#define VC_CAT_NETALT 26
13043 +#define VC_CAT_NETMIG 27
13044 +#define VC_CAT_NETCTRL 28
13046 +#define VC_CAT_TAGMIG 35
13047 +#define VC_CAT_DLIMIT 36
13048 +#define VC_CAT_INODE 38
13050 +#define VC_CAT_VSTAT 40
13051 +#define VC_CAT_VINFO 46
13052 +#define VC_CAT_EVENT 48
13054 +#define VC_CAT_FLAGS 52
13055 +#define VC_CAT_VSPACE 54
13056 +#define VC_CAT_DEBUG 56
13057 +#define VC_CAT_RLIMIT 60
13059 +#define VC_CAT_SYSTEST 61
13060 +#define VC_CAT_COMPAT 63
13062 +/* query version */
13064 +#define VCMD_get_version VC_CMD(VERSION, 0, 0)
13065 +#define VCMD_get_vci VC_CMD(VERSION, 1, 0)
13070 +#include <linux/errno.h>
13072 +#endif /* __KERNEL__ */
13074 +#endif /* _VX_SWITCH_H */
13076 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/tag_cmd.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/tag_cmd.h
13077 --- linux-2.6.29.1/include/linux/vserver/tag_cmd.h 1970-01-01 01:00:00.000000000 +0100
13078 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/tag_cmd.h 2009-02-22 22:54:26.000000000 +0100
13080 +#ifndef _VX_TAG_CMD_H
13081 +#define _VX_TAG_CMD_H
13084 +/* vinfo commands */
13086 +#define VCMD_task_tag VC_CMD(VINFO, 3, 0)
13089 +extern int vc_task_tag(uint32_t);
13091 +#endif /* __KERNEL__ */
13093 +/* context commands */
13095 +#define VCMD_tag_migrate VC_CMD(TAGMIG, 1, 0)
13098 +extern int vc_tag_migrate(uint32_t);
13100 +#endif /* __KERNEL__ */
13101 +#endif /* _VX_TAG_CMD_H */
13102 diff -NurpP --minimal linux-2.6.29.1/include/linux/vserver/tag.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/tag.h
13103 --- linux-2.6.29.1/include/linux/vserver/tag.h 1970-01-01 01:00:00.000000000 +0100
13104 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vserver/tag.h 2009-02-22 22:54:26.000000000 +0100
13109 +#include <linux/types.h>
13112 +#define DX_TAG(in) (IS_TAGGED(in))
13115 +#ifdef CONFIG_TAG_NFSD
13116 +#define DX_TAG_NFSD 1
13118 +#define DX_TAG_NFSD 0
13122 +#ifdef CONFIG_TAGGING_NONE
13124 +#define MAX_UID 0xFFFFFFFF
13125 +#define MAX_GID 0xFFFFFFFF
13127 +#define INOTAG_TAG(cond, uid, gid, tag) (0)
13129 +#define TAGINO_UID(cond, uid, tag) (uid)
13130 +#define TAGINO_GID(cond, gid, tag) (gid)
13135 +#ifdef CONFIG_TAGGING_GID16
13137 +#define MAX_UID 0xFFFFFFFF
13138 +#define MAX_GID 0x0000FFFF
13140 +#define INOTAG_TAG(cond, uid, gid, tag) \
13141 + ((cond) ? (((gid) >> 16) & 0xFFFF) : 0)
13143 +#define TAGINO_UID(cond, uid, tag) (uid)
13144 +#define TAGINO_GID(cond, gid, tag) \
13145 + ((cond) ? (((gid) & 0xFFFF) | ((tag) << 16)) : (gid))
13150 +#ifdef CONFIG_TAGGING_ID24
13152 +#define MAX_UID 0x00FFFFFF
13153 +#define MAX_GID 0x00FFFFFF
13155 +#define INOTAG_TAG(cond, uid, gid, tag) \
13156 + ((cond) ? ((((uid) >> 16) & 0xFF00) | (((gid) >> 24) & 0xFF)) : 0)
13158 +#define TAGINO_UID(cond, uid, tag) \
13159 + ((cond) ? (((uid) & 0xFFFFFF) | (((tag) & 0xFF00) << 16)) : (uid))
13160 +#define TAGINO_GID(cond, gid, tag) \
13161 + ((cond) ? (((gid) & 0xFFFFFF) | (((tag) & 0x00FF) << 24)) : (gid))
13166 +#ifdef CONFIG_TAGGING_UID16
13168 +#define MAX_UID 0x0000FFFF
13169 +#define MAX_GID 0xFFFFFFFF
13171 +#define INOTAG_TAG(cond, uid, gid, tag) \
13172 + ((cond) ? (((uid) >> 16) & 0xFFFF) : 0)
13174 +#define TAGINO_UID(cond, uid, tag) \
13175 + ((cond) ? (((uid) & 0xFFFF) | ((tag) << 16)) : (uid))
13176 +#define TAGINO_GID(cond, gid, tag) (gid)
13181 +#ifdef CONFIG_TAGGING_INTERN
13183 +#define MAX_UID 0xFFFFFFFF
13184 +#define MAX_GID 0xFFFFFFFF
13186 +#define INOTAG_TAG(cond, uid, gid, tag) \
13187 + ((cond) ? (tag) : 0)
13189 +#define TAGINO_UID(cond, uid, tag) (uid)
13190 +#define TAGINO_GID(cond, gid, tag) (gid)
13195 +#ifndef CONFIG_TAGGING_NONE
13196 +#define dx_current_fstag(sb) \
13197 + ((sb)->s_flags & MS_TAGGED ? dx_current_tag() : 0)
13199 +#define dx_current_fstag(sb) (0)
13202 +#ifndef CONFIG_TAGGING_INTERN
13203 +#define TAGINO_TAG(cond, tag) (0)
13205 +#define TAGINO_TAG(cond, tag) ((cond) ? (tag) : 0)
13208 +#define INOTAG_UID(cond, uid, gid) \
13209 + ((cond) ? ((uid) & MAX_UID) : (uid))
13210 +#define INOTAG_GID(cond, uid, gid) \
13211 + ((cond) ? ((gid) & MAX_GID) : (gid))
13214 +static inline uid_t dx_map_uid(uid_t uid)
13216 + if ((uid > MAX_UID) && (uid != -1))
13218 + return (uid & MAX_UID);
13221 +static inline gid_t dx_map_gid(gid_t gid)
13223 + if ((gid > MAX_GID) && (gid != -1))
13225 + return (gid & MAX_GID);
13233 +#define dx_notagcheck(sb) ((sb) && ((sb)->s_flags & MS_NOTAGCHECK))
13235 +int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
13236 + unsigned long *flags);
13238 +#ifdef CONFIG_PROPAGATE
13240 +void __dx_propagate_tag(struct nameidata *nd, struct inode *inode);
13242 +#define dx_propagate_tag(n, i) __dx_propagate_tag(n, i)
13245 +#define dx_propagate_tag(n, i) do { } while (0)
13248 +#endif /* _DX_TAG_H */
13249 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_inet6.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_inet6.h
13250 --- linux-2.6.29.1/include/linux/vs_inet6.h 1970-01-01 01:00:00.000000000 +0100
13251 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_inet6.h 2009-03-25 00:46:50.000000000 +0100
13253 +#ifndef _VS_INET6_H
13254 +#define _VS_INET6_H
13256 +#include "vserver/base.h"
13257 +#include "vserver/network.h"
13258 +#include "vserver/debug.h"
13260 +#include <net/ipv6.h>
13262 +#define NXAV6(a) &(a)->ip, &(a)->mask, (a)->prefix, (a)->type
13263 +#define NXAV6_FMT "[%pI6/%pI6/%d:%04x]"
13266 +#ifdef CONFIG_IPV6
13269 +int v6_addr_match(struct nx_addr_v6 *nxa,
13270 + const struct in6_addr *addr, uint16_t mask)
13274 + switch (nxa->type & mask) {
13275 + case NXA_TYPE_MASK:
13276 + ret = ipv6_masked_addr_cmp(&nxa->ip, &nxa->mask, addr);
13278 + case NXA_TYPE_ADDR:
13279 + ret = ipv6_addr_equal(&nxa->ip, addr);
13281 + case NXA_TYPE_ANY:
13285 + vxdprintk(VXD_CBIT(net, 0),
13286 + "v6_addr_match(%p" NXAV6_FMT ",%pI6,%04x) = %d",
13287 + nxa, NXAV6(nxa), addr, mask, ret);
13292 +int v6_addr_in_nx_info(struct nx_info *nxi,
13293 + const struct in6_addr *addr, uint16_t mask)
13295 + struct nx_addr_v6 *nxa;
13300 + for (nxa = &nxi->v6; nxa; nxa = nxa->next)
13301 + if (v6_addr_match(nxa, addr, mask))
13305 + vxdprintk(VXD_CBIT(net, 0),
13306 + "v6_addr_in_nx_info(%p[#%u],%pI6,%04x) = %d",
13307 + nxi, nxi ? nxi->nx_id : 0, addr, mask, ret);
13312 +int v6_nx_addr_match(struct nx_addr_v6 *nxa, struct nx_addr_v6 *addr, uint16_t mask)
13314 + /* FIXME: needs full range checks */
13315 + return v6_addr_match(nxa, &addr->ip, mask);
13319 +int v6_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v6 *nxa, uint16_t mask)
13321 + struct nx_addr_v6 *ptr;
13323 + for (ptr = &nxi->v6; ptr; ptr = ptr->next)
13324 + if (v6_nx_addr_match(ptr, nxa, mask))
13331 + * Check if a given address matches for a socket
13333 + * nxi: the socket's nx_info if any
13334 + * addr: to be verified address
13337 +int v6_sock_addr_match (
13338 + struct nx_info *nxi,
13339 + struct inet_sock *inet,
13340 + struct in6_addr *addr)
13342 + struct sock *sk = &inet->sk;
13343 + struct in6_addr *saddr = inet6_rcv_saddr(sk);
13345 + if (!ipv6_addr_any(addr) &&
13346 + ipv6_addr_equal(saddr, addr))
13348 + if (ipv6_addr_any(saddr))
13349 + return v6_addr_in_nx_info(nxi, addr, -1);
13354 + * check if address is covered by socket
13356 + * sk: the socket to check against
13357 + * addr: the address in question (must be != 0)
13361 +int __v6_addr_match_socket(const struct sock *sk, struct nx_addr_v6 *nxa)
13363 + struct nx_info *nxi = sk->sk_nx_info;
13364 + struct in6_addr *saddr = inet6_rcv_saddr(sk);
13366 + vxdprintk(VXD_CBIT(net, 5),
13367 + "__v6_addr_in_socket(%p," NXAV6_FMT ") %p:%pI6 %p;%lx",
13368 + sk, NXAV6(nxa), nxi, saddr, sk->sk_socket,
13369 + (sk->sk_socket?sk->sk_socket->flags:0));
13371 + if (!ipv6_addr_any(saddr)) { /* direct address match */
13372 + return v6_addr_match(nxa, saddr, -1);
13373 + } else if (nxi) { /* match against nx_info */
13374 + return v6_nx_addr_in_nx_info(nxi, nxa, -1);
13375 + } else { /* unrestricted any socket */
13381 +/* inet related checks and helpers */
13385 +struct net_device;
13389 +#include <linux/netdevice.h>
13390 +#include <linux/inetdevice.h>
13391 +#include <net/inet_timewait_sock.h>
13394 +int dev_in_nx_info(struct net_device *, struct nx_info *);
13395 +int v6_dev_in_nx_info(struct net_device *, struct nx_info *);
13396 +int nx_v6_addr_conflict(struct nx_info *, struct nx_info *);
13401 +int v6_ifa_in_nx_info(struct inet6_ifaddr *ifa, struct nx_info *nxi)
13407 + return v6_addr_in_nx_info(nxi, &ifa->addr, -1);
13411 +int nx_v6_ifa_visible(struct nx_info *nxi, struct inet6_ifaddr *ifa)
13413 + vxdprintk(VXD_CBIT(net, 1), "nx_v6_ifa_visible(%p[#%u],%p) %d",
13414 + nxi, nxi ? nxi->nx_id : 0, ifa,
13415 + nxi ? v6_ifa_in_nx_info(ifa, nxi) : 0);
13417 + if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13419 + if (v6_ifa_in_nx_info(ifa, nxi))
13425 +struct nx_v6_sock_addr {
13426 + struct in6_addr saddr; /* Address used for validation */
13427 + struct in6_addr baddr; /* Address used for socket bind */
13431 +int v6_map_sock_addr(struct inet_sock *inet, struct sockaddr_in6 *addr,
13432 + struct nx_v6_sock_addr *nsa)
13434 + // struct sock *sk = &inet->sk;
13435 + // struct nx_info *nxi = sk->sk_nx_info;
13436 + struct in6_addr saddr = addr->sin6_addr;
13437 + struct in6_addr baddr = saddr;
13439 + nsa->saddr = saddr;
13440 + nsa->baddr = baddr;
13445 +void v6_set_sock_addr(struct inet_sock *inet, struct nx_v6_sock_addr *nsa)
13447 + // struct sock *sk = &inet->sk;
13448 + // struct in6_addr *saddr = inet6_rcv_saddr(sk);
13450 + // *saddr = nsa->baddr;
13451 + // inet->saddr = nsa->baddr;
13455 +int nx_info_has_v6(struct nx_info *nxi)
13459 + if (NX_IPV6(nxi))
13464 +#else /* CONFIG_IPV6 */
13467 +int nx_v6_dev_visible(struct nx_info *n, struct net_device *d)
13474 +int nx_v6_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
13480 +int v6_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
13486 +int nx_info_has_v6(struct nx_info *nxi)
13491 +#endif /* CONFIG_IPV6 */
13493 +#define current_nx_info_has_v6() \
13494 + nx_info_has_v6(current_nx_info())
13497 +#warning duplicate inclusion
13499 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_inet.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_inet.h
13500 --- linux-2.6.29.1/include/linux/vs_inet.h 1970-01-01 01:00:00.000000000 +0100
13501 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_inet.h 2009-02-22 22:54:26.000000000 +0100
13503 +#ifndef _VS_INET_H
13504 +#define _VS_INET_H
13506 +#include "vserver/base.h"
13507 +#include "vserver/network.h"
13508 +#include "vserver/debug.h"
13510 +#define IPI_LOOPBACK htonl(INADDR_LOOPBACK)
13512 +#define NXAV4(a) NIPQUAD((a)->ip[0]), NIPQUAD((a)->ip[1]), \
13513 + NIPQUAD((a)->mask), (a)->type
13514 +#define NXAV4_FMT "[" NIPQUAD_FMT "-" NIPQUAD_FMT "/" NIPQUAD_FMT ":%04x]"
13518 +int v4_addr_match(struct nx_addr_v4 *nxa, __be32 addr, uint16_t tmask)
13520 + __be32 ip = nxa->ip[0].s_addr;
13521 + __be32 mask = nxa->mask.s_addr;
13522 + __be32 bcast = ip | ~mask;
13525 + switch (nxa->type & tmask) {
13526 + case NXA_TYPE_MASK:
13527 + ret = (ip == (addr & mask));
13529 + case NXA_TYPE_ADDR:
13533 + /* fall through to broadcast */
13534 + case NXA_MOD_BCAST:
13535 + ret = ((tmask & NXA_MOD_BCAST) && (addr == bcast));
13537 + case NXA_TYPE_RANGE:
13538 + ret = ((nxa->ip[0].s_addr <= addr) &&
13539 + (nxa->ip[1].s_addr > addr));
13541 + case NXA_TYPE_ANY:
13546 + vxdprintk(VXD_CBIT(net, 0),
13547 + "v4_addr_match(%p" NXAV4_FMT "," NIPQUAD_FMT ",%04x) = %d",
13548 + nxa, NXAV4(nxa), NIPQUAD(addr), tmask, ret);
13553 +int v4_addr_in_nx_info(struct nx_info *nxi, __be32 addr, uint16_t tmask)
13555 + struct nx_addr_v4 *nxa;
13562 + /* allow 127.0.0.1 when remapping lback */
13563 + if ((tmask & NXA_LOOPBACK) &&
13564 + (addr == IPI_LOOPBACK) &&
13565 + nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13568 + /* check for lback address */
13569 + if ((tmask & NXA_MOD_LBACK) &&
13570 + (nxi->v4_lback.s_addr == addr))
13573 + /* check for broadcast address */
13574 + if ((tmask & NXA_MOD_BCAST) &&
13575 + (nxi->v4_bcast.s_addr == addr))
13578 + /* check for v4 addresses */
13579 + for (nxa = &nxi->v4; nxa; nxa = nxa->next)
13580 + if (v4_addr_match(nxa, addr, tmask))
13584 + vxdprintk(VXD_CBIT(net, 0),
13585 + "v4_addr_in_nx_info(%p[#%u]," NIPQUAD_FMT ",%04x) = %d",
13586 + nxi, nxi ? nxi->nx_id : 0, NIPQUAD(addr), tmask, ret);
13591 +int v4_nx_addr_match(struct nx_addr_v4 *nxa, struct nx_addr_v4 *addr, uint16_t mask)
13593 + /* FIXME: needs full range checks */
13594 + return v4_addr_match(nxa, addr->ip[0].s_addr, mask);
13598 +int v4_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v4 *nxa, uint16_t mask)
13600 + struct nx_addr_v4 *ptr;
13602 + for (ptr = &nxi->v4; ptr; ptr = ptr->next)
13603 + if (v4_nx_addr_match(ptr, nxa, mask))
13608 +#include <net/inet_sock.h>
13611 + * Check if a given address matches for a socket
13613 + * nxi: the socket's nx_info if any
13614 + * addr: to be verified address
13617 +int v4_sock_addr_match (
13618 + struct nx_info *nxi,
13619 + struct inet_sock *inet,
13622 + __be32 saddr = inet->rcv_saddr;
13623 + __be32 bcast = nxi ? nxi->v4_bcast.s_addr : INADDR_BROADCAST;
13625 + if (addr && (saddr == addr || bcast == addr))
13628 + return v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND);
13633 +/* inet related checks and helpers */
13637 +struct net_device;
13640 +#ifdef CONFIG_INET
13642 +#include <linux/netdevice.h>
13643 +#include <linux/inetdevice.h>
13644 +#include <net/inet_sock.h>
13645 +#include <net/inet_timewait_sock.h>
13648 +int dev_in_nx_info(struct net_device *, struct nx_info *);
13649 +int v4_dev_in_nx_info(struct net_device *, struct nx_info *);
13650 +int nx_v4_addr_conflict(struct nx_info *, struct nx_info *);
13654 + * check if address is covered by socket
13656 + * sk: the socket to check against
13657 + * addr: the address in question (must be != 0)
13661 +int __v4_addr_match_socket(const struct sock *sk, struct nx_addr_v4 *nxa)
13663 + struct nx_info *nxi = sk->sk_nx_info;
13664 + __be32 saddr = inet_rcv_saddr(sk);
13666 + vxdprintk(VXD_CBIT(net, 5),
13667 + "__v4_addr_in_socket(%p," NXAV4_FMT ") %p:" NIPQUAD_FMT " %p;%lx",
13668 + sk, NXAV4(nxa), nxi, NIPQUAD(saddr), sk->sk_socket,
13669 + (sk->sk_socket?sk->sk_socket->flags:0));
13671 + if (saddr) { /* direct address match */
13672 + return v4_addr_match(nxa, saddr, -1);
13673 + } else if (nxi) { /* match against nx_info */
13674 + return v4_nx_addr_in_nx_info(nxi, nxa, -1);
13675 + } else { /* unrestricted any socket */
13683 +int nx_dev_visible(struct nx_info *nxi, struct net_device *dev)
13685 + vxdprintk(VXD_CBIT(net, 1), "nx_dev_visible(%p[#%u],%p »%s«) %d",
13686 + nxi, nxi ? nxi->nx_id : 0, dev, dev->name,
13687 + nxi ? dev_in_nx_info(dev, nxi) : 0);
13689 + if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13691 + if (dev_in_nx_info(dev, nxi))
13698 +int v4_ifa_in_nx_info(struct in_ifaddr *ifa, struct nx_info *nxi)
13704 + return v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW);
13708 +int nx_v4_ifa_visible(struct nx_info *nxi, struct in_ifaddr *ifa)
13710 + vxdprintk(VXD_CBIT(net, 1), "nx_v4_ifa_visible(%p[#%u],%p) %d",
13711 + nxi, nxi ? nxi->nx_id : 0, ifa,
13712 + nxi ? v4_ifa_in_nx_info(ifa, nxi) : 0);
13714 + if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13716 + if (v4_ifa_in_nx_info(ifa, nxi))
13722 +struct nx_v4_sock_addr {
13723 + __be32 saddr; /* Address used for validation */
13724 + __be32 baddr; /* Address used for socket bind */
13728 +int v4_map_sock_addr(struct inet_sock *inet, struct sockaddr_in *addr,
13729 + struct nx_v4_sock_addr *nsa)
13731 + struct sock *sk = &inet->sk;
13732 + struct nx_info *nxi = sk->sk_nx_info;
13733 + __be32 saddr = addr->sin_addr.s_addr;
13734 + __be32 baddr = saddr;
13736 + vxdprintk(VXD_CBIT(net, 3),
13737 + "inet_bind(%p)* %p,%p;%lx " NIPQUAD_FMT,
13738 + sk, sk->sk_nx_info, sk->sk_socket,
13739 + (sk->sk_socket ? sk->sk_socket->flags : 0),
13743 + if (saddr == INADDR_ANY) {
13744 + if (nx_info_flags(nxi, NXF_SINGLE_IP, 0))
13745 + baddr = nxi->v4.ip[0].s_addr;
13746 + } else if (saddr == IPI_LOOPBACK) {
13747 + if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13748 + baddr = nxi->v4_lback.s_addr;
13749 + } else { /* normal address bind */
13750 + if (!v4_addr_in_nx_info(nxi, saddr, NXA_MASK_BIND))
13751 + return -EADDRNOTAVAIL;
13755 + vxdprintk(VXD_CBIT(net, 3),
13756 + "inet_bind(%p) " NIPQUAD_FMT ", " NIPQUAD_FMT,
13757 + sk, NIPQUAD(saddr), NIPQUAD(baddr));
13759 + nsa->saddr = saddr;
13760 + nsa->baddr = baddr;
13765 +void v4_set_sock_addr(struct inet_sock *inet, struct nx_v4_sock_addr *nsa)
13767 + inet->saddr = nsa->baddr;
13768 + inet->rcv_saddr = nsa->baddr;
13773 + * helper to simplify inet_lookup_listener
13775 + * nxi: the socket's nx_info if any
13776 + * addr: to be verified address
13777 + * saddr: socket address
13779 +static inline int v4_inet_addr_match (
13780 + struct nx_info *nxi,
13784 + if (addr && (saddr == addr))
13787 + return nxi ? v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND) : 1;
13791 +static inline __be32 nx_map_sock_lback(struct nx_info *nxi, __be32 addr)
13793 + if (nx_info_flags(nxi, NXF_HIDE_LBACK, 0) &&
13794 + (addr == nxi->v4_lback.s_addr))
13795 + return IPI_LOOPBACK;
13800 +int nx_info_has_v4(struct nx_info *nxi)
13804 + if (NX_IPV4(nxi))
13806 + if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13811 +#else /* CONFIG_INET */
13814 +int nx_dev_visible(struct nx_info *n, struct net_device *d)
13820 +int nx_v4_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
13826 +int v4_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
13832 +int nx_info_has_v4(struct nx_info *nxi)
13837 +#endif /* CONFIG_INET */
13839 +#define current_nx_info_has_v4() \
13840 + nx_info_has_v4(current_nx_info())
13843 +// #warning duplicate inclusion
13845 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_limit.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_limit.h
13846 --- linux-2.6.29.1/include/linux/vs_limit.h 1970-01-01 01:00:00.000000000 +0100
13847 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_limit.h 2009-02-22 22:54:26.000000000 +0100
13849 +#ifndef _VS_LIMIT_H
13850 +#define _VS_LIMIT_H
13852 +#include "vserver/limit.h"
13853 +#include "vserver/base.h"
13854 +#include "vserver/context.h"
13855 +#include "vserver/debug.h"
13856 +#include "vserver/context.h"
13857 +#include "vserver/limit_int.h"
13860 +#define vx_acc_cres(v, d, p, r) \
13861 + __vx_acc_cres(v, r, d, p, __FILE__, __LINE__)
13863 +#define vx_acc_cres_cond(x, d, p, r) \
13864 + __vx_acc_cres(((x) == vx_current_xid()) ? current->vx_info : 0, \
13865 + r, d, p, __FILE__, __LINE__)
13868 +#define vx_add_cres(v, a, p, r) \
13869 + __vx_add_cres(v, r, a, p, __FILE__, __LINE__)
13870 +#define vx_sub_cres(v, a, p, r) vx_add_cres(v, -(a), p, r)
13872 +#define vx_add_cres_cond(x, a, p, r) \
13873 + __vx_add_cres(((x) == vx_current_xid()) ? current->vx_info : 0, \
13874 + r, a, p, __FILE__, __LINE__)
13875 +#define vx_sub_cres_cond(x, a, p, r) vx_add_cres_cond(x, -(a), p, r)
13878 +/* process and file limits */
13880 +#define vx_nproc_inc(p) \
13881 + vx_acc_cres((p)->vx_info, 1, p, RLIMIT_NPROC)
13883 +#define vx_nproc_dec(p) \
13884 + vx_acc_cres((p)->vx_info,-1, p, RLIMIT_NPROC)
13886 +#define vx_files_inc(f) \
13887 + vx_acc_cres_cond((f)->f_xid, 1, f, RLIMIT_NOFILE)
13889 +#define vx_files_dec(f) \
13890 + vx_acc_cres_cond((f)->f_xid,-1, f, RLIMIT_NOFILE)
13892 +#define vx_locks_inc(l) \
13893 + vx_acc_cres_cond((l)->fl_xid, 1, l, RLIMIT_LOCKS)
13895 +#define vx_locks_dec(l) \
13896 + vx_acc_cres_cond((l)->fl_xid,-1, l, RLIMIT_LOCKS)
13898 +#define vx_openfd_inc(f) \
13899 + vx_acc_cres(current->vx_info, 1, (void *)(long)(f), VLIMIT_OPENFD)
13901 +#define vx_openfd_dec(f) \
13902 + vx_acc_cres(current->vx_info,-1, (void *)(long)(f), VLIMIT_OPENFD)
13905 +#define vx_cres_avail(v, n, r) \
13906 + __vx_cres_avail(v, r, n, __FILE__, __LINE__)
13909 +#define vx_nproc_avail(n) \
13910 + vx_cres_avail(current->vx_info, n, RLIMIT_NPROC)
13912 +#define vx_files_avail(n) \
13913 + vx_cres_avail(current->vx_info, n, RLIMIT_NOFILE)
13915 +#define vx_locks_avail(n) \
13916 + vx_cres_avail(current->vx_info, n, RLIMIT_LOCKS)
13918 +#define vx_openfd_avail(n) \
13919 + vx_cres_avail(current->vx_info, n, VLIMIT_OPENFD)
13922 +/* dentry limits */
13924 +#define vx_dentry_inc(d) do { \
13925 + if (atomic_read(&d->d_count) == 1) \
13926 + vx_acc_cres(current->vx_info, 1, d, VLIMIT_DENTRY); \
13929 +#define vx_dentry_dec(d) do { \
13930 + if (atomic_read(&d->d_count) == 0) \
13931 + vx_acc_cres(current->vx_info,-1, d, VLIMIT_DENTRY); \
13934 +#define vx_dentry_avail(n) \
13935 + vx_cres_avail(current->vx_info, n, VLIMIT_DENTRY)
13938 +/* socket limits */
13940 +#define vx_sock_inc(s) \
13941 + vx_acc_cres((s)->sk_vx_info, 1, s, VLIMIT_NSOCK)
13943 +#define vx_sock_dec(s) \
13944 + vx_acc_cres((s)->sk_vx_info,-1, s, VLIMIT_NSOCK)
13946 +#define vx_sock_avail(n) \
13947 + vx_cres_avail(current->vx_info, n, VLIMIT_NSOCK)
13950 +/* ipc resource limits */
13952 +#define vx_ipcmsg_add(v, u, a) \
13953 + vx_add_cres(v, a, u, RLIMIT_MSGQUEUE)
13955 +#define vx_ipcmsg_sub(v, u, a) \
13956 + vx_sub_cres(v, a, u, RLIMIT_MSGQUEUE)
13958 +#define vx_ipcmsg_avail(v, a) \
13959 + vx_cres_avail(v, a, RLIMIT_MSGQUEUE)
13962 +#define vx_ipcshm_add(v, k, a) \
13963 + vx_add_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
13965 +#define vx_ipcshm_sub(v, k, a) \
13966 + vx_sub_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
13968 +#define vx_ipcshm_avail(v, a) \
13969 + vx_cres_avail(v, a, VLIMIT_SHMEM)
13972 +#define vx_semary_inc(a) \
13973 + vx_acc_cres(current->vx_info, 1, a, VLIMIT_SEMARY)
13975 +#define vx_semary_dec(a) \
13976 + vx_acc_cres(current->vx_info, -1, a, VLIMIT_SEMARY)
13979 +#define vx_nsems_add(a,n) \
13980 + vx_add_cres(current->vx_info, n, a, VLIMIT_NSEMS)
13982 +#define vx_nsems_sub(a,n) \
13983 + vx_sub_cres(current->vx_info, n, a, VLIMIT_NSEMS)
13987 +#warning duplicate inclusion
13989 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_memory.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_memory.h
13990 --- linux-2.6.29.1/include/linux/vs_memory.h 1970-01-01 01:00:00.000000000 +0100
13991 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_memory.h 2009-02-22 22:54:26.000000000 +0100
13993 +#ifndef _VS_MEMORY_H
13994 +#define _VS_MEMORY_H
13996 +#include "vserver/limit.h"
13997 +#include "vserver/base.h"
13998 +#include "vserver/context.h"
13999 +#include "vserver/debug.h"
14000 +#include "vserver/context.h"
14001 +#include "vserver/limit_int.h"
14004 +#define __acc_add_long(a, v) (*(v) += (a))
14005 +#define __acc_inc_long(v) (++*(v))
14006 +#define __acc_dec_long(v) (--*(v))
14008 +#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
14009 +#define __acc_add_atomic(a, v) atomic_long_add(a, v)
14010 +#define __acc_inc_atomic(v) atomic_long_inc(v)
14011 +#define __acc_dec_atomic(v) atomic_long_dec(v)
14012 +#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
14013 +#define __acc_add_atomic(a, v) __acc_add_long(a, v)
14014 +#define __acc_inc_atomic(v) __acc_inc_long(v)
14015 +#define __acc_dec_atomic(v) __acc_dec_long(v)
14016 +#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
14019 +#define vx_acc_page(m, d, v, r) do { \
14021 + __acc_inc_long(&(m)->v); \
14023 + __acc_dec_long(&(m)->v); \
14024 + __vx_acc_cres(m->mm_vx_info, r, d, m, __FILE__, __LINE__); \
14027 +#define vx_acc_page_atomic(m, d, v, r) do { \
14029 + __acc_inc_atomic(&(m)->v); \
14031 + __acc_dec_atomic(&(m)->v); \
14032 + __vx_acc_cres(m->mm_vx_info, r, d, m, __FILE__, __LINE__); \
14036 +#define vx_acc_pages(m, p, v, r) do { \
14037 + unsigned long __p = (p); \
14038 + __acc_add_long(__p, &(m)->v); \
14039 + __vx_add_cres(m->mm_vx_info, r, __p, m, __FILE__, __LINE__); \
14042 +#define vx_acc_pages_atomic(m, p, v, r) do { \
14043 + unsigned long __p = (p); \
14044 + __acc_add_atomic(__p, &(m)->v); \
14045 + __vx_add_cres(m->mm_vx_info, r, __p, m, __FILE__, __LINE__); \
14050 +#define vx_acc_vmpage(m, d) \
14051 + vx_acc_page(m, d, total_vm, RLIMIT_AS)
14052 +#define vx_acc_vmlpage(m, d) \
14053 + vx_acc_page(m, d, locked_vm, RLIMIT_MEMLOCK)
14054 +#define vx_acc_file_rsspage(m, d) \
14055 + vx_acc_page_atomic(m, d, _file_rss, VLIMIT_MAPPED)
14056 +#define vx_acc_anon_rsspage(m, d) \
14057 + vx_acc_page_atomic(m, d, _anon_rss, VLIMIT_ANON)
14059 +#define vx_acc_vmpages(m, p) \
14060 + vx_acc_pages(m, p, total_vm, RLIMIT_AS)
14061 +#define vx_acc_vmlpages(m, p) \
14062 + vx_acc_pages(m, p, locked_vm, RLIMIT_MEMLOCK)
14063 +#define vx_acc_file_rsspages(m, p) \
14064 + vx_acc_pages_atomic(m, p, _file_rss, VLIMIT_MAPPED)
14065 +#define vx_acc_anon_rsspages(m, p) \
14066 + vx_acc_pages_atomic(m, p, _anon_rss, VLIMIT_ANON)
14068 +#define vx_pages_add(s, r, p) __vx_add_cres(s, r, p, 0, __FILE__, __LINE__)
14069 +#define vx_pages_sub(s, r, p) vx_pages_add(s, r, -(p))
14071 +#define vx_vmpages_inc(m) vx_acc_vmpage(m, 1)
14072 +#define vx_vmpages_dec(m) vx_acc_vmpage(m, -1)
14073 +#define vx_vmpages_add(m, p) vx_acc_vmpages(m, p)
14074 +#define vx_vmpages_sub(m, p) vx_acc_vmpages(m, -(p))
14076 +#define vx_vmlocked_inc(m) vx_acc_vmlpage(m, 1)
14077 +#define vx_vmlocked_dec(m) vx_acc_vmlpage(m, -1)
14078 +#define vx_vmlocked_add(m, p) vx_acc_vmlpages(m, p)
14079 +#define vx_vmlocked_sub(m, p) vx_acc_vmlpages(m, -(p))
14081 +#define vx_file_rsspages_inc(m) vx_acc_file_rsspage(m, 1)
14082 +#define vx_file_rsspages_dec(m) vx_acc_file_rsspage(m, -1)
14083 +#define vx_file_rsspages_add(m, p) vx_acc_file_rsspages(m, p)
14084 +#define vx_file_rsspages_sub(m, p) vx_acc_file_rsspages(m, -(p))
14086 +#define vx_anon_rsspages_inc(m) vx_acc_anon_rsspage(m, 1)
14087 +#define vx_anon_rsspages_dec(m) vx_acc_anon_rsspage(m, -1)
14088 +#define vx_anon_rsspages_add(m, p) vx_acc_anon_rsspages(m, p)
14089 +#define vx_anon_rsspages_sub(m, p) vx_acc_anon_rsspages(m, -(p))
14092 +#define vx_pages_avail(m, p, r) \
14093 + __vx_cres_avail((m)->mm_vx_info, r, p, __FILE__, __LINE__)
14095 +#define vx_vmpages_avail(m, p) vx_pages_avail(m, p, RLIMIT_AS)
14096 +#define vx_vmlocked_avail(m, p) vx_pages_avail(m, p, RLIMIT_MEMLOCK)
14097 +#define vx_anon_avail(m, p) vx_pages_avail(m, p, VLIMIT_ANON)
14098 +#define vx_mapped_avail(m, p) vx_pages_avail(m, p, VLIMIT_MAPPED)
14100 +#define vx_rss_avail(m, p) \
14101 + __vx_cres_array_avail((m)->mm_vx_info, VLA_RSS, p, __FILE__, __LINE__)
14105 + VXPT_UNKNOWN = 0,
14114 +#define vx_page_fault(mm, vma, type, ret)
14118 +void __vx_page_fault(struct mm_struct *mm,
14119 + struct vm_area_struct *vma, int type, int ret)
14121 + struct vx_info *vxi = mm->mm_vx_info;
14124 + static char *page_type[6] =
14125 + { "UNKNOWN", "ANON", "NONE", "FILE", "SWAP", "WRITE" };
14126 + static char *page_what[4] =
14127 + { "FAULT_OOM", "FAULT_SIGBUS", "FAULT_MINOR", "FAULT_MAJOR" };
14133 + what = (ret & 0x3);
14135 +/* printk("[%d] page[%d][%d] %2x %s %s\n", vxi->vx_id,
14136 + type, what, ret, page_type[type], page_what[what]);
14138 + if (ret & VM_FAULT_WRITE)
14140 + atomic_inc(&vxi->cacct.page[type][what]);
14143 +#define vx_page_fault(mm, vma, type, ret) __vx_page_fault(mm, vma, type, ret)
14147 +extern unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm);
14150 +#warning duplicate inclusion
14152 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_network.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_network.h
14153 --- linux-2.6.29.1/include/linux/vs_network.h 1970-01-01 01:00:00.000000000 +0100
14154 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_network.h 2009-02-22 22:54:26.000000000 +0100
14156 +#ifndef _NX_VS_NETWORK_H
14157 +#define _NX_VS_NETWORK_H
14159 +#include "vserver/context.h"
14160 +#include "vserver/network.h"
14161 +#include "vserver/base.h"
14162 +#include "vserver/check.h"
14163 +#include "vserver/debug.h"
14165 +#include <linux/sched.h>
14168 +#define get_nx_info(i) __get_nx_info(i, __FILE__, __LINE__)
14170 +static inline struct nx_info *__get_nx_info(struct nx_info *nxi,
14171 + const char *_file, int _line)
14176 + vxlprintk(VXD_CBIT(nid, 2), "get_nx_info(%p[#%d.%d])",
14177 + nxi, nxi ? nxi->nx_id : 0,
14178 + nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14181 + atomic_inc(&nxi->nx_usecnt);
14186 +extern void free_nx_info(struct nx_info *);
14188 +#define put_nx_info(i) __put_nx_info(i, __FILE__, __LINE__)
14190 +static inline void __put_nx_info(struct nx_info *nxi, const char *_file, int _line)
14195 + vxlprintk(VXD_CBIT(nid, 2), "put_nx_info(%p[#%d.%d])",
14196 + nxi, nxi ? nxi->nx_id : 0,
14197 + nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14200 + if (atomic_dec_and_test(&nxi->nx_usecnt))
14201 + free_nx_info(nxi);
14205 +#define init_nx_info(p, i) __init_nx_info(p, i, __FILE__, __LINE__)
14207 +static inline void __init_nx_info(struct nx_info **nxp, struct nx_info *nxi,
14208 + const char *_file, int _line)
14211 + vxlprintk(VXD_CBIT(nid, 3),
14212 + "init_nx_info(%p[#%d.%d])",
14213 + nxi, nxi ? nxi->nx_id : 0,
14214 + nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14217 + atomic_inc(&nxi->nx_usecnt);
14223 +#define set_nx_info(p, i) __set_nx_info(p, i, __FILE__, __LINE__)
14225 +static inline void __set_nx_info(struct nx_info **nxp, struct nx_info *nxi,
14226 + const char *_file, int _line)
14228 + struct nx_info *nxo;
14233 + vxlprintk(VXD_CBIT(nid, 3), "set_nx_info(%p[#%d.%d])",
14234 + nxi, nxi ? nxi->nx_id : 0,
14235 + nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14238 + atomic_inc(&nxi->nx_usecnt);
14239 + nxo = xchg(nxp, nxi);
14243 +#define clr_nx_info(p) __clr_nx_info(p, __FILE__, __LINE__)
14245 +static inline void __clr_nx_info(struct nx_info **nxp,
14246 + const char *_file, int _line)
14248 + struct nx_info *nxo;
14250 + nxo = xchg(nxp, NULL);
14254 + vxlprintk(VXD_CBIT(nid, 3), "clr_nx_info(%p[#%d.%d])",
14255 + nxo, nxo ? nxo->nx_id : 0,
14256 + nxo ? atomic_read(&nxo->nx_usecnt) : 0,
14259 + if (atomic_dec_and_test(&nxo->nx_usecnt))
14260 + free_nx_info(nxo);
14264 +#define claim_nx_info(v, p) __claim_nx_info(v, p, __FILE__, __LINE__)
14266 +static inline void __claim_nx_info(struct nx_info *nxi,
14267 + struct task_struct *task, const char *_file, int _line)
14269 + vxlprintk(VXD_CBIT(nid, 3), "claim_nx_info(%p[#%d.%d.%d]) %p",
14270 + nxi, nxi ? nxi->nx_id : 0,
14271 + nxi?atomic_read(&nxi->nx_usecnt):0,
14272 + nxi?atomic_read(&nxi->nx_tasks):0,
14273 + task, _file, _line);
14275 + atomic_inc(&nxi->nx_tasks);
14279 +extern void unhash_nx_info(struct nx_info *);
14281 +#define release_nx_info(v, p) __release_nx_info(v, p, __FILE__, __LINE__)
14283 +static inline void __release_nx_info(struct nx_info *nxi,
14284 + struct task_struct *task, const char *_file, int _line)
14286 + vxlprintk(VXD_CBIT(nid, 3), "release_nx_info(%p[#%d.%d.%d]) %p",
14287 + nxi, nxi ? nxi->nx_id : 0,
14288 + nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14289 + nxi ? atomic_read(&nxi->nx_tasks) : 0,
14290 + task, _file, _line);
14294 + if (atomic_dec_and_test(&nxi->nx_tasks))
14295 + unhash_nx_info(nxi);
14299 +#define task_get_nx_info(i) __task_get_nx_info(i, __FILE__, __LINE__)
14301 +static __inline__ struct nx_info *__task_get_nx_info(struct task_struct *p,
14302 + const char *_file, int _line)
14304 + struct nx_info *nxi;
14307 + vxlprintk(VXD_CBIT(nid, 5), "task_get_nx_info(%p)",
14308 + p, _file, _line);
14309 + nxi = __get_nx_info(p->nx_info, _file, _line);
14315 +static inline void exit_nx_info(struct task_struct *p)
14318 + release_nx_info(p->nx_info, p);
14323 +#warning duplicate inclusion
14325 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_pid.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_pid.h
14326 --- linux-2.6.29.1/include/linux/vs_pid.h 1970-01-01 01:00:00.000000000 +0100
14327 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_pid.h 2009-02-22 22:54:26.000000000 +0100
14332 +#include "vserver/base.h"
14333 +#include "vserver/check.h"
14334 +#include "vserver/context.h"
14335 +#include "vserver/debug.h"
14336 +#include "vserver/pid.h"
14337 +#include <linux/pid_namespace.h>
14340 +#define VXF_FAKE_INIT (VXF_INFO_INIT | VXF_STATE_INIT)
14343 +int vx_proc_task_visible(struct task_struct *task)
14345 + if ((task->pid == 1) &&
14346 + !vx_flags(VXF_FAKE_INIT, VXF_FAKE_INIT))
14347 + /* show a blend through init */
14349 + if (vx_check(vx_task_xid(task), VS_WATCH | VS_IDENT))
14356 +#define find_task_by_real_pid(pid) find_task_by_pid_ns(pid, &init_pid_ns)
14361 +struct task_struct *vx_find_proc_task_by_pid(int pid)
14363 + struct task_struct *task = find_task_by_real_pid(pid);
14365 + if (task && !vx_proc_task_visible(task)) {
14366 + vxdprintk(VXD_CBIT(misc, 6),
14367 + "dropping task (find) %p[#%u,%u] for %p[#%u,%u]",
14368 + task, task->xid, task->pid,
14369 + current, current->xid, current->pid);
14378 +struct task_struct *vx_get_proc_task(struct inode *inode, struct pid *pid)
14380 + struct task_struct *task = get_pid_task(pid, PIDTYPE_PID);
14382 + if (task && !vx_proc_task_visible(task)) {
14383 + vxdprintk(VXD_CBIT(misc, 6),
14384 + "dropping task (get) %p[#%u,%u] for %p[#%u,%u]",
14385 + task, task->xid, task->pid,
14386 + current, current->xid, current->pid);
14387 + put_task_struct(task);
14396 +struct task_struct *vx_child_reaper(struct task_struct *p)
14398 + struct vx_info *vxi = p->vx_info;
14399 + struct task_struct *reaper = child_reaper(p);
14404 + BUG_ON(!p->vx_info->vx_reaper);
14406 + /* child reaper for the guest reaper */
14407 + if (vxi->vx_reaper == p)
14410 + reaper = vxi->vx_reaper;
14412 + vxdprintk(VXD_CBIT(xid, 7),
14413 + "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]",
14414 + p, p->xid, p->pid, reaper, reaper->xid, reaper->pid);
14422 +#warning duplicate inclusion
14424 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_sched.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_sched.h
14425 --- linux-2.6.29.1/include/linux/vs_sched.h 1970-01-01 01:00:00.000000000 +0100
14426 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_sched.h 2009-02-22 22:54:26.000000000 +0100
14428 +#ifndef _VS_SCHED_H
14429 +#define _VS_SCHED_H
14431 +#include "vserver/base.h"
14432 +#include "vserver/context.h"
14433 +#include "vserver/sched.h"
14436 +#define VAVAVOOM_RATIO 50
14438 +#define MAX_PRIO_BIAS 20
14439 +#define MIN_PRIO_BIAS -20
14442 +#ifdef CONFIG_VSERVER_HARDCPU
14445 + * effective_prio - return the priority that is based on the static
14446 + * priority but is modified by bonuses/penalties.
14448 + * We scale the actual sleep average [0 .... MAX_SLEEP_AVG]
14449 + * into a -4 ... 0 ... +4 bonus/penalty range.
14451 + * Additionally, we scale another amount based on the number of
14452 + * CPU tokens currently held by the context, if the process is
14453 + * part of a context (and the appropriate SCHED flag is set).
14454 + * This ranges from -5 ... 0 ... +15, quadratically.
14456 + * So, the total bonus is -9 .. 0 .. +19
14457 + * We use ~50% of the full 0...39 priority range so that:
14459 + * 1) nice +19 interactive tasks do not preempt nice 0 CPU hogs.
14460 + * 2) nice -20 CPU hogs do not get preempted by nice 0 tasks.
14461 + * unless that context is far exceeding its CPU allocation.
14463 + * Both properties are important to certain workloads.
14466 +int vx_effective_vavavoom(struct _vx_sched_pc *sched_pc, int max_prio)
14468 + int vavavoom, max;
14470 + /* lots of tokens = lots of vavavoom
14471 + * no tokens = no vavavoom */
14472 + if ((vavavoom = sched_pc->tokens) >= 0) {
14473 + max = sched_pc->tokens_max;
14474 + vavavoom = max - vavavoom;
14476 + vavavoom = max_prio * VAVAVOOM_RATIO / 100
14477 + * (vavavoom*vavavoom - (max >> 2)) / max;
14485 +int vx_adjust_prio(struct task_struct *p, int prio, int max_user)
14487 + struct vx_info *vxi = p->vx_info;
14488 + struct _vx_sched_pc *sched_pc;
14493 + sched_pc = &vx_cpu(vxi, sched_pc);
14494 + if (vx_info_flags(vxi, VXF_SCHED_PRIO, 0)) {
14495 + int vavavoom = vx_effective_vavavoom(sched_pc, max_user);
14497 + sched_pc->vavavoom = vavavoom;
14498 + prio += vavavoom;
14500 + prio += sched_pc->prio_bias;
14504 +#else /* !CONFIG_VSERVER_HARDCPU */
14507 +int vx_adjust_prio(struct task_struct *p, int prio, int max_user)
14509 + struct vx_info *vxi = p->vx_info;
14512 + prio += vx_cpu(vxi, sched_pc).prio_bias;
14516 +#endif /* CONFIG_VSERVER_HARDCPU */
14519 +static inline void vx_account_user(struct vx_info *vxi,
14520 + cputime_t cputime, int nice)
14524 + vx_cpu(vxi, sched_pc).user_ticks += cputime;
14527 +static inline void vx_account_system(struct vx_info *vxi,
14528 + cputime_t cputime, int idle)
14532 + vx_cpu(vxi, sched_pc).sys_ticks += cputime;
14536 +#warning duplicate inclusion
14538 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_socket.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_socket.h
14539 --- linux-2.6.29.1/include/linux/vs_socket.h 1970-01-01 01:00:00.000000000 +0100
14540 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_socket.h 2009-02-22 22:54:26.000000000 +0100
14542 +#ifndef _VS_SOCKET_H
14543 +#define _VS_SOCKET_H
14545 +#include "vserver/debug.h"
14546 +#include "vserver/base.h"
14547 +#include "vserver/cacct.h"
14548 +#include "vserver/context.h"
14549 +#include "vserver/tag.h"
14552 +/* socket accounting */
14554 +#include <linux/socket.h>
14556 +static inline int vx_sock_type(int family)
14558 + switch (family) {
14560 + return VXA_SOCK_UNSPEC;
14562 + return VXA_SOCK_UNIX;
14564 + return VXA_SOCK_INET;
14566 + return VXA_SOCK_INET6;
14568 + return VXA_SOCK_PACKET;
14570 + return VXA_SOCK_OTHER;
14574 +#define vx_acc_sock(v, f, p, s) \
14575 + __vx_acc_sock(v, f, p, s, __FILE__, __LINE__)
14577 +static inline void __vx_acc_sock(struct vx_info *vxi,
14578 + int family, int pos, int size, char *file, int line)
14581 + int type = vx_sock_type(family);
14583 + atomic_long_inc(&vxi->cacct.sock[type][pos].count);
14584 + atomic_long_add(size, &vxi->cacct.sock[type][pos].total);
14588 +#define vx_sock_recv(sk, s) \
14589 + vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 0, s)
14590 +#define vx_sock_send(sk, s) \
14591 + vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 1, s)
14592 +#define vx_sock_fail(sk, s) \
14593 + vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 2, s)
14596 +#define sock_vx_init(s) do { \
14597 + (s)->sk_xid = 0; \
14598 + (s)->sk_vx_info = NULL; \
14601 +#define sock_nx_init(s) do { \
14602 + (s)->sk_nid = 0; \
14603 + (s)->sk_nx_info = NULL; \
14607 +#warning duplicate inclusion
14609 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_tag.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_tag.h
14610 --- linux-2.6.29.1/include/linux/vs_tag.h 1970-01-01 01:00:00.000000000 +0100
14611 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_tag.h 2009-02-22 22:54:26.000000000 +0100
14616 +#include <linux/vserver/tag.h>
14618 +/* check conditions */
14620 +#define DX_ADMIN 0x0001
14621 +#define DX_WATCH 0x0002
14622 +#define DX_HOSTID 0x0008
14624 +#define DX_IDENT 0x0010
14626 +#define DX_ARG_MASK 0x0010
14629 +#define dx_task_tag(t) ((t)->tag)
14631 +#define dx_current_tag() dx_task_tag(current)
14633 +#define dx_check(c, m) __dx_check(dx_current_tag(), c, m)
14635 +#define dx_weak_check(c, m) ((m) ? dx_check(c, m) : 1)
14639 + * check current context for ADMIN/WATCH and
14640 + * optionally against supplied argument
14642 +static inline int __dx_check(tag_t cid, tag_t id, unsigned int mode)
14644 + if (mode & DX_ARG_MASK) {
14645 + if ((mode & DX_IDENT) && (id == cid))
14648 + return (((mode & DX_ADMIN) && (cid == 0)) ||
14649 + ((mode & DX_WATCH) && (cid == 1)) ||
14650 + ((mode & DX_HOSTID) && (id == 0)));
14654 +int dx_permission(struct inode *inode, int mask);
14658 +#warning duplicate inclusion
14660 diff -NurpP --minimal linux-2.6.29.1/include/linux/vs_time.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_time.h
14661 --- linux-2.6.29.1/include/linux/vs_time.h 1970-01-01 01:00:00.000000000 +0100
14662 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/linux/vs_time.h 2009-02-22 22:54:26.000000000 +0100
14664 +#ifndef _VS_TIME_H
14665 +#define _VS_TIME_H
14668 +/* time faking stuff */
14670 +#ifdef CONFIG_VSERVER_VTIME
14672 +extern void vx_gettimeofday(struct timeval *tv);
14673 +extern int vx_settimeofday(struct timespec *ts);
14676 +#define vx_gettimeofday(t) do_gettimeofday(t)
14677 +#define vx_settimeofday(t) do_settimeofday(t)
14681 +#warning duplicate inclusion
14683 diff -NurpP --minimal linux-2.6.29.1/include/net/addrconf.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/addrconf.h
14684 --- linux-2.6.29.1/include/net/addrconf.h 2008-12-25 00:26:37.000000000 +0100
14685 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/addrconf.h 2009-02-22 22:54:26.000000000 +0100
14686 @@ -84,7 +84,8 @@ extern int ipv6_dev_get_saddr(struct n
14687 struct net_device *dev,
14688 const struct in6_addr *daddr,
14689 unsigned int srcprefs,
14690 - struct in6_addr *saddr);
14691 + struct in6_addr *saddr,
14692 + struct nx_info *nxi);
14693 extern int ipv6_get_lladdr(struct net_device *dev,
14694 struct in6_addr *addr,
14695 unsigned char banned_flags);
14696 diff -NurpP --minimal linux-2.6.29.1/include/net/af_unix.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/af_unix.h
14697 --- linux-2.6.29.1/include/net/af_unix.h 2008-12-25 00:26:37.000000000 +0100
14698 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/af_unix.h 2009-02-22 22:54:26.000000000 +0100
14700 #include <linux/socket.h>
14701 #include <linux/un.h>
14702 #include <linux/mutex.h>
14703 +#include <linux/vs_base.h>
14704 #include <net/sock.h>
14706 extern void unix_inflight(struct file *fp);
14707 diff -NurpP --minimal linux-2.6.29.1/include/net/inet_timewait_sock.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/inet_timewait_sock.h
14708 --- linux-2.6.29.1/include/net/inet_timewait_sock.h 2009-03-24 14:22:44.000000000 +0100
14709 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/inet_timewait_sock.h 2009-03-24 14:48:36.000000000 +0100
14710 @@ -15,15 +15,14 @@
14711 #ifndef _INET_TIMEWAIT_SOCK_
14712 #define _INET_TIMEWAIT_SOCK_
14714 +// #include <net/inet_sock.h>
14715 +#include <net/sock.h>
14717 #include <linux/list.h>
14718 #include <linux/module.h>
14719 #include <linux/timer.h>
14720 #include <linux/types.h>
14721 #include <linux/workqueue.h>
14723 -#include <net/inet_sock.h>
14724 -#include <net/sock.h>
14725 #include <net/tcp_states.h>
14726 #include <net/timewait_sock.h>
14728 @@ -116,6 +115,10 @@ struct inet_timewait_sock {
14729 #define tw_hash __tw_common.skc_hash
14730 #define tw_prot __tw_common.skc_prot
14731 #define tw_net __tw_common.skc_net
14732 +#define tw_xid __tw_common.skc_xid
14733 +#define tw_vx_info __tw_common.skc_vx_info
14734 +#define tw_nid __tw_common.skc_nid
14735 +#define tw_nx_info __tw_common.skc_nx_info
14737 volatile unsigned char tw_substate;
14738 /* 3 bits hole, try to pack */
14739 diff -NurpP --minimal linux-2.6.29.1/include/net/route.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/route.h
14740 --- linux-2.6.29.1/include/net/route.h 2008-12-25 00:26:37.000000000 +0100
14741 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/route.h 2009-02-22 22:54:26.000000000 +0100
14742 @@ -135,6 +135,9 @@ static inline void ip_rt_put(struct rtab
14743 dst_release(&rt->u.dst);
14746 +#include <linux/vs_base.h>
14747 +#include <linux/vs_inet.h>
14749 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
14751 extern const __u8 ip_tos2prio[16];
14752 @@ -144,6 +147,9 @@ static inline char rt_tos2priority(u8 to
14753 return ip_tos2prio[IPTOS_TOS(tos)>>1];
14756 +extern int ip_v4_find_src(struct net *net, struct nx_info *,
14757 + struct rtable **, struct flowi *);
14759 static inline int ip_route_connect(struct rtable **rp, __be32 dst,
14760 __be32 src, u32 tos, int oif, u8 protocol,
14761 __be16 sport, __be16 dport, struct sock *sk,
14762 @@ -161,11 +167,24 @@ static inline int ip_route_connect(struc
14765 struct net *net = sock_net(sk);
14766 + struct nx_info *nx_info = current->nx_info;
14768 if (inet_sk(sk)->transparent)
14769 fl.flags |= FLOWI_FLAG_ANYSRC;
14771 - if (!dst || !src) {
14773 + nx_info = sk->sk_nx_info;
14775 + vxdprintk(VXD_CBIT(net, 4),
14776 + "ip_route_connect(%p) %p,%p;%lx",
14777 + sk, nx_info, sk->sk_socket,
14778 + (sk->sk_socket?sk->sk_socket->flags:0));
14780 + err = ip_v4_find_src(net, nx_info, rp, &fl);
14784 + if (!fl.fl4_dst || !fl.fl4_src) {
14785 err = __ip_route_output_key(net, rp, &fl);
14788 diff -NurpP --minimal linux-2.6.29.1/include/net/sock.h linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/sock.h
14789 --- linux-2.6.29.1/include/net/sock.h 2009-03-24 14:22:44.000000000 +0100
14790 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/include/net/sock.h 2009-03-24 14:48:36.000000000 +0100
14791 @@ -134,6 +134,10 @@ struct sock_common {
14792 #ifdef CONFIG_NET_NS
14793 struct net *skc_net;
14796 + struct vx_info *skc_vx_info;
14798 + struct nx_info *skc_nx_info;
14802 @@ -218,6 +222,10 @@ struct sock {
14803 #define sk_hash __sk_common.skc_hash
14804 #define sk_prot __sk_common.skc_prot
14805 #define sk_net __sk_common.skc_net
14806 +#define sk_xid __sk_common.skc_xid
14807 +#define sk_vx_info __sk_common.skc_vx_info
14808 +#define sk_nid __sk_common.skc_nid
14809 +#define sk_nx_info __sk_common.skc_nx_info
14810 unsigned char sk_shutdown : 2,
14813 diff -NurpP --minimal linux-2.6.29.1/init/main.c linux-2.6.29.1-vs2.3.0.36.9-pre5/init/main.c
14814 --- linux-2.6.29.1/init/main.c 2009-03-24 14:22:44.000000000 +0100
14815 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/init/main.c 2009-03-24 17:22:37.000000000 +0100
14817 #include <linux/ftrace.h>
14818 #include <linux/async.h>
14819 #include <trace/boot.h>
14820 +#include <linux/vserver/percpu.h>
14822 #include <asm/io.h>
14823 #include <asm/bugs.h>
14824 @@ -381,12 +382,14 @@ EXPORT_SYMBOL(__per_cpu_offset);
14826 static void __init setup_per_cpu_areas(void)
14828 - unsigned long size, i;
14829 + unsigned long size, vspc, i;
14831 unsigned long nr_possible_cpus = num_possible_cpus();
14833 + vspc = PERCPU_PERCTX * CONFIG_VSERVER_CONTEXTS;
14835 /* Copy section for each CPU (we discard the original) */
14836 - size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
14837 + size = ALIGN(PERCPU_ENOUGH_ROOM + vspc, PAGE_SIZE);
14838 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
14840 for_each_possible_cpu(i) {
14841 diff -NurpP --minimal linux-2.6.29.1/ipc/mqueue.c linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/mqueue.c
14842 --- linux-2.6.29.1/ipc/mqueue.c 2009-03-24 14:22:44.000000000 +0100
14843 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/mqueue.c 2009-03-24 17:24:19.000000000 +0100
14845 #include <linux/mutex.h>
14846 #include <linux/nsproxy.h>
14847 #include <linux/pid.h>
14848 +#include <linux/vs_context.h>
14849 +#include <linux/vs_limit.h>
14851 #include <net/sock.h>
14853 @@ -79,6 +81,7 @@ struct mqueue_inode_info {
14854 struct sigevent notify;
14855 struct pid* notify_owner;
14856 struct user_struct *user; /* user who created, for accounting */
14857 + struct vx_info *vxi;
14858 struct sock *notify_sock;
14859 struct sk_buff *notify_cookie;
14861 @@ -126,6 +129,7 @@ static struct inode *mqueue_get_inode(st
14862 if (S_ISREG(mode)) {
14863 struct mqueue_inode_info *info;
14864 struct task_struct *p = current;
14865 + struct vx_info *vxi = p->vx_info;
14866 unsigned long mq_bytes, mq_msg_tblsz;
14868 inode->i_fop = &mqueue_file_operations;
14869 @@ -140,6 +144,7 @@ static struct inode *mqueue_get_inode(st
14870 info->notify_owner = NULL;
14872 info->user = NULL; /* set when all is ok */
14873 + info->vxi = NULL;
14874 memset(&info->attr, 0, sizeof(info->attr));
14875 info->attr.mq_maxmsg = msg_max;
14876 info->attr.mq_msgsize = msgsize_max;
14877 @@ -154,22 +159,26 @@ static struct inode *mqueue_get_inode(st
14878 spin_lock(&mq_lock);
14879 if (u->mq_bytes + mq_bytes < u->mq_bytes ||
14880 u->mq_bytes + mq_bytes >
14881 - p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur) {
14882 + p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur ||
14883 + !vx_ipcmsg_avail(vxi, mq_bytes)) {
14884 spin_unlock(&mq_lock);
14887 u->mq_bytes += mq_bytes;
14888 + vx_ipcmsg_add(vxi, u, mq_bytes);
14889 spin_unlock(&mq_lock);
14891 info->messages = kmalloc(mq_msg_tblsz, GFP_KERNEL);
14892 if (!info->messages) {
14893 spin_lock(&mq_lock);
14894 u->mq_bytes -= mq_bytes;
14895 + vx_ipcmsg_sub(vxi, u, mq_bytes);
14896 spin_unlock(&mq_lock);
14900 info->user = get_uid(u);
14901 + info->vxi = get_vx_info(vxi);
14902 } else if (S_ISDIR(mode)) {
14904 /* Some things misbehave if size == 0 on a directory */
14905 @@ -260,10 +269,14 @@ static void mqueue_delete_inode(struct i
14906 (info->attr.mq_maxmsg * info->attr.mq_msgsize));
14909 + struct vx_info *vxi = info->vxi;
14911 spin_lock(&mq_lock);
14912 user->mq_bytes -= mq_bytes;
14913 + vx_ipcmsg_sub(vxi, user, mq_bytes);
14915 spin_unlock(&mq_lock);
14916 + put_vx_info(vxi);
14920 diff -NurpP --minimal linux-2.6.29.1/ipc/msg.c linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/msg.c
14921 --- linux-2.6.29.1/ipc/msg.c 2009-03-24 14:22:44.000000000 +0100
14922 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/msg.c 2009-02-22 22:54:26.000000000 +0100
14924 #include <linux/rwsem.h>
14925 #include <linux/nsproxy.h>
14926 #include <linux/ipc_namespace.h>
14927 +#include <linux/vs_base.h>
14929 #include <asm/current.h>
14930 #include <asm/uaccess.h>
14931 @@ -190,6 +191,7 @@ static int newque(struct ipc_namespace *
14933 msq->q_perm.mode = msgflg & S_IRWXUGO;
14934 msq->q_perm.key = key;
14935 + msq->q_perm.xid = vx_current_xid();
14937 msq->q_perm.security = NULL;
14938 retval = security_msg_queue_alloc(msq);
14939 diff -NurpP --minimal linux-2.6.29.1/ipc/namespace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/namespace.c
14940 --- linux-2.6.29.1/ipc/namespace.c 2008-12-25 00:26:37.000000000 +0100
14941 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/namespace.c 2009-02-22 22:54:26.000000000 +0100
14943 #include <linux/rcupdate.h>
14944 #include <linux/nsproxy.h>
14945 #include <linux/slab.h>
14946 +#include <linux/vs_base.h>
14947 +#include <linux/vserver/global.h>
14951 @@ -35,6 +37,7 @@ static struct ipc_namespace *clone_ipc_n
14952 register_ipcns_notifier(ns);
14954 kref_init(&ns->kref);
14955 + atomic_inc(&vs_global_ipc_ns);
14959 @@ -101,6 +104,7 @@ void free_ipc_ns(struct kref *kref)
14963 + atomic_dec(&vs_global_ipc_ns);
14965 atomic_dec(&nr_ipc_ns);
14967 diff -NurpP --minimal linux-2.6.29.1/ipc/sem.c linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/sem.c
14968 --- linux-2.6.29.1/ipc/sem.c 2009-03-24 14:22:44.000000000 +0100
14969 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/sem.c 2009-03-31 23:31:33.000000000 +0200
14971 #include <linux/rwsem.h>
14972 #include <linux/nsproxy.h>
14973 #include <linux/ipc_namespace.h>
14974 +#include <linux/vs_base.h>
14975 +#include <linux/vs_limit.h>
14977 #include <asm/uaccess.h>
14979 @@ -255,6 +257,7 @@ static int newary(struct ipc_namespace *
14981 sma->sem_perm.mode = (semflg & S_IRWXUGO);
14982 sma->sem_perm.key = key;
14983 + sma->sem_perm.xid = vx_current_xid();
14985 sma->sem_perm.security = NULL;
14986 retval = security_sem_alloc(sma);
14987 @@ -270,6 +273,9 @@ static int newary(struct ipc_namespace *
14990 ns->used_sems += nsems;
14991 + /* FIXME: obsoleted? */
14992 + vx_semary_inc(sma);
14993 + vx_nsems_add(sma, nsems);
14995 sma->sem_base = (struct sem *) &sma[1];
14996 INIT_LIST_HEAD(&sma->sem_pending);
14997 @@ -546,6 +552,9 @@ static void freeary(struct ipc_namespace
15000 ns->used_sems -= sma->sem_nsems;
15001 + /* FIXME: obsoleted? */
15002 + vx_nsems_sub(sma, sma->sem_nsems);
15003 + vx_semary_dec(sma);
15004 security_sem_free(sma);
15005 ipc_rcu_putref(sma);
15007 diff -NurpP --minimal linux-2.6.29.1/ipc/shm.c linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/shm.c
15008 --- linux-2.6.29.1/ipc/shm.c 2009-03-24 14:22:44.000000000 +0100
15009 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/ipc/shm.c 2009-03-24 14:48:36.000000000 +0100
15011 #include <linux/nsproxy.h>
15012 #include <linux/mount.h>
15013 #include <linux/ipc_namespace.h>
15014 +#include <linux/vs_context.h>
15015 +#include <linux/vs_limit.h>
15017 #include <asm/uaccess.h>
15019 @@ -168,7 +170,12 @@ static void shm_open(struct vm_area_stru
15021 static void shm_destroy(struct ipc_namespace *ns, struct shmid_kernel *shp)
15023 - ns->shm_tot -= (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15024 + struct vx_info *vxi = lookup_vx_info(shp->shm_perm.xid);
15025 + int numpages = (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15027 + vx_ipcshm_sub(vxi, shp, numpages);
15028 + ns->shm_tot -= numpages;
15032 if (!is_file_hugepages(shp->shm_file))
15033 @@ -178,6 +185,7 @@ static void shm_destroy(struct ipc_names
15035 fput (shp->shm_file);
15036 security_shm_free(shp);
15037 + put_vx_info(vxi);
15038 ipc_rcu_putref(shp);
15041 @@ -348,11 +356,15 @@ static int newseg(struct ipc_namespace *
15042 if (ns->shm_tot + numpages > ns->shm_ctlall)
15045 + if (!vx_ipcshm_avail(current->vx_info, numpages))
15048 shp = ipc_rcu_alloc(sizeof(*shp));
15052 shp->shm_perm.key = key;
15053 + shp->shm_perm.xid = vx_current_xid();
15054 shp->shm_perm.mode = (shmflg & S_IRWXUGO);
15055 shp->mlock_user = NULL;
15057 @@ -406,6 +418,7 @@ static int newseg(struct ipc_namespace *
15058 ns->shm_tot += numpages;
15059 error = shp->shm_perm.id;
15061 + vx_ipcshm_add(current->vx_info, key, numpages);
15065 diff -NurpP --minimal linux-2.6.29.1/kernel/capability.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/capability.c
15066 --- linux-2.6.29.1/kernel/capability.c 2009-03-24 14:22:44.000000000 +0100
15067 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/capability.c 2009-03-24 17:27:28.000000000 +0100
15069 #include <linux/security.h>
15070 #include <linux/syscalls.h>
15071 #include <linux/pid_namespace.h>
15072 +#include <linux/vs_context.h>
15073 #include <asm/uaccess.h>
15074 #include "cred-internals.h"
15076 @@ -122,6 +123,7 @@ static int cap_validate_magic(cap_user_h
15082 * The only thing that can change the capabilities of the current
15083 * process is the current process. As such, we can't be in this code
15084 @@ -289,6 +291,8 @@ error:
15088 +#include <linux/vserver/base.h>
15091 * capable - Determine if the current task has a superior capability in effect
15092 * @cap: The capability to be tested for
15093 @@ -301,6 +305,9 @@ error:
15095 int capable(int cap)
15097 + /* here for now so we don't require task locking */
15098 + if (vs_check_bit(VXC_CAP_MASK, cap) && !vx_mcaps(1L << cap))
15100 if (unlikely(!cap_valid(cap))) {
15101 printk(KERN_CRIT "capable() called with invalid cap=%u\n", cap);
15103 diff -NurpP --minimal linux-2.6.29.1/kernel/compat.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/compat.c
15104 --- linux-2.6.29.1/kernel/compat.c 2009-03-24 14:22:44.000000000 +0100
15105 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/compat.c 2009-03-24 14:48:36.000000000 +0100
15106 @@ -891,7 +891,7 @@ asmlinkage long compat_sys_time(compat_t
15110 - do_gettimeofday(&tv);
15111 + vx_gettimeofday(&tv);
15115 @@ -916,7 +916,7 @@ asmlinkage long compat_sys_stime(compat_
15119 - do_settimeofday(&tv);
15120 + vx_settimeofday(&tv);
15124 diff -NurpP --minimal linux-2.6.29.1/kernel/exit.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/exit.c
15125 --- linux-2.6.29.1/kernel/exit.c 2009-03-24 14:22:44.000000000 +0100
15126 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/exit.c 2009-03-24 17:28:40.000000000 +0100
15128 #include <linux/task_io_accounting_ops.h>
15129 #include <linux/tracehook.h>
15130 #include <linux/init_task.h>
15131 +#include <linux/vs_limit.h>
15132 +#include <linux/vs_context.h>
15133 +#include <linux/vs_network.h>
15134 +#include <linux/vs_pid.h>
15135 +#include <linux/vserver/global.h>
15136 #include <trace/sched.h>
15138 #include <asm/uaccess.h>
15139 @@ -502,9 +507,11 @@ static void close_files(struct files_str
15140 filp_close(file, files);
15143 + vx_openfd_dec(i);
15151 @@ -571,6 +578,7 @@ void put_fs_struct(struct fs_struct *fs)
15152 if (atomic_dec_and_test(&fs->count)) {
15153 path_put(&fs->root);
15154 path_put(&fs->pwd);
15155 + atomic_dec(&vs_global_fs);
15156 kmem_cache_free(fs_cachep, fs);
15159 @@ -1122,10 +1130,15 @@ NORET_TYPE void do_exit(long code)
15160 if (tsk->splice_pipe)
15161 __free_pipe_info(tsk->splice_pipe);
15163 + /* needs to stay after exit_notify() */
15164 + exit_vx_info(tsk, code);
15165 + exit_nx_info(tsk);
15168 /* causes final put_task_struct in finish_task_switch(). */
15169 tsk->state = TASK_DEAD;
15171 + printk("bad task: %p [%lx]\n", current, current->state);
15173 /* Avoid "noreturn function does return". */
15175 diff -NurpP --minimal linux-2.6.29.1/kernel/fork.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/fork.c
15176 --- linux-2.6.29.1/kernel/fork.c 2009-03-24 14:22:44.000000000 +0100
15177 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/fork.c 2009-03-24 17:31:42.000000000 +0100
15179 #include <linux/tty.h>
15180 #include <linux/proc_fs.h>
15181 #include <linux/blkdev.h>
15182 +#include <linux/vs_context.h>
15183 +#include <linux/vs_network.h>
15184 +#include <linux/vs_limit.h>
15185 +#include <linux/vs_memory.h>
15186 +#include <linux/vserver/global.h>
15187 #include <trace/sched.h>
15189 #include <asm/pgtable.h>
15190 @@ -140,6 +145,8 @@ void free_task(struct task_struct *tsk)
15191 prop_local_destroy_single(&tsk->dirties);
15192 free_thread_info(tsk->stack);
15193 rt_mutex_debug_task_free(tsk);
15194 + clr_vx_info(&tsk->vx_info);
15195 + clr_nx_info(&tsk->nx_info);
15196 ftrace_graph_exit_task(tsk);
15197 free_task_struct(tsk);
15199 @@ -279,6 +286,8 @@ static int dup_mmap(struct mm_struct *mm
15200 mm->free_area_cache = oldmm->mmap_base;
15201 mm->cached_hole_size = ~0UL;
15203 + __set_mm_counter(mm, file_rss, 0);
15204 + __set_mm_counter(mm, anon_rss, 0);
15205 cpus_clear(mm->cpu_vm_mask);
15206 mm->mm_rb = RB_ROOT;
15207 rb_link = &mm->mm_rb.rb_node;
15208 @@ -290,7 +299,7 @@ static int dup_mmap(struct mm_struct *mm
15210 if (mpnt->vm_flags & VM_DONTCOPY) {
15211 long pages = vma_pages(mpnt);
15212 - mm->total_vm -= pages;
15213 + vx_vmpages_sub(mm, pages);
15214 vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file,
15217 @@ -423,8 +432,8 @@ static struct mm_struct * mm_init(struct
15218 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter;
15219 mm->core_state = NULL;
15221 - set_mm_counter(mm, file_rss, 0);
15222 - set_mm_counter(mm, anon_rss, 0);
15223 + __set_mm_counter(mm, file_rss, 0);
15224 + __set_mm_counter(mm, anon_rss, 0);
15225 spin_lock_init(&mm->page_table_lock);
15226 spin_lock_init(&mm->ioctx_lock);
15227 INIT_HLIST_HEAD(&mm->ioctx_list);
15228 @@ -435,6 +444,7 @@ static struct mm_struct * mm_init(struct
15229 if (likely(!mm_alloc_pgd(mm))) {
15231 mmu_notifier_mm_init(mm);
15232 + set_vx_info(&mm->mm_vx_info, p->vx_info);
15236 @@ -468,6 +478,7 @@ void __mmdrop(struct mm_struct *mm)
15238 destroy_context(mm);
15239 mmu_notifier_mm_destroy(mm);
15240 + clr_vx_info(&mm->mm_vx_info);
15243 EXPORT_SYMBOL_GPL(__mmdrop);
15244 @@ -594,6 +605,7 @@ struct mm_struct *dup_mm(struct task_str
15247 memcpy(mm, oldmm, sizeof(*mm));
15248 + mm->mm_vx_info = NULL;
15250 /* Initializing for Swap token stuff */
15251 mm->token_priority = 0;
15252 @@ -627,6 +639,7 @@ fail_nocontext:
15253 * If init_new_context() failed, we cannot use mmput() to free the mm
15254 * because it calls destroy_context()
15256 + clr_vx_info(&mm->mm_vx_info);
15260 @@ -690,6 +703,7 @@ static struct fs_struct *__copy_fs_struc
15261 fs->pwd = old->pwd;
15262 path_get(&old->pwd);
15263 read_unlock(&old->lock);
15264 + atomic_inc(&vs_global_fs);
15268 @@ -955,6 +969,8 @@ static struct task_struct *copy_process(
15270 struct task_struct *p;
15271 int cgroup_callbacks_done = 0;
15272 + struct vx_info *vxi;
15273 + struct nx_info *nxi;
15275 if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
15276 return ERR_PTR(-EINVAL);
15277 @@ -989,12 +1005,28 @@ static struct task_struct *copy_process(
15278 DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
15279 DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
15281 + init_vx_info(&p->vx_info, current->vx_info);
15282 + init_nx_info(&p->nx_info, current->nx_info);
15284 + /* check vserver memory */
15285 + if (p->mm && !(clone_flags & CLONE_VM)) {
15286 + if (vx_vmpages_avail(p->mm, p->mm->total_vm))
15287 + vx_pages_add(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15289 + goto bad_fork_free;
15291 + if (p->mm && vx_flags(VXF_FORK_RSS, 0)) {
15292 + if (!vx_rss_avail(p->mm, get_mm_counter(p->mm, file_rss)))
15293 + goto bad_fork_cleanup_vm;
15296 + if (!vx_nproc_avail(1))
15297 + goto bad_fork_cleanup_vm;
15298 if (atomic_read(&p->real_cred->user->processes) >=
15299 p->signal->rlim[RLIMIT_NPROC].rlim_cur) {
15300 if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
15301 p->real_cred->user != INIT_USER)
15302 - goto bad_fork_free;
15303 + goto bad_fork_cleanup_vm;
15306 retval = copy_creds(p, clone_flags);
15307 @@ -1271,6 +1303,18 @@ static struct task_struct *copy_process(
15310 spin_unlock(¤t->sighand->siglock);
15312 + /* p is copy of current */
15313 + vxi = p->vx_info;
15315 + claim_vx_info(vxi, p);
15316 + atomic_inc(&vxi->cvirt.nr_threads);
15317 + atomic_inc(&vxi->cvirt.total_forks);
15320 + nxi = p->nx_info;
15322 + claim_nx_info(nxi, p);
15323 write_unlock_irq(&tasklist_lock);
15324 proc_fork_connector(p);
15325 cgroup_post_fork(p);
15326 @@ -1315,6 +1359,9 @@ bad_fork_cleanup_count:
15327 atomic_dec(&p->cred->user->processes);
15328 put_cred(p->real_cred);
15330 +bad_fork_cleanup_vm:
15331 + if (p->mm && !(clone_flags & CLONE_VM))
15332 + vx_pages_sub(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15336 diff -NurpP --minimal linux-2.6.29.1/kernel/kthread.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/kthread.c
15337 --- linux-2.6.29.1/kernel/kthread.c 2009-03-24 14:22:44.000000000 +0100
15338 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/kthread.c 2009-03-24 14:48:36.000000000 +0100
15340 #include <linux/file.h>
15341 #include <linux/module.h>
15342 #include <linux/mutex.h>
15343 +#include <linux/vs_pid.h>
15344 #include <trace/sched.h>
15346 #define KTHREAD_NICE_LEVEL (-5)
15347 @@ -102,7 +103,7 @@ static void create_kthread(struct kthrea
15348 struct sched_param param = { .sched_priority = 0 };
15349 wait_for_completion(&create->started);
15350 read_lock(&tasklist_lock);
15351 - create->result = find_task_by_pid_ns(pid, &init_pid_ns);
15352 + create->result = find_task_by_real_pid(pid);
15353 read_unlock(&tasklist_lock);
15355 * root may have changed our (kthreadd's) priority or CPU mask.
15356 diff -NurpP --minimal linux-2.6.29.1/kernel/Makefile linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/Makefile
15357 --- linux-2.6.29.1/kernel/Makefile 2009-03-24 14:22:44.000000000 +0100
15358 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/Makefile 2009-03-24 14:48:36.000000000 +0100
15359 @@ -22,6 +22,7 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg
15360 CFLAGS_REMOVE_sched_clock.o = -pg
15364 obj-$(CONFIG_FREEZER) += freezer.o
15365 obj-$(CONFIG_PROFILING) += profile.o
15366 obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o
15367 diff -NurpP --minimal linux-2.6.29.1/kernel/nsproxy.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/nsproxy.c
15368 --- linux-2.6.29.1/kernel/nsproxy.c 2009-03-24 14:22:44.000000000 +0100
15369 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/nsproxy.c 2009-03-25 00:39:02.000000000 +0100
15371 #include <linux/mnt_namespace.h>
15372 #include <linux/utsname.h>
15373 #include <linux/pid_namespace.h>
15374 +#include <linux/vserver/global.h>
15375 +#include <linux/vserver/debug.h>
15376 #include <net/net_namespace.h>
15377 #include <linux/ipc_namespace.h>
15379 @@ -37,6 +39,9 @@ static inline struct nsproxy *clone_nspr
15381 memcpy(ns, orig, sizeof(struct nsproxy));
15382 atomic_set(&ns->count, 1);
15383 + vxdprintk(VXD_CBIT(space, 2), "clone_nsproxy(%p[%u] = %p[1]",
15384 + orig, atomic_read(&orig->count), ns);
15385 + atomic_inc(&vs_global_nsproxy);
15389 @@ -46,41 +51,52 @@ static inline struct nsproxy *clone_nspr
15390 * Return the newly created nsproxy. Do not attach this to the task,
15391 * leave it to the caller to do proper locking and attach it to task.
15393 -static struct nsproxy *create_new_namespaces(unsigned long flags,
15394 - struct task_struct *tsk, struct fs_struct *new_fs)
15395 +static struct nsproxy *unshare_namespaces(unsigned long flags,
15396 + struct nsproxy *orig, struct fs_struct *new_fs)
15398 struct nsproxy *new_nsp;
15401 - new_nsp = clone_nsproxy(tsk->nsproxy);
15402 + vxdprintk(VXD_CBIT(space, 4),
15403 + "unshare_namespaces(0x%08lx,%p,%p)",
15404 + flags, orig, new_fs);
15406 + new_nsp = clone_nsproxy(orig);
15408 return ERR_PTR(-ENOMEM);
15410 - new_nsp->mnt_ns = copy_mnt_ns(flags, tsk->nsproxy->mnt_ns, new_fs);
15411 + new_nsp->mnt_ns = copy_mnt_ns(flags, orig->mnt_ns, new_fs);
15412 if (IS_ERR(new_nsp->mnt_ns)) {
15413 err = PTR_ERR(new_nsp->mnt_ns);
15417 - new_nsp->uts_ns = copy_utsname(flags, tsk->nsproxy->uts_ns);
15418 + new_nsp->uts_ns = copy_utsname(flags, orig->uts_ns);
15419 if (IS_ERR(new_nsp->uts_ns)) {
15420 err = PTR_ERR(new_nsp->uts_ns);
15424 - new_nsp->ipc_ns = copy_ipcs(flags, tsk->nsproxy->ipc_ns);
15425 + new_nsp->ipc_ns = copy_ipcs(flags, orig->ipc_ns);
15426 if (IS_ERR(new_nsp->ipc_ns)) {
15427 err = PTR_ERR(new_nsp->ipc_ns);
15431 - new_nsp->pid_ns = copy_pid_ns(flags, task_active_pid_ns(tsk));
15432 + new_nsp->pid_ns = copy_pid_ns(flags, orig->pid_ns);
15433 if (IS_ERR(new_nsp->pid_ns)) {
15434 err = PTR_ERR(new_nsp->pid_ns);
15438 - new_nsp->net_ns = copy_net_ns(flags, tsk->nsproxy->net_ns);
15440 + new_nsp->user_ns = copy_user_ns(flags, orig->user_ns);
15441 + if (IS_ERR(new_nsp->user_ns)) {
15442 + err = PTR_ERR(new_nsp->user_ns);
15446 + new_nsp->net_ns = copy_net_ns(flags, orig->net_ns);
15447 if (IS_ERR(new_nsp->net_ns)) {
15448 err = PTR_ERR(new_nsp->net_ns);
15450 @@ -105,6 +121,35 @@ out_ns:
15451 return ERR_PTR(err);
15454 +static struct nsproxy *create_new_namespaces(int flags, struct task_struct *tsk,
15455 + struct fs_struct *new_fs)
15457 + return unshare_namespaces(flags, tsk->nsproxy, new_fs);
15461 + * copies the nsproxy, setting refcount to 1, and grabbing a
15462 + * reference to all contained namespaces.
15464 +struct nsproxy *copy_nsproxy(struct nsproxy *orig)
15466 + struct nsproxy *ns = clone_nsproxy(orig);
15470 + get_mnt_ns(ns->mnt_ns);
15472 + get_uts_ns(ns->uts_ns);
15474 + get_ipc_ns(ns->ipc_ns);
15476 + get_pid_ns(ns->pid_ns);
15478 + get_net(ns->net_ns);
15484 * called from clone. This now handles copy for nsproxy and all
15485 * namespaces therein.
15486 @@ -112,9 +157,12 @@ out_ns:
15487 int copy_namespaces(unsigned long flags, struct task_struct *tsk)
15489 struct nsproxy *old_ns = tsk->nsproxy;
15490 - struct nsproxy *new_ns;
15491 + struct nsproxy *new_ns = NULL;
15494 + vxdprintk(VXD_CBIT(space, 7), "copy_namespaces(0x%08lx,%p[%p])",
15495 + flags, tsk, old_ns);
15500 @@ -151,6 +199,9 @@ int copy_namespaces(unsigned long flags,
15503 put_nsproxy(old_ns);
15504 + vxdprintk(VXD_CBIT(space, 3),
15505 + "copy_namespaces(0x%08lx,%p[%p]) = %d [%p]",
15506 + flags, tsk, old_ns, err, new_ns);
15510 @@ -164,7 +215,9 @@ void free_nsproxy(struct nsproxy *ns)
15511 put_ipc_ns(ns->ipc_ns);
15513 put_pid_ns(ns->pid_ns);
15514 - put_net(ns->net_ns);
15516 + put_net(ns->net_ns);
15517 + atomic_dec(&vs_global_nsproxy);
15518 kmem_cache_free(nsproxy_cachep, ns);
15521 @@ -177,6 +230,10 @@ int unshare_nsproxy_namespaces(unsigned
15525 + vxdprintk(VXD_CBIT(space, 4),
15526 + "unshare_nsproxy_namespaces(0x%08lx,[%p])",
15527 + unshare_flags, current->nsproxy);
15529 if (!(unshare_flags & (CLONE_NEWNS | CLONE_NEWUTS | CLONE_NEWIPC |
15532 diff -NurpP --minimal linux-2.6.29.1/kernel/pid.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/pid.c
15533 --- linux-2.6.29.1/kernel/pid.c 2009-03-24 14:22:44.000000000 +0100
15534 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/pid.c 2009-03-24 14:48:36.000000000 +0100
15536 #include <linux/pid_namespace.h>
15537 #include <linux/init_task.h>
15538 #include <linux/syscalls.h>
15539 +#include <linux/vs_pid.h>
15540 +#include <linux/vserver/global.h>
15542 #define pid_hashfn(nr, ns) \
15543 hash_long((unsigned long)nr + (unsigned long)ns, pidhash_shift)
15544 @@ -305,7 +307,7 @@ EXPORT_SYMBOL_GPL(find_pid_ns);
15546 struct pid *find_vpid(int nr)
15548 - return find_pid_ns(nr, current->nsproxy->pid_ns);
15549 + return find_pid_ns(vx_rmap_pid(nr), current->nsproxy->pid_ns);
15551 EXPORT_SYMBOL_GPL(find_vpid);
15553 @@ -365,6 +367,9 @@ void transfer_pid(struct task_struct *ol
15554 struct task_struct *pid_task(struct pid *pid, enum pid_type type)
15556 struct task_struct *result = NULL;
15558 + if (type == PIDTYPE_REALPID)
15559 + type = PIDTYPE_PID;
15561 struct hlist_node *first;
15562 first = rcu_dereference(pid->tasks[type].first);
15563 @@ -388,14 +393,14 @@ EXPORT_SYMBOL(find_task_by_pid_type_ns);
15565 struct task_struct *find_task_by_vpid(pid_t vnr)
15567 - return find_task_by_pid_type_ns(PIDTYPE_PID, vnr,
15568 + return find_task_by_pid_type_ns(PIDTYPE_PID, vx_rmap_pid(vnr),
15569 current->nsproxy->pid_ns);
15571 EXPORT_SYMBOL(find_task_by_vpid);
15573 struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns)
15575 - return find_task_by_pid_type_ns(PIDTYPE_PID, nr, ns);
15576 + return find_task_by_pid_type_ns(PIDTYPE_PID, vx_rmap_pid(nr), ns);
15578 EXPORT_SYMBOL(find_task_by_pid_ns);
15580 @@ -431,7 +436,7 @@ struct pid *find_get_pid(pid_t nr)
15582 EXPORT_SYMBOL_GPL(find_get_pid);
15584 -pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15585 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns)
15589 @@ -444,6 +449,11 @@ pid_t pid_nr_ns(struct pid *pid, struct
15593 +pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15595 + return vx_map_pid(pid_unmapped_nr_ns(pid, ns));
15598 pid_t pid_vnr(struct pid *pid)
15600 return pid_nr_ns(pid, current->nsproxy->pid_ns);
15601 diff -NurpP --minimal linux-2.6.29.1/kernel/pid_namespace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/pid_namespace.c
15602 --- linux-2.6.29.1/kernel/pid_namespace.c 2008-12-25 00:26:37.000000000 +0100
15603 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/pid_namespace.c 2009-02-22 22:54:26.000000000 +0100
15605 #include <linux/syscalls.h>
15606 #include <linux/err.h>
15607 #include <linux/acct.h>
15608 +#include <linux/vserver/global.h>
15610 #define BITS_PER_PAGE (PAGE_SIZE*8)
15612 @@ -85,6 +86,7 @@ static struct pid_namespace *create_pid_
15615 kref_init(&ns->kref);
15616 + atomic_inc(&vs_global_pid_ns);
15619 set_bit(0, ns->pidmap[0].page);
15620 @@ -109,6 +111,7 @@ static void destroy_pid_namespace(struct
15622 for (i = 0; i < PIDMAP_ENTRIES; i++)
15623 kfree(ns->pidmap[i].page);
15624 + atomic_dec(&vs_global_pid_ns);
15625 kmem_cache_free(pid_ns_cachep, ns);
15628 diff -NurpP --minimal linux-2.6.29.1/kernel/posix-timers.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/posix-timers.c
15629 --- linux-2.6.29.1/kernel/posix-timers.c 2009-03-24 14:22:44.000000000 +0100
15630 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/posix-timers.c 2009-03-24 17:36:18.000000000 +0100
15632 #include <linux/wait.h>
15633 #include <linux/workqueue.h>
15634 #include <linux/module.h>
15635 +#include <linux/vs_context.h>
15638 * Management arrays for POSIX timers. Timers are kept in slab memory
15639 @@ -321,6 +322,12 @@ int posix_timer_event(struct k_itimer *t
15641 struct task_struct *task;
15642 int shared, ret = -1;
15643 + struct vx_info_save vxis;
15644 + struct vx_info *vxi;
15646 + /* FIXME: move that down where we have the task */
15647 + vxi = task_get_vx_info(timr->it_process);
15648 + enter_vx_info(vxi, &vxis);
15650 * FIXME: if ->sigq is queued we can race with
15651 * dequeue_signal()->do_schedule_next_timer().
15652 @@ -341,6 +348,9 @@ int posix_timer_event(struct k_itimer *t
15653 ret = send_sigqueue(timr->sigq, task, shared);
15657 + leave_vx_info(&vxis);
15658 + put_vx_info(vxi);
15659 /* If we failed to send the signal the timer stops. */
15662 diff -NurpP --minimal linux-2.6.29.1/kernel/printk.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/printk.c
15663 --- linux-2.6.29.1/kernel/printk.c 2009-03-24 14:22:44.000000000 +0100
15664 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/printk.c 2009-03-24 14:48:36.000000000 +0100
15666 #include <linux/security.h>
15667 #include <linux/bootmem.h>
15668 #include <linux/syscalls.h>
15669 +#include <linux/vs_cvirt.h>
15671 #include <asm/uaccess.h>
15673 @@ -251,18 +252,13 @@ int do_syslog(int type, char __user *buf
15674 unsigned i, j, limit, count;
15680 error = security_syslog(type);
15685 - case 0: /* Close log */
15687 - case 1: /* Open log */
15689 - case 2: /* Read from log */
15690 + if ((type >= 2) && (type <= 4)) {
15692 if (!buf || len < 0)
15694 @@ -273,6 +269,16 @@ int do_syslog(int type, char __user *buf
15699 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
15700 + return vx_do_syslog(type, buf, len);
15703 + case 0: /* Close log */
15705 + case 1: /* Open log */
15707 + case 2: /* Read from log */
15708 error = wait_event_interruptible(log_wait,
15709 (log_start - log_end));
15711 @@ -297,16 +303,6 @@ int do_syslog(int type, char __user *buf
15714 case 3: /* Read last kernel messages */
15716 - if (!buf || len < 0)
15721 - if (!access_ok(VERIFY_WRITE, buf, len)) {
15726 if (count > log_buf_len)
15727 count = log_buf_len;
15728 diff -NurpP --minimal linux-2.6.29.1/kernel/ptrace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/ptrace.c
15729 --- linux-2.6.29.1/kernel/ptrace.c 2009-03-24 14:22:44.000000000 +0100
15730 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/ptrace.c 2009-03-24 14:48:36.000000000 +0100
15732 #include <linux/audit.h>
15733 #include <linux/pid_namespace.h>
15734 #include <linux/syscalls.h>
15735 +#include <linux/vs_context.h>
15737 #include <asm/pgtable.h>
15738 #include <asm/uaccess.h>
15739 @@ -159,6 +160,11 @@ int __ptrace_may_access(struct task_stru
15740 dumpable = get_dumpable(task->mm);
15741 if (!dumpable && !capable(CAP_SYS_PTRACE))
15743 + if (!vx_check(task->xid, VS_ADMIN_P|VS_IDENT))
15745 + if (!vx_check(task->xid, VS_IDENT) &&
15746 + !task_vx_flags(task, VXF_STATE_ADMIN, 0))
15749 return security_ptrace_may_access(task, mode);
15751 @@ -596,6 +602,10 @@ SYSCALL_DEFINE4(ptrace, long, request, l
15756 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
15757 + goto out_put_task_struct;
15759 if (request == PTRACE_ATTACH) {
15760 ret = ptrace_attach(child);
15762 diff -NurpP --minimal linux-2.6.29.1/kernel/sched.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched.c
15763 --- linux-2.6.29.1/kernel/sched.c 2009-03-24 14:22:44.000000000 +0100
15764 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched.c 2009-03-24 21:52:21.000000000 +0100
15766 #include <linux/debugfs.h>
15767 #include <linux/ctype.h>
15768 #include <linux/ftrace.h>
15769 +#include <linux/vs_sched.h>
15770 +#include <linux/vs_cvirt.h>
15771 #include <trace/sched.h>
15773 #include <asm/tlb.h>
15774 @@ -610,6 +612,16 @@ struct rq {
15776 struct hrtimer hrtick_timer;
15778 + unsigned long norm_time;
15779 + unsigned long idle_time;
15780 +#ifdef CONFIG_VSERVER_IDLETIME
15783 +#ifdef CONFIG_VSERVER_HARDCPU
15784 + struct list_head hold_queue;
15785 + unsigned long nr_onhold;
15789 #ifdef CONFIG_SCHEDSTATS
15790 /* latency stats */
15791 @@ -1705,6 +1717,7 @@ static void update_avg(u64 *avg, u64 sam
15793 static void enqueue_task(struct rq *rq, struct task_struct *p, int wakeup)
15795 + // BUG_ON(p->state & TASK_ONHOLD);
15796 sched_info_queued(p);
15797 p->sched_class->enqueue_task(rq, p, wakeup);
15799 @@ -1828,6 +1841,8 @@ static inline void check_class_changed(s
15800 p->sched_class->prio_changed(rq, p, oldprio, running);
15803 +#include "sched_mon.h"
15807 /* Used instead of source_load when we know the type == 0 */
15808 @@ -1915,6 +1930,7 @@ migrate_task(struct task_struct *p, int
15810 struct rq *rq = task_rq(p);
15812 + vxm_migrate_task(p, rq, dest_cpu);
15814 * If the task is not on a runqueue (and not running), then
15815 * it is sufficient to simply update the task's cpu field.
15816 @@ -2242,6 +2258,8 @@ static int sched_balance_self(int cpu, i
15818 #endif /* CONFIG_SMP */
15820 +#include "sched_hard.h"
15823 * try_to_wake_up - wake up a thread
15824 * @p: the to-be-woken-up thread
15825 @@ -2286,6 +2304,13 @@ static int try_to_wake_up(struct task_st
15826 rq = task_rq_lock(p, &flags);
15827 update_rq_clock(rq);
15828 old_state = p->state;
15830 + /* we need to unhold suspended tasks */
15831 + if (old_state & TASK_ONHOLD) {
15832 + vx_unhold_task(p, rq);
15833 + old_state = p->state;
15836 if (!(old_state & state))
15839 @@ -2307,6 +2332,12 @@ static int try_to_wake_up(struct task_st
15840 /* might preempt at this point */
15841 rq = task_rq_lock(p, &flags);
15842 old_state = p->state;
15844 + /* we need to unhold suspended tasks
15845 + if (old_state & TASK_ONHOLD) {
15846 + vx_unhold_task(p, rq);
15847 + old_state = p->state;
15849 if (!(old_state & state))
15852 @@ -4169,16 +4200,19 @@ void account_user_time(struct task_struc
15853 cputime_t cputime_scaled)
15855 struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15856 + struct vx_info *vxi = p->vx_info; /* p is _always_ current */
15858 + int nice = (TASK_NICE(p) > 0);
15860 /* Add user time to process. */
15861 p->utime = cputime_add(p->utime, cputime);
15862 p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
15863 + vx_account_user(vxi, cputime, nice);
15864 account_group_user_time(p, cputime);
15866 /* Add user time to cpustat. */
15867 tmp = cputime_to_cputime64(cputime);
15868 - if (TASK_NICE(p) > 0)
15870 cpustat->nice = cputime64_add(cpustat->nice, tmp);
15872 cpustat->user = cputime64_add(cpustat->user, tmp);
15873 @@ -4222,6 +4256,7 @@ void account_system_time(struct task_str
15874 cputime_t cputime, cputime_t cputime_scaled)
15876 struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15877 + struct vx_info *vxi = p->vx_info; /* p is _always_ current */
15880 if ((p->flags & PF_VCPU) && (irq_count() - hardirq_offset == 0)) {
15881 @@ -4232,6 +4267,7 @@ void account_system_time(struct task_str
15882 /* Add system time to process. */
15883 p->stime = cputime_add(p->stime, cputime);
15884 p->stimescaled = cputime_add(p->stimescaled, cputime_scaled);
15885 + vx_account_system(vxi, cputime, 0 /* do we have idle time? */);
15886 account_group_system_time(p, cputime);
15888 /* Add system time to cpustat. */
15889 @@ -4587,6 +4623,11 @@ need_resched_nonpreemptible:
15890 idle_balance(cpu, rq);
15892 prev->sched_class->put_prev_task(rq, prev);
15894 + vx_set_rq_time(rq, jiffies); /* update time */
15895 + vx_schedule(prev, rq, cpu); /* hold if over limit */
15896 + vx_try_unhold(rq, cpu); /* unhold if refilled */
15898 next = pick_next_task(rq, prev);
15900 if (likely(prev != next)) {
15901 @@ -5152,7 +5193,7 @@ SYSCALL_DEFINE1(nice, int, increment)
15904 if (increment < 0 && !can_nice(current, nice))
15906 + return vx_flags(VXF_IGNEG_NICE, 0) ? 0 : -EPERM;
15908 retval = security_task_setnice(current, nice);
15910 @@ -8428,7 +8469,10 @@ void __init sched_init(void)
15913 #endif /* CONFIG_FAIR_GROUP_SCHED */
15915 +#ifdef CONFIG_VSERVER_HARDCPU
15916 + INIT_LIST_HEAD(&rq->hold_queue);
15917 + rq->nr_onhold = 0;
15919 rq->rt.rt_runtime = def_rt_bandwidth.rt_runtime;
15920 #ifdef CONFIG_RT_GROUP_SCHED
15921 INIT_LIST_HEAD(&rq->leaf_rt_rq_list);
15922 diff -NurpP --minimal linux-2.6.29.1/kernel/sched_fair.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_fair.c
15923 --- linux-2.6.29.1/kernel/sched_fair.c 2009-03-24 14:22:45.000000000 +0100
15924 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_fair.c 2009-03-24 14:48:36.000000000 +0100
15925 @@ -717,6 +717,9 @@ enqueue_entity(struct cfs_rq *cfs_rq, st
15926 check_spread(cfs_rq, se);
15927 if (se != cfs_rq->curr)
15928 __enqueue_entity(cfs_rq, se);
15930 + if (entity_is_task(se))
15931 + vx_activate_task(task_of(se));
15934 static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
15935 @@ -760,6 +763,8 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
15937 if (se != cfs_rq->curr)
15938 __dequeue_entity(cfs_rq, se);
15939 + if (entity_is_task(se))
15940 + vx_deactivate_task(task_of(se));
15941 account_entity_dequeue(cfs_rq, se);
15942 update_min_vruntime(cfs_rq);
15944 diff -NurpP --minimal linux-2.6.29.1/kernel/sched_hard.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_hard.h
15945 --- linux-2.6.29.1/kernel/sched_hard.h 1970-01-01 01:00:00.000000000 +0100
15946 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_hard.h 2009-02-22 22:54:26.000000000 +0100
15949 +#ifdef CONFIG_VSERVER_IDLELIMIT
15952 + * vx_idle_resched - reschedule after maxidle
15955 +void vx_idle_resched(struct rq *rq)
15957 + /* maybe have a better criterion for paused */
15958 + if (!--rq->idle_tokens && !list_empty(&rq->hold_queue))
15959 + set_need_resched();
15962 +#else /* !CONFIG_VSERVER_IDLELIMIT */
15964 +#define vx_idle_resched(rq)
15966 +#endif /* CONFIG_VSERVER_IDLELIMIT */
15970 +#ifdef CONFIG_VSERVER_IDLETIME
15972 +#define vx_set_rq_min_skip(rq, min) \
15973 + (rq)->idle_skip = (min)
15975 +#define vx_save_min_skip(ret, min, val) \
15976 + __vx_save_min_skip(ret, min, val)
15979 +void __vx_save_min_skip(int ret, int *min, int val)
15983 + if ((*min > val) || !*min)
15988 +int vx_try_skip(struct rq *rq, int cpu)
15990 + /* artificially advance time */
15991 + if (rq->idle_skip > 0) {
15992 + vxdprintk(list_empty(&rq->hold_queue),
15993 + "hold queue empty on cpu %d", cpu);
15994 + rq->idle_time += rq->idle_skip;
15995 + vxm_idle_skip(rq, cpu);
16001 +#else /* !CONFIG_VSERVER_IDLETIME */
16003 +#define vx_set_rq_min_skip(rq, min) \
16004 + ({ int dummy = (min); dummy; })
16006 +#define vx_save_min_skip(ret, min, val)
16009 +int vx_try_skip(struct rq *rq, int cpu)
16014 +#endif /* CONFIG_VSERVER_IDLETIME */
16018 +#ifdef CONFIG_VSERVER_HARDCPU
16020 +#define vx_set_rq_max_idle(rq, max) \
16021 + (rq)->idle_tokens = (max)
16023 +#define vx_save_max_idle(ret, min, val) \
16024 + __vx_save_max_idle(ret, min, val)
16027 +void __vx_save_max_idle(int ret, int *min, int val)
16035 + * vx_hold_task - put a task on the hold queue
16038 +void vx_hold_task(struct task_struct *p, struct rq *rq)
16040 + // printk("@ hold_task(%p[%lx])\n", p, p->state);
16042 + /* ignore dead/killed tasks */
16043 + if (unlikely(p->state & (TASK_DEAD | TASK_WAKEKILL)))
16046 + /* ignore sleeping tasks */
16047 + if (unlikely(p->state & TASK_NORMAL))
16050 + /* remove task from runqueue */
16051 + if (likely(p->se.on_rq))
16052 + dequeue_task(rq, p, 0);
16054 + printk("@ woops, task %p not on runqueue?\n", p);
16056 + p->state |= TASK_ONHOLD;
16057 + /* a new one on hold */
16059 + vxm_hold_task(p, rq);
16060 + list_add_tail(&p->hq, &rq->hold_queue);
16061 + // list_add_tail(&p->run_list, &rq->hold_queue);
16065 + * vx_unhold_task - put a task back to the runqueue
16068 +void vx_unhold_task(struct task_struct *p, struct rq *rq)
16070 + // printk("@ unhold_task(%p[%lx])\n", p, p->state);
16071 + list_del_init(&p->hq);
16072 + // list_del(&p->run_list);
16073 + /* one less waiting */
16075 + p->state &= ~TASK_ONHOLD;
16076 + enqueue_task(rq, p, 0);
16077 + // ? inc_nr_running(p, rq);
16078 + vxm_unhold_task(p, rq);
16082 + * vx_remove_hold - remove a task from the hold queue
16085 +void vx_remove_hold(struct task_struct *p, struct rq *rq)
16087 + printk("@ remove_hold(%p[%lx])\n", p, p->state);
16088 + list_del_init(&p->hq);
16089 + // list_del(&p->run_list);
16090 + /* one less waiting */
16092 + p->state &= ~TASK_ONHOLD;
16095 +unsigned long nr_onhold(void)
16097 + unsigned long i, sum = 0;
16099 + for_each_online_cpu(i)
16100 + sum += cpu_rq(i)->nr_onhold;
16108 +int __vx_tokens_avail(struct _vx_sched_pc *sched_pc)
16110 + return sched_pc->tokens;
16114 +void __vx_consume_token(struct _vx_sched_pc *sched_pc)
16116 + sched_pc->tokens--;
16120 +int vx_need_resched(struct task_struct *p, int slice, int cpu)
16122 + struct vx_info *vxi = p->vx_info;
16124 + if (vx_info_flags(vxi, VXF_SCHED_HARD|VXF_SCHED_PRIO, 0)) {
16125 + struct _vx_sched_pc *sched_pc =
16126 + &vx_per_cpu(vxi, sched_pc, cpu);
16129 + /* maybe we can simplify that to decrement
16130 + the token counter unconditional? */
16132 + if ((tokens = __vx_tokens_avail(sched_pc)) > 0)
16133 + __vx_consume_token(sched_pc);
16135 + /* for tokens > 0, one token was consumed */
16139 + vxm_need_resched(p, slice, cpu);
16140 + return (slice == 0);
16144 +#define vx_set_rq_time(rq, time) do { \
16145 + rq->norm_time = time; \
16150 +void vx_try_unhold(struct rq *rq, int cpu)
16152 + struct vx_info *vxi = NULL;
16153 + struct list_head *l, *n;
16154 + int maxidle = HZ;
16157 + /* nothing to do? what about pause? */
16158 + if (list_empty(&rq->hold_queue))
16161 + list_for_each_safe(l, n, &rq->hold_queue) {
16162 + int ret, delta_min[2];
16163 + struct _vx_sched_pc *sched_pc;
16164 + struct task_struct *p;
16166 + p = list_entry(l, struct task_struct, hq);
16167 + /* don't bother with same context */
16168 + if (vxi == p->vx_info)
16171 + vxi = p->vx_info;
16172 + /* ignore paused contexts */
16173 + if (vx_info_flags(vxi, VXF_SCHED_PAUSE, 0))
16176 + sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16178 + /* recalc tokens */
16179 + vxm_sched_info(sched_pc, vxi, cpu);
16180 + ret = vx_tokens_recalc(sched_pc,
16181 + &rq->norm_time, &rq->idle_time, delta_min);
16182 + vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16185 + /* we found a runable context */
16186 + vx_unhold_task(p, rq);
16189 + vx_save_max_idle(ret, &maxidle, delta_min[0]);
16190 + vx_save_min_skip(ret, &minskip, delta_min[1]);
16192 + vx_set_rq_max_idle(rq, maxidle);
16193 + vx_set_rq_min_skip(rq, minskip);
16194 + vxm_rq_max_min(rq, cpu);
16199 +int vx_schedule(struct task_struct *next, struct rq *rq, int cpu)
16201 + struct vx_info *vxi = next->vx_info;
16202 + struct _vx_sched_pc *sched_pc;
16203 + int delta_min[2];
16209 + flags = vxi->vx_flags;
16211 + if (unlikely(vs_check_flags(flags, VXF_SCHED_PAUSE, 0)))
16212 + goto put_on_hold;
16213 + if (!vs_check_flags(flags, VXF_SCHED_HARD | VXF_SCHED_PRIO, 0))
16216 + sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16218 + /* update scheduler params */
16219 + if (cpu_isset(cpu, vxi->sched.update)) {
16220 + vx_update_sched_param(&vxi->sched, sched_pc);
16221 + vxm_update_sched(sched_pc, vxi, cpu);
16222 + cpu_clear(cpu, vxi->sched.update);
16225 + vxm_sched_info(sched_pc, vxi, cpu);
16226 + ret = vx_tokens_recalc(sched_pc,
16227 + &rq->norm_time, &rq->idle_time, delta_min);
16228 + vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16230 + if (!vs_check_flags(flags, VXF_SCHED_HARD, 0))
16233 + if (unlikely(ret < 0)) {
16234 + vx_save_max_idle(ret, &rq->idle_tokens, delta_min[0]);
16235 + vx_save_min_skip(ret, &rq->idle_skip, delta_min[1]);
16236 + vxm_rq_max_min(rq, cpu);
16238 + vx_hold_task(next, rq);
16245 +#else /* CONFIG_VSERVER_HARDCPU */
16248 +void vx_hold_task(struct task_struct *p, struct rq *rq)
16254 +void vx_unhold_task(struct task_struct *p, struct rq *rq)
16259 +unsigned long nr_onhold(void)
16266 +int vx_need_resched(struct task_struct *p, int slice, int cpu)
16268 + return (slice == 0);
16272 +#define vx_set_rq_time(rq, time)
16275 +void vx_try_unhold(struct rq *rq, int cpu)
16281 +int vx_schedule(struct task_struct *next, struct rq *rq, int cpu)
16283 + struct vx_info *vxi = next->vx_info;
16284 + struct _vx_sched_pc *sched_pc;
16285 + int delta_min[2];
16288 + if (!vx_info_flags(vxi, VXF_SCHED_PRIO, 0))
16291 + sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16292 + vxm_sched_info(sched_pc, vxi, cpu);
16293 + ret = vx_tokens_recalc(sched_pc,
16294 + &rq->norm_time, &rq->idle_time, delta_min);
16295 + vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16299 +#endif /* CONFIG_VSERVER_HARDCPU */
16301 diff -NurpP --minimal linux-2.6.29.1/kernel/sched_mon.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_mon.h
16302 --- linux-2.6.29.1/kernel/sched_mon.h 1970-01-01 01:00:00.000000000 +0100
16303 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sched_mon.h 2009-02-22 22:54:26.000000000 +0100
16306 +#include <linux/vserver/monitor.h>
16308 +#ifdef CONFIG_VSERVER_MONITOR
16310 +#ifdef CONFIG_VSERVER_HARDCPU
16311 +#define HARDCPU(x) (x)
16313 +#define HARDCPU(x) (0)
16316 +#ifdef CONFIG_VSERVER_IDLETIME
16317 +#define IDLETIME(x) (x)
16319 +#define IDLETIME(x) (0)
16322 +struct _vx_mon_entry *vxm_advance(int cpu);
16326 +void __vxm_basic(struct _vx_mon_entry *entry, xid_t xid, int type)
16328 + entry->type = type;
16329 + entry->xid = xid;
16333 +void __vxm_sync(int cpu)
16335 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16337 + __vxm_basic(entry, 0, VXM_SYNC);
16338 + entry->ev.sec = xtime.tv_sec;
16339 + entry->ev.nsec = xtime.tv_nsec;
16343 +void __vxm_task(struct task_struct *p, int type)
16345 + struct _vx_mon_entry *entry = vxm_advance(task_cpu(p));
16347 + __vxm_basic(entry, p->xid, type);
16348 + entry->ev.tsk.pid = p->pid;
16349 + entry->ev.tsk.state = p->state;
16353 +void __vxm_sched(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16355 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16357 + __vxm_basic(entry, vxi->vx_id, (VXM_SCHED | s->flags));
16358 + entry->sd.tokens = s->tokens;
16359 + entry->sd.norm_time = s->norm_time;
16360 + entry->sd.idle_time = s->idle_time;
16364 +void __vxm_rqinfo1(struct rq *q, int cpu)
16366 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16368 + entry->type = VXM_RQINFO_1;
16369 + entry->xid = ((unsigned long)q >> 16) & 0xffff;
16370 + entry->q1.running = q->nr_running;
16371 + entry->q1.onhold = HARDCPU(q->nr_onhold);
16372 + entry->q1.iowait = atomic_read(&q->nr_iowait);
16373 + entry->q1.uintr = q->nr_uninterruptible;
16374 + entry->q1.idle_tokens = IDLETIME(q->idle_tokens);
16378 +void __vxm_rqinfo2(struct rq *q, int cpu)
16380 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16382 + entry->type = VXM_RQINFO_2;
16383 + entry->xid = (unsigned long)q & 0xffff;
16384 + entry->q2.norm_time = q->norm_time;
16385 + entry->q2.idle_time = q->idle_time;
16386 + entry->q2.idle_skip = IDLETIME(q->idle_skip);
16390 +void __vxm_update(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16392 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16394 + __vxm_basic(entry, vxi->vx_id, VXM_UPDATE);
16395 + entry->ev.tokens = s->tokens;
16399 +void __vxm_update1(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16401 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16403 + __vxm_basic(entry, vxi->vx_id, VXM_UPDATE_1);
16404 + entry->u1.tokens_max = s->tokens_max;
16405 + entry->u1.fill_rate = s->fill_rate[0];
16406 + entry->u1.interval = s->interval[0];
16410 +void __vxm_update2(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16412 + struct _vx_mon_entry *entry = vxm_advance(cpu);
16414 + __vxm_basic(entry, vxi->vx_id, VXM_UPDATE_2);
16415 + entry->u2.tokens_min = s->tokens_min;
16416 + entry->u2.fill_rate = s->fill_rate[1];
16417 + entry->u2.interval = s->interval[1];
16421 +#define vxm_activate_task(p,q) __vxm_task(p, VXM_ACTIVATE)
16422 +#define vxm_activate_idle(p,q) __vxm_task(p, VXM_IDLE)
16423 +#define vxm_deactivate_task(p,q) __vxm_task(p, VXM_DEACTIVATE)
16424 +#define vxm_hold_task(p,q) __vxm_task(p, VXM_HOLD)
16425 +#define vxm_unhold_task(p,q) __vxm_task(p, VXM_UNHOLD)
16428 +void vxm_migrate_task(struct task_struct *p, struct rq *rq, int dest)
16430 + __vxm_task(p, VXM_MIGRATE);
16431 + __vxm_rqinfo1(rq, task_cpu(p));
16432 + __vxm_rqinfo2(rq, task_cpu(p));
16436 +void vxm_idle_skip(struct rq *rq, int cpu)
16438 + __vxm_rqinfo1(rq, cpu);
16439 + __vxm_rqinfo2(rq, cpu);
16443 +void vxm_need_resched(struct task_struct *p, int slice, int cpu)
16448 + __vxm_task(p, VXM_RESCHED);
16452 +void vxm_sync(unsigned long now, int cpu)
16454 + if (!CONFIG_VSERVER_MONITOR_SYNC ||
16455 + (now % CONFIG_VSERVER_MONITOR_SYNC))
16461 +#define vxm_sched_info(s,v,c) __vxm_sched(s,v,c)
16464 +void vxm_tokens_recalc(struct _vx_sched_pc *s, struct rq *rq,
16465 + struct vx_info *vxi, int cpu)
16467 + __vxm_sched(s, vxi, cpu);
16468 + __vxm_rqinfo2(rq, cpu);
16472 +void vxm_update_sched(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16474 + __vxm_sched(s, vxi, cpu);
16475 + __vxm_update(s, vxi, cpu);
16476 + __vxm_update1(s, vxi, cpu);
16477 + __vxm_update2(s, vxi, cpu);
16481 +void vxm_rq_max_min(struct rq *rq, int cpu)
16483 + __vxm_rqinfo1(rq, cpu);
16484 + __vxm_rqinfo2(rq, cpu);
16487 +#else /* CONFIG_VSERVER_MONITOR */
16489 +#define vxm_activate_task(t,q) do { } while (0)
16490 +#define vxm_activate_idle(t,q) do { } while (0)
16491 +#define vxm_deactivate_task(t,q) do { } while (0)
16492 +#define vxm_hold_task(t,q) do { } while (0)
16493 +#define vxm_unhold_task(t,q) do { } while (0)
16494 +#define vxm_migrate_task(t,q,d) do { } while (0)
16495 +#define vxm_idle_skip(q,c) do { } while (0)
16496 +#define vxm_need_resched(t,s,c) do { } while (0)
16497 +#define vxm_sync(s,c) do { } while (0)
16498 +#define vxm_sched_info(s,v,c) do { } while (0)
16499 +#define vxm_tokens_recalc(s,q,v,c) do { } while (0)
16500 +#define vxm_update_sched(s,v,c) do { } while (0)
16501 +#define vxm_rq_max_min(q,c) do { } while (0)
16503 +#endif /* CONFIG_VSERVER_MONITOR */
16505 diff -NurpP --minimal linux-2.6.29.1/kernel/signal.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/signal.c
16506 --- linux-2.6.29.1/kernel/signal.c 2009-03-24 14:22:45.000000000 +0100
16507 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/signal.c 2009-03-24 14:48:36.000000000 +0100
16509 #include <linux/freezer.h>
16510 #include <linux/pid_namespace.h>
16511 #include <linux/nsproxy.h>
16512 +#include <linux/vs_context.h>
16513 +#include <linux/vs_pid.h>
16514 #include <trace/sched.h>
16516 #include <asm/param.h>
16517 @@ -584,6 +586,14 @@ static int check_kill_permission(int sig
16518 if (!valid_signal(sig))
16521 + if ((info != SEND_SIG_NOINFO) &&
16522 + (is_si_special(info) || !SI_FROMUSER(info)))
16525 + vxdprintk(VXD_CBIT(misc, 7),
16526 + "check_kill_permission(%d,%p,%p[#%u,%u])",
16527 + sig, info, t, vx_task_xid(t), t->pid);
16529 if (info != SEND_SIG_NOINFO && (is_si_special(info) || SI_FROMKERNEL(info)))
16532 @@ -611,6 +621,18 @@ static int check_kill_permission(int sig
16537 + if (t->pid == 1 && current->xid)
16541 + if (!vx_check(vx_task_xid(t), VS_WATCH_P | VS_IDENT)) {
16542 + vxdprintk(current->xid || VXD_CBIT(misc, 7),
16543 + "signal %d[%p] xid mismatch %p[#%u,%u] xid=#%u",
16544 + sig, info, t, vx_task_xid(t), t->pid, current->xid);
16548 return security_task_kill(t, info, sig, 0);
16551 @@ -1076,7 +1098,7 @@ int kill_pid_info(int sig, struct siginf
16554 p = pid_task(pid, PIDTYPE_PID);
16556 + if (p && vx_check(vx_task_xid(p), VS_ADMIN | VS_IDENT)) {
16557 error = group_send_sig_info(sig, info, p);
16558 if (unlikely(error == -ESRCH))
16560 @@ -1115,7 +1137,7 @@ int kill_pid_info_as_uid(int sig, struct
16562 read_lock(&tasklist_lock);
16563 p = pid_task(pid, PIDTYPE_PID);
16565 + if (!p || !vx_check(vx_task_xid(p), VS_IDENT)) {
16569 @@ -1169,8 +1191,10 @@ static int kill_something_info(int sig,
16570 struct task_struct * p;
16572 for_each_process(p) {
16573 - if (task_pid_vnr(p) > 1 &&
16574 - !same_thread_group(p, current)) {
16575 + if (vx_check(vx_task_xid(p), VS_ADMIN|VS_IDENT) &&
16576 + task_pid_vnr(p) > 1 &&
16577 + !same_thread_group(p, current) &&
16578 + !vx_current_initpid(p->pid)) {
16579 int err = group_send_sig_info(sig, info, p);
16582 @@ -1849,6 +1873,11 @@ relock:
16583 !signal_group_exit(signal))
16586 + /* virtual init is protected against user signals */
16587 + if ((info->si_code == SI_USER) &&
16588 + vx_current_initpid(current->pid))
16591 if (sig_kernel_stop(signr)) {
16593 * The default action is to stop all threads in
16594 diff -NurpP --minimal linux-2.6.29.1/kernel/softirq.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/softirq.c
16595 --- linux-2.6.29.1/kernel/softirq.c 2009-03-24 14:22:45.000000000 +0100
16596 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/softirq.c 2009-03-24 14:48:36.000000000 +0100
16598 #include <linux/rcupdate.h>
16599 #include <linux/smp.h>
16600 #include <linux/tick.h>
16601 +#include <linux/vs_context.h>
16603 #include <asm/irq.h>
16605 diff -NurpP --minimal linux-2.6.29.1/kernel/sys.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sys.c
16606 --- linux-2.6.29.1/kernel/sys.c 2009-03-24 14:22:45.000000000 +0100
16607 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sys.c 2009-03-24 14:48:36.000000000 +0100
16609 #include <linux/syscalls.h>
16610 #include <linux/kprobes.h>
16611 #include <linux/user_namespace.h>
16612 +#include <linux/vs_pid.h>
16614 #include <asm/uaccess.h>
16615 #include <asm/io.h>
16616 @@ -128,7 +129,10 @@ static int set_one_prio(struct task_stru
16619 if (niceval < task_nice(p) && !can_nice(p, niceval)) {
16621 + if (vx_flags(VXF_IGNEG_NICE, 0))
16627 no_nice = security_task_setnice(p, niceval);
16628 @@ -177,6 +181,8 @@ SYSCALL_DEFINE3(setpriority, int, which,
16630 pgrp = task_pgrp(current);
16631 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
16632 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
16634 error = set_one_prio(p, niceval, error);
16635 } while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
16637 @@ -238,6 +244,8 @@ SYSCALL_DEFINE2(getpriority, int, which,
16639 pgrp = task_pgrp(current);
16640 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
16641 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
16643 niceval = 20 - task_nice(p);
16644 if (niceval > retval)
16646 @@ -347,6 +355,9 @@ void kernel_power_off(void)
16647 machine_power_off();
16649 EXPORT_SYMBOL_GPL(kernel_power_off);
16651 +long vs_reboot(unsigned int, void __user *);
16654 * Reboot system call: for obvious reasons only root may call it,
16655 * and even root needs to set up some magic numbers in the registers
16656 @@ -378,6 +389,9 @@ SYSCALL_DEFINE4(reboot, int, magic1, int
16657 if ((cmd == LINUX_REBOOT_CMD_POWER_OFF) && !pm_power_off)
16658 cmd = LINUX_REBOOT_CMD_HALT;
16660 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
16661 + return vs_reboot(cmd, arg);
16665 case LINUX_REBOOT_CMD_RESTART:
16666 @@ -1420,7 +1434,7 @@ SYSCALL_DEFINE2(sethostname, char __user
16668 char tmp[__NEW_UTS_LEN];
16670 - if (!capable(CAP_SYS_ADMIN))
16671 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
16673 if (len < 0 || len > __NEW_UTS_LEN)
16675 @@ -1469,7 +1483,7 @@ SYSCALL_DEFINE2(setdomainname, char __us
16677 char tmp[__NEW_UTS_LEN];
16679 - if (!capable(CAP_SYS_ADMIN))
16680 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
16682 if (len < 0 || len > __NEW_UTS_LEN)
16684 @@ -1538,7 +1552,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int,
16686 old_rlim = current->signal->rlim + resource;
16687 if ((new_rlim.rlim_max > old_rlim->rlim_max) &&
16688 - !capable(CAP_SYS_RESOURCE))
16689 + !vx_capable(CAP_SYS_RESOURCE, VXC_SET_RLIMIT))
16691 if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open)
16693 diff -NurpP --minimal linux-2.6.29.1/kernel/sysctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sysctl.c
16694 --- linux-2.6.29.1/kernel/sysctl.c 2009-03-24 14:22:45.000000000 +0100
16695 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sysctl.c 2009-03-24 14:48:36.000000000 +0100
16696 @@ -114,6 +114,7 @@ static int ngroups_max = NGROUPS_MAX;
16697 #ifdef CONFIG_MODULES
16698 extern char modprobe_path[];
16700 +extern char vshelper_path[];
16701 #ifdef CONFIG_CHR_DEV_SG
16702 extern int sg_big_buff;
16704 @@ -545,6 +546,15 @@ static struct ctl_table kern_table[] = {
16705 .strategy = &sysctl_string,
16709 + .ctl_name = KERN_VSHELPER,
16710 + .procname = "vshelper",
16711 + .data = &vshelper_path,
16714 + .proc_handler = &proc_dostring,
16715 + .strategy = &sysctl_string,
16717 #ifdef CONFIG_CHR_DEV_SG
16719 .ctl_name = KERN_SG_BIG_BUFF,
16720 diff -NurpP --minimal linux-2.6.29.1/kernel/sysctl_check.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sysctl_check.c
16721 --- linux-2.6.29.1/kernel/sysctl_check.c 2009-03-24 14:22:45.000000000 +0100
16722 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/sysctl_check.c 2009-03-24 14:48:36.000000000 +0100
16723 @@ -39,6 +39,7 @@ static const struct trans_ctl_table tran
16725 { KERN_PANIC, "panic" },
16726 { KERN_REALROOTDEV, "real-root-dev" },
16727 + { KERN_VSHELPER, "vshelper", },
16729 { KERN_SPARC_REBOOT, "reboot-cmd" },
16730 { KERN_CTLALTDEL, "ctrl-alt-del" },
16731 @@ -1216,6 +1217,22 @@ static const struct trans_ctl_table tran
16735 +static struct trans_ctl_table trans_vserver_table[] = {
16736 + { 1, "debug_switch" },
16737 + { 2, "debug_xid" },
16738 + { 3, "debug_nid" },
16739 + { 4, "debug_tag" },
16740 + { 5, "debug_net" },
16741 + { 6, "debug_limit" },
16742 + { 7, "debug_cres" },
16743 + { 8, "debug_dlim" },
16744 + { 9, "debug_quota" },
16745 + { 10, "debug_cvirt" },
16746 + { 11, "debug_space" },
16747 + { 12, "debug_misc" },
16751 static const struct trans_ctl_table trans_root_table[] = {
16752 { CTL_KERN, "kernel", trans_kern_table },
16753 { CTL_VM, "vm", trans_vm_table },
16754 @@ -1232,6 +1249,7 @@ static const struct trans_ctl_table tran
16755 { CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
16756 { CTL_PM, "pm", trans_pm_table },
16757 { CTL_FRV, "frv", trans_frv_table },
16758 + { CTL_VSERVER, "vserver", trans_vserver_table },
16762 diff -NurpP --minimal linux-2.6.29.1/kernel/time.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/time.c
16763 --- linux-2.6.29.1/kernel/time.c 2009-03-24 14:22:45.000000000 +0100
16764 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/time.c 2009-03-24 14:48:36.000000000 +0100
16765 @@ -63,6 +63,7 @@ EXPORT_SYMBOL(sys_tz);
16766 SYSCALL_DEFINE1(time, time_t __user *, tloc)
16768 time_t i = get_seconds();
16769 +/* FIXME: do_gettimeofday(&tv) -> vx_gettimeofday(&tv) */
16772 if (put_user(i,tloc))
16773 @@ -93,7 +94,7 @@ SYSCALL_DEFINE1(stime, time_t __user *,
16777 - do_settimeofday(&tv);
16778 + vx_settimeofday(&tv);
16782 @@ -104,7 +105,7 @@ SYSCALL_DEFINE2(gettimeofday, struct tim
16784 if (likely(tv != NULL)) {
16785 struct timeval ktv;
16786 - do_gettimeofday(&ktv);
16787 + vx_gettimeofday(&ktv);
16788 if (copy_to_user(tv, &ktv, sizeof(ktv)))
16791 @@ -179,7 +180,7 @@ int do_sys_settimeofday(struct timespec
16792 /* SMP safe, again the code in arch/foo/time.c should
16793 * globally block out interrupts when it runs.
16795 - return do_settimeofday(tv);
16796 + return vx_settimeofday(tv);
16800 @@ -311,7 +312,7 @@ void getnstimeofday(struct timespec *tv)
16804 - do_gettimeofday(&x);
16805 + vx_gettimeofday(&x);
16806 tv->tv_sec = x.tv_sec;
16807 tv->tv_nsec = x.tv_usec * NSEC_PER_USEC;
16809 diff -NurpP --minimal linux-2.6.29.1/kernel/timer.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/timer.c
16810 --- linux-2.6.29.1/kernel/timer.c 2009-03-24 14:22:45.000000000 +0100
16811 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/timer.c 2009-03-24 17:41:02.000000000 +0100
16813 #include <linux/delay.h>
16814 #include <linux/tick.h>
16815 #include <linux/kallsyms.h>
16816 +#include <linux/vs_base.h>
16817 +#include <linux/vs_cvirt.h>
16818 +#include <linux/vs_pid.h>
16819 +#include <linux/vserver/sched.h>
16821 #include <asm/uaccess.h>
16822 #include <asm/unistd.h>
16823 @@ -1018,6 +1022,25 @@ unsigned long get_next_timer_interrupt(u
16828 +void __vx_consume_token(struct _vx_sched_pc *sched_pc)
16830 + sched_pc->tokens--;
16834 +void vx_hard_tick(struct task_struct *p, int cpu)
16836 + struct vx_info *vxi = p->vx_info;
16838 + if (vx_info_flags(vxi, VXF_SCHED_HARD|VXF_SCHED_PRIO, 0)) {
16839 + struct _vx_sched_pc *sched_pc =
16840 + &vx_per_cpu(vxi, sched_pc, cpu);
16842 + __vx_consume_token(sched_pc);
16847 * Called from the timer interrupt handler to charge one tick to the current
16848 * process. user_tick is 1 if the tick is user time, 0 for system.
16849 @@ -1034,6 +1057,7 @@ void update_process_times(int user_tick)
16850 rcu_check_callbacks(cpu, user_tick);
16853 + vx_hard_tick(p, cpu);
16854 run_posix_cpu_timers(p);
16857 @@ -1136,12 +1160,6 @@ SYSCALL_DEFINE1(alarm, unsigned int, sec
16864 - * The Alpha uses getxpid, getxuid, and getxgid instead. Maybe this
16865 - * should be moved into arch/i386 instead?
16869 * sys_getpid - return the thread group id of the current process
16870 @@ -1170,10 +1188,23 @@ SYSCALL_DEFINE0(getppid)
16872 pid = task_tgid_vnr(current->real_parent);
16874 + return vx_map_pid(pid);
16881 + * The Alpha uses getxpid, getxuid, and getxgid instead.
16884 +asmlinkage long do_getxpid(long *ppid)
16886 + *ppid = sys_getppid();
16887 + return sys_getpid();
16890 +#else /* _alpha_ */
16892 SYSCALL_DEFINE0(getuid)
16894 /* Only we change this so SMP safe */
16895 @@ -1344,6 +1375,8 @@ int do_sysinfo(struct sysinfo *info)
16896 tp.tv_nsec = tp.tv_nsec - NSEC_PER_SEC;
16899 + if (vx_flags(VXF_VIRT_UPTIME, 0))
16900 + vx_vsi_uptime(&tp, NULL);
16901 info->uptime = tp.tv_sec + (tp.tv_nsec ? 1 : 0);
16903 info->loads[0] = avenrun[0] << (SI_LOAD_SHIFT - FSHIFT);
16904 diff -NurpP --minimal linux-2.6.29.1/kernel/user.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/user.c
16905 --- linux-2.6.29.1/kernel/user.c 2009-03-24 14:22:45.000000000 +0100
16906 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/user.c 2009-03-24 17:47:32.000000000 +0100
16907 @@ -249,10 +249,10 @@ static struct kobj_type uids_ktype = {
16909 * See Documentation/scheduler/sched-design-CFS.txt for ramifications.
16911 -static int uids_user_create(struct user_struct *up)
16912 +static int uids_user_create(struct user_namespace *ns, struct user_struct *up)
16914 struct kobject *kobj = &up->kobj;
16918 memset(kobj, 0, sizeof(struct kobject));
16919 if (up->user_ns != &init_user_ns)
16920 @@ -280,7 +280,7 @@ int __init uids_sysfs_init(void)
16924 - return uids_user_create(&root_user);
16925 + return uids_user_create(NULL, &root_user);
16928 /* work function to remove sysfs directory for a user and free up
16929 @@ -342,7 +342,8 @@ static void free_user(struct user_struct
16930 #else /* CONFIG_USER_SCHED && CONFIG_SYSFS */
16932 int uids_sysfs_init(void) { return 0; }
16933 -static inline int uids_user_create(struct user_struct *up) { return 0; }
16934 +static inline int uids_user_create(struct user_namespace *ns,
16935 + struct user_struct *up) { return 0; }
16936 static inline void uids_mutex_lock(void) { }
16937 static inline void uids_mutex_unlock(void) { }
16939 @@ -439,7 +440,7 @@ struct user_struct *alloc_uid(struct use
16941 new->user_ns = get_user_ns(ns);
16943 - if (uids_user_create(new))
16944 + if (uids_user_create(ns, new))
16945 goto out_destoy_sched;
16948 diff -NurpP --minimal linux-2.6.29.1/kernel/user_namespace.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/user_namespace.c
16949 --- linux-2.6.29.1/kernel/user_namespace.c 2009-03-24 14:22:45.000000000 +0100
16950 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/user_namespace.c 2009-03-24 18:05:19.000000000 +0100
16952 #include <linux/slab.h>
16953 #include <linux/user_namespace.h>
16954 #include <linux/cred.h>
16955 +#include <linux/vserver/global.h>
16958 * Create a new user namespace, deriving the creator from the user in the
16959 @@ -30,6 +31,7 @@ int create_user_ns(struct cred *new)
16962 kref_init(&ns->kref);
16963 + atomic_inc(&vs_global_user_ns);
16965 for (n = 0; n < UIDHASH_SZ; ++n)
16966 INIT_HLIST_HEAD(ns->uidhash_table + n);
16967 @@ -78,6 +80,8 @@ void free_user_ns(struct kref *kref)
16968 struct user_namespace *ns =
16969 container_of(kref, struct user_namespace, kref);
16971 + /* FIXME: maybe move into destroyer? */
16972 + atomic_dec(&vs_global_user_ns);
16973 INIT_WORK(&ns->destroyer, free_user_ns_work);
16974 schedule_work(&ns->destroyer);
16976 diff -NurpP --minimal linux-2.6.29.1/kernel/utsname.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/utsname.c
16977 --- linux-2.6.29.1/kernel/utsname.c 2008-12-25 00:26:37.000000000 +0100
16978 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/utsname.c 2009-02-22 22:54:26.000000000 +0100
16980 #include <linux/utsname.h>
16981 #include <linux/err.h>
16982 #include <linux/slab.h>
16983 +#include <linux/vserver/global.h>
16986 * Clone a new ns copying an original utsname, setting refcount to 1
16987 @@ -32,6 +33,7 @@ static struct uts_namespace *clone_uts_n
16988 memcpy(&ns->name, &old_ns->name, sizeof(ns->name));
16990 kref_init(&ns->kref);
16991 + atomic_inc(&vs_global_uts_ns);
16995 @@ -62,5 +64,6 @@ void free_uts_ns(struct kref *kref)
16996 struct uts_namespace *ns;
16998 ns = container_of(kref, struct uts_namespace, kref);
16999 + atomic_dec(&vs_global_uts_ns);
17002 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cacct.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct.c
17003 --- linux-2.6.29.1/kernel/vserver/cacct.c 1970-01-01 01:00:00.000000000 +0100
17004 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct.c 2009-02-22 22:54:26.000000000 +0100
17007 + * linux/kernel/vserver/cacct.c
17009 + * Virtual Server: Context Accounting
17011 + * Copyright (C) 2006-2007 Herbert Pötzl
17013 + * V0.01 added accounting stats
17017 +#include <linux/types.h>
17018 +#include <linux/vs_context.h>
17019 +#include <linux/vserver/cacct_cmd.h>
17020 +#include <linux/vserver/cacct_int.h>
17022 +#include <asm/errno.h>
17023 +#include <asm/uaccess.h>
17026 +int vc_sock_stat(struct vx_info *vxi, void __user *data)
17028 + struct vcmd_sock_stat_v0 vc_data;
17031 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17034 + field = vc_data.field;
17035 + if ((field < 0) || (field >= VXA_SOCK_SIZE))
17038 + for (j = 0; j < 3; j++) {
17039 + vc_data.count[j] = vx_sock_count(&vxi->cacct, field, j);
17040 + vc_data.total[j] = vx_sock_total(&vxi->cacct, field, j);
17043 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17048 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cacct_init.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct_init.h
17049 --- linux-2.6.29.1/kernel/vserver/cacct_init.h 1970-01-01 01:00:00.000000000 +0100
17050 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct_init.h 2009-02-22 22:54:26.000000000 +0100
17054 +static inline void vx_info_init_cacct(struct _vx_cacct *cacct)
17059 + for (i = 0; i < VXA_SOCK_SIZE; i++) {
17060 + for (j = 0; j < 3; j++) {
17061 + atomic_set(&cacct->sock[i][j].count, 0);
17062 + atomic_set(&cacct->sock[i][j].total, 0);
17065 + for (i = 0; i < 8; i++)
17066 + atomic_set(&cacct->slab[i], 0);
17067 + for (i = 0; i < 5; i++)
17068 + for (j = 0; j < 4; j++)
17069 + atomic_set(&cacct->page[i][j], 0);
17072 +static inline void vx_info_exit_cacct(struct _vx_cacct *cacct)
17077 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cacct_proc.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct_proc.h
17078 --- linux-2.6.29.1/kernel/vserver/cacct_proc.h 1970-01-01 01:00:00.000000000 +0100
17079 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cacct_proc.h 2009-02-22 22:54:26.000000000 +0100
17081 +#ifndef _VX_CACCT_PROC_H
17082 +#define _VX_CACCT_PROC_H
17084 +#include <linux/vserver/cacct_int.h>
17087 +#define VX_SOCKA_TOP \
17088 + "Type\t recv #/bytes\t\t send #/bytes\t\t fail #/bytes\n"
17090 +static inline int vx_info_proc_cacct(struct _vx_cacct *cacct, char *buffer)
17092 + int i, j, length = 0;
17093 + static char *type[VXA_SOCK_SIZE] = {
17094 + "UNSPEC", "UNIX", "INET", "INET6", "PACKET", "OTHER"
17097 + length += sprintf(buffer + length, VX_SOCKA_TOP);
17098 + for (i = 0; i < VXA_SOCK_SIZE; i++) {
17099 + length += sprintf(buffer + length, "%s:", type[i]);
17100 + for (j = 0; j < 3; j++) {
17101 + length += sprintf(buffer + length,
17102 + "\t%10lu/%-10lu",
17103 + vx_sock_count(cacct, i, j),
17104 + vx_sock_total(cacct, i, j));
17106 + buffer[length++] = '\n';
17109 + length += sprintf(buffer + length, "\n");
17110 + length += sprintf(buffer + length,
17111 + "slab:\t %8u %8u %8u %8u\n",
17112 + atomic_read(&cacct->slab[1]),
17113 + atomic_read(&cacct->slab[4]),
17114 + atomic_read(&cacct->slab[0]),
17115 + atomic_read(&cacct->slab[2]));
17117 + length += sprintf(buffer + length, "\n");
17118 + for (i = 0; i < 5; i++) {
17119 + length += sprintf(buffer + length,
17120 + "page[%d]: %8u %8u %8u %8u\t %8u %8u %8u %8u\n", i,
17121 + atomic_read(&cacct->page[i][0]),
17122 + atomic_read(&cacct->page[i][1]),
17123 + atomic_read(&cacct->page[i][2]),
17124 + atomic_read(&cacct->page[i][3]),
17125 + atomic_read(&cacct->page[i][4]),
17126 + atomic_read(&cacct->page[i][5]),
17127 + atomic_read(&cacct->page[i][6]),
17128 + atomic_read(&cacct->page[i][7]));
17133 +#endif /* _VX_CACCT_PROC_H */
17134 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/context.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/context.c
17135 --- linux-2.6.29.1/kernel/vserver/context.c 1970-01-01 01:00:00.000000000 +0100
17136 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/context.c 2009-04-03 21:25:31.000000000 +0200
17139 + * linux/kernel/vserver/context.c
17141 + * Virtual Server: Context Support
17143 + * Copyright (C) 2003-2007 Herbert Pötzl
17145 + * V0.01 context helper
17146 + * V0.02 vx_ctx_kill syscall command
17147 + * V0.03 replaced context_info calls
17148 + * V0.04 redesign of struct (de)alloc
17149 + * V0.05 rlimit basic implementation
17150 + * V0.06 task_xid and info commands
17151 + * V0.07 context flags and caps
17152 + * V0.08 switch to RCU based hash
17153 + * V0.09 revert to non RCU for now
17154 + * V0.10 and back to working RCU hash
17155 + * V0.11 and back to locking again
17156 + * V0.12 referenced context store
17157 + * V0.13 separate per cpu data
17158 + * V0.14 changed vcmds to vxi arg
17159 + * V0.15 added context stat
17160 + * V0.16 have __create claim() the vxi
17161 + * V0.17 removed older and legacy stuff
17165 +#include <linux/slab.h>
17166 +#include <linux/types.h>
17167 +#include <linux/security.h>
17168 +#include <linux/pid_namespace.h>
17170 +#include <linux/vserver/context.h>
17171 +#include <linux/vserver/network.h>
17172 +#include <linux/vserver/debug.h>
17173 +#include <linux/vserver/limit.h>
17174 +#include <linux/vserver/limit_int.h>
17175 +#include <linux/vserver/space.h>
17177 +#include <linux/vs_context.h>
17178 +#include <linux/vs_limit.h>
17179 +#include <linux/vs_pid.h>
17180 +#include <linux/vserver/context_cmd.h>
17182 +#include "cvirt_init.h"
17183 +#include "cacct_init.h"
17184 +#include "limit_init.h"
17185 +#include "sched_init.h"
17188 +atomic_t vx_global_ctotal = ATOMIC_INIT(0);
17189 +atomic_t vx_global_cactive = ATOMIC_INIT(0);
17192 +/* now inactive context structures */
17194 +static struct hlist_head vx_info_inactive = HLIST_HEAD_INIT;
17196 +static spinlock_t vx_info_inactive_lock = SPIN_LOCK_UNLOCKED;
17199 +/* __alloc_vx_info()
17201 + * allocate an initialized vx_info struct
17202 + * doesn't make it visible (hash) */
17204 +static struct vx_info *__alloc_vx_info(xid_t xid)
17206 + struct vx_info *new = NULL;
17209 + vxdprintk(VXD_CBIT(xid, 0), "alloc_vx_info(%d)*", xid);
17211 + /* would this benefit from a slab cache? */
17212 + new = kmalloc(sizeof(struct vx_info), GFP_KERNEL);
17216 + memset(new, 0, sizeof(struct vx_info));
17218 + new->ptr_pc = alloc_percpu(struct _vx_info_pc);
17219 + if (!new->ptr_pc)
17222 + new->vx_id = xid;
17223 + INIT_HLIST_NODE(&new->vx_hlist);
17224 + atomic_set(&new->vx_usecnt, 0);
17225 + atomic_set(&new->vx_tasks, 0);
17226 + new->vx_parent = NULL;
17227 + new->vx_state = 0;
17228 + init_waitqueue_head(&new->vx_wait);
17230 + /* prepare reaper */
17231 + get_task_struct(init_pid_ns.child_reaper);
17232 + new->vx_reaper = init_pid_ns.child_reaper;
17233 + new->vx_badness_bias = 0;
17235 + /* rest of init goes here */
17236 + vx_info_init_limit(&new->limit);
17237 + vx_info_init_sched(&new->sched);
17238 + vx_info_init_cvirt(&new->cvirt);
17239 + vx_info_init_cacct(&new->cacct);
17241 + /* per cpu data structures */
17242 + for_each_possible_cpu(cpu) {
17243 + vx_info_init_sched_pc(
17244 + &vx_per_cpu(new, sched_pc, cpu), cpu);
17245 + vx_info_init_cvirt_pc(
17246 + &vx_per_cpu(new, cvirt_pc, cpu), cpu);
17249 + new->vx_flags = VXF_INIT_SET;
17250 + cap_set_init_eff(new->vx_bcaps);
17251 + new->vx_ccaps = 0;
17252 + // new->vx_cap_bset = current->cap_bset;
17254 + new->reboot_cmd = 0;
17255 + new->exit_code = 0;
17257 + vxdprintk(VXD_CBIT(xid, 0),
17258 + "alloc_vx_info(%d) = %p", xid, new);
17259 + vxh_alloc_vx_info(new);
17260 + atomic_inc(&vx_global_ctotal);
17269 +/* __dealloc_vx_info()
17271 + * final disposal of vx_info */
17273 +static void __dealloc_vx_info(struct vx_info *vxi)
17275 + struct vx_info_save vxis;
17278 + vxdprintk(VXD_CBIT(xid, 0),
17279 + "dealloc_vx_info(%p)", vxi);
17280 + vxh_dealloc_vx_info(vxi);
17282 +#ifdef CONFIG_VSERVER_WARN
17283 + enter_vx_info(vxi, &vxis);
17284 + vx_info_exit_limit(&vxi->limit);
17285 + vx_info_exit_sched(&vxi->sched);
17286 + vx_info_exit_cvirt(&vxi->cvirt);
17287 + vx_info_exit_cacct(&vxi->cacct);
17289 + for_each_possible_cpu(cpu) {
17290 + vx_info_exit_sched_pc(
17291 + &vx_per_cpu(vxi, sched_pc, cpu), cpu);
17292 + vx_info_exit_cvirt_pc(
17293 + &vx_per_cpu(vxi, cvirt_pc, cpu), cpu);
17295 + leave_vx_info(&vxis);
17299 + vxi->vx_state |= VXS_RELEASED;
17302 + free_percpu(vxi->ptr_pc);
17305 + atomic_dec(&vx_global_ctotal);
17308 +static void __shutdown_vx_info(struct vx_info *vxi)
17310 + struct nsproxy *nsproxy;
17311 + struct fs_struct *fs;
17316 + vxi->vx_state |= VXS_SHUTDOWN;
17317 + vs_state_change(vxi, VSC_SHUTDOWN);
17319 + for (index = 0; index < VX_SPACES; index++) {
17320 + nsproxy = xchg(&vxi->vx_nsproxy[index], NULL);
17322 + put_nsproxy(nsproxy);
17324 + fs = xchg(&vxi->vx_fs[index], NULL);
17326 + put_fs_struct(fs);
17330 +/* exported stuff */
17332 +void free_vx_info(struct vx_info *vxi)
17334 + unsigned long flags;
17337 + /* check for reference counts first */
17338 + BUG_ON(atomic_read(&vxi->vx_usecnt));
17339 + BUG_ON(atomic_read(&vxi->vx_tasks));
17341 + /* context must not be hashed */
17342 + BUG_ON(vx_info_state(vxi, VXS_HASHED));
17344 + /* context shutdown is mandatory */
17345 + BUG_ON(!vx_info_state(vxi, VXS_SHUTDOWN));
17347 + /* nsproxy and fs check */
17348 + for (index = 0; index < VX_SPACES; index++) {
17349 + BUG_ON(vxi->vx_nsproxy[index]);
17350 + BUG_ON(vxi->vx_fs[index]);
17353 + spin_lock_irqsave(&vx_info_inactive_lock, flags);
17354 + hlist_del(&vxi->vx_hlist);
17355 + spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17357 + __dealloc_vx_info(vxi);
17361 +/* hash table for vx_info hash */
17363 +#define VX_HASH_SIZE 13
17365 +static struct hlist_head vx_info_hash[VX_HASH_SIZE] =
17366 + { [0 ... VX_HASH_SIZE-1] = HLIST_HEAD_INIT };
17368 +static spinlock_t vx_info_hash_lock = SPIN_LOCK_UNLOCKED;
17371 +static inline unsigned int __hashval(xid_t xid)
17373 + return (xid % VX_HASH_SIZE);
17378 +/* __hash_vx_info()
17380 + * add the vxi to the global hash table
17381 + * requires the hash_lock to be held */
17383 +static inline void __hash_vx_info(struct vx_info *vxi)
17385 + struct hlist_head *head;
17387 + vxd_assert_lock(&vx_info_hash_lock);
17388 + vxdprintk(VXD_CBIT(xid, 4),
17389 + "__hash_vx_info: %p[#%d]", vxi, vxi->vx_id);
17390 + vxh_hash_vx_info(vxi);
17392 + /* context must not be hashed */
17393 + BUG_ON(vx_info_state(vxi, VXS_HASHED));
17395 + vxi->vx_state |= VXS_HASHED;
17396 + head = &vx_info_hash[__hashval(vxi->vx_id)];
17397 + hlist_add_head(&vxi->vx_hlist, head);
17398 + atomic_inc(&vx_global_cactive);
17401 +/* __unhash_vx_info()
17403 + * remove the vxi from the global hash table
17404 + * requires the hash_lock to be held */
17406 +static inline void __unhash_vx_info(struct vx_info *vxi)
17408 + unsigned long flags;
17410 + vxd_assert_lock(&vx_info_hash_lock);
17411 + vxdprintk(VXD_CBIT(xid, 4),
17412 + "__unhash_vx_info: %p[#%d.%d.%d]", vxi, vxi->vx_id,
17413 + atomic_read(&vxi->vx_usecnt), atomic_read(&vxi->vx_tasks));
17414 + vxh_unhash_vx_info(vxi);
17416 + /* context must be hashed */
17417 + BUG_ON(!vx_info_state(vxi, VXS_HASHED));
17418 + /* but without tasks */
17419 + BUG_ON(atomic_read(&vxi->vx_tasks));
17421 + vxi->vx_state &= ~VXS_HASHED;
17422 + hlist_del_init(&vxi->vx_hlist);
17423 + spin_lock_irqsave(&vx_info_inactive_lock, flags);
17424 + hlist_add_head(&vxi->vx_hlist, &vx_info_inactive);
17425 + spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17426 + atomic_dec(&vx_global_cactive);
17430 +/* __lookup_vx_info()
17432 + * requires the hash_lock to be held
17433 + * doesn't increment the vx_refcnt */
17435 +static inline struct vx_info *__lookup_vx_info(xid_t xid)
17437 + struct hlist_head *head = &vx_info_hash[__hashval(xid)];
17438 + struct hlist_node *pos;
17439 + struct vx_info *vxi;
17441 + vxd_assert_lock(&vx_info_hash_lock);
17442 + hlist_for_each(pos, head) {
17443 + vxi = hlist_entry(pos, struct vx_info, vx_hlist);
17445 + if (vxi->vx_id == xid)
17450 + vxdprintk(VXD_CBIT(xid, 0),
17451 + "__lookup_vx_info(#%u): %p[#%u]",
17452 + xid, vxi, vxi ? vxi->vx_id : 0);
17453 + vxh_lookup_vx_info(vxi, xid);
17458 +/* __create_vx_info()
17460 + * create the requested context
17461 + * get(), claim() and hash it */
17463 +static struct vx_info *__create_vx_info(int id)
17465 + struct vx_info *new, *vxi = NULL;
17467 + vxdprintk(VXD_CBIT(xid, 1), "create_vx_info(%d)*", id);
17469 + if (!(new = __alloc_vx_info(id)))
17470 + return ERR_PTR(-ENOMEM);
17472 + /* required to make dynamic xids unique */
17473 + spin_lock(&vx_info_hash_lock);
17475 + /* static context requested */
17476 + if ((vxi = __lookup_vx_info(id))) {
17477 + vxdprintk(VXD_CBIT(xid, 0),
17478 + "create_vx_info(%d) = %p (already there)", id, vxi);
17479 + if (vx_info_flags(vxi, VXF_STATE_SETUP, 0))
17480 + vxi = ERR_PTR(-EBUSY);
17482 + vxi = ERR_PTR(-EEXIST);
17485 + /* new context */
17486 + vxdprintk(VXD_CBIT(xid, 0),
17487 + "create_vx_info(%d) = %p (new)", id, new);
17488 + claim_vx_info(new, NULL);
17489 + __hash_vx_info(get_vx_info(new));
17490 + vxi = new, new = NULL;
17493 + spin_unlock(&vx_info_hash_lock);
17494 + vxh_create_vx_info(IS_ERR(vxi) ? NULL : vxi, id);
17496 + __dealloc_vx_info(new);
17501 +/* exported stuff */
17504 +void unhash_vx_info(struct vx_info *vxi)
17506 + __shutdown_vx_info(vxi);
17507 + spin_lock(&vx_info_hash_lock);
17508 + __unhash_vx_info(vxi);
17509 + spin_unlock(&vx_info_hash_lock);
17510 + __wakeup_vx_info(vxi);
17514 +/* lookup_vx_info()
17516 + * search for a vx_info and get() it
17517 + * negative id means current */
17519 +struct vx_info *lookup_vx_info(int id)
17521 + struct vx_info *vxi = NULL;
17524 + vxi = get_vx_info(current->vx_info);
17525 + } else if (id > 1) {
17526 + spin_lock(&vx_info_hash_lock);
17527 + vxi = get_vx_info(__lookup_vx_info(id));
17528 + spin_unlock(&vx_info_hash_lock);
17533 +/* xid_is_hashed()
17535 + * verify that xid is still hashed */
17537 +int xid_is_hashed(xid_t xid)
17541 + spin_lock(&vx_info_hash_lock);
17542 + hashed = (__lookup_vx_info(xid) != NULL);
17543 + spin_unlock(&vx_info_hash_lock);
17547 +#ifdef CONFIG_PROC_FS
17551 + * get a subset of hashed xids for proc
17552 + * assumes size is at least one */
17554 +int get_xid_list(int index, unsigned int *xids, int size)
17556 + int hindex, nr_xids = 0;
17558 + /* only show current and children */
17559 + if (!vx_check(0, VS_ADMIN | VS_WATCH)) {
17562 + xids[nr_xids] = vx_current_xid();
17566 + for (hindex = 0; hindex < VX_HASH_SIZE; hindex++) {
17567 + struct hlist_head *head = &vx_info_hash[hindex];
17568 + struct hlist_node *pos;
17570 + spin_lock(&vx_info_hash_lock);
17571 + hlist_for_each(pos, head) {
17572 + struct vx_info *vxi;
17577 + vxi = hlist_entry(pos, struct vx_info, vx_hlist);
17578 + xids[nr_xids] = vxi->vx_id;
17579 + if (++nr_xids >= size) {
17580 + spin_unlock(&vx_info_hash_lock);
17584 + /* keep the lock time short */
17585 + spin_unlock(&vx_info_hash_lock);
17592 +#ifdef CONFIG_VSERVER_DEBUG
17594 +void dump_vx_info_inactive(int level)
17596 + struct hlist_node *entry, *next;
17598 + hlist_for_each_safe(entry, next, &vx_info_inactive) {
17599 + struct vx_info *vxi =
17600 + list_entry(entry, struct vx_info, vx_hlist);
17602 + dump_vx_info(vxi, level);
17609 +int vx_migrate_user(struct task_struct *p, struct vx_info *vxi)
17611 + struct user_struct *new_user, *old_user;
17616 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
17619 + new_user = alloc_uid(vxi->vx_id, p->uid);
17623 + old_user = p->user;
17624 + if (new_user != old_user) {
17625 + atomic_inc(&new_user->processes);
17626 + atomic_dec(&old_user->processes);
17627 + p->user = new_user;
17629 + free_uid(old_user);
17635 +void vx_mask_cap_bset(struct vx_info *vxi, struct task_struct *p)
17637 + // p->cap_effective &= vxi->vx_cap_bset;
17638 + p->cap_effective =
17639 + cap_intersect(p->cap_effective, vxi->cap_bset);
17640 + // p->cap_inheritable &= vxi->vx_cap_bset;
17641 + p->cap_inheritable =
17642 + cap_intersect(p->cap_inheritable, vxi->cap_bset);
17643 + // p->cap_permitted &= vxi->vx_cap_bset;
17644 + p->cap_permitted =
17645 + cap_intersect(p->cap_permitted, vxi->cap_bset);
17650 +#include <linux/file.h>
17651 +#include <linux/fdtable.h>
17653 +static int vx_openfd_task(struct task_struct *tsk)
17655 + struct files_struct *files = tsk->files;
17656 + struct fdtable *fdt;
17657 + const unsigned long *bptr;
17658 + int count, total;
17660 + /* no rcu_read_lock() because of spin_lock() */
17661 + spin_lock(&files->file_lock);
17662 + fdt = files_fdtable(files);
17663 + bptr = fdt->open_fds->fds_bits;
17664 + count = fdt->max_fds / (sizeof(unsigned long) * 8);
17665 + for (total = 0; count > 0; count--) {
17667 + total += hweight_long(*bptr);
17670 + spin_unlock(&files->file_lock);
17675 +/* for *space compatibility */
17677 +asmlinkage long sys_unshare(unsigned long);
17680 + * migrate task to new context
17681 + * gets vxi, puts old_vxi on change
17682 + * optionally unshares namespaces (hack)
17685 +int vx_migrate_task(struct task_struct *p, struct vx_info *vxi, int unshare)
17687 + struct vx_info *old_vxi;
17693 + vxdprintk(VXD_CBIT(xid, 5),
17694 + "vx_migrate_task(%p,%p[#%d.%d])", p, vxi,
17695 + vxi->vx_id, atomic_read(&vxi->vx_usecnt));
17697 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0) &&
17698 + !vx_info_flags(vxi, VXF_STATE_SETUP, 0))
17701 + if (vx_info_state(vxi, VXS_SHUTDOWN))
17704 + old_vxi = task_get_vx_info(p);
17705 + if (old_vxi == vxi)
17708 +// if (!(ret = vx_migrate_user(p, vxi))) {
17713 + openfd = vx_openfd_task(p);
17716 + atomic_dec(&old_vxi->cvirt.nr_threads);
17717 + atomic_dec(&old_vxi->cvirt.nr_running);
17718 + __rlim_dec(&old_vxi->limit, RLIMIT_NPROC);
17719 + /* FIXME: what about the struct files here? */
17720 + __rlim_sub(&old_vxi->limit, VLIMIT_OPENFD, openfd);
17721 + /* account for the executable */
17722 + __rlim_dec(&old_vxi->limit, VLIMIT_DENTRY);
17724 + atomic_inc(&vxi->cvirt.nr_threads);
17725 + atomic_inc(&vxi->cvirt.nr_running);
17726 + __rlim_inc(&vxi->limit, RLIMIT_NPROC);
17727 + /* FIXME: what about the struct files here? */
17728 + __rlim_add(&vxi->limit, VLIMIT_OPENFD, openfd);
17729 + /* account for the executable */
17730 + __rlim_inc(&vxi->limit, VLIMIT_DENTRY);
17733 + release_vx_info(old_vxi, p);
17734 + clr_vx_info(&p->vx_info);
17736 + claim_vx_info(vxi, p);
17737 + set_vx_info(&p->vx_info, vxi);
17738 + p->xid = vxi->vx_id;
17740 + vxdprintk(VXD_CBIT(xid, 5),
17741 + "moved task %p into vxi:%p[#%d]",
17742 + p, vxi, vxi->vx_id);
17744 + // vx_mask_cap_bset(vxi, p);
17747 + /* hack for *spaces to provide compatibility */
17749 + struct nsproxy *old_nsp, *new_nsp;
17751 + ret = unshare_nsproxy_namespaces(
17752 + CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER,
17757 + old_nsp = xchg(&p->nsproxy, new_nsp);
17758 + vx_set_space(vxi,
17759 + CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER, 0);
17760 + put_nsproxy(old_nsp);
17764 + put_vx_info(old_vxi);
17768 +int vx_set_reaper(struct vx_info *vxi, struct task_struct *p)
17770 + struct task_struct *old_reaper;
17775 + vxdprintk(VXD_CBIT(xid, 6),
17776 + "vx_set_reaper(%p[#%d],%p[#%d,%d])",
17777 + vxi, vxi->vx_id, p, p->xid, p->pid);
17779 + old_reaper = vxi->vx_reaper;
17780 + if (old_reaper == p)
17783 + /* set new child reaper */
17784 + get_task_struct(p);
17785 + vxi->vx_reaper = p;
17786 + put_task_struct(old_reaper);
17790 +int vx_set_init(struct vx_info *vxi, struct task_struct *p)
17795 + vxdprintk(VXD_CBIT(xid, 6),
17796 + "vx_set_init(%p[#%d],%p[#%d,%d,%d])",
17797 + vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
17799 + vxi->vx_flags &= ~VXF_STATE_INIT;
17800 + vxi->vx_initpid = p->tgid;
17804 +void vx_exit_init(struct vx_info *vxi, struct task_struct *p, int code)
17806 + vxdprintk(VXD_CBIT(xid, 6),
17807 + "vx_exit_init(%p[#%d],%p[#%d,%d,%d])",
17808 + vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
17810 + vxi->exit_code = code;
17811 + vxi->vx_initpid = 0;
17815 +void vx_set_persistent(struct vx_info *vxi)
17817 + vxdprintk(VXD_CBIT(xid, 6),
17818 + "vx_set_persistent(%p[#%d])", vxi, vxi->vx_id);
17820 + get_vx_info(vxi);
17821 + claim_vx_info(vxi, NULL);
17824 +void vx_clear_persistent(struct vx_info *vxi)
17826 + vxdprintk(VXD_CBIT(xid, 6),
17827 + "vx_clear_persistent(%p[#%d])", vxi, vxi->vx_id);
17829 + release_vx_info(vxi, NULL);
17830 + put_vx_info(vxi);
17833 +void vx_update_persistent(struct vx_info *vxi)
17835 + if (vx_info_flags(vxi, VXF_PERSISTENT, 0))
17836 + vx_set_persistent(vxi);
17838 + vx_clear_persistent(vxi);
17842 +/* task must be current or locked */
17844 +void exit_vx_info(struct task_struct *p, int code)
17846 + struct vx_info *vxi = p->vx_info;
17849 + atomic_dec(&vxi->cvirt.nr_threads);
17852 + vxi->exit_code = code;
17853 + release_vx_info(vxi, p);
17857 +void exit_vx_info_early(struct task_struct *p, int code)
17859 + struct vx_info *vxi = p->vx_info;
17862 + if (vxi->vx_initpid == p->tgid)
17863 + vx_exit_init(vxi, p, code);
17864 + if (vxi->vx_reaper == p)
17865 + vx_set_reaper(vxi, init_pid_ns.child_reaper);
17870 +/* vserver syscall commands below here */
17872 +/* taks xid and vx_info functions */
17874 +#include <asm/uaccess.h>
17877 +int vc_task_xid(uint32_t id)
17882 + struct task_struct *tsk;
17884 + read_lock(&tasklist_lock);
17885 + tsk = find_task_by_real_pid(id);
17886 + xid = (tsk) ? tsk->xid : -ESRCH;
17887 + read_unlock(&tasklist_lock);
17889 + xid = vx_current_xid();
17894 +int vc_vx_info(struct vx_info *vxi, void __user *data)
17896 + struct vcmd_vx_info_v0 vc_data;
17898 + vc_data.xid = vxi->vx_id;
17899 + vc_data.initpid = vxi->vx_initpid;
17901 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17907 +int vc_ctx_stat(struct vx_info *vxi, void __user *data)
17909 + struct vcmd_ctx_stat_v0 vc_data;
17911 + vc_data.usecnt = atomic_read(&vxi->vx_usecnt);
17912 + vc_data.tasks = atomic_read(&vxi->vx_tasks);
17914 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17920 +/* context functions */
17922 +int vc_ctx_create(uint32_t xid, void __user *data)
17924 + struct vcmd_ctx_create vc_data = { .flagword = VXF_INIT_SET };
17925 + struct vx_info *new_vxi;
17928 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
17931 + if ((xid > MAX_S_CONTEXT) || (xid < 2))
17934 + new_vxi = __create_vx_info(xid);
17935 + if (IS_ERR(new_vxi))
17936 + return PTR_ERR(new_vxi);
17938 + /* initial flags */
17939 + new_vxi->vx_flags = vc_data.flagword;
17942 + if (vs_state_change(new_vxi, VSC_STARTUP))
17945 + ret = vx_migrate_task(current, new_vxi, (!data));
17949 + /* return context id on success */
17950 + ret = new_vxi->vx_id;
17952 + /* get a reference for persistent contexts */
17953 + if ((vc_data.flagword & VXF_PERSISTENT))
17954 + vx_set_persistent(new_vxi);
17956 + release_vx_info(new_vxi, NULL);
17957 + put_vx_info(new_vxi);
17962 +int vc_ctx_migrate(struct vx_info *vxi, void __user *data)
17964 + struct vcmd_ctx_migrate vc_data = { .flagword = 0 };
17967 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
17970 + ret = vx_migrate_task(current, vxi, 0);
17973 + if (vc_data.flagword & VXM_SET_INIT)
17974 + ret = vx_set_init(vxi, current);
17977 + if (vc_data.flagword & VXM_SET_REAPER)
17978 + ret = vx_set_reaper(vxi, current);
17983 +int vc_get_cflags(struct vx_info *vxi, void __user *data)
17985 + struct vcmd_ctx_flags_v0 vc_data;
17987 + vc_data.flagword = vxi->vx_flags;
17989 + /* special STATE flag handling */
17990 + vc_data.mask = vs_mask_flags(~0ULL, vxi->vx_flags, VXF_ONE_TIME);
17992 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17997 +int vc_set_cflags(struct vx_info *vxi, void __user *data)
17999 + struct vcmd_ctx_flags_v0 vc_data;
18000 + uint64_t mask, trigger;
18002 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18005 + /* special STATE flag handling */
18006 + mask = vs_mask_mask(vc_data.mask, vxi->vx_flags, VXF_ONE_TIME);
18007 + trigger = (mask & vxi->vx_flags) ^ (mask & vc_data.flagword);
18009 + if (vxi == current->vx_info) {
18010 + /* if (trigger & VXF_STATE_SETUP)
18011 + vx_mask_cap_bset(vxi, current); */
18012 + if (trigger & VXF_STATE_INIT) {
18015 + ret = vx_set_init(vxi, current);
18018 + ret = vx_set_reaper(vxi, current);
18024 + vxi->vx_flags = vs_mask_flags(vxi->vx_flags,
18025 + vc_data.flagword, mask);
18026 + if (trigger & VXF_PERSISTENT)
18027 + vx_update_persistent(vxi);
18033 +static inline uint64_t caps_from_cap_t(kernel_cap_t c)
18035 + uint64_t v = c.cap[0] | ((uint64_t)c.cap[1] << 32);
18037 + // printk("caps_from_cap_t(%08x:%08x) = %016llx\n", c.cap[1], c.cap[0], v);
18041 +static inline kernel_cap_t cap_t_from_caps(uint64_t v)
18043 + kernel_cap_t c = __cap_empty_set;
18045 + c.cap[0] = v & 0xFFFFFFFF;
18046 + c.cap[1] = (v >> 32) & 0xFFFFFFFF;
18048 + // printk("cap_t_from_caps(%016llx) = %08x:%08x\n", v, c.cap[1], c.cap[0]);
18053 +static int do_get_caps(struct vx_info *vxi, uint64_t *bcaps, uint64_t *ccaps)
18056 + *bcaps = caps_from_cap_t(vxi->vx_bcaps);
18058 + *ccaps = vxi->vx_ccaps;
18063 +int vc_get_ccaps(struct vx_info *vxi, void __user *data)
18065 + struct vcmd_ctx_caps_v1 vc_data;
18068 + ret = do_get_caps(vxi, NULL, &vc_data.ccaps);
18071 + vc_data.cmask = ~0ULL;
18073 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18078 +static int do_set_caps(struct vx_info *vxi,
18079 + uint64_t bcaps, uint64_t bmask, uint64_t ccaps, uint64_t cmask)
18081 + uint64_t bcold = caps_from_cap_t(vxi->vx_bcaps);
18084 + printk("do_set_caps(%16llx, %16llx, %16llx, %16llx)\n",
18085 + bcaps, bmask, ccaps, cmask);
18087 + vxi->vx_bcaps = cap_t_from_caps(
18088 + vs_mask_flags(bcold, bcaps, bmask));
18089 + vxi->vx_ccaps = vs_mask_flags(vxi->vx_ccaps, ccaps, cmask);
18094 +int vc_set_ccaps(struct vx_info *vxi, void __user *data)
18096 + struct vcmd_ctx_caps_v1 vc_data;
18098 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18101 + return do_set_caps(vxi, 0, 0, vc_data.ccaps, vc_data.cmask);
18104 +int vc_get_bcaps(struct vx_info *vxi, void __user *data)
18106 + struct vcmd_bcaps vc_data;
18109 + ret = do_get_caps(vxi, &vc_data.bcaps, NULL);
18112 + vc_data.bmask = ~0ULL;
18114 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18119 +int vc_set_bcaps(struct vx_info *vxi, void __user *data)
18121 + struct vcmd_bcaps vc_data;
18123 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18126 + return do_set_caps(vxi, vc_data.bcaps, vc_data.bmask, 0, 0);
18130 +int vc_get_badness(struct vx_info *vxi, void __user *data)
18132 + struct vcmd_badness_v0 vc_data;
18134 + vc_data.bias = vxi->vx_badness_bias;
18136 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18141 +int vc_set_badness(struct vx_info *vxi, void __user *data)
18143 + struct vcmd_badness_v0 vc_data;
18145 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18148 + vxi->vx_badness_bias = vc_data.bias;
18152 +#include <linux/module.h>
18154 +EXPORT_SYMBOL_GPL(free_vx_info);
18156 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cvirt.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt.c
18157 --- linux-2.6.29.1/kernel/vserver/cvirt.c 1970-01-01 01:00:00.000000000 +0100
18158 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt.c 2009-02-22 22:54:26.000000000 +0100
18161 + * linux/kernel/vserver/cvirt.c
18163 + * Virtual Server: Context Virtualization
18165 + * Copyright (C) 2004-2007 Herbert Pötzl
18167 + * V0.01 broken out from limit.c
18168 + * V0.02 added utsname stuff
18169 + * V0.03 changed vcmds to vxi arg
18173 +#include <linux/types.h>
18174 +#include <linux/utsname.h>
18175 +#include <linux/vs_cvirt.h>
18176 +#include <linux/vserver/switch.h>
18177 +#include <linux/vserver/cvirt_cmd.h>
18179 +#include <asm/uaccess.h>
18182 +void vx_vsi_uptime(struct timespec *uptime, struct timespec *idle)
18184 + struct vx_info *vxi = current->vx_info;
18186 + set_normalized_timespec(uptime,
18187 + uptime->tv_sec - vxi->cvirt.bias_uptime.tv_sec,
18188 + uptime->tv_nsec - vxi->cvirt.bias_uptime.tv_nsec);
18191 + set_normalized_timespec(idle,
18192 + idle->tv_sec - vxi->cvirt.bias_idle.tv_sec,
18193 + idle->tv_nsec - vxi->cvirt.bias_idle.tv_nsec);
18197 +uint64_t vx_idle_jiffies(void)
18199 + return init_task.utime + init_task.stime;
18204 +static inline uint32_t __update_loadavg(uint32_t load,
18205 + int wsize, int delta, int n)
18207 + unsigned long long calc, prev;
18209 + /* just set it to n */
18210 + if (unlikely(delta >= wsize))
18211 + return (n << FSHIFT);
18213 + calc = delta * n;
18215 + prev = (wsize - delta);
18218 + do_div(calc, wsize);
18223 +void vx_update_load(struct vx_info *vxi)
18225 + uint32_t now, last, delta;
18226 + unsigned int nr_running, nr_uninterruptible;
18227 + unsigned int total;
18228 + unsigned long flags;
18230 + spin_lock_irqsave(&vxi->cvirt.load_lock, flags);
18233 + last = vxi->cvirt.load_last;
18234 + delta = now - last;
18236 + if (delta < 5*HZ)
18239 + nr_running = atomic_read(&vxi->cvirt.nr_running);
18240 + nr_uninterruptible = atomic_read(&vxi->cvirt.nr_uninterruptible);
18241 + total = nr_running + nr_uninterruptible;
18243 + vxi->cvirt.load[0] = __update_loadavg(vxi->cvirt.load[0],
18244 + 60*HZ, delta, total);
18245 + vxi->cvirt.load[1] = __update_loadavg(vxi->cvirt.load[1],
18246 + 5*60*HZ, delta, total);
18247 + vxi->cvirt.load[2] = __update_loadavg(vxi->cvirt.load[2],
18248 + 15*60*HZ, delta, total);
18250 + vxi->cvirt.load_last = now;
18252 + atomic_inc(&vxi->cvirt.load_updates);
18253 + spin_unlock_irqrestore(&vxi->cvirt.load_lock, flags);
18258 + * Commands to do_syslog:
18260 + * 0 -- Close the log. Currently a NOP.
18261 + * 1 -- Open the log. Currently a NOP.
18262 + * 2 -- Read from the log.
18263 + * 3 -- Read all messages remaining in the ring buffer.
18264 + * 4 -- Read and clear all messages remaining in the ring buffer
18265 + * 5 -- Clear ring buffer.
18266 + * 6 -- Disable printk's to console
18267 + * 7 -- Enable printk's to console
18268 + * 8 -- Set level of messages printed to console
18269 + * 9 -- Return number of unread characters in the log buffer
18270 + * 10 -- Return size of the log buffer
18272 +int vx_do_syslog(int type, char __user *buf, int len)
18275 + int do_clear = 0;
18276 + struct vx_info *vxi = current->vx_info;
18277 + struct _vx_syslog *log;
18281 + log = &vxi->cvirt.syslog;
18284 + case 0: /* Close log */
18285 + case 1: /* Open log */
18287 + case 2: /* Read from log */
18288 + error = wait_event_interruptible(log->log_wait,
18289 + (log->log_start - log->log_end));
18292 + spin_lock_irq(&log->logbuf_lock);
18293 + spin_unlock_irq(&log->logbuf_lock);
18295 + case 4: /* Read/clear last kernel messages */
18297 + /* fall through */
18298 + case 3: /* Read last kernel messages */
18301 + case 5: /* Clear ring buffer */
18304 + case 6: /* Disable logging to console */
18305 + case 7: /* Enable logging to console */
18306 + case 8: /* Set level of messages printed to console */
18309 + case 9: /* Number of chars in the log buffer */
18311 + case 10: /* Size of the log buffer */
18321 +/* virtual host info names */
18323 +static char *vx_vhi_name(struct vx_info *vxi, int id)
18325 + struct nsproxy *nsproxy;
18326 + struct uts_namespace *uts;
18328 + if (id == VHIN_CONTEXT)
18329 + return vxi->vx_name;
18331 + nsproxy = vxi->vx_nsproxy[0];
18335 + uts = nsproxy->uts_ns;
18340 + case VHIN_SYSNAME:
18341 + return uts->name.sysname;
18342 + case VHIN_NODENAME:
18343 + return uts->name.nodename;
18344 + case VHIN_RELEASE:
18345 + return uts->name.release;
18346 + case VHIN_VERSION:
18347 + return uts->name.version;
18348 + case VHIN_MACHINE:
18349 + return uts->name.machine;
18350 + case VHIN_DOMAINNAME:
18351 + return uts->name.domainname;
18358 +int vc_set_vhi_name(struct vx_info *vxi, void __user *data)
18360 + struct vcmd_vhi_name_v0 vc_data;
18363 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18366 + name = vx_vhi_name(vxi, vc_data.field);
18370 + memcpy(name, vc_data.name, 65);
18374 +int vc_get_vhi_name(struct vx_info *vxi, void __user *data)
18376 + struct vcmd_vhi_name_v0 vc_data;
18379 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18382 + name = vx_vhi_name(vxi, vc_data.field);
18386 + memcpy(vc_data.name, name, 65);
18387 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18393 +int vc_virt_stat(struct vx_info *vxi, void __user *data)
18395 + struct vcmd_virt_stat_v0 vc_data;
18396 + struct _vx_cvirt *cvirt = &vxi->cvirt;
18397 + struct timespec uptime;
18399 + do_posix_clock_monotonic_gettime(&uptime);
18400 + set_normalized_timespec(&uptime,
18401 + uptime.tv_sec - cvirt->bias_uptime.tv_sec,
18402 + uptime.tv_nsec - cvirt->bias_uptime.tv_nsec);
18404 + vc_data.offset = timeval_to_ns(&cvirt->bias_tv);
18405 + vc_data.uptime = timespec_to_ns(&uptime);
18406 + vc_data.nr_threads = atomic_read(&cvirt->nr_threads);
18407 + vc_data.nr_running = atomic_read(&cvirt->nr_running);
18408 + vc_data.nr_uninterruptible = atomic_read(&cvirt->nr_uninterruptible);
18409 + vc_data.nr_onhold = atomic_read(&cvirt->nr_onhold);
18410 + vc_data.nr_forks = atomic_read(&cvirt->total_forks);
18411 + vc_data.load[0] = cvirt->load[0];
18412 + vc_data.load[1] = cvirt->load[1];
18413 + vc_data.load[2] = cvirt->load[2];
18415 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18421 +#ifdef CONFIG_VSERVER_VTIME
18423 +/* virtualized time base */
18425 +void vx_gettimeofday(struct timeval *tv)
18427 + do_gettimeofday(tv);
18428 + if (!vx_flags(VXF_VIRT_TIME, 0))
18431 + tv->tv_sec += current->vx_info->cvirt.bias_tv.tv_sec;
18432 + tv->tv_usec += current->vx_info->cvirt.bias_tv.tv_usec;
18434 + if (tv->tv_usec >= USEC_PER_SEC) {
18436 + tv->tv_usec -= USEC_PER_SEC;
18437 + } else if (tv->tv_usec < 0) {
18439 + tv->tv_usec += USEC_PER_SEC;
18443 +int vx_settimeofday(struct timespec *ts)
18445 + struct timeval tv;
18447 + if (!vx_flags(VXF_VIRT_TIME, 0))
18448 + return do_settimeofday(ts);
18450 + do_gettimeofday(&tv);
18451 + current->vx_info->cvirt.bias_tv.tv_sec =
18452 + ts->tv_sec - tv.tv_sec;
18453 + current->vx_info->cvirt.bias_tv.tv_usec =
18454 + (ts->tv_nsec/NSEC_PER_USEC) - tv.tv_usec;
18460 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cvirt_init.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt_init.h
18461 --- linux-2.6.29.1/kernel/vserver/cvirt_init.h 1970-01-01 01:00:00.000000000 +0100
18462 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt_init.h 2009-02-22 22:54:26.000000000 +0100
18466 +extern uint64_t vx_idle_jiffies(void);
18468 +static inline void vx_info_init_cvirt(struct _vx_cvirt *cvirt)
18470 + uint64_t idle_jiffies = vx_idle_jiffies();
18471 + uint64_t nsuptime;
18473 + do_posix_clock_monotonic_gettime(&cvirt->bias_uptime);
18474 + nsuptime = (unsigned long long)cvirt->bias_uptime.tv_sec
18475 + * NSEC_PER_SEC + cvirt->bias_uptime.tv_nsec;
18476 + cvirt->bias_clock = nsec_to_clock_t(nsuptime);
18477 + cvirt->bias_tv.tv_sec = 0;
18478 + cvirt->bias_tv.tv_usec = 0;
18480 + jiffies_to_timespec(idle_jiffies, &cvirt->bias_idle);
18481 + atomic_set(&cvirt->nr_threads, 0);
18482 + atomic_set(&cvirt->nr_running, 0);
18483 + atomic_set(&cvirt->nr_uninterruptible, 0);
18484 + atomic_set(&cvirt->nr_onhold, 0);
18486 + spin_lock_init(&cvirt->load_lock);
18487 + cvirt->load_last = jiffies;
18488 + atomic_set(&cvirt->load_updates, 0);
18489 + cvirt->load[0] = 0;
18490 + cvirt->load[1] = 0;
18491 + cvirt->load[2] = 0;
18492 + atomic_set(&cvirt->total_forks, 0);
18494 + spin_lock_init(&cvirt->syslog.logbuf_lock);
18495 + init_waitqueue_head(&cvirt->syslog.log_wait);
18496 + cvirt->syslog.log_start = 0;
18497 + cvirt->syslog.log_end = 0;
18498 + cvirt->syslog.con_start = 0;
18499 + cvirt->syslog.logged_chars = 0;
18503 +void vx_info_init_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
18505 + // cvirt_pc->cpustat = { 0 };
18508 +static inline void vx_info_exit_cvirt(struct _vx_cvirt *cvirt)
18512 + vxwprintk_xid((value = atomic_read(&cvirt->nr_threads)),
18513 + "!!! cvirt: %p[nr_threads] = %d on exit.",
18515 + vxwprintk_xid((value = atomic_read(&cvirt->nr_running)),
18516 + "!!! cvirt: %p[nr_running] = %d on exit.",
18518 + vxwprintk_xid((value = atomic_read(&cvirt->nr_uninterruptible)),
18519 + "!!! cvirt: %p[nr_uninterruptible] = %d on exit.",
18521 + vxwprintk_xid((value = atomic_read(&cvirt->nr_onhold)),
18522 + "!!! cvirt: %p[nr_onhold] = %d on exit.",
18528 +void vx_info_exit_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
18533 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/cvirt_proc.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt_proc.h
18534 --- linux-2.6.29.1/kernel/vserver/cvirt_proc.h 1970-01-01 01:00:00.000000000 +0100
18535 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/cvirt_proc.h 2009-02-22 22:54:26.000000000 +0100
18537 +#ifndef _VX_CVIRT_PROC_H
18538 +#define _VX_CVIRT_PROC_H
18540 +#include <linux/nsproxy.h>
18541 +#include <linux/mnt_namespace.h>
18542 +#include <linux/ipc_namespace.h>
18543 +#include <linux/utsname.h>
18544 +#include <linux/ipc.h>
18548 +int vx_info_proc_nsproxy(struct nsproxy *nsproxy, char *buffer)
18550 + struct mnt_namespace *ns;
18551 + struct uts_namespace *uts;
18552 + struct ipc_namespace *ipc;
18553 + struct path path;
18554 + char *pstr, *root;
18560 + length += sprintf(buffer + length,
18561 + "NSProxy:\t%p [%p,%p,%p]\n",
18562 + nsproxy, nsproxy->mnt_ns,
18563 + nsproxy->uts_ns, nsproxy->ipc_ns);
18565 + ns = nsproxy->mnt_ns;
18569 + pstr = kmalloc(PATH_MAX, GFP_KERNEL);
18573 + path.mnt = ns->root;
18574 + path.dentry = ns->root->mnt_root;
18575 + root = d_path(&path, pstr, PATH_MAX - 2);
18576 + length += sprintf(buffer + length,
18577 + "Namespace:\t%p [#%u]\n"
18578 + "RootPath:\t%s\n",
18579 + ns, atomic_read(&ns->count),
18584 + uts = nsproxy->uts_ns;
18588 + length += sprintf(buffer + length,
18589 + "SysName:\t%.*s\n"
18590 + "NodeName:\t%.*s\n"
18591 + "Release:\t%.*s\n"
18592 + "Version:\t%.*s\n"
18593 + "Machine:\t%.*s\n"
18594 + "DomainName:\t%.*s\n",
18595 + __NEW_UTS_LEN, uts->name.sysname,
18596 + __NEW_UTS_LEN, uts->name.nodename,
18597 + __NEW_UTS_LEN, uts->name.release,
18598 + __NEW_UTS_LEN, uts->name.version,
18599 + __NEW_UTS_LEN, uts->name.machine,
18600 + __NEW_UTS_LEN, uts->name.domainname);
18603 + ipc = nsproxy->ipc_ns;
18607 + length += sprintf(buffer + length,
18608 + "SEMS:\t\t%d %d %d %d %d\n"
18609 + "MSG:\t\t%d %d %d\n"
18610 + "SHM:\t\t%lu %lu %d %d\n",
18611 + ipc->sem_ctls[0], ipc->sem_ctls[1],
18612 + ipc->sem_ctls[2], ipc->sem_ctls[3],
18614 + ipc->msg_ctlmax, ipc->msg_ctlmnb, ipc->msg_ctlmni,
18615 + (unsigned long)ipc->shm_ctlmax,
18616 + (unsigned long)ipc->shm_ctlall,
18617 + ipc->shm_ctlmni, ipc->shm_tot);
18624 +#include <linux/sched.h>
18626 +#define LOAD_INT(x) ((x) >> FSHIFT)
18627 +#define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1 - 1)) * 100)
18630 +int vx_info_proc_cvirt(struct _vx_cvirt *cvirt, char *buffer)
18635 + length += sprintf(buffer + length,
18636 + "BiasUptime:\t%lu.%02lu\n",
18637 + (unsigned long)cvirt->bias_uptime.tv_sec,
18638 + (cvirt->bias_uptime.tv_nsec / (NSEC_PER_SEC / 100)));
18640 + a = cvirt->load[0] + (FIXED_1 / 200);
18641 + b = cvirt->load[1] + (FIXED_1 / 200);
18642 + c = cvirt->load[2] + (FIXED_1 / 200);
18643 + length += sprintf(buffer + length,
18644 + "nr_threads:\t%d\n"
18645 + "nr_running:\t%d\n"
18646 + "nr_unintr:\t%d\n"
18647 + "nr_onhold:\t%d\n"
18648 + "load_updates:\t%d\n"
18649 + "loadavg:\t%d.%02d %d.%02d %d.%02d\n"
18650 + "total_forks:\t%d\n",
18651 + atomic_read(&cvirt->nr_threads),
18652 + atomic_read(&cvirt->nr_running),
18653 + atomic_read(&cvirt->nr_uninterruptible),
18654 + atomic_read(&cvirt->nr_onhold),
18655 + atomic_read(&cvirt->load_updates),
18656 + LOAD_INT(a), LOAD_FRAC(a),
18657 + LOAD_INT(b), LOAD_FRAC(b),
18658 + LOAD_INT(c), LOAD_FRAC(c),
18659 + atomic_read(&cvirt->total_forks));
18664 +int vx_info_proc_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc,
18665 + char *buffer, int cpu)
18671 +#endif /* _VX_CVIRT_PROC_H */
18672 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/debug.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/debug.c
18673 --- linux-2.6.29.1/kernel/vserver/debug.c 1970-01-01 01:00:00.000000000 +0100
18674 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/debug.c 2009-02-22 22:54:26.000000000 +0100
18677 + * kernel/vserver/debug.c
18679 + * Copyright (C) 2005-2007 Herbert Pötzl
18681 + * V0.01 vx_info dump support
18685 +#include <linux/module.h>
18687 +#include <linux/vserver/context.h>
18690 +void dump_vx_info(struct vx_info *vxi, int level)
18692 + printk("vx_info %p[#%d, %d.%d, %4x]\n", vxi, vxi->vx_id,
18693 + atomic_read(&vxi->vx_usecnt),
18694 + atomic_read(&vxi->vx_tasks),
18697 + __dump_vx_limit(&vxi->limit);
18698 + __dump_vx_sched(&vxi->sched);
18699 + __dump_vx_cvirt(&vxi->cvirt);
18700 + __dump_vx_cacct(&vxi->cacct);
18706 +EXPORT_SYMBOL_GPL(dump_vx_info);
18708 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/device.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/device.c
18709 --- linux-2.6.29.1/kernel/vserver/device.c 1970-01-01 01:00:00.000000000 +0100
18710 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/device.c 2009-02-22 22:54:26.000000000 +0100
18713 + * linux/kernel/vserver/device.c
18715 + * Linux-VServer: Device Support
18717 + * Copyright (C) 2006 Herbert Pötzl
18718 + * Copyright (C) 2007 Daniel Hokka Zakrisson
18720 + * V0.01 device mapping basics
18721 + * V0.02 added defaults
18725 +#include <linux/slab.h>
18726 +#include <linux/rcupdate.h>
18727 +#include <linux/fs.h>
18728 +#include <linux/namei.h>
18729 +#include <linux/hash.h>
18731 +#include <asm/errno.h>
18732 +#include <asm/uaccess.h>
18733 +#include <linux/vserver/base.h>
18734 +#include <linux/vserver/debug.h>
18735 +#include <linux/vserver/context.h>
18736 +#include <linux/vserver/device.h>
18737 +#include <linux/vserver/device_cmd.h>
18740 +#define DMAP_HASH_BITS 4
18743 +struct vs_mapping {
18745 + struct hlist_node hlist;
18746 + struct list_head list;
18748 +#define dm_hlist u.hlist
18749 +#define dm_list u.list
18752 + struct vx_dmap_target target;
18756 +static struct hlist_head dmap_main_hash[1 << DMAP_HASH_BITS];
18758 +static spinlock_t dmap_main_hash_lock = SPIN_LOCK_UNLOCKED;
18760 +static struct vx_dmap_target dmap_defaults[2] = {
18761 + { .flags = DATTR_OPEN },
18762 + { .flags = DATTR_OPEN },
18766 +struct kmem_cache *dmap_cachep __read_mostly;
18768 +int __init dmap_cache_init(void)
18770 + dmap_cachep = kmem_cache_create("dmap_cache",
18771 + sizeof(struct vs_mapping), 0,
18772 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
18776 +__initcall(dmap_cache_init);
18779 +static inline unsigned int __hashval(dev_t dev, int bits)
18781 + return hash_long((unsigned long)dev, bits);
18785 +/* __hash_mapping()
18786 + * add the mapping to the hash table
18788 +static inline void __hash_mapping(struct vx_info *vxi, struct vs_mapping *vdm)
18790 + spinlock_t *hash_lock = &dmap_main_hash_lock;
18791 + struct hlist_head *head, *hash = dmap_main_hash;
18792 + int device = vdm->device;
18794 + spin_lock(hash_lock);
18795 + vxdprintk(VXD_CBIT(misc, 8), "__hash_mapping: %p[#%d] %08x:%08x",
18796 + vxi, vxi ? vxi->vx_id : 0, device, vdm->target.target);
18798 + head = &hash[__hashval(device, DMAP_HASH_BITS)];
18799 + hlist_add_head(&vdm->dm_hlist, head);
18800 + spin_unlock(hash_lock);
18804 +static inline int __mode_to_default(umode_t mode)
18817 +/* __set_default()
18820 +static inline void __set_default(struct vx_info *vxi, umode_t mode,
18821 + struct vx_dmap_target *vdmt)
18823 + spinlock_t *hash_lock = &dmap_main_hash_lock;
18824 + spin_lock(hash_lock);
18827 + vxi->dmap.targets[__mode_to_default(mode)] = *vdmt;
18829 + dmap_defaults[__mode_to_default(mode)] = *vdmt;
18832 + spin_unlock(hash_lock);
18834 + vxdprintk(VXD_CBIT(misc, 8), "__set_default: %p[#%u] %08x %04x",
18835 + vxi, vxi ? vxi->vx_id : 0, vdmt->target, vdmt->flags);
18839 +/* __remove_default()
18840 + * remove a default
18842 +static inline int __remove_default(struct vx_info *vxi, umode_t mode)
18844 + spinlock_t *hash_lock = &dmap_main_hash_lock;
18845 + spin_lock(hash_lock);
18848 + vxi->dmap.targets[__mode_to_default(mode)].flags = 0;
18849 + else /* remove == reset */
18850 + dmap_defaults[__mode_to_default(mode)].flags = DATTR_OPEN | mode;
18852 + spin_unlock(hash_lock);
18857 +/* __find_mapping()
18858 + * find a mapping in the hash table
18860 + * caller must hold hash_lock
18862 +static inline int __find_mapping(xid_t xid, dev_t device, umode_t mode,
18863 + struct vs_mapping **local, struct vs_mapping **global)
18865 + struct hlist_head *hash = dmap_main_hash;
18866 + struct hlist_head *head = &hash[__hashval(device, DMAP_HASH_BITS)];
18867 + struct hlist_node *pos;
18868 + struct vs_mapping *vdm;
18874 + hlist_for_each(pos, head) {
18875 + vdm = hlist_entry(pos, struct vs_mapping, dm_hlist);
18877 + if ((vdm->device == device) &&
18878 + !((vdm->target.flags ^ mode) & S_IFMT)) {
18879 + if (vdm->xid == xid) {
18882 + } else if (global && vdm->xid == 0)
18887 + if (global && *global)
18894 +/* __lookup_mapping()
18895 + * find a mapping and store the result in target and flags
18897 +static inline int __lookup_mapping(struct vx_info *vxi,
18898 + dev_t device, dev_t *target, int *flags, umode_t mode)
18900 + spinlock_t *hash_lock = &dmap_main_hash_lock;
18901 + struct vs_mapping *vdm, *global;
18902 + struct vx_dmap_target *vdmt;
18904 + xid_t xid = vxi->vx_id;
18907 + spin_lock(hash_lock);
18908 + if (__find_mapping(xid, device, mode, &vdm, &global) > 0) {
18910 + vdmt = &vdm->target;
18914 + index = __mode_to_default(mode);
18915 + if (vxi && vxi->dmap.targets[index].flags) {
18917 + vdmt = &vxi->dmap.targets[index];
18918 + } else if (global) {
18920 + vdmt = &global->target;
18924 + vdmt = &dmap_defaults[index];
18928 + if (target && (vdmt->flags & DATTR_REMAP))
18929 + *target = vdmt->target;
18931 + *target = device;
18933 + *flags = vdmt->flags;
18935 + spin_unlock(hash_lock);
18941 +/* __remove_mapping()
18942 + * remove a mapping from the hash table
18944 +static inline int __remove_mapping(struct vx_info *vxi, dev_t device,
18947 + spinlock_t *hash_lock = &dmap_main_hash_lock;
18948 + struct vs_mapping *vdm = NULL;
18951 + spin_lock(hash_lock);
18953 + ret = __find_mapping((vxi ? vxi->vx_id : 0), device, mode, &vdm,
18955 + vxdprintk(VXD_CBIT(misc, 8), "__remove_mapping: %p[#%d] %08x %04x",
18956 + vxi, vxi ? vxi->vx_id : 0, device, mode);
18959 + hlist_del(&vdm->dm_hlist);
18962 + spin_unlock(hash_lock);
18964 + kmem_cache_free(dmap_cachep, vdm);
18970 +int vs_map_device(struct vx_info *vxi,
18971 + dev_t device, dev_t *target, umode_t mode)
18973 + int ret, flags = DATTR_MASK;
18977 + *target = device;
18980 + ret = __lookup_mapping(vxi, device, target, &flags, mode);
18981 + vxdprintk(VXD_CBIT(misc, 8), "vs_map_device: %08x target: %08x flags: %04x mode: %04x mapped=%d",
18982 + device, target ? *target : 0, flags, mode, ret);
18984 + return (flags & DATTR_MASK);
18989 +static int do_set_mapping(struct vx_info *vxi,
18990 + dev_t device, dev_t target, int flags, umode_t mode)
18993 + struct vs_mapping *new;
18995 + new = kmem_cache_alloc(dmap_cachep, GFP_KERNEL);
18999 + INIT_HLIST_NODE(&new->dm_hlist);
19000 + new->device = device;
19001 + new->target.target = target;
19002 + new->target.flags = flags | mode;
19003 + new->xid = (vxi ? vxi->vx_id : 0);
19005 + vxdprintk(VXD_CBIT(misc, 8), "do_set_mapping: %08x target: %08x flags: %04x", device, target, flags);
19006 + __hash_mapping(vxi, new);
19008 + struct vx_dmap_target new = {
19009 + .target = target,
19010 + .flags = flags | mode,
19012 + __set_default(vxi, mode, &new);
19018 +static int do_unset_mapping(struct vx_info *vxi,
19019 + dev_t device, dev_t target, int flags, umode_t mode)
19021 + int ret = -EINVAL;
19024 + ret = __remove_mapping(vxi, device, mode);
19028 + ret = __remove_default(vxi, mode);
19038 +static inline int __user_device(const char __user *name, dev_t *dev,
19041 + struct nameidata nd;
19048 + ret = user_lpath(name, &nd.path);
19051 + if (nd.path.dentry->d_inode) {
19052 + *dev = nd.path.dentry->d_inode->i_rdev;
19053 + *mode = nd.path.dentry->d_inode->i_mode;
19055 + path_put(&nd.path);
19059 +static inline int __mapping_mode(dev_t device, dev_t target,
19060 + umode_t device_mode, umode_t target_mode, umode_t *mode)
19063 + *mode = device_mode & S_IFMT;
19065 + *mode = target_mode & S_IFMT;
19069 + /* if both given, device and target mode have to match */
19070 + if (device && target &&
19071 + ((device_mode ^ target_mode) & S_IFMT))
19077 +static inline int do_mapping(struct vx_info *vxi, const char __user *device_path,
19078 + const char __user *target_path, int flags, int set)
19080 + dev_t device = ~0, target = ~0;
19081 + umode_t device_mode = 0, target_mode = 0, mode;
19084 + ret = __user_device(device_path, &device, &device_mode);
19087 + ret = __user_device(target_path, &target, &target_mode);
19091 + ret = __mapping_mode(device, target,
19092 + device_mode, target_mode, &mode);
19097 + return do_set_mapping(vxi, device, target,
19100 + return do_unset_mapping(vxi, device, target,
19105 +int vc_set_mapping(struct vx_info *vxi, void __user *data)
19107 + struct vcmd_set_mapping_v0 vc_data;
19109 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19112 + return do_mapping(vxi, vc_data.device, vc_data.target,
19113 + vc_data.flags, 1);
19116 +int vc_unset_mapping(struct vx_info *vxi, void __user *data)
19118 + struct vcmd_set_mapping_v0 vc_data;
19120 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19123 + return do_mapping(vxi, vc_data.device, vc_data.target,
19124 + vc_data.flags, 0);
19128 +#ifdef CONFIG_COMPAT
19130 +int vc_set_mapping_x32(struct vx_info *vxi, void __user *data)
19132 + struct vcmd_set_mapping_v0_x32 vc_data;
19134 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19137 + return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19138 + compat_ptr(vc_data.target_ptr), vc_data.flags, 1);
19141 +int vc_unset_mapping_x32(struct vx_info *vxi, void __user *data)
19143 + struct vcmd_set_mapping_v0_x32 vc_data;
19145 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19148 + return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19149 + compat_ptr(vc_data.target_ptr), vc_data.flags, 0);
19152 +#endif /* CONFIG_COMPAT */
19155 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/dlimit.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/dlimit.c
19156 --- linux-2.6.29.1/kernel/vserver/dlimit.c 1970-01-01 01:00:00.000000000 +0100
19157 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/dlimit.c 2009-02-22 22:54:26.000000000 +0100
19160 + * linux/kernel/vserver/dlimit.c
19162 + * Virtual Server: Context Disk Limits
19164 + * Copyright (C) 2004-2007 Herbert Pötzl
19166 + * V0.01 initial version
19167 + * V0.02 compat32 splitup
19171 +#include <linux/statfs.h>
19172 +#include <linux/sched.h>
19173 +#include <linux/namei.h>
19174 +#include <linux/vs_tag.h>
19175 +#include <linux/vs_dlimit.h>
19176 +#include <linux/vserver/dlimit_cmd.h>
19178 +#include <asm/uaccess.h>
19180 +/* __alloc_dl_info()
19182 + * allocate an initialized dl_info struct
19183 + * doesn't make it visible (hash) */
19185 +static struct dl_info *__alloc_dl_info(struct super_block *sb, tag_t tag)
19187 + struct dl_info *new = NULL;
19189 + vxdprintk(VXD_CBIT(dlim, 5),
19190 + "alloc_dl_info(%p,%d)*", sb, tag);
19192 + /* would this benefit from a slab cache? */
19193 + new = kmalloc(sizeof(struct dl_info), GFP_KERNEL);
19197 + memset(new, 0, sizeof(struct dl_info));
19198 + new->dl_tag = tag;
19200 + INIT_RCU_HEAD(&new->dl_rcu);
19201 + INIT_HLIST_NODE(&new->dl_hlist);
19202 + spin_lock_init(&new->dl_lock);
19203 + atomic_set(&new->dl_refcnt, 0);
19204 + atomic_set(&new->dl_usecnt, 0);
19206 + /* rest of init goes here */
19208 + vxdprintk(VXD_CBIT(dlim, 4),
19209 + "alloc_dl_info(%p,%d) = %p", sb, tag, new);
19213 +/* __dealloc_dl_info()
19215 + * final disposal of dl_info */
19217 +static void __dealloc_dl_info(struct dl_info *dli)
19219 + vxdprintk(VXD_CBIT(dlim, 4),
19220 + "dealloc_dl_info(%p)", dli);
19222 + dli->dl_hlist.next = LIST_POISON1;
19223 + dli->dl_tag = -1;
19226 + BUG_ON(atomic_read(&dli->dl_usecnt));
19227 + BUG_ON(atomic_read(&dli->dl_refcnt));
19233 +/* hash table for dl_info hash */
19235 +#define DL_HASH_SIZE 13
19237 +struct hlist_head dl_info_hash[DL_HASH_SIZE];
19239 +static spinlock_t dl_info_hash_lock = SPIN_LOCK_UNLOCKED;
19242 +static inline unsigned int __hashval(struct super_block *sb, tag_t tag)
19244 + return ((tag ^ (unsigned long)sb) % DL_HASH_SIZE);
19249 +/* __hash_dl_info()
19251 + * add the dli to the global hash table
19252 + * requires the hash_lock to be held */
19254 +static inline void __hash_dl_info(struct dl_info *dli)
19256 + struct hlist_head *head;
19258 + vxdprintk(VXD_CBIT(dlim, 6),
19259 + "__hash_dl_info: %p[#%d]", dli, dli->dl_tag);
19260 + get_dl_info(dli);
19261 + head = &dl_info_hash[__hashval(dli->dl_sb, dli->dl_tag)];
19262 + hlist_add_head_rcu(&dli->dl_hlist, head);
19265 +/* __unhash_dl_info()
19267 + * remove the dli from the global hash table
19268 + * requires the hash_lock to be held */
19270 +static inline void __unhash_dl_info(struct dl_info *dli)
19272 + vxdprintk(VXD_CBIT(dlim, 6),
19273 + "__unhash_dl_info: %p[#%d]", dli, dli->dl_tag);
19274 + hlist_del_rcu(&dli->dl_hlist);
19275 + put_dl_info(dli);
19279 +/* __lookup_dl_info()
19281 + * requires the rcu_read_lock()
19282 + * doesn't increment the dl_refcnt */
19284 +static inline struct dl_info *__lookup_dl_info(struct super_block *sb, tag_t tag)
19286 + struct hlist_head *head = &dl_info_hash[__hashval(sb, tag)];
19287 + struct hlist_node *pos;
19288 + struct dl_info *dli;
19290 + hlist_for_each_entry_rcu(dli, pos, head, dl_hlist) {
19292 + if (dli->dl_tag == tag && dli->dl_sb == sb) {
19300 +struct dl_info *locate_dl_info(struct super_block *sb, tag_t tag)
19302 + struct dl_info *dli;
19305 + dli = get_dl_info(__lookup_dl_info(sb, tag));
19306 + vxdprintk(VXD_CBIT(dlim, 7),
19307 + "locate_dl_info(%p,#%d) = %p", sb, tag, dli);
19308 + rcu_read_unlock();
19312 +void rcu_free_dl_info(struct rcu_head *head)
19314 + struct dl_info *dli = container_of(head, struct dl_info, dl_rcu);
19315 + int usecnt, refcnt;
19317 + BUG_ON(!dli || !head);
19319 + usecnt = atomic_read(&dli->dl_usecnt);
19320 + BUG_ON(usecnt < 0);
19322 + refcnt = atomic_read(&dli->dl_refcnt);
19323 + BUG_ON(refcnt < 0);
19325 + vxdprintk(VXD_CBIT(dlim, 3),
19326 + "rcu_free_dl_info(%p)", dli);
19328 + __dealloc_dl_info(dli);
19330 + printk("!!! rcu didn't free\n");
19336 +static int do_addrem_dlimit(uint32_t id, const char __user *name,
19337 + uint32_t flags, int add)
19339 + struct path path;
19342 + ret = user_lpath(name, &path);
19344 + struct super_block *sb;
19345 + struct dl_info *dli;
19348 + if (!path.dentry->d_inode)
19349 + goto out_release;
19350 + if (!(sb = path.dentry->d_inode->i_sb))
19351 + goto out_release;
19354 + dli = __alloc_dl_info(sb, id);
19355 + spin_lock(&dl_info_hash_lock);
19358 + if (__lookup_dl_info(sb, id))
19360 + __hash_dl_info(dli);
19363 + spin_lock(&dl_info_hash_lock);
19364 + dli = __lookup_dl_info(sb, id);
19369 + __unhash_dl_info(dli);
19373 + spin_unlock(&dl_info_hash_lock);
19375 + __dealloc_dl_info(dli);
19382 +int vc_add_dlimit(uint32_t id, void __user *data)
19384 + struct vcmd_ctx_dlimit_base_v0 vc_data;
19386 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19389 + return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 1);
19392 +int vc_rem_dlimit(uint32_t id, void __user *data)
19394 + struct vcmd_ctx_dlimit_base_v0 vc_data;
19396 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19399 + return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 0);
19402 +#ifdef CONFIG_COMPAT
19404 +int vc_add_dlimit_x32(uint32_t id, void __user *data)
19406 + struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19408 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19411 + return do_addrem_dlimit(id,
19412 + compat_ptr(vc_data.name_ptr), vc_data.flags, 1);
19415 +int vc_rem_dlimit_x32(uint32_t id, void __user *data)
19417 + struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19419 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19422 + return do_addrem_dlimit(id,
19423 + compat_ptr(vc_data.name_ptr), vc_data.flags, 0);
19426 +#endif /* CONFIG_COMPAT */
19430 +int do_set_dlimit(uint32_t id, const char __user *name,
19431 + uint32_t space_used, uint32_t space_total,
19432 + uint32_t inodes_used, uint32_t inodes_total,
19433 + uint32_t reserved, uint32_t flags)
19435 + struct path path;
19438 + ret = user_lpath(name, &path);
19440 + struct super_block *sb;
19441 + struct dl_info *dli;
19444 + if (!path.dentry->d_inode)
19445 + goto out_release;
19446 + if (!(sb = path.dentry->d_inode->i_sb))
19447 + goto out_release;
19448 + if ((reserved != CDLIM_KEEP &&
19449 + reserved > 100) ||
19450 + (inodes_used != CDLIM_KEEP &&
19451 + inodes_used > inodes_total) ||
19452 + (space_used != CDLIM_KEEP &&
19453 + space_used > space_total))
19454 + goto out_release;
19457 + dli = locate_dl_info(sb, id);
19459 + goto out_release;
19461 + spin_lock(&dli->dl_lock);
19463 + if (inodes_used != CDLIM_KEEP)
19464 + dli->dl_inodes_used = inodes_used;
19465 + if (inodes_total != CDLIM_KEEP)
19466 + dli->dl_inodes_total = inodes_total;
19467 + if (space_used != CDLIM_KEEP) {
19468 + dli->dl_space_used = space_used;
19469 + dli->dl_space_used <<= 10;
19471 + if (space_total == CDLIM_INFINITY)
19472 + dli->dl_space_total = DLIM_INFINITY;
19473 + else if (space_total != CDLIM_KEEP) {
19474 + dli->dl_space_total = space_total;
19475 + dli->dl_space_total <<= 10;
19477 + if (reserved != CDLIM_KEEP)
19478 + dli->dl_nrlmult = (1 << 10) * (100 - reserved) / 100;
19480 + spin_unlock(&dli->dl_lock);
19482 + put_dl_info(dli);
19491 +int vc_set_dlimit(uint32_t id, void __user *data)
19493 + struct vcmd_ctx_dlimit_v0 vc_data;
19495 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19498 + return do_set_dlimit(id, vc_data.name,
19499 + vc_data.space_used, vc_data.space_total,
19500 + vc_data.inodes_used, vc_data.inodes_total,
19501 + vc_data.reserved, vc_data.flags);
19504 +#ifdef CONFIG_COMPAT
19506 +int vc_set_dlimit_x32(uint32_t id, void __user *data)
19508 + struct vcmd_ctx_dlimit_v0_x32 vc_data;
19510 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19513 + return do_set_dlimit(id, compat_ptr(vc_data.name_ptr),
19514 + vc_data.space_used, vc_data.space_total,
19515 + vc_data.inodes_used, vc_data.inodes_total,
19516 + vc_data.reserved, vc_data.flags);
19519 +#endif /* CONFIG_COMPAT */
19523 +int do_get_dlimit(uint32_t id, const char __user *name,
19524 + uint32_t *space_used, uint32_t *space_total,
19525 + uint32_t *inodes_used, uint32_t *inodes_total,
19526 + uint32_t *reserved, uint32_t *flags)
19528 + struct path path;
19531 + ret = user_lpath(name, &path);
19533 + struct super_block *sb;
19534 + struct dl_info *dli;
19537 + if (!path.dentry->d_inode)
19538 + goto out_release;
19539 + if (!(sb = path.dentry->d_inode->i_sb))
19540 + goto out_release;
19543 + dli = locate_dl_info(sb, id);
19545 + goto out_release;
19547 + spin_lock(&dli->dl_lock);
19548 + *inodes_used = dli->dl_inodes_used;
19549 + *inodes_total = dli->dl_inodes_total;
19550 + *space_used = dli->dl_space_used >> 10;
19551 + if (dli->dl_space_total == DLIM_INFINITY)
19552 + *space_total = CDLIM_INFINITY;
19554 + *space_total = dli->dl_space_total >> 10;
19556 + *reserved = 100 - ((dli->dl_nrlmult * 100 + 512) >> 10);
19557 + spin_unlock(&dli->dl_lock);
19559 + put_dl_info(dli);
19570 +int vc_get_dlimit(uint32_t id, void __user *data)
19572 + struct vcmd_ctx_dlimit_v0 vc_data;
19575 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19578 + ret = do_get_dlimit(id, vc_data.name,
19579 + &vc_data.space_used, &vc_data.space_total,
19580 + &vc_data.inodes_used, &vc_data.inodes_total,
19581 + &vc_data.reserved, &vc_data.flags);
19585 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19590 +#ifdef CONFIG_COMPAT
19592 +int vc_get_dlimit_x32(uint32_t id, void __user *data)
19594 + struct vcmd_ctx_dlimit_v0_x32 vc_data;
19597 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19600 + ret = do_get_dlimit(id, compat_ptr(vc_data.name_ptr),
19601 + &vc_data.space_used, &vc_data.space_total,
19602 + &vc_data.inodes_used, &vc_data.inodes_total,
19603 + &vc_data.reserved, &vc_data.flags);
19607 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19612 +#endif /* CONFIG_COMPAT */
19615 +void vx_vsi_statfs(struct super_block *sb, struct kstatfs *buf)
19617 + struct dl_info *dli;
19618 + __u64 blimit, bfree, bavail;
19621 + dli = locate_dl_info(sb, dx_current_tag());
19625 + spin_lock(&dli->dl_lock);
19626 + if (dli->dl_inodes_total == (unsigned long)DLIM_INFINITY)
19629 + /* reduce max inodes available to limit */
19630 + if (buf->f_files > dli->dl_inodes_total)
19631 + buf->f_files = dli->dl_inodes_total;
19633 + ifree = dli->dl_inodes_total - dli->dl_inodes_used;
19634 + /* reduce free inodes to min */
19635 + if (ifree < buf->f_ffree)
19636 + buf->f_ffree = ifree;
19639 + if (dli->dl_space_total == DLIM_INFINITY)
19642 + blimit = dli->dl_space_total >> sb->s_blocksize_bits;
19644 + if (dli->dl_space_total < dli->dl_space_used)
19647 + bfree = (dli->dl_space_total - dli->dl_space_used)
19648 + >> sb->s_blocksize_bits;
19650 + bavail = ((dli->dl_space_total >> 10) * dli->dl_nrlmult);
19651 + if (bavail < dli->dl_space_used)
19654 + bavail = (bavail - dli->dl_space_used)
19655 + >> sb->s_blocksize_bits;
19657 + /* reduce max space available to limit */
19658 + if (buf->f_blocks > blimit)
19659 + buf->f_blocks = blimit;
19661 + /* reduce free space to min */
19662 + if (bfree < buf->f_bfree)
19663 + buf->f_bfree = bfree;
19665 + /* reduce avail space to min */
19666 + if (bavail < buf->f_bavail)
19667 + buf->f_bavail = bavail;
19670 + spin_unlock(&dli->dl_lock);
19671 + put_dl_info(dli);
19676 +#include <linux/module.h>
19678 +EXPORT_SYMBOL_GPL(locate_dl_info);
19679 +EXPORT_SYMBOL_GPL(rcu_free_dl_info);
19681 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/helper.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/helper.c
19682 --- linux-2.6.29.1/kernel/vserver/helper.c 1970-01-01 01:00:00.000000000 +0100
19683 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/helper.c 2009-03-25 00:40:43.000000000 +0100
19686 + * linux/kernel/vserver/helper.c
19688 + * Virtual Context Support
19690 + * Copyright (C) 2004-2007 Herbert Pötzl
19692 + * V0.01 basic helper
19696 +#include <linux/kmod.h>
19697 +#include <linux/reboot.h>
19698 +#include <linux/vs_context.h>
19699 +#include <linux/vs_network.h>
19700 +#include <linux/vserver/signal.h>
19703 +char vshelper_path[255] = "/sbin/vshelper";
19706 +static int do_vshelper(char *name, char *argv[], char *envp[], int sync)
19710 + if ((ret = call_usermodehelper(name, argv, envp, sync))) {
19711 + printk( KERN_WARNING
19712 + "%s: (%s %s) returned %s with %d\n",
19713 + name, argv[1], argv[2],
19714 + sync ? "sync" : "async", ret);
19716 + vxdprintk(VXD_CBIT(switch, 4),
19717 + "%s: (%s %s) returned %s with %d",
19718 + name, argv[1], argv[2], sync ? "sync" : "async", ret);
19723 + * vshelper path is set via /proc/sys
19724 + * invoked by vserver sys_reboot(), with
19725 + * the following arguments
19727 + * argv [0] = vshelper_path;
19728 + * argv [1] = action: "restart", "halt", "poweroff", ...
19729 + * argv [2] = context identifier
19731 + * envp [*] = type-specific parameters
19734 +long vs_reboot_helper(struct vx_info *vxi, int cmd, void __user *arg)
19736 + char id_buf[8], cmd_buf[16];
19737 + char uid_buf[16], pid_buf[16];
19740 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
19741 + char *envp[] = {"HOME=/", "TERM=linux",
19742 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
19743 + uid_buf, pid_buf, cmd_buf, 0};
19745 + if (vx_info_state(vxi, VXS_HELPER))
19747 + vxi->vx_state |= VXS_HELPER;
19749 + snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
19751 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19752 + snprintf(uid_buf, sizeof(uid_buf)-1, "VS_UID=%d", current_uid());
19753 + snprintf(pid_buf, sizeof(pid_buf)-1, "VS_PID=%d", current->pid);
19756 + case LINUX_REBOOT_CMD_RESTART:
19757 + argv[1] = "restart";
19760 + case LINUX_REBOOT_CMD_HALT:
19761 + argv[1] = "halt";
19764 + case LINUX_REBOOT_CMD_POWER_OFF:
19765 + argv[1] = "poweroff";
19768 + case LINUX_REBOOT_CMD_SW_SUSPEND:
19769 + argv[1] = "swsusp";
19773 + vxi->vx_state &= ~VXS_HELPER;
19777 + ret = do_vshelper(vshelper_path, argv, envp, 0);
19778 + vxi->vx_state &= ~VXS_HELPER;
19779 + __wakeup_vx_info(vxi);
19780 + return (ret) ? -EPERM : 0;
19784 +long vs_reboot(unsigned int cmd, void __user *arg)
19786 + struct vx_info *vxi = current->vx_info;
19789 + vxdprintk(VXD_CBIT(misc, 5),
19790 + "vs_reboot(%p[#%d],%d)",
19791 + vxi, vxi ? vxi->vx_id : 0, cmd);
19793 + ret = vs_reboot_helper(vxi, cmd, arg);
19797 + vxi->reboot_cmd = cmd;
19798 + if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
19800 + case LINUX_REBOOT_CMD_RESTART:
19801 + case LINUX_REBOOT_CMD_HALT:
19802 + case LINUX_REBOOT_CMD_POWER_OFF:
19803 + vx_info_kill(vxi, 0, SIGKILL);
19804 + vx_info_kill(vxi, 1, SIGKILL);
19814 + * argv [0] = vshelper_path;
19815 + * argv [1] = action: "startup", "shutdown"
19816 + * argv [2] = context identifier
19818 + * envp [*] = type-specific parameters
19821 +long vs_state_change(struct vx_info *vxi, unsigned int cmd)
19823 + char id_buf[8], cmd_buf[16];
19824 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
19825 + char *envp[] = {"HOME=/", "TERM=linux",
19826 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19828 + if (!vx_info_flags(vxi, VXF_SC_HELPER, 0))
19831 + snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
19832 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19835 + case VSC_STARTUP:
19836 + argv[1] = "startup";
19838 + case VSC_SHUTDOWN:
19839 + argv[1] = "shutdown";
19845 + return do_vshelper(vshelper_path, argv, envp, 1);
19850 + * argv [0] = vshelper_path;
19851 + * argv [1] = action: "netup", "netdown"
19852 + * argv [2] = context identifier
19854 + * envp [*] = type-specific parameters
19857 +long vs_net_change(struct nx_info *nxi, unsigned int cmd)
19859 + char id_buf[8], cmd_buf[16];
19860 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
19861 + char *envp[] = {"HOME=/", "TERM=linux",
19862 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19864 + if (!nx_info_flags(nxi, NXF_SC_HELPER, 0))
19867 + snprintf(id_buf, sizeof(id_buf)-1, "%d", nxi->nx_id);
19868 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19872 + argv[1] = "netup";
19874 + case VSC_NETDOWN:
19875 + argv[1] = "netdown";
19881 + return do_vshelper(vshelper_path, argv, envp, 1);
19884 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/history.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/history.c
19885 --- linux-2.6.29.1/kernel/vserver/history.c 1970-01-01 01:00:00.000000000 +0100
19886 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/history.c 2009-02-22 22:54:26.000000000 +0100
19889 + * kernel/vserver/history.c
19891 + * Virtual Context History Backtrace
19893 + * Copyright (C) 2004-2007 Herbert Pötzl
19895 + * V0.01 basic structure
19896 + * V0.02 hash/unhash and trace
19897 + * V0.03 preemption fixes
19901 +#include <linux/module.h>
19902 +#include <asm/uaccess.h>
19904 +#include <linux/vserver/context.h>
19905 +#include <linux/vserver/debug.h>
19906 +#include <linux/vserver/debug_cmd.h>
19907 +#include <linux/vserver/history.h>
19910 +#ifdef CONFIG_VSERVER_HISTORY
19911 +#define VXH_SIZE CONFIG_VSERVER_HISTORY_SIZE
19913 +#define VXH_SIZE 64
19916 +struct _vx_history {
19917 + unsigned int counter;
19919 + struct _vx_hist_entry entry[VXH_SIZE + 1];
19923 +DEFINE_PER_CPU(struct _vx_history, vx_history_buffer);
19925 +unsigned volatile int vxh_active = 1;
19927 +static atomic_t sequence = ATOMIC_INIT(0);
19932 + * requires disabled preemption */
19934 +struct _vx_hist_entry *vxh_advance(void *loc)
19936 + unsigned int cpu = smp_processor_id();
19937 + struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
19938 + struct _vx_hist_entry *entry;
19939 + unsigned int index;
19941 + index = vxh_active ? (hist->counter++ % VXH_SIZE) : VXH_SIZE;
19942 + entry = &hist->entry[index];
19944 + entry->seq = atomic_inc_return(&sequence);
19945 + entry->loc = loc;
19949 +EXPORT_SYMBOL_GPL(vxh_advance);
19952 +#define VXH_LOC_FMTS "(#%04x,*%d):%p"
19954 +#define VXH_LOC_ARGS(e) (e)->seq, cpu, (e)->loc
19957 +#define VXH_VXI_FMTS "%p[#%d,%d.%d]"
19959 +#define VXH_VXI_ARGS(e) (e)->vxi.ptr, \
19960 + (e)->vxi.ptr ? (e)->vxi.xid : 0, \
19961 + (e)->vxi.ptr ? (e)->vxi.usecnt : 0, \
19962 + (e)->vxi.ptr ? (e)->vxi.tasks : 0
19964 +void vxh_dump_entry(struct _vx_hist_entry *e, unsigned cpu)
19966 + switch (e->type) {
19967 + case VXH_THROW_OOPS:
19968 + printk( VXH_LOC_FMTS " oops \n", VXH_LOC_ARGS(e));
19971 + case VXH_GET_VX_INFO:
19972 + case VXH_PUT_VX_INFO:
19973 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
19975 + (e->type == VXH_GET_VX_INFO) ? "get" : "put",
19976 + VXH_VXI_ARGS(e));
19979 + case VXH_INIT_VX_INFO:
19980 + case VXH_SET_VX_INFO:
19981 + case VXH_CLR_VX_INFO:
19982 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
19984 + (e->type == VXH_INIT_VX_INFO) ? "init" :
19985 + ((e->type == VXH_SET_VX_INFO) ? "set" : "clr"),
19986 + VXH_VXI_ARGS(e), e->sc.data);
19989 + case VXH_CLAIM_VX_INFO:
19990 + case VXH_RELEASE_VX_INFO:
19991 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
19993 + (e->type == VXH_CLAIM_VX_INFO) ? "claim" : "release",
19994 + VXH_VXI_ARGS(e), e->sc.data);
19997 + case VXH_ALLOC_VX_INFO:
19998 + case VXH_DEALLOC_VX_INFO:
19999 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
20001 + (e->type == VXH_ALLOC_VX_INFO) ? "alloc" : "dealloc",
20002 + VXH_VXI_ARGS(e));
20005 + case VXH_HASH_VX_INFO:
20006 + case VXH_UNHASH_VX_INFO:
20007 + printk( VXH_LOC_FMTS " __%s_vx_info " VXH_VXI_FMTS "\n",
20009 + (e->type == VXH_HASH_VX_INFO) ? "hash" : "unhash",
20010 + VXH_VXI_ARGS(e));
20013 + case VXH_LOC_VX_INFO:
20014 + case VXH_LOOKUP_VX_INFO:
20015 + case VXH_CREATE_VX_INFO:
20016 + printk( VXH_LOC_FMTS " __%s_vx_info [#%d] -> " VXH_VXI_FMTS "\n",
20018 + (e->type == VXH_CREATE_VX_INFO) ? "create" :
20019 + ((e->type == VXH_LOC_VX_INFO) ? "loc" : "lookup"),
20020 + e->ll.arg, VXH_VXI_ARGS(e));
20025 +static void __vxh_dump_history(void)
20027 + unsigned int i, cpu;
20029 + printk("History:\tSEQ: %8x\tNR_CPUS: %d\n",
20030 + atomic_read(&sequence), NR_CPUS);
20032 + for (i = 0; i < VXH_SIZE; i++) {
20033 + for_each_online_cpu(cpu) {
20034 + struct _vx_history *hist =
20035 + &per_cpu(vx_history_buffer, cpu);
20036 + unsigned int index = (hist->counter - i) % VXH_SIZE;
20037 + struct _vx_hist_entry *entry = &hist->entry[index];
20039 + vxh_dump_entry(entry, cpu);
20044 +void vxh_dump_history(void)
20048 + local_irq_enable();
20050 + local_irq_disable();
20052 + __vxh_dump_history();
20056 +/* vserver syscall commands below here */
20059 +int vc_dump_history(uint32_t id)
20062 + __vxh_dump_history();
20069 +int do_read_history(struct __user _vx_hist_entry *data,
20070 + int cpu, uint32_t *index, uint32_t *count)
20072 + int pos, ret = 0;
20073 + struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
20074 + int end = hist->counter;
20075 + int start = end - VXH_SIZE + 2;
20076 + int idx = *index;
20078 + /* special case: get current pos */
20084 + /* have we lost some data? */
20088 + for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
20089 + struct _vx_hist_entry *entry =
20090 + &hist->entry[idx % VXH_SIZE];
20092 + /* send entry to userspace */
20093 + ret = copy_to_user(&data[pos], entry, sizeof(*entry));
20097 + /* save new index and count */
20100 + return ret ? ret : (*index < end);
20103 +int vc_read_history(uint32_t id, void __user *data)
20105 + struct vcmd_read_history_v0 vc_data;
20108 + if (id >= NR_CPUS)
20111 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20114 + ret = do_read_history((struct __user _vx_hist_entry *)vc_data.data,
20115 + id, &vc_data.index, &vc_data.count);
20117 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20122 +#ifdef CONFIG_COMPAT
20124 +int vc_read_history_x32(uint32_t id, void __user *data)
20126 + struct vcmd_read_history_v0_x32 vc_data;
20129 + if (id >= NR_CPUS)
20132 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20135 + ret = do_read_history((struct __user _vx_hist_entry *)
20136 + compat_ptr(vc_data.data_ptr),
20137 + id, &vc_data.index, &vc_data.count);
20139 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20144 +#endif /* CONFIG_COMPAT */
20146 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/inet.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/inet.c
20147 --- linux-2.6.29.1/kernel/vserver/inet.c 1970-01-01 01:00:00.000000000 +0100
20148 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/inet.c 2009-02-22 22:54:26.000000000 +0100
20151 +#include <linux/in.h>
20152 +#include <linux/inetdevice.h>
20153 +#include <linux/vs_inet.h>
20154 +#include <linux/vs_inet6.h>
20155 +#include <linux/vserver/debug.h>
20156 +#include <net/route.h>
20157 +#include <net/addrconf.h>
20160 +int nx_v4_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20164 + if (!nxi1 || !nxi2 || nxi1 == nxi2)
20167 + struct nx_addr_v4 *ptr;
20169 + for (ptr = &nxi1->v4; ptr; ptr = ptr->next) {
20170 + if (v4_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20177 + vxdprintk(VXD_CBIT(net, 2),
20178 + "nx_v4_addr_conflict(%p,%p): %d",
20179 + nxi1, nxi2, ret);
20185 +#ifdef CONFIG_IPV6
20187 +int nx_v6_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20191 + if (!nxi1 || !nxi2 || nxi1 == nxi2)
20194 + struct nx_addr_v6 *ptr;
20196 + for (ptr = &nxi1->v6; ptr; ptr = ptr->next) {
20197 + if (v6_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20204 + vxdprintk(VXD_CBIT(net, 2),
20205 + "nx_v6_addr_conflict(%p,%p): %d",
20206 + nxi1, nxi2, ret);
20213 +int v4_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20215 + struct in_device *in_dev;
20216 + struct in_ifaddr **ifap;
20217 + struct in_ifaddr *ifa;
20222 + in_dev = in_dev_get(dev);
20226 + for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
20227 + ifap = &ifa->ifa_next) {
20228 + if (v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW)) {
20233 + in_dev_put(in_dev);
20239 +#ifdef CONFIG_IPV6
20241 +int v6_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20243 + struct inet6_dev *in_dev;
20244 + struct inet6_ifaddr **ifap;
20245 + struct inet6_ifaddr *ifa;
20250 + in_dev = in6_dev_get(dev);
20254 + for (ifap = &in_dev->addr_list; (ifa = *ifap) != NULL;
20255 + ifap = &ifa->if_next) {
20256 + if (v6_addr_in_nx_info(nxi, &ifa->addr, -1)) {
20261 + in6_dev_put(in_dev);
20268 +int dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20274 + if (nxi->v4.type && v4_dev_in_nx_info(dev, nxi))
20276 +#ifdef CONFIG_IPV6
20278 + if (nxi->v6.type && v6_dev_in_nx_info(dev, nxi))
20283 + vxdprintk(VXD_CBIT(net, 3),
20284 + "dev_in_nx_info(%p,%p[#%d]) = %d",
20285 + dev, nxi, nxi ? nxi->nx_id : 0, ret);
20289 +int ip_v4_find_src(struct net *net, struct nx_info *nxi,
20290 + struct rtable **rp, struct flowi *fl)
20295 + /* FIXME: handle lback only case */
20296 + if (!NX_IPV4(nxi))
20299 + vxdprintk(VXD_CBIT(net, 4),
20300 + "ip_v4_find_src(%p[#%u]) " NIPQUAD_FMT " -> " NIPQUAD_FMT,
20301 + nxi, nxi ? nxi->nx_id : 0,
20302 + NIPQUAD(fl->fl4_src), NIPQUAD(fl->fl4_dst));
20304 + /* single IP is unconditional */
20305 + if (nx_info_flags(nxi, NXF_SINGLE_IP, 0) &&
20306 + (fl->fl4_src == INADDR_ANY))
20307 + fl->fl4_src = nxi->v4.ip[0].s_addr;
20309 + if (fl->fl4_src == INADDR_ANY) {
20310 + struct nx_addr_v4 *ptr;
20311 + __be32 found = 0;
20314 + err = __ip_route_output_key(net, rp, fl);
20316 + found = (*rp)->rt_src;
20318 + vxdprintk(VXD_CBIT(net, 4),
20319 + "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20320 + nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(found));
20321 + if (v4_addr_in_nx_info(nxi, found, NXA_MASK_BIND))
20325 + for (ptr = &nxi->v4; ptr; ptr = ptr->next) {
20326 + __be32 primary = ptr->ip[0].s_addr;
20327 + __be32 mask = ptr->mask.s_addr;
20328 + __be32 neta = primary & mask;
20330 + vxdprintk(VXD_CBIT(net, 4), "ip_v4_find_src(%p[#%u]) chk: "
20331 + NIPQUAD_FMT "/" NIPQUAD_FMT "/" NIPQUAD_FMT,
20332 + nxi, nxi ? nxi->nx_id : 0, NIPQUAD(primary),
20333 + NIPQUAD(mask), NIPQUAD(neta));
20334 + if ((found & mask) != neta)
20337 + fl->fl4_src = primary;
20338 + err = __ip_route_output_key(net, rp, fl);
20339 + vxdprintk(VXD_CBIT(net, 4),
20340 + "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20341 + nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(primary));
20343 + found = (*rp)->rt_src;
20345 + if (found == primary)
20349 + /* still no source ip? */
20350 + found = ipv4_is_loopback(fl->fl4_dst)
20351 + ? IPI_LOOPBACK : nxi->v4.ip[0].s_addr;
20353 + /* assign src ip to flow */
20354 + fl->fl4_src = found;
20357 + if (!v4_addr_in_nx_info(nxi, fl->fl4_src, NXA_MASK_BIND))
20361 + if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0)) {
20362 + if (ipv4_is_loopback(fl->fl4_dst))
20363 + fl->fl4_dst = nxi->v4_lback.s_addr;
20364 + if (ipv4_is_loopback(fl->fl4_src))
20365 + fl->fl4_src = nxi->v4_lback.s_addr;
20366 + } else if (ipv4_is_loopback(fl->fl4_dst) &&
20367 + !nx_info_flags(nxi, NXF_LBACK_ALLOW, 0))
20373 +EXPORT_SYMBOL_GPL(ip_v4_find_src);
20375 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/init.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/init.c
20376 --- linux-2.6.29.1/kernel/vserver/init.c 1970-01-01 01:00:00.000000000 +0100
20377 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/init.c 2009-02-22 22:54:26.000000000 +0100
20380 + * linux/kernel/init.c
20382 + * Virtual Server Init
20384 + * Copyright (C) 2004-2007 Herbert Pötzl
20386 + * V0.01 basic structure
20390 +#include <linux/init.h>
20392 +int vserver_register_sysctl(void);
20393 +void vserver_unregister_sysctl(void);
20396 +static int __init init_vserver(void)
20400 +#ifdef CONFIG_VSERVER_DEBUG
20401 + vserver_register_sysctl();
20407 +static void __exit exit_vserver(void)
20410 +#ifdef CONFIG_VSERVER_DEBUG
20411 + vserver_unregister_sysctl();
20416 +/* FIXME: GFP_ZONETYPES gone
20417 +long vx_slab[GFP_ZONETYPES]; */
20421 +module_init(init_vserver);
20422 +module_exit(exit_vserver);
20424 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/inode.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/inode.c
20425 --- linux-2.6.29.1/kernel/vserver/inode.c 1970-01-01 01:00:00.000000000 +0100
20426 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/inode.c 2009-02-22 22:54:26.000000000 +0100
20429 + * linux/kernel/vserver/inode.c
20431 + * Virtual Server: File System Support
20433 + * Copyright (C) 2004-2007 Herbert Pötzl
20435 + * V0.01 separated from vcontext V0.05
20436 + * V0.02 moved to tag (instead of xid)
20440 +#include <linux/tty.h>
20441 +#include <linux/proc_fs.h>
20442 +#include <linux/devpts_fs.h>
20443 +#include <linux/fs.h>
20444 +#include <linux/file.h>
20445 +#include <linux/mount.h>
20446 +#include <linux/parser.h>
20447 +#include <linux/namei.h>
20448 +#include <linux/vserver/inode.h>
20449 +#include <linux/vserver/inode_cmd.h>
20450 +#include <linux/vs_base.h>
20451 +#include <linux/vs_tag.h>
20453 +#include <asm/uaccess.h>
20456 +static int __vc_get_iattr(struct inode *in, uint32_t *tag, uint32_t *flags, uint32_t *mask)
20458 + struct proc_dir_entry *entry;
20460 + if (!in || !in->i_sb)
20463 + *flags = IATTR_TAG
20464 + | (IS_BARRIER(in) ? IATTR_BARRIER : 0)
20465 + | (IS_IXUNLINK(in) ? IATTR_IXUNLINK : 0)
20466 + | (IS_IMMUTABLE(in) ? IATTR_IMMUTABLE : 0);
20467 + *mask = IATTR_IXUNLINK | IATTR_IMMUTABLE;
20469 + if (S_ISDIR(in->i_mode))
20470 + *mask |= IATTR_BARRIER;
20472 + if (IS_TAGGED(in)) {
20473 + *tag = in->i_tag;
20474 + *mask |= IATTR_TAG;
20477 + switch (in->i_sb->s_magic) {
20478 + case PROC_SUPER_MAGIC:
20479 + entry = PROC_I(in)->pde;
20481 + /* check for specific inodes? */
20483 + *mask |= IATTR_FLAGS;
20485 + *flags |= (entry->vx_flags & IATTR_FLAGS);
20487 + *flags |= (PROC_I(in)->vx_flags & IATTR_FLAGS);
20490 + case DEVPTS_SUPER_MAGIC:
20491 + *tag = in->i_tag;
20492 + *mask |= IATTR_TAG;
20501 +int vc_get_iattr(void __user *data)
20503 + struct path path;
20504 + struct vcmd_ctx_iattr_v1 vc_data = { .tag = -1 };
20507 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20510 + ret = user_lpath(vc_data.name, &path);
20512 + ret = __vc_get_iattr(path.dentry->d_inode,
20513 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
20519 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20524 +#ifdef CONFIG_COMPAT
20526 +int vc_get_iattr_x32(void __user *data)
20528 + struct path path;
20529 + struct vcmd_ctx_iattr_v1_x32 vc_data = { .tag = -1 };
20532 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20535 + ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
20537 + ret = __vc_get_iattr(path.dentry->d_inode,
20538 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
20544 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20549 +#endif /* CONFIG_COMPAT */
20552 +int vc_fget_iattr(uint32_t fd, void __user *data)
20554 + struct file *filp;
20555 + struct vcmd_ctx_fiattr_v0 vc_data = { .tag = -1 };
20558 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20562 + if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
20565 + ret = __vc_get_iattr(filp->f_dentry->d_inode,
20566 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
20570 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20576 +static int __vc_set_iattr(struct dentry *de, uint32_t *tag, uint32_t *flags, uint32_t *mask)
20578 + struct inode *in = de->d_inode;
20579 + int error = 0, is_proc = 0, has_tag = 0;
20580 + struct iattr attr = { 0 };
20582 + if (!in || !in->i_sb)
20585 + is_proc = (in->i_sb->s_magic == PROC_SUPER_MAGIC);
20586 + if ((*mask & IATTR_FLAGS) && !is_proc)
20589 + has_tag = IS_TAGGED(in) ||
20590 + (in->i_sb->s_magic == DEVPTS_SUPER_MAGIC);
20591 + if ((*mask & IATTR_TAG) && !has_tag)
20594 + mutex_lock(&in->i_mutex);
20595 + if (*mask & IATTR_TAG) {
20596 + attr.ia_tag = *tag;
20597 + attr.ia_valid |= ATTR_TAG;
20600 + if (*mask & IATTR_FLAGS) {
20601 + struct proc_dir_entry *entry = PROC_I(in)->pde;
20602 + unsigned int iflags = PROC_I(in)->vx_flags;
20604 + iflags = (iflags & ~(*mask & IATTR_FLAGS))
20605 + | (*flags & IATTR_FLAGS);
20606 + PROC_I(in)->vx_flags = iflags;
20608 + entry->vx_flags = iflags;
20611 + if (*mask & (IATTR_BARRIER | IATTR_IXUNLINK | IATTR_IMMUTABLE)) {
20612 + if (*mask & IATTR_IMMUTABLE) {
20613 + if (*flags & IATTR_IMMUTABLE)
20614 + in->i_flags |= S_IMMUTABLE;
20616 + in->i_flags &= ~S_IMMUTABLE;
20618 + if (*mask & IATTR_IXUNLINK) {
20619 + if (*flags & IATTR_IXUNLINK)
20620 + in->i_flags |= S_IXUNLINK;
20622 + in->i_flags &= ~S_IXUNLINK;
20624 + if (S_ISDIR(in->i_mode) && (*mask & IATTR_BARRIER)) {
20625 + if (*flags & IATTR_BARRIER)
20626 + in->i_vflags |= V_BARRIER;
20628 + in->i_vflags &= ~V_BARRIER;
20630 + if (in->i_op && in->i_op->sync_flags) {
20631 + error = in->i_op->sync_flags(in);
20637 + if (attr.ia_valid) {
20638 + if (in->i_op && in->i_op->setattr)
20639 + error = in->i_op->setattr(de, &attr);
20641 + error = inode_change_ok(in, &attr);
20643 + error = inode_setattr(in, &attr);
20648 + mutex_unlock(&in->i_mutex);
20652 +int vc_set_iattr(void __user *data)
20654 + struct path path;
20655 + struct vcmd_ctx_iattr_v1 vc_data;
20658 + if (!capable(CAP_LINUX_IMMUTABLE))
20660 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20663 + ret = user_lpath(vc_data.name, &path);
20665 + ret = __vc_set_iattr(path.dentry,
20666 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
20670 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20675 +#ifdef CONFIG_COMPAT
20677 +int vc_set_iattr_x32(void __user *data)
20679 + struct path path;
20680 + struct vcmd_ctx_iattr_v1_x32 vc_data;
20683 + if (!capable(CAP_LINUX_IMMUTABLE))
20685 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20688 + ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
20690 + ret = __vc_set_iattr(path.dentry,
20691 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
20695 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20700 +#endif /* CONFIG_COMPAT */
20702 +int vc_fset_iattr(uint32_t fd, void __user *data)
20704 + struct file *filp;
20705 + struct vcmd_ctx_fiattr_v0 vc_data;
20708 + if (!capable(CAP_LINUX_IMMUTABLE))
20710 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20714 + if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
20717 + ret = __vc_set_iattr(filp->f_dentry, &vc_data.tag,
20718 + &vc_data.flags, &vc_data.mask);
20722 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20728 +enum { Opt_notagcheck, Opt_tag, Opt_notag, Opt_tagid, Opt_err };
20730 +static match_table_t tokens = {
20731 + {Opt_notagcheck, "notagcheck"},
20732 +#ifdef CONFIG_PROPAGATE
20733 + {Opt_notag, "notag"},
20734 + {Opt_tag, "tag"},
20735 + {Opt_tagid, "tagid=%u"},
20741 +static void __dx_parse_remove(char *string, char *opt)
20743 + char *p = strstr(string, opt);
20747 + while (*q != '\0' && *q != ',')
20756 +int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
20757 + unsigned long *flags)
20760 + substring_t args[MAX_OPT_ARGS];
20761 + int token, option = 0;
20762 + char *s, *p, *opts;
20766 + s = kstrdup(string, GFP_KERNEL | GFP_ATOMIC);
20771 + while ((p = strsep(&opts, ",")) != NULL) {
20772 + token = match_token(p, tokens, args);
20774 + vxdprintk(VXD_CBIT(tag, 7),
20775 + "dx_parse_tag(»%s«): %d:#%d",
20776 + p, token, option);
20779 +#ifdef CONFIG_PROPAGATE
20784 + __dx_parse_remove(s, "tag");
20785 + *mnt_flags |= MNT_TAGID;
20786 + set |= MNT_TAGID;
20790 + __dx_parse_remove(s, "notag");
20791 + *mnt_flags |= MNT_NOTAG;
20792 + set |= MNT_NOTAG;
20795 + if (tag && !match_int(args, &option))
20798 + __dx_parse_remove(s, "tagid");
20799 + *mnt_flags |= MNT_TAGID;
20800 + set |= MNT_TAGID;
20803 + case Opt_notagcheck:
20805 + __dx_parse_remove(s, "notagcheck");
20806 + *flags |= MS_NOTAGCHECK;
20807 + set |= MS_NOTAGCHECK;
20812 + strcpy(string, s);
20817 +#ifdef CONFIG_PROPAGATE
20819 +void __dx_propagate_tag(struct nameidata *nd, struct inode *inode)
20821 + tag_t new_tag = 0;
20822 + struct vfsmount *mnt;
20827 + mnt = nd->path.mnt;
20831 + propagate = (mnt->mnt_flags & MNT_TAGID);
20833 + new_tag = mnt->mnt_tag;
20835 + vxdprintk(VXD_CBIT(tag, 7),
20836 + "dx_propagate_tag(%p[#%lu.%d]): %d,%d",
20837 + inode, inode->i_ino, inode->i_tag,
20838 + new_tag, (propagate) ? 1 : 0);
20841 + inode->i_tag = new_tag;
20844 +#include <linux/module.h>
20846 +EXPORT_SYMBOL_GPL(__dx_propagate_tag);
20848 +#endif /* CONFIG_PROPAGATE */
20850 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/Kconfig
20851 --- linux-2.6.29.1/kernel/vserver/Kconfig 1970-01-01 01:00:00.000000000 +0100
20852 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/Kconfig 2009-02-22 22:54:26.000000000 +0100
20855 +# Linux VServer configuration
20858 +menu "Linux VServer"
20860 +config VSERVER_AUTO_LBACK
20861 + bool "Automatically Assign Loopback IP"
20864 + Automatically assign a guest specific loopback
20865 + IP and add it to the kernel network stack on
20868 +config VSERVER_AUTO_SINGLE
20869 + bool "Automatic Single IP Special Casing"
20870 + depends on EXPERIMENTAL
20873 + This allows network contexts with a single IP to
20874 + automatically remap 0.0.0.0 bindings to that IP,
20875 + avoiding further network checks and improving
20878 + (note: such guests do not allow to change the ip
20879 + on the fly and do not show loopback addresses)
20881 +config VSERVER_COWBL
20882 + bool "Enable COW Immutable Link Breaking"
20885 + This enables the COW (Copy-On-Write) link break code.
20886 + It allows you to treat unified files like normal files
20887 + when writing to them (which will implicitely break the
20888 + link and create a copy of the unified file)
20890 +config VSERVER_VTIME
20891 + bool "Enable Virtualized Guest Time"
20892 + depends on EXPERIMENTAL
20895 + This enables per guest time offsets to allow for
20896 + adjusting the system clock individually per guest.
20897 + this adds some overhead to the time functions and
20898 + therefore should not be enabled without good reason.
20900 +config VSERVER_DEVICE
20901 + bool "Enable Guest Device Mapping"
20902 + depends on EXPERIMENTAL
20905 + This enables generic device remapping.
20907 +config VSERVER_PROC_SECURE
20908 + bool "Enable Proc Security"
20909 + depends on PROC_FS
20912 + This configures ProcFS security to initially hide
20913 + non-process entries for all contexts except the main and
20914 + spectator context (i.e. for all guests), which is a secure
20917 + (note: on 1.2x the entries were visible by default)
20919 +config VSERVER_HARDCPU
20920 + bool "Enable Hard CPU Limits"
20923 + Activate the Hard CPU Limits
20925 + This will compile in code that allows the Token Bucket
20926 + Scheduler to put processes on hold when a context's
20927 + tokens are depleted (provided that its per-context
20928 + sched_hard flag is set).
20930 + Processes belonging to that context will not be able
20931 + to consume CPU resources again until a per-context
20932 + configured minimum of tokens has been reached.
20934 +config VSERVER_IDLETIME
20935 + bool "Avoid idle CPUs by skipping Time"
20936 + depends on VSERVER_HARDCPU
20939 + This option allows the scheduler to artificially
20940 + advance time (per cpu) when otherwise the idle
20941 + task would be scheduled, thus keeping the cpu
20942 + busy and sharing the available resources among
20943 + certain contexts.
20945 +config VSERVER_IDLELIMIT
20946 + bool "Limit the IDLE task"
20947 + depends on VSERVER_HARDCPU
20950 + Limit the idle slices, so the the next context
20951 + will be scheduled as soon as possible.
20953 + This might improve interactivity and latency, but
20954 + will also marginally increase scheduling overhead.
20957 + prompt "Persistent Inode Tagging"
20958 + default TAGGING_ID24
20960 + This adds persistent context information to filesystems
20961 + mounted with the tagxid option. Tagging is a requirement
20962 + for per-context disk limits and per-context quota.
20965 +config TAGGING_NONE
20968 + do not store per-context information in inodes.
20970 +config TAGGING_UID16
20971 + bool "UID16/GID32"
20973 + reduces UID to 16 bit, but leaves GID at 32 bit.
20975 +config TAGGING_GID16
20976 + bool "UID32/GID16"
20978 + reduces GID to 16 bit, but leaves UID at 32 bit.
20980 +config TAGGING_ID24
20981 + bool "UID24/GID24"
20983 + uses the upper 8bit from UID and GID for XID tagging
20984 + which leaves 24bit for UID/GID each, which should be
20985 + more than sufficient for normal use.
20987 +config TAGGING_INTERN
20988 + bool "UID32/GID32"
20990 + this uses otherwise reserved inode fields in the on
20991 + disk representation, which limits the use to a few
20992 + filesystems (currently ext2 and ext3)
20997 + bool "Tag NFSD User Auth and Files"
21000 + Enable this if you do want the in-kernel NFS
21001 + Server to use the tagging specified above.
21002 + (will require patched clients too)
21004 +config VSERVER_PRIVACY
21005 + bool "Honor Privacy Aspects of Guests"
21008 + When enabled, most context checks will disallow
21009 + access to structures assigned to a specific context,
21010 + like ptys or loop devices.
21012 +config VSERVER_CONTEXTS
21013 + int "Maximum number of Contexts (1-65533)" if EMBEDDED
21015 + default "768" if 64BIT
21018 + This setting will optimize certain data structures
21019 + and memory allocations according to the expected
21022 + note: this is not a strict upper limit.
21024 +config VSERVER_WARN
21025 + bool "VServer Warnings"
21028 + This enables various runtime warnings, which will
21029 + notify about potential manipulation attempts or
21030 + resource shortage. It is generally considered to
21031 + be a good idea to have that enabled.
21033 +config VSERVER_DEBUG
21034 + bool "VServer Debugging Code"
21037 + Set this to yes if you want to be able to activate
21038 + debugging output at runtime. It adds a very small
21039 + overhead to all vserver related functions and
21040 + increases the kernel size by about 20k.
21042 +config VSERVER_HISTORY
21043 + bool "VServer History Tracing"
21044 + depends on VSERVER_DEBUG
21047 + Set this to yes if you want to record the history of
21048 + linux-vserver activities, so they can be replayed in
21049 + the event of a kernel panic or oops.
21051 +config VSERVER_HISTORY_SIZE
21052 + int "Per-CPU History Size (32-65536)"
21053 + depends on VSERVER_HISTORY
21057 + This allows you to specify the number of entries in
21058 + the per-CPU history buffer.
21060 +config VSERVER_MONITOR
21061 + bool "VServer Scheduling Monitor"
21062 + depends on VSERVER_DISABLED
21065 + Set this to yes if you want to record the scheduling
21066 + decisions, so that they can be relayed to userspace
21067 + for detailed analysis.
21069 +config VSERVER_MONITOR_SIZE
21070 + int "Per-CPU Monitor Queue Size (32-65536)"
21071 + depends on VSERVER_MONITOR
21075 + This allows you to specify the number of entries in
21076 + the per-CPU scheduling monitor buffer.
21078 +config VSERVER_MONITOR_SYNC
21079 + int "Per-CPU Monitor Sync Interval (0-65536)"
21080 + depends on VSERVER_MONITOR
21084 + This allows you to specify the interval in ticks
21085 + when a time sync entry is inserted.
21093 + select NAMESPACES
21099 +config VSERVER_SECURITY
21101 + depends on SECURITY
21103 + select SECURITY_CAPABILITIES
21105 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/limit.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit.c
21106 --- linux-2.6.29.1/kernel/vserver/limit.c 1970-01-01 01:00:00.000000000 +0100
21107 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit.c 2009-02-22 22:54:26.000000000 +0100
21110 + * linux/kernel/vserver/limit.c
21112 + * Virtual Server: Context Limits
21114 + * Copyright (C) 2004-2007 Herbert Pötzl
21116 + * V0.01 broken out from vcontext V0.05
21117 + * V0.02 changed vcmds to vxi arg
21121 +#include <linux/sched.h>
21122 +#include <linux/module.h>
21123 +#include <linux/vs_limit.h>
21124 +#include <linux/vserver/limit.h>
21125 +#include <linux/vserver/limit_cmd.h>
21127 +#include <asm/uaccess.h>
21130 +const char *vlimit_name[NUM_LIMITS] = {
21131 + [RLIMIT_CPU] = "CPU",
21132 + [RLIMIT_RSS] = "RSS",
21133 + [RLIMIT_NPROC] = "NPROC",
21134 + [RLIMIT_NOFILE] = "NOFILE",
21135 + [RLIMIT_MEMLOCK] = "VML",
21136 + [RLIMIT_AS] = "VM",
21137 + [RLIMIT_LOCKS] = "LOCKS",
21138 + [RLIMIT_SIGPENDING] = "SIGP",
21139 + [RLIMIT_MSGQUEUE] = "MSGQ",
21141 + [VLIMIT_NSOCK] = "NSOCK",
21142 + [VLIMIT_OPENFD] = "OPENFD",
21143 + [VLIMIT_ANON] = "ANON",
21144 + [VLIMIT_SHMEM] = "SHMEM",
21145 + [VLIMIT_DENTRY] = "DENTRY",
21148 +EXPORT_SYMBOL_GPL(vlimit_name);
21150 +#define MASK_ENTRY(x) (1 << (x))
21152 +const struct vcmd_ctx_rlimit_mask_v0 vlimit_mask = {
21155 + , /* softlimit */
21156 + MASK_ENTRY( RLIMIT_RSS ) |
21157 + MASK_ENTRY( VLIMIT_ANON ) |
21160 + MASK_ENTRY( RLIMIT_RSS ) |
21161 + MASK_ENTRY( RLIMIT_NPROC ) |
21162 + MASK_ENTRY( RLIMIT_NOFILE ) |
21163 + MASK_ENTRY( RLIMIT_MEMLOCK ) |
21164 + MASK_ENTRY( RLIMIT_AS ) |
21165 + MASK_ENTRY( RLIMIT_LOCKS ) |
21166 + MASK_ENTRY( RLIMIT_MSGQUEUE ) |
21168 + MASK_ENTRY( VLIMIT_NSOCK ) |
21169 + MASK_ENTRY( VLIMIT_OPENFD ) |
21170 + MASK_ENTRY( VLIMIT_ANON ) |
21171 + MASK_ENTRY( VLIMIT_SHMEM ) |
21172 + MASK_ENTRY( VLIMIT_DENTRY ) |
21175 + /* accounting only */
21176 +uint32_t account_mask =
21177 + MASK_ENTRY( VLIMIT_SEMARY ) |
21178 + MASK_ENTRY( VLIMIT_NSEMS ) |
21179 + MASK_ENTRY( VLIMIT_MAPPED ) |
21183 +static int is_valid_vlimit(int id)
21185 + uint32_t mask = vlimit_mask.minimum |
21186 + vlimit_mask.softlimit | vlimit_mask.maximum;
21187 + return mask & (1 << id);
21190 +static int is_accounted_vlimit(int id)
21192 + if (is_valid_vlimit(id))
21194 + return account_mask & (1 << id);
21198 +static inline uint64_t vc_get_soft(struct vx_info *vxi, int id)
21200 + rlim_t limit = __rlim_soft(&vxi->limit, id);
21201 + return VX_VLIM(limit);
21204 +static inline uint64_t vc_get_hard(struct vx_info *vxi, int id)
21206 + rlim_t limit = __rlim_hard(&vxi->limit, id);
21207 + return VX_VLIM(limit);
21210 +static int do_get_rlimit(struct vx_info *vxi, uint32_t id,
21211 + uint64_t *minimum, uint64_t *softlimit, uint64_t *maximum)
21213 + if (!is_valid_vlimit(id))
21217 + *minimum = CRLIM_UNSET;
21219 + *softlimit = vc_get_soft(vxi, id);
21221 + *maximum = vc_get_hard(vxi, id);
21225 +int vc_get_rlimit(struct vx_info *vxi, void __user *data)
21227 + struct vcmd_ctx_rlimit_v0 vc_data;
21230 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21233 + ret = do_get_rlimit(vxi, vc_data.id,
21234 + &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21238 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21243 +static int do_set_rlimit(struct vx_info *vxi, uint32_t id,
21244 + uint64_t minimum, uint64_t softlimit, uint64_t maximum)
21246 + if (!is_valid_vlimit(id))
21249 + if (maximum != CRLIM_KEEP)
21250 + __rlim_hard(&vxi->limit, id) = VX_RLIM(maximum);
21251 + if (softlimit != CRLIM_KEEP)
21252 + __rlim_soft(&vxi->limit, id) = VX_RLIM(softlimit);
21254 + /* clamp soft limit */
21255 + if (__rlim_soft(&vxi->limit, id) > __rlim_hard(&vxi->limit, id))
21256 + __rlim_soft(&vxi->limit, id) = __rlim_hard(&vxi->limit, id);
21261 +int vc_set_rlimit(struct vx_info *vxi, void __user *data)
21263 + struct vcmd_ctx_rlimit_v0 vc_data;
21265 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21268 + return do_set_rlimit(vxi, vc_data.id,
21269 + vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21272 +#ifdef CONFIG_IA32_EMULATION
21274 +int vc_set_rlimit_x32(struct vx_info *vxi, void __user *data)
21276 + struct vcmd_ctx_rlimit_v0_x32 vc_data;
21278 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21281 + return do_set_rlimit(vxi, vc_data.id,
21282 + vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21285 +int vc_get_rlimit_x32(struct vx_info *vxi, void __user *data)
21287 + struct vcmd_ctx_rlimit_v0_x32 vc_data;
21290 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21293 + ret = do_get_rlimit(vxi, vc_data.id,
21294 + &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21298 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21303 +#endif /* CONFIG_IA32_EMULATION */
21306 +int vc_get_rlimit_mask(uint32_t id, void __user *data)
21308 + if (copy_to_user(data, &vlimit_mask, sizeof(vlimit_mask)))
21314 +static inline void vx_reset_minmax(struct _vx_limit *limit)
21319 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21320 + value = __rlim_get(limit, lim);
21321 + __rlim_rmax(limit, lim) = value;
21322 + __rlim_rmin(limit, lim) = value;
21327 +int vc_reset_minmax(struct vx_info *vxi, void __user *data)
21329 + vx_reset_minmax(&vxi->limit);
21334 +int vc_rlimit_stat(struct vx_info *vxi, void __user *data)
21336 + struct vcmd_rlimit_stat_v0 vc_data;
21337 + struct _vx_limit *limit = &vxi->limit;
21340 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21344 + if (!is_accounted_vlimit(id))
21347 + vx_limit_fixup(limit, id);
21348 + vc_data.hits = atomic_read(&__rlim_lhit(limit, id));
21349 + vc_data.value = __rlim_get(limit, id);
21350 + vc_data.minimum = __rlim_rmin(limit, id);
21351 + vc_data.maximum = __rlim_rmax(limit, id);
21353 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21359 +void vx_vsi_meminfo(struct sysinfo *val)
21361 + struct vx_info *vxi = current->vx_info;
21362 + unsigned long totalram, freeram;
21365 + /* we blindly accept the max */
21366 + v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21367 + totalram = (v != RLIM_INFINITY) ? v : val->totalram;
21369 + /* total minus used equals free */
21370 + v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21371 + freeram = (v < totalram) ? totalram - v : 0;
21373 + val->totalram = totalram;
21374 + val->freeram = freeram;
21375 + val->bufferram = 0;
21376 + val->totalhigh = 0;
21377 + val->freehigh = 0;
21381 +void vx_vsi_swapinfo(struct sysinfo *val)
21383 + struct vx_info *vxi = current->vx_info;
21384 + unsigned long totalswap, freeswap;
21387 + v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21388 + if (v == RLIM_INFINITY) {
21389 + val->freeswap = val->totalswap;
21393 + /* we blindly accept the max */
21394 + w = __rlim_hard(&vxi->limit, RLIMIT_RSS);
21395 + totalswap = (w != RLIM_INFINITY) ? (w - v) : val->totalswap;
21397 + /* currently 'used' swap */
21398 + w = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21399 + w -= (w > v) ? v : w;
21401 + /* total minus used equals free */
21402 + freeswap = (w < totalswap) ? totalswap - w : 0;
21404 + val->totalswap = totalswap;
21405 + val->freeswap = freeswap;
21410 +unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm)
21412 + struct vx_info *vxi = mm->mm_vx_info;
21413 + unsigned long points;
21419 + points = vxi->vx_badness_bias;
21421 + v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21422 + w = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21423 + points += (v > w) ? (v - w) : 0;
21428 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/limit_init.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit_init.h
21429 --- linux-2.6.29.1/kernel/vserver/limit_init.h 1970-01-01 01:00:00.000000000 +0100
21430 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit_init.h 2009-02-22 22:54:26.000000000 +0100
21434 +static inline void vx_info_init_limit(struct _vx_limit *limit)
21438 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21439 + __rlim_soft(limit, lim) = RLIM_INFINITY;
21440 + __rlim_hard(limit, lim) = RLIM_INFINITY;
21441 + __rlim_set(limit, lim, 0);
21442 + atomic_set(&__rlim_lhit(limit, lim), 0);
21443 + __rlim_rmin(limit, lim) = 0;
21444 + __rlim_rmax(limit, lim) = 0;
21448 +static inline void vx_info_exit_limit(struct _vx_limit *limit)
21453 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21454 + if ((1 << lim) & VLIM_NOCHECK)
21456 + value = __rlim_get(limit, lim);
21457 + vxwprintk_xid(value,
21458 + "!!! limit: %p[%s,%d] = %ld on exit.",
21459 + limit, vlimit_name[lim], lim, (long)value);
21463 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/limit_proc.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit_proc.h
21464 --- linux-2.6.29.1/kernel/vserver/limit_proc.h 1970-01-01 01:00:00.000000000 +0100
21465 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/limit_proc.h 2009-02-22 22:54:26.000000000 +0100
21467 +#ifndef _VX_LIMIT_PROC_H
21468 +#define _VX_LIMIT_PROC_H
21470 +#include <linux/vserver/limit_int.h>
21473 +#define VX_LIMIT_FMT ":\t%8ld\t%8ld/%8ld\t%8lld/%8lld\t%6d\n"
21474 +#define VX_LIMIT_TOP \
21475 + "Limit\t current\t min/max\t\t soft/hard\t\thits\n"
21477 +#define VX_LIMIT_ARG(r) \
21478 + (unsigned long)__rlim_get(limit, r), \
21479 + (unsigned long)__rlim_rmin(limit, r), \
21480 + (unsigned long)__rlim_rmax(limit, r), \
21481 + VX_VLIM(__rlim_soft(limit, r)), \
21482 + VX_VLIM(__rlim_hard(limit, r)), \
21483 + atomic_read(&__rlim_lhit(limit, r))
21485 +static inline int vx_info_proc_limit(struct _vx_limit *limit, char *buffer)
21487 + vx_limit_fixup(limit, -1);
21488 + return sprintf(buffer, VX_LIMIT_TOP
21489 + "PROC" VX_LIMIT_FMT
21490 + "VM" VX_LIMIT_FMT
21491 + "VML" VX_LIMIT_FMT
21492 + "RSS" VX_LIMIT_FMT
21493 + "ANON" VX_LIMIT_FMT
21494 + "RMAP" VX_LIMIT_FMT
21495 + "FILES" VX_LIMIT_FMT
21496 + "OFD" VX_LIMIT_FMT
21497 + "LOCKS" VX_LIMIT_FMT
21498 + "SOCK" VX_LIMIT_FMT
21499 + "MSGQ" VX_LIMIT_FMT
21500 + "SHM" VX_LIMIT_FMT
21501 + "SEMA" VX_LIMIT_FMT
21502 + "SEMS" VX_LIMIT_FMT
21503 + "DENT" VX_LIMIT_FMT,
21504 + VX_LIMIT_ARG(RLIMIT_NPROC),
21505 + VX_LIMIT_ARG(RLIMIT_AS),
21506 + VX_LIMIT_ARG(RLIMIT_MEMLOCK),
21507 + VX_LIMIT_ARG(RLIMIT_RSS),
21508 + VX_LIMIT_ARG(VLIMIT_ANON),
21509 + VX_LIMIT_ARG(VLIMIT_MAPPED),
21510 + VX_LIMIT_ARG(RLIMIT_NOFILE),
21511 + VX_LIMIT_ARG(VLIMIT_OPENFD),
21512 + VX_LIMIT_ARG(RLIMIT_LOCKS),
21513 + VX_LIMIT_ARG(VLIMIT_NSOCK),
21514 + VX_LIMIT_ARG(RLIMIT_MSGQUEUE),
21515 + VX_LIMIT_ARG(VLIMIT_SHMEM),
21516 + VX_LIMIT_ARG(VLIMIT_SEMARY),
21517 + VX_LIMIT_ARG(VLIMIT_NSEMS),
21518 + VX_LIMIT_ARG(VLIMIT_DENTRY));
21521 +#endif /* _VX_LIMIT_PROC_H */
21524 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/Makefile linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/Makefile
21525 --- linux-2.6.29.1/kernel/vserver/Makefile 1970-01-01 01:00:00.000000000 +0100
21526 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/Makefile 2009-02-22 22:54:26.000000000 +0100
21529 +# Makefile for the Linux vserver routines.
21533 +obj-y += vserver.o
21535 +vserver-y := switch.o context.o space.o sched.o network.o inode.o \
21536 + limit.o cvirt.o cacct.o signal.o helper.o init.o \
21539 +vserver-$(CONFIG_INET) += inet.o
21540 +vserver-$(CONFIG_PROC_FS) += proc.o
21541 +vserver-$(CONFIG_VSERVER_DEBUG) += sysctl.o debug.o
21542 +vserver-$(CONFIG_VSERVER_HISTORY) += history.o
21543 +vserver-$(CONFIG_VSERVER_MONITOR) += monitor.o
21544 +vserver-$(CONFIG_VSERVER_DEVICE) += device.o
21546 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/monitor.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/monitor.c
21547 --- linux-2.6.29.1/kernel/vserver/monitor.c 1970-01-01 01:00:00.000000000 +0100
21548 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/monitor.c 2009-02-22 22:54:26.000000000 +0100
21551 + * kernel/vserver/monitor.c
21553 + * Virtual Context Scheduler Monitor
21555 + * Copyright (C) 2006-2007 Herbert Pötzl
21557 + * V0.01 basic design
21561 +#include <linux/module.h>
21562 +#include <linux/jiffies.h>
21563 +#include <asm/uaccess.h>
21564 +#include <asm/atomic.h>
21566 +#include <linux/vserver/monitor.h>
21567 +#include <linux/vserver/debug_cmd.h>
21570 +#ifdef CONFIG_VSERVER_MONITOR
21571 +#define VXM_SIZE CONFIG_VSERVER_MONITOR_SIZE
21573 +#define VXM_SIZE 64
21576 +struct _vx_monitor {
21577 + unsigned int counter;
21579 + struct _vx_mon_entry entry[VXM_SIZE+1];
21583 +DEFINE_PER_CPU(struct _vx_monitor, vx_monitor_buffer);
21585 +unsigned volatile int vxm_active = 1;
21587 +static atomic_t sequence = ATOMIC_INIT(0);
21592 + * requires disabled preemption */
21594 +struct _vx_mon_entry *vxm_advance(int cpu)
21596 + struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
21597 + struct _vx_mon_entry *entry;
21598 + unsigned int index;
21600 + index = vxm_active ? (mon->counter++ % VXM_SIZE) : VXM_SIZE;
21601 + entry = &mon->entry[index];
21603 + entry->ev.seq = atomic_inc_return(&sequence);
21604 + entry->ev.jif = jiffies;
21608 +EXPORT_SYMBOL_GPL(vxm_advance);
21611 +int do_read_monitor(struct __user _vx_mon_entry *data,
21612 + int cpu, uint32_t *index, uint32_t *count)
21614 + int pos, ret = 0;
21615 + struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
21616 + int end = mon->counter;
21617 + int start = end - VXM_SIZE + 2;
21618 + int idx = *index;
21620 + /* special case: get current pos */
21626 + /* have we lost some data? */
21630 + for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
21631 + struct _vx_mon_entry *entry =
21632 + &mon->entry[idx % VXM_SIZE];
21634 + /* send entry to userspace */
21635 + ret = copy_to_user(&data[pos], entry, sizeof(*entry));
21639 + /* save new index and count */
21642 + return ret ? ret : (*index < end);
21645 +int vc_read_monitor(uint32_t id, void __user *data)
21647 + struct vcmd_read_monitor_v0 vc_data;
21650 + if (id >= NR_CPUS)
21653 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21656 + ret = do_read_monitor((struct __user _vx_mon_entry *)vc_data.data,
21657 + id, &vc_data.index, &vc_data.count);
21659 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21664 +#ifdef CONFIG_COMPAT
21666 +int vc_read_monitor_x32(uint32_t id, void __user *data)
21668 + struct vcmd_read_monitor_v0_x32 vc_data;
21671 + if (id >= NR_CPUS)
21674 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21677 + ret = do_read_monitor((struct __user _vx_mon_entry *)
21678 + compat_ptr(vc_data.data_ptr),
21679 + id, &vc_data.index, &vc_data.count);
21681 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21686 +#endif /* CONFIG_COMPAT */
21688 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/network.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/network.c
21689 --- linux-2.6.29.1/kernel/vserver/network.c 1970-01-01 01:00:00.000000000 +0100
21690 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/network.c 2009-02-22 22:54:26.000000000 +0100
21693 + * linux/kernel/vserver/network.c
21695 + * Virtual Server: Network Support
21697 + * Copyright (C) 2003-2007 Herbert Pötzl
21699 + * V0.01 broken out from vcontext V0.05
21700 + * V0.02 cleaned up implementation
21701 + * V0.03 added equiv nx commands
21702 + * V0.04 switch to RCU based hash
21703 + * V0.05 and back to locking again
21704 + * V0.06 changed vcmds to nxi arg
21705 + * V0.07 have __create claim() the nxi
21709 +#include <linux/err.h>
21710 +#include <linux/slab.h>
21711 +#include <linux/rcupdate.h>
21713 +#include <linux/vs_network.h>
21714 +#include <linux/vs_pid.h>
21715 +#include <linux/vserver/network_cmd.h>
21718 +atomic_t nx_global_ctotal = ATOMIC_INIT(0);
21719 +atomic_t nx_global_cactive = ATOMIC_INIT(0);
21721 +static struct kmem_cache *nx_addr_v4_cachep = NULL;
21722 +static struct kmem_cache *nx_addr_v6_cachep = NULL;
21725 +static int __init init_network(void)
21727 + nx_addr_v4_cachep = kmem_cache_create("nx_v4_addr_cache",
21728 + sizeof(struct nx_addr_v4), 0,
21729 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
21730 + nx_addr_v6_cachep = kmem_cache_create("nx_v6_addr_cache",
21731 + sizeof(struct nx_addr_v6), 0,
21732 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
21737 +/* __alloc_nx_addr_v4() */
21739 +static inline struct nx_addr_v4 *__alloc_nx_addr_v4(void)
21741 + struct nx_addr_v4 *nxa = kmem_cache_alloc(
21742 + nx_addr_v4_cachep, GFP_KERNEL);
21744 + if (!IS_ERR(nxa))
21745 + memset(nxa, 0, sizeof(*nxa));
21749 +/* __dealloc_nx_addr_v4() */
21751 +static inline void __dealloc_nx_addr_v4(struct nx_addr_v4 *nxa)
21753 + kmem_cache_free(nx_addr_v4_cachep, nxa);
21756 +/* __dealloc_nx_addr_v4_all() */
21758 +static inline void __dealloc_nx_addr_v4_all(struct nx_addr_v4 *nxa)
21761 + struct nx_addr_v4 *next = nxa->next;
21763 + __dealloc_nx_addr_v4(nxa);
21769 +#ifdef CONFIG_IPV6
21771 +/* __alloc_nx_addr_v6() */
21773 +static inline struct nx_addr_v6 *__alloc_nx_addr_v6(void)
21775 + struct nx_addr_v6 *nxa = kmem_cache_alloc(
21776 + nx_addr_v6_cachep, GFP_KERNEL);
21778 + if (!IS_ERR(nxa))
21779 + memset(nxa, 0, sizeof(*nxa));
21783 +/* __dealloc_nx_addr_v6() */
21785 +static inline void __dealloc_nx_addr_v6(struct nx_addr_v6 *nxa)
21787 + kmem_cache_free(nx_addr_v6_cachep, nxa);
21790 +/* __dealloc_nx_addr_v6_all() */
21792 +static inline void __dealloc_nx_addr_v6_all(struct nx_addr_v6 *nxa)
21795 + struct nx_addr_v6 *next = nxa->next;
21797 + __dealloc_nx_addr_v6(nxa);
21802 +#endif /* CONFIG_IPV6 */
21804 +/* __alloc_nx_info()
21806 + * allocate an initialized nx_info struct
21807 + * doesn't make it visible (hash) */
21809 +static struct nx_info *__alloc_nx_info(nid_t nid)
21811 + struct nx_info *new = NULL;
21813 + vxdprintk(VXD_CBIT(nid, 1), "alloc_nx_info(%d)*", nid);
21815 + /* would this benefit from a slab cache? */
21816 + new = kmalloc(sizeof(struct nx_info), GFP_KERNEL);
21820 + memset(new, 0, sizeof(struct nx_info));
21821 + new->nx_id = nid;
21822 + INIT_HLIST_NODE(&new->nx_hlist);
21823 + atomic_set(&new->nx_usecnt, 0);
21824 + atomic_set(&new->nx_tasks, 0);
21825 + new->nx_state = 0;
21827 + new->nx_flags = NXF_INIT_SET;
21829 + /* rest of init goes here */
21831 + new->v4_lback.s_addr = htonl(INADDR_LOOPBACK);
21832 + new->v4_bcast.s_addr = htonl(INADDR_BROADCAST);
21834 + vxdprintk(VXD_CBIT(nid, 0),
21835 + "alloc_nx_info(%d) = %p", nid, new);
21836 + atomic_inc(&nx_global_ctotal);
21840 +/* __dealloc_nx_info()
21842 + * final disposal of nx_info */
21844 +static void __dealloc_nx_info(struct nx_info *nxi)
21846 + vxdprintk(VXD_CBIT(nid, 0),
21847 + "dealloc_nx_info(%p)", nxi);
21849 + nxi->nx_hlist.next = LIST_POISON1;
21852 + BUG_ON(atomic_read(&nxi->nx_usecnt));
21853 + BUG_ON(atomic_read(&nxi->nx_tasks));
21855 + __dealloc_nx_addr_v4_all(nxi->v4.next);
21857 + nxi->nx_state |= NXS_RELEASED;
21859 + atomic_dec(&nx_global_ctotal);
21862 +static void __shutdown_nx_info(struct nx_info *nxi)
21864 + nxi->nx_state |= NXS_SHUTDOWN;
21865 + vs_net_change(nxi, VSC_NETDOWN);
21868 +/* exported stuff */
21870 +void free_nx_info(struct nx_info *nxi)
21872 + /* context shutdown is mandatory */
21873 + BUG_ON(nxi->nx_state != NXS_SHUTDOWN);
21875 + /* context must not be hashed */
21876 + BUG_ON(nxi->nx_state & NXS_HASHED);
21878 + BUG_ON(atomic_read(&nxi->nx_usecnt));
21879 + BUG_ON(atomic_read(&nxi->nx_tasks));
21881 + __dealloc_nx_info(nxi);
21885 +void __nx_set_lback(struct nx_info *nxi)
21887 + int nid = nxi->nx_id;
21888 + __be32 lback = htonl(INADDR_LOOPBACK ^ ((nid & 0xFFFF) << 8));
21890 + nxi->v4_lback.s_addr = lback;
21893 +extern int __nx_inet_add_lback(__be32 addr);
21894 +extern int __nx_inet_del_lback(__be32 addr);
21897 +/* hash table for nx_info hash */
21899 +#define NX_HASH_SIZE 13
21901 +struct hlist_head nx_info_hash[NX_HASH_SIZE];
21903 +static spinlock_t nx_info_hash_lock = SPIN_LOCK_UNLOCKED;
21906 +static inline unsigned int __hashval(nid_t nid)
21908 + return (nid % NX_HASH_SIZE);
21913 +/* __hash_nx_info()
21915 + * add the nxi to the global hash table
21916 + * requires the hash_lock to be held */
21918 +static inline void __hash_nx_info(struct nx_info *nxi)
21920 + struct hlist_head *head;
21922 + vxd_assert_lock(&nx_info_hash_lock);
21923 + vxdprintk(VXD_CBIT(nid, 4),
21924 + "__hash_nx_info: %p[#%d]", nxi, nxi->nx_id);
21926 + /* context must not be hashed */
21927 + BUG_ON(nx_info_state(nxi, NXS_HASHED));
21929 + nxi->nx_state |= NXS_HASHED;
21930 + head = &nx_info_hash[__hashval(nxi->nx_id)];
21931 + hlist_add_head(&nxi->nx_hlist, head);
21932 + atomic_inc(&nx_global_cactive);
21935 +/* __unhash_nx_info()
21937 + * remove the nxi from the global hash table
21938 + * requires the hash_lock to be held */
21940 +static inline void __unhash_nx_info(struct nx_info *nxi)
21942 + vxd_assert_lock(&nx_info_hash_lock);
21943 + vxdprintk(VXD_CBIT(nid, 4),
21944 + "__unhash_nx_info: %p[#%d.%d.%d]", nxi, nxi->nx_id,
21945 + atomic_read(&nxi->nx_usecnt), atomic_read(&nxi->nx_tasks));
21947 + /* context must be hashed */
21948 + BUG_ON(!nx_info_state(nxi, NXS_HASHED));
21949 + /* but without tasks */
21950 + BUG_ON(atomic_read(&nxi->nx_tasks));
21952 + nxi->nx_state &= ~NXS_HASHED;
21953 + hlist_del(&nxi->nx_hlist);
21954 + atomic_dec(&nx_global_cactive);
21958 +/* __lookup_nx_info()
21960 + * requires the hash_lock to be held
21961 + * doesn't increment the nx_refcnt */
21963 +static inline struct nx_info *__lookup_nx_info(nid_t nid)
21965 + struct hlist_head *head = &nx_info_hash[__hashval(nid)];
21966 + struct hlist_node *pos;
21967 + struct nx_info *nxi;
21969 + vxd_assert_lock(&nx_info_hash_lock);
21970 + hlist_for_each(pos, head) {
21971 + nxi = hlist_entry(pos, struct nx_info, nx_hlist);
21973 + if (nxi->nx_id == nid)
21978 + vxdprintk(VXD_CBIT(nid, 0),
21979 + "__lookup_nx_info(#%u): %p[#%u]",
21980 + nid, nxi, nxi ? nxi->nx_id : 0);
21985 +/* __create_nx_info()
21987 + * create the requested context
21988 + * get(), claim() and hash it */
21990 +static struct nx_info *__create_nx_info(int id)
21992 + struct nx_info *new, *nxi = NULL;
21994 + vxdprintk(VXD_CBIT(nid, 1), "create_nx_info(%d)*", id);
21996 + if (!(new = __alloc_nx_info(id)))
21997 + return ERR_PTR(-ENOMEM);
21999 + /* required to make dynamic xids unique */
22000 + spin_lock(&nx_info_hash_lock);
22002 + /* static context requested */
22003 + if ((nxi = __lookup_nx_info(id))) {
22004 + vxdprintk(VXD_CBIT(nid, 0),
22005 + "create_nx_info(%d) = %p (already there)", id, nxi);
22006 + if (nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22007 + nxi = ERR_PTR(-EBUSY);
22009 + nxi = ERR_PTR(-EEXIST);
22012 + /* new context */
22013 + vxdprintk(VXD_CBIT(nid, 0),
22014 + "create_nx_info(%d) = %p (new)", id, new);
22015 + claim_nx_info(new, NULL);
22016 + __nx_set_lback(new);
22017 + __hash_nx_info(get_nx_info(new));
22018 + nxi = new, new = NULL;
22021 + spin_unlock(&nx_info_hash_lock);
22023 + __dealloc_nx_info(new);
22029 +/* exported stuff */
22032 +void unhash_nx_info(struct nx_info *nxi)
22034 + __shutdown_nx_info(nxi);
22035 + spin_lock(&nx_info_hash_lock);
22036 + __unhash_nx_info(nxi);
22037 + spin_unlock(&nx_info_hash_lock);
22040 +/* lookup_nx_info()
22042 + * search for a nx_info and get() it
22043 + * negative id means current */
22045 +struct nx_info *lookup_nx_info(int id)
22047 + struct nx_info *nxi = NULL;
22050 + nxi = get_nx_info(current->nx_info);
22051 + } else if (id > 1) {
22052 + spin_lock(&nx_info_hash_lock);
22053 + nxi = get_nx_info(__lookup_nx_info(id));
22054 + spin_unlock(&nx_info_hash_lock);
22059 +/* nid_is_hashed()
22061 + * verify that nid is still hashed */
22063 +int nid_is_hashed(nid_t nid)
22067 + spin_lock(&nx_info_hash_lock);
22068 + hashed = (__lookup_nx_info(nid) != NULL);
22069 + spin_unlock(&nx_info_hash_lock);
22074 +#ifdef CONFIG_PROC_FS
22078 + * get a subset of hashed nids for proc
22079 + * assumes size is at least one */
22081 +int get_nid_list(int index, unsigned int *nids, int size)
22083 + int hindex, nr_nids = 0;
22085 + /* only show current and children */
22086 + if (!nx_check(0, VS_ADMIN | VS_WATCH)) {
22089 + nids[nr_nids] = nx_current_nid();
22093 + for (hindex = 0; hindex < NX_HASH_SIZE; hindex++) {
22094 + struct hlist_head *head = &nx_info_hash[hindex];
22095 + struct hlist_node *pos;
22097 + spin_lock(&nx_info_hash_lock);
22098 + hlist_for_each(pos, head) {
22099 + struct nx_info *nxi;
22104 + nxi = hlist_entry(pos, struct nx_info, nx_hlist);
22105 + nids[nr_nids] = nxi->nx_id;
22106 + if (++nr_nids >= size) {
22107 + spin_unlock(&nx_info_hash_lock);
22111 + /* keep the lock time short */
22112 + spin_unlock(&nx_info_hash_lock);
22121 + * migrate task to new network
22122 + * gets nxi, puts old_nxi on change
22125 +int nx_migrate_task(struct task_struct *p, struct nx_info *nxi)
22127 + struct nx_info *old_nxi;
22133 + vxdprintk(VXD_CBIT(nid, 5),
22134 + "nx_migrate_task(%p,%p[#%d.%d.%d])",
22135 + p, nxi, nxi->nx_id,
22136 + atomic_read(&nxi->nx_usecnt),
22137 + atomic_read(&nxi->nx_tasks));
22139 + if (nx_info_flags(nxi, NXF_INFO_PRIVATE, 0) &&
22140 + !nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22143 + if (nx_info_state(nxi, NXS_SHUTDOWN))
22146 + /* maybe disallow this completely? */
22147 + old_nxi = task_get_nx_info(p);
22148 + if (old_nxi == nxi)
22153 + clr_nx_info(&p->nx_info);
22154 + claim_nx_info(nxi, p);
22155 + set_nx_info(&p->nx_info, nxi);
22156 + p->nid = nxi->nx_id;
22159 + vxdprintk(VXD_CBIT(nid, 5),
22160 + "moved task %p into nxi:%p[#%d]",
22161 + p, nxi, nxi->nx_id);
22164 + release_nx_info(old_nxi, p);
22167 + put_nx_info(old_nxi);
22172 +void nx_set_persistent(struct nx_info *nxi)
22174 + vxdprintk(VXD_CBIT(nid, 6),
22175 + "nx_set_persistent(%p[#%d])", nxi, nxi->nx_id);
22177 + get_nx_info(nxi);
22178 + claim_nx_info(nxi, NULL);
22181 +void nx_clear_persistent(struct nx_info *nxi)
22183 + vxdprintk(VXD_CBIT(nid, 6),
22184 + "nx_clear_persistent(%p[#%d])", nxi, nxi->nx_id);
22186 + release_nx_info(nxi, NULL);
22187 + put_nx_info(nxi);
22190 +void nx_update_persistent(struct nx_info *nxi)
22192 + if (nx_info_flags(nxi, NXF_PERSISTENT, 0))
22193 + nx_set_persistent(nxi);
22195 + nx_clear_persistent(nxi);
22198 +/* vserver syscall commands below here */
22200 +/* taks nid and nx_info functions */
22202 +#include <asm/uaccess.h>
22205 +int vc_task_nid(uint32_t id)
22210 + struct task_struct *tsk;
22212 + read_lock(&tasklist_lock);
22213 + tsk = find_task_by_real_pid(id);
22214 + nid = (tsk) ? tsk->nid : -ESRCH;
22215 + read_unlock(&tasklist_lock);
22217 + nid = nx_current_nid();
22222 +int vc_nx_info(struct nx_info *nxi, void __user *data)
22224 + struct vcmd_nx_info_v0 vc_data;
22226 + vc_data.nid = nxi->nx_id;
22228 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22234 +/* network functions */
22236 +int vc_net_create(uint32_t nid, void __user *data)
22238 + struct vcmd_net_create vc_data = { .flagword = NXF_INIT_SET };
22239 + struct nx_info *new_nxi;
22242 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22245 + if ((nid > MAX_S_CONTEXT) || (nid < 2))
22248 + new_nxi = __create_nx_info(nid);
22249 + if (IS_ERR(new_nxi))
22250 + return PTR_ERR(new_nxi);
22252 + /* initial flags */
22253 + new_nxi->nx_flags = vc_data.flagword;
22256 + if (vs_net_change(new_nxi, VSC_NETUP))
22259 + ret = nx_migrate_task(current, new_nxi);
22263 + /* return context id on success */
22264 + ret = new_nxi->nx_id;
22266 + /* get a reference for persistent contexts */
22267 + if ((vc_data.flagword & NXF_PERSISTENT))
22268 + nx_set_persistent(new_nxi);
22270 + release_nx_info(new_nxi, NULL);
22271 + put_nx_info(new_nxi);
22276 +int vc_net_migrate(struct nx_info *nxi, void __user *data)
22278 + return nx_migrate_task(current, nxi);
22283 +int do_add_v4_addr(struct nx_info *nxi, __be32 ip, __be32 ip2, __be32 mask,
22284 + uint16_t type, uint16_t flags)
22286 + struct nx_addr_v4 *nxa = &nxi->v4;
22288 + if (NX_IPV4(nxi)) {
22289 + /* locate last entry */
22290 + for (; nxa->next; nxa = nxa->next);
22291 + nxa->next = __alloc_nx_addr_v4();
22295 + return PTR_ERR(nxa);
22298 + if (nxi->v4.next)
22299 + /* remove single ip for ip list */
22300 + nxi->nx_flags &= ~NXF_SINGLE_IP;
22302 + nxa->ip[0].s_addr = ip;
22303 + nxa->ip[1].s_addr = ip2;
22304 + nxa->mask.s_addr = mask;
22305 + nxa->type = type;
22306 + nxa->flags = flags;
22311 +int vc_net_add(struct nx_info *nxi, void __user *data)
22313 + struct vcmd_net_addr_v0 vc_data;
22314 + int index, ret = 0;
22316 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22319 + switch (vc_data.type) {
22320 + case NXA_TYPE_IPV4:
22321 + if ((vc_data.count < 1) || (vc_data.count > 4))
22325 + while (index < vc_data.count) {
22326 + ret = do_add_v4_addr(nxi, vc_data.ip[index].s_addr, 0,
22327 + vc_data.mask[index].s_addr, NXA_TYPE_ADDR, 0);
22335 + case NXA_TYPE_IPV4|NXA_MOD_BCAST:
22336 + nxi->v4_bcast = vc_data.ip[0];
22340 + case NXA_TYPE_IPV4|NXA_MOD_LBACK:
22341 + nxi->v4_lback = vc_data.ip[0];
22352 +int vc_net_remove(struct nx_info *nxi, void __user *data)
22354 + struct vcmd_net_addr_v0 vc_data;
22356 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22359 + switch (vc_data.type) {
22360 + case NXA_TYPE_ANY:
22361 + __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22362 + memset(&nxi->v4, 0, sizeof(nxi->v4));
22372 +int vc_net_add_ipv4(struct nx_info *nxi, void __user *data)
22374 + struct vcmd_net_addr_ipv4_v1 vc_data;
22376 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22379 + switch (vc_data.type) {
22380 + case NXA_TYPE_ADDR:
22381 + case NXA_TYPE_RANGE:
22382 + case NXA_TYPE_MASK:
22383 + return do_add_v4_addr(nxi, vc_data.ip.s_addr, 0,
22384 + vc_data.mask.s_addr, vc_data.type, vc_data.flags);
22386 + case NXA_TYPE_ADDR | NXA_MOD_BCAST:
22387 + nxi->v4_bcast = vc_data.ip;
22390 + case NXA_TYPE_ADDR | NXA_MOD_LBACK:
22391 + nxi->v4_lback = vc_data.ip;
22400 +int vc_net_remove_ipv4(struct nx_info *nxi, void __user *data)
22402 + struct vcmd_net_addr_ipv4_v1 vc_data;
22404 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22407 + switch (vc_data.type) {
22408 +/* case NXA_TYPE_ADDR:
22411 + case NXA_TYPE_ANY:
22412 + __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22413 + memset(&nxi->v4, 0, sizeof(nxi->v4));
22423 +#ifdef CONFIG_IPV6
22425 +int do_add_v6_addr(struct nx_info *nxi,
22426 + struct in6_addr *ip, struct in6_addr *mask,
22427 + uint32_t prefix, uint16_t type, uint16_t flags)
22429 + struct nx_addr_v6 *nxa = &nxi->v6;
22431 + if (NX_IPV6(nxi)) {
22432 + /* locate last entry */
22433 + for (; nxa->next; nxa = nxa->next);
22434 + nxa->next = __alloc_nx_addr_v6();
22438 + return PTR_ERR(nxa);
22442 + nxa->mask = *mask;
22443 + nxa->prefix = prefix;
22444 + nxa->type = type;
22445 + nxa->flags = flags;
22450 +int vc_net_add_ipv6(struct nx_info *nxi, void __user *data)
22452 + struct vcmd_net_addr_ipv6_v1 vc_data;
22454 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22457 + switch (vc_data.type) {
22458 + case NXA_TYPE_ADDR:
22459 + case NXA_TYPE_MASK:
22460 + return do_add_v6_addr(nxi, &vc_data.ip, &vc_data.mask,
22461 + vc_data.prefix, vc_data.type, vc_data.flags);
22468 +int vc_net_remove_ipv6(struct nx_info *nxi, void __user *data)
22470 + struct vcmd_net_addr_ipv6_v1 vc_data;
22472 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22475 + switch (vc_data.type) {
22476 + case NXA_TYPE_ANY:
22477 + __dealloc_nx_addr_v6_all(xchg(&nxi->v6.next, NULL));
22478 + memset(&nxi->v6, 0, sizeof(nxi->v6));
22487 +#endif /* CONFIG_IPV6 */
22490 +int vc_get_nflags(struct nx_info *nxi, void __user *data)
22492 + struct vcmd_net_flags_v0 vc_data;
22494 + vc_data.flagword = nxi->nx_flags;
22496 + /* special STATE flag handling */
22497 + vc_data.mask = vs_mask_flags(~0ULL, nxi->nx_flags, NXF_ONE_TIME);
22499 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22504 +int vc_set_nflags(struct nx_info *nxi, void __user *data)
22506 + struct vcmd_net_flags_v0 vc_data;
22507 + uint64_t mask, trigger;
22509 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22512 + /* special STATE flag handling */
22513 + mask = vs_mask_mask(vc_data.mask, nxi->nx_flags, NXF_ONE_TIME);
22514 + trigger = (mask & nxi->nx_flags) ^ (mask & vc_data.flagword);
22516 + nxi->nx_flags = vs_mask_flags(nxi->nx_flags,
22517 + vc_data.flagword, mask);
22518 + if (trigger & NXF_PERSISTENT)
22519 + nx_update_persistent(nxi);
22524 +int vc_get_ncaps(struct nx_info *nxi, void __user *data)
22526 + struct vcmd_net_caps_v0 vc_data;
22528 + vc_data.ncaps = nxi->nx_ncaps;
22529 + vc_data.cmask = ~0ULL;
22531 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22536 +int vc_set_ncaps(struct nx_info *nxi, void __user *data)
22538 + struct vcmd_net_caps_v0 vc_data;
22540 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22543 + nxi->nx_ncaps = vs_mask_flags(nxi->nx_ncaps,
22544 + vc_data.ncaps, vc_data.cmask);
22549 +#include <linux/module.h>
22551 +module_init(init_network);
22553 +EXPORT_SYMBOL_GPL(free_nx_info);
22554 +EXPORT_SYMBOL_GPL(unhash_nx_info);
22556 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/proc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/proc.c
22557 --- linux-2.6.29.1/kernel/vserver/proc.c 1970-01-01 01:00:00.000000000 +0100
22558 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/proc.c 2009-02-22 22:54:26.000000000 +0100
22561 + * linux/kernel/vserver/proc.c
22563 + * Virtual Context Support
22565 + * Copyright (C) 2003-2007 Herbert Pötzl
22567 + * V0.01 basic structure
22568 + * V0.02 adaptation vs1.3.0
22569 + * V0.03 proc permissions
22570 + * V0.04 locking/generic
22571 + * V0.05 next generation procfs
22572 + * V0.06 inode validation
22573 + * V0.07 generic rewrite vid
22574 + * V0.08 remove inode type
22578 +#include <linux/proc_fs.h>
22579 +#include <asm/unistd.h>
22581 +#include <linux/vs_context.h>
22582 +#include <linux/vs_network.h>
22583 +#include <linux/vs_cvirt.h>
22585 +#include <linux/in.h>
22586 +#include <linux/inetdevice.h>
22587 +#include <linux/vs_inet.h>
22588 +#include <linux/vs_inet6.h>
22590 +#include <linux/vserver/global.h>
22592 +#include "cvirt_proc.h"
22593 +#include "cacct_proc.h"
22594 +#include "limit_proc.h"
22595 +#include "sched_proc.h"
22596 +#include "vci_config.h"
22599 +static inline char *print_cap_t(char *buffer, kernel_cap_t *c)
22603 + CAP_FOR_EACH_U32(__capi) {
22604 + buffer += sprintf(buffer, "%08x",
22605 + c->cap[(_KERNEL_CAPABILITY_U32S-1) - __capi]);
22611 +static struct proc_dir_entry *proc_virtual;
22613 +static struct proc_dir_entry *proc_virtnet;
22616 +/* first the actual feeds */
22619 +static int proc_vci(char *buffer)
22621 + return sprintf(buffer,
22622 + "VCIVersion:\t%04x:%04x\n"
22623 + "VCISyscall:\t%d\n"
22624 + "VCIKernel:\t%08x\n",
22625 + VCI_VERSION >> 16,
22626 + VCI_VERSION & 0xFFFF,
22628 + vci_kernel_config());
22631 +static int proc_virtual_info(char *buffer)
22633 + return proc_vci(buffer);
22636 +static int proc_virtual_status(char *buffer)
22638 + return sprintf(buffer,
22640 + "#CActive:\t%d\n"
22641 + "#NSProxy:\t%d\t%d %d %d %d %d %d\n",
22642 + atomic_read(&vx_global_ctotal),
22643 + atomic_read(&vx_global_cactive),
22644 + atomic_read(&vs_global_nsproxy),
22645 + atomic_read(&vs_global_fs),
22646 + atomic_read(&vs_global_mnt_ns),
22647 + atomic_read(&vs_global_uts_ns),
22648 + atomic_read(&vs_global_ipc_ns),
22649 + atomic_read(&vs_global_user_ns),
22650 + atomic_read(&vs_global_pid_ns));
22654 +int proc_vxi_info(struct vx_info *vxi, char *buffer)
22658 + length = sprintf(buffer,
22666 + vxi->vx_badness_bias);
22670 +int proc_vxi_status(struct vx_info *vxi, char *buffer)
22672 + char *orig = buffer;
22674 + buffer += sprintf(buffer,
22677 + "Flags:\t%016llx\n",
22678 + atomic_read(&vxi->vx_usecnt),
22679 + atomic_read(&vxi->vx_tasks),
22680 + (unsigned long long)vxi->vx_flags);
22682 + buffer += sprintf(buffer, "BCaps:\t");
22683 + buffer = print_cap_t(buffer, &vxi->vx_bcaps);
22684 + buffer += sprintf(buffer, "\n");
22686 + buffer += sprintf(buffer,
22687 + "CCaps:\t%016llx\n"
22688 + "Spaces:\t%08lx %08lx\n",
22689 + (unsigned long long)vxi->vx_ccaps,
22690 + vxi->vx_nsmask[0], vxi->vx_nsmask[1]);
22691 + return buffer - orig;
22694 +int proc_vxi_limit(struct vx_info *vxi, char *buffer)
22696 + return vx_info_proc_limit(&vxi->limit, buffer);
22699 +int proc_vxi_sched(struct vx_info *vxi, char *buffer)
22703 + length = vx_info_proc_sched(&vxi->sched, buffer);
22704 + for_each_online_cpu(cpu) {
22705 + length += vx_info_proc_sched_pc(
22706 + &vx_per_cpu(vxi, sched_pc, cpu),
22707 + buffer + length, cpu);
22712 +int proc_vxi_nsproxy0(struct vx_info *vxi, char *buffer)
22714 + return vx_info_proc_nsproxy(vxi->vx_nsproxy[0], buffer);
22717 +int proc_vxi_nsproxy1(struct vx_info *vxi, char *buffer)
22719 + return vx_info_proc_nsproxy(vxi->vx_nsproxy[1], buffer);
22722 +int proc_vxi_cvirt(struct vx_info *vxi, char *buffer)
22726 + vx_update_load(vxi);
22727 + length = vx_info_proc_cvirt(&vxi->cvirt, buffer);
22728 + for_each_online_cpu(cpu) {
22729 + length += vx_info_proc_cvirt_pc(
22730 + &vx_per_cpu(vxi, cvirt_pc, cpu),
22731 + buffer + length, cpu);
22736 +int proc_vxi_cacct(struct vx_info *vxi, char *buffer)
22738 + return vx_info_proc_cacct(&vxi->cacct, buffer);
22742 +static int proc_virtnet_info(char *buffer)
22744 + return proc_vci(buffer);
22747 +static int proc_virtnet_status(char *buffer)
22749 + return sprintf(buffer,
22751 + "#CActive:\t%d\n",
22752 + atomic_read(&nx_global_ctotal),
22753 + atomic_read(&nx_global_cactive));
22756 +int proc_nxi_info(struct nx_info *nxi, char *buffer)
22758 + struct nx_addr_v4 *v4a;
22759 +#ifdef CONFIG_IPV6
22760 + struct nx_addr_v6 *v6a;
22764 + length = sprintf(buffer,
22767 + "Bcast:\t" NIPQUAD_FMT "\n"
22768 + "Lback:\t" NIPQUAD_FMT "\n",
22771 + NIPQUAD(nxi->v4_bcast.s_addr),
22772 + NIPQUAD(nxi->v4_lback.s_addr));
22774 + if (!NX_IPV4(nxi))
22776 + for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
22777 + length += sprintf(buffer + length, "%d:\t" NXAV4_FMT "\n",
22780 +#ifdef CONFIG_IPV6
22781 + if (!NX_IPV6(nxi))
22783 + for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
22784 + length += sprintf(buffer + length, "%d:\t" NXAV6_FMT "\n",
22791 +int proc_nxi_status(struct nx_info *nxi, char *buffer)
22795 + length = sprintf(buffer,
22798 + "Flags:\t%016llx\n"
22799 + "NCaps:\t%016llx\n",
22800 + atomic_read(&nxi->nx_usecnt),
22801 + atomic_read(&nxi->nx_tasks),
22802 + (unsigned long long)nxi->nx_flags,
22803 + (unsigned long long)nxi->nx_ncaps);
22809 +/* here the inode helpers */
22815 + struct inode_operations *iop;
22816 + struct file_operations *fop;
22817 + union proc_op op;
22820 +static struct inode *vs_proc_make_inode(struct super_block *sb, struct vs_entry *p)
22822 + struct inode *inode = new_inode(sb);
22827 + inode->i_mode = p->mode;
22829 + inode->i_op = p->iop;
22831 + inode->i_fop = p->fop;
22833 + inode->i_nlink = (p->mode & S_IFDIR) ? 2 : 1;
22834 + inode->i_flags |= S_IMMUTABLE;
22836 + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
22838 + inode->i_uid = 0;
22839 + inode->i_gid = 0;
22840 + inode->i_tag = 0;
22845 +static struct dentry *vs_proc_instantiate(struct inode *dir,
22846 + struct dentry *dentry, int id, void *ptr)
22848 + struct vs_entry *p = ptr;
22849 + struct inode *inode = vs_proc_make_inode(dir->i_sb, p);
22850 + struct dentry *error = ERR_PTR(-EINVAL);
22855 + PROC_I(inode)->op = p->op;
22856 + PROC_I(inode)->fd = id;
22857 + d_add(dentry, inode);
22865 +typedef struct dentry *instantiate_t(struct inode *, struct dentry *, int, void *);
22868 + * Fill a directory entry.
22870 + * If possible create the dcache entry and derive our inode number and
22871 + * file type from dcache entry.
22873 + * Since all of the proc inode numbers are dynamically generated, the inode
22874 + * numbers do not exist until the inode is cache. This means creating the
22875 + * the dcache entry in readdir is necessary to keep the inode numbers
22876 + * reported by readdir in sync with the inode numbers reported
22879 +static int proc_fill_cache(struct file *filp, void *dirent, filldir_t filldir,
22880 + char *name, int len, instantiate_t instantiate, int id, void *ptr)
22882 + struct dentry *child, *dir = filp->f_dentry;
22883 + struct inode *inode;
22884 + struct qstr qname;
22886 + unsigned type = DT_UNKNOWN;
22888 + qname.name = name;
22890 + qname.hash = full_name_hash(name, len);
22892 + child = d_lookup(dir, &qname);
22894 + struct dentry *new;
22895 + new = d_alloc(dir, &qname);
22897 + child = instantiate(dir->d_inode, new, id, ptr);
22904 + if (!child || IS_ERR(child) || !child->d_inode)
22905 + goto end_instantiate;
22906 + inode = child->d_inode;
22908 + ino = inode->i_ino;
22909 + type = inode->i_mode >> 12;
22914 + ino = find_inode_number(dir, &qname);
22917 + return filldir(dirent, name, len, filp->f_pos, ino, type);
22922 +/* get and revalidate vx_info/xid */
22925 +struct vx_info *get_proc_vx_info(struct inode *inode)
22927 + return lookup_vx_info(PROC_I(inode)->fd);
22930 +static int proc_xid_revalidate(struct dentry *dentry, struct nameidata *nd)
22932 + struct inode *inode = dentry->d_inode;
22933 + xid_t xid = PROC_I(inode)->fd;
22935 + if (!xid || xid_is_hashed(xid))
22942 +/* get and revalidate nx_info/nid */
22944 +static int proc_nid_revalidate(struct dentry *dentry, struct nameidata *nd)
22946 + struct inode *inode = dentry->d_inode;
22947 + nid_t nid = PROC_I(inode)->fd;
22949 + if (!nid || nid_is_hashed(nid))
22957 +#define PROC_BLOCK_SIZE (PAGE_SIZE - 1024)
22959 +static ssize_t proc_vs_info_read(struct file *file, char __user *buf,
22960 + size_t count, loff_t *ppos)
22962 + struct inode *inode = file->f_dentry->d_inode;
22963 + unsigned long page;
22964 + ssize_t length = 0;
22966 + if (count > PROC_BLOCK_SIZE)
22967 + count = PROC_BLOCK_SIZE;
22969 + /* fade that out as soon as stable */
22970 + WARN_ON(PROC_I(inode)->fd);
22972 + if (!(page = __get_free_page(GFP_KERNEL)))
22975 + BUG_ON(!PROC_I(inode)->op.proc_vs_read);
22976 + length = PROC_I(inode)->op.proc_vs_read((char *)page);
22979 + length = simple_read_from_buffer(buf, count, ppos,
22980 + (char *)page, length);
22986 +static ssize_t proc_vx_info_read(struct file *file, char __user *buf,
22987 + size_t count, loff_t *ppos)
22989 + struct inode *inode = file->f_dentry->d_inode;
22990 + struct vx_info *vxi = NULL;
22991 + xid_t xid = PROC_I(inode)->fd;
22992 + unsigned long page;
22993 + ssize_t length = 0;
22995 + if (count > PROC_BLOCK_SIZE)
22996 + count = PROC_BLOCK_SIZE;
22998 + /* fade that out as soon as stable */
23000 + vxi = lookup_vx_info(xid);
23004 + length = -ENOMEM;
23005 + if (!(page = __get_free_page(GFP_KERNEL)))
23008 + BUG_ON(!PROC_I(inode)->op.proc_vxi_read);
23009 + length = PROC_I(inode)->op.proc_vxi_read(vxi, (char *)page);
23012 + length = simple_read_from_buffer(buf, count, ppos,
23013 + (char *)page, length);
23017 + put_vx_info(vxi);
23022 +static ssize_t proc_nx_info_read(struct file *file, char __user *buf,
23023 + size_t count, loff_t *ppos)
23025 + struct inode *inode = file->f_dentry->d_inode;
23026 + struct nx_info *nxi = NULL;
23027 + nid_t nid = PROC_I(inode)->fd;
23028 + unsigned long page;
23029 + ssize_t length = 0;
23031 + if (count > PROC_BLOCK_SIZE)
23032 + count = PROC_BLOCK_SIZE;
23034 + /* fade that out as soon as stable */
23036 + nxi = lookup_nx_info(nid);
23040 + length = -ENOMEM;
23041 + if (!(page = __get_free_page(GFP_KERNEL)))
23044 + BUG_ON(!PROC_I(inode)->op.proc_nxi_read);
23045 + length = PROC_I(inode)->op.proc_nxi_read(nxi, (char *)page);
23048 + length = simple_read_from_buffer(buf, count, ppos,
23049 + (char *)page, length);
23053 + put_nx_info(nxi);
23060 +/* here comes the lower level */
23063 +#define NOD(NAME, MODE, IOP, FOP, OP) { \
23064 + .len = sizeof(NAME) - 1, \
23065 + .name = (NAME), \
23073 +#define DIR(NAME, MODE, OTYPE) \
23074 + NOD(NAME, (S_IFDIR | (MODE)), \
23075 + &proc_ ## OTYPE ## _inode_operations, \
23076 + &proc_ ## OTYPE ## _file_operations, { } )
23078 +#define INF(NAME, MODE, OTYPE) \
23079 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23080 + &proc_vs_info_file_operations, \
23081 + { .proc_vs_read = &proc_##OTYPE } )
23083 +#define VINF(NAME, MODE, OTYPE) \
23084 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23085 + &proc_vx_info_file_operations, \
23086 + { .proc_vxi_read = &proc_##OTYPE } )
23088 +#define NINF(NAME, MODE, OTYPE) \
23089 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23090 + &proc_nx_info_file_operations, \
23091 + { .proc_nxi_read = &proc_##OTYPE } )
23094 +static struct file_operations proc_vs_info_file_operations = {
23095 + .read = proc_vs_info_read,
23098 +static struct file_operations proc_vx_info_file_operations = {
23099 + .read = proc_vx_info_read,
23102 +static struct dentry_operations proc_xid_dentry_operations = {
23103 + .d_revalidate = proc_xid_revalidate,
23106 +static struct vs_entry vx_base_stuff[] = {
23107 + VINF("info", S_IRUGO, vxi_info),
23108 + VINF("status", S_IRUGO, vxi_status),
23109 + VINF("limit", S_IRUGO, vxi_limit),
23110 + VINF("sched", S_IRUGO, vxi_sched),
23111 + VINF("nsproxy", S_IRUGO, vxi_nsproxy0),
23112 + VINF("nsproxy1",S_IRUGO, vxi_nsproxy1),
23113 + VINF("cvirt", S_IRUGO, vxi_cvirt),
23114 + VINF("cacct", S_IRUGO, vxi_cacct),
23121 +static struct dentry *proc_xid_instantiate(struct inode *dir,
23122 + struct dentry *dentry, int id, void *ptr)
23124 + dentry->d_op = &proc_xid_dentry_operations;
23125 + return vs_proc_instantiate(dir, dentry, id, ptr);
23128 +static struct dentry *proc_xid_lookup(struct inode *dir,
23129 + struct dentry *dentry, struct nameidata *nd)
23131 + struct vs_entry *p = vx_base_stuff;
23132 + struct dentry *error = ERR_PTR(-ENOENT);
23134 + for (; p->name; p++) {
23135 + if (p->len != dentry->d_name.len)
23137 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23143 + error = proc_xid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23148 +static int proc_xid_readdir(struct file *filp,
23149 + void *dirent, filldir_t filldir)
23151 + struct dentry *dentry = filp->f_dentry;
23152 + struct inode *inode = dentry->d_inode;
23153 + struct vs_entry *p = vx_base_stuff;
23154 + int size = sizeof(vx_base_stuff) / sizeof(struct vs_entry);
23158 + pos = filp->f_pos;
23161 + ino = inode->i_ino;
23162 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23165 + /* fall through */
23167 + ino = parent_ino(dentry);
23168 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23171 + /* fall through */
23174 + if (index >= size)
23176 + for (p += index; p->name; p++) {
23177 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23178 + vs_proc_instantiate, PROC_I(inode)->fd, p))
23184 + filp->f_pos = pos;
23190 +static struct file_operations proc_nx_info_file_operations = {
23191 + .read = proc_nx_info_read,
23194 +static struct dentry_operations proc_nid_dentry_operations = {
23195 + .d_revalidate = proc_nid_revalidate,
23198 +static struct vs_entry nx_base_stuff[] = {
23199 + NINF("info", S_IRUGO, nxi_info),
23200 + NINF("status", S_IRUGO, nxi_status),
23205 +static struct dentry *proc_nid_instantiate(struct inode *dir,
23206 + struct dentry *dentry, int id, void *ptr)
23208 + dentry->d_op = &proc_nid_dentry_operations;
23209 + return vs_proc_instantiate(dir, dentry, id, ptr);
23212 +static struct dentry *proc_nid_lookup(struct inode *dir,
23213 + struct dentry *dentry, struct nameidata *nd)
23215 + struct vs_entry *p = nx_base_stuff;
23216 + struct dentry *error = ERR_PTR(-ENOENT);
23218 + for (; p->name; p++) {
23219 + if (p->len != dentry->d_name.len)
23221 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23227 + error = proc_nid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23232 +static int proc_nid_readdir(struct file *filp,
23233 + void *dirent, filldir_t filldir)
23235 + struct dentry *dentry = filp->f_dentry;
23236 + struct inode *inode = dentry->d_inode;
23237 + struct vs_entry *p = nx_base_stuff;
23238 + int size = sizeof(nx_base_stuff) / sizeof(struct vs_entry);
23242 + pos = filp->f_pos;
23245 + ino = inode->i_ino;
23246 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23249 + /* fall through */
23251 + ino = parent_ino(dentry);
23252 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23255 + /* fall through */
23258 + if (index >= size)
23260 + for (p += index; p->name; p++) {
23261 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23262 + vs_proc_instantiate, PROC_I(inode)->fd, p))
23268 + filp->f_pos = pos;
23273 +#define MAX_MULBY10 ((~0U - 9) / 10)
23275 +static inline int atovid(const char *str, int len)
23280 + while (len-- > 0) {
23285 + if (vid >= MAX_MULBY10)
23295 +/* now the upper level (virtual) */
23298 +static struct file_operations proc_xid_file_operations = {
23299 + .read = generic_read_dir,
23300 + .readdir = proc_xid_readdir,
23303 +static struct inode_operations proc_xid_inode_operations = {
23304 + .lookup = proc_xid_lookup,
23307 +static struct vs_entry vx_virtual_stuff[] = {
23308 + INF("info", S_IRUGO, virtual_info),
23309 + INF("status", S_IRUGO, virtual_status),
23310 + DIR(NULL, S_IRUGO | S_IXUGO, xid),
23314 +static struct dentry *proc_virtual_lookup(struct inode *dir,
23315 + struct dentry *dentry, struct nameidata *nd)
23317 + struct vs_entry *p = vx_virtual_stuff;
23318 + struct dentry *error = ERR_PTR(-ENOENT);
23321 + for (; p->name; p++) {
23322 + if (p->len != dentry->d_name.len)
23324 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23328 + goto instantiate;
23330 + id = atovid(dentry->d_name.name, dentry->d_name.len);
23331 + if ((id < 0) || !xid_is_hashed(id))
23335 + error = proc_xid_instantiate(dir, dentry, id, p);
23340 +static struct file_operations proc_nid_file_operations = {
23341 + .read = generic_read_dir,
23342 + .readdir = proc_nid_readdir,
23345 +static struct inode_operations proc_nid_inode_operations = {
23346 + .lookup = proc_nid_lookup,
23349 +static struct vs_entry nx_virtnet_stuff[] = {
23350 + INF("info", S_IRUGO, virtnet_info),
23351 + INF("status", S_IRUGO, virtnet_status),
23352 + DIR(NULL, S_IRUGO | S_IXUGO, nid),
23356 +static struct dentry *proc_virtnet_lookup(struct inode *dir,
23357 + struct dentry *dentry, struct nameidata *nd)
23359 + struct vs_entry *p = nx_virtnet_stuff;
23360 + struct dentry *error = ERR_PTR(-ENOENT);
23363 + for (; p->name; p++) {
23364 + if (p->len != dentry->d_name.len)
23366 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23370 + goto instantiate;
23372 + id = atovid(dentry->d_name.name, dentry->d_name.len);
23373 + if ((id < 0) || !nid_is_hashed(id))
23377 + error = proc_nid_instantiate(dir, dentry, id, p);
23383 +#define PROC_MAXVIDS 32
23385 +int proc_virtual_readdir(struct file *filp,
23386 + void *dirent, filldir_t filldir)
23388 + struct dentry *dentry = filp->f_dentry;
23389 + struct inode *inode = dentry->d_inode;
23390 + struct vs_entry *p = vx_virtual_stuff;
23391 + int size = sizeof(vx_virtual_stuff) / sizeof(struct vs_entry);
23393 + unsigned int xid_array[PROC_MAXVIDS];
23394 + char buf[PROC_NUMBUF];
23395 + unsigned int nr_xids, i;
23398 + pos = filp->f_pos;
23401 + ino = inode->i_ino;
23402 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23405 + /* fall through */
23407 + ino = parent_ino(dentry);
23408 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23411 + /* fall through */
23414 + if (index >= size)
23416 + for (p += index; p->name; p++) {
23417 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23418 + vs_proc_instantiate, 0, p))
23423 + index = pos - size;
23424 + p = &vx_virtual_stuff[size - 1];
23425 + nr_xids = get_xid_list(index, xid_array, PROC_MAXVIDS);
23426 + for (i = 0; i < nr_xids; i++) {
23427 + int n, xid = xid_array[i];
23428 + unsigned int j = PROC_NUMBUF;
23432 + buf[--j] = '0' + (n % 10);
23435 + if (proc_fill_cache(filp, dirent, filldir,
23436 + buf + j, PROC_NUMBUF - j,
23437 + vs_proc_instantiate, xid, p))
23443 + filp->f_pos = pos;
23447 +static int proc_virtual_getattr(struct vfsmount *mnt,
23448 + struct dentry *dentry, struct kstat *stat)
23450 + struct inode *inode = dentry->d_inode;
23452 + generic_fillattr(inode, stat);
23453 + stat->nlink = 2 + atomic_read(&vx_global_cactive);
23457 +static struct file_operations proc_virtual_dir_operations = {
23458 + .read = generic_read_dir,
23459 + .readdir = proc_virtual_readdir,
23462 +static struct inode_operations proc_virtual_dir_inode_operations = {
23463 + .getattr = proc_virtual_getattr,
23464 + .lookup = proc_virtual_lookup,
23471 +int proc_virtnet_readdir(struct file *filp,
23472 + void *dirent, filldir_t filldir)
23474 + struct dentry *dentry = filp->f_dentry;
23475 + struct inode *inode = dentry->d_inode;
23476 + struct vs_entry *p = nx_virtnet_stuff;
23477 + int size = sizeof(nx_virtnet_stuff) / sizeof(struct vs_entry);
23479 + unsigned int nid_array[PROC_MAXVIDS];
23480 + char buf[PROC_NUMBUF];
23481 + unsigned int nr_nids, i;
23484 + pos = filp->f_pos;
23487 + ino = inode->i_ino;
23488 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23491 + /* fall through */
23493 + ino = parent_ino(dentry);
23494 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23497 + /* fall through */
23500 + if (index >= size)
23502 + for (p += index; p->name; p++) {
23503 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23504 + vs_proc_instantiate, 0, p))
23509 + index = pos - size;
23510 + p = &nx_virtnet_stuff[size - 1];
23511 + nr_nids = get_nid_list(index, nid_array, PROC_MAXVIDS);
23512 + for (i = 0; i < nr_nids; i++) {
23513 + int n, nid = nid_array[i];
23514 + unsigned int j = PROC_NUMBUF;
23518 + buf[--j] = '0' + (n % 10);
23521 + if (proc_fill_cache(filp, dirent, filldir,
23522 + buf + j, PROC_NUMBUF - j,
23523 + vs_proc_instantiate, nid, p))
23529 + filp->f_pos = pos;
23533 +static int proc_virtnet_getattr(struct vfsmount *mnt,
23534 + struct dentry *dentry, struct kstat *stat)
23536 + struct inode *inode = dentry->d_inode;
23538 + generic_fillattr(inode, stat);
23539 + stat->nlink = 2 + atomic_read(&nx_global_cactive);
23543 +static struct file_operations proc_virtnet_dir_operations = {
23544 + .read = generic_read_dir,
23545 + .readdir = proc_virtnet_readdir,
23548 +static struct inode_operations proc_virtnet_dir_inode_operations = {
23549 + .getattr = proc_virtnet_getattr,
23550 + .lookup = proc_virtnet_lookup,
23555 +void proc_vx_init(void)
23557 + struct proc_dir_entry *ent;
23559 + ent = proc_mkdir("virtual", 0);
23561 + ent->proc_fops = &proc_virtual_dir_operations;
23562 + ent->proc_iops = &proc_virtual_dir_inode_operations;
23564 + proc_virtual = ent;
23566 + ent = proc_mkdir("virtnet", 0);
23568 + ent->proc_fops = &proc_virtnet_dir_operations;
23569 + ent->proc_iops = &proc_virtnet_dir_inode_operations;
23571 + proc_virtnet = ent;
23577 +/* per pid info */
23580 +int proc_pid_vx_info(struct task_struct *p, char *buffer)
23582 + struct vx_info *vxi;
23583 + char *orig = buffer;
23585 + buffer += sprintf(buffer, "XID:\t%d\n", vx_task_xid(p));
23587 + vxi = task_get_vx_info(p);
23591 + buffer += sprintf(buffer, "BCaps:\t");
23592 + buffer = print_cap_t(buffer, &vxi->vx_bcaps);
23593 + buffer += sprintf(buffer, "\n");
23594 + buffer += sprintf(buffer, "CCaps:\t%016llx\n",
23595 + (unsigned long long)vxi->vx_ccaps);
23596 + buffer += sprintf(buffer, "CFlags:\t%016llx\n",
23597 + (unsigned long long)vxi->vx_flags);
23598 + buffer += sprintf(buffer, "CIPid:\t%d\n", vxi->vx_initpid);
23600 + put_vx_info(vxi);
23602 + return buffer - orig;
23606 +int proc_pid_nx_info(struct task_struct *p, char *buffer)
23608 + struct nx_info *nxi;
23609 + struct nx_addr_v4 *v4a;
23610 +#ifdef CONFIG_IPV6
23611 + struct nx_addr_v6 *v6a;
23613 + char *orig = buffer;
23616 + buffer += sprintf(buffer, "NID:\t%d\n", nx_task_nid(p));
23618 + nxi = task_get_nx_info(p);
23622 + buffer += sprintf(buffer, "NCaps:\t%016llx\n",
23623 + (unsigned long long)nxi->nx_ncaps);
23624 + buffer += sprintf(buffer, "NFlags:\t%016llx\n",
23625 + (unsigned long long)nxi->nx_flags);
23627 + buffer += sprintf(buffer,
23628 + "V4Root[bcast]:\t" NIPQUAD_FMT "\n",
23629 + NIPQUAD(nxi->v4_bcast.s_addr));
23630 + buffer += sprintf (buffer,
23631 + "V4Root[lback]:\t" NIPQUAD_FMT "\n",
23632 + NIPQUAD(nxi->v4_lback.s_addr));
23633 + if (!NX_IPV4(nxi))
23635 + for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
23636 + buffer += sprintf(buffer, "V4Root[%d]:\t" NXAV4_FMT "\n",
23639 +#ifdef CONFIG_IPV6
23640 + if (!NX_IPV6(nxi))
23642 + for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
23643 + buffer += sprintf(buffer, "V6Root[%d]:\t" NXAV6_FMT "\n",
23647 + put_nx_info(nxi);
23649 + return buffer - orig;
23652 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/sched.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched.c
23653 --- linux-2.6.29.1/kernel/vserver/sched.c 1970-01-01 01:00:00.000000000 +0100
23654 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched.c 2009-02-22 22:54:26.000000000 +0100
23657 + * linux/kernel/vserver/sched.c
23659 + * Virtual Server: Scheduler Support
23661 + * Copyright (C) 2004-2007 Herbert Pötzl
23663 + * V0.01 adapted Sam Vilains version to 2.6.3
23664 + * V0.02 removed legacy interface
23665 + * V0.03 changed vcmds to vxi arg
23666 + * V0.04 removed older and legacy interfaces
23670 +#include <linux/vs_context.h>
23671 +#include <linux/vs_sched.h>
23672 +#include <linux/vserver/sched_cmd.h>
23674 +#include <asm/uaccess.h>
23677 +#define vxd_check_range(val, min, max) do { \
23678 + vxlprintk((val < min) || (val > max), \
23679 + "check_range(%ld,%ld,%ld)", \
23680 + (long)val, (long)min, (long)max, \
23681 + __FILE__, __LINE__); \
23685 +void vx_update_sched_param(struct _vx_sched *sched,
23686 + struct _vx_sched_pc *sched_pc)
23688 + unsigned int set_mask = sched->update_mask;
23690 + if (set_mask & VXSM_FILL_RATE)
23691 + sched_pc->fill_rate[0] = sched->fill_rate[0];
23692 + if (set_mask & VXSM_INTERVAL)
23693 + sched_pc->interval[0] = sched->interval[0];
23694 + if (set_mask & VXSM_FILL_RATE2)
23695 + sched_pc->fill_rate[1] = sched->fill_rate[1];
23696 + if (set_mask & VXSM_INTERVAL2)
23697 + sched_pc->interval[1] = sched->interval[1];
23698 + if (set_mask & VXSM_TOKENS)
23699 + sched_pc->tokens = sched->tokens;
23700 + if (set_mask & VXSM_TOKENS_MIN)
23701 + sched_pc->tokens_min = sched->tokens_min;
23702 + if (set_mask & VXSM_TOKENS_MAX)
23703 + sched_pc->tokens_max = sched->tokens_max;
23704 + if (set_mask & VXSM_PRIO_BIAS)
23705 + sched_pc->prio_bias = sched->prio_bias;
23707 + if (set_mask & VXSM_IDLE_TIME)
23708 + sched_pc->flags |= VXSF_IDLE_TIME;
23710 + sched_pc->flags &= ~VXSF_IDLE_TIME;
23713 + sched_pc->norm_time = jiffies;
23718 + * recalculate the context's scheduling tokens
23720 + * ret > 0 : number of tokens available
23721 + * ret < 0 : on hold, check delta_min[]
23722 + * -1 only jiffies
23723 + * -2 also idle time
23726 +int vx_tokens_recalc(struct _vx_sched_pc *sched_pc,
23727 + unsigned long *norm_time, unsigned long *idle_time, int delta_min[2])
23731 + int flags = sched_pc->flags;
23733 + /* how much time did pass? */
23734 + delta = *norm_time - sched_pc->norm_time;
23735 + // printk("@ %ld, %ld, %ld\n", *norm_time, sched_pc->norm_time, jiffies);
23736 + vxd_check_range(delta, 0, INT_MAX);
23738 + if (delta >= sched_pc->interval[0]) {
23739 + long tokens, integral;
23741 + /* calc integral token part */
23742 + tokens = delta / sched_pc->interval[0];
23743 + integral = tokens * sched_pc->interval[0];
23744 + tokens *= sched_pc->fill_rate[0];
23745 +#ifdef CONFIG_VSERVER_HARDCPU
23746 + delta_min[0] = delta - integral;
23747 + vxd_check_range(delta_min[0], 0, sched_pc->interval[0]);
23749 + /* advance time */
23750 + sched_pc->norm_time += delta;
23753 + sched_pc->tokens += tokens;
23754 + sched_pc->token_time += tokens;
23756 + delta_min[0] = delta;
23758 +#ifdef CONFIG_VSERVER_IDLETIME
23759 + if (!(flags & VXSF_IDLE_TIME))
23762 + /* how much was the idle skip? */
23763 + delta = *idle_time - sched_pc->idle_time;
23764 + vxd_check_range(delta, 0, INT_MAX);
23766 + if (delta >= sched_pc->interval[1]) {
23767 + long tokens, integral;
23769 + /* calc fair share token part */
23770 + tokens = delta / sched_pc->interval[1];
23771 + integral = tokens * sched_pc->interval[1];
23772 + tokens *= sched_pc->fill_rate[1];
23773 + delta_min[1] = delta - integral;
23774 + vxd_check_range(delta_min[1], 0, sched_pc->interval[1]);
23776 + /* advance idle time */
23777 + sched_pc->idle_time += integral;
23780 + sched_pc->tokens += tokens;
23781 + sched_pc->token_time += tokens;
23783 + delta_min[1] = delta;
23787 + /* clip at maximum */
23788 + if (sched_pc->tokens > sched_pc->tokens_max)
23789 + sched_pc->tokens = sched_pc->tokens_max;
23790 + tokens = sched_pc->tokens;
23792 + if ((flags & VXSF_ONHOLD)) {
23793 + /* can we unhold? */
23794 + if (tokens >= sched_pc->tokens_min) {
23795 + flags &= ~VXSF_ONHOLD;
23796 + sched_pc->hold_ticks +=
23797 + *norm_time - sched_pc->onhold;
23801 + /* put on hold? */
23802 + if (tokens <= 0) {
23803 + flags |= VXSF_ONHOLD;
23804 + sched_pc->onhold = *norm_time;
23808 + sched_pc->flags = flags;
23812 + tokens = sched_pc->tokens_min - tokens;
23813 + sched_pc->flags = flags;
23814 + // BUG_ON(tokens < 0); probably doesn't hold anymore
23816 +#ifdef CONFIG_VSERVER_HARDCPU
23817 + /* next interval? */
23818 + if (!sched_pc->fill_rate[0])
23819 + delta_min[0] = HZ;
23820 + else if (tokens > sched_pc->fill_rate[0])
23821 + delta_min[0] += sched_pc->interval[0] *
23822 + tokens / sched_pc->fill_rate[0];
23824 + delta_min[0] = sched_pc->interval[0] - delta_min[0];
23825 + vxd_check_range(delta_min[0], 0, INT_MAX);
23827 +#ifdef CONFIG_VSERVER_IDLETIME
23828 + if (!(flags & VXSF_IDLE_TIME))
23831 + /* next interval? */
23832 + if (!sched_pc->fill_rate[1])
23833 + delta_min[1] = HZ;
23834 + else if (tokens > sched_pc->fill_rate[1])
23835 + delta_min[1] += sched_pc->interval[1] *
23836 + tokens / sched_pc->fill_rate[1];
23838 + delta_min[1] = sched_pc->interval[1] - delta_min[1];
23839 + vxd_check_range(delta_min[1], 0, INT_MAX);
23844 +#endif /* CONFIG_VSERVER_IDLETIME */
23847 +#endif /* CONFIG_VSERVER_HARDCPU */
23850 +static inline unsigned long msec_to_ticks(unsigned long msec)
23852 + return msecs_to_jiffies(msec);
23855 +static inline unsigned long ticks_to_msec(unsigned long ticks)
23857 + return jiffies_to_msecs(ticks);
23860 +static inline unsigned long ticks_to_usec(unsigned long ticks)
23862 + return jiffies_to_usecs(ticks);
23866 +static int do_set_sched(struct vx_info *vxi, struct vcmd_sched_v5 *data)
23868 + unsigned int set_mask = data->mask;
23869 + unsigned int update_mask;
23872 + /* Sanity check data values */
23873 + if (data->tokens_max <= 0)
23874 + data->tokens_max = HZ;
23875 + if (data->tokens_min < 0)
23876 + data->tokens_min = HZ / 3;
23877 + if (data->tokens_min >= data->tokens_max)
23878 + data->tokens_min = data->tokens_max;
23880 + if (data->prio_bias > MAX_PRIO_BIAS)
23881 + data->prio_bias = MAX_PRIO_BIAS;
23882 + if (data->prio_bias < MIN_PRIO_BIAS)
23883 + data->prio_bias = MIN_PRIO_BIAS;
23885 + spin_lock(&vxi->sched.tokens_lock);
23887 + /* sync up on delayed updates */
23888 + for_each_cpu_mask(cpu, vxi->sched.update)
23889 + vx_update_sched_param(&vxi->sched,
23890 + &vx_per_cpu(vxi, sched_pc, cpu));
23892 + if (set_mask & VXSM_FILL_RATE)
23893 + vxi->sched.fill_rate[0] = data->fill_rate[0];
23894 + if (set_mask & VXSM_FILL_RATE2)
23895 + vxi->sched.fill_rate[1] = data->fill_rate[1];
23896 + if (set_mask & VXSM_INTERVAL)
23897 + vxi->sched.interval[0] = (set_mask & VXSM_MSEC) ?
23898 + msec_to_ticks(data->interval[0]) : data->interval[0];
23899 + if (set_mask & VXSM_INTERVAL2)
23900 + vxi->sched.interval[1] = (set_mask & VXSM_MSEC) ?
23901 + msec_to_ticks(data->interval[1]) : data->interval[1];
23902 + if (set_mask & VXSM_TOKENS)
23903 + vxi->sched.tokens = data->tokens;
23904 + if (set_mask & VXSM_TOKENS_MIN)
23905 + vxi->sched.tokens_min = data->tokens_min;
23906 + if (set_mask & VXSM_TOKENS_MAX)
23907 + vxi->sched.tokens_max = data->tokens_max;
23908 + if (set_mask & VXSM_PRIO_BIAS)
23909 + vxi->sched.prio_bias = data->prio_bias;
23911 + /* Sanity check rate/interval */
23912 + for (i = 0; i < 2; i++) {
23913 + if (data->fill_rate[i] < 0)
23914 + data->fill_rate[i] = 0;
23915 + if (data->interval[i] <= 0)
23916 + data->interval[i] = HZ;
23919 + update_mask = vxi->sched.update_mask & VXSM_SET_MASK;
23920 + update_mask |= (set_mask & (VXSM_SET_MASK | VXSM_IDLE_TIME));
23921 + vxi->sched.update_mask = update_mask;
23925 + if (set_mask & VXSM_CPU_ID) {
23926 + vxi->sched.update = cpumask_of_cpu(data->cpu_id);
23927 + cpus_and(vxi->sched.update, cpu_online_map,
23928 + vxi->sched.update);
23930 + vxi->sched.update = cpu_online_map;
23932 + /* forced reload? */
23933 + if (set_mask & VXSM_FORCE) {
23934 + for_each_cpu_mask(cpu, vxi->sched.update)
23935 + vx_update_sched_param(&vxi->sched,
23936 + &vx_per_cpu(vxi, sched_pc, cpu));
23937 + vxi->sched.update = CPU_MASK_NONE;
23940 + /* on UP we update immediately */
23941 + vx_update_sched_param(&vxi->sched,
23942 + &vx_per_cpu(vxi, sched_pc, 0));
23945 + spin_unlock(&vxi->sched.tokens_lock);
23950 +#define COPY_IDS(C) C(cpu_id); C(bucket_id)
23951 +#define COPY_PRI(C) C(prio_bias)
23952 +#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
23953 +#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
23954 + C(fill_rate[1]); C(interval[1]);
23956 +#define COPY_VALUE(name) vc_data.name = data->name
23958 +static int do_set_sched_v4(struct vx_info *vxi, struct vcmd_set_sched_v4 *data)
23960 + struct vcmd_sched_v5 vc_data;
23962 + vc_data.mask = data->set_mask;
23963 + COPY_IDS(COPY_VALUE);
23964 + COPY_PRI(COPY_VALUE);
23965 + COPY_TOK(COPY_VALUE);
23966 + vc_data.fill_rate[0] = vc_data.fill_rate[1] = data->fill_rate;
23967 + vc_data.interval[0] = vc_data.interval[1] = data->interval;
23968 + return do_set_sched(vxi, &vc_data);
23971 +int vc_set_sched_v4(struct vx_info *vxi, void __user *data)
23973 + struct vcmd_set_sched_v4 vc_data;
23975 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
23978 + return do_set_sched_v4(vxi, &vc_data);
23981 + /* latest interface is v5 */
23983 +int vc_set_sched(struct vx_info *vxi, void __user *data)
23985 + struct vcmd_sched_v5 vc_data;
23987 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
23990 + return do_set_sched(vxi, &vc_data);
23994 +#define COPY_PRI(C) C(prio_bias)
23995 +#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
23996 +#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
23997 + C(fill_rate[1]); C(interval[1]);
23999 +#define COPY_VALUE(name) vc_data.name = data->name
24002 +int vc_get_sched(struct vx_info *vxi, void __user *data)
24004 + struct vcmd_sched_v5 vc_data;
24006 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24009 + if (vc_data.mask & VXSM_CPU_ID) {
24010 + int cpu = vc_data.cpu_id;
24011 + struct _vx_sched_pc *data;
24013 + if (!cpu_possible(cpu))
24016 + data = &vx_per_cpu(vxi, sched_pc, cpu);
24017 + COPY_TOK(COPY_VALUE);
24018 + COPY_PRI(COPY_VALUE);
24019 + COPY_FRI(COPY_VALUE);
24021 + if (data->flags & VXSF_IDLE_TIME)
24022 + vc_data.mask |= VXSM_IDLE_TIME;
24024 + struct _vx_sched *data = &vxi->sched;
24026 + COPY_TOK(COPY_VALUE);
24027 + COPY_PRI(COPY_VALUE);
24028 + COPY_FRI(COPY_VALUE);
24031 + if (vc_data.mask & VXSM_MSEC) {
24032 + vc_data.interval[0] = ticks_to_msec(vc_data.interval[0]);
24033 + vc_data.interval[1] = ticks_to_msec(vc_data.interval[1]);
24036 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24042 +int vc_sched_info(struct vx_info *vxi, void __user *data)
24044 + struct vcmd_sched_info vc_data;
24047 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24050 + cpu = vc_data.cpu_id;
24051 + if (!cpu_possible(cpu))
24055 + struct _vx_sched_pc *sched_pc =
24056 + &vx_per_cpu(vxi, sched_pc, cpu);
24058 + vc_data.user_msec = ticks_to_msec(sched_pc->user_ticks);
24059 + vc_data.sys_msec = ticks_to_msec(sched_pc->sys_ticks);
24060 + vc_data.hold_msec = ticks_to_msec(sched_pc->hold_ticks);
24061 + vc_data.vavavoom = sched_pc->vavavoom;
24063 + vc_data.token_usec = ticks_to_usec(1);
24065 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24070 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/sched_init.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched_init.h
24071 --- linux-2.6.29.1/kernel/vserver/sched_init.h 1970-01-01 01:00:00.000000000 +0100
24072 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched_init.h 2009-02-22 22:54:26.000000000 +0100
24075 +static inline void vx_info_init_sched(struct _vx_sched *sched)
24077 + static struct lock_class_key tokens_lock_key;
24079 + /* scheduling; hard code starting values as constants */
24080 + sched->fill_rate[0] = 1;
24081 + sched->interval[0] = 4;
24082 + sched->fill_rate[1] = 1;
24083 + sched->interval[1] = 8;
24084 + sched->tokens = HZ >> 2;
24085 + sched->tokens_min = HZ >> 4;
24086 + sched->tokens_max = HZ >> 1;
24087 + sched->tokens_lock = SPIN_LOCK_UNLOCKED;
24088 + sched->prio_bias = 0;
24090 + lockdep_set_class(&sched->tokens_lock, &tokens_lock_key);
24094 +void vx_info_init_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24096 + sched_pc->fill_rate[0] = 1;
24097 + sched_pc->interval[0] = 4;
24098 + sched_pc->fill_rate[1] = 1;
24099 + sched_pc->interval[1] = 8;
24100 + sched_pc->tokens = HZ >> 2;
24101 + sched_pc->tokens_min = HZ >> 4;
24102 + sched_pc->tokens_max = HZ >> 1;
24103 + sched_pc->prio_bias = 0;
24104 + sched_pc->vavavoom = 0;
24105 + sched_pc->token_time = 0;
24106 + sched_pc->idle_time = 0;
24107 + sched_pc->norm_time = jiffies;
24109 + sched_pc->user_ticks = 0;
24110 + sched_pc->sys_ticks = 0;
24111 + sched_pc->hold_ticks = 0;
24114 +static inline void vx_info_exit_sched(struct _vx_sched *sched)
24120 +void vx_info_exit_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24124 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/sched_proc.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched_proc.h
24125 --- linux-2.6.29.1/kernel/vserver/sched_proc.h 1970-01-01 01:00:00.000000000 +0100
24126 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sched_proc.h 2009-02-22 22:54:26.000000000 +0100
24128 +#ifndef _VX_SCHED_PROC_H
24129 +#define _VX_SCHED_PROC_H
24133 +int vx_info_proc_sched(struct _vx_sched *sched, char *buffer)
24137 + length += sprintf(buffer,
24138 + "FillRate:\t%8d,%d\n"
24139 + "Interval:\t%8d,%d\n"
24140 + "TokensMin:\t%8d\n"
24141 + "TokensMax:\t%8d\n"
24142 + "PrioBias:\t%8d\n",
24143 + sched->fill_rate[0],
24144 + sched->fill_rate[1],
24145 + sched->interval[0],
24146 + sched->interval[1],
24147 + sched->tokens_min,
24148 + sched->tokens_max,
24149 + sched->prio_bias);
24154 +int vx_info_proc_sched_pc(struct _vx_sched_pc *sched_pc,
24155 + char *buffer, int cpu)
24159 + length += sprintf(buffer + length,
24160 + "cpu %d: %lld %lld %lld %ld %ld", cpu,
24161 + (unsigned long long)sched_pc->user_ticks,
24162 + (unsigned long long)sched_pc->sys_ticks,
24163 + (unsigned long long)sched_pc->hold_ticks,
24164 + sched_pc->token_time,
24165 + sched_pc->idle_time);
24166 + length += sprintf(buffer + length,
24167 + " %c%c %d %d %d %d/%d %d/%d",
24168 + (sched_pc->flags & VXSF_ONHOLD) ? 'H' : 'R',
24169 + (sched_pc->flags & VXSF_IDLE_TIME) ? 'I' : '-',
24170 + sched_pc->tokens,
24171 + sched_pc->tokens_min,
24172 + sched_pc->tokens_max,
24173 + sched_pc->fill_rate[0],
24174 + sched_pc->interval[0],
24175 + sched_pc->fill_rate[1],
24176 + sched_pc->interval[1]);
24177 + length += sprintf(buffer + length,
24179 + sched_pc->prio_bias,
24180 + sched_pc->vavavoom);
24184 +#endif /* _VX_SCHED_PROC_H */
24185 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/signal.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/signal.c
24186 --- linux-2.6.29.1/kernel/vserver/signal.c 1970-01-01 01:00:00.000000000 +0100
24187 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/signal.c 2009-02-22 22:54:26.000000000 +0100
24190 + * linux/kernel/vserver/signal.c
24192 + * Virtual Server: Signal Support
24194 + * Copyright (C) 2003-2007 Herbert Pötzl
24196 + * V0.01 broken out from vcontext V0.05
24197 + * V0.02 changed vcmds to vxi arg
24198 + * V0.03 adjusted siginfo for kill
24202 +#include <asm/uaccess.h>
24204 +#include <linux/vs_context.h>
24205 +#include <linux/vs_pid.h>
24206 +#include <linux/vserver/signal_cmd.h>
24209 +int vx_info_kill(struct vx_info *vxi, int pid, int sig)
24211 + int retval, count = 0;
24212 + struct task_struct *p;
24213 + struct siginfo *sip = SEND_SIG_PRIV;
24216 + vxdprintk(VXD_CBIT(misc, 4),
24217 + "vx_info_kill(%p[#%d],%d,%d)*",
24218 + vxi, vxi->vx_id, pid, sig);
24219 + read_lock(&tasklist_lock);
24223 + for_each_process(p) {
24226 + if (vx_task_xid(p) != vxi->vx_id || p->pid <= 1 ||
24227 + (pid && vxi->vx_initpid == p->pid))
24230 + err = group_send_sig_info(sig, sip, p);
24232 + if (err != -EPERM)
24238 + if (vxi->vx_initpid) {
24239 + pid = vxi->vx_initpid;
24240 + /* for now, only SIGINT to private init ... */
24241 + if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24242 + /* ... as long as there are tasks left */
24243 + (atomic_read(&vxi->vx_tasks) > 1))
24246 + /* fallthrough */
24248 + p = find_task_by_real_pid(pid);
24250 + if (vx_task_xid(p) == vxi->vx_id)
24251 + retval = group_send_sig_info(sig, sip, p);
24255 + read_unlock(&tasklist_lock);
24256 + vxdprintk(VXD_CBIT(misc, 4),
24257 + "vx_info_kill(%p[#%d],%d,%d,%ld) = %d",
24258 + vxi, vxi->vx_id, pid, sig, (long)sip, retval);
24262 +int vc_ctx_kill(struct vx_info *vxi, void __user *data)
24264 + struct vcmd_ctx_kill_v0 vc_data;
24266 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24269 + /* special check to allow guest shutdown */
24270 + if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24271 + /* forbid killall pid=0 when init is present */
24272 + (((vc_data.pid < 1) && vxi->vx_initpid) ||
24273 + (vc_data.pid > 1)))
24276 + return vx_info_kill(vxi, vc_data.pid, vc_data.sig);
24280 +static int __wait_exit(struct vx_info *vxi)
24282 + DECLARE_WAITQUEUE(wait, current);
24285 + add_wait_queue(&vxi->vx_wait, &wait);
24286 + set_current_state(TASK_INTERRUPTIBLE);
24289 + if (vx_info_state(vxi,
24290 + VXS_SHUTDOWN | VXS_HASHED | VXS_HELPER) == VXS_SHUTDOWN)
24292 + if (signal_pending(current)) {
24293 + ret = -ERESTARTSYS;
24300 + set_current_state(TASK_RUNNING);
24301 + remove_wait_queue(&vxi->vx_wait, &wait);
24307 +int vc_wait_exit(struct vx_info *vxi, void __user *data)
24309 + struct vcmd_wait_exit_v0 vc_data;
24312 + ret = __wait_exit(vxi);
24313 + vc_data.reboot_cmd = vxi->reboot_cmd;
24314 + vc_data.exit_code = vxi->exit_code;
24316 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24321 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/space.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/space.c
24322 --- linux-2.6.29.1/kernel/vserver/space.c 1970-01-01 01:00:00.000000000 +0100
24323 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/space.c 2009-04-03 21:25:31.000000000 +0200
24326 + * linux/kernel/vserver/space.c
24328 + * Virtual Server: Context Space Support
24330 + * Copyright (C) 2003-2007 Herbert Pötzl
24332 + * V0.01 broken out from context.c 0.07
24333 + * V0.02 added task locking for namespace
24334 + * V0.03 broken out vx_enter_namespace
24335 + * V0.04 added *space support and commands
24339 +#include <linux/utsname.h>
24340 +#include <linux/nsproxy.h>
24341 +#include <linux/err.h>
24342 +#include <asm/uaccess.h>
24344 +#include <linux/vs_context.h>
24345 +#include <linux/vserver/space.h>
24346 +#include <linux/vserver/space_cmd.h>
24348 +atomic_t vs_global_nsproxy = ATOMIC_INIT(0);
24349 +atomic_t vs_global_fs = ATOMIC_INIT(0);
24350 +atomic_t vs_global_mnt_ns = ATOMIC_INIT(0);
24351 +atomic_t vs_global_uts_ns = ATOMIC_INIT(0);
24352 +atomic_t vs_global_ipc_ns = ATOMIC_INIT(0);
24353 +atomic_t vs_global_user_ns = ATOMIC_INIT(0);
24354 +atomic_t vs_global_pid_ns = ATOMIC_INIT(0);
24357 +/* namespace functions */
24359 +#include <linux/mnt_namespace.h>
24360 +#include <linux/user_namespace.h>
24361 +#include <linux/pid_namespace.h>
24362 +#include <linux/ipc_namespace.h>
24363 +#include <net/net_namespace.h>
24366 +static const struct vcmd_space_mask_v1 space_mask_v0 = {
24367 + .mask = CLONE_FS |
24375 +static const struct vcmd_space_mask_v1 space_mask = {
24376 + .mask = CLONE_FS |
24381 +#ifdef CONFIG_PID_NS
24384 +#ifdef CONFIG_NET_NS
24390 +static const struct vcmd_space_mask_v1 default_space_mask = {
24391 + .mask = CLONE_FS |
24396 +#ifdef CONFIG_PID_NS
24403 + * build a new nsproxy mix
24404 + * assumes that both proxies are 'const'
24405 + * does not touch nsproxy refcounts
24406 + * will hold a reference on the result.
24409 +struct nsproxy *vs_mix_nsproxy(struct nsproxy *old_nsproxy,
24410 + struct nsproxy *new_nsproxy, unsigned long mask)
24412 + struct mnt_namespace *old_ns;
24413 + struct uts_namespace *old_uts;
24414 + struct ipc_namespace *old_ipc;
24415 +#ifdef CONFIG_PID_NS
24416 + struct pid_namespace *old_pid;
24418 +#ifdef CONFIG_NET_NS
24419 + struct net *old_net;
24421 + struct nsproxy *nsproxy;
24423 + nsproxy = copy_nsproxy(old_nsproxy);
24427 + if (mask & CLONE_NEWNS) {
24428 + old_ns = nsproxy->mnt_ns;
24429 + nsproxy->mnt_ns = new_nsproxy->mnt_ns;
24430 + if (nsproxy->mnt_ns)
24431 + get_mnt_ns(nsproxy->mnt_ns);
24435 + if (mask & CLONE_NEWUTS) {
24436 + old_uts = nsproxy->uts_ns;
24437 + nsproxy->uts_ns = new_nsproxy->uts_ns;
24438 + if (nsproxy->uts_ns)
24439 + get_uts_ns(nsproxy->uts_ns);
24443 + if (mask & CLONE_NEWIPC) {
24444 + old_ipc = nsproxy->ipc_ns;
24445 + nsproxy->ipc_ns = new_nsproxy->ipc_ns;
24446 + if (nsproxy->ipc_ns)
24447 + get_ipc_ns(nsproxy->ipc_ns);
24451 +#ifdef CONFIG_PID_NS
24452 + if (mask & CLONE_NEWPID) {
24453 + old_pid = nsproxy->pid_ns;
24454 + nsproxy->pid_ns = new_nsproxy->pid_ns;
24455 + if (nsproxy->pid_ns)
24456 + get_pid_ns(nsproxy->pid_ns);
24460 +#ifdef CONFIG_NET_NS
24461 + if (mask & CLONE_NEWNET) {
24462 + old_net = nsproxy->net_ns;
24463 + nsproxy->net_ns = new_nsproxy->net_ns;
24464 + if (nsproxy->net_ns)
24465 + get_net(nsproxy->net_ns);
24470 + put_mnt_ns(old_ns);
24472 + put_uts_ns(old_uts);
24474 + put_ipc_ns(old_ipc);
24475 +#ifdef CONFIG_PID_NS
24477 + put_pid_ns(old_pid);
24479 +#ifdef CONFIG_NET_NS
24481 + put_net(old_net);
24489 + * merge two nsproxy structs into a new one.
24490 + * will hold a reference on the result.
24494 +struct nsproxy *__vs_merge_nsproxy(struct nsproxy *old,
24495 + struct nsproxy *proxy, unsigned long mask)
24497 + struct nsproxy null_proxy = { .mnt_ns = NULL };
24503 + /* vs_mix_nsproxy returns with reference */
24504 + return vs_mix_nsproxy(old ? old : &null_proxy,
24507 + get_nsproxy(proxy);
24512 + * merge two fs structs into a new one.
24513 + * will take a reference on the result.
24517 +struct fs_struct *__vs_merge_fs(struct fs_struct *old,
24518 + struct fs_struct *fs, unsigned long mask)
24520 + if (!(mask & CLONE_FS)) {
24522 + atomic_inc(&old->count);
24529 + return copy_fs_struct(fs);
24533 +int vx_enter_space(struct vx_info *vxi, unsigned long mask, unsigned index)
24535 + struct nsproxy *proxy, *proxy_cur, *proxy_new;
24536 + struct fs_struct *fs, *fs_cur, *fs_new;
24539 + vxdprintk(VXD_CBIT(space, 8), "vx_enter_space(%p[#%u],0x%08lx,%d)",
24540 + vxi, vxi->vx_id, mask, index);
24542 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
24546 + mask = vxi->vx_nsmask[index];
24548 + if ((mask & vxi->vx_nsmask[index]) != mask)
24551 + proxy = vxi->vx_nsproxy[index];
24552 + fs = vxi->vx_fs[index];
24554 + task_lock(current);
24555 + fs_cur = current->fs;
24556 + atomic_inc(&fs_cur->count);
24557 + proxy_cur = current->nsproxy;
24558 + get_nsproxy(proxy_cur);
24559 + task_unlock(current);
24561 + fs_new = __vs_merge_fs(fs_cur, fs, mask);
24562 + if (IS_ERR(fs_new)) {
24563 + ret = PTR_ERR(fs_new);
24567 + proxy_new = __vs_merge_nsproxy(proxy_cur, proxy, mask);
24568 + if (IS_ERR(proxy_new)) {
24569 + ret = PTR_ERR(proxy_new);
24573 + fs_new = xchg(¤t->fs, fs_new);
24574 + proxy_new = xchg(¤t->nsproxy, proxy_new);
24578 + put_nsproxy(proxy_new);
24581 + put_fs_struct(fs_new);
24584 + put_nsproxy(proxy_cur);
24586 + put_fs_struct(fs_cur);
24591 +int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index)
24593 + struct nsproxy *proxy_vxi, *proxy_cur, *proxy_new;
24594 + struct fs_struct *fs_vxi, *fs_cur, *fs_new;
24597 + vxdprintk(VXD_CBIT(space, 8), "vx_set_space(%p[#%u],0x%08lx,%d)",
24598 + vxi, vxi->vx_id, mask, index);
24601 + mask = default_space_mask.mask;
24603 + if ((mask & space_mask.mask) != mask)
24606 + proxy_vxi = vxi->vx_nsproxy[index];
24607 + fs_vxi = vxi->vx_fs[index];
24609 + task_lock(current);
24610 + fs_cur = current->fs;
24611 + atomic_inc(&fs_cur->count);
24612 + proxy_cur = current->nsproxy;
24613 + get_nsproxy(proxy_cur);
24614 + task_unlock(current);
24616 + fs_new = __vs_merge_fs(fs_vxi, fs_cur, mask);
24617 + if (IS_ERR(fs_new)) {
24618 + ret = PTR_ERR(fs_new);
24622 + proxy_new = __vs_merge_nsproxy(proxy_vxi, proxy_cur, mask);
24623 + if (IS_ERR(proxy_new)) {
24624 + ret = PTR_ERR(proxy_new);
24628 + fs_new = xchg(&vxi->vx_fs[index], fs_new);
24629 + proxy_new = xchg(&vxi->vx_nsproxy[index], proxy_new);
24630 + vxi->vx_nsmask[index] |= mask;
24634 + put_nsproxy(proxy_new);
24637 + put_fs_struct(fs_new);
24640 + put_nsproxy(proxy_cur);
24642 + put_fs_struct(fs_cur);
24647 +int vc_enter_space_v1(struct vx_info *vxi, void __user *data)
24649 + struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
24651 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24654 + return vx_enter_space(vxi, vc_data.mask, 0);
24657 +int vc_enter_space(struct vx_info *vxi, void __user *data)
24659 + struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
24661 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24664 + if (vc_data.index >= VX_SPACES)
24667 + return vx_enter_space(vxi, vc_data.mask, vc_data.index);
24670 +int vc_set_space_v1(struct vx_info *vxi, void __user *data)
24672 + struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
24674 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24677 + return vx_set_space(vxi, vc_data.mask, 0);
24680 +int vc_set_space(struct vx_info *vxi, void __user *data)
24682 + struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
24684 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24687 + if (vc_data.index >= VX_SPACES)
24690 + return vx_set_space(vxi, vc_data.mask, vc_data.index);
24693 +int vc_get_space_mask(void __user *data, int type)
24695 + const struct vcmd_space_mask_v1 *mask;
24698 + mask = &space_mask_v0;
24699 + else if (type == 1)
24700 + mask = &space_mask;
24702 + mask = &default_space_mask;
24704 + vxdprintk(VXD_CBIT(space, 10),
24705 + "vc_get_space_mask(%d) = %08llx", type, mask->mask);
24707 + if (copy_to_user(data, mask, sizeof(*mask)))
24712 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/switch.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/switch.c
24713 --- linux-2.6.29.1/kernel/vserver/switch.c 1970-01-01 01:00:00.000000000 +0100
24714 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/switch.c 2009-03-24 14:59:18.000000000 +0100
24717 + * linux/kernel/vserver/switch.c
24719 + * Virtual Server: Syscall Switch
24721 + * Copyright (C) 2003-2007 Herbert Pötzl
24723 + * V0.01 syscall switch
24724 + * V0.02 added signal to context
24725 + * V0.03 added rlimit functions
24726 + * V0.04 added iattr, task/xid functions
24727 + * V0.05 added debug/history stuff
24728 + * V0.06 added compat32 layer
24729 + * V0.07 vcmd args and perms
24730 + * V0.08 added status commands
24731 + * V0.09 added tag commands
24732 + * V0.10 added oom bias
24733 + * V0.11 added device commands
24737 +#include <linux/vs_context.h>
24738 +#include <linux/vs_network.h>
24739 +#include <linux/vserver/switch.h>
24741 +#include "vci_config.h"
24745 +int vc_get_version(uint32_t id)
24747 + return VCI_VERSION;
24751 +int vc_get_vci(uint32_t id)
24753 + return vci_kernel_config();
24756 +#include <linux/vserver/context_cmd.h>
24757 +#include <linux/vserver/cvirt_cmd.h>
24758 +#include <linux/vserver/cacct_cmd.h>
24759 +#include <linux/vserver/limit_cmd.h>
24760 +#include <linux/vserver/network_cmd.h>
24761 +#include <linux/vserver/sched_cmd.h>
24762 +#include <linux/vserver/debug_cmd.h>
24763 +#include <linux/vserver/inode_cmd.h>
24764 +#include <linux/vserver/dlimit_cmd.h>
24765 +#include <linux/vserver/signal_cmd.h>
24766 +#include <linux/vserver/space_cmd.h>
24767 +#include <linux/vserver/tag_cmd.h>
24768 +#include <linux/vserver/device_cmd.h>
24770 +#include <linux/vserver/inode.h>
24771 +#include <linux/vserver/dlimit.h>
24774 +#ifdef CONFIG_COMPAT
24775 +#define __COMPAT(name, id, data, compat) \
24776 + (compat) ? name ## _x32(id, data) : name(id, data)
24777 +#define __COMPAT_NO_ID(name, data, compat) \
24778 + (compat) ? name ## _x32(data) : name(data)
24780 +#define __COMPAT(name, id, data, compat) \
24782 +#define __COMPAT_NO_ID(name, data, compat) \
24788 +long do_vcmd(uint32_t cmd, uint32_t id,
24789 + struct vx_info *vxi, struct nx_info *nxi,
24790 + void __user *data, int compat)
24794 + case VCMD_get_version:
24795 + return vc_get_version(id);
24796 + case VCMD_get_vci:
24797 + return vc_get_vci(id);
24799 + case VCMD_task_xid:
24800 + return vc_task_xid(id);
24801 + case VCMD_vx_info:
24802 + return vc_vx_info(vxi, data);
24804 + case VCMD_task_nid:
24805 + return vc_task_nid(id);
24806 + case VCMD_nx_info:
24807 + return vc_nx_info(nxi, data);
24809 + case VCMD_task_tag:
24810 + return vc_task_tag(id);
24812 + case VCMD_set_space_v1:
24813 + return vc_set_space_v1(vxi, data);
24814 + /* this is version 2 */
24815 + case VCMD_set_space:
24816 + return vc_set_space(vxi, data);
24818 + case VCMD_get_space_mask_v0:
24819 + return vc_get_space_mask(data, 0);
24820 + /* this is version 1 */
24821 + case VCMD_get_space_mask:
24822 + return vc_get_space_mask(data, 1);
24824 + case VCMD_get_space_default:
24825 + return vc_get_space_mask(data, -1);
24827 +#ifdef CONFIG_IA32_EMULATION
24828 + case VCMD_get_rlimit:
24829 + return __COMPAT(vc_get_rlimit, vxi, data, compat);
24830 + case VCMD_set_rlimit:
24831 + return __COMPAT(vc_set_rlimit, vxi, data, compat);
24833 + case VCMD_get_rlimit:
24834 + return vc_get_rlimit(vxi, data);
24835 + case VCMD_set_rlimit:
24836 + return vc_set_rlimit(vxi, data);
24838 + case VCMD_get_rlimit_mask:
24839 + return vc_get_rlimit_mask(id, data);
24840 + case VCMD_reset_minmax:
24841 + return vc_reset_minmax(vxi, data);
24843 + case VCMD_get_vhi_name:
24844 + return vc_get_vhi_name(vxi, data);
24845 + case VCMD_set_vhi_name:
24846 + return vc_set_vhi_name(vxi, data);
24848 + case VCMD_ctx_stat:
24849 + return vc_ctx_stat(vxi, data);
24850 + case VCMD_virt_stat:
24851 + return vc_virt_stat(vxi, data);
24852 + case VCMD_sock_stat:
24853 + return vc_sock_stat(vxi, data);
24854 + case VCMD_rlimit_stat:
24855 + return vc_rlimit_stat(vxi, data);
24857 + case VCMD_set_cflags:
24858 + return vc_set_cflags(vxi, data);
24859 + case VCMD_get_cflags:
24860 + return vc_get_cflags(vxi, data);
24862 + /* this is version 1 */
24863 + case VCMD_set_ccaps:
24864 + return vc_set_ccaps(vxi, data);
24865 + /* this is version 1 */
24866 + case VCMD_get_ccaps:
24867 + return vc_get_ccaps(vxi, data);
24868 + case VCMD_set_bcaps:
24869 + return vc_set_bcaps(vxi, data);
24870 + case VCMD_get_bcaps:
24871 + return vc_get_bcaps(vxi, data);
24873 + case VCMD_set_badness:
24874 + return vc_set_badness(vxi, data);
24875 + case VCMD_get_badness:
24876 + return vc_get_badness(vxi, data);
24878 + case VCMD_set_nflags:
24879 + return vc_set_nflags(nxi, data);
24880 + case VCMD_get_nflags:
24881 + return vc_get_nflags(nxi, data);
24883 + case VCMD_set_ncaps:
24884 + return vc_set_ncaps(nxi, data);
24885 + case VCMD_get_ncaps:
24886 + return vc_get_ncaps(nxi, data);
24888 + case VCMD_set_sched_v4:
24889 + return vc_set_sched_v4(vxi, data);
24890 + /* this is version 5 */
24891 + case VCMD_set_sched:
24892 + return vc_set_sched(vxi, data);
24893 + case VCMD_get_sched:
24894 + return vc_get_sched(vxi, data);
24895 + case VCMD_sched_info:
24896 + return vc_sched_info(vxi, data);
24898 + case VCMD_add_dlimit:
24899 + return __COMPAT(vc_add_dlimit, id, data, compat);
24900 + case VCMD_rem_dlimit:
24901 + return __COMPAT(vc_rem_dlimit, id, data, compat);
24902 + case VCMD_set_dlimit:
24903 + return __COMPAT(vc_set_dlimit, id, data, compat);
24904 + case VCMD_get_dlimit:
24905 + return __COMPAT(vc_get_dlimit, id, data, compat);
24907 + case VCMD_ctx_kill:
24908 + return vc_ctx_kill(vxi, data);
24910 + case VCMD_wait_exit:
24911 + return vc_wait_exit(vxi, data);
24913 + case VCMD_get_iattr:
24914 + return __COMPAT_NO_ID(vc_get_iattr, data, compat);
24915 + case VCMD_set_iattr:
24916 + return __COMPAT_NO_ID(vc_set_iattr, data, compat);
24918 + case VCMD_fget_iattr:
24919 + return vc_fget_iattr(id, data);
24920 + case VCMD_fset_iattr:
24921 + return vc_fset_iattr(id, data);
24923 + case VCMD_enter_space_v0:
24924 + return vc_enter_space_v1(vxi, NULL);
24925 + case VCMD_enter_space_v1:
24926 + return vc_enter_space_v1(vxi, data);
24927 + /* this is version 2 */
24928 + case VCMD_enter_space:
24929 + return vc_enter_space(vxi, data);
24931 + case VCMD_ctx_create_v0:
24932 + return vc_ctx_create(id, NULL);
24933 + case VCMD_ctx_create:
24934 + return vc_ctx_create(id, data);
24935 + case VCMD_ctx_migrate_v0:
24936 + return vc_ctx_migrate(vxi, NULL);
24937 + case VCMD_ctx_migrate:
24938 + return vc_ctx_migrate(vxi, data);
24940 + case VCMD_net_create_v0:
24941 + return vc_net_create(id, NULL);
24942 + case VCMD_net_create:
24943 + return vc_net_create(id, data);
24944 + case VCMD_net_migrate:
24945 + return vc_net_migrate(nxi, data);
24947 + case VCMD_tag_migrate:
24948 + return vc_tag_migrate(id);
24950 + case VCMD_net_add:
24951 + return vc_net_add(nxi, data);
24952 + case VCMD_net_remove:
24953 + return vc_net_remove(nxi, data);
24955 + case VCMD_net_add_ipv4:
24956 + return vc_net_add_ipv4(nxi, data);
24957 + case VCMD_net_remove_ipv4:
24958 + return vc_net_remove_ipv4(nxi, data);
24959 +#ifdef CONFIG_IPV6
24960 + case VCMD_net_add_ipv6:
24961 + return vc_net_add_ipv6(nxi, data);
24962 + case VCMD_net_remove_ipv6:
24963 + return vc_net_remove_ipv6(nxi, data);
24965 +/* case VCMD_add_match_ipv4:
24966 + return vc_add_match_ipv4(nxi, data);
24967 + case VCMD_get_match_ipv4:
24968 + return vc_get_match_ipv4(nxi, data);
24969 +#ifdef CONFIG_IPV6
24970 + case VCMD_add_match_ipv6:
24971 + return vc_add_match_ipv6(nxi, data);
24972 + case VCMD_get_match_ipv6:
24973 + return vc_get_match_ipv6(nxi, data);
24976 +#ifdef CONFIG_VSERVER_DEVICE
24977 + case VCMD_set_mapping:
24978 + return __COMPAT(vc_set_mapping, vxi, data, compat);
24979 + case VCMD_unset_mapping:
24980 + return __COMPAT(vc_unset_mapping, vxi, data, compat);
24982 +#ifdef CONFIG_VSERVER_HISTORY
24983 + case VCMD_dump_history:
24984 + return vc_dump_history(id);
24985 + case VCMD_read_history:
24986 + return __COMPAT(vc_read_history, id, data, compat);
24988 +#ifdef CONFIG_VSERVER_MONITOR
24989 + case VCMD_read_monitor:
24990 + return __COMPAT(vc_read_monitor, id, data, compat);
24993 + vxwprintk_task(1, "unimplemented VCMD_%02d_%d[%d]",
24994 + VC_CATEGORY(cmd), VC_COMMAND(cmd), VC_VERSION(cmd));
25000 +#define __VCMD(vcmd, _perm, _args, _flags) \
25001 + case VCMD_ ## vcmd: perm = _perm; \
25002 + args = _args; flags = _flags; break
25005 +#define VCA_NONE 0x00
25006 +#define VCA_VXI 0x01
25007 +#define VCA_NXI 0x02
25009 +#define VCF_NONE 0x00
25010 +#define VCF_INFO 0x01
25011 +#define VCF_ADMIN 0x02
25012 +#define VCF_ARES 0x06 /* includes admin */
25013 +#define VCF_SETUP 0x08
25015 +#define VCF_ZIDOK 0x10 /* zero id okay */
25019 +long do_vserver(uint32_t cmd, uint32_t id, void __user *data, int compat)
25022 + int permit = -1, state = 0;
25023 + int perm = -1, args = 0, flags = 0;
25024 + struct vx_info *vxi = NULL;
25025 + struct nx_info *nxi = NULL;
25028 + /* unpriviledged commands */
25029 + __VCMD(get_version, 0, VCA_NONE, 0);
25030 + __VCMD(get_vci, 0, VCA_NONE, 0);
25031 + __VCMD(get_rlimit_mask, 0, VCA_NONE, 0);
25032 + __VCMD(get_space_mask_v0,0, VCA_NONE, 0);
25033 + __VCMD(get_space_mask, 0, VCA_NONE, 0);
25034 + __VCMD(get_space_default,0, VCA_NONE, 0);
25036 + /* info commands */
25037 + __VCMD(task_xid, 2, VCA_NONE, 0);
25038 + __VCMD(reset_minmax, 2, VCA_VXI, 0);
25039 + __VCMD(vx_info, 3, VCA_VXI, VCF_INFO);
25040 + __VCMD(get_bcaps, 3, VCA_VXI, VCF_INFO);
25041 + __VCMD(get_ccaps, 3, VCA_VXI, VCF_INFO);
25042 + __VCMD(get_cflags, 3, VCA_VXI, VCF_INFO);
25043 + __VCMD(get_badness, 3, VCA_VXI, VCF_INFO);
25044 + __VCMD(get_vhi_name, 3, VCA_VXI, VCF_INFO);
25045 + __VCMD(get_rlimit, 3, VCA_VXI, VCF_INFO);
25047 + __VCMD(ctx_stat, 3, VCA_VXI, VCF_INFO);
25048 + __VCMD(virt_stat, 3, VCA_VXI, VCF_INFO);
25049 + __VCMD(sock_stat, 3, VCA_VXI, VCF_INFO);
25050 + __VCMD(rlimit_stat, 3, VCA_VXI, VCF_INFO);
25052 + __VCMD(task_nid, 2, VCA_NONE, 0);
25053 + __VCMD(nx_info, 3, VCA_NXI, VCF_INFO);
25054 + __VCMD(get_ncaps, 3, VCA_NXI, VCF_INFO);
25055 + __VCMD(get_nflags, 3, VCA_NXI, VCF_INFO);
25057 + __VCMD(task_tag, 2, VCA_NONE, 0);
25059 + __VCMD(get_iattr, 2, VCA_NONE, 0);
25060 + __VCMD(fget_iattr, 2, VCA_NONE, 0);
25061 + __VCMD(get_dlimit, 3, VCA_NONE, VCF_INFO);
25062 + __VCMD(get_sched, 3, VCA_VXI, VCF_INFO);
25063 + __VCMD(sched_info, 3, VCA_VXI, VCF_INFO | VCF_ZIDOK);
25065 + /* lower admin commands */
25066 + __VCMD(wait_exit, 4, VCA_VXI, VCF_INFO);
25067 + __VCMD(ctx_create_v0, 5, VCA_NONE, 0);
25068 + __VCMD(ctx_create, 5, VCA_NONE, 0);
25069 + __VCMD(ctx_migrate_v0, 5, VCA_VXI, VCF_ADMIN);
25070 + __VCMD(ctx_migrate, 5, VCA_VXI, VCF_ADMIN);
25071 + __VCMD(enter_space_v0, 5, VCA_VXI, VCF_ADMIN);
25072 + __VCMD(enter_space_v1, 5, VCA_VXI, VCF_ADMIN);
25073 + __VCMD(enter_space, 5, VCA_VXI, VCF_ADMIN);
25075 + __VCMD(net_create_v0, 5, VCA_NONE, 0);
25076 + __VCMD(net_create, 5, VCA_NONE, 0);
25077 + __VCMD(net_migrate, 5, VCA_NXI, VCF_ADMIN);
25079 + __VCMD(tag_migrate, 5, VCA_NONE, VCF_ADMIN);
25081 + /* higher admin commands */
25082 + __VCMD(ctx_kill, 6, VCA_VXI, VCF_ARES);
25083 + __VCMD(set_space_v1, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25084 + __VCMD(set_space, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25086 + __VCMD(set_ccaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25087 + __VCMD(set_bcaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25088 + __VCMD(set_cflags, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25089 + __VCMD(set_badness, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25091 + __VCMD(set_vhi_name, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25092 + __VCMD(set_rlimit, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25093 + __VCMD(set_sched, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25094 + __VCMD(set_sched_v4, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25096 + __VCMD(set_ncaps, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25097 + __VCMD(set_nflags, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25098 + __VCMD(net_add, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25099 + __VCMD(net_remove, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25100 + __VCMD(net_add_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25101 + __VCMD(net_remove_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25102 +#ifdef CONFIG_IPV6
25103 + __VCMD(net_add_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25104 + __VCMD(net_remove_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25106 + __VCMD(set_iattr, 7, VCA_NONE, 0);
25107 + __VCMD(fset_iattr, 7, VCA_NONE, 0);
25108 + __VCMD(set_dlimit, 7, VCA_NONE, VCF_ARES);
25109 + __VCMD(add_dlimit, 8, VCA_NONE, VCF_ARES);
25110 + __VCMD(rem_dlimit, 8, VCA_NONE, VCF_ARES);
25112 +#ifdef CONFIG_VSERVER_DEVICE
25113 + __VCMD(set_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25114 + __VCMD(unset_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25116 + /* debug level admin commands */
25117 +#ifdef CONFIG_VSERVER_HISTORY
25118 + __VCMD(dump_history, 9, VCA_NONE, 0);
25119 + __VCMD(read_history, 9, VCA_NONE, 0);
25121 +#ifdef CONFIG_VSERVER_MONITOR
25122 + __VCMD(read_monitor, 9, VCA_NONE, 0);
25129 + vxdprintk(VXD_CBIT(switch, 0),
25130 + "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]",
25131 + VC_CATEGORY(cmd), VC_COMMAND(cmd),
25132 + VC_VERSION(cmd), id, data, compat,
25133 + perm, args, flags);
25140 + if (!capable(CAP_CONTEXT))
25144 + /* moved here from the individual commands */
25146 + if ((perm > 1) && !capable(CAP_SYS_ADMIN))
25150 + /* vcmd involves resource management */
25152 + if ((flags & VCF_ARES) && !capable(CAP_SYS_RESOURCE))
25156 + /* various legacy exceptions */
25158 + /* will go away when spectator is a cap */
25159 + case VCMD_ctx_migrate_v0:
25160 + case VCMD_ctx_migrate:
25162 + current->xid = 1;
25168 + /* will go away when spectator is a cap */
25169 + case VCMD_net_migrate:
25171 + current->nid = 1;
25178 + /* vcmds are fine by default */
25181 + /* admin type vcmds require admin ... */
25182 + if (flags & VCF_ADMIN)
25183 + permit = vx_check(0, VS_ADMIN) ? 1 : 0;
25185 + /* ... but setup type vcmds override that */
25186 + if (!permit && (flags & VCF_SETUP))
25187 + permit = vx_flags(VXF_STATE_SETUP, 0) ? 2 : 0;
25195 + if (!id && (flags & VCF_ZIDOK))
25199 + if (args & VCA_VXI) {
25200 + vxi = lookup_vx_info(id);
25204 + if ((flags & VCF_ADMIN) &&
25205 + /* special case kill for shutdown */
25206 + (cmd != VCMD_ctx_kill) &&
25207 + /* can context be administrated? */
25208 + !vx_info_flags(vxi, VXF_STATE_ADMIN, 0)) {
25214 + if (args & VCA_NXI) {
25215 + nxi = lookup_nx_info(id);
25219 + if ((flags & VCF_ADMIN) &&
25220 + /* can context be administrated? */
25221 + !nx_info_flags(nxi, NXF_STATE_ADMIN, 0)) {
25228 + ret = do_vcmd(cmd, id, vxi, nxi, data, compat);
25231 + if ((args & VCA_NXI) && nxi)
25232 + put_nx_info(nxi);
25234 + if ((args & VCA_VXI) && vxi)
25235 + put_vx_info(vxi);
25237 + vxdprintk(VXD_CBIT(switch, 1),
25238 + "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]",
25239 + VC_CATEGORY(cmd), VC_COMMAND(cmd),
25240 + VC_VERSION(cmd), ret, ret, state, permit);
25245 +sys_vserver(uint32_t cmd, uint32_t id, void __user *data)
25247 + return do_vserver(cmd, id, data, 0);
25250 +#ifdef CONFIG_COMPAT
25253 +sys32_vserver(uint32_t cmd, uint32_t id, void __user *data)
25255 + return do_vserver(cmd, id, data, 1);
25258 +#endif /* CONFIG_COMPAT */
25259 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/sysctl.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sysctl.c
25260 --- linux-2.6.29.1/kernel/vserver/sysctl.c 1970-01-01 01:00:00.000000000 +0100
25261 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/sysctl.c 2009-02-22 22:54:26.000000000 +0100
25264 + * kernel/vserver/sysctl.c
25266 + * Virtual Context Support
25268 + * Copyright (C) 2004-2007 Herbert Pötzl
25270 + * V0.01 basic structure
25274 +#include <linux/module.h>
25275 +#include <linux/ctype.h>
25276 +#include <linux/sysctl.h>
25277 +#include <linux/parser.h>
25278 +#include <asm/uaccess.h>
25282 + CTL_DEBUG_ERROR = 0,
25283 + CTL_DEBUG_SWITCH = 1,
25298 +unsigned int vx_debug_switch = 0;
25299 +unsigned int vx_debug_xid = 0;
25300 +unsigned int vx_debug_nid = 0;
25301 +unsigned int vx_debug_tag = 0;
25302 +unsigned int vx_debug_net = 0;
25303 +unsigned int vx_debug_limit = 0;
25304 +unsigned int vx_debug_cres = 0;
25305 +unsigned int vx_debug_dlim = 0;
25306 +unsigned int vx_debug_quota = 0;
25307 +unsigned int vx_debug_cvirt = 0;
25308 +unsigned int vx_debug_space = 0;
25309 +unsigned int vx_debug_misc = 0;
25312 +static struct ctl_table_header *vserver_table_header;
25313 +static ctl_table vserver_root_table[];
25316 +void vserver_register_sysctl(void)
25318 + if (!vserver_table_header) {
25319 + vserver_table_header = register_sysctl_table(vserver_root_table);
25324 +void vserver_unregister_sysctl(void)
25326 + if (vserver_table_header) {
25327 + unregister_sysctl_table(vserver_table_header);
25328 + vserver_table_header = NULL;
25333 +static int proc_dodebug(ctl_table *table, int write,
25334 + struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
25336 + char tmpbuf[20], *p, c;
25337 + unsigned int value;
25338 + size_t left, len;
25340 + if ((*ppos && !write) || !*lenp) {
25348 + if (!access_ok(VERIFY_READ, buffer, left))
25350 + p = (char *)buffer;
25351 + while (left && __get_user(c, p) >= 0 && isspace(c))
25356 + if (left > sizeof(tmpbuf) - 1)
25358 + if (copy_from_user(tmpbuf, p, left))
25360 + tmpbuf[left] = '\0';
25362 + for (p = tmpbuf, value = 0; '0' <= *p && *p <= '9'; p++, left--)
25363 + value = 10 * value + (*p - '0');
25364 + if (*p && !isspace(*p))
25366 + while (left && isspace(*p))
25368 + *(unsigned int *)table->data = value;
25370 + if (!access_ok(VERIFY_WRITE, buffer, left))
25372 + len = sprintf(tmpbuf, "%d", *(unsigned int *)table->data);
25375 + if (__copy_to_user(buffer, tmpbuf, len))
25377 + if ((left -= len) > 0) {
25378 + if (put_user('\n', (char *)buffer + len))
25392 +#define CTL_ENTRY(ctl, name) \
25394 + .ctl_name = ctl, \
25395 + .procname = #name, \
25396 + .data = &vx_ ## name, \
25397 + .maxlen = sizeof(int), \
25399 + .proc_handler = &proc_dodebug, \
25400 + .strategy = &sysctl_intvec, \
25401 + .extra1 = &zero, \
25404 +static ctl_table vserver_debug_table[] = {
25405 + CTL_ENTRY(CTL_DEBUG_SWITCH, debug_switch),
25406 + CTL_ENTRY(CTL_DEBUG_XID, debug_xid),
25407 + CTL_ENTRY(CTL_DEBUG_NID, debug_nid),
25408 + CTL_ENTRY(CTL_DEBUG_TAG, debug_tag),
25409 + CTL_ENTRY(CTL_DEBUG_NET, debug_net),
25410 + CTL_ENTRY(CTL_DEBUG_LIMIT, debug_limit),
25411 + CTL_ENTRY(CTL_DEBUG_CRES, debug_cres),
25412 + CTL_ENTRY(CTL_DEBUG_DLIM, debug_dlim),
25413 + CTL_ENTRY(CTL_DEBUG_QUOTA, debug_quota),
25414 + CTL_ENTRY(CTL_DEBUG_CVIRT, debug_cvirt),
25415 + CTL_ENTRY(CTL_DEBUG_SPACE, debug_space),
25416 + CTL_ENTRY(CTL_DEBUG_MISC, debug_misc),
25417 + { .ctl_name = 0 }
25420 +static ctl_table vserver_root_table[] = {
25422 + .ctl_name = CTL_VSERVER,
25423 + .procname = "vserver",
25425 + .child = vserver_debug_table
25427 + { .ctl_name = 0 }
25431 +static match_table_t tokens = {
25432 + { CTL_DEBUG_SWITCH, "switch=%x" },
25433 + { CTL_DEBUG_XID, "xid=%x" },
25434 + { CTL_DEBUG_NID, "nid=%x" },
25435 + { CTL_DEBUG_TAG, "tag=%x" },
25436 + { CTL_DEBUG_NET, "net=%x" },
25437 + { CTL_DEBUG_LIMIT, "limit=%x" },
25438 + { CTL_DEBUG_CRES, "cres=%x" },
25439 + { CTL_DEBUG_DLIM, "dlim=%x" },
25440 + { CTL_DEBUG_QUOTA, "quota=%x" },
25441 + { CTL_DEBUG_CVIRT, "cvirt=%x" },
25442 + { CTL_DEBUG_SPACE, "space=%x" },
25443 + { CTL_DEBUG_MISC, "misc=%x" },
25444 + { CTL_DEBUG_ERROR, NULL }
25447 +#define HANDLE_CASE(id, name, val) \
25448 + case CTL_DEBUG_ ## id: \
25449 + vx_debug_ ## name = val; \
25450 + printk("vs_debug_" #name "=0x%x\n", val); \
25454 +static int __init vs_debug_setup(char *str)
25459 + printk("vs_debug_setup(%s)\n", str);
25460 + while ((p = strsep(&str, ",")) != NULL) {
25461 + substring_t args[MAX_OPT_ARGS];
25462 + unsigned int value;
25467 + token = match_token(p, tokens, args);
25468 + value = (token > 0) ? simple_strtoul(args[0].from, NULL, 0) : 0;
25471 + HANDLE_CASE(SWITCH, switch, value);
25472 + HANDLE_CASE(XID, xid, value);
25473 + HANDLE_CASE(NID, nid, value);
25474 + HANDLE_CASE(TAG, tag, value);
25475 + HANDLE_CASE(NET, net, value);
25476 + HANDLE_CASE(LIMIT, limit, value);
25477 + HANDLE_CASE(CRES, cres, value);
25478 + HANDLE_CASE(DLIM, dlim, value);
25479 + HANDLE_CASE(QUOTA, quota, value);
25480 + HANDLE_CASE(CVIRT, cvirt, value);
25481 + HANDLE_CASE(SPACE, space, value);
25482 + HANDLE_CASE(MISC, misc, value);
25491 +__setup("vsdebug=", vs_debug_setup);
25495 +EXPORT_SYMBOL_GPL(vx_debug_switch);
25496 +EXPORT_SYMBOL_GPL(vx_debug_xid);
25497 +EXPORT_SYMBOL_GPL(vx_debug_nid);
25498 +EXPORT_SYMBOL_GPL(vx_debug_net);
25499 +EXPORT_SYMBOL_GPL(vx_debug_limit);
25500 +EXPORT_SYMBOL_GPL(vx_debug_cres);
25501 +EXPORT_SYMBOL_GPL(vx_debug_dlim);
25502 +EXPORT_SYMBOL_GPL(vx_debug_quota);
25503 +EXPORT_SYMBOL_GPL(vx_debug_cvirt);
25504 +EXPORT_SYMBOL_GPL(vx_debug_space);
25505 +EXPORT_SYMBOL_GPL(vx_debug_misc);
25507 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/tag.c linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/tag.c
25508 --- linux-2.6.29.1/kernel/vserver/tag.c 1970-01-01 01:00:00.000000000 +0100
25509 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/tag.c 2009-02-22 22:54:26.000000000 +0100
25512 + * linux/kernel/vserver/tag.c
25514 + * Virtual Server: Shallow Tag Space
25516 + * Copyright (C) 2007 Herbert Pötzl
25518 + * V0.01 basic implementation
25522 +#include <linux/sched.h>
25523 +#include <linux/vserver/debug.h>
25524 +#include <linux/vs_pid.h>
25525 +#include <linux/vs_tag.h>
25527 +#include <linux/vserver/tag_cmd.h>
25530 +int dx_migrate_task(struct task_struct *p, tag_t tag)
25535 + vxdprintk(VXD_CBIT(tag, 5),
25536 + "dx_migrate_task(%p[#%d],#%d)", p, p->tag, tag);
25542 + vxdprintk(VXD_CBIT(tag, 5),
25543 + "moved task %p into [#%d]", p, tag);
25547 +/* vserver syscall commands below here */
25549 +/* taks xid and vx_info functions */
25552 +int vc_task_tag(uint32_t id)
25557 + struct task_struct *tsk;
25558 + read_lock(&tasklist_lock);
25559 + tsk = find_task_by_real_pid(id);
25560 + tag = (tsk) ? tsk->tag : -ESRCH;
25561 + read_unlock(&tasklist_lock);
25563 + tag = dx_current_tag();
25568 +int vc_tag_migrate(uint32_t tag)
25570 + return dx_migrate_task(current, tag & 0xFFFF);
25574 diff -NurpP --minimal linux-2.6.29.1/kernel/vserver/vci_config.h linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/vci_config.h
25575 --- linux-2.6.29.1/kernel/vserver/vci_config.h 1970-01-01 01:00:00.000000000 +0100
25576 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/kernel/vserver/vci_config.h 2009-02-22 22:54:26.000000000 +0100
25579 +/* interface version */
25581 +#define VCI_VERSION 0x00020304
25585 + VCI_KCBIT_NO_DYNAMIC = 0,
25587 + VCI_KCBIT_PROC_SECURE = 4,
25588 + VCI_KCBIT_HARDCPU = 5,
25589 + VCI_KCBIT_IDLELIMIT = 6,
25590 + VCI_KCBIT_IDLETIME = 7,
25592 + VCI_KCBIT_COWBL = 8,
25593 + VCI_KCBIT_FULLCOWBL = 9,
25594 + VCI_KCBIT_SPACES = 10,
25595 + VCI_KCBIT_NETV2 = 11,
25597 + VCI_KCBIT_DEBUG = 16,
25598 + VCI_KCBIT_HISTORY = 20,
25599 + VCI_KCBIT_TAGGED = 24,
25600 + VCI_KCBIT_PPTAG = 28,
25602 + VCI_KCBIT_MORE = 31,
25606 +static inline uint32_t vci_kernel_config(void)
25609 + (1 << VCI_KCBIT_NO_DYNAMIC) |
25611 + /* configured features */
25612 +#ifdef CONFIG_VSERVER_PROC_SECURE
25613 + (1 << VCI_KCBIT_PROC_SECURE) |
25615 +#ifdef CONFIG_VSERVER_HARDCPU
25616 + (1 << VCI_KCBIT_HARDCPU) |
25618 +#ifdef CONFIG_VSERVER_IDLELIMIT
25619 + (1 << VCI_KCBIT_IDLELIMIT) |
25621 +#ifdef CONFIG_VSERVER_IDLETIME
25622 + (1 << VCI_KCBIT_IDLETIME) |
25624 +#ifdef CONFIG_VSERVER_COWBL
25625 + (1 << VCI_KCBIT_COWBL) |
25626 + (1 << VCI_KCBIT_FULLCOWBL) |
25628 + (1 << VCI_KCBIT_SPACES) |
25629 + (1 << VCI_KCBIT_NETV2) |
25631 + /* debug options */
25632 +#ifdef CONFIG_VSERVER_DEBUG
25633 + (1 << VCI_KCBIT_DEBUG) |
25635 +#ifdef CONFIG_VSERVER_HISTORY
25636 + (1 << VCI_KCBIT_HISTORY) |
25639 + /* inode context tagging */
25640 +#if defined(CONFIG_TAGGING_NONE)
25641 + (0 << VCI_KCBIT_TAGGED) |
25642 +#elif defined(CONFIG_TAGGING_UID16)
25643 + (1 << VCI_KCBIT_TAGGED) |
25644 +#elif defined(CONFIG_TAGGING_GID16)
25645 + (2 << VCI_KCBIT_TAGGED) |
25646 +#elif defined(CONFIG_TAGGING_ID24)
25647 + (3 << VCI_KCBIT_TAGGED) |
25648 +#elif defined(CONFIG_TAGGING_INTERN)
25649 + (4 << VCI_KCBIT_TAGGED) |
25650 +#elif defined(CONFIG_TAGGING_RUNTIME)
25651 + (5 << VCI_KCBIT_TAGGED) |
25653 + (7 << VCI_KCBIT_TAGGED) |
25655 + (1 << VCI_KCBIT_PPTAG) |
25659 diff -NurpP --minimal linux-2.6.29.1/mm/filemap_xip.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/filemap_xip.c
25660 --- linux-2.6.29.1/mm/filemap_xip.c 2009-03-24 14:22:45.000000000 +0100
25661 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/filemap_xip.c 2009-03-24 14:48:36.000000000 +0100
25663 #include <linux/sched.h>
25664 #include <linux/seqlock.h>
25665 #include <linux/mutex.h>
25666 +#include <linux/vs_memory.h>
25667 #include <asm/tlbflush.h>
25668 #include <asm/io.h>
25670 diff -NurpP --minimal linux-2.6.29.1/mm/fremap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/fremap.c
25671 --- linux-2.6.29.1/mm/fremap.c 2009-03-24 14:22:45.000000000 +0100
25672 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/fremap.c 2009-03-24 14:48:36.000000000 +0100
25674 #include <linux/module.h>
25675 #include <linux/syscalls.h>
25676 #include <linux/mmu_notifier.h>
25677 +#include <linux/vs_memory.h>
25679 #include <asm/mmu_context.h>
25680 #include <asm/cacheflush.h>
25681 diff -NurpP --minimal linux-2.6.29.1/mm/hugetlb.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/hugetlb.c
25682 --- linux-2.6.29.1/mm/hugetlb.c 2009-03-24 14:22:45.000000000 +0100
25683 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/hugetlb.c 2009-03-24 14:48:36.000000000 +0100
25685 #include <asm/io.h>
25687 #include <linux/hugetlb.h>
25688 +#include <linux/vs_memory.h>
25689 #include "internal.h"
25691 const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
25692 diff -NurpP --minimal linux-2.6.29.1/mm/memory.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/memory.c
25693 --- linux-2.6.29.1/mm/memory.c 2009-04-15 22:53:12.000000000 +0200
25694 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/memory.c 2009-04-15 22:58:30.000000000 +0200
25696 #include <linux/kallsyms.h>
25697 #include <linux/swapops.h>
25698 #include <linux/elf.h>
25699 +// #include <linux/vs_memory.h>
25701 #include <asm/pgalloc.h>
25702 #include <asm/uaccess.h>
25703 @@ -612,6 +613,9 @@ static int copy_pte_range(struct mm_stru
25707 + if (!vx_rss_avail(dst_mm, ((end - addr)/PAGE_SIZE + 1)))
25711 rss[1] = rss[0] = 0;
25712 dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
25713 @@ -2435,6 +2439,11 @@ static int do_swap_page(struct mm_struct
25714 count_vm_event(PGMAJFAULT);
25717 + if (!vx_rss_avail(mm, 1)) {
25718 + ret = VM_FAULT_OOM;
25722 mark_page_accessed(page);
25725 @@ -2526,6 +2535,8 @@ static int do_anonymous_page(struct mm_s
25726 /* Allocate our own private page. */
25727 pte_unmap(page_table);
25729 + if (!vx_rss_avail(mm, 1))
25731 if (unlikely(anon_vma_prepare(vma)))
25733 page = alloc_zeroed_user_highpage_movable(vma, address);
25734 @@ -2793,6 +2804,7 @@ static inline int handle_pte_fault(struc
25738 + int ret = 0, type = VXPT_UNKNOWN;
25741 if (!pte_present(entry)) {
25742 @@ -2817,9 +2829,12 @@ static inline int handle_pte_fault(struc
25743 if (unlikely(!pte_same(*pte, entry)))
25745 if (write_access) {
25746 - if (!pte_write(entry))
25747 - return do_wp_page(mm, vma, address,
25748 + if (!pte_write(entry)) {
25749 + ret = do_wp_page(mm, vma, address,
25750 pte, pmd, ptl, entry);
25751 + type = VXPT_WRITE;
25754 entry = pte_mkdirty(entry);
25756 entry = pte_mkyoung(entry);
25757 @@ -2837,7 +2852,10 @@ static inline int handle_pte_fault(struc
25760 pte_unmap_unlock(pte, ptl);
25764 + vx_page_fault(mm, vma, type, ret);
25769 diff -NurpP --minimal linux-2.6.29.1/mm/mlock.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mlock.c
25770 --- linux-2.6.29.1/mm/mlock.c 2009-03-24 14:22:45.000000000 +0100
25771 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mlock.c 2009-03-28 05:08:26.000000000 +0100
25773 #include <linux/rmap.h>
25774 #include <linux/mmzone.h>
25775 #include <linux/hugetlb.h>
25776 +#include <linux/vs_memory.h>
25778 #include "internal.h"
25780 @@ -415,7 +416,7 @@ success:
25781 nr_pages = (end - start) >> PAGE_SHIFT;
25783 nr_pages = -nr_pages;
25784 - mm->locked_vm += nr_pages;
25785 + vx_vmlocked_add(mm, nr_pages);
25788 * vm_flags is protected by the mmap_sem held in write mode.
25789 @@ -492,7 +493,7 @@ static int do_mlock(unsigned long start,
25791 SYSCALL_DEFINE2(mlock, unsigned long, start, size_t, len)
25793 - unsigned long locked;
25794 + unsigned long locked, grow;
25795 unsigned long lock_limit;
25796 int error = -ENOMEM;
25798 @@ -505,8 +506,10 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
25799 len = PAGE_ALIGN(len + (start & ~PAGE_MASK));
25800 start &= PAGE_MASK;
25802 - locked = len >> PAGE_SHIFT;
25803 - locked += current->mm->locked_vm;
25804 + grow = len >> PAGE_SHIFT;
25805 + if (!vx_vmlocked_avail(current->mm, grow))
25807 + locked = current->mm->locked_vm + grow;
25809 lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
25810 lock_limit >>= PAGE_SHIFT;
25811 @@ -514,6 +517,7 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
25812 /* check against resource limits */
25813 if ((locked <= lock_limit) || capable(CAP_IPC_LOCK))
25814 error = do_mlock(start, len, 1);
25816 up_write(¤t->mm->mmap_sem);
25819 @@ -575,6 +579,8 @@ SYSCALL_DEFINE1(mlockall, int, flags)
25820 lock_limit >>= PAGE_SHIFT;
25823 + if (!vx_vmlocked_avail(current->mm, current->mm->total_vm))
25825 if (!(flags & MCL_CURRENT) || (current->mm->total_vm <= lock_limit) ||
25826 capable(CAP_IPC_LOCK))
25827 ret = do_mlockall(flags);
25828 @@ -652,8 +658,10 @@ void *alloc_locked_buffer(size_t size)
25832 - current->mm->total_vm += pgsz;
25833 - current->mm->locked_vm += pgsz;
25834 + // current->mm->total_vm += pgsz;
25835 + vx_vmpages_add(current->mm, pgsz);
25836 + // current->mm->locked_vm += pgsz;
25837 + vx_vmlocked_add(current->mm, pgsz);
25840 up_write(¤t->mm->mmap_sem);
25841 @@ -666,8 +674,10 @@ void release_locked_buffer(void *buffer,
25843 down_write(¤t->mm->mmap_sem);
25845 - current->mm->total_vm -= pgsz;
25846 - current->mm->locked_vm -= pgsz;
25847 + // current->mm->total_vm -= pgsz;
25848 + vx_vmpages_sub(current->mm, pgsz);
25849 + // current->mm->locked_vm -= pgsz;
25850 + vx_vmlocked_sub(current->mm, pgsz);
25852 up_write(¤t->mm->mmap_sem);
25854 diff -NurpP --minimal linux-2.6.29.1/mm/mmap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mmap.c
25855 --- linux-2.6.29.1/mm/mmap.c 2009-03-24 14:22:45.000000000 +0100
25856 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mmap.c 2009-03-24 14:48:36.000000000 +0100
25857 @@ -1219,7 +1219,8 @@ munmap_back:
25858 if (correct_wcount)
25859 atomic_inc(&inode->i_writecount);
25861 - mm->total_vm += len >> PAGE_SHIFT;
25862 + // mm->total_vm += len >> PAGE_SHIFT;
25863 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
25864 vm_stat_account(mm, vm_flags, file, len >> PAGE_SHIFT);
25865 if (vm_flags & VM_LOCKED) {
25867 @@ -1228,7 +1229,8 @@ out:
25868 long nr_pages = mlock_vma_pages_range(vma, addr, addr + len);
25870 return nr_pages; /* vma gone! */
25871 - mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
25872 + // mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
25873 + vx_vmlocked_add(mm, (len >> PAGE_SHIFT) - nr_pages);
25874 } else if ((flags & MAP_POPULATE) && !(flags & MAP_NONBLOCK))
25875 make_pages_present(addr, addr + len);
25877 @@ -1575,9 +1577,9 @@ static int acct_stack_growth(struct vm_a
25880 /* Ok, everything looks good - let it rip */
25881 - mm->total_vm += grow;
25882 + vx_vmpages_add(mm, grow);
25883 if (vma->vm_flags & VM_LOCKED)
25884 - mm->locked_vm += grow;
25885 + vx_vmlocked_add(mm, grow);
25886 vm_stat_account(mm, vma->vm_flags, vma->vm_file, grow);
25889 @@ -1752,7 +1754,8 @@ static void remove_vma_list(struct mm_st
25891 long nrpages = vma_pages(vma);
25893 - mm->total_vm -= nrpages;
25894 + // mm->total_vm -= nrpages;
25895 + vx_vmpages_sub(mm, nrpages);
25896 vm_stat_account(mm, vma->vm_flags, vma->vm_file, -nrpages);
25897 vma = remove_vma(vma);
25899 @@ -1924,7 +1927,8 @@ int do_munmap(struct mm_struct *mm, unsi
25900 struct vm_area_struct *tmp = vma;
25901 while (tmp && tmp->vm_start < end) {
25902 if (tmp->vm_flags & VM_LOCKED) {
25903 - mm->locked_vm -= vma_pages(tmp);
25904 + // mm->locked_vm -= vma_pages(tmp);
25905 + vx_vmlocked_sub(mm, vma_pages(tmp));
25906 munlock_vma_pages_all(tmp);
25908 tmp = tmp->vm_next;
25909 @@ -2013,6 +2017,8 @@ unsigned long do_brk(unsigned long addr,
25910 lock_limit >>= PAGE_SHIFT;
25911 if (locked > lock_limit && !capable(CAP_IPC_LOCK))
25913 + if (!vx_vmlocked_avail(mm, len >> PAGE_SHIFT))
25918 @@ -2039,7 +2045,8 @@ unsigned long do_brk(unsigned long addr,
25919 if (mm->map_count > sysctl_max_map_count)
25922 - if (security_vm_enough_memory(len >> PAGE_SHIFT))
25923 + if (security_vm_enough_memory(len >> PAGE_SHIFT) ||
25924 + !vx_vmpages_avail(mm, len >> PAGE_SHIFT))
25927 /* Can we just expand an old private anonymous mapping? */
25928 @@ -2065,10 +2072,13 @@ unsigned long do_brk(unsigned long addr,
25929 vma->vm_page_prot = vm_get_page_prot(flags);
25930 vma_link(mm, vma, prev, rb_link, rb_parent);
25932 - mm->total_vm += len >> PAGE_SHIFT;
25933 + // mm->total_vm += len >> PAGE_SHIFT;
25934 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
25936 if (flags & VM_LOCKED) {
25937 if (!mlock_vma_pages_range(vma, addr, addr + len))
25938 - mm->locked_vm += (len >> PAGE_SHIFT);
25939 + // mm->locked_vm += (len >> PAGE_SHIFT);
25940 + vx_vmlocked_add(mm, len >> PAGE_SHIFT);
25944 @@ -2111,6 +2121,11 @@ void exit_mmap(struct mm_struct *mm)
25945 free_pgtables(tlb, vma, FIRST_USER_ADDRESS, 0);
25946 tlb_finish_mmu(tlb, 0, end);
25948 + set_mm_counter(mm, file_rss, 0);
25949 + set_mm_counter(mm, anon_rss, 0);
25950 + vx_vmpages_sub(mm, mm->total_vm);
25951 + vx_vmlocked_sub(mm, mm->locked_vm);
25954 * Walk the list again, actually closing and freeing it,
25955 * with preemption enabled, without holding any MM locks.
25956 @@ -2150,7 +2165,8 @@ int insert_vm_struct(struct mm_struct *
25957 if (__vma && __vma->vm_start < vma->vm_end)
25959 if ((vma->vm_flags & VM_ACCOUNT) &&
25960 - security_vm_enough_memory_mm(mm, vma_pages(vma)))
25961 + (security_vm_enough_memory_mm(mm, vma_pages(vma)) ||
25962 + !vx_vmpages_avail(mm, vma_pages(vma))))
25964 vma_link(mm, vma, prev, rb_link, rb_parent);
25966 @@ -2226,6 +2242,8 @@ int may_expand_vm(struct mm_struct *mm,
25968 if (cur + npages > lim)
25970 + if (!vx_vmpages_avail(mm, npages))
25975 @@ -2303,8 +2321,7 @@ int install_special_mapping(struct mm_st
25979 - mm->total_vm += len >> PAGE_SHIFT;
25981 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
25985 diff -NurpP --minimal linux-2.6.29.1/mm/mremap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mremap.c
25986 --- linux-2.6.29.1/mm/mremap.c 2009-03-24 14:22:45.000000000 +0100
25987 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/mremap.c 2009-03-24 14:48:36.000000000 +0100
25989 #include <linux/security.h>
25990 #include <linux/syscalls.h>
25991 #include <linux/mmu_notifier.h>
25992 +#include <linux/vs_memory.h>
25994 #include <asm/uaccess.h>
25995 #include <asm/cacheflush.h>
25996 @@ -220,7 +221,7 @@ static unsigned long move_vma(struct vm_
25997 * If this were a serious issue, we'd add a flag to do_munmap().
25999 hiwater_vm = mm->hiwater_vm;
26000 - mm->total_vm += new_len >> PAGE_SHIFT;
26001 + vx_vmpages_add(mm, new_len >> PAGE_SHIFT);
26002 vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
26004 if (do_munmap(mm, old_addr, old_len) < 0) {
26005 @@ -238,7 +239,7 @@ static unsigned long move_vma(struct vm_
26008 if (vm_flags & VM_LOCKED) {
26009 - mm->locked_vm += new_len >> PAGE_SHIFT;
26010 + vx_vmlocked_add(mm, new_len >> PAGE_SHIFT);
26011 if (new_len > old_len)
26012 mlock_vma_pages_range(new_vma, new_addr + old_len,
26013 new_addr + new_len);
26014 @@ -349,6 +350,9 @@ unsigned long do_mremap(unsigned long ad
26016 if (locked > lock_limit && !capable(CAP_IPC_LOCK))
26018 + if (!vx_vmlocked_avail(current->mm,
26019 + (new_len - old_len) >> PAGE_SHIFT))
26022 if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
26024 @@ -377,10 +381,12 @@ unsigned long do_mremap(unsigned long ad
26025 vma_adjust(vma, vma->vm_start,
26026 addr + new_len, vma->vm_pgoff, NULL);
26028 - mm->total_vm += pages;
26029 + // mm->total_vm += pages;
26030 + vx_vmpages_add(mm, pages);
26031 vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
26032 if (vma->vm_flags & VM_LOCKED) {
26033 - mm->locked_vm += pages;
26034 + // mm->locked_vm += pages;
26035 + vx_vmlocked_add(mm, pages);
26036 mlock_vma_pages_range(vma, addr + old_len,
26039 diff -NurpP --minimal linux-2.6.29.1/mm/nommu.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/nommu.c
26040 --- linux-2.6.29.1/mm/nommu.c 2009-03-24 14:22:45.000000000 +0100
26041 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/nommu.c 2009-03-24 18:25:55.000000000 +0100
26042 @@ -1348,7 +1348,7 @@ unsigned long do_mmap_pgoff(struct file
26043 /* okay... we have a mapping; now we have to register it */
26044 result = vma->vm_start;
26046 - current->mm->total_vm += len >> PAGE_SHIFT;
26047 + vx_vmpages_add(current->mm, len >> PAGE_SHIFT);
26050 add_vma_to_mm(current->mm, vma);
26051 @@ -1614,7 +1614,7 @@ void exit_mmap(struct mm_struct *mm)
26055 - mm->total_vm = 0;
26056 + vx_vmpages_sub(mm, mm->total_vm);
26058 while ((vma = mm->mmap)) {
26059 mm->mmap = vma->vm_next;
26060 diff -NurpP --minimal linux-2.6.29.1/mm/oom_kill.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/oom_kill.c
26061 --- linux-2.6.29.1/mm/oom_kill.c 2009-03-24 14:22:45.000000000 +0100
26062 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/oom_kill.c 2009-03-24 14:48:36.000000000 +0100
26064 #include <linux/notifier.h>
26065 #include <linux/memcontrol.h>
26066 #include <linux/security.h>
26067 +#include <linux/vs_memory.h>
26069 int sysctl_panic_on_oom;
26070 int sysctl_oom_kill_allocating_task;
26071 @@ -72,6 +73,12 @@ unsigned long badness(struct task_struct
26072 points = mm->total_vm;
26075 + * add points for context badness
26078 + points += vx_badness(p, mm);
26081 * After this unlock we can no longer dereference local variable `mm'
26084 @@ -162,8 +169,8 @@ unsigned long badness(struct task_struct
26088 - printk(KERN_DEBUG "OOMkill: task %d (%s) got %lu points\n",
26089 - p->pid, p->comm, points);
26090 + printk(KERN_DEBUG "OOMkill: task %d:#%u (%s) got %d points\n",
26091 + task_pid_nr(p), p->xid, p->comm, points);
26095 @@ -326,8 +333,8 @@ static void __oom_kill_task(struct task_
26099 - printk(KERN_ERR "Killed process %d (%s)\n",
26100 - task_pid_nr(p), p->comm);
26101 + printk(KERN_ERR "Killed process %d:#%u (%s)\n",
26102 + task_pid_nr(p), p->xid, p->comm);
26105 * We give our sacrificial lamb high priority and access to
26106 @@ -410,8 +417,8 @@ static int oom_kill_process(struct task_
26110 - printk(KERN_ERR "%s: kill process %d (%s) score %li or a child\n",
26111 - message, task_pid_nr(p), p->comm, points);
26112 + printk(KERN_ERR "%s: kill process %d:#%u (%s) score %li or a child\n",
26113 + message, task_pid_nr(p), p->xid, p->comm, points);
26115 /* Try to kill a child first */
26116 list_for_each_entry(c, &p->children, sibling) {
26117 diff -NurpP --minimal linux-2.6.29.1/mm/page_alloc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/page_alloc.c
26118 --- linux-2.6.29.1/mm/page_alloc.c 2009-03-24 14:22:45.000000000 +0100
26119 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/page_alloc.c 2009-03-24 14:48:36.000000000 +0100
26121 #include <linux/page-isolation.h>
26122 #include <linux/page_cgroup.h>
26123 #include <linux/debugobjects.h>
26124 +#include <linux/vs_base.h>
26125 +#include <linux/vs_limit.h>
26127 #include <asm/tlbflush.h>
26128 #include <asm/div64.h>
26129 @@ -1839,6 +1841,9 @@ void si_meminfo(struct sysinfo *val)
26130 val->totalhigh = totalhigh_pages;
26131 val->freehigh = nr_free_highpages();
26132 val->mem_unit = PAGE_SIZE;
26134 + if (vx_flags(VXF_VIRT_MEM, 0))
26135 + vx_vsi_meminfo(val);
26138 EXPORT_SYMBOL(si_meminfo);
26139 @@ -1859,6 +1864,9 @@ void si_meminfo_node(struct sysinfo *val
26142 val->mem_unit = PAGE_SIZE;
26144 + if (vx_flags(VXF_VIRT_MEM, 0))
26145 + vx_vsi_meminfo(val);
26149 diff -NurpP --minimal linux-2.6.29.1/mm/rmap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/rmap.c
26150 --- linux-2.6.29.1/mm/rmap.c 2009-03-24 14:22:45.000000000 +0100
26151 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/rmap.c 2009-03-24 18:26:27.000000000 +0100
26153 #include <linux/memcontrol.h>
26154 #include <linux/mmu_notifier.h>
26155 #include <linux/migrate.h>
26156 +#include <linux/vs_memory.h>
26158 #include <asm/tlbflush.h>
26160 diff -NurpP --minimal linux-2.6.29.1/mm/shmem.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/shmem.c
26161 --- linux-2.6.29.1/mm/shmem.c 2009-03-24 14:22:45.000000000 +0100
26162 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/shmem.c 2009-03-24 14:48:36.000000000 +0100
26163 @@ -1757,7 +1757,7 @@ static int shmem_statfs(struct dentry *d
26165 struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
26167 - buf->f_type = TMPFS_MAGIC;
26168 + buf->f_type = TMPFS_SUPER_MAGIC;
26169 buf->f_bsize = PAGE_CACHE_SIZE;
26170 buf->f_namelen = NAME_MAX;
26171 spin_lock(&sbinfo->stat_lock);
26172 @@ -2326,7 +2326,7 @@ static int shmem_fill_super(struct super
26173 sb->s_maxbytes = SHMEM_MAX_BYTES;
26174 sb->s_blocksize = PAGE_CACHE_SIZE;
26175 sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
26176 - sb->s_magic = TMPFS_MAGIC;
26177 + sb->s_magic = TMPFS_SUPER_MAGIC;
26178 sb->s_op = &shmem_ops;
26179 sb->s_time_gran = 1;
26180 #ifdef CONFIG_TMPFS_POSIX_ACL
26181 diff -NurpP --minimal linux-2.6.29.1/mm/slab.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/slab.c
26182 --- linux-2.6.29.1/mm/slab.c 2009-03-24 14:22:45.000000000 +0100
26183 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/slab.c 2009-03-24 14:48:36.000000000 +0100
26184 @@ -509,6 +509,8 @@ struct kmem_cache {
26185 #define STATS_INC_FREEMISS(x) do { } while (0)
26188 +#include "slab_vs.h"
26193 @@ -3275,6 +3277,7 @@ retry:
26195 obj = slab_get_obj(cachep, slabp, nodeid);
26196 check_slabp(cachep, slabp);
26197 + vx_slab_alloc(cachep, flags);
26198 l3->free_objects--;
26199 /* move slabp to correct slabp list: */
26200 list_del(&slabp->list);
26201 @@ -3347,6 +3350,7 @@ __cache_alloc_node(struct kmem_cache *ca
26202 /* ___cache_alloc_node can fall back to other nodes */
26203 ptr = ____cache_alloc_node(cachep, flags, nodeid);
26205 + vx_slab_alloc(cachep, flags);
26206 local_irq_restore(save_flags);
26207 ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller);
26209 @@ -3518,6 +3522,7 @@ static inline void __cache_free(struct k
26212 objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0));
26213 + vx_slab_free(cachep);
26216 * Skip calling cache_free_alien() when the platform is not numa.
26217 diff -NurpP --minimal linux-2.6.29.1/mm/slab_vs.h linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/slab_vs.h
26218 --- linux-2.6.29.1/mm/slab_vs.h 1970-01-01 01:00:00.000000000 +0100
26219 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/slab_vs.h 2009-02-22 22:54:26.000000000 +0100
26222 +#include <linux/vserver/context.h>
26224 +#include <linux/vs_context.h>
26227 +void vx_slab_alloc(struct kmem_cache *cachep, gfp_t flags)
26229 + int what = gfp_zone(cachep->gfpflags);
26231 + if (!current->vx_info)
26234 + atomic_add(cachep->buffer_size, ¤t->vx_info->cacct.slab[what]);
26238 +void vx_slab_free(struct kmem_cache *cachep)
26240 + int what = gfp_zone(cachep->gfpflags);
26242 + if (!current->vx_info)
26245 + atomic_sub(cachep->buffer_size, ¤t->vx_info->cacct.slab[what]);
26248 diff -NurpP --minimal linux-2.6.29.1/mm/swapfile.c linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/swapfile.c
26249 --- linux-2.6.29.1/mm/swapfile.c 2009-03-24 14:22:45.000000000 +0100
26250 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/mm/swapfile.c 2009-03-24 18:27:15.000000000 +0100
26252 #include <asm/tlbflush.h>
26253 #include <linux/swapops.h>
26254 #include <linux/page_cgroup.h>
26255 +#include <linux/vs_base.h>
26256 +#include <linux/vs_memory.h>
26258 static DEFINE_SPINLOCK(swap_lock);
26259 static unsigned int nr_swapfiles;
26260 @@ -1935,6 +1937,8 @@ void si_swapinfo(struct sysinfo *val)
26261 val->freeswap = nr_swap_pages + nr_to_be_unused;
26262 val->totalswap = total_swap_pages + nr_to_be_unused;
26263 spin_unlock(&swap_lock);
26264 + if (vx_flags(VXF_VIRT_MEM, 0))
26265 + vx_vsi_swapinfo(val);
26269 diff -NurpP --minimal linux-2.6.29.1/net/core/dev.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/dev.c
26270 --- linux-2.6.29.1/net/core/dev.c 2009-04-15 22:53:12.000000000 +0200
26271 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/dev.c 2009-04-15 22:58:30.000000000 +0200
26272 @@ -126,6 +126,7 @@
26273 #include <linux/in.h>
26274 #include <linux/jhash.h>
26275 #include <linux/random.h>
26276 +#include <linux/vs_inet.h>
26278 #include "net-sysfs.h"
26280 @@ -2853,6 +2854,8 @@ static int dev_ifconf(struct net *net, c
26283 for_each_netdev(net, dev) {
26284 + if (!nx_dev_visible(current->nx_info, dev))
26286 for (i = 0; i < NPROTO; i++) {
26287 if (gifconf_list[i]) {
26289 @@ -2921,6 +2924,9 @@ static void dev_seq_printf_stats(struct
26291 const struct net_device_stats *stats = dev_get_stats(dev);
26293 + if (!nx_dev_visible(current->nx_info, dev))
26296 seq_printf(seq, "%6s:%8lu %7lu %4lu %4lu %4lu %5lu %10lu %9lu "
26297 "%8lu %7lu %4lu %4lu %4lu %5lu %7lu %10lu\n",
26298 dev->name, stats->rx_bytes, stats->rx_packets,
26299 @@ -4862,6 +4868,15 @@ int dev_change_net_namespace(struct net_
26303 +#ifdef CONFIG_SYSFS
26304 + /* Don't allow real devices to be moved when sysfs
26308 + if (dev->dev.parent)
26312 /* Ensure the device has been registrered */
26314 if (dev->reg_state != NETREG_REGISTERED)
26315 @@ -4921,6 +4936,8 @@ int dev_change_net_namespace(struct net_
26317 netdev_unregister_kobject(dev);
26319 + netdev_unregister_kobject(dev);
26321 /* Actually switch the network namespace */
26322 dev_net_set(dev, net);
26324 diff -NurpP --minimal linux-2.6.29.1/net/core/net-sysfs.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/net-sysfs.c
26325 --- linux-2.6.29.1/net/core/net-sysfs.c 2009-03-24 14:22:46.000000000 +0100
26326 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/net-sysfs.c 2009-03-24 18:30:01.000000000 +0100
26327 @@ -512,6 +512,9 @@ int netdev_register_kobject(struct net_d
26328 if (dev_net(net) != &init_net)
26331 + if (dev_net(net) != &init_net)
26334 return device_add(dev);
26337 diff -NurpP --minimal linux-2.6.29.1/net/core/rtnetlink.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/rtnetlink.c
26338 --- linux-2.6.29.1/net/core/rtnetlink.c 2009-03-24 14:22:46.000000000 +0100
26339 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/rtnetlink.c 2009-03-24 14:48:36.000000000 +0100
26340 @@ -690,6 +690,8 @@ static int rtnl_dump_ifinfo(struct sk_bu
26343 for_each_netdev(net, dev) {
26344 + if (!nx_dev_visible(skb->sk->sk_nx_info, dev))
26348 if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK,
26349 @@ -1235,6 +1237,9 @@ void rtmsg_ifinfo(int type, struct net_d
26350 struct sk_buff *skb;
26351 int err = -ENOBUFS;
26353 + if (!nx_dev_visible(current->nx_info, dev))
26356 skb = nlmsg_new(if_nlmsg_size(dev), GFP_KERNEL);
26359 diff -NurpP --minimal linux-2.6.29.1/net/core/sock.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/sock.c
26360 --- linux-2.6.29.1/net/core/sock.c 2009-03-24 14:22:46.000000000 +0100
26361 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/core/sock.c 2009-03-24 14:48:36.000000000 +0100
26362 @@ -124,6 +124,10 @@
26363 #include <linux/ipsec.h>
26365 #include <linux/filter.h>
26366 +#include <linux/vs_socket.h>
26367 +#include <linux/vs_limit.h>
26368 +#include <linux/vs_context.h>
26369 +#include <linux/vs_network.h>
26372 #include <net/tcp.h>
26373 @@ -900,6 +904,8 @@ static struct sock *sk_prot_alloc(struct
26374 if (!try_module_get(prot->owner))
26377 + sock_vx_init(sk);
26378 + sock_nx_init(sk);
26382 @@ -976,6 +982,11 @@ void sk_free(struct sock *sk)
26383 __func__, atomic_read(&sk->sk_omem_alloc));
26385 put_net(sock_net(sk));
26387 + clr_vx_info(&sk->sk_vx_info);
26389 + clr_nx_info(&sk->sk_nx_info);
26391 sk_prot_free(sk->sk_prot_creator, sk);
26394 @@ -1011,6 +1022,8 @@ struct sock *sk_clone(const struct sock
26397 get_net(sock_net(newsk));
26398 + sock_vx_init(newsk);
26399 + sock_nx_init(newsk);
26400 sk_node_init(&newsk->sk_node);
26401 sock_lock_init(newsk);
26402 bh_lock_sock(newsk);
26403 @@ -1057,6 +1070,12 @@ struct sock *sk_clone(const struct sock
26404 newsk->sk_priority = 0;
26405 atomic_set(&newsk->sk_refcnt, 2);
26407 + set_vx_info(&newsk->sk_vx_info, sk->sk_vx_info);
26408 + newsk->sk_xid = sk->sk_xid;
26409 + vx_sock_inc(newsk);
26410 + set_nx_info(&newsk->sk_nx_info, sk->sk_nx_info);
26411 + newsk->sk_nid = sk->sk_nid;
26414 * Increment the counter in the same struct proto as the master
26415 * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
26416 @@ -1744,6 +1763,11 @@ void sock_init_data(struct socket *sock,
26418 sk->sk_stamp = ktime_set(-1L, 0);
26420 + set_vx_info(&sk->sk_vx_info, current->vx_info);
26421 + sk->sk_xid = vx_current_xid();
26423 + set_nx_info(&sk->sk_nx_info, current->nx_info);
26424 + sk->sk_nid = nx_current_nid();
26425 atomic_set(&sk->sk_refcnt, 1);
26426 atomic_set(&sk->sk_drops, 0);
26428 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/af_inet.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/af_inet.c
26429 --- linux-2.6.29.1/net/ipv4/af_inet.c 2009-03-24 14:22:46.000000000 +0100
26430 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/af_inet.c 2009-03-24 14:48:37.000000000 +0100
26431 @@ -115,6 +115,7 @@
26432 #ifdef CONFIG_IP_MROUTE
26433 #include <linux/mroute.h>
26435 +#include <linux/vs_limit.h>
26437 extern void ip_mc_drop_socket(struct sock *sk);
26439 @@ -325,9 +326,12 @@ lookup_protocol:
26443 + if ((protocol == IPPROTO_ICMP) &&
26444 + nx_capable(answer->capability, NXC_RAW_ICMP))
26446 if (answer->capability > 0 && !capable(answer->capability))
26447 goto out_rcu_unlock;
26450 err = -EAFNOSUPPORT;
26451 if (!inet_netns_ok(net, protocol))
26452 goto out_rcu_unlock;
26453 @@ -445,6 +449,7 @@ int inet_bind(struct socket *sock, struc
26454 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
26455 struct sock *sk = sock->sk;
26456 struct inet_sock *inet = inet_sk(sk);
26457 + struct nx_v4_sock_addr nsa;
26458 unsigned short snum;
26461 @@ -458,7 +463,11 @@ int inet_bind(struct socket *sock, struc
26462 if (addr_len < sizeof(struct sockaddr_in))
26465 - chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
26466 + err = v4_map_sock_addr(inet, addr, &nsa);
26470 + chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
26472 /* Not specified by any standard per-se, however it breaks too
26473 * many applications when removed. It is unfortunate since
26474 @@ -470,7 +479,7 @@ int inet_bind(struct socket *sock, struc
26475 err = -EADDRNOTAVAIL;
26476 if (!sysctl_ip_nonlocal_bind &&
26477 !(inet->freebind || inet->transparent) &&
26478 - addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
26479 + nsa.saddr != htonl(INADDR_ANY) &&
26480 chk_addr_ret != RTN_LOCAL &&
26481 chk_addr_ret != RTN_MULTICAST &&
26482 chk_addr_ret != RTN_BROADCAST)
26483 @@ -495,7 +504,7 @@ int inet_bind(struct socket *sock, struc
26484 if (sk->sk_state != TCP_CLOSE || inet->num)
26485 goto out_release_sock;
26487 - inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
26488 + v4_set_sock_addr(inet, &nsa);
26489 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
26490 inet->saddr = 0; /* Use device */
26492 @@ -688,11 +697,13 @@ int inet_getname(struct socket *sock, st
26495 sin->sin_port = inet->dport;
26496 - sin->sin_addr.s_addr = inet->daddr;
26497 + sin->sin_addr.s_addr =
26498 + nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
26500 __be32 addr = inet->rcv_saddr;
26502 addr = inet->saddr;
26503 + addr = nx_map_sock_lback(sk->sk_nx_info, addr);
26504 sin->sin_port = inet->sport;
26505 sin->sin_addr.s_addr = addr;
26507 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/devinet.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/devinet.c
26508 --- linux-2.6.29.1/net/ipv4/devinet.c 2009-03-24 14:22:46.000000000 +0100
26509 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/devinet.c 2009-03-24 14:48:37.000000000 +0100
26510 @@ -413,6 +413,7 @@ struct in_device *inetdev_by_index(struc
26515 /* Called only from RTNL semaphored context. No locks. */
26517 struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix,
26518 @@ -653,6 +654,8 @@ int devinet_ioctl(struct net *net, unsig
26521 if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) {
26522 + struct nx_info *nxi = current->nx_info;
26524 if (tryaddrmatch) {
26525 /* Matthias Andree */
26526 /* compare label and address (4.4BSD style) */
26527 @@ -661,6 +664,8 @@ int devinet_ioctl(struct net *net, unsig
26528 This is checked above. */
26529 for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
26530 ifap = &ifa->ifa_next) {
26531 + if (!nx_v4_ifa_visible(nxi, ifa))
26533 if (!strcmp(ifr.ifr_name, ifa->ifa_label) &&
26534 sin_orig.sin_addr.s_addr ==
26535 ifa->ifa_address) {
26536 @@ -673,9 +678,12 @@ int devinet_ioctl(struct net *net, unsig
26537 comparing just the label */
26539 for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
26540 - ifap = &ifa->ifa_next)
26541 + ifap = &ifa->ifa_next) {
26542 + if (!nx_v4_ifa_visible(nxi, ifa))
26544 if (!strcmp(ifr.ifr_name, ifa->ifa_label))
26550 @@ -826,6 +834,8 @@ static int inet_gifconf(struct net_devic
26553 for (; ifa; ifa = ifa->ifa_next) {
26554 + if (!nx_v4_ifa_visible(current->nx_info, ifa))
26557 done += sizeof(ifr);
26559 @@ -1156,6 +1166,7 @@ static int inet_dump_ifaddr(struct sk_bu
26560 struct net_device *dev;
26561 struct in_device *in_dev;
26562 struct in_ifaddr *ifa;
26563 + struct sock *sk = skb->sk;
26564 int s_ip_idx, s_idx = cb->args[0];
26566 s_ip_idx = ip_idx = cb->args[1];
26567 @@ -1170,6 +1181,8 @@ static int inet_dump_ifaddr(struct sk_bu
26569 for (ifa = in_dev->ifa_list, ip_idx = 0; ifa;
26570 ifa = ifa->ifa_next, ip_idx++) {
26571 + if (sk && !nx_v4_ifa_visible(sk->sk_nx_info, ifa))
26573 if (ip_idx < s_ip_idx)
26575 if (inet_fill_ifaddr(skb, ifa, NETLINK_CB(cb->skb).pid,
26576 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/fib_hash.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/fib_hash.c
26577 --- linux-2.6.29.1/net/ipv4/fib_hash.c 2009-03-24 14:22:46.000000000 +0100
26578 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/fib_hash.c 2009-03-24 14:48:37.000000000 +0100
26579 @@ -1022,7 +1022,7 @@ static int fib_seq_show(struct seq_file
26580 prefix = f->fn_key;
26581 mask = FZ_MASK(iter->zone);
26582 flags = fib_flag_trans(fa->fa_type, mask, fi);
26584 + if (fi && nx_dev_visible(current->nx_info, fi->fib_dev))
26586 "%s\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u%n",
26587 fi->fib_dev ? fi->fib_dev->name : "*", prefix,
26588 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/inet_connection_sock.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_connection_sock.c
26589 --- linux-2.6.29.1/net/ipv4/inet_connection_sock.c 2009-03-24 14:22:46.000000000 +0100
26590 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_connection_sock.c 2009-03-24 14:48:37.000000000 +0100
26591 @@ -49,10 +49,40 @@ void inet_get_local_port_range(int *low,
26593 EXPORT_SYMBOL(inet_get_local_port_range);
26595 +int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
26597 + __be32 sk1_rcv_saddr = inet_rcv_saddr(sk1),
26598 + sk2_rcv_saddr = inet_rcv_saddr(sk2);
26600 + if (inet_v6_ipv6only(sk2))
26603 + if (sk1_rcv_saddr &&
26605 + sk1_rcv_saddr == sk2_rcv_saddr)
26608 + if (sk1_rcv_saddr &&
26609 + !sk2_rcv_saddr &&
26610 + v4_addr_in_nx_info(sk2->sk_nx_info, sk1_rcv_saddr, NXA_MASK_BIND))
26613 + if (sk2_rcv_saddr &&
26614 + !sk1_rcv_saddr &&
26615 + v4_addr_in_nx_info(sk1->sk_nx_info, sk2_rcv_saddr, NXA_MASK_BIND))
26618 + if (!sk1_rcv_saddr &&
26619 + !sk2_rcv_saddr &&
26620 + nx_v4_addr_conflict(sk1->sk_nx_info, sk2->sk_nx_info))
26626 int inet_csk_bind_conflict(const struct sock *sk,
26627 const struct inet_bind_bucket *tb)
26629 - const __be32 sk_rcv_saddr = inet_rcv_saddr(sk);
26631 struct hlist_node *node;
26632 int reuse = sk->sk_reuse;
26633 @@ -72,9 +102,7 @@ int inet_csk_bind_conflict(const struct
26634 sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
26635 if (!reuse || !sk2->sk_reuse ||
26636 sk2->sk_state == TCP_LISTEN) {
26637 - const __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
26638 - if (!sk2_rcv_saddr || !sk_rcv_saddr ||
26639 - sk2_rcv_saddr == sk_rcv_saddr)
26640 + if (ipv4_rcv_saddr_equal(sk, sk2))
26644 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/inet_diag.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_diag.c
26645 --- linux-2.6.29.1/net/ipv4/inet_diag.c 2009-03-24 14:22:46.000000000 +0100
26646 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_diag.c 2009-03-24 14:48:37.000000000 +0100
26648 #include <linux/stddef.h>
26650 #include <linux/inet_diag.h>
26651 +#include <linux/vs_network.h>
26652 +#include <linux/vs_inet.h>
26654 static const struct inet_diag_handler **inet_diag_table;
26656 @@ -118,8 +120,8 @@ static int inet_csk_diag_fill(struct soc
26658 r->id.idiag_sport = inet->sport;
26659 r->id.idiag_dport = inet->dport;
26660 - r->id.idiag_src[0] = inet->rcv_saddr;
26661 - r->id.idiag_dst[0] = inet->daddr;
26662 + r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, inet->rcv_saddr);
26663 + r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
26665 #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
26666 if (r->idiag_family == AF_INET6) {
26667 @@ -206,8 +208,8 @@ static int inet_twsk_diag_fill(struct in
26668 r->id.idiag_cookie[1] = (u32)(((unsigned long)tw >> 31) >> 1);
26669 r->id.idiag_sport = tw->tw_sport;
26670 r->id.idiag_dport = tw->tw_dport;
26671 - r->id.idiag_src[0] = tw->tw_rcv_saddr;
26672 - r->id.idiag_dst[0] = tw->tw_daddr;
26673 + r->id.idiag_src[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_rcv_saddr);
26674 + r->id.idiag_dst[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_daddr);
26675 r->idiag_state = tw->tw_substate;
26676 r->idiag_timer = 3;
26677 r->idiag_expires = DIV_ROUND_UP(tmo * 1000, HZ);
26678 @@ -264,6 +266,7 @@ static int inet_diag_get_exact(struct sk
26681 if (req->idiag_family == AF_INET) {
26682 + /* TODO: lback */
26683 sk = inet_lookup(&init_net, hashinfo, req->id.idiag_dst[0],
26684 req->id.idiag_dport, req->id.idiag_src[0],
26685 req->id.idiag_sport, req->id.idiag_if);
26686 @@ -506,6 +509,7 @@ static int inet_csk_diag_dump(struct soc
26690 + /* TODO: lback */
26691 entry.saddr = &inet->rcv_saddr;
26692 entry.daddr = &inet->daddr;
26694 @@ -542,6 +546,7 @@ static int inet_twsk_diag_dump(struct in
26698 + /* TODO: lback */
26699 entry.saddr = &tw->tw_rcv_saddr;
26700 entry.daddr = &tw->tw_daddr;
26702 @@ -588,8 +593,8 @@ static int inet_diag_fill_req(struct sk_
26704 r->id.idiag_sport = inet->sport;
26705 r->id.idiag_dport = ireq->rmt_port;
26706 - r->id.idiag_src[0] = ireq->loc_addr;
26707 - r->id.idiag_dst[0] = ireq->rmt_addr;
26708 + r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->loc_addr);
26709 + r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->rmt_addr);
26710 r->idiag_expires = jiffies_to_msecs(tmo);
26711 r->idiag_rqueue = 0;
26712 r->idiag_wqueue = 0;
26713 @@ -659,6 +664,7 @@ static int inet_diag_dump_reqs(struct sk
26717 + /* TODO: lback */
26719 #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
26720 (entry.family == AF_INET6) ?
26721 @@ -729,6 +735,8 @@ static int inet_diag_dump(struct sk_buff
26722 sk_nulls_for_each(sk, node, &ilb->head) {
26723 struct inet_sock *inet = inet_sk(sk);
26725 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26730 @@ -795,6 +803,8 @@ skip_listen_ht:
26731 sk_nulls_for_each(sk, node, &head->chain) {
26732 struct inet_sock *inet = inet_sk(sk);
26734 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26738 if (!(r->idiag_states & (1 << sk->sk_state)))
26739 @@ -819,6 +829,8 @@ next_normal:
26740 inet_twsk_for_each(tw, node,
26743 + if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
26747 if (r->id.idiag_sport != tw->tw_sport &&
26748 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/inet_hashtables.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_hashtables.c
26749 --- linux-2.6.29.1/net/ipv4/inet_hashtables.c 2009-03-24 14:22:46.000000000 +0100
26750 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/inet_hashtables.c 2009-03-24 19:35:10.000000000 +0100
26753 #include <net/inet_connection_sock.h>
26754 #include <net/inet_hashtables.h>
26755 +#include <net/route.h>
26756 #include <net/ip.h>
26759 @@ -142,7 +143,6 @@ static inline int compute_score(struct s
26760 * wildcarded during the search since they can never be otherwise.
26764 struct sock *__inet_lookup_listener(struct net *net,
26765 struct inet_hashinfo *hashinfo,
26766 const __be32 daddr, const unsigned short hnum,
26767 @@ -165,6 +165,7 @@ begin:
26773 * if the nulls value we got at the end of this lookup is
26774 * not the expected one, we must restart lookup.
26775 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/netfilter/nf_nat_helper.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/netfilter/nf_nat_helper.c
26776 --- linux-2.6.29.1/net/ipv4/netfilter/nf_nat_helper.c 2008-12-25 00:26:37.000000000 +0100
26777 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/netfilter/nf_nat_helper.c 2009-02-22 22:54:26.000000000 +0100
26779 #include <net/route.h>
26781 #include <linux/netfilter_ipv4.h>
26782 +#include <net/route.h>
26783 #include <net/netfilter/nf_conntrack.h>
26784 #include <net/netfilter/nf_conntrack_helper.h>
26785 #include <net/netfilter/nf_conntrack_ecache.h>
26786 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/netfilter.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/netfilter.c
26787 --- linux-2.6.29.1/net/ipv4/netfilter.c 2009-03-24 14:22:46.000000000 +0100
26788 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/netfilter.c 2009-03-24 14:48:37.000000000 +0100
26790 #include <linux/netfilter_ipv4.h>
26791 #include <linux/ip.h>
26792 #include <linux/skbuff.h>
26793 -#include <net/route.h>
26794 +// #include <net/route.h>
26795 #include <net/xfrm.h>
26796 #include <net/ip.h>
26797 #include <net/netfilter/nf_queue.h>
26798 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/raw.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/raw.c
26799 --- linux-2.6.29.1/net/ipv4/raw.c 2009-03-24 14:22:46.000000000 +0100
26800 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/raw.c 2009-03-24 14:48:37.000000000 +0100
26801 @@ -117,7 +117,7 @@ static struct sock *__raw_v4_lookup(stru
26803 if (net_eq(sock_net(sk), net) && inet->num == num &&
26804 !(inet->daddr && inet->daddr != raddr) &&
26805 - !(inet->rcv_saddr && inet->rcv_saddr != laddr) &&
26806 + v4_sock_addr_match(sk->sk_nx_info, inet, laddr) &&
26807 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
26808 goto found; /* gotcha */
26810 @@ -372,6 +372,12 @@ static int raw_send_hdrinc(struct sock *
26811 icmp_out_count(net, ((struct icmphdr *)
26812 skb_transport_header(skb))->type);
26815 + if (!nx_check(0, VS_ADMIN) && !capable(CAP_NET_RAW) &&
26816 + sk->sk_nx_info &&
26817 + !v4_addr_in_nx_info(sk->sk_nx_info, iph->saddr, NXA_MASK_BIND))
26820 err = NF_HOOK(PF_INET, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
26823 @@ -383,6 +389,7 @@ out:
26830 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
26831 @@ -550,6 +557,13 @@ static int raw_sendmsg(struct kiocb *ioc
26834 security_sk_classify_flow(sk, &fl);
26835 + if (sk->sk_nx_info) {
26836 + err = ip_v4_find_src(sock_net(sk),
26837 + sk->sk_nx_info, &rt, &fl);
26842 err = ip_route_output_flow(sock_net(sk), &rt, &fl, sk, 1);
26845 @@ -619,17 +633,19 @@ static int raw_bind(struct sock *sk, str
26847 struct inet_sock *inet = inet_sk(sk);
26848 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
26849 + struct nx_v4_sock_addr nsa = { 0 };
26853 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
26855 - chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
26856 + v4_map_sock_addr(inet, addr, &nsa);
26857 + chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
26858 ret = -EADDRNOTAVAIL;
26859 - if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
26860 + if (nsa.saddr && chk_addr_ret != RTN_LOCAL &&
26861 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
26863 - inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
26864 + v4_set_sock_addr(inet, &nsa);
26865 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
26866 inet->saddr = 0; /* Use device */
26868 @@ -681,7 +697,8 @@ static int raw_recvmsg(struct kiocb *ioc
26869 /* Copy the address. */
26871 sin->sin_family = AF_INET;
26872 - sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
26873 + sin->sin_addr.s_addr =
26874 + nx_map_sock_lback(sk->sk_nx_info, ip_hdr(skb)->saddr);
26876 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
26878 @@ -858,7 +875,8 @@ static struct sock *raw_get_first(struct
26879 struct hlist_node *node;
26881 sk_for_each(sk, node, &state->h->ht[state->bucket])
26882 - if (sock_net(sk) == seq_file_net(seq))
26883 + if ((sock_net(sk) == seq_file_net(seq)) &&
26884 + nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26888 @@ -874,7 +892,8 @@ static struct sock *raw_get_next(struct
26892 - } while (sk && sock_net(sk) != seq_file_net(seq));
26893 + } while (sk && ((sock_net(sk) != seq_file_net(seq)) ||
26894 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
26896 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
26897 sk = sk_head(&state->h->ht[state->bucket]);
26898 @@ -933,7 +952,10 @@ static void raw_sock_seq_show(struct seq
26900 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
26901 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
26902 - i, src, srcp, dest, destp, sp->sk_state,
26904 + nx_map_sock_lback(current_nx_info(), src), srcp,
26905 + nx_map_sock_lback(current_nx_info(), dest), destp,
26907 atomic_read(&sp->sk_wmem_alloc),
26908 atomic_read(&sp->sk_rmem_alloc),
26909 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
26910 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/tcp.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp.c
26911 --- linux-2.6.29.1/net/ipv4/tcp.c 2009-03-24 14:22:46.000000000 +0100
26912 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp.c 2009-03-24 14:48:37.000000000 +0100
26913 @@ -264,6 +264,7 @@
26914 #include <linux/cache.h>
26915 #include <linux/err.h>
26916 #include <linux/crypto.h>
26917 +#include <linux/in.h>
26919 #include <net/icmp.h>
26920 #include <net/tcp.h>
26921 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/tcp_ipv4.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp_ipv4.c
26922 --- linux-2.6.29.1/net/ipv4/tcp_ipv4.c 2009-03-24 14:22:46.000000000 +0100
26923 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp_ipv4.c 2009-03-24 19:51:34.000000000 +0100
26924 @@ -1894,6 +1894,12 @@ static void *listening_get_next(struct s
26925 req = req->dl_next;
26928 + vxdprintk(VXD_CBIT(net, 6),
26929 + "sk,req: %p [#%d] (from %d)", req->sk,
26930 + (req->sk)?req->sk->sk_nid:0, nx_current_nid());
26932 + !nx_check(req->sk->sk_nid, VS_WATCH_P | VS_IDENT))
26934 if (req->rsk_ops->family == st->family) {
26937 @@ -1918,6 +1924,10 @@ get_req:
26940 sk_nulls_for_each_from(sk, node) {
26941 + vxdprintk(VXD_CBIT(net, 6), "sk: %p [#%d] (from %d)",
26942 + sk, sk->sk_nid, nx_current_nid());
26943 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26945 if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
26948 @@ -1981,6 +1991,11 @@ static void *established_get_first(struc
26950 spin_lock_bh(lock);
26951 sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
26952 + vxdprintk(VXD_CBIT(net, 6),
26953 + "sk,egf: %p [#%d] (from %d)",
26954 + sk, sk->sk_nid, nx_current_nid());
26955 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26957 if (sk->sk_family != st->family ||
26958 !net_eq(sock_net(sk), net)) {
26960 @@ -1991,6 +2006,11 @@ static void *established_get_first(struc
26961 st->state = TCP_SEQ_STATE_TIME_WAIT;
26962 inet_twsk_for_each(tw, node,
26963 &tcp_hashinfo.ehash[st->bucket].twchain) {
26964 + vxdprintk(VXD_CBIT(net, 6),
26965 + "tw: %p [#%d] (from %d)",
26966 + tw, tw->tw_nid, nx_current_nid());
26967 + if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
26969 if (tw->tw_family != st->family ||
26970 !net_eq(twsk_net(tw), net)) {
26972 @@ -2019,7 +2039,9 @@ static void *established_get_next(struct
26976 - while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
26977 + while (tw && (tw->tw_family != st->family ||
26978 + !net_eq(twsk_net(tw), net) ||
26979 + !nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))) {
26983 @@ -2042,6 +2064,11 @@ get_tw:
26984 sk = sk_nulls_next(sk);
26986 sk_nulls_for_each_from(sk, node) {
26987 + vxdprintk(VXD_CBIT(net, 6),
26988 + "sk,egn: %p [#%d] (from %d)",
26989 + sk, sk->sk_nid, nx_current_nid());
26990 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26992 if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
26995 @@ -2193,9 +2220,9 @@ static void get_openreq4(struct sock *sk
26996 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
26997 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p%n",
27000 + nx_map_sock_lback(current_nx_info(), ireq->loc_addr),
27001 ntohs(inet_sk(sk)->sport),
27003 + nx_map_sock_lback(current_nx_info(), ireq->rmt_addr),
27004 ntohs(ireq->rmt_port),
27006 0, 0, /* could print option size, but that is af dependent. */
27007 @@ -2238,7 +2265,10 @@ static void get_tcp4_sock(struct sock *s
27009 seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
27010 "%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
27011 - i, src, srcp, dest, destp, sk->sk_state,
27013 + nx_map_sock_lback(current_nx_info(), src), srcp,
27014 + nx_map_sock_lback(current_nx_info(), dest), destp,
27016 tp->write_seq - tp->snd_una,
27017 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
27018 (tp->rcv_nxt - tp->copied_seq),
27019 @@ -2274,7 +2304,10 @@ static void get_timewait4_sock(struct in
27021 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27022 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p%n",
27023 - i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
27025 + nx_map_sock_lback(current_nx_info(), src), srcp,
27026 + nx_map_sock_lback(current_nx_info(), dest), destp,
27027 + tw->tw_substate, 0, 0,
27028 3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
27029 atomic_read(&tw->tw_refcnt), tw, len);
27031 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/tcp_minisocks.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp_minisocks.c
27032 --- linux-2.6.29.1/net/ipv4/tcp_minisocks.c 2009-03-24 14:22:46.000000000 +0100
27033 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/tcp_minisocks.c 2009-03-24 14:48:37.000000000 +0100
27035 #include <net/inet_common.h>
27036 #include <net/xfrm.h>
27038 +#include <linux/vs_limit.h>
27039 +#include <linux/vs_socket.h>
27040 +#include <linux/vs_context.h>
27042 #ifdef CONFIG_SYSCTL
27043 #define SYNC_INIT 0 /* let the user enable it */
27045 @@ -293,6 +297,11 @@ void tcp_time_wait(struct sock *sk, int
27046 tcptw->tw_ts_recent = tp->rx_opt.ts_recent;
27047 tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
27049 + tw->tw_xid = sk->sk_xid;
27050 + tw->tw_vx_info = NULL;
27051 + tw->tw_nid = sk->sk_nid;
27052 + tw->tw_nx_info = NULL;
27054 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
27055 if (tw->tw_family == PF_INET6) {
27056 struct ipv6_pinfo *np = inet6_sk(sk);
27057 diff -NurpP --minimal linux-2.6.29.1/net/ipv4/udp.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/udp.c
27058 --- linux-2.6.29.1/net/ipv4/udp.c 2009-04-15 22:53:12.000000000 +0200
27059 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv4/udp.c 2009-04-15 22:58:30.000000000 +0200
27060 @@ -222,14 +222,7 @@ fail:
27064 -static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
27066 - struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
27068 - return ( !ipv6_only_sock(sk2) &&
27069 - (!inet1->rcv_saddr || !inet2->rcv_saddr ||
27070 - inet1->rcv_saddr == inet2->rcv_saddr ));
27072 +extern int ipv4_rcv_saddr_equal(const struct sock *, const struct sock *);
27074 int udp_v4_get_port(struct sock *sk, unsigned short snum)
27076 @@ -251,6 +244,11 @@ static inline int compute_score(struct s
27077 if (inet->rcv_saddr != daddr)
27081 + /* block non nx_info ips */
27082 + if (!v4_addr_in_nx_info(sk->sk_nx_info,
27083 + daddr, NXA_MASK_BIND))
27087 if (inet->daddr != saddr)
27088 @@ -271,6 +269,7 @@ static inline int compute_score(struct s
27093 /* UDP is nearly always wildcards out the wazoo, it makes no sense to try
27094 * harder than this. -DaveM
27096 @@ -292,6 +291,11 @@ begin:
27097 sk_nulls_for_each_rcu(sk, node, &hslot->head) {
27098 score = compute_score(sk, net, saddr, hnum, sport,
27099 daddr, dport, dif);
27100 + /* FIXME: disabled?
27101 + if (score == 9) {
27105 if (score > badness) {
27108 @@ -305,6 +309,7 @@ begin:
27109 if (get_nulls_value(node) != hash)
27114 if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
27116 @@ -314,6 +319,7 @@ begin:
27124 @@ -356,7 +362,7 @@ static inline struct sock *udp_v4_mcast_
27125 s->sk_hash != hnum ||
27126 (inet->daddr && inet->daddr != rmt_addr) ||
27127 (inet->dport != rmt_port && inet->dport) ||
27128 - (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
27129 + !v4_sock_addr_match(sk->sk_nx_info, inet, loc_addr) ||
27130 ipv6_only_sock(s) ||
27131 (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
27133 @@ -694,8 +700,13 @@ int udp_sendmsg(struct kiocb *iocb, stru
27134 { .sport = inet->sport,
27135 .dport = dport } } };
27136 struct net *net = sock_net(sk);
27137 + struct nx_info *nxi = sk->sk_nx_info;
27139 security_sk_classify_flow(sk, &fl);
27140 + err = ip_v4_find_src(net, nxi, &rt, &fl);
27144 err = ip_route_output_flow(net, &rt, &fl, sk, 1);
27146 if (err == -ENETUNREACH)
27147 @@ -940,7 +951,8 @@ try_again:
27149 sin->sin_family = AF_INET;
27150 sin->sin_port = udp_hdr(skb)->source;
27151 - sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
27152 + sin->sin_addr.s_addr = nx_map_sock_lback(
27153 + skb->sk->sk_nx_info, ip_hdr(skb)->saddr);
27154 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
27156 if (inet->cmsg_flags)
27157 @@ -1594,6 +1606,8 @@ static struct sock *udp_get_first(struct
27158 sk_nulls_for_each(sk, node, &hslot->head) {
27159 if (!net_eq(sock_net(sk), net))
27161 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27163 if (sk->sk_family == state->family)
27166 @@ -1611,7 +1625,9 @@ static struct sock *udp_get_next(struct
27169 sk = sk_nulls_next(sk);
27170 - } while (sk && (!net_eq(sock_net(sk), net) || sk->sk_family != state->family));
27171 + } while (sk && (!net_eq(sock_net(sk), net) ||
27172 + sk->sk_family != state->family ||
27173 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
27176 if (state->bucket < UDP_HTABLE_SIZE)
27177 @@ -1716,7 +1732,10 @@ static void udp4_format_sock(struct sock
27179 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27180 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d%n",
27181 - bucket, src, srcp, dest, destp, sp->sk_state,
27183 + nx_map_sock_lback(current_nx_info(), src), srcp,
27184 + nx_map_sock_lback(current_nx_info(), dest), destp,
27186 atomic_read(&sp->sk_wmem_alloc),
27187 atomic_read(&sp->sk_rmem_alloc),
27188 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
27189 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/addrconf.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/addrconf.c
27190 --- linux-2.6.29.1/net/ipv6/addrconf.c 2009-03-24 14:22:46.000000000 +0100
27191 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/addrconf.c 2009-03-24 20:39:23.000000000 +0100
27194 #include <linux/proc_fs.h>
27195 #include <linux/seq_file.h>
27196 +#include <linux/vs_network.h>
27197 +#include <linux/vs_inet6.h>
27199 /* Set to 3 to get tracing... */
27200 #define ACONF_DEBUG 2
27201 @@ -1111,7 +1113,7 @@ out:
27203 int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
27204 const struct in6_addr *daddr, unsigned int prefs,
27205 - struct in6_addr *saddr)
27206 + struct in6_addr *saddr, struct nx_info *nxi)
27208 struct ipv6_saddr_score scores[2],
27209 *score = &scores[0], *hiscore = &scores[1];
27210 @@ -1184,6 +1186,8 @@ int ipv6_dev_get_saddr(struct net *net,
27214 + if (!v6_addr_in_nx_info(nxi, &score->ifa->addr, -1))
27218 bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
27219 @@ -1367,35 +1371,46 @@ struct inet6_ifaddr *ipv6_get_ifaddr(str
27223 +extern int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2);
27225 int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
27227 const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
27228 const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
27229 - __be32 sk_rcv_saddr = inet_sk(sk)->rcv_saddr;
27230 __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
27231 int sk_ipv6only = ipv6_only_sock(sk);
27232 int sk2_ipv6only = inet_v6_ipv6only(sk2);
27233 int addr_type = ipv6_addr_type(sk_rcv_saddr6);
27234 int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
27236 - if (!sk2_rcv_saddr && !sk_ipv6only)
27237 + /* FIXME: needs handling for v4 ANY */
27238 + if (!sk2_rcv_saddr && !sk_ipv6only && !sk2->sk_nx_info)
27241 if (addr_type2 == IPV6_ADDR_ANY &&
27242 - !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
27243 + !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED) &&
27244 + v6_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr6, -1))
27247 if (addr_type == IPV6_ADDR_ANY &&
27248 - !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
27249 + !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED) &&
27250 + (sk2_rcv_saddr6 && v6_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr6, -1)))
27253 + if (addr_type == IPV6_ADDR_ANY &&
27254 + addr_type2 == IPV6_ADDR_ANY &&
27255 + nx_v6_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info))
27258 if (sk2_rcv_saddr6 &&
27259 + addr_type != IPV6_ADDR_ANY &&
27260 + addr_type != IPV6_ADDR_ANY &&
27261 ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
27264 if (addr_type == IPV6_ADDR_MAPPED &&
27266 - (!sk2_rcv_saddr || !sk_rcv_saddr || sk_rcv_saddr == sk2_rcv_saddr))
27267 + ipv4_rcv_saddr_equal(sk, sk2))
27271 @@ -2993,7 +3008,10 @@ static void if6_seq_stop(struct seq_file
27272 static int if6_seq_show(struct seq_file *seq, void *v)
27274 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
27275 - seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27277 + if (nx_check(0, VS_ADMIN|VS_WATCH) ||
27278 + v6_addr_in_nx_info(current_nx_info(), &ifp->addr, -1))
27279 + seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27281 ifp->idev->dev->ifindex,
27283 @@ -3487,6 +3505,12 @@ static int inet6_dump_addr(struct sk_buf
27284 struct ifmcaddr6 *ifmca;
27285 struct ifacaddr6 *ifaca;
27286 struct net *net = sock_net(skb->sk);
27287 + struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27289 + /* disable ipv6 on non v6 guests */
27290 + if (nxi && !nx_info_has_v6(nxi))
27294 s_idx = cb->args[0];
27295 s_ip_idx = ip_idx = cb->args[1];
27296 @@ -3508,6 +3532,8 @@ static int inet6_dump_addr(struct sk_buf
27297 ifa = ifa->if_next, ip_idx++) {
27298 if (ip_idx < s_ip_idx)
27300 + if (!v6_addr_in_nx_info(nxi, &ifa->addr, -1))
27302 err = inet6_fill_ifaddr(skb, ifa,
27303 NETLINK_CB(cb->skb).pid,
27304 cb->nlh->nlmsg_seq,
27305 @@ -3521,6 +3547,8 @@ static int inet6_dump_addr(struct sk_buf
27306 ifmca = ifmca->next, ip_idx++) {
27307 if (ip_idx < s_ip_idx)
27309 + if (!v6_addr_in_nx_info(nxi, &ifmca->mca_addr, -1))
27311 err = inet6_fill_ifmcaddr(skb, ifmca,
27312 NETLINK_CB(cb->skb).pid,
27313 cb->nlh->nlmsg_seq,
27314 @@ -3534,6 +3562,8 @@ static int inet6_dump_addr(struct sk_buf
27315 ifaca = ifaca->aca_next, ip_idx++) {
27316 if (ip_idx < s_ip_idx)
27318 + if (!v6_addr_in_nx_info(nxi, &ifaca->aca_addr, -1))
27320 err = inet6_fill_ifacaddr(skb, ifaca,
27321 NETLINK_CB(cb->skb).pid,
27322 cb->nlh->nlmsg_seq,
27323 @@ -3819,12 +3849,19 @@ static int inet6_dump_ifinfo(struct sk_b
27324 int s_idx = cb->args[0];
27325 struct net_device *dev;
27326 struct inet6_dev *idev;
27327 + struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27329 + /* FIXME: maybe disable ipv6 on non v6 guests?
27330 + if (skb->sk && skb->sk->sk_vx_info)
27331 + return skb->len; */
27333 read_lock(&dev_base_lock);
27335 for_each_netdev(net, dev) {
27338 + if (!v6_dev_in_nx_info(dev, nxi))
27340 if ((idev = in6_dev_get(dev)) == NULL)
27342 err = inet6_fill_ifinfo(skb, idev, NETLINK_CB(cb->skb).pid,
27343 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/af_inet6.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/af_inet6.c
27344 --- linux-2.6.29.1/net/ipv6/af_inet6.c 2009-03-24 14:22:46.000000000 +0100
27345 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/af_inet6.c 2009-03-24 14:48:37.000000000 +0100
27347 #include <linux/netdevice.h>
27348 #include <linux/icmpv6.h>
27349 #include <linux/netfilter_ipv6.h>
27350 +#include <linux/vs_inet.h>
27351 +#include <linux/vs_inet6.h>
27353 #include <net/ip.h>
27354 #include <net/ipv6.h>
27356 #include <net/tcp.h>
27357 #include <net/ipip.h>
27358 #include <net/protocol.h>
27359 +#include <net/route.h>
27360 #include <net/inet_common.h>
27361 #include <net/route.h>
27362 #include <net/transp_v6.h>
27363 @@ -146,9 +149,12 @@ lookup_protocol:
27367 + if ((protocol == IPPROTO_ICMPV6) &&
27368 + nx_capable(answer->capability, NXC_RAW_ICMP))
27370 if (answer->capability > 0 && !capable(answer->capability))
27371 goto out_rcu_unlock;
27374 sock->ops = answer->ops;
27375 answer_prot = answer->prot;
27376 answer_no_check = answer->no_check;
27377 @@ -247,6 +253,7 @@ int inet6_bind(struct socket *sock, stru
27378 struct inet_sock *inet = inet_sk(sk);
27379 struct ipv6_pinfo *np = inet6_sk(sk);
27380 struct net *net = sock_net(sk);
27381 + struct nx_v6_sock_addr nsa;
27383 unsigned short snum;
27385 @@ -258,6 +265,11 @@ int inet6_bind(struct socket *sock, stru
27387 if (addr_len < SIN6_LEN_RFC2133)
27390 + err = v6_map_sock_addr(inet, addr, &nsa);
27394 addr_type = ipv6_addr_type(&addr->sin6_addr);
27395 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
27397 @@ -281,6 +293,10 @@ int inet6_bind(struct socket *sock, stru
27398 err = -EADDRNOTAVAIL;
27401 + if (!v4_addr_in_nx_info(sk->sk_nx_info, v4addr, NXA_MASK_BIND)) {
27402 + err = -EADDRNOTAVAIL;
27406 if (addr_type != IPV6_ADDR_ANY) {
27407 struct net_device *dev = NULL;
27408 @@ -306,6 +322,11 @@ int inet6_bind(struct socket *sock, stru
27412 + if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
27413 + err = -EADDRNOTAVAIL;
27417 /* ipv4 addr of the socket is invalid. Only the
27418 * unspecified and mapped address have a v4 equivalent.
27420 @@ -324,6 +345,8 @@ int inet6_bind(struct socket *sock, stru
27424 + v6_set_sock_addr(inet, &nsa);
27426 inet->rcv_saddr = v4addr;
27427 inet->saddr = v4addr;
27429 @@ -416,9 +439,11 @@ int inet6_getname(struct socket *sock, s
27431 sin->sin6_port = inet->dport;
27432 ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
27433 + /* FIXME: remap lback? */
27435 sin->sin6_flowinfo = np->flow_label;
27437 + /* FIXME: remap lback? */
27438 if (ipv6_addr_any(&np->rcv_saddr))
27439 ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
27441 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/fib6_rules.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/fib6_rules.c
27442 --- linux-2.6.29.1/net/ipv6/fib6_rules.c 2008-12-25 00:26:37.000000000 +0100
27443 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/fib6_rules.c 2009-02-22 22:54:26.000000000 +0100
27444 @@ -96,7 +96,7 @@ static int fib6_rule_action(struct fib_r
27445 if (ipv6_dev_get_saddr(net,
27446 ip6_dst_idev(&rt->u.dst)->dev,
27447 &flp->fl6_dst, srcprefs,
27451 if (!ipv6_prefix_equal(&saddr, &r->src.addr,
27453 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/inet6_hashtables.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/inet6_hashtables.c
27454 --- linux-2.6.29.1/net/ipv6/inet6_hashtables.c 2009-03-24 14:22:46.000000000 +0100
27455 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/inet6_hashtables.c 2009-03-24 20:50:24.000000000 +0100
27458 #include <linux/module.h>
27459 #include <linux/random.h>
27460 +#include <linux/vs_inet6.h>
27462 #include <net/inet_connection_sock.h>
27463 #include <net/inet_hashtables.h>
27464 @@ -76,7 +77,6 @@ struct sock *__inet6_lookup_established(
27465 unsigned int slot = hash & (hashinfo->ehash_size - 1);
27466 struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
27471 sk_nulls_for_each_rcu(sk, node, &head->chain) {
27472 @@ -88,7 +88,7 @@ begin:
27480 if (get_nulls_value(node) != slot)
27481 @@ -134,6 +134,9 @@ static int inline compute_score(struct s
27482 if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
27486 + if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
27489 if (sk->sk_bound_dev_if) {
27490 if (sk->sk_bound_dev_if != dif)
27491 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/ip6_output.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/ip6_output.c
27492 --- linux-2.6.29.1/net/ipv6/ip6_output.c 2009-03-24 14:22:47.000000000 +0100
27493 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/ip6_output.c 2009-03-24 14:48:37.000000000 +0100
27494 @@ -951,7 +951,7 @@ static int ip6_dst_lookup_tail(struct so
27495 err = ipv6_dev_get_saddr(net, ip6_dst_idev(*dst)->dev,
27497 sk ? inet6_sk(sk)->srcprefs : 0,
27499 + &fl->fl6_src, sk->sk_nx_info);
27501 goto out_err_release;
27503 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/Kconfig linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/Kconfig
27504 --- linux-2.6.29.1/net/ipv6/Kconfig 2008-12-25 00:26:37.000000000 +0100
27505 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/Kconfig 2009-02-22 22:54:26.000000000 +0100
27508 # IPv6 as module will cause a CRASH if you try to unload it
27510 - tristate "The IPv6 protocol"
27512 + bool "The IPv6 protocol"
27515 This is complemental support for the IP version 6.
27516 You will still be able to do traditional IPv4 networking as well.
27517 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/ndisc.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/ndisc.c
27518 --- linux-2.6.29.1/net/ipv6/ndisc.c 2009-03-24 14:22:47.000000000 +0100
27519 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/ndisc.c 2009-03-24 14:48:37.000000000 +0100
27520 @@ -589,7 +589,7 @@ static void ndisc_send_na(struct net_dev
27522 if (ipv6_dev_get_saddr(dev_net(dev), dev, daddr,
27523 inet6_sk(dev_net(dev)->ipv6.ndisc_sk)->srcprefs,
27525 + &tmpaddr, NULL /* FIXME: ? */ ))
27527 src_addr = &tmpaddr;
27529 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/raw.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/raw.c
27530 --- linux-2.6.29.1/net/ipv6/raw.c 2009-03-24 14:22:47.000000000 +0100
27531 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/raw.c 2009-03-24 14:48:37.000000000 +0100
27533 #include <linux/icmpv6.h>
27534 #include <linux/netfilter.h>
27535 #include <linux/netfilter_ipv6.h>
27536 +#include <linux/vs_inet6.h>
27537 #include <linux/skbuff.h>
27538 #include <asm/uaccess.h>
27539 #include <asm/ioctls.h>
27540 @@ -281,6 +282,13 @@ static int rawv6_bind(struct sock *sk, s
27544 + if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
27545 + err = -EADDRNOTAVAIL;
27551 /* ipv4 addr of the socket is invalid. Only the
27552 * unspecified and mapped address have a v4 equivalent.
27554 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/route.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/route.c
27555 --- linux-2.6.29.1/net/ipv6/route.c 2009-03-24 14:22:47.000000000 +0100
27556 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/route.c 2009-03-24 14:48:37.000000000 +0100
27557 @@ -2254,7 +2254,8 @@ static int rt6_fill_node(struct net *net
27558 struct inet6_dev *idev = ip6_dst_idev(&rt->u.dst);
27559 struct in6_addr saddr_buf;
27560 if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
27561 - dst, 0, &saddr_buf) == 0)
27562 + dst, 0, &saddr_buf,
27563 + (skb->sk ? skb->sk->sk_nx_info : NULL)) == 0)
27564 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
27567 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/tcp_ipv6.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/tcp_ipv6.c
27568 --- linux-2.6.29.1/net/ipv6/tcp_ipv6.c 2009-03-24 14:22:47.000000000 +0100
27569 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/tcp_ipv6.c 2009-03-24 14:48:37.000000000 +0100
27572 #include <linux/crypto.h>
27573 #include <linux/scatterlist.h>
27574 +#include <linux/vs_inet6.h>
27576 static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
27577 static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
27578 @@ -156,8 +157,15 @@ static int tcp_v6_connect(struct sock *s
27579 * connect() to INADDR_ANY means loopback (BSD'ism).
27582 - if(ipv6_addr_any(&usin->sin6_addr))
27583 - usin->sin6_addr.s6_addr[15] = 0x1;
27584 + if(ipv6_addr_any(&usin->sin6_addr)) {
27585 + struct nx_info *nxi = sk->sk_nx_info;
27587 + if (nxi && nx_info_has_v6(nxi))
27588 + /* FIXME: remap lback? */
27589 + usin->sin6_addr = nxi->v6.ip;
27591 + usin->sin6_addr.s6_addr[15] = 0x1;
27594 addr_type = ipv6_addr_type(&usin->sin6_addr);
27596 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/udp.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/udp.c
27597 --- linux-2.6.29.1/net/ipv6/udp.c 2009-03-24 14:22:47.000000000 +0100
27598 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/udp.c 2009-03-24 20:56:49.000000000 +0100
27601 #include <linux/proc_fs.h>
27602 #include <linux/seq_file.h>
27603 +#include <linux/vs_inet6.h>
27604 #include "udp_impl.h"
27606 int udp_v6_get_port(struct sock *sk, unsigned short snum)
27607 @@ -77,6 +78,10 @@ static inline int compute_score(struct s
27608 if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
27612 + /* block non nx_info ips */
27613 + if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
27616 if (!ipv6_addr_any(&np->daddr)) {
27617 if (!ipv6_addr_equal(&np->daddr, saddr))
27618 diff -NurpP --minimal linux-2.6.29.1/net/ipv6/xfrm6_policy.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/xfrm6_policy.c
27619 --- linux-2.6.29.1/net/ipv6/xfrm6_policy.c 2009-03-24 14:22:47.000000000 +0100
27620 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/ipv6/xfrm6_policy.c 2009-03-24 14:48:37.000000000 +0100
27621 @@ -63,7 +63,7 @@ static int xfrm6_get_saddr(struct net *n
27622 dev = ip6_dst_idev(dst)->dev;
27623 ipv6_dev_get_saddr(dev_net(dev), dev,
27624 (struct in6_addr *)&daddr->a6, 0,
27625 - (struct in6_addr *)&saddr->a6);
27626 + (struct in6_addr *)&saddr->a6, NULL);
27630 diff -NurpP --minimal linux-2.6.29.1/net/netlink/af_netlink.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/netlink/af_netlink.c
27631 --- linux-2.6.29.1/net/netlink/af_netlink.c 2009-03-24 14:22:47.000000000 +0100
27632 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/netlink/af_netlink.c 2009-03-24 14:48:37.000000000 +0100
27634 #include <linux/types.h>
27635 #include <linux/audit.h>
27636 #include <linux/mutex.h>
27637 +#include <linux/vs_context.h>
27638 +#include <linux/vs_network.h>
27639 +#include <linux/vs_limit.h>
27641 #include <net/net_namespace.h>
27642 #include <net/sock.h>
27643 @@ -1776,6 +1779,8 @@ static struct sock *netlink_seq_socket_i
27644 sk_for_each(s, node, &hash->table[j]) {
27645 if (sock_net(s) != seq_file_net(seq))
27647 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
27651 iter->hash_idx = j;
27652 @@ -1810,7 +1815,8 @@ static void *netlink_seq_next(struct seq
27656 - } while (s && sock_net(s) != seq_file_net(seq));
27657 + } while (s && (sock_net(s) != seq_file_net(seq) ||
27658 + !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)));
27662 @@ -1822,7 +1828,8 @@ static void *netlink_seq_next(struct seq
27664 for (; j <= hash->mask; j++) {
27665 s = sk_head(&hash->table[j]);
27666 - while (s && sock_net(s) != seq_file_net(seq))
27667 + while (s && (sock_net(s) != seq_file_net(seq) ||
27668 + !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)))
27672 diff -NurpP --minimal linux-2.6.29.1/net/sctp/ipv6.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sctp/ipv6.c
27673 --- linux-2.6.29.1/net/sctp/ipv6.c 2009-03-24 14:22:48.000000000 +0100
27674 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sctp/ipv6.c 2009-03-24 20:58:14.000000000 +0100
27675 @@ -317,7 +317,8 @@ static void sctp_v6_get_saddr(struct sct
27676 dst ? ip6_dst_idev(dst)->dev : NULL,
27677 &daddr->v6.sin6_addr,
27678 inet6_sk(&sk->inet.sk)->srcprefs,
27679 - &saddr->v6.sin6_addr);
27680 + &saddr->v6.sin6_addr,
27681 + asoc->base.sk->sk_nx_info);
27682 SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: %pI6\n",
27683 &saddr->v6.sin6_addr);
27685 diff -NurpP --minimal linux-2.6.29.1/net/socket.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/socket.c
27686 --- linux-2.6.29.1/net/socket.c 2009-03-24 14:22:48.000000000 +0100
27687 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/socket.c 2009-03-24 14:48:37.000000000 +0100
27690 #include <net/sock.h>
27691 #include <linux/netfilter.h>
27692 +#include <linux/vs_base.h>
27693 +#include <linux/vs_socket.h>
27694 +#include <linux/vs_inet.h>
27695 +#include <linux/vs_inet6.h>
27697 static int sock_no_open(struct inode *irrelevant, struct file *dontcare);
27698 static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
27699 @@ -549,7 +553,7 @@ static inline int __sock_sendmsg(struct
27700 struct msghdr *msg, size_t size)
27702 struct sock_iocb *si = kiocb_to_siocb(iocb);
27708 @@ -560,7 +564,22 @@ static inline int __sock_sendmsg(struct
27712 - return sock->ops->sendmsg(iocb, sock, msg, size);
27713 + len = sock->ops->sendmsg(iocb, sock, msg, size);
27716 + vx_sock_send(sock->sk, size);
27718 + vx_sock_fail(sock->sk, size);
27720 + vxdprintk(VXD_CBIT(net, 7),
27721 + "__sock_sendmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
27723 + (sock->sk)?sock->sk->sk_nx_info:0,
27724 + (sock->sk)?sock->sk->sk_vx_info:0,
27725 + (sock->sk)?sock->sk->sk_xid:0,
27726 + (sock->sk)?sock->sk->sk_nid:0,
27727 + (unsigned int)size, len);
27731 int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
27732 @@ -629,7 +648,7 @@ EXPORT_SYMBOL_GPL(__sock_recv_timestamp)
27733 static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock,
27734 struct msghdr *msg, size_t size, int flags)
27738 struct sock_iocb *si = kiocb_to_siocb(iocb);
27741 @@ -642,7 +661,18 @@ static inline int __sock_recvmsg(struct
27745 - return sock->ops->recvmsg(iocb, sock, msg, size, flags);
27746 + len = sock->ops->recvmsg(iocb, sock, msg, size, flags);
27747 + if ((len >= 0) && sock->sk)
27748 + vx_sock_recv(sock->sk, len);
27749 + vxdprintk(VXD_CBIT(net, 7),
27750 + "__sock_recvmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
27752 + (sock->sk)?sock->sk->sk_nx_info:0,
27753 + (sock->sk)?sock->sk->sk_vx_info:0,
27754 + (sock->sk)?sock->sk->sk_xid:0,
27755 + (sock->sk)?sock->sk->sk_nid:0,
27756 + (unsigned int)size, len);
27760 int sock_recvmsg(struct socket *sock, struct msghdr *msg,
27761 @@ -1106,6 +1136,13 @@ static int __sock_create(struct net *net
27762 if (type < 0 || type >= SOCK_MAX)
27765 + if (!nx_check(0, VS_ADMIN)) {
27766 + if (family == PF_INET && !current_nx_info_has_v4())
27767 + return -EAFNOSUPPORT;
27768 + if (family == PF_INET6 && !current_nx_info_has_v6())
27769 + return -EAFNOSUPPORT;
27774 This uglymoron is moved from INET layer to here to avoid
27775 @@ -1238,6 +1275,7 @@ SYSCALL_DEFINE3(socket, int, family, int
27779 + set_bit(SOCK_USER_SOCKET, &sock->flags);
27780 retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
27783 @@ -1279,10 +1317,12 @@ SYSCALL_DEFINE4(socketpair, int, family,
27784 err = sock_create(family, type, protocol, &sock1);
27787 + set_bit(SOCK_USER_SOCKET, &sock1->flags);
27789 err = sock_create(family, type, protocol, &sock2);
27791 goto out_release_1;
27792 + set_bit(SOCK_USER_SOCKET, &sock2->flags);
27794 err = sock1->ops->socketpair(sock1, sock2);
27796 diff -NurpP --minimal linux-2.6.29.1/net/sunrpc/auth.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/auth.c
27797 --- linux-2.6.29.1/net/sunrpc/auth.c 2009-03-24 14:22:48.000000000 +0100
27798 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/auth.c 2009-03-24 21:00:49.000000000 +0100
27800 #include <linux/hash.h>
27801 #include <linux/sunrpc/clnt.h>
27802 #include <linux/spinlock.h>
27803 +#include <linux/vs_tag.h>
27806 # define RPCDBG_FACILITY RPCDBG_AUTH
27807 @@ -360,6 +361,7 @@ rpcauth_lookupcred(struct rpc_auth *auth
27808 memset(&acred, 0, sizeof(acred));
27809 acred.uid = cred->fsuid;
27810 acred.gid = cred->fsgid;
27811 + acred.tag = dx_current_tag();
27812 acred.group_info = get_group_info(((struct cred *)cred)->group_info);
27814 ret = auth->au_ops->lookup_cred(auth, &acred, flags);
27815 @@ -400,6 +402,7 @@ rpcauth_bind_root_cred(struct rpc_task *
27816 struct auth_cred acred = {
27819 + .tag = dx_current_tag(),
27821 struct rpc_cred *ret;
27823 diff -NurpP --minimal linux-2.6.29.1/net/sunrpc/auth_unix.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/auth_unix.c
27824 --- linux-2.6.29.1/net/sunrpc/auth_unix.c 2008-12-25 00:26:37.000000000 +0100
27825 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/auth_unix.c 2009-02-22 22:54:26.000000000 +0100
27826 @@ -11,12 +11,14 @@
27827 #include <linux/module.h>
27828 #include <linux/sunrpc/clnt.h>
27829 #include <linux/sunrpc/auth.h>
27830 +#include <linux/vs_tag.h>
27832 #define NFS_NGROUPS 16
27835 struct rpc_cred uc_base;
27838 gid_t uc_gids[NFS_NGROUPS];
27840 #define uc_uid uc_base.cr_uid
27841 @@ -78,6 +80,7 @@ unx_create_cred(struct rpc_auth *auth, s
27842 groups = NFS_NGROUPS;
27844 cred->uc_gid = acred->gid;
27845 + cred->uc_tag = acred->tag;
27846 for (i = 0; i < groups; i++)
27847 cred->uc_gids[i] = GROUP_AT(acred->group_info, i);
27848 if (i < NFS_NGROUPS)
27849 @@ -119,7 +122,9 @@ unx_match(struct auth_cred *acred, struc
27853 - if (cred->uc_uid != acred->uid || cred->uc_gid != acred->gid)
27854 + if (cred->uc_uid != acred->uid ||
27855 + cred->uc_gid != acred->gid ||
27856 + cred->uc_tag != acred->tag)
27859 if (acred->group_info != NULL)
27860 @@ -142,7 +147,7 @@ unx_marshal(struct rpc_task *task, __be3
27861 struct rpc_clnt *clnt = task->tk_client;
27862 struct unx_cred *cred = container_of(task->tk_msg.rpc_cred, struct unx_cred, uc_base);
27863 __be32 *base, *hold;
27867 *p++ = htonl(RPC_AUTH_UNIX);
27869 @@ -152,9 +157,12 @@ unx_marshal(struct rpc_task *task, __be3
27870 * Copy the UTS nodename captured when the client was created.
27872 p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen);
27873 + tag = task->tk_client->cl_tag;
27875 - *p++ = htonl((u32) cred->uc_uid);
27876 - *p++ = htonl((u32) cred->uc_gid);
27877 + *p++ = htonl((u32) TAGINO_UID(tag,
27878 + cred->uc_uid, cred->uc_tag));
27879 + *p++ = htonl((u32) TAGINO_GID(tag,
27880 + cred->uc_gid, cred->uc_tag));
27882 for (i = 0; i < 16 && cred->uc_gids[i] != (gid_t) NOGROUP; i++)
27883 *p++ = htonl((u32) cred->uc_gids[i]);
27884 diff -NurpP --minimal linux-2.6.29.1/net/sunrpc/clnt.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/clnt.c
27885 --- linux-2.6.29.1/net/sunrpc/clnt.c 2009-03-24 14:22:48.000000000 +0100
27886 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/sunrpc/clnt.c 2009-03-24 14:48:37.000000000 +0100
27888 #include <linux/utsname.h>
27889 #include <linux/workqueue.h>
27890 #include <linux/in6.h>
27891 +#include <linux/vs_cvirt.h>
27893 #include <linux/sunrpc/clnt.h>
27894 #include <linux/sunrpc/rpc_pipe_fs.h>
27895 @@ -335,6 +336,9 @@ struct rpc_clnt *rpc_create(struct rpc_c
27896 if (!(args->flags & RPC_CLNT_CREATE_QUIET))
27897 clnt->cl_chatty = 1;
27899 + /* TODO: handle RPC_CLNT_CREATE_TAGGED
27900 + if (args->flags & RPC_CLNT_CREATE_TAGGED)
27901 + clnt->cl_tag = 1; */
27904 EXPORT_SYMBOL_GPL(rpc_create);
27905 diff -NurpP --minimal linux-2.6.29.1/net/unix/af_unix.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/unix/af_unix.c
27906 --- linux-2.6.29.1/net/unix/af_unix.c 2009-03-24 14:22:48.000000000 +0100
27907 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/unix/af_unix.c 2009-03-24 14:48:37.000000000 +0100
27908 @@ -114,6 +114,8 @@
27909 #include <linux/mount.h>
27910 #include <net/checksum.h>
27911 #include <linux/security.h>
27912 +#include <linux/vs_context.h>
27913 +#include <linux/vs_limit.h>
27915 static struct hlist_head unix_socket_table[UNIX_HASH_SIZE + 1];
27916 static DEFINE_SPINLOCK(unix_table_lock);
27917 @@ -258,6 +260,8 @@ static struct sock *__unix_find_socket_b
27918 if (!net_eq(sock_net(s), net))
27921 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
27923 if (u->addr->len == len &&
27924 !memcmp(u->addr->name, sunname, len))
27926 @@ -2110,6 +2114,8 @@ static struct sock *unix_seq_idx(struct
27927 for (s = first_unix_socket(&iter->i); s; s = next_unix_socket(&iter->i, s)) {
27928 if (sock_net(s) != seq_file_net(seq))
27930 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
27935 @@ -2134,7 +2140,8 @@ static void *unix_seq_next(struct seq_fi
27936 sk = first_unix_socket(&iter->i);
27938 sk = next_unix_socket(&iter->i, sk);
27939 - while (sk && (sock_net(sk) != seq_file_net(seq)))
27940 + while (sk && (sock_net(sk) != seq_file_net(seq) ||
27941 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)))
27942 sk = next_unix_socket(&iter->i, sk);
27945 diff -NurpP --minimal linux-2.6.29.1/net/x25/af_x25.c linux-2.6.29.1-vs2.3.0.36.9-pre5/net/x25/af_x25.c
27946 --- linux-2.6.29.1/net/x25/af_x25.c 2008-12-25 00:26:37.000000000 +0100
27947 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/net/x25/af_x25.c 2009-02-22 22:54:26.000000000 +0100
27948 @@ -506,7 +506,10 @@ static int x25_create(struct net *net, s
27952 - sock_init_data(sock, sk);
27953 + sk->sk_socket = sock;
27954 + sk->sk_type = sock->type;
27955 + sk->sk_sleep = &sock->wait;
27958 x25_init_timers(sk);
27960 diff -NurpP --minimal linux-2.6.29.1/scripts/checksyscalls.sh linux-2.6.29.1-vs2.3.0.36.9-pre5/scripts/checksyscalls.sh
27961 --- linux-2.6.29.1/scripts/checksyscalls.sh 2008-12-25 00:26:37.000000000 +0100
27962 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/scripts/checksyscalls.sh 2009-02-22 22:54:26.000000000 +0100
27963 @@ -108,7 +108,6 @@ cat << EOF
27964 #define __IGNORE_afs_syscall
27965 #define __IGNORE_getpmsg
27966 #define __IGNORE_putpmsg
27967 -#define __IGNORE_vserver
27971 diff -NurpP --minimal linux-2.6.29.1/security/commoncap.c linux-2.6.29.1-vs2.3.0.36.9-pre5/security/commoncap.c
27972 --- linux-2.6.29.1/security/commoncap.c 2009-03-24 14:23:21.000000000 +0100
27973 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/security/commoncap.c 2009-03-25 00:34:48.000000000 +0100
27974 @@ -27,10 +27,11 @@
27975 #include <linux/sched.h>
27976 #include <linux/prctl.h>
27977 #include <linux/securebits.h>
27978 +#include <linux/vs_context.h>
27980 int cap_netlink_send(struct sock *sk, struct sk_buff *skb)
27982 - NETLINK_CB(skb).eff_cap = current_cap();
27983 + NETLINK_CB(skb).eff_cap = vx_mbcaps(current_cap());
27987 @@ -40,6 +41,7 @@ int cap_netlink_recv(struct sk_buff *skb
27992 EXPORT_SYMBOL(cap_netlink_recv);
27995 @@ -60,7 +62,22 @@ EXPORT_SYMBOL(cap_netlink_recv);
27996 int cap_capable(struct task_struct *tsk, const struct cred *cred, int cap,
27999 - return cap_raised(cred->cap_effective, cap) ? 0 : -EPERM;
28000 + struct vx_info *vxi = tsk->vx_info;
28003 + printk("cap_capable() VXF_STATE_SETUP = %llx, raised = %x, eff = %08x:%08x\n",
28004 + vx_info_flags(vxi, VXF_STATE_SETUP, 0),
28005 + cap_raised(tsk->cap_effective, cap),
28006 + tsk->cap_effective.cap[1], tsk->cap_effective.cap[0]);
28009 + /* special case SETUP */
28010 + if (vx_info_flags(vxi, VXF_STATE_SETUP, 0) &&
28011 + /* FIXME: maybe use cred instead? */
28012 + cap_raised(tsk->cred->cap_effective, cap))
28015 + return vx_cap_raised(vxi, cred->cap_effective, cap) ? 0 : -EPERM;
28019 @@ -586,7 +603,7 @@ int cap_inode_setxattr(struct dentry *de
28021 if (!strncmp(name, XATTR_SECURITY_PREFIX,
28022 sizeof(XATTR_SECURITY_PREFIX) - 1) &&
28023 - !capable(CAP_SYS_ADMIN))
28024 + !vx_capable(CAP_SYS_ADMIN, VXC_FS_SECURITY))
28028 @@ -931,7 +948,8 @@ error:
28030 int cap_syslog(int type)
28032 - if ((type != 3 && type != 10) && !capable(CAP_SYS_ADMIN))
28033 + if ((type != 3 && type != 10) &&
28034 + !vx_capable(CAP_SYS_ADMIN, VXC_SYSLOG))
28038 @@ -953,3 +971,4 @@ int cap_vm_enough_memory(struct mm_struc
28040 return __vm_enough_memory(mm, pages, cap_sys_admin);
28043 diff -NurpP --minimal linux-2.6.29.1/security/selinux/hooks.c linux-2.6.29.1-vs2.3.0.36.9-pre5/security/selinux/hooks.c
28044 --- linux-2.6.29.1/security/selinux/hooks.c 2009-03-24 14:23:21.000000000 +0100
28045 +++ linux-2.6.29.1-vs2.3.0.36.9-pre5/security/selinux/hooks.c 2009-03-24 14:48:37.000000000 +0100
28047 #include <linux/dccp.h>
28048 #include <linux/quota.h>
28049 #include <linux/un.h> /* for Unix socket types */
28050 -#include <net/af_unix.h> /* for Unix socket types */
28051 #include <linux/parser.h>
28052 #include <linux/nfs_mount.h>
28053 #include <net/ipv6.h>