1 diff -NurpP --minimal linux-2.6.32.1/arch/alpha/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/Kconfig
2 --- linux-2.6.32.1/arch/alpha/Kconfig 2009-12-03 20:01:49.000000000 +0100
3 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/Kconfig 2009-12-03 20:04:56.000000000 +0100
4 @@ -674,6 +674,8 @@ config DUMMY_CONSOLE
8 +source "kernel/vserver/Kconfig"
10 source "security/Kconfig"
12 source "crypto/Kconfig"
13 diff -NurpP --minimal linux-2.6.32.1/arch/alpha/kernel/entry.S linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/entry.S
14 --- linux-2.6.32.1/arch/alpha/kernel/entry.S 2009-06-11 17:11:46.000000000 +0200
15 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/entry.S 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/alpha/kernel/osf_sys.c linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/osf_sys.c
49 --- linux-2.6.32.1/arch/alpha/kernel/osf_sys.c 2009-09-10 15:25:14.000000000 +0200
50 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/osf_sys.c 2009-12-03 20:04:56.000000000 +0100
51 @@ -872,7 +872,7 @@ SYSCALL_DEFINE2(osf_gettimeofday, struct
55 - do_gettimeofday(&ktv);
56 + vx_gettimeofday(&ktv);
57 if (put_tv32(tv, &ktv))
60 diff -NurpP --minimal linux-2.6.32.1/arch/alpha/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/ptrace.c
61 --- linux-2.6.32.1/arch/alpha/kernel/ptrace.c 2009-09-10 15:25:14.000000000 +0200
62 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/alpha/kernel/systbls.S linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/systbls.S
72 --- linux-2.6.32.1/arch/alpha/kernel/systbls.S 2009-03-24 14:18:08.000000000 +0100
73 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/systbls.S 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/alpha/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/traps.c
84 --- linux-2.6.32.1/arch/alpha/kernel/traps.c 2009-06-11 17:11:46.000000000 +0200
85 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/kernel/traps.c 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/alpha/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/mm/fault.c
97 --- linux-2.6.32.1/arch/alpha/mm/fault.c 2009-09-10 15:25:14.000000000 +0200
98 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/alpha/mm/fault.c 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/arm/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/arm/Kconfig
111 --- linux-2.6.32.1/arch/arm/Kconfig 2009-12-03 20:01:49.000000000 +0100
112 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/arm/Kconfig 2009-12-03 20:04:56.000000000 +0100
113 @@ -1512,6 +1512,8 @@ 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-2.6.32.1/arch/arm/kernel/calls.S linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/calls.S
123 --- linux-2.6.32.1/arch/arm/kernel/calls.S 2009-12-03 20:01:50.000000000 +0100
124 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/calls.S 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/arm/kernel/process.c linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/process.c
135 --- linux-2.6.32.1/arch/arm/kernel/process.c 2009-12-03 20:01:50.000000000 +0100
136 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/process.c 2009-12-03 20:04:56.000000000 +0100
137 @@ -269,7 +269,8 @@ 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-2.6.32.1/arch/arm/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/traps.c
148 --- linux-2.6.32.1/arch/arm/kernel/traps.c 2009-12-03 20:01:50.000000000 +0100
149 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/arm/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
150 @@ -234,8 +234,8 @@ static void __die(const char *str, int e
151 sysfs_printk_last_file();
154 - printk(KERN_EMERG "Process %.*s (pid: %d, stack limit = 0x%p)\n",
155 - TASK_COMM_LEN, tsk->comm, task_pid_nr(tsk), thread + 1);
156 + printk(KERN_EMERG "Process %.*s (pid: %d:#%u, stack limit = 0x%p)\n",
157 + TASK_COMM_LEN, tsk->comm, task_pid_nr(tsk), tsk->xid, thread + 1);
159 if (!user_mode(regs) || in_interrupt()) {
160 dump_mem(KERN_EMERG, "Stack: ", regs->ARM_sp,
161 diff -NurpP --minimal linux-2.6.32.1/arch/avr32/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/avr32/mm/fault.c
162 --- linux-2.6.32.1/arch/avr32/mm/fault.c 2009-09-10 15:25:20.000000000 +0200
163 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/avr32/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
164 @@ -216,7 +216,8 @@ out_of_memory:
165 down_read(&mm->mmap_sem);
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);
172 do_group_exit(SIGKILL);
174 diff -NurpP --minimal linux-2.6.32.1/arch/cris/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/cris/Kconfig
175 --- linux-2.6.32.1/arch/cris/Kconfig 2009-06-11 17:11:56.000000000 +0200
176 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/cris/Kconfig 2009-12-03 20:04:56.000000000 +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-2.6.32.1/arch/cris/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/cris/mm/fault.c
187 --- linux-2.6.32.1/arch/cris/mm/fault.c 2009-12-03 20:01:56.000000000 +0100
188 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/cris/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
189 @@ -245,7 +245,8 @@ do_page_fault(unsigned long address, str
192 up_read(&mm->mmap_sem);
193 - printk("VM: killing process %s\n", tsk->comm);
194 + printk("VM: killing process %s(%d:#%u)\n",
195 + tsk->comm, task_pid_nr(tsk), tsk->xid);
199 diff -NurpP --minimal linux-2.6.32.1/arch/frv/kernel/kernel_thread.S linux-2.6.32.1-vs2.3.0.36.27/arch/frv/kernel/kernel_thread.S
200 --- linux-2.6.32.1/arch/frv/kernel/kernel_thread.S 2008-12-25 00:26:37.000000000 +0100
201 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/frv/kernel/kernel_thread.S 2009-12-03 20:04:56.000000000 +0100
202 @@ -37,7 +37,7 @@ kernel_thread:
204 # start by forking the current process, but with shared VM
205 setlos.p #__NR_clone,gr7 ; syscall number
206 - ori gr10,#CLONE_VM,gr8 ; first syscall arg [clone_flags]
207 + ori gr10,#CLONE_KT,gr8 ; first syscall arg [clone_flags]
208 sethi.p #0xe4e4,gr9 ; second syscall arg [newsp]
210 setlos.p #0,gr10 ; third syscall arg [parent_tidptr]
211 diff -NurpP --minimal linux-2.6.32.1/arch/frv/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/frv/mm/fault.c
212 --- linux-2.6.32.1/arch/frv/mm/fault.c 2009-09-10 15:25:22.000000000 +0200
213 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/frv/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
214 @@ -257,7 +257,8 @@ asmlinkage void do_page_fault(int datamm
217 up_read(&mm->mmap_sem);
218 - printk("VM: killing process %s\n", current->comm);
219 + printk("VM: killing process %s(%d:#%u)\n",
220 + current->comm, task_pid_nr(current), current->xid);
221 if (user_mode(__frame))
222 do_group_exit(SIGKILL);
224 diff -NurpP --minimal linux-2.6.32.1/arch/h8300/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/h8300/Kconfig
225 --- linux-2.6.32.1/arch/h8300/Kconfig 2009-03-24 14:18:24.000000000 +0100
226 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/h8300/Kconfig 2009-12-03 20:04:56.000000000 +0100
227 @@ -226,6 +226,8 @@ source "fs/Kconfig"
229 source "arch/h8300/Kconfig.debug"
231 +source "kernel/vserver/Kconfig"
233 source "security/Kconfig"
235 source "crypto/Kconfig"
236 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/ia32/ia32_entry.S linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/ia32/ia32_entry.S
237 --- linux-2.6.32.1/arch/ia64/ia32/ia32_entry.S 2009-06-11 17:11:57.000000000 +0200
238 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/ia32/ia32_entry.S 2009-12-03 20:04:56.000000000 +0100
239 @@ -451,7 +451,7 @@ ia32_syscall_table:
240 data8 sys_tgkill /* 270 */
241 data8 compat_sys_utimes
242 data8 sys32_fadvise64_64
243 - data8 sys_ni_syscall
244 + data8 sys32_vserver
246 data8 sys_ni_syscall /* 275 */
248 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/Kconfig
249 --- linux-2.6.32.1/arch/ia64/Kconfig 2009-12-03 20:01:56.000000000 +0100
250 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/Kconfig 2009-12-03 20:04:56.000000000 +0100
251 @@ -685,6 +685,8 @@ source "fs/Kconfig"
253 source "arch/ia64/Kconfig.debug"
255 +source "kernel/vserver/Kconfig"
257 source "security/Kconfig"
259 source "crypto/Kconfig"
260 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/kernel/entry.S linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/entry.S
261 --- linux-2.6.32.1/arch/ia64/kernel/entry.S 2009-09-10 15:25:22.000000000 +0200
262 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/entry.S 2009-12-03 20:04:56.000000000 +0100
263 @@ -1753,7 +1753,7 @@ sys_call_table:
265 data8 sys_mq_getsetattr
267 - data8 sys_ni_syscall // reserved for vserver
269 data8 sys_waitid // 1270
271 data8 sys_request_key
272 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/kernel/perfmon.c linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/perfmon.c
273 --- linux-2.6.32.1/arch/ia64/kernel/perfmon.c 2009-09-10 15:25:22.000000000 +0200
274 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/perfmon.c 2009-12-03 20:04:56.000000000 +0100
276 #include <linux/rcupdate.h>
277 #include <linux/completion.h>
278 #include <linux/tracehook.h>
279 +#include <linux/vs_memory.h>
281 #include <asm/errno.h>
282 #include <asm/intrinsics.h>
283 @@ -2372,7 +2373,7 @@ pfm_smpl_buffer_alloc(struct task_struct
285 insert_vm_struct(mm, vma);
287 - mm->total_vm += size >> PAGE_SHIFT;
288 + vx_vmpages_add(mm, size >> PAGE_SHIFT);
289 vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
291 up_write(&task->mm->mmap_sem);
292 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/kernel/process.c linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/process.c
293 --- linux-2.6.32.1/arch/ia64/kernel/process.c 2009-12-03 20:01:56.000000000 +0100
294 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/process.c 2009-12-03 20:04:56.000000000 +0100
295 @@ -110,8 +110,8 @@ show_regs (struct pt_regs *regs)
296 unsigned long ip = regs->cr_iip + ia64_psr(regs)->ri;
299 - printk("\nPid: %d, CPU %d, comm: %20s\n", task_pid_nr(current),
300 - smp_processor_id(), current->comm);
301 + printk("\nPid: %d[#%u], CPU %d, comm: %20s\n", task_pid_nr(current),
302 + current->xid, smp_processor_id(), current->comm);
303 printk("psr : %016lx ifs : %016lx ip : [<%016lx>] %s (%s)\n",
304 regs->cr_ipsr, regs->cr_ifs, ip, print_tainted(),
305 init_utsname()->release);
306 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/ptrace.c
307 --- linux-2.6.32.1/arch/ia64/kernel/ptrace.c 2009-09-10 15:25:22.000000000 +0200
308 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +0100
310 #include <linux/regset.h>
311 #include <linux/elf.h>
312 #include <linux/tracehook.h>
313 +#include <linux/vs_base.h>
315 #include <asm/pgtable.h>
316 #include <asm/processor.h>
317 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/traps.c
318 --- linux-2.6.32.1/arch/ia64/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
319 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
320 @@ -60,8 +60,9 @@ die (const char *str, struct pt_regs *re
323 if (++die.lock_owner_depth < 3) {
324 - printk("%s[%d]: %s %ld [%d]\n",
325 - current->comm, task_pid_nr(current), str, err, ++die_counter);
326 + printk("%s[%d[#%u]]: %s %ld [%d]\n",
327 + current->comm, task_pid_nr(current), current->xid,
328 + str, err, ++die_counter);
329 if (notify_die(DIE_OOPS, str, regs, err, 255, SIGSEGV)
332 @@ -324,8 +325,9 @@ handle_fpu_swa (int fp_fault, struct pt_
333 if ((last.count & 15) < 5 && (ia64_fetchadd(1, &last.count, acq) & 15) < 5) {
334 last.time = current_jiffies + 5 * HZ;
336 - "%s(%d): floating-point assist fault at ip %016lx, isr %016lx\n",
337 - current->comm, task_pid_nr(current), regs->cr_iip + ia64_psr(regs)->ri, isr);
338 + "%s(%d[#%u]): floating-point assist fault at ip %016lx, isr %016lx\n",
339 + current->comm, task_pid_nr(current), current->xid,
340 + regs->cr_iip + ia64_psr(regs)->ri, isr);
344 diff -NurpP --minimal linux-2.6.32.1/arch/ia64/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/mm/fault.c
345 --- linux-2.6.32.1/arch/ia64/mm/fault.c 2009-09-10 15:25:23.000000000 +0200
346 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/ia64/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
348 #include <linux/interrupt.h>
349 #include <linux/kprobes.h>
350 #include <linux/kdebug.h>
351 +#include <linux/vs_memory.h>
353 #include <asm/pgtable.h>
354 #include <asm/processor.h>
355 @@ -281,7 +282,8 @@ ia64_do_page_fault (unsigned long addres
356 down_read(&mm->mmap_sem);
359 - printk(KERN_CRIT "VM: killing process %s\n", current->comm);
360 + printk(KERN_CRIT "VM: killing process %s(%d:#%u)\n",
361 + current->comm, task_pid_nr(current), current->xid);
363 do_group_exit(SIGKILL);
365 diff -NurpP --minimal linux-2.6.32.1/arch/m32r/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/m32r/kernel/traps.c
366 --- linux-2.6.32.1/arch/m32r/kernel/traps.c 2009-12-03 20:01:57.000000000 +0100
367 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m32r/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
368 @@ -196,8 +196,9 @@ static void show_registers(struct pt_reg
370 printk("SPI: %08lx\n", sp);
372 - printk("Process %s (pid: %d, process nr: %d, stackpage=%08lx)",
373 - current->comm, task_pid_nr(current), 0xffff & i, 4096+(unsigned long)current);
374 + printk("Process %s (pid: %d[#%u], process nr: %d, stackpage=%08lx)",
375 + current->comm, task_pid_nr(current), current->xid,
376 + 0xffff & i, 4096+(unsigned long)current);
379 * When in-kernel, we also print out the stack and code at the
380 diff -NurpP --minimal linux-2.6.32.1/arch/m32r/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/m32r/mm/fault.c
381 --- linux-2.6.32.1/arch/m32r/mm/fault.c 2009-09-10 15:25:23.000000000 +0200
382 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m32r/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
383 @@ -276,7 +276,8 @@ out_of_memory:
384 down_read(&mm->mmap_sem);
387 - printk("VM: killing process %s\n", tsk->comm);
388 + printk("VM: killing process %s(%d:#%u)\n",
389 + tsk->comm, task_pid_nr(tsk), tsk->xid);
390 if (error_code & ACE_USERMODE)
391 do_group_exit(SIGKILL);
393 diff -NurpP --minimal linux-2.6.32.1/arch/m68k/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/Kconfig
394 --- linux-2.6.32.1/arch/m68k/Kconfig 2009-12-03 20:01:57.000000000 +0100
395 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/Kconfig 2009-12-03 20:04:56.000000000 +0100
396 @@ -622,6 +622,8 @@ source "fs/Kconfig"
398 source "arch/m68k/Kconfig.debug"
400 +source "kernel/vserver/Kconfig"
402 source "security/Kconfig"
404 source "crypto/Kconfig"
405 diff -NurpP --minimal linux-2.6.32.1/arch/m68k/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/kernel/ptrace.c
406 --- linux-2.6.32.1/arch/m68k/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
407 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +0100
409 #include <linux/ptrace.h>
410 #include <linux/user.h>
411 #include <linux/signal.h>
412 +#include <linux/vs_base.h>
414 #include <asm/uaccess.h>
415 #include <asm/page.h>
416 @@ -269,6 +270,8 @@ long arch_ptrace(struct task_struct *chi
417 ret = ptrace_request(child, request, addr, data);
420 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
425 diff -NurpP --minimal linux-2.6.32.1/arch/m68k/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/kernel/traps.c
426 --- linux-2.6.32.1/arch/m68k/kernel/traps.c 2009-09-10 15:25:23.000000000 +0200
427 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
428 @@ -906,8 +906,8 @@ void show_registers(struct pt_regs *regs
429 printk("d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
430 regs->d4, regs->d5, regs->a0, regs->a1);
432 - printk("Process %s (pid: %d, task=%p)\n",
433 - current->comm, task_pid_nr(current), current);
434 + printk("Process %s (pid: %d[#%u], task=%p)\n",
435 + current->comm, task_pid_nr(current), current->xid, current);
436 addr = (unsigned long)&fp->un;
437 printk("Frame format=%X ", regs->format);
438 switch (regs->format) {
439 diff -NurpP --minimal linux-2.6.32.1/arch/m68k/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/mm/fault.c
440 --- linux-2.6.32.1/arch/m68k/mm/fault.c 2009-09-10 15:25:23.000000000 +0200
441 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68k/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
442 @@ -186,7 +186,8 @@ out_of_memory:
446 - printk("VM: killing process %s\n", current->comm);
447 + printk("VM: killing process %s(%d:#%u)\n",
448 + current->comm, task_pid_nr(current), current->xid);
450 do_group_exit(SIGKILL);
452 diff -NurpP --minimal linux-2.6.32.1/arch/m68knommu/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/m68knommu/Kconfig
453 --- linux-2.6.32.1/arch/m68knommu/Kconfig 2009-12-03 20:01:57.000000000 +0100
454 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68knommu/Kconfig 2009-12-03 20:04:56.000000000 +0100
455 @@ -727,6 +727,8 @@ source "fs/Kconfig"
457 source "arch/m68knommu/Kconfig.debug"
459 +source "kernel/vserver/Kconfig"
461 source "security/Kconfig"
463 source "crypto/Kconfig"
464 diff -NurpP --minimal linux-2.6.32.1/arch/m68knommu/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/m68knommu/kernel/traps.c
465 --- linux-2.6.32.1/arch/m68knommu/kernel/traps.c 2009-09-10 15:25:23.000000000 +0200
466 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/m68knommu/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
467 @@ -78,8 +78,9 @@ void die_if_kernel(char *str, struct pt_
468 printk(KERN_EMERG "d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
469 fp->d4, fp->d5, fp->a0, fp->a1);
471 - printk(KERN_EMERG "Process %s (pid: %d, stackpage=%08lx)\n",
472 - current->comm, current->pid, PAGE_SIZE+(unsigned long)current);
473 + printk(KERN_EMERG "Process %s (pid: %d[#%u], stackpage=%08lx)\n",
474 + current->comm, task_pid_nr(current), current->xid,
475 + PAGE_SIZE+(unsigned long)current);
476 show_stack(NULL, (unsigned long *)(fp + 1));
477 add_taint(TAINT_DIE);
479 diff -NurpP --minimal linux-2.6.32.1/arch/microblaze/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/microblaze/mm/fault.c
480 --- linux-2.6.32.1/arch/microblaze/mm/fault.c 2009-09-10 15:25:24.000000000 +0200
481 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/microblaze/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
482 @@ -279,7 +279,8 @@ out_of_memory:
485 up_read(&mm->mmap_sem);
486 - printk(KERN_WARNING "VM: killing process %s\n", current->comm);
487 + printk(KERN_WARNING "VM: killing process %s(%d:#%u)\n",
488 + current->comm, task_pid_nr(current), current->xid);
491 bad_page_fault(regs, address, SIGKILL);
492 diff -NurpP --minimal linux-2.6.32.1/arch/mips/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/mips/Kconfig
493 --- linux-2.6.32.1/arch/mips/Kconfig 2009-12-03 20:01:58.000000000 +0100
494 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/Kconfig 2009-12-03 20:04:56.000000000 +0100
495 @@ -2188,6 +2188,8 @@ source "fs/Kconfig"
497 source "arch/mips/Kconfig.debug"
499 +source "kernel/vserver/Kconfig"
501 source "security/Kconfig"
503 source "crypto/Kconfig"
504 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/ptrace.c
505 --- linux-2.6.32.1/arch/mips/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
506 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +0100
508 #include <linux/security.h>
509 #include <linux/audit.h>
510 #include <linux/seccomp.h>
511 +#include <linux/vs_base.h>
513 #include <asm/byteorder.h>
515 @@ -259,6 +260,9 @@ long arch_ptrace(struct task_struct *chi
519 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
523 /* when I and D space are separate, these will need to be fixed. */
524 case PTRACE_PEEKTEXT: /* read word at location addr. */
525 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/scall32-o32.S linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall32-o32.S
526 --- linux-2.6.32.1/arch/mips/kernel/scall32-o32.S 2009-12-03 20:01:59.000000000 +0100
527 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall32-o32.S 2009-12-03 20:04:56.000000000 +0100
528 @@ -525,7 +525,7 @@ einval: li v0, -ENOSYS
529 sys sys_mq_timedreceive 5
530 sys sys_mq_notify 2 /* 4275 */
531 sys sys_mq_getsetattr 3
532 - sys sys_ni_syscall 0 /* sys_vserver */
535 sys sys_ni_syscall 0 /* available, was setaltroot */
536 sys sys_add_key 5 /* 4280 */
537 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/scall64-64.S linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-64.S
538 --- linux-2.6.32.1/arch/mips/kernel/scall64-64.S 2009-12-03 20:01:59.000000000 +0100
539 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-64.S 2009-12-03 20:04:56.000000000 +0100
540 @@ -362,7 +362,7 @@ sys_call_table:
541 PTR sys_mq_timedreceive
543 PTR sys_mq_getsetattr /* 5235 */
544 - PTR sys_ni_syscall /* sys_vserver */
547 PTR sys_ni_syscall /* available, was setaltroot */
549 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/scall64-n32.S linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-n32.S
550 --- linux-2.6.32.1/arch/mips/kernel/scall64-n32.S 2009-12-03 20:01:59.000000000 +0100
551 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-n32.S 2009-12-03 20:04:56.000000000 +0100
552 @@ -360,7 +360,7 @@ EXPORT(sysn32_call_table)
553 PTR compat_sys_mq_timedreceive
554 PTR compat_sys_mq_notify
555 PTR compat_sys_mq_getsetattr
556 - PTR sys_ni_syscall /* 6240, sys_vserver */
557 + PTR sys32_vserver /* 6240 */
558 PTR compat_sys_waitid
559 PTR sys_ni_syscall /* available, was setaltroot */
561 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/scall64-o32.S linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-o32.S
562 --- linux-2.6.32.1/arch/mips/kernel/scall64-o32.S 2009-12-03 20:01:59.000000000 +0100
563 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/scall64-o32.S 2009-12-03 20:04:56.000000000 +0100
564 @@ -480,7 +480,7 @@ sys_call_table:
565 PTR compat_sys_mq_timedreceive
566 PTR compat_sys_mq_notify /* 4275 */
567 PTR compat_sys_mq_getsetattr
568 - PTR sys_ni_syscall /* sys_vserver */
571 PTR sys_ni_syscall /* available, was setaltroot */
572 PTR sys_add_key /* 4280 */
573 diff -NurpP --minimal linux-2.6.32.1/arch/mips/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/traps.c
574 --- linux-2.6.32.1/arch/mips/kernel/traps.c 2009-12-03 20:01:59.000000000 +0100
575 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mips/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
576 @@ -335,9 +335,10 @@ void show_registers(const struct pt_regs
580 - printk("Process %s (pid: %d, threadinfo=%p, task=%p, tls=%0*lx)\n",
581 - current->comm, current->pid, current_thread_info(), current,
582 - field, current_thread_info()->tp_value);
583 + printk("Process %s (pid: %d:#%u, threadinfo=%p, task=%p, tls=%0*lx)\n",
584 + current->comm, task_pid_nr(current), current->xid,
585 + current_thread_info(), current,
586 + field, current_thread_info()->tp_value);
587 if (cpu_has_userlocal) {
590 diff -NurpP --minimal linux-2.6.32.1/arch/mn10300/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/mn10300/mm/fault.c
591 --- linux-2.6.32.1/arch/mn10300/mm/fault.c 2009-09-10 15:25:39.000000000 +0200
592 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/mn10300/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
593 @@ -339,7 +339,8 @@ no_context:
595 up_read(&mm->mmap_sem);
596 monitor_signal(regs);
597 - printk(KERN_ALERT "VM: killing process %s\n", tsk->comm);
598 + printk(KERN_ALERT "VM: killing process %s(%d:#%u)\n",
599 + tsk->comm, task_pid_nr(tsk), tsk->xid);
600 if ((fault_code & MMUFCR_xFC_ACCESS) == MMUFCR_xFC_ACCESS_USR)
603 diff -NurpP --minimal linux-2.6.32.1/arch/parisc/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/Kconfig
604 --- linux-2.6.32.1/arch/parisc/Kconfig 2009-12-03 20:02:00.000000000 +0100
605 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/Kconfig 2009-12-03 20:04:56.000000000 +0100
606 @@ -294,6 +294,8 @@ source "fs/Kconfig"
608 source "arch/parisc/Kconfig.debug"
610 +source "kernel/vserver/Kconfig"
612 source "security/Kconfig"
614 source "crypto/Kconfig"
615 diff -NurpP --minimal linux-2.6.32.1/arch/parisc/kernel/syscall_table.S linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/kernel/syscall_table.S
616 --- linux-2.6.32.1/arch/parisc/kernel/syscall_table.S 2009-12-03 20:02:00.000000000 +0100
617 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/kernel/syscall_table.S 2009-12-03 20:04:56.000000000 +0100
619 ENTRY_COMP(mbind) /* 260 */
620 ENTRY_COMP(get_mempolicy)
621 ENTRY_COMP(set_mempolicy)
622 - ENTRY_SAME(ni_syscall) /* 263: reserved for vserver */
623 + ENTRY_DIFF(vserver)
625 ENTRY_SAME(request_key) /* 265 */
627 diff -NurpP --minimal linux-2.6.32.1/arch/parisc/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/kernel/traps.c
628 --- linux-2.6.32.1/arch/parisc/kernel/traps.c 2009-09-10 15:25:40.000000000 +0200
629 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
630 @@ -236,8 +236,9 @@ void die_if_kernel(char *str, struct pt_
634 - printk(KERN_CRIT "%s (pid %d): %s (code %ld) at " RFMT "\n",
635 - current->comm, task_pid_nr(current), str, err, regs->iaoq[0]);
636 + printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld) at " RFMT "\n",
637 + current->comm, task_pid_nr(current), current->xid,
638 + str, err, regs->iaoq[0]);
639 #ifdef PRINT_USER_FAULTS
640 /* XXX for debugging only */
642 @@ -270,8 +271,8 @@ void die_if_kernel(char *str, struct pt_
643 pdc_console_restart();
646 - printk(KERN_CRIT "%s (pid %d): %s (code %ld)\n",
647 - current->comm, task_pid_nr(current), str, err);
648 + printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld)\n",
649 + current->comm, task_pid_nr(current), current->xid, str, err);
651 /* Wot's wrong wif bein' racy? */
652 if (current->thread.flags & PARISC_KERNEL_DEATH) {
653 diff -NurpP --minimal linux-2.6.32.1/arch/parisc/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/mm/fault.c
654 --- linux-2.6.32.1/arch/parisc/mm/fault.c 2009-09-10 15:25:40.000000000 +0200
655 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/parisc/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
656 @@ -237,8 +237,9 @@ bad_area:
658 #ifdef PRINT_USER_FAULTS
659 printk(KERN_DEBUG "\n");
660 - printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n",
661 - task_pid_nr(tsk), tsk->comm, code, address);
662 + printk(KERN_DEBUG "do_page_fault() pid=%d:#%u "
663 + "command='%s' type=%lu address=0x%08lx\n",
664 + task_pid_nr(tsk), tsk->xid, tsk->comm, code, address);
666 printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n",
667 vma->vm_start, vma->vm_end);
668 @@ -264,7 +265,8 @@ no_context:
671 up_read(&mm->mmap_sem);
672 - printk(KERN_CRIT "VM: killing process %s\n", current->comm);
673 + printk(KERN_CRIT "VM: killing process %s(%d:#%u)\n",
674 + current->comm, current->pid, current->xid);
676 do_group_exit(SIGKILL);
678 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/include/asm/unistd.h linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/include/asm/unistd.h
679 --- linux-2.6.32.1/arch/powerpc/include/asm/unistd.h 2009-12-03 20:02:01.000000000 +0100
680 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/include/asm/unistd.h 2009-12-03 20:04:56.000000000 +0100
683 #define __NR_rtas 255
684 #define __NR_sys_debug_setcontext 256
685 -/* Number 257 is reserved for vserver */
686 +#define __NR_vserver 257
687 #define __NR_migrate_pages 258
688 #define __NR_mbind 259
689 #define __NR_get_mempolicy 260
690 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/Kconfig
691 --- linux-2.6.32.1/arch/powerpc/Kconfig 2009-12-03 20:02:00.000000000 +0100
692 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/Kconfig 2009-12-03 20:04:56.000000000 +0100
693 @@ -943,6 +943,8 @@ source "lib/Kconfig"
695 source "arch/powerpc/Kconfig.debug"
697 +source "kernel/vserver/Kconfig"
699 source "security/Kconfig"
702 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/kernel/irq.c linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/irq.c
703 --- linux-2.6.32.1/arch/powerpc/kernel/irq.c 2009-12-03 20:02:01.000000000 +0100
704 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/irq.c 2009-12-03 20:04:56.000000000 +0100
706 #include <linux/pci.h>
707 #include <linux/debugfs.h>
708 #include <linux/perf_event.h>
709 +// #include <linux/vs_context.h>
711 #include <asm/uaccess.h>
712 #include <asm/system.h>
713 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/kernel/process.c linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/process.c
714 --- linux-2.6.32.1/arch/powerpc/kernel/process.c 2009-12-03 20:02:02.000000000 +0100
715 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/process.c 2009-12-03 20:04:56.000000000 +0100
716 @@ -519,8 +519,9 @@ void show_regs(struct pt_regs * regs)
718 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
720 - printk("TASK = %p[%d] '%s' THREAD: %p",
721 - current, task_pid_nr(current), current->comm, task_thread_info(current));
722 + printk("TASK = %p[%d,#%u] '%s' THREAD: %p",
723 + current, task_pid_nr(current), current->xid,
724 + current->comm, task_thread_info(current));
727 printk(" CPU: %d", raw_smp_processor_id());
728 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/kernel/traps.c linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/traps.c
729 --- linux-2.6.32.1/arch/powerpc/kernel/traps.c 2009-09-10 15:25:41.000000000 +0200
730 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/traps.c 2009-12-03 20:04:56.000000000 +0100
731 @@ -931,8 +931,9 @@ void nonrecoverable_exception(struct pt_
733 void trace_syscall(struct pt_regs *regs)
735 - printk("Task: %p(%d), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
736 - current, task_pid_nr(current), regs->nip, regs->link, regs->gpr[0],
737 + printk("Task: %p(%d[#%u]), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
738 + current, task_pid_nr(current), current->xid,
739 + regs->nip, regs->link, regs->gpr[0],
740 regs->ccr&0x10000000?"Error=":"", regs->gpr[3], print_tainted());
743 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/kernel/vdso.c linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/vdso.c
744 --- linux-2.6.32.1/arch/powerpc/kernel/vdso.c 2009-12-03 20:02:02.000000000 +0100
745 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/kernel/vdso.c 2009-12-03 20:04:56.000000000 +0100
747 #include <linux/security.h>
748 #include <linux/bootmem.h>
749 #include <linux/lmb.h>
750 +#include <linux/vs_memory.h>
752 #include <asm/pgtable.h>
753 #include <asm/system.h>
754 diff -NurpP --minimal linux-2.6.32.1/arch/powerpc/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/mm/fault.c
755 --- linux-2.6.32.1/arch/powerpc/mm/fault.c 2009-12-03 20:02:02.000000000 +0100
756 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/powerpc/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
757 @@ -358,7 +358,8 @@ out_of_memory:
758 down_read(&mm->mmap_sem);
761 - printk("VM: killing process %s\n", current->comm);
762 + printk("VM: killing process %s(%d:#%u)\n",
763 + current->comm, current->pid, current->xid);
765 do_group_exit(SIGKILL);
767 diff -NurpP --minimal linux-2.6.32.1/arch/s390/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/s390/Kconfig
768 --- linux-2.6.32.1/arch/s390/Kconfig 2009-12-03 20:02:03.000000000 +0100
769 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/s390/Kconfig 2009-12-03 20:04:56.000000000 +0100
770 @@ -616,6 +616,8 @@ source "fs/Kconfig"
772 source "arch/s390/Kconfig.debug"
774 +source "kernel/vserver/Kconfig"
776 source "security/Kconfig"
778 source "crypto/Kconfig"
779 diff -NurpP --minimal linux-2.6.32.1/arch/s390/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/arch/s390/kernel/ptrace.c
780 --- linux-2.6.32.1/arch/s390/kernel/ptrace.c 2009-12-03 20:02:03.000000000 +0100
781 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/s390/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +0100
783 #include <linux/regset.h>
784 #include <linux/tracehook.h>
785 #include <linux/seccomp.h>
786 +#include <linux/vs_base.h>
787 #include <trace/syscall.h>
788 #include <asm/compat.h>
789 #include <asm/segment.h>
790 diff -NurpP --minimal linux-2.6.32.1/arch/s390/kernel/syscalls.S linux-2.6.32.1-vs2.3.0.36.27/arch/s390/kernel/syscalls.S
791 --- linux-2.6.32.1/arch/s390/kernel/syscalls.S 2009-12-03 20:02:03.000000000 +0100
792 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/s390/kernel/syscalls.S 2009-12-03 20:04:56.000000000 +0100
793 @@ -271,7 +271,7 @@ SYSCALL(sys_clock_settime,sys_clock_sett
794 SYSCALL(sys_clock_gettime,sys_clock_gettime,sys32_clock_gettime_wrapper) /* 260 */
795 SYSCALL(sys_clock_getres,sys_clock_getres,sys32_clock_getres_wrapper)
796 SYSCALL(sys_clock_nanosleep,sys_clock_nanosleep,sys32_clock_nanosleep_wrapper)
797 -NI_SYSCALL /* reserved for vserver */
798 +SYSCALL(sys_vserver,sys_vserver,sys32_vserver)
799 SYSCALL(sys_s390_fadvise64_64,sys_ni_syscall,sys32_fadvise64_64_wrapper)
800 SYSCALL(sys_statfs64,sys_statfs64,compat_sys_statfs64_wrapper)
801 SYSCALL(sys_fstatfs64,sys_fstatfs64,compat_sys_fstatfs64_wrapper)
802 diff -NurpP --minimal linux-2.6.32.1/arch/s390/lib/uaccess_pt.c linux-2.6.32.1-vs2.3.0.36.27/arch/s390/lib/uaccess_pt.c
803 --- linux-2.6.32.1/arch/s390/lib/uaccess_pt.c 2009-09-10 15:25:43.000000000 +0200
804 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/s390/lib/uaccess_pt.c 2009-12-03 20:04:56.000000000 +0100
805 @@ -90,7 +90,8 @@ out_of_memory:
806 down_read(&mm->mmap_sem);
809 - printk("VM: killing process %s\n", current->comm);
810 + printk("VM: killing process %s(%d:#%u)\n",
811 + current->comm, task_pid_nr(current), current->xid);
815 diff -NurpP --minimal linux-2.6.32.1/arch/sh/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/sh/Kconfig
816 --- linux-2.6.32.1/arch/sh/Kconfig 2009-12-03 20:02:03.000000000 +0100
817 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sh/Kconfig 2009-12-03 20:04:56.000000000 +0100
818 @@ -853,6 +853,8 @@ source "fs/Kconfig"
820 source "arch/sh/Kconfig.debug"
822 +source "kernel/vserver/Kconfig"
824 source "security/Kconfig"
826 source "crypto/Kconfig"
827 diff -NurpP --minimal linux-2.6.32.1/arch/sh/kernel/irq.c linux-2.6.32.1-vs2.3.0.36.27/arch/sh/kernel/irq.c
828 --- linux-2.6.32.1/arch/sh/kernel/irq.c 2009-12-03 20:02:10.000000000 +0100
829 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sh/kernel/irq.c 2009-12-03 20:04:56.000000000 +0100
831 #include <linux/kernel_stat.h>
832 #include <linux/seq_file.h>
833 #include <linux/ftrace.h>
834 +// #include <linux/vs_context.h>
835 #include <asm/processor.h>
836 #include <asm/machvec.h>
837 #include <asm/uaccess.h>
838 diff -NurpP --minimal linux-2.6.32.1/arch/sh/kernel/vsyscall/vsyscall.c linux-2.6.32.1-vs2.3.0.36.27/arch/sh/kernel/vsyscall/vsyscall.c
839 --- linux-2.6.32.1/arch/sh/kernel/vsyscall/vsyscall.c 2009-03-24 14:18:42.000000000 +0100
840 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sh/kernel/vsyscall/vsyscall.c 2009-12-03 20:04:56.000000000 +0100
842 #include <linux/elf.h>
843 #include <linux/sched.h>
844 #include <linux/err.h>
845 +#include <linux/vs_memory.h>
848 * Should the kernel map a VDSO page into processes and pass its
849 diff -NurpP --minimal linux-2.6.32.1/arch/sh/mm/fault_32.c linux-2.6.32.1-vs2.3.0.36.27/arch/sh/mm/fault_32.c
850 --- linux-2.6.32.1/arch/sh/mm/fault_32.c 2009-12-03 20:02:14.000000000 +0100
851 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sh/mm/fault_32.c 2009-12-03 20:04:56.000000000 +0100
852 @@ -292,7 +292,8 @@ out_of_memory:
853 down_read(&mm->mmap_sem);
856 - printk("VM: killing process %s\n", tsk->comm);
857 + printk("VM: killing process %s(%d:#%u)\n",
858 + tsk->comm, task_pid_nr(tsk), tsk->xid);
860 do_group_exit(SIGKILL);
862 diff -NurpP --minimal linux-2.6.32.1/arch/sh/mm/tlbflush_64.c linux-2.6.32.1-vs2.3.0.36.27/arch/sh/mm/tlbflush_64.c
863 --- linux-2.6.32.1/arch/sh/mm/tlbflush_64.c 2009-12-03 20:02:14.000000000 +0100
864 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sh/mm/tlbflush_64.c 2009-12-03 20:04:56.000000000 +0100
865 @@ -306,7 +306,8 @@ out_of_memory:
866 down_read(&mm->mmap_sem);
869 - printk("VM: killing process %s\n", tsk->comm);
870 + printk("VM: killing process %s(%d:#%u)\n",
871 + tsk->comm, task_pid_nr(tsk), tsk->xid);
873 do_group_exit(SIGKILL);
875 diff -NurpP --minimal linux-2.6.32.1/arch/sparc/include/asm/tlb_64.h linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/include/asm/tlb_64.h
876 --- linux-2.6.32.1/arch/sparc/include/asm/tlb_64.h 2009-09-10 15:25:45.000000000 +0200
877 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/include/asm/tlb_64.h 2009-12-03 20:04:56.000000000 +0100
880 #include <linux/swap.h>
881 #include <linux/pagemap.h>
882 +#include <linux/vs_memory.h>
883 #include <asm/pgalloc.h>
884 #include <asm/tlbflush.h>
885 #include <asm/mmu_context.h>
886 diff -NurpP --minimal linux-2.6.32.1/arch/sparc/include/asm/unistd.h linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/include/asm/unistd.h
887 --- linux-2.6.32.1/arch/sparc/include/asm/unistd.h 2009-12-03 20:02:15.000000000 +0100
888 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/include/asm/unistd.h 2009-12-03 20:04:56.000000000 +0100
890 #define __NR_timer_getoverrun 264
891 #define __NR_timer_delete 265
892 #define __NR_timer_create 266
893 -/* #define __NR_vserver 267 Reserved for VSERVER */
894 +#define __NR_vserver 267
895 #define __NR_io_setup 268
896 #define __NR_io_destroy 269
897 #define __NR_io_submit 270
898 diff -NurpP --minimal linux-2.6.32.1/arch/sparc/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/Kconfig
899 --- linux-2.6.32.1/arch/sparc/Kconfig 2009-12-03 20:02:14.000000000 +0100
900 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/Kconfig 2009-12-03 20:04:56.000000000 +0100
901 @@ -550,6 +550,8 @@ source "fs/Kconfig"
903 source "arch/sparc/Kconfig.debug"
905 +source "kernel/vserver/Kconfig"
907 source "security/Kconfig"
909 source "crypto/Kconfig"
910 diff -NurpP --minimal linux-2.6.32.1/arch/sparc/kernel/systbls_32.S linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/kernel/systbls_32.S
911 --- linux-2.6.32.1/arch/sparc/kernel/systbls_32.S 2009-12-03 20:02:15.000000000 +0100
912 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/kernel/systbls_32.S 2009-12-03 20:04:56.000000000 +0100
913 @@ -70,7 +70,7 @@ sys_call_table:
914 /*250*/ .long sparc_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
915 /*255*/ .long sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
916 /*260*/ .long sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
917 -/*265*/ .long sys_timer_delete, sys_timer_create, sys_nis_syscall, sys_io_setup, sys_io_destroy
918 +/*265*/ .long sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
919 /*270*/ .long sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
920 /*275*/ .long sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
921 /*280*/ .long sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
922 diff -NurpP --minimal linux-2.6.32.1/arch/sparc/kernel/systbls_64.S linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/kernel/systbls_64.S
923 --- linux-2.6.32.1/arch/sparc/kernel/systbls_64.S 2009-12-03 20:02:15.000000000 +0100
924 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/sparc/kernel/systbls_64.S 2009-12-03 20:04:56.000000000 +0100
925 @@ -71,7 +71,7 @@ sys_call_table32:
926 /*250*/ .word sys32_mremap, sys32_sysctl, sys32_getsid, sys_fdatasync, sys32_nfsservctl
927 .word sys32_sync_file_range, compat_sys_clock_settime, compat_sys_clock_gettime, compat_sys_clock_getres, sys32_clock_nanosleep
928 /*260*/ .word compat_sys_sched_getaffinity, compat_sys_sched_setaffinity, sys32_timer_settime, compat_sys_timer_gettime, sys_timer_getoverrun
929 - .word sys_timer_delete, compat_sys_timer_create, sys_ni_syscall, compat_sys_io_setup, sys_io_destroy
930 + .word sys_timer_delete, compat_sys_timer_create, sys32_vserver, compat_sys_io_setup, sys_io_destroy
931 /*270*/ .word sys32_io_submit, sys_io_cancel, compat_sys_io_getevents, sys32_mq_open, sys_mq_unlink
932 .word compat_sys_mq_timedsend, compat_sys_mq_timedreceive, compat_sys_mq_notify, compat_sys_mq_getsetattr, compat_sys_waitid
933 /*280*/ .word sys32_tee, sys_add_key, sys_request_key, sys_keyctl, compat_sys_openat
934 @@ -146,7 +146,7 @@ sys_call_table:
935 /*250*/ .word sys_64_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
936 .word sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
937 /*260*/ .word sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
938 - .word sys_timer_delete, sys_timer_create, sys_ni_syscall, sys_io_setup, sys_io_destroy
939 + .word sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
940 /*270*/ .word sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
941 .word sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
942 /*280*/ .word sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
943 diff -NurpP --minimal linux-2.6.32.1/arch/x86/ia32/ia32entry.S linux-2.6.32.1-vs2.3.0.36.27/arch/x86/ia32/ia32entry.S
944 --- linux-2.6.32.1/arch/x86/ia32/ia32entry.S 2009-12-03 20:02:15.000000000 +0100
945 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/x86/ia32/ia32entry.S 2009-12-03 20:04:56.000000000 +0100
946 @@ -777,7 +777,7 @@ ia32_sys_call_table:
947 .quad sys_tgkill /* 270 */
948 .quad compat_sys_utimes
949 .quad sys32_fadvise64_64
950 - .quad quiet_ni_syscall /* sys_vserver */
951 + .quad sys32_vserver
953 .quad compat_sys_get_mempolicy /* 275 */
954 .quad sys_set_mempolicy
955 diff -NurpP --minimal linux-2.6.32.1/arch/x86/include/asm/unistd_64.h linux-2.6.32.1-vs2.3.0.36.27/arch/x86/include/asm/unistd_64.h
956 --- linux-2.6.32.1/arch/x86/include/asm/unistd_64.h 2009-12-03 20:02:16.000000000 +0100
957 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/x86/include/asm/unistd_64.h 2009-12-03 20:04:56.000000000 +0100
958 @@ -535,7 +535,7 @@ __SYSCALL(__NR_tgkill, sys_tgkill)
959 #define __NR_utimes 235
960 __SYSCALL(__NR_utimes, sys_utimes)
961 #define __NR_vserver 236
962 -__SYSCALL(__NR_vserver, sys_ni_syscall)
963 +__SYSCALL(__NR_vserver, sys_vserver)
964 #define __NR_mbind 237
965 __SYSCALL(__NR_mbind, sys_mbind)
966 #define __NR_set_mempolicy 238
967 diff -NurpP --minimal linux-2.6.32.1/arch/x86/Kconfig linux-2.6.32.1-vs2.3.0.36.27/arch/x86/Kconfig
968 --- linux-2.6.32.1/arch/x86/Kconfig 2009-12-03 20:02:15.000000000 +0100
969 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/x86/Kconfig 2009-12-03 20:04:56.000000000 +0100
970 @@ -2085,6 +2085,8 @@ source "fs/Kconfig"
972 source "arch/x86/Kconfig.debug"
974 +source "kernel/vserver/Kconfig"
976 source "security/Kconfig"
978 source "crypto/Kconfig"
979 diff -NurpP --minimal linux-2.6.32.1/arch/x86/kernel/syscall_table_32.S linux-2.6.32.1-vs2.3.0.36.27/arch/x86/kernel/syscall_table_32.S
980 --- linux-2.6.32.1/arch/x86/kernel/syscall_table_32.S 2009-12-03 20:02:16.000000000 +0100
981 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/x86/kernel/syscall_table_32.S 2009-12-03 20:04:56.000000000 +0100
982 @@ -272,7 +272,7 @@ ENTRY(sys_call_table)
983 .long sys_tgkill /* 270 */
985 .long sys_fadvise64_64
986 - .long sys_ni_syscall /* sys_vserver */
989 .long sys_get_mempolicy
990 .long sys_set_mempolicy
991 diff -NurpP --minimal linux-2.6.32.1/arch/xtensa/mm/fault.c linux-2.6.32.1-vs2.3.0.36.27/arch/xtensa/mm/fault.c
992 --- linux-2.6.32.1/arch/xtensa/mm/fault.c 2009-09-10 15:25:48.000000000 +0200
993 +++ linux-2.6.32.1-vs2.3.0.36.27/arch/xtensa/mm/fault.c 2009-12-03 20:04:56.000000000 +0100
994 @@ -151,7 +151,8 @@ out_of_memory:
995 down_read(&mm->mmap_sem);
998 - printk("VM: killing process %s\n", current->comm);
999 + printk("VM: killing process %s(%d:#%u)\n",
1000 + current->comm, task_pid_nr(current), current->xid);
1001 if (user_mode(regs))
1002 do_group_exit(SIGKILL);
1003 bad_page_fault(regs, address, SIGKILL);
1004 diff -NurpP --minimal linux-2.6.32.1/Documentation/scheduler/sched-cfs-hard-limits.txt linux-2.6.32.1-vs2.3.0.36.27/Documentation/scheduler/sched-cfs-hard-limits.txt
1005 --- linux-2.6.32.1/Documentation/scheduler/sched-cfs-hard-limits.txt 1970-01-01 01:00:00.000000000 +0100
1006 +++ linux-2.6.32.1-vs2.3.0.36.27/Documentation/scheduler/sched-cfs-hard-limits.txt 2009-12-03 20:04:56.000000000 +0100
1008 +CPU HARD LIMITS FOR CFS GROUPS
1009 +==============================
1018 +CFS is a proportional share scheduler which tries to divide the CPU time
1019 +proportionately between tasks or groups of tasks (task group/cgroup) depending
1020 +on the priority/weight of the task or shares assigned to groups of tasks.
1021 +In CFS, a task/task group can get more than its share of CPU if there are
1022 +enough idle CPU cycles available in the system, due to the work conserving
1023 +nature of the scheduler. However in certain scenarios (like pay-per-use),
1024 +it is desirable not to provide extra time to a group even in the presence
1025 +of idle CPU cycles. This is where hard limiting can be of use.
1027 +Hard limits for task groups can be set by specifying how much CPU runtime a
1028 +group can consume within a given period. If the group consumes more CPU time
1029 +than the runtime in a given period, it gets throttled. None of the tasks of
1030 +the throttled group gets to run until the runtime of the group gets refreshed
1031 +at the beginning of the next period.
1036 +Hard limit feature adds 2 cgroup files for CFS group scheduler:
1038 +cfs_runtime_us: Hard limit for the group in microseconds.
1040 +cfs_period_us: Time period in microseconds within which hard limits is
1043 +A group gets created with default values for runtime (infinite runtime which
1044 +means hard limits disabled) and period (0.5s). Each group can set its own
1045 +values for runtime and period independent of other groups in the system.
1050 +# mount -t cgroup -ocpu none /cgroups/
1054 +# echo 250000 > cfs_runtime_us /* set a 250ms runtime or limit */
1055 +# echo 500000 > cfs_period_us /* set a 500ms period */
1056 diff -NurpP --minimal linux-2.6.32.1/Documentation/vserver/debug.txt linux-2.6.32.1-vs2.3.0.36.27/Documentation/vserver/debug.txt
1057 --- linux-2.6.32.1/Documentation/vserver/debug.txt 1970-01-01 01:00:00.000000000 +0100
1058 +++ linux-2.6.32.1-vs2.3.0.36.27/Documentation/vserver/debug.txt 2009-12-03 20:04:56.000000000 +0100
1063 + 2 4 "vx_map_tgid: %p/%llx: %d -> %d"
1064 + "vx_rmap_tgid: %p/%llx: %d -> %d"
1068 + 0 1 "ALLOC (%p,#%d)%c inode (%d)"
1069 + "FREE (%p,#%d)%c inode"
1070 + 1 2 "ALLOC (%p,#%d)%c %lld bytes (%d)"
1071 + "FREE (%p,#%d)%c %lld bytes"
1072 + 2 4 "ADJUST: %lld,%lld on %ld,%ld [mult=%d]"
1073 + 3 8 "ext3_has_free_blocks(%p): %lu<%lu+1, %c, %u!=%u r=%d"
1074 + "ext3_has_free_blocks(%p): free=%lu, root=%lu"
1075 + "rcu_free_dl_info(%p)"
1076 + 4 10 "alloc_dl_info(%p,%d) = %p"
1077 + "dealloc_dl_info(%p)"
1078 + "get_dl_info(%p[#%d.%d])"
1079 + "put_dl_info(%p[#%d.%d])"
1080 + 5 20 "alloc_dl_info(%p,%d)*"
1081 + 6 40 "__hash_dl_info: %p[#%d]"
1082 + "__unhash_dl_info: %p[#%d]"
1083 + 7 80 "locate_dl_info(%p,#%d) = %p"
1087 + 0 1 "destroy_dqhash: %p [#0x%08x] c=%d"
1088 + "new_dqhash: %p [#0x%08x]"
1089 + "vroot[%d]_clr_dev: dev=%p[%lu,%d:%d]"
1090 + "vroot[%d]_get_real_bdev: dev=%p[%lu,%d:%d]"
1091 + "vroot[%d]_set_dev: dev=%p[%lu,%d:%d]"
1092 + "vroot_get_real_bdev not set"
1093 + 1 2 "cow_break_link(»%s«)"
1094 + "temp copy »%s«"
1095 + 2 4 "dentry_open(new): %p"
1096 + "dentry_open(old): %p"
1097 + "lookup_create(new): %p"
1099 + "path_lookup(old): %d"
1100 + "vfs_create(new): %d"
1102 + "vfs_sendfile: %d"
1103 + 3 8 "fput(new_file=%p[#%d])"
1104 + "fput(old_file=%p[#%d])"
1105 + 4 10 "vx_info_kill(%p[#%d],%d,%d) = %d"
1106 + "vx_info_kill(%p[#%d],%d,%d)*"
1107 + 5 20 "vs_reboot(%p[#%d],%d)"
1108 + 6 40 "dropping task %p[#%u,%u] for %p[#%u,%u]"
1112 + 2 4 "nx_addr_conflict(%p,%p) %d.%d,%d.%d"
1113 + 3 8 "inet_bind(%p) %d.%d.%d.%d, %d.%d.%d.%d, %d.%d.%d.%d"
1114 + "inet_bind(%p)* %p,%p;%lx %d.%d.%d.%d"
1115 + 4 10 "ip_route_connect(%p) %p,%p;%lx"
1116 + 5 20 "__addr_in_socket(%p,%d.%d.%d.%d) %p:%d.%d.%d.%d %p;%lx"
1117 + 6 40 "sk,egf: %p [#%d] (from %d)"
1118 + "sk,egn: %p [#%d] (from %d)"
1119 + "sk,req: %p [#%d] (from %d)"
1120 + "sk: %p [#%d] (from %d)"
1121 + "tw: %p [#%d] (from %d)"
1122 + 7 80 "__sock_recvmsg: %p[%p,%p,%p;%d]:%d/%d"
1123 + "__sock_sendmsg: %p[%p,%p,%p;%d]:%d/%d"
1127 + 0 1 "__lookup_nx_info(#%u): %p[#%u]"
1128 + "alloc_nx_info(%d) = %p"
1129 + "create_nx_info(%d) (dynamic rejected)"
1130 + "create_nx_info(%d) = %p (already there)"
1131 + "create_nx_info(%d) = %p (new)"
1132 + "dealloc_nx_info(%p)"
1133 + 1 2 "alloc_nx_info(%d)*"
1134 + "create_nx_info(%d)*"
1135 + 2 4 "get_nx_info(%p[#%d.%d])"
1136 + "put_nx_info(%p[#%d.%d])"
1137 + 3 8 "claim_nx_info(%p[#%d.%d.%d]) %p"
1138 + "clr_nx_info(%p[#%d.%d])"
1139 + "init_nx_info(%p[#%d.%d])"
1140 + "release_nx_info(%p[#%d.%d.%d]) %p"
1141 + "set_nx_info(%p[#%d.%d])"
1142 + 4 10 "__hash_nx_info: %p[#%d]"
1143 + "__nx_dynamic_id: [#%d]"
1144 + "__unhash_nx_info: %p[#%d.%d.%d]"
1145 + 5 20 "moved task %p into nxi:%p[#%d]"
1146 + "nx_migrate_task(%p,%p[#%d.%d.%d])"
1147 + "task_get_nx_info(%p)"
1148 + 6 40 "nx_clear_persistent(%p[#%d])"
1152 + 0 1 "quota_sync_dqh(%p,%d) discard inode %p"
1153 + 1 2 "quota_sync_dqh(%p,%d)"
1154 + "sync_dquots(%p,%d)"
1155 + "sync_dquots_dqh(%p,%d)"
1156 + 3 8 "do_quotactl(%p,%d,cmd=%d,id=%d,%p)"
1160 + 0 1 "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]"
1161 + 1 2 "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]"
1162 + 4 10 "%s: (%s %s) returned %s with %d"
1166 + 7 80 "dx_parse_tag(»%s«): %d:#%d"
1167 + "dx_propagate_tag(%p[#%lu.%d]): %d,%d"
1171 + 0 1 "__lookup_vx_info(#%u): %p[#%u]"
1172 + "alloc_vx_info(%d) = %p"
1173 + "alloc_vx_info(%d)*"
1174 + "create_vx_info(%d) (dynamic rejected)"
1175 + "create_vx_info(%d) = %p (already there)"
1176 + "create_vx_info(%d) = %p (new)"
1177 + "dealloc_vx_info(%p)"
1178 + "loc_vx_info(%d) = %p (found)"
1179 + "loc_vx_info(%d) = %p (new)"
1180 + "loc_vx_info(%d) = %p (not available)"
1181 + 1 2 "create_vx_info(%d)*"
1182 + "loc_vx_info(%d)*"
1183 + 2 4 "get_vx_info(%p[#%d.%d])"
1184 + "put_vx_info(%p[#%d.%d])"
1185 + 3 8 "claim_vx_info(%p[#%d.%d.%d]) %p"
1186 + "clr_vx_info(%p[#%d.%d])"
1187 + "init_vx_info(%p[#%d.%d])"
1188 + "release_vx_info(%p[#%d.%d.%d]) %p"
1189 + "set_vx_info(%p[#%d.%d])"
1190 + 4 10 "__hash_vx_info: %p[#%d]"
1191 + "__unhash_vx_info: %p[#%d.%d.%d]"
1192 + "__vx_dynamic_id: [#%d]"
1193 + 5 20 "enter_vx_info(%p[#%d],%p) %p[#%d,%p]"
1194 + "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]"
1195 + "moved task %p into vxi:%p[#%d]"
1196 + "task_get_vx_info(%p)"
1197 + "vx_migrate_task(%p,%p[#%d.%d])"
1198 + 6 40 "vx_clear_persistent(%p[#%d])"
1199 + "vx_exit_init(%p[#%d],%p[#%d,%d,%d])"
1200 + "vx_set_init(%p[#%d],%p[#%d,%d,%d])"
1201 + "vx_set_persistent(%p[#%d])"
1202 + "vx_set_reaper(%p[#%d],%p[#%d,%d])"
1203 + 7 80 "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]"
1208 + n 2^n "vx_acc_cres[%5d,%s,%2d]: %5d%s"
1209 + "vx_cres_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1211 + m 2^m "vx_acc_page[%5d,%s,%2d]: %5d%s"
1212 + "vx_acc_pages[%5d,%s,%2d]: %5d += %5d"
1213 + "vx_pages_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1214 diff -NurpP --minimal linux-2.6.32.1/drivers/block/Kconfig linux-2.6.32.1-vs2.3.0.36.27/drivers/block/Kconfig
1215 --- linux-2.6.32.1/drivers/block/Kconfig 2009-09-10 15:25:49.000000000 +0200
1216 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/block/Kconfig 2009-12-03 20:04:56.000000000 +0100
1217 @@ -271,6 +271,13 @@ config BLK_DEV_CRYPTOLOOP
1218 instead, which can be configured to be on-disk compatible with the
1221 +config BLK_DEV_VROOT
1222 + tristate "Virtual Root device support"
1223 + depends on QUOTACTL
1225 + Saying Y here will allow you to use quota/fs ioctls on a shared
1226 + partition within a virtual server without compromising security.
1229 tristate "Network block device support"
1231 diff -NurpP --minimal linux-2.6.32.1/drivers/block/loop.c linux-2.6.32.1-vs2.3.0.36.27/drivers/block/loop.c
1232 --- linux-2.6.32.1/drivers/block/loop.c 2009-12-03 20:02:19.000000000 +0100
1233 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/block/loop.c 2009-12-03 20:04:56.000000000 +0100
1235 #include <linux/gfp.h>
1236 #include <linux/kthread.h>
1237 #include <linux/splice.h>
1238 +#include <linux/vs_context.h>
1240 #include <asm/uaccess.h>
1242 @@ -812,6 +813,7 @@ static int loop_set_fd(struct loop_devic
1243 lo->lo_blocksize = lo_blocksize;
1244 lo->lo_device = bdev;
1245 lo->lo_flags = lo_flags;
1246 + lo->lo_xid = vx_current_xid();
1247 lo->lo_backing_file = file;
1248 lo->transfer = transfer_none;
1250 @@ -937,6 +939,7 @@ static int loop_clr_fd(struct loop_devic
1251 lo->lo_encrypt_key_size = 0;
1253 lo->lo_thread = NULL;
1255 memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
1256 memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
1257 memset(lo->lo_file_name, 0, LO_NAME_SIZE);
1258 @@ -971,7 +974,7 @@ loop_set_status(struct loop_device *lo,
1260 if (lo->lo_encrypt_key_size &&
1261 lo->lo_key_owner != uid &&
1262 - !capable(CAP_SYS_ADMIN))
1263 + !vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP))
1265 if (lo->lo_state != Lo_bound)
1267 @@ -1055,7 +1058,8 @@ loop_get_status(struct loop_device *lo,
1268 memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
1269 info->lo_encrypt_type =
1270 lo->lo_encryption ? lo->lo_encryption->number : 0;
1271 - if (lo->lo_encrypt_key_size && capable(CAP_SYS_ADMIN)) {
1272 + if (lo->lo_encrypt_key_size &&
1273 + vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP)) {
1274 info->lo_encrypt_key_size = lo->lo_encrypt_key_size;
1275 memcpy(info->lo_encrypt_key, lo->lo_encrypt_key,
1276 lo->lo_encrypt_key_size);
1277 @@ -1399,6 +1403,9 @@ static int lo_open(struct block_device *
1279 struct loop_device *lo = bdev->bd_disk->private_data;
1281 + if (!vx_check(lo->lo_xid, VS_IDENT|VS_HOSTID|VS_ADMIN_P))
1284 mutex_lock(&lo->lo_ctl_mutex);
1286 mutex_unlock(&lo->lo_ctl_mutex);
1287 diff -NurpP --minimal linux-2.6.32.1/drivers/block/Makefile linux-2.6.32.1-vs2.3.0.36.27/drivers/block/Makefile
1288 --- linux-2.6.32.1/drivers/block/Makefile 2009-09-10 15:25:49.000000000 +0200
1289 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/block/Makefile 2009-12-03 20:04:56.000000000 +0100
1290 @@ -34,6 +34,7 @@ obj-$(CONFIG_VIODASD) += viodasd.o
1291 obj-$(CONFIG_BLK_DEV_SX8) += sx8.o
1292 obj-$(CONFIG_BLK_DEV_UB) += ub.o
1293 obj-$(CONFIG_BLK_DEV_HD) += hd.o
1294 +obj-$(CONFIG_BLK_DEV_VROOT) += vroot.o
1296 obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o
1298 diff -NurpP --minimal linux-2.6.32.1/drivers/block/vroot.c linux-2.6.32.1-vs2.3.0.36.27/drivers/block/vroot.c
1299 --- linux-2.6.32.1/drivers/block/vroot.c 1970-01-01 01:00:00.000000000 +0100
1300 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/block/vroot.c 2009-12-03 20:04:56.000000000 +0100
1303 + * linux/drivers/block/vroot.c
1305 + * written by Herbert Pötzl, 9/11/2002
1306 + * ported to 2.6.10 by Herbert Pötzl, 30/12/2004
1308 + * based on the loop.c code by Theodore Ts'o.
1310 + * Copyright (C) 2002-2007 by Herbert Pötzl.
1311 + * Redistribution of this file is permitted under the
1312 + * GNU General Public License.
1316 +#include <linux/module.h>
1317 +#include <linux/moduleparam.h>
1318 +#include <linux/file.h>
1319 +#include <linux/major.h>
1320 +#include <linux/blkdev.h>
1322 +#include <linux/vroot.h>
1323 +#include <linux/vs_context.h>
1326 +static int max_vroot = 8;
1328 +static struct vroot_device *vroot_dev;
1329 +static struct gendisk **disks;
1332 +static int vroot_set_dev(
1333 + struct vroot_device *vr,
1334 + struct block_device *bdev,
1337 + struct block_device *real_bdev;
1338 + struct file *file;
1339 + struct inode *inode;
1343 + if (vr->vr_state != Vr_unbound)
1352 + inode = file->f_dentry->d_inode;
1355 + if (S_ISBLK(inode->i_mode)) {
1356 + real_bdev = inode->i_bdev;
1357 + vr->vr_device = real_bdev;
1358 + __iget(real_bdev->bd_inode);
1362 + vxdprintk(VXD_CBIT(misc, 0),
1363 + "vroot[%d]_set_dev: dev=" VXF_DEV,
1364 + vr->vr_number, VXD_DEV(real_bdev));
1366 + vr->vr_state = Vr_bound;
1375 +static int vroot_clr_dev(
1376 + struct vroot_device *vr,
1377 + struct block_device *bdev)
1379 + struct block_device *real_bdev;
1381 + if (vr->vr_state != Vr_bound)
1383 + if (vr->vr_refcnt > 1) /* we needed one fd for the ioctl */
1386 + real_bdev = vr->vr_device;
1388 + vxdprintk(VXD_CBIT(misc, 0),
1389 + "vroot[%d]_clr_dev: dev=" VXF_DEV,
1390 + vr->vr_number, VXD_DEV(real_bdev));
1393 + vr->vr_state = Vr_unbound;
1394 + vr->vr_device = NULL;
1399 +static int vr_ioctl(struct block_device *bdev, fmode_t mode,
1400 + unsigned int cmd, unsigned long arg)
1402 + struct vroot_device *vr = bdev->bd_disk->private_data;
1405 + down(&vr->vr_ctl_mutex);
1407 + case VROOT_SET_DEV:
1408 + err = vroot_set_dev(vr, bdev, arg);
1410 + case VROOT_CLR_DEV:
1411 + err = vroot_clr_dev(vr, bdev);
1417 + up(&vr->vr_ctl_mutex);
1421 +static int vr_open(struct block_device *bdev, fmode_t mode)
1423 + struct vroot_device *vr = bdev->bd_disk->private_data;
1425 + down(&vr->vr_ctl_mutex);
1427 + up(&vr->vr_ctl_mutex);
1431 +static int vr_release(struct gendisk *disk, fmode_t mode)
1433 + struct vroot_device *vr = disk->private_data;
1435 + down(&vr->vr_ctl_mutex);
1437 + up(&vr->vr_ctl_mutex);
1441 +static struct block_device_operations vr_fops = {
1442 + .owner = THIS_MODULE,
1444 + .release = vr_release,
1445 + .ioctl = vr_ioctl,
1448 +struct block_device *__vroot_get_real_bdev(struct block_device *bdev)
1450 + struct inode *inode = bdev->bd_inode;
1451 + struct vroot_device *vr;
1452 + struct block_device *real_bdev;
1453 + int minor = iminor(inode);
1455 + vr = &vroot_dev[minor];
1456 + real_bdev = vr->vr_device;
1458 + vxdprintk(VXD_CBIT(misc, 0),
1459 + "vroot[%d]_get_real_bdev: dev=" VXF_DEV,
1460 + vr->vr_number, VXD_DEV(real_bdev));
1462 + if (vr->vr_state != Vr_bound)
1463 + return ERR_PTR(-ENXIO);
1465 + __iget(real_bdev->bd_inode);
1470 + * And now the modules code and kernel interface.
1473 +module_param(max_vroot, int, 0);
1475 +MODULE_PARM_DESC(max_vroot, "Maximum number of vroot devices (1-256)");
1476 +MODULE_LICENSE("GPL");
1477 +MODULE_ALIAS_BLOCKDEV_MAJOR(VROOT_MAJOR);
1479 +MODULE_AUTHOR ("Herbert Pötzl");
1480 +MODULE_DESCRIPTION ("Virtual Root Device Mapper");
1483 +int __init vroot_init(void)
1487 + if (max_vroot < 1 || max_vroot > 256) {
1488 + max_vroot = MAX_VROOT_DEFAULT;
1489 + printk(KERN_WARNING "vroot: invalid max_vroot "
1490 + "(must be between 1 and 256), "
1491 + "using default (%d)\n", max_vroot);
1494 + if (register_blkdev(VROOT_MAJOR, "vroot"))
1498 + vroot_dev = kmalloc(max_vroot * sizeof(struct vroot_device), GFP_KERNEL);
1501 + memset(vroot_dev, 0, max_vroot * sizeof(struct vroot_device));
1503 + disks = kmalloc(max_vroot * sizeof(struct gendisk *), GFP_KERNEL);
1507 + for (i = 0; i < max_vroot; i++) {
1508 + disks[i] = alloc_disk(1);
1511 + disks[i]->queue = blk_alloc_queue(GFP_KERNEL);
1512 + if (!disks[i]->queue)
1516 + for (i = 0; i < max_vroot; i++) {
1517 + struct vroot_device *vr = &vroot_dev[i];
1518 + struct gendisk *disk = disks[i];
1520 + memset(vr, 0, sizeof(*vr));
1521 + init_MUTEX(&vr->vr_ctl_mutex);
1522 + vr->vr_number = i;
1523 + disk->major = VROOT_MAJOR;
1524 + disk->first_minor = i;
1525 + disk->fops = &vr_fops;
1526 + sprintf(disk->disk_name, "vroot%d", i);
1527 + disk->private_data = vr;
1530 + err = register_vroot_grb(&__vroot_get_real_bdev);
1534 + for (i = 0; i < max_vroot; i++)
1535 + add_disk(disks[i]);
1536 + printk(KERN_INFO "vroot: loaded (max %d devices)\n", max_vroot);
1541 + put_disk(disks[i]);
1546 + unregister_blkdev(VROOT_MAJOR, "vroot");
1547 + printk(KERN_ERR "vroot: ran out of memory\n");
1551 +void vroot_exit(void)
1555 + if (unregister_vroot_grb(&__vroot_get_real_bdev))
1556 + printk(KERN_WARNING "vroot: cannot unregister grb\n");
1558 + for (i = 0; i < max_vroot; i++) {
1559 + del_gendisk(disks[i]);
1560 + put_disk(disks[i]);
1562 + unregister_blkdev(VROOT_MAJOR, "vroot");
1568 +module_init(vroot_init);
1569 +module_exit(vroot_exit);
1573 +static int __init max_vroot_setup(char *str)
1575 + max_vroot = simple_strtol(str, NULL, 0);
1579 +__setup("max_vroot=", max_vroot_setup);
1583 diff -NurpP --minimal linux-2.6.32.1/drivers/char/sysrq.c linux-2.6.32.1-vs2.3.0.36.27/drivers/char/sysrq.c
1584 --- linux-2.6.32.1/drivers/char/sysrq.c 2009-12-03 20:02:20.000000000 +0100
1585 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/char/sysrq.c 2009-12-03 20:04:56.000000000 +0100
1587 #include <linux/workqueue.h>
1588 #include <linux/hrtimer.h>
1589 #include <linux/oom.h>
1590 +#include <linux/vserver/debug.h>
1592 #include <asm/ptrace.h>
1593 #include <asm/irq_regs.h>
1594 @@ -391,6 +392,21 @@ static struct sysrq_key_op sysrq_unrt_op
1595 .enable_mask = SYSRQ_ENABLE_RTNICE,
1599 +#ifdef CONFIG_VSERVER_DEBUG
1600 +static void sysrq_handle_vxinfo(int key, struct tty_struct *tty)
1602 + dump_vx_info_inactive((key == 'x')?0:1);
1605 +static struct sysrq_key_op sysrq_showvxinfo_op = {
1606 + .handler = sysrq_handle_vxinfo,
1607 + .help_msg = "conteXt",
1608 + .action_msg = "Show Context Info",
1609 + .enable_mask = SYSRQ_ENABLE_DUMP,
1613 /* Key Operations table and lock */
1614 static DEFINE_SPINLOCK(sysrq_key_table_lock);
1616 @@ -445,7 +461,11 @@ static struct sysrq_key_op *sysrq_key_ta
1618 &sysrq_showstate_blocked_op, /* w */
1619 /* x: May be registered on ppc/powerpc for xmon */
1620 +#ifdef CONFIG_VSERVER_DEBUG
1621 + &sysrq_showvxinfo_op, /* x */
1625 /* y: May be registered on sparc64 for global register dump */
1627 &sysrq_ftrace_dump_op, /* z */
1628 @@ -460,6 +480,8 @@ static int sysrq_key_table_key2index(int
1630 else if ((key >= 'a') && (key <= 'z'))
1631 retval = key + 10 - 'a';
1632 + else if ((key >= 'A') && (key <= 'Z'))
1633 + retval = key + 10 - 'A';
1637 diff -NurpP --minimal linux-2.6.32.1/drivers/char/tty_io.c linux-2.6.32.1-vs2.3.0.36.27/drivers/char/tty_io.c
1638 --- linux-2.6.32.1/drivers/char/tty_io.c 2009-12-03 20:02:20.000000000 +0100
1639 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/char/tty_io.c 2009-12-03 20:04:56.000000000 +0100
1642 #include <linux/kmod.h>
1643 #include <linux/nsproxy.h>
1644 +#include <linux/vs_pid.h>
1646 #undef TTY_DEBUG_HANGUP
1648 @@ -1966,7 +1967,8 @@ static int tiocsti(struct tty_struct *tt
1650 struct tty_ldisc *ld;
1652 - if ((current->signal->tty != tty) && !capable(CAP_SYS_ADMIN))
1653 + if (((current->signal->tty != tty) && !capable(CAP_SYS_ADMIN)) ||
1654 + !vx_ccaps(VXC_TIOCSTI))
1656 if (get_user(ch, p))
1658 @@ -2254,6 +2256,7 @@ static int tiocspgrp(struct tty_struct *
1660 if (get_user(pgrp_nr, p))
1662 + pgrp_nr = vx_rmap_pid(pgrp_nr);
1666 diff -NurpP --minimal linux-2.6.32.1/drivers/infiniband/hw/ipath/ipath_user_pages.c linux-2.6.32.1-vs2.3.0.36.27/drivers/infiniband/hw/ipath/ipath_user_pages.c
1667 --- linux-2.6.32.1/drivers/infiniband/hw/ipath/ipath_user_pages.c 2009-12-03 20:02:23.000000000 +0100
1668 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/infiniband/hw/ipath/ipath_user_pages.c 2009-12-04 23:31:17.000000000 +0100
1670 #include <linux/mm.h>
1671 #include <linux/device.h>
1672 #include <linux/sched.h>
1673 +#include <linux/vs_memory.h>
1675 #include "ipath_kernel.h"
1677 @@ -62,7 +63,8 @@ static int __get_user_pages(unsigned lon
1678 lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >>
1681 - if (num_pages > lock_limit) {
1682 + if (num_pages > lock_limit ||
1683 + !vx_vmlocked_avail(current->mm, num_pages)) {
1687 @@ -79,7 +81,7 @@ static int __get_user_pages(unsigned lon
1691 - current->mm->locked_vm += num_pages;
1692 + vx_vmlocked_add(current->mm, num_pages);
1696 @@ -178,7 +180,7 @@ void ipath_release_user_pages(struct pag
1698 __ipath_release_user_pages(p, num_pages, 1);
1700 - current->mm->locked_vm -= num_pages;
1701 + vx_vmlocked_sub(current->mm, num_pages);
1703 up_write(¤t->mm->mmap_sem);
1705 @@ -195,7 +197,7 @@ static void user_pages_account(struct wo
1706 container_of(_work, struct ipath_user_pages_work, work);
1708 down_write(&work->mm->mmap_sem);
1709 - work->mm->locked_vm -= work->num_pages;
1710 + vx_vmlocked_sub(work->mm, work->num_pages);
1711 up_write(&work->mm->mmap_sem);
1714 diff -NurpP --minimal linux-2.6.32.1/drivers/md/dm.c linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm.c
1715 --- linux-2.6.32.1/drivers/md/dm.c 2009-12-03 20:02:25.000000000 +0100
1716 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm.c 2009-12-03 20:04:56.000000000 +0100
1718 #include <linux/slab.h>
1719 #include <linux/idr.h>
1720 #include <linux/hdreg.h>
1721 +#include <linux/vs_base.h>
1723 #include <trace/events/block.h>
1725 @@ -119,6 +120,7 @@ struct mapped_device {
1728 atomic_t open_count;
1731 unsigned long flags;
1733 @@ -323,6 +325,7 @@ static void __exit dm_exit(void)
1734 static int dm_blk_open(struct block_device *bdev, fmode_t mode)
1736 struct mapped_device *md;
1739 spin_lock(&_minor_lock);
1741 @@ -331,18 +334,19 @@ static int dm_blk_open(struct block_devi
1744 if (test_bit(DMF_FREEING, &md->flags) ||
1745 - test_bit(DMF_DELETING, &md->flags)) {
1747 + test_bit(DMF_DELETING, &md->flags))
1751 + if (!vx_check(md->xid, VS_IDENT|VS_HOSTID))
1756 atomic_inc(&md->open_count);
1760 spin_unlock(&_minor_lock);
1762 - return md ? 0 : -ENXIO;
1766 static int dm_blk_close(struct gendisk *disk, fmode_t mode)
1767 @@ -553,6 +557,14 @@ int dm_set_geometry(struct mapped_device
1772 + * Get the xid associated with a dm device
1774 +xid_t dm_get_xid(struct mapped_device *md)
1779 /*-----------------------------------------------------------------
1781 * A more elegant soln is in the works that uses the queue
1782 @@ -1762,6 +1774,7 @@ static struct mapped_device *alloc_dev(i
1783 INIT_LIST_HEAD(&md->uevent_list);
1784 spin_lock_init(&md->uevent_lock);
1786 + md->xid = vx_current_xid();
1787 md->queue = blk_init_queue(dm_request_fn, NULL);
1790 diff -NurpP --minimal linux-2.6.32.1/drivers/md/dm.h linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm.h
1791 --- linux-2.6.32.1/drivers/md/dm.h 2009-09-10 15:25:55.000000000 +0200
1792 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm.h 2009-12-03 20:04:56.000000000 +0100
1793 @@ -41,6 +41,8 @@ struct dm_dev_internal {
1795 struct dm_md_mempools;
1797 +xid_t dm_get_xid(struct mapped_device *md);
1799 /*-----------------------------------------------------------------
1800 * Internal table functions.
1801 *---------------------------------------------------------------*/
1802 diff -NurpP --minimal linux-2.6.32.1/drivers/md/dm-ioctl.c linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm-ioctl.c
1803 --- linux-2.6.32.1/drivers/md/dm-ioctl.c 2009-12-03 20:02:25.000000000 +0100
1804 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/md/dm-ioctl.c 2009-12-03 20:04:56.000000000 +0100
1806 #include <linux/dm-ioctl.h>
1807 #include <linux/hdreg.h>
1808 #include <linux/compat.h>
1809 +#include <linux/vs_context.h>
1811 #include <asm/uaccess.h>
1813 @@ -101,7 +102,8 @@ static struct hash_cell *__get_name_cell
1814 unsigned int h = hash_str(str);
1816 list_for_each_entry (hc, _name_buckets + h, name_list)
1817 - if (!strcmp(hc->name, str)) {
1818 + if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1819 + !strcmp(hc->name, str)) {
1823 @@ -115,7 +117,8 @@ static struct hash_cell *__get_uuid_cell
1824 unsigned int h = hash_str(str);
1826 list_for_each_entry (hc, _uuid_buckets + h, uuid_list)
1827 - if (!strcmp(hc->uuid, str)) {
1828 + if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1829 + !strcmp(hc->uuid, str)) {
1833 @@ -352,6 +355,9 @@ typedef int (*ioctl_fn)(struct dm_ioctl
1835 static int remove_all(struct dm_ioctl *param, size_t param_size)
1837 + if (!vx_check(0, VS_ADMIN))
1840 dm_hash_remove_all(1);
1841 param->data_size = 0;
1843 @@ -399,6 +405,8 @@ static int list_devices(struct dm_ioctl
1845 for (i = 0; i < NUM_BUCKETS; i++) {
1846 list_for_each_entry (hc, _name_buckets + i, name_list) {
1847 + if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1849 needed += sizeof(struct dm_name_list);
1850 needed += strlen(hc->name) + 1;
1851 needed += ALIGN_MASK;
1852 @@ -422,6 +430,8 @@ static int list_devices(struct dm_ioctl
1854 for (i = 0; i < NUM_BUCKETS; i++) {
1855 list_for_each_entry (hc, _name_buckets + i, name_list) {
1856 + if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1859 old_nl->next = (uint32_t) ((void *) nl -
1861 @@ -612,10 +622,11 @@ static struct hash_cell *__find_device_h
1865 - mdptr = dm_get_mdptr(md);
1866 + if (vx_check(dm_get_xid(md), VS_WATCH_P | VS_IDENT))
1867 + mdptr = dm_get_mdptr(md);
1875 @@ -1445,8 +1456,8 @@ static int ctl_ioctl(uint command, struc
1879 - /* only root can play with this */
1880 - if (!capable(CAP_SYS_ADMIN))
1881 + /* only root and certain contexts can play with this */
1882 + if (!vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_MAPPER))
1885 if (_IOC_TYPE(command) != DM_IOCTL)
1886 diff -NurpP --minimal linux-2.6.32.1/drivers/net/tun.c linux-2.6.32.1-vs2.3.0.36.27/drivers/net/tun.c
1887 --- linux-2.6.32.1/drivers/net/tun.c 2009-12-03 20:02:32.000000000 +0100
1888 +++ linux-2.6.32.1-vs2.3.0.36.27/drivers/net/tun.c 2009-12-03 20:04:56.000000000 +0100
1890 #include <linux/crc32.h>
1891 #include <linux/nsproxy.h>
1892 #include <linux/virtio_net.h>
1893 +#include <linux/vs_network.h>
1894 #include <net/net_namespace.h>
1895 #include <net/netns/generic.h>
1896 #include <net/rtnetlink.h>
1897 @@ -102,6 +103,7 @@ struct tun_struct {
1903 struct net_device *dev;
1904 struct fasync_struct *fasync;
1905 @@ -816,6 +818,7 @@ static void tun_setup(struct net_device
1909 + tun->nid = current->nid;
1911 dev->ethtool_ops = &tun_ethtool_ops;
1912 dev->destructor = tun_free_netdev;
1913 @@ -932,7 +935,7 @@ static int tun_set_iff(struct net *net,
1915 if (((tun->owner != -1 && cred->euid != tun->owner) ||
1916 (tun->group != -1 && !in_egroup_p(tun->group))) &&
1917 - !capable(CAP_NET_ADMIN))
1918 + !cap_raised(current_cap(), CAP_NET_ADMIN))
1920 err = security_tun_dev_attach(tun->socket.sk);
1922 @@ -946,7 +949,7 @@ static int tun_set_iff(struct net *net,
1924 unsigned long flags = 0;
1926 - if (!capable(CAP_NET_ADMIN))
1927 + if (!nx_capable(CAP_NET_ADMIN, NXC_TUN_CREATE))
1929 err = security_tun_dev_create();
1931 @@ -1013,6 +1016,9 @@ static int tun_set_iff(struct net *net,
1933 sk->sk_destruct = tun_sock_destruct;
1935 + if (!nx_check(tun->nid, VS_IDENT | VS_HOSTID | VS_ADMIN_P))
1938 err = tun_attach(tun, file);
1941 @@ -1202,6 +1208,16 @@ static long tun_chr_ioctl(struct file *f
1942 DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
1946 + if (!capable(CAP_CONTEXT))
1949 + /* Set nid owner of the device */
1950 + tun->nid = (nid_t) arg;
1952 + DBG(KERN_INFO "%s: nid owner set to %u\n", tun->dev->name, tun->nid);
1956 /* Only allow setting the type when the interface is down */
1957 if (tun->dev->flags & IFF_UP) {
1958 diff -NurpP --minimal linux-2.6.32.1/fs/attr.c linux-2.6.32.1-vs2.3.0.36.27/fs/attr.c
1959 --- linux-2.6.32.1/fs/attr.c 2009-12-03 20:02:51.000000000 +0100
1960 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/attr.c 2009-12-03 20:04:56.000000000 +0100
1962 #include <linux/fcntl.h>
1963 #include <linux/quotaops.h>
1964 #include <linux/security.h>
1965 +#include <linux/proc_fs.h>
1966 +#include <linux/devpts_fs.h>
1967 +#include <linux/vs_tag.h>
1969 /* Taken over from the old code... */
1971 @@ -55,6 +58,10 @@ int inode_change_ok(const struct inode *
1972 if (!is_owner_or_cap(inode))
1976 + if (dx_permission(inode, MAY_WRITE))
1982 @@ -120,6 +127,8 @@ int inode_setattr(struct inode * inode,
1983 inode->i_uid = attr->ia_uid;
1984 if (ia_valid & ATTR_GID)
1985 inode->i_gid = attr->ia_gid;
1986 + if ((ia_valid & ATTR_TAG) && IS_TAGGED(inode))
1987 + inode->i_tag = attr->ia_tag;
1988 if (ia_valid & ATTR_ATIME)
1989 inode->i_atime = timespec_trunc(attr->ia_atime,
1990 inode->i_sb->s_time_gran);
1991 @@ -214,7 +223,8 @@ int notify_change(struct dentry * dentry
1992 error = inode_change_ok(inode, attr);
1994 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
1995 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid))
1996 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
1997 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag))
1998 error = vfs_dq_transfer(inode, attr) ?
2001 diff -NurpP --minimal linux-2.6.32.1/fs/binfmt_aout.c linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_aout.c
2002 --- linux-2.6.32.1/fs/binfmt_aout.c 2009-03-24 14:22:24.000000000 +0100
2003 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_aout.c 2009-12-03 20:04:56.000000000 +0100
2005 #include <linux/binfmts.h>
2006 #include <linux/personality.h>
2007 #include <linux/init.h>
2008 +#include <linux/vs_memory.h>
2010 #include <asm/system.h>
2011 #include <asm/uaccess.h>
2012 diff -NurpP --minimal linux-2.6.32.1/fs/binfmt_elf.c linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_elf.c
2013 --- linux-2.6.32.1/fs/binfmt_elf.c 2009-12-03 20:02:51.000000000 +0100
2014 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_elf.c 2009-12-03 20:04:56.000000000 +0100
2016 #include <linux/random.h>
2017 #include <linux/elf.h>
2018 #include <linux/utsname.h>
2019 +#include <linux/vs_memory.h>
2020 #include <asm/uaccess.h>
2021 #include <asm/param.h>
2022 #include <asm/page.h>
2023 diff -NurpP --minimal linux-2.6.32.1/fs/binfmt_flat.c linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_flat.c
2024 --- linux-2.6.32.1/fs/binfmt_flat.c 2009-12-03 20:02:51.000000000 +0100
2025 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_flat.c 2009-12-03 20:04:56.000000000 +0100
2027 #include <linux/init.h>
2028 #include <linux/flat.h>
2029 #include <linux/syscalls.h>
2030 +#include <linux/vs_memory.h>
2032 #include <asm/byteorder.h>
2033 #include <asm/system.h>
2034 diff -NurpP --minimal linux-2.6.32.1/fs/binfmt_som.c linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_som.c
2035 --- linux-2.6.32.1/fs/binfmt_som.c 2009-06-11 17:13:02.000000000 +0200
2036 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/binfmt_som.c 2009-12-03 20:04:56.000000000 +0100
2038 #include <linux/shm.h>
2039 #include <linux/personality.h>
2040 #include <linux/init.h>
2041 +#include <linux/vs_memory.h>
2043 #include <asm/uaccess.h>
2044 #include <asm/pgtable.h>
2045 diff -NurpP --minimal linux-2.6.32.1/fs/block_dev.c linux-2.6.32.1-vs2.3.0.36.27/fs/block_dev.c
2046 --- linux-2.6.32.1/fs/block_dev.c 2009-12-03 20:02:51.000000000 +0100
2047 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/block_dev.c 2009-12-03 20:04:56.000000000 +0100
2049 #include <linux/namei.h>
2050 #include <linux/log2.h>
2051 #include <linux/kmemleak.h>
2052 +#include <linux/vs_device.h>
2053 #include <asm/uaccess.h>
2054 #include "internal.h"
2056 @@ -555,6 +556,7 @@ struct block_device *bdget(dev_t dev)
2057 bdev->bd_invalidated = 0;
2058 inode->i_mode = S_IFBLK;
2059 inode->i_rdev = dev;
2060 + inode->i_mdev = dev;
2061 inode->i_bdev = bdev;
2062 inode->i_data.a_ops = &def_blk_aops;
2063 mapping_set_gfp_mask(&inode->i_data, GFP_USER);
2064 @@ -601,6 +603,11 @@ EXPORT_SYMBOL(bdput);
2065 static struct block_device *bd_acquire(struct inode *inode)
2067 struct block_device *bdev;
2070 + if (!vs_map_blkdev(inode->i_rdev, &mdev, DATTR_OPEN))
2072 + inode->i_mdev = mdev;
2074 spin_lock(&bdev_lock);
2075 bdev = inode->i_bdev;
2076 @@ -611,7 +618,7 @@ static struct block_device *bd_acquire(s
2078 spin_unlock(&bdev_lock);
2080 - bdev = bdget(inode->i_rdev);
2081 + bdev = bdget(mdev);
2083 spin_lock(&bdev_lock);
2084 if (!inode->i_bdev) {
2085 diff -NurpP --minimal linux-2.6.32.1/fs/btrfs/ctree.h linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/ctree.h
2086 --- linux-2.6.32.1/fs/btrfs/ctree.h 2009-12-03 20:02:51.000000000 +0100
2087 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/ctree.h 2009-12-03 20:04:56.000000000 +0100
2088 @@ -544,11 +544,14 @@ struct btrfs_inode_item {
2089 /* modification sequence number for NFS */
2094 * a little future expansion, for more than this we can
2095 * just grow the inode item and version it
2097 - __le64 reserved[4];
2098 + __le16 reserved16;
2099 + __le32 reserved32;
2100 + __le64 reserved[3];
2101 struct btrfs_timespec atime;
2102 struct btrfs_timespec ctime;
2103 struct btrfs_timespec mtime;
2104 @@ -1155,6 +1158,8 @@ struct btrfs_root {
2105 #define BTRFS_MOUNT_NOSSD (1 << 9)
2106 #define BTRFS_MOUNT_DISCARD (1 << 10)
2108 +#define BTRFS_MOUNT_TAGGED (1 << 24)
2110 #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt)
2111 #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt)
2112 #define btrfs_test_opt(root, opt) ((root)->fs_info->mount_opt & \
2113 @@ -1174,6 +1179,10 @@ struct btrfs_root {
2114 #define BTRFS_INODE_NOATIME (1 << 9)
2115 #define BTRFS_INODE_DIRSYNC (1 << 10)
2117 +#define BTRFS_INODE_IXUNLINK (1 << 24)
2118 +#define BTRFS_INODE_BARRIER (1 << 25)
2119 +#define BTRFS_INODE_COW (1 << 26)
2122 /* some macros to generate set/get funcs for the struct fields. This
2123 * assumes there is a lefoo_to_cpu for every type, so lets make a simple
2124 @@ -1376,6 +1385,7 @@ BTRFS_SETGET_FUNCS(inode_block_group, st
2125 BTRFS_SETGET_FUNCS(inode_nlink, struct btrfs_inode_item, nlink, 32);
2126 BTRFS_SETGET_FUNCS(inode_uid, struct btrfs_inode_item, uid, 32);
2127 BTRFS_SETGET_FUNCS(inode_gid, struct btrfs_inode_item, gid, 32);
2128 +BTRFS_SETGET_FUNCS(inode_tag, struct btrfs_inode_item, tag, 16);
2129 BTRFS_SETGET_FUNCS(inode_mode, struct btrfs_inode_item, mode, 32);
2130 BTRFS_SETGET_FUNCS(inode_rdev, struct btrfs_inode_item, rdev, 64);
2131 BTRFS_SETGET_FUNCS(inode_flags, struct btrfs_inode_item, flags, 64);
2132 @@ -2338,6 +2348,7 @@ extern const struct dentry_operations bt
2133 long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2134 void btrfs_update_iflags(struct inode *inode);
2135 void btrfs_inherit_iflags(struct inode *inode, struct inode *dir);
2136 +int btrfs_sync_flags(struct inode *inode, int, int);
2139 int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync);
2140 diff -NurpP --minimal linux-2.6.32.1/fs/btrfs/disk-io.c linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/disk-io.c
2141 --- linux-2.6.32.1/fs/btrfs/disk-io.c 2009-12-03 20:02:51.000000000 +0100
2142 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/disk-io.c 2009-12-03 20:04:56.000000000 +0100
2143 @@ -1723,6 +1723,9 @@ struct btrfs_root *open_ctree(struct sup
2147 + if (btrfs_test_opt(tree_root, TAGGED))
2148 + sb->s_flags |= MS_TAGGED;
2150 features = btrfs_super_incompat_flags(disk_super) &
2151 ~BTRFS_FEATURE_INCOMPAT_SUPP;
2153 diff -NurpP --minimal linux-2.6.32.1/fs/btrfs/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/inode.c
2154 --- linux-2.6.32.1/fs/btrfs/inode.c 2009-12-03 20:02:51.000000000 +0100
2155 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/inode.c 2009-12-03 20:04:56.000000000 +0100
2157 #include <linux/xattr.h>
2158 #include <linux/posix_acl.h>
2159 #include <linux/falloc.h>
2160 +#include <linux/vs_tag.h>
2164 #include "disk-io.h"
2165 @@ -2242,6 +2244,8 @@ static void btrfs_read_locked_inode(stru
2167 u64 alloc_group_block;
2173 path = btrfs_alloc_path();
2174 @@ -2258,8 +2262,13 @@ static void btrfs_read_locked_inode(stru
2176 inode->i_mode = btrfs_inode_mode(leaf, inode_item);
2177 inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
2178 - inode->i_uid = btrfs_inode_uid(leaf, inode_item);
2179 - inode->i_gid = btrfs_inode_gid(leaf, inode_item);
2181 + uid = btrfs_inode_uid(leaf, inode_item);
2182 + gid = btrfs_inode_gid(leaf, inode_item);
2183 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2184 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2185 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2186 + btrfs_inode_tag(leaf, inode_item));
2187 btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item));
2189 tspec = btrfs_inode_atime(inode_item);
2190 @@ -2341,8 +2350,15 @@ static void fill_inode_item(struct btrfs
2191 struct btrfs_inode_item *item,
2192 struct inode *inode)
2194 - btrfs_set_inode_uid(leaf, item, inode->i_uid);
2195 - btrfs_set_inode_gid(leaf, item, inode->i_gid);
2196 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2197 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2199 + btrfs_set_inode_uid(leaf, item, uid);
2200 + btrfs_set_inode_gid(leaf, item, gid);
2201 +#ifdef CONFIG_TAGGING_INTERN
2202 + btrfs_set_inode_tag(leaf, item, inode->i_tag);
2205 btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size);
2206 btrfs_set_inode_mode(leaf, item, inode->i_mode);
2207 btrfs_set_inode_nlink(leaf, item, inode->i_nlink);
2208 @@ -4065,6 +4081,7 @@ static struct inode *btrfs_new_inode(str
2210 inode->i_gid = current_fsgid();
2212 + inode->i_tag = dx_current_fstag(root->fs_info->sb);
2213 inode->i_mode = mode;
2214 inode->i_ino = objectid;
2215 inode_set_bytes(inode, 0);
2216 @@ -5836,6 +5853,7 @@ static const struct inode_operations btr
2217 .listxattr = btrfs_listxattr,
2218 .removexattr = btrfs_removexattr,
2219 .permission = btrfs_permission,
2220 + .sync_flags = btrfs_sync_flags,
2222 static const struct inode_operations btrfs_dir_ro_inode_operations = {
2223 .lookup = btrfs_lookup,
2224 @@ -5911,6 +5929,7 @@ static const struct inode_operations btr
2225 .permission = btrfs_permission,
2226 .fallocate = btrfs_fallocate,
2227 .fiemap = btrfs_fiemap,
2228 + .sync_flags = btrfs_sync_flags,
2230 static const struct inode_operations btrfs_special_inode_operations = {
2231 .getattr = btrfs_getattr,
2232 diff -NurpP --minimal linux-2.6.32.1/fs/btrfs/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/ioctl.c
2233 --- linux-2.6.32.1/fs/btrfs/ioctl.c 2009-12-03 20:02:51.000000000 +0100
2234 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/ioctl.c 2009-12-03 20:04:56.000000000 +0100
2235 @@ -67,10 +67,13 @@ static unsigned int btrfs_flags_to_ioctl
2237 unsigned int iflags = 0;
2239 - if (flags & BTRFS_INODE_SYNC)
2240 - iflags |= FS_SYNC_FL;
2241 if (flags & BTRFS_INODE_IMMUTABLE)
2242 iflags |= FS_IMMUTABLE_FL;
2243 + if (flags & BTRFS_INODE_IXUNLINK)
2244 + iflags |= FS_IXUNLINK_FL;
2246 + if (flags & BTRFS_INODE_SYNC)
2247 + iflags |= FS_SYNC_FL;
2248 if (flags & BTRFS_INODE_APPEND)
2249 iflags |= FS_APPEND_FL;
2250 if (flags & BTRFS_INODE_NODUMP)
2251 @@ -80,28 +83,78 @@ static unsigned int btrfs_flags_to_ioctl
2252 if (flags & BTRFS_INODE_DIRSYNC)
2253 iflags |= FS_DIRSYNC_FL;
2255 + if (flags & BTRFS_INODE_BARRIER)
2256 + iflags |= FS_BARRIER_FL;
2257 + if (flags & BTRFS_INODE_COW)
2258 + iflags |= FS_COW_FL;
2263 - * Update inode->i_flags based on the btrfs internal flags.
2264 + * Update inode->i_(v)flags based on the btrfs internal flags.
2266 void btrfs_update_iflags(struct inode *inode)
2268 struct btrfs_inode *ip = BTRFS_I(inode);
2270 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2271 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2272 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2274 - if (ip->flags & BTRFS_INODE_SYNC)
2275 - inode->i_flags |= S_SYNC;
2276 if (ip->flags & BTRFS_INODE_IMMUTABLE)
2277 inode->i_flags |= S_IMMUTABLE;
2278 + if (ip->flags & BTRFS_INODE_IXUNLINK)
2279 + inode->i_flags |= S_IXUNLINK;
2281 + if (ip->flags & BTRFS_INODE_SYNC)
2282 + inode->i_flags |= S_SYNC;
2283 if (ip->flags & BTRFS_INODE_APPEND)
2284 inode->i_flags |= S_APPEND;
2285 if (ip->flags & BTRFS_INODE_NOATIME)
2286 inode->i_flags |= S_NOATIME;
2287 if (ip->flags & BTRFS_INODE_DIRSYNC)
2288 inode->i_flags |= S_DIRSYNC;
2290 + inode->i_vflags &= ~(V_BARRIER | V_COW);
2292 + if (ip->flags & BTRFS_INODE_BARRIER)
2293 + inode->i_vflags |= V_BARRIER;
2294 + if (ip->flags & BTRFS_INODE_COW)
2295 + inode->i_vflags |= V_COW;
2299 + * Update btrfs internal flags from inode->i_(v)flags.
2301 +void btrfs_update_flags(struct inode *inode)
2303 + struct btrfs_inode *ip = BTRFS_I(inode);
2305 + unsigned int flags = inode->i_flags;
2306 + unsigned int vflags = inode->i_vflags;
2308 + ip->flags &= ~(BTRFS_INODE_SYNC | BTRFS_INODE_APPEND |
2309 + BTRFS_INODE_IMMUTABLE | BTRFS_INODE_IXUNLINK |
2310 + BTRFS_INODE_NOATIME | BTRFS_INODE_DIRSYNC |
2311 + BTRFS_INODE_BARRIER | BTRFS_INODE_COW);
2313 + if (flags & S_IMMUTABLE)
2314 + ip->flags |= BTRFS_INODE_IMMUTABLE;
2315 + if (flags & S_IXUNLINK)
2316 + ip->flags |= BTRFS_INODE_IXUNLINK;
2318 + if (flags & S_SYNC)
2319 + ip->flags |= BTRFS_INODE_SYNC;
2320 + if (flags & S_APPEND)
2321 + ip->flags |= BTRFS_INODE_APPEND;
2322 + if (flags & S_NOATIME)
2323 + ip->flags |= BTRFS_INODE_NOATIME;
2324 + if (flags & S_DIRSYNC)
2325 + ip->flags |= BTRFS_INODE_DIRSYNC;
2327 + if (vflags & V_BARRIER)
2328 + ip->flags |= BTRFS_INODE_BARRIER;
2329 + if (vflags & V_COW)
2330 + ip->flags |= BTRFS_INODE_COW;
2334 @@ -119,7 +172,7 @@ void btrfs_inherit_iflags(struct inode *
2335 flags = BTRFS_I(dir)->flags;
2337 if (S_ISREG(inode->i_mode))
2338 - flags &= ~BTRFS_INODE_DIRSYNC;
2339 + flags &= ~(BTRFS_INODE_DIRSYNC | BTRFS_INODE_BARRIER);
2340 else if (!S_ISDIR(inode->i_mode))
2341 flags &= (BTRFS_INODE_NODUMP | BTRFS_INODE_NOATIME);
2343 @@ -127,6 +180,30 @@ void btrfs_inherit_iflags(struct inode *
2344 btrfs_update_iflags(inode);
2347 +int btrfs_sync_flags(struct inode *inode, int flags, int vflags)
2349 + struct btrfs_inode *ip = BTRFS_I(inode);
2350 + struct btrfs_root *root = ip->root;
2351 + struct btrfs_trans_handle *trans;
2354 + trans = btrfs_join_transaction(root, 1);
2357 + inode->i_flags = flags;
2358 + inode->i_vflags = vflags;
2359 + btrfs_update_flags(inode);
2361 + ret = btrfs_update_inode(trans, root, inode);
2364 + btrfs_update_iflags(inode);
2365 + inode->i_ctime = CURRENT_TIME;
2366 + btrfs_end_transaction(trans, root);
2371 static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
2373 struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode);
2374 @@ -149,6 +226,7 @@ static int btrfs_ioctl_setflags(struct f
2375 if (copy_from_user(&flags, arg, sizeof(flags)))
2378 + /* maybe add FS_IXUNLINK_FL ? */
2379 if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
2380 FS_NOATIME_FL | FS_NODUMP_FL | \
2381 FS_SYNC_FL | FS_DIRSYNC_FL))
2382 @@ -161,7 +239,8 @@ static int btrfs_ioctl_setflags(struct f
2384 flags = btrfs_mask_flags(inode->i_mode, flags);
2385 oldflags = btrfs_flags_to_ioctl(ip->flags);
2386 - if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
2387 + if ((flags ^ oldflags) & (FS_APPEND_FL |
2388 + FS_IMMUTABLE_FL | FS_IXUNLINK_FL)) {
2389 if (!capable(CAP_LINUX_IMMUTABLE)) {
2392 @@ -172,14 +251,19 @@ static int btrfs_ioctl_setflags(struct f
2396 - if (flags & FS_SYNC_FL)
2397 - ip->flags |= BTRFS_INODE_SYNC;
2399 - ip->flags &= ~BTRFS_INODE_SYNC;
2400 if (flags & FS_IMMUTABLE_FL)
2401 ip->flags |= BTRFS_INODE_IMMUTABLE;
2403 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
2404 + if (flags & FS_IXUNLINK_FL)
2405 + ip->flags |= BTRFS_INODE_IXUNLINK;
2407 + ip->flags &= ~BTRFS_INODE_IXUNLINK;
2409 + if (flags & FS_SYNC_FL)
2410 + ip->flags |= BTRFS_INODE_SYNC;
2412 + ip->flags &= ~BTRFS_INODE_SYNC;
2413 if (flags & FS_APPEND_FL)
2414 ip->flags |= BTRFS_INODE_APPEND;
2416 diff -NurpP --minimal linux-2.6.32.1/fs/btrfs/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/super.c
2417 --- linux-2.6.32.1/fs/btrfs/super.c 2009-12-03 20:02:51.000000000 +0100
2418 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/btrfs/super.c 2009-12-03 20:04:56.000000000 +0100
2419 @@ -67,7 +67,7 @@ enum {
2420 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
2421 Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
2422 Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
2423 - Opt_discard, Opt_err,
2424 + Opt_tag, Opt_notag, Opt_tagid, Opt_discard, Opt_err,
2427 static match_table_t tokens = {
2428 @@ -90,6 +90,9 @@ static match_table_t tokens = {
2429 {Opt_flushoncommit, "flushoncommit"},
2430 {Opt_ratio, "metadata_ratio=%d"},
2431 {Opt_discard, "discard"},
2433 + {Opt_notag, "notag"},
2434 + {Opt_tagid, "tagid=%u"},
2438 @@ -262,6 +265,22 @@ int btrfs_parse_options(struct btrfs_roo
2440 btrfs_set_opt(info->mount_opt, DISCARD);
2442 +#ifndef CONFIG_TAGGING_NONE
2444 + printk(KERN_INFO "btrfs: use tagging\n");
2445 + btrfs_set_opt(info->mount_opt, TAGGED);
2448 + printk(KERN_INFO "btrfs: disabled tagging\n");
2449 + btrfs_clear_opt(info->mount_opt, TAGGED);
2452 +#ifdef CONFIG_PROPAGATE
2455 + btrfs_set_opt(info->mount_opt, TAGGED);
2461 @@ -575,6 +594,12 @@ static int btrfs_remount(struct super_bl
2465 + if (btrfs_test_opt(root, TAGGED) && !(sb->s_flags & MS_TAGGED)) {
2466 + printk("btrfs: %s: tagging not permitted on remount.\n",
2471 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
2474 diff -NurpP --minimal linux-2.6.32.1/fs/char_dev.c linux-2.6.32.1-vs2.3.0.36.27/fs/char_dev.c
2475 --- linux-2.6.32.1/fs/char_dev.c 2009-12-03 20:02:51.000000000 +0100
2476 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/char_dev.c 2009-12-03 20:04:56.000000000 +0100
2478 #include <linux/cdev.h>
2479 #include <linux/mutex.h>
2480 #include <linux/backing-dev.h>
2481 +#include <linux/vs_context.h>
2482 +#include <linux/vs_device.h>
2484 #include "internal.h"
2486 @@ -368,14 +370,21 @@ static int chrdev_open(struct inode *ino
2488 struct cdev *new = NULL;
2492 + if (!vs_map_chrdev(inode->i_rdev, &mdev, DATTR_OPEN))
2494 + inode->i_mdev = mdev;
2496 spin_lock(&cdev_lock);
2499 struct kobject *kobj;
2502 spin_unlock(&cdev_lock);
2503 - kobj = kobj_lookup(cdev_map, inode->i_rdev, &idx);
2505 + kobj = kobj_lookup(cdev_map, mdev, &idx);
2508 new = container_of(kobj, struct cdev, kobj);
2509 diff -NurpP --minimal linux-2.6.32.1/fs/dcache.c linux-2.6.32.1-vs2.3.0.36.27/fs/dcache.c
2510 --- linux-2.6.32.1/fs/dcache.c 2009-12-03 20:02:51.000000000 +0100
2511 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/dcache.c 2009-12-03 20:04:56.000000000 +0100
2513 #include <linux/bootmem.h>
2514 #include <linux/fs_struct.h>
2515 #include <linux/hardirq.h>
2516 +#include <linux/vs_limit.h>
2517 #include "internal.h"
2519 int sysctl_vfs_cache_pressure __read_mostly = 100;
2520 @@ -230,6 +231,8 @@ repeat:
2524 + vx_dentry_dec(dentry);
2527 * AV: ->d_delete() is _NOT_ allowed to block now.
2529 @@ -321,6 +324,7 @@ static inline struct dentry * __dget_loc
2531 atomic_inc(&dentry->d_count);
2532 dentry_lru_del_init(dentry);
2533 + vx_dentry_inc(dentry);
2537 @@ -919,6 +923,9 @@ struct dentry *d_alloc(struct dentry * p
2538 struct dentry *dentry;
2541 + if (!vx_dentry_avail(1))
2544 dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL);
2547 @@ -964,6 +971,7 @@ struct dentry *d_alloc(struct dentry * p
2549 list_add(&dentry->d_u.d_child, &parent->d_subdirs);
2550 dentry_stat.nr_dentry++;
2551 + vx_dentry_inc(dentry);
2552 spin_unlock(&dcache_lock);
2555 @@ -1407,6 +1415,7 @@ struct dentry * __d_lookup(struct dentry
2558 atomic_inc(&dentry->d_count);
2559 + vx_dentry_inc(dentry);
2561 spin_unlock(&dentry->d_lock);
2563 diff -NurpP --minimal linux-2.6.32.1/fs/devpts/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/devpts/inode.c
2564 --- linux-2.6.32.1/fs/devpts/inode.c 2009-12-03 20:02:51.000000000 +0100
2565 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/devpts/inode.c 2009-12-03 20:04:56.000000000 +0100
2567 #include <linux/parser.h>
2568 #include <linux/fsnotify.h>
2569 #include <linux/seq_file.h>
2570 +#include <linux/vs_base.h>
2572 #define DEVPTS_DEFAULT_MODE 0600
2575 #define DEVPTS_DEFAULT_PTMX_MODE 0000
2576 #define PTMX_MINOR 2
2578 +static int devpts_permission(struct inode *inode, int mask)
2580 + int ret = -EACCES;
2582 + /* devpts is xid tagged */
2583 + if (vx_check((xid_t)inode->i_tag, VS_WATCH_P | VS_IDENT))
2584 + ret = generic_permission(inode, mask, NULL);
2588 +static struct inode_operations devpts_file_inode_operations = {
2589 + .permission = devpts_permission,
2592 extern int pty_limit; /* Config limit on Unix98 ptys */
2593 static DEFINE_MUTEX(allocated_ptys_lock);
2595 @@ -262,6 +277,25 @@ static int devpts_show_options(struct se
2599 +static int devpts_filter(struct dentry *de)
2601 + /* devpts is xid tagged */
2602 + return vx_check((xid_t)de->d_inode->i_tag, VS_WATCH_P | VS_IDENT);
2605 +static int devpts_readdir(struct file * filp, void * dirent, filldir_t filldir)
2607 + return dcache_readdir_filter(filp, dirent, filldir, devpts_filter);
2610 +static struct file_operations devpts_dir_operations = {
2611 + .open = dcache_dir_open,
2612 + .release = dcache_dir_close,
2613 + .llseek = dcache_dir_lseek,
2614 + .read = generic_read_dir,
2615 + .readdir = devpts_readdir,
2618 static const struct super_operations devpts_sops = {
2619 .statfs = simple_statfs,
2620 .remount_fs = devpts_remount,
2621 @@ -301,12 +335,15 @@ devpts_fill_super(struct super_block *s,
2622 inode = new_inode(s);
2627 inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2628 inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO | S_IWUSR;
2629 inode->i_op = &simple_dir_inode_operations;
2630 - inode->i_fop = &simple_dir_operations;
2631 + inode->i_fop = &devpts_dir_operations;
2633 + /* devpts is xid tagged */
2634 + inode->i_tag = (tag_t)vx_current_xid();
2636 s->s_root = d_alloc_root(inode);
2638 @@ -497,6 +534,9 @@ int devpts_pty_new(struct inode *ptmx_in
2639 inode->i_gid = opts->setgid ? opts->gid : current_fsgid();
2640 inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2641 init_special_inode(inode, S_IFCHR|opts->mode, device);
2642 + /* devpts is xid tagged */
2643 + inode->i_tag = (tag_t)vx_current_xid();
2644 + inode->i_op = &devpts_file_inode_operations;
2645 inode->i_private = tty;
2646 tty->driver_data = inode;
2648 diff -NurpP --minimal linux-2.6.32.1/fs/exec.c linux-2.6.32.1-vs2.3.0.36.27/fs/exec.c
2649 --- linux-2.6.32.1/fs/exec.c 2009-12-03 20:02:51.000000000 +0100
2650 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/exec.c 2009-12-03 20:04:56.000000000 +0100
2651 @@ -251,7 +251,9 @@ static int __bprm_mm_init(struct linux_b
2655 - mm->stack_vm = mm->total_vm = 1;
2657 + vx_vmpages_inc(mm);
2659 up_write(&mm->mmap_sem);
2660 bprm->p = vma->vm_end - sizeof(void *);
2662 @@ -1474,7 +1476,7 @@ static int format_corename(char *corenam
2663 /* UNIX time of coredump */
2666 - do_gettimeofday(&tv);
2667 + vx_gettimeofday(&tv);
2668 rc = snprintf(out_ptr, out_end - out_ptr,
2670 if (rc > out_end - out_ptr)
2671 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/balloc.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/balloc.c
2672 --- linux-2.6.32.1/fs/ext2/balloc.c 2009-06-11 17:13:03.000000000 +0200
2673 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/balloc.c 2009-12-03 20:04:56.000000000 +0100
2674 @@ -701,7 +701,6 @@ ext2_try_to_allocate(struct super_block
2676 end = EXT2_BLOCKS_PER_GROUP(sb);
2679 BUG_ON(start > EXT2_BLOCKS_PER_GROUP(sb));
2682 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/ext2.h linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ext2.h
2683 --- linux-2.6.32.1/fs/ext2/ext2.h 2009-09-10 15:26:21.000000000 +0200
2684 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ext2.h 2009-12-03 20:04:56.000000000 +0100
2685 @@ -131,6 +131,7 @@ extern int ext2_fiemap(struct inode *ino
2686 int __ext2_write_begin(struct file *file, struct address_space *mapping,
2687 loff_t pos, unsigned len, unsigned flags,
2688 struct page **pagep, void **fsdata);
2689 +extern int ext2_sync_flags(struct inode *, int, int);
2692 extern long ext2_ioctl(struct file *, unsigned int, unsigned long);
2693 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/file.c
2694 --- linux-2.6.32.1/fs/ext2/file.c 2009-12-03 20:02:51.000000000 +0100
2695 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/file.c 2009-12-03 20:04:56.000000000 +0100
2696 @@ -87,4 +87,5 @@ const struct inode_operations ext2_file_
2697 .setattr = ext2_setattr,
2698 .check_acl = ext2_check_acl,
2699 .fiemap = ext2_fiemap,
2700 + .sync_flags = ext2_sync_flags,
2702 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/ialloc.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ialloc.c
2703 --- linux-2.6.32.1/fs/ext2/ialloc.c 2009-06-11 17:13:03.000000000 +0200
2704 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ialloc.c 2009-12-03 20:04:56.000000000 +0100
2706 #include <linux/backing-dev.h>
2707 #include <linux/buffer_head.h>
2708 #include <linux/random.h>
2709 +#include <linux/vs_tag.h>
2713 @@ -560,6 +561,7 @@ got:
2715 inode->i_gid = current_fsgid();
2716 inode->i_mode = mode;
2717 + inode->i_tag = dx_current_fstag(sb);
2720 inode->i_blocks = 0;
2721 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/inode.c
2722 --- linux-2.6.32.1/fs/ext2/inode.c 2009-12-03 20:02:51.000000000 +0100
2723 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/inode.c 2009-12-03 20:04:56.000000000 +0100
2725 #include <linux/mpage.h>
2726 #include <linux/fiemap.h>
2727 #include <linux/namei.h>
2728 +#include <linux/vs_tag.h>
2732 @@ -1040,7 +1041,7 @@ void ext2_truncate(struct inode *inode)
2734 if (ext2_inode_is_fast_symlink(inode))
2736 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2737 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2740 blocksize = inode->i_sb->s_blocksize;
2741 @@ -1178,36 +1179,61 @@ void ext2_set_inode_flags(struct inode *
2743 unsigned int flags = EXT2_I(inode)->i_flags;
2745 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2746 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2747 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2750 + if (flags & EXT2_IMMUTABLE_FL)
2751 + inode->i_flags |= S_IMMUTABLE;
2752 + if (flags & EXT2_IXUNLINK_FL)
2753 + inode->i_flags |= S_IXUNLINK;
2755 if (flags & EXT2_SYNC_FL)
2756 inode->i_flags |= S_SYNC;
2757 if (flags & EXT2_APPEND_FL)
2758 inode->i_flags |= S_APPEND;
2759 - if (flags & EXT2_IMMUTABLE_FL)
2760 - inode->i_flags |= S_IMMUTABLE;
2761 if (flags & EXT2_NOATIME_FL)
2762 inode->i_flags |= S_NOATIME;
2763 if (flags & EXT2_DIRSYNC_FL)
2764 inode->i_flags |= S_DIRSYNC;
2766 + inode->i_vflags &= ~(V_BARRIER | V_COW);
2768 + if (flags & EXT2_BARRIER_FL)
2769 + inode->i_vflags |= V_BARRIER;
2770 + if (flags & EXT2_COW_FL)
2771 + inode->i_vflags |= V_COW;
2774 /* Propagate flags from i_flags to EXT2_I(inode)->i_flags */
2775 void ext2_get_inode_flags(struct ext2_inode_info *ei)
2777 unsigned int flags = ei->vfs_inode.i_flags;
2778 + unsigned int vflags = ei->vfs_inode.i_vflags;
2780 + ei->i_flags &= ~(EXT2_SYNC_FL | EXT2_APPEND_FL |
2781 + EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL |
2782 + EXT2_NOATIME_FL | EXT2_DIRSYNC_FL |
2783 + EXT2_BARRIER_FL | EXT2_COW_FL);
2785 + if (flags & S_IMMUTABLE)
2786 + ei->i_flags |= EXT2_IMMUTABLE_FL;
2787 + if (flags & S_IXUNLINK)
2788 + ei->i_flags |= EXT2_IXUNLINK_FL;
2790 - ei->i_flags &= ~(EXT2_SYNC_FL|EXT2_APPEND_FL|
2791 - EXT2_IMMUTABLE_FL|EXT2_NOATIME_FL|EXT2_DIRSYNC_FL);
2793 ei->i_flags |= EXT2_SYNC_FL;
2794 if (flags & S_APPEND)
2795 ei->i_flags |= EXT2_APPEND_FL;
2796 - if (flags & S_IMMUTABLE)
2797 - ei->i_flags |= EXT2_IMMUTABLE_FL;
2798 if (flags & S_NOATIME)
2799 ei->i_flags |= EXT2_NOATIME_FL;
2800 if (flags & S_DIRSYNC)
2801 ei->i_flags |= EXT2_DIRSYNC_FL;
2803 + if (vflags & V_BARRIER)
2804 + ei->i_flags |= EXT2_BARRIER_FL;
2805 + if (vflags & V_COW)
2806 + ei->i_flags |= EXT2_COW_FL;
2809 struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
2810 @@ -1217,6 +1243,8 @@ struct inode *ext2_iget (struct super_bl
2811 struct ext2_inode *raw_inode;
2812 struct inode *inode;
2818 inode = iget_locked(sb, ino);
2819 @@ -1235,12 +1263,17 @@ struct inode *ext2_iget (struct super_bl
2822 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2823 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2824 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2825 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2826 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2827 if (!(test_opt (inode->i_sb, NO_UID32))) {
2828 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2829 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2830 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2831 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2833 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2834 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2835 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2836 + le16_to_cpu(raw_inode->i_raw_tag));
2838 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2839 inode->i_size = le32_to_cpu(raw_inode->i_size);
2840 inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2841 @@ -1338,8 +1371,8 @@ int ext2_write_inode(struct inode *inode
2842 struct ext2_inode_info *ei = EXT2_I(inode);
2843 struct super_block *sb = inode->i_sb;
2844 ino_t ino = inode->i_ino;
2845 - uid_t uid = inode->i_uid;
2846 - gid_t gid = inode->i_gid;
2847 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2848 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2849 struct buffer_head * bh;
2850 struct ext2_inode * raw_inode = ext2_get_inode(sb, ino, &bh);
2852 @@ -1375,6 +1408,9 @@ int ext2_write_inode(struct inode *inode
2853 raw_inode->i_uid_high = 0;
2854 raw_inode->i_gid_high = 0;
2856 +#ifdef CONFIG_TAGGING_INTERN
2857 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2859 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2860 raw_inode->i_size = cpu_to_le32(inode->i_size);
2861 raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2862 @@ -1456,7 +1492,8 @@ int ext2_setattr(struct dentry *dentry,
2865 if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
2866 - (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
2867 + (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
2868 + (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
2869 error = vfs_dq_transfer(inode, iattr) ? -EDQUOT : 0;
2872 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ioctl.c
2873 --- linux-2.6.32.1/fs/ext2/ioctl.c 2009-09-10 15:26:21.000000000 +0200
2874 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/ioctl.c 2009-12-03 20:04:56.000000000 +0100
2876 #include <asm/uaccess.h>
2879 +int ext2_sync_flags(struct inode *inode, int flags, int vflags)
2881 + inode->i_flags = flags;
2882 + inode->i_vflags = vflags;
2883 + ext2_get_inode_flags(EXT2_I(inode));
2884 + inode->i_ctime = CURRENT_TIME_SEC;
2885 + mark_inode_dirty(inode);
2889 long ext2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
2891 struct inode *inode = filp->f_dentry->d_inode;
2892 @@ -51,6 +61,11 @@ long ext2_ioctl(struct file *filp, unsig
2894 flags = ext2_mask_flags(inode->i_mode, flags);
2896 + if (IS_BARRIER(inode)) {
2897 + vxwprintk_task(1, "messing with the barrier.");
2901 mutex_lock(&inode->i_mutex);
2902 /* Is it quota file? Do not allow user to mess with it */
2903 if (IS_NOQUOTA(inode)) {
2904 @@ -66,7 +81,9 @@ long ext2_ioctl(struct file *filp, unsig
2906 * This test looks nicer. Thanks to Pauline Middelink
2908 - if ((flags ^ oldflags) & (EXT2_APPEND_FL | EXT2_IMMUTABLE_FL)) {
2909 + if ((oldflags & EXT2_IMMUTABLE_FL) ||
2910 + ((flags ^ oldflags) & (EXT2_APPEND_FL |
2911 + EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL))) {
2912 if (!capable(CAP_LINUX_IMMUTABLE)) {
2913 mutex_unlock(&inode->i_mutex);
2915 @@ -74,7 +91,7 @@ long ext2_ioctl(struct file *filp, unsig
2919 - flags = flags & EXT2_FL_USER_MODIFIABLE;
2920 + flags &= EXT2_FL_USER_MODIFIABLE;
2921 flags |= oldflags & ~EXT2_FL_USER_MODIFIABLE;
2922 ei->i_flags = flags;
2923 mutex_unlock(&inode->i_mutex);
2924 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/namei.c
2925 --- linux-2.6.32.1/fs/ext2/namei.c 2009-12-03 20:02:51.000000000 +0100
2926 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/namei.c 2009-12-03 20:04:56.000000000 +0100
2930 #include <linux/pagemap.h>
2931 +#include <linux/vs_tag.h>
2935 @@ -74,6 +75,7 @@ static struct dentry *ext2_lookup(struct
2936 return ERR_PTR(-EIO);
2938 return ERR_CAST(inode);
2939 + dx_propagate_tag(nd, inode);
2943 @@ -401,6 +403,7 @@ const struct inode_operations ext2_dir_i
2945 .setattr = ext2_setattr,
2946 .check_acl = ext2_check_acl,
2947 + .sync_flags = ext2_sync_flags,
2950 const struct inode_operations ext2_special_inode_operations = {
2951 diff -NurpP --minimal linux-2.6.32.1/fs/ext2/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/super.c
2952 --- linux-2.6.32.1/fs/ext2/super.c 2009-09-10 15:26:21.000000000 +0200
2953 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext2/super.c 2009-12-03 20:04:56.000000000 +0100
2954 @@ -382,7 +382,8 @@ enum {
2955 Opt_err_ro, Opt_nouid32, Opt_nocheck, Opt_debug,
2956 Opt_oldalloc, Opt_orlov, Opt_nobh, Opt_user_xattr, Opt_nouser_xattr,
2957 Opt_acl, Opt_noacl, Opt_xip, Opt_ignore, Opt_err, Opt_quota,
2958 - Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation
2959 + Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation,
2960 + Opt_tag, Opt_notag, Opt_tagid
2963 static const match_table_t tokens = {
2964 @@ -410,6 +411,9 @@ static const match_table_t tokens = {
2966 {Opt_noacl, "noacl"},
2969 + {Opt_notag, "notag"},
2970 + {Opt_tagid, "tagid=%u"},
2971 {Opt_grpquota, "grpquota"},
2972 {Opt_ignore, "noquota"},
2973 {Opt_quota, "quota"},
2974 @@ -480,6 +484,20 @@ static int parse_options (char * options
2976 set_opt (sbi->s_mount_opt, NO_UID32);
2978 +#ifndef CONFIG_TAGGING_NONE
2980 + set_opt (sbi->s_mount_opt, TAGGED);
2983 + clear_opt (sbi->s_mount_opt, TAGGED);
2986 +#ifdef CONFIG_PROPAGATE
2989 + set_opt (sbi->s_mount_opt, TAGGED);
2993 clear_opt (sbi->s_mount_opt, CHECK);
2995 @@ -829,6 +847,8 @@ static int ext2_fill_super(struct super_
2996 if (!parse_options ((char *) data, sbi))
2999 + if (EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_TAGGED)
3000 + sb->s_flags |= MS_TAGGED;
3001 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3002 ((EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ?
3004 @@ -1175,6 +1195,14 @@ static int ext2_remount (struct super_bl
3008 + if ((sbi->s_mount_opt & EXT2_MOUNT_TAGGED) &&
3009 + !(sb->s_flags & MS_TAGGED)) {
3010 + printk("EXT2-fs: %s: tagging not permitted on remount.\n",
3013 + goto restore_opts;
3016 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3017 ((sbi->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3019 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/file.c
3020 --- linux-2.6.32.1/fs/ext3/file.c 2009-12-03 20:02:51.000000000 +0100
3021 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/file.c 2009-12-03 20:04:56.000000000 +0100
3022 @@ -80,5 +80,6 @@ const struct inode_operations ext3_file_
3024 .check_acl = ext3_check_acl,
3025 .fiemap = ext3_fiemap,
3026 + .sync_flags = ext3_sync_flags,
3029 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/ialloc.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/ialloc.c
3030 --- linux-2.6.32.1/fs/ext3/ialloc.c 2009-09-10 15:26:21.000000000 +0200
3031 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/ialloc.c 2009-12-03 20:04:56.000000000 +0100
3033 #include <linux/buffer_head.h>
3034 #include <linux/random.h>
3035 #include <linux/bitops.h>
3036 +#include <linux/vs_tag.h>
3038 #include <asm/byteorder.h>
3040 @@ -548,6 +549,7 @@ got:
3042 inode->i_gid = current_fsgid();
3043 inode->i_mode = mode;
3044 + inode->i_tag = dx_current_fstag(sb);
3047 /* This is the optimal IO size (for stat), not the fs block size */
3048 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/inode.c
3049 --- linux-2.6.32.1/fs/ext3/inode.c 2009-12-03 20:02:51.000000000 +0100
3050 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/inode.c 2009-12-03 20:04:56.000000000 +0100
3052 #include <linux/bio.h>
3053 #include <linux/fiemap.h>
3054 #include <linux/namei.h>
3055 +#include <linux/vs_tag.h>
3059 @@ -2333,7 +2334,7 @@ static void ext3_free_branches(handle_t
3061 int ext3_can_truncate(struct inode *inode)
3063 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3064 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3066 if (S_ISREG(inode->i_mode))
3068 @@ -2718,36 +2719,60 @@ void ext3_set_inode_flags(struct inode *
3070 unsigned int flags = EXT3_I(inode)->i_flags;
3072 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3073 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3074 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3076 + if (flags & EXT3_IMMUTABLE_FL)
3077 + inode->i_flags |= S_IMMUTABLE;
3078 + if (flags & EXT3_IXUNLINK_FL)
3079 + inode->i_flags |= S_IXUNLINK;
3081 if (flags & EXT3_SYNC_FL)
3082 inode->i_flags |= S_SYNC;
3083 if (flags & EXT3_APPEND_FL)
3084 inode->i_flags |= S_APPEND;
3085 - if (flags & EXT3_IMMUTABLE_FL)
3086 - inode->i_flags |= S_IMMUTABLE;
3087 if (flags & EXT3_NOATIME_FL)
3088 inode->i_flags |= S_NOATIME;
3089 if (flags & EXT3_DIRSYNC_FL)
3090 inode->i_flags |= S_DIRSYNC;
3092 + inode->i_vflags &= ~(V_BARRIER | V_COW);
3094 + if (flags & EXT3_BARRIER_FL)
3095 + inode->i_vflags |= V_BARRIER;
3096 + if (flags & EXT3_COW_FL)
3097 + inode->i_vflags |= V_COW;
3100 /* Propagate flags from i_flags to EXT3_I(inode)->i_flags */
3101 void ext3_get_inode_flags(struct ext3_inode_info *ei)
3103 unsigned int flags = ei->vfs_inode.i_flags;
3104 + unsigned int vflags = ei->vfs_inode.i_vflags;
3106 + ei->i_flags &= ~(EXT3_SYNC_FL | EXT3_APPEND_FL |
3107 + EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL |
3108 + EXT3_NOATIME_FL | EXT3_DIRSYNC_FL |
3109 + EXT3_BARRIER_FL | EXT3_COW_FL);
3111 + if (flags & S_IMMUTABLE)
3112 + ei->i_flags |= EXT3_IMMUTABLE_FL;
3113 + if (flags & S_IXUNLINK)
3114 + ei->i_flags |= EXT3_IXUNLINK_FL;
3116 - ei->i_flags &= ~(EXT3_SYNC_FL|EXT3_APPEND_FL|
3117 - EXT3_IMMUTABLE_FL|EXT3_NOATIME_FL|EXT3_DIRSYNC_FL);
3119 ei->i_flags |= EXT3_SYNC_FL;
3120 if (flags & S_APPEND)
3121 ei->i_flags |= EXT3_APPEND_FL;
3122 - if (flags & S_IMMUTABLE)
3123 - ei->i_flags |= EXT3_IMMUTABLE_FL;
3124 if (flags & S_NOATIME)
3125 ei->i_flags |= EXT3_NOATIME_FL;
3126 if (flags & S_DIRSYNC)
3127 ei->i_flags |= EXT3_DIRSYNC_FL;
3129 + if (vflags & V_BARRIER)
3130 + ei->i_flags |= EXT3_BARRIER_FL;
3131 + if (vflags & V_COW)
3132 + ei->i_flags |= EXT3_COW_FL;
3135 struct inode *ext3_iget(struct super_block *sb, unsigned long ino)
3136 @@ -2761,6 +2786,8 @@ struct inode *ext3_iget(struct super_blo
3137 transaction_t *transaction;
3143 inode = iget_locked(sb, ino);
3145 @@ -2777,12 +2804,17 @@ struct inode *ext3_iget(struct super_blo
3147 raw_inode = ext3_raw_inode(&iloc);
3148 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3149 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3150 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3151 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3152 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3153 if(!(test_opt (inode->i_sb, NO_UID32))) {
3154 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3155 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3156 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3157 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3159 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3160 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3161 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3162 + le16_to_cpu(raw_inode->i_raw_tag));
3164 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3165 inode->i_size = le32_to_cpu(raw_inode->i_size);
3166 inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
3167 @@ -2937,6 +2969,8 @@ static int ext3_do_update_inode(handle_t
3168 struct ext3_inode *raw_inode = ext3_raw_inode(iloc);
3169 struct ext3_inode_info *ei = EXT3_I(inode);
3170 struct buffer_head *bh = iloc->bh;
3171 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3172 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3173 int err = 0, rc, block;
3176 @@ -2951,29 +2985,32 @@ again:
3177 ext3_get_inode_flags(ei);
3178 raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3179 if(!(test_opt(inode->i_sb, NO_UID32))) {
3180 - raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3181 - raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3182 + raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3183 + raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3185 * Fix up interoperability with old kernels. Otherwise, old inodes get
3186 * re-used with the upper 16 bits of the uid/gid intact
3189 raw_inode->i_uid_high =
3190 - cpu_to_le16(high_16_bits(inode->i_uid));
3191 + cpu_to_le16(high_16_bits(uid));
3192 raw_inode->i_gid_high =
3193 - cpu_to_le16(high_16_bits(inode->i_gid));
3194 + cpu_to_le16(high_16_bits(gid));
3196 raw_inode->i_uid_high = 0;
3197 raw_inode->i_gid_high = 0;
3200 raw_inode->i_uid_low =
3201 - cpu_to_le16(fs_high2lowuid(inode->i_uid));
3202 + cpu_to_le16(fs_high2lowuid(uid));
3203 raw_inode->i_gid_low =
3204 - cpu_to_le16(fs_high2lowgid(inode->i_gid));
3205 + cpu_to_le16(fs_high2lowgid(gid));
3206 raw_inode->i_uid_high = 0;
3207 raw_inode->i_gid_high = 0;
3209 +#ifdef CONFIG_TAGGING_INTERN
3210 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3212 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3213 raw_inode->i_size = cpu_to_le32(ei->i_disksize);
3214 raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
3215 @@ -3131,7 +3168,8 @@ int ext3_setattr(struct dentry *dentry,
3218 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3219 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3220 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3221 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3224 /* (user+group)*(old+new) structure, inode write (sb,
3225 @@ -3153,6 +3191,8 @@ int ext3_setattr(struct dentry *dentry,
3226 inode->i_uid = attr->ia_uid;
3227 if (attr->ia_valid & ATTR_GID)
3228 inode->i_gid = attr->ia_gid;
3229 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3230 + inode->i_tag = attr->ia_tag;
3231 error = ext3_mark_inode_dirty(handle, inode);
3232 ext3_journal_stop(handle);
3234 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/ioctl.c
3235 --- linux-2.6.32.1/fs/ext3/ioctl.c 2009-06-11 17:13:03.000000000 +0200
3236 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/ioctl.c 2009-12-03 20:04:56.000000000 +0100
3240 #include <linux/fs.h>
3241 +#include <linux/mount.h>
3242 #include <linux/jbd.h>
3243 #include <linux/capability.h>
3244 #include <linux/ext3_fs.h>
3246 #include <linux/compat.h>
3247 #include <asm/uaccess.h>
3250 +int ext3_sync_flags(struct inode *inode, int flags, int vflags)
3252 + handle_t *handle = NULL;
3253 + struct ext3_iloc iloc;
3256 + handle = ext3_journal_start(inode, 1);
3257 + if (IS_ERR(handle))
3258 + return PTR_ERR(handle);
3260 + if (IS_SYNC(inode))
3261 + handle->h_sync = 1;
3262 + err = ext3_reserve_inode_write(handle, inode, &iloc);
3266 + inode->i_flags = flags;
3267 + inode->i_vflags = vflags;
3268 + ext3_get_inode_flags(EXT3_I(inode));
3269 + inode->i_ctime = CURRENT_TIME_SEC;
3271 + err = ext3_mark_iloc_dirty(handle, inode, &iloc);
3273 + ext3_journal_stop(handle);
3277 long ext3_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
3279 struct inode *inode = filp->f_dentry->d_inode;
3280 @@ -50,6 +79,11 @@ long ext3_ioctl(struct file *filp, unsig
3282 flags = ext3_mask_flags(inode->i_mode, flags);
3284 + if (IS_BARRIER(inode)) {
3285 + vxwprintk_task(1, "messing with the barrier.");
3289 mutex_lock(&inode->i_mutex);
3291 /* Is it quota file? Do not allow user to mess with it */
3292 @@ -68,7 +102,9 @@ long ext3_ioctl(struct file *filp, unsig
3294 * This test looks nicer. Thanks to Pauline Middelink
3296 - if ((flags ^ oldflags) & (EXT3_APPEND_FL | EXT3_IMMUTABLE_FL)) {
3297 + if ((oldflags & EXT3_IMMUTABLE_FL) ||
3298 + ((flags ^ oldflags) & (EXT3_APPEND_FL |
3299 + EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL))) {
3300 if (!capable(CAP_LINUX_IMMUTABLE))
3303 @@ -93,7 +129,7 @@ long ext3_ioctl(struct file *filp, unsig
3307 - flags = flags & EXT3_FL_USER_MODIFIABLE;
3308 + flags &= EXT3_FL_USER_MODIFIABLE;
3309 flags |= oldflags & ~EXT3_FL_USER_MODIFIABLE;
3310 ei->i_flags = flags;
3312 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/namei.c
3313 --- linux-2.6.32.1/fs/ext3/namei.c 2009-12-03 20:02:51.000000000 +0100
3314 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/namei.c 2009-12-03 20:04:56.000000000 +0100
3316 #include <linux/quotaops.h>
3317 #include <linux/buffer_head.h>
3318 #include <linux/bio.h>
3319 +#include <linux/vs_tag.h>
3323 @@ -912,6 +913,7 @@ restart:
3325 ll_rw_block(READ_META, 1, &bh);
3327 + dx_propagate_tag(nd, inode);
3329 if ((bh = bh_use[ra_ptr++]) == NULL)
3331 @@ -2446,6 +2448,7 @@ const struct inode_operations ext3_dir_i
3332 .removexattr = generic_removexattr,
3334 .check_acl = ext3_check_acl,
3335 + .sync_flags = ext3_sync_flags,
3338 const struct inode_operations ext3_special_inode_operations = {
3339 diff -NurpP --minimal linux-2.6.32.1/fs/ext3/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/super.c
3340 --- linux-2.6.32.1/fs/ext3/super.c 2009-12-03 20:02:51.000000000 +0100
3341 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext3/super.c 2009-12-03 20:04:56.000000000 +0100
3342 @@ -789,7 +789,7 @@ enum {
3343 Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
3344 Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
3345 Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
3347 + Opt_grpquota, Opt_tag, Opt_notag, Opt_tagid
3350 static const match_table_t tokens = {
3351 @@ -842,6 +842,9 @@ static const match_table_t tokens = {
3352 {Opt_usrquota, "usrquota"},
3353 {Opt_barrier, "barrier=%u"},
3354 {Opt_resize, "resize"},
3356 + {Opt_notag, "notag"},
3357 + {Opt_tagid, "tagid=%u"},
3361 @@ -934,6 +937,20 @@ static int parse_options (char *options,
3363 set_opt (sbi->s_mount_opt, NO_UID32);
3365 +#ifndef CONFIG_TAGGING_NONE
3367 + set_opt (sbi->s_mount_opt, TAGGED);
3370 + clear_opt (sbi->s_mount_opt, TAGGED);
3373 +#ifdef CONFIG_PROPAGATE
3376 + set_opt (sbi->s_mount_opt, TAGGED);
3380 clear_opt (sbi->s_mount_opt, CHECK);
3382 @@ -1658,6 +1675,9 @@ static int ext3_fill_super (struct super
3386 + if (EXT3_SB(sb)->s_mount_opt & EXT3_MOUNT_TAGGED)
3387 + sb->s_flags |= MS_TAGGED;
3389 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3390 ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3392 @@ -2527,6 +2547,14 @@ static int ext3_remount (struct super_bl
3393 if (sbi->s_mount_opt & EXT3_MOUNT_ABORT)
3394 ext3_abort(sb, __func__, "Abort forced by user");
3396 + if ((sbi->s_mount_opt & EXT3_MOUNT_TAGGED) &&
3397 + !(sb->s_flags & MS_TAGGED)) {
3398 + printk("EXT3-fs: %s: tagging not permitted on remount.\n",
3401 + goto restore_opts;
3404 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3405 ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3407 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/ext4.h linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ext4.h
3408 --- linux-2.6.32.1/fs/ext4/ext4.h 2009-12-14 21:29:46.000000000 +0100
3409 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ext4.h 2009-12-14 22:20:55.000000000 +0100
3410 @@ -284,8 +284,12 @@ struct flex_groups {
3411 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
3412 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
3413 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
3414 +#define EXT4_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
3415 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
3417 +#define EXT4_BARRIER_FL 0x04000000 /* Barrier for chroot() */
3418 +#define EXT4_COW_FL 0x20000000 /* Copy on Write marker */
3420 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
3421 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
3423 @@ -469,7 +473,8 @@ struct ext4_inode {
3424 __le16 l_i_file_acl_high;
3425 __le16 l_i_uid_high; /* these 2 fields */
3426 __le16 l_i_gid_high; /* were reserved2[0] */
3427 - __u32 l_i_reserved2;
3428 + __le16 l_i_tag; /* Context Tag */
3429 + __u16 l_i_reserved2;
3432 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
3433 @@ -583,6 +588,7 @@ do { \
3434 #define i_gid_low i_gid
3435 #define i_uid_high osd2.linux2.l_i_uid_high
3436 #define i_gid_high osd2.linux2.l_i_gid_high
3437 +#define i_raw_tag osd2.linux2.l_i_tag
3438 #define i_reserved2 osd2.linux2.l_i_reserved2
3440 #elif defined(__GNU__)
3441 @@ -751,6 +757,7 @@ struct ext4_inode_info {
3442 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
3443 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
3444 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
3445 +#define EXT4_MOUNT_TAGGED 0x400000 /* Enable Context Tags */
3446 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
3447 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
3448 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
3449 @@ -1742,6 +1749,7 @@ extern int ext4_get_blocks(handle_t *han
3450 struct buffer_head *bh, int flags);
3451 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3452 __u64 start, __u64 len);
3453 +extern int ext4_sync_flags(struct inode *, int, int);
3455 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3456 __u64 start_orig, __u64 start_donor,
3457 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/file.c
3458 --- linux-2.6.32.1/fs/ext4/file.c 2009-12-03 20:02:51.000000000 +0100
3459 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/file.c 2009-12-03 20:04:56.000000000 +0100
3460 @@ -161,5 +161,6 @@ const struct inode_operations ext4_file_
3461 .check_acl = ext4_check_acl,
3462 .fallocate = ext4_fallocate,
3463 .fiemap = ext4_fiemap,
3464 + .sync_flags = ext4_sync_flags,
3467 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/ialloc.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ialloc.c
3468 --- linux-2.6.32.1/fs/ext4/ialloc.c 2009-12-03 20:02:51.000000000 +0100
3469 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ialloc.c 2009-12-03 20:04:56.000000000 +0100
3471 #include <linux/random.h>
3472 #include <linux/bitops.h>
3473 #include <linux/blkdev.h>
3474 +#include <linux/vs_tag.h>
3475 #include <asm/byteorder.h>
3478 @@ -995,6 +996,7 @@ got:
3480 inode->i_gid = current_fsgid();
3481 inode->i_mode = mode;
3482 + inode->i_tag = dx_current_fstag(sb);
3484 inode->i_ino = ino + group * EXT4_INODES_PER_GROUP(sb);
3485 /* This is the optimal IO size (for stat), not the fs block size */
3486 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/inode.c
3487 --- linux-2.6.32.1/fs/ext4/inode.c 2009-12-14 21:29:46.000000000 +0100
3488 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/inode.c 2009-12-14 22:20:55.000000000 +0100
3490 #include <linux/uio.h>
3491 #include <linux/bio.h>
3492 #include <linux/workqueue.h>
3493 +#include <linux/vs_tag.h>
3495 #include "ext4_jbd2.h"
3497 @@ -4384,7 +4385,7 @@ static void ext4_free_branches(handle_t
3499 int ext4_can_truncate(struct inode *inode)
3501 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3502 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3504 if (S_ISREG(inode->i_mode))
3506 @@ -4735,36 +4736,60 @@ void ext4_set_inode_flags(struct inode *
3508 unsigned int flags = EXT4_I(inode)->i_flags;
3510 - inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3511 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3512 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3514 + if (flags & EXT4_IMMUTABLE_FL)
3515 + inode->i_flags |= S_IMMUTABLE;
3516 + if (flags & EXT4_IXUNLINK_FL)
3517 + inode->i_flags |= S_IXUNLINK;
3519 if (flags & EXT4_SYNC_FL)
3520 inode->i_flags |= S_SYNC;
3521 if (flags & EXT4_APPEND_FL)
3522 inode->i_flags |= S_APPEND;
3523 - if (flags & EXT4_IMMUTABLE_FL)
3524 - inode->i_flags |= S_IMMUTABLE;
3525 if (flags & EXT4_NOATIME_FL)
3526 inode->i_flags |= S_NOATIME;
3527 if (flags & EXT4_DIRSYNC_FL)
3528 inode->i_flags |= S_DIRSYNC;
3530 + inode->i_vflags &= ~(V_BARRIER | V_COW);
3532 + if (flags & EXT4_BARRIER_FL)
3533 + inode->i_vflags |= V_BARRIER;
3534 + if (flags & EXT4_COW_FL)
3535 + inode->i_vflags |= V_COW;
3538 /* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
3539 void ext4_get_inode_flags(struct ext4_inode_info *ei)
3541 unsigned int flags = ei->vfs_inode.i_flags;
3542 + unsigned int vflags = ei->vfs_inode.i_vflags;
3544 + ei->i_flags &= ~(EXT4_SYNC_FL | EXT4_APPEND_FL |
3545 + EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL |
3546 + EXT4_NOATIME_FL | EXT4_DIRSYNC_FL |
3547 + EXT4_BARRIER_FL | EXT4_COW_FL);
3549 + if (flags & S_IMMUTABLE)
3550 + ei->i_flags |= EXT4_IMMUTABLE_FL;
3551 + if (flags & S_IXUNLINK)
3552 + ei->i_flags |= EXT4_IXUNLINK_FL;
3554 - ei->i_flags &= ~(EXT4_SYNC_FL|EXT4_APPEND_FL|
3555 - EXT4_IMMUTABLE_FL|EXT4_NOATIME_FL|EXT4_DIRSYNC_FL);
3557 ei->i_flags |= EXT4_SYNC_FL;
3558 if (flags & S_APPEND)
3559 ei->i_flags |= EXT4_APPEND_FL;
3560 - if (flags & S_IMMUTABLE)
3561 - ei->i_flags |= EXT4_IMMUTABLE_FL;
3562 if (flags & S_NOATIME)
3563 ei->i_flags |= EXT4_NOATIME_FL;
3564 if (flags & S_DIRSYNC)
3565 ei->i_flags |= EXT4_DIRSYNC_FL;
3567 + if (vflags & V_BARRIER)
3568 + ei->i_flags |= EXT4_BARRIER_FL;
3569 + if (vflags & V_COW)
3570 + ei->i_flags |= EXT4_COW_FL;
3573 static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
3574 @@ -4799,6 +4824,8 @@ struct inode *ext4_iget(struct super_blo
3575 journal_t *journal = EXT4_SB(sb)->s_journal;
3581 inode = iget_locked(sb, ino);
3583 @@ -4814,12 +4841,16 @@ struct inode *ext4_iget(struct super_blo
3585 raw_inode = ext4_raw_inode(&iloc);
3586 inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3587 - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3588 - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3589 + uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3590 + gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3591 if (!(test_opt(inode->i_sb, NO_UID32))) {
3592 - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3593 - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3594 + uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3595 + gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3597 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3598 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3599 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3600 + le16_to_cpu(raw_inode->i_raw_tag));
3601 inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3604 @@ -5038,6 +5069,8 @@ static int ext4_do_update_inode(handle_t
3605 struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
3606 struct ext4_inode_info *ei = EXT4_I(inode);
3607 struct buffer_head *bh = iloc->bh;
3608 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3609 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3610 int err = 0, rc, block;
3612 /* For fields not not tracking in the in-memory inode,
3613 @@ -5048,29 +5081,32 @@ static int ext4_do_update_inode(handle_t
3614 ext4_get_inode_flags(ei);
3615 raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3616 if (!(test_opt(inode->i_sb, NO_UID32))) {
3617 - raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3618 - raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3619 + raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3620 + raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3622 * Fix up interoperability with old kernels. Otherwise, old inodes get
3623 * re-used with the upper 16 bits of the uid/gid intact
3626 raw_inode->i_uid_high =
3627 - cpu_to_le16(high_16_bits(inode->i_uid));
3628 + cpu_to_le16(high_16_bits(uid));
3629 raw_inode->i_gid_high =
3630 - cpu_to_le16(high_16_bits(inode->i_gid));
3631 + cpu_to_le16(high_16_bits(gid));
3633 raw_inode->i_uid_high = 0;
3634 raw_inode->i_gid_high = 0;
3637 raw_inode->i_uid_low =
3638 - cpu_to_le16(fs_high2lowuid(inode->i_uid));
3639 + cpu_to_le16(fs_high2lowuid(uid));
3640 raw_inode->i_gid_low =
3641 - cpu_to_le16(fs_high2lowgid(inode->i_gid));
3642 + cpu_to_le16(fs_high2lowgid(gid));
3643 raw_inode->i_uid_high = 0;
3644 raw_inode->i_gid_high = 0;
3646 +#ifdef CONFIG_TAGGING_INTERN
3647 + raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3649 raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3651 EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
3652 @@ -5255,7 +5291,8 @@ int ext4_setattr(struct dentry *dentry,
3655 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3656 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3657 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3658 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3661 /* (user+group)*(old+new) structure, inode write (sb,
3662 @@ -5277,6 +5314,8 @@ int ext4_setattr(struct dentry *dentry,
3663 inode->i_uid = attr->ia_uid;
3664 if (attr->ia_valid & ATTR_GID)
3665 inode->i_gid = attr->ia_gid;
3666 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3667 + inode->i_tag = attr->ia_tag;
3668 error = ext4_mark_inode_dirty(handle, inode);
3669 ext4_journal_stop(handle);
3671 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ioctl.c
3672 --- linux-2.6.32.1/fs/ext4/ioctl.c 2009-12-14 21:29:46.000000000 +0100
3673 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/ioctl.c 2009-12-14 22:20:55.000000000 +0100
3675 #include <linux/compat.h>
3676 #include <linux/mount.h>
3677 #include <linux/file.h>
3678 +#include <linux/vs_tag.h>
3679 #include <asm/uaccess.h>
3680 #include "ext4_jbd2.h"
3684 +int ext4_sync_flags(struct inode *inode, int flags, int vflags)
3686 + handle_t *handle = NULL;
3687 + struct ext4_iloc iloc;
3690 + handle = ext4_journal_start(inode, 1);
3691 + if (IS_ERR(handle))
3692 + return PTR_ERR(handle);
3694 + if (IS_SYNC(inode))
3695 + ext4_handle_sync(handle);
3696 + err = ext4_reserve_inode_write(handle, inode, &iloc);
3700 + inode->i_flags = flags;
3701 + inode->i_vflags = vflags;
3702 + ext4_get_inode_flags(EXT4_I(inode));
3703 + inode->i_ctime = ext4_current_time(inode);
3705 + err = ext4_mark_iloc_dirty(handle, inode, &iloc);
3707 + ext4_journal_stop(handle);
3711 long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
3713 struct inode *inode = filp->f_dentry->d_inode;
3714 @@ -50,6 +79,11 @@ long ext4_ioctl(struct file *filp, unsig
3716 flags = ext4_mask_flags(inode->i_mode, flags);
3718 + if (IS_BARRIER(inode)) {
3719 + vxwprintk_task(1, "messing with the barrier.");
3724 mutex_lock(&inode->i_mutex);
3725 /* Is it quota file? Do not allow user to mess with it */
3726 @@ -67,7 +101,9 @@ long ext4_ioctl(struct file *filp, unsig
3728 * This test looks nicer. Thanks to Pauline Middelink
3730 - if ((flags ^ oldflags) & (EXT4_APPEND_FL | EXT4_IMMUTABLE_FL)) {
3731 + if ((oldflags & EXT4_IMMUTABLE_FL) ||
3732 + ((flags ^ oldflags) & (EXT4_APPEND_FL |
3733 + EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL))) {
3734 if (!capable(CAP_LINUX_IMMUTABLE))
3737 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/namei.c
3738 --- linux-2.6.32.1/fs/ext4/namei.c 2009-12-14 21:29:46.000000000 +0100
3739 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/namei.c 2009-12-14 22:20:55.000000000 +0100
3741 #include <linux/quotaops.h>
3742 #include <linux/buffer_head.h>
3743 #include <linux/bio.h>
3744 +#include <linux/vs_tag.h>
3746 #include "ext4_jbd2.h"
3748 @@ -941,6 +942,7 @@ restart:
3750 ll_rw_block(READ_META, 1, &bh);
3752 + dx_propagate_tag(nd, inode);
3754 if ((bh = bh_use[ra_ptr++]) == NULL)
3756 @@ -2533,6 +2535,7 @@ const struct inode_operations ext4_dir_i
3758 .check_acl = ext4_check_acl,
3759 .fiemap = ext4_fiemap,
3760 + .sync_flags = ext4_sync_flags,
3763 const struct inode_operations ext4_special_inode_operations = {
3764 diff -NurpP --minimal linux-2.6.32.1/fs/ext4/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/super.c
3765 --- linux-2.6.32.1/fs/ext4/super.c 2009-12-14 21:29:46.000000000 +0100
3766 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ext4/super.c 2009-12-14 22:22:03.000000000 +0100
3767 @@ -1091,6 +1091,7 @@ enum {
3768 Opt_block_validity, Opt_noblock_validity,
3769 Opt_inode_readahead_blks, Opt_journal_ioprio,
3770 Opt_discard, Opt_nodiscard,
3771 + Opt_tag, Opt_notag, Opt_tagid
3774 static const match_table_t tokens = {
3775 @@ -1158,6 +1159,9 @@ static const match_table_t tokens = {
3776 {Opt_noauto_da_alloc, "noauto_da_alloc"},
3777 {Opt_discard, "discard"},
3778 {Opt_nodiscard, "nodiscard"},
3780 + {Opt_notag, "notag"},
3781 + {Opt_tagid, "tagid=%u"},
3785 @@ -1255,6 +1259,20 @@ static int parse_options(char *options,
3787 set_opt(sbi->s_mount_opt, NO_UID32);
3789 +#ifndef CONFIG_TAGGING_NONE
3791 + set_opt (sbi->s_mount_opt, TAGGED);
3794 + clear_opt (sbi->s_mount_opt, TAGGED);
3797 +#ifdef CONFIG_PROPAGATE
3800 + set_opt (sbi->s_mount_opt, TAGGED);
3804 set_opt(sbi->s_mount_opt, DEBUG);
3806 @@ -2455,6 +2473,9 @@ static int ext4_fill_super(struct super_
3807 &journal_ioprio, NULL, 0))
3810 + if (EXT4_SB(sb)->s_mount_opt & EXT4_MOUNT_TAGGED)
3811 + sb->s_flags |= MS_TAGGED;
3813 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3814 ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3816 @@ -3512,6 +3533,14 @@ static int ext4_remount(struct super_blo
3817 if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED)
3818 ext4_abort(sb, __func__, "Abort forced by user");
3820 + if ((sbi->s_mount_opt & EXT4_MOUNT_TAGGED) &&
3821 + !(sb->s_flags & MS_TAGGED)) {
3822 + printk("EXT4-fs: %s: tagging not permitted on remount.\n",
3825 + goto restore_opts;
3828 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3829 ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3831 diff -NurpP --minimal linux-2.6.32.1/fs/fcntl.c linux-2.6.32.1-vs2.3.0.36.27/fs/fcntl.c
3832 --- linux-2.6.32.1/fs/fcntl.c 2009-12-03 20:02:51.000000000 +0100
3833 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/fcntl.c 2009-12-03 20:04:56.000000000 +0100
3835 #include <linux/signal.h>
3836 #include <linux/rcupdate.h>
3837 #include <linux/pid_namespace.h>
3838 +#include <linux/vs_limit.h>
3840 #include <asm/poll.h>
3841 #include <asm/siginfo.h>
3842 @@ -102,6 +103,8 @@ SYSCALL_DEFINE3(dup3, unsigned int, oldf
3845 filp_close(tofree, files);
3847 + vx_openfd_inc(newfd); /* fd was unused */
3851 @@ -426,6 +429,8 @@ SYSCALL_DEFINE3(fcntl, unsigned int, fd,
3855 + if (!vx_files_avail(1))
3858 err = security_file_fcntl(filp, cmd, arg);
3860 diff -NurpP --minimal linux-2.6.32.1/fs/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/file.c
3861 --- linux-2.6.32.1/fs/file.c 2009-12-03 20:02:51.000000000 +0100
3862 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/file.c 2009-12-03 20:04:56.000000000 +0100
3864 #include <linux/spinlock.h>
3865 #include <linux/rcupdate.h>
3866 #include <linux/workqueue.h>
3867 +#include <linux/vs_limit.h>
3869 struct fdtable_defer {
3871 @@ -368,6 +369,8 @@ struct files_struct *dup_fd(struct files
3872 struct file *f = *old_fds++;
3875 + /* TODO: sum it first for check and performance */
3876 + vx_openfd_inc(open_files - i);
3879 * The fd may be claimed in the fd bitmap but not yet
3880 @@ -476,6 +479,7 @@ repeat:
3882 FD_CLR(fd, fdt->close_on_exec);
3884 + vx_openfd_inc(fd);
3887 if (rcu_dereference(fdt->fd[fd]) != NULL) {
3888 diff -NurpP --minimal linux-2.6.32.1/fs/file_table.c linux-2.6.32.1-vs2.3.0.36.27/fs/file_table.c
3889 --- linux-2.6.32.1/fs/file_table.c 2009-12-03 20:02:51.000000000 +0100
3890 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/file_table.c 2009-12-03 20:04:56.000000000 +0100
3892 #include <linux/fsnotify.h>
3893 #include <linux/sysctl.h>
3894 #include <linux/percpu_counter.h>
3895 +#include <linux/vs_limit.h>
3896 +#include <linux/vs_context.h>
3898 #include <asm/atomic.h>
3900 @@ -131,6 +133,8 @@ struct file *get_empty_filp(void)
3901 spin_lock_init(&f->f_lock);
3902 eventpoll_init_file(f);
3903 /* f->f_version: 0 */
3904 + f->f_xid = vx_current_xid();
3909 @@ -285,6 +289,8 @@ void __fput(struct file *file)
3910 cdev_put(inode->i_cdev);
3911 fops_put(file->f_op);
3912 put_pid(file->f_owner.pid);
3913 + vx_files_dec(file);
3916 if (file->f_mode & FMODE_WRITE)
3917 drop_file_write_access(file);
3918 @@ -352,6 +358,8 @@ void put_filp(struct file *file)
3920 if (atomic_long_dec_and_test(&file->f_count)) {
3921 security_file_free(file);
3922 + vx_files_dec(file);
3927 diff -NurpP --minimal linux-2.6.32.1/fs/fs_struct.c linux-2.6.32.1-vs2.3.0.36.27/fs/fs_struct.c
3928 --- linux-2.6.32.1/fs/fs_struct.c 2009-06-11 17:13:04.000000000 +0200
3929 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/fs_struct.c 2009-12-03 20:04:56.000000000 +0100
3931 #include <linux/path.h>
3932 #include <linux/slab.h>
3933 #include <linux/fs_struct.h>
3934 +#include <linux/vserver/global.h>
3937 * Replace the fs->{rootmnt,root} with {mnt,dentry}. Put the old values.
3938 @@ -77,6 +78,7 @@ void free_fs_struct(struct fs_struct *fs
3940 path_put(&fs->root);
3942 + atomic_dec(&vs_global_fs);
3943 kmem_cache_free(fs_cachep, fs);
3946 @@ -112,6 +114,7 @@ struct fs_struct *copy_fs_struct(struct
3948 path_get(&old->pwd);
3949 read_unlock(&old->lock);
3950 + atomic_inc(&vs_global_fs);
3954 diff -NurpP --minimal linux-2.6.32.1/fs/gfs2/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/file.c
3955 --- linux-2.6.32.1/fs/gfs2/file.c 2009-12-03 20:02:52.000000000 +0100
3956 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/file.c 2009-12-03 20:04:56.000000000 +0100
3957 @@ -132,6 +132,9 @@ static const u32 fsflags_to_gfs2[32] = {
3958 [7] = GFS2_DIF_NOATIME,
3959 [12] = GFS2_DIF_EXHASH,
3960 [14] = GFS2_DIF_INHERIT_JDATA,
3961 + [27] = GFS2_DIF_IXUNLINK,
3962 + [26] = GFS2_DIF_BARRIER,
3963 + [29] = GFS2_DIF_COW,
3966 static const u32 gfs2_to_fsflags[32] = {
3967 @@ -141,6 +144,9 @@ static const u32 gfs2_to_fsflags[32] = {
3968 [gfs2fl_NoAtime] = FS_NOATIME_FL,
3969 [gfs2fl_ExHash] = FS_INDEX_FL,
3970 [gfs2fl_InheritJdata] = FS_JOURNAL_DATA_FL,
3971 + [gfs2fl_IXUnlink] = FS_IXUNLINK_FL,
3972 + [gfs2fl_Barrier] = FS_BARRIER_FL,
3973 + [gfs2fl_Cow] = FS_COW_FL,
3976 static int gfs2_get_flags(struct file *filp, u32 __user *ptr)
3977 @@ -171,10 +177,16 @@ void gfs2_set_inode_flags(struct inode *
3979 struct gfs2_inode *ip = GFS2_I(inode);
3980 unsigned int flags = inode->i_flags;
3981 + unsigned int vflags = inode->i_vflags;
3983 + flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3984 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3986 - flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3987 if (ip->i_diskflags & GFS2_DIF_IMMUTABLE)
3988 flags |= S_IMMUTABLE;
3989 + if (ip->i_diskflags & GFS2_DIF_IXUNLINK)
3990 + flags |= S_IXUNLINK;
3992 if (ip->i_diskflags & GFS2_DIF_APPENDONLY)
3994 if (ip->i_diskflags & GFS2_DIF_NOATIME)
3995 @@ -182,6 +194,43 @@ void gfs2_set_inode_flags(struct inode *
3996 if (ip->i_diskflags & GFS2_DIF_SYNC)
3998 inode->i_flags = flags;
4000 + vflags &= ~(V_BARRIER | V_COW);
4002 + if (ip->i_diskflags & GFS2_DIF_BARRIER)
4003 + vflags |= V_BARRIER;
4004 + if (ip->i_diskflags & GFS2_DIF_COW)
4006 + inode->i_vflags = vflags;
4009 +void gfs2_get_inode_flags(struct inode *inode)
4011 + struct gfs2_inode *ip = GFS2_I(inode);
4012 + unsigned int flags = inode->i_flags;
4013 + unsigned int vflags = inode->i_vflags;
4015 + ip->i_diskflags &= ~(GFS2_DIF_APPENDONLY |
4016 + GFS2_DIF_NOATIME | GFS2_DIF_SYNC |
4017 + GFS2_DIF_IMMUTABLE | GFS2_DIF_IXUNLINK |
4018 + GFS2_DIF_BARRIER | GFS2_DIF_COW);
4020 + if (flags & S_IMMUTABLE)
4021 + ip->i_diskflags |= GFS2_DIF_IMMUTABLE;
4022 + if (flags & S_IXUNLINK)
4023 + ip->i_diskflags |= GFS2_DIF_IXUNLINK;
4025 + if (flags & S_APPEND)
4026 + ip->i_diskflags |= GFS2_DIF_APPENDONLY;
4027 + if (flags & S_NOATIME)
4028 + ip->i_diskflags |= GFS2_DIF_NOATIME;
4029 + if (flags & S_SYNC)
4030 + ip->i_diskflags |= GFS2_DIF_SYNC;
4032 + if (vflags & V_BARRIER)
4033 + ip->i_diskflags |= GFS2_DIF_BARRIER;
4034 + if (vflags & V_COW)
4035 + ip->i_diskflags |= GFS2_DIF_COW;
4038 /* Flags that can be set by user space */
4039 @@ -286,6 +335,37 @@ static int gfs2_set_flags(struct file *f
4040 return do_gfs2_set_flags(filp, gfsflags, ~GFS2_DIF_JDATA);
4043 +int gfs2_sync_flags(struct inode *inode, int flags, int vflags)
4045 + struct gfs2_inode *ip = GFS2_I(inode);
4046 + struct gfs2_sbd *sdp = GFS2_SB(inode);
4047 + struct buffer_head *bh;
4048 + struct gfs2_holder gh;
4051 + error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
4054 + error = gfs2_trans_begin(sdp, RES_DINODE, 0);
4057 + error = gfs2_meta_inode_buffer(ip, &bh);
4059 + goto out_trans_end;
4060 + gfs2_trans_add_bh(ip->i_gl, bh, 1);
4061 + inode->i_flags = flags;
4062 + inode->i_vflags = vflags;
4063 + gfs2_get_inode_flags(inode);
4064 + gfs2_dinode_out(ip, bh->b_data);
4066 + gfs2_set_aops(inode);
4068 + gfs2_trans_end(sdp);
4070 + gfs2_glock_dq_uninit(&gh);
4074 static long gfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
4077 diff -NurpP --minimal linux-2.6.32.1/fs/gfs2/inode.h linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/inode.h
4078 --- linux-2.6.32.1/fs/gfs2/inode.h 2009-09-10 15:26:22.000000000 +0200
4079 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/inode.h 2009-12-03 20:04:56.000000000 +0100
4080 @@ -109,6 +109,7 @@ extern const struct file_operations gfs2
4081 extern const struct file_operations gfs2_dir_fops_nolock;
4083 extern void gfs2_set_inode_flags(struct inode *inode);
4084 +extern int gfs2_sync_flags(struct inode *inode, int flags, int vflags);
4086 #ifdef CONFIG_GFS2_FS_LOCKING_DLM
4087 extern const struct file_operations gfs2_file_fops;
4088 diff -NurpP --minimal linux-2.6.32.1/fs/gfs2/ops_inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/ops_inode.c
4089 --- linux-2.6.32.1/fs/gfs2/ops_inode.c 2009-12-03 20:02:52.000000000 +0100
4090 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/gfs2/ops_inode.c 2009-12-03 20:04:56.000000000 +0100
4091 @@ -1400,6 +1400,7 @@ const struct inode_operations gfs2_file_
4092 .listxattr = gfs2_listxattr,
4093 .removexattr = gfs2_removexattr,
4094 .fiemap = gfs2_fiemap,
4095 + .sync_flags = gfs2_sync_flags,
4098 const struct inode_operations gfs2_dir_iops = {
4099 @@ -1420,6 +1421,7 @@ const struct inode_operations gfs2_dir_i
4100 .listxattr = gfs2_listxattr,
4101 .removexattr = gfs2_removexattr,
4102 .fiemap = gfs2_fiemap,
4103 + .sync_flags = gfs2_sync_flags,
4106 const struct inode_operations gfs2_symlink_iops = {
4107 diff -NurpP --minimal linux-2.6.32.1/fs/hfsplus/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/hfsplus/ioctl.c
4108 --- linux-2.6.32.1/fs/hfsplus/ioctl.c 2008-12-25 00:26:37.000000000 +0100
4109 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/hfsplus/ioctl.c 2009-12-03 20:04:56.000000000 +0100
4111 #include <linux/mount.h>
4112 #include <linux/sched.h>
4113 #include <linux/xattr.h>
4114 +#include <linux/mount.h>
4115 #include <asm/uaccess.h>
4116 #include "hfsplus_fs.h"
4118 diff -NurpP --minimal linux-2.6.32.1/fs/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/inode.c
4119 --- linux-2.6.32.1/fs/inode.c 2009-12-03 20:02:52.000000000 +0100
4120 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/inode.c 2009-12-03 20:04:56.000000000 +0100
4121 @@ -133,6 +133,9 @@ int inode_init_always(struct super_block
4122 struct address_space *const mapping = &inode->i_data;
4126 + /* essential because of inode slab reuse */
4128 inode->i_blkbits = sb->s_blocksize_bits;
4130 atomic_set(&inode->i_count, 1);
4131 @@ -153,6 +156,7 @@ int inode_init_always(struct super_block
4132 inode->i_bdev = NULL;
4133 inode->i_cdev = NULL;
4135 + inode->i_mdev = 0;
4136 inode->dirtied_when = 0;
4138 if (security_inode_alloc(inode))
4139 @@ -307,6 +311,8 @@ void __iget(struct inode *inode)
4140 inodes_stat.nr_unused--;
4143 +EXPORT_SYMBOL_GPL(__iget);
4146 * clear_inode - clear an inode
4147 * @inode: inode to clear
4148 @@ -1611,9 +1617,11 @@ void init_special_inode(struct inode *in
4149 if (S_ISCHR(mode)) {
4150 inode->i_fop = &def_chr_fops;
4151 inode->i_rdev = rdev;
4152 + inode->i_mdev = rdev;
4153 } else if (S_ISBLK(mode)) {
4154 inode->i_fop = &def_blk_fops;
4155 inode->i_rdev = rdev;
4156 + inode->i_mdev = rdev;
4157 } else if (S_ISFIFO(mode))
4158 inode->i_fop = &def_fifo_fops;
4159 else if (S_ISSOCK(mode))
4160 diff -NurpP --minimal linux-2.6.32.1/fs/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/ioctl.c
4161 --- linux-2.6.32.1/fs/ioctl.c 2009-12-03 20:02:52.000000000 +0100
4162 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ioctl.c 2009-12-03 20:04:56.000000000 +0100
4164 #include <linux/writeback.h>
4165 #include <linux/buffer_head.h>
4166 #include <linux/falloc.h>
4167 +#include <linux/proc_fs.h>
4168 +#include <linux/vserver/inode.h>
4169 +#include <linux/vs_tag.h>
4171 #include <asm/ioctls.h>
4173 diff -NurpP --minimal linux-2.6.32.1/fs/ioprio.c linux-2.6.32.1-vs2.3.0.36.27/fs/ioprio.c
4174 --- linux-2.6.32.1/fs/ioprio.c 2009-03-24 14:22:26.000000000 +0100
4175 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ioprio.c 2009-12-03 20:04:56.000000000 +0100
4177 #include <linux/syscalls.h>
4178 #include <linux/security.h>
4179 #include <linux/pid_namespace.h>
4180 +#include <linux/vs_base.h>
4182 int set_task_ioprio(struct task_struct *task, int ioprio)
4184 @@ -123,6 +124,8 @@ SYSCALL_DEFINE3(ioprio_set, int, which,
4186 pgrp = find_vpid(who);
4187 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
4188 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
4190 ret = set_task_ioprio(p, ioprio);
4193 @@ -212,6 +215,8 @@ SYSCALL_DEFINE2(ioprio_get, int, which,
4195 pgrp = find_vpid(who);
4196 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
4197 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
4199 tmpio = get_task_ioprio(p);
4202 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/acl.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/acl.c
4203 --- linux-2.6.32.1/fs/jfs/acl.c 2009-12-03 20:02:52.000000000 +0100
4204 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/acl.c 2009-12-03 20:04:56.000000000 +0100
4205 @@ -216,7 +216,8 @@ int jfs_setattr(struct dentry *dentry, s
4208 if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
4209 - (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
4210 + (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
4211 + (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
4212 if (vfs_dq_transfer(inode, iattr))
4215 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/file.c
4216 --- linux-2.6.32.1/fs/jfs/file.c 2009-12-03 20:02:52.000000000 +0100
4217 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/file.c 2009-12-03 20:04:56.000000000 +0100
4218 @@ -98,6 +98,7 @@ const struct inode_operations jfs_file_i
4219 .setattr = jfs_setattr,
4220 .check_acl = jfs_check_acl,
4222 + .sync_flags = jfs_sync_flags,
4225 const struct file_operations jfs_file_operations = {
4226 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/ioctl.c
4227 --- linux-2.6.32.1/fs/jfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
4228 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/ioctl.c 2009-12-03 20:04:56.000000000 +0100
4230 #include <linux/mount.h>
4231 #include <linux/time.h>
4232 #include <linux/sched.h>
4233 +#include <linux/mount.h>
4234 #include <asm/current.h>
4235 #include <asm/uaccess.h>
4237 @@ -52,6 +53,16 @@ static long jfs_map_ext2(unsigned long f
4241 +int jfs_sync_flags(struct inode *inode, int flags, int vflags)
4243 + inode->i_flags = flags;
4244 + inode->i_vflags = vflags;
4245 + jfs_get_inode_flags(JFS_IP(inode));
4246 + inode->i_ctime = CURRENT_TIME_SEC;
4247 + mark_inode_dirty(inode);
4251 long jfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
4253 struct inode *inode = filp->f_dentry->d_inode;
4254 @@ -85,6 +96,11 @@ long jfs_ioctl(struct file *filp, unsign
4255 if (!S_ISDIR(inode->i_mode))
4256 flags &= ~JFS_DIRSYNC_FL;
4258 + if (IS_BARRIER(inode)) {
4259 + vxwprintk_task(1, "messing with the barrier.");
4263 /* Is it quota file? Do not allow user to mess with it */
4264 if (IS_NOQUOTA(inode)) {
4266 @@ -102,8 +118,8 @@ long jfs_ioctl(struct file *filp, unsign
4267 * the relevant capability.
4269 if ((oldflags & JFS_IMMUTABLE_FL) ||
4270 - ((flags ^ oldflags) &
4271 - (JFS_APPEND_FL | JFS_IMMUTABLE_FL))) {
4272 + ((flags ^ oldflags) & (JFS_APPEND_FL |
4273 + JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL))) {
4274 if (!capable(CAP_LINUX_IMMUTABLE)) {
4275 mutex_unlock(&inode->i_mutex);
4277 @@ -111,7 +127,7 @@ long jfs_ioctl(struct file *filp, unsign
4281 - flags = flags & JFS_FL_USER_MODIFIABLE;
4282 + flags &= JFS_FL_USER_MODIFIABLE;
4283 flags |= oldflags & ~JFS_FL_USER_MODIFIABLE;
4284 jfs_inode->mode2 = flags;
4286 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/jfs_dinode.h linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_dinode.h
4287 --- linux-2.6.32.1/fs/jfs/jfs_dinode.h 2008-12-25 00:26:37.000000000 +0100
4288 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_dinode.h 2009-12-03 20:04:56.000000000 +0100
4289 @@ -161,9 +161,13 @@ struct dinode {
4291 #define JFS_APPEND_FL 0x01000000 /* writes to file may only append */
4292 #define JFS_IMMUTABLE_FL 0x02000000 /* Immutable file */
4293 +#define JFS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
4295 -#define JFS_FL_USER_VISIBLE 0x03F80000
4296 -#define JFS_FL_USER_MODIFIABLE 0x03F80000
4297 +#define JFS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
4298 +#define JFS_COW_FL 0x20000000 /* Copy on Write marker */
4300 +#define JFS_FL_USER_VISIBLE 0x07F80000
4301 +#define JFS_FL_USER_MODIFIABLE 0x07F80000
4302 #define JFS_FL_INHERIT 0x03C80000
4304 /* These are identical to EXT[23]_IOC_GETFLAGS/SETFLAGS */
4305 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/jfs_filsys.h linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_filsys.h
4306 --- linux-2.6.32.1/fs/jfs/jfs_filsys.h 2008-12-25 00:26:37.000000000 +0100
4307 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_filsys.h 2009-12-03 20:04:56.000000000 +0100
4309 #define JFS_NAME_MAX 255
4310 #define JFS_PATH_MAX BPSIZE
4312 +#define JFS_TAGGED 0x00800000 /* Context Tagging */
4315 * file system state (superblock state)
4316 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/jfs_imap.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_imap.c
4317 --- linux-2.6.32.1/fs/jfs/jfs_imap.c 2009-09-10 15:26:22.000000000 +0200
4318 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_imap.c 2009-12-03 20:04:56.000000000 +0100
4320 #include <linux/buffer_head.h>
4321 #include <linux/pagemap.h>
4322 #include <linux/quotaops.h>
4323 +#include <linux/vs_tag.h>
4325 #include "jfs_incore.h"
4326 #include "jfs_inode.h"
4327 @@ -3059,6 +3060,8 @@ static int copy_from_dinode(struct dinod
4329 struct jfs_inode_info *jfs_ip = JFS_IP(ip);
4330 struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
4334 jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
4335 jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
4336 @@ -3079,14 +3082,18 @@ static int copy_from_dinode(struct dinod
4338 ip->i_nlink = le32_to_cpu(dip->di_nlink);
4340 - jfs_ip->saved_uid = le32_to_cpu(dip->di_uid);
4341 + uid = le32_to_cpu(dip->di_uid);
4342 + gid = le32_to_cpu(dip->di_gid);
4343 + ip->i_tag = INOTAG_TAG(DX_TAG(ip), uid, gid, 0);
4345 + jfs_ip->saved_uid = INOTAG_UID(DX_TAG(ip), uid, gid);
4347 ip->i_uid = jfs_ip->saved_uid;
4349 ip->i_uid = sbi->uid;
4352 - jfs_ip->saved_gid = le32_to_cpu(dip->di_gid);
4353 + jfs_ip->saved_gid = INOTAG_GID(DX_TAG(ip), uid, gid);
4355 ip->i_gid = jfs_ip->saved_gid;
4357 @@ -3151,14 +3158,12 @@ static void copy_to_dinode(struct dinode
4358 dip->di_size = cpu_to_le64(ip->i_size);
4359 dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
4360 dip->di_nlink = cpu_to_le32(ip->i_nlink);
4361 - if (sbi->uid == -1)
4362 - dip->di_uid = cpu_to_le32(ip->i_uid);
4364 - dip->di_uid = cpu_to_le32(jfs_ip->saved_uid);
4365 - if (sbi->gid == -1)
4366 - dip->di_gid = cpu_to_le32(ip->i_gid);
4368 - dip->di_gid = cpu_to_le32(jfs_ip->saved_gid);
4370 + dip->di_uid = cpu_to_le32(TAGINO_UID(DX_TAG(ip),
4371 + (sbi->uid == -1) ? ip->i_uid : jfs_ip->saved_uid, ip->i_tag));
4372 + dip->di_gid = cpu_to_le32(TAGINO_GID(DX_TAG(ip),
4373 + (sbi->gid == -1) ? ip->i_gid : jfs_ip->saved_gid, ip->i_tag));
4375 jfs_get_inode_flags(jfs_ip);
4377 * mode2 is only needed for storing the higher order bits.
4378 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/jfs_inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_inode.c
4379 --- linux-2.6.32.1/fs/jfs/jfs_inode.c 2009-06-11 17:13:05.000000000 +0200
4380 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_inode.c 2009-12-03 20:04:56.000000000 +0100
4383 #include <linux/fs.h>
4384 #include <linux/quotaops.h>
4385 +#include <linux/vs_tag.h>
4386 #include "jfs_incore.h"
4387 #include "jfs_inode.h"
4388 #include "jfs_filsys.h"
4389 @@ -30,29 +31,46 @@ void jfs_set_inode_flags(struct inode *i
4391 unsigned int flags = JFS_IP(inode)->mode2;
4393 - inode->i_flags &= ~(S_IMMUTABLE | S_APPEND |
4394 - S_NOATIME | S_DIRSYNC | S_SYNC);
4395 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
4396 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
4398 if (flags & JFS_IMMUTABLE_FL)
4399 inode->i_flags |= S_IMMUTABLE;
4400 + if (flags & JFS_IXUNLINK_FL)
4401 + inode->i_flags |= S_IXUNLINK;
4403 + if (flags & JFS_SYNC_FL)
4404 + inode->i_flags |= S_SYNC;
4405 if (flags & JFS_APPEND_FL)
4406 inode->i_flags |= S_APPEND;
4407 if (flags & JFS_NOATIME_FL)
4408 inode->i_flags |= S_NOATIME;
4409 if (flags & JFS_DIRSYNC_FL)
4410 inode->i_flags |= S_DIRSYNC;
4411 - if (flags & JFS_SYNC_FL)
4412 - inode->i_flags |= S_SYNC;
4414 + inode->i_vflags &= ~(V_BARRIER | V_COW);
4416 + if (flags & JFS_BARRIER_FL)
4417 + inode->i_vflags |= V_BARRIER;
4418 + if (flags & JFS_COW_FL)
4419 + inode->i_vflags |= V_COW;
4422 void jfs_get_inode_flags(struct jfs_inode_info *jfs_ip)
4424 unsigned int flags = jfs_ip->vfs_inode.i_flags;
4425 + unsigned int vflags = jfs_ip->vfs_inode.i_vflags;
4427 + jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL |
4428 + JFS_APPEND_FL | JFS_NOATIME_FL |
4429 + JFS_DIRSYNC_FL | JFS_SYNC_FL |
4430 + JFS_BARRIER_FL | JFS_COW_FL);
4432 - jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_APPEND_FL | JFS_NOATIME_FL |
4433 - JFS_DIRSYNC_FL | JFS_SYNC_FL);
4434 if (flags & S_IMMUTABLE)
4435 jfs_ip->mode2 |= JFS_IMMUTABLE_FL;
4436 + if (flags & S_IXUNLINK)
4437 + jfs_ip->mode2 |= JFS_IXUNLINK_FL;
4439 if (flags & S_APPEND)
4440 jfs_ip->mode2 |= JFS_APPEND_FL;
4441 if (flags & S_NOATIME)
4442 @@ -61,6 +79,11 @@ void jfs_get_inode_flags(struct jfs_inod
4443 jfs_ip->mode2 |= JFS_DIRSYNC_FL;
4445 jfs_ip->mode2 |= JFS_SYNC_FL;
4447 + if (vflags & V_BARRIER)
4448 + jfs_ip->mode2 |= JFS_BARRIER_FL;
4449 + if (vflags & V_COW)
4450 + jfs_ip->mode2 |= JFS_COW_FL;
4454 @@ -105,6 +128,7 @@ struct inode *ialloc(struct inode *paren
4457 inode->i_gid = current_fsgid();
4458 + inode->i_tag = dx_current_fstag(sb);
4461 * New inodes need to save sane values on disk when
4462 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/jfs_inode.h linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_inode.h
4463 --- linux-2.6.32.1/fs/jfs/jfs_inode.h 2009-06-11 17:13:05.000000000 +0200
4464 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/jfs_inode.h 2009-12-03 20:04:56.000000000 +0100
4465 @@ -39,6 +39,7 @@ extern struct dentry *jfs_fh_to_dentry(s
4466 extern struct dentry *jfs_fh_to_parent(struct super_block *sb, struct fid *fid,
4467 int fh_len, int fh_type);
4468 extern void jfs_set_inode_flags(struct inode *);
4469 +extern int jfs_sync_flags(struct inode *, int, int);
4470 extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
4472 extern const struct address_space_operations jfs_aops;
4473 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/namei.c
4474 --- linux-2.6.32.1/fs/jfs/namei.c 2009-12-03 20:02:52.000000000 +0100
4475 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/namei.c 2009-12-03 20:04:56.000000000 +0100
4477 #include <linux/ctype.h>
4478 #include <linux/quotaops.h>
4479 #include <linux/exportfs.h>
4480 +#include <linux/vs_tag.h>
4481 #include "jfs_incore.h"
4482 #include "jfs_superblock.h"
4483 #include "jfs_inode.h"
4484 @@ -1476,6 +1477,7 @@ static struct dentry *jfs_lookup(struct
4485 return ERR_CAST(ip);
4488 + dx_propagate_tag(nd, ip);
4489 dentry = d_splice_alias(ip, dentry);
4491 if (dentry && (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2))
4492 @@ -1545,6 +1547,7 @@ const struct inode_operations jfs_dir_in
4493 .setattr = jfs_setattr,
4494 .check_acl = jfs_check_acl,
4496 + .sync_flags = jfs_sync_flags,
4499 const struct file_operations jfs_dir_operations = {
4500 diff -NurpP --minimal linux-2.6.32.1/fs/jfs/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/super.c
4501 --- linux-2.6.32.1/fs/jfs/super.c 2009-12-03 20:02:52.000000000 +0100
4502 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/jfs/super.c 2009-12-03 20:04:56.000000000 +0100
4503 @@ -192,7 +192,8 @@ static void jfs_put_super(struct super_b
4505 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
4506 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
4507 - Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
4508 + Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask,
4509 + Opt_tag, Opt_notag, Opt_tagid
4512 static const match_table_t tokens = {
4513 @@ -202,6 +203,10 @@ static const match_table_t tokens = {
4514 {Opt_resize, "resize=%u"},
4515 {Opt_resize_nosize, "resize"},
4516 {Opt_errors, "errors=%s"},
4518 + {Opt_notag, "notag"},
4519 + {Opt_tagid, "tagid=%u"},
4520 + {Opt_tag, "tagxid"},
4521 {Opt_ignore, "noquota"},
4522 {Opt_ignore, "quota"},
4523 {Opt_usrquota, "usrquota"},
4524 @@ -336,6 +341,20 @@ static int parse_options(char *options,
4528 +#ifndef CONFIG_TAGGING_NONE
4530 + *flag |= JFS_TAGGED;
4533 + *flag &= JFS_TAGGED;
4536 +#ifdef CONFIG_PROPAGATE
4539 + *flag |= JFS_TAGGED;
4543 printk("jfs: Unrecognized mount option \"%s\" "
4544 " or missing value\n", p);
4545 @@ -366,6 +385,12 @@ static int jfs_remount(struct super_bloc
4546 if (!parse_options(data, sb, &newLVSize, &flag)) {
4549 + if ((flag & JFS_TAGGED) && !(sb->s_flags & MS_TAGGED)) {
4550 + printk(KERN_ERR "JFS: %s: tagging not permitted on remount.\n",
4557 if (sb->s_flags & MS_RDONLY) {
4558 @@ -449,6 +474,9 @@ static int jfs_fill_super(struct super_b
4559 #ifdef CONFIG_JFS_POSIX_ACL
4560 sb->s_flags |= MS_POSIXACL;
4562 + /* map mount option tagxid */
4563 + if (sbi->flag & JFS_TAGGED)
4564 + sb->s_flags |= MS_TAGGED;
4567 printk(KERN_ERR "resize option for remount only\n");
4568 diff -NurpP --minimal linux-2.6.32.1/fs/libfs.c linux-2.6.32.1-vs2.3.0.36.27/fs/libfs.c
4569 --- linux-2.6.32.1/fs/libfs.c 2009-12-03 20:02:52.000000000 +0100
4570 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/libfs.c 2009-12-03 20:04:56.000000000 +0100
4571 @@ -127,7 +127,8 @@ static inline unsigned char dt_type(stru
4572 * both impossible due to the lock on directory.
4575 -int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
4576 +static inline int do_dcache_readdir_filter(struct file *filp,
4577 + void *dirent, filldir_t filldir, int (*filter)(struct dentry *dentry))
4579 struct dentry *dentry = filp->f_path.dentry;
4580 struct dentry *cursor = filp->private_data;
4581 @@ -160,6 +161,8 @@ int dcache_readdir(struct file * filp, v
4582 next = list_entry(p, struct dentry, d_u.d_child);
4583 if (d_unhashed(next) || !next->d_inode)
4585 + if (filter && !filter(next))
4588 spin_unlock(&dcache_lock);
4589 if (filldir(dirent, next->d_name.name,
4590 @@ -178,6 +181,18 @@ int dcache_readdir(struct file * filp, v
4594 +int dcache_readdir(struct file *filp, void *dirent, filldir_t filldir)
4596 + return do_dcache_readdir_filter(filp, dirent, filldir, NULL);
4599 +int dcache_readdir_filter(struct file *filp, void *dirent, filldir_t filldir,
4600 + int (*filter)(struct dentry *))
4602 + return do_dcache_readdir_filter(filp, dirent, filldir, filter);
4606 ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
4609 @@ -841,6 +856,7 @@ EXPORT_SYMBOL(dcache_dir_close);
4610 EXPORT_SYMBOL(dcache_dir_lseek);
4611 EXPORT_SYMBOL(dcache_dir_open);
4612 EXPORT_SYMBOL(dcache_readdir);
4613 +EXPORT_SYMBOL(dcache_readdir_filter);
4614 EXPORT_SYMBOL(generic_read_dir);
4615 EXPORT_SYMBOL(get_sb_pseudo);
4616 EXPORT_SYMBOL(simple_write_begin);
4617 diff -NurpP --minimal linux-2.6.32.1/fs/locks.c linux-2.6.32.1-vs2.3.0.36.27/fs/locks.c
4618 --- linux-2.6.32.1/fs/locks.c 2009-12-03 20:02:52.000000000 +0100
4619 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/locks.c 2009-12-03 20:04:56.000000000 +0100
4621 #include <linux/time.h>
4622 #include <linux/rcupdate.h>
4623 #include <linux/pid_namespace.h>
4624 +#include <linux/vs_base.h>
4625 +#include <linux/vs_limit.h>
4627 #include <asm/uaccess.h>
4629 @@ -148,6 +150,8 @@ static struct kmem_cache *filelock_cache
4630 /* Allocate an empty lock structure. */
4631 static struct file_lock *locks_alloc_lock(void)
4633 + if (!vx_locks_avail(1))
4635 return kmem_cache_alloc(filelock_cache, GFP_KERNEL);
4638 @@ -174,6 +178,7 @@ static void locks_free_lock(struct file_
4639 BUG_ON(!list_empty(&fl->fl_block));
4640 BUG_ON(!list_empty(&fl->fl_link));
4643 locks_release_private(fl);
4644 kmem_cache_free(filelock_cache, fl);
4646 @@ -194,6 +199,7 @@ void locks_init_lock(struct file_lock *f
4647 fl->fl_start = fl->fl_end = 0;
4649 fl->fl_lmops = NULL;
4653 EXPORT_SYMBOL(locks_init_lock);
4654 @@ -248,6 +254,7 @@ void locks_copy_lock(struct file_lock *n
4655 new->fl_file = fl->fl_file;
4656 new->fl_ops = fl->fl_ops;
4657 new->fl_lmops = fl->fl_lmops;
4658 + new->fl_xid = fl->fl_xid;
4660 locks_copy_private(new, fl);
4662 @@ -286,6 +293,11 @@ static int flock_make_lock(struct file *
4663 fl->fl_flags = FL_FLOCK;
4665 fl->fl_end = OFFSET_MAX;
4667 + vxd_assert(filp->f_xid == vx_current_xid(),
4668 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4669 + fl->fl_xid = filp->f_xid;
4674 @@ -451,6 +463,7 @@ static int lease_init(struct file *filp,
4676 fl->fl_owner = current->files;
4677 fl->fl_pid = current->tgid;
4678 + fl->fl_xid = vx_current_xid();
4681 fl->fl_flags = FL_LEASE;
4682 @@ -470,6 +483,11 @@ static struct file_lock *lease_alloc(str
4684 return ERR_PTR(error);
4686 + fl->fl_xid = vx_current_xid();
4688 + vxd_assert(filp->f_xid == fl->fl_xid,
4689 + "f_xid(%d) == fl_xid(%d)", filp->f_xid, fl->fl_xid);
4691 error = lease_init(filp, type, fl);
4693 locks_free_lock(fl);
4694 @@ -770,6 +788,7 @@ static int flock_lock_file(struct file *
4698 + new_fl->fl_xid = -1;
4700 for_each_lock(inode, before) {
4701 struct file_lock *fl = *before;
4702 @@ -790,6 +809,7 @@ find_conflict:
4704 locks_copy_lock(new_fl, request);
4705 locks_insert_lock(before, new_fl);
4706 + vx_locks_inc(new_fl);
4710 @@ -800,7 +820,8 @@ out:
4714 -static int __posix_lock_file(struct inode *inode, struct file_lock *request, struct file_lock *conflock)
4715 +static int __posix_lock_file(struct inode *inode, struct file_lock *request,
4716 + struct file_lock *conflock, xid_t xid)
4718 struct file_lock *fl;
4719 struct file_lock *new_fl = NULL;
4720 @@ -810,6 +831,8 @@ static int __posix_lock_file(struct inod
4721 struct file_lock **before;
4722 int error, added = 0;
4724 + vxd_assert(xid == vx_current_xid(),
4725 + "xid(%d) == current(%d)", xid, vx_current_xid());
4727 * We may need two file_lock structures for this operation,
4728 * so we get them in advance to avoid races.
4729 @@ -820,7 +843,11 @@ static int __posix_lock_file(struct inod
4730 (request->fl_type != F_UNLCK ||
4731 request->fl_start != 0 || request->fl_end != OFFSET_MAX)) {
4732 new_fl = locks_alloc_lock();
4733 + new_fl->fl_xid = xid;
4734 + vx_locks_inc(new_fl);
4735 new_fl2 = locks_alloc_lock();
4736 + new_fl2->fl_xid = xid;
4737 + vx_locks_inc(new_fl2);
4741 @@ -1019,7 +1046,8 @@ static int __posix_lock_file(struct inod
4742 int posix_lock_file(struct file *filp, struct file_lock *fl,
4743 struct file_lock *conflock)
4745 - return __posix_lock_file(filp->f_path.dentry->d_inode, fl, conflock);
4746 + return __posix_lock_file(filp->f_path.dentry->d_inode,
4747 + fl, conflock, filp->f_xid);
4749 EXPORT_SYMBOL(posix_lock_file);
4751 @@ -1109,7 +1137,7 @@ int locks_mandatory_area(int read_write,
4752 fl.fl_end = offset + count - 1;
4755 - error = __posix_lock_file(inode, &fl, NULL);
4756 + error = __posix_lock_file(inode, &fl, NULL, filp->f_xid);
4757 if (error != FILE_LOCK_DEFERRED)
4759 error = wait_event_interruptible(fl.fl_wait, !fl.fl_next);
4760 @@ -1424,6 +1452,7 @@ int generic_setlease(struct file *filp,
4762 locks_copy_lock(new_fl, lease);
4763 locks_insert_lock(before, new_fl);
4764 + vx_locks_inc(new_fl);
4768 @@ -1779,6 +1808,11 @@ int fcntl_setlk(unsigned int fd, struct
4769 if (file_lock == NULL)
4772 + vxd_assert(filp->f_xid == vx_current_xid(),
4773 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4774 + file_lock->fl_xid = filp->f_xid;
4775 + vx_locks_inc(file_lock);
4778 * This might block, so we do it before checking the inode.
4780 @@ -1897,6 +1931,11 @@ int fcntl_setlk64(unsigned int fd, struc
4781 if (file_lock == NULL)
4784 + vxd_assert(filp->f_xid == vx_current_xid(),
4785 + "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4786 + file_lock->fl_xid = filp->f_xid;
4787 + vx_locks_inc(file_lock);
4790 * This might block, so we do it before checking the inode.
4792 @@ -2162,8 +2201,11 @@ static int locks_show(struct seq_file *f
4794 lock_get_status(f, fl, (long)f->private, "");
4796 - list_for_each_entry(bfl, &fl->fl_block, fl_block)
4797 + list_for_each_entry(bfl, &fl->fl_block, fl_block) {
4798 + if (!vx_check(fl->fl_xid, VS_WATCH_P | VS_IDENT))
4800 lock_get_status(f, bfl, (long)f->private, " ->");
4805 diff -NurpP --minimal linux-2.6.32.1/fs/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/namei.c
4806 --- linux-2.6.32.1/fs/namei.c 2009-12-03 20:02:52.000000000 +0100
4807 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/namei.c 2009-12-03 20:04:56.000000000 +0100
4809 #include <linux/fcntl.h>
4810 #include <linux/device_cgroup.h>
4811 #include <linux/fs_struct.h>
4812 +#include <linux/proc_fs.h>
4813 +#include <linux/vserver/inode.h>
4814 +#include <linux/vs_base.h>
4815 +#include <linux/vs_tag.h>
4816 +#include <linux/vs_cowbl.h>
4817 +#include <linux/vs_device.h>
4818 +#include <linux/vs_context.h>
4819 +#include <linux/pid_namespace.h>
4820 #include <asm/uaccess.h>
4822 #define ACC_MODE(x) ("\000\004\002\006"[(x)&O_ACCMODE])
4823 @@ -169,6 +177,77 @@ void putname(const char *name)
4824 EXPORT_SYMBOL(putname);
4827 +static inline int dx_barrier(const struct inode *inode)
4829 + if (IS_BARRIER(inode) && !vx_check(0, VS_ADMIN | VS_WATCH)) {
4830 + vxwprintk_task(1, "did hit the barrier.");
4836 +static int __dx_permission(const struct inode *inode, int mask)
4838 + if (dx_barrier(inode))
4841 + if (inode->i_sb->s_magic == DEVPTS_SUPER_MAGIC) {
4842 + /* devpts is xid tagged */
4843 + if (S_ISDIR(inode->i_mode) ||
4844 + vx_check((xid_t)inode->i_tag, VS_IDENT | VS_WATCH_P))
4847 + else if (inode->i_sb->s_magic == PROC_SUPER_MAGIC) {
4848 + struct proc_dir_entry *de = PDE(inode);
4850 + if (de && !vx_hide_check(0, de->vx_flags))
4853 + if ((mask & (MAY_WRITE | MAY_APPEND))) {
4855 + struct task_struct *tsk;
4857 + if (vx_check(0, VS_ADMIN | VS_WATCH_P) ||
4858 + vx_flags(VXF_STATE_SETUP, 0))
4861 + pid = PROC_I(inode)->pid;
4865 + tsk = pid_task(pid, PIDTYPE_PID);
4866 + vxdprintk(VXD_CBIT(tag, 0), "accessing %p[#%u]",
4867 + tsk, (tsk ? vx_task_xid(tsk) : 0));
4868 + if (tsk && vx_check(vx_task_xid(tsk), VS_IDENT | VS_WATCH_P))
4872 + /* FIXME: Should we block some entries here? */
4877 + if (dx_notagcheck(inode->i_sb) ||
4878 + dx_check(inode->i_tag, DX_HOSTID | DX_ADMIN | DX_WATCH |
4887 +int dx_permission(const struct inode *inode, int mask)
4889 + int ret = __dx_permission(inode, mask);
4890 + if (unlikely(ret)) {
4891 + vxwprintk_task(1, "denied %x access to %s:%p[#%d,%lu]",
4892 + mask, inode->i_sb->s_id, inode, inode->i_tag,
4899 * This does basic POSIX ACL permission checking
4901 @@ -268,10 +347,14 @@ int inode_permission(struct inode *inode
4903 * Nobody gets write access to an immutable file.
4905 - if (IS_IMMUTABLE(inode))
4906 + if (IS_IMMUTABLE(inode) && !IS_COW(inode))
4910 + retval = dx_permission(inode, mask);
4914 if (inode->i_op->permission)
4915 retval = inode->i_op->permission(inode, mask);
4917 @@ -447,6 +530,9 @@ static int exec_permission_lite(struct i
4921 + if (dx_barrier(inode))
4924 if (inode->i_op->permission) {
4925 ret = inode->i_op->permission(inode, MAY_EXEC);
4927 @@ -762,7 +848,8 @@ static __always_inline void follow_dotdo
4929 if (nd->path.dentry == nd->root.dentry &&
4930 nd->path.mnt == nd->root.mnt) {
4932 + /* for sane '/' avoid follow_mount() */
4935 spin_lock(&dcache_lock);
4936 if (nd->path.dentry != nd->path.mnt->mnt_root) {
4937 @@ -798,16 +885,30 @@ static int do_lookup(struct nameidata *n
4939 struct vfsmount *mnt = nd->path.mnt;
4940 struct dentry *dentry = __d_lookup(nd->path.dentry, name);
4941 + struct inode *inode;
4945 if (dentry->d_op && dentry->d_op->d_revalidate)
4946 goto need_revalidate;
4947 + inode = dentry->d_inode;
4951 + if (__dx_permission(inode, MAY_ACCESS))
4956 path->dentry = dentry;
4957 __follow_mount(path);
4960 + vxwprintk_task(1, "did lookup hidden %s:%p[#%d,%lu] »%s/%.*s«.",
4961 + inode->i_sb->s_id, inode, inode->i_tag, inode->i_ino,
4962 + vxd_path(&nd->path), name->len, name->name);
4967 dentry = real_lookup(nd->path.dentry, name, nd);
4968 @@ -1389,7 +1490,7 @@ static int may_delete(struct inode *dir,
4971 if (check_sticky(dir, victim->d_inode)||IS_APPEND(victim->d_inode)||
4972 - IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
4973 + IS_IXORUNLINK(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
4976 if (!S_ISDIR(victim->d_inode->i_mode))
4977 @@ -1529,6 +1630,14 @@ int may_open(struct path *path, int acc_
4981 +#ifdef CONFIG_VSERVER_COWBL
4982 + if (IS_COW(inode) && (flag & FMODE_WRITE)) {
4983 + if (IS_COW_LINK(inode))
4985 + inode->i_flags &= ~(S_IXUNLINK|S_IMMUTABLE);
4986 + mark_inode_dirty(inode);
4989 error = inode_permission(inode, acc_mode);
4992 @@ -1677,7 +1786,11 @@ struct file *do_filp_open(int dfd, const
4995 int flag = open_to_namei_flags(open_flag);
4997 +#ifdef CONFIG_VSERVER_COWBL
5003 acc_mode = MAY_OPEN | ACC_MODE(flag);
5005 @@ -1825,6 +1938,25 @@ ok:
5008 error = may_open(&nd.path, acc_mode, flag);
5009 +#ifdef CONFIG_VSERVER_COWBL
5010 + if (error == -EMLINK) {
5011 + struct dentry *dentry;
5012 + dentry = cow_break_link(pathname);
5013 + if (IS_ERR(dentry)) {
5014 + error = PTR_ERR(dentry);
5019 + mnt_drop_write(nd.path.mnt);
5020 + release_open_intent(&nd);
5021 + path_put(&nd.path);
5030 mnt_drop_write(nd.path.mnt);
5031 @@ -1987,9 +2119,17 @@ int vfs_mknod(struct inode *dir, struct
5035 - if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
5036 + if (!(S_ISCHR(mode) || S_ISBLK(mode)))
5039 + if (!capable(CAP_MKNOD))
5042 + if (S_ISCHR(mode) && !vs_chrdev_perm(dev, DATTR_CREATE))
5044 + if (S_ISBLK(mode) && !vs_blkdev_perm(dev, DATTR_CREATE))
5047 if (!dir->i_op->mknod)
5050 @@ -2456,7 +2596,7 @@ int vfs_link(struct dentry *old_dentry,
5052 * A link to an append-only or immutable file cannot be created.
5054 - if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
5055 + if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
5057 if (!dir->i_op->link)
5059 @@ -2829,6 +2969,219 @@ int vfs_follow_link(struct nameidata *nd
5060 return __vfs_follow_link(nd, link);
5064 +#ifdef CONFIG_VSERVER_COWBL
5066 +#include <linux/file.h>
5069 +long do_cow_splice(struct file *in, struct file *out, size_t len)
5073 + return do_splice_direct(in, &ppos, out, len, 0);
5076 +struct dentry *cow_break_link(const char *pathname)
5078 + int ret, mode, pathlen, redo = 0;
5079 + struct nameidata old_nd, dir_nd;
5080 + struct path old_path, new_path;
5081 + struct dentry *dir, *res = NULL;
5082 + struct file *old_file;
5083 + struct file *new_file;
5084 + char *to, *path, pad='\251';
5087 + vxdprintk(VXD_CBIT(misc, 1), "cow_break_link(»%s«)", pathname);
5088 + path = kmalloc(PATH_MAX, GFP_KERNEL);
5093 + /* old_nd will have refs to dentry and mnt */
5094 + ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5095 + vxdprintk(VXD_CBIT(misc, 2), "path_lookup(old): %d", ret);
5097 + goto out_free_path;
5099 + old_path = old_nd.path;
5100 + mode = old_path.dentry->d_inode->i_mode;
5102 + to = d_path(&old_path, path, PATH_MAX-2);
5103 + pathlen = strlen(to);
5104 + vxdprintk(VXD_CBIT(misc, 2), "old path »%s« [»%.*s«:%d]", to,
5105 + old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5106 + old_path.dentry->d_name.len);
5108 + to[pathlen + 1] = 0;
5110 + to[pathlen] = pad--;
5112 + if (pad <= '\240')
5115 + vxdprintk(VXD_CBIT(misc, 1), "temp copy »%s«", to);
5116 + /* dir_nd will have refs to dentry and mnt */
5117 + ret = path_lookup(to,
5118 + LOOKUP_PARENT | LOOKUP_OPEN | LOOKUP_CREATE, &dir_nd);
5119 + vxdprintk(VXD_CBIT(misc, 2),
5120 + "path_lookup(new): %d", ret);
5124 + /* this puppy downs the inode mutex */
5125 + new_path.dentry = lookup_create(&dir_nd, 0);
5126 + if (!new_path.dentry || IS_ERR(new_path.dentry)) {
5127 + vxdprintk(VXD_CBIT(misc, 2),
5128 + "lookup_create(new): %p", new_path.dentry);
5129 + mutex_unlock(&dir_nd.path.dentry->d_inode->i_mutex);
5130 + path_put(&dir_nd.path);
5133 + vxdprintk(VXD_CBIT(misc, 2),
5134 + "lookup_create(new): %p [»%.*s«:%d]", new_path.dentry,
5135 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5136 + new_path.dentry->d_name.len);
5137 + dir = dir_nd.path.dentry;
5139 + ret = vfs_create(dir_nd.path.dentry->d_inode, new_path.dentry, mode, &dir_nd);
5140 + vxdprintk(VXD_CBIT(misc, 2),
5141 + "vfs_create(new): %d", ret);
5142 + if (ret == -EEXIST) {
5143 + mutex_unlock(&dir->d_inode->i_mutex);
5144 + dput(new_path.dentry);
5145 + path_put(&dir_nd.path);
5149 + goto out_unlock_new;
5151 + /* drop out early, ret passes ENOENT */
5153 + if ((redo = d_unhashed(old_path.dentry)))
5154 + goto out_unlock_new;
5156 + new_path.mnt = dir_nd.path.mnt;
5157 + dget(old_path.dentry);
5158 + mntget(old_path.mnt);
5159 + /* this one cleans up the dentry/mnt in case of failure */
5160 + old_file = dentry_open(old_path.dentry, old_path.mnt,
5161 + O_RDONLY, current_cred());
5162 + vxdprintk(VXD_CBIT(misc, 2),
5163 + "dentry_open(old): %p", old_file);
5164 + if (!old_file || IS_ERR(old_file)) {
5165 + res = IS_ERR(old_file) ? (void *) old_file : res;
5166 + goto out_unlock_new;
5169 + dget(new_path.dentry);
5170 + mntget(new_path.mnt);
5171 + /* this one cleans up the dentry/mnt in case of failure */
5172 + new_file = dentry_open(new_path.dentry, new_path.mnt,
5173 + O_WRONLY, current_cred());
5174 + vxdprintk(VXD_CBIT(misc, 2),
5175 + "dentry_open(new): %p", new_file);
5177 + ret = IS_ERR(new_file) ? PTR_ERR(new_file) : -ENOENT;
5178 + if (!new_file || IS_ERR(new_file))
5179 + goto out_fput_old;
5181 + size = i_size_read(old_file->f_dentry->d_inode);
5182 + ret = do_cow_splice(old_file, new_file, size);
5183 + vxdprintk(VXD_CBIT(misc, 2), "do_splice_direct: %d", ret);
5185 + goto out_fput_both;
5186 + } else if (ret < size) {
5188 + goto out_fput_both;
5190 + struct inode *old_inode = old_path.dentry->d_inode;
5191 + struct inode *new_inode = new_path.dentry->d_inode;
5192 + struct iattr attr = {
5193 + .ia_uid = old_inode->i_uid,
5194 + .ia_gid = old_inode->i_gid,
5195 + .ia_valid = ATTR_UID | ATTR_GID
5198 + ret = inode_setattr(new_inode, &attr);
5200 + goto out_fput_both;
5203 + mutex_lock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5205 + /* drop out late */
5207 + if ((redo = d_unhashed(old_path.dentry)))
5210 + vxdprintk(VXD_CBIT(misc, 2),
5211 + "vfs_rename: [»%*s«:%d] -> [»%*s«:%d]",
5212 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5213 + new_path.dentry->d_name.len,
5214 + old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5215 + old_path.dentry->d_name.len);
5216 + ret = vfs_rename(dir_nd.path.dentry->d_inode, new_path.dentry,
5217 + old_nd.path.dentry->d_parent->d_inode, old_path.dentry);
5218 + vxdprintk(VXD_CBIT(misc, 2), "vfs_rename: %d", ret);
5219 + res = new_path.dentry;
5222 + mutex_unlock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5225 + vxdprintk(VXD_CBIT(misc, 3),
5226 + "fput(new_file=%p[#%ld])", new_file,
5227 + atomic_long_read(&new_file->f_count));
5231 + vxdprintk(VXD_CBIT(misc, 3),
5232 + "fput(old_file=%p[#%ld])", old_file,
5233 + atomic_long_read(&old_file->f_count));
5237 + mutex_unlock(&dir->d_inode->i_mutex);
5241 + /* error path cleanup */
5242 + vfs_unlink(dir->d_inode, new_path.dentry);
5243 + dput(new_path.dentry);
5247 + goto out_rel_both;
5248 + /* lookup dentry once again */
5249 + path_put(&old_nd.path);
5250 + ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5252 + goto out_rel_both;
5254 + new_path.dentry = old_nd.path.dentry;
5255 + vxdprintk(VXD_CBIT(misc, 2),
5256 + "path_lookup(redo): %p [»%.*s«:%d]", new_path.dentry,
5257 + new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5258 + new_path.dentry->d_name.len);
5259 + dget(new_path.dentry);
5260 + res = new_path.dentry;
5263 + path_put(&dir_nd.path);
5265 + path_put(&old_nd.path);
5270 + res = ERR_PTR(ret);
5276 /* get the link contents into pagecache */
5277 static char *page_getlink(struct dentry * dentry, struct page **ppage)
5279 diff -NurpP --minimal linux-2.6.32.1/fs/namespace.c linux-2.6.32.1-vs2.3.0.36.27/fs/namespace.c
5280 --- linux-2.6.32.1/fs/namespace.c 2009-12-03 20:02:52.000000000 +0100
5281 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/namespace.c 2009-12-03 20:04:56.000000000 +0100
5283 #include <linux/log2.h>
5284 #include <linux/idr.h>
5285 #include <linux/fs_struct.h>
5286 +#include <linux/vs_base.h>
5287 +#include <linux/vs_context.h>
5288 +#include <linux/vs_tag.h>
5289 +#include <linux/vserver/space.h>
5290 +#include <linux/vserver/global.h>
5291 #include <asm/uaccess.h>
5292 #include <asm/unistd.h>
5294 @@ -567,6 +572,7 @@ static struct vfsmount *clone_mnt(struct
5295 mnt->mnt_root = dget(root);
5296 mnt->mnt_mountpoint = mnt->mnt_root;
5297 mnt->mnt_parent = mnt;
5298 + mnt->mnt_tag = old->mnt_tag;
5300 if (flag & CL_SLAVE) {
5301 list_add(&mnt->mnt_slave, &old->mnt_slave_list);
5302 @@ -661,6 +667,31 @@ static inline void mangle(struct seq_fil
5303 seq_escape(m, s, " \t\n\\");
5306 +static int mnt_is_reachable(struct vfsmount *mnt)
5309 + struct dentry *point;
5312 + if (mnt == mnt->mnt_ns->root)
5315 + spin_lock(&vfsmount_lock);
5316 + root = current->fs->root;
5317 + point = root.dentry;
5319 + while ((mnt != mnt->mnt_parent) && (mnt != root.mnt)) {
5320 + point = mnt->mnt_mountpoint;
5321 + mnt = mnt->mnt_parent;
5324 + ret = (mnt == root.mnt) && is_subdir(point, root.dentry);
5326 + spin_unlock(&vfsmount_lock);
5332 * Simple .show_options callback for filesystems which don't want to
5333 * implement more complex mount option showing.
5334 @@ -748,6 +779,8 @@ static int show_sb_opts(struct seq_file
5335 { MS_SYNCHRONOUS, ",sync" },
5336 { MS_DIRSYNC, ",dirsync" },
5337 { MS_MANDLOCK, ",mand" },
5338 + { MS_TAGGED, ",tag" },
5339 + { MS_NOTAGCHECK, ",notagcheck" },
5342 const struct proc_fs_info *fs_infop;
5343 @@ -795,10 +828,20 @@ static int show_vfsmnt(struct seq_file *
5345 struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5347 - mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5349 - seq_path(m, &mnt_path, " \t\n\\");
5351 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5353 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5356 + if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5357 + mnt == current->fs->root.mnt) {
5358 + seq_puts(m, "/dev/root / ");
5360 + mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5362 + seq_path(m, &mnt_path, " \t\n\\");
5365 show_type(m, mnt->mnt_sb);
5366 seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
5367 err = show_sb_opts(m, mnt->mnt_sb);
5368 @@ -828,6 +871,11 @@ static int show_mountinfo(struct seq_fil
5369 struct path root = p->root;
5372 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5374 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5377 seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
5378 MAJOR(sb->s_dev), MINOR(sb->s_dev));
5379 seq_dentry(m, mnt->mnt_root, " \t\n\\");
5380 @@ -886,17 +934,27 @@ static int show_vfsstat(struct seq_file
5381 struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5385 - if (mnt->mnt_devname) {
5386 - seq_puts(m, "device ");
5387 - mangle(m, mnt->mnt_devname);
5389 - seq_puts(m, "no device");
5390 + if (vx_flags(VXF_HIDE_MOUNT, 0))
5392 + if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5396 - seq_puts(m, " mounted on ");
5397 - seq_path(m, &mnt_path, " \t\n\\");
5399 + if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5400 + mnt == current->fs->root.mnt) {
5401 + seq_puts(m, "device /dev/root mounted on / ");
5404 + if (mnt->mnt_devname) {
5405 + seq_puts(m, "device ");
5406 + mangle(m, mnt->mnt_devname);
5408 + seq_puts(m, "no device");
5411 + seq_puts(m, " mounted on ");
5412 + seq_path(m, &mnt_path, " \t\n\\");
5416 /* file system type */
5417 seq_puts(m, "with fstype ");
5418 @@ -1130,7 +1188,7 @@ SYSCALL_DEFINE2(umount, char __user *, n
5422 - if (!capable(CAP_SYS_ADMIN))
5423 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5426 retval = do_umount(path.mnt, flags);
5427 @@ -1156,7 +1214,7 @@ SYSCALL_DEFINE1(oldumount, char __user *
5429 static int mount_is_safe(struct path *path)
5431 - if (capable(CAP_SYS_ADMIN))
5432 + if (vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5436 @@ -1420,7 +1478,7 @@ static int do_change_type(struct path *p
5437 int type = flag & ~MS_REC;
5440 - if (!capable(CAP_SYS_ADMIN))
5441 + if (!vx_capable(CAP_SYS_ADMIN, VXC_NAMESPACE))
5444 if (path->dentry != path->mnt->mnt_root)
5445 @@ -1447,11 +1505,13 @@ static int do_change_type(struct path *p
5446 * do loopback mount.
5448 static int do_loopback(struct path *path, char *old_name,
5450 + tag_t tag, unsigned long flags, int mnt_flags)
5452 struct path old_path;
5453 struct vfsmount *mnt = NULL;
5454 int err = mount_is_safe(path);
5455 + int recurse = flags & MS_REC;
5459 if (!old_name || !*old_name)
5460 @@ -1485,6 +1545,7 @@ static int do_loopback(struct path *path
5461 spin_unlock(&vfsmount_lock);
5462 release_mounts(&umount_list);
5464 + mnt->mnt_flags = mnt_flags;
5467 up_write(&namespace_sem);
5468 @@ -1515,12 +1576,12 @@ static int change_mount_flags(struct vfs
5469 * on it - tough luck.
5471 static int do_remount(struct path *path, int flags, int mnt_flags,
5473 + void *data, xid_t xid)
5476 struct super_block *sb = path->mnt->mnt_sb;
5478 - if (!capable(CAP_SYS_ADMIN))
5479 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_REMOUNT))
5482 if (!check_mnt(path->mnt))
5483 @@ -1562,7 +1623,7 @@ static int do_move_mount(struct path *pa
5484 struct path old_path, parent_path;
5487 - if (!capable(CAP_SYS_ADMIN))
5488 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5490 if (!old_name || !*old_name)
5492 @@ -1644,7 +1705,7 @@ static int do_new_mount(struct path *pat
5495 /* we need capabilities... */
5496 - if (!capable(CAP_SYS_ADMIN))
5497 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5501 @@ -1908,6 +1969,7 @@ long do_mount(char *dev_name, char *dir_
5508 if ((flags & MS_MGC_MSK) == MS_MGC_VAL)
5509 @@ -1925,6 +1987,12 @@ long do_mount(char *dev_name, char *dir_
5510 if (!(flags & MS_NOATIME))
5511 mnt_flags |= MNT_RELATIME;
5513 + if (dx_parse_tag(data_page, &tag, 1, &mnt_flags, &flags)) {
5514 + /* FIXME: bind and re-mounts get the tag flag? */
5515 + if (flags & (MS_BIND|MS_REMOUNT))
5516 + flags |= MS_TAGID;
5519 /* Separate the per-mountpoint flags */
5520 if (flags & MS_NOSUID)
5521 mnt_flags |= MNT_NOSUID;
5522 @@ -1941,6 +2009,8 @@ long do_mount(char *dev_name, char *dir_
5523 if (flags & MS_RDONLY)
5524 mnt_flags |= MNT_READONLY;
5526 + if (!capable(CAP_SYS_ADMIN))
5527 + mnt_flags |= MNT_NODEV;
5528 flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE |
5529 MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT |
5531 @@ -1957,9 +2027,9 @@ long do_mount(char *dev_name, char *dir_
5533 if (flags & MS_REMOUNT)
5534 retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
5537 else if (flags & MS_BIND)
5538 - retval = do_loopback(&path, dev_name, flags & MS_REC);
5539 + retval = do_loopback(&path, dev_name, tag, flags, mnt_flags);
5540 else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
5541 retval = do_change_type(&path, flags);
5542 else if (flags & MS_MOVE)
5543 @@ -2038,6 +2108,7 @@ static struct mnt_namespace *dup_mnt_ns(
5544 q = next_mnt(q, new_ns->root);
5546 up_write(&namespace_sem);
5547 + atomic_inc(&vs_global_mnt_ns);
5551 @@ -2182,9 +2253,10 @@ SYSCALL_DEFINE2(pivot_root, const char _
5552 down_write(&namespace_sem);
5553 mutex_lock(&old.dentry->d_inode->i_mutex);
5555 - if (IS_MNT_SHARED(old.mnt) ||
5556 + if ((IS_MNT_SHARED(old.mnt) ||
5557 IS_MNT_SHARED(new.mnt->mnt_parent) ||
5558 - IS_MNT_SHARED(root.mnt->mnt_parent))
5559 + IS_MNT_SHARED(root.mnt->mnt_parent)) &&
5560 + !vx_flags(VXF_STATE_SETUP, 0))
5562 if (!check_mnt(root.mnt))
5564 @@ -2320,6 +2392,7 @@ void put_mnt_ns(struct mnt_namespace *ns
5565 spin_unlock(&vfsmount_lock);
5566 up_write(&namespace_sem);
5567 release_mounts(&umount_list);
5568 + atomic_dec(&vs_global_mnt_ns);
5571 EXPORT_SYMBOL(put_mnt_ns);
5572 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/client.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/client.c
5573 --- linux-2.6.32.1/fs/nfs/client.c 2009-12-03 20:02:52.000000000 +0100
5574 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/client.c 2009-12-03 20:04:56.000000000 +0100
5575 @@ -738,6 +738,9 @@ static int nfs_init_server_rpcclient(str
5576 if (server->flags & NFS_MOUNT_SOFT)
5577 server->client->cl_softrtry = 1;
5579 + server->client->cl_tag = 0;
5580 + if (server->flags & NFS_MOUNT_TAGGED)
5581 + server->client->cl_tag = 1;
5585 @@ -909,6 +912,10 @@ static void nfs_server_set_fsinfo(struct
5586 server->acdirmin = server->acdirmax = 0;
5589 + /* FIXME: needs fsinfo
5590 + if (server->flags & NFS_MOUNT_TAGGED)
5591 + sb->s_flags |= MS_TAGGED; */
5593 server->maxfilesize = fsinfo->maxfilesize;
5595 /* We're airborne Set socket buffersize */
5596 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/dir.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/dir.c
5597 --- linux-2.6.32.1/fs/nfs/dir.c 2009-12-03 20:02:52.000000000 +0100
5598 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/dir.c 2009-12-03 20:04:56.000000000 +0100
5600 #include <linux/namei.h>
5601 #include <linux/mount.h>
5602 #include <linux/sched.h>
5603 +#include <linux/vs_tag.h>
5605 #include "nfs4_fs.h"
5606 #include "delegation.h"
5607 @@ -949,6 +950,7 @@ static struct dentry *nfs_lookup(struct
5609 goto out_unblock_sillyrename;
5611 + dx_propagate_tag(nd, inode);
5613 res = d_materialise_unique(dentry, inode);
5615 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/inode.c
5616 --- linux-2.6.32.1/fs/nfs/inode.c 2009-12-03 20:02:52.000000000 +0100
5617 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/inode.c 2009-12-03 20:04:56.000000000 +0100
5619 #include <linux/vfs.h>
5620 #include <linux/inet.h>
5621 #include <linux/nfs_xdr.h>
5622 +#include <linux/vs_tag.h>
5624 #include <asm/system.h>
5625 #include <asm/uaccess.h>
5626 @@ -279,6 +280,8 @@ nfs_fhget(struct super_block *sb, struct
5627 if (inode->i_state & I_NEW) {
5628 struct nfs_inode *nfsi = NFS_I(inode);
5629 unsigned long now = jiffies;
5633 /* We set i_ino for the few things that still rely on it,
5634 * such as stat(2) */
5635 @@ -327,8 +330,8 @@ nfs_fhget(struct super_block *sb, struct
5636 nfsi->change_attr = 0;
5639 - inode->i_uid = -2;
5640 - inode->i_gid = -2;
5643 inode->i_blocks = 0;
5644 memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
5646 @@ -365,13 +368,13 @@ nfs_fhget(struct super_block *sb, struct
5647 else if (nfs_server_capable(inode, NFS_CAP_NLINK))
5648 nfsi->cache_validity |= NFS_INO_INVALID_ATTR;
5649 if (fattr->valid & NFS_ATTR_FATTR_OWNER)
5650 - inode->i_uid = fattr->uid;
5652 else if (nfs_server_capable(inode, NFS_CAP_OWNER))
5653 nfsi->cache_validity |= NFS_INO_INVALID_ATTR
5654 | NFS_INO_INVALID_ACCESS
5655 | NFS_INO_INVALID_ACL;
5656 if (fattr->valid & NFS_ATTR_FATTR_GROUP)
5657 - inode->i_gid = fattr->gid;
5659 else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP))
5660 nfsi->cache_validity |= NFS_INO_INVALID_ATTR
5661 | NFS_INO_INVALID_ACCESS
5662 @@ -384,6 +387,11 @@ nfs_fhget(struct super_block *sb, struct
5664 inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
5666 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
5667 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
5668 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
5669 + /* maybe fattr->xid someday */
5671 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
5672 nfsi->attrtimeo_timestamp = now;
5673 nfsi->access_cache = RB_ROOT;
5674 @@ -496,6 +504,8 @@ void nfs_setattr_update_inode(struct ino
5675 inode->i_uid = attr->ia_uid;
5676 if ((attr->ia_valid & ATTR_GID) != 0)
5677 inode->i_gid = attr->ia_gid;
5678 + if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
5679 + inode->i_tag = attr->ia_tag;
5680 NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5681 spin_unlock(&inode->i_lock);
5683 @@ -906,6 +916,9 @@ static int nfs_check_inode_attributes(st
5684 struct nfs_inode *nfsi = NFS_I(inode);
5685 loff_t cur_size, new_isize;
5686 unsigned long invalid = 0;
5692 /* Has the inode gone and changed behind our back? */
5693 @@ -929,13 +942,18 @@ static int nfs_check_inode_attributes(st
5694 invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
5697 + uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5698 + gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5699 + tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5701 /* Have any file permissions changed? */
5702 if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
5703 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5704 - if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && inode->i_uid != fattr->uid)
5705 + if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && uid != fattr->uid)
5706 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5707 - if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && inode->i_gid != fattr->gid)
5708 + if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && gid != fattr->gid)
5709 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5710 + /* maybe check for tag too? */
5712 /* Has the link count changed? */
5713 if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink)
5714 @@ -1150,6 +1168,9 @@ static int nfs_update_inode(struct inode
5715 unsigned long invalid = 0;
5716 unsigned long now = jiffies;
5717 unsigned long save_cache_validity;
5722 dfprintk(VFS, "NFS: %s(%s/%ld ct=%d info=0x%x)\n",
5723 __func__, inode->i_sb->s_id, inode->i_ino,
5724 @@ -1252,6 +1273,9 @@ static int nfs_update_inode(struct inode
5725 | NFS_INO_REVAL_PAGECACHE
5726 | NFS_INO_REVAL_FORCED);
5728 + uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5729 + gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5730 + tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5732 if (fattr->valid & NFS_ATTR_FATTR_ATIME)
5733 memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
5734 @@ -1271,9 +1295,9 @@ static int nfs_update_inode(struct inode
5735 | NFS_INO_REVAL_FORCED);
5737 if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
5738 - if (inode->i_uid != fattr->uid) {
5739 + if (uid != fattr->uid) {
5740 invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5741 - inode->i_uid = fattr->uid;
5744 } else if (server->caps & NFS_CAP_OWNER)
5745 invalid |= save_cache_validity & (NFS_INO_INVALID_ATTR
5746 @@ -1282,9 +1306,9 @@ static int nfs_update_inode(struct inode
5747 | NFS_INO_REVAL_FORCED);
5749 if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
5750 - if (inode->i_gid != fattr->gid) {
5751 + if (gid != fattr->gid) {
5752 invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5753 - inode->i_gid = fattr->gid;
5756 } else if (server->caps & NFS_CAP_OWNER_GROUP)
5757 invalid |= save_cache_validity & (NFS_INO_INVALID_ATTR
5758 @@ -1292,6 +1316,10 @@ static int nfs_update_inode(struct inode
5759 | NFS_INO_INVALID_ACL
5760 | NFS_INO_REVAL_FORCED);
5762 + inode->i_uid = uid;
5763 + inode->i_gid = gid;
5764 + inode->i_tag = tag;
5766 if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
5767 if (inode->i_nlink != fattr->nlink) {
5768 invalid |= NFS_INO_INVALID_ATTR;
5769 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/nfs3xdr.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/nfs3xdr.c
5770 --- linux-2.6.32.1/fs/nfs/nfs3xdr.c 2009-12-03 20:02:52.000000000 +0100
5771 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/nfs3xdr.c 2009-12-03 20:04:56.000000000 +0100
5773 #include <linux/nfs3.h>
5774 #include <linux/nfs_fs.h>
5775 #include <linux/nfsacl.h>
5776 +#include <linux/vs_tag.h>
5777 #include "internal.h"
5779 #define NFSDBG_FACILITY NFSDBG_XDR
5780 @@ -176,7 +177,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_f
5783 static inline __be32 *
5784 -xdr_encode_sattr(__be32 *p, struct iattr *attr)
5785 +xdr_encode_sattr(__be32 *p, struct iattr *attr, int tag)
5787 if (attr->ia_valid & ATTR_MODE) {
5789 @@ -184,15 +185,17 @@ xdr_encode_sattr(__be32 *p, struct iattr
5793 - if (attr->ia_valid & ATTR_UID) {
5794 + if (attr->ia_valid & ATTR_UID ||
5795 + (tag && (attr->ia_valid & ATTR_TAG))) {
5797 - *p++ = htonl(attr->ia_uid);
5798 + *p++ = htonl(TAGINO_UID(tag, attr->ia_uid, attr->ia_tag));
5802 - if (attr->ia_valid & ATTR_GID) {
5803 + if (attr->ia_valid & ATTR_GID ||
5804 + (tag && (attr->ia_valid & ATTR_TAG))) {
5806 - *p++ = htonl(attr->ia_gid);
5807 + *p++ = htonl(TAGINO_GID(tag, attr->ia_gid, attr->ia_tag));
5811 @@ -279,7 +282,8 @@ static int
5812 nfs3_xdr_sattrargs(struct rpc_rqst *req, __be32 *p, struct nfs3_sattrargs *args)
5814 p = xdr_encode_fhandle(p, args->fh);
5815 - p = xdr_encode_sattr(p, args->sattr);
5816 + p = xdr_encode_sattr(p, args->sattr,
5817 + req->rq_task->tk_client->cl_tag);
5818 *p++ = htonl(args->guard);
5820 p = xdr_encode_time3(p, &args->guardtime);
5821 @@ -384,7 +388,8 @@ nfs3_xdr_createargs(struct rpc_rqst *req
5822 *p++ = args->verifier[0];
5823 *p++ = args->verifier[1];
5825 - p = xdr_encode_sattr(p, args->sattr);
5826 + p = xdr_encode_sattr(p, args->sattr,
5827 + req->rq_task->tk_client->cl_tag);
5829 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5831 @@ -398,7 +403,8 @@ nfs3_xdr_mkdirargs(struct rpc_rqst *req,
5833 p = xdr_encode_fhandle(p, args->fh);
5834 p = xdr_encode_array(p, args->name, args->len);
5835 - p = xdr_encode_sattr(p, args->sattr);
5836 + p = xdr_encode_sattr(p, args->sattr,
5837 + req->rq_task->tk_client->cl_tag);
5838 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5841 @@ -411,7 +417,8 @@ nfs3_xdr_symlinkargs(struct rpc_rqst *re
5843 p = xdr_encode_fhandle(p, args->fromfh);
5844 p = xdr_encode_array(p, args->fromname, args->fromlen);
5845 - p = xdr_encode_sattr(p, args->sattr);
5846 + p = xdr_encode_sattr(p, args->sattr,
5847 + req->rq_task->tk_client->cl_tag);
5848 *p++ = htonl(args->pathlen);
5849 req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5851 @@ -429,7 +436,8 @@ nfs3_xdr_mknodargs(struct rpc_rqst *req,
5852 p = xdr_encode_fhandle(p, args->fh);
5853 p = xdr_encode_array(p, args->name, args->len);
5854 *p++ = htonl(args->type);
5855 - p = xdr_encode_sattr(p, args->sattr);
5856 + p = xdr_encode_sattr(p, args->sattr,
5857 + req->rq_task->tk_client->cl_tag);
5858 if (args->type == NF3CHR || args->type == NF3BLK) {
5859 *p++ = htonl(MAJOR(args->rdev));
5860 *p++ = htonl(MINOR(args->rdev));
5861 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/nfsroot.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/nfsroot.c
5862 --- linux-2.6.32.1/fs/nfs/nfsroot.c 2009-09-10 15:26:23.000000000 +0200
5863 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/nfsroot.c 2009-12-03 20:04:56.000000000 +0100
5864 @@ -122,12 +122,12 @@ static int mount_port __initdata = 0; /
5866 /* Options that take integer arguments */
5867 Opt_port, Opt_rsize, Opt_wsize, Opt_timeo, Opt_retrans, Opt_acregmin,
5868 - Opt_acregmax, Opt_acdirmin, Opt_acdirmax,
5869 + Opt_acregmax, Opt_acdirmin, Opt_acdirmax, Opt_tagid,
5870 /* Options that take no arguments */
5871 Opt_soft, Opt_hard, Opt_intr,
5872 Opt_nointr, Opt_posix, Opt_noposix, Opt_cto, Opt_nocto, Opt_ac,
5873 Opt_noac, Opt_lock, Opt_nolock, Opt_v2, Opt_v3, Opt_udp, Opt_tcp,
5874 - Opt_acl, Opt_noacl,
5875 + Opt_acl, Opt_noacl, Opt_tag, Opt_notag,
5879 @@ -164,6 +164,9 @@ static const match_table_t tokens __init
5882 {Opt_noacl, "noacl"},
5884 + {Opt_notag, "notag"},
5885 + {Opt_tagid, "tagid=%u"},
5889 @@ -275,6 +278,20 @@ static int __init root_nfs_parse(char *n
5891 nfs_data.flags |= NFS_MOUNT_NOACL;
5893 +#ifndef CONFIG_TAGGING_NONE
5895 + nfs_data.flags |= NFS_MOUNT_TAGGED;
5898 + nfs_data.flags &= ~NFS_MOUNT_TAGGED;
5901 +#ifdef CONFIG_PROPAGATE
5904 + nfs_data.flags |= NFS_MOUNT_TAGGED;
5908 printk(KERN_WARNING "Root-NFS: unknown "
5910 diff -NurpP --minimal linux-2.6.32.1/fs/nfs/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/super.c
5911 --- linux-2.6.32.1/fs/nfs/super.c 2009-12-03 20:02:52.000000000 +0100
5912 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfs/super.c 2009-12-03 20:04:56.000000000 +0100
5914 #include <linux/nfs_xdr.h>
5915 #include <linux/magic.h>
5916 #include <linux/parser.h>
5917 +#include <linux/vs_tag.h>
5919 #include <asm/system.h>
5920 #include <asm/uaccess.h>
5921 @@ -551,6 +552,7 @@ static void nfs_show_mount_options(struc
5922 { NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
5923 { NFS_MOUNT_UNSHARED, ",nosharecache", "" },
5924 { NFS_MOUNT_NORESVPORT, ",noresvport", "" },
5925 + { NFS_MOUNT_TAGGED, ",tag", "" },
5928 const struct proc_nfs_info *nfs_infop;
5929 diff -NurpP --minimal linux-2.6.32.1/fs/nfsd/auth.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/auth.c
5930 --- linux-2.6.32.1/fs/nfsd/auth.c 2009-12-03 20:02:52.000000000 +0100
5931 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/auth.c 2009-12-03 20:04:56.000000000 +0100
5933 #include <linux/sunrpc/svcauth.h>
5934 #include <linux/nfsd/nfsd.h>
5935 #include <linux/nfsd/export.h>
5936 +#include <linux/vs_tag.h>
5939 int nfsexp_flags(struct svc_rqst *rqstp, struct svc_export *exp)
5940 @@ -44,6 +45,9 @@ int nfsd_setuser(struct svc_rqst *rqstp,
5942 new->fsuid = rqstp->rq_cred.cr_uid;
5943 new->fsgid = rqstp->rq_cred.cr_gid;
5944 + /* FIXME: this desperately needs a tag :)
5945 + new->xid = (xid_t)INOTAG_TAG(DX_TAG_NFSD, cred.cr_uid, cred.cr_gid, 0);
5948 rqgi = rqstp->rq_cred.cr_group_info;
5950 diff -NurpP --minimal linux-2.6.32.1/fs/nfsd/nfs3xdr.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfs3xdr.c
5951 --- linux-2.6.32.1/fs/nfsd/nfs3xdr.c 2009-12-03 20:02:52.000000000 +0100
5952 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfs3xdr.c 2009-12-03 20:04:56.000000000 +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-2.6.32.1/fs/nfsd/nfs4xdr.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfs4xdr.c
6004 --- linux-2.6.32.1/fs/nfsd/nfs4xdr.c 2009-12-03 20:02:52.000000000 +0100
6005 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfs4xdr.c 2009-12-03 20:04:56.000000000 +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 @@ -2050,14 +2051,18 @@ 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-2.6.32.1/fs/nfsd/nfsxdr.c linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfsxdr.c
6036 --- linux-2.6.32.1/fs/nfsd/nfsxdr.c 2008-12-25 00:26:37.000000000 +0100
6037 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/nfsd/nfsxdr.c 2009-12-03 20:04:56.000000000 +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-2.6.32.1/fs/ocfs2/dlm/dlmfs.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlm/dlmfs.c
6087 --- linux-2.6.32.1/fs/ocfs2/dlm/dlmfs.c 2009-12-03 20:02:53.000000000 +0100
6088 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlm/dlmfs.c 2009-12-03 20:04:56.000000000 +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 @@ -342,6 +343,7 @@ 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 @@ -367,6 +369,7 @@ 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-2.6.32.1/fs/ocfs2/dlmglue.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlmglue.c
6114 --- linux-2.6.32.1/fs/ocfs2/dlmglue.c 2009-12-03 20:02:53.000000000 +0100
6115 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlmglue.c 2009-12-03 20:04:56.000000000 +0100
6116 @@ -1991,6 +1991,7 @@ 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 @@ -2045,6 +2046,7 @@ 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-2.6.32.1/fs/ocfs2/dlmglue.h linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlmglue.h
6133 --- linux-2.6.32.1/fs/ocfs2/dlmglue.h 2009-12-03 20:02:53.000000000 +0100
6134 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/dlmglue.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/fs/ocfs2/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/file.c
6146 --- linux-2.6.32.1/fs/ocfs2/file.c 2009-12-03 20:02:53.000000000 +0100
6147 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/file.c 2009-12-03 20:04:56.000000000 +0100
6148 @@ -960,13 +960,15 @@ 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 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/inode.c
6166 --- linux-2.6.32.1/fs/ocfs2/inode.c 2009-12-03 20:02:53.000000000 +0100
6167 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/inode.c 2009-12-03 20:04:56.000000000 +0100
6169 #include <linux/highmem.h>
6170 #include <linux/pagemap.h>
6171 #include <linux/quotaops.h>
6172 +#include <linux/vs_tag.h>
6174 #include <asm/byteorder.h>
6176 @@ -79,11 +80,13 @@ void ocfs2_set_inode_flags(struct inode
6178 unsigned int flags = OCFS2_I(inode)->ip_attr;
6180 - inode->i_flags &= ~(S_IMMUTABLE |
6181 + inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
6182 S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
6184 if (flags & OCFS2_IMMUTABLE_FL)
6185 inode->i_flags |= S_IMMUTABLE;
6186 + if (flags & OCFS2_IXUNLINK_FL)
6187 + inode->i_flags |= S_IXUNLINK;
6189 if (flags & OCFS2_SYNC_FL)
6190 inode->i_flags |= S_SYNC;
6191 @@ -93,25 +96,44 @@ void ocfs2_set_inode_flags(struct inode
6192 inode->i_flags |= S_NOATIME;
6193 if (flags & OCFS2_DIRSYNC_FL)
6194 inode->i_flags |= S_DIRSYNC;
6196 + inode->i_vflags &= ~(V_BARRIER | V_COW);
6198 + if (flags & OCFS2_BARRIER_FL)
6199 + inode->i_vflags |= V_BARRIER;
6200 + if (flags & OCFS2_COW_FL)
6201 + inode->i_vflags |= V_COW;
6204 /* Propagate flags from i_flags to OCFS2_I(inode)->ip_attr */
6205 void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi)
6207 unsigned int flags = oi->vfs_inode.i_flags;
6208 + unsigned int vflags = oi->vfs_inode.i_vflags;
6210 + oi->ip_attr &= ~(OCFS2_SYNC_FL | OCFS2_APPEND_FL |
6211 + OCFS2_IMMUTABLE_FL | OCFS2_IXUNLINK_FL |
6212 + OCFS2_NOATIME_FL | OCFS2_DIRSYNC_FL |
6213 + OCFS2_BARRIER_FL | OCFS2_COW_FL);
6215 + if (flags & S_IMMUTABLE)
6216 + oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6217 + if (flags & S_IXUNLINK)
6218 + oi->ip_attr |= OCFS2_IXUNLINK_FL;
6220 - oi->ip_attr &= ~(OCFS2_SYNC_FL|OCFS2_APPEND_FL|
6221 - OCFS2_IMMUTABLE_FL|OCFS2_NOATIME_FL|OCFS2_DIRSYNC_FL);
6223 oi->ip_attr |= OCFS2_SYNC_FL;
6224 if (flags & S_APPEND)
6225 oi->ip_attr |= OCFS2_APPEND_FL;
6226 - if (flags & S_IMMUTABLE)
6227 - oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6228 if (flags & S_NOATIME)
6229 oi->ip_attr |= OCFS2_NOATIME_FL;
6230 if (flags & S_DIRSYNC)
6231 oi->ip_attr |= OCFS2_DIRSYNC_FL;
6233 + if (vflags & V_BARRIER)
6234 + oi->ip_attr |= OCFS2_BARRIER_FL;
6235 + if (vflags & V_COW)
6236 + oi->ip_attr |= OCFS2_COW_FL;
6239 struct inode *ocfs2_ilookup(struct super_block *sb, u64 blkno)
6240 @@ -246,6 +268,8 @@ void ocfs2_populate_inode(struct inode *
6241 struct super_block *sb;
6242 struct ocfs2_super *osb;
6247 mlog_entry("(0x%p, size:%llu)\n", inode,
6248 (unsigned long long)le64_to_cpu(fe->i_size));
6249 @@ -277,8 +301,12 @@ void ocfs2_populate_inode(struct inode *
6250 inode->i_generation = le32_to_cpu(fe->i_generation);
6251 inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
6252 inode->i_mode = le16_to_cpu(fe->i_mode);
6253 - inode->i_uid = le32_to_cpu(fe->i_uid);
6254 - inode->i_gid = le32_to_cpu(fe->i_gid);
6255 + uid = le32_to_cpu(fe->i_uid);
6256 + gid = le32_to_cpu(fe->i_gid);
6257 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6258 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6259 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6260 + /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6262 /* Fast symlinks will have i_size but no allocated clusters. */
6263 if (S_ISLNK(inode->i_mode) && !fe->i_clusters)
6264 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/inode.h linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/inode.h
6265 --- linux-2.6.32.1/fs/ocfs2/inode.h 2009-12-03 20:02:53.000000000 +0100
6266 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/inode.h 2009-12-03 20:04:56.000000000 +0100
6267 @@ -150,6 +150,7 @@ struct buffer_head *ocfs2_bread(struct i
6269 void ocfs2_set_inode_flags(struct inode *inode);
6270 void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi);
6271 +int ocfs2_sync_flags(struct inode *inode, int, int);
6273 static inline blkcnt_t ocfs2_inode_sector_count(struct inode *inode)
6275 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ioctl.c
6276 --- linux-2.6.32.1/fs/ocfs2/ioctl.c 2009-12-03 20:02:53.000000000 +0100
6277 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ioctl.c 2009-12-03 20:04:56.000000000 +0100
6278 @@ -42,7 +42,41 @@ static int ocfs2_get_inode_attr(struct i
6282 -static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6283 +int ocfs2_sync_flags(struct inode *inode, int flags, int vflags)
6285 + struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
6286 + struct buffer_head *bh = NULL;
6287 + handle_t *handle = NULL;
6290 + status = ocfs2_inode_lock(inode, &bh, 1);
6292 + mlog_errno(status);
6295 + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6296 + if (IS_ERR(handle)) {
6297 + status = PTR_ERR(handle);
6298 + mlog_errno(status);
6302 + inode->i_flags = flags;
6303 + inode->i_vflags = vflags;
6304 + ocfs2_get_inode_flags(OCFS2_I(inode));
6306 + status = ocfs2_mark_inode_dirty(handle, inode, bh);
6308 + mlog_errno(status);
6310 + ocfs2_commit_trans(osb, handle);
6312 + ocfs2_inode_unlock(inode, 1);
6317 +int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6320 struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6321 @@ -67,6 +101,11 @@ static int ocfs2_set_inode_attr(struct i
6322 if (!S_ISDIR(inode->i_mode))
6323 flags &= ~OCFS2_DIRSYNC_FL;
6325 + if (IS_BARRIER(inode)) {
6326 + vxwprintk_task(1, "messing with the barrier.");
6330 handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6331 if (IS_ERR(handle)) {
6332 status = PTR_ERR(handle);
6333 @@ -108,6 +147,7 @@ bail:
6338 long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
6340 struct inode *inode = filp->f_path.dentry->d_inode;
6341 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/namei.c
6342 --- linux-2.6.32.1/fs/ocfs2/namei.c 2009-12-03 20:02:53.000000000 +0100
6343 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/namei.c 2009-12-03 20:04:56.000000000 +0100
6345 #include <linux/slab.h>
6346 #include <linux/highmem.h>
6347 #include <linux/quotaops.h>
6348 +#include <linux/vs_tag.h>
6350 #define MLOG_MASK_PREFIX ML_NAMEI
6351 #include <cluster/masklog.h>
6352 @@ -481,6 +482,7 @@ static int ocfs2_mknod_locked(struct ocf
6360 @@ -524,8 +526,11 @@ static int ocfs2_mknod_locked(struct ocf
6361 fe->i_blkno = cpu_to_le64(fe_blkno);
6362 fe->i_suballoc_bit = cpu_to_le16(suballoc_bit);
6363 fe->i_suballoc_slot = cpu_to_le16(inode_ac->ac_alloc_slot);
6364 - fe->i_uid = cpu_to_le32(inode->i_uid);
6365 - fe->i_gid = cpu_to_le32(inode->i_gid);
6367 + tag = dx_current_fstag(osb->sb);
6368 + fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode), inode->i_uid, tag));
6369 + fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode), inode->i_gid, tag));
6370 + inode->i_tag = tag;
6371 fe->i_mode = cpu_to_le16(inode->i_mode);
6372 if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode))
6373 fe->id1.dev1.i_rdev = cpu_to_le64(huge_encode_dev(dev));
6374 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/ocfs2_fs.h linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ocfs2_fs.h
6375 --- linux-2.6.32.1/fs/ocfs2/ocfs2_fs.h 2009-12-03 20:02:53.000000000 +0100
6376 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ocfs2_fs.h 2009-12-03 20:04:56.000000000 +0100
6377 @@ -231,18 +231,23 @@
6378 #define OCFS2_HAS_REFCOUNT_FL (0x0010)
6380 /* Inode attributes, keep in sync with EXT2 */
6381 -#define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
6382 -#define OCFS2_UNRM_FL (0x00000002) /* Undelete */
6383 -#define OCFS2_COMPR_FL (0x00000004) /* Compress file */
6384 -#define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
6385 -#define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
6386 -#define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
6387 -#define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
6388 -#define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
6389 -#define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
6390 +#define OCFS2_SECRM_FL FS_SECRM_FL /* Secure deletion */
6391 +#define OCFS2_UNRM_FL FS_UNRM_FL /* Undelete */
6392 +#define OCFS2_COMPR_FL FS_COMPR_FL /* Compress file */
6393 +#define OCFS2_SYNC_FL FS_SYNC_FL /* Synchronous updates */
6394 +#define OCFS2_IMMUTABLE_FL FS_IMMUTABLE_FL /* Immutable file */
6395 +#define OCFS2_APPEND_FL FS_APPEND_FL /* writes to file may only append */
6396 +#define OCFS2_NODUMP_FL FS_NODUMP_FL /* do not dump file */
6397 +#define OCFS2_NOATIME_FL FS_NOATIME_FL /* do not update atime */
6399 -#define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
6400 -#define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
6401 +#define OCFS2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
6402 +#define OCFS2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
6404 +#define OCFS2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
6405 +#define OCFS2_COW_FL FS_COW_FL /* Copy on Write marker */
6407 +#define OCFS2_FL_VISIBLE (0x010300FF) /* User visible flags */
6408 +#define OCFS2_FL_MODIFIABLE (0x010300FF) /* User modifiable flags */
6411 * Extent record flags (e_node.leaf.flags)
6412 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/ocfs2.h linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ocfs2.h
6413 --- linux-2.6.32.1/fs/ocfs2/ocfs2.h 2009-12-03 20:02:53.000000000 +0100
6414 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/ocfs2.h 2009-12-03 20:04:56.000000000 +0100
6415 @@ -248,6 +248,7 @@ enum ocfs2_mount_options
6416 OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* POSIX access control lists */
6417 OCFS2_MOUNT_USRQUOTA = 1 << 9, /* We support user quotas */
6418 OCFS2_MOUNT_GRPQUOTA = 1 << 10, /* We support group quotas */
6419 + OCFS2_MOUNT_TAGGED = 1 << 11, /* use tagging */
6422 #define OCFS2_OSB_SOFT_RO 0x0001
6423 diff -NurpP --minimal linux-2.6.32.1/fs/ocfs2/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/super.c
6424 --- linux-2.6.32.1/fs/ocfs2/super.c 2009-12-03 20:02:53.000000000 +0100
6425 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/ocfs2/super.c 2009-12-03 20:04:56.000000000 +0100
6426 @@ -173,6 +173,7 @@ enum {
6430 + Opt_tag, Opt_notag, Opt_tagid,
6434 @@ -199,6 +200,9 @@ static const match_table_t tokens = {
6435 {Opt_noacl, "noacl"},
6436 {Opt_usrquota, "usrquota"},
6437 {Opt_grpquota, "grpquota"},
6439 + {Opt_notag, "notag"},
6440 + {Opt_tagid, "tagid=%u"},
6444 @@ -605,6 +609,13 @@ static int ocfs2_remount(struct super_bl
6448 + if ((osb->s_mount_opt & OCFS2_MOUNT_TAGGED) !=
6449 + (parsed_options.mount_opt & OCFS2_MOUNT_TAGGED)) {
6451 + mlog(ML_ERROR, "Cannot change tagging on remount\n");
6455 if ((osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) !=
6456 (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
6458 @@ -1148,6 +1159,9 @@ static int ocfs2_fill_super(struct super
6460 ocfs2_complete_mount_recovery(osb);
6462 + if (osb->s_mount_opt & OCFS2_MOUNT_TAGGED)
6463 + sb->s_flags |= MS_TAGGED;
6465 if (ocfs2_mount_local(osb))
6466 snprintf(nodestr, sizeof(nodestr), "local");
6468 @@ -1426,6 +1440,20 @@ static int ocfs2_parse_options(struct su
6469 printk(KERN_INFO "ocfs2 (no)acl options not supported\n");
6472 +#ifndef CONFIG_TAGGING_NONE
6474 + mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6477 + mopt->mount_opt &= ~OCFS2_MOUNT_TAGGED;
6480 +#ifdef CONFIG_PROPAGATE
6483 + mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6488 "Unrecognized mount option \"%s\" "
6489 diff -NurpP --minimal linux-2.6.32.1/fs/open.c linux-2.6.32.1-vs2.3.0.36.27/fs/open.c
6490 --- linux-2.6.32.1/fs/open.c 2009-12-03 20:02:53.000000000 +0100
6491 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/open.c 2009-12-03 20:04:56.000000000 +0100
6493 #include <linux/audit.h>
6494 #include <linux/falloc.h>
6495 #include <linux/fs_struct.h>
6496 +#include <linux/vs_base.h>
6497 +#include <linux/vs_limit.h>
6498 +#include <linux/vs_tag.h>
6499 +#include <linux/vs_cowbl.h>
6501 int vfs_statfs(struct dentry *dentry, struct kstatfs *buf)
6503 int retval = -ENODEV;
6506 + struct super_block *sb = dentry->d_sb;
6509 - if (dentry->d_sb->s_op->statfs) {
6510 + if (sb->s_op->statfs) {
6511 memset(buf, 0, sizeof(*buf));
6512 retval = security_sb_statfs(dentry);
6515 - retval = dentry->d_sb->s_op->statfs(dentry, buf);
6516 + retval = sb->s_op->statfs(dentry, buf);
6517 if (retval == 0 && buf->f_frsize == 0)
6518 buf->f_frsize = buf->f_bsize;
6520 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
6521 + vx_vsi_statfs(sb, buf);
6525 @@ -640,6 +648,10 @@ SYSCALL_DEFINE3(fchmodat, int, dfd, cons
6526 error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
6530 + error = cow_check_and_break(&path);
6532 + goto dput_and_out;
6533 inode = path.dentry->d_inode;
6535 error = mnt_want_write(path.mnt);
6536 @@ -673,11 +685,11 @@ static int chown_common(struct dentry *
6537 newattrs.ia_valid = ATTR_CTIME;
6538 if (user != (uid_t) -1) {
6539 newattrs.ia_valid |= ATTR_UID;
6540 - newattrs.ia_uid = user;
6541 + newattrs.ia_uid = dx_map_uid(user);
6543 if (group != (gid_t) -1) {
6544 newattrs.ia_valid |= ATTR_GID;
6545 - newattrs.ia_gid = group;
6546 + newattrs.ia_gid = dx_map_gid(group);
6548 if (!S_ISDIR(inode->i_mode))
6549 newattrs.ia_valid |=
6550 @@ -700,7 +712,11 @@ SYSCALL_DEFINE3(chown, const char __user
6551 error = mnt_want_write(path.mnt);
6554 - error = chown_common(path.dentry, user, group);
6555 +#ifdef CONFIG_VSERVER_COWBL
6556 + error = cow_check_and_break(&path);
6559 + error = chown_common(path.dentry, user, group);
6560 mnt_drop_write(path.mnt);
6563 @@ -725,7 +741,11 @@ SYSCALL_DEFINE5(fchownat, int, dfd, cons
6564 error = mnt_want_write(path.mnt);
6567 - error = chown_common(path.dentry, user, group);
6568 +#ifdef CONFIG_VSERVER_COWBL
6569 + error = cow_check_and_break(&path);
6572 + error = chown_common(path.dentry, user, group);
6573 mnt_drop_write(path.mnt);
6576 @@ -744,7 +764,11 @@ SYSCALL_DEFINE3(lchown, const char __use
6577 error = mnt_want_write(path.mnt);
6580 - error = chown_common(path.dentry, user, group);
6581 +#ifdef CONFIG_VSERVER_COWBL
6582 + error = cow_check_and_break(&path);
6585 + error = chown_common(path.dentry, user, group);
6586 mnt_drop_write(path.mnt);
6589 @@ -990,6 +1014,7 @@ static void __put_unused_fd(struct files
6590 __FD_CLR(fd, fdt->open_fds);
6591 if (fd < files->next_fd)
6592 files->next_fd = fd;
6593 + vx_openfd_dec(fd);
6596 void put_unused_fd(unsigned int fd)
6597 diff -NurpP --minimal linux-2.6.32.1/fs/proc/array.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/array.c
6598 --- linux-2.6.32.1/fs/proc/array.c 2009-12-03 20:02:53.000000000 +0100
6599 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/array.c 2009-12-03 20:04:56.000000000 +0100
6601 #include <linux/ptrace.h>
6602 #include <linux/tracehook.h>
6603 #include <linux/swapops.h>
6604 +#include <linux/vs_context.h>
6605 +#include <linux/vs_network.h>
6607 #include <asm/pgtable.h>
6608 #include <asm/processor.h>
6609 @@ -139,8 +141,9 @@ static const char *task_state_array[] =
6610 "D (disk sleep)", /* 2 */
6611 "T (stopped)", /* 4 */
6612 "T (tracing stop)", /* 8 */
6613 - "Z (zombie)", /* 16 */
6614 - "X (dead)" /* 32 */
6615 + "H (on hold)", /* 16 */
6616 + "Z (zombie)", /* 32 */
6617 + "X (dead)", /* 64 */
6620 static inline const char *get_task_state(struct task_struct *tsk)
6621 @@ -167,6 +170,9 @@ static inline void task_state(struct seq
6623 ppid = pid_alive(p) ?
6624 task_tgid_nr_ns(rcu_dereference(p->real_parent), ns) : 0;
6625 + if (unlikely(vx_current_initpid(p->pid)))
6630 struct task_struct *tracer = tracehook_tracer_task(p);
6631 @@ -282,7 +288,7 @@ static inline void task_sig(struct seq_f
6634 static void render_cap_t(struct seq_file *m, const char *header,
6636 + struct vx_info *vxi, kernel_cap_t *a)
6640 @@ -307,10 +313,11 @@ static inline void task_cap(struct seq_f
6641 cap_bset = cred->cap_bset;
6644 - render_cap_t(m, "CapInh:\t", &cap_inheritable);
6645 - render_cap_t(m, "CapPrm:\t", &cap_permitted);
6646 - render_cap_t(m, "CapEff:\t", &cap_effective);
6647 - render_cap_t(m, "CapBnd:\t", &cap_bset);
6648 + /* FIXME: maybe move the p->vx_info masking to __task_cred() ? */
6649 + render_cap_t(m, "CapInh:\t", p->vx_info, &cap_inheritable);
6650 + render_cap_t(m, "CapPrm:\t", p->vx_info, &cap_permitted);
6651 + render_cap_t(m, "CapEff:\t", p->vx_info, &cap_effective);
6652 + render_cap_t(m, "CapBnd:\t", p->vx_info, &cap_bset);
6655 static inline void task_context_switch_counts(struct seq_file *m,
6656 @@ -410,6 +417,42 @@ static void task_show_stack_usage(struct
6658 #endif /* CONFIG_MMU */
6660 +int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
6661 + struct pid *pid, struct task_struct *task)
6663 + seq_printf(m, "Proxy:\t%p(%c)\n"
6671 + (task->nsproxy == init_task.nsproxy ? 'I' : '-'),
6672 + atomic_read(&task->nsproxy->count),
6673 + task->nsproxy->uts_ns,
6674 + (task->nsproxy->uts_ns == init_task.nsproxy->uts_ns ? 'I' : '-'),
6675 + task->nsproxy->ipc_ns,
6676 + (task->nsproxy->ipc_ns == init_task.nsproxy->ipc_ns ? 'I' : '-'),
6677 + task->nsproxy->mnt_ns,
6678 + (task->nsproxy->mnt_ns == init_task.nsproxy->mnt_ns ? 'I' : '-'),
6679 + task->nsproxy->pid_ns,
6680 + (task->nsproxy->pid_ns == init_task.nsproxy->pid_ns ? 'I' : '-'),
6681 + task->nsproxy->net_ns,
6682 + (task->nsproxy->net_ns == init_task.nsproxy->net_ns ? 'I' : '-'));
6686 +void task_vs_id(struct seq_file *m, struct task_struct *task)
6688 + if (task_vx_flags(task, VXF_HIDE_VINFO, 0))
6691 + seq_printf(m, "VxID: %d\n", vx_task_xid(task));
6692 + seq_printf(m, "NxID: %d\n", nx_task_nid(task));
6696 int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
6697 struct pid *pid, struct task_struct *task)
6699 @@ -425,6 +468,7 @@ int proc_pid_status(struct seq_file *m,
6702 cpuset_task_status_allowed(m, task);
6703 + task_vs_id(m, task);
6704 #if defined(CONFIG_S390)
6705 task_show_regs(m, task);
6707 @@ -542,6 +586,17 @@ static int do_task_stat(struct seq_file
6708 /* convert nsec -> ticks */
6709 start_time = nsec_to_clock_t(start_time);
6711 + /* fixup start time for virt uptime */
6712 + if (vx_flags(VXF_VIRT_UPTIME, 0)) {
6713 + unsigned long long bias =
6714 + current->vx_info->cvirt.bias_clock;
6716 + if (start_time > bias)
6717 + start_time -= bias;
6722 seq_printf(m, "%d (%s) %c %d %d %d %d %d %u %lu \
6723 %lu %lu %lu %lu %lu %ld %ld %ld %ld %d 0 %llu %lu %ld %lu %lu %lu %lu %lu \
6724 %lu %lu %lu %lu %lu %lu %lu %lu %d %d %u %u %llu %lu %ld\n",
6725 diff -NurpP --minimal linux-2.6.32.1/fs/proc/base.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/base.c
6726 --- linux-2.6.32.1/fs/proc/base.c 2009-12-03 20:02:53.000000000 +0100
6727 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/base.c 2009-12-03 20:04:56.000000000 +0100
6729 #include <linux/elf.h>
6730 #include <linux/pid_namespace.h>
6731 #include <linux/fs_struct.h>
6732 +#include <linux/vs_context.h>
6733 +#include <linux/vs_network.h>
6734 #include "internal.h"
6737 @@ -1047,12 +1049,17 @@ static ssize_t oom_adjust_write(struct f
6741 - if (oom_adjust < task->signal->oom_adj && !capable(CAP_SYS_RESOURCE)) {
6742 + if (oom_adjust < task->signal->oom_adj &&
6743 + !vx_capable(CAP_SYS_RESOURCE, VXC_OOM_ADJUST)) {
6744 unlock_task_sighand(task, &flags);
6745 put_task_struct(task);
6749 + /* prevent guest processes from circumventing the oom killer */
6750 + if (vx_current_xid() && (oom_adjust == OOM_DISABLE))
6751 + oom_adjust = OOM_ADJUST_MIN;
6753 task->signal->oom_adj = oom_adjust;
6755 unlock_task_sighand(task, &flags);
6756 @@ -1092,7 +1099,7 @@ static ssize_t proc_loginuid_write(struc
6760 - if (!capable(CAP_AUDIT_CONTROL))
6761 + if (!vx_capable(CAP_AUDIT_CONTROL, VXC_AUDIT_CONTROL))
6764 if (current != pid_task(proc_pid(inode), PIDTYPE_PID))
6765 @@ -1458,6 +1465,8 @@ static struct inode *proc_pid_make_inode
6766 inode->i_gid = cred->egid;
6769 + /* procfs is xid tagged */
6770 + inode->i_tag = (tag_t)vx_task_xid(task);
6771 security_task_to_inode(task, inode);
6774 @@ -2008,6 +2017,13 @@ static struct dentry *proc_pident_lookup
6778 + /* TODO: maybe we can come up with a generic approach? */
6779 + if (task_vx_flags(task, VXF_HIDE_VINFO, 0) &&
6780 + (dentry->d_name.len == 5) &&
6781 + (!memcmp(dentry->d_name.name, "vinfo", 5) ||
6782 + !memcmp(dentry->d_name.name, "ninfo", 5)))
6786 * Yes, it does not scale. And it should not. Don't add
6787 * new entries into /proc/<tgid>/ without very good reasons.
6788 @@ -2399,7 +2415,7 @@ out_iput:
6789 static struct dentry *proc_base_lookup(struct inode *dir, struct dentry *dentry)
6791 struct dentry *error;
6792 - struct task_struct *task = get_proc_task(dir);
6793 + struct task_struct *task = get_proc_task_real(dir);
6794 const struct pid_entry *p, *last;
6796 error = ERR_PTR(-ENOENT);
6797 @@ -2489,6 +2505,9 @@ static int proc_pid_personality(struct s
6798 static const struct file_operations proc_task_operations;
6799 static const struct inode_operations proc_task_inode_operations;
6801 +extern int proc_pid_vx_info(struct task_struct *, char *);
6802 +extern int proc_pid_nx_info(struct task_struct *, char *);
6804 static const struct pid_entry tgid_base_stuff[] = {
6805 DIR("task", S_IRUGO|S_IXUGO, proc_task_inode_operations, proc_task_operations),
6806 DIR("fd", S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
6807 @@ -2547,6 +2566,8 @@ static const struct pid_entry tgid_base_
6808 #ifdef CONFIG_CGROUPS
6809 REG("cgroup", S_IRUGO, proc_cgroup_operations),
6811 + INF("vinfo", S_IRUGO, proc_pid_vx_info),
6812 + INF("ninfo", S_IRUGO, proc_pid_nx_info),
6813 INF("oom_score", S_IRUGO, proc_oom_score),
6814 REG("oom_adj", S_IRUGO|S_IWUSR, proc_oom_adjust_operations),
6815 #ifdef CONFIG_AUDITSYSCALL
6816 @@ -2562,6 +2583,7 @@ static const struct pid_entry tgid_base_
6817 #ifdef CONFIG_TASK_IO_ACCOUNTING
6818 INF("io", S_IRUGO, proc_tgid_io_accounting),
6820 + ONE("nsproxy", S_IRUGO, proc_pid_nsproxy),
6823 static int proc_tgid_base_readdir(struct file * filp,
6824 @@ -2753,7 +2775,7 @@ retry:
6826 pid = find_ge_pid(iter.tgid, ns);
6828 - iter.tgid = pid_nr_ns(pid, ns);
6829 + iter.tgid = pid_unmapped_nr_ns(pid, ns);
6830 iter.task = pid_task(pid, PIDTYPE_PID);
6831 /* What we to know is if the pid we have find is the
6832 * pid of a thread_group_leader. Testing for task
6833 @@ -2783,7 +2805,7 @@ static int proc_pid_fill_cache(struct fi
6834 struct tgid_iter iter)
6836 char name[PROC_NUMBUF];
6837 - int len = snprintf(name, sizeof(name), "%d", iter.tgid);
6838 + int len = snprintf(name, sizeof(name), "%d", vx_map_tgid(iter.tgid));
6839 return proc_fill_cache(filp, dirent, filldir, name, len,
6840 proc_pid_instantiate, iter.task, NULL);
6842 @@ -2792,7 +2814,7 @@ static int proc_pid_fill_cache(struct fi
6843 int proc_pid_readdir(struct file * filp, void * dirent, filldir_t filldir)
6845 unsigned int nr = filp->f_pos - FIRST_PROCESS_ENTRY;
6846 - struct task_struct *reaper = get_proc_task(filp->f_path.dentry->d_inode);
6847 + struct task_struct *reaper = get_proc_task_real(filp->f_path.dentry->d_inode);
6848 struct tgid_iter iter;
6849 struct pid_namespace *ns;
6851 @@ -2812,6 +2834,8 @@ int proc_pid_readdir(struct file * filp,
6853 iter.tgid += 1, iter = next_tgid(ns, iter)) {
6854 filp->f_pos = iter.tgid + TGID_OFFSET;
6855 + if (!vx_proc_task_visible(iter.task))
6857 if (proc_pid_fill_cache(filp, dirent, filldir, iter) < 0) {
6858 put_task_struct(iter.task);
6860 @@ -2958,6 +2982,8 @@ static struct dentry *proc_task_lookup(s
6861 tid = name_to_int(dentry);
6864 + if (vx_current_initpid(tid))
6867 ns = dentry->d_sb->s_fs_info;
6869 diff -NurpP --minimal linux-2.6.32.1/fs/proc/generic.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/generic.c
6870 --- linux-2.6.32.1/fs/proc/generic.c 2009-06-11 17:13:07.000000000 +0200
6871 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/generic.c 2009-12-03 20:04:56.000000000 +0100
6873 #include <linux/bitops.h>
6874 #include <linux/spinlock.h>
6875 #include <linux/completion.h>
6876 +#include <linux/vserver/inode.h>
6877 #include <asm/uaccess.h>
6879 #include "internal.h"
6880 @@ -425,6 +426,8 @@ struct dentry *proc_lookup_de(struct pro
6881 for (de = de->subdir; de ; de = de->next) {
6882 if (de->namelen != dentry->d_name.len)
6884 + if (!vx_hide_check(0, de->vx_flags))
6886 if (!memcmp(dentry->d_name.name, de->name, de->namelen)) {
6889 @@ -433,6 +436,8 @@ struct dentry *proc_lookup_de(struct pro
6890 spin_unlock(&proc_subdir_lock);
6892 inode = proc_get_inode(dir->i_sb, ino, de);
6893 + /* generic proc entries belong to the host */
6898 @@ -510,6 +515,8 @@ int proc_readdir_de(struct proc_dir_entr
6900 /* filldir passes info to user space */
6902 + if (!vx_hide_check(0, de->vx_flags))
6904 spin_unlock(&proc_subdir_lock);
6905 if (filldir(dirent, de->name, de->namelen, filp->f_pos,
6906 de->low_ino, de->mode >> 12) < 0) {
6907 @@ -517,6 +524,7 @@ int proc_readdir_de(struct proc_dir_entr
6910 spin_lock(&proc_subdir_lock);
6915 @@ -631,6 +639,7 @@ static struct proc_dir_entry *__proc_cre
6917 atomic_set(&ent->count, 1);
6919 + ent->vx_flags = IATTR_PROC_DEFAULT;
6920 spin_lock_init(&ent->pde_unload_lock);
6921 ent->pde_unload_completion = NULL;
6922 INIT_LIST_HEAD(&ent->pde_openers);
6923 @@ -654,7 +663,8 @@ struct proc_dir_entry *proc_symlink(cons
6929 + ent->vx_flags = IATTR_PROC_SYMLINK;
6933 diff -NurpP --minimal linux-2.6.32.1/fs/proc/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/inode.c
6934 --- linux-2.6.32.1/fs/proc/inode.c 2009-06-11 17:13:07.000000000 +0200
6935 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/inode.c 2009-12-03 20:04:56.000000000 +0100
6936 @@ -459,6 +459,8 @@ struct inode *proc_get_inode(struct supe
6937 inode->i_uid = de->uid;
6938 inode->i_gid = de->gid;
6941 + PROC_I(inode)->vx_flags = de->vx_flags;
6943 inode->i_size = de->size;
6945 diff -NurpP --minimal linux-2.6.32.1/fs/proc/internal.h linux-2.6.32.1-vs2.3.0.36.27/fs/proc/internal.h
6946 --- linux-2.6.32.1/fs/proc/internal.h 2009-09-10 15:26:23.000000000 +0200
6947 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/internal.h 2009-12-03 20:04:56.000000000 +0100
6951 #include <linux/proc_fs.h>
6952 +#include <linux/vs_pid.h>
6954 extern struct proc_dir_entry proc_root;
6955 #ifdef CONFIG_PROC_SYSCTL
6956 @@ -51,6 +52,9 @@ extern int proc_pid_status(struct seq_fi
6957 struct pid *pid, struct task_struct *task);
6958 extern int proc_pid_statm(struct seq_file *m, struct pid_namespace *ns,
6959 struct pid *pid, struct task_struct *task);
6960 +extern int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
6961 + struct pid *pid, struct task_struct *task);
6963 extern loff_t mem_lseek(struct file *file, loff_t offset, int orig);
6965 extern const struct file_operations proc_maps_operations;
6966 @@ -70,11 +74,16 @@ static inline struct pid *proc_pid(struc
6967 return PROC_I(inode)->pid;
6970 -static inline struct task_struct *get_proc_task(struct inode *inode)
6971 +static inline struct task_struct *get_proc_task_real(struct inode *inode)
6973 return get_pid_task(proc_pid(inode), PIDTYPE_PID);
6976 +static inline struct task_struct *get_proc_task(struct inode *inode)
6978 + return vx_get_proc_task(inode, proc_pid(inode));
6981 static inline int proc_fd(struct inode *inode)
6983 return PROC_I(inode)->fd;
6984 diff -NurpP --minimal linux-2.6.32.1/fs/proc/loadavg.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/loadavg.c
6985 --- linux-2.6.32.1/fs/proc/loadavg.c 2009-09-10 15:26:23.000000000 +0200
6986 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/loadavg.c 2009-12-03 20:04:56.000000000 +0100
6989 static int loadavg_proc_show(struct seq_file *m, void *v)
6991 + unsigned long running;
6992 + unsigned int threads;
6993 unsigned long avnrun[3];
6995 get_avenrun(avnrun, FIXED_1/200, 0);
6997 + if (vx_flags(VXF_VIRT_LOAD, 0)) {
6998 + struct vx_info *vxi = current_vx_info();
7000 + running = atomic_read(&vxi->cvirt.nr_running);
7001 + threads = atomic_read(&vxi->cvirt.nr_threads);
7003 + running = nr_running();
7004 + threads = nr_threads;
7007 seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %ld/%d %d\n",
7008 LOAD_INT(avnrun[0]), LOAD_FRAC(avnrun[0]),
7009 LOAD_INT(avnrun[1]), LOAD_FRAC(avnrun[1]),
7010 LOAD_INT(avnrun[2]), LOAD_FRAC(avnrun[2]),
7011 - nr_running(), nr_threads,
7013 task_active_pid_ns(current)->last_pid);
7016 diff -NurpP --minimal linux-2.6.32.1/fs/proc/meminfo.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/meminfo.c
7017 --- linux-2.6.32.1/fs/proc/meminfo.c 2009-12-03 20:02:53.000000000 +0100
7018 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/meminfo.c 2009-12-03 20:04:56.000000000 +0100
7019 @@ -41,7 +41,7 @@ static int meminfo_proc_show(struct seq_
7021 cached = global_page_state(NR_FILE_PAGES) -
7022 total_swapcache_pages - i.bufferram;
7024 + if (cached < 0 || vx_flags(VXF_VIRT_MEM, 0))
7027 get_vmalloc_info(&vmi);
7028 diff -NurpP --minimal linux-2.6.32.1/fs/proc/root.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/root.c
7029 --- linux-2.6.32.1/fs/proc/root.c 2009-06-11 17:13:07.000000000 +0200
7030 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/root.c 2009-12-03 20:04:56.000000000 +0100
7032 #include <linux/bitops.h>
7033 #include <linux/mount.h>
7034 #include <linux/pid_namespace.h>
7035 +#include <linux/vserver/inode.h>
7037 #include "internal.h"
7039 +struct proc_dir_entry *proc_virtual;
7041 +extern void proc_vx_init(void);
7043 static int proc_test_super(struct super_block *sb, void *data)
7045 return sb->s_fs_info == data;
7046 @@ -136,6 +141,7 @@ void __init proc_root_init(void)
7048 proc_mkdir("bus", NULL);
7053 static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat
7054 @@ -203,6 +209,7 @@ struct proc_dir_entry proc_root = {
7055 .proc_iops = &proc_root_inode_operations,
7056 .proc_fops = &proc_root_operations,
7057 .parent = &proc_root,
7058 + .vx_flags = IATTR_ADMIN | IATTR_WATCH,
7061 int pid_ns_prepare_proc(struct pid_namespace *ns)
7062 diff -NurpP --minimal linux-2.6.32.1/fs/proc/uptime.c linux-2.6.32.1-vs2.3.0.36.27/fs/proc/uptime.c
7063 --- linux-2.6.32.1/fs/proc/uptime.c 2009-12-03 20:02:53.000000000 +0100
7064 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/proc/uptime.c 2009-12-03 20:04:56.000000000 +0100
7066 #include <linux/sched.h>
7067 #include <linux/seq_file.h>
7068 #include <linux/time.h>
7069 -#include <linux/kernel_stat.h>
7070 +#include <linux/vserver/cvirt.h>
7071 #include <asm/cputime.h>
7073 static int uptime_proc_show(struct seq_file *m, void *v)
7075 struct timespec uptime;
7076 struct timespec idle;
7078 - cputime_t idletime = cputime_zero;
7080 - for_each_possible_cpu(i)
7081 - idletime = cputime64_add(idletime, kstat_cpu(i).cpustat.idle);
7082 + cputime_t idletime = cputime_add(init_task.utime, init_task.stime);
7084 do_posix_clock_monotonic_gettime(&uptime);
7085 monotonic_to_bootbased(&uptime);
7086 cputime_to_timespec(idletime, &idle);
7088 + if (vx_flags(VXF_VIRT_UPTIME, 0))
7089 + vx_vsi_uptime(&uptime, &idle);
7091 seq_printf(m, "%lu.%02lu %lu.%02lu\n",
7092 (unsigned long) uptime.tv_sec,
7093 (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
7094 diff -NurpP --minimal linux-2.6.32.1/fs/quota/quota.c linux-2.6.32.1-vs2.3.0.36.27/fs/quota/quota.c
7095 --- linux-2.6.32.1/fs/quota/quota.c 2009-09-10 15:26:24.000000000 +0200
7096 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/quota/quota.c 2009-12-03 20:04:56.000000000 +0100
7098 #include <linux/capability.h>
7099 #include <linux/quotaops.h>
7100 #include <linux/types.h>
7101 +#include <linux/vs_context.h>
7103 /* Check validity of generic quotactl commands */
7104 static int generic_quotactl_valid(struct super_block *sb, int type, int cmd,
7105 @@ -83,11 +84,11 @@ static int generic_quotactl_valid(struct
7106 if (cmd == Q_GETQUOTA) {
7107 if (((type == USRQUOTA && current_euid() != id) ||
7108 (type == GRPQUOTA && !in_egroup_p(id))) &&
7109 - !capable(CAP_SYS_ADMIN))
7110 + !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7113 else if (cmd != Q_GETFMT && cmd != Q_SYNC && cmd != Q_GETINFO)
7114 - if (!capable(CAP_SYS_ADMIN))
7115 + if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7119 @@ -135,10 +136,10 @@ static int xqm_quotactl_valid(struct sup
7120 if (cmd == Q_XGETQUOTA) {
7121 if (((type == XQM_USRQUOTA && current_euid() != id) ||
7122 (type == XQM_GRPQUOTA && !in_egroup_p(id))) &&
7123 - !capable(CAP_SYS_ADMIN))
7124 + !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7126 } else if (cmd != Q_XGETQSTAT && cmd != Q_XQUOTASYNC) {
7127 - if (!capable(CAP_SYS_ADMIN))
7128 + if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7132 @@ -351,6 +352,46 @@ static int do_quotactl(struct super_bloc
7136 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7138 +#include <linux/vroot.h>
7139 +#include <linux/major.h>
7140 +#include <linux/module.h>
7141 +#include <linux/kallsyms.h>
7142 +#include <linux/vserver/debug.h>
7144 +static vroot_grb_func *vroot_get_real_bdev = NULL;
7146 +static spinlock_t vroot_grb_lock = SPIN_LOCK_UNLOCKED;
7148 +int register_vroot_grb(vroot_grb_func *func) {
7151 + spin_lock(&vroot_grb_lock);
7152 + if (!vroot_get_real_bdev) {
7153 + vroot_get_real_bdev = func;
7156 + spin_unlock(&vroot_grb_lock);
7159 +EXPORT_SYMBOL(register_vroot_grb);
7161 +int unregister_vroot_grb(vroot_grb_func *func) {
7162 + int ret = -EINVAL;
7164 + spin_lock(&vroot_grb_lock);
7165 + if (vroot_get_real_bdev) {
7166 + vroot_get_real_bdev = NULL;
7169 + spin_unlock(&vroot_grb_lock);
7172 +EXPORT_SYMBOL(unregister_vroot_grb);
7177 * look up a superblock on which quota ops will be performed
7178 * - use the name of a block device to find the superblock thereon
7179 @@ -368,6 +409,22 @@ static struct super_block *quotactl_bloc
7182 return ERR_CAST(bdev);
7183 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7184 + if (bdev && bdev->bd_inode &&
7185 + imajor(bdev->bd_inode) == VROOT_MAJOR) {
7186 + struct block_device *bdnew = (void *)-EINVAL;
7188 + if (vroot_get_real_bdev)
7189 + bdnew = vroot_get_real_bdev(bdev);
7191 + vxdprintk(VXD_CBIT(misc, 0),
7192 + "vroot_get_real_bdev not set");
7194 + if (IS_ERR(bdnew))
7195 + return ERR_PTR(PTR_ERR(bdnew));
7199 sb = get_super(bdev);
7202 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/file.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/file.c
7203 --- linux-2.6.32.1/fs/reiserfs/file.c 2009-06-11 17:13:08.000000000 +0200
7204 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/file.c 2009-12-03 20:04:56.000000000 +0100
7205 @@ -307,4 +307,5 @@ const struct inode_operations reiserfs_f
7206 .listxattr = reiserfs_listxattr,
7207 .removexattr = reiserfs_removexattr,
7208 .permission = reiserfs_permission,
7209 + .sync_flags = reiserfs_sync_flags,
7211 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/inode.c
7212 --- linux-2.6.32.1/fs/reiserfs/inode.c 2009-09-10 15:26:24.000000000 +0200
7213 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/inode.c 2009-12-03 20:04:56.000000000 +0100
7215 #include <linux/writeback.h>
7216 #include <linux/quotaops.h>
7217 #include <linux/swap.h>
7218 +#include <linux/vs_tag.h>
7220 int reiserfs_commit_write(struct file *f, struct page *page,
7221 unsigned from, unsigned to);
7222 @@ -1117,6 +1118,8 @@ static void init_inode(struct inode *ino
7223 struct buffer_head *bh;
7224 struct item_head *ih;
7228 //int version = ITEM_VERSION_1;
7230 bh = PATH_PLAST_BUFFER(path);
7231 @@ -1138,12 +1141,13 @@ static void init_inode(struct inode *ino
7232 (struct stat_data_v1 *)B_I_PITEM(bh, ih);
7233 unsigned long blocks;
7235 + uid = sd_v1_uid(sd);
7236 + gid = sd_v1_gid(sd);
7238 set_inode_item_key_version(inode, KEY_FORMAT_3_5);
7239 set_inode_sd_version(inode, STAT_DATA_V1);
7240 inode->i_mode = sd_v1_mode(sd);
7241 inode->i_nlink = sd_v1_nlink(sd);
7242 - inode->i_uid = sd_v1_uid(sd);
7243 - inode->i_gid = sd_v1_gid(sd);
7244 inode->i_size = sd_v1_size(sd);
7245 inode->i_atime.tv_sec = sd_v1_atime(sd);
7246 inode->i_mtime.tv_sec = sd_v1_mtime(sd);
7247 @@ -1185,11 +1189,12 @@ static void init_inode(struct inode *ino
7248 // (directories and symlinks)
7249 struct stat_data *sd = (struct stat_data *)B_I_PITEM(bh, ih);
7251 + uid = sd_v2_uid(sd);
7252 + gid = sd_v2_gid(sd);
7254 inode->i_mode = sd_v2_mode(sd);
7255 inode->i_nlink = sd_v2_nlink(sd);
7256 - inode->i_uid = sd_v2_uid(sd);
7257 inode->i_size = sd_v2_size(sd);
7258 - inode->i_gid = sd_v2_gid(sd);
7259 inode->i_mtime.tv_sec = sd_v2_mtime(sd);
7260 inode->i_atime.tv_sec = sd_v2_atime(sd);
7261 inode->i_ctime.tv_sec = sd_v2_ctime(sd);
7262 @@ -1219,6 +1224,10 @@ static void init_inode(struct inode *ino
7263 sd_attrs_to_i_attrs(sd_v2_attrs(sd), inode);
7266 + inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
7267 + inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
7268 + inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
7271 if (S_ISREG(inode->i_mode)) {
7272 inode->i_op = &reiserfs_file_inode_operations;
7273 @@ -1241,13 +1250,15 @@ static void init_inode(struct inode *ino
7274 static void inode2sd(void *sd, struct inode *inode, loff_t size)
7276 struct stat_data *sd_v2 = (struct stat_data *)sd;
7277 + uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
7278 + gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
7281 + set_sd_v2_uid(sd_v2, uid);
7282 + set_sd_v2_gid(sd_v2, gid);
7283 set_sd_v2_mode(sd_v2, inode->i_mode);
7284 set_sd_v2_nlink(sd_v2, inode->i_nlink);
7285 - set_sd_v2_uid(sd_v2, inode->i_uid);
7286 set_sd_v2_size(sd_v2, size);
7287 - set_sd_v2_gid(sd_v2, inode->i_gid);
7288 set_sd_v2_mtime(sd_v2, inode->i_mtime.tv_sec);
7289 set_sd_v2_atime(sd_v2, inode->i_atime.tv_sec);
7290 set_sd_v2_ctime(sd_v2, inode->i_ctime.tv_sec);
7291 @@ -2828,14 +2839,19 @@ int reiserfs_commit_write(struct file *f
7292 void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode)
7294 if (reiserfs_attrs(inode->i_sb)) {
7295 - if (sd_attrs & REISERFS_SYNC_FL)
7296 - inode->i_flags |= S_SYNC;
7298 - inode->i_flags &= ~S_SYNC;
7299 if (sd_attrs & REISERFS_IMMUTABLE_FL)
7300 inode->i_flags |= S_IMMUTABLE;
7302 inode->i_flags &= ~S_IMMUTABLE;
7303 + if (sd_attrs & REISERFS_IXUNLINK_FL)
7304 + inode->i_flags |= S_IXUNLINK;
7306 + inode->i_flags &= ~S_IXUNLINK;
7308 + if (sd_attrs & REISERFS_SYNC_FL)
7309 + inode->i_flags |= S_SYNC;
7311 + inode->i_flags &= ~S_SYNC;
7312 if (sd_attrs & REISERFS_APPEND_FL)
7313 inode->i_flags |= S_APPEND;
7315 @@ -2848,6 +2864,15 @@ void sd_attrs_to_i_attrs(__u16 sd_attrs,
7316 REISERFS_I(inode)->i_flags |= i_nopack_mask;
7318 REISERFS_I(inode)->i_flags &= ~i_nopack_mask;
7320 + if (sd_attrs & REISERFS_BARRIER_FL)
7321 + inode->i_vflags |= V_BARRIER;
7323 + inode->i_vflags &= ~V_BARRIER;
7324 + if (sd_attrs & REISERFS_COW_FL)
7325 + inode->i_vflags |= V_COW;
7327 + inode->i_vflags &= ~V_COW;
7331 @@ -2858,6 +2883,11 @@ void i_attrs_to_sd_attrs(struct inode *i
7332 *sd_attrs |= REISERFS_IMMUTABLE_FL;
7334 *sd_attrs &= ~REISERFS_IMMUTABLE_FL;
7335 + if (inode->i_flags & S_IXUNLINK)
7336 + *sd_attrs |= REISERFS_IXUNLINK_FL;
7338 + *sd_attrs &= ~REISERFS_IXUNLINK_FL;
7340 if (inode->i_flags & S_SYNC)
7341 *sd_attrs |= REISERFS_SYNC_FL;
7343 @@ -2870,6 +2900,15 @@ void i_attrs_to_sd_attrs(struct inode *i
7344 *sd_attrs |= REISERFS_NOTAIL_FL;
7346 *sd_attrs &= ~REISERFS_NOTAIL_FL;
7348 + if (inode->i_vflags & V_BARRIER)
7349 + *sd_attrs |= REISERFS_BARRIER_FL;
7351 + *sd_attrs &= ~REISERFS_BARRIER_FL;
7352 + if (inode->i_vflags & V_COW)
7353 + *sd_attrs |= REISERFS_COW_FL;
7355 + *sd_attrs &= ~REISERFS_COW_FL;
7359 @@ -3090,9 +3129,11 @@ int reiserfs_setattr(struct dentry *dent
7362 error = inode_change_ok(inode, attr);
7365 if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
7366 - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
7367 + (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
7368 + (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
7369 error = reiserfs_chown_xattrs(inode, attr);
7372 @@ -3122,6 +3163,9 @@ int reiserfs_setattr(struct dentry *dent
7373 inode->i_uid = attr->ia_uid;
7374 if (attr->ia_valid & ATTR_GID)
7375 inode->i_gid = attr->ia_gid;
7376 + if ((attr->ia_valid & ATTR_TAG) &&
7378 + inode->i_tag = attr->ia_tag;
7379 mark_inode_dirty(inode);
7381 journal_end(&th, inode->i_sb, jbegin_count);
7382 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/ioctl.c
7383 --- linux-2.6.32.1/fs/reiserfs/ioctl.c 2009-06-11 17:13:08.000000000 +0200
7384 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/ioctl.c 2009-12-03 20:04:56.000000000 +0100
7386 #include <linux/mount.h>
7387 #include <linux/reiserfs_fs.h>
7388 #include <linux/time.h>
7389 +#include <linux/mount.h>
7390 #include <asm/uaccess.h>
7391 #include <linux/pagemap.h>
7392 #include <linux/smp_lock.h>
7393 #include <linux/compat.h>
7396 +int reiserfs_sync_flags(struct inode *inode, int flags, int vflags)
7398 + __u16 sd_attrs = 0;
7400 + inode->i_flags = flags;
7401 + inode->i_vflags = vflags;
7403 + i_attrs_to_sd_attrs(inode, &sd_attrs);
7404 + REISERFS_I(inode)->i_attrs = sd_attrs;
7405 + inode->i_ctime = CURRENT_TIME_SEC;
7406 + mark_inode_dirty(inode);
7411 ** reiserfs_ioctl - handler for ioctl for inode
7412 ** supported commands:
7414 int reiserfs_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
7417 - unsigned int flags;
7418 + unsigned int flags, oldflags;
7422 @@ -43,6 +59,7 @@ int reiserfs_ioctl(struct inode *inode,
7424 flags = REISERFS_I(inode)->i_attrs;
7425 i_attrs_to_sd_attrs(inode, (__u16 *) & flags);
7426 + flags &= REISERFS_FL_USER_VISIBLE;
7427 return put_user(flags, (int __user *)arg);
7428 case REISERFS_IOC_SETFLAGS:{
7429 if (!reiserfs_attrs(inode->i_sb))
7430 @@ -60,6 +77,10 @@ int reiserfs_ioctl(struct inode *inode,
7434 + if (IS_BARRIER(inode)) {
7435 + vxwprintk_task(1, "messing with the barrier.");
7439 * Is it quota file? Do not allow user to mess with it
7441 @@ -84,6 +105,10 @@ int reiserfs_ioctl(struct inode *inode,
7446 + oldflags = REISERFS_I(inode)->i_attrs;
7447 + flags &= REISERFS_FL_USER_MODIFIABLE;
7448 + flags |= oldflags & ~REISERFS_FL_USER_MODIFIABLE;
7449 sd_attrs_to_i_attrs(flags, inode);
7450 REISERFS_I(inode)->i_attrs = flags;
7451 inode->i_ctime = CURRENT_TIME_SEC;
7452 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/namei.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/namei.c
7453 --- linux-2.6.32.1/fs/reiserfs/namei.c 2009-06-11 17:13:08.000000000 +0200
7454 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/namei.c 2009-12-03 20:04:56.000000000 +0100
7456 #include <linux/reiserfs_acl.h>
7457 #include <linux/reiserfs_xattr.h>
7458 #include <linux/quotaops.h>
7459 +#include <linux/vs_tag.h>
7461 #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; }
7462 #define DEC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) drop_nlink(i);
7463 @@ -354,6 +355,7 @@ static struct dentry *reiserfs_lookup(st
7464 if (retval == IO_ERROR) {
7465 return ERR_PTR(-EIO);
7467 + dx_propagate_tag(nd, inode);
7469 return d_splice_alias(inode, dentry);
7471 @@ -570,6 +572,7 @@ static int new_inode_init(struct inode *
7473 inode->i_gid = current_fsgid();
7475 + inode->i_tag = dx_current_fstag(inode->i_sb);
7479 @@ -1515,6 +1518,7 @@ const struct inode_operations reiserfs_d
7480 .listxattr = reiserfs_listxattr,
7481 .removexattr = reiserfs_removexattr,
7482 .permission = reiserfs_permission,
7483 + .sync_flags = reiserfs_sync_flags,
7487 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/super.c
7488 --- linux-2.6.32.1/fs/reiserfs/super.c 2009-12-03 20:02:53.000000000 +0100
7489 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/super.c 2009-12-03 20:04:56.000000000 +0100
7490 @@ -884,6 +884,14 @@ static int reiserfs_parse_options(struct
7491 {"user_xattr",.setmask = 1 << REISERFS_UNSUPPORTED_OPT},
7492 {"nouser_xattr",.clrmask = 1 << REISERFS_UNSUPPORTED_OPT},
7494 +#ifndef CONFIG_TAGGING_NONE
7495 + {"tagxid",.setmask = 1 << REISERFS_TAGGED},
7496 + {"tag",.setmask = 1 << REISERFS_TAGGED},
7497 + {"notag",.clrmask = 1 << REISERFS_TAGGED},
7499 +#ifdef CONFIG_PROPAGATE
7500 + {"tag",.arg_required = 'T',.values = NULL},
7502 #ifdef CONFIG_REISERFS_FS_POSIX_ACL
7503 {"acl",.setmask = 1 << REISERFS_POSIXACL},
7504 {"noacl",.clrmask = 1 << REISERFS_POSIXACL},
7505 @@ -1190,6 +1198,14 @@ static int reiserfs_remount(struct super
7506 handle_quota_files(s, qf_names, &qfmt);
7509 + if ((mount_options & (1 << REISERFS_TAGGED)) &&
7510 + !(s->s_flags & MS_TAGGED)) {
7511 + reiserfs_warning(s, "super-vs01",
7512 + "reiserfs: tagging not permitted on remount.");
7519 /* Add options that are safe here */
7520 @@ -1652,6 +1668,10 @@ static int reiserfs_fill_super(struct su
7524 + /* map mount option tagxid */
7525 + if (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TAGGED))
7526 + s->s_flags |= MS_TAGGED;
7528 rs = SB_DISK_SUPER_BLOCK(s);
7529 /* Let's do basic sanity check to verify that underlying device is not
7530 smaller than the filesystem. If the check fails then abort and scream,
7531 diff -NurpP --minimal linux-2.6.32.1/fs/reiserfs/xattr.c linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/xattr.c
7532 --- linux-2.6.32.1/fs/reiserfs/xattr.c 2009-09-10 15:26:24.000000000 +0200
7533 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/reiserfs/xattr.c 2009-12-03 20:04:56.000000000 +0100
7535 #include <linux/namei.h>
7536 #include <linux/errno.h>
7537 #include <linux/fs.h>
7538 +#include <linux/mount.h>
7539 #include <linux/file.h>
7540 #include <linux/pagemap.h>
7541 #include <linux/xattr.h>
7542 diff -NurpP --minimal linux-2.6.32.1/fs/stat.c linux-2.6.32.1-vs2.3.0.36.27/fs/stat.c
7543 --- linux-2.6.32.1/fs/stat.c 2009-06-11 17:13:08.000000000 +0200
7544 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/stat.c 2009-12-03 20:04:56.000000000 +0100
7545 @@ -26,6 +26,7 @@ void generic_fillattr(struct inode *inod
7546 stat->nlink = inode->i_nlink;
7547 stat->uid = inode->i_uid;
7548 stat->gid = inode->i_gid;
7549 + stat->tag = inode->i_tag;
7550 stat->rdev = inode->i_rdev;
7551 stat->atime = inode->i_atime;
7552 stat->mtime = inode->i_mtime;
7553 diff -NurpP --minimal linux-2.6.32.1/fs/super.c linux-2.6.32.1-vs2.3.0.36.27/fs/super.c
7554 --- linux-2.6.32.1/fs/super.c 2009-12-03 20:02:53.000000000 +0100
7555 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/super.c 2009-12-03 20:04:56.000000000 +0100
7557 #include <linux/kobject.h>
7558 #include <linux/mutex.h>
7559 #include <linux/file.h>
7560 +#include <linux/devpts_fs.h>
7561 +#include <linux/proc_fs.h>
7562 +#include <linux/vs_context.h>
7563 #include <asm/uaccess.h>
7564 #include "internal.h"
7566 @@ -913,12 +916,18 @@ struct vfsmount *
7567 vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
7569 struct vfsmount *mnt;
7570 + struct super_block *sb;
7571 char *secdata = NULL;
7575 return ERR_PTR(-ENODEV);
7578 + if ((type->fs_flags & FS_BINARY_MOUNTDATA) &&
7579 + !vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT))
7583 mnt = alloc_vfsmnt(name);
7585 @@ -937,9 +946,17 @@ vfs_kern_mount(struct file_system_type *
7586 error = type->get_sb(type, flags, name, data, mnt);
7588 goto out_free_secdata;
7589 - BUG_ON(!mnt->mnt_sb);
7591 - error = security_sb_kern_mount(mnt->mnt_sb, flags, secdata);
7596 + if (!vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT) && !sb->s_bdev &&
7597 + (sb->s_magic != PROC_SUPER_MAGIC) &&
7598 + (sb->s_magic != DEVPTS_SUPER_MAGIC))
7601 + error = security_sb_kern_mount(sb, flags, secdata);
7605 diff -NurpP --minimal linux-2.6.32.1/fs/sysfs/mount.c linux-2.6.32.1-vs2.3.0.36.27/fs/sysfs/mount.c
7606 --- linux-2.6.32.1/fs/sysfs/mount.c 2009-06-11 17:13:08.000000000 +0200
7607 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/sysfs/mount.c 2009-12-03 20:04:56.000000000 +0100
7608 @@ -47,7 +47,7 @@ static int sysfs_fill_super(struct super
7610 sb->s_blocksize = PAGE_CACHE_SIZE;
7611 sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
7612 - sb->s_magic = SYSFS_MAGIC;
7613 + sb->s_magic = SYSFS_SUPER_MAGIC;
7614 sb->s_op = &sysfs_ops;
7615 sb->s_time_gran = 1;
7617 diff -NurpP --minimal linux-2.6.32.1/fs/utimes.c linux-2.6.32.1-vs2.3.0.36.27/fs/utimes.c
7618 --- linux-2.6.32.1/fs/utimes.c 2009-03-24 14:22:37.000000000 +0100
7619 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/utimes.c 2009-12-03 20:04:56.000000000 +0100
7621 #include <linux/stat.h>
7622 #include <linux/utime.h>
7623 #include <linux/syscalls.h>
7624 +#include <linux/mount.h>
7625 +#include <linux/vs_cowbl.h>
7626 #include <asm/uaccess.h>
7627 #include <asm/unistd.h>
7629 diff -NurpP --minimal linux-2.6.32.1/fs/xattr.c linux-2.6.32.1-vs2.3.0.36.27/fs/xattr.c
7630 --- linux-2.6.32.1/fs/xattr.c 2009-12-03 20:02:53.000000000 +0100
7631 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xattr.c 2009-12-03 20:04:56.000000000 +0100
7633 #include <linux/module.h>
7634 #include <linux/fsnotify.h>
7635 #include <linux/audit.h>
7636 +#include <linux/mount.h>
7637 #include <asm/uaccess.h>
7640 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/linux-2.6/xfs_ioctl.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_ioctl.c
7641 --- linux-2.6.32.1/fs/xfs/linux-2.6/xfs_ioctl.c 2009-09-10 15:26:24.000000000 +0200
7642 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_ioctl.c 2009-12-03 20:04:56.000000000 +0100
7644 #include "xfs_dir2_sf.h"
7645 #include "xfs_dinode.h"
7646 #include "xfs_inode.h"
7647 -#include "xfs_ioctl.h"
7648 #include "xfs_btree.h"
7649 #include "xfs_ialloc.h"
7650 #include "xfs_rtalloc.h"
7651 @@ -742,6 +741,10 @@ xfs_merge_ioc_xflags(
7652 xflags |= XFS_XFLAG_IMMUTABLE;
7654 xflags &= ~XFS_XFLAG_IMMUTABLE;
7655 + if (flags & FS_IXUNLINK_FL)
7656 + xflags |= XFS_XFLAG_IXUNLINK;
7658 + xflags &= ~XFS_XFLAG_IXUNLINK;
7659 if (flags & FS_APPEND_FL)
7660 xflags |= XFS_XFLAG_APPEND;
7662 @@ -770,6 +773,8 @@ xfs_di2lxflags(
7664 if (di_flags & XFS_DIFLAG_IMMUTABLE)
7665 flags |= FS_IMMUTABLE_FL;
7666 + if (di_flags & XFS_DIFLAG_IXUNLINK)
7667 + flags |= FS_IXUNLINK_FL;
7668 if (di_flags & XFS_DIFLAG_APPEND)
7669 flags |= FS_APPEND_FL;
7670 if (di_flags & XFS_DIFLAG_SYNC)
7671 @@ -828,6 +833,8 @@ xfs_set_diflags(
7672 di_flags = (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC);
7673 if (xflags & XFS_XFLAG_IMMUTABLE)
7674 di_flags |= XFS_DIFLAG_IMMUTABLE;
7675 + if (xflags & XFS_XFLAG_IXUNLINK)
7676 + di_flags |= XFS_DIFLAG_IXUNLINK;
7677 if (xflags & XFS_XFLAG_APPEND)
7678 di_flags |= XFS_DIFLAG_APPEND;
7679 if (xflags & XFS_XFLAG_SYNC)
7680 @@ -870,6 +877,10 @@ xfs_diflags_to_linux(
7681 inode->i_flags |= S_IMMUTABLE;
7683 inode->i_flags &= ~S_IMMUTABLE;
7684 + if (xflags & XFS_XFLAG_IXUNLINK)
7685 + inode->i_flags |= S_IXUNLINK;
7687 + inode->i_flags &= ~S_IXUNLINK;
7688 if (xflags & XFS_XFLAG_APPEND)
7689 inode->i_flags |= S_APPEND;
7691 @@ -1346,10 +1357,18 @@ xfs_file_ioctl(
7692 case XFS_IOC_FSGETXATTRA:
7693 return xfs_ioc_fsgetxattr(ip, 1, arg);
7694 case XFS_IOC_FSSETXATTR:
7695 + if (IS_BARRIER(inode)) {
7696 + vxwprintk_task(1, "messing with the barrier.");
7697 + return -XFS_ERROR(EACCES);
7699 return xfs_ioc_fssetxattr(ip, filp, arg);
7700 case XFS_IOC_GETXFLAGS:
7701 return xfs_ioc_getxflags(ip, arg);
7702 case XFS_IOC_SETXFLAGS:
7703 + if (IS_BARRIER(inode)) {
7704 + vxwprintk_task(1, "messing with the barrier.");
7705 + return -XFS_ERROR(EACCES);
7707 return xfs_ioc_setxflags(ip, filp, arg);
7709 case XFS_IOC_FSSETDM: {
7710 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/linux-2.6/xfs_ioctl.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_ioctl.h
7711 --- linux-2.6.32.1/fs/xfs/linux-2.6/xfs_ioctl.h 2009-03-24 14:22:37.000000000 +0100
7712 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_ioctl.h 2009-12-03 20:04:56.000000000 +0100
7713 @@ -70,6 +70,12 @@ xfs_handle_to_dentry(
7714 void __user *uhandle,
7719 + struct inode *inode,
7726 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/linux-2.6/xfs_iops.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_iops.c
7727 --- linux-2.6.32.1/fs/xfs/linux-2.6/xfs_iops.c 2009-12-03 20:02:53.000000000 +0100
7728 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_iops.c 2009-12-03 20:04:56.000000000 +0100
7730 #include "xfs_attr_sf.h"
7731 #include "xfs_dinode.h"
7732 #include "xfs_inode.h"
7733 +#include "xfs_ioctl.h"
7734 #include "xfs_bmap.h"
7735 #include "xfs_btree.h"
7736 #include "xfs_ialloc.h"
7738 #include <linux/security.h>
7739 #include <linux/falloc.h>
7740 #include <linux/fiemap.h>
7741 +#include <linux/vs_tag.h>
7744 * Bring the timestamps in the XFS inode uptodate.
7745 @@ -495,6 +497,7 @@ xfs_vn_getattr(
7746 stat->nlink = ip->i_d.di_nlink;
7747 stat->uid = ip->i_d.di_uid;
7748 stat->gid = ip->i_d.di_gid;
7749 + stat->tag = ip->i_d.di_tag;
7750 stat->ino = ip->i_ino;
7751 stat->atime = inode->i_atime;
7752 stat->mtime = inode->i_mtime;
7753 @@ -686,6 +689,7 @@ static const struct inode_operations xfs
7754 .listxattr = xfs_vn_listxattr,
7755 .fallocate = xfs_vn_fallocate,
7756 .fiemap = xfs_vn_fiemap,
7757 + .sync_flags = xfs_sync_flags,
7760 static const struct inode_operations xfs_dir_inode_operations = {
7761 @@ -711,6 +715,7 @@ static const struct inode_operations xfs
7762 .getxattr = generic_getxattr,
7763 .removexattr = generic_removexattr,
7764 .listxattr = xfs_vn_listxattr,
7765 + .sync_flags = xfs_sync_flags,
7768 static const struct inode_operations xfs_dir_ci_inode_operations = {
7769 @@ -760,6 +765,10 @@ xfs_diflags_to_iflags(
7770 inode->i_flags |= S_IMMUTABLE;
7772 inode->i_flags &= ~S_IMMUTABLE;
7773 + if (ip->i_d.di_flags & XFS_DIFLAG_IXUNLINK)
7774 + inode->i_flags |= S_IXUNLINK;
7776 + inode->i_flags &= ~S_IXUNLINK;
7777 if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
7778 inode->i_flags |= S_APPEND;
7780 @@ -772,6 +781,15 @@ xfs_diflags_to_iflags(
7781 inode->i_flags |= S_NOATIME;
7783 inode->i_flags &= ~S_NOATIME;
7785 + if (ip->i_d.di_vflags & XFS_DIVFLAG_BARRIER)
7786 + inode->i_vflags |= V_BARRIER;
7788 + inode->i_vflags &= ~V_BARRIER;
7789 + if (ip->i_d.di_vflags & XFS_DIVFLAG_COW)
7790 + inode->i_vflags |= V_COW;
7792 + inode->i_vflags &= ~V_COW;
7796 @@ -800,6 +818,7 @@ xfs_setup_inode(
7797 inode->i_nlink = ip->i_d.di_nlink;
7798 inode->i_uid = ip->i_d.di_uid;
7799 inode->i_gid = ip->i_d.di_gid;
7800 + inode->i_tag = ip->i_d.di_tag;
7802 switch (inode->i_mode & S_IFMT) {
7804 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/linux-2.6/xfs_linux.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_linux.h
7805 --- linux-2.6.32.1/fs/xfs/linux-2.6/xfs_linux.h 2009-09-10 15:26:24.000000000 +0200
7806 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_linux.h 2009-12-03 20:04:56.000000000 +0100
7809 #define current_cpu() (raw_smp_processor_id())
7810 #define current_pid() (current->pid)
7811 +#define current_fstag(cred,vp) (dx_current_fstag((vp)->i_sb))
7812 #define current_test_flags(f) (current->flags & (f))
7813 #define current_set_flags_nested(sp, f) \
7814 (*(sp) = current->flags, current->flags |= (f))
7815 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/linux-2.6/xfs_super.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_super.c
7816 --- linux-2.6.32.1/fs/xfs/linux-2.6/xfs_super.c 2009-12-03 20:02:53.000000000 +0100
7817 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/linux-2.6/xfs_super.c 2009-12-03 20:04:56.000000000 +0100
7818 @@ -117,6 +117,9 @@ mempool_t *xfs_ioend_pool;
7819 #define MNTOPT_DMAPI "dmapi" /* DMI enabled (DMAPI / XDSM) */
7820 #define MNTOPT_XDSM "xdsm" /* DMI enabled (DMAPI / XDSM) */
7821 #define MNTOPT_DMI "dmi" /* DMI enabled (DMAPI / XDSM) */
7822 +#define MNTOPT_TAGXID "tagxid" /* context tagging for inodes */
7823 +#define MNTOPT_TAGGED "tag" /* context tagging for inodes */
7824 +#define MNTOPT_NOTAGTAG "notag" /* do not use context tagging */
7827 * Table driven mount option parser.
7828 @@ -125,10 +128,14 @@ mempool_t *xfs_ioend_pool;
7829 * in the future, too.
7832 + Opt_tag, Opt_notag,
7833 Opt_barrier, Opt_nobarrier, Opt_err
7836 static const match_table_t tokens = {
7837 + {Opt_tag, "tagxid"},
7839 + {Opt_notag, "notag"},
7840 {Opt_barrier, "barrier"},
7841 {Opt_nobarrier, "nobarrier"},
7843 @@ -382,6 +389,19 @@ xfs_parseargs(
7844 } else if (!strcmp(this_char, "irixsgid")) {
7846 "XFS: irixsgid is now a sysctl(2) variable, option is deprecated.");
7847 +#ifndef CONFIG_TAGGING_NONE
7848 + } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
7849 + mp->m_flags |= XFS_MOUNT_TAGGED;
7850 + } else if (!strcmp(this_char, MNTOPT_NOTAGTAG)) {
7851 + mp->m_flags &= ~XFS_MOUNT_TAGGED;
7852 + } else if (!strcmp(this_char, MNTOPT_TAGXID)) {
7853 + mp->m_flags |= XFS_MOUNT_TAGGED;
7855 +#ifdef CONFIG_PROPAGATE
7856 + } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
7858 + mp->m_flags |= XFS_MOUNT_TAGGED;
7862 "XFS: unknown mount option [%s].", this_char);
7863 @@ -1270,6 +1290,16 @@ xfs_fs_remount(
7865 mp->m_flags &= ~XFS_MOUNT_BARRIER;
7868 + if (!(sb->s_flags & MS_TAGGED)) {
7870 + "XFS: %s: tagging not permitted on remount.\n",
7879 * Logically we would return an error here to prevent
7880 @@ -1477,6 +1507,9 @@ xfs_fs_fill_super(
7882 XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, mtpt, mp->m_fsname);
7884 + if (mp->m_flags & XFS_MOUNT_TAGGED)
7885 + sb->s_flags |= MS_TAGGED;
7887 sb->s_magic = XFS_SB_MAGIC;
7888 sb->s_blocksize = mp->m_sb.sb_blocksize;
7889 sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1;
7890 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_dinode.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_dinode.h
7891 --- linux-2.6.32.1/fs/xfs/xfs_dinode.h 2009-06-11 17:13:09.000000000 +0200
7892 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_dinode.h 2009-12-03 20:04:56.000000000 +0100
7893 @@ -50,7 +50,9 @@ typedef struct xfs_dinode {
7894 __be32 di_gid; /* owner's group id */
7895 __be32 di_nlink; /* number of links to file */
7896 __be16 di_projid; /* owner's project id */
7897 - __u8 di_pad[8]; /* unused, zeroed space */
7898 + __be16 di_tag; /* context tagging */
7899 + __be16 di_vflags; /* vserver specific flags */
7900 + __u8 di_pad[4]; /* unused, zeroed space */
7901 __be16 di_flushiter; /* incremented on flush */
7902 xfs_timestamp_t di_atime; /* time last accessed */
7903 xfs_timestamp_t di_mtime; /* time last modified */
7904 @@ -183,6 +185,8 @@ static inline void xfs_dinode_put_rdev(s
7905 #define XFS_DIFLAG_EXTSZINHERIT_BIT 12 /* inherit inode extent size */
7906 #define XFS_DIFLAG_NODEFRAG_BIT 13 /* do not reorganize/defragment */
7907 #define XFS_DIFLAG_FILESTREAM_BIT 14 /* use filestream allocator */
7908 +#define XFS_DIFLAG_IXUNLINK_BIT 15 /* Immutable inver on unlink */
7910 #define XFS_DIFLAG_REALTIME (1 << XFS_DIFLAG_REALTIME_BIT)
7911 #define XFS_DIFLAG_PREALLOC (1 << XFS_DIFLAG_PREALLOC_BIT)
7912 #define XFS_DIFLAG_NEWRTBM (1 << XFS_DIFLAG_NEWRTBM_BIT)
7913 @@ -198,6 +202,7 @@ static inline void xfs_dinode_put_rdev(s
7914 #define XFS_DIFLAG_EXTSZINHERIT (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
7915 #define XFS_DIFLAG_NODEFRAG (1 << XFS_DIFLAG_NODEFRAG_BIT)
7916 #define XFS_DIFLAG_FILESTREAM (1 << XFS_DIFLAG_FILESTREAM_BIT)
7917 +#define XFS_DIFLAG_IXUNLINK (1 << XFS_DIFLAG_IXUNLINK_BIT)
7919 #ifdef CONFIG_XFS_RT
7920 #define XFS_IS_REALTIME_INODE(ip) ((ip)->i_d.di_flags & XFS_DIFLAG_REALTIME)
7921 @@ -210,6 +215,10 @@ static inline void xfs_dinode_put_rdev(s
7922 XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
7923 XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
7924 XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
7925 - XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
7926 + XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM | \
7927 + XFS_DIFLAG_IXUNLINK)
7929 +#define XFS_DIVFLAG_BARRIER 0x01
7930 +#define XFS_DIVFLAG_COW 0x02
7932 #endif /* __XFS_DINODE_H__ */
7933 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_fs.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_fs.h
7934 --- linux-2.6.32.1/fs/xfs/xfs_fs.h 2009-12-03 20:02:53.000000000 +0100
7935 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_fs.h 2009-12-03 20:04:56.000000000 +0100
7936 @@ -67,6 +67,9 @@ struct fsxattr {
7937 #define XFS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
7938 #define XFS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
7939 #define XFS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
7940 +#define XFS_XFLAG_IXUNLINK 0x00008000 /* immutable invert on unlink */
7941 +#define XFS_XFLAG_BARRIER 0x10000000 /* chroot() barrier */
7942 +#define XFS_XFLAG_COW 0x20000000 /* copy on write mark */
7943 #define XFS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
7946 @@ -292,7 +295,8 @@ typedef struct xfs_bstat {
7947 __s32 bs_extents; /* number of extents */
7948 __u32 bs_gen; /* generation count */
7949 __u16 bs_projid; /* project id */
7950 - unsigned char bs_pad[14]; /* pad space, unused */
7951 + __u16 bs_tag; /* context tagging */
7952 + unsigned char bs_pad[12]; /* pad space, unused */
7953 __u32 bs_dmevmask; /* DMIG event mask */
7954 __u16 bs_dmstate; /* DMIG state info */
7955 __u16 bs_aextents; /* attribute number of extents */
7956 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_ialloc.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_ialloc.c
7957 --- linux-2.6.32.1/fs/xfs/xfs_ialloc.c 2009-12-03 20:02:53.000000000 +0100
7958 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_ialloc.c 2009-12-03 20:04:56.000000000 +0100
7960 #include "xfs_error.h"
7961 #include "xfs_bmap.h"
7965 * Allocation group level functions.
7967 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_inode.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_inode.c
7968 --- linux-2.6.32.1/fs/xfs/xfs_inode.c 2009-12-03 20:02:54.000000000 +0100
7969 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_inode.c 2009-12-03 20:04:56.000000000 +0100
7970 @@ -249,6 +249,7 @@ xfs_inotobp(
7974 +#include <linux/vs_tag.h>
7977 * This routine is called to map an inode to the buffer containing
7978 @@ -654,15 +655,25 @@ xfs_iformat_btree(
7980 xfs_dinode_from_disk(
7982 - xfs_dinode_t *from)
7983 + xfs_dinode_t *from,
7986 + uint32_t uid, gid, tag;
7988 to->di_magic = be16_to_cpu(from->di_magic);
7989 to->di_mode = be16_to_cpu(from->di_mode);
7990 to->di_version = from ->di_version;
7991 to->di_format = from->di_format;
7992 to->di_onlink = be16_to_cpu(from->di_onlink);
7993 - to->di_uid = be32_to_cpu(from->di_uid);
7994 - to->di_gid = be32_to_cpu(from->di_gid);
7996 + uid = be32_to_cpu(from->di_uid);
7997 + gid = be32_to_cpu(from->di_gid);
7998 + tag = be16_to_cpu(from->di_tag);
8000 + to->di_uid = INOTAG_UID(tagged, uid, gid);
8001 + to->di_gid = INOTAG_GID(tagged, uid, gid);
8002 + to->di_tag = INOTAG_TAG(tagged, uid, gid, tag);
8004 to->di_nlink = be32_to_cpu(from->di_nlink);
8005 to->di_projid = be16_to_cpu(from->di_projid);
8006 memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8007 @@ -683,21 +694,26 @@ xfs_dinode_from_disk(
8008 to->di_dmevmask = be32_to_cpu(from->di_dmevmask);
8009 to->di_dmstate = be16_to_cpu(from->di_dmstate);
8010 to->di_flags = be16_to_cpu(from->di_flags);
8011 + to->di_vflags = be16_to_cpu(from->di_vflags);
8012 to->di_gen = be32_to_cpu(from->di_gen);
8018 - xfs_icdinode_t *from)
8019 + xfs_icdinode_t *from,
8022 to->di_magic = cpu_to_be16(from->di_magic);
8023 to->di_mode = cpu_to_be16(from->di_mode);
8024 to->di_version = from ->di_version;
8025 to->di_format = from->di_format;
8026 to->di_onlink = cpu_to_be16(from->di_onlink);
8027 - to->di_uid = cpu_to_be32(from->di_uid);
8028 - to->di_gid = cpu_to_be32(from->di_gid);
8030 + to->di_uid = cpu_to_be32(TAGINO_UID(tagged, from->di_uid, from->di_tag));
8031 + to->di_gid = cpu_to_be32(TAGINO_GID(tagged, from->di_gid, from->di_tag));
8032 + to->di_tag = cpu_to_be16(TAGINO_TAG(tagged, from->di_tag));
8034 to->di_nlink = cpu_to_be32(from->di_nlink);
8035 to->di_projid = cpu_to_be16(from->di_projid);
8036 memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8037 @@ -718,12 +734,14 @@ xfs_dinode_to_disk(
8038 to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
8039 to->di_dmstate = cpu_to_be16(from->di_dmstate);
8040 to->di_flags = cpu_to_be16(from->di_flags);
8041 + to->di_vflags = cpu_to_be16(from->di_vflags);
8042 to->di_gen = cpu_to_be32(from->di_gen);
8047 - __uint16_t di_flags)
8048 + __uint16_t di_flags,
8049 + __uint16_t di_vflags)
8053 @@ -734,6 +752,8 @@ _xfs_dic2xflags(
8054 flags |= XFS_XFLAG_PREALLOC;
8055 if (di_flags & XFS_DIFLAG_IMMUTABLE)
8056 flags |= XFS_XFLAG_IMMUTABLE;
8057 + if (di_flags & XFS_DIFLAG_IXUNLINK)
8058 + flags |= XFS_XFLAG_IXUNLINK;
8059 if (di_flags & XFS_DIFLAG_APPEND)
8060 flags |= XFS_XFLAG_APPEND;
8061 if (di_flags & XFS_DIFLAG_SYNC)
8062 @@ -758,6 +778,10 @@ _xfs_dic2xflags(
8063 flags |= XFS_XFLAG_FILESTREAM;
8066 + if (di_vflags & XFS_DIVFLAG_BARRIER)
8067 + flags |= FS_BARRIER_FL;
8068 + if (di_vflags & XFS_DIVFLAG_COW)
8069 + flags |= FS_COW_FL;
8073 @@ -767,7 +791,7 @@ xfs_ip2xflags(
8075 xfs_icdinode_t *dic = &ip->i_d;
8077 - return _xfs_dic2xflags(dic->di_flags) |
8078 + return _xfs_dic2xflags(dic->di_flags, dic->di_vflags) |
8079 (XFS_IFORK_Q(ip) ? XFS_XFLAG_HASATTR : 0);
8082 @@ -775,7 +799,8 @@ uint
8086 - return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
8087 + return _xfs_dic2xflags(be16_to_cpu(dip->di_flags),
8088 + be16_to_cpu(dip->di_vflags)) |
8089 (XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
8092 @@ -811,7 +836,6 @@ xfs_iread(
8095 dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
8098 * If we got something that isn't an inode it means someone
8099 * (nfs or dmi) has a stale handle.
8100 @@ -836,7 +860,8 @@ xfs_iread(
8101 * Otherwise, just get the truly permanent information.
8104 - xfs_dinode_from_disk(&ip->i_d, dip);
8105 + xfs_dinode_from_disk(&ip->i_d, dip,
8106 + mp->m_flags & XFS_MOUNT_TAGGED);
8107 error = xfs_iformat(ip, dip);
8110 @@ -1036,6 +1061,7 @@ xfs_ialloc(
8111 ASSERT(ip->i_d.di_nlink == nlink);
8112 ip->i_d.di_uid = current_fsuid();
8113 ip->i_d.di_gid = current_fsgid();
8114 + ip->i_d.di_tag = current_fstag(cr, &ip->i_vnode);
8115 ip->i_d.di_projid = prid;
8116 memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
8118 @@ -1096,6 +1122,7 @@ xfs_ialloc(
8119 ip->i_d.di_dmevmask = 0;
8120 ip->i_d.di_dmstate = 0;
8121 ip->i_d.di_flags = 0;
8122 + ip->i_d.di_vflags = 0;
8123 flags = XFS_ILOG_CORE;
8124 switch (mode & S_IFMT) {
8126 @@ -2172,6 +2199,7 @@ xfs_ifree(
8128 ip->i_d.di_mode = 0; /* mark incore inode as free */
8129 ip->i_d.di_flags = 0;
8130 + ip->i_d.di_vflags = 0;
8131 ip->i_d.di_dmevmask = 0;
8132 ip->i_d.di_forkoff = 0; /* mark the attr fork not in use */
8133 ip->i_df.if_ext_max =
8134 @@ -3139,7 +3167,8 @@ xfs_iflush_int(
8135 * because if the inode is dirty at all the core must
8138 - xfs_dinode_to_disk(dip, &ip->i_d);
8139 + xfs_dinode_to_disk(dip, &ip->i_d,
8140 + mp->m_flags & XFS_MOUNT_TAGGED);
8142 /* Wrap, we never let the log put out DI_MAX_FLUSH */
8143 if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
8144 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_inode.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_inode.h
8145 --- linux-2.6.32.1/fs/xfs/xfs_inode.h 2009-12-03 20:02:54.000000000 +0100
8146 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_inode.h 2009-12-03 20:04:56.000000000 +0100
8147 @@ -135,7 +135,9 @@ typedef struct xfs_icdinode {
8148 __uint32_t di_gid; /* owner's group id */
8149 __uint32_t di_nlink; /* number of links to file */
8150 __uint16_t di_projid; /* owner's project id */
8151 - __uint8_t di_pad[8]; /* unused, zeroed space */
8152 + __uint16_t di_tag; /* context tagging */
8153 + __uint16_t di_vflags; /* vserver specific flags */
8154 + __uint8_t di_pad[4]; /* unused, zeroed space */
8155 __uint16_t di_flushiter; /* incremented on flush */
8156 xfs_ictimestamp_t di_atime; /* time last accessed */
8157 xfs_ictimestamp_t di_mtime; /* time last modified */
8158 @@ -569,7 +571,7 @@ int xfs_itobp(struct xfs_mount *, struc
8159 int xfs_iread(struct xfs_mount *, struct xfs_trans *,
8160 struct xfs_inode *, xfs_daddr_t, uint);
8161 void xfs_dinode_to_disk(struct xfs_dinode *,
8162 - struct xfs_icdinode *);
8163 + struct xfs_icdinode *, int);
8164 void xfs_idestroy_fork(struct xfs_inode *, int);
8165 void xfs_idata_realloc(struct xfs_inode *, int, int);
8166 void xfs_iroot_realloc(struct xfs_inode *, int, int);
8167 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_itable.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_itable.c
8168 --- linux-2.6.32.1/fs/xfs/xfs_itable.c 2009-12-03 20:02:54.000000000 +0100
8169 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_itable.c 2009-12-03 20:04:56.000000000 +0100
8170 @@ -84,6 +84,7 @@ xfs_bulkstat_one_iget(
8171 buf->bs_mode = dic->di_mode;
8172 buf->bs_uid = dic->di_uid;
8173 buf->bs_gid = dic->di_gid;
8174 + buf->bs_tag = dic->di_tag;
8175 buf->bs_size = dic->di_size;
8178 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_log_recover.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_log_recover.c
8179 --- linux-2.6.32.1/fs/xfs/xfs_log_recover.c 2009-12-03 20:02:54.000000000 +0100
8180 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_log_recover.c 2009-12-03 20:04:56.000000000 +0100
8181 @@ -2467,7 +2467,8 @@ xlog_recover_do_inode_trans(
8184 /* The core is in in-core format */
8185 - xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr);
8186 + xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr,
8187 + mp->m_flags & XFS_MOUNT_TAGGED);
8189 /* the rest is in on-disk format */
8190 if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
8191 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_mount.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_mount.h
8192 --- linux-2.6.32.1/fs/xfs/xfs_mount.h 2009-12-03 20:02:54.000000000 +0100
8193 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_mount.h 2009-12-03 20:04:56.000000000 +0100
8194 @@ -283,6 +283,7 @@ typedef struct xfs_mount {
8196 #define XFS_MOUNT_NOATTR2 (1ULL << 25) /* disable use of attr2 format */
8198 +#define XFS_MOUNT_TAGGED (1ULL << 31) /* context tagging */
8201 * Default minimum read and write sizes.
8202 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_vnodeops.c linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_vnodeops.c
8203 --- linux-2.6.32.1/fs/xfs/xfs_vnodeops.c 2009-12-03 20:02:54.000000000 +0100
8204 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_vnodeops.c 2009-12-03 20:04:56.000000000 +0100
8206 #include "xfs_filestream.h"
8207 #include "xfs_vnodeops.h"
8211 +xfs_get_inode_flags(
8214 + struct inode *inode = VFS_I(ip);
8215 + unsigned int flags = inode->i_flags;
8216 + unsigned int vflags = inode->i_vflags;
8218 + if (flags & S_IMMUTABLE)
8219 + ip->i_d.di_flags |= XFS_DIFLAG_IMMUTABLE;
8221 + ip->i_d.di_flags &= ~XFS_DIFLAG_IMMUTABLE;
8222 + if (flags & S_IXUNLINK)
8223 + ip->i_d.di_flags |= XFS_DIFLAG_IXUNLINK;
8225 + ip->i_d.di_flags &= ~XFS_DIFLAG_IXUNLINK;
8227 + if (vflags & V_BARRIER)
8228 + ip->i_d.di_vflags |= XFS_DIVFLAG_BARRIER;
8230 + ip->i_d.di_vflags &= ~XFS_DIVFLAG_BARRIER;
8231 + if (vflags & V_COW)
8232 + ip->i_d.di_vflags |= XFS_DIVFLAG_COW;
8234 + ip->i_d.di_vflags &= ~XFS_DIVFLAG_COW;
8239 + struct inode *inode,
8243 + struct xfs_inode *ip = XFS_I(inode);
8244 + struct xfs_mount *mp = ip->i_mount;
8245 + struct xfs_trans *tp;
8246 + unsigned int lock_flags = 0;
8249 + tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
8250 + code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
8254 + lock_flags = XFS_ILOCK_EXCL;
8255 + xfs_ilock(ip, lock_flags);
8257 + xfs_trans_ijoin(tp, ip, lock_flags);
8258 + xfs_trans_ihold(tp, ip);
8260 + inode->i_flags = flags;
8261 + inode->i_vflags = vflags;
8262 + xfs_get_inode_flags(ip);
8264 + xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
8265 + xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
8267 + XFS_STATS_INC(xs_ig_attrchg);
8269 + if (mp->m_flags & XFS_MOUNT_WSYNC)
8270 + xfs_trans_set_sync(tp);
8271 + code = xfs_trans_commit(tp, 0);
8272 + xfs_iunlock(ip, lock_flags);
8276 + xfs_trans_cancel(tp, 0);
8278 + xfs_iunlock(ip, lock_flags);
8285 struct xfs_inode *ip,
8286 @@ -69,6 +143,7 @@ xfs_setattr(
8287 uint commit_flags=0;
8288 uid_t uid=0, iuid=0;
8289 gid_t gid=0, igid=0;
8290 + tag_t tag=0, itag=0;
8292 struct xfs_dquot *udqp, *gdqp, *olddquot1, *olddquot2;
8293 int need_iolock = 1;
8294 @@ -165,7 +240,7 @@ xfs_setattr(
8296 * Change file ownership. Must be the owner or privileged.
8298 - if (mask & (ATTR_UID|ATTR_GID)) {
8299 + if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8301 * These IDs could have changed since we last looked at them.
8302 * But, we're assured that if the ownership did change
8303 @@ -174,8 +249,10 @@ xfs_setattr(
8305 iuid = ip->i_d.di_uid;
8306 igid = ip->i_d.di_gid;
8307 + itag = ip->i_d.di_tag;
8308 gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
8309 uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
8310 + tag = (mask & ATTR_TAG) ? iattr->ia_tag : itag;
8313 * Do a quota reservation only if uid/gid is actually
8314 @@ -183,7 +260,8 @@ xfs_setattr(
8316 if (XFS_IS_QUOTA_RUNNING(mp) &&
8317 ((XFS_IS_UQUOTA_ON(mp) && iuid != uid) ||
8318 - (XFS_IS_GQUOTA_ON(mp) && igid != gid))) {
8319 + (XFS_IS_GQUOTA_ON(mp) && igid != gid) ||
8320 + (XFS_IS_GQUOTA_ON(mp) && itag != tag))) {
8322 code = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
8323 capable(CAP_FOWNER) ?
8324 @@ -336,7 +414,7 @@ xfs_setattr(
8326 * Change file ownership. Must be the owner or privileged.
8328 - if (mask & (ATTR_UID|ATTR_GID)) {
8329 + if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8331 * CAP_FSETID overrides the following restrictions:
8333 @@ -352,6 +430,10 @@ xfs_setattr(
8334 * Change the ownerships and register quota modifications
8335 * in the transaction.
8337 + if (itag != tag) {
8338 + ip->i_d.di_tag = tag;
8339 + inode->i_tag = tag;
8342 if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_UQUOTA_ON(mp)) {
8343 ASSERT(mask & ATTR_UID);
8344 diff -NurpP --minimal linux-2.6.32.1/fs/xfs/xfs_vnodeops.h linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_vnodeops.h
8345 --- linux-2.6.32.1/fs/xfs/xfs_vnodeops.h 2009-09-10 15:26:24.000000000 +0200
8346 +++ linux-2.6.32.1-vs2.3.0.36.27/fs/xfs/xfs_vnodeops.h 2009-12-03 20:04:56.000000000 +0100
8347 @@ -14,6 +14,7 @@ struct xfs_inode;
8351 +int xfs_sync_xflags(struct xfs_inode *ip);
8352 int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
8353 #define XFS_ATTR_DMI 0x01 /* invocation from a DMI function */
8354 #define XFS_ATTR_NONBLOCK 0x02 /* return EAGAIN if operation would block */
8355 diff -NurpP --minimal linux-2.6.32.1/include/asm-generic/tlb.h linux-2.6.32.1-vs2.3.0.36.27/include/asm-generic/tlb.h
8356 --- linux-2.6.32.1/include/asm-generic/tlb.h 2009-09-10 15:26:24.000000000 +0200
8357 +++ linux-2.6.32.1-vs2.3.0.36.27/include/asm-generic/tlb.h 2009-12-03 20:04:56.000000000 +0100
8359 #define _ASM_GENERIC__TLB_H
8361 #include <linux/swap.h>
8362 +#include <linux/vs_memory.h>
8363 #include <asm/pgalloc.h>
8364 #include <asm/tlbflush.h>
8366 diff -NurpP --minimal linux-2.6.32.1/include/linux/capability.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/capability.h
8367 --- linux-2.6.32.1/include/linux/capability.h 2009-12-03 20:02:54.000000000 +0100
8368 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/capability.h 2009-12-03 20:04:56.000000000 +0100
8369 @@ -285,6 +285,7 @@ struct cpu_vfs_cap_data {
8370 arbitrary SCSI commands */
8371 /* Allow setting encryption key on loopback filesystem */
8372 /* Allow setting zone reclaim policy */
8373 +/* Allow the selection of a security context */
8375 #define CAP_SYS_ADMIN 21
8377 @@ -357,7 +358,13 @@ struct cpu_vfs_cap_data {
8379 #define CAP_MAC_ADMIN 33
8381 -#define CAP_LAST_CAP CAP_MAC_ADMIN
8382 +/* Allow context manipulations */
8383 +/* Allow changing context info on files */
8385 +#define CAP_CONTEXT 34
8388 +#define CAP_LAST_CAP CAP_CONTEXT
8390 #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
8392 diff -NurpP --minimal linux-2.6.32.1/include/linux/devpts_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/devpts_fs.h
8393 --- linux-2.6.32.1/include/linux/devpts_fs.h 2008-12-25 00:26:37.000000000 +0100
8394 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/devpts_fs.h 2009-12-03 20:04:56.000000000 +0100
8395 @@ -45,5 +45,4 @@ static inline void devpts_pty_kill(struc
8400 #endif /* _LINUX_DEVPTS_FS_H */
8401 diff -NurpP --minimal linux-2.6.32.1/include/linux/ext2_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/ext2_fs.h
8402 --- linux-2.6.32.1/include/linux/ext2_fs.h 2009-03-24 14:22:41.000000000 +0100
8403 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/ext2_fs.h 2009-12-03 20:04:56.000000000 +0100
8404 @@ -189,8 +189,12 @@ struct ext2_group_desc
8405 #define EXT2_NOTAIL_FL FS_NOTAIL_FL /* file tail should not be merged */
8406 #define EXT2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
8407 #define EXT2_TOPDIR_FL FS_TOPDIR_FL /* Top of directory hierarchies*/
8408 +#define EXT2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
8409 #define EXT2_RESERVED_FL FS_RESERVED_FL /* reserved for ext2 lib */
8411 +#define EXT2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
8412 +#define EXT2_COW_FL FS_COW_FL /* Copy on Write marker */
8414 #define EXT2_FL_USER_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */
8415 #define EXT2_FL_USER_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */
8417 @@ -274,7 +278,8 @@ struct ext2_inode {
8419 __le16 l_i_uid_high; /* these 2 fields */
8420 __le16 l_i_gid_high; /* were reserved2[0] */
8421 - __u32 l_i_reserved2;
8422 + __le16 l_i_tag; /* Context Tag */
8423 + __u16 l_i_reserved2;
8426 __u8 h_i_frag; /* Fragment number */
8427 @@ -303,6 +308,7 @@ struct ext2_inode {
8428 #define i_gid_low i_gid
8429 #define i_uid_high osd2.linux2.l_i_uid_high
8430 #define i_gid_high osd2.linux2.l_i_gid_high
8431 +#define i_raw_tag osd2.linux2.l_i_tag
8432 #define i_reserved2 osd2.linux2.l_i_reserved2
8435 @@ -347,6 +353,7 @@ struct ext2_inode {
8436 #define EXT2_MOUNT_USRQUOTA 0x020000 /* user quota */
8437 #define EXT2_MOUNT_GRPQUOTA 0x040000 /* group quota */
8438 #define EXT2_MOUNT_RESERVATION 0x080000 /* Preallocation */
8439 +#define EXT2_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8442 #define clear_opt(o, opt) o &= ~EXT2_MOUNT_##opt
8443 diff -NurpP --minimal linux-2.6.32.1/include/linux/ext3_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/ext3_fs.h
8444 --- linux-2.6.32.1/include/linux/ext3_fs.h 2009-09-10 15:26:25.000000000 +0200
8445 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/ext3_fs.h 2009-12-03 20:04:56.000000000 +0100
8446 @@ -173,10 +173,14 @@ struct ext3_group_desc
8447 #define EXT3_NOTAIL_FL 0x00008000 /* file tail should not be merged */
8448 #define EXT3_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
8449 #define EXT3_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8450 +#define EXT3_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8451 #define EXT3_RESERVED_FL 0x80000000 /* reserved for ext3 lib */
8453 -#define EXT3_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8454 -#define EXT3_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8455 +#define EXT3_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8456 +#define EXT3_COW_FL 0x20000000 /* Copy on Write marker */
8458 +#define EXT3_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8459 +#define EXT3_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8461 /* Flags that should be inherited by new inodes from their parent. */
8462 #define EXT3_FL_INHERITED (EXT3_SECRM_FL | EXT3_UNRM_FL | EXT3_COMPR_FL |\
8463 @@ -320,7 +324,8 @@ struct ext3_inode {
8465 __le16 l_i_uid_high; /* these 2 fields */
8466 __le16 l_i_gid_high; /* were reserved2[0] */
8467 - __u32 l_i_reserved2;
8468 + __le16 l_i_tag; /* Context Tag */
8469 + __u16 l_i_reserved2;
8472 __u8 h_i_frag; /* Fragment number */
8473 @@ -351,6 +356,7 @@ struct ext3_inode {
8474 #define i_gid_low i_gid
8475 #define i_uid_high osd2.linux2.l_i_uid_high
8476 #define i_gid_high osd2.linux2.l_i_gid_high
8477 +#define i_raw_tag osd2.linux2.l_i_tag
8478 #define i_reserved2 osd2.linux2.l_i_reserved2
8480 #elif defined(__GNU__)
8481 @@ -414,6 +420,7 @@ struct ext3_inode {
8482 #define EXT3_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
8483 #define EXT3_MOUNT_DATA_ERR_ABORT 0x400000 /* Abort on file data write
8484 * error in ordered mode */
8485 +#define EXT3_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8487 /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
8488 #ifndef _LINUX_EXT2_FS_H
8489 @@ -892,6 +899,7 @@ extern void ext3_get_inode_flags(struct
8490 extern void ext3_set_aops(struct inode *inode);
8491 extern int ext3_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
8492 u64 start, u64 len);
8493 +extern int ext3_sync_flags(struct inode *, int, int);
8496 extern long ext3_ioctl(struct file *, unsigned int, unsigned long);
8497 diff -NurpP --minimal linux-2.6.32.1/include/linux/fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/fs.h
8498 --- linux-2.6.32.1/include/linux/fs.h 2009-12-03 20:02:55.000000000 +0100
8499 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/fs.h 2009-12-03 20:04:56.000000000 +0100
8500 @@ -205,6 +205,9 @@ struct inodes_stat_t {
8501 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
8502 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
8503 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
8504 +#define MS_TAGGED (1<<25) /* use generic inode tagging */
8505 +#define MS_TAGID (1<<26) /* use specific tag for this mount */
8506 +#define MS_NOTAGCHECK (1<<27) /* don't check tags */
8507 #define MS_ACTIVE (1<<30)
8508 #define MS_NOUSER (1<<31)
8510 @@ -231,6 +234,14 @@ struct inodes_stat_t {
8511 #define S_NOCMTIME 128 /* Do not update file c/mtime */
8512 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
8513 #define S_PRIVATE 512 /* Inode is fs-internal */
8514 +#define S_IXUNLINK 1024 /* Immutable Invert on unlink */
8516 +/* Linux-VServer related Inode flags */
8520 +#define V_BARRIER 4 /* Barrier for chroot() */
8521 +#define V_COW 8 /* Copy on Write */
8524 * Note that nosuid etc flags are inode-specific: setting some file-system
8525 @@ -253,12 +264,15 @@ struct inodes_stat_t {
8526 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
8527 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
8528 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
8529 -#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8530 -#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8531 +#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8532 +#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8533 +#define IS_TAGGED(inode) __IS_FLG(inode, MS_TAGGED)
8535 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
8536 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
8537 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
8538 +#define IS_IXUNLINK(inode) ((inode)->i_flags & S_IXUNLINK)
8539 +#define IS_IXORUNLINK(inode) ((IS_IXUNLINK(inode) ? S_IMMUTABLE : 0) ^ IS_IMMUTABLE(inode))
8540 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
8542 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
8543 @@ -266,6 +280,16 @@ struct inodes_stat_t {
8544 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
8545 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
8547 +#define IS_BARRIER(inode) (S_ISDIR((inode)->i_mode) && ((inode)->i_vflags & V_BARRIER))
8549 +#ifdef CONFIG_VSERVER_COWBL
8550 +# define IS_COW(inode) (IS_IXUNLINK(inode) && IS_IMMUTABLE(inode))
8551 +# define IS_COW_LINK(inode) (S_ISREG((inode)->i_mode) && ((inode)->i_nlink > 1))
8553 +# define IS_COW(inode) (0)
8554 +# define IS_COW_LINK(inode) (0)
8557 /* the read-only stuff doesn't really belong here, but any other place is
8558 probably as bad and I don't want to create yet another include file. */
8560 @@ -347,11 +371,14 @@ struct inodes_stat_t {
8561 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8562 #define FS_EXTENT_FL 0x00080000 /* Extents */
8563 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
8564 +#define FS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8565 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
8567 -#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8568 -#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8569 +#define FS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8570 +#define FS_COW_FL 0x20000000 /* Copy on Write marker */
8572 +#define FS_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8573 +#define FS_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8575 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
8576 #define SYNC_FILE_RANGE_WRITE 2
8577 @@ -433,6 +460,7 @@ typedef void (dio_iodone_t)(struct kiocb
8578 #define ATTR_KILL_PRIV (1 << 14)
8579 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
8580 #define ATTR_TIMES_SET (1 << 16)
8581 +#define ATTR_TAG (1 << 17)
8584 * This is the Inode Attributes structure, used for notify_change(). It
8585 @@ -448,6 +476,7 @@ struct iattr {
8591 struct timespec ia_atime;
8592 struct timespec ia_mtime;
8593 @@ -461,6 +490,9 @@ struct iattr {
8594 struct file *ia_file;
8597 +#define ATTR_FLAG_BARRIER 512 /* Barrier for chroot() */
8598 +#define ATTR_FLAG_IXUNLINK 1024 /* Immutable invert on unlink */
8601 * Includes for diskquotas.
8603 @@ -726,7 +758,9 @@ struct inode {
8604 unsigned int i_nlink;
8612 #ifdef __NEED_I_SIZE_ORDERED
8613 @@ -773,7 +807,8 @@ struct inode {
8614 unsigned long i_state;
8615 unsigned long dirtied_when; /* jiffies of first dirtying */
8617 - unsigned int i_flags;
8618 + unsigned short i_flags;
8619 + unsigned short i_vflags;
8621 atomic_t i_writecount;
8622 #ifdef CONFIG_SECURITY
8623 @@ -861,12 +896,12 @@ static inline void i_size_write(struct i
8625 static inline unsigned iminor(const struct inode *inode)
8627 - return MINOR(inode->i_rdev);
8628 + return MINOR(inode->i_mdev);
8631 static inline unsigned imajor(const struct inode *inode)
8633 - return MAJOR(inode->i_rdev);
8634 + return MAJOR(inode->i_mdev);
8637 extern struct block_device *I_BDEV(struct inode *inode);
8638 @@ -925,6 +960,7 @@ struct file {
8640 struct fown_struct f_owner;
8641 const struct cred *f_cred;
8643 struct file_ra_state f_ra;
8646 @@ -1066,6 +1102,7 @@ struct file_lock {
8647 struct file *fl_file;
8652 struct fasync_struct * fl_fasync; /* for lease break notifications */
8653 unsigned long fl_break_time; /* for nonblocking lease breaks */
8654 @@ -1531,6 +1568,7 @@ struct inode_operations {
8655 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
8656 ssize_t (*listxattr) (struct dentry *, char *, size_t);
8657 int (*removexattr) (struct dentry *, const char *);
8658 + int (*sync_flags) (struct inode *, int, int);
8659 void (*truncate_range)(struct inode *, loff_t, loff_t);
8660 long (*fallocate)(struct inode *inode, int mode, loff_t offset,
8662 @@ -1551,6 +1589,7 @@ extern ssize_t vfs_readv(struct file *,
8663 unsigned long, loff_t *);
8664 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
8665 unsigned long, loff_t *);
8666 +ssize_t vfs_sendfile(struct file *, struct file *, loff_t *, size_t, loff_t);
8668 struct super_operations {
8669 struct inode *(*alloc_inode)(struct super_block *sb);
8670 @@ -2347,6 +2386,7 @@ extern int dcache_dir_open(struct inode
8671 extern int dcache_dir_close(struct inode *, struct file *);
8672 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
8673 extern int dcache_readdir(struct file *, void *, filldir_t);
8674 +extern int dcache_readdir_filter(struct file *, void *, filldir_t, int (*)(struct dentry *));
8675 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
8676 extern int simple_statfs(struct dentry *, struct kstatfs *);
8677 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
8678 diff -NurpP --minimal linux-2.6.32.1/include/linux/gfs2_ondisk.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/gfs2_ondisk.h
8679 --- linux-2.6.32.1/include/linux/gfs2_ondisk.h 2009-12-03 20:02:55.000000000 +0100
8680 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/gfs2_ondisk.h 2009-12-03 20:04:56.000000000 +0100
8681 @@ -235,6 +235,9 @@ enum {
8685 + gfs2fl_IXUnlink = 16,
8686 + gfs2fl_Barrier = 17,
8688 gfs2fl_TruncInProg = 29,
8689 gfs2fl_InheritDirectio = 30,
8690 gfs2fl_InheritJdata = 31,
8691 @@ -251,6 +254,9 @@ enum {
8692 #define GFS2_DIF_NOATIME 0x00000080
8693 #define GFS2_DIF_SYNC 0x00000100
8694 #define GFS2_DIF_SYSTEM 0x00000200 /* New in gfs2 */
8695 +#define GFS2_DIF_IXUNLINK 0x00010000
8696 +#define GFS2_DIF_BARRIER 0x00020000
8697 +#define GFS2_DIF_COW 0x00040000
8698 #define GFS2_DIF_TRUNC_IN_PROG 0x20000000 /* New in gfs2 */
8699 #define GFS2_DIF_INHERIT_DIRECTIO 0x40000000
8700 #define GFS2_DIF_INHERIT_JDATA 0x80000000
8701 diff -NurpP --minimal linux-2.6.32.1/include/linux/if_tun.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/if_tun.h
8702 --- linux-2.6.32.1/include/linux/if_tun.h 2009-12-03 20:02:55.000000000 +0100
8703 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/if_tun.h 2009-12-03 20:04:56.000000000 +0100
8705 #define TUNGETIFF _IOR('T', 210, unsigned int)
8706 #define TUNGETSNDBUF _IOR('T', 211, int)
8707 #define TUNSETSNDBUF _IOW('T', 212, int)
8708 +#define TUNSETNID _IOW('T', 215, int)
8710 /* TUNSETIFF ifr flags */
8711 #define IFF_TUN 0x0001
8712 diff -NurpP --minimal linux-2.6.32.1/include/linux/init_task.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/init_task.h
8713 --- linux-2.6.32.1/include/linux/init_task.h 2009-12-03 20:02:55.000000000 +0100
8714 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/init_task.h 2009-12-03 20:04:56.000000000 +0100
8715 @@ -184,6 +184,10 @@ extern struct cred init_cred;
8717 INIT_TRACE_RECURSION \
8718 INIT_TASK_RCU_PREEMPT(tsk) \
8720 + .vx_info = NULL, \
8722 + .nx_info = NULL, \
8726 diff -NurpP --minimal linux-2.6.32.1/include/linux/ipc.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/ipc.h
8727 --- linux-2.6.32.1/include/linux/ipc.h 2009-12-03 20:02:55.000000000 +0100
8728 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/ipc.h 2009-12-03 20:04:56.000000000 +0100
8729 @@ -91,6 +91,7 @@ struct kern_ipc_perm
8737 diff -NurpP --minimal linux-2.6.32.1/include/linux/Kbuild linux-2.6.32.1-vs2.3.0.36.27/include/linux/Kbuild
8738 --- linux-2.6.32.1/include/linux/Kbuild 2009-12-03 20:02:54.000000000 +0100
8739 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/Kbuild 2009-12-03 20:04:56.000000000 +0100
8740 @@ -382,5 +382,8 @@ unifdef-y += xattr.h
8743 objhdr-y += version.h
8745 +header-y += vserver/
8749 diff -NurpP --minimal linux-2.6.32.1/include/linux/loop.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/loop.h
8750 --- linux-2.6.32.1/include/linux/loop.h 2009-09-10 15:26:25.000000000 +0200
8751 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/loop.h 2009-12-03 20:04:56.000000000 +0100
8752 @@ -45,6 +45,7 @@ struct loop_device {
8753 struct loop_func_table *lo_encryption;
8755 uid_t lo_key_owner; /* Who set the key */
8757 int (*ioctl)(struct loop_device *, int cmd,
8760 diff -NurpP --minimal linux-2.6.32.1/include/linux/magic.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/magic.h
8761 --- linux-2.6.32.1/include/linux/magic.h 2009-12-03 20:02:55.000000000 +0100
8762 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/magic.h 2009-12-03 20:04:56.000000000 +0100
8765 #define ADFS_SUPER_MAGIC 0xadf5
8766 #define AFFS_SUPER_MAGIC 0xadff
8767 -#define AFS_SUPER_MAGIC 0x5346414F
8768 +#define AFS_SUPER_MAGIC 0x5346414F
8769 #define AUTOFS_SUPER_MAGIC 0x0187
8770 #define CODA_SUPER_MAGIC 0x73757245
8771 #define CRAMFS_MAGIC 0x28cd3d45 /* some random number */
8773 #define NFS_SUPER_MAGIC 0x6969
8774 #define OPENPROM_SUPER_MAGIC 0x9fa1
8775 #define PROC_SUPER_MAGIC 0x9fa0
8776 +#define DEVPTS_SUPER_MAGIC 0x1cd1
8777 #define QNX4_SUPER_MAGIC 0x002f /* qnx4 fs detection */
8779 #define REISERFS_SUPER_MAGIC 0x52654973 /* used by gcc */
8780 diff -NurpP --minimal linux-2.6.32.1/include/linux/major.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/major.h
8781 --- linux-2.6.32.1/include/linux/major.h 2009-09-10 15:26:25.000000000 +0200
8782 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/major.h 2009-12-03 20:04:56.000000000 +0100
8784 #define HD_MAJOR IDE0_MAJOR
8785 #define PTY_SLAVE_MAJOR 3
8787 +#define VROOT_MAJOR 4
8788 #define TTYAUX_MAJOR 5
8791 diff -NurpP --minimal linux-2.6.32.1/include/linux/mm_types.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/mm_types.h
8792 --- linux-2.6.32.1/include/linux/mm_types.h 2009-12-03 20:02:55.000000000 +0100
8793 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/mm_types.h 2009-12-03 20:04:56.000000000 +0100
8794 @@ -246,6 +246,7 @@ struct mm_struct {
8796 /* Architecture-specific MM context */
8797 mm_context_t context;
8798 + struct vx_info *mm_vx_info;
8800 /* Swap token stuff */
8802 diff -NurpP --minimal linux-2.6.32.1/include/linux/mount.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/mount.h
8803 --- linux-2.6.32.1/include/linux/mount.h 2009-09-10 15:26:25.000000000 +0200
8804 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/mount.h 2009-12-03 20:04:56.000000000 +0100
8805 @@ -36,6 +36,9 @@ struct mnt_namespace;
8806 #define MNT_UNBINDABLE 0x2000 /* if the vfsmount is a unbindable mount */
8807 #define MNT_PNODE_MASK 0x3000 /* propagation flag mask */
8809 +#define MNT_TAGID 0x10000
8810 +#define MNT_NOTAG 0x20000
8813 struct list_head mnt_hash;
8814 struct vfsmount *mnt_parent; /* fs we are mounted on */
8815 @@ -70,6 +73,7 @@ struct vfsmount {
8819 + tag_t mnt_tag; /* tagging used for vfsmount */
8822 static inline int *get_mnt_writers_ptr(struct vfsmount *mnt)
8823 diff -NurpP --minimal linux-2.6.32.1/include/linux/net.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/net.h
8824 --- linux-2.6.32.1/include/linux/net.h 2009-12-03 20:02:55.000000000 +0100
8825 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/net.h 2009-12-03 20:04:56.000000000 +0100
8826 @@ -69,6 +69,7 @@ struct net;
8827 #define SOCK_NOSPACE 2
8828 #define SOCK_PASSCRED 3
8829 #define SOCK_PASSSEC 4
8830 +#define SOCK_USER_SOCKET 5
8832 #ifndef ARCH_HAS_SOCKET_TYPES
8834 diff -NurpP --minimal linux-2.6.32.1/include/linux/nfs_mount.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/nfs_mount.h
8835 --- linux-2.6.32.1/include/linux/nfs_mount.h 2009-03-24 14:22:43.000000000 +0100
8836 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/nfs_mount.h 2009-12-03 20:04:56.000000000 +0100
8837 @@ -63,7 +63,8 @@ struct nfs_mount_data {
8838 #define NFS_MOUNT_SECFLAVOUR 0x2000 /* 5 */
8839 #define NFS_MOUNT_NORDIRPLUS 0x4000 /* 5 */
8840 #define NFS_MOUNT_UNSHARED 0x8000 /* 5 */
8841 -#define NFS_MOUNT_FLAGMASK 0xFFFF
8842 +#define NFS_MOUNT_TAGGED 0x10000 /* context tagging */
8843 +#define NFS_MOUNT_FLAGMASK 0x1FFFF
8845 /* The following are for internal use only */
8846 #define NFS_MOUNT_LOOKUP_CACHE_NONEG 0x10000
8847 diff -NurpP --minimal linux-2.6.32.1/include/linux/nsproxy.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/nsproxy.h
8848 --- linux-2.6.32.1/include/linux/nsproxy.h 2009-06-11 17:13:17.000000000 +0200
8849 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/nsproxy.h 2009-12-03 20:04:56.000000000 +0100
8852 #include <linux/spinlock.h>
8853 #include <linux/sched.h>
8854 +#include <linux/vserver/debug.h>
8856 struct mnt_namespace;
8857 struct uts_namespace;
8858 @@ -63,22 +64,33 @@ static inline struct nsproxy *task_nspro
8861 int copy_namespaces(unsigned long flags, struct task_struct *tsk);
8862 +struct nsproxy *copy_nsproxy(struct nsproxy *orig);
8863 void exit_task_namespaces(struct task_struct *tsk);
8864 void switch_task_namespaces(struct task_struct *tsk, struct nsproxy *new);
8865 void free_nsproxy(struct nsproxy *ns);
8866 int unshare_nsproxy_namespaces(unsigned long, struct nsproxy **,
8867 struct fs_struct *);
8869 -static inline void put_nsproxy(struct nsproxy *ns)
8870 +#define get_nsproxy(n) __get_nsproxy(n, __FILE__, __LINE__)
8872 +static inline void __get_nsproxy(struct nsproxy *ns,
8873 + const char *_file, int _line)
8875 - if (atomic_dec_and_test(&ns->count)) {
8878 + vxlprintk(VXD_CBIT(space, 0), "get_nsproxy(%p[%u])",
8879 + ns, atomic_read(&ns->count), _file, _line);
8880 + atomic_inc(&ns->count);
8883 -static inline void get_nsproxy(struct nsproxy *ns)
8884 +#define put_nsproxy(n) __put_nsproxy(n, __FILE__, __LINE__)
8886 +static inline void __put_nsproxy(struct nsproxy *ns,
8887 + const char *_file, int _line)
8889 - atomic_inc(&ns->count);
8890 + vxlprintk(VXD_CBIT(space, 0), "put_nsproxy(%p[%u])",
8891 + ns, atomic_read(&ns->count), _file, _line);
8892 + if (atomic_dec_and_test(&ns->count)) {
8897 #ifdef CONFIG_CGROUP_NS
8898 diff -NurpP --minimal linux-2.6.32.1/include/linux/pid.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/pid.h
8899 --- linux-2.6.32.1/include/linux/pid.h 2009-03-24 14:22:43.000000000 +0100
8900 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/pid.h 2009-12-03 20:04:56.000000000 +0100
8901 @@ -8,7 +8,8 @@ enum pid_type
8911 @@ -160,6 +161,7 @@ static inline pid_t pid_nr(struct pid *p
8914 pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns);
8915 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns);
8916 pid_t pid_vnr(struct pid *pid);
8918 #define do_each_pid_task(pid, type, task) \
8919 diff -NurpP --minimal linux-2.6.32.1/include/linux/proc_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/proc_fs.h
8920 --- linux-2.6.32.1/include/linux/proc_fs.h 2009-12-03 20:02:56.000000000 +0100
8921 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/proc_fs.h 2009-12-03 20:04:56.000000000 +0100
8922 @@ -56,6 +56,7 @@ struct proc_dir_entry {
8928 const struct inode_operations *proc_iops;
8930 @@ -250,12 +251,18 @@ kclist_add(struct kcore_list *new, void
8931 extern void kclist_add(struct kcore_list *, void *, size_t, int type);
8938 int (*proc_get_link)(struct inode *, struct path *);
8939 int (*proc_read)(struct task_struct *task, char *page);
8940 int (*proc_show)(struct seq_file *m,
8941 struct pid_namespace *ns, struct pid *pid,
8942 struct task_struct *task);
8943 + int (*proc_vs_read)(char *page);
8944 + int (*proc_vxi_read)(struct vx_info *vxi, char *page);
8945 + int (*proc_nxi_read)(struct nx_info *nxi, char *page);
8948 struct ctl_table_header;
8949 @@ -263,6 +270,7 @@ struct ctl_table;
8956 struct proc_dir_entry *pde;
8957 diff -NurpP --minimal linux-2.6.32.1/include/linux/quotaops.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/quotaops.h
8958 --- linux-2.6.32.1/include/linux/quotaops.h 2009-12-03 20:02:56.000000000 +0100
8959 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/quotaops.h 2009-12-03 20:04:56.000000000 +0100
8961 #define _LINUX_QUOTAOPS_
8963 #include <linux/fs.h>
8964 +#include <linux/vs_dlimit.h>
8966 static inline struct quota_info *sb_dqopt(struct super_block *sb)
8968 @@ -154,10 +155,14 @@ static inline void vfs_dq_init(struct in
8969 * a transaction (deadlocks possible otherwise) */
8970 static inline int vfs_dq_prealloc_space_nodirty(struct inode *inode, qsize_t nr)
8972 + if (dl_alloc_space(inode, nr))
8974 if (sb_any_quota_active(inode->i_sb)) {
8975 /* Used space is updated in alloc_space() */
8976 - if (inode->i_sb->dq_op->alloc_space(inode, nr, 1) == NO_QUOTA)
8977 + if (inode->i_sb->dq_op->alloc_space(inode, nr, 1) == NO_QUOTA) {
8978 + dl_free_space(inode, nr);
8983 inode_add_bytes(inode, nr);
8984 @@ -174,10 +179,14 @@ static inline int vfs_dq_prealloc_space(
8986 static inline int vfs_dq_alloc_space_nodirty(struct inode *inode, qsize_t nr)
8988 + if (dl_alloc_space(inode, nr))
8990 if (sb_any_quota_active(inode->i_sb)) {
8991 /* Used space is updated in alloc_space() */
8992 - if (inode->i_sb->dq_op->alloc_space(inode, nr, 0) == NO_QUOTA)
8993 + if (inode->i_sb->dq_op->alloc_space(inode, nr, 0) == NO_QUOTA) {
8994 + dl_free_space(inode, nr);
8999 inode_add_bytes(inode, nr);
9000 @@ -194,20 +203,28 @@ static inline int vfs_dq_alloc_space(str
9002 static inline int vfs_dq_reserve_space(struct inode *inode, qsize_t nr)
9004 + if (dl_reserve_space(inode, nr))
9006 if (sb_any_quota_active(inode->i_sb)) {
9007 /* Used space is updated in alloc_space() */
9008 - if (inode->i_sb->dq_op->reserve_space(inode, nr, 0) == NO_QUOTA)
9009 + if (inode->i_sb->dq_op->reserve_space(inode, nr, 0) == NO_QUOTA) {
9010 + dl_release_space(inode, nr);
9017 static inline int vfs_dq_alloc_inode(struct inode *inode)
9019 + if (dl_alloc_inode(inode))
9021 if (sb_any_quota_active(inode->i_sb)) {
9023 - if (inode->i_sb->dq_op->alloc_inode(inode, 1) == NO_QUOTA)
9024 + if (inode->i_sb->dq_op->alloc_inode(inode, 1) == NO_QUOTA) {
9025 + dl_free_inode(inode);
9031 @@ -217,9 +234,13 @@ static inline int vfs_dq_alloc_inode(str
9033 static inline int vfs_dq_claim_space(struct inode *inode, qsize_t nr)
9035 + if (dl_claim_space(inode, nr))
9037 if (sb_any_quota_active(inode->i_sb)) {
9038 - if (inode->i_sb->dq_op->claim_space(inode, nr) == NO_QUOTA)
9039 + if (inode->i_sb->dq_op->claim_space(inode, nr) == NO_QUOTA) {
9040 + dl_release_space(inode, nr);
9044 inode_add_bytes(inode, nr);
9046 @@ -235,6 +256,7 @@ void vfs_dq_release_reservation_space(st
9048 if (sb_any_quota_active(inode->i_sb))
9049 inode->i_sb->dq_op->release_rsv(inode, nr);
9050 + dl_release_space(inode, nr);
9053 static inline void vfs_dq_free_space_nodirty(struct inode *inode, qsize_t nr)
9054 @@ -243,6 +265,7 @@ static inline void vfs_dq_free_space_nod
9055 inode->i_sb->dq_op->free_space(inode, nr);
9057 inode_sub_bytes(inode, nr);
9058 + dl_free_space(inode, nr);
9061 static inline void vfs_dq_free_space(struct inode *inode, qsize_t nr)
9062 @@ -255,6 +278,7 @@ static inline void vfs_dq_free_inode(str
9064 if (sb_any_quota_active(inode->i_sb))
9065 inode->i_sb->dq_op->free_inode(inode, 1);
9066 + dl_free_inode(inode);
9069 /* Cannot be called inside a transaction */
9070 @@ -358,6 +382,8 @@ static inline int vfs_dq_transfer(struct
9072 static inline int vfs_dq_prealloc_space_nodirty(struct inode *inode, qsize_t nr)
9074 + if (dl_alloc_space(inode, nr))
9076 inode_add_bytes(inode, nr);
9079 @@ -371,6 +397,8 @@ static inline int vfs_dq_prealloc_space(
9081 static inline int vfs_dq_alloc_space_nodirty(struct inode *inode, qsize_t nr)
9083 + if (dl_alloc_space(inode, nr))
9085 inode_add_bytes(inode, nr);
9088 @@ -384,22 +412,28 @@ static inline int vfs_dq_alloc_space(str
9090 static inline int vfs_dq_reserve_space(struct inode *inode, qsize_t nr)
9092 + if (dl_reserve_space(inode, nr))
9097 static inline int vfs_dq_claim_space(struct inode *inode, qsize_t nr)
9099 + if (dl_claim_space(inode, nr))
9101 return vfs_dq_alloc_space(inode, nr);
9105 int vfs_dq_release_reservation_space(struct inode *inode, qsize_t nr)
9107 + dl_release_space(inode, nr);
9111 static inline void vfs_dq_free_space_nodirty(struct inode *inode, qsize_t nr)
9113 + dl_free_space(inode, nr);
9114 inode_sub_bytes(inode, nr);
9117 diff -NurpP --minimal linux-2.6.32.1/include/linux/reboot.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/reboot.h
9118 --- linux-2.6.32.1/include/linux/reboot.h 2008-12-25 00:26:37.000000000 +0100
9119 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/reboot.h 2009-12-03 22:06:59.000000000 +0100
9121 #define LINUX_REBOOT_CMD_RESTART2 0xA1B2C3D4
9122 #define LINUX_REBOOT_CMD_SW_SUSPEND 0xD000FCE2
9123 #define LINUX_REBOOT_CMD_KEXEC 0x45584543
9124 +#define LINUX_REBOOT_CMD_OOM 0xDEADBEEF
9128 diff -NurpP --minimal linux-2.6.32.1/include/linux/reiserfs_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/reiserfs_fs.h
9129 --- linux-2.6.32.1/include/linux/reiserfs_fs.h 2009-09-10 15:26:26.000000000 +0200
9130 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/reiserfs_fs.h 2009-12-03 20:04:56.000000000 +0100
9131 @@ -899,6 +899,11 @@ struct stat_data_v1 {
9132 #define REISERFS_COMPR_FL FS_COMPR_FL
9133 #define REISERFS_NOTAIL_FL FS_NOTAIL_FL
9135 +/* unfortunately reiserfs sdattr is only 16 bit */
9136 +#define REISERFS_IXUNLINK_FL (FS_IXUNLINK_FL >> 16)
9137 +#define REISERFS_BARRIER_FL (FS_BARRIER_FL >> 16)
9138 +#define REISERFS_COW_FL (FS_COW_FL >> 16)
9140 /* persistent flags that file inherits from the parent directory */
9141 #define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL | \
9142 REISERFS_SYNC_FL | \
9143 @@ -908,6 +913,9 @@ struct stat_data_v1 {
9144 REISERFS_COMPR_FL | \
9145 REISERFS_NOTAIL_FL )
9147 +#define REISERFS_FL_USER_VISIBLE 0x80FF
9148 +#define REISERFS_FL_USER_MODIFIABLE 0x80FF
9150 /* Stat Data on disk (reiserfs version of UFS disk inode minus the
9153 @@ -1989,6 +1997,7 @@ static inline void reiserfs_update_sd(st
9154 void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
9155 void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
9156 int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);
9157 +int reiserfs_sync_flags(struct inode *inode, int, int);
9160 void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
9161 diff -NurpP --minimal linux-2.6.32.1/include/linux/reiserfs_fs_sb.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/reiserfs_fs_sb.h
9162 --- linux-2.6.32.1/include/linux/reiserfs_fs_sb.h 2009-09-10 15:26:26.000000000 +0200
9163 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/reiserfs_fs_sb.h 2009-12-03 20:04:56.000000000 +0100
9164 @@ -456,6 +456,7 @@ enum reiserfs_mount_options {
9165 REISERFS_EXPOSE_PRIVROOT,
9166 REISERFS_BARRIER_NONE,
9167 REISERFS_BARRIER_FLUSH,
9170 /* Actions on error */
9171 REISERFS_ERROR_PANIC,
9172 diff -NurpP --minimal linux-2.6.32.1/include/linux/sched.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/sched.h
9173 --- linux-2.6.32.1/include/linux/sched.h 2009-12-14 21:29:46.000000000 +0100
9174 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/sched.h 2009-12-14 22:20:55.000000000 +0100
9175 @@ -390,25 +390,28 @@ extern void arch_unmap_area_topdown(stru
9176 * The mm counters are not protected by its page_table_lock,
9177 * so must be incremented atomically.
9179 -#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
9180 -#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
9181 -#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
9182 -#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
9183 -#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
9184 +#define __set_mm_counter(mm, member, value) \
9185 + atomic_long_set(&(mm)->_##member, value)
9186 +#define get_mm_counter(mm, member) \
9187 + ((unsigned long)atomic_long_read(&(mm)->_##member))
9189 #else /* !USE_SPLIT_PTLOCKS */
9191 * The mm counters are protected by its page_table_lock,
9192 * so can be incremented directly.
9194 -#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
9195 +#define __set_mm_counter(mm, member, value) (mm)->_##member = (value)
9196 #define get_mm_counter(mm, member) ((mm)->_##member)
9197 -#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
9198 -#define inc_mm_counter(mm, member) (mm)->_##member++
9199 -#define dec_mm_counter(mm, member) (mm)->_##member--
9201 #endif /* !USE_SPLIT_PTLOCKS */
9203 +#define set_mm_counter(mm, member, value) \
9204 + vx_ ## member ## pages_sub((mm), (get_mm_counter(mm, member) - value))
9205 +#define add_mm_counter(mm, member, value) \
9206 + vx_ ## member ## pages_add((mm), (value))
9207 +#define inc_mm_counter(mm, member) vx_ ## member ## pages_inc((mm))
9208 +#define dec_mm_counter(mm, member) vx_ ## member ## pages_dec((mm))
9210 #define get_mm_rss(mm) \
9211 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
9212 #define update_hiwater_rss(mm) do { \
9213 @@ -1183,6 +1186,12 @@ struct sched_entity {
9214 u64 nr_wakeups_affine_attempts;
9215 u64 nr_wakeups_passive;
9216 u64 nr_wakeups_idle;
9217 +#ifdef CONFIG_CFS_HARD_LIMITS
9218 + u64 throttle_start;
9220 + u64 throttle_count;
9225 #ifdef CONFIG_FAIR_GROUP_SCHED
9226 @@ -1393,6 +1402,14 @@ struct task_struct {
9230 +/* vserver context data */
9231 + struct vx_info *vx_info;
9232 + struct nx_info *nx_info;
9238 /* Thread group tracking */
9241 @@ -1618,6 +1635,11 @@ struct pid_namespace;
9242 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
9243 struct pid_namespace *ns);
9245 +#include <linux/vserver/base.h>
9246 +#include <linux/vserver/context.h>
9247 +#include <linux/vserver/debug.h>
9248 +#include <linux/vserver/pid.h>
9250 static inline pid_t task_pid_nr(struct task_struct *tsk)
9253 @@ -1631,7 +1653,8 @@ static inline pid_t task_pid_nr_ns(struc
9255 static inline pid_t task_pid_vnr(struct task_struct *tsk)
9257 - return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
9258 + // return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
9259 + return vx_map_pid(__task_pid_nr_ns(tsk, PIDTYPE_PID, NULL));
9263 @@ -1644,7 +1667,7 @@ pid_t task_tgid_nr_ns(struct task_struct
9265 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
9267 - return pid_vnr(task_tgid(tsk));
9268 + return vx_map_tgid(pid_vnr(task_tgid(tsk)));
9272 diff -NurpP --minimal linux-2.6.32.1/include/linux/shmem_fs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/shmem_fs.h
9273 --- linux-2.6.32.1/include/linux/shmem_fs.h 2009-12-03 20:02:56.000000000 +0100
9274 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/shmem_fs.h 2009-12-03 20:04:56.000000000 +0100
9277 #define SHMEM_NR_DIRECT 16
9279 +#define TMPFS_SUPER_MAGIC 0x01021994
9282 struct shmem_inode_info {
9284 unsigned long flags;
9285 diff -NurpP --minimal linux-2.6.32.1/include/linux/stat.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/stat.h
9286 --- linux-2.6.32.1/include/linux/stat.h 2008-12-25 00:26:37.000000000 +0100
9287 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/stat.h 2009-12-03 20:04:56.000000000 +0100
9288 @@ -66,6 +66,7 @@ struct kstat {
9295 struct timespec atime;
9296 diff -NurpP --minimal linux-2.6.32.1/include/linux/sunrpc/auth.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/sunrpc/auth.h
9297 --- linux-2.6.32.1/include/linux/sunrpc/auth.h 2009-12-03 20:02:56.000000000 +0100
9298 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/sunrpc/auth.h 2009-12-03 20:04:56.000000000 +0100
9304 struct group_info *group_info;
9305 unsigned char machine_cred : 1;
9307 diff -NurpP --minimal linux-2.6.32.1/include/linux/sunrpc/clnt.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/sunrpc/clnt.h
9308 --- linux-2.6.32.1/include/linux/sunrpc/clnt.h 2009-12-03 20:02:56.000000000 +0100
9309 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/sunrpc/clnt.h 2009-12-03 20:04:56.000000000 +0100
9310 @@ -49,7 +49,8 @@ struct rpc_clnt {
9311 unsigned int cl_softrtry : 1,/* soft timeouts */
9312 cl_discrtry : 1,/* disconnect before retry */
9313 cl_autobind : 1,/* use getport() */
9314 - cl_chatty : 1;/* be verbose */
9315 + cl_chatty : 1,/* be verbose */
9316 + cl_tag : 1;/* context tagging */
9318 struct rpc_rtt * cl_rtt; /* RTO estimator data */
9319 const struct rpc_timeout *cl_timeout; /* Timeout strategy */
9320 diff -NurpP --minimal linux-2.6.32.1/include/linux/syscalls.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/syscalls.h
9321 --- linux-2.6.32.1/include/linux/syscalls.h 2009-12-03 20:02:56.000000000 +0100
9322 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/syscalls.h 2009-12-03 20:04:56.000000000 +0100
9323 @@ -546,6 +546,8 @@ asmlinkage long sys_symlink(const char _
9324 asmlinkage long sys_unlink(const char __user *pathname);
9325 asmlinkage long sys_rename(const char __user *oldname,
9326 const char __user *newname);
9327 +asmlinkage long sys_copyfile(const char __user *from, const char __user *to,
9329 asmlinkage long sys_chmod(const char __user *filename, mode_t mode);
9330 asmlinkage long sys_fchmod(unsigned int fd, mode_t mode);
9332 diff -NurpP --minimal linux-2.6.32.1/include/linux/sysctl.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/sysctl.h
9333 --- linux-2.6.32.1/include/linux/sysctl.h 2009-12-03 20:02:56.000000000 +0100
9334 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/sysctl.h 2009-12-03 20:04:56.000000000 +0100
9335 @@ -69,6 +69,7 @@ enum
9336 CTL_ABI=9, /* Binary emulation */
9337 CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
9338 CTL_ARLAN=254, /* arlan wireless driver */
9339 + CTL_VSERVER=4242, /* Linux-VServer debug */
9340 CTL_S390DBF=5677, /* s390 debug */
9341 CTL_SUNRPC=7249, /* sunrpc debug */
9342 CTL_PM=9899, /* frv power management */
9343 @@ -103,6 +104,7 @@ enum
9345 KERN_PANIC=15, /* int: panic timeout */
9346 KERN_REALROOTDEV=16, /* real root device to mount after initrd */
9347 + KERN_VSHELPER=17, /* string: path to vshelper policy agent */
9349 KERN_SPARC_REBOOT=21, /* reboot command on Sparc */
9350 KERN_CTLALTDEL=22, /* int: allow ctl-alt-del to reboot */
9351 diff -NurpP --minimal linux-2.6.32.1/include/linux/sysfs.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/sysfs.h
9352 --- linux-2.6.32.1/include/linux/sysfs.h 2008-12-25 00:26:37.000000000 +0100
9353 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/sysfs.h 2009-12-03 20:04:56.000000000 +0100
9355 #include <linux/list.h>
9356 #include <asm/atomic.h>
9358 +#define SYSFS_SUPER_MAGIC 0x62656572
9363 diff -NurpP --minimal linux-2.6.32.1/include/linux/time.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/time.h
9364 --- linux-2.6.32.1/include/linux/time.h 2009-12-03 20:02:56.000000000 +0100
9365 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/time.h 2009-12-03 20:04:56.000000000 +0100
9366 @@ -237,6 +237,9 @@ static __always_inline void timespec_add
9367 a->tv_sec += __iter_div_u64_rem(a->tv_nsec + ns, NSEC_PER_SEC, &ns);
9371 +#include <linux/vs_time.h>
9373 #endif /* __KERNEL__ */
9375 #define NFDBITS __NFDBITS
9376 diff -NurpP --minimal linux-2.6.32.1/include/linux/types.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/types.h
9377 --- linux-2.6.32.1/include/linux/types.h 2009-09-10 15:26:26.000000000 +0200
9378 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/types.h 2009-12-03 20:04:56.000000000 +0100
9379 @@ -37,6 +37,9 @@ typedef __kernel_uid32_t uid_t;
9380 typedef __kernel_gid32_t gid_t;
9381 typedef __kernel_uid16_t uid16_t;
9382 typedef __kernel_gid16_t gid16_t;
9383 +typedef unsigned int xid_t;
9384 +typedef unsigned int nid_t;
9385 +typedef unsigned int tag_t;
9387 typedef unsigned long uintptr_t;
9389 diff -NurpP --minimal linux-2.6.32.1/include/linux/vroot.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vroot.h
9390 --- linux-2.6.32.1/include/linux/vroot.h 1970-01-01 01:00:00.000000000 +0100
9391 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vroot.h 2009-12-03 20:04:56.000000000 +0100
9395 + * include/linux/vroot.h
9397 + * written by Herbert Pötzl, 9/11/2002
9398 + * ported to 2.6 by Herbert Pötzl, 30/12/2004
9400 + * Copyright (C) 2002-2007 by Herbert Pötzl.
9401 + * Redistribution of this file is permitted under the
9402 + * GNU General Public License.
9405 +#ifndef _LINUX_VROOT_H
9406 +#define _LINUX_VROOT_H
9411 +/* Possible states of device */
9417 +struct vroot_device {
9421 + struct semaphore vr_ctl_mutex;
9422 + struct block_device *vr_device;
9427 +typedef struct block_device *(vroot_grb_func)(struct block_device *);
9429 +extern int register_vroot_grb(vroot_grb_func *);
9430 +extern int unregister_vroot_grb(vroot_grb_func *);
9432 +#endif /* __KERNEL__ */
9434 +#define MAX_VROOT_DEFAULT 8
9437 + * IOCTL commands --- we will commandeer 0x56 ('V')
9440 +#define VROOT_SET_DEV 0x5600
9441 +#define VROOT_CLR_DEV 0x5601
9443 +#endif /* _LINUX_VROOT_H */
9444 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_base.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_base.h
9445 --- linux-2.6.32.1/include/linux/vs_base.h 1970-01-01 01:00:00.000000000 +0100
9446 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_base.h 2009-12-03 20:04:56.000000000 +0100
9451 +#include "vserver/base.h"
9452 +#include "vserver/check.h"
9453 +#include "vserver/debug.h"
9456 +#warning duplicate inclusion
9458 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_context.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_context.h
9459 --- linux-2.6.32.1/include/linux/vs_context.h 1970-01-01 01:00:00.000000000 +0100
9460 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_context.h 2009-12-03 20:04:56.000000000 +0100
9462 +#ifndef _VS_CONTEXT_H
9463 +#define _VS_CONTEXT_H
9465 +#include "vserver/base.h"
9466 +#include "vserver/check.h"
9467 +#include "vserver/context.h"
9468 +#include "vserver/history.h"
9469 +#include "vserver/debug.h"
9471 +#include <linux/sched.h>
9474 +#define get_vx_info(i) __get_vx_info(i, __FILE__, __LINE__, __HERE__)
9476 +static inline struct vx_info *__get_vx_info(struct vx_info *vxi,
9477 + const char *_file, int _line, void *_here)
9482 + vxlprintk(VXD_CBIT(xid, 2), "get_vx_info(%p[#%d.%d])",
9483 + vxi, vxi ? vxi->vx_id : 0,
9484 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9486 + __vxh_get_vx_info(vxi, _here);
9488 + atomic_inc(&vxi->vx_usecnt);
9493 +extern void free_vx_info(struct vx_info *);
9495 +#define put_vx_info(i) __put_vx_info(i, __FILE__, __LINE__, __HERE__)
9497 +static inline void __put_vx_info(struct vx_info *vxi,
9498 + const char *_file, int _line, void *_here)
9503 + vxlprintk(VXD_CBIT(xid, 2), "put_vx_info(%p[#%d.%d])",
9504 + vxi, vxi ? vxi->vx_id : 0,
9505 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9507 + __vxh_put_vx_info(vxi, _here);
9509 + if (atomic_dec_and_test(&vxi->vx_usecnt))
9510 + free_vx_info(vxi);
9514 +#define init_vx_info(p, i) \
9515 + __init_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9517 +static inline void __init_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9518 + const char *_file, int _line, void *_here)
9521 + vxlprintk(VXD_CBIT(xid, 3),
9522 + "init_vx_info(%p[#%d.%d])",
9523 + vxi, vxi ? vxi->vx_id : 0,
9524 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9526 + __vxh_init_vx_info(vxi, vxp, _here);
9528 + atomic_inc(&vxi->vx_usecnt);
9534 +#define set_vx_info(p, i) \
9535 + __set_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9537 +static inline void __set_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9538 + const char *_file, int _line, void *_here)
9540 + struct vx_info *vxo;
9545 + vxlprintk(VXD_CBIT(xid, 3), "set_vx_info(%p[#%d.%d])",
9546 + vxi, vxi ? vxi->vx_id : 0,
9547 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9549 + __vxh_set_vx_info(vxi, vxp, _here);
9551 + atomic_inc(&vxi->vx_usecnt);
9552 + vxo = xchg(vxp, vxi);
9557 +#define clr_vx_info(p) __clr_vx_info(p, __FILE__, __LINE__, __HERE__)
9559 +static inline void __clr_vx_info(struct vx_info **vxp,
9560 + const char *_file, int _line, void *_here)
9562 + struct vx_info *vxo;
9564 + vxo = xchg(vxp, NULL);
9568 + vxlprintk(VXD_CBIT(xid, 3), "clr_vx_info(%p[#%d.%d])",
9569 + vxo, vxo ? vxo->vx_id : 0,
9570 + vxo ? atomic_read(&vxo->vx_usecnt) : 0,
9572 + __vxh_clr_vx_info(vxo, vxp, _here);
9574 + if (atomic_dec_and_test(&vxo->vx_usecnt))
9575 + free_vx_info(vxo);
9579 +#define claim_vx_info(v, p) \
9580 + __claim_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9582 +static inline void __claim_vx_info(struct vx_info *vxi,
9583 + struct task_struct *task,
9584 + const char *_file, int _line, void *_here)
9586 + vxlprintk(VXD_CBIT(xid, 3), "claim_vx_info(%p[#%d.%d.%d]) %p",
9587 + vxi, vxi ? vxi->vx_id : 0,
9588 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9589 + vxi ? atomic_read(&vxi->vx_tasks) : 0,
9590 + task, _file, _line);
9591 + __vxh_claim_vx_info(vxi, task, _here);
9593 + atomic_inc(&vxi->vx_tasks);
9597 +extern void unhash_vx_info(struct vx_info *);
9599 +#define release_vx_info(v, p) \
9600 + __release_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9602 +static inline void __release_vx_info(struct vx_info *vxi,
9603 + struct task_struct *task,
9604 + const char *_file, int _line, void *_here)
9606 + vxlprintk(VXD_CBIT(xid, 3), "release_vx_info(%p[#%d.%d.%d]) %p",
9607 + vxi, vxi ? vxi->vx_id : 0,
9608 + vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9609 + vxi ? atomic_read(&vxi->vx_tasks) : 0,
9610 + task, _file, _line);
9611 + __vxh_release_vx_info(vxi, task, _here);
9615 + if (atomic_dec_and_test(&vxi->vx_tasks))
9616 + unhash_vx_info(vxi);
9620 +#define task_get_vx_info(p) \
9621 + __task_get_vx_info(p, __FILE__, __LINE__, __HERE__)
9623 +static inline struct vx_info *__task_get_vx_info(struct task_struct *p,
9624 + const char *_file, int _line, void *_here)
9626 + struct vx_info *vxi;
9629 + vxlprintk(VXD_CBIT(xid, 5), "task_get_vx_info(%p)",
9631 + vxi = __get_vx_info(p->vx_info, _file, _line, _here);
9637 +static inline void __wakeup_vx_info(struct vx_info *vxi)
9639 + if (waitqueue_active(&vxi->vx_wait))
9640 + wake_up_interruptible(&vxi->vx_wait);
9644 +#define enter_vx_info(v, s) __enter_vx_info(v, s, __FILE__, __LINE__)
9646 +static inline void __enter_vx_info(struct vx_info *vxi,
9647 + struct vx_info_save *vxis, const char *_file, int _line)
9649 + vxlprintk(VXD_CBIT(xid, 5), "enter_vx_info(%p[#%d],%p) %p[#%d,%p]",
9650 + vxi, vxi ? vxi->vx_id : 0, vxis, current,
9651 + current->xid, current->vx_info, _file, _line);
9652 + vxis->vxi = xchg(¤t->vx_info, vxi);
9653 + vxis->xid = current->xid;
9654 + current->xid = vxi ? vxi->vx_id : 0;
9657 +#define leave_vx_info(s) __leave_vx_info(s, __FILE__, __LINE__)
9659 +static inline void __leave_vx_info(struct vx_info_save *vxis,
9660 + const char *_file, int _line)
9662 + vxlprintk(VXD_CBIT(xid, 5), "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]",
9663 + vxis, vxis->xid, vxis->vxi, current,
9664 + current->xid, current->vx_info, _file, _line);
9665 + (void)xchg(¤t->vx_info, vxis->vxi);
9666 + current->xid = vxis->xid;
9670 +static inline void __enter_vx_admin(struct vx_info_save *vxis)
9672 + vxis->vxi = xchg(¤t->vx_info, NULL);
9673 + vxis->xid = xchg(¤t->xid, (xid_t)0);
9676 +static inline void __leave_vx_admin(struct vx_info_save *vxis)
9678 + (void)xchg(¤t->xid, vxis->xid);
9679 + (void)xchg(¤t->vx_info, vxis->vxi);
9682 +#define task_is_init(p) \
9683 + __task_is_init(p, __FILE__, __LINE__, __HERE__)
9685 +static inline int __task_is_init(struct task_struct *p,
9686 + const char *_file, int _line, void *_here)
9688 + int is_init = is_global_init(p);
9692 + is_init = p->vx_info->vx_initpid == p->pid;
9697 +extern void exit_vx_info(struct task_struct *, int);
9698 +extern void exit_vx_info_early(struct task_struct *, int);
9702 +#warning duplicate inclusion
9704 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_cowbl.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_cowbl.h
9705 --- linux-2.6.32.1/include/linux/vs_cowbl.h 1970-01-01 01:00:00.000000000 +0100
9706 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_cowbl.h 2009-12-03 20:04:56.000000000 +0100
9708 +#ifndef _VS_COWBL_H
9709 +#define _VS_COWBL_H
9711 +#include <linux/fs.h>
9712 +#include <linux/dcache.h>
9713 +#include <linux/namei.h>
9715 +extern struct dentry *cow_break_link(const char *pathname);
9717 +static inline int cow_check_and_break(struct path *path)
9719 + struct inode *inode = path->dentry->d_inode;
9722 + /* do we need this check? */
9723 + if (IS_RDONLY(inode))
9726 + if (IS_COW(inode)) {
9727 + if (IS_COW_LINK(inode)) {
9728 + struct dentry *new_dentry, *old_dentry = path->dentry;
9731 + buf = kmalloc(PATH_MAX, GFP_KERNEL);
9735 + pp = d_path(path, buf, PATH_MAX);
9736 + new_dentry = cow_break_link(pp);
9738 + if (!IS_ERR(new_dentry)) {
9739 + path->dentry = new_dentry;
9742 + error = PTR_ERR(new_dentry);
9744 + inode->i_flags &= ~(S_IXUNLINK | S_IMMUTABLE);
9745 + inode->i_ctime = CURRENT_TIME;
9746 + mark_inode_dirty(inode);
9753 +#warning duplicate inclusion
9755 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_cvirt.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_cvirt.h
9756 --- linux-2.6.32.1/include/linux/vs_cvirt.h 1970-01-01 01:00:00.000000000 +0100
9757 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_cvirt.h 2009-12-03 20:04:56.000000000 +0100
9759 +#ifndef _VS_CVIRT_H
9760 +#define _VS_CVIRT_H
9762 +#include "vserver/cvirt.h"
9763 +#include "vserver/context.h"
9764 +#include "vserver/base.h"
9765 +#include "vserver/check.h"
9766 +#include "vserver/debug.h"
9769 +static inline void vx_activate_task(struct task_struct *p)
9771 + struct vx_info *vxi;
9773 + if ((vxi = p->vx_info)) {
9774 + vx_update_load(vxi);
9775 + atomic_inc(&vxi->cvirt.nr_running);
9779 +static inline void vx_deactivate_task(struct task_struct *p)
9781 + struct vx_info *vxi;
9783 + if ((vxi = p->vx_info)) {
9784 + vx_update_load(vxi);
9785 + atomic_dec(&vxi->cvirt.nr_running);
9789 +static inline void vx_uninterruptible_inc(struct task_struct *p)
9791 + struct vx_info *vxi;
9793 + if ((vxi = p->vx_info))
9794 + atomic_inc(&vxi->cvirt.nr_uninterruptible);
9797 +static inline void vx_uninterruptible_dec(struct task_struct *p)
9799 + struct vx_info *vxi;
9801 + if ((vxi = p->vx_info))
9802 + atomic_dec(&vxi->cvirt.nr_uninterruptible);
9807 +#warning duplicate inclusion
9809 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_device.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_device.h
9810 --- linux-2.6.32.1/include/linux/vs_device.h 1970-01-01 01:00:00.000000000 +0100
9811 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_device.h 2009-12-03 20:04:56.000000000 +0100
9813 +#ifndef _VS_DEVICE_H
9814 +#define _VS_DEVICE_H
9816 +#include "vserver/base.h"
9817 +#include "vserver/device.h"
9818 +#include "vserver/debug.h"
9821 +#ifdef CONFIG_VSERVER_DEVICE
9823 +int vs_map_device(struct vx_info *, dev_t, dev_t *, umode_t);
9825 +#define vs_device_perm(v, d, m, p) \
9826 + ((vs_map_device(current_vx_info(), d, NULL, m) & (p)) == (p))
9831 +int vs_map_device(struct vx_info *vxi,
9832 + dev_t device, dev_t *target, umode_t mode)
9839 +#define vs_device_perm(v, d, m, p) ((p) == (p))
9844 +#define vs_map_chrdev(d, t, p) \
9845 + ((vs_map_device(current_vx_info(), d, t, S_IFCHR) & (p)) == (p))
9846 +#define vs_map_blkdev(d, t, p) \
9847 + ((vs_map_device(current_vx_info(), d, t, S_IFBLK) & (p)) == (p))
9849 +#define vs_chrdev_perm(d, p) \
9850 + vs_device_perm(current_vx_info(), d, S_IFCHR, p)
9851 +#define vs_blkdev_perm(d, p) \
9852 + vs_device_perm(current_vx_info(), d, S_IFBLK, p)
9856 +#warning duplicate inclusion
9858 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_dlimit.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_dlimit.h
9859 --- linux-2.6.32.1/include/linux/vs_dlimit.h 1970-01-01 01:00:00.000000000 +0100
9860 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_dlimit.h 2009-12-03 20:04:56.000000000 +0100
9862 +#ifndef _VS_DLIMIT_H
9863 +#define _VS_DLIMIT_H
9865 +#include <linux/fs.h>
9867 +#include "vserver/dlimit.h"
9868 +#include "vserver/base.h"
9869 +#include "vserver/debug.h"
9872 +#define get_dl_info(i) __get_dl_info(i, __FILE__, __LINE__)
9874 +static inline struct dl_info *__get_dl_info(struct dl_info *dli,
9875 + const char *_file, int _line)
9879 + vxlprintk(VXD_CBIT(dlim, 4), "get_dl_info(%p[#%d.%d])",
9880 + dli, dli ? dli->dl_tag : 0,
9881 + dli ? atomic_read(&dli->dl_usecnt) : 0,
9883 + atomic_inc(&dli->dl_usecnt);
9888 +#define free_dl_info(i) \
9889 + call_rcu(&(i)->dl_rcu, rcu_free_dl_info)
9891 +#define put_dl_info(i) __put_dl_info(i, __FILE__, __LINE__)
9893 +static inline void __put_dl_info(struct dl_info *dli,
9894 + const char *_file, int _line)
9898 + vxlprintk(VXD_CBIT(dlim, 4), "put_dl_info(%p[#%d.%d])",
9899 + dli, dli ? dli->dl_tag : 0,
9900 + dli ? atomic_read(&dli->dl_usecnt) : 0,
9902 + if (atomic_dec_and_test(&dli->dl_usecnt))
9903 + free_dl_info(dli);
9907 +#define __dlimit_char(d) ((d) ? '*' : ' ')
9909 +static inline int __dl_alloc_space(struct super_block *sb,
9910 + tag_t tag, dlsize_t nr, const char *file, int line)
9912 + struct dl_info *dli = NULL;
9917 + dli = locate_dl_info(sb, tag);
9921 + spin_lock(&dli->dl_lock);
9922 + ret = (dli->dl_space_used + nr > dli->dl_space_total);
9924 + dli->dl_space_used += nr;
9925 + spin_unlock(&dli->dl_lock);
9928 + vxlprintk(VXD_CBIT(dlim, 1),
9929 + "ALLOC (%p,#%d)%c %lld bytes (%d)",
9930 + sb, tag, __dlimit_char(dli), (long long)nr,
9935 +static inline void __dl_free_space(struct super_block *sb,
9936 + tag_t tag, dlsize_t nr, const char *_file, int _line)
9938 + struct dl_info *dli = NULL;
9942 + dli = locate_dl_info(sb, tag);
9946 + spin_lock(&dli->dl_lock);
9947 + if (dli->dl_space_used > nr)
9948 + dli->dl_space_used -= nr;
9950 + dli->dl_space_used = 0;
9951 + spin_unlock(&dli->dl_lock);
9954 + vxlprintk(VXD_CBIT(dlim, 1),
9955 + "FREE (%p,#%d)%c %lld bytes",
9956 + sb, tag, __dlimit_char(dli), (long long)nr,
9960 +static inline int __dl_alloc_inode(struct super_block *sb,
9961 + tag_t tag, const char *_file, int _line)
9963 + struct dl_info *dli;
9966 + dli = locate_dl_info(sb, tag);
9970 + spin_lock(&dli->dl_lock);
9971 + ret = (dli->dl_inodes_used >= dli->dl_inodes_total);
9973 + dli->dl_inodes_used++;
9974 + spin_unlock(&dli->dl_lock);
9977 + vxlprintk(VXD_CBIT(dlim, 0),
9978 + "ALLOC (%p,#%d)%c inode (%d)",
9979 + sb, tag, __dlimit_char(dli), ret, _file, _line);
9983 +static inline void __dl_free_inode(struct super_block *sb,
9984 + tag_t tag, const char *_file, int _line)
9986 + struct dl_info *dli;
9988 + dli = locate_dl_info(sb, tag);
9992 + spin_lock(&dli->dl_lock);
9993 + if (dli->dl_inodes_used > 1)
9994 + dli->dl_inodes_used--;
9996 + dli->dl_inodes_used = 0;
9997 + spin_unlock(&dli->dl_lock);
10000 + vxlprintk(VXD_CBIT(dlim, 0),
10001 + "FREE (%p,#%d)%c inode",
10002 + sb, tag, __dlimit_char(dli), _file, _line);
10005 +static inline void __dl_adjust_block(struct super_block *sb, tag_t tag,
10006 + unsigned long long *free_blocks, unsigned long long *root_blocks,
10007 + const char *_file, int _line)
10009 + struct dl_info *dli;
10010 + uint64_t broot, bfree;
10012 + dli = locate_dl_info(sb, tag);
10016 + spin_lock(&dli->dl_lock);
10017 + broot = (dli->dl_space_total -
10018 + (dli->dl_space_total >> 10) * dli->dl_nrlmult)
10019 + >> sb->s_blocksize_bits;
10020 + bfree = (dli->dl_space_total - dli->dl_space_used)
10021 + >> sb->s_blocksize_bits;
10022 + spin_unlock(&dli->dl_lock);
10024 + vxlprintk(VXD_CBIT(dlim, 2),
10025 + "ADJUST: %lld,%lld on %lld,%lld [mult=%d]",
10026 + (long long)bfree, (long long)broot,
10027 + *free_blocks, *root_blocks, dli->dl_nrlmult,
10029 + if (free_blocks) {
10030 + if (*free_blocks > bfree)
10031 + *free_blocks = bfree;
10033 + if (root_blocks) {
10034 + if (*root_blocks > broot)
10035 + *root_blocks = broot;
10037 + put_dl_info(dli);
10040 +#define dl_prealloc_space(in, bytes) \
10041 + __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10042 + __FILE__, __LINE__ )
10044 +#define dl_alloc_space(in, bytes) \
10045 + __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10046 + __FILE__, __LINE__ )
10048 +#define dl_reserve_space(in, bytes) \
10049 + __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10050 + __FILE__, __LINE__ )
10052 +#define dl_claim_space(in, bytes) (0)
10054 +#define dl_release_space(in, bytes) \
10055 + __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10056 + __FILE__, __LINE__ )
10058 +#define dl_free_space(in, bytes) \
10059 + __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10060 + __FILE__, __LINE__ )
10064 +#define dl_alloc_inode(in) \
10065 + __dl_alloc_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10067 +#define dl_free_inode(in) \
10068 + __dl_free_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10071 +#define dl_adjust_block(sb, tag, fb, rb) \
10072 + __dl_adjust_block(sb, tag, fb, rb, __FILE__, __LINE__ )
10076 +#warning duplicate inclusion
10078 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/base.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/base.h
10079 --- linux-2.6.32.1/include/linux/vserver/base.h 1970-01-01 01:00:00.000000000 +0100
10080 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/base.h 2009-12-03 20:04:56.000000000 +0100
10082 +#ifndef _VX_BASE_H
10083 +#define _VX_BASE_H
10086 +/* context state changes */
10098 +#define vx_task_xid(t) ((t)->xid)
10100 +#define vx_current_xid() vx_task_xid(current)
10102 +#define current_vx_info() (current->vx_info)
10105 +#define nx_task_nid(t) ((t)->nid)
10107 +#define nx_current_nid() nx_task_nid(current)
10109 +#define current_nx_info() (current->nx_info)
10112 +/* generic flag merging */
10114 +#define vs_check_flags(v, m, f) (((v) & (m)) ^ (f))
10116 +#define vs_mask_flags(v, f, m) (((v) & ~(m)) | ((f) & (m)))
10118 +#define vs_mask_mask(v, f, m) (((v) & ~(m)) | ((v) & (f) & (m)))
10120 +#define vs_check_bit(v, n) ((v) & (1LL << (n)))
10123 +/* context flags */
10125 +#define __vx_flags(v) ((v) ? (v)->vx_flags : 0)
10127 +#define vx_current_flags() __vx_flags(current_vx_info())
10129 +#define vx_info_flags(v, m, f) \
10130 + vs_check_flags(__vx_flags(v), m, f)
10132 +#define task_vx_flags(t, m, f) \
10133 + ((t) && vx_info_flags((t)->vx_info, m, f))
10135 +#define vx_flags(m, f) vx_info_flags(current_vx_info(), m, f)
10138 +/* context caps */
10140 +#define __vx_ccaps(v) ((v) ? (v)->vx_ccaps : 0)
10142 +#define vx_current_ccaps() __vx_ccaps(current_vx_info())
10144 +#define vx_info_ccaps(v, c) (__vx_ccaps(v) & (c))
10146 +#define vx_ccaps(c) vx_info_ccaps(current_vx_info(), (c))
10150 +/* network flags */
10152 +#define __nx_flags(n) ((n) ? (n)->nx_flags : 0)
10154 +#define nx_current_flags() __nx_flags(current_nx_info())
10156 +#define nx_info_flags(n, m, f) \
10157 + vs_check_flags(__nx_flags(n), m, f)
10159 +#define task_nx_flags(t, m, f) \
10160 + ((t) && nx_info_flags((t)->nx_info, m, f))
10162 +#define nx_flags(m, f) nx_info_flags(current_nx_info(), m, f)
10165 +/* network caps */
10167 +#define __nx_ncaps(n) ((n) ? (n)->nx_ncaps : 0)
10169 +#define nx_current_ncaps() __nx_ncaps(current_nx_info())
10171 +#define nx_info_ncaps(n, c) (__nx_ncaps(n) & (c))
10173 +#define nx_ncaps(c) nx_info_ncaps(current_nx_info(), c)
10176 +/* context mask capabilities */
10178 +#define __vx_mcaps(v) ((v) ? (v)->vx_ccaps >> 32UL : ~0 )
10180 +#define vx_info_mcaps(v, c) (__vx_mcaps(v) & (c))
10182 +#define vx_mcaps(c) vx_info_mcaps(current_vx_info(), c)
10185 +/* context bcap mask */
10187 +#define __vx_bcaps(v) ((v)->vx_bcaps)
10189 +#define vx_current_bcaps() __vx_bcaps(current_vx_info())
10192 +/* mask given bcaps */
10194 +#define vx_info_mbcaps(v, c) ((v) ? cap_intersect(__vx_bcaps(v), c) : c)
10196 +#define vx_mbcaps(c) vx_info_mbcaps(current_vx_info(), c)
10199 +/* masked cap_bset */
10201 +#define vx_info_cap_bset(v) vx_info_mbcaps(v, current->cap_bset)
10203 +#define vx_current_cap_bset() vx_info_cap_bset(current_vx_info())
10206 +#define vx_info_mbcap(v, b) \
10207 + (!vx_info_flags(v, VXF_STATE_SETUP, 0) ? \
10208 + vx_info_bcaps(v, b) : (b))
10210 +#define task_vx_mbcap(t, b) \
10211 + vx_info_mbcap((t)->vx_info, (t)->b)
10213 +#define vx_mbcap(b) task_vx_mbcap(current, b)
10216 +#define vx_cap_raised(v, c, f) cap_raised(vx_info_mbcaps(v, c), f)
10218 +#define vx_capable(b, c) (capable(b) || \
10219 + (cap_raised(current_cap(), b) && vx_ccaps(c)))
10221 +#define nx_capable(b, c) (capable(b) || \
10222 + (cap_raised(current_cap(), b) && nx_ncaps(c)))
10224 +#define vx_task_initpid(t, n) \
10225 + ((t)->vx_info && \
10226 + ((t)->vx_info->vx_initpid == (n)))
10228 +#define vx_current_initpid(n) vx_task_initpid(current, n)
10231 +/* context unshare mask */
10233 +#define __vx_umask(v) ((v)->vx_umask)
10235 +#define vx_current_umask() __vx_umask(current_vx_info())
10237 +#define vx_can_unshare(b, f) (capable(b) || \
10238 + (cap_raised(current_cap(), b) && \
10239 + !((f) & ~vx_current_umask())))
10242 +#define __vx_state(v) ((v) ? ((v)->vx_state) : 0)
10244 +#define vx_info_state(v, m) (__vx_state(v) & (m))
10247 +#define __nx_state(n) ((n) ? ((n)->nx_state) : 0)
10249 +#define nx_info_state(n, m) (__nx_state(n) & (m))
10252 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cacct_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_cmd.h
10253 --- linux-2.6.32.1/include/linux/vserver/cacct_cmd.h 1970-01-01 01:00:00.000000000 +0100
10254 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_cmd.h 2009-12-03 20:04:56.000000000 +0100
10256 +#ifndef _VX_CACCT_CMD_H
10257 +#define _VX_CACCT_CMD_H
10260 +/* virtual host info name commands */
10262 +#define VCMD_sock_stat VC_CMD(VSTAT, 5, 0)
10264 +struct vcmd_sock_stat_v0 {
10266 + uint32_t count[3];
10267 + uint64_t total[3];
10273 +#include <linux/compiler.h>
10275 +extern int vc_sock_stat(struct vx_info *, void __user *);
10277 +#endif /* __KERNEL__ */
10278 +#endif /* _VX_CACCT_CMD_H */
10279 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cacct_def.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_def.h
10280 --- linux-2.6.32.1/include/linux/vserver/cacct_def.h 1970-01-01 01:00:00.000000000 +0100
10281 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_def.h 2009-12-03 20:04:56.000000000 +0100
10283 +#ifndef _VX_CACCT_DEF_H
10284 +#define _VX_CACCT_DEF_H
10286 +#include <asm/atomic.h>
10287 +#include <linux/vserver/cacct.h>
10290 +struct _vx_sock_acc {
10291 + atomic_long_t count;
10292 + atomic_long_t total;
10295 +/* context sub struct */
10297 +struct _vx_cacct {
10298 + struct _vx_sock_acc sock[VXA_SOCK_SIZE][3];
10299 + atomic_t slab[8];
10300 + atomic_t page[6][8];
10303 +#ifdef CONFIG_VSERVER_DEBUG
10305 +static inline void __dump_vx_cacct(struct _vx_cacct *cacct)
10309 + printk("\t_vx_cacct:");
10310 + for (i = 0; i < 6; i++) {
10311 + struct _vx_sock_acc *ptr = cacct->sock[i];
10313 + printk("\t [%d] =", i);
10314 + for (j = 0; j < 3; j++) {
10315 + printk(" [%d] = %8lu, %8lu", j,
10316 + atomic_long_read(&ptr[j].count),
10317 + atomic_long_read(&ptr[j].total));
10325 +#endif /* _VX_CACCT_DEF_H */
10326 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cacct.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct.h
10327 --- linux-2.6.32.1/include/linux/vserver/cacct.h 1970-01-01 01:00:00.000000000 +0100
10328 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct.h 2009-12-03 20:04:56.000000000 +0100
10330 +#ifndef _VX_CACCT_H
10331 +#define _VX_CACCT_H
10334 +enum sock_acc_field {
10335 + VXA_SOCK_UNSPEC = 0,
10341 + VXA_SOCK_SIZE /* array size */
10344 +#endif /* _VX_CACCT_H */
10345 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cacct_int.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_int.h
10346 --- linux-2.6.32.1/include/linux/vserver/cacct_int.h 1970-01-01 01:00:00.000000000 +0100
10347 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cacct_int.h 2009-12-03 20:04:56.000000000 +0100
10349 +#ifndef _VX_CACCT_INT_H
10350 +#define _VX_CACCT_INT_H
10356 +unsigned long vx_sock_count(struct _vx_cacct *cacct, int type, int pos)
10358 + return atomic_long_read(&cacct->sock[type][pos].count);
10363 +unsigned long vx_sock_total(struct _vx_cacct *cacct, int type, int pos)
10365 + return atomic_long_read(&cacct->sock[type][pos].total);
10368 +#endif /* __KERNEL__ */
10369 +#endif /* _VX_CACCT_INT_H */
10370 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/check.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/check.h
10371 --- linux-2.6.32.1/include/linux/vserver/check.h 1970-01-01 01:00:00.000000000 +0100
10372 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/check.h 2009-12-03 20:04:56.000000000 +0100
10374 +#ifndef _VS_CHECK_H
10375 +#define _VS_CHECK_H
10378 +#define MAX_S_CONTEXT 65535 /* Arbitrary limit */
10380 +#ifdef CONFIG_VSERVER_DYNAMIC_IDS
10381 +#define MIN_D_CONTEXT 49152 /* dynamic contexts start here */
10383 +#define MIN_D_CONTEXT 65536
10386 +/* check conditions */
10388 +#define VS_ADMIN 0x0001
10389 +#define VS_WATCH 0x0002
10390 +#define VS_HIDE 0x0004
10391 +#define VS_HOSTID 0x0008
10393 +#define VS_IDENT 0x0010
10394 +#define VS_EQUIV 0x0020
10395 +#define VS_PARENT 0x0040
10396 +#define VS_CHILD 0x0080
10398 +#define VS_ARG_MASK 0x00F0
10400 +#define VS_DYNAMIC 0x0100
10401 +#define VS_STATIC 0x0200
10403 +#define VS_ATR_MASK 0x0F00
10405 +#ifdef CONFIG_VSERVER_PRIVACY
10406 +#define VS_ADMIN_P (0)
10407 +#define VS_WATCH_P (0)
10409 +#define VS_ADMIN_P VS_ADMIN
10410 +#define VS_WATCH_P VS_WATCH
10413 +#define VS_HARDIRQ 0x1000
10414 +#define VS_SOFTIRQ 0x2000
10415 +#define VS_IRQ 0x4000
10417 +#define VS_IRQ_MASK 0xF000
10419 +#include <linux/hardirq.h>
10422 + * check current context for ADMIN/WATCH and
10423 + * optionally against supplied argument
10425 +static inline int __vs_check(int cid, int id, unsigned int mode)
10427 + if (mode & VS_ARG_MASK) {
10428 + if ((mode & VS_IDENT) && (id == cid))
10431 + if (mode & VS_ATR_MASK) {
10432 + if ((mode & VS_DYNAMIC) &&
10433 + (id >= MIN_D_CONTEXT) &&
10434 + (id <= MAX_S_CONTEXT))
10436 + if ((mode & VS_STATIC) &&
10437 + (id > 1) && (id < MIN_D_CONTEXT))
10440 + if (mode & VS_IRQ_MASK) {
10441 + if ((mode & VS_IRQ) && unlikely(in_interrupt()))
10443 + if ((mode & VS_HARDIRQ) && unlikely(in_irq()))
10445 + if ((mode & VS_SOFTIRQ) && unlikely(in_softirq()))
10448 + return (((mode & VS_ADMIN) && (cid == 0)) ||
10449 + ((mode & VS_WATCH) && (cid == 1)) ||
10450 + ((mode & VS_HOSTID) && (id == 0)));
10453 +#define vx_check(c, m) __vs_check(vx_current_xid(), c, (m) | VS_IRQ)
10455 +#define vx_weak_check(c, m) ((m) ? vx_check(c, m) : 1)
10458 +#define nx_check(c, m) __vs_check(nx_current_nid(), c, m)
10460 +#define nx_weak_check(c, m) ((m) ? nx_check(c, m) : 1)
10463 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/context_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/context_cmd.h
10464 --- linux-2.6.32.1/include/linux/vserver/context_cmd.h 1970-01-01 01:00:00.000000000 +0100
10465 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/context_cmd.h 2009-12-03 20:04:56.000000000 +0100
10467 +#ifndef _VX_CONTEXT_CMD_H
10468 +#define _VX_CONTEXT_CMD_H
10471 +/* vinfo commands */
10473 +#define VCMD_task_xid VC_CMD(VINFO, 1, 0)
10476 +extern int vc_task_xid(uint32_t);
10478 +#endif /* __KERNEL__ */
10480 +#define VCMD_vx_info VC_CMD(VINFO, 5, 0)
10482 +struct vcmd_vx_info_v0 {
10484 + uint32_t initpid;
10485 + /* more to come */
10489 +extern int vc_vx_info(struct vx_info *, void __user *);
10491 +#endif /* __KERNEL__ */
10493 +#define VCMD_ctx_stat VC_CMD(VSTAT, 0, 0)
10495 +struct vcmd_ctx_stat_v0 {
10498 + /* more to come */
10502 +extern int vc_ctx_stat(struct vx_info *, void __user *);
10504 +#endif /* __KERNEL__ */
10506 +/* context commands */
10508 +#define VCMD_ctx_create_v0 VC_CMD(VPROC, 1, 0)
10509 +#define VCMD_ctx_create VC_CMD(VPROC, 1, 1)
10511 +struct vcmd_ctx_create {
10512 + uint64_t flagword;
10515 +#define VCMD_ctx_migrate_v0 VC_CMD(PROCMIG, 1, 0)
10516 +#define VCMD_ctx_migrate VC_CMD(PROCMIG, 1, 1)
10518 +struct vcmd_ctx_migrate {
10519 + uint64_t flagword;
10523 +extern int vc_ctx_create(uint32_t, void __user *);
10524 +extern int vc_ctx_migrate(struct vx_info *, void __user *);
10526 +#endif /* __KERNEL__ */
10529 +/* flag commands */
10531 +#define VCMD_get_cflags VC_CMD(FLAGS, 1, 0)
10532 +#define VCMD_set_cflags VC_CMD(FLAGS, 2, 0)
10534 +struct vcmd_ctx_flags_v0 {
10535 + uint64_t flagword;
10540 +extern int vc_get_cflags(struct vx_info *, void __user *);
10541 +extern int vc_set_cflags(struct vx_info *, void __user *);
10543 +#endif /* __KERNEL__ */
10546 +/* context caps commands */
10548 +#define VCMD_get_ccaps VC_CMD(FLAGS, 3, 1)
10549 +#define VCMD_set_ccaps VC_CMD(FLAGS, 4, 1)
10551 +struct vcmd_ctx_caps_v1 {
10557 +extern int vc_get_ccaps(struct vx_info *, void __user *);
10558 +extern int vc_set_ccaps(struct vx_info *, void __user *);
10560 +#endif /* __KERNEL__ */
10563 +/* bcaps commands */
10565 +#define VCMD_get_bcaps VC_CMD(FLAGS, 9, 0)
10566 +#define VCMD_set_bcaps VC_CMD(FLAGS, 10, 0)
10568 +struct vcmd_bcaps {
10574 +extern int vc_get_bcaps(struct vx_info *, void __user *);
10575 +extern int vc_set_bcaps(struct vx_info *, void __user *);
10577 +#endif /* __KERNEL__ */
10580 +/* umask commands */
10582 +#define VCMD_get_umask VC_CMD(FLAGS, 13, 0)
10583 +#define VCMD_set_umask VC_CMD(FLAGS, 14, 0)
10585 +struct vcmd_umask {
10591 +extern int vc_get_umask(struct vx_info *, void __user *);
10592 +extern int vc_set_umask(struct vx_info *, void __user *);
10594 +#endif /* __KERNEL__ */
10599 +#define VCMD_get_badness VC_CMD(MEMCTRL, 5, 0)
10600 +#define VCMD_set_badness VC_CMD(MEMCTRL, 6, 0)
10602 +struct vcmd_badness_v0 {
10607 +extern int vc_get_badness(struct vx_info *, void __user *);
10608 +extern int vc_set_badness(struct vx_info *, void __user *);
10610 +#endif /* __KERNEL__ */
10611 +#endif /* _VX_CONTEXT_CMD_H */
10612 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/context.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/context.h
10613 --- linux-2.6.32.1/include/linux/vserver/context.h 1970-01-01 01:00:00.000000000 +0100
10614 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/context.h 2009-12-03 20:04:56.000000000 +0100
10616 +#ifndef _VX_CONTEXT_H
10617 +#define _VX_CONTEXT_H
10619 +#include <linux/types.h>
10620 +#include <linux/capability.h>
10623 +/* context flags */
10625 +#define VXF_INFO_SCHED 0x00000002
10626 +#define VXF_INFO_NPROC 0x00000004
10627 +#define VXF_INFO_PRIVATE 0x00000008
10629 +#define VXF_INFO_INIT 0x00000010
10630 +#define VXF_INFO_HIDE 0x00000020
10631 +#define VXF_INFO_ULIMIT 0x00000040
10632 +#define VXF_INFO_NSPACE 0x00000080
10634 +#define VXF_SCHED_HARD 0x00000100
10635 +#define VXF_SCHED_PRIO 0x00000200
10636 +#define VXF_SCHED_PAUSE 0x00000400
10638 +#define VXF_VIRT_MEM 0x00010000
10639 +#define VXF_VIRT_UPTIME 0x00020000
10640 +#define VXF_VIRT_CPU 0x00040000
10641 +#define VXF_VIRT_LOAD 0x00080000
10642 +#define VXF_VIRT_TIME 0x00100000
10644 +#define VXF_HIDE_MOUNT 0x01000000
10645 +/* was VXF_HIDE_NETIF 0x02000000 */
10646 +#define VXF_HIDE_VINFO 0x04000000
10648 +#define VXF_STATE_SETUP (1ULL << 32)
10649 +#define VXF_STATE_INIT (1ULL << 33)
10650 +#define VXF_STATE_ADMIN (1ULL << 34)
10652 +#define VXF_SC_HELPER (1ULL << 36)
10653 +#define VXF_REBOOT_KILL (1ULL << 37)
10654 +#define VXF_PERSISTENT (1ULL << 38)
10656 +#define VXF_FORK_RSS (1ULL << 48)
10657 +#define VXF_PROLIFIC (1ULL << 49)
10659 +#define VXF_IGNEG_NICE (1ULL << 52)
10661 +#define VXF_ONE_TIME (0x0007ULL << 32)
10663 +#define VXF_INIT_SET (VXF_STATE_SETUP | VXF_STATE_INIT | VXF_STATE_ADMIN)
10666 +/* context migration */
10668 +#define VXM_SET_INIT 0x00000001
10669 +#define VXM_SET_REAPER 0x00000002
10671 +/* context caps */
10673 +#define VXC_CAP_MASK 0x00000000
10675 +#define VXC_SET_UTSNAME 0x00000001
10676 +#define VXC_SET_RLIMIT 0x00000002
10677 +#define VXC_FS_SECURITY 0x00000004
10678 +#define VXC_TIOCSTI 0x00000010
10680 +/* was VXC_RAW_ICMP 0x00000100 */
10681 +#define VXC_SYSLOG 0x00001000
10682 +#define VXC_OOM_ADJUST 0x00002000
10683 +#define VXC_AUDIT_CONTROL 0x00004000
10685 +#define VXC_SECURE_MOUNT 0x00010000
10686 +#define VXC_SECURE_REMOUNT 0x00020000
10687 +#define VXC_BINARY_MOUNT 0x00040000
10689 +#define VXC_QUOTA_CTL 0x00100000
10690 +#define VXC_ADMIN_MAPPER 0x00200000
10691 +#define VXC_ADMIN_CLOOP 0x00400000
10693 +#define VXC_KTHREAD 0x01000000
10694 +#define VXC_NAMESPACE 0x02000000
10699 +#include <linux/list.h>
10700 +#include <linux/spinlock.h>
10701 +#include <linux/rcupdate.h>
10703 +#include "limit_def.h"
10704 +#include "sched_def.h"
10705 +#include "cvirt_def.h"
10706 +#include "cacct_def.h"
10707 +#include "device_def.h"
10709 +#define VX_SPACES 2
10711 +struct _vx_info_pc {
10712 + struct _vx_sched_pc sched_pc;
10713 + struct _vx_cvirt_pc cvirt_pc;
10717 + struct hlist_node vx_hlist; /* linked list of contexts */
10718 + xid_t vx_id; /* context id */
10719 + atomic_t vx_usecnt; /* usage count */
10720 + atomic_t vx_tasks; /* tasks count */
10721 + struct vx_info *vx_parent; /* parent context */
10722 + int vx_state; /* context state */
10724 + unsigned long vx_nsmask[VX_SPACES]; /* assignment mask */
10725 + struct nsproxy *vx_nsproxy[VX_SPACES]; /* private namespaces */
10726 + struct fs_struct *vx_fs[VX_SPACES]; /* private namespace fs */
10728 + uint64_t vx_flags; /* context flags */
10729 + uint64_t vx_ccaps; /* context caps (vserver) */
10730 + kernel_cap_t vx_bcaps; /* bounding caps (system) */
10731 + unsigned long vx_umask; /* unshare mask (guest) */
10733 + struct task_struct *vx_reaper; /* guest reaper process */
10734 + pid_t vx_initpid; /* PID of guest init */
10735 + int64_t vx_badness_bias; /* OOM points bias */
10737 + struct _vx_limit limit; /* vserver limits */
10738 + struct _vx_sched sched; /* vserver scheduler */
10739 + struct _vx_cvirt cvirt; /* virtual/bias stuff */
10740 + struct _vx_cacct cacct; /* context accounting */
10742 + struct _vx_device dmap; /* default device map targets */
10744 +#ifndef CONFIG_SMP
10745 + struct _vx_info_pc info_pc; /* per cpu data */
10747 + struct _vx_info_pc *ptr_pc; /* per cpu array */
10750 + wait_queue_head_t vx_wait; /* context exit waitqueue */
10751 + int reboot_cmd; /* last sys_reboot() cmd */
10752 + int exit_code; /* last process exit code */
10754 + char vx_name[65]; /* vserver name */
10757 +#ifndef CONFIG_SMP
10758 +#define vx_ptr_pc(vxi) (&(vxi)->info_pc)
10759 +#define vx_per_cpu(vxi, v, id) vx_ptr_pc(vxi)->v
10761 +#define vx_ptr_pc(vxi) ((vxi)->ptr_pc)
10762 +#define vx_per_cpu(vxi, v, id) per_cpu_ptr(vx_ptr_pc(vxi), id)->v
10765 +#define vx_cpu(vxi, v) vx_per_cpu(vxi, v, smp_processor_id())
10768 +struct vx_info_save {
10769 + struct vx_info *vxi;
10774 +/* status flags */
10776 +#define VXS_HASHED 0x0001
10777 +#define VXS_PAUSED 0x0010
10778 +#define VXS_SHUTDOWN 0x0100
10779 +#define VXS_HELPER 0x1000
10780 +#define VXS_RELEASED 0x8000
10783 +extern void claim_vx_info(struct vx_info *, struct task_struct *);
10784 +extern void release_vx_info(struct vx_info *, struct task_struct *);
10786 +extern struct vx_info *lookup_vx_info(int);
10787 +extern struct vx_info *lookup_or_create_vx_info(int);
10789 +extern int get_xid_list(int, unsigned int *, int);
10790 +extern int xid_is_hashed(xid_t);
10792 +extern int vx_migrate_task(struct task_struct *, struct vx_info *, int);
10794 +extern long vs_state_change(struct vx_info *, unsigned int);
10797 +#endif /* __KERNEL__ */
10798 +#endif /* _VX_CONTEXT_H */
10799 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cvirt_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt_cmd.h
10800 --- linux-2.6.32.1/include/linux/vserver/cvirt_cmd.h 1970-01-01 01:00:00.000000000 +0100
10801 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt_cmd.h 2009-12-03 20:04:56.000000000 +0100
10803 +#ifndef _VX_CVIRT_CMD_H
10804 +#define _VX_CVIRT_CMD_H
10807 +/* virtual host info name commands */
10809 +#define VCMD_set_vhi_name VC_CMD(VHOST, 1, 0)
10810 +#define VCMD_get_vhi_name VC_CMD(VHOST, 2, 0)
10812 +struct vcmd_vhi_name_v0 {
10818 +enum vhi_name_field {
10819 + VHIN_CONTEXT = 0,
10831 +#include <linux/compiler.h>
10833 +extern int vc_set_vhi_name(struct vx_info *, void __user *);
10834 +extern int vc_get_vhi_name(struct vx_info *, void __user *);
10836 +#endif /* __KERNEL__ */
10838 +#define VCMD_virt_stat VC_CMD(VSTAT, 3, 0)
10840 +struct vcmd_virt_stat_v0 {
10843 + uint32_t nr_threads;
10844 + uint32_t nr_running;
10845 + uint32_t nr_uninterruptible;
10846 + uint32_t nr_onhold;
10847 + uint32_t nr_forks;
10848 + uint32_t load[3];
10852 +extern int vc_virt_stat(struct vx_info *, void __user *);
10854 +#endif /* __KERNEL__ */
10855 +#endif /* _VX_CVIRT_CMD_H */
10856 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cvirt_def.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt_def.h
10857 --- linux-2.6.32.1/include/linux/vserver/cvirt_def.h 1970-01-01 01:00:00.000000000 +0100
10858 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt_def.h 2009-12-03 20:04:56.000000000 +0100
10860 +#ifndef _VX_CVIRT_DEF_H
10861 +#define _VX_CVIRT_DEF_H
10863 +#include <linux/jiffies.h>
10864 +#include <linux/spinlock.h>
10865 +#include <linux/wait.h>
10866 +#include <linux/time.h>
10867 +#include <asm/atomic.h>
10870 +struct _vx_usage_stat {
10874 + uint64_t softirq;
10880 +struct _vx_syslog {
10881 + wait_queue_head_t log_wait;
10882 + spinlock_t logbuf_lock; /* lock for the log buffer */
10884 + unsigned long log_start; /* next char to be read by syslog() */
10885 + unsigned long con_start; /* next char to be sent to consoles */
10886 + unsigned long log_end; /* most-recently-written-char + 1 */
10887 + unsigned long logged_chars; /* #chars since last read+clear operation */
10889 + char log_buf[1024];
10893 +/* context sub struct */
10895 +struct _vx_cvirt {
10896 + atomic_t nr_threads; /* number of current threads */
10897 + atomic_t nr_running; /* number of running threads */
10898 + atomic_t nr_uninterruptible; /* number of uninterruptible threads */
10900 + atomic_t nr_onhold; /* processes on hold */
10901 + uint32_t onhold_last; /* jiffies when put on hold */
10903 + struct timeval bias_tv; /* time offset to the host */
10904 + struct timespec bias_idle;
10905 + struct timespec bias_uptime; /* context creation point */
10906 + uint64_t bias_clock; /* offset in clock_t */
10908 + spinlock_t load_lock; /* lock for the load averages */
10909 + atomic_t load_updates; /* nr of load updates done so far */
10910 + uint32_t load_last; /* last time load was calculated */
10911 + uint32_t load[3]; /* load averages 1,5,15 */
10913 + atomic_t total_forks; /* number of forks so far */
10915 + struct _vx_syslog syslog;
10918 +struct _vx_cvirt_pc {
10919 + struct _vx_usage_stat cpustat;
10923 +#ifdef CONFIG_VSERVER_DEBUG
10925 +static inline void __dump_vx_cvirt(struct _vx_cvirt *cvirt)
10927 + printk("\t_vx_cvirt:\n");
10928 + printk("\t threads: %4d, %4d, %4d, %4d\n",
10929 + atomic_read(&cvirt->nr_threads),
10930 + atomic_read(&cvirt->nr_running),
10931 + atomic_read(&cvirt->nr_uninterruptible),
10932 + atomic_read(&cvirt->nr_onhold));
10933 + /* add rest here */
10934 + printk("\t total_forks = %d\n", atomic_read(&cvirt->total_forks));
10939 +#endif /* _VX_CVIRT_DEF_H */
10940 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/cvirt.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt.h
10941 --- linux-2.6.32.1/include/linux/vserver/cvirt.h 1970-01-01 01:00:00.000000000 +0100
10942 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/cvirt.h 2009-12-03 20:04:56.000000000 +0100
10944 +#ifndef _VX_CVIRT_H
10945 +#define _VX_CVIRT_H
10952 +void vx_vsi_uptime(struct timespec *, struct timespec *);
10957 +void vx_update_load(struct vx_info *);
10960 +int vx_do_syslog(int, char __user *, int);
10962 +#endif /* __KERNEL__ */
10963 +#endif /* _VX_CVIRT_H */
10964 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/debug_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/debug_cmd.h
10965 --- linux-2.6.32.1/include/linux/vserver/debug_cmd.h 1970-01-01 01:00:00.000000000 +0100
10966 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/debug_cmd.h 2009-12-03 20:04:56.000000000 +0100
10968 +#ifndef _VX_DEBUG_CMD_H
10969 +#define _VX_DEBUG_CMD_H
10972 +/* debug commands */
10974 +#define VCMD_dump_history VC_CMD(DEBUG, 1, 0)
10976 +#define VCMD_read_history VC_CMD(DEBUG, 5, 0)
10977 +#define VCMD_read_monitor VC_CMD(DEBUG, 6, 0)
10979 +struct vcmd_read_history_v0 {
10982 + char __user *data;
10985 +struct vcmd_read_monitor_v0 {
10988 + char __user *data;
10994 +#ifdef CONFIG_COMPAT
10996 +#include <asm/compat.h>
10998 +struct vcmd_read_history_v0_x32 {
11001 + compat_uptr_t data_ptr;
11004 +struct vcmd_read_monitor_v0_x32 {
11007 + compat_uptr_t data_ptr;
11010 +#endif /* CONFIG_COMPAT */
11012 +extern int vc_dump_history(uint32_t);
11014 +extern int vc_read_history(uint32_t, void __user *);
11015 +extern int vc_read_monitor(uint32_t, void __user *);
11017 +#ifdef CONFIG_COMPAT
11019 +extern int vc_read_history_x32(uint32_t, void __user *);
11020 +extern int vc_read_monitor_x32(uint32_t, void __user *);
11022 +#endif /* CONFIG_COMPAT */
11024 +#endif /* __KERNEL__ */
11025 +#endif /* _VX_DEBUG_CMD_H */
11026 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/debug.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/debug.h
11027 --- linux-2.6.32.1/include/linux/vserver/debug.h 1970-01-01 01:00:00.000000000 +0100
11028 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/debug.h 2009-12-03 20:04:56.000000000 +0100
11030 +#ifndef _VX_DEBUG_H
11031 +#define _VX_DEBUG_H
11034 +#define VXD_CBIT(n, m) (vx_debug_ ## n & (1 << (m)))
11035 +#define VXD_CMIN(n, m) (vx_debug_ ## n > (m))
11036 +#define VXD_MASK(n, m) (vx_debug_ ## n & (m))
11038 +#define VXD_DEV(d) (d), (d)->bd_inode->i_ino, \
11039 + imajor((d)->bd_inode), iminor((d)->bd_inode)
11040 +#define VXF_DEV "%p[%lu,%d:%d]"
11043 +#define vxd_path(p) \
11044 + ({ static char _buffer[PATH_MAX]; \
11045 + d_path(p, _buffer, sizeof(_buffer)); })
11047 +#define vxd_cond_path(n) \
11048 + ((n) ? vxd_path(&(n)->path) : "<null>" )
11051 +#ifdef CONFIG_VSERVER_DEBUG
11053 +extern unsigned int vx_debug_switch;
11054 +extern unsigned int vx_debug_xid;
11055 +extern unsigned int vx_debug_nid;
11056 +extern unsigned int vx_debug_tag;
11057 +extern unsigned int vx_debug_net;
11058 +extern unsigned int vx_debug_limit;
11059 +extern unsigned int vx_debug_cres;
11060 +extern unsigned int vx_debug_dlim;
11061 +extern unsigned int vx_debug_quota;
11062 +extern unsigned int vx_debug_cvirt;
11063 +extern unsigned int vx_debug_space;
11064 +extern unsigned int vx_debug_misc;
11067 +#define VX_LOGLEVEL "vxD: "
11068 +#define VX_PROC_FMT "%p: "
11069 +#define VX_PROCESS current
11071 +#define vxdprintk(c, f, x...) \
11074 + printk(VX_LOGLEVEL VX_PROC_FMT f "\n", \
11075 + VX_PROCESS , ##x); \
11078 +#define vxlprintk(c, f, x...) \
11081 + printk(VX_LOGLEVEL f " @%s:%d\n", x); \
11084 +#define vxfprintk(c, f, x...) \
11087 + printk(VX_LOGLEVEL f " %s@%s:%d\n", x); \
11093 +void dump_vx_info(struct vx_info *, int);
11094 +void dump_vx_info_inactive(int);
11096 +#else /* CONFIG_VSERVER_DEBUG */
11098 +#define vx_debug_switch 0
11099 +#define vx_debug_xid 0
11100 +#define vx_debug_nid 0
11101 +#define vx_debug_tag 0
11102 +#define vx_debug_net 0
11103 +#define vx_debug_limit 0
11104 +#define vx_debug_cres 0
11105 +#define vx_debug_dlim 0
11106 +#define vx_debug_cvirt 0
11108 +#define vxdprintk(x...) do { } while (0)
11109 +#define vxlprintk(x...) do { } while (0)
11110 +#define vxfprintk(x...) do { } while (0)
11112 +#endif /* CONFIG_VSERVER_DEBUG */
11115 +#ifdef CONFIG_VSERVER_WARN
11117 +#define VX_WARNLEVEL KERN_WARNING "vxW: "
11118 +#define VX_WARN_TASK "[»%s«,%u:#%u|%u|%u] "
11119 +#define VX_WARN_XID "[xid #%u] "
11120 +#define VX_WARN_NID "[nid #%u] "
11121 +#define VX_WARN_TAG "[tag #%u] "
11123 +#define vxwprintk(c, f, x...) \
11126 + printk(VX_WARNLEVEL f "\n", ##x); \
11129 +#else /* CONFIG_VSERVER_WARN */
11131 +#define vxwprintk(x...) do { } while (0)
11133 +#endif /* CONFIG_VSERVER_WARN */
11135 +#define vxwprintk_task(c, f, x...) \
11136 + vxwprintk(c, VX_WARN_TASK f, \
11137 + current->comm, current->pid, \
11138 + current->xid, current->nid, current->tag, ##x)
11139 +#define vxwprintk_xid(c, f, x...) \
11140 + vxwprintk(c, VX_WARN_XID f, current->xid, x)
11141 +#define vxwprintk_nid(c, f, x...) \
11142 + vxwprintk(c, VX_WARN_NID f, current->nid, x)
11143 +#define vxwprintk_tag(c, f, x...) \
11144 + vxwprintk(c, VX_WARN_TAG f, current->tag, x)
11146 +#ifdef CONFIG_VSERVER_DEBUG
11147 +#define vxd_assert_lock(l) assert_spin_locked(l)
11148 +#define vxd_assert(c, f, x...) vxlprintk(!(c), \
11149 + "assertion [" f "] failed.", ##x, __FILE__, __LINE__)
11151 +#define vxd_assert_lock(l) do { } while (0)
11152 +#define vxd_assert(c, f, x...) do { } while (0)
11156 +#endif /* _VX_DEBUG_H */
11157 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/device_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device_cmd.h
11158 --- linux-2.6.32.1/include/linux/vserver/device_cmd.h 1970-01-01 01:00:00.000000000 +0100
11159 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device_cmd.h 2009-12-03 20:04:56.000000000 +0100
11161 +#ifndef _VX_DEVICE_CMD_H
11162 +#define _VX_DEVICE_CMD_H
11165 +/* device vserver commands */
11167 +#define VCMD_set_mapping VC_CMD(DEVICE, 1, 0)
11168 +#define VCMD_unset_mapping VC_CMD(DEVICE, 2, 0)
11170 +struct vcmd_set_mapping_v0 {
11171 + const char __user *device;
11172 + const char __user *target;
11179 +#ifdef CONFIG_COMPAT
11181 +#include <asm/compat.h>
11183 +struct vcmd_set_mapping_v0_x32 {
11184 + compat_uptr_t device_ptr;
11185 + compat_uptr_t target_ptr;
11189 +#endif /* CONFIG_COMPAT */
11191 +#include <linux/compiler.h>
11193 +extern int vc_set_mapping(struct vx_info *, void __user *);
11194 +extern int vc_unset_mapping(struct vx_info *, void __user *);
11196 +#ifdef CONFIG_COMPAT
11198 +extern int vc_set_mapping_x32(struct vx_info *, void __user *);
11199 +extern int vc_unset_mapping_x32(struct vx_info *, void __user *);
11201 +#endif /* CONFIG_COMPAT */
11203 +#endif /* __KERNEL__ */
11204 +#endif /* _VX_DEVICE_CMD_H */
11205 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/device_def.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device_def.h
11206 --- linux-2.6.32.1/include/linux/vserver/device_def.h 1970-01-01 01:00:00.000000000 +0100
11207 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device_def.h 2009-12-03 20:04:56.000000000 +0100
11209 +#ifndef _VX_DEVICE_DEF_H
11210 +#define _VX_DEVICE_DEF_H
11212 +#include <linux/types.h>
11214 +struct vx_dmap_target {
11219 +struct _vx_device {
11220 +#ifdef CONFIG_VSERVER_DEVICE
11221 + struct vx_dmap_target targets[2];
11225 +#endif /* _VX_DEVICE_DEF_H */
11226 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/device.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device.h
11227 --- linux-2.6.32.1/include/linux/vserver/device.h 1970-01-01 01:00:00.000000000 +0100
11228 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/device.h 2009-12-03 20:04:56.000000000 +0100
11230 +#ifndef _VX_DEVICE_H
11231 +#define _VX_DEVICE_H
11234 +#define DATTR_CREATE 0x00000001
11235 +#define DATTR_OPEN 0x00000002
11237 +#define DATTR_REMAP 0x00000010
11239 +#define DATTR_MASK 0x00000013
11242 +#else /* _VX_DEVICE_H */
11243 +#warning duplicate inclusion
11244 +#endif /* _VX_DEVICE_H */
11245 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/dlimit_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/dlimit_cmd.h
11246 --- linux-2.6.32.1/include/linux/vserver/dlimit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11247 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/dlimit_cmd.h 2009-12-10 16:43:43.000000000 +0100
11249 +#ifndef _VX_DLIMIT_CMD_H
11250 +#define _VX_DLIMIT_CMD_H
11253 +/* dlimit vserver commands */
11255 +#define VCMD_add_dlimit VC_CMD(DLIMIT, 1, 0)
11256 +#define VCMD_rem_dlimit VC_CMD(DLIMIT, 2, 0)
11258 +#define VCMD_set_dlimit VC_CMD(DLIMIT, 5, 0)
11259 +#define VCMD_get_dlimit VC_CMD(DLIMIT, 6, 0)
11261 +struct vcmd_ctx_dlimit_base_v0 {
11262 + const char __user *name;
11266 +struct vcmd_ctx_dlimit_v0 {
11267 + const char __user *name;
11268 + uint32_t space_used; /* used space in kbytes */
11269 + uint32_t space_total; /* maximum space in kbytes */
11270 + uint32_t inodes_used; /* used inodes */
11271 + uint32_t inodes_total; /* maximum inodes */
11272 + uint32_t reserved; /* reserved for root in % */
11276 +#define CDLIM_UNSET ((uint32_t)0UL)
11277 +#define CDLIM_INFINITY ((uint32_t)~0UL)
11278 +#define CDLIM_KEEP ((uint32_t)~1UL)
11280 +#define DLIME_UNIT 0
11281 +#define DLIME_KILO 1
11282 +#define DLIME_MEGA 2
11283 +#define DLIME_GIGA 3
11285 +#define DLIMF_SHIFT 0x10
11287 +#define DLIMS_USED 0
11288 +#define DLIMS_TOTAL 2
11291 +uint64_t dlimit_space_32to64(uint32_t val, uint32_t flags, int shift)
11293 + int exp = (flags & DLIMF_SHIFT) ?
11294 + (flags >> shift) & DLIME_GIGA : DLIME_KILO;
11295 + return ((uint64_t)val) << (10 * exp);
11299 +uint32_t dlimit_space_64to32(uint64_t val, uint32_t *flags, int shift)
11303 + if (*flags & DLIMF_SHIFT) {
11304 + while (val > (1LL << 32) && (exp < 3)) {
11308 + *flags &= ~(DLIME_GIGA << shift);
11309 + *flags |= exp << shift;
11317 +#ifdef CONFIG_COMPAT
11319 +#include <asm/compat.h>
11321 +struct vcmd_ctx_dlimit_base_v0_x32 {
11322 + compat_uptr_t name_ptr;
11326 +struct vcmd_ctx_dlimit_v0_x32 {
11327 + compat_uptr_t name_ptr;
11328 + uint32_t space_used; /* used space in kbytes */
11329 + uint32_t space_total; /* maximum space in kbytes */
11330 + uint32_t inodes_used; /* used inodes */
11331 + uint32_t inodes_total; /* maximum inodes */
11332 + uint32_t reserved; /* reserved for root in % */
11336 +#endif /* CONFIG_COMPAT */
11338 +#include <linux/compiler.h>
11340 +extern int vc_add_dlimit(uint32_t, void __user *);
11341 +extern int vc_rem_dlimit(uint32_t, void __user *);
11343 +extern int vc_set_dlimit(uint32_t, void __user *);
11344 +extern int vc_get_dlimit(uint32_t, void __user *);
11346 +#ifdef CONFIG_COMPAT
11348 +extern int vc_add_dlimit_x32(uint32_t, void __user *);
11349 +extern int vc_rem_dlimit_x32(uint32_t, void __user *);
11351 +extern int vc_set_dlimit_x32(uint32_t, void __user *);
11352 +extern int vc_get_dlimit_x32(uint32_t, void __user *);
11354 +#endif /* CONFIG_COMPAT */
11356 +#endif /* __KERNEL__ */
11357 +#endif /* _VX_DLIMIT_CMD_H */
11358 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/dlimit.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/dlimit.h
11359 --- linux-2.6.32.1/include/linux/vserver/dlimit.h 1970-01-01 01:00:00.000000000 +0100
11360 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/dlimit.h 2009-12-03 20:04:56.000000000 +0100
11362 +#ifndef _VX_DLIMIT_H
11363 +#define _VX_DLIMIT_H
11365 +#include "switch.h"
11370 +/* keep in sync with CDLIM_INFINITY */
11372 +#define DLIM_INFINITY (~0ULL)
11374 +#include <linux/spinlock.h>
11375 +#include <linux/rcupdate.h>
11377 +struct super_block;
11380 + struct hlist_node dl_hlist; /* linked list of contexts */
11381 + struct rcu_head dl_rcu; /* the rcu head */
11382 + tag_t dl_tag; /* context tag */
11383 + atomic_t dl_usecnt; /* usage count */
11384 + atomic_t dl_refcnt; /* reference count */
11386 + struct super_block *dl_sb; /* associated superblock */
11388 + spinlock_t dl_lock; /* protect the values */
11390 + unsigned long long dl_space_used; /* used space in bytes */
11391 + unsigned long long dl_space_total; /* maximum space in bytes */
11392 + unsigned long dl_inodes_used; /* used inodes */
11393 + unsigned long dl_inodes_total; /* maximum inodes */
11395 + unsigned int dl_nrlmult; /* non root limit mult */
11400 +extern void rcu_free_dl_info(struct rcu_head *);
11401 +extern void unhash_dl_info(struct dl_info *);
11403 +extern struct dl_info *locate_dl_info(struct super_block *, tag_t);
11408 +extern void vx_vsi_statfs(struct super_block *, struct kstatfs *);
11410 +typedef uint64_t dlsize_t;
11412 +#endif /* __KERNEL__ */
11413 +#else /* _VX_DLIMIT_H */
11414 +#warning duplicate inclusion
11415 +#endif /* _VX_DLIMIT_H */
11416 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/global.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/global.h
11417 --- linux-2.6.32.1/include/linux/vserver/global.h 1970-01-01 01:00:00.000000000 +0100
11418 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/global.h 2009-12-03 20:04:56.000000000 +0100
11420 +#ifndef _VX_GLOBAL_H
11421 +#define _VX_GLOBAL_H
11424 +extern atomic_t vx_global_ctotal;
11425 +extern atomic_t vx_global_cactive;
11427 +extern atomic_t nx_global_ctotal;
11428 +extern atomic_t nx_global_cactive;
11430 +extern atomic_t vs_global_nsproxy;
11431 +extern atomic_t vs_global_fs;
11432 +extern atomic_t vs_global_mnt_ns;
11433 +extern atomic_t vs_global_uts_ns;
11434 +extern atomic_t vs_global_user_ns;
11435 +extern atomic_t vs_global_pid_ns;
11438 +#endif /* _VX_GLOBAL_H */
11439 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/history.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/history.h
11440 --- linux-2.6.32.1/include/linux/vserver/history.h 1970-01-01 01:00:00.000000000 +0100
11441 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/history.h 2009-12-03 20:04:56.000000000 +0100
11443 +#ifndef _VX_HISTORY_H
11444 +#define _VX_HISTORY_H
11449 + VXH_THROW_OOPS = 1,
11453 + VXH_INIT_VX_INFO,
11456 + VXH_CLAIM_VX_INFO,
11457 + VXH_RELEASE_VX_INFO,
11458 + VXH_ALLOC_VX_INFO,
11459 + VXH_DEALLOC_VX_INFO,
11460 + VXH_HASH_VX_INFO,
11461 + VXH_UNHASH_VX_INFO,
11463 + VXH_LOOKUP_VX_INFO,
11464 + VXH_CREATE_VX_INFO,
11467 +struct _vxhe_vxi {
11468 + struct vx_info *ptr;
11474 +struct _vxhe_set_clr {
11478 +struct _vxhe_loc_lookup {
11482 +struct _vx_hist_entry {
11484 + unsigned short seq;
11485 + unsigned short type;
11486 + struct _vxhe_vxi vxi;
11488 + struct _vxhe_set_clr sc;
11489 + struct _vxhe_loc_lookup ll;
11493 +#ifdef CONFIG_VSERVER_HISTORY
11495 +extern unsigned volatile int vxh_active;
11497 +struct _vx_hist_entry *vxh_advance(void *loc);
11501 +void __vxh_copy_vxi(struct _vx_hist_entry *entry, struct vx_info *vxi)
11503 + entry->vxi.ptr = vxi;
11505 + entry->vxi.usecnt = atomic_read(&vxi->vx_usecnt);
11506 + entry->vxi.tasks = atomic_read(&vxi->vx_tasks);
11507 + entry->vxi.xid = vxi->vx_id;
11512 +#define __HERE__ current_text_addr()
11514 +#define __VXH_BODY(__type, __data, __here) \
11515 + struct _vx_hist_entry *entry; \
11517 + preempt_disable(); \
11518 + entry = vxh_advance(__here); \
11520 + entry->type = __type; \
11521 + preempt_enable();
11524 + /* pass vxi only */
11526 +#define __VXH_SMPL \
11527 + __vxh_copy_vxi(entry, vxi)
11530 +void __vxh_smpl(struct vx_info *vxi, int __type, void *__here)
11532 + __VXH_BODY(__type, __VXH_SMPL, __here)
11535 + /* pass vxi and data (void *) */
11537 +#define __VXH_DATA \
11538 + __vxh_copy_vxi(entry, vxi); \
11539 + entry->sc.data = data
11542 +void __vxh_data(struct vx_info *vxi, void *data,
11543 + int __type, void *__here)
11545 + __VXH_BODY(__type, __VXH_DATA, __here)
11548 + /* pass vxi and arg (long) */
11550 +#define __VXH_LONG \
11551 + __vxh_copy_vxi(entry, vxi); \
11552 + entry->ll.arg = arg
11555 +void __vxh_long(struct vx_info *vxi, long arg,
11556 + int __type, void *__here)
11558 + __VXH_BODY(__type, __VXH_LONG, __here)
11563 +void __vxh_throw_oops(void *__here)
11565 + __VXH_BODY(VXH_THROW_OOPS, {}, __here);
11566 + /* prevent further acquisition */
11571 +#define vxh_throw_oops() __vxh_throw_oops(__HERE__);
11573 +#define __vxh_get_vx_info(v, h) __vxh_smpl(v, VXH_GET_VX_INFO, h);
11574 +#define __vxh_put_vx_info(v, h) __vxh_smpl(v, VXH_PUT_VX_INFO, h);
11576 +#define __vxh_init_vx_info(v, d, h) \
11577 + __vxh_data(v, d, VXH_INIT_VX_INFO, h);
11578 +#define __vxh_set_vx_info(v, d, h) \
11579 + __vxh_data(v, d, VXH_SET_VX_INFO, h);
11580 +#define __vxh_clr_vx_info(v, d, h) \
11581 + __vxh_data(v, d, VXH_CLR_VX_INFO, h);
11583 +#define __vxh_claim_vx_info(v, d, h) \
11584 + __vxh_data(v, d, VXH_CLAIM_VX_INFO, h);
11585 +#define __vxh_release_vx_info(v, d, h) \
11586 + __vxh_data(v, d, VXH_RELEASE_VX_INFO, h);
11588 +#define vxh_alloc_vx_info(v) \
11589 + __vxh_smpl(v, VXH_ALLOC_VX_INFO, __HERE__);
11590 +#define vxh_dealloc_vx_info(v) \
11591 + __vxh_smpl(v, VXH_DEALLOC_VX_INFO, __HERE__);
11593 +#define vxh_hash_vx_info(v) \
11594 + __vxh_smpl(v, VXH_HASH_VX_INFO, __HERE__);
11595 +#define vxh_unhash_vx_info(v) \
11596 + __vxh_smpl(v, VXH_UNHASH_VX_INFO, __HERE__);
11598 +#define vxh_loc_vx_info(v, l) \
11599 + __vxh_long(v, l, VXH_LOC_VX_INFO, __HERE__);
11600 +#define vxh_lookup_vx_info(v, l) \
11601 + __vxh_long(v, l, VXH_LOOKUP_VX_INFO, __HERE__);
11602 +#define vxh_create_vx_info(v, l) \
11603 + __vxh_long(v, l, VXH_CREATE_VX_INFO, __HERE__);
11605 +extern void vxh_dump_history(void);
11608 +#else /* CONFIG_VSERVER_HISTORY */
11610 +#define __HERE__ 0
11612 +#define vxh_throw_oops() do { } while (0)
11614 +#define __vxh_get_vx_info(v, h) do { } while (0)
11615 +#define __vxh_put_vx_info(v, h) do { } while (0)
11617 +#define __vxh_init_vx_info(v, d, h) do { } while (0)
11618 +#define __vxh_set_vx_info(v, d, h) do { } while (0)
11619 +#define __vxh_clr_vx_info(v, d, h) do { } while (0)
11621 +#define __vxh_claim_vx_info(v, d, h) do { } while (0)
11622 +#define __vxh_release_vx_info(v, d, h) do { } while (0)
11624 +#define vxh_alloc_vx_info(v) do { } while (0)
11625 +#define vxh_dealloc_vx_info(v) do { } while (0)
11627 +#define vxh_hash_vx_info(v) do { } while (0)
11628 +#define vxh_unhash_vx_info(v) do { } while (0)
11630 +#define vxh_loc_vx_info(v, l) do { } while (0)
11631 +#define vxh_lookup_vx_info(v, l) do { } while (0)
11632 +#define vxh_create_vx_info(v, l) do { } while (0)
11634 +#define vxh_dump_history() do { } while (0)
11637 +#endif /* CONFIG_VSERVER_HISTORY */
11639 +#endif /* _VX_HISTORY_H */
11640 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/inode_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/inode_cmd.h
11641 --- linux-2.6.32.1/include/linux/vserver/inode_cmd.h 1970-01-01 01:00:00.000000000 +0100
11642 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/inode_cmd.h 2009-12-03 20:04:56.000000000 +0100
11644 +#ifndef _VX_INODE_CMD_H
11645 +#define _VX_INODE_CMD_H
11648 +/* inode vserver commands */
11650 +#define VCMD_get_iattr VC_CMD(INODE, 1, 1)
11651 +#define VCMD_set_iattr VC_CMD(INODE, 2, 1)
11653 +#define VCMD_fget_iattr VC_CMD(INODE, 3, 0)
11654 +#define VCMD_fset_iattr VC_CMD(INODE, 4, 0)
11656 +struct vcmd_ctx_iattr_v1 {
11657 + const char __user *name;
11663 +struct vcmd_ctx_fiattr_v0 {
11673 +#ifdef CONFIG_COMPAT
11675 +#include <asm/compat.h>
11677 +struct vcmd_ctx_iattr_v1_x32 {
11678 + compat_uptr_t name_ptr;
11684 +#endif /* CONFIG_COMPAT */
11686 +#include <linux/compiler.h>
11688 +extern int vc_get_iattr(void __user *);
11689 +extern int vc_set_iattr(void __user *);
11691 +extern int vc_fget_iattr(uint32_t, void __user *);
11692 +extern int vc_fset_iattr(uint32_t, void __user *);
11694 +#ifdef CONFIG_COMPAT
11696 +extern int vc_get_iattr_x32(void __user *);
11697 +extern int vc_set_iattr_x32(void __user *);
11699 +#endif /* CONFIG_COMPAT */
11701 +#endif /* __KERNEL__ */
11702 +#endif /* _VX_INODE_CMD_H */
11703 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/inode.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/inode.h
11704 --- linux-2.6.32.1/include/linux/vserver/inode.h 1970-01-01 01:00:00.000000000 +0100
11705 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/inode.h 2009-12-03 20:04:56.000000000 +0100
11707 +#ifndef _VX_INODE_H
11708 +#define _VX_INODE_H
11711 +#define IATTR_TAG 0x01000000
11713 +#define IATTR_ADMIN 0x00000001
11714 +#define IATTR_WATCH 0x00000002
11715 +#define IATTR_HIDE 0x00000004
11716 +#define IATTR_FLAGS 0x00000007
11718 +#define IATTR_BARRIER 0x00010000
11719 +#define IATTR_IXUNLINK 0x00020000
11720 +#define IATTR_IMMUTABLE 0x00040000
11721 +#define IATTR_COW 0x00080000
11726 +#ifdef CONFIG_VSERVER_PROC_SECURE
11727 +#define IATTR_PROC_DEFAULT ( IATTR_ADMIN | IATTR_HIDE )
11728 +#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11730 +#define IATTR_PROC_DEFAULT ( IATTR_ADMIN )
11731 +#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11734 +#define vx_hide_check(c, m) (((m) & IATTR_HIDE) ? vx_check(c, m) : 1)
11736 +#endif /* __KERNEL__ */
11738 +/* inode ioctls */
11740 +#define FIOC_GETXFLG _IOR('x', 5, long)
11741 +#define FIOC_SETXFLG _IOW('x', 6, long)
11743 +#else /* _VX_INODE_H */
11744 +#warning duplicate inclusion
11745 +#endif /* _VX_INODE_H */
11746 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/Kbuild linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/Kbuild
11747 --- linux-2.6.32.1/include/linux/vserver/Kbuild 1970-01-01 01:00:00.000000000 +0100
11748 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/Kbuild 2009-12-03 20:04:56.000000000 +0100
11751 +unifdef-y += context_cmd.h network_cmd.h space_cmd.h \
11752 + cacct_cmd.h cvirt_cmd.h limit_cmd.h dlimit_cmd.h \
11753 + inode_cmd.h tag_cmd.h sched_cmd.h signal_cmd.h \
11754 + debug_cmd.h device_cmd.h
11756 +unifdef-y += switch.h network.h monitor.h inode.h device.h
11758 diff -NurpP --minimal linux-2.6.32.1/include/linux/vserver/limit_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_cmd.h
11759 --- linux-2.6.32.1/include/linux/vserver/limit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11760 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_cmd.h 2009-12-03 20:04:56.000000000 +0100
11762 +#ifndef _VX_LIMIT_CMD_H
11763 +#define _VX_LIMIT_CMD_H
11766 +/* rlimit vserver commands */
11768 +#define VCMD_get_rlimit VC_CMD(RLIMIT, 1, 0)
11769 +#define VCMD_set_rlimit VC_CMD(RLIMIT, 2, 0)
11770 +#define VCMD_get_rlimit_mask VC_CMD(RLIMIT, 3, 0)
11771 +#define VCMD_reset_hits VC_CMD(RLIMIT, 7, 0)
11772 +#define VCMD_reset_minmax VC_CMD(RLIMIT, 9, 0)
11774 +struct vcmd_ctx_rlimit_v0 {
11776 + uint64_t minimum;
11777 + uint64_t softlimit;
11778 + uint64_t maximum;
11781 +struct vcmd_ctx_rlimit_mask_v0 {
11782 + uint32_t minimum;
11783 + uint32_t softlimit;
11784 + uint32_t maximum;
11787 +#define VCMD_rlimit_stat VC_CMD(VSTAT, 1, 0)
11789 +struct vcmd_rlimit_stat_v0 {
11793 + uint64_t minimum;
11794 + uint64_t maximum;
11797 +#define CRLIM_UNSET (0ULL)
11798 +#define CRLIM_INFINITY (~0ULL)
11799 +#define CRLIM_KEEP (~1ULL)
11803 +#ifdef CONFIG_IA32_EMULATION
11805 +struct vcmd_ctx_rlimit_v0_x32 {
11807 + uint64_t minimum;
11808 + uint64_t softlimit;
11809 + uint64_t maximum;
11810 +} __attribute__ ((packed));
11812 +#endif /* CONFIG_IA32_EMULATION */
11814 +#include <linux/compiler.h>
11816 +extern int vc_get_rlimit_mask(uint32_t, void __user *);
11817 +extern int vc_get_rlimit(struct vx_info *, void __user *);
11818 +extern int vc_set_rlimit(struct vx_info *, void __user *);
11819 +extern int vc_reset_hits(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-2.6.32.1/include/linux/vserver/limit_def.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_def.h
11834 --- linux-2.6.32.1/include/linux/vserver/limit_def.h 1970-01-01 01:00:00.000000000 +0100
11835 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_def.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/limit.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit.h
11885 --- linux-2.6.32.1/include/linux/vserver/limit.h 1970-01-01 01:00:00.000000000 +0100
11886 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/limit_int.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_int.h
11959 --- linux-2.6.32.1/include/linux/vserver/limit_int.h 1970-01-01 01:00:00.000000000 +0100
11960 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/limit_int.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/monitor.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/monitor.h
12161 --- linux-2.6.32.1/include/linux/vserver/monitor.h 1970-01-01 01:00:00.000000000 +0100
12162 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/monitor.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/network_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/network_cmd.h
12261 --- linux-2.6.32.1/include/linux/vserver/network_cmd.h 1970-01-01 01:00:00.000000000 +0100
12262 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/network_cmd.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/network.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/network.h
12415 --- linux-2.6.32.1/include/linux/vserver/network.h 1970-01-01 01:00:00.000000000 +0100
12416 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/network.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/percpu.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/percpu.h
12565 --- linux-2.6.32.1/include/linux/vserver/percpu.h 1970-01-01 01:00:00.000000000 +0100
12566 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/percpu.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/pid.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/pid.h
12583 --- linux-2.6.32.1/include/linux/vserver/pid.h 1970-01-01 01:00:00.000000000 +0100
12584 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/pid.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/sched_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched_cmd.h
12638 --- linux-2.6.32.1/include/linux/vserver/sched_cmd.h 1970-01-01 01:00:00.000000000 +0100
12639 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched_cmd.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/sched_def.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched_def.h
12750 --- linux-2.6.32.1/include/linux/vserver/sched_def.h 1970-01-01 01:00:00.000000000 +0100
12751 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched_def.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/sched.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched.h
12822 --- linux-2.6.32.1/include/linux/vserver/sched.h 1970-01-01 01:00:00.000000000 +0100
12823 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/sched.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/signal_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/signal_cmd.h
12852 --- linux-2.6.32.1/include/linux/vserver/signal_cmd.h 1970-01-01 01:00:00.000000000 +0100
12853 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/signal_cmd.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/signal.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/signal.h
12899 --- linux-2.6.32.1/include/linux/vserver/signal.h 1970-01-01 01:00:00.000000000 +0100
12900 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/signal.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/space_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/space_cmd.h
12917 --- linux-2.6.32.1/include/linux/vserver/space_cmd.h 1970-01-01 01:00:00.000000000 +0100
12918 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/space_cmd.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/space.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/space.h
12959 --- linux-2.6.32.1/include/linux/vserver/space.h 1970-01-01 01:00:00.000000000 +0100
12960 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/space.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/switch.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/switch.h
12975 --- linux-2.6.32.1/include/linux/vserver/switch.h 1970-01-01 01:00:00.000000000 +0100
12976 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/switch.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/tag_cmd.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/tag_cmd.h
13077 --- linux-2.6.32.1/include/linux/vserver/tag_cmd.h 1970-01-01 01:00:00.000000000 +0100
13078 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/tag_cmd.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vserver/tag.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/tag.h
13103 --- linux-2.6.32.1/include/linux/vserver/tag.h 1970-01-01 01:00:00.000000000 +0100
13104 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vserver/tag.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_inet6.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_inet6.h
13250 --- linux-2.6.32.1/include/linux/vs_inet6.h 1970-01-01 01:00:00.000000000 +0100
13251 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_inet6.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_inet.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_inet.h
13500 --- linux-2.6.32.1/include/linux/vs_inet.h 1970-01-01 01:00:00.000000000 +0100
13501 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_inet.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_limit.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_limit.h
13846 --- linux-2.6.32.1/include/linux/vs_limit.h 1970-01-01 01:00:00.000000000 +0100
13847 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_limit.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_memory.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_memory.h
13990 --- linux-2.6.32.1/include/linux/vs_memory.h 1970-01-01 01:00:00.000000000 +0100
13991 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_memory.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_network.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_network.h
14153 --- linux-2.6.32.1/include/linux/vs_network.h 1970-01-01 01:00:00.000000000 +0100
14154 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_network.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_pid.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_pid.h
14326 --- linux-2.6.32.1/include/linux/vs_pid.h 1970-01-01 01:00:00.000000000 +0100
14327 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_pid.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_sched.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_sched.h
14425 --- linux-2.6.32.1/include/linux/vs_sched.h 1970-01-01 01:00:00.000000000 +0100
14426 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_sched.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_socket.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_socket.h
14539 --- linux-2.6.32.1/include/linux/vs_socket.h 1970-01-01 01:00:00.000000000 +0100
14540 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_socket.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/linux/vs_tag.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_tag.h
14610 --- linux-2.6.32.1/include/linux/vs_tag.h 1970-01-01 01:00:00.000000000 +0100
14611 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_tag.h 2009-12-03 20:04:56.000000000 +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(const struct inode *inode, int mask);
14658 +#warning duplicate inclusion
14660 diff -NurpP --minimal linux-2.6.32.1/include/linux/vs_time.h linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_time.h
14661 --- linux-2.6.32.1/include/linux/vs_time.h 1970-01-01 01:00:00.000000000 +0100
14662 +++ linux-2.6.32.1-vs2.3.0.36.27/include/linux/vs_time.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/net/addrconf.h linux-2.6.32.1-vs2.3.0.36.27/include/net/addrconf.h
14684 --- linux-2.6.32.1/include/net/addrconf.h 2009-12-03 20:02:57.000000000 +0100
14685 +++ linux-2.6.32.1-vs2.3.0.36.27/include/net/addrconf.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/net/af_unix.h linux-2.6.32.1-vs2.3.0.36.27/include/net/af_unix.h
14697 --- linux-2.6.32.1/include/net/af_unix.h 2008-12-25 00:26:37.000000000 +0100
14698 +++ linux-2.6.32.1-vs2.3.0.36.27/include/net/af_unix.h 2009-12-03 20:04:56.000000000 +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-2.6.32.1/include/net/inet_timewait_sock.h linux-2.6.32.1-vs2.3.0.36.27/include/net/inet_timewait_sock.h
14708 --- linux-2.6.32.1/include/net/inet_timewait_sock.h 2009-12-03 20:02:57.000000000 +0100
14709 +++ linux-2.6.32.1-vs2.3.0.36.27/include/net/inet_timewait_sock.h 2009-12-03 20:04:56.000000000 +0100
14710 @@ -117,6 +117,10 @@ struct inet_timewait_sock {
14711 #define tw_hash __tw_common.skc_hash
14712 #define tw_prot __tw_common.skc_prot
14713 #define tw_net __tw_common.skc_net
14714 +#define tw_xid __tw_common.skc_xid
14715 +#define tw_vx_info __tw_common.skc_vx_info
14716 +#define tw_nid __tw_common.skc_nid
14717 +#define tw_nx_info __tw_common.skc_nx_info
14719 volatile unsigned char tw_substate;
14720 /* 3 bits hole, try to pack */
14721 diff -NurpP --minimal linux-2.6.32.1/include/net/route.h linux-2.6.32.1-vs2.3.0.36.27/include/net/route.h
14722 --- linux-2.6.32.1/include/net/route.h 2009-09-10 15:26:27.000000000 +0200
14723 +++ linux-2.6.32.1-vs2.3.0.36.27/include/net/route.h 2009-12-03 20:04:56.000000000 +0100
14724 @@ -135,6 +135,9 @@ static inline void ip_rt_put(struct rtab
14725 dst_release(&rt->u.dst);
14728 +#include <linux/vs_base.h>
14729 +#include <linux/vs_inet.h>
14731 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
14733 extern const __u8 ip_tos2prio[16];
14734 @@ -144,6 +147,9 @@ static inline char rt_tos2priority(u8 to
14735 return ip_tos2prio[IPTOS_TOS(tos)>>1];
14738 +extern int ip_v4_find_src(struct net *net, struct nx_info *,
14739 + struct rtable **, struct flowi *);
14741 static inline int ip_route_connect(struct rtable **rp, __be32 dst,
14742 __be32 src, u32 tos, int oif, u8 protocol,
14743 __be16 sport, __be16 dport, struct sock *sk,
14744 @@ -161,11 +167,24 @@ static inline int ip_route_connect(struc
14747 struct net *net = sock_net(sk);
14748 + struct nx_info *nx_info = current_nx_info();
14750 if (inet_sk(sk)->transparent)
14751 fl.flags |= FLOWI_FLAG_ANYSRC;
14753 - if (!dst || !src) {
14755 + nx_info = sk->sk_nx_info;
14757 + vxdprintk(VXD_CBIT(net, 4),
14758 + "ip_route_connect(%p) %p,%p;%lx",
14759 + sk, nx_info, sk->sk_socket,
14760 + (sk->sk_socket?sk->sk_socket->flags:0));
14762 + err = ip_v4_find_src(net, nx_info, rp, &fl);
14766 + if (!fl.fl4_dst || !fl.fl4_src) {
14767 err = __ip_route_output_key(net, rp, &fl);
14770 diff -NurpP --minimal linux-2.6.32.1/include/net/sock.h linux-2.6.32.1-vs2.3.0.36.27/include/net/sock.h
14771 --- linux-2.6.32.1/include/net/sock.h 2009-12-03 20:02:57.000000000 +0100
14772 +++ linux-2.6.32.1-vs2.3.0.36.27/include/net/sock.h 2009-12-03 20:04:56.000000000 +0100
14773 @@ -139,6 +139,10 @@ struct sock_common {
14774 #ifdef CONFIG_NET_NS
14775 struct net *skc_net;
14778 + struct vx_info *skc_vx_info;
14780 + struct nx_info *skc_nx_info;
14784 @@ -225,6 +229,10 @@ struct sock {
14785 #define sk_bind_node __sk_common.skc_bind_node
14786 #define sk_prot __sk_common.skc_prot
14787 #define sk_net __sk_common.skc_net
14788 +#define sk_xid __sk_common.skc_xid
14789 +#define sk_vx_info __sk_common.skc_vx_info
14790 +#define sk_nid __sk_common.skc_nid
14791 +#define sk_nx_info __sk_common.skc_nx_info
14792 kmemcheck_bitfield_begin(flags);
14793 unsigned int sk_shutdown : 2,
14795 diff -NurpP --minimal linux-2.6.32.1/init/Kconfig linux-2.6.32.1-vs2.3.0.36.27/init/Kconfig
14796 --- linux-2.6.32.1/init/Kconfig 2009-12-03 20:02:57.000000000 +0100
14797 +++ linux-2.6.32.1-vs2.3.0.36.27/init/Kconfig 2009-12-03 20:04:56.000000000 +0100
14798 @@ -477,6 +477,19 @@ config CGROUP_SCHED
14802 +config CFS_HARD_LIMITS
14803 + bool "Hard Limits for CFS Group Scheduler"
14804 + depends on EXPERIMENTAL
14805 + depends on FAIR_GROUP_SCHED && CGROUP_SCHED
14808 + This option enables hard limiting of CPU time obtained by
14809 + a fair task group. Use this if you want to throttle a group of tasks
14810 + based on its CPU usage. For more details refer to
14811 + Documentation/scheduler/sched-cfs-hard-limits.txt
14816 boolean "Control Group support"
14818 diff -NurpP --minimal linux-2.6.32.1/init/main.c linux-2.6.32.1-vs2.3.0.36.27/init/main.c
14819 --- linux-2.6.32.1/init/main.c 2009-12-03 20:02:57.000000000 +0100
14820 +++ linux-2.6.32.1-vs2.3.0.36.27/init/main.c 2009-12-03 20:04:56.000000000 +0100
14822 #include <linux/sfi.h>
14823 #include <linux/shmem_fs.h>
14824 #include <trace/boot.h>
14825 +#include <linux/vserver/percpu.h>
14827 #include <asm/io.h>
14828 #include <asm/bugs.h>
14829 diff -NurpP --minimal linux-2.6.32.1/ipc/mqueue.c linux-2.6.32.1-vs2.3.0.36.27/ipc/mqueue.c
14830 --- linux-2.6.32.1/ipc/mqueue.c 2009-12-03 20:02:57.000000000 +0100
14831 +++ linux-2.6.32.1-vs2.3.0.36.27/ipc/mqueue.c 2009-12-03 20:04:56.000000000 +0100
14833 #include <linux/pid.h>
14834 #include <linux/ipc_namespace.h>
14835 #include <linux/ima.h>
14836 +#include <linux/vs_context.h>
14837 +#include <linux/vs_limit.h>
14839 #include <net/sock.h>
14841 @@ -66,6 +68,7 @@ struct mqueue_inode_info {
14842 struct sigevent notify;
14843 struct pid* notify_owner;
14844 struct user_struct *user; /* user who created, for accounting */
14845 + struct vx_info *vxi;
14846 struct sock *notify_sock;
14847 struct sk_buff *notify_cookie;
14849 @@ -125,6 +128,7 @@ static struct inode *mqueue_get_inode(st
14850 if (S_ISREG(mode)) {
14851 struct mqueue_inode_info *info;
14852 struct task_struct *p = current;
14853 + struct vx_info *vxi = p->vx_info;
14854 unsigned long mq_bytes, mq_msg_tblsz;
14856 inode->i_fop = &mqueue_file_operations;
14857 @@ -139,6 +143,7 @@ static struct inode *mqueue_get_inode(st
14858 info->notify_owner = NULL;
14860 info->user = NULL; /* set when all is ok */
14861 + info->vxi = NULL;
14862 memset(&info->attr, 0, sizeof(info->attr));
14863 info->attr.mq_maxmsg = ipc_ns->mq_msg_max;
14864 info->attr.mq_msgsize = ipc_ns->mq_msgsize_max;
14865 @@ -153,22 +158,26 @@ static struct inode *mqueue_get_inode(st
14866 spin_lock(&mq_lock);
14867 if (u->mq_bytes + mq_bytes < u->mq_bytes ||
14868 u->mq_bytes + mq_bytes >
14869 - p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur) {
14870 + p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur ||
14871 + !vx_ipcmsg_avail(vxi, mq_bytes)) {
14872 spin_unlock(&mq_lock);
14875 u->mq_bytes += mq_bytes;
14876 + vx_ipcmsg_add(vxi, u, mq_bytes);
14877 spin_unlock(&mq_lock);
14879 info->messages = kmalloc(mq_msg_tblsz, GFP_KERNEL);
14880 if (!info->messages) {
14881 spin_lock(&mq_lock);
14882 u->mq_bytes -= mq_bytes;
14883 + vx_ipcmsg_sub(vxi, u, mq_bytes);
14884 spin_unlock(&mq_lock);
14888 info->user = get_uid(u);
14889 + info->vxi = get_vx_info(vxi);
14890 } else if (S_ISDIR(mode)) {
14892 /* Some things misbehave if size == 0 on a directory */
14893 @@ -269,8 +278,11 @@ static void mqueue_delete_inode(struct i
14894 (info->attr.mq_maxmsg * info->attr.mq_msgsize));
14897 + struct vx_info *vxi = info->vxi;
14899 spin_lock(&mq_lock);
14900 user->mq_bytes -= mq_bytes;
14901 + vx_ipcmsg_sub(vxi, user, mq_bytes);
14903 * get_ns_from_inode() ensures that the
14904 * (ipc_ns = sb->s_fs_info) is either a valid ipc_ns
14905 @@ -280,6 +292,7 @@ static void mqueue_delete_inode(struct i
14907 ipc_ns->mq_queues_count--;
14908 spin_unlock(&mq_lock);
14909 + put_vx_info(vxi);
14913 diff -NurpP --minimal linux-2.6.32.1/ipc/msg.c linux-2.6.32.1-vs2.3.0.36.27/ipc/msg.c
14914 --- linux-2.6.32.1/ipc/msg.c 2009-03-24 14:22:44.000000000 +0100
14915 +++ linux-2.6.32.1-vs2.3.0.36.27/ipc/msg.c 2009-12-03 20:04:56.000000000 +0100
14917 #include <linux/rwsem.h>
14918 #include <linux/nsproxy.h>
14919 #include <linux/ipc_namespace.h>
14920 +#include <linux/vs_base.h>
14922 #include <asm/current.h>
14923 #include <asm/uaccess.h>
14924 @@ -190,6 +191,7 @@ static int newque(struct ipc_namespace *
14926 msq->q_perm.mode = msgflg & S_IRWXUGO;
14927 msq->q_perm.key = key;
14928 + msq->q_perm.xid = vx_current_xid();
14930 msq->q_perm.security = NULL;
14931 retval = security_msg_queue_alloc(msq);
14932 diff -NurpP --minimal linux-2.6.32.1/ipc/namespace.c linux-2.6.32.1-vs2.3.0.36.27/ipc/namespace.c
14933 --- linux-2.6.32.1/ipc/namespace.c 2009-09-10 15:26:27.000000000 +0200
14934 +++ linux-2.6.32.1-vs2.3.0.36.27/ipc/namespace.c 2009-12-03 20:04:56.000000000 +0100
14936 #include <linux/slab.h>
14937 #include <linux/fs.h>
14938 #include <linux/mount.h>
14939 +#include <linux/vs_base.h>
14940 +#include <linux/vserver/global.h>
14944 diff -NurpP --minimal linux-2.6.32.1/ipc/sem.c linux-2.6.32.1-vs2.3.0.36.27/ipc/sem.c
14945 --- linux-2.6.32.1/ipc/sem.c 2009-09-10 15:26:27.000000000 +0200
14946 +++ linux-2.6.32.1-vs2.3.0.36.27/ipc/sem.c 2009-12-03 20:04:56.000000000 +0100
14948 #include <linux/rwsem.h>
14949 #include <linux/nsproxy.h>
14950 #include <linux/ipc_namespace.h>
14951 +#include <linux/vs_base.h>
14952 +#include <linux/vs_limit.h>
14954 #include <asm/uaccess.h>
14956 @@ -255,6 +257,7 @@ static int newary(struct ipc_namespace *
14958 sma->sem_perm.mode = (semflg & S_IRWXUGO);
14959 sma->sem_perm.key = key;
14960 + sma->sem_perm.xid = vx_current_xid();
14962 sma->sem_perm.security = NULL;
14963 retval = security_sem_alloc(sma);
14964 @@ -270,6 +273,9 @@ static int newary(struct ipc_namespace *
14967 ns->used_sems += nsems;
14968 + /* FIXME: obsoleted? */
14969 + vx_semary_inc(sma);
14970 + vx_nsems_add(sma, nsems);
14972 sma->sem_base = (struct sem *) &sma[1];
14973 INIT_LIST_HEAD(&sma->sem_pending);
14974 @@ -546,6 +552,9 @@ static void freeary(struct ipc_namespace
14977 ns->used_sems -= sma->sem_nsems;
14978 + /* FIXME: obsoleted? */
14979 + vx_nsems_sub(sma, sma->sem_nsems);
14980 + vx_semary_dec(sma);
14981 security_sem_free(sma);
14982 ipc_rcu_putref(sma);
14984 diff -NurpP --minimal linux-2.6.32.1/ipc/shm.c linux-2.6.32.1-vs2.3.0.36.27/ipc/shm.c
14985 --- linux-2.6.32.1/ipc/shm.c 2009-12-03 20:02:57.000000000 +0100
14986 +++ linux-2.6.32.1-vs2.3.0.36.27/ipc/shm.c 2009-12-03 20:04:56.000000000 +0100
14988 #include <linux/mount.h>
14989 #include <linux/ipc_namespace.h>
14990 #include <linux/ima.h>
14991 +#include <linux/vs_context.h>
14992 +#include <linux/vs_limit.h>
14994 #include <asm/uaccess.h>
14996 @@ -169,7 +171,12 @@ static void shm_open(struct vm_area_stru
14998 static void shm_destroy(struct ipc_namespace *ns, struct shmid_kernel *shp)
15000 - ns->shm_tot -= (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15001 + struct vx_info *vxi = lookup_vx_info(shp->shm_perm.xid);
15002 + int numpages = (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15004 + vx_ipcshm_sub(vxi, shp, numpages);
15005 + ns->shm_tot -= numpages;
15009 if (!is_file_hugepages(shp->shm_file))
15010 @@ -179,6 +186,7 @@ static void shm_destroy(struct ipc_names
15012 fput (shp->shm_file);
15013 security_shm_free(shp);
15014 + put_vx_info(vxi);
15015 ipc_rcu_putref(shp);
15018 @@ -349,11 +357,15 @@ static int newseg(struct ipc_namespace *
15019 if (ns->shm_tot + numpages > ns->shm_ctlall)
15022 + if (!vx_ipcshm_avail(current_vx_info(), numpages))
15025 shp = ipc_rcu_alloc(sizeof(*shp));
15029 shp->shm_perm.key = key;
15030 + shp->shm_perm.xid = vx_current_xid();
15031 shp->shm_perm.mode = (shmflg & S_IRWXUGO);
15032 shp->mlock_user = NULL;
15034 @@ -407,6 +419,7 @@ static int newseg(struct ipc_namespace *
15035 ns->shm_tot += numpages;
15036 error = shp->shm_perm.id;
15038 + vx_ipcshm_add(current_vx_info(), key, numpages);
15042 diff -NurpP --minimal linux-2.6.32.1/kernel/capability.c linux-2.6.32.1-vs2.3.0.36.27/kernel/capability.c
15043 --- linux-2.6.32.1/kernel/capability.c 2009-03-24 14:22:44.000000000 +0100
15044 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/capability.c 2009-12-03 20:04:56.000000000 +0100
15046 #include <linux/security.h>
15047 #include <linux/syscalls.h>
15048 #include <linux/pid_namespace.h>
15049 +#include <linux/vs_context.h>
15050 #include <asm/uaccess.h>
15051 #include "cred-internals.h"
15053 @@ -122,6 +123,7 @@ static int cap_validate_magic(cap_user_h
15059 * The only thing that can change the capabilities of the current
15060 * process is the current process. As such, we can't be in this code
15061 @@ -289,6 +291,8 @@ error:
15065 +#include <linux/vserver/base.h>
15068 * capable - Determine if the current task has a superior capability in effect
15069 * @cap: The capability to be tested for
15070 @@ -301,6 +305,9 @@ error:
15072 int capable(int cap)
15074 + /* here for now so we don't require task locking */
15075 + if (vs_check_bit(VXC_CAP_MASK, cap) && !vx_mcaps(1L << cap))
15077 if (unlikely(!cap_valid(cap))) {
15078 printk(KERN_CRIT "capable() called with invalid cap=%u\n", cap);
15080 diff -NurpP --minimal linux-2.6.32.1/kernel/compat.c linux-2.6.32.1-vs2.3.0.36.27/kernel/compat.c
15081 --- linux-2.6.32.1/kernel/compat.c 2009-09-10 15:26:27.000000000 +0200
15082 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/compat.c 2009-12-03 20:04:56.000000000 +0100
15083 @@ -902,7 +902,7 @@ asmlinkage long compat_sys_time(compat_t
15087 - do_gettimeofday(&tv);
15088 + vx_gettimeofday(&tv);
15092 @@ -927,7 +927,7 @@ asmlinkage long compat_sys_stime(compat_
15096 - do_settimeofday(&tv);
15097 + vx_settimeofday(&tv);
15101 diff -NurpP --minimal linux-2.6.32.1/kernel/exit.c linux-2.6.32.1-vs2.3.0.36.27/kernel/exit.c
15102 --- linux-2.6.32.1/kernel/exit.c 2009-12-03 20:02:57.000000000 +0100
15103 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/exit.c 2009-12-03 20:04:56.000000000 +0100
15105 #include <linux/fs_struct.h>
15106 #include <linux/init_task.h>
15107 #include <linux/perf_event.h>
15108 +#include <linux/vs_limit.h>
15109 +#include <linux/vs_context.h>
15110 +#include <linux/vs_network.h>
15111 +#include <linux/vs_pid.h>
15112 #include <trace/events/sched.h>
15114 #include <asm/uaccess.h>
15115 @@ -488,9 +492,11 @@ static void close_files(struct files_str
15116 filp_close(file, files);
15119 + vx_openfd_dec(i);
15127 @@ -1011,11 +1017,16 @@ NORET_TYPE void do_exit(long code)
15129 validate_creds_for_do_exit(tsk);
15131 + /* needs to stay after exit_notify() */
15132 + exit_vx_info(tsk, code);
15133 + exit_nx_info(tsk);
15137 /* causes final put_task_struct in finish_task_switch(). */
15138 tsk->state = TASK_DEAD;
15140 + printk("bad task: %p [%lx]\n", current, current->state);
15142 /* Avoid "noreturn function does return". */
15144 diff -NurpP --minimal linux-2.6.32.1/kernel/fork.c linux-2.6.32.1-vs2.3.0.36.27/kernel/fork.c
15145 --- linux-2.6.32.1/kernel/fork.c 2009-12-03 20:02:57.000000000 +0100
15146 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/fork.c 2009-12-03 20:04:56.000000000 +0100
15148 #include <linux/magic.h>
15149 #include <linux/perf_event.h>
15150 #include <linux/posix-timers.h>
15151 +#include <linux/vs_context.h>
15152 +#include <linux/vs_network.h>
15153 +#include <linux/vs_limit.h>
15154 +#include <linux/vs_memory.h>
15156 #include <asm/pgtable.h>
15157 #include <asm/pgalloc.h>
15158 @@ -151,6 +155,8 @@ void free_task(struct task_struct *tsk)
15159 account_kernel_stack(tsk->stack, -1);
15160 free_thread_info(tsk->stack);
15161 rt_mutex_debug_task_free(tsk);
15162 + clr_vx_info(&tsk->vx_info);
15163 + clr_nx_info(&tsk->nx_info);
15164 ftrace_graph_exit_task(tsk);
15165 free_task_struct(tsk);
15167 @@ -296,6 +302,8 @@ static int dup_mmap(struct mm_struct *mm
15168 mm->free_area_cache = oldmm->mmap_base;
15169 mm->cached_hole_size = ~0UL;
15171 + __set_mm_counter(mm, file_rss, 0);
15172 + __set_mm_counter(mm, anon_rss, 0);
15173 cpumask_clear(mm_cpumask(mm));
15174 mm->mm_rb = RB_ROOT;
15175 rb_link = &mm->mm_rb.rb_node;
15176 @@ -310,7 +318,7 @@ static int dup_mmap(struct mm_struct *mm
15178 if (mpnt->vm_flags & VM_DONTCOPY) {
15179 long pages = vma_pages(mpnt);
15180 - mm->total_vm -= pages;
15181 + vx_vmpages_sub(mm, pages);
15182 vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file,
15185 @@ -452,8 +460,8 @@ static struct mm_struct * mm_init(struct
15186 (current->mm->flags & MMF_INIT_MASK) : default_dump_filter;
15187 mm->core_state = NULL;
15189 - set_mm_counter(mm, file_rss, 0);
15190 - set_mm_counter(mm, anon_rss, 0);
15191 + __set_mm_counter(mm, file_rss, 0);
15192 + __set_mm_counter(mm, anon_rss, 0);
15193 spin_lock_init(&mm->page_table_lock);
15194 mm->free_area_cache = TASK_UNMAPPED_BASE;
15195 mm->cached_hole_size = ~0UL;
15196 @@ -463,6 +471,7 @@ static struct mm_struct * mm_init(struct
15197 if (likely(!mm_alloc_pgd(mm))) {
15199 mmu_notifier_mm_init(mm);
15200 + set_vx_info(&mm->mm_vx_info, p->vx_info);
15204 @@ -496,6 +505,7 @@ void __mmdrop(struct mm_struct *mm)
15206 destroy_context(mm);
15207 mmu_notifier_mm_destroy(mm);
15208 + clr_vx_info(&mm->mm_vx_info);
15211 EXPORT_SYMBOL_GPL(__mmdrop);
15212 @@ -631,6 +641,7 @@ struct mm_struct *dup_mm(struct task_str
15215 memcpy(mm, oldmm, sizeof(*mm));
15216 + mm->mm_vx_info = NULL;
15218 /* Initializing for Swap token stuff */
15219 mm->token_priority = 0;
15220 @@ -669,6 +680,7 @@ fail_nocontext:
15221 * If init_new_context() failed, we cannot use mmput() to free the mm
15222 * because it calls destroy_context()
15224 + clr_vx_info(&mm->mm_vx_info);
15228 @@ -980,6 +992,8 @@ static struct task_struct *copy_process(
15230 struct task_struct *p;
15231 int cgroup_callbacks_done = 0;
15232 + struct vx_info *vxi;
15233 + struct nx_info *nxi;
15235 if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
15236 return ERR_PTR(-EINVAL);
15237 @@ -1026,12 +1040,28 @@ static struct task_struct *copy_process(
15238 DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
15239 DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
15241 + init_vx_info(&p->vx_info, current_vx_info());
15242 + init_nx_info(&p->nx_info, current_nx_info());
15244 + /* check vserver memory */
15245 + if (p->mm && !(clone_flags & CLONE_VM)) {
15246 + if (vx_vmpages_avail(p->mm, p->mm->total_vm))
15247 + vx_pages_add(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15249 + goto bad_fork_free;
15251 + if (p->mm && vx_flags(VXF_FORK_RSS, 0)) {
15252 + if (!vx_rss_avail(p->mm, get_mm_counter(p->mm, file_rss)))
15253 + goto bad_fork_cleanup_vm;
15256 + if (!vx_nproc_avail(1))
15257 + goto bad_fork_cleanup_vm;
15258 if (atomic_read(&p->real_cred->user->processes) >=
15259 p->signal->rlim[RLIMIT_NPROC].rlim_cur) {
15260 if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
15261 p->real_cred->user != INIT_USER)
15262 - goto bad_fork_free;
15263 + goto bad_fork_cleanup_vm;
15266 retval = copy_creds(p, clone_flags);
15267 @@ -1300,6 +1330,18 @@ static struct task_struct *copy_process(
15270 spin_unlock(¤t->sighand->siglock);
15272 + /* p is copy of current */
15273 + vxi = p->vx_info;
15275 + claim_vx_info(vxi, p);
15276 + atomic_inc(&vxi->cvirt.nr_threads);
15277 + atomic_inc(&vxi->cvirt.total_forks);
15280 + nxi = p->nx_info;
15282 + claim_nx_info(nxi, p);
15283 write_unlock_irq(&tasklist_lock);
15284 proc_fork_connector(p);
15285 cgroup_post_fork(p);
15286 @@ -1341,6 +1383,9 @@ bad_fork_cleanup_cgroup:
15287 bad_fork_cleanup_count:
15288 atomic_dec(&p->cred->user->processes);
15290 +bad_fork_cleanup_vm:
15291 + if (p->mm && !(clone_flags & CLONE_VM))
15292 + vx_pages_sub(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15296 diff -NurpP --minimal linux-2.6.32.1/kernel/kthread.c linux-2.6.32.1-vs2.3.0.36.27/kernel/kthread.c
15297 --- linux-2.6.32.1/kernel/kthread.c 2009-12-03 20:02:58.000000000 +0100
15298 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/kthread.c 2009-12-03 20:04:56.000000000 +0100
15300 #include <linux/file.h>
15301 #include <linux/module.h>
15302 #include <linux/mutex.h>
15303 +#include <linux/vs_pid.h>
15304 #include <trace/events/sched.h>
15306 static DEFINE_SPINLOCK(kthread_create_lock);
15307 diff -NurpP --minimal linux-2.6.32.1/kernel/Makefile linux-2.6.32.1-vs2.3.0.36.27/kernel/Makefile
15308 --- linux-2.6.32.1/kernel/Makefile 2009-12-03 20:02:57.000000000 +0100
15309 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/Makefile 2009-12-03 20:04:56.000000000 +0100
15310 @@ -23,6 +23,7 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg
15311 CFLAGS_REMOVE_sched_clock.o = -pg
15315 obj-$(CONFIG_FREEZER) += freezer.o
15316 obj-$(CONFIG_PROFILING) += profile.o
15317 obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o
15318 diff -NurpP --minimal linux-2.6.32.1/kernel/nsproxy.c linux-2.6.32.1-vs2.3.0.36.27/kernel/nsproxy.c
15319 --- linux-2.6.32.1/kernel/nsproxy.c 2009-09-10 15:26:28.000000000 +0200
15320 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/nsproxy.c 2009-12-03 20:04:56.000000000 +0100
15322 #include <linux/mnt_namespace.h>
15323 #include <linux/utsname.h>
15324 #include <linux/pid_namespace.h>
15325 +#include <linux/vserver/global.h>
15326 +#include <linux/vserver/debug.h>
15327 #include <net/net_namespace.h>
15328 #include <linux/ipc_namespace.h>
15330 @@ -31,8 +33,11 @@ static inline struct nsproxy *create_nsp
15331 struct nsproxy *nsproxy;
15333 nsproxy = kmem_cache_alloc(nsproxy_cachep, GFP_KERNEL);
15336 atomic_set(&nsproxy->count, 1);
15337 + atomic_inc(&vs_global_nsproxy);
15339 + vxdprintk(VXD_CBIT(space, 2), "create_nsproxy = %p[1]", nsproxy);
15343 @@ -41,41 +46,52 @@ static inline struct nsproxy *create_nsp
15344 * Return the newly created nsproxy. Do not attach this to the task,
15345 * leave it to the caller to do proper locking and attach it to task.
15347 -static struct nsproxy *create_new_namespaces(unsigned long flags,
15348 - struct task_struct *tsk, struct fs_struct *new_fs)
15349 +static struct nsproxy *unshare_namespaces(unsigned long flags,
15350 + struct nsproxy *orig, struct fs_struct *new_fs)
15352 struct nsproxy *new_nsp;
15355 + vxdprintk(VXD_CBIT(space, 4),
15356 + "unshare_namespaces(0x%08lx,%p,%p)",
15357 + flags, orig, new_fs);
15359 new_nsp = create_nsproxy();
15361 return ERR_PTR(-ENOMEM);
15363 - new_nsp->mnt_ns = copy_mnt_ns(flags, tsk->nsproxy->mnt_ns, new_fs);
15364 + new_nsp->mnt_ns = copy_mnt_ns(flags, orig->mnt_ns, new_fs);
15365 if (IS_ERR(new_nsp->mnt_ns)) {
15366 err = PTR_ERR(new_nsp->mnt_ns);
15370 - new_nsp->uts_ns = copy_utsname(flags, tsk->nsproxy->uts_ns);
15371 + new_nsp->uts_ns = copy_utsname(flags, orig->uts_ns);
15372 if (IS_ERR(new_nsp->uts_ns)) {
15373 err = PTR_ERR(new_nsp->uts_ns);
15377 - new_nsp->ipc_ns = copy_ipcs(flags, tsk->nsproxy->ipc_ns);
15378 + new_nsp->ipc_ns = copy_ipcs(flags, orig->ipc_ns);
15379 if (IS_ERR(new_nsp->ipc_ns)) {
15380 err = PTR_ERR(new_nsp->ipc_ns);
15384 - new_nsp->pid_ns = copy_pid_ns(flags, task_active_pid_ns(tsk));
15385 + new_nsp->pid_ns = copy_pid_ns(flags, orig->pid_ns);
15386 if (IS_ERR(new_nsp->pid_ns)) {
15387 err = PTR_ERR(new_nsp->pid_ns);
15391 - new_nsp->net_ns = copy_net_ns(flags, tsk->nsproxy->net_ns);
15393 + new_nsp->user_ns = copy_user_ns(flags, orig->user_ns);
15394 + if (IS_ERR(new_nsp->user_ns)) {
15395 + err = PTR_ERR(new_nsp->user_ns);
15399 + new_nsp->net_ns = copy_net_ns(flags, orig->net_ns);
15400 if (IS_ERR(new_nsp->net_ns)) {
15401 err = PTR_ERR(new_nsp->net_ns);
15403 @@ -100,6 +116,38 @@ out_ns:
15404 return ERR_PTR(err);
15407 +static struct nsproxy *create_new_namespaces(int flags, struct task_struct *tsk,
15408 + struct fs_struct *new_fs)
15410 + return unshare_namespaces(flags, tsk->nsproxy, new_fs);
15414 + * copies the nsproxy, setting refcount to 1, and grabbing a
15415 + * reference to all contained namespaces.
15417 +struct nsproxy *copy_nsproxy(struct nsproxy *orig)
15419 + struct nsproxy *ns = create_nsproxy();
15422 + memcpy(ns, orig, sizeof(struct nsproxy));
15423 + atomic_set(&ns->count, 1);
15426 + get_mnt_ns(ns->mnt_ns);
15428 + get_uts_ns(ns->uts_ns);
15430 + get_ipc_ns(ns->ipc_ns);
15432 + get_pid_ns(ns->pid_ns);
15434 + get_net(ns->net_ns);
15440 * called from clone. This now handles copy for nsproxy and all
15441 * namespaces therein.
15442 @@ -107,9 +155,12 @@ out_ns:
15443 int copy_namespaces(unsigned long flags, struct task_struct *tsk)
15445 struct nsproxy *old_ns = tsk->nsproxy;
15446 - struct nsproxy *new_ns;
15447 + struct nsproxy *new_ns = NULL;
15450 + vxdprintk(VXD_CBIT(space, 7), "copy_namespaces(0x%08lx,%p[%p])",
15451 + flags, tsk, old_ns);
15456 @@ -119,7 +170,7 @@ int copy_namespaces(unsigned long flags,
15457 CLONE_NEWPID | CLONE_NEWNET)))
15460 - if (!capable(CAP_SYS_ADMIN)) {
15461 + if (!vx_can_unshare(CAP_SYS_ADMIN, flags)) {
15465 @@ -146,6 +197,9 @@ int copy_namespaces(unsigned long flags,
15468 put_nsproxy(old_ns);
15469 + vxdprintk(VXD_CBIT(space, 3),
15470 + "copy_namespaces(0x%08lx,%p[%p]) = %d [%p]",
15471 + flags, tsk, old_ns, err, new_ns);
15475 @@ -159,7 +213,9 @@ void free_nsproxy(struct nsproxy *ns)
15476 put_ipc_ns(ns->ipc_ns);
15478 put_pid_ns(ns->pid_ns);
15479 - put_net(ns->net_ns);
15481 + put_net(ns->net_ns);
15482 + atomic_dec(&vs_global_nsproxy);
15483 kmem_cache_free(nsproxy_cachep, ns);
15486 @@ -172,11 +228,15 @@ int unshare_nsproxy_namespaces(unsigned
15490 + vxdprintk(VXD_CBIT(space, 4),
15491 + "unshare_nsproxy_namespaces(0x%08lx,[%p])",
15492 + unshare_flags, current->nsproxy);
15494 if (!(unshare_flags & (CLONE_NEWNS | CLONE_NEWUTS | CLONE_NEWIPC |
15498 - if (!capable(CAP_SYS_ADMIN))
15499 + if (!vx_can_unshare(CAP_SYS_ADMIN, unshare_flags))
15502 *new_nsp = create_new_namespaces(unshare_flags, current,
15503 diff -NurpP --minimal linux-2.6.32.1/kernel/pid.c linux-2.6.32.1-vs2.3.0.36.27/kernel/pid.c
15504 --- linux-2.6.32.1/kernel/pid.c 2009-12-03 20:02:58.000000000 +0100
15505 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/pid.c 2009-12-03 20:04:56.000000000 +0100
15507 #include <linux/pid_namespace.h>
15508 #include <linux/init_task.h>
15509 #include <linux/syscalls.h>
15510 +#include <linux/vs_pid.h>
15512 #define pid_hashfn(nr, ns) \
15513 hash_long((unsigned long)nr + (unsigned long)ns, pidhash_shift)
15514 @@ -305,7 +306,7 @@ EXPORT_SYMBOL_GPL(find_pid_ns);
15516 struct pid *find_vpid(int nr)
15518 - return find_pid_ns(nr, current->nsproxy->pid_ns);
15519 + return find_pid_ns(vx_rmap_pid(nr), current->nsproxy->pid_ns);
15521 EXPORT_SYMBOL_GPL(find_vpid);
15523 @@ -365,6 +366,9 @@ void transfer_pid(struct task_struct *ol
15524 struct task_struct *pid_task(struct pid *pid, enum pid_type type)
15526 struct task_struct *result = NULL;
15528 + if (type == PIDTYPE_REALPID)
15529 + type = PIDTYPE_PID;
15531 struct hlist_node *first;
15532 first = rcu_dereference(pid->tasks[type].first);
15533 @@ -380,7 +384,7 @@ EXPORT_SYMBOL(pid_task);
15535 struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns)
15537 - return pid_task(find_pid_ns(nr, ns), PIDTYPE_PID);
15538 + return pid_task(find_pid_ns(vx_rmap_pid(nr), ns), PIDTYPE_PID);
15541 struct task_struct *find_task_by_vpid(pid_t vnr)
15542 @@ -422,7 +426,7 @@ struct pid *find_get_pid(pid_t nr)
15544 EXPORT_SYMBOL_GPL(find_get_pid);
15546 -pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15547 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns)
15551 @@ -435,6 +439,11 @@ pid_t pid_nr_ns(struct pid *pid, struct
15555 +pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15557 + return vx_map_pid(pid_unmapped_nr_ns(pid, ns));
15560 pid_t pid_vnr(struct pid *pid)
15562 return pid_nr_ns(pid, current->nsproxy->pid_ns);
15563 diff -NurpP --minimal linux-2.6.32.1/kernel/pid_namespace.c linux-2.6.32.1-vs2.3.0.36.27/kernel/pid_namespace.c
15564 --- linux-2.6.32.1/kernel/pid_namespace.c 2009-12-03 20:02:58.000000000 +0100
15565 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/pid_namespace.c 2009-12-03 20:04:56.000000000 +0100
15567 #include <linux/syscalls.h>
15568 #include <linux/err.h>
15569 #include <linux/acct.h>
15570 +#include <linux/vserver/global.h>
15572 #define BITS_PER_PAGE (PAGE_SIZE*8)
15574 @@ -86,6 +87,7 @@ static struct pid_namespace *create_pid_
15577 kref_init(&ns->kref);
15578 + atomic_inc(&vs_global_pid_ns);
15580 ns->parent = get_pid_ns(parent_pid_ns);
15582 @@ -111,6 +113,7 @@ static void destroy_pid_namespace(struct
15584 for (i = 0; i < PIDMAP_ENTRIES; i++)
15585 kfree(ns->pidmap[i].page);
15586 + atomic_dec(&vs_global_pid_ns);
15587 kmem_cache_free(pid_ns_cachep, ns);
15590 diff -NurpP --minimal linux-2.6.32.1/kernel/posix-timers.c linux-2.6.32.1-vs2.3.0.36.27/kernel/posix-timers.c
15591 --- linux-2.6.32.1/kernel/posix-timers.c 2009-12-03 20:02:58.000000000 +0100
15592 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/posix-timers.c 2009-12-03 20:04:56.000000000 +0100
15594 #include <linux/wait.h>
15595 #include <linux/workqueue.h>
15596 #include <linux/module.h>
15597 +#include <linux/vs_context.h>
15600 * Management arrays for POSIX timers. Timers are kept in slab memory
15601 @@ -363,6 +364,7 @@ int posix_timer_event(struct k_itimer *t
15603 struct task_struct *task;
15604 int shared, ret = -1;
15607 * FIXME: if ->sigq is queued we can race with
15608 * dequeue_signal()->do_schedule_next_timer().
15609 @@ -379,10 +381,18 @@ int posix_timer_event(struct k_itimer *t
15611 task = pid_task(timr->it_pid, PIDTYPE_PID);
15613 + struct vx_info_save vxis;
15614 + struct vx_info *vxi;
15616 + vxi = get_vx_info(task->vx_info);
15617 + enter_vx_info(vxi, &vxis);
15618 shared = !(timr->it_sigev_notify & SIGEV_THREAD_ID);
15619 ret = send_sigqueue(timr->sigq, task, shared);
15620 + leave_vx_info(&vxis);
15621 + put_vx_info(vxi);
15625 /* If we failed to send the signal the timer stops. */
15628 diff -NurpP --minimal linux-2.6.32.1/kernel/printk.c linux-2.6.32.1-vs2.3.0.36.27/kernel/printk.c
15629 --- linux-2.6.32.1/kernel/printk.c 2009-12-03 20:02:58.000000000 +0100
15630 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/printk.c 2009-12-03 20:04:56.000000000 +0100
15632 #include <linux/bootmem.h>
15633 #include <linux/syscalls.h>
15634 #include <linux/kexec.h>
15635 +#include <linux/vs_cvirt.h>
15637 #include <asm/uaccess.h>
15639 @@ -276,18 +277,13 @@ int do_syslog(int type, char __user *buf
15640 unsigned i, j, limit, count;
15646 error = security_syslog(type);
15651 - case 0: /* Close log */
15653 - case 1: /* Open log */
15655 - case 2: /* Read from log */
15656 + if ((type >= 2) && (type <= 4)) {
15658 if (!buf || len < 0)
15660 @@ -298,6 +294,16 @@ int do_syslog(int type, char __user *buf
15665 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
15666 + return vx_do_syslog(type, buf, len);
15669 + case 0: /* Close log */
15671 + case 1: /* Open log */
15673 + case 2: /* Read from log */
15674 error = wait_event_interruptible(log_wait,
15675 (log_start - log_end));
15677 @@ -322,16 +328,6 @@ int do_syslog(int type, char __user *buf
15680 case 3: /* Read last kernel messages */
15682 - if (!buf || len < 0)
15687 - if (!access_ok(VERIFY_WRITE, buf, len)) {
15692 if (count > log_buf_len)
15693 count = log_buf_len;
15694 diff -NurpP --minimal linux-2.6.32.1/kernel/ptrace.c linux-2.6.32.1-vs2.3.0.36.27/kernel/ptrace.c
15695 --- linux-2.6.32.1/kernel/ptrace.c 2009-12-03 20:02:58.000000000 +0100
15696 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/ptrace.c 2009-12-03 20:04:56.000000000 +0100
15698 #include <linux/pid_namespace.h>
15699 #include <linux/syscalls.h>
15700 #include <linux/uaccess.h>
15701 +#include <linux/vs_context.h>
15705 @@ -151,6 +152,11 @@ int __ptrace_may_access(struct task_stru
15706 dumpable = get_dumpable(task->mm);
15707 if (!dumpable && !capable(CAP_SYS_PTRACE))
15709 + if (!vx_check(task->xid, VS_ADMIN_P|VS_IDENT))
15711 + if (!vx_check(task->xid, VS_IDENT) &&
15712 + !task_vx_flags(task, VXF_STATE_ADMIN, 0))
15715 return security_ptrace_access_check(task, mode);
15717 @@ -621,6 +627,10 @@ SYSCALL_DEFINE4(ptrace, long, request, l
15722 + if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
15723 + goto out_put_task_struct;
15725 if (request == PTRACE_ATTACH) {
15726 ret = ptrace_attach(child);
15728 diff -NurpP --minimal linux-2.6.32.1/kernel/sched.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sched.c
15729 --- linux-2.6.32.1/kernel/sched.c 2009-12-03 20:02:58.000000000 +0100
15730 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sched.c 2009-12-03 20:04:56.000000000 +0100
15732 #include <linux/debugfs.h>
15733 #include <linux/ctype.h>
15734 #include <linux/ftrace.h>
15735 +#include <linux/vs_sched.h>
15736 +#include <linux/vs_cvirt.h>
15738 #include <asm/tlb.h>
15739 #include <asm/irq_regs.h>
15740 @@ -237,6 +239,15 @@ static DEFINE_MUTEX(sched_domains_mutex)
15742 #include <linux/cgroup.h>
15744 +#if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_CFS_HARD_LIMITS)
15745 +struct cfs_bandwidth {
15746 + spinlock_t cfs_runtime_lock;
15747 + ktime_t cfs_period;
15749 + struct hrtimer cfs_period_timer;
15755 static LIST_HEAD(task_groups);
15756 @@ -257,6 +268,9 @@ struct task_group {
15757 /* runqueue "owned" by this group on each cpu */
15758 struct cfs_rq **cfs_rq;
15759 unsigned long shares;
15760 +#ifdef CONFIG_CFS_HARD_LIMITS
15761 + struct cfs_bandwidth cfs_bandwidth;
15765 #ifdef CONFIG_RT_GROUP_SCHED
15766 @@ -446,6 +460,19 @@ struct cfs_rq {
15767 unsigned long rq_weight;
15770 +#ifdef CONFIG_CFS_HARD_LIMITS
15771 + /* set when the group is throttled on this cpu */
15772 + int cfs_throttled;
15774 + /* runtime currently consumed by the group on this rq */
15777 + /* runtime available to the group on this rq */
15780 + /* Protects the cfs runtime related fields of this cfs_rq */
15781 + spinlock_t cfs_runtime_lock;
15785 /* Real-Time classes' related field in a runqueue: */
15786 @@ -1607,6 +1634,7 @@ static void update_group_shares_cpu(stru
15790 +static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq);
15792 * Re-compute the task group their per cpu shares over the given domain.
15793 * This needs to be done in a bottom-up fashion because the rq weight of a
15794 @@ -1634,8 +1662,10 @@ static int tg_shares_up(struct task_grou
15795 * If there are currently no tasks on the cpu pretend there
15796 * is one of average load so that when a new task gets to
15797 * run here it will not get delayed by group starvation.
15798 + * Also if the group is throttled on this cpu, pretend that
15799 + * it has no tasks.
15802 + if (!weight || cfs_rq_throttled(tg->cfs_rq[i]))
15803 weight = NICE_0_LOAD;
15805 rq_weight += weight;
15806 @@ -1811,6 +1841,175 @@ static void cfs_rq_set_shares(struct cfs
15808 static void calc_load_account_active(struct rq *this_rq);
15811 +#if defined(CONFIG_RT_GROUP_SCHED) || defined(CONFIG_FAIR_GROUP_SCHED)
15814 +static inline const struct cpumask *sched_bw_period_mask(void)
15816 + return cpu_rq(smp_processor_id())->rd->span;
15818 +#else /* !CONFIG_SMP */
15819 +static inline const struct cpumask *sched_bw_period_mask(void)
15821 + return cpu_online_mask;
15823 +#endif /* CONFIG_SMP */
15826 +static inline const struct cpumask *sched_bw_period_mask(void)
15828 + return cpu_online_mask;
15833 +#ifdef CONFIG_FAIR_GROUP_SCHED
15834 +#ifdef CONFIG_CFS_HARD_LIMITS
15837 + * Runtime allowed for a cfs group before it is hard limited.
15838 + * default: Infinite which means no hard limiting.
15840 +u64 sched_cfs_runtime = RUNTIME_INF;
15843 + * period over which we hard limit the cfs group's bandwidth.
15846 +u64 sched_cfs_period = 500000;
15848 +static inline u64 global_cfs_period(void)
15850 + return sched_cfs_period * NSEC_PER_USEC;
15853 +static inline u64 global_cfs_runtime(void)
15855 + return RUNTIME_INF;
15858 +void do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b);
15860 +static inline void cfs_rq_runtime_lock(struct cfs_rq *cfs_rq)
15862 + spin_lock(&cfs_rq->cfs_runtime_lock);
15865 +static inline void cfs_rq_runtime_unlock(struct cfs_rq *cfs_rq)
15867 + spin_unlock(&cfs_rq->cfs_runtime_lock);
15871 + * Refresh the runtimes of the throttled groups.
15872 + * But nothing much to do now, will populate this in later patches.
15874 +static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
15876 + struct cfs_bandwidth *cfs_b =
15877 + container_of(timer, struct cfs_bandwidth, cfs_period_timer);
15879 + do_sched_cfs_period_timer(cfs_b);
15880 + hrtimer_add_expires_ns(timer, ktime_to_ns(cfs_b->cfs_period));
15881 + return HRTIMER_RESTART;
15885 + * TODO: Check if this kind of timer setup is sufficient for cfs or
15886 + * should we do what rt is doing.
15888 +static void start_cfs_bandwidth(struct task_group *tg)
15890 + struct cfs_bandwidth *cfs_b = &tg->cfs_bandwidth;
15893 + * Timer isn't setup for groups with infinite runtime
15895 + if (cfs_b->cfs_runtime == RUNTIME_INF)
15898 + if (hrtimer_active(&cfs_b->cfs_period_timer))
15901 + hrtimer_start_range_ns(&cfs_b->cfs_period_timer, cfs_b->cfs_period,
15902 + 0, HRTIMER_MODE_REL);
15905 +static void init_cfs_bandwidth(struct task_group *tg)
15907 + struct cfs_bandwidth *cfs_b = &tg->cfs_bandwidth;
15909 + cfs_b->cfs_period = ns_to_ktime(global_cfs_period());
15910 + cfs_b->cfs_runtime = global_cfs_runtime();
15912 + spin_lock_init(&cfs_b->cfs_runtime_lock);
15914 + hrtimer_init(&cfs_b->cfs_period_timer,
15915 + CLOCK_MONOTONIC, HRTIMER_MODE_REL);
15916 + cfs_b->cfs_period_timer.function = &sched_cfs_period_timer;
15919 +static inline void destroy_cfs_bandwidth(struct task_group *tg)
15921 + hrtimer_cancel(&tg->cfs_bandwidth.cfs_period_timer);
15924 +static void init_cfs_hard_limits(struct cfs_rq *cfs_rq, struct task_group *tg)
15926 + cfs_rq->cfs_time = 0;
15927 + cfs_rq->cfs_throttled = 0;
15928 + cfs_rq->cfs_runtime = tg->cfs_bandwidth.cfs_runtime;
15929 + spin_lock_init(&cfs_rq->cfs_runtime_lock);
15932 +#else /* !CONFIG_CFS_HARD_LIMITS */
15934 +static void init_cfs_bandwidth(struct task_group *tg)
15939 +static inline void destroy_cfs_bandwidth(struct task_group *tg)
15944 +static void init_cfs_hard_limits(struct cfs_rq *cfs_rq, struct task_group *tg)
15949 +static inline void cfs_rq_runtime_lock(struct cfs_rq *cfs_rq)
15954 +static inline void cfs_rq_runtime_unlock(struct cfs_rq *cfs_rq)
15959 +#endif /* CONFIG_CFS_HARD_LIMITS */
15960 +#else /* !CONFIG_FAIR_GROUP_SCHED */
15962 +static inline void cfs_rq_runtime_lock(struct cfs_rq *cfs_rq)
15967 +static inline void cfs_rq_runtime_unlock(struct cfs_rq *cfs_rq)
15972 +static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
15977 +#endif /* CONFIG_FAIR_GROUP_SCHED */
15979 #include "sched_stats.h"
15980 #include "sched_idletask.c"
15981 #include "sched_fair.c"
15982 @@ -2965,9 +3164,17 @@ EXPORT_SYMBOL(avenrun);
15984 void get_avenrun(unsigned long *loads, unsigned long offset, int shift)
15986 - loads[0] = (avenrun[0] + offset) << shift;
15987 - loads[1] = (avenrun[1] + offset) << shift;
15988 - loads[2] = (avenrun[2] + offset) << shift;
15989 + if (vx_flags(VXF_VIRT_LOAD, 0)) {
15990 + struct vx_info *vxi = current_vx_info();
15992 + loads[0] = (vxi->cvirt.load[0] + offset) << shift;
15993 + loads[1] = (vxi->cvirt.load[1] + offset) << shift;
15994 + loads[2] = (vxi->cvirt.load[2] + offset) << shift;
15996 + loads[0] = (avenrun[0] + offset) << shift;
15997 + loads[1] = (avenrun[1] + offset) << shift;
15998 + loads[2] = (avenrun[2] + offset) << shift;
16002 static unsigned long
16003 @@ -5006,16 +5213,19 @@ void account_user_time(struct task_struc
16004 cputime_t cputime_scaled)
16006 struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
16007 + struct vx_info *vxi = p->vx_info; /* p is _always_ current */
16009 + int nice = (TASK_NICE(p) > 0);
16011 /* Add user time to process. */
16012 p->utime = cputime_add(p->utime, cputime);
16013 p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
16014 + vx_account_user(vxi, cputime, nice);
16015 account_group_user_time(p, cputime);
16017 /* Add user time to cpustat. */
16018 tmp = cputime_to_cputime64(cputime);
16019 - if (TASK_NICE(p) > 0)
16021 cpustat->nice = cputime64_add(cpustat->nice, tmp);
16023 cpustat->user = cputime64_add(cpustat->user, tmp);
16024 @@ -5061,6 +5271,7 @@ void account_system_time(struct task_str
16025 cputime_t cputime, cputime_t cputime_scaled)
16027 struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
16028 + struct vx_info *vxi = p->vx_info; /* p is _always_ current */
16031 if ((p->flags & PF_VCPU) && (irq_count() - hardirq_offset == 0)) {
16032 @@ -5071,6 +5282,7 @@ void account_system_time(struct task_str
16033 /* Add system time to process. */
16034 p->stime = cputime_add(p->stime, cputime);
16035 p->stimescaled = cputime_add(p->stimescaled, cputime_scaled);
16036 + vx_account_system(vxi, cputime, 0 /* do we have idle time? */);
16037 account_group_system_time(p, cputime);
16039 /* Add system time to cpustat. */
16040 @@ -6106,7 +6318,7 @@ SYSCALL_DEFINE1(nice, int, increment)
16043 if (increment < 0 && !can_nice(current, nice))
16045 + return vx_flags(VXF_IGNEG_NICE, 0) ? 0 : -EPERM;
16047 retval = security_task_setnice(current, nice);
16049 @@ -9164,6 +9376,32 @@ static int update_sched_domains(struct n
16054 +static void disable_runtime(struct rq *rq)
16056 + unsigned long flags;
16058 + spin_lock_irqsave(&rq->lock, flags);
16059 +#if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_CFS_HARD_LIMITS)
16060 + disable_runtime_cfs(rq);
16062 + disable_runtime_rt(rq);
16063 + spin_unlock_irqrestore(&rq->lock, flags);
16066 +static void enable_runtime(struct rq *rq)
16068 + unsigned long flags;
16070 + spin_lock_irqsave(&rq->lock, flags);
16071 +#if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_CFS_HARD_LIMITS)
16072 + enable_runtime_cfs(rq);
16074 + enable_runtime_rt(rq);
16075 + spin_unlock_irqrestore(&rq->lock, flags);
16079 static int update_runtime(struct notifier_block *nfb,
16080 unsigned long action, void *hcpu)
16082 @@ -9296,6 +9534,7 @@ static void init_tg_cfs_entry(struct tas
16083 struct rq *rq = cpu_rq(cpu);
16084 tg->cfs_rq[cpu] = cfs_rq;
16085 init_cfs_rq(cfs_rq, rq);
16086 + init_cfs_hard_limits(cfs_rq, tg);
16089 list_add(&cfs_rq->leaf_cfs_rq_list, &rq->leaf_cfs_rq_list);
16090 @@ -9425,6 +9664,10 @@ void __init sched_init(void)
16091 #endif /* CONFIG_USER_SCHED */
16092 #endif /* CONFIG_RT_GROUP_SCHED */
16094 +#ifdef CONFIG_FAIR_GROUP_SCHED
16095 + init_cfs_bandwidth(&init_task_group);
16098 #ifdef CONFIG_GROUP_SCHED
16099 list_add(&init_task_group.list, &task_groups);
16100 INIT_LIST_HEAD(&init_task_group.children);
16101 @@ -9451,6 +9694,7 @@ void __init sched_init(void)
16102 init_cfs_rq(&rq->cfs, rq);
16103 init_rt_rq(&rq->rt, rq);
16104 #ifdef CONFIG_FAIR_GROUP_SCHED
16105 + init_cfs_hard_limits(&rq->cfs, &init_task_group);
16106 init_task_group.shares = init_task_group_load;
16107 INIT_LIST_HEAD(&rq->leaf_cfs_rq_list);
16108 #ifdef CONFIG_CGROUP_SCHED
16109 @@ -9726,6 +9970,7 @@ static void free_fair_sched_group(struct
16113 + destroy_cfs_bandwidth(tg);
16114 for_each_possible_cpu(i) {
16116 kfree(tg->cfs_rq[i]);
16117 @@ -9752,6 +9997,7 @@ int alloc_fair_sched_group(struct task_g
16121 + init_cfs_bandwidth(tg);
16122 tg->shares = NICE_0_LOAD;
16124 for_each_possible_cpu(i) {
16125 @@ -10475,6 +10721,100 @@ static u64 cpu_shares_read_u64(struct cg
16127 return (u64) tg->shares;
16130 +#ifdef CONFIG_CFS_HARD_LIMITS
16132 +static int tg_set_cfs_bandwidth(struct task_group *tg,
16133 + u64 cfs_period, u64 cfs_runtime)
16137 + spin_lock_irq(&tg->cfs_bandwidth.cfs_runtime_lock);
16138 + tg->cfs_bandwidth.cfs_period = ns_to_ktime(cfs_period);
16139 + tg->cfs_bandwidth.cfs_runtime = cfs_runtime;
16141 + for_each_possible_cpu(i) {
16142 + struct cfs_rq *cfs_rq = tg->cfs_rq[i];
16144 + cfs_rq_runtime_lock(cfs_rq);
16145 + cfs_rq->cfs_runtime = cfs_runtime;
16146 + cfs_rq_runtime_unlock(cfs_rq);
16149 + start_cfs_bandwidth(tg);
16150 + spin_unlock_irq(&tg->cfs_bandwidth.cfs_runtime_lock);
16154 +int tg_set_cfs_runtime(struct task_group *tg, long cfs_runtime_us)
16156 + u64 cfs_runtime, cfs_period;
16158 + cfs_period = ktime_to_ns(tg->cfs_bandwidth.cfs_period);
16159 + cfs_runtime = (u64)cfs_runtime_us * NSEC_PER_USEC;
16160 + if (cfs_runtime_us < 0)
16161 + cfs_runtime = RUNTIME_INF;
16163 + return tg_set_cfs_bandwidth(tg, cfs_period, cfs_runtime);
16166 +long tg_get_cfs_runtime(struct task_group *tg)
16168 + u64 cfs_runtime_us;
16170 + if (tg->cfs_bandwidth.cfs_runtime == RUNTIME_INF)
16173 + cfs_runtime_us = tg->cfs_bandwidth.cfs_runtime;
16174 + do_div(cfs_runtime_us, NSEC_PER_USEC);
16175 + return cfs_runtime_us;
16178 +int tg_set_cfs_period(struct task_group *tg, long cfs_period_us)
16180 + u64 cfs_runtime, cfs_period;
16182 + cfs_period = (u64)cfs_period_us * NSEC_PER_USEC;
16183 + cfs_runtime = tg->cfs_bandwidth.cfs_runtime;
16185 + if (cfs_period == 0)
16188 + return tg_set_cfs_bandwidth(tg, cfs_period, cfs_runtime);
16191 +long tg_get_cfs_period(struct task_group *tg)
16193 + u64 cfs_period_us;
16195 + cfs_period_us = ktime_to_ns(tg->cfs_bandwidth.cfs_period);
16196 + do_div(cfs_period_us, NSEC_PER_USEC);
16197 + return cfs_period_us;
16200 +static s64 cpu_cfs_runtime_read_s64(struct cgroup *cgrp, struct cftype *cft)
16202 + return tg_get_cfs_runtime(cgroup_tg(cgrp));
16205 +static int cpu_cfs_runtime_write_s64(struct cgroup *cgrp, struct cftype *cftype,
16206 + s64 cfs_runtime_us)
16208 + return tg_set_cfs_runtime(cgroup_tg(cgrp), cfs_runtime_us);
16211 +static u64 cpu_cfs_period_read_u64(struct cgroup *cgrp, struct cftype *cft)
16213 + return tg_get_cfs_period(cgroup_tg(cgrp));
16216 +static int cpu_cfs_period_write_u64(struct cgroup *cgrp, struct cftype *cftype,
16217 + u64 cfs_period_us)
16219 + return tg_set_cfs_period(cgroup_tg(cgrp), cfs_period_us);
16222 +#endif /* CONFIG_CFS_HARD_LIMITS */
16223 #endif /* CONFIG_FAIR_GROUP_SCHED */
16225 #ifdef CONFIG_RT_GROUP_SCHED
16226 @@ -10508,6 +10848,18 @@ static struct cftype cpu_files[] = {
16227 .read_u64 = cpu_shares_read_u64,
16228 .write_u64 = cpu_shares_write_u64,
16230 +#ifdef CONFIG_CFS_HARD_LIMITS
16232 + .name = "cfs_runtime_us",
16233 + .read_s64 = cpu_cfs_runtime_read_s64,
16234 + .write_s64 = cpu_cfs_runtime_write_s64,
16237 + .name = "cfs_period_us",
16238 + .read_u64 = cpu_cfs_period_read_u64,
16239 + .write_u64 = cpu_cfs_period_write_u64,
16241 +#endif /* CONFIG_CFS_HARD_LIMITS */
16243 #ifdef CONFIG_RT_GROUP_SCHED
16245 diff -NurpP --minimal linux-2.6.32.1/kernel/sched_debug.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_debug.c
16246 --- linux-2.6.32.1/kernel/sched_debug.c 2009-12-03 20:02:58.000000000 +0100
16247 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_debug.c 2009-12-03 20:04:56.000000000 +0100
16248 @@ -80,6 +80,11 @@ static void print_cfs_group_stats(struct
16252 +#ifdef CONFIG_CFS_HARD_LIMITS
16253 + PN(se->throttle_max);
16254 + PN(se->throttle_sum);
16255 + P(se->throttle_count);
16258 P(se->load.weight);
16260 @@ -214,6 +219,16 @@ void print_cfs_rq(struct seq_file *m, in
16262 SEQ_printf(m, " .%-30s: %lu\n", "shares", cfs_rq->shares);
16264 +#ifdef CONFIG_CFS_HARD_LIMITS
16265 + spin_lock_irqsave(&rq->lock, flags);
16266 + SEQ_printf(m, " .%-30s: %d\n", "cfs_throttled",
16267 + cfs_rq->cfs_throttled);
16268 + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "cfs_time",
16269 + SPLIT_NS(cfs_rq->cfs_time));
16270 + SEQ_printf(m, " .%-30s: %Ld.%06ld\n", "cfs_runtime",
16271 + SPLIT_NS(cfs_rq->cfs_runtime));
16272 + spin_unlock_irqrestore(&rq->lock, flags);
16273 +#endif /* CONFIG_CFS_HARD_LIMITS */
16274 print_cfs_group_stats(m, cpu, cfs_rq->tg);
16277 @@ -310,7 +325,7 @@ static int sched_debug_show(struct seq_f
16278 u64 now = ktime_to_ns(ktime_get());
16281 - SEQ_printf(m, "Sched Debug Version: v0.09, %s %.*s\n",
16282 + SEQ_printf(m, "Sched Debug Version: v0.10, %s %.*s\n",
16283 init_utsname()->release,
16284 (int)strcspn(init_utsname()->version, " "),
16285 init_utsname()->version);
16286 diff -NurpP --minimal linux-2.6.32.1/kernel/sched_fair.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_fair.c
16287 --- linux-2.6.32.1/kernel/sched_fair.c 2009-12-03 20:02:58.000000000 +0100
16288 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_fair.c 2009-12-03 20:04:56.000000000 +0100
16289 @@ -189,7 +189,308 @@ find_matching_se(struct sched_entity **s
16293 -#else /* !CONFIG_FAIR_GROUP_SCHED */
16294 +#ifdef CONFIG_CFS_HARD_LIMITS
16296 +static inline void update_stats_throttle_start(struct cfs_rq *cfs_rq,
16297 + struct sched_entity *se)
16299 + schedstat_set(se->throttle_start, rq_of(cfs_rq)->clock);
16302 +static inline void update_stats_throttle_end(struct cfs_rq *cfs_rq,
16303 + struct sched_entity *se)
16305 + schedstat_set(se->throttle_max, max(se->throttle_max,
16306 + rq_of(cfs_rq)->clock - se->throttle_start));
16307 + schedstat_set(se->throttle_count, se->throttle_count + 1);
16308 + schedstat_set(se->throttle_sum, se->throttle_sum +
16309 + rq_of(cfs_rq)->clock - se->throttle_start);
16310 + schedstat_set(se->throttle_start, 0);
16314 +struct cfs_rq *sched_cfs_period_cfs_rq(struct cfs_bandwidth *cfs_b, int cpu)
16316 + return container_of(cfs_b, struct task_group,
16317 + cfs_bandwidth)->cfs_rq[cpu];
16320 +static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
16322 + return cfs_rq->cfs_throttled;
16327 + * Ensure this RQ takes back all the runtime it lend to its neighbours.
16329 +static void disable_runtime_cfs(struct rq *rq)
16331 + struct root_domain *rd = rq->rd;
16332 + struct cfs_rq *cfs_rq;
16334 + if (unlikely(!scheduler_running))
16337 + for_each_leaf_cfs_rq(rq, cfs_rq) {
16338 + struct cfs_bandwidth *cfs_b = &cfs_rq->tg->cfs_bandwidth;
16342 + spin_lock(&cfs_b->cfs_runtime_lock);
16343 + spin_lock(&cfs_rq->cfs_runtime_lock);
16346 + * Either we're all are infinity and nobody needs to borrow,
16347 + * or we're already disabled and this have nothing to do, or
16348 + * we have exactly the right amount of runtime to take out.
16350 + if (cfs_rq->cfs_runtime == RUNTIME_INF ||
16351 + cfs_rq->cfs_runtime == cfs_b->cfs_runtime)
16353 + spin_unlock(&cfs_rq->cfs_runtime_lock);
16356 + * Calculate the difference between what we started out with
16357 + * and what we current have, that's the amount of runtime
16358 + * we lend and now have to reclaim.
16360 + want = cfs_b->cfs_runtime - cfs_rq->cfs_runtime;
16363 + * Greedy reclaim, take back as much as possible.
16365 + for_each_cpu(i, rd->span) {
16366 + struct cfs_rq *iter = sched_cfs_period_cfs_rq(cfs_b, i);
16370 + * Can't reclaim from ourselves or disabled runqueues.
16372 + if (iter == cfs_rq || iter->cfs_runtime == RUNTIME_INF)
16375 + spin_lock(&iter->cfs_runtime_lock);
16377 + diff = min_t(s64, iter->cfs_runtime, want);
16378 + iter->cfs_runtime -= diff;
16381 + iter->cfs_runtime -= want;
16385 + spin_unlock(&iter->cfs_runtime_lock);
16390 + spin_lock(&cfs_rq->cfs_runtime_lock);
16392 + * We cannot be left wanting - that would mean some
16393 + * runtime leaked out of the system.
16398 + * Disable all the borrow logic by pretending we have infinite
16399 + * runtime - in which case borrowing doesn't make sense.
16401 + cfs_rq->cfs_runtime = RUNTIME_INF;
16402 + spin_unlock(&cfs_rq->cfs_runtime_lock);
16403 + spin_unlock(&cfs_b->cfs_runtime_lock);
16407 +static void enable_runtime_cfs(struct rq *rq)
16409 + struct cfs_rq *cfs_rq;
16411 + if (unlikely(!scheduler_running))
16415 + * Reset each runqueue's bandwidth settings
16417 + for_each_leaf_cfs_rq(rq, cfs_rq) {
16418 + struct cfs_bandwidth *cfs_b = &cfs_rq->tg->cfs_bandwidth;
16420 + spin_lock(&cfs_b->cfs_runtime_lock);
16421 + spin_lock(&cfs_rq->cfs_runtime_lock);
16422 + cfs_rq->cfs_runtime = cfs_b->cfs_runtime;
16423 + cfs_rq->cfs_time = 0;
16424 + cfs_rq->cfs_throttled = 0;
16425 + spin_unlock(&cfs_rq->cfs_runtime_lock);
16426 + spin_unlock(&cfs_b->cfs_runtime_lock);
16431 + * Ran out of runtime, check if we can borrow some from others
16432 + * instead of getting throttled right away.
16434 +static void do_cfs_balance_runtime(struct cfs_rq *cfs_rq)
16436 + struct cfs_bandwidth *cfs_b = &cfs_rq->tg->cfs_bandwidth;
16437 + const struct cpumask *span = sched_bw_period_mask();
16441 + weight = cpumask_weight(span);
16442 + spin_lock(&cfs_b->cfs_runtime_lock);
16443 + cfs_period = ktime_to_ns(cfs_b->cfs_period);
16445 + for_each_cpu(i, span) {
16446 + struct cfs_rq *borrow_cfs_rq =
16447 + sched_cfs_period_cfs_rq(cfs_b, i);
16450 + if (borrow_cfs_rq == cfs_rq)
16453 + cfs_rq_runtime_lock(borrow_cfs_rq);
16454 + if (borrow_cfs_rq->cfs_runtime == RUNTIME_INF) {
16455 + cfs_rq_runtime_unlock(borrow_cfs_rq);
16459 + diff = borrow_cfs_rq->cfs_runtime - borrow_cfs_rq->cfs_time;
16461 + diff = div_u64((u64)diff, weight);
16462 + if (cfs_rq->cfs_runtime + diff > cfs_period)
16463 + diff = cfs_period - cfs_rq->cfs_runtime;
16464 + borrow_cfs_rq->cfs_runtime -= diff;
16465 + cfs_rq->cfs_runtime += diff;
16466 + if (cfs_rq->cfs_runtime == cfs_period) {
16467 + cfs_rq_runtime_unlock(borrow_cfs_rq);
16471 + cfs_rq_runtime_unlock(borrow_cfs_rq);
16473 + spin_unlock(&cfs_b->cfs_runtime_lock);
16477 + * Called with rq->runtime_lock held.
16479 +static void cfs_balance_runtime(struct cfs_rq *cfs_rq)
16481 + cfs_rq_runtime_unlock(cfs_rq);
16482 + do_cfs_balance_runtime(cfs_rq);
16483 + cfs_rq_runtime_lock(cfs_rq);
16486 +#else /* !CONFIG_SMP */
16488 +static void cfs_balance_runtime(struct cfs_rq *cfs_rq)
16492 +#endif /* CONFIG_SMP */
16495 + * Check if group entity exceeded its runtime. If so, mark the cfs_rq as
16496 + * throttled mark the current task for reschedling.
16498 +static void sched_cfs_runtime_exceeded(struct sched_entity *se,
16499 + struct task_struct *tsk_curr, unsigned long delta_exec)
16501 + struct cfs_rq *cfs_rq;
16503 + cfs_rq = group_cfs_rq(se);
16505 + if (cfs_rq->cfs_runtime == RUNTIME_INF)
16508 + cfs_rq->cfs_time += delta_exec;
16510 + if (cfs_rq_throttled(cfs_rq))
16513 + if (cfs_rq->cfs_time > cfs_rq->cfs_runtime)
16514 + cfs_balance_runtime(cfs_rq);
16516 + if (cfs_rq->cfs_time > cfs_rq->cfs_runtime) {
16517 + cfs_rq->cfs_throttled = 1;
16518 + update_stats_throttle_start(cfs_rq, se);
16519 + resched_task(tsk_curr);
16523 +static inline void update_curr_group(struct sched_entity *curr,
16524 + unsigned long delta_exec, struct task_struct *tsk_curr)
16526 + sched_cfs_runtime_exceeded(curr, tsk_curr, delta_exec);
16529 +static void enqueue_entity_locked(struct cfs_rq *cfs_rq,
16530 + struct sched_entity *se, int wakeup);
16532 +static void enqueue_throttled_entity(struct rq *rq, struct sched_entity *se)
16534 + for_each_sched_entity(se) {
16535 + struct cfs_rq *gcfs_rq = group_cfs_rq(se);
16537 + if (se->on_rq || cfs_rq_throttled(gcfs_rq) ||
16538 + !gcfs_rq->nr_running)
16540 + enqueue_entity_locked(cfs_rq_of(se), se, 0);
16545 + * Refresh runtimes of all cfs_rqs in this group, i,e.,
16546 + * refresh runtimes of the representative cfs_rq of this
16547 + * tg on all cpus. Enqueue any throttled entity back.
16549 +void do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b)
16552 + const struct cpumask *span = sched_bw_period_mask();
16553 + unsigned long flags;
16555 + for_each_cpu(i, span) {
16556 + struct rq *rq = cpu_rq(i);
16557 + struct cfs_rq *cfs_rq = sched_cfs_period_cfs_rq(cfs_b, i);
16558 + struct sched_entity *se = cfs_rq->tg->se[i];
16560 + spin_lock_irqsave(&rq->lock, flags);
16561 + cfs_rq_runtime_lock(cfs_rq);
16562 + cfs_rq->cfs_time = 0;
16563 + if (cfs_rq_throttled(cfs_rq)) {
16564 + update_rq_clock(rq);
16565 + update_stats_throttle_end(cfs_rq, se);
16566 + cfs_rq->cfs_throttled = 0;
16567 + enqueue_throttled_entity(rq, se);
16569 + cfs_rq_runtime_unlock(cfs_rq);
16570 + spin_unlock_irqrestore(&rq->lock, flags);
16576 +static inline void update_curr_group(struct sched_entity *curr,
16577 + unsigned long delta_exec, struct task_struct *tsk_curr)
16582 +static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
16587 +#endif /* CONFIG_CFS_HARD_LIMITS */
16589 +#else /* CONFIG_FAIR_GROUP_SCHED */
16591 +static inline void update_curr_group(struct sched_entity *curr,
16592 + unsigned long delta_exec, struct task_struct *tsk_curr)
16597 static inline struct task_struct *task_of(struct sched_entity *se)
16599 @@ -251,7 +552,6 @@ find_matching_se(struct sched_entity **s
16601 #endif /* CONFIG_FAIR_GROUP_SCHED */
16604 /**************************************************************
16605 * Scheduling class tree data structure manipulation methods:
16607 @@ -489,14 +789,25 @@ __update_curr(struct cfs_rq *cfs_rq, str
16608 update_min_vruntime(cfs_rq);
16611 -static void update_curr(struct cfs_rq *cfs_rq)
16612 +static void update_curr_task(struct sched_entity *curr,
16613 + unsigned long delta_exec)
16615 + struct task_struct *curtask = task_of(curr);
16617 + trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
16618 + cpuacct_charge(curtask, delta_exec);
16619 + account_group_exec_runtime(curtask, delta_exec);
16622 +static int update_curr_common(struct cfs_rq *cfs_rq, unsigned long *delta)
16624 struct sched_entity *curr = cfs_rq->curr;
16625 - u64 now = rq_of(cfs_rq)->clock;
16626 + struct rq *rq = rq_of(cfs_rq);
16627 + u64 now = rq->clock;
16628 unsigned long delta_exec;
16630 if (unlikely(!curr))
16635 * Get the amount of time the current task was running
16636 @@ -505,20 +816,47 @@ static void update_curr(struct cfs_rq *c
16638 delta_exec = (unsigned long)(now - curr->exec_start);
16643 __update_curr(cfs_rq, curr, delta_exec);
16644 curr->exec_start = now;
16645 + *delta = delta_exec;
16649 - if (entity_is_task(curr)) {
16650 - struct task_struct *curtask = task_of(curr);
16651 +static void update_curr(struct cfs_rq *cfs_rq)
16653 + struct sched_entity *curr = cfs_rq->curr;
16654 + struct rq *rq = rq_of(cfs_rq);
16655 + unsigned long delta_exec;
16657 - trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
16658 - cpuacct_charge(curtask, delta_exec);
16659 - account_group_exec_runtime(curtask, delta_exec);
16660 + if (update_curr_common(cfs_rq, &delta_exec))
16663 + if (entity_is_task(curr))
16664 + update_curr_task(curr, delta_exec);
16666 + cfs_rq_runtime_lock(group_cfs_rq(curr));
16667 + update_curr_group(curr, delta_exec, rq->curr);
16668 + cfs_rq_runtime_unlock(group_cfs_rq(curr));
16672 +static void update_curr_locked(struct cfs_rq *cfs_rq)
16674 + struct sched_entity *curr = cfs_rq->curr;
16675 + struct rq *rq = rq_of(cfs_rq);
16676 + unsigned long delta_exec;
16678 + if (update_curr_common(cfs_rq, &delta_exec))
16681 + if (entity_is_task(curr))
16682 + update_curr_task(curr, delta_exec);
16684 + update_curr_group(curr, delta_exec, rq->curr);
16688 update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
16690 @@ -740,13 +1078,9 @@ place_entity(struct cfs_rq *cfs_rq, stru
16691 se->vruntime = vruntime;
16695 -enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int wakeup)
16696 +static void enqueue_entity_common(struct cfs_rq *cfs_rq,
16697 + struct sched_entity *se, int wakeup)
16700 - * Update run-time statistics of the 'current'.
16702 - update_curr(cfs_rq);
16703 account_entity_enqueue(cfs_rq, se);
16706 @@ -758,6 +1092,29 @@ enqueue_entity(struct cfs_rq *cfs_rq, st
16707 check_spread(cfs_rq, se);
16708 if (se != cfs_rq->curr)
16709 __enqueue_entity(cfs_rq, se);
16711 + if (entity_is_task(se))
16712 + vx_activate_task(task_of(se));
16715 +static void enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
16719 + * Update run-time statistics of the 'current'.
16721 + update_curr(cfs_rq);
16722 + enqueue_entity_common(cfs_rq, se, wakeup);
16725 +static void enqueue_entity_locked(struct cfs_rq *cfs_rq,
16726 + struct sched_entity *se, int wakeup)
16729 + * Update run-time statistics of the 'current'.
16731 + update_curr_locked(cfs_rq);
16732 + enqueue_entity_common(cfs_rq, se, wakeup);
16735 static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
16736 @@ -801,6 +1158,8 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
16738 if (se != cfs_rq->curr)
16739 __dequeue_entity(cfs_rq, se);
16740 + if (entity_is_task(se))
16741 + vx_deactivate_task(task_of(se));
16742 account_entity_dequeue(cfs_rq, se);
16743 update_min_vruntime(cfs_rq);
16745 @@ -897,6 +1256,32 @@ static struct sched_entity *pick_next_en
16750 + * Called from put_prev_entity()
16751 + * If a group entity (@se) is found to be throttled, it will not be put back
16752 + * on @cfs_rq, which is equivalent to dequeing it.
16754 +static int dequeue_throttled_entity(struct cfs_rq *cfs_rq,
16755 + struct sched_entity *se)
16757 + struct cfs_rq *gcfs_rq = group_cfs_rq(se);
16759 + if (entity_is_task(se))
16762 + cfs_rq_runtime_lock(gcfs_rq);
16763 + if (!cfs_rq_throttled(gcfs_rq) && gcfs_rq->nr_running) {
16764 + cfs_rq_runtime_unlock(gcfs_rq);
16768 + __clear_buddies(cfs_rq, se);
16769 + account_entity_dequeue(cfs_rq, se);
16770 + cfs_rq->curr = NULL;
16771 + cfs_rq_runtime_unlock(gcfs_rq);
16775 static void put_prev_entity(struct cfs_rq *cfs_rq, struct sched_entity *prev)
16778 @@ -908,6 +1293,8 @@ static void put_prev_entity(struct cfs_r
16780 check_spread(cfs_rq, prev);
16782 + if (dequeue_throttled_entity(cfs_rq, prev))
16784 update_stats_wait_start(cfs_rq, prev);
16785 /* Put 'current' back into the tree. */
16786 __enqueue_entity(cfs_rq, prev);
16787 @@ -1004,10 +1391,28 @@ static inline void hrtick_update(struct
16791 +static int enqueue_group_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
16794 + struct cfs_rq *gcfs_rq = group_cfs_rq(se);
16797 + cfs_rq_runtime_lock(gcfs_rq);
16798 + if (cfs_rq_throttled(gcfs_rq)) {
16802 + enqueue_entity_locked(cfs_rq, se, wakeup);
16804 + cfs_rq_runtime_unlock(gcfs_rq);
16809 * The enqueue_task method is called before nr_running is
16810 * increased. Here we update the fair scheduling stats and
16811 * then put the task into the rbtree:
16812 + * Don't enqueue a throttled entity further into the hierarchy.
16814 static void enqueue_task_fair(struct rq *rq, struct task_struct *p, int wakeup)
16816 @@ -1017,11 +1422,15 @@ static void enqueue_task_fair(struct rq
16817 for_each_sched_entity(se) {
16821 cfs_rq = cfs_rq_of(se);
16822 - enqueue_entity(cfs_rq, se, wakeup);
16823 + if (entity_is_task(se))
16824 + enqueue_entity(cfs_rq, se, wakeup);
16826 + if (enqueue_group_entity(cfs_rq, se, wakeup))
16834 @@ -1041,6 +1450,17 @@ static void dequeue_task_fair(struct rq
16835 /* Don't dequeue parent if it has other entities besides us */
16836 if (cfs_rq->load.weight)
16840 + * If this cfs_rq is throttled, then it is already
16843 + cfs_rq_runtime_lock(cfs_rq);
16844 + if (cfs_rq_throttled(cfs_rq)) {
16845 + cfs_rq_runtime_unlock(cfs_rq);
16848 + cfs_rq_runtime_unlock(cfs_rq);
16852 @@ -1788,9 +2208,10 @@ load_balance_fair(struct rq *this_rq, in
16853 u64 rem_load, moved_load;
16857 + * empty group or throttled group
16859 - if (!busiest_cfs_rq->task_weight)
16860 + if (!busiest_cfs_rq->task_weight ||
16861 + cfs_rq_throttled(busiest_cfs_rq))
16864 rem_load = (u64)rem_load_move * busiest_weight;
16865 @@ -1839,6 +2260,12 @@ move_one_task_fair(struct rq *this_rq, i
16867 for_each_leaf_cfs_rq(busiest, busy_cfs_rq) {
16869 + * Don't move task from a throttled cfs_rq
16871 + if (cfs_rq_throttled(busy_cfs_rq))
16875 * pass busy_cfs_rq argument into
16876 * load_balance_[start|next]_fair iterators
16878 diff -NurpP --minimal linux-2.6.32.1/kernel/sched_rt.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_rt.c
16879 --- linux-2.6.32.1/kernel/sched_rt.c 2009-12-03 20:02:58.000000000 +0100
16880 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sched_rt.c 2009-12-03 20:04:56.000000000 +0100
16881 @@ -235,18 +235,6 @@ static int rt_se_boosted(struct sched_rt
16882 return p->prio != p->normal_prio;
16886 -static inline const struct cpumask *sched_rt_period_mask(void)
16888 - return cpu_rq(smp_processor_id())->rd->span;
16891 -static inline const struct cpumask *sched_rt_period_mask(void)
16893 - return cpu_online_mask;
16898 struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu)
16900 @@ -296,11 +284,6 @@ static inline int rt_rq_throttled(struct
16901 return rt_rq->rt_throttled;
16904 -static inline const struct cpumask *sched_rt_period_mask(void)
16906 - return cpu_online_mask;
16910 struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu)
16912 @@ -373,7 +356,7 @@ next:
16914 * Ensure this RQ takes back all the runtime it lend to its neighbours.
16916 -static void __disable_runtime(struct rq *rq)
16917 +static void disable_runtime_rt(struct rq *rq)
16919 struct root_domain *rd = rq->rd;
16920 struct rt_rq *rt_rq;
16921 @@ -450,16 +433,7 @@ balanced:
16925 -static void disable_runtime(struct rq *rq)
16927 - unsigned long flags;
16929 - spin_lock_irqsave(&rq->lock, flags);
16930 - __disable_runtime(rq);
16931 - spin_unlock_irqrestore(&rq->lock, flags);
16934 -static void __enable_runtime(struct rq *rq)
16935 +static void enable_runtime_rt(struct rq *rq)
16937 struct rt_rq *rt_rq;
16939 @@ -482,15 +456,6 @@ static void __enable_runtime(struct rq *
16943 -static void enable_runtime(struct rq *rq)
16945 - unsigned long flags;
16947 - spin_lock_irqsave(&rq->lock, flags);
16948 - __enable_runtime(rq);
16949 - spin_unlock_irqrestore(&rq->lock, flags);
16952 static int balance_runtime(struct rt_rq *rt_rq)
16955 @@ -518,7 +483,7 @@ static int do_sched_rt_period_timer(stru
16956 if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF)
16959 - span = sched_rt_period_mask();
16960 + span = sched_bw_period_mask();
16961 for_each_cpu(i, span) {
16963 struct rt_rq *rt_rq = sched_rt_period_rt_rq(rt_b, i);
16964 @@ -1564,7 +1529,7 @@ static void rq_online_rt(struct rq *rq)
16965 if (rq->rt.overloaded)
16966 rt_set_overload(rq);
16968 - __enable_runtime(rq);
16969 + enable_runtime_rt(rq);
16971 cpupri_set(&rq->rd->cpupri, rq->cpu, rq->rt.highest_prio.curr);
16973 @@ -1575,7 +1540,7 @@ static void rq_offline_rt(struct rq *rq)
16974 if (rq->rt.overloaded)
16975 rt_clear_overload(rq);
16977 - __disable_runtime(rq);
16978 + disable_runtime_rt(rq);
16980 cpupri_set(&rq->rd->cpupri, rq->cpu, CPUPRI_INVALID);
16982 diff -NurpP --minimal linux-2.6.32.1/kernel/signal.c linux-2.6.32.1-vs2.3.0.36.27/kernel/signal.c
16983 --- linux-2.6.32.1/kernel/signal.c 2009-12-03 20:02:58.000000000 +0100
16984 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/signal.c 2009-12-03 20:04:56.000000000 +0100
16986 #include <linux/freezer.h>
16987 #include <linux/pid_namespace.h>
16988 #include <linux/nsproxy.h>
16989 +#include <linux/vs_context.h>
16990 +#include <linux/vs_pid.h>
16991 #include <trace/events/sched.h>
16993 #include <asm/param.h>
16994 @@ -598,6 +600,14 @@ static int check_kill_permission(int sig
16995 if (!valid_signal(sig))
16998 + if ((info != SEND_SIG_NOINFO) &&
16999 + (is_si_special(info) || !SI_FROMUSER(info)))
17002 + vxdprintk(VXD_CBIT(misc, 7),
17003 + "check_kill_permission(%d,%p,%p[#%u,%u])",
17004 + sig, info, t, vx_task_xid(t), t->pid);
17006 if (info != SEND_SIG_NOINFO && (is_si_special(info) || SI_FROMKERNEL(info)))
17009 @@ -625,6 +635,20 @@ static int check_kill_permission(int sig
17014 + if (t->pid == 1 && current->xid)
17018 + /* FIXME: we shouldn't return ESRCH ever, to avoid
17019 + loops, maybe ENOENT or EACCES? */
17020 + if (!vx_check(vx_task_xid(t), VS_WATCH_P | VS_IDENT)) {
17021 + vxdprintk(current->xid || VXD_CBIT(misc, 7),
17022 + "signal %d[%p] xid mismatch %p[#%u,%u] xid=#%u",
17023 + sig, info, t, vx_task_xid(t), t->pid, current->xid);
17027 return security_task_kill(t, info, sig, 0);
17030 @@ -1112,7 +1136,7 @@ int kill_pid_info(int sig, struct siginf
17033 p = pid_task(pid, PIDTYPE_PID);
17035 + if (p && vx_check(vx_task_xid(p), VS_IDENT)) {
17036 error = group_send_sig_info(sig, info, p);
17037 if (unlikely(error == -ESRCH))
17039 @@ -1151,7 +1175,7 @@ int kill_pid_info_as_uid(int sig, struct
17041 read_lock(&tasklist_lock);
17042 p = pid_task(pid, PIDTYPE_PID);
17044 + if (!p || !vx_check(vx_task_xid(p), VS_IDENT)) {
17048 @@ -1205,8 +1229,10 @@ static int kill_something_info(int sig,
17049 struct task_struct * p;
17051 for_each_process(p) {
17052 - if (task_pid_vnr(p) > 1 &&
17053 - !same_thread_group(p, current)) {
17054 + if (vx_check(vx_task_xid(p), VS_ADMIN|VS_IDENT) &&
17055 + task_pid_vnr(p) > 1 &&
17056 + !same_thread_group(p, current) &&
17057 + !vx_current_initpid(p->pid)) {
17058 int err = group_send_sig_info(sig, info, p);
17061 @@ -1871,6 +1897,11 @@ relock:
17062 !sig_kernel_only(signr))
17065 + /* virtual init is protected against user signals */
17066 + if ((info->si_code == SI_USER) &&
17067 + vx_current_initpid(current->pid))
17070 if (sig_kernel_stop(signr)) {
17072 * The default action is to stop all threads in
17073 diff -NurpP --minimal linux-2.6.32.1/kernel/softirq.c linux-2.6.32.1-vs2.3.0.36.27/kernel/softirq.c
17074 --- linux-2.6.32.1/kernel/softirq.c 2009-12-03 20:02:58.000000000 +0100
17075 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/softirq.c 2009-12-03 20:04:56.000000000 +0100
17077 #include <linux/ftrace.h>
17078 #include <linux/smp.h>
17079 #include <linux/tick.h>
17080 +#include <linux/vs_context.h>
17082 #define CREATE_TRACE_POINTS
17083 #include <trace/events/irq.h>
17084 diff -NurpP --minimal linux-2.6.32.1/kernel/sys.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sys.c
17085 --- linux-2.6.32.1/kernel/sys.c 2009-12-03 20:02:58.000000000 +0100
17086 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sys.c 2009-12-03 20:04:56.000000000 +0100
17088 #include <linux/syscalls.h>
17089 #include <linux/kprobes.h>
17090 #include <linux/user_namespace.h>
17091 +#include <linux/vs_pid.h>
17093 #include <asm/uaccess.h>
17094 #include <asm/io.h>
17095 @@ -130,7 +131,10 @@ static int set_one_prio(struct task_stru
17098 if (niceval < task_nice(p) && !can_nice(p, niceval)) {
17100 + if (vx_flags(VXF_IGNEG_NICE, 0))
17106 no_nice = security_task_setnice(p, niceval);
17107 @@ -179,6 +183,8 @@ SYSCALL_DEFINE3(setpriority, int, which,
17109 pgrp = task_pgrp(current);
17110 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
17111 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
17113 error = set_one_prio(p, niceval, error);
17114 } while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
17116 @@ -240,6 +246,8 @@ SYSCALL_DEFINE2(getpriority, int, which,
17118 pgrp = task_pgrp(current);
17119 do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
17120 + if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
17122 niceval = 20 - task_nice(p);
17123 if (niceval > retval)
17125 @@ -349,6 +357,9 @@ void kernel_power_off(void)
17126 machine_power_off();
17128 EXPORT_SYMBOL_GPL(kernel_power_off);
17130 +long vs_reboot(unsigned int, void __user *);
17133 * Reboot system call: for obvious reasons only root may call it,
17134 * and even root needs to set up some magic numbers in the registers
17135 @@ -381,6 +392,9 @@ SYSCALL_DEFINE4(reboot, int, magic1, int
17136 if ((cmd == LINUX_REBOOT_CMD_POWER_OFF) && !pm_power_off)
17137 cmd = LINUX_REBOOT_CMD_HALT;
17139 + if (!vx_check(0, VS_ADMIN|VS_WATCH))
17140 + return vs_reboot(cmd, arg);
17144 case LINUX_REBOOT_CMD_RESTART:
17145 @@ -1133,7 +1147,7 @@ SYSCALL_DEFINE2(sethostname, char __user
17147 char tmp[__NEW_UTS_LEN];
17149 - if (!capable(CAP_SYS_ADMIN))
17150 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
17152 if (len < 0 || len > __NEW_UTS_LEN)
17154 @@ -1182,7 +1196,7 @@ SYSCALL_DEFINE2(setdomainname, char __us
17156 char tmp[__NEW_UTS_LEN];
17158 - if (!capable(CAP_SYS_ADMIN))
17159 + if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
17161 if (len < 0 || len > __NEW_UTS_LEN)
17163 @@ -1251,7 +1265,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int,
17165 old_rlim = current->signal->rlim + resource;
17166 if ((new_rlim.rlim_max > old_rlim->rlim_max) &&
17167 - !capable(CAP_SYS_RESOURCE))
17168 + !vx_capable(CAP_SYS_RESOURCE, VXC_SET_RLIMIT))
17170 if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open)
17172 diff -NurpP --minimal linux-2.6.32.1/kernel/sysctl.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sysctl.c
17173 --- linux-2.6.32.1/kernel/sysctl.c 2009-12-03 20:02:58.000000000 +0100
17174 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sysctl.c 2009-12-03 20:04:56.000000000 +0100
17175 @@ -124,6 +124,7 @@ static int ngroups_max = NGROUPS_MAX;
17176 extern char modprobe_path[];
17177 extern int modules_disabled;
17179 +extern char vshelper_path[];
17180 #ifdef CONFIG_CHR_DEV_SG
17181 extern int sg_big_buff;
17183 @@ -593,6 +594,15 @@ static struct ctl_table kern_table[] = {
17184 .strategy = &sysctl_string,
17188 + .ctl_name = KERN_VSHELPER,
17189 + .procname = "vshelper",
17190 + .data = &vshelper_path,
17193 + .proc_handler = &proc_dostring,
17194 + .strategy = &sysctl_string,
17196 #ifdef CONFIG_CHR_DEV_SG
17198 .ctl_name = KERN_SG_BIG_BUFF,
17199 diff -NurpP --minimal linux-2.6.32.1/kernel/sysctl_check.c linux-2.6.32.1-vs2.3.0.36.27/kernel/sysctl_check.c
17200 --- linux-2.6.32.1/kernel/sysctl_check.c 2009-12-03 20:02:58.000000000 +0100
17201 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/sysctl_check.c 2009-12-03 20:04:56.000000000 +0100
17202 @@ -39,6 +39,7 @@ static const struct trans_ctl_table tran
17204 { KERN_PANIC, "panic" },
17205 { KERN_REALROOTDEV, "real-root-dev" },
17206 + { KERN_VSHELPER, "vshelper", },
17208 { KERN_SPARC_REBOOT, "reboot-cmd" },
17209 { KERN_CTLALTDEL, "ctrl-alt-del" },
17210 @@ -1217,6 +1218,22 @@ static const struct trans_ctl_table tran
17214 +static struct trans_ctl_table trans_vserver_table[] = {
17215 + { 1, "debug_switch" },
17216 + { 2, "debug_xid" },
17217 + { 3, "debug_nid" },
17218 + { 4, "debug_tag" },
17219 + { 5, "debug_net" },
17220 + { 6, "debug_limit" },
17221 + { 7, "debug_cres" },
17222 + { 8, "debug_dlim" },
17223 + { 9, "debug_quota" },
17224 + { 10, "debug_cvirt" },
17225 + { 11, "debug_space" },
17226 + { 12, "debug_misc" },
17230 static const struct trans_ctl_table trans_root_table[] = {
17231 { CTL_KERN, "kernel", trans_kern_table },
17232 { CTL_VM, "vm", trans_vm_table },
17233 @@ -1233,6 +1250,7 @@ static const struct trans_ctl_table tran
17234 { CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
17235 { CTL_PM, "pm", trans_pm_table },
17236 { CTL_FRV, "frv", trans_frv_table },
17237 + { CTL_VSERVER, "vserver", trans_vserver_table },
17241 diff -NurpP --minimal linux-2.6.32.1/kernel/time.c linux-2.6.32.1-vs2.3.0.36.27/kernel/time.c
17242 --- linux-2.6.32.1/kernel/time.c 2009-12-03 20:02:58.000000000 +0100
17243 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/time.c 2009-12-03 20:04:56.000000000 +0100
17244 @@ -63,6 +63,7 @@ EXPORT_SYMBOL(sys_tz);
17245 SYSCALL_DEFINE1(time, time_t __user *, tloc)
17247 time_t i = get_seconds();
17248 +/* FIXME: do_gettimeofday(&tv) -> vx_gettimeofday(&tv) */
17251 if (put_user(i,tloc))
17252 @@ -93,7 +94,7 @@ SYSCALL_DEFINE1(stime, time_t __user *,
17256 - do_settimeofday(&tv);
17257 + vx_settimeofday(&tv);
17261 @@ -104,7 +105,7 @@ SYSCALL_DEFINE2(gettimeofday, struct tim
17263 if (likely(tv != NULL)) {
17264 struct timeval ktv;
17265 - do_gettimeofday(&ktv);
17266 + vx_gettimeofday(&ktv);
17267 if (copy_to_user(tv, &ktv, sizeof(ktv)))
17270 @@ -179,7 +180,7 @@ int do_sys_settimeofday(struct timespec
17271 /* SMP safe, again the code in arch/foo/time.c should
17272 * globally block out interrupts when it runs.
17274 - return do_settimeofday(tv);
17275 + return vx_settimeofday(tv);
17279 @@ -311,7 +312,7 @@ void getnstimeofday(struct timespec *tv)
17283 - do_gettimeofday(&x);
17284 + vx_gettimeofday(&x);
17285 tv->tv_sec = x.tv_sec;
17286 tv->tv_nsec = x.tv_usec * NSEC_PER_USEC;
17288 diff -NurpP --minimal linux-2.6.32.1/kernel/timer.c linux-2.6.32.1-vs2.3.0.36.27/kernel/timer.c
17289 --- linux-2.6.32.1/kernel/timer.c 2009-12-03 20:02:58.000000000 +0100
17290 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/timer.c 2009-12-03 20:04:56.000000000 +0100
17292 #include <linux/kallsyms.h>
17293 #include <linux/perf_event.h>
17294 #include <linux/sched.h>
17295 +#include <linux/vs_base.h>
17296 +#include <linux/vs_cvirt.h>
17297 +#include <linux/vs_pid.h>
17298 +#include <linux/vserver/sched.h>
17300 #include <asm/uaccess.h>
17301 #include <asm/unistd.h>
17302 @@ -1255,12 +1259,6 @@ SYSCALL_DEFINE1(alarm, unsigned int, sec
17309 - * The Alpha uses getxpid, getxuid, and getxgid instead. Maybe this
17310 - * should be moved into arch/i386 instead?
17314 * sys_getpid - return the thread group id of the current process
17315 @@ -1289,10 +1287,23 @@ SYSCALL_DEFINE0(getppid)
17317 pid = task_tgid_vnr(current->real_parent);
17319 + return vx_map_pid(pid);
17326 + * The Alpha uses getxpid, getxuid, and getxgid instead.
17329 +asmlinkage long do_getxpid(long *ppid)
17331 + *ppid = sys_getppid();
17332 + return sys_getpid();
17335 +#else /* _alpha_ */
17337 SYSCALL_DEFINE0(getuid)
17339 /* Only we change this so SMP safe */
17340 diff -NurpP --minimal linux-2.6.32.1/kernel/user.c linux-2.6.32.1-vs2.3.0.36.27/kernel/user.c
17341 --- linux-2.6.32.1/kernel/user.c 2009-12-03 20:02:58.000000000 +0100
17342 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/user.c 2009-12-03 20:04:56.000000000 +0100
17343 @@ -251,10 +251,10 @@ static struct kobj_type uids_ktype = {
17345 * See Documentation/scheduler/sched-design-CFS.txt for ramifications.
17347 -static int uids_user_create(struct user_struct *up)
17348 +static int uids_user_create(struct user_namespace *ns, struct user_struct *up)
17350 struct kobject *kobj = &up->kobj;
17354 memset(kobj, 0, sizeof(struct kobject));
17355 if (up->user_ns != &init_user_ns)
17356 @@ -282,7 +282,7 @@ int __init uids_sysfs_init(void)
17360 - return uids_user_create(&root_user);
17361 + return uids_user_create(NULL, &root_user);
17364 /* delayed work function to remove sysfs directory for a user and free up
17365 @@ -353,7 +353,8 @@ static struct user_struct *uid_hash_find
17368 int uids_sysfs_init(void) { return 0; }
17369 -static inline int uids_user_create(struct user_struct *up) { return 0; }
17370 +static inline int uids_user_create(struct user_namespace *ns,
17371 + struct user_struct *up) { return 0; }
17372 static inline void uids_mutex_lock(void) { }
17373 static inline void uids_mutex_unlock(void) { }
17375 @@ -450,7 +451,7 @@ struct user_struct *alloc_uid(struct use
17377 new->user_ns = get_user_ns(ns);
17379 - if (uids_user_create(new))
17380 + if (uids_user_create(ns, new))
17381 goto out_destoy_sched;
17384 diff -NurpP --minimal linux-2.6.32.1/kernel/user_namespace.c linux-2.6.32.1-vs2.3.0.36.27/kernel/user_namespace.c
17385 --- linux-2.6.32.1/kernel/user_namespace.c 2009-03-24 14:22:45.000000000 +0100
17386 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/user_namespace.c 2009-12-03 20:04:56.000000000 +0100
17388 #include <linux/slab.h>
17389 #include <linux/user_namespace.h>
17390 #include <linux/cred.h>
17391 +#include <linux/vserver/global.h>
17394 * Create a new user namespace, deriving the creator from the user in the
17395 @@ -30,6 +31,7 @@ int create_user_ns(struct cred *new)
17398 kref_init(&ns->kref);
17399 + atomic_inc(&vs_global_user_ns);
17401 for (n = 0; n < UIDHASH_SZ; ++n)
17402 INIT_HLIST_HEAD(ns->uidhash_table + n);
17403 @@ -78,6 +80,8 @@ void free_user_ns(struct kref *kref)
17404 struct user_namespace *ns =
17405 container_of(kref, struct user_namespace, kref);
17407 + /* FIXME: maybe move into destroyer? */
17408 + atomic_dec(&vs_global_user_ns);
17409 INIT_WORK(&ns->destroyer, free_user_ns_work);
17410 schedule_work(&ns->destroyer);
17412 diff -NurpP --minimal linux-2.6.32.1/kernel/utsname.c linux-2.6.32.1-vs2.3.0.36.27/kernel/utsname.c
17413 --- linux-2.6.32.1/kernel/utsname.c 2009-09-10 15:26:28.000000000 +0200
17414 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/utsname.c 2009-12-03 20:04:56.000000000 +0100
17415 @@ -14,14 +14,17 @@
17416 #include <linux/utsname.h>
17417 #include <linux/err.h>
17418 #include <linux/slab.h>
17419 +#include <linux/vserver/global.h>
17421 static struct uts_namespace *create_uts_ns(void)
17423 struct uts_namespace *uts_ns;
17425 uts_ns = kmalloc(sizeof(struct uts_namespace), GFP_KERNEL);
17428 kref_init(&uts_ns->kref);
17429 + atomic_inc(&vs_global_uts_ns);
17434 @@ -71,5 +74,6 @@ void free_uts_ns(struct kref *kref)
17435 struct uts_namespace *ns;
17437 ns = container_of(kref, struct uts_namespace, kref);
17438 + atomic_dec(&vs_global_uts_ns);
17441 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cacct.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct.c
17442 --- linux-2.6.32.1/kernel/vserver/cacct.c 1970-01-01 01:00:00.000000000 +0100
17443 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct.c 2009-12-03 20:04:56.000000000 +0100
17446 + * linux/kernel/vserver/cacct.c
17448 + * Virtual Server: Context Accounting
17450 + * Copyright (C) 2006-2007 Herbert Pötzl
17452 + * V0.01 added accounting stats
17456 +#include <linux/types.h>
17457 +#include <linux/vs_context.h>
17458 +#include <linux/vserver/cacct_cmd.h>
17459 +#include <linux/vserver/cacct_int.h>
17461 +#include <asm/errno.h>
17462 +#include <asm/uaccess.h>
17465 +int vc_sock_stat(struct vx_info *vxi, void __user *data)
17467 + struct vcmd_sock_stat_v0 vc_data;
17470 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17473 + field = vc_data.field;
17474 + if ((field < 0) || (field >= VXA_SOCK_SIZE))
17477 + for (j = 0; j < 3; j++) {
17478 + vc_data.count[j] = vx_sock_count(&vxi->cacct, field, j);
17479 + vc_data.total[j] = vx_sock_total(&vxi->cacct, field, j);
17482 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17487 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cacct_init.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct_init.h
17488 --- linux-2.6.32.1/kernel/vserver/cacct_init.h 1970-01-01 01:00:00.000000000 +0100
17489 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct_init.h 2009-12-03 20:04:56.000000000 +0100
17493 +static inline void vx_info_init_cacct(struct _vx_cacct *cacct)
17498 + for (i = 0; i < VXA_SOCK_SIZE; i++) {
17499 + for (j = 0; j < 3; j++) {
17500 + atomic_long_set(&cacct->sock[i][j].count, 0);
17501 + atomic_long_set(&cacct->sock[i][j].total, 0);
17504 + for (i = 0; i < 8; i++)
17505 + atomic_set(&cacct->slab[i], 0);
17506 + for (i = 0; i < 5; i++)
17507 + for (j = 0; j < 4; j++)
17508 + atomic_set(&cacct->page[i][j], 0);
17511 +static inline void vx_info_exit_cacct(struct _vx_cacct *cacct)
17516 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cacct_proc.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct_proc.h
17517 --- linux-2.6.32.1/kernel/vserver/cacct_proc.h 1970-01-01 01:00:00.000000000 +0100
17518 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cacct_proc.h 2009-12-03 20:04:56.000000000 +0100
17520 +#ifndef _VX_CACCT_PROC_H
17521 +#define _VX_CACCT_PROC_H
17523 +#include <linux/vserver/cacct_int.h>
17526 +#define VX_SOCKA_TOP \
17527 + "Type\t recv #/bytes\t\t send #/bytes\t\t fail #/bytes\n"
17529 +static inline int vx_info_proc_cacct(struct _vx_cacct *cacct, char *buffer)
17531 + int i, j, length = 0;
17532 + static char *type[VXA_SOCK_SIZE] = {
17533 + "UNSPEC", "UNIX", "INET", "INET6", "PACKET", "OTHER"
17536 + length += sprintf(buffer + length, VX_SOCKA_TOP);
17537 + for (i = 0; i < VXA_SOCK_SIZE; i++) {
17538 + length += sprintf(buffer + length, "%s:", type[i]);
17539 + for (j = 0; j < 3; j++) {
17540 + length += sprintf(buffer + length,
17541 + "\t%10lu/%-10lu",
17542 + vx_sock_count(cacct, i, j),
17543 + vx_sock_total(cacct, i, j));
17545 + buffer[length++] = '\n';
17548 + length += sprintf(buffer + length, "\n");
17549 + length += sprintf(buffer + length,
17550 + "slab:\t %8u %8u %8u %8u\n",
17551 + atomic_read(&cacct->slab[1]),
17552 + atomic_read(&cacct->slab[4]),
17553 + atomic_read(&cacct->slab[0]),
17554 + atomic_read(&cacct->slab[2]));
17556 + length += sprintf(buffer + length, "\n");
17557 + for (i = 0; i < 5; i++) {
17558 + length += sprintf(buffer + length,
17559 + "page[%d]: %8u %8u %8u %8u\t %8u %8u %8u %8u\n", i,
17560 + atomic_read(&cacct->page[i][0]),
17561 + atomic_read(&cacct->page[i][1]),
17562 + atomic_read(&cacct->page[i][2]),
17563 + atomic_read(&cacct->page[i][3]),
17564 + atomic_read(&cacct->page[i][4]),
17565 + atomic_read(&cacct->page[i][5]),
17566 + atomic_read(&cacct->page[i][6]),
17567 + atomic_read(&cacct->page[i][7]));
17572 +#endif /* _VX_CACCT_PROC_H */
17573 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/context.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/context.c
17574 --- linux-2.6.32.1/kernel/vserver/context.c 1970-01-01 01:00:00.000000000 +0100
17575 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/context.c 2009-12-03 20:04:56.000000000 +0100
17578 + * linux/kernel/vserver/context.c
17580 + * Virtual Server: Context Support
17582 + * Copyright (C) 2003-2007 Herbert Pötzl
17584 + * V0.01 context helper
17585 + * V0.02 vx_ctx_kill syscall command
17586 + * V0.03 replaced context_info calls
17587 + * V0.04 redesign of struct (de)alloc
17588 + * V0.05 rlimit basic implementation
17589 + * V0.06 task_xid and info commands
17590 + * V0.07 context flags and caps
17591 + * V0.08 switch to RCU based hash
17592 + * V0.09 revert to non RCU for now
17593 + * V0.10 and back to working RCU hash
17594 + * V0.11 and back to locking again
17595 + * V0.12 referenced context store
17596 + * V0.13 separate per cpu data
17597 + * V0.14 changed vcmds to vxi arg
17598 + * V0.15 added context stat
17599 + * V0.16 have __create claim() the vxi
17600 + * V0.17 removed older and legacy stuff
17604 +#include <linux/slab.h>
17605 +#include <linux/types.h>
17606 +#include <linux/security.h>
17607 +#include <linux/pid_namespace.h>
17609 +#include <linux/vserver/context.h>
17610 +#include <linux/vserver/network.h>
17611 +#include <linux/vserver/debug.h>
17612 +#include <linux/vserver/limit.h>
17613 +#include <linux/vserver/limit_int.h>
17614 +#include <linux/vserver/space.h>
17615 +#include <linux/init_task.h>
17616 +#include <linux/fs_struct.h>
17618 +#include <linux/vs_context.h>
17619 +#include <linux/vs_limit.h>
17620 +#include <linux/vs_pid.h>
17621 +#include <linux/vserver/context_cmd.h>
17623 +#include "cvirt_init.h"
17624 +#include "cacct_init.h"
17625 +#include "limit_init.h"
17626 +#include "sched_init.h"
17629 +atomic_t vx_global_ctotal = ATOMIC_INIT(0);
17630 +atomic_t vx_global_cactive = ATOMIC_INIT(0);
17633 +/* now inactive context structures */
17635 +static struct hlist_head vx_info_inactive = HLIST_HEAD_INIT;
17637 +static spinlock_t vx_info_inactive_lock = SPIN_LOCK_UNLOCKED;
17640 +/* __alloc_vx_info()
17642 + * allocate an initialized vx_info struct
17643 + * doesn't make it visible (hash) */
17645 +static struct vx_info *__alloc_vx_info(xid_t xid)
17647 + struct vx_info *new = NULL;
17650 + vxdprintk(VXD_CBIT(xid, 0), "alloc_vx_info(%d)*", xid);
17652 + /* would this benefit from a slab cache? */
17653 + new = kmalloc(sizeof(struct vx_info), GFP_KERNEL);
17657 + memset(new, 0, sizeof(struct vx_info));
17659 + new->ptr_pc = alloc_percpu(struct _vx_info_pc);
17660 + if (!new->ptr_pc)
17663 + new->vx_id = xid;
17664 + INIT_HLIST_NODE(&new->vx_hlist);
17665 + atomic_set(&new->vx_usecnt, 0);
17666 + atomic_set(&new->vx_tasks, 0);
17667 + new->vx_parent = NULL;
17668 + new->vx_state = 0;
17669 + init_waitqueue_head(&new->vx_wait);
17671 + /* prepare reaper */
17672 + get_task_struct(init_pid_ns.child_reaper);
17673 + new->vx_reaper = init_pid_ns.child_reaper;
17674 + new->vx_badness_bias = 0;
17676 + /* rest of init goes here */
17677 + vx_info_init_limit(&new->limit);
17678 + vx_info_init_sched(&new->sched);
17679 + vx_info_init_cvirt(&new->cvirt);
17680 + vx_info_init_cacct(&new->cacct);
17682 + /* per cpu data structures */
17683 + for_each_possible_cpu(cpu) {
17684 + vx_info_init_sched_pc(
17685 + &vx_per_cpu(new, sched_pc, cpu), cpu);
17686 + vx_info_init_cvirt_pc(
17687 + &vx_per_cpu(new, cvirt_pc, cpu), cpu);
17690 + new->vx_flags = VXF_INIT_SET;
17691 + cap_set_init_eff(new->vx_bcaps);
17692 + new->vx_ccaps = 0;
17693 + new->vx_umask = 0;
17695 + new->reboot_cmd = 0;
17696 + new->exit_code = 0;
17698 + // preconfig fs entries
17699 + for (index = 0; index < VX_SPACES; index++) {
17700 + write_lock(&init_fs.lock);
17702 + write_unlock(&init_fs.lock);
17703 + new->vx_fs[index] = &init_fs;
17706 + vxdprintk(VXD_CBIT(xid, 0),
17707 + "alloc_vx_info(%d) = %p", xid, new);
17708 + vxh_alloc_vx_info(new);
17709 + atomic_inc(&vx_global_ctotal);
17718 +/* __dealloc_vx_info()
17720 + * final disposal of vx_info */
17722 +static void __dealloc_vx_info(struct vx_info *vxi)
17724 + struct vx_info_save vxis;
17727 + vxdprintk(VXD_CBIT(xid, 0),
17728 + "dealloc_vx_info(%p)", vxi);
17729 + vxh_dealloc_vx_info(vxi);
17731 +#ifdef CONFIG_VSERVER_WARN
17732 + enter_vx_info(vxi, &vxis);
17733 + vx_info_exit_limit(&vxi->limit);
17734 + vx_info_exit_sched(&vxi->sched);
17735 + vx_info_exit_cvirt(&vxi->cvirt);
17736 + vx_info_exit_cacct(&vxi->cacct);
17738 + for_each_possible_cpu(cpu) {
17739 + vx_info_exit_sched_pc(
17740 + &vx_per_cpu(vxi, sched_pc, cpu), cpu);
17741 + vx_info_exit_cvirt_pc(
17742 + &vx_per_cpu(vxi, cvirt_pc, cpu), cpu);
17744 + leave_vx_info(&vxis);
17748 + vxi->vx_state |= VXS_RELEASED;
17751 + free_percpu(vxi->ptr_pc);
17754 + atomic_dec(&vx_global_ctotal);
17757 +static void __shutdown_vx_info(struct vx_info *vxi)
17759 + struct nsproxy *nsproxy;
17760 + struct fs_struct *fs;
17765 + vxi->vx_state |= VXS_SHUTDOWN;
17766 + vs_state_change(vxi, VSC_SHUTDOWN);
17768 + for (index = 0; index < VX_SPACES; index++) {
17769 + nsproxy = xchg(&vxi->vx_nsproxy[index], NULL);
17771 + put_nsproxy(nsproxy);
17773 + fs = xchg(&vxi->vx_fs[index], NULL);
17774 + write_lock(&fs->lock);
17775 + kill = !--fs->users;
17776 + write_unlock(&fs->lock);
17778 + free_fs_struct(fs);
17782 +/* exported stuff */
17784 +void free_vx_info(struct vx_info *vxi)
17786 + unsigned long flags;
17789 + /* check for reference counts first */
17790 + BUG_ON(atomic_read(&vxi->vx_usecnt));
17791 + BUG_ON(atomic_read(&vxi->vx_tasks));
17793 + /* context must not be hashed */
17794 + BUG_ON(vx_info_state(vxi, VXS_HASHED));
17796 + /* context shutdown is mandatory */
17797 + BUG_ON(!vx_info_state(vxi, VXS_SHUTDOWN));
17799 + /* nsproxy and fs check */
17800 + for (index = 0; index < VX_SPACES; index++) {
17801 + BUG_ON(vxi->vx_nsproxy[index]);
17802 + BUG_ON(vxi->vx_fs[index]);
17805 + spin_lock_irqsave(&vx_info_inactive_lock, flags);
17806 + hlist_del(&vxi->vx_hlist);
17807 + spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17809 + __dealloc_vx_info(vxi);
17813 +/* hash table for vx_info hash */
17815 +#define VX_HASH_SIZE 13
17817 +static struct hlist_head vx_info_hash[VX_HASH_SIZE] =
17818 + { [0 ... VX_HASH_SIZE-1] = HLIST_HEAD_INIT };
17820 +static spinlock_t vx_info_hash_lock = SPIN_LOCK_UNLOCKED;
17823 +static inline unsigned int __hashval(xid_t xid)
17825 + return (xid % VX_HASH_SIZE);
17830 +/* __hash_vx_info()
17832 + * add the vxi to the global hash table
17833 + * requires the hash_lock to be held */
17835 +static inline void __hash_vx_info(struct vx_info *vxi)
17837 + struct hlist_head *head;
17839 + vxd_assert_lock(&vx_info_hash_lock);
17840 + vxdprintk(VXD_CBIT(xid, 4),
17841 + "__hash_vx_info: %p[#%d]", vxi, vxi->vx_id);
17842 + vxh_hash_vx_info(vxi);
17844 + /* context must not be hashed */
17845 + BUG_ON(vx_info_state(vxi, VXS_HASHED));
17847 + vxi->vx_state |= VXS_HASHED;
17848 + head = &vx_info_hash[__hashval(vxi->vx_id)];
17849 + hlist_add_head(&vxi->vx_hlist, head);
17850 + atomic_inc(&vx_global_cactive);
17853 +/* __unhash_vx_info()
17855 + * remove the vxi from the global hash table
17856 + * requires the hash_lock to be held */
17858 +static inline void __unhash_vx_info(struct vx_info *vxi)
17860 + unsigned long flags;
17862 + vxd_assert_lock(&vx_info_hash_lock);
17863 + vxdprintk(VXD_CBIT(xid, 4),
17864 + "__unhash_vx_info: %p[#%d.%d.%d]", vxi, vxi->vx_id,
17865 + atomic_read(&vxi->vx_usecnt), atomic_read(&vxi->vx_tasks));
17866 + vxh_unhash_vx_info(vxi);
17868 + /* context must be hashed */
17869 + BUG_ON(!vx_info_state(vxi, VXS_HASHED));
17870 + /* but without tasks */
17871 + BUG_ON(atomic_read(&vxi->vx_tasks));
17873 + vxi->vx_state &= ~VXS_HASHED;
17874 + hlist_del_init(&vxi->vx_hlist);
17875 + spin_lock_irqsave(&vx_info_inactive_lock, flags);
17876 + hlist_add_head(&vxi->vx_hlist, &vx_info_inactive);
17877 + spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17878 + atomic_dec(&vx_global_cactive);
17882 +/* __lookup_vx_info()
17884 + * requires the hash_lock to be held
17885 + * doesn't increment the vx_refcnt */
17887 +static inline struct vx_info *__lookup_vx_info(xid_t xid)
17889 + struct hlist_head *head = &vx_info_hash[__hashval(xid)];
17890 + struct hlist_node *pos;
17891 + struct vx_info *vxi;
17893 + vxd_assert_lock(&vx_info_hash_lock);
17894 + hlist_for_each(pos, head) {
17895 + vxi = hlist_entry(pos, struct vx_info, vx_hlist);
17897 + if (vxi->vx_id == xid)
17902 + vxdprintk(VXD_CBIT(xid, 0),
17903 + "__lookup_vx_info(#%u): %p[#%u]",
17904 + xid, vxi, vxi ? vxi->vx_id : 0);
17905 + vxh_lookup_vx_info(vxi, xid);
17910 +/* __create_vx_info()
17912 + * create the requested context
17913 + * get(), claim() and hash it */
17915 +static struct vx_info *__create_vx_info(int id)
17917 + struct vx_info *new, *vxi = NULL;
17919 + vxdprintk(VXD_CBIT(xid, 1), "create_vx_info(%d)*", id);
17921 + if (!(new = __alloc_vx_info(id)))
17922 + return ERR_PTR(-ENOMEM);
17924 + /* required to make dynamic xids unique */
17925 + spin_lock(&vx_info_hash_lock);
17927 + /* static context requested */
17928 + if ((vxi = __lookup_vx_info(id))) {
17929 + vxdprintk(VXD_CBIT(xid, 0),
17930 + "create_vx_info(%d) = %p (already there)", id, vxi);
17931 + if (vx_info_flags(vxi, VXF_STATE_SETUP, 0))
17932 + vxi = ERR_PTR(-EBUSY);
17934 + vxi = ERR_PTR(-EEXIST);
17937 + /* new context */
17938 + vxdprintk(VXD_CBIT(xid, 0),
17939 + "create_vx_info(%d) = %p (new)", id, new);
17940 + claim_vx_info(new, NULL);
17941 + __hash_vx_info(get_vx_info(new));
17942 + vxi = new, new = NULL;
17945 + spin_unlock(&vx_info_hash_lock);
17946 + vxh_create_vx_info(IS_ERR(vxi) ? NULL : vxi, id);
17948 + __dealloc_vx_info(new);
17953 +/* exported stuff */
17956 +void unhash_vx_info(struct vx_info *vxi)
17958 + __shutdown_vx_info(vxi);
17959 + spin_lock(&vx_info_hash_lock);
17960 + __unhash_vx_info(vxi);
17961 + spin_unlock(&vx_info_hash_lock);
17962 + __wakeup_vx_info(vxi);
17966 +/* lookup_vx_info()
17968 + * search for a vx_info and get() it
17969 + * negative id means current */
17971 +struct vx_info *lookup_vx_info(int id)
17973 + struct vx_info *vxi = NULL;
17976 + vxi = get_vx_info(current_vx_info());
17977 + } else if (id > 1) {
17978 + spin_lock(&vx_info_hash_lock);
17979 + vxi = get_vx_info(__lookup_vx_info(id));
17980 + spin_unlock(&vx_info_hash_lock);
17985 +/* xid_is_hashed()
17987 + * verify that xid is still hashed */
17989 +int xid_is_hashed(xid_t xid)
17993 + spin_lock(&vx_info_hash_lock);
17994 + hashed = (__lookup_vx_info(xid) != NULL);
17995 + spin_unlock(&vx_info_hash_lock);
17999 +#ifdef CONFIG_PROC_FS
18003 + * get a subset of hashed xids for proc
18004 + * assumes size is at least one */
18006 +int get_xid_list(int index, unsigned int *xids, int size)
18008 + int hindex, nr_xids = 0;
18010 + /* only show current and children */
18011 + if (!vx_check(0, VS_ADMIN | VS_WATCH)) {
18014 + xids[nr_xids] = vx_current_xid();
18018 + for (hindex = 0; hindex < VX_HASH_SIZE; hindex++) {
18019 + struct hlist_head *head = &vx_info_hash[hindex];
18020 + struct hlist_node *pos;
18022 + spin_lock(&vx_info_hash_lock);
18023 + hlist_for_each(pos, head) {
18024 + struct vx_info *vxi;
18029 + vxi = hlist_entry(pos, struct vx_info, vx_hlist);
18030 + xids[nr_xids] = vxi->vx_id;
18031 + if (++nr_xids >= size) {
18032 + spin_unlock(&vx_info_hash_lock);
18036 + /* keep the lock time short */
18037 + spin_unlock(&vx_info_hash_lock);
18044 +#ifdef CONFIG_VSERVER_DEBUG
18046 +void dump_vx_info_inactive(int level)
18048 + struct hlist_node *entry, *next;
18050 + hlist_for_each_safe(entry, next, &vx_info_inactive) {
18051 + struct vx_info *vxi =
18052 + list_entry(entry, struct vx_info, vx_hlist);
18054 + dump_vx_info(vxi, level);
18061 +int vx_migrate_user(struct task_struct *p, struct vx_info *vxi)
18063 + struct user_struct *new_user, *old_user;
18068 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
18071 + new_user = alloc_uid(vxi->vx_id, p->uid);
18075 + old_user = p->user;
18076 + if (new_user != old_user) {
18077 + atomic_inc(&new_user->processes);
18078 + atomic_dec(&old_user->processes);
18079 + p->user = new_user;
18081 + free_uid(old_user);
18087 +void vx_mask_cap_bset(struct vx_info *vxi, struct task_struct *p)
18089 + // p->cap_effective &= vxi->vx_cap_bset;
18090 + p->cap_effective =
18091 + cap_intersect(p->cap_effective, vxi->cap_bset);
18092 + // p->cap_inheritable &= vxi->vx_cap_bset;
18093 + p->cap_inheritable =
18094 + cap_intersect(p->cap_inheritable, vxi->cap_bset);
18095 + // p->cap_permitted &= vxi->vx_cap_bset;
18096 + p->cap_permitted =
18097 + cap_intersect(p->cap_permitted, vxi->cap_bset);
18102 +#include <linux/file.h>
18103 +#include <linux/fdtable.h>
18105 +static int vx_openfd_task(struct task_struct *tsk)
18107 + struct files_struct *files = tsk->files;
18108 + struct fdtable *fdt;
18109 + const unsigned long *bptr;
18110 + int count, total;
18112 + /* no rcu_read_lock() because of spin_lock() */
18113 + spin_lock(&files->file_lock);
18114 + fdt = files_fdtable(files);
18115 + bptr = fdt->open_fds->fds_bits;
18116 + count = fdt->max_fds / (sizeof(unsigned long) * 8);
18117 + for (total = 0; count > 0; count--) {
18119 + total += hweight_long(*bptr);
18122 + spin_unlock(&files->file_lock);
18127 +/* for *space compatibility */
18129 +asmlinkage long sys_unshare(unsigned long);
18132 + * migrate task to new context
18133 + * gets vxi, puts old_vxi on change
18134 + * optionally unshares namespaces (hack)
18137 +int vx_migrate_task(struct task_struct *p, struct vx_info *vxi, int unshare)
18139 + struct vx_info *old_vxi;
18145 + vxdprintk(VXD_CBIT(xid, 5),
18146 + "vx_migrate_task(%p,%p[#%d.%d])", p, vxi,
18147 + vxi->vx_id, atomic_read(&vxi->vx_usecnt));
18149 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0) &&
18150 + !vx_info_flags(vxi, VXF_STATE_SETUP, 0))
18153 + if (vx_info_state(vxi, VXS_SHUTDOWN))
18156 + old_vxi = task_get_vx_info(p);
18157 + if (old_vxi == vxi)
18160 +// if (!(ret = vx_migrate_user(p, vxi))) {
18165 + openfd = vx_openfd_task(p);
18168 + atomic_dec(&old_vxi->cvirt.nr_threads);
18169 + atomic_dec(&old_vxi->cvirt.nr_running);
18170 + __rlim_dec(&old_vxi->limit, RLIMIT_NPROC);
18171 + /* FIXME: what about the struct files here? */
18172 + __rlim_sub(&old_vxi->limit, VLIMIT_OPENFD, openfd);
18173 + /* account for the executable */
18174 + __rlim_dec(&old_vxi->limit, VLIMIT_DENTRY);
18176 + atomic_inc(&vxi->cvirt.nr_threads);
18177 + atomic_inc(&vxi->cvirt.nr_running);
18178 + __rlim_inc(&vxi->limit, RLIMIT_NPROC);
18179 + /* FIXME: what about the struct files here? */
18180 + __rlim_add(&vxi->limit, VLIMIT_OPENFD, openfd);
18181 + /* account for the executable */
18182 + __rlim_inc(&vxi->limit, VLIMIT_DENTRY);
18185 + release_vx_info(old_vxi, p);
18186 + clr_vx_info(&p->vx_info);
18188 + claim_vx_info(vxi, p);
18189 + set_vx_info(&p->vx_info, vxi);
18190 + p->xid = vxi->vx_id;
18192 + vxdprintk(VXD_CBIT(xid, 5),
18193 + "moved task %p into vxi:%p[#%d]",
18194 + p, vxi, vxi->vx_id);
18196 + // vx_mask_cap_bset(vxi, p);
18199 + /* hack for *spaces to provide compatibility */
18201 + struct nsproxy *old_nsp, *new_nsp;
18203 + ret = unshare_nsproxy_namespaces(
18204 + CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER,
18209 + old_nsp = xchg(&p->nsproxy, new_nsp);
18210 + vx_set_space(vxi,
18211 + CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER, 0);
18212 + put_nsproxy(old_nsp);
18216 + put_vx_info(old_vxi);
18220 +int vx_set_reaper(struct vx_info *vxi, struct task_struct *p)
18222 + struct task_struct *old_reaper;
18227 + vxdprintk(VXD_CBIT(xid, 6),
18228 + "vx_set_reaper(%p[#%d],%p[#%d,%d])",
18229 + vxi, vxi->vx_id, p, p->xid, p->pid);
18231 + old_reaper = vxi->vx_reaper;
18232 + if (old_reaper == p)
18235 + /* set new child reaper */
18236 + get_task_struct(p);
18237 + vxi->vx_reaper = p;
18238 + put_task_struct(old_reaper);
18242 +int vx_set_init(struct vx_info *vxi, struct task_struct *p)
18247 + vxdprintk(VXD_CBIT(xid, 6),
18248 + "vx_set_init(%p[#%d],%p[#%d,%d,%d])",
18249 + vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
18251 + vxi->vx_flags &= ~VXF_STATE_INIT;
18252 + // vxi->vx_initpid = p->tgid;
18253 + vxi->vx_initpid = p->pid;
18257 +void vx_exit_init(struct vx_info *vxi, struct task_struct *p, int code)
18259 + vxdprintk(VXD_CBIT(xid, 6),
18260 + "vx_exit_init(%p[#%d],%p[#%d,%d,%d])",
18261 + vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
18263 + vxi->exit_code = code;
18264 + vxi->vx_initpid = 0;
18268 +void vx_set_persistent(struct vx_info *vxi)
18270 + vxdprintk(VXD_CBIT(xid, 6),
18271 + "vx_set_persistent(%p[#%d])", vxi, vxi->vx_id);
18273 + get_vx_info(vxi);
18274 + claim_vx_info(vxi, NULL);
18277 +void vx_clear_persistent(struct vx_info *vxi)
18279 + vxdprintk(VXD_CBIT(xid, 6),
18280 + "vx_clear_persistent(%p[#%d])", vxi, vxi->vx_id);
18282 + release_vx_info(vxi, NULL);
18283 + put_vx_info(vxi);
18286 +void vx_update_persistent(struct vx_info *vxi)
18288 + if (vx_info_flags(vxi, VXF_PERSISTENT, 0))
18289 + vx_set_persistent(vxi);
18291 + vx_clear_persistent(vxi);
18295 +/* task must be current or locked */
18297 +void exit_vx_info(struct task_struct *p, int code)
18299 + struct vx_info *vxi = p->vx_info;
18302 + atomic_dec(&vxi->cvirt.nr_threads);
18305 + vxi->exit_code = code;
18306 + release_vx_info(vxi, p);
18310 +void exit_vx_info_early(struct task_struct *p, int code)
18312 + struct vx_info *vxi = p->vx_info;
18315 + if (vxi->vx_initpid == p->pid)
18316 + vx_exit_init(vxi, p, code);
18317 + if (vxi->vx_reaper == p)
18318 + vx_set_reaper(vxi, init_pid_ns.child_reaper);
18323 +/* vserver syscall commands below here */
18325 +/* taks xid and vx_info functions */
18327 +#include <asm/uaccess.h>
18330 +int vc_task_xid(uint32_t id)
18335 + struct task_struct *tsk;
18337 + read_lock(&tasklist_lock);
18338 + tsk = find_task_by_real_pid(id);
18339 + xid = (tsk) ? tsk->xid : -ESRCH;
18340 + read_unlock(&tasklist_lock);
18342 + xid = vx_current_xid();
18347 +int vc_vx_info(struct vx_info *vxi, void __user *data)
18349 + struct vcmd_vx_info_v0 vc_data;
18351 + vc_data.xid = vxi->vx_id;
18352 + vc_data.initpid = vxi->vx_initpid;
18354 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18360 +int vc_ctx_stat(struct vx_info *vxi, void __user *data)
18362 + struct vcmd_ctx_stat_v0 vc_data;
18364 + vc_data.usecnt = atomic_read(&vxi->vx_usecnt);
18365 + vc_data.tasks = atomic_read(&vxi->vx_tasks);
18367 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18373 +/* context functions */
18375 +int vc_ctx_create(uint32_t xid, void __user *data)
18377 + struct vcmd_ctx_create vc_data = { .flagword = VXF_INIT_SET };
18378 + struct vx_info *new_vxi;
18381 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
18384 + if ((xid > MAX_S_CONTEXT) || (xid < 2))
18387 + new_vxi = __create_vx_info(xid);
18388 + if (IS_ERR(new_vxi))
18389 + return PTR_ERR(new_vxi);
18391 + /* initial flags */
18392 + new_vxi->vx_flags = vc_data.flagword;
18395 + if (vs_state_change(new_vxi, VSC_STARTUP))
18398 + ret = vx_migrate_task(current, new_vxi, (!data));
18402 + /* return context id on success */
18403 + ret = new_vxi->vx_id;
18405 + /* get a reference for persistent contexts */
18406 + if ((vc_data.flagword & VXF_PERSISTENT))
18407 + vx_set_persistent(new_vxi);
18409 + release_vx_info(new_vxi, NULL);
18410 + put_vx_info(new_vxi);
18415 +int vc_ctx_migrate(struct vx_info *vxi, void __user *data)
18417 + struct vcmd_ctx_migrate vc_data = { .flagword = 0 };
18420 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
18423 + ret = vx_migrate_task(current, vxi, 0);
18426 + if (vc_data.flagword & VXM_SET_INIT)
18427 + ret = vx_set_init(vxi, current);
18430 + if (vc_data.flagword & VXM_SET_REAPER)
18431 + ret = vx_set_reaper(vxi, current);
18436 +int vc_get_cflags(struct vx_info *vxi, void __user *data)
18438 + struct vcmd_ctx_flags_v0 vc_data;
18440 + vc_data.flagword = vxi->vx_flags;
18442 + /* special STATE flag handling */
18443 + vc_data.mask = vs_mask_flags(~0ULL, vxi->vx_flags, VXF_ONE_TIME);
18445 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18450 +int vc_set_cflags(struct vx_info *vxi, void __user *data)
18452 + struct vcmd_ctx_flags_v0 vc_data;
18453 + uint64_t mask, trigger;
18455 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18458 + /* special STATE flag handling */
18459 + mask = vs_mask_mask(vc_data.mask, vxi->vx_flags, VXF_ONE_TIME);
18460 + trigger = (mask & vxi->vx_flags) ^ (mask & vc_data.flagword);
18462 + if (vxi == current_vx_info()) {
18463 + /* if (trigger & VXF_STATE_SETUP)
18464 + vx_mask_cap_bset(vxi, current); */
18465 + if (trigger & VXF_STATE_INIT) {
18468 + ret = vx_set_init(vxi, current);
18471 + ret = vx_set_reaper(vxi, current);
18477 + vxi->vx_flags = vs_mask_flags(vxi->vx_flags,
18478 + vc_data.flagword, mask);
18479 + if (trigger & VXF_PERSISTENT)
18480 + vx_update_persistent(vxi);
18486 +static inline uint64_t caps_from_cap_t(kernel_cap_t c)
18488 + uint64_t v = c.cap[0] | ((uint64_t)c.cap[1] << 32);
18490 + // printk("caps_from_cap_t(%08x:%08x) = %016llx\n", c.cap[1], c.cap[0], v);
18494 +static inline kernel_cap_t cap_t_from_caps(uint64_t v)
18496 + kernel_cap_t c = __cap_empty_set;
18498 + c.cap[0] = v & 0xFFFFFFFF;
18499 + c.cap[1] = (v >> 32) & 0xFFFFFFFF;
18501 + // printk("cap_t_from_caps(%016llx) = %08x:%08x\n", v, c.cap[1], c.cap[0]);
18506 +static int do_get_caps(struct vx_info *vxi, uint64_t *bcaps, uint64_t *ccaps)
18509 + *bcaps = caps_from_cap_t(vxi->vx_bcaps);
18511 + *ccaps = vxi->vx_ccaps;
18516 +int vc_get_ccaps(struct vx_info *vxi, void __user *data)
18518 + struct vcmd_ctx_caps_v1 vc_data;
18521 + ret = do_get_caps(vxi, NULL, &vc_data.ccaps);
18524 + vc_data.cmask = ~0ULL;
18526 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18531 +static int do_set_caps(struct vx_info *vxi,
18532 + uint64_t bcaps, uint64_t bmask, uint64_t ccaps, uint64_t cmask)
18534 + uint64_t bcold = caps_from_cap_t(vxi->vx_bcaps);
18537 + printk("do_set_caps(%16llx, %16llx, %16llx, %16llx)\n",
18538 + bcaps, bmask, ccaps, cmask);
18540 + vxi->vx_bcaps = cap_t_from_caps(
18541 + vs_mask_flags(bcold, bcaps, bmask));
18542 + vxi->vx_ccaps = vs_mask_flags(vxi->vx_ccaps, ccaps, cmask);
18547 +int vc_set_ccaps(struct vx_info *vxi, void __user *data)
18549 + struct vcmd_ctx_caps_v1 vc_data;
18551 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18554 + return do_set_caps(vxi, 0, 0, vc_data.ccaps, vc_data.cmask);
18557 +int vc_get_bcaps(struct vx_info *vxi, void __user *data)
18559 + struct vcmd_bcaps vc_data;
18562 + ret = do_get_caps(vxi, &vc_data.bcaps, NULL);
18565 + vc_data.bmask = ~0ULL;
18567 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18572 +int vc_set_bcaps(struct vx_info *vxi, void __user *data)
18574 + struct vcmd_bcaps vc_data;
18576 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18579 + return do_set_caps(vxi, vc_data.bcaps, vc_data.bmask, 0, 0);
18583 +int vc_get_umask(struct vx_info *vxi, void __user *data)
18585 + struct vcmd_umask vc_data;
18587 + vc_data.umask = vxi->vx_umask;
18588 + vc_data.mask = ~0ULL;
18590 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18595 +int vc_set_umask(struct vx_info *vxi, void __user *data)
18597 + struct vcmd_umask vc_data;
18599 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18602 + vxi->vx_umask = vs_mask_flags(vxi->vx_umask,
18603 + vc_data.umask, vc_data.mask);
18608 +int vc_get_badness(struct vx_info *vxi, void __user *data)
18610 + struct vcmd_badness_v0 vc_data;
18612 + vc_data.bias = vxi->vx_badness_bias;
18614 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18619 +int vc_set_badness(struct vx_info *vxi, void __user *data)
18621 + struct vcmd_badness_v0 vc_data;
18623 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18626 + vxi->vx_badness_bias = vc_data.bias;
18630 +#include <linux/module.h>
18632 +EXPORT_SYMBOL_GPL(free_vx_info);
18634 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cvirt.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt.c
18635 --- linux-2.6.32.1/kernel/vserver/cvirt.c 1970-01-01 01:00:00.000000000 +0100
18636 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt.c 2009-12-03 20:04:56.000000000 +0100
18639 + * linux/kernel/vserver/cvirt.c
18641 + * Virtual Server: Context Virtualization
18643 + * Copyright (C) 2004-2007 Herbert Pötzl
18645 + * V0.01 broken out from limit.c
18646 + * V0.02 added utsname stuff
18647 + * V0.03 changed vcmds to vxi arg
18651 +#include <linux/types.h>
18652 +#include <linux/utsname.h>
18653 +#include <linux/vs_cvirt.h>
18654 +#include <linux/vserver/switch.h>
18655 +#include <linux/vserver/cvirt_cmd.h>
18657 +#include <asm/uaccess.h>
18660 +void vx_vsi_uptime(struct timespec *uptime, struct timespec *idle)
18662 + struct vx_info *vxi = current_vx_info();
18664 + set_normalized_timespec(uptime,
18665 + uptime->tv_sec - vxi->cvirt.bias_uptime.tv_sec,
18666 + uptime->tv_nsec - vxi->cvirt.bias_uptime.tv_nsec);
18669 + set_normalized_timespec(idle,
18670 + idle->tv_sec - vxi->cvirt.bias_idle.tv_sec,
18671 + idle->tv_nsec - vxi->cvirt.bias_idle.tv_nsec);
18675 +uint64_t vx_idle_jiffies(void)
18677 + return init_task.utime + init_task.stime;
18682 +static inline uint32_t __update_loadavg(uint32_t load,
18683 + int wsize, int delta, int n)
18685 + unsigned long long calc, prev;
18687 + /* just set it to n */
18688 + if (unlikely(delta >= wsize))
18689 + return (n << FSHIFT);
18691 + calc = delta * n;
18693 + prev = (wsize - delta);
18696 + do_div(calc, wsize);
18701 +void vx_update_load(struct vx_info *vxi)
18703 + uint32_t now, last, delta;
18704 + unsigned int nr_running, nr_uninterruptible;
18705 + unsigned int total;
18706 + unsigned long flags;
18708 + spin_lock_irqsave(&vxi->cvirt.load_lock, flags);
18711 + last = vxi->cvirt.load_last;
18712 + delta = now - last;
18714 + if (delta < 5*HZ)
18717 + nr_running = atomic_read(&vxi->cvirt.nr_running);
18718 + nr_uninterruptible = atomic_read(&vxi->cvirt.nr_uninterruptible);
18719 + total = nr_running + nr_uninterruptible;
18721 + vxi->cvirt.load[0] = __update_loadavg(vxi->cvirt.load[0],
18722 + 60*HZ, delta, total);
18723 + vxi->cvirt.load[1] = __update_loadavg(vxi->cvirt.load[1],
18724 + 5*60*HZ, delta, total);
18725 + vxi->cvirt.load[2] = __update_loadavg(vxi->cvirt.load[2],
18726 + 15*60*HZ, delta, total);
18728 + vxi->cvirt.load_last = now;
18730 + atomic_inc(&vxi->cvirt.load_updates);
18731 + spin_unlock_irqrestore(&vxi->cvirt.load_lock, flags);
18736 + * Commands to do_syslog:
18738 + * 0 -- Close the log. Currently a NOP.
18739 + * 1 -- Open the log. Currently a NOP.
18740 + * 2 -- Read from the log.
18741 + * 3 -- Read all messages remaining in the ring buffer.
18742 + * 4 -- Read and clear all messages remaining in the ring buffer
18743 + * 5 -- Clear ring buffer.
18744 + * 6 -- Disable printk's to console
18745 + * 7 -- Enable printk's to console
18746 + * 8 -- Set level of messages printed to console
18747 + * 9 -- Return number of unread characters in the log buffer
18748 + * 10 -- Return size of the log buffer
18750 +int vx_do_syslog(int type, char __user *buf, int len)
18753 + int do_clear = 0;
18754 + struct vx_info *vxi = current_vx_info();
18755 + struct _vx_syslog *log;
18759 + log = &vxi->cvirt.syslog;
18762 + case 0: /* Close log */
18763 + case 1: /* Open log */
18765 + case 2: /* Read from log */
18766 + error = wait_event_interruptible(log->log_wait,
18767 + (log->log_start - log->log_end));
18770 + spin_lock_irq(&log->logbuf_lock);
18771 + spin_unlock_irq(&log->logbuf_lock);
18773 + case 4: /* Read/clear last kernel messages */
18775 + /* fall through */
18776 + case 3: /* Read last kernel messages */
18779 + case 5: /* Clear ring buffer */
18782 + case 6: /* Disable logging to console */
18783 + case 7: /* Enable logging to console */
18784 + case 8: /* Set level of messages printed to console */
18787 + case 9: /* Number of chars in the log buffer */
18789 + case 10: /* Size of the log buffer */
18799 +/* virtual host info names */
18801 +static char *vx_vhi_name(struct vx_info *vxi, int id)
18803 + struct nsproxy *nsproxy;
18804 + struct uts_namespace *uts;
18806 + if (id == VHIN_CONTEXT)
18807 + return vxi->vx_name;
18809 + nsproxy = vxi->vx_nsproxy[0];
18813 + uts = nsproxy->uts_ns;
18818 + case VHIN_SYSNAME:
18819 + return uts->name.sysname;
18820 + case VHIN_NODENAME:
18821 + return uts->name.nodename;
18822 + case VHIN_RELEASE:
18823 + return uts->name.release;
18824 + case VHIN_VERSION:
18825 + return uts->name.version;
18826 + case VHIN_MACHINE:
18827 + return uts->name.machine;
18828 + case VHIN_DOMAINNAME:
18829 + return uts->name.domainname;
18836 +int vc_set_vhi_name(struct vx_info *vxi, void __user *data)
18838 + struct vcmd_vhi_name_v0 vc_data;
18841 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18844 + name = vx_vhi_name(vxi, vc_data.field);
18848 + memcpy(name, vc_data.name, 65);
18852 +int vc_get_vhi_name(struct vx_info *vxi, void __user *data)
18854 + struct vcmd_vhi_name_v0 vc_data;
18857 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18860 + name = vx_vhi_name(vxi, vc_data.field);
18864 + memcpy(vc_data.name, name, 65);
18865 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18871 +int vc_virt_stat(struct vx_info *vxi, void __user *data)
18873 + struct vcmd_virt_stat_v0 vc_data;
18874 + struct _vx_cvirt *cvirt = &vxi->cvirt;
18875 + struct timespec uptime;
18877 + do_posix_clock_monotonic_gettime(&uptime);
18878 + set_normalized_timespec(&uptime,
18879 + uptime.tv_sec - cvirt->bias_uptime.tv_sec,
18880 + uptime.tv_nsec - cvirt->bias_uptime.tv_nsec);
18882 + vc_data.offset = timeval_to_ns(&cvirt->bias_tv);
18883 + vc_data.uptime = timespec_to_ns(&uptime);
18884 + vc_data.nr_threads = atomic_read(&cvirt->nr_threads);
18885 + vc_data.nr_running = atomic_read(&cvirt->nr_running);
18886 + vc_data.nr_uninterruptible = atomic_read(&cvirt->nr_uninterruptible);
18887 + vc_data.nr_onhold = atomic_read(&cvirt->nr_onhold);
18888 + vc_data.nr_forks = atomic_read(&cvirt->total_forks);
18889 + vc_data.load[0] = cvirt->load[0];
18890 + vc_data.load[1] = cvirt->load[1];
18891 + vc_data.load[2] = cvirt->load[2];
18893 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18899 +#ifdef CONFIG_VSERVER_VTIME
18901 +/* virtualized time base */
18903 +void vx_gettimeofday(struct timeval *tv)
18905 + struct vx_info *vxi;
18907 + do_gettimeofday(tv);
18908 + if (!vx_flags(VXF_VIRT_TIME, 0))
18911 + vxi = current_vx_info();
18912 + tv->tv_sec += vxi->cvirt.bias_tv.tv_sec;
18913 + tv->tv_usec += vxi->cvirt.bias_tv.tv_usec;
18915 + if (tv->tv_usec >= USEC_PER_SEC) {
18917 + tv->tv_usec -= USEC_PER_SEC;
18918 + } else if (tv->tv_usec < 0) {
18920 + tv->tv_usec += USEC_PER_SEC;
18924 +int vx_settimeofday(struct timespec *ts)
18926 + struct timeval tv;
18927 + struct vx_info *vxi;
18929 + if (!vx_flags(VXF_VIRT_TIME, 0))
18930 + return do_settimeofday(ts);
18932 + do_gettimeofday(&tv);
18933 + vxi = current_vx_info();
18934 + vxi->cvirt.bias_tv.tv_sec = ts->tv_sec - tv.tv_sec;
18935 + vxi->cvirt.bias_tv.tv_usec =
18936 + (ts->tv_nsec/NSEC_PER_USEC) - tv.tv_usec;
18942 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cvirt_init.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt_init.h
18943 --- linux-2.6.32.1/kernel/vserver/cvirt_init.h 1970-01-01 01:00:00.000000000 +0100
18944 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt_init.h 2009-12-03 20:04:56.000000000 +0100
18948 +extern uint64_t vx_idle_jiffies(void);
18950 +static inline void vx_info_init_cvirt(struct _vx_cvirt *cvirt)
18952 + uint64_t idle_jiffies = vx_idle_jiffies();
18953 + uint64_t nsuptime;
18955 + do_posix_clock_monotonic_gettime(&cvirt->bias_uptime);
18956 + nsuptime = (unsigned long long)cvirt->bias_uptime.tv_sec
18957 + * NSEC_PER_SEC + cvirt->bias_uptime.tv_nsec;
18958 + cvirt->bias_clock = nsec_to_clock_t(nsuptime);
18959 + cvirt->bias_tv.tv_sec = 0;
18960 + cvirt->bias_tv.tv_usec = 0;
18962 + jiffies_to_timespec(idle_jiffies, &cvirt->bias_idle);
18963 + atomic_set(&cvirt->nr_threads, 0);
18964 + atomic_set(&cvirt->nr_running, 0);
18965 + atomic_set(&cvirt->nr_uninterruptible, 0);
18966 + atomic_set(&cvirt->nr_onhold, 0);
18968 + spin_lock_init(&cvirt->load_lock);
18969 + cvirt->load_last = jiffies;
18970 + atomic_set(&cvirt->load_updates, 0);
18971 + cvirt->load[0] = 0;
18972 + cvirt->load[1] = 0;
18973 + cvirt->load[2] = 0;
18974 + atomic_set(&cvirt->total_forks, 0);
18976 + spin_lock_init(&cvirt->syslog.logbuf_lock);
18977 + init_waitqueue_head(&cvirt->syslog.log_wait);
18978 + cvirt->syslog.log_start = 0;
18979 + cvirt->syslog.log_end = 0;
18980 + cvirt->syslog.con_start = 0;
18981 + cvirt->syslog.logged_chars = 0;
18985 +void vx_info_init_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
18987 + // cvirt_pc->cpustat = { 0 };
18990 +static inline void vx_info_exit_cvirt(struct _vx_cvirt *cvirt)
18994 + vxwprintk_xid((value = atomic_read(&cvirt->nr_threads)),
18995 + "!!! cvirt: %p[nr_threads] = %d on exit.",
18997 + vxwprintk_xid((value = atomic_read(&cvirt->nr_running)),
18998 + "!!! cvirt: %p[nr_running] = %d on exit.",
19000 + vxwprintk_xid((value = atomic_read(&cvirt->nr_uninterruptible)),
19001 + "!!! cvirt: %p[nr_uninterruptible] = %d on exit.",
19003 + vxwprintk_xid((value = atomic_read(&cvirt->nr_onhold)),
19004 + "!!! cvirt: %p[nr_onhold] = %d on exit.",
19010 +void vx_info_exit_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
19015 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/cvirt_proc.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt_proc.h
19016 --- linux-2.6.32.1/kernel/vserver/cvirt_proc.h 1970-01-01 01:00:00.000000000 +0100
19017 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/cvirt_proc.h 2009-12-03 20:04:56.000000000 +0100
19019 +#ifndef _VX_CVIRT_PROC_H
19020 +#define _VX_CVIRT_PROC_H
19022 +#include <linux/nsproxy.h>
19023 +#include <linux/mnt_namespace.h>
19024 +#include <linux/ipc_namespace.h>
19025 +#include <linux/utsname.h>
19026 +#include <linux/ipc.h>
19030 +int vx_info_proc_nsproxy(struct nsproxy *nsproxy, char *buffer)
19032 + struct mnt_namespace *ns;
19033 + struct uts_namespace *uts;
19034 + struct ipc_namespace *ipc;
19035 + struct path path;
19036 + char *pstr, *root;
19042 + length += sprintf(buffer + length,
19043 + "NSProxy:\t%p [%p,%p,%p]\n",
19044 + nsproxy, nsproxy->mnt_ns,
19045 + nsproxy->uts_ns, nsproxy->ipc_ns);
19047 + ns = nsproxy->mnt_ns;
19051 + pstr = kmalloc(PATH_MAX, GFP_KERNEL);
19055 + path.mnt = ns->root;
19056 + path.dentry = ns->root->mnt_root;
19057 + root = d_path(&path, pstr, PATH_MAX - 2);
19058 + length += sprintf(buffer + length,
19059 + "Namespace:\t%p [#%u]\n"
19060 + "RootPath:\t%s\n",
19061 + ns, atomic_read(&ns->count),
19066 + uts = nsproxy->uts_ns;
19070 + length += sprintf(buffer + length,
19071 + "SysName:\t%.*s\n"
19072 + "NodeName:\t%.*s\n"
19073 + "Release:\t%.*s\n"
19074 + "Version:\t%.*s\n"
19075 + "Machine:\t%.*s\n"
19076 + "DomainName:\t%.*s\n",
19077 + __NEW_UTS_LEN, uts->name.sysname,
19078 + __NEW_UTS_LEN, uts->name.nodename,
19079 + __NEW_UTS_LEN, uts->name.release,
19080 + __NEW_UTS_LEN, uts->name.version,
19081 + __NEW_UTS_LEN, uts->name.machine,
19082 + __NEW_UTS_LEN, uts->name.domainname);
19085 + ipc = nsproxy->ipc_ns;
19089 + length += sprintf(buffer + length,
19090 + "SEMS:\t\t%d %d %d %d %d\n"
19091 + "MSG:\t\t%d %d %d\n"
19092 + "SHM:\t\t%lu %lu %d %d\n",
19093 + ipc->sem_ctls[0], ipc->sem_ctls[1],
19094 + ipc->sem_ctls[2], ipc->sem_ctls[3],
19096 + ipc->msg_ctlmax, ipc->msg_ctlmnb, ipc->msg_ctlmni,
19097 + (unsigned long)ipc->shm_ctlmax,
19098 + (unsigned long)ipc->shm_ctlall,
19099 + ipc->shm_ctlmni, ipc->shm_tot);
19106 +#include <linux/sched.h>
19108 +#define LOAD_INT(x) ((x) >> FSHIFT)
19109 +#define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1 - 1)) * 100)
19112 +int vx_info_proc_cvirt(struct _vx_cvirt *cvirt, char *buffer)
19117 + length += sprintf(buffer + length,
19118 + "BiasUptime:\t%lu.%02lu\n",
19119 + (unsigned long)cvirt->bias_uptime.tv_sec,
19120 + (cvirt->bias_uptime.tv_nsec / (NSEC_PER_SEC / 100)));
19122 + a = cvirt->load[0] + (FIXED_1 / 200);
19123 + b = cvirt->load[1] + (FIXED_1 / 200);
19124 + c = cvirt->load[2] + (FIXED_1 / 200);
19125 + length += sprintf(buffer + length,
19126 + "nr_threads:\t%d\n"
19127 + "nr_running:\t%d\n"
19128 + "nr_unintr:\t%d\n"
19129 + "nr_onhold:\t%d\n"
19130 + "load_updates:\t%d\n"
19131 + "loadavg:\t%d.%02d %d.%02d %d.%02d\n"
19132 + "total_forks:\t%d\n",
19133 + atomic_read(&cvirt->nr_threads),
19134 + atomic_read(&cvirt->nr_running),
19135 + atomic_read(&cvirt->nr_uninterruptible),
19136 + atomic_read(&cvirt->nr_onhold),
19137 + atomic_read(&cvirt->load_updates),
19138 + LOAD_INT(a), LOAD_FRAC(a),
19139 + LOAD_INT(b), LOAD_FRAC(b),
19140 + LOAD_INT(c), LOAD_FRAC(c),
19141 + atomic_read(&cvirt->total_forks));
19146 +int vx_info_proc_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc,
19147 + char *buffer, int cpu)
19153 +#endif /* _VX_CVIRT_PROC_H */
19154 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/debug.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/debug.c
19155 --- linux-2.6.32.1/kernel/vserver/debug.c 1970-01-01 01:00:00.000000000 +0100
19156 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/debug.c 2009-12-03 20:04:56.000000000 +0100
19159 + * kernel/vserver/debug.c
19161 + * Copyright (C) 2005-2007 Herbert Pötzl
19163 + * V0.01 vx_info dump support
19167 +#include <linux/module.h>
19169 +#include <linux/vserver/context.h>
19172 +void dump_vx_info(struct vx_info *vxi, int level)
19174 + printk("vx_info %p[#%d, %d.%d, %4x]\n", vxi, vxi->vx_id,
19175 + atomic_read(&vxi->vx_usecnt),
19176 + atomic_read(&vxi->vx_tasks),
19179 + __dump_vx_limit(&vxi->limit);
19180 + __dump_vx_sched(&vxi->sched);
19181 + __dump_vx_cvirt(&vxi->cvirt);
19182 + __dump_vx_cacct(&vxi->cacct);
19188 +EXPORT_SYMBOL_GPL(dump_vx_info);
19190 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/device.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/device.c
19191 --- linux-2.6.32.1/kernel/vserver/device.c 1970-01-01 01:00:00.000000000 +0100
19192 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/device.c 2009-12-03 20:04:56.000000000 +0100
19195 + * linux/kernel/vserver/device.c
19197 + * Linux-VServer: Device Support
19199 + * Copyright (C) 2006 Herbert Pötzl
19200 + * Copyright (C) 2007 Daniel Hokka Zakrisson
19202 + * V0.01 device mapping basics
19203 + * V0.02 added defaults
19207 +#include <linux/slab.h>
19208 +#include <linux/rcupdate.h>
19209 +#include <linux/fs.h>
19210 +#include <linux/namei.h>
19211 +#include <linux/hash.h>
19213 +#include <asm/errno.h>
19214 +#include <asm/uaccess.h>
19215 +#include <linux/vserver/base.h>
19216 +#include <linux/vserver/debug.h>
19217 +#include <linux/vserver/context.h>
19218 +#include <linux/vserver/device.h>
19219 +#include <linux/vserver/device_cmd.h>
19222 +#define DMAP_HASH_BITS 4
19225 +struct vs_mapping {
19227 + struct hlist_node hlist;
19228 + struct list_head list;
19230 +#define dm_hlist u.hlist
19231 +#define dm_list u.list
19234 + struct vx_dmap_target target;
19238 +static struct hlist_head dmap_main_hash[1 << DMAP_HASH_BITS];
19240 +static spinlock_t dmap_main_hash_lock = SPIN_LOCK_UNLOCKED;
19242 +static struct vx_dmap_target dmap_defaults[2] = {
19243 + { .flags = DATTR_OPEN },
19244 + { .flags = DATTR_OPEN },
19248 +struct kmem_cache *dmap_cachep __read_mostly;
19250 +int __init dmap_cache_init(void)
19252 + dmap_cachep = kmem_cache_create("dmap_cache",
19253 + sizeof(struct vs_mapping), 0,
19254 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
19258 +__initcall(dmap_cache_init);
19261 +static inline unsigned int __hashval(dev_t dev, int bits)
19263 + return hash_long((unsigned long)dev, bits);
19267 +/* __hash_mapping()
19268 + * add the mapping to the hash table
19270 +static inline void __hash_mapping(struct vx_info *vxi, struct vs_mapping *vdm)
19272 + spinlock_t *hash_lock = &dmap_main_hash_lock;
19273 + struct hlist_head *head, *hash = dmap_main_hash;
19274 + int device = vdm->device;
19276 + spin_lock(hash_lock);
19277 + vxdprintk(VXD_CBIT(misc, 8), "__hash_mapping: %p[#%d] %08x:%08x",
19278 + vxi, vxi ? vxi->vx_id : 0, device, vdm->target.target);
19280 + head = &hash[__hashval(device, DMAP_HASH_BITS)];
19281 + hlist_add_head(&vdm->dm_hlist, head);
19282 + spin_unlock(hash_lock);
19286 +static inline int __mode_to_default(umode_t mode)
19299 +/* __set_default()
19302 +static inline void __set_default(struct vx_info *vxi, umode_t mode,
19303 + struct vx_dmap_target *vdmt)
19305 + spinlock_t *hash_lock = &dmap_main_hash_lock;
19306 + spin_lock(hash_lock);
19309 + vxi->dmap.targets[__mode_to_default(mode)] = *vdmt;
19311 + dmap_defaults[__mode_to_default(mode)] = *vdmt;
19314 + spin_unlock(hash_lock);
19316 + vxdprintk(VXD_CBIT(misc, 8), "__set_default: %p[#%u] %08x %04x",
19317 + vxi, vxi ? vxi->vx_id : 0, vdmt->target, vdmt->flags);
19321 +/* __remove_default()
19322 + * remove a default
19324 +static inline int __remove_default(struct vx_info *vxi, umode_t mode)
19326 + spinlock_t *hash_lock = &dmap_main_hash_lock;
19327 + spin_lock(hash_lock);
19330 + vxi->dmap.targets[__mode_to_default(mode)].flags = 0;
19331 + else /* remove == reset */
19332 + dmap_defaults[__mode_to_default(mode)].flags = DATTR_OPEN | mode;
19334 + spin_unlock(hash_lock);
19339 +/* __find_mapping()
19340 + * find a mapping in the hash table
19342 + * caller must hold hash_lock
19344 +static inline int __find_mapping(xid_t xid, dev_t device, umode_t mode,
19345 + struct vs_mapping **local, struct vs_mapping **global)
19347 + struct hlist_head *hash = dmap_main_hash;
19348 + struct hlist_head *head = &hash[__hashval(device, DMAP_HASH_BITS)];
19349 + struct hlist_node *pos;
19350 + struct vs_mapping *vdm;
19356 + hlist_for_each(pos, head) {
19357 + vdm = hlist_entry(pos, struct vs_mapping, dm_hlist);
19359 + if ((vdm->device == device) &&
19360 + !((vdm->target.flags ^ mode) & S_IFMT)) {
19361 + if (vdm->xid == xid) {
19364 + } else if (global && vdm->xid == 0)
19369 + if (global && *global)
19376 +/* __lookup_mapping()
19377 + * find a mapping and store the result in target and flags
19379 +static inline int __lookup_mapping(struct vx_info *vxi,
19380 + dev_t device, dev_t *target, int *flags, umode_t mode)
19382 + spinlock_t *hash_lock = &dmap_main_hash_lock;
19383 + struct vs_mapping *vdm, *global;
19384 + struct vx_dmap_target *vdmt;
19386 + xid_t xid = vxi->vx_id;
19389 + spin_lock(hash_lock);
19390 + if (__find_mapping(xid, device, mode, &vdm, &global) > 0) {
19392 + vdmt = &vdm->target;
19396 + index = __mode_to_default(mode);
19397 + if (vxi && vxi->dmap.targets[index].flags) {
19399 + vdmt = &vxi->dmap.targets[index];
19400 + } else if (global) {
19402 + vdmt = &global->target;
19406 + vdmt = &dmap_defaults[index];
19410 + if (target && (vdmt->flags & DATTR_REMAP))
19411 + *target = vdmt->target;
19413 + *target = device;
19415 + *flags = vdmt->flags;
19417 + spin_unlock(hash_lock);
19423 +/* __remove_mapping()
19424 + * remove a mapping from the hash table
19426 +static inline int __remove_mapping(struct vx_info *vxi, dev_t device,
19429 + spinlock_t *hash_lock = &dmap_main_hash_lock;
19430 + struct vs_mapping *vdm = NULL;
19433 + spin_lock(hash_lock);
19435 + ret = __find_mapping((vxi ? vxi->vx_id : 0), device, mode, &vdm,
19437 + vxdprintk(VXD_CBIT(misc, 8), "__remove_mapping: %p[#%d] %08x %04x",
19438 + vxi, vxi ? vxi->vx_id : 0, device, mode);
19441 + hlist_del(&vdm->dm_hlist);
19444 + spin_unlock(hash_lock);
19446 + kmem_cache_free(dmap_cachep, vdm);
19452 +int vs_map_device(struct vx_info *vxi,
19453 + dev_t device, dev_t *target, umode_t mode)
19455 + int ret, flags = DATTR_MASK;
19459 + *target = device;
19462 + ret = __lookup_mapping(vxi, device, target, &flags, mode);
19463 + vxdprintk(VXD_CBIT(misc, 8), "vs_map_device: %08x target: %08x flags: %04x mode: %04x mapped=%d",
19464 + device, target ? *target : 0, flags, mode, ret);
19466 + return (flags & DATTR_MASK);
19471 +static int do_set_mapping(struct vx_info *vxi,
19472 + dev_t device, dev_t target, int flags, umode_t mode)
19475 + struct vs_mapping *new;
19477 + new = kmem_cache_alloc(dmap_cachep, GFP_KERNEL);
19481 + INIT_HLIST_NODE(&new->dm_hlist);
19482 + new->device = device;
19483 + new->target.target = target;
19484 + new->target.flags = flags | mode;
19485 + new->xid = (vxi ? vxi->vx_id : 0);
19487 + vxdprintk(VXD_CBIT(misc, 8), "do_set_mapping: %08x target: %08x flags: %04x", device, target, flags);
19488 + __hash_mapping(vxi, new);
19490 + struct vx_dmap_target new = {
19491 + .target = target,
19492 + .flags = flags | mode,
19494 + __set_default(vxi, mode, &new);
19500 +static int do_unset_mapping(struct vx_info *vxi,
19501 + dev_t device, dev_t target, int flags, umode_t mode)
19503 + int ret = -EINVAL;
19506 + ret = __remove_mapping(vxi, device, mode);
19510 + ret = __remove_default(vxi, mode);
19520 +static inline int __user_device(const char __user *name, dev_t *dev,
19523 + struct nameidata nd;
19530 + ret = user_lpath(name, &nd.path);
19533 + if (nd.path.dentry->d_inode) {
19534 + *dev = nd.path.dentry->d_inode->i_rdev;
19535 + *mode = nd.path.dentry->d_inode->i_mode;
19537 + path_put(&nd.path);
19541 +static inline int __mapping_mode(dev_t device, dev_t target,
19542 + umode_t device_mode, umode_t target_mode, umode_t *mode)
19545 + *mode = device_mode & S_IFMT;
19547 + *mode = target_mode & S_IFMT;
19551 + /* if both given, device and target mode have to match */
19552 + if (device && target &&
19553 + ((device_mode ^ target_mode) & S_IFMT))
19559 +static inline int do_mapping(struct vx_info *vxi, const char __user *device_path,
19560 + const char __user *target_path, int flags, int set)
19562 + dev_t device = ~0, target = ~0;
19563 + umode_t device_mode = 0, target_mode = 0, mode;
19566 + ret = __user_device(device_path, &device, &device_mode);
19569 + ret = __user_device(target_path, &target, &target_mode);
19573 + ret = __mapping_mode(device, target,
19574 + device_mode, target_mode, &mode);
19579 + return do_set_mapping(vxi, device, target,
19582 + return do_unset_mapping(vxi, device, target,
19587 +int vc_set_mapping(struct vx_info *vxi, void __user *data)
19589 + struct vcmd_set_mapping_v0 vc_data;
19591 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19594 + return do_mapping(vxi, vc_data.device, vc_data.target,
19595 + vc_data.flags, 1);
19598 +int vc_unset_mapping(struct vx_info *vxi, void __user *data)
19600 + struct vcmd_set_mapping_v0 vc_data;
19602 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19605 + return do_mapping(vxi, vc_data.device, vc_data.target,
19606 + vc_data.flags, 0);
19610 +#ifdef CONFIG_COMPAT
19612 +int vc_set_mapping_x32(struct vx_info *vxi, void __user *data)
19614 + struct vcmd_set_mapping_v0_x32 vc_data;
19616 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19619 + return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19620 + compat_ptr(vc_data.target_ptr), vc_data.flags, 1);
19623 +int vc_unset_mapping_x32(struct vx_info *vxi, void __user *data)
19625 + struct vcmd_set_mapping_v0_x32 vc_data;
19627 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19630 + return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19631 + compat_ptr(vc_data.target_ptr), vc_data.flags, 0);
19634 +#endif /* CONFIG_COMPAT */
19637 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/dlimit.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/dlimit.c
19638 --- linux-2.6.32.1/kernel/vserver/dlimit.c 1970-01-01 01:00:00.000000000 +0100
19639 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/dlimit.c 2009-12-03 20:04:56.000000000 +0100
19642 + * linux/kernel/vserver/dlimit.c
19644 + * Virtual Server: Context Disk Limits
19646 + * Copyright (C) 2004-2009 Herbert Pötzl
19648 + * V0.01 initial version
19649 + * V0.02 compat32 splitup
19650 + * V0.03 extended interface
19654 +#include <linux/statfs.h>
19655 +#include <linux/sched.h>
19656 +#include <linux/namei.h>
19657 +#include <linux/vs_tag.h>
19658 +#include <linux/vs_dlimit.h>
19659 +#include <linux/vserver/dlimit_cmd.h>
19661 +#include <asm/uaccess.h>
19663 +/* __alloc_dl_info()
19665 + * allocate an initialized dl_info struct
19666 + * doesn't make it visible (hash) */
19668 +static struct dl_info *__alloc_dl_info(struct super_block *sb, tag_t tag)
19670 + struct dl_info *new = NULL;
19672 + vxdprintk(VXD_CBIT(dlim, 5),
19673 + "alloc_dl_info(%p,%d)*", sb, tag);
19675 + /* would this benefit from a slab cache? */
19676 + new = kmalloc(sizeof(struct dl_info), GFP_KERNEL);
19680 + memset(new, 0, sizeof(struct dl_info));
19681 + new->dl_tag = tag;
19683 + INIT_RCU_HEAD(&new->dl_rcu);
19684 + INIT_HLIST_NODE(&new->dl_hlist);
19685 + spin_lock_init(&new->dl_lock);
19686 + atomic_set(&new->dl_refcnt, 0);
19687 + atomic_set(&new->dl_usecnt, 0);
19689 + /* rest of init goes here */
19691 + vxdprintk(VXD_CBIT(dlim, 4),
19692 + "alloc_dl_info(%p,%d) = %p", sb, tag, new);
19696 +/* __dealloc_dl_info()
19698 + * final disposal of dl_info */
19700 +static void __dealloc_dl_info(struct dl_info *dli)
19702 + vxdprintk(VXD_CBIT(dlim, 4),
19703 + "dealloc_dl_info(%p)", dli);
19705 + dli->dl_hlist.next = LIST_POISON1;
19706 + dli->dl_tag = -1;
19709 + BUG_ON(atomic_read(&dli->dl_usecnt));
19710 + BUG_ON(atomic_read(&dli->dl_refcnt));
19716 +/* hash table for dl_info hash */
19718 +#define DL_HASH_SIZE 13
19720 +struct hlist_head dl_info_hash[DL_HASH_SIZE];
19722 +static spinlock_t dl_info_hash_lock = SPIN_LOCK_UNLOCKED;
19725 +static inline unsigned int __hashval(struct super_block *sb, tag_t tag)
19727 + return ((tag ^ (unsigned long)sb) % DL_HASH_SIZE);
19732 +/* __hash_dl_info()
19734 + * add the dli to the global hash table
19735 + * requires the hash_lock to be held */
19737 +static inline void __hash_dl_info(struct dl_info *dli)
19739 + struct hlist_head *head;
19741 + vxdprintk(VXD_CBIT(dlim, 6),
19742 + "__hash_dl_info: %p[#%d]", dli, dli->dl_tag);
19743 + get_dl_info(dli);
19744 + head = &dl_info_hash[__hashval(dli->dl_sb, dli->dl_tag)];
19745 + hlist_add_head_rcu(&dli->dl_hlist, head);
19748 +/* __unhash_dl_info()
19750 + * remove the dli from the global hash table
19751 + * requires the hash_lock to be held */
19753 +static inline void __unhash_dl_info(struct dl_info *dli)
19755 + vxdprintk(VXD_CBIT(dlim, 6),
19756 + "__unhash_dl_info: %p[#%d]", dli, dli->dl_tag);
19757 + hlist_del_rcu(&dli->dl_hlist);
19758 + put_dl_info(dli);
19762 +/* __lookup_dl_info()
19764 + * requires the rcu_read_lock()
19765 + * doesn't increment the dl_refcnt */
19767 +static inline struct dl_info *__lookup_dl_info(struct super_block *sb, tag_t tag)
19769 + struct hlist_head *head = &dl_info_hash[__hashval(sb, tag)];
19770 + struct hlist_node *pos;
19771 + struct dl_info *dli;
19773 + hlist_for_each_entry_rcu(dli, pos, head, dl_hlist) {
19775 + if (dli->dl_tag == tag && dli->dl_sb == sb) {
19783 +struct dl_info *locate_dl_info(struct super_block *sb, tag_t tag)
19785 + struct dl_info *dli;
19788 + dli = get_dl_info(__lookup_dl_info(sb, tag));
19789 + vxdprintk(VXD_CBIT(dlim, 7),
19790 + "locate_dl_info(%p,#%d) = %p", sb, tag, dli);
19791 + rcu_read_unlock();
19795 +void rcu_free_dl_info(struct rcu_head *head)
19797 + struct dl_info *dli = container_of(head, struct dl_info, dl_rcu);
19798 + int usecnt, refcnt;
19800 + BUG_ON(!dli || !head);
19802 + usecnt = atomic_read(&dli->dl_usecnt);
19803 + BUG_ON(usecnt < 0);
19805 + refcnt = atomic_read(&dli->dl_refcnt);
19806 + BUG_ON(refcnt < 0);
19808 + vxdprintk(VXD_CBIT(dlim, 3),
19809 + "rcu_free_dl_info(%p)", dli);
19811 + __dealloc_dl_info(dli);
19813 + printk("!!! rcu didn't free\n");
19819 +static int do_addrem_dlimit(uint32_t id, const char __user *name,
19820 + uint32_t flags, int add)
19822 + struct path path;
19825 + ret = user_lpath(name, &path);
19827 + struct super_block *sb;
19828 + struct dl_info *dli;
19831 + if (!path.dentry->d_inode)
19832 + goto out_release;
19833 + if (!(sb = path.dentry->d_inode->i_sb))
19834 + goto out_release;
19837 + dli = __alloc_dl_info(sb, id);
19838 + spin_lock(&dl_info_hash_lock);
19841 + if (__lookup_dl_info(sb, id))
19843 + __hash_dl_info(dli);
19846 + spin_lock(&dl_info_hash_lock);
19847 + dli = __lookup_dl_info(sb, id);
19852 + __unhash_dl_info(dli);
19856 + spin_unlock(&dl_info_hash_lock);
19858 + __dealloc_dl_info(dli);
19865 +int vc_add_dlimit(uint32_t id, void __user *data)
19867 + struct vcmd_ctx_dlimit_base_v0 vc_data;
19869 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19872 + return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 1);
19875 +int vc_rem_dlimit(uint32_t id, void __user *data)
19877 + struct vcmd_ctx_dlimit_base_v0 vc_data;
19879 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19882 + return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 0);
19885 +#ifdef CONFIG_COMPAT
19887 +int vc_add_dlimit_x32(uint32_t id, void __user *data)
19889 + struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19891 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19894 + return do_addrem_dlimit(id,
19895 + compat_ptr(vc_data.name_ptr), vc_data.flags, 1);
19898 +int vc_rem_dlimit_x32(uint32_t id, void __user *data)
19900 + struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19902 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19905 + return do_addrem_dlimit(id,
19906 + compat_ptr(vc_data.name_ptr), vc_data.flags, 0);
19909 +#endif /* CONFIG_COMPAT */
19913 +int do_set_dlimit(uint32_t id, const char __user *name,
19914 + uint32_t space_used, uint32_t space_total,
19915 + uint32_t inodes_used, uint32_t inodes_total,
19916 + uint32_t reserved, uint32_t flags)
19918 + struct path path;
19921 + ret = user_lpath(name, &path);
19923 + struct super_block *sb;
19924 + struct dl_info *dli;
19927 + if (!path.dentry->d_inode)
19928 + goto out_release;
19929 + if (!(sb = path.dentry->d_inode->i_sb))
19930 + goto out_release;
19932 + /* sanity checks */
19933 + if ((reserved != CDLIM_KEEP &&
19934 + reserved > 100) ||
19935 + (inodes_used != CDLIM_KEEP &&
19936 + inodes_used > inodes_total) ||
19937 + (space_used != CDLIM_KEEP &&
19938 + space_used > space_total))
19939 + goto out_release;
19942 + dli = locate_dl_info(sb, id);
19944 + goto out_release;
19946 + spin_lock(&dli->dl_lock);
19948 + if (inodes_used != CDLIM_KEEP)
19949 + dli->dl_inodes_used = inodes_used;
19950 + if (inodes_total != CDLIM_KEEP)
19951 + dli->dl_inodes_total = inodes_total;
19952 + if (space_used != CDLIM_KEEP)
19953 + dli->dl_space_used = dlimit_space_32to64(
19954 + space_used, flags, DLIMS_USED);
19956 + if (space_total == CDLIM_INFINITY)
19957 + dli->dl_space_total = DLIM_INFINITY;
19958 + else if (space_total != CDLIM_KEEP)
19959 + dli->dl_space_total = dlimit_space_32to64(
19960 + space_total, flags, DLIMS_TOTAL);
19962 + if (reserved != CDLIM_KEEP)
19963 + dli->dl_nrlmult = (1 << 10) * (100 - reserved) / 100;
19965 + spin_unlock(&dli->dl_lock);
19967 + put_dl_info(dli);
19976 +int vc_set_dlimit(uint32_t id, void __user *data)
19978 + struct vcmd_ctx_dlimit_v0 vc_data;
19980 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19983 + return do_set_dlimit(id, vc_data.name,
19984 + vc_data.space_used, vc_data.space_total,
19985 + vc_data.inodes_used, vc_data.inodes_total,
19986 + vc_data.reserved, vc_data.flags);
19989 +#ifdef CONFIG_COMPAT
19991 +int vc_set_dlimit_x32(uint32_t id, void __user *data)
19993 + struct vcmd_ctx_dlimit_v0_x32 vc_data;
19995 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19998 + return do_set_dlimit(id, compat_ptr(vc_data.name_ptr),
19999 + vc_data.space_used, vc_data.space_total,
20000 + vc_data.inodes_used, vc_data.inodes_total,
20001 + vc_data.reserved, vc_data.flags);
20004 +#endif /* CONFIG_COMPAT */
20008 +int do_get_dlimit(uint32_t id, const char __user *name,
20009 + uint32_t *space_used, uint32_t *space_total,
20010 + uint32_t *inodes_used, uint32_t *inodes_total,
20011 + uint32_t *reserved, uint32_t *flags)
20013 + struct path path;
20016 + ret = user_lpath(name, &path);
20018 + struct super_block *sb;
20019 + struct dl_info *dli;
20022 + if (!path.dentry->d_inode)
20023 + goto out_release;
20024 + if (!(sb = path.dentry->d_inode->i_sb))
20025 + goto out_release;
20028 + dli = locate_dl_info(sb, id);
20030 + goto out_release;
20032 + spin_lock(&dli->dl_lock);
20033 + *inodes_used = dli->dl_inodes_used;
20034 + *inodes_total = dli->dl_inodes_total;
20036 + *space_used = dlimit_space_64to32(
20037 + dli->dl_space_used, flags, DLIMS_USED);
20039 + if (dli->dl_space_total == DLIM_INFINITY)
20040 + *space_total = CDLIM_INFINITY;
20042 + *space_total = dlimit_space_64to32(
20043 + dli->dl_space_total, flags, DLIMS_TOTAL);
20045 + *reserved = 100 - ((dli->dl_nrlmult * 100 + 512) >> 10);
20046 + spin_unlock(&dli->dl_lock);
20048 + put_dl_info(dli);
20059 +int vc_get_dlimit(uint32_t id, void __user *data)
20061 + struct vcmd_ctx_dlimit_v0 vc_data;
20064 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20067 + ret = do_get_dlimit(id, vc_data.name,
20068 + &vc_data.space_used, &vc_data.space_total,
20069 + &vc_data.inodes_used, &vc_data.inodes_total,
20070 + &vc_data.reserved, &vc_data.flags);
20074 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20079 +#ifdef CONFIG_COMPAT
20081 +int vc_get_dlimit_x32(uint32_t id, void __user *data)
20083 + struct vcmd_ctx_dlimit_v0_x32 vc_data;
20086 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20089 + ret = do_get_dlimit(id, compat_ptr(vc_data.name_ptr),
20090 + &vc_data.space_used, &vc_data.space_total,
20091 + &vc_data.inodes_used, &vc_data.inodes_total,
20092 + &vc_data.reserved, &vc_data.flags);
20096 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20101 +#endif /* CONFIG_COMPAT */
20104 +void vx_vsi_statfs(struct super_block *sb, struct kstatfs *buf)
20106 + struct dl_info *dli;
20107 + __u64 blimit, bfree, bavail;
20110 + dli = locate_dl_info(sb, dx_current_tag());
20114 + spin_lock(&dli->dl_lock);
20115 + if (dli->dl_inodes_total == (unsigned long)DLIM_INFINITY)
20118 + /* reduce max inodes available to limit */
20119 + if (buf->f_files > dli->dl_inodes_total)
20120 + buf->f_files = dli->dl_inodes_total;
20122 + ifree = dli->dl_inodes_total - dli->dl_inodes_used;
20123 + /* reduce free inodes to min */
20124 + if (ifree < buf->f_ffree)
20125 + buf->f_ffree = ifree;
20128 + if (dli->dl_space_total == DLIM_INFINITY)
20131 + blimit = dli->dl_space_total >> sb->s_blocksize_bits;
20133 + if (dli->dl_space_total < dli->dl_space_used)
20136 + bfree = (dli->dl_space_total - dli->dl_space_used)
20137 + >> sb->s_blocksize_bits;
20139 + bavail = ((dli->dl_space_total >> 10) * dli->dl_nrlmult);
20140 + if (bavail < dli->dl_space_used)
20143 + bavail = (bavail - dli->dl_space_used)
20144 + >> sb->s_blocksize_bits;
20146 + /* reduce max space available to limit */
20147 + if (buf->f_blocks > blimit)
20148 + buf->f_blocks = blimit;
20150 + /* reduce free space to min */
20151 + if (bfree < buf->f_bfree)
20152 + buf->f_bfree = bfree;
20154 + /* reduce avail space to min */
20155 + if (bavail < buf->f_bavail)
20156 + buf->f_bavail = bavail;
20159 + spin_unlock(&dli->dl_lock);
20160 + put_dl_info(dli);
20165 +#include <linux/module.h>
20167 +EXPORT_SYMBOL_GPL(locate_dl_info);
20168 +EXPORT_SYMBOL_GPL(rcu_free_dl_info);
20170 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/helper.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/helper.c
20171 --- linux-2.6.32.1/kernel/vserver/helper.c 1970-01-01 01:00:00.000000000 +0100
20172 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/helper.c 2009-12-03 22:20:22.000000000 +0100
20175 + * linux/kernel/vserver/helper.c
20177 + * Virtual Context Support
20179 + * Copyright (C) 2004-2007 Herbert Pötzl
20181 + * V0.01 basic helper
20185 +#include <linux/kmod.h>
20186 +#include <linux/reboot.h>
20187 +#include <linux/vs_context.h>
20188 +#include <linux/vs_network.h>
20189 +#include <linux/vserver/signal.h>
20192 +char vshelper_path[255] = "/sbin/vshelper";
20195 +static int do_vshelper(char *name, char *argv[], char *envp[], int sync)
20199 + if ((ret = call_usermodehelper(name, argv, envp, sync))) {
20200 + printk( KERN_WARNING
20201 + "%s: (%s %s) returned %s with %d\n",
20202 + name, argv[1], argv[2],
20203 + sync ? "sync" : "async", ret);
20205 + vxdprintk(VXD_CBIT(switch, 4),
20206 + "%s: (%s %s) returned %s with %d",
20207 + name, argv[1], argv[2], sync ? "sync" : "async", ret);
20212 + * vshelper path is set via /proc/sys
20213 + * invoked by vserver sys_reboot(), with
20214 + * the following arguments
20216 + * argv [0] = vshelper_path;
20217 + * argv [1] = action: "restart", "halt", "poweroff", ...
20218 + * argv [2] = context identifier
20220 + * envp [*] = type-specific parameters
20223 +long vs_reboot_helper(struct vx_info *vxi, int cmd, void __user *arg)
20225 + char id_buf[8], cmd_buf[16];
20226 + char uid_buf[16], pid_buf[16];
20229 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
20230 + char *envp[] = {"HOME=/", "TERM=linux",
20231 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
20232 + uid_buf, pid_buf, cmd_buf, 0};
20234 + if (vx_info_state(vxi, VXS_HELPER))
20236 + vxi->vx_state |= VXS_HELPER;
20238 + snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
20240 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
20241 + snprintf(uid_buf, sizeof(uid_buf)-1, "VS_UID=%d", current_uid());
20242 + snprintf(pid_buf, sizeof(pid_buf)-1, "VS_PID=%d", current->pid);
20245 + case LINUX_REBOOT_CMD_RESTART:
20246 + argv[1] = "restart";
20249 + case LINUX_REBOOT_CMD_HALT:
20250 + argv[1] = "halt";
20253 + case LINUX_REBOOT_CMD_POWER_OFF:
20254 + argv[1] = "poweroff";
20257 + case LINUX_REBOOT_CMD_SW_SUSPEND:
20258 + argv[1] = "swsusp";
20261 + case LINUX_REBOOT_CMD_OOM:
20266 + vxi->vx_state &= ~VXS_HELPER;
20270 + ret = do_vshelper(vshelper_path, argv, envp, 0);
20271 + vxi->vx_state &= ~VXS_HELPER;
20272 + __wakeup_vx_info(vxi);
20273 + return (ret) ? -EPERM : 0;
20277 +long vs_reboot(unsigned int cmd, void __user *arg)
20279 + struct vx_info *vxi = current_vx_info();
20282 + vxdprintk(VXD_CBIT(misc, 5),
20283 + "vs_reboot(%p[#%d],%u)",
20284 + vxi, vxi ? vxi->vx_id : 0, cmd);
20286 + ret = vs_reboot_helper(vxi, cmd, arg);
20290 + vxi->reboot_cmd = cmd;
20291 + if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
20293 + case LINUX_REBOOT_CMD_RESTART:
20294 + case LINUX_REBOOT_CMD_HALT:
20295 + case LINUX_REBOOT_CMD_POWER_OFF:
20296 + vx_info_kill(vxi, 0, SIGKILL);
20297 + vx_info_kill(vxi, 1, SIGKILL);
20305 +long vs_oom_action(unsigned int cmd)
20307 + struct vx_info *vxi = current_vx_info();
20310 + vxdprintk(VXD_CBIT(misc, 5),
20311 + "vs_oom_action(%p[#%d],%u)",
20312 + vxi, vxi ? vxi->vx_id : 0, cmd);
20314 + ret = vs_reboot_helper(vxi, cmd, NULL);
20318 + vxi->reboot_cmd = cmd;
20319 + if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
20320 + vx_info_kill(vxi, 0, SIGKILL);
20321 + vx_info_kill(vxi, 1, SIGKILL);
20327 + * argv [0] = vshelper_path;
20328 + * argv [1] = action: "startup", "shutdown"
20329 + * argv [2] = context identifier
20331 + * envp [*] = type-specific parameters
20334 +long vs_state_change(struct vx_info *vxi, unsigned int cmd)
20336 + char id_buf[8], cmd_buf[16];
20337 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
20338 + char *envp[] = {"HOME=/", "TERM=linux",
20339 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
20341 + if (!vx_info_flags(vxi, VXF_SC_HELPER, 0))
20344 + snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
20345 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
20348 + case VSC_STARTUP:
20349 + argv[1] = "startup";
20351 + case VSC_SHUTDOWN:
20352 + argv[1] = "shutdown";
20358 + return do_vshelper(vshelper_path, argv, envp, 1);
20363 + * argv [0] = vshelper_path;
20364 + * argv [1] = action: "netup", "netdown"
20365 + * argv [2] = context identifier
20367 + * envp [*] = type-specific parameters
20370 +long vs_net_change(struct nx_info *nxi, unsigned int cmd)
20372 + char id_buf[8], cmd_buf[16];
20373 + char *argv[] = {vshelper_path, NULL, id_buf, 0};
20374 + char *envp[] = {"HOME=/", "TERM=linux",
20375 + "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
20377 + if (!nx_info_flags(nxi, NXF_SC_HELPER, 0))
20380 + snprintf(id_buf, sizeof(id_buf)-1, "%d", nxi->nx_id);
20381 + snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
20385 + argv[1] = "netup";
20387 + case VSC_NETDOWN:
20388 + argv[1] = "netdown";
20394 + return do_vshelper(vshelper_path, argv, envp, 1);
20397 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/history.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/history.c
20398 --- linux-2.6.32.1/kernel/vserver/history.c 1970-01-01 01:00:00.000000000 +0100
20399 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/history.c 2009-12-03 20:04:56.000000000 +0100
20402 + * kernel/vserver/history.c
20404 + * Virtual Context History Backtrace
20406 + * Copyright (C) 2004-2007 Herbert Pötzl
20408 + * V0.01 basic structure
20409 + * V0.02 hash/unhash and trace
20410 + * V0.03 preemption fixes
20414 +#include <linux/module.h>
20415 +#include <asm/uaccess.h>
20417 +#include <linux/vserver/context.h>
20418 +#include <linux/vserver/debug.h>
20419 +#include <linux/vserver/debug_cmd.h>
20420 +#include <linux/vserver/history.h>
20423 +#ifdef CONFIG_VSERVER_HISTORY
20424 +#define VXH_SIZE CONFIG_VSERVER_HISTORY_SIZE
20426 +#define VXH_SIZE 64
20429 +struct _vx_history {
20430 + unsigned int counter;
20432 + struct _vx_hist_entry entry[VXH_SIZE + 1];
20436 +DEFINE_PER_CPU(struct _vx_history, vx_history_buffer);
20438 +unsigned volatile int vxh_active = 1;
20440 +static atomic_t sequence = ATOMIC_INIT(0);
20445 + * requires disabled preemption */
20447 +struct _vx_hist_entry *vxh_advance(void *loc)
20449 + unsigned int cpu = smp_processor_id();
20450 + struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
20451 + struct _vx_hist_entry *entry;
20452 + unsigned int index;
20454 + index = vxh_active ? (hist->counter++ % VXH_SIZE) : VXH_SIZE;
20455 + entry = &hist->entry[index];
20457 + entry->seq = atomic_inc_return(&sequence);
20458 + entry->loc = loc;
20462 +EXPORT_SYMBOL_GPL(vxh_advance);
20465 +#define VXH_LOC_FMTS "(#%04x,*%d):%p"
20467 +#define VXH_LOC_ARGS(e) (e)->seq, cpu, (e)->loc
20470 +#define VXH_VXI_FMTS "%p[#%d,%d.%d]"
20472 +#define VXH_VXI_ARGS(e) (e)->vxi.ptr, \
20473 + (e)->vxi.ptr ? (e)->vxi.xid : 0, \
20474 + (e)->vxi.ptr ? (e)->vxi.usecnt : 0, \
20475 + (e)->vxi.ptr ? (e)->vxi.tasks : 0
20477 +void vxh_dump_entry(struct _vx_hist_entry *e, unsigned cpu)
20479 + switch (e->type) {
20480 + case VXH_THROW_OOPS:
20481 + printk( VXH_LOC_FMTS " oops \n", VXH_LOC_ARGS(e));
20484 + case VXH_GET_VX_INFO:
20485 + case VXH_PUT_VX_INFO:
20486 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
20488 + (e->type == VXH_GET_VX_INFO) ? "get" : "put",
20489 + VXH_VXI_ARGS(e));
20492 + case VXH_INIT_VX_INFO:
20493 + case VXH_SET_VX_INFO:
20494 + case VXH_CLR_VX_INFO:
20495 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
20497 + (e->type == VXH_INIT_VX_INFO) ? "init" :
20498 + ((e->type == VXH_SET_VX_INFO) ? "set" : "clr"),
20499 + VXH_VXI_ARGS(e), e->sc.data);
20502 + case VXH_CLAIM_VX_INFO:
20503 + case VXH_RELEASE_VX_INFO:
20504 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
20506 + (e->type == VXH_CLAIM_VX_INFO) ? "claim" : "release",
20507 + VXH_VXI_ARGS(e), e->sc.data);
20510 + case VXH_ALLOC_VX_INFO:
20511 + case VXH_DEALLOC_VX_INFO:
20512 + printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
20514 + (e->type == VXH_ALLOC_VX_INFO) ? "alloc" : "dealloc",
20515 + VXH_VXI_ARGS(e));
20518 + case VXH_HASH_VX_INFO:
20519 + case VXH_UNHASH_VX_INFO:
20520 + printk( VXH_LOC_FMTS " __%s_vx_info " VXH_VXI_FMTS "\n",
20522 + (e->type == VXH_HASH_VX_INFO) ? "hash" : "unhash",
20523 + VXH_VXI_ARGS(e));
20526 + case VXH_LOC_VX_INFO:
20527 + case VXH_LOOKUP_VX_INFO:
20528 + case VXH_CREATE_VX_INFO:
20529 + printk( VXH_LOC_FMTS " __%s_vx_info [#%d] -> " VXH_VXI_FMTS "\n",
20531 + (e->type == VXH_CREATE_VX_INFO) ? "create" :
20532 + ((e->type == VXH_LOC_VX_INFO) ? "loc" : "lookup"),
20533 + e->ll.arg, VXH_VXI_ARGS(e));
20538 +static void __vxh_dump_history(void)
20540 + unsigned int i, cpu;
20542 + printk("History:\tSEQ: %8x\tNR_CPUS: %d\n",
20543 + atomic_read(&sequence), NR_CPUS);
20545 + for (i = 0; i < VXH_SIZE; i++) {
20546 + for_each_online_cpu(cpu) {
20547 + struct _vx_history *hist =
20548 + &per_cpu(vx_history_buffer, cpu);
20549 + unsigned int index = (hist->counter - i) % VXH_SIZE;
20550 + struct _vx_hist_entry *entry = &hist->entry[index];
20552 + vxh_dump_entry(entry, cpu);
20557 +void vxh_dump_history(void)
20561 + local_irq_enable();
20563 + local_irq_disable();
20565 + __vxh_dump_history();
20569 +/* vserver syscall commands below here */
20572 +int vc_dump_history(uint32_t id)
20575 + __vxh_dump_history();
20582 +int do_read_history(struct __user _vx_hist_entry *data,
20583 + int cpu, uint32_t *index, uint32_t *count)
20585 + int pos, ret = 0;
20586 + struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
20587 + int end = hist->counter;
20588 + int start = end - VXH_SIZE + 2;
20589 + int idx = *index;
20591 + /* special case: get current pos */
20597 + /* have we lost some data? */
20601 + for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
20602 + struct _vx_hist_entry *entry =
20603 + &hist->entry[idx % VXH_SIZE];
20605 + /* send entry to userspace */
20606 + ret = copy_to_user(&data[pos], entry, sizeof(*entry));
20610 + /* save new index and count */
20613 + return ret ? ret : (*index < end);
20616 +int vc_read_history(uint32_t id, void __user *data)
20618 + struct vcmd_read_history_v0 vc_data;
20621 + if (id >= NR_CPUS)
20624 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20627 + ret = do_read_history((struct __user _vx_hist_entry *)vc_data.data,
20628 + id, &vc_data.index, &vc_data.count);
20630 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20635 +#ifdef CONFIG_COMPAT
20637 +int vc_read_history_x32(uint32_t id, void __user *data)
20639 + struct vcmd_read_history_v0_x32 vc_data;
20642 + if (id >= NR_CPUS)
20645 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20648 + ret = do_read_history((struct __user _vx_hist_entry *)
20649 + compat_ptr(vc_data.data_ptr),
20650 + id, &vc_data.index, &vc_data.count);
20652 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20657 +#endif /* CONFIG_COMPAT */
20659 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/inet.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/inet.c
20660 --- linux-2.6.32.1/kernel/vserver/inet.c 1970-01-01 01:00:00.000000000 +0100
20661 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/inet.c 2009-12-03 20:04:56.000000000 +0100
20664 +#include <linux/in.h>
20665 +#include <linux/inetdevice.h>
20666 +#include <linux/vs_inet.h>
20667 +#include <linux/vs_inet6.h>
20668 +#include <linux/vserver/debug.h>
20669 +#include <net/route.h>
20670 +#include <net/addrconf.h>
20673 +int nx_v4_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20677 + if (!nxi1 || !nxi2 || nxi1 == nxi2)
20680 + struct nx_addr_v4 *ptr;
20682 + for (ptr = &nxi1->v4; ptr; ptr = ptr->next) {
20683 + if (v4_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20690 + vxdprintk(VXD_CBIT(net, 2),
20691 + "nx_v4_addr_conflict(%p,%p): %d",
20692 + nxi1, nxi2, ret);
20698 +#ifdef CONFIG_IPV6
20700 +int nx_v6_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20704 + if (!nxi1 || !nxi2 || nxi1 == nxi2)
20707 + struct nx_addr_v6 *ptr;
20709 + for (ptr = &nxi1->v6; ptr; ptr = ptr->next) {
20710 + if (v6_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20717 + vxdprintk(VXD_CBIT(net, 2),
20718 + "nx_v6_addr_conflict(%p,%p): %d",
20719 + nxi1, nxi2, ret);
20726 +int v4_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20728 + struct in_device *in_dev;
20729 + struct in_ifaddr **ifap;
20730 + struct in_ifaddr *ifa;
20735 + in_dev = in_dev_get(dev);
20739 + for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
20740 + ifap = &ifa->ifa_next) {
20741 + if (v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW)) {
20746 + in_dev_put(in_dev);
20752 +#ifdef CONFIG_IPV6
20754 +int v6_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20756 + struct inet6_dev *in_dev;
20757 + struct inet6_ifaddr **ifap;
20758 + struct inet6_ifaddr *ifa;
20763 + in_dev = in6_dev_get(dev);
20767 + for (ifap = &in_dev->addr_list; (ifa = *ifap) != NULL;
20768 + ifap = &ifa->if_next) {
20769 + if (v6_addr_in_nx_info(nxi, &ifa->addr, -1)) {
20774 + in6_dev_put(in_dev);
20781 +int dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20787 + if (nxi->v4.type && v4_dev_in_nx_info(dev, nxi))
20789 +#ifdef CONFIG_IPV6
20791 + if (nxi->v6.type && v6_dev_in_nx_info(dev, nxi))
20796 + vxdprintk(VXD_CBIT(net, 3),
20797 + "dev_in_nx_info(%p,%p[#%d]) = %d",
20798 + dev, nxi, nxi ? nxi->nx_id : 0, ret);
20802 +int ip_v4_find_src(struct net *net, struct nx_info *nxi,
20803 + struct rtable **rp, struct flowi *fl)
20808 + /* FIXME: handle lback only case */
20809 + if (!NX_IPV4(nxi))
20812 + vxdprintk(VXD_CBIT(net, 4),
20813 + "ip_v4_find_src(%p[#%u]) " NIPQUAD_FMT " -> " NIPQUAD_FMT,
20814 + nxi, nxi ? nxi->nx_id : 0,
20815 + NIPQUAD(fl->fl4_src), NIPQUAD(fl->fl4_dst));
20817 + /* single IP is unconditional */
20818 + if (nx_info_flags(nxi, NXF_SINGLE_IP, 0) &&
20819 + (fl->fl4_src == INADDR_ANY))
20820 + fl->fl4_src = nxi->v4.ip[0].s_addr;
20822 + if (fl->fl4_src == INADDR_ANY) {
20823 + struct nx_addr_v4 *ptr;
20824 + __be32 found = 0;
20827 + err = __ip_route_output_key(net, rp, fl);
20829 + found = (*rp)->rt_src;
20831 + vxdprintk(VXD_CBIT(net, 4),
20832 + "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20833 + nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(found));
20834 + if (v4_addr_in_nx_info(nxi, found, NXA_MASK_BIND))
20838 + for (ptr = &nxi->v4; ptr; ptr = ptr->next) {
20839 + __be32 primary = ptr->ip[0].s_addr;
20840 + __be32 mask = ptr->mask.s_addr;
20841 + __be32 neta = primary & mask;
20843 + vxdprintk(VXD_CBIT(net, 4), "ip_v4_find_src(%p[#%u]) chk: "
20844 + NIPQUAD_FMT "/" NIPQUAD_FMT "/" NIPQUAD_FMT,
20845 + nxi, nxi ? nxi->nx_id : 0, NIPQUAD(primary),
20846 + NIPQUAD(mask), NIPQUAD(neta));
20847 + if ((found & mask) != neta)
20850 + fl->fl4_src = primary;
20851 + err = __ip_route_output_key(net, rp, fl);
20852 + vxdprintk(VXD_CBIT(net, 4),
20853 + "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20854 + nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(primary));
20856 + found = (*rp)->rt_src;
20858 + if (found == primary)
20862 + /* still no source ip? */
20863 + found = ipv4_is_loopback(fl->fl4_dst)
20864 + ? IPI_LOOPBACK : nxi->v4.ip[0].s_addr;
20866 + /* assign src ip to flow */
20867 + fl->fl4_src = found;
20870 + if (!v4_addr_in_nx_info(nxi, fl->fl4_src, NXA_MASK_BIND))
20874 + if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0)) {
20875 + if (ipv4_is_loopback(fl->fl4_dst))
20876 + fl->fl4_dst = nxi->v4_lback.s_addr;
20877 + if (ipv4_is_loopback(fl->fl4_src))
20878 + fl->fl4_src = nxi->v4_lback.s_addr;
20879 + } else if (ipv4_is_loopback(fl->fl4_dst) &&
20880 + !nx_info_flags(nxi, NXF_LBACK_ALLOW, 0))
20886 +EXPORT_SYMBOL_GPL(ip_v4_find_src);
20888 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/init.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/init.c
20889 --- linux-2.6.32.1/kernel/vserver/init.c 1970-01-01 01:00:00.000000000 +0100
20890 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/init.c 2009-12-03 20:04:56.000000000 +0100
20893 + * linux/kernel/init.c
20895 + * Virtual Server Init
20897 + * Copyright (C) 2004-2007 Herbert Pötzl
20899 + * V0.01 basic structure
20903 +#include <linux/init.h>
20905 +int vserver_register_sysctl(void);
20906 +void vserver_unregister_sysctl(void);
20909 +static int __init init_vserver(void)
20913 +#ifdef CONFIG_VSERVER_DEBUG
20914 + vserver_register_sysctl();
20920 +static void __exit exit_vserver(void)
20923 +#ifdef CONFIG_VSERVER_DEBUG
20924 + vserver_unregister_sysctl();
20929 +/* FIXME: GFP_ZONETYPES gone
20930 +long vx_slab[GFP_ZONETYPES]; */
20934 +module_init(init_vserver);
20935 +module_exit(exit_vserver);
20937 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/inode.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/inode.c
20938 --- linux-2.6.32.1/kernel/vserver/inode.c 1970-01-01 01:00:00.000000000 +0100
20939 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/inode.c 2009-12-03 20:04:56.000000000 +0100
20942 + * linux/kernel/vserver/inode.c
20944 + * Virtual Server: File System Support
20946 + * Copyright (C) 2004-2007 Herbert Pötzl
20948 + * V0.01 separated from vcontext V0.05
20949 + * V0.02 moved to tag (instead of xid)
20953 +#include <linux/tty.h>
20954 +#include <linux/proc_fs.h>
20955 +#include <linux/devpts_fs.h>
20956 +#include <linux/fs.h>
20957 +#include <linux/file.h>
20958 +#include <linux/mount.h>
20959 +#include <linux/parser.h>
20960 +#include <linux/namei.h>
20961 +#include <linux/vserver/inode.h>
20962 +#include <linux/vserver/inode_cmd.h>
20963 +#include <linux/vs_base.h>
20964 +#include <linux/vs_tag.h>
20966 +#include <asm/uaccess.h>
20969 +static int __vc_get_iattr(struct inode *in, uint32_t *tag, uint32_t *flags, uint32_t *mask)
20971 + struct proc_dir_entry *entry;
20973 + if (!in || !in->i_sb)
20976 + *flags = IATTR_TAG
20977 + | (IS_IMMUTABLE(in) ? IATTR_IMMUTABLE : 0)
20978 + | (IS_IXUNLINK(in) ? IATTR_IXUNLINK : 0)
20979 + | (IS_BARRIER(in) ? IATTR_BARRIER : 0)
20980 + | (IS_COW(in) ? IATTR_COW : 0);
20981 + *mask = IATTR_IXUNLINK | IATTR_IMMUTABLE | IATTR_COW;
20983 + if (S_ISDIR(in->i_mode))
20984 + *mask |= IATTR_BARRIER;
20986 + if (IS_TAGGED(in)) {
20987 + *tag = in->i_tag;
20988 + *mask |= IATTR_TAG;
20991 + switch (in->i_sb->s_magic) {
20992 + case PROC_SUPER_MAGIC:
20993 + entry = PROC_I(in)->pde;
20995 + /* check for specific inodes? */
20997 + *mask |= IATTR_FLAGS;
20999 + *flags |= (entry->vx_flags & IATTR_FLAGS);
21001 + *flags |= (PROC_I(in)->vx_flags & IATTR_FLAGS);
21004 + case DEVPTS_SUPER_MAGIC:
21005 + *tag = in->i_tag;
21006 + *mask |= IATTR_TAG;
21015 +int vc_get_iattr(void __user *data)
21017 + struct path path;
21018 + struct vcmd_ctx_iattr_v1 vc_data = { .tag = -1 };
21021 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21024 + ret = user_lpath(vc_data.name, &path);
21026 + ret = __vc_get_iattr(path.dentry->d_inode,
21027 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
21033 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21038 +#ifdef CONFIG_COMPAT
21040 +int vc_get_iattr_x32(void __user *data)
21042 + struct path path;
21043 + struct vcmd_ctx_iattr_v1_x32 vc_data = { .tag = -1 };
21046 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21049 + ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
21051 + ret = __vc_get_iattr(path.dentry->d_inode,
21052 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
21058 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21063 +#endif /* CONFIG_COMPAT */
21066 +int vc_fget_iattr(uint32_t fd, void __user *data)
21068 + struct file *filp;
21069 + struct vcmd_ctx_fiattr_v0 vc_data = { .tag = -1 };
21072 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21076 + if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
21079 + ret = __vc_get_iattr(filp->f_dentry->d_inode,
21080 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
21084 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21090 +static int __vc_set_iattr(struct dentry *de, uint32_t *tag, uint32_t *flags, uint32_t *mask)
21092 + struct inode *in = de->d_inode;
21093 + int error = 0, is_proc = 0, has_tag = 0;
21094 + struct iattr attr = { 0 };
21096 + if (!in || !in->i_sb)
21099 + is_proc = (in->i_sb->s_magic == PROC_SUPER_MAGIC);
21100 + if ((*mask & IATTR_FLAGS) && !is_proc)
21103 + has_tag = IS_TAGGED(in) ||
21104 + (in->i_sb->s_magic == DEVPTS_SUPER_MAGIC);
21105 + if ((*mask & IATTR_TAG) && !has_tag)
21108 + mutex_lock(&in->i_mutex);
21109 + if (*mask & IATTR_TAG) {
21110 + attr.ia_tag = *tag;
21111 + attr.ia_valid |= ATTR_TAG;
21114 + if (*mask & IATTR_FLAGS) {
21115 + struct proc_dir_entry *entry = PROC_I(in)->pde;
21116 + unsigned int iflags = PROC_I(in)->vx_flags;
21118 + iflags = (iflags & ~(*mask & IATTR_FLAGS))
21119 + | (*flags & IATTR_FLAGS);
21120 + PROC_I(in)->vx_flags = iflags;
21122 + entry->vx_flags = iflags;
21125 + if (*mask & (IATTR_IMMUTABLE | IATTR_IXUNLINK |
21126 + IATTR_BARRIER | IATTR_COW)) {
21127 + int iflags = in->i_flags;
21128 + int vflags = in->i_vflags;
21130 + if (*mask & IATTR_IMMUTABLE) {
21131 + if (*flags & IATTR_IMMUTABLE)
21132 + iflags |= S_IMMUTABLE;
21134 + iflags &= ~S_IMMUTABLE;
21136 + if (*mask & IATTR_IXUNLINK) {
21137 + if (*flags & IATTR_IXUNLINK)
21138 + iflags |= S_IXUNLINK;
21140 + iflags &= ~S_IXUNLINK;
21142 + if (S_ISDIR(in->i_mode) && (*mask & IATTR_BARRIER)) {
21143 + if (*flags & IATTR_BARRIER)
21144 + vflags |= V_BARRIER;
21146 + vflags &= ~V_BARRIER;
21148 + if (S_ISREG(in->i_mode) && (*mask & IATTR_COW)) {
21149 + if (*flags & IATTR_COW)
21152 + vflags &= ~V_COW;
21154 + if (in->i_op && in->i_op->sync_flags) {
21155 + error = in->i_op->sync_flags(in, iflags, vflags);
21161 + if (attr.ia_valid) {
21162 + if (in->i_op && in->i_op->setattr)
21163 + error = in->i_op->setattr(de, &attr);
21165 + error = inode_change_ok(in, &attr);
21167 + error = inode_setattr(in, &attr);
21172 + mutex_unlock(&in->i_mutex);
21176 +int vc_set_iattr(void __user *data)
21178 + struct path path;
21179 + struct vcmd_ctx_iattr_v1 vc_data;
21182 + if (!capable(CAP_LINUX_IMMUTABLE))
21184 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21187 + ret = user_lpath(vc_data.name, &path);
21189 + ret = __vc_set_iattr(path.dentry,
21190 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
21194 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21199 +#ifdef CONFIG_COMPAT
21201 +int vc_set_iattr_x32(void __user *data)
21203 + struct path path;
21204 + struct vcmd_ctx_iattr_v1_x32 vc_data;
21207 + if (!capable(CAP_LINUX_IMMUTABLE))
21209 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21212 + ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
21214 + ret = __vc_set_iattr(path.dentry,
21215 + &vc_data.tag, &vc_data.flags, &vc_data.mask);
21219 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21224 +#endif /* CONFIG_COMPAT */
21226 +int vc_fset_iattr(uint32_t fd, void __user *data)
21228 + struct file *filp;
21229 + struct vcmd_ctx_fiattr_v0 vc_data;
21232 + if (!capable(CAP_LINUX_IMMUTABLE))
21234 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21238 + if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
21241 + ret = __vc_set_iattr(filp->f_dentry, &vc_data.tag,
21242 + &vc_data.flags, &vc_data.mask);
21246 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21252 +enum { Opt_notagcheck, Opt_tag, Opt_notag, Opt_tagid, Opt_err };
21254 +static match_table_t tokens = {
21255 + {Opt_notagcheck, "notagcheck"},
21256 +#ifdef CONFIG_PROPAGATE
21257 + {Opt_notag, "notag"},
21258 + {Opt_tag, "tag"},
21259 + {Opt_tagid, "tagid=%u"},
21265 +static void __dx_parse_remove(char *string, char *opt)
21267 + char *p = strstr(string, opt);
21271 + while (*q != '\0' && *q != ',')
21280 +int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
21281 + unsigned long *flags)
21284 + substring_t args[MAX_OPT_ARGS];
21285 + int token, option = 0;
21286 + char *s, *p, *opts;
21290 + s = kstrdup(string, GFP_KERNEL | GFP_ATOMIC);
21295 + while ((p = strsep(&opts, ",")) != NULL) {
21296 + token = match_token(p, tokens, args);
21298 + vxdprintk(VXD_CBIT(tag, 7),
21299 + "dx_parse_tag(»%s«): %d:#%d",
21300 + p, token, option);
21303 +#ifdef CONFIG_PROPAGATE
21308 + __dx_parse_remove(s, "tag");
21309 + *mnt_flags |= MNT_TAGID;
21310 + set |= MNT_TAGID;
21314 + __dx_parse_remove(s, "notag");
21315 + *mnt_flags |= MNT_NOTAG;
21316 + set |= MNT_NOTAG;
21319 + if (tag && !match_int(args, &option))
21322 + __dx_parse_remove(s, "tagid");
21323 + *mnt_flags |= MNT_TAGID;
21324 + set |= MNT_TAGID;
21327 + case Opt_notagcheck:
21329 + __dx_parse_remove(s, "notagcheck");
21330 + *flags |= MS_NOTAGCHECK;
21331 + set |= MS_NOTAGCHECK;
21336 + strcpy(string, s);
21341 +#ifdef CONFIG_PROPAGATE
21343 +void __dx_propagate_tag(struct nameidata *nd, struct inode *inode)
21345 + tag_t new_tag = 0;
21346 + struct vfsmount *mnt;
21351 + mnt = nd->path.mnt;
21355 + propagate = (mnt->mnt_flags & MNT_TAGID);
21357 + new_tag = mnt->mnt_tag;
21359 + vxdprintk(VXD_CBIT(tag, 7),
21360 + "dx_propagate_tag(%p[#%lu.%d]): %d,%d",
21361 + inode, inode->i_ino, inode->i_tag,
21362 + new_tag, (propagate) ? 1 : 0);
21365 + inode->i_tag = new_tag;
21368 +#include <linux/module.h>
21370 +EXPORT_SYMBOL_GPL(__dx_propagate_tag);
21372 +#endif /* CONFIG_PROPAGATE */
21374 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/Kconfig linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/Kconfig
21375 --- linux-2.6.32.1/kernel/vserver/Kconfig 1970-01-01 01:00:00.000000000 +0100
21376 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/Kconfig 2009-12-03 20:04:56.000000000 +0100
21379 +# Linux VServer configuration
21382 +menu "Linux VServer"
21384 +config VSERVER_AUTO_LBACK
21385 + bool "Automatically Assign Loopback IP"
21388 + Automatically assign a guest specific loopback
21389 + IP and add it to the kernel network stack on
21392 +config VSERVER_AUTO_SINGLE
21393 + bool "Automatic Single IP Special Casing"
21394 + depends on EXPERIMENTAL
21397 + This allows network contexts with a single IP to
21398 + automatically remap 0.0.0.0 bindings to that IP,
21399 + avoiding further network checks and improving
21402 + (note: such guests do not allow to change the ip
21403 + on the fly and do not show loopback addresses)
21405 +config VSERVER_COWBL
21406 + bool "Enable COW Immutable Link Breaking"
21409 + This enables the COW (Copy-On-Write) link break code.
21410 + It allows you to treat unified files like normal files
21411 + when writing to them (which will implicitely break the
21412 + link and create a copy of the unified file)
21414 +config VSERVER_VTIME
21415 + bool "Enable Virtualized Guest Time"
21416 + depends on EXPERIMENTAL
21419 + This enables per guest time offsets to allow for
21420 + adjusting the system clock individually per guest.
21421 + this adds some overhead to the time functions and
21422 + therefore should not be enabled without good reason.
21424 +config VSERVER_DEVICE
21425 + bool "Enable Guest Device Mapping"
21426 + depends on EXPERIMENTAL
21429 + This enables generic device remapping.
21431 +config VSERVER_PROC_SECURE
21432 + bool "Enable Proc Security"
21433 + depends on PROC_FS
21436 + This configures ProcFS security to initially hide
21437 + non-process entries for all contexts except the main and
21438 + spectator context (i.e. for all guests), which is a secure
21441 + (note: on 1.2x the entries were visible by default)
21443 +config VSERVER_HARDCPU
21444 + bool "Enable Hard CPU Limits"
21447 + Activate the Hard CPU Limits
21449 + This will compile in code that allows the Token Bucket
21450 + Scheduler to put processes on hold when a context's
21451 + tokens are depleted (provided that its per-context
21452 + sched_hard flag is set).
21454 + Processes belonging to that context will not be able
21455 + to consume CPU resources again until a per-context
21456 + configured minimum of tokens has been reached.
21458 +config VSERVER_IDLETIME
21459 + bool "Avoid idle CPUs by skipping Time"
21460 + depends on VSERVER_HARDCPU
21463 + This option allows the scheduler to artificially
21464 + advance time (per cpu) when otherwise the idle
21465 + task would be scheduled, thus keeping the cpu
21466 + busy and sharing the available resources among
21467 + certain contexts.
21469 +config VSERVER_IDLELIMIT
21470 + bool "Limit the IDLE task"
21471 + depends on VSERVER_HARDCPU
21474 + Limit the idle slices, so the the next context
21475 + will be scheduled as soon as possible.
21477 + This might improve interactivity and latency, but
21478 + will also marginally increase scheduling overhead.
21481 + prompt "Persistent Inode Tagging"
21482 + default TAGGING_ID24
21484 + This adds persistent context information to filesystems
21485 + mounted with the tagxid option. Tagging is a requirement
21486 + for per-context disk limits and per-context quota.
21489 +config TAGGING_NONE
21492 + do not store per-context information in inodes.
21494 +config TAGGING_UID16
21495 + bool "UID16/GID32"
21497 + reduces UID to 16 bit, but leaves GID at 32 bit.
21499 +config TAGGING_GID16
21500 + bool "UID32/GID16"
21502 + reduces GID to 16 bit, but leaves UID at 32 bit.
21504 +config TAGGING_ID24
21505 + bool "UID24/GID24"
21507 + uses the upper 8bit from UID and GID for XID tagging
21508 + which leaves 24bit for UID/GID each, which should be
21509 + more than sufficient for normal use.
21511 +config TAGGING_INTERN
21512 + bool "UID32/GID32"
21514 + this uses otherwise reserved inode fields in the on
21515 + disk representation, which limits the use to a few
21516 + filesystems (currently ext2 and ext3)
21521 + bool "Tag NFSD User Auth and Files"
21524 + Enable this if you do want the in-kernel NFS
21525 + Server to use the tagging specified above.
21526 + (will require patched clients too)
21528 +config VSERVER_PRIVACY
21529 + bool "Honor Privacy Aspects of Guests"
21532 + When enabled, most context checks will disallow
21533 + access to structures assigned to a specific context,
21534 + like ptys or loop devices.
21536 +config VSERVER_CONTEXTS
21537 + int "Maximum number of Contexts (1-65533)" if EMBEDDED
21539 + default "768" if 64BIT
21542 + This setting will optimize certain data structures
21543 + and memory allocations according to the expected
21546 + note: this is not a strict upper limit.
21548 +config VSERVER_WARN
21549 + bool "VServer Warnings"
21552 + This enables various runtime warnings, which will
21553 + notify about potential manipulation attempts or
21554 + resource shortage. It is generally considered to
21555 + be a good idea to have that enabled.
21557 +config VSERVER_DEBUG
21558 + bool "VServer Debugging Code"
21561 + Set this to yes if you want to be able to activate
21562 + debugging output at runtime. It adds a very small
21563 + overhead to all vserver related functions and
21564 + increases the kernel size by about 20k.
21566 +config VSERVER_HISTORY
21567 + bool "VServer History Tracing"
21568 + depends on VSERVER_DEBUG
21571 + Set this to yes if you want to record the history of
21572 + linux-vserver activities, so they can be replayed in
21573 + the event of a kernel panic or oops.
21575 +config VSERVER_HISTORY_SIZE
21576 + int "Per-CPU History Size (32-65536)"
21577 + depends on VSERVER_HISTORY
21581 + This allows you to specify the number of entries in
21582 + the per-CPU history buffer.
21584 +config VSERVER_MONITOR
21585 + bool "VServer Scheduling Monitor"
21586 + depends on VSERVER_DISABLED
21589 + Set this to yes if you want to record the scheduling
21590 + decisions, so that they can be relayed to userspace
21591 + for detailed analysis.
21593 +config VSERVER_MONITOR_SIZE
21594 + int "Per-CPU Monitor Queue Size (32-65536)"
21595 + depends on VSERVER_MONITOR
21599 + This allows you to specify the number of entries in
21600 + the per-CPU scheduling monitor buffer.
21602 +config VSERVER_MONITOR_SYNC
21603 + int "Per-CPU Monitor Sync Interval (0-65536)"
21604 + depends on VSERVER_MONITOR
21608 + This allows you to specify the interval in ticks
21609 + when a time sync entry is inserted.
21617 + select NAMESPACES
21623 +config VSERVER_SECURITY
21625 + depends on SECURITY
21627 + select SECURITY_CAPABILITIES
21629 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/limit.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit.c
21630 --- linux-2.6.32.1/kernel/vserver/limit.c 1970-01-01 01:00:00.000000000 +0100
21631 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit.c 2009-12-03 20:04:56.000000000 +0100
21634 + * linux/kernel/vserver/limit.c
21636 + * Virtual Server: Context Limits
21638 + * Copyright (C) 2004-2007 Herbert Pötzl
21640 + * V0.01 broken out from vcontext V0.05
21641 + * V0.02 changed vcmds to vxi arg
21645 +#include <linux/sched.h>
21646 +#include <linux/module.h>
21647 +#include <linux/vs_limit.h>
21648 +#include <linux/vserver/limit.h>
21649 +#include <linux/vserver/limit_cmd.h>
21651 +#include <asm/uaccess.h>
21654 +const char *vlimit_name[NUM_LIMITS] = {
21655 + [RLIMIT_CPU] = "CPU",
21656 + [RLIMIT_RSS] = "RSS",
21657 + [RLIMIT_NPROC] = "NPROC",
21658 + [RLIMIT_NOFILE] = "NOFILE",
21659 + [RLIMIT_MEMLOCK] = "VML",
21660 + [RLIMIT_AS] = "VM",
21661 + [RLIMIT_LOCKS] = "LOCKS",
21662 + [RLIMIT_SIGPENDING] = "SIGP",
21663 + [RLIMIT_MSGQUEUE] = "MSGQ",
21665 + [VLIMIT_NSOCK] = "NSOCK",
21666 + [VLIMIT_OPENFD] = "OPENFD",
21667 + [VLIMIT_ANON] = "ANON",
21668 + [VLIMIT_SHMEM] = "SHMEM",
21669 + [VLIMIT_DENTRY] = "DENTRY",
21672 +EXPORT_SYMBOL_GPL(vlimit_name);
21674 +#define MASK_ENTRY(x) (1 << (x))
21676 +const struct vcmd_ctx_rlimit_mask_v0 vlimit_mask = {
21679 + , /* softlimit */
21680 + MASK_ENTRY( RLIMIT_RSS ) |
21681 + MASK_ENTRY( VLIMIT_ANON ) |
21684 + MASK_ENTRY( RLIMIT_RSS ) |
21685 + MASK_ENTRY( RLIMIT_NPROC ) |
21686 + MASK_ENTRY( RLIMIT_NOFILE ) |
21687 + MASK_ENTRY( RLIMIT_MEMLOCK ) |
21688 + MASK_ENTRY( RLIMIT_AS ) |
21689 + MASK_ENTRY( RLIMIT_LOCKS ) |
21690 + MASK_ENTRY( RLIMIT_MSGQUEUE ) |
21692 + MASK_ENTRY( VLIMIT_NSOCK ) |
21693 + MASK_ENTRY( VLIMIT_OPENFD ) |
21694 + MASK_ENTRY( VLIMIT_ANON ) |
21695 + MASK_ENTRY( VLIMIT_SHMEM ) |
21696 + MASK_ENTRY( VLIMIT_DENTRY ) |
21699 + /* accounting only */
21700 +uint32_t account_mask =
21701 + MASK_ENTRY( VLIMIT_SEMARY ) |
21702 + MASK_ENTRY( VLIMIT_NSEMS ) |
21703 + MASK_ENTRY( VLIMIT_MAPPED ) |
21707 +static int is_valid_vlimit(int id)
21709 + uint32_t mask = vlimit_mask.minimum |
21710 + vlimit_mask.softlimit | vlimit_mask.maximum;
21711 + return mask & (1 << id);
21714 +static int is_accounted_vlimit(int id)
21716 + if (is_valid_vlimit(id))
21718 + return account_mask & (1 << id);
21722 +static inline uint64_t vc_get_soft(struct vx_info *vxi, int id)
21724 + rlim_t limit = __rlim_soft(&vxi->limit, id);
21725 + return VX_VLIM(limit);
21728 +static inline uint64_t vc_get_hard(struct vx_info *vxi, int id)
21730 + rlim_t limit = __rlim_hard(&vxi->limit, id);
21731 + return VX_VLIM(limit);
21734 +static int do_get_rlimit(struct vx_info *vxi, uint32_t id,
21735 + uint64_t *minimum, uint64_t *softlimit, uint64_t *maximum)
21737 + if (!is_valid_vlimit(id))
21741 + *minimum = CRLIM_UNSET;
21743 + *softlimit = vc_get_soft(vxi, id);
21745 + *maximum = vc_get_hard(vxi, id);
21749 +int vc_get_rlimit(struct vx_info *vxi, void __user *data)
21751 + struct vcmd_ctx_rlimit_v0 vc_data;
21754 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21757 + ret = do_get_rlimit(vxi, vc_data.id,
21758 + &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21762 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21767 +static int do_set_rlimit(struct vx_info *vxi, uint32_t id,
21768 + uint64_t minimum, uint64_t softlimit, uint64_t maximum)
21770 + if (!is_valid_vlimit(id))
21773 + if (maximum != CRLIM_KEEP)
21774 + __rlim_hard(&vxi->limit, id) = VX_RLIM(maximum);
21775 + if (softlimit != CRLIM_KEEP)
21776 + __rlim_soft(&vxi->limit, id) = VX_RLIM(softlimit);
21778 + /* clamp soft limit */
21779 + if (__rlim_soft(&vxi->limit, id) > __rlim_hard(&vxi->limit, id))
21780 + __rlim_soft(&vxi->limit, id) = __rlim_hard(&vxi->limit, id);
21785 +int vc_set_rlimit(struct vx_info *vxi, void __user *data)
21787 + struct vcmd_ctx_rlimit_v0 vc_data;
21789 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21792 + return do_set_rlimit(vxi, vc_data.id,
21793 + vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21796 +#ifdef CONFIG_IA32_EMULATION
21798 +int vc_set_rlimit_x32(struct vx_info *vxi, void __user *data)
21800 + struct vcmd_ctx_rlimit_v0_x32 vc_data;
21802 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21805 + return do_set_rlimit(vxi, vc_data.id,
21806 + vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21809 +int vc_get_rlimit_x32(struct vx_info *vxi, void __user *data)
21811 + struct vcmd_ctx_rlimit_v0_x32 vc_data;
21814 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21817 + ret = do_get_rlimit(vxi, vc_data.id,
21818 + &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21822 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21827 +#endif /* CONFIG_IA32_EMULATION */
21830 +int vc_get_rlimit_mask(uint32_t id, void __user *data)
21832 + if (copy_to_user(data, &vlimit_mask, sizeof(vlimit_mask)))
21838 +static inline void vx_reset_hits(struct _vx_limit *limit)
21842 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21843 + atomic_set(&__rlim_lhit(limit, lim), 0);
21847 +int vc_reset_hits(struct vx_info *vxi, void __user *data)
21849 + vx_reset_hits(&vxi->limit);
21853 +static inline void vx_reset_minmax(struct _vx_limit *limit)
21858 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21859 + value = __rlim_get(limit, lim);
21860 + __rlim_rmax(limit, lim) = value;
21861 + __rlim_rmin(limit, lim) = value;
21865 +int vc_reset_minmax(struct vx_info *vxi, void __user *data)
21867 + vx_reset_minmax(&vxi->limit);
21872 +int vc_rlimit_stat(struct vx_info *vxi, void __user *data)
21874 + struct vcmd_rlimit_stat_v0 vc_data;
21875 + struct _vx_limit *limit = &vxi->limit;
21878 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21882 + if (!is_accounted_vlimit(id))
21885 + vx_limit_fixup(limit, id);
21886 + vc_data.hits = atomic_read(&__rlim_lhit(limit, id));
21887 + vc_data.value = __rlim_get(limit, id);
21888 + vc_data.minimum = __rlim_rmin(limit, id);
21889 + vc_data.maximum = __rlim_rmax(limit, id);
21891 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21897 +void vx_vsi_meminfo(struct sysinfo *val)
21899 + struct vx_info *vxi = current_vx_info();
21900 + unsigned long totalram, freeram;
21903 + /* we blindly accept the max */
21904 + v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21905 + totalram = (v != RLIM_INFINITY) ? v : val->totalram;
21907 + /* total minus used equals free */
21908 + v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21909 + freeram = (v < totalram) ? totalram - v : 0;
21911 + val->totalram = totalram;
21912 + val->freeram = freeram;
21913 + val->bufferram = 0;
21914 + val->totalhigh = 0;
21915 + val->freehigh = 0;
21919 +void vx_vsi_swapinfo(struct sysinfo *val)
21921 + struct vx_info *vxi = current_vx_info();
21922 + unsigned long totalswap, freeswap;
21925 + v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21926 + if (v == RLIM_INFINITY) {
21927 + val->freeswap = val->totalswap;
21931 + /* we blindly accept the max */
21932 + w = __rlim_hard(&vxi->limit, RLIMIT_RSS);
21933 + totalswap = (w != RLIM_INFINITY) ? (w - v) : val->totalswap;
21935 + /* currently 'used' swap */
21936 + w = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21937 + w -= (w > v) ? v : w;
21939 + /* total minus used equals free */
21940 + freeswap = (w < totalswap) ? totalswap - w : 0;
21942 + val->totalswap = totalswap;
21943 + val->freeswap = freeswap;
21948 +unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm)
21950 + struct vx_info *vxi = mm->mm_vx_info;
21951 + unsigned long points;
21957 + points = vxi->vx_badness_bias;
21959 + v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21960 + w = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21961 + points += (v > w) ? (v - w) : 0;
21966 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/limit_init.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit_init.h
21967 --- linux-2.6.32.1/kernel/vserver/limit_init.h 1970-01-01 01:00:00.000000000 +0100
21968 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit_init.h 2009-12-03 20:04:56.000000000 +0100
21972 +static inline void vx_info_init_limit(struct _vx_limit *limit)
21976 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21977 + __rlim_soft(limit, lim) = RLIM_INFINITY;
21978 + __rlim_hard(limit, lim) = RLIM_INFINITY;
21979 + __rlim_set(limit, lim, 0);
21980 + atomic_set(&__rlim_lhit(limit, lim), 0);
21981 + __rlim_rmin(limit, lim) = 0;
21982 + __rlim_rmax(limit, lim) = 0;
21986 +static inline void vx_info_exit_limit(struct _vx_limit *limit)
21991 + for (lim = 0; lim < NUM_LIMITS; lim++) {
21992 + if ((1 << lim) & VLIM_NOCHECK)
21994 + value = __rlim_get(limit, lim);
21995 + vxwprintk_xid(value,
21996 + "!!! limit: %p[%s,%d] = %ld on exit.",
21997 + limit, vlimit_name[lim], lim, (long)value);
22001 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/limit_proc.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit_proc.h
22002 --- linux-2.6.32.1/kernel/vserver/limit_proc.h 1970-01-01 01:00:00.000000000 +0100
22003 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/limit_proc.h 2009-12-03 20:04:56.000000000 +0100
22005 +#ifndef _VX_LIMIT_PROC_H
22006 +#define _VX_LIMIT_PROC_H
22008 +#include <linux/vserver/limit_int.h>
22011 +#define VX_LIMIT_FMT ":\t%8ld\t%8ld/%8ld\t%8lld/%8lld\t%6d\n"
22012 +#define VX_LIMIT_TOP \
22013 + "Limit\t current\t min/max\t\t soft/hard\t\thits\n"
22015 +#define VX_LIMIT_ARG(r) \
22016 + (unsigned long)__rlim_get(limit, r), \
22017 + (unsigned long)__rlim_rmin(limit, r), \
22018 + (unsigned long)__rlim_rmax(limit, r), \
22019 + VX_VLIM(__rlim_soft(limit, r)), \
22020 + VX_VLIM(__rlim_hard(limit, r)), \
22021 + atomic_read(&__rlim_lhit(limit, r))
22023 +static inline int vx_info_proc_limit(struct _vx_limit *limit, char *buffer)
22025 + vx_limit_fixup(limit, -1);
22026 + return sprintf(buffer, VX_LIMIT_TOP
22027 + "PROC" VX_LIMIT_FMT
22028 + "VM" VX_LIMIT_FMT
22029 + "VML" VX_LIMIT_FMT
22030 + "RSS" VX_LIMIT_FMT
22031 + "ANON" VX_LIMIT_FMT
22032 + "RMAP" VX_LIMIT_FMT
22033 + "FILES" VX_LIMIT_FMT
22034 + "OFD" VX_LIMIT_FMT
22035 + "LOCKS" VX_LIMIT_FMT
22036 + "SOCK" VX_LIMIT_FMT
22037 + "MSGQ" VX_LIMIT_FMT
22038 + "SHM" VX_LIMIT_FMT
22039 + "SEMA" VX_LIMIT_FMT
22040 + "SEMS" VX_LIMIT_FMT
22041 + "DENT" VX_LIMIT_FMT,
22042 + VX_LIMIT_ARG(RLIMIT_NPROC),
22043 + VX_LIMIT_ARG(RLIMIT_AS),
22044 + VX_LIMIT_ARG(RLIMIT_MEMLOCK),
22045 + VX_LIMIT_ARG(RLIMIT_RSS),
22046 + VX_LIMIT_ARG(VLIMIT_ANON),
22047 + VX_LIMIT_ARG(VLIMIT_MAPPED),
22048 + VX_LIMIT_ARG(RLIMIT_NOFILE),
22049 + VX_LIMIT_ARG(VLIMIT_OPENFD),
22050 + VX_LIMIT_ARG(RLIMIT_LOCKS),
22051 + VX_LIMIT_ARG(VLIMIT_NSOCK),
22052 + VX_LIMIT_ARG(RLIMIT_MSGQUEUE),
22053 + VX_LIMIT_ARG(VLIMIT_SHMEM),
22054 + VX_LIMIT_ARG(VLIMIT_SEMARY),
22055 + VX_LIMIT_ARG(VLIMIT_NSEMS),
22056 + VX_LIMIT_ARG(VLIMIT_DENTRY));
22059 +#endif /* _VX_LIMIT_PROC_H */
22062 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/Makefile linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/Makefile
22063 --- linux-2.6.32.1/kernel/vserver/Makefile 1970-01-01 01:00:00.000000000 +0100
22064 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/Makefile 2009-12-03 20:04:56.000000000 +0100
22067 +# Makefile for the Linux vserver routines.
22071 +obj-y += vserver.o
22073 +vserver-y := switch.o context.o space.o sched.o network.o inode.o \
22074 + limit.o cvirt.o cacct.o signal.o helper.o init.o \
22077 +vserver-$(CONFIG_INET) += inet.o
22078 +vserver-$(CONFIG_PROC_FS) += proc.o
22079 +vserver-$(CONFIG_VSERVER_DEBUG) += sysctl.o debug.o
22080 +vserver-$(CONFIG_VSERVER_HISTORY) += history.o
22081 +vserver-$(CONFIG_VSERVER_MONITOR) += monitor.o
22082 +vserver-$(CONFIG_VSERVER_DEVICE) += device.o
22084 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/monitor.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/monitor.c
22085 --- linux-2.6.32.1/kernel/vserver/monitor.c 1970-01-01 01:00:00.000000000 +0100
22086 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/monitor.c 2009-12-03 20:04:56.000000000 +0100
22089 + * kernel/vserver/monitor.c
22091 + * Virtual Context Scheduler Monitor
22093 + * Copyright (C) 2006-2007 Herbert Pötzl
22095 + * V0.01 basic design
22099 +#include <linux/module.h>
22100 +#include <linux/jiffies.h>
22101 +#include <asm/uaccess.h>
22102 +#include <asm/atomic.h>
22104 +#include <linux/vserver/monitor.h>
22105 +#include <linux/vserver/debug_cmd.h>
22108 +#ifdef CONFIG_VSERVER_MONITOR
22109 +#define VXM_SIZE CONFIG_VSERVER_MONITOR_SIZE
22111 +#define VXM_SIZE 64
22114 +struct _vx_monitor {
22115 + unsigned int counter;
22117 + struct _vx_mon_entry entry[VXM_SIZE+1];
22121 +DEFINE_PER_CPU(struct _vx_monitor, vx_monitor_buffer);
22123 +unsigned volatile int vxm_active = 1;
22125 +static atomic_t sequence = ATOMIC_INIT(0);
22130 + * requires disabled preemption */
22132 +struct _vx_mon_entry *vxm_advance(int cpu)
22134 + struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
22135 + struct _vx_mon_entry *entry;
22136 + unsigned int index;
22138 + index = vxm_active ? (mon->counter++ % VXM_SIZE) : VXM_SIZE;
22139 + entry = &mon->entry[index];
22141 + entry->ev.seq = atomic_inc_return(&sequence);
22142 + entry->ev.jif = jiffies;
22146 +EXPORT_SYMBOL_GPL(vxm_advance);
22149 +int do_read_monitor(struct __user _vx_mon_entry *data,
22150 + int cpu, uint32_t *index, uint32_t *count)
22152 + int pos, ret = 0;
22153 + struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
22154 + int end = mon->counter;
22155 + int start = end - VXM_SIZE + 2;
22156 + int idx = *index;
22158 + /* special case: get current pos */
22164 + /* have we lost some data? */
22168 + for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
22169 + struct _vx_mon_entry *entry =
22170 + &mon->entry[idx % VXM_SIZE];
22172 + /* send entry to userspace */
22173 + ret = copy_to_user(&data[pos], entry, sizeof(*entry));
22177 + /* save new index and count */
22180 + return ret ? ret : (*index < end);
22183 +int vc_read_monitor(uint32_t id, void __user *data)
22185 + struct vcmd_read_monitor_v0 vc_data;
22188 + if (id >= NR_CPUS)
22191 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22194 + ret = do_read_monitor((struct __user _vx_mon_entry *)vc_data.data,
22195 + id, &vc_data.index, &vc_data.count);
22197 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22202 +#ifdef CONFIG_COMPAT
22204 +int vc_read_monitor_x32(uint32_t id, void __user *data)
22206 + struct vcmd_read_monitor_v0_x32 vc_data;
22209 + if (id >= NR_CPUS)
22212 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22215 + ret = do_read_monitor((struct __user _vx_mon_entry *)
22216 + compat_ptr(vc_data.data_ptr),
22217 + id, &vc_data.index, &vc_data.count);
22219 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22224 +#endif /* CONFIG_COMPAT */
22226 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/network.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/network.c
22227 --- linux-2.6.32.1/kernel/vserver/network.c 1970-01-01 01:00:00.000000000 +0100
22228 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/network.c 2009-12-03 20:04:56.000000000 +0100
22231 + * linux/kernel/vserver/network.c
22233 + * Virtual Server: Network Support
22235 + * Copyright (C) 2003-2007 Herbert Pötzl
22237 + * V0.01 broken out from vcontext V0.05
22238 + * V0.02 cleaned up implementation
22239 + * V0.03 added equiv nx commands
22240 + * V0.04 switch to RCU based hash
22241 + * V0.05 and back to locking again
22242 + * V0.06 changed vcmds to nxi arg
22243 + * V0.07 have __create claim() the nxi
22247 +#include <linux/err.h>
22248 +#include <linux/slab.h>
22249 +#include <linux/rcupdate.h>
22251 +#include <linux/vs_network.h>
22252 +#include <linux/vs_pid.h>
22253 +#include <linux/vserver/network_cmd.h>
22256 +atomic_t nx_global_ctotal = ATOMIC_INIT(0);
22257 +atomic_t nx_global_cactive = ATOMIC_INIT(0);
22259 +static struct kmem_cache *nx_addr_v4_cachep = NULL;
22260 +static struct kmem_cache *nx_addr_v6_cachep = NULL;
22263 +static int __init init_network(void)
22265 + nx_addr_v4_cachep = kmem_cache_create("nx_v4_addr_cache",
22266 + sizeof(struct nx_addr_v4), 0,
22267 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
22268 + nx_addr_v6_cachep = kmem_cache_create("nx_v6_addr_cache",
22269 + sizeof(struct nx_addr_v6), 0,
22270 + SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
22275 +/* __alloc_nx_addr_v4() */
22277 +static inline struct nx_addr_v4 *__alloc_nx_addr_v4(void)
22279 + struct nx_addr_v4 *nxa = kmem_cache_alloc(
22280 + nx_addr_v4_cachep, GFP_KERNEL);
22282 + if (!IS_ERR(nxa))
22283 + memset(nxa, 0, sizeof(*nxa));
22287 +/* __dealloc_nx_addr_v4() */
22289 +static inline void __dealloc_nx_addr_v4(struct nx_addr_v4 *nxa)
22291 + kmem_cache_free(nx_addr_v4_cachep, nxa);
22294 +/* __dealloc_nx_addr_v4_all() */
22296 +static inline void __dealloc_nx_addr_v4_all(struct nx_addr_v4 *nxa)
22299 + struct nx_addr_v4 *next = nxa->next;
22301 + __dealloc_nx_addr_v4(nxa);
22307 +#ifdef CONFIG_IPV6
22309 +/* __alloc_nx_addr_v6() */
22311 +static inline struct nx_addr_v6 *__alloc_nx_addr_v6(void)
22313 + struct nx_addr_v6 *nxa = kmem_cache_alloc(
22314 + nx_addr_v6_cachep, GFP_KERNEL);
22316 + if (!IS_ERR(nxa))
22317 + memset(nxa, 0, sizeof(*nxa));
22321 +/* __dealloc_nx_addr_v6() */
22323 +static inline void __dealloc_nx_addr_v6(struct nx_addr_v6 *nxa)
22325 + kmem_cache_free(nx_addr_v6_cachep, nxa);
22328 +/* __dealloc_nx_addr_v6_all() */
22330 +static inline void __dealloc_nx_addr_v6_all(struct nx_addr_v6 *nxa)
22333 + struct nx_addr_v6 *next = nxa->next;
22335 + __dealloc_nx_addr_v6(nxa);
22340 +#endif /* CONFIG_IPV6 */
22342 +/* __alloc_nx_info()
22344 + * allocate an initialized nx_info struct
22345 + * doesn't make it visible (hash) */
22347 +static struct nx_info *__alloc_nx_info(nid_t nid)
22349 + struct nx_info *new = NULL;
22351 + vxdprintk(VXD_CBIT(nid, 1), "alloc_nx_info(%d)*", nid);
22353 + /* would this benefit from a slab cache? */
22354 + new = kmalloc(sizeof(struct nx_info), GFP_KERNEL);
22358 + memset(new, 0, sizeof(struct nx_info));
22359 + new->nx_id = nid;
22360 + INIT_HLIST_NODE(&new->nx_hlist);
22361 + atomic_set(&new->nx_usecnt, 0);
22362 + atomic_set(&new->nx_tasks, 0);
22363 + new->nx_state = 0;
22365 + new->nx_flags = NXF_INIT_SET;
22367 + /* rest of init goes here */
22369 + new->v4_lback.s_addr = htonl(INADDR_LOOPBACK);
22370 + new->v4_bcast.s_addr = htonl(INADDR_BROADCAST);
22372 + vxdprintk(VXD_CBIT(nid, 0),
22373 + "alloc_nx_info(%d) = %p", nid, new);
22374 + atomic_inc(&nx_global_ctotal);
22378 +/* __dealloc_nx_info()
22380 + * final disposal of nx_info */
22382 +static void __dealloc_nx_info(struct nx_info *nxi)
22384 + vxdprintk(VXD_CBIT(nid, 0),
22385 + "dealloc_nx_info(%p)", nxi);
22387 + nxi->nx_hlist.next = LIST_POISON1;
22390 + BUG_ON(atomic_read(&nxi->nx_usecnt));
22391 + BUG_ON(atomic_read(&nxi->nx_tasks));
22393 + __dealloc_nx_addr_v4_all(nxi->v4.next);
22395 + nxi->nx_state |= NXS_RELEASED;
22397 + atomic_dec(&nx_global_ctotal);
22400 +static void __shutdown_nx_info(struct nx_info *nxi)
22402 + nxi->nx_state |= NXS_SHUTDOWN;
22403 + vs_net_change(nxi, VSC_NETDOWN);
22406 +/* exported stuff */
22408 +void free_nx_info(struct nx_info *nxi)
22410 + /* context shutdown is mandatory */
22411 + BUG_ON(nxi->nx_state != NXS_SHUTDOWN);
22413 + /* context must not be hashed */
22414 + BUG_ON(nxi->nx_state & NXS_HASHED);
22416 + BUG_ON(atomic_read(&nxi->nx_usecnt));
22417 + BUG_ON(atomic_read(&nxi->nx_tasks));
22419 + __dealloc_nx_info(nxi);
22423 +void __nx_set_lback(struct nx_info *nxi)
22425 + int nid = nxi->nx_id;
22426 + __be32 lback = htonl(INADDR_LOOPBACK ^ ((nid & 0xFFFF) << 8));
22428 + nxi->v4_lback.s_addr = lback;
22431 +extern int __nx_inet_add_lback(__be32 addr);
22432 +extern int __nx_inet_del_lback(__be32 addr);
22435 +/* hash table for nx_info hash */
22437 +#define NX_HASH_SIZE 13
22439 +struct hlist_head nx_info_hash[NX_HASH_SIZE];
22441 +static spinlock_t nx_info_hash_lock = SPIN_LOCK_UNLOCKED;
22444 +static inline unsigned int __hashval(nid_t nid)
22446 + return (nid % NX_HASH_SIZE);
22451 +/* __hash_nx_info()
22453 + * add the nxi to the global hash table
22454 + * requires the hash_lock to be held */
22456 +static inline void __hash_nx_info(struct nx_info *nxi)
22458 + struct hlist_head *head;
22460 + vxd_assert_lock(&nx_info_hash_lock);
22461 + vxdprintk(VXD_CBIT(nid, 4),
22462 + "__hash_nx_info: %p[#%d]", nxi, nxi->nx_id);
22464 + /* context must not be hashed */
22465 + BUG_ON(nx_info_state(nxi, NXS_HASHED));
22467 + nxi->nx_state |= NXS_HASHED;
22468 + head = &nx_info_hash[__hashval(nxi->nx_id)];
22469 + hlist_add_head(&nxi->nx_hlist, head);
22470 + atomic_inc(&nx_global_cactive);
22473 +/* __unhash_nx_info()
22475 + * remove the nxi from the global hash table
22476 + * requires the hash_lock to be held */
22478 +static inline void __unhash_nx_info(struct nx_info *nxi)
22480 + vxd_assert_lock(&nx_info_hash_lock);
22481 + vxdprintk(VXD_CBIT(nid, 4),
22482 + "__unhash_nx_info: %p[#%d.%d.%d]", nxi, nxi->nx_id,
22483 + atomic_read(&nxi->nx_usecnt), atomic_read(&nxi->nx_tasks));
22485 + /* context must be hashed */
22486 + BUG_ON(!nx_info_state(nxi, NXS_HASHED));
22487 + /* but without tasks */
22488 + BUG_ON(atomic_read(&nxi->nx_tasks));
22490 + nxi->nx_state &= ~NXS_HASHED;
22491 + hlist_del(&nxi->nx_hlist);
22492 + atomic_dec(&nx_global_cactive);
22496 +/* __lookup_nx_info()
22498 + * requires the hash_lock to be held
22499 + * doesn't increment the nx_refcnt */
22501 +static inline struct nx_info *__lookup_nx_info(nid_t nid)
22503 + struct hlist_head *head = &nx_info_hash[__hashval(nid)];
22504 + struct hlist_node *pos;
22505 + struct nx_info *nxi;
22507 + vxd_assert_lock(&nx_info_hash_lock);
22508 + hlist_for_each(pos, head) {
22509 + nxi = hlist_entry(pos, struct nx_info, nx_hlist);
22511 + if (nxi->nx_id == nid)
22516 + vxdprintk(VXD_CBIT(nid, 0),
22517 + "__lookup_nx_info(#%u): %p[#%u]",
22518 + nid, nxi, nxi ? nxi->nx_id : 0);
22523 +/* __create_nx_info()
22525 + * create the requested context
22526 + * get(), claim() and hash it */
22528 +static struct nx_info *__create_nx_info(int id)
22530 + struct nx_info *new, *nxi = NULL;
22532 + vxdprintk(VXD_CBIT(nid, 1), "create_nx_info(%d)*", id);
22534 + if (!(new = __alloc_nx_info(id)))
22535 + return ERR_PTR(-ENOMEM);
22537 + /* required to make dynamic xids unique */
22538 + spin_lock(&nx_info_hash_lock);
22540 + /* static context requested */
22541 + if ((nxi = __lookup_nx_info(id))) {
22542 + vxdprintk(VXD_CBIT(nid, 0),
22543 + "create_nx_info(%d) = %p (already there)", id, nxi);
22544 + if (nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22545 + nxi = ERR_PTR(-EBUSY);
22547 + nxi = ERR_PTR(-EEXIST);
22550 + /* new context */
22551 + vxdprintk(VXD_CBIT(nid, 0),
22552 + "create_nx_info(%d) = %p (new)", id, new);
22553 + claim_nx_info(new, NULL);
22554 + __nx_set_lback(new);
22555 + __hash_nx_info(get_nx_info(new));
22556 + nxi = new, new = NULL;
22559 + spin_unlock(&nx_info_hash_lock);
22561 + __dealloc_nx_info(new);
22567 +/* exported stuff */
22570 +void unhash_nx_info(struct nx_info *nxi)
22572 + __shutdown_nx_info(nxi);
22573 + spin_lock(&nx_info_hash_lock);
22574 + __unhash_nx_info(nxi);
22575 + spin_unlock(&nx_info_hash_lock);
22578 +/* lookup_nx_info()
22580 + * search for a nx_info and get() it
22581 + * negative id means current */
22583 +struct nx_info *lookup_nx_info(int id)
22585 + struct nx_info *nxi = NULL;
22588 + nxi = get_nx_info(current_nx_info());
22589 + } else if (id > 1) {
22590 + spin_lock(&nx_info_hash_lock);
22591 + nxi = get_nx_info(__lookup_nx_info(id));
22592 + spin_unlock(&nx_info_hash_lock);
22597 +/* nid_is_hashed()
22599 + * verify that nid is still hashed */
22601 +int nid_is_hashed(nid_t nid)
22605 + spin_lock(&nx_info_hash_lock);
22606 + hashed = (__lookup_nx_info(nid) != NULL);
22607 + spin_unlock(&nx_info_hash_lock);
22612 +#ifdef CONFIG_PROC_FS
22616 + * get a subset of hashed nids for proc
22617 + * assumes size is at least one */
22619 +int get_nid_list(int index, unsigned int *nids, int size)
22621 + int hindex, nr_nids = 0;
22623 + /* only show current and children */
22624 + if (!nx_check(0, VS_ADMIN | VS_WATCH)) {
22627 + nids[nr_nids] = nx_current_nid();
22631 + for (hindex = 0; hindex < NX_HASH_SIZE; hindex++) {
22632 + struct hlist_head *head = &nx_info_hash[hindex];
22633 + struct hlist_node *pos;
22635 + spin_lock(&nx_info_hash_lock);
22636 + hlist_for_each(pos, head) {
22637 + struct nx_info *nxi;
22642 + nxi = hlist_entry(pos, struct nx_info, nx_hlist);
22643 + nids[nr_nids] = nxi->nx_id;
22644 + if (++nr_nids >= size) {
22645 + spin_unlock(&nx_info_hash_lock);
22649 + /* keep the lock time short */
22650 + spin_unlock(&nx_info_hash_lock);
22659 + * migrate task to new network
22660 + * gets nxi, puts old_nxi on change
22663 +int nx_migrate_task(struct task_struct *p, struct nx_info *nxi)
22665 + struct nx_info *old_nxi;
22671 + vxdprintk(VXD_CBIT(nid, 5),
22672 + "nx_migrate_task(%p,%p[#%d.%d.%d])",
22673 + p, nxi, nxi->nx_id,
22674 + atomic_read(&nxi->nx_usecnt),
22675 + atomic_read(&nxi->nx_tasks));
22677 + if (nx_info_flags(nxi, NXF_INFO_PRIVATE, 0) &&
22678 + !nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22681 + if (nx_info_state(nxi, NXS_SHUTDOWN))
22684 + /* maybe disallow this completely? */
22685 + old_nxi = task_get_nx_info(p);
22686 + if (old_nxi == nxi)
22691 + clr_nx_info(&p->nx_info);
22692 + claim_nx_info(nxi, p);
22693 + set_nx_info(&p->nx_info, nxi);
22694 + p->nid = nxi->nx_id;
22697 + vxdprintk(VXD_CBIT(nid, 5),
22698 + "moved task %p into nxi:%p[#%d]",
22699 + p, nxi, nxi->nx_id);
22702 + release_nx_info(old_nxi, p);
22705 + put_nx_info(old_nxi);
22710 +void nx_set_persistent(struct nx_info *nxi)
22712 + vxdprintk(VXD_CBIT(nid, 6),
22713 + "nx_set_persistent(%p[#%d])", nxi, nxi->nx_id);
22715 + get_nx_info(nxi);
22716 + claim_nx_info(nxi, NULL);
22719 +void nx_clear_persistent(struct nx_info *nxi)
22721 + vxdprintk(VXD_CBIT(nid, 6),
22722 + "nx_clear_persistent(%p[#%d])", nxi, nxi->nx_id);
22724 + release_nx_info(nxi, NULL);
22725 + put_nx_info(nxi);
22728 +void nx_update_persistent(struct nx_info *nxi)
22730 + if (nx_info_flags(nxi, NXF_PERSISTENT, 0))
22731 + nx_set_persistent(nxi);
22733 + nx_clear_persistent(nxi);
22736 +/* vserver syscall commands below here */
22738 +/* taks nid and nx_info functions */
22740 +#include <asm/uaccess.h>
22743 +int vc_task_nid(uint32_t id)
22748 + struct task_struct *tsk;
22750 + read_lock(&tasklist_lock);
22751 + tsk = find_task_by_real_pid(id);
22752 + nid = (tsk) ? tsk->nid : -ESRCH;
22753 + read_unlock(&tasklist_lock);
22755 + nid = nx_current_nid();
22760 +int vc_nx_info(struct nx_info *nxi, void __user *data)
22762 + struct vcmd_nx_info_v0 vc_data;
22764 + vc_data.nid = nxi->nx_id;
22766 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22772 +/* network functions */
22774 +int vc_net_create(uint32_t nid, void __user *data)
22776 + struct vcmd_net_create vc_data = { .flagword = NXF_INIT_SET };
22777 + struct nx_info *new_nxi;
22780 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22783 + if ((nid > MAX_S_CONTEXT) || (nid < 2))
22786 + new_nxi = __create_nx_info(nid);
22787 + if (IS_ERR(new_nxi))
22788 + return PTR_ERR(new_nxi);
22790 + /* initial flags */
22791 + new_nxi->nx_flags = vc_data.flagword;
22794 + if (vs_net_change(new_nxi, VSC_NETUP))
22797 + ret = nx_migrate_task(current, new_nxi);
22801 + /* return context id on success */
22802 + ret = new_nxi->nx_id;
22804 + /* get a reference for persistent contexts */
22805 + if ((vc_data.flagword & NXF_PERSISTENT))
22806 + nx_set_persistent(new_nxi);
22808 + release_nx_info(new_nxi, NULL);
22809 + put_nx_info(new_nxi);
22814 +int vc_net_migrate(struct nx_info *nxi, void __user *data)
22816 + return nx_migrate_task(current, nxi);
22821 +int do_add_v4_addr(struct nx_info *nxi, __be32 ip, __be32 ip2, __be32 mask,
22822 + uint16_t type, uint16_t flags)
22824 + struct nx_addr_v4 *nxa = &nxi->v4;
22826 + if (NX_IPV4(nxi)) {
22827 + /* locate last entry */
22828 + for (; nxa->next; nxa = nxa->next);
22829 + nxa->next = __alloc_nx_addr_v4();
22833 + return PTR_ERR(nxa);
22836 + if (nxi->v4.next)
22837 + /* remove single ip for ip list */
22838 + nxi->nx_flags &= ~NXF_SINGLE_IP;
22840 + nxa->ip[0].s_addr = ip;
22841 + nxa->ip[1].s_addr = ip2;
22842 + nxa->mask.s_addr = mask;
22843 + nxa->type = type;
22844 + nxa->flags = flags;
22849 +int vc_net_add(struct nx_info *nxi, void __user *data)
22851 + struct vcmd_net_addr_v0 vc_data;
22852 + int index, ret = 0;
22854 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22857 + switch (vc_data.type) {
22858 + case NXA_TYPE_IPV4:
22859 + if ((vc_data.count < 1) || (vc_data.count > 4))
22863 + while (index < vc_data.count) {
22864 + ret = do_add_v4_addr(nxi, vc_data.ip[index].s_addr, 0,
22865 + vc_data.mask[index].s_addr, NXA_TYPE_ADDR, 0);
22873 + case NXA_TYPE_IPV4|NXA_MOD_BCAST:
22874 + nxi->v4_bcast = vc_data.ip[0];
22878 + case NXA_TYPE_IPV4|NXA_MOD_LBACK:
22879 + nxi->v4_lback = vc_data.ip[0];
22890 +int vc_net_remove(struct nx_info *nxi, void __user *data)
22892 + struct vcmd_net_addr_v0 vc_data;
22894 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22897 + switch (vc_data.type) {
22898 + case NXA_TYPE_ANY:
22899 + __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22900 + memset(&nxi->v4, 0, sizeof(nxi->v4));
22910 +int vc_net_add_ipv4(struct nx_info *nxi, void __user *data)
22912 + struct vcmd_net_addr_ipv4_v1 vc_data;
22914 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22917 + switch (vc_data.type) {
22918 + case NXA_TYPE_ADDR:
22919 + case NXA_TYPE_RANGE:
22920 + case NXA_TYPE_MASK:
22921 + return do_add_v4_addr(nxi, vc_data.ip.s_addr, 0,
22922 + vc_data.mask.s_addr, vc_data.type, vc_data.flags);
22924 + case NXA_TYPE_ADDR | NXA_MOD_BCAST:
22925 + nxi->v4_bcast = vc_data.ip;
22928 + case NXA_TYPE_ADDR | NXA_MOD_LBACK:
22929 + nxi->v4_lback = vc_data.ip;
22938 +int vc_net_remove_ipv4(struct nx_info *nxi, void __user *data)
22940 + struct vcmd_net_addr_ipv4_v1 vc_data;
22942 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22945 + switch (vc_data.type) {
22946 +/* case NXA_TYPE_ADDR:
22949 + case NXA_TYPE_ANY:
22950 + __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22951 + memset(&nxi->v4, 0, sizeof(nxi->v4));
22961 +#ifdef CONFIG_IPV6
22963 +int do_add_v6_addr(struct nx_info *nxi,
22964 + struct in6_addr *ip, struct in6_addr *mask,
22965 + uint32_t prefix, uint16_t type, uint16_t flags)
22967 + struct nx_addr_v6 *nxa = &nxi->v6;
22969 + if (NX_IPV6(nxi)) {
22970 + /* locate last entry */
22971 + for (; nxa->next; nxa = nxa->next);
22972 + nxa->next = __alloc_nx_addr_v6();
22976 + return PTR_ERR(nxa);
22980 + nxa->mask = *mask;
22981 + nxa->prefix = prefix;
22982 + nxa->type = type;
22983 + nxa->flags = flags;
22988 +int vc_net_add_ipv6(struct nx_info *nxi, void __user *data)
22990 + struct vcmd_net_addr_ipv6_v1 vc_data;
22992 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22995 + switch (vc_data.type) {
22996 + case NXA_TYPE_ADDR:
22997 + case NXA_TYPE_MASK:
22998 + return do_add_v6_addr(nxi, &vc_data.ip, &vc_data.mask,
22999 + vc_data.prefix, vc_data.type, vc_data.flags);
23006 +int vc_net_remove_ipv6(struct nx_info *nxi, void __user *data)
23008 + struct vcmd_net_addr_ipv6_v1 vc_data;
23010 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
23013 + switch (vc_data.type) {
23014 + case NXA_TYPE_ANY:
23015 + __dealloc_nx_addr_v6_all(xchg(&nxi->v6.next, NULL));
23016 + memset(&nxi->v6, 0, sizeof(nxi->v6));
23025 +#endif /* CONFIG_IPV6 */
23028 +int vc_get_nflags(struct nx_info *nxi, void __user *data)
23030 + struct vcmd_net_flags_v0 vc_data;
23032 + vc_data.flagword = nxi->nx_flags;
23034 + /* special STATE flag handling */
23035 + vc_data.mask = vs_mask_flags(~0ULL, nxi->nx_flags, NXF_ONE_TIME);
23037 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
23042 +int vc_set_nflags(struct nx_info *nxi, void __user *data)
23044 + struct vcmd_net_flags_v0 vc_data;
23045 + uint64_t mask, trigger;
23047 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
23050 + /* special STATE flag handling */
23051 + mask = vs_mask_mask(vc_data.mask, nxi->nx_flags, NXF_ONE_TIME);
23052 + trigger = (mask & nxi->nx_flags) ^ (mask & vc_data.flagword);
23054 + nxi->nx_flags = vs_mask_flags(nxi->nx_flags,
23055 + vc_data.flagword, mask);
23056 + if (trigger & NXF_PERSISTENT)
23057 + nx_update_persistent(nxi);
23062 +int vc_get_ncaps(struct nx_info *nxi, void __user *data)
23064 + struct vcmd_net_caps_v0 vc_data;
23066 + vc_data.ncaps = nxi->nx_ncaps;
23067 + vc_data.cmask = ~0ULL;
23069 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
23074 +int vc_set_ncaps(struct nx_info *nxi, void __user *data)
23076 + struct vcmd_net_caps_v0 vc_data;
23078 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
23081 + nxi->nx_ncaps = vs_mask_flags(nxi->nx_ncaps,
23082 + vc_data.ncaps, vc_data.cmask);
23087 +#include <linux/module.h>
23089 +module_init(init_network);
23091 +EXPORT_SYMBOL_GPL(free_nx_info);
23092 +EXPORT_SYMBOL_GPL(unhash_nx_info);
23094 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/proc.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/proc.c
23095 --- linux-2.6.32.1/kernel/vserver/proc.c 1970-01-01 01:00:00.000000000 +0100
23096 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/proc.c 2009-12-03 20:04:56.000000000 +0100
23099 + * linux/kernel/vserver/proc.c
23101 + * Virtual Context Support
23103 + * Copyright (C) 2003-2007 Herbert Pötzl
23105 + * V0.01 basic structure
23106 + * V0.02 adaptation vs1.3.0
23107 + * V0.03 proc permissions
23108 + * V0.04 locking/generic
23109 + * V0.05 next generation procfs
23110 + * V0.06 inode validation
23111 + * V0.07 generic rewrite vid
23112 + * V0.08 remove inode type
23116 +#include <linux/proc_fs.h>
23117 +#include <linux/fs_struct.h>
23118 +#include <linux/mount.h>
23119 +#include <asm/unistd.h>
23121 +#include <linux/vs_context.h>
23122 +#include <linux/vs_network.h>
23123 +#include <linux/vs_cvirt.h>
23125 +#include <linux/in.h>
23126 +#include <linux/inetdevice.h>
23127 +#include <linux/vs_inet.h>
23128 +#include <linux/vs_inet6.h>
23130 +#include <linux/vserver/global.h>
23132 +#include "cvirt_proc.h"
23133 +#include "cacct_proc.h"
23134 +#include "limit_proc.h"
23135 +#include "sched_proc.h"
23136 +#include "vci_config.h"
23139 +static inline char *print_cap_t(char *buffer, kernel_cap_t *c)
23143 + CAP_FOR_EACH_U32(__capi) {
23144 + buffer += sprintf(buffer, "%08x",
23145 + c->cap[(_KERNEL_CAPABILITY_U32S-1) - __capi]);
23151 +static struct proc_dir_entry *proc_virtual;
23153 +static struct proc_dir_entry *proc_virtnet;
23156 +/* first the actual feeds */
23159 +static int proc_vci(char *buffer)
23161 + return sprintf(buffer,
23162 + "VCIVersion:\t%04x:%04x\n"
23163 + "VCISyscall:\t%d\n"
23164 + "VCIKernel:\t%08x\n",
23165 + VCI_VERSION >> 16,
23166 + VCI_VERSION & 0xFFFF,
23168 + vci_kernel_config());
23171 +static int proc_virtual_info(char *buffer)
23173 + return proc_vci(buffer);
23176 +static int proc_virtual_status(char *buffer)
23178 + return sprintf(buffer,
23180 + "#CActive:\t%d\n"
23181 + "#NSProxy:\t%d\t%d %d %d %d %d %d\n"
23182 + "#InitTask:\t%d\t%d %d\n",
23183 + atomic_read(&vx_global_ctotal),
23184 + atomic_read(&vx_global_cactive),
23185 + atomic_read(&vs_global_nsproxy),
23186 + atomic_read(&vs_global_fs),
23187 + atomic_read(&vs_global_mnt_ns),
23188 + atomic_read(&vs_global_uts_ns),
23189 + atomic_read(&nr_ipc_ns),
23190 + atomic_read(&vs_global_user_ns),
23191 + atomic_read(&vs_global_pid_ns),
23192 + atomic_read(&init_task.usage),
23193 + atomic_read(&init_task.nsproxy->count),
23194 + init_task.fs->users);
23198 +int proc_vxi_info(struct vx_info *vxi, char *buffer)
23202 + length = sprintf(buffer,
23210 + vxi->vx_badness_bias);
23214 +int proc_vxi_status(struct vx_info *vxi, char *buffer)
23216 + char *orig = buffer;
23218 + buffer += sprintf(buffer,
23221 + "Flags:\t%016llx\n",
23222 + atomic_read(&vxi->vx_usecnt),
23223 + atomic_read(&vxi->vx_tasks),
23224 + (unsigned long long)vxi->vx_flags);
23226 + buffer += sprintf(buffer, "BCaps:\t");
23227 + buffer = print_cap_t(buffer, &vxi->vx_bcaps);
23228 + buffer += sprintf(buffer, "\n");
23230 + buffer += sprintf(buffer,
23231 + "CCaps:\t%016llx\n"
23232 + "Spaces:\t%08lx %08lx\n",
23233 + (unsigned long long)vxi->vx_ccaps,
23234 + vxi->vx_nsmask[0], vxi->vx_nsmask[1]);
23235 + return buffer - orig;
23238 +int proc_vxi_limit(struct vx_info *vxi, char *buffer)
23240 + return vx_info_proc_limit(&vxi->limit, buffer);
23243 +int proc_vxi_sched(struct vx_info *vxi, char *buffer)
23247 + length = vx_info_proc_sched(&vxi->sched, buffer);
23248 + for_each_online_cpu(cpu) {
23249 + length += vx_info_proc_sched_pc(
23250 + &vx_per_cpu(vxi, sched_pc, cpu),
23251 + buffer + length, cpu);
23256 +int proc_vxi_nsproxy0(struct vx_info *vxi, char *buffer)
23258 + return vx_info_proc_nsproxy(vxi->vx_nsproxy[0], buffer);
23261 +int proc_vxi_nsproxy1(struct vx_info *vxi, char *buffer)
23263 + return vx_info_proc_nsproxy(vxi->vx_nsproxy[1], buffer);
23266 +int proc_vxi_cvirt(struct vx_info *vxi, char *buffer)
23270 + vx_update_load(vxi);
23271 + length = vx_info_proc_cvirt(&vxi->cvirt, buffer);
23272 + for_each_online_cpu(cpu) {
23273 + length += vx_info_proc_cvirt_pc(
23274 + &vx_per_cpu(vxi, cvirt_pc, cpu),
23275 + buffer + length, cpu);
23280 +int proc_vxi_cacct(struct vx_info *vxi, char *buffer)
23282 + return vx_info_proc_cacct(&vxi->cacct, buffer);
23286 +static int proc_virtnet_info(char *buffer)
23288 + return proc_vci(buffer);
23291 +static int proc_virtnet_status(char *buffer)
23293 + return sprintf(buffer,
23295 + "#CActive:\t%d\n",
23296 + atomic_read(&nx_global_ctotal),
23297 + atomic_read(&nx_global_cactive));
23300 +int proc_nxi_info(struct nx_info *nxi, char *buffer)
23302 + struct nx_addr_v4 *v4a;
23303 +#ifdef CONFIG_IPV6
23304 + struct nx_addr_v6 *v6a;
23308 + length = sprintf(buffer,
23311 + "Bcast:\t" NIPQUAD_FMT "\n"
23312 + "Lback:\t" NIPQUAD_FMT "\n",
23315 + NIPQUAD(nxi->v4_bcast.s_addr),
23316 + NIPQUAD(nxi->v4_lback.s_addr));
23318 + if (!NX_IPV4(nxi))
23320 + for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
23321 + length += sprintf(buffer + length, "%d:\t" NXAV4_FMT "\n",
23324 +#ifdef CONFIG_IPV6
23325 + if (!NX_IPV6(nxi))
23327 + for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
23328 + length += sprintf(buffer + length, "%d:\t" NXAV6_FMT "\n",
23335 +int proc_nxi_status(struct nx_info *nxi, char *buffer)
23339 + length = sprintf(buffer,
23342 + "Flags:\t%016llx\n"
23343 + "NCaps:\t%016llx\n",
23344 + atomic_read(&nxi->nx_usecnt),
23345 + atomic_read(&nxi->nx_tasks),
23346 + (unsigned long long)nxi->nx_flags,
23347 + (unsigned long long)nxi->nx_ncaps);
23353 +/* here the inode helpers */
23359 + struct inode_operations *iop;
23360 + struct file_operations *fop;
23361 + union proc_op op;
23364 +static struct inode *vs_proc_make_inode(struct super_block *sb, struct vs_entry *p)
23366 + struct inode *inode = new_inode(sb);
23371 + inode->i_mode = p->mode;
23373 + inode->i_op = p->iop;
23375 + inode->i_fop = p->fop;
23377 + inode->i_nlink = (p->mode & S_IFDIR) ? 2 : 1;
23378 + inode->i_flags |= S_IMMUTABLE;
23380 + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
23382 + inode->i_uid = 0;
23383 + inode->i_gid = 0;
23384 + inode->i_tag = 0;
23389 +static struct dentry *vs_proc_instantiate(struct inode *dir,
23390 + struct dentry *dentry, int id, void *ptr)
23392 + struct vs_entry *p = ptr;
23393 + struct inode *inode = vs_proc_make_inode(dir->i_sb, p);
23394 + struct dentry *error = ERR_PTR(-EINVAL);
23399 + PROC_I(inode)->op = p->op;
23400 + PROC_I(inode)->fd = id;
23401 + d_add(dentry, inode);
23409 +typedef struct dentry *instantiate_t(struct inode *, struct dentry *, int, void *);
23412 + * Fill a directory entry.
23414 + * If possible create the dcache entry and derive our inode number and
23415 + * file type from dcache entry.
23417 + * Since all of the proc inode numbers are dynamically generated, the inode
23418 + * numbers do not exist until the inode is cache. This means creating the
23419 + * the dcache entry in readdir is necessary to keep the inode numbers
23420 + * reported by readdir in sync with the inode numbers reported
23423 +static int proc_fill_cache(struct file *filp, void *dirent, filldir_t filldir,
23424 + char *name, int len, instantiate_t instantiate, int id, void *ptr)
23426 + struct dentry *child, *dir = filp->f_dentry;
23427 + struct inode *inode;
23428 + struct qstr qname;
23430 + unsigned type = DT_UNKNOWN;
23432 + qname.name = name;
23434 + qname.hash = full_name_hash(name, len);
23436 + child = d_lookup(dir, &qname);
23438 + struct dentry *new;
23439 + new = d_alloc(dir, &qname);
23441 + child = instantiate(dir->d_inode, new, id, ptr);
23448 + if (!child || IS_ERR(child) || !child->d_inode)
23449 + goto end_instantiate;
23450 + inode = child->d_inode;
23452 + ino = inode->i_ino;
23453 + type = inode->i_mode >> 12;
23458 + ino = find_inode_number(dir, &qname);
23461 + return filldir(dirent, name, len, filp->f_pos, ino, type);
23466 +/* get and revalidate vx_info/xid */
23469 +struct vx_info *get_proc_vx_info(struct inode *inode)
23471 + return lookup_vx_info(PROC_I(inode)->fd);
23474 +static int proc_xid_revalidate(struct dentry *dentry, struct nameidata *nd)
23476 + struct inode *inode = dentry->d_inode;
23477 + xid_t xid = PROC_I(inode)->fd;
23479 + if (!xid || xid_is_hashed(xid))
23486 +/* get and revalidate nx_info/nid */
23488 +static int proc_nid_revalidate(struct dentry *dentry, struct nameidata *nd)
23490 + struct inode *inode = dentry->d_inode;
23491 + nid_t nid = PROC_I(inode)->fd;
23493 + if (!nid || nid_is_hashed(nid))
23501 +#define PROC_BLOCK_SIZE (PAGE_SIZE - 1024)
23503 +static ssize_t proc_vs_info_read(struct file *file, char __user *buf,
23504 + size_t count, loff_t *ppos)
23506 + struct inode *inode = file->f_dentry->d_inode;
23507 + unsigned long page;
23508 + ssize_t length = 0;
23510 + if (count > PROC_BLOCK_SIZE)
23511 + count = PROC_BLOCK_SIZE;
23513 + /* fade that out as soon as stable */
23514 + WARN_ON(PROC_I(inode)->fd);
23516 + if (!(page = __get_free_page(GFP_KERNEL)))
23519 + BUG_ON(!PROC_I(inode)->op.proc_vs_read);
23520 + length = PROC_I(inode)->op.proc_vs_read((char *)page);
23523 + length = simple_read_from_buffer(buf, count, ppos,
23524 + (char *)page, length);
23530 +static ssize_t proc_vx_info_read(struct file *file, char __user *buf,
23531 + size_t count, loff_t *ppos)
23533 + struct inode *inode = file->f_dentry->d_inode;
23534 + struct vx_info *vxi = NULL;
23535 + xid_t xid = PROC_I(inode)->fd;
23536 + unsigned long page;
23537 + ssize_t length = 0;
23539 + if (count > PROC_BLOCK_SIZE)
23540 + count = PROC_BLOCK_SIZE;
23542 + /* fade that out as soon as stable */
23544 + vxi = lookup_vx_info(xid);
23548 + length = -ENOMEM;
23549 + if (!(page = __get_free_page(GFP_KERNEL)))
23552 + BUG_ON(!PROC_I(inode)->op.proc_vxi_read);
23553 + length = PROC_I(inode)->op.proc_vxi_read(vxi, (char *)page);
23556 + length = simple_read_from_buffer(buf, count, ppos,
23557 + (char *)page, length);
23561 + put_vx_info(vxi);
23566 +static ssize_t proc_nx_info_read(struct file *file, char __user *buf,
23567 + size_t count, loff_t *ppos)
23569 + struct inode *inode = file->f_dentry->d_inode;
23570 + struct nx_info *nxi = NULL;
23571 + nid_t nid = PROC_I(inode)->fd;
23572 + unsigned long page;
23573 + ssize_t length = 0;
23575 + if (count > PROC_BLOCK_SIZE)
23576 + count = PROC_BLOCK_SIZE;
23578 + /* fade that out as soon as stable */
23580 + nxi = lookup_nx_info(nid);
23584 + length = -ENOMEM;
23585 + if (!(page = __get_free_page(GFP_KERNEL)))
23588 + BUG_ON(!PROC_I(inode)->op.proc_nxi_read);
23589 + length = PROC_I(inode)->op.proc_nxi_read(nxi, (char *)page);
23592 + length = simple_read_from_buffer(buf, count, ppos,
23593 + (char *)page, length);
23597 + put_nx_info(nxi);
23604 +/* here comes the lower level */
23607 +#define NOD(NAME, MODE, IOP, FOP, OP) { \
23608 + .len = sizeof(NAME) - 1, \
23609 + .name = (NAME), \
23617 +#define DIR(NAME, MODE, OTYPE) \
23618 + NOD(NAME, (S_IFDIR | (MODE)), \
23619 + &proc_ ## OTYPE ## _inode_operations, \
23620 + &proc_ ## OTYPE ## _file_operations, { } )
23622 +#define INF(NAME, MODE, OTYPE) \
23623 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23624 + &proc_vs_info_file_operations, \
23625 + { .proc_vs_read = &proc_##OTYPE } )
23627 +#define VINF(NAME, MODE, OTYPE) \
23628 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23629 + &proc_vx_info_file_operations, \
23630 + { .proc_vxi_read = &proc_##OTYPE } )
23632 +#define NINF(NAME, MODE, OTYPE) \
23633 + NOD(NAME, (S_IFREG | (MODE)), NULL, \
23634 + &proc_nx_info_file_operations, \
23635 + { .proc_nxi_read = &proc_##OTYPE } )
23638 +static struct file_operations proc_vs_info_file_operations = {
23639 + .read = proc_vs_info_read,
23642 +static struct file_operations proc_vx_info_file_operations = {
23643 + .read = proc_vx_info_read,
23646 +static struct dentry_operations proc_xid_dentry_operations = {
23647 + .d_revalidate = proc_xid_revalidate,
23650 +static struct vs_entry vx_base_stuff[] = {
23651 + VINF("info", S_IRUGO, vxi_info),
23652 + VINF("status", S_IRUGO, vxi_status),
23653 + VINF("limit", S_IRUGO, vxi_limit),
23654 + VINF("sched", S_IRUGO, vxi_sched),
23655 + VINF("nsproxy", S_IRUGO, vxi_nsproxy0),
23656 + VINF("nsproxy1",S_IRUGO, vxi_nsproxy1),
23657 + VINF("cvirt", S_IRUGO, vxi_cvirt),
23658 + VINF("cacct", S_IRUGO, vxi_cacct),
23665 +static struct dentry *proc_xid_instantiate(struct inode *dir,
23666 + struct dentry *dentry, int id, void *ptr)
23668 + dentry->d_op = &proc_xid_dentry_operations;
23669 + return vs_proc_instantiate(dir, dentry, id, ptr);
23672 +static struct dentry *proc_xid_lookup(struct inode *dir,
23673 + struct dentry *dentry, struct nameidata *nd)
23675 + struct vs_entry *p = vx_base_stuff;
23676 + struct dentry *error = ERR_PTR(-ENOENT);
23678 + for (; p->name; p++) {
23679 + if (p->len != dentry->d_name.len)
23681 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23687 + error = proc_xid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23692 +static int proc_xid_readdir(struct file *filp,
23693 + void *dirent, filldir_t filldir)
23695 + struct dentry *dentry = filp->f_dentry;
23696 + struct inode *inode = dentry->d_inode;
23697 + struct vs_entry *p = vx_base_stuff;
23698 + int size = sizeof(vx_base_stuff) / sizeof(struct vs_entry);
23702 + pos = filp->f_pos;
23705 + ino = inode->i_ino;
23706 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23709 + /* fall through */
23711 + ino = parent_ino(dentry);
23712 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23715 + /* fall through */
23718 + if (index >= size)
23720 + for (p += index; p->name; p++) {
23721 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23722 + vs_proc_instantiate, PROC_I(inode)->fd, p))
23728 + filp->f_pos = pos;
23734 +static struct file_operations proc_nx_info_file_operations = {
23735 + .read = proc_nx_info_read,
23738 +static struct dentry_operations proc_nid_dentry_operations = {
23739 + .d_revalidate = proc_nid_revalidate,
23742 +static struct vs_entry nx_base_stuff[] = {
23743 + NINF("info", S_IRUGO, nxi_info),
23744 + NINF("status", S_IRUGO, nxi_status),
23749 +static struct dentry *proc_nid_instantiate(struct inode *dir,
23750 + struct dentry *dentry, int id, void *ptr)
23752 + dentry->d_op = &proc_nid_dentry_operations;
23753 + return vs_proc_instantiate(dir, dentry, id, ptr);
23756 +static struct dentry *proc_nid_lookup(struct inode *dir,
23757 + struct dentry *dentry, struct nameidata *nd)
23759 + struct vs_entry *p = nx_base_stuff;
23760 + struct dentry *error = ERR_PTR(-ENOENT);
23762 + for (; p->name; p++) {
23763 + if (p->len != dentry->d_name.len)
23765 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23771 + error = proc_nid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23776 +static int proc_nid_readdir(struct file *filp,
23777 + void *dirent, filldir_t filldir)
23779 + struct dentry *dentry = filp->f_dentry;
23780 + struct inode *inode = dentry->d_inode;
23781 + struct vs_entry *p = nx_base_stuff;
23782 + int size = sizeof(nx_base_stuff) / sizeof(struct vs_entry);
23786 + pos = filp->f_pos;
23789 + ino = inode->i_ino;
23790 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23793 + /* fall through */
23795 + ino = parent_ino(dentry);
23796 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23799 + /* fall through */
23802 + if (index >= size)
23804 + for (p += index; p->name; p++) {
23805 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23806 + vs_proc_instantiate, PROC_I(inode)->fd, p))
23812 + filp->f_pos = pos;
23817 +#define MAX_MULBY10 ((~0U - 9) / 10)
23819 +static inline int atovid(const char *str, int len)
23824 + while (len-- > 0) {
23829 + if (vid >= MAX_MULBY10)
23839 +/* now the upper level (virtual) */
23842 +static struct file_operations proc_xid_file_operations = {
23843 + .read = generic_read_dir,
23844 + .readdir = proc_xid_readdir,
23847 +static struct inode_operations proc_xid_inode_operations = {
23848 + .lookup = proc_xid_lookup,
23851 +static struct vs_entry vx_virtual_stuff[] = {
23852 + INF("info", S_IRUGO, virtual_info),
23853 + INF("status", S_IRUGO, virtual_status),
23854 + DIR(NULL, S_IRUGO | S_IXUGO, xid),
23858 +static struct dentry *proc_virtual_lookup(struct inode *dir,
23859 + struct dentry *dentry, struct nameidata *nd)
23861 + struct vs_entry *p = vx_virtual_stuff;
23862 + struct dentry *error = ERR_PTR(-ENOENT);
23865 + for (; p->name; p++) {
23866 + if (p->len != dentry->d_name.len)
23868 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23872 + goto instantiate;
23874 + id = atovid(dentry->d_name.name, dentry->d_name.len);
23875 + if ((id < 0) || !xid_is_hashed(id))
23879 + error = proc_xid_instantiate(dir, dentry, id, p);
23884 +static struct file_operations proc_nid_file_operations = {
23885 + .read = generic_read_dir,
23886 + .readdir = proc_nid_readdir,
23889 +static struct inode_operations proc_nid_inode_operations = {
23890 + .lookup = proc_nid_lookup,
23893 +static struct vs_entry nx_virtnet_stuff[] = {
23894 + INF("info", S_IRUGO, virtnet_info),
23895 + INF("status", S_IRUGO, virtnet_status),
23896 + DIR(NULL, S_IRUGO | S_IXUGO, nid),
23900 +static struct dentry *proc_virtnet_lookup(struct inode *dir,
23901 + struct dentry *dentry, struct nameidata *nd)
23903 + struct vs_entry *p = nx_virtnet_stuff;
23904 + struct dentry *error = ERR_PTR(-ENOENT);
23907 + for (; p->name; p++) {
23908 + if (p->len != dentry->d_name.len)
23910 + if (!memcmp(dentry->d_name.name, p->name, p->len))
23914 + goto instantiate;
23916 + id = atovid(dentry->d_name.name, dentry->d_name.len);
23917 + if ((id < 0) || !nid_is_hashed(id))
23921 + error = proc_nid_instantiate(dir, dentry, id, p);
23927 +#define PROC_MAXVIDS 32
23929 +int proc_virtual_readdir(struct file *filp,
23930 + void *dirent, filldir_t filldir)
23932 + struct dentry *dentry = filp->f_dentry;
23933 + struct inode *inode = dentry->d_inode;
23934 + struct vs_entry *p = vx_virtual_stuff;
23935 + int size = sizeof(vx_virtual_stuff) / sizeof(struct vs_entry);
23937 + unsigned int xid_array[PROC_MAXVIDS];
23938 + char buf[PROC_NUMBUF];
23939 + unsigned int nr_xids, i;
23942 + pos = filp->f_pos;
23945 + ino = inode->i_ino;
23946 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23949 + /* fall through */
23951 + ino = parent_ino(dentry);
23952 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23955 + /* fall through */
23958 + if (index >= size)
23960 + for (p += index; p->name; p++) {
23961 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23962 + vs_proc_instantiate, 0, p))
23967 + index = pos - size;
23968 + p = &vx_virtual_stuff[size - 1];
23969 + nr_xids = get_xid_list(index, xid_array, PROC_MAXVIDS);
23970 + for (i = 0; i < nr_xids; i++) {
23971 + int n, xid = xid_array[i];
23972 + unsigned int j = PROC_NUMBUF;
23976 + buf[--j] = '0' + (n % 10);
23979 + if (proc_fill_cache(filp, dirent, filldir,
23980 + buf + j, PROC_NUMBUF - j,
23981 + vs_proc_instantiate, xid, p))
23987 + filp->f_pos = pos;
23991 +static int proc_virtual_getattr(struct vfsmount *mnt,
23992 + struct dentry *dentry, struct kstat *stat)
23994 + struct inode *inode = dentry->d_inode;
23996 + generic_fillattr(inode, stat);
23997 + stat->nlink = 2 + atomic_read(&vx_global_cactive);
24001 +static struct file_operations proc_virtual_dir_operations = {
24002 + .read = generic_read_dir,
24003 + .readdir = proc_virtual_readdir,
24006 +static struct inode_operations proc_virtual_dir_inode_operations = {
24007 + .getattr = proc_virtual_getattr,
24008 + .lookup = proc_virtual_lookup,
24015 +int proc_virtnet_readdir(struct file *filp,
24016 + void *dirent, filldir_t filldir)
24018 + struct dentry *dentry = filp->f_dentry;
24019 + struct inode *inode = dentry->d_inode;
24020 + struct vs_entry *p = nx_virtnet_stuff;
24021 + int size = sizeof(nx_virtnet_stuff) / sizeof(struct vs_entry);
24023 + unsigned int nid_array[PROC_MAXVIDS];
24024 + char buf[PROC_NUMBUF];
24025 + unsigned int nr_nids, i;
24028 + pos = filp->f_pos;
24031 + ino = inode->i_ino;
24032 + if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
24035 + /* fall through */
24037 + ino = parent_ino(dentry);
24038 + if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
24041 + /* fall through */
24044 + if (index >= size)
24046 + for (p += index; p->name; p++) {
24047 + if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
24048 + vs_proc_instantiate, 0, p))
24053 + index = pos - size;
24054 + p = &nx_virtnet_stuff[size - 1];
24055 + nr_nids = get_nid_list(index, nid_array, PROC_MAXVIDS);
24056 + for (i = 0; i < nr_nids; i++) {
24057 + int n, nid = nid_array[i];
24058 + unsigned int j = PROC_NUMBUF;
24062 + buf[--j] = '0' + (n % 10);
24065 + if (proc_fill_cache(filp, dirent, filldir,
24066 + buf + j, PROC_NUMBUF - j,
24067 + vs_proc_instantiate, nid, p))
24073 + filp->f_pos = pos;
24077 +static int proc_virtnet_getattr(struct vfsmount *mnt,
24078 + struct dentry *dentry, struct kstat *stat)
24080 + struct inode *inode = dentry->d_inode;
24082 + generic_fillattr(inode, stat);
24083 + stat->nlink = 2 + atomic_read(&nx_global_cactive);
24087 +static struct file_operations proc_virtnet_dir_operations = {
24088 + .read = generic_read_dir,
24089 + .readdir = proc_virtnet_readdir,
24092 +static struct inode_operations proc_virtnet_dir_inode_operations = {
24093 + .getattr = proc_virtnet_getattr,
24094 + .lookup = proc_virtnet_lookup,
24099 +void proc_vx_init(void)
24101 + struct proc_dir_entry *ent;
24103 + ent = proc_mkdir("virtual", 0);
24105 + ent->proc_fops = &proc_virtual_dir_operations;
24106 + ent->proc_iops = &proc_virtual_dir_inode_operations;
24108 + proc_virtual = ent;
24110 + ent = proc_mkdir("virtnet", 0);
24112 + ent->proc_fops = &proc_virtnet_dir_operations;
24113 + ent->proc_iops = &proc_virtnet_dir_inode_operations;
24115 + proc_virtnet = ent;
24121 +/* per pid info */
24124 +int proc_pid_vx_info(struct task_struct *p, char *buffer)
24126 + struct vx_info *vxi;
24127 + char *orig = buffer;
24129 + buffer += sprintf(buffer, "XID:\t%d\n", vx_task_xid(p));
24131 + vxi = task_get_vx_info(p);
24135 + buffer += sprintf(buffer, "BCaps:\t");
24136 + buffer = print_cap_t(buffer, &vxi->vx_bcaps);
24137 + buffer += sprintf(buffer, "\n");
24138 + buffer += sprintf(buffer, "CCaps:\t%016llx\n",
24139 + (unsigned long long)vxi->vx_ccaps);
24140 + buffer += sprintf(buffer, "CFlags:\t%016llx\n",
24141 + (unsigned long long)vxi->vx_flags);
24142 + buffer += sprintf(buffer, "CIPid:\t%d\n", vxi->vx_initpid);
24144 + put_vx_info(vxi);
24146 + return buffer - orig;
24150 +int proc_pid_nx_info(struct task_struct *p, char *buffer)
24152 + struct nx_info *nxi;
24153 + struct nx_addr_v4 *v4a;
24154 +#ifdef CONFIG_IPV6
24155 + struct nx_addr_v6 *v6a;
24157 + char *orig = buffer;
24160 + buffer += sprintf(buffer, "NID:\t%d\n", nx_task_nid(p));
24162 + nxi = task_get_nx_info(p);
24166 + buffer += sprintf(buffer, "NCaps:\t%016llx\n",
24167 + (unsigned long long)nxi->nx_ncaps);
24168 + buffer += sprintf(buffer, "NFlags:\t%016llx\n",
24169 + (unsigned long long)nxi->nx_flags);
24171 + buffer += sprintf(buffer,
24172 + "V4Root[bcast]:\t" NIPQUAD_FMT "\n",
24173 + NIPQUAD(nxi->v4_bcast.s_addr));
24174 + buffer += sprintf (buffer,
24175 + "V4Root[lback]:\t" NIPQUAD_FMT "\n",
24176 + NIPQUAD(nxi->v4_lback.s_addr));
24177 + if (!NX_IPV4(nxi))
24179 + for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
24180 + buffer += sprintf(buffer, "V4Root[%d]:\t" NXAV4_FMT "\n",
24183 +#ifdef CONFIG_IPV6
24184 + if (!NX_IPV6(nxi))
24186 + for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
24187 + buffer += sprintf(buffer, "V6Root[%d]:\t" NXAV6_FMT "\n",
24191 + put_nx_info(nxi);
24193 + return buffer - orig;
24196 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/sched.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched.c
24197 --- linux-2.6.32.1/kernel/vserver/sched.c 1970-01-01 01:00:00.000000000 +0100
24198 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched.c 2009-12-03 20:04:56.000000000 +0100
24201 + * linux/kernel/vserver/sched.c
24203 + * Virtual Server: Scheduler Support
24205 + * Copyright (C) 2004-2007 Herbert Pötzl
24207 + * V0.01 adapted Sam Vilains version to 2.6.3
24208 + * V0.02 removed legacy interface
24209 + * V0.03 changed vcmds to vxi arg
24210 + * V0.04 removed older and legacy interfaces
24214 +#include <linux/vs_context.h>
24215 +#include <linux/vs_sched.h>
24216 +#include <linux/vserver/sched_cmd.h>
24218 +#include <asm/uaccess.h>
24221 +#define vxd_check_range(val, min, max) do { \
24222 + vxlprintk((val < min) || (val > max), \
24223 + "check_range(%ld,%ld,%ld)", \
24224 + (long)val, (long)min, (long)max, \
24225 + __FILE__, __LINE__); \
24229 +void vx_update_sched_param(struct _vx_sched *sched,
24230 + struct _vx_sched_pc *sched_pc)
24232 + unsigned int set_mask = sched->update_mask;
24234 + if (set_mask & VXSM_FILL_RATE)
24235 + sched_pc->fill_rate[0] = sched->fill_rate[0];
24236 + if (set_mask & VXSM_INTERVAL)
24237 + sched_pc->interval[0] = sched->interval[0];
24238 + if (set_mask & VXSM_FILL_RATE2)
24239 + sched_pc->fill_rate[1] = sched->fill_rate[1];
24240 + if (set_mask & VXSM_INTERVAL2)
24241 + sched_pc->interval[1] = sched->interval[1];
24242 + if (set_mask & VXSM_TOKENS)
24243 + sched_pc->tokens = sched->tokens;
24244 + if (set_mask & VXSM_TOKENS_MIN)
24245 + sched_pc->tokens_min = sched->tokens_min;
24246 + if (set_mask & VXSM_TOKENS_MAX)
24247 + sched_pc->tokens_max = sched->tokens_max;
24248 + if (set_mask & VXSM_PRIO_BIAS)
24249 + sched_pc->prio_bias = sched->prio_bias;
24251 + if (set_mask & VXSM_IDLE_TIME)
24252 + sched_pc->flags |= VXSF_IDLE_TIME;
24254 + sched_pc->flags &= ~VXSF_IDLE_TIME;
24257 + sched_pc->norm_time = jiffies;
24262 + * recalculate the context's scheduling tokens
24264 + * ret > 0 : number of tokens available
24265 + * ret < 0 : on hold, check delta_min[]
24266 + * -1 only jiffies
24267 + * -2 also idle time
24270 +int vx_tokens_recalc(struct _vx_sched_pc *sched_pc,
24271 + unsigned long *norm_time, unsigned long *idle_time, int delta_min[2])
24275 + int flags = sched_pc->flags;
24277 + /* how much time did pass? */
24278 + delta = *norm_time - sched_pc->norm_time;
24279 + // printk("@ %ld, %ld, %ld\n", *norm_time, sched_pc->norm_time, jiffies);
24280 + vxd_check_range(delta, 0, INT_MAX);
24282 + if (delta >= sched_pc->interval[0]) {
24283 + long tokens, integral;
24285 + /* calc integral token part */
24286 + tokens = delta / sched_pc->interval[0];
24287 + integral = tokens * sched_pc->interval[0];
24288 + tokens *= sched_pc->fill_rate[0];
24289 +#ifdef CONFIG_VSERVER_HARDCPU
24290 + delta_min[0] = delta - integral;
24291 + vxd_check_range(delta_min[0], 0, sched_pc->interval[0]);
24293 + /* advance time */
24294 + sched_pc->norm_time += delta;
24297 + sched_pc->tokens += tokens;
24298 + sched_pc->token_time += tokens;
24300 + delta_min[0] = delta;
24302 +#ifdef CONFIG_VSERVER_IDLETIME
24303 + if (!(flags & VXSF_IDLE_TIME))
24306 + /* how much was the idle skip? */
24307 + delta = *idle_time - sched_pc->idle_time;
24308 + vxd_check_range(delta, 0, INT_MAX);
24310 + if (delta >= sched_pc->interval[1]) {
24311 + long tokens, integral;
24313 + /* calc fair share token part */
24314 + tokens = delta / sched_pc->interval[1];
24315 + integral = tokens * sched_pc->interval[1];
24316 + tokens *= sched_pc->fill_rate[1];
24317 + delta_min[1] = delta - integral;
24318 + vxd_check_range(delta_min[1], 0, sched_pc->interval[1]);
24320 + /* advance idle time */
24321 + sched_pc->idle_time += integral;
24324 + sched_pc->tokens += tokens;
24325 + sched_pc->token_time += tokens;
24327 + delta_min[1] = delta;
24331 + /* clip at maximum */
24332 + if (sched_pc->tokens > sched_pc->tokens_max)
24333 + sched_pc->tokens = sched_pc->tokens_max;
24334 + tokens = sched_pc->tokens;
24336 + if ((flags & VXSF_ONHOLD)) {
24337 + /* can we unhold? */
24338 + if (tokens >= sched_pc->tokens_min) {
24339 + flags &= ~VXSF_ONHOLD;
24340 + sched_pc->hold_ticks +=
24341 + *norm_time - sched_pc->onhold;
24345 + /* put on hold? */
24346 + if (tokens <= 0) {
24347 + flags |= VXSF_ONHOLD;
24348 + sched_pc->onhold = *norm_time;
24352 + sched_pc->flags = flags;
24356 + tokens = sched_pc->tokens_min - tokens;
24357 + sched_pc->flags = flags;
24358 + // BUG_ON(tokens < 0); probably doesn't hold anymore
24360 +#ifdef CONFIG_VSERVER_HARDCPU
24361 + /* next interval? */
24362 + if (!sched_pc->fill_rate[0])
24363 + delta_min[0] = HZ;
24364 + else if (tokens > sched_pc->fill_rate[0])
24365 + delta_min[0] += sched_pc->interval[0] *
24366 + tokens / sched_pc->fill_rate[0];
24368 + delta_min[0] = sched_pc->interval[0] - delta_min[0];
24369 + vxd_check_range(delta_min[0], 0, INT_MAX);
24371 +#ifdef CONFIG_VSERVER_IDLETIME
24372 + if (!(flags & VXSF_IDLE_TIME))
24375 + /* next interval? */
24376 + if (!sched_pc->fill_rate[1])
24377 + delta_min[1] = HZ;
24378 + else if (tokens > sched_pc->fill_rate[1])
24379 + delta_min[1] += sched_pc->interval[1] *
24380 + tokens / sched_pc->fill_rate[1];
24382 + delta_min[1] = sched_pc->interval[1] - delta_min[1];
24383 + vxd_check_range(delta_min[1], 0, INT_MAX);
24388 +#endif /* CONFIG_VSERVER_IDLETIME */
24391 +#endif /* CONFIG_VSERVER_HARDCPU */
24394 +static inline unsigned long msec_to_ticks(unsigned long msec)
24396 + return msecs_to_jiffies(msec);
24399 +static inline unsigned long ticks_to_msec(unsigned long ticks)
24401 + return jiffies_to_msecs(ticks);
24404 +static inline unsigned long ticks_to_usec(unsigned long ticks)
24406 + return jiffies_to_usecs(ticks);
24410 +static int do_set_sched(struct vx_info *vxi, struct vcmd_sched_v5 *data)
24412 + unsigned int set_mask = data->mask;
24413 + unsigned int update_mask;
24416 + /* Sanity check data values */
24417 + if (data->tokens_max <= 0)
24418 + data->tokens_max = HZ;
24419 + if (data->tokens_min < 0)
24420 + data->tokens_min = HZ / 3;
24421 + if (data->tokens_min >= data->tokens_max)
24422 + data->tokens_min = data->tokens_max;
24424 + if (data->prio_bias > MAX_PRIO_BIAS)
24425 + data->prio_bias = MAX_PRIO_BIAS;
24426 + if (data->prio_bias < MIN_PRIO_BIAS)
24427 + data->prio_bias = MIN_PRIO_BIAS;
24429 + spin_lock(&vxi->sched.tokens_lock);
24431 + /* sync up on delayed updates */
24432 + for_each_cpu_mask(cpu, vxi->sched.update)
24433 + vx_update_sched_param(&vxi->sched,
24434 + &vx_per_cpu(vxi, sched_pc, cpu));
24436 + if (set_mask & VXSM_FILL_RATE)
24437 + vxi->sched.fill_rate[0] = data->fill_rate[0];
24438 + if (set_mask & VXSM_FILL_RATE2)
24439 + vxi->sched.fill_rate[1] = data->fill_rate[1];
24440 + if (set_mask & VXSM_INTERVAL)
24441 + vxi->sched.interval[0] = (set_mask & VXSM_MSEC) ?
24442 + msec_to_ticks(data->interval[0]) : data->interval[0];
24443 + if (set_mask & VXSM_INTERVAL2)
24444 + vxi->sched.interval[1] = (set_mask & VXSM_MSEC) ?
24445 + msec_to_ticks(data->interval[1]) : data->interval[1];
24446 + if (set_mask & VXSM_TOKENS)
24447 + vxi->sched.tokens = data->tokens;
24448 + if (set_mask & VXSM_TOKENS_MIN)
24449 + vxi->sched.tokens_min = data->tokens_min;
24450 + if (set_mask & VXSM_TOKENS_MAX)
24451 + vxi->sched.tokens_max = data->tokens_max;
24452 + if (set_mask & VXSM_PRIO_BIAS)
24453 + vxi->sched.prio_bias = data->prio_bias;
24455 + /* Sanity check rate/interval */
24456 + for (i = 0; i < 2; i++) {
24457 + if (data->fill_rate[i] < 0)
24458 + data->fill_rate[i] = 0;
24459 + if (data->interval[i] <= 0)
24460 + data->interval[i] = HZ;
24463 + update_mask = vxi->sched.update_mask & VXSM_SET_MASK;
24464 + update_mask |= (set_mask & (VXSM_SET_MASK | VXSM_IDLE_TIME));
24465 + vxi->sched.update_mask = update_mask;
24469 + if (set_mask & VXSM_CPU_ID) {
24470 + vxi->sched.update = cpumask_of_cpu(data->cpu_id);
24471 + cpus_and(vxi->sched.update, cpu_online_map,
24472 + vxi->sched.update);
24474 + vxi->sched.update = cpu_online_map;
24476 + /* forced reload? */
24477 + if (set_mask & VXSM_FORCE) {
24478 + for_each_cpu_mask(cpu, vxi->sched.update)
24479 + vx_update_sched_param(&vxi->sched,
24480 + &vx_per_cpu(vxi, sched_pc, cpu));
24481 + vxi->sched.update = CPU_MASK_NONE;
24484 + /* on UP we update immediately */
24485 + vx_update_sched_param(&vxi->sched,
24486 + &vx_per_cpu(vxi, sched_pc, 0));
24489 + spin_unlock(&vxi->sched.tokens_lock);
24494 +#define COPY_IDS(C) C(cpu_id); C(bucket_id)
24495 +#define COPY_PRI(C) C(prio_bias)
24496 +#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
24497 +#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
24498 + C(fill_rate[1]); C(interval[1]);
24500 +#define COPY_VALUE(name) vc_data.name = data->name
24502 +static int do_set_sched_v4(struct vx_info *vxi, struct vcmd_set_sched_v4 *data)
24504 + struct vcmd_sched_v5 vc_data;
24506 + vc_data.mask = data->set_mask;
24507 + COPY_IDS(COPY_VALUE);
24508 + COPY_PRI(COPY_VALUE);
24509 + COPY_TOK(COPY_VALUE);
24510 + vc_data.fill_rate[0] = vc_data.fill_rate[1] = data->fill_rate;
24511 + vc_data.interval[0] = vc_data.interval[1] = data->interval;
24512 + return do_set_sched(vxi, &vc_data);
24515 +int vc_set_sched_v4(struct vx_info *vxi, void __user *data)
24517 + struct vcmd_set_sched_v4 vc_data;
24519 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24522 + return do_set_sched_v4(vxi, &vc_data);
24525 + /* latest interface is v5 */
24527 +int vc_set_sched(struct vx_info *vxi, void __user *data)
24529 + struct vcmd_sched_v5 vc_data;
24531 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24534 + return do_set_sched(vxi, &vc_data);
24538 +#define COPY_PRI(C) C(prio_bias)
24539 +#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
24540 +#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
24541 + C(fill_rate[1]); C(interval[1]);
24543 +#define COPY_VALUE(name) vc_data.name = data->name
24546 +int vc_get_sched(struct vx_info *vxi, void __user *data)
24548 + struct vcmd_sched_v5 vc_data;
24550 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24553 + if (vc_data.mask & VXSM_CPU_ID) {
24554 + int cpu = vc_data.cpu_id;
24555 + struct _vx_sched_pc *data;
24557 + if (!cpu_possible(cpu))
24560 + data = &vx_per_cpu(vxi, sched_pc, cpu);
24561 + COPY_TOK(COPY_VALUE);
24562 + COPY_PRI(COPY_VALUE);
24563 + COPY_FRI(COPY_VALUE);
24565 + if (data->flags & VXSF_IDLE_TIME)
24566 + vc_data.mask |= VXSM_IDLE_TIME;
24568 + struct _vx_sched *data = &vxi->sched;
24570 + COPY_TOK(COPY_VALUE);
24571 + COPY_PRI(COPY_VALUE);
24572 + COPY_FRI(COPY_VALUE);
24575 + if (vc_data.mask & VXSM_MSEC) {
24576 + vc_data.interval[0] = ticks_to_msec(vc_data.interval[0]);
24577 + vc_data.interval[1] = ticks_to_msec(vc_data.interval[1]);
24580 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24586 +int vc_sched_info(struct vx_info *vxi, void __user *data)
24588 + struct vcmd_sched_info vc_data;
24591 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24594 + cpu = vc_data.cpu_id;
24595 + if (!cpu_possible(cpu))
24599 + struct _vx_sched_pc *sched_pc =
24600 + &vx_per_cpu(vxi, sched_pc, cpu);
24602 + vc_data.user_msec = ticks_to_msec(sched_pc->user_ticks);
24603 + vc_data.sys_msec = ticks_to_msec(sched_pc->sys_ticks);
24604 + vc_data.hold_msec = ticks_to_msec(sched_pc->hold_ticks);
24605 + vc_data.vavavoom = sched_pc->vavavoom;
24607 + vc_data.token_usec = ticks_to_usec(1);
24609 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24614 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/sched_init.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched_init.h
24615 --- linux-2.6.32.1/kernel/vserver/sched_init.h 1970-01-01 01:00:00.000000000 +0100
24616 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched_init.h 2009-12-03 20:04:56.000000000 +0100
24619 +static inline void vx_info_init_sched(struct _vx_sched *sched)
24621 + static struct lock_class_key tokens_lock_key;
24623 + /* scheduling; hard code starting values as constants */
24624 + sched->fill_rate[0] = 1;
24625 + sched->interval[0] = 4;
24626 + sched->fill_rate[1] = 1;
24627 + sched->interval[1] = 8;
24628 + sched->tokens = HZ >> 2;
24629 + sched->tokens_min = HZ >> 4;
24630 + sched->tokens_max = HZ >> 1;
24631 + sched->tokens_lock = SPIN_LOCK_UNLOCKED;
24632 + sched->prio_bias = 0;
24634 + lockdep_set_class(&sched->tokens_lock, &tokens_lock_key);
24638 +void vx_info_init_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24640 + sched_pc->fill_rate[0] = 1;
24641 + sched_pc->interval[0] = 4;
24642 + sched_pc->fill_rate[1] = 1;
24643 + sched_pc->interval[1] = 8;
24644 + sched_pc->tokens = HZ >> 2;
24645 + sched_pc->tokens_min = HZ >> 4;
24646 + sched_pc->tokens_max = HZ >> 1;
24647 + sched_pc->prio_bias = 0;
24648 + sched_pc->vavavoom = 0;
24649 + sched_pc->token_time = 0;
24650 + sched_pc->idle_time = 0;
24651 + sched_pc->norm_time = jiffies;
24653 + sched_pc->user_ticks = 0;
24654 + sched_pc->sys_ticks = 0;
24655 + sched_pc->hold_ticks = 0;
24658 +static inline void vx_info_exit_sched(struct _vx_sched *sched)
24664 +void vx_info_exit_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24668 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/sched_proc.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched_proc.h
24669 --- linux-2.6.32.1/kernel/vserver/sched_proc.h 1970-01-01 01:00:00.000000000 +0100
24670 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sched_proc.h 2009-12-03 20:04:56.000000000 +0100
24672 +#ifndef _VX_SCHED_PROC_H
24673 +#define _VX_SCHED_PROC_H
24677 +int vx_info_proc_sched(struct _vx_sched *sched, char *buffer)
24681 + length += sprintf(buffer,
24682 + "FillRate:\t%8d,%d\n"
24683 + "Interval:\t%8d,%d\n"
24684 + "TokensMin:\t%8d\n"
24685 + "TokensMax:\t%8d\n"
24686 + "PrioBias:\t%8d\n",
24687 + sched->fill_rate[0],
24688 + sched->fill_rate[1],
24689 + sched->interval[0],
24690 + sched->interval[1],
24691 + sched->tokens_min,
24692 + sched->tokens_max,
24693 + sched->prio_bias);
24698 +int vx_info_proc_sched_pc(struct _vx_sched_pc *sched_pc,
24699 + char *buffer, int cpu)
24703 + length += sprintf(buffer + length,
24704 + "cpu %d: %lld %lld %lld %ld %ld", cpu,
24705 + (unsigned long long)sched_pc->user_ticks,
24706 + (unsigned long long)sched_pc->sys_ticks,
24707 + (unsigned long long)sched_pc->hold_ticks,
24708 + sched_pc->token_time,
24709 + sched_pc->idle_time);
24710 + length += sprintf(buffer + length,
24711 + " %c%c %d %d %d %d/%d %d/%d",
24712 + (sched_pc->flags & VXSF_ONHOLD) ? 'H' : 'R',
24713 + (sched_pc->flags & VXSF_IDLE_TIME) ? 'I' : '-',
24714 + sched_pc->tokens,
24715 + sched_pc->tokens_min,
24716 + sched_pc->tokens_max,
24717 + sched_pc->fill_rate[0],
24718 + sched_pc->interval[0],
24719 + sched_pc->fill_rate[1],
24720 + sched_pc->interval[1]);
24721 + length += sprintf(buffer + length,
24723 + sched_pc->prio_bias,
24724 + sched_pc->vavavoom);
24728 +#endif /* _VX_SCHED_PROC_H */
24729 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/signal.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/signal.c
24730 --- linux-2.6.32.1/kernel/vserver/signal.c 1970-01-01 01:00:00.000000000 +0100
24731 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/signal.c 2009-12-03 20:04:56.000000000 +0100
24734 + * linux/kernel/vserver/signal.c
24736 + * Virtual Server: Signal Support
24738 + * Copyright (C) 2003-2007 Herbert Pötzl
24740 + * V0.01 broken out from vcontext V0.05
24741 + * V0.02 changed vcmds to vxi arg
24742 + * V0.03 adjusted siginfo for kill
24746 +#include <asm/uaccess.h>
24748 +#include <linux/vs_context.h>
24749 +#include <linux/vs_pid.h>
24750 +#include <linux/vserver/signal_cmd.h>
24753 +int vx_info_kill(struct vx_info *vxi, int pid, int sig)
24755 + int retval, count = 0;
24756 + struct task_struct *p;
24757 + struct siginfo *sip = SEND_SIG_PRIV;
24760 + vxdprintk(VXD_CBIT(misc, 4),
24761 + "vx_info_kill(%p[#%d],%d,%d)*",
24762 + vxi, vxi->vx_id, pid, sig);
24763 + read_lock(&tasklist_lock);
24767 + for_each_process(p) {
24770 + if (vx_task_xid(p) != vxi->vx_id || p->pid <= 1 ||
24771 + (pid && vxi->vx_initpid == p->pid))
24774 + err = group_send_sig_info(sig, sip, p);
24776 + if (err != -EPERM)
24782 + if (vxi->vx_initpid) {
24783 + pid = vxi->vx_initpid;
24784 + /* for now, only SIGINT to private init ... */
24785 + if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24786 + /* ... as long as there are tasks left */
24787 + (atomic_read(&vxi->vx_tasks) > 1))
24790 + /* fallthrough */
24792 + p = find_task_by_real_pid(pid);
24794 + if (vx_task_xid(p) == vxi->vx_id)
24795 + retval = group_send_sig_info(sig, sip, p);
24799 + read_unlock(&tasklist_lock);
24800 + vxdprintk(VXD_CBIT(misc, 4),
24801 + "vx_info_kill(%p[#%d],%d,%d,%ld) = %d",
24802 + vxi, vxi->vx_id, pid, sig, (long)sip, retval);
24806 +int vc_ctx_kill(struct vx_info *vxi, void __user *data)
24808 + struct vcmd_ctx_kill_v0 vc_data;
24810 + if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24813 + /* special check to allow guest shutdown */
24814 + if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24815 + /* forbid killall pid=0 when init is present */
24816 + (((vc_data.pid < 1) && vxi->vx_initpid) ||
24817 + (vc_data.pid > 1)))
24820 + return vx_info_kill(vxi, vc_data.pid, vc_data.sig);
24824 +static int __wait_exit(struct vx_info *vxi)
24826 + DECLARE_WAITQUEUE(wait, current);
24829 + add_wait_queue(&vxi->vx_wait, &wait);
24830 + set_current_state(TASK_INTERRUPTIBLE);
24833 + if (vx_info_state(vxi,
24834 + VXS_SHUTDOWN | VXS_HASHED | VXS_HELPER) == VXS_SHUTDOWN)
24836 + if (signal_pending(current)) {
24837 + ret = -ERESTARTSYS;
24844 + set_current_state(TASK_RUNNING);
24845 + remove_wait_queue(&vxi->vx_wait, &wait);
24851 +int vc_wait_exit(struct vx_info *vxi, void __user *data)
24853 + struct vcmd_wait_exit_v0 vc_data;
24856 + ret = __wait_exit(vxi);
24857 + vc_data.reboot_cmd = vxi->reboot_cmd;
24858 + vc_data.exit_code = vxi->exit_code;
24860 + if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24865 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/space.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/space.c
24866 --- linux-2.6.32.1/kernel/vserver/space.c 1970-01-01 01:00:00.000000000 +0100
24867 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/space.c 2009-12-03 20:04:56.000000000 +0100
24870 + * linux/kernel/vserver/space.c
24872 + * Virtual Server: Context Space Support
24874 + * Copyright (C) 2003-2007 Herbert Pötzl
24876 + * V0.01 broken out from context.c 0.07
24877 + * V0.02 added task locking for namespace
24878 + * V0.03 broken out vx_enter_namespace
24879 + * V0.04 added *space support and commands
24883 +#include <linux/utsname.h>
24884 +#include <linux/nsproxy.h>
24885 +#include <linux/err.h>
24886 +#include <linux/fs_struct.h>
24887 +#include <asm/uaccess.h>
24889 +#include <linux/vs_context.h>
24890 +#include <linux/vserver/space.h>
24891 +#include <linux/vserver/space_cmd.h>
24893 +atomic_t vs_global_nsproxy = ATOMIC_INIT(0);
24894 +atomic_t vs_global_fs = ATOMIC_INIT(0);
24895 +atomic_t vs_global_mnt_ns = ATOMIC_INIT(0);
24896 +atomic_t vs_global_uts_ns = ATOMIC_INIT(0);
24897 +atomic_t vs_global_user_ns = ATOMIC_INIT(0);
24898 +atomic_t vs_global_pid_ns = ATOMIC_INIT(0);
24901 +/* namespace functions */
24903 +#include <linux/mnt_namespace.h>
24904 +#include <linux/user_namespace.h>
24905 +#include <linux/pid_namespace.h>
24906 +#include <linux/ipc_namespace.h>
24907 +#include <net/net_namespace.h>
24910 +static const struct vcmd_space_mask_v1 space_mask_v0 = {
24911 + .mask = CLONE_FS |
24919 +static const struct vcmd_space_mask_v1 space_mask = {
24920 + .mask = CLONE_FS |
24925 +#ifdef CONFIG_PID_NS
24928 +#ifdef CONFIG_NET_NS
24934 +static const struct vcmd_space_mask_v1 default_space_mask = {
24935 + .mask = CLONE_FS |
24940 +#ifdef CONFIG_PID_NS
24947 + * build a new nsproxy mix
24948 + * assumes that both proxies are 'const'
24949 + * does not touch nsproxy refcounts
24950 + * will hold a reference on the result.
24953 +struct nsproxy *vs_mix_nsproxy(struct nsproxy *old_nsproxy,
24954 + struct nsproxy *new_nsproxy, unsigned long mask)
24956 + struct mnt_namespace *old_ns;
24957 + struct uts_namespace *old_uts;
24958 + struct ipc_namespace *old_ipc;
24959 +#ifdef CONFIG_PID_NS
24960 + struct pid_namespace *old_pid;
24962 +#ifdef CONFIG_NET_NS
24963 + struct net *old_net;
24965 + struct nsproxy *nsproxy;
24967 + nsproxy = copy_nsproxy(old_nsproxy);
24971 + if (mask & CLONE_NEWNS) {
24972 + old_ns = nsproxy->mnt_ns;
24973 + nsproxy->mnt_ns = new_nsproxy->mnt_ns;
24974 + if (nsproxy->mnt_ns)
24975 + get_mnt_ns(nsproxy->mnt_ns);
24979 + if (mask & CLONE_NEWUTS) {
24980 + old_uts = nsproxy->uts_ns;
24981 + nsproxy->uts_ns = new_nsproxy->uts_ns;
24982 + if (nsproxy->uts_ns)
24983 + get_uts_ns(nsproxy->uts_ns);
24987 + if (mask & CLONE_NEWIPC) {
24988 + old_ipc = nsproxy->ipc_ns;
24989 + nsproxy->ipc_ns = new_nsproxy->ipc_ns;
24990 + if (nsproxy->ipc_ns)
24991 + get_ipc_ns(nsproxy->ipc_ns);
24995 +#ifdef CONFIG_PID_NS
24996 + if (mask & CLONE_NEWPID) {
24997 + old_pid = nsproxy->pid_ns;
24998 + nsproxy->pid_ns = new_nsproxy->pid_ns;
24999 + if (nsproxy->pid_ns)
25000 + get_pid_ns(nsproxy->pid_ns);
25004 +#ifdef CONFIG_NET_NS
25005 + if (mask & CLONE_NEWNET) {
25006 + old_net = nsproxy->net_ns;
25007 + nsproxy->net_ns = new_nsproxy->net_ns;
25008 + if (nsproxy->net_ns)
25009 + get_net(nsproxy->net_ns);
25014 + put_mnt_ns(old_ns);
25016 + put_uts_ns(old_uts);
25018 + put_ipc_ns(old_ipc);
25019 +#ifdef CONFIG_PID_NS
25021 + put_pid_ns(old_pid);
25023 +#ifdef CONFIG_NET_NS
25025 + put_net(old_net);
25033 + * merge two nsproxy structs into a new one.
25034 + * will hold a reference on the result.
25038 +struct nsproxy *__vs_merge_nsproxy(struct nsproxy *old,
25039 + struct nsproxy *proxy, unsigned long mask)
25041 + struct nsproxy null_proxy = { .mnt_ns = NULL };
25047 + /* vs_mix_nsproxy returns with reference */
25048 + return vs_mix_nsproxy(old ? old : &null_proxy,
25051 + get_nsproxy(proxy);
25056 +int vx_enter_space(struct vx_info *vxi, unsigned long mask, unsigned index)
25058 + struct nsproxy *proxy, *proxy_cur, *proxy_new;
25059 + struct fs_struct *fs_cur, *fs = NULL;
25060 + int ret, kill = 0;
25062 + vxdprintk(VXD_CBIT(space, 8), "vx_enter_space(%p[#%u],0x%08lx,%d)",
25063 + vxi, vxi->vx_id, mask, index);
25065 + if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
25069 + mask = vxi->vx_nsmask[index];
25071 + if ((mask & vxi->vx_nsmask[index]) != mask)
25074 + if (mask & CLONE_FS) {
25075 + fs = copy_fs_struct(vxi->vx_fs[index]);
25079 + proxy = vxi->vx_nsproxy[index];
25081 + vxdprintk(VXD_CBIT(space, 9),
25082 + "vx_enter_space(%p[#%u],0x%08lx,%d) -> (%p,%p)",
25083 + vxi, vxi->vx_id, mask, index, proxy, fs);
25085 + task_lock(current);
25086 + fs_cur = current->fs;
25088 + if (mask & CLONE_FS) {
25089 + write_lock(&fs_cur->lock);
25090 + current->fs = fs;
25091 + kill = !--fs_cur->users;
25092 + write_unlock(&fs_cur->lock);
25095 + proxy_cur = current->nsproxy;
25096 + get_nsproxy(proxy_cur);
25097 + task_unlock(current);
25100 + free_fs_struct(fs_cur);
25102 + proxy_new = __vs_merge_nsproxy(proxy_cur, proxy, mask);
25103 + if (IS_ERR(proxy_new)) {
25104 + ret = PTR_ERR(proxy_new);
25108 + proxy_new = xchg(¤t->nsproxy, proxy_new);
25112 + put_nsproxy(proxy_new);
25115 + put_nsproxy(proxy_cur);
25120 +int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index)
25122 + struct nsproxy *proxy_vxi, *proxy_cur, *proxy_new;
25123 + struct fs_struct *fs_vxi, *fs;
25124 + int ret, kill = 0;
25126 + vxdprintk(VXD_CBIT(space, 8), "vx_set_space(%p[#%u],0x%08lx,%d)",
25127 + vxi, vxi->vx_id, mask, index);
25130 + mask = default_space_mask.mask;
25132 + if ((mask & space_mask.mask) != mask)
25135 + proxy_vxi = vxi->vx_nsproxy[index];
25136 + fs_vxi = vxi->vx_fs[index];
25138 + if (mask & CLONE_FS) {
25139 + fs = copy_fs_struct(current->fs);
25144 + task_lock(current);
25146 + if (mask & CLONE_FS) {
25147 + write_lock(&fs_vxi->lock);
25148 + vxi->vx_fs[index] = fs;
25149 + kill = !--fs_vxi->users;
25150 + write_unlock(&fs_vxi->lock);
25153 + proxy_cur = current->nsproxy;
25154 + get_nsproxy(proxy_cur);
25155 + task_unlock(current);
25158 + free_fs_struct(fs_vxi);
25160 + proxy_new = __vs_merge_nsproxy(proxy_vxi, proxy_cur, mask);
25161 + if (IS_ERR(proxy_new)) {
25162 + ret = PTR_ERR(proxy_new);
25166 + proxy_new = xchg(&vxi->vx_nsproxy[index], proxy_new);
25167 + vxi->vx_nsmask[index] |= mask;
25171 + put_nsproxy(proxy_new);
25174 + put_nsproxy(proxy_cur);
25179 +int vc_enter_space_v1(struct vx_info *vxi, void __user *data)
25181 + struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
25183 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
25186 + return vx_enter_space(vxi, vc_data.mask, 0);
25189 +int vc_enter_space(struct vx_info *vxi, void __user *data)
25191 + struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
25193 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
25196 + if (vc_data.index >= VX_SPACES)
25199 + return vx_enter_space(vxi, vc_data.mask, vc_data.index);
25202 +int vc_set_space_v1(struct vx_info *vxi, void __user *data)
25204 + struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
25206 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
25209 + return vx_set_space(vxi, vc_data.mask, 0);
25212 +int vc_set_space(struct vx_info *vxi, void __user *data)
25214 + struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
25216 + if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
25219 + if (vc_data.index >= VX_SPACES)
25222 + return vx_set_space(vxi, vc_data.mask, vc_data.index);
25225 +int vc_get_space_mask(void __user *data, int type)
25227 + const struct vcmd_space_mask_v1 *mask;
25230 + mask = &space_mask_v0;
25231 + else if (type == 1)
25232 + mask = &space_mask;
25234 + mask = &default_space_mask;
25236 + vxdprintk(VXD_CBIT(space, 10),
25237 + "vc_get_space_mask(%d) = %08llx", type, mask->mask);
25239 + if (copy_to_user(data, mask, sizeof(*mask)))
25244 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/switch.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/switch.c
25245 --- linux-2.6.32.1/kernel/vserver/switch.c 1970-01-01 01:00:00.000000000 +0100
25246 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/switch.c 2009-12-03 20:04:56.000000000 +0100
25249 + * linux/kernel/vserver/switch.c
25251 + * Virtual Server: Syscall Switch
25253 + * Copyright (C) 2003-2007 Herbert Pötzl
25255 + * V0.01 syscall switch
25256 + * V0.02 added signal to context
25257 + * V0.03 added rlimit functions
25258 + * V0.04 added iattr, task/xid functions
25259 + * V0.05 added debug/history stuff
25260 + * V0.06 added compat32 layer
25261 + * V0.07 vcmd args and perms
25262 + * V0.08 added status commands
25263 + * V0.09 added tag commands
25264 + * V0.10 added oom bias
25265 + * V0.11 added device commands
25269 +#include <linux/vs_context.h>
25270 +#include <linux/vs_network.h>
25271 +#include <linux/vserver/switch.h>
25273 +#include "vci_config.h"
25277 +int vc_get_version(uint32_t id)
25279 + return VCI_VERSION;
25283 +int vc_get_vci(uint32_t id)
25285 + return vci_kernel_config();
25288 +#include <linux/vserver/context_cmd.h>
25289 +#include <linux/vserver/cvirt_cmd.h>
25290 +#include <linux/vserver/cacct_cmd.h>
25291 +#include <linux/vserver/limit_cmd.h>
25292 +#include <linux/vserver/network_cmd.h>
25293 +#include <linux/vserver/sched_cmd.h>
25294 +#include <linux/vserver/debug_cmd.h>
25295 +#include <linux/vserver/inode_cmd.h>
25296 +#include <linux/vserver/dlimit_cmd.h>
25297 +#include <linux/vserver/signal_cmd.h>
25298 +#include <linux/vserver/space_cmd.h>
25299 +#include <linux/vserver/tag_cmd.h>
25300 +#include <linux/vserver/device_cmd.h>
25302 +#include <linux/vserver/inode.h>
25303 +#include <linux/vserver/dlimit.h>
25306 +#ifdef CONFIG_COMPAT
25307 +#define __COMPAT(name, id, data, compat) \
25308 + (compat) ? name ## _x32(id, data) : name(id, data)
25309 +#define __COMPAT_NO_ID(name, data, compat) \
25310 + (compat) ? name ## _x32(data) : name(data)
25312 +#define __COMPAT(name, id, data, compat) \
25314 +#define __COMPAT_NO_ID(name, data, compat) \
25320 +long do_vcmd(uint32_t cmd, uint32_t id,
25321 + struct vx_info *vxi, struct nx_info *nxi,
25322 + void __user *data, int compat)
25326 + case VCMD_get_version:
25327 + return vc_get_version(id);
25328 + case VCMD_get_vci:
25329 + return vc_get_vci(id);
25331 + case VCMD_task_xid:
25332 + return vc_task_xid(id);
25333 + case VCMD_vx_info:
25334 + return vc_vx_info(vxi, data);
25336 + case VCMD_task_nid:
25337 + return vc_task_nid(id);
25338 + case VCMD_nx_info:
25339 + return vc_nx_info(nxi, data);
25341 + case VCMD_task_tag:
25342 + return vc_task_tag(id);
25344 + case VCMD_set_space_v1:
25345 + return vc_set_space_v1(vxi, data);
25346 + /* this is version 2 */
25347 + case VCMD_set_space:
25348 + return vc_set_space(vxi, data);
25350 + case VCMD_get_space_mask_v0:
25351 + return vc_get_space_mask(data, 0);
25352 + /* this is version 1 */
25353 + case VCMD_get_space_mask:
25354 + return vc_get_space_mask(data, 1);
25356 + case VCMD_get_space_default:
25357 + return vc_get_space_mask(data, -1);
25359 +#ifdef CONFIG_IA32_EMULATION
25360 + case VCMD_get_rlimit:
25361 + return __COMPAT(vc_get_rlimit, vxi, data, compat);
25362 + case VCMD_set_rlimit:
25363 + return __COMPAT(vc_set_rlimit, vxi, data, compat);
25365 + case VCMD_get_rlimit:
25366 + return vc_get_rlimit(vxi, data);
25367 + case VCMD_set_rlimit:
25368 + return vc_set_rlimit(vxi, data);
25370 + case VCMD_get_rlimit_mask:
25371 + return vc_get_rlimit_mask(id, data);
25372 + case VCMD_reset_hits:
25373 + return vc_reset_hits(vxi, data);
25374 + case VCMD_reset_minmax:
25375 + return vc_reset_minmax(vxi, data);
25377 + case VCMD_get_vhi_name:
25378 + return vc_get_vhi_name(vxi, data);
25379 + case VCMD_set_vhi_name:
25380 + return vc_set_vhi_name(vxi, data);
25382 + case VCMD_ctx_stat:
25383 + return vc_ctx_stat(vxi, data);
25384 + case VCMD_virt_stat:
25385 + return vc_virt_stat(vxi, data);
25386 + case VCMD_sock_stat:
25387 + return vc_sock_stat(vxi, data);
25388 + case VCMD_rlimit_stat:
25389 + return vc_rlimit_stat(vxi, data);
25391 + case VCMD_set_cflags:
25392 + return vc_set_cflags(vxi, data);
25393 + case VCMD_get_cflags:
25394 + return vc_get_cflags(vxi, data);
25396 + /* this is version 1 */
25397 + case VCMD_set_ccaps:
25398 + return vc_set_ccaps(vxi, data);
25399 + /* this is version 1 */
25400 + case VCMD_get_ccaps:
25401 + return vc_get_ccaps(vxi, data);
25402 + case VCMD_set_bcaps:
25403 + return vc_set_bcaps(vxi, data);
25404 + case VCMD_get_bcaps:
25405 + return vc_get_bcaps(vxi, data);
25407 + case VCMD_set_badness:
25408 + return vc_set_badness(vxi, data);
25409 + case VCMD_get_badness:
25410 + return vc_get_badness(vxi, data);
25412 + case VCMD_set_nflags:
25413 + return vc_set_nflags(nxi, data);
25414 + case VCMD_get_nflags:
25415 + return vc_get_nflags(nxi, data);
25417 + case VCMD_set_ncaps:
25418 + return vc_set_ncaps(nxi, data);
25419 + case VCMD_get_ncaps:
25420 + return vc_get_ncaps(nxi, data);
25422 + case VCMD_set_sched_v4:
25423 + return vc_set_sched_v4(vxi, data);
25424 + /* this is version 5 */
25425 + case VCMD_set_sched:
25426 + return vc_set_sched(vxi, data);
25427 + case VCMD_get_sched:
25428 + return vc_get_sched(vxi, data);
25429 + case VCMD_sched_info:
25430 + return vc_sched_info(vxi, data);
25432 + case VCMD_add_dlimit:
25433 + return __COMPAT(vc_add_dlimit, id, data, compat);
25434 + case VCMD_rem_dlimit:
25435 + return __COMPAT(vc_rem_dlimit, id, data, compat);
25436 + case VCMD_set_dlimit:
25437 + return __COMPAT(vc_set_dlimit, id, data, compat);
25438 + case VCMD_get_dlimit:
25439 + return __COMPAT(vc_get_dlimit, id, data, compat);
25441 + case VCMD_ctx_kill:
25442 + return vc_ctx_kill(vxi, data);
25444 + case VCMD_wait_exit:
25445 + return vc_wait_exit(vxi, data);
25447 + case VCMD_get_iattr:
25448 + return __COMPAT_NO_ID(vc_get_iattr, data, compat);
25449 + case VCMD_set_iattr:
25450 + return __COMPAT_NO_ID(vc_set_iattr, data, compat);
25452 + case VCMD_fget_iattr:
25453 + return vc_fget_iattr(id, data);
25454 + case VCMD_fset_iattr:
25455 + return vc_fset_iattr(id, data);
25457 + case VCMD_enter_space_v0:
25458 + return vc_enter_space_v1(vxi, NULL);
25459 + case VCMD_enter_space_v1:
25460 + return vc_enter_space_v1(vxi, data);
25461 + /* this is version 2 */
25462 + case VCMD_enter_space:
25463 + return vc_enter_space(vxi, data);
25465 + case VCMD_ctx_create_v0:
25466 + return vc_ctx_create(id, NULL);
25467 + case VCMD_ctx_create:
25468 + return vc_ctx_create(id, data);
25469 + case VCMD_ctx_migrate_v0:
25470 + return vc_ctx_migrate(vxi, NULL);
25471 + case VCMD_ctx_migrate:
25472 + return vc_ctx_migrate(vxi, data);
25474 + case VCMD_net_create_v0:
25475 + return vc_net_create(id, NULL);
25476 + case VCMD_net_create:
25477 + return vc_net_create(id, data);
25478 + case VCMD_net_migrate:
25479 + return vc_net_migrate(nxi, data);
25481 + case VCMD_tag_migrate:
25482 + return vc_tag_migrate(id);
25484 + case VCMD_net_add:
25485 + return vc_net_add(nxi, data);
25486 + case VCMD_net_remove:
25487 + return vc_net_remove(nxi, data);
25489 + case VCMD_net_add_ipv4:
25490 + return vc_net_add_ipv4(nxi, data);
25491 + case VCMD_net_remove_ipv4:
25492 + return vc_net_remove_ipv4(nxi, data);
25493 +#ifdef CONFIG_IPV6
25494 + case VCMD_net_add_ipv6:
25495 + return vc_net_add_ipv6(nxi, data);
25496 + case VCMD_net_remove_ipv6:
25497 + return vc_net_remove_ipv6(nxi, data);
25499 +/* case VCMD_add_match_ipv4:
25500 + return vc_add_match_ipv4(nxi, data);
25501 + case VCMD_get_match_ipv4:
25502 + return vc_get_match_ipv4(nxi, data);
25503 +#ifdef CONFIG_IPV6
25504 + case VCMD_add_match_ipv6:
25505 + return vc_add_match_ipv6(nxi, data);
25506 + case VCMD_get_match_ipv6:
25507 + return vc_get_match_ipv6(nxi, data);
25510 +#ifdef CONFIG_VSERVER_DEVICE
25511 + case VCMD_set_mapping:
25512 + return __COMPAT(vc_set_mapping, vxi, data, compat);
25513 + case VCMD_unset_mapping:
25514 + return __COMPAT(vc_unset_mapping, vxi, data, compat);
25516 +#ifdef CONFIG_VSERVER_HISTORY
25517 + case VCMD_dump_history:
25518 + return vc_dump_history(id);
25519 + case VCMD_read_history:
25520 + return __COMPAT(vc_read_history, id, data, compat);
25522 +#ifdef CONFIG_VSERVER_MONITOR
25523 + case VCMD_read_monitor:
25524 + return __COMPAT(vc_read_monitor, id, data, compat);
25527 + vxwprintk_task(1, "unimplemented VCMD_%02d_%d[%d]",
25528 + VC_CATEGORY(cmd), VC_COMMAND(cmd), VC_VERSION(cmd));
25534 +#define __VCMD(vcmd, _perm, _args, _flags) \
25535 + case VCMD_ ## vcmd: perm = _perm; \
25536 + args = _args; flags = _flags; break
25539 +#define VCA_NONE 0x00
25540 +#define VCA_VXI 0x01
25541 +#define VCA_NXI 0x02
25543 +#define VCF_NONE 0x00
25544 +#define VCF_INFO 0x01
25545 +#define VCF_ADMIN 0x02
25546 +#define VCF_ARES 0x06 /* includes admin */
25547 +#define VCF_SETUP 0x08
25549 +#define VCF_ZIDOK 0x10 /* zero id okay */
25553 +long do_vserver(uint32_t cmd, uint32_t id, void __user *data, int compat)
25556 + int permit = -1, state = 0;
25557 + int perm = -1, args = 0, flags = 0;
25558 + struct vx_info *vxi = NULL;
25559 + struct nx_info *nxi = NULL;
25562 + /* unpriviledged commands */
25563 + __VCMD(get_version, 0, VCA_NONE, 0);
25564 + __VCMD(get_vci, 0, VCA_NONE, 0);
25565 + __VCMD(get_rlimit_mask, 0, VCA_NONE, 0);
25566 + __VCMD(get_space_mask_v0,0, VCA_NONE, 0);
25567 + __VCMD(get_space_mask, 0, VCA_NONE, 0);
25568 + __VCMD(get_space_default,0, VCA_NONE, 0);
25570 + /* info commands */
25571 + __VCMD(task_xid, 2, VCA_NONE, 0);
25572 + __VCMD(reset_hits, 2, VCA_VXI, 0);
25573 + __VCMD(reset_minmax, 2, VCA_VXI, 0);
25574 + __VCMD(vx_info, 3, VCA_VXI, VCF_INFO);
25575 + __VCMD(get_bcaps, 3, VCA_VXI, VCF_INFO);
25576 + __VCMD(get_ccaps, 3, VCA_VXI, VCF_INFO);
25577 + __VCMD(get_cflags, 3, VCA_VXI, VCF_INFO);
25578 + __VCMD(get_badness, 3, VCA_VXI, VCF_INFO);
25579 + __VCMD(get_vhi_name, 3, VCA_VXI, VCF_INFO);
25580 + __VCMD(get_rlimit, 3, VCA_VXI, VCF_INFO);
25582 + __VCMD(ctx_stat, 3, VCA_VXI, VCF_INFO);
25583 + __VCMD(virt_stat, 3, VCA_VXI, VCF_INFO);
25584 + __VCMD(sock_stat, 3, VCA_VXI, VCF_INFO);
25585 + __VCMD(rlimit_stat, 3, VCA_VXI, VCF_INFO);
25587 + __VCMD(task_nid, 2, VCA_NONE, 0);
25588 + __VCMD(nx_info, 3, VCA_NXI, VCF_INFO);
25589 + __VCMD(get_ncaps, 3, VCA_NXI, VCF_INFO);
25590 + __VCMD(get_nflags, 3, VCA_NXI, VCF_INFO);
25592 + __VCMD(task_tag, 2, VCA_NONE, 0);
25594 + __VCMD(get_iattr, 2, VCA_NONE, 0);
25595 + __VCMD(fget_iattr, 2, VCA_NONE, 0);
25596 + __VCMD(get_dlimit, 3, VCA_NONE, VCF_INFO);
25597 + __VCMD(get_sched, 3, VCA_VXI, VCF_INFO);
25598 + __VCMD(sched_info, 3, VCA_VXI, VCF_INFO | VCF_ZIDOK);
25600 + /* lower admin commands */
25601 + __VCMD(wait_exit, 4, VCA_VXI, VCF_INFO);
25602 + __VCMD(ctx_create_v0, 5, VCA_NONE, 0);
25603 + __VCMD(ctx_create, 5, VCA_NONE, 0);
25604 + __VCMD(ctx_migrate_v0, 5, VCA_VXI, VCF_ADMIN);
25605 + __VCMD(ctx_migrate, 5, VCA_VXI, VCF_ADMIN);
25606 + __VCMD(enter_space_v0, 5, VCA_VXI, VCF_ADMIN);
25607 + __VCMD(enter_space_v1, 5, VCA_VXI, VCF_ADMIN);
25608 + __VCMD(enter_space, 5, VCA_VXI, VCF_ADMIN);
25610 + __VCMD(net_create_v0, 5, VCA_NONE, 0);
25611 + __VCMD(net_create, 5, VCA_NONE, 0);
25612 + __VCMD(net_migrate, 5, VCA_NXI, VCF_ADMIN);
25614 + __VCMD(tag_migrate, 5, VCA_NONE, VCF_ADMIN);
25616 + /* higher admin commands */
25617 + __VCMD(ctx_kill, 6, VCA_VXI, VCF_ARES);
25618 + __VCMD(set_space_v1, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25619 + __VCMD(set_space, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25621 + __VCMD(set_ccaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25622 + __VCMD(set_bcaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25623 + __VCMD(set_cflags, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25624 + __VCMD(set_badness, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25626 + __VCMD(set_vhi_name, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25627 + __VCMD(set_rlimit, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25628 + __VCMD(set_sched, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25629 + __VCMD(set_sched_v4, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25631 + __VCMD(set_ncaps, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25632 + __VCMD(set_nflags, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25633 + __VCMD(net_add, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25634 + __VCMD(net_remove, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25635 + __VCMD(net_add_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25636 + __VCMD(net_remove_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25637 +#ifdef CONFIG_IPV6
25638 + __VCMD(net_add_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25639 + __VCMD(net_remove_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25641 + __VCMD(set_iattr, 7, VCA_NONE, 0);
25642 + __VCMD(fset_iattr, 7, VCA_NONE, 0);
25643 + __VCMD(set_dlimit, 7, VCA_NONE, VCF_ARES);
25644 + __VCMD(add_dlimit, 8, VCA_NONE, VCF_ARES);
25645 + __VCMD(rem_dlimit, 8, VCA_NONE, VCF_ARES);
25647 +#ifdef CONFIG_VSERVER_DEVICE
25648 + __VCMD(set_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25649 + __VCMD(unset_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25651 + /* debug level admin commands */
25652 +#ifdef CONFIG_VSERVER_HISTORY
25653 + __VCMD(dump_history, 9, VCA_NONE, 0);
25654 + __VCMD(read_history, 9, VCA_NONE, 0);
25656 +#ifdef CONFIG_VSERVER_MONITOR
25657 + __VCMD(read_monitor, 9, VCA_NONE, 0);
25664 + vxdprintk(VXD_CBIT(switch, 0),
25665 + "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]",
25666 + VC_CATEGORY(cmd), VC_COMMAND(cmd),
25667 + VC_VERSION(cmd), id, data, compat,
25668 + perm, args, flags);
25675 + if (!capable(CAP_CONTEXT))
25679 + /* moved here from the individual commands */
25681 + if ((perm > 1) && !capable(CAP_SYS_ADMIN))
25685 + /* vcmd involves resource management */
25687 + if ((flags & VCF_ARES) && !capable(CAP_SYS_RESOURCE))
25691 + /* various legacy exceptions */
25693 + /* will go away when spectator is a cap */
25694 + case VCMD_ctx_migrate_v0:
25695 + case VCMD_ctx_migrate:
25697 + current->xid = 1;
25703 + /* will go away when spectator is a cap */
25704 + case VCMD_net_migrate:
25706 + current->nid = 1;
25713 + /* vcmds are fine by default */
25716 + /* admin type vcmds require admin ... */
25717 + if (flags & VCF_ADMIN)
25718 + permit = vx_check(0, VS_ADMIN) ? 1 : 0;
25720 + /* ... but setup type vcmds override that */
25721 + if (!permit && (flags & VCF_SETUP))
25722 + permit = vx_flags(VXF_STATE_SETUP, 0) ? 2 : 0;
25730 + if (!id && (flags & VCF_ZIDOK))
25734 + if (args & VCA_VXI) {
25735 + vxi = lookup_vx_info(id);
25739 + if ((flags & VCF_ADMIN) &&
25740 + /* special case kill for shutdown */
25741 + (cmd != VCMD_ctx_kill) &&
25742 + /* can context be administrated? */
25743 + !vx_info_flags(vxi, VXF_STATE_ADMIN, 0)) {
25749 + if (args & VCA_NXI) {
25750 + nxi = lookup_nx_info(id);
25754 + if ((flags & VCF_ADMIN) &&
25755 + /* can context be administrated? */
25756 + !nx_info_flags(nxi, NXF_STATE_ADMIN, 0)) {
25763 + ret = do_vcmd(cmd, id, vxi, nxi, data, compat);
25766 + if ((args & VCA_NXI) && nxi)
25767 + put_nx_info(nxi);
25769 + if ((args & VCA_VXI) && vxi)
25770 + put_vx_info(vxi);
25772 + vxdprintk(VXD_CBIT(switch, 1),
25773 + "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]",
25774 + VC_CATEGORY(cmd), VC_COMMAND(cmd),
25775 + VC_VERSION(cmd), ret, ret, state, permit);
25780 +sys_vserver(uint32_t cmd, uint32_t id, void __user *data)
25782 + return do_vserver(cmd, id, data, 0);
25785 +#ifdef CONFIG_COMPAT
25788 +sys32_vserver(uint32_t cmd, uint32_t id, void __user *data)
25790 + return do_vserver(cmd, id, data, 1);
25793 +#endif /* CONFIG_COMPAT */
25794 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/sysctl.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sysctl.c
25795 --- linux-2.6.32.1/kernel/vserver/sysctl.c 1970-01-01 01:00:00.000000000 +0100
25796 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/sysctl.c 2009-12-03 20:04:56.000000000 +0100
25799 + * kernel/vserver/sysctl.c
25801 + * Virtual Context Support
25803 + * Copyright (C) 2004-2007 Herbert Pötzl
25805 + * V0.01 basic structure
25809 +#include <linux/module.h>
25810 +#include <linux/ctype.h>
25811 +#include <linux/sysctl.h>
25812 +#include <linux/parser.h>
25813 +#include <asm/uaccess.h>
25817 + CTL_DEBUG_ERROR = 0,
25818 + CTL_DEBUG_SWITCH = 1,
25833 +unsigned int vx_debug_switch = 0;
25834 +unsigned int vx_debug_xid = 0;
25835 +unsigned int vx_debug_nid = 0;
25836 +unsigned int vx_debug_tag = 0;
25837 +unsigned int vx_debug_net = 0;
25838 +unsigned int vx_debug_limit = 0;
25839 +unsigned int vx_debug_cres = 0;
25840 +unsigned int vx_debug_dlim = 0;
25841 +unsigned int vx_debug_quota = 0;
25842 +unsigned int vx_debug_cvirt = 0;
25843 +unsigned int vx_debug_space = 0;
25844 +unsigned int vx_debug_misc = 0;
25847 +static struct ctl_table_header *vserver_table_header;
25848 +static ctl_table vserver_root_table[];
25851 +void vserver_register_sysctl(void)
25853 + if (!vserver_table_header) {
25854 + vserver_table_header = register_sysctl_table(vserver_root_table);
25859 +void vserver_unregister_sysctl(void)
25861 + if (vserver_table_header) {
25862 + unregister_sysctl_table(vserver_table_header);
25863 + vserver_table_header = NULL;
25868 +static int proc_dodebug(ctl_table *table, int write,
25869 + struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
25871 + char tmpbuf[20], *p, c;
25872 + unsigned int value;
25873 + size_t left, len;
25875 + if ((*ppos && !write) || !*lenp) {
25883 + if (!access_ok(VERIFY_READ, buffer, left))
25885 + p = (char *)buffer;
25886 + while (left && __get_user(c, p) >= 0 && isspace(c))
25891 + if (left > sizeof(tmpbuf) - 1)
25893 + if (copy_from_user(tmpbuf, p, left))
25895 + tmpbuf[left] = '\0';
25897 + for (p = tmpbuf, value = 0; '0' <= *p && *p <= '9'; p++, left--)
25898 + value = 10 * value + (*p - '0');
25899 + if (*p && !isspace(*p))
25901 + while (left && isspace(*p))
25903 + *(unsigned int *)table->data = value;
25905 + if (!access_ok(VERIFY_WRITE, buffer, left))
25907 + len = sprintf(tmpbuf, "%d", *(unsigned int *)table->data);
25910 + if (__copy_to_user(buffer, tmpbuf, len))
25912 + if ((left -= len) > 0) {
25913 + if (put_user('\n', (char *)buffer + len))
25927 +#define CTL_ENTRY(ctl, name) \
25929 + .ctl_name = ctl, \
25930 + .procname = #name, \
25931 + .data = &vx_ ## name, \
25932 + .maxlen = sizeof(int), \
25934 + .proc_handler = &proc_dodebug, \
25935 + .strategy = &sysctl_intvec, \
25936 + .extra1 = &zero, \
25939 +static ctl_table vserver_debug_table[] = {
25940 + CTL_ENTRY(CTL_DEBUG_SWITCH, debug_switch),
25941 + CTL_ENTRY(CTL_DEBUG_XID, debug_xid),
25942 + CTL_ENTRY(CTL_DEBUG_NID, debug_nid),
25943 + CTL_ENTRY(CTL_DEBUG_TAG, debug_tag),
25944 + CTL_ENTRY(CTL_DEBUG_NET, debug_net),
25945 + CTL_ENTRY(CTL_DEBUG_LIMIT, debug_limit),
25946 + CTL_ENTRY(CTL_DEBUG_CRES, debug_cres),
25947 + CTL_ENTRY(CTL_DEBUG_DLIM, debug_dlim),
25948 + CTL_ENTRY(CTL_DEBUG_QUOTA, debug_quota),
25949 + CTL_ENTRY(CTL_DEBUG_CVIRT, debug_cvirt),
25950 + CTL_ENTRY(CTL_DEBUG_SPACE, debug_space),
25951 + CTL_ENTRY(CTL_DEBUG_MISC, debug_misc),
25952 + { .ctl_name = 0 }
25955 +static ctl_table vserver_root_table[] = {
25957 + .ctl_name = CTL_VSERVER,
25958 + .procname = "vserver",
25960 + .child = vserver_debug_table
25962 + { .ctl_name = 0 }
25966 +static match_table_t tokens = {
25967 + { CTL_DEBUG_SWITCH, "switch=%x" },
25968 + { CTL_DEBUG_XID, "xid=%x" },
25969 + { CTL_DEBUG_NID, "nid=%x" },
25970 + { CTL_DEBUG_TAG, "tag=%x" },
25971 + { CTL_DEBUG_NET, "net=%x" },
25972 + { CTL_DEBUG_LIMIT, "limit=%x" },
25973 + { CTL_DEBUG_CRES, "cres=%x" },
25974 + { CTL_DEBUG_DLIM, "dlim=%x" },
25975 + { CTL_DEBUG_QUOTA, "quota=%x" },
25976 + { CTL_DEBUG_CVIRT, "cvirt=%x" },
25977 + { CTL_DEBUG_SPACE, "space=%x" },
25978 + { CTL_DEBUG_MISC, "misc=%x" },
25979 + { CTL_DEBUG_ERROR, NULL }
25982 +#define HANDLE_CASE(id, name, val) \
25983 + case CTL_DEBUG_ ## id: \
25984 + vx_debug_ ## name = val; \
25985 + printk("vs_debug_" #name "=0x%x\n", val); \
25989 +static int __init vs_debug_setup(char *str)
25994 + printk("vs_debug_setup(%s)\n", str);
25995 + while ((p = strsep(&str, ",")) != NULL) {
25996 + substring_t args[MAX_OPT_ARGS];
25997 + unsigned int value;
26002 + token = match_token(p, tokens, args);
26003 + value = (token > 0) ? simple_strtoul(args[0].from, NULL, 0) : 0;
26006 + HANDLE_CASE(SWITCH, switch, value);
26007 + HANDLE_CASE(XID, xid, value);
26008 + HANDLE_CASE(NID, nid, value);
26009 + HANDLE_CASE(TAG, tag, value);
26010 + HANDLE_CASE(NET, net, value);
26011 + HANDLE_CASE(LIMIT, limit, value);
26012 + HANDLE_CASE(CRES, cres, value);
26013 + HANDLE_CASE(DLIM, dlim, value);
26014 + HANDLE_CASE(QUOTA, quota, value);
26015 + HANDLE_CASE(CVIRT, cvirt, value);
26016 + HANDLE_CASE(SPACE, space, value);
26017 + HANDLE_CASE(MISC, misc, value);
26026 +__setup("vsdebug=", vs_debug_setup);
26030 +EXPORT_SYMBOL_GPL(vx_debug_switch);
26031 +EXPORT_SYMBOL_GPL(vx_debug_xid);
26032 +EXPORT_SYMBOL_GPL(vx_debug_nid);
26033 +EXPORT_SYMBOL_GPL(vx_debug_net);
26034 +EXPORT_SYMBOL_GPL(vx_debug_limit);
26035 +EXPORT_SYMBOL_GPL(vx_debug_cres);
26036 +EXPORT_SYMBOL_GPL(vx_debug_dlim);
26037 +EXPORT_SYMBOL_GPL(vx_debug_quota);
26038 +EXPORT_SYMBOL_GPL(vx_debug_cvirt);
26039 +EXPORT_SYMBOL_GPL(vx_debug_space);
26040 +EXPORT_SYMBOL_GPL(vx_debug_misc);
26042 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/tag.c linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/tag.c
26043 --- linux-2.6.32.1/kernel/vserver/tag.c 1970-01-01 01:00:00.000000000 +0100
26044 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/tag.c 2009-12-03 20:04:56.000000000 +0100
26047 + * linux/kernel/vserver/tag.c
26049 + * Virtual Server: Shallow Tag Space
26051 + * Copyright (C) 2007 Herbert Pötzl
26053 + * V0.01 basic implementation
26057 +#include <linux/sched.h>
26058 +#include <linux/vserver/debug.h>
26059 +#include <linux/vs_pid.h>
26060 +#include <linux/vs_tag.h>
26062 +#include <linux/vserver/tag_cmd.h>
26065 +int dx_migrate_task(struct task_struct *p, tag_t tag)
26070 + vxdprintk(VXD_CBIT(tag, 5),
26071 + "dx_migrate_task(%p[#%d],#%d)", p, p->tag, tag);
26077 + vxdprintk(VXD_CBIT(tag, 5),
26078 + "moved task %p into [#%d]", p, tag);
26082 +/* vserver syscall commands below here */
26084 +/* taks xid and vx_info functions */
26087 +int vc_task_tag(uint32_t id)
26092 + struct task_struct *tsk;
26093 + read_lock(&tasklist_lock);
26094 + tsk = find_task_by_real_pid(id);
26095 + tag = (tsk) ? tsk->tag : -ESRCH;
26096 + read_unlock(&tasklist_lock);
26098 + tag = dx_current_tag();
26103 +int vc_tag_migrate(uint32_t tag)
26105 + return dx_migrate_task(current, tag & 0xFFFF);
26109 diff -NurpP --minimal linux-2.6.32.1/kernel/vserver/vci_config.h linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/vci_config.h
26110 --- linux-2.6.32.1/kernel/vserver/vci_config.h 1970-01-01 01:00:00.000000000 +0100
26111 +++ linux-2.6.32.1-vs2.3.0.36.27/kernel/vserver/vci_config.h 2009-12-03 20:04:56.000000000 +0100
26114 +/* interface version */
26116 +#define VCI_VERSION 0x00020305
26120 + VCI_KCBIT_NO_DYNAMIC = 0,
26122 + VCI_KCBIT_PROC_SECURE = 4,
26123 + VCI_KCBIT_HARDCPU = 5,
26124 + VCI_KCBIT_IDLELIMIT = 6,
26125 + VCI_KCBIT_IDLETIME = 7,
26127 + VCI_KCBIT_COWBL = 8,
26128 + VCI_KCBIT_FULLCOWBL = 9,
26129 + VCI_KCBIT_SPACES = 10,
26130 + VCI_KCBIT_NETV2 = 11,
26132 + VCI_KCBIT_DEBUG = 16,
26133 + VCI_KCBIT_HISTORY = 20,
26134 + VCI_KCBIT_TAGGED = 24,
26135 + VCI_KCBIT_PPTAG = 28,
26137 + VCI_KCBIT_MORE = 31,
26141 +static inline uint32_t vci_kernel_config(void)
26144 + (1 << VCI_KCBIT_NO_DYNAMIC) |
26146 + /* configured features */
26147 +#ifdef CONFIG_VSERVER_PROC_SECURE
26148 + (1 << VCI_KCBIT_PROC_SECURE) |
26150 +#ifdef CONFIG_VSERVER_HARDCPU
26151 + (1 << VCI_KCBIT_HARDCPU) |
26153 +#ifdef CONFIG_VSERVER_IDLELIMIT
26154 + (1 << VCI_KCBIT_IDLELIMIT) |
26156 +#ifdef CONFIG_VSERVER_IDLETIME
26157 + (1 << VCI_KCBIT_IDLETIME) |
26159 +#ifdef CONFIG_VSERVER_COWBL
26160 + (1 << VCI_KCBIT_COWBL) |
26161 + (1 << VCI_KCBIT_FULLCOWBL) |
26163 + (1 << VCI_KCBIT_SPACES) |
26164 + (1 << VCI_KCBIT_NETV2) |
26166 + /* debug options */
26167 +#ifdef CONFIG_VSERVER_DEBUG
26168 + (1 << VCI_KCBIT_DEBUG) |
26170 +#ifdef CONFIG_VSERVER_HISTORY
26171 + (1 << VCI_KCBIT_HISTORY) |
26174 + /* inode context tagging */
26175 +#if defined(CONFIG_TAGGING_NONE)
26176 + (0 << VCI_KCBIT_TAGGED) |
26177 +#elif defined(CONFIG_TAGGING_UID16)
26178 + (1 << VCI_KCBIT_TAGGED) |
26179 +#elif defined(CONFIG_TAGGING_GID16)
26180 + (2 << VCI_KCBIT_TAGGED) |
26181 +#elif defined(CONFIG_TAGGING_ID24)
26182 + (3 << VCI_KCBIT_TAGGED) |
26183 +#elif defined(CONFIG_TAGGING_INTERN)
26184 + (4 << VCI_KCBIT_TAGGED) |
26185 +#elif defined(CONFIG_TAGGING_RUNTIME)
26186 + (5 << VCI_KCBIT_TAGGED) |
26188 + (7 << VCI_KCBIT_TAGGED) |
26190 + (1 << VCI_KCBIT_PPTAG) |
26194 diff -NurpP --minimal linux-2.6.32.1/mm/allocpercpu.c linux-2.6.32.1-vs2.3.0.36.27/mm/allocpercpu.c
26195 --- linux-2.6.32.1/mm/allocpercpu.c 2009-12-03 20:02:58.000000000 +0100
26196 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/allocpercpu.c 2009-12-03 20:04:56.000000000 +0100
26197 @@ -160,12 +160,14 @@ EXPORT_SYMBOL(__per_cpu_offset);
26199 void __init setup_per_cpu_areas(void)
26201 - unsigned long size, i;
26202 + unsigned long size, vspc, i;
26204 unsigned long nr_possible_cpus = num_possible_cpus();
26206 + vspc = PERCPU_PERCTX * CONFIG_VSERVER_CONTEXTS;
26208 /* Copy section for each CPU (we discard the original) */
26209 - size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
26210 + size = ALIGN(PERCPU_ENOUGH_ROOM + vspc, PAGE_SIZE);
26211 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
26213 for_each_possible_cpu(i) {
26214 diff -NurpP --minimal linux-2.6.32.1/mm/filemap_xip.c linux-2.6.32.1-vs2.3.0.36.27/mm/filemap_xip.c
26215 --- linux-2.6.32.1/mm/filemap_xip.c 2009-12-03 20:02:58.000000000 +0100
26216 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/filemap_xip.c 2009-12-03 20:04:56.000000000 +0100
26218 #include <linux/sched.h>
26219 #include <linux/seqlock.h>
26220 #include <linux/mutex.h>
26221 +#include <linux/vs_memory.h>
26222 #include <asm/tlbflush.h>
26223 #include <asm/io.h>
26225 diff -NurpP --minimal linux-2.6.32.1/mm/fremap.c linux-2.6.32.1-vs2.3.0.36.27/mm/fremap.c
26226 --- linux-2.6.32.1/mm/fremap.c 2009-03-24 14:22:45.000000000 +0100
26227 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/fremap.c 2009-12-03 20:04:56.000000000 +0100
26229 #include <linux/module.h>
26230 #include <linux/syscalls.h>
26231 #include <linux/mmu_notifier.h>
26232 +#include <linux/vs_memory.h>
26234 #include <asm/mmu_context.h>
26235 #include <asm/cacheflush.h>
26236 diff -NurpP --minimal linux-2.6.32.1/mm/hugetlb.c linux-2.6.32.1-vs2.3.0.36.27/mm/hugetlb.c
26237 --- linux-2.6.32.1/mm/hugetlb.c 2009-12-03 20:02:58.000000000 +0100
26238 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/hugetlb.c 2009-12-03 20:04:56.000000000 +0100
26240 #include <asm/io.h>
26242 #include <linux/hugetlb.h>
26243 +#include <linux/vs_memory.h>
26244 #include "internal.h"
26246 const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
26247 diff -NurpP --minimal linux-2.6.32.1/mm/memory.c linux-2.6.32.1-vs2.3.0.36.27/mm/memory.c
26248 --- linux-2.6.32.1/mm/memory.c 2009-12-03 20:02:58.000000000 +0100
26249 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/memory.c 2009-12-03 20:04:56.000000000 +0100
26251 #include <linux/kallsyms.h>
26252 #include <linux/swapops.h>
26253 #include <linux/elf.h>
26254 +// #include <linux/vs_memory.h>
26256 #include <asm/io.h>
26257 #include <asm/pgalloc.h>
26258 @@ -647,6 +648,9 @@ static int copy_pte_range(struct mm_stru
26262 + if (!vx_rss_avail(dst_mm, ((end - addr)/PAGE_SIZE + 1)))
26266 rss[1] = rss[0] = 0;
26267 dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
26268 @@ -2645,6 +2649,8 @@ static int do_anonymous_page(struct mm_s
26269 /* Allocate our own private page. */
26270 pte_unmap(page_table);
26272 + if (!vx_rss_avail(mm, 1))
26274 if (unlikely(anon_vma_prepare(vma)))
26276 page = alloc_zeroed_user_highpage_movable(vma, address);
26277 @@ -2936,6 +2942,7 @@ static inline int handle_pte_fault(struc
26281 + int ret = 0, type = VXPT_UNKNOWN;
26284 if (!pte_present(entry)) {
26285 @@ -2960,9 +2967,12 @@ static inline int handle_pte_fault(struc
26286 if (unlikely(!pte_same(*pte, entry)))
26288 if (flags & FAULT_FLAG_WRITE) {
26289 - if (!pte_write(entry))
26290 - return do_wp_page(mm, vma, address,
26291 + if (!pte_write(entry)) {
26292 + ret = do_wp_page(mm, vma, address,
26293 pte, pmd, ptl, entry);
26294 + type = VXPT_WRITE;
26297 entry = pte_mkdirty(entry);
26299 entry = pte_mkyoung(entry);
26300 @@ -2980,7 +2990,10 @@ static inline int handle_pte_fault(struc
26303 pte_unmap_unlock(pte, ptl);
26307 + vx_page_fault(mm, vma, type, ret);
26312 diff -NurpP --minimal linux-2.6.32.1/mm/mlock.c linux-2.6.32.1-vs2.3.0.36.27/mm/mlock.c
26313 --- linux-2.6.32.1/mm/mlock.c 2009-12-03 20:02:58.000000000 +0100
26314 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/mlock.c 2009-12-03 20:04:56.000000000 +0100
26316 #include <linux/rmap.h>
26317 #include <linux/mmzone.h>
26318 #include <linux/hugetlb.h>
26319 +#include <linux/vs_memory.h>
26321 #include "internal.h"
26323 @@ -401,7 +402,7 @@ success:
26324 nr_pages = (end - start) >> PAGE_SHIFT;
26326 nr_pages = -nr_pages;
26327 - mm->locked_vm += nr_pages;
26328 + vx_vmlocked_add(mm, nr_pages);
26331 * vm_flags is protected by the mmap_sem held in write mode.
26332 @@ -474,7 +475,7 @@ static int do_mlock(unsigned long start,
26334 SYSCALL_DEFINE2(mlock, unsigned long, start, size_t, len)
26336 - unsigned long locked;
26337 + unsigned long locked, grow;
26338 unsigned long lock_limit;
26339 int error = -ENOMEM;
26341 @@ -487,8 +488,10 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
26342 len = PAGE_ALIGN(len + (start & ~PAGE_MASK));
26343 start &= PAGE_MASK;
26345 - locked = len >> PAGE_SHIFT;
26346 - locked += current->mm->locked_vm;
26347 + grow = len >> PAGE_SHIFT;
26348 + if (!vx_vmlocked_avail(current->mm, grow))
26350 + locked = current->mm->locked_vm + grow;
26352 lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
26353 lock_limit >>= PAGE_SHIFT;
26354 @@ -496,6 +499,7 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
26355 /* check against resource limits */
26356 if ((locked <= lock_limit) || capable(CAP_IPC_LOCK))
26357 error = do_mlock(start, len, 1);
26359 up_write(¤t->mm->mmap_sem);
26362 @@ -557,6 +561,8 @@ SYSCALL_DEFINE1(mlockall, int, flags)
26363 lock_limit >>= PAGE_SHIFT;
26366 + if (!vx_vmlocked_avail(current->mm, current->mm->total_vm))
26368 if (!(flags & MCL_CURRENT) || (current->mm->total_vm <= lock_limit) ||
26369 capable(CAP_IPC_LOCK))
26370 ret = do_mlockall(flags);
26371 @@ -631,8 +637,10 @@ int account_locked_memory(struct mm_stru
26375 - mm->total_vm += pgsz;
26376 - mm->locked_vm += pgsz;
26377 + // mm->total_vm += pgsz;
26378 + vx_vmpages_add(mm, pgsz);
26379 + // mm->locked_vm += pgsz;
26380 + vx_vmlocked_add(mm, pgsz);
26384 @@ -646,8 +654,10 @@ void refund_locked_memory(struct mm_stru
26386 down_write(&mm->mmap_sem);
26388 - mm->total_vm -= pgsz;
26389 - mm->locked_vm -= pgsz;
26390 + // mm->total_vm -= pgsz;
26391 + vx_vmpages_sub(mm, pgsz);
26392 + // mm->locked_vm -= pgsz;
26393 + vx_vmlocked_sub(mm, pgsz);
26395 up_write(&mm->mmap_sem);
26397 diff -NurpP --minimal linux-2.6.32.1/mm/mmap.c linux-2.6.32.1-vs2.3.0.36.27/mm/mmap.c
26398 --- linux-2.6.32.1/mm/mmap.c 2009-12-03 20:02:58.000000000 +0100
26399 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/mmap.c 2009-12-03 20:04:56.000000000 +0100
26400 @@ -1236,7 +1236,8 @@ munmap_back:
26402 perf_event_mmap(vma);
26404 - mm->total_vm += len >> PAGE_SHIFT;
26405 + // mm->total_vm += len >> PAGE_SHIFT;
26406 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
26407 vm_stat_account(mm, vm_flags, file, len >> PAGE_SHIFT);
26408 if (vm_flags & VM_LOCKED) {
26410 @@ -1245,7 +1246,8 @@ out:
26411 long nr_pages = mlock_vma_pages_range(vma, addr, addr + len);
26413 return nr_pages; /* vma gone! */
26414 - mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
26415 + // mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
26416 + vx_vmlocked_add(mm, (len >> PAGE_SHIFT) - nr_pages);
26417 } else if ((flags & MAP_POPULATE) && !(flags & MAP_NONBLOCK))
26418 make_pages_present(addr, addr + len);
26420 @@ -1592,9 +1594,9 @@ static int acct_stack_growth(struct vm_a
26423 /* Ok, everything looks good - let it rip */
26424 - mm->total_vm += grow;
26425 + vx_vmpages_add(mm, grow);
26426 if (vma->vm_flags & VM_LOCKED)
26427 - mm->locked_vm += grow;
26428 + vx_vmlocked_add(mm, grow);
26429 vm_stat_account(mm, vma->vm_flags, vma->vm_file, grow);
26432 @@ -1769,7 +1771,8 @@ static void remove_vma_list(struct mm_st
26434 long nrpages = vma_pages(vma);
26436 - mm->total_vm -= nrpages;
26437 + // mm->total_vm -= nrpages;
26438 + vx_vmpages_sub(mm, nrpages);
26439 vm_stat_account(mm, vma->vm_flags, vma->vm_file, -nrpages);
26440 vma = remove_vma(vma);
26442 @@ -1941,7 +1944,8 @@ int do_munmap(struct mm_struct *mm, unsi
26443 struct vm_area_struct *tmp = vma;
26444 while (tmp && tmp->vm_start < end) {
26445 if (tmp->vm_flags & VM_LOCKED) {
26446 - mm->locked_vm -= vma_pages(tmp);
26447 + // mm->locked_vm -= vma_pages(tmp);
26448 + vx_vmlocked_sub(mm, vma_pages(tmp));
26449 munlock_vma_pages_all(tmp);
26451 tmp = tmp->vm_next;
26452 @@ -2030,6 +2034,8 @@ unsigned long do_brk(unsigned long addr,
26453 lock_limit >>= PAGE_SHIFT;
26454 if (locked > lock_limit && !capable(CAP_IPC_LOCK))
26456 + if (!vx_vmlocked_avail(mm, len >> PAGE_SHIFT))
26461 @@ -2056,7 +2062,8 @@ unsigned long do_brk(unsigned long addr,
26462 if (mm->map_count > sysctl_max_map_count)
26465 - if (security_vm_enough_memory(len >> PAGE_SHIFT))
26466 + if (security_vm_enough_memory(len >> PAGE_SHIFT) ||
26467 + !vx_vmpages_avail(mm, len >> PAGE_SHIFT))
26470 /* Can we just expand an old private anonymous mapping? */
26471 @@ -2082,10 +2089,13 @@ unsigned long do_brk(unsigned long addr,
26472 vma->vm_page_prot = vm_get_page_prot(flags);
26473 vma_link(mm, vma, prev, rb_link, rb_parent);
26475 - mm->total_vm += len >> PAGE_SHIFT;
26476 + // mm->total_vm += len >> PAGE_SHIFT;
26477 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
26479 if (flags & VM_LOCKED) {
26480 if (!mlock_vma_pages_range(vma, addr, addr + len))
26481 - mm->locked_vm += (len >> PAGE_SHIFT);
26482 + // mm->locked_vm += (len >> PAGE_SHIFT);
26483 + vx_vmlocked_add(mm, len >> PAGE_SHIFT);
26487 @@ -2129,6 +2139,11 @@ void exit_mmap(struct mm_struct *mm)
26488 free_pgtables(tlb, vma, FIRST_USER_ADDRESS, 0);
26489 tlb_finish_mmu(tlb, 0, end);
26491 + set_mm_counter(mm, file_rss, 0);
26492 + set_mm_counter(mm, anon_rss, 0);
26493 + vx_vmpages_sub(mm, mm->total_vm);
26494 + vx_vmlocked_sub(mm, mm->locked_vm);
26497 * Walk the list again, actually closing and freeing it,
26498 * with preemption enabled, without holding any MM locks.
26499 @@ -2168,7 +2183,8 @@ int insert_vm_struct(struct mm_struct *
26500 if (__vma && __vma->vm_start < vma->vm_end)
26502 if ((vma->vm_flags & VM_ACCOUNT) &&
26503 - security_vm_enough_memory_mm(mm, vma_pages(vma)))
26504 + (security_vm_enough_memory_mm(mm, vma_pages(vma)) ||
26505 + !vx_vmpages_avail(mm, vma_pages(vma))))
26507 vma_link(mm, vma, prev, rb_link, rb_parent);
26509 @@ -2244,6 +2260,8 @@ int may_expand_vm(struct mm_struct *mm,
26511 if (cur + npages > lim)
26513 + if (!vx_vmpages_avail(mm, npages))
26518 @@ -2321,7 +2339,7 @@ int install_special_mapping(struct mm_st
26522 - mm->total_vm += len >> PAGE_SHIFT;
26523 + vx_vmpages_add(mm, len >> PAGE_SHIFT);
26525 perf_event_mmap(vma);
26527 diff -NurpP --minimal linux-2.6.32.1/mm/mremap.c linux-2.6.32.1-vs2.3.0.36.27/mm/mremap.c
26528 --- linux-2.6.32.1/mm/mremap.c 2009-12-03 20:02:58.000000000 +0100
26529 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/mremap.c 2009-12-03 20:04:56.000000000 +0100
26531 #include <linux/security.h>
26532 #include <linux/syscalls.h>
26533 #include <linux/mmu_notifier.h>
26534 +#include <linux/vs_memory.h>
26536 #include <asm/uaccess.h>
26537 #include <asm/cacheflush.h>
26538 @@ -234,7 +235,7 @@ static unsigned long move_vma(struct vm_
26539 * If this were a serious issue, we'd add a flag to do_munmap().
26541 hiwater_vm = mm->hiwater_vm;
26542 - mm->total_vm += new_len >> PAGE_SHIFT;
26543 + vx_vmpages_add(mm, new_len >> PAGE_SHIFT);
26544 vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
26546 if (do_munmap(mm, old_addr, old_len) < 0) {
26547 @@ -252,7 +253,7 @@ static unsigned long move_vma(struct vm_
26550 if (vm_flags & VM_LOCKED) {
26551 - mm->locked_vm += new_len >> PAGE_SHIFT;
26552 + vx_vmlocked_add(mm, new_len >> PAGE_SHIFT);
26553 if (new_len > old_len)
26554 mlock_vma_pages_range(new_vma, new_addr + old_len,
26555 new_addr + new_len);
26556 @@ -363,6 +364,9 @@ unsigned long do_mremap(unsigned long ad
26558 if (locked > lock_limit && !capable(CAP_IPC_LOCK))
26560 + if (!vx_vmlocked_avail(current->mm,
26561 + (new_len - old_len) >> PAGE_SHIFT))
26564 if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
26566 @@ -391,10 +395,12 @@ unsigned long do_mremap(unsigned long ad
26567 vma_adjust(vma, vma->vm_start,
26568 addr + new_len, vma->vm_pgoff, NULL);
26570 - mm->total_vm += pages;
26571 + // mm->total_vm += pages;
26572 + vx_vmpages_add(mm, pages);
26573 vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
26574 if (vma->vm_flags & VM_LOCKED) {
26575 - mm->locked_vm += pages;
26576 + // mm->locked_vm += pages;
26577 + vx_vmlocked_add(mm, pages);
26578 mlock_vma_pages_range(vma, addr + old_len,
26581 diff -NurpP --minimal linux-2.6.32.1/mm/nommu.c linux-2.6.32.1-vs2.3.0.36.27/mm/nommu.c
26582 --- linux-2.6.32.1/mm/nommu.c 2009-12-03 20:02:58.000000000 +0100
26583 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/nommu.c 2009-12-03 20:04:56.000000000 +0100
26584 @@ -1346,7 +1346,7 @@ unsigned long do_mmap_pgoff(struct file
26585 /* okay... we have a mapping; now we have to register it */
26586 result = vma->vm_start;
26588 - current->mm->total_vm += len >> PAGE_SHIFT;
26589 + vx_vmpages_add(current->mm, len >> PAGE_SHIFT);
26592 add_vma_to_mm(current->mm, vma);
26593 @@ -1606,7 +1606,7 @@ void exit_mmap(struct mm_struct *mm)
26597 - mm->total_vm = 0;
26598 + vx_vmpages_sub(mm, mm->total_vm);
26600 while ((vma = mm->mmap)) {
26601 mm->mmap = vma->vm_next;
26602 diff -NurpP --minimal linux-2.6.32.1/mm/oom_kill.c linux-2.6.32.1-vs2.3.0.36.27/mm/oom_kill.c
26603 --- linux-2.6.32.1/mm/oom_kill.c 2009-12-03 20:02:58.000000000 +0100
26604 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/oom_kill.c 2009-12-03 22:26:55.000000000 +0100
26606 #include <linux/notifier.h>
26607 #include <linux/memcontrol.h>
26608 #include <linux/security.h>
26609 +#include <linux/reboot.h>
26610 +#include <linux/vs_memory.h>
26611 +#include <linux/vs_context.h>
26613 int sysctl_panic_on_oom;
26614 int sysctl_oom_kill_allocating_task;
26615 @@ -186,9 +189,21 @@ unsigned long badness(struct task_struct
26616 points >>= -(oom_adj);
26620 + * add points for context badness and
26621 + * reduce badness for processes belonging to
26622 + * a different context
26625 + points += vx_badness(p, mm);
26627 + if ((vx_current_xid() > 1) &&
26628 + vx_current_xid() != vx_task_xid(p))
26632 - printk(KERN_DEBUG "OOMkill: task %d (%s) got %lu points\n",
26633 - p->pid, p->comm, points);
26634 + printk(KERN_DEBUG "OOMkill: task %d:#%u (%s) got %d points\n",
26635 + task_pid_nr(p), p->xid, p->comm, points);
26639 @@ -230,6 +245,7 @@ static struct task_struct *select_bad_pr
26640 struct task_struct *p;
26641 struct task_struct *chosen = NULL;
26642 struct timespec uptime;
26643 + unsigned xid = vx_current_xid();
26646 do_posix_clock_monotonic_gettime(&uptime);
26647 @@ -242,11 +258,14 @@ static struct task_struct *select_bad_pr
26651 - /* skip the init task */
26652 - if (is_global_init(p))
26653 + /* skip the init task, global and per guest */
26654 + if (task_is_init(p))
26656 if (mem && !task_in_mem_cgroup(p, mem))
26658 + /* skip other guest and host processes if oom in guest */
26659 + if (xid && vx_task_xid(p) != xid)
26663 * This task already has access to memory reserves and is
26664 @@ -357,8 +376,8 @@ static void __oom_kill_task(struct task_
26668 - printk(KERN_ERR "Killed process %d (%s)\n",
26669 - task_pid_nr(p), p->comm);
26670 + printk(KERN_ERR "Killed process %s(%d:#%u)\n",
26671 + p->comm, task_pid_nr(p), p->xid);
26674 * We give our sacrificial lamb high priority and access to
26675 @@ -419,8 +438,8 @@ static int oom_kill_process(struct task_
26679 - printk(KERN_ERR "%s: kill process %d (%s) score %li or a child\n",
26680 - message, task_pid_nr(p), p->comm, points);
26681 + printk(KERN_ERR "%s: kill process %s(%d:#%u) score %li or a child\n",
26682 + message, p->comm, task_pid_nr(p), p->xid, points);
26684 /* Try to kill a child first */
26685 list_for_each_entry(c, &p->children, sibling) {
26686 @@ -519,6 +538,8 @@ void clear_zonelist_oom(struct zonelist
26687 spin_unlock(&zone_scan_lock);
26690 +long vs_oom_action(unsigned int);
26693 * Must be called with tasklist_lock held for read.
26695 @@ -544,7 +565,11 @@ retry:
26696 /* Found nothing?!?! Either we hang forever, or we panic. */
26698 read_unlock(&tasklist_lock);
26699 - panic("Out of memory and no killable processes...\n");
26700 + /* avoid panic for guest OOM */
26701 + if (current->xid)
26702 + vs_oom_action(LINUX_REBOOT_CMD_OOM);
26704 + panic("Out of memory and no killable processes...\n");
26707 if (oom_kill_process(p, gfp_mask, order, points, NULL,
26708 diff -NurpP --minimal linux-2.6.32.1/mm/page_alloc.c linux-2.6.32.1-vs2.3.0.36.27/mm/page_alloc.c
26709 --- linux-2.6.32.1/mm/page_alloc.c 2009-12-03 20:02:58.000000000 +0100
26710 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/page_alloc.c 2009-12-03 20:04:56.000000000 +0100
26712 #include <linux/page_cgroup.h>
26713 #include <linux/debugobjects.h>
26714 #include <linux/kmemleak.h>
26715 +#include <linux/vs_base.h>
26716 +#include <linux/vs_limit.h>
26717 #include <trace/events/kmem.h>
26719 #include <asm/tlbflush.h>
26720 @@ -2130,6 +2132,9 @@ void si_meminfo(struct sysinfo *val)
26721 val->totalhigh = totalhigh_pages;
26722 val->freehigh = nr_free_highpages();
26723 val->mem_unit = PAGE_SIZE;
26725 + if (vx_flags(VXF_VIRT_MEM, 0))
26726 + vx_vsi_meminfo(val);
26729 EXPORT_SYMBOL(si_meminfo);
26730 @@ -2150,6 +2155,9 @@ void si_meminfo_node(struct sysinfo *val
26733 val->mem_unit = PAGE_SIZE;
26735 + if (vx_flags(VXF_VIRT_MEM, 0))
26736 + vx_vsi_meminfo(val);
26740 diff -NurpP --minimal linux-2.6.32.1/mm/rmap.c linux-2.6.32.1-vs2.3.0.36.27/mm/rmap.c
26741 --- linux-2.6.32.1/mm/rmap.c 2009-12-03 20:02:58.000000000 +0100
26742 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/rmap.c 2009-12-03 20:04:56.000000000 +0100
26744 #include <linux/memcontrol.h>
26745 #include <linux/mmu_notifier.h>
26746 #include <linux/migrate.h>
26747 +#include <linux/vs_memory.h>
26749 #include <asm/tlbflush.h>
26751 diff -NurpP --minimal linux-2.6.32.1/mm/shmem.c linux-2.6.32.1-vs2.3.0.36.27/mm/shmem.c
26752 --- linux-2.6.32.1/mm/shmem.c 2009-12-03 20:02:58.000000000 +0100
26753 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/shmem.c 2009-12-03 20:04:56.000000000 +0100
26754 @@ -1781,7 +1781,7 @@ static int shmem_statfs(struct dentry *d
26756 struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
26758 - buf->f_type = TMPFS_MAGIC;
26759 + buf->f_type = TMPFS_SUPER_MAGIC;
26760 buf->f_bsize = PAGE_CACHE_SIZE;
26761 buf->f_namelen = NAME_MAX;
26762 spin_lock(&sbinfo->stat_lock);
26763 @@ -2346,7 +2346,7 @@ int shmem_fill_super(struct super_block
26764 sb->s_maxbytes = SHMEM_MAX_BYTES;
26765 sb->s_blocksize = PAGE_CACHE_SIZE;
26766 sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
26767 - sb->s_magic = TMPFS_MAGIC;
26768 + sb->s_magic = TMPFS_SUPER_MAGIC;
26769 sb->s_op = &shmem_ops;
26770 sb->s_time_gran = 1;
26771 #ifdef CONFIG_TMPFS_POSIX_ACL
26772 diff -NurpP --minimal linux-2.6.32.1/mm/slab.c linux-2.6.32.1-vs2.3.0.36.27/mm/slab.c
26773 --- linux-2.6.32.1/mm/slab.c 2009-12-03 20:02:58.000000000 +0100
26774 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/slab.c 2009-12-03 20:04:56.000000000 +0100
26775 @@ -431,6 +431,8 @@ static void kmem_list3_init(struct kmem_
26776 #define STATS_INC_FREEMISS(x) do { } while (0)
26779 +#include "slab_vs.h"
26784 @@ -3253,6 +3255,7 @@ retry:
26786 obj = slab_get_obj(cachep, slabp, nodeid);
26787 check_slabp(cachep, slabp);
26788 + vx_slab_alloc(cachep, flags);
26789 l3->free_objects--;
26790 /* move slabp to correct slabp list: */
26791 list_del(&slabp->list);
26792 @@ -3329,6 +3332,7 @@ __cache_alloc_node(struct kmem_cache *ca
26793 /* ___cache_alloc_node can fall back to other nodes */
26794 ptr = ____cache_alloc_node(cachep, flags, nodeid);
26796 + vx_slab_alloc(cachep, flags);
26797 local_irq_restore(save_flags);
26798 ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller);
26799 kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags,
26800 @@ -3515,6 +3519,7 @@ static inline void __cache_free(struct k
26802 kmemleak_free_recursive(objp, cachep->flags);
26803 objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0));
26804 + vx_slab_free(cachep);
26806 kmemcheck_slab_free(cachep, objp, obj_size(cachep));
26808 diff -NurpP --minimal linux-2.6.32.1/mm/slab_vs.h linux-2.6.32.1-vs2.3.0.36.27/mm/slab_vs.h
26809 --- linux-2.6.32.1/mm/slab_vs.h 1970-01-01 01:00:00.000000000 +0100
26810 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/slab_vs.h 2009-12-03 20:04:56.000000000 +0100
26813 +#include <linux/vserver/context.h>
26815 +#include <linux/vs_context.h>
26818 +void vx_slab_alloc(struct kmem_cache *cachep, gfp_t flags)
26820 + int what = gfp_zone(cachep->gfpflags);
26821 + struct vx_info *vxi = current_vx_info();
26826 + atomic_add(cachep->buffer_size, &vxi->cacct.slab[what]);
26830 +void vx_slab_free(struct kmem_cache *cachep)
26832 + int what = gfp_zone(cachep->gfpflags);
26833 + struct vx_info *vxi = current_vx_info();
26838 + atomic_sub(cachep->buffer_size, &vxi->cacct.slab[what]);
26841 diff -NurpP --minimal linux-2.6.32.1/mm/swapfile.c linux-2.6.32.1-vs2.3.0.36.27/mm/swapfile.c
26842 --- linux-2.6.32.1/mm/swapfile.c 2009-12-03 20:02:58.000000000 +0100
26843 +++ linux-2.6.32.1-vs2.3.0.36.27/mm/swapfile.c 2009-12-03 20:04:56.000000000 +0100
26845 #include <asm/tlbflush.h>
26846 #include <linux/swapops.h>
26847 #include <linux/page_cgroup.h>
26848 +#include <linux/vs_base.h>
26849 +#include <linux/vs_memory.h>
26851 static DEFINE_SPINLOCK(swap_lock);
26852 static unsigned int nr_swapfiles;
26853 @@ -1680,6 +1682,8 @@ static void *swap_next(struct seq_file *
26854 if (v == SEQ_START_TOKEN)
26857 + if (vx_flags(VXF_VIRT_MEM, 0))
26862 @@ -1707,6 +1711,16 @@ static int swap_show(struct seq_file *sw
26864 if (ptr == SEQ_START_TOKEN) {
26865 seq_puts(swap,"Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
26866 + if (vx_flags(VXF_VIRT_MEM, 0)) {
26867 + struct sysinfo si;
26869 + vx_vsi_swapinfo(&si);
26870 + if (si.totalswap < (1 << 10))
26872 + seq_printf(swap, "%s\t\t\t\t\t%s\t%lu\t%lu\t%d\n",
26873 + "hdv0", "partition", si.totalswap >> 10,
26874 + (si.totalswap - si.freeswap) >> 10, -1);
26879 @@ -2064,6 +2078,8 @@ void si_swapinfo(struct sysinfo *val)
26880 val->freeswap = nr_swap_pages + nr_to_be_unused;
26881 val->totalswap = total_swap_pages + nr_to_be_unused;
26882 spin_unlock(&swap_lock);
26883 + if (vx_flags(VXF_VIRT_MEM, 0))
26884 + vx_vsi_swapinfo(val);
26888 diff -NurpP --minimal linux-2.6.32.1/net/core/dev.c linux-2.6.32.1-vs2.3.0.36.27/net/core/dev.c
26889 --- linux-2.6.32.1/net/core/dev.c 2009-12-03 20:02:59.000000000 +0100
26890 +++ linux-2.6.32.1-vs2.3.0.36.27/net/core/dev.c 2009-12-03 20:04:56.000000000 +0100
26891 @@ -126,6 +126,7 @@
26892 #include <linux/in.h>
26893 #include <linux/jhash.h>
26894 #include <linux/random.h>
26895 +#include <linux/vs_inet.h>
26896 #include <trace/events/napi.h>
26898 #include "net-sysfs.h"
26899 @@ -591,7 +592,8 @@ struct net_device *__dev_get_by_name(str
26900 hlist_for_each(p, dev_name_hash(net, name)) {
26901 struct net_device *dev
26902 = hlist_entry(p, struct net_device, name_hlist);
26903 - if (!strncmp(dev->name, name, IFNAMSIZ))
26904 + if (!strncmp(dev->name, name, IFNAMSIZ) &&
26905 + nx_dev_visible(current_nx_info(), dev))
26909 @@ -642,7 +644,8 @@ struct net_device *__dev_get_by_index(st
26910 hlist_for_each(p, dev_index_hash(net, ifindex)) {
26911 struct net_device *dev
26912 = hlist_entry(p, struct net_device, index_hlist);
26913 - if (dev->ifindex == ifindex)
26914 + if ((dev->ifindex == ifindex) &&
26915 + nx_dev_visible(current_nx_info(), dev))
26919 @@ -695,10 +698,12 @@ struct net_device *dev_getbyhwaddr(struc
26923 - for_each_netdev(net, dev)
26924 + for_each_netdev(net, dev) {
26925 if (dev->type == type &&
26926 - !memcmp(dev->dev_addr, ha, dev->addr_len))
26927 + !memcmp(dev->dev_addr, ha, dev->addr_len) &&
26928 + nx_dev_visible(current_nx_info(), dev))
26934 @@ -709,9 +714,11 @@ struct net_device *__dev_getfirstbyhwtyp
26935 struct net_device *dev;
26938 - for_each_netdev(net, dev)
26939 - if (dev->type == type)
26940 + for_each_netdev(net, dev) {
26941 + if ((dev->type == type) &&
26942 + nx_dev_visible(current_nx_info(), dev))
26948 @@ -830,6 +837,8 @@ static int __dev_alloc_name(struct net *
26950 if (i < 0 || i >= max_netdevices)
26952 + if (!nx_dev_visible(current_nx_info(), d))
26955 /* avoid cases where sscanf is not exact inverse of printf */
26956 snprintf(buf, IFNAMSIZ, name, i);
26957 @@ -2984,6 +2993,8 @@ static int dev_ifconf(struct net *net, c
26960 for_each_netdev(net, dev) {
26961 + if (!nx_dev_visible(current_nx_info(), dev))
26963 for (i = 0; i < NPROTO; i++) {
26964 if (gifconf_list[i]) {
26966 @@ -3052,6 +3063,9 @@ static void dev_seq_printf_stats(struct
26968 const struct net_device_stats *stats = dev_get_stats(dev);
26970 + if (!nx_dev_visible(current_nx_info(), dev))
26973 seq_printf(seq, "%6s:%8lu %7lu %4lu %4lu %4lu %5lu %10lu %9lu "
26974 "%8lu %7lu %4lu %4lu %4lu %5lu %7lu %10lu\n",
26975 dev->name, stats->rx_bytes, stats->rx_packets,
26976 @@ -5313,6 +5327,15 @@ int dev_change_net_namespace(struct net_
26980 +#ifdef CONFIG_SYSFS
26981 + /* Don't allow real devices to be moved when sysfs
26985 + if (dev->dev.parent)
26989 /* Ensure the device has been registrered */
26991 if (dev->reg_state != NETREG_REGISTERED)
26992 diff -NurpP --minimal linux-2.6.32.1/net/core/rtnetlink.c linux-2.6.32.1-vs2.3.0.36.27/net/core/rtnetlink.c
26993 --- linux-2.6.32.1/net/core/rtnetlink.c 2009-12-03 20:02:59.000000000 +0100
26994 +++ linux-2.6.32.1-vs2.3.0.36.27/net/core/rtnetlink.c 2009-12-03 20:04:56.000000000 +0100
26995 @@ -688,6 +688,8 @@ static int rtnl_dump_ifinfo(struct sk_bu
26998 for_each_netdev(net, dev) {
26999 + if (!nx_dev_visible(skb->sk->sk_nx_info, dev))
27003 if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK,
27004 @@ -1222,6 +1224,9 @@ void rtmsg_ifinfo(int type, struct net_d
27005 struct sk_buff *skb;
27006 int err = -ENOBUFS;
27008 + if (!nx_dev_visible(current_nx_info(), dev))
27011 skb = nlmsg_new(if_nlmsg_size(dev), GFP_KERNEL);
27014 diff -NurpP --minimal linux-2.6.32.1/net/core/sock.c linux-2.6.32.1-vs2.3.0.36.27/net/core/sock.c
27015 --- linux-2.6.32.1/net/core/sock.c 2009-12-03 20:02:59.000000000 +0100
27016 +++ linux-2.6.32.1-vs2.3.0.36.27/net/core/sock.c 2009-12-03 20:04:56.000000000 +0100
27017 @@ -125,6 +125,10 @@
27018 #include <linux/ipsec.h>
27020 #include <linux/filter.h>
27021 +#include <linux/vs_socket.h>
27022 +#include <linux/vs_limit.h>
27023 +#include <linux/vs_context.h>
27024 +#include <linux/vs_network.h>
27027 #include <net/tcp.h>
27028 @@ -984,6 +988,8 @@ static struct sock *sk_prot_alloc(struct
27029 if (!try_module_get(prot->owner))
27032 + sock_vx_init(sk);
27033 + sock_nx_init(sk);
27037 @@ -1063,6 +1069,11 @@ static void __sk_free(struct sock *sk)
27038 __func__, atomic_read(&sk->sk_omem_alloc));
27040 put_net(sock_net(sk));
27042 + clr_vx_info(&sk->sk_vx_info);
27044 + clr_nx_info(&sk->sk_nx_info);
27046 sk_prot_free(sk->sk_prot_creator, sk);
27049 @@ -1110,6 +1121,8 @@ struct sock *sk_clone(const struct sock
27052 get_net(sock_net(newsk));
27053 + sock_vx_init(newsk);
27054 + sock_nx_init(newsk);
27055 sk_node_init(&newsk->sk_node);
27056 sock_lock_init(newsk);
27057 bh_lock_sock(newsk);
27058 @@ -1164,6 +1177,12 @@ struct sock *sk_clone(const struct sock
27060 atomic_set(&newsk->sk_refcnt, 2);
27062 + set_vx_info(&newsk->sk_vx_info, sk->sk_vx_info);
27063 + newsk->sk_xid = sk->sk_xid;
27064 + vx_sock_inc(newsk);
27065 + set_nx_info(&newsk->sk_nx_info, sk->sk_nx_info);
27066 + newsk->sk_nid = sk->sk_nid;
27069 * Increment the counter in the same struct proto as the master
27070 * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
27071 @@ -1882,6 +1901,12 @@ void sock_init_data(struct socket *sock,
27073 sk->sk_stamp = ktime_set(-1L, 0);
27075 + set_vx_info(&sk->sk_vx_info, current_vx_info());
27076 + sk->sk_xid = vx_current_xid();
27078 + set_nx_info(&sk->sk_nx_info, current_nx_info());
27079 + sk->sk_nid = nx_current_nid();
27082 * Before updating sk_refcnt, we must commit prior changes to memory
27083 * (Documentation/RCU/rculist_nulls.txt for details)
27084 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/af_inet.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/af_inet.c
27085 --- linux-2.6.32.1/net/ipv4/af_inet.c 2009-12-03 20:02:59.000000000 +0100
27086 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/af_inet.c 2009-12-03 20:04:56.000000000 +0100
27087 @@ -115,6 +115,7 @@
27088 #ifdef CONFIG_IP_MROUTE
27089 #include <linux/mroute.h>
27091 +#include <linux/vs_limit.h>
27094 /* The inetsw table contains everything that inet_create needs to
27095 @@ -325,9 +326,12 @@ lookup_protocol:
27099 + if ((protocol == IPPROTO_ICMP) &&
27100 + nx_capable(answer->capability, NXC_RAW_ICMP))
27102 if (answer->capability > 0 && !capable(answer->capability))
27103 goto out_rcu_unlock;
27106 err = -EAFNOSUPPORT;
27107 if (!inet_netns_ok(net, protocol))
27108 goto out_rcu_unlock;
27109 @@ -447,6 +451,7 @@ int inet_bind(struct socket *sock, struc
27110 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
27111 struct sock *sk = sock->sk;
27112 struct inet_sock *inet = inet_sk(sk);
27113 + struct nx_v4_sock_addr nsa;
27114 unsigned short snum;
27117 @@ -460,7 +465,11 @@ int inet_bind(struct socket *sock, struc
27118 if (addr_len < sizeof(struct sockaddr_in))
27121 - chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
27122 + err = v4_map_sock_addr(inet, addr, &nsa);
27126 + chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
27128 /* Not specified by any standard per-se, however it breaks too
27129 * many applications when removed. It is unfortunate since
27130 @@ -472,7 +481,7 @@ int inet_bind(struct socket *sock, struc
27131 err = -EADDRNOTAVAIL;
27132 if (!sysctl_ip_nonlocal_bind &&
27133 !(inet->freebind || inet->transparent) &&
27134 - addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
27135 + nsa.saddr != htonl(INADDR_ANY) &&
27136 chk_addr_ret != RTN_LOCAL &&
27137 chk_addr_ret != RTN_MULTICAST &&
27138 chk_addr_ret != RTN_BROADCAST)
27139 @@ -497,7 +506,7 @@ int inet_bind(struct socket *sock, struc
27140 if (sk->sk_state != TCP_CLOSE || inet->num)
27141 goto out_release_sock;
27143 - inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
27144 + v4_set_sock_addr(inet, &nsa);
27145 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
27146 inet->saddr = 0; /* Use device */
27148 @@ -694,11 +703,13 @@ int inet_getname(struct socket *sock, st
27151 sin->sin_port = inet->dport;
27152 - sin->sin_addr.s_addr = inet->daddr;
27153 + sin->sin_addr.s_addr =
27154 + nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
27156 __be32 addr = inet->rcv_saddr;
27158 addr = inet->saddr;
27159 + addr = nx_map_sock_lback(sk->sk_nx_info, addr);
27160 sin->sin_port = inet->sport;
27161 sin->sin_addr.s_addr = addr;
27163 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/devinet.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/devinet.c
27164 --- linux-2.6.32.1/net/ipv4/devinet.c 2009-12-03 20:02:59.000000000 +0100
27165 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/devinet.c 2009-12-03 20:04:56.000000000 +0100
27166 @@ -413,6 +413,7 @@ struct in_device *inetdev_by_index(struc
27171 /* Called only from RTNL semaphored context. No locks. */
27173 struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix,
27174 @@ -653,6 +654,8 @@ int devinet_ioctl(struct net *net, unsig
27177 if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) {
27178 + struct nx_info *nxi = current_nx_info();
27180 if (tryaddrmatch) {
27181 /* Matthias Andree */
27182 /* compare label and address (4.4BSD style) */
27183 @@ -661,6 +664,8 @@ int devinet_ioctl(struct net *net, unsig
27184 This is checked above. */
27185 for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
27186 ifap = &ifa->ifa_next) {
27187 + if (!nx_v4_ifa_visible(nxi, ifa))
27189 if (!strcmp(ifr.ifr_name, ifa->ifa_label) &&
27190 sin_orig.sin_addr.s_addr ==
27191 ifa->ifa_address) {
27192 @@ -673,9 +678,12 @@ int devinet_ioctl(struct net *net, unsig
27193 comparing just the label */
27195 for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
27196 - ifap = &ifa->ifa_next)
27197 + ifap = &ifa->ifa_next) {
27198 + if (!nx_v4_ifa_visible(nxi, ifa))
27200 if (!strcmp(ifr.ifr_name, ifa->ifa_label))
27206 @@ -826,6 +834,8 @@ static int inet_gifconf(struct net_devic
27209 for (; ifa; ifa = ifa->ifa_next) {
27210 + if (!nx_v4_ifa_visible(current_nx_info(), ifa))
27213 done += sizeof(ifr);
27215 @@ -1174,6 +1184,7 @@ static int inet_dump_ifaddr(struct sk_bu
27216 struct net_device *dev;
27217 struct in_device *in_dev;
27218 struct in_ifaddr *ifa;
27219 + struct sock *sk = skb->sk;
27220 int s_ip_idx, s_idx = cb->args[0];
27222 s_ip_idx = ip_idx = cb->args[1];
27223 @@ -1188,6 +1199,8 @@ static int inet_dump_ifaddr(struct sk_bu
27225 for (ifa = in_dev->ifa_list, ip_idx = 0; ifa;
27226 ifa = ifa->ifa_next, ip_idx++) {
27227 + if (sk && !nx_v4_ifa_visible(sk->sk_nx_info, ifa))
27229 if (ip_idx < s_ip_idx)
27231 if (inet_fill_ifaddr(skb, ifa, NETLINK_CB(cb->skb).pid,
27232 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/fib_hash.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/fib_hash.c
27233 --- linux-2.6.32.1/net/ipv4/fib_hash.c 2009-09-10 15:26:29.000000000 +0200
27234 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/fib_hash.c 2009-12-03 20:04:56.000000000 +0100
27235 @@ -1021,7 +1021,7 @@ static int fib_seq_show(struct seq_file
27236 prefix = f->fn_key;
27237 mask = FZ_MASK(iter->zone);
27238 flags = fib_flag_trans(fa->fa_type, mask, fi);
27240 + if (fi && nx_dev_visible(current_nx_info(), fi->fib_dev))
27242 "%s\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u%n",
27243 fi->fib_dev ? fi->fib_dev->name : "*", prefix,
27244 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/inet_connection_sock.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_connection_sock.c
27245 --- linux-2.6.32.1/net/ipv4/inet_connection_sock.c 2009-12-03 20:02:59.000000000 +0100
27246 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_connection_sock.c 2009-12-03 20:04:56.000000000 +0100
27247 @@ -49,10 +49,40 @@ void inet_get_local_port_range(int *low,
27249 EXPORT_SYMBOL(inet_get_local_port_range);
27251 +int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
27253 + __be32 sk1_rcv_saddr = inet_rcv_saddr(sk1),
27254 + sk2_rcv_saddr = inet_rcv_saddr(sk2);
27256 + if (inet_v6_ipv6only(sk2))
27259 + if (sk1_rcv_saddr &&
27261 + sk1_rcv_saddr == sk2_rcv_saddr)
27264 + if (sk1_rcv_saddr &&
27265 + !sk2_rcv_saddr &&
27266 + v4_addr_in_nx_info(sk2->sk_nx_info, sk1_rcv_saddr, NXA_MASK_BIND))
27269 + if (sk2_rcv_saddr &&
27270 + !sk1_rcv_saddr &&
27271 + v4_addr_in_nx_info(sk1->sk_nx_info, sk2_rcv_saddr, NXA_MASK_BIND))
27274 + if (!sk1_rcv_saddr &&
27275 + !sk2_rcv_saddr &&
27276 + nx_v4_addr_conflict(sk1->sk_nx_info, sk2->sk_nx_info))
27282 int inet_csk_bind_conflict(const struct sock *sk,
27283 const struct inet_bind_bucket *tb)
27285 - const __be32 sk_rcv_saddr = inet_rcv_saddr(sk);
27287 struct hlist_node *node;
27288 int reuse = sk->sk_reuse;
27289 @@ -72,9 +102,7 @@ int inet_csk_bind_conflict(const struct
27290 sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
27291 if (!reuse || !sk2->sk_reuse ||
27292 sk2->sk_state == TCP_LISTEN) {
27293 - const __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
27294 - if (!sk2_rcv_saddr || !sk_rcv_saddr ||
27295 - sk2_rcv_saddr == sk_rcv_saddr)
27296 + if (ipv4_rcv_saddr_equal(sk, sk2))
27300 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/inet_diag.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_diag.c
27301 --- linux-2.6.32.1/net/ipv4/inet_diag.c 2009-09-10 15:26:29.000000000 +0200
27302 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_diag.c 2009-12-03 20:04:56.000000000 +0100
27304 #include <linux/stddef.h>
27306 #include <linux/inet_diag.h>
27307 +#include <linux/vs_network.h>
27308 +#include <linux/vs_inet.h>
27310 static const struct inet_diag_handler **inet_diag_table;
27312 @@ -118,8 +120,8 @@ static int inet_csk_diag_fill(struct soc
27314 r->id.idiag_sport = inet->sport;
27315 r->id.idiag_dport = inet->dport;
27316 - r->id.idiag_src[0] = inet->rcv_saddr;
27317 - r->id.idiag_dst[0] = inet->daddr;
27318 + r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, inet->rcv_saddr);
27319 + r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
27321 #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
27322 if (r->idiag_family == AF_INET6) {
27323 @@ -204,8 +206,8 @@ static int inet_twsk_diag_fill(struct in
27324 r->id.idiag_cookie[1] = (u32)(((unsigned long)tw >> 31) >> 1);
27325 r->id.idiag_sport = tw->tw_sport;
27326 r->id.idiag_dport = tw->tw_dport;
27327 - r->id.idiag_src[0] = tw->tw_rcv_saddr;
27328 - r->id.idiag_dst[0] = tw->tw_daddr;
27329 + r->id.idiag_src[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_rcv_saddr);
27330 + r->id.idiag_dst[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_daddr);
27331 r->idiag_state = tw->tw_substate;
27332 r->idiag_timer = 3;
27333 r->idiag_expires = DIV_ROUND_UP(tmo * 1000, HZ);
27334 @@ -262,6 +264,7 @@ static int inet_diag_get_exact(struct sk
27337 if (req->idiag_family == AF_INET) {
27338 + /* TODO: lback */
27339 sk = inet_lookup(&init_net, hashinfo, req->id.idiag_dst[0],
27340 req->id.idiag_dport, req->id.idiag_src[0],
27341 req->id.idiag_sport, req->id.idiag_if);
27342 @@ -504,6 +507,7 @@ static int inet_csk_diag_dump(struct soc
27346 + /* TODO: lback */
27347 entry.saddr = &inet->rcv_saddr;
27348 entry.daddr = &inet->daddr;
27350 @@ -540,6 +544,7 @@ static int inet_twsk_diag_dump(struct in
27354 + /* TODO: lback */
27355 entry.saddr = &tw->tw_rcv_saddr;
27356 entry.daddr = &tw->tw_daddr;
27358 @@ -586,8 +591,8 @@ static int inet_diag_fill_req(struct sk_
27360 r->id.idiag_sport = inet->sport;
27361 r->id.idiag_dport = ireq->rmt_port;
27362 - r->id.idiag_src[0] = ireq->loc_addr;
27363 - r->id.idiag_dst[0] = ireq->rmt_addr;
27364 + r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->loc_addr);
27365 + r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->rmt_addr);
27366 r->idiag_expires = jiffies_to_msecs(tmo);
27367 r->idiag_rqueue = 0;
27368 r->idiag_wqueue = 0;
27369 @@ -657,6 +662,7 @@ static int inet_diag_dump_reqs(struct sk
27373 + /* TODO: lback */
27375 #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
27376 (entry.family == AF_INET6) ?
27377 @@ -727,6 +733,8 @@ static int inet_diag_dump(struct sk_buff
27378 sk_nulls_for_each(sk, node, &ilb->head) {
27379 struct inet_sock *inet = inet_sk(sk);
27381 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27386 @@ -793,6 +801,8 @@ skip_listen_ht:
27387 sk_nulls_for_each(sk, node, &head->chain) {
27388 struct inet_sock *inet = inet_sk(sk);
27390 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27394 if (!(r->idiag_states & (1 << sk->sk_state)))
27395 @@ -817,6 +827,8 @@ next_normal:
27396 inet_twsk_for_each(tw, node,
27399 + if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
27403 if (r->id.idiag_sport != tw->tw_sport &&
27404 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/inet_hashtables.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_hashtables.c
27405 --- linux-2.6.32.1/net/ipv4/inet_hashtables.c 2009-06-11 17:13:29.000000000 +0200
27406 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/inet_hashtables.c 2009-12-03 20:04:56.000000000 +0100
27409 #include <net/inet_connection_sock.h>
27410 #include <net/inet_hashtables.h>
27411 +#include <net/route.h>
27412 #include <net/ip.h>
27415 @@ -134,6 +135,11 @@ static inline int compute_score(struct s
27416 if (rcv_saddr != daddr)
27420 + /* block non nx_info ips */
27421 + if (!v4_addr_in_nx_info(sk->sk_nx_info,
27422 + daddr, NXA_MASK_BIND))
27425 if (sk->sk_bound_dev_if) {
27426 if (sk->sk_bound_dev_if != dif)
27427 @@ -151,7 +157,6 @@ static inline int compute_score(struct s
27428 * wildcarded during the search since they can never be otherwise.
27432 struct sock *__inet_lookup_listener(struct net *net,
27433 struct inet_hashinfo *hashinfo,
27434 const __be32 daddr, const unsigned short hnum,
27435 @@ -174,6 +179,7 @@ begin:
27441 * if the nulls value we got at the end of this lookup is
27442 * not the expected one, we must restart lookup.
27443 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/netfilter/nf_nat_helper.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/netfilter/nf_nat_helper.c
27444 --- linux-2.6.32.1/net/ipv4/netfilter/nf_nat_helper.c 2009-12-03 20:02:59.000000000 +0100
27445 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/netfilter/nf_nat_helper.c 2009-12-03 20:04:56.000000000 +0100
27447 #include <net/route.h>
27449 #include <linux/netfilter_ipv4.h>
27450 +#include <net/route.h>
27451 #include <net/netfilter/nf_conntrack.h>
27452 #include <net/netfilter/nf_conntrack_helper.h>
27453 #include <net/netfilter/nf_conntrack_ecache.h>
27454 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/netfilter.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/netfilter.c
27455 --- linux-2.6.32.1/net/ipv4/netfilter.c 2009-09-10 15:26:29.000000000 +0200
27456 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/netfilter.c 2009-12-03 20:04:56.000000000 +0100
27458 #include <linux/netfilter_ipv4.h>
27459 #include <linux/ip.h>
27460 #include <linux/skbuff.h>
27461 -#include <net/route.h>
27462 +// #include <net/route.h>
27463 #include <net/xfrm.h>
27464 #include <net/ip.h>
27465 #include <net/netfilter/nf_queue.h>
27466 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/raw.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/raw.c
27467 --- linux-2.6.32.1/net/ipv4/raw.c 2009-12-03 20:02:59.000000000 +0100
27468 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/raw.c 2009-12-03 20:04:56.000000000 +0100
27469 @@ -117,7 +117,7 @@ static struct sock *__raw_v4_lookup(stru
27471 if (net_eq(sock_net(sk), net) && inet->num == num &&
27472 !(inet->daddr && inet->daddr != raddr) &&
27473 - !(inet->rcv_saddr && inet->rcv_saddr != laddr) &&
27474 + v4_sock_addr_match(sk->sk_nx_info, inet, laddr) &&
27475 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
27476 goto found; /* gotcha */
27478 @@ -383,6 +383,12 @@ static int raw_send_hdrinc(struct sock *
27479 icmp_out_count(net, ((struct icmphdr *)
27480 skb_transport_header(skb))->type);
27483 + if (!nx_check(0, VS_ADMIN) && !capable(CAP_NET_RAW) &&
27484 + sk->sk_nx_info &&
27485 + !v4_addr_in_nx_info(sk->sk_nx_info, iph->saddr, NXA_MASK_BIND))
27488 err = NF_HOOK(PF_INET, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
27491 @@ -563,6 +569,13 @@ static int raw_sendmsg(struct kiocb *ioc
27494 security_sk_classify_flow(sk, &fl);
27495 + if (sk->sk_nx_info) {
27496 + err = ip_v4_find_src(sock_net(sk),
27497 + sk->sk_nx_info, &rt, &fl);
27502 err = ip_route_output_flow(sock_net(sk), &rt, &fl, sk, 1);
27505 @@ -635,17 +648,19 @@ static int raw_bind(struct sock *sk, str
27507 struct inet_sock *inet = inet_sk(sk);
27508 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
27509 + struct nx_v4_sock_addr nsa = { 0 };
27513 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
27515 - chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
27516 + v4_map_sock_addr(inet, addr, &nsa);
27517 + chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
27518 ret = -EADDRNOTAVAIL;
27519 - if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
27520 + if (nsa.saddr && chk_addr_ret != RTN_LOCAL &&
27521 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
27523 - inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
27524 + v4_set_sock_addr(inet, &nsa);
27525 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
27526 inet->saddr = 0; /* Use device */
27528 @@ -697,7 +712,8 @@ static int raw_recvmsg(struct kiocb *ioc
27529 /* Copy the address. */
27531 sin->sin_family = AF_INET;
27532 - sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
27533 + sin->sin_addr.s_addr =
27534 + nx_map_sock_lback(sk->sk_nx_info, ip_hdr(skb)->saddr);
27536 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
27538 @@ -875,7 +891,8 @@ static struct sock *raw_get_first(struct
27539 struct hlist_node *node;
27541 sk_for_each(sk, node, &state->h->ht[state->bucket])
27542 - if (sock_net(sk) == seq_file_net(seq))
27543 + if ((sock_net(sk) == seq_file_net(seq)) &&
27544 + nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27548 @@ -891,7 +908,8 @@ static struct sock *raw_get_next(struct
27552 - } while (sk && sock_net(sk) != seq_file_net(seq));
27553 + } while (sk && ((sock_net(sk) != seq_file_net(seq)) ||
27554 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
27556 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
27557 sk = sk_head(&state->h->ht[state->bucket]);
27558 @@ -950,7 +968,10 @@ static void raw_sock_seq_show(struct seq
27560 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
27561 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
27562 - i, src, srcp, dest, destp, sp->sk_state,
27564 + nx_map_sock_lback(current_nx_info(), src), srcp,
27565 + nx_map_sock_lback(current_nx_info(), dest), destp,
27567 sk_wmem_alloc_get(sp),
27568 sk_rmem_alloc_get(sp),
27569 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
27570 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/tcp.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp.c
27571 --- linux-2.6.32.1/net/ipv4/tcp.c 2009-12-03 20:02:59.000000000 +0100
27572 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp.c 2009-12-03 20:04:56.000000000 +0100
27573 @@ -264,6 +264,7 @@
27574 #include <linux/cache.h>
27575 #include <linux/err.h>
27576 #include <linux/crypto.h>
27577 +#include <linux/in.h>
27579 #include <net/icmp.h>
27580 #include <net/tcp.h>
27581 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/tcp_ipv4.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp_ipv4.c
27582 --- linux-2.6.32.1/net/ipv4/tcp_ipv4.c 2009-12-03 20:03:00.000000000 +0100
27583 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp_ipv4.c 2009-12-03 20:04:56.000000000 +0100
27584 @@ -1925,6 +1925,12 @@ static void *listening_get_next(struct s
27585 req = req->dl_next;
27588 + vxdprintk(VXD_CBIT(net, 6),
27589 + "sk,req: %p [#%d] (from %d)", req->sk,
27590 + (req->sk)?req->sk->sk_nid:0, nx_current_nid());
27592 + !nx_check(req->sk->sk_nid, VS_WATCH_P | VS_IDENT))
27594 if (req->rsk_ops->family == st->family) {
27597 @@ -1949,6 +1955,10 @@ get_req:
27600 sk_nulls_for_each_from(sk, node) {
27601 + vxdprintk(VXD_CBIT(net, 6), "sk: %p [#%d] (from %d)",
27602 + sk, sk->sk_nid, nx_current_nid());
27603 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27605 if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
27608 @@ -2012,6 +2022,11 @@ static void *established_get_first(struc
27610 spin_lock_bh(lock);
27611 sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
27612 + vxdprintk(VXD_CBIT(net, 6),
27613 + "sk,egf: %p [#%d] (from %d)",
27614 + sk, sk->sk_nid, nx_current_nid());
27615 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27617 if (sk->sk_family != st->family ||
27618 !net_eq(sock_net(sk), net)) {
27620 @@ -2022,6 +2037,11 @@ static void *established_get_first(struc
27621 st->state = TCP_SEQ_STATE_TIME_WAIT;
27622 inet_twsk_for_each(tw, node,
27623 &tcp_hashinfo.ehash[st->bucket].twchain) {
27624 + vxdprintk(VXD_CBIT(net, 6),
27625 + "tw: %p [#%d] (from %d)",
27626 + tw, tw->tw_nid, nx_current_nid());
27627 + if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
27629 if (tw->tw_family != st->family ||
27630 !net_eq(twsk_net(tw), net)) {
27632 @@ -2050,7 +2070,9 @@ static void *established_get_next(struct
27636 - while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
27637 + while (tw && (tw->tw_family != st->family ||
27638 + !net_eq(twsk_net(tw), net) ||
27639 + !nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))) {
27643 @@ -2073,6 +2095,11 @@ get_tw:
27644 sk = sk_nulls_next(sk);
27646 sk_nulls_for_each_from(sk, node) {
27647 + vxdprintk(VXD_CBIT(net, 6),
27648 + "sk,egn: %p [#%d] (from %d)",
27649 + sk, sk->sk_nid, nx_current_nid());
27650 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27652 if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
27655 @@ -2224,9 +2251,9 @@ static void get_openreq4(struct sock *sk
27656 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27657 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p%n",
27660 + nx_map_sock_lback(current_nx_info(), ireq->loc_addr),
27661 ntohs(inet_sk(sk)->sport),
27663 + nx_map_sock_lback(current_nx_info(), ireq->rmt_addr),
27664 ntohs(ireq->rmt_port),
27666 0, 0, /* could print option size, but that is af dependent. */
27667 @@ -2269,7 +2296,10 @@ static void get_tcp4_sock(struct sock *s
27669 seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
27670 "%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
27671 - i, src, srcp, dest, destp, sk->sk_state,
27673 + nx_map_sock_lback(current_nx_info(), src), srcp,
27674 + nx_map_sock_lback(current_nx_info(), dest), destp,
27676 tp->write_seq - tp->snd_una,
27677 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
27678 (tp->rcv_nxt - tp->copied_seq),
27679 @@ -2305,7 +2335,10 @@ static void get_timewait4_sock(struct in
27681 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27682 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p%n",
27683 - i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
27685 + nx_map_sock_lback(current_nx_info(), src), srcp,
27686 + nx_map_sock_lback(current_nx_info(), dest), destp,
27687 + tw->tw_substate, 0, 0,
27688 3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
27689 atomic_read(&tw->tw_refcnt), tw, len);
27691 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/tcp_minisocks.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp_minisocks.c
27692 --- linux-2.6.32.1/net/ipv4/tcp_minisocks.c 2009-12-03 20:03:00.000000000 +0100
27693 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/tcp_minisocks.c 2009-12-03 20:04:56.000000000 +0100
27695 #include <net/inet_common.h>
27696 #include <net/xfrm.h>
27698 +#include <linux/vs_limit.h>
27699 +#include <linux/vs_socket.h>
27700 +#include <linux/vs_context.h>
27702 #ifdef CONFIG_SYSCTL
27703 #define SYNC_INIT 0 /* let the user enable it */
27705 @@ -294,6 +298,11 @@ void tcp_time_wait(struct sock *sk, int
27706 tcptw->tw_ts_recent = tp->rx_opt.ts_recent;
27707 tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
27709 + tw->tw_xid = sk->sk_xid;
27710 + tw->tw_vx_info = NULL;
27711 + tw->tw_nid = sk->sk_nid;
27712 + tw->tw_nx_info = NULL;
27714 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
27715 if (tw->tw_family == PF_INET6) {
27716 struct ipv6_pinfo *np = inet6_sk(sk);
27717 diff -NurpP --minimal linux-2.6.32.1/net/ipv4/udp.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/udp.c
27718 --- linux-2.6.32.1/net/ipv4/udp.c 2009-12-03 20:03:00.000000000 +0100
27719 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv4/udp.c 2009-12-03 20:04:56.000000000 +0100
27720 @@ -224,14 +224,7 @@ fail:
27722 EXPORT_SYMBOL(udp_lib_get_port);
27724 -static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
27726 - struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
27728 - return (!ipv6_only_sock(sk2) &&
27729 - (!inet1->rcv_saddr || !inet2->rcv_saddr ||
27730 - inet1->rcv_saddr == inet2->rcv_saddr));
27732 +extern int ipv4_rcv_saddr_equal(const struct sock *, const struct sock *);
27734 int udp_v4_get_port(struct sock *sk, unsigned short snum)
27736 @@ -253,6 +246,11 @@ static inline int compute_score(struct s
27737 if (inet->rcv_saddr != daddr)
27741 + /* block non nx_info ips */
27742 + if (!v4_addr_in_nx_info(sk->sk_nx_info,
27743 + daddr, NXA_MASK_BIND))
27747 if (inet->daddr != saddr)
27748 @@ -273,6 +271,7 @@ static inline int compute_score(struct s
27753 /* UDP is nearly always wildcards out the wazoo, it makes no sense to try
27754 * harder than this. -DaveM
27756 @@ -294,6 +293,11 @@ begin:
27757 sk_nulls_for_each_rcu(sk, node, &hslot->head) {
27758 score = compute_score(sk, net, saddr, hnum, sport,
27759 daddr, dport, dif);
27760 + /* FIXME: disabled?
27761 + if (score == 9) {
27765 if (score > badness) {
27768 @@ -307,6 +311,7 @@ begin:
27769 if (get_nulls_value(node) != hash)
27774 if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
27776 @@ -316,6 +321,7 @@ begin:
27784 @@ -358,7 +364,7 @@ static inline struct sock *udp_v4_mcast_
27785 s->sk_hash != hnum ||
27786 (inet->daddr && inet->daddr != rmt_addr) ||
27787 (inet->dport != rmt_port && inet->dport) ||
27788 - (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
27789 + !v4_sock_addr_match(sk->sk_nx_info, inet, loc_addr) ||
27790 ipv6_only_sock(s) ||
27791 (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
27793 @@ -707,8 +713,13 @@ int udp_sendmsg(struct kiocb *iocb, stru
27794 { .sport = inet->sport,
27795 .dport = dport } } };
27796 struct net *net = sock_net(sk);
27797 + struct nx_info *nxi = sk->sk_nx_info;
27799 security_sk_classify_flow(sk, &fl);
27800 + err = ip_v4_find_src(net, nxi, &rt, &fl);
27804 err = ip_route_output_flow(net, &rt, &fl, sk, 1);
27806 if (err == -ENETUNREACH)
27807 @@ -988,7 +999,8 @@ try_again:
27809 sin->sin_family = AF_INET;
27810 sin->sin_port = udp_hdr(skb)->source;
27811 - sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
27812 + sin->sin_addr.s_addr = nx_map_sock_lback(
27813 + skb->sk->sk_nx_info, ip_hdr(skb)->saddr);
27814 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
27816 if (inet->cmsg_flags)
27817 @@ -1627,6 +1639,8 @@ static struct sock *udp_get_first(struct
27818 sk_nulls_for_each(sk, node, &hslot->head) {
27819 if (!net_eq(sock_net(sk), net))
27821 + if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27823 if (sk->sk_family == state->family)
27826 @@ -1644,7 +1658,9 @@ static struct sock *udp_get_next(struct
27829 sk = sk_nulls_next(sk);
27830 - } while (sk && (!net_eq(sock_net(sk), net) || sk->sk_family != state->family));
27831 + } while (sk && (!net_eq(sock_net(sk), net) ||
27832 + sk->sk_family != state->family ||
27833 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
27836 if (state->bucket < UDP_HTABLE_SIZE)
27837 @@ -1751,7 +1767,10 @@ static void udp4_format_sock(struct sock
27839 seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27840 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d%n",
27841 - bucket, src, srcp, dest, destp, sp->sk_state,
27843 + nx_map_sock_lback(current_nx_info(), src), srcp,
27844 + nx_map_sock_lback(current_nx_info(), dest), destp,
27846 sk_wmem_alloc_get(sp),
27847 sk_rmem_alloc_get(sp),
27848 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
27849 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/addrconf.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/addrconf.c
27850 --- linux-2.6.32.1/net/ipv6/addrconf.c 2009-12-03 20:03:00.000000000 +0100
27851 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/addrconf.c 2009-12-03 20:04:56.000000000 +0100
27854 #include <linux/proc_fs.h>
27855 #include <linux/seq_file.h>
27856 +#include <linux/vs_network.h>
27857 +#include <linux/vs_inet6.h>
27859 /* Set to 3 to get tracing... */
27860 #define ACONF_DEBUG 2
27861 @@ -1119,7 +1121,7 @@ out:
27863 int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
27864 const struct in6_addr *daddr, unsigned int prefs,
27865 - struct in6_addr *saddr)
27866 + struct in6_addr *saddr, struct nx_info *nxi)
27868 struct ipv6_saddr_score scores[2],
27869 *score = &scores[0], *hiscore = &scores[1];
27870 @@ -1192,6 +1194,8 @@ int ipv6_dev_get_saddr(struct net *net,
27874 + if (!v6_addr_in_nx_info(nxi, &score->ifa->addr, -1))
27878 bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
27879 @@ -3000,7 +3004,10 @@ static void if6_seq_stop(struct seq_file
27880 static int if6_seq_show(struct seq_file *seq, void *v)
27882 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
27883 - seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27885 + if (nx_check(0, VS_ADMIN|VS_WATCH) ||
27886 + v6_addr_in_nx_info(current_nx_info(), &ifp->addr, -1))
27887 + seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27889 ifp->idev->dev->ifindex,
27891 @@ -3497,6 +3504,12 @@ static int inet6_dump_addr(struct sk_buf
27892 struct ifmcaddr6 *ifmca;
27893 struct ifacaddr6 *ifaca;
27894 struct net *net = sock_net(skb->sk);
27895 + struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27897 + /* disable ipv6 on non v6 guests */
27898 + if (nxi && !nx_info_has_v6(nxi))
27902 s_idx = cb->args[0];
27903 s_ip_idx = ip_idx = cb->args[1];
27904 @@ -3518,6 +3531,8 @@ static int inet6_dump_addr(struct sk_buf
27905 ifa = ifa->if_next, ip_idx++) {
27906 if (ip_idx < s_ip_idx)
27908 + if (!v6_addr_in_nx_info(nxi, &ifa->addr, -1))
27910 err = inet6_fill_ifaddr(skb, ifa,
27911 NETLINK_CB(cb->skb).pid,
27912 cb->nlh->nlmsg_seq,
27913 @@ -3531,6 +3546,8 @@ static int inet6_dump_addr(struct sk_buf
27914 ifmca = ifmca->next, ip_idx++) {
27915 if (ip_idx < s_ip_idx)
27917 + if (!v6_addr_in_nx_info(nxi, &ifmca->mca_addr, -1))
27919 err = inet6_fill_ifmcaddr(skb, ifmca,
27920 NETLINK_CB(cb->skb).pid,
27921 cb->nlh->nlmsg_seq,
27922 @@ -3544,6 +3561,8 @@ static int inet6_dump_addr(struct sk_buf
27923 ifaca = ifaca->aca_next, ip_idx++) {
27924 if (ip_idx < s_ip_idx)
27926 + if (!v6_addr_in_nx_info(nxi, &ifaca->aca_addr, -1))
27928 err = inet6_fill_ifacaddr(skb, ifaca,
27929 NETLINK_CB(cb->skb).pid,
27930 cb->nlh->nlmsg_seq,
27931 @@ -3830,12 +3849,19 @@ static int inet6_dump_ifinfo(struct sk_b
27932 int s_idx = cb->args[0];
27933 struct net_device *dev;
27934 struct inet6_dev *idev;
27935 + struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27937 + /* FIXME: maybe disable ipv6 on non v6 guests?
27938 + if (skb->sk && skb->sk->sk_vx_info)
27939 + return skb->len; */
27941 read_lock(&dev_base_lock);
27943 for_each_netdev(net, dev) {
27946 + if (!v6_dev_in_nx_info(dev, nxi))
27948 if ((idev = in6_dev_get(dev)) == NULL)
27950 err = inet6_fill_ifinfo(skb, idev, NETLINK_CB(cb->skb).pid,
27951 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/af_inet6.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/af_inet6.c
27952 --- linux-2.6.32.1/net/ipv6/af_inet6.c 2009-12-03 20:03:00.000000000 +0100
27953 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/af_inet6.c 2009-12-03 20:04:56.000000000 +0100
27955 #include <linux/netdevice.h>
27956 #include <linux/icmpv6.h>
27957 #include <linux/netfilter_ipv6.h>
27958 +#include <linux/vs_inet.h>
27959 +#include <linux/vs_inet6.h>
27961 #include <net/ip.h>
27962 #include <net/ipv6.h>
27963 @@ -158,9 +160,12 @@ lookup_protocol:
27967 + if ((protocol == IPPROTO_ICMPV6) &&
27968 + nx_capable(answer->capability, NXC_RAW_ICMP))
27970 if (answer->capability > 0 && !capable(answer->capability))
27971 goto out_rcu_unlock;
27974 sock->ops = answer->ops;
27975 answer_prot = answer->prot;
27976 answer_no_check = answer->no_check;
27977 @@ -259,6 +264,7 @@ int inet6_bind(struct socket *sock, stru
27978 struct inet_sock *inet = inet_sk(sk);
27979 struct ipv6_pinfo *np = inet6_sk(sk);
27980 struct net *net = sock_net(sk);
27981 + struct nx_v6_sock_addr nsa;
27983 unsigned short snum;
27985 @@ -270,6 +276,11 @@ int inet6_bind(struct socket *sock, stru
27987 if (addr_len < SIN6_LEN_RFC2133)
27990 + err = v6_map_sock_addr(inet, addr, &nsa);
27994 addr_type = ipv6_addr_type(&addr->sin6_addr);
27995 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
27997 @@ -301,6 +312,7 @@ int inet6_bind(struct socket *sock, stru
27998 /* Reproduce AF_INET checks to make the bindings consitant */
27999 v4addr = addr->sin6_addr.s6_addr32[3];
28000 chk_addr_ret = inet_addr_type(net, v4addr);
28002 if (!sysctl_ip_nonlocal_bind &&
28003 !(inet->freebind || inet->transparent) &&
28004 v4addr != htonl(INADDR_ANY) &&
28005 @@ -310,6 +322,10 @@ int inet6_bind(struct socket *sock, stru
28006 err = -EADDRNOTAVAIL;
28009 + if (!v4_addr_in_nx_info(sk->sk_nx_info, v4addr, NXA_MASK_BIND)) {
28010 + err = -EADDRNOTAVAIL;
28014 if (addr_type != IPV6_ADDR_ANY) {
28015 struct net_device *dev = NULL;
28016 @@ -335,6 +351,11 @@ int inet6_bind(struct socket *sock, stru
28020 + if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
28021 + err = -EADDRNOTAVAIL;
28025 /* ipv4 addr of the socket is invalid. Only the
28026 * unspecified and mapped address have a v4 equivalent.
28028 @@ -353,6 +374,8 @@ int inet6_bind(struct socket *sock, stru
28032 + v6_set_sock_addr(inet, &nsa);
28034 inet->rcv_saddr = v4addr;
28035 inet->saddr = v4addr;
28037 @@ -448,9 +471,11 @@ int inet6_getname(struct socket *sock, s
28039 sin->sin6_port = inet->dport;
28040 ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
28041 + /* FIXME: remap lback? */
28043 sin->sin6_flowinfo = np->flow_label;
28045 + /* FIXME: remap lback? */
28046 if (ipv6_addr_any(&np->rcv_saddr))
28047 ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
28049 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/fib6_rules.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/fib6_rules.c
28050 --- linux-2.6.32.1/net/ipv6/fib6_rules.c 2009-09-10 15:26:30.000000000 +0200
28051 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/fib6_rules.c 2009-12-03 20:04:56.000000000 +0100
28052 @@ -96,7 +96,7 @@ static int fib6_rule_action(struct fib_r
28053 if (ipv6_dev_get_saddr(net,
28054 ip6_dst_idev(&rt->u.dst)->dev,
28055 &flp->fl6_dst, srcprefs,
28059 if (!ipv6_prefix_equal(&saddr, &r->src.addr,
28061 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/inet6_hashtables.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/inet6_hashtables.c
28062 --- linux-2.6.32.1/net/ipv6/inet6_hashtables.c 2009-03-24 14:22:46.000000000 +0100
28063 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/inet6_hashtables.c 2009-12-03 20:04:56.000000000 +0100
28066 #include <linux/module.h>
28067 #include <linux/random.h>
28068 +#include <linux/vs_inet6.h>
28070 #include <net/inet_connection_sock.h>
28071 #include <net/inet_hashtables.h>
28072 @@ -76,7 +77,6 @@ struct sock *__inet6_lookup_established(
28073 unsigned int slot = hash & (hashinfo->ehash_size - 1);
28074 struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
28079 sk_nulls_for_each_rcu(sk, node, &head->chain) {
28080 @@ -88,7 +88,7 @@ begin:
28088 if (get_nulls_value(node) != slot)
28089 @@ -134,6 +134,9 @@ static int inline compute_score(struct s
28090 if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
28094 + if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
28097 if (sk->sk_bound_dev_if) {
28098 if (sk->sk_bound_dev_if != dif)
28099 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/ip6_output.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/ip6_output.c
28100 --- linux-2.6.32.1/net/ipv6/ip6_output.c 2009-12-03 20:03:00.000000000 +0100
28101 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/ip6_output.c 2009-12-03 20:04:56.000000000 +0100
28102 @@ -934,7 +934,7 @@ static int ip6_dst_lookup_tail(struct so
28103 err = ipv6_dev_get_saddr(net, ip6_dst_idev(*dst)->dev,
28105 sk ? inet6_sk(sk)->srcprefs : 0,
28107 + &fl->fl6_src, sk->sk_nx_info);
28109 goto out_err_release;
28111 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/Kconfig linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/Kconfig
28112 --- linux-2.6.32.1/net/ipv6/Kconfig 2009-09-10 15:26:30.000000000 +0200
28113 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/Kconfig 2009-12-03 20:04:56.000000000 +0100
28116 # IPv6 as module will cause a CRASH if you try to unload it
28118 - tristate "The IPv6 protocol"
28120 + bool "The IPv6 protocol"
28123 This is complemental support for the IP version 6.
28124 You will still be able to do traditional IPv4 networking as well.
28125 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/ndisc.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/ndisc.c
28126 --- linux-2.6.32.1/net/ipv6/ndisc.c 2009-12-03 20:03:00.000000000 +0100
28127 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/ndisc.c 2009-12-03 20:04:56.000000000 +0100
28128 @@ -589,7 +589,7 @@ static void ndisc_send_na(struct net_dev
28130 if (ipv6_dev_get_saddr(dev_net(dev), dev, daddr,
28131 inet6_sk(dev_net(dev)->ipv6.ndisc_sk)->srcprefs,
28133 + &tmpaddr, NULL /* FIXME: ? */ ))
28135 src_addr = &tmpaddr;
28137 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/raw.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/raw.c
28138 --- linux-2.6.32.1/net/ipv6/raw.c 2009-12-03 20:03:00.000000000 +0100
28139 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/raw.c 2009-12-03 20:04:56.000000000 +0100
28141 #include <linux/icmpv6.h>
28142 #include <linux/netfilter.h>
28143 #include <linux/netfilter_ipv6.h>
28144 +#include <linux/vs_inet6.h>
28145 #include <linux/skbuff.h>
28146 #include <asm/uaccess.h>
28147 #include <asm/ioctls.h>
28148 @@ -281,6 +282,13 @@ static int rawv6_bind(struct sock *sk, s
28152 + if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
28153 + err = -EADDRNOTAVAIL;
28159 /* ipv4 addr of the socket is invalid. Only the
28160 * unspecified and mapped address have a v4 equivalent.
28162 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/route.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/route.c
28163 --- linux-2.6.32.1/net/ipv6/route.c 2009-12-03 20:03:00.000000000 +0100
28164 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/route.c 2009-12-03 20:04:56.000000000 +0100
28165 @@ -2257,7 +2257,8 @@ static int rt6_fill_node(struct net *net
28166 struct inet6_dev *idev = ip6_dst_idev(&rt->u.dst);
28167 struct in6_addr saddr_buf;
28168 if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
28169 - dst, 0, &saddr_buf) == 0)
28170 + dst, 0, &saddr_buf,
28171 + (skb->sk ? skb->sk->sk_nx_info : NULL)) == 0)
28172 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
28175 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/tcp_ipv6.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/tcp_ipv6.c
28176 --- linux-2.6.32.1/net/ipv6/tcp_ipv6.c 2009-12-03 20:03:00.000000000 +0100
28177 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/tcp_ipv6.c 2009-12-03 20:04:56.000000000 +0100
28180 #include <linux/crypto.h>
28181 #include <linux/scatterlist.h>
28182 +#include <linux/vs_inet6.h>
28184 static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
28185 static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
28186 @@ -156,8 +157,15 @@ static int tcp_v6_connect(struct sock *s
28187 * connect() to INADDR_ANY means loopback (BSD'ism).
28190 - if(ipv6_addr_any(&usin->sin6_addr))
28191 - usin->sin6_addr.s6_addr[15] = 0x1;
28192 + if(ipv6_addr_any(&usin->sin6_addr)) {
28193 + struct nx_info *nxi = sk->sk_nx_info;
28195 + if (nxi && nx_info_has_v6(nxi))
28196 + /* FIXME: remap lback? */
28197 + usin->sin6_addr = nxi->v6.ip;
28199 + usin->sin6_addr.s6_addr[15] = 0x1;
28202 addr_type = ipv6_addr_type(&usin->sin6_addr);
28204 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/udp.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/udp.c
28205 --- linux-2.6.32.1/net/ipv6/udp.c 2009-12-03 20:03:00.000000000 +0100
28206 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/udp.c 2009-12-03 20:04:56.000000000 +0100
28209 #include <linux/proc_fs.h>
28210 #include <linux/seq_file.h>
28211 +#include <linux/vs_inet6.h>
28212 #include "udp_impl.h"
28214 int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
28215 @@ -61,24 +62,49 @@ int ipv6_rcv_saddr_equal(const struct so
28216 int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
28218 /* if both are mapped, treat as IPv4 */
28219 - if (addr_type == IPV6_ADDR_MAPPED && addr_type2 == IPV6_ADDR_MAPPED)
28220 - return (!sk2_ipv6only &&
28221 + if (addr_type == IPV6_ADDR_MAPPED && addr_type2 == IPV6_ADDR_MAPPED) {
28222 + if (!sk2_ipv6only &&
28223 (!sk_rcv_saddr || !sk2_rcv_saddr ||
28224 - sk_rcv_saddr == sk2_rcv_saddr));
28225 + sk_rcv_saddr == sk2_rcv_saddr))
28231 if (addr_type2 == IPV6_ADDR_ANY &&
28232 !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
28236 if (addr_type == IPV6_ADDR_ANY &&
28237 !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
28241 if (sk2_rcv_saddr6 &&
28242 ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
28249 + if (!sk_rcv_saddr && !sk2_rcv_saddr)
28250 + return nx_v4_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
28251 + if (!sk2_rcv_saddr)
28252 + return v4_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr, -1);
28253 + if (!sk_rcv_saddr)
28254 + return v4_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr, -1);
28257 + if (addr_type2 == IPV6_ADDR_ANY && addr_type == IPV6_ADDR_ANY)
28258 + return nx_v6_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
28259 + else if (addr_type2 == IPV6_ADDR_ANY)
28260 + return v6_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr6, -1);
28261 + else if (addr_type == IPV6_ADDR_ANY) {
28262 + if (addr_type2 == IPV6_ADDR_MAPPED)
28263 + return nx_v4_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
28265 + return v6_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr6, -1);
28270 int udp_v6_get_port(struct sock *sk, unsigned short snum)
28271 @@ -109,6 +135,10 @@ static inline int compute_score(struct s
28272 if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
28276 + /* block non nx_info ips */
28277 + if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
28280 if (!ipv6_addr_any(&np->daddr)) {
28281 if (!ipv6_addr_equal(&np->daddr, saddr))
28282 diff -NurpP --minimal linux-2.6.32.1/net/ipv6/xfrm6_policy.c linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/xfrm6_policy.c
28283 --- linux-2.6.32.1/net/ipv6/xfrm6_policy.c 2009-12-03 20:03:00.000000000 +0100
28284 +++ linux-2.6.32.1-vs2.3.0.36.27/net/ipv6/xfrm6_policy.c 2009-12-03 20:04:56.000000000 +0100
28285 @@ -63,7 +63,7 @@ static int xfrm6_get_saddr(struct net *n
28286 dev = ip6_dst_idev(dst)->dev;
28287 ipv6_dev_get_saddr(dev_net(dev), dev,
28288 (struct in6_addr *)&daddr->a6, 0,
28289 - (struct in6_addr *)&saddr->a6);
28290 + (struct in6_addr *)&saddr->a6, NULL);
28294 diff -NurpP --minimal linux-2.6.32.1/net/netlink/af_netlink.c linux-2.6.32.1-vs2.3.0.36.27/net/netlink/af_netlink.c
28295 --- linux-2.6.32.1/net/netlink/af_netlink.c 2009-12-03 20:03:01.000000000 +0100
28296 +++ linux-2.6.32.1-vs2.3.0.36.27/net/netlink/af_netlink.c 2009-12-03 20:04:56.000000000 +0100
28298 #include <linux/types.h>
28299 #include <linux/audit.h>
28300 #include <linux/mutex.h>
28301 +#include <linux/vs_context.h>
28302 +#include <linux/vs_network.h>
28303 +#include <linux/vs_limit.h>
28305 #include <net/net_namespace.h>
28306 #include <net/sock.h>
28307 @@ -1899,6 +1902,8 @@ static struct sock *netlink_seq_socket_i
28308 sk_for_each(s, node, &hash->table[j]) {
28309 if (sock_net(s) != seq_file_net(seq))
28311 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
28315 iter->hash_idx = j;
28316 @@ -1933,7 +1938,8 @@ static void *netlink_seq_next(struct seq
28320 - } while (s && sock_net(s) != seq_file_net(seq));
28321 + } while (s && (sock_net(s) != seq_file_net(seq) ||
28322 + !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)));
28326 @@ -1945,7 +1951,8 @@ static void *netlink_seq_next(struct seq
28328 for (; j <= hash->mask; j++) {
28329 s = sk_head(&hash->table[j]);
28330 - while (s && sock_net(s) != seq_file_net(seq))
28331 + while (s && (sock_net(s) != seq_file_net(seq) ||
28332 + !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)))
28336 diff -NurpP --minimal linux-2.6.32.1/net/sctp/ipv6.c linux-2.6.32.1-vs2.3.0.36.27/net/sctp/ipv6.c
28337 --- linux-2.6.32.1/net/sctp/ipv6.c 2009-12-03 20:03:01.000000000 +0100
28338 +++ linux-2.6.32.1-vs2.3.0.36.27/net/sctp/ipv6.c 2009-12-03 20:04:56.000000000 +0100
28339 @@ -316,7 +316,8 @@ static void sctp_v6_get_saddr(struct sct
28340 dst ? ip6_dst_idev(dst)->dev : NULL,
28341 &daddr->v6.sin6_addr,
28342 inet6_sk(&sk->inet.sk)->srcprefs,
28343 - &saddr->v6.sin6_addr);
28344 + &saddr->v6.sin6_addr,
28345 + asoc->base.sk->sk_nx_info);
28346 SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: %pI6\n",
28347 &saddr->v6.sin6_addr);
28349 diff -NurpP --minimal linux-2.6.32.1/net/socket.c linux-2.6.32.1-vs2.3.0.36.27/net/socket.c
28350 --- linux-2.6.32.1/net/socket.c 2009-12-03 20:03:01.000000000 +0100
28351 +++ linux-2.6.32.1-vs2.3.0.36.27/net/socket.c 2009-12-03 20:04:56.000000000 +0100
28354 #include <net/sock.h>
28355 #include <linux/netfilter.h>
28356 +#include <linux/vs_base.h>
28357 +#include <linux/vs_socket.h>
28358 +#include <linux/vs_inet.h>
28359 +#include <linux/vs_inet6.h>
28361 static int sock_no_open(struct inode *irrelevant, struct file *dontcare);
28362 static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
28363 @@ -559,7 +563,7 @@ static inline int __sock_sendmsg(struct
28364 struct msghdr *msg, size_t size)
28366 struct sock_iocb *si = kiocb_to_siocb(iocb);
28372 @@ -570,7 +574,22 @@ static inline int __sock_sendmsg(struct
28376 - return sock->ops->sendmsg(iocb, sock, msg, size);
28377 + len = sock->ops->sendmsg(iocb, sock, msg, size);
28380 + vx_sock_send(sock->sk, size);
28382 + vx_sock_fail(sock->sk, size);
28384 + vxdprintk(VXD_CBIT(net, 7),
28385 + "__sock_sendmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
28387 + (sock->sk)?sock->sk->sk_nx_info:0,
28388 + (sock->sk)?sock->sk->sk_vx_info:0,
28389 + (sock->sk)?sock->sk->sk_xid:0,
28390 + (sock->sk)?sock->sk->sk_nid:0,
28391 + (unsigned int)size, len);
28395 int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
28396 @@ -671,7 +690,7 @@ EXPORT_SYMBOL_GPL(__sock_recv_timestamp)
28397 static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock,
28398 struct msghdr *msg, size_t size, int flags)
28402 struct sock_iocb *si = kiocb_to_siocb(iocb);
28405 @@ -684,7 +703,18 @@ static inline int __sock_recvmsg(struct
28409 - return sock->ops->recvmsg(iocb, sock, msg, size, flags);
28410 + len = sock->ops->recvmsg(iocb, sock, msg, size, flags);
28411 + if ((len >= 0) && sock->sk)
28412 + vx_sock_recv(sock->sk, len);
28413 + vxdprintk(VXD_CBIT(net, 7),
28414 + "__sock_recvmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
28416 + (sock->sk)?sock->sk->sk_nx_info:0,
28417 + (sock->sk)?sock->sk->sk_vx_info:0,
28418 + (sock->sk)?sock->sk->sk_xid:0,
28419 + (sock->sk)?sock->sk->sk_nid:0,
28420 + (unsigned int)size, len);
28424 int sock_recvmsg(struct socket *sock, struct msghdr *msg,
28425 @@ -1155,6 +1185,13 @@ static int __sock_create(struct net *net
28426 if (type < 0 || type >= SOCK_MAX)
28429 + if (!nx_check(0, VS_ADMIN)) {
28430 + if (family == PF_INET && !current_nx_info_has_v4())
28431 + return -EAFNOSUPPORT;
28432 + if (family == PF_INET6 && !current_nx_info_has_v6())
28433 + return -EAFNOSUPPORT;
28438 This uglymoron is moved from INET layer to here to avoid
28439 @@ -1287,6 +1324,7 @@ SYSCALL_DEFINE3(socket, int, family, int
28443 + set_bit(SOCK_USER_SOCKET, &sock->flags);
28444 retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
28447 @@ -1328,10 +1366,12 @@ SYSCALL_DEFINE4(socketpair, int, family,
28448 err = sock_create(family, type, protocol, &sock1);
28451 + set_bit(SOCK_USER_SOCKET, &sock1->flags);
28453 err = sock_create(family, type, protocol, &sock2);
28455 goto out_release_1;
28456 + set_bit(SOCK_USER_SOCKET, &sock2->flags);
28458 err = sock1->ops->socketpair(sock1, sock2);
28460 diff -NurpP --minimal linux-2.6.32.1/net/sunrpc/auth.c linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/auth.c
28461 --- linux-2.6.32.1/net/sunrpc/auth.c 2009-12-03 20:03:01.000000000 +0100
28462 +++ linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/auth.c 2009-12-03 20:04:56.000000000 +0100
28464 #include <linux/hash.h>
28465 #include <linux/sunrpc/clnt.h>
28466 #include <linux/spinlock.h>
28467 +#include <linux/vs_tag.h>
28470 # define RPCDBG_FACILITY RPCDBG_AUTH
28471 @@ -360,6 +361,7 @@ rpcauth_lookupcred(struct rpc_auth *auth
28472 memset(&acred, 0, sizeof(acred));
28473 acred.uid = cred->fsuid;
28474 acred.gid = cred->fsgid;
28475 + acred.tag = dx_current_tag();
28476 acred.group_info = get_group_info(((struct cred *)cred)->group_info);
28478 ret = auth->au_ops->lookup_cred(auth, &acred, flags);
28479 @@ -400,6 +402,7 @@ rpcauth_bind_root_cred(struct rpc_task *
28480 struct auth_cred acred = {
28483 + .tag = dx_current_tag(),
28485 struct rpc_cred *ret;
28487 diff -NurpP --minimal linux-2.6.32.1/net/sunrpc/auth_unix.c linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/auth_unix.c
28488 --- linux-2.6.32.1/net/sunrpc/auth_unix.c 2008-12-25 00:26:37.000000000 +0100
28489 +++ linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/auth_unix.c 2009-12-03 20:04:56.000000000 +0100
28490 @@ -11,12 +11,14 @@
28491 #include <linux/module.h>
28492 #include <linux/sunrpc/clnt.h>
28493 #include <linux/sunrpc/auth.h>
28494 +#include <linux/vs_tag.h>
28496 #define NFS_NGROUPS 16
28499 struct rpc_cred uc_base;
28502 gid_t uc_gids[NFS_NGROUPS];
28504 #define uc_uid uc_base.cr_uid
28505 @@ -78,6 +80,7 @@ unx_create_cred(struct rpc_auth *auth, s
28506 groups = NFS_NGROUPS;
28508 cred->uc_gid = acred->gid;
28509 + cred->uc_tag = acred->tag;
28510 for (i = 0; i < groups; i++)
28511 cred->uc_gids[i] = GROUP_AT(acred->group_info, i);
28512 if (i < NFS_NGROUPS)
28513 @@ -119,7 +122,9 @@ unx_match(struct auth_cred *acred, struc
28517 - if (cred->uc_uid != acred->uid || cred->uc_gid != acred->gid)
28518 + if (cred->uc_uid != acred->uid ||
28519 + cred->uc_gid != acred->gid ||
28520 + cred->uc_tag != acred->tag)
28523 if (acred->group_info != NULL)
28524 @@ -142,7 +147,7 @@ unx_marshal(struct rpc_task *task, __be3
28525 struct rpc_clnt *clnt = task->tk_client;
28526 struct unx_cred *cred = container_of(task->tk_msg.rpc_cred, struct unx_cred, uc_base);
28527 __be32 *base, *hold;
28531 *p++ = htonl(RPC_AUTH_UNIX);
28533 @@ -152,9 +157,12 @@ unx_marshal(struct rpc_task *task, __be3
28534 * Copy the UTS nodename captured when the client was created.
28536 p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen);
28537 + tag = task->tk_client->cl_tag;
28539 - *p++ = htonl((u32) cred->uc_uid);
28540 - *p++ = htonl((u32) cred->uc_gid);
28541 + *p++ = htonl((u32) TAGINO_UID(tag,
28542 + cred->uc_uid, cred->uc_tag));
28543 + *p++ = htonl((u32) TAGINO_GID(tag,
28544 + cred->uc_gid, cred->uc_tag));
28546 for (i = 0; i < 16 && cred->uc_gids[i] != (gid_t) NOGROUP; i++)
28547 *p++ = htonl((u32) cred->uc_gids[i]);
28548 diff -NurpP --minimal linux-2.6.32.1/net/sunrpc/clnt.c linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/clnt.c
28549 --- linux-2.6.32.1/net/sunrpc/clnt.c 2009-12-03 20:03:01.000000000 +0100
28550 +++ linux-2.6.32.1-vs2.3.0.36.27/net/sunrpc/clnt.c 2009-12-03 20:04:56.000000000 +0100
28552 #include <linux/utsname.h>
28553 #include <linux/workqueue.h>
28554 #include <linux/in6.h>
28555 +#include <linux/vs_cvirt.h>
28557 #include <linux/sunrpc/clnt.h>
28558 #include <linux/sunrpc/rpc_pipe_fs.h>
28559 @@ -358,6 +359,9 @@ struct rpc_clnt *rpc_create(struct rpc_c
28560 if (!(args->flags & RPC_CLNT_CREATE_QUIET))
28561 clnt->cl_chatty = 1;
28563 + /* TODO: handle RPC_CLNT_CREATE_TAGGED
28564 + if (args->flags & RPC_CLNT_CREATE_TAGGED)
28565 + clnt->cl_tag = 1; */
28568 EXPORT_SYMBOL_GPL(rpc_create);
28569 diff -NurpP --minimal linux-2.6.32.1/net/unix/af_unix.c linux-2.6.32.1-vs2.3.0.36.27/net/unix/af_unix.c
28570 --- linux-2.6.32.1/net/unix/af_unix.c 2009-12-03 20:03:01.000000000 +0100
28571 +++ linux-2.6.32.1-vs2.3.0.36.27/net/unix/af_unix.c 2009-12-03 20:04:56.000000000 +0100
28572 @@ -114,6 +114,8 @@
28573 #include <linux/mount.h>
28574 #include <net/checksum.h>
28575 #include <linux/security.h>
28576 +#include <linux/vs_context.h>
28577 +#include <linux/vs_limit.h>
28579 static struct hlist_head unix_socket_table[UNIX_HASH_SIZE + 1];
28580 static DEFINE_SPINLOCK(unix_table_lock);
28581 @@ -258,6 +260,8 @@ static struct sock *__unix_find_socket_b
28582 if (!net_eq(sock_net(s), net))
28585 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
28587 if (u->addr->len == len &&
28588 !memcmp(u->addr->name, sunname, len))
28590 @@ -2114,6 +2118,8 @@ static struct sock *unix_seq_idx(struct
28591 for (s = first_unix_socket(&iter->i); s; s = next_unix_socket(&iter->i, s)) {
28592 if (sock_net(s) != seq_file_net(seq))
28594 + if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
28599 @@ -2138,7 +2144,8 @@ static void *unix_seq_next(struct seq_fi
28600 sk = first_unix_socket(&iter->i);
28602 sk = next_unix_socket(&iter->i, sk);
28603 - while (sk && (sock_net(sk) != seq_file_net(seq)))
28604 + while (sk && (sock_net(sk) != seq_file_net(seq) ||
28605 + !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)))
28606 sk = next_unix_socket(&iter->i, sk);
28609 diff -NurpP --minimal linux-2.6.32.1/net/x25/af_x25.c linux-2.6.32.1-vs2.3.0.36.27/net/x25/af_x25.c
28610 --- linux-2.6.32.1/net/x25/af_x25.c 2009-12-03 20:03:01.000000000 +0100
28611 +++ linux-2.6.32.1-vs2.3.0.36.27/net/x25/af_x25.c 2009-12-03 20:04:56.000000000 +0100
28612 @@ -519,7 +519,10 @@ static int x25_create(struct net *net, s
28616 - sock_init_data(sock, sk);
28617 + sk->sk_socket = sock;
28618 + sk->sk_type = sock->type;
28619 + sk->sk_sleep = &sock->wait;
28622 x25_init_timers(sk);
28624 diff -NurpP --minimal linux-2.6.32.1/scripts/checksyscalls.sh linux-2.6.32.1-vs2.3.0.36.27/scripts/checksyscalls.sh
28625 --- linux-2.6.32.1/scripts/checksyscalls.sh 2009-09-10 15:26:31.000000000 +0200
28626 +++ linux-2.6.32.1-vs2.3.0.36.27/scripts/checksyscalls.sh 2009-12-03 20:04:56.000000000 +0100
28627 @@ -194,7 +194,6 @@ cat << EOF
28628 #define __IGNORE_afs_syscall
28629 #define __IGNORE_getpmsg
28630 #define __IGNORE_putpmsg
28631 -#define __IGNORE_vserver
28635 diff -NurpP --minimal linux-2.6.32.1/security/commoncap.c linux-2.6.32.1-vs2.3.0.36.27/security/commoncap.c
28636 --- linux-2.6.32.1/security/commoncap.c 2009-12-03 20:03:02.000000000 +0100
28637 +++ linux-2.6.32.1-vs2.3.0.36.27/security/commoncap.c 2009-12-03 20:04:56.000000000 +0100
28639 #include <linux/sched.h>
28640 #include <linux/prctl.h>
28641 #include <linux/securebits.h>
28642 +#include <linux/vs_context.h>
28645 * If a non-root user executes a setuid-root binary in
28646 @@ -52,7 +53,7 @@ static void warn_setuid_and_fcaps_mixed(
28648 int cap_netlink_send(struct sock *sk, struct sk_buff *skb)
28650 - NETLINK_CB(skb).eff_cap = current_cap();
28651 + NETLINK_CB(skb).eff_cap = vx_mbcaps(current_cap());
28655 @@ -62,6 +63,7 @@ int cap_netlink_recv(struct sk_buff *skb
28660 EXPORT_SYMBOL(cap_netlink_recv);
28663 @@ -82,7 +84,22 @@ EXPORT_SYMBOL(cap_netlink_recv);
28664 int cap_capable(struct task_struct *tsk, const struct cred *cred, int cap,
28667 - return cap_raised(cred->cap_effective, cap) ? 0 : -EPERM;
28668 + struct vx_info *vxi = tsk->vx_info;
28671 + printk("cap_capable() VXF_STATE_SETUP = %llx, raised = %x, eff = %08x:%08x\n",
28672 + vx_info_flags(vxi, VXF_STATE_SETUP, 0),
28673 + cap_raised(tsk->cap_effective, cap),
28674 + tsk->cap_effective.cap[1], tsk->cap_effective.cap[0]);
28677 + /* special case SETUP */
28678 + if (vx_info_flags(vxi, VXF_STATE_SETUP, 0) &&
28679 + /* FIXME: maybe use cred instead? */
28680 + cap_raised(tsk->cred->cap_effective, cap))
28683 + return vx_cap_raised(vxi, cred->cap_effective, cap) ? 0 : -EPERM;
28687 @@ -618,7 +635,7 @@ int cap_inode_setxattr(struct dentry *de
28689 if (!strncmp(name, XATTR_SECURITY_PREFIX,
28690 sizeof(XATTR_SECURITY_PREFIX) - 1) &&
28691 - !capable(CAP_SYS_ADMIN))
28692 + !vx_capable(CAP_SYS_ADMIN, VXC_FS_SECURITY))
28696 @@ -962,7 +979,8 @@ error:
28698 int cap_syslog(int type)
28700 - if ((type != 3 && type != 10) && !capable(CAP_SYS_ADMIN))
28701 + if ((type != 3 && type != 10) &&
28702 + !vx_capable(CAP_SYS_ADMIN, VXC_SYSLOG))
28706 @@ -1014,3 +1032,4 @@ int cap_file_mmap(struct file *file, uns
28711 diff -NurpP --minimal linux-2.6.32.1/security/selinux/hooks.c linux-2.6.32.1-vs2.3.0.36.27/security/selinux/hooks.c
28712 --- linux-2.6.32.1/security/selinux/hooks.c 2009-12-03 20:03:02.000000000 +0100
28713 +++ linux-2.6.32.1-vs2.3.0.36.27/security/selinux/hooks.c 2009-12-03 20:04:56.000000000 +0100
28715 #include <linux/dccp.h>
28716 #include <linux/quota.h>
28717 #include <linux/un.h> /* for Unix socket types */
28718 -#include <net/af_unix.h> /* for Unix socket types */
28719 #include <linux/parser.h>
28720 #include <linux/nfs_mount.h>
28721 #include <net/ipv6.h>
28722 diff -NurpP --minimal linux-2.6.32.1/security/selinux/include/av_permissions.h linux-2.6.32.1-vs2.3.0.36.27/security/selinux/include/av_permissions.h
28723 --- linux-2.6.32.1/security/selinux/include/av_permissions.h 2009-12-03 20:03:02.000000000 +0100
28724 +++ linux-2.6.32.1-vs2.3.0.36.27/security/selinux/include/av_permissions.h 2009-12-03 20:04:56.000000000 +0100
28725 @@ -565,6 +565,7 @@
28726 #define CAPABILITY__SETFCAP 0x80000000UL
28727 #define CAPABILITY2__MAC_OVERRIDE 0x00000001UL
28728 #define CAPABILITY2__MAC_ADMIN 0x00000002UL
28729 +#define CAPABILITY2__CONTEXT 0x00000004UL
28730 #define NETLINK_ROUTE_SOCKET__IOCTL 0x00000001UL
28731 #define NETLINK_ROUTE_SOCKET__READ 0x00000002UL
28732 #define NETLINK_ROUTE_SOCKET__WRITE 0x00000004UL
28733 diff -NurpP --minimal linux-2.6.32.1/security/selinux/include/av_perm_to_string.h linux-2.6.32.1-vs2.3.0.36.27/security/selinux/include/av_perm_to_string.h
28734 --- linux-2.6.32.1/security/selinux/include/av_perm_to_string.h 2009-12-03 20:03:02.000000000 +0100
28735 +++ linux-2.6.32.1-vs2.3.0.36.27/security/selinux/include/av_perm_to_string.h 2009-12-03 20:04:56.000000000 +0100
28736 @@ -142,6 +142,7 @@
28737 S_(SECCLASS_CAPABILITY, CAPABILITY__SETFCAP, "setfcap")
28738 S_(SECCLASS_CAPABILITY2, CAPABILITY2__MAC_OVERRIDE, "mac_override")
28739 S_(SECCLASS_CAPABILITY2, CAPABILITY2__MAC_ADMIN, "mac_admin")
28740 + S_(SECCLASS_CAPABILITY2, CAPABILITY2__CONTEXT, "context")
28741 S_(SECCLASS_NETLINK_ROUTE_SOCKET, NETLINK_ROUTE_SOCKET__NLMSG_READ, "nlmsg_read")
28742 S_(SECCLASS_NETLINK_ROUTE_SOCKET, NETLINK_ROUTE_SOCKET__NLMSG_WRITE, "nlmsg_write")
28743 S_(SECCLASS_NETLINK_FIREWALL_SOCKET, NETLINK_FIREWALL_SOCKET__NLMSG_READ, "nlmsg_read")