Get Gentoo!
gentoo.org sites
gentoo.org Wiki Bugs Forums Packages
Planet Archives Sources
Infra Status

Gentoo Archives: gentoo-commits

From: "Benedikt Boehm (hollow)" <hollow@g.o>
To: gentoo-commits@l.g.o
Subject: [gentoo-commits] vps r562 - in vserver-sources: . 2.3.0.36.14
Date: Sun, 28 Jun 2009 17:36:18
Message-Id: E1MKyI0-0007Fl-W8@stork.gentoo.org
1 Author: hollow
2 Date: 2009-06-28 17:35:40 +0000 (Sun, 28 Jun 2009)
3 New Revision: 562
4
5 Added:
6 vserver-sources/2.3.0.36.14/
7 vserver-sources/2.3.0.36.14/4410_vs2.3.0.36.14.patch
8 Log:
9 version bump
10
11 Added: vserver-sources/2.3.0.36.14/4410_vs2.3.0.36.14.patch
12 ===================================================================
13 --- vserver-sources/2.3.0.36.14/4410_vs2.3.0.36.14.patch (rev 0)
14 +++ vserver-sources/2.3.0.36.14/4410_vs2.3.0.36.14.patch 2009-06-28 17:35:40 UTC (rev 562)
15 @@ -0,0 +1,28140 @@
16 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/Kconfig
17 +--- linux-2.6.29.4/arch/alpha/Kconfig 2009-03-24 14:18:07.000000000 +0100
18 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/Kconfig 2009-03-24 14:48:16.000000000 +0100
19 +@@ -666,6 +666,8 @@ config DUMMY_CONSOLE
20 + depends on VGA_HOSE
21 + default y
22 +
23 ++source "kernel/vserver/Kconfig"
24 ++
25 + source "security/Kconfig"
26 +
27 + source "crypto/Kconfig"
28 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/kernel/entry.S linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/entry.S
29 +--- linux-2.6.29.4/arch/alpha/kernel/entry.S 2009-03-24 14:18:07.000000000 +0100
30 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/entry.S 2009-03-24 14:48:16.000000000 +0100
31 +@@ -874,24 +874,15 @@ sys_getxgid:
32 + .globl sys_getxpid
33 + .ent sys_getxpid
34 + sys_getxpid:
35 ++ lda $sp, -16($sp)
36 ++ stq $26, 0($sp)
37 + .prologue 0
38 +- ldq $2, TI_TASK($8)
39 +
40 +- /* See linux/kernel/timer.c sys_getppid for discussion
41 +- about this loop. */
42 +- ldq $3, TASK_GROUP_LEADER($2)
43 +- ldq $4, TASK_REAL_PARENT($3)
44 +- ldl $0, TASK_TGID($2)
45 +-1: ldl $1, TASK_TGID($4)
46 +-#ifdef CONFIG_SMP
47 +- mov $4, $5
48 +- mb
49 +- ldq $3, TASK_GROUP_LEADER($2)
50 +- ldq $4, TASK_REAL_PARENT($3)
51 +- cmpeq $4, $5, $5
52 +- beq $5, 1b
53 +-#endif
54 +- stq $1, 80($sp)
55 ++ lda $16, 96($sp)
56 ++ jsr $26, do_getxpid
57 ++ ldq $26, 0($sp)
58 ++
59 ++ lda $sp, 16($sp)
60 + ret
61 + .end sys_getxpid
62 +
63 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/kernel/osf_sys.c linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/osf_sys.c
64 +--- linux-2.6.29.4/arch/alpha/kernel/osf_sys.c 2009-03-24 14:18:07.000000000 +0100
65 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/osf_sys.c 2009-03-24 14:48:16.000000000 +0100
66 +@@ -877,7 +877,7 @@ SYSCALL_DEFINE2(osf_gettimeofday, struct
67 + {
68 + if (tv) {
69 + struct timeval ktv;
70 +- do_gettimeofday(&ktv);
71 ++ vx_gettimeofday(&ktv);
72 + if (put_tv32(tv, &ktv))
73 + return -EFAULT;
74 + }
75 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/ptrace.c
76 +--- linux-2.6.29.4/arch/alpha/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
77 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
78 +@@ -15,6 +15,7 @@
79 + #include <linux/slab.h>
80 + #include <linux/security.h>
81 + #include <linux/signal.h>
82 ++#include <linux/vs_base.h>
83 +
84 + #include <asm/uaccess.h>
85 + #include <asm/pgtable.h>
86 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/kernel/systbls.S linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/systbls.S
87 +--- linux-2.6.29.4/arch/alpha/kernel/systbls.S 2009-03-24 14:18:08.000000000 +0100
88 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/systbls.S 2009-03-24 14:48:16.000000000 +0100
89 +@@ -446,7 +446,7 @@ sys_call_table:
90 + .quad sys_stat64 /* 425 */
91 + .quad sys_lstat64
92 + .quad sys_fstat64
93 +- .quad sys_ni_syscall /* sys_vserver */
94 ++ .quad sys_vserver /* sys_vserver */
95 + .quad sys_ni_syscall /* sys_mbind */
96 + .quad sys_ni_syscall /* sys_get_mempolicy */
97 + .quad sys_ni_syscall /* sys_set_mempolicy */
98 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/traps.c
99 +--- linux-2.6.29.4/arch/alpha/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
100 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
101 +@@ -183,7 +183,8 @@ die_if_kernel(char * str, struct pt_regs
102 + #ifdef CONFIG_SMP
103 + printk("CPU %d ", hard_smp_processor_id());
104 + #endif
105 +- printk("%s(%d): %s %ld\n", current->comm, task_pid_nr(current), str, err);
106 ++ printk("%s(%d[#%u]): %s %ld\n", current->comm,
107 ++ task_pid_nr(current), current->xid, str, err);
108 + dik_show_regs(regs, r9_15);
109 + add_taint(TAINT_DIE);
110 + dik_show_trace((unsigned long *)(regs+1));
111 +diff -NurpP --minimal linux-2.6.29.4/arch/alpha/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/mm/fault.c
112 +--- linux-2.6.29.4/arch/alpha/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
113 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/alpha/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
114 +@@ -193,8 +193,8 @@ do_page_fault(unsigned long address, uns
115 + down_read(&mm->mmap_sem);
116 + goto survive;
117 + }
118 +- printk(KERN_ALERT "VM: killing process %s(%d)\n",
119 +- current->comm, task_pid_nr(current));
120 ++ printk(KERN_ALERT "VM: killing process %s(%d:#%u)\n",
121 ++ current->comm, task_pid_nr(current), current->xid);
122 + if (!user_mode(regs))
123 + goto no_context;
124 + do_group_exit(SIGKILL);
125 +diff -NurpP --minimal linux-2.6.29.4/arch/arm/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/arm/Kconfig
126 +--- linux-2.6.29.4/arch/arm/Kconfig 2009-03-24 14:18:08.000000000 +0100
127 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/arm/Kconfig 2009-03-24 14:48:16.000000000 +0100
128 +@@ -1333,6 +1333,8 @@ source "fs/Kconfig"
129 +
130 + source "arch/arm/Kconfig.debug"
131 +
132 ++source "kernel/vserver/Kconfig"
133 ++
134 + source "security/Kconfig"
135 +
136 + source "crypto/Kconfig"
137 +diff -NurpP --minimal linux-2.6.29.4/arch/arm/kernel/calls.S linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/calls.S
138 +--- linux-2.6.29.4/arch/arm/kernel/calls.S 2009-03-24 14:18:09.000000000 +0100
139 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/calls.S 2009-02-22 22:54:24.000000000 +0100
140 +@@ -322,7 +322,7 @@
141 + /* 310 */ CALL(sys_request_key)
142 + CALL(sys_keyctl)
143 + CALL(ABI(sys_semtimedop, sys_oabi_semtimedop))
144 +-/* vserver */ CALL(sys_ni_syscall)
145 ++ CALL(sys_vserver)
146 + CALL(sys_ioprio_set)
147 + /* 315 */ CALL(sys_ioprio_get)
148 + CALL(sys_inotify_init)
149 +diff -NurpP --minimal linux-2.6.29.4/arch/arm/kernel/process.c linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/process.c
150 +--- linux-2.6.29.4/arch/arm/kernel/process.c 2008-12-25 00:26:37.000000000 +0100
151 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/process.c 2009-02-22 22:54:24.000000000 +0100
152 +@@ -262,7 +262,8 @@ void __show_regs(struct pt_regs *regs)
153 + void show_regs(struct pt_regs * regs)
154 + {
155 + printk("\n");
156 +- printk("Pid: %d, comm: %20s\n", task_pid_nr(current), current->comm);
157 ++ printk("Pid: %d[#%u], comm: %20s\n",
158 ++ task_pid_nr(current), current->xid, current->comm);
159 + __show_regs(regs);
160 + __backtrace();
161 + }
162 +diff -NurpP --minimal linux-2.6.29.4/arch/arm/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/traps.c
163 +--- linux-2.6.29.4/arch/arm/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
164 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/arm/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
165 +@@ -214,8 +214,8 @@ static void __die(const char *str, int e
166 + str, err, ++die_counter);
167 + print_modules();
168 + __show_regs(regs);
169 +- printk("Process %s (pid: %d, stack limit = 0x%p)\n",
170 +- tsk->comm, task_pid_nr(tsk), thread + 1);
171 ++ printk("Process %s (pid: %d:#%u, stack limit = 0x%p)\n",
172 ++ tsk->comm, task_pid_nr(tsk), tsk->xid, thread + 1);
173 +
174 + if (!user_mode(regs) || in_interrupt()) {
175 + dump_mem("Stack: ", regs->ARM_sp,
176 +diff -NurpP --minimal linux-2.6.29.4/arch/arm/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/arm/mm/fault.c
177 +--- linux-2.6.29.4/arch/arm/mm/fault.c 2009-03-24 14:18:17.000000000 +0100
178 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/arm/mm/fault.c 2009-03-24 14:48:17.000000000 +0100
179 +@@ -294,7 +294,8 @@ do_page_fault(unsigned long addr, unsign
180 + * happened to us that made us unable to handle
181 + * the page fault gracefully.
182 + */
183 +- printk("VM: killing process %s\n", tsk->comm);
184 ++ printk("VM: killing process %s(%d:#%u)\n",
185 ++ tsk->comm, task_pid_nr(tsk), tsk->xid);
186 + do_group_exit(SIGKILL);
187 + return 0;
188 + }
189 +diff -NurpP --minimal linux-2.6.29.4/arch/cris/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/cris/Kconfig
190 +--- linux-2.6.29.4/arch/cris/Kconfig 2009-03-24 14:18:23.000000000 +0100
191 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/cris/Kconfig 2009-03-24 14:48:19.000000000 +0100
192 +@@ -685,6 +685,8 @@ source "drivers/staging/Kconfig"
193 +
194 + source "arch/cris/Kconfig.debug"
195 +
196 ++source "kernel/vserver/Kconfig"
197 ++
198 + source "security/Kconfig"
199 +
200 + source "crypto/Kconfig"
201 +diff -NurpP --minimal linux-2.6.29.4/arch/frv/kernel/kernel_thread.S linux-2.6.29.4-vs2.3.0.36.14/arch/frv/kernel/kernel_thread.S
202 +--- linux-2.6.29.4/arch/frv/kernel/kernel_thread.S 2008-12-25 00:26:37.000000000 +0100
203 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/frv/kernel/kernel_thread.S 2009-02-22 22:54:24.000000000 +0100
204 +@@ -37,7 +37,7 @@ kernel_thread:
205 +
206 + # start by forking the current process, but with shared VM
207 + setlos.p #__NR_clone,gr7 ; syscall number
208 +- ori gr10,#CLONE_VM,gr8 ; first syscall arg [clone_flags]
209 ++ ori gr10,#CLONE_KT,gr8 ; first syscall arg [clone_flags]
210 + sethi.p #0xe4e4,gr9 ; second syscall arg [newsp]
211 + setlo #0xe4e4,gr9
212 + setlos.p #0,gr10 ; third syscall arg [parent_tidptr]
213 +diff -NurpP --minimal linux-2.6.29.4/arch/h8300/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/h8300/Kconfig
214 +--- linux-2.6.29.4/arch/h8300/Kconfig 2009-03-24 14:18:24.000000000 +0100
215 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/h8300/Kconfig 2009-03-24 14:48:19.000000000 +0100
216 +@@ -226,6 +226,8 @@ source "fs/Kconfig"
217 +
218 + source "arch/h8300/Kconfig.debug"
219 +
220 ++source "kernel/vserver/Kconfig"
221 ++
222 + source "security/Kconfig"
223 +
224 + source "crypto/Kconfig"
225 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/ia32/ia32_entry.S linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/ia32/ia32_entry.S
226 +--- linux-2.6.29.4/arch/ia64/ia32/ia32_entry.S 2009-03-24 14:18:24.000000000 +0100
227 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/ia32/ia32_entry.S 2009-02-22 22:54:24.000000000 +0100
228 +@@ -451,7 +451,7 @@ ia32_syscall_table:
229 + data8 sys_tgkill /* 270 */
230 + data8 compat_sys_utimes
231 + data8 sys32_fadvise64_64
232 +- data8 sys_ni_syscall
233 ++ data8 sys32_vserver
234 + data8 sys_ni_syscall
235 + data8 sys_ni_syscall /* 275 */
236 + data8 sys_ni_syscall
237 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/Kconfig
238 +--- linux-2.6.29.4/arch/ia64/Kconfig 2009-03-24 14:18:24.000000000 +0100
239 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/Kconfig 2009-03-24 14:48:19.000000000 +0100
240 +@@ -672,6 +672,8 @@ source "fs/Kconfig"
241 +
242 + source "arch/ia64/Kconfig.debug"
243 +
244 ++source "kernel/vserver/Kconfig"
245 ++
246 + source "security/Kconfig"
247 +
248 + source "crypto/Kconfig"
249 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/kernel/entry.S linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/entry.S
250 +--- linux-2.6.29.4/arch/ia64/kernel/entry.S 2009-03-24 14:18:25.000000000 +0100
251 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/entry.S 2009-02-22 22:54:24.000000000 +0100
252 +@@ -1653,7 +1653,7 @@ sys_call_table:
253 + data8 sys_mq_notify
254 + data8 sys_mq_getsetattr
255 + data8 sys_kexec_load
256 +- data8 sys_ni_syscall // reserved for vserver
257 ++ data8 sys_vserver
258 + data8 sys_waitid // 1270
259 + data8 sys_add_key
260 + data8 sys_request_key
261 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/kernel/perfmon.c linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/perfmon.c
262 +--- linux-2.6.29.4/arch/ia64/kernel/perfmon.c 2009-03-24 14:18:25.000000000 +0100
263 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/perfmon.c 2009-03-24 14:48:19.000000000 +0100
264 +@@ -41,6 +41,7 @@
265 + #include <linux/rcupdate.h>
266 + #include <linux/completion.h>
267 + #include <linux/tracehook.h>
268 ++#include <linux/vs_memory.h>
269 +
270 + #include <asm/errno.h>
271 + #include <asm/intrinsics.h>
272 +@@ -2372,7 +2373,7 @@ pfm_smpl_buffer_alloc(struct task_struct
273 + */
274 + insert_vm_struct(mm, vma);
275 +
276 +- mm->total_vm += size >> PAGE_SHIFT;
277 ++ vx_vmpages_add(mm, size >> PAGE_SHIFT);
278 + vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
279 + vma_pages(vma));
280 + up_write(&task->mm->mmap_sem);
281 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/kernel/process.c linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/process.c
282 +--- linux-2.6.29.4/arch/ia64/kernel/process.c 2008-12-25 00:26:37.000000000 +0100
283 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/process.c 2009-02-22 22:54:24.000000000 +0100
284 +@@ -110,8 +110,8 @@ show_regs (struct pt_regs *regs)
285 + unsigned long ip = regs->cr_iip + ia64_psr(regs)->ri;
286 +
287 + print_modules();
288 +- printk("\nPid: %d, CPU %d, comm: %20s\n", task_pid_nr(current),
289 +- smp_processor_id(), current->comm);
290 ++ printk("\nPid: %d[#%u], CPU %d, comm: %20s\n", task_pid_nr(current),
291 ++ current->xid, smp_processor_id(), current->comm);
292 + printk("psr : %016lx ifs : %016lx ip : [<%016lx>] %s (%s)\n",
293 + regs->cr_ipsr, regs->cr_ifs, ip, print_tainted(),
294 + init_utsname()->release);
295 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/ptrace.c
296 +--- linux-2.6.29.4/arch/ia64/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
297 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
298 +@@ -23,6 +23,7 @@
299 + #include <linux/regset.h>
300 + #include <linux/elf.h>
301 + #include <linux/tracehook.h>
302 ++#include <linux/vs_base.h>
303 +
304 + #include <asm/pgtable.h>
305 + #include <asm/processor.h>
306 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/traps.c
307 +--- linux-2.6.29.4/arch/ia64/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
308 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
309 +@@ -60,8 +60,9 @@ die (const char *str, struct pt_regs *re
310 + put_cpu();
311 +
312 + if (++die.lock_owner_depth < 3) {
313 +- printk("%s[%d]: %s %ld [%d]\n",
314 +- current->comm, task_pid_nr(current), str, err, ++die_counter);
315 ++ printk("%s[%d[#%u]]: %s %ld [%d]\n",
316 ++ current->comm, task_pid_nr(current), current->xid,
317 ++ str, err, ++die_counter);
318 + if (notify_die(DIE_OOPS, str, regs, err, 255, SIGSEGV)
319 + != NOTIFY_STOP)
320 + show_regs(regs);
321 +@@ -324,8 +325,9 @@ handle_fpu_swa (int fp_fault, struct pt_
322 + if ((last.count & 15) < 5 && (ia64_fetchadd(1, &last.count, acq) & 15) < 5) {
323 + last.time = current_jiffies + 5 * HZ;
324 + printk(KERN_WARNING
325 +- "%s(%d): floating-point assist fault at ip %016lx, isr %016lx\n",
326 +- current->comm, task_pid_nr(current), regs->cr_iip + ia64_psr(regs)->ri, isr);
327 ++ "%s(%d[#%u]): floating-point assist fault at ip %016lx, isr %016lx\n",
328 ++ current->comm, task_pid_nr(current), current->xid,
329 ++ regs->cr_iip + ia64_psr(regs)->ri, isr);
330 + }
331 + }
332 + }
333 +diff -NurpP --minimal linux-2.6.29.4/arch/ia64/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/mm/fault.c
334 +--- linux-2.6.29.4/arch/ia64/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
335 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/ia64/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
336 +@@ -10,6 +10,7 @@
337 + #include <linux/interrupt.h>
338 + #include <linux/kprobes.h>
339 + #include <linux/kdebug.h>
340 ++#include <linux/vs_memory.h>
341 +
342 + #include <asm/pgtable.h>
343 + #include <asm/processor.h>
344 +diff -NurpP --minimal linux-2.6.29.4/arch/m32r/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/m32r/kernel/traps.c
345 +--- linux-2.6.29.4/arch/m32r/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
346 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m32r/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
347 +@@ -196,8 +196,9 @@ static void show_registers(struct pt_reg
348 + } else {
349 + printk("SPI: %08lx\n", sp);
350 + }
351 +- printk("Process %s (pid: %d, process nr: %d, stackpage=%08lx)",
352 +- current->comm, task_pid_nr(current), 0xffff & i, 4096+(unsigned long)current);
353 ++ printk("Process %s (pid: %d[#%u], process nr: %d, stackpage=%08lx)",
354 ++ current->comm, task_pid_nr(current), current->xid,
355 ++ 0xffff & i, 4096+(unsigned long)current);
356 +
357 + /*
358 + * When in-kernel, we also print out the stack and code at the
359 +diff -NurpP --minimal linux-2.6.29.4/arch/m68k/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/Kconfig
360 +--- linux-2.6.29.4/arch/m68k/Kconfig 2009-03-24 14:18:26.000000000 +0100
361 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/Kconfig 2009-03-24 14:48:20.000000000 +0100
362 +@@ -616,6 +616,8 @@ source "fs/Kconfig"
363 +
364 + source "arch/m68k/Kconfig.debug"
365 +
366 ++source "kernel/vserver/Kconfig"
367 ++
368 + source "security/Kconfig"
369 +
370 + source "crypto/Kconfig"
371 +diff -NurpP --minimal linux-2.6.29.4/arch/m68k/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/kernel/ptrace.c
372 +--- linux-2.6.29.4/arch/m68k/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
373 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
374 +@@ -18,6 +18,7 @@
375 + #include <linux/ptrace.h>
376 + #include <linux/user.h>
377 + #include <linux/signal.h>
378 ++#include <linux/vs_base.h>
379 +
380 + #include <asm/uaccess.h>
381 + #include <asm/page.h>
382 +@@ -269,6 +270,8 @@ long arch_ptrace(struct task_struct *chi
383 + ret = ptrace_request(child, request, addr, data);
384 + break;
385 + }
386 ++ if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
387 ++ goto out_tsk;
388 +
389 + return ret;
390 + out_eio:
391 +diff -NurpP --minimal linux-2.6.29.4/arch/m68k/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/kernel/traps.c
392 +--- linux-2.6.29.4/arch/m68k/kernel/traps.c 2009-03-24 14:18:26.000000000 +0100
393 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m68k/kernel/traps.c 2009-03-24 14:48:20.000000000 +0100
394 +@@ -906,8 +906,8 @@ void show_registers(struct pt_regs *regs
395 + printk("d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
396 + regs->d4, regs->d5, regs->a0, regs->a1);
397 +
398 +- printk("Process %s (pid: %d, task=%p)\n",
399 +- current->comm, task_pid_nr(current), current);
400 ++ printk("Process %s (pid: %d[#%u], task=%p)\n",
401 ++ current->comm, task_pid_nr(current), current->xid, current);
402 + addr = (unsigned long)&fp->un;
403 + printk("Frame format=%X ", regs->format);
404 + switch (regs->format) {
405 +diff -NurpP --minimal linux-2.6.29.4/arch/m68knommu/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/m68knommu/Kconfig
406 +--- linux-2.6.29.4/arch/m68knommu/Kconfig 2009-03-24 14:18:27.000000000 +0100
407 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m68knommu/Kconfig 2009-03-24 14:48:20.000000000 +0100
408 +@@ -720,6 +720,8 @@ source "fs/Kconfig"
409 +
410 + source "arch/m68knommu/Kconfig.debug"
411 +
412 ++source "kernel/vserver/Kconfig"
413 ++
414 + source "security/Kconfig"
415 +
416 + source "crypto/Kconfig"
417 +diff -NurpP --minimal linux-2.6.29.4/arch/m68knommu/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/m68knommu/kernel/traps.c
418 +--- linux-2.6.29.4/arch/m68knommu/kernel/traps.c 2008-12-25 00:26:37.000000000 +0100
419 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/m68knommu/kernel/traps.c 2009-02-22 22:54:24.000000000 +0100
420 +@@ -78,8 +78,9 @@ void die_if_kernel(char *str, struct pt_
421 + printk(KERN_EMERG "d4: %08lx d5: %08lx a0: %08lx a1: %08lx\n",
422 + fp->d4, fp->d5, fp->a0, fp->a1);
423 +
424 +- printk(KERN_EMERG "Process %s (pid: %d, stackpage=%08lx)\n",
425 +- current->comm, current->pid, PAGE_SIZE+(unsigned long)current);
426 ++ printk(KERN_EMERG "Process %s (pid: %d[#%u], stackpage=%08lx)\n",
427 ++ current->comm, task_pid_nr(current), current->xid,
428 ++ PAGE_SIZE+(unsigned long)current);
429 + show_stack(NULL, (unsigned long *)(fp + 1));
430 + add_taint(TAINT_DIE);
431 + do_exit(SIGSEGV);
432 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/mips/Kconfig
433 +--- linux-2.6.29.4/arch/mips/Kconfig 2009-03-24 14:18:29.000000000 +0100
434 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/Kconfig 2009-03-24 14:48:21.000000000 +0100
435 +@@ -2142,6 +2142,8 @@ source "fs/Kconfig"
436 +
437 + source "arch/mips/Kconfig.debug"
438 +
439 ++source "kernel/vserver/Kconfig"
440 ++
441 + source "security/Kconfig"
442 +
443 + source "crypto/Kconfig"
444 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/ptrace.c
445 +--- linux-2.6.29.4/arch/mips/kernel/ptrace.c 2008-12-25 00:26:37.000000000 +0100
446 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/ptrace.c 2009-02-22 22:54:24.000000000 +0100
447 +@@ -25,6 +25,7 @@
448 + #include <linux/security.h>
449 + #include <linux/audit.h>
450 + #include <linux/seccomp.h>
451 ++#include <linux/vs_base.h>
452 +
453 + #include <asm/byteorder.h>
454 + #include <asm/cpu.h>
455 +@@ -259,6 +260,9 @@ long arch_ptrace(struct task_struct *chi
456 + {
457 + int ret;
458 +
459 ++ if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
460 ++ goto out;
461 ++
462 + switch (request) {
463 + /* when I and D space are separate, these will need to be fixed. */
464 + case PTRACE_PEEKTEXT: /* read word at location addr. */
465 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/scall32-o32.S linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall32-o32.S
466 +--- linux-2.6.29.4/arch/mips/kernel/scall32-o32.S 2009-03-24 14:18:31.000000000 +0100
467 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall32-o32.S 2009-03-24 14:48:21.000000000 +0100
468 +@@ -597,7 +597,7 @@ einval: li v0, -ENOSYS
469 + sys sys_mq_timedreceive 5
470 + sys sys_mq_notify 2 /* 4275 */
471 + sys sys_mq_getsetattr 3
472 +- sys sys_ni_syscall 0 /* sys_vserver */
473 ++ sys sys_vserver 3
474 + sys sys_waitid 5
475 + sys sys_ni_syscall 0 /* available, was setaltroot */
476 + sys sys_add_key 5 /* 4280 */
477 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/scall64-64.S linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-64.S
478 +--- linux-2.6.29.4/arch/mips/kernel/scall64-64.S 2009-03-24 14:18:31.000000000 +0100
479 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-64.S 2009-03-24 14:48:21.000000000 +0100
480 +@@ -434,7 +434,7 @@ sys_call_table:
481 + PTR sys_mq_timedreceive
482 + PTR sys_mq_notify
483 + PTR sys_mq_getsetattr /* 5235 */
484 +- PTR sys_ni_syscall /* sys_vserver */
485 ++ PTR sys_vserver
486 + PTR sys_waitid
487 + PTR sys_ni_syscall /* available, was setaltroot */
488 + PTR sys_add_key
489 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/scall64-n32.S linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-n32.S
490 +--- linux-2.6.29.4/arch/mips/kernel/scall64-n32.S 2009-03-24 14:18:31.000000000 +0100
491 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-n32.S 2009-03-24 14:48:21.000000000 +0100
492 +@@ -360,7 +360,7 @@ EXPORT(sysn32_call_table)
493 + PTR compat_sys_mq_timedreceive
494 + PTR compat_sys_mq_notify
495 + PTR compat_sys_mq_getsetattr
496 +- PTR sys_ni_syscall /* 6240, sys_vserver */
497 ++ PTR sys32_vserver /* 6240 */
498 + PTR compat_sys_waitid
499 + PTR sys_ni_syscall /* available, was setaltroot */
500 + PTR sys_add_key
501 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/scall64-o32.S linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-o32.S
502 +--- linux-2.6.29.4/arch/mips/kernel/scall64-o32.S 2009-03-24 14:18:31.000000000 +0100
503 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/scall64-o32.S 2009-03-24 14:59:48.000000000 +0100
504 +@@ -480,7 +480,7 @@ sys_call_table:
505 + PTR compat_sys_mq_timedreceive
506 + PTR compat_sys_mq_notify /* 4275 */
507 + PTR compat_sys_mq_getsetattr
508 +- PTR sys_ni_syscall /* sys_vserver */
509 ++ PTR sys32_vserver
510 + PTR sys_32_waitid
511 + PTR sys_ni_syscall /* available, was setaltroot */
512 + PTR sys_add_key /* 4280 */
513 +diff -NurpP --minimal linux-2.6.29.4/arch/mips/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/traps.c
514 +--- linux-2.6.29.4/arch/mips/kernel/traps.c 2009-03-24 14:18:31.000000000 +0100
515 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/mips/kernel/traps.c 2009-03-24 14:48:21.000000000 +0100
516 +@@ -335,9 +335,10 @@ void show_registers(const struct pt_regs
517 +
518 + __show_regs(regs);
519 + print_modules();
520 +- printk("Process %s (pid: %d, threadinfo=%p, task=%p, tls=%0*lx)\n",
521 +- current->comm, current->pid, current_thread_info(), current,
522 +- field, current_thread_info()->tp_value);
523 ++ printk("Process %s (pid: %d:#%u, threadinfo=%p, task=%p, tls=%0*lx)\n",
524 ++ current->comm, task_pid_nr(current), current->xid,
525 ++ current_thread_info(), current,
526 ++ field, current_thread_info()->tp_value);
527 + if (cpu_has_userlocal) {
528 + unsigned long tls;
529 +
530 +diff -NurpP --minimal linux-2.6.29.4/arch/parisc/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/Kconfig
531 +--- linux-2.6.29.4/arch/parisc/Kconfig 2009-03-24 14:18:32.000000000 +0100
532 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/Kconfig 2009-03-24 14:48:21.000000000 +0100
533 +@@ -281,6 +281,8 @@ source "fs/Kconfig"
534 +
535 + source "arch/parisc/Kconfig.debug"
536 +
537 ++source "kernel/vserver/Kconfig"
538 ++
539 + source "security/Kconfig"
540 +
541 + source "crypto/Kconfig"
542 +diff -NurpP --minimal linux-2.6.29.4/arch/parisc/kernel/syscall_table.S linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/kernel/syscall_table.S
543 +--- linux-2.6.29.4/arch/parisc/kernel/syscall_table.S 2008-12-25 00:26:37.000000000 +0100
544 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/kernel/syscall_table.S 2009-02-22 22:54:24.000000000 +0100
545 +@@ -361,7 +361,7 @@
546 + ENTRY_COMP(mbind) /* 260 */
547 + ENTRY_COMP(get_mempolicy)
548 + ENTRY_COMP(set_mempolicy)
549 +- ENTRY_SAME(ni_syscall) /* 263: reserved for vserver */
550 ++ ENTRY_DIFF(vserver)
551 + ENTRY_SAME(add_key)
552 + ENTRY_SAME(request_key) /* 265 */
553 + ENTRY_SAME(keyctl)
554 +diff -NurpP --minimal linux-2.6.29.4/arch/parisc/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/kernel/traps.c
555 +--- linux-2.6.29.4/arch/parisc/kernel/traps.c 2009-03-24 14:18:32.000000000 +0100
556 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/kernel/traps.c 2009-03-24 14:48:21.000000000 +0100
557 +@@ -236,8 +236,9 @@ void die_if_kernel(char *str, struct pt_
558 + if (err == 0)
559 + return; /* STFU */
560 +
561 +- printk(KERN_CRIT "%s (pid %d): %s (code %ld) at " RFMT "\n",
562 +- current->comm, task_pid_nr(current), str, err, regs->iaoq[0]);
563 ++ printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld) at " RFMT "\n",
564 ++ current->comm, task_pid_nr(current), current->xid,
565 ++ str, err, regs->iaoq[0]);
566 + #ifdef PRINT_USER_FAULTS
567 + /* XXX for debugging only */
568 + show_regs(regs);
569 +@@ -269,8 +270,8 @@ KERN_CRIT " || |
570 + pdc_console_restart();
571 +
572 + if (err)
573 +- printk(KERN_CRIT "%s (pid %d): %s (code %ld)\n",
574 +- current->comm, task_pid_nr(current), str, err);
575 ++ printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld)\n",
576 ++ current->comm, task_pid_nr(current), current->xid, str, err);
577 +
578 + /* Wot's wrong wif bein' racy? */
579 + if (current->thread.flags & PARISC_KERNEL_DEATH) {
580 +diff -NurpP --minimal linux-2.6.29.4/arch/parisc/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/mm/fault.c
581 +--- linux-2.6.29.4/arch/parisc/mm/fault.c 2009-03-24 14:18:33.000000000 +0100
582 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/parisc/mm/fault.c 2009-03-24 14:48:21.000000000 +0100
583 +@@ -238,8 +238,9 @@ bad_area:
584 +
585 + #ifdef PRINT_USER_FAULTS
586 + printk(KERN_DEBUG "\n");
587 +- printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n",
588 +- task_pid_nr(tsk), tsk->comm, code, address);
589 ++ printk(KERN_DEBUG "do_page_fault() pid=%d:#%u "
590 ++ "command='%s' type=%lu address=0x%08lx\n",
591 ++ task_pid_nr(tsk), tsk->xid, tsk->comm, code, address);
592 + if (vma) {
593 + printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n",
594 + vma->vm_start, vma->vm_end);
595 +@@ -265,7 +266,8 @@ no_context:
596 +
597 + out_of_memory:
598 + up_read(&mm->mmap_sem);
599 +- printk(KERN_CRIT "VM: killing process %s\n", current->comm);
600 ++ printk(KERN_CRIT "VM: killing process %s(%d:#%u)\n",
601 ++ current->comm, current->pid, current->xid);
602 + if (user_mode(regs))
603 + do_group_exit(SIGKILL);
604 + goto no_context;
605 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/Kconfig
606 +--- linux-2.6.29.4/arch/powerpc/Kconfig 2009-03-24 14:18:33.000000000 +0100
607 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/Kconfig 2009-03-24 14:48:22.000000000 +0100
608 +@@ -882,6 +882,8 @@ source "lib/Kconfig"
609 +
610 + source "arch/powerpc/Kconfig.debug"
611 +
612 ++source "kernel/vserver/Kconfig"
613 ++
614 + source "security/Kconfig"
615 +
616 + config KEYS_COMPAT
617 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/kernel/irq.c linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/irq.c
618 +--- linux-2.6.29.4/arch/powerpc/kernel/irq.c 2009-03-24 14:18:35.000000000 +0100
619 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/irq.c 2009-03-24 14:48:22.000000000 +0100
620 +@@ -53,6 +53,7 @@
621 + #include <linux/bootmem.h>
622 + #include <linux/pci.h>
623 + #include <linux/debugfs.h>
624 ++#include <linux/vs_context.h>
625 +
626 + #include <asm/uaccess.h>
627 + #include <asm/system.h>
628 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/kernel/process.c linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/process.c
629 +--- linux-2.6.29.4/arch/powerpc/kernel/process.c 2009-03-24 14:18:35.000000000 +0100
630 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/process.c 2009-03-24 14:48:22.000000000 +0100
631 +@@ -516,8 +516,9 @@ void show_regs(struct pt_regs * regs)
632 + #else
633 + printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
634 + #endif
635 +- printk("TASK = %p[%d] '%s' THREAD: %p",
636 +- current, task_pid_nr(current), current->comm, task_thread_info(current));
637 ++ printk("TASK = %p[%d,#%u] '%s' THREAD: %p",
638 ++ current, task_pid_nr(current), current->xid,
639 ++ current->comm, task_thread_info(current));
640 +
641 + #ifdef CONFIG_SMP
642 + printk(" CPU: %d", raw_smp_processor_id());
643 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/kernel/traps.c linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/traps.c
644 +--- linux-2.6.29.4/arch/powerpc/kernel/traps.c 2009-03-24 14:18:35.000000000 +0100
645 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/traps.c 2009-03-24 14:48:22.000000000 +0100
646 +@@ -940,8 +940,9 @@ void nonrecoverable_exception(struct pt_
647 +
648 + void trace_syscall(struct pt_regs *regs)
649 + {
650 +- printk("Task: %p(%d), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
651 +- current, task_pid_nr(current), regs->nip, regs->link, regs->gpr[0],
652 ++ printk("Task: %p(%d[#%u]), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld %s\n",
653 ++ current, task_pid_nr(current), current->xid,
654 ++ regs->nip, regs->link, regs->gpr[0],
655 + regs->ccr&0x10000000?"Error=":"", regs->gpr[3], print_tainted());
656 + }
657 +
658 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/kernel/vdso.c linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/vdso.c
659 +--- linux-2.6.29.4/arch/powerpc/kernel/vdso.c 2009-03-24 14:18:35.000000000 +0100
660 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/kernel/vdso.c 2009-03-24 14:48:22.000000000 +0100
661 +@@ -22,6 +22,7 @@
662 + #include <linux/security.h>
663 + #include <linux/bootmem.h>
664 + #include <linux/lmb.h>
665 ++#include <linux/vs_memory.h>
666 +
667 + #include <asm/pgtable.h>
668 + #include <asm/system.h>
669 +diff -NurpP --minimal linux-2.6.29.4/arch/powerpc/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/mm/fault.c
670 +--- linux-2.6.29.4/arch/powerpc/mm/fault.c 2009-03-24 14:18:36.000000000 +0100
671 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/powerpc/mm/fault.c 2009-03-24 14:48:22.000000000 +0100
672 +@@ -362,7 +362,8 @@ out_of_memory:
673 + down_read(&mm->mmap_sem);
674 + goto survive;
675 + }
676 +- printk("VM: killing process %s\n", current->comm);
677 ++ printk("VM: killing process %s(%d:#%u)\n",
678 ++ current->comm, current->pid, current->xid);
679 + if (user_mode(regs))
680 + do_group_exit(SIGKILL);
681 + return SIGKILL;
682 +diff -NurpP --minimal linux-2.6.29.4/arch/s390/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/s390/Kconfig
683 +--- linux-2.6.29.4/arch/s390/Kconfig 2009-03-24 14:18:38.000000000 +0100
684 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/s390/Kconfig 2009-03-24 14:48:22.000000000 +0100
685 +@@ -586,6 +586,8 @@ source "fs/Kconfig"
686 +
687 + source "arch/s390/Kconfig.debug"
688 +
689 ++source "kernel/vserver/Kconfig"
690 ++
691 + source "security/Kconfig"
692 +
693 + source "crypto/Kconfig"
694 +diff -NurpP --minimal linux-2.6.29.4/arch/s390/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/arch/s390/kernel/ptrace.c
695 +--- linux-2.6.29.4/arch/s390/kernel/ptrace.c 2009-03-24 14:18:40.000000000 +0100
696 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/s390/kernel/ptrace.c 2009-03-24 14:48:22.000000000 +0100
697 +@@ -36,6 +36,7 @@
698 + #include <linux/elf.h>
699 + #include <linux/regset.h>
700 + #include <linux/tracehook.h>
701 ++#include <linux/vs_base.h>
702 +
703 + #include <asm/segment.h>
704 + #include <asm/page.h>
705 +diff -NurpP --minimal linux-2.6.29.4/arch/s390/kernel/syscalls.S linux-2.6.29.4-vs2.3.0.36.14/arch/s390/kernel/syscalls.S
706 +--- linux-2.6.29.4/arch/s390/kernel/syscalls.S 2009-03-24 14:18:40.000000000 +0100
707 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/s390/kernel/syscalls.S 2009-03-24 14:48:22.000000000 +0100
708 +@@ -271,7 +271,7 @@ SYSCALL(sys_clock_settime,sys_clock_sett
709 + SYSCALL(sys_clock_gettime,sys_clock_gettime,sys32_clock_gettime_wrapper) /* 260 */
710 + SYSCALL(sys_clock_getres,sys_clock_getres,sys32_clock_getres_wrapper)
711 + SYSCALL(sys_clock_nanosleep,sys_clock_nanosleep,sys32_clock_nanosleep_wrapper)
712 +-NI_SYSCALL /* reserved for vserver */
713 ++SYSCALL(sys_vserver,sys_vserver,sys32_vserver)
714 + SYSCALL(sys_s390_fadvise64_64,sys_ni_syscall,sys32_fadvise64_64_wrapper)
715 + SYSCALL(sys_statfs64,sys_statfs64,compat_sys_statfs64_wrapper)
716 + SYSCALL(sys_fstatfs64,sys_fstatfs64,compat_sys_fstatfs64_wrapper)
717 +diff -NurpP --minimal linux-2.6.29.4/arch/s390/mm/fault.c linux-2.6.29.4-vs2.3.0.36.14/arch/s390/mm/fault.c
718 +--- linux-2.6.29.4/arch/s390/mm/fault.c 2008-12-25 00:26:37.000000000 +0100
719 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/s390/mm/fault.c 2009-02-22 22:54:24.000000000 +0100
720 +@@ -216,7 +216,8 @@ static int do_out_of_memory(struct pt_re
721 + down_read(&mm->mmap_sem);
722 + return 1;
723 + }
724 +- printk("VM: killing process %s\n", tsk->comm);
725 ++ printk("VM: killing process %s(%d:#%u)\n",
726 ++ tsk->comm, tsk->pid, tsk->xid);
727 + if (regs->psw.mask & PSW_MASK_PSTATE)
728 + do_group_exit(SIGKILL);
729 + do_no_context(regs, error_code, address);
730 +diff -NurpP --minimal linux-2.6.29.4/arch/sh/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/sh/Kconfig
731 +--- linux-2.6.29.4/arch/sh/Kconfig 2009-03-24 14:18:40.000000000 +0100
732 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sh/Kconfig 2009-03-24 14:48:22.000000000 +0100
733 +@@ -694,6 +694,8 @@ source "fs/Kconfig"
734 +
735 + source "arch/sh/Kconfig.debug"
736 +
737 ++source "kernel/vserver/Kconfig"
738 ++
739 + source "security/Kconfig"
740 +
741 + source "crypto/Kconfig"
742 +diff -NurpP --minimal linux-2.6.29.4/arch/sh/kernel/irq.c linux-2.6.29.4-vs2.3.0.36.14/arch/sh/kernel/irq.c
743 +--- linux-2.6.29.4/arch/sh/kernel/irq.c 2008-12-25 00:26:37.000000000 +0100
744 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sh/kernel/irq.c 2009-02-22 22:54:24.000000000 +0100
745 +@@ -11,6 +11,7 @@
746 + #include <linux/module.h>
747 + #include <linux/kernel_stat.h>
748 + #include <linux/seq_file.h>
749 ++#include <linux/vs_context.h>
750 + #include <asm/processor.h>
751 + #include <asm/machvec.h>
752 + #include <asm/uaccess.h>
753 +diff -NurpP --minimal linux-2.6.29.4/arch/sh/kernel/vsyscall/vsyscall.c linux-2.6.29.4-vs2.3.0.36.14/arch/sh/kernel/vsyscall/vsyscall.c
754 +--- linux-2.6.29.4/arch/sh/kernel/vsyscall/vsyscall.c 2009-03-24 14:18:42.000000000 +0100
755 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sh/kernel/vsyscall/vsyscall.c 2009-03-24 14:48:22.000000000 +0100
756 +@@ -19,6 +19,7 @@
757 + #include <linux/elf.h>
758 + #include <linux/sched.h>
759 + #include <linux/err.h>
760 ++#include <linux/vs_memory.h>
761 +
762 + /*
763 + * Should the kernel map a VDSO page into processes and pass its
764 +diff -NurpP --minimal linux-2.6.29.4/arch/sparc/include/asm/tlb_64.h linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/include/asm/tlb_64.h
765 +--- linux-2.6.29.4/arch/sparc/include/asm/tlb_64.h 2009-05-23 23:16:50.000000000 +0200
766 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/include/asm/tlb_64.h 2009-04-30 12:14:53.000000000 +0200
767 +@@ -3,6 +3,7 @@
768 +
769 + #include <linux/swap.h>
770 + #include <linux/pagemap.h>
771 ++#include <linux/vs_memory.h>
772 + #include <asm/pgalloc.h>
773 + #include <asm/tlbflush.h>
774 + #include <asm/mmu_context.h>
775 +diff -NurpP --minimal linux-2.6.29.4/arch/sparc/include/asm/unistd.h linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/include/asm/unistd.h
776 +--- linux-2.6.29.4/arch/sparc/include/asm/unistd.h 2009-03-24 14:18:44.000000000 +0100
777 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/include/asm/unistd.h 2009-04-08 15:52:53.000000000 +0200
778 +@@ -335,7 +335,7 @@
779 + #define __NR_timer_getoverrun 264
780 + #define __NR_timer_delete 265
781 + #define __NR_timer_create 266
782 +-/* #define __NR_vserver 267 Reserved for VSERVER */
783 ++#define __NR_vserver 267
784 + #define __NR_io_setup 268
785 + #define __NR_io_destroy 269
786 + #define __NR_io_submit 270
787 +diff -NurpP --minimal linux-2.6.29.4/arch/sparc/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/Kconfig
788 +--- linux-2.6.29.4/arch/sparc/Kconfig 2009-03-24 14:18:43.000000000 +0100
789 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/Kconfig 2009-03-24 14:48:22.000000000 +0100
790 +@@ -522,6 +522,8 @@ source "fs/Kconfig"
791 +
792 + source "arch/sparc/Kconfig.debug"
793 +
794 ++source "kernel/vserver/Kconfig"
795 ++
796 + source "security/Kconfig"
797 +
798 + source "crypto/Kconfig"
799 +diff -NurpP --minimal linux-2.6.29.4/arch/sparc/kernel/systbls_32.S linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/kernel/systbls_32.S
800 +--- linux-2.6.29.4/arch/sparc/kernel/systbls_32.S 2009-03-24 14:18:45.000000000 +0100
801 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/kernel/systbls_32.S 2009-04-08 15:57:59.000000000 +0200
802 +@@ -70,7 +70,7 @@ sys_call_table:
803 + /*250*/ .long sparc_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
804 + /*255*/ .long sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
805 + /*260*/ .long sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
806 +-/*265*/ .long sys_timer_delete, sys_timer_create, sys_nis_syscall, sys_io_setup, sys_io_destroy
807 ++/*265*/ .long sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
808 + /*270*/ .long sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
809 + /*275*/ .long sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
810 + /*280*/ .long sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
811 +diff -NurpP --minimal linux-2.6.29.4/arch/sparc/kernel/systbls_64.S linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/kernel/systbls_64.S
812 +--- linux-2.6.29.4/arch/sparc/kernel/systbls_64.S 2009-03-24 14:18:45.000000000 +0100
813 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/sparc/kernel/systbls_64.S 2009-04-08 15:56:23.000000000 +0200
814 +@@ -71,7 +71,7 @@ sys_call_table32:
815 + /*250*/ .word sys32_mremap, sys32_sysctl, sys32_getsid, sys_fdatasync, sys32_nfsservctl
816 + .word sys32_sync_file_range, compat_sys_clock_settime, compat_sys_clock_gettime, compat_sys_clock_getres, sys32_clock_nanosleep
817 + /*260*/ .word compat_sys_sched_getaffinity, compat_sys_sched_setaffinity, sys32_timer_settime, compat_sys_timer_gettime, sys_timer_getoverrun
818 +- .word sys_timer_delete, compat_sys_timer_create, sys_ni_syscall, compat_sys_io_setup, sys_io_destroy
819 ++ .word sys_timer_delete, compat_sys_timer_create, sys32_vserver, compat_sys_io_setup, sys_io_destroy
820 + /*270*/ .word sys32_io_submit, sys_io_cancel, compat_sys_io_getevents, sys32_mq_open, sys_mq_unlink
821 + .word compat_sys_mq_timedsend, compat_sys_mq_timedreceive, compat_sys_mq_notify, compat_sys_mq_getsetattr, compat_sys_waitid
822 + /*280*/ .word sys32_tee, sys_add_key, sys_request_key, sys_keyctl, compat_sys_openat
823 +@@ -145,7 +145,7 @@ sys_call_table:
824 + /*250*/ .word sys_64_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
825 + .word sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
826 + /*260*/ .word sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
827 +- .word sys_timer_delete, sys_timer_create, sys_ni_syscall, sys_io_setup, sys_io_destroy
828 ++ .word sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
829 + /*270*/ .word sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
830 + .word sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
831 + /*280*/ .word sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
832 +diff -NurpP --minimal linux-2.6.29.4/arch/x86/ia32/ia32entry.S linux-2.6.29.4-vs2.3.0.36.14/arch/x86/ia32/ia32entry.S
833 +--- linux-2.6.29.4/arch/x86/ia32/ia32entry.S 2009-03-24 14:18:48.000000000 +0100
834 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/x86/ia32/ia32entry.S 2009-03-22 23:29:11.000000000 +0100
835 +@@ -768,7 +768,7 @@ ia32_sys_call_table:
836 + .quad sys_tgkill /* 270 */
837 + .quad compat_sys_utimes
838 + .quad sys32_fadvise64_64
839 +- .quad quiet_ni_syscall /* sys_vserver */
840 ++ .quad sys32_vserver
841 + .quad sys_mbind
842 + .quad compat_sys_get_mempolicy /* 275 */
843 + .quad sys_set_mempolicy
844 +diff -NurpP --minimal linux-2.6.29.4/arch/x86/include/asm/unistd_64.h linux-2.6.29.4-vs2.3.0.36.14/arch/x86/include/asm/unistd_64.h
845 +--- linux-2.6.29.4/arch/x86/include/asm/unistd_64.h 2008-12-25 00:26:37.000000000 +0100
846 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/x86/include/asm/unistd_64.h 2009-02-22 22:54:24.000000000 +0100
847 +@@ -535,7 +535,7 @@ __SYSCALL(__NR_tgkill, sys_tgkill)
848 + #define __NR_utimes 235
849 + __SYSCALL(__NR_utimes, sys_utimes)
850 + #define __NR_vserver 236
851 +-__SYSCALL(__NR_vserver, sys_ni_syscall)
852 ++__SYSCALL(__NR_vserver, sys_vserver)
853 + #define __NR_mbind 237
854 + __SYSCALL(__NR_mbind, sys_mbind)
855 + #define __NR_set_mempolicy 238
856 +diff -NurpP --minimal linux-2.6.29.4/arch/x86/Kconfig linux-2.6.29.4-vs2.3.0.36.14/arch/x86/Kconfig
857 +--- linux-2.6.29.4/arch/x86/Kconfig 2009-03-24 14:18:47.000000000 +0100
858 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/x86/Kconfig 2009-03-24 14:48:23.000000000 +0100
859 +@@ -1990,6 +1990,8 @@ source "fs/Kconfig"
860 +
861 + source "arch/x86/Kconfig.debug"
862 +
863 ++source "kernel/vserver/Kconfig"
864 ++
865 + source "security/Kconfig"
866 +
867 + source "crypto/Kconfig"
868 +diff -NurpP --minimal linux-2.6.29.4/arch/x86/kernel/syscall_table_32.S linux-2.6.29.4-vs2.3.0.36.14/arch/x86/kernel/syscall_table_32.S
869 +--- linux-2.6.29.4/arch/x86/kernel/syscall_table_32.S 2009-03-24 14:18:51.000000000 +0100
870 ++++ linux-2.6.29.4-vs2.3.0.36.14/arch/x86/kernel/syscall_table_32.S 2009-02-22 22:54:24.000000000 +0100
871 +@@ -272,7 +272,7 @@ ENTRY(sys_call_table)
872 + .long sys_tgkill /* 270 */
873 + .long sys_utimes
874 + .long sys_fadvise64_64
875 +- .long sys_ni_syscall /* sys_vserver */
876 ++ .long sys_vserver
877 + .long sys_mbind
878 + .long sys_get_mempolicy
879 + .long sys_set_mempolicy
880 +diff -NurpP --minimal linux-2.6.29.4/Documentation/vserver/debug.txt linux-2.6.29.4-vs2.3.0.36.14/Documentation/vserver/debug.txt
881 +--- linux-2.6.29.4/Documentation/vserver/debug.txt 1970-01-01 01:00:00.000000000 +0100
882 ++++ linux-2.6.29.4-vs2.3.0.36.14/Documentation/vserver/debug.txt 2009-02-22 22:54:24.000000000 +0100
883 +@@ -0,0 +1,154 @@
884 ++
885 ++debug_cvirt:
886 ++
887 ++ 2 4 "vx_map_tgid: %p/%llx: %d -> %d"
888 ++ "vx_rmap_tgid: %p/%llx: %d -> %d"
889 ++
890 ++debug_dlim:
891 ++
892 ++ 0 1 "ALLOC (%p,#%d)%c inode (%d)"
893 ++ "FREE (%p,#%d)%c inode"
894 ++ 1 2 "ALLOC (%p,#%d)%c %lld bytes (%d)"
895 ++ "FREE (%p,#%d)%c %lld bytes"
896 ++ 2 4 "ADJUST: %lld,%lld on %ld,%ld [mult=%d]"
897 ++ 3 8 "ext3_has_free_blocks(%p): %lu<%lu+1, %c, %u!=%u r=%d"
898 ++ "ext3_has_free_blocks(%p): free=%lu, root=%lu"
899 ++ "rcu_free_dl_info(%p)"
900 ++ 4 10 "alloc_dl_info(%p,%d) = %p"
901 ++ "dealloc_dl_info(%p)"
902 ++ "get_dl_info(%p[#%d.%d])"
903 ++ "put_dl_info(%p[#%d.%d])"
904 ++ 5 20 "alloc_dl_info(%p,%d)*"
905 ++ 6 40 "__hash_dl_info: %p[#%d]"
906 ++ "__unhash_dl_info: %p[#%d]"
907 ++ 7 80 "locate_dl_info(%p,#%d) = %p"
908 ++
909 ++debug_misc:
910 ++
911 ++ 0 1 "destroy_dqhash: %p [#0x%08x] c=%d"
912 ++ "new_dqhash: %p [#0x%08x]"
913 ++ "vroot[%d]_clr_dev: dev=%p[%lu,%d:%d]"
914 ++ "vroot[%d]_get_real_bdev: dev=%p[%lu,%d:%d]"
915 ++ "vroot[%d]_set_dev: dev=%p[%lu,%d:%d]"
916 ++ "vroot_get_real_bdev not set"
917 ++ 1 2 "cow_break_link(»%s«)"
918 ++ "temp copy »%s«"
919 ++ 2 4 "dentry_open(new): %p"
920 ++ "dentry_open(old): %p"
921 ++ "lookup_create(new): %p"
922 ++ "old path »%s«"
923 ++ "path_lookup(old): %d"
924 ++ "vfs_create(new): %d"
925 ++ "vfs_rename: %d"
926 ++ "vfs_sendfile: %d"
927 ++ 3 8 "fput(new_file=%p[#%d])"
928 ++ "fput(old_file=%p[#%d])"
929 ++ 4 10 "vx_info_kill(%p[#%d],%d,%d) = %d"
930 ++ "vx_info_kill(%p[#%d],%d,%d)*"
931 ++ 5 20 "vs_reboot(%p[#%d],%d)"
932 ++ 6 40 "dropping task %p[#%u,%u] for %p[#%u,%u]"
933 ++
934 ++debug_net:
935 ++
936 ++ 2 4 "nx_addr_conflict(%p,%p) %d.%d,%d.%d"
937 ++ 3 8 "inet_bind(%p) %d.%d.%d.%d, %d.%d.%d.%d, %d.%d.%d.%d"
938 ++ "inet_bind(%p)* %p,%p;%lx %d.%d.%d.%d"
939 ++ 4 10 "ip_route_connect(%p) %p,%p;%lx"
940 ++ 5 20 "__addr_in_socket(%p,%d.%d.%d.%d) %p:%d.%d.%d.%d %p;%lx"
941 ++ 6 40 "sk,egf: %p [#%d] (from %d)"
942 ++ "sk,egn: %p [#%d] (from %d)"
943 ++ "sk,req: %p [#%d] (from %d)"
944 ++ "sk: %p [#%d] (from %d)"
945 ++ "tw: %p [#%d] (from %d)"
946 ++ 7 80 "__sock_recvmsg: %p[%p,%p,%p;%d]:%d/%d"
947 ++ "__sock_sendmsg: %p[%p,%p,%p;%d]:%d/%d"
948 ++
949 ++debug_nid:
950 ++
951 ++ 0 1 "__lookup_nx_info(#%u): %p[#%u]"
952 ++ "alloc_nx_info(%d) = %p"
953 ++ "create_nx_info(%d) (dynamic rejected)"
954 ++ "create_nx_info(%d) = %p (already there)"
955 ++ "create_nx_info(%d) = %p (new)"
956 ++ "dealloc_nx_info(%p)"
957 ++ 1 2 "alloc_nx_info(%d)*"
958 ++ "create_nx_info(%d)*"
959 ++ 2 4 "get_nx_info(%p[#%d.%d])"
960 ++ "put_nx_info(%p[#%d.%d])"
961 ++ 3 8 "claim_nx_info(%p[#%d.%d.%d]) %p"
962 ++ "clr_nx_info(%p[#%d.%d])"
963 ++ "init_nx_info(%p[#%d.%d])"
964 ++ "release_nx_info(%p[#%d.%d.%d]) %p"
965 ++ "set_nx_info(%p[#%d.%d])"
966 ++ 4 10 "__hash_nx_info: %p[#%d]"
967 ++ "__nx_dynamic_id: [#%d]"
968 ++ "__unhash_nx_info: %p[#%d.%d.%d]"
969 ++ 5 20 "moved task %p into nxi:%p[#%d]"
970 ++ "nx_migrate_task(%p,%p[#%d.%d.%d])"
971 ++ "task_get_nx_info(%p)"
972 ++ 6 40 "nx_clear_persistent(%p[#%d])"
973 ++
974 ++debug_quota:
975 ++
976 ++ 0 1 "quota_sync_dqh(%p,%d) discard inode %p"
977 ++ 1 2 "quota_sync_dqh(%p,%d)"
978 ++ "sync_dquots(%p,%d)"
979 ++ "sync_dquots_dqh(%p,%d)"
980 ++ 3 8 "do_quotactl(%p,%d,cmd=%d,id=%d,%p)"
981 ++
982 ++debug_switch:
983 ++
984 ++ 0 1 "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]"
985 ++ 1 2 "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]"
986 ++ 4 10 "%s: (%s %s) returned %s with %d"
987 ++
988 ++debug_tag:
989 ++
990 ++ 7 80 "dx_parse_tag(»%s«): %d:#%d"
991 ++ "dx_propagate_tag(%p[#%lu.%d]): %d,%d"
992 ++
993 ++debug_xid:
994 ++
995 ++ 0 1 "__lookup_vx_info(#%u): %p[#%u]"
996 ++ "alloc_vx_info(%d) = %p"
997 ++ "alloc_vx_info(%d)*"
998 ++ "create_vx_info(%d) (dynamic rejected)"
999 ++ "create_vx_info(%d) = %p (already there)"
1000 ++ "create_vx_info(%d) = %p (new)"
1001 ++ "dealloc_vx_info(%p)"
1002 ++ "loc_vx_info(%d) = %p (found)"
1003 ++ "loc_vx_info(%d) = %p (new)"
1004 ++ "loc_vx_info(%d) = %p (not available)"
1005 ++ 1 2 "create_vx_info(%d)*"
1006 ++ "loc_vx_info(%d)*"
1007 ++ 2 4 "get_vx_info(%p[#%d.%d])"
1008 ++ "put_vx_info(%p[#%d.%d])"
1009 ++ 3 8 "claim_vx_info(%p[#%d.%d.%d]) %p"
1010 ++ "clr_vx_info(%p[#%d.%d])"
1011 ++ "init_vx_info(%p[#%d.%d])"
1012 ++ "release_vx_info(%p[#%d.%d.%d]) %p"
1013 ++ "set_vx_info(%p[#%d.%d])"
1014 ++ 4 10 "__hash_vx_info: %p[#%d]"
1015 ++ "__unhash_vx_info: %p[#%d.%d.%d]"
1016 ++ "__vx_dynamic_id: [#%d]"
1017 ++ 5 20 "enter_vx_info(%p[#%d],%p) %p[#%d,%p]"
1018 ++ "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]"
1019 ++ "moved task %p into vxi:%p[#%d]"
1020 ++ "task_get_vx_info(%p)"
1021 ++ "vx_migrate_task(%p,%p[#%d.%d])"
1022 ++ 6 40 "vx_clear_persistent(%p[#%d])"
1023 ++ "vx_exit_init(%p[#%d],%p[#%d,%d,%d])"
1024 ++ "vx_set_init(%p[#%d],%p[#%d,%d,%d])"
1025 ++ "vx_set_persistent(%p[#%d])"
1026 ++ "vx_set_reaper(%p[#%d],%p[#%d,%d])"
1027 ++ 7 80 "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]"
1028 ++
1029 ++
1030 ++debug_limit:
1031 ++
1032 ++ n 2^n "vx_acc_cres[%5d,%s,%2d]: %5d%s"
1033 ++ "vx_cres_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1034 ++
1035 ++ m 2^m "vx_acc_page[%5d,%s,%2d]: %5d%s"
1036 ++ "vx_acc_pages[%5d,%s,%2d]: %5d += %5d"
1037 ++ "vx_pages_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
1038 +diff -NurpP --minimal linux-2.6.29.4/drivers/block/Kconfig linux-2.6.29.4-vs2.3.0.36.14/drivers/block/Kconfig
1039 +--- linux-2.6.29.4/drivers/block/Kconfig 2008-12-25 00:26:37.000000000 +0100
1040 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/block/Kconfig 2009-02-22 22:54:24.000000000 +0100
1041 +@@ -264,6 +264,13 @@ config BLK_DEV_CRYPTOLOOP
1042 + instead, which can be configured to be on-disk compatible with the
1043 + cryptoloop device.
1044 +
1045 ++config BLK_DEV_VROOT
1046 ++ tristate "Virtual Root device support"
1047 ++ depends on QUOTACTL
1048 ++ ---help---
1049 ++ Saying Y here will allow you to use quota/fs ioctls on a shared
1050 ++ partition within a virtual server without compromising security.
1051 ++
1052 + config BLK_DEV_NBD
1053 + tristate "Network block device support"
1054 + depends on NET
1055 +diff -NurpP --minimal linux-2.6.29.4/drivers/block/loop.c linux-2.6.29.4-vs2.3.0.36.14/drivers/block/loop.c
1056 +--- linux-2.6.29.4/drivers/block/loop.c 2009-03-24 14:18:56.000000000 +0100
1057 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/block/loop.c 2009-03-24 15:09:29.000000000 +0100
1058 +@@ -75,6 +75,7 @@
1059 + #include <linux/gfp.h>
1060 + #include <linux/kthread.h>
1061 + #include <linux/splice.h>
1062 ++#include <linux/vs_context.h>
1063 +
1064 + #include <asm/uaccess.h>
1065 +
1066 +@@ -809,6 +810,7 @@ static int loop_set_fd(struct loop_devic
1067 + lo->lo_blocksize = lo_blocksize;
1068 + lo->lo_device = bdev;
1069 + lo->lo_flags = lo_flags;
1070 ++ lo->lo_xid = vx_current_xid();
1071 + lo->lo_backing_file = file;
1072 + lo->transfer = transfer_none;
1073 + lo->ioctl = NULL;
1074 +@@ -931,6 +933,7 @@ static int loop_clr_fd(struct loop_devic
1075 + lo->lo_encrypt_key_size = 0;
1076 + lo->lo_flags = 0;
1077 + lo->lo_thread = NULL;
1078 ++ lo->lo_xid = 0;
1079 + memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
1080 + memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
1081 + memset(lo->lo_file_name, 0, LO_NAME_SIZE);
1082 +@@ -958,7 +961,7 @@ loop_set_status(struct loop_device *lo,
1083 +
1084 + if (lo->lo_encrypt_key_size &&
1085 + lo->lo_key_owner != uid &&
1086 +- !capable(CAP_SYS_ADMIN))
1087 ++ !vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP))
1088 + return -EPERM;
1089 + if (lo->lo_state != Lo_bound)
1090 + return -ENXIO;
1091 +@@ -1042,7 +1045,8 @@ loop_get_status(struct loop_device *lo,
1092 + memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
1093 + info->lo_encrypt_type =
1094 + lo->lo_encryption ? lo->lo_encryption->number : 0;
1095 +- if (lo->lo_encrypt_key_size && capable(CAP_SYS_ADMIN)) {
1096 ++ if (lo->lo_encrypt_key_size &&
1097 ++ vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP)) {
1098 + info->lo_encrypt_key_size = lo->lo_encrypt_key_size;
1099 + memcpy(info->lo_encrypt_key, lo->lo_encrypt_key,
1100 + lo->lo_encrypt_key_size);
1101 +@@ -1351,6 +1355,9 @@ static int lo_open(struct block_device *
1102 + {
1103 + struct loop_device *lo = bdev->bd_disk->private_data;
1104 +
1105 ++ if (!vx_check(lo->lo_xid, VS_IDENT|VS_HOSTID|VS_ADMIN_P))
1106 ++ return -EACCES;
1107 ++
1108 + mutex_lock(&lo->lo_ctl_mutex);
1109 + lo->lo_refcnt++;
1110 + mutex_unlock(&lo->lo_ctl_mutex);
1111 +diff -NurpP --minimal linux-2.6.29.4/drivers/block/Makefile linux-2.6.29.4-vs2.3.0.36.14/drivers/block/Makefile
1112 +--- linux-2.6.29.4/drivers/block/Makefile 2009-03-24 14:18:55.000000000 +0100
1113 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/block/Makefile 2009-03-24 14:48:25.000000000 +0100
1114 +@@ -31,5 +31,6 @@ obj-$(CONFIG_VIODASD) += viodasd.o
1115 + obj-$(CONFIG_BLK_DEV_SX8) += sx8.o
1116 + obj-$(CONFIG_BLK_DEV_UB) += ub.o
1117 + obj-$(CONFIG_BLK_DEV_HD) += hd.o
1118 ++obj-$(CONFIG_BLK_DEV_VROOT) += vroot.o
1119 +
1120 + obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o
1121 +diff -NurpP --minimal linux-2.6.29.4/drivers/block/vroot.c linux-2.6.29.4-vs2.3.0.36.14/drivers/block/vroot.c
1122 +--- linux-2.6.29.4/drivers/block/vroot.c 1970-01-01 01:00:00.000000000 +0100
1123 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/block/vroot.c 2009-02-22 22:54:24.000000000 +0100
1124 +@@ -0,0 +1,281 @@
1125 ++/*
1126 ++ * linux/drivers/block/vroot.c
1127 ++ *
1128 ++ * written by Herbert Pötzl, 9/11/2002
1129 ++ * ported to 2.6.10 by Herbert Pötzl, 30/12/2004
1130 ++ *
1131 ++ * based on the loop.c code by Theodore Ts'o.
1132 ++ *
1133 ++ * Copyright (C) 2002-2007 by Herbert Pötzl.
1134 ++ * Redistribution of this file is permitted under the
1135 ++ * GNU General Public License.
1136 ++ *
1137 ++ */
1138 ++
1139 ++#include <linux/module.h>
1140 ++#include <linux/moduleparam.h>
1141 ++#include <linux/file.h>
1142 ++#include <linux/major.h>
1143 ++#include <linux/blkdev.h>
1144 ++
1145 ++#include <linux/vroot.h>
1146 ++#include <linux/vs_context.h>
1147 ++
1148 ++
1149 ++static int max_vroot = 8;
1150 ++
1151 ++static struct vroot_device *vroot_dev;
1152 ++static struct gendisk **disks;
1153 ++
1154 ++
1155 ++static int vroot_set_dev(
1156 ++ struct vroot_device *vr,
1157 ++ struct block_device *bdev,
1158 ++ unsigned int arg)
1159 ++{
1160 ++ struct block_device *real_bdev;
1161 ++ struct file *file;
1162 ++ struct inode *inode;
1163 ++ int error;
1164 ++
1165 ++ error = -EBUSY;
1166 ++ if (vr->vr_state != Vr_unbound)
1167 ++ goto out;
1168 ++
1169 ++ error = -EBADF;
1170 ++ file = fget(arg);
1171 ++ if (!file)
1172 ++ goto out;
1173 ++
1174 ++ error = -EINVAL;
1175 ++ inode = file->f_dentry->d_inode;
1176 ++
1177 ++
1178 ++ if (S_ISBLK(inode->i_mode)) {
1179 ++ real_bdev = inode->i_bdev;
1180 ++ vr->vr_device = real_bdev;
1181 ++ __iget(real_bdev->bd_inode);
1182 ++ } else
1183 ++ goto out_fput;
1184 ++
1185 ++ vxdprintk(VXD_CBIT(misc, 0),
1186 ++ "vroot[%d]_set_dev: dev=" VXF_DEV,
1187 ++ vr->vr_number, VXD_DEV(real_bdev));
1188 ++
1189 ++ vr->vr_state = Vr_bound;
1190 ++ error = 0;
1191 ++
1192 ++ out_fput:
1193 ++ fput(file);
1194 ++ out:
1195 ++ return error;
1196 ++}
1197 ++
1198 ++static int vroot_clr_dev(
1199 ++ struct vroot_device *vr,
1200 ++ struct block_device *bdev)
1201 ++{
1202 ++ struct block_device *real_bdev;
1203 ++
1204 ++ if (vr->vr_state != Vr_bound)
1205 ++ return -ENXIO;
1206 ++ if (vr->vr_refcnt > 1) /* we needed one fd for the ioctl */
1207 ++ return -EBUSY;
1208 ++
1209 ++ real_bdev = vr->vr_device;
1210 ++
1211 ++ vxdprintk(VXD_CBIT(misc, 0),
1212 ++ "vroot[%d]_clr_dev: dev=" VXF_DEV,
1213 ++ vr->vr_number, VXD_DEV(real_bdev));
1214 ++
1215 ++ bdput(real_bdev);
1216 ++ vr->vr_state = Vr_unbound;
1217 ++ vr->vr_device = NULL;
1218 ++ return 0;
1219 ++}
1220 ++
1221 ++
1222 ++static int vr_ioctl(struct block_device *bdev, fmode_t mode,
1223 ++ unsigned int cmd, unsigned long arg)
1224 ++{
1225 ++ struct vroot_device *vr = bdev->bd_disk->private_data;
1226 ++ int err;
1227 ++
1228 ++ down(&vr->vr_ctl_mutex);
1229 ++ switch (cmd) {
1230 ++ case VROOT_SET_DEV:
1231 ++ err = vroot_set_dev(vr, bdev, arg);
1232 ++ break;
1233 ++ case VROOT_CLR_DEV:
1234 ++ err = vroot_clr_dev(vr, bdev);
1235 ++ break;
1236 ++ default:
1237 ++ err = -EINVAL;
1238 ++ break;
1239 ++ }
1240 ++ up(&vr->vr_ctl_mutex);
1241 ++ return err;
1242 ++}
1243 ++
1244 ++static int vr_open(struct block_device *bdev, fmode_t mode)
1245 ++{
1246 ++ struct vroot_device *vr = bdev->bd_disk->private_data;
1247 ++
1248 ++ down(&vr->vr_ctl_mutex);
1249 ++ vr->vr_refcnt++;
1250 ++ up(&vr->vr_ctl_mutex);
1251 ++ return 0;
1252 ++}
1253 ++
1254 ++static int vr_release(struct gendisk *disk, fmode_t mode)
1255 ++{
1256 ++ struct vroot_device *vr = disk->private_data;
1257 ++
1258 ++ down(&vr->vr_ctl_mutex);
1259 ++ --vr->vr_refcnt;
1260 ++ up(&vr->vr_ctl_mutex);
1261 ++ return 0;
1262 ++}
1263 ++
1264 ++static struct block_device_operations vr_fops = {
1265 ++ .owner = THIS_MODULE,
1266 ++ .open = vr_open,
1267 ++ .release = vr_release,
1268 ++ .ioctl = vr_ioctl,
1269 ++};
1270 ++
1271 ++struct block_device *__vroot_get_real_bdev(struct block_device *bdev)
1272 ++{
1273 ++ struct inode *inode = bdev->bd_inode;
1274 ++ struct vroot_device *vr;
1275 ++ struct block_device *real_bdev;
1276 ++ int minor = iminor(inode);
1277 ++
1278 ++ vr = &vroot_dev[minor];
1279 ++ real_bdev = vr->vr_device;
1280 ++
1281 ++ vxdprintk(VXD_CBIT(misc, 0),
1282 ++ "vroot[%d]_get_real_bdev: dev=" VXF_DEV,
1283 ++ vr->vr_number, VXD_DEV(real_bdev));
1284 ++
1285 ++ if (vr->vr_state != Vr_bound)
1286 ++ return ERR_PTR(-ENXIO);
1287 ++
1288 ++ __iget(real_bdev->bd_inode);
1289 ++ return real_bdev;
1290 ++}
1291 ++
1292 ++/*
1293 ++ * And now the modules code and kernel interface.
1294 ++ */
1295 ++
1296 ++module_param(max_vroot, int, 0);
1297 ++
1298 ++MODULE_PARM_DESC(max_vroot, "Maximum number of vroot devices (1-256)");
1299 ++MODULE_LICENSE("GPL");
1300 ++MODULE_ALIAS_BLOCKDEV_MAJOR(VROOT_MAJOR);
1301 ++
1302 ++MODULE_AUTHOR ("Herbert Pötzl");
1303 ++MODULE_DESCRIPTION ("Virtual Root Device Mapper");
1304 ++
1305 ++
1306 ++int __init vroot_init(void)
1307 ++{
1308 ++ int err, i;
1309 ++
1310 ++ if (max_vroot < 1 || max_vroot > 256) {
1311 ++ max_vroot = MAX_VROOT_DEFAULT;
1312 ++ printk(KERN_WARNING "vroot: invalid max_vroot "
1313 ++ "(must be between 1 and 256), "
1314 ++ "using default (%d)\n", max_vroot);
1315 ++ }
1316 ++
1317 ++ if (register_blkdev(VROOT_MAJOR, "vroot"))
1318 ++ return -EIO;
1319 ++
1320 ++ err = -ENOMEM;
1321 ++ vroot_dev = kmalloc(max_vroot * sizeof(struct vroot_device), GFP_KERNEL);
1322 ++ if (!vroot_dev)
1323 ++ goto out_mem1;
1324 ++ memset(vroot_dev, 0, max_vroot * sizeof(struct vroot_device));
1325 ++
1326 ++ disks = kmalloc(max_vroot * sizeof(struct gendisk *), GFP_KERNEL);
1327 ++ if (!disks)
1328 ++ goto out_mem2;
1329 ++
1330 ++ for (i = 0; i < max_vroot; i++) {
1331 ++ disks[i] = alloc_disk(1);
1332 ++ if (!disks[i])
1333 ++ goto out_mem3;
1334 ++ disks[i]->queue = blk_alloc_queue(GFP_KERNEL);
1335 ++ if (!disks[i]->queue)
1336 ++ goto out_mem3;
1337 ++ }
1338 ++
1339 ++ for (i = 0; i < max_vroot; i++) {
1340 ++ struct vroot_device *vr = &vroot_dev[i];
1341 ++ struct gendisk *disk = disks[i];
1342 ++
1343 ++ memset(vr, 0, sizeof(*vr));
1344 ++ init_MUTEX(&vr->vr_ctl_mutex);
1345 ++ vr->vr_number = i;
1346 ++ disk->major = VROOT_MAJOR;
1347 ++ disk->first_minor = i;
1348 ++ disk->fops = &vr_fops;
1349 ++ sprintf(disk->disk_name, "vroot%d", i);
1350 ++ disk->private_data = vr;
1351 ++ }
1352 ++
1353 ++ err = register_vroot_grb(&__vroot_get_real_bdev);
1354 ++ if (err)
1355 ++ goto out_mem3;
1356 ++
1357 ++ for (i = 0; i < max_vroot; i++)
1358 ++ add_disk(disks[i]);
1359 ++ printk(KERN_INFO "vroot: loaded (max %d devices)\n", max_vroot);
1360 ++ return 0;
1361 ++
1362 ++out_mem3:
1363 ++ while (i--)
1364 ++ put_disk(disks[i]);
1365 ++ kfree(disks);
1366 ++out_mem2:
1367 ++ kfree(vroot_dev);
1368 ++out_mem1:
1369 ++ unregister_blkdev(VROOT_MAJOR, "vroot");
1370 ++ printk(KERN_ERR "vroot: ran out of memory\n");
1371 ++ return err;
1372 ++}
1373 ++
1374 ++void vroot_exit(void)
1375 ++{
1376 ++ int i;
1377 ++
1378 ++ if (unregister_vroot_grb(&__vroot_get_real_bdev))
1379 ++ printk(KERN_WARNING "vroot: cannot unregister grb\n");
1380 ++
1381 ++ for (i = 0; i < max_vroot; i++) {
1382 ++ del_gendisk(disks[i]);
1383 ++ put_disk(disks[i]);
1384 ++ }
1385 ++ unregister_blkdev(VROOT_MAJOR, "vroot");
1386 ++
1387 ++ kfree(disks);
1388 ++ kfree(vroot_dev);
1389 ++}
1390 ++
1391 ++module_init(vroot_init);
1392 ++module_exit(vroot_exit);
1393 ++
1394 ++#ifndef MODULE
1395 ++
1396 ++static int __init max_vroot_setup(char *str)
1397 ++{
1398 ++ max_vroot = simple_strtol(str, NULL, 0);
1399 ++ return 1;
1400 ++}
1401 ++
1402 ++__setup("max_vroot=", max_vroot_setup);
1403 ++
1404 ++#endif
1405 ++
1406 +diff -NurpP --minimal linux-2.6.29.4/drivers/char/sysrq.c linux-2.6.29.4-vs2.3.0.36.14/drivers/char/sysrq.c
1407 +--- linux-2.6.29.4/drivers/char/sysrq.c 2009-03-24 14:18:57.000000000 +0100
1408 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/char/sysrq.c 2009-03-24 15:15:27.000000000 +0100
1409 +@@ -38,6 +38,7 @@
1410 + #include <linux/irq.h>
1411 + #include <linux/hrtimer.h>
1412 + #include <linux/oom.h>
1413 ++#include <linux/vserver/debug.h>
1414 +
1415 + #include <asm/ptrace.h>
1416 + #include <asm/irq_regs.h>
1417 +@@ -369,6 +370,21 @@ static struct sysrq_key_op sysrq_unrt_op
1418 + .enable_mask = SYSRQ_ENABLE_RTNICE,
1419 + };
1420 +
1421 ++
1422 ++#ifdef CONFIG_VSERVER_DEBUG
1423 ++static void sysrq_handle_vxinfo(int key, struct tty_struct *tty)
1424 ++{
1425 ++ dump_vx_info_inactive((key == 'x')?0:1);
1426 ++}
1427 ++
1428 ++static struct sysrq_key_op sysrq_showvxinfo_op = {
1429 ++ .handler = sysrq_handle_vxinfo,
1430 ++ .help_msg = "conteXt",
1431 ++ .action_msg = "Show Context Info",
1432 ++ .enable_mask = SYSRQ_ENABLE_DUMP,
1433 ++};
1434 ++#endif
1435 ++
1436 + /* Key Operations table and lock */
1437 + static DEFINE_SPINLOCK(sysrq_key_table_lock);
1438 +
1439 +@@ -419,7 +435,11 @@ static struct sysrq_key_op *sysrq_key_ta
1440 + NULL, /* v */
1441 + &sysrq_showstate_blocked_op, /* w */
1442 + /* x: May be registered on ppc/powerpc for xmon */
1443 ++#ifdef CONFIG_VSERVER_DEBUG
1444 ++ &sysrq_showvxinfo_op, /* x */
1445 ++#else
1446 + NULL, /* x */
1447 ++#endif
1448 + /* y: May be registered on sparc64 for global register dump */
1449 + NULL, /* y */
1450 + &sysrq_ftrace_dump_op, /* z */
1451 +@@ -434,6 +454,8 @@ static int sysrq_key_table_key2index(int
1452 + retval = key - '0';
1453 + else if ((key >= 'a') && (key <= 'z'))
1454 + retval = key + 10 - 'a';
1455 ++ else if ((key >= 'A') && (key <= 'Z'))
1456 ++ retval = key + 10 - 'A';
1457 + else
1458 + retval = -1;
1459 + return retval;
1460 +diff -NurpP --minimal linux-2.6.29.4/drivers/char/tty_io.c linux-2.6.29.4-vs2.3.0.36.14/drivers/char/tty_io.c
1461 +--- linux-2.6.29.4/drivers/char/tty_io.c 2009-03-24 14:18:57.000000000 +0100
1462 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/char/tty_io.c 2009-03-24 14:48:25.000000000 +0100
1463 +@@ -106,6 +106,7 @@
1464 +
1465 + #include <linux/kmod.h>
1466 + #include <linux/nsproxy.h>
1467 ++#include <linux/vs_pid.h>
1468 +
1469 + #undef TTY_DEBUG_HANGUP
1470 +
1471 +@@ -2303,6 +2304,7 @@ static int tiocspgrp(struct tty_struct *
1472 + return -ENOTTY;
1473 + if (get_user(pgrp_nr, p))
1474 + return -EFAULT;
1475 ++ pgrp_nr = vx_rmap_pid(pgrp_nr);
1476 + if (pgrp_nr < 0)
1477 + return -EINVAL;
1478 + rcu_read_lock();
1479 +diff -NurpP --minimal linux-2.6.29.4/drivers/infiniband/hw/ipath/ipath_user_pages.c linux-2.6.29.4-vs2.3.0.36.14/drivers/infiniband/hw/ipath/ipath_user_pages.c
1480 +--- linux-2.6.29.4/drivers/infiniband/hw/ipath/ipath_user_pages.c 2008-12-25 00:26:37.000000000 +0100
1481 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/infiniband/hw/ipath/ipath_user_pages.c 2009-02-22 22:54:24.000000000 +0100
1482 +@@ -33,6 +33,7 @@
1483 +
1484 + #include <linux/mm.h>
1485 + #include <linux/device.h>
1486 ++#include <linux/vs_memory.h>
1487 +
1488 + #include "ipath_kernel.h"
1489 +
1490 +@@ -61,7 +62,8 @@ static int __get_user_pages(unsigned lon
1491 + lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >>
1492 + PAGE_SHIFT;
1493 +
1494 +- if (num_pages > lock_limit) {
1495 ++ if (num_pages > lock_limit ||
1496 ++ !vx_vmlocked_avail(current->mm, num_pages)) {
1497 + ret = -ENOMEM;
1498 + goto bail;
1499 + }
1500 +@@ -78,7 +80,7 @@ static int __get_user_pages(unsigned lon
1501 + goto bail_release;
1502 + }
1503 +
1504 +- current->mm->locked_vm += num_pages;
1505 ++ vx_vmlocked_add(current->mm, num_pages);
1506 +
1507 + ret = 0;
1508 + goto bail;
1509 +@@ -177,7 +179,7 @@ void ipath_release_user_pages(struct pag
1510 +
1511 + __ipath_release_user_pages(p, num_pages, 1);
1512 +
1513 +- current->mm->locked_vm -= num_pages;
1514 ++ vx_vmlocked_sub(current->mm, num_pages);
1515 +
1516 + up_write(&current->mm->mmap_sem);
1517 + }
1518 +@@ -194,7 +196,7 @@ static void user_pages_account(struct wo
1519 + container_of(_work, struct ipath_user_pages_work, work);
1520 +
1521 + down_write(&work->mm->mmap_sem);
1522 +- work->mm->locked_vm -= work->num_pages;
1523 ++ vx_vmlocked_sub(work->mm, work->num_pages);
1524 + up_write(&work->mm->mmap_sem);
1525 + mmput(work->mm);
1526 + kfree(work);
1527 +diff -NurpP --minimal linux-2.6.29.4/drivers/md/dm.c linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm.c
1528 +--- linux-2.6.29.4/drivers/md/dm.c 2009-03-24 14:19:05.000000000 +0100
1529 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm.c 2009-03-24 15:15:57.000000000 +0100
1530 +@@ -22,6 +22,7 @@
1531 + #include <linux/hdreg.h>
1532 + #include <linux/blktrace_api.h>
1533 + #include <trace/block.h>
1534 ++#include <linux/vs_base.h>
1535 +
1536 + #define DM_MSG_PREFIX "core"
1537 +
1538 +@@ -115,6 +116,7 @@ struct mapped_device {
1539 + rwlock_t map_lock;
1540 + atomic_t holders;
1541 + atomic_t open_count;
1542 ++ xid_t xid;
1543 +
1544 + unsigned long flags;
1545 +
1546 +@@ -295,6 +297,7 @@ static void __exit dm_exit(void)
1547 + static int dm_blk_open(struct block_device *bdev, fmode_t mode)
1548 + {
1549 + struct mapped_device *md;
1550 ++ int ret = -ENXIO;
1551 +
1552 + spin_lock(&_minor_lock);
1553 +
1554 +@@ -303,18 +306,19 @@ static int dm_blk_open(struct block_devi
1555 + goto out;
1556 +
1557 + if (test_bit(DMF_FREEING, &md->flags) ||
1558 +- test_bit(DMF_DELETING, &md->flags)) {
1559 +- md = NULL;
1560 ++ test_bit(DMF_DELETING, &md->flags))
1561 ++ goto out;
1562 ++
1563 ++ ret = -EACCES;
1564 ++ if (!vx_check(md->xid, VS_IDENT|VS_HOSTID))
1565 + goto out;
1566 +- }
1567 +
1568 + dm_get(md);
1569 + atomic_inc(&md->open_count);
1570 +-
1571 ++ ret = 0;
1572 + out:
1573 + spin_unlock(&_minor_lock);
1574 +-
1575 +- return md ? 0 : -ENXIO;
1576 ++ return ret;
1577 + }
1578 +
1579 + static int dm_blk_close(struct gendisk *disk, fmode_t mode)
1580 +@@ -504,6 +508,14 @@ int dm_set_geometry(struct mapped_device
1581 + return 0;
1582 + }
1583 +
1584 ++/*
1585 ++ * Get the xid associated with a dm device
1586 ++ */
1587 ++xid_t dm_get_xid(struct mapped_device *md)
1588 ++{
1589 ++ return md->xid;
1590 ++}
1591 ++
1592 + /*-----------------------------------------------------------------
1593 + * CRUD START:
1594 + * A more elegant soln is in the works that uses the queue
1595 +@@ -1110,6 +1122,7 @@ static struct mapped_device *alloc_dev(i
1596 + INIT_LIST_HEAD(&md->uevent_list);
1597 + spin_lock_init(&md->uevent_lock);
1598 +
1599 ++ md->xid = vx_current_xid();
1600 + md->queue = blk_alloc_queue(GFP_KERNEL);
1601 + if (!md->queue)
1602 + goto bad_queue;
1603 +diff -NurpP --minimal linux-2.6.29.4/drivers/md/dm.h linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm.h
1604 +--- linux-2.6.29.4/drivers/md/dm.h 2009-03-24 14:19:05.000000000 +0100
1605 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm.h 2009-03-24 14:48:26.000000000 +0100
1606 +@@ -54,6 +54,8 @@ int dm_table_any_congested(struct dm_tab
1607 + #define dm_target_is_valid(t) ((t)->table)
1608 + int dm_table_barrier_ok(struct dm_table *t);
1609 +
1610 ++xid_t dm_get_xid(struct mapped_device *md);
1611 ++
1612 + /*-----------------------------------------------------------------
1613 + * A registry of target types.
1614 + *---------------------------------------------------------------*/
1615 +diff -NurpP --minimal linux-2.6.29.4/drivers/md/dm-ioctl.c linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm-ioctl.c
1616 +--- linux-2.6.29.4/drivers/md/dm-ioctl.c 2009-03-24 14:19:05.000000000 +0100
1617 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/md/dm-ioctl.c 2009-03-24 14:48:26.000000000 +0100
1618 +@@ -16,6 +16,7 @@
1619 + #include <linux/dm-ioctl.h>
1620 + #include <linux/hdreg.h>
1621 + #include <linux/compat.h>
1622 ++#include <linux/vs_context.h>
1623 +
1624 + #include <asm/uaccess.h>
1625 +
1626 +@@ -101,7 +102,8 @@ static struct hash_cell *__get_name_cell
1627 + unsigned int h = hash_str(str);
1628 +
1629 + list_for_each_entry (hc, _name_buckets + h, name_list)
1630 +- if (!strcmp(hc->name, str)) {
1631 ++ if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1632 ++ !strcmp(hc->name, str)) {
1633 + dm_get(hc->md);
1634 + return hc;
1635 + }
1636 +@@ -115,7 +117,8 @@ static struct hash_cell *__get_uuid_cell
1637 + unsigned int h = hash_str(str);
1638 +
1639 + list_for_each_entry (hc, _uuid_buckets + h, uuid_list)
1640 +- if (!strcmp(hc->uuid, str)) {
1641 ++ if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1642 ++ !strcmp(hc->uuid, str)) {
1643 + dm_get(hc->md);
1644 + return hc;
1645 + }
1646 +@@ -352,6 +355,9 @@ typedef int (*ioctl_fn)(struct dm_ioctl
1647 +
1648 + static int remove_all(struct dm_ioctl *param, size_t param_size)
1649 + {
1650 ++ if (!vx_check(0, VS_ADMIN))
1651 ++ return -EPERM;
1652 ++
1653 + dm_hash_remove_all(1);
1654 + param->data_size = 0;
1655 + return 0;
1656 +@@ -399,6 +405,8 @@ static int list_devices(struct dm_ioctl
1657 + */
1658 + for (i = 0; i < NUM_BUCKETS; i++) {
1659 + list_for_each_entry (hc, _name_buckets + i, name_list) {
1660 ++ if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1661 ++ continue;
1662 + needed += sizeof(struct dm_name_list);
1663 + needed += strlen(hc->name) + 1;
1664 + needed += ALIGN_MASK;
1665 +@@ -422,6 +430,8 @@ static int list_devices(struct dm_ioctl
1666 + */
1667 + for (i = 0; i < NUM_BUCKETS; i++) {
1668 + list_for_each_entry (hc, _name_buckets + i, name_list) {
1669 ++ if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1670 ++ continue;
1671 + if (old_nl)
1672 + old_nl->next = (uint32_t) ((void *) nl -
1673 + (void *) old_nl);
1674 +@@ -612,10 +622,11 @@ static struct hash_cell *__find_device_h
1675 + if (!md)
1676 + goto out;
1677 +
1678 +- mdptr = dm_get_mdptr(md);
1679 ++ if (vx_check(dm_get_xid(md), VS_WATCH_P | VS_IDENT))
1680 ++ mdptr = dm_get_mdptr(md);
1681 ++
1682 + if (!mdptr)
1683 + dm_put(md);
1684 +-
1685 + out:
1686 + return mdptr;
1687 + }
1688 +@@ -1405,8 +1416,8 @@ static int ctl_ioctl(uint command, struc
1689 + ioctl_fn fn = NULL;
1690 + size_t param_size;
1691 +
1692 +- /* only root can play with this */
1693 +- if (!capable(CAP_SYS_ADMIN))
1694 ++ /* only root and certain contexts can play with this */
1695 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_MAPPER))
1696 + return -EACCES;
1697 +
1698 + if (_IOC_TYPE(command) != DM_IOCTL)
1699 +diff -NurpP --minimal linux-2.6.29.4/drivers/net/tun.c linux-2.6.29.4-vs2.3.0.36.14/drivers/net/tun.c
1700 +--- linux-2.6.29.4/drivers/net/tun.c 2009-03-24 14:19:23.000000000 +0100
1701 ++++ linux-2.6.29.4-vs2.3.0.36.14/drivers/net/tun.c 2009-03-25 01:51:59.000000000 +0100
1702 +@@ -61,6 +61,7 @@
1703 + #include <linux/crc32.h>
1704 + #include <linux/nsproxy.h>
1705 + #include <linux/virtio_net.h>
1706 ++#include <linux/vs_network.h>
1707 + #include <net/net_namespace.h>
1708 + #include <net/netns/generic.h>
1709 +
1710 +@@ -93,6 +94,7 @@ struct tun_struct {
1711 + int attached;
1712 + uid_t owner;
1713 + gid_t group;
1714 ++ nid_t nid;
1715 +
1716 + wait_queue_head_t read_wait;
1717 + struct sk_buff_head readq;
1718 +@@ -697,6 +699,7 @@ static void tun_setup(struct net_device
1719 +
1720 + tun->owner = -1;
1721 + tun->group = -1;
1722 ++ tun->nid = current->nid;
1723 +
1724 + dev->ethtool_ops = &tun_ethtool_ops;
1725 + dev->destructor = free_netdev;
1726 +@@ -727,6 +730,9 @@ static int tun_set_iff(struct net *net,
1727 + tn = net_generic(net, tun_net_id);
1728 + tun = tun_get_by_name(tn, ifr->ifr_name);
1729 + if (tun) {
1730 ++ if (!nx_check(tun->nid, VS_IDENT | VS_HOSTID | VS_ADMIN_P))
1731 ++ return -EPERM;
1732 ++
1733 + if (tun->attached)
1734 + return -EBUSY;
1735 +
1736 +@@ -735,7 +741,7 @@ static int tun_set_iff(struct net *net,
1737 + cred->euid != tun->owner) ||
1738 + (tun->group != -1 &&
1739 + cred->egid != tun->group)) &&
1740 +- !capable(CAP_NET_ADMIN)) {
1741 ++ !cap_raised(current_cap(), CAP_NET_ADMIN)) {
1742 + return -EPERM;
1743 + }
1744 + }
1745 +@@ -747,7 +753,7 @@ static int tun_set_iff(struct net *net,
1746 +
1747 + err = -EINVAL;
1748 +
1749 +- if (!capable(CAP_NET_ADMIN))
1750 ++ if (!nx_capable(CAP_NET_ADMIN, NXC_TUN_CREATE))
1751 + return -EPERM;
1752 +
1753 + /* Set dev type */
1754 +@@ -987,6 +993,16 @@ static int tun_chr_ioctl(struct inode *i
1755 + DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
1756 + break;
1757 +
1758 ++ case TUNSETNID:
1759 ++ if (!capable(CAP_CONTEXT))
1760 ++ return -EPERM;
1761 ++
1762 ++ /* Set nid owner of the device */
1763 ++ tun->nid = (nid_t) arg;
1764 ++
1765 ++ DBG(KERN_INFO "%s: nid owner set to %u\n", tun->dev->name, tun->nid);
1766 ++ break;
1767 ++
1768 + case TUNSETLINK:
1769 + /* Only allow setting the type when the interface is down */
1770 + rtnl_lock();
1771 +diff -NurpP --minimal linux-2.6.29.4/fs/attr.c linux-2.6.29.4-vs2.3.0.36.14/fs/attr.c
1772 +--- linux-2.6.29.4/fs/attr.c 2009-03-24 14:22:24.000000000 +0100
1773 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/attr.c 2009-03-24 14:48:34.000000000 +0100
1774 +@@ -14,6 +14,9 @@
1775 + #include <linux/fcntl.h>
1776 + #include <linux/quotaops.h>
1777 + #include <linux/security.h>
1778 ++#include <linux/proc_fs.h>
1779 ++#include <linux/devpts_fs.h>
1780 ++#include <linux/vs_tag.h>
1781 +
1782 + /* Taken over from the old code... */
1783 +
1784 +@@ -55,6 +58,10 @@ int inode_change_ok(struct inode *inode,
1785 + if (!is_owner_or_cap(inode))
1786 + goto error;
1787 + }
1788 ++
1789 ++ if (dx_permission(inode, MAY_WRITE))
1790 ++ goto error;
1791 ++
1792 + fine:
1793 + retval = 0;
1794 + error:
1795 +@@ -78,6 +85,8 @@ int inode_setattr(struct inode * inode,
1796 + inode->i_uid = attr->ia_uid;
1797 + if (ia_valid & ATTR_GID)
1798 + inode->i_gid = attr->ia_gid;
1799 ++ if ((ia_valid & ATTR_TAG) && IS_TAGGED(inode))
1800 ++ inode->i_tag = attr->ia_tag;
1801 + if (ia_valid & ATTR_ATIME)
1802 + inode->i_atime = timespec_trunc(attr->ia_atime,
1803 + inode->i_sb->s_time_gran);
1804 +@@ -172,7 +181,8 @@ int notify_change(struct dentry * dentry
1805 + error = inode_change_ok(inode, attr);
1806 + if (!error) {
1807 + if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
1808 +- (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid))
1809 ++ (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
1810 ++ (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag))
1811 + error = DQUOT_TRANSFER(inode, attr) ? -EDQUOT : 0;
1812 + if (!error)
1813 + error = inode_setattr(inode, attr);
1814 +diff -NurpP --minimal linux-2.6.29.4/fs/binfmt_aout.c linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_aout.c
1815 +--- linux-2.6.29.4/fs/binfmt_aout.c 2009-03-24 14:22:24.000000000 +0100
1816 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_aout.c 2009-03-24 14:48:34.000000000 +0100
1817 +@@ -24,6 +24,7 @@
1818 + #include <linux/binfmts.h>
1819 + #include <linux/personality.h>
1820 + #include <linux/init.h>
1821 ++#include <linux/vs_memory.h>
1822 +
1823 + #include <asm/system.h>
1824 + #include <asm/uaccess.h>
1825 +diff -NurpP --minimal linux-2.6.29.4/fs/binfmt_elf.c linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_elf.c
1826 +--- linux-2.6.29.4/fs/binfmt_elf.c 2009-03-24 14:22:24.000000000 +0100
1827 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_elf.c 2009-03-24 14:48:34.000000000 +0100
1828 +@@ -38,6 +38,7 @@
1829 + #include <linux/random.h>
1830 + #include <linux/elf.h>
1831 + #include <linux/utsname.h>
1832 ++#include <linux/vs_memory.h>
1833 + #include <asm/uaccess.h>
1834 + #include <asm/param.h>
1835 + #include <asm/page.h>
1836 +diff -NurpP --minimal linux-2.6.29.4/fs/binfmt_flat.c linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_flat.c
1837 +--- linux-2.6.29.4/fs/binfmt_flat.c 2009-03-24 14:22:24.000000000 +0100
1838 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_flat.c 2009-03-24 14:48:34.000000000 +0100
1839 +@@ -35,6 +35,7 @@
1840 + #include <linux/init.h>
1841 + #include <linux/flat.h>
1842 + #include <linux/syscalls.h>
1843 ++#include <linux/vs_memory.h>
1844 +
1845 + #include <asm/byteorder.h>
1846 + #include <asm/system.h>
1847 +diff -NurpP --minimal linux-2.6.29.4/fs/binfmt_som.c linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_som.c
1848 +--- linux-2.6.29.4/fs/binfmt_som.c 2009-03-24 14:22:24.000000000 +0100
1849 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/binfmt_som.c 2009-03-24 14:48:34.000000000 +0100
1850 +@@ -28,6 +28,7 @@
1851 + #include <linux/shm.h>
1852 + #include <linux/personality.h>
1853 + #include <linux/init.h>
1854 ++#include <linux/vs_memory.h>
1855 +
1856 + #include <asm/uaccess.h>
1857 + #include <asm/pgtable.h>
1858 +diff -NurpP --minimal linux-2.6.29.4/fs/block_dev.c linux-2.6.29.4-vs2.3.0.36.14/fs/block_dev.c
1859 +--- linux-2.6.29.4/fs/block_dev.c 2009-03-24 14:22:24.000000000 +0100
1860 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/block_dev.c 2009-03-24 14:48:34.000000000 +0100
1861 +@@ -24,6 +24,7 @@
1862 + #include <linux/uio.h>
1863 + #include <linux/namei.h>
1864 + #include <linux/log2.h>
1865 ++#include <linux/vs_device.h>
1866 + #include <asm/uaccess.h>
1867 + #include "internal.h"
1868 +
1869 +@@ -392,6 +393,7 @@ struct block_device *bdget(dev_t dev)
1870 + bdev->bd_invalidated = 0;
1871 + inode->i_mode = S_IFBLK;
1872 + inode->i_rdev = dev;
1873 ++ inode->i_mdev = dev;
1874 + inode->i_bdev = bdev;
1875 + inode->i_data.a_ops = &def_blk_aops;
1876 + mapping_set_gfp_mask(&inode->i_data, GFP_USER);
1877 +@@ -428,6 +430,11 @@ EXPORT_SYMBOL(bdput);
1878 + static struct block_device *bd_acquire(struct inode *inode)
1879 + {
1880 + struct block_device *bdev;
1881 ++ dev_t mdev;
1882 ++
1883 ++ if (!vs_map_blkdev(inode->i_rdev, &mdev, DATTR_OPEN))
1884 ++ return NULL;
1885 ++ inode->i_mdev = mdev;
1886 +
1887 + spin_lock(&bdev_lock);
1888 + bdev = inode->i_bdev;
1889 +@@ -438,7 +445,7 @@ static struct block_device *bd_acquire(s
1890 + }
1891 + spin_unlock(&bdev_lock);
1892 +
1893 +- bdev = bdget(inode->i_rdev);
1894 ++ bdev = bdget(mdev);
1895 + if (bdev) {
1896 + spin_lock(&bdev_lock);
1897 + if (!inode->i_bdev) {
1898 +diff -NurpP --minimal linux-2.6.29.4/fs/char_dev.c linux-2.6.29.4-vs2.3.0.36.14/fs/char_dev.c
1899 +--- linux-2.6.29.4/fs/char_dev.c 2009-03-24 14:22:25.000000000 +0100
1900 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/char_dev.c 2009-03-24 14:48:35.000000000 +0100
1901 +@@ -21,6 +21,8 @@
1902 + #include <linux/cdev.h>
1903 + #include <linux/mutex.h>
1904 + #include <linux/backing-dev.h>
1905 ++#include <linux/vs_context.h>
1906 ++#include <linux/vs_device.h>
1907 +
1908 + #include "internal.h"
1909 +
1910 +@@ -358,14 +360,21 @@ static int chrdev_open(struct inode *ino
1911 + struct cdev *p;
1912 + struct cdev *new = NULL;
1913 + int ret = 0;
1914 ++ dev_t mdev;
1915 ++
1916 ++ if (!vs_map_chrdev(inode->i_rdev, &mdev, DATTR_OPEN))
1917 ++ return -EPERM;
1918 ++ inode->i_mdev = mdev;
1919 +
1920 + spin_lock(&cdev_lock);
1921 + p = inode->i_cdev;
1922 + if (!p) {
1923 + struct kobject *kobj;
1924 + int idx;
1925 ++
1926 + spin_unlock(&cdev_lock);
1927 +- kobj = kobj_lookup(cdev_map, inode->i_rdev, &idx);
1928 ++
1929 ++ kobj = kobj_lookup(cdev_map, mdev, &idx);
1930 + if (!kobj)
1931 + return -ENXIO;
1932 + new = container_of(kobj, struct cdev, kobj);
1933 +diff -NurpP --minimal linux-2.6.29.4/fs/dcache.c linux-2.6.29.4-vs2.3.0.36.14/fs/dcache.c
1934 +--- linux-2.6.29.4/fs/dcache.c 2009-03-24 14:22:25.000000000 +0100
1935 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/dcache.c 2009-03-24 14:48:35.000000000 +0100
1936 +@@ -32,6 +32,7 @@
1937 + #include <linux/seqlock.h>
1938 + #include <linux/swap.h>
1939 + #include <linux/bootmem.h>
1940 ++#include <linux/vs_limit.h>
1941 + #include "internal.h"
1942 +
1943 + int sysctl_vfs_cache_pressure __read_mostly = 100;
1944 +@@ -229,6 +230,8 @@ repeat:
1945 + return;
1946 + }
1947 +
1948 ++ vx_dentry_dec(dentry);
1949 ++
1950 + /*
1951 + * AV: ->d_delete() is _NOT_ allowed to block now.
1952 + */
1953 +@@ -320,6 +323,7 @@ static inline struct dentry * __dget_loc
1954 + {
1955 + atomic_inc(&dentry->d_count);
1956 + dentry_lru_del_init(dentry);
1957 ++ vx_dentry_inc(dentry);
1958 + return dentry;
1959 + }
1960 +
1961 +@@ -918,6 +922,9 @@ struct dentry *d_alloc(struct dentry * p
1962 + struct dentry *dentry;
1963 + char *dname;
1964 +
1965 ++ if (!vx_dentry_avail(1))
1966 ++ return NULL;
1967 ++
1968 + dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL);
1969 + if (!dentry)
1970 + return NULL;
1971 +@@ -963,6 +970,7 @@ struct dentry *d_alloc(struct dentry * p
1972 + if (parent)
1973 + list_add(&dentry->d_u.d_child, &parent->d_subdirs);
1974 + dentry_stat.nr_dentry++;
1975 ++ vx_dentry_inc(dentry);
1976 + spin_unlock(&dcache_lock);
1977 +
1978 + return dentry;
1979 +@@ -1418,6 +1426,7 @@ struct dentry * __d_lookup(struct dentry
1980 + }
1981 +
1982 + atomic_inc(&dentry->d_count);
1983 ++ vx_dentry_inc(dentry);
1984 + found = dentry;
1985 + spin_unlock(&dentry->d_lock);
1986 + break;
1987 +diff -NurpP --minimal linux-2.6.29.4/fs/devpts/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/devpts/inode.c
1988 +--- linux-2.6.29.4/fs/devpts/inode.c 2009-03-24 14:22:25.000000000 +0100
1989 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/devpts/inode.c 2009-03-24 15:32:47.000000000 +0100
1990 +@@ -19,12 +19,12 @@
1991 + #include <linux/tty.h>
1992 + #include <linux/mutex.h>
1993 + #include <linux/idr.h>
1994 ++#include <linux/magic.h>
1995 + #include <linux/devpts_fs.h>
1996 + #include <linux/parser.h>
1997 + #include <linux/fsnotify.h>
1998 + #include <linux/seq_file.h>
1999 +-
2000 +-#define DEVPTS_SUPER_MAGIC 0x1cd1
2001 ++#include <linux/vs_base.h>
2002 +
2003 + #define DEVPTS_DEFAULT_MODE 0600
2004 + /*
2005 +@@ -36,6 +36,20 @@
2006 + #define DEVPTS_DEFAULT_PTMX_MODE 0000
2007 + #define PTMX_MINOR 2
2008 +
2009 ++static int devpts_permission(struct inode *inode, int mask)
2010 ++{
2011 ++ int ret = -EACCES;
2012 ++
2013 ++ /* devpts is xid tagged */
2014 ++ if (vx_check((xid_t)inode->i_tag, VS_WATCH_P | VS_IDENT))
2015 ++ ret = generic_permission(inode, mask, NULL);
2016 ++ return ret;
2017 ++}
2018 ++
2019 ++static struct inode_operations devpts_file_inode_operations = {
2020 ++ .permission = devpts_permission,
2021 ++};
2022 ++
2023 + extern int pty_limit; /* Config limit on Unix98 ptys */
2024 + static DEFINE_MUTEX(allocated_ptys_lock);
2025 +
2026 +@@ -254,6 +268,25 @@ static int devpts_show_options(struct se
2027 + return 0;
2028 + }
2029 +
2030 ++static int devpts_filter(struct dentry *de)
2031 ++{
2032 ++ /* devpts is xid tagged */
2033 ++ return vx_check((xid_t)de->d_inode->i_tag, VS_WATCH_P | VS_IDENT);
2034 ++}
2035 ++
2036 ++static int devpts_readdir(struct file * filp, void * dirent, filldir_t filldir)
2037 ++{
2038 ++ return dcache_readdir_filter(filp, dirent, filldir, devpts_filter);
2039 ++}
2040 ++
2041 ++static struct file_operations devpts_dir_operations = {
2042 ++ .open = dcache_dir_open,
2043 ++ .release = dcache_dir_close,
2044 ++ .llseek = dcache_dir_lseek,
2045 ++ .read = generic_read_dir,
2046 ++ .readdir = devpts_readdir,
2047 ++};
2048 ++
2049 + static const struct super_operations devpts_sops = {
2050 + .statfs = simple_statfs,
2051 + .remount_fs = devpts_remount,
2052 +@@ -293,12 +326,15 @@ devpts_fill_super(struct super_block *s,
2053 + inode = new_inode(s);
2054 + if (!inode)
2055 + goto free_fsi;
2056 ++
2057 + inode->i_ino = 1;
2058 + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2059 + inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO | S_IWUSR;
2060 + inode->i_op = &simple_dir_inode_operations;
2061 +- inode->i_fop = &simple_dir_operations;
2062 ++ inode->i_fop = &devpts_dir_operations;
2063 + inode->i_nlink = 2;
2064 ++ /* devpts is xid tagged */
2065 ++ inode->i_tag = (tag_t)vx_current_xid();
2066 +
2067 + s->s_root = d_alloc_root(inode);
2068 + if (s->s_root)
2069 +@@ -479,6 +515,7 @@ static int init_pts_mount(struct file_sy
2070 + return err;
2071 + }
2072 +
2073 ++
2074 + static int devpts_get_sb(struct file_system_type *fs_type,
2075 + int flags, const char *dev_name, void *data, struct vfsmount *mnt)
2076 + {
2077 +@@ -590,6 +627,9 @@ int devpts_pty_new(struct inode *ptmx_in
2078 + inode->i_gid = opts->setgid ? opts->gid : current_fsgid();
2079 + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2080 + init_special_inode(inode, S_IFCHR|opts->mode, device);
2081 ++ /* devpts is xid tagged */
2082 ++ inode->i_tag = (tag_t)vx_current_xid();
2083 ++ inode->i_op = &devpts_file_inode_operations;
2084 + inode->i_private = tty;
2085 + tty->driver_data = inode;
2086 +
2087 +diff -NurpP --minimal linux-2.6.29.4/fs/exec.c linux-2.6.29.4-vs2.3.0.36.14/fs/exec.c
2088 +--- linux-2.6.29.4/fs/exec.c 2009-05-23 23:16:52.000000000 +0200
2089 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/exec.c 2009-05-10 23:42:01.000000000 +0200
2090 +@@ -257,7 +257,9 @@ static int __bprm_mm_init(struct linux_b
2091 + if (err)
2092 + goto err;
2093 +
2094 +- mm->stack_vm = mm->total_vm = 1;
2095 ++ mm->total_vm = 0;
2096 ++ vx_vmpages_inc(mm);
2097 ++ mm->stack_vm = 1;
2098 + up_write(&mm->mmap_sem);
2099 + bprm->p = vma->vm_end - sizeof(void *);
2100 + return 0;
2101 +@@ -1463,7 +1465,7 @@ static int format_corename(char *corenam
2102 + /* UNIX time of coredump */
2103 + case 't': {
2104 + struct timeval tv;
2105 +- do_gettimeofday(&tv);
2106 ++ vx_gettimeofday(&tv);
2107 + rc = snprintf(out_ptr, out_end - out_ptr,
2108 + "%lu", tv.tv_sec);
2109 + if (rc > out_end - out_ptr)
2110 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/balloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/balloc.c
2111 +--- linux-2.6.29.4/fs/ext2/balloc.c 2009-03-24 14:22:25.000000000 +0100
2112 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/balloc.c 2009-03-24 14:48:35.000000000 +0100
2113 +@@ -16,6 +16,8 @@
2114 + #include <linux/sched.h>
2115 + #include <linux/buffer_head.h>
2116 + #include <linux/capability.h>
2117 ++#include <linux/vs_dlimit.h>
2118 ++#include <linux/vs_tag.h>
2119 +
2120 + /*
2121 + * balloc.c contains the blocks allocation and deallocation routines
2122 +@@ -569,6 +571,7 @@ do_more:
2123 + }
2124 + error_return:
2125 + brelse(bitmap_bh);
2126 ++ DLIMIT_FREE_BLOCK(inode, freed);
2127 + release_blocks(sb, freed);
2128 + DQUOT_FREE_BLOCK(inode, freed);
2129 + }
2130 +@@ -701,7 +704,6 @@ ext2_try_to_allocate(struct super_block
2131 + start = 0;
2132 + end = EXT2_BLOCKS_PER_GROUP(sb);
2133 + }
2134 +-
2135 + BUG_ON(start > EXT2_BLOCKS_PER_GROUP(sb));
2136 +
2137 + repeat:
2138 +@@ -1251,6 +1253,11 @@ ext2_fsblk_t ext2_new_blocks(struct inod
2139 + *errp = -EDQUOT;
2140 + return 0;
2141 + }
2142 ++ if (DLIMIT_ALLOC_BLOCK(inode, num)) {
2143 ++ *errp = -ENOSPC;
2144 ++ DQUOT_FREE_BLOCK(inode, num);
2145 ++ return 0;
2146 ++ }
2147 +
2148 + sbi = EXT2_SB(sb);
2149 + es = EXT2_SB(sb)->s_es;
2150 +@@ -1409,6 +1416,7 @@ allocated:
2151 +
2152 + *errp = 0;
2153 + brelse(bitmap_bh);
2154 ++ DLIMIT_FREE_BLOCK(inode, *count-num);
2155 + DQUOT_FREE_BLOCK(inode, *count-num);
2156 + *count = num;
2157 + return ret_block;
2158 +@@ -1419,8 +1427,10 @@ out:
2159 + /*
2160 + * Undo the block allocation
2161 + */
2162 +- if (!performed_allocation)
2163 ++ if (!performed_allocation) {
2164 ++ DLIMIT_FREE_BLOCK(inode, *count);
2165 + DQUOT_FREE_BLOCK(inode, *count);
2166 ++ }
2167 + brelse(bitmap_bh);
2168 + return 0;
2169 + }
2170 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/ext2.h linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ext2.h
2171 +--- linux-2.6.29.4/fs/ext2/ext2.h 2008-12-25 00:26:37.000000000 +0100
2172 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ext2.h 2009-02-22 22:54:25.000000000 +0100
2173 +@@ -170,6 +170,7 @@ extern const struct file_operations ext2
2174 + extern const struct address_space_operations ext2_aops;
2175 + extern const struct address_space_operations ext2_aops_xip;
2176 + extern const struct address_space_operations ext2_nobh_aops;
2177 ++extern int ext2_sync_flags(struct inode *inode);
2178 +
2179 + /* namei.c */
2180 + extern const struct inode_operations ext2_dir_inode_operations;
2181 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/file.c
2182 +--- linux-2.6.29.4/fs/ext2/file.c 2008-12-25 00:26:37.000000000 +0100
2183 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/file.c 2009-02-22 22:54:25.000000000 +0100
2184 +@@ -87,4 +87,5 @@ const struct inode_operations ext2_file_
2185 + .setattr = ext2_setattr,
2186 + .permission = ext2_permission,
2187 + .fiemap = ext2_fiemap,
2188 ++ .sync_flags = ext2_sync_flags,
2189 + };
2190 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/ialloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ialloc.c
2191 +--- linux-2.6.29.4/fs/ext2/ialloc.c 2009-03-24 14:22:25.000000000 +0100
2192 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ialloc.c 2009-03-24 15:36:12.000000000 +0100
2193 +@@ -17,6 +17,8 @@
2194 + #include <linux/backing-dev.h>
2195 + #include <linux/buffer_head.h>
2196 + #include <linux/random.h>
2197 ++#include <linux/vs_dlimit.h>
2198 ++#include <linux/vs_tag.h>
2199 + #include "ext2.h"
2200 + #include "xattr.h"
2201 + #include "acl.h"
2202 +@@ -123,6 +125,7 @@ void ext2_free_inode (struct inode * ino
2203 + ext2_xattr_delete_inode(inode);
2204 + DQUOT_FREE_INODE(inode);
2205 + DQUOT_DROP(inode);
2206 ++ DLIMIT_FREE_INODE(inode);
2207 + }
2208 +
2209 + es = EXT2_SB(sb)->s_es;
2210 +@@ -454,6 +457,11 @@ struct inode *ext2_new_inode(struct inod
2211 + if (!inode)
2212 + return ERR_PTR(-ENOMEM);
2213 +
2214 ++ inode->i_tag = dx_current_fstag(sb);
2215 ++ if (DLIMIT_ALLOC_INODE(inode)) {
2216 ++ err = -ENOSPC;
2217 ++ goto fail_dlim;
2218 ++ }
2219 + ei = EXT2_I(inode);
2220 + sbi = EXT2_SB(sb);
2221 + es = sbi->s_es;
2222 +@@ -609,6 +617,7 @@ fail_free_drop:
2223 +
2224 + fail_drop:
2225 + DQUOT_DROP(inode);
2226 ++ DLIMIT_FREE_INODE(inode);
2227 + inode->i_flags |= S_NOQUOTA;
2228 + inode->i_nlink = 0;
2229 + unlock_new_inode(inode);
2230 +@@ -616,6 +625,8 @@ fail_drop:
2231 + return ERR_PTR(err);
2232 +
2233 + fail:
2234 ++ DLIMIT_FREE_INODE(inode);
2235 ++fail_dlim:
2236 + make_bad_inode(inode);
2237 + iput(inode);
2238 + return ERR_PTR(err);
2239 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/inode.c
2240 +--- linux-2.6.29.4/fs/ext2/inode.c 2009-03-24 14:22:25.000000000 +0100
2241 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/inode.c 2009-03-24 15:41:38.000000000 +0100
2242 +@@ -33,6 +33,7 @@
2243 + #include <linux/mpage.h>
2244 + #include <linux/fiemap.h>
2245 + #include <linux/namei.h>
2246 ++#include <linux/vs_tag.h>
2247 + #include "ext2.h"
2248 + #include "acl.h"
2249 + #include "xip.h"
2250 +@@ -1018,7 +1019,7 @@ void ext2_truncate(struct inode *inode)
2251 + return;
2252 + if (ext2_inode_is_fast_symlink(inode))
2253 + return;
2254 +- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2255 ++ if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2256 + return;
2257 +
2258 + blocksize = inode->i_sb->s_blocksize;
2259 +@@ -1156,38 +1157,72 @@ void ext2_set_inode_flags(struct inode *
2260 + {
2261 + unsigned int flags = EXT2_I(inode)->i_flags;
2262 +
2263 +- inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2264 ++ inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2265 ++ S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2266 ++
2267 ++
2268 ++ if (flags & EXT2_IMMUTABLE_FL)
2269 ++ inode->i_flags |= S_IMMUTABLE;
2270 ++ if (flags & EXT2_IXUNLINK_FL)
2271 ++ inode->i_flags |= S_IXUNLINK;
2272 ++
2273 + if (flags & EXT2_SYNC_FL)
2274 + inode->i_flags |= S_SYNC;
2275 + if (flags & EXT2_APPEND_FL)
2276 + inode->i_flags |= S_APPEND;
2277 +- if (flags & EXT2_IMMUTABLE_FL)
2278 +- inode->i_flags |= S_IMMUTABLE;
2279 + if (flags & EXT2_NOATIME_FL)
2280 + inode->i_flags |= S_NOATIME;
2281 + if (flags & EXT2_DIRSYNC_FL)
2282 + inode->i_flags |= S_DIRSYNC;
2283 ++
2284 ++ inode->i_vflags &= ~(V_BARRIER | V_COW);
2285 ++
2286 ++ if (flags & EXT2_BARRIER_FL)
2287 ++ inode->i_vflags |= V_BARRIER;
2288 ++ if (flags & EXT2_COW_FL)
2289 ++ inode->i_vflags |= V_COW;
2290 + }
2291 +
2292 + /* Propagate flags from i_flags to EXT2_I(inode)->i_flags */
2293 + void ext2_get_inode_flags(struct ext2_inode_info *ei)
2294 + {
2295 + unsigned int flags = ei->vfs_inode.i_flags;
2296 ++ unsigned int vflags = ei->vfs_inode.i_vflags;
2297 ++
2298 ++ ei->i_flags &= ~(EXT2_SYNC_FL | EXT2_APPEND_FL |
2299 ++ EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL |
2300 ++ EXT2_NOATIME_FL | EXT2_DIRSYNC_FL |
2301 ++ EXT2_BARRIER_FL | EXT2_COW_FL);
2302 ++
2303 ++ if (flags & S_IMMUTABLE)
2304 ++ ei->i_flags |= EXT2_IMMUTABLE_FL;
2305 ++ if (flags & S_IXUNLINK)
2306 ++ ei->i_flags |= EXT2_IXUNLINK_FL;
2307 +
2308 +- ei->i_flags &= ~(EXT2_SYNC_FL|EXT2_APPEND_FL|
2309 +- EXT2_IMMUTABLE_FL|EXT2_NOATIME_FL|EXT2_DIRSYNC_FL);
2310 + if (flags & S_SYNC)
2311 + ei->i_flags |= EXT2_SYNC_FL;
2312 + if (flags & S_APPEND)
2313 + ei->i_flags |= EXT2_APPEND_FL;
2314 +- if (flags & S_IMMUTABLE)
2315 +- ei->i_flags |= EXT2_IMMUTABLE_FL;
2316 + if (flags & S_NOATIME)
2317 + ei->i_flags |= EXT2_NOATIME_FL;
2318 + if (flags & S_DIRSYNC)
2319 + ei->i_flags |= EXT2_DIRSYNC_FL;
2320 ++
2321 ++ if (vflags & V_BARRIER)
2322 ++ ei->i_flags |= EXT2_BARRIER_FL;
2323 ++ if (vflags & V_COW)
2324 ++ ei->i_flags |= EXT2_COW_FL;
2325 + }
2326 +
2327 ++int ext2_sync_flags(struct inode *inode)
2328 ++{
2329 ++ ext2_get_inode_flags(EXT2_I(inode));
2330 ++ inode->i_ctime = CURRENT_TIME;
2331 ++ mark_inode_dirty(inode);
2332 ++ return 0;
2333 ++}
2334 ++
2335 ++
2336 + struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
2337 + {
2338 + struct ext2_inode_info *ei;
2339 +@@ -1195,6 +1230,8 @@ struct inode *ext2_iget (struct super_bl
2340 + struct ext2_inode *raw_inode;
2341 + struct inode *inode;
2342 + long ret = -EIO;
2343 ++ uid_t uid;
2344 ++ gid_t gid;
2345 + int n;
2346 +
2347 + inode = iget_locked(sb, ino);
2348 +@@ -1217,12 +1254,17 @@ struct inode *ext2_iget (struct super_bl
2349 + }
2350 +
2351 + inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2352 +- inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2353 +- inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2354 ++ uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2355 ++ gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2356 + if (!(test_opt (inode->i_sb, NO_UID32))) {
2357 +- inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2358 +- inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2359 ++ uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2360 ++ gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2361 + }
2362 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2363 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2364 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2365 ++ le16_to_cpu(raw_inode->i_raw_tag));
2366 ++
2367 + inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2368 + inode->i_size = le32_to_cpu(raw_inode->i_size);
2369 + inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2370 +@@ -1320,8 +1362,8 @@ static int ext2_update_inode(struct inod
2371 + struct ext2_inode_info *ei = EXT2_I(inode);
2372 + struct super_block *sb = inode->i_sb;
2373 + ino_t ino = inode->i_ino;
2374 +- uid_t uid = inode->i_uid;
2375 +- gid_t gid = inode->i_gid;
2376 ++ uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2377 ++ gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2378 + struct buffer_head * bh;
2379 + struct ext2_inode * raw_inode = ext2_get_inode(sb, ino, &bh);
2380 + int n;
2381 +@@ -1357,6 +1399,9 @@ static int ext2_update_inode(struct inod
2382 + raw_inode->i_uid_high = 0;
2383 + raw_inode->i_gid_high = 0;
2384 + }
2385 ++#ifdef CONFIG_TAGGING_INTERN
2386 ++ raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2387 ++#endif
2388 + raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2389 + raw_inode->i_size = cpu_to_le32(inode->i_size);
2390 + raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2391 +@@ -1443,7 +1488,8 @@ int ext2_setattr(struct dentry *dentry,
2392 + if (error)
2393 + return error;
2394 + if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
2395 +- (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
2396 ++ (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
2397 ++ (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
2398 + error = DQUOT_TRANSFER(inode, iattr) ? -EDQUOT : 0;
2399 + if (error)
2400 + return error;
2401 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ioctl.c
2402 +--- linux-2.6.29.4/fs/ext2/ioctl.c 2009-03-24 14:22:25.000000000 +0100
2403 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/ioctl.c 2009-03-24 14:48:35.000000000 +0100
2404 +@@ -14,6 +14,7 @@
2405 + #include <linux/compat.h>
2406 + #include <linux/mount.h>
2407 + #include <linux/smp_lock.h>
2408 ++#include <linux/mount.h>
2409 + #include <asm/current.h>
2410 + #include <asm/uaccess.h>
2411 +
2412 +@@ -52,6 +53,11 @@ long ext2_ioctl(struct file *filp, unsig
2413 +
2414 + flags = ext2_mask_flags(inode->i_mode, flags);
2415 +
2416 ++ if (IS_BARRIER(inode)) {
2417 ++ vxwprintk_task(1, "messing with the barrier.");
2418 ++ return -EACCES;
2419 ++ }
2420 ++
2421 + mutex_lock(&inode->i_mutex);
2422 + /* Is it quota file? Do not allow user to mess with it */
2423 + if (IS_NOQUOTA(inode)) {
2424 +@@ -67,7 +73,9 @@ long ext2_ioctl(struct file *filp, unsig
2425 + *
2426 + * This test looks nicer. Thanks to Pauline Middelink
2427 + */
2428 +- if ((flags ^ oldflags) & (EXT2_APPEND_FL | EXT2_IMMUTABLE_FL)) {
2429 ++ if ((oldflags & EXT2_IMMUTABLE_FL) ||
2430 ++ ((flags ^ oldflags) & (EXT2_APPEND_FL |
2431 ++ EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL))) {
2432 + if (!capable(CAP_LINUX_IMMUTABLE)) {
2433 + mutex_unlock(&inode->i_mutex);
2434 + ret = -EPERM;
2435 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/namei.c
2436 +--- linux-2.6.29.4/fs/ext2/namei.c 2009-03-24 14:22:25.000000000 +0100
2437 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/namei.c 2009-03-24 14:48:35.000000000 +0100
2438 +@@ -31,6 +31,7 @@
2439 + */
2440 +
2441 + #include <linux/pagemap.h>
2442 ++#include <linux/vs_tag.h>
2443 + #include "ext2.h"
2444 + #include "xattr.h"
2445 + #include "acl.h"
2446 +@@ -68,6 +69,7 @@ static struct dentry *ext2_lookup(struct
2447 + inode = ext2_iget(dir->i_sb, ino);
2448 + if (IS_ERR(inode))
2449 + return ERR_CAST(inode);
2450 ++ dx_propagate_tag(nd, inode);
2451 + }
2452 + return d_splice_alias(inode, dentry);
2453 + }
2454 +@@ -388,6 +390,7 @@ const struct inode_operations ext2_dir_i
2455 + #endif
2456 + .setattr = ext2_setattr,
2457 + .permission = ext2_permission,
2458 ++ .sync_flags = ext2_sync_flags,
2459 + };
2460 +
2461 + const struct inode_operations ext2_special_inode_operations = {
2462 +@@ -399,4 +402,5 @@ const struct inode_operations ext2_speci
2463 + #endif
2464 + .setattr = ext2_setattr,
2465 + .permission = ext2_permission,
2466 ++ .sync_flags = ext2_sync_flags,
2467 + };
2468 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/super.c
2469 +--- linux-2.6.29.4/fs/ext2/super.c 2009-03-24 14:22:25.000000000 +0100
2470 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/super.c 2009-03-24 14:48:35.000000000 +0100
2471 +@@ -391,7 +391,8 @@ enum {
2472 + Opt_err_ro, Opt_nouid32, Opt_nocheck, Opt_debug,
2473 + Opt_oldalloc, Opt_orlov, Opt_nobh, Opt_user_xattr, Opt_nouser_xattr,
2474 + Opt_acl, Opt_noacl, Opt_xip, Opt_ignore, Opt_err, Opt_quota,
2475 +- Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation
2476 ++ Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation,
2477 ++ Opt_tag, Opt_notag, Opt_tagid
2478 + };
2479 +
2480 + static const match_table_t tokens = {
2481 +@@ -419,6 +420,9 @@ static const match_table_t tokens = {
2482 + {Opt_acl, "acl"},
2483 + {Opt_noacl, "noacl"},
2484 + {Opt_xip, "xip"},
2485 ++ {Opt_tag, "tag"},
2486 ++ {Opt_notag, "notag"},
2487 ++ {Opt_tagid, "tagid=%u"},
2488 + {Opt_grpquota, "grpquota"},
2489 + {Opt_ignore, "noquota"},
2490 + {Opt_quota, "quota"},
2491 +@@ -489,6 +493,20 @@ static int parse_options (char * options
2492 + case Opt_nouid32:
2493 + set_opt (sbi->s_mount_opt, NO_UID32);
2494 + break;
2495 ++#ifndef CONFIG_TAGGING_NONE
2496 ++ case Opt_tag:
2497 ++ set_opt (sbi->s_mount_opt, TAGGED);
2498 ++ break;
2499 ++ case Opt_notag:
2500 ++ clear_opt (sbi->s_mount_opt, TAGGED);
2501 ++ break;
2502 ++#endif
2503 ++#ifdef CONFIG_PROPAGATE
2504 ++ case Opt_tagid:
2505 ++ /* use args[0] */
2506 ++ set_opt (sbi->s_mount_opt, TAGGED);
2507 ++ break;
2508 ++#endif
2509 + case Opt_nocheck:
2510 + clear_opt (sbi->s_mount_opt, CHECK);
2511 + break;
2512 +@@ -838,6 +856,8 @@ static int ext2_fill_super(struct super_
2513 + if (!parse_options ((char *) data, sbi))
2514 + goto failed_mount;
2515 +
2516 ++ if (EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_TAGGED)
2517 ++ sb->s_flags |= MS_TAGGED;
2518 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
2519 + ((EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ?
2520 + MS_POSIXACL : 0);
2521 +@@ -1170,6 +1190,13 @@ static int ext2_remount (struct super_bl
2522 + goto restore_opts;
2523 + }
2524 +
2525 ++ if ((sbi->s_mount_opt & EXT2_MOUNT_TAGGED) &&
2526 ++ !(sb->s_flags & MS_TAGGED)) {
2527 ++ printk("EXT2-fs: %s: tagging not permitted on remount.\n",
2528 ++ sb->s_id);
2529 ++ return -EINVAL;
2530 ++ }
2531 ++
2532 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
2533 + ((sbi->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
2534 +
2535 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/symlink.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/symlink.c
2536 +--- linux-2.6.29.4/fs/ext2/symlink.c 2008-12-25 00:26:37.000000000 +0100
2537 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/symlink.c 2009-02-22 22:54:25.000000000 +0100
2538 +@@ -38,6 +38,7 @@ const struct inode_operations ext2_symli
2539 + .listxattr = ext2_listxattr,
2540 + .removexattr = generic_removexattr,
2541 + #endif
2542 ++ .sync_flags = ext2_sync_flags,
2543 + };
2544 +
2545 + const struct inode_operations ext2_fast_symlink_inode_operations = {
2546 +@@ -49,4 +50,5 @@ const struct inode_operations ext2_fast_
2547 + .listxattr = ext2_listxattr,
2548 + .removexattr = generic_removexattr,
2549 + #endif
2550 ++ .sync_flags = ext2_sync_flags,
2551 + };
2552 +diff -NurpP --minimal linux-2.6.29.4/fs/ext2/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/xattr.c
2553 +--- linux-2.6.29.4/fs/ext2/xattr.c 2008-12-25 00:26:37.000000000 +0100
2554 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext2/xattr.c 2009-02-22 22:54:25.000000000 +0100
2555 +@@ -60,6 +60,7 @@
2556 + #include <linux/mbcache.h>
2557 + #include <linux/quotaops.h>
2558 + #include <linux/rwsem.h>
2559 ++#include <linux/vs_dlimit.h>
2560 + #include "ext2.h"
2561 + #include "xattr.h"
2562 + #include "acl.h"
2563 +@@ -641,8 +642,12 @@ ext2_xattr_set2(struct inode *inode, str
2564 + the inode. */
2565 + ea_bdebug(new_bh, "reusing block");
2566 +
2567 ++ error = -ENOSPC;
2568 ++ if (DLIMIT_ALLOC_BLOCK(inode, 1))
2569 ++ goto cleanup;
2570 + error = -EDQUOT;
2571 + if (DQUOT_ALLOC_BLOCK(inode, 1)) {
2572 ++ DLIMIT_FREE_BLOCK(inode, 1);
2573 + unlock_buffer(new_bh);
2574 + goto cleanup;
2575 + }
2576 +@@ -731,6 +736,7 @@ ext2_xattr_set2(struct inode *inode, str
2577 + le32_add_cpu(&HDR(old_bh)->h_refcount, -1);
2578 + if (ce)
2579 + mb_cache_entry_release(ce);
2580 ++ DLIMIT_FREE_BLOCK(inode, 1);
2581 + DQUOT_FREE_BLOCK(inode, 1);
2582 + mark_buffer_dirty(old_bh);
2583 + ea_bdebug(old_bh, "refcount now=%d",
2584 +@@ -794,6 +800,7 @@ ext2_xattr_delete_inode(struct inode *in
2585 + mark_buffer_dirty(bh);
2586 + if (IS_SYNC(inode))
2587 + sync_dirty_buffer(bh);
2588 ++ DLIMIT_FREE_BLOCK(inode, 1);
2589 + DQUOT_FREE_BLOCK(inode, 1);
2590 + }
2591 + EXT2_I(inode)->i_file_acl = 0;
2592 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/balloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/balloc.c
2593 +--- linux-2.6.29.4/fs/ext3/balloc.c 2009-03-24 14:22:25.000000000 +0100
2594 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/balloc.c 2009-03-25 00:53:20.000000000 +0100
2595 +@@ -19,6 +19,8 @@
2596 + #include <linux/ext3_jbd.h>
2597 + #include <linux/quotaops.h>
2598 + #include <linux/buffer_head.h>
2599 ++#include <linux/vs_dlimit.h>
2600 ++#include <linux/vs_tag.h>
2601 +
2602 + /*
2603 + * balloc.c contains the blocks allocation and deallocation routines
2604 +@@ -675,8 +677,10 @@ void ext3_free_blocks(handle_t *handle,
2605 + return;
2606 + }
2607 + ext3_free_blocks_sb(handle, sb, block, count, &dquot_freed_blocks);
2608 +- if (dquot_freed_blocks)
2609 ++ if (dquot_freed_blocks) {
2610 ++ DLIMIT_FREE_BLOCK(inode, dquot_freed_blocks);
2611 + DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
2612 ++ }
2613 + return;
2614 + }
2615 +
2616 +@@ -1415,18 +1419,33 @@ out:
2617 + *
2618 + * Check if filesystem has at least 1 free block available for allocation.
2619 + */
2620 +-static int ext3_has_free_blocks(struct ext3_sb_info *sbi)
2621 ++static int ext3_has_free_blocks(struct super_block *sb)
2622 + {
2623 +- ext3_fsblk_t free_blocks, root_blocks;
2624 ++ struct ext3_sb_info *sbi = EXT3_SB(sb);
2625 ++ unsigned long long free_blocks, root_blocks;
2626 ++ int cond;
2627 +
2628 + free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
2629 + root_blocks = le32_to_cpu(sbi->s_es->s_r_blocks_count);
2630 +- if (free_blocks < root_blocks + 1 && !capable(CAP_SYS_RESOURCE) &&
2631 ++
2632 ++ vxdprintk(VXD_CBIT(dlim, 3),
2633 ++ "ext3_has_free_blocks(%p): free=%llu, root=%llu",
2634 ++ sb, free_blocks, root_blocks);
2635 ++
2636 ++ DLIMIT_ADJUST_BLOCK(sb, dx_current_tag(), &free_blocks, &root_blocks);
2637 ++
2638 ++ cond = (free_blocks < root_blocks + 1 &&
2639 ++ !capable(CAP_SYS_RESOURCE) &&
2640 + sbi->s_resuid != current_fsuid() &&
2641 +- (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))) {
2642 +- return 0;
2643 +- }
2644 +- return 1;
2645 ++ (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid)));
2646 ++
2647 ++ vxdprintk(VXD_CBIT(dlim, 3),
2648 ++ "ext3_has_free_blocks(%p): %llu<%llu+1, %c, %u!=%u r=%d",
2649 ++ sb, free_blocks, root_blocks,
2650 ++ !capable(CAP_SYS_RESOURCE)?'1':'0',
2651 ++ sbi->s_resuid, current_fsuid(), cond?0:1);
2652 ++
2653 ++ return (cond ? 0 : 1);
2654 + }
2655 +
2656 + /**
2657 +@@ -1443,7 +1462,7 @@ static int ext3_has_free_blocks(struct e
2658 + */
2659 + int ext3_should_retry_alloc(struct super_block *sb, int *retries)
2660 + {
2661 +- if (!ext3_has_free_blocks(EXT3_SB(sb)) || (*retries)++ > 3)
2662 ++ if (!ext3_has_free_blocks(sb) || (*retries)++ > 3)
2663 + return 0;
2664 +
2665 + jbd_debug(1, "%s: retrying operation after ENOSPC\n", sb->s_id);
2666 +@@ -1506,6 +1525,8 @@ ext3_fsblk_t ext3_new_blocks(handle_t *h
2667 + *errp = -EDQUOT;
2668 + return 0;
2669 + }
2670 ++ if (DLIMIT_ALLOC_BLOCK(inode, num))
2671 ++ goto out_dlimit;
2672 +
2673 + sbi = EXT3_SB(sb);
2674 + es = EXT3_SB(sb)->s_es;
2675 +@@ -1522,7 +1543,7 @@ ext3_fsblk_t ext3_new_blocks(handle_t *h
2676 + if (block_i && ((windowsz = block_i->rsv_window_node.rsv_goal_size) > 0))
2677 + my_rsv = &block_i->rsv_window_node;
2678 +
2679 +- if (!ext3_has_free_blocks(sbi)) {
2680 ++ if (!ext3_has_free_blocks(sb)) {
2681 + *errp = -ENOSPC;
2682 + goto out;
2683 + }
2684 +@@ -1715,12 +1736,16 @@ allocated:
2685 + *errp = 0;
2686 + brelse(bitmap_bh);
2687 + DQUOT_FREE_BLOCK(inode, *count-num);
2688 ++ DLIMIT_FREE_BLOCK(inode, *count-num);
2689 + *count = num;
2690 + return ret_block;
2691 +
2692 + io_error:
2693 + *errp = -EIO;
2694 + out:
2695 ++ if (!performed_allocation)
2696 ++ DLIMIT_FREE_BLOCK(inode, *count);
2697 ++out_dlimit:
2698 + if (fatal) {
2699 + *errp = fatal;
2700 + ext3_std_error(sb, fatal);
2701 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/file.c
2702 +--- linux-2.6.29.4/fs/ext3/file.c 2008-12-25 00:26:37.000000000 +0100
2703 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/file.c 2009-02-22 22:54:25.000000000 +0100
2704 +@@ -135,5 +135,6 @@ const struct inode_operations ext3_file_
2705 + #endif
2706 + .permission = ext3_permission,
2707 + .fiemap = ext3_fiemap,
2708 ++ .sync_flags = ext3_sync_flags,
2709 + };
2710 +
2711 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/ialloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/ialloc.c
2712 +--- linux-2.6.29.4/fs/ext3/ialloc.c 2009-03-24 14:22:25.000000000 +0100
2713 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/ialloc.c 2009-03-24 15:44:06.000000000 +0100
2714 +@@ -23,6 +23,8 @@
2715 + #include <linux/buffer_head.h>
2716 + #include <linux/random.h>
2717 + #include <linux/bitops.h>
2718 ++#include <linux/vs_dlimit.h>
2719 ++#include <linux/vs_tag.h>
2720 +
2721 + #include <asm/byteorder.h>
2722 +
2723 +@@ -127,6 +129,7 @@ void ext3_free_inode (handle_t *handle,
2724 + ext3_xattr_delete_inode(handle, inode);
2725 + DQUOT_FREE_INODE(inode);
2726 + DQUOT_DROP(inode);
2727 ++ DLIMIT_FREE_INODE(inode);
2728 +
2729 + is_directory = S_ISDIR(inode->i_mode);
2730 +
2731 +@@ -440,6 +443,12 @@ struct inode *ext3_new_inode(handle_t *h
2732 + inode = new_inode(sb);
2733 + if (!inode)
2734 + return ERR_PTR(-ENOMEM);
2735 ++
2736 ++ inode->i_tag = dx_current_fstag(sb);
2737 ++ if (DLIMIT_ALLOC_INODE(inode)) {
2738 ++ err = -ENOSPC;
2739 ++ goto out_dlimit;
2740 ++ }
2741 + ei = EXT3_I(inode);
2742 +
2743 + sbi = EXT3_SB(sb);
2744 +@@ -613,6 +622,8 @@ got:
2745 + fail:
2746 + ext3_std_error(sb, err);
2747 + out:
2748 ++ DLIMIT_FREE_INODE(inode);
2749 ++out_dlimit:
2750 + iput(inode);
2751 + ret = ERR_PTR(err);
2752 + really_out:
2753 +@@ -624,6 +635,7 @@ fail_free_drop:
2754 +
2755 + fail_drop:
2756 + DQUOT_DROP(inode);
2757 ++ DLIMIT_FREE_INODE(inode);
2758 + inode->i_flags |= S_NOQUOTA;
2759 + inode->i_nlink = 0;
2760 + unlock_new_inode(inode);
2761 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/inode.c
2762 +--- linux-2.6.29.4/fs/ext3/inode.c 2009-03-24 14:22:25.000000000 +0100
2763 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/inode.c 2009-03-24 15:44:29.000000000 +0100
2764 +@@ -38,6 +38,7 @@
2765 + #include <linux/bio.h>
2766 + #include <linux/fiemap.h>
2767 + #include <linux/namei.h>
2768 ++#include <linux/vs_tag.h>
2769 + #include "xattr.h"
2770 + #include "acl.h"
2771 +
2772 +@@ -2288,7 +2289,7 @@ static void ext3_free_branches(handle_t
2773 +
2774 + int ext3_can_truncate(struct inode *inode)
2775 + {
2776 +- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2777 ++ if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2778 + return 0;
2779 + if (S_ISREG(inode->i_mode))
2780 + return 1;
2781 +@@ -2662,36 +2663,84 @@ void ext3_set_inode_flags(struct inode *
2782 + {
2783 + unsigned int flags = EXT3_I(inode)->i_flags;
2784 +
2785 +- inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2786 ++ inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2787 ++ S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2788 ++
2789 ++ if (flags & EXT3_IMMUTABLE_FL)
2790 ++ inode->i_flags |= S_IMMUTABLE;
2791 ++ if (flags & EXT3_IXUNLINK_FL)
2792 ++ inode->i_flags |= S_IXUNLINK;
2793 ++
2794 + if (flags & EXT3_SYNC_FL)
2795 + inode->i_flags |= S_SYNC;
2796 + if (flags & EXT3_APPEND_FL)
2797 + inode->i_flags |= S_APPEND;
2798 +- if (flags & EXT3_IMMUTABLE_FL)
2799 +- inode->i_flags |= S_IMMUTABLE;
2800 + if (flags & EXT3_NOATIME_FL)
2801 + inode->i_flags |= S_NOATIME;
2802 + if (flags & EXT3_DIRSYNC_FL)
2803 + inode->i_flags |= S_DIRSYNC;
2804 ++
2805 ++ inode->i_vflags &= ~(V_BARRIER | V_COW);
2806 ++
2807 ++ if (flags & EXT3_BARRIER_FL)
2808 ++ inode->i_vflags |= V_BARRIER;
2809 ++ if (flags & EXT3_COW_FL)
2810 ++ inode->i_vflags |= V_COW;
2811 + }
2812 +
2813 + /* Propagate flags from i_flags to EXT3_I(inode)->i_flags */
2814 + void ext3_get_inode_flags(struct ext3_inode_info *ei)
2815 + {
2816 + unsigned int flags = ei->vfs_inode.i_flags;
2817 ++ unsigned int vflags = ei->vfs_inode.i_vflags;
2818 ++
2819 ++ ei->i_flags &= ~(EXT3_SYNC_FL | EXT3_APPEND_FL |
2820 ++ EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL |
2821 ++ EXT3_NOATIME_FL | EXT3_DIRSYNC_FL |
2822 ++ EXT3_BARRIER_FL | EXT3_COW_FL);
2823 ++
2824 ++ if (flags & S_IMMUTABLE)
2825 ++ ei->i_flags |= EXT3_IMMUTABLE_FL;
2826 ++ if (flags & S_IXUNLINK)
2827 ++ ei->i_flags |= EXT3_IXUNLINK_FL;
2828 +
2829 +- ei->i_flags &= ~(EXT3_SYNC_FL|EXT3_APPEND_FL|
2830 +- EXT3_IMMUTABLE_FL|EXT3_NOATIME_FL|EXT3_DIRSYNC_FL);
2831 + if (flags & S_SYNC)
2832 + ei->i_flags |= EXT3_SYNC_FL;
2833 + if (flags & S_APPEND)
2834 + ei->i_flags |= EXT3_APPEND_FL;
2835 +- if (flags & S_IMMUTABLE)
2836 +- ei->i_flags |= EXT3_IMMUTABLE_FL;
2837 + if (flags & S_NOATIME)
2838 + ei->i_flags |= EXT3_NOATIME_FL;
2839 + if (flags & S_DIRSYNC)
2840 + ei->i_flags |= EXT3_DIRSYNC_FL;
2841 ++
2842 ++ if (vflags & V_BARRIER)
2843 ++ ei->i_flags |= EXT3_BARRIER_FL;
2844 ++ if (vflags & V_COW)
2845 ++ ei->i_flags |= EXT3_COW_FL;
2846 ++}
2847 ++
2848 ++int ext3_sync_flags(struct inode *inode)
2849 ++{
2850 ++ struct ext3_iloc iloc;
2851 ++ handle_t *handle;
2852 ++ int err;
2853 ++
2854 ++ handle = ext3_journal_start(inode, 1);
2855 ++ if (IS_ERR(handle))
2856 ++ return PTR_ERR(handle);
2857 ++ if (IS_SYNC(inode))
2858 ++ handle->h_sync = 1;
2859 ++ err = ext3_reserve_inode_write(handle, inode, &iloc);
2860 ++ if (err)
2861 ++ goto flags_err;
2862 ++
2863 ++ ext3_get_inode_flags(EXT3_I(inode));
2864 ++ inode->i_ctime = CURRENT_TIME;
2865 ++
2866 ++ err = ext3_mark_iloc_dirty(handle, inode, &iloc);
2867 ++flags_err:
2868 ++ ext3_journal_stop(handle);
2869 ++ return err;
2870 + }
2871 +
2872 + struct inode *ext3_iget(struct super_block *sb, unsigned long ino)
2873 +@@ -2703,6 +2752,8 @@ struct inode *ext3_iget(struct super_blo
2874 + struct inode *inode;
2875 + long ret;
2876 + int block;
2877 ++ uid_t uid;
2878 ++ gid_t gid;
2879 +
2880 + inode = iget_locked(sb, ino);
2881 + if (!inode)
2882 +@@ -2723,12 +2774,17 @@ struct inode *ext3_iget(struct super_blo
2883 + bh = iloc.bh;
2884 + raw_inode = ext3_raw_inode(&iloc);
2885 + inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2886 +- inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2887 +- inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2888 ++ uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2889 ++ gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2890 + if(!(test_opt (inode->i_sb, NO_UID32))) {
2891 +- inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2892 +- inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2893 ++ uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2894 ++ gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2895 + }
2896 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2897 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2898 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2899 ++ le16_to_cpu(raw_inode->i_raw_tag));
2900 ++
2901 + inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2902 + inode->i_size = le32_to_cpu(raw_inode->i_size);
2903 + inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2904 +@@ -2859,6 +2915,8 @@ static int ext3_do_update_inode(handle_t
2905 + struct ext3_inode *raw_inode = ext3_raw_inode(iloc);
2906 + struct ext3_inode_info *ei = EXT3_I(inode);
2907 + struct buffer_head *bh = iloc->bh;
2908 ++ uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2909 ++ gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2910 + int err = 0, rc, block;
2911 +
2912 + /* For fields not not tracking in the in-memory inode,
2913 +@@ -2869,29 +2927,32 @@ static int ext3_do_update_inode(handle_t
2914 + ext3_get_inode_flags(ei);
2915 + raw_inode->i_mode = cpu_to_le16(inode->i_mode);
2916 + if(!(test_opt(inode->i_sb, NO_UID32))) {
2917 +- raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
2918 +- raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
2919 ++ raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
2920 ++ raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
2921 + /*
2922 + * Fix up interoperability with old kernels. Otherwise, old inodes get
2923 + * re-used with the upper 16 bits of the uid/gid intact
2924 + */
2925 + if(!ei->i_dtime) {
2926 + raw_inode->i_uid_high =
2927 +- cpu_to_le16(high_16_bits(inode->i_uid));
2928 ++ cpu_to_le16(high_16_bits(uid));
2929 + raw_inode->i_gid_high =
2930 +- cpu_to_le16(high_16_bits(inode->i_gid));
2931 ++ cpu_to_le16(high_16_bits(gid));
2932 + } else {
2933 + raw_inode->i_uid_high = 0;
2934 + raw_inode->i_gid_high = 0;
2935 + }
2936 + } else {
2937 + raw_inode->i_uid_low =
2938 +- cpu_to_le16(fs_high2lowuid(inode->i_uid));
2939 ++ cpu_to_le16(fs_high2lowuid(uid));
2940 + raw_inode->i_gid_low =
2941 +- cpu_to_le16(fs_high2lowgid(inode->i_gid));
2942 ++ cpu_to_le16(fs_high2lowgid(gid));
2943 + raw_inode->i_uid_high = 0;
2944 + raw_inode->i_gid_high = 0;
2945 + }
2946 ++#ifdef CONFIG_TAGGING_INTERN
2947 ++ raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2948 ++#endif
2949 + raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2950 + raw_inode->i_size = cpu_to_le32(ei->i_disksize);
2951 + raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2952 +@@ -3044,7 +3105,8 @@ int ext3_setattr(struct dentry *dentry,
2953 + return error;
2954 +
2955 + if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
2956 +- (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
2957 ++ (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
2958 ++ (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
2959 + handle_t *handle;
2960 +
2961 + /* (user+group)*(old+new) structure, inode write (sb,
2962 +@@ -3066,6 +3128,8 @@ int ext3_setattr(struct dentry *dentry,
2963 + inode->i_uid = attr->ia_uid;
2964 + if (attr->ia_valid & ATTR_GID)
2965 + inode->i_gid = attr->ia_gid;
2966 ++ if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
2967 ++ inode->i_tag = attr->ia_tag;
2968 + error = ext3_mark_inode_dirty(handle, inode);
2969 + ext3_journal_stop(handle);
2970 + }
2971 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/ioctl.c
2972 +--- linux-2.6.29.4/fs/ext3/ioctl.c 2009-03-24 14:22:25.000000000 +0100
2973 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/ioctl.c 2009-03-24 14:48:35.000000000 +0100
2974 +@@ -8,6 +8,7 @@
2975 + */
2976 +
2977 + #include <linux/fs.h>
2978 ++#include <linux/mount.h>
2979 + #include <linux/jbd.h>
2980 + #include <linux/capability.h>
2981 + #include <linux/ext3_fs.h>
2982 +@@ -16,6 +17,7 @@
2983 + #include <linux/time.h>
2984 + #include <linux/compat.h>
2985 + #include <linux/smp_lock.h>
2986 ++#include <linux/vs_tag.h>
2987 + #include <asm/uaccess.h>
2988 +
2989 + int ext3_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
2990 +@@ -55,6 +57,11 @@ int ext3_ioctl (struct inode * inode, st
2991 +
2992 + flags = ext3_mask_flags(inode->i_mode, flags);
2993 +
2994 ++ if (IS_BARRIER(inode)) {
2995 ++ vxwprintk_task(1, "messing with the barrier.");
2996 ++ return -EACCES;
2997 ++ }
2998 ++
2999 + mutex_lock(&inode->i_mutex);
3000 + /* Is it quota file? Do not allow user to mess with it */
3001 + if (IS_NOQUOTA(inode)) {
3002 +@@ -73,7 +80,9 @@ int ext3_ioctl (struct inode * inode, st
3003 + *
3004 + * This test looks nicer. Thanks to Pauline Middelink
3005 + */
3006 +- if ((flags ^ oldflags) & (EXT3_APPEND_FL | EXT3_IMMUTABLE_FL)) {
3007 ++ if ((oldflags & EXT3_IMMUTABLE_FL) ||
3008 ++ ((flags ^ oldflags) & (EXT3_APPEND_FL |
3009 ++ EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL))) {
3010 + if (!capable(CAP_LINUX_IMMUTABLE)) {
3011 + mutex_unlock(&inode->i_mutex);
3012 + err = -EPERM;
3013 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/namei.c
3014 +--- linux-2.6.29.4/fs/ext3/namei.c 2009-03-24 14:22:25.000000000 +0100
3015 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/namei.c 2009-03-24 14:48:35.000000000 +0100
3016 +@@ -36,6 +36,7 @@
3017 + #include <linux/quotaops.h>
3018 + #include <linux/buffer_head.h>
3019 + #include <linux/bio.h>
3020 ++#include <linux/vs_tag.h>
3021 +
3022 + #include "namei.h"
3023 + #include "xattr.h"
3024 +@@ -912,6 +913,7 @@ restart:
3025 + if (bh)
3026 + ll_rw_block(READ_META, 1, &bh);
3027 + }
3028 ++ dx_propagate_tag(nd, inode);
3029 + }
3030 + if ((bh = bh_use[ra_ptr++]) == NULL)
3031 + goto next;
3032 +@@ -2433,6 +2435,7 @@ const struct inode_operations ext3_dir_i
3033 + .removexattr = generic_removexattr,
3034 + #endif
3035 + .permission = ext3_permission,
3036 ++ .sync_flags = ext3_sync_flags,
3037 + };
3038 +
3039 + const struct inode_operations ext3_special_inode_operations = {
3040 +@@ -2444,4 +2447,5 @@ const struct inode_operations ext3_speci
3041 + .removexattr = generic_removexattr,
3042 + #endif
3043 + .permission = ext3_permission,
3044 ++ .sync_flags = ext3_sync_flags,
3045 + };
3046 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/super.c
3047 +--- linux-2.6.29.4/fs/ext3/super.c 2009-03-24 14:22:25.000000000 +0100
3048 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/super.c 2009-03-24 14:48:35.000000000 +0100
3049 +@@ -790,7 +790,7 @@ enum {
3050 + Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
3051 + Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
3052 + Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
3053 +- Opt_grpquota
3054 ++ Opt_grpquota, Opt_tag, Opt_notag, Opt_tagid
3055 + };
3056 +
3057 + static const match_table_t tokens = {
3058 +@@ -843,6 +843,9 @@ static const match_table_t tokens = {
3059 + {Opt_usrquota, "usrquota"},
3060 + {Opt_barrier, "barrier=%u"},
3061 + {Opt_resize, "resize"},
3062 ++ {Opt_tag, "tag"},
3063 ++ {Opt_notag, "notag"},
3064 ++ {Opt_tagid, "tagid=%u"},
3065 + {Opt_err, NULL},
3066 + };
3067 +
3068 +@@ -935,6 +938,20 @@ static int parse_options (char *options,
3069 + case Opt_nouid32:
3070 + set_opt (sbi->s_mount_opt, NO_UID32);
3071 + break;
3072 ++#ifndef CONFIG_TAGGING_NONE
3073 ++ case Opt_tag:
3074 ++ set_opt (sbi->s_mount_opt, TAGGED);
3075 ++ break;
3076 ++ case Opt_notag:
3077 ++ clear_opt (sbi->s_mount_opt, TAGGED);
3078 ++ break;
3079 ++#endif
3080 ++#ifdef CONFIG_PROPAGATE
3081 ++ case Opt_tagid:
3082 ++ /* use args[0] */
3083 ++ set_opt (sbi->s_mount_opt, TAGGED);
3084 ++ break;
3085 ++#endif
3086 + case Opt_nocheck:
3087 + clear_opt (sbi->s_mount_opt, CHECK);
3088 + break;
3089 +@@ -1653,6 +1670,9 @@ static int ext3_fill_super (struct super
3090 + NULL, 0))
3091 + goto failed_mount;
3092 +
3093 ++ if (EXT3_SB(sb)->s_mount_opt & EXT3_MOUNT_TAGGED)
3094 ++ sb->s_flags |= MS_TAGGED;
3095 ++
3096 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3097 + ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3098 +
3099 +@@ -2527,6 +2547,13 @@ static int ext3_remount (struct super_bl
3100 + if (sbi->s_mount_opt & EXT3_MOUNT_ABORT)
3101 + ext3_abort(sb, __func__, "Abort forced by user");
3102 +
3103 ++ if ((sbi->s_mount_opt & EXT3_MOUNT_TAGGED) &&
3104 ++ !(sb->s_flags & MS_TAGGED)) {
3105 ++ printk("EXT3-fs: %s: tagging not permitted on remount.\n",
3106 ++ sb->s_id);
3107 ++ return -EINVAL;
3108 ++ }
3109 ++
3110 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3111 + ((sbi->s_mount_opt & EXT3_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3112 +
3113 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/symlink.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/symlink.c
3114 +--- linux-2.6.29.4/fs/ext3/symlink.c 2008-12-25 00:26:37.000000000 +0100
3115 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/symlink.c 2009-02-22 22:54:25.000000000 +0100
3116 +@@ -40,6 +40,7 @@ const struct inode_operations ext3_symli
3117 + .listxattr = ext3_listxattr,
3118 + .removexattr = generic_removexattr,
3119 + #endif
3120 ++ .sync_flags = ext3_sync_flags,
3121 + };
3122 +
3123 + const struct inode_operations ext3_fast_symlink_inode_operations = {
3124 +@@ -51,4 +52,5 @@ const struct inode_operations ext3_fast_
3125 + .listxattr = ext3_listxattr,
3126 + .removexattr = generic_removexattr,
3127 + #endif
3128 ++ .sync_flags = ext3_sync_flags,
3129 + };
3130 +diff -NurpP --minimal linux-2.6.29.4/fs/ext3/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/xattr.c
3131 +--- linux-2.6.29.4/fs/ext3/xattr.c 2008-12-25 00:26:37.000000000 +0100
3132 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext3/xattr.c 2009-02-22 22:54:25.000000000 +0100
3133 +@@ -58,6 +58,7 @@
3134 + #include <linux/mbcache.h>
3135 + #include <linux/quotaops.h>
3136 + #include <linux/rwsem.h>
3137 ++#include <linux/vs_dlimit.h>
3138 + #include "xattr.h"
3139 + #include "acl.h"
3140 +
3141 +@@ -498,6 +499,7 @@ ext3_xattr_release_block(handle_t *handl
3142 + error = ext3_journal_dirty_metadata(handle, bh);
3143 + if (IS_SYNC(inode))
3144 + handle->h_sync = 1;
3145 ++ DLIMIT_FREE_BLOCK(inode, 1);
3146 + DQUOT_FREE_BLOCK(inode, 1);
3147 + ea_bdebug(bh, "refcount now=%d; releasing",
3148 + le32_to_cpu(BHDR(bh)->h_refcount));
3149 +@@ -771,11 +773,14 @@ inserted:
3150 + if (new_bh == bs->bh)
3151 + ea_bdebug(new_bh, "keeping");
3152 + else {
3153 ++ error = -ENOSPC;
3154 ++ if (DLIMIT_ALLOC_BLOCK(inode, 1))
3155 ++ goto cleanup;
3156 + /* The old block is released after updating
3157 + the inode. */
3158 + error = -EDQUOT;
3159 + if (DQUOT_ALLOC_BLOCK(inode, 1))
3160 +- goto cleanup;
3161 ++ goto cleanup_dlimit;
3162 + error = ext3_journal_get_write_access(handle,
3163 + new_bh);
3164 + if (error)
3165 +@@ -849,6 +854,8 @@ cleanup:
3166 +
3167 + cleanup_dquot:
3168 + DQUOT_FREE_BLOCK(inode, 1);
3169 ++cleanup_dlimit:
3170 ++ DLIMIT_FREE_BLOCK(inode, 1);
3171 + goto cleanup;
3172 +
3173 + bad_block:
3174 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/balloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/balloc.c
3175 +--- linux-2.6.29.4/fs/ext4/balloc.c 2009-03-24 14:22:25.000000000 +0100
3176 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/balloc.c 2009-03-24 14:48:35.000000000 +0100
3177 +@@ -17,6 +17,8 @@
3178 + #include <linux/jbd2.h>
3179 + #include <linux/quotaops.h>
3180 + #include <linux/buffer_head.h>
3181 ++#include <linux/vs_dlimit.h>
3182 ++#include <linux/vs_tag.h>
3183 + #include "ext4.h"
3184 + #include "ext4_jbd2.h"
3185 + #include "group.h"
3186 +@@ -535,8 +537,10 @@ void ext4_free_blocks(handle_t *handle,
3187 +
3188 + ext4_mb_free_blocks(handle, inode, block, count,
3189 + metadata, &dquot_freed_blocks);
3190 +- if (dquot_freed_blocks)
3191 ++ if (dquot_freed_blocks) {
3192 ++ DLIMIT_FREE_BLOCK(inode, dquot_freed_blocks);
3193 + DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
3194 ++ }
3195 + return;
3196 + }
3197 +
3198 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/ext4.h linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ext4.h
3199 +--- linux-2.6.29.4/fs/ext4/ext4.h 2009-05-23 23:16:52.000000000 +0200
3200 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ext4.h 2009-05-23 23:19:11.000000000 +0200
3201 +@@ -243,8 +243,12 @@ struct flex_groups {
3202 + #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
3203 + #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
3204 + #define EXT4_EXT_MIGRATE 0x00100000 /* Inode is migrating */
3205 ++#define EXT4_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
3206 + #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
3207 +
3208 ++#define EXT4_BARRIER_FL 0x04000000 /* Barrier for chroot() */
3209 ++#define EXT4_COW_FL 0x20000000 /* Copy on Write marker */
3210 ++
3211 + #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
3212 + #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
3213 +
3214 +@@ -541,6 +545,7 @@ do { \
3215 + #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
3216 + #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
3217 + #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
3218 ++#define EXT4_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
3219 +
3220 + /* Compatibility, for having both ext2_fs.h and ext4_fs.h included at once */
3221 + #ifndef _LINUX_EXT2_FS_H
3222 +@@ -1076,6 +1081,7 @@ struct buffer_head *ext4_bread(handle_t
3223 + ext4_lblk_t, int, int *);
3224 + int ext4_get_block(struct inode *inode, sector_t iblock,
3225 + struct buffer_head *bh_result, int create);
3226 ++extern int ext4_sync_flags(struct inode *inode);
3227 +
3228 + extern struct inode *ext4_iget(struct super_block *, unsigned long);
3229 + extern int ext4_write_inode(struct inode *, int);
3230 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/file.c
3231 +--- linux-2.6.29.4/fs/ext4/file.c 2009-03-24 14:22:25.000000000 +0100
3232 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/file.c 2009-03-24 14:48:35.000000000 +0100
3233 +@@ -171,5 +171,6 @@ const struct inode_operations ext4_file_
3234 + .permission = ext4_permission,
3235 + .fallocate = ext4_fallocate,
3236 + .fiemap = ext4_fiemap,
3237 ++ .sync_flags = ext4_sync_flags,
3238 + };
3239 +
3240 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/ialloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ialloc.c
3241 +--- linux-2.6.29.4/fs/ext4/ialloc.c 2009-03-24 14:22:25.000000000 +0100
3242 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ialloc.c 2009-03-24 14:48:35.000000000 +0100
3243 +@@ -22,6 +22,8 @@
3244 + #include <linux/random.h>
3245 + #include <linux/bitops.h>
3246 + #include <linux/blkdev.h>
3247 ++#include <linux/vs_dlimit.h>
3248 ++#include <linux/vs_tag.h>
3249 + #include <asm/byteorder.h>
3250 + #include "ext4.h"
3251 + #include "ext4_jbd2.h"
3252 +@@ -224,6 +226,7 @@ void ext4_free_inode(handle_t *handle, s
3253 + ext4_xattr_delete_inode(handle, inode);
3254 + DQUOT_FREE_INODE(inode);
3255 + DQUOT_DROP(inode);
3256 ++ DLIMIT_FREE_INODE(inode);
3257 +
3258 + is_directory = S_ISDIR(inode->i_mode);
3259 +
3260 +@@ -711,6 +714,12 @@ struct inode *ext4_new_inode(handle_t *h
3261 + inode = new_inode(sb);
3262 + if (!inode)
3263 + return ERR_PTR(-ENOMEM);
3264 ++
3265 ++ inode->i_tag = dx_current_fstag(sb);
3266 ++ if (DLIMIT_ALLOC_INODE(inode)) {
3267 ++ err = -ENOSPC;
3268 ++ goto out_dlimit;
3269 ++ }
3270 + ei = EXT4_I(inode);
3271 +
3272 + sbi = EXT4_SB(sb);
3273 +@@ -889,7 +898,8 @@ got:
3274 + * newly created directory and file only if -o extent mount option is
3275 + * specified
3276 + */
3277 +- ei->i_flags = EXT4_I(dir)->i_flags & ~(EXT4_INDEX_FL|EXT4_EXTENTS_FL);
3278 ++ ei->i_flags = EXT4_I(dir)->i_flags &
3279 ++ ~(EXT4_INDEX_FL|EXT4_EXTENTS_FL|EXT4_IXUNLINK_FL|EXT4_BARRIER_FL);
3280 + if (S_ISLNK(mode))
3281 + ei->i_flags &= ~(EXT4_IMMUTABLE_FL|EXT4_APPEND_FL);
3282 + /* dirsync only applies to directories */
3283 +@@ -949,6 +959,8 @@ got:
3284 + fail:
3285 + ext4_std_error(sb, err);
3286 + out:
3287 ++ DLIMIT_FREE_INODE(inode);
3288 ++out_dlimit:
3289 + iput(inode);
3290 + ret = ERR_PTR(err);
3291 + really_out:
3292 +@@ -960,6 +972,7 @@ fail_free_drop:
3293 +
3294 + fail_drop:
3295 + DQUOT_DROP(inode);
3296 ++ DLIMIT_FREE_INODE(inode);
3297 + inode->i_flags |= S_NOQUOTA;
3298 + inode->i_nlink = 0;
3299 + unlock_new_inode(inode);
3300 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/inode.c
3301 +--- linux-2.6.29.4/fs/ext4/inode.c 2009-05-23 23:16:52.000000000 +0200
3302 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/inode.c 2009-05-23 23:19:11.000000000 +0200
3303 +@@ -37,6 +37,7 @@
3304 + #include <linux/namei.h>
3305 + #include <linux/uio.h>
3306 + #include <linux/bio.h>
3307 ++#include <linux/vs_tag.h>
3308 + #include "ext4_jbd2.h"
3309 + #include "xattr.h"
3310 + #include "acl.h"
3311 +@@ -3781,7 +3782,7 @@ static void ext4_free_branches(handle_t
3312 +
3313 + int ext4_can_truncate(struct inode *inode)
3314 + {
3315 +- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3316 ++ if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3317 + return 0;
3318 + if (S_ISREG(inode->i_mode))
3319 + return 1;
3320 +@@ -4134,37 +4135,86 @@ void ext4_set_inode_flags(struct inode *
3321 + {
3322 + unsigned int flags = EXT4_I(inode)->i_flags;
3323 +
3324 +- inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3325 ++ inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3326 ++ S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3327 ++
3328 ++ if (flags & EXT4_IMMUTABLE_FL)
3329 ++ inode->i_flags |= S_IMMUTABLE;
3330 ++ if (flags & EXT4_IXUNLINK_FL)
3331 ++ inode->i_flags |= S_IXUNLINK;
3332 ++
3333 + if (flags & EXT4_SYNC_FL)
3334 + inode->i_flags |= S_SYNC;
3335 + if (flags & EXT4_APPEND_FL)
3336 + inode->i_flags |= S_APPEND;
3337 +- if (flags & EXT4_IMMUTABLE_FL)
3338 +- inode->i_flags |= S_IMMUTABLE;
3339 + if (flags & EXT4_NOATIME_FL)
3340 + inode->i_flags |= S_NOATIME;
3341 + if (flags & EXT4_DIRSYNC_FL)
3342 + inode->i_flags |= S_DIRSYNC;
3343 ++
3344 ++ inode->i_vflags &= ~(V_BARRIER | V_COW);
3345 ++
3346 ++ if (flags & EXT4_BARRIER_FL)
3347 ++ inode->i_vflags |= V_BARRIER;
3348 ++ if (flags & EXT4_COW_FL)
3349 ++ inode->i_vflags |= V_COW;
3350 + }
3351 +
3352 + /* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
3353 + void ext4_get_inode_flags(struct ext4_inode_info *ei)
3354 + {
3355 + unsigned int flags = ei->vfs_inode.i_flags;
3356 ++ unsigned int vflags = ei->vfs_inode.i_vflags;
3357 ++
3358 ++ ei->i_flags &= ~(EXT4_SYNC_FL | EXT4_APPEND_FL |
3359 ++ EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL |
3360 ++ EXT4_NOATIME_FL | EXT4_DIRSYNC_FL |
3361 ++ EXT4_BARRIER_FL | EXT4_COW_FL);
3362 ++
3363 ++ if (flags & S_IMMUTABLE)
3364 ++ ei->i_flags |= EXT4_IMMUTABLE_FL;
3365 ++ if (flags & S_IXUNLINK)
3366 ++ ei->i_flags |= EXT4_IXUNLINK_FL;
3367 +
3368 +- ei->i_flags &= ~(EXT4_SYNC_FL|EXT4_APPEND_FL|
3369 +- EXT4_IMMUTABLE_FL|EXT4_NOATIME_FL|EXT4_DIRSYNC_FL);
3370 + if (flags & S_SYNC)
3371 + ei->i_flags |= EXT4_SYNC_FL;
3372 + if (flags & S_APPEND)
3373 + ei->i_flags |= EXT4_APPEND_FL;
3374 +- if (flags & S_IMMUTABLE)
3375 +- ei->i_flags |= EXT4_IMMUTABLE_FL;
3376 + if (flags & S_NOATIME)
3377 + ei->i_flags |= EXT4_NOATIME_FL;
3378 + if (flags & S_DIRSYNC)
3379 + ei->i_flags |= EXT4_DIRSYNC_FL;
3380 ++
3381 ++ if (vflags & V_BARRIER)
3382 ++ ei->i_flags |= EXT4_BARRIER_FL;
3383 ++ if (vflags & V_COW)
3384 ++ ei->i_flags |= EXT4_COW_FL;
3385 ++}
3386 ++
3387 ++int ext4_sync_flags(struct inode *inode)
3388 ++{
3389 ++ struct ext4_iloc iloc;
3390 ++ handle_t *handle;
3391 ++ int err;
3392 ++
3393 ++ handle = ext4_journal_start(inode, 1);
3394 ++ if (IS_ERR(handle))
3395 ++ return PTR_ERR(handle);
3396 ++ if (IS_SYNC(inode))
3397 ++ handle->h_sync = 1;
3398 ++ err = ext4_reserve_inode_write(handle, inode, &iloc);
3399 ++ if (err)
3400 ++ goto flags_err;
3401 ++
3402 ++ ext4_get_inode_flags(EXT4_I(inode));
3403 ++ inode->i_ctime = CURRENT_TIME;
3404 ++
3405 ++ err = ext4_mark_iloc_dirty(handle, inode, &iloc);
3406 ++flags_err:
3407 ++ ext4_journal_stop(handle);
3408 ++ return err;
3409 + }
3410 ++
3411 + static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
3412 + struct ext4_inode_info *ei)
3413 + {
3414 +@@ -4197,6 +4247,8 @@ struct inode *ext4_iget(struct super_blo
3415 + struct inode *inode;
3416 + long ret;
3417 + int block;
3418 ++ uid_t uid;
3419 ++ gid_t gid;
3420 +
3421 + inode = iget_locked(sb, ino);
3422 + if (!inode)
3423 +@@ -4216,12 +4268,16 @@ struct inode *ext4_iget(struct super_blo
3424 + bh = iloc.bh;
3425 + raw_inode = ext4_raw_inode(&iloc);
3426 + inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3427 +- inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3428 +- inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3429 ++ uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3430 ++ gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3431 + if (!(test_opt(inode->i_sb, NO_UID32))) {
3432 +- inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3433 +- inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3434 ++ uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3435 ++ gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3436 + }
3437 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3438 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3439 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3440 ++ le16_to_cpu(raw_inode->i_raw_tag));
3441 + inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3442 +
3443 + ei->i_state = 0;
3444 +@@ -4387,6 +4443,8 @@ static int ext4_do_update_inode(handle_t
3445 + struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
3446 + struct ext4_inode_info *ei = EXT4_I(inode);
3447 + struct buffer_head *bh = iloc->bh;
3448 ++ uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3449 ++ gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3450 + int err = 0, rc, block;
3451 +
3452 + /* For fields not not tracking in the in-memory inode,
3453 +@@ -4397,29 +4455,32 @@ static int ext4_do_update_inode(handle_t
3454 + ext4_get_inode_flags(ei);
3455 + raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3456 + if (!(test_opt(inode->i_sb, NO_UID32))) {
3457 +- raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3458 +- raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3459 ++ raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3460 ++ raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3461 + /*
3462 + * Fix up interoperability with old kernels. Otherwise, old inodes get
3463 + * re-used with the upper 16 bits of the uid/gid intact
3464 + */
3465 + if (!ei->i_dtime) {
3466 + raw_inode->i_uid_high =
3467 +- cpu_to_le16(high_16_bits(inode->i_uid));
3468 ++ cpu_to_le16(high_16_bits(uid));
3469 + raw_inode->i_gid_high =
3470 +- cpu_to_le16(high_16_bits(inode->i_gid));
3471 ++ cpu_to_le16(high_16_bits(gid));
3472 + } else {
3473 + raw_inode->i_uid_high = 0;
3474 + raw_inode->i_gid_high = 0;
3475 + }
3476 + } else {
3477 + raw_inode->i_uid_low =
3478 +- cpu_to_le16(fs_high2lowuid(inode->i_uid));
3479 ++ cpu_to_le16(fs_high2lowuid(uid));
3480 + raw_inode->i_gid_low =
3481 +- cpu_to_le16(fs_high2lowgid(inode->i_gid));
3482 ++ cpu_to_le16(fs_high2lowgid(gid));
3483 + raw_inode->i_uid_high = 0;
3484 + raw_inode->i_gid_high = 0;
3485 + }
3486 ++#ifdef CONFIG_TAGGING_INTERN
3487 ++ raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3488 ++#endif
3489 + raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3490 +
3491 + EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
3492 +@@ -4601,7 +4662,8 @@ int ext4_setattr(struct dentry *dentry,
3493 + return error;
3494 +
3495 + if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3496 +- (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3497 ++ (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3498 ++ (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3499 + handle_t *handle;
3500 +
3501 + /* (user+group)*(old+new) structure, inode write (sb,
3502 +@@ -4623,6 +4685,8 @@ int ext4_setattr(struct dentry *dentry,
3503 + inode->i_uid = attr->ia_uid;
3504 + if (attr->ia_valid & ATTR_GID)
3505 + inode->i_gid = attr->ia_gid;
3506 ++ if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3507 ++ inode->i_tag = attr->ia_tag;
3508 + error = ext4_mark_inode_dirty(handle, inode);
3509 + ext4_journal_stop(handle);
3510 + }
3511 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ioctl.c
3512 +--- linux-2.6.29.4/fs/ext4/ioctl.c 2009-03-24 14:22:25.000000000 +0100
3513 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/ioctl.c 2009-03-24 14:48:35.000000000 +0100
3514 +@@ -8,12 +8,14 @@
3515 + */
3516 +
3517 + #include <linux/fs.h>
3518 ++#include <linux/mount.h>
3519 + #include <linux/jbd2.h>
3520 + #include <linux/capability.h>
3521 + #include <linux/time.h>
3522 + #include <linux/compat.h>
3523 + #include <linux/smp_lock.h>
3524 + #include <linux/mount.h>
3525 ++#include <linux/vs_tag.h>
3526 + #include <asm/uaccess.h>
3527 + #include "ext4_jbd2.h"
3528 + #include "ext4.h"
3529 +@@ -51,6 +53,11 @@ long ext4_ioctl(struct file *filp, unsig
3530 + if (!S_ISDIR(inode->i_mode))
3531 + flags &= ~EXT4_DIRSYNC_FL;
3532 +
3533 ++ if (IS_BARRIER(inode)) {
3534 ++ vxwprintk_task(1, "messing with the barrier.");
3535 ++ return -EACCES;
3536 ++ }
3537 ++
3538 + err = -EPERM;
3539 + mutex_lock(&inode->i_mutex);
3540 + /* Is it quota file? Do not allow user to mess with it */
3541 +@@ -68,7 +75,9 @@ long ext4_ioctl(struct file *filp, unsig
3542 + *
3543 + * This test looks nicer. Thanks to Pauline Middelink
3544 + */
3545 +- if ((flags ^ oldflags) & (EXT4_APPEND_FL | EXT4_IMMUTABLE_FL)) {
3546 ++ if ((oldflags & EXT4_IMMUTABLE_FL) ||
3547 ++ ((flags ^ oldflags) & (EXT4_APPEND_FL |
3548 ++ EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL))) {
3549 + if (!capable(CAP_LINUX_IMMUTABLE))
3550 + goto flags_out;
3551 + }
3552 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/namei.c
3553 +--- linux-2.6.29.4/fs/ext4/namei.c 2009-03-24 14:22:25.000000000 +0100
3554 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/namei.c 2009-03-24 14:48:35.000000000 +0100
3555 +@@ -34,6 +34,7 @@
3556 + #include <linux/quotaops.h>
3557 + #include <linux/buffer_head.h>
3558 + #include <linux/bio.h>
3559 ++#include <linux/vs_tag.h>
3560 + #include "ext4.h"
3561 + #include "ext4_jbd2.h"
3562 +
3563 +@@ -917,6 +918,7 @@ restart:
3564 + if (bh)
3565 + ll_rw_block(READ_META, 1, &bh);
3566 + }
3567 ++ dx_propagate_tag(nd, inode);
3568 + }
3569 + if ((bh = bh_use[ra_ptr++]) == NULL)
3570 + goto next;
3571 +@@ -2481,6 +2483,7 @@ const struct inode_operations ext4_dir_i
3572 + .removexattr = generic_removexattr,
3573 + #endif
3574 + .permission = ext4_permission,
3575 ++ .sync_flags = ext4_sync_flags,
3576 + };
3577 +
3578 + const struct inode_operations ext4_special_inode_operations = {
3579 +@@ -2492,4 +2495,5 @@ const struct inode_operations ext4_speci
3580 + .removexattr = generic_removexattr,
3581 + #endif
3582 + .permission = ext4_permission,
3583 ++ .sync_flags = ext4_sync_flags,
3584 + };
3585 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/super.c
3586 +--- linux-2.6.29.4/fs/ext4/super.c 2009-03-24 14:22:25.000000000 +0100
3587 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/super.c 2009-03-24 15:46:48.000000000 +0100
3588 +@@ -1013,7 +1013,8 @@ enum {
3589 + Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
3590 + Opt_grpquota, Opt_i_version,
3591 + Opt_stripe, Opt_delalloc, Opt_nodelalloc,
3592 +- Opt_inode_readahead_blks, Opt_journal_ioprio
3593 ++ Opt_inode_readahead_blks, Opt_journal_ioprio,
3594 ++ Opt_tag, Opt_notag, Opt_tagid
3595 + };
3596 +
3597 + static const match_table_t tokens = {
3598 +@@ -1073,6 +1074,9 @@ static const match_table_t tokens = {
3599 + {Opt_nodelalloc, "nodelalloc"},
3600 + {Opt_inode_readahead_blks, "inode_readahead_blks=%u"},
3601 + {Opt_journal_ioprio, "journal_ioprio=%u"},
3602 ++ {Opt_tag, "tag"},
3603 ++ {Opt_notag, "notag"},
3604 ++ {Opt_tagid, "tagid=%u"},
3605 + {Opt_err, NULL},
3606 + };
3607 +
3608 +@@ -1168,6 +1172,20 @@ static int parse_options(char *options,
3609 + case Opt_nouid32:
3610 + set_opt(sbi->s_mount_opt, NO_UID32);
3611 + break;
3612 ++#ifndef CONFIG_TAGGING_NONE
3613 ++ case Opt_tag:
3614 ++ set_opt (sbi->s_mount_opt, TAGGED);
3615 ++ break;
3616 ++ case Opt_notag:
3617 ++ clear_opt (sbi->s_mount_opt, TAGGED);
3618 ++ break;
3619 ++#endif
3620 ++#ifdef CONFIG_PROPAGATE
3621 ++ case Opt_tagid:
3622 ++ /* use args[0] */
3623 ++ set_opt (sbi->s_mount_opt, TAGGED);
3624 ++ break;
3625 ++#endif
3626 + case Opt_debug:
3627 + set_opt(sbi->s_mount_opt, DEBUG);
3628 + break;
3629 +@@ -2113,6 +2131,9 @@ static int ext4_fill_super(struct super_
3630 + &journal_ioprio, NULL, 0))
3631 + goto failed_mount;
3632 +
3633 ++ if (EXT4_SB(sb)->s_mount_opt & EXT4_MOUNT_TAGGED)
3634 ++ sb->s_flags |= MS_TAGGED;
3635 ++
3636 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3637 + ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3638 +
3639 +@@ -3160,6 +3181,13 @@ static int ext4_remount(struct super_blo
3640 + if (sbi->s_mount_opt & EXT4_MOUNT_ABORT)
3641 + ext4_abort(sb, __func__, "Abort forced by user");
3642 +
3643 ++ if ((sbi->s_mount_opt & EXT4_MOUNT_TAGGED) &&
3644 ++ !(sb->s_flags & MS_TAGGED)) {
3645 ++ printk("EXT4-fs: %s: tagging not permitted on remount.\n",
3646 ++ sb->s_id);
3647 ++ return -EINVAL;
3648 ++ }
3649 ++
3650 + sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3651 + ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3652 +
3653 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/symlink.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/symlink.c
3654 +--- linux-2.6.29.4/fs/ext4/symlink.c 2008-12-25 00:26:37.000000000 +0100
3655 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/symlink.c 2009-02-22 22:54:25.000000000 +0100
3656 +@@ -40,6 +40,7 @@ const struct inode_operations ext4_symli
3657 + .listxattr = ext4_listxattr,
3658 + .removexattr = generic_removexattr,
3659 + #endif
3660 ++ .sync_flags = ext4_sync_flags,
3661 + };
3662 +
3663 + const struct inode_operations ext4_fast_symlink_inode_operations = {
3664 +@@ -51,4 +52,5 @@ const struct inode_operations ext4_fast_
3665 + .listxattr = ext4_listxattr,
3666 + .removexattr = generic_removexattr,
3667 + #endif
3668 ++ .sync_flags = ext4_sync_flags,
3669 + };
3670 +diff -NurpP --minimal linux-2.6.29.4/fs/ext4/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/xattr.c
3671 +--- linux-2.6.29.4/fs/ext4/xattr.c 2009-03-24 14:22:25.000000000 +0100
3672 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ext4/xattr.c 2009-03-24 15:47:44.000000000 +0100
3673 +@@ -56,6 +56,7 @@
3674 + #include <linux/mbcache.h>
3675 + #include <linux/quotaops.h>
3676 + #include <linux/rwsem.h>
3677 ++#include <linux/vs_dlimit.h>
3678 + #include "ext4_jbd2.h"
3679 + #include "ext4.h"
3680 + #include "xattr.h"
3681 +@@ -490,6 +491,7 @@ ext4_xattr_release_block(handle_t *handl
3682 + error = ext4_handle_dirty_metadata(handle, inode, bh);
3683 + if (IS_SYNC(inode))
3684 + ext4_handle_sync(handle);
3685 ++ DLIMIT_FREE_BLOCK(inode, 1);
3686 + DQUOT_FREE_BLOCK(inode, 1);
3687 + ea_bdebug(bh, "refcount now=%d; releasing",
3688 + le32_to_cpu(BHDR(bh)->h_refcount));
3689 +@@ -781,11 +783,14 @@ inserted:
3690 + if (new_bh == bs->bh)
3691 + ea_bdebug(new_bh, "keeping");
3692 + else {
3693 ++ error = -ENOSPC;
3694 ++ if (DLIMIT_ALLOC_BLOCK(inode, 1))
3695 ++ goto cleanup;
3696 + /* The old block is released after updating
3697 + the inode. */
3698 + error = -EDQUOT;
3699 + if (DQUOT_ALLOC_BLOCK(inode, 1))
3700 +- goto cleanup;
3701 ++ goto cleanup_dlimit;
3702 + error = ext4_journal_get_write_access(handle,
3703 + new_bh);
3704 + if (error)
3705 +@@ -861,6 +866,8 @@ cleanup:
3706 +
3707 + cleanup_dquot:
3708 + DQUOT_FREE_BLOCK(inode, 1);
3709 ++cleanup_dlimit:
3710 ++ DLIMIT_FREE_BLOCK(inode, 1);
3711 + goto cleanup;
3712 +
3713 + bad_block:
3714 +diff -NurpP --minimal linux-2.6.29.4/fs/fcntl.c linux-2.6.29.4-vs2.3.0.36.14/fs/fcntl.c
3715 +--- linux-2.6.29.4/fs/fcntl.c 2009-05-23 23:16:52.000000000 +0200
3716 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/fcntl.c 2009-05-23 23:19:11.000000000 +0200
3717 +@@ -20,6 +20,7 @@
3718 + #include <linux/rcupdate.h>
3719 + #include <linux/pid_namespace.h>
3720 + #include <linux/smp_lock.h>
3721 ++#include <linux/vs_limit.h>
3722 +
3723 + #include <asm/poll.h>
3724 + #include <asm/siginfo.h>
3725 +@@ -103,6 +104,8 @@ SYSCALL_DEFINE3(dup3, unsigned int, oldf
3726 +
3727 + if (tofree)
3728 + filp_close(tofree, files);
3729 ++ else
3730 ++ vx_openfd_inc(newfd); /* fd was unused */
3731 +
3732 + return newfd;
3733 +
3734 +@@ -345,6 +348,8 @@ SYSCALL_DEFINE3(fcntl, unsigned int, fd,
3735 + filp = fget(fd);
3736 + if (!filp)
3737 + goto out;
3738 ++ if (!vx_files_avail(1))
3739 ++ goto out;
3740 +
3741 + err = security_file_fcntl(filp, cmd, arg);
3742 + if (err) {
3743 +diff -NurpP --minimal linux-2.6.29.4/fs/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/file.c
3744 +--- linux-2.6.29.4/fs/file.c 2008-12-25 00:26:37.000000000 +0100
3745 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/file.c 2009-02-22 22:54:25.000000000 +0100
3746 +@@ -19,6 +19,7 @@
3747 + #include <linux/spinlock.h>
3748 + #include <linux/rcupdate.h>
3749 + #include <linux/workqueue.h>
3750 ++#include <linux/vs_limit.h>
3751 +
3752 + struct fdtable_defer {
3753 + spinlock_t lock;
3754 +@@ -367,6 +368,8 @@ struct files_struct *dup_fd(struct files
3755 + struct file *f = *old_fds++;
3756 + if (f) {
3757 + get_file(f);
3758 ++ /* TODO: sum it first for check and performance */
3759 ++ vx_openfd_inc(open_files - i);
3760 + } else {
3761 + /*
3762 + * The fd may be claimed in the fd bitmap but not yet
3763 +@@ -475,6 +478,7 @@ repeat:
3764 + else
3765 + FD_CLR(fd, fdt->close_on_exec);
3766 + error = fd;
3767 ++ vx_openfd_inc(fd);
3768 + #if 1
3769 + /* Sanity check */
3770 + if (rcu_dereference(fdt->fd[fd]) != NULL) {
3771 +diff -NurpP --minimal linux-2.6.29.4/fs/file_table.c linux-2.6.29.4-vs2.3.0.36.14/fs/file_table.c
3772 +--- linux-2.6.29.4/fs/file_table.c 2009-03-24 14:22:25.000000000 +0100
3773 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/file_table.c 2009-03-24 14:48:35.000000000 +0100
3774 +@@ -21,6 +21,8 @@
3775 + #include <linux/fsnotify.h>
3776 + #include <linux/sysctl.h>
3777 + #include <linux/percpu_counter.h>
3778 ++#include <linux/vs_limit.h>
3779 ++#include <linux/vs_context.h>
3780 +
3781 + #include <asm/atomic.h>
3782 +
3783 +@@ -129,6 +131,8 @@ struct file *get_empty_filp(void)
3784 + f->f_cred = get_cred(cred);
3785 + eventpoll_init_file(f);
3786 + /* f->f_version: 0 */
3787 ++ f->f_xid = vx_current_xid();
3788 ++ vx_files_inc(f);
3789 + return f;
3790 +
3791 + over:
3792 +@@ -283,6 +287,8 @@ void __fput(struct file *file)
3793 + cdev_put(inode->i_cdev);
3794 + fops_put(file->f_op);
3795 + put_pid(file->f_owner.pid);
3796 ++ vx_files_dec(file);
3797 ++ file->f_xid = 0;
3798 + file_kill(file);
3799 + if (file->f_mode & FMODE_WRITE)
3800 + drop_file_write_access(file);
3801 +@@ -350,6 +356,8 @@ void put_filp(struct file *file)
3802 + {
3803 + if (atomic_long_dec_and_test(&file->f_count)) {
3804 + security_file_free(file);
3805 ++ vx_files_dec(file);
3806 ++ file->f_xid = 0;
3807 + file_kill(file);
3808 + file_free(file);
3809 + }
3810 +diff -NurpP --minimal linux-2.6.29.4/fs/fs_struct.c linux-2.6.29.4-vs2.3.0.36.14/fs/fs_struct.c
3811 +--- linux-2.6.29.4/fs/fs_struct.c 2009-05-23 23:16:52.000000000 +0200
3812 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/fs_struct.c 2009-05-20 01:29:56.000000000 +0200
3813 +@@ -3,6 +3,7 @@
3814 + #include <linux/fs.h>
3815 + #include <linux/path.h>
3816 + #include <linux/slab.h>
3817 ++#include <linux/vserver/global.h>
3818 +
3819 + /*
3820 + * Replace the fs->{rootmnt,root} with {mnt,dentry}. Put the old values.
3821 +@@ -76,6 +77,7 @@ void free_fs_struct(struct fs_struct *fs
3822 + {
3823 + path_put(&fs->root);
3824 + path_put(&fs->pwd);
3825 ++ atomic_dec(&vs_global_fs);
3826 + kmem_cache_free(fs_cachep, fs);
3827 + }
3828 +
3829 +@@ -111,6 +113,7 @@ struct fs_struct *copy_fs_struct(struct
3830 + fs->pwd = old->pwd;
3831 + path_get(&old->pwd);
3832 + read_unlock(&old->lock);
3833 ++ atomic_inc(&vs_global_fs);
3834 + }
3835 + return fs;
3836 + }
3837 +diff -NurpP --minimal linux-2.6.29.4/fs/hfsplus/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/hfsplus/ioctl.c
3838 +--- linux-2.6.29.4/fs/hfsplus/ioctl.c 2008-12-25 00:26:37.000000000 +0100
3839 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/hfsplus/ioctl.c 2009-02-22 22:54:25.000000000 +0100
3840 +@@ -17,6 +17,7 @@
3841 + #include <linux/mount.h>
3842 + #include <linux/sched.h>
3843 + #include <linux/xattr.h>
3844 ++#include <linux/mount.h>
3845 + #include <asm/uaccess.h>
3846 + #include "hfsplus_fs.h"
3847 +
3848 +diff -NurpP --minimal linux-2.6.29.4/fs/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/inode.c
3849 +--- linux-2.6.29.4/fs/inode.c 2009-03-24 14:22:26.000000000 +0100
3850 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/inode.c 2009-03-25 00:51:55.000000000 +0100
3851 +@@ -126,6 +126,9 @@ struct inode *inode_init_always(struct s
3852 + struct address_space * const mapping = &inode->i_data;
3853 +
3854 + inode->i_sb = sb;
3855 ++
3856 ++ /* essential because of inode slab reuse */
3857 ++ inode->i_tag = 0;
3858 + inode->i_blkbits = sb->s_blocksize_bits;
3859 + inode->i_flags = 0;
3860 + atomic_set(&inode->i_count, 1);
3861 +@@ -146,6 +149,7 @@ struct inode *inode_init_always(struct s
3862 + inode->i_bdev = NULL;
3863 + inode->i_cdev = NULL;
3864 + inode->i_rdev = 0;
3865 ++ inode->i_mdev = 0;
3866 + inode->dirtied_when = 0;
3867 + if (security_inode_alloc(inode)) {
3868 + if (inode->i_sb->s_op->destroy_inode)
3869 +@@ -267,6 +271,8 @@ void __iget(struct inode * inode)
3870 + inodes_stat.nr_unused--;
3871 + }
3872 +
3873 ++EXPORT_SYMBOL_GPL(__iget);
3874 ++
3875 + /**
3876 + * clear_inode - clear an inode
3877 + * @inode: inode to clear
3878 +@@ -1539,9 +1545,11 @@ void init_special_inode(struct inode *in
3879 + if (S_ISCHR(mode)) {
3880 + inode->i_fop = &def_chr_fops;
3881 + inode->i_rdev = rdev;
3882 ++ inode->i_mdev = rdev;
3883 + } else if (S_ISBLK(mode)) {
3884 + inode->i_fop = &def_blk_fops;
3885 + inode->i_rdev = rdev;
3886 ++ inode->i_mdev = rdev;
3887 + } else if (S_ISFIFO(mode))
3888 + inode->i_fop = &def_fifo_fops;
3889 + else if (S_ISSOCK(mode))
3890 +diff -NurpP --minimal linux-2.6.29.4/fs/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/ioctl.c
3891 +--- linux-2.6.29.4/fs/ioctl.c 2009-05-23 23:16:52.000000000 +0200
3892 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ioctl.c 2009-05-23 23:19:11.000000000 +0200
3893 +@@ -15,6 +15,9 @@
3894 + #include <linux/uaccess.h>
3895 + #include <linux/writeback.h>
3896 + #include <linux/buffer_head.h>
3897 ++#include <linux/proc_fs.h>
3898 ++#include <linux/vserver/inode.h>
3899 ++#include <linux/vs_tag.h>
3900 +
3901 + #include <asm/ioctls.h>
3902 +
3903 +diff -NurpP --minimal linux-2.6.29.4/fs/ioprio.c linux-2.6.29.4-vs2.3.0.36.14/fs/ioprio.c
3904 +--- linux-2.6.29.4/fs/ioprio.c 2009-03-24 14:22:26.000000000 +0100
3905 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ioprio.c 2009-03-24 14:48:35.000000000 +0100
3906 +@@ -26,6 +26,7 @@
3907 + #include <linux/syscalls.h>
3908 + #include <linux/security.h>
3909 + #include <linux/pid_namespace.h>
3910 ++#include <linux/vs_base.h>
3911 +
3912 + int set_task_ioprio(struct task_struct *task, int ioprio)
3913 + {
3914 +@@ -123,6 +124,8 @@ SYSCALL_DEFINE3(ioprio_set, int, which,
3915 + else
3916 + pgrp = find_vpid(who);
3917 + do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
3918 ++ if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
3919 ++ continue;
3920 + ret = set_task_ioprio(p, ioprio);
3921 + if (ret)
3922 + break;
3923 +@@ -212,6 +215,8 @@ SYSCALL_DEFINE2(ioprio_get, int, which,
3924 + else
3925 + pgrp = find_vpid(who);
3926 + do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
3927 ++ if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
3928 ++ continue;
3929 + tmpio = get_task_ioprio(p);
3930 + if (tmpio < 0)
3931 + continue;
3932 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/acl.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/acl.c
3933 +--- linux-2.6.29.4/fs/jfs/acl.c 2008-12-25 00:26:37.000000000 +0100
3934 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/acl.c 2009-02-22 22:54:25.000000000 +0100
3935 +@@ -232,7 +232,8 @@ int jfs_setattr(struct dentry *dentry, s
3936 + return rc;
3937 +
3938 + if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
3939 +- (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
3940 ++ (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
3941 ++ (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
3942 + if (DQUOT_TRANSFER(inode, iattr))
3943 + return -EDQUOT;
3944 + }
3945 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/file.c
3946 +--- linux-2.6.29.4/fs/jfs/file.c 2008-12-25 00:26:37.000000000 +0100
3947 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/file.c 2009-02-22 22:54:25.000000000 +0100
3948 +@@ -98,6 +98,7 @@ const struct inode_operations jfs_file_i
3949 + .setattr = jfs_setattr,
3950 + .permission = jfs_permission,
3951 + #endif
3952 ++ .sync_flags = jfs_sync_flags,
3953 + };
3954 +
3955 + const struct file_operations jfs_file_operations = {
3956 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/inode.c
3957 +--- linux-2.6.29.4/fs/jfs/inode.c 2009-03-24 14:22:26.000000000 +0100
3958 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/inode.c 2009-03-24 14:48:35.000000000 +0100
3959 +@@ -22,6 +22,7 @@
3960 + #include <linux/buffer_head.h>
3961 + #include <linux/pagemap.h>
3962 + #include <linux/quotaops.h>
3963 ++#include <linux/vs_dlimit.h>
3964 + #include "jfs_incore.h"
3965 + #include "jfs_inode.h"
3966 + #include "jfs_filsys.h"
3967 +@@ -161,6 +162,7 @@ void jfs_delete_inode(struct inode *inod
3968 + DQUOT_INIT(inode);
3969 + DQUOT_FREE_INODE(inode);
3970 + DQUOT_DROP(inode);
3971 ++ DLIMIT_FREE_INODE(inode);
3972 + }
3973 +
3974 + clear_inode(inode);
3975 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/ioctl.c
3976 +--- linux-2.6.29.4/fs/jfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
3977 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/ioctl.c 2009-02-22 22:54:25.000000000 +0100
3978 +@@ -11,6 +11,7 @@
3979 + #include <linux/mount.h>
3980 + #include <linux/time.h>
3981 + #include <linux/sched.h>
3982 ++#include <linux/mount.h>
3983 + #include <asm/current.h>
3984 + #include <asm/uaccess.h>
3985 +
3986 +@@ -85,6 +86,11 @@ long jfs_ioctl(struct file *filp, unsign
3987 + if (!S_ISDIR(inode->i_mode))
3988 + flags &= ~JFS_DIRSYNC_FL;
3989 +
3990 ++ if (IS_BARRIER(inode)) {
3991 ++ vxwprintk_task(1, "messing with the barrier.");
3992 ++ return -EACCES;
3993 ++ }
3994 ++
3995 + /* Is it quota file? Do not allow user to mess with it */
3996 + if (IS_NOQUOTA(inode)) {
3997 + err = -EPERM;
3998 +@@ -102,8 +108,8 @@ long jfs_ioctl(struct file *filp, unsign
3999 + * the relevant capability.
4000 + */
4001 + if ((oldflags & JFS_IMMUTABLE_FL) ||
4002 +- ((flags ^ oldflags) &
4003 +- (JFS_APPEND_FL | JFS_IMMUTABLE_FL))) {
4004 ++ ((flags ^ oldflags) & (JFS_APPEND_FL |
4005 ++ JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL))) {
4006 + if (!capable(CAP_LINUX_IMMUTABLE)) {
4007 + mutex_unlock(&inode->i_mutex);
4008 + err = -EPERM;
4009 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_dinode.h linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_dinode.h
4010 +--- linux-2.6.29.4/fs/jfs/jfs_dinode.h 2008-12-25 00:26:37.000000000 +0100
4011 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_dinode.h 2009-02-22 22:54:25.000000000 +0100
4012 +@@ -161,9 +161,13 @@ struct dinode {
4013 +
4014 + #define JFS_APPEND_FL 0x01000000 /* writes to file may only append */
4015 + #define JFS_IMMUTABLE_FL 0x02000000 /* Immutable file */
4016 ++#define JFS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
4017 +
4018 +-#define JFS_FL_USER_VISIBLE 0x03F80000
4019 +-#define JFS_FL_USER_MODIFIABLE 0x03F80000
4020 ++#define JFS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
4021 ++#define JFS_COW_FL 0x20000000 /* Copy on Write marker */
4022 ++
4023 ++#define JFS_FL_USER_VISIBLE 0x07F80000
4024 ++#define JFS_FL_USER_MODIFIABLE 0x07F80000
4025 + #define JFS_FL_INHERIT 0x03C80000
4026 +
4027 + /* These are identical to EXT[23]_IOC_GETFLAGS/SETFLAGS */
4028 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_dtree.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_dtree.c
4029 +--- linux-2.6.29.4/fs/jfs/jfs_dtree.c 2008-12-25 00:26:37.000000000 +0100
4030 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_dtree.c 2009-02-22 22:54:25.000000000 +0100
4031 +@@ -102,6 +102,7 @@
4032 +
4033 + #include <linux/fs.h>
4034 + #include <linux/quotaops.h>
4035 ++#include <linux/vs_dlimit.h>
4036 + #include "jfs_incore.h"
4037 + #include "jfs_superblock.h"
4038 + #include "jfs_filsys.h"
4039 +@@ -383,10 +384,10 @@ static u32 add_index(tid_t tid, struct i
4040 + */
4041 + if (DQUOT_ALLOC_BLOCK(ip, sbi->nbperpage))
4042 + goto clean_up;
4043 +- if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr)) {
4044 +- DQUOT_FREE_BLOCK(ip, sbi->nbperpage);
4045 +- goto clean_up;
4046 +- }
4047 ++ if (DLIMIT_ALLOC_BLOCK(ip, sbi->nbperpage))
4048 ++ goto clean_up_dquot;
4049 ++ if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr))
4050 ++ goto clean_up_dlimit;
4051 +
4052 + /*
4053 + * Save the table, we're going to overwrite it with the
4054 +@@ -480,6 +481,12 @@ static u32 add_index(tid_t tid, struct i
4055 +
4056 + return index;
4057 +
4058 ++ clean_up_dlimit:
4059 ++ DLIMIT_FREE_BLOCK(ip, sbi->nbperpage);
4060 ++
4061 ++ clean_up_dquot:
4062 ++ DQUOT_FREE_BLOCK(ip, sbi->nbperpage);
4063 ++
4064 + clean_up:
4065 +
4066 + jfs_ip->next_index--;
4067 +@@ -951,6 +958,7 @@ static int dtSplitUp(tid_t tid,
4068 + struct tlock *tlck;
4069 + struct lv *lv;
4070 + int quota_allocation = 0;
4071 ++ int dlimit_allocation = 0;
4072 +
4073 + /* get split page */
4074 + smp = split->mp;
4075 +@@ -1033,6 +1041,12 @@ static int dtSplitUp(tid_t tid,
4076 + }
4077 + quota_allocation += n;
4078 +
4079 ++ if (DLIMIT_ALLOC_BLOCK(ip, n)) {
4080 ++ rc = -ENOSPC;
4081 ++ goto extendOut;
4082 ++ }
4083 ++ dlimit_allocation += n;
4084 ++
4085 + if ((rc = dbReAlloc(sbi->ipbmap, xaddr, (s64) xlen,
4086 + (s64) n, &nxaddr)))
4087 + goto extendOut;
4088 +@@ -1306,6 +1320,9 @@ static int dtSplitUp(tid_t tid,
4089 + freeKeyName:
4090 + kfree(key.name);
4091 +
4092 ++ /* Rollback dlimit allocation */
4093 ++ if (rc && dlimit_allocation)
4094 ++ DLIMIT_FREE_BLOCK(ip, dlimit_allocation);
4095 + /* Rollback quota allocation */
4096 + if (rc && quota_allocation)
4097 + DQUOT_FREE_BLOCK(ip, quota_allocation);
4098 +@@ -1373,6 +1390,12 @@ static int dtSplitPage(tid_t tid, struct
4099 + release_metapage(rmp);
4100 + return -EDQUOT;
4101 + }
4102 ++ /* Allocate blocks to dlimit. */
4103 ++ if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4104 ++ DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4105 ++ release_metapage(rmp);
4106 ++ return -ENOSPC;
4107 ++ }
4108 +
4109 + jfs_info("dtSplitPage: ip:0x%p smp:0x%p rmp:0x%p", ip, smp, rmp);
4110 +
4111 +@@ -1920,6 +1943,12 @@ static int dtSplitRoot(tid_t tid,
4112 + release_metapage(rmp);
4113 + return -EDQUOT;
4114 + }
4115 ++ /* Allocate blocks to dlimit. */
4116 ++ if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4117 ++ DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4118 ++ release_metapage(rmp);
4119 ++ return -ENOSPC;
4120 ++ }
4121 +
4122 + BT_MARK_DIRTY(rmp, ip);
4123 + /*
4124 +@@ -2286,6 +2315,8 @@ static int dtDeleteUp(tid_t tid, struct
4125 +
4126 + xlen = lengthPXD(&fp->header.self);
4127 +
4128 ++ /* Free dlimit allocation. */
4129 ++ DLIMIT_FREE_BLOCK(ip, xlen);
4130 + /* Free quota allocation. */
4131 + DQUOT_FREE_BLOCK(ip, xlen);
4132 +
4133 +@@ -2362,6 +2393,8 @@ static int dtDeleteUp(tid_t tid, struct
4134 +
4135 + xlen = lengthPXD(&p->header.self);
4136 +
4137 ++ /* Free dlimit allocation */
4138 ++ DLIMIT_FREE_BLOCK(ip, xlen);
4139 + /* Free quota allocation */
4140 + DQUOT_FREE_BLOCK(ip, xlen);
4141 +
4142 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_extent.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_extent.c
4143 +--- linux-2.6.29.4/fs/jfs/jfs_extent.c 2008-12-25 00:26:37.000000000 +0100
4144 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_extent.c 2009-02-22 22:54:25.000000000 +0100
4145 +@@ -18,6 +18,7 @@
4146 +
4147 + #include <linux/fs.h>
4148 + #include <linux/quotaops.h>
4149 ++#include <linux/vs_dlimit.h>
4150 + #include "jfs_incore.h"
4151 + #include "jfs_inode.h"
4152 + #include "jfs_superblock.h"
4153 +@@ -147,6 +148,14 @@ extAlloc(struct inode *ip, s64 xlen, s64
4154 + return -EDQUOT;
4155 + }
4156 +
4157 ++ /* Allocate blocks to dlimit. */
4158 ++ if (DLIMIT_ALLOC_BLOCK(ip, nxlen)) {
4159 ++ DQUOT_FREE_BLOCK(ip, nxlen);
4160 ++ dbFree(ip, nxaddr, (s64) nxlen);
4161 ++ mutex_unlock(&JFS_IP(ip)->commit_mutex);
4162 ++ return -ENOSPC;
4163 ++ }
4164 ++
4165 + /* determine the value of the extent flag */
4166 + xflag = abnr ? XAD_NOTRECORDED : 0;
4167 +
4168 +@@ -164,6 +173,7 @@ extAlloc(struct inode *ip, s64 xlen, s64
4169 + */
4170 + if (rc) {
4171 + dbFree(ip, nxaddr, nxlen);
4172 ++ DLIMIT_FREE_BLOCK(ip, nxlen);
4173 + DQUOT_FREE_BLOCK(ip, nxlen);
4174 + mutex_unlock(&JFS_IP(ip)->commit_mutex);
4175 + return (rc);
4176 +@@ -261,6 +271,13 @@ int extRealloc(struct inode *ip, s64 nxl
4177 + mutex_unlock(&JFS_IP(ip)->commit_mutex);
4178 + return -EDQUOT;
4179 + }
4180 ++ /* Allocate blocks to dlimit. */
4181 ++ if (DLIMIT_ALLOC_BLOCK(ip, nxlen)) {
4182 ++ DQUOT_FREE_BLOCK(ip, nxlen);
4183 ++ dbFree(ip, nxaddr, (s64) nxlen);
4184 ++ up(&JFS_IP(ip)->commit_sem);
4185 ++ return -ENOSPC;
4186 ++ }
4187 +
4188 + delta = nxlen - xlen;
4189 +
4190 +@@ -297,6 +314,7 @@ int extRealloc(struct inode *ip, s64 nxl
4191 + /* extend the extent */
4192 + if ((rc = xtExtend(0, ip, xoff + xlen, (int) nextend, 0))) {
4193 + dbFree(ip, xaddr + xlen, delta);
4194 ++ DLIMIT_FREE_BLOCK(ip, nxlen);
4195 + DQUOT_FREE_BLOCK(ip, nxlen);
4196 + goto exit;
4197 + }
4198 +@@ -308,6 +326,7 @@ int extRealloc(struct inode *ip, s64 nxl
4199 + */
4200 + if ((rc = xtTailgate(0, ip, xoff, (int) ntail, nxaddr, 0))) {
4201 + dbFree(ip, nxaddr, nxlen);
4202 ++ DLIMIT_FREE_BLOCK(ip, nxlen);
4203 + DQUOT_FREE_BLOCK(ip, nxlen);
4204 + goto exit;
4205 + }
4206 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_filsys.h linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_filsys.h
4207 +--- linux-2.6.29.4/fs/jfs/jfs_filsys.h 2008-12-25 00:26:37.000000000 +0100
4208 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_filsys.h 2009-02-22 22:54:25.000000000 +0100
4209 +@@ -263,6 +263,7 @@
4210 + #define JFS_NAME_MAX 255
4211 + #define JFS_PATH_MAX BPSIZE
4212 +
4213 ++#define JFS_TAGGED 0x00800000 /* Context Tagging */
4214 +
4215 + /*
4216 + * file system state (superblock state)
4217 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_imap.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_imap.c
4218 +--- linux-2.6.29.4/fs/jfs/jfs_imap.c 2009-03-24 14:22:26.000000000 +0100
4219 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_imap.c 2009-03-24 14:48:35.000000000 +0100
4220 +@@ -45,6 +45,7 @@
4221 + #include <linux/buffer_head.h>
4222 + #include <linux/pagemap.h>
4223 + #include <linux/quotaops.h>
4224 ++#include <linux/vs_tag.h>
4225 +
4226 + #include "jfs_incore.h"
4227 + #include "jfs_inode.h"
4228 +@@ -3062,6 +3063,8 @@ static int copy_from_dinode(struct dinod
4229 + {
4230 + struct jfs_inode_info *jfs_ip = JFS_IP(ip);
4231 + struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
4232 ++ uid_t uid;
4233 ++ gid_t gid;
4234 +
4235 + jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
4236 + jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
4237 +@@ -3082,14 +3085,18 @@ static int copy_from_dinode(struct dinod
4238 + }
4239 + ip->i_nlink = le32_to_cpu(dip->di_nlink);
4240 +
4241 +- jfs_ip->saved_uid = le32_to_cpu(dip->di_uid);
4242 ++ uid = le32_to_cpu(dip->di_uid);
4243 ++ gid = le32_to_cpu(dip->di_gid);
4244 ++ ip->i_tag = INOTAG_TAG(DX_TAG(ip), uid, gid, 0);
4245 ++
4246 ++ jfs_ip->saved_uid = INOTAG_UID(DX_TAG(ip), uid, gid);
4247 + if (sbi->uid == -1)
4248 + ip->i_uid = jfs_ip->saved_uid;
4249 + else {
4250 + ip->i_uid = sbi->uid;
4251 + }
4252 +
4253 +- jfs_ip->saved_gid = le32_to_cpu(dip->di_gid);
4254 ++ jfs_ip->saved_gid = INOTAG_GID(DX_TAG(ip), uid, gid);
4255 + if (sbi->gid == -1)
4256 + ip->i_gid = jfs_ip->saved_gid;
4257 + else {
4258 +@@ -3154,14 +3161,12 @@ static void copy_to_dinode(struct dinode
4259 + dip->di_size = cpu_to_le64(ip->i_size);
4260 + dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
4261 + dip->di_nlink = cpu_to_le32(ip->i_nlink);
4262 +- if (sbi->uid == -1)
4263 +- dip->di_uid = cpu_to_le32(ip->i_uid);
4264 +- else
4265 +- dip->di_uid = cpu_to_le32(jfs_ip->saved_uid);
4266 +- if (sbi->gid == -1)
4267 +- dip->di_gid = cpu_to_le32(ip->i_gid);
4268 +- else
4269 +- dip->di_gid = cpu_to_le32(jfs_ip->saved_gid);
4270 ++
4271 ++ dip->di_uid = cpu_to_le32(TAGINO_UID(DX_TAG(ip),
4272 ++ (sbi->uid == -1) ? ip->i_uid : jfs_ip->saved_uid, ip->i_tag));
4273 ++ dip->di_gid = cpu_to_le32(TAGINO_GID(DX_TAG(ip),
4274 ++ (sbi->gid == -1) ? ip->i_gid : jfs_ip->saved_gid, ip->i_tag));
4275 ++
4276 + jfs_get_inode_flags(jfs_ip);
4277 + /*
4278 + * mode2 is only needed for storing the higher order bits.
4279 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_inode.c
4280 +--- linux-2.6.29.4/fs/jfs/jfs_inode.c 2009-03-24 14:22:26.000000000 +0100
4281 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_inode.c 2009-03-24 15:58:01.000000000 +0100
4282 +@@ -18,6 +18,8 @@
4283 +
4284 + #include <linux/fs.h>
4285 + #include <linux/quotaops.h>
4286 ++#include <linux/vs_dlimit.h>
4287 ++#include <linux/vs_tag.h>
4288 + #include "jfs_incore.h"
4289 + #include "jfs_inode.h"
4290 + #include "jfs_filsys.h"
4291 +@@ -30,29 +32,46 @@ void jfs_set_inode_flags(struct inode *i
4292 + {
4293 + unsigned int flags = JFS_IP(inode)->mode2;
4294 +
4295 +- inode->i_flags &= ~(S_IMMUTABLE | S_APPEND |
4296 +- S_NOATIME | S_DIRSYNC | S_SYNC);
4297 ++ inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
4298 ++ S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
4299 +
4300 + if (flags & JFS_IMMUTABLE_FL)
4301 + inode->i_flags |= S_IMMUTABLE;
4302 ++ if (flags & JFS_IXUNLINK_FL)
4303 ++ inode->i_flags |= S_IXUNLINK;
4304 ++
4305 ++ if (flags & JFS_SYNC_FL)
4306 ++ inode->i_flags |= S_SYNC;
4307 + if (flags & JFS_APPEND_FL)
4308 + inode->i_flags |= S_APPEND;
4309 + if (flags & JFS_NOATIME_FL)
4310 + inode->i_flags |= S_NOATIME;
4311 + if (flags & JFS_DIRSYNC_FL)
4312 + inode->i_flags |= S_DIRSYNC;
4313 +- if (flags & JFS_SYNC_FL)
4314 +- inode->i_flags |= S_SYNC;
4315 ++
4316 ++ inode->i_vflags &= ~(V_BARRIER | V_COW);
4317 ++
4318 ++ if (flags & JFS_BARRIER_FL)
4319 ++ inode->i_vflags |= V_BARRIER;
4320 ++ if (flags & JFS_COW_FL)
4321 ++ inode->i_vflags |= V_COW;
4322 + }
4323 +
4324 + void jfs_get_inode_flags(struct jfs_inode_info *jfs_ip)
4325 + {
4326 + unsigned int flags = jfs_ip->vfs_inode.i_flags;
4327 ++ unsigned int vflags = jfs_ip->vfs_inode.i_vflags;
4328 ++
4329 ++ jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL |
4330 ++ JFS_APPEND_FL | JFS_NOATIME_FL |
4331 ++ JFS_DIRSYNC_FL | JFS_SYNC_FL |
4332 ++ JFS_BARRIER_FL | JFS_COW_FL);
4333 +
4334 +- jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_APPEND_FL | JFS_NOATIME_FL |
4335 +- JFS_DIRSYNC_FL | JFS_SYNC_FL);
4336 + if (flags & S_IMMUTABLE)
4337 + jfs_ip->mode2 |= JFS_IMMUTABLE_FL;
4338 ++ if (flags & S_IXUNLINK)
4339 ++ jfs_ip->mode2 |= JFS_IXUNLINK_FL;
4340 ++
4341 + if (flags & S_APPEND)
4342 + jfs_ip->mode2 |= JFS_APPEND_FL;
4343 + if (flags & S_NOATIME)
4344 +@@ -61,6 +80,19 @@ void jfs_get_inode_flags(struct jfs_inod
4345 + jfs_ip->mode2 |= JFS_DIRSYNC_FL;
4346 + if (flags & S_SYNC)
4347 + jfs_ip->mode2 |= JFS_SYNC_FL;
4348 ++
4349 ++ if (vflags & V_BARRIER)
4350 ++ jfs_ip->mode2 |= JFS_BARRIER_FL;
4351 ++ if (vflags & V_COW)
4352 ++ jfs_ip->mode2 |= JFS_COW_FL;
4353 ++}
4354 ++
4355 ++int jfs_sync_flags(struct inode *inode)
4356 ++{
4357 ++ jfs_get_inode_flags(JFS_IP(inode));
4358 ++ inode->i_ctime = CURRENT_TIME;
4359 ++ mark_inode_dirty(inode);
4360 ++ return 0;
4361 + }
4362 +
4363 + /*
4364 +@@ -113,6 +145,12 @@ struct inode *ialloc(struct inode *paren
4365 + jfs_inode->saved_uid = inode->i_uid;
4366 + jfs_inode->saved_gid = inode->i_gid;
4367 +
4368 ++ inode->i_tag = dx_current_fstag(sb);
4369 ++ if (DLIMIT_ALLOC_INODE(inode)) {
4370 ++ rc = -ENOSPC;
4371 ++ goto fail_drop2;
4372 ++ }
4373 ++
4374 + /*
4375 + * Allocate inode to quota.
4376 + */
4377 +@@ -162,6 +200,8 @@ struct inode *ialloc(struct inode *paren
4378 + return inode;
4379 +
4380 + fail_drop:
4381 ++ DLIMIT_FREE_INODE(inode);
4382 ++fail_drop2:
4383 + DQUOT_DROP(inode);
4384 + inode->i_flags |= S_NOQUOTA;
4385 + fail_unlock:
4386 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_inode.h linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_inode.h
4387 +--- linux-2.6.29.4/fs/jfs/jfs_inode.h 2008-12-25 00:26:37.000000000 +0100
4388 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_inode.h 2009-02-22 22:54:25.000000000 +0100
4389 +@@ -39,6 +39,7 @@ extern struct dentry *jfs_fh_to_dentry(s
4390 + extern struct dentry *jfs_fh_to_parent(struct super_block *sb, struct fid *fid,
4391 + int fh_len, int fh_type);
4392 + extern void jfs_set_inode_flags(struct inode *);
4393 ++extern int jfs_sync_flags(struct inode *);
4394 + extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
4395 +
4396 + extern const struct address_space_operations jfs_aops;
4397 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/jfs_xtree.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_xtree.c
4398 +--- linux-2.6.29.4/fs/jfs/jfs_xtree.c 2008-12-25 00:26:37.000000000 +0100
4399 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/jfs_xtree.c 2009-02-22 22:54:25.000000000 +0100
4400 +@@ -23,6 +23,7 @@
4401 + #include <linux/module.h>
4402 + #include <linux/quotaops.h>
4403 + #include <linux/seq_file.h>
4404 ++#include <linux/vs_dlimit.h>
4405 + #include "jfs_incore.h"
4406 + #include "jfs_filsys.h"
4407 + #include "jfs_metapage.h"
4408 +@@ -848,7 +849,12 @@ int xtInsert(tid_t tid, /* transaction
4409 + hint = 0;
4410 + if ((rc = DQUOT_ALLOC_BLOCK(ip, xlen)))
4411 + goto out;
4412 ++ if ((rc = DLIMIT_ALLOC_BLOCK(ip, xlen))) {
4413 ++ DQUOT_FREE_BLOCK(ip, xlen);
4414 ++ goto out;
4415 ++ }
4416 + if ((rc = dbAlloc(ip, hint, (s64) xlen, &xaddr))) {
4417 ++ DLIMIT_FREE_BLOCK(ip, xlen);
4418 + DQUOT_FREE_BLOCK(ip, xlen);
4419 + goto out;
4420 + }
4421 +@@ -878,6 +884,7 @@ int xtInsert(tid_t tid, /* transaction
4422 + /* undo data extent allocation */
4423 + if (*xaddrp == 0) {
4424 + dbFree(ip, xaddr, (s64) xlen);
4425 ++ DLIMIT_FREE_BLOCK(ip, xlen);
4426 + DQUOT_FREE_BLOCK(ip, xlen);
4427 + }
4428 + return rc;
4429 +@@ -1234,6 +1241,7 @@ xtSplitPage(tid_t tid, struct inode *ip,
4430 + struct tlock *tlck;
4431 + struct xtlock *sxtlck = NULL, *rxtlck = NULL;
4432 + int quota_allocation = 0;
4433 ++ int dlimit_allocation = 0;
4434 +
4435 + smp = split->mp;
4436 + sp = XT_PAGE(ip, smp);
4437 +@@ -1253,6 +1261,13 @@ xtSplitPage(tid_t tid, struct inode *ip,
4438 +
4439 + quota_allocation += lengthPXD(pxd);
4440 +
4441 ++ /* Allocate blocks to dlimit. */
4442 ++ if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4443 ++ rc = -ENOSPC;
4444 ++ goto clean_up;
4445 ++ }
4446 ++ dlimit_allocation += lengthPXD(pxd);
4447 ++
4448 + /*
4449 + * allocate the new right page for the split
4450 + */
4451 +@@ -1454,6 +1469,9 @@ xtSplitPage(tid_t tid, struct inode *ip,
4452 +
4453 + clean_up:
4454 +
4455 ++ /* Rollback dlimit allocation. */
4456 ++ if (dlimit_allocation)
4457 ++ DLIMIT_FREE_BLOCK(ip, dlimit_allocation);
4458 + /* Rollback quota allocation. */
4459 + if (quota_allocation)
4460 + DQUOT_FREE_BLOCK(ip, quota_allocation);
4461 +@@ -1517,6 +1535,12 @@ xtSplitRoot(tid_t tid,
4462 + release_metapage(rmp);
4463 + return -EDQUOT;
4464 + }
4465 ++ /* Allocate blocks to dlimit. */
4466 ++ if (DLIMIT_ALLOC_BLOCK(ip, lengthPXD(pxd))) {
4467 ++ DQUOT_FREE_BLOCK(ip, lengthPXD(pxd));
4468 ++ release_metapage(rmp);
4469 ++ return -ENOSPC;
4470 ++ }
4471 +
4472 + jfs_info("xtSplitRoot: ip:0x%p rmp:0x%p", ip, rmp);
4473 +
4474 +@@ -3940,6 +3964,8 @@ s64 xtTruncate(tid_t tid, struct inode *
4475 + else
4476 + ip->i_size = newsize;
4477 +
4478 ++ /* update dlimit allocation to reflect freed blocks */
4479 ++ DLIMIT_FREE_BLOCK(ip, nfreed);
4480 + /* update quota allocation to reflect freed blocks */
4481 + DQUOT_FREE_BLOCK(ip, nfreed);
4482 +
4483 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/namei.c
4484 +--- linux-2.6.29.4/fs/jfs/namei.c 2009-03-24 14:22:26.000000000 +0100
4485 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/namei.c 2009-03-24 14:48:35.000000000 +0100
4486 +@@ -21,6 +21,7 @@
4487 + #include <linux/ctype.h>
4488 + #include <linux/quotaops.h>
4489 + #include <linux/exportfs.h>
4490 ++#include <linux/vs_tag.h>
4491 + #include "jfs_incore.h"
4492 + #include "jfs_superblock.h"
4493 + #include "jfs_inode.h"
4494 +@@ -1476,6 +1477,7 @@ static struct dentry *jfs_lookup(struct
4495 + return ERR_CAST(ip);
4496 + }
4497 +
4498 ++ dx_propagate_tag(nd, ip);
4499 + dentry = d_splice_alias(ip, dentry);
4500 +
4501 + if (dentry && (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2))
4502 +@@ -1545,6 +1547,7 @@ const struct inode_operations jfs_dir_in
4503 + .setattr = jfs_setattr,
4504 + .permission = jfs_permission,
4505 + #endif
4506 ++ .sync_flags = jfs_sync_flags,
4507 + };
4508 +
4509 + const struct file_operations jfs_dir_operations = {
4510 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/super.c
4511 +--- linux-2.6.29.4/fs/jfs/super.c 2009-03-24 14:22:26.000000000 +0100
4512 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/super.c 2009-03-24 14:48:35.000000000 +0100
4513 +@@ -196,7 +196,8 @@ static void jfs_put_super(struct super_b
4514 + enum {
4515 + Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
4516 + Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
4517 +- Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
4518 ++ Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask,
4519 ++ Opt_tag, Opt_notag, Opt_tagid
4520 + };
4521 +
4522 + static const match_table_t tokens = {
4523 +@@ -206,6 +207,10 @@ static const match_table_t tokens = {
4524 + {Opt_resize, "resize=%u"},
4525 + {Opt_resize_nosize, "resize"},
4526 + {Opt_errors, "errors=%s"},
4527 ++ {Opt_tag, "tag"},
4528 ++ {Opt_notag, "notag"},
4529 ++ {Opt_tagid, "tagid=%u"},
4530 ++ {Opt_tag, "tagxid"},
4531 + {Opt_ignore, "noquota"},
4532 + {Opt_ignore, "quota"},
4533 + {Opt_usrquota, "usrquota"},
4534 +@@ -340,6 +345,20 @@ static int parse_options(char *options,
4535 + }
4536 + break;
4537 + }
4538 ++#ifndef CONFIG_TAGGING_NONE
4539 ++ case Opt_tag:
4540 ++ *flag |= JFS_TAGGED;
4541 ++ break;
4542 ++ case Opt_notag:
4543 ++ *flag &= JFS_TAGGED;
4544 ++ break;
4545 ++#endif
4546 ++#ifdef CONFIG_PROPAGATE
4547 ++ case Opt_tagid:
4548 ++ /* use args[0] */
4549 ++ *flag |= JFS_TAGGED;
4550 ++ break;
4551 ++#endif
4552 + default:
4553 + printk("jfs: Unrecognized mount option \"%s\" "
4554 + " or missing value\n", p);
4555 +@@ -370,6 +389,13 @@ static int jfs_remount(struct super_bloc
4556 + if (!parse_options(data, sb, &newLVSize, &flag)) {
4557 + return -EINVAL;
4558 + }
4559 ++
4560 ++ if ((flag & JFS_TAGGED) && !(sb->s_flags & MS_TAGGED)) {
4561 ++ printk(KERN_ERR "JFS: %s: tagging not permitted on remount.\n",
4562 ++ sb->s_id);
4563 ++ return -EINVAL;
4564 ++ }
4565 ++
4566 + if (newLVSize) {
4567 + if (sb->s_flags & MS_RDONLY) {
4568 + printk(KERN_ERR
4569 +@@ -441,6 +467,9 @@ static int jfs_fill_super(struct super_b
4570 + #ifdef CONFIG_JFS_POSIX_ACL
4571 + sb->s_flags |= MS_POSIXACL;
4572 + #endif
4573 ++ /* map mount option tagxid */
4574 ++ if (sbi->flag & JFS_TAGGED)
4575 ++ sb->s_flags |= MS_TAGGED;
4576 +
4577 + if (newLVSize) {
4578 + printk(KERN_ERR "resize option for remount only\n");
4579 +diff -NurpP --minimal linux-2.6.29.4/fs/jfs/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/xattr.c
4580 +--- linux-2.6.29.4/fs/jfs/xattr.c 2008-12-25 00:26:37.000000000 +0100
4581 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/jfs/xattr.c 2009-02-22 22:54:25.000000000 +0100
4582 +@@ -23,6 +23,7 @@
4583 + #include <linux/posix_acl_xattr.h>
4584 + #include <linux/quotaops.h>
4585 + #include <linux/security.h>
4586 ++#include <linux/vs_dlimit.h>
4587 + #include "jfs_incore.h"
4588 + #include "jfs_superblock.h"
4589 + #include "jfs_dmap.h"
4590 +@@ -263,9 +264,16 @@ static int ea_write(struct inode *ip, st
4591 + if (DQUOT_ALLOC_BLOCK(ip, nblocks)) {
4592 + return -EDQUOT;
4593 + }
4594 ++ /* Allocate new blocks to dlimit. */
4595 ++ if (DLIMIT_ALLOC_BLOCK(ip, nblocks)) {
4596 ++ DQUOT_FREE_BLOCK(ip, nblocks);
4597 ++ return -ENOSPC;
4598 ++ }
4599 +
4600 + rc = dbAlloc(ip, INOHINT(ip), nblocks, &blkno);
4601 + if (rc) {
4602 ++ /*Rollback dlimit allocation. */
4603 ++ DLIMIT_FREE_BLOCK(ip, nblocks);
4604 + /*Rollback quota allocation. */
4605 + DQUOT_FREE_BLOCK(ip, nblocks);
4606 + return rc;
4607 +@@ -332,6 +340,8 @@ static int ea_write(struct inode *ip, st
4608 +
4609 + failed:
4610 + /* Rollback quota allocation. */
4611 ++ DLIMIT_FREE_BLOCK(ip, nblocks);
4612 ++ /* Rollback quota allocation. */
4613 + DQUOT_FREE_BLOCK(ip, nblocks);
4614 +
4615 + dbFree(ip, blkno, nblocks);
4616 +@@ -468,6 +478,7 @@ static int ea_get(struct inode *inode, s
4617 + s64 blkno;
4618 + int rc;
4619 + int quota_allocation = 0;
4620 ++ int dlimit_allocation = 0;
4621 +
4622 + /* When fsck.jfs clears a bad ea, it doesn't clear the size */
4623 + if (ji->ea.flag == 0)
4624 +@@ -543,6 +554,12 @@ static int ea_get(struct inode *inode, s
4625 +
4626 + quota_allocation = blocks_needed;
4627 +
4628 ++ /* Allocate new blocks to dlimit. */
4629 ++ rc = -ENOSPC;
4630 ++ if (DLIMIT_ALLOC_BLOCK(inode, blocks_needed))
4631 ++ goto clean_up;
4632 ++ dlimit_allocation = blocks_needed;
4633 ++
4634 + rc = dbAlloc(inode, INOHINT(inode), (s64) blocks_needed,
4635 + &blkno);
4636 + if (rc)
4637 +@@ -600,6 +617,9 @@ static int ea_get(struct inode *inode, s
4638 + return ea_size;
4639 +
4640 + clean_up:
4641 ++ /* Rollback dlimit allocation */
4642 ++ if (dlimit_allocation)
4643 ++ DLIMIT_FREE_BLOCK(inode, dlimit_allocation);
4644 + /* Rollback quota allocation */
4645 + if (quota_allocation)
4646 + DQUOT_FREE_BLOCK(inode, quota_allocation);
4647 +@@ -676,8 +696,10 @@ static int ea_put(tid_t tid, struct inod
4648 + }
4649 +
4650 + /* If old blocks exist, they must be removed from quota allocation. */
4651 +- if (old_blocks)
4652 ++ if (old_blocks) {
4653 ++ DLIMIT_FREE_BLOCK(inode, old_blocks);
4654 + DQUOT_FREE_BLOCK(inode, old_blocks);
4655 ++ }
4656 +
4657 + inode->i_ctime = CURRENT_TIME;
4658 +
4659 +diff -NurpP --minimal linux-2.6.29.4/fs/libfs.c linux-2.6.29.4-vs2.3.0.36.14/fs/libfs.c
4660 +--- linux-2.6.29.4/fs/libfs.c 2009-03-24 14:22:26.000000000 +0100
4661 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/libfs.c 2009-03-24 14:48:35.000000000 +0100
4662 +@@ -125,7 +125,8 @@ static inline unsigned char dt_type(stru
4663 + * both impossible due to the lock on directory.
4664 + */
4665 +
4666 +-int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
4667 ++static inline int do_dcache_readdir_filter(struct file *filp,
4668 ++ void *dirent, filldir_t filldir, int (*filter)(struct dentry *dentry))
4669 + {
4670 + struct dentry *dentry = filp->f_path.dentry;
4671 + struct dentry *cursor = filp->private_data;
4672 +@@ -158,6 +159,8 @@ int dcache_readdir(struct file * filp, v
4673 + next = list_entry(p, struct dentry, d_u.d_child);
4674 + if (d_unhashed(next) || !next->d_inode)
4675 + continue;
4676 ++ if (filter && !filter(next))
4677 ++ continue;
4678 +
4679 + spin_unlock(&dcache_lock);
4680 + if (filldir(dirent, next->d_name.name,
4681 +@@ -176,6 +179,18 @@ int dcache_readdir(struct file * filp, v
4682 + return 0;
4683 + }
4684 +
4685 ++int dcache_readdir(struct file *filp, void *dirent, filldir_t filldir)
4686 ++{
4687 ++ return do_dcache_readdir_filter(filp, dirent, filldir, NULL);
4688 ++}
4689 ++
4690 ++int dcache_readdir_filter(struct file *filp, void *dirent, filldir_t filldir,
4691 ++ int (*filter)(struct dentry *))
4692 ++{
4693 ++ return do_dcache_readdir_filter(filp, dirent, filldir, filter);
4694 ++}
4695 ++
4696 ++
4697 + ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
4698 + {
4699 + return -EISDIR;
4700 +@@ -796,6 +811,7 @@ EXPORT_SYMBOL(dcache_dir_close);
4701 + EXPORT_SYMBOL(dcache_dir_lseek);
4702 + EXPORT_SYMBOL(dcache_dir_open);
4703 + EXPORT_SYMBOL(dcache_readdir);
4704 ++EXPORT_SYMBOL(dcache_readdir_filter);
4705 + EXPORT_SYMBOL(generic_read_dir);
4706 + EXPORT_SYMBOL(get_sb_pseudo);
4707 + EXPORT_SYMBOL(simple_write_begin);
4708 +diff -NurpP --minimal linux-2.6.29.4/fs/locks.c linux-2.6.29.4-vs2.3.0.36.14/fs/locks.c
4709 +--- linux-2.6.29.4/fs/locks.c 2009-03-24 14:22:26.000000000 +0100
4710 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/locks.c 2009-03-24 14:48:35.000000000 +0100
4711 +@@ -127,6 +127,8 @@
4712 + #include <linux/time.h>
4713 + #include <linux/rcupdate.h>
4714 + #include <linux/pid_namespace.h>
4715 ++#include <linux/vs_base.h>
4716 ++#include <linux/vs_limit.h>
4717 +
4718 + #include <asm/uaccess.h>
4719 +
4720 +@@ -148,6 +150,8 @@ static struct kmem_cache *filelock_cache
4721 + /* Allocate an empty lock structure. */
4722 + static struct file_lock *locks_alloc_lock(void)
4723 + {
4724 ++ if (!vx_locks_avail(1))
4725 ++ return NULL;
4726 + return kmem_cache_alloc(filelock_cache, GFP_KERNEL);
4727 + }
4728 +
4729 +@@ -173,6 +177,7 @@ static void locks_free_lock(struct file_
4730 + BUG_ON(!list_empty(&fl->fl_block));
4731 + BUG_ON(!list_empty(&fl->fl_link));
4732 +
4733 ++ vx_locks_dec(fl);
4734 + locks_release_private(fl);
4735 + kmem_cache_free(filelock_cache, fl);
4736 + }
4737 +@@ -193,6 +198,7 @@ void locks_init_lock(struct file_lock *f
4738 + fl->fl_start = fl->fl_end = 0;
4739 + fl->fl_ops = NULL;
4740 + fl->fl_lmops = NULL;
4741 ++ fl->fl_xid = -1;
4742 + }
4743 +
4744 + EXPORT_SYMBOL(locks_init_lock);
4745 +@@ -247,6 +253,7 @@ void locks_copy_lock(struct file_lock *n
4746 + new->fl_file = fl->fl_file;
4747 + new->fl_ops = fl->fl_ops;
4748 + new->fl_lmops = fl->fl_lmops;
4749 ++ new->fl_xid = fl->fl_xid;
4750 +
4751 + locks_copy_private(new, fl);
4752 + }
4753 +@@ -285,6 +292,11 @@ static int flock_make_lock(struct file *
4754 + fl->fl_flags = FL_FLOCK;
4755 + fl->fl_type = type;
4756 + fl->fl_end = OFFSET_MAX;
4757 ++
4758 ++ vxd_assert(filp->f_xid == vx_current_xid(),
4759 ++ "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4760 ++ fl->fl_xid = filp->f_xid;
4761 ++ vx_locks_inc(fl);
4762 +
4763 + *lock = fl;
4764 + return 0;
4765 +@@ -450,6 +462,7 @@ static int lease_init(struct file *filp,
4766 +
4767 + fl->fl_owner = current->files;
4768 + fl->fl_pid = current->tgid;
4769 ++ fl->fl_xid = vx_current_xid();
4770 +
4771 + fl->fl_file = filp;
4772 + fl->fl_flags = FL_LEASE;
4773 +@@ -469,6 +482,11 @@ static struct file_lock *lease_alloc(str
4774 + if (fl == NULL)
4775 + return ERR_PTR(error);
4776 +
4777 ++ fl->fl_xid = vx_current_xid();
4778 ++ if (filp)
4779 ++ vxd_assert(filp->f_xid == fl->fl_xid,
4780 ++ "f_xid(%d) == fl_xid(%d)", filp->f_xid, fl->fl_xid);
4781 ++ vx_locks_inc(fl);
4782 + error = lease_init(filp, type, fl);
4783 + if (error) {
4784 + locks_free_lock(fl);
4785 +@@ -769,6 +787,7 @@ static int flock_lock_file(struct file *
4786 + if (found)
4787 + cond_resched_bkl();
4788 +
4789 ++ new_fl->fl_xid = -1;
4790 + find_conflict:
4791 + for_each_lock(inode, before) {
4792 + struct file_lock *fl = *before;
4793 +@@ -789,6 +808,7 @@ find_conflict:
4794 + goto out;
4795 + locks_copy_lock(new_fl, request);
4796 + locks_insert_lock(before, new_fl);
4797 ++ vx_locks_inc(new_fl);
4798 + new_fl = NULL;
4799 + error = 0;
4800 +
4801 +@@ -799,7 +819,8 @@ out:
4802 + return error;
4803 + }
4804 +
4805 +-static int __posix_lock_file(struct inode *inode, struct file_lock *request, struct file_lock *conflock)
4806 ++static int __posix_lock_file(struct inode *inode, struct file_lock *request,
4807 ++ struct file_lock *conflock, xid_t xid)
4808 + {
4809 + struct file_lock *fl;
4810 + struct file_lock *new_fl = NULL;
4811 +@@ -809,6 +830,8 @@ static int __posix_lock_file(struct inod
4812 + struct file_lock **before;
4813 + int error, added = 0;
4814 +
4815 ++ vxd_assert(xid == vx_current_xid(),
4816 ++ "xid(%d) == current(%d)", xid, vx_current_xid());
4817 + /*
4818 + * We may need two file_lock structures for this operation,
4819 + * so we get them in advance to avoid races.
4820 +@@ -819,7 +842,11 @@ static int __posix_lock_file(struct inod
4821 + (request->fl_type != F_UNLCK ||
4822 + request->fl_start != 0 || request->fl_end != OFFSET_MAX)) {
4823 + new_fl = locks_alloc_lock();
4824 ++ new_fl->fl_xid = xid;
4825 ++ vx_locks_inc(new_fl);
4826 + new_fl2 = locks_alloc_lock();
4827 ++ new_fl2->fl_xid = xid;
4828 ++ vx_locks_inc(new_fl2);
4829 + }
4830 +
4831 + lock_kernel();
4832 +@@ -1018,7 +1045,8 @@ static int __posix_lock_file(struct inod
4833 + int posix_lock_file(struct file *filp, struct file_lock *fl,
4834 + struct file_lock *conflock)
4835 + {
4836 +- return __posix_lock_file(filp->f_path.dentry->d_inode, fl, conflock);
4837 ++ return __posix_lock_file(filp->f_path.dentry->d_inode,
4838 ++ fl, conflock, filp->f_xid);
4839 + }
4840 + EXPORT_SYMBOL(posix_lock_file);
4841 +
4842 +@@ -1108,7 +1136,7 @@ int locks_mandatory_area(int read_write,
4843 + fl.fl_end = offset + count - 1;
4844 +
4845 + for (;;) {
4846 +- error = __posix_lock_file(inode, &fl, NULL);
4847 ++ error = __posix_lock_file(inode, &fl, NULL, filp->f_xid);
4848 + if (error != FILE_LOCK_DEFERRED)
4849 + break;
4850 + error = wait_event_interruptible(fl.fl_wait, !fl.fl_next);
4851 +@@ -1423,6 +1451,7 @@ int generic_setlease(struct file *filp,
4852 +
4853 + locks_copy_lock(new_fl, lease);
4854 + locks_insert_lock(before, new_fl);
4855 ++ vx_locks_inc(new_fl);
4856 +
4857 + *flp = new_fl;
4858 + return 0;
4859 +@@ -1778,6 +1807,11 @@ int fcntl_setlk(unsigned int fd, struct
4860 + if (file_lock == NULL)
4861 + return -ENOLCK;
4862 +
4863 ++ vxd_assert(filp->f_xid == vx_current_xid(),
4864 ++ "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4865 ++ file_lock->fl_xid = filp->f_xid;
4866 ++ vx_locks_inc(file_lock);
4867 ++
4868 + /*
4869 + * This might block, so we do it before checking the inode.
4870 + */
4871 +@@ -1896,6 +1930,11 @@ int fcntl_setlk64(unsigned int fd, struc
4872 + if (file_lock == NULL)
4873 + return -ENOLCK;
4874 +
4875 ++ vxd_assert(filp->f_xid == vx_current_xid(),
4876 ++ "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4877 ++ file_lock->fl_xid = filp->f_xid;
4878 ++ vx_locks_inc(file_lock);
4879 ++
4880 + /*
4881 + * This might block, so we do it before checking the inode.
4882 + */
4883 +@@ -2161,8 +2200,11 @@ static int locks_show(struct seq_file *f
4884 +
4885 + lock_get_status(f, fl, (long)f->private, "");
4886 +
4887 +- list_for_each_entry(bfl, &fl->fl_block, fl_block)
4888 ++ list_for_each_entry(bfl, &fl->fl_block, fl_block) {
4889 ++ if (!vx_check(fl->fl_xid, VS_WATCH_P | VS_IDENT))
4890 ++ continue;
4891 + lock_get_status(f, bfl, (long)f->private, " ->");
4892 ++ }
4893 +
4894 + f->private++;
4895 + return 0;
4896 +diff -NurpP --minimal linux-2.6.29.4/fs/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/namei.c
4897 +--- linux-2.6.29.4/fs/namei.c 2009-05-23 23:16:52.000000000 +0200
4898 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/namei.c 2009-05-23 22:57:27.000000000 +0200
4899 +@@ -31,6 +31,14 @@
4900 + #include <linux/file.h>
4901 + #include <linux/fcntl.h>
4902 + #include <linux/device_cgroup.h>
4903 ++#include <linux/proc_fs.h>
4904 ++#include <linux/vserver/inode.h>
4905 ++#include <linux/vs_base.h>
4906 ++#include <linux/vs_tag.h>
4907 ++#include <linux/vs_cowbl.h>
4908 ++#include <linux/vs_device.h>
4909 ++#include <linux/vs_context.h>
4910 ++#include <linux/pid_namespace.h>
4911 + #include <asm/uaccess.h>
4912 +
4913 + #define ACC_MODE(x) ("\000\004\002\006"[(x)&O_ACCMODE])
4914 +@@ -167,6 +175,77 @@ void putname(const char *name)
4915 + EXPORT_SYMBOL(putname);
4916 + #endif
4917 +
4918 ++static inline int dx_barrier(struct inode *inode)
4919 ++{
4920 ++ if (IS_BARRIER(inode) && !vx_check(0, VS_ADMIN | VS_WATCH)) {
4921 ++ vxwprintk_task(1, "did hit the barrier.");
4922 ++ return 1;
4923 ++ }
4924 ++ return 0;
4925 ++}
4926 ++
4927 ++static int __dx_permission(struct inode *inode, int mask)
4928 ++{
4929 ++ if (dx_barrier(inode))
4930 ++ return -EACCES;
4931 ++
4932 ++ if (inode->i_sb->s_magic == DEVPTS_SUPER_MAGIC) {
4933 ++ /* devpts is xid tagged */
4934 ++ if (S_ISDIR(inode->i_mode) ||
4935 ++ vx_check((xid_t)inode->i_tag, VS_IDENT | VS_WATCH_P))
4936 ++ return 0;
4937 ++ }
4938 ++ else if (inode->i_sb->s_magic == PROC_SUPER_MAGIC) {
4939 ++ struct proc_dir_entry *de = PDE(inode);
4940 ++
4941 ++ if (de && !vx_hide_check(0, de->vx_flags))
4942 ++ goto out;
4943 ++
4944 ++ if ((mask & (MAY_WRITE | MAY_APPEND))) {
4945 ++ struct pid *pid;
4946 ++ struct task_struct *tsk;
4947 ++
4948 ++ if (vx_check(0, VS_ADMIN | VS_WATCH_P) ||
4949 ++ vx_flags(VXF_STATE_SETUP, 0))
4950 ++ return 0;
4951 ++
4952 ++ pid = PROC_I(inode)->pid;
4953 ++ if (!pid)
4954 ++ goto out;
4955 ++
4956 ++ tsk = pid_task(pid, PIDTYPE_PID);
4957 ++ vxdprintk(VXD_CBIT(tag, 0), "accessing %p[#%u]",
4958 ++ tsk, (tsk ? vx_task_xid(tsk) : 0));
4959 ++ if (tsk && vx_check(vx_task_xid(tsk), VS_IDENT | VS_WATCH_P))
4960 ++ return 0;
4961 ++ }
4962 ++ else {
4963 ++ /* FIXME: Should we block some entries here? */
4964 ++ return 0;
4965 ++ }
4966 ++ }
4967 ++ else {
4968 ++ if (dx_notagcheck(inode->i_sb) ||
4969 ++ dx_check(inode->i_tag, DX_HOSTID | DX_ADMIN | DX_WATCH |
4970 ++ DX_IDENT))
4971 ++ return 0;
4972 ++ }
4973 ++
4974 ++out:
4975 ++ return -EACCES;
4976 ++}
4977 ++
4978 ++int dx_permission(struct inode *inode, int mask)
4979 ++{
4980 ++ int ret = __dx_permission(inode, mask);
4981 ++ if (unlikely(ret)) {
4982 ++ vxwprintk_task(1, "denied %x access to %s:%p[#%d,%lu]",
4983 ++ mask, inode->i_sb->s_id, inode, inode->i_tag,
4984 ++ inode->i_ino);
4985 ++ }
4986 ++ return ret;
4987 ++}
4988 ++
4989 +
4990 + /**
4991 + * generic_permission - check for access rights on a Posix-like filesystem
4992 +@@ -253,10 +332,14 @@ int inode_permission(struct inode *inode
4993 + /*
4994 + * Nobody gets write access to an immutable file.
4995 + */
4996 +- if (IS_IMMUTABLE(inode))
4997 ++ if (IS_IMMUTABLE(inode) && !IS_COW(inode))
4998 + return -EACCES;
4999 + }
5000 +
5001 ++ retval = dx_permission(inode, mask);
5002 ++ if (retval)
5003 ++ return retval;
5004 ++
5005 + if (inode->i_op->permission)
5006 + retval = inode->i_op->permission(inode, mask);
5007 + else
5008 +@@ -432,6 +515,8 @@ static int exec_permission_lite(struct i
5009 + {
5010 + umode_t mode = inode->i_mode;
5011 +
5012 ++ if (dx_barrier(inode))
5013 ++ return -EACCES;
5014 + if (inode->i_op->permission)
5015 + return -EAGAIN;
5016 +
5017 +@@ -749,7 +834,8 @@ static __always_inline void follow_dotdo
5018 + if (nd->path.dentry == fs->root.dentry &&
5019 + nd->path.mnt == fs->root.mnt) {
5020 + read_unlock(&fs->lock);
5021 +- break;
5022 ++ /* for sane '/' avoid follow_mount() */
5023 ++ return;
5024 + }
5025 + read_unlock(&fs->lock);
5026 + spin_lock(&dcache_lock);
5027 +@@ -786,16 +872,30 @@ static int do_lookup(struct nameidata *n
5028 + {
5029 + struct vfsmount *mnt = nd->path.mnt;
5030 + struct dentry *dentry = __d_lookup(nd->path.dentry, name);
5031 ++ struct inode *inode;
5032 +
5033 + if (!dentry)
5034 + goto need_lookup;
5035 + if (dentry->d_op && dentry->d_op->d_revalidate)
5036 + goto need_revalidate;
5037 ++ inode = dentry->d_inode;
5038 ++ if (!inode)
5039 ++ goto done;
5040 ++
5041 ++ if (__dx_permission(inode, MAY_ACCESS))
5042 ++ goto hidden;
5043 ++
5044 + done:
5045 + path->mnt = mnt;
5046 + path->dentry = dentry;
5047 + __follow_mount(path);
5048 + return 0;
5049 ++hidden:
5050 ++ vxwprintk_task(1, "did lookup hidden %s:%p[#%d,%lu] »%s/%.*s«.",
5051 ++ inode->i_sb->s_id, inode, inode->i_tag, inode->i_ino,
5052 ++ vxd_path(&nd->path), name->len, name->name);
5053 ++ dput(dentry);
5054 ++ return -ENOENT;
5055 +
5056 + need_lookup:
5057 + dentry = real_lookup(nd->path.dentry, name, nd);
5058 +@@ -1364,7 +1464,7 @@ static int may_delete(struct inode *dir,
5059 + if (IS_APPEND(dir))
5060 + return -EPERM;
5061 + if (check_sticky(dir, victim->d_inode)||IS_APPEND(victim->d_inode)||
5062 +- IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5063 ++ IS_IXORUNLINK(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5064 + return -EPERM;
5065 + if (isdir) {
5066 + if (!S_ISDIR(victim->d_inode->i_mode))
5067 +@@ -1506,6 +1606,14 @@ int may_open(struct path *path, int acc_
5068 + flag &= ~O_TRUNC;
5069 + }
5070 +
5071 ++#ifdef CONFIG_VSERVER_COWBL
5072 ++ if (IS_COW(inode) && (flag & FMODE_WRITE)) {
5073 ++ if (IS_COW_LINK(inode))
5074 ++ return -EMLINK;
5075 ++ inode->i_flags &= ~(S_IXUNLINK|S_IMMUTABLE);
5076 ++ mark_inode_dirty(inode);
5077 ++ }
5078 ++#endif
5079 + error = inode_permission(inode, acc_mode);
5080 + if (error)
5081 + return error;
5082 +@@ -1639,6 +1747,11 @@ struct file *do_filp_open(int dfd, const
5083 + int will_write;
5084 + int flag = open_to_namei_flags(open_flag);
5085 +
5086 ++#ifdef CONFIG_VSERVER_COWBL
5087 ++ int rflag = flag;
5088 ++ int rmode = mode;
5089 ++restart:
5090 ++#endif
5091 + acc_mode = MAY_OPEN | ACC_MODE(flag);
5092 +
5093 + /* O_TRUNC implies we need access checks for write permissions */
5094 +@@ -1771,6 +1884,25 @@ ok:
5095 + goto exit;
5096 + }
5097 + error = may_open(&nd.path, acc_mode, flag);
5098 ++#ifdef CONFIG_VSERVER_COWBL
5099 ++ if (error == -EMLINK) {
5100 ++ struct dentry *dentry;
5101 ++ dentry = cow_break_link(pathname);
5102 ++ if (IS_ERR(dentry)) {
5103 ++ error = PTR_ERR(dentry);
5104 ++ goto exit_cow;
5105 ++ }
5106 ++ dput(dentry);
5107 ++ if (will_write)
5108 ++ mnt_drop_write(nd.path.mnt);
5109 ++ release_open_intent(&nd);
5110 ++ path_put(&nd.path);
5111 ++ flag = rflag;
5112 ++ mode = rmode;
5113 ++ goto restart;
5114 ++ }
5115 ++exit_cow:
5116 ++#endif
5117 + if (error) {
5118 + if (will_write)
5119 + mnt_drop_write(nd.path.mnt);
5120 +@@ -1924,9 +2056,17 @@ int vfs_mknod(struct inode *dir, struct
5121 + if (error)
5122 + return error;
5123 +
5124 +- if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
5125 ++ if (!(S_ISCHR(mode) || S_ISBLK(mode)))
5126 ++ goto okay;
5127 ++
5128 ++ if (!capable(CAP_MKNOD))
5129 + return -EPERM;
5130 +
5131 ++ if (S_ISCHR(mode) && !vs_chrdev_perm(dev, DATTR_CREATE))
5132 ++ return -EPERM;
5133 ++ if (S_ISBLK(mode) && !vs_blkdev_perm(dev, DATTR_CREATE))
5134 ++ return -EPERM;
5135 ++okay:
5136 + if (!dir->i_op->mknod)
5137 + return -EPERM;
5138 +
5139 +@@ -2393,7 +2533,7 @@ int vfs_link(struct dentry *old_dentry,
5140 + /*
5141 + * A link to an append-only or immutable file cannot be created.
5142 + */
5143 +- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
5144 ++ if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
5145 + return -EPERM;
5146 + if (!dir->i_op->link)
5147 + return -EPERM;
5148 +@@ -2766,6 +2906,219 @@ int vfs_follow_link(struct nameidata *nd
5149 + return __vfs_follow_link(nd, link);
5150 + }
5151 +
5152 ++
5153 ++#ifdef CONFIG_VSERVER_COWBL
5154 ++
5155 ++#include <linux/file.h>
5156 ++
5157 ++static inline
5158 ++long do_cow_splice(struct file *in, struct file *out, size_t len)
5159 ++{
5160 ++ loff_t ppos = 0;
5161 ++
5162 ++ return do_splice_direct(in, &ppos, out, len, 0);
5163 ++}
5164 ++
5165 ++struct dentry *cow_break_link(const char *pathname)
5166 ++{
5167 ++ int ret, mode, pathlen, redo = 0;
5168 ++ struct nameidata old_nd, dir_nd;
5169 ++ struct path old_path, new_path;
5170 ++ struct dentry *dir, *res = NULL;
5171 ++ struct file *old_file;
5172 ++ struct file *new_file;
5173 ++ char *to, *path, pad='\251';
5174 ++ loff_t size;
5175 ++
5176 ++ vxdprintk(VXD_CBIT(misc, 1), "cow_break_link(»%s«)", pathname);
5177 ++ path = kmalloc(PATH_MAX, GFP_KERNEL);
5178 ++ ret = -ENOMEM;
5179 ++ if (!path)
5180 ++ goto out;
5181 ++
5182 ++ /* old_nd will have refs to dentry and mnt */
5183 ++ ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5184 ++ vxdprintk(VXD_CBIT(misc, 2), "path_lookup(old): %d", ret);
5185 ++ if (ret < 0)
5186 ++ goto out_free_path;
5187 ++
5188 ++ old_path = old_nd.path;
5189 ++ mode = old_path.dentry->d_inode->i_mode;
5190 ++
5191 ++ to = d_path(&old_path, path, PATH_MAX-2);
5192 ++ pathlen = strlen(to);
5193 ++ vxdprintk(VXD_CBIT(misc, 2), "old path »%s« [»%.*s«:%d]", to,
5194 ++ old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5195 ++ old_path.dentry->d_name.len);
5196 ++
5197 ++ to[pathlen + 1] = 0;
5198 ++retry:
5199 ++ to[pathlen] = pad--;
5200 ++ ret = -EMLINK;
5201 ++ if (pad <= '\240')
5202 ++ goto out_rel_old;
5203 ++
5204 ++ vxdprintk(VXD_CBIT(misc, 1), "temp copy »%s«", to);
5205 ++ /* dir_nd will have refs to dentry and mnt */
5206 ++ ret = path_lookup(to,
5207 ++ LOOKUP_PARENT | LOOKUP_OPEN | LOOKUP_CREATE, &dir_nd);
5208 ++ vxdprintk(VXD_CBIT(misc, 2),
5209 ++ "path_lookup(new): %d", ret);
5210 ++ if (ret < 0)
5211 ++ goto retry;
5212 ++
5213 ++ /* this puppy downs the inode mutex */
5214 ++ new_path.dentry = lookup_create(&dir_nd, 0);
5215 ++ if (!new_path.dentry || IS_ERR(new_path.dentry)) {
5216 ++ vxdprintk(VXD_CBIT(misc, 2),
5217 ++ "lookup_create(new): %p", new_path.dentry);
5218 ++ mutex_unlock(&dir_nd.path.dentry->d_inode->i_mutex);
5219 ++ path_put(&dir_nd.path);
5220 ++ goto retry;
5221 ++ }
5222 ++ vxdprintk(VXD_CBIT(misc, 2),
5223 ++ "lookup_create(new): %p [»%.*s«:%d]", new_path.dentry,
5224 ++ new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5225 ++ new_path.dentry->d_name.len);
5226 ++ dir = dir_nd.path.dentry;
5227 ++
5228 ++ ret = vfs_create(dir_nd.path.dentry->d_inode, new_path.dentry, mode, &dir_nd);
5229 ++ vxdprintk(VXD_CBIT(misc, 2),
5230 ++ "vfs_create(new): %d", ret);
5231 ++ if (ret == -EEXIST) {
5232 ++ mutex_unlock(&dir->d_inode->i_mutex);
5233 ++ dput(new_path.dentry);
5234 ++ path_put(&dir_nd.path);
5235 ++ goto retry;
5236 ++ }
5237 ++ else if (ret < 0)
5238 ++ goto out_unlock_new;
5239 ++
5240 ++ /* drop out early, ret passes ENOENT */
5241 ++ ret = -ENOENT;
5242 ++ if ((redo = d_unhashed(old_path.dentry)))
5243 ++ goto out_unlock_new;
5244 ++
5245 ++ new_path.mnt = dir_nd.path.mnt;
5246 ++ dget(old_path.dentry);
5247 ++ mntget(old_path.mnt);
5248 ++ /* this one cleans up the dentry/mnt in case of failure */
5249 ++ old_file = dentry_open(old_path.dentry, old_path.mnt,
5250 ++ O_RDONLY, current_cred());
5251 ++ vxdprintk(VXD_CBIT(misc, 2),
5252 ++ "dentry_open(old): %p", old_file);
5253 ++ if (!old_file || IS_ERR(old_file)) {
5254 ++ res = IS_ERR(old_file) ? (void *) old_file : res;
5255 ++ goto out_unlock_new;
5256 ++ }
5257 ++
5258 ++ dget(new_path.dentry);
5259 ++ mntget(new_path.mnt);
5260 ++ /* this one cleans up the dentry/mnt in case of failure */
5261 ++ new_file = dentry_open(new_path.dentry, new_path.mnt,
5262 ++ O_WRONLY, current_cred());
5263 ++ vxdprintk(VXD_CBIT(misc, 2),
5264 ++ "dentry_open(new): %p", new_file);
5265 ++
5266 ++ ret = IS_ERR(new_file) ? PTR_ERR(new_file) : -ENOENT;
5267 ++ if (!new_file || IS_ERR(new_file))
5268 ++ goto out_fput_old;
5269 ++
5270 ++ size = i_size_read(old_file->f_dentry->d_inode);
5271 ++ ret = do_cow_splice(old_file, new_file, size);
5272 ++ vxdprintk(VXD_CBIT(misc, 2), "do_splice_direct: %d", ret);
5273 ++ if (ret < 0) {
5274 ++ goto out_fput_both;
5275 ++ } else if (ret < size) {
5276 ++ ret = -ENOSPC;
5277 ++ goto out_fput_both;
5278 ++ } else {
5279 ++ struct inode *old_inode = old_path.dentry->d_inode;
5280 ++ struct inode *new_inode = new_path.dentry->d_inode;
5281 ++ struct iattr attr = {
5282 ++ .ia_uid = old_inode->i_uid,
5283 ++ .ia_gid = old_inode->i_gid,
5284 ++ .ia_valid = ATTR_UID | ATTR_GID
5285 ++ };
5286 ++
5287 ++ ret = inode_setattr(new_inode, &attr);
5288 ++ if (ret)
5289 ++ goto out_fput_both;
5290 ++ }
5291 ++
5292 ++ mutex_lock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5293 ++
5294 ++ /* drop out late */
5295 ++ ret = -ENOENT;
5296 ++ if ((redo = d_unhashed(old_path.dentry)))
5297 ++ goto out_unlock;
5298 ++
5299 ++ vxdprintk(VXD_CBIT(misc, 2),
5300 ++ "vfs_rename: [»%*s«:%d] -> [»%*s«:%d]",
5301 ++ new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5302 ++ new_path.dentry->d_name.len,
5303 ++ old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5304 ++ old_path.dentry->d_name.len);
5305 ++ ret = vfs_rename(dir_nd.path.dentry->d_inode, new_path.dentry,
5306 ++ old_nd.path.dentry->d_parent->d_inode, old_path.dentry);
5307 ++ vxdprintk(VXD_CBIT(misc, 2), "vfs_rename: %d", ret);
5308 ++ res = new_path.dentry;
5309 ++
5310 ++out_unlock:
5311 ++ mutex_unlock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5312 ++
5313 ++out_fput_both:
5314 ++ vxdprintk(VXD_CBIT(misc, 3),
5315 ++ "fput(new_file=%p[#%ld])", new_file,
5316 ++ atomic_read(&new_file->f_count));
5317 ++ fput(new_file);
5318 ++
5319 ++out_fput_old:
5320 ++ vxdprintk(VXD_CBIT(misc, 3),
5321 ++ "fput(old_file=%p[#%ld])", old_file,
5322 ++ atomic_read(&old_file->f_count));
5323 ++ fput(old_file);
5324 ++
5325 ++out_unlock_new:
5326 ++ mutex_unlock(&dir->d_inode->i_mutex);
5327 ++ if (!ret)
5328 ++ goto out_redo;
5329 ++
5330 ++ /* error path cleanup */
5331 ++ vfs_unlink(dir->d_inode, new_path.dentry);
5332 ++ dput(new_path.dentry);
5333 ++
5334 ++out_redo:
5335 ++ if (!redo)
5336 ++ goto out_rel_both;
5337 ++ /* lookup dentry once again */
5338 ++ path_put(&old_nd.path);
5339 ++ ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5340 ++ if (ret)
5341 ++ goto out_rel_both;
5342 ++
5343 ++ new_path.dentry = old_nd.path.dentry;
5344 ++ vxdprintk(VXD_CBIT(misc, 2),
5345 ++ "path_lookup(redo): %p [»%.*s«:%d]", new_path.dentry,
5346 ++ new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5347 ++ new_path.dentry->d_name.len);
5348 ++ dget(new_path.dentry);
5349 ++ res = new_path.dentry;
5350 ++
5351 ++out_rel_both:
5352 ++ path_put(&dir_nd.path);
5353 ++out_rel_old:
5354 ++ path_put(&old_nd.path);
5355 ++out_free_path:
5356 ++ kfree(path);
5357 ++out:
5358 ++ if (ret)
5359 ++ res = ERR_PTR(ret);
5360 ++ return res;
5361 ++}
5362 ++
5363 ++#endif
5364 ++
5365 + /* get the link contents into pagecache */
5366 + static char *page_getlink(struct dentry * dentry, struct page **ppage)
5367 + {
5368 +diff -NurpP --minimal linux-2.6.29.4/fs/namespace.c linux-2.6.29.4-vs2.3.0.36.14/fs/namespace.c
5369 +--- linux-2.6.29.4/fs/namespace.c 2009-05-23 23:16:52.000000000 +0200
5370 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/namespace.c 2009-05-10 23:42:01.000000000 +0200
5371 +@@ -27,6 +27,11 @@
5372 + #include <linux/ramfs.h>
5373 + #include <linux/log2.h>
5374 + #include <linux/idr.h>
5375 ++#include <linux/vs_base.h>
5376 ++#include <linux/vs_context.h>
5377 ++#include <linux/vs_tag.h>
5378 ++#include <linux/vserver/space.h>
5379 ++#include <linux/vserver/global.h>
5380 + #include <asm/uaccess.h>
5381 + #include <asm/unistd.h>
5382 + #include "pnode.h"
5383 +@@ -573,6 +578,7 @@ static struct vfsmount *clone_mnt(struct
5384 + mnt->mnt_root = dget(root);
5385 + mnt->mnt_mountpoint = mnt->mnt_root;
5386 + mnt->mnt_parent = mnt;
5387 ++ mnt->mnt_tag = old->mnt_tag;
5388 +
5389 + if (flag & CL_SLAVE) {
5390 + list_add(&mnt->mnt_slave, &old->mnt_slave_list);
5391 +@@ -687,6 +693,31 @@ static inline void mangle(struct seq_fil
5392 + seq_escape(m, s, " \t\n\\");
5393 + }
5394 +
5395 ++static int mnt_is_reachable(struct vfsmount *mnt)
5396 ++{
5397 ++ struct path root;
5398 ++ struct dentry *point;
5399 ++ int ret;
5400 ++
5401 ++ if (mnt == mnt->mnt_ns->root)
5402 ++ return 1;
5403 ++
5404 ++ spin_lock(&vfsmount_lock);
5405 ++ root = current->fs->root;
5406 ++ point = root.dentry;
5407 ++
5408 ++ while ((mnt != mnt->mnt_parent) && (mnt != root.mnt)) {
5409 ++ point = mnt->mnt_mountpoint;
5410 ++ mnt = mnt->mnt_parent;
5411 ++ }
5412 ++
5413 ++ ret = (mnt == root.mnt) && is_subdir(point, root.dentry);
5414 ++
5415 ++ spin_unlock(&vfsmount_lock);
5416 ++
5417 ++ return ret;
5418 ++}
5419 ++
5420 + /*
5421 + * Simple .show_options callback for filesystems which don't want to
5422 + * implement more complex mount option showing.
5423 +@@ -759,6 +790,8 @@ static int show_sb_opts(struct seq_file
5424 + { MS_SYNCHRONOUS, ",sync" },
5425 + { MS_DIRSYNC, ",dirsync" },
5426 + { MS_MANDLOCK, ",mand" },
5427 ++ { MS_TAGGED, ",tag" },
5428 ++ { MS_NOTAGCHECK, ",notagcheck" },
5429 + { 0, NULL }
5430 + };
5431 + const struct proc_fs_info *fs_infop;
5432 +@@ -805,10 +838,20 @@ static int show_vfsmnt(struct seq_file *
5433 + int err = 0;
5434 + struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5435 +
5436 +- mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5437 +- seq_putc(m, ' ');
5438 +- seq_path(m, &mnt_path, " \t\n\\");
5439 +- seq_putc(m, ' ');
5440 ++ if (vx_flags(VXF_HIDE_MOUNT, 0))
5441 ++ return SEQ_SKIP;
5442 ++ if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5443 ++ return SEQ_SKIP;
5444 ++
5445 ++ if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5446 ++ mnt == current->fs->root.mnt) {
5447 ++ seq_puts(m, "/dev/root / ");
5448 ++ } else {
5449 ++ mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5450 ++ seq_putc(m, ' ');
5451 ++ seq_path(m, &mnt_path, " \t\n\\");
5452 ++ seq_putc(m, ' ');
5453 ++ }
5454 + show_type(m, mnt->mnt_sb);
5455 + seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
5456 + err = show_sb_opts(m, mnt->mnt_sb);
5457 +@@ -838,6 +881,11 @@ static int show_mountinfo(struct seq_fil
5458 + struct path root = p->root;
5459 + int err = 0;
5460 +
5461 ++ if (vx_flags(VXF_HIDE_MOUNT, 0))
5462 ++ return SEQ_SKIP;
5463 ++ if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5464 ++ return SEQ_SKIP;
5465 ++
5466 + seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
5467 + MAJOR(sb->s_dev), MINOR(sb->s_dev));
5468 + seq_dentry(m, mnt->mnt_root, " \t\n\\");
5469 +@@ -896,17 +944,27 @@ static int show_vfsstat(struct seq_file
5470 + struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5471 + int err = 0;
5472 +
5473 +- /* device */
5474 +- if (mnt->mnt_devname) {
5475 +- seq_puts(m, "device ");
5476 +- mangle(m, mnt->mnt_devname);
5477 +- } else
5478 +- seq_puts(m, "no device");
5479 ++ if (vx_flags(VXF_HIDE_MOUNT, 0))
5480 ++ return SEQ_SKIP;
5481 ++ if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5482 ++ return SEQ_SKIP;
5483 +
5484 +- /* mount point */
5485 +- seq_puts(m, " mounted on ");
5486 +- seq_path(m, &mnt_path, " \t\n\\");
5487 +- seq_putc(m, ' ');
5488 ++ if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5489 ++ mnt == current->fs->root.mnt) {
5490 ++ seq_puts(m, "device /dev/root mounted on / ");
5491 ++ } else {
5492 ++ /* device */
5493 ++ if (mnt->mnt_devname) {
5494 ++ seq_puts(m, "device ");
5495 ++ mangle(m, mnt->mnt_devname);
5496 ++ } else
5497 ++ seq_puts(m, "no device");
5498 ++
5499 ++ /* mount point */
5500 ++ seq_puts(m, " mounted on ");
5501 ++ seq_path(m, &mnt_path, " \t\n\\");
5502 ++ seq_putc(m, ' ');
5503 ++ }
5504 +
5505 + /* file system type */
5506 + seq_puts(m, "with fstype ");
5507 +@@ -1145,7 +1203,7 @@ SYSCALL_DEFINE2(umount, char __user *, n
5508 + goto dput_and_out;
5509 +
5510 + retval = -EPERM;
5511 +- if (!capable(CAP_SYS_ADMIN))
5512 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5513 + goto dput_and_out;
5514 +
5515 + retval = do_umount(path.mnt, flags);
5516 +@@ -1171,7 +1229,7 @@ SYSCALL_DEFINE1(oldumount, char __user *
5517 +
5518 + static int mount_is_safe(struct path *path)
5519 + {
5520 +- if (capable(CAP_SYS_ADMIN))
5521 ++ if (vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5522 + return 0;
5523 + return -EPERM;
5524 + #ifdef notyet
5525 +@@ -1462,11 +1520,13 @@ static int do_change_type(struct path *p
5526 + * do loopback mount.
5527 + */
5528 + static int do_loopback(struct path *path, char *old_name,
5529 +- int recurse)
5530 ++ tag_t tag, unsigned long flags, int mnt_flags)
5531 + {
5532 + struct path old_path;
5533 + struct vfsmount *mnt = NULL;
5534 + int err = mount_is_safe(path);
5535 ++ int recurse = flags & MS_REC;
5536 ++
5537 + if (err)
5538 + return err;
5539 + if (!old_name || !*old_name)
5540 +@@ -1500,6 +1560,7 @@ static int do_loopback(struct path *path
5541 + spin_unlock(&vfsmount_lock);
5542 + release_mounts(&umount_list);
5543 + }
5544 ++ mnt->mnt_flags = mnt_flags;
5545 +
5546 + out:
5547 + up_write(&namespace_sem);
5548 +@@ -1530,12 +1591,12 @@ static int change_mount_flags(struct vfs
5549 + * on it - tough luck.
5550 + */
5551 + static int do_remount(struct path *path, int flags, int mnt_flags,
5552 +- void *data)
5553 ++ void *data, xid_t xid)
5554 + {
5555 + int err;
5556 + struct super_block *sb = path->mnt->mnt_sb;
5557 +
5558 +- if (!capable(CAP_SYS_ADMIN))
5559 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_REMOUNT))
5560 + return -EPERM;
5561 +
5562 + if (!check_mnt(path->mnt))
5563 +@@ -1577,7 +1638,7 @@ static int do_move_mount(struct path *pa
5564 + struct path old_path, parent_path;
5565 + struct vfsmount *p;
5566 + int err = 0;
5567 +- if (!capable(CAP_SYS_ADMIN))
5568 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5569 + return -EPERM;
5570 + if (!old_name || !*old_name)
5571 + return -EINVAL;
5572 +@@ -1659,7 +1720,7 @@ static int do_new_mount(struct path *pat
5573 + return -EINVAL;
5574 +
5575 + /* we need capabilities... */
5576 +- if (!capable(CAP_SYS_ADMIN))
5577 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5578 + return -EPERM;
5579 +
5580 + mnt = do_kern_mount(type, flags, name, data);
5581 +@@ -1904,6 +1965,7 @@ long do_mount(char *dev_name, char *dir_
5582 + struct path path;
5583 + int retval = 0;
5584 + int mnt_flags = 0;
5585 ++ tag_t tag = 0;
5586 +
5587 + /* Discard magic */
5588 + if ((flags & MS_MGC_MSK) == MS_MGC_VAL)
5589 +@@ -1919,6 +1981,12 @@ long do_mount(char *dev_name, char *dir_
5590 + if (data_page)
5591 + ((char *)data_page)[PAGE_SIZE - 1] = 0;
5592 +
5593 ++ if (dx_parse_tag(data_page, &tag, 1, &mnt_flags, &flags)) {
5594 ++ /* FIXME: bind and re-mounts get the tag flag? */
5595 ++ if (flags & (MS_BIND|MS_REMOUNT))
5596 ++ flags |= MS_TAGID;
5597 ++ }
5598 ++
5599 + /* Separate the per-mountpoint flags */
5600 + if (flags & MS_NOSUID)
5601 + mnt_flags |= MNT_NOSUID;
5602 +@@ -1935,6 +2003,8 @@ long do_mount(char *dev_name, char *dir_
5603 + if (flags & MS_RDONLY)
5604 + mnt_flags |= MNT_READONLY;
5605 +
5606 ++ if (!capable(CAP_SYS_ADMIN))
5607 ++ mnt_flags |= MNT_NODEV;
5608 + flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE |
5609 + MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT);
5610 +
5611 +@@ -1950,9 +2020,9 @@ long do_mount(char *dev_name, char *dir_
5612 +
5613 + if (flags & MS_REMOUNT)
5614 + retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
5615 +- data_page);
5616 ++ data_page, tag);
5617 + else if (flags & MS_BIND)
5618 +- retval = do_loopback(&path, dev_name, flags & MS_REC);
5619 ++ retval = do_loopback(&path, dev_name, tag, flags, mnt_flags);
5620 + else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
5621 + retval = do_change_type(&path, flags);
5622 + else if (flags & MS_MOVE)
5623 +@@ -2021,6 +2091,7 @@ static struct mnt_namespace *dup_mnt_ns(
5624 + q = next_mnt(q, new_ns->root);
5625 + }
5626 + up_write(&namespace_sem);
5627 ++ atomic_inc(&vs_global_mnt_ns);
5628 +
5629 + if (rootmnt)
5630 + mntput(rootmnt);
5631 +@@ -2147,9 +2218,10 @@ SYSCALL_DEFINE2(pivot_root, const char _
5632 + down_write(&namespace_sem);
5633 + mutex_lock(&old.dentry->d_inode->i_mutex);
5634 + error = -EINVAL;
5635 +- if (IS_MNT_SHARED(old.mnt) ||
5636 ++ if ((IS_MNT_SHARED(old.mnt) ||
5637 + IS_MNT_SHARED(new.mnt->mnt_parent) ||
5638 +- IS_MNT_SHARED(root.mnt->mnt_parent))
5639 ++ IS_MNT_SHARED(root.mnt->mnt_parent)) &&
5640 ++ !vx_flags(VXF_STATE_SETUP, 0))
5641 + goto out2;
5642 + if (!check_mnt(root.mnt))
5643 + goto out2;
5644 +@@ -2288,5 +2360,6 @@ void __put_mnt_ns(struct mnt_namespace *
5645 + spin_unlock(&vfsmount_lock);
5646 + up_write(&namespace_sem);
5647 + release_mounts(&umount_list);
5648 ++ atomic_dec(&vs_global_mnt_ns);
5649 + kfree(ns);
5650 + }
5651 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/client.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/client.c
5652 +--- linux-2.6.29.4/fs/nfs/client.c 2009-03-24 14:22:26.000000000 +0100
5653 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/client.c 2009-03-24 14:48:35.000000000 +0100
5654 +@@ -696,6 +696,9 @@ static int nfs_init_server_rpcclient(str
5655 + if (server->flags & NFS_MOUNT_SOFT)
5656 + server->client->cl_softrtry = 1;
5657 +
5658 ++ server->client->cl_tag = 0;
5659 ++ if (server->flags & NFS_MOUNT_TAGGED)
5660 ++ server->client->cl_tag = 1;
5661 + return 0;
5662 + }
5663 +
5664 +@@ -862,6 +865,10 @@ static void nfs_server_set_fsinfo(struct
5665 + server->acdirmin = server->acdirmax = 0;
5666 + }
5667 +
5668 ++ /* FIXME: needs fsinfo
5669 ++ if (server->flags & NFS_MOUNT_TAGGED)
5670 ++ sb->s_flags |= MS_TAGGED; */
5671 ++
5672 + server->maxfilesize = fsinfo->maxfilesize;
5673 +
5674 + /* We're airborne Set socket buffersize */
5675 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/dir.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/dir.c
5676 +--- linux-2.6.29.4/fs/nfs/dir.c 2009-05-23 23:16:52.000000000 +0200
5677 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/dir.c 2009-05-23 23:19:11.000000000 +0200
5678 +@@ -34,6 +34,7 @@
5679 + #include <linux/namei.h>
5680 + #include <linux/mount.h>
5681 + #include <linux/sched.h>
5682 ++#include <linux/vs_tag.h>
5683 +
5684 + #include "nfs4_fs.h"
5685 + #include "delegation.h"
5686 +@@ -950,6 +951,7 @@ static struct dentry *nfs_lookup(struct
5687 + if (IS_ERR(res))
5688 + goto out_unblock_sillyrename;
5689 +
5690 ++ dx_propagate_tag(nd, inode);
5691 + no_entry:
5692 + res = d_materialise_unique(dentry, inode);
5693 + if (res != NULL) {
5694 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/inode.c
5695 +--- linux-2.6.29.4/fs/nfs/inode.c 2009-03-24 14:22:26.000000000 +0100
5696 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/inode.c 2009-03-24 14:48:35.000000000 +0100
5697 +@@ -37,6 +37,7 @@
5698 + #include <linux/vfs.h>
5699 + #include <linux/inet.h>
5700 + #include <linux/nfs_xdr.h>
5701 ++#include <linux/vs_tag.h>
5702 +
5703 + #include <asm/system.h>
5704 + #include <asm/uaccess.h>
5705 +@@ -313,8 +314,10 @@ nfs_fhget(struct super_block *sb, struct
5706 + nfsi->change_attr = fattr->change_attr;
5707 + inode->i_size = nfs_size_to_loff_t(fattr->size);
5708 + inode->i_nlink = fattr->nlink;
5709 +- inode->i_uid = fattr->uid;
5710 +- inode->i_gid = fattr->gid;
5711 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5712 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5713 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5714 ++ /* maybe fattr->xid someday */
5715 + if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
5716 + /*
5717 + * report the blocks in 512byte units
5718 +@@ -462,6 +465,8 @@ void nfs_setattr_update_inode(struct ino
5719 + inode->i_uid = attr->ia_uid;
5720 + if ((attr->ia_valid & ATTR_GID) != 0)
5721 + inode->i_gid = attr->ia_gid;
5722 ++ if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
5723 ++ inode->i_tag = attr->ia_tag;
5724 + NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5725 + spin_unlock(&inode->i_lock);
5726 + }
5727 +@@ -850,6 +855,9 @@ static int nfs_check_inode_attributes(st
5728 + struct nfs_inode *nfsi = NFS_I(inode);
5729 + loff_t cur_size, new_isize;
5730 + unsigned long invalid = 0;
5731 ++ uid_t uid;
5732 ++ gid_t gid;
5733 ++ tag_t tag;
5734 +
5735 +
5736 + /* Has the inode gone and changed behind our back? */
5737 +@@ -871,10 +879,15 @@ static int nfs_check_inode_attributes(st
5738 + if (cur_size != new_isize && nfsi->npages == 0)
5739 + invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
5740 +
5741 ++ uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5742 ++ gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5743 ++ tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5744 ++
5745 + /* Have any file permissions changed? */
5746 + if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)
5747 +- || inode->i_uid != fattr->uid
5748 +- || inode->i_gid != fattr->gid)
5749 ++ || inode->i_uid != uid
5750 ++ || inode->i_gid != gid
5751 ++ || inode->i_tag != tag)
5752 + invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5753 +
5754 + /* Has the link count changed? */
5755 +@@ -1073,6 +1086,9 @@ static int nfs_update_inode(struct inode
5756 + loff_t cur_isize, new_isize;
5757 + unsigned long invalid = 0;
5758 + unsigned long now = jiffies;
5759 ++ uid_t uid;
5760 ++ gid_t gid;
5761 ++ tag_t tag;
5762 +
5763 + dfprintk(VFS, "NFS: %s(%s/%ld ct=%d info=0x%x)\n",
5764 + __func__, inode->i_sb->s_id, inode->i_ino,
5765 +@@ -1146,9 +1162,14 @@ static int nfs_update_inode(struct inode
5766 + memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
5767 + nfsi->change_attr = fattr->change_attr;
5768 +
5769 ++ uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5770 ++ gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5771 ++ tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5772 ++
5773 + if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO) ||
5774 +- inode->i_uid != fattr->uid ||
5775 +- inode->i_gid != fattr->gid)
5776 ++ inode->i_uid != uid ||
5777 ++ inode->i_gid != gid ||
5778 ++ inode->i_tag != tag)
5779 + invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5780 +
5781 + if (inode->i_nlink != fattr->nlink)
5782 +@@ -1156,8 +1177,9 @@ static int nfs_update_inode(struct inode
5783 +
5784 + inode->i_mode = fattr->mode;
5785 + inode->i_nlink = fattr->nlink;
5786 +- inode->i_uid = fattr->uid;
5787 +- inode->i_gid = fattr->gid;
5788 ++ inode->i_uid = uid;
5789 ++ inode->i_gid = gid;
5790 ++ inode->i_tag = tag;
5791 +
5792 + if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
5793 + /*
5794 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/nfs3xdr.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/nfs3xdr.c
5795 +--- linux-2.6.29.4/fs/nfs/nfs3xdr.c 2009-05-23 23:16:53.000000000 +0200
5796 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/nfs3xdr.c 2009-04-30 12:14:53.000000000 +0200
5797 +@@ -22,6 +22,7 @@
5798 + #include <linux/nfs3.h>
5799 + #include <linux/nfs_fs.h>
5800 + #include <linux/nfsacl.h>
5801 ++#include <linux/vs_tag.h>
5802 + #include "internal.h"
5803 +
5804 + #define NFSDBG_FACILITY NFSDBG_XDR
5805 +@@ -182,7 +183,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_f
5806 + }
5807 +
5808 + static inline __be32 *
5809 +-xdr_encode_sattr(__be32 *p, struct iattr *attr)
5810 ++xdr_encode_sattr(__be32 *p, struct iattr *attr, int tag)
5811 + {
5812 + if (attr->ia_valid & ATTR_MODE) {
5813 + *p++ = xdr_one;
5814 +@@ -190,15 +191,17 @@ xdr_encode_sattr(__be32 *p, struct iattr
5815 + } else {
5816 + *p++ = xdr_zero;
5817 + }
5818 +- if (attr->ia_valid & ATTR_UID) {
5819 ++ if (attr->ia_valid & ATTR_UID ||
5820 ++ (tag && (attr->ia_valid & ATTR_TAG))) {
5821 + *p++ = xdr_one;
5822 +- *p++ = htonl(attr->ia_uid);
5823 ++ *p++ = htonl(TAGINO_UID(tag, attr->ia_uid, attr->ia_tag));
5824 + } else {
5825 + *p++ = xdr_zero;
5826 + }
5827 +- if (attr->ia_valid & ATTR_GID) {
5828 ++ if (attr->ia_valid & ATTR_GID ||
5829 ++ (tag && (attr->ia_valid & ATTR_TAG))) {
5830 + *p++ = xdr_one;
5831 +- *p++ = htonl(attr->ia_gid);
5832 ++ *p++ = htonl(TAGINO_GID(tag, attr->ia_gid, attr->ia_tag));
5833 + } else {
5834 + *p++ = xdr_zero;
5835 + }
5836 +@@ -283,7 +286,8 @@ static int
5837 + nfs3_xdr_sattrargs(struct rpc_rqst *req, __be32 *p, struct nfs3_sattrargs *args)
5838 + {
5839 + p = xdr_encode_fhandle(p, args->fh);
5840 +- p = xdr_encode_sattr(p, args->sattr);
5841 ++ p = xdr_encode_sattr(p, args->sattr,
5842 ++ req->rq_task->tk_client->cl_tag);
5843 + *p++ = htonl(args->guard);
5844 + if (args->guard)
5845 + p = xdr_encode_time3(p, &args->guardtime);
5846 +@@ -388,7 +392,8 @@ nfs3_xdr_createargs(struct rpc_rqst *req
5847 + *p++ = args->verifier[0];
5848 + *p++ = args->verifier[1];
5849 + } else
5850 +- p = xdr_encode_sattr(p, args->sattr);
5851 ++ p = xdr_encode_sattr(p, args->sattr,
5852 ++ req->rq_task->tk_client->cl_tag);
5853 +
5854 + req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5855 + return 0;
5856 +@@ -402,7 +407,8 @@ nfs3_xdr_mkdirargs(struct rpc_rqst *req,
5857 + {
5858 + p = xdr_encode_fhandle(p, args->fh);
5859 + p = xdr_encode_array(p, args->name, args->len);
5860 +- p = xdr_encode_sattr(p, args->sattr);
5861 ++ p = xdr_encode_sattr(p, args->sattr,
5862 ++ req->rq_task->tk_client->cl_tag);
5863 + req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5864 + return 0;
5865 + }
5866 +@@ -415,7 +421,8 @@ nfs3_xdr_symlinkargs(struct rpc_rqst *re
5867 + {
5868 + p = xdr_encode_fhandle(p, args->fromfh);
5869 + p = xdr_encode_array(p, args->fromname, args->fromlen);
5870 +- p = xdr_encode_sattr(p, args->sattr);
5871 ++ p = xdr_encode_sattr(p, args->sattr,
5872 ++ req->rq_task->tk_client->cl_tag);
5873 + *p++ = htonl(args->pathlen);
5874 + req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5875 +
5876 +@@ -433,7 +440,8 @@ nfs3_xdr_mknodargs(struct rpc_rqst *req,
5877 + p = xdr_encode_fhandle(p, args->fh);
5878 + p = xdr_encode_array(p, args->name, args->len);
5879 + *p++ = htonl(args->type);
5880 +- p = xdr_encode_sattr(p, args->sattr);
5881 ++ p = xdr_encode_sattr(p, args->sattr,
5882 ++ req->rq_task->tk_client->cl_tag);
5883 + if (args->type == NF3CHR || args->type == NF3BLK) {
5884 + *p++ = htonl(MAJOR(args->rdev));
5885 + *p++ = htonl(MINOR(args->rdev));
5886 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/nfsroot.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/nfsroot.c
5887 +--- linux-2.6.29.4/fs/nfs/nfsroot.c 2009-03-24 14:22:26.000000000 +0100
5888 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/nfsroot.c 2009-03-24 14:48:35.000000000 +0100
5889 +@@ -119,12 +119,12 @@ static int mount_port __initdata = 0; /
5890 + enum {
5891 + /* Options that take integer arguments */
5892 + Opt_port, Opt_rsize, Opt_wsize, Opt_timeo, Opt_retrans, Opt_acregmin,
5893 +- Opt_acregmax, Opt_acdirmin, Opt_acdirmax,
5894 ++ Opt_acregmax, Opt_acdirmin, Opt_acdirmax, Opt_tagid,
5895 + /* Options that take no arguments */
5896 + Opt_soft, Opt_hard, Opt_intr,
5897 + Opt_nointr, Opt_posix, Opt_noposix, Opt_cto, Opt_nocto, Opt_ac,
5898 + Opt_noac, Opt_lock, Opt_nolock, Opt_v2, Opt_v3, Opt_udp, Opt_tcp,
5899 +- Opt_acl, Opt_noacl,
5900 ++ Opt_acl, Opt_noacl, Opt_tag, Opt_notag,
5901 + /* Error token */
5902 + Opt_err
5903 + };
5904 +@@ -161,6 +161,9 @@ static match_table_t __initconst tokens
5905 + {Opt_tcp, "tcp"},
5906 + {Opt_acl, "acl"},
5907 + {Opt_noacl, "noacl"},
5908 ++ {Opt_tag, "tag"},
5909 ++ {Opt_notag, "notag"},
5910 ++ {Opt_tagid, "tagid=%u"},
5911 + {Opt_err, NULL}
5912 +
5913 + };
5914 +@@ -272,6 +275,20 @@ static int __init root_nfs_parse(char *n
5915 + case Opt_noacl:
5916 + nfs_data.flags |= NFS_MOUNT_NOACL;
5917 + break;
5918 ++#ifndef CONFIG_TAGGING_NONE
5919 ++ case Opt_tag:
5920 ++ nfs_data.flags |= NFS_MOUNT_TAGGED;
5921 ++ break;
5922 ++ case Opt_notag:
5923 ++ nfs_data.flags &= ~NFS_MOUNT_TAGGED;
5924 ++ break;
5925 ++#endif
5926 ++#ifdef CONFIG_PROPAGATE
5927 ++ case Opt_tagid:
5928 ++ /* use args[0] */
5929 ++ nfs_data.flags |= NFS_MOUNT_TAGGED;
5930 ++ break;
5931 ++#endif
5932 + default:
5933 + printk(KERN_WARNING "Root-NFS: unknown "
5934 + "option: %s\n", p);
5935 +diff -NurpP --minimal linux-2.6.29.4/fs/nfs/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/super.c
5936 +--- linux-2.6.29.4/fs/nfs/super.c 2009-03-24 14:22:26.000000000 +0100
5937 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfs/super.c 2009-03-24 16:02:06.000000000 +0100
5938 +@@ -51,6 +51,7 @@
5939 + #include <linux/nfs_xdr.h>
5940 + #include <linux/magic.h>
5941 + #include <linux/parser.h>
5942 ++#include <linux/vs_tag.h>
5943 +
5944 + #include <asm/system.h>
5945 + #include <asm/uaccess.h>
5946 +@@ -517,6 +518,7 @@ static void nfs_show_mount_options(struc
5947 + { NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
5948 + { NFS_MOUNT_UNSHARED, ",nosharecache", "" },
5949 + { NFS_MOUNT_NORESVPORT, ",noresvport", "" },
5950 ++ { NFS_MOUNT_TAGGED, ",tag", "" },
5951 + { 0, NULL, NULL }
5952 + };
5953 + const struct proc_nfs_info *nfs_infop;
5954 +diff -NurpP --minimal linux-2.6.29.4/fs/nfsd/auth.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/auth.c
5955 +--- linux-2.6.29.4/fs/nfsd/auth.c 2009-03-24 14:22:26.000000000 +0100
5956 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/auth.c 2009-03-24 16:09:39.000000000 +0100
5957 +@@ -10,6 +10,7 @@
5958 + #include <linux/sunrpc/svcauth.h>
5959 + #include <linux/nfsd/nfsd.h>
5960 + #include <linux/nfsd/export.h>
5961 ++#include <linux/vs_tag.h>
5962 + #include "auth.h"
5963 +
5964 + int nfsexp_flags(struct svc_rqst *rqstp, struct svc_export *exp)
5965 +@@ -42,6 +43,9 @@ int nfsd_setuser(struct svc_rqst *rqstp,
5966 +
5967 + new->fsuid = rqstp->rq_cred.cr_uid;
5968 + new->fsgid = rqstp->rq_cred.cr_gid;
5969 ++ /* FIXME: this desperately needs a tag :)
5970 ++ new->xid = (xid_t)INOTAG_TAG(DX_TAG_NFSD, cred.cr_uid, cred.cr_gid, 0);
5971 ++ */
5972 +
5973 + rqgi = rqstp->rq_cred.cr_group_info;
5974 +
5975 +@@ -69,7 +73,7 @@ int nfsd_setuser(struct svc_rqst *rqstp,
5976 + }
5977 + } else {
5978 + gi = get_group_info(rqgi);
5979 +- }
5980 ++ }
5981 +
5982 + if (new->fsuid == (uid_t) -1)
5983 + new->fsuid = exp->ex_anon_uid;
5984 +@@ -94,6 +98,6 @@ oom:
5985 + ret = -ENOMEM;
5986 + error:
5987 + abort_creds(new);
5988 +- return ret;
5989 ++ return ret;
5990 + }
5991 +
5992 +diff -NurpP --minimal linux-2.6.29.4/fs/nfsd/nfs3xdr.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfs3xdr.c
5993 +--- linux-2.6.29.4/fs/nfsd/nfs3xdr.c 2008-12-25 00:26:37.000000000 +0100
5994 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfs3xdr.c 2009-02-22 22:54:25.000000000 +0100
5995 +@@ -21,6 +21,7 @@
5996 + #include <linux/sunrpc/svc.h>
5997 + #include <linux/nfsd/nfsd.h>
5998 + #include <linux/nfsd/xdr3.h>
5999 ++#include <linux/vs_tag.h>
6000 + #include "auth.h"
6001 +
6002 + #define NFSDDBG_FACILITY NFSDDBG_XDR
6003 +@@ -108,6 +109,8 @@ static __be32 *
6004 + decode_sattr3(__be32 *p, struct iattr *iap)
6005 + {
6006 + u32 tmp;
6007 ++ uid_t uid = 0;
6008 ++ gid_t gid = 0;
6009 +
6010 + iap->ia_valid = 0;
6011 +
6012 +@@ -117,12 +120,15 @@ decode_sattr3(__be32 *p, struct iattr *i
6013 + }
6014 + if (*p++) {
6015 + iap->ia_valid |= ATTR_UID;
6016 +- iap->ia_uid = ntohl(*p++);
6017 ++ uid = ntohl(*p++);
6018 + }
6019 + if (*p++) {
6020 + iap->ia_valid |= ATTR_GID;
6021 +- iap->ia_gid = ntohl(*p++);
6022 ++ gid = ntohl(*p++);
6023 + }
6024 ++ iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
6025 ++ iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
6026 ++ iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
6027 + if (*p++) {
6028 + u64 newsize;
6029 +
6030 +@@ -178,8 +184,12 @@ encode_fattr3(struct svc_rqst *rqstp, __
6031 + *p++ = htonl(nfs3_ftypes[(stat->mode & S_IFMT) >> 12]);
6032 + *p++ = htonl((u32) stat->mode);
6033 + *p++ = htonl((u32) stat->nlink);
6034 +- *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
6035 +- *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
6036 ++ *p++ = htonl((u32) nfsd_ruid(rqstp,
6037 ++ TAGINO_UID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
6038 ++ stat->uid, stat->tag)));
6039 ++ *p++ = htonl((u32) nfsd_rgid(rqstp,
6040 ++ TAGINO_GID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
6041 ++ stat->gid, stat->tag)));
6042 + if (S_ISLNK(stat->mode) && stat->size > NFS3_MAXPATHLEN) {
6043 + p = xdr_encode_hyper(p, (u64) NFS3_MAXPATHLEN);
6044 + } else {
6045 +diff -NurpP --minimal linux-2.6.29.4/fs/nfsd/nfs4xdr.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfs4xdr.c
6046 +--- linux-2.6.29.4/fs/nfsd/nfs4xdr.c 2009-05-23 23:16:53.000000000 +0200
6047 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfs4xdr.c 2009-05-23 23:19:11.000000000 +0200
6048 +@@ -56,6 +56,7 @@
6049 + #include <linux/nfs4_acl.h>
6050 + #include <linux/sunrpc/gss_api.h>
6051 + #include <linux/sunrpc/svcauth_gss.h>
6052 ++#include <linux/vs_tag.h>
6053 +
6054 + #define NFSDDBG_FACILITY NFSDDBG_XDR
6055 +
6056 +@@ -1714,14 +1715,18 @@ out_acl:
6057 + WRITE32(stat.nlink);
6058 + }
6059 + if (bmval1 & FATTR4_WORD1_OWNER) {
6060 +- status = nfsd4_encode_user(rqstp, stat.uid, &p, &buflen);
6061 ++ status = nfsd4_encode_user(rqstp,
6062 ++ TAGINO_UID(DX_TAG(dentry->d_inode),
6063 ++ stat.uid, stat.tag), &p, &buflen);
6064 + if (status == nfserr_resource)
6065 + goto out_resource;
6066 + if (status)
6067 + goto out;
6068 + }
6069 + if (bmval1 & FATTR4_WORD1_OWNER_GROUP) {
6070 +- status = nfsd4_encode_group(rqstp, stat.gid, &p, &buflen);
6071 ++ status = nfsd4_encode_group(rqstp,
6072 ++ TAGINO_GID(DX_TAG(dentry->d_inode),
6073 ++ stat.gid, stat.tag), &p, &buflen);
6074 + if (status == nfserr_resource)
6075 + goto out_resource;
6076 + if (status)
6077 +diff -NurpP --minimal linux-2.6.29.4/fs/nfsd/nfsxdr.c linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfsxdr.c
6078 +--- linux-2.6.29.4/fs/nfsd/nfsxdr.c 2008-12-25 00:26:37.000000000 +0100
6079 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/nfsd/nfsxdr.c 2009-02-22 22:54:25.000000000 +0100
6080 +@@ -15,6 +15,7 @@
6081 + #include <linux/nfsd/nfsd.h>
6082 + #include <linux/nfsd/xdr.h>
6083 + #include <linux/mm.h>
6084 ++#include <linux/vs_tag.h>
6085 + #include "auth.h"
6086 +
6087 + #define NFSDDBG_FACILITY NFSDDBG_XDR
6088 +@@ -98,6 +99,8 @@ static __be32 *
6089 + decode_sattr(__be32 *p, struct iattr *iap)
6090 + {
6091 + u32 tmp, tmp1;
6092 ++ uid_t uid = 0;
6093 ++ gid_t gid = 0;
6094 +
6095 + iap->ia_valid = 0;
6096 +
6097 +@@ -111,12 +114,15 @@ decode_sattr(__be32 *p, struct iattr *ia
6098 + }
6099 + if ((tmp = ntohl(*p++)) != (u32)-1) {
6100 + iap->ia_valid |= ATTR_UID;
6101 +- iap->ia_uid = tmp;
6102 ++ uid = tmp;
6103 + }
6104 + if ((tmp = ntohl(*p++)) != (u32)-1) {
6105 + iap->ia_valid |= ATTR_GID;
6106 +- iap->ia_gid = tmp;
6107 ++ gid = tmp;
6108 + }
6109 ++ iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
6110 ++ iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
6111 ++ iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
6112 + if ((tmp = ntohl(*p++)) != (u32)-1) {
6113 + iap->ia_valid |= ATTR_SIZE;
6114 + iap->ia_size = tmp;
6115 +@@ -161,8 +167,10 @@ encode_fattr(struct svc_rqst *rqstp, __b
6116 + *p++ = htonl(nfs_ftypes[type >> 12]);
6117 + *p++ = htonl((u32) stat->mode);
6118 + *p++ = htonl((u32) stat->nlink);
6119 +- *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
6120 +- *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
6121 ++ *p++ = htonl((u32) nfsd_ruid(rqstp,
6122 ++ TAGINO_UID(DX_TAG(dentry->d_inode), stat->uid, stat->tag)));
6123 ++ *p++ = htonl((u32) nfsd_rgid(rqstp,
6124 ++ TAGINO_GID(DX_TAG(dentry->d_inode), stat->gid, stat->tag)));
6125 +
6126 + if (S_ISLNK(type) && stat->size > NFS_MAXPATHLEN) {
6127 + *p++ = htonl(NFS_MAXPATHLEN);
6128 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/dlm/dlmfs.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlm/dlmfs.c
6129 +--- linux-2.6.29.4/fs/ocfs2/dlm/dlmfs.c 2009-03-24 14:22:27.000000000 +0100
6130 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlm/dlmfs.c 2009-03-24 16:10:48.000000000 +0100
6131 +@@ -43,6 +43,7 @@
6132 + #include <linux/init.h>
6133 + #include <linux/string.h>
6134 + #include <linux/backing-dev.h>
6135 ++#include <linux/vs_tag.h>
6136 +
6137 + #include <asm/uaccess.h>
6138 +
6139 +@@ -341,6 +342,7 @@ static struct inode *dlmfs_get_root_inod
6140 + inode->i_mode = mode;
6141 + inode->i_uid = current_fsuid();
6142 + inode->i_gid = current_fsgid();
6143 ++ inode->i_tag = dx_current_fstag(sb);
6144 + inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
6145 + inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
6146 + inc_nlink(inode);
6147 +@@ -366,6 +368,7 @@ static struct inode *dlmfs_get_inode(str
6148 + inode->i_mode = mode;
6149 + inode->i_uid = current_fsuid();
6150 + inode->i_gid = current_fsgid();
6151 ++ inode->i_tag = dx_current_fstag(sb);
6152 + inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
6153 + inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
6154 +
6155 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/dlmglue.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlmglue.c
6156 +--- linux-2.6.29.4/fs/ocfs2/dlmglue.c 2009-03-24 14:22:27.000000000 +0100
6157 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlmglue.c 2009-03-24 14:48:35.000000000 +0100
6158 +@@ -1885,6 +1885,7 @@ static void __ocfs2_stuff_meta_lvb(struc
6159 + lvb->lvb_iclusters = cpu_to_be32(oi->ip_clusters);
6160 + lvb->lvb_iuid = cpu_to_be32(inode->i_uid);
6161 + lvb->lvb_igid = cpu_to_be32(inode->i_gid);
6162 ++ lvb->lvb_itag = cpu_to_be16(inode->i_tag);
6163 + lvb->lvb_imode = cpu_to_be16(inode->i_mode);
6164 + lvb->lvb_inlink = cpu_to_be16(inode->i_nlink);
6165 + lvb->lvb_iatime_packed =
6166 +@@ -1939,6 +1940,7 @@ static void ocfs2_refresh_inode_from_lvb
6167 +
6168 + inode->i_uid = be32_to_cpu(lvb->lvb_iuid);
6169 + inode->i_gid = be32_to_cpu(lvb->lvb_igid);
6170 ++ inode->i_tag = be16_to_cpu(lvb->lvb_itag);
6171 + inode->i_mode = be16_to_cpu(lvb->lvb_imode);
6172 + inode->i_nlink = be16_to_cpu(lvb->lvb_inlink);
6173 + ocfs2_unpack_timespec(&inode->i_atime,
6174 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/dlmglue.h linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlmglue.h
6175 +--- linux-2.6.29.4/fs/ocfs2/dlmglue.h 2009-03-24 14:22:27.000000000 +0100
6176 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/dlmglue.h 2009-03-24 14:48:35.000000000 +0100
6177 +@@ -46,7 +46,8 @@ struct ocfs2_meta_lvb {
6178 + __be16 lvb_inlink;
6179 + __be32 lvb_iattr;
6180 + __be32 lvb_igeneration;
6181 +- __be32 lvb_reserved2;
6182 ++ __be16 lvb_itag;
6183 ++ __be16 lvb_reserved2;
6184 + };
6185 +
6186 + #define OCFS2_QINFO_LVB_VERSION 1
6187 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/file.c
6188 +--- linux-2.6.29.4/fs/ocfs2/file.c 2009-05-23 23:16:53.000000000 +0200
6189 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/file.c 2009-05-23 23:19:11.000000000 +0200
6190 +@@ -911,13 +911,15 @@ int ocfs2_setattr(struct dentry *dentry,
6191 + mlog(0, "uid change: %d\n", attr->ia_uid);
6192 + if (attr->ia_valid & ATTR_GID)
6193 + mlog(0, "gid change: %d\n", attr->ia_gid);
6194 ++ if (attr->ia_valid & ATTR_TAG)
6195 ++ mlog(0, "tag change: %d\n", attr->ia_tag);
6196 + if (attr->ia_valid & ATTR_SIZE)
6197 + mlog(0, "size change...\n");
6198 + if (attr->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME))
6199 + mlog(0, "time change...\n");
6200 +
6201 + #define OCFS2_VALID_ATTRS (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME | ATTR_SIZE \
6202 +- | ATTR_GID | ATTR_UID | ATTR_MODE)
6203 ++ | ATTR_GID | ATTR_UID | ATTR_TAG | ATTR_MODE)
6204 + if (!(attr->ia_valid & OCFS2_VALID_ATTRS)) {
6205 + mlog(0, "can't handle attrs: 0x%x\n", attr->ia_valid);
6206 + return 0;
6207 +@@ -2128,6 +2130,7 @@ const struct inode_operations ocfs2_file
6208 + .removexattr = generic_removexattr,
6209 + .fallocate = ocfs2_fallocate,
6210 + .fiemap = ocfs2_fiemap,
6211 ++ .sync_flags = ocfs2_sync_flags,
6212 + };
6213 +
6214 + const struct inode_operations ocfs2_special_file_iops = {
6215 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/inode.c
6216 +--- linux-2.6.29.4/fs/ocfs2/inode.c 2009-03-24 14:22:27.000000000 +0100
6217 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/inode.c 2009-03-24 16:11:11.000000000 +0100
6218 +@@ -29,6 +29,7 @@
6219 + #include <linux/highmem.h>
6220 + #include <linux/pagemap.h>
6221 + #include <linux/quotaops.h>
6222 ++#include <linux/vs_tag.h>
6223 +
6224 + #include <asm/byteorder.h>
6225 +
6226 +@@ -44,6 +45,7 @@
6227 + #include "file.h"
6228 + #include "heartbeat.h"
6229 + #include "inode.h"
6230 ++#include "ioctl.h"
6231 + #include "journal.h"
6232 + #include "namei.h"
6233 + #include "suballoc.h"
6234 +@@ -77,11 +79,13 @@ void ocfs2_set_inode_flags(struct inode
6235 + {
6236 + unsigned int flags = OCFS2_I(inode)->ip_attr;
6237 +
6238 +- inode->i_flags &= ~(S_IMMUTABLE |
6239 ++ inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
6240 + S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
6241 +
6242 + if (flags & OCFS2_IMMUTABLE_FL)
6243 + inode->i_flags |= S_IMMUTABLE;
6244 ++ if (flags & OCFS2_IXUNLINK_FL)
6245 ++ inode->i_flags |= S_IXUNLINK;
6246 +
6247 + if (flags & OCFS2_SYNC_FL)
6248 + inode->i_flags |= S_SYNC;
6249 +@@ -91,25 +95,85 @@ void ocfs2_set_inode_flags(struct inode
6250 + inode->i_flags |= S_NOATIME;
6251 + if (flags & OCFS2_DIRSYNC_FL)
6252 + inode->i_flags |= S_DIRSYNC;
6253 ++
6254 ++ inode->i_vflags &= ~(V_BARRIER | V_COW);
6255 ++
6256 ++ if (flags & OCFS2_BARRIER_FL)
6257 ++ inode->i_vflags |= V_BARRIER;
6258 ++ if (flags & OCFS2_COW_FL)
6259 ++ inode->i_vflags |= V_COW;
6260 + }
6261 +
6262 + /* Propagate flags from i_flags to OCFS2_I(inode)->ip_attr */
6263 + void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi)
6264 + {
6265 + unsigned int flags = oi->vfs_inode.i_flags;
6266 ++ unsigned int vflags = oi->vfs_inode.i_vflags;
6267 ++
6268 ++ oi->ip_attr &= ~(OCFS2_SYNC_FL | OCFS2_APPEND_FL |
6269 ++ OCFS2_IMMUTABLE_FL | OCFS2_IXUNLINK_FL |
6270 ++ OCFS2_NOATIME_FL | OCFS2_DIRSYNC_FL |
6271 ++ OCFS2_BARRIER_FL | OCFS2_COW_FL);
6272 ++
6273 ++ if (flags & S_IMMUTABLE)
6274 ++ oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6275 ++ if (flags & S_IXUNLINK)
6276 ++ oi->ip_attr |= OCFS2_IXUNLINK_FL;
6277 +
6278 +- oi->ip_attr &= ~(OCFS2_SYNC_FL|OCFS2_APPEND_FL|
6279 +- OCFS2_IMMUTABLE_FL|OCFS2_NOATIME_FL|OCFS2_DIRSYNC_FL);
6280 + if (flags & S_SYNC)
6281 + oi->ip_attr |= OCFS2_SYNC_FL;
6282 + if (flags & S_APPEND)
6283 + oi->ip_attr |= OCFS2_APPEND_FL;
6284 +- if (flags & S_IMMUTABLE)
6285 +- oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6286 + if (flags & S_NOATIME)
6287 + oi->ip_attr |= OCFS2_NOATIME_FL;
6288 + if (flags & S_DIRSYNC)
6289 + oi->ip_attr |= OCFS2_DIRSYNC_FL;
6290 ++
6291 ++ if (vflags & V_BARRIER)
6292 ++ oi->ip_attr |= OCFS2_BARRIER_FL;
6293 ++ if (vflags & V_COW)
6294 ++ oi->ip_attr |= OCFS2_COW_FL;
6295 ++}
6296 ++
6297 ++int ocfs2_sync_flags(struct inode *inode)
6298 ++{
6299 ++ struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6300 ++ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
6301 ++ handle_t *handle = NULL;
6302 ++ struct buffer_head *bh = NULL;
6303 ++ int status;
6304 ++
6305 ++ status = ocfs2_inode_lock(inode, &bh, 1);
6306 ++ if (status < 0) {
6307 ++ mlog_errno(status);
6308 ++ goto bail;
6309 ++ }
6310 ++
6311 ++ status = -EROFS;
6312 ++ if (IS_RDONLY(inode))
6313 ++ goto bail_unlock;
6314 ++
6315 ++ handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6316 ++ if (IS_ERR(handle)) {
6317 ++ status = PTR_ERR(handle);
6318 ++ mlog_errno(status);
6319 ++ goto bail_unlock;
6320 ++ }
6321 ++
6322 ++ ocfs2_get_inode_flags(ocfs2_inode);
6323 ++ status = ocfs2_mark_inode_dirty(handle, inode, bh);
6324 ++ if (status < 0)
6325 ++ mlog_errno(status);
6326 ++
6327 ++ ocfs2_commit_trans(osb, handle);
6328 ++bail_unlock:
6329 ++ ocfs2_inode_unlock(inode, 1);
6330 ++bail:
6331 ++ if (bh)
6332 ++ brelse(bh);
6333 ++
6334 ++ mlog_exit(status);
6335 ++ return status;
6336 + }
6337 +
6338 + struct inode *ocfs2_iget(struct ocfs2_super *osb, u64 blkno, unsigned flags,
6339 +@@ -222,6 +286,8 @@ void ocfs2_populate_inode(struct inode *
6340 + struct super_block *sb;
6341 + struct ocfs2_super *osb;
6342 + int use_plocks = 1;
6343 ++ uid_t uid;
6344 ++ gid_t gid;
6345 +
6346 + mlog_entry("(0x%p, size:%llu)\n", inode,
6347 + (unsigned long long)le64_to_cpu(fe->i_size));
6348 +@@ -253,8 +319,12 @@ void ocfs2_populate_inode(struct inode *
6349 + inode->i_generation = le32_to_cpu(fe->i_generation);
6350 + inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
6351 + inode->i_mode = le16_to_cpu(fe->i_mode);
6352 +- inode->i_uid = le32_to_cpu(fe->i_uid);
6353 +- inode->i_gid = le32_to_cpu(fe->i_gid);
6354 ++ uid = le32_to_cpu(fe->i_uid);
6355 ++ gid = le32_to_cpu(fe->i_gid);
6356 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6357 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6358 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6359 ++ /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6360 +
6361 + /* Fast symlinks will have i_size but no allocated clusters. */
6362 + if (S_ISLNK(inode->i_mode) && !fe->i_clusters)
6363 +@@ -1206,8 +1276,11 @@ int ocfs2_mark_inode_dirty(handle_t *han
6364 +
6365 + fe->i_size = cpu_to_le64(i_size_read(inode));
6366 + fe->i_links_count = cpu_to_le16(inode->i_nlink);
6367 +- fe->i_uid = cpu_to_le32(inode->i_uid);
6368 +- fe->i_gid = cpu_to_le32(inode->i_gid);
6369 ++ fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode),
6370 ++ inode->i_uid, inode->i_tag));
6371 ++ fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode),
6372 ++ inode->i_gid, inode->i_tag));
6373 ++ /* i_tag = = cpu_to_le16(inode->i_tag); */
6374 + fe->i_mode = cpu_to_le16(inode->i_mode);
6375 + fe->i_atime = cpu_to_le64(inode->i_atime.tv_sec);
6376 + fe->i_atime_nsec = cpu_to_le32(inode->i_atime.tv_nsec);
6377 +@@ -1235,16 +1308,25 @@ leave:
6378 + void ocfs2_refresh_inode(struct inode *inode,
6379 + struct ocfs2_dinode *fe)
6380 + {
6381 ++ uid_t uid;
6382 ++ gid_t gid;
6383 ++
6384 + spin_lock(&OCFS2_I(inode)->ip_lock);
6385 +
6386 + OCFS2_I(inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
6387 + OCFS2_I(inode)->ip_attr = le32_to_cpu(fe->i_attr);
6388 ++ /* OCFS2_I(inode)->ip_flags &= ~OCFS2_FL_MASK;
6389 ++ OCFS2_I(inode)->ip_flags |= le32_to_cpu(fe->i_flags) & OCFS2_FL_MASK; */
6390 + OCFS2_I(inode)->ip_dyn_features = le16_to_cpu(fe->i_dyn_features);
6391 + ocfs2_set_inode_flags(inode);
6392 + i_size_write(inode, le64_to_cpu(fe->i_size));
6393 + inode->i_nlink = le16_to_cpu(fe->i_links_count);
6394 +- inode->i_uid = le32_to_cpu(fe->i_uid);
6395 +- inode->i_gid = le32_to_cpu(fe->i_gid);
6396 ++ uid = le32_to_cpu(fe->i_uid);
6397 ++ gid = le32_to_cpu(fe->i_gid);
6398 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6399 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6400 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6401 ++ /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6402 + inode->i_mode = le16_to_cpu(fe->i_mode);
6403 + if (S_ISLNK(inode->i_mode) && le32_to_cpu(fe->i_clusters) == 0)
6404 + inode->i_blocks = 0;
6405 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/inode.h linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/inode.h
6406 +--- linux-2.6.29.4/fs/ocfs2/inode.h 2009-03-24 14:22:27.000000000 +0100
6407 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/inode.h 2009-03-24 14:48:35.000000000 +0100
6408 +@@ -147,6 +147,7 @@ struct buffer_head *ocfs2_bread(struct i
6409 +
6410 + void ocfs2_set_inode_flags(struct inode *inode);
6411 + void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi);
6412 ++int ocfs2_sync_flags(struct inode *inode);
6413 +
6414 + static inline blkcnt_t ocfs2_inode_sector_count(struct inode *inode)
6415 + {
6416 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ioctl.c
6417 +--- linux-2.6.29.4/fs/ocfs2/ioctl.c 2008-12-25 00:26:37.000000000 +0100
6418 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ioctl.c 2009-02-22 22:54:25.000000000 +0100
6419 +@@ -42,7 +42,7 @@ static int ocfs2_get_inode_attr(struct i
6420 + return status;
6421 + }
6422 +
6423 +-static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6424 ++int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6425 + unsigned mask)
6426 + {
6427 + struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6428 +@@ -67,6 +67,11 @@ static int ocfs2_set_inode_attr(struct i
6429 + if (!S_ISDIR(inode->i_mode))
6430 + flags &= ~OCFS2_DIRSYNC_FL;
6431 +
6432 ++ if (IS_BARRIER(inode)) {
6433 ++ vxwprintk_task(1, "messing with the barrier.");
6434 ++ goto bail_unlock;
6435 ++ }
6436 ++
6437 + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6438 + if (IS_ERR(handle)) {
6439 + status = PTR_ERR(handle);
6440 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/ioctl.h linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ioctl.h
6441 +--- linux-2.6.29.4/fs/ocfs2/ioctl.h 2008-12-25 00:26:37.000000000 +0100
6442 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ioctl.h 2009-02-22 22:54:25.000000000 +0100
6443 +@@ -10,6 +10,9 @@
6444 + #ifndef OCFS2_IOCTL_H
6445 + #define OCFS2_IOCTL_H
6446 +
6447 ++int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6448 ++ unsigned mask);
6449 ++
6450 + long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
6451 + long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg);
6452 +
6453 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/namei.c
6454 +--- linux-2.6.29.4/fs/ocfs2/namei.c 2009-03-24 14:22:27.000000000 +0100
6455 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/namei.c 2009-03-25 01:04:31.000000000 +0100
6456 +@@ -41,6 +41,7 @@
6457 + #include <linux/slab.h>
6458 + #include <linux/highmem.h>
6459 + #include <linux/quotaops.h>
6460 ++#include <linux/vs_tag.h>
6461 +
6462 + #define MLOG_MASK_PREFIX ML_NAMEI
6463 + #include <cluster/masklog.h>
6464 +@@ -462,6 +463,7 @@ static int ocfs2_mknod_locked(struct ocf
6465 + struct ocfs2_extent_list *fel;
6466 + u64 fe_blkno = 0;
6467 + u16 suballoc_bit;
6468 ++ tag_t tag;
6469 +
6470 + mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry,
6471 + inode->i_mode, (unsigned long)dev, dentry->d_name.len,
6472 +@@ -508,8 +510,11 @@ static int ocfs2_mknod_locked(struct ocf
6473 + fe->i_blkno = cpu_to_le64(fe_blkno);
6474 + fe->i_suballoc_bit = cpu_to_le16(suballoc_bit);
6475 + fe->i_suballoc_slot = cpu_to_le16(inode_ac->ac_alloc_slot);
6476 +- fe->i_uid = cpu_to_le32(inode->i_uid);
6477 +- fe->i_gid = cpu_to_le32(inode->i_gid);
6478 ++
6479 ++ tag = dx_current_fstag(osb->sb);
6480 ++ fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode), inode->i_uid, tag));
6481 ++ fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode), inode->i_gid, tag));
6482 ++ inode->i_tag = tag;
6483 + fe->i_mode = cpu_to_le16(inode->i_mode);
6484 + if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode))
6485 + fe->id1.dev1.i_rdev = cpu_to_le64(huge_encode_dev(dev));
6486 +@@ -2025,6 +2030,7 @@ const struct inode_operations ocfs2_dir_
6487 + .rename = ocfs2_rename,
6488 + .setattr = ocfs2_setattr,
6489 + .getattr = ocfs2_getattr,
6490 ++ .sync_flags = ocfs2_sync_flags,
6491 + .permission = ocfs2_permission,
6492 + .setxattr = generic_setxattr,
6493 + .getxattr = generic_getxattr,
6494 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/ocfs2_fs.h linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ocfs2_fs.h
6495 +--- linux-2.6.29.4/fs/ocfs2/ocfs2_fs.h 2009-03-24 14:22:27.000000000 +0100
6496 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ocfs2_fs.h 2009-03-24 14:48:35.000000000 +0100
6497 +@@ -219,18 +219,23 @@
6498 + #define OCFS2_INDEXED_DIR_FL (0x0008)
6499 +
6500 + /* Inode attributes, keep in sync with EXT2 */
6501 +-#define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
6502 +-#define OCFS2_UNRM_FL (0x00000002) /* Undelete */
6503 +-#define OCFS2_COMPR_FL (0x00000004) /* Compress file */
6504 +-#define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
6505 +-#define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
6506 +-#define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
6507 +-#define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
6508 +-#define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
6509 +-#define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
6510 ++#define OCFS2_SECRM_FL FS_SECRM_FL /* Secure deletion */
6511 ++#define OCFS2_UNRM_FL FS_UNRM_FL /* Undelete */
6512 ++#define OCFS2_COMPR_FL FS_COMPR_FL /* Compress file */
6513 ++#define OCFS2_SYNC_FL FS_SYNC_FL /* Synchronous updates */
6514 ++#define OCFS2_IMMUTABLE_FL FS_IMMUTABLE_FL /* Immutable file */
6515 ++#define OCFS2_APPEND_FL FS_APPEND_FL /* writes to file may only append */
6516 ++#define OCFS2_NODUMP_FL FS_NODUMP_FL /* do not dump file */
6517 ++#define OCFS2_NOATIME_FL FS_NOATIME_FL /* do not update atime */
6518 +
6519 +-#define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
6520 +-#define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
6521 ++#define OCFS2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
6522 ++#define OCFS2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
6523 ++
6524 ++#define OCFS2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
6525 ++#define OCFS2_COW_FL FS_COW_FL /* Copy on Write marker */
6526 ++
6527 ++#define OCFS2_FL_VISIBLE (0x010300FF) /* User visible flags */
6528 ++#define OCFS2_FL_MODIFIABLE (0x010300FF) /* User modifiable flags */
6529 +
6530 + /*
6531 + * Extent record flags (e_node.leaf.flags)
6532 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/ocfs2.h linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ocfs2.h
6533 +--- linux-2.6.29.4/fs/ocfs2/ocfs2.h 2009-03-24 14:22:27.000000000 +0100
6534 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/ocfs2.h 2009-03-24 16:18:22.000000000 +0100
6535 +@@ -199,6 +199,7 @@ enum ocfs2_mount_options
6536 + OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* POSIX access control lists */
6537 + OCFS2_MOUNT_USRQUOTA = 1 << 9, /* We support user quotas */
6538 + OCFS2_MOUNT_GRPQUOTA = 1 << 10, /* We support group quotas */
6539 ++ OCFS2_MOUNT_TAGGED = 1 << 11, /* use tagging */
6540 + };
6541 +
6542 + #define OCFS2_OSB_SOFT_RO 0x0001
6543 +diff -NurpP --minimal linux-2.6.29.4/fs/ocfs2/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/super.c
6544 +--- linux-2.6.29.4/fs/ocfs2/super.c 2009-03-24 14:22:27.000000000 +0100
6545 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/ocfs2/super.c 2009-03-24 16:19:44.000000000 +0100
6546 +@@ -172,6 +172,7 @@ enum {
6547 + Opt_noacl,
6548 + Opt_usrquota,
6549 + Opt_grpquota,
6550 ++ Opt_tag, Opt_notag, Opt_tagid,
6551 + Opt_err,
6552 + };
6553 +
6554 +@@ -198,6 +199,9 @@ static const match_table_t tokens = {
6555 + {Opt_noacl, "noacl"},
6556 + {Opt_usrquota, "usrquota"},
6557 + {Opt_grpquota, "grpquota"},
6558 ++ {Opt_tag, "tag"},
6559 ++ {Opt_notag, "notag"},
6560 ++ {Opt_tagid, "tagid=%u"},
6561 + {Opt_err, NULL}
6562 + };
6563 +
6564 +@@ -436,6 +440,13 @@ static int ocfs2_remount(struct super_bl
6565 + goto out;
6566 + }
6567 +
6568 ++ if ((parsed_options.mount_opt & OCFS2_MOUNT_TAGGED) &&
6569 ++ !(sb->s_flags & MS_TAGGED)) {
6570 ++ ret = -EINVAL;
6571 ++ mlog(ML_ERROR, "Cannot change tagging on remount\n");
6572 ++ goto out;
6573 ++ }
6574 ++
6575 + if ((osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) !=
6576 + (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
6577 + ret = -EINVAL;
6578 +@@ -950,6 +961,9 @@ static int ocfs2_fill_super(struct super
6579 +
6580 + ocfs2_complete_mount_recovery(osb);
6581 +
6582 ++ if (osb->s_mount_opt & OCFS2_MOUNT_TAGGED)
6583 ++ sb->s_flags |= MS_TAGGED;
6584 ++
6585 + if (ocfs2_mount_local(osb))
6586 + snprintf(nodestr, sizeof(nodestr), "local");
6587 + else
6588 +@@ -1208,6 +1222,20 @@ static int ocfs2_parse_options(struct su
6589 + printk(KERN_INFO "ocfs2 (no)acl options not supported\n");
6590 + break;
6591 + #endif
6592 ++#ifndef CONFIG_TAGGING_NONE
6593 ++ case Opt_tag:
6594 ++ mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6595 ++ break;
6596 ++ case Opt_notag:
6597 ++ mopt->mount_opt &= ~OCFS2_MOUNT_TAGGED;
6598 ++ break;
6599 ++#endif
6600 ++#ifdef CONFIG_PROPAGATE
6601 ++ case Opt_tagid:
6602 ++ /* use args[0] */
6603 ++ mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6604 ++ break;
6605 ++#endif
6606 + default:
6607 + mlog(ML_ERROR,
6608 + "Unrecognized mount option \"%s\" "
6609 +diff -NurpP --minimal linux-2.6.29.4/fs/open.c linux-2.6.29.4-vs2.3.0.36.14/fs/open.c
6610 +--- linux-2.6.29.4/fs/open.c 2009-03-24 14:22:27.000000000 +0100
6611 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/open.c 2009-03-24 14:48:35.000000000 +0100
6612 +@@ -29,22 +29,31 @@
6613 + #include <linux/rcupdate.h>
6614 + #include <linux/audit.h>
6615 + #include <linux/falloc.h>
6616 ++#include <linux/vs_base.h>
6617 ++#include <linux/vs_limit.h>
6618 ++#include <linux/vs_dlimit.h>
6619 ++#include <linux/vs_tag.h>
6620 ++#include <linux/vs_cowbl.h>
6621 +
6622 + int vfs_statfs(struct dentry *dentry, struct kstatfs *buf)
6623 + {
6624 + int retval = -ENODEV;
6625 +
6626 + if (dentry) {
6627 ++ struct super_block *sb = dentry->d_sb;
6628 ++
6629 + retval = -ENOSYS;
6630 +- if (dentry->d_sb->s_op->statfs) {
6631 ++ if (sb->s_op->statfs) {
6632 + memset(buf, 0, sizeof(*buf));
6633 + retval = security_sb_statfs(dentry);
6634 + if (retval)
6635 + return retval;
6636 +- retval = dentry->d_sb->s_op->statfs(dentry, buf);
6637 ++ retval = sb->s_op->statfs(dentry, buf);
6638 + if (retval == 0 && buf->f_frsize == 0)
6639 + buf->f_frsize = buf->f_bsize;
6640 + }
6641 ++ if (!vx_check(0, VS_ADMIN|VS_WATCH))
6642 ++ vx_vsi_statfs(sb, buf);
6643 + }
6644 + return retval;
6645 + }
6646 +@@ -638,6 +647,10 @@ SYSCALL_DEFINE3(fchmodat, int, dfd, cons
6647 + error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
6648 + if (error)
6649 + goto out;
6650 ++
6651 ++ error = cow_check_and_break(&path);
6652 ++ if (error)
6653 ++ goto dput_and_out;
6654 + inode = path.dentry->d_inode;
6655 +
6656 + error = mnt_want_write(path.mnt);
6657 +@@ -671,11 +684,11 @@ static int chown_common(struct dentry *
6658 + newattrs.ia_valid = ATTR_CTIME;
6659 + if (user != (uid_t) -1) {
6660 + newattrs.ia_valid |= ATTR_UID;
6661 +- newattrs.ia_uid = user;
6662 ++ newattrs.ia_uid = dx_map_uid(user);
6663 + }
6664 + if (group != (gid_t) -1) {
6665 + newattrs.ia_valid |= ATTR_GID;
6666 +- newattrs.ia_gid = group;
6667 ++ newattrs.ia_gid = dx_map_gid(group);
6668 + }
6669 + if (!S_ISDIR(inode->i_mode))
6670 + newattrs.ia_valid |=
6671 +@@ -698,7 +711,11 @@ SYSCALL_DEFINE3(chown, const char __user
6672 + error = mnt_want_write(path.mnt);
6673 + if (error)
6674 + goto out_release;
6675 +- error = chown_common(path.dentry, user, group);
6676 ++#ifdef CONFIG_VSERVER_COWBL
6677 ++ error = cow_check_and_break(&path);
6678 ++ if (!error)
6679 ++#endif
6680 ++ error = chown_common(path.dentry, user, group);
6681 + mnt_drop_write(path.mnt);
6682 + out_release:
6683 + path_put(&path);
6684 +@@ -723,7 +740,11 @@ SYSCALL_DEFINE5(fchownat, int, dfd, cons
6685 + error = mnt_want_write(path.mnt);
6686 + if (error)
6687 + goto out_release;
6688 +- error = chown_common(path.dentry, user, group);
6689 ++#ifdef CONFIG_VSERVER_COWBL
6690 ++ error = cow_check_and_break(&path);
6691 ++ if (!error)
6692 ++#endif
6693 ++ error = chown_common(path.dentry, user, group);
6694 + mnt_drop_write(path.mnt);
6695 + out_release:
6696 + path_put(&path);
6697 +@@ -742,7 +763,11 @@ SYSCALL_DEFINE3(lchown, const char __use
6698 + error = mnt_want_write(path.mnt);
6699 + if (error)
6700 + goto out_release;
6701 +- error = chown_common(path.dentry, user, group);
6702 ++#ifdef CONFIG_VSERVER_COWBL
6703 ++ error = cow_check_and_break(&path);
6704 ++ if (!error)
6705 ++#endif
6706 ++ error = chown_common(path.dentry, user, group);
6707 + mnt_drop_write(path.mnt);
6708 + out_release:
6709 + path_put(&path);
6710 +@@ -986,6 +1011,7 @@ static void __put_unused_fd(struct files
6711 + __FD_CLR(fd, fdt->open_fds);
6712 + if (fd < files->next_fd)
6713 + files->next_fd = fd;
6714 ++ vx_openfd_dec(fd);
6715 + }
6716 +
6717 + void put_unused_fd(unsigned int fd)
6718 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/array.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/array.c
6719 +--- linux-2.6.29.4/fs/proc/array.c 2009-05-23 23:16:53.000000000 +0200
6720 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/array.c 2009-05-10 23:42:01.000000000 +0200
6721 +@@ -82,6 +82,8 @@
6722 + #include <linux/pid_namespace.h>
6723 + #include <linux/ptrace.h>
6724 + #include <linux/tracehook.h>
6725 ++#include <linux/vs_context.h>
6726 ++#include <linux/vs_network.h>
6727 +
6728 + #include <asm/pgtable.h>
6729 + #include <asm/processor.h>
6730 +@@ -138,8 +140,9 @@ static const char *task_state_array[] =
6731 + "D (disk sleep)", /* 2 */
6732 + "T (stopped)", /* 4 */
6733 + "T (tracing stop)", /* 8 */
6734 +- "Z (zombie)", /* 16 */
6735 +- "X (dead)" /* 32 */
6736 ++ "H (on hold)", /* 16 */
6737 ++ "Z (zombie)", /* 32 */
6738 ++ "X (dead)", /* 64 */
6739 + };
6740 +
6741 + static inline const char *get_task_state(struct task_struct *tsk)
6742 +@@ -166,6 +169,9 @@ static inline void task_state(struct seq
6743 + rcu_read_lock();
6744 + ppid = pid_alive(p) ?
6745 + task_tgid_nr_ns(rcu_dereference(p->real_parent), ns) : 0;
6746 ++ if (unlikely(vx_current_initpid(p->pid)))
6747 ++ ppid = 0;
6748 ++
6749 + tpid = 0;
6750 + if (pid_alive(p)) {
6751 + struct task_struct *tracer = tracehook_tracer_task(p);
6752 +@@ -281,7 +287,7 @@ static inline void task_sig(struct seq_f
6753 + }
6754 +
6755 + static void render_cap_t(struct seq_file *m, const char *header,
6756 +- kernel_cap_t *a)
6757 ++ struct vx_info *vxi, kernel_cap_t *a)
6758 + {
6759 + unsigned __capi;
6760 +
6761 +@@ -306,10 +312,11 @@ static inline void task_cap(struct seq_f
6762 + cap_bset = cred->cap_bset;
6763 + rcu_read_unlock();
6764 +
6765 +- render_cap_t(m, "CapInh:\t", &cap_inheritable);
6766 +- render_cap_t(m, "CapPrm:\t", &cap_permitted);
6767 +- render_cap_t(m, "CapEff:\t", &cap_effective);
6768 +- render_cap_t(m, "CapBnd:\t", &cap_bset);
6769 ++ /* FIXME: maybe move the p->vx_info masking to __task_cred() ? */
6770 ++ render_cap_t(m, "CapInh:\t", p->vx_info, &cap_inheritable);
6771 ++ render_cap_t(m, "CapPrm:\t", p->vx_info, &cap_permitted);
6772 ++ render_cap_t(m, "CapEff:\t", p->vx_info, &cap_effective);
6773 ++ render_cap_t(m, "CapBnd:\t", p->vx_info, &cap_bset);
6774 + }
6775 +
6776 + static inline void task_context_switch_counts(struct seq_file *m,
6777 +@@ -321,6 +328,42 @@ static inline void task_context_switch_c
6778 + p->nivcsw);
6779 + }
6780 +
6781 ++int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
6782 ++ struct pid *pid, struct task_struct *task)
6783 ++{
6784 ++ seq_printf(m, "Proxy:\t%p(%c)\n"
6785 ++ "Count:\t%u\n"
6786 ++ "uts:\t%p(%c)\n"
6787 ++ "ipc:\t%p(%c)\n"
6788 ++ "mnt:\t%p(%c)\n"
6789 ++ "pid:\t%p(%c)\n"
6790 ++ "net:\t%p(%c)\n",
6791 ++ task->nsproxy,
6792 ++ (task->nsproxy == init_task.nsproxy ? 'I' : '-'),
6793 ++ atomic_read(&task->nsproxy->count),
6794 ++ task->nsproxy->uts_ns,
6795 ++ (task->nsproxy->uts_ns == init_task.nsproxy->uts_ns ? 'I' : '-'),
6796 ++ task->nsproxy->ipc_ns,
6797 ++ (task->nsproxy->ipc_ns == init_task.nsproxy->ipc_ns ? 'I' : '-'),
6798 ++ task->nsproxy->mnt_ns,
6799 ++ (task->nsproxy->mnt_ns == init_task.nsproxy->mnt_ns ? 'I' : '-'),
6800 ++ task->nsproxy->pid_ns,
6801 ++ (task->nsproxy->pid_ns == init_task.nsproxy->pid_ns ? 'I' : '-'),
6802 ++ task->nsproxy->net_ns,
6803 ++ (task->nsproxy->net_ns == init_task.nsproxy->net_ns ? 'I' : '-'));
6804 ++ return 0;
6805 ++}
6806 ++
6807 ++void task_vs_id(struct seq_file *m, struct task_struct *task)
6808 ++{
6809 ++ if (task_vx_flags(task, VXF_HIDE_VINFO, 0))
6810 ++ return;
6811 ++
6812 ++ seq_printf(m, "VxID: %d\n", vx_task_xid(task));
6813 ++ seq_printf(m, "NxID: %d\n", nx_task_nid(task));
6814 ++}
6815 ++
6816 ++
6817 + int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
6818 + struct pid *pid, struct task_struct *task)
6819 + {
6820 +@@ -336,6 +379,7 @@ int proc_pid_status(struct seq_file *m,
6821 + task_sig(m, task);
6822 + task_cap(m, task);
6823 + cpuset_task_status_allowed(m, task);
6824 ++ task_vs_id(m, task);
6825 + #if defined(CONFIG_S390)
6826 + task_show_regs(m, task);
6827 + #endif
6828 +@@ -452,6 +496,17 @@ static int do_task_stat(struct seq_file
6829 + /* convert nsec -> ticks */
6830 + start_time = nsec_to_clock_t(start_time);
6831 +
6832 ++ /* fixup start time for virt uptime */
6833 ++ if (vx_flags(VXF_VIRT_UPTIME, 0)) {
6834 ++ unsigned long long bias =
6835 ++ current->vx_info->cvirt.bias_clock;
6836 ++
6837 ++ if (start_time > bias)
6838 ++ start_time -= bias;
6839 ++ else
6840 ++ start_time = 0;
6841 ++ }
6842 ++
6843 + seq_printf(m, "%d (%s) %c %d %d %d %d %d %u %lu \
6844 + %lu %lu %lu %lu %lu %ld %ld %ld %ld %d 0 %llu %lu %ld %lu %lu %lu %lu %lu \
6845 + %lu %lu %lu %lu %lu %lu %lu %lu %d %d %u %u %llu %lu %ld\n",
6846 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/base.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/base.c
6847 +--- linux-2.6.29.4/fs/proc/base.c 2009-05-23 23:16:53.000000000 +0200
6848 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/base.c 2009-05-10 23:42:01.000000000 +0200
6849 +@@ -80,6 +80,8 @@
6850 + #include <linux/oom.h>
6851 + #include <linux/elf.h>
6852 + #include <linux/pid_namespace.h>
6853 ++#include <linux/vs_context.h>
6854 ++#include <linux/vs_network.h>
6855 + #include "internal.h"
6856 +
6857 + /* NOTE:
6858 +@@ -1443,6 +1445,8 @@ static struct inode *proc_pid_make_inode
6859 + inode->i_gid = cred->egid;
6860 + rcu_read_unlock();
6861 + }
6862 ++ /* procfs is xid tagged */
6863 ++ inode->i_tag = (tag_t)vx_task_xid(task);
6864 + security_task_to_inode(task, inode);
6865 +
6866 + out:
6867 +@@ -1993,6 +1997,13 @@ static struct dentry *proc_pident_lookup
6868 + if (!task)
6869 + goto out_no_task;
6870 +
6871 ++ /* TODO: maybe we can come up with a generic approach? */
6872 ++ if (task_vx_flags(task, VXF_HIDE_VINFO, 0) &&
6873 ++ (dentry->d_name.len == 5) &&
6874 ++ (!memcmp(dentry->d_name.name, "vinfo", 5) ||
6875 ++ !memcmp(dentry->d_name.name, "ninfo", 5)))
6876 ++ goto out;
6877 ++
6878 + /*
6879 + * Yes, it does not scale. And it should not. Don't add
6880 + * new entries into /proc/<tgid>/ without very good reasons.
6881 +@@ -2378,7 +2389,7 @@ out_iput:
6882 + static struct dentry *proc_base_lookup(struct inode *dir, struct dentry *dentry)
6883 + {
6884 + struct dentry *error;
6885 +- struct task_struct *task = get_proc_task(dir);
6886 ++ struct task_struct *task = get_proc_task_real(dir);
6887 + const struct pid_entry *p, *last;
6888 +
6889 + error = ERR_PTR(-ENOENT);
6890 +@@ -2468,6 +2479,9 @@ static int proc_pid_personality(struct s
6891 + static const struct file_operations proc_task_operations;
6892 + static const struct inode_operations proc_task_inode_operations;
6893 +
6894 ++extern int proc_pid_vx_info(struct task_struct *, char *);
6895 ++extern int proc_pid_nx_info(struct task_struct *, char *);
6896 ++
6897 + static const struct pid_entry tgid_base_stuff[] = {
6898 + DIR("task", S_IRUGO|S_IXUGO, proc_task_inode_operations, proc_task_operations),
6899 + DIR("fd", S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
6900 +@@ -2526,6 +2540,8 @@ static const struct pid_entry tgid_base_
6901 + #ifdef CONFIG_CGROUPS
6902 + REG("cgroup", S_IRUGO, proc_cgroup_operations),
6903 + #endif
6904 ++ INF("vinfo", S_IRUGO, proc_pid_vx_info),
6905 ++ INF("ninfo", S_IRUGO, proc_pid_nx_info),
6906 + INF("oom_score", S_IRUGO, proc_oom_score),
6907 + REG("oom_adj", S_IRUGO|S_IWUSR, proc_oom_adjust_operations),
6908 + #ifdef CONFIG_AUDITSYSCALL
6909 +@@ -2541,6 +2557,7 @@ static const struct pid_entry tgid_base_
6910 + #ifdef CONFIG_TASK_IO_ACCOUNTING
6911 + INF("io", S_IRUGO, proc_tgid_io_accounting),
6912 + #endif
6913 ++ ONE("nsproxy", S_IRUGO, proc_pid_nsproxy),
6914 + };
6915 +
6916 + static int proc_tgid_base_readdir(struct file * filp,
6917 +@@ -2737,7 +2754,7 @@ retry:
6918 + iter.task = NULL;
6919 + pid = find_ge_pid(iter.tgid, ns);
6920 + if (pid) {
6921 +- iter.tgid = pid_nr_ns(pid, ns);
6922 ++ iter.tgid = pid_unmapped_nr_ns(pid, ns);
6923 + iter.task = pid_task(pid, PIDTYPE_PID);
6924 + /* What we to know is if the pid we have find is the
6925 + * pid of a thread_group_leader. Testing for task
6926 +@@ -2767,7 +2784,7 @@ static int proc_pid_fill_cache(struct fi
6927 + struct tgid_iter iter)
6928 + {
6929 + char name[PROC_NUMBUF];
6930 +- int len = snprintf(name, sizeof(name), "%d", iter.tgid);
6931 ++ int len = snprintf(name, sizeof(name), "%d", vx_map_tgid(iter.tgid));
6932 + return proc_fill_cache(filp, dirent, filldir, name, len,
6933 + proc_pid_instantiate, iter.task, NULL);
6934 + }
6935 +@@ -2776,7 +2793,7 @@ static int proc_pid_fill_cache(struct fi
6936 + int proc_pid_readdir(struct file * filp, void * dirent, filldir_t filldir)
6937 + {
6938 + unsigned int nr = filp->f_pos - FIRST_PROCESS_ENTRY;
6939 +- struct task_struct *reaper = get_proc_task(filp->f_path.dentry->d_inode);
6940 ++ struct task_struct *reaper = get_proc_task_real(filp->f_path.dentry->d_inode);
6941 + struct tgid_iter iter;
6942 + struct pid_namespace *ns;
6943 +
6944 +@@ -2796,6 +2813,8 @@ int proc_pid_readdir(struct file * filp,
6945 + iter.task;
6946 + iter.tgid += 1, iter = next_tgid(ns, iter)) {
6947 + filp->f_pos = iter.tgid + TGID_OFFSET;
6948 ++ if (!vx_proc_task_visible(iter.task))
6949 ++ continue;
6950 + if (proc_pid_fill_cache(filp, dirent, filldir, iter) < 0) {
6951 + put_task_struct(iter.task);
6952 + goto out;
6953 +@@ -2942,6 +2961,8 @@ static struct dentry *proc_task_lookup(s
6954 + tid = name_to_int(dentry);
6955 + if (tid == ~0U)
6956 + goto out;
6957 ++ if (vx_current_initpid(tid))
6958 ++ goto out;
6959 +
6960 + ns = dentry->d_sb->s_fs_info;
6961 + rcu_read_lock();
6962 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/generic.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/generic.c
6963 +--- linux-2.6.29.4/fs/proc/generic.c 2009-03-24 14:22:27.000000000 +0100
6964 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/generic.c 2009-03-24 14:48:35.000000000 +0100
6965 +@@ -20,6 +20,7 @@
6966 + #include <linux/bitops.h>
6967 + #include <linux/spinlock.h>
6968 + #include <linux/completion.h>
6969 ++#include <linux/vserver/inode.h>
6970 + #include <asm/uaccess.h>
6971 +
6972 + #include "internal.h"
6973 +@@ -382,6 +383,8 @@ struct dentry *proc_lookup_de(struct pro
6974 + for (de = de->subdir; de ; de = de->next) {
6975 + if (de->namelen != dentry->d_name.len)
6976 + continue;
6977 ++ if (!vx_hide_check(0, de->vx_flags))
6978 ++ continue;
6979 + if (!memcmp(dentry->d_name.name, de->name, de->namelen)) {
6980 + unsigned int ino;
6981 +
6982 +@@ -390,6 +393,8 @@ struct dentry *proc_lookup_de(struct pro
6983 + spin_unlock(&proc_subdir_lock);
6984 + error = -EINVAL;
6985 + inode = proc_get_inode(dir->i_sb, ino, de);
6986 ++ /* generic proc entries belong to the host */
6987 ++ inode->i_tag = 0;
6988 + goto out_unlock;
6989 + }
6990 + }
6991 +@@ -467,6 +472,8 @@ int proc_readdir_de(struct proc_dir_entr
6992 +
6993 + /* filldir passes info to user space */
6994 + de_get(de);
6995 ++ if (!vx_hide_check(0, de->vx_flags))
6996 ++ goto skip;
6997 + spin_unlock(&proc_subdir_lock);
6998 + if (filldir(dirent, de->name, de->namelen, filp->f_pos,
6999 + de->low_ino, de->mode >> 12) < 0) {
7000 +@@ -474,6 +481,7 @@ int proc_readdir_de(struct proc_dir_entr
7001 + goto out;
7002 + }
7003 + spin_lock(&proc_subdir_lock);
7004 ++ skip:
7005 + filp->f_pos++;
7006 + next = de->next;
7007 + de_put(de);
7008 +@@ -588,6 +596,7 @@ static struct proc_dir_entry *__proc_cre
7009 + ent->nlink = nlink;
7010 + atomic_set(&ent->count, 1);
7011 + ent->pde_users = 0;
7012 ++ ent->vx_flags = IATTR_PROC_DEFAULT;
7013 + spin_lock_init(&ent->pde_unload_lock);
7014 + ent->pde_unload_completion = NULL;
7015 + INIT_LIST_HEAD(&ent->pde_openers);
7016 +@@ -611,7 +620,8 @@ struct proc_dir_entry *proc_symlink(cons
7017 + kfree(ent->data);
7018 + kfree(ent);
7019 + ent = NULL;
7020 +- }
7021 ++ } else
7022 ++ ent->vx_flags = IATTR_PROC_SYMLINK;
7023 + } else {
7024 + kfree(ent);
7025 + ent = NULL;
7026 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/inode.c
7027 +--- linux-2.6.29.4/fs/proc/inode.c 2009-03-24 14:22:27.000000000 +0100
7028 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/inode.c 2009-03-24 14:48:35.000000000 +0100
7029 +@@ -465,6 +465,8 @@ struct inode *proc_get_inode(struct supe
7030 + inode->i_uid = de->uid;
7031 + inode->i_gid = de->gid;
7032 + }
7033 ++ if (de->vx_flags)
7034 ++ PROC_I(inode)->vx_flags = de->vx_flags;
7035 + if (de->size)
7036 + inode->i_size = de->size;
7037 + if (de->nlink)
7038 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/internal.h linux-2.6.29.4-vs2.3.0.36.14/fs/proc/internal.h
7039 +--- linux-2.6.29.4/fs/proc/internal.h 2009-03-24 14:22:27.000000000 +0100
7040 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/internal.h 2009-03-24 14:48:35.000000000 +0100
7041 +@@ -10,6 +10,7 @@
7042 + */
7043 +
7044 + #include <linux/proc_fs.h>
7045 ++#include <linux/vs_pid.h>
7046 +
7047 + extern struct proc_dir_entry proc_root;
7048 + #ifdef CONFIG_PROC_SYSCTL
7049 +@@ -51,6 +52,9 @@ extern int proc_pid_status(struct seq_fi
7050 + struct pid *pid, struct task_struct *task);
7051 + extern int proc_pid_statm(struct seq_file *m, struct pid_namespace *ns,
7052 + struct pid *pid, struct task_struct *task);
7053 ++extern int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
7054 ++ struct pid *pid, struct task_struct *task);
7055 ++
7056 + extern loff_t mem_lseek(struct file *file, loff_t offset, int orig);
7057 +
7058 + extern const struct file_operations proc_maps_operations;
7059 +@@ -70,11 +74,16 @@ static inline struct pid *proc_pid(struc
7060 + return PROC_I(inode)->pid;
7061 + }
7062 +
7063 +-static inline struct task_struct *get_proc_task(struct inode *inode)
7064 ++static inline struct task_struct *get_proc_task_real(struct inode *inode)
7065 + {
7066 + return get_pid_task(proc_pid(inode), PIDTYPE_PID);
7067 + }
7068 +
7069 ++static inline struct task_struct *get_proc_task(struct inode *inode)
7070 ++{
7071 ++ return vx_get_proc_task(inode, proc_pid(inode));
7072 ++}
7073 ++
7074 + static inline int proc_fd(struct inode *inode)
7075 + {
7076 + return PROC_I(inode)->fd;
7077 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/loadavg.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/loadavg.c
7078 +--- linux-2.6.29.4/fs/proc/loadavg.c 2008-12-25 00:26:37.000000000 +0100
7079 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/loadavg.c 2009-05-20 00:24:34.000000000 +0200
7080 +@@ -12,21 +12,37 @@
7081 +
7082 + static int loadavg_proc_show(struct seq_file *m, void *v)
7083 + {
7084 ++ unsigned long running;
7085 ++ unsigned int threads;
7086 + int a, b, c;
7087 + unsigned long seq;
7088 +
7089 + do {
7090 + seq = read_seqbegin(&xtime_lock);
7091 +- a = avenrun[0] + (FIXED_1/200);
7092 +- b = avenrun[1] + (FIXED_1/200);
7093 +- c = avenrun[2] + (FIXED_1/200);
7094 ++ if (vx_flags(VXF_VIRT_LOAD, 0)) {
7095 ++ struct vx_info *vxi = current->vx_info;
7096 ++
7097 ++ a = vxi->cvirt.load[0] + (FIXED_1/200);
7098 ++ b = vxi->cvirt.load[1] + (FIXED_1/200);
7099 ++ c = vxi->cvirt.load[2] + (FIXED_1/200);
7100 ++
7101 ++ running = atomic_read(&vxi->cvirt.nr_running);
7102 ++ threads = atomic_read(&vxi->cvirt.nr_threads);
7103 ++ } else {
7104 ++ a = avenrun[0] + (FIXED_1/200);
7105 ++ b = avenrun[1] + (FIXED_1/200);
7106 ++ c = avenrun[2] + (FIXED_1/200);
7107 ++
7108 ++ running = nr_running();
7109 ++ threads = nr_threads;
7110 ++ }
7111 + } while (read_seqretry(&xtime_lock, seq));
7112 +
7113 + seq_printf(m, "%d.%02d %d.%02d %d.%02d %ld/%d %d\n",
7114 + LOAD_INT(a), LOAD_FRAC(a),
7115 + LOAD_INT(b), LOAD_FRAC(b),
7116 + LOAD_INT(c), LOAD_FRAC(c),
7117 +- nr_running(), nr_threads,
7118 ++ running, threads,
7119 + task_active_pid_ns(current)->last_pid);
7120 + return 0;
7121 + }
7122 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/meminfo.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/meminfo.c
7123 +--- linux-2.6.29.4/fs/proc/meminfo.c 2009-05-23 23:16:53.000000000 +0200
7124 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/meminfo.c 2009-05-10 23:42:01.000000000 +0200
7125 +@@ -41,7 +41,7 @@ static int meminfo_proc_show(struct seq_
7126 +
7127 + cached = global_page_state(NR_FILE_PAGES) -
7128 + total_swapcache_pages - i.bufferram;
7129 +- if (cached < 0)
7130 ++ if (cached < 0 || vx_flags(VXF_VIRT_MEM, 0))
7131 + cached = 0;
7132 +
7133 + get_vmalloc_info(&vmi);
7134 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/root.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/root.c
7135 +--- linux-2.6.29.4/fs/proc/root.c 2009-03-24 14:22:27.000000000 +0100
7136 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/root.c 2009-03-24 14:48:35.000000000 +0100
7137 +@@ -18,9 +18,14 @@
7138 + #include <linux/bitops.h>
7139 + #include <linux/mount.h>
7140 + #include <linux/pid_namespace.h>
7141 ++#include <linux/vserver/inode.h>
7142 +
7143 + #include "internal.h"
7144 +
7145 ++struct proc_dir_entry *proc_virtual;
7146 ++
7147 ++extern void proc_vx_init(void);
7148 ++
7149 + static int proc_test_super(struct super_block *sb, void *data)
7150 + {
7151 + return sb->s_fs_info == data;
7152 +@@ -136,6 +141,7 @@ void __init proc_root_init(void)
7153 + #endif
7154 + proc_mkdir("bus", NULL);
7155 + proc_sys_init();
7156 ++ proc_vx_init();
7157 + }
7158 +
7159 + static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat
7160 +@@ -203,6 +209,7 @@ struct proc_dir_entry proc_root = {
7161 + .proc_iops = &proc_root_inode_operations,
7162 + .proc_fops = &proc_root_operations,
7163 + .parent = &proc_root,
7164 ++ .vx_flags = IATTR_ADMIN | IATTR_WATCH,
7165 + };
7166 +
7167 + int pid_ns_prepare_proc(struct pid_namespace *ns)
7168 +diff -NurpP --minimal linux-2.6.29.4/fs/proc/uptime.c linux-2.6.29.4-vs2.3.0.36.14/fs/proc/uptime.c
7169 +--- linux-2.6.29.4/fs/proc/uptime.c 2008-12-25 00:26:37.000000000 +0100
7170 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/proc/uptime.c 2009-02-22 22:54:25.000000000 +0100
7171 +@@ -2,6 +2,7 @@
7172 + #include <linux/proc_fs.h>
7173 + #include <linux/sched.h>
7174 + #include <linux/time.h>
7175 ++#include <linux/vserver/cvirt.h>
7176 + #include <asm/cputime.h>
7177 +
7178 + static int proc_calc_metrics(char *page, char **start, off_t off,
7179 +@@ -29,6 +30,10 @@ static int uptime_read_proc(char *page,
7180 + do_posix_clock_monotonic_gettime(&uptime);
7181 + monotonic_to_bootbased(&uptime);
7182 + cputime_to_timespec(idletime, &idle);
7183 ++
7184 ++ if (vx_flags(VXF_VIRT_UPTIME, 0))
7185 ++ vx_vsi_uptime(&uptime, &idle);
7186 ++
7187 + len = sprintf(page, "%lu.%02lu %lu.%02lu\n",
7188 + (unsigned long) uptime.tv_sec,
7189 + (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
7190 +diff -NurpP --minimal linux-2.6.29.4/fs/quota.c linux-2.6.29.4-vs2.3.0.36.14/fs/quota.c
7191 +--- linux-2.6.29.4/fs/quota.c 2009-03-24 14:22:27.000000000 +0100
7192 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/quota.c 2009-03-24 14:48:35.000000000 +0100
7193 +@@ -18,6 +18,7 @@
7194 + #include <linux/capability.h>
7195 + #include <linux/quotaops.h>
7196 + #include <linux/types.h>
7197 ++#include <linux/vs_context.h>
7198 +
7199 + /* Check validity of generic quotactl commands */
7200 + static int generic_quotactl_valid(struct super_block *sb, int type, int cmd, qid_t id)
7201 +@@ -81,11 +82,11 @@ static int generic_quotactl_valid(struct
7202 + if (cmd == Q_GETQUOTA) {
7203 + if (((type == USRQUOTA && current_euid() != id) ||
7204 + (type == GRPQUOTA && !in_egroup_p(id))) &&
7205 +- !capable(CAP_SYS_ADMIN))
7206 ++ !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7207 + return -EPERM;
7208 + }
7209 + else if (cmd != Q_GETFMT && cmd != Q_SYNC && cmd != Q_GETINFO)
7210 +- if (!capable(CAP_SYS_ADMIN))
7211 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7212 + return -EPERM;
7213 +
7214 + return 0;
7215 +@@ -132,10 +133,10 @@ static int xqm_quotactl_valid(struct sup
7216 + if (cmd == Q_XGETQUOTA) {
7217 + if (((type == XQM_USRQUOTA && current_euid() != id) ||
7218 + (type == XQM_GRPQUOTA && !in_egroup_p(id))) &&
7219 +- !capable(CAP_SYS_ADMIN))
7220 ++ !vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7221 + return -EPERM;
7222 + } else if (cmd != Q_XGETQSTAT && cmd != Q_XQUOTASYNC) {
7223 +- if (!capable(CAP_SYS_ADMIN))
7224 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7225 + return -EPERM;
7226 + }
7227 +
7228 +@@ -337,6 +338,46 @@ static int do_quotactl(struct super_bloc
7229 + return 0;
7230 + }
7231 +
7232 ++#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7233 ++
7234 ++#include <linux/vroot.h>
7235 ++#include <linux/major.h>
7236 ++#include <linux/module.h>
7237 ++#include <linux/kallsyms.h>
7238 ++#include <linux/vserver/debug.h>
7239 ++
7240 ++static vroot_grb_func *vroot_get_real_bdev = NULL;
7241 ++
7242 ++static spinlock_t vroot_grb_lock = SPIN_LOCK_UNLOCKED;
7243 ++
7244 ++int register_vroot_grb(vroot_grb_func *func) {
7245 ++ int ret = -EBUSY;
7246 ++
7247 ++ spin_lock(&vroot_grb_lock);
7248 ++ if (!vroot_get_real_bdev) {
7249 ++ vroot_get_real_bdev = func;
7250 ++ ret = 0;
7251 ++ }
7252 ++ spin_unlock(&vroot_grb_lock);
7253 ++ return ret;
7254 ++}
7255 ++EXPORT_SYMBOL(register_vroot_grb);
7256 ++
7257 ++int unregister_vroot_grb(vroot_grb_func *func) {
7258 ++ int ret = -EINVAL;
7259 ++
7260 ++ spin_lock(&vroot_grb_lock);
7261 ++ if (vroot_get_real_bdev) {
7262 ++ vroot_get_real_bdev = NULL;
7263 ++ ret = 0;
7264 ++ }
7265 ++ spin_unlock(&vroot_grb_lock);
7266 ++ return ret;
7267 ++}
7268 ++EXPORT_SYMBOL(unregister_vroot_grb);
7269 ++
7270 ++#endif
7271 ++
7272 + /*
7273 + * look up a superblock on which quota ops will be performed
7274 + * - use the name of a block device to find the superblock thereon
7275 +@@ -354,6 +395,22 @@ static inline struct super_block *quotac
7276 + putname(tmp);
7277 + if (IS_ERR(bdev))
7278 + return ERR_CAST(bdev);
7279 ++#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7280 ++ if (bdev && bdev->bd_inode &&
7281 ++ imajor(bdev->bd_inode) == VROOT_MAJOR) {
7282 ++ struct block_device *bdnew = (void *)-EINVAL;
7283 ++
7284 ++ if (vroot_get_real_bdev)
7285 ++ bdnew = vroot_get_real_bdev(bdev);
7286 ++ else
7287 ++ vxdprintk(VXD_CBIT(misc, 0),
7288 ++ "vroot_get_real_bdev not set");
7289 ++ bdput(bdev);
7290 ++ if (IS_ERR(bdnew))
7291 ++ return ERR_PTR(PTR_ERR(bdnew));
7292 ++ bdev = bdnew;
7293 ++ }
7294 ++#endif
7295 + sb = get_super(bdev);
7296 + bdput(bdev);
7297 + if (!sb)
7298 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/bitmap.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/bitmap.c
7299 +--- linux-2.6.29.4/fs/reiserfs/bitmap.c 2008-12-25 00:26:37.000000000 +0100
7300 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/bitmap.c 2009-02-22 22:54:25.000000000 +0100
7301 +@@ -13,6 +13,7 @@
7302 + #include <linux/reiserfs_fs_sb.h>
7303 + #include <linux/reiserfs_fs_i.h>
7304 + #include <linux/quotaops.h>
7305 ++#include <linux/vs_dlimit.h>
7306 +
7307 + #define PREALLOCATION_SIZE 9
7308 +
7309 +@@ -429,8 +430,10 @@ static void _reiserfs_free_block(struct
7310 + set_sb_free_blocks(rs, sb_free_blocks(rs) + 1);
7311 +
7312 + journal_mark_dirty(th, s, sbh);
7313 +- if (for_unformatted)
7314 ++ if (for_unformatted) {
7315 ++ DLIMIT_FREE_BLOCK(inode, 1);
7316 + DQUOT_FREE_BLOCK_NODIRTY(inode, 1);
7317 ++ }
7318 + }
7319 +
7320 + void reiserfs_free_block(struct reiserfs_transaction_handle *th,
7321 +@@ -1045,6 +1048,7 @@ static inline int blocknrs_and_prealloc_
7322 + b_blocknr_t finish = SB_BLOCK_COUNT(s) - 1;
7323 + int passno = 0;
7324 + int nr_allocated = 0;
7325 ++ int blocks;
7326 +
7327 + determine_prealloc_size(hint);
7328 + if (!hint->formatted_node) {
7329 +@@ -1054,19 +1058,30 @@ static inline int blocknrs_and_prealloc_
7330 + "reiserquota: allocating %d blocks id=%u",
7331 + amount_needed, hint->inode->i_uid);
7332 + #endif
7333 +- quota_ret =
7334 +- DQUOT_ALLOC_BLOCK_NODIRTY(hint->inode, amount_needed);
7335 +- if (quota_ret) /* Quota exceeded? */
7336 ++ quota_ret = DQUOT_ALLOC_BLOCK_NODIRTY(hint->inode,
7337 ++ amount_needed);
7338 ++ if (quota_ret)
7339 + return QUOTA_EXCEEDED;
7340 ++ if (DLIMIT_ALLOC_BLOCK(hint->inode, amount_needed)) {
7341 ++ DQUOT_FREE_BLOCK_NODIRTY(hint->inode,
7342 ++ amount_needed);
7343 ++ return NO_DISK_SPACE;
7344 ++ }
7345 ++
7346 + if (hint->preallocate && hint->prealloc_size) {
7347 + #ifdef REISERQUOTA_DEBUG
7348 + reiserfs_debug(s, REISERFS_DEBUG_CODE,
7349 + "reiserquota: allocating (prealloc) %d blocks id=%u",
7350 + hint->prealloc_size, hint->inode->i_uid);
7351 + #endif
7352 +- quota_ret =
7353 +- DQUOT_PREALLOC_BLOCK_NODIRTY(hint->inode,
7354 +- hint->prealloc_size);
7355 ++ quota_ret = DQUOT_PREALLOC_BLOCK_NODIRTY(hint->inode,
7356 ++ hint->prealloc_size);
7357 ++ if (!quota_ret &&
7358 ++ DLIMIT_ALLOC_BLOCK(hint->inode, hint->prealloc_size)) {
7359 ++ DQUOT_FREE_BLOCK_NODIRTY(hint->inode,
7360 ++ hint->prealloc_size);
7361 ++ quota_ret = 1;
7362 ++ }
7363 + if (quota_ret)
7364 + hint->preallocate = hint->prealloc_size = 0;
7365 + }
7366 +@@ -1098,7 +1113,10 @@ static inline int blocknrs_and_prealloc_
7367 + nr_allocated,
7368 + hint->inode->i_uid);
7369 + #endif
7370 +- DQUOT_FREE_BLOCK_NODIRTY(hint->inode, amount_needed + hint->prealloc_size - nr_allocated); /* Free not allocated blocks */
7371 ++ /* Free not allocated blocks */
7372 ++ blocks = amount_needed + hint->prealloc_size - nr_allocated;
7373 ++ DLIMIT_FREE_BLOCK(hint->inode, blocks);
7374 ++ DQUOT_FREE_BLOCK_NODIRTY(hint->inode, blocks);
7375 + }
7376 + while (nr_allocated--)
7377 + reiserfs_free_block(hint->th, hint->inode,
7378 +@@ -1129,10 +1147,10 @@ static inline int blocknrs_and_prealloc_
7379 + REISERFS_I(hint->inode)->i_prealloc_count,
7380 + hint->inode->i_uid);
7381 + #endif
7382 +- DQUOT_FREE_BLOCK_NODIRTY(hint->inode, amount_needed +
7383 +- hint->prealloc_size - nr_allocated -
7384 +- REISERFS_I(hint->inode)->
7385 +- i_prealloc_count);
7386 ++ blocks = amount_needed + hint->prealloc_size - nr_allocated -
7387 ++ REISERFS_I(hint->inode)->i_prealloc_count;
7388 ++ DLIMIT_FREE_BLOCK(hint->inode, blocks);
7389 ++ DQUOT_FREE_BLOCK_NODIRTY(hint->inode, blocks);
7390 + }
7391 +
7392 + return CARRY_ON;
7393 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/file.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/file.c
7394 +--- linux-2.6.29.4/fs/reiserfs/file.c 2008-12-25 00:26:37.000000000 +0100
7395 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/file.c 2009-02-22 22:54:25.000000000 +0100
7396 +@@ -307,4 +307,5 @@ const struct inode_operations reiserfs_f
7397 + .listxattr = reiserfs_listxattr,
7398 + .removexattr = reiserfs_removexattr,
7399 + .permission = reiserfs_permission,
7400 ++ .sync_flags = reiserfs_sync_flags,
7401 + };
7402 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/inode.c
7403 +--- linux-2.6.29.4/fs/reiserfs/inode.c 2009-03-24 14:22:27.000000000 +0100
7404 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/inode.c 2009-03-24 14:48:35.000000000 +0100
7405 +@@ -18,6 +18,8 @@
7406 + #include <linux/writeback.h>
7407 + #include <linux/quotaops.h>
7408 + #include <linux/swap.h>
7409 ++#include <linux/vs_dlimit.h>
7410 ++#include <linux/vs_tag.h>
7411 +
7412 + int reiserfs_commit_write(struct file *f, struct page *page,
7413 + unsigned from, unsigned to);
7414 +@@ -54,6 +56,7 @@ void reiserfs_delete_inode(struct inode
7415 + * stat data deletion */
7416 + if (!err)
7417 + DQUOT_FREE_INODE(inode);
7418 ++ DLIMIT_FREE_INODE(inode);
7419 +
7420 + if (journal_end(&th, inode->i_sb, jbegin_count))
7421 + goto out;
7422 +@@ -1116,6 +1119,8 @@ static void init_inode(struct inode *ino
7423 + struct buffer_head *bh;
7424 + struct item_head *ih;
7425 + __u32 rdev;
7426 ++ uid_t uid;
7427 ++ gid_t gid;
7428 + //int version = ITEM_VERSION_1;
7429 +
7430 + bh = PATH_PLAST_BUFFER(path);
7431 +@@ -1139,12 +1144,13 @@ static void init_inode(struct inode *ino
7432 + (struct stat_data_v1 *)B_I_PITEM(bh, ih);
7433 + unsigned long blocks;
7434 +
7435 ++ uid = sd_v1_uid(sd);
7436 ++ gid = sd_v1_gid(sd);
7437 ++
7438 + set_inode_item_key_version(inode, KEY_FORMAT_3_5);
7439 + set_inode_sd_version(inode, STAT_DATA_V1);
7440 + inode->i_mode = sd_v1_mode(sd);
7441 + inode->i_nlink = sd_v1_nlink(sd);
7442 +- inode->i_uid = sd_v1_uid(sd);
7443 +- inode->i_gid = sd_v1_gid(sd);
7444 + inode->i_size = sd_v1_size(sd);
7445 + inode->i_atime.tv_sec = sd_v1_atime(sd);
7446 + inode->i_mtime.tv_sec = sd_v1_mtime(sd);
7447 +@@ -1186,11 +1192,12 @@ static void init_inode(struct inode *ino
7448 + // (directories and symlinks)
7449 + struct stat_data *sd = (struct stat_data *)B_I_PITEM(bh, ih);
7450 +
7451 ++ uid = sd_v2_uid(sd);
7452 ++ gid = sd_v2_gid(sd);
7453 ++
7454 + inode->i_mode = sd_v2_mode(sd);
7455 + inode->i_nlink = sd_v2_nlink(sd);
7456 +- inode->i_uid = sd_v2_uid(sd);
7457 + inode->i_size = sd_v2_size(sd);
7458 +- inode->i_gid = sd_v2_gid(sd);
7459 + inode->i_mtime.tv_sec = sd_v2_mtime(sd);
7460 + inode->i_atime.tv_sec = sd_v2_atime(sd);
7461 + inode->i_ctime.tv_sec = sd_v2_ctime(sd);
7462 +@@ -1220,6 +1227,10 @@ static void init_inode(struct inode *ino
7463 + sd_attrs_to_i_attrs(sd_v2_attrs(sd), inode);
7464 + }
7465 +
7466 ++ inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
7467 ++ inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
7468 ++ inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
7469 ++
7470 + pathrelse(path);
7471 + if (S_ISREG(inode->i_mode)) {
7472 + inode->i_op = &reiserfs_file_inode_operations;
7473 +@@ -1242,13 +1253,15 @@ static void init_inode(struct inode *ino
7474 + static void inode2sd(void *sd, struct inode *inode, loff_t size)
7475 + {
7476 + struct stat_data *sd_v2 = (struct stat_data *)sd;
7477 ++ uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
7478 ++ gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
7479 + __u16 flags;
7480 +
7481 ++ set_sd_v2_uid(sd_v2, uid);
7482 ++ set_sd_v2_gid(sd_v2, gid);
7483 + set_sd_v2_mode(sd_v2, inode->i_mode);
7484 + set_sd_v2_nlink(sd_v2, inode->i_nlink);
7485 +- set_sd_v2_uid(sd_v2, inode->i_uid);
7486 + set_sd_v2_size(sd_v2, size);
7487 +- set_sd_v2_gid(sd_v2, inode->i_gid);
7488 + set_sd_v2_mtime(sd_v2, inode->i_mtime.tv_sec);
7489 + set_sd_v2_atime(sd_v2, inode->i_atime.tv_sec);
7490 + set_sd_v2_ctime(sd_v2, inode->i_ctime.tv_sec);
7491 +@@ -1763,6 +1776,10 @@ int reiserfs_new_inode(struct reiserfs_t
7492 +
7493 + BUG_ON(!th->t_trans_id);
7494 +
7495 ++ if (DLIMIT_ALLOC_INODE(inode)) {
7496 ++ err = -ENOSPC;
7497 ++ goto out_bad_dlimit;
7498 ++ }
7499 + if (DQUOT_ALLOC_INODE(inode)) {
7500 + err = -EDQUOT;
7501 + goto out_end_trans;
7502 +@@ -1950,6 +1967,9 @@ int reiserfs_new_inode(struct reiserfs_t
7503 + DQUOT_FREE_INODE(inode);
7504 +
7505 + out_end_trans:
7506 ++ DLIMIT_FREE_INODE(inode);
7507 ++
7508 ++ out_bad_dlimit:
7509 + journal_end(th, th->t_super, th->t_blocks_allocated);
7510 + /* Drop can be outside and it needs more credits so it's better to have it outside */
7511 + DQUOT_DROP(inode);
7512 +@@ -2837,14 +2857,19 @@ int reiserfs_commit_write(struct file *f
7513 + void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode)
7514 + {
7515 + if (reiserfs_attrs(inode->i_sb)) {
7516 +- if (sd_attrs & REISERFS_SYNC_FL)
7517 +- inode->i_flags |= S_SYNC;
7518 +- else
7519 +- inode->i_flags &= ~S_SYNC;
7520 + if (sd_attrs & REISERFS_IMMUTABLE_FL)
7521 + inode->i_flags |= S_IMMUTABLE;
7522 + else
7523 + inode->i_flags &= ~S_IMMUTABLE;
7524 ++ if (sd_attrs & REISERFS_IXUNLINK_FL)
7525 ++ inode->i_flags |= S_IXUNLINK;
7526 ++ else
7527 ++ inode->i_flags &= ~S_IXUNLINK;
7528 ++
7529 ++ if (sd_attrs & REISERFS_SYNC_FL)
7530 ++ inode->i_flags |= S_SYNC;
7531 ++ else
7532 ++ inode->i_flags &= ~S_SYNC;
7533 + if (sd_attrs & REISERFS_APPEND_FL)
7534 + inode->i_flags |= S_APPEND;
7535 + else
7536 +@@ -2857,6 +2882,15 @@ void sd_attrs_to_i_attrs(__u16 sd_attrs,
7537 + REISERFS_I(inode)->i_flags |= i_nopack_mask;
7538 + else
7539 + REISERFS_I(inode)->i_flags &= ~i_nopack_mask;
7540 ++
7541 ++ if (sd_attrs & REISERFS_BARRIER_FL)
7542 ++ inode->i_vflags |= V_BARRIER;
7543 ++ else
7544 ++ inode->i_vflags &= ~V_BARRIER;
7545 ++ if (sd_attrs & REISERFS_COW_FL)
7546 ++ inode->i_vflags |= V_COW;
7547 ++ else
7548 ++ inode->i_vflags &= ~V_COW;
7549 + }
7550 + }
7551 +
7552 +@@ -2867,6 +2901,11 @@ void i_attrs_to_sd_attrs(struct inode *i
7553 + *sd_attrs |= REISERFS_IMMUTABLE_FL;
7554 + else
7555 + *sd_attrs &= ~REISERFS_IMMUTABLE_FL;
7556 ++ if (inode->i_flags & S_IXUNLINK)
7557 ++ *sd_attrs |= REISERFS_IXUNLINK_FL;
7558 ++ else
7559 ++ *sd_attrs &= ~REISERFS_IXUNLINK_FL;
7560 ++
7561 + if (inode->i_flags & S_SYNC)
7562 + *sd_attrs |= REISERFS_SYNC_FL;
7563 + else
7564 +@@ -2879,6 +2918,15 @@ void i_attrs_to_sd_attrs(struct inode *i
7565 + *sd_attrs |= REISERFS_NOTAIL_FL;
7566 + else
7567 + *sd_attrs &= ~REISERFS_NOTAIL_FL;
7568 ++
7569 ++ if (inode->i_vflags & V_BARRIER)
7570 ++ *sd_attrs |= REISERFS_BARRIER_FL;
7571 ++ else
7572 ++ *sd_attrs &= ~REISERFS_BARRIER_FL;
7573 ++ if (inode->i_vflags & V_COW)
7574 ++ *sd_attrs |= REISERFS_COW_FL;
7575 ++ else
7576 ++ *sd_attrs &= ~REISERFS_COW_FL;
7577 + }
7578 + }
7579 +
7580 +@@ -3046,6 +3094,22 @@ static ssize_t reiserfs_direct_IO(int rw
7581 + reiserfs_get_blocks_direct_io, NULL);
7582 + }
7583 +
7584 ++int reiserfs_sync_flags(struct inode *inode)
7585 ++{
7586 ++ u16 oldflags, newflags;
7587 ++
7588 ++ oldflags = REISERFS_I(inode)->i_attrs;
7589 ++ newflags = oldflags;
7590 ++ i_attrs_to_sd_attrs(inode, &newflags);
7591 ++
7592 ++ if (oldflags ^ newflags) {
7593 ++ REISERFS_I(inode)->i_attrs = newflags;
7594 ++ inode->i_ctime = CURRENT_TIME_SEC;
7595 ++ mark_inode_dirty(inode);
7596 ++ }
7597 ++ return 0;
7598 ++}
7599 ++
7600 + int reiserfs_setattr(struct dentry *dentry, struct iattr *attr)
7601 + {
7602 + struct inode *inode = dentry->d_inode;
7603 +@@ -3099,9 +3163,11 @@ int reiserfs_setattr(struct dentry *dent
7604 + }
7605 +
7606 + error = inode_change_ok(inode, attr);
7607 ++
7608 + if (!error) {
7609 + if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
7610 +- (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
7611 ++ (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
7612 ++ (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
7613 + error = reiserfs_chown_xattrs(inode, attr);
7614 +
7615 + if (!error) {
7616 +@@ -3131,6 +3197,9 @@ int reiserfs_setattr(struct dentry *dent
7617 + inode->i_uid = attr->ia_uid;
7618 + if (attr->ia_valid & ATTR_GID)
7619 + inode->i_gid = attr->ia_gid;
7620 ++ if ((attr->ia_valid & ATTR_TAG) &&
7621 ++ IS_TAGGED(inode))
7622 ++ inode->i_tag = attr->ia_tag;
7623 + mark_inode_dirty(inode);
7624 + error =
7625 + journal_end(&th, inode->i_sb, jbegin_count);
7626 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/ioctl.c
7627 +--- linux-2.6.29.4/fs/reiserfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
7628 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/ioctl.c 2009-02-22 22:54:25.000000000 +0100
7629 +@@ -7,6 +7,7 @@
7630 + #include <linux/mount.h>
7631 + #include <linux/reiserfs_fs.h>
7632 + #include <linux/time.h>
7633 ++#include <linux/mount.h>
7634 + #include <asm/uaccess.h>
7635 + #include <linux/pagemap.h>
7636 + #include <linux/smp_lock.h>
7637 +@@ -23,7 +24,7 @@
7638 + int reiserfs_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
7639 + unsigned long arg)
7640 + {
7641 +- unsigned int flags;
7642 ++ unsigned int flags, oldflags;
7643 + int err = 0;
7644 +
7645 + switch (cmd) {
7646 +@@ -43,6 +44,7 @@ int reiserfs_ioctl(struct inode *inode,
7647 +
7648 + flags = REISERFS_I(inode)->i_attrs;
7649 + i_attrs_to_sd_attrs(inode, (__u16 *) & flags);
7650 ++ flags &= REISERFS_FL_USER_VISIBLE;
7651 + return put_user(flags, (int __user *)arg);
7652 + case REISERFS_IOC_SETFLAGS:{
7653 + if (!reiserfs_attrs(inode->i_sb))
7654 +@@ -60,6 +62,10 @@ int reiserfs_ioctl(struct inode *inode,
7655 + err = -EFAULT;
7656 + goto setflags_out;
7657 + }
7658 ++ if (IS_BARRIER(inode)) {
7659 ++ vxwprintk_task(1, "messing with the barrier.");
7660 ++ return -EACCES;
7661 ++ }
7662 + /*
7663 + * Is it quota file? Do not allow user to mess with it
7664 + */
7665 +@@ -84,6 +90,10 @@ int reiserfs_ioctl(struct inode *inode,
7666 + goto setflags_out;
7667 + }
7668 + }
7669 ++
7670 ++ oldflags = REISERFS_I(inode)->i_attrs;
7671 ++ flags = flags & REISERFS_FL_USER_MODIFIABLE;
7672 ++ flags |= oldflags & ~REISERFS_FL_USER_MODIFIABLE;
7673 + sd_attrs_to_i_attrs(flags, inode);
7674 + REISERFS_I(inode)->i_attrs = flags;
7675 + inode->i_ctime = CURRENT_TIME_SEC;
7676 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/namei.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/namei.c
7677 +--- linux-2.6.29.4/fs/reiserfs/namei.c 2009-03-24 14:22:27.000000000 +0100
7678 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/namei.c 2009-03-24 14:48:35.000000000 +0100
7679 +@@ -17,6 +17,7 @@
7680 + #include <linux/reiserfs_acl.h>
7681 + #include <linux/reiserfs_xattr.h>
7682 + #include <linux/quotaops.h>
7683 ++#include <linux/vs_tag.h>
7684 +
7685 + #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; }
7686 + #define DEC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) drop_nlink(i);
7687 +@@ -360,6 +361,7 @@ static struct dentry *reiserfs_lookup(st
7688 + reiserfs_write_unlock(dir->i_sb);
7689 + return ERR_PTR(-EACCES);
7690 + }
7691 ++ dx_propagate_tag(nd, inode);
7692 +
7693 + /* Propogate the priv_object flag so we know we're in the priv tree */
7694 + if (is_reiserfs_priv_object(dir))
7695 +@@ -586,6 +588,7 @@ static int new_inode_init(struct inode *
7696 + } else {
7697 + inode->i_gid = current_fsgid();
7698 + }
7699 ++ inode->i_tag = dx_current_fstag(inode->i_sb);
7700 + DQUOT_INIT(inode);
7701 + return 0;
7702 + }
7703 +@@ -1540,6 +1543,7 @@ const struct inode_operations reiserfs_d
7704 + .listxattr = reiserfs_listxattr,
7705 + .removexattr = reiserfs_removexattr,
7706 + .permission = reiserfs_permission,
7707 ++ .sync_flags = reiserfs_sync_flags,
7708 + };
7709 +
7710 + /*
7711 +@@ -1556,6 +1560,7 @@ const struct inode_operations reiserfs_s
7712 + .listxattr = reiserfs_listxattr,
7713 + .removexattr = reiserfs_removexattr,
7714 + .permission = reiserfs_permission,
7715 ++ .sync_flags = reiserfs_sync_flags,
7716 +
7717 + };
7718 +
7719 +@@ -1569,5 +1574,6 @@ const struct inode_operations reiserfs_s
7720 + .listxattr = reiserfs_listxattr,
7721 + .removexattr = reiserfs_removexattr,
7722 + .permission = reiserfs_permission,
7723 ++ .sync_flags = reiserfs_sync_flags,
7724 +
7725 + };
7726 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/stree.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/stree.c
7727 +--- linux-2.6.29.4/fs/reiserfs/stree.c 2008-12-25 00:26:37.000000000 +0100
7728 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/stree.c 2009-02-22 22:54:25.000000000 +0100
7729 +@@ -55,6 +55,7 @@
7730 + #include <linux/reiserfs_fs.h>
7731 + #include <linux/buffer_head.h>
7732 + #include <linux/quotaops.h>
7733 ++#include <linux/vs_dlimit.h>
7734 +
7735 + /* Does the buffer contain a disk block which is in the tree. */
7736 + inline int B_IS_IN_TREE(const struct buffer_head *p_s_bh)
7737 +@@ -1297,6 +1298,7 @@ int reiserfs_delete_item(struct reiserfs
7738 + "reiserquota delete_item(): freeing %u, id=%u type=%c",
7739 + quota_cut_bytes, p_s_inode->i_uid, head2type(&s_ih));
7740 + #endif
7741 ++ DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
7742 + DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
7743 +
7744 + /* Return deleted body length */
7745 +@@ -1385,6 +1387,7 @@ void reiserfs_delete_solid_item(struct r
7746 + #endif
7747 + DQUOT_FREE_SPACE_NODIRTY(inode,
7748 + quota_cut_bytes);
7749 ++ DLIMIT_FREE_SPACE(inode, quota_cut_bytes);
7750 + }
7751 + break;
7752 + }
7753 +@@ -1734,6 +1737,7 @@ int reiserfs_cut_from_item(struct reiser
7754 + "reiserquota cut_from_item(): freeing %u id=%u type=%c",
7755 + quota_cut_bytes, p_s_inode->i_uid, '?');
7756 + #endif
7757 ++ DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
7758 + DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
7759 + return n_ret_value;
7760 + }
7761 +@@ -1975,6 +1979,11 @@ int reiserfs_paste_into_item(struct reis
7762 + pathrelse(p_s_search_path);
7763 + return -EDQUOT;
7764 + }
7765 ++ if (DLIMIT_ALLOC_SPACE(inode, n_pasted_size)) {
7766 ++ DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
7767 ++ pathrelse(p_s_search_path);
7768 ++ return -ENOSPC;
7769 ++ }
7770 + init_tb_struct(th, &s_paste_balance, th->t_super, p_s_search_path,
7771 + n_pasted_size);
7772 + #ifdef DISPLACE_NEW_PACKING_LOCALITIES
7773 +@@ -2027,6 +2036,7 @@ int reiserfs_paste_into_item(struct reis
7774 + n_pasted_size, inode->i_uid,
7775 + key2type(&(p_s_key->on_disk_key)));
7776 + #endif
7777 ++ DLIMIT_FREE_SPACE(inode, n_pasted_size);
7778 + DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
7779 + return retval;
7780 + }
7781 +@@ -2064,6 +2074,11 @@ int reiserfs_insert_item(struct reiserfs
7782 + pathrelse(p_s_path);
7783 + return -EDQUOT;
7784 + }
7785 ++ if (DLIMIT_ALLOC_SPACE(inode, quota_bytes)) {
7786 ++ DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);
7787 ++ pathrelse(p_s_path);
7788 ++ return -ENOSPC;
7789 ++ }
7790 + }
7791 + init_tb_struct(th, &s_ins_balance, th->t_super, p_s_path,
7792 + IH_SIZE + ih_item_len(p_s_ih));
7793 +@@ -2111,7 +2126,9 @@ int reiserfs_insert_item(struct reiserfs
7794 + "reiserquota insert_item(): freeing %u id=%u type=%c",
7795 + quota_bytes, inode->i_uid, head2type(p_s_ih));
7796 + #endif
7797 +- if (inode)
7798 ++ if (inode) {
7799 ++ DLIMIT_FREE_SPACE(inode, quota_bytes);
7800 + DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);
7801 ++ }
7802 + return retval;
7803 + }
7804 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/super.c
7805 +--- linux-2.6.29.4/fs/reiserfs/super.c 2009-03-24 14:22:27.000000000 +0100
7806 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/super.c 2009-03-24 14:48:35.000000000 +0100
7807 +@@ -902,6 +902,14 @@ static int reiserfs_parse_options(struct
7808 + {"user_xattr",.setmask = 1 << REISERFS_UNSUPPORTED_OPT},
7809 + {"nouser_xattr",.clrmask = 1 << REISERFS_UNSUPPORTED_OPT},
7810 + #endif
7811 ++#ifndef CONFIG_TAGGING_NONE
7812 ++ {"tagxid",.setmask = 1 << REISERFS_TAGGED},
7813 ++ {"tag",.setmask = 1 << REISERFS_TAGGED},
7814 ++ {"notag",.clrmask = 1 << REISERFS_TAGGED},
7815 ++#endif
7816 ++#ifdef CONFIG_PROPAGATE
7817 ++ {"tag",.arg_required = 'T',.values = NULL},
7818 ++#endif
7819 + #ifdef CONFIG_REISERFS_FS_POSIX_ACL
7820 + {"acl",.setmask = 1 << REISERFS_POSIXACL},
7821 + {"noacl",.clrmask = 1 << REISERFS_POSIXACL},
7822 +@@ -1195,6 +1203,12 @@ static int reiserfs_remount(struct super
7823 + handle_quota_files(s, qf_names, &qfmt);
7824 + #endif
7825 +
7826 ++ if ((mount_options & (1 << REISERFS_TAGGED)) &&
7827 ++ !(s->s_flags & MS_TAGGED)) {
7828 ++ reiserfs_warning(s, "reiserfs: tagging not permitted on remount.");
7829 ++ return -EINVAL;
7830 ++ }
7831 ++
7832 + handle_attrs(s);
7833 +
7834 + /* Add options that are safe here */
7835 +@@ -1659,6 +1673,10 @@ static int reiserfs_fill_super(struct su
7836 + goto error;
7837 + }
7838 +
7839 ++ /* map mount option tagxid */
7840 ++ if (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TAGGED))
7841 ++ s->s_flags |= MS_TAGGED;
7842 ++
7843 + rs = SB_DISK_SUPER_BLOCK(s);
7844 + /* Let's do basic sanity check to verify that underlying device is not
7845 + smaller than the filesystem. If the check fails then abort and scream,
7846 +diff -NurpP --minimal linux-2.6.29.4/fs/reiserfs/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/xattr.c
7847 +--- linux-2.6.29.4/fs/reiserfs/xattr.c 2008-12-25 00:26:37.000000000 +0100
7848 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/reiserfs/xattr.c 2009-02-22 22:54:25.000000000 +0100
7849 +@@ -35,6 +35,7 @@
7850 + #include <linux/namei.h>
7851 + #include <linux/errno.h>
7852 + #include <linux/fs.h>
7853 ++#include <linux/mount.h>
7854 + #include <linux/file.h>
7855 + #include <linux/pagemap.h>
7856 + #include <linux/xattr.h>
7857 +diff -NurpP --minimal linux-2.6.29.4/fs/stat.c linux-2.6.29.4-vs2.3.0.36.14/fs/stat.c
7858 +--- linux-2.6.29.4/fs/stat.c 2009-03-24 14:22:36.000000000 +0100
7859 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/stat.c 2009-03-24 14:48:35.000000000 +0100
7860 +@@ -26,6 +26,7 @@ void generic_fillattr(struct inode *inod
7861 + stat->nlink = inode->i_nlink;
7862 + stat->uid = inode->i_uid;
7863 + stat->gid = inode->i_gid;
7864 ++ stat->tag = inode->i_tag;
7865 + stat->rdev = inode->i_rdev;
7866 + stat->atime = inode->i_atime;
7867 + stat->mtime = inode->i_mtime;
7868 +diff -NurpP --minimal linux-2.6.29.4/fs/super.c linux-2.6.29.4-vs2.3.0.36.14/fs/super.c
7869 +--- linux-2.6.29.4/fs/super.c 2009-03-24 14:22:36.000000000 +0100
7870 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/super.c 2009-03-24 16:32:06.000000000 +0100
7871 +@@ -39,6 +39,9 @@
7872 + #include <linux/mutex.h>
7873 + #include <linux/file.h>
7874 + #include <linux/async.h>
7875 ++#include <linux/devpts_fs.h>
7876 ++#include <linux/proc_fs.h>
7877 ++#include <linux/vs_context.h>
7878 + #include <asm/uaccess.h>
7879 + #include "internal.h"
7880 +
7881 +@@ -918,12 +921,18 @@ struct vfsmount *
7882 + vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
7883 + {
7884 + struct vfsmount *mnt;
7885 ++ struct super_block *sb;
7886 + char *secdata = NULL;
7887 + int error;
7888 +
7889 + if (!type)
7890 + return ERR_PTR(-ENODEV);
7891 +
7892 ++ error = -EPERM;
7893 ++ if ((type->fs_flags & FS_BINARY_MOUNTDATA) &&
7894 ++ !vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT))
7895 ++ goto out;
7896 ++
7897 + error = -ENOMEM;
7898 + mnt = alloc_vfsmnt(name);
7899 + if (!mnt)
7900 +@@ -942,9 +951,17 @@ vfs_kern_mount(struct file_system_type *
7901 + error = type->get_sb(type, flags, name, data, mnt);
7902 + if (error < 0)
7903 + goto out_free_secdata;
7904 +- BUG_ON(!mnt->mnt_sb);
7905 +
7906 +- error = security_sb_kern_mount(mnt->mnt_sb, flags, secdata);
7907 ++ sb = mnt->mnt_sb;
7908 ++ BUG_ON(!sb);
7909 ++
7910 ++ error = -EPERM;
7911 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT) && !sb->s_bdev &&
7912 ++ (sb->s_magic != PROC_SUPER_MAGIC) &&
7913 ++ (sb->s_magic != DEVPTS_SUPER_MAGIC))
7914 ++ goto out_sb;
7915 ++
7916 ++ error = security_sb_kern_mount(sb, flags, secdata);
7917 + if (error)
7918 + goto out_sb;
7919 +
7920 +diff -NurpP --minimal linux-2.6.29.4/fs/sysfs/mount.c linux-2.6.29.4-vs2.3.0.36.14/fs/sysfs/mount.c
7921 +--- linux-2.6.29.4/fs/sysfs/mount.c 2008-12-25 00:26:37.000000000 +0100
7922 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/sysfs/mount.c 2009-02-22 22:54:25.000000000 +0100
7923 +@@ -20,8 +20,6 @@
7924 +
7925 + #include "sysfs.h"
7926 +
7927 +-/* Random magic number */
7928 +-#define SYSFS_MAGIC 0x62656572
7929 +
7930 + static struct vfsmount *sysfs_mount;
7931 + struct super_block * sysfs_sb = NULL;
7932 +@@ -47,7 +45,7 @@ static int sysfs_fill_super(struct super
7933 +
7934 + sb->s_blocksize = PAGE_CACHE_SIZE;
7935 + sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
7936 +- sb->s_magic = SYSFS_MAGIC;
7937 ++ sb->s_magic = SYSFS_SUPER_MAGIC;
7938 + sb->s_op = &sysfs_ops;
7939 + sb->s_time_gran = 1;
7940 + sysfs_sb = sb;
7941 +diff -NurpP --minimal linux-2.6.29.4/fs/utimes.c linux-2.6.29.4-vs2.3.0.36.14/fs/utimes.c
7942 +--- linux-2.6.29.4/fs/utimes.c 2009-03-24 14:22:37.000000000 +0100
7943 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/utimes.c 2009-02-22 22:54:25.000000000 +0100
7944 +@@ -8,6 +8,8 @@
7945 + #include <linux/stat.h>
7946 + #include <linux/utime.h>
7947 + #include <linux/syscalls.h>
7948 ++#include <linux/mount.h>
7949 ++#include <linux/vs_cowbl.h>
7950 + #include <asm/uaccess.h>
7951 + #include <asm/unistd.h>
7952 +
7953 +diff -NurpP --minimal linux-2.6.29.4/fs/xattr.c linux-2.6.29.4-vs2.3.0.36.14/fs/xattr.c
7954 +--- linux-2.6.29.4/fs/xattr.c 2009-03-24 14:22:37.000000000 +0100
7955 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xattr.c 2009-03-24 14:48:35.000000000 +0100
7956 +@@ -18,6 +18,7 @@
7957 + #include <linux/module.h>
7958 + #include <linux/fsnotify.h>
7959 + #include <linux/audit.h>
7960 ++#include <linux/mount.h>
7961 + #include <asm/uaccess.h>
7962 +
7963 +
7964 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/linux-2.6/xfs_ioctl.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_ioctl.c
7965 +--- linux-2.6.29.4/fs/xfs/linux-2.6/xfs_ioctl.c 2009-03-24 14:22:37.000000000 +0100
7966 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_ioctl.c 2009-03-24 14:48:35.000000000 +0100
7967 +@@ -769,6 +769,10 @@ xfs_merge_ioc_xflags(
7968 + xflags |= XFS_XFLAG_IMMUTABLE;
7969 + else
7970 + xflags &= ~XFS_XFLAG_IMMUTABLE;
7971 ++ if (flags & FS_IXUNLINK_FL)
7972 ++ xflags |= XFS_XFLAG_IXUNLINK;
7973 ++ else
7974 ++ xflags &= ~XFS_XFLAG_IXUNLINK;
7975 + if (flags & FS_APPEND_FL)
7976 + xflags |= XFS_XFLAG_APPEND;
7977 + else
7978 +@@ -797,6 +801,8 @@ xfs_di2lxflags(
7979 +
7980 + if (di_flags & XFS_DIFLAG_IMMUTABLE)
7981 + flags |= FS_IMMUTABLE_FL;
7982 ++ if (di_flags & XFS_DIFLAG_IXUNLINK)
7983 ++ flags |= FS_IXUNLINK_FL;
7984 + if (di_flags & XFS_DIFLAG_APPEND)
7985 + flags |= FS_APPEND_FL;
7986 + if (di_flags & XFS_DIFLAG_SYNC)
7987 +@@ -855,6 +861,8 @@ xfs_set_diflags(
7988 + di_flags = (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC);
7989 + if (xflags & XFS_XFLAG_IMMUTABLE)
7990 + di_flags |= XFS_DIFLAG_IMMUTABLE;
7991 ++ if (xflags & XFS_XFLAG_IXUNLINK)
7992 ++ di_flags |= XFS_DIFLAG_IXUNLINK;
7993 + if (xflags & XFS_XFLAG_APPEND)
7994 + di_flags |= XFS_DIFLAG_APPEND;
7995 + if (xflags & XFS_XFLAG_SYNC)
7996 +@@ -897,6 +905,10 @@ xfs_diflags_to_linux(
7997 + inode->i_flags |= S_IMMUTABLE;
7998 + else
7999 + inode->i_flags &= ~S_IMMUTABLE;
8000 ++ if (xflags & XFS_XFLAG_IXUNLINK)
8001 ++ inode->i_flags |= S_IXUNLINK;
8002 ++ else
8003 ++ inode->i_flags &= ~S_IXUNLINK;
8004 + if (xflags & XFS_XFLAG_APPEND)
8005 + inode->i_flags |= S_APPEND;
8006 + else
8007 +@@ -1371,10 +1383,18 @@ xfs_file_ioctl(
8008 + case XFS_IOC_FSGETXATTRA:
8009 + return xfs_ioc_fsgetxattr(ip, 1, arg);
8010 + case XFS_IOC_FSSETXATTR:
8011 ++ if (IS_BARRIER(inode)) {
8012 ++ vxwprintk_task(1, "messing with the barrier.");
8013 ++ return -XFS_ERROR(EACCES);
8014 ++ }
8015 + return xfs_ioc_fssetxattr(ip, filp, arg);
8016 + case XFS_IOC_GETXFLAGS:
8017 + return xfs_ioc_getxflags(ip, arg);
8018 + case XFS_IOC_SETXFLAGS:
8019 ++ if (IS_BARRIER(inode)) {
8020 ++ vxwprintk_task(1, "messing with the barrier.");
8021 ++ return -XFS_ERROR(EACCES);
8022 ++ }
8023 + return xfs_ioc_setxflags(ip, filp, arg);
8024 +
8025 + case XFS_IOC_FSSETDM: {
8026 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/linux-2.6/xfs_iops.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_iops.c
8027 +--- linux-2.6.29.4/fs/xfs/linux-2.6/xfs_iops.c 2009-03-24 14:22:37.000000000 +0100
8028 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_iops.c 2009-03-24 16:33:10.000000000 +0100
8029 +@@ -54,6 +54,7 @@
8030 + #include <linux/security.h>
8031 + #include <linux/falloc.h>
8032 + #include <linux/fiemap.h>
8033 ++#include <linux/vs_tag.h>
8034 +
8035 + /*
8036 + * Bring the atime in the XFS inode uptodate.
8037 +@@ -552,6 +553,7 @@ xfs_vn_getattr(
8038 + stat->nlink = ip->i_d.di_nlink;
8039 + stat->uid = ip->i_d.di_uid;
8040 + stat->gid = ip->i_d.di_gid;
8041 ++ stat->tag = ip->i_d.di_tag;
8042 + stat->ino = ip->i_ino;
8043 + #if XFS_BIG_INUMS
8044 + stat->ino += mp->m_inoadd;
8045 +@@ -591,6 +593,12 @@ xfs_vn_getattr(
8046 + }
8047 +
8048 + STATIC int
8049 ++xfs_vn_sync_xflags(struct inode *inode)
8050 ++{
8051 ++ return -xfs_sync_xflags(XFS_I(inode));
8052 ++}
8053 ++
8054 ++STATIC int
8055 + xfs_vn_setattr(
8056 + struct dentry *dentry,
8057 + struct iattr *iattr)
8058 +@@ -748,6 +756,7 @@ static const struct inode_operations xfs
8059 + .listxattr = xfs_vn_listxattr,
8060 + .fallocate = xfs_vn_fallocate,
8061 + .fiemap = xfs_vn_fiemap,
8062 ++ .sync_flags = xfs_vn_sync_xflags,
8063 + };
8064 +
8065 + static const struct inode_operations xfs_dir_inode_operations = {
8066 +@@ -773,6 +782,7 @@ static const struct inode_operations xfs
8067 + .getxattr = generic_getxattr,
8068 + .removexattr = generic_removexattr,
8069 + .listxattr = xfs_vn_listxattr,
8070 ++ .sync_flags = xfs_vn_sync_xflags,
8071 + };
8072 +
8073 + static const struct inode_operations xfs_dir_ci_inode_operations = {
8074 +@@ -822,6 +832,10 @@ xfs_diflags_to_iflags(
8075 + inode->i_flags |= S_IMMUTABLE;
8076 + else
8077 + inode->i_flags &= ~S_IMMUTABLE;
8078 ++ if (ip->i_d.di_flags & XFS_DIFLAG_IXUNLINK)
8079 ++ inode->i_flags |= S_IXUNLINK;
8080 ++ else
8081 ++ inode->i_flags &= ~S_IXUNLINK;
8082 + if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
8083 + inode->i_flags |= S_APPEND;
8084 + else
8085 +@@ -834,6 +848,15 @@ xfs_diflags_to_iflags(
8086 + inode->i_flags |= S_NOATIME;
8087 + else
8088 + inode->i_flags &= ~S_NOATIME;
8089 ++
8090 ++ if (ip->i_d.di_vflags & XFS_DIVFLAG_BARRIER)
8091 ++ inode->i_vflags |= V_BARRIER;
8092 ++ else
8093 ++ inode->i_vflags &= ~V_BARRIER;
8094 ++ if (ip->i_d.di_vflags & XFS_DIVFLAG_COW)
8095 ++ inode->i_vflags |= V_COW;
8096 ++ else
8097 ++ inode->i_vflags &= ~V_COW;
8098 + }
8099 +
8100 + /*
8101 +@@ -862,6 +885,7 @@ xfs_setup_inode(
8102 + inode->i_nlink = ip->i_d.di_nlink;
8103 + inode->i_uid = ip->i_d.di_uid;
8104 + inode->i_gid = ip->i_d.di_gid;
8105 ++ inode->i_tag = ip->i_d.di_tag;
8106 +
8107 + switch (inode->i_mode & S_IFMT) {
8108 + case S_IFBLK:
8109 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/linux-2.6/xfs_linux.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_linux.h
8110 +--- linux-2.6.29.4/fs/xfs/linux-2.6/xfs_linux.h 2009-03-24 14:22:37.000000000 +0100
8111 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_linux.h 2009-03-25 01:26:04.000000000 +0100
8112 +@@ -119,6 +119,7 @@
8113 +
8114 + #define current_cpu() (raw_smp_processor_id())
8115 + #define current_pid() (current->pid)
8116 ++#define current_fstag(cred,vp) (dx_current_fstag((vp)->i_sb))
8117 + #define current_test_flags(f) (current->flags & (f))
8118 + #define current_set_flags_nested(sp, f) \
8119 + (*(sp) = current->flags, current->flags |= (f))
8120 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/linux-2.6/xfs_super.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_super.c
8121 +--- linux-2.6.29.4/fs/xfs/linux-2.6/xfs_super.c 2009-03-24 14:22:37.000000000 +0100
8122 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/linux-2.6/xfs_super.c 2009-03-25 01:48:51.000000000 +0100
8123 +@@ -120,6 +120,9 @@ mempool_t *xfs_ioend_pool;
8124 + #define MNTOPT_DMAPI "dmapi" /* DMI enabled (DMAPI / XDSM) */
8125 + #define MNTOPT_XDSM "xdsm" /* DMI enabled (DMAPI / XDSM) */
8126 + #define MNTOPT_DMI "dmi" /* DMI enabled (DMAPI / XDSM) */
8127 ++#define MNTOPT_TAGXID "tagxid" /* context tagging for inodes */
8128 ++#define MNTOPT_TAGGED "tag" /* context tagging for inodes */
8129 ++#define MNTOPT_NOTAGTAG "notag" /* do not use context tagging */
8130 +
8131 + /*
8132 + * Table driven mount option parser.
8133 +@@ -128,10 +131,14 @@ mempool_t *xfs_ioend_pool;
8134 + * in the future, too.
8135 + */
8136 + enum {
8137 ++ Opt_tag, Opt_notag,
8138 + Opt_barrier, Opt_nobarrier, Opt_err
8139 + };
8140 +
8141 + static const match_table_t tokens = {
8142 ++ {Opt_tag, "tagxid"},
8143 ++ {Opt_tag, "tag"},
8144 ++ {Opt_notag, "notag"},
8145 + {Opt_barrier, "barrier"},
8146 + {Opt_nobarrier, "nobarrier"},
8147 + {Opt_err, NULL}
8148 +@@ -395,6 +402,19 @@ xfs_parseargs(
8149 + } else if (!strcmp(this_char, "irixsgid")) {
8150 + cmn_err(CE_WARN,
8151 + "XFS: irixsgid is now a sysctl(2) variable, option is deprecated.");
8152 ++#ifndef CONFIG_TAGGING_NONE
8153 ++ } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
8154 ++ mp->m_flags |= XFS_MOUNT_TAGGED;
8155 ++ } else if (!strcmp(this_char, MNTOPT_NOTAGTAG)) {
8156 ++ mp->m_flags &= ~XFS_MOUNT_TAGGED;
8157 ++ } else if (!strcmp(this_char, MNTOPT_TAGXID)) {
8158 ++ mp->m_flags |= XFS_MOUNT_TAGGED;
8159 ++#endif
8160 ++#ifdef CONFIG_PROPAGATE
8161 ++ } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
8162 ++ /* use value */
8163 ++ mp->m_flags |= XFS_MOUNT_TAGGED;
8164 ++#endif
8165 + } else {
8166 + cmn_err(CE_WARN,
8167 + "XFS: unknown mount option [%s].", this_char);
8168 +@@ -1221,6 +1241,16 @@ xfs_fs_remount(
8169 + case Opt_nobarrier:
8170 + mp->m_flags &= ~XFS_MOUNT_BARRIER;
8171 + break;
8172 ++ case Opt_tag:
8173 ++ if (!(sb->s_flags & MS_TAGGED)) {
8174 ++ printk(KERN_INFO
8175 ++ "XFS: %s: tagging not permitted on remount.\n",
8176 ++ sb->s_id);
8177 ++ return -EINVAL;
8178 ++ }
8179 ++ break;
8180 ++ case Opt_notag:
8181 ++ break;
8182 + default:
8183 + /*
8184 + * Logically we would return an error here to prevent
8185 +@@ -1480,6 +1510,9 @@ xfs_fs_fill_super(
8186 +
8187 + XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, mtpt, mp->m_fsname);
8188 +
8189 ++ if (mp->m_flags & XFS_MOUNT_TAGGED)
8190 ++ sb->s_flags |= MS_TAGGED;
8191 ++
8192 + sb->s_dirt = 1;
8193 + sb->s_magic = XFS_SB_MAGIC;
8194 + sb->s_blocksize = mp->m_sb.sb_blocksize;
8195 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/quota/xfs_qm_syscalls.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/quota/xfs_qm_syscalls.c
8196 +--- linux-2.6.29.4/fs/xfs/quota/xfs_qm_syscalls.c 2009-03-24 14:22:37.000000000 +0100
8197 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/quota/xfs_qm_syscalls.c 2009-03-24 14:48:35.000000000 +0100
8198 +@@ -426,7 +426,7 @@ xfs_qm_scall_quotaon(
8199 + uint accflags;
8200 + __int64_t sbflags;
8201 +
8202 +- if (!capable(CAP_SYS_ADMIN))
8203 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
8204 + return XFS_ERROR(EPERM);
8205 +
8206 + flags &= (XFS_ALL_QUOTA_ACCT | XFS_ALL_QUOTA_ENFD);
8207 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_dinode.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_dinode.h
8208 +--- linux-2.6.29.4/fs/xfs/xfs_dinode.h 2009-03-24 14:22:37.000000000 +0100
8209 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_dinode.h 2009-03-24 16:38:27.000000000 +0100
8210 +@@ -50,7 +50,9 @@ typedef struct xfs_dinode {
8211 + __be32 di_gid; /* owner's group id */
8212 + __be32 di_nlink; /* number of links to file */
8213 + __be16 di_projid; /* owner's project id */
8214 +- __u8 di_pad[8]; /* unused, zeroed space */
8215 ++ __be16 di_tag; /* context tagging */
8216 ++ __be16 di_vflags; /* vserver specific flags */
8217 ++ __u8 di_pad[4]; /* unused, zeroed space */
8218 + __be16 di_flushiter; /* incremented on flush */
8219 + xfs_timestamp_t di_atime; /* time last accessed */
8220 + xfs_timestamp_t di_mtime; /* time last modified */
8221 +@@ -181,6 +183,8 @@ static inline void xfs_dinode_put_rdev(s
8222 + #define XFS_DIFLAG_EXTSZINHERIT_BIT 12 /* inherit inode extent size */
8223 + #define XFS_DIFLAG_NODEFRAG_BIT 13 /* do not reorganize/defragment */
8224 + #define XFS_DIFLAG_FILESTREAM_BIT 14 /* use filestream allocator */
8225 ++#define XFS_DIFLAG_IXUNLINK_BIT 15 /* Immutable inver on unlink */
8226 ++
8227 + #define XFS_DIFLAG_REALTIME (1 << XFS_DIFLAG_REALTIME_BIT)
8228 + #define XFS_DIFLAG_PREALLOC (1 << XFS_DIFLAG_PREALLOC_BIT)
8229 + #define XFS_DIFLAG_NEWRTBM (1 << XFS_DIFLAG_NEWRTBM_BIT)
8230 +@@ -196,6 +200,7 @@ static inline void xfs_dinode_put_rdev(s
8231 + #define XFS_DIFLAG_EXTSZINHERIT (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
8232 + #define XFS_DIFLAG_NODEFRAG (1 << XFS_DIFLAG_NODEFRAG_BIT)
8233 + #define XFS_DIFLAG_FILESTREAM (1 << XFS_DIFLAG_FILESTREAM_BIT)
8234 ++#define XFS_DIFLAG_IXUNLINK (1 << XFS_DIFLAG_IXUNLINK_BIT)
8235 +
8236 + #ifdef CONFIG_XFS_RT
8237 + #define XFS_IS_REALTIME_INODE(ip) ((ip)->i_d.di_flags & XFS_DIFLAG_REALTIME)
8238 +@@ -208,6 +213,10 @@ static inline void xfs_dinode_put_rdev(s
8239 + XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
8240 + XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
8241 + XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
8242 +- XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
8243 ++ XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM | \
8244 ++ XFS_DIFLAG_IXUNLINK)
8245 ++
8246 ++#define XFS_DIVFLAG_BARRIER 0x01
8247 ++#define XFS_DIVFLAG_COW 0x02
8248 +
8249 + #endif /* __XFS_DINODE_H__ */
8250 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_fs.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_fs.h
8251 +--- linux-2.6.29.4/fs/xfs/xfs_fs.h 2009-03-24 14:22:37.000000000 +0100
8252 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_fs.h 2009-03-24 14:48:35.000000000 +0100
8253 +@@ -67,6 +67,9 @@ struct fsxattr {
8254 + #define XFS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
8255 + #define XFS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
8256 + #define XFS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
8257 ++#define XFS_XFLAG_IXUNLINK 0x00008000 /* immutable invert on unlink */
8258 ++#define XFS_XFLAG_BARRIER 0x10000000 /* chroot() barrier */
8259 ++#define XFS_XFLAG_COW 0x20000000 /* copy on write mark */
8260 + #define XFS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
8261 +
8262 + /*
8263 +@@ -289,7 +292,8 @@ typedef struct xfs_bstat {
8264 + __s32 bs_extents; /* number of extents */
8265 + __u32 bs_gen; /* generation count */
8266 + __u16 bs_projid; /* project id */
8267 +- unsigned char bs_pad[14]; /* pad space, unused */
8268 ++ __u16 bs_tag; /* context tagging */
8269 ++ unsigned char bs_pad[12]; /* pad space, unused */
8270 + __u32 bs_dmevmask; /* DMIG event mask */
8271 + __u16 bs_dmstate; /* DMIG state info */
8272 + __u16 bs_aextents; /* attribute number of extents */
8273 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_ialloc.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_ialloc.c
8274 +--- linux-2.6.29.4/fs/xfs/xfs_ialloc.c 2009-03-24 14:22:37.000000000 +0100
8275 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_ialloc.c 2009-03-24 17:03:17.000000000 +0100
8276 +@@ -41,7 +41,6 @@
8277 + #include "xfs_error.h"
8278 + #include "xfs_bmap.h"
8279 +
8280 +-
8281 + /*
8282 + * Allocation group level functions.
8283 + */
8284 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_inode.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_inode.c
8285 +--- linux-2.6.29.4/fs/xfs/xfs_inode.c 2009-03-24 14:22:37.000000000 +0100
8286 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_inode.c 2009-03-25 01:42:50.000000000 +0100
8287 +@@ -250,6 +250,7 @@ xfs_inotobp(
8288 + return 0;
8289 + }
8290 +
8291 ++#include <linux/vs_tag.h>
8292 +
8293 + /*
8294 + * This routine is called to map an inode to the buffer containing
8295 +@@ -645,15 +646,25 @@ xfs_iformat_btree(
8296 + void
8297 + xfs_dinode_from_disk(
8298 + xfs_icdinode_t *to,
8299 +- xfs_dinode_t *from)
8300 ++ xfs_dinode_t *from,
8301 ++ int tagged)
8302 + {
8303 ++ uint32_t uid, gid, tag;
8304 ++
8305 + to->di_magic = be16_to_cpu(from->di_magic);
8306 + to->di_mode = be16_to_cpu(from->di_mode);
8307 + to->di_version = from ->di_version;
8308 + to->di_format = from->di_format;
8309 + to->di_onlink = be16_to_cpu(from->di_onlink);
8310 +- to->di_uid = be32_to_cpu(from->di_uid);
8311 +- to->di_gid = be32_to_cpu(from->di_gid);
8312 ++
8313 ++ uid = be32_to_cpu(from->di_uid);
8314 ++ gid = be32_to_cpu(from->di_gid);
8315 ++ tag = be16_to_cpu(from->di_tag);
8316 ++
8317 ++ to->di_uid = INOTAG_UID(tagged, uid, gid);
8318 ++ to->di_gid = INOTAG_GID(tagged, uid, gid);
8319 ++ to->di_tag = INOTAG_TAG(tagged, uid, gid, tag);
8320 ++
8321 + to->di_nlink = be32_to_cpu(from->di_nlink);
8322 + to->di_projid = be16_to_cpu(from->di_projid);
8323 + memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8324 +@@ -674,21 +685,26 @@ xfs_dinode_from_disk(
8325 + to->di_dmevmask = be32_to_cpu(from->di_dmevmask);
8326 + to->di_dmstate = be16_to_cpu(from->di_dmstate);
8327 + to->di_flags = be16_to_cpu(from->di_flags);
8328 ++ to->di_vflags = be16_to_cpu(from->di_vflags);
8329 + to->di_gen = be32_to_cpu(from->di_gen);
8330 + }
8331 +
8332 + void
8333 + xfs_dinode_to_disk(
8334 + xfs_dinode_t *to,
8335 +- xfs_icdinode_t *from)
8336 ++ xfs_icdinode_t *from,
8337 ++ int tagged)
8338 + {
8339 + to->di_magic = cpu_to_be16(from->di_magic);
8340 + to->di_mode = cpu_to_be16(from->di_mode);
8341 + to->di_version = from ->di_version;
8342 + to->di_format = from->di_format;
8343 + to->di_onlink = cpu_to_be16(from->di_onlink);
8344 +- to->di_uid = cpu_to_be32(from->di_uid);
8345 +- to->di_gid = cpu_to_be32(from->di_gid);
8346 ++
8347 ++ to->di_uid = cpu_to_be32(TAGINO_UID(tagged, from->di_uid, from->di_tag));
8348 ++ to->di_gid = cpu_to_be32(TAGINO_GID(tagged, from->di_gid, from->di_tag));
8349 ++ to->di_tag = cpu_to_be16(TAGINO_TAG(tagged, from->di_tag));
8350 ++
8351 + to->di_nlink = cpu_to_be32(from->di_nlink);
8352 + to->di_projid = cpu_to_be16(from->di_projid);
8353 + memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8354 +@@ -709,12 +725,14 @@ xfs_dinode_to_disk(
8355 + to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
8356 + to->di_dmstate = cpu_to_be16(from->di_dmstate);
8357 + to->di_flags = cpu_to_be16(from->di_flags);
8358 ++ to->di_vflags = cpu_to_be16(from->di_vflags);
8359 + to->di_gen = cpu_to_be32(from->di_gen);
8360 + }
8361 +
8362 + STATIC uint
8363 + _xfs_dic2xflags(
8364 +- __uint16_t di_flags)
8365 ++ __uint16_t di_flags,
8366 ++ __uint16_t di_vflags)
8367 + {
8368 + uint flags = 0;
8369 +
8370 +@@ -725,6 +743,8 @@ _xfs_dic2xflags(
8371 + flags |= XFS_XFLAG_PREALLOC;
8372 + if (di_flags & XFS_DIFLAG_IMMUTABLE)
8373 + flags |= XFS_XFLAG_IMMUTABLE;
8374 ++ if (di_flags & XFS_DIFLAG_IXUNLINK)
8375 ++ flags |= XFS_XFLAG_IXUNLINK;
8376 + if (di_flags & XFS_DIFLAG_APPEND)
8377 + flags |= XFS_XFLAG_APPEND;
8378 + if (di_flags & XFS_DIFLAG_SYNC)
8379 +@@ -749,6 +769,10 @@ _xfs_dic2xflags(
8380 + flags |= XFS_XFLAG_FILESTREAM;
8381 + }
8382 +
8383 ++ if (di_vflags & XFS_DIVFLAG_BARRIER)
8384 ++ flags |= FS_BARRIER_FL;
8385 ++ if (di_vflags & XFS_DIVFLAG_COW)
8386 ++ flags |= FS_COW_FL;
8387 + return flags;
8388 + }
8389 +
8390 +@@ -758,7 +782,7 @@ xfs_ip2xflags(
8391 + {
8392 + xfs_icdinode_t *dic = &ip->i_d;
8393 +
8394 +- return _xfs_dic2xflags(dic->di_flags) |
8395 ++ return _xfs_dic2xflags(dic->di_flags, dic->di_vflags) |
8396 + (XFS_IFORK_Q(ip) ? XFS_XFLAG_HASATTR : 0);
8397 + }
8398 +
8399 +@@ -766,7 +790,8 @@ uint
8400 + xfs_dic2xflags(
8401 + xfs_dinode_t *dip)
8402 + {
8403 +- return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
8404 ++ return _xfs_dic2xflags(be16_to_cpu(dip->di_flags),
8405 ++ be16_to_cpu(dip->di_vflags)) |
8406 + (XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
8407 + }
8408 +
8409 +@@ -802,7 +827,6 @@ xfs_iread(
8410 + if (error)
8411 + return error;
8412 + dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
8413 +-
8414 + /*
8415 + * If we got something that isn't an inode it means someone
8416 + * (nfs or dmi) has a stale handle.
8417 +@@ -827,7 +851,8 @@ xfs_iread(
8418 + * Otherwise, just get the truly permanent information.
8419 + */
8420 + if (dip->di_mode) {
8421 +- xfs_dinode_from_disk(&ip->i_d, dip);
8422 ++ xfs_dinode_from_disk(&ip->i_d, dip,
8423 ++ mp->m_flags & XFS_MOUNT_TAGGED);
8424 + error = xfs_iformat(ip, dip);
8425 + if (error) {
8426 + #ifdef DEBUG
8427 +@@ -1027,6 +1052,7 @@ xfs_ialloc(
8428 + ASSERT(ip->i_d.di_nlink == nlink);
8429 + ip->i_d.di_uid = current_fsuid();
8430 + ip->i_d.di_gid = current_fsgid();
8431 ++ ip->i_d.di_tag = current_fstag(cr, &ip->i_vnode);
8432 + ip->i_d.di_projid = prid;
8433 + memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
8434 +
8435 +@@ -1087,6 +1113,7 @@ xfs_ialloc(
8436 + ip->i_d.di_dmevmask = 0;
8437 + ip->i_d.di_dmstate = 0;
8438 + ip->i_d.di_flags = 0;
8439 ++ ip->i_d.di_vflags = 0;
8440 + flags = XFS_ILOG_CORE;
8441 + switch (mode & S_IFMT) {
8442 + case S_IFIFO:
8443 +@@ -2161,6 +2188,7 @@ xfs_ifree(
8444 + }
8445 + ip->i_d.di_mode = 0; /* mark incore inode as free */
8446 + ip->i_d.di_flags = 0;
8447 ++ ip->i_d.di_vflags = 0;
8448 + ip->i_d.di_dmevmask = 0;
8449 + ip->i_d.di_forkoff = 0; /* mark the attr fork not in use */
8450 + ip->i_df.if_ext_max =
8451 +@@ -3128,7 +3156,8 @@ xfs_iflush_int(
8452 + * because if the inode is dirty at all the core must
8453 + * be.
8454 + */
8455 +- xfs_dinode_to_disk(dip, &ip->i_d);
8456 ++ xfs_dinode_to_disk(dip, &ip->i_d,
8457 ++ mp->m_flags & XFS_MOUNT_TAGGED);
8458 +
8459 + /* Wrap, we never let the log put out DI_MAX_FLUSH */
8460 + if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
8461 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_inode.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_inode.h
8462 +--- linux-2.6.29.4/fs/xfs/xfs_inode.h 2009-03-24 14:22:37.000000000 +0100
8463 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_inode.h 2009-03-25 01:10:35.000000000 +0100
8464 +@@ -134,7 +134,9 @@ typedef struct xfs_icdinode {
8465 + __uint32_t di_gid; /* owner's group id */
8466 + __uint32_t di_nlink; /* number of links to file */
8467 + __uint16_t di_projid; /* owner's project id */
8468 +- __uint8_t di_pad[8]; /* unused, zeroed space */
8469 ++ __uint16_t di_tag; /* context tagging */
8470 ++ __uint16_t di_vflags; /* vserver specific flags */
8471 ++ __uint8_t di_pad[4]; /* unused, zeroed space */
8472 + __uint16_t di_flushiter; /* incremented on flush */
8473 + xfs_ictimestamp_t di_atime; /* time last accessed */
8474 + xfs_ictimestamp_t di_mtime; /* time last modified */
8475 +@@ -589,9 +591,9 @@ int xfs_itobp(struct xfs_mount *, struc
8476 + int xfs_iread(struct xfs_mount *, struct xfs_trans *,
8477 + struct xfs_inode *, xfs_daddr_t, uint);
8478 + void xfs_dinode_from_disk(struct xfs_icdinode *,
8479 +- struct xfs_dinode *);
8480 ++ struct xfs_dinode *, int);
8481 + void xfs_dinode_to_disk(struct xfs_dinode *,
8482 +- struct xfs_icdinode *);
8483 ++ struct xfs_icdinode *, int);
8484 + void xfs_idestroy_fork(struct xfs_inode *, int);
8485 + void xfs_idata_realloc(struct xfs_inode *, int, int);
8486 + void xfs_iroot_realloc(struct xfs_inode *, int, int);
8487 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_itable.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_itable.c
8488 +--- linux-2.6.29.4/fs/xfs/xfs_itable.c 2009-03-24 14:22:37.000000000 +0100
8489 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_itable.c 2009-03-24 14:48:35.000000000 +0100
8490 +@@ -82,6 +82,7 @@ xfs_bulkstat_one_iget(
8491 + buf->bs_mode = dic->di_mode;
8492 + buf->bs_uid = dic->di_uid;
8493 + buf->bs_gid = dic->di_gid;
8494 ++ buf->bs_tag = dic->di_tag;
8495 + buf->bs_size = dic->di_size;
8496 + vn_atime_to_bstime(VFS_I(ip), &buf->bs_atime);
8497 + buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
8498 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_log_recover.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_log_recover.c
8499 +--- linux-2.6.29.4/fs/xfs/xfs_log_recover.c 2009-03-24 14:22:37.000000000 +0100
8500 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_log_recover.c 2009-03-24 17:17:15.000000000 +0100
8501 +@@ -2421,7 +2421,8 @@ xlog_recover_do_inode_trans(
8502 + }
8503 +
8504 + /* The core is in in-core format */
8505 +- xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr);
8506 ++ xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr,
8507 ++ mp->m_flags & XFS_MOUNT_TAGGED);
8508 +
8509 + /* the rest is in on-disk format */
8510 + if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
8511 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_mount.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_mount.h
8512 +--- linux-2.6.29.4/fs/xfs/xfs_mount.h 2009-03-24 14:22:37.000000000 +0100
8513 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_mount.h 2009-03-24 14:48:35.000000000 +0100
8514 +@@ -371,6 +371,7 @@ typedef struct xfs_mount {
8515 + allocator */
8516 + #define XFS_MOUNT_NOATTR2 (1ULL << 25) /* disable use of attr2 format */
8517 +
8518 ++#define XFS_MOUNT_TAGGED (1ULL << 31) /* context tagging */
8519 +
8520 + /*
8521 + * Default minimum read and write sizes.
8522 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_vnodeops.c linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_vnodeops.c
8523 +--- linux-2.6.29.4/fs/xfs/xfs_vnodeops.c 2009-03-24 14:22:37.000000000 +0100
8524 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_vnodeops.c 2009-03-24 17:18:40.000000000 +0100
8525 +@@ -53,6 +53,90 @@
8526 + #include "xfs_filestream.h"
8527 + #include "xfs_vnodeops.h"
8528 +
8529 ++
8530 ++STATIC void
8531 ++xfs_get_inode_flags(
8532 ++ xfs_inode_t *ip)
8533 ++{
8534 ++ struct inode *inode = VFS_I(ip);
8535 ++ unsigned int flags = inode->i_flags;
8536 ++ unsigned int vflags = inode->i_vflags;
8537 ++
8538 ++ if (flags & S_IMMUTABLE)
8539 ++ ip->i_d.di_flags |= XFS_DIFLAG_IMMUTABLE;
8540 ++ else
8541 ++ ip->i_d.di_flags &= ~XFS_DIFLAG_IMMUTABLE;
8542 ++ if (flags & S_IXUNLINK)
8543 ++ ip->i_d.di_flags |= XFS_DIFLAG_IXUNLINK;
8544 ++ else
8545 ++ ip->i_d.di_flags &= ~XFS_DIFLAG_IXUNLINK;
8546 ++
8547 ++ if (vflags & V_BARRIER)
8548 ++ ip->i_d.di_vflags |= XFS_DIVFLAG_BARRIER;
8549 ++ else
8550 ++ ip->i_d.di_vflags &= ~XFS_DIVFLAG_BARRIER;
8551 ++ if (vflags & V_COW)
8552 ++ ip->i_d.di_vflags |= XFS_DIVFLAG_COW;
8553 ++ else
8554 ++ ip->i_d.di_vflags &= ~XFS_DIVFLAG_COW;
8555 ++}
8556 ++
8557 ++int
8558 ++xfs_sync_xflags(
8559 ++ xfs_inode_t *ip)
8560 ++{
8561 ++ struct xfs_mount *mp = ip->i_mount;
8562 ++ struct xfs_trans *tp;
8563 ++ unsigned int lock_flags = 0;
8564 ++ int code;
8565 ++
8566 ++ xfs_itrace_entry(ip);
8567 ++
8568 ++ if (mp->m_flags & XFS_MOUNT_RDONLY)
8569 ++ return XFS_ERROR(EROFS);
8570 ++
8571 ++ /*
8572 ++ * we acquire the inode lock and do an error checking pass.
8573 ++ */
8574 ++ tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
8575 ++ code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
8576 ++ if (code)
8577 ++ goto error_return;
8578 ++
8579 ++ lock_flags = XFS_ILOCK_EXCL;
8580 ++ xfs_ilock(ip, lock_flags);
8581 ++
8582 ++ xfs_trans_ijoin(tp, ip, lock_flags);
8583 ++ xfs_trans_ihold(tp, ip);
8584 ++
8585 ++ xfs_get_inode_flags(ip);
8586 ++ // xfs_diflags_to_linux(ip);
8587 ++
8588 ++ xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
8589 ++ xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
8590 ++
8591 ++ XFS_STATS_INC(xs_ig_attrchg);
8592 ++
8593 ++ /*
8594 ++ * If this is a synchronous mount, make sure that the
8595 ++ * transaction goes to disk before returning to the user.
8596 ++ */
8597 ++ if (mp->m_flags & XFS_MOUNT_WSYNC)
8598 ++ xfs_trans_set_sync(tp);
8599 ++ code = xfs_trans_commit(tp, 0);
8600 ++ xfs_iunlock(ip, lock_flags);
8601 ++
8602 ++ if (code)
8603 ++ return code;
8604 ++ return 0;
8605 ++
8606 ++ error_return:
8607 ++ xfs_trans_cancel(tp, 0);
8608 ++ if (lock_flags)
8609 ++ xfs_iunlock(ip, lock_flags);
8610 ++ return code;
8611 ++}
8612 ++
8613 + int
8614 + xfs_setattr(
8615 + struct xfs_inode *ip,
8616 +@@ -68,6 +152,7 @@ xfs_setattr(
8617 + uint commit_flags=0;
8618 + uid_t uid=0, iuid=0;
8619 + gid_t gid=0, igid=0;
8620 ++ tag_t tag=0, itag=0;
8621 + int timeflags = 0;
8622 + struct xfs_dquot *udqp, *gdqp, *olddquot1, *olddquot2;
8623 + int need_iolock = 1;
8624 +@@ -164,7 +249,7 @@ xfs_setattr(
8625 + /*
8626 + * Change file ownership. Must be the owner or privileged.
8627 + */
8628 +- if (mask & (ATTR_UID|ATTR_GID)) {
8629 ++ if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8630 + /*
8631 + * These IDs could have changed since we last looked at them.
8632 + * But, we're assured that if the ownership did change
8633 +@@ -173,15 +258,19 @@ xfs_setattr(
8634 + */
8635 + iuid = ip->i_d.di_uid;
8636 + igid = ip->i_d.di_gid;
8637 ++ itag = ip->i_d.di_tag;
8638 + gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
8639 + uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
8640 ++ tag = (mask & ATTR_TAG) ? iattr->ia_tag : itag;
8641 +
8642 + /*
8643 + * Do a quota reservation only if uid/gid is actually
8644 + * going to change.
8645 + */
8646 + if ((XFS_IS_UQUOTA_ON(mp) && iuid != uid) ||
8647 +- (XFS_IS_GQUOTA_ON(mp) && igid != gid)) {
8648 ++ (XFS_IS_GQUOTA_ON(mp) && igid != gid) ||
8649 ++ (XFS_IS_GQUOTA_ON(mp) && itag != tag)) {
8650 ++ /* TODO: handle tagging? */
8651 + ASSERT(tp);
8652 + code = XFS_QM_DQVOPCHOWNRESV(mp, tp, ip, udqp, gdqp,
8653 + capable(CAP_FOWNER) ?
8654 +@@ -334,7 +423,7 @@ xfs_setattr(
8655 + /*
8656 + * Change file ownership. Must be the owner or privileged.
8657 + */
8658 +- if (mask & (ATTR_UID|ATTR_GID)) {
8659 ++ if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8660 + /*
8661 + * CAP_FSETID overrides the following restrictions:
8662 + *
8663 +@@ -350,6 +439,10 @@ xfs_setattr(
8664 + * Change the ownerships and register quota modifications
8665 + * in the transaction.
8666 + */
8667 ++ if (itag != tag) {
8668 ++ ip->i_d.di_tag = tag;
8669 ++ inode->i_tag = tag;
8670 ++ }
8671 + if (iuid != uid) {
8672 + if (XFS_IS_UQUOTA_ON(mp)) {
8673 + ASSERT(mask & ATTR_UID);
8674 +diff -NurpP --minimal linux-2.6.29.4/fs/xfs/xfs_vnodeops.h linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_vnodeops.h
8675 +--- linux-2.6.29.4/fs/xfs/xfs_vnodeops.h 2009-03-24 14:22:37.000000000 +0100
8676 ++++ linux-2.6.29.4-vs2.3.0.36.14/fs/xfs/xfs_vnodeops.h 2009-03-24 17:19:30.000000000 +0100
8677 +@@ -14,6 +14,7 @@ struct xfs_inode;
8678 + struct xfs_iomap;
8679 +
8680 +
8681 ++int xfs_sync_xflags(struct xfs_inode *ip);
8682 + int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
8683 + #define XFS_ATTR_DMI 0x01 /* invocation from a DMI function */
8684 + #define XFS_ATTR_NONBLOCK 0x02 /* return EAGAIN if operation would block */
8685 +diff -NurpP --minimal linux-2.6.29.4/include/asm-generic/tlb.h linux-2.6.29.4-vs2.3.0.36.14/include/asm-generic/tlb.h
8686 +--- linux-2.6.29.4/include/asm-generic/tlb.h 2008-12-25 00:26:37.000000000 +0100
8687 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/asm-generic/tlb.h 2009-04-08 15:50:06.000000000 +0200
8688 +@@ -14,6 +14,7 @@
8689 + #define _ASM_GENERIC__TLB_H
8690 +
8691 + #include <linux/swap.h>
8692 ++#include <linux/vs_memory.h>
8693 + #include <asm/pgalloc.h>
8694 + #include <asm/tlbflush.h>
8695 +
8696 +diff -NurpP --minimal linux-2.6.29.4/include/linux/capability.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/capability.h
8697 +--- linux-2.6.29.4/include/linux/capability.h 2009-05-23 23:16:53.000000000 +0200
8698 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/capability.h 2009-04-30 12:14:53.000000000 +0200
8699 +@@ -285,6 +285,7 @@ struct cpu_vfs_cap_data {
8700 + arbitrary SCSI commands */
8701 + /* Allow setting encryption key on loopback filesystem */
8702 + /* Allow setting zone reclaim policy */
8703 ++/* Allow the selection of a security context */
8704 +
8705 + #define CAP_SYS_ADMIN 21
8706 +
8707 +@@ -357,7 +358,13 @@ struct cpu_vfs_cap_data {
8708 +
8709 + #define CAP_MAC_ADMIN 33
8710 +
8711 +-#define CAP_LAST_CAP CAP_MAC_ADMIN
8712 ++/* Allow context manipulations */
8713 ++/* Allow changing context info on files */
8714 ++
8715 ++#define CAP_CONTEXT 34
8716 ++
8717 ++
8718 ++#define CAP_LAST_CAP CAP_CONTEXT
8719 +
8720 + #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
8721 +
8722 +diff -NurpP --minimal linux-2.6.29.4/include/linux/devpts_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/devpts_fs.h
8723 +--- linux-2.6.29.4/include/linux/devpts_fs.h 2008-12-25 00:26:37.000000000 +0100
8724 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/devpts_fs.h 2009-02-22 22:54:26.000000000 +0100
8725 +@@ -45,5 +45,4 @@ static inline void devpts_pty_kill(struc
8726 +
8727 + #endif
8728 +
8729 +-
8730 + #endif /* _LINUX_DEVPTS_FS_H */
8731 +diff -NurpP --minimal linux-2.6.29.4/include/linux/ext2_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/ext2_fs.h
8732 +--- linux-2.6.29.4/include/linux/ext2_fs.h 2009-03-24 14:22:41.000000000 +0100
8733 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/ext2_fs.h 2009-03-24 15:39:54.000000000 +0100
8734 +@@ -189,8 +189,12 @@ struct ext2_group_desc
8735 + #define EXT2_NOTAIL_FL FS_NOTAIL_FL /* file tail should not be merged */
8736 + #define EXT2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
8737 + #define EXT2_TOPDIR_FL FS_TOPDIR_FL /* Top of directory hierarchies*/
8738 ++#define EXT2_IXUNLINK_FL FS_IXUNLINK_FL /* Immutable invert on unlink */
8739 + #define EXT2_RESERVED_FL FS_RESERVED_FL /* reserved for ext2 lib */
8740 +
8741 ++#define EXT2_BARRIER_FL FS_BARRIER_FL /* Barrier for chroot() */
8742 ++#define EXT2_COW_FL FS_COW_FL /* Copy on Write marker */
8743 ++
8744 + #define EXT2_FL_USER_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */
8745 + #define EXT2_FL_USER_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */
8746 +
8747 +@@ -271,7 +275,7 @@ struct ext2_inode {
8748 + struct {
8749 + __u8 l_i_frag; /* Fragment number */
8750 + __u8 l_i_fsize; /* Fragment size */
8751 +- __u16 i_pad1;
8752 ++ __u16 l_i_tag; /* Context Tag */
8753 + __le16 l_i_uid_high; /* these 2 fields */
8754 + __le16 l_i_gid_high; /* were reserved2[0] */
8755 + __u32 l_i_reserved2;
8756 +@@ -303,6 +307,7 @@ struct ext2_inode {
8757 + #define i_gid_low i_gid
8758 + #define i_uid_high osd2.linux2.l_i_uid_high
8759 + #define i_gid_high osd2.linux2.l_i_gid_high
8760 ++#define i_raw_tag osd2.linux2.l_i_tag
8761 + #define i_reserved2 osd2.linux2.l_i_reserved2
8762 + #endif
8763 +
8764 +@@ -347,6 +352,7 @@ struct ext2_inode {
8765 + #define EXT2_MOUNT_USRQUOTA 0x020000 /* user quota */
8766 + #define EXT2_MOUNT_GRPQUOTA 0x040000 /* group quota */
8767 + #define EXT2_MOUNT_RESERVATION 0x080000 /* Preallocation */
8768 ++#define EXT2_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8769 +
8770 +
8771 + #define clear_opt(o, opt) o &= ~EXT2_MOUNT_##opt
8772 +diff -NurpP --minimal linux-2.6.29.4/include/linux/ext3_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/ext3_fs.h
8773 +--- linux-2.6.29.4/include/linux/ext3_fs.h 2009-03-24 14:22:41.000000000 +0100
8774 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/ext3_fs.h 2009-03-24 14:48:36.000000000 +0100
8775 +@@ -173,10 +173,14 @@ struct ext3_group_desc
8776 + #define EXT3_NOTAIL_FL 0x00008000 /* file tail should not be merged */
8777 + #define EXT3_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
8778 + #define EXT3_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8779 ++#define EXT3_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8780 + #define EXT3_RESERVED_FL 0x80000000 /* reserved for ext3 lib */
8781 +
8782 +-#define EXT3_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8783 +-#define EXT3_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8784 ++#define EXT3_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8785 ++#define EXT3_COW_FL 0x20000000 /* Copy on Write marker */
8786 ++
8787 ++#define EXT3_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8788 ++#define EXT3_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8789 +
8790 + /* Flags that should be inherited by new inodes from their parent. */
8791 + #define EXT3_FL_INHERITED (EXT3_SECRM_FL | EXT3_UNRM_FL | EXT3_COMPR_FL |\
8792 +@@ -316,7 +320,7 @@ struct ext3_inode {
8793 + struct {
8794 + __u8 l_i_frag; /* Fragment number */
8795 + __u8 l_i_fsize; /* Fragment size */
8796 +- __u16 i_pad1;
8797 ++ __u16 l_i_tag; /* Context Tag */
8798 + __le16 l_i_uid_high; /* these 2 fields */
8799 + __le16 l_i_gid_high; /* were reserved2[0] */
8800 + __u32 l_i_reserved2;
8801 +@@ -350,6 +354,7 @@ struct ext3_inode {
8802 + #define i_gid_low i_gid
8803 + #define i_uid_high osd2.linux2.l_i_uid_high
8804 + #define i_gid_high osd2.linux2.l_i_gid_high
8805 ++#define i_raw_tag osd2.linux2.l_i_tag
8806 + #define i_reserved2 osd2.linux2.l_i_reserved2
8807 +
8808 + #elif defined(__GNU__)
8809 +@@ -413,6 +418,7 @@ struct ext3_inode {
8810 + #define EXT3_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
8811 + #define EXT3_MOUNT_DATA_ERR_ABORT 0x400000 /* Abort on file data write
8812 + * error in ordered mode */
8813 ++#define EXT3_MOUNT_TAGGED (1<<24) /* Enable Context Tags */
8814 +
8815 + /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
8816 + #ifndef _LINUX_EXT2_FS_H
8817 +@@ -874,6 +880,7 @@ struct buffer_head * ext3_bread (handle_
8818 + int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
8819 + sector_t iblock, unsigned long maxblocks, struct buffer_head *bh_result,
8820 + int create, int extend_disksize);
8821 ++extern int ext3_sync_flags(struct inode *inode);
8822 +
8823 + extern struct inode *ext3_iget(struct super_block *, unsigned long);
8824 + extern int ext3_write_inode (struct inode *, int);
8825 +diff -NurpP --minimal linux-2.6.29.4/include/linux/fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/fs.h
8826 +--- linux-2.6.29.4/include/linux/fs.h 2009-03-24 14:22:41.000000000 +0100
8827 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/fs.h 2009-03-24 17:21:20.000000000 +0100
8828 +@@ -141,6 +141,9 @@ struct inodes_stat_t {
8829 + #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
8830 + #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
8831 + #define MS_I_VERSION (1<<23) /* Update inode I_version field */
8832 ++#define MS_TAGGED (1<<24) /* use generic inode tagging */
8833 ++#define MS_TAGID (1<<25) /* use specific tag for this mount */
8834 ++#define MS_NOTAGCHECK (1<<26) /* don't check tags */
8835 + #define MS_ACTIVE (1<<30)
8836 + #define MS_NOUSER (1<<31)
8837 +
8838 +@@ -167,6 +170,14 @@ struct inodes_stat_t {
8839 + #define S_NOCMTIME 128 /* Do not update file c/mtime */
8840 + #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
8841 + #define S_PRIVATE 512 /* Inode is fs-internal */
8842 ++#define S_IXUNLINK 1024 /* Immutable Invert on unlink */
8843 ++
8844 ++/* Linux-VServer related Inode flags */
8845 ++
8846 ++#define V_VALID 1
8847 ++#define V_XATTR 2
8848 ++#define V_BARRIER 4 /* Barrier for chroot() */
8849 ++#define V_COW 8 /* Copy on Write */
8850 +
8851 + /*
8852 + * Note that nosuid etc flags are inode-specific: setting some file-system
8853 +@@ -189,12 +200,15 @@ struct inodes_stat_t {
8854 + #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
8855 + ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
8856 + #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
8857 +-#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8858 +-#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8859 ++#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8860 ++#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
8861 ++#define IS_TAGGED(inode) __IS_FLG(inode, MS_TAGGED)
8862 +
8863 + #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
8864 + #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
8865 + #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
8866 ++#define IS_IXUNLINK(inode) ((inode)->i_flags & S_IXUNLINK)
8867 ++#define IS_IXORUNLINK(inode) ((IS_IXUNLINK(inode) ? S_IMMUTABLE : 0) ^ IS_IMMUTABLE(inode))
8868 + #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
8869 +
8870 + #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
8871 +@@ -202,6 +216,16 @@ struct inodes_stat_t {
8872 + #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
8873 + #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
8874 +
8875 ++#define IS_BARRIER(inode) (S_ISDIR((inode)->i_mode) && ((inode)->i_vflags & V_BARRIER))
8876 ++
8877 ++#ifdef CONFIG_VSERVER_COWBL
8878 ++# define IS_COW(inode) (IS_IXUNLINK(inode) && IS_IMMUTABLE(inode))
8879 ++# define IS_COW_LINK(inode) (S_ISREG((inode)->i_mode) && ((inode)->i_nlink > 1))
8880 ++#else
8881 ++# define IS_COW(inode) (0)
8882 ++# define IS_COW_LINK(inode) (0)
8883 ++#endif
8884 ++
8885 + /* the read-only stuff doesn't really belong here, but any other place is
8886 + probably as bad and I don't want to create yet another include file. */
8887 +
8888 +@@ -279,11 +303,14 @@ struct inodes_stat_t {
8889 + #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8890 + #define FS_EXTENT_FL 0x00080000 /* Extents */
8891 + #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
8892 ++#define FS_IXUNLINK_FL 0x08000000 /* Immutable invert on unlink */
8893 + #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
8894 +
8895 +-#define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
8896 +-#define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
8897 ++#define FS_BARRIER_FL 0x04000000 /* Barrier for chroot() */
8898 ++#define FS_COW_FL 0x20000000 /* Copy on Write marker */
8899 +
8900 ++#define FS_FL_USER_VISIBLE 0x0103DFFF /* User visible flags */
8901 ++#define FS_FL_USER_MODIFIABLE 0x010380FF /* User modifiable flags */
8902 +
8903 + #define SYNC_FILE_RANGE_WAIT_BEFORE 1
8904 + #define SYNC_FILE_RANGE_WRITE 2
8905 +@@ -365,6 +392,7 @@ typedef void (dio_iodone_t)(struct kiocb
8906 + #define ATTR_KILL_PRIV (1 << 14)
8907 + #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
8908 + #define ATTR_TIMES_SET (1 << 16)
8909 ++#define ATTR_TAG (1 << 17)
8910 +
8911 + /*
8912 + * This is the Inode Attributes structure, used for notify_change(). It
8913 +@@ -380,6 +408,7 @@ struct iattr {
8914 + umode_t ia_mode;
8915 + uid_t ia_uid;
8916 + gid_t ia_gid;
8917 ++ tag_t ia_tag;
8918 + loff_t ia_size;
8919 + struct timespec ia_atime;
8920 + struct timespec ia_mtime;
8921 +@@ -393,6 +422,9 @@ struct iattr {
8922 + struct file *ia_file;
8923 + };
8924 +
8925 ++#define ATTR_FLAG_BARRIER 512 /* Barrier for chroot() */
8926 ++#define ATTR_FLAG_IXUNLINK 1024 /* Immutable invert on unlink */
8927 ++
8928 + /*
8929 + * Includes for diskquotas.
8930 + */
8931 +@@ -656,7 +688,9 @@ struct inode {
8932 + unsigned int i_nlink;
8933 + uid_t i_uid;
8934 + gid_t i_gid;
8935 ++ tag_t i_tag;
8936 + dev_t i_rdev;
8937 ++ dev_t i_mdev;
8938 + u64 i_version;
8939 + loff_t i_size;
8940 + #ifdef __NEED_I_SIZE_ORDERED
8941 +@@ -704,7 +738,8 @@ struct inode {
8942 + unsigned long i_state;
8943 + unsigned long dirtied_when; /* jiffies of first dirtying */
8944 +
8945 +- unsigned int i_flags;
8946 ++ unsigned short i_flags;
8947 ++ unsigned short i_vflags;
8948 +
8949 + atomic_t i_writecount;
8950 + #ifdef CONFIG_SECURITY
8951 +@@ -791,12 +826,12 @@ static inline void i_size_write(struct i
8952 +
8953 + static inline unsigned iminor(const struct inode *inode)
8954 + {
8955 +- return MINOR(inode->i_rdev);
8956 ++ return MINOR(inode->i_mdev);
8957 + }
8958 +
8959 + static inline unsigned imajor(const struct inode *inode)
8960 + {
8961 +- return MAJOR(inode->i_rdev);
8962 ++ return MAJOR(inode->i_mdev);
8963 + }
8964 +
8965 + extern struct block_device *I_BDEV(struct inode *inode);
8966 +@@ -854,6 +889,7 @@ struct file {
8967 + loff_t f_pos;
8968 + struct fown_struct f_owner;
8969 + const struct cred *f_cred;
8970 ++ xid_t f_xid;
8971 + struct file_ra_state f_ra;
8972 +
8973 + u64 f_version;
8974 +@@ -996,6 +1032,7 @@ struct file_lock {
8975 + struct file *fl_file;
8976 + loff_t fl_start;
8977 + loff_t fl_end;
8978 ++ xid_t fl_xid;
8979 +
8980 + struct fasync_struct * fl_fasync; /* for lease break notifications */
8981 + unsigned long fl_break_time; /* for nonblocking lease breaks */
8982 +@@ -1363,6 +1400,7 @@ struct inode_operations {
8983 + loff_t len);
8984 + int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
8985 + u64 len);
8986 ++ int (*sync_flags) (struct inode *);
8987 + };
8988 +
8989 + struct seq_file;
8990 +@@ -1378,6 +1416,7 @@ extern ssize_t vfs_readv(struct file *,
8991 + unsigned long, loff_t *);
8992 + extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
8993 + unsigned long, loff_t *);
8994 ++ssize_t vfs_sendfile(struct file *, struct file *, loff_t *, size_t, loff_t);
8995 +
8996 + struct super_operations {
8997 + struct inode *(*alloc_inode)(struct super_block *sb);
8998 +@@ -2096,6 +2135,7 @@ extern int dcache_dir_open(struct inode
8999 + extern int dcache_dir_close(struct inode *, struct file *);
9000 + extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
9001 + extern int dcache_readdir(struct file *, void *, filldir_t);
9002 ++extern int dcache_readdir_filter(struct file *, void *, filldir_t, int (*)(struct dentry *));
9003 + extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
9004 + extern int simple_statfs(struct dentry *, struct kstatfs *);
9005 + extern int simple_link(struct dentry *, struct inode *, struct dentry *);
9006 +diff -NurpP --minimal linux-2.6.29.4/include/linux/if_tun.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/if_tun.h
9007 +--- linux-2.6.29.4/include/linux/if_tun.h 2008-12-25 00:26:37.000000000 +0100
9008 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/if_tun.h 2009-02-22 22:54:26.000000000 +0100
9009 +@@ -46,6 +46,7 @@
9010 + #define TUNSETOFFLOAD _IOW('T', 208, unsigned int)
9011 + #define TUNSETTXFILTER _IOW('T', 209, unsigned int)
9012 + #define TUNGETIFF _IOR('T', 210, unsigned int)
9013 ++#define TUNSETNID _IOW('T', 215, int)
9014 +
9015 + /* TUNSETIFF ifr flags */
9016 + #define IFF_TUN 0x0001
9017 +diff -NurpP --minimal linux-2.6.29.4/include/linux/init_task.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/init_task.h
9018 +--- linux-2.6.29.4/include/linux/init_task.h 2009-03-24 14:22:41.000000000 +0100
9019 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/init_task.h 2009-03-24 14:48:36.000000000 +0100
9020 +@@ -184,6 +184,10 @@ extern struct cred init_cred;
9021 + INIT_IDS \
9022 + INIT_TRACE_IRQFLAGS \
9023 + INIT_LOCKDEP \
9024 ++ .xid = 0, \
9025 ++ .vx_info = NULL, \
9026 ++ .nid = 0, \
9027 ++ .nx_info = NULL, \
9028 + }
9029 +
9030 +
9031 +diff -NurpP --minimal linux-2.6.29.4/include/linux/interrupt.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/interrupt.h
9032 +--- linux-2.6.29.4/include/linux/interrupt.h 2009-05-23 23:16:53.000000000 +0200
9033 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/interrupt.h 2009-04-30 12:14:53.000000000 +0200
9034 +@@ -9,8 +9,8 @@
9035 + #include <linux/cpumask.h>
9036 + #include <linux/irqreturn.h>
9037 + #include <linux/irqnr.h>
9038 +-#include <linux/hardirq.h>
9039 + #include <linux/sched.h>
9040 ++#include <linux/hardirq.h>
9041 + #include <linux/irqflags.h>
9042 + #include <linux/smp.h>
9043 + #include <linux/percpu.h>
9044 +diff -NurpP --minimal linux-2.6.29.4/include/linux/ipc.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/ipc.h
9045 +--- linux-2.6.29.4/include/linux/ipc.h 2008-12-25 00:26:37.000000000 +0100
9046 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/ipc.h 2009-02-22 22:54:26.000000000 +0100
9047 +@@ -93,6 +93,7 @@ struct kern_ipc_perm
9048 + key_t key;
9049 + uid_t uid;
9050 + gid_t gid;
9051 ++ xid_t xid;
9052 + uid_t cuid;
9053 + gid_t cgid;
9054 + mode_t mode;
9055 +diff -NurpP --minimal linux-2.6.29.4/include/linux/Kbuild linux-2.6.29.4-vs2.3.0.36.14/include/linux/Kbuild
9056 +--- linux-2.6.29.4/include/linux/Kbuild 2009-03-24 14:22:40.000000000 +0100
9057 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/Kbuild 2009-03-24 17:22:07.000000000 +0100
9058 +@@ -373,5 +373,8 @@ unifdef-y += xattr.h
9059 + unifdef-y += xfrm.h
9060 +
9061 + objhdr-y += version.h
9062 ++
9063 ++header-y += vserver/
9064 + header-y += wimax.h
9065 + header-y += wimax/
9066 ++
9067 +diff -NurpP --minimal linux-2.6.29.4/include/linux/loop.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/loop.h
9068 +--- linux-2.6.29.4/include/linux/loop.h 2009-03-24 14:22:42.000000000 +0100
9069 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/loop.h 2009-03-24 14:48:36.000000000 +0100
9070 +@@ -45,6 +45,7 @@ struct loop_device {
9071 + struct loop_func_table *lo_encryption;
9072 + __u32 lo_init[2];
9073 + uid_t lo_key_owner; /* Who set the key */
9074 ++ xid_t lo_xid;
9075 + int (*ioctl)(struct loop_device *, int cmd,
9076 + unsigned long arg);
9077 +
9078 +diff -NurpP --minimal linux-2.6.29.4/include/linux/magic.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/magic.h
9079 +--- linux-2.6.29.4/include/linux/magic.h 2009-03-24 14:22:42.000000000 +0100
9080 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/magic.h 2009-03-24 14:48:36.000000000 +0100
9081 +@@ -3,7 +3,7 @@
9082 +
9083 + #define ADFS_SUPER_MAGIC 0xadf5
9084 + #define AFFS_SUPER_MAGIC 0xadff
9085 +-#define AFS_SUPER_MAGIC 0x5346414F
9086 ++#define AFS_SUPER_MAGIC 0x5346414F
9087 + #define AUTOFS_SUPER_MAGIC 0x0187
9088 + #define CODA_SUPER_MAGIC 0x73757245
9089 + #define DEBUGFS_MAGIC 0x64626720
9090 +@@ -33,6 +33,7 @@
9091 + #define NFS_SUPER_MAGIC 0x6969
9092 + #define OPENPROM_SUPER_MAGIC 0x9fa1
9093 + #define PROC_SUPER_MAGIC 0x9fa0
9094 ++#define DEVPTS_SUPER_MAGIC 0x1cd1
9095 + #define QNX4_SUPER_MAGIC 0x002f /* qnx4 fs detection */
9096 +
9097 + #define REISERFS_SUPER_MAGIC 0x52654973 /* used by gcc */
9098 +diff -NurpP --minimal linux-2.6.29.4/include/linux/major.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/major.h
9099 +--- linux-2.6.29.4/include/linux/major.h 2008-12-25 00:26:37.000000000 +0100
9100 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/major.h 2009-02-22 22:54:26.000000000 +0100
9101 +@@ -15,6 +15,7 @@
9102 + #define HD_MAJOR IDE0_MAJOR
9103 + #define PTY_SLAVE_MAJOR 3
9104 + #define TTY_MAJOR 4
9105 ++#define VROOT_MAJOR 4
9106 + #define TTYAUX_MAJOR 5
9107 + #define LP_MAJOR 6
9108 + #define VCS_MAJOR 7
9109 +diff -NurpP --minimal linux-2.6.29.4/include/linux/mm_types.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/mm_types.h
9110 +--- linux-2.6.29.4/include/linux/mm_types.h 2009-03-24 14:22:42.000000000 +0100
9111 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/mm_types.h 2009-03-24 14:48:36.000000000 +0100
9112 +@@ -232,6 +232,7 @@ struct mm_struct {
9113 +
9114 + /* Architecture-specific MM context */
9115 + mm_context_t context;
9116 ++ struct vx_info *mm_vx_info;
9117 +
9118 + /* Swap token stuff */
9119 + /*
9120 +diff -NurpP --minimal linux-2.6.29.4/include/linux/mount.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/mount.h
9121 +--- linux-2.6.29.4/include/linux/mount.h 2008-12-25 00:26:37.000000000 +0100
9122 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/mount.h 2009-02-22 22:54:26.000000000 +0100
9123 +@@ -35,6 +35,9 @@ struct mnt_namespace;
9124 + #define MNT_UNBINDABLE 0x2000 /* if the vfsmount is a unbindable mount */
9125 + #define MNT_PNODE_MASK 0x3000 /* propagation flag mask */
9126 +
9127 ++#define MNT_TAGID 0x10000
9128 ++#define MNT_NOTAG 0x20000
9129 ++
9130 + struct vfsmount {
9131 + struct list_head mnt_hash;
9132 + struct vfsmount *mnt_parent; /* fs we are mounted on */
9133 +@@ -69,6 +72,7 @@ struct vfsmount {
9134 + * are held, and all mnt_writer[]s on this mount have 0 as their ->count
9135 + */
9136 + atomic_t __mnt_writers;
9137 ++ tag_t mnt_tag; /* tagging used for vfsmount */
9138 + };
9139 +
9140 + static inline struct vfsmount *mntget(struct vfsmount *mnt)
9141 +diff -NurpP --minimal linux-2.6.29.4/include/linux/net.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/net.h
9142 +--- linux-2.6.29.4/include/linux/net.h 2008-12-25 00:26:37.000000000 +0100
9143 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/net.h 2009-02-22 22:54:26.000000000 +0100
9144 +@@ -68,6 +68,7 @@ struct net;
9145 + #define SOCK_NOSPACE 2
9146 + #define SOCK_PASSCRED 3
9147 + #define SOCK_PASSSEC 4
9148 ++#define SOCK_USER_SOCKET 5
9149 +
9150 + #ifndef ARCH_HAS_SOCKET_TYPES
9151 + /**
9152 +diff -NurpP --minimal linux-2.6.29.4/include/linux/nfs_mount.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/nfs_mount.h
9153 +--- linux-2.6.29.4/include/linux/nfs_mount.h 2009-03-24 14:22:43.000000000 +0100
9154 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/nfs_mount.h 2009-03-24 14:48:36.000000000 +0100
9155 +@@ -63,7 +63,8 @@ struct nfs_mount_data {
9156 + #define NFS_MOUNT_SECFLAVOUR 0x2000 /* 5 */
9157 + #define NFS_MOUNT_NORDIRPLUS 0x4000 /* 5 */
9158 + #define NFS_MOUNT_UNSHARED 0x8000 /* 5 */
9159 +-#define NFS_MOUNT_FLAGMASK 0xFFFF
9160 ++#define NFS_MOUNT_TAGGED 0x10000 /* context tagging */
9161 ++#define NFS_MOUNT_FLAGMASK 0x1FFFF
9162 +
9163 + /* The following are for internal use only */
9164 + #define NFS_MOUNT_LOOKUP_CACHE_NONEG 0x10000
9165 +diff -NurpP --minimal linux-2.6.29.4/include/linux/nsproxy.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/nsproxy.h
9166 +--- linux-2.6.29.4/include/linux/nsproxy.h 2009-03-24 14:22:43.000000000 +0100
9167 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/nsproxy.h 2009-03-24 14:48:36.000000000 +0100
9168 +@@ -3,6 +3,7 @@
9169 +
9170 + #include <linux/spinlock.h>
9171 + #include <linux/sched.h>
9172 ++#include <linux/vserver/debug.h>
9173 +
9174 + struct mnt_namespace;
9175 + struct uts_namespace;
9176 +@@ -62,22 +63,33 @@ static inline struct nsproxy *task_nspro
9177 + }
9178 +
9179 + int copy_namespaces(unsigned long flags, struct task_struct *tsk);
9180 ++struct nsproxy *copy_nsproxy(struct nsproxy *orig);
9181 + void exit_task_namespaces(struct task_struct *tsk);
9182 + void switch_task_namespaces(struct task_struct *tsk, struct nsproxy *new);
9183 + void free_nsproxy(struct nsproxy *ns);
9184 + int unshare_nsproxy_namespaces(unsigned long, struct nsproxy **,
9185 + struct fs_struct *);
9186 +
9187 +-static inline void put_nsproxy(struct nsproxy *ns)
9188 ++#define get_nsproxy(n) __get_nsproxy(n, __FILE__, __LINE__)
9189 ++
9190 ++static inline void __get_nsproxy(struct nsproxy *ns,
9191 ++ const char *_file, int _line)
9192 + {
9193 +- if (atomic_dec_and_test(&ns->count)) {
9194 +- free_nsproxy(ns);
9195 +- }
9196 ++ vxlprintk(VXD_CBIT(space, 0), "get_nsproxy(%p[%u])",
9197 ++ ns, atomic_read(&ns->count), _file, _line);
9198 ++ atomic_inc(&ns->count);
9199 + }
9200 +
9201 +-static inline void get_nsproxy(struct nsproxy *ns)
9202 ++#define put_nsproxy(n) __put_nsproxy(n, __FILE__, __LINE__)
9203 ++
9204 ++static inline void __put_nsproxy(struct nsproxy *ns,
9205 ++ const char *_file, int _line)
9206 + {
9207 +- atomic_inc(&ns->count);
9208 ++ vxlprintk(VXD_CBIT(space, 0), "put_nsproxy(%p[%u])",
9209 ++ ns, atomic_read(&ns->count), _file, _line);
9210 ++ if (atomic_dec_and_test(&ns->count)) {
9211 ++ free_nsproxy(ns);
9212 ++ }
9213 + }
9214 +
9215 + #ifdef CONFIG_CGROUP_NS
9216 +diff -NurpP --minimal linux-2.6.29.4/include/linux/pid.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/pid.h
9217 +--- linux-2.6.29.4/include/linux/pid.h 2009-03-24 14:22:43.000000000 +0100
9218 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/pid.h 2009-03-24 14:48:36.000000000 +0100
9219 +@@ -8,7 +8,8 @@ enum pid_type
9220 + PIDTYPE_PID,
9221 + PIDTYPE_PGID,
9222 + PIDTYPE_SID,
9223 +- PIDTYPE_MAX
9224 ++ PIDTYPE_MAX,
9225 ++ PIDTYPE_REALPID
9226 + };
9227 +
9228 + /*
9229 +@@ -160,6 +161,7 @@ static inline pid_t pid_nr(struct pid *p
9230 + }
9231 +
9232 + pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns);
9233 ++pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns);
9234 + pid_t pid_vnr(struct pid *pid);
9235 +
9236 + #define do_each_pid_task(pid, type, task) \
9237 +diff -NurpP --minimal linux-2.6.29.4/include/linux/proc_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/proc_fs.h
9238 +--- linux-2.6.29.4/include/linux/proc_fs.h 2008-12-25 00:26:37.000000000 +0100
9239 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/proc_fs.h 2009-02-22 22:54:26.000000000 +0100
9240 +@@ -59,6 +59,7 @@ struct proc_dir_entry {
9241 + nlink_t nlink;
9242 + uid_t uid;
9243 + gid_t gid;
9244 ++ int vx_flags;
9245 + loff_t size;
9246 + const struct inode_operations *proc_iops;
9247 + /*
9248 +@@ -268,12 +269,18 @@ static inline void kclist_add(struct kco
9249 + extern void kclist_add(struct kcore_list *, void *, size_t);
9250 + #endif
9251 +
9252 ++struct vx_info;
9253 ++struct nx_info;
9254 ++
9255 + union proc_op {
9256 + int (*proc_get_link)(struct inode *, struct path *);
9257 + int (*proc_read)(struct task_struct *task, char *page);
9258 + int (*proc_show)(struct seq_file *m,
9259 + struct pid_namespace *ns, struct pid *pid,
9260 + struct task_struct *task);
9261 ++ int (*proc_vs_read)(char *page);
9262 ++ int (*proc_vxi_read)(struct vx_info *vxi, char *page);
9263 ++ int (*proc_nxi_read)(struct nx_info *nxi, char *page);
9264 + };
9265 +
9266 + struct ctl_table_header;
9267 +@@ -281,6 +288,7 @@ struct ctl_table;
9268 +
9269 + struct proc_inode {
9270 + struct pid *pid;
9271 ++ int vx_flags;
9272 + int fd;
9273 + union proc_op op;
9274 + struct proc_dir_entry *pde;
9275 +diff -NurpP --minimal linux-2.6.29.4/include/linux/reiserfs_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/reiserfs_fs.h
9276 +--- linux-2.6.29.4/include/linux/reiserfs_fs.h 2008-12-25 00:26:37.000000000 +0100
9277 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/reiserfs_fs.h 2009-02-22 22:54:26.000000000 +0100
9278 +@@ -837,6 +837,11 @@ struct stat_data_v1 {
9279 + #define REISERFS_COMPR_FL FS_COMPR_FL
9280 + #define REISERFS_NOTAIL_FL FS_NOTAIL_FL
9281 +
9282 ++/* unfortunately reiserfs sdattr is only 16 bit */
9283 ++#define REISERFS_IXUNLINK_FL (FS_IXUNLINK_FL >> 16)
9284 ++#define REISERFS_BARRIER_FL (FS_BARRIER_FL >> 16)
9285 ++#define REISERFS_COW_FL (FS_COW_FL >> 16)
9286 ++
9287 + /* persistent flags that file inherits from the parent directory */
9288 + #define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL | \
9289 + REISERFS_SYNC_FL | \
9290 +@@ -846,6 +851,9 @@ struct stat_data_v1 {
9291 + REISERFS_COMPR_FL | \
9292 + REISERFS_NOTAIL_FL )
9293 +
9294 ++#define REISERFS_FL_USER_VISIBLE 0x80FF
9295 ++#define REISERFS_FL_USER_MODIFIABLE 0x80FF
9296 ++
9297 + /* Stat Data on disk (reiserfs version of UFS disk inode minus the
9298 + address blocks) */
9299 + struct stat_data {
9300 +@@ -1911,6 +1919,7 @@ static inline void reiserfs_update_sd(st
9301 + void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
9302 + void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
9303 + int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);
9304 ++int reiserfs_sync_flags(struct inode *inode);
9305 +
9306 + /* namei.c */
9307 + void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
9308 +diff -NurpP --minimal linux-2.6.29.4/include/linux/reiserfs_fs_sb.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/reiserfs_fs_sb.h
9309 +--- linux-2.6.29.4/include/linux/reiserfs_fs_sb.h 2008-12-25 00:26:37.000000000 +0100
9310 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/reiserfs_fs_sb.h 2009-02-22 22:54:26.000000000 +0100
9311 +@@ -456,6 +456,7 @@ enum reiserfs_mount_options {
9312 + REISERFS_POSIXACL,
9313 + REISERFS_BARRIER_NONE,
9314 + REISERFS_BARRIER_FLUSH,
9315 ++ REISERFS_TAGGED,
9316 +
9317 + /* Actions on error */
9318 + REISERFS_ERROR_PANIC,
9319 +diff -NurpP --minimal linux-2.6.29.4/include/linux/sched.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/sched.h
9320 +--- linux-2.6.29.4/include/linux/sched.h 2009-05-23 23:16:53.000000000 +0200
9321 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/sched.h 2009-04-30 12:14:53.000000000 +0200
9322 +@@ -71,7 +71,6 @@ struct sched_param {
9323 + #include <linux/fs_struct.h>
9324 + #include <linux/compiler.h>
9325 + #include <linux/completion.h>
9326 +-#include <linux/pid.h>
9327 + #include <linux/percpu.h>
9328 + #include <linux/topology.h>
9329 + #include <linux/proportions.h>
9330 +@@ -88,6 +87,7 @@ struct sched_param {
9331 + #include <linux/kobject.h>
9332 + #include <linux/latencytop.h>
9333 + #include <linux/cred.h>
9334 ++#include <linux/pid.h>
9335 +
9336 + #include <asm/processor.h>
9337 +
9338 +@@ -176,12 +176,13 @@ extern unsigned long long time_sync_thre
9339 + #define TASK_UNINTERRUPTIBLE 2
9340 + #define __TASK_STOPPED 4
9341 + #define __TASK_TRACED 8
9342 ++#define TASK_ONHOLD 16
9343 + /* in tsk->exit_state */
9344 +-#define EXIT_ZOMBIE 16
9345 +-#define EXIT_DEAD 32
9346 ++#define EXIT_ZOMBIE 32
9347 ++#define EXIT_DEAD 64
9348 + /* in tsk->state again */
9349 +-#define TASK_DEAD 64
9350 +-#define TASK_WAKEKILL 128
9351 ++#define TASK_DEAD 128
9352 ++#define TASK_WAKEKILL 256
9353 +
9354 + /* Convenience macros for the sake of set_task_state */
9355 + #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
9356 +@@ -359,25 +360,28 @@ extern void arch_unmap_area_topdown(stru
9357 + * The mm counters are not protected by its page_table_lock,
9358 + * so must be incremented atomically.
9359 + */
9360 +-#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
9361 +-#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
9362 +-#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
9363 +-#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
9364 +-#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
9365 ++#define __set_mm_counter(mm, member, value) \
9366 ++ atomic_long_set(&(mm)->_##member, value)
9367 ++#define get_mm_counter(mm, member) \
9368 ++ ((unsigned long)atomic_long_read(&(mm)->_##member))
9369 +
9370 + #else /* !USE_SPLIT_PTLOCKS */
9371 + /*
9372 + * The mm counters are protected by its page_table_lock,
9373 + * so can be incremented directly.
9374 + */
9375 +-#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
9376 ++#define __set_mm_counter(mm, member, value) (mm)->_##member = (value)
9377 + #define get_mm_counter(mm, member) ((mm)->_##member)
9378 +-#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
9379 +-#define inc_mm_counter(mm, member) (mm)->_##member++
9380 +-#define dec_mm_counter(mm, member) (mm)->_##member--
9381 +
9382 + #endif /* !USE_SPLIT_PTLOCKS */
9383 +
9384 ++#define set_mm_counter(mm, member, value) \
9385 ++ vx_ ## member ## pages_sub((mm), (get_mm_counter(mm, member) - value))
9386 ++#define add_mm_counter(mm, member, value) \
9387 ++ vx_ ## member ## pages_add((mm), (value))
9388 ++#define inc_mm_counter(mm, member) vx_ ## member ## pages_inc((mm))
9389 ++#define dec_mm_counter(mm, member) vx_ ## member ## pages_dec((mm))
9390 ++
9391 + #define get_mm_rss(mm) \
9392 + (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
9393 + #define update_hiwater_rss(mm) do { \
9394 +@@ -1132,7 +1136,9 @@ struct task_struct {
9395 + const struct sched_class *sched_class;
9396 + struct sched_entity se;
9397 + struct sched_rt_entity rt;
9398 +-
9399 ++#ifdef CONFIG_VSERVER_HARDCPU
9400 ++ struct list_head hq;
9401 ++#endif
9402 + #ifdef CONFIG_PREEMPT_NOTIFIERS
9403 + /* list of struct preempt_notifier: */
9404 + struct hlist_head preempt_notifiers;
9405 +@@ -1288,6 +1294,14 @@ struct task_struct {
9406 + #endif
9407 + seccomp_t seccomp;
9408 +
9409 ++/* vserver context data */
9410 ++ struct vx_info *vx_info;
9411 ++ struct nx_info *nx_info;
9412 ++
9413 ++ xid_t xid;
9414 ++ nid_t nid;
9415 ++ tag_t tag;
9416 ++
9417 + /* Thread group tracking */
9418 + u32 parent_exec_id;
9419 + u32 self_exec_id;
9420 +@@ -1500,6 +1514,11 @@ struct pid_namespace;
9421 + * see also pid_nr() etc in include/linux/pid.h
9422 + */
9423 +
9424 ++#include <linux/vserver/base.h>
9425 ++#include <linux/vserver/context.h>
9426 ++#include <linux/vserver/debug.h>
9427 ++#include <linux/vserver/pid.h>
9428 ++
9429 + static inline pid_t task_pid_nr(struct task_struct *tsk)
9430 + {
9431 + return tsk->pid;
9432 +@@ -1509,7 +1528,7 @@ pid_t task_pid_nr_ns(struct task_struct
9433 +
9434 + static inline pid_t task_pid_vnr(struct task_struct *tsk)
9435 + {
9436 +- return pid_vnr(task_pid(tsk));
9437 ++ return vx_map_pid(pid_vnr(task_pid(tsk)));
9438 + }
9439 +
9440 +
9441 +@@ -1522,7 +1541,7 @@ pid_t task_tgid_nr_ns(struct task_struct
9442 +
9443 + static inline pid_t task_tgid_vnr(struct task_struct *tsk)
9444 + {
9445 +- return pid_vnr(task_tgid(tsk));
9446 ++ return vx_map_tgid(pid_vnr(task_tgid(tsk)));
9447 + }
9448 +
9449 +
9450 +diff -NurpP --minimal linux-2.6.29.4/include/linux/shmem_fs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/shmem_fs.h
9451 +--- linux-2.6.29.4/include/linux/shmem_fs.h 2008-12-25 00:26:37.000000000 +0100
9452 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/shmem_fs.h 2009-02-22 22:54:26.000000000 +0100
9453 +@@ -8,6 +8,9 @@
9454 +
9455 + #define SHMEM_NR_DIRECT 16
9456 +
9457 ++#define TMPFS_SUPER_MAGIC 0x01021994
9458 ++
9459 ++
9460 + struct shmem_inode_info {
9461 + spinlock_t lock;
9462 + unsigned long flags;
9463 +diff -NurpP --minimal linux-2.6.29.4/include/linux/stat.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/stat.h
9464 +--- linux-2.6.29.4/include/linux/stat.h 2008-12-25 00:26:37.000000000 +0100
9465 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/stat.h 2009-02-22 22:54:26.000000000 +0100
9466 +@@ -66,6 +66,7 @@ struct kstat {
9467 + unsigned int nlink;
9468 + uid_t uid;
9469 + gid_t gid;
9470 ++ tag_t tag;
9471 + dev_t rdev;
9472 + loff_t size;
9473 + struct timespec atime;
9474 +diff -NurpP --minimal linux-2.6.29.4/include/linux/sunrpc/auth.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/sunrpc/auth.h
9475 +--- linux-2.6.29.4/include/linux/sunrpc/auth.h 2008-12-25 00:26:37.000000000 +0100
9476 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/sunrpc/auth.h 2009-02-22 22:54:26.000000000 +0100
9477 +@@ -25,6 +25,7 @@
9478 + struct auth_cred {
9479 + uid_t uid;
9480 + gid_t gid;
9481 ++ tag_t tag;
9482 + struct group_info *group_info;
9483 + unsigned char machine_cred : 1;
9484 + };
9485 +diff -NurpP --minimal linux-2.6.29.4/include/linux/sunrpc/clnt.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/sunrpc/clnt.h
9486 +--- linux-2.6.29.4/include/linux/sunrpc/clnt.h 2009-03-24 14:22:43.000000000 +0100
9487 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/sunrpc/clnt.h 2009-03-24 14:48:36.000000000 +0100
9488 +@@ -43,7 +43,8 @@ struct rpc_clnt {
9489 + unsigned int cl_softrtry : 1,/* soft timeouts */
9490 + cl_discrtry : 1,/* disconnect before retry */
9491 + cl_autobind : 1,/* use getport() */
9492 +- cl_chatty : 1;/* be verbose */
9493 ++ cl_chatty : 1,/* be verbose */
9494 ++ cl_tag : 1;/* context tagging */
9495 +
9496 + struct rpc_rtt * cl_rtt; /* RTO estimator data */
9497 + const struct rpc_timeout *cl_timeout; /* Timeout strategy */
9498 +diff -NurpP --minimal linux-2.6.29.4/include/linux/syscalls.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/syscalls.h
9499 +--- linux-2.6.29.4/include/linux/syscalls.h 2009-03-24 14:22:43.000000000 +0100
9500 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/syscalls.h 2009-03-24 14:48:36.000000000 +0100
9501 +@@ -369,6 +369,8 @@ asmlinkage long sys_symlink(const char _
9502 + asmlinkage long sys_unlink(const char __user *pathname);
9503 + asmlinkage long sys_rename(const char __user *oldname,
9504 + const char __user *newname);
9505 ++asmlinkage long sys_copyfile(const char __user *from, const char __user *to,
9506 ++ umode_t mode);
9507 + asmlinkage long sys_chmod(const char __user *filename, mode_t mode);
9508 + asmlinkage long sys_fchmod(unsigned int fd, mode_t mode);
9509 +
9510 +diff -NurpP --minimal linux-2.6.29.4/include/linux/sysctl.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/sysctl.h
9511 +--- linux-2.6.29.4/include/linux/sysctl.h 2008-12-25 00:26:37.000000000 +0100
9512 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/sysctl.h 2009-02-22 22:54:26.000000000 +0100
9513 +@@ -70,6 +70,7 @@ enum
9514 + CTL_ABI=9, /* Binary emulation */
9515 + CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
9516 + CTL_ARLAN=254, /* arlan wireless driver */
9517 ++ CTL_VSERVER=4242, /* Linux-VServer debug */
9518 + CTL_S390DBF=5677, /* s390 debug */
9519 + CTL_SUNRPC=7249, /* sunrpc debug */
9520 + CTL_PM=9899, /* frv power management */
9521 +@@ -104,6 +105,7 @@ enum
9522 +
9523 + KERN_PANIC=15, /* int: panic timeout */
9524 + KERN_REALROOTDEV=16, /* real root device to mount after initrd */
9525 ++ KERN_VSHELPER=17, /* string: path to vshelper policy agent */
9526 +
9527 + KERN_SPARC_REBOOT=21, /* reboot command on Sparc */
9528 + KERN_CTLALTDEL=22, /* int: allow ctl-alt-del to reboot */
9529 +diff -NurpP --minimal linux-2.6.29.4/include/linux/sysfs.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/sysfs.h
9530 +--- linux-2.6.29.4/include/linux/sysfs.h 2008-12-25 00:26:37.000000000 +0100
9531 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/sysfs.h 2009-02-22 22:54:26.000000000 +0100
9532 +@@ -17,6 +17,8 @@
9533 + #include <linux/list.h>
9534 + #include <asm/atomic.h>
9535 +
9536 ++#define SYSFS_SUPER_MAGIC 0x62656572
9537 ++
9538 + struct kobject;
9539 + struct module;
9540 +
9541 +diff -NurpP --minimal linux-2.6.29.4/include/linux/time.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/time.h
9542 +--- linux-2.6.29.4/include/linux/time.h 2009-03-24 14:22:43.000000000 +0100
9543 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/time.h 2009-02-22 22:54:26.000000000 +0100
9544 +@@ -190,6 +190,9 @@ static __always_inline void timespec_add
9545 + a->tv_sec += __iter_div_u64_rem(a->tv_nsec + ns, NSEC_PER_SEC, &ns);
9546 + a->tv_nsec = ns;
9547 + }
9548 ++
9549 ++#include <linux/vs_time.h>
9550 ++
9551 + #endif /* __KERNEL__ */
9552 +
9553 + #define NFDBITS __NFDBITS
9554 +diff -NurpP --minimal linux-2.6.29.4/include/linux/types.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/types.h
9555 +--- linux-2.6.29.4/include/linux/types.h 2009-03-24 14:22:43.000000000 +0100
9556 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/types.h 2009-03-24 14:48:36.000000000 +0100
9557 +@@ -36,6 +36,9 @@ typedef __kernel_uid32_t uid_t;
9558 + typedef __kernel_gid32_t gid_t;
9559 + typedef __kernel_uid16_t uid16_t;
9560 + typedef __kernel_gid16_t gid16_t;
9561 ++typedef unsigned int xid_t;
9562 ++typedef unsigned int nid_t;
9563 ++typedef unsigned int tag_t;
9564 +
9565 + typedef unsigned long uintptr_t;
9566 +
9567 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vroot.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vroot.h
9568 +--- linux-2.6.29.4/include/linux/vroot.h 1970-01-01 01:00:00.000000000 +0100
9569 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vroot.h 2009-02-22 22:54:26.000000000 +0100
9570 +@@ -0,0 +1,51 @@
9571 ++
9572 ++/*
9573 ++ * include/linux/vroot.h
9574 ++ *
9575 ++ * written by Herbert Pötzl, 9/11/2002
9576 ++ * ported to 2.6 by Herbert Pötzl, 30/12/2004
9577 ++ *
9578 ++ * Copyright (C) 2002-2007 by Herbert Pötzl.
9579 ++ * Redistribution of this file is permitted under the
9580 ++ * GNU General Public License.
9581 ++ */
9582 ++
9583 ++#ifndef _LINUX_VROOT_H
9584 ++#define _LINUX_VROOT_H
9585 ++
9586 ++
9587 ++#ifdef __KERNEL__
9588 ++
9589 ++/* Possible states of device */
9590 ++enum {
9591 ++ Vr_unbound,
9592 ++ Vr_bound,
9593 ++};
9594 ++
9595 ++struct vroot_device {
9596 ++ int vr_number;
9597 ++ int vr_refcnt;
9598 ++
9599 ++ struct semaphore vr_ctl_mutex;
9600 ++ struct block_device *vr_device;
9601 ++ int vr_state;
9602 ++};
9603 ++
9604 ++
9605 ++typedef struct block_device *(vroot_grb_func)(struct block_device *);
9606 ++
9607 ++extern int register_vroot_grb(vroot_grb_func *);
9608 ++extern int unregister_vroot_grb(vroot_grb_func *);
9609 ++
9610 ++#endif /* __KERNEL__ */
9611 ++
9612 ++#define MAX_VROOT_DEFAULT 8
9613 ++
9614 ++/*
9615 ++ * IOCTL commands --- we will commandeer 0x56 ('V')
9616 ++ */
9617 ++
9618 ++#define VROOT_SET_DEV 0x5600
9619 ++#define VROOT_CLR_DEV 0x5601
9620 ++
9621 ++#endif /* _LINUX_VROOT_H */
9622 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_base.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_base.h
9623 +--- linux-2.6.29.4/include/linux/vs_base.h 1970-01-01 01:00:00.000000000 +0100
9624 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_base.h 2009-02-22 22:54:26.000000000 +0100
9625 +@@ -0,0 +1,10 @@
9626 ++#ifndef _VS_BASE_H
9627 ++#define _VS_BASE_H
9628 ++
9629 ++#include "vserver/base.h"
9630 ++#include "vserver/check.h"
9631 ++#include "vserver/debug.h"
9632 ++
9633 ++#else
9634 ++#warning duplicate inclusion
9635 ++#endif
9636 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_context.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_context.h
9637 +--- linux-2.6.29.4/include/linux/vs_context.h 1970-01-01 01:00:00.000000000 +0100
9638 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_context.h 2009-02-22 22:54:26.000000000 +0100
9639 +@@ -0,0 +1,227 @@
9640 ++#ifndef _VS_CONTEXT_H
9641 ++#define _VS_CONTEXT_H
9642 ++
9643 ++#include "vserver/base.h"
9644 ++#include "vserver/check.h"
9645 ++#include "vserver/context.h"
9646 ++#include "vserver/history.h"
9647 ++#include "vserver/debug.h"
9648 ++
9649 ++#include <linux/sched.h>
9650 ++
9651 ++
9652 ++#define get_vx_info(i) __get_vx_info(i, __FILE__, __LINE__, __HERE__)
9653 ++
9654 ++static inline struct vx_info *__get_vx_info(struct vx_info *vxi,
9655 ++ const char *_file, int _line, void *_here)
9656 ++{
9657 ++ if (!vxi)
9658 ++ return NULL;
9659 ++
9660 ++ vxlprintk(VXD_CBIT(xid, 2), "get_vx_info(%p[#%d.%d])",
9661 ++ vxi, vxi ? vxi->vx_id : 0,
9662 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9663 ++ _file, _line);
9664 ++ __vxh_get_vx_info(vxi, _here);
9665 ++
9666 ++ atomic_inc(&vxi->vx_usecnt);
9667 ++ return vxi;
9668 ++}
9669 ++
9670 ++
9671 ++extern void free_vx_info(struct vx_info *);
9672 ++
9673 ++#define put_vx_info(i) __put_vx_info(i, __FILE__, __LINE__, __HERE__)
9674 ++
9675 ++static inline void __put_vx_info(struct vx_info *vxi,
9676 ++ const char *_file, int _line, void *_here)
9677 ++{
9678 ++ if (!vxi)
9679 ++ return;
9680 ++
9681 ++ vxlprintk(VXD_CBIT(xid, 2), "put_vx_info(%p[#%d.%d])",
9682 ++ vxi, vxi ? vxi->vx_id : 0,
9683 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9684 ++ _file, _line);
9685 ++ __vxh_put_vx_info(vxi, _here);
9686 ++
9687 ++ if (atomic_dec_and_test(&vxi->vx_usecnt))
9688 ++ free_vx_info(vxi);
9689 ++}
9690 ++
9691 ++
9692 ++#define init_vx_info(p, i) \
9693 ++ __init_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9694 ++
9695 ++static inline void __init_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9696 ++ const char *_file, int _line, void *_here)
9697 ++{
9698 ++ if (vxi) {
9699 ++ vxlprintk(VXD_CBIT(xid, 3),
9700 ++ "init_vx_info(%p[#%d.%d])",
9701 ++ vxi, vxi ? vxi->vx_id : 0,
9702 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9703 ++ _file, _line);
9704 ++ __vxh_init_vx_info(vxi, vxp, _here);
9705 ++
9706 ++ atomic_inc(&vxi->vx_usecnt);
9707 ++ }
9708 ++ *vxp = vxi;
9709 ++}
9710 ++
9711 ++
9712 ++#define set_vx_info(p, i) \
9713 ++ __set_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9714 ++
9715 ++static inline void __set_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9716 ++ const char *_file, int _line, void *_here)
9717 ++{
9718 ++ struct vx_info *vxo;
9719 ++
9720 ++ if (!vxi)
9721 ++ return;
9722 ++
9723 ++ vxlprintk(VXD_CBIT(xid, 3), "set_vx_info(%p[#%d.%d])",
9724 ++ vxi, vxi ? vxi->vx_id : 0,
9725 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9726 ++ _file, _line);
9727 ++ __vxh_set_vx_info(vxi, vxp, _here);
9728 ++
9729 ++ atomic_inc(&vxi->vx_usecnt);
9730 ++ vxo = xchg(vxp, vxi);
9731 ++ BUG_ON(vxo);
9732 ++}
9733 ++
9734 ++
9735 ++#define clr_vx_info(p) __clr_vx_info(p, __FILE__, __LINE__, __HERE__)
9736 ++
9737 ++static inline void __clr_vx_info(struct vx_info **vxp,
9738 ++ const char *_file, int _line, void *_here)
9739 ++{
9740 ++ struct vx_info *vxo;
9741 ++
9742 ++ vxo = xchg(vxp, NULL);
9743 ++ if (!vxo)
9744 ++ return;
9745 ++
9746 ++ vxlprintk(VXD_CBIT(xid, 3), "clr_vx_info(%p[#%d.%d])",
9747 ++ vxo, vxo ? vxo->vx_id : 0,
9748 ++ vxo ? atomic_read(&vxo->vx_usecnt) : 0,
9749 ++ _file, _line);
9750 ++ __vxh_clr_vx_info(vxo, vxp, _here);
9751 ++
9752 ++ if (atomic_dec_and_test(&vxo->vx_usecnt))
9753 ++ free_vx_info(vxo);
9754 ++}
9755 ++
9756 ++
9757 ++#define claim_vx_info(v, p) \
9758 ++ __claim_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9759 ++
9760 ++static inline void __claim_vx_info(struct vx_info *vxi,
9761 ++ struct task_struct *task,
9762 ++ const char *_file, int _line, void *_here)
9763 ++{
9764 ++ vxlprintk(VXD_CBIT(xid, 3), "claim_vx_info(%p[#%d.%d.%d]) %p",
9765 ++ vxi, vxi ? vxi->vx_id : 0,
9766 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9767 ++ vxi ? atomic_read(&vxi->vx_tasks) : 0,
9768 ++ task, _file, _line);
9769 ++ __vxh_claim_vx_info(vxi, task, _here);
9770 ++
9771 ++ atomic_inc(&vxi->vx_tasks);
9772 ++}
9773 ++
9774 ++
9775 ++extern void unhash_vx_info(struct vx_info *);
9776 ++
9777 ++#define release_vx_info(v, p) \
9778 ++ __release_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9779 ++
9780 ++static inline void __release_vx_info(struct vx_info *vxi,
9781 ++ struct task_struct *task,
9782 ++ const char *_file, int _line, void *_here)
9783 ++{
9784 ++ vxlprintk(VXD_CBIT(xid, 3), "release_vx_info(%p[#%d.%d.%d]) %p",
9785 ++ vxi, vxi ? vxi->vx_id : 0,
9786 ++ vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9787 ++ vxi ? atomic_read(&vxi->vx_tasks) : 0,
9788 ++ task, _file, _line);
9789 ++ __vxh_release_vx_info(vxi, task, _here);
9790 ++
9791 ++ might_sleep();
9792 ++
9793 ++ if (atomic_dec_and_test(&vxi->vx_tasks))
9794 ++ unhash_vx_info(vxi);
9795 ++}
9796 ++
9797 ++
9798 ++#define task_get_vx_info(p) \
9799 ++ __task_get_vx_info(p, __FILE__, __LINE__, __HERE__)
9800 ++
9801 ++static inline struct vx_info *__task_get_vx_info(struct task_struct *p,
9802 ++ const char *_file, int _line, void *_here)
9803 ++{
9804 ++ struct vx_info *vxi;
9805 ++
9806 ++ task_lock(p);
9807 ++ vxlprintk(VXD_CBIT(xid, 5), "task_get_vx_info(%p)",
9808 ++ p, _file, _line);
9809 ++ vxi = __get_vx_info(p->vx_info, _file, _line, _here);
9810 ++ task_unlock(p);
9811 ++ return vxi;
9812 ++}
9813 ++
9814 ++
9815 ++static inline void __wakeup_vx_info(struct vx_info *vxi)
9816 ++{
9817 ++ if (waitqueue_active(&vxi->vx_wait))
9818 ++ wake_up_interruptible(&vxi->vx_wait);
9819 ++}
9820 ++
9821 ++
9822 ++#define enter_vx_info(v, s) __enter_vx_info(v, s, __FILE__, __LINE__)
9823 ++
9824 ++static inline void __enter_vx_info(struct vx_info *vxi,
9825 ++ struct vx_info_save *vxis, const char *_file, int _line)
9826 ++{
9827 ++ vxlprintk(VXD_CBIT(xid, 5), "enter_vx_info(%p[#%d],%p) %p[#%d,%p]",
9828 ++ vxi, vxi ? vxi->vx_id : 0, vxis, current,
9829 ++ current->xid, current->vx_info, _file, _line);
9830 ++ vxis->vxi = xchg(&current->vx_info, vxi);
9831 ++ vxis->xid = current->xid;
9832 ++ current->xid = vxi ? vxi->vx_id : 0;
9833 ++}
9834 ++
9835 ++#define leave_vx_info(s) __leave_vx_info(s, __FILE__, __LINE__)
9836 ++
9837 ++static inline void __leave_vx_info(struct vx_info_save *vxis,
9838 ++ const char *_file, int _line)
9839 ++{
9840 ++ vxlprintk(VXD_CBIT(xid, 5), "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]",
9841 ++ vxis, vxis->xid, vxis->vxi, current,
9842 ++ current->xid, current->vx_info, _file, _line);
9843 ++ (void)xchg(&current->vx_info, vxis->vxi);
9844 ++ current->xid = vxis->xid;
9845 ++}
9846 ++
9847 ++
9848 ++static inline void __enter_vx_admin(struct vx_info_save *vxis)
9849 ++{
9850 ++ vxis->vxi = xchg(&current->vx_info, NULL);
9851 ++ vxis->xid = xchg(&current->xid, (xid_t)0);
9852 ++}
9853 ++
9854 ++static inline void __leave_vx_admin(struct vx_info_save *vxis)
9855 ++{
9856 ++ (void)xchg(&current->xid, vxis->xid);
9857 ++ (void)xchg(&current->vx_info, vxis->vxi);
9858 ++}
9859 ++
9860 ++extern void exit_vx_info(struct task_struct *, int);
9861 ++extern void exit_vx_info_early(struct task_struct *, int);
9862 ++
9863 ++
9864 ++#else
9865 ++#warning duplicate inclusion
9866 ++#endif
9867 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_cowbl.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_cowbl.h
9868 +--- linux-2.6.29.4/include/linux/vs_cowbl.h 1970-01-01 01:00:00.000000000 +0100
9869 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_cowbl.h 2009-02-22 22:54:26.000000000 +0100
9870 +@@ -0,0 +1,47 @@
9871 ++#ifndef _VS_COWBL_H
9872 ++#define _VS_COWBL_H
9873 ++
9874 ++#include <linux/fs.h>
9875 ++#include <linux/dcache.h>
9876 ++#include <linux/namei.h>
9877 ++
9878 ++extern struct dentry *cow_break_link(const char *pathname);
9879 ++
9880 ++static inline int cow_check_and_break(struct path *path)
9881 ++{
9882 ++ struct inode *inode = path->dentry->d_inode;
9883 ++ int error = 0;
9884 ++
9885 ++ /* do we need this check? */
9886 ++ if (IS_RDONLY(inode))
9887 ++ return -EROFS;
9888 ++
9889 ++ if (IS_COW(inode)) {
9890 ++ if (IS_COW_LINK(inode)) {
9891 ++ struct dentry *new_dentry, *old_dentry = path->dentry;
9892 ++ char *pp, *buf;
9893 ++
9894 ++ buf = kmalloc(PATH_MAX, GFP_KERNEL);
9895 ++ if (!buf) {
9896 ++ return -ENOMEM;
9897 ++ }
9898 ++ pp = d_path(path, buf, PATH_MAX);
9899 ++ new_dentry = cow_break_link(pp);
9900 ++ kfree(buf);
9901 ++ if (!IS_ERR(new_dentry)) {
9902 ++ path->dentry = new_dentry;
9903 ++ dput(old_dentry);
9904 ++ } else
9905 ++ error = PTR_ERR(new_dentry);
9906 ++ } else {
9907 ++ inode->i_flags &= ~(S_IXUNLINK | S_IMMUTABLE);
9908 ++ inode->i_ctime = CURRENT_TIME;
9909 ++ mark_inode_dirty(inode);
9910 ++ }
9911 ++ }
9912 ++ return error;
9913 ++}
9914 ++
9915 ++#else
9916 ++#warning duplicate inclusion
9917 ++#endif
9918 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_cvirt.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_cvirt.h
9919 +--- linux-2.6.29.4/include/linux/vs_cvirt.h 1970-01-01 01:00:00.000000000 +0100
9920 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_cvirt.h 2009-02-22 22:54:26.000000000 +0100
9921 +@@ -0,0 +1,50 @@
9922 ++#ifndef _VS_CVIRT_H
9923 ++#define _VS_CVIRT_H
9924 ++
9925 ++#include "vserver/cvirt.h"
9926 ++#include "vserver/context.h"
9927 ++#include "vserver/base.h"
9928 ++#include "vserver/check.h"
9929 ++#include "vserver/debug.h"
9930 ++
9931 ++
9932 ++static inline void vx_activate_task(struct task_struct *p)
9933 ++{
9934 ++ struct vx_info *vxi;
9935 ++
9936 ++ if ((vxi = p->vx_info)) {
9937 ++ vx_update_load(vxi);
9938 ++ atomic_inc(&vxi->cvirt.nr_running);
9939 ++ }
9940 ++}
9941 ++
9942 ++static inline void vx_deactivate_task(struct task_struct *p)
9943 ++{
9944 ++ struct vx_info *vxi;
9945 ++
9946 ++ if ((vxi = p->vx_info)) {
9947 ++ vx_update_load(vxi);
9948 ++ atomic_dec(&vxi->cvirt.nr_running);
9949 ++ }
9950 ++}
9951 ++
9952 ++static inline void vx_uninterruptible_inc(struct task_struct *p)
9953 ++{
9954 ++ struct vx_info *vxi;
9955 ++
9956 ++ if ((vxi = p->vx_info))
9957 ++ atomic_inc(&vxi->cvirt.nr_uninterruptible);
9958 ++}
9959 ++
9960 ++static inline void vx_uninterruptible_dec(struct task_struct *p)
9961 ++{
9962 ++ struct vx_info *vxi;
9963 ++
9964 ++ if ((vxi = p->vx_info))
9965 ++ atomic_dec(&vxi->cvirt.nr_uninterruptible);
9966 ++}
9967 ++
9968 ++
9969 ++#else
9970 ++#warning duplicate inclusion
9971 ++#endif
9972 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_device.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_device.h
9973 +--- linux-2.6.29.4/include/linux/vs_device.h 1970-01-01 01:00:00.000000000 +0100
9974 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_device.h 2009-02-22 22:54:26.000000000 +0100
9975 +@@ -0,0 +1,45 @@
9976 ++#ifndef _VS_DEVICE_H
9977 ++#define _VS_DEVICE_H
9978 ++
9979 ++#include "vserver/base.h"
9980 ++#include "vserver/device.h"
9981 ++#include "vserver/debug.h"
9982 ++
9983 ++
9984 ++#ifdef CONFIG_VSERVER_DEVICE
9985 ++
9986 ++int vs_map_device(struct vx_info *, dev_t, dev_t *, umode_t);
9987 ++
9988 ++#define vs_device_perm(v, d, m, p) \
9989 ++ ((vs_map_device(current_vx_info(), d, NULL, m) & (p)) == (p))
9990 ++
9991 ++#else
9992 ++
9993 ++static inline
9994 ++int vs_map_device(struct vx_info *vxi,
9995 ++ dev_t device, dev_t *target, umode_t mode)
9996 ++{
9997 ++ if (target)
9998 ++ *target = device;
9999 ++ return ~0;
10000 ++}
10001 ++
10002 ++#define vs_device_perm(v, d, m, p) ((p) == (p))
10003 ++
10004 ++#endif
10005 ++
10006 ++
10007 ++#define vs_map_chrdev(d, t, p) \
10008 ++ ((vs_map_device(current_vx_info(), d, t, S_IFCHR) & (p)) == (p))
10009 ++#define vs_map_blkdev(d, t, p) \
10010 ++ ((vs_map_device(current_vx_info(), d, t, S_IFBLK) & (p)) == (p))
10011 ++
10012 ++#define vs_chrdev_perm(d, p) \
10013 ++ vs_device_perm(current_vx_info(), d, S_IFCHR, p)
10014 ++#define vs_blkdev_perm(d, p) \
10015 ++ vs_device_perm(current_vx_info(), d, S_IFBLK, p)
10016 ++
10017 ++
10018 ++#else
10019 ++#warning duplicate inclusion
10020 ++#endif
10021 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_dlimit.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_dlimit.h
10022 +--- linux-2.6.29.4/include/linux/vs_dlimit.h 1970-01-01 01:00:00.000000000 +0100
10023 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_dlimit.h 2009-02-22 22:54:26.000000000 +0100
10024 +@@ -0,0 +1,211 @@
10025 ++#ifndef _VS_DLIMIT_H
10026 ++#define _VS_DLIMIT_H
10027 ++
10028 ++#include <linux/fs.h>
10029 ++
10030 ++#include "vserver/dlimit.h"
10031 ++#include "vserver/base.h"
10032 ++#include "vserver/debug.h"
10033 ++
10034 ++
10035 ++#define get_dl_info(i) __get_dl_info(i, __FILE__, __LINE__)
10036 ++
10037 ++static inline struct dl_info *__get_dl_info(struct dl_info *dli,
10038 ++ const char *_file, int _line)
10039 ++{
10040 ++ if (!dli)
10041 ++ return NULL;
10042 ++ vxlprintk(VXD_CBIT(dlim, 4), "get_dl_info(%p[#%d.%d])",
10043 ++ dli, dli ? dli->dl_tag : 0,
10044 ++ dli ? atomic_read(&dli->dl_usecnt) : 0,
10045 ++ _file, _line);
10046 ++ atomic_inc(&dli->dl_usecnt);
10047 ++ return dli;
10048 ++}
10049 ++
10050 ++
10051 ++#define free_dl_info(i) \
10052 ++ call_rcu(&(i)->dl_rcu, rcu_free_dl_info)
10053 ++
10054 ++#define put_dl_info(i) __put_dl_info(i, __FILE__, __LINE__)
10055 ++
10056 ++static inline void __put_dl_info(struct dl_info *dli,
10057 ++ const char *_file, int _line)
10058 ++{
10059 ++ if (!dli)
10060 ++ return;
10061 ++ vxlprintk(VXD_CBIT(dlim, 4), "put_dl_info(%p[#%d.%d])",
10062 ++ dli, dli ? dli->dl_tag : 0,
10063 ++ dli ? atomic_read(&dli->dl_usecnt) : 0,
10064 ++ _file, _line);
10065 ++ if (atomic_dec_and_test(&dli->dl_usecnt))
10066 ++ free_dl_info(dli);
10067 ++}
10068 ++
10069 ++
10070 ++#define __dlimit_char(d) ((d) ? '*' : ' ')
10071 ++
10072 ++static inline int __dl_alloc_space(struct super_block *sb,
10073 ++ tag_t tag, dlsize_t nr, const char *file, int line)
10074 ++{
10075 ++ struct dl_info *dli = NULL;
10076 ++ int ret = 0;
10077 ++
10078 ++ if (nr == 0)
10079 ++ goto out;
10080 ++ dli = locate_dl_info(sb, tag);
10081 ++ if (!dli)
10082 ++ goto out;
10083 ++
10084 ++ spin_lock(&dli->dl_lock);
10085 ++ ret = (dli->dl_space_used + nr > dli->dl_space_total);
10086 ++ if (!ret)
10087 ++ dli->dl_space_used += nr;
10088 ++ spin_unlock(&dli->dl_lock);
10089 ++ put_dl_info(dli);
10090 ++out:
10091 ++ vxlprintk(VXD_CBIT(dlim, 1),
10092 ++ "ALLOC (%p,#%d)%c %lld bytes (%d)",
10093 ++ sb, tag, __dlimit_char(dli), (long long)nr,
10094 ++ ret, file, line);
10095 ++ return ret;
10096 ++}
10097 ++
10098 ++static inline void __dl_free_space(struct super_block *sb,
10099 ++ tag_t tag, dlsize_t nr, const char *_file, int _line)
10100 ++{
10101 ++ struct dl_info *dli = NULL;
10102 ++
10103 ++ if (nr == 0)
10104 ++ goto out;
10105 ++ dli = locate_dl_info(sb, tag);
10106 ++ if (!dli)
10107 ++ goto out;
10108 ++
10109 ++ spin_lock(&dli->dl_lock);
10110 ++ if (dli->dl_space_used > nr)
10111 ++ dli->dl_space_used -= nr;
10112 ++ else
10113 ++ dli->dl_space_used = 0;
10114 ++ spin_unlock(&dli->dl_lock);
10115 ++ put_dl_info(dli);
10116 ++out:
10117 ++ vxlprintk(VXD_CBIT(dlim, 1),
10118 ++ "FREE (%p,#%d)%c %lld bytes",
10119 ++ sb, tag, __dlimit_char(dli), (long long)nr,
10120 ++ _file, _line);
10121 ++}
10122 ++
10123 ++static inline int __dl_alloc_inode(struct super_block *sb,
10124 ++ tag_t tag, const char *_file, int _line)
10125 ++{
10126 ++ struct dl_info *dli;
10127 ++ int ret = 0;
10128 ++
10129 ++ dli = locate_dl_info(sb, tag);
10130 ++ if (!dli)
10131 ++ goto out;
10132 ++
10133 ++ spin_lock(&dli->dl_lock);
10134 ++ ret = (dli->dl_inodes_used >= dli->dl_inodes_total);
10135 ++ if (!ret)
10136 ++ dli->dl_inodes_used++;
10137 ++ spin_unlock(&dli->dl_lock);
10138 ++ put_dl_info(dli);
10139 ++out:
10140 ++ vxlprintk(VXD_CBIT(dlim, 0),
10141 ++ "ALLOC (%p,#%d)%c inode (%d)",
10142 ++ sb, tag, __dlimit_char(dli), ret, _file, _line);
10143 ++ return ret;
10144 ++}
10145 ++
10146 ++static inline void __dl_free_inode(struct super_block *sb,
10147 ++ tag_t tag, const char *_file, int _line)
10148 ++{
10149 ++ struct dl_info *dli;
10150 ++
10151 ++ dli = locate_dl_info(sb, tag);
10152 ++ if (!dli)
10153 ++ goto out;
10154 ++
10155 ++ spin_lock(&dli->dl_lock);
10156 ++ if (dli->dl_inodes_used > 1)
10157 ++ dli->dl_inodes_used--;
10158 ++ else
10159 ++ dli->dl_inodes_used = 0;
10160 ++ spin_unlock(&dli->dl_lock);
10161 ++ put_dl_info(dli);
10162 ++out:
10163 ++ vxlprintk(VXD_CBIT(dlim, 0),
10164 ++ "FREE (%p,#%d)%c inode",
10165 ++ sb, tag, __dlimit_char(dli), _file, _line);
10166 ++}
10167 ++
10168 ++static inline void __dl_adjust_block(struct super_block *sb, tag_t tag,
10169 ++ unsigned long long *free_blocks, unsigned long long *root_blocks,
10170 ++ const char *_file, int _line)
10171 ++{
10172 ++ struct dl_info *dli;
10173 ++ uint64_t broot, bfree;
10174 ++
10175 ++ dli = locate_dl_info(sb, tag);
10176 ++ if (!dli)
10177 ++ return;
10178 ++
10179 ++ spin_lock(&dli->dl_lock);
10180 ++ broot = (dli->dl_space_total -
10181 ++ (dli->dl_space_total >> 10) * dli->dl_nrlmult)
10182 ++ >> sb->s_blocksize_bits;
10183 ++ bfree = (dli->dl_space_total - dli->dl_space_used)
10184 ++ >> sb->s_blocksize_bits;
10185 ++ spin_unlock(&dli->dl_lock);
10186 ++
10187 ++ vxlprintk(VXD_CBIT(dlim, 2),
10188 ++ "ADJUST: %lld,%lld on %lld,%lld [mult=%d]",
10189 ++ (long long)bfree, (long long)broot,
10190 ++ *free_blocks, *root_blocks, dli->dl_nrlmult,
10191 ++ _file, _line);
10192 ++ if (free_blocks) {
10193 ++ if (*free_blocks > bfree)
10194 ++ *free_blocks = bfree;
10195 ++ }
10196 ++ if (root_blocks) {
10197 ++ if (*root_blocks > broot)
10198 ++ *root_blocks = broot;
10199 ++ }
10200 ++ put_dl_info(dli);
10201 ++}
10202 ++
10203 ++#define DLIMIT_ALLOC_SPACE(in, bytes) \
10204 ++ __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10205 ++ __FILE__, __LINE__ )
10206 ++
10207 ++#define DLIMIT_FREE_SPACE(in, bytes) \
10208 ++ __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10209 ++ __FILE__, __LINE__ )
10210 ++
10211 ++#define DLIMIT_ALLOC_BLOCK(in, nr) \
10212 ++ __dl_alloc_space((in)->i_sb, (in)->i_tag, \
10213 ++ ((dlsize_t)(nr)) << (in)->i_sb->s_blocksize_bits, \
10214 ++ __FILE__, __LINE__ )
10215 ++
10216 ++#define DLIMIT_FREE_BLOCK(in, nr) \
10217 ++ __dl_free_space((in)->i_sb, (in)->i_tag, \
10218 ++ ((dlsize_t)(nr)) << (in)->i_sb->s_blocksize_bits, \
10219 ++ __FILE__, __LINE__ )
10220 ++
10221 ++
10222 ++#define DLIMIT_ALLOC_INODE(in) \
10223 ++ __dl_alloc_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10224 ++
10225 ++#define DLIMIT_FREE_INODE(in) \
10226 ++ __dl_free_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10227 ++
10228 ++
10229 ++#define DLIMIT_ADJUST_BLOCK(sb, tag, fb, rb) \
10230 ++ __dl_adjust_block(sb, tag, fb, rb, __FILE__, __LINE__ )
10231 ++
10232 ++
10233 ++#else
10234 ++#warning duplicate inclusion
10235 ++#endif
10236 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/base.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/base.h
10237 +--- linux-2.6.29.4/include/linux/vserver/base.h 1970-01-01 01:00:00.000000000 +0100
10238 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/base.h 2009-03-25 00:36:15.000000000 +0100
10239 +@@ -0,0 +1,157 @@
10240 ++#ifndef _VX_BASE_H
10241 ++#define _VX_BASE_H
10242 ++
10243 ++
10244 ++/* context state changes */
10245 ++
10246 ++enum {
10247 ++ VSC_STARTUP = 1,
10248 ++ VSC_SHUTDOWN,
10249 ++
10250 ++ VSC_NETUP,
10251 ++ VSC_NETDOWN,
10252 ++};
10253 ++
10254 ++
10255 ++
10256 ++#define vx_task_xid(t) ((t)->xid)
10257 ++
10258 ++#define vx_current_xid() vx_task_xid(current)
10259 ++
10260 ++#define current_vx_info() (current->vx_info)
10261 ++
10262 ++
10263 ++#define nx_task_nid(t) ((t)->nid)
10264 ++
10265 ++#define nx_current_nid() nx_task_nid(current)
10266 ++
10267 ++#define current_nx_info() (current->nx_info)
10268 ++
10269 ++
10270 ++/* generic flag merging */
10271 ++
10272 ++#define vs_check_flags(v, m, f) (((v) & (m)) ^ (f))
10273 ++
10274 ++#define vs_mask_flags(v, f, m) (((v) & ~(m)) | ((f) & (m)))
10275 ++
10276 ++#define vs_mask_mask(v, f, m) (((v) & ~(m)) | ((v) & (f) & (m)))
10277 ++
10278 ++#define vs_check_bit(v, n) ((v) & (1LL << (n)))
10279 ++
10280 ++
10281 ++/* context flags */
10282 ++
10283 ++#define __vx_flags(v) ((v) ? (v)->vx_flags : 0)
10284 ++
10285 ++#define vx_current_flags() __vx_flags(current->vx_info)
10286 ++
10287 ++#define vx_info_flags(v, m, f) \
10288 ++ vs_check_flags(__vx_flags(v), m, f)
10289 ++
10290 ++#define task_vx_flags(t, m, f) \
10291 ++ ((t) && vx_info_flags((t)->vx_info, m, f))
10292 ++
10293 ++#define vx_flags(m, f) vx_info_flags(current->vx_info, m, f)
10294 ++
10295 ++
10296 ++/* context caps */
10297 ++
10298 ++#define __vx_ccaps(v) ((v) ? (v)->vx_ccaps : 0)
10299 ++
10300 ++#define vx_current_ccaps() __vx_ccaps(current->vx_info)
10301 ++
10302 ++#define vx_info_ccaps(v, c) (__vx_ccaps(v) & (c))
10303 ++
10304 ++#define vx_ccaps(c) vx_info_ccaps(current->vx_info, (c))
10305 ++
10306 ++
10307 ++
10308 ++/* network flags */
10309 ++
10310 ++#define __nx_flags(n) ((n) ? (n)->nx_flags : 0)
10311 ++
10312 ++#define nx_current_flags() __nx_flags(current->nx_info)
10313 ++
10314 ++#define nx_info_flags(n, m, f) \
10315 ++ vs_check_flags(__nx_flags(n), m, f)
10316 ++
10317 ++#define task_nx_flags(t, m, f) \
10318 ++ ((t) && nx_info_flags((t)->nx_info, m, f))
10319 ++
10320 ++#define nx_flags(m, f) nx_info_flags(current->nx_info, m, f)
10321 ++
10322 ++
10323 ++/* network caps */
10324 ++
10325 ++#define __nx_ncaps(n) ((n) ? (n)->nx_ncaps : 0)
10326 ++
10327 ++#define nx_current_ncaps() __nx_ncaps(current->nx_info)
10328 ++
10329 ++#define nx_info_ncaps(n, c) (__nx_ncaps(n) & (c))
10330 ++
10331 ++#define nx_ncaps(c) nx_info_ncaps(current->nx_info, c)
10332 ++
10333 ++
10334 ++/* context mask capabilities */
10335 ++
10336 ++#define __vx_mcaps(v) ((v) ? (v)->vx_ccaps >> 32UL : ~0 )
10337 ++
10338 ++#define vx_info_mcaps(v, c) (__vx_mcaps(v) & (c))
10339 ++
10340 ++#define vx_mcaps(c) vx_info_mcaps(current->vx_info, c)
10341 ++
10342 ++
10343 ++/* context bcap mask */
10344 ++
10345 ++#define __vx_bcaps(v) ((v)->vx_bcaps)
10346 ++
10347 ++#define vx_current_bcaps() __vx_bcaps(current->vx_info)
10348 ++
10349 ++
10350 ++/* mask given bcaps */
10351 ++
10352 ++#define vx_info_mbcaps(v, c) ((v) ? cap_intersect(__vx_bcaps(v), c) : c)
10353 ++
10354 ++#define vx_mbcaps(c) vx_info_mbcaps(current->vx_info, c)
10355 ++
10356 ++
10357 ++/* masked cap_bset */
10358 ++
10359 ++#define vx_info_cap_bset(v) vx_info_mbcaps(v, current->cap_bset)
10360 ++
10361 ++#define vx_current_cap_bset() vx_info_cap_bset(current->vx_info)
10362 ++
10363 ++#if 0
10364 ++#define vx_info_mbcap(v, b) \
10365 ++ (!vx_info_flags(v, VXF_STATE_SETUP, 0) ? \
10366 ++ vx_info_bcaps(v, b) : (b))
10367 ++
10368 ++#define task_vx_mbcap(t, b) \
10369 ++ vx_info_mbcap((t)->vx_info, (t)->b)
10370 ++
10371 ++#define vx_mbcap(b) task_vx_mbcap(current, b)
10372 ++#endif
10373 ++
10374 ++#define vx_cap_raised(v, c, f) cap_raised(vx_info_mbcaps(v, c), f)
10375 ++
10376 ++#define vx_capable(b, c) (capable(b) || \
10377 ++ (cap_raised(current_cap(), b) && vx_ccaps(c)))
10378 ++
10379 ++#define nx_capable(b, c) (capable(b) || \
10380 ++ (cap_raised(current_cap(), b) && nx_ncaps(c)))
10381 ++
10382 ++#define vx_current_initpid(n) \
10383 ++ (current->vx_info && \
10384 ++ (current->vx_info->vx_initpid == (n)))
10385 ++
10386 ++
10387 ++#define __vx_state(v) ((v) ? ((v)->vx_state) : 0)
10388 ++
10389 ++#define vx_info_state(v, m) (__vx_state(v) & (m))
10390 ++
10391 ++
10392 ++#define __nx_state(n) ((n) ? ((n)->nx_state) : 0)
10393 ++
10394 ++#define nx_info_state(n, m) (__nx_state(n) & (m))
10395 ++
10396 ++#endif
10397 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cacct_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_cmd.h
10398 +--- linux-2.6.29.4/include/linux/vserver/cacct_cmd.h 1970-01-01 01:00:00.000000000 +0100
10399 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_cmd.h 2009-02-22 22:54:26.000000000 +0100
10400 +@@ -0,0 +1,23 @@
10401 ++#ifndef _VX_CACCT_CMD_H
10402 ++#define _VX_CACCT_CMD_H
10403 ++
10404 ++
10405 ++/* virtual host info name commands */
10406 ++
10407 ++#define VCMD_sock_stat VC_CMD(VSTAT, 5, 0)
10408 ++
10409 ++struct vcmd_sock_stat_v0 {
10410 ++ uint32_t field;
10411 ++ uint32_t count[3];
10412 ++ uint64_t total[3];
10413 ++};
10414 ++
10415 ++
10416 ++#ifdef __KERNEL__
10417 ++
10418 ++#include <linux/compiler.h>
10419 ++
10420 ++extern int vc_sock_stat(struct vx_info *, void __user *);
10421 ++
10422 ++#endif /* __KERNEL__ */
10423 ++#endif /* _VX_CACCT_CMD_H */
10424 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cacct_def.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_def.h
10425 +--- linux-2.6.29.4/include/linux/vserver/cacct_def.h 1970-01-01 01:00:00.000000000 +0100
10426 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_def.h 2009-02-22 22:54:26.000000000 +0100
10427 +@@ -0,0 +1,43 @@
10428 ++#ifndef _VX_CACCT_DEF_H
10429 ++#define _VX_CACCT_DEF_H
10430 ++
10431 ++#include <asm/atomic.h>
10432 ++#include <linux/vserver/cacct.h>
10433 ++
10434 ++
10435 ++struct _vx_sock_acc {
10436 ++ atomic_long_t count;
10437 ++ atomic_long_t total;
10438 ++};
10439 ++
10440 ++/* context sub struct */
10441 ++
10442 ++struct _vx_cacct {
10443 ++ struct _vx_sock_acc sock[VXA_SOCK_SIZE][3];
10444 ++ atomic_t slab[8];
10445 ++ atomic_t page[6][8];
10446 ++};
10447 ++
10448 ++#ifdef CONFIG_VSERVER_DEBUG
10449 ++
10450 ++static inline void __dump_vx_cacct(struct _vx_cacct *cacct)
10451 ++{
10452 ++ int i, j;
10453 ++
10454 ++ printk("\t_vx_cacct:");
10455 ++ for (i = 0; i < 6; i++) {
10456 ++ struct _vx_sock_acc *ptr = cacct->sock[i];
10457 ++
10458 ++ printk("\t [%d] =", i);
10459 ++ for (j = 0; j < 3; j++) {
10460 ++ printk(" [%d] = %8lu, %8lu", j,
10461 ++ atomic_long_read(&ptr[j].count),
10462 ++ atomic_long_read(&ptr[j].total));
10463 ++ }
10464 ++ printk("\n");
10465 ++ }
10466 ++}
10467 ++
10468 ++#endif
10469 ++
10470 ++#endif /* _VX_CACCT_DEF_H */
10471 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cacct.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct.h
10472 +--- linux-2.6.29.4/include/linux/vserver/cacct.h 1970-01-01 01:00:00.000000000 +0100
10473 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct.h 2009-02-22 22:54:26.000000000 +0100
10474 +@@ -0,0 +1,15 @@
10475 ++#ifndef _VX_CACCT_H
10476 ++#define _VX_CACCT_H
10477 ++
10478 ++
10479 ++enum sock_acc_field {
10480 ++ VXA_SOCK_UNSPEC = 0,
10481 ++ VXA_SOCK_UNIX,
10482 ++ VXA_SOCK_INET,
10483 ++ VXA_SOCK_INET6,
10484 ++ VXA_SOCK_PACKET,
10485 ++ VXA_SOCK_OTHER,
10486 ++ VXA_SOCK_SIZE /* array size */
10487 ++};
10488 ++
10489 ++#endif /* _VX_CACCT_H */
10490 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cacct_int.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_int.h
10491 +--- linux-2.6.29.4/include/linux/vserver/cacct_int.h 1970-01-01 01:00:00.000000000 +0100
10492 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cacct_int.h 2009-02-22 22:54:26.000000000 +0100
10493 +@@ -0,0 +1,21 @@
10494 ++#ifndef _VX_CACCT_INT_H
10495 ++#define _VX_CACCT_INT_H
10496 ++
10497 ++
10498 ++#ifdef __KERNEL__
10499 ++
10500 ++static inline
10501 ++unsigned long vx_sock_count(struct _vx_cacct *cacct, int type, int pos)
10502 ++{
10503 ++ return atomic_long_read(&cacct->sock[type][pos].count);
10504 ++}
10505 ++
10506 ++
10507 ++static inline
10508 ++unsigned long vx_sock_total(struct _vx_cacct *cacct, int type, int pos)
10509 ++{
10510 ++ return atomic_long_read(&cacct->sock[type][pos].total);
10511 ++}
10512 ++
10513 ++#endif /* __KERNEL__ */
10514 ++#endif /* _VX_CACCT_INT_H */
10515 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/check.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/check.h
10516 +--- linux-2.6.29.4/include/linux/vserver/check.h 1970-01-01 01:00:00.000000000 +0100
10517 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/check.h 2009-02-22 22:54:26.000000000 +0100
10518 +@@ -0,0 +1,89 @@
10519 ++#ifndef _VS_CHECK_H
10520 ++#define _VS_CHECK_H
10521 ++
10522 ++
10523 ++#define MAX_S_CONTEXT 65535 /* Arbitrary limit */
10524 ++
10525 ++#ifdef CONFIG_VSERVER_DYNAMIC_IDS
10526 ++#define MIN_D_CONTEXT 49152 /* dynamic contexts start here */
10527 ++#else
10528 ++#define MIN_D_CONTEXT 65536
10529 ++#endif
10530 ++
10531 ++/* check conditions */
10532 ++
10533 ++#define VS_ADMIN 0x0001
10534 ++#define VS_WATCH 0x0002
10535 ++#define VS_HIDE 0x0004
10536 ++#define VS_HOSTID 0x0008
10537 ++
10538 ++#define VS_IDENT 0x0010
10539 ++#define VS_EQUIV 0x0020
10540 ++#define VS_PARENT 0x0040
10541 ++#define VS_CHILD 0x0080
10542 ++
10543 ++#define VS_ARG_MASK 0x00F0
10544 ++
10545 ++#define VS_DYNAMIC 0x0100
10546 ++#define VS_STATIC 0x0200
10547 ++
10548 ++#define VS_ATR_MASK 0x0F00
10549 ++
10550 ++#ifdef CONFIG_VSERVER_PRIVACY
10551 ++#define VS_ADMIN_P (0)
10552 ++#define VS_WATCH_P (0)
10553 ++#else
10554 ++#define VS_ADMIN_P VS_ADMIN
10555 ++#define VS_WATCH_P VS_WATCH
10556 ++#endif
10557 ++
10558 ++#define VS_HARDIRQ 0x1000
10559 ++#define VS_SOFTIRQ 0x2000
10560 ++#define VS_IRQ 0x4000
10561 ++
10562 ++#define VS_IRQ_MASK 0xF000
10563 ++
10564 ++#include <linux/hardirq.h>
10565 ++
10566 ++/*
10567 ++ * check current context for ADMIN/WATCH and
10568 ++ * optionally against supplied argument
10569 ++ */
10570 ++static inline int __vs_check(int cid, int id, unsigned int mode)
10571 ++{
10572 ++ if (mode & VS_ARG_MASK) {
10573 ++ if ((mode & VS_IDENT) && (id == cid))
10574 ++ return 1;
10575 ++ }
10576 ++ if (mode & VS_ATR_MASK) {
10577 ++ if ((mode & VS_DYNAMIC) &&
10578 ++ (id >= MIN_D_CONTEXT) &&
10579 ++ (id <= MAX_S_CONTEXT))
10580 ++ return 1;
10581 ++ if ((mode & VS_STATIC) &&
10582 ++ (id > 1) && (id < MIN_D_CONTEXT))
10583 ++ return 1;
10584 ++ }
10585 ++ if (mode & VS_IRQ_MASK) {
10586 ++ if ((mode & VS_IRQ) && unlikely(in_interrupt()))
10587 ++ return 1;
10588 ++ if ((mode & VS_HARDIRQ) && unlikely(in_irq()))
10589 ++ return 1;
10590 ++ if ((mode & VS_SOFTIRQ) && unlikely(in_softirq()))
10591 ++ return 1;
10592 ++ }
10593 ++ return (((mode & VS_ADMIN) && (cid == 0)) ||
10594 ++ ((mode & VS_WATCH) && (cid == 1)) ||
10595 ++ ((mode & VS_HOSTID) && (id == 0)));
10596 ++}
10597 ++
10598 ++#define vx_check(c, m) __vs_check(vx_current_xid(), c, (m) | VS_IRQ)
10599 ++
10600 ++#define vx_weak_check(c, m) ((m) ? vx_check(c, m) : 1)
10601 ++
10602 ++
10603 ++#define nx_check(c, m) __vs_check(nx_current_nid(), c, m)
10604 ++
10605 ++#define nx_weak_check(c, m) ((m) ? nx_check(c, m) : 1)
10606 ++
10607 ++#endif
10608 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/context_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/context_cmd.h
10609 +--- linux-2.6.29.4/include/linux/vserver/context_cmd.h 1970-01-01 01:00:00.000000000 +0100
10610 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/context_cmd.h 2009-02-22 22:54:26.000000000 +0100
10611 +@@ -0,0 +1,128 @@
10612 ++#ifndef _VX_CONTEXT_CMD_H
10613 ++#define _VX_CONTEXT_CMD_H
10614 ++
10615 ++
10616 ++/* vinfo commands */
10617 ++
10618 ++#define VCMD_task_xid VC_CMD(VINFO, 1, 0)
10619 ++
10620 ++#ifdef __KERNEL__
10621 ++extern int vc_task_xid(uint32_t);
10622 ++
10623 ++#endif /* __KERNEL__ */
10624 ++
10625 ++#define VCMD_vx_info VC_CMD(VINFO, 5, 0)
10626 ++
10627 ++struct vcmd_vx_info_v0 {
10628 ++ uint32_t xid;
10629 ++ uint32_t initpid;
10630 ++ /* more to come */
10631 ++};
10632 ++
10633 ++#ifdef __KERNEL__
10634 ++extern int vc_vx_info(struct vx_info *, void __user *);
10635 ++
10636 ++#endif /* __KERNEL__ */
10637 ++
10638 ++#define VCMD_ctx_stat VC_CMD(VSTAT, 0, 0)
10639 ++
10640 ++struct vcmd_ctx_stat_v0 {
10641 ++ uint32_t usecnt;
10642 ++ uint32_t tasks;
10643 ++ /* more to come */
10644 ++};
10645 ++
10646 ++#ifdef __KERNEL__
10647 ++extern int vc_ctx_stat(struct vx_info *, void __user *);
10648 ++
10649 ++#endif /* __KERNEL__ */
10650 ++
10651 ++/* context commands */
10652 ++
10653 ++#define VCMD_ctx_create_v0 VC_CMD(VPROC, 1, 0)
10654 ++#define VCMD_ctx_create VC_CMD(VPROC, 1, 1)
10655 ++
10656 ++struct vcmd_ctx_create {
10657 ++ uint64_t flagword;
10658 ++};
10659 ++
10660 ++#define VCMD_ctx_migrate_v0 VC_CMD(PROCMIG, 1, 0)
10661 ++#define VCMD_ctx_migrate VC_CMD(PROCMIG, 1, 1)
10662 ++
10663 ++struct vcmd_ctx_migrate {
10664 ++ uint64_t flagword;
10665 ++};
10666 ++
10667 ++#ifdef __KERNEL__
10668 ++extern int vc_ctx_create(uint32_t, void __user *);
10669 ++extern int vc_ctx_migrate(struct vx_info *, void __user *);
10670 ++
10671 ++#endif /* __KERNEL__ */
10672 ++
10673 ++
10674 ++/* flag commands */
10675 ++
10676 ++#define VCMD_get_cflags VC_CMD(FLAGS, 1, 0)
10677 ++#define VCMD_set_cflags VC_CMD(FLAGS, 2, 0)
10678 ++
10679 ++struct vcmd_ctx_flags_v0 {
10680 ++ uint64_t flagword;
10681 ++ uint64_t mask;
10682 ++};
10683 ++
10684 ++#ifdef __KERNEL__
10685 ++extern int vc_get_cflags(struct vx_info *, void __user *);
10686 ++extern int vc_set_cflags(struct vx_info *, void __user *);
10687 ++
10688 ++#endif /* __KERNEL__ */
10689 ++
10690 ++
10691 ++/* context caps commands */
10692 ++
10693 ++#define VCMD_get_ccaps VC_CMD(FLAGS, 3, 1)
10694 ++#define VCMD_set_ccaps VC_CMD(FLAGS, 4, 1)
10695 ++
10696 ++struct vcmd_ctx_caps_v1 {
10697 ++ uint64_t ccaps;
10698 ++ uint64_t cmask;
10699 ++};
10700 ++
10701 ++#ifdef __KERNEL__
10702 ++extern int vc_get_ccaps(struct vx_info *, void __user *);
10703 ++extern int vc_set_ccaps(struct vx_info *, void __user *);
10704 ++
10705 ++#endif /* __KERNEL__ */
10706 ++
10707 ++
10708 ++/* bcaps commands */
10709 ++
10710 ++#define VCMD_get_bcaps VC_CMD(FLAGS, 9, 0)
10711 ++#define VCMD_set_bcaps VC_CMD(FLAGS, 10, 0)
10712 ++
10713 ++struct vcmd_bcaps {
10714 ++ uint64_t bcaps;
10715 ++ uint64_t bmask;
10716 ++};
10717 ++
10718 ++#ifdef __KERNEL__
10719 ++extern int vc_get_bcaps(struct vx_info *, void __user *);
10720 ++extern int vc_set_bcaps(struct vx_info *, void __user *);
10721 ++
10722 ++#endif /* __KERNEL__ */
10723 ++
10724 ++
10725 ++/* OOM badness */
10726 ++
10727 ++#define VCMD_get_badness VC_CMD(MEMCTRL, 5, 0)
10728 ++#define VCMD_set_badness VC_CMD(MEMCTRL, 6, 0)
10729 ++
10730 ++struct vcmd_badness_v0 {
10731 ++ int64_t bias;
10732 ++};
10733 ++
10734 ++#ifdef __KERNEL__
10735 ++extern int vc_get_badness(struct vx_info *, void __user *);
10736 ++extern int vc_set_badness(struct vx_info *, void __user *);
10737 ++
10738 ++#endif /* __KERNEL__ */
10739 ++#endif /* _VX_CONTEXT_CMD_H */
10740 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/context.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/context.h
10741 +--- linux-2.6.29.4/include/linux/vserver/context.h 1970-01-01 01:00:00.000000000 +0100
10742 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/context.h 2009-02-22 22:54:26.000000000 +0100
10743 +@@ -0,0 +1,179 @@
10744 ++#ifndef _VX_CONTEXT_H
10745 ++#define _VX_CONTEXT_H
10746 ++
10747 ++#include <linux/types.h>
10748 ++#include <linux/capability.h>
10749 ++
10750 ++
10751 ++/* context flags */
10752 ++
10753 ++#define VXF_INFO_SCHED 0x00000002
10754 ++#define VXF_INFO_NPROC 0x00000004
10755 ++#define VXF_INFO_PRIVATE 0x00000008
10756 ++
10757 ++#define VXF_INFO_INIT 0x00000010
10758 ++#define VXF_INFO_HIDE 0x00000020
10759 ++#define VXF_INFO_ULIMIT 0x00000040
10760 ++#define VXF_INFO_NSPACE 0x00000080
10761 ++
10762 ++#define VXF_SCHED_HARD 0x00000100
10763 ++#define VXF_SCHED_PRIO 0x00000200
10764 ++#define VXF_SCHED_PAUSE 0x00000400
10765 ++
10766 ++#define VXF_VIRT_MEM 0x00010000
10767 ++#define VXF_VIRT_UPTIME 0x00020000
10768 ++#define VXF_VIRT_CPU 0x00040000
10769 ++#define VXF_VIRT_LOAD 0x00080000
10770 ++#define VXF_VIRT_TIME 0x00100000
10771 ++
10772 ++#define VXF_HIDE_MOUNT 0x01000000
10773 ++/* was VXF_HIDE_NETIF 0x02000000 */
10774 ++#define VXF_HIDE_VINFO 0x04000000
10775 ++
10776 ++#define VXF_STATE_SETUP (1ULL << 32)
10777 ++#define VXF_STATE_INIT (1ULL << 33)
10778 ++#define VXF_STATE_ADMIN (1ULL << 34)
10779 ++
10780 ++#define VXF_SC_HELPER (1ULL << 36)
10781 ++#define VXF_REBOOT_KILL (1ULL << 37)
10782 ++#define VXF_PERSISTENT (1ULL << 38)
10783 ++
10784 ++#define VXF_FORK_RSS (1ULL << 48)
10785 ++#define VXF_PROLIFIC (1ULL << 49)
10786 ++
10787 ++#define VXF_IGNEG_NICE (1ULL << 52)
10788 ++
10789 ++#define VXF_ONE_TIME (0x0007ULL << 32)
10790 ++
10791 ++#define VXF_INIT_SET (VXF_STATE_SETUP | VXF_STATE_INIT | VXF_STATE_ADMIN)
10792 ++
10793 ++
10794 ++/* context migration */
10795 ++
10796 ++#define VXM_SET_INIT 0x00000001
10797 ++#define VXM_SET_REAPER 0x00000002
10798 ++
10799 ++/* context caps */
10800 ++
10801 ++#define VXC_CAP_MASK 0x00000000
10802 ++
10803 ++#define VXC_SET_UTSNAME 0x00000001
10804 ++#define VXC_SET_RLIMIT 0x00000002
10805 ++#define VXC_FS_SECURITY 0x00000004
10806 ++
10807 ++/* was VXC_RAW_ICMP 0x00000100 */
10808 ++#define VXC_SYSLOG 0x00001000
10809 ++
10810 ++#define VXC_SECURE_MOUNT 0x00010000
10811 ++#define VXC_SECURE_REMOUNT 0x00020000
10812 ++#define VXC_BINARY_MOUNT 0x00040000
10813 ++
10814 ++#define VXC_QUOTA_CTL 0x00100000
10815 ++#define VXC_ADMIN_MAPPER 0x00200000
10816 ++#define VXC_ADMIN_CLOOP 0x00400000
10817 ++
10818 ++#define VXC_KTHREAD 0x01000000
10819 ++
10820 ++
10821 ++#ifdef __KERNEL__
10822 ++
10823 ++#include <linux/list.h>
10824 ++#include <linux/spinlock.h>
10825 ++#include <linux/rcupdate.h>
10826 ++
10827 ++#include "limit_def.h"
10828 ++#include "sched_def.h"
10829 ++#include "cvirt_def.h"
10830 ++#include "cacct_def.h"
10831 ++#include "device_def.h"
10832 ++
10833 ++#define VX_SPACES 2
10834 ++
10835 ++struct _vx_info_pc {
10836 ++ struct _vx_sched_pc sched_pc;
10837 ++ struct _vx_cvirt_pc cvirt_pc;
10838 ++};
10839 ++
10840 ++struct vx_info {
10841 ++ struct hlist_node vx_hlist; /* linked list of contexts */
10842 ++ xid_t vx_id; /* context id */
10843 ++ atomic_t vx_usecnt; /* usage count */
10844 ++ atomic_t vx_tasks; /* tasks count */
10845 ++ struct vx_info *vx_parent; /* parent context */
10846 ++ int vx_state; /* context state */
10847 ++
10848 ++ unsigned long vx_nsmask[VX_SPACES]; /* assignment mask */
10849 ++ struct nsproxy *vx_nsproxy[VX_SPACES]; /* private namespaces */
10850 ++ struct fs_struct *vx_fs[VX_SPACES]; /* private namespace fs */
10851 ++
10852 ++ uint64_t vx_flags; /* context flags */
10853 ++ uint64_t vx_ccaps; /* context caps (vserver) */
10854 ++ kernel_cap_t vx_bcaps; /* bounding caps (system) */
10855 ++ // kernel_cap_t vx_cap_bset; /* the guest's bset */
10856 ++
10857 ++ struct task_struct *vx_reaper; /* guest reaper process */
10858 ++ pid_t vx_initpid; /* PID of guest init */
10859 ++ int64_t vx_badness_bias; /* OOM points bias */
10860 ++
10861 ++ struct _vx_limit limit; /* vserver limits */
10862 ++ struct _vx_sched sched; /* vserver scheduler */
10863 ++ struct _vx_cvirt cvirt; /* virtual/bias stuff */
10864 ++ struct _vx_cacct cacct; /* context accounting */
10865 ++
10866 ++ struct _vx_device dmap; /* default device map targets */
10867 ++
10868 ++#ifndef CONFIG_SMP
10869 ++ struct _vx_info_pc info_pc; /* per cpu data */
10870 ++#else
10871 ++ struct _vx_info_pc *ptr_pc; /* per cpu array */
10872 ++#endif
10873 ++
10874 ++ wait_queue_head_t vx_wait; /* context exit waitqueue */
10875 ++ int reboot_cmd; /* last sys_reboot() cmd */
10876 ++ int exit_code; /* last process exit code */
10877 ++
10878 ++ char vx_name[65]; /* vserver name */
10879 ++};
10880 ++
10881 ++#ifndef CONFIG_SMP
10882 ++#define vx_ptr_pc(vxi) (&(vxi)->info_pc)
10883 ++#define vx_per_cpu(vxi, v, id) vx_ptr_pc(vxi)->v
10884 ++#else
10885 ++#define vx_ptr_pc(vxi) ((vxi)->ptr_pc)
10886 ++#define vx_per_cpu(vxi, v, id) per_cpu_ptr(vx_ptr_pc(vxi), id)->v
10887 ++#endif
10888 ++
10889 ++#define vx_cpu(vxi, v) vx_per_cpu(vxi, v, smp_processor_id())
10890 ++
10891 ++
10892 ++struct vx_info_save {
10893 ++ struct vx_info *vxi;
10894 ++ xid_t xid;
10895 ++};
10896 ++
10897 ++
10898 ++/* status flags */
10899 ++
10900 ++#define VXS_HASHED 0x0001
10901 ++#define VXS_PAUSED 0x0010
10902 ++#define VXS_SHUTDOWN 0x0100
10903 ++#define VXS_HELPER 0x1000
10904 ++#define VXS_RELEASED 0x8000
10905 ++
10906 ++
10907 ++extern void claim_vx_info(struct vx_info *, struct task_struct *);
10908 ++extern void release_vx_info(struct vx_info *, struct task_struct *);
10909 ++
10910 ++extern struct vx_info *lookup_vx_info(int);
10911 ++extern struct vx_info *lookup_or_create_vx_info(int);
10912 ++
10913 ++extern int get_xid_list(int, unsigned int *, int);
10914 ++extern int xid_is_hashed(xid_t);
10915 ++
10916 ++extern int vx_migrate_task(struct task_struct *, struct vx_info *, int);
10917 ++
10918 ++extern long vs_state_change(struct vx_info *, unsigned int);
10919 ++
10920 ++
10921 ++#endif /* __KERNEL__ */
10922 ++#endif /* _VX_CONTEXT_H */
10923 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cvirt_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt_cmd.h
10924 +--- linux-2.6.29.4/include/linux/vserver/cvirt_cmd.h 1970-01-01 01:00:00.000000000 +0100
10925 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt_cmd.h 2009-02-22 22:54:26.000000000 +0100
10926 +@@ -0,0 +1,53 @@
10927 ++#ifndef _VX_CVIRT_CMD_H
10928 ++#define _VX_CVIRT_CMD_H
10929 ++
10930 ++
10931 ++/* virtual host info name commands */
10932 ++
10933 ++#define VCMD_set_vhi_name VC_CMD(VHOST, 1, 0)
10934 ++#define VCMD_get_vhi_name VC_CMD(VHOST, 2, 0)
10935 ++
10936 ++struct vcmd_vhi_name_v0 {
10937 ++ uint32_t field;
10938 ++ char name[65];
10939 ++};
10940 ++
10941 ++
10942 ++enum vhi_name_field {
10943 ++ VHIN_CONTEXT = 0,
10944 ++ VHIN_SYSNAME,
10945 ++ VHIN_NODENAME,
10946 ++ VHIN_RELEASE,
10947 ++ VHIN_VERSION,
10948 ++ VHIN_MACHINE,
10949 ++ VHIN_DOMAINNAME,
10950 ++};
10951 ++
10952 ++
10953 ++#ifdef __KERNEL__
10954 ++
10955 ++#include <linux/compiler.h>
10956 ++
10957 ++extern int vc_set_vhi_name(struct vx_info *, void __user *);
10958 ++extern int vc_get_vhi_name(struct vx_info *, void __user *);
10959 ++
10960 ++#endif /* __KERNEL__ */
10961 ++
10962 ++#define VCMD_virt_stat VC_CMD(VSTAT, 3, 0)
10963 ++
10964 ++struct vcmd_virt_stat_v0 {
10965 ++ uint64_t offset;
10966 ++ uint64_t uptime;
10967 ++ uint32_t nr_threads;
10968 ++ uint32_t nr_running;
10969 ++ uint32_t nr_uninterruptible;
10970 ++ uint32_t nr_onhold;
10971 ++ uint32_t nr_forks;
10972 ++ uint32_t load[3];
10973 ++};
10974 ++
10975 ++#ifdef __KERNEL__
10976 ++extern int vc_virt_stat(struct vx_info *, void __user *);
10977 ++
10978 ++#endif /* __KERNEL__ */
10979 ++#endif /* _VX_CVIRT_CMD_H */
10980 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cvirt_def.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt_def.h
10981 +--- linux-2.6.29.4/include/linux/vserver/cvirt_def.h 1970-01-01 01:00:00.000000000 +0100
10982 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt_def.h 2009-02-22 22:54:26.000000000 +0100
10983 +@@ -0,0 +1,80 @@
10984 ++#ifndef _VX_CVIRT_DEF_H
10985 ++#define _VX_CVIRT_DEF_H
10986 ++
10987 ++#include <linux/jiffies.h>
10988 ++#include <linux/spinlock.h>
10989 ++#include <linux/wait.h>
10990 ++#include <linux/time.h>
10991 ++#include <asm/atomic.h>
10992 ++
10993 ++
10994 ++struct _vx_usage_stat {
10995 ++ uint64_t user;
10996 ++ uint64_t nice;
10997 ++ uint64_t system;
10998 ++ uint64_t softirq;
10999 ++ uint64_t irq;
11000 ++ uint64_t idle;
11001 ++ uint64_t iowait;
11002 ++};
11003 ++
11004 ++struct _vx_syslog {
11005 ++ wait_queue_head_t log_wait;
11006 ++ spinlock_t logbuf_lock; /* lock for the log buffer */
11007 ++
11008 ++ unsigned long log_start; /* next char to be read by syslog() */
11009 ++ unsigned long con_start; /* next char to be sent to consoles */
11010 ++ unsigned long log_end; /* most-recently-written-char + 1 */
11011 ++ unsigned long logged_chars; /* #chars since last read+clear operation */
11012 ++
11013 ++ char log_buf[1024];
11014 ++};
11015 ++
11016 ++
11017 ++/* context sub struct */
11018 ++
11019 ++struct _vx_cvirt {
11020 ++ atomic_t nr_threads; /* number of current threads */
11021 ++ atomic_t nr_running; /* number of running threads */
11022 ++ atomic_t nr_uninterruptible; /* number of uninterruptible threads */
11023 ++
11024 ++ atomic_t nr_onhold; /* processes on hold */
11025 ++ uint32_t onhold_last; /* jiffies when put on hold */
11026 ++
11027 ++ struct timeval bias_tv; /* time offset to the host */
11028 ++ struct timespec bias_idle;
11029 ++ struct timespec bias_uptime; /* context creation point */
11030 ++ uint64_t bias_clock; /* offset in clock_t */
11031 ++
11032 ++ spinlock_t load_lock; /* lock for the load averages */
11033 ++ atomic_t load_updates; /* nr of load updates done so far */
11034 ++ uint32_t load_last; /* last time load was calculated */
11035 ++ uint32_t load[3]; /* load averages 1,5,15 */
11036 ++
11037 ++ atomic_t total_forks; /* number of forks so far */
11038 ++
11039 ++ struct _vx_syslog syslog;
11040 ++};
11041 ++
11042 ++struct _vx_cvirt_pc {
11043 ++ struct _vx_usage_stat cpustat;
11044 ++};
11045 ++
11046 ++
11047 ++#ifdef CONFIG_VSERVER_DEBUG
11048 ++
11049 ++static inline void __dump_vx_cvirt(struct _vx_cvirt *cvirt)
11050 ++{
11051 ++ printk("\t_vx_cvirt:\n");
11052 ++ printk("\t threads: %4d, %4d, %4d, %4d\n",
11053 ++ atomic_read(&cvirt->nr_threads),
11054 ++ atomic_read(&cvirt->nr_running),
11055 ++ atomic_read(&cvirt->nr_uninterruptible),
11056 ++ atomic_read(&cvirt->nr_onhold));
11057 ++ /* add rest here */
11058 ++ printk("\t total_forks = %d\n", atomic_read(&cvirt->total_forks));
11059 ++}
11060 ++
11061 ++#endif
11062 ++
11063 ++#endif /* _VX_CVIRT_DEF_H */
11064 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/cvirt.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt.h
11065 +--- linux-2.6.29.4/include/linux/vserver/cvirt.h 1970-01-01 01:00:00.000000000 +0100
11066 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/cvirt.h 2009-02-22 22:54:26.000000000 +0100
11067 +@@ -0,0 +1,20 @@
11068 ++#ifndef _VX_CVIRT_H
11069 ++#define _VX_CVIRT_H
11070 ++
11071 ++
11072 ++#ifdef __KERNEL__
11073 ++
11074 ++struct timespec;
11075 ++
11076 ++void vx_vsi_uptime(struct timespec *, struct timespec *);
11077 ++
11078 ++
11079 ++struct vx_info;
11080 ++
11081 ++void vx_update_load(struct vx_info *);
11082 ++
11083 ++
11084 ++int vx_do_syslog(int, char __user *, int);
11085 ++
11086 ++#endif /* __KERNEL__ */
11087 ++#endif /* _VX_CVIRT_H */
11088 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/debug_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/debug_cmd.h
11089 +--- linux-2.6.29.4/include/linux/vserver/debug_cmd.h 1970-01-01 01:00:00.000000000 +0100
11090 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/debug_cmd.h 2009-02-22 22:54:26.000000000 +0100
11091 +@@ -0,0 +1,58 @@
11092 ++#ifndef _VX_DEBUG_CMD_H
11093 ++#define _VX_DEBUG_CMD_H
11094 ++
11095 ++
11096 ++/* debug commands */
11097 ++
11098 ++#define VCMD_dump_history VC_CMD(DEBUG, 1, 0)
11099 ++
11100 ++#define VCMD_read_history VC_CMD(DEBUG, 5, 0)
11101 ++#define VCMD_read_monitor VC_CMD(DEBUG, 6, 0)
11102 ++
11103 ++struct vcmd_read_history_v0 {
11104 ++ uint32_t index;
11105 ++ uint32_t count;
11106 ++ char __user *data;
11107 ++};
11108 ++
11109 ++struct vcmd_read_monitor_v0 {
11110 ++ uint32_t index;
11111 ++ uint32_t count;
11112 ++ char __user *data;
11113 ++};
11114 ++
11115 ++
11116 ++#ifdef __KERNEL__
11117 ++
11118 ++#ifdef CONFIG_COMPAT
11119 ++
11120 ++#include <asm/compat.h>
11121 ++
11122 ++struct vcmd_read_history_v0_x32 {
11123 ++ uint32_t index;
11124 ++ uint32_t count;
11125 ++ compat_uptr_t data_ptr;
11126 ++};
11127 ++
11128 ++struct vcmd_read_monitor_v0_x32 {
11129 ++ uint32_t index;
11130 ++ uint32_t count;
11131 ++ compat_uptr_t data_ptr;
11132 ++};
11133 ++
11134 ++#endif /* CONFIG_COMPAT */
11135 ++
11136 ++extern int vc_dump_history(uint32_t);
11137 ++
11138 ++extern int vc_read_history(uint32_t, void __user *);
11139 ++extern int vc_read_monitor(uint32_t, void __user *);
11140 ++
11141 ++#ifdef CONFIG_COMPAT
11142 ++
11143 ++extern int vc_read_history_x32(uint32_t, void __user *);
11144 ++extern int vc_read_monitor_x32(uint32_t, void __user *);
11145 ++
11146 ++#endif /* CONFIG_COMPAT */
11147 ++
11148 ++#endif /* __KERNEL__ */
11149 ++#endif /* _VX_DEBUG_CMD_H */
11150 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/debug.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/debug.h
11151 +--- linux-2.6.29.4/include/linux/vserver/debug.h 1970-01-01 01:00:00.000000000 +0100
11152 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/debug.h 2009-02-22 22:54:26.000000000 +0100
11153 +@@ -0,0 +1,127 @@
11154 ++#ifndef _VX_DEBUG_H
11155 ++#define _VX_DEBUG_H
11156 ++
11157 ++
11158 ++#define VXD_CBIT(n, m) (vx_debug_ ## n & (1 << (m)))
11159 ++#define VXD_CMIN(n, m) (vx_debug_ ## n > (m))
11160 ++#define VXD_MASK(n, m) (vx_debug_ ## n & (m))
11161 ++
11162 ++#define VXD_DEV(d) (d), (d)->bd_inode->i_ino, \
11163 ++ imajor((d)->bd_inode), iminor((d)->bd_inode)
11164 ++#define VXF_DEV "%p[%lu,%d:%d]"
11165 ++
11166 ++
11167 ++#define vxd_path(p) \
11168 ++ ({ static char _buffer[PATH_MAX]; \
11169 ++ d_path(p, _buffer, sizeof(_buffer)); })
11170 ++
11171 ++#define vxd_cond_path(n) \
11172 ++ ((n) ? vxd_path(&(n)->path) : "<null>" )
11173 ++
11174 ++
11175 ++#ifdef CONFIG_VSERVER_DEBUG
11176 ++
11177 ++extern unsigned int vx_debug_switch;
11178 ++extern unsigned int vx_debug_xid;
11179 ++extern unsigned int vx_debug_nid;
11180 ++extern unsigned int vx_debug_tag;
11181 ++extern unsigned int vx_debug_net;
11182 ++extern unsigned int vx_debug_limit;
11183 ++extern unsigned int vx_debug_cres;
11184 ++extern unsigned int vx_debug_dlim;
11185 ++extern unsigned int vx_debug_quota;
11186 ++extern unsigned int vx_debug_cvirt;
11187 ++extern unsigned int vx_debug_space;
11188 ++extern unsigned int vx_debug_misc;
11189 ++
11190 ++
11191 ++#define VX_LOGLEVEL "vxD: "
11192 ++#define VX_PROC_FMT "%p: "
11193 ++#define VX_PROCESS current
11194 ++
11195 ++#define vxdprintk(c, f, x...) \
11196 ++ do { \
11197 ++ if (c) \
11198 ++ printk(VX_LOGLEVEL VX_PROC_FMT f "\n", \
11199 ++ VX_PROCESS , ##x); \
11200 ++ } while (0)
11201 ++
11202 ++#define vxlprintk(c, f, x...) \
11203 ++ do { \
11204 ++ if (c) \
11205 ++ printk(VX_LOGLEVEL f " @%s:%d\n", x); \
11206 ++ } while (0)
11207 ++
11208 ++#define vxfprintk(c, f, x...) \
11209 ++ do { \
11210 ++ if (c) \
11211 ++ printk(VX_LOGLEVEL f " %s@%s:%d\n", x); \
11212 ++ } while (0)
11213 ++
11214 ++
11215 ++struct vx_info;
11216 ++
11217 ++void dump_vx_info(struct vx_info *, int);
11218 ++void dump_vx_info_inactive(int);
11219 ++
11220 ++#else /* CONFIG_VSERVER_DEBUG */
11221 ++
11222 ++#define vx_debug_switch 0
11223 ++#define vx_debug_xid 0
11224 ++#define vx_debug_nid 0
11225 ++#define vx_debug_tag 0
11226 ++#define vx_debug_net 0
11227 ++#define vx_debug_limit 0
11228 ++#define vx_debug_cres 0
11229 ++#define vx_debug_dlim 0
11230 ++#define vx_debug_cvirt 0
11231 ++
11232 ++#define vxdprintk(x...) do { } while (0)
11233 ++#define vxlprintk(x...) do { } while (0)
11234 ++#define vxfprintk(x...) do { } while (0)
11235 ++
11236 ++#endif /* CONFIG_VSERVER_DEBUG */
11237 ++
11238 ++
11239 ++#ifdef CONFIG_VSERVER_WARN
11240 ++
11241 ++#define VX_WARNLEVEL KERN_WARNING "vxW: "
11242 ++#define VX_WARN_TASK "[»%s«,%u:#%u|%u|%u] "
11243 ++#define VX_WARN_XID "[xid #%u] "
11244 ++#define VX_WARN_NID "[nid #%u] "
11245 ++#define VX_WARN_TAG "[tag #%u] "
11246 ++
11247 ++#define vxwprintk(c, f, x...) \
11248 ++ do { \
11249 ++ if (c) \
11250 ++ printk(VX_WARNLEVEL f "\n", ##x); \
11251 ++ } while (0)
11252 ++
11253 ++#else /* CONFIG_VSERVER_WARN */
11254 ++
11255 ++#define vxwprintk(x...) do { } while (0)
11256 ++
11257 ++#endif /* CONFIG_VSERVER_WARN */
11258 ++
11259 ++#define vxwprintk_task(c, f, x...) \
11260 ++ vxwprintk(c, VX_WARN_TASK f, \
11261 ++ current->comm, current->pid, \
11262 ++ current->xid, current->nid, current->tag, ##x)
11263 ++#define vxwprintk_xid(c, f, x...) \
11264 ++ vxwprintk(c, VX_WARN_XID f, current->xid, x)
11265 ++#define vxwprintk_nid(c, f, x...) \
11266 ++ vxwprintk(c, VX_WARN_NID f, current->nid, x)
11267 ++#define vxwprintk_tag(c, f, x...) \
11268 ++ vxwprintk(c, VX_WARN_TAG f, current->tag, x)
11269 ++
11270 ++#ifdef CONFIG_VSERVER_DEBUG
11271 ++#define vxd_assert_lock(l) assert_spin_locked(l)
11272 ++#define vxd_assert(c, f, x...) vxlprintk(!(c), \
11273 ++ "assertion [" f "] failed.", ##x, __FILE__, __LINE__)
11274 ++#else
11275 ++#define vxd_assert_lock(l) do { } while (0)
11276 ++#define vxd_assert(c, f, x...) do { } while (0)
11277 ++#endif
11278 ++
11279 ++
11280 ++#endif /* _VX_DEBUG_H */
11281 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/device_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device_cmd.h
11282 +--- linux-2.6.29.4/include/linux/vserver/device_cmd.h 1970-01-01 01:00:00.000000000 +0100
11283 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device_cmd.h 2009-02-22 22:54:26.000000000 +0100
11284 +@@ -0,0 +1,44 @@
11285 ++#ifndef _VX_DEVICE_CMD_H
11286 ++#define _VX_DEVICE_CMD_H
11287 ++
11288 ++
11289 ++/* device vserver commands */
11290 ++
11291 ++#define VCMD_set_mapping VC_CMD(DEVICE, 1, 0)
11292 ++#define VCMD_unset_mapping VC_CMD(DEVICE, 2, 0)
11293 ++
11294 ++struct vcmd_set_mapping_v0 {
11295 ++ const char __user *device;
11296 ++ const char __user *target;
11297 ++ uint32_t flags;
11298 ++};
11299 ++
11300 ++
11301 ++#ifdef __KERNEL__
11302 ++
11303 ++#ifdef CONFIG_COMPAT
11304 ++
11305 ++#include <asm/compat.h>
11306 ++
11307 ++struct vcmd_set_mapping_v0_x32 {
11308 ++ compat_uptr_t device_ptr;
11309 ++ compat_uptr_t target_ptr;
11310 ++ uint32_t flags;
11311 ++};
11312 ++
11313 ++#endif /* CONFIG_COMPAT */
11314 ++
11315 ++#include <linux/compiler.h>
11316 ++
11317 ++extern int vc_set_mapping(struct vx_info *, void __user *);
11318 ++extern int vc_unset_mapping(struct vx_info *, void __user *);
11319 ++
11320 ++#ifdef CONFIG_COMPAT
11321 ++
11322 ++extern int vc_set_mapping_x32(struct vx_info *, void __user *);
11323 ++extern int vc_unset_mapping_x32(struct vx_info *, void __user *);
11324 ++
11325 ++#endif /* CONFIG_COMPAT */
11326 ++
11327 ++#endif /* __KERNEL__ */
11328 ++#endif /* _VX_DEVICE_CMD_H */
11329 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/device_def.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device_def.h
11330 +--- linux-2.6.29.4/include/linux/vserver/device_def.h 1970-01-01 01:00:00.000000000 +0100
11331 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device_def.h 2009-02-22 22:54:26.000000000 +0100
11332 +@@ -0,0 +1,17 @@
11333 ++#ifndef _VX_DEVICE_DEF_H
11334 ++#define _VX_DEVICE_DEF_H
11335 ++
11336 ++#include <linux/types.h>
11337 ++
11338 ++struct vx_dmap_target {
11339 ++ dev_t target;
11340 ++ uint32_t flags;
11341 ++};
11342 ++
11343 ++struct _vx_device {
11344 ++#ifdef CONFIG_VSERVER_DEVICE
11345 ++ struct vx_dmap_target targets[2];
11346 ++#endif
11347 ++};
11348 ++
11349 ++#endif /* _VX_DEVICE_DEF_H */
11350 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/device.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device.h
11351 +--- linux-2.6.29.4/include/linux/vserver/device.h 1970-01-01 01:00:00.000000000 +0100
11352 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/device.h 2009-02-22 22:54:26.000000000 +0100
11353 +@@ -0,0 +1,15 @@
11354 ++#ifndef _VX_DEVICE_H
11355 ++#define _VX_DEVICE_H
11356 ++
11357 ++
11358 ++#define DATTR_CREATE 0x00000001
11359 ++#define DATTR_OPEN 0x00000002
11360 ++
11361 ++#define DATTR_REMAP 0x00000010
11362 ++
11363 ++#define DATTR_MASK 0x00000013
11364 ++
11365 ++
11366 ++#else /* _VX_DEVICE_H */
11367 ++#warning duplicate inclusion
11368 ++#endif /* _VX_DEVICE_H */
11369 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/dlimit_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/dlimit_cmd.h
11370 +--- linux-2.6.29.4/include/linux/vserver/dlimit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11371 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/dlimit_cmd.h 2009-02-22 22:54:26.000000000 +0100
11372 +@@ -0,0 +1,74 @@
11373 ++#ifndef _VX_DLIMIT_CMD_H
11374 ++#define _VX_DLIMIT_CMD_H
11375 ++
11376 ++
11377 ++/* dlimit vserver commands */
11378 ++
11379 ++#define VCMD_add_dlimit VC_CMD(DLIMIT, 1, 0)
11380 ++#define VCMD_rem_dlimit VC_CMD(DLIMIT, 2, 0)
11381 ++
11382 ++#define VCMD_set_dlimit VC_CMD(DLIMIT, 5, 0)
11383 ++#define VCMD_get_dlimit VC_CMD(DLIMIT, 6, 0)
11384 ++
11385 ++struct vcmd_ctx_dlimit_base_v0 {
11386 ++ const char __user *name;
11387 ++ uint32_t flags;
11388 ++};
11389 ++
11390 ++struct vcmd_ctx_dlimit_v0 {
11391 ++ const char __user *name;
11392 ++ uint32_t space_used; /* used space in kbytes */
11393 ++ uint32_t space_total; /* maximum space in kbytes */
11394 ++ uint32_t inodes_used; /* used inodes */
11395 ++ uint32_t inodes_total; /* maximum inodes */
11396 ++ uint32_t reserved; /* reserved for root in % */
11397 ++ uint32_t flags;
11398 ++};
11399 ++
11400 ++#define CDLIM_UNSET ((uint32_t)0UL)
11401 ++#define CDLIM_INFINITY ((uint32_t)~0UL)
11402 ++#define CDLIM_KEEP ((uint32_t)~1UL)
11403 ++
11404 ++#ifdef __KERNEL__
11405 ++
11406 ++#ifdef CONFIG_COMPAT
11407 ++
11408 ++#include <asm/compat.h>
11409 ++
11410 ++struct vcmd_ctx_dlimit_base_v0_x32 {
11411 ++ compat_uptr_t name_ptr;
11412 ++ uint32_t flags;
11413 ++};
11414 ++
11415 ++struct vcmd_ctx_dlimit_v0_x32 {
11416 ++ compat_uptr_t name_ptr;
11417 ++ uint32_t space_used; /* used space in kbytes */
11418 ++ uint32_t space_total; /* maximum space in kbytes */
11419 ++ uint32_t inodes_used; /* used inodes */
11420 ++ uint32_t inodes_total; /* maximum inodes */
11421 ++ uint32_t reserved; /* reserved for root in % */
11422 ++ uint32_t flags;
11423 ++};
11424 ++
11425 ++#endif /* CONFIG_COMPAT */
11426 ++
11427 ++#include <linux/compiler.h>
11428 ++
11429 ++extern int vc_add_dlimit(uint32_t, void __user *);
11430 ++extern int vc_rem_dlimit(uint32_t, void __user *);
11431 ++
11432 ++extern int vc_set_dlimit(uint32_t, void __user *);
11433 ++extern int vc_get_dlimit(uint32_t, void __user *);
11434 ++
11435 ++#ifdef CONFIG_COMPAT
11436 ++
11437 ++extern int vc_add_dlimit_x32(uint32_t, void __user *);
11438 ++extern int vc_rem_dlimit_x32(uint32_t, void __user *);
11439 ++
11440 ++extern int vc_set_dlimit_x32(uint32_t, void __user *);
11441 ++extern int vc_get_dlimit_x32(uint32_t, void __user *);
11442 ++
11443 ++#endif /* CONFIG_COMPAT */
11444 ++
11445 ++#endif /* __KERNEL__ */
11446 ++#endif /* _VX_DLIMIT_CMD_H */
11447 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/dlimit.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/dlimit.h
11448 +--- linux-2.6.29.4/include/linux/vserver/dlimit.h 1970-01-01 01:00:00.000000000 +0100
11449 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/dlimit.h 2009-02-22 22:54:26.000000000 +0100
11450 +@@ -0,0 +1,54 @@
11451 ++#ifndef _VX_DLIMIT_H
11452 ++#define _VX_DLIMIT_H
11453 ++
11454 ++#include "switch.h"
11455 ++
11456 ++
11457 ++#ifdef __KERNEL__
11458 ++
11459 ++/* keep in sync with CDLIM_INFINITY */
11460 ++
11461 ++#define DLIM_INFINITY (~0ULL)
11462 ++
11463 ++#include <linux/spinlock.h>
11464 ++#include <linux/rcupdate.h>
11465 ++
11466 ++struct super_block;
11467 ++
11468 ++struct dl_info {
11469 ++ struct hlist_node dl_hlist; /* linked list of contexts */
11470 ++ struct rcu_head dl_rcu; /* the rcu head */
11471 ++ tag_t dl_tag; /* context tag */
11472 ++ atomic_t dl_usecnt; /* usage count */
11473 ++ atomic_t dl_refcnt; /* reference count */
11474 ++
11475 ++ struct super_block *dl_sb; /* associated superblock */
11476 ++
11477 ++ spinlock_t dl_lock; /* protect the values */
11478 ++
11479 ++ unsigned long long dl_space_used; /* used space in bytes */
11480 ++ unsigned long long dl_space_total; /* maximum space in bytes */
11481 ++ unsigned long dl_inodes_used; /* used inodes */
11482 ++ unsigned long dl_inodes_total; /* maximum inodes */
11483 ++
11484 ++ unsigned int dl_nrlmult; /* non root limit mult */
11485 ++};
11486 ++
11487 ++struct rcu_head;
11488 ++
11489 ++extern void rcu_free_dl_info(struct rcu_head *);
11490 ++extern void unhash_dl_info(struct dl_info *);
11491 ++
11492 ++extern struct dl_info *locate_dl_info(struct super_block *, tag_t);
11493 ++
11494 ++
11495 ++struct kstatfs;
11496 ++
11497 ++extern void vx_vsi_statfs(struct super_block *, struct kstatfs *);
11498 ++
11499 ++typedef uint64_t dlsize_t;
11500 ++
11501 ++#endif /* __KERNEL__ */
11502 ++#else /* _VX_DLIMIT_H */
11503 ++#warning duplicate inclusion
11504 ++#endif /* _VX_DLIMIT_H */
11505 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/global.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/global.h
11506 +--- linux-2.6.29.4/include/linux/vserver/global.h 1970-01-01 01:00:00.000000000 +0100
11507 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/global.h 2009-02-22 22:54:26.000000000 +0100
11508 +@@ -0,0 +1,20 @@
11509 ++#ifndef _VX_GLOBAL_H
11510 ++#define _VX_GLOBAL_H
11511 ++
11512 ++
11513 ++extern atomic_t vx_global_ctotal;
11514 ++extern atomic_t vx_global_cactive;
11515 ++
11516 ++extern atomic_t nx_global_ctotal;
11517 ++extern atomic_t nx_global_cactive;
11518 ++
11519 ++extern atomic_t vs_global_nsproxy;
11520 ++extern atomic_t vs_global_fs;
11521 ++extern atomic_t vs_global_mnt_ns;
11522 ++extern atomic_t vs_global_uts_ns;
11523 ++extern atomic_t vs_global_ipc_ns;
11524 ++extern atomic_t vs_global_user_ns;
11525 ++extern atomic_t vs_global_pid_ns;
11526 ++
11527 ++
11528 ++#endif /* _VX_GLOBAL_H */
11529 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/history.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/history.h
11530 +--- linux-2.6.29.4/include/linux/vserver/history.h 1970-01-01 01:00:00.000000000 +0100
11531 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/history.h 2009-02-22 22:54:26.000000000 +0100
11532 +@@ -0,0 +1,197 @@
11533 ++#ifndef _VX_HISTORY_H
11534 ++#define _VX_HISTORY_H
11535 ++
11536 ++
11537 ++enum {
11538 ++ VXH_UNUSED = 0,
11539 ++ VXH_THROW_OOPS = 1,
11540 ++
11541 ++ VXH_GET_VX_INFO,
11542 ++ VXH_PUT_VX_INFO,
11543 ++ VXH_INIT_VX_INFO,
11544 ++ VXH_SET_VX_INFO,
11545 ++ VXH_CLR_VX_INFO,
11546 ++ VXH_CLAIM_VX_INFO,
11547 ++ VXH_RELEASE_VX_INFO,
11548 ++ VXH_ALLOC_VX_INFO,
11549 ++ VXH_DEALLOC_VX_INFO,
11550 ++ VXH_HASH_VX_INFO,
11551 ++ VXH_UNHASH_VX_INFO,
11552 ++ VXH_LOC_VX_INFO,
11553 ++ VXH_LOOKUP_VX_INFO,
11554 ++ VXH_CREATE_VX_INFO,
11555 ++};
11556 ++
11557 ++struct _vxhe_vxi {
11558 ++ struct vx_info *ptr;
11559 ++ unsigned xid;
11560 ++ unsigned usecnt;
11561 ++ unsigned tasks;
11562 ++};
11563 ++
11564 ++struct _vxhe_set_clr {
11565 ++ void *data;
11566 ++};
11567 ++
11568 ++struct _vxhe_loc_lookup {
11569 ++ unsigned arg;
11570 ++};
11571 ++
11572 ++struct _vx_hist_entry {
11573 ++ void *loc;
11574 ++ unsigned short seq;
11575 ++ unsigned short type;
11576 ++ struct _vxhe_vxi vxi;
11577 ++ union {
11578 ++ struct _vxhe_set_clr sc;
11579 ++ struct _vxhe_loc_lookup ll;
11580 ++ };
11581 ++};
11582 ++
11583 ++#ifdef CONFIG_VSERVER_HISTORY
11584 ++
11585 ++extern unsigned volatile int vxh_active;
11586 ++
11587 ++struct _vx_hist_entry *vxh_advance(void *loc);
11588 ++
11589 ++
11590 ++static inline
11591 ++void __vxh_copy_vxi(struct _vx_hist_entry *entry, struct vx_info *vxi)
11592 ++{
11593 ++ entry->vxi.ptr = vxi;
11594 ++ if (vxi) {
11595 ++ entry->vxi.usecnt = atomic_read(&vxi->vx_usecnt);
11596 ++ entry->vxi.tasks = atomic_read(&vxi->vx_tasks);
11597 ++ entry->vxi.xid = vxi->vx_id;
11598 ++ }
11599 ++}
11600 ++
11601 ++
11602 ++#define __HERE__ current_text_addr()
11603 ++
11604 ++#define __VXH_BODY(__type, __data, __here) \
11605 ++ struct _vx_hist_entry *entry; \
11606 ++ \
11607 ++ preempt_disable(); \
11608 ++ entry = vxh_advance(__here); \
11609 ++ __data; \
11610 ++ entry->type = __type; \
11611 ++ preempt_enable();
11612 ++
11613 ++
11614 ++ /* pass vxi only */
11615 ++
11616 ++#define __VXH_SMPL \
11617 ++ __vxh_copy_vxi(entry, vxi)
11618 ++
11619 ++static inline
11620 ++void __vxh_smpl(struct vx_info *vxi, int __type, void *__here)
11621 ++{
11622 ++ __VXH_BODY(__type, __VXH_SMPL, __here)
11623 ++}
11624 ++
11625 ++ /* pass vxi and data (void *) */
11626 ++
11627 ++#define __VXH_DATA \
11628 ++ __vxh_copy_vxi(entry, vxi); \
11629 ++ entry->sc.data = data
11630 ++
11631 ++static inline
11632 ++void __vxh_data(struct vx_info *vxi, void *data,
11633 ++ int __type, void *__here)
11634 ++{
11635 ++ __VXH_BODY(__type, __VXH_DATA, __here)
11636 ++}
11637 ++
11638 ++ /* pass vxi and arg (long) */
11639 ++
11640 ++#define __VXH_LONG \
11641 ++ __vxh_copy_vxi(entry, vxi); \
11642 ++ entry->ll.arg = arg
11643 ++
11644 ++static inline
11645 ++void __vxh_long(struct vx_info *vxi, long arg,
11646 ++ int __type, void *__here)
11647 ++{
11648 ++ __VXH_BODY(__type, __VXH_LONG, __here)
11649 ++}
11650 ++
11651 ++
11652 ++static inline
11653 ++void __vxh_throw_oops(void *__here)
11654 ++{
11655 ++ __VXH_BODY(VXH_THROW_OOPS, {}, __here);
11656 ++ /* prevent further acquisition */
11657 ++ vxh_active = 0;
11658 ++}
11659 ++
11660 ++
11661 ++#define vxh_throw_oops() __vxh_throw_oops(__HERE__);
11662 ++
11663 ++#define __vxh_get_vx_info(v, h) __vxh_smpl(v, VXH_GET_VX_INFO, h);
11664 ++#define __vxh_put_vx_info(v, h) __vxh_smpl(v, VXH_PUT_VX_INFO, h);
11665 ++
11666 ++#define __vxh_init_vx_info(v, d, h) \
11667 ++ __vxh_data(v, d, VXH_INIT_VX_INFO, h);
11668 ++#define __vxh_set_vx_info(v, d, h) \
11669 ++ __vxh_data(v, d, VXH_SET_VX_INFO, h);
11670 ++#define __vxh_clr_vx_info(v, d, h) \
11671 ++ __vxh_data(v, d, VXH_CLR_VX_INFO, h);
11672 ++
11673 ++#define __vxh_claim_vx_info(v, d, h) \
11674 ++ __vxh_data(v, d, VXH_CLAIM_VX_INFO, h);
11675 ++#define __vxh_release_vx_info(v, d, h) \
11676 ++ __vxh_data(v, d, VXH_RELEASE_VX_INFO, h);
11677 ++
11678 ++#define vxh_alloc_vx_info(v) \
11679 ++ __vxh_smpl(v, VXH_ALLOC_VX_INFO, __HERE__);
11680 ++#define vxh_dealloc_vx_info(v) \
11681 ++ __vxh_smpl(v, VXH_DEALLOC_VX_INFO, __HERE__);
11682 ++
11683 ++#define vxh_hash_vx_info(v) \
11684 ++ __vxh_smpl(v, VXH_HASH_VX_INFO, __HERE__);
11685 ++#define vxh_unhash_vx_info(v) \
11686 ++ __vxh_smpl(v, VXH_UNHASH_VX_INFO, __HERE__);
11687 ++
11688 ++#define vxh_loc_vx_info(v, l) \
11689 ++ __vxh_long(v, l, VXH_LOC_VX_INFO, __HERE__);
11690 ++#define vxh_lookup_vx_info(v, l) \
11691 ++ __vxh_long(v, l, VXH_LOOKUP_VX_INFO, __HERE__);
11692 ++#define vxh_create_vx_info(v, l) \
11693 ++ __vxh_long(v, l, VXH_CREATE_VX_INFO, __HERE__);
11694 ++
11695 ++extern void vxh_dump_history(void);
11696 ++
11697 ++
11698 ++#else /* CONFIG_VSERVER_HISTORY */
11699 ++
11700 ++#define __HERE__ 0
11701 ++
11702 ++#define vxh_throw_oops() do { } while (0)
11703 ++
11704 ++#define __vxh_get_vx_info(v, h) do { } while (0)
11705 ++#define __vxh_put_vx_info(v, h) do { } while (0)
11706 ++
11707 ++#define __vxh_init_vx_info(v, d, h) do { } while (0)
11708 ++#define __vxh_set_vx_info(v, d, h) do { } while (0)
11709 ++#define __vxh_clr_vx_info(v, d, h) do { } while (0)
11710 ++
11711 ++#define __vxh_claim_vx_info(v, d, h) do { } while (0)
11712 ++#define __vxh_release_vx_info(v, d, h) do { } while (0)
11713 ++
11714 ++#define vxh_alloc_vx_info(v) do { } while (0)
11715 ++#define vxh_dealloc_vx_info(v) do { } while (0)
11716 ++
11717 ++#define vxh_hash_vx_info(v) do { } while (0)
11718 ++#define vxh_unhash_vx_info(v) do { } while (0)
11719 ++
11720 ++#define vxh_loc_vx_info(v, l) do { } while (0)
11721 ++#define vxh_lookup_vx_info(v, l) do { } while (0)
11722 ++#define vxh_create_vx_info(v, l) do { } while (0)
11723 ++
11724 ++#define vxh_dump_history() do { } while (0)
11725 ++
11726 ++
11727 ++#endif /* CONFIG_VSERVER_HISTORY */
11728 ++
11729 ++#endif /* _VX_HISTORY_H */
11730 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/inode_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/inode_cmd.h
11731 +--- linux-2.6.29.4/include/linux/vserver/inode_cmd.h 1970-01-01 01:00:00.000000000 +0100
11732 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/inode_cmd.h 2009-02-22 22:54:26.000000000 +0100
11733 +@@ -0,0 +1,59 @@
11734 ++#ifndef _VX_INODE_CMD_H
11735 ++#define _VX_INODE_CMD_H
11736 ++
11737 ++
11738 ++/* inode vserver commands */
11739 ++
11740 ++#define VCMD_get_iattr VC_CMD(INODE, 1, 1)
11741 ++#define VCMD_set_iattr VC_CMD(INODE, 2, 1)
11742 ++
11743 ++#define VCMD_fget_iattr VC_CMD(INODE, 3, 0)
11744 ++#define VCMD_fset_iattr VC_CMD(INODE, 4, 0)
11745 ++
11746 ++struct vcmd_ctx_iattr_v1 {
11747 ++ const char __user *name;
11748 ++ uint32_t tag;
11749 ++ uint32_t flags;
11750 ++ uint32_t mask;
11751 ++};
11752 ++
11753 ++struct vcmd_ctx_fiattr_v0 {
11754 ++ uint32_t tag;
11755 ++ uint32_t flags;
11756 ++ uint32_t mask;
11757 ++};
11758 ++
11759 ++
11760 ++#ifdef __KERNEL__
11761 ++
11762 ++
11763 ++#ifdef CONFIG_COMPAT
11764 ++
11765 ++#include <asm/compat.h>
11766 ++
11767 ++struct vcmd_ctx_iattr_v1_x32 {
11768 ++ compat_uptr_t name_ptr;
11769 ++ uint32_t tag;
11770 ++ uint32_t flags;
11771 ++ uint32_t mask;
11772 ++};
11773 ++
11774 ++#endif /* CONFIG_COMPAT */
11775 ++
11776 ++#include <linux/compiler.h>
11777 ++
11778 ++extern int vc_get_iattr(void __user *);
11779 ++extern int vc_set_iattr(void __user *);
11780 ++
11781 ++extern int vc_fget_iattr(uint32_t, void __user *);
11782 ++extern int vc_fset_iattr(uint32_t, void __user *);
11783 ++
11784 ++#ifdef CONFIG_COMPAT
11785 ++
11786 ++extern int vc_get_iattr_x32(void __user *);
11787 ++extern int vc_set_iattr_x32(void __user *);
11788 ++
11789 ++#endif /* CONFIG_COMPAT */
11790 ++
11791 ++#endif /* __KERNEL__ */
11792 ++#endif /* _VX_INODE_CMD_H */
11793 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/inode.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/inode.h
11794 +--- linux-2.6.29.4/include/linux/vserver/inode.h 1970-01-01 01:00:00.000000000 +0100
11795 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/inode.h 2009-02-22 22:54:26.000000000 +0100
11796 +@@ -0,0 +1,38 @@
11797 ++#ifndef _VX_INODE_H
11798 ++#define _VX_INODE_H
11799 ++
11800 ++
11801 ++#define IATTR_TAG 0x01000000
11802 ++
11803 ++#define IATTR_ADMIN 0x00000001
11804 ++#define IATTR_WATCH 0x00000002
11805 ++#define IATTR_HIDE 0x00000004
11806 ++#define IATTR_FLAGS 0x00000007
11807 ++
11808 ++#define IATTR_BARRIER 0x00010000
11809 ++#define IATTR_IXUNLINK 0x00020000
11810 ++#define IATTR_IMMUTABLE 0x00040000
11811 ++
11812 ++#ifdef __KERNEL__
11813 ++
11814 ++
11815 ++#ifdef CONFIG_VSERVER_PROC_SECURE
11816 ++#define IATTR_PROC_DEFAULT ( IATTR_ADMIN | IATTR_HIDE )
11817 ++#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11818 ++#else
11819 ++#define IATTR_PROC_DEFAULT ( IATTR_ADMIN )
11820 ++#define IATTR_PROC_SYMLINK ( IATTR_ADMIN )
11821 ++#endif
11822 ++
11823 ++#define vx_hide_check(c, m) (((m) & IATTR_HIDE) ? vx_check(c, m) : 1)
11824 ++
11825 ++#endif /* __KERNEL__ */
11826 ++
11827 ++/* inode ioctls */
11828 ++
11829 ++#define FIOC_GETXFLG _IOR('x', 5, long)
11830 ++#define FIOC_SETXFLG _IOW('x', 6, long)
11831 ++
11832 ++#else /* _VX_INODE_H */
11833 ++#warning duplicate inclusion
11834 ++#endif /* _VX_INODE_H */
11835 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/Kbuild linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/Kbuild
11836 +--- linux-2.6.29.4/include/linux/vserver/Kbuild 1970-01-01 01:00:00.000000000 +0100
11837 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/Kbuild 2009-02-22 22:54:26.000000000 +0100
11838 +@@ -0,0 +1,8 @@
11839 ++
11840 ++unifdef-y += context_cmd.h network_cmd.h space_cmd.h \
11841 ++ cacct_cmd.h cvirt_cmd.h limit_cmd.h dlimit_cmd.h \
11842 ++ inode_cmd.h tag_cmd.h sched_cmd.h signal_cmd.h \
11843 ++ debug_cmd.h device_cmd.h
11844 ++
11845 ++unifdef-y += switch.h network.h monitor.h inode.h device.h
11846 ++
11847 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/limit_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_cmd.h
11848 +--- linux-2.6.29.4/include/linux/vserver/limit_cmd.h 1970-01-01 01:00:00.000000000 +0100
11849 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_cmd.h 2009-02-22 22:54:26.000000000 +0100
11850 +@@ -0,0 +1,69 @@
11851 ++#ifndef _VX_LIMIT_CMD_H
11852 ++#define _VX_LIMIT_CMD_H
11853 ++
11854 ++
11855 ++/* rlimit vserver commands */
11856 ++
11857 ++#define VCMD_get_rlimit VC_CMD(RLIMIT, 1, 0)
11858 ++#define VCMD_set_rlimit VC_CMD(RLIMIT, 2, 0)
11859 ++#define VCMD_get_rlimit_mask VC_CMD(RLIMIT, 3, 0)
11860 ++#define VCMD_reset_minmax VC_CMD(RLIMIT, 9, 0)
11861 ++
11862 ++struct vcmd_ctx_rlimit_v0 {
11863 ++ uint32_t id;
11864 ++ uint64_t minimum;
11865 ++ uint64_t softlimit;
11866 ++ uint64_t maximum;
11867 ++};
11868 ++
11869 ++struct vcmd_ctx_rlimit_mask_v0 {
11870 ++ uint32_t minimum;
11871 ++ uint32_t softlimit;
11872 ++ uint32_t maximum;
11873 ++};
11874 ++
11875 ++#define VCMD_rlimit_stat VC_CMD(VSTAT, 1, 0)
11876 ++
11877 ++struct vcmd_rlimit_stat_v0 {
11878 ++ uint32_t id;
11879 ++ uint32_t hits;
11880 ++ uint64_t value;
11881 ++ uint64_t minimum;
11882 ++ uint64_t maximum;
11883 ++};
11884 ++
11885 ++#define CRLIM_UNSET (0ULL)
11886 ++#define CRLIM_INFINITY (~0ULL)
11887 ++#define CRLIM_KEEP (~1ULL)
11888 ++
11889 ++#ifdef __KERNEL__
11890 ++
11891 ++#ifdef CONFIG_IA32_EMULATION
11892 ++
11893 ++struct vcmd_ctx_rlimit_v0_x32 {
11894 ++ uint32_t id;
11895 ++ uint64_t minimum;
11896 ++ uint64_t softlimit;
11897 ++ uint64_t maximum;
11898 ++} __attribute__ ((packed));
11899 ++
11900 ++#endif /* CONFIG_IA32_EMULATION */
11901 ++
11902 ++#include <linux/compiler.h>
11903 ++
11904 ++extern int vc_get_rlimit_mask(uint32_t, void __user *);
11905 ++extern int vc_get_rlimit(struct vx_info *, void __user *);
11906 ++extern int vc_set_rlimit(struct vx_info *, void __user *);
11907 ++extern int vc_reset_minmax(struct vx_info *, void __user *);
11908 ++
11909 ++extern int vc_rlimit_stat(struct vx_info *, void __user *);
11910 ++
11911 ++#ifdef CONFIG_IA32_EMULATION
11912 ++
11913 ++extern int vc_get_rlimit_x32(struct vx_info *, void __user *);
11914 ++extern int vc_set_rlimit_x32(struct vx_info *, void __user *);
11915 ++
11916 ++#endif /* CONFIG_IA32_EMULATION */
11917 ++
11918 ++#endif /* __KERNEL__ */
11919 ++#endif /* _VX_LIMIT_CMD_H */
11920 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/limit_def.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_def.h
11921 +--- linux-2.6.29.4/include/linux/vserver/limit_def.h 1970-01-01 01:00:00.000000000 +0100
11922 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_def.h 2009-02-22 22:54:26.000000000 +0100
11923 +@@ -0,0 +1,47 @@
11924 ++#ifndef _VX_LIMIT_DEF_H
11925 ++#define _VX_LIMIT_DEF_H
11926 ++
11927 ++#include <asm/atomic.h>
11928 ++#include <asm/resource.h>
11929 ++
11930 ++#include "limit.h"
11931 ++
11932 ++
11933 ++struct _vx_res_limit {
11934 ++ rlim_t soft; /* Context soft limit */
11935 ++ rlim_t hard; /* Context hard limit */
11936 ++
11937 ++ rlim_atomic_t rcur; /* Current value */
11938 ++ rlim_t rmin; /* Context minimum */
11939 ++ rlim_t rmax; /* Context maximum */
11940 ++
11941 ++ atomic_t lhit; /* Limit hits */
11942 ++};
11943 ++
11944 ++/* context sub struct */
11945 ++
11946 ++struct _vx_limit {
11947 ++ struct _vx_res_limit res[NUM_LIMITS];
11948 ++};
11949 ++
11950 ++#ifdef CONFIG_VSERVER_DEBUG
11951 ++
11952 ++static inline void __dump_vx_limit(struct _vx_limit *limit)
11953 ++{
11954 ++ int i;
11955 ++
11956 ++ printk("\t_vx_limit:");
11957 ++ for (i = 0; i < NUM_LIMITS; i++) {
11958 ++ printk("\t [%2d] = %8lu %8lu/%8lu, %8ld/%8ld, %8d\n",
11959 ++ i, (unsigned long)__rlim_get(limit, i),
11960 ++ (unsigned long)__rlim_rmin(limit, i),
11961 ++ (unsigned long)__rlim_rmax(limit, i),
11962 ++ (long)__rlim_soft(limit, i),
11963 ++ (long)__rlim_hard(limit, i),
11964 ++ atomic_read(&__rlim_lhit(limit, i)));
11965 ++ }
11966 ++}
11967 ++
11968 ++#endif
11969 ++
11970 ++#endif /* _VX_LIMIT_DEF_H */
11971 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/limit.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit.h
11972 +--- linux-2.6.29.4/include/linux/vserver/limit.h 1970-01-01 01:00:00.000000000 +0100
11973 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit.h 2009-02-22 22:54:26.000000000 +0100
11974 +@@ -0,0 +1,70 @@
11975 ++#ifndef _VX_LIMIT_H
11976 ++#define _VX_LIMIT_H
11977 ++
11978 ++#define VLIMIT_NSOCK 16
11979 ++#define VLIMIT_OPENFD 17
11980 ++#define VLIMIT_ANON 18
11981 ++#define VLIMIT_SHMEM 19
11982 ++#define VLIMIT_SEMARY 20
11983 ++#define VLIMIT_NSEMS 21
11984 ++#define VLIMIT_DENTRY 22
11985 ++#define VLIMIT_MAPPED 23
11986 ++
11987 ++
11988 ++#ifdef __KERNEL__
11989 ++
11990 ++#define VLIM_NOCHECK ((1L << VLIMIT_DENTRY) | (1L << RLIMIT_RSS))
11991 ++
11992 ++/* keep in sync with CRLIM_INFINITY */
11993 ++
11994 ++#define VLIM_INFINITY (~0ULL)
11995 ++
11996 ++#include <asm/atomic.h>
11997 ++#include <asm/resource.h>
11998 ++
11999 ++#ifndef RLIM_INFINITY
12000 ++#warning RLIM_INFINITY is undefined
12001 ++#endif
12002 ++
12003 ++#define __rlim_val(l, r, v) ((l)->res[r].v)
12004 ++
12005 ++#define __rlim_soft(l, r) __rlim_val(l, r, soft)
12006 ++#define __rlim_hard(l, r) __rlim_val(l, r, hard)
12007 ++
12008 ++#define __rlim_rcur(l, r) __rlim_val(l, r, rcur)
12009 ++#define __rlim_rmin(l, r) __rlim_val(l, r, rmin)
12010 ++#define __rlim_rmax(l, r) __rlim_val(l, r, rmax)
12011 ++
12012 ++#define __rlim_lhit(l, r) __rlim_val(l, r, lhit)
12013 ++#define __rlim_hit(l, r) atomic_inc(&__rlim_lhit(l, r))
12014 ++
12015 ++typedef atomic_long_t rlim_atomic_t;
12016 ++typedef unsigned long rlim_t;
12017 ++
12018 ++#define __rlim_get(l, r) atomic_long_read(&__rlim_rcur(l, r))
12019 ++#define __rlim_set(l, r, v) atomic_long_set(&__rlim_rcur(l, r), v)
12020 ++#define __rlim_inc(l, r) atomic_long_inc(&__rlim_rcur(l, r))
12021 ++#define __rlim_dec(l, r) atomic_long_dec(&__rlim_rcur(l, r))
12022 ++#define __rlim_add(l, r, v) atomic_long_add(v, &__rlim_rcur(l, r))
12023 ++#define __rlim_sub(l, r, v) atomic_long_sub(v, &__rlim_rcur(l, r))
12024 ++
12025 ++
12026 ++#if (RLIM_INFINITY == VLIM_INFINITY)
12027 ++#define VX_VLIM(r) ((long long)(long)(r))
12028 ++#define VX_RLIM(v) ((rlim_t)(v))
12029 ++#else
12030 ++#define VX_VLIM(r) (((r) == RLIM_INFINITY) \
12031 ++ ? VLIM_INFINITY : (long long)(r))
12032 ++#define VX_RLIM(v) (((v) == VLIM_INFINITY) \
12033 ++ ? RLIM_INFINITY : (rlim_t)(v))
12034 ++#endif
12035 ++
12036 ++struct sysinfo;
12037 ++
12038 ++void vx_vsi_meminfo(struct sysinfo *);
12039 ++void vx_vsi_swapinfo(struct sysinfo *);
12040 ++
12041 ++#define NUM_LIMITS 24
12042 ++
12043 ++#endif /* __KERNEL__ */
12044 ++#endif /* _VX_LIMIT_H */
12045 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/limit_int.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_int.h
12046 +--- linux-2.6.29.4/include/linux/vserver/limit_int.h 1970-01-01 01:00:00.000000000 +0100
12047 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/limit_int.h 2009-02-22 22:54:26.000000000 +0100
12048 +@@ -0,0 +1,198 @@
12049 ++#ifndef _VX_LIMIT_INT_H
12050 ++#define _VX_LIMIT_INT_H
12051 ++
12052 ++#include "context.h"
12053 ++
12054 ++#ifdef __KERNEL__
12055 ++
12056 ++#define VXD_RCRES_COND(r) VXD_CBIT(cres, r)
12057 ++#define VXD_RLIMIT_COND(r) VXD_CBIT(limit, r)
12058 ++
12059 ++extern const char *vlimit_name[NUM_LIMITS];
12060 ++
12061 ++static inline void __vx_acc_cres(struct vx_info *vxi,
12062 ++ int res, int dir, void *_data, char *_file, int _line)
12063 ++{
12064 ++ if (VXD_RCRES_COND(res))
12065 ++ vxlprintk(1, "vx_acc_cres[%5d,%s,%2d]: %5ld%s (%p)",
12066 ++ (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
12067 ++ (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
12068 ++ (dir > 0) ? "++" : "--", _data, _file, _line);
12069 ++ if (!vxi)
12070 ++ return;
12071 ++
12072 ++ if (dir > 0)
12073 ++ __rlim_inc(&vxi->limit, res);
12074 ++ else
12075 ++ __rlim_dec(&vxi->limit, res);
12076 ++}
12077 ++
12078 ++static inline void __vx_add_cres(struct vx_info *vxi,
12079 ++ int res, int amount, void *_data, char *_file, int _line)
12080 ++{
12081 ++ if (VXD_RCRES_COND(res))
12082 ++ vxlprintk(1, "vx_add_cres[%5d,%s,%2d]: %5ld += %5d (%p)",
12083 ++ (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
12084 ++ (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
12085 ++ amount, _data, _file, _line);
12086 ++ if (amount == 0)
12087 ++ return;
12088 ++ if (!vxi)
12089 ++ return;
12090 ++ __rlim_add(&vxi->limit, res, amount);
12091 ++}
12092 ++
12093 ++static inline
12094 ++int __vx_cres_adjust_max(struct _vx_limit *limit, int res, rlim_t value)
12095 ++{
12096 ++ int cond = (value > __rlim_rmax(limit, res));
12097 ++
12098 ++ if (cond)
12099 ++ __rlim_rmax(limit, res) = value;
12100 ++ return cond;
12101 ++}
12102 ++
12103 ++static inline
12104 ++int __vx_cres_adjust_min(struct _vx_limit *limit, int res, rlim_t value)
12105 ++{
12106 ++ int cond = (value < __rlim_rmin(limit, res));
12107 ++
12108 ++ if (cond)
12109 ++ __rlim_rmin(limit, res) = value;
12110 ++ return cond;
12111 ++}
12112 ++
12113 ++static inline
12114 ++void __vx_cres_fixup(struct _vx_limit *limit, int res, rlim_t value)
12115 ++{
12116 ++ if (!__vx_cres_adjust_max(limit, res, value))
12117 ++ __vx_cres_adjust_min(limit, res, value);
12118 ++}
12119 ++
12120 ++
12121 ++/* return values:
12122 ++ +1 ... no limit hit
12123 ++ -1 ... over soft limit
12124 ++ 0 ... over hard limit */
12125 ++
12126 ++static inline int __vx_cres_avail(struct vx_info *vxi,
12127 ++ int res, int num, char *_file, int _line)
12128 ++{
12129 ++ struct _vx_limit *limit;
12130 ++ rlim_t value;
12131 ++
12132 ++ if (VXD_RLIMIT_COND(res))
12133 ++ vxlprintk(1, "vx_cres_avail[%5d,%s,%2d]: %5ld/%5ld > %5ld + %5d",
12134 ++ (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
12135 ++ (vxi ? (long)__rlim_soft(&vxi->limit, res) : -1),
12136 ++ (vxi ? (long)__rlim_hard(&vxi->limit, res) : -1),
12137 ++ (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
12138 ++ num, _file, _line);
12139 ++ if (!vxi)
12140 ++ return 1;
12141 ++
12142 ++ limit = &vxi->limit;
12143 ++ value = __rlim_get(limit, res);
12144 ++
12145 ++ if (!__vx_cres_adjust_max(limit, res, value))
12146 ++ __vx_cres_adjust_min(limit, res, value);
12147 ++
12148 ++ if (num == 0)
12149 ++ return 1;
12150 ++
12151 ++ if (__rlim_soft(limit, res) == RLIM_INFINITY)
12152 ++ return -1;
12153 ++ if (value + num <= __rlim_soft(limit, res))
12154 ++ return -1;
12155 ++
12156 ++ if (__rlim_hard(limit, res) == RLIM_INFINITY)
12157 ++ return 1;
12158 ++ if (value + num <= __rlim_hard(limit, res))
12159 ++ return 1;
12160 ++
12161 ++ __rlim_hit(limit, res);
12162 ++ return 0;
12163 ++}
12164 ++
12165 ++
12166 ++static const int VLA_RSS[] = { RLIMIT_RSS, VLIMIT_ANON, VLIMIT_MAPPED, 0 };
12167 ++
12168 ++static inline
12169 ++rlim_t __vx_cres_array_sum(struct _vx_limit *limit, const int *array)
12170 ++{
12171 ++ rlim_t value, sum = 0;
12172 ++ int res;
12173 ++
12174 ++ while ((res = *array++)) {
12175 ++ value = __rlim_get(limit, res);
12176 ++ __vx_cres_fixup(limit, res, value);
12177 ++ sum += value;
12178 ++ }
12179 ++ return sum;
12180 ++}
12181 ++
12182 ++static inline
12183 ++rlim_t __vx_cres_array_fixup(struct _vx_limit *limit, const int *array)
12184 ++{
12185 ++ rlim_t value = __vx_cres_array_sum(limit, array + 1);
12186 ++ int res = *array;
12187 ++
12188 ++ if (value == __rlim_get(limit, res))
12189 ++ return value;
12190 ++
12191 ++ __rlim_set(limit, res, value);
12192 ++ /* now adjust min/max */
12193 ++ if (!__vx_cres_adjust_max(limit, res, value))
12194 ++ __vx_cres_adjust_min(limit, res, value);
12195 ++
12196 ++ return value;
12197 ++}
12198 ++
12199 ++static inline int __vx_cres_array_avail(struct vx_info *vxi,
12200 ++ const int *array, int num, char *_file, int _line)
12201 ++{
12202 ++ struct _vx_limit *limit;
12203 ++ rlim_t value = 0;
12204 ++ int res;
12205 ++
12206 ++ if (num == 0)
12207 ++ return 1;
12208 ++ if (!vxi)
12209 ++ return 1;
12210 ++
12211 ++ limit = &vxi->limit;
12212 ++ res = *array;
12213 ++ value = __vx_cres_array_sum(limit, array + 1);
12214 ++
12215 ++ __rlim_set(limit, res, value);
12216 ++ __vx_cres_fixup(limit, res, value);
12217 ++
12218 ++ return __vx_cres_avail(vxi, res, num, _file, _line);
12219 ++}
12220 ++
12221 ++
12222 ++static inline void vx_limit_fixup(struct _vx_limit *limit, int id)
12223 ++{
12224 ++ rlim_t value;
12225 ++ int res;
12226 ++
12227 ++ /* complex resources first */
12228 ++ if ((id < 0) || (id == RLIMIT_RSS))
12229 ++ __vx_cres_array_fixup(limit, VLA_RSS);
12230 ++
12231 ++ for (res = 0; res < NUM_LIMITS; res++) {
12232 ++ if ((id > 0) && (res != id))
12233 ++ continue;
12234 ++
12235 ++ value = __rlim_get(limit, res);
12236 ++ __vx_cres_fixup(limit, res, value);
12237 ++
12238 ++ /* not supposed to happen, maybe warn? */
12239 ++ if (__rlim_rmax(limit, res) > __rlim_hard(limit, res))
12240 ++ __rlim_rmax(limit, res) = __rlim_hard(limit, res);
12241 ++ }
12242 ++}
12243 ++
12244 ++
12245 ++#endif /* __KERNEL__ */
12246 ++#endif /* _VX_LIMIT_INT_H */
12247 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/monitor.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/monitor.h
12248 +--- linux-2.6.29.4/include/linux/vserver/monitor.h 1970-01-01 01:00:00.000000000 +0100
12249 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/monitor.h 2009-02-22 22:54:26.000000000 +0100
12250 +@@ -0,0 +1,96 @@
12251 ++#ifndef _VX_MONITOR_H
12252 ++#define _VX_MONITOR_H
12253 ++
12254 ++#include <linux/types.h>
12255 ++
12256 ++enum {
12257 ++ VXM_UNUSED = 0,
12258 ++
12259 ++ VXM_SYNC = 0x10,
12260 ++
12261 ++ VXM_UPDATE = 0x20,
12262 ++ VXM_UPDATE_1,
12263 ++ VXM_UPDATE_2,
12264 ++
12265 ++ VXM_RQINFO_1 = 0x24,
12266 ++ VXM_RQINFO_2,
12267 ++
12268 ++ VXM_ACTIVATE = 0x40,
12269 ++ VXM_DEACTIVATE,
12270 ++ VXM_IDLE,
12271 ++
12272 ++ VXM_HOLD = 0x44,
12273 ++ VXM_UNHOLD,
12274 ++
12275 ++ VXM_MIGRATE = 0x48,
12276 ++ VXM_RESCHED,
12277 ++
12278 ++ /* all other bits are flags */
12279 ++ VXM_SCHED = 0x80,
12280 ++};
12281 ++
12282 ++struct _vxm_update_1 {
12283 ++ uint32_t tokens_max;
12284 ++ uint32_t fill_rate;
12285 ++ uint32_t interval;
12286 ++};
12287 ++
12288 ++struct _vxm_update_2 {
12289 ++ uint32_t tokens_min;
12290 ++ uint32_t fill_rate;
12291 ++ uint32_t interval;
12292 ++};
12293 ++
12294 ++struct _vxm_rqinfo_1 {
12295 ++ uint16_t running;
12296 ++ uint16_t onhold;
12297 ++ uint16_t iowait;
12298 ++ uint16_t uintr;
12299 ++ uint32_t idle_tokens;
12300 ++};
12301 ++
12302 ++struct _vxm_rqinfo_2 {
12303 ++ uint32_t norm_time;
12304 ++ uint32_t idle_time;
12305 ++ uint32_t idle_skip;
12306 ++};
12307 ++
12308 ++struct _vxm_sched {
12309 ++ uint32_t tokens;
12310 ++ uint32_t norm_time;
12311 ++ uint32_t idle_time;
12312 ++};
12313 ++
12314 ++struct _vxm_task {
12315 ++ uint16_t pid;
12316 ++ uint16_t state;
12317 ++};
12318 ++
12319 ++struct _vxm_event {
12320 ++ uint32_t jif;
12321 ++ union {
12322 ++ uint32_t seq;
12323 ++ uint32_t sec;
12324 ++ };
12325 ++ union {
12326 ++ uint32_t tokens;
12327 ++ uint32_t nsec;
12328 ++ struct _vxm_task tsk;
12329 ++ };
12330 ++};
12331 ++
12332 ++struct _vx_mon_entry {
12333 ++ uint16_t type;
12334 ++ uint16_t xid;
12335 ++ union {
12336 ++ struct _vxm_event ev;
12337 ++ struct _vxm_sched sd;
12338 ++ struct _vxm_update_1 u1;
12339 ++ struct _vxm_update_2 u2;
12340 ++ struct _vxm_rqinfo_1 q1;
12341 ++ struct _vxm_rqinfo_2 q2;
12342 ++ };
12343 ++};
12344 ++
12345 ++
12346 ++#endif /* _VX_MONITOR_H */
12347 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/network_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/network_cmd.h
12348 +--- linux-2.6.29.4/include/linux/vserver/network_cmd.h 1970-01-01 01:00:00.000000000 +0100
12349 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/network_cmd.h 2009-02-22 22:54:26.000000000 +0100
12350 +@@ -0,0 +1,150 @@
12351 ++#ifndef _VX_NETWORK_CMD_H
12352 ++#define _VX_NETWORK_CMD_H
12353 ++
12354 ++
12355 ++/* vinfo commands */
12356 ++
12357 ++#define VCMD_task_nid VC_CMD(VINFO, 2, 0)
12358 ++
12359 ++#ifdef __KERNEL__
12360 ++extern int vc_task_nid(uint32_t);
12361 ++
12362 ++#endif /* __KERNEL__ */
12363 ++
12364 ++#define VCMD_nx_info VC_CMD(VINFO, 6, 0)
12365 ++
12366 ++struct vcmd_nx_info_v0 {
12367 ++ uint32_t nid;
12368 ++ /* more to come */
12369 ++};
12370 ++
12371 ++#ifdef __KERNEL__
12372 ++extern int vc_nx_info(struct nx_info *, void __user *);
12373 ++
12374 ++#endif /* __KERNEL__ */
12375 ++
12376 ++#include <linux/in.h>
12377 ++#include <linux/in6.h>
12378 ++
12379 ++#define VCMD_net_create_v0 VC_CMD(VNET, 1, 0)
12380 ++#define VCMD_net_create VC_CMD(VNET, 1, 1)
12381 ++
12382 ++struct vcmd_net_create {
12383 ++ uint64_t flagword;
12384 ++};
12385 ++
12386 ++#define VCMD_net_migrate VC_CMD(NETMIG, 1, 0)
12387 ++
12388 ++#define VCMD_net_add VC_CMD(NETALT, 1, 0)
12389 ++#define VCMD_net_remove VC_CMD(NETALT, 2, 0)
12390 ++
12391 ++struct vcmd_net_addr_v0 {
12392 ++ uint16_t type;
12393 ++ uint16_t count;
12394 ++ struct in_addr ip[4];
12395 ++ struct in_addr mask[4];
12396 ++};
12397 ++
12398 ++#define VCMD_net_add_ipv4 VC_CMD(NETALT, 1, 1)
12399 ++#define VCMD_net_remove_ipv4 VC_CMD(NETALT, 2, 1)
12400 ++
12401 ++struct vcmd_net_addr_ipv4_v1 {
12402 ++ uint16_t type;
12403 ++ uint16_t flags;
12404 ++ struct in_addr ip;
12405 ++ struct in_addr mask;
12406 ++};
12407 ++
12408 ++#define VCMD_net_add_ipv6 VC_CMD(NETALT, 3, 1)
12409 ++#define VCMD_net_remove_ipv6 VC_CMD(NETALT, 4, 1)
12410 ++
12411 ++struct vcmd_net_addr_ipv6_v1 {
12412 ++ uint16_t type;
12413 ++ uint16_t flags;
12414 ++ uint32_t prefix;
12415 ++ struct in6_addr ip;
12416 ++ struct in6_addr mask;
12417 ++};
12418 ++
12419 ++#define VCMD_add_match_ipv4 VC_CMD(NETALT, 5, 0)
12420 ++#define VCMD_get_match_ipv4 VC_CMD(NETALT, 6, 0)
12421 ++
12422 ++struct vcmd_match_ipv4_v0 {
12423 ++ uint16_t type;
12424 ++ uint16_t flags;
12425 ++ uint16_t parent;
12426 ++ uint16_t prefix;
12427 ++ struct in_addr ip;
12428 ++ struct in_addr ip2;
12429 ++ struct in_addr mask;
12430 ++};
12431 ++
12432 ++#define VCMD_add_match_ipv6 VC_CMD(NETALT, 7, 0)
12433 ++#define VCMD_get_match_ipv6 VC_CMD(NETALT, 8, 0)
12434 ++
12435 ++struct vcmd_match_ipv6_v0 {
12436 ++ uint16_t type;
12437 ++ uint16_t flags;
12438 ++ uint16_t parent;
12439 ++ uint16_t prefix;
12440 ++ struct in6_addr ip;
12441 ++ struct in6_addr ip2;
12442 ++ struct in6_addr mask;
12443 ++};
12444 ++
12445 ++
12446 ++#ifdef __KERNEL__
12447 ++extern int vc_net_create(uint32_t, void __user *);
12448 ++extern int vc_net_migrate(struct nx_info *, void __user *);
12449 ++
12450 ++extern int vc_net_add(struct nx_info *, void __user *);
12451 ++extern int vc_net_remove(struct nx_info *, void __user *);
12452 ++
12453 ++extern int vc_net_add_ipv4(struct nx_info *, void __user *);
12454 ++extern int vc_net_remove_ipv4(struct nx_info *, void __user *);
12455 ++
12456 ++extern int vc_net_add_ipv6(struct nx_info *, void __user *);
12457 ++extern int vc_net_remove_ipv6(struct nx_info *, void __user *);
12458 ++
12459 ++extern int vc_add_match_ipv4(struct nx_info *, void __user *);
12460 ++extern int vc_get_match_ipv4(struct nx_info *, void __user *);
12461 ++
12462 ++extern int vc_add_match_ipv6(struct nx_info *, void __user *);
12463 ++extern int vc_get_match_ipv6(struct nx_info *, void __user *);
12464 ++
12465 ++#endif /* __KERNEL__ */
12466 ++
12467 ++
12468 ++/* flag commands */
12469 ++
12470 ++#define VCMD_get_nflags VC_CMD(FLAGS, 5, 0)
12471 ++#define VCMD_set_nflags VC_CMD(FLAGS, 6, 0)
12472 ++
12473 ++struct vcmd_net_flags_v0 {
12474 ++ uint64_t flagword;
12475 ++ uint64_t mask;
12476 ++};
12477 ++
12478 ++#ifdef __KERNEL__
12479 ++extern int vc_get_nflags(struct nx_info *, void __user *);
12480 ++extern int vc_set_nflags(struct nx_info *, void __user *);
12481 ++
12482 ++#endif /* __KERNEL__ */
12483 ++
12484 ++
12485 ++/* network caps commands */
12486 ++
12487 ++#define VCMD_get_ncaps VC_CMD(FLAGS, 7, 0)
12488 ++#define VCMD_set_ncaps VC_CMD(FLAGS, 8, 0)
12489 ++
12490 ++struct vcmd_net_caps_v0 {
12491 ++ uint64_t ncaps;
12492 ++ uint64_t cmask;
12493 ++};
12494 ++
12495 ++#ifdef __KERNEL__
12496 ++extern int vc_get_ncaps(struct nx_info *, void __user *);
12497 ++extern int vc_set_ncaps(struct nx_info *, void __user *);
12498 ++
12499 ++#endif /* __KERNEL__ */
12500 ++#endif /* _VX_CONTEXT_CMD_H */
12501 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/network.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/network.h
12502 +--- linux-2.6.29.4/include/linux/vserver/network.h 1970-01-01 01:00:00.000000000 +0100
12503 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/network.h 2009-02-22 22:54:26.000000000 +0100
12504 +@@ -0,0 +1,146 @@
12505 ++#ifndef _VX_NETWORK_H
12506 ++#define _VX_NETWORK_H
12507 ++
12508 ++#include <linux/types.h>
12509 ++
12510 ++
12511 ++#define MAX_N_CONTEXT 65535 /* Arbitrary limit */
12512 ++
12513 ++
12514 ++/* network flags */
12515 ++
12516 ++#define NXF_INFO_PRIVATE 0x00000008
12517 ++
12518 ++#define NXF_SINGLE_IP 0x00000100
12519 ++#define NXF_LBACK_REMAP 0x00000200
12520 ++#define NXF_LBACK_ALLOW 0x00000400
12521 ++
12522 ++#define NXF_HIDE_NETIF 0x02000000
12523 ++#define NXF_HIDE_LBACK 0x04000000
12524 ++
12525 ++#define NXF_STATE_SETUP (1ULL << 32)
12526 ++#define NXF_STATE_ADMIN (1ULL << 34)
12527 ++
12528 ++#define NXF_SC_HELPER (1ULL << 36)
12529 ++#define NXF_PERSISTENT (1ULL << 38)
12530 ++
12531 ++#define NXF_ONE_TIME (0x0005ULL << 32)
12532 ++
12533 ++
12534 ++#define NXF_INIT_SET (__nxf_init_set())
12535 ++
12536 ++static inline uint64_t __nxf_init_set(void) {
12537 ++ return NXF_STATE_ADMIN
12538 ++#ifdef CONFIG_VSERVER_AUTO_LBACK
12539 ++ | NXF_LBACK_REMAP
12540 ++ | NXF_HIDE_LBACK
12541 ++#endif
12542 ++#ifdef CONFIG_VSERVER_AUTO_SINGLE
12543 ++ | NXF_SINGLE_IP
12544 ++#endif
12545 ++ | NXF_HIDE_NETIF;
12546 ++}
12547 ++
12548 ++
12549 ++/* network caps */
12550 ++
12551 ++#define NXC_TUN_CREATE 0x00000001
12552 ++
12553 ++#define NXC_RAW_ICMP 0x00000100
12554 ++
12555 ++
12556 ++/* address types */
12557 ++
12558 ++#define NXA_TYPE_IPV4 0x0001
12559 ++#define NXA_TYPE_IPV6 0x0002
12560 ++
12561 ++#define NXA_TYPE_NONE 0x0000
12562 ++#define NXA_TYPE_ANY 0x00FF
12563 ++
12564 ++#define NXA_TYPE_ADDR 0x0010
12565 ++#define NXA_TYPE_MASK 0x0020
12566 ++#define NXA_TYPE_RANGE 0x0040
12567 ++
12568 ++#define NXA_MASK_ALL (NXA_TYPE_ADDR | NXA_TYPE_MASK | NXA_TYPE_RANGE)
12569 ++
12570 ++#define NXA_MOD_BCAST 0x0100
12571 ++#define NXA_MOD_LBACK 0x0200
12572 ++
12573 ++#define NXA_LOOPBACK 0x1000
12574 ++
12575 ++#define NXA_MASK_BIND (NXA_MASK_ALL | NXA_MOD_BCAST | NXA_MOD_LBACK)
12576 ++#define NXA_MASK_SHOW (NXA_MASK_ALL | NXA_LOOPBACK)
12577 ++
12578 ++#ifdef __KERNEL__
12579 ++
12580 ++#include <linux/list.h>
12581 ++#include <linux/spinlock.h>
12582 ++#include <linux/rcupdate.h>
12583 ++#include <linux/in.h>
12584 ++#include <linux/in6.h>
12585 ++#include <asm/atomic.h>
12586 ++
12587 ++struct nx_addr_v4 {
12588 ++ struct nx_addr_v4 *next;
12589 ++ struct in_addr ip[2];
12590 ++ struct in_addr mask;
12591 ++ uint16_t type;
12592 ++ uint16_t flags;
12593 ++};
12594 ++
12595 ++struct nx_addr_v6 {
12596 ++ struct nx_addr_v6 *next;
12597 ++ struct in6_addr ip;
12598 ++ struct in6_addr mask;
12599 ++ uint32_t prefix;
12600 ++ uint16_t type;
12601 ++ uint16_t flags;
12602 ++};
12603 ++
12604 ++struct nx_info {
12605 ++ struct hlist_node nx_hlist; /* linked list of nxinfos */
12606 ++ nid_t nx_id; /* vnet id */
12607 ++ atomic_t nx_usecnt; /* usage count */
12608 ++ atomic_t nx_tasks; /* tasks count */
12609 ++ int nx_state; /* context state */
12610 ++
12611 ++ uint64_t nx_flags; /* network flag word */
12612 ++ uint64_t nx_ncaps; /* network capabilities */
12613 ++
12614 ++ struct in_addr v4_lback; /* Loopback address */
12615 ++ struct in_addr v4_bcast; /* Broadcast address */
12616 ++ struct nx_addr_v4 v4; /* First/Single ipv4 address */
12617 ++#ifdef CONFIG_IPV6
12618 ++ struct nx_addr_v6 v6; /* First/Single ipv6 address */
12619 ++#endif
12620 ++ char nx_name[65]; /* network context name */
12621 ++};
12622 ++
12623 ++
12624 ++/* status flags */
12625 ++
12626 ++#define NXS_HASHED 0x0001
12627 ++#define NXS_SHUTDOWN 0x0100
12628 ++#define NXS_RELEASED 0x8000
12629 ++
12630 ++extern struct nx_info *lookup_nx_info(int);
12631 ++
12632 ++extern int get_nid_list(int, unsigned int *, int);
12633 ++extern int nid_is_hashed(nid_t);
12634 ++
12635 ++extern int nx_migrate_task(struct task_struct *, struct nx_info *);
12636 ++
12637 ++extern long vs_net_change(struct nx_info *, unsigned int);
12638 ++
12639 ++struct sock;
12640 ++
12641 ++
12642 ++#define NX_IPV4(n) ((n)->v4.type != NXA_TYPE_NONE)
12643 ++#ifdef CONFIG_IPV6
12644 ++#define NX_IPV6(n) ((n)->v6.type != NXA_TYPE_NONE)
12645 ++#else
12646 ++#define NX_IPV6(n) (0)
12647 ++#endif
12648 ++
12649 ++#endif /* __KERNEL__ */
12650 ++#endif /* _VX_NETWORK_H */
12651 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/percpu.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/percpu.h
12652 +--- linux-2.6.29.4/include/linux/vserver/percpu.h 1970-01-01 01:00:00.000000000 +0100
12653 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/percpu.h 2009-02-22 22:54:26.000000000 +0100
12654 +@@ -0,0 +1,14 @@
12655 ++#ifndef _VX_PERCPU_H
12656 ++#define _VX_PERCPU_H
12657 ++
12658 ++#include "cvirt_def.h"
12659 ++#include "sched_def.h"
12660 ++
12661 ++struct _vx_percpu {
12662 ++ struct _vx_cvirt_pc cvirt;
12663 ++ struct _vx_sched_pc sched;
12664 ++};
12665 ++
12666 ++#define PERCPU_PERCTX (sizeof(struct _vx_percpu))
12667 ++
12668 ++#endif /* _VX_PERCPU_H */
12669 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/pid.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/pid.h
12670 +--- linux-2.6.29.4/include/linux/vserver/pid.h 1970-01-01 01:00:00.000000000 +0100
12671 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/pid.h 2009-02-22 22:54:26.000000000 +0100
12672 +@@ -0,0 +1,51 @@
12673 ++#ifndef _VSERVER_PID_H
12674 ++#define _VSERVER_PID_H
12675 ++
12676 ++/* pid faking stuff */
12677 ++
12678 ++#define vx_info_map_pid(v, p) \
12679 ++ __vx_info_map_pid((v), (p), __func__, __FILE__, __LINE__)
12680 ++#define vx_info_map_tgid(v,p) vx_info_map_pid(v,p)
12681 ++#define vx_map_pid(p) vx_info_map_pid(current->vx_info, p)
12682 ++#define vx_map_tgid(p) vx_map_pid(p)
12683 ++
12684 ++static inline int __vx_info_map_pid(struct vx_info *vxi, int pid,
12685 ++ const char *func, const char *file, int line)
12686 ++{
12687 ++ if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
12688 ++ vxfprintk(VXD_CBIT(cvirt, 2),
12689 ++ "vx_map_tgid: %p/%llx: %d -> %d",
12690 ++ vxi, (long long)vxi->vx_flags, pid,
12691 ++ (pid && pid == vxi->vx_initpid) ? 1 : pid,
12692 ++ func, file, line);
12693 ++ if (pid == 0)
12694 ++ return 0;
12695 ++ if (pid == vxi->vx_initpid)
12696 ++ return 1;
12697 ++ }
12698 ++ return pid;
12699 ++}
12700 ++
12701 ++#define vx_info_rmap_pid(v, p) \
12702 ++ __vx_info_rmap_pid((v), (p), __func__, __FILE__, __LINE__)
12703 ++#define vx_rmap_pid(p) vx_info_rmap_pid(current->vx_info, p)
12704 ++#define vx_rmap_tgid(p) vx_rmap_pid(p)
12705 ++
12706 ++static inline int __vx_info_rmap_pid(struct vx_info *vxi, int pid,
12707 ++ const char *func, const char *file, int line)
12708 ++{
12709 ++ if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
12710 ++ vxfprintk(VXD_CBIT(cvirt, 2),
12711 ++ "vx_rmap_tgid: %p/%llx: %d -> %d",
12712 ++ vxi, (long long)vxi->vx_flags, pid,
12713 ++ (pid == 1) ? vxi->vx_initpid : pid,
12714 ++ func, file, line);
12715 ++ if ((pid == 1) && vxi->vx_initpid)
12716 ++ return vxi->vx_initpid;
12717 ++ if (pid == vxi->vx_initpid)
12718 ++ return ~0U;
12719 ++ }
12720 ++ return pid;
12721 ++}
12722 ++
12723 ++#endif
12724 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/sched_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched_cmd.h
12725 +--- linux-2.6.29.4/include/linux/vserver/sched_cmd.h 1970-01-01 01:00:00.000000000 +0100
12726 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched_cmd.h 2009-02-22 22:54:26.000000000 +0100
12727 +@@ -0,0 +1,108 @@
12728 ++#ifndef _VX_SCHED_CMD_H
12729 ++#define _VX_SCHED_CMD_H
12730 ++
12731 ++
12732 ++/* sched vserver commands */
12733 ++
12734 ++#define VCMD_set_sched_v2 VC_CMD(SCHED, 1, 2)
12735 ++#define VCMD_set_sched_v3 VC_CMD(SCHED, 1, 3)
12736 ++#define VCMD_set_sched_v4 VC_CMD(SCHED, 1, 4)
12737 ++
12738 ++struct vcmd_set_sched_v2 {
12739 ++ int32_t fill_rate;
12740 ++ int32_t interval;
12741 ++ int32_t tokens;
12742 ++ int32_t tokens_min;
12743 ++ int32_t tokens_max;
12744 ++ uint64_t cpu_mask;
12745 ++};
12746 ++
12747 ++struct vcmd_set_sched_v3 {
12748 ++ uint32_t set_mask;
12749 ++ int32_t fill_rate;
12750 ++ int32_t interval;
12751 ++ int32_t tokens;
12752 ++ int32_t tokens_min;
12753 ++ int32_t tokens_max;
12754 ++ int32_t priority_bias;
12755 ++};
12756 ++
12757 ++struct vcmd_set_sched_v4 {
12758 ++ uint32_t set_mask;
12759 ++ int32_t fill_rate;
12760 ++ int32_t interval;
12761 ++ int32_t tokens;
12762 ++ int32_t tokens_min;
12763 ++ int32_t tokens_max;
12764 ++ int32_t prio_bias;
12765 ++ int32_t cpu_id;
12766 ++ int32_t bucket_id;
12767 ++};
12768 ++
12769 ++#define VCMD_set_sched VC_CMD(SCHED, 1, 5)
12770 ++#define VCMD_get_sched VC_CMD(SCHED, 2, 5)
12771 ++
12772 ++struct vcmd_sched_v5 {
12773 ++ uint32_t mask;
12774 ++ int32_t cpu_id;
12775 ++ int32_t bucket_id;
12776 ++ int32_t fill_rate[2];
12777 ++ int32_t interval[2];
12778 ++ int32_t tokens;
12779 ++ int32_t tokens_min;
12780 ++ int32_t tokens_max;
12781 ++ int32_t prio_bias;
12782 ++};
12783 ++
12784 ++#define VXSM_FILL_RATE 0x0001
12785 ++#define VXSM_INTERVAL 0x0002
12786 ++#define VXSM_FILL_RATE2 0x0004
12787 ++#define VXSM_INTERVAL2 0x0008
12788 ++#define VXSM_TOKENS 0x0010
12789 ++#define VXSM_TOKENS_MIN 0x0020
12790 ++#define VXSM_TOKENS_MAX 0x0040
12791 ++#define VXSM_PRIO_BIAS 0x0100
12792 ++
12793 ++#define VXSM_IDLE_TIME 0x0200
12794 ++#define VXSM_FORCE 0x0400
12795 ++
12796 ++#define VXSM_V3_MASK 0x0173
12797 ++#define VXSM_SET_MASK 0x01FF
12798 ++
12799 ++#define VXSM_CPU_ID 0x1000
12800 ++#define VXSM_BUCKET_ID 0x2000
12801 ++
12802 ++#define VXSM_MSEC 0x4000
12803 ++
12804 ++#define SCHED_KEEP (-2) /* only for v2 */
12805 ++
12806 ++#ifdef __KERNEL__
12807 ++
12808 ++#include <linux/compiler.h>
12809 ++
12810 ++extern int vc_set_sched_v2(struct vx_info *, void __user *);
12811 ++extern int vc_set_sched_v3(struct vx_info *, void __user *);
12812 ++extern int vc_set_sched_v4(struct vx_info *, void __user *);
12813 ++extern int vc_set_sched(struct vx_info *, void __user *);
12814 ++extern int vc_get_sched(struct vx_info *, void __user *);
12815 ++
12816 ++#endif /* __KERNEL__ */
12817 ++
12818 ++#define VCMD_sched_info VC_CMD(SCHED, 3, 0)
12819 ++
12820 ++struct vcmd_sched_info {
12821 ++ int32_t cpu_id;
12822 ++ int32_t bucket_id;
12823 ++ uint64_t user_msec;
12824 ++ uint64_t sys_msec;
12825 ++ uint64_t hold_msec;
12826 ++ uint32_t token_usec;
12827 ++ int32_t vavavoom;
12828 ++};
12829 ++
12830 ++#ifdef __KERNEL__
12831 ++
12832 ++extern int vc_sched_info(struct vx_info *, void __user *);
12833 ++
12834 ++#endif /* __KERNEL__ */
12835 ++#endif /* _VX_SCHED_CMD_H */
12836 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/sched_def.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched_def.h
12837 +--- linux-2.6.29.4/include/linux/vserver/sched_def.h 1970-01-01 01:00:00.000000000 +0100
12838 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched_def.h 2009-02-22 22:54:26.000000000 +0100
12839 +@@ -0,0 +1,68 @@
12840 ++#ifndef _VX_SCHED_DEF_H
12841 ++#define _VX_SCHED_DEF_H
12842 ++
12843 ++#include <linux/spinlock.h>
12844 ++#include <linux/jiffies.h>
12845 ++#include <linux/cpumask.h>
12846 ++#include <asm/atomic.h>
12847 ++#include <asm/param.h>
12848 ++
12849 ++
12850 ++/* context sub struct */
12851 ++
12852 ++struct _vx_sched {
12853 ++ spinlock_t tokens_lock; /* lock for token bucket */
12854 ++
12855 ++ int tokens; /* number of CPU tokens */
12856 ++ int fill_rate[2]; /* Fill rate: add X tokens... */
12857 ++ int interval[2]; /* Divisor: per Y jiffies */
12858 ++ int tokens_min; /* Limit: minimum for unhold */
12859 ++ int tokens_max; /* Limit: no more than N tokens */
12860 ++
12861 ++ int prio_bias; /* bias offset for priority */
12862 ++
12863 ++ unsigned update_mask; /* which features should be updated */
12864 ++ cpumask_t update; /* CPUs which should update */
12865 ++};
12866 ++
12867 ++struct _vx_sched_pc {
12868 ++ int tokens; /* number of CPU tokens */
12869 ++ int flags; /* bucket flags */
12870 ++
12871 ++ int fill_rate[2]; /* Fill rate: add X tokens... */
12872 ++ int interval[2]; /* Divisor: per Y jiffies */
12873 ++ int tokens_min; /* Limit: minimum for unhold */
12874 ++ int tokens_max; /* Limit: no more than N tokens */
12875 ++
12876 ++ int prio_bias; /* bias offset for priority */
12877 ++ int vavavoom; /* last calculated vavavoom */
12878 ++
12879 ++ unsigned long norm_time; /* last time accounted */
12880 ++ unsigned long idle_time; /* non linear time for fair sched */
12881 ++ unsigned long token_time; /* token time for accounting */
12882 ++ unsigned long onhold; /* jiffies when put on hold */
12883 ++
12884 ++ uint64_t user_ticks; /* token tick events */
12885 ++ uint64_t sys_ticks; /* token tick events */
12886 ++ uint64_t hold_ticks; /* token ticks paused */
12887 ++};
12888 ++
12889 ++
12890 ++#define VXSF_ONHOLD 0x0001
12891 ++#define VXSF_IDLE_TIME 0x0100
12892 ++
12893 ++#ifdef CONFIG_VSERVER_DEBUG
12894 ++
12895 ++static inline void __dump_vx_sched(struct _vx_sched *sched)
12896 ++{
12897 ++ printk("\t_vx_sched:\n");
12898 ++ printk("\t tokens: %4d/%4d, %4d/%4d, %4d, %4d\n",
12899 ++ sched->fill_rate[0], sched->interval[0],
12900 ++ sched->fill_rate[1], sched->interval[1],
12901 ++ sched->tokens_min, sched->tokens_max);
12902 ++ printk("\t priority = %4d\n", sched->prio_bias);
12903 ++}
12904 ++
12905 ++#endif
12906 ++
12907 ++#endif /* _VX_SCHED_DEF_H */
12908 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/sched.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched.h
12909 +--- linux-2.6.29.4/include/linux/vserver/sched.h 1970-01-01 01:00:00.000000000 +0100
12910 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/sched.h 2009-02-22 22:54:26.000000000 +0100
12911 +@@ -0,0 +1,26 @@
12912 ++#ifndef _VX_SCHED_H
12913 ++#define _VX_SCHED_H
12914 ++
12915 ++
12916 ++#ifdef __KERNEL__
12917 ++
12918 ++struct timespec;
12919 ++
12920 ++void vx_vsi_uptime(struct timespec *, struct timespec *);
12921 ++
12922 ++
12923 ++struct vx_info;
12924 ++
12925 ++void vx_update_load(struct vx_info *);
12926 ++
12927 ++
12928 ++int vx_tokens_recalc(struct _vx_sched_pc *,
12929 ++ unsigned long *, unsigned long *, int [2]);
12930 ++
12931 ++void vx_update_sched_param(struct _vx_sched *sched,
12932 ++ struct _vx_sched_pc *sched_pc);
12933 ++
12934 ++#endif /* __KERNEL__ */
12935 ++#else /* _VX_SCHED_H */
12936 ++#warning duplicate inclusion
12937 ++#endif /* _VX_SCHED_H */
12938 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/signal_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/signal_cmd.h
12939 +--- linux-2.6.29.4/include/linux/vserver/signal_cmd.h 1970-01-01 01:00:00.000000000 +0100
12940 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/signal_cmd.h 2009-02-22 22:54:26.000000000 +0100
12941 +@@ -0,0 +1,43 @@
12942 ++#ifndef _VX_SIGNAL_CMD_H
12943 ++#define _VX_SIGNAL_CMD_H
12944 ++
12945 ++
12946 ++/* signalling vserver commands */
12947 ++
12948 ++#define VCMD_ctx_kill VC_CMD(PROCTRL, 1, 0)
12949 ++#define VCMD_wait_exit VC_CMD(EVENT, 99, 0)
12950 ++
12951 ++struct vcmd_ctx_kill_v0 {
12952 ++ int32_t pid;
12953 ++ int32_t sig;
12954 ++};
12955 ++
12956 ++struct vcmd_wait_exit_v0 {
12957 ++ int32_t reboot_cmd;
12958 ++ int32_t exit_code;
12959 ++};
12960 ++
12961 ++#ifdef __KERNEL__
12962 ++
12963 ++extern int vc_ctx_kill(struct vx_info *, void __user *);
12964 ++extern int vc_wait_exit(struct vx_info *, void __user *);
12965 ++
12966 ++#endif /* __KERNEL__ */
12967 ++
12968 ++/* process alteration commands */
12969 ++
12970 ++#define VCMD_get_pflags VC_CMD(PROCALT, 5, 0)
12971 ++#define VCMD_set_pflags VC_CMD(PROCALT, 6, 0)
12972 ++
12973 ++struct vcmd_pflags_v0 {
12974 ++ uint32_t flagword;
12975 ++ uint32_t mask;
12976 ++};
12977 ++
12978 ++#ifdef __KERNEL__
12979 ++
12980 ++extern int vc_get_pflags(uint32_t pid, void __user *);
12981 ++extern int vc_set_pflags(uint32_t pid, void __user *);
12982 ++
12983 ++#endif /* __KERNEL__ */
12984 ++#endif /* _VX_SIGNAL_CMD_H */
12985 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/signal.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/signal.h
12986 +--- linux-2.6.29.4/include/linux/vserver/signal.h 1970-01-01 01:00:00.000000000 +0100
12987 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/signal.h 2009-02-22 22:54:26.000000000 +0100
12988 +@@ -0,0 +1,14 @@
12989 ++#ifndef _VX_SIGNAL_H
12990 ++#define _VX_SIGNAL_H
12991 ++
12992 ++
12993 ++#ifdef __KERNEL__
12994 ++
12995 ++struct vx_info;
12996 ++
12997 ++int vx_info_kill(struct vx_info *, int, int);
12998 ++
12999 ++#endif /* __KERNEL__ */
13000 ++#else /* _VX_SIGNAL_H */
13001 ++#warning duplicate inclusion
13002 ++#endif /* _VX_SIGNAL_H */
13003 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/space_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/space_cmd.h
13004 +--- linux-2.6.29.4/include/linux/vserver/space_cmd.h 1970-01-01 01:00:00.000000000 +0100
13005 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/space_cmd.h 2009-02-22 22:54:26.000000000 +0100
13006 +@@ -0,0 +1,38 @@
13007 ++#ifndef _VX_SPACE_CMD_H
13008 ++#define _VX_SPACE_CMD_H
13009 ++
13010 ++
13011 ++#define VCMD_enter_space_v0 VC_CMD(PROCALT, 1, 0)
13012 ++#define VCMD_enter_space_v1 VC_CMD(PROCALT, 1, 1)
13013 ++#define VCMD_enter_space VC_CMD(PROCALT, 1, 2)
13014 ++
13015 ++#define VCMD_set_space_v0 VC_CMD(PROCALT, 3, 0)
13016 ++#define VCMD_set_space_v1 VC_CMD(PROCALT, 3, 1)
13017 ++#define VCMD_set_space VC_CMD(PROCALT, 3, 2)
13018 ++
13019 ++#define VCMD_get_space_mask_v0 VC_CMD(PROCALT, 4, 0)
13020 ++
13021 ++#define VCMD_get_space_mask VC_CMD(VSPACE, 0, 1)
13022 ++#define VCMD_get_space_default VC_CMD(VSPACE, 1, 0)
13023 ++
13024 ++
13025 ++struct vcmd_space_mask_v1 {
13026 ++ uint64_t mask;
13027 ++};
13028 ++
13029 ++struct vcmd_space_mask_v2 {
13030 ++ uint64_t mask;
13031 ++ uint32_t index;
13032 ++};
13033 ++
13034 ++
13035 ++#ifdef __KERNEL__
13036 ++
13037 ++extern int vc_enter_space_v1(struct vx_info *, void __user *);
13038 ++extern int vc_set_space_v1(struct vx_info *, void __user *);
13039 ++extern int vc_enter_space(struct vx_info *, void __user *);
13040 ++extern int vc_set_space(struct vx_info *, void __user *);
13041 ++extern int vc_get_space_mask(void __user *, int);
13042 ++
13043 ++#endif /* __KERNEL__ */
13044 ++#endif /* _VX_SPACE_CMD_H */
13045 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/space.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/space.h
13046 +--- linux-2.6.29.4/include/linux/vserver/space.h 1970-01-01 01:00:00.000000000 +0100
13047 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/space.h 2009-02-22 22:54:26.000000000 +0100
13048 +@@ -0,0 +1,12 @@
13049 ++#ifndef _VX_SPACE_H
13050 ++#define _VX_SPACE_H
13051 ++
13052 ++#include <linux/types.h>
13053 ++
13054 ++struct vx_info;
13055 ++
13056 ++int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index);
13057 ++
13058 ++#else /* _VX_SPACE_H */
13059 ++#warning duplicate inclusion
13060 ++#endif /* _VX_SPACE_H */
13061 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/switch.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/switch.h
13062 +--- linux-2.6.29.4/include/linux/vserver/switch.h 1970-01-01 01:00:00.000000000 +0100
13063 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/switch.h 2009-02-22 22:54:26.000000000 +0100
13064 +@@ -0,0 +1,98 @@
13065 ++#ifndef _VX_SWITCH_H
13066 ++#define _VX_SWITCH_H
13067 ++
13068 ++#include <linux/types.h>
13069 ++
13070 ++
13071 ++#define VC_CATEGORY(c) (((c) >> 24) & 0x3F)
13072 ++#define VC_COMMAND(c) (((c) >> 16) & 0xFF)
13073 ++#define VC_VERSION(c) ((c) & 0xFFF)
13074 ++
13075 ++#define VC_CMD(c, i, v) ((((VC_CAT_ ## c) & 0x3F) << 24) \
13076 ++ | (((i) & 0xFF) << 16) | ((v) & 0xFFF))
13077 ++
13078 ++/*
13079 ++
13080 ++ Syscall Matrix V2.8
13081 ++
13082 ++ |VERSION|CREATE |MODIFY |MIGRATE|CONTROL|EXPERIM| |SPECIAL|SPECIAL|
13083 ++ |STATS |DESTROY|ALTER |CHANGE |LIMIT |TEST | | | |
13084 ++ |INFO |SETUP | |MOVE | | | | | |
13085 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13086 ++ SYSTEM |VERSION|VSETUP |VHOST | | | | |DEVICE | |
13087 ++ HOST | 00| 01| 02| 03| 04| 05| | 06| 07|
13088 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13089 ++ CPU | |VPROC |PROCALT|PROCMIG|PROCTRL| | |SCHED. | |
13090 ++ PROCESS| 08| 09| 10| 11| 12| 13| | 14| 15|
13091 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13092 ++ MEMORY | | | | |MEMCTRL| | |SWAP | |
13093 ++ | 16| 17| 18| 19| 20| 21| | 22| 23|
13094 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13095 ++ NETWORK| |VNET |NETALT |NETMIG |NETCTL | | |SERIAL | |
13096 ++ | 24| 25| 26| 27| 28| 29| | 30| 31|
13097 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13098 ++ DISK | | | |TAGMIG |DLIMIT | | |INODE | |
13099 ++ VFS | 32| 33| 34| 35| 36| 37| | 38| 39|
13100 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13101 ++ OTHER |VSTAT | | | | | | |VINFO | |
13102 ++ | 40| 41| 42| 43| 44| 45| | 46| 47|
13103 ++ =======+=======+=======+=======+=======+=======+=======+ +=======+=======+
13104 ++ SPECIAL|EVENT | | | |FLAGS | | |VSPACE | |
13105 ++ | 48| 49| 50| 51| 52| 53| | 54| 55|
13106 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13107 ++ SPECIAL|DEBUG | | | |RLIMIT |SYSCALL| | |COMPAT |
13108 ++ | 56| 57| 58| 59| 60|TEST 61| | 62| 63|
13109 ++ -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
13110 ++
13111 ++*/
13112 ++
13113 ++#define VC_CAT_VERSION 0
13114 ++
13115 ++#define VC_CAT_VSETUP 1
13116 ++#define VC_CAT_VHOST 2
13117 ++
13118 ++#define VC_CAT_DEVICE 6
13119 ++
13120 ++#define VC_CAT_VPROC 9
13121 ++#define VC_CAT_PROCALT 10
13122 ++#define VC_CAT_PROCMIG 11
13123 ++#define VC_CAT_PROCTRL 12
13124 ++
13125 ++#define VC_CAT_SCHED 14
13126 ++#define VC_CAT_MEMCTRL 20
13127 ++
13128 ++#define VC_CAT_VNET 25
13129 ++#define VC_CAT_NETALT 26
13130 ++#define VC_CAT_NETMIG 27
13131 ++#define VC_CAT_NETCTRL 28
13132 ++
13133 ++#define VC_CAT_TAGMIG 35
13134 ++#define VC_CAT_DLIMIT 36
13135 ++#define VC_CAT_INODE 38
13136 ++
13137 ++#define VC_CAT_VSTAT 40
13138 ++#define VC_CAT_VINFO 46
13139 ++#define VC_CAT_EVENT 48
13140 ++
13141 ++#define VC_CAT_FLAGS 52
13142 ++#define VC_CAT_VSPACE 54
13143 ++#define VC_CAT_DEBUG 56
13144 ++#define VC_CAT_RLIMIT 60
13145 ++
13146 ++#define VC_CAT_SYSTEST 61
13147 ++#define VC_CAT_COMPAT 63
13148 ++
13149 ++/* query version */
13150 ++
13151 ++#define VCMD_get_version VC_CMD(VERSION, 0, 0)
13152 ++#define VCMD_get_vci VC_CMD(VERSION, 1, 0)
13153 ++
13154 ++
13155 ++#ifdef __KERNEL__
13156 ++
13157 ++#include <linux/errno.h>
13158 ++
13159 ++#endif /* __KERNEL__ */
13160 ++
13161 ++#endif /* _VX_SWITCH_H */
13162 ++
13163 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/tag_cmd.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/tag_cmd.h
13164 +--- linux-2.6.29.4/include/linux/vserver/tag_cmd.h 1970-01-01 01:00:00.000000000 +0100
13165 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/tag_cmd.h 2009-02-22 22:54:26.000000000 +0100
13166 +@@ -0,0 +1,22 @@
13167 ++#ifndef _VX_TAG_CMD_H
13168 ++#define _VX_TAG_CMD_H
13169 ++
13170 ++
13171 ++/* vinfo commands */
13172 ++
13173 ++#define VCMD_task_tag VC_CMD(VINFO, 3, 0)
13174 ++
13175 ++#ifdef __KERNEL__
13176 ++extern int vc_task_tag(uint32_t);
13177 ++
13178 ++#endif /* __KERNEL__ */
13179 ++
13180 ++/* context commands */
13181 ++
13182 ++#define VCMD_tag_migrate VC_CMD(TAGMIG, 1, 0)
13183 ++
13184 ++#ifdef __KERNEL__
13185 ++extern int vc_tag_migrate(uint32_t);
13186 ++
13187 ++#endif /* __KERNEL__ */
13188 ++#endif /* _VX_TAG_CMD_H */
13189 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vserver/tag.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/tag.h
13190 +--- linux-2.6.29.4/include/linux/vserver/tag.h 1970-01-01 01:00:00.000000000 +0100
13191 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vserver/tag.h 2009-02-22 22:54:26.000000000 +0100
13192 +@@ -0,0 +1,143 @@
13193 ++#ifndef _DX_TAG_H
13194 ++#define _DX_TAG_H
13195 ++
13196 ++#include <linux/types.h>
13197 ++
13198 ++
13199 ++#define DX_TAG(in) (IS_TAGGED(in))
13200 ++
13201 ++
13202 ++#ifdef CONFIG_TAG_NFSD
13203 ++#define DX_TAG_NFSD 1
13204 ++#else
13205 ++#define DX_TAG_NFSD 0
13206 ++#endif
13207 ++
13208 ++
13209 ++#ifdef CONFIG_TAGGING_NONE
13210 ++
13211 ++#define MAX_UID 0xFFFFFFFF
13212 ++#define MAX_GID 0xFFFFFFFF
13213 ++
13214 ++#define INOTAG_TAG(cond, uid, gid, tag) (0)
13215 ++
13216 ++#define TAGINO_UID(cond, uid, tag) (uid)
13217 ++#define TAGINO_GID(cond, gid, tag) (gid)
13218 ++
13219 ++#endif
13220 ++
13221 ++
13222 ++#ifdef CONFIG_TAGGING_GID16
13223 ++
13224 ++#define MAX_UID 0xFFFFFFFF
13225 ++#define MAX_GID 0x0000FFFF
13226 ++
13227 ++#define INOTAG_TAG(cond, uid, gid, tag) \
13228 ++ ((cond) ? (((gid) >> 16) & 0xFFFF) : 0)
13229 ++
13230 ++#define TAGINO_UID(cond, uid, tag) (uid)
13231 ++#define TAGINO_GID(cond, gid, tag) \
13232 ++ ((cond) ? (((gid) & 0xFFFF) | ((tag) << 16)) : (gid))
13233 ++
13234 ++#endif
13235 ++
13236 ++
13237 ++#ifdef CONFIG_TAGGING_ID24
13238 ++
13239 ++#define MAX_UID 0x00FFFFFF
13240 ++#define MAX_GID 0x00FFFFFF
13241 ++
13242 ++#define INOTAG_TAG(cond, uid, gid, tag) \
13243 ++ ((cond) ? ((((uid) >> 16) & 0xFF00) | (((gid) >> 24) & 0xFF)) : 0)
13244 ++
13245 ++#define TAGINO_UID(cond, uid, tag) \
13246 ++ ((cond) ? (((uid) & 0xFFFFFF) | (((tag) & 0xFF00) << 16)) : (uid))
13247 ++#define TAGINO_GID(cond, gid, tag) \
13248 ++ ((cond) ? (((gid) & 0xFFFFFF) | (((tag) & 0x00FF) << 24)) : (gid))
13249 ++
13250 ++#endif
13251 ++
13252 ++
13253 ++#ifdef CONFIG_TAGGING_UID16
13254 ++
13255 ++#define MAX_UID 0x0000FFFF
13256 ++#define MAX_GID 0xFFFFFFFF
13257 ++
13258 ++#define INOTAG_TAG(cond, uid, gid, tag) \
13259 ++ ((cond) ? (((uid) >> 16) & 0xFFFF) : 0)
13260 ++
13261 ++#define TAGINO_UID(cond, uid, tag) \
13262 ++ ((cond) ? (((uid) & 0xFFFF) | ((tag) << 16)) : (uid))
13263 ++#define TAGINO_GID(cond, gid, tag) (gid)
13264 ++
13265 ++#endif
13266 ++
13267 ++
13268 ++#ifdef CONFIG_TAGGING_INTERN
13269 ++
13270 ++#define MAX_UID 0xFFFFFFFF
13271 ++#define MAX_GID 0xFFFFFFFF
13272 ++
13273 ++#define INOTAG_TAG(cond, uid, gid, tag) \
13274 ++ ((cond) ? (tag) : 0)
13275 ++
13276 ++#define TAGINO_UID(cond, uid, tag) (uid)
13277 ++#define TAGINO_GID(cond, gid, tag) (gid)
13278 ++
13279 ++#endif
13280 ++
13281 ++
13282 ++#ifndef CONFIG_TAGGING_NONE
13283 ++#define dx_current_fstag(sb) \
13284 ++ ((sb)->s_flags & MS_TAGGED ? dx_current_tag() : 0)
13285 ++#else
13286 ++#define dx_current_fstag(sb) (0)
13287 ++#endif
13288 ++
13289 ++#ifndef CONFIG_TAGGING_INTERN
13290 ++#define TAGINO_TAG(cond, tag) (0)
13291 ++#else
13292 ++#define TAGINO_TAG(cond, tag) ((cond) ? (tag) : 0)
13293 ++#endif
13294 ++
13295 ++#define INOTAG_UID(cond, uid, gid) \
13296 ++ ((cond) ? ((uid) & MAX_UID) : (uid))
13297 ++#define INOTAG_GID(cond, uid, gid) \
13298 ++ ((cond) ? ((gid) & MAX_GID) : (gid))
13299 ++
13300 ++
13301 ++static inline uid_t dx_map_uid(uid_t uid)
13302 ++{
13303 ++ if ((uid > MAX_UID) && (uid != -1))
13304 ++ uid = -2;
13305 ++ return (uid & MAX_UID);
13306 ++}
13307 ++
13308 ++static inline gid_t dx_map_gid(gid_t gid)
13309 ++{
13310 ++ if ((gid > MAX_GID) && (gid != -1))
13311 ++ gid = -2;
13312 ++ return (gid & MAX_GID);
13313 ++}
13314 ++
13315 ++struct peer_tag {
13316 ++ int32_t xid;
13317 ++ int32_t nid;
13318 ++};
13319 ++
13320 ++#define dx_notagcheck(sb) ((sb) && ((sb)->s_flags & MS_NOTAGCHECK))
13321 ++
13322 ++int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
13323 ++ unsigned long *flags);
13324 ++
13325 ++#ifdef CONFIG_PROPAGATE
13326 ++
13327 ++void __dx_propagate_tag(struct nameidata *nd, struct inode *inode);
13328 ++
13329 ++#define dx_propagate_tag(n, i) __dx_propagate_tag(n, i)
13330 ++
13331 ++#else
13332 ++#define dx_propagate_tag(n, i) do { } while (0)
13333 ++#endif
13334 ++
13335 ++#endif /* _DX_TAG_H */
13336 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_inet6.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_inet6.h
13337 +--- linux-2.6.29.4/include/linux/vs_inet6.h 1970-01-01 01:00:00.000000000 +0100
13338 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_inet6.h 2009-03-25 00:46:50.000000000 +0100
13339 +@@ -0,0 +1,246 @@
13340 ++#ifndef _VS_INET6_H
13341 ++#define _VS_INET6_H
13342 ++
13343 ++#include "vserver/base.h"
13344 ++#include "vserver/network.h"
13345 ++#include "vserver/debug.h"
13346 ++
13347 ++#include <net/ipv6.h>
13348 ++
13349 ++#define NXAV6(a) &(a)->ip, &(a)->mask, (a)->prefix, (a)->type
13350 ++#define NXAV6_FMT "[%pI6/%pI6/%d:%04x]"
13351 ++
13352 ++
13353 ++#ifdef CONFIG_IPV6
13354 ++
13355 ++static inline
13356 ++int v6_addr_match(struct nx_addr_v6 *nxa,
13357 ++ const struct in6_addr *addr, uint16_t mask)
13358 ++{
13359 ++ int ret = 0;
13360 ++
13361 ++ switch (nxa->type & mask) {
13362 ++ case NXA_TYPE_MASK:
13363 ++ ret = ipv6_masked_addr_cmp(&nxa->ip, &nxa->mask, addr);
13364 ++ break;
13365 ++ case NXA_TYPE_ADDR:
13366 ++ ret = ipv6_addr_equal(&nxa->ip, addr);
13367 ++ break;
13368 ++ case NXA_TYPE_ANY:
13369 ++ ret = 1;
13370 ++ break;
13371 ++ }
13372 ++ vxdprintk(VXD_CBIT(net, 0),
13373 ++ "v6_addr_match(%p" NXAV6_FMT ",%pI6,%04x) = %d",
13374 ++ nxa, NXAV6(nxa), addr, mask, ret);
13375 ++ return ret;
13376 ++}
13377 ++
13378 ++static inline
13379 ++int v6_addr_in_nx_info(struct nx_info *nxi,
13380 ++ const struct in6_addr *addr, uint16_t mask)
13381 ++{
13382 ++ struct nx_addr_v6 *nxa;
13383 ++ int ret = 1;
13384 ++
13385 ++ if (!nxi)
13386 ++ goto out;
13387 ++ for (nxa = &nxi->v6; nxa; nxa = nxa->next)
13388 ++ if (v6_addr_match(nxa, addr, mask))
13389 ++ goto out;
13390 ++ ret = 0;
13391 ++out:
13392 ++ vxdprintk(VXD_CBIT(net, 0),
13393 ++ "v6_addr_in_nx_info(%p[#%u],%pI6,%04x) = %d",
13394 ++ nxi, nxi ? nxi->nx_id : 0, addr, mask, ret);
13395 ++ return ret;
13396 ++}
13397 ++
13398 ++static inline
13399 ++int v6_nx_addr_match(struct nx_addr_v6 *nxa, struct nx_addr_v6 *addr, uint16_t mask)
13400 ++{
13401 ++ /* FIXME: needs full range checks */
13402 ++ return v6_addr_match(nxa, &addr->ip, mask);
13403 ++}
13404 ++
13405 ++static inline
13406 ++int v6_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v6 *nxa, uint16_t mask)
13407 ++{
13408 ++ struct nx_addr_v6 *ptr;
13409 ++
13410 ++ for (ptr = &nxi->v6; ptr; ptr = ptr->next)
13411 ++ if (v6_nx_addr_match(ptr, nxa, mask))
13412 ++ return 1;
13413 ++ return 0;
13414 ++}
13415 ++
13416 ++
13417 ++/*
13418 ++ * Check if a given address matches for a socket
13419 ++ *
13420 ++ * nxi: the socket's nx_info if any
13421 ++ * addr: to be verified address
13422 ++ */
13423 ++static inline
13424 ++int v6_sock_addr_match (
13425 ++ struct nx_info *nxi,
13426 ++ struct inet_sock *inet,
13427 ++ struct in6_addr *addr)
13428 ++{
13429 ++ struct sock *sk = &inet->sk;
13430 ++ struct in6_addr *saddr = inet6_rcv_saddr(sk);
13431 ++
13432 ++ if (!ipv6_addr_any(addr) &&
13433 ++ ipv6_addr_equal(saddr, addr))
13434 ++ return 1;
13435 ++ if (ipv6_addr_any(saddr))
13436 ++ return v6_addr_in_nx_info(nxi, addr, -1);
13437 ++ return 0;
13438 ++}
13439 ++
13440 ++/*
13441 ++ * check if address is covered by socket
13442 ++ *
13443 ++ * sk: the socket to check against
13444 ++ * addr: the address in question (must be != 0)
13445 ++ */
13446 ++
13447 ++static inline
13448 ++int __v6_addr_match_socket(const struct sock *sk, struct nx_addr_v6 *nxa)
13449 ++{
13450 ++ struct nx_info *nxi = sk->sk_nx_info;
13451 ++ struct in6_addr *saddr = inet6_rcv_saddr(sk);
13452 ++
13453 ++ vxdprintk(VXD_CBIT(net, 5),
13454 ++ "__v6_addr_in_socket(%p," NXAV6_FMT ") %p:%pI6 %p;%lx",
13455 ++ sk, NXAV6(nxa), nxi, saddr, sk->sk_socket,
13456 ++ (sk->sk_socket?sk->sk_socket->flags:0));
13457 ++
13458 ++ if (!ipv6_addr_any(saddr)) { /* direct address match */
13459 ++ return v6_addr_match(nxa, saddr, -1);
13460 ++ } else if (nxi) { /* match against nx_info */
13461 ++ return v6_nx_addr_in_nx_info(nxi, nxa, -1);
13462 ++ } else { /* unrestricted any socket */
13463 ++ return 1;
13464 ++ }
13465 ++}
13466 ++
13467 ++
13468 ++/* inet related checks and helpers */
13469 ++
13470 ++
13471 ++struct in_ifaddr;
13472 ++struct net_device;
13473 ++struct sock;
13474 ++
13475 ++
13476 ++#include <linux/netdevice.h>
13477 ++#include <linux/inetdevice.h>
13478 ++#include <net/inet_timewait_sock.h>
13479 ++
13480 ++
13481 ++int dev_in_nx_info(struct net_device *, struct nx_info *);
13482 ++int v6_dev_in_nx_info(struct net_device *, struct nx_info *);
13483 ++int nx_v6_addr_conflict(struct nx_info *, struct nx_info *);
13484 ++
13485 ++
13486 ++
13487 ++static inline
13488 ++int v6_ifa_in_nx_info(struct inet6_ifaddr *ifa, struct nx_info *nxi)
13489 ++{
13490 ++ if (!nxi)
13491 ++ return 1;
13492 ++ if (!ifa)
13493 ++ return 0;
13494 ++ return v6_addr_in_nx_info(nxi, &ifa->addr, -1);
13495 ++}
13496 ++
13497 ++static inline
13498 ++int nx_v6_ifa_visible(struct nx_info *nxi, struct inet6_ifaddr *ifa)
13499 ++{
13500 ++ vxdprintk(VXD_CBIT(net, 1), "nx_v6_ifa_visible(%p[#%u],%p) %d",
13501 ++ nxi, nxi ? nxi->nx_id : 0, ifa,
13502 ++ nxi ? v6_ifa_in_nx_info(ifa, nxi) : 0);
13503 ++
13504 ++ if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13505 ++ return 1;
13506 ++ if (v6_ifa_in_nx_info(ifa, nxi))
13507 ++ return 1;
13508 ++ return 0;
13509 ++}
13510 ++
13511 ++
13512 ++struct nx_v6_sock_addr {
13513 ++ struct in6_addr saddr; /* Address used for validation */
13514 ++ struct in6_addr baddr; /* Address used for socket bind */
13515 ++};
13516 ++
13517 ++static inline
13518 ++int v6_map_sock_addr(struct inet_sock *inet, struct sockaddr_in6 *addr,
13519 ++ struct nx_v6_sock_addr *nsa)
13520 ++{
13521 ++ // struct sock *sk = &inet->sk;
13522 ++ // struct nx_info *nxi = sk->sk_nx_info;
13523 ++ struct in6_addr saddr = addr->sin6_addr;
13524 ++ struct in6_addr baddr = saddr;
13525 ++
13526 ++ nsa->saddr = saddr;
13527 ++ nsa->baddr = baddr;
13528 ++ return 0;
13529 ++}
13530 ++
13531 ++static inline
13532 ++void v6_set_sock_addr(struct inet_sock *inet, struct nx_v6_sock_addr *nsa)
13533 ++{
13534 ++ // struct sock *sk = &inet->sk;
13535 ++ // struct in6_addr *saddr = inet6_rcv_saddr(sk);
13536 ++
13537 ++ // *saddr = nsa->baddr;
13538 ++ // inet->saddr = nsa->baddr;
13539 ++}
13540 ++
13541 ++static inline
13542 ++int nx_info_has_v6(struct nx_info *nxi)
13543 ++{
13544 ++ if (!nxi)
13545 ++ return 1;
13546 ++ if (NX_IPV6(nxi))
13547 ++ return 1;
13548 ++ return 0;
13549 ++}
13550 ++
13551 ++#else /* CONFIG_IPV6 */
13552 ++
13553 ++static inline
13554 ++int nx_v6_dev_visible(struct nx_info *n, struct net_device *d)
13555 ++{
13556 ++ return 1;
13557 ++}
13558 ++
13559 ++
13560 ++static inline
13561 ++int nx_v6_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
13562 ++{
13563 ++ return 1;
13564 ++}
13565 ++
13566 ++static inline
13567 ++int v6_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
13568 ++{
13569 ++ return 1;
13570 ++}
13571 ++
13572 ++static inline
13573 ++int nx_info_has_v6(struct nx_info *nxi)
13574 ++{
13575 ++ return 0;
13576 ++}
13577 ++
13578 ++#endif /* CONFIG_IPV6 */
13579 ++
13580 ++#define current_nx_info_has_v6() \
13581 ++ nx_info_has_v6(current_nx_info())
13582 ++
13583 ++#else
13584 ++#warning duplicate inclusion
13585 ++#endif
13586 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_inet.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_inet.h
13587 +--- linux-2.6.29.4/include/linux/vs_inet.h 1970-01-01 01:00:00.000000000 +0100
13588 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_inet.h 2009-02-22 22:54:26.000000000 +0100
13589 +@@ -0,0 +1,342 @@
13590 ++#ifndef _VS_INET_H
13591 ++#define _VS_INET_H
13592 ++
13593 ++#include "vserver/base.h"
13594 ++#include "vserver/network.h"
13595 ++#include "vserver/debug.h"
13596 ++
13597 ++#define IPI_LOOPBACK htonl(INADDR_LOOPBACK)
13598 ++
13599 ++#define NXAV4(a) NIPQUAD((a)->ip[0]), NIPQUAD((a)->ip[1]), \
13600 ++ NIPQUAD((a)->mask), (a)->type
13601 ++#define NXAV4_FMT "[" NIPQUAD_FMT "-" NIPQUAD_FMT "/" NIPQUAD_FMT ":%04x]"
13602 ++
13603 ++
13604 ++static inline
13605 ++int v4_addr_match(struct nx_addr_v4 *nxa, __be32 addr, uint16_t tmask)
13606 ++{
13607 ++ __be32 ip = nxa->ip[0].s_addr;
13608 ++ __be32 mask = nxa->mask.s_addr;
13609 ++ __be32 bcast = ip | ~mask;
13610 ++ int ret = 0;
13611 ++
13612 ++ switch (nxa->type & tmask) {
13613 ++ case NXA_TYPE_MASK:
13614 ++ ret = (ip == (addr & mask));
13615 ++ break;
13616 ++ case NXA_TYPE_ADDR:
13617 ++ ret = 3;
13618 ++ if (addr == ip)
13619 ++ break;
13620 ++ /* fall through to broadcast */
13621 ++ case NXA_MOD_BCAST:
13622 ++ ret = ((tmask & NXA_MOD_BCAST) && (addr == bcast));
13623 ++ break;
13624 ++ case NXA_TYPE_RANGE:
13625 ++ ret = ((nxa->ip[0].s_addr <= addr) &&
13626 ++ (nxa->ip[1].s_addr > addr));
13627 ++ break;
13628 ++ case NXA_TYPE_ANY:
13629 ++ ret = 2;
13630 ++ break;
13631 ++ }
13632 ++
13633 ++ vxdprintk(VXD_CBIT(net, 0),
13634 ++ "v4_addr_match(%p" NXAV4_FMT "," NIPQUAD_FMT ",%04x) = %d",
13635 ++ nxa, NXAV4(nxa), NIPQUAD(addr), tmask, ret);
13636 ++ return ret;
13637 ++}
13638 ++
13639 ++static inline
13640 ++int v4_addr_in_nx_info(struct nx_info *nxi, __be32 addr, uint16_t tmask)
13641 ++{
13642 ++ struct nx_addr_v4 *nxa;
13643 ++ int ret = 1;
13644 ++
13645 ++ if (!nxi)
13646 ++ goto out;
13647 ++
13648 ++ ret = 2;
13649 ++ /* allow 127.0.0.1 when remapping lback */
13650 ++ if ((tmask & NXA_LOOPBACK) &&
13651 ++ (addr == IPI_LOOPBACK) &&
13652 ++ nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13653 ++ goto out;
13654 ++ ret = 3;
13655 ++ /* check for lback address */
13656 ++ if ((tmask & NXA_MOD_LBACK) &&
13657 ++ (nxi->v4_lback.s_addr == addr))
13658 ++ goto out;
13659 ++ ret = 4;
13660 ++ /* check for broadcast address */
13661 ++ if ((tmask & NXA_MOD_BCAST) &&
13662 ++ (nxi->v4_bcast.s_addr == addr))
13663 ++ goto out;
13664 ++ ret = 5;
13665 ++ /* check for v4 addresses */
13666 ++ for (nxa = &nxi->v4; nxa; nxa = nxa->next)
13667 ++ if (v4_addr_match(nxa, addr, tmask))
13668 ++ goto out;
13669 ++ ret = 0;
13670 ++out:
13671 ++ vxdprintk(VXD_CBIT(net, 0),
13672 ++ "v4_addr_in_nx_info(%p[#%u]," NIPQUAD_FMT ",%04x) = %d",
13673 ++ nxi, nxi ? nxi->nx_id : 0, NIPQUAD(addr), tmask, ret);
13674 ++ return ret;
13675 ++}
13676 ++
13677 ++static inline
13678 ++int v4_nx_addr_match(struct nx_addr_v4 *nxa, struct nx_addr_v4 *addr, uint16_t mask)
13679 ++{
13680 ++ /* FIXME: needs full range checks */
13681 ++ return v4_addr_match(nxa, addr->ip[0].s_addr, mask);
13682 ++}
13683 ++
13684 ++static inline
13685 ++int v4_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v4 *nxa, uint16_t mask)
13686 ++{
13687 ++ struct nx_addr_v4 *ptr;
13688 ++
13689 ++ for (ptr = &nxi->v4; ptr; ptr = ptr->next)
13690 ++ if (v4_nx_addr_match(ptr, nxa, mask))
13691 ++ return 1;
13692 ++ return 0;
13693 ++}
13694 ++
13695 ++#include <net/inet_sock.h>
13696 ++
13697 ++/*
13698 ++ * Check if a given address matches for a socket
13699 ++ *
13700 ++ * nxi: the socket's nx_info if any
13701 ++ * addr: to be verified address
13702 ++ */
13703 ++static inline
13704 ++int v4_sock_addr_match (
13705 ++ struct nx_info *nxi,
13706 ++ struct inet_sock *inet,
13707 ++ __be32 addr)
13708 ++{
13709 ++ __be32 saddr = inet->rcv_saddr;
13710 ++ __be32 bcast = nxi ? nxi->v4_bcast.s_addr : INADDR_BROADCAST;
13711 ++
13712 ++ if (addr && (saddr == addr || bcast == addr))
13713 ++ return 1;
13714 ++ if (!saddr)
13715 ++ return v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND);
13716 ++ return 0;
13717 ++}
13718 ++
13719 ++
13720 ++/* inet related checks and helpers */
13721 ++
13722 ++
13723 ++struct in_ifaddr;
13724 ++struct net_device;
13725 ++struct sock;
13726 ++
13727 ++#ifdef CONFIG_INET
13728 ++
13729 ++#include <linux/netdevice.h>
13730 ++#include <linux/inetdevice.h>
13731 ++#include <net/inet_sock.h>
13732 ++#include <net/inet_timewait_sock.h>
13733 ++
13734 ++
13735 ++int dev_in_nx_info(struct net_device *, struct nx_info *);
13736 ++int v4_dev_in_nx_info(struct net_device *, struct nx_info *);
13737 ++int nx_v4_addr_conflict(struct nx_info *, struct nx_info *);
13738 ++
13739 ++
13740 ++/*
13741 ++ * check if address is covered by socket
13742 ++ *
13743 ++ * sk: the socket to check against
13744 ++ * addr: the address in question (must be != 0)
13745 ++ */
13746 ++
13747 ++static inline
13748 ++int __v4_addr_match_socket(const struct sock *sk, struct nx_addr_v4 *nxa)
13749 ++{
13750 ++ struct nx_info *nxi = sk->sk_nx_info;
13751 ++ __be32 saddr = inet_rcv_saddr(sk);
13752 ++
13753 ++ vxdprintk(VXD_CBIT(net, 5),
13754 ++ "__v4_addr_in_socket(%p," NXAV4_FMT ") %p:" NIPQUAD_FMT " %p;%lx",
13755 ++ sk, NXAV4(nxa), nxi, NIPQUAD(saddr), sk->sk_socket,
13756 ++ (sk->sk_socket?sk->sk_socket->flags:0));
13757 ++
13758 ++ if (saddr) { /* direct address match */
13759 ++ return v4_addr_match(nxa, saddr, -1);
13760 ++ } else if (nxi) { /* match against nx_info */
13761 ++ return v4_nx_addr_in_nx_info(nxi, nxa, -1);
13762 ++ } else { /* unrestricted any socket */
13763 ++ return 1;
13764 ++ }
13765 ++}
13766 ++
13767 ++
13768 ++
13769 ++static inline
13770 ++int nx_dev_visible(struct nx_info *nxi, struct net_device *dev)
13771 ++{
13772 ++ vxdprintk(VXD_CBIT(net, 1), "nx_dev_visible(%p[#%u],%p »%s«) %d",
13773 ++ nxi, nxi ? nxi->nx_id : 0, dev, dev->name,
13774 ++ nxi ? dev_in_nx_info(dev, nxi) : 0);
13775 ++
13776 ++ if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13777 ++ return 1;
13778 ++ if (dev_in_nx_info(dev, nxi))
13779 ++ return 1;
13780 ++ return 0;
13781 ++}
13782 ++
13783 ++
13784 ++static inline
13785 ++int v4_ifa_in_nx_info(struct in_ifaddr *ifa, struct nx_info *nxi)
13786 ++{
13787 ++ if (!nxi)
13788 ++ return 1;
13789 ++ if (!ifa)
13790 ++ return 0;
13791 ++ return v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW);
13792 ++}
13793 ++
13794 ++static inline
13795 ++int nx_v4_ifa_visible(struct nx_info *nxi, struct in_ifaddr *ifa)
13796 ++{
13797 ++ vxdprintk(VXD_CBIT(net, 1), "nx_v4_ifa_visible(%p[#%u],%p) %d",
13798 ++ nxi, nxi ? nxi->nx_id : 0, ifa,
13799 ++ nxi ? v4_ifa_in_nx_info(ifa, nxi) : 0);
13800 ++
13801 ++ if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
13802 ++ return 1;
13803 ++ if (v4_ifa_in_nx_info(ifa, nxi))
13804 ++ return 1;
13805 ++ return 0;
13806 ++}
13807 ++
13808 ++
13809 ++struct nx_v4_sock_addr {
13810 ++ __be32 saddr; /* Address used for validation */
13811 ++ __be32 baddr; /* Address used for socket bind */
13812 ++};
13813 ++
13814 ++static inline
13815 ++int v4_map_sock_addr(struct inet_sock *inet, struct sockaddr_in *addr,
13816 ++ struct nx_v4_sock_addr *nsa)
13817 ++{
13818 ++ struct sock *sk = &inet->sk;
13819 ++ struct nx_info *nxi = sk->sk_nx_info;
13820 ++ __be32 saddr = addr->sin_addr.s_addr;
13821 ++ __be32 baddr = saddr;
13822 ++
13823 ++ vxdprintk(VXD_CBIT(net, 3),
13824 ++ "inet_bind(%p)* %p,%p;%lx " NIPQUAD_FMT,
13825 ++ sk, sk->sk_nx_info, sk->sk_socket,
13826 ++ (sk->sk_socket ? sk->sk_socket->flags : 0),
13827 ++ NIPQUAD(saddr));
13828 ++
13829 ++ if (nxi) {
13830 ++ if (saddr == INADDR_ANY) {
13831 ++ if (nx_info_flags(nxi, NXF_SINGLE_IP, 0))
13832 ++ baddr = nxi->v4.ip[0].s_addr;
13833 ++ } else if (saddr == IPI_LOOPBACK) {
13834 ++ if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13835 ++ baddr = nxi->v4_lback.s_addr;
13836 ++ } else { /* normal address bind */
13837 ++ if (!v4_addr_in_nx_info(nxi, saddr, NXA_MASK_BIND))
13838 ++ return -EADDRNOTAVAIL;
13839 ++ }
13840 ++ }
13841 ++
13842 ++ vxdprintk(VXD_CBIT(net, 3),
13843 ++ "inet_bind(%p) " NIPQUAD_FMT ", " NIPQUAD_FMT,
13844 ++ sk, NIPQUAD(saddr), NIPQUAD(baddr));
13845 ++
13846 ++ nsa->saddr = saddr;
13847 ++ nsa->baddr = baddr;
13848 ++ return 0;
13849 ++}
13850 ++
13851 ++static inline
13852 ++void v4_set_sock_addr(struct inet_sock *inet, struct nx_v4_sock_addr *nsa)
13853 ++{
13854 ++ inet->saddr = nsa->baddr;
13855 ++ inet->rcv_saddr = nsa->baddr;
13856 ++}
13857 ++
13858 ++
13859 ++/*
13860 ++ * helper to simplify inet_lookup_listener
13861 ++ *
13862 ++ * nxi: the socket's nx_info if any
13863 ++ * addr: to be verified address
13864 ++ * saddr: socket address
13865 ++ */
13866 ++static inline int v4_inet_addr_match (
13867 ++ struct nx_info *nxi,
13868 ++ __be32 addr,
13869 ++ __be32 saddr)
13870 ++{
13871 ++ if (addr && (saddr == addr))
13872 ++ return 1;
13873 ++ if (!saddr)
13874 ++ return nxi ? v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND) : 1;
13875 ++ return 0;
13876 ++}
13877 ++
13878 ++static inline __be32 nx_map_sock_lback(struct nx_info *nxi, __be32 addr)
13879 ++{
13880 ++ if (nx_info_flags(nxi, NXF_HIDE_LBACK, 0) &&
13881 ++ (addr == nxi->v4_lback.s_addr))
13882 ++ return IPI_LOOPBACK;
13883 ++ return addr;
13884 ++}
13885 ++
13886 ++static inline
13887 ++int nx_info_has_v4(struct nx_info *nxi)
13888 ++{
13889 ++ if (!nxi)
13890 ++ return 1;
13891 ++ if (NX_IPV4(nxi))
13892 ++ return 1;
13893 ++ if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
13894 ++ return 1;
13895 ++ return 0;
13896 ++}
13897 ++
13898 ++#else /* CONFIG_INET */
13899 ++
13900 ++static inline
13901 ++int nx_dev_visible(struct nx_info *n, struct net_device *d)
13902 ++{
13903 ++ return 1;
13904 ++}
13905 ++
13906 ++static inline
13907 ++int nx_v4_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
13908 ++{
13909 ++ return 1;
13910 ++}
13911 ++
13912 ++static inline
13913 ++int v4_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
13914 ++{
13915 ++ return 1;
13916 ++}
13917 ++
13918 ++static inline
13919 ++int nx_info_has_v4(struct nx_info *nxi)
13920 ++{
13921 ++ return 0;
13922 ++}
13923 ++
13924 ++#endif /* CONFIG_INET */
13925 ++
13926 ++#define current_nx_info_has_v4() \
13927 ++ nx_info_has_v4(current_nx_info())
13928 ++
13929 ++#else
13930 ++// #warning duplicate inclusion
13931 ++#endif
13932 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_limit.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_limit.h
13933 +--- linux-2.6.29.4/include/linux/vs_limit.h 1970-01-01 01:00:00.000000000 +0100
13934 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_limit.h 2009-02-22 22:54:26.000000000 +0100
13935 +@@ -0,0 +1,140 @@
13936 ++#ifndef _VS_LIMIT_H
13937 ++#define _VS_LIMIT_H
13938 ++
13939 ++#include "vserver/limit.h"
13940 ++#include "vserver/base.h"
13941 ++#include "vserver/context.h"
13942 ++#include "vserver/debug.h"
13943 ++#include "vserver/context.h"
13944 ++#include "vserver/limit_int.h"
13945 ++
13946 ++
13947 ++#define vx_acc_cres(v, d, p, r) \
13948 ++ __vx_acc_cres(v, r, d, p, __FILE__, __LINE__)
13949 ++
13950 ++#define vx_acc_cres_cond(x, d, p, r) \
13951 ++ __vx_acc_cres(((x) == vx_current_xid()) ? current->vx_info : 0, \
13952 ++ r, d, p, __FILE__, __LINE__)
13953 ++
13954 ++
13955 ++#define vx_add_cres(v, a, p, r) \
13956 ++ __vx_add_cres(v, r, a, p, __FILE__, __LINE__)
13957 ++#define vx_sub_cres(v, a, p, r) vx_add_cres(v, -(a), p, r)
13958 ++
13959 ++#define vx_add_cres_cond(x, a, p, r) \
13960 ++ __vx_add_cres(((x) == vx_current_xid()) ? current->vx_info : 0, \
13961 ++ r, a, p, __FILE__, __LINE__)
13962 ++#define vx_sub_cres_cond(x, a, p, r) vx_add_cres_cond(x, -(a), p, r)
13963 ++
13964 ++
13965 ++/* process and file limits */
13966 ++
13967 ++#define vx_nproc_inc(p) \
13968 ++ vx_acc_cres((p)->vx_info, 1, p, RLIMIT_NPROC)
13969 ++
13970 ++#define vx_nproc_dec(p) \
13971 ++ vx_acc_cres((p)->vx_info,-1, p, RLIMIT_NPROC)
13972 ++
13973 ++#define vx_files_inc(f) \
13974 ++ vx_acc_cres_cond((f)->f_xid, 1, f, RLIMIT_NOFILE)
13975 ++
13976 ++#define vx_files_dec(f) \
13977 ++ vx_acc_cres_cond((f)->f_xid,-1, f, RLIMIT_NOFILE)
13978 ++
13979 ++#define vx_locks_inc(l) \
13980 ++ vx_acc_cres_cond((l)->fl_xid, 1, l, RLIMIT_LOCKS)
13981 ++
13982 ++#define vx_locks_dec(l) \
13983 ++ vx_acc_cres_cond((l)->fl_xid,-1, l, RLIMIT_LOCKS)
13984 ++
13985 ++#define vx_openfd_inc(f) \
13986 ++ vx_acc_cres(current->vx_info, 1, (void *)(long)(f), VLIMIT_OPENFD)
13987 ++
13988 ++#define vx_openfd_dec(f) \
13989 ++ vx_acc_cres(current->vx_info,-1, (void *)(long)(f), VLIMIT_OPENFD)
13990 ++
13991 ++
13992 ++#define vx_cres_avail(v, n, r) \
13993 ++ __vx_cres_avail(v, r, n, __FILE__, __LINE__)
13994 ++
13995 ++
13996 ++#define vx_nproc_avail(n) \
13997 ++ vx_cres_avail(current->vx_info, n, RLIMIT_NPROC)
13998 ++
13999 ++#define vx_files_avail(n) \
14000 ++ vx_cres_avail(current->vx_info, n, RLIMIT_NOFILE)
14001 ++
14002 ++#define vx_locks_avail(n) \
14003 ++ vx_cres_avail(current->vx_info, n, RLIMIT_LOCKS)
14004 ++
14005 ++#define vx_openfd_avail(n) \
14006 ++ vx_cres_avail(current->vx_info, n, VLIMIT_OPENFD)
14007 ++
14008 ++
14009 ++/* dentry limits */
14010 ++
14011 ++#define vx_dentry_inc(d) do { \
14012 ++ if (atomic_read(&d->d_count) == 1) \
14013 ++ vx_acc_cres(current->vx_info, 1, d, VLIMIT_DENTRY); \
14014 ++ } while (0)
14015 ++
14016 ++#define vx_dentry_dec(d) do { \
14017 ++ if (atomic_read(&d->d_count) == 0) \
14018 ++ vx_acc_cres(current->vx_info,-1, d, VLIMIT_DENTRY); \
14019 ++ } while (0)
14020 ++
14021 ++#define vx_dentry_avail(n) \
14022 ++ vx_cres_avail(current->vx_info, n, VLIMIT_DENTRY)
14023 ++
14024 ++
14025 ++/* socket limits */
14026 ++
14027 ++#define vx_sock_inc(s) \
14028 ++ vx_acc_cres((s)->sk_vx_info, 1, s, VLIMIT_NSOCK)
14029 ++
14030 ++#define vx_sock_dec(s) \
14031 ++ vx_acc_cres((s)->sk_vx_info,-1, s, VLIMIT_NSOCK)
14032 ++
14033 ++#define vx_sock_avail(n) \
14034 ++ vx_cres_avail(current->vx_info, n, VLIMIT_NSOCK)
14035 ++
14036 ++
14037 ++/* ipc resource limits */
14038 ++
14039 ++#define vx_ipcmsg_add(v, u, a) \
14040 ++ vx_add_cres(v, a, u, RLIMIT_MSGQUEUE)
14041 ++
14042 ++#define vx_ipcmsg_sub(v, u, a) \
14043 ++ vx_sub_cres(v, a, u, RLIMIT_MSGQUEUE)
14044 ++
14045 ++#define vx_ipcmsg_avail(v, a) \
14046 ++ vx_cres_avail(v, a, RLIMIT_MSGQUEUE)
14047 ++
14048 ++
14049 ++#define vx_ipcshm_add(v, k, a) \
14050 ++ vx_add_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
14051 ++
14052 ++#define vx_ipcshm_sub(v, k, a) \
14053 ++ vx_sub_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
14054 ++
14055 ++#define vx_ipcshm_avail(v, a) \
14056 ++ vx_cres_avail(v, a, VLIMIT_SHMEM)
14057 ++
14058 ++
14059 ++#define vx_semary_inc(a) \
14060 ++ vx_acc_cres(current->vx_info, 1, a, VLIMIT_SEMARY)
14061 ++
14062 ++#define vx_semary_dec(a) \
14063 ++ vx_acc_cres(current->vx_info, -1, a, VLIMIT_SEMARY)
14064 ++
14065 ++
14066 ++#define vx_nsems_add(a,n) \
14067 ++ vx_add_cres(current->vx_info, n, a, VLIMIT_NSEMS)
14068 ++
14069 ++#define vx_nsems_sub(a,n) \
14070 ++ vx_sub_cres(current->vx_info, n, a, VLIMIT_NSEMS)
14071 ++
14072 ++
14073 ++#else
14074 ++#warning duplicate inclusion
14075 ++#endif
14076 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_memory.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_memory.h
14077 +--- linux-2.6.29.4/include/linux/vs_memory.h 1970-01-01 01:00:00.000000000 +0100
14078 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_memory.h 2009-02-22 22:54:26.000000000 +0100
14079 +@@ -0,0 +1,159 @@
14080 ++#ifndef _VS_MEMORY_H
14081 ++#define _VS_MEMORY_H
14082 ++
14083 ++#include "vserver/limit.h"
14084 ++#include "vserver/base.h"
14085 ++#include "vserver/context.h"
14086 ++#include "vserver/debug.h"
14087 ++#include "vserver/context.h"
14088 ++#include "vserver/limit_int.h"
14089 ++
14090 ++
14091 ++#define __acc_add_long(a, v) (*(v) += (a))
14092 ++#define __acc_inc_long(v) (++*(v))
14093 ++#define __acc_dec_long(v) (--*(v))
14094 ++
14095 ++#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
14096 ++#define __acc_add_atomic(a, v) atomic_long_add(a, v)
14097 ++#define __acc_inc_atomic(v) atomic_long_inc(v)
14098 ++#define __acc_dec_atomic(v) atomic_long_dec(v)
14099 ++#else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
14100 ++#define __acc_add_atomic(a, v) __acc_add_long(a, v)
14101 ++#define __acc_inc_atomic(v) __acc_inc_long(v)
14102 ++#define __acc_dec_atomic(v) __acc_dec_long(v)
14103 ++#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
14104 ++
14105 ++
14106 ++#define vx_acc_page(m, d, v, r) do { \
14107 ++ if ((d) > 0) \
14108 ++ __acc_inc_long(&(m)->v); \
14109 ++ else \
14110 ++ __acc_dec_long(&(m)->v); \
14111 ++ __vx_acc_cres(m->mm_vx_info, r, d, m, __FILE__, __LINE__); \
14112 ++} while (0)
14113 ++
14114 ++#define vx_acc_page_atomic(m, d, v, r) do { \
14115 ++ if ((d) > 0) \
14116 ++ __acc_inc_atomic(&(m)->v); \
14117 ++ else \
14118 ++ __acc_dec_atomic(&(m)->v); \
14119 ++ __vx_acc_cres(m->mm_vx_info, r, d, m, __FILE__, __LINE__); \
14120 ++} while (0)
14121 ++
14122 ++
14123 ++#define vx_acc_pages(m, p, v, r) do { \
14124 ++ unsigned long __p = (p); \
14125 ++ __acc_add_long(__p, &(m)->v); \
14126 ++ __vx_add_cres(m->mm_vx_info, r, __p, m, __FILE__, __LINE__); \
14127 ++} while (0)
14128 ++
14129 ++#define vx_acc_pages_atomic(m, p, v, r) do { \
14130 ++ unsigned long __p = (p); \
14131 ++ __acc_add_atomic(__p, &(m)->v); \
14132 ++ __vx_add_cres(m->mm_vx_info, r, __p, m, __FILE__, __LINE__); \
14133 ++} while (0)
14134 ++
14135 ++
14136 ++
14137 ++#define vx_acc_vmpage(m, d) \
14138 ++ vx_acc_page(m, d, total_vm, RLIMIT_AS)
14139 ++#define vx_acc_vmlpage(m, d) \
14140 ++ vx_acc_page(m, d, locked_vm, RLIMIT_MEMLOCK)
14141 ++#define vx_acc_file_rsspage(m, d) \
14142 ++ vx_acc_page_atomic(m, d, _file_rss, VLIMIT_MAPPED)
14143 ++#define vx_acc_anon_rsspage(m, d) \
14144 ++ vx_acc_page_atomic(m, d, _anon_rss, VLIMIT_ANON)
14145 ++
14146 ++#define vx_acc_vmpages(m, p) \
14147 ++ vx_acc_pages(m, p, total_vm, RLIMIT_AS)
14148 ++#define vx_acc_vmlpages(m, p) \
14149 ++ vx_acc_pages(m, p, locked_vm, RLIMIT_MEMLOCK)
14150 ++#define vx_acc_file_rsspages(m, p) \
14151 ++ vx_acc_pages_atomic(m, p, _file_rss, VLIMIT_MAPPED)
14152 ++#define vx_acc_anon_rsspages(m, p) \
14153 ++ vx_acc_pages_atomic(m, p, _anon_rss, VLIMIT_ANON)
14154 ++
14155 ++#define vx_pages_add(s, r, p) __vx_add_cres(s, r, p, 0, __FILE__, __LINE__)
14156 ++#define vx_pages_sub(s, r, p) vx_pages_add(s, r, -(p))
14157 ++
14158 ++#define vx_vmpages_inc(m) vx_acc_vmpage(m, 1)
14159 ++#define vx_vmpages_dec(m) vx_acc_vmpage(m, -1)
14160 ++#define vx_vmpages_add(m, p) vx_acc_vmpages(m, p)
14161 ++#define vx_vmpages_sub(m, p) vx_acc_vmpages(m, -(p))
14162 ++
14163 ++#define vx_vmlocked_inc(m) vx_acc_vmlpage(m, 1)
14164 ++#define vx_vmlocked_dec(m) vx_acc_vmlpage(m, -1)
14165 ++#define vx_vmlocked_add(m, p) vx_acc_vmlpages(m, p)
14166 ++#define vx_vmlocked_sub(m, p) vx_acc_vmlpages(m, -(p))
14167 ++
14168 ++#define vx_file_rsspages_inc(m) vx_acc_file_rsspage(m, 1)
14169 ++#define vx_file_rsspages_dec(m) vx_acc_file_rsspage(m, -1)
14170 ++#define vx_file_rsspages_add(m, p) vx_acc_file_rsspages(m, p)
14171 ++#define vx_file_rsspages_sub(m, p) vx_acc_file_rsspages(m, -(p))
14172 ++
14173 ++#define vx_anon_rsspages_inc(m) vx_acc_anon_rsspage(m, 1)
14174 ++#define vx_anon_rsspages_dec(m) vx_acc_anon_rsspage(m, -1)
14175 ++#define vx_anon_rsspages_add(m, p) vx_acc_anon_rsspages(m, p)
14176 ++#define vx_anon_rsspages_sub(m, p) vx_acc_anon_rsspages(m, -(p))
14177 ++
14178 ++
14179 ++#define vx_pages_avail(m, p, r) \
14180 ++ __vx_cres_avail((m)->mm_vx_info, r, p, __FILE__, __LINE__)
14181 ++
14182 ++#define vx_vmpages_avail(m, p) vx_pages_avail(m, p, RLIMIT_AS)
14183 ++#define vx_vmlocked_avail(m, p) vx_pages_avail(m, p, RLIMIT_MEMLOCK)
14184 ++#define vx_anon_avail(m, p) vx_pages_avail(m, p, VLIMIT_ANON)
14185 ++#define vx_mapped_avail(m, p) vx_pages_avail(m, p, VLIMIT_MAPPED)
14186 ++
14187 ++#define vx_rss_avail(m, p) \
14188 ++ __vx_cres_array_avail((m)->mm_vx_info, VLA_RSS, p, __FILE__, __LINE__)
14189 ++
14190 ++
14191 ++enum {
14192 ++ VXPT_UNKNOWN = 0,
14193 ++ VXPT_ANON,
14194 ++ VXPT_NONE,
14195 ++ VXPT_FILE,
14196 ++ VXPT_SWAP,
14197 ++ VXPT_WRITE
14198 ++};
14199 ++
14200 ++#if 0
14201 ++#define vx_page_fault(mm, vma, type, ret)
14202 ++#else
14203 ++
14204 ++static inline
14205 ++void __vx_page_fault(struct mm_struct *mm,
14206 ++ struct vm_area_struct *vma, int type, int ret)
14207 ++{
14208 ++ struct vx_info *vxi = mm->mm_vx_info;
14209 ++ int what;
14210 ++/*
14211 ++ static char *page_type[6] =
14212 ++ { "UNKNOWN", "ANON", "NONE", "FILE", "SWAP", "WRITE" };
14213 ++ static char *page_what[4] =
14214 ++ { "FAULT_OOM", "FAULT_SIGBUS", "FAULT_MINOR", "FAULT_MAJOR" };
14215 ++*/
14216 ++
14217 ++ if (!vxi)
14218 ++ return;
14219 ++
14220 ++ what = (ret & 0x3);
14221 ++
14222 ++/* printk("[%d] page[%d][%d] %2x %s %s\n", vxi->vx_id,
14223 ++ type, what, ret, page_type[type], page_what[what]);
14224 ++*/
14225 ++ if (ret & VM_FAULT_WRITE)
14226 ++ what |= 0x4;
14227 ++ atomic_inc(&vxi->cacct.page[type][what]);
14228 ++}
14229 ++
14230 ++#define vx_page_fault(mm, vma, type, ret) __vx_page_fault(mm, vma, type, ret)
14231 ++#endif
14232 ++
14233 ++
14234 ++extern unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm);
14235 ++
14236 ++#else
14237 ++#warning duplicate inclusion
14238 ++#endif
14239 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_network.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_network.h
14240 +--- linux-2.6.29.4/include/linux/vs_network.h 1970-01-01 01:00:00.000000000 +0100
14241 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_network.h 2009-02-22 22:54:26.000000000 +0100
14242 +@@ -0,0 +1,169 @@
14243 ++#ifndef _NX_VS_NETWORK_H
14244 ++#define _NX_VS_NETWORK_H
14245 ++
14246 ++#include "vserver/context.h"
14247 ++#include "vserver/network.h"
14248 ++#include "vserver/base.h"
14249 ++#include "vserver/check.h"
14250 ++#include "vserver/debug.h"
14251 ++
14252 ++#include <linux/sched.h>
14253 ++
14254 ++
14255 ++#define get_nx_info(i) __get_nx_info(i, __FILE__, __LINE__)
14256 ++
14257 ++static inline struct nx_info *__get_nx_info(struct nx_info *nxi,
14258 ++ const char *_file, int _line)
14259 ++{
14260 ++ if (!nxi)
14261 ++ return NULL;
14262 ++
14263 ++ vxlprintk(VXD_CBIT(nid, 2), "get_nx_info(%p[#%d.%d])",
14264 ++ nxi, nxi ? nxi->nx_id : 0,
14265 ++ nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14266 ++ _file, _line);
14267 ++
14268 ++ atomic_inc(&nxi->nx_usecnt);
14269 ++ return nxi;
14270 ++}
14271 ++
14272 ++
14273 ++extern void free_nx_info(struct nx_info *);
14274 ++
14275 ++#define put_nx_info(i) __put_nx_info(i, __FILE__, __LINE__)
14276 ++
14277 ++static inline void __put_nx_info(struct nx_info *nxi, const char *_file, int _line)
14278 ++{
14279 ++ if (!nxi)
14280 ++ return;
14281 ++
14282 ++ vxlprintk(VXD_CBIT(nid, 2), "put_nx_info(%p[#%d.%d])",
14283 ++ nxi, nxi ? nxi->nx_id : 0,
14284 ++ nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14285 ++ _file, _line);
14286 ++
14287 ++ if (atomic_dec_and_test(&nxi->nx_usecnt))
14288 ++ free_nx_info(nxi);
14289 ++}
14290 ++
14291 ++
14292 ++#define init_nx_info(p, i) __init_nx_info(p, i, __FILE__, __LINE__)
14293 ++
14294 ++static inline void __init_nx_info(struct nx_info **nxp, struct nx_info *nxi,
14295 ++ const char *_file, int _line)
14296 ++{
14297 ++ if (nxi) {
14298 ++ vxlprintk(VXD_CBIT(nid, 3),
14299 ++ "init_nx_info(%p[#%d.%d])",
14300 ++ nxi, nxi ? nxi->nx_id : 0,
14301 ++ nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14302 ++ _file, _line);
14303 ++
14304 ++ atomic_inc(&nxi->nx_usecnt);
14305 ++ }
14306 ++ *nxp = nxi;
14307 ++}
14308 ++
14309 ++
14310 ++#define set_nx_info(p, i) __set_nx_info(p, i, __FILE__, __LINE__)
14311 ++
14312 ++static inline void __set_nx_info(struct nx_info **nxp, struct nx_info *nxi,
14313 ++ const char *_file, int _line)
14314 ++{
14315 ++ struct nx_info *nxo;
14316 ++
14317 ++ if (!nxi)
14318 ++ return;
14319 ++
14320 ++ vxlprintk(VXD_CBIT(nid, 3), "set_nx_info(%p[#%d.%d])",
14321 ++ nxi, nxi ? nxi->nx_id : 0,
14322 ++ nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14323 ++ _file, _line);
14324 ++
14325 ++ atomic_inc(&nxi->nx_usecnt);
14326 ++ nxo = xchg(nxp, nxi);
14327 ++ BUG_ON(nxo);
14328 ++}
14329 ++
14330 ++#define clr_nx_info(p) __clr_nx_info(p, __FILE__, __LINE__)
14331 ++
14332 ++static inline void __clr_nx_info(struct nx_info **nxp,
14333 ++ const char *_file, int _line)
14334 ++{
14335 ++ struct nx_info *nxo;
14336 ++
14337 ++ nxo = xchg(nxp, NULL);
14338 ++ if (!nxo)
14339 ++ return;
14340 ++
14341 ++ vxlprintk(VXD_CBIT(nid, 3), "clr_nx_info(%p[#%d.%d])",
14342 ++ nxo, nxo ? nxo->nx_id : 0,
14343 ++ nxo ? atomic_read(&nxo->nx_usecnt) : 0,
14344 ++ _file, _line);
14345 ++
14346 ++ if (atomic_dec_and_test(&nxo->nx_usecnt))
14347 ++ free_nx_info(nxo);
14348 ++}
14349 ++
14350 ++
14351 ++#define claim_nx_info(v, p) __claim_nx_info(v, p, __FILE__, __LINE__)
14352 ++
14353 ++static inline void __claim_nx_info(struct nx_info *nxi,
14354 ++ struct task_struct *task, const char *_file, int _line)
14355 ++{
14356 ++ vxlprintk(VXD_CBIT(nid, 3), "claim_nx_info(%p[#%d.%d.%d]) %p",
14357 ++ nxi, nxi ? nxi->nx_id : 0,
14358 ++ nxi?atomic_read(&nxi->nx_usecnt):0,
14359 ++ nxi?atomic_read(&nxi->nx_tasks):0,
14360 ++ task, _file, _line);
14361 ++
14362 ++ atomic_inc(&nxi->nx_tasks);
14363 ++}
14364 ++
14365 ++
14366 ++extern void unhash_nx_info(struct nx_info *);
14367 ++
14368 ++#define release_nx_info(v, p) __release_nx_info(v, p, __FILE__, __LINE__)
14369 ++
14370 ++static inline void __release_nx_info(struct nx_info *nxi,
14371 ++ struct task_struct *task, const char *_file, int _line)
14372 ++{
14373 ++ vxlprintk(VXD_CBIT(nid, 3), "release_nx_info(%p[#%d.%d.%d]) %p",
14374 ++ nxi, nxi ? nxi->nx_id : 0,
14375 ++ nxi ? atomic_read(&nxi->nx_usecnt) : 0,
14376 ++ nxi ? atomic_read(&nxi->nx_tasks) : 0,
14377 ++ task, _file, _line);
14378 ++
14379 ++ might_sleep();
14380 ++
14381 ++ if (atomic_dec_and_test(&nxi->nx_tasks))
14382 ++ unhash_nx_info(nxi);
14383 ++}
14384 ++
14385 ++
14386 ++#define task_get_nx_info(i) __task_get_nx_info(i, __FILE__, __LINE__)
14387 ++
14388 ++static __inline__ struct nx_info *__task_get_nx_info(struct task_struct *p,
14389 ++ const char *_file, int _line)
14390 ++{
14391 ++ struct nx_info *nxi;
14392 ++
14393 ++ task_lock(p);
14394 ++ vxlprintk(VXD_CBIT(nid, 5), "task_get_nx_info(%p)",
14395 ++ p, _file, _line);
14396 ++ nxi = __get_nx_info(p->nx_info, _file, _line);
14397 ++ task_unlock(p);
14398 ++ return nxi;
14399 ++}
14400 ++
14401 ++
14402 ++static inline void exit_nx_info(struct task_struct *p)
14403 ++{
14404 ++ if (p->nx_info)
14405 ++ release_nx_info(p->nx_info, p);
14406 ++}
14407 ++
14408 ++
14409 ++#else
14410 ++#warning duplicate inclusion
14411 ++#endif
14412 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_pid.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_pid.h
14413 +--- linux-2.6.29.4/include/linux/vs_pid.h 1970-01-01 01:00:00.000000000 +0100
14414 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_pid.h 2009-02-22 22:54:26.000000000 +0100
14415 +@@ -0,0 +1,95 @@
14416 ++#ifndef _VS_PID_H
14417 ++#define _VS_PID_H
14418 ++
14419 ++#include "vserver/base.h"
14420 ++#include "vserver/check.h"
14421 ++#include "vserver/context.h"
14422 ++#include "vserver/debug.h"
14423 ++#include "vserver/pid.h"
14424 ++#include <linux/pid_namespace.h>
14425 ++
14426 ++
14427 ++#define VXF_FAKE_INIT (VXF_INFO_INIT | VXF_STATE_INIT)
14428 ++
14429 ++static inline
14430 ++int vx_proc_task_visible(struct task_struct *task)
14431 ++{
14432 ++ if ((task->pid == 1) &&
14433 ++ !vx_flags(VXF_FAKE_INIT, VXF_FAKE_INIT))
14434 ++ /* show a blend through init */
14435 ++ goto visible;
14436 ++ if (vx_check(vx_task_xid(task), VS_WATCH | VS_IDENT))
14437 ++ goto visible;
14438 ++ return 0;
14439 ++visible:
14440 ++ return 1;
14441 ++}
14442 ++
14443 ++#define find_task_by_real_pid(pid) find_task_by_pid_ns(pid, &init_pid_ns)
14444 ++
14445 ++#if 0
14446 ++
14447 ++static inline
14448 ++struct task_struct *vx_find_proc_task_by_pid(int pid)
14449 ++{
14450 ++ struct task_struct *task = find_task_by_real_pid(pid);
14451 ++
14452 ++ if (task && !vx_proc_task_visible(task)) {
14453 ++ vxdprintk(VXD_CBIT(misc, 6),
14454 ++ "dropping task (find) %p[#%u,%u] for %p[#%u,%u]",
14455 ++ task, task->xid, task->pid,
14456 ++ current, current->xid, current->pid);
14457 ++ task = NULL;
14458 ++ }
14459 ++ return task;
14460 ++}
14461 ++
14462 ++#endif
14463 ++
14464 ++static inline
14465 ++struct task_struct *vx_get_proc_task(struct inode *inode, struct pid *pid)
14466 ++{
14467 ++ struct task_struct *task = get_pid_task(pid, PIDTYPE_PID);
14468 ++
14469 ++ if (task && !vx_proc_task_visible(task)) {
14470 ++ vxdprintk(VXD_CBIT(misc, 6),
14471 ++ "dropping task (get) %p[#%u,%u] for %p[#%u,%u]",
14472 ++ task, task->xid, task->pid,
14473 ++ current, current->xid, current->pid);
14474 ++ put_task_struct(task);
14475 ++ task = NULL;
14476 ++ }
14477 ++ return task;
14478 ++}
14479 ++
14480 ++#if 0
14481 ++
14482 ++static inline
14483 ++struct task_struct *vx_child_reaper(struct task_struct *p)
14484 ++{
14485 ++ struct vx_info *vxi = p->vx_info;
14486 ++ struct task_struct *reaper = child_reaper(p);
14487 ++
14488 ++ if (!vxi)
14489 ++ goto out;
14490 ++
14491 ++ BUG_ON(!p->vx_info->vx_reaper);
14492 ++
14493 ++ /* child reaper for the guest reaper */
14494 ++ if (vxi->vx_reaper == p)
14495 ++ goto out;
14496 ++
14497 ++ reaper = vxi->vx_reaper;
14498 ++out:
14499 ++ vxdprintk(VXD_CBIT(xid, 7),
14500 ++ "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]",
14501 ++ p, p->xid, p->pid, reaper, reaper->xid, reaper->pid);
14502 ++ return reaper;
14503 ++}
14504 ++
14505 ++#endif
14506 ++
14507 ++
14508 ++#else
14509 ++#warning duplicate inclusion
14510 ++#endif
14511 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_sched.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_sched.h
14512 +--- linux-2.6.29.4/include/linux/vs_sched.h 1970-01-01 01:00:00.000000000 +0100
14513 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_sched.h 2009-02-22 22:54:26.000000000 +0100
14514 +@@ -0,0 +1,110 @@
14515 ++#ifndef _VS_SCHED_H
14516 ++#define _VS_SCHED_H
14517 ++
14518 ++#include "vserver/base.h"
14519 ++#include "vserver/context.h"
14520 ++#include "vserver/sched.h"
14521 ++
14522 ++
14523 ++#define VAVAVOOM_RATIO 50
14524 ++
14525 ++#define MAX_PRIO_BIAS 20
14526 ++#define MIN_PRIO_BIAS -20
14527 ++
14528 ++
14529 ++#ifdef CONFIG_VSERVER_HARDCPU
14530 ++
14531 ++/*
14532 ++ * effective_prio - return the priority that is based on the static
14533 ++ * priority but is modified by bonuses/penalties.
14534 ++ *
14535 ++ * We scale the actual sleep average [0 .... MAX_SLEEP_AVG]
14536 ++ * into a -4 ... 0 ... +4 bonus/penalty range.
14537 ++ *
14538 ++ * Additionally, we scale another amount based on the number of
14539 ++ * CPU tokens currently held by the context, if the process is
14540 ++ * part of a context (and the appropriate SCHED flag is set).
14541 ++ * This ranges from -5 ... 0 ... +15, quadratically.
14542 ++ *
14543 ++ * So, the total bonus is -9 .. 0 .. +19
14544 ++ * We use ~50% of the full 0...39 priority range so that:
14545 ++ *
14546 ++ * 1) nice +19 interactive tasks do not preempt nice 0 CPU hogs.
14547 ++ * 2) nice -20 CPU hogs do not get preempted by nice 0 tasks.
14548 ++ * unless that context is far exceeding its CPU allocation.
14549 ++ *
14550 ++ * Both properties are important to certain workloads.
14551 ++ */
14552 ++static inline
14553 ++int vx_effective_vavavoom(struct _vx_sched_pc *sched_pc, int max_prio)
14554 ++{
14555 ++ int vavavoom, max;
14556 ++
14557 ++ /* lots of tokens = lots of vavavoom
14558 ++ * no tokens = no vavavoom */
14559 ++ if ((vavavoom = sched_pc->tokens) >= 0) {
14560 ++ max = sched_pc->tokens_max;
14561 ++ vavavoom = max - vavavoom;
14562 ++ max = max * max;
14563 ++ vavavoom = max_prio * VAVAVOOM_RATIO / 100
14564 ++ * (vavavoom*vavavoom - (max >> 2)) / max;
14565 ++ return vavavoom;
14566 ++ }
14567 ++ return 0;
14568 ++}
14569 ++
14570 ++
14571 ++static inline
14572 ++int vx_adjust_prio(struct task_struct *p, int prio, int max_user)
14573 ++{
14574 ++ struct vx_info *vxi = p->vx_info;
14575 ++ struct _vx_sched_pc *sched_pc;
14576 ++
14577 ++ if (!vxi)
14578 ++ return prio;
14579 ++
14580 ++ sched_pc = &vx_cpu(vxi, sched_pc);
14581 ++ if (vx_info_flags(vxi, VXF_SCHED_PRIO, 0)) {
14582 ++ int vavavoom = vx_effective_vavavoom(sched_pc, max_user);
14583 ++
14584 ++ sched_pc->vavavoom = vavavoom;
14585 ++ prio += vavavoom;
14586 ++ }
14587 ++ prio += sched_pc->prio_bias;
14588 ++ return prio;
14589 ++}
14590 ++
14591 ++#else /* !CONFIG_VSERVER_HARDCPU */
14592 ++
14593 ++static inline
14594 ++int vx_adjust_prio(struct task_struct *p, int prio, int max_user)
14595 ++{
14596 ++ struct vx_info *vxi = p->vx_info;
14597 ++
14598 ++ if (vxi)
14599 ++ prio += vx_cpu(vxi, sched_pc).prio_bias;
14600 ++ return prio;
14601 ++}
14602 ++
14603 ++#endif /* CONFIG_VSERVER_HARDCPU */
14604 ++
14605 ++
14606 ++static inline void vx_account_user(struct vx_info *vxi,
14607 ++ cputime_t cputime, int nice)
14608 ++{
14609 ++ if (!vxi)
14610 ++ return;
14611 ++ vx_cpu(vxi, sched_pc).user_ticks += cputime;
14612 ++}
14613 ++
14614 ++static inline void vx_account_system(struct vx_info *vxi,
14615 ++ cputime_t cputime, int idle)
14616 ++{
14617 ++ if (!vxi)
14618 ++ return;
14619 ++ vx_cpu(vxi, sched_pc).sys_ticks += cputime;
14620 ++}
14621 ++
14622 ++#else
14623 ++#warning duplicate inclusion
14624 ++#endif
14625 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_socket.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_socket.h
14626 +--- linux-2.6.29.4/include/linux/vs_socket.h 1970-01-01 01:00:00.000000000 +0100
14627 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_socket.h 2009-02-22 22:54:26.000000000 +0100
14628 +@@ -0,0 +1,67 @@
14629 ++#ifndef _VS_SOCKET_H
14630 ++#define _VS_SOCKET_H
14631 ++
14632 ++#include "vserver/debug.h"
14633 ++#include "vserver/base.h"
14634 ++#include "vserver/cacct.h"
14635 ++#include "vserver/context.h"
14636 ++#include "vserver/tag.h"
14637 ++
14638 ++
14639 ++/* socket accounting */
14640 ++
14641 ++#include <linux/socket.h>
14642 ++
14643 ++static inline int vx_sock_type(int family)
14644 ++{
14645 ++ switch (family) {
14646 ++ case PF_UNSPEC:
14647 ++ return VXA_SOCK_UNSPEC;
14648 ++ case PF_UNIX:
14649 ++ return VXA_SOCK_UNIX;
14650 ++ case PF_INET:
14651 ++ return VXA_SOCK_INET;
14652 ++ case PF_INET6:
14653 ++ return VXA_SOCK_INET6;
14654 ++ case PF_PACKET:
14655 ++ return VXA_SOCK_PACKET;
14656 ++ default:
14657 ++ return VXA_SOCK_OTHER;
14658 ++ }
14659 ++}
14660 ++
14661 ++#define vx_acc_sock(v, f, p, s) \
14662 ++ __vx_acc_sock(v, f, p, s, __FILE__, __LINE__)
14663 ++
14664 ++static inline void __vx_acc_sock(struct vx_info *vxi,
14665 ++ int family, int pos, int size, char *file, int line)
14666 ++{
14667 ++ if (vxi) {
14668 ++ int type = vx_sock_type(family);
14669 ++
14670 ++ atomic_long_inc(&vxi->cacct.sock[type][pos].count);
14671 ++ atomic_long_add(size, &vxi->cacct.sock[type][pos].total);
14672 ++ }
14673 ++}
14674 ++
14675 ++#define vx_sock_recv(sk, s) \
14676 ++ vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 0, s)
14677 ++#define vx_sock_send(sk, s) \
14678 ++ vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 1, s)
14679 ++#define vx_sock_fail(sk, s) \
14680 ++ vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 2, s)
14681 ++
14682 ++
14683 ++#define sock_vx_init(s) do { \
14684 ++ (s)->sk_xid = 0; \
14685 ++ (s)->sk_vx_info = NULL; \
14686 ++ } while (0)
14687 ++
14688 ++#define sock_nx_init(s) do { \
14689 ++ (s)->sk_nid = 0; \
14690 ++ (s)->sk_nx_info = NULL; \
14691 ++ } while (0)
14692 ++
14693 ++#else
14694 ++#warning duplicate inclusion
14695 ++#endif
14696 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_tag.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_tag.h
14697 +--- linux-2.6.29.4/include/linux/vs_tag.h 1970-01-01 01:00:00.000000000 +0100
14698 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_tag.h 2009-02-22 22:54:26.000000000 +0100
14699 +@@ -0,0 +1,47 @@
14700 ++#ifndef _VS_TAG_H
14701 ++#define _VS_TAG_H
14702 ++
14703 ++#include <linux/vserver/tag.h>
14704 ++
14705 ++/* check conditions */
14706 ++
14707 ++#define DX_ADMIN 0x0001
14708 ++#define DX_WATCH 0x0002
14709 ++#define DX_HOSTID 0x0008
14710 ++
14711 ++#define DX_IDENT 0x0010
14712 ++
14713 ++#define DX_ARG_MASK 0x0010
14714 ++
14715 ++
14716 ++#define dx_task_tag(t) ((t)->tag)
14717 ++
14718 ++#define dx_current_tag() dx_task_tag(current)
14719 ++
14720 ++#define dx_check(c, m) __dx_check(dx_current_tag(), c, m)
14721 ++
14722 ++#define dx_weak_check(c, m) ((m) ? dx_check(c, m) : 1)
14723 ++
14724 ++
14725 ++/*
14726 ++ * check current context for ADMIN/WATCH and
14727 ++ * optionally against supplied argument
14728 ++ */
14729 ++static inline int __dx_check(tag_t cid, tag_t id, unsigned int mode)
14730 ++{
14731 ++ if (mode & DX_ARG_MASK) {
14732 ++ if ((mode & DX_IDENT) && (id == cid))
14733 ++ return 1;
14734 ++ }
14735 ++ return (((mode & DX_ADMIN) && (cid == 0)) ||
14736 ++ ((mode & DX_WATCH) && (cid == 1)) ||
14737 ++ ((mode & DX_HOSTID) && (id == 0)));
14738 ++}
14739 ++
14740 ++struct inode;
14741 ++int dx_permission(struct inode *inode, int mask);
14742 ++
14743 ++
14744 ++#else
14745 ++#warning duplicate inclusion
14746 ++#endif
14747 +diff -NurpP --minimal linux-2.6.29.4/include/linux/vs_time.h linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_time.h
14748 +--- linux-2.6.29.4/include/linux/vs_time.h 1970-01-01 01:00:00.000000000 +0100
14749 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/linux/vs_time.h 2009-02-22 22:54:26.000000000 +0100
14750 +@@ -0,0 +1,19 @@
14751 ++#ifndef _VS_TIME_H
14752 ++#define _VS_TIME_H
14753 ++
14754 ++
14755 ++/* time faking stuff */
14756 ++
14757 ++#ifdef CONFIG_VSERVER_VTIME
14758 ++
14759 ++extern void vx_gettimeofday(struct timeval *tv);
14760 ++extern int vx_settimeofday(struct timespec *ts);
14761 ++
14762 ++#else
14763 ++#define vx_gettimeofday(t) do_gettimeofday(t)
14764 ++#define vx_settimeofday(t) do_settimeofday(t)
14765 ++#endif
14766 ++
14767 ++#else
14768 ++#warning duplicate inclusion
14769 ++#endif
14770 +diff -NurpP --minimal linux-2.6.29.4/include/net/addrconf.h linux-2.6.29.4-vs2.3.0.36.14/include/net/addrconf.h
14771 +--- linux-2.6.29.4/include/net/addrconf.h 2008-12-25 00:26:37.000000000 +0100
14772 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/net/addrconf.h 2009-02-22 22:54:26.000000000 +0100
14773 +@@ -84,7 +84,8 @@ extern int ipv6_dev_get_saddr(struct n
14774 + struct net_device *dev,
14775 + const struct in6_addr *daddr,
14776 + unsigned int srcprefs,
14777 +- struct in6_addr *saddr);
14778 ++ struct in6_addr *saddr,
14779 ++ struct nx_info *nxi);
14780 + extern int ipv6_get_lladdr(struct net_device *dev,
14781 + struct in6_addr *addr,
14782 + unsigned char banned_flags);
14783 +diff -NurpP --minimal linux-2.6.29.4/include/net/af_unix.h linux-2.6.29.4-vs2.3.0.36.14/include/net/af_unix.h
14784 +--- linux-2.6.29.4/include/net/af_unix.h 2008-12-25 00:26:37.000000000 +0100
14785 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/net/af_unix.h 2009-02-22 22:54:26.000000000 +0100
14786 +@@ -4,6 +4,7 @@
14787 + #include <linux/socket.h>
14788 + #include <linux/un.h>
14789 + #include <linux/mutex.h>
14790 ++#include <linux/vs_base.h>
14791 + #include <net/sock.h>
14792 +
14793 + extern void unix_inflight(struct file *fp);
14794 +diff -NurpP --minimal linux-2.6.29.4/include/net/inet_timewait_sock.h linux-2.6.29.4-vs2.3.0.36.14/include/net/inet_timewait_sock.h
14795 +--- linux-2.6.29.4/include/net/inet_timewait_sock.h 2009-03-24 14:22:44.000000000 +0100
14796 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/net/inet_timewait_sock.h 2009-03-24 14:48:36.000000000 +0100
14797 +@@ -15,15 +15,14 @@
14798 + #ifndef _INET_TIMEWAIT_SOCK_
14799 + #define _INET_TIMEWAIT_SOCK_
14800 +
14801 ++// #include <net/inet_sock.h>
14802 ++#include <net/sock.h>
14803 +
14804 + #include <linux/list.h>
14805 + #include <linux/module.h>
14806 + #include <linux/timer.h>
14807 + #include <linux/types.h>
14808 + #include <linux/workqueue.h>
14809 +-
14810 +-#include <net/inet_sock.h>
14811 +-#include <net/sock.h>
14812 + #include <net/tcp_states.h>
14813 + #include <net/timewait_sock.h>
14814 +
14815 +@@ -116,6 +115,10 @@ struct inet_timewait_sock {
14816 + #define tw_hash __tw_common.skc_hash
14817 + #define tw_prot __tw_common.skc_prot
14818 + #define tw_net __tw_common.skc_net
14819 ++#define tw_xid __tw_common.skc_xid
14820 ++#define tw_vx_info __tw_common.skc_vx_info
14821 ++#define tw_nid __tw_common.skc_nid
14822 ++#define tw_nx_info __tw_common.skc_nx_info
14823 + int tw_timeout;
14824 + volatile unsigned char tw_substate;
14825 + /* 3 bits hole, try to pack */
14826 +diff -NurpP --minimal linux-2.6.29.4/include/net/route.h linux-2.6.29.4-vs2.3.0.36.14/include/net/route.h
14827 +--- linux-2.6.29.4/include/net/route.h 2008-12-25 00:26:37.000000000 +0100
14828 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/net/route.h 2009-02-22 22:54:26.000000000 +0100
14829 +@@ -135,6 +135,9 @@ static inline void ip_rt_put(struct rtab
14830 + dst_release(&rt->u.dst);
14831 + }
14832 +
14833 ++#include <linux/vs_base.h>
14834 ++#include <linux/vs_inet.h>
14835 ++
14836 + #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
14837 +
14838 + extern const __u8 ip_tos2prio[16];
14839 +@@ -144,6 +147,9 @@ static inline char rt_tos2priority(u8 to
14840 + return ip_tos2prio[IPTOS_TOS(tos)>>1];
14841 + }
14842 +
14843 ++extern int ip_v4_find_src(struct net *net, struct nx_info *,
14844 ++ struct rtable **, struct flowi *);
14845 ++
14846 + static inline int ip_route_connect(struct rtable **rp, __be32 dst,
14847 + __be32 src, u32 tos, int oif, u8 protocol,
14848 + __be16 sport, __be16 dport, struct sock *sk,
14849 +@@ -161,11 +167,24 @@ static inline int ip_route_connect(struc
14850 +
14851 + int err;
14852 + struct net *net = sock_net(sk);
14853 ++ struct nx_info *nx_info = current->nx_info;
14854 +
14855 + if (inet_sk(sk)->transparent)
14856 + fl.flags |= FLOWI_FLAG_ANYSRC;
14857 +
14858 +- if (!dst || !src) {
14859 ++ if (sk)
14860 ++ nx_info = sk->sk_nx_info;
14861 ++
14862 ++ vxdprintk(VXD_CBIT(net, 4),
14863 ++ "ip_route_connect(%p) %p,%p;%lx",
14864 ++ sk, nx_info, sk->sk_socket,
14865 ++ (sk->sk_socket?sk->sk_socket->flags:0));
14866 ++
14867 ++ err = ip_v4_find_src(net, nx_info, rp, &fl);
14868 ++ if (err)
14869 ++ return err;
14870 ++
14871 ++ if (!fl.fl4_dst || !fl.fl4_src) {
14872 + err = __ip_route_output_key(net, rp, &fl);
14873 + if (err)
14874 + return err;
14875 +diff -NurpP --minimal linux-2.6.29.4/include/net/sock.h linux-2.6.29.4-vs2.3.0.36.14/include/net/sock.h
14876 +--- linux-2.6.29.4/include/net/sock.h 2009-03-24 14:22:44.000000000 +0100
14877 ++++ linux-2.6.29.4-vs2.3.0.36.14/include/net/sock.h 2009-03-24 14:48:36.000000000 +0100
14878 +@@ -134,6 +134,10 @@ struct sock_common {
14879 + #ifdef CONFIG_NET_NS
14880 + struct net *skc_net;
14881 + #endif
14882 ++ xid_t skc_xid;
14883 ++ struct vx_info *skc_vx_info;
14884 ++ nid_t skc_nid;
14885 ++ struct nx_info *skc_nx_info;
14886 + };
14887 +
14888 + /**
14889 +@@ -218,6 +222,10 @@ struct sock {
14890 + #define sk_hash __sk_common.skc_hash
14891 + #define sk_prot __sk_common.skc_prot
14892 + #define sk_net __sk_common.skc_net
14893 ++#define sk_xid __sk_common.skc_xid
14894 ++#define sk_vx_info __sk_common.skc_vx_info
14895 ++#define sk_nid __sk_common.skc_nid
14896 ++#define sk_nx_info __sk_common.skc_nx_info
14897 + unsigned char sk_shutdown : 2,
14898 + sk_no_check : 2,
14899 + sk_userlocks : 4;
14900 +diff -NurpP --minimal linux-2.6.29.4/init/main.c linux-2.6.29.4-vs2.3.0.36.14/init/main.c
14901 +--- linux-2.6.29.4/init/main.c 2009-03-24 14:22:44.000000000 +0100
14902 ++++ linux-2.6.29.4-vs2.3.0.36.14/init/main.c 2009-03-24 17:22:37.000000000 +0100
14903 +@@ -64,6 +64,7 @@
14904 + #include <linux/ftrace.h>
14905 + #include <linux/async.h>
14906 + #include <trace/boot.h>
14907 ++#include <linux/vserver/percpu.h>
14908 +
14909 + #include <asm/io.h>
14910 + #include <asm/bugs.h>
14911 +@@ -381,12 +382,14 @@ EXPORT_SYMBOL(__per_cpu_offset);
14912 +
14913 + static void __init setup_per_cpu_areas(void)
14914 + {
14915 +- unsigned long size, i;
14916 ++ unsigned long size, vspc, i;
14917 + char *ptr;
14918 + unsigned long nr_possible_cpus = num_possible_cpus();
14919 +
14920 ++ vspc = PERCPU_PERCTX * CONFIG_VSERVER_CONTEXTS;
14921 ++
14922 + /* Copy section for each CPU (we discard the original) */
14923 +- size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
14924 ++ size = ALIGN(PERCPU_ENOUGH_ROOM + vspc, PAGE_SIZE);
14925 + ptr = alloc_bootmem_pages(size * nr_possible_cpus);
14926 +
14927 + for_each_possible_cpu(i) {
14928 +diff -NurpP --minimal linux-2.6.29.4/ipc/mqueue.c linux-2.6.29.4-vs2.3.0.36.14/ipc/mqueue.c
14929 +--- linux-2.6.29.4/ipc/mqueue.c 2009-03-24 14:22:44.000000000 +0100
14930 ++++ linux-2.6.29.4-vs2.3.0.36.14/ipc/mqueue.c 2009-03-24 17:24:19.000000000 +0100
14931 +@@ -31,6 +31,8 @@
14932 + #include <linux/mutex.h>
14933 + #include <linux/nsproxy.h>
14934 + #include <linux/pid.h>
14935 ++#include <linux/vs_context.h>
14936 ++#include <linux/vs_limit.h>
14937 +
14938 + #include <net/sock.h>
14939 + #include "util.h"
14940 +@@ -79,6 +81,7 @@ struct mqueue_inode_info {
14941 + struct sigevent notify;
14942 + struct pid* notify_owner;
14943 + struct user_struct *user; /* user who created, for accounting */
14944 ++ struct vx_info *vxi;
14945 + struct sock *notify_sock;
14946 + struct sk_buff *notify_cookie;
14947 +
14948 +@@ -126,6 +129,7 @@ static struct inode *mqueue_get_inode(st
14949 + if (S_ISREG(mode)) {
14950 + struct mqueue_inode_info *info;
14951 + struct task_struct *p = current;
14952 ++ struct vx_info *vxi = p->vx_info;
14953 + unsigned long mq_bytes, mq_msg_tblsz;
14954 +
14955 + inode->i_fop = &mqueue_file_operations;
14956 +@@ -140,6 +144,7 @@ static struct inode *mqueue_get_inode(st
14957 + info->notify_owner = NULL;
14958 + info->qsize = 0;
14959 + info->user = NULL; /* set when all is ok */
14960 ++ info->vxi = NULL;
14961 + memset(&info->attr, 0, sizeof(info->attr));
14962 + info->attr.mq_maxmsg = msg_max;
14963 + info->attr.mq_msgsize = msgsize_max;
14964 +@@ -154,22 +159,26 @@ static struct inode *mqueue_get_inode(st
14965 + spin_lock(&mq_lock);
14966 + if (u->mq_bytes + mq_bytes < u->mq_bytes ||
14967 + u->mq_bytes + mq_bytes >
14968 +- p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur) {
14969 ++ p->signal->rlim[RLIMIT_MSGQUEUE].rlim_cur ||
14970 ++ !vx_ipcmsg_avail(vxi, mq_bytes)) {
14971 + spin_unlock(&mq_lock);
14972 + goto out_inode;
14973 + }
14974 + u->mq_bytes += mq_bytes;
14975 ++ vx_ipcmsg_add(vxi, u, mq_bytes);
14976 + spin_unlock(&mq_lock);
14977 +
14978 + info->messages = kmalloc(mq_msg_tblsz, GFP_KERNEL);
14979 + if (!info->messages) {
14980 + spin_lock(&mq_lock);
14981 + u->mq_bytes -= mq_bytes;
14982 ++ vx_ipcmsg_sub(vxi, u, mq_bytes);
14983 + spin_unlock(&mq_lock);
14984 + goto out_inode;
14985 + }
14986 + /* all is ok */
14987 + info->user = get_uid(u);
14988 ++ info->vxi = get_vx_info(vxi);
14989 + } else if (S_ISDIR(mode)) {
14990 + inc_nlink(inode);
14991 + /* Some things misbehave if size == 0 on a directory */
14992 +@@ -260,10 +269,14 @@ static void mqueue_delete_inode(struct i
14993 + (info->attr.mq_maxmsg * info->attr.mq_msgsize));
14994 + user = info->user;
14995 + if (user) {
14996 ++ struct vx_info *vxi = info->vxi;
14997 ++
14998 + spin_lock(&mq_lock);
14999 + user->mq_bytes -= mq_bytes;
15000 ++ vx_ipcmsg_sub(vxi, user, mq_bytes);
15001 + queues_count--;
15002 + spin_unlock(&mq_lock);
15003 ++ put_vx_info(vxi);
15004 + free_uid(user);
15005 + }
15006 + }
15007 +diff -NurpP --minimal linux-2.6.29.4/ipc/msg.c linux-2.6.29.4-vs2.3.0.36.14/ipc/msg.c
15008 +--- linux-2.6.29.4/ipc/msg.c 2009-03-24 14:22:44.000000000 +0100
15009 ++++ linux-2.6.29.4-vs2.3.0.36.14/ipc/msg.c 2009-02-22 22:54:26.000000000 +0100
15010 +@@ -38,6 +38,7 @@
15011 + #include <linux/rwsem.h>
15012 + #include <linux/nsproxy.h>
15013 + #include <linux/ipc_namespace.h>
15014 ++#include <linux/vs_base.h>
15015 +
15016 + #include <asm/current.h>
15017 + #include <asm/uaccess.h>
15018 +@@ -190,6 +191,7 @@ static int newque(struct ipc_namespace *
15019 +
15020 + msq->q_perm.mode = msgflg & S_IRWXUGO;
15021 + msq->q_perm.key = key;
15022 ++ msq->q_perm.xid = vx_current_xid();
15023 +
15024 + msq->q_perm.security = NULL;
15025 + retval = security_msg_queue_alloc(msq);
15026 +diff -NurpP --minimal linux-2.6.29.4/ipc/namespace.c linux-2.6.29.4-vs2.3.0.36.14/ipc/namespace.c
15027 +--- linux-2.6.29.4/ipc/namespace.c 2008-12-25 00:26:37.000000000 +0100
15028 ++++ linux-2.6.29.4-vs2.3.0.36.14/ipc/namespace.c 2009-02-22 22:54:26.000000000 +0100
15029 +@@ -9,6 +9,8 @@
15030 + #include <linux/rcupdate.h>
15031 + #include <linux/nsproxy.h>
15032 + #include <linux/slab.h>
15033 ++#include <linux/vs_base.h>
15034 ++#include <linux/vserver/global.h>
15035 +
15036 + #include "util.h"
15037 +
15038 +@@ -35,6 +37,7 @@ static struct ipc_namespace *clone_ipc_n
15039 + register_ipcns_notifier(ns);
15040 +
15041 + kref_init(&ns->kref);
15042 ++ atomic_inc(&vs_global_ipc_ns);
15043 + return ns;
15044 + }
15045 +
15046 +@@ -101,6 +104,7 @@ void free_ipc_ns(struct kref *kref)
15047 + sem_exit_ns(ns);
15048 + msg_exit_ns(ns);
15049 + shm_exit_ns(ns);
15050 ++ atomic_dec(&vs_global_ipc_ns);
15051 + kfree(ns);
15052 + atomic_dec(&nr_ipc_ns);
15053 +
15054 +diff -NurpP --minimal linux-2.6.29.4/ipc/sem.c linux-2.6.29.4-vs2.3.0.36.14/ipc/sem.c
15055 +--- linux-2.6.29.4/ipc/sem.c 2009-03-24 14:22:44.000000000 +0100
15056 ++++ linux-2.6.29.4-vs2.3.0.36.14/ipc/sem.c 2009-03-31 23:31:33.000000000 +0200
15057 +@@ -83,6 +83,8 @@
15058 + #include <linux/rwsem.h>
15059 + #include <linux/nsproxy.h>
15060 + #include <linux/ipc_namespace.h>
15061 ++#include <linux/vs_base.h>
15062 ++#include <linux/vs_limit.h>
15063 +
15064 + #include <asm/uaccess.h>
15065 + #include "util.h"
15066 +@@ -255,6 +257,7 @@ static int newary(struct ipc_namespace *
15067 +
15068 + sma->sem_perm.mode = (semflg & S_IRWXUGO);
15069 + sma->sem_perm.key = key;
15070 ++ sma->sem_perm.xid = vx_current_xid();
15071 +
15072 + sma->sem_perm.security = NULL;
15073 + retval = security_sem_alloc(sma);
15074 +@@ -270,6 +273,9 @@ static int newary(struct ipc_namespace *
15075 + return id;
15076 + }
15077 + ns->used_sems += nsems;
15078 ++ /* FIXME: obsoleted? */
15079 ++ vx_semary_inc(sma);
15080 ++ vx_nsems_add(sma, nsems);
15081 +
15082 + sma->sem_base = (struct sem *) &sma[1];
15083 + INIT_LIST_HEAD(&sma->sem_pending);
15084 +@@ -546,6 +552,9 @@ static void freeary(struct ipc_namespace
15085 + sem_unlock(sma);
15086 +
15087 + ns->used_sems -= sma->sem_nsems;
15088 ++ /* FIXME: obsoleted? */
15089 ++ vx_nsems_sub(sma, sma->sem_nsems);
15090 ++ vx_semary_dec(sma);
15091 + security_sem_free(sma);
15092 + ipc_rcu_putref(sma);
15093 + }
15094 +diff -NurpP --minimal linux-2.6.29.4/ipc/shm.c linux-2.6.29.4-vs2.3.0.36.14/ipc/shm.c
15095 +--- linux-2.6.29.4/ipc/shm.c 2009-03-24 14:22:44.000000000 +0100
15096 ++++ linux-2.6.29.4-vs2.3.0.36.14/ipc/shm.c 2009-03-24 14:48:36.000000000 +0100
15097 +@@ -39,6 +39,8 @@
15098 + #include <linux/nsproxy.h>
15099 + #include <linux/mount.h>
15100 + #include <linux/ipc_namespace.h>
15101 ++#include <linux/vs_context.h>
15102 ++#include <linux/vs_limit.h>
15103 +
15104 + #include <asm/uaccess.h>
15105 +
15106 +@@ -168,7 +170,12 @@ static void shm_open(struct vm_area_stru
15107 + */
15108 + static void shm_destroy(struct ipc_namespace *ns, struct shmid_kernel *shp)
15109 + {
15110 +- ns->shm_tot -= (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15111 ++ struct vx_info *vxi = lookup_vx_info(shp->shm_perm.xid);
15112 ++ int numpages = (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
15113 ++
15114 ++ vx_ipcshm_sub(vxi, shp, numpages);
15115 ++ ns->shm_tot -= numpages;
15116 ++
15117 + shm_rmid(ns, shp);
15118 + shm_unlock(shp);
15119 + if (!is_file_hugepages(shp->shm_file))
15120 +@@ -178,6 +185,7 @@ static void shm_destroy(struct ipc_names
15121 + shp->mlock_user);
15122 + fput (shp->shm_file);
15123 + security_shm_free(shp);
15124 ++ put_vx_info(vxi);
15125 + ipc_rcu_putref(shp);
15126 + }
15127 +
15128 +@@ -348,11 +356,15 @@ static int newseg(struct ipc_namespace *
15129 + if (ns->shm_tot + numpages > ns->shm_ctlall)
15130 + return -ENOSPC;
15131 +
15132 ++ if (!vx_ipcshm_avail(current->vx_info, numpages))
15133 ++ return -ENOSPC;
15134 ++
15135 + shp = ipc_rcu_alloc(sizeof(*shp));
15136 + if (!shp)
15137 + return -ENOMEM;
15138 +
15139 + shp->shm_perm.key = key;
15140 ++ shp->shm_perm.xid = vx_current_xid();
15141 + shp->shm_perm.mode = (shmflg & S_IRWXUGO);
15142 + shp->mlock_user = NULL;
15143 +
15144 +@@ -406,6 +418,7 @@ static int newseg(struct ipc_namespace *
15145 + ns->shm_tot += numpages;
15146 + error = shp->shm_perm.id;
15147 + shm_unlock(shp);
15148 ++ vx_ipcshm_add(current->vx_info, key, numpages);
15149 + return error;
15150 +
15151 + no_id:
15152 +diff -NurpP --minimal linux-2.6.29.4/kernel/capability.c linux-2.6.29.4-vs2.3.0.36.14/kernel/capability.c
15153 +--- linux-2.6.29.4/kernel/capability.c 2009-03-24 14:22:44.000000000 +0100
15154 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/capability.c 2009-03-24 17:27:28.000000000 +0100
15155 +@@ -14,6 +14,7 @@
15156 + #include <linux/security.h>
15157 + #include <linux/syscalls.h>
15158 + #include <linux/pid_namespace.h>
15159 ++#include <linux/vs_context.h>
15160 + #include <asm/uaccess.h>
15161 + #include "cred-internals.h"
15162 +
15163 +@@ -122,6 +123,7 @@ static int cap_validate_magic(cap_user_h
15164 + return 0;
15165 + }
15166 +
15167 ++
15168 + /*
15169 + * The only thing that can change the capabilities of the current
15170 + * process is the current process. As such, we can't be in this code
15171 +@@ -289,6 +291,8 @@ error:
15172 + return ret;
15173 + }
15174 +
15175 ++#include <linux/vserver/base.h>
15176 ++
15177 + /**
15178 + * capable - Determine if the current task has a superior capability in effect
15179 + * @cap: The capability to be tested for
15180 +@@ -301,6 +305,9 @@ error:
15181 + */
15182 + int capable(int cap)
15183 + {
15184 ++ /* here for now so we don't require task locking */
15185 ++ if (vs_check_bit(VXC_CAP_MASK, cap) && !vx_mcaps(1L << cap))
15186 ++ return 0;
15187 + if (unlikely(!cap_valid(cap))) {
15188 + printk(KERN_CRIT "capable() called with invalid cap=%u\n", cap);
15189 + BUG();
15190 +diff -NurpP --minimal linux-2.6.29.4/kernel/compat.c linux-2.6.29.4-vs2.3.0.36.14/kernel/compat.c
15191 +--- linux-2.6.29.4/kernel/compat.c 2009-03-24 14:22:44.000000000 +0100
15192 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/compat.c 2009-03-24 14:48:36.000000000 +0100
15193 +@@ -891,7 +891,7 @@ asmlinkage long compat_sys_time(compat_t
15194 + compat_time_t i;
15195 + struct timeval tv;
15196 +
15197 +- do_gettimeofday(&tv);
15198 ++ vx_gettimeofday(&tv);
15199 + i = tv.tv_sec;
15200 +
15201 + if (tloc) {
15202 +@@ -916,7 +916,7 @@ asmlinkage long compat_sys_stime(compat_
15203 + if (err)
15204 + return err;
15205 +
15206 +- do_settimeofday(&tv);
15207 ++ vx_settimeofday(&tv);
15208 + return 0;
15209 + }
15210 +
15211 +diff -NurpP --minimal linux-2.6.29.4/kernel/exit.c linux-2.6.29.4-vs2.3.0.36.14/kernel/exit.c
15212 +--- linux-2.6.29.4/kernel/exit.c 2009-05-23 23:16:53.000000000 +0200
15213 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/exit.c 2009-05-10 23:58:02.000000000 +0200
15214 +@@ -47,6 +47,10 @@
15215 + #include <linux/task_io_accounting_ops.h>
15216 + #include <linux/tracehook.h>
15217 + #include <linux/init_task.h>
15218 ++#include <linux/vs_limit.h>
15219 ++#include <linux/vs_context.h>
15220 ++#include <linux/vs_network.h>
15221 ++#include <linux/vs_pid.h>
15222 + #include <trace/sched.h>
15223 +
15224 + #include <asm/uaccess.h>
15225 +@@ -497,9 +501,11 @@ static void close_files(struct files_str
15226 + filp_close(file, files);
15227 + cond_resched();
15228 + }
15229 ++ vx_openfd_dec(i);
15230 + }
15231 + i++;
15232 + set >>= 1;
15233 ++ cond_resched();
15234 + }
15235 + }
15236 + }
15237 +@@ -1092,10 +1098,15 @@ NORET_TYPE void do_exit(long code)
15238 + if (tsk->splice_pipe)
15239 + __free_pipe_info(tsk->splice_pipe);
15240 +
15241 ++ /* needs to stay after exit_notify() */
15242 ++ exit_vx_info(tsk, code);
15243 ++ exit_nx_info(tsk);
15244 ++
15245 + preempt_disable();
15246 + /* causes final put_task_struct in finish_task_switch(). */
15247 + tsk->state = TASK_DEAD;
15248 + schedule();
15249 ++ printk("bad task: %p [%lx]\n", current, current->state);
15250 + BUG();
15251 + /* Avoid "noreturn function does return". */
15252 + for (;;)
15253 +diff -NurpP --minimal linux-2.6.29.4/kernel/fork.c linux-2.6.29.4-vs2.3.0.36.14/kernel/fork.c
15254 +--- linux-2.6.29.4/kernel/fork.c 2009-05-23 23:16:53.000000000 +0200
15255 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/fork.c 2009-05-10 23:58:22.000000000 +0200
15256 +@@ -60,6 +60,10 @@
15257 + #include <linux/tty.h>
15258 + #include <linux/proc_fs.h>
15259 + #include <linux/blkdev.h>
15260 ++#include <linux/vs_context.h>
15261 ++#include <linux/vs_network.h>
15262 ++#include <linux/vs_limit.h>
15263 ++#include <linux/vs_memory.h>
15264 + #include <trace/sched.h>
15265 +
15266 + #include <asm/pgtable.h>
15267 +@@ -140,6 +144,8 @@ void free_task(struct task_struct *tsk)
15268 + prop_local_destroy_single(&tsk->dirties);
15269 + free_thread_info(tsk->stack);
15270 + rt_mutex_debug_task_free(tsk);
15271 ++ clr_vx_info(&tsk->vx_info);
15272 ++ clr_nx_info(&tsk->nx_info);
15273 + ftrace_graph_exit_task(tsk);
15274 + free_task_struct(tsk);
15275 + }
15276 +@@ -279,6 +285,8 @@ static int dup_mmap(struct mm_struct *mm
15277 + mm->free_area_cache = oldmm->mmap_base;
15278 + mm->cached_hole_size = ~0UL;
15279 + mm->map_count = 0;
15280 ++ __set_mm_counter(mm, file_rss, 0);
15281 ++ __set_mm_counter(mm, anon_rss, 0);
15282 + cpus_clear(mm->cpu_vm_mask);
15283 + mm->mm_rb = RB_ROOT;
15284 + rb_link = &mm->mm_rb.rb_node;
15285 +@@ -290,7 +298,7 @@ static int dup_mmap(struct mm_struct *mm
15286 +
15287 + if (mpnt->vm_flags & VM_DONTCOPY) {
15288 + long pages = vma_pages(mpnt);
15289 +- mm->total_vm -= pages;
15290 ++ vx_vmpages_sub(mm, pages);
15291 + vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file,
15292 + -pages);
15293 + continue;
15294 +@@ -423,8 +431,8 @@ static struct mm_struct * mm_init(struct
15295 + mm->flags = (current->mm) ? current->mm->flags : default_dump_filter;
15296 + mm->core_state = NULL;
15297 + mm->nr_ptes = 0;
15298 +- set_mm_counter(mm, file_rss, 0);
15299 +- set_mm_counter(mm, anon_rss, 0);
15300 ++ __set_mm_counter(mm, file_rss, 0);
15301 ++ __set_mm_counter(mm, anon_rss, 0);
15302 + spin_lock_init(&mm->page_table_lock);
15303 + spin_lock_init(&mm->ioctx_lock);
15304 + INIT_HLIST_HEAD(&mm->ioctx_list);
15305 +@@ -435,6 +443,7 @@ static struct mm_struct * mm_init(struct
15306 + if (likely(!mm_alloc_pgd(mm))) {
15307 + mm->def_flags = 0;
15308 + mmu_notifier_mm_init(mm);
15309 ++ set_vx_info(&mm->mm_vx_info, p->vx_info);
15310 + return mm;
15311 + }
15312 +
15313 +@@ -468,6 +477,7 @@ void __mmdrop(struct mm_struct *mm)
15314 + mm_free_pgd(mm);
15315 + destroy_context(mm);
15316 + mmu_notifier_mm_destroy(mm);
15317 ++ clr_vx_info(&mm->mm_vx_info);
15318 + free_mm(mm);
15319 + }
15320 + EXPORT_SYMBOL_GPL(__mmdrop);
15321 +@@ -594,6 +604,7 @@ struct mm_struct *dup_mm(struct task_str
15322 + goto fail_nomem;
15323 +
15324 + memcpy(mm, oldmm, sizeof(*mm));
15325 ++ mm->mm_vx_info = NULL;
15326 +
15327 + /* Initializing for Swap token stuff */
15328 + mm->token_priority = 0;
15329 +@@ -627,6 +638,7 @@ fail_nocontext:
15330 + * If init_new_context() failed, we cannot use mmput() to free the mm
15331 + * because it calls destroy_context()
15332 + */
15333 ++ clr_vx_info(&mm->mm_vx_info);
15334 + mm_free_pgd(mm);
15335 + free_mm(mm);
15336 + return NULL;
15337 +@@ -943,6 +955,8 @@ static struct task_struct *copy_process(
15338 + int retval;
15339 + struct task_struct *p;
15340 + int cgroup_callbacks_done = 0;
15341 ++ struct vx_info *vxi;
15342 ++ struct nx_info *nxi;
15343 +
15344 + if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
15345 + return ERR_PTR(-EINVAL);
15346 +@@ -977,12 +991,28 @@ static struct task_struct *copy_process(
15347 + DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
15348 + DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
15349 + #endif
15350 ++ init_vx_info(&p->vx_info, current->vx_info);
15351 ++ init_nx_info(&p->nx_info, current->nx_info);
15352 ++
15353 ++ /* check vserver memory */
15354 ++ if (p->mm && !(clone_flags & CLONE_VM)) {
15355 ++ if (vx_vmpages_avail(p->mm, p->mm->total_vm))
15356 ++ vx_pages_add(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15357 ++ else
15358 ++ goto bad_fork_free;
15359 ++ }
15360 ++ if (p->mm && vx_flags(VXF_FORK_RSS, 0)) {
15361 ++ if (!vx_rss_avail(p->mm, get_mm_counter(p->mm, file_rss)))
15362 ++ goto bad_fork_cleanup_vm;
15363 ++ }
15364 + retval = -EAGAIN;
15365 ++ if (!vx_nproc_avail(1))
15366 ++ goto bad_fork_cleanup_vm;
15367 + if (atomic_read(&p->real_cred->user->processes) >=
15368 + p->signal->rlim[RLIMIT_NPROC].rlim_cur) {
15369 + if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
15370 + p->real_cred->user != INIT_USER)
15371 +- goto bad_fork_free;
15372 ++ goto bad_fork_cleanup_vm;
15373 + }
15374 +
15375 + retval = copy_creds(p, clone_flags);
15376 +@@ -1259,6 +1289,18 @@ static struct task_struct *copy_process(
15377 +
15378 + total_forks++;
15379 + spin_unlock(&current->sighand->siglock);
15380 ++
15381 ++ /* p is copy of current */
15382 ++ vxi = p->vx_info;
15383 ++ if (vxi) {
15384 ++ claim_vx_info(vxi, p);
15385 ++ atomic_inc(&vxi->cvirt.nr_threads);
15386 ++ atomic_inc(&vxi->cvirt.total_forks);
15387 ++ vx_nproc_inc(p);
15388 ++ }
15389 ++ nxi = p->nx_info;
15390 ++ if (nxi)
15391 ++ claim_nx_info(nxi, p);
15392 + write_unlock_irq(&tasklist_lock);
15393 + proc_fork_connector(p);
15394 + cgroup_post_fork(p);
15395 +@@ -1303,6 +1345,9 @@ bad_fork_cleanup_count:
15396 + atomic_dec(&p->cred->user->processes);
15397 + put_cred(p->real_cred);
15398 + put_cred(p->cred);
15399 ++bad_fork_cleanup_vm:
15400 ++ if (p->mm && !(clone_flags & CLONE_VM))
15401 ++ vx_pages_sub(p->vx_info, RLIMIT_AS, p->mm->total_vm);
15402 + bad_fork_free:
15403 + free_task(p);
15404 + fork_out:
15405 +diff -NurpP --minimal linux-2.6.29.4/kernel/kthread.c linux-2.6.29.4-vs2.3.0.36.14/kernel/kthread.c
15406 +--- linux-2.6.29.4/kernel/kthread.c 2009-03-24 14:22:44.000000000 +0100
15407 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/kthread.c 2009-03-24 14:48:36.000000000 +0100
15408 +@@ -13,6 +13,7 @@
15409 + #include <linux/file.h>
15410 + #include <linux/module.h>
15411 + #include <linux/mutex.h>
15412 ++#include <linux/vs_pid.h>
15413 + #include <trace/sched.h>
15414 +
15415 + #define KTHREAD_NICE_LEVEL (-5)
15416 +@@ -102,7 +103,7 @@ static void create_kthread(struct kthrea
15417 + struct sched_param param = { .sched_priority = 0 };
15418 + wait_for_completion(&create->started);
15419 + read_lock(&tasklist_lock);
15420 +- create->result = find_task_by_pid_ns(pid, &init_pid_ns);
15421 ++ create->result = find_task_by_real_pid(pid);
15422 + read_unlock(&tasklist_lock);
15423 + /*
15424 + * root may have changed our (kthreadd's) priority or CPU mask.
15425 +diff -NurpP --minimal linux-2.6.29.4/kernel/Makefile linux-2.6.29.4-vs2.3.0.36.14/kernel/Makefile
15426 +--- linux-2.6.29.4/kernel/Makefile 2009-03-24 14:22:44.000000000 +0100
15427 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/Makefile 2009-03-24 14:48:36.000000000 +0100
15428 +@@ -22,6 +22,7 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg
15429 + CFLAGS_REMOVE_sched_clock.o = -pg
15430 + endif
15431 +
15432 ++obj-y += vserver/
15433 + obj-$(CONFIG_FREEZER) += freezer.o
15434 + obj-$(CONFIG_PROFILING) += profile.o
15435 + obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o
15436 +diff -NurpP --minimal linux-2.6.29.4/kernel/nsproxy.c linux-2.6.29.4-vs2.3.0.36.14/kernel/nsproxy.c
15437 +--- linux-2.6.29.4/kernel/nsproxy.c 2009-03-24 14:22:44.000000000 +0100
15438 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/nsproxy.c 2009-03-25 00:39:02.000000000 +0100
15439 +@@ -19,6 +19,8 @@
15440 + #include <linux/mnt_namespace.h>
15441 + #include <linux/utsname.h>
15442 + #include <linux/pid_namespace.h>
15443 ++#include <linux/vserver/global.h>
15444 ++#include <linux/vserver/debug.h>
15445 + #include <net/net_namespace.h>
15446 + #include <linux/ipc_namespace.h>
15447 +
15448 +@@ -37,6 +39,9 @@ static inline struct nsproxy *clone_nspr
15449 + if (ns) {
15450 + memcpy(ns, orig, sizeof(struct nsproxy));
15451 + atomic_set(&ns->count, 1);
15452 ++ vxdprintk(VXD_CBIT(space, 2), "clone_nsproxy(%p[%u] = %p[1]",
15453 ++ orig, atomic_read(&orig->count), ns);
15454 ++ atomic_inc(&vs_global_nsproxy);
15455 + }
15456 + return ns;
15457 + }
15458 +@@ -46,41 +51,52 @@ static inline struct nsproxy *clone_nspr
15459 + * Return the newly created nsproxy. Do not attach this to the task,
15460 + * leave it to the caller to do proper locking and attach it to task.
15461 + */
15462 +-static struct nsproxy *create_new_namespaces(unsigned long flags,
15463 +- struct task_struct *tsk, struct fs_struct *new_fs)
15464 ++static struct nsproxy *unshare_namespaces(unsigned long flags,
15465 ++ struct nsproxy *orig, struct fs_struct *new_fs)
15466 + {
15467 + struct nsproxy *new_nsp;
15468 + int err;
15469 +
15470 +- new_nsp = clone_nsproxy(tsk->nsproxy);
15471 ++ vxdprintk(VXD_CBIT(space, 4),
15472 ++ "unshare_namespaces(0x%08lx,%p,%p)",
15473 ++ flags, orig, new_fs);
15474 ++
15475 ++ new_nsp = clone_nsproxy(orig);
15476 + if (!new_nsp)
15477 + return ERR_PTR(-ENOMEM);
15478 +
15479 +- new_nsp->mnt_ns = copy_mnt_ns(flags, tsk->nsproxy->mnt_ns, new_fs);
15480 ++ new_nsp->mnt_ns = copy_mnt_ns(flags, orig->mnt_ns, new_fs);
15481 + if (IS_ERR(new_nsp->mnt_ns)) {
15482 + err = PTR_ERR(new_nsp->mnt_ns);
15483 + goto out_ns;
15484 + }
15485 +
15486 +- new_nsp->uts_ns = copy_utsname(flags, tsk->nsproxy->uts_ns);
15487 ++ new_nsp->uts_ns = copy_utsname(flags, orig->uts_ns);
15488 + if (IS_ERR(new_nsp->uts_ns)) {
15489 + err = PTR_ERR(new_nsp->uts_ns);
15490 + goto out_uts;
15491 + }
15492 +
15493 +- new_nsp->ipc_ns = copy_ipcs(flags, tsk->nsproxy->ipc_ns);
15494 ++ new_nsp->ipc_ns = copy_ipcs(flags, orig->ipc_ns);
15495 + if (IS_ERR(new_nsp->ipc_ns)) {
15496 + err = PTR_ERR(new_nsp->ipc_ns);
15497 + goto out_ipc;
15498 + }
15499 +
15500 +- new_nsp->pid_ns = copy_pid_ns(flags, task_active_pid_ns(tsk));
15501 ++ new_nsp->pid_ns = copy_pid_ns(flags, orig->pid_ns);
15502 + if (IS_ERR(new_nsp->pid_ns)) {
15503 + err = PTR_ERR(new_nsp->pid_ns);
15504 + goto out_pid;
15505 + }
15506 +
15507 +- new_nsp->net_ns = copy_net_ns(flags, tsk->nsproxy->net_ns);
15508 ++ /* disabled now?
15509 ++ new_nsp->user_ns = copy_user_ns(flags, orig->user_ns);
15510 ++ if (IS_ERR(new_nsp->user_ns)) {
15511 ++ err = PTR_ERR(new_nsp->user_ns);
15512 ++ goto out_user;
15513 ++ } */
15514 ++
15515 ++ new_nsp->net_ns = copy_net_ns(flags, orig->net_ns);
15516 + if (IS_ERR(new_nsp->net_ns)) {
15517 + err = PTR_ERR(new_nsp->net_ns);
15518 + goto out_net;
15519 +@@ -105,6 +121,35 @@ out_ns:
15520 + return ERR_PTR(err);
15521 + }
15522 +
15523 ++static struct nsproxy *create_new_namespaces(int flags, struct task_struct *tsk,
15524 ++ struct fs_struct *new_fs)
15525 ++{
15526 ++ return unshare_namespaces(flags, tsk->nsproxy, new_fs);
15527 ++}
15528 ++
15529 ++/*
15530 ++ * copies the nsproxy, setting refcount to 1, and grabbing a
15531 ++ * reference to all contained namespaces.
15532 ++ */
15533 ++struct nsproxy *copy_nsproxy(struct nsproxy *orig)
15534 ++{
15535 ++ struct nsproxy *ns = clone_nsproxy(orig);
15536 ++
15537 ++ if (ns) {
15538 ++ if (ns->mnt_ns)
15539 ++ get_mnt_ns(ns->mnt_ns);
15540 ++ if (ns->uts_ns)
15541 ++ get_uts_ns(ns->uts_ns);
15542 ++ if (ns->ipc_ns)
15543 ++ get_ipc_ns(ns->ipc_ns);
15544 ++ if (ns->pid_ns)
15545 ++ get_pid_ns(ns->pid_ns);
15546 ++ if (ns->net_ns)
15547 ++ get_net(ns->net_ns);
15548 ++ }
15549 ++ return ns;
15550 ++}
15551 ++
15552 + /*
15553 + * called from clone. This now handles copy for nsproxy and all
15554 + * namespaces therein.
15555 +@@ -112,9 +157,12 @@ out_ns:
15556 + int copy_namespaces(unsigned long flags, struct task_struct *tsk)
15557 + {
15558 + struct nsproxy *old_ns = tsk->nsproxy;
15559 +- struct nsproxy *new_ns;
15560 ++ struct nsproxy *new_ns = NULL;
15561 + int err = 0;
15562 +
15563 ++ vxdprintk(VXD_CBIT(space, 7), "copy_namespaces(0x%08lx,%p[%p])",
15564 ++ flags, tsk, old_ns);
15565 ++
15566 + if (!old_ns)
15567 + return 0;
15568 +
15569 +@@ -151,6 +199,9 @@ int copy_namespaces(unsigned long flags,
15570 +
15571 + out:
15572 + put_nsproxy(old_ns);
15573 ++ vxdprintk(VXD_CBIT(space, 3),
15574 ++ "copy_namespaces(0x%08lx,%p[%p]) = %d [%p]",
15575 ++ flags, tsk, old_ns, err, new_ns);
15576 + return err;
15577 + }
15578 +
15579 +@@ -164,7 +215,9 @@ void free_nsproxy(struct nsproxy *ns)
15580 + put_ipc_ns(ns->ipc_ns);
15581 + if (ns->pid_ns)
15582 + put_pid_ns(ns->pid_ns);
15583 +- put_net(ns->net_ns);
15584 ++ if (ns->net_ns)
15585 ++ put_net(ns->net_ns);
15586 ++ atomic_dec(&vs_global_nsproxy);
15587 + kmem_cache_free(nsproxy_cachep, ns);
15588 + }
15589 +
15590 +@@ -177,6 +230,10 @@ int unshare_nsproxy_namespaces(unsigned
15591 + {
15592 + int err = 0;
15593 +
15594 ++ vxdprintk(VXD_CBIT(space, 4),
15595 ++ "unshare_nsproxy_namespaces(0x%08lx,[%p])",
15596 ++ unshare_flags, current->nsproxy);
15597 ++
15598 + if (!(unshare_flags & (CLONE_NEWNS | CLONE_NEWUTS | CLONE_NEWIPC |
15599 + CLONE_NEWNET)))
15600 + return 0;
15601 +diff -NurpP --minimal linux-2.6.29.4/kernel/pid.c linux-2.6.29.4-vs2.3.0.36.14/kernel/pid.c
15602 +--- linux-2.6.29.4/kernel/pid.c 2009-03-24 14:22:44.000000000 +0100
15603 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/pid.c 2009-05-10 23:59:04.000000000 +0200
15604 +@@ -36,6 +36,7 @@
15605 + #include <linux/pid_namespace.h>
15606 + #include <linux/init_task.h>
15607 + #include <linux/syscalls.h>
15608 ++#include <linux/vs_pid.h>
15609 +
15610 + #define pid_hashfn(nr, ns) \
15611 + hash_long((unsigned long)nr + (unsigned long)ns, pidhash_shift)
15612 +@@ -305,7 +306,7 @@ EXPORT_SYMBOL_GPL(find_pid_ns);
15613 +
15614 + struct pid *find_vpid(int nr)
15615 + {
15616 +- return find_pid_ns(nr, current->nsproxy->pid_ns);
15617 ++ return find_pid_ns(vx_rmap_pid(nr), current->nsproxy->pid_ns);
15618 + }
15619 + EXPORT_SYMBOL_GPL(find_vpid);
15620 +
15621 +@@ -365,6 +366,9 @@ void transfer_pid(struct task_struct *ol
15622 + struct task_struct *pid_task(struct pid *pid, enum pid_type type)
15623 + {
15624 + struct task_struct *result = NULL;
15625 ++
15626 ++ if (type == PIDTYPE_REALPID)
15627 ++ type = PIDTYPE_PID;
15628 + if (pid) {
15629 + struct hlist_node *first;
15630 + first = rcu_dereference(pid->tasks[type].first);
15631 +@@ -388,14 +392,14 @@ EXPORT_SYMBOL(find_task_by_pid_type_ns);
15632 +
15633 + struct task_struct *find_task_by_vpid(pid_t vnr)
15634 + {
15635 +- return find_task_by_pid_type_ns(PIDTYPE_PID, vnr,
15636 ++ return find_task_by_pid_type_ns(PIDTYPE_PID, vx_rmap_pid(vnr),
15637 + current->nsproxy->pid_ns);
15638 + }
15639 + EXPORT_SYMBOL(find_task_by_vpid);
15640 +
15641 + struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns)
15642 + {
15643 +- return find_task_by_pid_type_ns(PIDTYPE_PID, nr, ns);
15644 ++ return find_task_by_pid_type_ns(PIDTYPE_PID, vx_rmap_pid(nr), ns);
15645 + }
15646 + EXPORT_SYMBOL(find_task_by_pid_ns);
15647 +
15648 +@@ -431,7 +435,7 @@ struct pid *find_get_pid(pid_t nr)
15649 + }
15650 + EXPORT_SYMBOL_GPL(find_get_pid);
15651 +
15652 +-pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15653 ++pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns)
15654 + {
15655 + struct upid *upid;
15656 + pid_t nr = 0;
15657 +@@ -444,6 +448,11 @@ pid_t pid_nr_ns(struct pid *pid, struct
15658 + return nr;
15659 + }
15660 +
15661 ++pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15662 ++{
15663 ++ return vx_map_pid(pid_unmapped_nr_ns(pid, ns));
15664 ++}
15665 ++
15666 + pid_t pid_vnr(struct pid *pid)
15667 + {
15668 + return pid_nr_ns(pid, current->nsproxy->pid_ns);
15669 +diff -NurpP --minimal linux-2.6.29.4/kernel/pid_namespace.c linux-2.6.29.4-vs2.3.0.36.14/kernel/pid_namespace.c
15670 +--- linux-2.6.29.4/kernel/pid_namespace.c 2008-12-25 00:26:37.000000000 +0100
15671 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/pid_namespace.c 2009-02-22 22:54:26.000000000 +0100
15672 +@@ -13,6 +13,7 @@
15673 + #include <linux/syscalls.h>
15674 + #include <linux/err.h>
15675 + #include <linux/acct.h>
15676 ++#include <linux/vserver/global.h>
15677 +
15678 + #define BITS_PER_PAGE (PAGE_SIZE*8)
15679 +
15680 +@@ -85,6 +86,7 @@ static struct pid_namespace *create_pid_
15681 + goto out_free_map;
15682 +
15683 + kref_init(&ns->kref);
15684 ++ atomic_inc(&vs_global_pid_ns);
15685 + ns->level = level;
15686 +
15687 + set_bit(0, ns->pidmap[0].page);
15688 +@@ -109,6 +111,7 @@ static void destroy_pid_namespace(struct
15689 +
15690 + for (i = 0; i < PIDMAP_ENTRIES; i++)
15691 + kfree(ns->pidmap[i].page);
15692 ++ atomic_dec(&vs_global_pid_ns);
15693 + kmem_cache_free(pid_ns_cachep, ns);
15694 + }
15695 +
15696 +diff -NurpP --minimal linux-2.6.29.4/kernel/posix-timers.c linux-2.6.29.4-vs2.3.0.36.14/kernel/posix-timers.c
15697 +--- linux-2.6.29.4/kernel/posix-timers.c 2009-03-24 14:22:44.000000000 +0100
15698 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/posix-timers.c 2009-05-06 21:28:16.000000000 +0200
15699 +@@ -46,6 +46,7 @@
15700 + #include <linux/wait.h>
15701 + #include <linux/workqueue.h>
15702 + #include <linux/module.h>
15703 ++#include <linux/vs_context.h>
15704 +
15705 + /*
15706 + * Management arrays for POSIX timers. Timers are kept in slab memory
15707 +@@ -321,6 +322,7 @@ int posix_timer_event(struct k_itimer *t
15708 + {
15709 + struct task_struct *task;
15710 + int shared, ret = -1;
15711 ++
15712 + /*
15713 + * FIXME: if ->sigq is queued we can race with
15714 + * dequeue_signal()->do_schedule_next_timer().
15715 +@@ -337,10 +339,18 @@ int posix_timer_event(struct k_itimer *t
15716 + rcu_read_lock();
15717 + task = pid_task(timr->it_pid, PIDTYPE_PID);
15718 + if (task) {
15719 ++ struct vx_info_save vxis;
15720 ++ struct vx_info *vxi;
15721 ++
15722 ++ vxi = task_get_vx_info(task);
15723 ++ enter_vx_info(vxi, &vxis);
15724 + shared = !(timr->it_sigev_notify & SIGEV_THREAD_ID);
15725 + ret = send_sigqueue(timr->sigq, task, shared);
15726 ++ leave_vx_info(&vxis);
15727 ++ put_vx_info(vxi);
15728 + }
15729 + rcu_read_unlock();
15730 ++
15731 + /* If we failed to send the signal the timer stops. */
15732 + return ret > 0;
15733 + }
15734 +diff -NurpP --minimal linux-2.6.29.4/kernel/printk.c linux-2.6.29.4-vs2.3.0.36.14/kernel/printk.c
15735 +--- linux-2.6.29.4/kernel/printk.c 2009-03-24 14:22:44.000000000 +0100
15736 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/printk.c 2009-03-24 14:48:36.000000000 +0100
15737 +@@ -32,6 +32,7 @@
15738 + #include <linux/security.h>
15739 + #include <linux/bootmem.h>
15740 + #include <linux/syscalls.h>
15741 ++#include <linux/vs_cvirt.h>
15742 +
15743 + #include <asm/uaccess.h>
15744 +
15745 +@@ -251,18 +252,13 @@ int do_syslog(int type, char __user *buf
15746 + unsigned i, j, limit, count;
15747 + int do_clear = 0;
15748 + char c;
15749 +- int error = 0;
15750 ++ int error;
15751 +
15752 + error = security_syslog(type);
15753 + if (error)
15754 + return error;
15755 +
15756 +- switch (type) {
15757 +- case 0: /* Close log */
15758 +- break;
15759 +- case 1: /* Open log */
15760 +- break;
15761 +- case 2: /* Read from log */
15762 ++ if ((type >= 2) && (type <= 4)) {
15763 + error = -EINVAL;
15764 + if (!buf || len < 0)
15765 + goto out;
15766 +@@ -273,6 +269,16 @@ int do_syslog(int type, char __user *buf
15767 + error = -EFAULT;
15768 + goto out;
15769 + }
15770 ++ }
15771 ++ if (!vx_check(0, VS_ADMIN|VS_WATCH))
15772 ++ return vx_do_syslog(type, buf, len);
15773 ++
15774 ++ switch (type) {
15775 ++ case 0: /* Close log */
15776 ++ break;
15777 ++ case 1: /* Open log */
15778 ++ break;
15779 ++ case 2: /* Read from log */
15780 + error = wait_event_interruptible(log_wait,
15781 + (log_start - log_end));
15782 + if (error)
15783 +@@ -297,16 +303,6 @@ int do_syslog(int type, char __user *buf
15784 + do_clear = 1;
15785 + /* FALL THRU */
15786 + case 3: /* Read last kernel messages */
15787 +- error = -EINVAL;
15788 +- if (!buf || len < 0)
15789 +- goto out;
15790 +- error = 0;
15791 +- if (!len)
15792 +- goto out;
15793 +- if (!access_ok(VERIFY_WRITE, buf, len)) {
15794 +- error = -EFAULT;
15795 +- goto out;
15796 +- }
15797 + count = len;
15798 + if (count > log_buf_len)
15799 + count = log_buf_len;
15800 +diff -NurpP --minimal linux-2.6.29.4/kernel/ptrace.c linux-2.6.29.4-vs2.3.0.36.14/kernel/ptrace.c
15801 +--- linux-2.6.29.4/kernel/ptrace.c 2009-05-23 23:16:53.000000000 +0200
15802 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/ptrace.c 2009-05-10 23:42:01.000000000 +0200
15803 +@@ -21,6 +21,7 @@
15804 + #include <linux/audit.h>
15805 + #include <linux/pid_namespace.h>
15806 + #include <linux/syscalls.h>
15807 ++#include <linux/vs_context.h>
15808 +
15809 + #include <asm/pgtable.h>
15810 + #include <asm/uaccess.h>
15811 +@@ -159,6 +160,11 @@ int __ptrace_may_access(struct task_stru
15812 + dumpable = get_dumpable(task->mm);
15813 + if (!dumpable && !capable(CAP_SYS_PTRACE))
15814 + return -EPERM;
15815 ++ if (!vx_check(task->xid, VS_ADMIN_P|VS_IDENT))
15816 ++ return -EPERM;
15817 ++ if (!vx_check(task->xid, VS_IDENT) &&
15818 ++ !task_vx_flags(task, VXF_STATE_ADMIN, 0))
15819 ++ return -EACCES;
15820 +
15821 + return security_ptrace_may_access(task, mode);
15822 + }
15823 +@@ -596,6 +602,10 @@ SYSCALL_DEFINE4(ptrace, long, request, l
15824 + goto out;
15825 + }
15826 +
15827 ++ ret = -EPERM;
15828 ++ if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
15829 ++ goto out_put_task_struct;
15830 ++
15831 + if (request == PTRACE_ATTACH) {
15832 + ret = ptrace_attach(child);
15833 + /*
15834 +diff -NurpP --minimal linux-2.6.29.4/kernel/sched.c linux-2.6.29.4-vs2.3.0.36.14/kernel/sched.c
15835 +--- linux-2.6.29.4/kernel/sched.c 2009-05-23 23:16:53.000000000 +0200
15836 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sched.c 2009-05-10 23:42:01.000000000 +0200
15837 +@@ -72,6 +72,8 @@
15838 + #include <linux/debugfs.h>
15839 + #include <linux/ctype.h>
15840 + #include <linux/ftrace.h>
15841 ++#include <linux/vs_sched.h>
15842 ++#include <linux/vs_cvirt.h>
15843 + #include <trace/sched.h>
15844 +
15845 + #include <asm/tlb.h>
15846 +@@ -617,6 +619,16 @@ struct rq {
15847 + #endif
15848 + struct hrtimer hrtick_timer;
15849 + #endif
15850 ++ unsigned long norm_time;
15851 ++ unsigned long idle_time;
15852 ++#ifdef CONFIG_VSERVER_IDLETIME
15853 ++ int idle_skip;
15854 ++#endif
15855 ++#ifdef CONFIG_VSERVER_HARDCPU
15856 ++ struct list_head hold_queue;
15857 ++ unsigned long nr_onhold;
15858 ++ int idle_tokens;
15859 ++#endif
15860 +
15861 + #ifdef CONFIG_SCHEDSTATS
15862 + /* latency stats */
15863 +@@ -1713,6 +1725,7 @@ static void update_avg(u64 *avg, u64 sam
15864 +
15865 + static void enqueue_task(struct rq *rq, struct task_struct *p, int wakeup)
15866 + {
15867 ++ // BUG_ON(p->state & TASK_ONHOLD);
15868 + sched_info_queued(p);
15869 + p->sched_class->enqueue_task(rq, p, wakeup);
15870 + p->se.on_rq = 1;
15871 +@@ -1836,6 +1849,8 @@ static inline void check_class_changed(s
15872 + p->sched_class->prio_changed(rq, p, oldprio, running);
15873 + }
15874 +
15875 ++#include "sched_mon.h"
15876 ++
15877 + #ifdef CONFIG_SMP
15878 +
15879 + /* Used instead of source_load when we know the type == 0 */
15880 +@@ -1923,6 +1938,7 @@ migrate_task(struct task_struct *p, int
15881 + {
15882 + struct rq *rq = task_rq(p);
15883 +
15884 ++ vxm_migrate_task(p, rq, dest_cpu);
15885 + /*
15886 + * If the task is not on a runqueue (and not running), then
15887 + * it is sufficient to simply update the task's cpu field.
15888 +@@ -2250,6 +2266,8 @@ static int sched_balance_self(int cpu, i
15889 +
15890 + #endif /* CONFIG_SMP */
15891 +
15892 ++#include "sched_hard.h"
15893 ++
15894 + /***
15895 + * try_to_wake_up - wake up a thread
15896 + * @p: the to-be-woken-up thread
15897 +@@ -2294,6 +2312,13 @@ static int try_to_wake_up(struct task_st
15898 + rq = task_rq_lock(p, &flags);
15899 + update_rq_clock(rq);
15900 + old_state = p->state;
15901 ++
15902 ++ /* we need to unhold suspended tasks */
15903 ++ if (old_state & TASK_ONHOLD) {
15904 ++ vx_unhold_task(p, rq);
15905 ++ old_state = p->state;
15906 ++ }
15907 ++
15908 + if (!(old_state & state))
15909 + goto out;
15910 +
15911 +@@ -2315,6 +2340,12 @@ static int try_to_wake_up(struct task_st
15912 + /* might preempt at this point */
15913 + rq = task_rq_lock(p, &flags);
15914 + old_state = p->state;
15915 ++
15916 ++ /* we need to unhold suspended tasks
15917 ++ if (old_state & TASK_ONHOLD) {
15918 ++ vx_unhold_task(p, rq);
15919 ++ old_state = p->state;
15920 ++ } */
15921 + if (!(old_state & state))
15922 + goto out;
15923 + if (p->se.on_rq)
15924 +@@ -4226,16 +4257,19 @@ void account_user_time(struct task_struc
15925 + cputime_t cputime_scaled)
15926 + {
15927 + struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15928 ++ struct vx_info *vxi = p->vx_info; /* p is _always_ current */
15929 + cputime64_t tmp;
15930 ++ int nice = (TASK_NICE(p) > 0);
15931 +
15932 + /* Add user time to process. */
15933 + p->utime = cputime_add(p->utime, cputime);
15934 + p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
15935 ++ vx_account_user(vxi, cputime, nice);
15936 + account_group_user_time(p, cputime);
15937 +
15938 + /* Add user time to cpustat. */
15939 + tmp = cputime_to_cputime64(cputime);
15940 +- if (TASK_NICE(p) > 0)
15941 ++ if (nice)
15942 + cpustat->nice = cputime64_add(cpustat->nice, tmp);
15943 + else
15944 + cpustat->user = cputime64_add(cpustat->user, tmp);
15945 +@@ -4279,6 +4313,7 @@ void account_system_time(struct task_str
15946 + cputime_t cputime, cputime_t cputime_scaled)
15947 + {
15948 + struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15949 ++ struct vx_info *vxi = p->vx_info; /* p is _always_ current */
15950 + cputime64_t tmp;
15951 +
15952 + if ((p->flags & PF_VCPU) && (irq_count() - hardirq_offset == 0)) {
15953 +@@ -4289,6 +4324,7 @@ void account_system_time(struct task_str
15954 + /* Add system time to process. */
15955 + p->stime = cputime_add(p->stime, cputime);
15956 + p->stimescaled = cputime_add(p->stimescaled, cputime_scaled);
15957 ++ vx_account_system(vxi, cputime, 0 /* do we have idle time? */);
15958 + account_group_system_time(p, cputime);
15959 +
15960 + /* Add system time to cpustat. */
15961 +@@ -4644,6 +4680,11 @@ need_resched_nonpreemptible:
15962 + idle_balance(cpu, rq);
15963 +
15964 + prev->sched_class->put_prev_task(rq, prev);
15965 ++
15966 ++ vx_set_rq_time(rq, jiffies); /* update time */
15967 ++ vx_schedule(prev, rq, cpu); /* hold if over limit */
15968 ++ vx_try_unhold(rq, cpu); /* unhold if refilled */
15969 ++
15970 + next = pick_next_task(rq, prev);
15971 +
15972 + if (likely(prev != next)) {
15973 +@@ -5209,7 +5250,7 @@ SYSCALL_DEFINE1(nice, int, increment)
15974 + nice = 19;
15975 +
15976 + if (increment < 0 && !can_nice(current, nice))
15977 +- return -EPERM;
15978 ++ return vx_flags(VXF_IGNEG_NICE, 0) ? 0 : -EPERM;
15979 +
15980 + retval = security_task_setnice(current, nice);
15981 + if (retval)
15982 +@@ -8485,7 +8526,10 @@ void __init sched_init(void)
15983 +
15984 + #endif
15985 + #endif /* CONFIG_FAIR_GROUP_SCHED */
15986 +-
15987 ++#ifdef CONFIG_VSERVER_HARDCPU
15988 ++ INIT_LIST_HEAD(&rq->hold_queue);
15989 ++ rq->nr_onhold = 0;
15990 ++#endif
15991 + rq->rt.rt_runtime = def_rt_bandwidth.rt_runtime;
15992 + #ifdef CONFIG_RT_GROUP_SCHED
15993 + INIT_LIST_HEAD(&rq->leaf_rt_rq_list);
15994 +diff -NurpP --minimal linux-2.6.29.4/kernel/sched_fair.c linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_fair.c
15995 +--- linux-2.6.29.4/kernel/sched_fair.c 2009-03-24 14:22:45.000000000 +0100
15996 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_fair.c 2009-03-24 14:48:36.000000000 +0100
15997 +@@ -717,6 +717,9 @@ enqueue_entity(struct cfs_rq *cfs_rq, st
15998 + check_spread(cfs_rq, se);
15999 + if (se != cfs_rq->curr)
16000 + __enqueue_entity(cfs_rq, se);
16001 ++
16002 ++ if (entity_is_task(se))
16003 ++ vx_activate_task(task_of(se));
16004 + }
16005 +
16006 + static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
16007 +@@ -760,6 +763,8 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
16008 +
16009 + if (se != cfs_rq->curr)
16010 + __dequeue_entity(cfs_rq, se);
16011 ++ if (entity_is_task(se))
16012 ++ vx_deactivate_task(task_of(se));
16013 + account_entity_dequeue(cfs_rq, se);
16014 + update_min_vruntime(cfs_rq);
16015 + }
16016 +diff -NurpP --minimal linux-2.6.29.4/kernel/sched_hard.h linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_hard.h
16017 +--- linux-2.6.29.4/kernel/sched_hard.h 1970-01-01 01:00:00.000000000 +0100
16018 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_hard.h 2009-02-22 22:54:26.000000000 +0100
16019 +@@ -0,0 +1,353 @@
16020 ++
16021 ++#ifdef CONFIG_VSERVER_IDLELIMIT
16022 ++
16023 ++/*
16024 ++ * vx_idle_resched - reschedule after maxidle
16025 ++ */
16026 ++static inline
16027 ++void vx_idle_resched(struct rq *rq)
16028 ++{
16029 ++ /* maybe have a better criterion for paused */
16030 ++ if (!--rq->idle_tokens && !list_empty(&rq->hold_queue))
16031 ++ set_need_resched();
16032 ++}
16033 ++
16034 ++#else /* !CONFIG_VSERVER_IDLELIMIT */
16035 ++
16036 ++#define vx_idle_resched(rq)
16037 ++
16038 ++#endif /* CONFIG_VSERVER_IDLELIMIT */
16039 ++
16040 ++
16041 ++
16042 ++#ifdef CONFIG_VSERVER_IDLETIME
16043 ++
16044 ++#define vx_set_rq_min_skip(rq, min) \
16045 ++ (rq)->idle_skip = (min)
16046 ++
16047 ++#define vx_save_min_skip(ret, min, val) \
16048 ++ __vx_save_min_skip(ret, min, val)
16049 ++
16050 ++static inline
16051 ++void __vx_save_min_skip(int ret, int *min, int val)
16052 ++{
16053 ++ if (ret > -2)
16054 ++ return;
16055 ++ if ((*min > val) || !*min)
16056 ++ *min = val;
16057 ++}
16058 ++
16059 ++static inline
16060 ++int vx_try_skip(struct rq *rq, int cpu)
16061 ++{
16062 ++ /* artificially advance time */
16063 ++ if (rq->idle_skip > 0) {
16064 ++ vxdprintk(list_empty(&rq->hold_queue),
16065 ++ "hold queue empty on cpu %d", cpu);
16066 ++ rq->idle_time += rq->idle_skip;
16067 ++ vxm_idle_skip(rq, cpu);
16068 ++ return 1;
16069 ++ }
16070 ++ return 0;
16071 ++}
16072 ++
16073 ++#else /* !CONFIG_VSERVER_IDLETIME */
16074 ++
16075 ++#define vx_set_rq_min_skip(rq, min) \
16076 ++ ({ int dummy = (min); dummy; })
16077 ++
16078 ++#define vx_save_min_skip(ret, min, val)
16079 ++
16080 ++static inline
16081 ++int vx_try_skip(struct rq *rq, int cpu)
16082 ++{
16083 ++ return 0;
16084 ++}
16085 ++
16086 ++#endif /* CONFIG_VSERVER_IDLETIME */
16087 ++
16088 ++
16089 ++
16090 ++#ifdef CONFIG_VSERVER_HARDCPU
16091 ++
16092 ++#define vx_set_rq_max_idle(rq, max) \
16093 ++ (rq)->idle_tokens = (max)
16094 ++
16095 ++#define vx_save_max_idle(ret, min, val) \
16096 ++ __vx_save_max_idle(ret, min, val)
16097 ++
16098 ++static inline
16099 ++void __vx_save_max_idle(int ret, int *min, int val)
16100 ++{
16101 ++ if (*min > val)
16102 ++ *min = val;
16103 ++}
16104 ++
16105 ++
16106 ++/*
16107 ++ * vx_hold_task - put a task on the hold queue
16108 ++ */
16109 ++static inline
16110 ++void vx_hold_task(struct task_struct *p, struct rq *rq)
16111 ++{
16112 ++ // printk("@ hold_task(%p[%lx])\n", p, p->state);
16113 ++
16114 ++ /* ignore dead/killed tasks */
16115 ++ if (unlikely(p->state & (TASK_DEAD | TASK_WAKEKILL)))
16116 ++ return;
16117 ++
16118 ++ /* ignore sleeping tasks */
16119 ++ if (unlikely(p->state & TASK_NORMAL))
16120 ++ return;
16121 ++
16122 ++ /* remove task from runqueue */
16123 ++ if (likely(p->se.on_rq))
16124 ++ dequeue_task(rq, p, 0);
16125 ++ else
16126 ++ printk("@ woops, task %p not on runqueue?\n", p);
16127 ++
16128 ++ p->state |= TASK_ONHOLD;
16129 ++ /* a new one on hold */
16130 ++ rq->nr_onhold++;
16131 ++ vxm_hold_task(p, rq);
16132 ++ list_add_tail(&p->hq, &rq->hold_queue);
16133 ++ // list_add_tail(&p->run_list, &rq->hold_queue);
16134 ++}
16135 ++
16136 ++/*
16137 ++ * vx_unhold_task - put a task back to the runqueue
16138 ++ */
16139 ++static inline
16140 ++void vx_unhold_task(struct task_struct *p, struct rq *rq)
16141 ++{
16142 ++ // printk("@ unhold_task(%p[%lx])\n", p, p->state);
16143 ++ list_del_init(&p->hq);
16144 ++ // list_del(&p->run_list);
16145 ++ /* one less waiting */
16146 ++ rq->nr_onhold--;
16147 ++ p->state &= ~TASK_ONHOLD;
16148 ++ enqueue_task(rq, p, 0);
16149 ++ // ? inc_nr_running(p, rq);
16150 ++ vxm_unhold_task(p, rq);
16151 ++}
16152 ++
16153 ++/*
16154 ++ * vx_remove_hold - remove a task from the hold queue
16155 ++ */
16156 ++static inline
16157 ++void vx_remove_hold(struct task_struct *p, struct rq *rq)
16158 ++{
16159 ++ printk("@ remove_hold(%p[%lx])\n", p, p->state);
16160 ++ list_del_init(&p->hq);
16161 ++ // list_del(&p->run_list);
16162 ++ /* one less waiting */
16163 ++ rq->nr_onhold--;
16164 ++ p->state &= ~TASK_ONHOLD;
16165 ++}
16166 ++
16167 ++unsigned long nr_onhold(void)
16168 ++{
16169 ++ unsigned long i, sum = 0;
16170 ++
16171 ++ for_each_online_cpu(i)
16172 ++ sum += cpu_rq(i)->nr_onhold;
16173 ++
16174 ++ return sum;
16175 ++}
16176 ++
16177 ++
16178 ++
16179 ++static inline
16180 ++int __vx_tokens_avail(struct _vx_sched_pc *sched_pc)
16181 ++{
16182 ++ return sched_pc->tokens;
16183 ++}
16184 ++
16185 ++static inline
16186 ++void __vx_consume_token(struct _vx_sched_pc *sched_pc)
16187 ++{
16188 ++ sched_pc->tokens--;
16189 ++}
16190 ++
16191 ++static inline
16192 ++int vx_need_resched(struct task_struct *p, int slice, int cpu)
16193 ++{
16194 ++ struct vx_info *vxi = p->vx_info;
16195 ++
16196 ++ if (vx_info_flags(vxi, VXF_SCHED_HARD|VXF_SCHED_PRIO, 0)) {
16197 ++ struct _vx_sched_pc *sched_pc =
16198 ++ &vx_per_cpu(vxi, sched_pc, cpu);
16199 ++ int tokens;
16200 ++
16201 ++ /* maybe we can simplify that to decrement
16202 ++ the token counter unconditional? */
16203 ++
16204 ++ if ((tokens = __vx_tokens_avail(sched_pc)) > 0)
16205 ++ __vx_consume_token(sched_pc);
16206 ++
16207 ++ /* for tokens > 0, one token was consumed */
16208 ++ if (tokens < 2)
16209 ++ slice = 0;
16210 ++ }
16211 ++ vxm_need_resched(p, slice, cpu);
16212 ++ return (slice == 0);
16213 ++}
16214 ++
16215 ++
16216 ++#define vx_set_rq_time(rq, time) do { \
16217 ++ rq->norm_time = time; \
16218 ++} while (0)
16219 ++
16220 ++
16221 ++static inline
16222 ++void vx_try_unhold(struct rq *rq, int cpu)
16223 ++{
16224 ++ struct vx_info *vxi = NULL;
16225 ++ struct list_head *l, *n;
16226 ++ int maxidle = HZ;
16227 ++ int minskip = 0;
16228 ++
16229 ++ /* nothing to do? what about pause? */
16230 ++ if (list_empty(&rq->hold_queue))
16231 ++ return;
16232 ++
16233 ++ list_for_each_safe(l, n, &rq->hold_queue) {
16234 ++ int ret, delta_min[2];
16235 ++ struct _vx_sched_pc *sched_pc;
16236 ++ struct task_struct *p;
16237 ++
16238 ++ p = list_entry(l, struct task_struct, hq);
16239 ++ /* don't bother with same context */
16240 ++ if (vxi == p->vx_info)
16241 ++ continue;
16242 ++
16243 ++ vxi = p->vx_info;
16244 ++ /* ignore paused contexts */
16245 ++ if (vx_info_flags(vxi, VXF_SCHED_PAUSE, 0))
16246 ++ continue;
16247 ++
16248 ++ sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16249 ++
16250 ++ /* recalc tokens */
16251 ++ vxm_sched_info(sched_pc, vxi, cpu);
16252 ++ ret = vx_tokens_recalc(sched_pc,
16253 ++ &rq->norm_time, &rq->idle_time, delta_min);
16254 ++ vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16255 ++
16256 ++ if (ret > 0) {
16257 ++ /* we found a runable context */
16258 ++ vx_unhold_task(p, rq);
16259 ++ break;
16260 ++ }
16261 ++ vx_save_max_idle(ret, &maxidle, delta_min[0]);
16262 ++ vx_save_min_skip(ret, &minskip, delta_min[1]);
16263 ++ }
16264 ++ vx_set_rq_max_idle(rq, maxidle);
16265 ++ vx_set_rq_min_skip(rq, minskip);
16266 ++ vxm_rq_max_min(rq, cpu);
16267 ++}
16268 ++
16269 ++
16270 ++static inline
16271 ++int vx_schedule(struct task_struct *next, struct rq *rq, int cpu)
16272 ++{
16273 ++ struct vx_info *vxi = next->vx_info;
16274 ++ struct _vx_sched_pc *sched_pc;
16275 ++ int delta_min[2];
16276 ++ int flags, ret;
16277 ++
16278 ++ if (!vxi)
16279 ++ return 1;
16280 ++
16281 ++ flags = vxi->vx_flags;
16282 ++
16283 ++ if (unlikely(vs_check_flags(flags, VXF_SCHED_PAUSE, 0)))
16284 ++ goto put_on_hold;
16285 ++ if (!vs_check_flags(flags, VXF_SCHED_HARD | VXF_SCHED_PRIO, 0))
16286 ++ return 1;
16287 ++
16288 ++ sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16289 ++#ifdef CONFIG_SMP
16290 ++ /* update scheduler params */
16291 ++ if (cpu_isset(cpu, vxi->sched.update)) {
16292 ++ vx_update_sched_param(&vxi->sched, sched_pc);
16293 ++ vxm_update_sched(sched_pc, vxi, cpu);
16294 ++ cpu_clear(cpu, vxi->sched.update);
16295 ++ }
16296 ++#endif
16297 ++ vxm_sched_info(sched_pc, vxi, cpu);
16298 ++ ret = vx_tokens_recalc(sched_pc,
16299 ++ &rq->norm_time, &rq->idle_time, delta_min);
16300 ++ vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16301 ++
16302 ++ if (!vs_check_flags(flags, VXF_SCHED_HARD, 0))
16303 ++ return 1;
16304 ++
16305 ++ if (unlikely(ret < 0)) {
16306 ++ vx_save_max_idle(ret, &rq->idle_tokens, delta_min[0]);
16307 ++ vx_save_min_skip(ret, &rq->idle_skip, delta_min[1]);
16308 ++ vxm_rq_max_min(rq, cpu);
16309 ++ put_on_hold:
16310 ++ vx_hold_task(next, rq);
16311 ++ return 0;
16312 ++ }
16313 ++ return 1;
16314 ++}
16315 ++
16316 ++
16317 ++#else /* CONFIG_VSERVER_HARDCPU */
16318 ++
16319 ++static inline
16320 ++void vx_hold_task(struct task_struct *p, struct rq *rq)
16321 ++{
16322 ++ return;
16323 ++}
16324 ++
16325 ++static inline
16326 ++void vx_unhold_task(struct task_struct *p, struct rq *rq)
16327 ++{
16328 ++ return;
16329 ++}
16330 ++
16331 ++unsigned long nr_onhold(void)
16332 ++{
16333 ++ return 0;
16334 ++}
16335 ++
16336 ++
16337 ++static inline
16338 ++int vx_need_resched(struct task_struct *p, int slice, int cpu)
16339 ++{
16340 ++ return (slice == 0);
16341 ++}
16342 ++
16343 ++
16344 ++#define vx_set_rq_time(rq, time)
16345 ++
16346 ++static inline
16347 ++void vx_try_unhold(struct rq *rq, int cpu)
16348 ++{
16349 ++ return;
16350 ++}
16351 ++
16352 ++static inline
16353 ++int vx_schedule(struct task_struct *next, struct rq *rq, int cpu)
16354 ++{
16355 ++ struct vx_info *vxi = next->vx_info;
16356 ++ struct _vx_sched_pc *sched_pc;
16357 ++ int delta_min[2];
16358 ++ int ret;
16359 ++
16360 ++ if (!vx_info_flags(vxi, VXF_SCHED_PRIO, 0))
16361 ++ return 1;
16362 ++
16363 ++ sched_pc = &vx_per_cpu(vxi, sched_pc, cpu);
16364 ++ vxm_sched_info(sched_pc, vxi, cpu);
16365 ++ ret = vx_tokens_recalc(sched_pc,
16366 ++ &rq->norm_time, &rq->idle_time, delta_min);
16367 ++ vxm_tokens_recalc(sched_pc, rq, vxi, cpu);
16368 ++ return 1;
16369 ++}
16370 ++
16371 ++#endif /* CONFIG_VSERVER_HARDCPU */
16372 ++
16373 +diff -NurpP --minimal linux-2.6.29.4/kernel/sched_mon.h linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_mon.h
16374 +--- linux-2.6.29.4/kernel/sched_mon.h 1970-01-01 01:00:00.000000000 +0100
16375 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sched_mon.h 2009-02-22 22:54:26.000000000 +0100
16376 +@@ -0,0 +1,200 @@
16377 ++
16378 ++#include <linux/vserver/monitor.h>
16379 ++
16380 ++#ifdef CONFIG_VSERVER_MONITOR
16381 ++
16382 ++#ifdef CONFIG_VSERVER_HARDCPU
16383 ++#define HARDCPU(x) (x)
16384 ++#else
16385 ++#define HARDCPU(x) (0)
16386 ++#endif
16387 ++
16388 ++#ifdef CONFIG_VSERVER_IDLETIME
16389 ++#define IDLETIME(x) (x)
16390 ++#else
16391 ++#define IDLETIME(x) (0)
16392 ++#endif
16393 ++
16394 ++struct _vx_mon_entry *vxm_advance(int cpu);
16395 ++
16396 ++
16397 ++static inline
16398 ++void __vxm_basic(struct _vx_mon_entry *entry, xid_t xid, int type)
16399 ++{
16400 ++ entry->type = type;
16401 ++ entry->xid = xid;
16402 ++}
16403 ++
16404 ++static inline
16405 ++void __vxm_sync(int cpu)
16406 ++{
16407 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16408 ++
16409 ++ __vxm_basic(entry, 0, VXM_SYNC);
16410 ++ entry->ev.sec = xtime.tv_sec;
16411 ++ entry->ev.nsec = xtime.tv_nsec;
16412 ++}
16413 ++
16414 ++static inline
16415 ++void __vxm_task(struct task_struct *p, int type)
16416 ++{
16417 ++ struct _vx_mon_entry *entry = vxm_advance(task_cpu(p));
16418 ++
16419 ++ __vxm_basic(entry, p->xid, type);
16420 ++ entry->ev.tsk.pid = p->pid;
16421 ++ entry->ev.tsk.state = p->state;
16422 ++}
16423 ++
16424 ++static inline
16425 ++void __vxm_sched(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16426 ++{
16427 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16428 ++
16429 ++ __vxm_basic(entry, vxi->vx_id, (VXM_SCHED | s->flags));
16430 ++ entry->sd.tokens = s->tokens;
16431 ++ entry->sd.norm_time = s->norm_time;
16432 ++ entry->sd.idle_time = s->idle_time;
16433 ++}
16434 ++
16435 ++static inline
16436 ++void __vxm_rqinfo1(struct rq *q, int cpu)
16437 ++{
16438 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16439 ++
16440 ++ entry->type = VXM_RQINFO_1;
16441 ++ entry->xid = ((unsigned long)q >> 16) & 0xffff;
16442 ++ entry->q1.running = q->nr_running;
16443 ++ entry->q1.onhold = HARDCPU(q->nr_onhold);
16444 ++ entry->q1.iowait = atomic_read(&q->nr_iowait);
16445 ++ entry->q1.uintr = q->nr_uninterruptible;
16446 ++ entry->q1.idle_tokens = IDLETIME(q->idle_tokens);
16447 ++}
16448 ++
16449 ++static inline
16450 ++void __vxm_rqinfo2(struct rq *q, int cpu)
16451 ++{
16452 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16453 ++
16454 ++ entry->type = VXM_RQINFO_2;
16455 ++ entry->xid = (unsigned long)q & 0xffff;
16456 ++ entry->q2.norm_time = q->norm_time;
16457 ++ entry->q2.idle_time = q->idle_time;
16458 ++ entry->q2.idle_skip = IDLETIME(q->idle_skip);
16459 ++}
16460 ++
16461 ++static inline
16462 ++void __vxm_update(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16463 ++{
16464 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16465 ++
16466 ++ __vxm_basic(entry, vxi->vx_id, VXM_UPDATE);
16467 ++ entry->ev.tokens = s->tokens;
16468 ++}
16469 ++
16470 ++static inline
16471 ++void __vxm_update1(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16472 ++{
16473 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16474 ++
16475 ++ __vxm_basic(entry, vxi->vx_id, VXM_UPDATE_1);
16476 ++ entry->u1.tokens_max = s->tokens_max;
16477 ++ entry->u1.fill_rate = s->fill_rate[0];
16478 ++ entry->u1.interval = s->interval[0];
16479 ++}
16480 ++
16481 ++static inline
16482 ++void __vxm_update2(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16483 ++{
16484 ++ struct _vx_mon_entry *entry = vxm_advance(cpu);
16485 ++
16486 ++ __vxm_basic(entry, vxi->vx_id, VXM_UPDATE_2);
16487 ++ entry->u2.tokens_min = s->tokens_min;
16488 ++ entry->u2.fill_rate = s->fill_rate[1];
16489 ++ entry->u2.interval = s->interval[1];
16490 ++}
16491 ++
16492 ++
16493 ++#define vxm_activate_task(p,q) __vxm_task(p, VXM_ACTIVATE)
16494 ++#define vxm_activate_idle(p,q) __vxm_task(p, VXM_IDLE)
16495 ++#define vxm_deactivate_task(p,q) __vxm_task(p, VXM_DEACTIVATE)
16496 ++#define vxm_hold_task(p,q) __vxm_task(p, VXM_HOLD)
16497 ++#define vxm_unhold_task(p,q) __vxm_task(p, VXM_UNHOLD)
16498 ++
16499 ++static inline
16500 ++void vxm_migrate_task(struct task_struct *p, struct rq *rq, int dest)
16501 ++{
16502 ++ __vxm_task(p, VXM_MIGRATE);
16503 ++ __vxm_rqinfo1(rq, task_cpu(p));
16504 ++ __vxm_rqinfo2(rq, task_cpu(p));
16505 ++}
16506 ++
16507 ++static inline
16508 ++void vxm_idle_skip(struct rq *rq, int cpu)
16509 ++{
16510 ++ __vxm_rqinfo1(rq, cpu);
16511 ++ __vxm_rqinfo2(rq, cpu);
16512 ++}
16513 ++
16514 ++static inline
16515 ++void vxm_need_resched(struct task_struct *p, int slice, int cpu)
16516 ++{
16517 ++ if (slice)
16518 ++ return;
16519 ++
16520 ++ __vxm_task(p, VXM_RESCHED);
16521 ++}
16522 ++
16523 ++static inline
16524 ++void vxm_sync(unsigned long now, int cpu)
16525 ++{
16526 ++ if (!CONFIG_VSERVER_MONITOR_SYNC ||
16527 ++ (now % CONFIG_VSERVER_MONITOR_SYNC))
16528 ++ return;
16529 ++
16530 ++ __vxm_sync(cpu);
16531 ++}
16532 ++
16533 ++#define vxm_sched_info(s,v,c) __vxm_sched(s,v,c)
16534 ++
16535 ++static inline
16536 ++void vxm_tokens_recalc(struct _vx_sched_pc *s, struct rq *rq,
16537 ++ struct vx_info *vxi, int cpu)
16538 ++{
16539 ++ __vxm_sched(s, vxi, cpu);
16540 ++ __vxm_rqinfo2(rq, cpu);
16541 ++}
16542 ++
16543 ++static inline
16544 ++void vxm_update_sched(struct _vx_sched_pc *s, struct vx_info *vxi, int cpu)
16545 ++{
16546 ++ __vxm_sched(s, vxi, cpu);
16547 ++ __vxm_update(s, vxi, cpu);
16548 ++ __vxm_update1(s, vxi, cpu);
16549 ++ __vxm_update2(s, vxi, cpu);
16550 ++}
16551 ++
16552 ++static inline
16553 ++void vxm_rq_max_min(struct rq *rq, int cpu)
16554 ++{
16555 ++ __vxm_rqinfo1(rq, cpu);
16556 ++ __vxm_rqinfo2(rq, cpu);
16557 ++}
16558 ++
16559 ++#else /* CONFIG_VSERVER_MONITOR */
16560 ++
16561 ++#define vxm_activate_task(t,q) do { } while (0)
16562 ++#define vxm_activate_idle(t,q) do { } while (0)
16563 ++#define vxm_deactivate_task(t,q) do { } while (0)
16564 ++#define vxm_hold_task(t,q) do { } while (0)
16565 ++#define vxm_unhold_task(t,q) do { } while (0)
16566 ++#define vxm_migrate_task(t,q,d) do { } while (0)
16567 ++#define vxm_idle_skip(q,c) do { } while (0)
16568 ++#define vxm_need_resched(t,s,c) do { } while (0)
16569 ++#define vxm_sync(s,c) do { } while (0)
16570 ++#define vxm_sched_info(s,v,c) do { } while (0)
16571 ++#define vxm_tokens_recalc(s,q,v,c) do { } while (0)
16572 ++#define vxm_update_sched(s,v,c) do { } while (0)
16573 ++#define vxm_rq_max_min(q,c) do { } while (0)
16574 ++
16575 ++#endif /* CONFIG_VSERVER_MONITOR */
16576 ++
16577 +diff -NurpP --minimal linux-2.6.29.4/kernel/signal.c linux-2.6.29.4-vs2.3.0.36.14/kernel/signal.c
16578 +--- linux-2.6.29.4/kernel/signal.c 2009-03-24 14:22:45.000000000 +0100
16579 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/signal.c 2009-04-20 23:37:12.000000000 +0200
16580 +@@ -27,6 +27,8 @@
16581 + #include <linux/freezer.h>
16582 + #include <linux/pid_namespace.h>
16583 + #include <linux/nsproxy.h>
16584 ++#include <linux/vs_context.h>
16585 ++#include <linux/vs_pid.h>
16586 + #include <trace/sched.h>
16587 +
16588 + #include <asm/param.h>
16589 +@@ -584,6 +586,14 @@ static int check_kill_permission(int sig
16590 + if (!valid_signal(sig))
16591 + return -EINVAL;
16592 +
16593 ++ if ((info != SEND_SIG_NOINFO) &&
16594 ++ (is_si_special(info) || !SI_FROMUSER(info)))
16595 ++ goto skip;
16596 ++
16597 ++ vxdprintk(VXD_CBIT(misc, 7),
16598 ++ "check_kill_permission(%d,%p,%p[#%u,%u])",
16599 ++ sig, info, t, vx_task_xid(t), t->pid);
16600 ++
16601 + if (info != SEND_SIG_NOINFO && (is_si_special(info) || SI_FROMKERNEL(info)))
16602 + return 0;
16603 +
16604 +@@ -611,6 +621,20 @@ static int check_kill_permission(int sig
16605 + }
16606 + }
16607 +
16608 ++ error = -EPERM;
16609 ++ if (t->pid == 1 && current->xid)
16610 ++ return error;
16611 ++
16612 ++ error = -ESRCH;
16613 ++ /* FIXME: we shouldn't return ESRCH ever, to avoid
16614 ++ loops, maybe ENOENT or EACCES? */
16615 ++ if (!vx_check(vx_task_xid(t), VS_WATCH_P | VS_IDENT)) {
16616 ++ vxdprintk(current->xid || VXD_CBIT(misc, 7),
16617 ++ "signal %d[%p] xid mismatch %p[#%u,%u] xid=#%u",
16618 ++ sig, info, t, vx_task_xid(t), t->pid, current->xid);
16619 ++ return error;
16620 ++ }
16621 ++skip:
16622 + return security_task_kill(t, info, sig, 0);
16623 + }
16624 +
16625 +@@ -1076,7 +1100,7 @@ int kill_pid_info(int sig, struct siginf
16626 + rcu_read_lock();
16627 + retry:
16628 + p = pid_task(pid, PIDTYPE_PID);
16629 +- if (p) {
16630 ++ if (p && vx_check(vx_task_xid(p), VS_IDENT)) {
16631 + error = group_send_sig_info(sig, info, p);
16632 + if (unlikely(error == -ESRCH))
16633 + /*
16634 +@@ -1115,7 +1139,7 @@ int kill_pid_info_as_uid(int sig, struct
16635 +
16636 + read_lock(&tasklist_lock);
16637 + p = pid_task(pid, PIDTYPE_PID);
16638 +- if (!p) {
16639 ++ if (!p || !vx_check(vx_task_xid(p), VS_IDENT)) {
16640 + ret = -ESRCH;
16641 + goto out_unlock;
16642 + }
16643 +@@ -1169,8 +1193,10 @@ static int kill_something_info(int sig,
16644 + struct task_struct * p;
16645 +
16646 + for_each_process(p) {
16647 +- if (task_pid_vnr(p) > 1 &&
16648 +- !same_thread_group(p, current)) {
16649 ++ if (vx_check(vx_task_xid(p), VS_ADMIN|VS_IDENT) &&
16650 ++ task_pid_vnr(p) > 1 &&
16651 ++ !same_thread_group(p, current) &&
16652 ++ !vx_current_initpid(p->pid)) {
16653 + int err = group_send_sig_info(sig, info, p);
16654 + ++count;
16655 + if (err != -EPERM)
16656 +@@ -1849,6 +1875,11 @@ relock:
16657 + !signal_group_exit(signal))
16658 + continue;
16659 +
16660 ++ /* virtual init is protected against user signals */
16661 ++ if ((info->si_code == SI_USER) &&
16662 ++ vx_current_initpid(current->pid))
16663 ++ continue;
16664 ++
16665 + if (sig_kernel_stop(signr)) {
16666 + /*
16667 + * The default action is to stop all threads in
16668 +diff -NurpP --minimal linux-2.6.29.4/kernel/softirq.c linux-2.6.29.4-vs2.3.0.36.14/kernel/softirq.c
16669 +--- linux-2.6.29.4/kernel/softirq.c 2009-05-23 23:16:53.000000000 +0200
16670 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/softirq.c 2009-04-30 12:14:53.000000000 +0200
16671 +@@ -23,6 +23,7 @@
16672 + #include <linux/rcupdate.h>
16673 + #include <linux/smp.h>
16674 + #include <linux/tick.h>
16675 ++#include <linux/vs_context.h>
16676 +
16677 + #include <asm/irq.h>
16678 + /*
16679 +diff -NurpP --minimal linux-2.6.29.4/kernel/sys.c linux-2.6.29.4-vs2.3.0.36.14/kernel/sys.c
16680 +--- linux-2.6.29.4/kernel/sys.c 2009-03-24 14:22:45.000000000 +0100
16681 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sys.c 2009-03-24 14:48:36.000000000 +0100
16682 +@@ -39,6 +39,7 @@
16683 + #include <linux/syscalls.h>
16684 + #include <linux/kprobes.h>
16685 + #include <linux/user_namespace.h>
16686 ++#include <linux/vs_pid.h>
16687 +
16688 + #include <asm/uaccess.h>
16689 + #include <asm/io.h>
16690 +@@ -128,7 +129,10 @@ static int set_one_prio(struct task_stru
16691 + goto out;
16692 + }
16693 + if (niceval < task_nice(p) && !can_nice(p, niceval)) {
16694 +- error = -EACCES;
16695 ++ if (vx_flags(VXF_IGNEG_NICE, 0))
16696 ++ error = 0;
16697 ++ else
16698 ++ error = -EACCES;
16699 + goto out;
16700 + }
16701 + no_nice = security_task_setnice(p, niceval);
16702 +@@ -177,6 +181,8 @@ SYSCALL_DEFINE3(setpriority, int, which,
16703 + else
16704 + pgrp = task_pgrp(current);
16705 + do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
16706 ++ if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
16707 ++ continue;
16708 + error = set_one_prio(p, niceval, error);
16709 + } while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
16710 + break;
16711 +@@ -238,6 +244,8 @@ SYSCALL_DEFINE2(getpriority, int, which,
16712 + else
16713 + pgrp = task_pgrp(current);
16714 + do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
16715 ++ if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
16716 ++ continue;
16717 + niceval = 20 - task_nice(p);
16718 + if (niceval > retval)
16719 + retval = niceval;
16720 +@@ -347,6 +355,9 @@ void kernel_power_off(void)
16721 + machine_power_off();
16722 + }
16723 + EXPORT_SYMBOL_GPL(kernel_power_off);
16724 ++
16725 ++long vs_reboot(unsigned int, void __user *);
16726 ++
16727 + /*
16728 + * Reboot system call: for obvious reasons only root may call it,
16729 + * and even root needs to set up some magic numbers in the registers
16730 +@@ -378,6 +389,9 @@ SYSCALL_DEFINE4(reboot, int, magic1, int
16731 + if ((cmd == LINUX_REBOOT_CMD_POWER_OFF) && !pm_power_off)
16732 + cmd = LINUX_REBOOT_CMD_HALT;
16733 +
16734 ++ if (!vx_check(0, VS_ADMIN|VS_WATCH))
16735 ++ return vs_reboot(cmd, arg);
16736 ++
16737 + lock_kernel();
16738 + switch (cmd) {
16739 + case LINUX_REBOOT_CMD_RESTART:
16740 +@@ -1420,7 +1434,7 @@ SYSCALL_DEFINE2(sethostname, char __user
16741 + int errno;
16742 + char tmp[__NEW_UTS_LEN];
16743 +
16744 +- if (!capable(CAP_SYS_ADMIN))
16745 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
16746 + return -EPERM;
16747 + if (len < 0 || len > __NEW_UTS_LEN)
16748 + return -EINVAL;
16749 +@@ -1469,7 +1483,7 @@ SYSCALL_DEFINE2(setdomainname, char __us
16750 + int errno;
16751 + char tmp[__NEW_UTS_LEN];
16752 +
16753 +- if (!capable(CAP_SYS_ADMIN))
16754 ++ if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
16755 + return -EPERM;
16756 + if (len < 0 || len > __NEW_UTS_LEN)
16757 + return -EINVAL;
16758 +@@ -1538,7 +1552,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int,
16759 + return -EINVAL;
16760 + old_rlim = current->signal->rlim + resource;
16761 + if ((new_rlim.rlim_max > old_rlim->rlim_max) &&
16762 +- !capable(CAP_SYS_RESOURCE))
16763 ++ !vx_capable(CAP_SYS_RESOURCE, VXC_SET_RLIMIT))
16764 + return -EPERM;
16765 + if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open)
16766 + return -EPERM;
16767 +diff -NurpP --minimal linux-2.6.29.4/kernel/sysctl.c linux-2.6.29.4-vs2.3.0.36.14/kernel/sysctl.c
16768 +--- linux-2.6.29.4/kernel/sysctl.c 2009-05-23 23:16:53.000000000 +0200
16769 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sysctl.c 2009-04-30 12:14:53.000000000 +0200
16770 +@@ -111,6 +111,7 @@ static int ngroups_max = NGROUPS_MAX;
16771 + #ifdef CONFIG_MODULES
16772 + extern char modprobe_path[];
16773 + #endif
16774 ++extern char vshelper_path[];
16775 + #ifdef CONFIG_CHR_DEV_SG
16776 + extern int sg_big_buff;
16777 + #endif
16778 +@@ -542,6 +543,15 @@ static struct ctl_table kern_table[] = {
16779 + .strategy = &sysctl_string,
16780 + },
16781 + #endif
16782 ++ {
16783 ++ .ctl_name = KERN_VSHELPER,
16784 ++ .procname = "vshelper",
16785 ++ .data = &vshelper_path,
16786 ++ .maxlen = 256,
16787 ++ .mode = 0644,
16788 ++ .proc_handler = &proc_dostring,
16789 ++ .strategy = &sysctl_string,
16790 ++ },
16791 + #ifdef CONFIG_CHR_DEV_SG
16792 + {
16793 + .ctl_name = KERN_SG_BIG_BUFF,
16794 +diff -NurpP --minimal linux-2.6.29.4/kernel/sysctl_check.c linux-2.6.29.4-vs2.3.0.36.14/kernel/sysctl_check.c
16795 +--- linux-2.6.29.4/kernel/sysctl_check.c 2009-03-24 14:22:45.000000000 +0100
16796 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/sysctl_check.c 2009-03-24 14:48:36.000000000 +0100
16797 +@@ -39,6 +39,7 @@ static const struct trans_ctl_table tran
16798 +
16799 + { KERN_PANIC, "panic" },
16800 + { KERN_REALROOTDEV, "real-root-dev" },
16801 ++ { KERN_VSHELPER, "vshelper", },
16802 +
16803 + { KERN_SPARC_REBOOT, "reboot-cmd" },
16804 + { KERN_CTLALTDEL, "ctrl-alt-del" },
16805 +@@ -1216,6 +1217,22 @@ static const struct trans_ctl_table tran
16806 + {}
16807 + };
16808 +
16809 ++static struct trans_ctl_table trans_vserver_table[] = {
16810 ++ { 1, "debug_switch" },
16811 ++ { 2, "debug_xid" },
16812 ++ { 3, "debug_nid" },
16813 ++ { 4, "debug_tag" },
16814 ++ { 5, "debug_net" },
16815 ++ { 6, "debug_limit" },
16816 ++ { 7, "debug_cres" },
16817 ++ { 8, "debug_dlim" },
16818 ++ { 9, "debug_quota" },
16819 ++ { 10, "debug_cvirt" },
16820 ++ { 11, "debug_space" },
16821 ++ { 12, "debug_misc" },
16822 ++ {}
16823 ++};
16824 ++
16825 + static const struct trans_ctl_table trans_root_table[] = {
16826 + { CTL_KERN, "kernel", trans_kern_table },
16827 + { CTL_VM, "vm", trans_vm_table },
16828 +@@ -1232,6 +1249,7 @@ static const struct trans_ctl_table tran
16829 + { CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
16830 + { CTL_PM, "pm", trans_pm_table },
16831 + { CTL_FRV, "frv", trans_frv_table },
16832 ++ { CTL_VSERVER, "vserver", trans_vserver_table },
16833 + {}
16834 + };
16835 +
16836 +diff -NurpP --minimal linux-2.6.29.4/kernel/time.c linux-2.6.29.4-vs2.3.0.36.14/kernel/time.c
16837 +--- linux-2.6.29.4/kernel/time.c 2009-03-24 14:22:45.000000000 +0100
16838 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/time.c 2009-03-24 14:48:36.000000000 +0100
16839 +@@ -63,6 +63,7 @@ EXPORT_SYMBOL(sys_tz);
16840 + SYSCALL_DEFINE1(time, time_t __user *, tloc)
16841 + {
16842 + time_t i = get_seconds();
16843 ++/* FIXME: do_gettimeofday(&tv) -> vx_gettimeofday(&tv) */
16844 +
16845 + if (tloc) {
16846 + if (put_user(i,tloc))
16847 +@@ -93,7 +94,7 @@ SYSCALL_DEFINE1(stime, time_t __user *,
16848 + if (err)
16849 + return err;
16850 +
16851 +- do_settimeofday(&tv);
16852 ++ vx_settimeofday(&tv);
16853 + return 0;
16854 + }
16855 +
16856 +@@ -104,7 +105,7 @@ SYSCALL_DEFINE2(gettimeofday, struct tim
16857 + {
16858 + if (likely(tv != NULL)) {
16859 + struct timeval ktv;
16860 +- do_gettimeofday(&ktv);
16861 ++ vx_gettimeofday(&ktv);
16862 + if (copy_to_user(tv, &ktv, sizeof(ktv)))
16863 + return -EFAULT;
16864 + }
16865 +@@ -179,7 +180,7 @@ int do_sys_settimeofday(struct timespec
16866 + /* SMP safe, again the code in arch/foo/time.c should
16867 + * globally block out interrupts when it runs.
16868 + */
16869 +- return do_settimeofday(tv);
16870 ++ return vx_settimeofday(tv);
16871 + }
16872 + return 0;
16873 + }
16874 +@@ -311,7 +312,7 @@ void getnstimeofday(struct timespec *tv)
16875 + {
16876 + struct timeval x;
16877 +
16878 +- do_gettimeofday(&x);
16879 ++ vx_gettimeofday(&x);
16880 + tv->tv_sec = x.tv_sec;
16881 + tv->tv_nsec = x.tv_usec * NSEC_PER_USEC;
16882 + }
16883 +diff -NurpP --minimal linux-2.6.29.4/kernel/timer.c linux-2.6.29.4-vs2.3.0.36.14/kernel/timer.c
16884 +--- linux-2.6.29.4/kernel/timer.c 2009-03-24 14:22:45.000000000 +0100
16885 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/timer.c 2009-03-24 17:41:02.000000000 +0100
16886 +@@ -37,6 +37,10 @@
16887 + #include <linux/delay.h>
16888 + #include <linux/tick.h>
16889 + #include <linux/kallsyms.h>
16890 ++#include <linux/vs_base.h>
16891 ++#include <linux/vs_cvirt.h>
16892 ++#include <linux/vs_pid.h>
16893 ++#include <linux/vserver/sched.h>
16894 +
16895 + #include <asm/uaccess.h>
16896 + #include <asm/unistd.h>
16897 +@@ -1018,6 +1022,25 @@ unsigned long get_next_timer_interrupt(u
16898 + }
16899 + #endif
16900 +
16901 ++static inline
16902 ++void __vx_consume_token(struct _vx_sched_pc *sched_pc)
16903 ++{
16904 ++ sched_pc->tokens--;
16905 ++}
16906 ++
16907 ++static inline
16908 ++void vx_hard_tick(struct task_struct *p, int cpu)
16909 ++{
16910 ++ struct vx_info *vxi = p->vx_info;
16911 ++
16912 ++ if (vx_info_flags(vxi, VXF_SCHED_HARD|VXF_SCHED_PRIO, 0)) {
16913 ++ struct _vx_sched_pc *sched_pc =
16914 ++ &vx_per_cpu(vxi, sched_pc, cpu);
16915 ++
16916 ++ __vx_consume_token(sched_pc);
16917 ++ }
16918 ++}
16919 ++
16920 + /*
16921 + * Called from the timer interrupt handler to charge one tick to the current
16922 + * process. user_tick is 1 if the tick is user time, 0 for system.
16923 +@@ -1034,6 +1057,7 @@ void update_process_times(int user_tick)
16924 + rcu_check_callbacks(cpu, user_tick);
16925 + printk_tick();
16926 + scheduler_tick();
16927 ++ vx_hard_tick(p, cpu);
16928 + run_posix_cpu_timers(p);
16929 + }
16930 +
16931 +@@ -1136,12 +1160,6 @@ SYSCALL_DEFINE1(alarm, unsigned int, sec
16932 +
16933 + #endif
16934 +
16935 +-#ifndef __alpha__
16936 +-
16937 +-/*
16938 +- * The Alpha uses getxpid, getxuid, and getxgid instead. Maybe this
16939 +- * should be moved into arch/i386 instead?
16940 +- */
16941 +
16942 + /**
16943 + * sys_getpid - return the thread group id of the current process
16944 +@@ -1170,10 +1188,23 @@ SYSCALL_DEFINE0(getppid)
16945 + rcu_read_lock();
16946 + pid = task_tgid_vnr(current->real_parent);
16947 + rcu_read_unlock();
16948 ++ return vx_map_pid(pid);
16949 ++}
16950 +
16951 +- return pid;
16952 ++#ifdef __alpha__
16953 ++
16954 ++/*
16955 ++ * The Alpha uses getxpid, getxuid, and getxgid instead.
16956 ++ */
16957 ++
16958 ++asmlinkage long do_getxpid(long *ppid)
16959 ++{
16960 ++ *ppid = sys_getppid();
16961 ++ return sys_getpid();
16962 + }
16963 +
16964 ++#else /* _alpha_ */
16965 ++
16966 + SYSCALL_DEFINE0(getuid)
16967 + {
16968 + /* Only we change this so SMP safe */
16969 +@@ -1344,6 +1375,8 @@ int do_sysinfo(struct sysinfo *info)
16970 + tp.tv_nsec = tp.tv_nsec - NSEC_PER_SEC;
16971 + tp.tv_sec++;
16972 + }
16973 ++ if (vx_flags(VXF_VIRT_UPTIME, 0))
16974 ++ vx_vsi_uptime(&tp, NULL);
16975 + info->uptime = tp.tv_sec + (tp.tv_nsec ? 1 : 0);
16976 +
16977 + info->loads[0] = avenrun[0] << (SI_LOAD_SHIFT - FSHIFT);
16978 +diff -NurpP --minimal linux-2.6.29.4/kernel/user.c linux-2.6.29.4-vs2.3.0.36.14/kernel/user.c
16979 +--- linux-2.6.29.4/kernel/user.c 2009-03-24 14:22:45.000000000 +0100
16980 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/user.c 2009-03-24 17:47:32.000000000 +0100
16981 +@@ -249,10 +249,10 @@ static struct kobj_type uids_ktype = {
16982 + *
16983 + * See Documentation/scheduler/sched-design-CFS.txt for ramifications.
16984 + */
16985 +-static int uids_user_create(struct user_struct *up)
16986 ++static int uids_user_create(struct user_namespace *ns, struct user_struct *up)
16987 + {
16988 + struct kobject *kobj = &up->kobj;
16989 +- int error;
16990 ++ int error = 0;
16991 +
16992 + memset(kobj, 0, sizeof(struct kobject));
16993 + if (up->user_ns != &init_user_ns)
16994 +@@ -280,7 +280,7 @@ int __init uids_sysfs_init(void)
16995 + if (!uids_kset)
16996 + return -ENOMEM;
16997 +
16998 +- return uids_user_create(&root_user);
16999 ++ return uids_user_create(NULL, &root_user);
17000 + }
17001 +
17002 + /* work function to remove sysfs directory for a user and free up
17003 +@@ -342,7 +342,8 @@ static void free_user(struct user_struct
17004 + #else /* CONFIG_USER_SCHED && CONFIG_SYSFS */
17005 +
17006 + int uids_sysfs_init(void) { return 0; }
17007 +-static inline int uids_user_create(struct user_struct *up) { return 0; }
17008 ++static inline int uids_user_create(struct user_namespace *ns,
17009 ++ struct user_struct *up) { return 0; }
17010 + static inline void uids_mutex_lock(void) { }
17011 + static inline void uids_mutex_unlock(void) { }
17012 +
17013 +@@ -439,7 +440,7 @@ struct user_struct *alloc_uid(struct use
17014 +
17015 + new->user_ns = get_user_ns(ns);
17016 +
17017 +- if (uids_user_create(new))
17018 ++ if (uids_user_create(ns, new))
17019 + goto out_destoy_sched;
17020 +
17021 + /*
17022 +diff -NurpP --minimal linux-2.6.29.4/kernel/user_namespace.c linux-2.6.29.4-vs2.3.0.36.14/kernel/user_namespace.c
17023 +--- linux-2.6.29.4/kernel/user_namespace.c 2009-03-24 14:22:45.000000000 +0100
17024 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/user_namespace.c 2009-03-24 18:05:19.000000000 +0100
17025 +@@ -10,6 +10,7 @@
17026 + #include <linux/slab.h>
17027 + #include <linux/user_namespace.h>
17028 + #include <linux/cred.h>
17029 ++#include <linux/vserver/global.h>
17030 +
17031 + /*
17032 + * Create a new user namespace, deriving the creator from the user in the
17033 +@@ -30,6 +31,7 @@ int create_user_ns(struct cred *new)
17034 + return -ENOMEM;
17035 +
17036 + kref_init(&ns->kref);
17037 ++ atomic_inc(&vs_global_user_ns);
17038 +
17039 + for (n = 0; n < UIDHASH_SZ; ++n)
17040 + INIT_HLIST_HEAD(ns->uidhash_table + n);
17041 +@@ -78,6 +80,8 @@ void free_user_ns(struct kref *kref)
17042 + struct user_namespace *ns =
17043 + container_of(kref, struct user_namespace, kref);
17044 +
17045 ++ /* FIXME: maybe move into destroyer? */
17046 ++ atomic_dec(&vs_global_user_ns);
17047 + INIT_WORK(&ns->destroyer, free_user_ns_work);
17048 + schedule_work(&ns->destroyer);
17049 + }
17050 +diff -NurpP --minimal linux-2.6.29.4/kernel/utsname.c linux-2.6.29.4-vs2.3.0.36.14/kernel/utsname.c
17051 +--- linux-2.6.29.4/kernel/utsname.c 2008-12-25 00:26:37.000000000 +0100
17052 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/utsname.c 2009-02-22 22:54:26.000000000 +0100
17053 +@@ -14,6 +14,7 @@
17054 + #include <linux/utsname.h>
17055 + #include <linux/err.h>
17056 + #include <linux/slab.h>
17057 ++#include <linux/vserver/global.h>
17058 +
17059 + /*
17060 + * Clone a new ns copying an original utsname, setting refcount to 1
17061 +@@ -32,6 +33,7 @@ static struct uts_namespace *clone_uts_n
17062 + memcpy(&ns->name, &old_ns->name, sizeof(ns->name));
17063 + up_read(&uts_sem);
17064 + kref_init(&ns->kref);
17065 ++ atomic_inc(&vs_global_uts_ns);
17066 + return ns;
17067 + }
17068 +
17069 +@@ -62,5 +64,6 @@ void free_uts_ns(struct kref *kref)
17070 + struct uts_namespace *ns;
17071 +
17072 + ns = container_of(kref, struct uts_namespace, kref);
17073 ++ atomic_dec(&vs_global_uts_ns);
17074 + kfree(ns);
17075 + }
17076 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cacct.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct.c
17077 +--- linux-2.6.29.4/kernel/vserver/cacct.c 1970-01-01 01:00:00.000000000 +0100
17078 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct.c 2009-02-22 22:54:26.000000000 +0100
17079 +@@ -0,0 +1,42 @@
17080 ++/*
17081 ++ * linux/kernel/vserver/cacct.c
17082 ++ *
17083 ++ * Virtual Server: Context Accounting
17084 ++ *
17085 ++ * Copyright (C) 2006-2007 Herbert Pötzl
17086 ++ *
17087 ++ * V0.01 added accounting stats
17088 ++ *
17089 ++ */
17090 ++
17091 ++#include <linux/types.h>
17092 ++#include <linux/vs_context.h>
17093 ++#include <linux/vserver/cacct_cmd.h>
17094 ++#include <linux/vserver/cacct_int.h>
17095 ++
17096 ++#include <asm/errno.h>
17097 ++#include <asm/uaccess.h>
17098 ++
17099 ++
17100 ++int vc_sock_stat(struct vx_info *vxi, void __user *data)
17101 ++{
17102 ++ struct vcmd_sock_stat_v0 vc_data;
17103 ++ int j, field;
17104 ++
17105 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17106 ++ return -EFAULT;
17107 ++
17108 ++ field = vc_data.field;
17109 ++ if ((field < 0) || (field >= VXA_SOCK_SIZE))
17110 ++ return -EINVAL;
17111 ++
17112 ++ for (j = 0; j < 3; j++) {
17113 ++ vc_data.count[j] = vx_sock_count(&vxi->cacct, field, j);
17114 ++ vc_data.total[j] = vx_sock_total(&vxi->cacct, field, j);
17115 ++ }
17116 ++
17117 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17118 ++ return -EFAULT;
17119 ++ return 0;
17120 ++}
17121 ++
17122 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cacct_init.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct_init.h
17123 +--- linux-2.6.29.4/kernel/vserver/cacct_init.h 1970-01-01 01:00:00.000000000 +0100
17124 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct_init.h 2009-02-22 22:54:26.000000000 +0100
17125 +@@ -0,0 +1,25 @@
17126 ++
17127 ++
17128 ++static inline void vx_info_init_cacct(struct _vx_cacct *cacct)
17129 ++{
17130 ++ int i, j;
17131 ++
17132 ++
17133 ++ for (i = 0; i < VXA_SOCK_SIZE; i++) {
17134 ++ for (j = 0; j < 3; j++) {
17135 ++ atomic_set(&cacct->sock[i][j].count, 0);
17136 ++ atomic_set(&cacct->sock[i][j].total, 0);
17137 ++ }
17138 ++ }
17139 ++ for (i = 0; i < 8; i++)
17140 ++ atomic_set(&cacct->slab[i], 0);
17141 ++ for (i = 0; i < 5; i++)
17142 ++ for (j = 0; j < 4; j++)
17143 ++ atomic_set(&cacct->page[i][j], 0);
17144 ++}
17145 ++
17146 ++static inline void vx_info_exit_cacct(struct _vx_cacct *cacct)
17147 ++{
17148 ++ return;
17149 ++}
17150 ++
17151 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cacct_proc.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct_proc.h
17152 +--- linux-2.6.29.4/kernel/vserver/cacct_proc.h 1970-01-01 01:00:00.000000000 +0100
17153 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cacct_proc.h 2009-02-22 22:54:26.000000000 +0100
17154 +@@ -0,0 +1,53 @@
17155 ++#ifndef _VX_CACCT_PROC_H
17156 ++#define _VX_CACCT_PROC_H
17157 ++
17158 ++#include <linux/vserver/cacct_int.h>
17159 ++
17160 ++
17161 ++#define VX_SOCKA_TOP \
17162 ++ "Type\t recv #/bytes\t\t send #/bytes\t\t fail #/bytes\n"
17163 ++
17164 ++static inline int vx_info_proc_cacct(struct _vx_cacct *cacct, char *buffer)
17165 ++{
17166 ++ int i, j, length = 0;
17167 ++ static char *type[VXA_SOCK_SIZE] = {
17168 ++ "UNSPEC", "UNIX", "INET", "INET6", "PACKET", "OTHER"
17169 ++ };
17170 ++
17171 ++ length += sprintf(buffer + length, VX_SOCKA_TOP);
17172 ++ for (i = 0; i < VXA_SOCK_SIZE; i++) {
17173 ++ length += sprintf(buffer + length, "%s:", type[i]);
17174 ++ for (j = 0; j < 3; j++) {
17175 ++ length += sprintf(buffer + length,
17176 ++ "\t%10lu/%-10lu",
17177 ++ vx_sock_count(cacct, i, j),
17178 ++ vx_sock_total(cacct, i, j));
17179 ++ }
17180 ++ buffer[length++] = '\n';
17181 ++ }
17182 ++
17183 ++ length += sprintf(buffer + length, "\n");
17184 ++ length += sprintf(buffer + length,
17185 ++ "slab:\t %8u %8u %8u %8u\n",
17186 ++ atomic_read(&cacct->slab[1]),
17187 ++ atomic_read(&cacct->slab[4]),
17188 ++ atomic_read(&cacct->slab[0]),
17189 ++ atomic_read(&cacct->slab[2]));
17190 ++
17191 ++ length += sprintf(buffer + length, "\n");
17192 ++ for (i = 0; i < 5; i++) {
17193 ++ length += sprintf(buffer + length,
17194 ++ "page[%d]: %8u %8u %8u %8u\t %8u %8u %8u %8u\n", i,
17195 ++ atomic_read(&cacct->page[i][0]),
17196 ++ atomic_read(&cacct->page[i][1]),
17197 ++ atomic_read(&cacct->page[i][2]),
17198 ++ atomic_read(&cacct->page[i][3]),
17199 ++ atomic_read(&cacct->page[i][4]),
17200 ++ atomic_read(&cacct->page[i][5]),
17201 ++ atomic_read(&cacct->page[i][6]),
17202 ++ atomic_read(&cacct->page[i][7]));
17203 ++ }
17204 ++ return length;
17205 ++}
17206 ++
17207 ++#endif /* _VX_CACCT_PROC_H */
17208 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/context.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/context.c
17209 +--- linux-2.6.29.4/kernel/vserver/context.c 1970-01-01 01:00:00.000000000 +0100
17210 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/context.c 2009-05-23 22:56:58.000000000 +0200
17211 +@@ -0,0 +1,1030 @@
17212 ++/*
17213 ++ * linux/kernel/vserver/context.c
17214 ++ *
17215 ++ * Virtual Server: Context Support
17216 ++ *
17217 ++ * Copyright (C) 2003-2007 Herbert Pötzl
17218 ++ *
17219 ++ * V0.01 context helper
17220 ++ * V0.02 vx_ctx_kill syscall command
17221 ++ * V0.03 replaced context_info calls
17222 ++ * V0.04 redesign of struct (de)alloc
17223 ++ * V0.05 rlimit basic implementation
17224 ++ * V0.06 task_xid and info commands
17225 ++ * V0.07 context flags and caps
17226 ++ * V0.08 switch to RCU based hash
17227 ++ * V0.09 revert to non RCU for now
17228 ++ * V0.10 and back to working RCU hash
17229 ++ * V0.11 and back to locking again
17230 ++ * V0.12 referenced context store
17231 ++ * V0.13 separate per cpu data
17232 ++ * V0.14 changed vcmds to vxi arg
17233 ++ * V0.15 added context stat
17234 ++ * V0.16 have __create claim() the vxi
17235 ++ * V0.17 removed older and legacy stuff
17236 ++ *
17237 ++ */
17238 ++
17239 ++#include <linux/slab.h>
17240 ++#include <linux/types.h>
17241 ++#include <linux/security.h>
17242 ++#include <linux/pid_namespace.h>
17243 ++
17244 ++#include <linux/vserver/context.h>
17245 ++#include <linux/vserver/network.h>
17246 ++#include <linux/vserver/debug.h>
17247 ++#include <linux/vserver/limit.h>
17248 ++#include <linux/vserver/limit_int.h>
17249 ++#include <linux/vserver/space.h>
17250 ++#include <linux/init_task.h>
17251 ++
17252 ++#include <linux/vs_context.h>
17253 ++#include <linux/vs_limit.h>
17254 ++#include <linux/vs_pid.h>
17255 ++#include <linux/vserver/context_cmd.h>
17256 ++
17257 ++#include "cvirt_init.h"
17258 ++#include "cacct_init.h"
17259 ++#include "limit_init.h"
17260 ++#include "sched_init.h"
17261 ++
17262 ++
17263 ++atomic_t vx_global_ctotal = ATOMIC_INIT(0);
17264 ++atomic_t vx_global_cactive = ATOMIC_INIT(0);
17265 ++
17266 ++
17267 ++/* now inactive context structures */
17268 ++
17269 ++static struct hlist_head vx_info_inactive = HLIST_HEAD_INIT;
17270 ++
17271 ++static spinlock_t vx_info_inactive_lock = SPIN_LOCK_UNLOCKED;
17272 ++
17273 ++
17274 ++/* __alloc_vx_info()
17275 ++
17276 ++ * allocate an initialized vx_info struct
17277 ++ * doesn't make it visible (hash) */
17278 ++
17279 ++static struct vx_info *__alloc_vx_info(xid_t xid)
17280 ++{
17281 ++ struct vx_info *new = NULL;
17282 ++ int cpu, index;
17283 ++
17284 ++ vxdprintk(VXD_CBIT(xid, 0), "alloc_vx_info(%d)*", xid);
17285 ++
17286 ++ /* would this benefit from a slab cache? */
17287 ++ new = kmalloc(sizeof(struct vx_info), GFP_KERNEL);
17288 ++ if (!new)
17289 ++ return 0;
17290 ++
17291 ++ memset(new, 0, sizeof(struct vx_info));
17292 ++#ifdef CONFIG_SMP
17293 ++ new->ptr_pc = alloc_percpu(struct _vx_info_pc);
17294 ++ if (!new->ptr_pc)
17295 ++ goto error;
17296 ++#endif
17297 ++ new->vx_id = xid;
17298 ++ INIT_HLIST_NODE(&new->vx_hlist);
17299 ++ atomic_set(&new->vx_usecnt, 0);
17300 ++ atomic_set(&new->vx_tasks, 0);
17301 ++ new->vx_parent = NULL;
17302 ++ new->vx_state = 0;
17303 ++ init_waitqueue_head(&new->vx_wait);
17304 ++
17305 ++ /* prepare reaper */
17306 ++ get_task_struct(init_pid_ns.child_reaper);
17307 ++ new->vx_reaper = init_pid_ns.child_reaper;
17308 ++ new->vx_badness_bias = 0;
17309 ++
17310 ++ /* rest of init goes here */
17311 ++ vx_info_init_limit(&new->limit);
17312 ++ vx_info_init_sched(&new->sched);
17313 ++ vx_info_init_cvirt(&new->cvirt);
17314 ++ vx_info_init_cacct(&new->cacct);
17315 ++
17316 ++ /* per cpu data structures */
17317 ++ for_each_possible_cpu(cpu) {
17318 ++ vx_info_init_sched_pc(
17319 ++ &vx_per_cpu(new, sched_pc, cpu), cpu);
17320 ++ vx_info_init_cvirt_pc(
17321 ++ &vx_per_cpu(new, cvirt_pc, cpu), cpu);
17322 ++ }
17323 ++
17324 ++ new->vx_flags = VXF_INIT_SET;
17325 ++ cap_set_init_eff(new->vx_bcaps);
17326 ++ new->vx_ccaps = 0;
17327 ++ // new->vx_cap_bset = current->cap_bset;
17328 ++
17329 ++ new->reboot_cmd = 0;
17330 ++ new->exit_code = 0;
17331 ++
17332 ++ // preconfig fs entries
17333 ++ for (index = 0; index < VX_SPACES; index++) {
17334 ++ write_lock(&init_fs.lock);
17335 ++ init_fs.users++;
17336 ++ write_unlock(&init_fs.lock);
17337 ++ new->vx_fs[index] = &init_fs;
17338 ++ }
17339 ++
17340 ++ vxdprintk(VXD_CBIT(xid, 0),
17341 ++ "alloc_vx_info(%d) = %p", xid, new);
17342 ++ vxh_alloc_vx_info(new);
17343 ++ atomic_inc(&vx_global_ctotal);
17344 ++ return new;
17345 ++#ifdef CONFIG_SMP
17346 ++error:
17347 ++ kfree(new);
17348 ++ return 0;
17349 ++#endif
17350 ++}
17351 ++
17352 ++/* __dealloc_vx_info()
17353 ++
17354 ++ * final disposal of vx_info */
17355 ++
17356 ++static void __dealloc_vx_info(struct vx_info *vxi)
17357 ++{
17358 ++ struct vx_info_save vxis;
17359 ++ int cpu;
17360 ++
17361 ++ vxdprintk(VXD_CBIT(xid, 0),
17362 ++ "dealloc_vx_info(%p)", vxi);
17363 ++ vxh_dealloc_vx_info(vxi);
17364 ++
17365 ++#ifdef CONFIG_VSERVER_WARN
17366 ++ enter_vx_info(vxi, &vxis);
17367 ++ vx_info_exit_limit(&vxi->limit);
17368 ++ vx_info_exit_sched(&vxi->sched);
17369 ++ vx_info_exit_cvirt(&vxi->cvirt);
17370 ++ vx_info_exit_cacct(&vxi->cacct);
17371 ++
17372 ++ for_each_possible_cpu(cpu) {
17373 ++ vx_info_exit_sched_pc(
17374 ++ &vx_per_cpu(vxi, sched_pc, cpu), cpu);
17375 ++ vx_info_exit_cvirt_pc(
17376 ++ &vx_per_cpu(vxi, cvirt_pc, cpu), cpu);
17377 ++ }
17378 ++ leave_vx_info(&vxis);
17379 ++#endif
17380 ++
17381 ++ vxi->vx_id = -1;
17382 ++ vxi->vx_state |= VXS_RELEASED;
17383 ++
17384 ++#ifdef CONFIG_SMP
17385 ++ free_percpu(vxi->ptr_pc);
17386 ++#endif
17387 ++ kfree(vxi);
17388 ++ atomic_dec(&vx_global_ctotal);
17389 ++}
17390 ++
17391 ++static void __shutdown_vx_info(struct vx_info *vxi)
17392 ++{
17393 ++ struct nsproxy *nsproxy;
17394 ++ struct fs_struct *fs;
17395 ++ int index, kill;
17396 ++
17397 ++ might_sleep();
17398 ++
17399 ++ vxi->vx_state |= VXS_SHUTDOWN;
17400 ++ vs_state_change(vxi, VSC_SHUTDOWN);
17401 ++
17402 ++ for (index = 0; index < VX_SPACES; index++) {
17403 ++ nsproxy = xchg(&vxi->vx_nsproxy[index], NULL);
17404 ++ if (nsproxy)
17405 ++ put_nsproxy(nsproxy);
17406 ++
17407 ++ fs = xchg(&vxi->vx_fs[index], NULL);
17408 ++ write_lock(&fs->lock);
17409 ++ kill = !--fs->users;
17410 ++ write_unlock(&fs->lock);
17411 ++ if (kill)
17412 ++ free_fs_struct(fs);
17413 ++ }
17414 ++}
17415 ++
17416 ++/* exported stuff */
17417 ++
17418 ++void free_vx_info(struct vx_info *vxi)
17419 ++{
17420 ++ unsigned long flags;
17421 ++ unsigned index;
17422 ++
17423 ++ /* check for reference counts first */
17424 ++ BUG_ON(atomic_read(&vxi->vx_usecnt));
17425 ++ BUG_ON(atomic_read(&vxi->vx_tasks));
17426 ++
17427 ++ /* context must not be hashed */
17428 ++ BUG_ON(vx_info_state(vxi, VXS_HASHED));
17429 ++
17430 ++ /* context shutdown is mandatory */
17431 ++ BUG_ON(!vx_info_state(vxi, VXS_SHUTDOWN));
17432 ++
17433 ++ /* nsproxy and fs check */
17434 ++ for (index = 0; index < VX_SPACES; index++) {
17435 ++ BUG_ON(vxi->vx_nsproxy[index]);
17436 ++ BUG_ON(vxi->vx_fs[index]);
17437 ++ }
17438 ++
17439 ++ spin_lock_irqsave(&vx_info_inactive_lock, flags);
17440 ++ hlist_del(&vxi->vx_hlist);
17441 ++ spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17442 ++
17443 ++ __dealloc_vx_info(vxi);
17444 ++}
17445 ++
17446 ++
17447 ++/* hash table for vx_info hash */
17448 ++
17449 ++#define VX_HASH_SIZE 13
17450 ++
17451 ++static struct hlist_head vx_info_hash[VX_HASH_SIZE] =
17452 ++ { [0 ... VX_HASH_SIZE-1] = HLIST_HEAD_INIT };
17453 ++
17454 ++static spinlock_t vx_info_hash_lock = SPIN_LOCK_UNLOCKED;
17455 ++
17456 ++
17457 ++static inline unsigned int __hashval(xid_t xid)
17458 ++{
17459 ++ return (xid % VX_HASH_SIZE);
17460 ++}
17461 ++
17462 ++
17463 ++
17464 ++/* __hash_vx_info()
17465 ++
17466 ++ * add the vxi to the global hash table
17467 ++ * requires the hash_lock to be held */
17468 ++
17469 ++static inline void __hash_vx_info(struct vx_info *vxi)
17470 ++{
17471 ++ struct hlist_head *head;
17472 ++
17473 ++ vxd_assert_lock(&vx_info_hash_lock);
17474 ++ vxdprintk(VXD_CBIT(xid, 4),
17475 ++ "__hash_vx_info: %p[#%d]", vxi, vxi->vx_id);
17476 ++ vxh_hash_vx_info(vxi);
17477 ++
17478 ++ /* context must not be hashed */
17479 ++ BUG_ON(vx_info_state(vxi, VXS_HASHED));
17480 ++
17481 ++ vxi->vx_state |= VXS_HASHED;
17482 ++ head = &vx_info_hash[__hashval(vxi->vx_id)];
17483 ++ hlist_add_head(&vxi->vx_hlist, head);
17484 ++ atomic_inc(&vx_global_cactive);
17485 ++}
17486 ++
17487 ++/* __unhash_vx_info()
17488 ++
17489 ++ * remove the vxi from the global hash table
17490 ++ * requires the hash_lock to be held */
17491 ++
17492 ++static inline void __unhash_vx_info(struct vx_info *vxi)
17493 ++{
17494 ++ unsigned long flags;
17495 ++
17496 ++ vxd_assert_lock(&vx_info_hash_lock);
17497 ++ vxdprintk(VXD_CBIT(xid, 4),
17498 ++ "__unhash_vx_info: %p[#%d.%d.%d]", vxi, vxi->vx_id,
17499 ++ atomic_read(&vxi->vx_usecnt), atomic_read(&vxi->vx_tasks));
17500 ++ vxh_unhash_vx_info(vxi);
17501 ++
17502 ++ /* context must be hashed */
17503 ++ BUG_ON(!vx_info_state(vxi, VXS_HASHED));
17504 ++ /* but without tasks */
17505 ++ BUG_ON(atomic_read(&vxi->vx_tasks));
17506 ++
17507 ++ vxi->vx_state &= ~VXS_HASHED;
17508 ++ hlist_del_init(&vxi->vx_hlist);
17509 ++ spin_lock_irqsave(&vx_info_inactive_lock, flags);
17510 ++ hlist_add_head(&vxi->vx_hlist, &vx_info_inactive);
17511 ++ spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
17512 ++ atomic_dec(&vx_global_cactive);
17513 ++}
17514 ++
17515 ++
17516 ++/* __lookup_vx_info()
17517 ++
17518 ++ * requires the hash_lock to be held
17519 ++ * doesn't increment the vx_refcnt */
17520 ++
17521 ++static inline struct vx_info *__lookup_vx_info(xid_t xid)
17522 ++{
17523 ++ struct hlist_head *head = &vx_info_hash[__hashval(xid)];
17524 ++ struct hlist_node *pos;
17525 ++ struct vx_info *vxi;
17526 ++
17527 ++ vxd_assert_lock(&vx_info_hash_lock);
17528 ++ hlist_for_each(pos, head) {
17529 ++ vxi = hlist_entry(pos, struct vx_info, vx_hlist);
17530 ++
17531 ++ if (vxi->vx_id == xid)
17532 ++ goto found;
17533 ++ }
17534 ++ vxi = NULL;
17535 ++found:
17536 ++ vxdprintk(VXD_CBIT(xid, 0),
17537 ++ "__lookup_vx_info(#%u): %p[#%u]",
17538 ++ xid, vxi, vxi ? vxi->vx_id : 0);
17539 ++ vxh_lookup_vx_info(vxi, xid);
17540 ++ return vxi;
17541 ++}
17542 ++
17543 ++
17544 ++/* __create_vx_info()
17545 ++
17546 ++ * create the requested context
17547 ++ * get(), claim() and hash it */
17548 ++
17549 ++static struct vx_info *__create_vx_info(int id)
17550 ++{
17551 ++ struct vx_info *new, *vxi = NULL;
17552 ++
17553 ++ vxdprintk(VXD_CBIT(xid, 1), "create_vx_info(%d)*", id);
17554 ++
17555 ++ if (!(new = __alloc_vx_info(id)))
17556 ++ return ERR_PTR(-ENOMEM);
17557 ++
17558 ++ /* required to make dynamic xids unique */
17559 ++ spin_lock(&vx_info_hash_lock);
17560 ++
17561 ++ /* static context requested */
17562 ++ if ((vxi = __lookup_vx_info(id))) {
17563 ++ vxdprintk(VXD_CBIT(xid, 0),
17564 ++ "create_vx_info(%d) = %p (already there)", id, vxi);
17565 ++ if (vx_info_flags(vxi, VXF_STATE_SETUP, 0))
17566 ++ vxi = ERR_PTR(-EBUSY);
17567 ++ else
17568 ++ vxi = ERR_PTR(-EEXIST);
17569 ++ goto out_unlock;
17570 ++ }
17571 ++ /* new context */
17572 ++ vxdprintk(VXD_CBIT(xid, 0),
17573 ++ "create_vx_info(%d) = %p (new)", id, new);
17574 ++ claim_vx_info(new, NULL);
17575 ++ __hash_vx_info(get_vx_info(new));
17576 ++ vxi = new, new = NULL;
17577 ++
17578 ++out_unlock:
17579 ++ spin_unlock(&vx_info_hash_lock);
17580 ++ vxh_create_vx_info(IS_ERR(vxi) ? NULL : vxi, id);
17581 ++ if (new)
17582 ++ __dealloc_vx_info(new);
17583 ++ return vxi;
17584 ++}
17585 ++
17586 ++
17587 ++/* exported stuff */
17588 ++
17589 ++
17590 ++void unhash_vx_info(struct vx_info *vxi)
17591 ++{
17592 ++ __shutdown_vx_info(vxi);
17593 ++ spin_lock(&vx_info_hash_lock);
17594 ++ __unhash_vx_info(vxi);
17595 ++ spin_unlock(&vx_info_hash_lock);
17596 ++ __wakeup_vx_info(vxi);
17597 ++}
17598 ++
17599 ++
17600 ++/* lookup_vx_info()
17601 ++
17602 ++ * search for a vx_info and get() it
17603 ++ * negative id means current */
17604 ++
17605 ++struct vx_info *lookup_vx_info(int id)
17606 ++{
17607 ++ struct vx_info *vxi = NULL;
17608 ++
17609 ++ if (id < 0) {
17610 ++ vxi = get_vx_info(current->vx_info);
17611 ++ } else if (id > 1) {
17612 ++ spin_lock(&vx_info_hash_lock);
17613 ++ vxi = get_vx_info(__lookup_vx_info(id));
17614 ++ spin_unlock(&vx_info_hash_lock);
17615 ++ }
17616 ++ return vxi;
17617 ++}
17618 ++
17619 ++/* xid_is_hashed()
17620 ++
17621 ++ * verify that xid is still hashed */
17622 ++
17623 ++int xid_is_hashed(xid_t xid)
17624 ++{
17625 ++ int hashed;
17626 ++
17627 ++ spin_lock(&vx_info_hash_lock);
17628 ++ hashed = (__lookup_vx_info(xid) != NULL);
17629 ++ spin_unlock(&vx_info_hash_lock);
17630 ++ return hashed;
17631 ++}
17632 ++
17633 ++#ifdef CONFIG_PROC_FS
17634 ++
17635 ++/* get_xid_list()
17636 ++
17637 ++ * get a subset of hashed xids for proc
17638 ++ * assumes size is at least one */
17639 ++
17640 ++int get_xid_list(int index, unsigned int *xids, int size)
17641 ++{
17642 ++ int hindex, nr_xids = 0;
17643 ++
17644 ++ /* only show current and children */
17645 ++ if (!vx_check(0, VS_ADMIN | VS_WATCH)) {
17646 ++ if (index > 0)
17647 ++ return 0;
17648 ++ xids[nr_xids] = vx_current_xid();
17649 ++ return 1;
17650 ++ }
17651 ++
17652 ++ for (hindex = 0; hindex < VX_HASH_SIZE; hindex++) {
17653 ++ struct hlist_head *head = &vx_info_hash[hindex];
17654 ++ struct hlist_node *pos;
17655 ++
17656 ++ spin_lock(&vx_info_hash_lock);
17657 ++ hlist_for_each(pos, head) {
17658 ++ struct vx_info *vxi;
17659 ++
17660 ++ if (--index > 0)
17661 ++ continue;
17662 ++
17663 ++ vxi = hlist_entry(pos, struct vx_info, vx_hlist);
17664 ++ xids[nr_xids] = vxi->vx_id;
17665 ++ if (++nr_xids >= size) {
17666 ++ spin_unlock(&vx_info_hash_lock);
17667 ++ goto out;
17668 ++ }
17669 ++ }
17670 ++ /* keep the lock time short */
17671 ++ spin_unlock(&vx_info_hash_lock);
17672 ++ }
17673 ++out:
17674 ++ return nr_xids;
17675 ++}
17676 ++#endif
17677 ++
17678 ++#ifdef CONFIG_VSERVER_DEBUG
17679 ++
17680 ++void dump_vx_info_inactive(int level)
17681 ++{
17682 ++ struct hlist_node *entry, *next;
17683 ++
17684 ++ hlist_for_each_safe(entry, next, &vx_info_inactive) {
17685 ++ struct vx_info *vxi =
17686 ++ list_entry(entry, struct vx_info, vx_hlist);
17687 ++
17688 ++ dump_vx_info(vxi, level);
17689 ++ }
17690 ++}
17691 ++
17692 ++#endif
17693 ++
17694 ++#if 0
17695 ++int vx_migrate_user(struct task_struct *p, struct vx_info *vxi)
17696 ++{
17697 ++ struct user_struct *new_user, *old_user;
17698 ++
17699 ++ if (!p || !vxi)
17700 ++ BUG();
17701 ++
17702 ++ if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
17703 ++ return -EACCES;
17704 ++
17705 ++ new_user = alloc_uid(vxi->vx_id, p->uid);
17706 ++ if (!new_user)
17707 ++ return -ENOMEM;
17708 ++
17709 ++ old_user = p->user;
17710 ++ if (new_user != old_user) {
17711 ++ atomic_inc(&new_user->processes);
17712 ++ atomic_dec(&old_user->processes);
17713 ++ p->user = new_user;
17714 ++ }
17715 ++ free_uid(old_user);
17716 ++ return 0;
17717 ++}
17718 ++#endif
17719 ++
17720 ++#if 0
17721 ++void vx_mask_cap_bset(struct vx_info *vxi, struct task_struct *p)
17722 ++{
17723 ++ // p->cap_effective &= vxi->vx_cap_bset;
17724 ++ p->cap_effective =
17725 ++ cap_intersect(p->cap_effective, vxi->cap_bset);
17726 ++ // p->cap_inheritable &= vxi->vx_cap_bset;
17727 ++ p->cap_inheritable =
17728 ++ cap_intersect(p->cap_inheritable, vxi->cap_bset);
17729 ++ // p->cap_permitted &= vxi->vx_cap_bset;
17730 ++ p->cap_permitted =
17731 ++ cap_intersect(p->cap_permitted, vxi->cap_bset);
17732 ++}
17733 ++#endif
17734 ++
17735 ++
17736 ++#include <linux/file.h>
17737 ++#include <linux/fdtable.h>
17738 ++
17739 ++static int vx_openfd_task(struct task_struct *tsk)
17740 ++{
17741 ++ struct files_struct *files = tsk->files;
17742 ++ struct fdtable *fdt;
17743 ++ const unsigned long *bptr;
17744 ++ int count, total;
17745 ++
17746 ++ /* no rcu_read_lock() because of spin_lock() */
17747 ++ spin_lock(&files->file_lock);
17748 ++ fdt = files_fdtable(files);
17749 ++ bptr = fdt->open_fds->fds_bits;
17750 ++ count = fdt->max_fds / (sizeof(unsigned long) * 8);
17751 ++ for (total = 0; count > 0; count--) {
17752 ++ if (*bptr)
17753 ++ total += hweight_long(*bptr);
17754 ++ bptr++;
17755 ++ }
17756 ++ spin_unlock(&files->file_lock);
17757 ++ return total;
17758 ++}
17759 ++
17760 ++
17761 ++/* for *space compatibility */
17762 ++
17763 ++asmlinkage long sys_unshare(unsigned long);
17764 ++
17765 ++/*
17766 ++ * migrate task to new context
17767 ++ * gets vxi, puts old_vxi on change
17768 ++ * optionally unshares namespaces (hack)
17769 ++ */
17770 ++
17771 ++int vx_migrate_task(struct task_struct *p, struct vx_info *vxi, int unshare)
17772 ++{
17773 ++ struct vx_info *old_vxi;
17774 ++ int ret = 0;
17775 ++
17776 ++ if (!p || !vxi)
17777 ++ BUG();
17778 ++
17779 ++ vxdprintk(VXD_CBIT(xid, 5),
17780 ++ "vx_migrate_task(%p,%p[#%d.%d])", p, vxi,
17781 ++ vxi->vx_id, atomic_read(&vxi->vx_usecnt));
17782 ++
17783 ++ if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0) &&
17784 ++ !vx_info_flags(vxi, VXF_STATE_SETUP, 0))
17785 ++ return -EACCES;
17786 ++
17787 ++ if (vx_info_state(vxi, VXS_SHUTDOWN))
17788 ++ return -EFAULT;
17789 ++
17790 ++ old_vxi = task_get_vx_info(p);
17791 ++ if (old_vxi == vxi)
17792 ++ goto out;
17793 ++
17794 ++// if (!(ret = vx_migrate_user(p, vxi))) {
17795 ++ {
17796 ++ int openfd;
17797 ++
17798 ++ task_lock(p);
17799 ++ openfd = vx_openfd_task(p);
17800 ++
17801 ++ if (old_vxi) {
17802 ++ atomic_dec(&old_vxi->cvirt.nr_threads);
17803 ++ atomic_dec(&old_vxi->cvirt.nr_running);
17804 ++ __rlim_dec(&old_vxi->limit, RLIMIT_NPROC);
17805 ++ /* FIXME: what about the struct files here? */
17806 ++ __rlim_sub(&old_vxi->limit, VLIMIT_OPENFD, openfd);
17807 ++ /* account for the executable */
17808 ++ __rlim_dec(&old_vxi->limit, VLIMIT_DENTRY);
17809 ++ }
17810 ++ atomic_inc(&vxi->cvirt.nr_threads);
17811 ++ atomic_inc(&vxi->cvirt.nr_running);
17812 ++ __rlim_inc(&vxi->limit, RLIMIT_NPROC);
17813 ++ /* FIXME: what about the struct files here? */
17814 ++ __rlim_add(&vxi->limit, VLIMIT_OPENFD, openfd);
17815 ++ /* account for the executable */
17816 ++ __rlim_inc(&vxi->limit, VLIMIT_DENTRY);
17817 ++
17818 ++ if (old_vxi) {
17819 ++ release_vx_info(old_vxi, p);
17820 ++ clr_vx_info(&p->vx_info);
17821 ++ }
17822 ++ claim_vx_info(vxi, p);
17823 ++ set_vx_info(&p->vx_info, vxi);
17824 ++ p->xid = vxi->vx_id;
17825 ++
17826 ++ vxdprintk(VXD_CBIT(xid, 5),
17827 ++ "moved task %p into vxi:%p[#%d]",
17828 ++ p, vxi, vxi->vx_id);
17829 ++
17830 ++ // vx_mask_cap_bset(vxi, p);
17831 ++ task_unlock(p);
17832 ++
17833 ++ /* hack for *spaces to provide compatibility */
17834 ++ if (unshare) {
17835 ++ struct nsproxy *old_nsp, *new_nsp;
17836 ++
17837 ++ ret = unshare_nsproxy_namespaces(
17838 ++ CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER,
17839 ++ &new_nsp, NULL);
17840 ++ if (ret)
17841 ++ goto out;
17842 ++
17843 ++ old_nsp = xchg(&p->nsproxy, new_nsp);
17844 ++ vx_set_space(vxi,
17845 ++ CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER, 0);
17846 ++ put_nsproxy(old_nsp);
17847 ++ }
17848 ++ }
17849 ++out:
17850 ++ put_vx_info(old_vxi);
17851 ++ return ret;
17852 ++}
17853 ++
17854 ++int vx_set_reaper(struct vx_info *vxi, struct task_struct *p)
17855 ++{
17856 ++ struct task_struct *old_reaper;
17857 ++
17858 ++ if (!vxi)
17859 ++ return -EINVAL;
17860 ++
17861 ++ vxdprintk(VXD_CBIT(xid, 6),
17862 ++ "vx_set_reaper(%p[#%d],%p[#%d,%d])",
17863 ++ vxi, vxi->vx_id, p, p->xid, p->pid);
17864 ++
17865 ++ old_reaper = vxi->vx_reaper;
17866 ++ if (old_reaper == p)
17867 ++ return 0;
17868 ++
17869 ++ /* set new child reaper */
17870 ++ get_task_struct(p);
17871 ++ vxi->vx_reaper = p;
17872 ++ put_task_struct(old_reaper);
17873 ++ return 0;
17874 ++}
17875 ++
17876 ++int vx_set_init(struct vx_info *vxi, struct task_struct *p)
17877 ++{
17878 ++ if (!vxi)
17879 ++ return -EINVAL;
17880 ++
17881 ++ vxdprintk(VXD_CBIT(xid, 6),
17882 ++ "vx_set_init(%p[#%d],%p[#%d,%d,%d])",
17883 ++ vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
17884 ++
17885 ++ vxi->vx_flags &= ~VXF_STATE_INIT;
17886 ++ vxi->vx_initpid = p->tgid;
17887 ++ return 0;
17888 ++}
17889 ++
17890 ++void vx_exit_init(struct vx_info *vxi, struct task_struct *p, int code)
17891 ++{
17892 ++ vxdprintk(VXD_CBIT(xid, 6),
17893 ++ "vx_exit_init(%p[#%d],%p[#%d,%d,%d])",
17894 ++ vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
17895 ++
17896 ++ vxi->exit_code = code;
17897 ++ vxi->vx_initpid = 0;
17898 ++}
17899 ++
17900 ++
17901 ++void vx_set_persistent(struct vx_info *vxi)
17902 ++{
17903 ++ vxdprintk(VXD_CBIT(xid, 6),
17904 ++ "vx_set_persistent(%p[#%d])", vxi, vxi->vx_id);
17905 ++
17906 ++ get_vx_info(vxi);
17907 ++ claim_vx_info(vxi, NULL);
17908 ++}
17909 ++
17910 ++void vx_clear_persistent(struct vx_info *vxi)
17911 ++{
17912 ++ vxdprintk(VXD_CBIT(xid, 6),
17913 ++ "vx_clear_persistent(%p[#%d])", vxi, vxi->vx_id);
17914 ++
17915 ++ release_vx_info(vxi, NULL);
17916 ++ put_vx_info(vxi);
17917 ++}
17918 ++
17919 ++void vx_update_persistent(struct vx_info *vxi)
17920 ++{
17921 ++ if (vx_info_flags(vxi, VXF_PERSISTENT, 0))
17922 ++ vx_set_persistent(vxi);
17923 ++ else
17924 ++ vx_clear_persistent(vxi);
17925 ++}
17926 ++
17927 ++
17928 ++/* task must be current or locked */
17929 ++
17930 ++void exit_vx_info(struct task_struct *p, int code)
17931 ++{
17932 ++ struct vx_info *vxi = p->vx_info;
17933 ++
17934 ++ if (vxi) {
17935 ++ atomic_dec(&vxi->cvirt.nr_threads);
17936 ++ vx_nproc_dec(p);
17937 ++
17938 ++ vxi->exit_code = code;
17939 ++ release_vx_info(vxi, p);
17940 ++ }
17941 ++}
17942 ++
17943 ++void exit_vx_info_early(struct task_struct *p, int code)
17944 ++{
17945 ++ struct vx_info *vxi = p->vx_info;
17946 ++
17947 ++ if (vxi) {
17948 ++ if (vxi->vx_initpid == p->tgid)
17949 ++ vx_exit_init(vxi, p, code);
17950 ++ if (vxi->vx_reaper == p)
17951 ++ vx_set_reaper(vxi, init_pid_ns.child_reaper);
17952 ++ }
17953 ++}
17954 ++
17955 ++
17956 ++/* vserver syscall commands below here */
17957 ++
17958 ++/* taks xid and vx_info functions */
17959 ++
17960 ++#include <asm/uaccess.h>
17961 ++
17962 ++
17963 ++int vc_task_xid(uint32_t id)
17964 ++{
17965 ++ xid_t xid;
17966 ++
17967 ++ if (id) {
17968 ++ struct task_struct *tsk;
17969 ++
17970 ++ read_lock(&tasklist_lock);
17971 ++ tsk = find_task_by_real_pid(id);
17972 ++ xid = (tsk) ? tsk->xid : -ESRCH;
17973 ++ read_unlock(&tasklist_lock);
17974 ++ } else
17975 ++ xid = vx_current_xid();
17976 ++ return xid;
17977 ++}
17978 ++
17979 ++
17980 ++int vc_vx_info(struct vx_info *vxi, void __user *data)
17981 ++{
17982 ++ struct vcmd_vx_info_v0 vc_data;
17983 ++
17984 ++ vc_data.xid = vxi->vx_id;
17985 ++ vc_data.initpid = vxi->vx_initpid;
17986 ++
17987 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17988 ++ return -EFAULT;
17989 ++ return 0;
17990 ++}
17991 ++
17992 ++
17993 ++int vc_ctx_stat(struct vx_info *vxi, void __user *data)
17994 ++{
17995 ++ struct vcmd_ctx_stat_v0 vc_data;
17996 ++
17997 ++ vc_data.usecnt = atomic_read(&vxi->vx_usecnt);
17998 ++ vc_data.tasks = atomic_read(&vxi->vx_tasks);
17999 ++
18000 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18001 ++ return -EFAULT;
18002 ++ return 0;
18003 ++}
18004 ++
18005 ++
18006 ++/* context functions */
18007 ++
18008 ++int vc_ctx_create(uint32_t xid, void __user *data)
18009 ++{
18010 ++ struct vcmd_ctx_create vc_data = { .flagword = VXF_INIT_SET };
18011 ++ struct vx_info *new_vxi;
18012 ++ int ret;
18013 ++
18014 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
18015 ++ return -EFAULT;
18016 ++
18017 ++ if ((xid > MAX_S_CONTEXT) || (xid < 2))
18018 ++ return -EINVAL;
18019 ++
18020 ++ new_vxi = __create_vx_info(xid);
18021 ++ if (IS_ERR(new_vxi))
18022 ++ return PTR_ERR(new_vxi);
18023 ++
18024 ++ /* initial flags */
18025 ++ new_vxi->vx_flags = vc_data.flagword;
18026 ++
18027 ++ ret = -ENOEXEC;
18028 ++ if (vs_state_change(new_vxi, VSC_STARTUP))
18029 ++ goto out;
18030 ++
18031 ++ ret = vx_migrate_task(current, new_vxi, (!data));
18032 ++ if (ret)
18033 ++ goto out;
18034 ++
18035 ++ /* return context id on success */
18036 ++ ret = new_vxi->vx_id;
18037 ++
18038 ++ /* get a reference for persistent contexts */
18039 ++ if ((vc_data.flagword & VXF_PERSISTENT))
18040 ++ vx_set_persistent(new_vxi);
18041 ++out:
18042 ++ release_vx_info(new_vxi, NULL);
18043 ++ put_vx_info(new_vxi);
18044 ++ return ret;
18045 ++}
18046 ++
18047 ++
18048 ++int vc_ctx_migrate(struct vx_info *vxi, void __user *data)
18049 ++{
18050 ++ struct vcmd_ctx_migrate vc_data = { .flagword = 0 };
18051 ++ int ret;
18052 ++
18053 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
18054 ++ return -EFAULT;
18055 ++
18056 ++ ret = vx_migrate_task(current, vxi, 0);
18057 ++ if (ret)
18058 ++ return ret;
18059 ++ if (vc_data.flagword & VXM_SET_INIT)
18060 ++ ret = vx_set_init(vxi, current);
18061 ++ if (ret)
18062 ++ return ret;
18063 ++ if (vc_data.flagword & VXM_SET_REAPER)
18064 ++ ret = vx_set_reaper(vxi, current);
18065 ++ return ret;
18066 ++}
18067 ++
18068 ++
18069 ++int vc_get_cflags(struct vx_info *vxi, void __user *data)
18070 ++{
18071 ++ struct vcmd_ctx_flags_v0 vc_data;
18072 ++
18073 ++ vc_data.flagword = vxi->vx_flags;
18074 ++
18075 ++ /* special STATE flag handling */
18076 ++ vc_data.mask = vs_mask_flags(~0ULL, vxi->vx_flags, VXF_ONE_TIME);
18077 ++
18078 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18079 ++ return -EFAULT;
18080 ++ return 0;
18081 ++}
18082 ++
18083 ++int vc_set_cflags(struct vx_info *vxi, void __user *data)
18084 ++{
18085 ++ struct vcmd_ctx_flags_v0 vc_data;
18086 ++ uint64_t mask, trigger;
18087 ++
18088 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18089 ++ return -EFAULT;
18090 ++
18091 ++ /* special STATE flag handling */
18092 ++ mask = vs_mask_mask(vc_data.mask, vxi->vx_flags, VXF_ONE_TIME);
18093 ++ trigger = (mask & vxi->vx_flags) ^ (mask & vc_data.flagword);
18094 ++
18095 ++ if (vxi == current->vx_info) {
18096 ++ /* if (trigger & VXF_STATE_SETUP)
18097 ++ vx_mask_cap_bset(vxi, current); */
18098 ++ if (trigger & VXF_STATE_INIT) {
18099 ++ int ret;
18100 ++
18101 ++ ret = vx_set_init(vxi, current);
18102 ++ if (ret)
18103 ++ return ret;
18104 ++ ret = vx_set_reaper(vxi, current);
18105 ++ if (ret)
18106 ++ return ret;
18107 ++ }
18108 ++ }
18109 ++
18110 ++ vxi->vx_flags = vs_mask_flags(vxi->vx_flags,
18111 ++ vc_data.flagword, mask);
18112 ++ if (trigger & VXF_PERSISTENT)
18113 ++ vx_update_persistent(vxi);
18114 ++
18115 ++ return 0;
18116 ++}
18117 ++
18118 ++
18119 ++static inline uint64_t caps_from_cap_t(kernel_cap_t c)
18120 ++{
18121 ++ uint64_t v = c.cap[0] | ((uint64_t)c.cap[1] << 32);
18122 ++
18123 ++ // printk("caps_from_cap_t(%08x:%08x) = %016llx\n", c.cap[1], c.cap[0], v);
18124 ++ return v;
18125 ++}
18126 ++
18127 ++static inline kernel_cap_t cap_t_from_caps(uint64_t v)
18128 ++{
18129 ++ kernel_cap_t c = __cap_empty_set;
18130 ++
18131 ++ c.cap[0] = v & 0xFFFFFFFF;
18132 ++ c.cap[1] = (v >> 32) & 0xFFFFFFFF;
18133 ++
18134 ++ // printk("cap_t_from_caps(%016llx) = %08x:%08x\n", v, c.cap[1], c.cap[0]);
18135 ++ return c;
18136 ++}
18137 ++
18138 ++
18139 ++static int do_get_caps(struct vx_info *vxi, uint64_t *bcaps, uint64_t *ccaps)
18140 ++{
18141 ++ if (bcaps)
18142 ++ *bcaps = caps_from_cap_t(vxi->vx_bcaps);
18143 ++ if (ccaps)
18144 ++ *ccaps = vxi->vx_ccaps;
18145 ++
18146 ++ return 0;
18147 ++}
18148 ++
18149 ++int vc_get_ccaps(struct vx_info *vxi, void __user *data)
18150 ++{
18151 ++ struct vcmd_ctx_caps_v1 vc_data;
18152 ++ int ret;
18153 ++
18154 ++ ret = do_get_caps(vxi, NULL, &vc_data.ccaps);
18155 ++ if (ret)
18156 ++ return ret;
18157 ++ vc_data.cmask = ~0ULL;
18158 ++
18159 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18160 ++ return -EFAULT;
18161 ++ return 0;
18162 ++}
18163 ++
18164 ++static int do_set_caps(struct vx_info *vxi,
18165 ++ uint64_t bcaps, uint64_t bmask, uint64_t ccaps, uint64_t cmask)
18166 ++{
18167 ++ uint64_t bcold = caps_from_cap_t(vxi->vx_bcaps);
18168 ++
18169 ++#if 0
18170 ++ printk("do_set_caps(%16llx, %16llx, %16llx, %16llx)\n",
18171 ++ bcaps, bmask, ccaps, cmask);
18172 ++#endif
18173 ++ vxi->vx_bcaps = cap_t_from_caps(
18174 ++ vs_mask_flags(bcold, bcaps, bmask));
18175 ++ vxi->vx_ccaps = vs_mask_flags(vxi->vx_ccaps, ccaps, cmask);
18176 ++
18177 ++ return 0;
18178 ++}
18179 ++
18180 ++int vc_set_ccaps(struct vx_info *vxi, void __user *data)
18181 ++{
18182 ++ struct vcmd_ctx_caps_v1 vc_data;
18183 ++
18184 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18185 ++ return -EFAULT;
18186 ++
18187 ++ return do_set_caps(vxi, 0, 0, vc_data.ccaps, vc_data.cmask);
18188 ++}
18189 ++
18190 ++int vc_get_bcaps(struct vx_info *vxi, void __user *data)
18191 ++{
18192 ++ struct vcmd_bcaps vc_data;
18193 ++ int ret;
18194 ++
18195 ++ ret = do_get_caps(vxi, &vc_data.bcaps, NULL);
18196 ++ if (ret)
18197 ++ return ret;
18198 ++ vc_data.bmask = ~0ULL;
18199 ++
18200 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18201 ++ return -EFAULT;
18202 ++ return 0;
18203 ++}
18204 ++
18205 ++int vc_set_bcaps(struct vx_info *vxi, void __user *data)
18206 ++{
18207 ++ struct vcmd_bcaps vc_data;
18208 ++
18209 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18210 ++ return -EFAULT;
18211 ++
18212 ++ return do_set_caps(vxi, vc_data.bcaps, vc_data.bmask, 0, 0);
18213 ++}
18214 ++
18215 ++
18216 ++int vc_get_badness(struct vx_info *vxi, void __user *data)
18217 ++{
18218 ++ struct vcmd_badness_v0 vc_data;
18219 ++
18220 ++ vc_data.bias = vxi->vx_badness_bias;
18221 ++
18222 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18223 ++ return -EFAULT;
18224 ++ return 0;
18225 ++}
18226 ++
18227 ++int vc_set_badness(struct vx_info *vxi, void __user *data)
18228 ++{
18229 ++ struct vcmd_badness_v0 vc_data;
18230 ++
18231 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18232 ++ return -EFAULT;
18233 ++
18234 ++ vxi->vx_badness_bias = vc_data.bias;
18235 ++ return 0;
18236 ++}
18237 ++
18238 ++#include <linux/module.h>
18239 ++
18240 ++EXPORT_SYMBOL_GPL(free_vx_info);
18241 ++
18242 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cvirt.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt.c
18243 +--- linux-2.6.29.4/kernel/vserver/cvirt.c 1970-01-01 01:00:00.000000000 +0100
18244 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt.c 2009-02-22 22:54:26.000000000 +0100
18245 +@@ -0,0 +1,300 @@
18246 ++/*
18247 ++ * linux/kernel/vserver/cvirt.c
18248 ++ *
18249 ++ * Virtual Server: Context Virtualization
18250 ++ *
18251 ++ * Copyright (C) 2004-2007 Herbert Pötzl
18252 ++ *
18253 ++ * V0.01 broken out from limit.c
18254 ++ * V0.02 added utsname stuff
18255 ++ * V0.03 changed vcmds to vxi arg
18256 ++ *
18257 ++ */
18258 ++
18259 ++#include <linux/types.h>
18260 ++#include <linux/utsname.h>
18261 ++#include <linux/vs_cvirt.h>
18262 ++#include <linux/vserver/switch.h>
18263 ++#include <linux/vserver/cvirt_cmd.h>
18264 ++
18265 ++#include <asm/uaccess.h>
18266 ++
18267 ++
18268 ++void vx_vsi_uptime(struct timespec *uptime, struct timespec *idle)
18269 ++{
18270 ++ struct vx_info *vxi = current->vx_info;
18271 ++
18272 ++ set_normalized_timespec(uptime,
18273 ++ uptime->tv_sec - vxi->cvirt.bias_uptime.tv_sec,
18274 ++ uptime->tv_nsec - vxi->cvirt.bias_uptime.tv_nsec);
18275 ++ if (!idle)
18276 ++ return;
18277 ++ set_normalized_timespec(idle,
18278 ++ idle->tv_sec - vxi->cvirt.bias_idle.tv_sec,
18279 ++ idle->tv_nsec - vxi->cvirt.bias_idle.tv_nsec);
18280 ++ return;
18281 ++}
18282 ++
18283 ++uint64_t vx_idle_jiffies(void)
18284 ++{
18285 ++ return init_task.utime + init_task.stime;
18286 ++}
18287 ++
18288 ++
18289 ++
18290 ++static inline uint32_t __update_loadavg(uint32_t load,
18291 ++ int wsize, int delta, int n)
18292 ++{
18293 ++ unsigned long long calc, prev;
18294 ++
18295 ++ /* just set it to n */
18296 ++ if (unlikely(delta >= wsize))
18297 ++ return (n << FSHIFT);
18298 ++
18299 ++ calc = delta * n;
18300 ++ calc <<= FSHIFT;
18301 ++ prev = (wsize - delta);
18302 ++ prev *= load;
18303 ++ calc += prev;
18304 ++ do_div(calc, wsize);
18305 ++ return calc;
18306 ++}
18307 ++
18308 ++
18309 ++void vx_update_load(struct vx_info *vxi)
18310 ++{
18311 ++ uint32_t now, last, delta;
18312 ++ unsigned int nr_running, nr_uninterruptible;
18313 ++ unsigned int total;
18314 ++ unsigned long flags;
18315 ++
18316 ++ spin_lock_irqsave(&vxi->cvirt.load_lock, flags);
18317 ++
18318 ++ now = jiffies;
18319 ++ last = vxi->cvirt.load_last;
18320 ++ delta = now - last;
18321 ++
18322 ++ if (delta < 5*HZ)
18323 ++ goto out;
18324 ++
18325 ++ nr_running = atomic_read(&vxi->cvirt.nr_running);
18326 ++ nr_uninterruptible = atomic_read(&vxi->cvirt.nr_uninterruptible);
18327 ++ total = nr_running + nr_uninterruptible;
18328 ++
18329 ++ vxi->cvirt.load[0] = __update_loadavg(vxi->cvirt.load[0],
18330 ++ 60*HZ, delta, total);
18331 ++ vxi->cvirt.load[1] = __update_loadavg(vxi->cvirt.load[1],
18332 ++ 5*60*HZ, delta, total);
18333 ++ vxi->cvirt.load[2] = __update_loadavg(vxi->cvirt.load[2],
18334 ++ 15*60*HZ, delta, total);
18335 ++
18336 ++ vxi->cvirt.load_last = now;
18337 ++out:
18338 ++ atomic_inc(&vxi->cvirt.load_updates);
18339 ++ spin_unlock_irqrestore(&vxi->cvirt.load_lock, flags);
18340 ++}
18341 ++
18342 ++
18343 ++/*
18344 ++ * Commands to do_syslog:
18345 ++ *
18346 ++ * 0 -- Close the log. Currently a NOP.
18347 ++ * 1 -- Open the log. Currently a NOP.
18348 ++ * 2 -- Read from the log.
18349 ++ * 3 -- Read all messages remaining in the ring buffer.
18350 ++ * 4 -- Read and clear all messages remaining in the ring buffer
18351 ++ * 5 -- Clear ring buffer.
18352 ++ * 6 -- Disable printk's to console
18353 ++ * 7 -- Enable printk's to console
18354 ++ * 8 -- Set level of messages printed to console
18355 ++ * 9 -- Return number of unread characters in the log buffer
18356 ++ * 10 -- Return size of the log buffer
18357 ++ */
18358 ++int vx_do_syslog(int type, char __user *buf, int len)
18359 ++{
18360 ++ int error = 0;
18361 ++ int do_clear = 0;
18362 ++ struct vx_info *vxi = current->vx_info;
18363 ++ struct _vx_syslog *log;
18364 ++
18365 ++ if (!vxi)
18366 ++ return -EINVAL;
18367 ++ log = &vxi->cvirt.syslog;
18368 ++
18369 ++ switch (type) {
18370 ++ case 0: /* Close log */
18371 ++ case 1: /* Open log */
18372 ++ break;
18373 ++ case 2: /* Read from log */
18374 ++ error = wait_event_interruptible(log->log_wait,
18375 ++ (log->log_start - log->log_end));
18376 ++ if (error)
18377 ++ break;
18378 ++ spin_lock_irq(&log->logbuf_lock);
18379 ++ spin_unlock_irq(&log->logbuf_lock);
18380 ++ break;
18381 ++ case 4: /* Read/clear last kernel messages */
18382 ++ do_clear = 1;
18383 ++ /* fall through */
18384 ++ case 3: /* Read last kernel messages */
18385 ++ return 0;
18386 ++
18387 ++ case 5: /* Clear ring buffer */
18388 ++ return 0;
18389 ++
18390 ++ case 6: /* Disable logging to console */
18391 ++ case 7: /* Enable logging to console */
18392 ++ case 8: /* Set level of messages printed to console */
18393 ++ break;
18394 ++
18395 ++ case 9: /* Number of chars in the log buffer */
18396 ++ return 0;
18397 ++ case 10: /* Size of the log buffer */
18398 ++ return 0;
18399 ++ default:
18400 ++ error = -EINVAL;
18401 ++ break;
18402 ++ }
18403 ++ return error;
18404 ++}
18405 ++
18406 ++
18407 ++/* virtual host info names */
18408 ++
18409 ++static char *vx_vhi_name(struct vx_info *vxi, int id)
18410 ++{
18411 ++ struct nsproxy *nsproxy;
18412 ++ struct uts_namespace *uts;
18413 ++
18414 ++ if (id == VHIN_CONTEXT)
18415 ++ return vxi->vx_name;
18416 ++
18417 ++ nsproxy = vxi->vx_nsproxy[0];
18418 ++ if (!nsproxy)
18419 ++ return NULL;
18420 ++
18421 ++ uts = nsproxy->uts_ns;
18422 ++ if (!uts)
18423 ++ return NULL;
18424 ++
18425 ++ switch (id) {
18426 ++ case VHIN_SYSNAME:
18427 ++ return uts->name.sysname;
18428 ++ case VHIN_NODENAME:
18429 ++ return uts->name.nodename;
18430 ++ case VHIN_RELEASE:
18431 ++ return uts->name.release;
18432 ++ case VHIN_VERSION:
18433 ++ return uts->name.version;
18434 ++ case VHIN_MACHINE:
18435 ++ return uts->name.machine;
18436 ++ case VHIN_DOMAINNAME:
18437 ++ return uts->name.domainname;
18438 ++ default:
18439 ++ return NULL;
18440 ++ }
18441 ++ return NULL;
18442 ++}
18443 ++
18444 ++int vc_set_vhi_name(struct vx_info *vxi, void __user *data)
18445 ++{
18446 ++ struct vcmd_vhi_name_v0 vc_data;
18447 ++ char *name;
18448 ++
18449 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18450 ++ return -EFAULT;
18451 ++
18452 ++ name = vx_vhi_name(vxi, vc_data.field);
18453 ++ if (!name)
18454 ++ return -EINVAL;
18455 ++
18456 ++ memcpy(name, vc_data.name, 65);
18457 ++ return 0;
18458 ++}
18459 ++
18460 ++int vc_get_vhi_name(struct vx_info *vxi, void __user *data)
18461 ++{
18462 ++ struct vcmd_vhi_name_v0 vc_data;
18463 ++ char *name;
18464 ++
18465 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18466 ++ return -EFAULT;
18467 ++
18468 ++ name = vx_vhi_name(vxi, vc_data.field);
18469 ++ if (!name)
18470 ++ return -EINVAL;
18471 ++
18472 ++ memcpy(vc_data.name, name, 65);
18473 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18474 ++ return -EFAULT;
18475 ++ return 0;
18476 ++}
18477 ++
18478 ++
18479 ++int vc_virt_stat(struct vx_info *vxi, void __user *data)
18480 ++{
18481 ++ struct vcmd_virt_stat_v0 vc_data;
18482 ++ struct _vx_cvirt *cvirt = &vxi->cvirt;
18483 ++ struct timespec uptime;
18484 ++
18485 ++ do_posix_clock_monotonic_gettime(&uptime);
18486 ++ set_normalized_timespec(&uptime,
18487 ++ uptime.tv_sec - cvirt->bias_uptime.tv_sec,
18488 ++ uptime.tv_nsec - cvirt->bias_uptime.tv_nsec);
18489 ++
18490 ++ vc_data.offset = timeval_to_ns(&cvirt->bias_tv);
18491 ++ vc_data.uptime = timespec_to_ns(&uptime);
18492 ++ vc_data.nr_threads = atomic_read(&cvirt->nr_threads);
18493 ++ vc_data.nr_running = atomic_read(&cvirt->nr_running);
18494 ++ vc_data.nr_uninterruptible = atomic_read(&cvirt->nr_uninterruptible);
18495 ++ vc_data.nr_onhold = atomic_read(&cvirt->nr_onhold);
18496 ++ vc_data.nr_forks = atomic_read(&cvirt->total_forks);
18497 ++ vc_data.load[0] = cvirt->load[0];
18498 ++ vc_data.load[1] = cvirt->load[1];
18499 ++ vc_data.load[2] = cvirt->load[2];
18500 ++
18501 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18502 ++ return -EFAULT;
18503 ++ return 0;
18504 ++}
18505 ++
18506 ++
18507 ++#ifdef CONFIG_VSERVER_VTIME
18508 ++
18509 ++/* virtualized time base */
18510 ++
18511 ++void vx_gettimeofday(struct timeval *tv)
18512 ++{
18513 ++ do_gettimeofday(tv);
18514 ++ if (!vx_flags(VXF_VIRT_TIME, 0))
18515 ++ return;
18516 ++
18517 ++ tv->tv_sec += current->vx_info->cvirt.bias_tv.tv_sec;
18518 ++ tv->tv_usec += current->vx_info->cvirt.bias_tv.tv_usec;
18519 ++
18520 ++ if (tv->tv_usec >= USEC_PER_SEC) {
18521 ++ tv->tv_sec++;
18522 ++ tv->tv_usec -= USEC_PER_SEC;
18523 ++ } else if (tv->tv_usec < 0) {
18524 ++ tv->tv_sec--;
18525 ++ tv->tv_usec += USEC_PER_SEC;
18526 ++ }
18527 ++}
18528 ++
18529 ++int vx_settimeofday(struct timespec *ts)
18530 ++{
18531 ++ struct timeval tv;
18532 ++
18533 ++ if (!vx_flags(VXF_VIRT_TIME, 0))
18534 ++ return do_settimeofday(ts);
18535 ++
18536 ++ do_gettimeofday(&tv);
18537 ++ current->vx_info->cvirt.bias_tv.tv_sec =
18538 ++ ts->tv_sec - tv.tv_sec;
18539 ++ current->vx_info->cvirt.bias_tv.tv_usec =
18540 ++ (ts->tv_nsec/NSEC_PER_USEC) - tv.tv_usec;
18541 ++ return 0;
18542 ++}
18543 ++
18544 ++#endif
18545 ++
18546 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cvirt_init.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt_init.h
18547 +--- linux-2.6.29.4/kernel/vserver/cvirt_init.h 1970-01-01 01:00:00.000000000 +0100
18548 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt_init.h 2009-02-22 22:54:26.000000000 +0100
18549 +@@ -0,0 +1,69 @@
18550 ++
18551 ++
18552 ++extern uint64_t vx_idle_jiffies(void);
18553 ++
18554 ++static inline void vx_info_init_cvirt(struct _vx_cvirt *cvirt)
18555 ++{
18556 ++ uint64_t idle_jiffies = vx_idle_jiffies();
18557 ++ uint64_t nsuptime;
18558 ++
18559 ++ do_posix_clock_monotonic_gettime(&cvirt->bias_uptime);
18560 ++ nsuptime = (unsigned long long)cvirt->bias_uptime.tv_sec
18561 ++ * NSEC_PER_SEC + cvirt->bias_uptime.tv_nsec;
18562 ++ cvirt->bias_clock = nsec_to_clock_t(nsuptime);
18563 ++ cvirt->bias_tv.tv_sec = 0;
18564 ++ cvirt->bias_tv.tv_usec = 0;
18565 ++
18566 ++ jiffies_to_timespec(idle_jiffies, &cvirt->bias_idle);
18567 ++ atomic_set(&cvirt->nr_threads, 0);
18568 ++ atomic_set(&cvirt->nr_running, 0);
18569 ++ atomic_set(&cvirt->nr_uninterruptible, 0);
18570 ++ atomic_set(&cvirt->nr_onhold, 0);
18571 ++
18572 ++ spin_lock_init(&cvirt->load_lock);
18573 ++ cvirt->load_last = jiffies;
18574 ++ atomic_set(&cvirt->load_updates, 0);
18575 ++ cvirt->load[0] = 0;
18576 ++ cvirt->load[1] = 0;
18577 ++ cvirt->load[2] = 0;
18578 ++ atomic_set(&cvirt->total_forks, 0);
18579 ++
18580 ++ spin_lock_init(&cvirt->syslog.logbuf_lock);
18581 ++ init_waitqueue_head(&cvirt->syslog.log_wait);
18582 ++ cvirt->syslog.log_start = 0;
18583 ++ cvirt->syslog.log_end = 0;
18584 ++ cvirt->syslog.con_start = 0;
18585 ++ cvirt->syslog.logged_chars = 0;
18586 ++}
18587 ++
18588 ++static inline
18589 ++void vx_info_init_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
18590 ++{
18591 ++ // cvirt_pc->cpustat = { 0 };
18592 ++}
18593 ++
18594 ++static inline void vx_info_exit_cvirt(struct _vx_cvirt *cvirt)
18595 ++{
18596 ++ int value;
18597 ++
18598 ++ vxwprintk_xid((value = atomic_read(&cvirt->nr_threads)),
18599 ++ "!!! cvirt: %p[nr_threads] = %d on exit.",
18600 ++ cvirt, value);
18601 ++ vxwprintk_xid((value = atomic_read(&cvirt->nr_running)),
18602 ++ "!!! cvirt: %p[nr_running] = %d on exit.",
18603 ++ cvirt, value);
18604 ++ vxwprintk_xid((value = atomic_read(&cvirt->nr_uninterruptible)),
18605 ++ "!!! cvirt: %p[nr_uninterruptible] = %d on exit.",
18606 ++ cvirt, value);
18607 ++ vxwprintk_xid((value = atomic_read(&cvirt->nr_onhold)),
18608 ++ "!!! cvirt: %p[nr_onhold] = %d on exit.",
18609 ++ cvirt, value);
18610 ++ return;
18611 ++}
18612 ++
18613 ++static inline
18614 ++void vx_info_exit_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
18615 ++{
18616 ++ return;
18617 ++}
18618 ++
18619 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/cvirt_proc.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt_proc.h
18620 +--- linux-2.6.29.4/kernel/vserver/cvirt_proc.h 1970-01-01 01:00:00.000000000 +0100
18621 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/cvirt_proc.h 2009-02-22 22:54:26.000000000 +0100
18622 +@@ -0,0 +1,135 @@
18623 ++#ifndef _VX_CVIRT_PROC_H
18624 ++#define _VX_CVIRT_PROC_H
18625 ++
18626 ++#include <linux/nsproxy.h>
18627 ++#include <linux/mnt_namespace.h>
18628 ++#include <linux/ipc_namespace.h>
18629 ++#include <linux/utsname.h>
18630 ++#include <linux/ipc.h>
18631 ++
18632 ++
18633 ++static inline
18634 ++int vx_info_proc_nsproxy(struct nsproxy *nsproxy, char *buffer)
18635 ++{
18636 ++ struct mnt_namespace *ns;
18637 ++ struct uts_namespace *uts;
18638 ++ struct ipc_namespace *ipc;
18639 ++ struct path path;
18640 ++ char *pstr, *root;
18641 ++ int length = 0;
18642 ++
18643 ++ if (!nsproxy)
18644 ++ goto out;
18645 ++
18646 ++ length += sprintf(buffer + length,
18647 ++ "NSProxy:\t%p [%p,%p,%p]\n",
18648 ++ nsproxy, nsproxy->mnt_ns,
18649 ++ nsproxy->uts_ns, nsproxy->ipc_ns);
18650 ++
18651 ++ ns = nsproxy->mnt_ns;
18652 ++ if (!ns)
18653 ++ goto skip_ns;
18654 ++
18655 ++ pstr = kmalloc(PATH_MAX, GFP_KERNEL);
18656 ++ if (!pstr)
18657 ++ goto skip_ns;
18658 ++
18659 ++ path.mnt = ns->root;
18660 ++ path.dentry = ns->root->mnt_root;
18661 ++ root = d_path(&path, pstr, PATH_MAX - 2);
18662 ++ length += sprintf(buffer + length,
18663 ++ "Namespace:\t%p [#%u]\n"
18664 ++ "RootPath:\t%s\n",
18665 ++ ns, atomic_read(&ns->count),
18666 ++ root);
18667 ++ kfree(pstr);
18668 ++skip_ns:
18669 ++
18670 ++ uts = nsproxy->uts_ns;
18671 ++ if (!uts)
18672 ++ goto skip_uts;
18673 ++
18674 ++ length += sprintf(buffer + length,
18675 ++ "SysName:\t%.*s\n"
18676 ++ "NodeName:\t%.*s\n"
18677 ++ "Release:\t%.*s\n"
18678 ++ "Version:\t%.*s\n"
18679 ++ "Machine:\t%.*s\n"
18680 ++ "DomainName:\t%.*s\n",
18681 ++ __NEW_UTS_LEN, uts->name.sysname,
18682 ++ __NEW_UTS_LEN, uts->name.nodename,
18683 ++ __NEW_UTS_LEN, uts->name.release,
18684 ++ __NEW_UTS_LEN, uts->name.version,
18685 ++ __NEW_UTS_LEN, uts->name.machine,
18686 ++ __NEW_UTS_LEN, uts->name.domainname);
18687 ++skip_uts:
18688 ++
18689 ++ ipc = nsproxy->ipc_ns;
18690 ++ if (!ipc)
18691 ++ goto skip_ipc;
18692 ++
18693 ++ length += sprintf(buffer + length,
18694 ++ "SEMS:\t\t%d %d %d %d %d\n"
18695 ++ "MSG:\t\t%d %d %d\n"
18696 ++ "SHM:\t\t%lu %lu %d %d\n",
18697 ++ ipc->sem_ctls[0], ipc->sem_ctls[1],
18698 ++ ipc->sem_ctls[2], ipc->sem_ctls[3],
18699 ++ ipc->used_sems,
18700 ++ ipc->msg_ctlmax, ipc->msg_ctlmnb, ipc->msg_ctlmni,
18701 ++ (unsigned long)ipc->shm_ctlmax,
18702 ++ (unsigned long)ipc->shm_ctlall,
18703 ++ ipc->shm_ctlmni, ipc->shm_tot);
18704 ++skip_ipc:
18705 ++out:
18706 ++ return length;
18707 ++}
18708 ++
18709 ++
18710 ++#include <linux/sched.h>
18711 ++
18712 ++#define LOAD_INT(x) ((x) >> FSHIFT)
18713 ++#define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1 - 1)) * 100)
18714 ++
18715 ++static inline
18716 ++int vx_info_proc_cvirt(struct _vx_cvirt *cvirt, char *buffer)
18717 ++{
18718 ++ int length = 0;
18719 ++ int a, b, c;
18720 ++
18721 ++ length += sprintf(buffer + length,
18722 ++ "BiasUptime:\t%lu.%02lu\n",
18723 ++ (unsigned long)cvirt->bias_uptime.tv_sec,
18724 ++ (cvirt->bias_uptime.tv_nsec / (NSEC_PER_SEC / 100)));
18725 ++
18726 ++ a = cvirt->load[0] + (FIXED_1 / 200);
18727 ++ b = cvirt->load[1] + (FIXED_1 / 200);
18728 ++ c = cvirt->load[2] + (FIXED_1 / 200);
18729 ++ length += sprintf(buffer + length,
18730 ++ "nr_threads:\t%d\n"
18731 ++ "nr_running:\t%d\n"
18732 ++ "nr_unintr:\t%d\n"
18733 ++ "nr_onhold:\t%d\n"
18734 ++ "load_updates:\t%d\n"
18735 ++ "loadavg:\t%d.%02d %d.%02d %d.%02d\n"
18736 ++ "total_forks:\t%d\n",
18737 ++ atomic_read(&cvirt->nr_threads),
18738 ++ atomic_read(&cvirt->nr_running),
18739 ++ atomic_read(&cvirt->nr_uninterruptible),
18740 ++ atomic_read(&cvirt->nr_onhold),
18741 ++ atomic_read(&cvirt->load_updates),
18742 ++ LOAD_INT(a), LOAD_FRAC(a),
18743 ++ LOAD_INT(b), LOAD_FRAC(b),
18744 ++ LOAD_INT(c), LOAD_FRAC(c),
18745 ++ atomic_read(&cvirt->total_forks));
18746 ++ return length;
18747 ++}
18748 ++
18749 ++static inline
18750 ++int vx_info_proc_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc,
18751 ++ char *buffer, int cpu)
18752 ++{
18753 ++ int length = 0;
18754 ++ return length;
18755 ++}
18756 ++
18757 ++#endif /* _VX_CVIRT_PROC_H */
18758 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/debug.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/debug.c
18759 +--- linux-2.6.29.4/kernel/vserver/debug.c 1970-01-01 01:00:00.000000000 +0100
18760 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/debug.c 2009-02-22 22:54:26.000000000 +0100
18761 +@@ -0,0 +1,32 @@
18762 ++/*
18763 ++ * kernel/vserver/debug.c
18764 ++ *
18765 ++ * Copyright (C) 2005-2007 Herbert Pötzl
18766 ++ *
18767 ++ * V0.01 vx_info dump support
18768 ++ *
18769 ++ */
18770 ++
18771 ++#include <linux/module.h>
18772 ++
18773 ++#include <linux/vserver/context.h>
18774 ++
18775 ++
18776 ++void dump_vx_info(struct vx_info *vxi, int level)
18777 ++{
18778 ++ printk("vx_info %p[#%d, %d.%d, %4x]\n", vxi, vxi->vx_id,
18779 ++ atomic_read(&vxi->vx_usecnt),
18780 ++ atomic_read(&vxi->vx_tasks),
18781 ++ vxi->vx_state);
18782 ++ if (level > 0) {
18783 ++ __dump_vx_limit(&vxi->limit);
18784 ++ __dump_vx_sched(&vxi->sched);
18785 ++ __dump_vx_cvirt(&vxi->cvirt);
18786 ++ __dump_vx_cacct(&vxi->cacct);
18787 ++ }
18788 ++ printk("---\n");
18789 ++}
18790 ++
18791 ++
18792 ++EXPORT_SYMBOL_GPL(dump_vx_info);
18793 ++
18794 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/device.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/device.c
18795 +--- linux-2.6.29.4/kernel/vserver/device.c 1970-01-01 01:00:00.000000000 +0100
18796 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/device.c 2009-02-22 22:54:26.000000000 +0100
18797 +@@ -0,0 +1,443 @@
18798 ++/*
18799 ++ * linux/kernel/vserver/device.c
18800 ++ *
18801 ++ * Linux-VServer: Device Support
18802 ++ *
18803 ++ * Copyright (C) 2006 Herbert Pötzl
18804 ++ * Copyright (C) 2007 Daniel Hokka Zakrisson
18805 ++ *
18806 ++ * V0.01 device mapping basics
18807 ++ * V0.02 added defaults
18808 ++ *
18809 ++ */
18810 ++
18811 ++#include <linux/slab.h>
18812 ++#include <linux/rcupdate.h>
18813 ++#include <linux/fs.h>
18814 ++#include <linux/namei.h>
18815 ++#include <linux/hash.h>
18816 ++
18817 ++#include <asm/errno.h>
18818 ++#include <asm/uaccess.h>
18819 ++#include <linux/vserver/base.h>
18820 ++#include <linux/vserver/debug.h>
18821 ++#include <linux/vserver/context.h>
18822 ++#include <linux/vserver/device.h>
18823 ++#include <linux/vserver/device_cmd.h>
18824 ++
18825 ++
18826 ++#define DMAP_HASH_BITS 4
18827 ++
18828 ++
18829 ++struct vs_mapping {
18830 ++ union {
18831 ++ struct hlist_node hlist;
18832 ++ struct list_head list;
18833 ++ } u;
18834 ++#define dm_hlist u.hlist
18835 ++#define dm_list u.list
18836 ++ xid_t xid;
18837 ++ dev_t device;
18838 ++ struct vx_dmap_target target;
18839 ++};
18840 ++
18841 ++
18842 ++static struct hlist_head dmap_main_hash[1 << DMAP_HASH_BITS];
18843 ++
18844 ++static spinlock_t dmap_main_hash_lock = SPIN_LOCK_UNLOCKED;
18845 ++
18846 ++static struct vx_dmap_target dmap_defaults[2] = {
18847 ++ { .flags = DATTR_OPEN },
18848 ++ { .flags = DATTR_OPEN },
18849 ++};
18850 ++
18851 ++
18852 ++struct kmem_cache *dmap_cachep __read_mostly;
18853 ++
18854 ++int __init dmap_cache_init(void)
18855 ++{
18856 ++ dmap_cachep = kmem_cache_create("dmap_cache",
18857 ++ sizeof(struct vs_mapping), 0,
18858 ++ SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
18859 ++ return 0;
18860 ++}
18861 ++
18862 ++__initcall(dmap_cache_init);
18863 ++
18864 ++
18865 ++static inline unsigned int __hashval(dev_t dev, int bits)
18866 ++{
18867 ++ return hash_long((unsigned long)dev, bits);
18868 ++}
18869 ++
18870 ++
18871 ++/* __hash_mapping()
18872 ++ * add the mapping to the hash table
18873 ++ */
18874 ++static inline void __hash_mapping(struct vx_info *vxi, struct vs_mapping *vdm)
18875 ++{
18876 ++ spinlock_t *hash_lock = &dmap_main_hash_lock;
18877 ++ struct hlist_head *head, *hash = dmap_main_hash;
18878 ++ int device = vdm->device;
18879 ++
18880 ++ spin_lock(hash_lock);
18881 ++ vxdprintk(VXD_CBIT(misc, 8), "__hash_mapping: %p[#%d] %08x:%08x",
18882 ++ vxi, vxi ? vxi->vx_id : 0, device, vdm->target.target);
18883 ++
18884 ++ head = &hash[__hashval(device, DMAP_HASH_BITS)];
18885 ++ hlist_add_head(&vdm->dm_hlist, head);
18886 ++ spin_unlock(hash_lock);
18887 ++}
18888 ++
18889 ++
18890 ++static inline int __mode_to_default(umode_t mode)
18891 ++{
18892 ++ switch (mode) {
18893 ++ case S_IFBLK:
18894 ++ return 0;
18895 ++ case S_IFCHR:
18896 ++ return 1;
18897 ++ default:
18898 ++ BUG();
18899 ++ }
18900 ++}
18901 ++
18902 ++
18903 ++/* __set_default()
18904 ++ * set a default
18905 ++ */
18906 ++static inline void __set_default(struct vx_info *vxi, umode_t mode,
18907 ++ struct vx_dmap_target *vdmt)
18908 ++{
18909 ++ spinlock_t *hash_lock = &dmap_main_hash_lock;
18910 ++ spin_lock(hash_lock);
18911 ++
18912 ++ if (vxi)
18913 ++ vxi->dmap.targets[__mode_to_default(mode)] = *vdmt;
18914 ++ else
18915 ++ dmap_defaults[__mode_to_default(mode)] = *vdmt;
18916 ++
18917 ++
18918 ++ spin_unlock(hash_lock);
18919 ++
18920 ++ vxdprintk(VXD_CBIT(misc, 8), "__set_default: %p[#%u] %08x %04x",
18921 ++ vxi, vxi ? vxi->vx_id : 0, vdmt->target, vdmt->flags);
18922 ++}
18923 ++
18924 ++
18925 ++/* __remove_default()
18926 ++ * remove a default
18927 ++ */
18928 ++static inline int __remove_default(struct vx_info *vxi, umode_t mode)
18929 ++{
18930 ++ spinlock_t *hash_lock = &dmap_main_hash_lock;
18931 ++ spin_lock(hash_lock);
18932 ++
18933 ++ if (vxi)
18934 ++ vxi->dmap.targets[__mode_to_default(mode)].flags = 0;
18935 ++ else /* remove == reset */
18936 ++ dmap_defaults[__mode_to_default(mode)].flags = DATTR_OPEN | mode;
18937 ++
18938 ++ spin_unlock(hash_lock);
18939 ++ return 0;
18940 ++}
18941 ++
18942 ++
18943 ++/* __find_mapping()
18944 ++ * find a mapping in the hash table
18945 ++ *
18946 ++ * caller must hold hash_lock
18947 ++ */
18948 ++static inline int __find_mapping(xid_t xid, dev_t device, umode_t mode,
18949 ++ struct vs_mapping **local, struct vs_mapping **global)
18950 ++{
18951 ++ struct hlist_head *hash = dmap_main_hash;
18952 ++ struct hlist_head *head = &hash[__hashval(device, DMAP_HASH_BITS)];
18953 ++ struct hlist_node *pos;
18954 ++ struct vs_mapping *vdm;
18955 ++
18956 ++ *local = NULL;
18957 ++ if (global)
18958 ++ *global = NULL;
18959 ++
18960 ++ hlist_for_each(pos, head) {
18961 ++ vdm = hlist_entry(pos, struct vs_mapping, dm_hlist);
18962 ++
18963 ++ if ((vdm->device == device) &&
18964 ++ !((vdm->target.flags ^ mode) & S_IFMT)) {
18965 ++ if (vdm->xid == xid) {
18966 ++ *local = vdm;
18967 ++ return 1;
18968 ++ } else if (global && vdm->xid == 0)
18969 ++ *global = vdm;
18970 ++ }
18971 ++ }
18972 ++
18973 ++ if (global && *global)
18974 ++ return 0;
18975 ++ else
18976 ++ return -ENOENT;
18977 ++}
18978 ++
18979 ++
18980 ++/* __lookup_mapping()
18981 ++ * find a mapping and store the result in target and flags
18982 ++ */
18983 ++static inline int __lookup_mapping(struct vx_info *vxi,
18984 ++ dev_t device, dev_t *target, int *flags, umode_t mode)
18985 ++{
18986 ++ spinlock_t *hash_lock = &dmap_main_hash_lock;
18987 ++ struct vs_mapping *vdm, *global;
18988 ++ struct vx_dmap_target *vdmt;
18989 ++ int ret = 0;
18990 ++ xid_t xid = vxi->vx_id;
18991 ++ int index;
18992 ++
18993 ++ spin_lock(hash_lock);
18994 ++ if (__find_mapping(xid, device, mode, &vdm, &global) > 0) {
18995 ++ ret = 1;
18996 ++ vdmt = &vdm->target;
18997 ++ goto found;
18998 ++ }
18999 ++
19000 ++ index = __mode_to_default(mode);
19001 ++ if (vxi && vxi->dmap.targets[index].flags) {
19002 ++ ret = 2;
19003 ++ vdmt = &vxi->dmap.targets[index];
19004 ++ } else if (global) {
19005 ++ ret = 3;
19006 ++ vdmt = &global->target;
19007 ++ goto found;
19008 ++ } else {
19009 ++ ret = 4;
19010 ++ vdmt = &dmap_defaults[index];
19011 ++ }
19012 ++
19013 ++found:
19014 ++ if (target && (vdmt->flags & DATTR_REMAP))
19015 ++ *target = vdmt->target;
19016 ++ else if (target)
19017 ++ *target = device;
19018 ++ if (flags)
19019 ++ *flags = vdmt->flags;
19020 ++
19021 ++ spin_unlock(hash_lock);
19022 ++
19023 ++ return ret;
19024 ++}
19025 ++
19026 ++
19027 ++/* __remove_mapping()
19028 ++ * remove a mapping from the hash table
19029 ++ */
19030 ++static inline int __remove_mapping(struct vx_info *vxi, dev_t device,
19031 ++ umode_t mode)
19032 ++{
19033 ++ spinlock_t *hash_lock = &dmap_main_hash_lock;
19034 ++ struct vs_mapping *vdm = NULL;
19035 ++ int ret = 0;
19036 ++
19037 ++ spin_lock(hash_lock);
19038 ++
19039 ++ ret = __find_mapping((vxi ? vxi->vx_id : 0), device, mode, &vdm,
19040 ++ NULL);
19041 ++ vxdprintk(VXD_CBIT(misc, 8), "__remove_mapping: %p[#%d] %08x %04x",
19042 ++ vxi, vxi ? vxi->vx_id : 0, device, mode);
19043 ++ if (ret < 0)
19044 ++ goto out;
19045 ++ hlist_del(&vdm->dm_hlist);
19046 ++
19047 ++out:
19048 ++ spin_unlock(hash_lock);
19049 ++ if (vdm)
19050 ++ kmem_cache_free(dmap_cachep, vdm);
19051 ++ return ret;
19052 ++}
19053 ++
19054 ++
19055 ++
19056 ++int vs_map_device(struct vx_info *vxi,
19057 ++ dev_t device, dev_t *target, umode_t mode)
19058 ++{
19059 ++ int ret, flags = DATTR_MASK;
19060 ++
19061 ++ if (!vxi) {
19062 ++ if (target)
19063 ++ *target = device;
19064 ++ goto out;
19065 ++ }
19066 ++ ret = __lookup_mapping(vxi, device, target, &flags, mode);
19067 ++ vxdprintk(VXD_CBIT(misc, 8), "vs_map_device: %08x target: %08x flags: %04x mode: %04x mapped=%d",
19068 ++ device, target ? *target : 0, flags, mode, ret);
19069 ++out:
19070 ++ return (flags & DATTR_MASK);
19071 ++}
19072 ++
19073 ++
19074 ++
19075 ++static int do_set_mapping(struct vx_info *vxi,
19076 ++ dev_t device, dev_t target, int flags, umode_t mode)
19077 ++{
19078 ++ if (device) {
19079 ++ struct vs_mapping *new;
19080 ++
19081 ++ new = kmem_cache_alloc(dmap_cachep, GFP_KERNEL);
19082 ++ if (!new)
19083 ++ return -ENOMEM;
19084 ++
19085 ++ INIT_HLIST_NODE(&new->dm_hlist);
19086 ++ new->device = device;
19087 ++ new->target.target = target;
19088 ++ new->target.flags = flags | mode;
19089 ++ new->xid = (vxi ? vxi->vx_id : 0);
19090 ++
19091 ++ vxdprintk(VXD_CBIT(misc, 8), "do_set_mapping: %08x target: %08x flags: %04x", device, target, flags);
19092 ++ __hash_mapping(vxi, new);
19093 ++ } else {
19094 ++ struct vx_dmap_target new = {
19095 ++ .target = target,
19096 ++ .flags = flags | mode,
19097 ++ };
19098 ++ __set_default(vxi, mode, &new);
19099 ++ }
19100 ++ return 0;
19101 ++}
19102 ++
19103 ++
19104 ++static int do_unset_mapping(struct vx_info *vxi,
19105 ++ dev_t device, dev_t target, int flags, umode_t mode)
19106 ++{
19107 ++ int ret = -EINVAL;
19108 ++
19109 ++ if (device) {
19110 ++ ret = __remove_mapping(vxi, device, mode);
19111 ++ if (ret < 0)
19112 ++ goto out;
19113 ++ } else {
19114 ++ ret = __remove_default(vxi, mode);
19115 ++ if (ret < 0)
19116 ++ goto out;
19117 ++ }
19118 ++
19119 ++out:
19120 ++ return ret;
19121 ++}
19122 ++
19123 ++
19124 ++static inline int __user_device(const char __user *name, dev_t *dev,
19125 ++ umode_t *mode)
19126 ++{
19127 ++ struct nameidata nd;
19128 ++ int ret;
19129 ++
19130 ++ if (!name) {
19131 ++ *dev = 0;
19132 ++ return 0;
19133 ++ }
19134 ++ ret = user_lpath(name, &nd.path);
19135 ++ if (ret)
19136 ++ return ret;
19137 ++ if (nd.path.dentry->d_inode) {
19138 ++ *dev = nd.path.dentry->d_inode->i_rdev;
19139 ++ *mode = nd.path.dentry->d_inode->i_mode;
19140 ++ }
19141 ++ path_put(&nd.path);
19142 ++ return 0;
19143 ++}
19144 ++
19145 ++static inline int __mapping_mode(dev_t device, dev_t target,
19146 ++ umode_t device_mode, umode_t target_mode, umode_t *mode)
19147 ++{
19148 ++ if (device)
19149 ++ *mode = device_mode & S_IFMT;
19150 ++ else if (target)
19151 ++ *mode = target_mode & S_IFMT;
19152 ++ else
19153 ++ return -EINVAL;
19154 ++
19155 ++ /* if both given, device and target mode have to match */
19156 ++ if (device && target &&
19157 ++ ((device_mode ^ target_mode) & S_IFMT))
19158 ++ return -EINVAL;
19159 ++ return 0;
19160 ++}
19161 ++
19162 ++
19163 ++static inline int do_mapping(struct vx_info *vxi, const char __user *device_path,
19164 ++ const char __user *target_path, int flags, int set)
19165 ++{
19166 ++ dev_t device = ~0, target = ~0;
19167 ++ umode_t device_mode = 0, target_mode = 0, mode;
19168 ++ int ret;
19169 ++
19170 ++ ret = __user_device(device_path, &device, &device_mode);
19171 ++ if (ret)
19172 ++ return ret;
19173 ++ ret = __user_device(target_path, &target, &target_mode);
19174 ++ if (ret)
19175 ++ return ret;
19176 ++
19177 ++ ret = __mapping_mode(device, target,
19178 ++ device_mode, target_mode, &mode);
19179 ++ if (ret)
19180 ++ return ret;
19181 ++
19182 ++ if (set)
19183 ++ return do_set_mapping(vxi, device, target,
19184 ++ flags, mode);
19185 ++ else
19186 ++ return do_unset_mapping(vxi, device, target,
19187 ++ flags, mode);
19188 ++}
19189 ++
19190 ++
19191 ++int vc_set_mapping(struct vx_info *vxi, void __user *data)
19192 ++{
19193 ++ struct vcmd_set_mapping_v0 vc_data;
19194 ++
19195 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19196 ++ return -EFAULT;
19197 ++
19198 ++ return do_mapping(vxi, vc_data.device, vc_data.target,
19199 ++ vc_data.flags, 1);
19200 ++}
19201 ++
19202 ++int vc_unset_mapping(struct vx_info *vxi, void __user *data)
19203 ++{
19204 ++ struct vcmd_set_mapping_v0 vc_data;
19205 ++
19206 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19207 ++ return -EFAULT;
19208 ++
19209 ++ return do_mapping(vxi, vc_data.device, vc_data.target,
19210 ++ vc_data.flags, 0);
19211 ++}
19212 ++
19213 ++
19214 ++#ifdef CONFIG_COMPAT
19215 ++
19216 ++int vc_set_mapping_x32(struct vx_info *vxi, void __user *data)
19217 ++{
19218 ++ struct vcmd_set_mapping_v0_x32 vc_data;
19219 ++
19220 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19221 ++ return -EFAULT;
19222 ++
19223 ++ return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19224 ++ compat_ptr(vc_data.target_ptr), vc_data.flags, 1);
19225 ++}
19226 ++
19227 ++int vc_unset_mapping_x32(struct vx_info *vxi, void __user *data)
19228 ++{
19229 ++ struct vcmd_set_mapping_v0_x32 vc_data;
19230 ++
19231 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19232 ++ return -EFAULT;
19233 ++
19234 ++ return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
19235 ++ compat_ptr(vc_data.target_ptr), vc_data.flags, 0);
19236 ++}
19237 ++
19238 ++#endif /* CONFIG_COMPAT */
19239 ++
19240 ++
19241 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/dlimit.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/dlimit.c
19242 +--- linux-2.6.29.4/kernel/vserver/dlimit.c 1970-01-01 01:00:00.000000000 +0100
19243 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/dlimit.c 2009-02-22 22:54:26.000000000 +0100
19244 +@@ -0,0 +1,522 @@
19245 ++/*
19246 ++ * linux/kernel/vserver/dlimit.c
19247 ++ *
19248 ++ * Virtual Server: Context Disk Limits
19249 ++ *
19250 ++ * Copyright (C) 2004-2007 Herbert Pötzl
19251 ++ *
19252 ++ * V0.01 initial version
19253 ++ * V0.02 compat32 splitup
19254 ++ *
19255 ++ */
19256 ++
19257 ++#include <linux/statfs.h>
19258 ++#include <linux/sched.h>
19259 ++#include <linux/namei.h>
19260 ++#include <linux/vs_tag.h>
19261 ++#include <linux/vs_dlimit.h>
19262 ++#include <linux/vserver/dlimit_cmd.h>
19263 ++
19264 ++#include <asm/uaccess.h>
19265 ++
19266 ++/* __alloc_dl_info()
19267 ++
19268 ++ * allocate an initialized dl_info struct
19269 ++ * doesn't make it visible (hash) */
19270 ++
19271 ++static struct dl_info *__alloc_dl_info(struct super_block *sb, tag_t tag)
19272 ++{
19273 ++ struct dl_info *new = NULL;
19274 ++
19275 ++ vxdprintk(VXD_CBIT(dlim, 5),
19276 ++ "alloc_dl_info(%p,%d)*", sb, tag);
19277 ++
19278 ++ /* would this benefit from a slab cache? */
19279 ++ new = kmalloc(sizeof(struct dl_info), GFP_KERNEL);
19280 ++ if (!new)
19281 ++ return 0;
19282 ++
19283 ++ memset(new, 0, sizeof(struct dl_info));
19284 ++ new->dl_tag = tag;
19285 ++ new->dl_sb = sb;
19286 ++ INIT_RCU_HEAD(&new->dl_rcu);
19287 ++ INIT_HLIST_NODE(&new->dl_hlist);
19288 ++ spin_lock_init(&new->dl_lock);
19289 ++ atomic_set(&new->dl_refcnt, 0);
19290 ++ atomic_set(&new->dl_usecnt, 0);
19291 ++
19292 ++ /* rest of init goes here */
19293 ++
19294 ++ vxdprintk(VXD_CBIT(dlim, 4),
19295 ++ "alloc_dl_info(%p,%d) = %p", sb, tag, new);
19296 ++ return new;
19297 ++}
19298 ++
19299 ++/* __dealloc_dl_info()
19300 ++
19301 ++ * final disposal of dl_info */
19302 ++
19303 ++static void __dealloc_dl_info(struct dl_info *dli)
19304 ++{
19305 ++ vxdprintk(VXD_CBIT(dlim, 4),
19306 ++ "dealloc_dl_info(%p)", dli);
19307 ++
19308 ++ dli->dl_hlist.next = LIST_POISON1;
19309 ++ dli->dl_tag = -1;
19310 ++ dli->dl_sb = 0;
19311 ++
19312 ++ BUG_ON(atomic_read(&dli->dl_usecnt));
19313 ++ BUG_ON(atomic_read(&dli->dl_refcnt));
19314 ++
19315 ++ kfree(dli);
19316 ++}
19317 ++
19318 ++
19319 ++/* hash table for dl_info hash */
19320 ++
19321 ++#define DL_HASH_SIZE 13
19322 ++
19323 ++struct hlist_head dl_info_hash[DL_HASH_SIZE];
19324 ++
19325 ++static spinlock_t dl_info_hash_lock = SPIN_LOCK_UNLOCKED;
19326 ++
19327 ++
19328 ++static inline unsigned int __hashval(struct super_block *sb, tag_t tag)
19329 ++{
19330 ++ return ((tag ^ (unsigned long)sb) % DL_HASH_SIZE);
19331 ++}
19332 ++
19333 ++
19334 ++
19335 ++/* __hash_dl_info()
19336 ++
19337 ++ * add the dli to the global hash table
19338 ++ * requires the hash_lock to be held */
19339 ++
19340 ++static inline void __hash_dl_info(struct dl_info *dli)
19341 ++{
19342 ++ struct hlist_head *head;
19343 ++
19344 ++ vxdprintk(VXD_CBIT(dlim, 6),
19345 ++ "__hash_dl_info: %p[#%d]", dli, dli->dl_tag);
19346 ++ get_dl_info(dli);
19347 ++ head = &dl_info_hash[__hashval(dli->dl_sb, dli->dl_tag)];
19348 ++ hlist_add_head_rcu(&dli->dl_hlist, head);
19349 ++}
19350 ++
19351 ++/* __unhash_dl_info()
19352 ++
19353 ++ * remove the dli from the global hash table
19354 ++ * requires the hash_lock to be held */
19355 ++
19356 ++static inline void __unhash_dl_info(struct dl_info *dli)
19357 ++{
19358 ++ vxdprintk(VXD_CBIT(dlim, 6),
19359 ++ "__unhash_dl_info: %p[#%d]", dli, dli->dl_tag);
19360 ++ hlist_del_rcu(&dli->dl_hlist);
19361 ++ put_dl_info(dli);
19362 ++}
19363 ++
19364 ++
19365 ++/* __lookup_dl_info()
19366 ++
19367 ++ * requires the rcu_read_lock()
19368 ++ * doesn't increment the dl_refcnt */
19369 ++
19370 ++static inline struct dl_info *__lookup_dl_info(struct super_block *sb, tag_t tag)
19371 ++{
19372 ++ struct hlist_head *head = &dl_info_hash[__hashval(sb, tag)];
19373 ++ struct hlist_node *pos;
19374 ++ struct dl_info *dli;
19375 ++
19376 ++ hlist_for_each_entry_rcu(dli, pos, head, dl_hlist) {
19377 ++
19378 ++ if (dli->dl_tag == tag && dli->dl_sb == sb) {
19379 ++ return dli;
19380 ++ }
19381 ++ }
19382 ++ return NULL;
19383 ++}
19384 ++
19385 ++
19386 ++struct dl_info *locate_dl_info(struct super_block *sb, tag_t tag)
19387 ++{
19388 ++ struct dl_info *dli;
19389 ++
19390 ++ rcu_read_lock();
19391 ++ dli = get_dl_info(__lookup_dl_info(sb, tag));
19392 ++ vxdprintk(VXD_CBIT(dlim, 7),
19393 ++ "locate_dl_info(%p,#%d) = %p", sb, tag, dli);
19394 ++ rcu_read_unlock();
19395 ++ return dli;
19396 ++}
19397 ++
19398 ++void rcu_free_dl_info(struct rcu_head *head)
19399 ++{
19400 ++ struct dl_info *dli = container_of(head, struct dl_info, dl_rcu);
19401 ++ int usecnt, refcnt;
19402 ++
19403 ++ BUG_ON(!dli || !head);
19404 ++
19405 ++ usecnt = atomic_read(&dli->dl_usecnt);
19406 ++ BUG_ON(usecnt < 0);
19407 ++
19408 ++ refcnt = atomic_read(&dli->dl_refcnt);
19409 ++ BUG_ON(refcnt < 0);
19410 ++
19411 ++ vxdprintk(VXD_CBIT(dlim, 3),
19412 ++ "rcu_free_dl_info(%p)", dli);
19413 ++ if (!usecnt)
19414 ++ __dealloc_dl_info(dli);
19415 ++ else
19416 ++ printk("!!! rcu didn't free\n");
19417 ++}
19418 ++
19419 ++
19420 ++
19421 ++
19422 ++static int do_addrem_dlimit(uint32_t id, const char __user *name,
19423 ++ uint32_t flags, int add)
19424 ++{
19425 ++ struct path path;
19426 ++ int ret;
19427 ++
19428 ++ ret = user_lpath(name, &path);
19429 ++ if (!ret) {
19430 ++ struct super_block *sb;
19431 ++ struct dl_info *dli;
19432 ++
19433 ++ ret = -EINVAL;
19434 ++ if (!path.dentry->d_inode)
19435 ++ goto out_release;
19436 ++ if (!(sb = path.dentry->d_inode->i_sb))
19437 ++ goto out_release;
19438 ++
19439 ++ if (add) {
19440 ++ dli = __alloc_dl_info(sb, id);
19441 ++ spin_lock(&dl_info_hash_lock);
19442 ++
19443 ++ ret = -EEXIST;
19444 ++ if (__lookup_dl_info(sb, id))
19445 ++ goto out_unlock;
19446 ++ __hash_dl_info(dli);
19447 ++ dli = NULL;
19448 ++ } else {
19449 ++ spin_lock(&dl_info_hash_lock);
19450 ++ dli = __lookup_dl_info(sb, id);
19451 ++
19452 ++ ret = -ESRCH;
19453 ++ if (!dli)
19454 ++ goto out_unlock;
19455 ++ __unhash_dl_info(dli);
19456 ++ }
19457 ++ ret = 0;
19458 ++ out_unlock:
19459 ++ spin_unlock(&dl_info_hash_lock);
19460 ++ if (add && dli)
19461 ++ __dealloc_dl_info(dli);
19462 ++ out_release:
19463 ++ path_put(&path);
19464 ++ }
19465 ++ return ret;
19466 ++}
19467 ++
19468 ++int vc_add_dlimit(uint32_t id, void __user *data)
19469 ++{
19470 ++ struct vcmd_ctx_dlimit_base_v0 vc_data;
19471 ++
19472 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19473 ++ return -EFAULT;
19474 ++
19475 ++ return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 1);
19476 ++}
19477 ++
19478 ++int vc_rem_dlimit(uint32_t id, void __user *data)
19479 ++{
19480 ++ struct vcmd_ctx_dlimit_base_v0 vc_data;
19481 ++
19482 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19483 ++ return -EFAULT;
19484 ++
19485 ++ return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 0);
19486 ++}
19487 ++
19488 ++#ifdef CONFIG_COMPAT
19489 ++
19490 ++int vc_add_dlimit_x32(uint32_t id, void __user *data)
19491 ++{
19492 ++ struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19493 ++
19494 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19495 ++ return -EFAULT;
19496 ++
19497 ++ return do_addrem_dlimit(id,
19498 ++ compat_ptr(vc_data.name_ptr), vc_data.flags, 1);
19499 ++}
19500 ++
19501 ++int vc_rem_dlimit_x32(uint32_t id, void __user *data)
19502 ++{
19503 ++ struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
19504 ++
19505 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19506 ++ return -EFAULT;
19507 ++
19508 ++ return do_addrem_dlimit(id,
19509 ++ compat_ptr(vc_data.name_ptr), vc_data.flags, 0);
19510 ++}
19511 ++
19512 ++#endif /* CONFIG_COMPAT */
19513 ++
19514 ++
19515 ++static inline
19516 ++int do_set_dlimit(uint32_t id, const char __user *name,
19517 ++ uint32_t space_used, uint32_t space_total,
19518 ++ uint32_t inodes_used, uint32_t inodes_total,
19519 ++ uint32_t reserved, uint32_t flags)
19520 ++{
19521 ++ struct path path;
19522 ++ int ret;
19523 ++
19524 ++ ret = user_lpath(name, &path);
19525 ++ if (!ret) {
19526 ++ struct super_block *sb;
19527 ++ struct dl_info *dli;
19528 ++
19529 ++ ret = -EINVAL;
19530 ++ if (!path.dentry->d_inode)
19531 ++ goto out_release;
19532 ++ if (!(sb = path.dentry->d_inode->i_sb))
19533 ++ goto out_release;
19534 ++ if ((reserved != CDLIM_KEEP &&
19535 ++ reserved > 100) ||
19536 ++ (inodes_used != CDLIM_KEEP &&
19537 ++ inodes_used > inodes_total) ||
19538 ++ (space_used != CDLIM_KEEP &&
19539 ++ space_used > space_total))
19540 ++ goto out_release;
19541 ++
19542 ++ ret = -ESRCH;
19543 ++ dli = locate_dl_info(sb, id);
19544 ++ if (!dli)
19545 ++ goto out_release;
19546 ++
19547 ++ spin_lock(&dli->dl_lock);
19548 ++
19549 ++ if (inodes_used != CDLIM_KEEP)
19550 ++ dli->dl_inodes_used = inodes_used;
19551 ++ if (inodes_total != CDLIM_KEEP)
19552 ++ dli->dl_inodes_total = inodes_total;
19553 ++ if (space_used != CDLIM_KEEP) {
19554 ++ dli->dl_space_used = space_used;
19555 ++ dli->dl_space_used <<= 10;
19556 ++ }
19557 ++ if (space_total == CDLIM_INFINITY)
19558 ++ dli->dl_space_total = DLIM_INFINITY;
19559 ++ else if (space_total != CDLIM_KEEP) {
19560 ++ dli->dl_space_total = space_total;
19561 ++ dli->dl_space_total <<= 10;
19562 ++ }
19563 ++ if (reserved != CDLIM_KEEP)
19564 ++ dli->dl_nrlmult = (1 << 10) * (100 - reserved) / 100;
19565 ++
19566 ++ spin_unlock(&dli->dl_lock);
19567 ++
19568 ++ put_dl_info(dli);
19569 ++ ret = 0;
19570 ++
19571 ++ out_release:
19572 ++ path_put(&path);
19573 ++ }
19574 ++ return ret;
19575 ++}
19576 ++
19577 ++int vc_set_dlimit(uint32_t id, void __user *data)
19578 ++{
19579 ++ struct vcmd_ctx_dlimit_v0 vc_data;
19580 ++
19581 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19582 ++ return -EFAULT;
19583 ++
19584 ++ return do_set_dlimit(id, vc_data.name,
19585 ++ vc_data.space_used, vc_data.space_total,
19586 ++ vc_data.inodes_used, vc_data.inodes_total,
19587 ++ vc_data.reserved, vc_data.flags);
19588 ++}
19589 ++
19590 ++#ifdef CONFIG_COMPAT
19591 ++
19592 ++int vc_set_dlimit_x32(uint32_t id, void __user *data)
19593 ++{
19594 ++ struct vcmd_ctx_dlimit_v0_x32 vc_data;
19595 ++
19596 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19597 ++ return -EFAULT;
19598 ++
19599 ++ return do_set_dlimit(id, compat_ptr(vc_data.name_ptr),
19600 ++ vc_data.space_used, vc_data.space_total,
19601 ++ vc_data.inodes_used, vc_data.inodes_total,
19602 ++ vc_data.reserved, vc_data.flags);
19603 ++}
19604 ++
19605 ++#endif /* CONFIG_COMPAT */
19606 ++
19607 ++
19608 ++static inline
19609 ++int do_get_dlimit(uint32_t id, const char __user *name,
19610 ++ uint32_t *space_used, uint32_t *space_total,
19611 ++ uint32_t *inodes_used, uint32_t *inodes_total,
19612 ++ uint32_t *reserved, uint32_t *flags)
19613 ++{
19614 ++ struct path path;
19615 ++ int ret;
19616 ++
19617 ++ ret = user_lpath(name, &path);
19618 ++ if (!ret) {
19619 ++ struct super_block *sb;
19620 ++ struct dl_info *dli;
19621 ++
19622 ++ ret = -EINVAL;
19623 ++ if (!path.dentry->d_inode)
19624 ++ goto out_release;
19625 ++ if (!(sb = path.dentry->d_inode->i_sb))
19626 ++ goto out_release;
19627 ++
19628 ++ ret = -ESRCH;
19629 ++ dli = locate_dl_info(sb, id);
19630 ++ if (!dli)
19631 ++ goto out_release;
19632 ++
19633 ++ spin_lock(&dli->dl_lock);
19634 ++ *inodes_used = dli->dl_inodes_used;
19635 ++ *inodes_total = dli->dl_inodes_total;
19636 ++ *space_used = dli->dl_space_used >> 10;
19637 ++ if (dli->dl_space_total == DLIM_INFINITY)
19638 ++ *space_total = CDLIM_INFINITY;
19639 ++ else
19640 ++ *space_total = dli->dl_space_total >> 10;
19641 ++
19642 ++ *reserved = 100 - ((dli->dl_nrlmult * 100 + 512) >> 10);
19643 ++ spin_unlock(&dli->dl_lock);
19644 ++
19645 ++ put_dl_info(dli);
19646 ++ ret = -EFAULT;
19647 ++
19648 ++ ret = 0;
19649 ++ out_release:
19650 ++ path_put(&path);
19651 ++ }
19652 ++ return ret;
19653 ++}
19654 ++
19655 ++
19656 ++int vc_get_dlimit(uint32_t id, void __user *data)
19657 ++{
19658 ++ struct vcmd_ctx_dlimit_v0 vc_data;
19659 ++ int ret;
19660 ++
19661 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19662 ++ return -EFAULT;
19663 ++
19664 ++ ret = do_get_dlimit(id, vc_data.name,
19665 ++ &vc_data.space_used, &vc_data.space_total,
19666 ++ &vc_data.inodes_used, &vc_data.inodes_total,
19667 ++ &vc_data.reserved, &vc_data.flags);
19668 ++ if (ret)
19669 ++ return ret;
19670 ++
19671 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19672 ++ return -EFAULT;
19673 ++ return 0;
19674 ++}
19675 ++
19676 ++#ifdef CONFIG_COMPAT
19677 ++
19678 ++int vc_get_dlimit_x32(uint32_t id, void __user *data)
19679 ++{
19680 ++ struct vcmd_ctx_dlimit_v0_x32 vc_data;
19681 ++ int ret;
19682 ++
19683 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19684 ++ return -EFAULT;
19685 ++
19686 ++ ret = do_get_dlimit(id, compat_ptr(vc_data.name_ptr),
19687 ++ &vc_data.space_used, &vc_data.space_total,
19688 ++ &vc_data.inodes_used, &vc_data.inodes_total,
19689 ++ &vc_data.reserved, &vc_data.flags);
19690 ++ if (ret)
19691 ++ return ret;
19692 ++
19693 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19694 ++ return -EFAULT;
19695 ++ return 0;
19696 ++}
19697 ++
19698 ++#endif /* CONFIG_COMPAT */
19699 ++
19700 ++
19701 ++void vx_vsi_statfs(struct super_block *sb, struct kstatfs *buf)
19702 ++{
19703 ++ struct dl_info *dli;
19704 ++ __u64 blimit, bfree, bavail;
19705 ++ __u32 ifree;
19706 ++
19707 ++ dli = locate_dl_info(sb, dx_current_tag());
19708 ++ if (!dli)
19709 ++ return;
19710 ++
19711 ++ spin_lock(&dli->dl_lock);
19712 ++ if (dli->dl_inodes_total == (unsigned long)DLIM_INFINITY)
19713 ++ goto no_ilim;
19714 ++
19715 ++ /* reduce max inodes available to limit */
19716 ++ if (buf->f_files > dli->dl_inodes_total)
19717 ++ buf->f_files = dli->dl_inodes_total;
19718 ++
19719 ++ ifree = dli->dl_inodes_total - dli->dl_inodes_used;
19720 ++ /* reduce free inodes to min */
19721 ++ if (ifree < buf->f_ffree)
19722 ++ buf->f_ffree = ifree;
19723 ++
19724 ++no_ilim:
19725 ++ if (dli->dl_space_total == DLIM_INFINITY)
19726 ++ goto no_blim;
19727 ++
19728 ++ blimit = dli->dl_space_total >> sb->s_blocksize_bits;
19729 ++
19730 ++ if (dli->dl_space_total < dli->dl_space_used)
19731 ++ bfree = 0;
19732 ++ else
19733 ++ bfree = (dli->dl_space_total - dli->dl_space_used)
19734 ++ >> sb->s_blocksize_bits;
19735 ++
19736 ++ bavail = ((dli->dl_space_total >> 10) * dli->dl_nrlmult);
19737 ++ if (bavail < dli->dl_space_used)
19738 ++ bavail = 0;
19739 ++ else
19740 ++ bavail = (bavail - dli->dl_space_used)
19741 ++ >> sb->s_blocksize_bits;
19742 ++
19743 ++ /* reduce max space available to limit */
19744 ++ if (buf->f_blocks > blimit)
19745 ++ buf->f_blocks = blimit;
19746 ++
19747 ++ /* reduce free space to min */
19748 ++ if (bfree < buf->f_bfree)
19749 ++ buf->f_bfree = bfree;
19750 ++
19751 ++ /* reduce avail space to min */
19752 ++ if (bavail < buf->f_bavail)
19753 ++ buf->f_bavail = bavail;
19754 ++
19755 ++no_blim:
19756 ++ spin_unlock(&dli->dl_lock);
19757 ++ put_dl_info(dli);
19758 ++
19759 ++ return;
19760 ++}
19761 ++
19762 ++#include <linux/module.h>
19763 ++
19764 ++EXPORT_SYMBOL_GPL(locate_dl_info);
19765 ++EXPORT_SYMBOL_GPL(rcu_free_dl_info);
19766 ++
19767 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/helper.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/helper.c
19768 +--- linux-2.6.29.4/kernel/vserver/helper.c 1970-01-01 01:00:00.000000000 +0100
19769 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/helper.c 2009-03-25 00:40:43.000000000 +0100
19770 +@@ -0,0 +1,199 @@
19771 ++/*
19772 ++ * linux/kernel/vserver/helper.c
19773 ++ *
19774 ++ * Virtual Context Support
19775 ++ *
19776 ++ * Copyright (C) 2004-2007 Herbert Pötzl
19777 ++ *
19778 ++ * V0.01 basic helper
19779 ++ *
19780 ++ */
19781 ++
19782 ++#include <linux/kmod.h>
19783 ++#include <linux/reboot.h>
19784 ++#include <linux/vs_context.h>
19785 ++#include <linux/vs_network.h>
19786 ++#include <linux/vserver/signal.h>
19787 ++
19788 ++
19789 ++char vshelper_path[255] = "/sbin/vshelper";
19790 ++
19791 ++
19792 ++static int do_vshelper(char *name, char *argv[], char *envp[], int sync)
19793 ++{
19794 ++ int ret;
19795 ++
19796 ++ if ((ret = call_usermodehelper(name, argv, envp, sync))) {
19797 ++ printk( KERN_WARNING
19798 ++ "%s: (%s %s) returned %s with %d\n",
19799 ++ name, argv[1], argv[2],
19800 ++ sync ? "sync" : "async", ret);
19801 ++ }
19802 ++ vxdprintk(VXD_CBIT(switch, 4),
19803 ++ "%s: (%s %s) returned %s with %d",
19804 ++ name, argv[1], argv[2], sync ? "sync" : "async", ret);
19805 ++ return ret;
19806 ++}
19807 ++
19808 ++/*
19809 ++ * vshelper path is set via /proc/sys
19810 ++ * invoked by vserver sys_reboot(), with
19811 ++ * the following arguments
19812 ++ *
19813 ++ * argv [0] = vshelper_path;
19814 ++ * argv [1] = action: "restart", "halt", "poweroff", ...
19815 ++ * argv [2] = context identifier
19816 ++ *
19817 ++ * envp [*] = type-specific parameters
19818 ++ */
19819 ++
19820 ++long vs_reboot_helper(struct vx_info *vxi, int cmd, void __user *arg)
19821 ++{
19822 ++ char id_buf[8], cmd_buf[16];
19823 ++ char uid_buf[16], pid_buf[16];
19824 ++ int ret;
19825 ++
19826 ++ char *argv[] = {vshelper_path, NULL, id_buf, 0};
19827 ++ char *envp[] = {"HOME=/", "TERM=linux",
19828 ++ "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
19829 ++ uid_buf, pid_buf, cmd_buf, 0};
19830 ++
19831 ++ if (vx_info_state(vxi, VXS_HELPER))
19832 ++ return -EAGAIN;
19833 ++ vxi->vx_state |= VXS_HELPER;
19834 ++
19835 ++ snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
19836 ++
19837 ++ snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19838 ++ snprintf(uid_buf, sizeof(uid_buf)-1, "VS_UID=%d", current_uid());
19839 ++ snprintf(pid_buf, sizeof(pid_buf)-1, "VS_PID=%d", current->pid);
19840 ++
19841 ++ switch (cmd) {
19842 ++ case LINUX_REBOOT_CMD_RESTART:
19843 ++ argv[1] = "restart";
19844 ++ break;
19845 ++
19846 ++ case LINUX_REBOOT_CMD_HALT:
19847 ++ argv[1] = "halt";
19848 ++ break;
19849 ++
19850 ++ case LINUX_REBOOT_CMD_POWER_OFF:
19851 ++ argv[1] = "poweroff";
19852 ++ break;
19853 ++
19854 ++ case LINUX_REBOOT_CMD_SW_SUSPEND:
19855 ++ argv[1] = "swsusp";
19856 ++ break;
19857 ++
19858 ++ default:
19859 ++ vxi->vx_state &= ~VXS_HELPER;
19860 ++ return 0;
19861 ++ }
19862 ++
19863 ++ ret = do_vshelper(vshelper_path, argv, envp, 0);
19864 ++ vxi->vx_state &= ~VXS_HELPER;
19865 ++ __wakeup_vx_info(vxi);
19866 ++ return (ret) ? -EPERM : 0;
19867 ++}
19868 ++
19869 ++
19870 ++long vs_reboot(unsigned int cmd, void __user *arg)
19871 ++{
19872 ++ struct vx_info *vxi = current->vx_info;
19873 ++ long ret = 0;
19874 ++
19875 ++ vxdprintk(VXD_CBIT(misc, 5),
19876 ++ "vs_reboot(%p[#%d],%d)",
19877 ++ vxi, vxi ? vxi->vx_id : 0, cmd);
19878 ++
19879 ++ ret = vs_reboot_helper(vxi, cmd, arg);
19880 ++ if (ret)
19881 ++ return ret;
19882 ++
19883 ++ vxi->reboot_cmd = cmd;
19884 ++ if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
19885 ++ switch (cmd) {
19886 ++ case LINUX_REBOOT_CMD_RESTART:
19887 ++ case LINUX_REBOOT_CMD_HALT:
19888 ++ case LINUX_REBOOT_CMD_POWER_OFF:
19889 ++ vx_info_kill(vxi, 0, SIGKILL);
19890 ++ vx_info_kill(vxi, 1, SIGKILL);
19891 ++ default:
19892 ++ break;
19893 ++ }
19894 ++ }
19895 ++ return 0;
19896 ++}
19897 ++
19898 ++
19899 ++/*
19900 ++ * argv [0] = vshelper_path;
19901 ++ * argv [1] = action: "startup", "shutdown"
19902 ++ * argv [2] = context identifier
19903 ++ *
19904 ++ * envp [*] = type-specific parameters
19905 ++ */
19906 ++
19907 ++long vs_state_change(struct vx_info *vxi, unsigned int cmd)
19908 ++{
19909 ++ char id_buf[8], cmd_buf[16];
19910 ++ char *argv[] = {vshelper_path, NULL, id_buf, 0};
19911 ++ char *envp[] = {"HOME=/", "TERM=linux",
19912 ++ "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19913 ++
19914 ++ if (!vx_info_flags(vxi, VXF_SC_HELPER, 0))
19915 ++ return 0;
19916 ++
19917 ++ snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
19918 ++ snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19919 ++
19920 ++ switch (cmd) {
19921 ++ case VSC_STARTUP:
19922 ++ argv[1] = "startup";
19923 ++ break;
19924 ++ case VSC_SHUTDOWN:
19925 ++ argv[1] = "shutdown";
19926 ++ break;
19927 ++ default:
19928 ++ return 0;
19929 ++ }
19930 ++
19931 ++ return do_vshelper(vshelper_path, argv, envp, 1);
19932 ++}
19933 ++
19934 ++
19935 ++/*
19936 ++ * argv [0] = vshelper_path;
19937 ++ * argv [1] = action: "netup", "netdown"
19938 ++ * argv [2] = context identifier
19939 ++ *
19940 ++ * envp [*] = type-specific parameters
19941 ++ */
19942 ++
19943 ++long vs_net_change(struct nx_info *nxi, unsigned int cmd)
19944 ++{
19945 ++ char id_buf[8], cmd_buf[16];
19946 ++ char *argv[] = {vshelper_path, NULL, id_buf, 0};
19947 ++ char *envp[] = {"HOME=/", "TERM=linux",
19948 ++ "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19949 ++
19950 ++ if (!nx_info_flags(nxi, NXF_SC_HELPER, 0))
19951 ++ return 0;
19952 ++
19953 ++ snprintf(id_buf, sizeof(id_buf)-1, "%d", nxi->nx_id);
19954 ++ snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19955 ++
19956 ++ switch (cmd) {
19957 ++ case VSC_NETUP:
19958 ++ argv[1] = "netup";
19959 ++ break;
19960 ++ case VSC_NETDOWN:
19961 ++ argv[1] = "netdown";
19962 ++ break;
19963 ++ default:
19964 ++ return 0;
19965 ++ }
19966 ++
19967 ++ return do_vshelper(vshelper_path, argv, envp, 1);
19968 ++}
19969 ++
19970 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/history.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/history.c
19971 +--- linux-2.6.29.4/kernel/vserver/history.c 1970-01-01 01:00:00.000000000 +0100
19972 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/history.c 2009-02-22 22:54:26.000000000 +0100
19973 +@@ -0,0 +1,258 @@
19974 ++/*
19975 ++ * kernel/vserver/history.c
19976 ++ *
19977 ++ * Virtual Context History Backtrace
19978 ++ *
19979 ++ * Copyright (C) 2004-2007 Herbert Pötzl
19980 ++ *
19981 ++ * V0.01 basic structure
19982 ++ * V0.02 hash/unhash and trace
19983 ++ * V0.03 preemption fixes
19984 ++ *
19985 ++ */
19986 ++
19987 ++#include <linux/module.h>
19988 ++#include <asm/uaccess.h>
19989 ++
19990 ++#include <linux/vserver/context.h>
19991 ++#include <linux/vserver/debug.h>
19992 ++#include <linux/vserver/debug_cmd.h>
19993 ++#include <linux/vserver/history.h>
19994 ++
19995 ++
19996 ++#ifdef CONFIG_VSERVER_HISTORY
19997 ++#define VXH_SIZE CONFIG_VSERVER_HISTORY_SIZE
19998 ++#else
19999 ++#define VXH_SIZE 64
20000 ++#endif
20001 ++
20002 ++struct _vx_history {
20003 ++ unsigned int counter;
20004 ++
20005 ++ struct _vx_hist_entry entry[VXH_SIZE + 1];
20006 ++};
20007 ++
20008 ++
20009 ++DEFINE_PER_CPU(struct _vx_history, vx_history_buffer);
20010 ++
20011 ++unsigned volatile int vxh_active = 1;
20012 ++
20013 ++static atomic_t sequence = ATOMIC_INIT(0);
20014 ++
20015 ++
20016 ++/* vxh_advance()
20017 ++
20018 ++ * requires disabled preemption */
20019 ++
20020 ++struct _vx_hist_entry *vxh_advance(void *loc)
20021 ++{
20022 ++ unsigned int cpu = smp_processor_id();
20023 ++ struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
20024 ++ struct _vx_hist_entry *entry;
20025 ++ unsigned int index;
20026 ++
20027 ++ index = vxh_active ? (hist->counter++ % VXH_SIZE) : VXH_SIZE;
20028 ++ entry = &hist->entry[index];
20029 ++
20030 ++ entry->seq = atomic_inc_return(&sequence);
20031 ++ entry->loc = loc;
20032 ++ return entry;
20033 ++}
20034 ++
20035 ++EXPORT_SYMBOL_GPL(vxh_advance);
20036 ++
20037 ++
20038 ++#define VXH_LOC_FMTS "(#%04x,*%d):%p"
20039 ++
20040 ++#define VXH_LOC_ARGS(e) (e)->seq, cpu, (e)->loc
20041 ++
20042 ++
20043 ++#define VXH_VXI_FMTS "%p[#%d,%d.%d]"
20044 ++
20045 ++#define VXH_VXI_ARGS(e) (e)->vxi.ptr, \
20046 ++ (e)->vxi.ptr ? (e)->vxi.xid : 0, \
20047 ++ (e)->vxi.ptr ? (e)->vxi.usecnt : 0, \
20048 ++ (e)->vxi.ptr ? (e)->vxi.tasks : 0
20049 ++
20050 ++void vxh_dump_entry(struct _vx_hist_entry *e, unsigned cpu)
20051 ++{
20052 ++ switch (e->type) {
20053 ++ case VXH_THROW_OOPS:
20054 ++ printk( VXH_LOC_FMTS " oops \n", VXH_LOC_ARGS(e));
20055 ++ break;
20056 ++
20057 ++ case VXH_GET_VX_INFO:
20058 ++ case VXH_PUT_VX_INFO:
20059 ++ printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
20060 ++ VXH_LOC_ARGS(e),
20061 ++ (e->type == VXH_GET_VX_INFO) ? "get" : "put",
20062 ++ VXH_VXI_ARGS(e));
20063 ++ break;
20064 ++
20065 ++ case VXH_INIT_VX_INFO:
20066 ++ case VXH_SET_VX_INFO:
20067 ++ case VXH_CLR_VX_INFO:
20068 ++ printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
20069 ++ VXH_LOC_ARGS(e),
20070 ++ (e->type == VXH_INIT_VX_INFO) ? "init" :
20071 ++ ((e->type == VXH_SET_VX_INFO) ? "set" : "clr"),
20072 ++ VXH_VXI_ARGS(e), e->sc.data);
20073 ++ break;
20074 ++
20075 ++ case VXH_CLAIM_VX_INFO:
20076 ++ case VXH_RELEASE_VX_INFO:
20077 ++ printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
20078 ++ VXH_LOC_ARGS(e),
20079 ++ (e->type == VXH_CLAIM_VX_INFO) ? "claim" : "release",
20080 ++ VXH_VXI_ARGS(e), e->sc.data);
20081 ++ break;
20082 ++
20083 ++ case VXH_ALLOC_VX_INFO:
20084 ++ case VXH_DEALLOC_VX_INFO:
20085 ++ printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
20086 ++ VXH_LOC_ARGS(e),
20087 ++ (e->type == VXH_ALLOC_VX_INFO) ? "alloc" : "dealloc",
20088 ++ VXH_VXI_ARGS(e));
20089 ++ break;
20090 ++
20091 ++ case VXH_HASH_VX_INFO:
20092 ++ case VXH_UNHASH_VX_INFO:
20093 ++ printk( VXH_LOC_FMTS " __%s_vx_info " VXH_VXI_FMTS "\n",
20094 ++ VXH_LOC_ARGS(e),
20095 ++ (e->type == VXH_HASH_VX_INFO) ? "hash" : "unhash",
20096 ++ VXH_VXI_ARGS(e));
20097 ++ break;
20098 ++
20099 ++ case VXH_LOC_VX_INFO:
20100 ++ case VXH_LOOKUP_VX_INFO:
20101 ++ case VXH_CREATE_VX_INFO:
20102 ++ printk( VXH_LOC_FMTS " __%s_vx_info [#%d] -> " VXH_VXI_FMTS "\n",
20103 ++ VXH_LOC_ARGS(e),
20104 ++ (e->type == VXH_CREATE_VX_INFO) ? "create" :
20105 ++ ((e->type == VXH_LOC_VX_INFO) ? "loc" : "lookup"),
20106 ++ e->ll.arg, VXH_VXI_ARGS(e));
20107 ++ break;
20108 ++ }
20109 ++}
20110 ++
20111 ++static void __vxh_dump_history(void)
20112 ++{
20113 ++ unsigned int i, cpu;
20114 ++
20115 ++ printk("History:\tSEQ: %8x\tNR_CPUS: %d\n",
20116 ++ atomic_read(&sequence), NR_CPUS);
20117 ++
20118 ++ for (i = 0; i < VXH_SIZE; i++) {
20119 ++ for_each_online_cpu(cpu) {
20120 ++ struct _vx_history *hist =
20121 ++ &per_cpu(vx_history_buffer, cpu);
20122 ++ unsigned int index = (hist->counter - i) % VXH_SIZE;
20123 ++ struct _vx_hist_entry *entry = &hist->entry[index];
20124 ++
20125 ++ vxh_dump_entry(entry, cpu);
20126 ++ }
20127 ++ }
20128 ++}
20129 ++
20130 ++void vxh_dump_history(void)
20131 ++{
20132 ++ vxh_active = 0;
20133 ++#ifdef CONFIG_SMP
20134 ++ local_irq_enable();
20135 ++ smp_send_stop();
20136 ++ local_irq_disable();
20137 ++#endif
20138 ++ __vxh_dump_history();
20139 ++}
20140 ++
20141 ++
20142 ++/* vserver syscall commands below here */
20143 ++
20144 ++
20145 ++int vc_dump_history(uint32_t id)
20146 ++{
20147 ++ vxh_active = 0;
20148 ++ __vxh_dump_history();
20149 ++ vxh_active = 1;
20150 ++
20151 ++ return 0;
20152 ++}
20153 ++
20154 ++
20155 ++int do_read_history(struct __user _vx_hist_entry *data,
20156 ++ int cpu, uint32_t *index, uint32_t *count)
20157 ++{
20158 ++ int pos, ret = 0;
20159 ++ struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
20160 ++ int end = hist->counter;
20161 ++ int start = end - VXH_SIZE + 2;
20162 ++ int idx = *index;
20163 ++
20164 ++ /* special case: get current pos */
20165 ++ if (!*count) {
20166 ++ *index = end;
20167 ++ return 0;
20168 ++ }
20169 ++
20170 ++ /* have we lost some data? */
20171 ++ if (idx < start)
20172 ++ idx = start;
20173 ++
20174 ++ for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
20175 ++ struct _vx_hist_entry *entry =
20176 ++ &hist->entry[idx % VXH_SIZE];
20177 ++
20178 ++ /* send entry to userspace */
20179 ++ ret = copy_to_user(&data[pos], entry, sizeof(*entry));
20180 ++ if (ret)
20181 ++ break;
20182 ++ }
20183 ++ /* save new index and count */
20184 ++ *index = idx;
20185 ++ *count = pos;
20186 ++ return ret ? ret : (*index < end);
20187 ++}
20188 ++
20189 ++int vc_read_history(uint32_t id, void __user *data)
20190 ++{
20191 ++ struct vcmd_read_history_v0 vc_data;
20192 ++ int ret;
20193 ++
20194 ++ if (id >= NR_CPUS)
20195 ++ return -EINVAL;
20196 ++
20197 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20198 ++ return -EFAULT;
20199 ++
20200 ++ ret = do_read_history((struct __user _vx_hist_entry *)vc_data.data,
20201 ++ id, &vc_data.index, &vc_data.count);
20202 ++
20203 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20204 ++ return -EFAULT;
20205 ++ return ret;
20206 ++}
20207 ++
20208 ++#ifdef CONFIG_COMPAT
20209 ++
20210 ++int vc_read_history_x32(uint32_t id, void __user *data)
20211 ++{
20212 ++ struct vcmd_read_history_v0_x32 vc_data;
20213 ++ int ret;
20214 ++
20215 ++ if (id >= NR_CPUS)
20216 ++ return -EINVAL;
20217 ++
20218 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20219 ++ return -EFAULT;
20220 ++
20221 ++ ret = do_read_history((struct __user _vx_hist_entry *)
20222 ++ compat_ptr(vc_data.data_ptr),
20223 ++ id, &vc_data.index, &vc_data.count);
20224 ++
20225 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20226 ++ return -EFAULT;
20227 ++ return ret;
20228 ++}
20229 ++
20230 ++#endif /* CONFIG_COMPAT */
20231 ++
20232 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/inet.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/inet.c
20233 +--- linux-2.6.29.4/kernel/vserver/inet.c 1970-01-01 01:00:00.000000000 +0100
20234 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/inet.c 2009-02-22 22:54:26.000000000 +0100
20235 +@@ -0,0 +1,225 @@
20236 ++
20237 ++#include <linux/in.h>
20238 ++#include <linux/inetdevice.h>
20239 ++#include <linux/vs_inet.h>
20240 ++#include <linux/vs_inet6.h>
20241 ++#include <linux/vserver/debug.h>
20242 ++#include <net/route.h>
20243 ++#include <net/addrconf.h>
20244 ++
20245 ++
20246 ++int nx_v4_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20247 ++{
20248 ++ int ret = 0;
20249 ++
20250 ++ if (!nxi1 || !nxi2 || nxi1 == nxi2)
20251 ++ ret = 1;
20252 ++ else {
20253 ++ struct nx_addr_v4 *ptr;
20254 ++
20255 ++ for (ptr = &nxi1->v4; ptr; ptr = ptr->next) {
20256 ++ if (v4_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20257 ++ ret = 1;
20258 ++ break;
20259 ++ }
20260 ++ }
20261 ++ }
20262 ++
20263 ++ vxdprintk(VXD_CBIT(net, 2),
20264 ++ "nx_v4_addr_conflict(%p,%p): %d",
20265 ++ nxi1, nxi2, ret);
20266 ++
20267 ++ return ret;
20268 ++}
20269 ++
20270 ++
20271 ++#ifdef CONFIG_IPV6
20272 ++
20273 ++int nx_v6_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
20274 ++{
20275 ++ int ret = 0;
20276 ++
20277 ++ if (!nxi1 || !nxi2 || nxi1 == nxi2)
20278 ++ ret = 1;
20279 ++ else {
20280 ++ struct nx_addr_v6 *ptr;
20281 ++
20282 ++ for (ptr = &nxi1->v6; ptr; ptr = ptr->next) {
20283 ++ if (v6_nx_addr_in_nx_info(nxi2, ptr, -1)) {
20284 ++ ret = 1;
20285 ++ break;
20286 ++ }
20287 ++ }
20288 ++ }
20289 ++
20290 ++ vxdprintk(VXD_CBIT(net, 2),
20291 ++ "nx_v6_addr_conflict(%p,%p): %d",
20292 ++ nxi1, nxi2, ret);
20293 ++
20294 ++ return ret;
20295 ++}
20296 ++
20297 ++#endif
20298 ++
20299 ++int v4_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20300 ++{
20301 ++ struct in_device *in_dev;
20302 ++ struct in_ifaddr **ifap;
20303 ++ struct in_ifaddr *ifa;
20304 ++ int ret = 0;
20305 ++
20306 ++ if (!dev)
20307 ++ goto out;
20308 ++ in_dev = in_dev_get(dev);
20309 ++ if (!in_dev)
20310 ++ goto out;
20311 ++
20312 ++ for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
20313 ++ ifap = &ifa->ifa_next) {
20314 ++ if (v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW)) {
20315 ++ ret = 1;
20316 ++ break;
20317 ++ }
20318 ++ }
20319 ++ in_dev_put(in_dev);
20320 ++out:
20321 ++ return ret;
20322 ++}
20323 ++
20324 ++
20325 ++#ifdef CONFIG_IPV6
20326 ++
20327 ++int v6_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20328 ++{
20329 ++ struct inet6_dev *in_dev;
20330 ++ struct inet6_ifaddr **ifap;
20331 ++ struct inet6_ifaddr *ifa;
20332 ++ int ret = 0;
20333 ++
20334 ++ if (!dev)
20335 ++ goto out;
20336 ++ in_dev = in6_dev_get(dev);
20337 ++ if (!in_dev)
20338 ++ goto out;
20339 ++
20340 ++ for (ifap = &in_dev->addr_list; (ifa = *ifap) != NULL;
20341 ++ ifap = &ifa->if_next) {
20342 ++ if (v6_addr_in_nx_info(nxi, &ifa->addr, -1)) {
20343 ++ ret = 1;
20344 ++ break;
20345 ++ }
20346 ++ }
20347 ++ in6_dev_put(in_dev);
20348 ++out:
20349 ++ return ret;
20350 ++}
20351 ++
20352 ++#endif
20353 ++
20354 ++int dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
20355 ++{
20356 ++ int ret = 1;
20357 ++
20358 ++ if (!nxi)
20359 ++ goto out;
20360 ++ if (nxi->v4.type && v4_dev_in_nx_info(dev, nxi))
20361 ++ goto out;
20362 ++#ifdef CONFIG_IPV6
20363 ++ ret = 2;
20364 ++ if (nxi->v6.type && v6_dev_in_nx_info(dev, nxi))
20365 ++ goto out;
20366 ++#endif
20367 ++ ret = 0;
20368 ++out:
20369 ++ vxdprintk(VXD_CBIT(net, 3),
20370 ++ "dev_in_nx_info(%p,%p[#%d]) = %d",
20371 ++ dev, nxi, nxi ? nxi->nx_id : 0, ret);
20372 ++ return ret;
20373 ++}
20374 ++
20375 ++int ip_v4_find_src(struct net *net, struct nx_info *nxi,
20376 ++ struct rtable **rp, struct flowi *fl)
20377 ++{
20378 ++ if (!nxi)
20379 ++ return 0;
20380 ++
20381 ++ /* FIXME: handle lback only case */
20382 ++ if (!NX_IPV4(nxi))
20383 ++ return -EPERM;
20384 ++
20385 ++ vxdprintk(VXD_CBIT(net, 4),
20386 ++ "ip_v4_find_src(%p[#%u]) " NIPQUAD_FMT " -> " NIPQUAD_FMT,
20387 ++ nxi, nxi ? nxi->nx_id : 0,
20388 ++ NIPQUAD(fl->fl4_src), NIPQUAD(fl->fl4_dst));
20389 ++
20390 ++ /* single IP is unconditional */
20391 ++ if (nx_info_flags(nxi, NXF_SINGLE_IP, 0) &&
20392 ++ (fl->fl4_src == INADDR_ANY))
20393 ++ fl->fl4_src = nxi->v4.ip[0].s_addr;
20394 ++
20395 ++ if (fl->fl4_src == INADDR_ANY) {
20396 ++ struct nx_addr_v4 *ptr;
20397 ++ __be32 found = 0;
20398 ++ int err;
20399 ++
20400 ++ err = __ip_route_output_key(net, rp, fl);
20401 ++ if (!err) {
20402 ++ found = (*rp)->rt_src;
20403 ++ ip_rt_put(*rp);
20404 ++ vxdprintk(VXD_CBIT(net, 4),
20405 ++ "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20406 ++ nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(found));
20407 ++ if (v4_addr_in_nx_info(nxi, found, NXA_MASK_BIND))
20408 ++ goto found;
20409 ++ }
20410 ++
20411 ++ for (ptr = &nxi->v4; ptr; ptr = ptr->next) {
20412 ++ __be32 primary = ptr->ip[0].s_addr;
20413 ++ __be32 mask = ptr->mask.s_addr;
20414 ++ __be32 neta = primary & mask;
20415 ++
20416 ++ vxdprintk(VXD_CBIT(net, 4), "ip_v4_find_src(%p[#%u]) chk: "
20417 ++ NIPQUAD_FMT "/" NIPQUAD_FMT "/" NIPQUAD_FMT,
20418 ++ nxi, nxi ? nxi->nx_id : 0, NIPQUAD(primary),
20419 ++ NIPQUAD(mask), NIPQUAD(neta));
20420 ++ if ((found & mask) != neta)
20421 ++ continue;
20422 ++
20423 ++ fl->fl4_src = primary;
20424 ++ err = __ip_route_output_key(net, rp, fl);
20425 ++ vxdprintk(VXD_CBIT(net, 4),
20426 ++ "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
20427 ++ nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(primary));
20428 ++ if (!err) {
20429 ++ found = (*rp)->rt_src;
20430 ++ ip_rt_put(*rp);
20431 ++ if (found == primary)
20432 ++ goto found;
20433 ++ }
20434 ++ }
20435 ++ /* still no source ip? */
20436 ++ found = ipv4_is_loopback(fl->fl4_dst)
20437 ++ ? IPI_LOOPBACK : nxi->v4.ip[0].s_addr;
20438 ++ found:
20439 ++ /* assign src ip to flow */
20440 ++ fl->fl4_src = found;
20441 ++
20442 ++ } else {
20443 ++ if (!v4_addr_in_nx_info(nxi, fl->fl4_src, NXA_MASK_BIND))
20444 ++ return -EPERM;
20445 ++ }
20446 ++
20447 ++ if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0)) {
20448 ++ if (ipv4_is_loopback(fl->fl4_dst))
20449 ++ fl->fl4_dst = nxi->v4_lback.s_addr;
20450 ++ if (ipv4_is_loopback(fl->fl4_src))
20451 ++ fl->fl4_src = nxi->v4_lback.s_addr;
20452 ++ } else if (ipv4_is_loopback(fl->fl4_dst) &&
20453 ++ !nx_info_flags(nxi, NXF_LBACK_ALLOW, 0))
20454 ++ return -EPERM;
20455 ++
20456 ++ return 0;
20457 ++}
20458 ++
20459 ++EXPORT_SYMBOL_GPL(ip_v4_find_src);
20460 ++
20461 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/init.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/init.c
20462 +--- linux-2.6.29.4/kernel/vserver/init.c 1970-01-01 01:00:00.000000000 +0100
20463 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/init.c 2009-02-22 22:54:26.000000000 +0100
20464 +@@ -0,0 +1,45 @@
20465 ++/*
20466 ++ * linux/kernel/init.c
20467 ++ *
20468 ++ * Virtual Server Init
20469 ++ *
20470 ++ * Copyright (C) 2004-2007 Herbert Pötzl
20471 ++ *
20472 ++ * V0.01 basic structure
20473 ++ *
20474 ++ */
20475 ++
20476 ++#include <linux/init.h>
20477 ++
20478 ++int vserver_register_sysctl(void);
20479 ++void vserver_unregister_sysctl(void);
20480 ++
20481 ++
20482 ++static int __init init_vserver(void)
20483 ++{
20484 ++ int ret = 0;
20485 ++
20486 ++#ifdef CONFIG_VSERVER_DEBUG
20487 ++ vserver_register_sysctl();
20488 ++#endif
20489 ++ return ret;
20490 ++}
20491 ++
20492 ++
20493 ++static void __exit exit_vserver(void)
20494 ++{
20495 ++
20496 ++#ifdef CONFIG_VSERVER_DEBUG
20497 ++ vserver_unregister_sysctl();
20498 ++#endif
20499 ++ return;
20500 ++}
20501 ++
20502 ++/* FIXME: GFP_ZONETYPES gone
20503 ++long vx_slab[GFP_ZONETYPES]; */
20504 ++long vx_area;
20505 ++
20506 ++
20507 ++module_init(init_vserver);
20508 ++module_exit(exit_vserver);
20509 ++
20510 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/inode.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/inode.c
20511 +--- linux-2.6.29.4/kernel/vserver/inode.c 1970-01-01 01:00:00.000000000 +0100
20512 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/inode.c 2009-02-22 22:54:26.000000000 +0100
20513 +@@ -0,0 +1,422 @@
20514 ++/*
20515 ++ * linux/kernel/vserver/inode.c
20516 ++ *
20517 ++ * Virtual Server: File System Support
20518 ++ *
20519 ++ * Copyright (C) 2004-2007 Herbert Pötzl
20520 ++ *
20521 ++ * V0.01 separated from vcontext V0.05
20522 ++ * V0.02 moved to tag (instead of xid)
20523 ++ *
20524 ++ */
20525 ++
20526 ++#include <linux/tty.h>
20527 ++#include <linux/proc_fs.h>
20528 ++#include <linux/devpts_fs.h>
20529 ++#include <linux/fs.h>
20530 ++#include <linux/file.h>
20531 ++#include <linux/mount.h>
20532 ++#include <linux/parser.h>
20533 ++#include <linux/namei.h>
20534 ++#include <linux/vserver/inode.h>
20535 ++#include <linux/vserver/inode_cmd.h>
20536 ++#include <linux/vs_base.h>
20537 ++#include <linux/vs_tag.h>
20538 ++
20539 ++#include <asm/uaccess.h>
20540 ++
20541 ++
20542 ++static int __vc_get_iattr(struct inode *in, uint32_t *tag, uint32_t *flags, uint32_t *mask)
20543 ++{
20544 ++ struct proc_dir_entry *entry;
20545 ++
20546 ++ if (!in || !in->i_sb)
20547 ++ return -ESRCH;
20548 ++
20549 ++ *flags = IATTR_TAG
20550 ++ | (IS_BARRIER(in) ? IATTR_BARRIER : 0)
20551 ++ | (IS_IXUNLINK(in) ? IATTR_IXUNLINK : 0)
20552 ++ | (IS_IMMUTABLE(in) ? IATTR_IMMUTABLE : 0);
20553 ++ *mask = IATTR_IXUNLINK | IATTR_IMMUTABLE;
20554 ++
20555 ++ if (S_ISDIR(in->i_mode))
20556 ++ *mask |= IATTR_BARRIER;
20557 ++
20558 ++ if (IS_TAGGED(in)) {
20559 ++ *tag = in->i_tag;
20560 ++ *mask |= IATTR_TAG;
20561 ++ }
20562 ++
20563 ++ switch (in->i_sb->s_magic) {
20564 ++ case PROC_SUPER_MAGIC:
20565 ++ entry = PROC_I(in)->pde;
20566 ++
20567 ++ /* check for specific inodes? */
20568 ++ if (entry)
20569 ++ *mask |= IATTR_FLAGS;
20570 ++ if (entry)
20571 ++ *flags |= (entry->vx_flags & IATTR_FLAGS);
20572 ++ else
20573 ++ *flags |= (PROC_I(in)->vx_flags & IATTR_FLAGS);
20574 ++ break;
20575 ++
20576 ++ case DEVPTS_SUPER_MAGIC:
20577 ++ *tag = in->i_tag;
20578 ++ *mask |= IATTR_TAG;
20579 ++ break;
20580 ++
20581 ++ default:
20582 ++ break;
20583 ++ }
20584 ++ return 0;
20585 ++}
20586 ++
20587 ++int vc_get_iattr(void __user *data)
20588 ++{
20589 ++ struct path path;
20590 ++ struct vcmd_ctx_iattr_v1 vc_data = { .tag = -1 };
20591 ++ int ret;
20592 ++
20593 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20594 ++ return -EFAULT;
20595 ++
20596 ++ ret = user_lpath(vc_data.name, &path);
20597 ++ if (!ret) {
20598 ++ ret = __vc_get_iattr(path.dentry->d_inode,
20599 ++ &vc_data.tag, &vc_data.flags, &vc_data.mask);
20600 ++ path_put(&path);
20601 ++ }
20602 ++ if (ret)
20603 ++ return ret;
20604 ++
20605 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20606 ++ ret = -EFAULT;
20607 ++ return ret;
20608 ++}
20609 ++
20610 ++#ifdef CONFIG_COMPAT
20611 ++
20612 ++int vc_get_iattr_x32(void __user *data)
20613 ++{
20614 ++ struct path path;
20615 ++ struct vcmd_ctx_iattr_v1_x32 vc_data = { .tag = -1 };
20616 ++ int ret;
20617 ++
20618 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20619 ++ return -EFAULT;
20620 ++
20621 ++ ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
20622 ++ if (!ret) {
20623 ++ ret = __vc_get_iattr(path.dentry->d_inode,
20624 ++ &vc_data.tag, &vc_data.flags, &vc_data.mask);
20625 ++ path_put(&path);
20626 ++ }
20627 ++ if (ret)
20628 ++ return ret;
20629 ++
20630 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20631 ++ ret = -EFAULT;
20632 ++ return ret;
20633 ++}
20634 ++
20635 ++#endif /* CONFIG_COMPAT */
20636 ++
20637 ++
20638 ++int vc_fget_iattr(uint32_t fd, void __user *data)
20639 ++{
20640 ++ struct file *filp;
20641 ++ struct vcmd_ctx_fiattr_v0 vc_data = { .tag = -1 };
20642 ++ int ret;
20643 ++
20644 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20645 ++ return -EFAULT;
20646 ++
20647 ++ filp = fget(fd);
20648 ++ if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
20649 ++ return -EBADF;
20650 ++
20651 ++ ret = __vc_get_iattr(filp->f_dentry->d_inode,
20652 ++ &vc_data.tag, &vc_data.flags, &vc_data.mask);
20653 ++
20654 ++ fput(filp);
20655 ++
20656 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20657 ++ ret = -EFAULT;
20658 ++ return ret;
20659 ++}
20660 ++
20661 ++
20662 ++static int __vc_set_iattr(struct dentry *de, uint32_t *tag, uint32_t *flags, uint32_t *mask)
20663 ++{
20664 ++ struct inode *in = de->d_inode;
20665 ++ int error = 0, is_proc = 0, has_tag = 0;
20666 ++ struct iattr attr = { 0 };
20667 ++
20668 ++ if (!in || !in->i_sb)
20669 ++ return -ESRCH;
20670 ++
20671 ++ is_proc = (in->i_sb->s_magic == PROC_SUPER_MAGIC);
20672 ++ if ((*mask & IATTR_FLAGS) && !is_proc)
20673 ++ return -EINVAL;
20674 ++
20675 ++ has_tag = IS_TAGGED(in) ||
20676 ++ (in->i_sb->s_magic == DEVPTS_SUPER_MAGIC);
20677 ++ if ((*mask & IATTR_TAG) && !has_tag)
20678 ++ return -EINVAL;
20679 ++
20680 ++ mutex_lock(&in->i_mutex);
20681 ++ if (*mask & IATTR_TAG) {
20682 ++ attr.ia_tag = *tag;
20683 ++ attr.ia_valid |= ATTR_TAG;
20684 ++ }
20685 ++
20686 ++ if (*mask & IATTR_FLAGS) {
20687 ++ struct proc_dir_entry *entry = PROC_I(in)->pde;
20688 ++ unsigned int iflags = PROC_I(in)->vx_flags;
20689 ++
20690 ++ iflags = (iflags & ~(*mask & IATTR_FLAGS))
20691 ++ | (*flags & IATTR_FLAGS);
20692 ++ PROC_I(in)->vx_flags = iflags;
20693 ++ if (entry)
20694 ++ entry->vx_flags = iflags;
20695 ++ }
20696 ++
20697 ++ if (*mask & (IATTR_BARRIER | IATTR_IXUNLINK | IATTR_IMMUTABLE)) {
20698 ++ if (*mask & IATTR_IMMUTABLE) {
20699 ++ if (*flags & IATTR_IMMUTABLE)
20700 ++ in->i_flags |= S_IMMUTABLE;
20701 ++ else
20702 ++ in->i_flags &= ~S_IMMUTABLE;
20703 ++ }
20704 ++ if (*mask & IATTR_IXUNLINK) {
20705 ++ if (*flags & IATTR_IXUNLINK)
20706 ++ in->i_flags |= S_IXUNLINK;
20707 ++ else
20708 ++ in->i_flags &= ~S_IXUNLINK;
20709 ++ }
20710 ++ if (S_ISDIR(in->i_mode) && (*mask & IATTR_BARRIER)) {
20711 ++ if (*flags & IATTR_BARRIER)
20712 ++ in->i_vflags |= V_BARRIER;
20713 ++ else
20714 ++ in->i_vflags &= ~V_BARRIER;
20715 ++ }
20716 ++ if (in->i_op && in->i_op->sync_flags) {
20717 ++ error = in->i_op->sync_flags(in);
20718 ++ if (error)
20719 ++ goto out;
20720 ++ }
20721 ++ }
20722 ++
20723 ++ if (attr.ia_valid) {
20724 ++ if (in->i_op && in->i_op->setattr)
20725 ++ error = in->i_op->setattr(de, &attr);
20726 ++ else {
20727 ++ error = inode_change_ok(in, &attr);
20728 ++ if (!error)
20729 ++ error = inode_setattr(in, &attr);
20730 ++ }
20731 ++ }
20732 ++
20733 ++out:
20734 ++ mutex_unlock(&in->i_mutex);
20735 ++ return error;
20736 ++}
20737 ++
20738 ++int vc_set_iattr(void __user *data)
20739 ++{
20740 ++ struct path path;
20741 ++ struct vcmd_ctx_iattr_v1 vc_data;
20742 ++ int ret;
20743 ++
20744 ++ if (!capable(CAP_LINUX_IMMUTABLE))
20745 ++ return -EPERM;
20746 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20747 ++ return -EFAULT;
20748 ++
20749 ++ ret = user_lpath(vc_data.name, &path);
20750 ++ if (!ret) {
20751 ++ ret = __vc_set_iattr(path.dentry,
20752 ++ &vc_data.tag, &vc_data.flags, &vc_data.mask);
20753 ++ path_put(&path);
20754 ++ }
20755 ++
20756 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20757 ++ ret = -EFAULT;
20758 ++ return ret;
20759 ++}
20760 ++
20761 ++#ifdef CONFIG_COMPAT
20762 ++
20763 ++int vc_set_iattr_x32(void __user *data)
20764 ++{
20765 ++ struct path path;
20766 ++ struct vcmd_ctx_iattr_v1_x32 vc_data;
20767 ++ int ret;
20768 ++
20769 ++ if (!capable(CAP_LINUX_IMMUTABLE))
20770 ++ return -EPERM;
20771 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20772 ++ return -EFAULT;
20773 ++
20774 ++ ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
20775 ++ if (!ret) {
20776 ++ ret = __vc_set_iattr(path.dentry,
20777 ++ &vc_data.tag, &vc_data.flags, &vc_data.mask);
20778 ++ path_put(&path);
20779 ++ }
20780 ++
20781 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20782 ++ ret = -EFAULT;
20783 ++ return ret;
20784 ++}
20785 ++
20786 ++#endif /* CONFIG_COMPAT */
20787 ++
20788 ++int vc_fset_iattr(uint32_t fd, void __user *data)
20789 ++{
20790 ++ struct file *filp;
20791 ++ struct vcmd_ctx_fiattr_v0 vc_data;
20792 ++ int ret;
20793 ++
20794 ++ if (!capable(CAP_LINUX_IMMUTABLE))
20795 ++ return -EPERM;
20796 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20797 ++ return -EFAULT;
20798 ++
20799 ++ filp = fget(fd);
20800 ++ if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
20801 ++ return -EBADF;
20802 ++
20803 ++ ret = __vc_set_iattr(filp->f_dentry, &vc_data.tag,
20804 ++ &vc_data.flags, &vc_data.mask);
20805 ++
20806 ++ fput(filp);
20807 ++
20808 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20809 ++ return -EFAULT;
20810 ++ return ret;
20811 ++}
20812 ++
20813 ++
20814 ++enum { Opt_notagcheck, Opt_tag, Opt_notag, Opt_tagid, Opt_err };
20815 ++
20816 ++static match_table_t tokens = {
20817 ++ {Opt_notagcheck, "notagcheck"},
20818 ++#ifdef CONFIG_PROPAGATE
20819 ++ {Opt_notag, "notag"},
20820 ++ {Opt_tag, "tag"},
20821 ++ {Opt_tagid, "tagid=%u"},
20822 ++#endif
20823 ++ {Opt_err, NULL}
20824 ++};
20825 ++
20826 ++
20827 ++static void __dx_parse_remove(char *string, char *opt)
20828 ++{
20829 ++ char *p = strstr(string, opt);
20830 ++ char *q = p;
20831 ++
20832 ++ if (p) {
20833 ++ while (*q != '\0' && *q != ',')
20834 ++ q++;
20835 ++ while (*q)
20836 ++ *p++ = *q++;
20837 ++ while (*p)
20838 ++ *p++ = '\0';
20839 ++ }
20840 ++}
20841 ++
20842 ++int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
20843 ++ unsigned long *flags)
20844 ++{
20845 ++ int set = 0;
20846 ++ substring_t args[MAX_OPT_ARGS];
20847 ++ int token, option = 0;
20848 ++ char *s, *p, *opts;
20849 ++
20850 ++ if (!string)
20851 ++ return 0;
20852 ++ s = kstrdup(string, GFP_KERNEL | GFP_ATOMIC);
20853 ++ if (!s)
20854 ++ return 0;
20855 ++
20856 ++ opts = s;
20857 ++ while ((p = strsep(&opts, ",")) != NULL) {
20858 ++ token = match_token(p, tokens, args);
20859 ++
20860 ++ vxdprintk(VXD_CBIT(tag, 7),
20861 ++ "dx_parse_tag(»%s«): %d:#%d",
20862 ++ p, token, option);
20863 ++
20864 ++ switch (token) {
20865 ++#ifdef CONFIG_PROPAGATE
20866 ++ case Opt_tag:
20867 ++ if (tag)
20868 ++ *tag = 0;
20869 ++ if (remove)
20870 ++ __dx_parse_remove(s, "tag");
20871 ++ *mnt_flags |= MNT_TAGID;
20872 ++ set |= MNT_TAGID;
20873 ++ break;
20874 ++ case Opt_notag:
20875 ++ if (remove)
20876 ++ __dx_parse_remove(s, "notag");
20877 ++ *mnt_flags |= MNT_NOTAG;
20878 ++ set |= MNT_NOTAG;
20879 ++ break;
20880 ++ case Opt_tagid:
20881 ++ if (tag && !match_int(args, &option))
20882 ++ *tag = option;
20883 ++ if (remove)
20884 ++ __dx_parse_remove(s, "tagid");
20885 ++ *mnt_flags |= MNT_TAGID;
20886 ++ set |= MNT_TAGID;
20887 ++ break;
20888 ++#endif
20889 ++ case Opt_notagcheck:
20890 ++ if (remove)
20891 ++ __dx_parse_remove(s, "notagcheck");
20892 ++ *flags |= MS_NOTAGCHECK;
20893 ++ set |= MS_NOTAGCHECK;
20894 ++ break;
20895 ++ }
20896 ++ }
20897 ++ if (set)
20898 ++ strcpy(string, s);
20899 ++ kfree(s);
20900 ++ return set;
20901 ++}
20902 ++
20903 ++#ifdef CONFIG_PROPAGATE
20904 ++
20905 ++void __dx_propagate_tag(struct nameidata *nd, struct inode *inode)
20906 ++{
20907 ++ tag_t new_tag = 0;
20908 ++ struct vfsmount *mnt;
20909 ++ int propagate;
20910 ++
20911 ++ if (!nd)
20912 ++ return;
20913 ++ mnt = nd->path.mnt;
20914 ++ if (!mnt)
20915 ++ return;
20916 ++
20917 ++ propagate = (mnt->mnt_flags & MNT_TAGID);
20918 ++ if (propagate)
20919 ++ new_tag = mnt->mnt_tag;
20920 ++
20921 ++ vxdprintk(VXD_CBIT(tag, 7),
20922 ++ "dx_propagate_tag(%p[#%lu.%d]): %d,%d",
20923 ++ inode, inode->i_ino, inode->i_tag,
20924 ++ new_tag, (propagate) ? 1 : 0);
20925 ++
20926 ++ if (propagate)
20927 ++ inode->i_tag = new_tag;
20928 ++}
20929 ++
20930 ++#include <linux/module.h>
20931 ++
20932 ++EXPORT_SYMBOL_GPL(__dx_propagate_tag);
20933 ++
20934 ++#endif /* CONFIG_PROPAGATE */
20935 ++
20936 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/Kconfig linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/Kconfig
20937 +--- linux-2.6.29.4/kernel/vserver/Kconfig 1970-01-01 01:00:00.000000000 +0100
20938 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/Kconfig 2009-02-22 22:54:26.000000000 +0100
20939 +@@ -0,0 +1,251 @@
20940 ++#
20941 ++# Linux VServer configuration
20942 ++#
20943 ++
20944 ++menu "Linux VServer"
20945 ++
20946 ++config VSERVER_AUTO_LBACK
20947 ++ bool "Automatically Assign Loopback IP"
20948 ++ default y
20949 ++ help
20950 ++ Automatically assign a guest specific loopback
20951 ++ IP and add it to the kernel network stack on
20952 ++ startup.
20953 ++
20954 ++config VSERVER_AUTO_SINGLE
20955 ++ bool "Automatic Single IP Special Casing"
20956 ++ depends on EXPERIMENTAL
20957 ++ default y
20958 ++ help
20959 ++ This allows network contexts with a single IP to
20960 ++ automatically remap 0.0.0.0 bindings to that IP,
20961 ++ avoiding further network checks and improving
20962 ++ performance.
20963 ++
20964 ++ (note: such guests do not allow to change the ip
20965 ++ on the fly and do not show loopback addresses)
20966 ++
20967 ++config VSERVER_COWBL
20968 ++ bool "Enable COW Immutable Link Breaking"
20969 ++ default y
20970 ++ help
20971 ++ This enables the COW (Copy-On-Write) link break code.
20972 ++ It allows you to treat unified files like normal files
20973 ++ when writing to them (which will implicitely break the
20974 ++ link and create a copy of the unified file)
20975 ++
20976 ++config VSERVER_VTIME
20977 ++ bool "Enable Virtualized Guest Time"
20978 ++ depends on EXPERIMENTAL
20979 ++ default n
20980 ++ help
20981 ++ This enables per guest time offsets to allow for
20982 ++ adjusting the system clock individually per guest.
20983 ++ this adds some overhead to the time functions and
20984 ++ therefore should not be enabled without good reason.
20985 ++
20986 ++config VSERVER_DEVICE
20987 ++ bool "Enable Guest Device Mapping"
20988 ++ depends on EXPERIMENTAL
20989 ++ default n
20990 ++ help
20991 ++ This enables generic device remapping.
20992 ++
20993 ++config VSERVER_PROC_SECURE
20994 ++ bool "Enable Proc Security"
20995 ++ depends on PROC_FS
20996 ++ default y
20997 ++ help
20998 ++ This configures ProcFS security to initially hide
20999 ++ non-process entries for all contexts except the main and
21000 ++ spectator context (i.e. for all guests), which is a secure
21001 ++ default.
21002 ++
21003 ++ (note: on 1.2x the entries were visible by default)
21004 ++
21005 ++config VSERVER_HARDCPU
21006 ++ bool "Enable Hard CPU Limits"
21007 ++ default y
21008 ++ help
21009 ++ Activate the Hard CPU Limits
21010 ++
21011 ++ This will compile in code that allows the Token Bucket
21012 ++ Scheduler to put processes on hold when a context's
21013 ++ tokens are depleted (provided that its per-context
21014 ++ sched_hard flag is set).
21015 ++
21016 ++ Processes belonging to that context will not be able
21017 ++ to consume CPU resources again until a per-context
21018 ++ configured minimum of tokens has been reached.
21019 ++
21020 ++config VSERVER_IDLETIME
21021 ++ bool "Avoid idle CPUs by skipping Time"
21022 ++ depends on VSERVER_HARDCPU
21023 ++ default y
21024 ++ help
21025 ++ This option allows the scheduler to artificially
21026 ++ advance time (per cpu) when otherwise the idle
21027 ++ task would be scheduled, thus keeping the cpu
21028 ++ busy and sharing the available resources among
21029 ++ certain contexts.
21030 ++
21031 ++config VSERVER_IDLELIMIT
21032 ++ bool "Limit the IDLE task"
21033 ++ depends on VSERVER_HARDCPU
21034 ++ default n
21035 ++ help
21036 ++ Limit the idle slices, so the the next context
21037 ++ will be scheduled as soon as possible.
21038 ++
21039 ++ This might improve interactivity and latency, but
21040 ++ will also marginally increase scheduling overhead.
21041 ++
21042 ++choice
21043 ++ prompt "Persistent Inode Tagging"
21044 ++ default TAGGING_ID24
21045 ++ help
21046 ++ This adds persistent context information to filesystems
21047 ++ mounted with the tagxid option. Tagging is a requirement
21048 ++ for per-context disk limits and per-context quota.
21049 ++
21050 ++
21051 ++config TAGGING_NONE
21052 ++ bool "Disabled"
21053 ++ help
21054 ++ do not store per-context information in inodes.
21055 ++
21056 ++config TAGGING_UID16
21057 ++ bool "UID16/GID32"
21058 ++ help
21059 ++ reduces UID to 16 bit, but leaves GID at 32 bit.
21060 ++
21061 ++config TAGGING_GID16
21062 ++ bool "UID32/GID16"
21063 ++ help
21064 ++ reduces GID to 16 bit, but leaves UID at 32 bit.
21065 ++
21066 ++config TAGGING_ID24
21067 ++ bool "UID24/GID24"
21068 ++ help
21069 ++ uses the upper 8bit from UID and GID for XID tagging
21070 ++ which leaves 24bit for UID/GID each, which should be
21071 ++ more than sufficient for normal use.
21072 ++
21073 ++config TAGGING_INTERN
21074 ++ bool "UID32/GID32"
21075 ++ help
21076 ++ this uses otherwise reserved inode fields in the on
21077 ++ disk representation, which limits the use to a few
21078 ++ filesystems (currently ext2 and ext3)
21079 ++
21080 ++endchoice
21081 ++
21082 ++config TAG_NFSD
21083 ++ bool "Tag NFSD User Auth and Files"
21084 ++ default n
21085 ++ help
21086 ++ Enable this if you do want the in-kernel NFS
21087 ++ Server to use the tagging specified above.
21088 ++ (will require patched clients too)
21089 ++
21090 ++config VSERVER_PRIVACY
21091 ++ bool "Honor Privacy Aspects of Guests"
21092 ++ default n
21093 ++ help
21094 ++ When enabled, most context checks will disallow
21095 ++ access to structures assigned to a specific context,
21096 ++ like ptys or loop devices.
21097 ++
21098 ++config VSERVER_CONTEXTS
21099 ++ int "Maximum number of Contexts (1-65533)" if EMBEDDED
21100 ++ range 1 65533
21101 ++ default "768" if 64BIT
21102 ++ default "256"
21103 ++ help
21104 ++ This setting will optimize certain data structures
21105 ++ and memory allocations according to the expected
21106 ++ maximum.
21107 ++
21108 ++ note: this is not a strict upper limit.
21109 ++
21110 ++config VSERVER_WARN
21111 ++ bool "VServer Warnings"
21112 ++ default y
21113 ++ help
21114 ++ This enables various runtime warnings, which will
21115 ++ notify about potential manipulation attempts or
21116 ++ resource shortage. It is generally considered to
21117 ++ be a good idea to have that enabled.
21118 ++
21119 ++config VSERVER_DEBUG
21120 ++ bool "VServer Debugging Code"
21121 ++ default n
21122 ++ help
21123 ++ Set this to yes if you want to be able to activate
21124 ++ debugging output at runtime. It adds a very small
21125 ++ overhead to all vserver related functions and
21126 ++ increases the kernel size by about 20k.
21127 ++
21128 ++config VSERVER_HISTORY
21129 ++ bool "VServer History Tracing"
21130 ++ depends on VSERVER_DEBUG
21131 ++ default n
21132 ++ help
21133 ++ Set this to yes if you want to record the history of
21134 ++ linux-vserver activities, so they can be replayed in
21135 ++ the event of a kernel panic or oops.
21136 ++
21137 ++config VSERVER_HISTORY_SIZE
21138 ++ int "Per-CPU History Size (32-65536)"
21139 ++ depends on VSERVER_HISTORY
21140 ++ range 32 65536
21141 ++ default 64
21142 ++ help
21143 ++ This allows you to specify the number of entries in
21144 ++ the per-CPU history buffer.
21145 ++
21146 ++config VSERVER_MONITOR
21147 ++ bool "VServer Scheduling Monitor"
21148 ++ depends on VSERVER_DISABLED
21149 ++ default n
21150 ++ help
21151 ++ Set this to yes if you want to record the scheduling
21152 ++ decisions, so that they can be relayed to userspace
21153 ++ for detailed analysis.
21154 ++
21155 ++config VSERVER_MONITOR_SIZE
21156 ++ int "Per-CPU Monitor Queue Size (32-65536)"
21157 ++ depends on VSERVER_MONITOR
21158 ++ range 32 65536
21159 ++ default 1024
21160 ++ help
21161 ++ This allows you to specify the number of entries in
21162 ++ the per-CPU scheduling monitor buffer.
21163 ++
21164 ++config VSERVER_MONITOR_SYNC
21165 ++ int "Per-CPU Monitor Sync Interval (0-65536)"
21166 ++ depends on VSERVER_MONITOR
21167 ++ range 0 65536
21168 ++ default 256
21169 ++ help
21170 ++ This allows you to specify the interval in ticks
21171 ++ when a time sync entry is inserted.
21172 ++
21173 ++endmenu
21174 ++
21175 ++
21176 ++config VSERVER
21177 ++ bool
21178 ++ default y
21179 ++ select NAMESPACES
21180 ++ select UTS_NS
21181 ++ select IPC_NS
21182 ++ select USER_NS
21183 ++ select SYSVIPC
21184 ++
21185 ++config VSERVER_SECURITY
21186 ++ bool
21187 ++ depends on SECURITY
21188 ++ default y
21189 ++ select SECURITY_CAPABILITIES
21190 ++
21191 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/limit.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit.c
21192 +--- linux-2.6.29.4/kernel/vserver/limit.c 1970-01-01 01:00:00.000000000 +0100
21193 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit.c 2009-02-22 22:54:26.000000000 +0100
21194 +@@ -0,0 +1,319 @@
21195 ++/*
21196 ++ * linux/kernel/vserver/limit.c
21197 ++ *
21198 ++ * Virtual Server: Context Limits
21199 ++ *
21200 ++ * Copyright (C) 2004-2007 Herbert Pötzl
21201 ++ *
21202 ++ * V0.01 broken out from vcontext V0.05
21203 ++ * V0.02 changed vcmds to vxi arg
21204 ++ *
21205 ++ */
21206 ++
21207 ++#include <linux/sched.h>
21208 ++#include <linux/module.h>
21209 ++#include <linux/vs_limit.h>
21210 ++#include <linux/vserver/limit.h>
21211 ++#include <linux/vserver/limit_cmd.h>
21212 ++
21213 ++#include <asm/uaccess.h>
21214 ++
21215 ++
21216 ++const char *vlimit_name[NUM_LIMITS] = {
21217 ++ [RLIMIT_CPU] = "CPU",
21218 ++ [RLIMIT_RSS] = "RSS",
21219 ++ [RLIMIT_NPROC] = "NPROC",
21220 ++ [RLIMIT_NOFILE] = "NOFILE",
21221 ++ [RLIMIT_MEMLOCK] = "VML",
21222 ++ [RLIMIT_AS] = "VM",
21223 ++ [RLIMIT_LOCKS] = "LOCKS",
21224 ++ [RLIMIT_SIGPENDING] = "SIGP",
21225 ++ [RLIMIT_MSGQUEUE] = "MSGQ",
21226 ++
21227 ++ [VLIMIT_NSOCK] = "NSOCK",
21228 ++ [VLIMIT_OPENFD] = "OPENFD",
21229 ++ [VLIMIT_ANON] = "ANON",
21230 ++ [VLIMIT_SHMEM] = "SHMEM",
21231 ++ [VLIMIT_DENTRY] = "DENTRY",
21232 ++};
21233 ++
21234 ++EXPORT_SYMBOL_GPL(vlimit_name);
21235 ++
21236 ++#define MASK_ENTRY(x) (1 << (x))
21237 ++
21238 ++const struct vcmd_ctx_rlimit_mask_v0 vlimit_mask = {
21239 ++ /* minimum */
21240 ++ 0
21241 ++ , /* softlimit */
21242 ++ MASK_ENTRY( RLIMIT_RSS ) |
21243 ++ MASK_ENTRY( VLIMIT_ANON ) |
21244 ++ 0
21245 ++ , /* maximum */
21246 ++ MASK_ENTRY( RLIMIT_RSS ) |
21247 ++ MASK_ENTRY( RLIMIT_NPROC ) |
21248 ++ MASK_ENTRY( RLIMIT_NOFILE ) |
21249 ++ MASK_ENTRY( RLIMIT_MEMLOCK ) |
21250 ++ MASK_ENTRY( RLIMIT_AS ) |
21251 ++ MASK_ENTRY( RLIMIT_LOCKS ) |
21252 ++ MASK_ENTRY( RLIMIT_MSGQUEUE ) |
21253 ++
21254 ++ MASK_ENTRY( VLIMIT_NSOCK ) |
21255 ++ MASK_ENTRY( VLIMIT_OPENFD ) |
21256 ++ MASK_ENTRY( VLIMIT_ANON ) |
21257 ++ MASK_ENTRY( VLIMIT_SHMEM ) |
21258 ++ MASK_ENTRY( VLIMIT_DENTRY ) |
21259 ++ 0
21260 ++};
21261 ++ /* accounting only */
21262 ++uint32_t account_mask =
21263 ++ MASK_ENTRY( VLIMIT_SEMARY ) |
21264 ++ MASK_ENTRY( VLIMIT_NSEMS ) |
21265 ++ MASK_ENTRY( VLIMIT_MAPPED ) |
21266 ++ 0;
21267 ++
21268 ++
21269 ++static int is_valid_vlimit(int id)
21270 ++{
21271 ++ uint32_t mask = vlimit_mask.minimum |
21272 ++ vlimit_mask.softlimit | vlimit_mask.maximum;
21273 ++ return mask & (1 << id);
21274 ++}
21275 ++
21276 ++static int is_accounted_vlimit(int id)
21277 ++{
21278 ++ if (is_valid_vlimit(id))
21279 ++ return 1;
21280 ++ return account_mask & (1 << id);
21281 ++}
21282 ++
21283 ++
21284 ++static inline uint64_t vc_get_soft(struct vx_info *vxi, int id)
21285 ++{
21286 ++ rlim_t limit = __rlim_soft(&vxi->limit, id);
21287 ++ return VX_VLIM(limit);
21288 ++}
21289 ++
21290 ++static inline uint64_t vc_get_hard(struct vx_info *vxi, int id)
21291 ++{
21292 ++ rlim_t limit = __rlim_hard(&vxi->limit, id);
21293 ++ return VX_VLIM(limit);
21294 ++}
21295 ++
21296 ++static int do_get_rlimit(struct vx_info *vxi, uint32_t id,
21297 ++ uint64_t *minimum, uint64_t *softlimit, uint64_t *maximum)
21298 ++{
21299 ++ if (!is_valid_vlimit(id))
21300 ++ return -EINVAL;
21301 ++
21302 ++ if (minimum)
21303 ++ *minimum = CRLIM_UNSET;
21304 ++ if (softlimit)
21305 ++ *softlimit = vc_get_soft(vxi, id);
21306 ++ if (maximum)
21307 ++ *maximum = vc_get_hard(vxi, id);
21308 ++ return 0;
21309 ++}
21310 ++
21311 ++int vc_get_rlimit(struct vx_info *vxi, void __user *data)
21312 ++{
21313 ++ struct vcmd_ctx_rlimit_v0 vc_data;
21314 ++ int ret;
21315 ++
21316 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21317 ++ return -EFAULT;
21318 ++
21319 ++ ret = do_get_rlimit(vxi, vc_data.id,
21320 ++ &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21321 ++ if (ret)
21322 ++ return ret;
21323 ++
21324 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21325 ++ return -EFAULT;
21326 ++ return 0;
21327 ++}
21328 ++
21329 ++static int do_set_rlimit(struct vx_info *vxi, uint32_t id,
21330 ++ uint64_t minimum, uint64_t softlimit, uint64_t maximum)
21331 ++{
21332 ++ if (!is_valid_vlimit(id))
21333 ++ return -EINVAL;
21334 ++
21335 ++ if (maximum != CRLIM_KEEP)
21336 ++ __rlim_hard(&vxi->limit, id) = VX_RLIM(maximum);
21337 ++ if (softlimit != CRLIM_KEEP)
21338 ++ __rlim_soft(&vxi->limit, id) = VX_RLIM(softlimit);
21339 ++
21340 ++ /* clamp soft limit */
21341 ++ if (__rlim_soft(&vxi->limit, id) > __rlim_hard(&vxi->limit, id))
21342 ++ __rlim_soft(&vxi->limit, id) = __rlim_hard(&vxi->limit, id);
21343 ++
21344 ++ return 0;
21345 ++}
21346 ++
21347 ++int vc_set_rlimit(struct vx_info *vxi, void __user *data)
21348 ++{
21349 ++ struct vcmd_ctx_rlimit_v0 vc_data;
21350 ++
21351 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21352 ++ return -EFAULT;
21353 ++
21354 ++ return do_set_rlimit(vxi, vc_data.id,
21355 ++ vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21356 ++}
21357 ++
21358 ++#ifdef CONFIG_IA32_EMULATION
21359 ++
21360 ++int vc_set_rlimit_x32(struct vx_info *vxi, void __user *data)
21361 ++{
21362 ++ struct vcmd_ctx_rlimit_v0_x32 vc_data;
21363 ++
21364 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21365 ++ return -EFAULT;
21366 ++
21367 ++ return do_set_rlimit(vxi, vc_data.id,
21368 ++ vc_data.minimum, vc_data.softlimit, vc_data.maximum);
21369 ++}
21370 ++
21371 ++int vc_get_rlimit_x32(struct vx_info *vxi, void __user *data)
21372 ++{
21373 ++ struct vcmd_ctx_rlimit_v0_x32 vc_data;
21374 ++ int ret;
21375 ++
21376 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21377 ++ return -EFAULT;
21378 ++
21379 ++ ret = do_get_rlimit(vxi, vc_data.id,
21380 ++ &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
21381 ++ if (ret)
21382 ++ return ret;
21383 ++
21384 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21385 ++ return -EFAULT;
21386 ++ return 0;
21387 ++}
21388 ++
21389 ++#endif /* CONFIG_IA32_EMULATION */
21390 ++
21391 ++
21392 ++int vc_get_rlimit_mask(uint32_t id, void __user *data)
21393 ++{
21394 ++ if (copy_to_user(data, &vlimit_mask, sizeof(vlimit_mask)))
21395 ++ return -EFAULT;
21396 ++ return 0;
21397 ++}
21398 ++
21399 ++
21400 ++static inline void vx_reset_minmax(struct _vx_limit *limit)
21401 ++{
21402 ++ rlim_t value;
21403 ++ int lim;
21404 ++
21405 ++ for (lim = 0; lim < NUM_LIMITS; lim++) {
21406 ++ value = __rlim_get(limit, lim);
21407 ++ __rlim_rmax(limit, lim) = value;
21408 ++ __rlim_rmin(limit, lim) = value;
21409 ++ }
21410 ++}
21411 ++
21412 ++
21413 ++int vc_reset_minmax(struct vx_info *vxi, void __user *data)
21414 ++{
21415 ++ vx_reset_minmax(&vxi->limit);
21416 ++ return 0;
21417 ++}
21418 ++
21419 ++
21420 ++int vc_rlimit_stat(struct vx_info *vxi, void __user *data)
21421 ++{
21422 ++ struct vcmd_rlimit_stat_v0 vc_data;
21423 ++ struct _vx_limit *limit = &vxi->limit;
21424 ++ int id;
21425 ++
21426 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21427 ++ return -EFAULT;
21428 ++
21429 ++ id = vc_data.id;
21430 ++ if (!is_accounted_vlimit(id))
21431 ++ return -EINVAL;
21432 ++
21433 ++ vx_limit_fixup(limit, id);
21434 ++ vc_data.hits = atomic_read(&__rlim_lhit(limit, id));
21435 ++ vc_data.value = __rlim_get(limit, id);
21436 ++ vc_data.minimum = __rlim_rmin(limit, id);
21437 ++ vc_data.maximum = __rlim_rmax(limit, id);
21438 ++
21439 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21440 ++ return -EFAULT;
21441 ++ return 0;
21442 ++}
21443 ++
21444 ++
21445 ++void vx_vsi_meminfo(struct sysinfo *val)
21446 ++{
21447 ++ struct vx_info *vxi = current->vx_info;
21448 ++ unsigned long totalram, freeram;
21449 ++ rlim_t v;
21450 ++
21451 ++ /* we blindly accept the max */
21452 ++ v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21453 ++ totalram = (v != RLIM_INFINITY) ? v : val->totalram;
21454 ++
21455 ++ /* total minus used equals free */
21456 ++ v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21457 ++ freeram = (v < totalram) ? totalram - v : 0;
21458 ++
21459 ++ val->totalram = totalram;
21460 ++ val->freeram = freeram;
21461 ++ val->bufferram = 0;
21462 ++ val->totalhigh = 0;
21463 ++ val->freehigh = 0;
21464 ++ return;
21465 ++}
21466 ++
21467 ++void vx_vsi_swapinfo(struct sysinfo *val)
21468 ++{
21469 ++ struct vx_info *vxi = current->vx_info;
21470 ++ unsigned long totalswap, freeswap;
21471 ++ rlim_t v, w;
21472 ++
21473 ++ v = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21474 ++ if (v == RLIM_INFINITY) {
21475 ++ val->freeswap = val->totalswap;
21476 ++ return;
21477 ++ }
21478 ++
21479 ++ /* we blindly accept the max */
21480 ++ w = __rlim_hard(&vxi->limit, RLIMIT_RSS);
21481 ++ totalswap = (w != RLIM_INFINITY) ? (w - v) : val->totalswap;
21482 ++
21483 ++ /* currently 'used' swap */
21484 ++ w = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21485 ++ w -= (w > v) ? v : w;
21486 ++
21487 ++ /* total minus used equals free */
21488 ++ freeswap = (w < totalswap) ? totalswap - w : 0;
21489 ++
21490 ++ val->totalswap = totalswap;
21491 ++ val->freeswap = freeswap;
21492 ++ return;
21493 ++}
21494 ++
21495 ++
21496 ++unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm)
21497 ++{
21498 ++ struct vx_info *vxi = mm->mm_vx_info;
21499 ++ unsigned long points;
21500 ++ rlim_t v, w;
21501 ++
21502 ++ if (!vxi)
21503 ++ return 0;
21504 ++
21505 ++ points = vxi->vx_badness_bias;
21506 ++
21507 ++ v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
21508 ++ w = __rlim_soft(&vxi->limit, RLIMIT_RSS);
21509 ++ points += (v > w) ? (v - w) : 0;
21510 ++
21511 ++ return points;
21512 ++}
21513 ++
21514 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/limit_init.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit_init.h
21515 +--- linux-2.6.29.4/kernel/vserver/limit_init.h 1970-01-01 01:00:00.000000000 +0100
21516 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit_init.h 2009-02-22 22:54:26.000000000 +0100
21517 +@@ -0,0 +1,31 @@
21518 ++
21519 ++
21520 ++static inline void vx_info_init_limit(struct _vx_limit *limit)
21521 ++{
21522 ++ int lim;
21523 ++
21524 ++ for (lim = 0; lim < NUM_LIMITS; lim++) {
21525 ++ __rlim_soft(limit, lim) = RLIM_INFINITY;
21526 ++ __rlim_hard(limit, lim) = RLIM_INFINITY;
21527 ++ __rlim_set(limit, lim, 0);
21528 ++ atomic_set(&__rlim_lhit(limit, lim), 0);
21529 ++ __rlim_rmin(limit, lim) = 0;
21530 ++ __rlim_rmax(limit, lim) = 0;
21531 ++ }
21532 ++}
21533 ++
21534 ++static inline void vx_info_exit_limit(struct _vx_limit *limit)
21535 ++{
21536 ++ rlim_t value;
21537 ++ int lim;
21538 ++
21539 ++ for (lim = 0; lim < NUM_LIMITS; lim++) {
21540 ++ if ((1 << lim) & VLIM_NOCHECK)
21541 ++ continue;
21542 ++ value = __rlim_get(limit, lim);
21543 ++ vxwprintk_xid(value,
21544 ++ "!!! limit: %p[%s,%d] = %ld on exit.",
21545 ++ limit, vlimit_name[lim], lim, (long)value);
21546 ++ }
21547 ++}
21548 ++
21549 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/limit_proc.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit_proc.h
21550 +--- linux-2.6.29.4/kernel/vserver/limit_proc.h 1970-01-01 01:00:00.000000000 +0100
21551 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/limit_proc.h 2009-02-22 22:54:26.000000000 +0100
21552 +@@ -0,0 +1,57 @@
21553 ++#ifndef _VX_LIMIT_PROC_H
21554 ++#define _VX_LIMIT_PROC_H
21555 ++
21556 ++#include <linux/vserver/limit_int.h>
21557 ++
21558 ++
21559 ++#define VX_LIMIT_FMT ":\t%8ld\t%8ld/%8ld\t%8lld/%8lld\t%6d\n"
21560 ++#define VX_LIMIT_TOP \
21561 ++ "Limit\t current\t min/max\t\t soft/hard\t\thits\n"
21562 ++
21563 ++#define VX_LIMIT_ARG(r) \
21564 ++ (unsigned long)__rlim_get(limit, r), \
21565 ++ (unsigned long)__rlim_rmin(limit, r), \
21566 ++ (unsigned long)__rlim_rmax(limit, r), \
21567 ++ VX_VLIM(__rlim_soft(limit, r)), \
21568 ++ VX_VLIM(__rlim_hard(limit, r)), \
21569 ++ atomic_read(&__rlim_lhit(limit, r))
21570 ++
21571 ++static inline int vx_info_proc_limit(struct _vx_limit *limit, char *buffer)
21572 ++{
21573 ++ vx_limit_fixup(limit, -1);
21574 ++ return sprintf(buffer, VX_LIMIT_TOP
21575 ++ "PROC" VX_LIMIT_FMT
21576 ++ "VM" VX_LIMIT_FMT
21577 ++ "VML" VX_LIMIT_FMT
21578 ++ "RSS" VX_LIMIT_FMT
21579 ++ "ANON" VX_LIMIT_FMT
21580 ++ "RMAP" VX_LIMIT_FMT
21581 ++ "FILES" VX_LIMIT_FMT
21582 ++ "OFD" VX_LIMIT_FMT
21583 ++ "LOCKS" VX_LIMIT_FMT
21584 ++ "SOCK" VX_LIMIT_FMT
21585 ++ "MSGQ" VX_LIMIT_FMT
21586 ++ "SHM" VX_LIMIT_FMT
21587 ++ "SEMA" VX_LIMIT_FMT
21588 ++ "SEMS" VX_LIMIT_FMT
21589 ++ "DENT" VX_LIMIT_FMT,
21590 ++ VX_LIMIT_ARG(RLIMIT_NPROC),
21591 ++ VX_LIMIT_ARG(RLIMIT_AS),
21592 ++ VX_LIMIT_ARG(RLIMIT_MEMLOCK),
21593 ++ VX_LIMIT_ARG(RLIMIT_RSS),
21594 ++ VX_LIMIT_ARG(VLIMIT_ANON),
21595 ++ VX_LIMIT_ARG(VLIMIT_MAPPED),
21596 ++ VX_LIMIT_ARG(RLIMIT_NOFILE),
21597 ++ VX_LIMIT_ARG(VLIMIT_OPENFD),
21598 ++ VX_LIMIT_ARG(RLIMIT_LOCKS),
21599 ++ VX_LIMIT_ARG(VLIMIT_NSOCK),
21600 ++ VX_LIMIT_ARG(RLIMIT_MSGQUEUE),
21601 ++ VX_LIMIT_ARG(VLIMIT_SHMEM),
21602 ++ VX_LIMIT_ARG(VLIMIT_SEMARY),
21603 ++ VX_LIMIT_ARG(VLIMIT_NSEMS),
21604 ++ VX_LIMIT_ARG(VLIMIT_DENTRY));
21605 ++}
21606 ++
21607 ++#endif /* _VX_LIMIT_PROC_H */
21608 ++
21609 ++
21610 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/Makefile linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/Makefile
21611 +--- linux-2.6.29.4/kernel/vserver/Makefile 1970-01-01 01:00:00.000000000 +0100
21612 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/Makefile 2009-02-22 22:54:26.000000000 +0100
21613 +@@ -0,0 +1,18 @@
21614 ++#
21615 ++# Makefile for the Linux vserver routines.
21616 ++#
21617 ++
21618 ++
21619 ++obj-y += vserver.o
21620 ++
21621 ++vserver-y := switch.o context.o space.o sched.o network.o inode.o \
21622 ++ limit.o cvirt.o cacct.o signal.o helper.o init.o \
21623 ++ dlimit.o tag.o
21624 ++
21625 ++vserver-$(CONFIG_INET) += inet.o
21626 ++vserver-$(CONFIG_PROC_FS) += proc.o
21627 ++vserver-$(CONFIG_VSERVER_DEBUG) += sysctl.o debug.o
21628 ++vserver-$(CONFIG_VSERVER_HISTORY) += history.o
21629 ++vserver-$(CONFIG_VSERVER_MONITOR) += monitor.o
21630 ++vserver-$(CONFIG_VSERVER_DEVICE) += device.o
21631 ++
21632 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/monitor.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/monitor.c
21633 +--- linux-2.6.29.4/kernel/vserver/monitor.c 1970-01-01 01:00:00.000000000 +0100
21634 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/monitor.c 2009-02-22 22:54:26.000000000 +0100
21635 +@@ -0,0 +1,138 @@
21636 ++/*
21637 ++ * kernel/vserver/monitor.c
21638 ++ *
21639 ++ * Virtual Context Scheduler Monitor
21640 ++ *
21641 ++ * Copyright (C) 2006-2007 Herbert Pötzl
21642 ++ *
21643 ++ * V0.01 basic design
21644 ++ *
21645 ++ */
21646 ++
21647 ++#include <linux/module.h>
21648 ++#include <linux/jiffies.h>
21649 ++#include <asm/uaccess.h>
21650 ++#include <asm/atomic.h>
21651 ++
21652 ++#include <linux/vserver/monitor.h>
21653 ++#include <linux/vserver/debug_cmd.h>
21654 ++
21655 ++
21656 ++#ifdef CONFIG_VSERVER_MONITOR
21657 ++#define VXM_SIZE CONFIG_VSERVER_MONITOR_SIZE
21658 ++#else
21659 ++#define VXM_SIZE 64
21660 ++#endif
21661 ++
21662 ++struct _vx_monitor {
21663 ++ unsigned int counter;
21664 ++
21665 ++ struct _vx_mon_entry entry[VXM_SIZE+1];
21666 ++};
21667 ++
21668 ++
21669 ++DEFINE_PER_CPU(struct _vx_monitor, vx_monitor_buffer);
21670 ++
21671 ++unsigned volatile int vxm_active = 1;
21672 ++
21673 ++static atomic_t sequence = ATOMIC_INIT(0);
21674 ++
21675 ++
21676 ++/* vxm_advance()
21677 ++
21678 ++ * requires disabled preemption */
21679 ++
21680 ++struct _vx_mon_entry *vxm_advance(int cpu)
21681 ++{
21682 ++ struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
21683 ++ struct _vx_mon_entry *entry;
21684 ++ unsigned int index;
21685 ++
21686 ++ index = vxm_active ? (mon->counter++ % VXM_SIZE) : VXM_SIZE;
21687 ++ entry = &mon->entry[index];
21688 ++
21689 ++ entry->ev.seq = atomic_inc_return(&sequence);
21690 ++ entry->ev.jif = jiffies;
21691 ++ return entry;
21692 ++}
21693 ++
21694 ++EXPORT_SYMBOL_GPL(vxm_advance);
21695 ++
21696 ++
21697 ++int do_read_monitor(struct __user _vx_mon_entry *data,
21698 ++ int cpu, uint32_t *index, uint32_t *count)
21699 ++{
21700 ++ int pos, ret = 0;
21701 ++ struct _vx_monitor *mon = &per_cpu(vx_monitor_buffer, cpu);
21702 ++ int end = mon->counter;
21703 ++ int start = end - VXM_SIZE + 2;
21704 ++ int idx = *index;
21705 ++
21706 ++ /* special case: get current pos */
21707 ++ if (!*count) {
21708 ++ *index = end;
21709 ++ return 0;
21710 ++ }
21711 ++
21712 ++ /* have we lost some data? */
21713 ++ if (idx < start)
21714 ++ idx = start;
21715 ++
21716 ++ for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
21717 ++ struct _vx_mon_entry *entry =
21718 ++ &mon->entry[idx % VXM_SIZE];
21719 ++
21720 ++ /* send entry to userspace */
21721 ++ ret = copy_to_user(&data[pos], entry, sizeof(*entry));
21722 ++ if (ret)
21723 ++ break;
21724 ++ }
21725 ++ /* save new index and count */
21726 ++ *index = idx;
21727 ++ *count = pos;
21728 ++ return ret ? ret : (*index < end);
21729 ++}
21730 ++
21731 ++int vc_read_monitor(uint32_t id, void __user *data)
21732 ++{
21733 ++ struct vcmd_read_monitor_v0 vc_data;
21734 ++ int ret;
21735 ++
21736 ++ if (id >= NR_CPUS)
21737 ++ return -EINVAL;
21738 ++
21739 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21740 ++ return -EFAULT;
21741 ++
21742 ++ ret = do_read_monitor((struct __user _vx_mon_entry *)vc_data.data,
21743 ++ id, &vc_data.index, &vc_data.count);
21744 ++
21745 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21746 ++ return -EFAULT;
21747 ++ return ret;
21748 ++}
21749 ++
21750 ++#ifdef CONFIG_COMPAT
21751 ++
21752 ++int vc_read_monitor_x32(uint32_t id, void __user *data)
21753 ++{
21754 ++ struct vcmd_read_monitor_v0_x32 vc_data;
21755 ++ int ret;
21756 ++
21757 ++ if (id >= NR_CPUS)
21758 ++ return -EINVAL;
21759 ++
21760 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21761 ++ return -EFAULT;
21762 ++
21763 ++ ret = do_read_monitor((struct __user _vx_mon_entry *)
21764 ++ compat_ptr(vc_data.data_ptr),
21765 ++ id, &vc_data.index, &vc_data.count);
21766 ++
21767 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21768 ++ return -EFAULT;
21769 ++ return ret;
21770 ++}
21771 ++
21772 ++#endif /* CONFIG_COMPAT */
21773 ++
21774 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/network.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/network.c
21775 +--- linux-2.6.29.4/kernel/vserver/network.c 1970-01-01 01:00:00.000000000 +0100
21776 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/network.c 2009-02-22 22:54:26.000000000 +0100
21777 +@@ -0,0 +1,864 @@
21778 ++/*
21779 ++ * linux/kernel/vserver/network.c
21780 ++ *
21781 ++ * Virtual Server: Network Support
21782 ++ *
21783 ++ * Copyright (C) 2003-2007 Herbert Pötzl
21784 ++ *
21785 ++ * V0.01 broken out from vcontext V0.05
21786 ++ * V0.02 cleaned up implementation
21787 ++ * V0.03 added equiv nx commands
21788 ++ * V0.04 switch to RCU based hash
21789 ++ * V0.05 and back to locking again
21790 ++ * V0.06 changed vcmds to nxi arg
21791 ++ * V0.07 have __create claim() the nxi
21792 ++ *
21793 ++ */
21794 ++
21795 ++#include <linux/err.h>
21796 ++#include <linux/slab.h>
21797 ++#include <linux/rcupdate.h>
21798 ++
21799 ++#include <linux/vs_network.h>
21800 ++#include <linux/vs_pid.h>
21801 ++#include <linux/vserver/network_cmd.h>
21802 ++
21803 ++
21804 ++atomic_t nx_global_ctotal = ATOMIC_INIT(0);
21805 ++atomic_t nx_global_cactive = ATOMIC_INIT(0);
21806 ++
21807 ++static struct kmem_cache *nx_addr_v4_cachep = NULL;
21808 ++static struct kmem_cache *nx_addr_v6_cachep = NULL;
21809 ++
21810 ++
21811 ++static int __init init_network(void)
21812 ++{
21813 ++ nx_addr_v4_cachep = kmem_cache_create("nx_v4_addr_cache",
21814 ++ sizeof(struct nx_addr_v4), 0,
21815 ++ SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
21816 ++ nx_addr_v6_cachep = kmem_cache_create("nx_v6_addr_cache",
21817 ++ sizeof(struct nx_addr_v6), 0,
21818 ++ SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
21819 ++ return 0;
21820 ++}
21821 ++
21822 ++
21823 ++/* __alloc_nx_addr_v4() */
21824 ++
21825 ++static inline struct nx_addr_v4 *__alloc_nx_addr_v4(void)
21826 ++{
21827 ++ struct nx_addr_v4 *nxa = kmem_cache_alloc(
21828 ++ nx_addr_v4_cachep, GFP_KERNEL);
21829 ++
21830 ++ if (!IS_ERR(nxa))
21831 ++ memset(nxa, 0, sizeof(*nxa));
21832 ++ return nxa;
21833 ++}
21834 ++
21835 ++/* __dealloc_nx_addr_v4() */
21836 ++
21837 ++static inline void __dealloc_nx_addr_v4(struct nx_addr_v4 *nxa)
21838 ++{
21839 ++ kmem_cache_free(nx_addr_v4_cachep, nxa);
21840 ++}
21841 ++
21842 ++/* __dealloc_nx_addr_v4_all() */
21843 ++
21844 ++static inline void __dealloc_nx_addr_v4_all(struct nx_addr_v4 *nxa)
21845 ++{
21846 ++ while (nxa) {
21847 ++ struct nx_addr_v4 *next = nxa->next;
21848 ++
21849 ++ __dealloc_nx_addr_v4(nxa);
21850 ++ nxa = next;
21851 ++ }
21852 ++}
21853 ++
21854 ++
21855 ++#ifdef CONFIG_IPV6
21856 ++
21857 ++/* __alloc_nx_addr_v6() */
21858 ++
21859 ++static inline struct nx_addr_v6 *__alloc_nx_addr_v6(void)
21860 ++{
21861 ++ struct nx_addr_v6 *nxa = kmem_cache_alloc(
21862 ++ nx_addr_v6_cachep, GFP_KERNEL);
21863 ++
21864 ++ if (!IS_ERR(nxa))
21865 ++ memset(nxa, 0, sizeof(*nxa));
21866 ++ return nxa;
21867 ++}
21868 ++
21869 ++/* __dealloc_nx_addr_v6() */
21870 ++
21871 ++static inline void __dealloc_nx_addr_v6(struct nx_addr_v6 *nxa)
21872 ++{
21873 ++ kmem_cache_free(nx_addr_v6_cachep, nxa);
21874 ++}
21875 ++
21876 ++/* __dealloc_nx_addr_v6_all() */
21877 ++
21878 ++static inline void __dealloc_nx_addr_v6_all(struct nx_addr_v6 *nxa)
21879 ++{
21880 ++ while (nxa) {
21881 ++ struct nx_addr_v6 *next = nxa->next;
21882 ++
21883 ++ __dealloc_nx_addr_v6(nxa);
21884 ++ nxa = next;
21885 ++ }
21886 ++}
21887 ++
21888 ++#endif /* CONFIG_IPV6 */
21889 ++
21890 ++/* __alloc_nx_info()
21891 ++
21892 ++ * allocate an initialized nx_info struct
21893 ++ * doesn't make it visible (hash) */
21894 ++
21895 ++static struct nx_info *__alloc_nx_info(nid_t nid)
21896 ++{
21897 ++ struct nx_info *new = NULL;
21898 ++
21899 ++ vxdprintk(VXD_CBIT(nid, 1), "alloc_nx_info(%d)*", nid);
21900 ++
21901 ++ /* would this benefit from a slab cache? */
21902 ++ new = kmalloc(sizeof(struct nx_info), GFP_KERNEL);
21903 ++ if (!new)
21904 ++ return 0;
21905 ++
21906 ++ memset(new, 0, sizeof(struct nx_info));
21907 ++ new->nx_id = nid;
21908 ++ INIT_HLIST_NODE(&new->nx_hlist);
21909 ++ atomic_set(&new->nx_usecnt, 0);
21910 ++ atomic_set(&new->nx_tasks, 0);
21911 ++ new->nx_state = 0;
21912 ++
21913 ++ new->nx_flags = NXF_INIT_SET;
21914 ++
21915 ++ /* rest of init goes here */
21916 ++
21917 ++ new->v4_lback.s_addr = htonl(INADDR_LOOPBACK);
21918 ++ new->v4_bcast.s_addr = htonl(INADDR_BROADCAST);
21919 ++
21920 ++ vxdprintk(VXD_CBIT(nid, 0),
21921 ++ "alloc_nx_info(%d) = %p", nid, new);
21922 ++ atomic_inc(&nx_global_ctotal);
21923 ++ return new;
21924 ++}
21925 ++
21926 ++/* __dealloc_nx_info()
21927 ++
21928 ++ * final disposal of nx_info */
21929 ++
21930 ++static void __dealloc_nx_info(struct nx_info *nxi)
21931 ++{
21932 ++ vxdprintk(VXD_CBIT(nid, 0),
21933 ++ "dealloc_nx_info(%p)", nxi);
21934 ++
21935 ++ nxi->nx_hlist.next = LIST_POISON1;
21936 ++ nxi->nx_id = -1;
21937 ++
21938 ++ BUG_ON(atomic_read(&nxi->nx_usecnt));
21939 ++ BUG_ON(atomic_read(&nxi->nx_tasks));
21940 ++
21941 ++ __dealloc_nx_addr_v4_all(nxi->v4.next);
21942 ++
21943 ++ nxi->nx_state |= NXS_RELEASED;
21944 ++ kfree(nxi);
21945 ++ atomic_dec(&nx_global_ctotal);
21946 ++}
21947 ++
21948 ++static void __shutdown_nx_info(struct nx_info *nxi)
21949 ++{
21950 ++ nxi->nx_state |= NXS_SHUTDOWN;
21951 ++ vs_net_change(nxi, VSC_NETDOWN);
21952 ++}
21953 ++
21954 ++/* exported stuff */
21955 ++
21956 ++void free_nx_info(struct nx_info *nxi)
21957 ++{
21958 ++ /* context shutdown is mandatory */
21959 ++ BUG_ON(nxi->nx_state != NXS_SHUTDOWN);
21960 ++
21961 ++ /* context must not be hashed */
21962 ++ BUG_ON(nxi->nx_state & NXS_HASHED);
21963 ++
21964 ++ BUG_ON(atomic_read(&nxi->nx_usecnt));
21965 ++ BUG_ON(atomic_read(&nxi->nx_tasks));
21966 ++
21967 ++ __dealloc_nx_info(nxi);
21968 ++}
21969 ++
21970 ++
21971 ++void __nx_set_lback(struct nx_info *nxi)
21972 ++{
21973 ++ int nid = nxi->nx_id;
21974 ++ __be32 lback = htonl(INADDR_LOOPBACK ^ ((nid & 0xFFFF) << 8));
21975 ++
21976 ++ nxi->v4_lback.s_addr = lback;
21977 ++}
21978 ++
21979 ++extern int __nx_inet_add_lback(__be32 addr);
21980 ++extern int __nx_inet_del_lback(__be32 addr);
21981 ++
21982 ++
21983 ++/* hash table for nx_info hash */
21984 ++
21985 ++#define NX_HASH_SIZE 13
21986 ++
21987 ++struct hlist_head nx_info_hash[NX_HASH_SIZE];
21988 ++
21989 ++static spinlock_t nx_info_hash_lock = SPIN_LOCK_UNLOCKED;
21990 ++
21991 ++
21992 ++static inline unsigned int __hashval(nid_t nid)
21993 ++{
21994 ++ return (nid % NX_HASH_SIZE);
21995 ++}
21996 ++
21997 ++
21998 ++
21999 ++/* __hash_nx_info()
22000 ++
22001 ++ * add the nxi to the global hash table
22002 ++ * requires the hash_lock to be held */
22003 ++
22004 ++static inline void __hash_nx_info(struct nx_info *nxi)
22005 ++{
22006 ++ struct hlist_head *head;
22007 ++
22008 ++ vxd_assert_lock(&nx_info_hash_lock);
22009 ++ vxdprintk(VXD_CBIT(nid, 4),
22010 ++ "__hash_nx_info: %p[#%d]", nxi, nxi->nx_id);
22011 ++
22012 ++ /* context must not be hashed */
22013 ++ BUG_ON(nx_info_state(nxi, NXS_HASHED));
22014 ++
22015 ++ nxi->nx_state |= NXS_HASHED;
22016 ++ head = &nx_info_hash[__hashval(nxi->nx_id)];
22017 ++ hlist_add_head(&nxi->nx_hlist, head);
22018 ++ atomic_inc(&nx_global_cactive);
22019 ++}
22020 ++
22021 ++/* __unhash_nx_info()
22022 ++
22023 ++ * remove the nxi from the global hash table
22024 ++ * requires the hash_lock to be held */
22025 ++
22026 ++static inline void __unhash_nx_info(struct nx_info *nxi)
22027 ++{
22028 ++ vxd_assert_lock(&nx_info_hash_lock);
22029 ++ vxdprintk(VXD_CBIT(nid, 4),
22030 ++ "__unhash_nx_info: %p[#%d.%d.%d]", nxi, nxi->nx_id,
22031 ++ atomic_read(&nxi->nx_usecnt), atomic_read(&nxi->nx_tasks));
22032 ++
22033 ++ /* context must be hashed */
22034 ++ BUG_ON(!nx_info_state(nxi, NXS_HASHED));
22035 ++ /* but without tasks */
22036 ++ BUG_ON(atomic_read(&nxi->nx_tasks));
22037 ++
22038 ++ nxi->nx_state &= ~NXS_HASHED;
22039 ++ hlist_del(&nxi->nx_hlist);
22040 ++ atomic_dec(&nx_global_cactive);
22041 ++}
22042 ++
22043 ++
22044 ++/* __lookup_nx_info()
22045 ++
22046 ++ * requires the hash_lock to be held
22047 ++ * doesn't increment the nx_refcnt */
22048 ++
22049 ++static inline struct nx_info *__lookup_nx_info(nid_t nid)
22050 ++{
22051 ++ struct hlist_head *head = &nx_info_hash[__hashval(nid)];
22052 ++ struct hlist_node *pos;
22053 ++ struct nx_info *nxi;
22054 ++
22055 ++ vxd_assert_lock(&nx_info_hash_lock);
22056 ++ hlist_for_each(pos, head) {
22057 ++ nxi = hlist_entry(pos, struct nx_info, nx_hlist);
22058 ++
22059 ++ if (nxi->nx_id == nid)
22060 ++ goto found;
22061 ++ }
22062 ++ nxi = NULL;
22063 ++found:
22064 ++ vxdprintk(VXD_CBIT(nid, 0),
22065 ++ "__lookup_nx_info(#%u): %p[#%u]",
22066 ++ nid, nxi, nxi ? nxi->nx_id : 0);
22067 ++ return nxi;
22068 ++}
22069 ++
22070 ++
22071 ++/* __create_nx_info()
22072 ++
22073 ++ * create the requested context
22074 ++ * get(), claim() and hash it */
22075 ++
22076 ++static struct nx_info *__create_nx_info(int id)
22077 ++{
22078 ++ struct nx_info *new, *nxi = NULL;
22079 ++
22080 ++ vxdprintk(VXD_CBIT(nid, 1), "create_nx_info(%d)*", id);
22081 ++
22082 ++ if (!(new = __alloc_nx_info(id)))
22083 ++ return ERR_PTR(-ENOMEM);
22084 ++
22085 ++ /* required to make dynamic xids unique */
22086 ++ spin_lock(&nx_info_hash_lock);
22087 ++
22088 ++ /* static context requested */
22089 ++ if ((nxi = __lookup_nx_info(id))) {
22090 ++ vxdprintk(VXD_CBIT(nid, 0),
22091 ++ "create_nx_info(%d) = %p (already there)", id, nxi);
22092 ++ if (nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22093 ++ nxi = ERR_PTR(-EBUSY);
22094 ++ else
22095 ++ nxi = ERR_PTR(-EEXIST);
22096 ++ goto out_unlock;
22097 ++ }
22098 ++ /* new context */
22099 ++ vxdprintk(VXD_CBIT(nid, 0),
22100 ++ "create_nx_info(%d) = %p (new)", id, new);
22101 ++ claim_nx_info(new, NULL);
22102 ++ __nx_set_lback(new);
22103 ++ __hash_nx_info(get_nx_info(new));
22104 ++ nxi = new, new = NULL;
22105 ++
22106 ++out_unlock:
22107 ++ spin_unlock(&nx_info_hash_lock);
22108 ++ if (new)
22109 ++ __dealloc_nx_info(new);
22110 ++ return nxi;
22111 ++}
22112 ++
22113 ++
22114 ++
22115 ++/* exported stuff */
22116 ++
22117 ++
22118 ++void unhash_nx_info(struct nx_info *nxi)
22119 ++{
22120 ++ __shutdown_nx_info(nxi);
22121 ++ spin_lock(&nx_info_hash_lock);
22122 ++ __unhash_nx_info(nxi);
22123 ++ spin_unlock(&nx_info_hash_lock);
22124 ++}
22125 ++
22126 ++/* lookup_nx_info()
22127 ++
22128 ++ * search for a nx_info and get() it
22129 ++ * negative id means current */
22130 ++
22131 ++struct nx_info *lookup_nx_info(int id)
22132 ++{
22133 ++ struct nx_info *nxi = NULL;
22134 ++
22135 ++ if (id < 0) {
22136 ++ nxi = get_nx_info(current->nx_info);
22137 ++ } else if (id > 1) {
22138 ++ spin_lock(&nx_info_hash_lock);
22139 ++ nxi = get_nx_info(__lookup_nx_info(id));
22140 ++ spin_unlock(&nx_info_hash_lock);
22141 ++ }
22142 ++ return nxi;
22143 ++}
22144 ++
22145 ++/* nid_is_hashed()
22146 ++
22147 ++ * verify that nid is still hashed */
22148 ++
22149 ++int nid_is_hashed(nid_t nid)
22150 ++{
22151 ++ int hashed;
22152 ++
22153 ++ spin_lock(&nx_info_hash_lock);
22154 ++ hashed = (__lookup_nx_info(nid) != NULL);
22155 ++ spin_unlock(&nx_info_hash_lock);
22156 ++ return hashed;
22157 ++}
22158 ++
22159 ++
22160 ++#ifdef CONFIG_PROC_FS
22161 ++
22162 ++/* get_nid_list()
22163 ++
22164 ++ * get a subset of hashed nids for proc
22165 ++ * assumes size is at least one */
22166 ++
22167 ++int get_nid_list(int index, unsigned int *nids, int size)
22168 ++{
22169 ++ int hindex, nr_nids = 0;
22170 ++
22171 ++ /* only show current and children */
22172 ++ if (!nx_check(0, VS_ADMIN | VS_WATCH)) {
22173 ++ if (index > 0)
22174 ++ return 0;
22175 ++ nids[nr_nids] = nx_current_nid();
22176 ++ return 1;
22177 ++ }
22178 ++
22179 ++ for (hindex = 0; hindex < NX_HASH_SIZE; hindex++) {
22180 ++ struct hlist_head *head = &nx_info_hash[hindex];
22181 ++ struct hlist_node *pos;
22182 ++
22183 ++ spin_lock(&nx_info_hash_lock);
22184 ++ hlist_for_each(pos, head) {
22185 ++ struct nx_info *nxi;
22186 ++
22187 ++ if (--index > 0)
22188 ++ continue;
22189 ++
22190 ++ nxi = hlist_entry(pos, struct nx_info, nx_hlist);
22191 ++ nids[nr_nids] = nxi->nx_id;
22192 ++ if (++nr_nids >= size) {
22193 ++ spin_unlock(&nx_info_hash_lock);
22194 ++ goto out;
22195 ++ }
22196 ++ }
22197 ++ /* keep the lock time short */
22198 ++ spin_unlock(&nx_info_hash_lock);
22199 ++ }
22200 ++out:
22201 ++ return nr_nids;
22202 ++}
22203 ++#endif
22204 ++
22205 ++
22206 ++/*
22207 ++ * migrate task to new network
22208 ++ * gets nxi, puts old_nxi on change
22209 ++ */
22210 ++
22211 ++int nx_migrate_task(struct task_struct *p, struct nx_info *nxi)
22212 ++{
22213 ++ struct nx_info *old_nxi;
22214 ++ int ret = 0;
22215 ++
22216 ++ if (!p || !nxi)
22217 ++ BUG();
22218 ++
22219 ++ vxdprintk(VXD_CBIT(nid, 5),
22220 ++ "nx_migrate_task(%p,%p[#%d.%d.%d])",
22221 ++ p, nxi, nxi->nx_id,
22222 ++ atomic_read(&nxi->nx_usecnt),
22223 ++ atomic_read(&nxi->nx_tasks));
22224 ++
22225 ++ if (nx_info_flags(nxi, NXF_INFO_PRIVATE, 0) &&
22226 ++ !nx_info_flags(nxi, NXF_STATE_SETUP, 0))
22227 ++ return -EACCES;
22228 ++
22229 ++ if (nx_info_state(nxi, NXS_SHUTDOWN))
22230 ++ return -EFAULT;
22231 ++
22232 ++ /* maybe disallow this completely? */
22233 ++ old_nxi = task_get_nx_info(p);
22234 ++ if (old_nxi == nxi)
22235 ++ goto out;
22236 ++
22237 ++ task_lock(p);
22238 ++ if (old_nxi)
22239 ++ clr_nx_info(&p->nx_info);
22240 ++ claim_nx_info(nxi, p);
22241 ++ set_nx_info(&p->nx_info, nxi);
22242 ++ p->nid = nxi->nx_id;
22243 ++ task_unlock(p);
22244 ++
22245 ++ vxdprintk(VXD_CBIT(nid, 5),
22246 ++ "moved task %p into nxi:%p[#%d]",
22247 ++ p, nxi, nxi->nx_id);
22248 ++
22249 ++ if (old_nxi)
22250 ++ release_nx_info(old_nxi, p);
22251 ++ ret = 0;
22252 ++out:
22253 ++ put_nx_info(old_nxi);
22254 ++ return ret;
22255 ++}
22256 ++
22257 ++
22258 ++void nx_set_persistent(struct nx_info *nxi)
22259 ++{
22260 ++ vxdprintk(VXD_CBIT(nid, 6),
22261 ++ "nx_set_persistent(%p[#%d])", nxi, nxi->nx_id);
22262 ++
22263 ++ get_nx_info(nxi);
22264 ++ claim_nx_info(nxi, NULL);
22265 ++}
22266 ++
22267 ++void nx_clear_persistent(struct nx_info *nxi)
22268 ++{
22269 ++ vxdprintk(VXD_CBIT(nid, 6),
22270 ++ "nx_clear_persistent(%p[#%d])", nxi, nxi->nx_id);
22271 ++
22272 ++ release_nx_info(nxi, NULL);
22273 ++ put_nx_info(nxi);
22274 ++}
22275 ++
22276 ++void nx_update_persistent(struct nx_info *nxi)
22277 ++{
22278 ++ if (nx_info_flags(nxi, NXF_PERSISTENT, 0))
22279 ++ nx_set_persistent(nxi);
22280 ++ else
22281 ++ nx_clear_persistent(nxi);
22282 ++}
22283 ++
22284 ++/* vserver syscall commands below here */
22285 ++
22286 ++/* taks nid and nx_info functions */
22287 ++
22288 ++#include <asm/uaccess.h>
22289 ++
22290 ++
22291 ++int vc_task_nid(uint32_t id)
22292 ++{
22293 ++ nid_t nid;
22294 ++
22295 ++ if (id) {
22296 ++ struct task_struct *tsk;
22297 ++
22298 ++ read_lock(&tasklist_lock);
22299 ++ tsk = find_task_by_real_pid(id);
22300 ++ nid = (tsk) ? tsk->nid : -ESRCH;
22301 ++ read_unlock(&tasklist_lock);
22302 ++ } else
22303 ++ nid = nx_current_nid();
22304 ++ return nid;
22305 ++}
22306 ++
22307 ++
22308 ++int vc_nx_info(struct nx_info *nxi, void __user *data)
22309 ++{
22310 ++ struct vcmd_nx_info_v0 vc_data;
22311 ++
22312 ++ vc_data.nid = nxi->nx_id;
22313 ++
22314 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22315 ++ return -EFAULT;
22316 ++ return 0;
22317 ++}
22318 ++
22319 ++
22320 ++/* network functions */
22321 ++
22322 ++int vc_net_create(uint32_t nid, void __user *data)
22323 ++{
22324 ++ struct vcmd_net_create vc_data = { .flagword = NXF_INIT_SET };
22325 ++ struct nx_info *new_nxi;
22326 ++ int ret;
22327 ++
22328 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22329 ++ return -EFAULT;
22330 ++
22331 ++ if ((nid > MAX_S_CONTEXT) || (nid < 2))
22332 ++ return -EINVAL;
22333 ++
22334 ++ new_nxi = __create_nx_info(nid);
22335 ++ if (IS_ERR(new_nxi))
22336 ++ return PTR_ERR(new_nxi);
22337 ++
22338 ++ /* initial flags */
22339 ++ new_nxi->nx_flags = vc_data.flagword;
22340 ++
22341 ++ ret = -ENOEXEC;
22342 ++ if (vs_net_change(new_nxi, VSC_NETUP))
22343 ++ goto out;
22344 ++
22345 ++ ret = nx_migrate_task(current, new_nxi);
22346 ++ if (ret)
22347 ++ goto out;
22348 ++
22349 ++ /* return context id on success */
22350 ++ ret = new_nxi->nx_id;
22351 ++
22352 ++ /* get a reference for persistent contexts */
22353 ++ if ((vc_data.flagword & NXF_PERSISTENT))
22354 ++ nx_set_persistent(new_nxi);
22355 ++out:
22356 ++ release_nx_info(new_nxi, NULL);
22357 ++ put_nx_info(new_nxi);
22358 ++ return ret;
22359 ++}
22360 ++
22361 ++
22362 ++int vc_net_migrate(struct nx_info *nxi, void __user *data)
22363 ++{
22364 ++ return nx_migrate_task(current, nxi);
22365 ++}
22366 ++
22367 ++
22368 ++
22369 ++int do_add_v4_addr(struct nx_info *nxi, __be32 ip, __be32 ip2, __be32 mask,
22370 ++ uint16_t type, uint16_t flags)
22371 ++{
22372 ++ struct nx_addr_v4 *nxa = &nxi->v4;
22373 ++
22374 ++ if (NX_IPV4(nxi)) {
22375 ++ /* locate last entry */
22376 ++ for (; nxa->next; nxa = nxa->next);
22377 ++ nxa->next = __alloc_nx_addr_v4();
22378 ++ nxa = nxa->next;
22379 ++
22380 ++ if (IS_ERR(nxa))
22381 ++ return PTR_ERR(nxa);
22382 ++ }
22383 ++
22384 ++ if (nxi->v4.next)
22385 ++ /* remove single ip for ip list */
22386 ++ nxi->nx_flags &= ~NXF_SINGLE_IP;
22387 ++
22388 ++ nxa->ip[0].s_addr = ip;
22389 ++ nxa->ip[1].s_addr = ip2;
22390 ++ nxa->mask.s_addr = mask;
22391 ++ nxa->type = type;
22392 ++ nxa->flags = flags;
22393 ++ return 0;
22394 ++}
22395 ++
22396 ++
22397 ++int vc_net_add(struct nx_info *nxi, void __user *data)
22398 ++{
22399 ++ struct vcmd_net_addr_v0 vc_data;
22400 ++ int index, ret = 0;
22401 ++
22402 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22403 ++ return -EFAULT;
22404 ++
22405 ++ switch (vc_data.type) {
22406 ++ case NXA_TYPE_IPV4:
22407 ++ if ((vc_data.count < 1) || (vc_data.count > 4))
22408 ++ return -EINVAL;
22409 ++
22410 ++ index = 0;
22411 ++ while (index < vc_data.count) {
22412 ++ ret = do_add_v4_addr(nxi, vc_data.ip[index].s_addr, 0,
22413 ++ vc_data.mask[index].s_addr, NXA_TYPE_ADDR, 0);
22414 ++ if (ret)
22415 ++ return ret;
22416 ++ index++;
22417 ++ }
22418 ++ ret = index;
22419 ++ break;
22420 ++
22421 ++ case NXA_TYPE_IPV4|NXA_MOD_BCAST:
22422 ++ nxi->v4_bcast = vc_data.ip[0];
22423 ++ ret = 1;
22424 ++ break;
22425 ++
22426 ++ case NXA_TYPE_IPV4|NXA_MOD_LBACK:
22427 ++ nxi->v4_lback = vc_data.ip[0];
22428 ++ ret = 1;
22429 ++ break;
22430 ++
22431 ++ default:
22432 ++ ret = -EINVAL;
22433 ++ break;
22434 ++ }
22435 ++ return ret;
22436 ++}
22437 ++
22438 ++int vc_net_remove(struct nx_info *nxi, void __user *data)
22439 ++{
22440 ++ struct vcmd_net_addr_v0 vc_data;
22441 ++
22442 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22443 ++ return -EFAULT;
22444 ++
22445 ++ switch (vc_data.type) {
22446 ++ case NXA_TYPE_ANY:
22447 ++ __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22448 ++ memset(&nxi->v4, 0, sizeof(nxi->v4));
22449 ++ break;
22450 ++
22451 ++ default:
22452 ++ return -EINVAL;
22453 ++ }
22454 ++ return 0;
22455 ++}
22456 ++
22457 ++
22458 ++int vc_net_add_ipv4(struct nx_info *nxi, void __user *data)
22459 ++{
22460 ++ struct vcmd_net_addr_ipv4_v1 vc_data;
22461 ++
22462 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22463 ++ return -EFAULT;
22464 ++
22465 ++ switch (vc_data.type) {
22466 ++ case NXA_TYPE_ADDR:
22467 ++ case NXA_TYPE_RANGE:
22468 ++ case NXA_TYPE_MASK:
22469 ++ return do_add_v4_addr(nxi, vc_data.ip.s_addr, 0,
22470 ++ vc_data.mask.s_addr, vc_data.type, vc_data.flags);
22471 ++
22472 ++ case NXA_TYPE_ADDR | NXA_MOD_BCAST:
22473 ++ nxi->v4_bcast = vc_data.ip;
22474 ++ break;
22475 ++
22476 ++ case NXA_TYPE_ADDR | NXA_MOD_LBACK:
22477 ++ nxi->v4_lback = vc_data.ip;
22478 ++ break;
22479 ++
22480 ++ default:
22481 ++ return -EINVAL;
22482 ++ }
22483 ++ return 0;
22484 ++}
22485 ++
22486 ++int vc_net_remove_ipv4(struct nx_info *nxi, void __user *data)
22487 ++{
22488 ++ struct vcmd_net_addr_ipv4_v1 vc_data;
22489 ++
22490 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22491 ++ return -EFAULT;
22492 ++
22493 ++ switch (vc_data.type) {
22494 ++/* case NXA_TYPE_ADDR:
22495 ++ break; */
22496 ++
22497 ++ case NXA_TYPE_ANY:
22498 ++ __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
22499 ++ memset(&nxi->v4, 0, sizeof(nxi->v4));
22500 ++ break;
22501 ++
22502 ++ default:
22503 ++ return -EINVAL;
22504 ++ }
22505 ++ return 0;
22506 ++}
22507 ++
22508 ++
22509 ++#ifdef CONFIG_IPV6
22510 ++
22511 ++int do_add_v6_addr(struct nx_info *nxi,
22512 ++ struct in6_addr *ip, struct in6_addr *mask,
22513 ++ uint32_t prefix, uint16_t type, uint16_t flags)
22514 ++{
22515 ++ struct nx_addr_v6 *nxa = &nxi->v6;
22516 ++
22517 ++ if (NX_IPV6(nxi)) {
22518 ++ /* locate last entry */
22519 ++ for (; nxa->next; nxa = nxa->next);
22520 ++ nxa->next = __alloc_nx_addr_v6();
22521 ++ nxa = nxa->next;
22522 ++
22523 ++ if (IS_ERR(nxa))
22524 ++ return PTR_ERR(nxa);
22525 ++ }
22526 ++
22527 ++ nxa->ip = *ip;
22528 ++ nxa->mask = *mask;
22529 ++ nxa->prefix = prefix;
22530 ++ nxa->type = type;
22531 ++ nxa->flags = flags;
22532 ++ return 0;
22533 ++}
22534 ++
22535 ++
22536 ++int vc_net_add_ipv6(struct nx_info *nxi, void __user *data)
22537 ++{
22538 ++ struct vcmd_net_addr_ipv6_v1 vc_data;
22539 ++
22540 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22541 ++ return -EFAULT;
22542 ++
22543 ++ switch (vc_data.type) {
22544 ++ case NXA_TYPE_ADDR:
22545 ++ case NXA_TYPE_MASK:
22546 ++ return do_add_v6_addr(nxi, &vc_data.ip, &vc_data.mask,
22547 ++ vc_data.prefix, vc_data.type, vc_data.flags);
22548 ++ default:
22549 ++ return -EINVAL;
22550 ++ }
22551 ++ return 0;
22552 ++}
22553 ++
22554 ++int vc_net_remove_ipv6(struct nx_info *nxi, void __user *data)
22555 ++{
22556 ++ struct vcmd_net_addr_ipv6_v1 vc_data;
22557 ++
22558 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
22559 ++ return -EFAULT;
22560 ++
22561 ++ switch (vc_data.type) {
22562 ++ case NXA_TYPE_ANY:
22563 ++ __dealloc_nx_addr_v6_all(xchg(&nxi->v6.next, NULL));
22564 ++ memset(&nxi->v6, 0, sizeof(nxi->v6));
22565 ++ break;
22566 ++
22567 ++ default:
22568 ++ return -EINVAL;
22569 ++ }
22570 ++ return 0;
22571 ++}
22572 ++
22573 ++#endif /* CONFIG_IPV6 */
22574 ++
22575 ++
22576 ++int vc_get_nflags(struct nx_info *nxi, void __user *data)
22577 ++{
22578 ++ struct vcmd_net_flags_v0 vc_data;
22579 ++
22580 ++ vc_data.flagword = nxi->nx_flags;
22581 ++
22582 ++ /* special STATE flag handling */
22583 ++ vc_data.mask = vs_mask_flags(~0ULL, nxi->nx_flags, NXF_ONE_TIME);
22584 ++
22585 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22586 ++ return -EFAULT;
22587 ++ return 0;
22588 ++}
22589 ++
22590 ++int vc_set_nflags(struct nx_info *nxi, void __user *data)
22591 ++{
22592 ++ struct vcmd_net_flags_v0 vc_data;
22593 ++ uint64_t mask, trigger;
22594 ++
22595 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22596 ++ return -EFAULT;
22597 ++
22598 ++ /* special STATE flag handling */
22599 ++ mask = vs_mask_mask(vc_data.mask, nxi->nx_flags, NXF_ONE_TIME);
22600 ++ trigger = (mask & nxi->nx_flags) ^ (mask & vc_data.flagword);
22601 ++
22602 ++ nxi->nx_flags = vs_mask_flags(nxi->nx_flags,
22603 ++ vc_data.flagword, mask);
22604 ++ if (trigger & NXF_PERSISTENT)
22605 ++ nx_update_persistent(nxi);
22606 ++
22607 ++ return 0;
22608 ++}
22609 ++
22610 ++int vc_get_ncaps(struct nx_info *nxi, void __user *data)
22611 ++{
22612 ++ struct vcmd_net_caps_v0 vc_data;
22613 ++
22614 ++ vc_data.ncaps = nxi->nx_ncaps;
22615 ++ vc_data.cmask = ~0ULL;
22616 ++
22617 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22618 ++ return -EFAULT;
22619 ++ return 0;
22620 ++}
22621 ++
22622 ++int vc_set_ncaps(struct nx_info *nxi, void __user *data)
22623 ++{
22624 ++ struct vcmd_net_caps_v0 vc_data;
22625 ++
22626 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22627 ++ return -EFAULT;
22628 ++
22629 ++ nxi->nx_ncaps = vs_mask_flags(nxi->nx_ncaps,
22630 ++ vc_data.ncaps, vc_data.cmask);
22631 ++ return 0;
22632 ++}
22633 ++
22634 ++
22635 ++#include <linux/module.h>
22636 ++
22637 ++module_init(init_network);
22638 ++
22639 ++EXPORT_SYMBOL_GPL(free_nx_info);
22640 ++EXPORT_SYMBOL_GPL(unhash_nx_info);
22641 ++
22642 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/proc.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/proc.c
22643 +--- linux-2.6.29.4/kernel/vserver/proc.c 1970-01-01 01:00:00.000000000 +0100
22644 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/proc.c 2009-05-23 22:57:16.000000000 +0200
22645 +@@ -0,0 +1,1096 @@
22646 ++/*
22647 ++ * linux/kernel/vserver/proc.c
22648 ++ *
22649 ++ * Virtual Context Support
22650 ++ *
22651 ++ * Copyright (C) 2003-2007 Herbert Pötzl
22652 ++ *
22653 ++ * V0.01 basic structure
22654 ++ * V0.02 adaptation vs1.3.0
22655 ++ * V0.03 proc permissions
22656 ++ * V0.04 locking/generic
22657 ++ * V0.05 next generation procfs
22658 ++ * V0.06 inode validation
22659 ++ * V0.07 generic rewrite vid
22660 ++ * V0.08 remove inode type
22661 ++ *
22662 ++ */
22663 ++
22664 ++#include <linux/proc_fs.h>
22665 ++#include <asm/unistd.h>
22666 ++
22667 ++#include <linux/vs_context.h>
22668 ++#include <linux/vs_network.h>
22669 ++#include <linux/vs_cvirt.h>
22670 ++
22671 ++#include <linux/in.h>
22672 ++#include <linux/inetdevice.h>
22673 ++#include <linux/vs_inet.h>
22674 ++#include <linux/vs_inet6.h>
22675 ++
22676 ++#include <linux/vserver/global.h>
22677 ++
22678 ++#include "cvirt_proc.h"
22679 ++#include "cacct_proc.h"
22680 ++#include "limit_proc.h"
22681 ++#include "sched_proc.h"
22682 ++#include "vci_config.h"
22683 ++
22684 ++
22685 ++static inline char *print_cap_t(char *buffer, kernel_cap_t *c)
22686 ++{
22687 ++ unsigned __capi;
22688 ++
22689 ++ CAP_FOR_EACH_U32(__capi) {
22690 ++ buffer += sprintf(buffer, "%08x",
22691 ++ c->cap[(_KERNEL_CAPABILITY_U32S-1) - __capi]);
22692 ++ }
22693 ++ return buffer;
22694 ++}
22695 ++
22696 ++
22697 ++static struct proc_dir_entry *proc_virtual;
22698 ++
22699 ++static struct proc_dir_entry *proc_virtnet;
22700 ++
22701 ++
22702 ++/* first the actual feeds */
22703 ++
22704 ++
22705 ++static int proc_vci(char *buffer)
22706 ++{
22707 ++ return sprintf(buffer,
22708 ++ "VCIVersion:\t%04x:%04x\n"
22709 ++ "VCISyscall:\t%d\n"
22710 ++ "VCIKernel:\t%08x\n",
22711 ++ VCI_VERSION >> 16,
22712 ++ VCI_VERSION & 0xFFFF,
22713 ++ __NR_vserver,
22714 ++ vci_kernel_config());
22715 ++}
22716 ++
22717 ++static int proc_virtual_info(char *buffer)
22718 ++{
22719 ++ return proc_vci(buffer);
22720 ++}
22721 ++
22722 ++static int proc_virtual_status(char *buffer)
22723 ++{
22724 ++ return sprintf(buffer,
22725 ++ "#CTotal:\t%d\n"
22726 ++ "#CActive:\t%d\n"
22727 ++ "#NSProxy:\t%d\t%d %d %d %d %d %d\n"
22728 ++ "#InitTask:\t%d\t%d %d\n",
22729 ++ atomic_read(&vx_global_ctotal),
22730 ++ atomic_read(&vx_global_cactive),
22731 ++ atomic_read(&vs_global_nsproxy),
22732 ++ atomic_read(&vs_global_fs),
22733 ++ atomic_read(&vs_global_mnt_ns),
22734 ++ atomic_read(&vs_global_uts_ns),
22735 ++ atomic_read(&vs_global_ipc_ns),
22736 ++ atomic_read(&vs_global_user_ns),
22737 ++ atomic_read(&vs_global_pid_ns),
22738 ++ atomic_read(&init_task.usage),
22739 ++ atomic_read(&init_task.nsproxy->count),
22740 ++ init_task.fs->users);
22741 ++}
22742 ++
22743 ++
22744 ++int proc_vxi_info(struct vx_info *vxi, char *buffer)
22745 ++{
22746 ++ int length;
22747 ++
22748 ++ length = sprintf(buffer,
22749 ++ "ID:\t%d\n"
22750 ++ "Info:\t%p\n"
22751 ++ "Init:\t%d\n"
22752 ++ "OOM:\t%lld\n",
22753 ++ vxi->vx_id,
22754 ++ vxi,
22755 ++ vxi->vx_initpid,
22756 ++ vxi->vx_badness_bias);
22757 ++ return length;
22758 ++}
22759 ++
22760 ++int proc_vxi_status(struct vx_info *vxi, char *buffer)
22761 ++{
22762 ++ char *orig = buffer;
22763 ++
22764 ++ buffer += sprintf(buffer,
22765 ++ "UseCnt:\t%d\n"
22766 ++ "Tasks:\t%d\n"
22767 ++ "Flags:\t%016llx\n",
22768 ++ atomic_read(&vxi->vx_usecnt),
22769 ++ atomic_read(&vxi->vx_tasks),
22770 ++ (unsigned long long)vxi->vx_flags);
22771 ++
22772 ++ buffer += sprintf(buffer, "BCaps:\t");
22773 ++ buffer = print_cap_t(buffer, &vxi->vx_bcaps);
22774 ++ buffer += sprintf(buffer, "\n");
22775 ++
22776 ++ buffer += sprintf(buffer,
22777 ++ "CCaps:\t%016llx\n"
22778 ++ "Spaces:\t%08lx %08lx\n",
22779 ++ (unsigned long long)vxi->vx_ccaps,
22780 ++ vxi->vx_nsmask[0], vxi->vx_nsmask[1]);
22781 ++ return buffer - orig;
22782 ++}
22783 ++
22784 ++int proc_vxi_limit(struct vx_info *vxi, char *buffer)
22785 ++{
22786 ++ return vx_info_proc_limit(&vxi->limit, buffer);
22787 ++}
22788 ++
22789 ++int proc_vxi_sched(struct vx_info *vxi, char *buffer)
22790 ++{
22791 ++ int cpu, length;
22792 ++
22793 ++ length = vx_info_proc_sched(&vxi->sched, buffer);
22794 ++ for_each_online_cpu(cpu) {
22795 ++ length += vx_info_proc_sched_pc(
22796 ++ &vx_per_cpu(vxi, sched_pc, cpu),
22797 ++ buffer + length, cpu);
22798 ++ }
22799 ++ return length;
22800 ++}
22801 ++
22802 ++int proc_vxi_nsproxy0(struct vx_info *vxi, char *buffer)
22803 ++{
22804 ++ return vx_info_proc_nsproxy(vxi->vx_nsproxy[0], buffer);
22805 ++}
22806 ++
22807 ++int proc_vxi_nsproxy1(struct vx_info *vxi, char *buffer)
22808 ++{
22809 ++ return vx_info_proc_nsproxy(vxi->vx_nsproxy[1], buffer);
22810 ++}
22811 ++
22812 ++int proc_vxi_cvirt(struct vx_info *vxi, char *buffer)
22813 ++{
22814 ++ int cpu, length;
22815 ++
22816 ++ vx_update_load(vxi);
22817 ++ length = vx_info_proc_cvirt(&vxi->cvirt, buffer);
22818 ++ for_each_online_cpu(cpu) {
22819 ++ length += vx_info_proc_cvirt_pc(
22820 ++ &vx_per_cpu(vxi, cvirt_pc, cpu),
22821 ++ buffer + length, cpu);
22822 ++ }
22823 ++ return length;
22824 ++}
22825 ++
22826 ++int proc_vxi_cacct(struct vx_info *vxi, char *buffer)
22827 ++{
22828 ++ return vx_info_proc_cacct(&vxi->cacct, buffer);
22829 ++}
22830 ++
22831 ++
22832 ++static int proc_virtnet_info(char *buffer)
22833 ++{
22834 ++ return proc_vci(buffer);
22835 ++}
22836 ++
22837 ++static int proc_virtnet_status(char *buffer)
22838 ++{
22839 ++ return sprintf(buffer,
22840 ++ "#CTotal:\t%d\n"
22841 ++ "#CActive:\t%d\n",
22842 ++ atomic_read(&nx_global_ctotal),
22843 ++ atomic_read(&nx_global_cactive));
22844 ++}
22845 ++
22846 ++int proc_nxi_info(struct nx_info *nxi, char *buffer)
22847 ++{
22848 ++ struct nx_addr_v4 *v4a;
22849 ++#ifdef CONFIG_IPV6
22850 ++ struct nx_addr_v6 *v6a;
22851 ++#endif
22852 ++ int length, i;
22853 ++
22854 ++ length = sprintf(buffer,
22855 ++ "ID:\t%d\n"
22856 ++ "Info:\t%p\n"
22857 ++ "Bcast:\t" NIPQUAD_FMT "\n"
22858 ++ "Lback:\t" NIPQUAD_FMT "\n",
22859 ++ nxi->nx_id,
22860 ++ nxi,
22861 ++ NIPQUAD(nxi->v4_bcast.s_addr),
22862 ++ NIPQUAD(nxi->v4_lback.s_addr));
22863 ++
22864 ++ if (!NX_IPV4(nxi))
22865 ++ goto skip_v4;
22866 ++ for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
22867 ++ length += sprintf(buffer + length, "%d:\t" NXAV4_FMT "\n",
22868 ++ i, NXAV4(v4a));
22869 ++skip_v4:
22870 ++#ifdef CONFIG_IPV6
22871 ++ if (!NX_IPV6(nxi))
22872 ++ goto skip_v6;
22873 ++ for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
22874 ++ length += sprintf(buffer + length, "%d:\t" NXAV6_FMT "\n",
22875 ++ i, NXAV6(v6a));
22876 ++skip_v6:
22877 ++#endif
22878 ++ return length;
22879 ++}
22880 ++
22881 ++int proc_nxi_status(struct nx_info *nxi, char *buffer)
22882 ++{
22883 ++ int length;
22884 ++
22885 ++ length = sprintf(buffer,
22886 ++ "UseCnt:\t%d\n"
22887 ++ "Tasks:\t%d\n"
22888 ++ "Flags:\t%016llx\n"
22889 ++ "NCaps:\t%016llx\n",
22890 ++ atomic_read(&nxi->nx_usecnt),
22891 ++ atomic_read(&nxi->nx_tasks),
22892 ++ (unsigned long long)nxi->nx_flags,
22893 ++ (unsigned long long)nxi->nx_ncaps);
22894 ++ return length;
22895 ++}
22896 ++
22897 ++
22898 ++
22899 ++/* here the inode helpers */
22900 ++
22901 ++struct vs_entry {
22902 ++ int len;
22903 ++ char *name;
22904 ++ mode_t mode;
22905 ++ struct inode_operations *iop;
22906 ++ struct file_operations *fop;
22907 ++ union proc_op op;
22908 ++};
22909 ++
22910 ++static struct inode *vs_proc_make_inode(struct super_block *sb, struct vs_entry *p)
22911 ++{
22912 ++ struct inode *inode = new_inode(sb);
22913 ++
22914 ++ if (!inode)
22915 ++ goto out;
22916 ++
22917 ++ inode->i_mode = p->mode;
22918 ++ if (p->iop)
22919 ++ inode->i_op = p->iop;
22920 ++ if (p->fop)
22921 ++ inode->i_fop = p->fop;
22922 ++
22923 ++ inode->i_nlink = (p->mode & S_IFDIR) ? 2 : 1;
22924 ++ inode->i_flags |= S_IMMUTABLE;
22925 ++
22926 ++ inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
22927 ++
22928 ++ inode->i_uid = 0;
22929 ++ inode->i_gid = 0;
22930 ++ inode->i_tag = 0;
22931 ++out:
22932 ++ return inode;
22933 ++}
22934 ++
22935 ++static struct dentry *vs_proc_instantiate(struct inode *dir,
22936 ++ struct dentry *dentry, int id, void *ptr)
22937 ++{
22938 ++ struct vs_entry *p = ptr;
22939 ++ struct inode *inode = vs_proc_make_inode(dir->i_sb, p);
22940 ++ struct dentry *error = ERR_PTR(-EINVAL);
22941 ++
22942 ++ if (!inode)
22943 ++ goto out;
22944 ++
22945 ++ PROC_I(inode)->op = p->op;
22946 ++ PROC_I(inode)->fd = id;
22947 ++ d_add(dentry, inode);
22948 ++ error = NULL;
22949 ++out:
22950 ++ return error;
22951 ++}
22952 ++
22953 ++/* Lookups */
22954 ++
22955 ++typedef struct dentry *instantiate_t(struct inode *, struct dentry *, int, void *);
22956 ++
22957 ++/*
22958 ++ * Fill a directory entry.
22959 ++ *
22960 ++ * If possible create the dcache entry and derive our inode number and
22961 ++ * file type from dcache entry.
22962 ++ *
22963 ++ * Since all of the proc inode numbers are dynamically generated, the inode
22964 ++ * numbers do not exist until the inode is cache. This means creating the
22965 ++ * the dcache entry in readdir is necessary to keep the inode numbers
22966 ++ * reported by readdir in sync with the inode numbers reported
22967 ++ * by stat.
22968 ++ */
22969 ++static int proc_fill_cache(struct file *filp, void *dirent, filldir_t filldir,
22970 ++ char *name, int len, instantiate_t instantiate, int id, void *ptr)
22971 ++{
22972 ++ struct dentry *child, *dir = filp->f_dentry;
22973 ++ struct inode *inode;
22974 ++ struct qstr qname;
22975 ++ ino_t ino = 0;
22976 ++ unsigned type = DT_UNKNOWN;
22977 ++
22978 ++ qname.name = name;
22979 ++ qname.len = len;
22980 ++ qname.hash = full_name_hash(name, len);
22981 ++
22982 ++ child = d_lookup(dir, &qname);
22983 ++ if (!child) {
22984 ++ struct dentry *new;
22985 ++ new = d_alloc(dir, &qname);
22986 ++ if (new) {
22987 ++ child = instantiate(dir->d_inode, new, id, ptr);
22988 ++ if (child)
22989 ++ dput(new);
22990 ++ else
22991 ++ child = new;
22992 ++ }
22993 ++ }
22994 ++ if (!child || IS_ERR(child) || !child->d_inode)
22995 ++ goto end_instantiate;
22996 ++ inode = child->d_inode;
22997 ++ if (inode) {
22998 ++ ino = inode->i_ino;
22999 ++ type = inode->i_mode >> 12;
23000 ++ }
23001 ++ dput(child);
23002 ++end_instantiate:
23003 ++ if (!ino)
23004 ++ ino = find_inode_number(dir, &qname);
23005 ++ if (!ino)
23006 ++ ino = 1;
23007 ++ return filldir(dirent, name, len, filp->f_pos, ino, type);
23008 ++}
23009 ++
23010 ++
23011 ++
23012 ++/* get and revalidate vx_info/xid */
23013 ++
23014 ++static inline
23015 ++struct vx_info *get_proc_vx_info(struct inode *inode)
23016 ++{
23017 ++ return lookup_vx_info(PROC_I(inode)->fd);
23018 ++}
23019 ++
23020 ++static int proc_xid_revalidate(struct dentry *dentry, struct nameidata *nd)
23021 ++{
23022 ++ struct inode *inode = dentry->d_inode;
23023 ++ xid_t xid = PROC_I(inode)->fd;
23024 ++
23025 ++ if (!xid || xid_is_hashed(xid))
23026 ++ return 1;
23027 ++ d_drop(dentry);
23028 ++ return 0;
23029 ++}
23030 ++
23031 ++
23032 ++/* get and revalidate nx_info/nid */
23033 ++
23034 ++static int proc_nid_revalidate(struct dentry *dentry, struct nameidata *nd)
23035 ++{
23036 ++ struct inode *inode = dentry->d_inode;
23037 ++ nid_t nid = PROC_I(inode)->fd;
23038 ++
23039 ++ if (!nid || nid_is_hashed(nid))
23040 ++ return 1;
23041 ++ d_drop(dentry);
23042 ++ return 0;
23043 ++}
23044 ++
23045 ++
23046 ++
23047 ++#define PROC_BLOCK_SIZE (PAGE_SIZE - 1024)
23048 ++
23049 ++static ssize_t proc_vs_info_read(struct file *file, char __user *buf,
23050 ++ size_t count, loff_t *ppos)
23051 ++{
23052 ++ struct inode *inode = file->f_dentry->d_inode;
23053 ++ unsigned long page;
23054 ++ ssize_t length = 0;
23055 ++
23056 ++ if (count > PROC_BLOCK_SIZE)
23057 ++ count = PROC_BLOCK_SIZE;
23058 ++
23059 ++ /* fade that out as soon as stable */
23060 ++ WARN_ON(PROC_I(inode)->fd);
23061 ++
23062 ++ if (!(page = __get_free_page(GFP_KERNEL)))
23063 ++ return -ENOMEM;
23064 ++
23065 ++ BUG_ON(!PROC_I(inode)->op.proc_vs_read);
23066 ++ length = PROC_I(inode)->op.proc_vs_read((char *)page);
23067 ++
23068 ++ if (length >= 0)
23069 ++ length = simple_read_from_buffer(buf, count, ppos,
23070 ++ (char *)page, length);
23071 ++
23072 ++ free_page(page);
23073 ++ return length;
23074 ++}
23075 ++
23076 ++static ssize_t proc_vx_info_read(struct file *file, char __user *buf,
23077 ++ size_t count, loff_t *ppos)
23078 ++{
23079 ++ struct inode *inode = file->f_dentry->d_inode;
23080 ++ struct vx_info *vxi = NULL;
23081 ++ xid_t xid = PROC_I(inode)->fd;
23082 ++ unsigned long page;
23083 ++ ssize_t length = 0;
23084 ++
23085 ++ if (count > PROC_BLOCK_SIZE)
23086 ++ count = PROC_BLOCK_SIZE;
23087 ++
23088 ++ /* fade that out as soon as stable */
23089 ++ WARN_ON(!xid);
23090 ++ vxi = lookup_vx_info(xid);
23091 ++ if (!vxi)
23092 ++ goto out;
23093 ++
23094 ++ length = -ENOMEM;
23095 ++ if (!(page = __get_free_page(GFP_KERNEL)))
23096 ++ goto out_put;
23097 ++
23098 ++ BUG_ON(!PROC_I(inode)->op.proc_vxi_read);
23099 ++ length = PROC_I(inode)->op.proc_vxi_read(vxi, (char *)page);
23100 ++
23101 ++ if (length >= 0)
23102 ++ length = simple_read_from_buffer(buf, count, ppos,
23103 ++ (char *)page, length);
23104 ++
23105 ++ free_page(page);
23106 ++out_put:
23107 ++ put_vx_info(vxi);
23108 ++out:
23109 ++ return length;
23110 ++}
23111 ++
23112 ++static ssize_t proc_nx_info_read(struct file *file, char __user *buf,
23113 ++ size_t count, loff_t *ppos)
23114 ++{
23115 ++ struct inode *inode = file->f_dentry->d_inode;
23116 ++ struct nx_info *nxi = NULL;
23117 ++ nid_t nid = PROC_I(inode)->fd;
23118 ++ unsigned long page;
23119 ++ ssize_t length = 0;
23120 ++
23121 ++ if (count > PROC_BLOCK_SIZE)
23122 ++ count = PROC_BLOCK_SIZE;
23123 ++
23124 ++ /* fade that out as soon as stable */
23125 ++ WARN_ON(!nid);
23126 ++ nxi = lookup_nx_info(nid);
23127 ++ if (!nxi)
23128 ++ goto out;
23129 ++
23130 ++ length = -ENOMEM;
23131 ++ if (!(page = __get_free_page(GFP_KERNEL)))
23132 ++ goto out_put;
23133 ++
23134 ++ BUG_ON(!PROC_I(inode)->op.proc_nxi_read);
23135 ++ length = PROC_I(inode)->op.proc_nxi_read(nxi, (char *)page);
23136 ++
23137 ++ if (length >= 0)
23138 ++ length = simple_read_from_buffer(buf, count, ppos,
23139 ++ (char *)page, length);
23140 ++
23141 ++ free_page(page);
23142 ++out_put:
23143 ++ put_nx_info(nxi);
23144 ++out:
23145 ++ return length;
23146 ++}
23147 ++
23148 ++
23149 ++
23150 ++/* here comes the lower level */
23151 ++
23152 ++
23153 ++#define NOD(NAME, MODE, IOP, FOP, OP) { \
23154 ++ .len = sizeof(NAME) - 1, \
23155 ++ .name = (NAME), \
23156 ++ .mode = MODE, \
23157 ++ .iop = IOP, \
23158 ++ .fop = FOP, \
23159 ++ .op = OP, \
23160 ++}
23161 ++
23162 ++
23163 ++#define DIR(NAME, MODE, OTYPE) \
23164 ++ NOD(NAME, (S_IFDIR | (MODE)), \
23165 ++ &proc_ ## OTYPE ## _inode_operations, \
23166 ++ &proc_ ## OTYPE ## _file_operations, { } )
23167 ++
23168 ++#define INF(NAME, MODE, OTYPE) \
23169 ++ NOD(NAME, (S_IFREG | (MODE)), NULL, \
23170 ++ &proc_vs_info_file_operations, \
23171 ++ { .proc_vs_read = &proc_##OTYPE } )
23172 ++
23173 ++#define VINF(NAME, MODE, OTYPE) \
23174 ++ NOD(NAME, (S_IFREG | (MODE)), NULL, \
23175 ++ &proc_vx_info_file_operations, \
23176 ++ { .proc_vxi_read = &proc_##OTYPE } )
23177 ++
23178 ++#define NINF(NAME, MODE, OTYPE) \
23179 ++ NOD(NAME, (S_IFREG | (MODE)), NULL, \
23180 ++ &proc_nx_info_file_operations, \
23181 ++ { .proc_nxi_read = &proc_##OTYPE } )
23182 ++
23183 ++
23184 ++static struct file_operations proc_vs_info_file_operations = {
23185 ++ .read = proc_vs_info_read,
23186 ++};
23187 ++
23188 ++static struct file_operations proc_vx_info_file_operations = {
23189 ++ .read = proc_vx_info_read,
23190 ++};
23191 ++
23192 ++static struct dentry_operations proc_xid_dentry_operations = {
23193 ++ .d_revalidate = proc_xid_revalidate,
23194 ++};
23195 ++
23196 ++static struct vs_entry vx_base_stuff[] = {
23197 ++ VINF("info", S_IRUGO, vxi_info),
23198 ++ VINF("status", S_IRUGO, vxi_status),
23199 ++ VINF("limit", S_IRUGO, vxi_limit),
23200 ++ VINF("sched", S_IRUGO, vxi_sched),
23201 ++ VINF("nsproxy", S_IRUGO, vxi_nsproxy0),
23202 ++ VINF("nsproxy1",S_IRUGO, vxi_nsproxy1),
23203 ++ VINF("cvirt", S_IRUGO, vxi_cvirt),
23204 ++ VINF("cacct", S_IRUGO, vxi_cacct),
23205 ++ {}
23206 ++};
23207 ++
23208 ++
23209 ++
23210 ++
23211 ++static struct dentry *proc_xid_instantiate(struct inode *dir,
23212 ++ struct dentry *dentry, int id, void *ptr)
23213 ++{
23214 ++ dentry->d_op = &proc_xid_dentry_operations;
23215 ++ return vs_proc_instantiate(dir, dentry, id, ptr);
23216 ++}
23217 ++
23218 ++static struct dentry *proc_xid_lookup(struct inode *dir,
23219 ++ struct dentry *dentry, struct nameidata *nd)
23220 ++{
23221 ++ struct vs_entry *p = vx_base_stuff;
23222 ++ struct dentry *error = ERR_PTR(-ENOENT);
23223 ++
23224 ++ for (; p->name; p++) {
23225 ++ if (p->len != dentry->d_name.len)
23226 ++ continue;
23227 ++ if (!memcmp(dentry->d_name.name, p->name, p->len))
23228 ++ break;
23229 ++ }
23230 ++ if (!p->name)
23231 ++ goto out;
23232 ++
23233 ++ error = proc_xid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23234 ++out:
23235 ++ return error;
23236 ++}
23237 ++
23238 ++static int proc_xid_readdir(struct file *filp,
23239 ++ void *dirent, filldir_t filldir)
23240 ++{
23241 ++ struct dentry *dentry = filp->f_dentry;
23242 ++ struct inode *inode = dentry->d_inode;
23243 ++ struct vs_entry *p = vx_base_stuff;
23244 ++ int size = sizeof(vx_base_stuff) / sizeof(struct vs_entry);
23245 ++ int pos, index;
23246 ++ u64 ino;
23247 ++
23248 ++ pos = filp->f_pos;
23249 ++ switch (pos) {
23250 ++ case 0:
23251 ++ ino = inode->i_ino;
23252 ++ if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23253 ++ goto out;
23254 ++ pos++;
23255 ++ /* fall through */
23256 ++ case 1:
23257 ++ ino = parent_ino(dentry);
23258 ++ if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23259 ++ goto out;
23260 ++ pos++;
23261 ++ /* fall through */
23262 ++ default:
23263 ++ index = pos - 2;
23264 ++ if (index >= size)
23265 ++ goto out;
23266 ++ for (p += index; p->name; p++) {
23267 ++ if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23268 ++ vs_proc_instantiate, PROC_I(inode)->fd, p))
23269 ++ goto out;
23270 ++ pos++;
23271 ++ }
23272 ++ }
23273 ++out:
23274 ++ filp->f_pos = pos;
23275 ++ return 1;
23276 ++}
23277 ++
23278 ++
23279 ++
23280 ++static struct file_operations proc_nx_info_file_operations = {
23281 ++ .read = proc_nx_info_read,
23282 ++};
23283 ++
23284 ++static struct dentry_operations proc_nid_dentry_operations = {
23285 ++ .d_revalidate = proc_nid_revalidate,
23286 ++};
23287 ++
23288 ++static struct vs_entry nx_base_stuff[] = {
23289 ++ NINF("info", S_IRUGO, nxi_info),
23290 ++ NINF("status", S_IRUGO, nxi_status),
23291 ++ {}
23292 ++};
23293 ++
23294 ++
23295 ++static struct dentry *proc_nid_instantiate(struct inode *dir,
23296 ++ struct dentry *dentry, int id, void *ptr)
23297 ++{
23298 ++ dentry->d_op = &proc_nid_dentry_operations;
23299 ++ return vs_proc_instantiate(dir, dentry, id, ptr);
23300 ++}
23301 ++
23302 ++static struct dentry *proc_nid_lookup(struct inode *dir,
23303 ++ struct dentry *dentry, struct nameidata *nd)
23304 ++{
23305 ++ struct vs_entry *p = nx_base_stuff;
23306 ++ struct dentry *error = ERR_PTR(-ENOENT);
23307 ++
23308 ++ for (; p->name; p++) {
23309 ++ if (p->len != dentry->d_name.len)
23310 ++ continue;
23311 ++ if (!memcmp(dentry->d_name.name, p->name, p->len))
23312 ++ break;
23313 ++ }
23314 ++ if (!p->name)
23315 ++ goto out;
23316 ++
23317 ++ error = proc_nid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
23318 ++out:
23319 ++ return error;
23320 ++}
23321 ++
23322 ++static int proc_nid_readdir(struct file *filp,
23323 ++ void *dirent, filldir_t filldir)
23324 ++{
23325 ++ struct dentry *dentry = filp->f_dentry;
23326 ++ struct inode *inode = dentry->d_inode;
23327 ++ struct vs_entry *p = nx_base_stuff;
23328 ++ int size = sizeof(nx_base_stuff) / sizeof(struct vs_entry);
23329 ++ int pos, index;
23330 ++ u64 ino;
23331 ++
23332 ++ pos = filp->f_pos;
23333 ++ switch (pos) {
23334 ++ case 0:
23335 ++ ino = inode->i_ino;
23336 ++ if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23337 ++ goto out;
23338 ++ pos++;
23339 ++ /* fall through */
23340 ++ case 1:
23341 ++ ino = parent_ino(dentry);
23342 ++ if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23343 ++ goto out;
23344 ++ pos++;
23345 ++ /* fall through */
23346 ++ default:
23347 ++ index = pos - 2;
23348 ++ if (index >= size)
23349 ++ goto out;
23350 ++ for (p += index; p->name; p++) {
23351 ++ if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23352 ++ vs_proc_instantiate, PROC_I(inode)->fd, p))
23353 ++ goto out;
23354 ++ pos++;
23355 ++ }
23356 ++ }
23357 ++out:
23358 ++ filp->f_pos = pos;
23359 ++ return 1;
23360 ++}
23361 ++
23362 ++
23363 ++#define MAX_MULBY10 ((~0U - 9) / 10)
23364 ++
23365 ++static inline int atovid(const char *str, int len)
23366 ++{
23367 ++ int vid, c;
23368 ++
23369 ++ vid = 0;
23370 ++ while (len-- > 0) {
23371 ++ c = *str - '0';
23372 ++ str++;
23373 ++ if (c > 9)
23374 ++ return -1;
23375 ++ if (vid >= MAX_MULBY10)
23376 ++ return -1;
23377 ++ vid *= 10;
23378 ++ vid += c;
23379 ++ if (!vid)
23380 ++ return -1;
23381 ++ }
23382 ++ return vid;
23383 ++}
23384 ++
23385 ++/* now the upper level (virtual) */
23386 ++
23387 ++
23388 ++static struct file_operations proc_xid_file_operations = {
23389 ++ .read = generic_read_dir,
23390 ++ .readdir = proc_xid_readdir,
23391 ++};
23392 ++
23393 ++static struct inode_operations proc_xid_inode_operations = {
23394 ++ .lookup = proc_xid_lookup,
23395 ++};
23396 ++
23397 ++static struct vs_entry vx_virtual_stuff[] = {
23398 ++ INF("info", S_IRUGO, virtual_info),
23399 ++ INF("status", S_IRUGO, virtual_status),
23400 ++ DIR(NULL, S_IRUGO | S_IXUGO, xid),
23401 ++};
23402 ++
23403 ++
23404 ++static struct dentry *proc_virtual_lookup(struct inode *dir,
23405 ++ struct dentry *dentry, struct nameidata *nd)
23406 ++{
23407 ++ struct vs_entry *p = vx_virtual_stuff;
23408 ++ struct dentry *error = ERR_PTR(-ENOENT);
23409 ++ int id = 0;
23410 ++
23411 ++ for (; p->name; p++) {
23412 ++ if (p->len != dentry->d_name.len)
23413 ++ continue;
23414 ++ if (!memcmp(dentry->d_name.name, p->name, p->len))
23415 ++ break;
23416 ++ }
23417 ++ if (p->name)
23418 ++ goto instantiate;
23419 ++
23420 ++ id = atovid(dentry->d_name.name, dentry->d_name.len);
23421 ++ if ((id < 0) || !xid_is_hashed(id))
23422 ++ goto out;
23423 ++
23424 ++instantiate:
23425 ++ error = proc_xid_instantiate(dir, dentry, id, p);
23426 ++out:
23427 ++ return error;
23428 ++}
23429 ++
23430 ++static struct file_operations proc_nid_file_operations = {
23431 ++ .read = generic_read_dir,
23432 ++ .readdir = proc_nid_readdir,
23433 ++};
23434 ++
23435 ++static struct inode_operations proc_nid_inode_operations = {
23436 ++ .lookup = proc_nid_lookup,
23437 ++};
23438 ++
23439 ++static struct vs_entry nx_virtnet_stuff[] = {
23440 ++ INF("info", S_IRUGO, virtnet_info),
23441 ++ INF("status", S_IRUGO, virtnet_status),
23442 ++ DIR(NULL, S_IRUGO | S_IXUGO, nid),
23443 ++};
23444 ++
23445 ++
23446 ++static struct dentry *proc_virtnet_lookup(struct inode *dir,
23447 ++ struct dentry *dentry, struct nameidata *nd)
23448 ++{
23449 ++ struct vs_entry *p = nx_virtnet_stuff;
23450 ++ struct dentry *error = ERR_PTR(-ENOENT);
23451 ++ int id = 0;
23452 ++
23453 ++ for (; p->name; p++) {
23454 ++ if (p->len != dentry->d_name.len)
23455 ++ continue;
23456 ++ if (!memcmp(dentry->d_name.name, p->name, p->len))
23457 ++ break;
23458 ++ }
23459 ++ if (p->name)
23460 ++ goto instantiate;
23461 ++
23462 ++ id = atovid(dentry->d_name.name, dentry->d_name.len);
23463 ++ if ((id < 0) || !nid_is_hashed(id))
23464 ++ goto out;
23465 ++
23466 ++instantiate:
23467 ++ error = proc_nid_instantiate(dir, dentry, id, p);
23468 ++out:
23469 ++ return error;
23470 ++}
23471 ++
23472 ++
23473 ++#define PROC_MAXVIDS 32
23474 ++
23475 ++int proc_virtual_readdir(struct file *filp,
23476 ++ void *dirent, filldir_t filldir)
23477 ++{
23478 ++ struct dentry *dentry = filp->f_dentry;
23479 ++ struct inode *inode = dentry->d_inode;
23480 ++ struct vs_entry *p = vx_virtual_stuff;
23481 ++ int size = sizeof(vx_virtual_stuff) / sizeof(struct vs_entry);
23482 ++ int pos, index;
23483 ++ unsigned int xid_array[PROC_MAXVIDS];
23484 ++ char buf[PROC_NUMBUF];
23485 ++ unsigned int nr_xids, i;
23486 ++ u64 ino;
23487 ++
23488 ++ pos = filp->f_pos;
23489 ++ switch (pos) {
23490 ++ case 0:
23491 ++ ino = inode->i_ino;
23492 ++ if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23493 ++ goto out;
23494 ++ pos++;
23495 ++ /* fall through */
23496 ++ case 1:
23497 ++ ino = parent_ino(dentry);
23498 ++ if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23499 ++ goto out;
23500 ++ pos++;
23501 ++ /* fall through */
23502 ++ default:
23503 ++ index = pos - 2;
23504 ++ if (index >= size)
23505 ++ goto entries;
23506 ++ for (p += index; p->name; p++) {
23507 ++ if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23508 ++ vs_proc_instantiate, 0, p))
23509 ++ goto out;
23510 ++ pos++;
23511 ++ }
23512 ++ entries:
23513 ++ index = pos - size;
23514 ++ p = &vx_virtual_stuff[size - 1];
23515 ++ nr_xids = get_xid_list(index, xid_array, PROC_MAXVIDS);
23516 ++ for (i = 0; i < nr_xids; i++) {
23517 ++ int n, xid = xid_array[i];
23518 ++ unsigned int j = PROC_NUMBUF;
23519 ++
23520 ++ n = xid;
23521 ++ do
23522 ++ buf[--j] = '0' + (n % 10);
23523 ++ while (n /= 10);
23524 ++
23525 ++ if (proc_fill_cache(filp, dirent, filldir,
23526 ++ buf + j, PROC_NUMBUF - j,
23527 ++ vs_proc_instantiate, xid, p))
23528 ++ goto out;
23529 ++ pos++;
23530 ++ }
23531 ++ }
23532 ++out:
23533 ++ filp->f_pos = pos;
23534 ++ return 0;
23535 ++}
23536 ++
23537 ++static int proc_virtual_getattr(struct vfsmount *mnt,
23538 ++ struct dentry *dentry, struct kstat *stat)
23539 ++{
23540 ++ struct inode *inode = dentry->d_inode;
23541 ++
23542 ++ generic_fillattr(inode, stat);
23543 ++ stat->nlink = 2 + atomic_read(&vx_global_cactive);
23544 ++ return 0;
23545 ++}
23546 ++
23547 ++static struct file_operations proc_virtual_dir_operations = {
23548 ++ .read = generic_read_dir,
23549 ++ .readdir = proc_virtual_readdir,
23550 ++};
23551 ++
23552 ++static struct inode_operations proc_virtual_dir_inode_operations = {
23553 ++ .getattr = proc_virtual_getattr,
23554 ++ .lookup = proc_virtual_lookup,
23555 ++};
23556 ++
23557 ++
23558 ++
23559 ++
23560 ++
23561 ++int proc_virtnet_readdir(struct file *filp,
23562 ++ void *dirent, filldir_t filldir)
23563 ++{
23564 ++ struct dentry *dentry = filp->f_dentry;
23565 ++ struct inode *inode = dentry->d_inode;
23566 ++ struct vs_entry *p = nx_virtnet_stuff;
23567 ++ int size = sizeof(nx_virtnet_stuff) / sizeof(struct vs_entry);
23568 ++ int pos, index;
23569 ++ unsigned int nid_array[PROC_MAXVIDS];
23570 ++ char buf[PROC_NUMBUF];
23571 ++ unsigned int nr_nids, i;
23572 ++ u64 ino;
23573 ++
23574 ++ pos = filp->f_pos;
23575 ++ switch (pos) {
23576 ++ case 0:
23577 ++ ino = inode->i_ino;
23578 ++ if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
23579 ++ goto out;
23580 ++ pos++;
23581 ++ /* fall through */
23582 ++ case 1:
23583 ++ ino = parent_ino(dentry);
23584 ++ if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
23585 ++ goto out;
23586 ++ pos++;
23587 ++ /* fall through */
23588 ++ default:
23589 ++ index = pos - 2;
23590 ++ if (index >= size)
23591 ++ goto entries;
23592 ++ for (p += index; p->name; p++) {
23593 ++ if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
23594 ++ vs_proc_instantiate, 0, p))
23595 ++ goto out;
23596 ++ pos++;
23597 ++ }
23598 ++ entries:
23599 ++ index = pos - size;
23600 ++ p = &nx_virtnet_stuff[size - 1];
23601 ++ nr_nids = get_nid_list(index, nid_array, PROC_MAXVIDS);
23602 ++ for (i = 0; i < nr_nids; i++) {
23603 ++ int n, nid = nid_array[i];
23604 ++ unsigned int j = PROC_NUMBUF;
23605 ++
23606 ++ n = nid;
23607 ++ do
23608 ++ buf[--j] = '0' + (n % 10);
23609 ++ while (n /= 10);
23610 ++
23611 ++ if (proc_fill_cache(filp, dirent, filldir,
23612 ++ buf + j, PROC_NUMBUF - j,
23613 ++ vs_proc_instantiate, nid, p))
23614 ++ goto out;
23615 ++ pos++;
23616 ++ }
23617 ++ }
23618 ++out:
23619 ++ filp->f_pos = pos;
23620 ++ return 0;
23621 ++}
23622 ++
23623 ++static int proc_virtnet_getattr(struct vfsmount *mnt,
23624 ++ struct dentry *dentry, struct kstat *stat)
23625 ++{
23626 ++ struct inode *inode = dentry->d_inode;
23627 ++
23628 ++ generic_fillattr(inode, stat);
23629 ++ stat->nlink = 2 + atomic_read(&nx_global_cactive);
23630 ++ return 0;
23631 ++}
23632 ++
23633 ++static struct file_operations proc_virtnet_dir_operations = {
23634 ++ .read = generic_read_dir,
23635 ++ .readdir = proc_virtnet_readdir,
23636 ++};
23637 ++
23638 ++static struct inode_operations proc_virtnet_dir_inode_operations = {
23639 ++ .getattr = proc_virtnet_getattr,
23640 ++ .lookup = proc_virtnet_lookup,
23641 ++};
23642 ++
23643 ++
23644 ++
23645 ++void proc_vx_init(void)
23646 ++{
23647 ++ struct proc_dir_entry *ent;
23648 ++
23649 ++ ent = proc_mkdir("virtual", 0);
23650 ++ if (ent) {
23651 ++ ent->proc_fops = &proc_virtual_dir_operations;
23652 ++ ent->proc_iops = &proc_virtual_dir_inode_operations;
23653 ++ }
23654 ++ proc_virtual = ent;
23655 ++
23656 ++ ent = proc_mkdir("virtnet", 0);
23657 ++ if (ent) {
23658 ++ ent->proc_fops = &proc_virtnet_dir_operations;
23659 ++ ent->proc_iops = &proc_virtnet_dir_inode_operations;
23660 ++ }
23661 ++ proc_virtnet = ent;
23662 ++}
23663 ++
23664 ++
23665 ++
23666 ++
23667 ++/* per pid info */
23668 ++
23669 ++
23670 ++int proc_pid_vx_info(struct task_struct *p, char *buffer)
23671 ++{
23672 ++ struct vx_info *vxi;
23673 ++ char *orig = buffer;
23674 ++
23675 ++ buffer += sprintf(buffer, "XID:\t%d\n", vx_task_xid(p));
23676 ++
23677 ++ vxi = task_get_vx_info(p);
23678 ++ if (!vxi)
23679 ++ goto out;
23680 ++
23681 ++ buffer += sprintf(buffer, "BCaps:\t");
23682 ++ buffer = print_cap_t(buffer, &vxi->vx_bcaps);
23683 ++ buffer += sprintf(buffer, "\n");
23684 ++ buffer += sprintf(buffer, "CCaps:\t%016llx\n",
23685 ++ (unsigned long long)vxi->vx_ccaps);
23686 ++ buffer += sprintf(buffer, "CFlags:\t%016llx\n",
23687 ++ (unsigned long long)vxi->vx_flags);
23688 ++ buffer += sprintf(buffer, "CIPid:\t%d\n", vxi->vx_initpid);
23689 ++
23690 ++ put_vx_info(vxi);
23691 ++out:
23692 ++ return buffer - orig;
23693 ++}
23694 ++
23695 ++
23696 ++int proc_pid_nx_info(struct task_struct *p, char *buffer)
23697 ++{
23698 ++ struct nx_info *nxi;
23699 ++ struct nx_addr_v4 *v4a;
23700 ++#ifdef CONFIG_IPV6
23701 ++ struct nx_addr_v6 *v6a;
23702 ++#endif
23703 ++ char *orig = buffer;
23704 ++ int i;
23705 ++
23706 ++ buffer += sprintf(buffer, "NID:\t%d\n", nx_task_nid(p));
23707 ++
23708 ++ nxi = task_get_nx_info(p);
23709 ++ if (!nxi)
23710 ++ goto out;
23711 ++
23712 ++ buffer += sprintf(buffer, "NCaps:\t%016llx\n",
23713 ++ (unsigned long long)nxi->nx_ncaps);
23714 ++ buffer += sprintf(buffer, "NFlags:\t%016llx\n",
23715 ++ (unsigned long long)nxi->nx_flags);
23716 ++
23717 ++ buffer += sprintf(buffer,
23718 ++ "V4Root[bcast]:\t" NIPQUAD_FMT "\n",
23719 ++ NIPQUAD(nxi->v4_bcast.s_addr));
23720 ++ buffer += sprintf (buffer,
23721 ++ "V4Root[lback]:\t" NIPQUAD_FMT "\n",
23722 ++ NIPQUAD(nxi->v4_lback.s_addr));
23723 ++ if (!NX_IPV4(nxi))
23724 ++ goto skip_v4;
23725 ++ for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
23726 ++ buffer += sprintf(buffer, "V4Root[%d]:\t" NXAV4_FMT "\n",
23727 ++ i, NXAV4(v4a));
23728 ++skip_v4:
23729 ++#ifdef CONFIG_IPV6
23730 ++ if (!NX_IPV6(nxi))
23731 ++ goto skip_v6;
23732 ++ for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
23733 ++ buffer += sprintf(buffer, "V6Root[%d]:\t" NXAV6_FMT "\n",
23734 ++ i, NXAV6(v6a));
23735 ++skip_v6:
23736 ++#endif
23737 ++ put_nx_info(nxi);
23738 ++out:
23739 ++ return buffer - orig;
23740 ++}
23741 ++
23742 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/sched.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched.c
23743 +--- linux-2.6.29.4/kernel/vserver/sched.c 1970-01-01 01:00:00.000000000 +0100
23744 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched.c 2009-02-22 22:54:26.000000000 +0100
23745 +@@ -0,0 +1,414 @@
23746 ++/*
23747 ++ * linux/kernel/vserver/sched.c
23748 ++ *
23749 ++ * Virtual Server: Scheduler Support
23750 ++ *
23751 ++ * Copyright (C) 2004-2007 Herbert Pötzl
23752 ++ *
23753 ++ * V0.01 adapted Sam Vilains version to 2.6.3
23754 ++ * V0.02 removed legacy interface
23755 ++ * V0.03 changed vcmds to vxi arg
23756 ++ * V0.04 removed older and legacy interfaces
23757 ++ *
23758 ++ */
23759 ++
23760 ++#include <linux/vs_context.h>
23761 ++#include <linux/vs_sched.h>
23762 ++#include <linux/vserver/sched_cmd.h>
23763 ++
23764 ++#include <asm/uaccess.h>
23765 ++
23766 ++
23767 ++#define vxd_check_range(val, min, max) do { \
23768 ++ vxlprintk((val < min) || (val > max), \
23769 ++ "check_range(%ld,%ld,%ld)", \
23770 ++ (long)val, (long)min, (long)max, \
23771 ++ __FILE__, __LINE__); \
23772 ++ } while (0)
23773 ++
23774 ++
23775 ++void vx_update_sched_param(struct _vx_sched *sched,
23776 ++ struct _vx_sched_pc *sched_pc)
23777 ++{
23778 ++ unsigned int set_mask = sched->update_mask;
23779 ++
23780 ++ if (set_mask & VXSM_FILL_RATE)
23781 ++ sched_pc->fill_rate[0] = sched->fill_rate[0];
23782 ++ if (set_mask & VXSM_INTERVAL)
23783 ++ sched_pc->interval[0] = sched->interval[0];
23784 ++ if (set_mask & VXSM_FILL_RATE2)
23785 ++ sched_pc->fill_rate[1] = sched->fill_rate[1];
23786 ++ if (set_mask & VXSM_INTERVAL2)
23787 ++ sched_pc->interval[1] = sched->interval[1];
23788 ++ if (set_mask & VXSM_TOKENS)
23789 ++ sched_pc->tokens = sched->tokens;
23790 ++ if (set_mask & VXSM_TOKENS_MIN)
23791 ++ sched_pc->tokens_min = sched->tokens_min;
23792 ++ if (set_mask & VXSM_TOKENS_MAX)
23793 ++ sched_pc->tokens_max = sched->tokens_max;
23794 ++ if (set_mask & VXSM_PRIO_BIAS)
23795 ++ sched_pc->prio_bias = sched->prio_bias;
23796 ++
23797 ++ if (set_mask & VXSM_IDLE_TIME)
23798 ++ sched_pc->flags |= VXSF_IDLE_TIME;
23799 ++ else
23800 ++ sched_pc->flags &= ~VXSF_IDLE_TIME;
23801 ++
23802 ++ /* reset time */
23803 ++ sched_pc->norm_time = jiffies;
23804 ++}
23805 ++
23806 ++
23807 ++/*
23808 ++ * recalculate the context's scheduling tokens
23809 ++ *
23810 ++ * ret > 0 : number of tokens available
23811 ++ * ret < 0 : on hold, check delta_min[]
23812 ++ * -1 only jiffies
23813 ++ * -2 also idle time
23814 ++ *
23815 ++ */
23816 ++int vx_tokens_recalc(struct _vx_sched_pc *sched_pc,
23817 ++ unsigned long *norm_time, unsigned long *idle_time, int delta_min[2])
23818 ++{
23819 ++ long delta;
23820 ++ long tokens = 0;
23821 ++ int flags = sched_pc->flags;
23822 ++
23823 ++ /* how much time did pass? */
23824 ++ delta = *norm_time - sched_pc->norm_time;
23825 ++ // printk("@ %ld, %ld, %ld\n", *norm_time, sched_pc->norm_time, jiffies);
23826 ++ vxd_check_range(delta, 0, INT_MAX);
23827 ++
23828 ++ if (delta >= sched_pc->interval[0]) {
23829 ++ long tokens, integral;
23830 ++
23831 ++ /* calc integral token part */
23832 ++ tokens = delta / sched_pc->interval[0];
23833 ++ integral = tokens * sched_pc->interval[0];
23834 ++ tokens *= sched_pc->fill_rate[0];
23835 ++#ifdef CONFIG_VSERVER_HARDCPU
23836 ++ delta_min[0] = delta - integral;
23837 ++ vxd_check_range(delta_min[0], 0, sched_pc->interval[0]);
23838 ++#endif
23839 ++ /* advance time */
23840 ++ sched_pc->norm_time += delta;
23841 ++
23842 ++ /* add tokens */
23843 ++ sched_pc->tokens += tokens;
23844 ++ sched_pc->token_time += tokens;
23845 ++ } else
23846 ++ delta_min[0] = delta;
23847 ++
23848 ++#ifdef CONFIG_VSERVER_IDLETIME
23849 ++ if (!(flags & VXSF_IDLE_TIME))
23850 ++ goto skip_idle;
23851 ++
23852 ++ /* how much was the idle skip? */
23853 ++ delta = *idle_time - sched_pc->idle_time;
23854 ++ vxd_check_range(delta, 0, INT_MAX);
23855 ++
23856 ++ if (delta >= sched_pc->interval[1]) {
23857 ++ long tokens, integral;
23858 ++
23859 ++ /* calc fair share token part */
23860 ++ tokens = delta / sched_pc->interval[1];
23861 ++ integral = tokens * sched_pc->interval[1];
23862 ++ tokens *= sched_pc->fill_rate[1];
23863 ++ delta_min[1] = delta - integral;
23864 ++ vxd_check_range(delta_min[1], 0, sched_pc->interval[1]);
23865 ++
23866 ++ /* advance idle time */
23867 ++ sched_pc->idle_time += integral;
23868 ++
23869 ++ /* add tokens */
23870 ++ sched_pc->tokens += tokens;
23871 ++ sched_pc->token_time += tokens;
23872 ++ } else
23873 ++ delta_min[1] = delta;
23874 ++skip_idle:
23875 ++#endif
23876 ++
23877 ++ /* clip at maximum */
23878 ++ if (sched_pc->tokens > sched_pc->tokens_max)
23879 ++ sched_pc->tokens = sched_pc->tokens_max;
23880 ++ tokens = sched_pc->tokens;
23881 ++
23882 ++ if ((flags & VXSF_ONHOLD)) {
23883 ++ /* can we unhold? */
23884 ++ if (tokens >= sched_pc->tokens_min) {
23885 ++ flags &= ~VXSF_ONHOLD;
23886 ++ sched_pc->hold_ticks +=
23887 ++ *norm_time - sched_pc->onhold;
23888 ++ } else
23889 ++ goto on_hold;
23890 ++ } else {
23891 ++ /* put on hold? */
23892 ++ if (tokens <= 0) {
23893 ++ flags |= VXSF_ONHOLD;
23894 ++ sched_pc->onhold = *norm_time;
23895 ++ goto on_hold;
23896 ++ }
23897 ++ }
23898 ++ sched_pc->flags = flags;
23899 ++ return tokens;
23900 ++
23901 ++on_hold:
23902 ++ tokens = sched_pc->tokens_min - tokens;
23903 ++ sched_pc->flags = flags;
23904 ++ // BUG_ON(tokens < 0); probably doesn't hold anymore
23905 ++
23906 ++#ifdef CONFIG_VSERVER_HARDCPU
23907 ++ /* next interval? */
23908 ++ if (!sched_pc->fill_rate[0])
23909 ++ delta_min[0] = HZ;
23910 ++ else if (tokens > sched_pc->fill_rate[0])
23911 ++ delta_min[0] += sched_pc->interval[0] *
23912 ++ tokens / sched_pc->fill_rate[0];
23913 ++ else
23914 ++ delta_min[0] = sched_pc->interval[0] - delta_min[0];
23915 ++ vxd_check_range(delta_min[0], 0, INT_MAX);
23916 ++
23917 ++#ifdef CONFIG_VSERVER_IDLETIME
23918 ++ if (!(flags & VXSF_IDLE_TIME))
23919 ++ return -1;
23920 ++
23921 ++ /* next interval? */
23922 ++ if (!sched_pc->fill_rate[1])
23923 ++ delta_min[1] = HZ;
23924 ++ else if (tokens > sched_pc->fill_rate[1])
23925 ++ delta_min[1] += sched_pc->interval[1] *
23926 ++ tokens / sched_pc->fill_rate[1];
23927 ++ else
23928 ++ delta_min[1] = sched_pc->interval[1] - delta_min[1];
23929 ++ vxd_check_range(delta_min[1], 0, INT_MAX);
23930 ++
23931 ++ return -2;
23932 ++#else
23933 ++ return -1;
23934 ++#endif /* CONFIG_VSERVER_IDLETIME */
23935 ++#else
23936 ++ return 0;
23937 ++#endif /* CONFIG_VSERVER_HARDCPU */
23938 ++}
23939 ++
23940 ++static inline unsigned long msec_to_ticks(unsigned long msec)
23941 ++{
23942 ++ return msecs_to_jiffies(msec);
23943 ++}
23944 ++
23945 ++static inline unsigned long ticks_to_msec(unsigned long ticks)
23946 ++{
23947 ++ return jiffies_to_msecs(ticks);
23948 ++}
23949 ++
23950 ++static inline unsigned long ticks_to_usec(unsigned long ticks)
23951 ++{
23952 ++ return jiffies_to_usecs(ticks);
23953 ++}
23954 ++
23955 ++
23956 ++static int do_set_sched(struct vx_info *vxi, struct vcmd_sched_v5 *data)
23957 ++{
23958 ++ unsigned int set_mask = data->mask;
23959 ++ unsigned int update_mask;
23960 ++ int i, cpu;
23961 ++
23962 ++ /* Sanity check data values */
23963 ++ if (data->tokens_max <= 0)
23964 ++ data->tokens_max = HZ;
23965 ++ if (data->tokens_min < 0)
23966 ++ data->tokens_min = HZ / 3;
23967 ++ if (data->tokens_min >= data->tokens_max)
23968 ++ data->tokens_min = data->tokens_max;
23969 ++
23970 ++ if (data->prio_bias > MAX_PRIO_BIAS)
23971 ++ data->prio_bias = MAX_PRIO_BIAS;
23972 ++ if (data->prio_bias < MIN_PRIO_BIAS)
23973 ++ data->prio_bias = MIN_PRIO_BIAS;
23974 ++
23975 ++ spin_lock(&vxi->sched.tokens_lock);
23976 ++
23977 ++ /* sync up on delayed updates */
23978 ++ for_each_cpu_mask(cpu, vxi->sched.update)
23979 ++ vx_update_sched_param(&vxi->sched,
23980 ++ &vx_per_cpu(vxi, sched_pc, cpu));
23981 ++
23982 ++ if (set_mask & VXSM_FILL_RATE)
23983 ++ vxi->sched.fill_rate[0] = data->fill_rate[0];
23984 ++ if (set_mask & VXSM_FILL_RATE2)
23985 ++ vxi->sched.fill_rate[1] = data->fill_rate[1];
23986 ++ if (set_mask & VXSM_INTERVAL)
23987 ++ vxi->sched.interval[0] = (set_mask & VXSM_MSEC) ?
23988 ++ msec_to_ticks(data->interval[0]) : data->interval[0];
23989 ++ if (set_mask & VXSM_INTERVAL2)
23990 ++ vxi->sched.interval[1] = (set_mask & VXSM_MSEC) ?
23991 ++ msec_to_ticks(data->interval[1]) : data->interval[1];
23992 ++ if (set_mask & VXSM_TOKENS)
23993 ++ vxi->sched.tokens = data->tokens;
23994 ++ if (set_mask & VXSM_TOKENS_MIN)
23995 ++ vxi->sched.tokens_min = data->tokens_min;
23996 ++ if (set_mask & VXSM_TOKENS_MAX)
23997 ++ vxi->sched.tokens_max = data->tokens_max;
23998 ++ if (set_mask & VXSM_PRIO_BIAS)
23999 ++ vxi->sched.prio_bias = data->prio_bias;
24000 ++
24001 ++ /* Sanity check rate/interval */
24002 ++ for (i = 0; i < 2; i++) {
24003 ++ if (data->fill_rate[i] < 0)
24004 ++ data->fill_rate[i] = 0;
24005 ++ if (data->interval[i] <= 0)
24006 ++ data->interval[i] = HZ;
24007 ++ }
24008 ++
24009 ++ update_mask = vxi->sched.update_mask & VXSM_SET_MASK;
24010 ++ update_mask |= (set_mask & (VXSM_SET_MASK | VXSM_IDLE_TIME));
24011 ++ vxi->sched.update_mask = update_mask;
24012 ++
24013 ++#ifdef CONFIG_SMP
24014 ++ rmb();
24015 ++ if (set_mask & VXSM_CPU_ID) {
24016 ++ vxi->sched.update = cpumask_of_cpu(data->cpu_id);
24017 ++ cpus_and(vxi->sched.update, cpu_online_map,
24018 ++ vxi->sched.update);
24019 ++ } else
24020 ++ vxi->sched.update = cpu_online_map;
24021 ++
24022 ++ /* forced reload? */
24023 ++ if (set_mask & VXSM_FORCE) {
24024 ++ for_each_cpu_mask(cpu, vxi->sched.update)
24025 ++ vx_update_sched_param(&vxi->sched,
24026 ++ &vx_per_cpu(vxi, sched_pc, cpu));
24027 ++ vxi->sched.update = CPU_MASK_NONE;
24028 ++ }
24029 ++#else
24030 ++ /* on UP we update immediately */
24031 ++ vx_update_sched_param(&vxi->sched,
24032 ++ &vx_per_cpu(vxi, sched_pc, 0));
24033 ++#endif
24034 ++
24035 ++ spin_unlock(&vxi->sched.tokens_lock);
24036 ++ return 0;
24037 ++}
24038 ++
24039 ++
24040 ++#define COPY_IDS(C) C(cpu_id); C(bucket_id)
24041 ++#define COPY_PRI(C) C(prio_bias)
24042 ++#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
24043 ++#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
24044 ++ C(fill_rate[1]); C(interval[1]);
24045 ++
24046 ++#define COPY_VALUE(name) vc_data.name = data->name
24047 ++
24048 ++static int do_set_sched_v4(struct vx_info *vxi, struct vcmd_set_sched_v4 *data)
24049 ++{
24050 ++ struct vcmd_sched_v5 vc_data;
24051 ++
24052 ++ vc_data.mask = data->set_mask;
24053 ++ COPY_IDS(COPY_VALUE);
24054 ++ COPY_PRI(COPY_VALUE);
24055 ++ COPY_TOK(COPY_VALUE);
24056 ++ vc_data.fill_rate[0] = vc_data.fill_rate[1] = data->fill_rate;
24057 ++ vc_data.interval[0] = vc_data.interval[1] = data->interval;
24058 ++ return do_set_sched(vxi, &vc_data);
24059 ++}
24060 ++
24061 ++int vc_set_sched_v4(struct vx_info *vxi, void __user *data)
24062 ++{
24063 ++ struct vcmd_set_sched_v4 vc_data;
24064 ++
24065 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24066 ++ return -EFAULT;
24067 ++
24068 ++ return do_set_sched_v4(vxi, &vc_data);
24069 ++}
24070 ++
24071 ++ /* latest interface is v5 */
24072 ++
24073 ++int vc_set_sched(struct vx_info *vxi, void __user *data)
24074 ++{
24075 ++ struct vcmd_sched_v5 vc_data;
24076 ++
24077 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24078 ++ return -EFAULT;
24079 ++
24080 ++ return do_set_sched(vxi, &vc_data);
24081 ++}
24082 ++
24083 ++
24084 ++#define COPY_PRI(C) C(prio_bias)
24085 ++#define COPY_TOK(C) C(tokens); C(tokens_min); C(tokens_max)
24086 ++#define COPY_FRI(C) C(fill_rate[0]); C(interval[0]); \
24087 ++ C(fill_rate[1]); C(interval[1]);
24088 ++
24089 ++#define COPY_VALUE(name) vc_data.name = data->name
24090 ++
24091 ++
24092 ++int vc_get_sched(struct vx_info *vxi, void __user *data)
24093 ++{
24094 ++ struct vcmd_sched_v5 vc_data;
24095 ++
24096 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24097 ++ return -EFAULT;
24098 ++
24099 ++ if (vc_data.mask & VXSM_CPU_ID) {
24100 ++ int cpu = vc_data.cpu_id;
24101 ++ struct _vx_sched_pc *data;
24102 ++
24103 ++ if (!cpu_possible(cpu))
24104 ++ return -EINVAL;
24105 ++
24106 ++ data = &vx_per_cpu(vxi, sched_pc, cpu);
24107 ++ COPY_TOK(COPY_VALUE);
24108 ++ COPY_PRI(COPY_VALUE);
24109 ++ COPY_FRI(COPY_VALUE);
24110 ++
24111 ++ if (data->flags & VXSF_IDLE_TIME)
24112 ++ vc_data.mask |= VXSM_IDLE_TIME;
24113 ++ } else {
24114 ++ struct _vx_sched *data = &vxi->sched;
24115 ++
24116 ++ COPY_TOK(COPY_VALUE);
24117 ++ COPY_PRI(COPY_VALUE);
24118 ++ COPY_FRI(COPY_VALUE);
24119 ++ }
24120 ++
24121 ++ if (vc_data.mask & VXSM_MSEC) {
24122 ++ vc_data.interval[0] = ticks_to_msec(vc_data.interval[0]);
24123 ++ vc_data.interval[1] = ticks_to_msec(vc_data.interval[1]);
24124 ++ }
24125 ++
24126 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24127 ++ return -EFAULT;
24128 ++ return 0;
24129 ++}
24130 ++
24131 ++
24132 ++int vc_sched_info(struct vx_info *vxi, void __user *data)
24133 ++{
24134 ++ struct vcmd_sched_info vc_data;
24135 ++ int cpu;
24136 ++
24137 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24138 ++ return -EFAULT;
24139 ++
24140 ++ cpu = vc_data.cpu_id;
24141 ++ if (!cpu_possible(cpu))
24142 ++ return -EINVAL;
24143 ++
24144 ++ if (vxi) {
24145 ++ struct _vx_sched_pc *sched_pc =
24146 ++ &vx_per_cpu(vxi, sched_pc, cpu);
24147 ++
24148 ++ vc_data.user_msec = ticks_to_msec(sched_pc->user_ticks);
24149 ++ vc_data.sys_msec = ticks_to_msec(sched_pc->sys_ticks);
24150 ++ vc_data.hold_msec = ticks_to_msec(sched_pc->hold_ticks);
24151 ++ vc_data.vavavoom = sched_pc->vavavoom;
24152 ++ }
24153 ++ vc_data.token_usec = ticks_to_usec(1);
24154 ++
24155 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24156 ++ return -EFAULT;
24157 ++ return 0;
24158 ++}
24159 ++
24160 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/sched_init.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched_init.h
24161 +--- linux-2.6.29.4/kernel/vserver/sched_init.h 1970-01-01 01:00:00.000000000 +0100
24162 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched_init.h 2009-02-22 22:54:26.000000000 +0100
24163 +@@ -0,0 +1,50 @@
24164 ++
24165 ++static inline void vx_info_init_sched(struct _vx_sched *sched)
24166 ++{
24167 ++ static struct lock_class_key tokens_lock_key;
24168 ++
24169 ++ /* scheduling; hard code starting values as constants */
24170 ++ sched->fill_rate[0] = 1;
24171 ++ sched->interval[0] = 4;
24172 ++ sched->fill_rate[1] = 1;
24173 ++ sched->interval[1] = 8;
24174 ++ sched->tokens = HZ >> 2;
24175 ++ sched->tokens_min = HZ >> 4;
24176 ++ sched->tokens_max = HZ >> 1;
24177 ++ sched->tokens_lock = SPIN_LOCK_UNLOCKED;
24178 ++ sched->prio_bias = 0;
24179 ++
24180 ++ lockdep_set_class(&sched->tokens_lock, &tokens_lock_key);
24181 ++}
24182 ++
24183 ++static inline
24184 ++void vx_info_init_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24185 ++{
24186 ++ sched_pc->fill_rate[0] = 1;
24187 ++ sched_pc->interval[0] = 4;
24188 ++ sched_pc->fill_rate[1] = 1;
24189 ++ sched_pc->interval[1] = 8;
24190 ++ sched_pc->tokens = HZ >> 2;
24191 ++ sched_pc->tokens_min = HZ >> 4;
24192 ++ sched_pc->tokens_max = HZ >> 1;
24193 ++ sched_pc->prio_bias = 0;
24194 ++ sched_pc->vavavoom = 0;
24195 ++ sched_pc->token_time = 0;
24196 ++ sched_pc->idle_time = 0;
24197 ++ sched_pc->norm_time = jiffies;
24198 ++
24199 ++ sched_pc->user_ticks = 0;
24200 ++ sched_pc->sys_ticks = 0;
24201 ++ sched_pc->hold_ticks = 0;
24202 ++}
24203 ++
24204 ++static inline void vx_info_exit_sched(struct _vx_sched *sched)
24205 ++{
24206 ++ return;
24207 ++}
24208 ++
24209 ++static inline
24210 ++void vx_info_exit_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
24211 ++{
24212 ++ return;
24213 ++}
24214 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/sched_proc.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched_proc.h
24215 +--- linux-2.6.29.4/kernel/vserver/sched_proc.h 1970-01-01 01:00:00.000000000 +0100
24216 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sched_proc.h 2009-02-22 22:54:26.000000000 +0100
24217 +@@ -0,0 +1,57 @@
24218 ++#ifndef _VX_SCHED_PROC_H
24219 ++#define _VX_SCHED_PROC_H
24220 ++
24221 ++
24222 ++static inline
24223 ++int vx_info_proc_sched(struct _vx_sched *sched, char *buffer)
24224 ++{
24225 ++ int length = 0;
24226 ++
24227 ++ length += sprintf(buffer,
24228 ++ "FillRate:\t%8d,%d\n"
24229 ++ "Interval:\t%8d,%d\n"
24230 ++ "TokensMin:\t%8d\n"
24231 ++ "TokensMax:\t%8d\n"
24232 ++ "PrioBias:\t%8d\n",
24233 ++ sched->fill_rate[0],
24234 ++ sched->fill_rate[1],
24235 ++ sched->interval[0],
24236 ++ sched->interval[1],
24237 ++ sched->tokens_min,
24238 ++ sched->tokens_max,
24239 ++ sched->prio_bias);
24240 ++ return length;
24241 ++}
24242 ++
24243 ++static inline
24244 ++int vx_info_proc_sched_pc(struct _vx_sched_pc *sched_pc,
24245 ++ char *buffer, int cpu)
24246 ++{
24247 ++ int length = 0;
24248 ++
24249 ++ length += sprintf(buffer + length,
24250 ++ "cpu %d: %lld %lld %lld %ld %ld", cpu,
24251 ++ (unsigned long long)sched_pc->user_ticks,
24252 ++ (unsigned long long)sched_pc->sys_ticks,
24253 ++ (unsigned long long)sched_pc->hold_ticks,
24254 ++ sched_pc->token_time,
24255 ++ sched_pc->idle_time);
24256 ++ length += sprintf(buffer + length,
24257 ++ " %c%c %d %d %d %d/%d %d/%d",
24258 ++ (sched_pc->flags & VXSF_ONHOLD) ? 'H' : 'R',
24259 ++ (sched_pc->flags & VXSF_IDLE_TIME) ? 'I' : '-',
24260 ++ sched_pc->tokens,
24261 ++ sched_pc->tokens_min,
24262 ++ sched_pc->tokens_max,
24263 ++ sched_pc->fill_rate[0],
24264 ++ sched_pc->interval[0],
24265 ++ sched_pc->fill_rate[1],
24266 ++ sched_pc->interval[1]);
24267 ++ length += sprintf(buffer + length,
24268 ++ " %d %d\n",
24269 ++ sched_pc->prio_bias,
24270 ++ sched_pc->vavavoom);
24271 ++ return length;
24272 ++}
24273 ++
24274 ++#endif /* _VX_SCHED_PROC_H */
24275 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/signal.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/signal.c
24276 +--- linux-2.6.29.4/kernel/vserver/signal.c 1970-01-01 01:00:00.000000000 +0100
24277 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/signal.c 2009-02-22 22:54:26.000000000 +0100
24278 +@@ -0,0 +1,132 @@
24279 ++/*
24280 ++ * linux/kernel/vserver/signal.c
24281 ++ *
24282 ++ * Virtual Server: Signal Support
24283 ++ *
24284 ++ * Copyright (C) 2003-2007 Herbert Pötzl
24285 ++ *
24286 ++ * V0.01 broken out from vcontext V0.05
24287 ++ * V0.02 changed vcmds to vxi arg
24288 ++ * V0.03 adjusted siginfo for kill
24289 ++ *
24290 ++ */
24291 ++
24292 ++#include <asm/uaccess.h>
24293 ++
24294 ++#include <linux/vs_context.h>
24295 ++#include <linux/vs_pid.h>
24296 ++#include <linux/vserver/signal_cmd.h>
24297 ++
24298 ++
24299 ++int vx_info_kill(struct vx_info *vxi, int pid, int sig)
24300 ++{
24301 ++ int retval, count = 0;
24302 ++ struct task_struct *p;
24303 ++ struct siginfo *sip = SEND_SIG_PRIV;
24304 ++
24305 ++ retval = -ESRCH;
24306 ++ vxdprintk(VXD_CBIT(misc, 4),
24307 ++ "vx_info_kill(%p[#%d],%d,%d)*",
24308 ++ vxi, vxi->vx_id, pid, sig);
24309 ++ read_lock(&tasklist_lock);
24310 ++ switch (pid) {
24311 ++ case 0:
24312 ++ case -1:
24313 ++ for_each_process(p) {
24314 ++ int err = 0;
24315 ++
24316 ++ if (vx_task_xid(p) != vxi->vx_id || p->pid <= 1 ||
24317 ++ (pid && vxi->vx_initpid == p->pid))
24318 ++ continue;
24319 ++
24320 ++ err = group_send_sig_info(sig, sip, p);
24321 ++ ++count;
24322 ++ if (err != -EPERM)
24323 ++ retval = err;
24324 ++ }
24325 ++ break;
24326 ++
24327 ++ case 1:
24328 ++ if (vxi->vx_initpid) {
24329 ++ pid = vxi->vx_initpid;
24330 ++ /* for now, only SIGINT to private init ... */
24331 ++ if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24332 ++ /* ... as long as there are tasks left */
24333 ++ (atomic_read(&vxi->vx_tasks) > 1))
24334 ++ sig = SIGINT;
24335 ++ }
24336 ++ /* fallthrough */
24337 ++ default:
24338 ++ p = find_task_by_real_pid(pid);
24339 ++ if (p) {
24340 ++ if (vx_task_xid(p) == vxi->vx_id)
24341 ++ retval = group_send_sig_info(sig, sip, p);
24342 ++ }
24343 ++ break;
24344 ++ }
24345 ++ read_unlock(&tasklist_lock);
24346 ++ vxdprintk(VXD_CBIT(misc, 4),
24347 ++ "vx_info_kill(%p[#%d],%d,%d,%ld) = %d",
24348 ++ vxi, vxi->vx_id, pid, sig, (long)sip, retval);
24349 ++ return retval;
24350 ++}
24351 ++
24352 ++int vc_ctx_kill(struct vx_info *vxi, void __user *data)
24353 ++{
24354 ++ struct vcmd_ctx_kill_v0 vc_data;
24355 ++
24356 ++ if (copy_from_user(&vc_data, data, sizeof(vc_data)))
24357 ++ return -EFAULT;
24358 ++
24359 ++ /* special check to allow guest shutdown */
24360 ++ if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
24361 ++ /* forbid killall pid=0 when init is present */
24362 ++ (((vc_data.pid < 1) && vxi->vx_initpid) ||
24363 ++ (vc_data.pid > 1)))
24364 ++ return -EACCES;
24365 ++
24366 ++ return vx_info_kill(vxi, vc_data.pid, vc_data.sig);
24367 ++}
24368 ++
24369 ++
24370 ++static int __wait_exit(struct vx_info *vxi)
24371 ++{
24372 ++ DECLARE_WAITQUEUE(wait, current);
24373 ++ int ret = 0;
24374 ++
24375 ++ add_wait_queue(&vxi->vx_wait, &wait);
24376 ++ set_current_state(TASK_INTERRUPTIBLE);
24377 ++
24378 ++wait:
24379 ++ if (vx_info_state(vxi,
24380 ++ VXS_SHUTDOWN | VXS_HASHED | VXS_HELPER) == VXS_SHUTDOWN)
24381 ++ goto out;
24382 ++ if (signal_pending(current)) {
24383 ++ ret = -ERESTARTSYS;
24384 ++ goto out;
24385 ++ }
24386 ++ schedule();
24387 ++ goto wait;
24388 ++
24389 ++out:
24390 ++ set_current_state(TASK_RUNNING);
24391 ++ remove_wait_queue(&vxi->vx_wait, &wait);
24392 ++ return ret;
24393 ++}
24394 ++
24395 ++
24396 ++
24397 ++int vc_wait_exit(struct vx_info *vxi, void __user *data)
24398 ++{
24399 ++ struct vcmd_wait_exit_v0 vc_data;
24400 ++ int ret;
24401 ++
24402 ++ ret = __wait_exit(vxi);
24403 ++ vc_data.reboot_cmd = vxi->reboot_cmd;
24404 ++ vc_data.exit_code = vxi->exit_code;
24405 ++
24406 ++ if (copy_to_user(data, &vc_data, sizeof(vc_data)))
24407 ++ ret = -EFAULT;
24408 ++ return ret;
24409 ++}
24410 ++
24411 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/space.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/space.c
24412 +--- linux-2.6.29.4/kernel/vserver/space.c 1970-01-01 01:00:00.000000000 +0100
24413 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/space.c 2009-05-29 18:59:49.000000000 +0200
24414 +@@ -0,0 +1,375 @@
24415 ++/*
24416 ++ * linux/kernel/vserver/space.c
24417 ++ *
24418 ++ * Virtual Server: Context Space Support
24419 ++ *
24420 ++ * Copyright (C) 2003-2007 Herbert Pötzl
24421 ++ *
24422 ++ * V0.01 broken out from context.c 0.07
24423 ++ * V0.02 added task locking for namespace
24424 ++ * V0.03 broken out vx_enter_namespace
24425 ++ * V0.04 added *space support and commands
24426 ++ *
24427 ++ */
24428 ++
24429 ++#include <linux/utsname.h>
24430 ++#include <linux/nsproxy.h>
24431 ++#include <linux/err.h>
24432 ++#include <asm/uaccess.h>
24433 ++
24434 ++#include <linux/vs_context.h>
24435 ++#include <linux/vserver/space.h>
24436 ++#include <linux/vserver/space_cmd.h>
24437 ++
24438 ++atomic_t vs_global_nsproxy = ATOMIC_INIT(0);
24439 ++atomic_t vs_global_fs = ATOMIC_INIT(0);
24440 ++atomic_t vs_global_mnt_ns = ATOMIC_INIT(0);
24441 ++atomic_t vs_global_uts_ns = ATOMIC_INIT(0);
24442 ++atomic_t vs_global_ipc_ns = ATOMIC_INIT(0);
24443 ++atomic_t vs_global_user_ns = ATOMIC_INIT(0);
24444 ++atomic_t vs_global_pid_ns = ATOMIC_INIT(0);
24445 ++
24446 ++
24447 ++/* namespace functions */
24448 ++
24449 ++#include <linux/mnt_namespace.h>
24450 ++#include <linux/user_namespace.h>
24451 ++#include <linux/pid_namespace.h>
24452 ++#include <linux/ipc_namespace.h>
24453 ++#include <net/net_namespace.h>
24454 ++
24455 ++
24456 ++static const struct vcmd_space_mask_v1 space_mask_v0 = {
24457 ++ .mask = CLONE_FS |
24458 ++ CLONE_NEWNS |
24459 ++ CLONE_NEWUTS |
24460 ++ CLONE_NEWIPC |
24461 ++ CLONE_NEWUSER |
24462 ++ 0
24463 ++};
24464 ++
24465 ++static const struct vcmd_space_mask_v1 space_mask = {
24466 ++ .mask = CLONE_FS |
24467 ++ CLONE_NEWNS |
24468 ++ CLONE_NEWUTS |
24469 ++ CLONE_NEWIPC |
24470 ++ CLONE_NEWUSER |
24471 ++#ifdef CONFIG_PID_NS
24472 ++ CLONE_NEWPID |
24473 ++#endif
24474 ++#ifdef CONFIG_NET_NS
24475 ++ CLONE_NEWNET |
24476 ++#endif
24477 ++ 0
24478 ++};
24479 ++
24480 ++static const struct vcmd_space_mask_v1 default_space_mask = {
24481 ++ .mask = CLONE_FS |
24482 ++ CLONE_NEWNS |
24483 ++ CLONE_NEWUTS |
24484 ++ CLONE_NEWIPC |
24485 ++ CLONE_NEWUSER |
24486 ++#ifdef CONFIG_PID_NS
24487 ++// CLONE_NEWPID |
24488 ++#endif
24489 ++ 0
24490 ++};
24491 ++
24492 ++/*
24493 ++ * build a new nsproxy mix
24494 ++ * assumes that both proxies are 'const'
24495 ++ * does not touch nsproxy refcounts
24496 ++ * will hold a reference on the result.
24497 ++ */
24498 ++
24499 ++struct nsproxy *vs_mix_nsproxy(struct nsproxy *old_nsproxy,
24500 ++ struct nsproxy *new_nsproxy, unsigned long mask)
24501 ++{
24502 ++ struct mnt_namespace *old_ns;
24503 ++ struct uts_namespace *old_uts;
24504 ++ struct ipc_namespace *old_ipc;
24505 ++#ifdef CONFIG_PID_NS
24506 ++ struct pid_namespace *old_pid;
24507 ++#endif
24508 ++#ifdef CONFIG_NET_NS
24509 ++ struct net *old_net;
24510 ++#endif
24511 ++ struct nsproxy *nsproxy;
24512 ++
24513 ++ nsproxy = copy_nsproxy(old_nsproxy);
24514 ++ if (!nsproxy)
24515 ++ goto out;
24516 ++
24517 ++ if (mask & CLONE_NEWNS) {
24518 ++ old_ns = nsproxy->mnt_ns;
24519 ++ nsproxy->mnt_ns = new_nsproxy->mnt_ns;
24520 ++ if (nsproxy->mnt_ns)
24521 ++ get_mnt_ns(nsproxy->mnt_ns);
24522 ++ } else
24523 ++ old_ns = NULL;
24524 ++
24525 ++ if (mask & CLONE_NEWUTS) {
24526 ++ old_uts = nsproxy->uts_ns;
24527 ++ nsproxy->uts_ns = new_nsproxy->uts_ns;
24528 ++ if (nsproxy->uts_ns)
24529 ++ get_uts_ns(nsproxy->uts_ns);
24530 ++ } else
24531 ++ old_uts = NULL;
24532 ++
24533 ++ if (mask & CLONE_NEWIPC) {
24534 ++ old_ipc = nsproxy->ipc_ns;
24535 ++ nsproxy->ipc_ns = new_nsproxy->ipc_ns;
24536 ++ if (nsproxy->ipc_ns)
24537 ++ get_ipc_ns(nsproxy->ipc_ns);
24538 ++ } else
24539 ++ old_ipc = NULL;
24540 ++
24541 ++#ifdef CONFIG_PID_NS
24542 ++ if (mask & CLONE_NEWPID) {
24543 ++ old_pid = nsproxy->pid_ns;
24544 ++ nsproxy->pid_ns = new_nsproxy->pid_ns;
24545 ++ if (nsproxy->pid_ns)
24546 ++ get_pid_ns(nsproxy->pid_ns);
24547 ++ } else
24548 ++ old_pid = NULL;
24549 ++#endif
24550 ++#ifdef CONFIG_NET_NS
24551 ++ if (mask & CLONE_NEWNET) {
24552 ++ old_net = nsproxy->net_ns;
24553 ++ nsproxy->net_ns = new_nsproxy->net_ns;
24554 ++ if (nsproxy->net_ns)
24555 ++ get_net(nsproxy->net_ns);
24556 ++ } else
24557 ++ old_net = NULL;
24558 ++#endif
24559 ++ if (old_ns)
24560 ++ put_mnt_ns(old_ns);
24561 ++ if (old_uts)
24562 ++ put_uts_ns(old_uts);
24563 ++ if (old_ipc)
24564 ++ put_ipc_ns(old_ipc);
24565 ++#ifdef CONFIG_PID_NS
24566 ++ if (old_pid)
24567 ++ put_pid_ns(old_pid);
24568 ++#endif
24569 ++#ifdef CONFIG_NET_NS
24570 ++ if (old_net)
24571 ++ put_net(old_net);
24572 ++#endif
24573 ++out:
24574 ++ return nsproxy;
24575 ++}
24576 ++
24577 ++
24578 ++/*
24579 ++ * merge two nsproxy structs into a new one.
24580 ++ * will hold a reference on the result.
24581 ++ */
24582 ++
24583 ++static inline
24584 ++struct nsproxy *__vs_merge_nsproxy(struct nsproxy *old,
24585 ++ struct nsproxy *proxy, unsigned long mask)
24586 ++{
24587 ++ struct nsproxy null_proxy = { .mnt_ns = NULL };
24588 ++
24589 ++ if (!proxy)
24590 ++ return NULL;
24591 ++
24592 ++ if (mask) {
24593 ++ /* vs_mix_nsproxy returns with reference */
24594 ++ return vs_mix_nsproxy(old ? old : &null_proxy,
24595 ++ proxy, mask);
24596 ++ }
24597 ++ get_nsproxy(proxy);
24598 ++ return proxy;
24599 ++}
24600 ++
24601 ++
24602 ++int vx_enter_space(struct vx_info *vxi, unsigned long mask, unsigned index)
24603 ++{
24604 ++ struct nsproxy *proxy, *proxy_cur, *proxy_new;
24605 ++ struct fs_struct *fs_cur, *fs = NULL;
24606 ++ int ret, kill = 0;
24607 ++
24608 ++ vxdprintk(VXD_CBIT(space, 8), "vx_enter_space(%p[#%u],0x%08lx,%d)",
24609 ++ vxi, vxi->vx_id, mask, index);
24610 ++
24611 ++ if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
24612 ++ return -EACCES;
24613 ++
24614 ++ if (!mask)
24615 ++ mask = vxi->vx_nsmask[index];
24616 ++
24617 ++ if ((mask & vxi->vx_nsmask[index]) != mask)
24618 ++ return -EINVAL;
24619 ++
24620 ++ if (mask & CLONE_FS) {
24621 ++ fs = copy_fs_struct(vxi->vx_fs[index]);
24622 ++ if (!fs)
24623 ++ return -ENOMEM;
24624 ++ }
24625 ++ proxy = vxi->vx_nsproxy[index];
24626 ++
24627 ++ vxdprintk(VXD_CBIT(space, 9),
24628 ++ "vx_enter_space(%p[#%u],0x%08lx,%d) -> (%p,%p)",
24629 ++ vxi, vxi->vx_id, mask, index, proxy, fs);
24630 ++
24631 ++ task_lock(current);
24632 ++ fs_cur = current->fs;
24633 ++
24634 ++ if (mask & CLONE_FS) {
24635 ++ write_lock(&fs_cur->lock);
24636 ++ current->fs = fs;
24637 ++ kill = !--fs_cur->users;
24638 ++ write_unlock(&fs_cur->lock);
24639 ++ }
24640 ++
24641 ++ proxy_cur = current->nsproxy;
24642 ++ get_nsproxy(proxy_cur);
24643 ++ task_unlock(current);
24644 ++
24645 ++ if (kill)
24646 ++ free_fs_struct(fs_cur);
24647 ++
24648 ++ proxy_new = __vs_merge_nsproxy(proxy_cur, proxy, mask);
24649 ++ if (IS_ERR(proxy_new)) {
24650 ++ ret = PTR_ERR(proxy_new);
24651 ++ goto out_put;
24652 ++ }
24653 ++
24654 ++ proxy_new = xchg(&current->nsproxy, proxy_new);
24655 ++ ret = 0;
24656 ++
24657 ++ if (proxy_new)
24658 ++ put_nsproxy(proxy_new);
24659 ++out_put:
24660 ++ if (proxy_cur)
24661 ++ put_nsproxy(proxy_cur);
24662 ++ return ret;
24663 ++}
24664 ++
24665 ++
24666 ++int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index)
24667 ++{
24668 ++ struct nsproxy *proxy_vxi, *proxy_cur, *proxy_new;
24669 ++ struct fs_struct *fs_vxi, *fs;
24670 ++ int ret, kill = 0;
24671 ++
24672 ++ vxdprintk(VXD_CBIT(space, 8), "vx_set_space(%p[#%u],0x%08lx,%d)",
24673 ++ vxi, vxi->vx_id, mask, index);
24674 ++#if 0
24675 ++ if (!mask)
24676 ++ mask = default_space_mask.mask;
24677 ++#endif
24678 ++ if ((mask & space_mask.mask) != mask)
24679 ++ return -EINVAL;
24680 ++
24681 ++ proxy_vxi = vxi->vx_nsproxy[index];
24682 ++ fs_vxi = vxi->vx_fs[index];
24683 ++
24684 ++ if (mask & CLONE_FS) {
24685 ++ fs = copy_fs_struct(current->fs);
24686 ++ if (!fs)
24687 ++ return -ENOMEM;
24688 ++ }
24689 ++
24690 ++ task_lock(current);
24691 ++
24692 ++ if (mask & CLONE_FS) {
24693 ++ write_lock(&fs_vxi->lock);
24694 ++ vxi->vx_fs[index] = fs;
24695 ++ kill = !--fs_vxi->users;
24696 ++ write_unlock(&fs_vxi->lock);
24697 ++ }
24698 ++
24699 ++ proxy_cur = current->nsproxy;
24700 ++ get_nsproxy(proxy_cur);
24701 ++ task_unlock(current);
24702 ++
24703 ++ if (kill)
24704 ++ free_fs_struct(fs_vxi);
24705 ++
24706 ++ proxy_new = __vs_merge_nsproxy(proxy_vxi, proxy_cur, mask);
24707 ++ if (IS_ERR(proxy_new)) {
24708 ++ ret = PTR_ERR(proxy_new);
24709 ++ goto out_put;
24710 ++ }
24711 ++
24712 ++ proxy_new = xchg(&vxi->vx_nsproxy[index], proxy_new);
24713 ++ vxi->vx_nsmask[index] |= mask;
24714 ++ ret = 0;
24715 ++
24716 ++ if (proxy_new)
24717 ++ put_nsproxy(proxy_new);
24718 ++out_put:
24719 ++ if (proxy_cur)
24720 ++ put_nsproxy(proxy_cur);
24721 ++ return ret;
24722 ++}
24723 ++
24724 ++
24725 ++int vc_enter_space_v1(struct vx_info *vxi, void __user *data)
24726 ++{
24727 ++ struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
24728 ++
24729 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24730 ++ return -EFAULT;
24731 ++
24732 ++ return vx_enter_space(vxi, vc_data.mask, 0);
24733 ++}
24734 ++
24735 ++int vc_enter_space(struct vx_info *vxi, void __user *data)
24736 ++{
24737 ++ struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
24738 ++
24739 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24740 ++ return -EFAULT;
24741 ++
24742 ++ if (vc_data.index >= VX_SPACES)
24743 ++ return -EINVAL;
24744 ++
24745 ++ return vx_enter_space(vxi, vc_data.mask, vc_data.index);
24746 ++}
24747 ++
24748 ++int vc_set_space_v1(struct vx_info *vxi, void __user *data)
24749 ++{
24750 ++ struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
24751 ++
24752 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24753 ++ return -EFAULT;
24754 ++
24755 ++ return vx_set_space(vxi, vc_data.mask, 0);
24756 ++}
24757 ++
24758 ++int vc_set_space(struct vx_info *vxi, void __user *data)
24759 ++{
24760 ++ struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
24761 ++
24762 ++ if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
24763 ++ return -EFAULT;
24764 ++
24765 ++ if (vc_data.index >= VX_SPACES)
24766 ++ return -EINVAL;
24767 ++
24768 ++ return vx_set_space(vxi, vc_data.mask, vc_data.index);
24769 ++}
24770 ++
24771 ++int vc_get_space_mask(void __user *data, int type)
24772 ++{
24773 ++ const struct vcmd_space_mask_v1 *mask;
24774 ++
24775 ++ if (type == 0)
24776 ++ mask = &space_mask_v0;
24777 ++ else if (type == 1)
24778 ++ mask = &space_mask;
24779 ++ else
24780 ++ mask = &default_space_mask;
24781 ++
24782 ++ vxdprintk(VXD_CBIT(space, 10),
24783 ++ "vc_get_space_mask(%d) = %08llx", type, mask->mask);
24784 ++
24785 ++ if (copy_to_user(data, mask, sizeof(*mask)))
24786 ++ return -EFAULT;
24787 ++ return 0;
24788 ++}
24789 ++
24790 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/switch.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/switch.c
24791 +--- linux-2.6.29.4/kernel/vserver/switch.c 1970-01-01 01:00:00.000000000 +0100
24792 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/switch.c 2009-03-24 14:59:18.000000000 +0100
24793 +@@ -0,0 +1,543 @@
24794 ++/*
24795 ++ * linux/kernel/vserver/switch.c
24796 ++ *
24797 ++ * Virtual Server: Syscall Switch
24798 ++ *
24799 ++ * Copyright (C) 2003-2007 Herbert Pötzl
24800 ++ *
24801 ++ * V0.01 syscall switch
24802 ++ * V0.02 added signal to context
24803 ++ * V0.03 added rlimit functions
24804 ++ * V0.04 added iattr, task/xid functions
24805 ++ * V0.05 added debug/history stuff
24806 ++ * V0.06 added compat32 layer
24807 ++ * V0.07 vcmd args and perms
24808 ++ * V0.08 added status commands
24809 ++ * V0.09 added tag commands
24810 ++ * V0.10 added oom bias
24811 ++ * V0.11 added device commands
24812 ++ *
24813 ++ */
24814 ++
24815 ++#include <linux/vs_context.h>
24816 ++#include <linux/vs_network.h>
24817 ++#include <linux/vserver/switch.h>
24818 ++
24819 ++#include "vci_config.h"
24820 ++
24821 ++
24822 ++static inline
24823 ++int vc_get_version(uint32_t id)
24824 ++{
24825 ++ return VCI_VERSION;
24826 ++}
24827 ++
24828 ++static inline
24829 ++int vc_get_vci(uint32_t id)
24830 ++{
24831 ++ return vci_kernel_config();
24832 ++}
24833 ++
24834 ++#include <linux/vserver/context_cmd.h>
24835 ++#include <linux/vserver/cvirt_cmd.h>
24836 ++#include <linux/vserver/cacct_cmd.h>
24837 ++#include <linux/vserver/limit_cmd.h>
24838 ++#include <linux/vserver/network_cmd.h>
24839 ++#include <linux/vserver/sched_cmd.h>
24840 ++#include <linux/vserver/debug_cmd.h>
24841 ++#include <linux/vserver/inode_cmd.h>
24842 ++#include <linux/vserver/dlimit_cmd.h>
24843 ++#include <linux/vserver/signal_cmd.h>
24844 ++#include <linux/vserver/space_cmd.h>
24845 ++#include <linux/vserver/tag_cmd.h>
24846 ++#include <linux/vserver/device_cmd.h>
24847 ++
24848 ++#include <linux/vserver/inode.h>
24849 ++#include <linux/vserver/dlimit.h>
24850 ++
24851 ++
24852 ++#ifdef CONFIG_COMPAT
24853 ++#define __COMPAT(name, id, data, compat) \
24854 ++ (compat) ? name ## _x32(id, data) : name(id, data)
24855 ++#define __COMPAT_NO_ID(name, data, compat) \
24856 ++ (compat) ? name ## _x32(data) : name(data)
24857 ++#else
24858 ++#define __COMPAT(name, id, data, compat) \
24859 ++ name(id, data)
24860 ++#define __COMPAT_NO_ID(name, data, compat) \
24861 ++ name(data)
24862 ++#endif
24863 ++
24864 ++
24865 ++static inline
24866 ++long do_vcmd(uint32_t cmd, uint32_t id,
24867 ++ struct vx_info *vxi, struct nx_info *nxi,
24868 ++ void __user *data, int compat)
24869 ++{
24870 ++ switch (cmd) {
24871 ++
24872 ++ case VCMD_get_version:
24873 ++ return vc_get_version(id);
24874 ++ case VCMD_get_vci:
24875 ++ return vc_get_vci(id);
24876 ++
24877 ++ case VCMD_task_xid:
24878 ++ return vc_task_xid(id);
24879 ++ case VCMD_vx_info:
24880 ++ return vc_vx_info(vxi, data);
24881 ++
24882 ++ case VCMD_task_nid:
24883 ++ return vc_task_nid(id);
24884 ++ case VCMD_nx_info:
24885 ++ return vc_nx_info(nxi, data);
24886 ++
24887 ++ case VCMD_task_tag:
24888 ++ return vc_task_tag(id);
24889 ++
24890 ++ case VCMD_set_space_v1:
24891 ++ return vc_set_space_v1(vxi, data);
24892 ++ /* this is version 2 */
24893 ++ case VCMD_set_space:
24894 ++ return vc_set_space(vxi, data);
24895 ++
24896 ++ case VCMD_get_space_mask_v0:
24897 ++ return vc_get_space_mask(data, 0);
24898 ++ /* this is version 1 */
24899 ++ case VCMD_get_space_mask:
24900 ++ return vc_get_space_mask(data, 1);
24901 ++
24902 ++ case VCMD_get_space_default:
24903 ++ return vc_get_space_mask(data, -1);
24904 ++
24905 ++#ifdef CONFIG_IA32_EMULATION
24906 ++ case VCMD_get_rlimit:
24907 ++ return __COMPAT(vc_get_rlimit, vxi, data, compat);
24908 ++ case VCMD_set_rlimit:
24909 ++ return __COMPAT(vc_set_rlimit, vxi, data, compat);
24910 ++#else
24911 ++ case VCMD_get_rlimit:
24912 ++ return vc_get_rlimit(vxi, data);
24913 ++ case VCMD_set_rlimit:
24914 ++ return vc_set_rlimit(vxi, data);
24915 ++#endif
24916 ++ case VCMD_get_rlimit_mask:
24917 ++ return vc_get_rlimit_mask(id, data);
24918 ++ case VCMD_reset_minmax:
24919 ++ return vc_reset_minmax(vxi, data);
24920 ++
24921 ++ case VCMD_get_vhi_name:
24922 ++ return vc_get_vhi_name(vxi, data);
24923 ++ case VCMD_set_vhi_name:
24924 ++ return vc_set_vhi_name(vxi, data);
24925 ++
24926 ++ case VCMD_ctx_stat:
24927 ++ return vc_ctx_stat(vxi, data);
24928 ++ case VCMD_virt_stat:
24929 ++ return vc_virt_stat(vxi, data);
24930 ++ case VCMD_sock_stat:
24931 ++ return vc_sock_stat(vxi, data);
24932 ++ case VCMD_rlimit_stat:
24933 ++ return vc_rlimit_stat(vxi, data);
24934 ++
24935 ++ case VCMD_set_cflags:
24936 ++ return vc_set_cflags(vxi, data);
24937 ++ case VCMD_get_cflags:
24938 ++ return vc_get_cflags(vxi, data);
24939 ++
24940 ++ /* this is version 1 */
24941 ++ case VCMD_set_ccaps:
24942 ++ return vc_set_ccaps(vxi, data);
24943 ++ /* this is version 1 */
24944 ++ case VCMD_get_ccaps:
24945 ++ return vc_get_ccaps(vxi, data);
24946 ++ case VCMD_set_bcaps:
24947 ++ return vc_set_bcaps(vxi, data);
24948 ++ case VCMD_get_bcaps:
24949 ++ return vc_get_bcaps(vxi, data);
24950 ++
24951 ++ case VCMD_set_badness:
24952 ++ return vc_set_badness(vxi, data);
24953 ++ case VCMD_get_badness:
24954 ++ return vc_get_badness(vxi, data);
24955 ++
24956 ++ case VCMD_set_nflags:
24957 ++ return vc_set_nflags(nxi, data);
24958 ++ case VCMD_get_nflags:
24959 ++ return vc_get_nflags(nxi, data);
24960 ++
24961 ++ case VCMD_set_ncaps:
24962 ++ return vc_set_ncaps(nxi, data);
24963 ++ case VCMD_get_ncaps:
24964 ++ return vc_get_ncaps(nxi, data);
24965 ++
24966 ++ case VCMD_set_sched_v4:
24967 ++ return vc_set_sched_v4(vxi, data);
24968 ++ /* this is version 5 */
24969 ++ case VCMD_set_sched:
24970 ++ return vc_set_sched(vxi, data);
24971 ++ case VCMD_get_sched:
24972 ++ return vc_get_sched(vxi, data);
24973 ++ case VCMD_sched_info:
24974 ++ return vc_sched_info(vxi, data);
24975 ++
24976 ++ case VCMD_add_dlimit:
24977 ++ return __COMPAT(vc_add_dlimit, id, data, compat);
24978 ++ case VCMD_rem_dlimit:
24979 ++ return __COMPAT(vc_rem_dlimit, id, data, compat);
24980 ++ case VCMD_set_dlimit:
24981 ++ return __COMPAT(vc_set_dlimit, id, data, compat);
24982 ++ case VCMD_get_dlimit:
24983 ++ return __COMPAT(vc_get_dlimit, id, data, compat);
24984 ++
24985 ++ case VCMD_ctx_kill:
24986 ++ return vc_ctx_kill(vxi, data);
24987 ++
24988 ++ case VCMD_wait_exit:
24989 ++ return vc_wait_exit(vxi, data);
24990 ++
24991 ++ case VCMD_get_iattr:
24992 ++ return __COMPAT_NO_ID(vc_get_iattr, data, compat);
24993 ++ case VCMD_set_iattr:
24994 ++ return __COMPAT_NO_ID(vc_set_iattr, data, compat);
24995 ++
24996 ++ case VCMD_fget_iattr:
24997 ++ return vc_fget_iattr(id, data);
24998 ++ case VCMD_fset_iattr:
24999 ++ return vc_fset_iattr(id, data);
25000 ++
25001 ++ case VCMD_enter_space_v0:
25002 ++ return vc_enter_space_v1(vxi, NULL);
25003 ++ case VCMD_enter_space_v1:
25004 ++ return vc_enter_space_v1(vxi, data);
25005 ++ /* this is version 2 */
25006 ++ case VCMD_enter_space:
25007 ++ return vc_enter_space(vxi, data);
25008 ++
25009 ++ case VCMD_ctx_create_v0:
25010 ++ return vc_ctx_create(id, NULL);
25011 ++ case VCMD_ctx_create:
25012 ++ return vc_ctx_create(id, data);
25013 ++ case VCMD_ctx_migrate_v0:
25014 ++ return vc_ctx_migrate(vxi, NULL);
25015 ++ case VCMD_ctx_migrate:
25016 ++ return vc_ctx_migrate(vxi, data);
25017 ++
25018 ++ case VCMD_net_create_v0:
25019 ++ return vc_net_create(id, NULL);
25020 ++ case VCMD_net_create:
25021 ++ return vc_net_create(id, data);
25022 ++ case VCMD_net_migrate:
25023 ++ return vc_net_migrate(nxi, data);
25024 ++
25025 ++ case VCMD_tag_migrate:
25026 ++ return vc_tag_migrate(id);
25027 ++
25028 ++ case VCMD_net_add:
25029 ++ return vc_net_add(nxi, data);
25030 ++ case VCMD_net_remove:
25031 ++ return vc_net_remove(nxi, data);
25032 ++
25033 ++ case VCMD_net_add_ipv4:
25034 ++ return vc_net_add_ipv4(nxi, data);
25035 ++ case VCMD_net_remove_ipv4:
25036 ++ return vc_net_remove_ipv4(nxi, data);
25037 ++#ifdef CONFIG_IPV6
25038 ++ case VCMD_net_add_ipv6:
25039 ++ return vc_net_add_ipv6(nxi, data);
25040 ++ case VCMD_net_remove_ipv6:
25041 ++ return vc_net_remove_ipv6(nxi, data);
25042 ++#endif
25043 ++/* case VCMD_add_match_ipv4:
25044 ++ return vc_add_match_ipv4(nxi, data);
25045 ++ case VCMD_get_match_ipv4:
25046 ++ return vc_get_match_ipv4(nxi, data);
25047 ++#ifdef CONFIG_IPV6
25048 ++ case VCMD_add_match_ipv6:
25049 ++ return vc_add_match_ipv6(nxi, data);
25050 ++ case VCMD_get_match_ipv6:
25051 ++ return vc_get_match_ipv6(nxi, data);
25052 ++#endif */
25053 ++
25054 ++#ifdef CONFIG_VSERVER_DEVICE
25055 ++ case VCMD_set_mapping:
25056 ++ return __COMPAT(vc_set_mapping, vxi, data, compat);
25057 ++ case VCMD_unset_mapping:
25058 ++ return __COMPAT(vc_unset_mapping, vxi, data, compat);
25059 ++#endif
25060 ++#ifdef CONFIG_VSERVER_HISTORY
25061 ++ case VCMD_dump_history:
25062 ++ return vc_dump_history(id);
25063 ++ case VCMD_read_history:
25064 ++ return __COMPAT(vc_read_history, id, data, compat);
25065 ++#endif
25066 ++#ifdef CONFIG_VSERVER_MONITOR
25067 ++ case VCMD_read_monitor:
25068 ++ return __COMPAT(vc_read_monitor, id, data, compat);
25069 ++#endif
25070 ++ default:
25071 ++ vxwprintk_task(1, "unimplemented VCMD_%02d_%d[%d]",
25072 ++ VC_CATEGORY(cmd), VC_COMMAND(cmd), VC_VERSION(cmd));
25073 ++ }
25074 ++ return -ENOSYS;
25075 ++}
25076 ++
25077 ++
25078 ++#define __VCMD(vcmd, _perm, _args, _flags) \
25079 ++ case VCMD_ ## vcmd: perm = _perm; \
25080 ++ args = _args; flags = _flags; break
25081 ++
25082 ++
25083 ++#define VCA_NONE 0x00
25084 ++#define VCA_VXI 0x01
25085 ++#define VCA_NXI 0x02
25086 ++
25087 ++#define VCF_NONE 0x00
25088 ++#define VCF_INFO 0x01
25089 ++#define VCF_ADMIN 0x02
25090 ++#define VCF_ARES 0x06 /* includes admin */
25091 ++#define VCF_SETUP 0x08
25092 ++
25093 ++#define VCF_ZIDOK 0x10 /* zero id okay */
25094 ++
25095 ++
25096 ++static inline
25097 ++long do_vserver(uint32_t cmd, uint32_t id, void __user *data, int compat)
25098 ++{
25099 ++ long ret;
25100 ++ int permit = -1, state = 0;
25101 ++ int perm = -1, args = 0, flags = 0;
25102 ++ struct vx_info *vxi = NULL;
25103 ++ struct nx_info *nxi = NULL;
25104 ++
25105 ++ switch (cmd) {
25106 ++ /* unpriviledged commands */
25107 ++ __VCMD(get_version, 0, VCA_NONE, 0);
25108 ++ __VCMD(get_vci, 0, VCA_NONE, 0);
25109 ++ __VCMD(get_rlimit_mask, 0, VCA_NONE, 0);
25110 ++ __VCMD(get_space_mask_v0,0, VCA_NONE, 0);
25111 ++ __VCMD(get_space_mask, 0, VCA_NONE, 0);
25112 ++ __VCMD(get_space_default,0, VCA_NONE, 0);
25113 ++
25114 ++ /* info commands */
25115 ++ __VCMD(task_xid, 2, VCA_NONE, 0);
25116 ++ __VCMD(reset_minmax, 2, VCA_VXI, 0);
25117 ++ __VCMD(vx_info, 3, VCA_VXI, VCF_INFO);
25118 ++ __VCMD(get_bcaps, 3, VCA_VXI, VCF_INFO);
25119 ++ __VCMD(get_ccaps, 3, VCA_VXI, VCF_INFO);
25120 ++ __VCMD(get_cflags, 3, VCA_VXI, VCF_INFO);
25121 ++ __VCMD(get_badness, 3, VCA_VXI, VCF_INFO);
25122 ++ __VCMD(get_vhi_name, 3, VCA_VXI, VCF_INFO);
25123 ++ __VCMD(get_rlimit, 3, VCA_VXI, VCF_INFO);
25124 ++
25125 ++ __VCMD(ctx_stat, 3, VCA_VXI, VCF_INFO);
25126 ++ __VCMD(virt_stat, 3, VCA_VXI, VCF_INFO);
25127 ++ __VCMD(sock_stat, 3, VCA_VXI, VCF_INFO);
25128 ++ __VCMD(rlimit_stat, 3, VCA_VXI, VCF_INFO);
25129 ++
25130 ++ __VCMD(task_nid, 2, VCA_NONE, 0);
25131 ++ __VCMD(nx_info, 3, VCA_NXI, VCF_INFO);
25132 ++ __VCMD(get_ncaps, 3, VCA_NXI, VCF_INFO);
25133 ++ __VCMD(get_nflags, 3, VCA_NXI, VCF_INFO);
25134 ++
25135 ++ __VCMD(task_tag, 2, VCA_NONE, 0);
25136 ++
25137 ++ __VCMD(get_iattr, 2, VCA_NONE, 0);
25138 ++ __VCMD(fget_iattr, 2, VCA_NONE, 0);
25139 ++ __VCMD(get_dlimit, 3, VCA_NONE, VCF_INFO);
25140 ++ __VCMD(get_sched, 3, VCA_VXI, VCF_INFO);
25141 ++ __VCMD(sched_info, 3, VCA_VXI, VCF_INFO | VCF_ZIDOK);
25142 ++
25143 ++ /* lower admin commands */
25144 ++ __VCMD(wait_exit, 4, VCA_VXI, VCF_INFO);
25145 ++ __VCMD(ctx_create_v0, 5, VCA_NONE, 0);
25146 ++ __VCMD(ctx_create, 5, VCA_NONE, 0);
25147 ++ __VCMD(ctx_migrate_v0, 5, VCA_VXI, VCF_ADMIN);
25148 ++ __VCMD(ctx_migrate, 5, VCA_VXI, VCF_ADMIN);
25149 ++ __VCMD(enter_space_v0, 5, VCA_VXI, VCF_ADMIN);
25150 ++ __VCMD(enter_space_v1, 5, VCA_VXI, VCF_ADMIN);
25151 ++ __VCMD(enter_space, 5, VCA_VXI, VCF_ADMIN);
25152 ++
25153 ++ __VCMD(net_create_v0, 5, VCA_NONE, 0);
25154 ++ __VCMD(net_create, 5, VCA_NONE, 0);
25155 ++ __VCMD(net_migrate, 5, VCA_NXI, VCF_ADMIN);
25156 ++
25157 ++ __VCMD(tag_migrate, 5, VCA_NONE, VCF_ADMIN);
25158 ++
25159 ++ /* higher admin commands */
25160 ++ __VCMD(ctx_kill, 6, VCA_VXI, VCF_ARES);
25161 ++ __VCMD(set_space_v1, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25162 ++ __VCMD(set_space, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25163 ++
25164 ++ __VCMD(set_ccaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25165 ++ __VCMD(set_bcaps, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25166 ++ __VCMD(set_cflags, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25167 ++ __VCMD(set_badness, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25168 ++
25169 ++ __VCMD(set_vhi_name, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25170 ++ __VCMD(set_rlimit, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25171 ++ __VCMD(set_sched, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25172 ++ __VCMD(set_sched_v4, 7, VCA_VXI, VCF_ARES | VCF_SETUP);
25173 ++
25174 ++ __VCMD(set_ncaps, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25175 ++ __VCMD(set_nflags, 7, VCA_NXI, VCF_ARES | VCF_SETUP);
25176 ++ __VCMD(net_add, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25177 ++ __VCMD(net_remove, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25178 ++ __VCMD(net_add_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25179 ++ __VCMD(net_remove_ipv4, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25180 ++#ifdef CONFIG_IPV6
25181 ++ __VCMD(net_add_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25182 ++ __VCMD(net_remove_ipv6, 8, VCA_NXI, VCF_ARES | VCF_SETUP);
25183 ++#endif
25184 ++ __VCMD(set_iattr, 7, VCA_NONE, 0);
25185 ++ __VCMD(fset_iattr, 7, VCA_NONE, 0);
25186 ++ __VCMD(set_dlimit, 7, VCA_NONE, VCF_ARES);
25187 ++ __VCMD(add_dlimit, 8, VCA_NONE, VCF_ARES);
25188 ++ __VCMD(rem_dlimit, 8, VCA_NONE, VCF_ARES);
25189 ++
25190 ++#ifdef CONFIG_VSERVER_DEVICE
25191 ++ __VCMD(set_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25192 ++ __VCMD(unset_mapping, 8, VCA_VXI, VCF_ARES|VCF_ZIDOK);
25193 ++#endif
25194 ++ /* debug level admin commands */
25195 ++#ifdef CONFIG_VSERVER_HISTORY
25196 ++ __VCMD(dump_history, 9, VCA_NONE, 0);
25197 ++ __VCMD(read_history, 9, VCA_NONE, 0);
25198 ++#endif
25199 ++#ifdef CONFIG_VSERVER_MONITOR
25200 ++ __VCMD(read_monitor, 9, VCA_NONE, 0);
25201 ++#endif
25202 ++
25203 ++ default:
25204 ++ perm = -1;
25205 ++ }
25206 ++
25207 ++ vxdprintk(VXD_CBIT(switch, 0),
25208 ++ "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]",
25209 ++ VC_CATEGORY(cmd), VC_COMMAND(cmd),
25210 ++ VC_VERSION(cmd), id, data, compat,
25211 ++ perm, args, flags);
25212 ++
25213 ++ ret = -ENOSYS;
25214 ++ if (perm < 0)
25215 ++ goto out;
25216 ++
25217 ++ state = 1;
25218 ++ if (!capable(CAP_CONTEXT))
25219 ++ goto out;
25220 ++
25221 ++ state = 2;
25222 ++ /* moved here from the individual commands */
25223 ++ ret = -EPERM;
25224 ++ if ((perm > 1) && !capable(CAP_SYS_ADMIN))
25225 ++ goto out;
25226 ++
25227 ++ state = 3;
25228 ++ /* vcmd involves resource management */
25229 ++ ret = -EPERM;
25230 ++ if ((flags & VCF_ARES) && !capable(CAP_SYS_RESOURCE))
25231 ++ goto out;
25232 ++
25233 ++ state = 4;
25234 ++ /* various legacy exceptions */
25235 ++ switch (cmd) {
25236 ++ /* will go away when spectator is a cap */
25237 ++ case VCMD_ctx_migrate_v0:
25238 ++ case VCMD_ctx_migrate:
25239 ++ if (id == 1) {
25240 ++ current->xid = 1;
25241 ++ ret = 1;
25242 ++ goto out;
25243 ++ }
25244 ++ break;
25245 ++
25246 ++ /* will go away when spectator is a cap */
25247 ++ case VCMD_net_migrate:
25248 ++ if (id == 1) {
25249 ++ current->nid = 1;
25250 ++ ret = 1;
25251 ++ goto out;
25252 ++ }
25253 ++ break;
25254 ++ }
25255 ++
25256 ++ /* vcmds are fine by default */
25257 ++ permit = 1;
25258 ++
25259 ++ /* admin type vcmds require admin ... */
25260 ++ if (flags & VCF_ADMIN)
25261 ++ permit = vx_check(0, VS_ADMIN) ? 1 : 0;
25262 ++
25263 ++ /* ... but setup type vcmds override that */
25264 ++ if (!permit && (flags & VCF_SETUP))
25265 ++ permit = vx_flags(VXF_STATE_SETUP, 0) ? 2 : 0;
25266 ++
25267 ++ state = 5;
25268 ++ ret = -EPERM;
25269 ++ if (!permit)
25270 ++ goto out;
25271 ++
25272 ++ state = 6;
25273 ++ if (!id && (flags & VCF_ZIDOK))
25274 ++ goto skip_id;
25275 ++
25276 ++ ret = -ESRCH;
25277 ++ if (args & VCA_VXI) {
25278 ++ vxi = lookup_vx_info(id);
25279 ++ if (!vxi)
25280 ++ goto out;
25281 ++
25282 ++ if ((flags & VCF_ADMIN) &&
25283 ++ /* special case kill for shutdown */
25284 ++ (cmd != VCMD_ctx_kill) &&
25285 ++ /* can context be administrated? */
25286 ++ !vx_info_flags(vxi, VXF_STATE_ADMIN, 0)) {
25287 ++ ret = -EACCES;
25288 ++ goto out_vxi;
25289 ++ }
25290 ++ }
25291 ++ state = 7;
25292 ++ if (args & VCA_NXI) {
25293 ++ nxi = lookup_nx_info(id);
25294 ++ if (!nxi)
25295 ++ goto out_vxi;
25296 ++
25297 ++ if ((flags & VCF_ADMIN) &&
25298 ++ /* can context be administrated? */
25299 ++ !nx_info_flags(nxi, NXF_STATE_ADMIN, 0)) {
25300 ++ ret = -EACCES;
25301 ++ goto out_nxi;
25302 ++ }
25303 ++ }
25304 ++skip_id:
25305 ++ state = 8;
25306 ++ ret = do_vcmd(cmd, id, vxi, nxi, data, compat);
25307 ++
25308 ++out_nxi:
25309 ++ if ((args & VCA_NXI) && nxi)
25310 ++ put_nx_info(nxi);
25311 ++out_vxi:
25312 ++ if ((args & VCA_VXI) && vxi)
25313 ++ put_vx_info(vxi);
25314 ++out:
25315 ++ vxdprintk(VXD_CBIT(switch, 1),
25316 ++ "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]",
25317 ++ VC_CATEGORY(cmd), VC_COMMAND(cmd),
25318 ++ VC_VERSION(cmd), ret, ret, state, permit);
25319 ++ return ret;
25320 ++}
25321 ++
25322 ++asmlinkage long
25323 ++sys_vserver(uint32_t cmd, uint32_t id, void __user *data)
25324 ++{
25325 ++ return do_vserver(cmd, id, data, 0);
25326 ++}
25327 ++
25328 ++#ifdef CONFIG_COMPAT
25329 ++
25330 ++asmlinkage long
25331 ++sys32_vserver(uint32_t cmd, uint32_t id, void __user *data)
25332 ++{
25333 ++ return do_vserver(cmd, id, data, 1);
25334 ++}
25335 ++
25336 ++#endif /* CONFIG_COMPAT */
25337 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/sysctl.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sysctl.c
25338 +--- linux-2.6.29.4/kernel/vserver/sysctl.c 1970-01-01 01:00:00.000000000 +0100
25339 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/sysctl.c 2009-02-22 22:54:26.000000000 +0100
25340 +@@ -0,0 +1,244 @@
25341 ++/*
25342 ++ * kernel/vserver/sysctl.c
25343 ++ *
25344 ++ * Virtual Context Support
25345 ++ *
25346 ++ * Copyright (C) 2004-2007 Herbert Pötzl
25347 ++ *
25348 ++ * V0.01 basic structure
25349 ++ *
25350 ++ */
25351 ++
25352 ++#include <linux/module.h>
25353 ++#include <linux/ctype.h>
25354 ++#include <linux/sysctl.h>
25355 ++#include <linux/parser.h>
25356 ++#include <asm/uaccess.h>
25357 ++
25358 ++
25359 ++enum {
25360 ++ CTL_DEBUG_ERROR = 0,
25361 ++ CTL_DEBUG_SWITCH = 1,
25362 ++ CTL_DEBUG_XID,
25363 ++ CTL_DEBUG_NID,
25364 ++ CTL_DEBUG_TAG,
25365 ++ CTL_DEBUG_NET,
25366 ++ CTL_DEBUG_LIMIT,
25367 ++ CTL_DEBUG_CRES,
25368 ++ CTL_DEBUG_DLIM,
25369 ++ CTL_DEBUG_QUOTA,
25370 ++ CTL_DEBUG_CVIRT,
25371 ++ CTL_DEBUG_SPACE,
25372 ++ CTL_DEBUG_MISC,
25373 ++};
25374 ++
25375 ++
25376 ++unsigned int vx_debug_switch = 0;
25377 ++unsigned int vx_debug_xid = 0;
25378 ++unsigned int vx_debug_nid = 0;
25379 ++unsigned int vx_debug_tag = 0;
25380 ++unsigned int vx_debug_net = 0;
25381 ++unsigned int vx_debug_limit = 0;
25382 ++unsigned int vx_debug_cres = 0;
25383 ++unsigned int vx_debug_dlim = 0;
25384 ++unsigned int vx_debug_quota = 0;
25385 ++unsigned int vx_debug_cvirt = 0;
25386 ++unsigned int vx_debug_space = 0;
25387 ++unsigned int vx_debug_misc = 0;
25388 ++
25389 ++
25390 ++static struct ctl_table_header *vserver_table_header;
25391 ++static ctl_table vserver_root_table[];
25392 ++
25393 ++
25394 ++void vserver_register_sysctl(void)
25395 ++{
25396 ++ if (!vserver_table_header) {
25397 ++ vserver_table_header = register_sysctl_table(vserver_root_table);
25398 ++ }
25399 ++
25400 ++}
25401 ++
25402 ++void vserver_unregister_sysctl(void)
25403 ++{
25404 ++ if (vserver_table_header) {
25405 ++ unregister_sysctl_table(vserver_table_header);
25406 ++ vserver_table_header = NULL;
25407 ++ }
25408 ++}
25409 ++
25410 ++
25411 ++static int proc_dodebug(ctl_table *table, int write,
25412 ++ struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
25413 ++{
25414 ++ char tmpbuf[20], *p, c;
25415 ++ unsigned int value;
25416 ++ size_t left, len;
25417 ++
25418 ++ if ((*ppos && !write) || !*lenp) {
25419 ++ *lenp = 0;
25420 ++ return 0;
25421 ++ }
25422 ++
25423 ++ left = *lenp;
25424 ++
25425 ++ if (write) {
25426 ++ if (!access_ok(VERIFY_READ, buffer, left))
25427 ++ return -EFAULT;
25428 ++ p = (char *)buffer;
25429 ++ while (left && __get_user(c, p) >= 0 && isspace(c))
25430 ++ left--, p++;
25431 ++ if (!left)
25432 ++ goto done;
25433 ++
25434 ++ if (left > sizeof(tmpbuf) - 1)
25435 ++ return -EINVAL;
25436 ++ if (copy_from_user(tmpbuf, p, left))
25437 ++ return -EFAULT;
25438 ++ tmpbuf[left] = '\0';
25439 ++
25440 ++ for (p = tmpbuf, value = 0; '0' <= *p && *p <= '9'; p++, left--)
25441 ++ value = 10 * value + (*p - '0');
25442 ++ if (*p && !isspace(*p))
25443 ++ return -EINVAL;
25444 ++ while (left && isspace(*p))
25445 ++ left--, p++;
25446 ++ *(unsigned int *)table->data = value;
25447 ++ } else {
25448 ++ if (!access_ok(VERIFY_WRITE, buffer, left))
25449 ++ return -EFAULT;
25450 ++ len = sprintf(tmpbuf, "%d", *(unsigned int *)table->data);
25451 ++ if (len > left)
25452 ++ len = left;
25453 ++ if (__copy_to_user(buffer, tmpbuf, len))
25454 ++ return -EFAULT;
25455 ++ if ((left -= len) > 0) {
25456 ++ if (put_user('\n', (char *)buffer + len))
25457 ++ return -EFAULT;
25458 ++ left--;
25459 ++ }
25460 ++ }
25461 ++
25462 ++done:
25463 ++ *lenp -= left;
25464 ++ *ppos += *lenp;
25465 ++ return 0;
25466 ++}
25467 ++
25468 ++static int zero;
25469 ++
25470 ++#define CTL_ENTRY(ctl, name) \
25471 ++ { \
25472 ++ .ctl_name = ctl, \
25473 ++ .procname = #name, \
25474 ++ .data = &vx_ ## name, \
25475 ++ .maxlen = sizeof(int), \
25476 ++ .mode = 0644, \
25477 ++ .proc_handler = &proc_dodebug, \
25478 ++ .strategy = &sysctl_intvec, \
25479 ++ .extra1 = &zero, \
25480 ++ }
25481 ++
25482 ++static ctl_table vserver_debug_table[] = {
25483 ++ CTL_ENTRY(CTL_DEBUG_SWITCH, debug_switch),
25484 ++ CTL_ENTRY(CTL_DEBUG_XID, debug_xid),
25485 ++ CTL_ENTRY(CTL_DEBUG_NID, debug_nid),
25486 ++ CTL_ENTRY(CTL_DEBUG_TAG, debug_tag),
25487 ++ CTL_ENTRY(CTL_DEBUG_NET, debug_net),
25488 ++ CTL_ENTRY(CTL_DEBUG_LIMIT, debug_limit),
25489 ++ CTL_ENTRY(CTL_DEBUG_CRES, debug_cres),
25490 ++ CTL_ENTRY(CTL_DEBUG_DLIM, debug_dlim),
25491 ++ CTL_ENTRY(CTL_DEBUG_QUOTA, debug_quota),
25492 ++ CTL_ENTRY(CTL_DEBUG_CVIRT, debug_cvirt),
25493 ++ CTL_ENTRY(CTL_DEBUG_SPACE, debug_space),
25494 ++ CTL_ENTRY(CTL_DEBUG_MISC, debug_misc),
25495 ++ { .ctl_name = 0 }
25496 ++};
25497 ++
25498 ++static ctl_table vserver_root_table[] = {
25499 ++ {
25500 ++ .ctl_name = CTL_VSERVER,
25501 ++ .procname = "vserver",
25502 ++ .mode = 0555,
25503 ++ .child = vserver_debug_table
25504 ++ },
25505 ++ { .ctl_name = 0 }
25506 ++};
25507 ++
25508 ++
25509 ++static match_table_t tokens = {
25510 ++ { CTL_DEBUG_SWITCH, "switch=%x" },
25511 ++ { CTL_DEBUG_XID, "xid=%x" },
25512 ++ { CTL_DEBUG_NID, "nid=%x" },
25513 ++ { CTL_DEBUG_TAG, "tag=%x" },
25514 ++ { CTL_DEBUG_NET, "net=%x" },
25515 ++ { CTL_DEBUG_LIMIT, "limit=%x" },
25516 ++ { CTL_DEBUG_CRES, "cres=%x" },
25517 ++ { CTL_DEBUG_DLIM, "dlim=%x" },
25518 ++ { CTL_DEBUG_QUOTA, "quota=%x" },
25519 ++ { CTL_DEBUG_CVIRT, "cvirt=%x" },
25520 ++ { CTL_DEBUG_SPACE, "space=%x" },
25521 ++ { CTL_DEBUG_MISC, "misc=%x" },
25522 ++ { CTL_DEBUG_ERROR, NULL }
25523 ++};
25524 ++
25525 ++#define HANDLE_CASE(id, name, val) \
25526 ++ case CTL_DEBUG_ ## id: \
25527 ++ vx_debug_ ## name = val; \
25528 ++ printk("vs_debug_" #name "=0x%x\n", val); \
25529 ++ break
25530 ++
25531 ++
25532 ++static int __init vs_debug_setup(char *str)
25533 ++{
25534 ++ char *p;
25535 ++ int token;
25536 ++
25537 ++ printk("vs_debug_setup(%s)\n", str);
25538 ++ while ((p = strsep(&str, ",")) != NULL) {
25539 ++ substring_t args[MAX_OPT_ARGS];
25540 ++ unsigned int value;
25541 ++
25542 ++ if (!*p)
25543 ++ continue;
25544 ++
25545 ++ token = match_token(p, tokens, args);
25546 ++ value = (token > 0) ? simple_strtoul(args[0].from, NULL, 0) : 0;
25547 ++
25548 ++ switch (token) {
25549 ++ HANDLE_CASE(SWITCH, switch, value);
25550 ++ HANDLE_CASE(XID, xid, value);
25551 ++ HANDLE_CASE(NID, nid, value);
25552 ++ HANDLE_CASE(TAG, tag, value);
25553 ++ HANDLE_CASE(NET, net, value);
25554 ++ HANDLE_CASE(LIMIT, limit, value);
25555 ++ HANDLE_CASE(CRES, cres, value);
25556 ++ HANDLE_CASE(DLIM, dlim, value);
25557 ++ HANDLE_CASE(QUOTA, quota, value);
25558 ++ HANDLE_CASE(CVIRT, cvirt, value);
25559 ++ HANDLE_CASE(SPACE, space, value);
25560 ++ HANDLE_CASE(MISC, misc, value);
25561 ++ default:
25562 ++ return -EINVAL;
25563 ++ break;
25564 ++ }
25565 ++ }
25566 ++ return 1;
25567 ++}
25568 ++
25569 ++__setup("vsdebug=", vs_debug_setup);
25570 ++
25571 ++
25572 ++
25573 ++EXPORT_SYMBOL_GPL(vx_debug_switch);
25574 ++EXPORT_SYMBOL_GPL(vx_debug_xid);
25575 ++EXPORT_SYMBOL_GPL(vx_debug_nid);
25576 ++EXPORT_SYMBOL_GPL(vx_debug_net);
25577 ++EXPORT_SYMBOL_GPL(vx_debug_limit);
25578 ++EXPORT_SYMBOL_GPL(vx_debug_cres);
25579 ++EXPORT_SYMBOL_GPL(vx_debug_dlim);
25580 ++EXPORT_SYMBOL_GPL(vx_debug_quota);
25581 ++EXPORT_SYMBOL_GPL(vx_debug_cvirt);
25582 ++EXPORT_SYMBOL_GPL(vx_debug_space);
25583 ++EXPORT_SYMBOL_GPL(vx_debug_misc);
25584 ++
25585 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/tag.c linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/tag.c
25586 +--- linux-2.6.29.4/kernel/vserver/tag.c 1970-01-01 01:00:00.000000000 +0100
25587 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/tag.c 2009-02-22 22:54:26.000000000 +0100
25588 +@@ -0,0 +1,63 @@
25589 ++/*
25590 ++ * linux/kernel/vserver/tag.c
25591 ++ *
25592 ++ * Virtual Server: Shallow Tag Space
25593 ++ *
25594 ++ * Copyright (C) 2007 Herbert Pötzl
25595 ++ *
25596 ++ * V0.01 basic implementation
25597 ++ *
25598 ++ */
25599 ++
25600 ++#include <linux/sched.h>
25601 ++#include <linux/vserver/debug.h>
25602 ++#include <linux/vs_pid.h>
25603 ++#include <linux/vs_tag.h>
25604 ++
25605 ++#include <linux/vserver/tag_cmd.h>
25606 ++
25607 ++
25608 ++int dx_migrate_task(struct task_struct *p, tag_t tag)
25609 ++{
25610 ++ if (!p)
25611 ++ BUG();
25612 ++
25613 ++ vxdprintk(VXD_CBIT(tag, 5),
25614 ++ "dx_migrate_task(%p[#%d],#%d)", p, p->tag, tag);
25615 ++
25616 ++ task_lock(p);
25617 ++ p->tag = tag;
25618 ++ task_unlock(p);
25619 ++
25620 ++ vxdprintk(VXD_CBIT(tag, 5),
25621 ++ "moved task %p into [#%d]", p, tag);
25622 ++ return 0;
25623 ++}
25624 ++
25625 ++/* vserver syscall commands below here */
25626 ++
25627 ++/* taks xid and vx_info functions */
25628 ++
25629 ++
25630 ++int vc_task_tag(uint32_t id)
25631 ++{
25632 ++ tag_t tag;
25633 ++
25634 ++ if (id) {
25635 ++ struct task_struct *tsk;
25636 ++ read_lock(&tasklist_lock);
25637 ++ tsk = find_task_by_real_pid(id);
25638 ++ tag = (tsk) ? tsk->tag : -ESRCH;
25639 ++ read_unlock(&tasklist_lock);
25640 ++ } else
25641 ++ tag = dx_current_tag();
25642 ++ return tag;
25643 ++}
25644 ++
25645 ++
25646 ++int vc_tag_migrate(uint32_t tag)
25647 ++{
25648 ++ return dx_migrate_task(current, tag & 0xFFFF);
25649 ++}
25650 ++
25651 ++
25652 +diff -NurpP --minimal linux-2.6.29.4/kernel/vserver/vci_config.h linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/vci_config.h
25653 +--- linux-2.6.29.4/kernel/vserver/vci_config.h 1970-01-01 01:00:00.000000000 +0100
25654 ++++ linux-2.6.29.4-vs2.3.0.36.14/kernel/vserver/vci_config.h 2009-02-22 22:54:26.000000000 +0100
25655 +@@ -0,0 +1,81 @@
25656 ++
25657 ++/* interface version */
25658 ++
25659 ++#define VCI_VERSION 0x00020304
25660 ++
25661 ++
25662 ++enum {
25663 ++ VCI_KCBIT_NO_DYNAMIC = 0,
25664 ++
25665 ++ VCI_KCBIT_PROC_SECURE = 4,
25666 ++ VCI_KCBIT_HARDCPU = 5,
25667 ++ VCI_KCBIT_IDLELIMIT = 6,
25668 ++ VCI_KCBIT_IDLETIME = 7,
25669 ++
25670 ++ VCI_KCBIT_COWBL = 8,
25671 ++ VCI_KCBIT_FULLCOWBL = 9,
25672 ++ VCI_KCBIT_SPACES = 10,
25673 ++ VCI_KCBIT_NETV2 = 11,
25674 ++
25675 ++ VCI_KCBIT_DEBUG = 16,
25676 ++ VCI_KCBIT_HISTORY = 20,
25677 ++ VCI_KCBIT_TAGGED = 24,
25678 ++ VCI_KCBIT_PPTAG = 28,
25679 ++
25680 ++ VCI_KCBIT_MORE = 31,
25681 ++};
25682 ++
25683 ++
25684 ++static inline uint32_t vci_kernel_config(void)
25685 ++{
25686 ++ return
25687 ++ (1 << VCI_KCBIT_NO_DYNAMIC) |
25688 ++
25689 ++ /* configured features */
25690 ++#ifdef CONFIG_VSERVER_PROC_SECURE
25691 ++ (1 << VCI_KCBIT_PROC_SECURE) |
25692 ++#endif
25693 ++#ifdef CONFIG_VSERVER_HARDCPU
25694 ++ (1 << VCI_KCBIT_HARDCPU) |
25695 ++#endif
25696 ++#ifdef CONFIG_VSERVER_IDLELIMIT
25697 ++ (1 << VCI_KCBIT_IDLELIMIT) |
25698 ++#endif
25699 ++#ifdef CONFIG_VSERVER_IDLETIME
25700 ++ (1 << VCI_KCBIT_IDLETIME) |
25701 ++#endif
25702 ++#ifdef CONFIG_VSERVER_COWBL
25703 ++ (1 << VCI_KCBIT_COWBL) |
25704 ++ (1 << VCI_KCBIT_FULLCOWBL) |
25705 ++#endif
25706 ++ (1 << VCI_KCBIT_SPACES) |
25707 ++ (1 << VCI_KCBIT_NETV2) |
25708 ++
25709 ++ /* debug options */
25710 ++#ifdef CONFIG_VSERVER_DEBUG
25711 ++ (1 << VCI_KCBIT_DEBUG) |
25712 ++#endif
25713 ++#ifdef CONFIG_VSERVER_HISTORY
25714 ++ (1 << VCI_KCBIT_HISTORY) |
25715 ++#endif
25716 ++
25717 ++ /* inode context tagging */
25718 ++#if defined(CONFIG_TAGGING_NONE)
25719 ++ (0 << VCI_KCBIT_TAGGED) |
25720 ++#elif defined(CONFIG_TAGGING_UID16)
25721 ++ (1 << VCI_KCBIT_TAGGED) |
25722 ++#elif defined(CONFIG_TAGGING_GID16)
25723 ++ (2 << VCI_KCBIT_TAGGED) |
25724 ++#elif defined(CONFIG_TAGGING_ID24)
25725 ++ (3 << VCI_KCBIT_TAGGED) |
25726 ++#elif defined(CONFIG_TAGGING_INTERN)
25727 ++ (4 << VCI_KCBIT_TAGGED) |
25728 ++#elif defined(CONFIG_TAGGING_RUNTIME)
25729 ++ (5 << VCI_KCBIT_TAGGED) |
25730 ++#else
25731 ++ (7 << VCI_KCBIT_TAGGED) |
25732 ++#endif
25733 ++ (1 << VCI_KCBIT_PPTAG) |
25734 ++ 0;
25735 ++}
25736 ++
25737 +diff -NurpP --minimal linux-2.6.29.4/Makefile linux-2.6.29.4-vs2.3.0.36.14/Makefile
25738 +--- linux-2.6.29.4/Makefile 2009-05-23 23:16:50.000000000 +0200
25739 ++++ linux-2.6.29.4-vs2.3.0.36.14/Makefile 2009-05-29 19:03:54.000000000 +0200
25740 +@@ -1,7 +1,7 @@
25741 + VERSION = 2
25742 + PATCHLEVEL = 6
25743 + SUBLEVEL = 29
25744 +-EXTRAVERSION =
25745 ++EXTRAVERSION = -vs2.3.0.36.14-gentoo
25746 + NAME = Temporary Tasmanian Devil
25747 +
25748 + # *DOCUMENTATION*
25749 +diff -NurpP --minimal linux-2.6.29.4/mm/filemap_xip.c linux-2.6.29.4-vs2.3.0.36.14/mm/filemap_xip.c
25750 +--- linux-2.6.29.4/mm/filemap_xip.c 2009-05-23 23:16:53.000000000 +0200
25751 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/filemap_xip.c 2009-04-30 12:14:53.000000000 +0200
25752 +@@ -17,6 +17,7 @@
25753 + #include <linux/sched.h>
25754 + #include <linux/seqlock.h>
25755 + #include <linux/mutex.h>
25756 ++#include <linux/vs_memory.h>
25757 + #include <asm/tlbflush.h>
25758 + #include <asm/io.h>
25759 +
25760 +diff -NurpP --minimal linux-2.6.29.4/mm/fremap.c linux-2.6.29.4-vs2.3.0.36.14/mm/fremap.c
25761 +--- linux-2.6.29.4/mm/fremap.c 2009-03-24 14:22:45.000000000 +0100
25762 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/fremap.c 2009-03-24 14:48:36.000000000 +0100
25763 +@@ -16,6 +16,7 @@
25764 + #include <linux/module.h>
25765 + #include <linux/syscalls.h>
25766 + #include <linux/mmu_notifier.h>
25767 ++#include <linux/vs_memory.h>
25768 +
25769 + #include <asm/mmu_context.h>
25770 + #include <asm/cacheflush.h>
25771 +diff -NurpP --minimal linux-2.6.29.4/mm/hugetlb.c linux-2.6.29.4-vs2.3.0.36.14/mm/hugetlb.c
25772 +--- linux-2.6.29.4/mm/hugetlb.c 2009-03-24 14:22:45.000000000 +0100
25773 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/hugetlb.c 2009-03-24 14:48:36.000000000 +0100
25774 +@@ -24,6 +24,7 @@
25775 + #include <asm/io.h>
25776 +
25777 + #include <linux/hugetlb.h>
25778 ++#include <linux/vs_memory.h>
25779 + #include "internal.h"
25780 +
25781 + const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
25782 +diff -NurpP --minimal linux-2.6.29.4/mm/memory.c linux-2.6.29.4-vs2.3.0.36.14/mm/memory.c
25783 +--- linux-2.6.29.4/mm/memory.c 2009-05-23 23:16:53.000000000 +0200
25784 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/memory.c 2009-05-23 23:19:11.000000000 +0200
25785 +@@ -55,6 +55,7 @@
25786 + #include <linux/kallsyms.h>
25787 + #include <linux/swapops.h>
25788 + #include <linux/elf.h>
25789 ++// #include <linux/vs_memory.h>
25790 +
25791 + #include <asm/pgalloc.h>
25792 + #include <asm/uaccess.h>
25793 +@@ -612,6 +613,9 @@ static int copy_pte_range(struct mm_stru
25794 + int progress = 0;
25795 + int rss[2];
25796 +
25797 ++ if (!vx_rss_avail(dst_mm, ((end - addr)/PAGE_SIZE + 1)))
25798 ++ return -ENOMEM;
25799 ++
25800 + again:
25801 + rss[1] = rss[0] = 0;
25802 + dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
25803 +@@ -2481,6 +2485,11 @@ static int do_swap_page(struct mm_struct
25804 + count_vm_event(PGMAJFAULT);
25805 + }
25806 +
25807 ++ if (!vx_rss_avail(mm, 1)) {
25808 ++ ret = VM_FAULT_OOM;
25809 ++ goto out;
25810 ++ }
25811 ++
25812 + mark_page_accessed(page);
25813 +
25814 + lock_page(page);
25815 +@@ -2572,6 +2581,8 @@ static int do_anonymous_page(struct mm_s
25816 + /* Allocate our own private page. */
25817 + pte_unmap(page_table);
25818 +
25819 ++ if (!vx_rss_avail(mm, 1))
25820 ++ goto oom;
25821 + if (unlikely(anon_vma_prepare(vma)))
25822 + goto oom;
25823 + page = alloc_zeroed_user_highpage_movable(vma, address);
25824 +@@ -2855,6 +2866,7 @@ static inline int handle_pte_fault(struc
25825 + {
25826 + pte_t entry;
25827 + spinlock_t *ptl;
25828 ++ int ret = 0, type = VXPT_UNKNOWN;
25829 +
25830 + entry = *pte;
25831 + if (!pte_present(entry)) {
25832 +@@ -2879,9 +2891,12 @@ static inline int handle_pte_fault(struc
25833 + if (unlikely(!pte_same(*pte, entry)))
25834 + goto unlock;
25835 + if (write_access) {
25836 +- if (!pte_write(entry))
25837 +- return do_wp_page(mm, vma, address,
25838 ++ if (!pte_write(entry)) {
25839 ++ ret = do_wp_page(mm, vma, address,
25840 + pte, pmd, ptl, entry);
25841 ++ type = VXPT_WRITE;
25842 ++ goto out;
25843 ++ }
25844 + entry = pte_mkdirty(entry);
25845 + }
25846 + entry = pte_mkyoung(entry);
25847 +@@ -2899,7 +2914,10 @@ static inline int handle_pte_fault(struc
25848 + }
25849 + unlock:
25850 + pte_unmap_unlock(pte, ptl);
25851 +- return 0;
25852 ++ ret = 0;
25853 ++out:
25854 ++ vx_page_fault(mm, vma, type, ret);
25855 ++ return ret;
25856 + }
25857 +
25858 + /*
25859 +diff -NurpP --minimal linux-2.6.29.4/mm/mlock.c linux-2.6.29.4-vs2.3.0.36.14/mm/mlock.c
25860 +--- linux-2.6.29.4/mm/mlock.c 2009-03-24 14:22:45.000000000 +0100
25861 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/mlock.c 2009-03-28 05:08:26.000000000 +0100
25862 +@@ -18,6 +18,7 @@
25863 + #include <linux/rmap.h>
25864 + #include <linux/mmzone.h>
25865 + #include <linux/hugetlb.h>
25866 ++#include <linux/vs_memory.h>
25867 +
25868 + #include "internal.h"
25869 +
25870 +@@ -415,7 +416,7 @@ success:
25871 + nr_pages = (end - start) >> PAGE_SHIFT;
25872 + if (!lock)
25873 + nr_pages = -nr_pages;
25874 +- mm->locked_vm += nr_pages;
25875 ++ vx_vmlocked_add(mm, nr_pages);
25876 +
25877 + /*
25878 + * vm_flags is protected by the mmap_sem held in write mode.
25879 +@@ -492,7 +493,7 @@ static int do_mlock(unsigned long start,
25880 +
25881 + SYSCALL_DEFINE2(mlock, unsigned long, start, size_t, len)
25882 + {
25883 +- unsigned long locked;
25884 ++ unsigned long locked, grow;
25885 + unsigned long lock_limit;
25886 + int error = -ENOMEM;
25887 +
25888 +@@ -505,8 +506,10 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
25889 + len = PAGE_ALIGN(len + (start & ~PAGE_MASK));
25890 + start &= PAGE_MASK;
25891 +
25892 +- locked = len >> PAGE_SHIFT;
25893 +- locked += current->mm->locked_vm;
25894 ++ grow = len >> PAGE_SHIFT;
25895 ++ if (!vx_vmlocked_avail(current->mm, grow))
25896 ++ goto out;
25897 ++ locked = current->mm->locked_vm + grow;
25898 +
25899 + lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
25900 + lock_limit >>= PAGE_SHIFT;
25901 +@@ -514,6 +517,7 @@ SYSCALL_DEFINE2(mlock, unsigned long, st
25902 + /* check against resource limits */
25903 + if ((locked <= lock_limit) || capable(CAP_IPC_LOCK))
25904 + error = do_mlock(start, len, 1);
25905 ++out:
25906 + up_write(&current->mm->mmap_sem);
25907 + return error;
25908 + }
25909 +@@ -575,6 +579,8 @@ SYSCALL_DEFINE1(mlockall, int, flags)
25910 + lock_limit >>= PAGE_SHIFT;
25911 +
25912 + ret = -ENOMEM;
25913 ++ if (!vx_vmlocked_avail(current->mm, current->mm->total_vm))
25914 ++ goto out;
25915 + if (!(flags & MCL_CURRENT) || (current->mm->total_vm <= lock_limit) ||
25916 + capable(CAP_IPC_LOCK))
25917 + ret = do_mlockall(flags);
25918 +@@ -652,8 +658,10 @@ void *alloc_locked_buffer(size_t size)
25919 + if (!buffer)
25920 + goto out;
25921 +
25922 +- current->mm->total_vm += pgsz;
25923 +- current->mm->locked_vm += pgsz;
25924 ++ // current->mm->total_vm += pgsz;
25925 ++ vx_vmpages_add(current->mm, pgsz);
25926 ++ // current->mm->locked_vm += pgsz;
25927 ++ vx_vmlocked_add(current->mm, pgsz);
25928 +
25929 + out:
25930 + up_write(&current->mm->mmap_sem);
25931 +@@ -666,8 +674,10 @@ void release_locked_buffer(void *buffer,
25932 +
25933 + down_write(&current->mm->mmap_sem);
25934 +
25935 +- current->mm->total_vm -= pgsz;
25936 +- current->mm->locked_vm -= pgsz;
25937 ++ // current->mm->total_vm -= pgsz;
25938 ++ vx_vmpages_sub(current->mm, pgsz);
25939 ++ // current->mm->locked_vm -= pgsz;
25940 ++ vx_vmlocked_sub(current->mm, pgsz);
25941 +
25942 + up_write(&current->mm->mmap_sem);
25943 + }
25944 +diff -NurpP --minimal linux-2.6.29.4/mm/mmap.c linux-2.6.29.4-vs2.3.0.36.14/mm/mmap.c
25945 +--- linux-2.6.29.4/mm/mmap.c 2009-05-23 23:16:53.000000000 +0200
25946 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/mmap.c 2009-05-10 23:42:01.000000000 +0200
25947 +@@ -1215,7 +1215,8 @@ munmap_back:
25948 + if (correct_wcount)
25949 + atomic_inc(&inode->i_writecount);
25950 + out:
25951 +- mm->total_vm += len >> PAGE_SHIFT;
25952 ++ // mm->total_vm += len >> PAGE_SHIFT;
25953 ++ vx_vmpages_add(mm, len >> PAGE_SHIFT);
25954 + vm_stat_account(mm, vm_flags, file, len >> PAGE_SHIFT);
25955 + if (vm_flags & VM_LOCKED) {
25956 + /*
25957 +@@ -1224,7 +1225,8 @@ out:
25958 + long nr_pages = mlock_vma_pages_range(vma, addr, addr + len);
25959 + if (nr_pages < 0)
25960 + return nr_pages; /* vma gone! */
25961 +- mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
25962 ++ // mm->locked_vm += (len >> PAGE_SHIFT) - nr_pages;
25963 ++ vx_vmlocked_add(mm, (len >> PAGE_SHIFT) - nr_pages);
25964 + } else if ((flags & MAP_POPULATE) && !(flags & MAP_NONBLOCK))
25965 + make_pages_present(addr, addr + len);
25966 + return addr;
25967 +@@ -1571,9 +1573,9 @@ static int acct_stack_growth(struct vm_a
25968 + return -ENOMEM;
25969 +
25970 + /* Ok, everything looks good - let it rip */
25971 +- mm->total_vm += grow;
25972 ++ vx_vmpages_add(mm, grow);
25973 + if (vma->vm_flags & VM_LOCKED)
25974 +- mm->locked_vm += grow;
25975 ++ vx_vmlocked_add(mm, grow);
25976 + vm_stat_account(mm, vma->vm_flags, vma->vm_file, grow);
25977 + return 0;
25978 + }
25979 +@@ -1748,7 +1750,8 @@ static void remove_vma_list(struct mm_st
25980 + do {
25981 + long nrpages = vma_pages(vma);
25982 +
25983 +- mm->total_vm -= nrpages;
25984 ++ // mm->total_vm -= nrpages;
25985 ++ vx_vmpages_sub(mm, nrpages);
25986 + vm_stat_account(mm, vma->vm_flags, vma->vm_file, -nrpages);
25987 + vma = remove_vma(vma);
25988 + } while (vma);
25989 +@@ -1920,7 +1923,8 @@ int do_munmap(struct mm_struct *mm, unsi
25990 + struct vm_area_struct *tmp = vma;
25991 + while (tmp && tmp->vm_start < end) {
25992 + if (tmp->vm_flags & VM_LOCKED) {
25993 +- mm->locked_vm -= vma_pages(tmp);
25994 ++ // mm->locked_vm -= vma_pages(tmp);
25995 ++ vx_vmlocked_sub(mm, vma_pages(tmp));
25996 + munlock_vma_pages_all(tmp);
25997 + }
25998 + tmp = tmp->vm_next;
25999 +@@ -2009,6 +2013,8 @@ unsigned long do_brk(unsigned long addr,
26000 + lock_limit >>= PAGE_SHIFT;
26001 + if (locked > lock_limit && !capable(CAP_IPC_LOCK))
26002 + return -EAGAIN;
26003 ++ if (!vx_vmlocked_avail(mm, len >> PAGE_SHIFT))
26004 ++ return -ENOMEM;
26005 + }
26006 +
26007 + /*
26008 +@@ -2035,7 +2041,8 @@ unsigned long do_brk(unsigned long addr,
26009 + if (mm->map_count > sysctl_max_map_count)
26010 + return -ENOMEM;
26011 +
26012 +- if (security_vm_enough_memory(len >> PAGE_SHIFT))
26013 ++ if (security_vm_enough_memory(len >> PAGE_SHIFT) ||
26014 ++ !vx_vmpages_avail(mm, len >> PAGE_SHIFT))
26015 + return -ENOMEM;
26016 +
26017 + /* Can we just expand an old private anonymous mapping? */
26018 +@@ -2061,10 +2068,13 @@ unsigned long do_brk(unsigned long addr,
26019 + vma->vm_page_prot = vm_get_page_prot(flags);
26020 + vma_link(mm, vma, prev, rb_link, rb_parent);
26021 + out:
26022 +- mm->total_vm += len >> PAGE_SHIFT;
26023 ++ // mm->total_vm += len >> PAGE_SHIFT;
26024 ++ vx_vmpages_add(mm, len >> PAGE_SHIFT);
26025 ++
26026 + if (flags & VM_LOCKED) {
26027 + if (!mlock_vma_pages_range(vma, addr, addr + len))
26028 +- mm->locked_vm += (len >> PAGE_SHIFT);
26029 ++ // mm->locked_vm += (len >> PAGE_SHIFT);
26030 ++ vx_vmlocked_add(mm, len >> PAGE_SHIFT);
26031 + }
26032 + return addr;
26033 + }
26034 +@@ -2107,6 +2117,11 @@ void exit_mmap(struct mm_struct *mm)
26035 + free_pgtables(tlb, vma, FIRST_USER_ADDRESS, 0);
26036 + tlb_finish_mmu(tlb, 0, end);
26037 +
26038 ++ set_mm_counter(mm, file_rss, 0);
26039 ++ set_mm_counter(mm, anon_rss, 0);
26040 ++ vx_vmpages_sub(mm, mm->total_vm);
26041 ++ vx_vmlocked_sub(mm, mm->locked_vm);
26042 ++
26043 + /*
26044 + * Walk the list again, actually closing and freeing it,
26045 + * with preemption enabled, without holding any MM locks.
26046 +@@ -2146,7 +2161,8 @@ int insert_vm_struct(struct mm_struct *
26047 + if (__vma && __vma->vm_start < vma->vm_end)
26048 + return -ENOMEM;
26049 + if ((vma->vm_flags & VM_ACCOUNT) &&
26050 +- security_vm_enough_memory_mm(mm, vma_pages(vma)))
26051 ++ (security_vm_enough_memory_mm(mm, vma_pages(vma)) ||
26052 ++ !vx_vmpages_avail(mm, vma_pages(vma))))
26053 + return -ENOMEM;
26054 + vma_link(mm, vma, prev, rb_link, rb_parent);
26055 + return 0;
26056 +@@ -2222,6 +2238,8 @@ int may_expand_vm(struct mm_struct *mm,
26057 +
26058 + if (cur + npages > lim)
26059 + return 0;
26060 ++ if (!vx_vmpages_avail(mm, npages))
26061 ++ return 0;
26062 + return 1;
26063 + }
26064 +
26065 +@@ -2299,8 +2317,7 @@ int install_special_mapping(struct mm_st
26066 + return -ENOMEM;
26067 + }
26068 +
26069 +- mm->total_vm += len >> PAGE_SHIFT;
26070 +-
26071 ++ vx_vmpages_add(mm, len >> PAGE_SHIFT);
26072 + return 0;
26073 + }
26074 +
26075 +diff -NurpP --minimal linux-2.6.29.4/mm/mremap.c linux-2.6.29.4-vs2.3.0.36.14/mm/mremap.c
26076 +--- linux-2.6.29.4/mm/mremap.c 2009-03-24 14:22:45.000000000 +0100
26077 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/mremap.c 2009-03-24 14:48:36.000000000 +0100
26078 +@@ -19,6 +19,7 @@
26079 + #include <linux/security.h>
26080 + #include <linux/syscalls.h>
26081 + #include <linux/mmu_notifier.h>
26082 ++#include <linux/vs_memory.h>
26083 +
26084 + #include <asm/uaccess.h>
26085 + #include <asm/cacheflush.h>
26086 +@@ -220,7 +221,7 @@ static unsigned long move_vma(struct vm_
26087 + * If this were a serious issue, we'd add a flag to do_munmap().
26088 + */
26089 + hiwater_vm = mm->hiwater_vm;
26090 +- mm->total_vm += new_len >> PAGE_SHIFT;
26091 ++ vx_vmpages_add(mm, new_len >> PAGE_SHIFT);
26092 + vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
26093 +
26094 + if (do_munmap(mm, old_addr, old_len) < 0) {
26095 +@@ -238,7 +239,7 @@ static unsigned long move_vma(struct vm_
26096 + }
26097 +
26098 + if (vm_flags & VM_LOCKED) {
26099 +- mm->locked_vm += new_len >> PAGE_SHIFT;
26100 ++ vx_vmlocked_add(mm, new_len >> PAGE_SHIFT);
26101 + if (new_len > old_len)
26102 + mlock_vma_pages_range(new_vma, new_addr + old_len,
26103 + new_addr + new_len);
26104 +@@ -349,6 +350,9 @@ unsigned long do_mremap(unsigned long ad
26105 + ret = -EAGAIN;
26106 + if (locked > lock_limit && !capable(CAP_IPC_LOCK))
26107 + goto out;
26108 ++ if (!vx_vmlocked_avail(current->mm,
26109 ++ (new_len - old_len) >> PAGE_SHIFT))
26110 ++ goto out;
26111 + }
26112 + if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
26113 + ret = -ENOMEM;
26114 +@@ -377,10 +381,12 @@ unsigned long do_mremap(unsigned long ad
26115 + vma_adjust(vma, vma->vm_start,
26116 + addr + new_len, vma->vm_pgoff, NULL);
26117 +
26118 +- mm->total_vm += pages;
26119 ++ // mm->total_vm += pages;
26120 ++ vx_vmpages_add(mm, pages);
26121 + vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
26122 + if (vma->vm_flags & VM_LOCKED) {
26123 +- mm->locked_vm += pages;
26124 ++ // mm->locked_vm += pages;
26125 ++ vx_vmlocked_add(mm, pages);
26126 + mlock_vma_pages_range(vma, addr + old_len,
26127 + addr + new_len);
26128 + }
26129 +diff -NurpP --minimal linux-2.6.29.4/mm/nommu.c linux-2.6.29.4-vs2.3.0.36.14/mm/nommu.c
26130 +--- linux-2.6.29.4/mm/nommu.c 2009-05-23 23:16:53.000000000 +0200
26131 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/nommu.c 2009-05-10 23:42:01.000000000 +0200
26132 +@@ -1352,7 +1352,7 @@ unsigned long do_mmap_pgoff(struct file
26133 + /* okay... we have a mapping; now we have to register it */
26134 + result = vma->vm_start;
26135 +
26136 +- current->mm->total_vm += len >> PAGE_SHIFT;
26137 ++ vx_vmpages_add(current->mm, len >> PAGE_SHIFT);
26138 +
26139 + share:
26140 + add_vma_to_mm(current->mm, vma);
26141 +@@ -1618,7 +1618,7 @@ void exit_mmap(struct mm_struct *mm)
26142 +
26143 + kenter("");
26144 +
26145 +- mm->total_vm = 0;
26146 ++ vx_vmpages_sub(mm, mm->total_vm);
26147 +
26148 + while ((vma = mm->mmap)) {
26149 + mm->mmap = vma->vm_next;
26150 +diff -NurpP --minimal linux-2.6.29.4/mm/oom_kill.c linux-2.6.29.4-vs2.3.0.36.14/mm/oom_kill.c
26151 +--- linux-2.6.29.4/mm/oom_kill.c 2009-03-24 14:22:45.000000000 +0100
26152 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/oom_kill.c 2009-03-24 14:48:36.000000000 +0100
26153 +@@ -27,6 +27,7 @@
26154 + #include <linux/notifier.h>
26155 + #include <linux/memcontrol.h>
26156 + #include <linux/security.h>
26157 ++#include <linux/vs_memory.h>
26158 +
26159 + int sysctl_panic_on_oom;
26160 + int sysctl_oom_kill_allocating_task;
26161 +@@ -72,6 +73,12 @@ unsigned long badness(struct task_struct
26162 + points = mm->total_vm;
26163 +
26164 + /*
26165 ++ * add points for context badness
26166 ++ */
26167 ++
26168 ++ points += vx_badness(p, mm);
26169 ++
26170 ++ /*
26171 + * After this unlock we can no longer dereference local variable `mm'
26172 + */
26173 + task_unlock(p);
26174 +@@ -162,8 +169,8 @@ unsigned long badness(struct task_struct
26175 + }
26176 +
26177 + #ifdef DEBUG
26178 +- printk(KERN_DEBUG "OOMkill: task %d (%s) got %lu points\n",
26179 +- p->pid, p->comm, points);
26180 ++ printk(KERN_DEBUG "OOMkill: task %d:#%u (%s) got %d points\n",
26181 ++ task_pid_nr(p), p->xid, p->comm, points);
26182 + #endif
26183 + return points;
26184 + }
26185 +@@ -326,8 +333,8 @@ static void __oom_kill_task(struct task_
26186 + }
26187 +
26188 + if (verbose)
26189 +- printk(KERN_ERR "Killed process %d (%s)\n",
26190 +- task_pid_nr(p), p->comm);
26191 ++ printk(KERN_ERR "Killed process %d:#%u (%s)\n",
26192 ++ task_pid_nr(p), p->xid, p->comm);
26193 +
26194 + /*
26195 + * We give our sacrificial lamb high priority and access to
26196 +@@ -410,8 +417,8 @@ static int oom_kill_process(struct task_
26197 + return 0;
26198 + }
26199 +
26200 +- printk(KERN_ERR "%s: kill process %d (%s) score %li or a child\n",
26201 +- message, task_pid_nr(p), p->comm, points);
26202 ++ printk(KERN_ERR "%s: kill process %d:#%u (%s) score %li or a child\n",
26203 ++ message, task_pid_nr(p), p->xid, p->comm, points);
26204 +
26205 + /* Try to kill a child first */
26206 + list_for_each_entry(c, &p->children, sibling) {
26207 +diff -NurpP --minimal linux-2.6.29.4/mm/page_alloc.c linux-2.6.29.4-vs2.3.0.36.14/mm/page_alloc.c
26208 +--- linux-2.6.29.4/mm/page_alloc.c 2009-03-24 14:22:45.000000000 +0100
26209 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/page_alloc.c 2009-03-24 14:48:36.000000000 +0100
26210 +@@ -46,6 +46,8 @@
26211 + #include <linux/page-isolation.h>
26212 + #include <linux/page_cgroup.h>
26213 + #include <linux/debugobjects.h>
26214 ++#include <linux/vs_base.h>
26215 ++#include <linux/vs_limit.h>
26216 +
26217 + #include <asm/tlbflush.h>
26218 + #include <asm/div64.h>
26219 +@@ -1839,6 +1841,9 @@ void si_meminfo(struct sysinfo *val)
26220 + val->totalhigh = totalhigh_pages;
26221 + val->freehigh = nr_free_highpages();
26222 + val->mem_unit = PAGE_SIZE;
26223 ++
26224 ++ if (vx_flags(VXF_VIRT_MEM, 0))
26225 ++ vx_vsi_meminfo(val);
26226 + }
26227 +
26228 + EXPORT_SYMBOL(si_meminfo);
26229 +@@ -1859,6 +1864,9 @@ void si_meminfo_node(struct sysinfo *val
26230 + val->freehigh = 0;
26231 + #endif
26232 + val->mem_unit = PAGE_SIZE;
26233 ++
26234 ++ if (vx_flags(VXF_VIRT_MEM, 0))
26235 ++ vx_vsi_meminfo(val);
26236 + }
26237 + #endif
26238 +
26239 +diff -NurpP --minimal linux-2.6.29.4/mm/rmap.c linux-2.6.29.4-vs2.3.0.36.14/mm/rmap.c
26240 +--- linux-2.6.29.4/mm/rmap.c 2009-03-24 14:22:45.000000000 +0100
26241 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/rmap.c 2009-03-24 18:26:27.000000000 +0100
26242 +@@ -50,6 +50,7 @@
26243 + #include <linux/memcontrol.h>
26244 + #include <linux/mmu_notifier.h>
26245 + #include <linux/migrate.h>
26246 ++#include <linux/vs_memory.h>
26247 +
26248 + #include <asm/tlbflush.h>
26249 +
26250 +diff -NurpP --minimal linux-2.6.29.4/mm/shmem.c linux-2.6.29.4-vs2.3.0.36.14/mm/shmem.c
26251 +--- linux-2.6.29.4/mm/shmem.c 2009-03-24 14:22:45.000000000 +0100
26252 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/shmem.c 2009-03-24 14:48:36.000000000 +0100
26253 +@@ -1757,7 +1757,7 @@ static int shmem_statfs(struct dentry *d
26254 + {
26255 + struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
26256 +
26257 +- buf->f_type = TMPFS_MAGIC;
26258 ++ buf->f_type = TMPFS_SUPER_MAGIC;
26259 + buf->f_bsize = PAGE_CACHE_SIZE;
26260 + buf->f_namelen = NAME_MAX;
26261 + spin_lock(&sbinfo->stat_lock);
26262 +@@ -2326,7 +2326,7 @@ static int shmem_fill_super(struct super
26263 + sb->s_maxbytes = SHMEM_MAX_BYTES;
26264 + sb->s_blocksize = PAGE_CACHE_SIZE;
26265 + sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
26266 +- sb->s_magic = TMPFS_MAGIC;
26267 ++ sb->s_magic = TMPFS_SUPER_MAGIC;
26268 + sb->s_op = &shmem_ops;
26269 + sb->s_time_gran = 1;
26270 + #ifdef CONFIG_TMPFS_POSIX_ACL
26271 +diff -NurpP --minimal linux-2.6.29.4/mm/slab.c linux-2.6.29.4-vs2.3.0.36.14/mm/slab.c
26272 +--- linux-2.6.29.4/mm/slab.c 2009-03-24 14:22:45.000000000 +0100
26273 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/slab.c 2009-03-24 14:48:36.000000000 +0100
26274 +@@ -509,6 +509,8 @@ struct kmem_cache {
26275 + #define STATS_INC_FREEMISS(x) do { } while (0)
26276 + #endif
26277 +
26278 ++#include "slab_vs.h"
26279 ++
26280 + #if DEBUG
26281 +
26282 + /*
26283 +@@ -3275,6 +3277,7 @@ retry:
26284 +
26285 + obj = slab_get_obj(cachep, slabp, nodeid);
26286 + check_slabp(cachep, slabp);
26287 ++ vx_slab_alloc(cachep, flags);
26288 + l3->free_objects--;
26289 + /* move slabp to correct slabp list: */
26290 + list_del(&slabp->list);
26291 +@@ -3347,6 +3350,7 @@ __cache_alloc_node(struct kmem_cache *ca
26292 + /* ___cache_alloc_node can fall back to other nodes */
26293 + ptr = ____cache_alloc_node(cachep, flags, nodeid);
26294 + out:
26295 ++ vx_slab_alloc(cachep, flags);
26296 + local_irq_restore(save_flags);
26297 + ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller);
26298 +
26299 +@@ -3518,6 +3522,7 @@ static inline void __cache_free(struct k
26300 +
26301 + check_irq_off();
26302 + objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0));
26303 ++ vx_slab_free(cachep);
26304 +
26305 + /*
26306 + * Skip calling cache_free_alien() when the platform is not numa.
26307 +diff -NurpP --minimal linux-2.6.29.4/mm/slab_vs.h linux-2.6.29.4-vs2.3.0.36.14/mm/slab_vs.h
26308 +--- linux-2.6.29.4/mm/slab_vs.h 1970-01-01 01:00:00.000000000 +0100
26309 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/slab_vs.h 2009-02-22 22:54:26.000000000 +0100
26310 +@@ -0,0 +1,27 @@
26311 ++
26312 ++#include <linux/vserver/context.h>
26313 ++
26314 ++#include <linux/vs_context.h>
26315 ++
26316 ++static inline
26317 ++void vx_slab_alloc(struct kmem_cache *cachep, gfp_t flags)
26318 ++{
26319 ++ int what = gfp_zone(cachep->gfpflags);
26320 ++
26321 ++ if (!current->vx_info)
26322 ++ return;
26323 ++
26324 ++ atomic_add(cachep->buffer_size, &current->vx_info->cacct.slab[what]);
26325 ++}
26326 ++
26327 ++static inline
26328 ++void vx_slab_free(struct kmem_cache *cachep)
26329 ++{
26330 ++ int what = gfp_zone(cachep->gfpflags);
26331 ++
26332 ++ if (!current->vx_info)
26333 ++ return;
26334 ++
26335 ++ atomic_sub(cachep->buffer_size, &current->vx_info->cacct.slab[what]);
26336 ++}
26337 ++
26338 +diff -NurpP --minimal linux-2.6.29.4/mm/swapfile.c linux-2.6.29.4-vs2.3.0.36.14/mm/swapfile.c
26339 +--- linux-2.6.29.4/mm/swapfile.c 2009-03-24 14:22:45.000000000 +0100
26340 ++++ linux-2.6.29.4-vs2.3.0.36.14/mm/swapfile.c 2009-03-24 18:27:15.000000000 +0100
26341 +@@ -34,6 +34,8 @@
26342 + #include <asm/tlbflush.h>
26343 + #include <linux/swapops.h>
26344 + #include <linux/page_cgroup.h>
26345 ++#include <linux/vs_base.h>
26346 ++#include <linux/vs_memory.h>
26347 +
26348 + static DEFINE_SPINLOCK(swap_lock);
26349 + static unsigned int nr_swapfiles;
26350 +@@ -1935,6 +1937,8 @@ void si_swapinfo(struct sysinfo *val)
26351 + val->freeswap = nr_swap_pages + nr_to_be_unused;
26352 + val->totalswap = total_swap_pages + nr_to_be_unused;
26353 + spin_unlock(&swap_lock);
26354 ++ if (vx_flags(VXF_VIRT_MEM, 0))
26355 ++ vx_vsi_swapinfo(val);
26356 + }
26357 +
26358 + /*
26359 +diff -NurpP --minimal linux-2.6.29.4/net/core/dev.c linux-2.6.29.4-vs2.3.0.36.14/net/core/dev.c
26360 +--- linux-2.6.29.4/net/core/dev.c 2009-05-23 23:16:53.000000000 +0200
26361 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/core/dev.c 2009-04-15 22:58:30.000000000 +0200
26362 +@@ -126,6 +126,7 @@
26363 + #include <linux/in.h>
26364 + #include <linux/jhash.h>
26365 + #include <linux/random.h>
26366 ++#include <linux/vs_inet.h>
26367 +
26368 + #include "net-sysfs.h"
26369 +
26370 +@@ -2853,6 +2854,8 @@ static int dev_ifconf(struct net *net, c
26371 +
26372 + total = 0;
26373 + for_each_netdev(net, dev) {
26374 ++ if (!nx_dev_visible(current->nx_info, dev))
26375 ++ continue;
26376 + for (i = 0; i < NPROTO; i++) {
26377 + if (gifconf_list[i]) {
26378 + int done;
26379 +@@ -2921,6 +2924,9 @@ static void dev_seq_printf_stats(struct
26380 + {
26381 + const struct net_device_stats *stats = dev_get_stats(dev);
26382 +
26383 ++ if (!nx_dev_visible(current->nx_info, dev))
26384 ++ return;
26385 ++
26386 + seq_printf(seq, "%6s:%8lu %7lu %4lu %4lu %4lu %5lu %10lu %9lu "
26387 + "%8lu %7lu %4lu %4lu %4lu %5lu %7lu %10lu\n",
26388 + dev->name, stats->rx_bytes, stats->rx_packets,
26389 +@@ -4862,6 +4868,15 @@ int dev_change_net_namespace(struct net_
26390 + goto out;
26391 + #endif
26392 +
26393 ++#ifdef CONFIG_SYSFS
26394 ++ /* Don't allow real devices to be moved when sysfs
26395 ++ * is enabled.
26396 ++ */
26397 ++ err = -EINVAL;
26398 ++ if (dev->dev.parent)
26399 ++ goto out;
26400 ++#endif
26401 ++
26402 + /* Ensure the device has been registrered */
26403 + err = -EINVAL;
26404 + if (dev->reg_state != NETREG_REGISTERED)
26405 +@@ -4921,6 +4936,8 @@ int dev_change_net_namespace(struct net_
26406 +
26407 + netdev_unregister_kobject(dev);
26408 +
26409 ++ netdev_unregister_kobject(dev);
26410 ++
26411 + /* Actually switch the network namespace */
26412 + dev_net_set(dev, net);
26413 +
26414 +diff -NurpP --minimal linux-2.6.29.4/net/core/net-sysfs.c linux-2.6.29.4-vs2.3.0.36.14/net/core/net-sysfs.c
26415 +--- linux-2.6.29.4/net/core/net-sysfs.c 2009-03-24 14:22:46.000000000 +0100
26416 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/core/net-sysfs.c 2009-03-24 18:30:01.000000000 +0100
26417 +@@ -512,6 +512,9 @@ int netdev_register_kobject(struct net_d
26418 + if (dev_net(net) != &init_net)
26419 + return 0;
26420 +
26421 ++ if (dev_net(net) != &init_net)
26422 ++ return 0;
26423 ++
26424 + return device_add(dev);
26425 + }
26426 +
26427 +diff -NurpP --minimal linux-2.6.29.4/net/core/rtnetlink.c linux-2.6.29.4-vs2.3.0.36.14/net/core/rtnetlink.c
26428 +--- linux-2.6.29.4/net/core/rtnetlink.c 2009-03-24 14:22:46.000000000 +0100
26429 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/core/rtnetlink.c 2009-03-24 14:48:36.000000000 +0100
26430 +@@ -690,6 +690,8 @@ static int rtnl_dump_ifinfo(struct sk_bu
26431 +
26432 + idx = 0;
26433 + for_each_netdev(net, dev) {
26434 ++ if (!nx_dev_visible(skb->sk->sk_nx_info, dev))
26435 ++ continue;
26436 + if (idx < s_idx)
26437 + goto cont;
26438 + if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK,
26439 +@@ -1235,6 +1237,9 @@ void rtmsg_ifinfo(int type, struct net_d
26440 + struct sk_buff *skb;
26441 + int err = -ENOBUFS;
26442 +
26443 ++ if (!nx_dev_visible(current->nx_info, dev))
26444 ++ return;
26445 ++
26446 + skb = nlmsg_new(if_nlmsg_size(dev), GFP_KERNEL);
26447 + if (skb == NULL)
26448 + goto errout;
26449 +diff -NurpP --minimal linux-2.6.29.4/net/core/sock.c linux-2.6.29.4-vs2.3.0.36.14/net/core/sock.c
26450 +--- linux-2.6.29.4/net/core/sock.c 2009-03-24 14:22:46.000000000 +0100
26451 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/core/sock.c 2009-03-24 14:48:36.000000000 +0100
26452 +@@ -124,6 +124,10 @@
26453 + #include <linux/ipsec.h>
26454 +
26455 + #include <linux/filter.h>
26456 ++#include <linux/vs_socket.h>
26457 ++#include <linux/vs_limit.h>
26458 ++#include <linux/vs_context.h>
26459 ++#include <linux/vs_network.h>
26460 +
26461 + #ifdef CONFIG_INET
26462 + #include <net/tcp.h>
26463 +@@ -900,6 +904,8 @@ static struct sock *sk_prot_alloc(struct
26464 + if (!try_module_get(prot->owner))
26465 + goto out_free_sec;
26466 + }
26467 ++ sock_vx_init(sk);
26468 ++ sock_nx_init(sk);
26469 +
26470 + return sk;
26471 +
26472 +@@ -976,6 +982,11 @@ void sk_free(struct sock *sk)
26473 + __func__, atomic_read(&sk->sk_omem_alloc));
26474 +
26475 + put_net(sock_net(sk));
26476 ++ vx_sock_dec(sk);
26477 ++ clr_vx_info(&sk->sk_vx_info);
26478 ++ sk->sk_xid = -1;
26479 ++ clr_nx_info(&sk->sk_nx_info);
26480 ++ sk->sk_nid = -1;
26481 + sk_prot_free(sk->sk_prot_creator, sk);
26482 + }
26483 +
26484 +@@ -1011,6 +1022,8 @@ struct sock *sk_clone(const struct sock
26485 +
26486 + /* SANITY */
26487 + get_net(sock_net(newsk));
26488 ++ sock_vx_init(newsk);
26489 ++ sock_nx_init(newsk);
26490 + sk_node_init(&newsk->sk_node);
26491 + sock_lock_init(newsk);
26492 + bh_lock_sock(newsk);
26493 +@@ -1057,6 +1070,12 @@ struct sock *sk_clone(const struct sock
26494 + newsk->sk_priority = 0;
26495 + atomic_set(&newsk->sk_refcnt, 2);
26496 +
26497 ++ set_vx_info(&newsk->sk_vx_info, sk->sk_vx_info);
26498 ++ newsk->sk_xid = sk->sk_xid;
26499 ++ vx_sock_inc(newsk);
26500 ++ set_nx_info(&newsk->sk_nx_info, sk->sk_nx_info);
26501 ++ newsk->sk_nid = sk->sk_nid;
26502 ++
26503 + /*
26504 + * Increment the counter in the same struct proto as the master
26505 + * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
26506 +@@ -1744,6 +1763,11 @@ void sock_init_data(struct socket *sock,
26507 +
26508 + sk->sk_stamp = ktime_set(-1L, 0);
26509 +
26510 ++ set_vx_info(&sk->sk_vx_info, current->vx_info);
26511 ++ sk->sk_xid = vx_current_xid();
26512 ++ vx_sock_inc(sk);
26513 ++ set_nx_info(&sk->sk_nx_info, current->nx_info);
26514 ++ sk->sk_nid = nx_current_nid();
26515 + atomic_set(&sk->sk_refcnt, 1);
26516 + atomic_set(&sk->sk_drops, 0);
26517 + }
26518 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/af_inet.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/af_inet.c
26519 +--- linux-2.6.29.4/net/ipv4/af_inet.c 2009-03-24 14:22:46.000000000 +0100
26520 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/af_inet.c 2009-03-24 14:48:37.000000000 +0100
26521 +@@ -115,6 +115,7 @@
26522 + #ifdef CONFIG_IP_MROUTE
26523 + #include <linux/mroute.h>
26524 + #endif
26525 ++#include <linux/vs_limit.h>
26526 +
26527 + extern void ip_mc_drop_socket(struct sock *sk);
26528 +
26529 +@@ -325,9 +326,12 @@ lookup_protocol:
26530 + }
26531 +
26532 + err = -EPERM;
26533 ++ if ((protocol == IPPROTO_ICMP) &&
26534 ++ nx_capable(answer->capability, NXC_RAW_ICMP))
26535 ++ goto override;
26536 + if (answer->capability > 0 && !capable(answer->capability))
26537 + goto out_rcu_unlock;
26538 +-
26539 ++override:
26540 + err = -EAFNOSUPPORT;
26541 + if (!inet_netns_ok(net, protocol))
26542 + goto out_rcu_unlock;
26543 +@@ -445,6 +449,7 @@ int inet_bind(struct socket *sock, struc
26544 + struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
26545 + struct sock *sk = sock->sk;
26546 + struct inet_sock *inet = inet_sk(sk);
26547 ++ struct nx_v4_sock_addr nsa;
26548 + unsigned short snum;
26549 + int chk_addr_ret;
26550 + int err;
26551 +@@ -458,7 +463,11 @@ int inet_bind(struct socket *sock, struc
26552 + if (addr_len < sizeof(struct sockaddr_in))
26553 + goto out;
26554 +
26555 +- chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
26556 ++ err = v4_map_sock_addr(inet, addr, &nsa);
26557 ++ if (err)
26558 ++ goto out;
26559 ++
26560 ++ chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
26561 +
26562 + /* Not specified by any standard per-se, however it breaks too
26563 + * many applications when removed. It is unfortunate since
26564 +@@ -470,7 +479,7 @@ int inet_bind(struct socket *sock, struc
26565 + err = -EADDRNOTAVAIL;
26566 + if (!sysctl_ip_nonlocal_bind &&
26567 + !(inet->freebind || inet->transparent) &&
26568 +- addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
26569 ++ nsa.saddr != htonl(INADDR_ANY) &&
26570 + chk_addr_ret != RTN_LOCAL &&
26571 + chk_addr_ret != RTN_MULTICAST &&
26572 + chk_addr_ret != RTN_BROADCAST)
26573 +@@ -495,7 +504,7 @@ int inet_bind(struct socket *sock, struc
26574 + if (sk->sk_state != TCP_CLOSE || inet->num)
26575 + goto out_release_sock;
26576 +
26577 +- inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
26578 ++ v4_set_sock_addr(inet, &nsa);
26579 + if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
26580 + inet->saddr = 0; /* Use device */
26581 +
26582 +@@ -688,11 +697,13 @@ int inet_getname(struct socket *sock, st
26583 + peer == 1))
26584 + return -ENOTCONN;
26585 + sin->sin_port = inet->dport;
26586 +- sin->sin_addr.s_addr = inet->daddr;
26587 ++ sin->sin_addr.s_addr =
26588 ++ nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
26589 + } else {
26590 + __be32 addr = inet->rcv_saddr;
26591 + if (!addr)
26592 + addr = inet->saddr;
26593 ++ addr = nx_map_sock_lback(sk->sk_nx_info, addr);
26594 + sin->sin_port = inet->sport;
26595 + sin->sin_addr.s_addr = addr;
26596 + }
26597 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/devinet.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/devinet.c
26598 +--- linux-2.6.29.4/net/ipv4/devinet.c 2009-03-24 14:22:46.000000000 +0100
26599 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/devinet.c 2009-03-24 14:48:37.000000000 +0100
26600 +@@ -413,6 +413,7 @@ struct in_device *inetdev_by_index(struc
26601 + return in_dev;
26602 + }
26603 +
26604 ++
26605 + /* Called only from RTNL semaphored context. No locks. */
26606 +
26607 + struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix,
26608 +@@ -653,6 +654,8 @@ int devinet_ioctl(struct net *net, unsig
26609 + *colon = ':';
26610 +
26611 + if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) {
26612 ++ struct nx_info *nxi = current->nx_info;
26613 ++
26614 + if (tryaddrmatch) {
26615 + /* Matthias Andree */
26616 + /* compare label and address (4.4BSD style) */
26617 +@@ -661,6 +664,8 @@ int devinet_ioctl(struct net *net, unsig
26618 + This is checked above. */
26619 + for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
26620 + ifap = &ifa->ifa_next) {
26621 ++ if (!nx_v4_ifa_visible(nxi, ifa))
26622 ++ continue;
26623 + if (!strcmp(ifr.ifr_name, ifa->ifa_label) &&
26624 + sin_orig.sin_addr.s_addr ==
26625 + ifa->ifa_address) {
26626 +@@ -673,9 +678,12 @@ int devinet_ioctl(struct net *net, unsig
26627 + comparing just the label */
26628 + if (!ifa) {
26629 + for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
26630 +- ifap = &ifa->ifa_next)
26631 ++ ifap = &ifa->ifa_next) {
26632 ++ if (!nx_v4_ifa_visible(nxi, ifa))
26633 ++ continue;
26634 + if (!strcmp(ifr.ifr_name, ifa->ifa_label))
26635 + break;
26636 ++ }
26637 + }
26638 + }
26639 +
26640 +@@ -826,6 +834,8 @@ static int inet_gifconf(struct net_devic
26641 + goto out;
26642 +
26643 + for (; ifa; ifa = ifa->ifa_next) {
26644 ++ if (!nx_v4_ifa_visible(current->nx_info, ifa))
26645 ++ continue;
26646 + if (!buf) {
26647 + done += sizeof(ifr);
26648 + continue;
26649 +@@ -1156,6 +1166,7 @@ static int inet_dump_ifaddr(struct sk_bu
26650 + struct net_device *dev;
26651 + struct in_device *in_dev;
26652 + struct in_ifaddr *ifa;
26653 ++ struct sock *sk = skb->sk;
26654 + int s_ip_idx, s_idx = cb->args[0];
26655 +
26656 + s_ip_idx = ip_idx = cb->args[1];
26657 +@@ -1170,6 +1181,8 @@ static int inet_dump_ifaddr(struct sk_bu
26658 +
26659 + for (ifa = in_dev->ifa_list, ip_idx = 0; ifa;
26660 + ifa = ifa->ifa_next, ip_idx++) {
26661 ++ if (sk && !nx_v4_ifa_visible(sk->sk_nx_info, ifa))
26662 ++ continue;
26663 + if (ip_idx < s_ip_idx)
26664 + continue;
26665 + if (inet_fill_ifaddr(skb, ifa, NETLINK_CB(cb->skb).pid,
26666 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/fib_hash.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/fib_hash.c
26667 +--- linux-2.6.29.4/net/ipv4/fib_hash.c 2009-03-24 14:22:46.000000000 +0100
26668 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/fib_hash.c 2009-03-24 14:48:37.000000000 +0100
26669 +@@ -1022,7 +1022,7 @@ static int fib_seq_show(struct seq_file
26670 + prefix = f->fn_key;
26671 + mask = FZ_MASK(iter->zone);
26672 + flags = fib_flag_trans(fa->fa_type, mask, fi);
26673 +- if (fi)
26674 ++ if (fi && nx_dev_visible(current->nx_info, fi->fib_dev))
26675 + seq_printf(seq,
26676 + "%s\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u%n",
26677 + fi->fib_dev ? fi->fib_dev->name : "*", prefix,
26678 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/inet_connection_sock.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_connection_sock.c
26679 +--- linux-2.6.29.4/net/ipv4/inet_connection_sock.c 2009-03-24 14:22:46.000000000 +0100
26680 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_connection_sock.c 2009-03-24 14:48:37.000000000 +0100
26681 +@@ -49,10 +49,40 @@ void inet_get_local_port_range(int *low,
26682 + }
26683 + EXPORT_SYMBOL(inet_get_local_port_range);
26684 +
26685 ++int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
26686 ++{
26687 ++ __be32 sk1_rcv_saddr = inet_rcv_saddr(sk1),
26688 ++ sk2_rcv_saddr = inet_rcv_saddr(sk2);
26689 ++
26690 ++ if (inet_v6_ipv6only(sk2))
26691 ++ return 0;
26692 ++
26693 ++ if (sk1_rcv_saddr &&
26694 ++ sk2_rcv_saddr &&
26695 ++ sk1_rcv_saddr == sk2_rcv_saddr)
26696 ++ return 1;
26697 ++
26698 ++ if (sk1_rcv_saddr &&
26699 ++ !sk2_rcv_saddr &&
26700 ++ v4_addr_in_nx_info(sk2->sk_nx_info, sk1_rcv_saddr, NXA_MASK_BIND))
26701 ++ return 1;
26702 ++
26703 ++ if (sk2_rcv_saddr &&
26704 ++ !sk1_rcv_saddr &&
26705 ++ v4_addr_in_nx_info(sk1->sk_nx_info, sk2_rcv_saddr, NXA_MASK_BIND))
26706 ++ return 1;
26707 ++
26708 ++ if (!sk1_rcv_saddr &&
26709 ++ !sk2_rcv_saddr &&
26710 ++ nx_v4_addr_conflict(sk1->sk_nx_info, sk2->sk_nx_info))
26711 ++ return 1;
26712 ++
26713 ++ return 0;
26714 ++}
26715 ++
26716 + int inet_csk_bind_conflict(const struct sock *sk,
26717 + const struct inet_bind_bucket *tb)
26718 + {
26719 +- const __be32 sk_rcv_saddr = inet_rcv_saddr(sk);
26720 + struct sock *sk2;
26721 + struct hlist_node *node;
26722 + int reuse = sk->sk_reuse;
26723 +@@ -72,9 +102,7 @@ int inet_csk_bind_conflict(const struct
26724 + sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
26725 + if (!reuse || !sk2->sk_reuse ||
26726 + sk2->sk_state == TCP_LISTEN) {
26727 +- const __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
26728 +- if (!sk2_rcv_saddr || !sk_rcv_saddr ||
26729 +- sk2_rcv_saddr == sk_rcv_saddr)
26730 ++ if (ipv4_rcv_saddr_equal(sk, sk2))
26731 + break;
26732 + }
26733 + }
26734 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/inet_diag.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_diag.c
26735 +--- linux-2.6.29.4/net/ipv4/inet_diag.c 2009-03-24 14:22:46.000000000 +0100
26736 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_diag.c 2009-03-24 14:48:37.000000000 +0100
26737 +@@ -32,6 +32,8 @@
26738 + #include <linux/stddef.h>
26739 +
26740 + #include <linux/inet_diag.h>
26741 ++#include <linux/vs_network.h>
26742 ++#include <linux/vs_inet.h>
26743 +
26744 + static const struct inet_diag_handler **inet_diag_table;
26745 +
26746 +@@ -118,8 +120,8 @@ static int inet_csk_diag_fill(struct soc
26747 +
26748 + r->id.idiag_sport = inet->sport;
26749 + r->id.idiag_dport = inet->dport;
26750 +- r->id.idiag_src[0] = inet->rcv_saddr;
26751 +- r->id.idiag_dst[0] = inet->daddr;
26752 ++ r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, inet->rcv_saddr);
26753 ++ r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, inet->daddr);
26754 +
26755 + #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
26756 + if (r->idiag_family == AF_INET6) {
26757 +@@ -206,8 +208,8 @@ static int inet_twsk_diag_fill(struct in
26758 + r->id.idiag_cookie[1] = (u32)(((unsigned long)tw >> 31) >> 1);
26759 + r->id.idiag_sport = tw->tw_sport;
26760 + r->id.idiag_dport = tw->tw_dport;
26761 +- r->id.idiag_src[0] = tw->tw_rcv_saddr;
26762 +- r->id.idiag_dst[0] = tw->tw_daddr;
26763 ++ r->id.idiag_src[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_rcv_saddr);
26764 ++ r->id.idiag_dst[0] = nx_map_sock_lback(tw->tw_nx_info, tw->tw_daddr);
26765 + r->idiag_state = tw->tw_substate;
26766 + r->idiag_timer = 3;
26767 + r->idiag_expires = DIV_ROUND_UP(tmo * 1000, HZ);
26768 +@@ -264,6 +266,7 @@ static int inet_diag_get_exact(struct sk
26769 + err = -EINVAL;
26770 +
26771 + if (req->idiag_family == AF_INET) {
26772 ++ /* TODO: lback */
26773 + sk = inet_lookup(&init_net, hashinfo, req->id.idiag_dst[0],
26774 + req->id.idiag_dport, req->id.idiag_src[0],
26775 + req->id.idiag_sport, req->id.idiag_if);
26776 +@@ -506,6 +509,7 @@ static int inet_csk_diag_dump(struct soc
26777 + } else
26778 + #endif
26779 + {
26780 ++ /* TODO: lback */
26781 + entry.saddr = &inet->rcv_saddr;
26782 + entry.daddr = &inet->daddr;
26783 + }
26784 +@@ -542,6 +546,7 @@ static int inet_twsk_diag_dump(struct in
26785 + } else
26786 + #endif
26787 + {
26788 ++ /* TODO: lback */
26789 + entry.saddr = &tw->tw_rcv_saddr;
26790 + entry.daddr = &tw->tw_daddr;
26791 + }
26792 +@@ -588,8 +593,8 @@ static int inet_diag_fill_req(struct sk_
26793 +
26794 + r->id.idiag_sport = inet->sport;
26795 + r->id.idiag_dport = ireq->rmt_port;
26796 +- r->id.idiag_src[0] = ireq->loc_addr;
26797 +- r->id.idiag_dst[0] = ireq->rmt_addr;
26798 ++ r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->loc_addr);
26799 ++ r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->rmt_addr);
26800 + r->idiag_expires = jiffies_to_msecs(tmo);
26801 + r->idiag_rqueue = 0;
26802 + r->idiag_wqueue = 0;
26803 +@@ -659,6 +664,7 @@ static int inet_diag_dump_reqs(struct sk
26804 + continue;
26805 +
26806 + if (bc) {
26807 ++ /* TODO: lback */
26808 + entry.saddr =
26809 + #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
26810 + (entry.family == AF_INET6) ?
26811 +@@ -729,6 +735,8 @@ static int inet_diag_dump(struct sk_buff
26812 + sk_nulls_for_each(sk, node, &ilb->head) {
26813 + struct inet_sock *inet = inet_sk(sk);
26814 +
26815 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26816 ++ continue;
26817 + if (num < s_num) {
26818 + num++;
26819 + continue;
26820 +@@ -795,6 +803,8 @@ skip_listen_ht:
26821 + sk_nulls_for_each(sk, node, &head->chain) {
26822 + struct inet_sock *inet = inet_sk(sk);
26823 +
26824 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26825 ++ continue;
26826 + if (num < s_num)
26827 + goto next_normal;
26828 + if (!(r->idiag_states & (1 << sk->sk_state)))
26829 +@@ -819,6 +829,8 @@ next_normal:
26830 + inet_twsk_for_each(tw, node,
26831 + &head->twchain) {
26832 +
26833 ++ if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
26834 ++ continue;
26835 + if (num < s_num)
26836 + goto next_dying;
26837 + if (r->id.idiag_sport != tw->tw_sport &&
26838 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/inet_hashtables.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_hashtables.c
26839 +--- linux-2.6.29.4/net/ipv4/inet_hashtables.c 2009-03-24 14:22:46.000000000 +0100
26840 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/inet_hashtables.c 2009-04-17 15:35:48.000000000 +0200
26841 +@@ -21,6 +21,7 @@
26842 +
26843 + #include <net/inet_connection_sock.h>
26844 + #include <net/inet_hashtables.h>
26845 ++#include <net/route.h>
26846 + #include <net/ip.h>
26847 +
26848 + /*
26849 +@@ -125,6 +126,11 @@ static inline int compute_score(struct s
26850 + if (rcv_saddr != daddr)
26851 + return -1;
26852 + score += 2;
26853 ++ } else {
26854 ++ /* block non nx_info ips */
26855 ++ if (!v4_addr_in_nx_info(sk->sk_nx_info,
26856 ++ daddr, NXA_MASK_BIND))
26857 ++ return -1;
26858 + }
26859 + if (sk->sk_bound_dev_if) {
26860 + if (sk->sk_bound_dev_if != dif)
26861 +@@ -142,7 +148,6 @@ static inline int compute_score(struct s
26862 + * wildcarded during the search since they can never be otherwise.
26863 + */
26864 +
26865 +-
26866 + struct sock *__inet_lookup_listener(struct net *net,
26867 + struct inet_hashinfo *hashinfo,
26868 + const __be32 daddr, const unsigned short hnum,
26869 +@@ -165,6 +170,7 @@ begin:
26870 + hiscore = score;
26871 + }
26872 + }
26873 ++
26874 + /*
26875 + * if the nulls value we got at the end of this lookup is
26876 + * not the expected one, we must restart lookup.
26877 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/netfilter/nf_nat_helper.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/netfilter/nf_nat_helper.c
26878 +--- linux-2.6.29.4/net/ipv4/netfilter/nf_nat_helper.c 2008-12-25 00:26:37.000000000 +0100
26879 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/netfilter/nf_nat_helper.c 2009-02-22 22:54:26.000000000 +0100
26880 +@@ -19,6 +19,7 @@
26881 + #include <net/route.h>
26882 +
26883 + #include <linux/netfilter_ipv4.h>
26884 ++#include <net/route.h>
26885 + #include <net/netfilter/nf_conntrack.h>
26886 + #include <net/netfilter/nf_conntrack_helper.h>
26887 + #include <net/netfilter/nf_conntrack_ecache.h>
26888 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/netfilter.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/netfilter.c
26889 +--- linux-2.6.29.4/net/ipv4/netfilter.c 2009-03-24 14:22:46.000000000 +0100
26890 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/netfilter.c 2009-03-24 14:48:37.000000000 +0100
26891 +@@ -4,7 +4,7 @@
26892 + #include <linux/netfilter_ipv4.h>
26893 + #include <linux/ip.h>
26894 + #include <linux/skbuff.h>
26895 +-#include <net/route.h>
26896 ++// #include <net/route.h>
26897 + #include <net/xfrm.h>
26898 + #include <net/ip.h>
26899 + #include <net/netfilter/nf_queue.h>
26900 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/raw.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/raw.c
26901 +--- linux-2.6.29.4/net/ipv4/raw.c 2009-03-24 14:22:46.000000000 +0100
26902 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/raw.c 2009-03-24 14:48:37.000000000 +0100
26903 +@@ -117,7 +117,7 @@ static struct sock *__raw_v4_lookup(stru
26904 +
26905 + if (net_eq(sock_net(sk), net) && inet->num == num &&
26906 + !(inet->daddr && inet->daddr != raddr) &&
26907 +- !(inet->rcv_saddr && inet->rcv_saddr != laddr) &&
26908 ++ v4_sock_addr_match(sk->sk_nx_info, inet, laddr) &&
26909 + !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
26910 + goto found; /* gotcha */
26911 + }
26912 +@@ -372,6 +372,12 @@ static int raw_send_hdrinc(struct sock *
26913 + icmp_out_count(net, ((struct icmphdr *)
26914 + skb_transport_header(skb))->type);
26915 +
26916 ++ err = -EPERM;
26917 ++ if (!nx_check(0, VS_ADMIN) && !capable(CAP_NET_RAW) &&
26918 ++ sk->sk_nx_info &&
26919 ++ !v4_addr_in_nx_info(sk->sk_nx_info, iph->saddr, NXA_MASK_BIND))
26920 ++ goto error_free;
26921 ++
26922 + err = NF_HOOK(PF_INET, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
26923 + dst_output);
26924 + if (err > 0)
26925 +@@ -383,6 +389,7 @@ out:
26926 +
26927 + error_fault:
26928 + err = -EFAULT;
26929 ++error_free:
26930 + kfree_skb(skb);
26931 + error:
26932 + IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
26933 +@@ -550,6 +557,13 @@ static int raw_sendmsg(struct kiocb *ioc
26934 + }
26935 +
26936 + security_sk_classify_flow(sk, &fl);
26937 ++ if (sk->sk_nx_info) {
26938 ++ err = ip_v4_find_src(sock_net(sk),
26939 ++ sk->sk_nx_info, &rt, &fl);
26940 ++
26941 ++ if (err)
26942 ++ goto done;
26943 ++ }
26944 + err = ip_route_output_flow(sock_net(sk), &rt, &fl, sk, 1);
26945 + }
26946 + if (err)
26947 +@@ -619,17 +633,19 @@ static int raw_bind(struct sock *sk, str
26948 + {
26949 + struct inet_sock *inet = inet_sk(sk);
26950 + struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
26951 ++ struct nx_v4_sock_addr nsa = { 0 };
26952 + int ret = -EINVAL;
26953 + int chk_addr_ret;
26954 +
26955 + if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
26956 + goto out;
26957 +- chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
26958 ++ v4_map_sock_addr(inet, addr, &nsa);
26959 ++ chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
26960 + ret = -EADDRNOTAVAIL;
26961 +- if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
26962 ++ if (nsa.saddr && chk_addr_ret != RTN_LOCAL &&
26963 + chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
26964 + goto out;
26965 +- inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
26966 ++ v4_set_sock_addr(inet, &nsa);
26967 + if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
26968 + inet->saddr = 0; /* Use device */
26969 + sk_dst_reset(sk);
26970 +@@ -681,7 +697,8 @@ static int raw_recvmsg(struct kiocb *ioc
26971 + /* Copy the address. */
26972 + if (sin) {
26973 + sin->sin_family = AF_INET;
26974 +- sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
26975 ++ sin->sin_addr.s_addr =
26976 ++ nx_map_sock_lback(sk->sk_nx_info, ip_hdr(skb)->saddr);
26977 + sin->sin_port = 0;
26978 + memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
26979 + }
26980 +@@ -858,7 +875,8 @@ static struct sock *raw_get_first(struct
26981 + struct hlist_node *node;
26982 +
26983 + sk_for_each(sk, node, &state->h->ht[state->bucket])
26984 +- if (sock_net(sk) == seq_file_net(seq))
26985 ++ if ((sock_net(sk) == seq_file_net(seq)) &&
26986 ++ nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
26987 + goto found;
26988 + }
26989 + sk = NULL;
26990 +@@ -874,7 +892,8 @@ static struct sock *raw_get_next(struct
26991 + sk = sk_next(sk);
26992 + try_again:
26993 + ;
26994 +- } while (sk && sock_net(sk) != seq_file_net(seq));
26995 ++ } while (sk && ((sock_net(sk) != seq_file_net(seq)) ||
26996 ++ !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
26997 +
26998 + if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
26999 + sk = sk_head(&state->h->ht[state->bucket]);
27000 +@@ -933,7 +952,10 @@ static void raw_sock_seq_show(struct seq
27001 +
27002 + seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
27003 + " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
27004 +- i, src, srcp, dest, destp, sp->sk_state,
27005 ++ i,
27006 ++ nx_map_sock_lback(current_nx_info(), src), srcp,
27007 ++ nx_map_sock_lback(current_nx_info(), dest), destp,
27008 ++ sp->sk_state,
27009 + atomic_read(&sp->sk_wmem_alloc),
27010 + atomic_read(&sp->sk_rmem_alloc),
27011 + 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
27012 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/tcp.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp.c
27013 +--- linux-2.6.29.4/net/ipv4/tcp.c 2009-03-24 14:22:46.000000000 +0100
27014 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp.c 2009-03-24 14:48:37.000000000 +0100
27015 +@@ -264,6 +264,7 @@
27016 + #include <linux/cache.h>
27017 + #include <linux/err.h>
27018 + #include <linux/crypto.h>
27019 ++#include <linux/in.h>
27020 +
27021 + #include <net/icmp.h>
27022 + #include <net/tcp.h>
27023 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/tcp_ipv4.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp_ipv4.c
27024 +--- linux-2.6.29.4/net/ipv4/tcp_ipv4.c 2009-05-23 23:16:53.000000000 +0200
27025 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp_ipv4.c 2009-05-23 23:19:11.000000000 +0200
27026 +@@ -1895,6 +1895,12 @@ static void *listening_get_next(struct s
27027 + req = req->dl_next;
27028 + while (1) {
27029 + while (req) {
27030 ++ vxdprintk(VXD_CBIT(net, 6),
27031 ++ "sk,req: %p [#%d] (from %d)", req->sk,
27032 ++ (req->sk)?req->sk->sk_nid:0, nx_current_nid());
27033 ++ if (req->sk &&
27034 ++ !nx_check(req->sk->sk_nid, VS_WATCH_P | VS_IDENT))
27035 ++ continue;
27036 + if (req->rsk_ops->family == st->family) {
27037 + cur = req;
27038 + goto out;
27039 +@@ -1919,6 +1925,10 @@ get_req:
27040 + }
27041 + get_sk:
27042 + sk_nulls_for_each_from(sk, node) {
27043 ++ vxdprintk(VXD_CBIT(net, 6), "sk: %p [#%d] (from %d)",
27044 ++ sk, sk->sk_nid, nx_current_nid());
27045 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27046 ++ continue;
27047 + if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
27048 + cur = sk;
27049 + goto out;
27050 +@@ -1982,6 +1992,11 @@ static void *established_get_first(struc
27051 +
27052 + spin_lock_bh(lock);
27053 + sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
27054 ++ vxdprintk(VXD_CBIT(net, 6),
27055 ++ "sk,egf: %p [#%d] (from %d)",
27056 ++ sk, sk->sk_nid, nx_current_nid());
27057 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27058 ++ continue;
27059 + if (sk->sk_family != st->family ||
27060 + !net_eq(sock_net(sk), net)) {
27061 + continue;
27062 +@@ -1992,6 +2007,11 @@ static void *established_get_first(struc
27063 + st->state = TCP_SEQ_STATE_TIME_WAIT;
27064 + inet_twsk_for_each(tw, node,
27065 + &tcp_hashinfo.ehash[st->bucket].twchain) {
27066 ++ vxdprintk(VXD_CBIT(net, 6),
27067 ++ "tw: %p [#%d] (from %d)",
27068 ++ tw, tw->tw_nid, nx_current_nid());
27069 ++ if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
27070 ++ continue;
27071 + if (tw->tw_family != st->family ||
27072 + !net_eq(twsk_net(tw), net)) {
27073 + continue;
27074 +@@ -2020,7 +2040,9 @@ static void *established_get_next(struct
27075 + tw = cur;
27076 + tw = tw_next(tw);
27077 + get_tw:
27078 +- while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
27079 ++ while (tw && (tw->tw_family != st->family ||
27080 ++ !net_eq(twsk_net(tw), net) ||
27081 ++ !nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))) {
27082 + tw = tw_next(tw);
27083 + }
27084 + if (tw) {
27085 +@@ -2043,6 +2065,11 @@ get_tw:
27086 + sk = sk_nulls_next(sk);
27087 +
27088 + sk_nulls_for_each_from(sk, node) {
27089 ++ vxdprintk(VXD_CBIT(net, 6),
27090 ++ "sk,egn: %p [#%d] (from %d)",
27091 ++ sk, sk->sk_nid, nx_current_nid());
27092 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27093 ++ continue;
27094 + if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
27095 + goto found;
27096 + }
27097 +@@ -2194,9 +2221,9 @@ static void get_openreq4(struct sock *sk
27098 + seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27099 + " %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p%n",
27100 + i,
27101 +- ireq->loc_addr,
27102 ++ nx_map_sock_lback(current_nx_info(), ireq->loc_addr),
27103 + ntohs(inet_sk(sk)->sport),
27104 +- ireq->rmt_addr,
27105 ++ nx_map_sock_lback(current_nx_info(), ireq->rmt_addr),
27106 + ntohs(ireq->rmt_port),
27107 + TCP_SYN_RECV,
27108 + 0, 0, /* could print option size, but that is af dependent. */
27109 +@@ -2239,7 +2266,10 @@ static void get_tcp4_sock(struct sock *s
27110 +
27111 + seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
27112 + "%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
27113 +- i, src, srcp, dest, destp, sk->sk_state,
27114 ++ i,
27115 ++ nx_map_sock_lback(current_nx_info(), src), srcp,
27116 ++ nx_map_sock_lback(current_nx_info(), dest), destp,
27117 ++ sk->sk_state,
27118 + tp->write_seq - tp->snd_una,
27119 + sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
27120 + (tp->rcv_nxt - tp->copied_seq),
27121 +@@ -2275,7 +2305,10 @@ static void get_timewait4_sock(struct in
27122 +
27123 + seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27124 + " %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p%n",
27125 +- i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
27126 ++ i,
27127 ++ nx_map_sock_lback(current_nx_info(), src), srcp,
27128 ++ nx_map_sock_lback(current_nx_info(), dest), destp,
27129 ++ tw->tw_substate, 0, 0,
27130 + 3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
27131 + atomic_read(&tw->tw_refcnt), tw, len);
27132 + }
27133 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/tcp_minisocks.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp_minisocks.c
27134 +--- linux-2.6.29.4/net/ipv4/tcp_minisocks.c 2009-03-24 14:22:46.000000000 +0100
27135 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/tcp_minisocks.c 2009-03-24 14:48:37.000000000 +0100
27136 +@@ -26,6 +26,10 @@
27137 + #include <net/inet_common.h>
27138 + #include <net/xfrm.h>
27139 +
27140 ++#include <linux/vs_limit.h>
27141 ++#include <linux/vs_socket.h>
27142 ++#include <linux/vs_context.h>
27143 ++
27144 + #ifdef CONFIG_SYSCTL
27145 + #define SYNC_INIT 0 /* let the user enable it */
27146 + #else
27147 +@@ -293,6 +297,11 @@ void tcp_time_wait(struct sock *sk, int
27148 + tcptw->tw_ts_recent = tp->rx_opt.ts_recent;
27149 + tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
27150 +
27151 ++ tw->tw_xid = sk->sk_xid;
27152 ++ tw->tw_vx_info = NULL;
27153 ++ tw->tw_nid = sk->sk_nid;
27154 ++ tw->tw_nx_info = NULL;
27155 ++
27156 + #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
27157 + if (tw->tw_family == PF_INET6) {
27158 + struct ipv6_pinfo *np = inet6_sk(sk);
27159 +diff -NurpP --minimal linux-2.6.29.4/net/ipv4/udp.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/udp.c
27160 +--- linux-2.6.29.4/net/ipv4/udp.c 2009-05-23 23:16:53.000000000 +0200
27161 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv4/udp.c 2009-04-15 22:58:30.000000000 +0200
27162 +@@ -222,14 +222,7 @@ fail:
27163 + return error;
27164 + }
27165 +
27166 +-static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
27167 +-{
27168 +- struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
27169 +-
27170 +- return ( !ipv6_only_sock(sk2) &&
27171 +- (!inet1->rcv_saddr || !inet2->rcv_saddr ||
27172 +- inet1->rcv_saddr == inet2->rcv_saddr ));
27173 +-}
27174 ++extern int ipv4_rcv_saddr_equal(const struct sock *, const struct sock *);
27175 +
27176 + int udp_v4_get_port(struct sock *sk, unsigned short snum)
27177 + {
27178 +@@ -251,6 +244,11 @@ static inline int compute_score(struct s
27179 + if (inet->rcv_saddr != daddr)
27180 + return -1;
27181 + score += 2;
27182 ++ } else {
27183 ++ /* block non nx_info ips */
27184 ++ if (!v4_addr_in_nx_info(sk->sk_nx_info,
27185 ++ daddr, NXA_MASK_BIND))
27186 ++ return -1;
27187 + }
27188 + if (inet->daddr) {
27189 + if (inet->daddr != saddr)
27190 +@@ -271,6 +269,7 @@ static inline int compute_score(struct s
27191 + return score;
27192 + }
27193 +
27194 ++
27195 + /* UDP is nearly always wildcards out the wazoo, it makes no sense to try
27196 + * harder than this. -DaveM
27197 + */
27198 +@@ -292,6 +291,11 @@ begin:
27199 + sk_nulls_for_each_rcu(sk, node, &hslot->head) {
27200 + score = compute_score(sk, net, saddr, hnum, sport,
27201 + daddr, dport, dif);
27202 ++ /* FIXME: disabled?
27203 ++ if (score == 9) {
27204 ++ result = sk;
27205 ++ break;
27206 ++ } else */
27207 + if (score > badness) {
27208 + result = sk;
27209 + badness = score;
27210 +@@ -305,6 +309,7 @@ begin:
27211 + if (get_nulls_value(node) != hash)
27212 + goto begin;
27213 +
27214 ++
27215 + if (result) {
27216 + if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
27217 + result = NULL;
27218 +@@ -314,6 +319,7 @@ begin:
27219 + goto begin;
27220 + }
27221 + }
27222 ++
27223 + rcu_read_unlock();
27224 + return result;
27225 + }
27226 +@@ -356,7 +362,7 @@ static inline struct sock *udp_v4_mcast_
27227 + s->sk_hash != hnum ||
27228 + (inet->daddr && inet->daddr != rmt_addr) ||
27229 + (inet->dport != rmt_port && inet->dport) ||
27230 +- (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
27231 ++ !v4_sock_addr_match(sk->sk_nx_info, inet, loc_addr) ||
27232 + ipv6_only_sock(s) ||
27233 + (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
27234 + continue;
27235 +@@ -694,8 +700,13 @@ int udp_sendmsg(struct kiocb *iocb, stru
27236 + { .sport = inet->sport,
27237 + .dport = dport } } };
27238 + struct net *net = sock_net(sk);
27239 ++ struct nx_info *nxi = sk->sk_nx_info;
27240 +
27241 + security_sk_classify_flow(sk, &fl);
27242 ++ err = ip_v4_find_src(net, nxi, &rt, &fl);
27243 ++ if (err)
27244 ++ goto out;
27245 ++
27246 + err = ip_route_output_flow(net, &rt, &fl, sk, 1);
27247 + if (err) {
27248 + if (err == -ENETUNREACH)
27249 +@@ -940,7 +951,8 @@ try_again:
27250 + {
27251 + sin->sin_family = AF_INET;
27252 + sin->sin_port = udp_hdr(skb)->source;
27253 +- sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
27254 ++ sin->sin_addr.s_addr = nx_map_sock_lback(
27255 ++ skb->sk->sk_nx_info, ip_hdr(skb)->saddr);
27256 + memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
27257 + }
27258 + if (inet->cmsg_flags)
27259 +@@ -1594,6 +1606,8 @@ static struct sock *udp_get_first(struct
27260 + sk_nulls_for_each(sk, node, &hslot->head) {
27261 + if (!net_eq(sock_net(sk), net))
27262 + continue;
27263 ++ if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
27264 ++ continue;
27265 + if (sk->sk_family == state->family)
27266 + goto found;
27267 + }
27268 +@@ -1611,7 +1625,9 @@ static struct sock *udp_get_next(struct
27269 +
27270 + do {
27271 + sk = sk_nulls_next(sk);
27272 +- } while (sk && (!net_eq(sock_net(sk), net) || sk->sk_family != state->family));
27273 ++ } while (sk && (!net_eq(sock_net(sk), net) ||
27274 ++ sk->sk_family != state->family ||
27275 ++ !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
27276 +
27277 + if (!sk) {
27278 + if (state->bucket < UDP_HTABLE_SIZE)
27279 +@@ -1716,7 +1732,10 @@ static void udp4_format_sock(struct sock
27280 +
27281 + seq_printf(f, "%4d: %08X:%04X %08X:%04X"
27282 + " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d%n",
27283 +- bucket, src, srcp, dest, destp, sp->sk_state,
27284 ++ bucket,
27285 ++ nx_map_sock_lback(current_nx_info(), src), srcp,
27286 ++ nx_map_sock_lback(current_nx_info(), dest), destp,
27287 ++ sp->sk_state,
27288 + atomic_read(&sp->sk_wmem_alloc),
27289 + atomic_read(&sp->sk_rmem_alloc),
27290 + 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
27291 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/addrconf.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/addrconf.c
27292 +--- linux-2.6.29.4/net/ipv6/addrconf.c 2009-03-24 14:22:46.000000000 +0100
27293 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/addrconf.c 2009-03-24 20:39:23.000000000 +0100
27294 +@@ -85,6 +85,8 @@
27295 +
27296 + #include <linux/proc_fs.h>
27297 + #include <linux/seq_file.h>
27298 ++#include <linux/vs_network.h>
27299 ++#include <linux/vs_inet6.h>
27300 +
27301 + /* Set to 3 to get tracing... */
27302 + #define ACONF_DEBUG 2
27303 +@@ -1111,7 +1113,7 @@ out:
27304 +
27305 + int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
27306 + const struct in6_addr *daddr, unsigned int prefs,
27307 +- struct in6_addr *saddr)
27308 ++ struct in6_addr *saddr, struct nx_info *nxi)
27309 + {
27310 + struct ipv6_saddr_score scores[2],
27311 + *score = &scores[0], *hiscore = &scores[1];
27312 +@@ -1184,6 +1186,8 @@ int ipv6_dev_get_saddr(struct net *net,
27313 + dev->name);
27314 + continue;
27315 + }
27316 ++ if (!v6_addr_in_nx_info(nxi, &score->ifa->addr, -1))
27317 ++ continue;
27318 +
27319 + score->rule = -1;
27320 + bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
27321 +@@ -1367,35 +1371,46 @@ struct inet6_ifaddr *ipv6_get_ifaddr(str
27322 + return ifp;
27323 + }
27324 +
27325 ++extern int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2);
27326 ++
27327 + int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
27328 + {
27329 + const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
27330 + const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
27331 +- __be32 sk_rcv_saddr = inet_sk(sk)->rcv_saddr;
27332 + __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
27333 + int sk_ipv6only = ipv6_only_sock(sk);
27334 + int sk2_ipv6only = inet_v6_ipv6only(sk2);
27335 + int addr_type = ipv6_addr_type(sk_rcv_saddr6);
27336 + int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
27337 +
27338 +- if (!sk2_rcv_saddr && !sk_ipv6only)
27339 ++ /* FIXME: needs handling for v4 ANY */
27340 ++ if (!sk2_rcv_saddr && !sk_ipv6only && !sk2->sk_nx_info)
27341 + return 1;
27342 +
27343 + if (addr_type2 == IPV6_ADDR_ANY &&
27344 +- !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
27345 ++ !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED) &&
27346 ++ v6_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr6, -1))
27347 + return 1;
27348 +
27349 + if (addr_type == IPV6_ADDR_ANY &&
27350 +- !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
27351 ++ !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED) &&
27352 ++ (sk2_rcv_saddr6 && v6_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr6, -1)))
27353 ++ return 1;
27354 ++
27355 ++ if (addr_type == IPV6_ADDR_ANY &&
27356 ++ addr_type2 == IPV6_ADDR_ANY &&
27357 ++ nx_v6_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info))
27358 + return 1;
27359 +
27360 + if (sk2_rcv_saddr6 &&
27361 ++ addr_type != IPV6_ADDR_ANY &&
27362 ++ addr_type != IPV6_ADDR_ANY &&
27363 + ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
27364 + return 1;
27365 +
27366 + if (addr_type == IPV6_ADDR_MAPPED &&
27367 + !sk2_ipv6only &&
27368 +- (!sk2_rcv_saddr || !sk_rcv_saddr || sk_rcv_saddr == sk2_rcv_saddr))
27369 ++ ipv4_rcv_saddr_equal(sk, sk2))
27370 + return 1;
27371 +
27372 + return 0;
27373 +@@ -2993,7 +3008,10 @@ static void if6_seq_stop(struct seq_file
27374 + static int if6_seq_show(struct seq_file *seq, void *v)
27375 + {
27376 + struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
27377 +- seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27378 ++
27379 ++ if (nx_check(0, VS_ADMIN|VS_WATCH) ||
27380 ++ v6_addr_in_nx_info(current_nx_info(), &ifp->addr, -1))
27381 ++ seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
27382 + &ifp->addr,
27383 + ifp->idev->dev->ifindex,
27384 + ifp->prefix_len,
27385 +@@ -3487,6 +3505,12 @@ static int inet6_dump_addr(struct sk_buf
27386 + struct ifmcaddr6 *ifmca;
27387 + struct ifacaddr6 *ifaca;
27388 + struct net *net = sock_net(skb->sk);
27389 ++ struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27390 ++
27391 ++ /* disable ipv6 on non v6 guests */
27392 ++ if (nxi && !nx_info_has_v6(nxi))
27393 ++ return skb->len;
27394 ++
27395 +
27396 + s_idx = cb->args[0];
27397 + s_ip_idx = ip_idx = cb->args[1];
27398 +@@ -3508,6 +3532,8 @@ static int inet6_dump_addr(struct sk_buf
27399 + ifa = ifa->if_next, ip_idx++) {
27400 + if (ip_idx < s_ip_idx)
27401 + continue;
27402 ++ if (!v6_addr_in_nx_info(nxi, &ifa->addr, -1))
27403 ++ continue;
27404 + err = inet6_fill_ifaddr(skb, ifa,
27405 + NETLINK_CB(cb->skb).pid,
27406 + cb->nlh->nlmsg_seq,
27407 +@@ -3521,6 +3547,8 @@ static int inet6_dump_addr(struct sk_buf
27408 + ifmca = ifmca->next, ip_idx++) {
27409 + if (ip_idx < s_ip_idx)
27410 + continue;
27411 ++ if (!v6_addr_in_nx_info(nxi, &ifmca->mca_addr, -1))
27412 ++ continue;
27413 + err = inet6_fill_ifmcaddr(skb, ifmca,
27414 + NETLINK_CB(cb->skb).pid,
27415 + cb->nlh->nlmsg_seq,
27416 +@@ -3534,6 +3562,8 @@ static int inet6_dump_addr(struct sk_buf
27417 + ifaca = ifaca->aca_next, ip_idx++) {
27418 + if (ip_idx < s_ip_idx)
27419 + continue;
27420 ++ if (!v6_addr_in_nx_info(nxi, &ifaca->aca_addr, -1))
27421 ++ continue;
27422 + err = inet6_fill_ifacaddr(skb, ifaca,
27423 + NETLINK_CB(cb->skb).pid,
27424 + cb->nlh->nlmsg_seq,
27425 +@@ -3819,12 +3849,19 @@ static int inet6_dump_ifinfo(struct sk_b
27426 + int s_idx = cb->args[0];
27427 + struct net_device *dev;
27428 + struct inet6_dev *idev;
27429 ++ struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
27430 ++
27431 ++ /* FIXME: maybe disable ipv6 on non v6 guests?
27432 ++ if (skb->sk && skb->sk->sk_vx_info)
27433 ++ return skb->len; */
27434 +
27435 + read_lock(&dev_base_lock);
27436 + idx = 0;
27437 + for_each_netdev(net, dev) {
27438 + if (idx < s_idx)
27439 + goto cont;
27440 ++ if (!v6_dev_in_nx_info(dev, nxi))
27441 ++ goto cont;
27442 + if ((idev = in6_dev_get(dev)) == NULL)
27443 + goto cont;
27444 + err = inet6_fill_ifinfo(skb, idev, NETLINK_CB(cb->skb).pid,
27445 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/af_inet6.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/af_inet6.c
27446 +--- linux-2.6.29.4/net/ipv6/af_inet6.c 2009-03-24 14:22:46.000000000 +0100
27447 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/af_inet6.c 2009-03-24 14:48:37.000000000 +0100
27448 +@@ -41,6 +41,8 @@
27449 + #include <linux/netdevice.h>
27450 + #include <linux/icmpv6.h>
27451 + #include <linux/netfilter_ipv6.h>
27452 ++#include <linux/vs_inet.h>
27453 ++#include <linux/vs_inet6.h>
27454 +
27455 + #include <net/ip.h>
27456 + #include <net/ipv6.h>
27457 +@@ -49,6 +51,7 @@
27458 + #include <net/tcp.h>
27459 + #include <net/ipip.h>
27460 + #include <net/protocol.h>
27461 ++#include <net/route.h>
27462 + #include <net/inet_common.h>
27463 + #include <net/route.h>
27464 + #include <net/transp_v6.h>
27465 +@@ -146,9 +149,12 @@ lookup_protocol:
27466 + }
27467 +
27468 + err = -EPERM;
27469 ++ if ((protocol == IPPROTO_ICMPV6) &&
27470 ++ nx_capable(answer->capability, NXC_RAW_ICMP))
27471 ++ goto override;
27472 + if (answer->capability > 0 && !capable(answer->capability))
27473 + goto out_rcu_unlock;
27474 +-
27475 ++override:
27476 + sock->ops = answer->ops;
27477 + answer_prot = answer->prot;
27478 + answer_no_check = answer->no_check;
27479 +@@ -247,6 +253,7 @@ int inet6_bind(struct socket *sock, stru
27480 + struct inet_sock *inet = inet_sk(sk);
27481 + struct ipv6_pinfo *np = inet6_sk(sk);
27482 + struct net *net = sock_net(sk);
27483 ++ struct nx_v6_sock_addr nsa;
27484 + __be32 v4addr = 0;
27485 + unsigned short snum;
27486 + int addr_type = 0;
27487 +@@ -258,6 +265,11 @@ int inet6_bind(struct socket *sock, stru
27488 +
27489 + if (addr_len < SIN6_LEN_RFC2133)
27490 + return -EINVAL;
27491 ++
27492 ++ err = v6_map_sock_addr(inet, addr, &nsa);
27493 ++ if (err)
27494 ++ return err;
27495 ++
27496 + addr_type = ipv6_addr_type(&addr->sin6_addr);
27497 + if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
27498 + return -EINVAL;
27499 +@@ -281,6 +293,10 @@ int inet6_bind(struct socket *sock, stru
27500 + err = -EADDRNOTAVAIL;
27501 + goto out;
27502 + }
27503 ++ if (!v4_addr_in_nx_info(sk->sk_nx_info, v4addr, NXA_MASK_BIND)) {
27504 ++ err = -EADDRNOTAVAIL;
27505 ++ goto out;
27506 ++ }
27507 + } else {
27508 + if (addr_type != IPV6_ADDR_ANY) {
27509 + struct net_device *dev = NULL;
27510 +@@ -306,6 +322,11 @@ int inet6_bind(struct socket *sock, stru
27511 + }
27512 + }
27513 +
27514 ++ if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
27515 ++ err = -EADDRNOTAVAIL;
27516 ++ goto out;
27517 ++ }
27518 ++
27519 + /* ipv4 addr of the socket is invalid. Only the
27520 + * unspecified and mapped address have a v4 equivalent.
27521 + */
27522 +@@ -324,6 +345,8 @@ int inet6_bind(struct socket *sock, stru
27523 + }
27524 + }
27525 +
27526 ++ v6_set_sock_addr(inet, &nsa);
27527 ++
27528 + inet->rcv_saddr = v4addr;
27529 + inet->saddr = v4addr;
27530 +
27531 +@@ -416,9 +439,11 @@ int inet6_getname(struct socket *sock, s
27532 + return -ENOTCONN;
27533 + sin->sin6_port = inet->dport;
27534 + ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
27535 ++ /* FIXME: remap lback? */
27536 + if (np->sndflow)
27537 + sin->sin6_flowinfo = np->flow_label;
27538 + } else {
27539 ++ /* FIXME: remap lback? */
27540 + if (ipv6_addr_any(&np->rcv_saddr))
27541 + ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
27542 + else
27543 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/fib6_rules.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/fib6_rules.c
27544 +--- linux-2.6.29.4/net/ipv6/fib6_rules.c 2008-12-25 00:26:37.000000000 +0100
27545 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/fib6_rules.c 2009-02-22 22:54:26.000000000 +0100
27546 +@@ -96,7 +96,7 @@ static int fib6_rule_action(struct fib_r
27547 + if (ipv6_dev_get_saddr(net,
27548 + ip6_dst_idev(&rt->u.dst)->dev,
27549 + &flp->fl6_dst, srcprefs,
27550 +- &saddr))
27551 ++ &saddr, NULL))
27552 + goto again;
27553 + if (!ipv6_prefix_equal(&saddr, &r->src.addr,
27554 + r->src.plen))
27555 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/inet6_hashtables.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/inet6_hashtables.c
27556 +--- linux-2.6.29.4/net/ipv6/inet6_hashtables.c 2009-03-24 14:22:46.000000000 +0100
27557 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/inet6_hashtables.c 2009-03-24 20:50:24.000000000 +0100
27558 +@@ -16,6 +16,7 @@
27559 +
27560 + #include <linux/module.h>
27561 + #include <linux/random.h>
27562 ++#include <linux/vs_inet6.h>
27563 +
27564 + #include <net/inet_connection_sock.h>
27565 + #include <net/inet_hashtables.h>
27566 +@@ -76,7 +77,6 @@ struct sock *__inet6_lookup_established(
27567 + unsigned int slot = hash & (hashinfo->ehash_size - 1);
27568 + struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
27569 +
27570 +-
27571 + rcu_read_lock();
27572 + begin:
27573 + sk_nulls_for_each_rcu(sk, node, &head->chain) {
27574 +@@ -88,7 +88,7 @@ begin:
27575 + sock_put(sk);
27576 + goto begin;
27577 + }
27578 +- goto out;
27579 ++ goto out;
27580 + }
27581 + }
27582 + if (get_nulls_value(node) != slot)
27583 +@@ -134,6 +134,9 @@ static int inline compute_score(struct s
27584 + if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
27585 + return -1;
27586 + score++;
27587 ++ } else {
27588 ++ if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
27589 ++ return -1;
27590 + }
27591 + if (sk->sk_bound_dev_if) {
27592 + if (sk->sk_bound_dev_if != dif)
27593 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/ip6_output.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/ip6_output.c
27594 +--- linux-2.6.29.4/net/ipv6/ip6_output.c 2009-03-24 14:22:47.000000000 +0100
27595 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/ip6_output.c 2009-03-24 14:48:37.000000000 +0100
27596 +@@ -951,7 +951,7 @@ static int ip6_dst_lookup_tail(struct so
27597 + err = ipv6_dev_get_saddr(net, ip6_dst_idev(*dst)->dev,
27598 + &fl->fl6_dst,
27599 + sk ? inet6_sk(sk)->srcprefs : 0,
27600 +- &fl->fl6_src);
27601 ++ &fl->fl6_src, sk->sk_nx_info);
27602 + if (err)
27603 + goto out_err_release;
27604 + }
27605 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/Kconfig linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/Kconfig
27606 +--- linux-2.6.29.4/net/ipv6/Kconfig 2008-12-25 00:26:37.000000000 +0100
27607 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/Kconfig 2009-02-22 22:54:26.000000000 +0100
27608 +@@ -4,8 +4,8 @@
27609 +
27610 + # IPv6 as module will cause a CRASH if you try to unload it
27611 + menuconfig IPV6
27612 +- tristate "The IPv6 protocol"
27613 +- default m
27614 ++ bool "The IPv6 protocol"
27615 ++ default n
27616 + ---help---
27617 + This is complemental support for the IP version 6.
27618 + You will still be able to do traditional IPv4 networking as well.
27619 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/ndisc.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/ndisc.c
27620 +--- linux-2.6.29.4/net/ipv6/ndisc.c 2009-03-24 14:22:47.000000000 +0100
27621 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/ndisc.c 2009-03-24 14:48:37.000000000 +0100
27622 +@@ -589,7 +589,7 @@ static void ndisc_send_na(struct net_dev
27623 + } else {
27624 + if (ipv6_dev_get_saddr(dev_net(dev), dev, daddr,
27625 + inet6_sk(dev_net(dev)->ipv6.ndisc_sk)->srcprefs,
27626 +- &tmpaddr))
27627 ++ &tmpaddr, NULL /* FIXME: ? */ ))
27628 + return;
27629 + src_addr = &tmpaddr;
27630 + }
27631 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/raw.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/raw.c
27632 +--- linux-2.6.29.4/net/ipv6/raw.c 2009-03-24 14:22:47.000000000 +0100
27633 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/raw.c 2009-03-24 14:48:37.000000000 +0100
27634 +@@ -29,6 +29,7 @@
27635 + #include <linux/icmpv6.h>
27636 + #include <linux/netfilter.h>
27637 + #include <linux/netfilter_ipv6.h>
27638 ++#include <linux/vs_inet6.h>
27639 + #include <linux/skbuff.h>
27640 + #include <asm/uaccess.h>
27641 + #include <asm/ioctls.h>
27642 +@@ -281,6 +282,13 @@ static int rawv6_bind(struct sock *sk, s
27643 + }
27644 + }
27645 +
27646 ++ if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
27647 ++ err = -EADDRNOTAVAIL;
27648 ++ if (dev)
27649 ++ dev_put(dev);
27650 ++ goto out;
27651 ++ }
27652 ++
27653 + /* ipv4 addr of the socket is invalid. Only the
27654 + * unspecified and mapped address have a v4 equivalent.
27655 + */
27656 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/route.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/route.c
27657 +--- linux-2.6.29.4/net/ipv6/route.c 2009-03-24 14:22:47.000000000 +0100
27658 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/route.c 2009-03-24 14:48:37.000000000 +0100
27659 +@@ -2254,7 +2254,8 @@ static int rt6_fill_node(struct net *net
27660 + struct inet6_dev *idev = ip6_dst_idev(&rt->u.dst);
27661 + struct in6_addr saddr_buf;
27662 + if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
27663 +- dst, 0, &saddr_buf) == 0)
27664 ++ dst, 0, &saddr_buf,
27665 ++ (skb->sk ? skb->sk->sk_nx_info : NULL)) == 0)
27666 + NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
27667 + }
27668 +
27669 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/tcp_ipv6.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/tcp_ipv6.c
27670 +--- linux-2.6.29.4/net/ipv6/tcp_ipv6.c 2009-03-24 14:22:47.000000000 +0100
27671 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/tcp_ipv6.c 2009-03-24 14:48:37.000000000 +0100
27672 +@@ -68,6 +68,7 @@
27673 +
27674 + #include <linux/crypto.h>
27675 + #include <linux/scatterlist.h>
27676 ++#include <linux/vs_inet6.h>
27677 +
27678 + static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
27679 + static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
27680 +@@ -156,8 +157,15 @@ static int tcp_v6_connect(struct sock *s
27681 + * connect() to INADDR_ANY means loopback (BSD'ism).
27682 + */
27683 +
27684 +- if(ipv6_addr_any(&usin->sin6_addr))
27685 +- usin->sin6_addr.s6_addr[15] = 0x1;
27686 ++ if(ipv6_addr_any(&usin->sin6_addr)) {
27687 ++ struct nx_info *nxi = sk->sk_nx_info;
27688 ++
27689 ++ if (nxi && nx_info_has_v6(nxi))
27690 ++ /* FIXME: remap lback? */
27691 ++ usin->sin6_addr = nxi->v6.ip;
27692 ++ else
27693 ++ usin->sin6_addr.s6_addr[15] = 0x1;
27694 ++ }
27695 +
27696 + addr_type = ipv6_addr_type(&usin->sin6_addr);
27697 +
27698 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/udp.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/udp.c
27699 +--- linux-2.6.29.4/net/ipv6/udp.c 2009-03-24 14:22:47.000000000 +0100
27700 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/udp.c 2009-03-24 20:56:49.000000000 +0100
27701 +@@ -47,6 +47,7 @@
27702 +
27703 + #include <linux/proc_fs.h>
27704 + #include <linux/seq_file.h>
27705 ++#include <linux/vs_inet6.h>
27706 + #include "udp_impl.h"
27707 +
27708 + int udp_v6_get_port(struct sock *sk, unsigned short snum)
27709 +@@ -77,6 +78,10 @@ static inline int compute_score(struct s
27710 + if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
27711 + return -1;
27712 + score++;
27713 ++ } else {
27714 ++ /* block non nx_info ips */
27715 ++ if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
27716 ++ return -1;
27717 + }
27718 + if (!ipv6_addr_any(&np->daddr)) {
27719 + if (!ipv6_addr_equal(&np->daddr, saddr))
27720 +diff -NurpP --minimal linux-2.6.29.4/net/ipv6/xfrm6_policy.c linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/xfrm6_policy.c
27721 +--- linux-2.6.29.4/net/ipv6/xfrm6_policy.c 2009-03-24 14:22:47.000000000 +0100
27722 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/ipv6/xfrm6_policy.c 2009-03-24 14:48:37.000000000 +0100
27723 +@@ -63,7 +63,7 @@ static int xfrm6_get_saddr(struct net *n
27724 + dev = ip6_dst_idev(dst)->dev;
27725 + ipv6_dev_get_saddr(dev_net(dev), dev,
27726 + (struct in6_addr *)&daddr->a6, 0,
27727 +- (struct in6_addr *)&saddr->a6);
27728 ++ (struct in6_addr *)&saddr->a6, NULL);
27729 + dst_release(dst);
27730 + return 0;
27731 + }
27732 +diff -NurpP --minimal linux-2.6.29.4/net/netlink/af_netlink.c linux-2.6.29.4-vs2.3.0.36.14/net/netlink/af_netlink.c
27733 +--- linux-2.6.29.4/net/netlink/af_netlink.c 2009-03-24 14:22:47.000000000 +0100
27734 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/netlink/af_netlink.c 2009-03-24 14:48:37.000000000 +0100
27735 +@@ -55,6 +55,9 @@
27736 + #include <linux/types.h>
27737 + #include <linux/audit.h>
27738 + #include <linux/mutex.h>
27739 ++#include <linux/vs_context.h>
27740 ++#include <linux/vs_network.h>
27741 ++#include <linux/vs_limit.h>
27742 +
27743 + #include <net/net_namespace.h>
27744 + #include <net/sock.h>
27745 +@@ -1776,6 +1779,8 @@ static struct sock *netlink_seq_socket_i
27746 + sk_for_each(s, node, &hash->table[j]) {
27747 + if (sock_net(s) != seq_file_net(seq))
27748 + continue;
27749 ++ if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
27750 ++ continue;
27751 + if (off == pos) {
27752 + iter->link = i;
27753 + iter->hash_idx = j;
27754 +@@ -1810,7 +1815,8 @@ static void *netlink_seq_next(struct seq
27755 + s = v;
27756 + do {
27757 + s = sk_next(s);
27758 +- } while (s && sock_net(s) != seq_file_net(seq));
27759 ++ } while (s && (sock_net(s) != seq_file_net(seq) ||
27760 ++ !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)));
27761 + if (s)
27762 + return s;
27763 +
27764 +@@ -1822,7 +1828,8 @@ static void *netlink_seq_next(struct seq
27765 +
27766 + for (; j <= hash->mask; j++) {
27767 + s = sk_head(&hash->table[j]);
27768 +- while (s && sock_net(s) != seq_file_net(seq))
27769 ++ while (s && (sock_net(s) != seq_file_net(seq) ||
27770 ++ !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)))
27771 + s = sk_next(s);
27772 + if (s) {
27773 + iter->link = i;
27774 +diff -NurpP --minimal linux-2.6.29.4/net/sctp/ipv6.c linux-2.6.29.4-vs2.3.0.36.14/net/sctp/ipv6.c
27775 +--- linux-2.6.29.4/net/sctp/ipv6.c 2009-03-24 14:22:48.000000000 +0100
27776 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/sctp/ipv6.c 2009-03-24 20:58:14.000000000 +0100
27777 +@@ -317,7 +317,8 @@ static void sctp_v6_get_saddr(struct sct
27778 + dst ? ip6_dst_idev(dst)->dev : NULL,
27779 + &daddr->v6.sin6_addr,
27780 + inet6_sk(&sk->inet.sk)->srcprefs,
27781 +- &saddr->v6.sin6_addr);
27782 ++ &saddr->v6.sin6_addr,
27783 ++ asoc->base.sk->sk_nx_info);
27784 + SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: %pI6\n",
27785 + &saddr->v6.sin6_addr);
27786 + return;
27787 +diff -NurpP --minimal linux-2.6.29.4/net/socket.c linux-2.6.29.4-vs2.3.0.36.14/net/socket.c
27788 +--- linux-2.6.29.4/net/socket.c 2009-03-24 14:22:48.000000000 +0100
27789 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/socket.c 2009-03-24 14:48:37.000000000 +0100
27790 +@@ -95,6 +95,10 @@
27791 +
27792 + #include <net/sock.h>
27793 + #include <linux/netfilter.h>
27794 ++#include <linux/vs_base.h>
27795 ++#include <linux/vs_socket.h>
27796 ++#include <linux/vs_inet.h>
27797 ++#include <linux/vs_inet6.h>
27798 +
27799 + static int sock_no_open(struct inode *irrelevant, struct file *dontcare);
27800 + static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
27801 +@@ -549,7 +553,7 @@ static inline int __sock_sendmsg(struct
27802 + struct msghdr *msg, size_t size)
27803 + {
27804 + struct sock_iocb *si = kiocb_to_siocb(iocb);
27805 +- int err;
27806 ++ int err, len;
27807 +
27808 + si->sock = sock;
27809 + si->scm = NULL;
27810 +@@ -560,7 +564,22 @@ static inline int __sock_sendmsg(struct
27811 + if (err)
27812 + return err;
27813 +
27814 +- return sock->ops->sendmsg(iocb, sock, msg, size);
27815 ++ len = sock->ops->sendmsg(iocb, sock, msg, size);
27816 ++ if (sock->sk) {
27817 ++ if (len == size)
27818 ++ vx_sock_send(sock->sk, size);
27819 ++ else
27820 ++ vx_sock_fail(sock->sk, size);
27821 ++ }
27822 ++ vxdprintk(VXD_CBIT(net, 7),
27823 ++ "__sock_sendmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
27824 ++ sock, sock->sk,
27825 ++ (sock->sk)?sock->sk->sk_nx_info:0,
27826 ++ (sock->sk)?sock->sk->sk_vx_info:0,
27827 ++ (sock->sk)?sock->sk->sk_xid:0,
27828 ++ (sock->sk)?sock->sk->sk_nid:0,
27829 ++ (unsigned int)size, len);
27830 ++ return len;
27831 + }
27832 +
27833 + int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
27834 +@@ -629,7 +648,7 @@ EXPORT_SYMBOL_GPL(__sock_recv_timestamp)
27835 + static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock,
27836 + struct msghdr *msg, size_t size, int flags)
27837 + {
27838 +- int err;
27839 ++ int err, len;
27840 + struct sock_iocb *si = kiocb_to_siocb(iocb);
27841 +
27842 + si->sock = sock;
27843 +@@ -642,7 +661,18 @@ static inline int __sock_recvmsg(struct
27844 + if (err)
27845 + return err;
27846 +
27847 +- return sock->ops->recvmsg(iocb, sock, msg, size, flags);
27848 ++ len = sock->ops->recvmsg(iocb, sock, msg, size, flags);
27849 ++ if ((len >= 0) && sock->sk)
27850 ++ vx_sock_recv(sock->sk, len);
27851 ++ vxdprintk(VXD_CBIT(net, 7),
27852 ++ "__sock_recvmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
27853 ++ sock, sock->sk,
27854 ++ (sock->sk)?sock->sk->sk_nx_info:0,
27855 ++ (sock->sk)?sock->sk->sk_vx_info:0,
27856 ++ (sock->sk)?sock->sk->sk_xid:0,
27857 ++ (sock->sk)?sock->sk->sk_nid:0,
27858 ++ (unsigned int)size, len);
27859 ++ return len;
27860 + }
27861 +
27862 + int sock_recvmsg(struct socket *sock, struct msghdr *msg,
27863 +@@ -1106,6 +1136,13 @@ static int __sock_create(struct net *net
27864 + if (type < 0 || type >= SOCK_MAX)
27865 + return -EINVAL;
27866 +
27867 ++ if (!nx_check(0, VS_ADMIN)) {
27868 ++ if (family == PF_INET && !current_nx_info_has_v4())
27869 ++ return -EAFNOSUPPORT;
27870 ++ if (family == PF_INET6 && !current_nx_info_has_v6())
27871 ++ return -EAFNOSUPPORT;
27872 ++ }
27873 ++
27874 + /* Compatibility.
27875 +
27876 + This uglymoron is moved from INET layer to here to avoid
27877 +@@ -1238,6 +1275,7 @@ SYSCALL_DEFINE3(socket, int, family, int
27878 + if (retval < 0)
27879 + goto out;
27880 +
27881 ++ set_bit(SOCK_USER_SOCKET, &sock->flags);
27882 + retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
27883 + if (retval < 0)
27884 + goto out_release;
27885 +@@ -1279,10 +1317,12 @@ SYSCALL_DEFINE4(socketpair, int, family,
27886 + err = sock_create(family, type, protocol, &sock1);
27887 + if (err < 0)
27888 + goto out;
27889 ++ set_bit(SOCK_USER_SOCKET, &sock1->flags);
27890 +
27891 + err = sock_create(family, type, protocol, &sock2);
27892 + if (err < 0)
27893 + goto out_release_1;
27894 ++ set_bit(SOCK_USER_SOCKET, &sock2->flags);
27895 +
27896 + err = sock1->ops->socketpair(sock1, sock2);
27897 + if (err < 0)
27898 +diff -NurpP --minimal linux-2.6.29.4/net/sunrpc/auth.c linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/auth.c
27899 +--- linux-2.6.29.4/net/sunrpc/auth.c 2009-03-24 14:22:48.000000000 +0100
27900 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/auth.c 2009-03-24 21:00:49.000000000 +0100
27901 +@@ -14,6 +14,7 @@
27902 + #include <linux/hash.h>
27903 + #include <linux/sunrpc/clnt.h>
27904 + #include <linux/spinlock.h>
27905 ++#include <linux/vs_tag.h>
27906 +
27907 + #ifdef RPC_DEBUG
27908 + # define RPCDBG_FACILITY RPCDBG_AUTH
27909 +@@ -360,6 +361,7 @@ rpcauth_lookupcred(struct rpc_auth *auth
27910 + memset(&acred, 0, sizeof(acred));
27911 + acred.uid = cred->fsuid;
27912 + acred.gid = cred->fsgid;
27913 ++ acred.tag = dx_current_tag();
27914 + acred.group_info = get_group_info(((struct cred *)cred)->group_info);
27915 +
27916 + ret = auth->au_ops->lookup_cred(auth, &acred, flags);
27917 +@@ -400,6 +402,7 @@ rpcauth_bind_root_cred(struct rpc_task *
27918 + struct auth_cred acred = {
27919 + .uid = 0,
27920 + .gid = 0,
27921 ++ .tag = dx_current_tag(),
27922 + };
27923 + struct rpc_cred *ret;
27924 +
27925 +diff -NurpP --minimal linux-2.6.29.4/net/sunrpc/auth_unix.c linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/auth_unix.c
27926 +--- linux-2.6.29.4/net/sunrpc/auth_unix.c 2008-12-25 00:26:37.000000000 +0100
27927 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/auth_unix.c 2009-02-22 22:54:26.000000000 +0100
27928 +@@ -11,12 +11,14 @@
27929 + #include <linux/module.h>
27930 + #include <linux/sunrpc/clnt.h>
27931 + #include <linux/sunrpc/auth.h>
27932 ++#include <linux/vs_tag.h>
27933 +
27934 + #define NFS_NGROUPS 16
27935 +
27936 + struct unx_cred {
27937 + struct rpc_cred uc_base;
27938 + gid_t uc_gid;
27939 ++ tag_t uc_tag;
27940 + gid_t uc_gids[NFS_NGROUPS];
27941 + };
27942 + #define uc_uid uc_base.cr_uid
27943 +@@ -78,6 +80,7 @@ unx_create_cred(struct rpc_auth *auth, s
27944 + groups = NFS_NGROUPS;
27945 +
27946 + cred->uc_gid = acred->gid;
27947 ++ cred->uc_tag = acred->tag;
27948 + for (i = 0; i < groups; i++)
27949 + cred->uc_gids[i] = GROUP_AT(acred->group_info, i);
27950 + if (i < NFS_NGROUPS)
27951 +@@ -119,7 +122,9 @@ unx_match(struct auth_cred *acred, struc
27952 + unsigned int i;
27953 +
27954 +
27955 +- if (cred->uc_uid != acred->uid || cred->uc_gid != acred->gid)
27956 ++ if (cred->uc_uid != acred->uid ||
27957 ++ cred->uc_gid != acred->gid ||
27958 ++ cred->uc_tag != acred->tag)
27959 + return 0;
27960 +
27961 + if (acred->group_info != NULL)
27962 +@@ -142,7 +147,7 @@ unx_marshal(struct rpc_task *task, __be3
27963 + struct rpc_clnt *clnt = task->tk_client;
27964 + struct unx_cred *cred = container_of(task->tk_msg.rpc_cred, struct unx_cred, uc_base);
27965 + __be32 *base, *hold;
27966 +- int i;
27967 ++ int i, tag;
27968 +
27969 + *p++ = htonl(RPC_AUTH_UNIX);
27970 + base = p++;
27971 +@@ -152,9 +157,12 @@ unx_marshal(struct rpc_task *task, __be3
27972 + * Copy the UTS nodename captured when the client was created.
27973 + */
27974 + p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen);
27975 ++ tag = task->tk_client->cl_tag;
27976 +
27977 +- *p++ = htonl((u32) cred->uc_uid);
27978 +- *p++ = htonl((u32) cred->uc_gid);
27979 ++ *p++ = htonl((u32) TAGINO_UID(tag,
27980 ++ cred->uc_uid, cred->uc_tag));
27981 ++ *p++ = htonl((u32) TAGINO_GID(tag,
27982 ++ cred->uc_gid, cred->uc_tag));
27983 + hold = p++;
27984 + for (i = 0; i < 16 && cred->uc_gids[i] != (gid_t) NOGROUP; i++)
27985 + *p++ = htonl((u32) cred->uc_gids[i]);
27986 +diff -NurpP --minimal linux-2.6.29.4/net/sunrpc/clnt.c linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/clnt.c
27987 +--- linux-2.6.29.4/net/sunrpc/clnt.c 2009-03-24 14:22:48.000000000 +0100
27988 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/sunrpc/clnt.c 2009-03-24 14:48:37.000000000 +0100
27989 +@@ -32,6 +32,7 @@
27990 + #include <linux/utsname.h>
27991 + #include <linux/workqueue.h>
27992 + #include <linux/in6.h>
27993 ++#include <linux/vs_cvirt.h>
27994 +
27995 + #include <linux/sunrpc/clnt.h>
27996 + #include <linux/sunrpc/rpc_pipe_fs.h>
27997 +@@ -335,6 +336,9 @@ struct rpc_clnt *rpc_create(struct rpc_c
27998 + if (!(args->flags & RPC_CLNT_CREATE_QUIET))
27999 + clnt->cl_chatty = 1;
28000 +
28001 ++ /* TODO: handle RPC_CLNT_CREATE_TAGGED
28002 ++ if (args->flags & RPC_CLNT_CREATE_TAGGED)
28003 ++ clnt->cl_tag = 1; */
28004 + return clnt;
28005 + }
28006 + EXPORT_SYMBOL_GPL(rpc_create);
28007 +diff -NurpP --minimal linux-2.6.29.4/net/unix/af_unix.c linux-2.6.29.4-vs2.3.0.36.14/net/unix/af_unix.c
28008 +--- linux-2.6.29.4/net/unix/af_unix.c 2009-03-24 14:22:48.000000000 +0100
28009 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/unix/af_unix.c 2009-03-24 14:48:37.000000000 +0100
28010 +@@ -114,6 +114,8 @@
28011 + #include <linux/mount.h>
28012 + #include <net/checksum.h>
28013 + #include <linux/security.h>
28014 ++#include <linux/vs_context.h>
28015 ++#include <linux/vs_limit.h>
28016 +
28017 + static struct hlist_head unix_socket_table[UNIX_HASH_SIZE + 1];
28018 + static DEFINE_SPINLOCK(unix_table_lock);
28019 +@@ -258,6 +260,8 @@ static struct sock *__unix_find_socket_b
28020 + if (!net_eq(sock_net(s), net))
28021 + continue;
28022 +
28023 ++ if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
28024 ++ continue;
28025 + if (u->addr->len == len &&
28026 + !memcmp(u->addr->name, sunname, len))
28027 + goto found;
28028 +@@ -2110,6 +2114,8 @@ static struct sock *unix_seq_idx(struct
28029 + for (s = first_unix_socket(&iter->i); s; s = next_unix_socket(&iter->i, s)) {
28030 + if (sock_net(s) != seq_file_net(seq))
28031 + continue;
28032 ++ if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
28033 ++ continue;
28034 + if (off == pos)
28035 + return s;
28036 + ++off;
28037 +@@ -2134,7 +2140,8 @@ static void *unix_seq_next(struct seq_fi
28038 + sk = first_unix_socket(&iter->i);
28039 + else
28040 + sk = next_unix_socket(&iter->i, sk);
28041 +- while (sk && (sock_net(sk) != seq_file_net(seq)))
28042 ++ while (sk && (sock_net(sk) != seq_file_net(seq) ||
28043 ++ !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)))
28044 + sk = next_unix_socket(&iter->i, sk);
28045 + return sk;
28046 + }
28047 +diff -NurpP --minimal linux-2.6.29.4/net/x25/af_x25.c linux-2.6.29.4-vs2.3.0.36.14/net/x25/af_x25.c
28048 +--- linux-2.6.29.4/net/x25/af_x25.c 2009-05-23 23:16:53.000000000 +0200
28049 ++++ linux-2.6.29.4-vs2.3.0.36.14/net/x25/af_x25.c 2009-04-30 12:14:53.000000000 +0200
28050 +@@ -506,7 +506,10 @@ static int x25_create(struct net *net, s
28051 +
28052 + x25 = x25_sk(sk);
28053 +
28054 +- sock_init_data(sock, sk);
28055 ++ sk->sk_socket = sock;
28056 ++ sk->sk_type = sock->type;
28057 ++ sk->sk_sleep = &sock->wait;
28058 ++ sock->sk = sk;
28059 +
28060 + x25_init_timers(sk);
28061 +
28062 +diff -NurpP --minimal linux-2.6.29.4/scripts/checksyscalls.sh linux-2.6.29.4-vs2.3.0.36.14/scripts/checksyscalls.sh
28063 +--- linux-2.6.29.4/scripts/checksyscalls.sh 2008-12-25 00:26:37.000000000 +0100
28064 ++++ linux-2.6.29.4-vs2.3.0.36.14/scripts/checksyscalls.sh 2009-02-22 22:54:26.000000000 +0100
28065 +@@ -108,7 +108,6 @@ cat << EOF
28066 + #define __IGNORE_afs_syscall
28067 + #define __IGNORE_getpmsg
28068 + #define __IGNORE_putpmsg
28069 +-#define __IGNORE_vserver
28070 + EOF
28071 + }
28072 +
28073 +diff -NurpP --minimal linux-2.6.29.4/security/commoncap.c linux-2.6.29.4-vs2.3.0.36.14/security/commoncap.c
28074 +--- linux-2.6.29.4/security/commoncap.c 2009-05-23 23:16:53.000000000 +0200
28075 ++++ linux-2.6.29.4-vs2.3.0.36.14/security/commoncap.c 2009-04-30 12:14:53.000000000 +0200
28076 +@@ -27,10 +27,11 @@
28077 + #include <linux/sched.h>
28078 + #include <linux/prctl.h>
28079 + #include <linux/securebits.h>
28080 ++#include <linux/vs_context.h>
28081 +
28082 + int cap_netlink_send(struct sock *sk, struct sk_buff *skb)
28083 + {
28084 +- NETLINK_CB(skb).eff_cap = current_cap();
28085 ++ NETLINK_CB(skb).eff_cap = vx_mbcaps(current_cap());
28086 + return 0;
28087 + }
28088 +
28089 +@@ -40,6 +41,7 @@ int cap_netlink_recv(struct sk_buff *skb
28090 + return -EPERM;
28091 + return 0;
28092 + }
28093 ++
28094 + EXPORT_SYMBOL(cap_netlink_recv);
28095 +
28096 + /**
28097 +@@ -60,7 +62,22 @@ EXPORT_SYMBOL(cap_netlink_recv);
28098 + int cap_capable(struct task_struct *tsk, const struct cred *cred, int cap,
28099 + int audit)
28100 + {
28101 +- return cap_raised(cred->cap_effective, cap) ? 0 : -EPERM;
28102 ++ struct vx_info *vxi = tsk->vx_info;
28103 ++
28104 ++#if 0
28105 ++ printk("cap_capable() VXF_STATE_SETUP = %llx, raised = %x, eff = %08x:%08x\n",
28106 ++ vx_info_flags(vxi, VXF_STATE_SETUP, 0),
28107 ++ cap_raised(tsk->cap_effective, cap),
28108 ++ tsk->cap_effective.cap[1], tsk->cap_effective.cap[0]);
28109 ++#endif
28110 ++
28111 ++ /* special case SETUP */
28112 ++ if (vx_info_flags(vxi, VXF_STATE_SETUP, 0) &&
28113 ++ /* FIXME: maybe use cred instead? */
28114 ++ cap_raised(tsk->cred->cap_effective, cap))
28115 ++ return 0;
28116 ++
28117 ++ return vx_cap_raised(vxi, cred->cap_effective, cap) ? 0 : -EPERM;
28118 + }
28119 +
28120 + /**
28121 +@@ -586,7 +603,7 @@ int cap_inode_setxattr(struct dentry *de
28122 +
28123 + if (!strncmp(name, XATTR_SECURITY_PREFIX,
28124 + sizeof(XATTR_SECURITY_PREFIX) - 1) &&
28125 +- !capable(CAP_SYS_ADMIN))
28126 ++ !vx_capable(CAP_SYS_ADMIN, VXC_FS_SECURITY))
28127 + return -EPERM;
28128 + return 0;
28129 + }
28130 +@@ -930,7 +947,8 @@ error:
28131 + */
28132 + int cap_syslog(int type)
28133 + {
28134 +- if ((type != 3 && type != 10) && !capable(CAP_SYS_ADMIN))
28135 ++ if ((type != 3 && type != 10) &&
28136 ++ !vx_capable(CAP_SYS_ADMIN, VXC_SYSLOG))
28137 + return -EPERM;
28138 + return 0;
28139 + }
28140 +@@ -952,3 +970,4 @@ int cap_vm_enough_memory(struct mm_struc
28141 + cap_sys_admin = 1;
28142 + return __vm_enough_memory(mm, pages, cap_sys_admin);
28143 + }
28144 ++
28145 +diff -NurpP --minimal linux-2.6.29.4/security/selinux/hooks.c linux-2.6.29.4-vs2.3.0.36.14/security/selinux/hooks.c
28146 +--- linux-2.6.29.4/security/selinux/hooks.c 2009-05-23 23:16:53.000000000 +0200
28147 ++++ linux-2.6.29.4-vs2.3.0.36.14/security/selinux/hooks.c 2009-05-23 23:19:11.000000000 +0200
28148 +@@ -64,7 +64,6 @@
28149 + #include <linux/dccp.h>
28150 + #include <linux/quota.h>
28151 + #include <linux/un.h> /* for Unix socket types */
28152 +-#include <net/af_unix.h> /* for Unix socket types */
28153 + #include <linux/parser.h>
28154 + #include <linux/nfs_mount.h>
28155 + #include <net/ipv6.h>
Report Message
Find on MARC Find on Google Groups