kernel-vserver-2.3.patch

   1 diff -NurpP --minimal linux-2.6.36/Documentation/vserver/debug.txt linux-2.6.36-vs2.3.0.36.38/Documentation/vserver/debug.txt
   2 --- linux-2.6.36/Documentation/vserver/debug.txt        1970-01-01 01:00:00.000000000 +0100
   3 +++ linux-2.6.36-vs2.3.0.36.38/Documentation/vserver/debug.txt  2010-10-21 13:09:36.000000000 +0200
   4 @@ -0,0 +1,154 @@
   5 +
   6 +debug_cvirt:
   7 +
   8 + 2   4 "vx_map_tgid: %p/%llx: %d -> %d"
   9 +       "vx_rmap_tgid: %p/%llx: %d -> %d"
  10 +
  11 +debug_dlim:
  12 +
  13 + 0   1 "ALLOC (%p,#%d)%c inode (%d)"
  14 +       "FREE  (%p,#%d)%c inode"
  15 + 1   2 "ALLOC (%p,#%d)%c %lld bytes (%d)"
  16 +       "FREE  (%p,#%d)%c %lld bytes"
  17 + 2   4 "ADJUST: %lld,%lld on %ld,%ld [mult=%d]"
  18 + 3   8 "ext3_has_free_blocks(%p): %lu<%lu+1, %c, %u!=%u r=%d"
  19 +       "ext3_has_free_blocks(%p): free=%lu, root=%lu"
  20 +       "rcu_free_dl_info(%p)"
  21 + 4  10 "alloc_dl_info(%p,%d) = %p"
  22 +       "dealloc_dl_info(%p)"
  23 +       "get_dl_info(%p[#%d.%d])"
  24 +       "put_dl_info(%p[#%d.%d])"
  25 + 5  20 "alloc_dl_info(%p,%d)*"
  26 + 6  40 "__hash_dl_info: %p[#%d]"
  27 +       "__unhash_dl_info: %p[#%d]"
  28 + 7  80 "locate_dl_info(%p,#%d) = %p"
  29 +
  30 +debug_misc:
  31 +
  32 + 0   1 "destroy_dqhash: %p [#0x%08x] c=%d"
  33 +       "new_dqhash: %p [#0x%08x]"
  34 +       "vroot[%d]_clr_dev: dev=%p[%lu,%d:%d]"
  35 +       "vroot[%d]_get_real_bdev: dev=%p[%lu,%d:%d]"
  36 +       "vroot[%d]_set_dev: dev=%p[%lu,%d:%d]"
  37 +       "vroot_get_real_bdev not set"
  38 + 1   2 "cow_break_link(»%s«)"
  39 +       "temp copy »%s«"
  40 + 2   4 "dentry_open(new): %p"
  41 +       "dentry_open(old): %p"
  42 +       "lookup_create(new): %p"
  43 +       "old path »%s«"
  44 +       "path_lookup(old): %d"
  45 +       "vfs_create(new): %d"
  46 +       "vfs_rename: %d"
  47 +       "vfs_sendfile: %d"
  48 + 3   8 "fput(new_file=%p[#%d])"
  49 +       "fput(old_file=%p[#%d])"
  50 + 4  10 "vx_info_kill(%p[#%d],%d,%d) = %d"
  51 +       "vx_info_kill(%p[#%d],%d,%d)*"
  52 + 5  20 "vs_reboot(%p[#%d],%d)"
  53 + 6  40 "dropping task %p[#%u,%u] for %p[#%u,%u]"
  54 +
  55 +debug_net:
  56 +
  57 + 2   4 "nx_addr_conflict(%p,%p) %d.%d,%d.%d"
  58 + 3   8 "inet_bind(%p) %d.%d.%d.%d, %d.%d.%d.%d, %d.%d.%d.%d"
  59 +       "inet_bind(%p)* %p,%p;%lx %d.%d.%d.%d"
  60 + 4  10 "ip_route_connect(%p) %p,%p;%lx"
  61 + 5  20 "__addr_in_socket(%p,%d.%d.%d.%d) %p:%d.%d.%d.%d %p;%lx"
  62 + 6  40 "sk,egf: %p [#%d] (from %d)"
  63 +       "sk,egn: %p [#%d] (from %d)"
  64 +       "sk,req: %p [#%d] (from %d)"
  65 +       "sk: %p [#%d] (from %d)"
  66 +       "tw: %p [#%d] (from %d)"
  67 + 7  80 "__sock_recvmsg: %p[%p,%p,%p;%d]:%d/%d"
  68 +       "__sock_sendmsg: %p[%p,%p,%p;%d]:%d/%d"
  69 +
  70 +debug_nid:
  71 +
  72 + 0   1 "__lookup_nx_info(#%u): %p[#%u]"
  73 +       "alloc_nx_info(%d) = %p"
  74 +       "create_nx_info(%d) (dynamic rejected)"
  75 +       "create_nx_info(%d) = %p (already there)"
  76 +       "create_nx_info(%d) = %p (new)"
  77 +       "dealloc_nx_info(%p)"
  78 + 1   2 "alloc_nx_info(%d)*"
  79 +       "create_nx_info(%d)*"
  80 + 2   4 "get_nx_info(%p[#%d.%d])"
  81 +       "put_nx_info(%p[#%d.%d])"
  82 + 3   8 "claim_nx_info(%p[#%d.%d.%d]) %p"
  83 +       "clr_nx_info(%p[#%d.%d])"
  84 +       "init_nx_info(%p[#%d.%d])"
  85 +       "release_nx_info(%p[#%d.%d.%d]) %p"
  86 +       "set_nx_info(%p[#%d.%d])"
  87 + 4  10 "__hash_nx_info: %p[#%d]"
  88 +       "__nx_dynamic_id: [#%d]"
  89 +       "__unhash_nx_info: %p[#%d.%d.%d]"
  90 + 5  20 "moved task %p into nxi:%p[#%d]"
  91 +       "nx_migrate_task(%p,%p[#%d.%d.%d])"
  92 +       "task_get_nx_info(%p)"
  93 + 6  40 "nx_clear_persistent(%p[#%d])"
  94 +
  95 +debug_quota:
  96 +
  97 + 0   1 "quota_sync_dqh(%p,%d) discard inode %p"
  98 + 1   2 "quota_sync_dqh(%p,%d)"
  99 +       "sync_dquots(%p,%d)"
 100 +       "sync_dquots_dqh(%p,%d)"
 101 + 3   8 "do_quotactl(%p,%d,cmd=%d,id=%d,%p)"
 102 +
 103 +debug_switch:
 104 +
 105 + 0   1 "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]"
 106 + 1   2 "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]"
 107 + 4  10 "%s: (%s %s) returned %s with %d"
 108 +
 109 +debug_tag:
 110 +
 111 + 7  80 "dx_parse_tag(»%s«): %d:#%d"
 112 +       "dx_propagate_tag(%p[#%lu.%d]): %d,%d"
 113 +
 114 +debug_xid:
 115 +
 116 + 0   1 "__lookup_vx_info(#%u): %p[#%u]"
 117 +       "alloc_vx_info(%d) = %p"
 118 +       "alloc_vx_info(%d)*"
 119 +       "create_vx_info(%d) (dynamic rejected)"
 120 +       "create_vx_info(%d) = %p (already there)"
 121 +       "create_vx_info(%d) = %p (new)"
 122 +       "dealloc_vx_info(%p)"
 123 +       "loc_vx_info(%d) = %p (found)"
 124 +       "loc_vx_info(%d) = %p (new)"
 125 +       "loc_vx_info(%d) = %p (not available)"
 126 + 1   2 "create_vx_info(%d)*"
 127 +       "loc_vx_info(%d)*"
 128 + 2   4 "get_vx_info(%p[#%d.%d])"
 129 +       "put_vx_info(%p[#%d.%d])"
 130 + 3   8 "claim_vx_info(%p[#%d.%d.%d]) %p"
 131 +       "clr_vx_info(%p[#%d.%d])"
 132 +       "init_vx_info(%p[#%d.%d])"
 133 +       "release_vx_info(%p[#%d.%d.%d]) %p"
 134 +       "set_vx_info(%p[#%d.%d])"
 135 + 4  10 "__hash_vx_info: %p[#%d]"
 136 +       "__unhash_vx_info: %p[#%d.%d.%d]"
 137 +       "__vx_dynamic_id: [#%d]"
 138 + 5  20 "enter_vx_info(%p[#%d],%p) %p[#%d,%p]"
 139 +       "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]"
 140 +       "moved task %p into vxi:%p[#%d]"
 141 +       "task_get_vx_info(%p)"
 142 +       "vx_migrate_task(%p,%p[#%d.%d])"
 143 + 6  40 "vx_clear_persistent(%p[#%d])"
 144 +       "vx_exit_init(%p[#%d],%p[#%d,%d,%d])"
 145 +       "vx_set_init(%p[#%d],%p[#%d,%d,%d])"
 146 +       "vx_set_persistent(%p[#%d])"
 147 +       "vx_set_reaper(%p[#%d],%p[#%d,%d])"
 148 + 7  80 "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]"
 149 +
 150 +
 151 +debug_limit:
 152 +
 153 + n 2^n "vx_acc_cres[%5d,%s,%2d]: %5d%s"
 154 +       "vx_cres_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
 155 +
 156 + m 2^m "vx_acc_page[%5d,%s,%2d]: %5d%s"
 157 +       "vx_acc_pages[%5d,%s,%2d]: %5d += %5d"
 158 +       "vx_pages_avail[%5d,%s,%2d]: %5ld > %5d + %5d"
 159 diff -NurpP --minimal linux-2.6.36/arch/alpha/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/alpha/Kconfig
 160 --- linux-2.6.36/arch/alpha/Kconfig     2010-10-21 13:06:45.000000000 +0200
 161 +++ linux-2.6.36-vs2.3.0.36.38/arch/alpha/Kconfig       2010-10-21 13:09:36.000000000 +0200
 162 @@ -677,6 +677,8 @@ config DUMMY_CONSOLE
 163         depends on VGA_HOSE
 164         default y
 165
 166 +source "kernel/vserver/Kconfig"
 167 +
 168  source "security/Kconfig"
 169
 170  source "crypto/Kconfig"
 171 diff -NurpP --minimal linux-2.6.36/arch/alpha/kernel/entry.S linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/entry.S
 172 --- linux-2.6.36/arch/alpha/kernel/entry.S      2010-10-21 13:06:45.000000000 +0200
 173 +++ linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/entry.S        2010-10-21 13:09:36.000000000 +0200
 174 @@ -860,24 +860,15 @@ sys_getxgid:
 175         .globl  sys_getxpid
 176         .ent    sys_getxpid
 177  sys_getxpid:
 178 +       lda     $sp, -16($sp)
 179 +       stq     $26, 0($sp)
 180         .prologue 0
 181 -       ldq     $2, TI_TASK($8)
 182
 183 -       /* See linux/kernel/timer.c sys_getppid for discussion
 184 -          about this loop.  */
 185 -       ldq     $3, TASK_GROUP_LEADER($2)
 186 -       ldq     $4, TASK_REAL_PARENT($3)
 187 -       ldl     $0, TASK_TGID($2)
 188 -1:     ldl     $1, TASK_TGID($4)
 189 -#ifdef CONFIG_SMP
 190 -       mov     $4, $5
 191 -       mb
 192 -       ldq     $3, TASK_GROUP_LEADER($2)
 193 -       ldq     $4, TASK_REAL_PARENT($3)
 194 -       cmpeq   $4, $5, $5
 195 -       beq     $5, 1b
 196 -#endif
 197 -       stq     $1, 80($sp)
 198 +       lda     $16, 96($sp)
 199 +       jsr     $26, do_getxpid
 200 +       ldq     $26, 0($sp)
 201 +
 202 +       lda     $sp, 16($sp)
 203         ret
 204  .end sys_getxpid
 205
 206 diff -NurpP --minimal linux-2.6.36/arch/alpha/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/ptrace.c
 207 --- linux-2.6.36/arch/alpha/kernel/ptrace.c     2010-07-07 18:30:51.000000000 +0200
 208 +++ linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/ptrace.c       2010-10-21 13:09:36.000000000 +0200
 209 @@ -13,6 +13,7 @@
 210  #include <linux/user.h>
 211  #include <linux/security.h>
 212  #include <linux/signal.h>
 213 +#include <linux/vs_base.h>
 214
 215  #include <asm/uaccess.h>
 216  #include <asm/pgtable.h>
 217 diff -NurpP --minimal linux-2.6.36/arch/alpha/kernel/systbls.S linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/systbls.S
 218 --- linux-2.6.36/arch/alpha/kernel/systbls.S    2010-10-21 13:06:46.000000000 +0200
 219 +++ linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/systbls.S      2010-10-21 13:09:36.000000000 +0200
 220 @@ -446,7 +446,7 @@ sys_call_table:
 221         .quad sys_stat64                        /* 425 */
 222         .quad sys_lstat64
 223         .quad sys_fstat64
 224 -       .quad sys_ni_syscall                    /* sys_vserver */
 225 +       .quad sys_vserver                       /* sys_vserver */
 226         .quad sys_ni_syscall                    /* sys_mbind */
 227         .quad sys_ni_syscall                    /* sys_get_mempolicy */
 228         .quad sys_ni_syscall                    /* sys_set_mempolicy */
 229 diff -NurpP --minimal linux-2.6.36/arch/alpha/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/traps.c
 230 --- linux-2.6.36/arch/alpha/kernel/traps.c      2010-10-21 13:06:46.000000000 +0200
 231 +++ linux-2.6.36-vs2.3.0.36.38/arch/alpha/kernel/traps.c        2010-10-21 13:09:36.000000000 +0200
 232 @@ -183,7 +183,8 @@ die_if_kernel(char * str, struct pt_regs
 233  #ifdef CONFIG_SMP
 234         printk("CPU %d ", hard_smp_processor_id());
 235  #endif
 236 -       printk("%s(%d): %s %ld\n", current->comm, task_pid_nr(current), str, err);
 237 +       printk("%s(%d[#%u]): %s %ld\n", current->comm,
 238 +               task_pid_nr(current), current->xid, str, err);
 239         dik_show_regs(regs, r9_15);
 240         add_taint(TAINT_DIE);
 241         dik_show_trace((unsigned long *)(regs+1));
 242 diff -NurpP --minimal linux-2.6.36/arch/arm/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/arm/Kconfig
 243 --- linux-2.6.36/arch/arm/Kconfig       2010-10-21 13:06:46.000000000 +0200
 244 +++ linux-2.6.36-vs2.3.0.36.38/arch/arm/Kconfig 2010-10-21 13:09:36.000000000 +0200
 245 @@ -1803,6 +1803,8 @@ source "fs/Kconfig"
 246
 247  source "arch/arm/Kconfig.debug"
 248
 249 +source "kernel/vserver/Kconfig"
 250 +
 251  source "security/Kconfig"
 252
 253  source "crypto/Kconfig"
 254 diff -NurpP --minimal linux-2.6.36/arch/arm/include/asm/tlb.h linux-2.6.36-vs2.3.0.36.38/arch/arm/include/asm/tlb.h
 255 --- linux-2.6.36/arch/arm/include/asm/tlb.h     2009-09-10 15:25:15.000000000 +0200
 256 +++ linux-2.6.36-vs2.3.0.36.38/arch/arm/include/asm/tlb.h       2010-10-21 13:09:36.000000000 +0200
 257 @@ -27,6 +27,7 @@
 258
 259  #else /* !CONFIG_MMU */
 260
 261 +#include <linux/vs_memory.h>
 262  #include <asm/pgalloc.h>
 263
 264  /*
 265 diff -NurpP --minimal linux-2.6.36/arch/arm/kernel/calls.S linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/calls.S
 266 --- linux-2.6.36/arch/arm/kernel/calls.S        2010-10-21 13:06:46.000000000 +0200
 267 +++ linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/calls.S  2010-10-21 13:09:36.000000000 +0200
 268 @@ -322,7 +322,7 @@
 269  /* 310 */      CALL(sys_request_key)
 270                 CALL(sys_keyctl)
 271                 CALL(ABI(sys_semtimedop, sys_oabi_semtimedop))
 272 -/* vserver */  CALL(sys_ni_syscall)
 273 +               CALL(sys_vserver)
 274                 CALL(sys_ioprio_set)
 275  /* 315 */      CALL(sys_ioprio_get)
 276                 CALL(sys_inotify_init)
 277 diff -NurpP --minimal linux-2.6.36/arch/arm/kernel/process.c linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/process.c
 278 --- linux-2.6.36/arch/arm/kernel/process.c      2010-10-21 13:06:46.000000000 +0200
 279 +++ linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/process.c        2010-10-21 13:09:36.000000000 +0200
 280 @@ -295,7 +295,8 @@ void __show_regs(struct pt_regs *regs)
 281  void show_regs(struct pt_regs * regs)
 282  {
 283         printk("\n");
 284 -       printk("Pid: %d, comm: %20s\n", task_pid_nr(current), current->comm);
 285 +       printk("Pid: %d[#%u], comm: %20s\n",
 286 +               task_pid_nr(current), current->xid, current->comm);
 287         __show_regs(regs);
 288         __backtrace();
 289  }
 290 diff -NurpP --minimal linux-2.6.36/arch/arm/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/traps.c
 291 --- linux-2.6.36/arch/arm/kernel/traps.c        2010-10-21 13:06:46.000000000 +0200
 292 +++ linux-2.6.36-vs2.3.0.36.38/arch/arm/kernel/traps.c  2010-10-21 13:09:36.000000000 +0200
 293 @@ -244,8 +244,8 @@ static int __die(const char *str, int er
 294
 295         print_modules();
 296         __show_regs(regs);
 297 -       printk(KERN_EMERG "Process %.*s (pid: %d, stack limit = 0x%p)\n",
 298 -               TASK_COMM_LEN, tsk->comm, task_pid_nr(tsk), thread + 1);
 299 +       printk(KERN_EMERG "Process %.*s (pid: %d:#%u, stack limit = 0x%p)\n",
 300 +               TASK_COMM_LEN, tsk->comm, task_pid_nr(tsk), tsk->xid, thread + 1);
 301
 302         if (!user_mode(regs) || in_interrupt()) {
 303                 dump_mem(KERN_EMERG, "Stack: ", regs->ARM_sp,
 304 diff -NurpP --minimal linux-2.6.36/arch/cris/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/cris/Kconfig
 305 --- linux-2.6.36/arch/cris/Kconfig      2010-10-21 13:06:51.000000000 +0200
 306 +++ linux-2.6.36-vs2.3.0.36.38/arch/cris/Kconfig        2010-10-21 13:09:36.000000000 +0200
 307 @@ -694,6 +694,8 @@ source "drivers/staging/Kconfig"
 308
 309  source "arch/cris/Kconfig.debug"
 310
 311 +source "kernel/vserver/Kconfig"
 312 +
 313  source "security/Kconfig"
 314
 315  source "crypto/Kconfig"
 316 diff -NurpP --minimal linux-2.6.36/arch/frv/kernel/kernel_thread.S linux-2.6.36-vs2.3.0.36.38/arch/frv/kernel/kernel_thread.S
 317 --- linux-2.6.36/arch/frv/kernel/kernel_thread.S        2008-12-25 00:26:37.000000000 +0100
 318 +++ linux-2.6.36-vs2.3.0.36.38/arch/frv/kernel/kernel_thread.S  2010-10-21 13:09:36.000000000 +0200
 319 @@ -37,7 +37,7 @@ kernel_thread:
 320
 321         # start by forking the current process, but with shared VM
 322         setlos.p        #__NR_clone,gr7         ; syscall number
 323 -       ori             gr10,#CLONE_VM,gr8      ; first syscall arg     [clone_flags]
 324 +       ori             gr10,#CLONE_KT,gr8      ; first syscall arg     [clone_flags]
 325         sethi.p         #0xe4e4,gr9             ; second syscall arg    [newsp]
 326         setlo           #0xe4e4,gr9
 327         setlos.p        #0,gr10                 ; third syscall arg     [parent_tidptr]
 328 diff -NurpP --minimal linux-2.6.36/arch/h8300/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/h8300/Kconfig
 329 --- linux-2.6.36/arch/h8300/Kconfig     2010-10-21 13:06:51.000000000 +0200
 330 +++ linux-2.6.36-vs2.3.0.36.38/arch/h8300/Kconfig       2010-10-21 13:09:36.000000000 +0200
 331 @@ -226,6 +226,8 @@ source "fs/Kconfig"
 332
 333  source "arch/h8300/Kconfig.debug"
 334
 335 +source "kernel/vserver/Kconfig"
 336 +
 337  source "security/Kconfig"
 338
 339  source "crypto/Kconfig"
 340 diff -NurpP --minimal linux-2.6.36/arch/ia64/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/ia64/Kconfig
 341 --- linux-2.6.36/arch/ia64/Kconfig      2010-10-21 13:06:52.000000000 +0200
 342 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/Kconfig        2010-10-21 13:09:36.000000000 +0200
 343 @@ -671,6 +671,8 @@ source "fs/Kconfig"
 344
 345  source "arch/ia64/Kconfig.debug"
 346
 347 +source "kernel/vserver/Kconfig"
 348 +
 349  source "security/Kconfig"
 350
 351  source "crypto/Kconfig"
 352 diff -NurpP --minimal linux-2.6.36/arch/ia64/include/asm/tlb.h linux-2.6.36-vs2.3.0.36.38/arch/ia64/include/asm/tlb.h
 353 --- linux-2.6.36/arch/ia64/include/asm/tlb.h    2010-02-25 11:51:26.000000000 +0100
 354 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/include/asm/tlb.h      2010-10-21 13:09:36.000000000 +0200
 355 @@ -40,6 +40,7 @@
 356  #include <linux/mm.h>
 357  #include <linux/pagemap.h>
 358  #include <linux/swap.h>
 359 +#include <linux/vs_memory.h>
 360
 361  #include <asm/pgalloc.h>
 362  #include <asm/processor.h>
 363 diff -NurpP --minimal linux-2.6.36/arch/ia64/kernel/entry.S linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/entry.S
 364 --- linux-2.6.36/arch/ia64/kernel/entry.S       2010-10-21 13:06:52.000000000 +0200
 365 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/entry.S 2010-10-21 13:09:36.000000000 +0200
 366 @@ -1714,7 +1714,7 @@ sys_call_table:
 367         data8 sys_mq_notify
 368         data8 sys_mq_getsetattr
 369         data8 sys_kexec_load
 370 -       data8 sys_ni_syscall                    // reserved for vserver
 371 +       data8 sys_vserver
 372         data8 sys_waitid                        // 1270
 373         data8 sys_add_key
 374         data8 sys_request_key
 375 diff -NurpP --minimal linux-2.6.36/arch/ia64/kernel/perfmon.c linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/perfmon.c
 376 --- linux-2.6.36/arch/ia64/kernel/perfmon.c     2010-10-21 13:06:52.000000000 +0200
 377 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/perfmon.c       2010-10-21 13:09:36.000000000 +0200
 378 @@ -42,6 +42,7 @@
 379  #include <linux/completion.h>
 380  #include <linux/tracehook.h>
 381  #include <linux/slab.h>
 382 +#include <linux/vs_memory.h>
 383
 384  #include <asm/errno.h>
 385  #include <asm/intrinsics.h>
 386 diff -NurpP --minimal linux-2.6.36/arch/ia64/kernel/process.c linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/process.c
 387 --- linux-2.6.36/arch/ia64/kernel/process.c     2010-10-21 13:06:52.000000000 +0200
 388 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/process.c       2010-10-21 13:09:36.000000000 +0200
 389 @@ -113,8 +113,8 @@ show_regs (struct pt_regs *regs)
 390         unsigned long ip = regs->cr_iip + ia64_psr(regs)->ri;
 391
 392         print_modules();
 393 -       printk("\nPid: %d, CPU %d, comm: %20s\n", task_pid_nr(current),
 394 -                       smp_processor_id(), current->comm);
 395 +       printk("\nPid: %d[#%u], CPU %d, comm: %20s\n", task_pid_nr(current),
 396 +                       current->xid, smp_processor_id(), current->comm);
 397         printk("psr : %016lx ifs : %016lx ip  : [<%016lx>]    %s (%s)\n",
 398                regs->cr_ipsr, regs->cr_ifs, ip, print_tainted(),
 399                init_utsname()->release);
 400 diff -NurpP --minimal linux-2.6.36/arch/ia64/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/ptrace.c
 401 --- linux-2.6.36/arch/ia64/kernel/ptrace.c      2010-08-02 16:52:04.000000000 +0200
 402 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/ptrace.c        2010-10-21 13:09:36.000000000 +0200
 403 @@ -21,6 +21,7 @@
 404  #include <linux/regset.h>
 405  #include <linux/elf.h>
 406  #include <linux/tracehook.h>
 407 +#include <linux/vs_base.h>
 408
 409  #include <asm/pgtable.h>
 410  #include <asm/processor.h>
 411 diff -NurpP --minimal linux-2.6.36/arch/ia64/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/traps.c
 412 --- linux-2.6.36/arch/ia64/kernel/traps.c       2010-07-07 18:31:01.000000000 +0200
 413 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/kernel/traps.c 2010-10-21 13:09:36.000000000 +0200
 414 @@ -59,8 +59,9 @@ die (const char *str, struct pt_regs *re
 415         put_cpu();
 416
 417         if (++die.lock_owner_depth < 3) {
 418 -               printk("%s[%d]: %s %ld [%d]\n",
 419 -               current->comm, task_pid_nr(current), str, err, ++die_counter);
 420 +               printk("%s[%d[#%u]]: %s %ld [%d]\n",
 421 +                       current->comm, task_pid_nr(current), current->xid,
 422 +                       str, err, ++die_counter);
 423                 if (notify_die(DIE_OOPS, str, regs, err, 255, SIGSEGV)
 424                     != NOTIFY_STOP)
 425                         show_regs(regs);
 426 @@ -323,8 +324,9 @@ handle_fpu_swa (int fp_fault, struct pt_
 427                         if ((last.count & 15) < 5 && (ia64_fetchadd(1, &last.count, acq) & 15) < 5) {
 428                                 last.time = current_jiffies + 5 * HZ;
 429                                 printk(KERN_WARNING
 430 -                                       "%s(%d): floating-point assist fault at ip %016lx, isr %016lx\n",
 431 -                                       current->comm, task_pid_nr(current), regs->cr_iip + ia64_psr(regs)->ri, isr);
 432 +                                       "%s(%d[#%u]): floating-point assist fault at ip %016lx, isr %016lx\n",
 433 +                                       current->comm, task_pid_nr(current), current->xid,
 434 +                                       regs->cr_iip + ia64_psr(regs)->ri, isr);
 435                         }
 436                 }
 437         }
 438 diff -NurpP --minimal linux-2.6.36/arch/ia64/mm/fault.c linux-2.6.36-vs2.3.0.36.38/arch/ia64/mm/fault.c
 439 --- linux-2.6.36/arch/ia64/mm/fault.c   2010-08-02 16:52:04.000000000 +0200
 440 +++ linux-2.6.36-vs2.3.0.36.38/arch/ia64/mm/fault.c     2010-10-21 13:09:36.000000000 +0200
 441 @@ -10,6 +10,7 @@
 442  #include <linux/interrupt.h>
 443  #include <linux/kprobes.h>
 444  #include <linux/kdebug.h>
 445 +#include <linux/vs_memory.h>
 446
 447  #include <asm/pgtable.h>
 448  #include <asm/processor.h>
 449 diff -NurpP --minimal linux-2.6.36/arch/m32r/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/m32r/kernel/traps.c
 450 --- linux-2.6.36/arch/m32r/kernel/traps.c       2009-12-03 20:01:57.000000000 +0100
 451 +++ linux-2.6.36-vs2.3.0.36.38/arch/m32r/kernel/traps.c 2010-10-21 13:09:36.000000000 +0200
 452 @@ -196,8 +196,9 @@ static void show_registers(struct pt_reg
 453         } else {
 454                 printk("SPI: %08lx\n", sp);
 455         }
 456 -       printk("Process %s (pid: %d, process nr: %d, stackpage=%08lx)",
 457 -               current->comm, task_pid_nr(current), 0xffff & i, 4096+(unsigned long)current);
 458 +       printk("Process %s (pid: %d[#%u], process nr: %d, stackpage=%08lx)",
 459 +               current->comm, task_pid_nr(current), current->xid,
 460 +               0xffff & i, 4096+(unsigned long)current);
 461
 462         /*
 463          * When in-kernel, we also print out the stack and code at the
 464 diff -NurpP --minimal linux-2.6.36/arch/m68k/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/m68k/Kconfig
 465 --- linux-2.6.36/arch/m68k/Kconfig      2010-10-21 13:06:52.000000000 +0200
 466 +++ linux-2.6.36-vs2.3.0.36.38/arch/m68k/Kconfig        2010-10-21 13:09:36.000000000 +0200
 467 @@ -616,6 +616,8 @@ source "fs/Kconfig"
 468
 469  source "arch/m68k/Kconfig.debug"
 470
 471 +source "kernel/vserver/Kconfig"
 472 +
 473  source "security/Kconfig"
 474
 475  source "crypto/Kconfig"
 476 diff -NurpP --minimal linux-2.6.36/arch/m68k/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/arch/m68k/kernel/ptrace.c
 477 --- linux-2.6.36/arch/m68k/kernel/ptrace.c      2010-07-07 18:31:02.000000000 +0200
 478 +++ linux-2.6.36-vs2.3.0.36.38/arch/m68k/kernel/ptrace.c        2010-10-21 13:09:36.000000000 +0200
 479 @@ -18,6 +18,7 @@
 480  #include <linux/ptrace.h>
 481  #include <linux/user.h>
 482  #include <linux/signal.h>
 483 +#include <linux/vs_base.h>
 484
 485  #include <asm/uaccess.h>
 486  #include <asm/page.h>
 487 @@ -254,6 +255,8 @@ long arch_ptrace(struct task_struct *chi
 488                 ret = ptrace_request(child, request, addr, data);
 489                 break;
 490         }
 491 +       if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
 492 +               goto out_tsk;
 493
 494         return ret;
 495  out_eio:
 496 diff -NurpP --minimal linux-2.6.36/arch/m68k/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/m68k/kernel/traps.c
 497 --- linux-2.6.36/arch/m68k/kernel/traps.c       2010-08-02 16:52:04.000000000 +0200
 498 +++ linux-2.6.36-vs2.3.0.36.38/arch/m68k/kernel/traps.c 2010-10-21 13:09:36.000000000 +0200
 499 @@ -906,8 +906,8 @@ void show_registers(struct pt_regs *regs
 500         printk("d4: %08lx    d5: %08lx    a0: %08lx    a1: %08lx\n",
 501                regs->d4, regs->d5, regs->a0, regs->a1);
 502
 503 -       printk("Process %s (pid: %d, task=%p)\n",
 504 -               current->comm, task_pid_nr(current), current);
 505 +       printk("Process %s (pid: %d[#%u], task=%p)\n",
 506 +               current->comm, task_pid_nr(current), current->xid, current);
 507         addr = (unsigned long)&fp->un;
 508         printk("Frame format=%X ", regs->format);
 509         switch (regs->format) {
 510 diff -NurpP --minimal linux-2.6.36/arch/m68knommu/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/m68knommu/Kconfig
 511 --- linux-2.6.36/arch/m68knommu/Kconfig 2010-10-21 13:06:53.000000000 +0200
 512 +++ linux-2.6.36-vs2.3.0.36.38/arch/m68knommu/Kconfig   2010-10-21 13:09:36.000000000 +0200
 513 @@ -730,6 +730,8 @@ source "fs/Kconfig"
 514
 515  source "arch/m68knommu/Kconfig.debug"
 516
 517 +source "kernel/vserver/Kconfig"
 518 +
 519  source "security/Kconfig"
 520
 521  source "crypto/Kconfig"
 522 diff -NurpP --minimal linux-2.6.36/arch/m68knommu/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/m68knommu/kernel/traps.c
 523 --- linux-2.6.36/arch/m68knommu/kernel/traps.c  2009-09-10 15:25:23.000000000 +0200
 524 +++ linux-2.6.36-vs2.3.0.36.38/arch/m68knommu/kernel/traps.c    2010-10-21 13:09:36.000000000 +0200
 525 @@ -78,8 +78,9 @@ void die_if_kernel(char *str, struct pt_
 526         printk(KERN_EMERG "d4: %08lx    d5: %08lx    a0: %08lx    a1: %08lx\n",
 527                fp->d4, fp->d5, fp->a0, fp->a1);
 528
 529 -       printk(KERN_EMERG "Process %s (pid: %d, stackpage=%08lx)\n",
 530 -               current->comm, current->pid, PAGE_SIZE+(unsigned long)current);
 531 +       printk(KERN_EMERG "Process %s (pid: %d[#%u], stackpage=%08lx)\n",
 532 +               current->comm, task_pid_nr(current), current->xid,
 533 +               PAGE_SIZE+(unsigned long)current);
 534         show_stack(NULL, (unsigned long *)(fp + 1));
 535         add_taint(TAINT_DIE);
 536         do_exit(SIGSEGV);
 537 diff -NurpP --minimal linux-2.6.36/arch/mips/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/mips/Kconfig
 538 --- linux-2.6.36/arch/mips/Kconfig      2010-10-21 13:06:53.000000000 +0200
 539 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/Kconfig        2010-10-21 13:09:36.000000000 +0200
 540 @@ -2298,6 +2298,8 @@ source "fs/Kconfig"
 541
 542  source "arch/mips/Kconfig.debug"
 543
 544 +source "kernel/vserver/Kconfig"
 545 +
 546  source "security/Kconfig"
 547
 548  source "crypto/Kconfig"
 549 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/ptrace.c
 550 --- linux-2.6.36/arch/mips/kernel/ptrace.c      2010-10-21 13:06:54.000000000 +0200
 551 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/ptrace.c        2010-10-21 13:09:36.000000000 +0200
 552 @@ -25,6 +25,7 @@
 553  #include <linux/security.h>
 554  #include <linux/audit.h>
 555  #include <linux/seccomp.h>
 556 +#include <linux/vs_base.h>
 557
 558  #include <asm/byteorder.h>
 559  #include <asm/cpu.h>
 560 @@ -259,6 +260,9 @@ long arch_ptrace(struct task_struct *chi
 561  {
 562         int ret;
 563
 564 +       if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
 565 +               goto out;
 566 +
 567         switch (request) {
 568         /* when I and D space are separate, these will need to be fixed. */
 569         case PTRACE_PEEKTEXT: /* read word at location addr. */
 570 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/scall32-o32.S linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall32-o32.S
 571 --- linux-2.6.36/arch/mips/kernel/scall32-o32.S 2010-10-21 13:06:54.000000000 +0200
 572 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall32-o32.S   2010-10-21 13:09:36.000000000 +0200
 573 @@ -524,7 +524,7 @@ einval:     li      v0, -ENOSYS
 574         sys     sys_mq_timedreceive     5
 575         sys     sys_mq_notify           2       /* 4275 */
 576         sys     sys_mq_getsetattr       3
 577 -       sys     sys_ni_syscall          0       /* sys_vserver */
 578 +       sys     sys_vserver             3
 579         sys     sys_waitid              5
 580         sys     sys_ni_syscall          0       /* available, was setaltroot */
 581         sys     sys_add_key             5       /* 4280 */
 582 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/scall64-64.S linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-64.S
 583 --- linux-2.6.36/arch/mips/kernel/scall64-64.S  2010-10-21 13:06:54.000000000 +0200
 584 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-64.S    2010-10-21 13:09:36.000000000 +0200
 585 @@ -363,7 +363,7 @@ sys_call_table:
 586         PTR     sys_mq_timedreceive
 587         PTR     sys_mq_notify
 588         PTR     sys_mq_getsetattr               /* 5235 */
 589 -       PTR     sys_ni_syscall                  /* sys_vserver */
 590 +       PTR     sys_vserver
 591         PTR     sys_waitid
 592         PTR     sys_ni_syscall                  /* available, was setaltroot */
 593         PTR     sys_add_key
 594 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/scall64-n32.S linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-n32.S
 595 --- linux-2.6.36/arch/mips/kernel/scall64-n32.S 2010-10-21 13:06:54.000000000 +0200
 596 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-n32.S   2010-10-21 13:09:36.000000000 +0200
 597 @@ -362,7 +362,7 @@ EXPORT(sysn32_call_table)
 598         PTR     compat_sys_mq_timedreceive
 599         PTR     compat_sys_mq_notify
 600         PTR     compat_sys_mq_getsetattr
 601 -       PTR     sys_ni_syscall                  /* 6240, sys_vserver */
 602 +       PTR     sys32_vserver                   /* 6240 */
 603         PTR     compat_sys_waitid
 604         PTR     sys_ni_syscall                  /* available, was setaltroot */
 605         PTR     sys_add_key
 606 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/scall64-o32.S linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-o32.S
 607 --- linux-2.6.36/arch/mips/kernel/scall64-o32.S 2010-10-21 13:06:54.000000000 +0200
 608 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/scall64-o32.S   2010-10-21 13:09:36.000000000 +0200
 609 @@ -481,7 +481,7 @@ sys_call_table:
 610         PTR     compat_sys_mq_timedreceive
 611         PTR     compat_sys_mq_notify            /* 4275 */
 612         PTR     compat_sys_mq_getsetattr
 613 -       PTR     sys_ni_syscall                  /* sys_vserver */
 614 +       PTR     sys32_vserver
 615         PTR     sys_32_waitid
 616         PTR     sys_ni_syscall                  /* available, was setaltroot */
 617         PTR     sys_add_key                     /* 4280 */
 618 diff -NurpP --minimal linux-2.6.36/arch/mips/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/traps.c
 619 --- linux-2.6.36/arch/mips/kernel/traps.c       2010-10-21 13:06:54.000000000 +0200
 620 +++ linux-2.6.36-vs2.3.0.36.38/arch/mips/kernel/traps.c 2010-10-21 13:09:36.000000000 +0200
 621 @@ -341,9 +341,10 @@ void show_registers(struct pt_regs *regs
 622
 623         __show_regs(regs);
 624         print_modules();
 625 -       printk("Process %s (pid: %d, threadinfo=%p, task=%p, tls=%0*lx)\n",
 626 -              current->comm, current->pid, current_thread_info(), current,
 627 -             field, current_thread_info()->tp_value);
 628 +       printk("Process %s (pid: %d:#%u, threadinfo=%p, task=%p, tls=%0*lx)\n",
 629 +               current->comm, task_pid_nr(current), current->xid,
 630 +               current_thread_info(), current,
 631 +               field, current_thread_info()->tp_value);
 632         if (cpu_has_userlocal) {
 633                 unsigned long tls;
 634
 635 diff -NurpP --minimal linux-2.6.36/arch/parisc/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/parisc/Kconfig
 636 --- linux-2.6.36/arch/parisc/Kconfig    2010-10-21 13:06:55.000000000 +0200
 637 +++ linux-2.6.36-vs2.3.0.36.38/arch/parisc/Kconfig      2010-10-21 13:09:36.000000000 +0200
 638 @@ -295,6 +295,8 @@ source "fs/Kconfig"
 639
 640  source "arch/parisc/Kconfig.debug"
 641
 642 +source "kernel/vserver/Kconfig"
 643 +
 644  source "security/Kconfig"
 645
 646  source "crypto/Kconfig"
 647 diff -NurpP --minimal linux-2.6.36/arch/parisc/kernel/syscall_table.S linux-2.6.36-vs2.3.0.36.38/arch/parisc/kernel/syscall_table.S
 648 --- linux-2.6.36/arch/parisc/kernel/syscall_table.S     2010-07-07 18:31:04.000000000 +0200
 649 +++ linux-2.6.36-vs2.3.0.36.38/arch/parisc/kernel/syscall_table.S       2010-10-21 13:09:36.000000000 +0200
 650 @@ -361,7 +361,7 @@
 651         ENTRY_COMP(mbind)               /* 260 */
 652         ENTRY_COMP(get_mempolicy)
 653         ENTRY_COMP(set_mempolicy)
 654 -       ENTRY_SAME(ni_syscall)  /* 263: reserved for vserver */
 655 +       ENTRY_DIFF(vserver)
 656         ENTRY_SAME(add_key)
 657         ENTRY_SAME(request_key)         /* 265 */
 658         ENTRY_SAME(keyctl)
 659 diff -NurpP --minimal linux-2.6.36/arch/parisc/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/parisc/kernel/traps.c
 660 --- linux-2.6.36/arch/parisc/kernel/traps.c     2009-09-10 15:25:40.000000000 +0200
 661 +++ linux-2.6.36-vs2.3.0.36.38/arch/parisc/kernel/traps.c       2010-10-21 13:09:36.000000000 +0200
 662 @@ -236,8 +236,9 @@ void die_if_kernel(char *str, struct pt_
 663                 if (err == 0)
 664                         return; /* STFU */
 665
 666 -               printk(KERN_CRIT "%s (pid %d): %s (code %ld) at " RFMT "\n",
 667 -                       current->comm, task_pid_nr(current), str, err, regs->iaoq[0]);
 668 +               printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld) at " RFMT "\n",
 669 +                       current->comm, task_pid_nr(current), current->xid,
 670 +                       str, err, regs->iaoq[0]);
 671  #ifdef PRINT_USER_FAULTS
 672                 /* XXX for debugging only */
 673                 show_regs(regs);
 674 @@ -270,8 +271,8 @@ void die_if_kernel(char *str, struct pt_
 675                 pdc_console_restart();
 676
 677         if (err)
 678 -               printk(KERN_CRIT "%s (pid %d): %s (code %ld)\n",
 679 -                       current->comm, task_pid_nr(current), str, err);
 680 +               printk(KERN_CRIT "%s (pid %d:#%u): %s (code %ld)\n",
 681 +                       current->comm, task_pid_nr(current), current->xid, str, err);
 682
 683         /* Wot's wrong wif bein' racy? */
 684         if (current->thread.flags & PARISC_KERNEL_DEATH) {
 685 diff -NurpP --minimal linux-2.6.36/arch/parisc/mm/fault.c linux-2.6.36-vs2.3.0.36.38/arch/parisc/mm/fault.c
 686 --- linux-2.6.36/arch/parisc/mm/fault.c 2010-08-02 16:52:06.000000000 +0200
 687 +++ linux-2.6.36-vs2.3.0.36.38/arch/parisc/mm/fault.c   2010-10-21 13:09:36.000000000 +0200
 688 @@ -237,8 +237,9 @@ bad_area:
 689
 690  #ifdef PRINT_USER_FAULTS
 691                 printk(KERN_DEBUG "\n");
 692 -               printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n",
 693 -                   task_pid_nr(tsk), tsk->comm, code, address);
 694 +               printk(KERN_DEBUG "do_page_fault() pid=%d:#%u "
 695 +                   "command='%s' type=%lu address=0x%08lx\n",
 696 +                   task_pid_nr(tsk), tsk->xid, tsk->comm, code, address);
 697                 if (vma) {
 698                         printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n",
 699                                         vma->vm_start, vma->vm_end);
 700 diff -NurpP --minimal linux-2.6.36/arch/powerpc/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/powerpc/Kconfig
 701 --- linux-2.6.36/arch/powerpc/Kconfig   2010-10-21 13:06:55.000000000 +0200
 702 +++ linux-2.6.36-vs2.3.0.36.38/arch/powerpc/Kconfig     2010-10-21 13:09:36.000000000 +0200
 703 @@ -974,6 +974,8 @@ source "lib/Kconfig"
 704
 705  source "arch/powerpc/Kconfig.debug"
 706
 707 +source "kernel/vserver/Kconfig"
 708 +
 709  source "security/Kconfig"
 710
 711  config KEYS_COMPAT
 712 diff -NurpP --minimal linux-2.6.36/arch/powerpc/include/asm/unistd.h linux-2.6.36-vs2.3.0.36.38/arch/powerpc/include/asm/unistd.h
 713 --- linux-2.6.36/arch/powerpc/include/asm/unistd.h      2010-10-21 13:06:56.000000000 +0200
 714 +++ linux-2.6.36-vs2.3.0.36.38/arch/powerpc/include/asm/unistd.h        2010-10-21 13:09:36.000000000 +0200
 715 @@ -275,7 +275,7 @@
 716  #endif
 717  #define __NR_rtas              255
 718  #define __NR_sys_debug_setcontext 256
 719 -/* Number 257 is reserved for vserver */
 720 +#define __NR_vserver           257
 721  #define __NR_migrate_pages     258
 722  #define __NR_mbind             259
 723  #define __NR_get_mempolicy     260
 724 diff -NurpP --minimal linux-2.6.36/arch/powerpc/kernel/process.c linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/process.c
 725 --- linux-2.6.36/arch/powerpc/kernel/process.c  2010-10-21 13:06:56.000000000 +0200
 726 +++ linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/process.c    2010-10-21 13:09:36.000000000 +0200
 727 @@ -634,8 +634,9 @@ void show_regs(struct pt_regs * regs)
 728  #else
 729                 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
 730  #endif
 731 -       printk("TASK = %p[%d] '%s' THREAD: %p",
 732 -              current, task_pid_nr(current), current->comm, task_thread_info(current));
 733 +       printk("TASK = %p[%d,#%u] '%s' THREAD: %p",
 734 +              current, task_pid_nr(current), current->xid,
 735 +              current->comm, task_thread_info(current));
 736
 737  #ifdef CONFIG_SMP
 738         printk(" CPU: %d", raw_smp_processor_id());
 739 diff -NurpP --minimal linux-2.6.36/arch/powerpc/kernel/traps.c linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/traps.c
 740 --- linux-2.6.36/arch/powerpc/kernel/traps.c    2010-10-21 13:06:57.000000000 +0200
 741 +++ linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/traps.c      2010-10-21 13:09:36.000000000 +0200
 742 @@ -1048,8 +1048,9 @@ void nonrecoverable_exception(struct pt_
 743
 744  void trace_syscall(struct pt_regs *regs)
 745  {
 746 -       printk("Task: %p(%d), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld    %s\n",
 747 -              current, task_pid_nr(current), regs->nip, regs->link, regs->gpr[0],
 748 +       printk("Task: %p(%d[#%u]), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld    %s\n",
 749 +              current, task_pid_nr(current), current->xid,
 750 +              regs->nip, regs->link, regs->gpr[0],
 751                regs->ccr&0x10000000?"Error=":"", regs->gpr[3], print_tainted());
 752  }
 753
 754 diff -NurpP --minimal linux-2.6.36/arch/powerpc/kernel/vdso.c linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/vdso.c
 755 --- linux-2.6.36/arch/powerpc/kernel/vdso.c     2010-08-02 16:52:07.000000000 +0200
 756 +++ linux-2.6.36-vs2.3.0.36.38/arch/powerpc/kernel/vdso.c       2010-10-21 13:09:36.000000000 +0200
 757 @@ -23,6 +23,7 @@
 758  #include <linux/security.h>
 759  #include <linux/bootmem.h>
 760  #include <linux/memblock.h>
 761 +#include <linux/vs_memory.h>
 762
 763  #include <asm/pgtable.h>
 764  #include <asm/system.h>
 765 diff -NurpP --minimal linux-2.6.36/arch/s390/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/s390/Kconfig
 766 --- linux-2.6.36/arch/s390/Kconfig      2010-10-21 13:06:57.000000000 +0200
 767 +++ linux-2.6.36-vs2.3.0.36.38/arch/s390/Kconfig        2010-10-21 13:09:36.000000000 +0200
 768 @@ -620,6 +620,8 @@ source "fs/Kconfig"
 769
 770  source "arch/s390/Kconfig.debug"
 771
 772 +source "kernel/vserver/Kconfig"
 773 +
 774  source "security/Kconfig"
 775
 776  source "crypto/Kconfig"
 777 diff -NurpP --minimal linux-2.6.36/arch/s390/include/asm/tlb.h linux-2.6.36-vs2.3.0.36.38/arch/s390/include/asm/tlb.h
 778 --- linux-2.6.36/arch/s390/include/asm/tlb.h    2010-10-21 13:06:58.000000000 +0200
 779 +++ linux-2.6.36-vs2.3.0.36.38/arch/s390/include/asm/tlb.h      2010-10-21 13:09:36.000000000 +0200
 780 @@ -23,6 +23,8 @@
 781
 782  #include <linux/mm.h>
 783  #include <linux/swap.h>
 784 +#include <linux/vs_memory.h>
 785 +
 786  #include <asm/processor.h>
 787  #include <asm/pgalloc.h>
 788  #include <asm/smp.h>
 789 diff -NurpP --minimal linux-2.6.36/arch/s390/include/asm/unistd.h linux-2.6.36-vs2.3.0.36.38/arch/s390/include/asm/unistd.h
 790 --- linux-2.6.36/arch/s390/include/asm/unistd.h 2010-10-21 13:06:58.000000000 +0200
 791 +++ linux-2.6.36-vs2.3.0.36.38/arch/s390/include/asm/unistd.h   2010-10-21 13:09:36.000000000 +0200
 792 @@ -202,7 +202,7 @@
 793  #define __NR_clock_gettime     (__NR_timer_create+6)
 794  #define __NR_clock_getres      (__NR_timer_create+7)
 795  #define __NR_clock_nanosleep   (__NR_timer_create+8)
 796 -/* Number 263 is reserved for vserver */
 797 +#define __NR_vserver           263
 798  #define __NR_statfs64          265
 799  #define __NR_fstatfs64         266
 800  #define __NR_remap_file_pages  267
 801 diff -NurpP --minimal linux-2.6.36/arch/s390/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/arch/s390/kernel/ptrace.c
 802 --- linux-2.6.36/arch/s390/kernel/ptrace.c      2010-08-02 16:52:09.000000000 +0200
 803 +++ linux-2.6.36-vs2.3.0.36.38/arch/s390/kernel/ptrace.c        2010-10-21 13:09:36.000000000 +0200
 804 @@ -36,6 +36,7 @@
 805  #include <linux/regset.h>
 806  #include <linux/tracehook.h>
 807  #include <linux/seccomp.h>
 808 +#include <linux/vs_base.h>
 809  #include <trace/syscall.h>
 810  #include <asm/compat.h>
 811  #include <asm/segment.h>
 812 diff -NurpP --minimal linux-2.6.36/arch/s390/kernel/syscalls.S linux-2.6.36-vs2.3.0.36.38/arch/s390/kernel/syscalls.S
 813 --- linux-2.6.36/arch/s390/kernel/syscalls.S    2010-10-21 13:06:58.000000000 +0200
 814 +++ linux-2.6.36-vs2.3.0.36.38/arch/s390/kernel/syscalls.S      2010-10-21 13:09:36.000000000 +0200
 815 @@ -271,7 +271,7 @@ SYSCALL(sys_clock_settime,sys_clock_sett
 816  SYSCALL(sys_clock_gettime,sys_clock_gettime,sys32_clock_gettime_wrapper)       /* 260 */
 817  SYSCALL(sys_clock_getres,sys_clock_getres,sys32_clock_getres_wrapper)
 818  SYSCALL(sys_clock_nanosleep,sys_clock_nanosleep,sys32_clock_nanosleep_wrapper)
 819 -NI_SYSCALL                                                     /* reserved for vserver */
 820 +SYSCALL(sys_vserver,sys_vserver,sys32_vserver)
 821  SYSCALL(sys_s390_fadvise64_64,sys_ni_syscall,sys32_fadvise64_64_wrapper)
 822  SYSCALL(sys_statfs64,sys_statfs64,compat_sys_statfs64_wrapper)
 823  SYSCALL(sys_fstatfs64,sys_fstatfs64,compat_sys_fstatfs64_wrapper)
 824 diff -NurpP --minimal linux-2.6.36/arch/sh/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/sh/Kconfig
 825 --- linux-2.6.36/arch/sh/Kconfig        2010-10-21 13:06:58.000000000 +0200
 826 +++ linux-2.6.36-vs2.3.0.36.38/arch/sh/Kconfig  2010-10-21 13:09:36.000000000 +0200
 827 @@ -878,6 +878,8 @@ source "fs/Kconfig"
 828
 829  source "arch/sh/Kconfig.debug"
 830
 831 +source "kernel/vserver/Kconfig"
 832 +
 833  source "security/Kconfig"
 834
 835  source "crypto/Kconfig"
 836 diff -NurpP --minimal linux-2.6.36/arch/sh/kernel/irq.c linux-2.6.36-vs2.3.0.36.38/arch/sh/kernel/irq.c
 837 --- linux-2.6.36/arch/sh/kernel/irq.c   2010-08-02 16:52:10.000000000 +0200
 838 +++ linux-2.6.36-vs2.3.0.36.38/arch/sh/kernel/irq.c     2010-10-21 13:09:36.000000000 +0200
 839 @@ -13,6 +13,7 @@
 840  #include <linux/seq_file.h>
 841  #include <linux/ftrace.h>
 842  #include <linux/delay.h>
 843 +// #include <linux/vs_context.h>
 844  #include <asm/processor.h>
 845  #include <asm/machvec.h>
 846  #include <asm/uaccess.h>
 847 diff -NurpP --minimal linux-2.6.36/arch/sh/kernel/vsyscall/vsyscall.c linux-2.6.36-vs2.3.0.36.38/arch/sh/kernel/vsyscall/vsyscall.c
 848 --- linux-2.6.36/arch/sh/kernel/vsyscall/vsyscall.c     2010-07-07 18:31:10.000000000 +0200
 849 +++ linux-2.6.36-vs2.3.0.36.38/arch/sh/kernel/vsyscall/vsyscall.c       2010-10-21 13:09:36.000000000 +0200
 850 @@ -18,6 +18,7 @@
 851  #include <linux/elf.h>
 852  #include <linux/sched.h>
 853  #include <linux/err.h>
 854 +#include <linux/vs_memory.h>
 855
 856  /*
 857   * Should the kernel map a VDSO page into processes and pass its
 858 diff -NurpP --minimal linux-2.6.36/arch/sparc/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/sparc/Kconfig
 859 --- linux-2.6.36/arch/sparc/Kconfig     2010-10-21 13:06:58.000000000 +0200
 860 +++ linux-2.6.36-vs2.3.0.36.38/arch/sparc/Kconfig       2010-10-21 13:09:36.000000000 +0200
 861 @@ -565,6 +565,8 @@ source "fs/Kconfig"
 862
 863  source "arch/sparc/Kconfig.debug"
 864
 865 +source "kernel/vserver/Kconfig"
 866 +
 867  source "security/Kconfig"
 868
 869  source "crypto/Kconfig"
 870 diff -NurpP --minimal linux-2.6.36/arch/sparc/include/asm/tlb_64.h linux-2.6.36-vs2.3.0.36.38/arch/sparc/include/asm/tlb_64.h
 871 --- linux-2.6.36/arch/sparc/include/asm/tlb_64.h        2009-09-10 15:25:45.000000000 +0200
 872 +++ linux-2.6.36-vs2.3.0.36.38/arch/sparc/include/asm/tlb_64.h  2010-10-21 13:09:36.000000000 +0200
 873 @@ -3,6 +3,7 @@
 874
 875  #include <linux/swap.h>
 876  #include <linux/pagemap.h>
 877 +#include <linux/vs_memory.h>
 878  #include <asm/pgalloc.h>
 879  #include <asm/tlbflush.h>
 880  #include <asm/mmu_context.h>
 881 diff -NurpP --minimal linux-2.6.36/arch/sparc/include/asm/unistd.h linux-2.6.36-vs2.3.0.36.38/arch/sparc/include/asm/unistd.h
 882 --- linux-2.6.36/arch/sparc/include/asm/unistd.h        2010-10-21 13:06:58.000000000 +0200
 883 +++ linux-2.6.36-vs2.3.0.36.38/arch/sparc/include/asm/unistd.h  2010-10-21 13:09:36.000000000 +0200
 884 @@ -335,7 +335,7 @@
 885  #define __NR_timer_getoverrun  264
 886  #define __NR_timer_delete      265
 887  #define __NR_timer_create      266
 888 -/* #define __NR_vserver                267 Reserved for VSERVER */
 889 +#define __NR_vserver           267
 890  #define __NR_io_setup          268
 891  #define __NR_io_destroy                269
 892  #define __NR_io_submit         270
 893 diff -NurpP --minimal linux-2.6.36/arch/sparc/kernel/systbls_32.S linux-2.6.36-vs2.3.0.36.38/arch/sparc/kernel/systbls_32.S
 894 --- linux-2.6.36/arch/sparc/kernel/systbls_32.S 2010-10-21 13:06:59.000000000 +0200
 895 +++ linux-2.6.36-vs2.3.0.36.38/arch/sparc/kernel/systbls_32.S   2010-10-21 13:09:36.000000000 +0200
 896 @@ -70,7 +70,7 @@ sys_call_table:
 897  /*250*/        .long sys_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
 898  /*255*/        .long sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
 899  /*260*/        .long sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
 900 -/*265*/        .long sys_timer_delete, sys_timer_create, sys_nis_syscall, sys_io_setup, sys_io_destroy
 901 +/*265*/        .long sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
 902  /*270*/        .long sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
 903  /*275*/        .long sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
 904  /*280*/        .long sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
 905 diff -NurpP --minimal linux-2.6.36/arch/sparc/kernel/systbls_64.S linux-2.6.36-vs2.3.0.36.38/arch/sparc/kernel/systbls_64.S
 906 --- linux-2.6.36/arch/sparc/kernel/systbls_64.S 2010-10-21 13:06:59.000000000 +0200
 907 +++ linux-2.6.36-vs2.3.0.36.38/arch/sparc/kernel/systbls_64.S   2010-10-21 13:09:36.000000000 +0200
 908 @@ -71,7 +71,7 @@ sys_call_table32:
 909  /*250*/        .word sys_mremap, compat_sys_sysctl, sys32_getsid, sys_fdatasync, sys32_nfsservctl
 910         .word sys32_sync_file_range, compat_sys_clock_settime, compat_sys_clock_gettime, compat_sys_clock_getres, sys32_clock_nanosleep
 911  /*260*/        .word compat_sys_sched_getaffinity, compat_sys_sched_setaffinity, sys32_timer_settime, compat_sys_timer_gettime, sys_timer_getoverrun
 912 -       .word sys_timer_delete, compat_sys_timer_create, sys_ni_syscall, compat_sys_io_setup, sys_io_destroy
 913 +       .word sys_timer_delete, compat_sys_timer_create, sys32_vserver, compat_sys_io_setup, sys_io_destroy
 914  /*270*/        .word sys32_io_submit, sys_io_cancel, compat_sys_io_getevents, sys32_mq_open, sys_mq_unlink
 915         .word compat_sys_mq_timedsend, compat_sys_mq_timedreceive, compat_sys_mq_notify, compat_sys_mq_getsetattr, compat_sys_waitid
 916  /*280*/        .word sys32_tee, sys_add_key, sys_request_key, sys_keyctl, compat_sys_openat
 917 @@ -147,7 +147,7 @@ sys_call_table:
 918  /*250*/        .word sys_64_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
 919         .word sys_sync_file_range, sys_clock_settime, sys_clock_gettime, sys_clock_getres, sys_clock_nanosleep
 920  /*260*/        .word sys_sched_getaffinity, sys_sched_setaffinity, sys_timer_settime, sys_timer_gettime, sys_timer_getoverrun
 921 -       .word sys_timer_delete, sys_timer_create, sys_ni_syscall, sys_io_setup, sys_io_destroy
 922 +       .word sys_timer_delete, sys_timer_create, sys_vserver, sys_io_setup, sys_io_destroy
 923  /*270*/        .word sys_io_submit, sys_io_cancel, sys_io_getevents, sys_mq_open, sys_mq_unlink
 924         .word sys_mq_timedsend, sys_mq_timedreceive, sys_mq_notify, sys_mq_getsetattr, sys_waitid
 925  /*280*/        .word sys_tee, sys_add_key, sys_request_key, sys_keyctl, sys_openat
 926 diff -NurpP --minimal linux-2.6.36/arch/um/Kconfig.rest linux-2.6.36-vs2.3.0.36.38/arch/um/Kconfig.rest
 927 --- linux-2.6.36/arch/um/Kconfig.rest   2009-06-11 17:12:19.000000000 +0200
 928 +++ linux-2.6.36-vs2.3.0.36.38/arch/um/Kconfig.rest     2010-10-21 13:09:36.000000000 +0200
 929 @@ -18,6 +18,8 @@ source "drivers/connector/Kconfig"
 930
 931  source "fs/Kconfig"
 932
 933 +source "kernel/vserver/Kconfig"
 934 +
 935  source "security/Kconfig"
 936
 937  source "crypto/Kconfig"
 938 diff -NurpP --minimal linux-2.6.36/arch/um/include/asm/tlb.h linux-2.6.36-vs2.3.0.36.38/arch/um/include/asm/tlb.h
 939 --- linux-2.6.36/arch/um/include/asm/tlb.h      2009-09-10 15:25:46.000000000 +0200
 940 +++ linux-2.6.36-vs2.3.0.36.38/arch/um/include/asm/tlb.h        2010-10-21 13:09:36.000000000 +0200
 941 @@ -3,6 +3,7 @@
 942
 943  #include <linux/pagemap.h>
 944  #include <linux/swap.h>
 945 +#include <linux/vs_memory.h>
 946  #include <asm/percpu.h>
 947  #include <asm/pgalloc.h>
 948  #include <asm/tlbflush.h>
 949 diff -NurpP --minimal linux-2.6.36/arch/um/include/shared/kern_constants.h linux-2.6.36-vs2.3.0.36.38/arch/um/include/shared/kern_constants.h
 950 --- linux-2.6.36/arch/um/include/shared/kern_constants.h        1970-01-01 01:00:00.000000000 +0100
 951 +++ linux-2.6.36-vs2.3.0.36.38/arch/um/include/shared/kern_constants.h  2010-10-21 13:09:36.000000000 +0200
 952 @@ -0,0 +1 @@
 953 +#include "../../../../include/generated/asm-offsets.h"
 954 diff -NurpP --minimal linux-2.6.36/arch/um/include/shared/user_constants.h linux-2.6.36-vs2.3.0.36.38/arch/um/include/shared/user_constants.h
 955 --- linux-2.6.36/arch/um/include/shared/user_constants.h        1970-01-01 01:00:00.000000000 +0100
 956 +++ linux-2.6.36-vs2.3.0.36.38/arch/um/include/shared/user_constants.h  2010-10-21 13:09:36.000000000 +0200
 957 @@ -0,0 +1,40 @@
 958 +/*
 959 + * DO NOT MODIFY.
 960 + *
 961 + * This file was generated by arch/um/Makefile
 962 + *
 963 + */
 964 +
 965 +#define HOST_SC_CR2 176 /* offsetof(struct sigcontext, cr2)    # */
 966 +#define HOST_SC_ERR 152 /* offsetof(struct sigcontext, err)    # */
 967 +#define HOST_SC_TRAPNO 160 /* offsetof(struct sigcontext, trapno)      # */
 968 +#define HOST_FP_SIZE 64 /* sizeof(struct _fpstate) / sizeof(unsigned long)     # */
 969 +#define HOST_RBX 5 /* RBX      # */
 970 +#define HOST_RCX 11 /* RCX     # */
 971 +#define HOST_RDI 14 /* RDI     # */
 972 +#define HOST_RSI 13 /* RSI     # */
 973 +#define HOST_RDX 12 /* RDX     # */
 974 +#define HOST_RBP 4 /* RBP      # */
 975 +#define HOST_RAX 10 /* RAX     # */
 976 +#define HOST_R8 9 /* R8        # */
 977 +#define HOST_R9 8 /* R9        # */
 978 +#define HOST_R10 7 /* R10      # */
 979 +#define HOST_R11 6 /* R11      # */
 980 +#define HOST_R12 3 /* R12      # */
 981 +#define HOST_R13 2 /* R13      # */
 982 +#define HOST_R14 1 /* R14      # */
 983 +#define HOST_R15 0 /* R15      # */
 984 +#define HOST_ORIG_RAX 15 /* ORIG_RAX   # */
 985 +#define HOST_CS 17 /* CS       # */
 986 +#define HOST_SS 20 /* SS       # */
 987 +#define HOST_EFLAGS 18 /* EFLAGS       # */
 988 +#define HOST_IP 16 /* RIP      # */
 989 +#define HOST_SP 19 /* RSP      # */
 990 +#define UM_FRAME_SIZE 216 /* sizeof(struct user_regs_struct)   # */
 991 +#define UM_POLLIN 1 /* POLLIN  # */
 992 +#define UM_POLLPRI 2 /* POLLPRI        # */
 993 +#define UM_POLLOUT 4 /* POLLOUT        # */
 994 +#define UM_PROT_READ 1 /* PROT_READ    # */
 995 +#define UM_PROT_WRITE 2 /* PROT_WRITE  # */
 996 +#define UM_PROT_EXEC 4 /* PROT_EXEC    # */
 997 +
 998 diff -NurpP --minimal linux-2.6.36/arch/x86/Kconfig linux-2.6.36-vs2.3.0.36.38/arch/x86/Kconfig
 999 --- linux-2.6.36/arch/x86/Kconfig       2010-10-21 13:06:59.000000000 +0200
1000 +++ linux-2.6.36-vs2.3.0.36.38/arch/x86/Kconfig 2010-10-21 13:09:36.000000000 +0200
1001 @@ -2135,6 +2135,8 @@ source "fs/Kconfig"
1002
1003  source "arch/x86/Kconfig.debug"
1004
1005 +source "kernel/vserver/Kconfig"
1006 +
1007  source "security/Kconfig"
1008
1009  source "crypto/Kconfig"
1010 diff -NurpP --minimal linux-2.6.36/arch/x86/ia32/ia32entry.S linux-2.6.36-vs2.3.0.36.38/arch/x86/ia32/ia32entry.S
1011 --- linux-2.6.36/arch/x86/ia32/ia32entry.S      2010-10-21 13:06:59.000000000 +0200
1012 +++ linux-2.6.36-vs2.3.0.36.38/arch/x86/ia32/ia32entry.S        2010-10-21 13:09:36.000000000 +0200
1013 @@ -783,7 +783,7 @@ ia32_sys_call_table:
1014         .quad sys_tgkill                /* 270 */
1015         .quad compat_sys_utimes
1016         .quad sys32_fadvise64_64
1017 -       .quad quiet_ni_syscall  /* sys_vserver */
1018 +       .quad sys32_vserver
1019         .quad sys_mbind
1020         .quad compat_sys_get_mempolicy  /* 275 */
1021         .quad sys_set_mempolicy
1022 diff -NurpP --minimal linux-2.6.36/arch/x86/include/asm/unistd_64.h linux-2.6.36-vs2.3.0.36.38/arch/x86/include/asm/unistd_64.h
1023 --- linux-2.6.36/arch/x86/include/asm/unistd_64.h       2010-10-21 13:06:59.000000000 +0200
1024 +++ linux-2.6.36-vs2.3.0.36.38/arch/x86/include/asm/unistd_64.h 2010-10-21 13:09:36.000000000 +0200
1025 @@ -535,7 +535,7 @@ __SYSCALL(__NR_tgkill, sys_tgkill)
1026  #define __NR_utimes                            235
1027  __SYSCALL(__NR_utimes, sys_utimes)
1028  #define __NR_vserver                           236
1029 -__SYSCALL(__NR_vserver, sys_ni_syscall)
1030 +__SYSCALL(__NR_vserver, sys_vserver)
1031  #define __NR_mbind                             237
1032  __SYSCALL(__NR_mbind, sys_mbind)
1033  #define __NR_set_mempolicy                     238
1034 diff -NurpP --minimal linux-2.6.36/arch/x86/kernel/syscall_table_32.S linux-2.6.36-vs2.3.0.36.38/arch/x86/kernel/syscall_table_32.S
1035 --- linux-2.6.36/arch/x86/kernel/syscall_table_32.S     2010-10-21 13:07:00.000000000 +0200
1036 +++ linux-2.6.36-vs2.3.0.36.38/arch/x86/kernel/syscall_table_32.S       2010-10-21 13:09:36.000000000 +0200
1037 @@ -272,7 +272,7 @@ ENTRY(sys_call_table)
1038         .long sys_tgkill        /* 270 */
1039         .long sys_utimes
1040         .long sys_fadvise64_64
1041 -       .long sys_ni_syscall    /* sys_vserver */
1042 +       .long sys_vserver
1043         .long sys_mbind
1044         .long sys_get_mempolicy
1045         .long sys_set_mempolicy
1046 diff -NurpP --minimal linux-2.6.36/drivers/block/Kconfig linux-2.6.36-vs2.3.0.36.38/drivers/block/Kconfig
1047 --- linux-2.6.36/drivers/block/Kconfig  2010-08-02 16:52:14.000000000 +0200
1048 +++ linux-2.6.36-vs2.3.0.36.38/drivers/block/Kconfig    2010-10-21 13:09:36.000000000 +0200
1049 @@ -273,6 +273,13 @@ config BLK_DEV_CRYPTOLOOP
1050
1051  source "drivers/block/drbd/Kconfig"
1052
1053 +config BLK_DEV_VROOT
1054 +       tristate "Virtual Root device support"
1055 +       depends on QUOTACTL
1056 +       ---help---
1057 +         Saying Y here will allow you to use quota/fs ioctls on a shared
1058 +         partition within a virtual server without compromising security.
1059 +
1060  config BLK_DEV_NBD
1061         tristate "Network block device support"
1062         depends on NET
1063 diff -NurpP --minimal linux-2.6.36/drivers/block/Makefile linux-2.6.36-vs2.3.0.36.38/drivers/block/Makefile
1064 --- linux-2.6.36/drivers/block/Makefile 2010-02-25 11:51:36.000000000 +0100
1065 +++ linux-2.6.36-vs2.3.0.36.38/drivers/block/Makefile   2010-10-21 13:09:36.000000000 +0200
1066 @@ -34,6 +34,7 @@ obj-$(CONFIG_VIODASD)         += viodasd.o
1067  obj-$(CONFIG_BLK_DEV_SX8)      += sx8.o
1068  obj-$(CONFIG_BLK_DEV_UB)       += ub.o
1069  obj-$(CONFIG_BLK_DEV_HD)       += hd.o
1070 +obj-$(CONFIG_BLK_DEV_VROOT)    += vroot.o
1071
1072  obj-$(CONFIG_XEN_BLKDEV_FRONTEND)      += xen-blkfront.o
1073  obj-$(CONFIG_BLK_DEV_DRBD)     += drbd/
1074 diff -NurpP --minimal linux-2.6.36/drivers/block/loop.c linux-2.6.36-vs2.3.0.36.38/drivers/block/loop.c
1075 --- linux-2.6.36/drivers/block/loop.c   2010-10-21 13:07:02.000000000 +0200
1076 +++ linux-2.6.36-vs2.3.0.36.38/drivers/block/loop.c     2010-10-21 13:11:25.000000000 +0200
1077 @@ -74,6 +74,7 @@
1078  #include <linux/highmem.h>
1079  #include <linux/kthread.h>
1080  #include <linux/splice.h>
1081 +#include <linux/vs_context.h>
1082
1083  #include <asm/uaccess.h>
1084
1085 @@ -814,6 +815,7 @@ static int loop_set_fd(struct loop_devic
1086         lo->lo_blocksize = lo_blocksize;
1087         lo->lo_device = bdev;
1088         lo->lo_flags = lo_flags;
1089 +       lo->lo_xid = vx_current_xid();
1090         lo->lo_backing_file = file;
1091         lo->transfer = transfer_none;
1092         lo->ioctl = NULL;
1093 @@ -942,6 +944,7 @@ static int loop_clr_fd(struct loop_devic
1094         lo->lo_encrypt_key_size = 0;
1095         lo->lo_flags = 0;
1096         lo->lo_thread = NULL;
1097 +       lo->lo_xid = 0;
1098         memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
1099         memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
1100         memset(lo->lo_file_name, 0, LO_NAME_SIZE);
1101 @@ -979,7 +982,7 @@ loop_set_status(struct loop_device *lo,
1102
1103         if (lo->lo_encrypt_key_size &&
1104             lo->lo_key_owner != uid &&
1105 -           !capable(CAP_SYS_ADMIN))
1106 +           !vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP))
1107                 return -EPERM;
1108         if (lo->lo_state != Lo_bound)
1109                 return -ENXIO;
1110 @@ -1063,7 +1066,8 @@ loop_get_status(struct loop_device *lo,
1111         memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
1112         info->lo_encrypt_type =
1113                 lo->lo_encryption ? lo->lo_encryption->number : 0;
1114 -       if (lo->lo_encrypt_key_size && capable(CAP_SYS_ADMIN)) {
1115 +       if (lo->lo_encrypt_key_size &&
1116 +               vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_CLOOP)) {
1117                 info->lo_encrypt_key_size = lo->lo_encrypt_key_size;
1118                 memcpy(info->lo_encrypt_key, lo->lo_encrypt_key,
1119                        lo->lo_encrypt_key_size);
1120 @@ -1409,6 +1413,9 @@ static int lo_open(struct block_device *
1121  {
1122         struct loop_device *lo = bdev->bd_disk->private_data;
1123
1124 +       if (!vx_check(lo->lo_xid, VS_IDENT|VS_HOSTID|VS_ADMIN_P))
1125 +               return -EACCES;
1126 +
1127         lock_kernel();
1128         mutex_lock(&lo->lo_ctl_mutex);
1129         lo->lo_refcnt++;
1130 diff -NurpP --minimal linux-2.6.36/drivers/block/vroot.c linux-2.6.36-vs2.3.0.36.38/drivers/block/vroot.c
1131 --- linux-2.6.36/drivers/block/vroot.c  1970-01-01 01:00:00.000000000 +0100
1132 +++ linux-2.6.36-vs2.3.0.36.38/drivers/block/vroot.c    2010-10-21 13:09:36.000000000 +0200
1133 @@ -0,0 +1,282 @@
1134 +/*
1135 + *  linux/drivers/block/vroot.c
1136 + *
1137 + *  written by Herbert Pötzl, 9/11/2002
1138 + *  ported to 2.6.10 by Herbert Pötzl, 30/12/2004
1139 + *
1140 + *  based on the loop.c code by Theodore Ts'o.
1141 + *
1142 + * Copyright (C) 2002-2007 by Herbert Pötzl.
1143 + * Redistribution of this file is permitted under the
1144 + * GNU General Public License.
1145 + *
1146 + */
1147 +
1148 +#include <linux/module.h>
1149 +#include <linux/moduleparam.h>
1150 +#include <linux/file.h>
1151 +#include <linux/major.h>
1152 +#include <linux/blkdev.h>
1153 +#include <linux/slab.h>
1154 +
1155 +#include <linux/vroot.h>
1156 +#include <linux/vs_context.h>
1157 +
1158 +
1159 +static int max_vroot = 8;
1160 +
1161 +static struct vroot_device *vroot_dev;
1162 +static struct gendisk **disks;
1163 +
1164 +
1165 +static int vroot_set_dev(
1166 +       struct vroot_device *vr,
1167 +       struct block_device *bdev,
1168 +       unsigned int arg)
1169 +{
1170 +       struct block_device *real_bdev;
1171 +       struct file *file;
1172 +       struct inode *inode;
1173 +       int error;
1174 +
1175 +       error = -EBUSY;
1176 +       if (vr->vr_state != Vr_unbound)
1177 +               goto out;
1178 +
1179 +       error = -EBADF;
1180 +       file = fget(arg);
1181 +       if (!file)
1182 +               goto out;
1183 +
1184 +       error = -EINVAL;
1185 +       inode = file->f_dentry->d_inode;
1186 +
1187 +
1188 +       if (S_ISBLK(inode->i_mode)) {
1189 +               real_bdev = inode->i_bdev;
1190 +               vr->vr_device = real_bdev;
1191 +               __iget(real_bdev->bd_inode);
1192 +       } else
1193 +               goto out_fput;
1194 +
1195 +       vxdprintk(VXD_CBIT(misc, 0),
1196 +               "vroot[%d]_set_dev: dev=" VXF_DEV,
1197 +               vr->vr_number, VXD_DEV(real_bdev));
1198 +
1199 +       vr->vr_state = Vr_bound;
1200 +       error = 0;
1201 +
1202 + out_fput:
1203 +       fput(file);
1204 + out:
1205 +       return error;
1206 +}
1207 +
1208 +static int vroot_clr_dev(
1209 +       struct vroot_device *vr,
1210 +       struct block_device *bdev)
1211 +{
1212 +       struct block_device *real_bdev;
1213 +
1214 +       if (vr->vr_state != Vr_bound)
1215 +               return -ENXIO;
1216 +       if (vr->vr_refcnt > 1)  /* we needed one fd for the ioctl */
1217 +               return -EBUSY;
1218 +
1219 +       real_bdev = vr->vr_device;
1220 +
1221 +       vxdprintk(VXD_CBIT(misc, 0),
1222 +               "vroot[%d]_clr_dev: dev=" VXF_DEV,
1223 +               vr->vr_number, VXD_DEV(real_bdev));
1224 +
1225 +       bdput(real_bdev);
1226 +       vr->vr_state = Vr_unbound;
1227 +       vr->vr_device = NULL;
1228 +       return 0;
1229 +}
1230 +
1231 +
1232 +static int vr_ioctl(struct block_device *bdev, fmode_t mode,
1233 +       unsigned int cmd, unsigned long arg)
1234 +{
1235 +       struct vroot_device *vr = bdev->bd_disk->private_data;
1236 +       int err;
1237 +
1238 +       down(&vr->vr_ctl_mutex);
1239 +       switch (cmd) {
1240 +       case VROOT_SET_DEV:
1241 +               err = vroot_set_dev(vr, bdev, arg);
1242 +               break;
1243 +       case VROOT_CLR_DEV:
1244 +               err = vroot_clr_dev(vr, bdev);
1245 +               break;
1246 +       default:
1247 +               err = -EINVAL;
1248 +               break;
1249 +       }
1250 +       up(&vr->vr_ctl_mutex);
1251 +       return err;
1252 +}
1253 +
1254 +static int vr_open(struct block_device *bdev, fmode_t mode)
1255 +{
1256 +       struct vroot_device *vr = bdev->bd_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 int vr_release(struct gendisk *disk, fmode_t mode)
1265 +{
1266 +       struct vroot_device *vr = disk->private_data;
1267 +
1268 +       down(&vr->vr_ctl_mutex);
1269 +       --vr->vr_refcnt;
1270 +       up(&vr->vr_ctl_mutex);
1271 +       return 0;
1272 +}
1273 +
1274 +static struct block_device_operations vr_fops = {
1275 +       .owner =        THIS_MODULE,
1276 +       .open =         vr_open,
1277 +       .release =      vr_release,
1278 +       .ioctl =        vr_ioctl,
1279 +};
1280 +
1281 +struct block_device *__vroot_get_real_bdev(struct block_device *bdev)
1282 +{
1283 +       struct inode *inode = bdev->bd_inode;
1284 +       struct vroot_device *vr;
1285 +       struct block_device *real_bdev;
1286 +       int minor = iminor(inode);
1287 +
1288 +       vr = &vroot_dev[minor];
1289 +       real_bdev = vr->vr_device;
1290 +
1291 +       vxdprintk(VXD_CBIT(misc, 0),
1292 +               "vroot[%d]_get_real_bdev: dev=" VXF_DEV,
1293 +               vr->vr_number, VXD_DEV(real_bdev));
1294 +
1295 +       if (vr->vr_state != Vr_bound)
1296 +               return ERR_PTR(-ENXIO);
1297 +
1298 +       __iget(real_bdev->bd_inode);
1299 +       return real_bdev;
1300 +}
1301 +
1302 +/*
1303 + * And now the modules code and kernel interface.
1304 + */
1305 +
1306 +module_param(max_vroot, int, 0);
1307 +
1308 +MODULE_PARM_DESC(max_vroot, "Maximum number of vroot devices (1-256)");
1309 +MODULE_LICENSE("GPL");
1310 +MODULE_ALIAS_BLOCKDEV_MAJOR(VROOT_MAJOR);
1311 +
1312 +MODULE_AUTHOR ("Herbert Pötzl");
1313 +MODULE_DESCRIPTION ("Virtual Root Device Mapper");
1314 +
1315 +
1316 +int __init vroot_init(void)
1317 +{
1318 +       int err, i;
1319 +
1320 +       if (max_vroot < 1 || max_vroot > 256) {
1321 +               max_vroot = MAX_VROOT_DEFAULT;
1322 +               printk(KERN_WARNING "vroot: invalid max_vroot "
1323 +                       "(must be between 1 and 256), "
1324 +                       "using default (%d)\n", max_vroot);
1325 +       }
1326 +
1327 +       if (register_blkdev(VROOT_MAJOR, "vroot"))
1328 +               return -EIO;
1329 +
1330 +       err = -ENOMEM;
1331 +       vroot_dev = kmalloc(max_vroot * sizeof(struct vroot_device), GFP_KERNEL);
1332 +       if (!vroot_dev)
1333 +               goto out_mem1;
1334 +       memset(vroot_dev, 0, max_vroot * sizeof(struct vroot_device));
1335 +
1336 +       disks = kmalloc(max_vroot * sizeof(struct gendisk *), GFP_KERNEL);
1337 +       if (!disks)
1338 +               goto out_mem2;
1339 +
1340 +       for (i = 0; i < max_vroot; i++) {
1341 +               disks[i] = alloc_disk(1);
1342 +               if (!disks[i])
1343 +                       goto out_mem3;
1344 +               disks[i]->queue = blk_alloc_queue(GFP_KERNEL);
1345 +               if (!disks[i]->queue)
1346 +                       goto out_mem3;
1347 +       }
1348 +
1349 +       for (i = 0; i < max_vroot; i++) {
1350 +               struct vroot_device *vr = &vroot_dev[i];
1351 +               struct gendisk *disk = disks[i];
1352 +
1353 +               memset(vr, 0, sizeof(*vr));
1354 +               init_MUTEX(&vr->vr_ctl_mutex);
1355 +               vr->vr_number = i;
1356 +               disk->major = VROOT_MAJOR;
1357 +               disk->first_minor = i;
1358 +               disk->fops = &vr_fops;
1359 +               sprintf(disk->disk_name, "vroot%d", i);
1360 +               disk->private_data = vr;
1361 +       }
1362 +
1363 +       err = register_vroot_grb(&__vroot_get_real_bdev);
1364 +       if (err)
1365 +               goto out_mem3;
1366 +
1367 +       for (i = 0; i < max_vroot; i++)
1368 +               add_disk(disks[i]);
1369 +       printk(KERN_INFO "vroot: loaded (max %d devices)\n", max_vroot);
1370 +       return 0;
1371 +
1372 +out_mem3:
1373 +       while (i--)
1374 +               put_disk(disks[i]);
1375 +       kfree(disks);
1376 +out_mem2:
1377 +       kfree(vroot_dev);
1378 +out_mem1:
1379 +       unregister_blkdev(VROOT_MAJOR, "vroot");
1380 +       printk(KERN_ERR "vroot: ran out of memory\n");
1381 +       return err;
1382 +}
1383 +
1384 +void vroot_exit(void)
1385 +{
1386 +       int i;
1387 +
1388 +       if (unregister_vroot_grb(&__vroot_get_real_bdev))
1389 +               printk(KERN_WARNING "vroot: cannot unregister grb\n");
1390 +
1391 +       for (i = 0; i < max_vroot; i++) {
1392 +               del_gendisk(disks[i]);
1393 +               put_disk(disks[i]);
1394 +       }
1395 +       unregister_blkdev(VROOT_MAJOR, "vroot");
1396 +
1397 +       kfree(disks);
1398 +       kfree(vroot_dev);
1399 +}
1400 +
1401 +module_init(vroot_init);
1402 +module_exit(vroot_exit);
1403 +
1404 +#ifndef MODULE
1405 +
1406 +static int __init max_vroot_setup(char *str)
1407 +{
1408 +       max_vroot = simple_strtol(str, NULL, 0);
1409 +       return 1;
1410 +}
1411 +
1412 +__setup("max_vroot=", max_vroot_setup);
1413 +
1414 +#endif
1415 +
1416 diff -NurpP --minimal linux-2.6.36/drivers/char/sysrq.c linux-2.6.36-vs2.3.0.36.38/drivers/char/sysrq.c
1417 --- linux-2.6.36/drivers/char/sysrq.c   2010-10-21 13:07:02.000000000 +0200
1418 +++ linux-2.6.36-vs2.3.0.36.38/drivers/char/sysrq.c     2010-10-28 18:08:06.000000000 +0200
1419 @@ -41,6 +41,7 @@
1420  #include <linux/oom.h>
1421  #include <linux/slab.h>
1422  #include <linux/input.h>
1423 +#include <linux/vserver/debug.h>
1424
1425  #include <asm/ptrace.h>
1426  #include <asm/irq_regs.h>
1427 @@ -395,6 +396,21 @@ static struct sysrq_key_op sysrq_unrt_op
1428         .enable_mask    = SYSRQ_ENABLE_RTNICE,
1429  };
1430
1431 +
1432 +#ifdef CONFIG_VSERVER_DEBUG
1433 +static void sysrq_handle_vxinfo(int key)
1434 +{
1435 +       dump_vx_info_inactive((key == 'x') ? 0 : 1);
1436 +}
1437 +
1438 +static struct sysrq_key_op sysrq_showvxinfo_op = {
1439 +       .handler        = sysrq_handle_vxinfo,
1440 +       .help_msg       = "conteXt",
1441 +       .action_msg     = "Show Context Info",
1442 +       .enable_mask    = SYSRQ_ENABLE_DUMP,
1443 +};
1444 +#endif
1445 +
1446  /* Key Operations table and lock */
1447  static DEFINE_SPINLOCK(sysrq_key_table_lock);
1448
1449 @@ -449,7 +465,11 @@ static struct sysrq_key_op *sysrq_key_ta
1450         NULL,                           /* v */
1451         &sysrq_showstate_blocked_op,    /* w */
1452         /* x: May be registered on ppc/powerpc for xmon */
1453 +#ifdef CONFIG_VSERVER_DEBUG
1454 +       &sysrq_showvxinfo_op,           /* x */
1455 +#else
1456         NULL,                           /* x */
1457 +#endif
1458         /* y: May be registered on sparc64 for global register dump */
1459         NULL,                           /* y */
1460         &sysrq_ftrace_dump_op,          /* z */
1461 @@ -464,6 +484,8 @@ static int sysrq_key_table_key2index(int
1462                 retval = key - '0';
1463         else if ((key >= 'a') && (key <= 'z'))
1464                 retval = key + 10 - 'a';
1465 +       else if ((key >= 'A') && (key <= 'Z'))
1466 +               retval = key + 10 - 'A';
1467         else
1468                 retval = -1;
1469         return retval;
1470 diff -NurpP --minimal linux-2.6.36/drivers/char/tty_io.c linux-2.6.36-vs2.3.0.36.38/drivers/char/tty_io.c
1471 --- linux-2.6.36/drivers/char/tty_io.c  2010-10-21 13:07:02.000000000 +0200
1472 +++ linux-2.6.36-vs2.3.0.36.38/drivers/char/tty_io.c    2010-10-21 13:09:36.000000000 +0200
1473 @@ -106,6 +106,7 @@
1474
1475  #include <linux/kmod.h>
1476  #include <linux/nsproxy.h>
1477 +#include <linux/vs_pid.h>
1478
1479  #undef TTY_DEBUG_HANGUP
1480
1481 @@ -2038,7 +2039,8 @@ static int tiocsti(struct tty_struct *tt
1482         char ch, mbz = 0;
1483         struct tty_ldisc *ld;
1484
1485 -       if ((current->signal->tty != tty) && !capable(CAP_SYS_ADMIN))
1486 +       if (((current->signal->tty != tty) &&
1487 +               !vx_capable(CAP_SYS_ADMIN, VXC_TIOCSTI)))
1488                 return -EPERM;
1489         if (get_user(ch, p))
1490                 return -EFAULT;
1491 @@ -2326,6 +2328,7 @@ static int tiocspgrp(struct tty_struct *
1492                 return -ENOTTY;
1493         if (get_user(pgrp_nr, p))
1494                 return -EFAULT;
1495 +       pgrp_nr = vx_rmap_pid(pgrp_nr);
1496         if (pgrp_nr < 0)
1497                 return -EINVAL;
1498         rcu_read_lock();
1499 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/r100_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r100_reg_safe.h
1500 --- linux-2.6.36/drivers/gpu/drm/radeon/r100_reg_safe.h 1970-01-01 01:00:00.000000000 +0100
1501 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r100_reg_safe.h   2010-10-21 13:09:36.000000000 +0200
1502 @@ -0,0 +1,28 @@
1503 +static const unsigned r100_reg_safe_bm[102] = {
1504 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1505 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1506 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1507 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1508 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1509 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1510 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1511 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1512 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1513 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1514 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1515 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1516 +       0xFFFFFFCF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1517 +       0xFFFFFF9F, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1518 +       0x38E7FE1F, 0xFFC3FF8E, 0x7FF8FFFF, 0xFFFF803C,
1519 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1520 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFEFFFF, 0xFFFFFFFF,
1521 +       0x00000000, 0xFFFFFFFD, 0xFFFFFFFF, 0xFFFFFFFF,
1522 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1523 +       0xFFFFFFFF, 0xFFFCFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1524 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1525 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1526 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1527 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1528 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1529 +       0xFFFFFFFF, 0xFFFFFFEF,
1530 +};
1531 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/r200_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r200_reg_safe.h
1532 --- linux-2.6.36/drivers/gpu/drm/radeon/r200_reg_safe.h 1970-01-01 01:00:00.000000000 +0100
1533 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r200_reg_safe.h   2010-10-21 13:09:36.000000000 +0200
1534 @@ -0,0 +1,28 @@
1535 +static const unsigned r200_reg_safe_bm[102] = {
1536 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1537 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1538 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1539 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1540 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1541 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1542 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1543 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1544 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1545 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1546 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1547 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1548 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1549 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1550 +       0xFFE7FE1F, 0xF003FFFF, 0x7EFFFFFF, 0xFFFF803C,
1551 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1552 +       0xFFFFFFFF, 0xFFFFEFCE, 0xFFFEFFFF, 0xFFFFFFFE,
1553 +       0x020E0FF0, 0xFFCC83FD, 0xFFFFFFFF, 0xFFFFFFFF,
1554 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1555 +       0xFFFBFFFF, 0xEFFCFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1556 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1557 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1558 +       0xDFDFDFDF, 0x3FFDDFDF, 0xFFFFFFFF, 0xFFFFFF7F,
1559 +       0xFFFFFFFF, 0x00FFFFFF, 0x00000000, 0x00000000,
1560 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1561 +       0xFFFFFE3F, 0xFFFFFFEF,
1562 +};
1563 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/r300_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r300_reg_safe.h
1564 --- linux-2.6.36/drivers/gpu/drm/radeon/r300_reg_safe.h 1970-01-01 01:00:00.000000000 +0100
1565 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r300_reg_safe.h   2010-10-21 13:09:36.000000000 +0200
1566 @@ -0,0 +1,42 @@
1567 +static const unsigned r300_reg_safe_bm[159] = {
1568 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1569 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1570 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1571 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1572 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1573 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1574 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1575 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1576 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1577 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1578 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1579 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1580 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1581 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1582 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFF03F,
1583 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1584 +       0xFFFFFFFF, 0xFFFFEFCE, 0xF00EBFFF, 0x007C0000,
1585 +       0xF0000078, 0xFF000009, 0xFFFFFFFF, 0xFFFFFFFF,
1586 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1587 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1588 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1589 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1590 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1591 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1592 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1593 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1594 +       0xFFFFF7FF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1595 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1596 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1597 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1598 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1599 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1600 +       0xFFFFFC78, 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF,
1601 +       0x38FF8F50, 0xFFF88082, 0xF000000C, 0xFAE009FF,
1602 +       0x0000FFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000,
1603 +       0x00000000, 0x0000C100, 0x00000000, 0x00000000,
1604 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1605 +       0x00000000, 0xFFFF0000, 0xFFFFFFFF, 0xFF80FFFF,
1606 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1607 +       0x0003FC01, 0xFFFFFCF8, 0xFF800B19,
1608 +};
1609 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/r420_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r420_reg_safe.h
1610 --- linux-2.6.36/drivers/gpu/drm/radeon/r420_reg_safe.h 1970-01-01 01:00:00.000000000 +0100
1611 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/r420_reg_safe.h   2010-10-21 13:09:36.000000000 +0200
1612 @@ -0,0 +1,42 @@
1613 +static const unsigned r420_reg_safe_bm[159] = {
1614 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1615 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1616 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1617 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1618 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1619 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1620 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1621 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1622 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1623 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1624 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1625 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1626 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1627 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1628 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFF03F,
1629 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1630 +       0xFFFFFFFF, 0xFFFFEFCE, 0xF00EBFFF, 0x007C0000,
1631 +       0xF0000078, 0xFF000009, 0xFFFFFFFF, 0xFFFFFFFF,
1632 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1633 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1634 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1635 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1636 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1637 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1638 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1639 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1640 +       0xFFFFF7FF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1641 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1642 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1643 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1644 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1645 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1646 +       0xFFFFFC78, 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF,
1647 +       0x38FF8F50, 0xFFF88082, 0xF000000C, 0xFAE009FF,
1648 +       0x0000FFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000,
1649 +       0x00000000, 0x00000100, 0x00000000, 0x00000000,
1650 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1651 +       0x00000000, 0x00000000, 0x00000000, 0xFF800000,
1652 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1653 +       0x0003FC01, 0xFFFFFCF8, 0xFF800B19,
1654 +};
1655 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/rn50_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rn50_reg_safe.h
1656 --- linux-2.6.36/drivers/gpu/drm/radeon/rn50_reg_safe.h 1970-01-01 01:00:00.000000000 +0100
1657 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rn50_reg_safe.h   2010-10-21 13:09:36.000000000 +0200
1658 @@ -0,0 +1,28 @@
1659 +static const unsigned rn50_reg_safe_bm[102] = {
1660 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1661 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1662 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1663 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1664 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1665 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1666 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1667 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1668 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1669 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1670 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1671 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1672 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1673 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1674 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1675 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1676 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1677 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1678 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1679 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1680 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1681 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1682 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1683 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1684 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1685 +       0xFFFFFFFF, 0xFFFFFFFF,
1686 +};
1687 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/rs600_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rs600_reg_safe.h
1688 --- linux-2.6.36/drivers/gpu/drm/radeon/rs600_reg_safe.h        1970-01-01 01:00:00.000000000 +0100
1689 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rs600_reg_safe.h  2010-10-21 13:09:36.000000000 +0200
1690 @@ -0,0 +1,57 @@
1691 +static const unsigned rs600_reg_safe_bm[219] = {
1692 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1693 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1694 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1695 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1696 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1697 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1698 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1699 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1700 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1701 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1702 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1703 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1704 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1705 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1706 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFF03F,
1707 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1708 +       0xFFFFFFFF, 0xFFFFEFCE, 0xF00EBFFF, 0x007C0000,
1709 +       0xF0000078, 0xFF000009, 0xFFFFFFFF, 0xFFFFFFFF,
1710 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1711 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1712 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1713 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1714 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1715 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1716 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1717 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1718 +       0xFFFFF7FF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1719 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1720 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1721 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1722 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1723 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1724 +       0xFFFFFC78, 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF,
1725 +       0x38FF8F50, 0xFFF88082, 0xF000000C, 0xFAE009FF,
1726 +       0x0000FFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000,
1727 +       0x00000000, 0x00000100, 0x00000000, 0x00000000,
1728 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1729 +       0x00000000, 0x00000000, 0x00000000, 0xFF800000,
1730 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1731 +       0x0003FC01, 0xFFFFFCF8, 0xFF800B19, 0xFFFFFFFF,
1732 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1733 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1734 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1735 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1736 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1737 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1738 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1739 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1740 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1741 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1742 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1743 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1744 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1745 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1746 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1747 +};
1748 diff -NurpP --minimal linux-2.6.36/drivers/gpu/drm/radeon/rv515_reg_safe.h linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rv515_reg_safe.h
1749 --- linux-2.6.36/drivers/gpu/drm/radeon/rv515_reg_safe.h        1970-01-01 01:00:00.000000000 +0100
1750 +++ linux-2.6.36-vs2.3.0.36.38/drivers/gpu/drm/radeon/rv515_reg_safe.h  2010-10-21 13:09:36.000000000 +0200
1751 @@ -0,0 +1,57 @@
1752 +static const unsigned rv515_reg_safe_bm[219] = {
1753 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1754 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1755 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1756 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1757 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1758 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1759 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1760 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1761 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1762 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1763 +       0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF,
1764 +       0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF,
1765 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1766 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1767 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFF03F,
1768 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1769 +       0xFFFFFFFF, 0xFFFFEFCE, 0xF00EBFFF, 0x007C0000,
1770 +       0xF0000038, 0xFF000009, 0xFFFFFFFF, 0xFFFFFFFF,
1771 +       0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, 0xFFFFFFFF,
1772 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1773 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1774 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1775 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1776 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1777 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1778 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1779 +       0xFFFFF7FF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1780 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1781 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1782 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1783 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1784 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1785 +       0x1FFFF878, 0xFFFFE000, 0xFFFFFE1E, 0xFFFFFFFF,
1786 +       0x388F8F50, 0xFFF88082, 0xFF0000FC, 0xFAE009FF,
1787 +       0x0000FFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000,
1788 +       0xFFFF8CFC, 0xFFFFC1FF, 0xFFFFFFFF, 0xFFFFFFFF,
1789 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1790 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFF80FFFF,
1791 +       0x00000000, 0x00000000, 0x00000000, 0x00000000,
1792 +       0x0003FC01, 0x3FFFFCF8, 0xFF800B19, 0xFFDFFFFF,
1793 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1794 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1795 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1796 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1797 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1798 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1799 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1800 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1801 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1802 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1803 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1804 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1805 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1806 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1807 +       0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
1808 +};
1809 diff -NurpP --minimal linux-2.6.36/drivers/infiniband/core/addr.c linux-2.6.36-vs2.3.0.36.38/drivers/infiniband/core/addr.c
1810 --- linux-2.6.36/drivers/infiniband/core/addr.c 2010-10-21 13:07:05.000000000 +0200
1811 +++ linux-2.6.36-vs2.3.0.36.38/drivers/infiniband/core/addr.c   2010-10-21 13:09:36.000000000 +0200
1812 @@ -252,7 +252,7 @@ static int addr6_resolve(struct sockaddr
1813
1814         if (ipv6_addr_any(&fl.fl6_src)) {
1815                 ret = ipv6_dev_get_saddr(&init_net, ip6_dst_idev(dst)->dev,
1816 -                                        &fl.fl6_dst, 0, &fl.fl6_src);
1817 +                                        &fl.fl6_dst, 0, &fl.fl6_src, NULL);
1818                 if (ret)
1819                         goto put;
1820
1821 diff -NurpP --minimal linux-2.6.36/drivers/infiniband/hw/ipath/ipath_user_pages.c linux-2.6.36-vs2.3.0.36.38/drivers/infiniband/hw/ipath/ipath_user_pages.c
1822 --- linux-2.6.36/drivers/infiniband/hw/ipath/ipath_user_pages.c 2010-07-07 18:31:19.000000000 +0200
1823 +++ linux-2.6.36-vs2.3.0.36.38/drivers/infiniband/hw/ipath/ipath_user_pages.c   2010-10-21 13:09:36.000000000 +0200
1824 @@ -35,6 +35,7 @@
1825  #include <linux/device.h>
1826  #include <linux/slab.h>
1827  #include <linux/sched.h>
1828 +#include <linux/vs_memory.h>
1829
1830  #include "ipath_kernel.h"
1831
1832 diff -NurpP --minimal linux-2.6.36/drivers/md/dm-ioctl.c linux-2.6.36-vs2.3.0.36.38/drivers/md/dm-ioctl.c
1833 --- linux-2.6.36/drivers/md/dm-ioctl.c  2010-10-21 13:07:07.000000000 +0200
1834 +++ linux-2.6.36-vs2.3.0.36.38/drivers/md/dm-ioctl.c    2010-10-21 13:09:36.000000000 +0200
1835 @@ -16,6 +16,7 @@
1836  #include <linux/dm-ioctl.h>
1837  #include <linux/hdreg.h>
1838  #include <linux/compat.h>
1839 +#include <linux/vs_context.h>
1840
1841  #include <asm/uaccess.h>
1842
1843 @@ -106,7 +107,8 @@ static struct hash_cell *__get_name_cell
1844         unsigned int h = hash_str(str);
1845
1846         list_for_each_entry (hc, _name_buckets + h, name_list)
1847 -               if (!strcmp(hc->name, str)) {
1848 +               if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1849 +                       !strcmp(hc->name, str)) {
1850                         dm_get(hc->md);
1851                         return hc;
1852                 }
1853 @@ -120,7 +122,8 @@ static struct hash_cell *__get_uuid_cell
1854         unsigned int h = hash_str(str);
1855
1856         list_for_each_entry (hc, _uuid_buckets + h, uuid_list)
1857 -               if (!strcmp(hc->uuid, str)) {
1858 +               if (vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT) &&
1859 +                       !strcmp(hc->uuid, str)) {
1860                         dm_get(hc->md);
1861                         return hc;
1862                 }
1863 @@ -377,6 +380,9 @@ typedef int (*ioctl_fn)(struct dm_ioctl
1864
1865  static int remove_all(struct dm_ioctl *param, size_t param_size)
1866  {
1867 +       if (!vx_check(0, VS_ADMIN))
1868 +               return -EPERM;
1869 +
1870         dm_hash_remove_all(1);
1871         param->data_size = 0;
1872         return 0;
1873 @@ -424,6 +430,8 @@ static int list_devices(struct dm_ioctl
1874          */
1875         for (i = 0; i < NUM_BUCKETS; i++) {
1876                 list_for_each_entry (hc, _name_buckets + i, name_list) {
1877 +                       if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1878 +                               continue;
1879                         needed += sizeof(struct dm_name_list);
1880                         needed += strlen(hc->name) + 1;
1881                         needed += ALIGN_MASK;
1882 @@ -447,6 +455,8 @@ static int list_devices(struct dm_ioctl
1883          */
1884         for (i = 0; i < NUM_BUCKETS; i++) {
1885                 list_for_each_entry (hc, _name_buckets + i, name_list) {
1886 +                       if (!vx_check(dm_get_xid(hc->md), VS_WATCH_P | VS_IDENT))
1887 +                               continue;
1888                         if (old_nl)
1889                                 old_nl->next = (uint32_t) ((void *) nl -
1890                                                            (void *) old_nl);
1891 @@ -681,10 +691,11 @@ static struct hash_cell *__find_device_h
1892         if (!md)
1893                 goto out;
1894
1895 -       mdptr = dm_get_mdptr(md);
1896 +       if (vx_check(dm_get_xid(md), VS_WATCH_P | VS_IDENT))
1897 +               mdptr = dm_get_mdptr(md);
1898 +
1899         if (!mdptr)
1900                 dm_put(md);
1901 -
1902  out:
1903         return mdptr;
1904  }
1905 @@ -1513,8 +1524,8 @@ static int ctl_ioctl(uint command, struc
1906         ioctl_fn fn = NULL;
1907         size_t param_size;
1908
1909 -       /* only root can play with this */
1910 -       if (!capable(CAP_SYS_ADMIN))
1911 +       /* only root and certain contexts can play with this */
1912 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_ADMIN_MAPPER))
1913                 return -EACCES;
1914
1915         if (_IOC_TYPE(command) != DM_IOCTL)
1916 diff -NurpP --minimal linux-2.6.36/drivers/md/dm.c linux-2.6.36-vs2.3.0.36.38/drivers/md/dm.c
1917 --- linux-2.6.36/drivers/md/dm.c        2010-10-21 13:07:07.000000000 +0200
1918 +++ linux-2.6.36-vs2.3.0.36.38/drivers/md/dm.c  2010-10-21 15:19:30.000000000 +0200
1919 @@ -21,6 +21,7 @@
1920  #include <linux/idr.h>
1921  #include <linux/hdreg.h>
1922  #include <linux/delay.h>
1923 +#include <linux/vs_base.h>
1924
1925  #include <trace/events/block.h>
1926
1927 @@ -121,6 +122,7 @@ struct mapped_device {
1928         rwlock_t map_lock;
1929         atomic_t holders;
1930         atomic_t open_count;
1931 +       xid_t xid;
1932
1933         unsigned long flags;
1934
1935 @@ -343,6 +345,7 @@ int dm_deleting_md(struct mapped_device
1936  static int dm_blk_open(struct block_device *bdev, fmode_t mode)
1937  {
1938         struct mapped_device *md;
1939 +       int ret = -ENXIO;
1940
1941         lock_kernel();
1942         spin_lock(&_minor_lock);
1943 @@ -352,19 +355,20 @@ static int dm_blk_open(struct block_devi
1944                 goto out;
1945
1946         if (test_bit(DMF_FREEING, &md->flags) ||
1947 -           dm_deleting_md(md)) {
1948 -               md = NULL;
1949 +           dm_deleting_md(md))
1950 +               goto out;
1951 +
1952 +       ret = -EACCES;
1953 +       if (!vx_check(md->xid, VS_IDENT|VS_HOSTID))
1954                 goto out;
1955 -       }
1956
1957         dm_get(md);
1958         atomic_inc(&md->open_count);
1959 -
1960 +       ret = 0;
1961  out:
1962         spin_unlock(&_minor_lock);
1963         unlock_kernel();
1964 -
1965 -       return md ? 0 : -ENXIO;
1966 +       return ret;
1967  }
1968
1969  static int dm_blk_close(struct gendisk *disk, fmode_t mode)
1970 @@ -586,6 +590,14 @@ int dm_set_geometry(struct mapped_device
1971         return 0;
1972  }
1973
1974 +/*
1975 + * Get the xid associated with a dm device
1976 + */
1977 +xid_t dm_get_xid(struct mapped_device *md)
1978 +{
1979 +       return md->xid;
1980 +}
1981 +
1982  /*-----------------------------------------------------------------
1983   * CRUD START:
1984   *   A more elegant soln is in the works that uses the queue
1985 @@ -1981,6 +1993,7 @@ static struct mapped_device *alloc_dev(i
1986         INIT_LIST_HEAD(&md->uevent_list);
1987         spin_lock_init(&md->uevent_lock);
1988
1989 +       md->xid = vx_current_xid();
1990         md->queue = blk_alloc_queue(GFP_KERNEL);
1991         if (!md->queue)
1992                 goto bad_queue;
1993 diff -NurpP --minimal linux-2.6.36/drivers/md/dm.h linux-2.6.36-vs2.3.0.36.38/drivers/md/dm.h
1994 --- linux-2.6.36/drivers/md/dm.h        2010-10-21 13:07:07.000000000 +0200
1995 +++ linux-2.6.36-vs2.3.0.36.38/drivers/md/dm.h  2010-10-21 13:09:36.000000000 +0200
1996 @@ -41,6 +41,8 @@ struct dm_dev_internal {
1997  struct dm_table;
1998  struct dm_md_mempools;
1999
2000 +xid_t dm_get_xid(struct mapped_device *md);
2001 +
2002  /*-----------------------------------------------------------------
2003   * Internal table functions.
2004   *---------------------------------------------------------------*/
2005 diff -NurpP --minimal linux-2.6.36/drivers/net/tun.c linux-2.6.36-vs2.3.0.36.38/drivers/net/tun.c
2006 --- linux-2.6.36/drivers/net/tun.c      2010-10-21 13:07:15.000000000 +0200
2007 +++ linux-2.6.36-vs2.3.0.36.38/drivers/net/tun.c        2010-10-21 13:09:36.000000000 +0200
2008 @@ -62,6 +62,7 @@
2009  #include <linux/nsproxy.h>
2010  #include <linux/virtio_net.h>
2011  #include <linux/rcupdate.h>
2012 +#include <linux/vs_network.h>
2013  #include <net/net_namespace.h>
2014  #include <net/netns/generic.h>
2015  #include <net/rtnetlink.h>
2016 @@ -103,6 +104,7 @@ struct tun_struct {
2017         unsigned int            flags;
2018         uid_t                   owner;
2019         gid_t                   group;
2020 +       nid_t                   nid;
2021
2022         struct net_device       *dev;
2023         struct fasync_struct    *fasync;
2024 @@ -853,6 +855,7 @@ static void tun_setup(struct net_device
2025
2026         tun->owner = -1;
2027         tun->group = -1;
2028 +       tun->nid = current->nid;
2029
2030         dev->ethtool_ops = &tun_ethtool_ops;
2031         dev->destructor = tun_free_netdev;
2032 @@ -1003,7 +1006,7 @@ static int tun_set_iff(struct net *net,
2033
2034                 if (((tun->owner != -1 && cred->euid != tun->owner) ||
2035                      (tun->group != -1 && !in_egroup_p(tun->group))) &&
2036 -                   !capable(CAP_NET_ADMIN))
2037 +               !cap_raised(current_cap(), CAP_NET_ADMIN))
2038                         return -EPERM;
2039                 err = security_tun_dev_attach(tun->socket.sk);
2040                 if (err < 0)
2041 @@ -1017,7 +1020,7 @@ static int tun_set_iff(struct net *net,
2042                 char *name;
2043                 unsigned long flags = 0;
2044
2045 -               if (!capable(CAP_NET_ADMIN))
2046 +               if (!nx_capable(CAP_NET_ADMIN, NXC_TUN_CREATE))
2047                         return -EPERM;
2048                 err = security_tun_dev_create();
2049                 if (err < 0)
2050 @@ -1087,6 +1090,9 @@ static int tun_set_iff(struct net *net,
2051
2052                 sk->sk_destruct = tun_sock_destruct;
2053
2054 +               if (!nx_check(tun->nid, VS_IDENT | VS_HOSTID | VS_ADMIN_P))
2055 +                       return -EPERM;
2056 +
2057                 err = tun_attach(tun, file);
2058                 if (err < 0)
2059                         goto failed;
2060 @@ -1278,6 +1284,16 @@ static long __tun_chr_ioctl(struct file
2061                 DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
2062                 break;
2063
2064 +       case TUNSETNID:
2065 +               if (!capable(CAP_CONTEXT))
2066 +                       return -EPERM;
2067 +
2068 +               /* Set nid owner of the device */
2069 +               tun->nid = (nid_t) arg;
2070 +
2071 +               DBG(KERN_INFO "%s: nid owner set to %u\n", tun->dev->name, tun->nid);
2072 +               break;
2073 +
2074         case TUNSETLINK:
2075                 /* Only allow setting the type when the interface is down */
2076                 if (tun->dev->flags & IFF_UP) {
2077 diff -NurpP --minimal linux-2.6.36/fs/attr.c linux-2.6.36-vs2.3.0.36.38/fs/attr.c
2078 --- linux-2.6.36/fs/attr.c      2010-10-21 13:07:47.000000000 +0200
2079 +++ linux-2.6.36-vs2.3.0.36.38/fs/attr.c        2010-10-21 14:04:40.000000000 +0200
2080 @@ -13,6 +13,9 @@
2081  #include <linux/fsnotify.h>
2082  #include <linux/fcntl.h>
2083  #include <linux/security.h>
2084 +#include <linux/proc_fs.h>
2085 +#include <linux/devpts_fs.h>
2086 +#include <linux/vs_tag.h>
2087
2088  /**
2089   * inode_change_ok - check if attribute changes to an inode are allowed
2090 @@ -73,6 +76,10 @@ int inode_change_ok(const struct inode *
2091                         return -EPERM;
2092         }
2093
2094 +       /* check for inode tag permission */
2095 +       if (dx_permission(inode, MAY_WRITE))
2096 +               return -EACCES;
2097 +
2098         return 0;
2099  }
2100  EXPORT_SYMBOL(inode_change_ok);
2101 @@ -143,6 +150,8 @@ void setattr_copy(struct inode *inode, c
2102                 inode->i_uid = attr->ia_uid;
2103         if (ia_valid & ATTR_GID)
2104                 inode->i_gid = attr->ia_gid;
2105 +       if ((ia_valid & ATTR_TAG) && IS_TAGGED(inode))
2106 +               inode->i_tag = attr->ia_tag;
2107         if (ia_valid & ATTR_ATIME)
2108                 inode->i_atime = timespec_trunc(attr->ia_atime,
2109                                                 inode->i_sb->s_time_gran);
2110 @@ -170,7 +179,8 @@ int notify_change(struct dentry * dentry
2111         struct timespec now;
2112         unsigned int ia_valid = attr->ia_valid;
2113
2114 -       if (ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID | ATTR_TIMES_SET)) {
2115 +       if (ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID |
2116 +               ATTR_TAG | ATTR_TIMES_SET)) {
2117                 if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
2118                         return -EPERM;
2119         }
2120 diff -NurpP --minimal linux-2.6.36/fs/binfmt_aout.c linux-2.6.36-vs2.3.0.36.38/fs/binfmt_aout.c
2121 --- linux-2.6.36/fs/binfmt_aout.c       2010-10-21 13:07:47.000000000 +0200
2122 +++ linux-2.6.36-vs2.3.0.36.38/fs/binfmt_aout.c 2010-10-21 13:09:36.000000000 +0200
2123 @@ -25,6 +25,7 @@
2124  #include <linux/init.h>
2125  #include <linux/coredump.h>
2126  #include <linux/slab.h>
2127 +#include <linux/vs_memory.h>
2128
2129  #include <asm/system.h>
2130  #include <asm/uaccess.h>
2131 diff -NurpP --minimal linux-2.6.36/fs/binfmt_elf.c linux-2.6.36-vs2.3.0.36.38/fs/binfmt_elf.c
2132 --- linux-2.6.36/fs/binfmt_elf.c        2010-07-07 18:31:50.000000000 +0200
2133 +++ linux-2.6.36-vs2.3.0.36.38/fs/binfmt_elf.c  2010-10-21 13:09:36.000000000 +0200
2134 @@ -32,6 +32,7 @@
2135  #include <linux/elf.h>
2136  #include <linux/utsname.h>
2137  #include <linux/coredump.h>
2138 +#include <linux/vs_memory.h>
2139  #include <asm/uaccess.h>
2140  #include <asm/param.h>
2141  #include <asm/page.h>
2142 diff -NurpP --minimal linux-2.6.36/fs/binfmt_flat.c linux-2.6.36-vs2.3.0.36.38/fs/binfmt_flat.c
2143 --- linux-2.6.36/fs/binfmt_flat.c       2010-08-02 16:52:47.000000000 +0200
2144 +++ linux-2.6.36-vs2.3.0.36.38/fs/binfmt_flat.c 2010-10-21 13:09:36.000000000 +0200
2145 @@ -35,6 +35,7 @@
2146  #include <linux/init.h>
2147  #include <linux/flat.h>
2148  #include <linux/syscalls.h>
2149 +#include <linux/vs_memory.h>
2150
2151  #include <asm/byteorder.h>
2152  #include <asm/system.h>
2153 diff -NurpP --minimal linux-2.6.36/fs/binfmt_som.c linux-2.6.36-vs2.3.0.36.38/fs/binfmt_som.c
2154 --- linux-2.6.36/fs/binfmt_som.c        2010-02-25 11:52:04.000000000 +0100
2155 +++ linux-2.6.36-vs2.3.0.36.38/fs/binfmt_som.c  2010-10-21 13:09:36.000000000 +0200
2156 @@ -28,6 +28,7 @@
2157  #include <linux/shm.h>
2158  #include <linux/personality.h>
2159  #include <linux/init.h>
2160 +#include <linux/vs_memory.h>
2161
2162  #include <asm/uaccess.h>
2163  #include <asm/pgtable.h>
2164 diff -NurpP --minimal linux-2.6.36/fs/block_dev.c linux-2.6.36-vs2.3.0.36.38/fs/block_dev.c
2165 --- linux-2.6.36/fs/block_dev.c 2010-10-21 13:07:47.000000000 +0200
2166 +++ linux-2.6.36-vs2.3.0.36.38/fs/block_dev.c   2010-10-21 13:09:36.000000000 +0200
2167 @@ -26,6 +26,7 @@
2168  #include <linux/namei.h>
2169  #include <linux/log2.h>
2170  #include <linux/kmemleak.h>
2171 +#include <linux/vs_device.h>
2172  #include <asm/uaccess.h>
2173  #include "internal.h"
2174
2175 @@ -530,6 +531,7 @@ struct block_device *bdget(dev_t dev)
2176                 bdev->bd_invalidated = 0;
2177                 inode->i_mode = S_IFBLK;
2178                 inode->i_rdev = dev;
2179 +               inode->i_mdev = dev;
2180                 inode->i_bdev = bdev;
2181                 inode->i_data.a_ops = &def_blk_aops;
2182                 mapping_set_gfp_mask(&inode->i_data, GFP_USER);
2183 @@ -576,6 +578,11 @@ EXPORT_SYMBOL(bdput);
2184  static struct block_device *bd_acquire(struct inode *inode)
2185  {
2186         struct block_device *bdev;
2187 +       dev_t mdev;
2188 +
2189 +       if (!vs_map_blkdev(inode->i_rdev, &mdev, DATTR_OPEN))
2190 +               return NULL;
2191 +       inode->i_mdev = mdev;
2192
2193         spin_lock(&bdev_lock);
2194         bdev = inode->i_bdev;
2195 @@ -586,7 +593,7 @@ static struct block_device *bd_acquire(s
2196         }
2197         spin_unlock(&bdev_lock);
2198
2199 -       bdev = bdget(inode->i_rdev);
2200 +       bdev = bdget(mdev);
2201         if (bdev) {
2202                 spin_lock(&bdev_lock);
2203                 if (!inode->i_bdev) {
2204 diff -NurpP --minimal linux-2.6.36/fs/btrfs/ctree.h linux-2.6.36-vs2.3.0.36.38/fs/btrfs/ctree.h
2205 --- linux-2.6.36/fs/btrfs/ctree.h       2010-10-21 13:07:48.000000000 +0200
2206 +++ linux-2.6.36-vs2.3.0.36.38/fs/btrfs/ctree.h 2010-10-21 13:09:36.000000000 +0200
2207 @@ -551,11 +551,14 @@ struct btrfs_inode_item {
2208         /* modification sequence number for NFS */
2209         __le64 sequence;
2210
2211 +       __le16 tag;
2212         /*
2213          * a little future expansion, for more than this we can
2214          * just grow the inode item and version it
2215          */
2216 -       __le64 reserved[4];
2217 +       __le16 reserved16;
2218 +       __le32 reserved32;
2219 +       __le64 reserved[3];
2220         struct btrfs_timespec atime;
2221         struct btrfs_timespec ctime;
2222         struct btrfs_timespec mtime;
2223 @@ -1193,6 +1196,8 @@ struct btrfs_root {
2224  #define BTRFS_MOUNT_DISCARD            (1 << 10)
2225  #define BTRFS_MOUNT_FORCE_COMPRESS      (1 << 11)
2226
2227 +#define BTRFS_MOUNT_TAGGED             (1 << 24)
2228 +
2229  #define btrfs_clear_opt(o, opt)                ((o) &= ~BTRFS_MOUNT_##opt)
2230  #define btrfs_set_opt(o, opt)          ((o) |= BTRFS_MOUNT_##opt)
2231  #define btrfs_test_opt(root, opt)      ((root)->fs_info->mount_opt & \
2232 @@ -1413,6 +1418,7 @@ BTRFS_SETGET_FUNCS(inode_block_group, st
2233  BTRFS_SETGET_FUNCS(inode_nlink, struct btrfs_inode_item, nlink, 32);
2234  BTRFS_SETGET_FUNCS(inode_uid, struct btrfs_inode_item, uid, 32);
2235  BTRFS_SETGET_FUNCS(inode_gid, struct btrfs_inode_item, gid, 32);
2236 +BTRFS_SETGET_FUNCS(inode_tag, struct btrfs_inode_item, tag, 16);
2237  BTRFS_SETGET_FUNCS(inode_mode, struct btrfs_inode_item, mode, 32);
2238  BTRFS_SETGET_FUNCS(inode_rdev, struct btrfs_inode_item, rdev, 64);
2239  BTRFS_SETGET_FUNCS(inode_flags, struct btrfs_inode_item, flags, 64);
2240 @@ -1474,6 +1480,10 @@ BTRFS_SETGET_FUNCS(extent_flags, struct
2241
2242  BTRFS_SETGET_FUNCS(extent_refs_v0, struct btrfs_extent_item_v0, refs, 32);
2243
2244 +#define BTRFS_INODE_IXUNLINK           (1 << 24)
2245 +#define BTRFS_INODE_BARRIER            (1 << 25)
2246 +#define BTRFS_INODE_COW                        (1 << 26)
2247 +
2248
2249  BTRFS_SETGET_FUNCS(tree_block_level, struct btrfs_tree_block_info, level, 8);
2250
2251 @@ -2432,6 +2442,7 @@ extern const struct dentry_operations bt
2252  long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2253  void btrfs_update_iflags(struct inode *inode);
2254  void btrfs_inherit_iflags(struct inode *inode, struct inode *dir);
2255 +int btrfs_sync_flags(struct inode *inode, int, int);
2256
2257  /* file.c */
2258  int btrfs_sync_file(struct file *file, int datasync);
2259 diff -NurpP --minimal linux-2.6.36/fs/btrfs/disk-io.c linux-2.6.36-vs2.3.0.36.38/fs/btrfs/disk-io.c
2260 --- linux-2.6.36/fs/btrfs/disk-io.c     2010-10-21 13:07:48.000000000 +0200
2261 +++ linux-2.6.36-vs2.3.0.36.38/fs/btrfs/disk-io.c       2010-10-21 13:09:36.000000000 +0200
2262 @@ -1707,6 +1707,9 @@ struct btrfs_root *open_ctree(struct sup
2263                 goto fail_iput;
2264         }
2265
2266 +       if (btrfs_test_opt(tree_root, TAGGED))
2267 +               sb->s_flags |= MS_TAGGED;
2268 +
2269         features = btrfs_super_incompat_flags(disk_super) &
2270                 ~BTRFS_FEATURE_INCOMPAT_SUPP;
2271         if (features) {
2272 diff -NurpP --minimal linux-2.6.36/fs/btrfs/inode.c linux-2.6.36-vs2.3.0.36.38/fs/btrfs/inode.c
2273 --- linux-2.6.36/fs/btrfs/inode.c       2010-10-21 13:07:48.000000000 +0200
2274 +++ linux-2.6.36-vs2.3.0.36.38/fs/btrfs/inode.c 2010-10-21 13:09:36.000000000 +0200
2275 @@ -37,6 +37,8 @@
2276  #include <linux/posix_acl.h>
2277  #include <linux/falloc.h>
2278  #include <linux/slab.h>
2279 +#include <linux/vs_tag.h>
2280 +
2281  #include "compat.h"
2282  #include "ctree.h"
2283  #include "disk-io.h"
2284 @@ -2415,6 +2417,8 @@ static void btrfs_read_locked_inode(stru
2285         int maybe_acls;
2286         u64 alloc_group_block;
2287         u32 rdev;
2288 +       uid_t uid;
2289 +       gid_t gid;
2290         int ret;
2291
2292         path = btrfs_alloc_path();
2293 @@ -2431,8 +2435,13 @@ static void btrfs_read_locked_inode(stru
2294
2295         inode->i_mode = btrfs_inode_mode(leaf, inode_item);
2296         inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
2297 -       inode->i_uid = btrfs_inode_uid(leaf, inode_item);
2298 -       inode->i_gid = btrfs_inode_gid(leaf, inode_item);
2299 +
2300 +       uid = btrfs_inode_uid(leaf, inode_item);
2301 +       gid = btrfs_inode_gid(leaf, inode_item);
2302 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2303 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2304 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2305 +               btrfs_inode_tag(leaf, inode_item));
2306         btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item));
2307
2308         tspec = btrfs_inode_atime(inode_item);
2309 @@ -2514,8 +2523,15 @@ static void fill_inode_item(struct btrfs
2310                             struct btrfs_inode_item *item,
2311                             struct inode *inode)
2312  {
2313 -       btrfs_set_inode_uid(leaf, item, inode->i_uid);
2314 -       btrfs_set_inode_gid(leaf, item, inode->i_gid);
2315 +       uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2316 +       gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2317 +
2318 +       btrfs_set_inode_uid(leaf, item, uid);
2319 +       btrfs_set_inode_gid(leaf, item, gid);
2320 +#ifdef CONFIG_TAGGING_INTERN
2321 +       btrfs_set_inode_tag(leaf, item, inode->i_tag);
2322 +#endif
2323 +
2324         btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size);
2325         btrfs_set_inode_mode(leaf, item, inode->i_mode);
2326         btrfs_set_inode_nlink(leaf, item, inode->i_nlink);
2327 @@ -6946,6 +6962,7 @@ static const struct inode_operations btr
2328         .listxattr      = btrfs_listxattr,
2329         .removexattr    = btrfs_removexattr,
2330         .permission     = btrfs_permission,
2331 +       .sync_flags     = btrfs_sync_flags,
2332  };
2333  static const struct inode_operations btrfs_dir_ro_inode_operations = {
2334         .lookup         = btrfs_lookup,
2335 @@ -7021,6 +7038,7 @@ static const struct inode_operations btr
2336         .permission     = btrfs_permission,
2337         .fallocate      = btrfs_fallocate,
2338         .fiemap         = btrfs_fiemap,
2339 +       .sync_flags     = btrfs_sync_flags,
2340  };
2341  static const struct inode_operations btrfs_special_inode_operations = {
2342         .getattr        = btrfs_getattr,
2343 diff -NurpP --minimal linux-2.6.36/fs/btrfs/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/btrfs/ioctl.c
2344 --- linux-2.6.36/fs/btrfs/ioctl.c       2010-08-02 16:52:47.000000000 +0200
2345 +++ linux-2.6.36-vs2.3.0.36.38/fs/btrfs/ioctl.c 2010-10-21 13:09:36.000000000 +0200
2346 @@ -68,10 +68,13 @@ static unsigned int btrfs_flags_to_ioctl
2347  {
2348         unsigned int iflags = 0;
2349
2350 -       if (flags & BTRFS_INODE_SYNC)
2351 -               iflags |= FS_SYNC_FL;
2352         if (flags & BTRFS_INODE_IMMUTABLE)
2353                 iflags |= FS_IMMUTABLE_FL;
2354 +       if (flags & BTRFS_INODE_IXUNLINK)
2355 +               iflags |= FS_IXUNLINK_FL;
2356 +
2357 +       if (flags & BTRFS_INODE_SYNC)
2358 +               iflags |= FS_SYNC_FL;
2359         if (flags & BTRFS_INODE_APPEND)
2360                 iflags |= FS_APPEND_FL;
2361         if (flags & BTRFS_INODE_NODUMP)
2362 @@ -81,28 +84,78 @@ static unsigned int btrfs_flags_to_ioctl
2363         if (flags & BTRFS_INODE_DIRSYNC)
2364                 iflags |= FS_DIRSYNC_FL;
2365
2366 +       if (flags & BTRFS_INODE_BARRIER)
2367 +               iflags |= FS_BARRIER_FL;
2368 +       if (flags & BTRFS_INODE_COW)
2369 +               iflags |= FS_COW_FL;
2370         return iflags;
2371  }
2372
2373  /*
2374 - * Update inode->i_flags based on the btrfs internal flags.
2375 + * Update inode->i_(v)flags based on the btrfs internal flags.
2376   */
2377  void btrfs_update_iflags(struct inode *inode)
2378  {
2379         struct btrfs_inode *ip = BTRFS_I(inode);
2380
2381 -       inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2382 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2383 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2384
2385 -       if (ip->flags & BTRFS_INODE_SYNC)
2386 -               inode->i_flags |= S_SYNC;
2387         if (ip->flags & BTRFS_INODE_IMMUTABLE)
2388                 inode->i_flags |= S_IMMUTABLE;
2389 +       if (ip->flags & BTRFS_INODE_IXUNLINK)
2390 +               inode->i_flags |= S_IXUNLINK;
2391 +
2392 +       if (ip->flags & BTRFS_INODE_SYNC)
2393 +               inode->i_flags |= S_SYNC;
2394         if (ip->flags & BTRFS_INODE_APPEND)
2395                 inode->i_flags |= S_APPEND;
2396         if (ip->flags & BTRFS_INODE_NOATIME)
2397                 inode->i_flags |= S_NOATIME;
2398         if (ip->flags & BTRFS_INODE_DIRSYNC)
2399                 inode->i_flags |= S_DIRSYNC;
2400 +
2401 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
2402 +
2403 +       if (ip->flags & BTRFS_INODE_BARRIER)
2404 +               inode->i_vflags |= V_BARRIER;
2405 +       if (ip->flags & BTRFS_INODE_COW)
2406 +               inode->i_vflags |= V_COW;
2407 +}
2408 +
2409 +/*
2410 + * Update btrfs internal flags from inode->i_(v)flags.
2411 + */
2412 +void btrfs_update_flags(struct inode *inode)
2413 +{
2414 +       struct btrfs_inode *ip = BTRFS_I(inode);
2415 +
2416 +       unsigned int flags = inode->i_flags;
2417 +       unsigned int vflags = inode->i_vflags;
2418 +
2419 +       ip->flags &= ~(BTRFS_INODE_SYNC | BTRFS_INODE_APPEND |
2420 +                       BTRFS_INODE_IMMUTABLE | BTRFS_INODE_IXUNLINK |
2421 +                       BTRFS_INODE_NOATIME | BTRFS_INODE_DIRSYNC |
2422 +                       BTRFS_INODE_BARRIER | BTRFS_INODE_COW);
2423 +
2424 +       if (flags & S_IMMUTABLE)
2425 +               ip->flags |= BTRFS_INODE_IMMUTABLE;
2426 +       if (flags & S_IXUNLINK)
2427 +               ip->flags |= BTRFS_INODE_IXUNLINK;
2428 +
2429 +       if (flags & S_SYNC)
2430 +               ip->flags |= BTRFS_INODE_SYNC;
2431 +       if (flags & S_APPEND)
2432 +               ip->flags |= BTRFS_INODE_APPEND;
2433 +       if (flags & S_NOATIME)
2434 +               ip->flags |= BTRFS_INODE_NOATIME;
2435 +       if (flags & S_DIRSYNC)
2436 +               ip->flags |= BTRFS_INODE_DIRSYNC;
2437 +
2438 +       if (vflags & V_BARRIER)
2439 +               ip->flags |= BTRFS_INODE_BARRIER;
2440 +       if (vflags & V_COW)
2441 +               ip->flags |= BTRFS_INODE_COW;
2442  }
2443
2444  /*
2445 @@ -120,7 +173,7 @@ void btrfs_inherit_iflags(struct inode *
2446         flags = BTRFS_I(dir)->flags;
2447
2448         if (S_ISREG(inode->i_mode))
2449 -               flags &= ~BTRFS_INODE_DIRSYNC;
2450 +               flags &= ~(BTRFS_INODE_DIRSYNC | BTRFS_INODE_BARRIER);
2451         else if (!S_ISDIR(inode->i_mode))
2452                 flags &= (BTRFS_INODE_NODUMP | BTRFS_INODE_NOATIME);
2453
2454 @@ -128,6 +181,30 @@ void btrfs_inherit_iflags(struct inode *
2455         btrfs_update_iflags(inode);
2456  }
2457
2458 +int btrfs_sync_flags(struct inode *inode, int flags, int vflags)
2459 +{
2460 +       struct btrfs_inode *ip = BTRFS_I(inode);
2461 +       struct btrfs_root *root = ip->root;
2462 +       struct btrfs_trans_handle *trans;
2463 +       int ret;
2464 +
2465 +       trans = btrfs_join_transaction(root, 1);
2466 +       BUG_ON(!trans);
2467 +
2468 +       inode->i_flags = flags;
2469 +       inode->i_vflags = vflags;
2470 +       btrfs_update_flags(inode);
2471 +
2472 +       ret = btrfs_update_inode(trans, root, inode);
2473 +       BUG_ON(ret);
2474 +
2475 +       btrfs_update_iflags(inode);
2476 +       inode->i_ctime = CURRENT_TIME;
2477 +       btrfs_end_transaction(trans, root);
2478 +
2479 +       return 0;
2480 +}
2481 +
2482  static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
2483  {
2484         struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode);
2485 @@ -150,6 +227,7 @@ static int btrfs_ioctl_setflags(struct f
2486         if (copy_from_user(&flags, arg, sizeof(flags)))
2487                 return -EFAULT;
2488
2489 +       /* maybe add FS_IXUNLINK_FL ? */
2490         if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
2491                       FS_NOATIME_FL | FS_NODUMP_FL | \
2492                       FS_SYNC_FL | FS_DIRSYNC_FL))
2493 @@ -162,7 +240,8 @@ static int btrfs_ioctl_setflags(struct f
2494
2495         flags = btrfs_mask_flags(inode->i_mode, flags);
2496         oldflags = btrfs_flags_to_ioctl(ip->flags);
2497 -       if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
2498 +       if ((flags ^ oldflags) & (FS_APPEND_FL |
2499 +               FS_IMMUTABLE_FL | FS_IXUNLINK_FL)) {
2500                 if (!capable(CAP_LINUX_IMMUTABLE)) {
2501                         ret = -EPERM;
2502                         goto out_unlock;
2503 @@ -173,14 +252,19 @@ static int btrfs_ioctl_setflags(struct f
2504         if (ret)
2505                 goto out_unlock;
2506
2507 -       if (flags & FS_SYNC_FL)
2508 -               ip->flags |= BTRFS_INODE_SYNC;
2509 -       else
2510 -               ip->flags &= ~BTRFS_INODE_SYNC;
2511         if (flags & FS_IMMUTABLE_FL)
2512                 ip->flags |= BTRFS_INODE_IMMUTABLE;
2513         else
2514                 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
2515 +       if (flags & FS_IXUNLINK_FL)
2516 +               ip->flags |= BTRFS_INODE_IXUNLINK;
2517 +       else
2518 +               ip->flags &= ~BTRFS_INODE_IXUNLINK;
2519 +
2520 +       if (flags & FS_SYNC_FL)
2521 +               ip->flags |= BTRFS_INODE_SYNC;
2522 +       else
2523 +               ip->flags &= ~BTRFS_INODE_SYNC;
2524         if (flags & FS_APPEND_FL)
2525                 ip->flags |= BTRFS_INODE_APPEND;
2526         else
2527 diff -NurpP --minimal linux-2.6.36/fs/btrfs/super.c linux-2.6.36-vs2.3.0.36.38/fs/btrfs/super.c
2528 --- linux-2.6.36/fs/btrfs/super.c       2010-10-21 13:07:48.000000000 +0200
2529 +++ linux-2.6.36-vs2.3.0.36.38/fs/btrfs/super.c 2010-10-21 13:09:36.000000000 +0200
2530 @@ -68,7 +68,7 @@ enum {
2531         Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
2532         Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
2533         Opt_compress_force, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
2534 -       Opt_discard, Opt_err,
2535 +       Opt_tag, Opt_notag, Opt_tagid, Opt_discard, Opt_err,
2536  };
2537
2538  static match_table_t tokens = {
2539 @@ -92,6 +92,9 @@ static match_table_t tokens = {
2540         {Opt_flushoncommit, "flushoncommit"},
2541         {Opt_ratio, "metadata_ratio=%d"},
2542         {Opt_discard, "discard"},
2543 +       {Opt_tag, "tag"},
2544 +       {Opt_notag, "notag"},
2545 +       {Opt_tagid, "tagid=%u"},
2546         {Opt_err, NULL},
2547  };
2548
2549 @@ -235,6 +238,22 @@ int btrfs_parse_options(struct btrfs_roo
2550                 case Opt_discard:
2551                         btrfs_set_opt(info->mount_opt, DISCARD);
2552                         break;
2553 +#ifndef CONFIG_TAGGING_NONE
2554 +               case Opt_tag:
2555 +                       printk(KERN_INFO "btrfs: use tagging\n");
2556 +                       btrfs_set_opt(info->mount_opt, TAGGED);
2557 +                       break;
2558 +               case Opt_notag:
2559 +                       printk(KERN_INFO "btrfs: disabled tagging\n");
2560 +                       btrfs_clear_opt(info->mount_opt, TAGGED);
2561 +                       break;
2562 +#endif
2563 +#ifdef CONFIG_PROPAGATE
2564 +               case Opt_tagid:
2565 +                       /* use args[0] */
2566 +                       btrfs_set_opt(info->mount_opt, TAGGED);
2567 +                       break;
2568 +#endif
2569                 case Opt_err:
2570                         printk(KERN_INFO "btrfs: unrecognized mount option "
2571                                "'%s'\n", p);
2572 @@ -681,6 +700,12 @@ static int btrfs_remount(struct super_bl
2573         if (ret)
2574                 return -EINVAL;
2575
2576 +       if (btrfs_test_opt(root, TAGGED) && !(sb->s_flags & MS_TAGGED)) {
2577 +               printk("btrfs: %s: tagging not permitted on remount.\n",
2578 +                       sb->s_id);
2579 +               return -EINVAL;
2580 +       }
2581 +
2582         if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
2583                 return 0;
2584
2585 diff -NurpP --minimal linux-2.6.36/fs/char_dev.c linux-2.6.36-vs2.3.0.36.38/fs/char_dev.c
2586 --- linux-2.6.36/fs/char_dev.c  2010-10-21 13:07:48.000000000 +0200
2587 +++ linux-2.6.36-vs2.3.0.36.38/fs/char_dev.c    2010-10-21 13:09:36.000000000 +0200
2588 @@ -21,6 +21,8 @@
2589  #include <linux/mutex.h>
2590  #include <linux/backing-dev.h>
2591  #include <linux/tty.h>
2592 +#include <linux/vs_context.h>
2593 +#include <linux/vs_device.h>
2594
2595  #include "internal.h"
2596
2597 @@ -371,14 +373,21 @@ static int chrdev_open(struct inode *ino
2598         struct cdev *p;
2599         struct cdev *new = NULL;
2600         int ret = 0;
2601 +       dev_t mdev;
2602 +
2603 +       if (!vs_map_chrdev(inode->i_rdev, &mdev, DATTR_OPEN))
2604 +               return -EPERM;
2605 +       inode->i_mdev = mdev;
2606
2607         spin_lock(&cdev_lock);
2608         p = inode->i_cdev;
2609         if (!p) {
2610                 struct kobject *kobj;
2611                 int idx;
2612 +
2613                 spin_unlock(&cdev_lock);
2614 -               kobj = kobj_lookup(cdev_map, inode->i_rdev, &idx);
2615 +
2616 +               kobj = kobj_lookup(cdev_map, mdev, &idx);
2617                 if (!kobj)
2618                         return -ENXIO;
2619                 new = container_of(kobj, struct cdev, kobj);
2620 diff -NurpP --minimal linux-2.6.36/fs/dcache.c linux-2.6.36-vs2.3.0.36.38/fs/dcache.c
2621 --- linux-2.6.36/fs/dcache.c    2010-10-21 13:07:48.000000000 +0200
2622 +++ linux-2.6.36-vs2.3.0.36.38/fs/dcache.c      2010-10-21 13:09:36.000000000 +0200
2623 @@ -33,6 +33,7 @@
2624  #include <linux/bootmem.h>
2625  #include <linux/fs_struct.h>
2626  #include <linux/hardirq.h>
2627 +#include <linux/vs_limit.h>
2628  #include "internal.h"
2629
2630  int sysctl_vfs_cache_pressure __read_mostly = 100;
2631 @@ -230,6 +231,8 @@ repeat:
2632                 return;
2633         }
2634
2635 +       vx_dentry_dec(dentry);
2636 +
2637         /*
2638          * AV: ->d_delete() is _NOT_ allowed to block now.
2639          */
2640 @@ -323,6 +326,7 @@ static inline struct dentry * __dget_loc
2641  {
2642         atomic_inc(&dentry->d_count);
2643         dentry_lru_del_init(dentry);
2644 +       vx_dentry_inc(dentry);
2645         return dentry;
2646  }
2647
2648 @@ -928,6 +932,9 @@ struct dentry *d_alloc(struct dentry * p
2649         struct dentry *dentry;
2650         char *dname;
2651
2652 +       if (!vx_dentry_avail(1))
2653 +               return NULL;
2654 +
2655         dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL);
2656         if (!dentry)
2657                 return NULL;
2658 @@ -973,6 +980,7 @@ struct dentry *d_alloc(struct dentry * p
2659         if (parent)
2660                 list_add(&dentry->d_u.d_child, &parent->d_subdirs);
2661         dentry_stat.nr_dentry++;
2662 +       vx_dentry_inc(dentry);
2663         spin_unlock(&dcache_lock);
2664
2665         return dentry;
2666 @@ -1432,6 +1440,7 @@ struct dentry * __d_lookup(struct dentry
2667                 }
2668
2669                 atomic_inc(&dentry->d_count);
2670 +               vx_dentry_inc(dentry);
2671                 found = dentry;
2672                 spin_unlock(&dentry->d_lock);
2673                 break;
2674 diff -NurpP --minimal linux-2.6.36/fs/devpts/inode.c linux-2.6.36-vs2.3.0.36.38/fs/devpts/inode.c
2675 --- linux-2.6.36/fs/devpts/inode.c      2010-08-02 16:52:48.000000000 +0200
2676 +++ linux-2.6.36-vs2.3.0.36.38/fs/devpts/inode.c        2010-10-21 22:25:15.000000000 +0200
2677 @@ -25,6 +25,7 @@
2678  #include <linux/parser.h>
2679  #include <linux/fsnotify.h>
2680  #include <linux/seq_file.h>
2681 +#include <linux/vs_base.h>
2682
2683  #define DEVPTS_DEFAULT_MODE 0600
2684  /*
2685 @@ -36,6 +37,20 @@
2686  #define DEVPTS_DEFAULT_PTMX_MODE 0000
2687  #define PTMX_MINOR     2
2688
2689 +static int devpts_permission(struct inode *inode, int mask)
2690 +{
2691 +       int ret = -EACCES;
2692 +
2693 +       /* devpts is xid tagged */
2694 +       if (vx_check((xid_t)inode->i_tag, VS_WATCH_P | VS_IDENT))
2695 +               ret = generic_permission(inode, mask, NULL);
2696 +       return ret;
2697 +}
2698 +
2699 +static struct inode_operations devpts_file_inode_operations = {
2700 +       .permission     = devpts_permission,
2701 +};
2702 +
2703  extern int pty_limit;                  /* Config limit on Unix98 ptys */
2704  static DEFINE_MUTEX(allocated_ptys_lock);
2705
2706 @@ -263,6 +278,25 @@ static int devpts_show_options(struct se
2707         return 0;
2708  }
2709
2710 +static int devpts_filter(struct dentry *de)
2711 +{
2712 +       /* devpts is xid tagged */
2713 +       return vx_check((xid_t)de->d_inode->i_tag, VS_WATCH_P | VS_IDENT);
2714 +}
2715 +
2716 +static int devpts_readdir(struct file * filp, void * dirent, filldir_t filldir)
2717 +{
2718 +       return dcache_readdir_filter(filp, dirent, filldir, devpts_filter);
2719 +}
2720 +
2721 +static struct file_operations devpts_dir_operations = {
2722 +       .open           = dcache_dir_open,
2723 +       .release        = dcache_dir_close,
2724 +       .llseek         = dcache_dir_lseek,
2725 +       .read           = generic_read_dir,
2726 +       .readdir        = devpts_readdir,
2727 +};
2728 +
2729  static const struct super_operations devpts_sops = {
2730         .statfs         = simple_statfs,
2731         .remount_fs     = devpts_remount,
2732 @@ -302,12 +336,15 @@ devpts_fill_super(struct super_block *s,
2733         inode = new_inode(s);
2734         if (!inode)
2735                 goto free_fsi;
2736 +
2737         inode->i_ino = 1;
2738         inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2739         inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO | S_IWUSR;
2740         inode->i_op = &simple_dir_inode_operations;
2741 -       inode->i_fop = &simple_dir_operations;
2742 +       inode->i_fop = &devpts_dir_operations;
2743         inode->i_nlink = 2;
2744 +       /* devpts is xid tagged */
2745 +       inode->i_tag = (tag_t)vx_current_xid();
2746
2747         s->s_root = d_alloc_root(inode);
2748         if (s->s_root)
2749 @@ -495,6 +532,9 @@ int devpts_pty_new(struct inode *ptmx_in
2750         inode->i_gid = opts->setgid ? opts->gid : current_fsgid();
2751         inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2752         init_special_inode(inode, S_IFCHR|opts->mode, device);
2753 +       /* devpts is xid tagged */
2754 +       inode->i_tag = (tag_t)vx_current_xid();
2755 +       inode->i_op = &devpts_file_inode_operations;
2756         inode->i_private = tty;
2757         tty->driver_data = inode;
2758
2759 diff -NurpP --minimal linux-2.6.36/fs/ext2/balloc.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/balloc.c
2760 --- linux-2.6.36/fs/ext2/balloc.c       2010-10-21 13:07:48.000000000 +0200
2761 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/balloc.c 2010-10-21 13:09:36.000000000 +0200
2762 @@ -702,7 +702,6 @@ ext2_try_to_allocate(struct super_block
2763                         start = 0;
2764                 end = EXT2_BLOCKS_PER_GROUP(sb);
2765         }
2766 -
2767         BUG_ON(start > EXT2_BLOCKS_PER_GROUP(sb));
2768
2769  repeat:
2770 diff -NurpP --minimal linux-2.6.36/fs/ext2/ext2.h linux-2.6.36-vs2.3.0.36.38/fs/ext2/ext2.h
2771 --- linux-2.6.36/fs/ext2/ext2.h 2010-10-21 13:07:48.000000000 +0200
2772 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/ext2.h   2010-10-21 14:10:55.000000000 +0200
2773 @@ -127,6 +127,7 @@ extern void ext2_set_inode_flags(struct
2774  extern void ext2_get_inode_flags(struct ext2_inode_info *);
2775  extern int ext2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2776                        u64 start, u64 len);
2777 +extern int ext2_sync_flags(struct inode *, int, int);
2778
2779  /* ioctl.c */
2780  extern long ext2_ioctl(struct file *, unsigned int, unsigned long);
2781 diff -NurpP --minimal linux-2.6.36/fs/ext2/file.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/file.c
2782 --- linux-2.6.36/fs/ext2/file.c 2010-08-02 16:52:48.000000000 +0200
2783 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/file.c   2010-10-21 13:09:36.000000000 +0200
2784 @@ -104,4 +104,5 @@ const struct inode_operations ext2_file_
2785         .setattr        = ext2_setattr,
2786         .check_acl      = ext2_check_acl,
2787         .fiemap         = ext2_fiemap,
2788 +       .sync_flags     = ext2_sync_flags,
2789  };
2790 diff -NurpP --minimal linux-2.6.36/fs/ext2/ialloc.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/ialloc.c
2791 --- linux-2.6.36/fs/ext2/ialloc.c       2010-10-21 13:07:48.000000000 +0200
2792 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/ialloc.c 2010-10-21 13:09:36.000000000 +0200
2793 @@ -17,6 +17,7 @@
2794  #include <linux/backing-dev.h>
2795  #include <linux/buffer_head.h>
2796  #include <linux/random.h>
2797 +#include <linux/vs_tag.h>
2798  #include "ext2.h"
2799  #include "xattr.h"
2800  #include "acl.h"
2801 @@ -548,6 +549,7 @@ got:
2802                 inode->i_mode = mode;
2803                 inode->i_uid = current_fsuid();
2804                 inode->i_gid = dir->i_gid;
2805 +               inode->i_tag = dx_current_fstag(sb);
2806         } else
2807                 inode_init_owner(inode, dir, mode);
2808
2809 diff -NurpP --minimal linux-2.6.36/fs/ext2/inode.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/inode.c
2810 --- linux-2.6.36/fs/ext2/inode.c        2010-10-21 13:07:48.000000000 +0200
2811 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/inode.c  2010-10-21 13:09:36.000000000 +0200
2812 @@ -32,6 +32,7 @@
2813  #include <linux/mpage.h>
2814  #include <linux/fiemap.h>
2815  #include <linux/namei.h>
2816 +#include <linux/vs_tag.h>
2817  #include "ext2.h"
2818  #include "acl.h"
2819  #include "xip.h"
2820 @@ -1169,7 +1170,7 @@ static void ext2_truncate_blocks(struct
2821                 return;
2822         if (ext2_inode_is_fast_symlink(inode))
2823                 return;
2824 -       if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
2825 +       if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
2826                 return;
2827         __ext2_truncate_blocks(inode, offset);
2828  }
2829 @@ -1258,36 +1259,61 @@ void ext2_set_inode_flags(struct inode *
2830  {
2831         unsigned int flags = EXT2_I(inode)->i_flags;
2832
2833 -       inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2834 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
2835 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
2836 +
2837 +
2838 +       if (flags & EXT2_IMMUTABLE_FL)
2839 +               inode->i_flags |= S_IMMUTABLE;
2840 +       if (flags & EXT2_IXUNLINK_FL)
2841 +               inode->i_flags |= S_IXUNLINK;
2842 +
2843         if (flags & EXT2_SYNC_FL)
2844                 inode->i_flags |= S_SYNC;
2845         if (flags & EXT2_APPEND_FL)
2846                 inode->i_flags |= S_APPEND;
2847 -       if (flags & EXT2_IMMUTABLE_FL)
2848 -               inode->i_flags |= S_IMMUTABLE;
2849         if (flags & EXT2_NOATIME_FL)
2850                 inode->i_flags |= S_NOATIME;
2851         if (flags & EXT2_DIRSYNC_FL)
2852                 inode->i_flags |= S_DIRSYNC;
2853 +
2854 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
2855 +
2856 +       if (flags & EXT2_BARRIER_FL)
2857 +               inode->i_vflags |= V_BARRIER;
2858 +       if (flags & EXT2_COW_FL)
2859 +               inode->i_vflags |= V_COW;
2860  }
2861
2862  /* Propagate flags from i_flags to EXT2_I(inode)->i_flags */
2863  void ext2_get_inode_flags(struct ext2_inode_info *ei)
2864  {
2865         unsigned int flags = ei->vfs_inode.i_flags;
2866 +       unsigned int vflags = ei->vfs_inode.i_vflags;
2867 +
2868 +       ei->i_flags &= ~(EXT2_SYNC_FL | EXT2_APPEND_FL |
2869 +                       EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL |
2870 +                       EXT2_NOATIME_FL | EXT2_DIRSYNC_FL |
2871 +                       EXT2_BARRIER_FL | EXT2_COW_FL);
2872 +
2873 +       if (flags & S_IMMUTABLE)
2874 +               ei->i_flags |= EXT2_IMMUTABLE_FL;
2875 +       if (flags & S_IXUNLINK)
2876 +               ei->i_flags |= EXT2_IXUNLINK_FL;
2877
2878 -       ei->i_flags &= ~(EXT2_SYNC_FL|EXT2_APPEND_FL|
2879 -                       EXT2_IMMUTABLE_FL|EXT2_NOATIME_FL|EXT2_DIRSYNC_FL);
2880         if (flags & S_SYNC)
2881                 ei->i_flags |= EXT2_SYNC_FL;
2882         if (flags & S_APPEND)
2883                 ei->i_flags |= EXT2_APPEND_FL;
2884 -       if (flags & S_IMMUTABLE)
2885 -               ei->i_flags |= EXT2_IMMUTABLE_FL;
2886         if (flags & S_NOATIME)
2887                 ei->i_flags |= EXT2_NOATIME_FL;
2888         if (flags & S_DIRSYNC)
2889                 ei->i_flags |= EXT2_DIRSYNC_FL;
2890 +
2891 +       if (vflags & V_BARRIER)
2892 +               ei->i_flags |= EXT2_BARRIER_FL;
2893 +       if (vflags & V_COW)
2894 +               ei->i_flags |= EXT2_COW_FL;
2895  }
2896
2897  struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
2898 @@ -1297,6 +1323,8 @@ struct inode *ext2_iget (struct super_bl
2899         struct ext2_inode *raw_inode;
2900         struct inode *inode;
2901         long ret = -EIO;
2902 +       uid_t uid;
2903 +       gid_t gid;
2904         int n;
2905
2906         inode = iget_locked(sb, ino);
2907 @@ -1315,12 +1343,17 @@ struct inode *ext2_iget (struct super_bl
2908         }
2909
2910         inode->i_mode = le16_to_cpu(raw_inode->i_mode);
2911 -       inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2912 -       inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2913 +       uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
2914 +       gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
2915         if (!(test_opt (inode->i_sb, NO_UID32))) {
2916 -               inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2917 -               inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2918 +               uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
2919 +               gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
2920         }
2921 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
2922 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
2923 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
2924 +               le16_to_cpu(raw_inode->i_raw_tag));
2925 +
2926         inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
2927         inode->i_size = le32_to_cpu(raw_inode->i_size);
2928         inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
2929 @@ -1418,8 +1451,8 @@ static int __ext2_write_inode(struct ino
2930         struct ext2_inode_info *ei = EXT2_I(inode);
2931         struct super_block *sb = inode->i_sb;
2932         ino_t ino = inode->i_ino;
2933 -       uid_t uid = inode->i_uid;
2934 -       gid_t gid = inode->i_gid;
2935 +       uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
2936 +       gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
2937         struct buffer_head * bh;
2938         struct ext2_inode * raw_inode = ext2_get_inode(sb, ino, &bh);
2939         int n;
2940 @@ -1455,6 +1488,9 @@ static int __ext2_write_inode(struct ino
2941                 raw_inode->i_uid_high = 0;
2942                 raw_inode->i_gid_high = 0;
2943         }
2944 +#ifdef CONFIG_TAGGING_INTERN
2945 +       raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
2946 +#endif
2947         raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
2948         raw_inode->i_size = cpu_to_le32(inode->i_size);
2949         raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
2950 @@ -1544,7 +1580,8 @@ int ext2_setattr(struct dentry *dentry,
2951         if (is_quota_modification(inode, iattr))
2952                 dquot_initialize(inode);
2953         if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
2954 -           (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
2955 +           (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
2956 +           (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
2957                 error = dquot_transfer(inode, iattr);
2958                 if (error)
2959                         return error;
2960 diff -NurpP --minimal linux-2.6.36/fs/ext2/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/ioctl.c
2961 --- linux-2.6.36/fs/ext2/ioctl.c        2009-09-10 15:26:21.000000000 +0200
2962 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/ioctl.c  2010-10-21 13:09:36.000000000 +0200
2963 @@ -17,6 +17,16 @@
2964  #include <asm/uaccess.h>
2965
2966
2967 +int ext2_sync_flags(struct inode *inode, int flags, int vflags)
2968 +{
2969 +       inode->i_flags = flags;
2970 +       inode->i_vflags = vflags;
2971 +       ext2_get_inode_flags(EXT2_I(inode));
2972 +       inode->i_ctime = CURRENT_TIME_SEC;
2973 +       mark_inode_dirty(inode);
2974 +       return 0;
2975 +}
2976 +
2977  long ext2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
2978  {
2979         struct inode *inode = filp->f_dentry->d_inode;
2980 @@ -51,6 +61,11 @@ long ext2_ioctl(struct file *filp, unsig
2981
2982                 flags = ext2_mask_flags(inode->i_mode, flags);
2983
2984 +               if (IS_BARRIER(inode)) {
2985 +                       vxwprintk_task(1, "messing with the barrier.");
2986 +                       return -EACCES;
2987 +               }
2988 +
2989                 mutex_lock(&inode->i_mutex);
2990                 /* Is it quota file? Do not allow user to mess with it */
2991                 if (IS_NOQUOTA(inode)) {
2992 @@ -66,7 +81,9 @@ long ext2_ioctl(struct file *filp, unsig
2993                  *
2994                  * This test looks nicer. Thanks to Pauline Middelink
2995                  */
2996 -               if ((flags ^ oldflags) & (EXT2_APPEND_FL | EXT2_IMMUTABLE_FL)) {
2997 +               if ((oldflags & EXT2_IMMUTABLE_FL) ||
2998 +                       ((flags ^ oldflags) & (EXT2_APPEND_FL |
2999 +                       EXT2_IMMUTABLE_FL | EXT2_IXUNLINK_FL))) {
3000                         if (!capable(CAP_LINUX_IMMUTABLE)) {
3001                                 mutex_unlock(&inode->i_mutex);
3002                                 ret = -EPERM;
3003 @@ -74,7 +91,7 @@ long ext2_ioctl(struct file *filp, unsig
3004                         }
3005                 }
3006
3007 -               flags = flags & EXT2_FL_USER_MODIFIABLE;
3008 +               flags &= EXT2_FL_USER_MODIFIABLE;
3009                 flags |= oldflags & ~EXT2_FL_USER_MODIFIABLE;
3010                 ei->i_flags = flags;
3011                 mutex_unlock(&inode->i_mutex);
3012 diff -NurpP --minimal linux-2.6.36/fs/ext2/namei.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/namei.c
3013 --- linux-2.6.36/fs/ext2/namei.c        2010-07-07 18:31:51.000000000 +0200
3014 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/namei.c  2010-10-21 13:09:36.000000000 +0200
3015 @@ -32,6 +32,7 @@
3016
3017  #include <linux/pagemap.h>
3018  #include <linux/quotaops.h>
3019 +#include <linux/vs_tag.h>
3020  #include "ext2.h"
3021  #include "xattr.h"
3022  #include "acl.h"
3023 @@ -75,6 +76,7 @@ static struct dentry *ext2_lookup(struct
3024                                 return ERR_PTR(-EIO);
3025                         } else {
3026                                 return ERR_CAST(inode);
3027 +               dx_propagate_tag(nd, inode);
3028                         }
3029                 }
3030         }
3031 @@ -418,6 +420,7 @@ const struct inode_operations ext2_dir_i
3032  #endif
3033         .setattr        = ext2_setattr,
3034         .check_acl      = ext2_check_acl,
3035 +       .sync_flags     = ext2_sync_flags,
3036  };
3037
3038  const struct inode_operations ext2_special_inode_operations = {
3039 diff -NurpP --minimal linux-2.6.36/fs/ext2/super.c linux-2.6.36-vs2.3.0.36.38/fs/ext2/super.c
3040 --- linux-2.6.36/fs/ext2/super.c        2010-10-21 13:07:48.000000000 +0200
3041 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext2/super.c  2010-10-21 13:09:36.000000000 +0200
3042 @@ -378,7 +378,8 @@ enum {
3043         Opt_err_ro, Opt_nouid32, Opt_nocheck, Opt_debug,
3044         Opt_oldalloc, Opt_orlov, Opt_nobh, Opt_user_xattr, Opt_nouser_xattr,
3045         Opt_acl, Opt_noacl, Opt_xip, Opt_ignore, Opt_err, Opt_quota,
3046 -       Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation
3047 +       Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation,
3048 +       Opt_tag, Opt_notag, Opt_tagid
3049  };
3050
3051  static const match_table_t tokens = {
3052 @@ -406,6 +407,9 @@ static const match_table_t tokens = {
3053         {Opt_acl, "acl"},
3054         {Opt_noacl, "noacl"},
3055         {Opt_xip, "xip"},
3056 +       {Opt_tag, "tag"},
3057 +       {Opt_notag, "notag"},
3058 +       {Opt_tagid, "tagid=%u"},
3059         {Opt_grpquota, "grpquota"},
3060         {Opt_ignore, "noquota"},
3061         {Opt_quota, "quota"},
3062 @@ -476,6 +480,20 @@ static int parse_options(char *options,
3063                 case Opt_nouid32:
3064                         set_opt (sbi->s_mount_opt, NO_UID32);
3065                         break;
3066 +#ifndef CONFIG_TAGGING_NONE
3067 +               case Opt_tag:
3068 +                       set_opt (sbi->s_mount_opt, TAGGED);
3069 +                       break;
3070 +               case Opt_notag:
3071 +                       clear_opt (sbi->s_mount_opt, TAGGED);
3072 +                       break;
3073 +#endif
3074 +#ifdef CONFIG_PROPAGATE
3075 +               case Opt_tagid:
3076 +                       /* use args[0] */
3077 +                       set_opt (sbi->s_mount_opt, TAGGED);
3078 +                       break;
3079 +#endif
3080                 case Opt_nocheck:
3081                         clear_opt (sbi->s_mount_opt, CHECK);
3082                         break;
3083 @@ -833,6 +851,8 @@ static int ext2_fill_super(struct super_
3084         if (!parse_options((char *) data, sb))
3085                 goto failed_mount;
3086
3087 +       if (EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_TAGGED)
3088 +               sb->s_flags |= MS_TAGGED;
3089         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3090                 ((EXT2_SB(sb)->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ?
3091                  MS_POSIXACL : 0);
3092 @@ -1205,6 +1225,14 @@ static int ext2_remount (struct super_bl
3093                 goto restore_opts;
3094         }
3095
3096 +       if ((sbi->s_mount_opt & EXT2_MOUNT_TAGGED) &&
3097 +               !(sb->s_flags & MS_TAGGED)) {
3098 +               printk("EXT2-fs: %s: tagging not permitted on remount.\n",
3099 +                      sb->s_id);
3100 +               err = -EINVAL;
3101 +               goto restore_opts;
3102 +       }
3103 +
3104         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3105                 ((sbi->s_mount_opt & EXT2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3106
3107 diff -NurpP --minimal linux-2.6.36/fs/ext3/file.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/file.c
3108 --- linux-2.6.36/fs/ext3/file.c 2010-07-07 18:31:51.000000000 +0200
3109 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/file.c   2010-10-21 13:09:36.000000000 +0200
3110 @@ -81,5 +81,6 @@ const struct inode_operations ext3_file_
3111  #endif
3112         .check_acl      = ext3_check_acl,
3113         .fiemap         = ext3_fiemap,
3114 +       .sync_flags     = ext3_sync_flags,
3115  };
3116
3117 diff -NurpP --minimal linux-2.6.36/fs/ext3/ialloc.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/ialloc.c
3118 --- linux-2.6.36/fs/ext3/ialloc.c       2010-10-21 13:07:48.000000000 +0200
3119 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/ialloc.c 2010-10-21 13:09:36.000000000 +0200
3120 @@ -23,6 +23,7 @@
3121  #include <linux/buffer_head.h>
3122  #include <linux/random.h>
3123  #include <linux/bitops.h>
3124 +#include <linux/vs_tag.h>
3125
3126  #include <asm/byteorder.h>
3127
3128 @@ -531,6 +532,7 @@ got:
3129                 inode->i_mode = mode;
3130                 inode->i_uid = current_fsuid();
3131                 inode->i_gid = dir->i_gid;
3132 +               inode->i_tag = dx_current_fstag(sb);
3133         } else
3134                 inode_init_owner(inode, dir, mode);
3135
3136 diff -NurpP --minimal linux-2.6.36/fs/ext3/inode.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/inode.c
3137 --- linux-2.6.36/fs/ext3/inode.c        2010-10-21 13:07:48.000000000 +0200
3138 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/inode.c  2010-10-21 13:09:36.000000000 +0200
3139 @@ -38,6 +38,7 @@
3140  #include <linux/bio.h>
3141  #include <linux/fiemap.h>
3142  #include <linux/namei.h>
3143 +#include <linux/vs_tag.h>
3144  #include "xattr.h"
3145  #include "acl.h"
3146
3147 @@ -2387,7 +2388,7 @@ static void ext3_free_branches(handle_t
3148
3149  int ext3_can_truncate(struct inode *inode)
3150  {
3151 -       if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3152 +       if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3153                 return 0;
3154         if (S_ISREG(inode->i_mode))
3155                 return 1;
3156 @@ -2772,36 +2773,60 @@ void ext3_set_inode_flags(struct inode *
3157  {
3158         unsigned int flags = EXT3_I(inode)->i_flags;
3159
3160 -       inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3161 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3162 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3163 +
3164 +       if (flags & EXT3_IMMUTABLE_FL)
3165 +               inode->i_flags |= S_IMMUTABLE;
3166 +       if (flags & EXT3_IXUNLINK_FL)
3167 +               inode->i_flags |= S_IXUNLINK;
3168 +
3169         if (flags & EXT3_SYNC_FL)
3170                 inode->i_flags |= S_SYNC;
3171         if (flags & EXT3_APPEND_FL)
3172                 inode->i_flags |= S_APPEND;
3173 -       if (flags & EXT3_IMMUTABLE_FL)
3174 -               inode->i_flags |= S_IMMUTABLE;
3175         if (flags & EXT3_NOATIME_FL)
3176                 inode->i_flags |= S_NOATIME;
3177         if (flags & EXT3_DIRSYNC_FL)
3178                 inode->i_flags |= S_DIRSYNC;
3179 +
3180 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
3181 +
3182 +       if (flags & EXT3_BARRIER_FL)
3183 +               inode->i_vflags |= V_BARRIER;
3184 +       if (flags & EXT3_COW_FL)
3185 +               inode->i_vflags |= V_COW;
3186  }
3187
3188  /* Propagate flags from i_flags to EXT3_I(inode)->i_flags */
3189  void ext3_get_inode_flags(struct ext3_inode_info *ei)
3190  {
3191         unsigned int flags = ei->vfs_inode.i_flags;
3192 +       unsigned int vflags = ei->vfs_inode.i_vflags;
3193 +
3194 +       ei->i_flags &= ~(EXT3_SYNC_FL | EXT3_APPEND_FL |
3195 +                       EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL |
3196 +                       EXT3_NOATIME_FL | EXT3_DIRSYNC_FL |
3197 +                       EXT3_BARRIER_FL | EXT3_COW_FL);
3198 +
3199 +       if (flags & S_IMMUTABLE)
3200 +               ei->i_flags |= EXT3_IMMUTABLE_FL;
3201 +       if (flags & S_IXUNLINK)
3202 +               ei->i_flags |= EXT3_IXUNLINK_FL;
3203
3204 -       ei->i_flags &= ~(EXT3_SYNC_FL|EXT3_APPEND_FL|
3205 -                       EXT3_IMMUTABLE_FL|EXT3_NOATIME_FL|EXT3_DIRSYNC_FL);
3206         if (flags & S_SYNC)
3207                 ei->i_flags |= EXT3_SYNC_FL;
3208         if (flags & S_APPEND)
3209                 ei->i_flags |= EXT3_APPEND_FL;
3210 -       if (flags & S_IMMUTABLE)
3211 -               ei->i_flags |= EXT3_IMMUTABLE_FL;
3212         if (flags & S_NOATIME)
3213                 ei->i_flags |= EXT3_NOATIME_FL;
3214         if (flags & S_DIRSYNC)
3215                 ei->i_flags |= EXT3_DIRSYNC_FL;
3216 +
3217 +       if (vflags & V_BARRIER)
3218 +               ei->i_flags |= EXT3_BARRIER_FL;
3219 +       if (vflags & V_COW)
3220 +               ei->i_flags |= EXT3_COW_FL;
3221  }
3222
3223  struct inode *ext3_iget(struct super_block *sb, unsigned long ino)
3224 @@ -2815,6 +2840,8 @@ struct inode *ext3_iget(struct super_blo
3225         transaction_t *transaction;
3226         long ret;
3227         int block;
3228 +       uid_t uid;
3229 +       gid_t gid;
3230
3231         inode = iget_locked(sb, ino);
3232         if (!inode)
3233 @@ -2831,12 +2858,17 @@ struct inode *ext3_iget(struct super_blo
3234         bh = iloc.bh;
3235         raw_inode = ext3_raw_inode(&iloc);
3236         inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3237 -       inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3238 -       inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3239 +       uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3240 +       gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3241         if(!(test_opt (inode->i_sb, NO_UID32))) {
3242 -               inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3243 -               inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3244 +               uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3245 +               gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3246         }
3247 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3248 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3249 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3250 +               le16_to_cpu(raw_inode->i_raw_tag));
3251 +
3252         inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3253         inode->i_size = le32_to_cpu(raw_inode->i_size);
3254         inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime);
3255 @@ -2991,6 +3023,8 @@ static int ext3_do_update_inode(handle_t
3256         struct ext3_inode *raw_inode = ext3_raw_inode(iloc);
3257         struct ext3_inode_info *ei = EXT3_I(inode);
3258         struct buffer_head *bh = iloc->bh;
3259 +       uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3260 +       gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3261         int err = 0, rc, block;
3262
3263  again:
3264 @@ -3005,29 +3039,32 @@ again:
3265         ext3_get_inode_flags(ei);
3266         raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3267         if(!(test_opt(inode->i_sb, NO_UID32))) {
3268 -               raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3269 -               raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3270 +               raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3271 +               raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3272  /*
3273   * Fix up interoperability with old kernels. Otherwise, old inodes get
3274   * re-used with the upper 16 bits of the uid/gid intact
3275   */
3276                 if(!ei->i_dtime) {
3277                         raw_inode->i_uid_high =
3278 -                               cpu_to_le16(high_16_bits(inode->i_uid));
3279 +                               cpu_to_le16(high_16_bits(uid));
3280                         raw_inode->i_gid_high =
3281 -                               cpu_to_le16(high_16_bits(inode->i_gid));
3282 +                               cpu_to_le16(high_16_bits(gid));
3283                 } else {
3284                         raw_inode->i_uid_high = 0;
3285                         raw_inode->i_gid_high = 0;
3286                 }
3287         } else {
3288                 raw_inode->i_uid_low =
3289 -                       cpu_to_le16(fs_high2lowuid(inode->i_uid));
3290 +                       cpu_to_le16(fs_high2lowuid(uid));
3291                 raw_inode->i_gid_low =
3292 -                       cpu_to_le16(fs_high2lowgid(inode->i_gid));
3293 +                       cpu_to_le16(fs_high2lowgid(gid));
3294                 raw_inode->i_uid_high = 0;
3295                 raw_inode->i_gid_high = 0;
3296         }
3297 +#ifdef CONFIG_TAGGING_INTERN
3298 +       raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3299 +#endif
3300         raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3301         raw_inode->i_size = cpu_to_le32(ei->i_disksize);
3302         raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
3303 @@ -3187,7 +3224,8 @@ int ext3_setattr(struct dentry *dentry,
3304         if (is_quota_modification(inode, attr))
3305                 dquot_initialize(inode);
3306         if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3307 -               (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3308 +               (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3309 +               (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3310                 handle_t *handle;
3311
3312                 /* (user+group)*(old+new) structure, inode write (sb,
3313 @@ -3209,6 +3247,8 @@ int ext3_setattr(struct dentry *dentry,
3314                         inode->i_uid = attr->ia_uid;
3315                 if (attr->ia_valid & ATTR_GID)
3316                         inode->i_gid = attr->ia_gid;
3317 +               if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3318 +                       inode->i_tag = attr->ia_tag;
3319                 error = ext3_mark_inode_dirty(handle, inode);
3320                 ext3_journal_stop(handle);
3321         }
3322 diff -NurpP --minimal linux-2.6.36/fs/ext3/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/ioctl.c
3323 --- linux-2.6.36/fs/ext3/ioctl.c        2009-06-11 17:13:03.000000000 +0200
3324 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/ioctl.c  2010-10-21 13:09:36.000000000 +0200
3325 @@ -8,6 +8,7 @@
3326   */
3327
3328  #include <linux/fs.h>
3329 +#include <linux/mount.h>
3330  #include <linux/jbd.h>
3331  #include <linux/capability.h>
3332  #include <linux/ext3_fs.h>
3333 @@ -17,6 +18,34 @@
3334  #include <linux/compat.h>
3335  #include <asm/uaccess.h>
3336
3337 +
3338 +int ext3_sync_flags(struct inode *inode, int flags, int vflags)
3339 +{
3340 +       handle_t *handle = NULL;
3341 +       struct ext3_iloc iloc;
3342 +       int err;
3343 +
3344 +       handle = ext3_journal_start(inode, 1);
3345 +       if (IS_ERR(handle))
3346 +               return PTR_ERR(handle);
3347 +
3348 +       if (IS_SYNC(inode))
3349 +               handle->h_sync = 1;
3350 +       err = ext3_reserve_inode_write(handle, inode, &iloc);
3351 +       if (err)
3352 +               goto flags_err;
3353 +
3354 +       inode->i_flags = flags;
3355 +       inode->i_vflags = vflags;
3356 +       ext3_get_inode_flags(EXT3_I(inode));
3357 +       inode->i_ctime = CURRENT_TIME_SEC;
3358 +
3359 +       err = ext3_mark_iloc_dirty(handle, inode, &iloc);
3360 +flags_err:
3361 +       ext3_journal_stop(handle);
3362 +       return err;
3363 +}
3364 +
3365  long ext3_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
3366  {
3367         struct inode *inode = filp->f_dentry->d_inode;
3368 @@ -50,6 +79,11 @@ long ext3_ioctl(struct file *filp, unsig
3369
3370                 flags = ext3_mask_flags(inode->i_mode, flags);
3371
3372 +               if (IS_BARRIER(inode)) {
3373 +                       vxwprintk_task(1, "messing with the barrier.");
3374 +                       return -EACCES;
3375 +               }
3376 +
3377                 mutex_lock(&inode->i_mutex);
3378
3379                 /* Is it quota file? Do not allow user to mess with it */
3380 @@ -68,7 +102,9 @@ long ext3_ioctl(struct file *filp, unsig
3381                  *
3382                  * This test looks nicer. Thanks to Pauline Middelink
3383                  */
3384 -               if ((flags ^ oldflags) & (EXT3_APPEND_FL | EXT3_IMMUTABLE_FL)) {
3385 +               if ((oldflags & EXT3_IMMUTABLE_FL) ||
3386 +                       ((flags ^ oldflags) & (EXT3_APPEND_FL |
3387 +                       EXT3_IMMUTABLE_FL | EXT3_IXUNLINK_FL))) {
3388                         if (!capable(CAP_LINUX_IMMUTABLE))
3389                                 goto flags_out;
3390                 }
3391 @@ -93,7 +129,7 @@ long ext3_ioctl(struct file *filp, unsig
3392                 if (err)
3393                         goto flags_err;
3394
3395 -               flags = flags & EXT3_FL_USER_MODIFIABLE;
3396 +               flags &= EXT3_FL_USER_MODIFIABLE;
3397                 flags |= oldflags & ~EXT3_FL_USER_MODIFIABLE;
3398                 ei->i_flags = flags;
3399
3400 diff -NurpP --minimal linux-2.6.36/fs/ext3/namei.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/namei.c
3401 --- linux-2.6.36/fs/ext3/namei.c        2010-10-21 13:07:48.000000000 +0200
3402 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/namei.c  2010-10-21 13:09:36.000000000 +0200
3403 @@ -36,6 +36,7 @@
3404  #include <linux/quotaops.h>
3405  #include <linux/buffer_head.h>
3406  #include <linux/bio.h>
3407 +#include <linux/vs_tag.h>
3408
3409  #include "namei.h"
3410  #include "xattr.h"
3411 @@ -912,6 +913,7 @@ restart:
3412                                 if (bh)
3413                                         ll_rw_block(READ_META, 1, &bh);
3414                         }
3415 +               dx_propagate_tag(nd, inode);
3416                 }
3417                 if ((bh = bh_use[ra_ptr++]) == NULL)
3418                         goto next;
3419 @@ -2465,6 +2467,7 @@ const struct inode_operations ext3_dir_i
3420         .removexattr    = generic_removexattr,
3421  #endif
3422         .check_acl      = ext3_check_acl,
3423 +       .sync_flags     = ext3_sync_flags,
3424  };
3425
3426  const struct inode_operations ext3_special_inode_operations = {
3427 diff -NurpP --minimal linux-2.6.36/fs/ext3/super.c linux-2.6.36-vs2.3.0.36.38/fs/ext3/super.c
3428 --- linux-2.6.36/fs/ext3/super.c        2010-10-21 13:07:48.000000000 +0200
3429 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext3/super.c  2010-10-21 13:09:36.000000000 +0200
3430 @@ -802,7 +802,8 @@ enum {
3431         Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
3432         Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_jqfmt_vfsv1, Opt_quota,
3433         Opt_noquota, Opt_ignore, Opt_barrier, Opt_nobarrier, Opt_err,
3434 -       Opt_resize, Opt_usrquota, Opt_grpquota
3435 +       Opt_resize, Opt_usrquota, Opt_grpquota,
3436 +       Opt_tag, Opt_notag, Opt_tagid
3437  };
3438
3439  static const match_table_t tokens = {
3440 @@ -859,6 +860,9 @@ static const match_table_t tokens = {
3441         {Opt_barrier, "barrier"},
3442         {Opt_nobarrier, "nobarrier"},
3443         {Opt_resize, "resize"},
3444 +       {Opt_tag, "tag"},
3445 +       {Opt_notag, "notag"},
3446 +       {Opt_tagid, "tagid=%u"},
3447         {Opt_err, NULL},
3448  };
3449
3450 @@ -1011,6 +1015,20 @@ static int parse_options (char *options,
3451                 case Opt_nouid32:
3452                         set_opt (sbi->s_mount_opt, NO_UID32);
3453                         break;
3454 +#ifndef CONFIG_TAGGING_NONE
3455 +               case Opt_tag:
3456 +                       set_opt (sbi->s_mount_opt, TAGGED);
3457 +                       break;
3458 +               case Opt_notag:
3459 +                       clear_opt (sbi->s_mount_opt, TAGGED);
3460 +                       break;
3461 +#endif
3462 +#ifdef CONFIG_PROPAGATE
3463 +               case Opt_tagid:
3464 +                       /* use args[0] */
3465 +                       set_opt (sbi->s_mount_opt, TAGGED);
3466 +                       break;
3467 +#endif
3468                 case Opt_nocheck:
3469                         clear_opt (sbi->s_mount_opt, CHECK);
3470                         break;
3471 @@ -1699,6 +1717,9 @@ static int ext3_fill_super (struct super
3472                             NULL, 0))
3473                 goto failed_mount;
3474
3475 +       if (EXT3_SB(sb)->s_mount_opt & EXT3_MOUNT_TAGGED)
3476 +               sb->s_flags |= MS_TAGGED;
3477 +
3478         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3479                 (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0);
3480
3481 @@ -2564,6 +2585,14 @@ static int ext3_remount (struct super_bl
3482         if (test_opt(sb, ABORT))
3483                 ext3_abort(sb, __func__, "Abort forced by user");
3484
3485 +       if ((sbi->s_mount_opt & EXT3_MOUNT_TAGGED) &&
3486 +               !(sb->s_flags & MS_TAGGED)) {
3487 +               printk("EXT3-fs: %s: tagging not permitted on remount.\n",
3488 +                       sb->s_id);
3489 +               err = -EINVAL;
3490 +               goto restore_opts;
3491 +       }
3492 +
3493         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3494                 (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0);
3495
3496 diff -NurpP --minimal linux-2.6.36/fs/ext4/ext4.h linux-2.6.36-vs2.3.0.36.38/fs/ext4/ext4.h
3497 --- linux-2.6.36/fs/ext4/ext4.h 2010-10-21 13:07:48.000000000 +0200
3498 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/ext4.h   2010-10-21 13:09:36.000000000 +0200
3499 @@ -323,8 +323,12 @@ struct flex_groups {
3500  #define EXT4_EXTENTS_FL                        0x00080000 /* Inode uses extents */
3501  #define EXT4_EA_INODE_FL               0x00200000 /* Inode used for large EA */
3502  #define EXT4_EOFBLOCKS_FL              0x00400000 /* Blocks allocated beyond EOF */
3503 +#define EXT4_IXUNLINK_FL               0x08000000 /* Immutable invert on unlink */
3504  #define EXT4_RESERVED_FL               0x80000000 /* reserved for ext4 lib */
3505
3506 +#define EXT4_BARRIER_FL                        0x04000000 /* Barrier for chroot() */
3507 +#define EXT4_COW_FL                    0x20000000 /* Copy on Write marker */
3508 +
3509  #define EXT4_FL_USER_VISIBLE           0x004BDFFF /* User visible flags */
3510  #define EXT4_FL_USER_MODIFIABLE                0x004B80FF /* User modifiable flags */
3511
3512 @@ -593,7 +597,8 @@ struct ext4_inode {
3513                         __le16  l_i_file_acl_high;
3514                         __le16  l_i_uid_high;   /* these 2 fields */
3515                         __le16  l_i_gid_high;   /* were reserved2[0] */
3516 -                       __u32   l_i_reserved2;
3517 +                       __le16  l_i_tag;        /* Context Tag */
3518 +                       __u16   l_i_reserved2;
3519                 } linux2;
3520                 struct {
3521                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
3522 @@ -707,6 +712,7 @@ do {                                                                               \
3523  #define i_gid_low      i_gid
3524  #define i_uid_high     osd2.linux2.l_i_uid_high
3525  #define i_gid_high     osd2.linux2.l_i_gid_high
3526 +#define i_raw_tag      osd2.linux2.l_i_tag
3527  #define i_reserved2    osd2.linux2.l_i_reserved2
3528
3529  #elif defined(__GNU__)
3530 @@ -885,6 +891,7 @@ struct ext4_inode_info {
3531  #define EXT4_MOUNT_JOURNAL_CHECKSUM    0x800000 /* Journal checksums */
3532  #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT        0x1000000 /* Journal Async Commit */
3533  #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
3534 +#define EXT4_MOUNT_TAGGED              0x4000000 /* Enable Context Tags */
3535  #define EXT4_MOUNT_DELALLOC            0x8000000 /* Delalloc support */
3536  #define EXT4_MOUNT_DATA_ERR_ABORT      0x10000000 /* Abort on file data write */
3537  #define EXT4_MOUNT_BLOCK_VALIDITY      0x20000000 /* Block validity checking */
3538 @@ -1997,6 +2004,7 @@ extern int ext4_map_blocks(handle_t *han
3539                            struct ext4_map_blocks *map, int flags);
3540  extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3541                         __u64 start, __u64 len);
3542 +extern int ext4_sync_flags(struct inode *, int, int);
3543  /* move_extent.c */
3544  extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3545                              __u64 start_orig, __u64 start_donor,
3546 diff -NurpP --minimal linux-2.6.36/fs/ext4/file.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/file.c
3547 --- linux-2.6.36/fs/ext4/file.c 2010-10-21 13:07:48.000000000 +0200
3548 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/file.c   2010-10-21 13:09:36.000000000 +0200
3549 @@ -161,5 +161,6 @@ const struct inode_operations ext4_file_
3550         .check_acl      = ext4_check_acl,
3551         .fallocate      = ext4_fallocate,
3552         .fiemap         = ext4_fiemap,
3553 +       .sync_flags     = ext4_sync_flags,
3554  };
3555
3556 diff -NurpP --minimal linux-2.6.36/fs/ext4/ialloc.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/ialloc.c
3557 --- linux-2.6.36/fs/ext4/ialloc.c       2010-10-21 13:07:48.000000000 +0200
3558 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/ialloc.c 2010-10-21 13:09:36.000000000 +0200
3559 @@ -22,6 +22,7 @@
3560  #include <linux/random.h>
3561  #include <linux/bitops.h>
3562  #include <linux/blkdev.h>
3563 +#include <linux/vs_tag.h>
3564  #include <asm/byteorder.h>
3565
3566  #include "ext4.h"
3567 @@ -976,6 +977,7 @@ got:
3568                 inode->i_mode = mode;
3569                 inode->i_uid = current_fsuid();
3570                 inode->i_gid = dir->i_gid;
3571 +               inode->i_tag = dx_current_fstag(sb);
3572         } else
3573                 inode_init_owner(inode, dir, mode);
3574
3575 diff -NurpP --minimal linux-2.6.36/fs/ext4/inode.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/inode.c
3576 --- linux-2.6.36/fs/ext4/inode.c        2010-10-21 13:07:48.000000000 +0200
3577 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/inode.c  2010-10-21 13:09:36.000000000 +0200
3578 @@ -40,6 +40,7 @@
3579  #include <linux/workqueue.h>
3580  #include <linux/kernel.h>
3581  #include <linux/slab.h>
3582 +#include <linux/vs_tag.h>
3583
3584  #include "ext4_jbd2.h"
3585  #include "xattr.h"
3586 @@ -4596,7 +4597,7 @@ static void ext4_free_branches(handle_t
3587
3588  int ext4_can_truncate(struct inode *inode)
3589  {
3590 -       if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
3591 +       if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
3592                 return 0;
3593         if (S_ISREG(inode->i_mode))
3594                 return 1;
3595 @@ -4947,41 +4948,64 @@ void ext4_set_inode_flags(struct inode *
3596  {
3597         unsigned int flags = EXT4_I(inode)->i_flags;
3598
3599 -       inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3600 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
3601 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
3602 +
3603 +       if (flags & EXT4_IMMUTABLE_FL)
3604 +               inode->i_flags |= S_IMMUTABLE;
3605 +       if (flags & EXT4_IXUNLINK_FL)
3606 +               inode->i_flags |= S_IXUNLINK;
3607 +
3608         if (flags & EXT4_SYNC_FL)
3609                 inode->i_flags |= S_SYNC;
3610         if (flags & EXT4_APPEND_FL)
3611                 inode->i_flags |= S_APPEND;
3612 -       if (flags & EXT4_IMMUTABLE_FL)
3613 -               inode->i_flags |= S_IMMUTABLE;
3614         if (flags & EXT4_NOATIME_FL)
3615                 inode->i_flags |= S_NOATIME;
3616         if (flags & EXT4_DIRSYNC_FL)
3617                 inode->i_flags |= S_DIRSYNC;
3618 +
3619 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
3620 +
3621 +       if (flags & EXT4_BARRIER_FL)
3622 +               inode->i_vflags |= V_BARRIER;
3623 +       if (flags & EXT4_COW_FL)
3624 +               inode->i_vflags |= V_COW;
3625  }
3626
3627  /* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
3628  void ext4_get_inode_flags(struct ext4_inode_info *ei)
3629  {
3630 -       unsigned int vfs_fl;
3631 +       unsigned int vfs_fl, vfs_vf;
3632         unsigned long old_fl, new_fl;
3633
3634         do {
3635                 vfs_fl = ei->vfs_inode.i_flags;
3636 +               vfs_vf = ei->vfs_inode.i_vflags;
3637                 old_fl = ei->i_flags;
3638                 new_fl = old_fl & ~(EXT4_SYNC_FL|EXT4_APPEND_FL|
3639                                 EXT4_IMMUTABLE_FL|EXT4_NOATIME_FL|
3640 -                               EXT4_DIRSYNC_FL);
3641 +                               EXT4_DIRSYNC_FL|EXT4_BARRIER_FL|
3642 +                               EXT4_COW_FL);
3643 +
3644 +               if (vfs_fl & S_IMMUTABLE)
3645 +                       new_fl |= EXT4_IMMUTABLE_FL;
3646 +               if (vfs_fl & S_IXUNLINK)
3647 +                       new_fl |= EXT4_IXUNLINK_FL;
3648 +
3649                 if (vfs_fl & S_SYNC)
3650                         new_fl |= EXT4_SYNC_FL;
3651                 if (vfs_fl & S_APPEND)
3652                         new_fl |= EXT4_APPEND_FL;
3653 -               if (vfs_fl & S_IMMUTABLE)
3654 -                       new_fl |= EXT4_IMMUTABLE_FL;
3655                 if (vfs_fl & S_NOATIME)
3656                         new_fl |= EXT4_NOATIME_FL;
3657                 if (vfs_fl & S_DIRSYNC)
3658                         new_fl |= EXT4_DIRSYNC_FL;
3659 +
3660 +               if (vfs_vf & V_BARRIER)
3661 +                       new_fl |= EXT4_BARRIER_FL;
3662 +               if (vfs_vf & V_COW)
3663 +                       new_fl |= EXT4_COW_FL;
3664         } while (cmpxchg(&ei->i_flags, old_fl, new_fl) != old_fl);
3665  }
3666
3667 @@ -5017,6 +5041,8 @@ struct inode *ext4_iget(struct super_blo
3668         journal_t *journal = EXT4_SB(sb)->s_journal;
3669         long ret;
3670         int block;
3671 +       uid_t uid;
3672 +       gid_t gid;
3673
3674         inode = iget_locked(sb, ino);
3675         if (!inode)
3676 @@ -5032,12 +5058,16 @@ struct inode *ext4_iget(struct super_blo
3677                 goto bad_inode;
3678         raw_inode = ext4_raw_inode(&iloc);
3679         inode->i_mode = le16_to_cpu(raw_inode->i_mode);
3680 -       inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3681 -       inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3682 +       uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
3683 +       gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
3684         if (!(test_opt(inode->i_sb, NO_UID32))) {
3685 -               inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3686 -               inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3687 +               uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
3688 +               gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
3689         }
3690 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
3691 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
3692 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
3693 +               le16_to_cpu(raw_inode->i_raw_tag));
3694         inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);
3695
3696         ei->i_state_flags = 0;
3697 @@ -5256,6 +5286,8 @@ static int ext4_do_update_inode(handle_t
3698         struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
3699         struct ext4_inode_info *ei = EXT4_I(inode);
3700         struct buffer_head *bh = iloc->bh;
3701 +       uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
3702 +       gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
3703         int err = 0, rc, block;
3704
3705         /* For fields not not tracking in the in-memory inode,
3706 @@ -5266,29 +5298,32 @@ static int ext4_do_update_inode(handle_t
3707         ext4_get_inode_flags(ei);
3708         raw_inode->i_mode = cpu_to_le16(inode->i_mode);
3709         if (!(test_opt(inode->i_sb, NO_UID32))) {
3710 -               raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
3711 -               raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
3712 +               raw_inode->i_uid_low = cpu_to_le16(low_16_bits(uid));
3713 +               raw_inode->i_gid_low = cpu_to_le16(low_16_bits(gid));
3714  /*
3715   * Fix up interoperability with old kernels. Otherwise, old inodes get
3716   * re-used with the upper 16 bits of the uid/gid intact
3717   */
3718                 if (!ei->i_dtime) {
3719                         raw_inode->i_uid_high =
3720 -                               cpu_to_le16(high_16_bits(inode->i_uid));
3721 +                               cpu_to_le16(high_16_bits(uid));
3722                         raw_inode->i_gid_high =
3723 -                               cpu_to_le16(high_16_bits(inode->i_gid));
3724 +                               cpu_to_le16(high_16_bits(gid));
3725                 } else {
3726                         raw_inode->i_uid_high = 0;
3727                         raw_inode->i_gid_high = 0;
3728                 }
3729         } else {
3730                 raw_inode->i_uid_low =
3731 -                       cpu_to_le16(fs_high2lowuid(inode->i_uid));
3732 +                       cpu_to_le16(fs_high2lowuid(uid));
3733                 raw_inode->i_gid_low =
3734 -                       cpu_to_le16(fs_high2lowgid(inode->i_gid));
3735 +                       cpu_to_le16(fs_high2lowgid(gid));
3736                 raw_inode->i_uid_high = 0;
3737                 raw_inode->i_gid_high = 0;
3738         }
3739 +#ifdef CONFIG_TAGGING_INTERN
3740 +       raw_inode->i_raw_tag = cpu_to_le16(inode->i_tag);
3741 +#endif
3742         raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
3743
3744         EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
3745 @@ -5473,7 +5508,8 @@ int ext4_setattr(struct dentry *dentry,
3746         if (is_quota_modification(inode, attr))
3747                 dquot_initialize(inode);
3748         if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
3749 -               (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
3750 +               (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
3751 +               (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
3752                 handle_t *handle;
3753
3754                 /* (user+group)*(old+new) structure, inode write (sb,
3755 @@ -5495,6 +5531,8 @@ int ext4_setattr(struct dentry *dentry,
3756                         inode->i_uid = attr->ia_uid;
3757                 if (attr->ia_valid & ATTR_GID)
3758                         inode->i_gid = attr->ia_gid;
3759 +               if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
3760 +                       inode->i_tag = attr->ia_tag;
3761                 error = ext4_mark_inode_dirty(handle, inode);
3762                 ext4_journal_stop(handle);
3763         }
3764 diff -NurpP --minimal linux-2.6.36/fs/ext4/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/ioctl.c
3765 --- linux-2.6.36/fs/ext4/ioctl.c        2010-08-02 16:52:48.000000000 +0200
3766 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/ioctl.c  2010-10-21 13:09:36.000000000 +0200
3767 @@ -14,10 +14,39 @@
3768  #include <linux/compat.h>
3769  #include <linux/mount.h>
3770  #include <linux/file.h>
3771 +#include <linux/vs_tag.h>
3772  #include <asm/uaccess.h>
3773  #include "ext4_jbd2.h"
3774  #include "ext4.h"
3775
3776 +
3777 +int ext4_sync_flags(struct inode *inode, int flags, int vflags)
3778 +{
3779 +       handle_t *handle = NULL;
3780 +       struct ext4_iloc iloc;
3781 +       int err;
3782 +
3783 +       handle = ext4_journal_start(inode, 1);
3784 +       if (IS_ERR(handle))
3785 +               return PTR_ERR(handle);
3786 +
3787 +       if (IS_SYNC(inode))
3788 +               ext4_handle_sync(handle);
3789 +       err = ext4_reserve_inode_write(handle, inode, &iloc);
3790 +       if (err)
3791 +               goto flags_err;
3792 +
3793 +       inode->i_flags = flags;
3794 +       inode->i_vflags = vflags;
3795 +       ext4_get_inode_flags(EXT4_I(inode));
3796 +       inode->i_ctime = ext4_current_time(inode);
3797 +
3798 +       err = ext4_mark_iloc_dirty(handle, inode, &iloc);
3799 +flags_err:
3800 +       ext4_journal_stop(handle);
3801 +       return err;
3802 +}
3803 +
3804  long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
3805  {
3806         struct inode *inode = filp->f_dentry->d_inode;
3807 @@ -50,6 +79,11 @@ long ext4_ioctl(struct file *filp, unsig
3808
3809                 flags = ext4_mask_flags(inode->i_mode, flags);
3810
3811 +               if (IS_BARRIER(inode)) {
3812 +                       vxwprintk_task(1, "messing with the barrier.");
3813 +                       return -EACCES;
3814 +               }
3815 +
3816                 err = -EPERM;
3817                 mutex_lock(&inode->i_mutex);
3818                 /* Is it quota file? Do not allow user to mess with it */
3819 @@ -67,7 +101,9 @@ long ext4_ioctl(struct file *filp, unsig
3820                  *
3821                  * This test looks nicer. Thanks to Pauline Middelink
3822                  */
3823 -               if ((flags ^ oldflags) & (EXT4_APPEND_FL | EXT4_IMMUTABLE_FL)) {
3824 +               if ((oldflags & EXT4_IMMUTABLE_FL) ||
3825 +                       ((flags ^ oldflags) & (EXT4_APPEND_FL |
3826 +                       EXT4_IMMUTABLE_FL | EXT4_IXUNLINK_FL))) {
3827                         if (!capable(CAP_LINUX_IMMUTABLE))
3828                                 goto flags_out;
3829                 }
3830 diff -NurpP --minimal linux-2.6.36/fs/ext4/namei.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/namei.c
3831 --- linux-2.6.36/fs/ext4/namei.c        2010-10-21 13:07:48.000000000 +0200
3832 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/namei.c  2010-10-21 13:09:36.000000000 +0200
3833 @@ -34,6 +34,7 @@
3834  #include <linux/quotaops.h>
3835  #include <linux/buffer_head.h>
3836  #include <linux/bio.h>
3837 +#include <linux/vs_tag.h>
3838  #include "ext4.h"
3839  #include "ext4_jbd2.h"
3840
3841 @@ -912,6 +913,7 @@ restart:
3842                                 if (bh)
3843                                         ll_rw_block(READ_META, 1, &bh);
3844                         }
3845 +               dx_propagate_tag(nd, inode);
3846                 }
3847                 if ((bh = bh_use[ra_ptr++]) == NULL)
3848                         goto next;
3849 @@ -2522,6 +2524,7 @@ const struct inode_operations ext4_dir_i
3850  #endif
3851         .check_acl      = ext4_check_acl,
3852         .fiemap         = ext4_fiemap,
3853 +       .sync_flags     = ext4_sync_flags,
3854  };
3855
3856  const struct inode_operations ext4_special_inode_operations = {
3857 diff -NurpP --minimal linux-2.6.36/fs/ext4/super.c linux-2.6.36-vs2.3.0.36.38/fs/ext4/super.c
3858 --- linux-2.6.36/fs/ext4/super.c        2010-10-21 13:07:48.000000000 +0200
3859 +++ linux-2.6.36-vs2.3.0.36.38/fs/ext4/super.c  2010-10-21 13:09:36.000000000 +0200
3860 @@ -1219,6 +1219,7 @@ enum {
3861         Opt_inode_readahead_blks, Opt_journal_ioprio,
3862         Opt_dioread_nolock, Opt_dioread_lock,
3863         Opt_discard, Opt_nodiscard,
3864 +       Opt_tag, Opt_notag, Opt_tagid
3865  };
3866
3867  static const match_table_t tokens = {
3868 @@ -1289,6 +1290,9 @@ static const match_table_t tokens = {
3869         {Opt_dioread_lock, "dioread_lock"},
3870         {Opt_discard, "discard"},
3871         {Opt_nodiscard, "nodiscard"},
3872 +       {Opt_tag, "tag"},
3873 +       {Opt_notag, "notag"},
3874 +       {Opt_tagid, "tagid=%u"},
3875         {Opt_err, NULL},
3876  };
3877
3878 @@ -1457,6 +1461,20 @@ static int parse_options(char *options,
3879                 case Opt_nouid32:
3880                         set_opt(sbi->s_mount_opt, NO_UID32);
3881                         break;
3882 +#ifndef CONFIG_TAGGING_NONE
3883 +               case Opt_tag:
3884 +                       set_opt (sbi->s_mount_opt, TAGGED);
3885 +                       break;
3886 +               case Opt_notag:
3887 +                       clear_opt (sbi->s_mount_opt, TAGGED);
3888 +                       break;
3889 +#endif
3890 +#ifdef CONFIG_PROPAGATE
3891 +               case Opt_tagid:
3892 +                       /* use args[0] */
3893 +                       set_opt (sbi->s_mount_opt, TAGGED);
3894 +                       break;
3895 +#endif
3896                 case Opt_debug:
3897                         set_opt(sbi->s_mount_opt, DEBUG);
3898                         break;
3899 @@ -2691,6 +2709,9 @@ static int ext4_fill_super(struct super_
3900                            &journal_ioprio, NULL, 0))
3901                 goto failed_mount;
3902
3903 +       if (EXT4_SB(sb)->s_mount_opt & EXT4_MOUNT_TAGGED)
3904 +               sb->s_flags |= MS_TAGGED;
3905 +
3906         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3907                 (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0);
3908
3909 @@ -3753,6 +3774,14 @@ static int ext4_remount(struct super_blo
3910         if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED)
3911                 ext4_abort(sb, "Abort forced by user");
3912
3913 +       if ((sbi->s_mount_opt & EXT4_MOUNT_TAGGED) &&
3914 +               !(sb->s_flags & MS_TAGGED)) {
3915 +               printk("EXT4-fs: %s: tagging not permitted on remount.\n",
3916 +                       sb->s_id);
3917 +               err = -EINVAL;
3918 +               goto restore_opts;
3919 +       }
3920 +
3921         sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3922                 (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0);
3923
3924 diff -NurpP --minimal linux-2.6.36/fs/fcntl.c linux-2.6.36-vs2.3.0.36.38/fs/fcntl.c
3925 --- linux-2.6.36/fs/fcntl.c     2010-10-21 13:07:48.000000000 +0200
3926 +++ linux-2.6.36-vs2.3.0.36.38/fs/fcntl.c       2010-10-21 13:09:36.000000000 +0200
3927 @@ -20,6 +20,7 @@
3928  #include <linux/signal.h>
3929  #include <linux/rcupdate.h>
3930  #include <linux/pid_namespace.h>
3931 +#include <linux/vs_limit.h>
3932
3933  #include <asm/poll.h>
3934  #include <asm/siginfo.h>
3935 @@ -103,6 +104,8 @@ SYSCALL_DEFINE3(dup3, unsigned int, oldf
3936
3937         if (tofree)
3938                 filp_close(tofree, files);
3939 +       else
3940 +               vx_openfd_inc(newfd);   /* fd was unused */
3941
3942         return newfd;
3943
3944 @@ -434,6 +437,8 @@ SYSCALL_DEFINE3(fcntl, unsigned int, fd,
3945         filp = fget(fd);
3946         if (!filp)
3947                 goto out;
3948 +       if (!vx_files_avail(1))
3949 +               goto out;
3950
3951         err = security_file_fcntl(filp, cmd, arg);
3952         if (err) {
3953 diff -NurpP --minimal linux-2.6.36/fs/file.c linux-2.6.36-vs2.3.0.36.38/fs/file.c
3954 --- linux-2.6.36/fs/file.c      2010-10-21 13:07:48.000000000 +0200
3955 +++ linux-2.6.36-vs2.3.0.36.38/fs/file.c        2010-10-21 13:09:36.000000000 +0200
3956 @@ -20,6 +20,7 @@
3957  #include <linux/spinlock.h>
3958  #include <linux/rcupdate.h>
3959  #include <linux/workqueue.h>
3960 +#include <linux/vs_limit.h>
3961
3962  struct fdtable_defer {
3963         spinlock_t lock;
3964 @@ -355,6 +356,8 @@ struct files_struct *dup_fd(struct files
3965                 struct file *f = *old_fds++;
3966                 if (f) {
3967                         get_file(f);
3968 +                       /* TODO: sum it first for check and performance */
3969 +                       vx_openfd_inc(open_files - i);
3970                 } else {
3971                         /*
3972                          * The fd may be claimed in the fd bitmap but not yet
3973 @@ -462,6 +465,7 @@ repeat:
3974         else
3975                 FD_CLR(fd, fdt->close_on_exec);
3976         error = fd;
3977 +       vx_openfd_inc(fd);
3978  #if 1
3979         /* Sanity check */
3980         if (rcu_dereference_raw(fdt->fd[fd]) != NULL) {
3981 diff -NurpP --minimal linux-2.6.36/fs/file_table.c linux-2.6.36-vs2.3.0.36.38/fs/file_table.c
3982 --- linux-2.6.36/fs/file_table.c        2010-10-21 13:07:48.000000000 +0200
3983 +++ linux-2.6.36-vs2.3.0.36.38/fs/file_table.c  2010-10-21 14:10:28.000000000 +0200
3984 @@ -24,6 +24,8 @@
3985  #include <linux/percpu_counter.h>
3986  #include <linux/percpu.h>
3987  #include <linux/ima.h>
3988 +#include <linux/vs_limit.h>
3989 +#include <linux/vs_context.h>
3990
3991  #include <asm/atomic.h>
3992
3993 @@ -135,6 +137,8 @@ struct file *get_empty_filp(void)
3994         spin_lock_init(&f->f_lock);
3995         eventpoll_init_file(f);
3996         /* f->f_version: 0 */
3997 +       f->f_xid = vx_current_xid();
3998 +       vx_files_inc(f);
3999         return f;
4000
4001  over:
4002 @@ -251,6 +255,8 @@ static void __fput(struct file *file)
4003                 cdev_put(inode->i_cdev);
4004         fops_put(file->f_op);
4005         put_pid(file->f_owner.pid);
4006 +       vx_files_dec(file);
4007 +       file->f_xid = 0;
4008         file_sb_list_del(file);
4009         if (file->f_mode & FMODE_WRITE)
4010                 drop_file_write_access(file);
4011 @@ -334,6 +340,8 @@ void put_filp(struct file *file)
4012  {
4013         if (atomic_long_dec_and_test(&file->f_count)) {
4014                 security_file_free(file);
4015 +               vx_files_dec(file);
4016 +               file->f_xid = 0;
4017                 file_sb_list_del(file);
4018                 file_free(file);
4019         }
4020 diff -NurpP --minimal linux-2.6.36/fs/fs_struct.c linux-2.6.36-vs2.3.0.36.38/fs/fs_struct.c
4021 --- linux-2.6.36/fs/fs_struct.c 2010-10-21 13:07:48.000000000 +0200
4022 +++ linux-2.6.36-vs2.3.0.36.38/fs/fs_struct.c   2010-10-21 13:56:13.000000000 +0200
4023 @@ -4,6 +4,7 @@
4024  #include <linux/path.h>
4025  #include <linux/slab.h>
4026  #include <linux/fs_struct.h>
4027 +#include <linux/vserver/global.h>
4028
4029  /*
4030   * Replace the fs->{rootmnt,root} with {mnt,dentry}. Put the old values.
4031 @@ -77,6 +78,7 @@ void free_fs_struct(struct fs_struct *fs
4032  {
4033         path_put(&fs->root);
4034         path_put(&fs->pwd);
4035 +       atomic_dec(&vs_global_fs);
4036         kmem_cache_free(fs_cachep, fs);
4037  }
4038
4039 @@ -107,6 +109,7 @@ struct fs_struct *copy_fs_struct(struct
4040                 spin_lock_init(&fs->lock);
4041                 fs->umask = old->umask;
4042                 get_fs_root_and_pwd(old, &fs->root, &fs->pwd);
4043 +               atomic_inc(&vs_global_fs);
4044         }
4045         return fs;
4046  }
4047 diff -NurpP --minimal linux-2.6.36/fs/gfs2/file.c linux-2.6.36-vs2.3.0.36.38/fs/gfs2/file.c
4048 --- linux-2.6.36/fs/gfs2/file.c 2010-10-21 13:07:49.000000000 +0200
4049 +++ linux-2.6.36-vs2.3.0.36.38/fs/gfs2/file.c   2010-10-21 13:09:36.000000000 +0200
4050 @@ -132,6 +132,9 @@ static const u32 fsflags_to_gfs2[32] = {
4051         [7] = GFS2_DIF_NOATIME,
4052         [12] = GFS2_DIF_EXHASH,
4053         [14] = GFS2_DIF_INHERIT_JDATA,
4054 +       [27] = GFS2_DIF_IXUNLINK,
4055 +       [26] = GFS2_DIF_BARRIER,
4056 +       [29] = GFS2_DIF_COW,
4057  };
4058
4059  static const u32 gfs2_to_fsflags[32] = {
4060 @@ -141,6 +144,9 @@ static const u32 gfs2_to_fsflags[32] = {
4061         [gfs2fl_NoAtime] = FS_NOATIME_FL,
4062         [gfs2fl_ExHash] = FS_INDEX_FL,
4063         [gfs2fl_InheritJdata] = FS_JOURNAL_DATA_FL,
4064 +       [gfs2fl_IXUnlink] = FS_IXUNLINK_FL,
4065 +       [gfs2fl_Barrier] = FS_BARRIER_FL,
4066 +       [gfs2fl_Cow] = FS_COW_FL,
4067  };
4068
4069  static int gfs2_get_flags(struct file *filp, u32 __user *ptr)
4070 @@ -171,10 +177,16 @@ void gfs2_set_inode_flags(struct inode *
4071  {
4072         struct gfs2_inode *ip = GFS2_I(inode);
4073         unsigned int flags = inode->i_flags;
4074 +       unsigned int vflags = inode->i_vflags;
4075 +
4076 +       flags &= ~(S_IMMUTABLE | S_IXUNLINK |
4077 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
4078
4079 -       flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
4080         if (ip->i_diskflags & GFS2_DIF_IMMUTABLE)
4081                 flags |= S_IMMUTABLE;
4082 +       if (ip->i_diskflags & GFS2_DIF_IXUNLINK)
4083 +               flags |= S_IXUNLINK;
4084 +
4085         if (ip->i_diskflags & GFS2_DIF_APPENDONLY)
4086                 flags |= S_APPEND;
4087         if (ip->i_diskflags & GFS2_DIF_NOATIME)
4088 @@ -182,6 +194,43 @@ void gfs2_set_inode_flags(struct inode *
4089         if (ip->i_diskflags & GFS2_DIF_SYNC)
4090                 flags |= S_SYNC;
4091         inode->i_flags = flags;
4092 +
4093 +       vflags &= ~(V_BARRIER | V_COW);
4094 +
4095 +       if (ip->i_diskflags & GFS2_DIF_BARRIER)
4096 +               vflags |= V_BARRIER;
4097 +       if (ip->i_diskflags & GFS2_DIF_COW)
4098 +               vflags |= V_COW;
4099 +       inode->i_vflags = vflags;
4100 +}
4101 +
4102 +void gfs2_get_inode_flags(struct inode *inode)
4103 +{
4104 +       struct gfs2_inode *ip = GFS2_I(inode);
4105 +       unsigned int flags = inode->i_flags;
4106 +       unsigned int vflags = inode->i_vflags;
4107 +
4108 +       ip->i_diskflags &= ~(GFS2_DIF_APPENDONLY |
4109 +                       GFS2_DIF_NOATIME | GFS2_DIF_SYNC |
4110 +                       GFS2_DIF_IMMUTABLE | GFS2_DIF_IXUNLINK |
4111 +                       GFS2_DIF_BARRIER | GFS2_DIF_COW);
4112 +
4113 +       if (flags & S_IMMUTABLE)
4114 +               ip->i_diskflags |= GFS2_DIF_IMMUTABLE;
4115 +       if (flags & S_IXUNLINK)
4116 +               ip->i_diskflags |= GFS2_DIF_IXUNLINK;
4117 +
4118 +       if (flags & S_APPEND)
4119 +               ip->i_diskflags |= GFS2_DIF_APPENDONLY;
4120 +       if (flags & S_NOATIME)
4121 +               ip->i_diskflags |= GFS2_DIF_NOATIME;
4122 +       if (flags & S_SYNC)
4123 +               ip->i_diskflags |= GFS2_DIF_SYNC;
4124 +
4125 +       if (vflags & V_BARRIER)
4126 +               ip->i_diskflags |= GFS2_DIF_BARRIER;
4127 +       if (vflags & V_COW)
4128 +               ip->i_diskflags |= GFS2_DIF_COW;
4129  }
4130
4131  /* Flags that can be set by user space */
4132 @@ -293,6 +342,37 @@ static int gfs2_set_flags(struct file *f
4133         return do_gfs2_set_flags(filp, gfsflags, ~GFS2_DIF_JDATA);
4134  }
4135
4136 +int gfs2_sync_flags(struct inode *inode, int flags, int vflags)
4137 +{
4138 +       struct gfs2_inode *ip = GFS2_I(inode);
4139 +       struct gfs2_sbd *sdp = GFS2_SB(inode);
4140 +       struct buffer_head *bh;
4141 +       struct gfs2_holder gh;
4142 +       int error;
4143 +
4144 +       error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
4145 +       if (error)
4146 +               return error;
4147 +       error = gfs2_trans_begin(sdp, RES_DINODE, 0);
4148 +       if (error)
4149 +               goto out;
4150 +       error = gfs2_meta_inode_buffer(ip, &bh);
4151 +       if (error)
4152 +               goto out_trans_end;
4153 +       gfs2_trans_add_bh(ip->i_gl, bh, 1);
4154 +       inode->i_flags = flags;
4155 +       inode->i_vflags = vflags;
4156 +       gfs2_get_inode_flags(inode);
4157 +       gfs2_dinode_out(ip, bh->b_data);
4158 +       brelse(bh);
4159 +       gfs2_set_aops(inode);
4160 +out_trans_end:
4161 +       gfs2_trans_end(sdp);
4162 +out:
4163 +       gfs2_glock_dq_uninit(&gh);
4164 +       return error;
4165 +}
4166 +
4167  static long gfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
4168  {
4169         switch(cmd) {
4170 diff -NurpP --minimal linux-2.6.36/fs/gfs2/inode.h linux-2.6.36-vs2.3.0.36.38/fs/gfs2/inode.h
4171 --- linux-2.6.36/fs/gfs2/inode.h        2010-08-02 16:52:49.000000000 +0200
4172 +++ linux-2.6.36-vs2.3.0.36.38/fs/gfs2/inode.h  2010-10-21 13:09:36.000000000 +0200
4173 @@ -109,6 +109,7 @@ extern const struct file_operations gfs2
4174  extern const struct file_operations gfs2_dir_fops_nolock;
4175
4176  extern void gfs2_set_inode_flags(struct inode *inode);
4177 +extern int gfs2_sync_flags(struct inode *inode, int flags, int vflags);
4178
4179  #ifdef CONFIG_GFS2_FS_LOCKING_DLM
4180  extern const struct file_operations gfs2_file_fops;
4181 diff -NurpP --minimal linux-2.6.36/fs/gfs2/ops_inode.c linux-2.6.36-vs2.3.0.36.38/fs/gfs2/ops_inode.c
4182 --- linux-2.6.36/fs/gfs2/ops_inode.c    2010-10-21 13:07:49.000000000 +0200
4183 +++ linux-2.6.36-vs2.3.0.36.38/fs/gfs2/ops_inode.c      2010-10-21 13:09:36.000000000 +0200
4184 @@ -1352,6 +1352,7 @@ const struct inode_operations gfs2_file_
4185         .listxattr = gfs2_listxattr,
4186         .removexattr = gfs2_removexattr,
4187         .fiemap = gfs2_fiemap,
4188 +       .sync_flags = gfs2_sync_flags,
4189  };
4190
4191  const struct inode_operations gfs2_dir_iops = {
4192 @@ -1372,6 +1373,7 @@ const struct inode_operations gfs2_dir_i
4193         .listxattr = gfs2_listxattr,
4194         .removexattr = gfs2_removexattr,
4195         .fiemap = gfs2_fiemap,
4196 +       .sync_flags = gfs2_sync_flags,
4197  };
4198
4199  const struct inode_operations gfs2_symlink_iops = {
4200 diff -NurpP --minimal linux-2.6.36/fs/hfsplus/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/hfsplus/ioctl.c
4201 --- linux-2.6.36/fs/hfsplus/ioctl.c     2010-08-02 16:52:49.000000000 +0200
4202 +++ linux-2.6.36-vs2.3.0.36.38/fs/hfsplus/ioctl.c       2010-10-21 13:09:36.000000000 +0200
4203 @@ -18,6 +18,7 @@
4204  #include <linux/sched.h>
4205  #include <linux/xattr.h>
4206  #include <linux/smp_lock.h>
4207 +// #include <linux/mount.h>
4208  #include <asm/uaccess.h>
4209  #include "hfsplus_fs.h"
4210
4211 diff -NurpP --minimal linux-2.6.36/fs/inode.c linux-2.6.36-vs2.3.0.36.38/fs/inode.c
4212 --- linux-2.6.36/fs/inode.c     2010-10-21 13:07:49.000000000 +0200
4213 +++ linux-2.6.36-vs2.3.0.36.38/fs/inode.c       2010-10-21 13:09:36.000000000 +0200
4214 @@ -24,6 +24,7 @@
4215  #include <linux/mount.h>
4216  #include <linux/async.h>
4217  #include <linux/posix_acl.h>
4218 +#include <linux/vs_tag.h>
4219
4220  /*
4221   * This is needed for the following functions:
4222 @@ -130,6 +131,9 @@ int inode_init_always(struct super_block
4223         struct address_space *const mapping = &inode->i_data;
4224
4225         inode->i_sb = sb;
4226 +
4227 +       /* essential because of inode slab reuse */
4228 +       inode->i_tag = 0;
4229         inode->i_blkbits = sb->s_blocksize_bits;
4230         inode->i_flags = 0;
4231         atomic_set(&inode->i_count, 1);
4232 @@ -150,6 +154,7 @@ int inode_init_always(struct super_block
4233         inode->i_bdev = NULL;
4234         inode->i_cdev = NULL;
4235         inode->i_rdev = 0;
4236 +       inode->i_mdev = 0;
4237         inode->dirtied_when = 0;
4238
4239         if (security_inode_alloc(inode))
4240 @@ -393,6 +398,8 @@ static int invalidate_list(struct list_h
4241         return busy;
4242  }
4243
4244 +EXPORT_SYMBOL_GPL(__iget);
4245 +
4246  /**
4247   *     invalidate_inodes       - discard the inodes on a device
4248   *     @sb: superblock
4249 @@ -1528,9 +1535,11 @@ void init_special_inode(struct inode *in
4250         if (S_ISCHR(mode)) {
4251                 inode->i_fop = &def_chr_fops;
4252                 inode->i_rdev = rdev;
4253 +               inode->i_mdev = rdev;
4254         } else if (S_ISBLK(mode)) {
4255                 inode->i_fop = &def_blk_fops;
4256                 inode->i_rdev = rdev;
4257 +               inode->i_mdev = rdev;
4258         } else if (S_ISFIFO(mode))
4259                 inode->i_fop = &def_fifo_fops;
4260         else if (S_ISSOCK(mode))
4261 @@ -1559,5 +1568,6 @@ void inode_init_owner(struct inode *inod
4262         } else
4263                 inode->i_gid = current_fsgid();
4264         inode->i_mode = mode;
4265 +       inode->i_tag = dx_current_fstag(inode->i_sb);
4266  }
4267  EXPORT_SYMBOL(inode_init_owner);
4268 diff -NurpP --minimal linux-2.6.36/fs/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/ioctl.c
4269 --- linux-2.6.36/fs/ioctl.c     2010-10-21 13:07:49.000000000 +0200
4270 +++ linux-2.6.36-vs2.3.0.36.38/fs/ioctl.c       2010-10-21 13:09:36.000000000 +0200
4271 @@ -16,6 +16,9 @@
4272  #include <linux/writeback.h>
4273  #include <linux/buffer_head.h>
4274  #include <linux/falloc.h>
4275 +#include <linux/proc_fs.h>
4276 +#include <linux/vserver/inode.h>
4277 +#include <linux/vs_tag.h>
4278
4279  #include <asm/ioctls.h>
4280
4281 diff -NurpP --minimal linux-2.6.36/fs/ioprio.c linux-2.6.36-vs2.3.0.36.38/fs/ioprio.c
4282 --- linux-2.6.36/fs/ioprio.c    2010-07-07 18:31:52.000000000 +0200
4283 +++ linux-2.6.36-vs2.3.0.36.38/fs/ioprio.c      2010-10-21 13:09:36.000000000 +0200
4284 @@ -27,6 +27,7 @@
4285  #include <linux/syscalls.h>
4286  #include <linux/security.h>
4287  #include <linux/pid_namespace.h>
4288 +#include <linux/vs_base.h>
4289
4290  int set_task_ioprio(struct task_struct *task, int ioprio)
4291  {
4292 @@ -124,6 +125,8 @@ SYSCALL_DEFINE3(ioprio_set, int, which,
4293                         else
4294                                 pgrp = find_vpid(who);
4295                         do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
4296 +                               if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
4297 +                                       continue;
4298                                 ret = set_task_ioprio(p, ioprio);
4299                                 if (ret)
4300                                         break;
4301 @@ -213,6 +216,8 @@ SYSCALL_DEFINE2(ioprio_get, int, which,
4302                         else
4303                                 pgrp = find_vpid(who);
4304                         do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
4305 +                               if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
4306 +                                       continue;
4307                                 tmpio = get_task_ioprio(p);
4308                                 if (tmpio < 0)
4309                                         continue;
4310 diff -NurpP --minimal linux-2.6.36/fs/jfs/file.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/file.c
4311 --- linux-2.6.36/fs/jfs/file.c  2010-10-21 13:07:50.000000000 +0200
4312 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/file.c    2010-10-21 13:09:36.000000000 +0200
4313 @@ -102,7 +102,8 @@ int jfs_setattr(struct dentry *dentry, s
4314         if (is_quota_modification(inode, iattr))
4315                 dquot_initialize(inode);
4316         if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
4317 -           (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
4318 +           (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid) ||
4319 +           (iattr->ia_valid & ATTR_TAG && iattr->ia_tag != inode->i_tag)) {
4320                 rc = dquot_transfer(inode, iattr);
4321                 if (rc)
4322                         return rc;
4323 @@ -133,6 +134,7 @@ const struct inode_operations jfs_file_i
4324  #ifdef CONFIG_JFS_POSIX_ACL
4325         .check_acl      = jfs_check_acl,
4326  #endif
4327 +       .sync_flags     = jfs_sync_flags,
4328  };
4329
4330  const struct file_operations jfs_file_operations = {
4331 diff -NurpP --minimal linux-2.6.36/fs/jfs/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/ioctl.c
4332 --- linux-2.6.36/fs/jfs/ioctl.c 2008-12-25 00:26:37.000000000 +0100
4333 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/ioctl.c   2010-10-21 13:09:36.000000000 +0200
4334 @@ -11,6 +11,7 @@
4335  #include <linux/mount.h>
4336  #include <linux/time.h>
4337  #include <linux/sched.h>
4338 +#include <linux/mount.h>
4339  #include <asm/current.h>
4340  #include <asm/uaccess.h>
4341
4342 @@ -52,6 +53,16 @@ static long jfs_map_ext2(unsigned long f
4343  }
4344
4345
4346 +int jfs_sync_flags(struct inode *inode, int flags, int vflags)
4347 +{
4348 +       inode->i_flags = flags;
4349 +       inode->i_vflags = vflags;
4350 +       jfs_get_inode_flags(JFS_IP(inode));
4351 +       inode->i_ctime = CURRENT_TIME_SEC;
4352 +       mark_inode_dirty(inode);
4353 +       return 0;
4354 +}
4355 +
4356  long jfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
4357  {
4358         struct inode *inode = filp->f_dentry->d_inode;
4359 @@ -85,6 +96,11 @@ long jfs_ioctl(struct file *filp, unsign
4360                 if (!S_ISDIR(inode->i_mode))
4361                         flags &= ~JFS_DIRSYNC_FL;
4362
4363 +               if (IS_BARRIER(inode)) {
4364 +                       vxwprintk_task(1, "messing with the barrier.");
4365 +                       return -EACCES;
4366 +               }
4367 +
4368                 /* Is it quota file? Do not allow user to mess with it */
4369                 if (IS_NOQUOTA(inode)) {
4370                         err = -EPERM;
4371 @@ -102,8 +118,8 @@ long jfs_ioctl(struct file *filp, unsign
4372                  * the relevant capability.
4373                  */
4374                 if ((oldflags & JFS_IMMUTABLE_FL) ||
4375 -                       ((flags ^ oldflags) &
4376 -                       (JFS_APPEND_FL | JFS_IMMUTABLE_FL))) {
4377 +                       ((flags ^ oldflags) & (JFS_APPEND_FL |
4378 +                       JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL))) {
4379                         if (!capable(CAP_LINUX_IMMUTABLE)) {
4380                                 mutex_unlock(&inode->i_mutex);
4381                                 err = -EPERM;
4382 @@ -111,7 +127,7 @@ long jfs_ioctl(struct file *filp, unsign
4383                         }
4384                 }
4385
4386 -               flags = flags & JFS_FL_USER_MODIFIABLE;
4387 +               flags &= JFS_FL_USER_MODIFIABLE;
4388                 flags |= oldflags & ~JFS_FL_USER_MODIFIABLE;
4389                 jfs_inode->mode2 = flags;
4390
4391 diff -NurpP --minimal linux-2.6.36/fs/jfs/jfs_dinode.h linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_dinode.h
4392 --- linux-2.6.36/fs/jfs/jfs_dinode.h    2008-12-25 00:26:37.000000000 +0100
4393 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_dinode.h      2010-10-21 13:09:36.000000000 +0200
4394 @@ -161,9 +161,13 @@ struct dinode {
4395
4396  #define JFS_APPEND_FL          0x01000000 /* writes to file may only append */
4397  #define JFS_IMMUTABLE_FL       0x02000000 /* Immutable file */
4398 +#define JFS_IXUNLINK_FL                0x08000000 /* Immutable invert on unlink */
4399
4400 -#define JFS_FL_USER_VISIBLE    0x03F80000
4401 -#define JFS_FL_USER_MODIFIABLE 0x03F80000
4402 +#define JFS_BARRIER_FL         0x04000000 /* Barrier for chroot() */
4403 +#define JFS_COW_FL             0x20000000 /* Copy on Write marker */
4404 +
4405 +#define JFS_FL_USER_VISIBLE    0x07F80000
4406 +#define JFS_FL_USER_MODIFIABLE 0x07F80000
4407  #define JFS_FL_INHERIT         0x03C80000
4408
4409  /* These are identical to EXT[23]_IOC_GETFLAGS/SETFLAGS */
4410 diff -NurpP --minimal linux-2.6.36/fs/jfs/jfs_filsys.h linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_filsys.h
4411 --- linux-2.6.36/fs/jfs/jfs_filsys.h    2008-12-25 00:26:37.000000000 +0100
4412 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_filsys.h      2010-10-21 13:09:36.000000000 +0200
4413 @@ -263,6 +263,7 @@
4414  #define JFS_NAME_MAX   255
4415  #define JFS_PATH_MAX   BPSIZE
4416
4417 +#define JFS_TAGGED             0x00800000      /* Context Tagging */
4418
4419  /*
4420   *     file system state (superblock state)
4421 diff -NurpP --minimal linux-2.6.36/fs/jfs/jfs_imap.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_imap.c
4422 --- linux-2.6.36/fs/jfs/jfs_imap.c      2010-07-07 18:31:52.000000000 +0200
4423 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_imap.c        2010-10-21 13:09:36.000000000 +0200
4424 @@ -46,6 +46,7 @@
4425  #include <linux/pagemap.h>
4426  #include <linux/quotaops.h>
4427  #include <linux/slab.h>
4428 +#include <linux/vs_tag.h>
4429
4430  #include "jfs_incore.h"
4431  #include "jfs_inode.h"
4432 @@ -3060,6 +3061,8 @@ static int copy_from_dinode(struct dinod
4433  {
4434         struct jfs_inode_info *jfs_ip = JFS_IP(ip);
4435         struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
4436 +       uid_t uid;
4437 +       gid_t gid;
4438
4439         jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
4440         jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
4441 @@ -3080,14 +3083,18 @@ static int copy_from_dinode(struct dinod
4442         }
4443         ip->i_nlink = le32_to_cpu(dip->di_nlink);
4444
4445 -       jfs_ip->saved_uid = le32_to_cpu(dip->di_uid);
4446 +       uid = le32_to_cpu(dip->di_uid);
4447 +       gid = le32_to_cpu(dip->di_gid);
4448 +       ip->i_tag = INOTAG_TAG(DX_TAG(ip), uid, gid, 0);
4449 +
4450 +       jfs_ip->saved_uid = INOTAG_UID(DX_TAG(ip), uid, gid);
4451         if (sbi->uid == -1)
4452                 ip->i_uid = jfs_ip->saved_uid;
4453         else {
4454                 ip->i_uid = sbi->uid;
4455         }
4456
4457 -       jfs_ip->saved_gid = le32_to_cpu(dip->di_gid);
4458 +       jfs_ip->saved_gid = INOTAG_GID(DX_TAG(ip), uid, gid);
4459         if (sbi->gid == -1)
4460                 ip->i_gid = jfs_ip->saved_gid;
4461         else {
4462 @@ -3152,14 +3159,12 @@ static void copy_to_dinode(struct dinode
4463         dip->di_size = cpu_to_le64(ip->i_size);
4464         dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
4465         dip->di_nlink = cpu_to_le32(ip->i_nlink);
4466 -       if (sbi->uid == -1)
4467 -               dip->di_uid = cpu_to_le32(ip->i_uid);
4468 -       else
4469 -               dip->di_uid = cpu_to_le32(jfs_ip->saved_uid);
4470 -       if (sbi->gid == -1)
4471 -               dip->di_gid = cpu_to_le32(ip->i_gid);
4472 -       else
4473 -               dip->di_gid = cpu_to_le32(jfs_ip->saved_gid);
4474 +
4475 +       dip->di_uid = cpu_to_le32(TAGINO_UID(DX_TAG(ip),
4476 +               (sbi->uid == -1) ? ip->i_uid : jfs_ip->saved_uid, ip->i_tag));
4477 +       dip->di_gid = cpu_to_le32(TAGINO_GID(DX_TAG(ip),
4478 +               (sbi->gid == -1) ? ip->i_gid : jfs_ip->saved_gid, ip->i_tag));
4479 +
4480         jfs_get_inode_flags(jfs_ip);
4481         /*
4482          * mode2 is only needed for storing the higher order bits.
4483 diff -NurpP --minimal linux-2.6.36/fs/jfs/jfs_inode.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_inode.c
4484 --- linux-2.6.36/fs/jfs/jfs_inode.c     2010-08-02 16:52:49.000000000 +0200
4485 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_inode.c       2010-10-21 13:09:36.000000000 +0200
4486 @@ -18,6 +18,7 @@
4487
4488  #include <linux/fs.h>
4489  #include <linux/quotaops.h>
4490 +#include <linux/vs_tag.h>
4491  #include "jfs_incore.h"
4492  #include "jfs_inode.h"
4493  #include "jfs_filsys.h"
4494 @@ -30,29 +31,46 @@ void jfs_set_inode_flags(struct inode *i
4495  {
4496         unsigned int flags = JFS_IP(inode)->mode2;
4497
4498 -       inode->i_flags &= ~(S_IMMUTABLE | S_APPEND |
4499 -               S_NOATIME | S_DIRSYNC | S_SYNC);
4500 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
4501 +               S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
4502
4503         if (flags & JFS_IMMUTABLE_FL)
4504                 inode->i_flags |= S_IMMUTABLE;
4505 +       if (flags & JFS_IXUNLINK_FL)
4506 +               inode->i_flags |= S_IXUNLINK;
4507 +
4508 +       if (flags & JFS_SYNC_FL)
4509 +               inode->i_flags |= S_SYNC;
4510         if (flags & JFS_APPEND_FL)
4511                 inode->i_flags |= S_APPEND;
4512         if (flags & JFS_NOATIME_FL)
4513                 inode->i_flags |= S_NOATIME;
4514         if (flags & JFS_DIRSYNC_FL)
4515                 inode->i_flags |= S_DIRSYNC;
4516 -       if (flags & JFS_SYNC_FL)
4517 -               inode->i_flags |= S_SYNC;
4518 +
4519 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
4520 +
4521 +       if (flags & JFS_BARRIER_FL)
4522 +               inode->i_vflags |= V_BARRIER;
4523 +       if (flags & JFS_COW_FL)
4524 +               inode->i_vflags |= V_COW;
4525  }
4526
4527  void jfs_get_inode_flags(struct jfs_inode_info *jfs_ip)
4528  {
4529         unsigned int flags = jfs_ip->vfs_inode.i_flags;
4530 +       unsigned int vflags = jfs_ip->vfs_inode.i_vflags;
4531 +
4532 +       jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_IXUNLINK_FL |
4533 +                          JFS_APPEND_FL | JFS_NOATIME_FL |
4534 +                          JFS_DIRSYNC_FL | JFS_SYNC_FL |
4535 +                          JFS_BARRIER_FL | JFS_COW_FL);
4536
4537 -       jfs_ip->mode2 &= ~(JFS_IMMUTABLE_FL | JFS_APPEND_FL | JFS_NOATIME_FL |
4538 -                          JFS_DIRSYNC_FL | JFS_SYNC_FL);
4539         if (flags & S_IMMUTABLE)
4540                 jfs_ip->mode2 |= JFS_IMMUTABLE_FL;
4541 +       if (flags & S_IXUNLINK)
4542 +               jfs_ip->mode2 |= JFS_IXUNLINK_FL;
4543 +
4544         if (flags & S_APPEND)
4545                 jfs_ip->mode2 |= JFS_APPEND_FL;
4546         if (flags & S_NOATIME)
4547 @@ -61,6 +79,11 @@ void jfs_get_inode_flags(struct jfs_inod
4548                 jfs_ip->mode2 |= JFS_DIRSYNC_FL;
4549         if (flags & S_SYNC)
4550                 jfs_ip->mode2 |= JFS_SYNC_FL;
4551 +
4552 +       if (vflags & V_BARRIER)
4553 +               jfs_ip->mode2 |= JFS_BARRIER_FL;
4554 +       if (vflags & V_COW)
4555 +               jfs_ip->mode2 |= JFS_COW_FL;
4556  }
4557
4558  /*
4559 diff -NurpP --minimal linux-2.6.36/fs/jfs/jfs_inode.h linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_inode.h
4560 --- linux-2.6.36/fs/jfs/jfs_inode.h     2010-10-21 13:07:50.000000000 +0200
4561 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/jfs_inode.h       2010-10-21 13:09:36.000000000 +0200
4562 @@ -39,6 +39,7 @@ extern struct dentry *jfs_fh_to_dentry(s
4563  extern struct dentry *jfs_fh_to_parent(struct super_block *sb, struct fid *fid,
4564         int fh_len, int fh_type);
4565  extern void jfs_set_inode_flags(struct inode *);
4566 +extern int jfs_sync_flags(struct inode *, int, int);
4567  extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
4568  extern int jfs_setattr(struct dentry *, struct iattr *);
4569
4570 diff -NurpP --minimal linux-2.6.36/fs/jfs/namei.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/namei.c
4571 --- linux-2.6.36/fs/jfs/namei.c 2010-07-07 18:31:52.000000000 +0200
4572 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/namei.c   2010-10-21 13:09:36.000000000 +0200
4573 @@ -21,6 +21,7 @@
4574  #include <linux/ctype.h>
4575  #include <linux/quotaops.h>
4576  #include <linux/exportfs.h>
4577 +#include <linux/vs_tag.h>
4578  #include "jfs_incore.h"
4579  #include "jfs_superblock.h"
4580  #include "jfs_inode.h"
4581 @@ -1491,6 +1492,7 @@ static struct dentry *jfs_lookup(struct
4582                 return ERR_CAST(ip);
4583         }
4584
4585 +       dx_propagate_tag(nd, ip);
4586         dentry = d_splice_alias(ip, dentry);
4587
4588         if (dentry && (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2))
4589 @@ -1560,6 +1562,7 @@ const struct inode_operations jfs_dir_in
4590  #ifdef CONFIG_JFS_POSIX_ACL
4591         .check_acl      = jfs_check_acl,
4592  #endif
4593 +       .sync_flags     = jfs_sync_flags,
4594  };
4595
4596  const struct file_operations jfs_dir_operations = {
4597 diff -NurpP --minimal linux-2.6.36/fs/jfs/super.c linux-2.6.36-vs2.3.0.36.38/fs/jfs/super.c
4598 --- linux-2.6.36/fs/jfs/super.c 2010-10-21 13:07:50.000000000 +0200
4599 +++ linux-2.6.36-vs2.3.0.36.38/fs/jfs/super.c   2010-10-21 13:09:36.000000000 +0200
4600 @@ -195,7 +195,8 @@ static void jfs_put_super(struct super_b
4601  enum {
4602         Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
4603         Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
4604 -       Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
4605 +       Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask,
4606 +       Opt_tag, Opt_notag, Opt_tagid
4607  };
4608
4609  static const match_table_t tokens = {
4610 @@ -205,6 +206,10 @@ static const match_table_t tokens = {
4611         {Opt_resize, "resize=%u"},
4612         {Opt_resize_nosize, "resize"},
4613         {Opt_errors, "errors=%s"},
4614 +       {Opt_tag, "tag"},
4615 +       {Opt_notag, "notag"},
4616 +       {Opt_tagid, "tagid=%u"},
4617 +       {Opt_tag, "tagxid"},
4618         {Opt_ignore, "noquota"},
4619         {Opt_ignore, "quota"},
4620         {Opt_usrquota, "usrquota"},
4621 @@ -339,6 +344,20 @@ static int parse_options(char *options,
4622                         }
4623                         break;
4624                 }
4625 +#ifndef CONFIG_TAGGING_NONE
4626 +               case Opt_tag:
4627 +                       *flag |= JFS_TAGGED;
4628 +                       break;
4629 +               case Opt_notag:
4630 +                       *flag &= JFS_TAGGED;
4631 +                       break;
4632 +#endif
4633 +#ifdef CONFIG_PROPAGATE
4634 +               case Opt_tagid:
4635 +                       /* use args[0] */
4636 +                       *flag |= JFS_TAGGED;
4637 +                       break;
4638 +#endif
4639                 default:
4640                         printk("jfs: Unrecognized mount option \"%s\" "
4641                                         " or missing value\n", p);
4642 @@ -369,6 +388,12 @@ static int jfs_remount(struct super_bloc
4643         if (!parse_options(data, sb, &newLVSize, &flag)) {
4644                 return -EINVAL;
4645         }
4646 +       if ((flag & JFS_TAGGED) && !(sb->s_flags & MS_TAGGED)) {
4647 +               printk(KERN_ERR "JFS: %s: tagging not permitted on remount.\n",
4648 +                       sb->s_id);
4649 +               return -EINVAL;
4650 +       }
4651 +
4652         lock_kernel();
4653         if (newLVSize) {
4654                 if (sb->s_flags & MS_RDONLY) {
4655 @@ -460,6 +485,9 @@ static int jfs_fill_super(struct super_b
4656  #ifdef CONFIG_JFS_POSIX_ACL
4657         sb->s_flags |= MS_POSIXACL;
4658  #endif
4659 +       /* map mount option tagxid */
4660 +       if (sbi->flag & JFS_TAGGED)
4661 +               sb->s_flags |= MS_TAGGED;
4662
4663         if (newLVSize) {
4664                 printk(KERN_ERR "resize option for remount only\n");
4665 diff -NurpP --minimal linux-2.6.36/fs/libfs.c linux-2.6.36-vs2.3.0.36.38/fs/libfs.c
4666 --- linux-2.6.36/fs/libfs.c     2010-10-21 13:07:50.000000000 +0200
4667 +++ linux-2.6.36-vs2.3.0.36.38/fs/libfs.c       2010-10-21 13:09:36.000000000 +0200
4668 @@ -124,7 +124,8 @@ static inline unsigned char dt_type(stru
4669   * both impossible due to the lock on directory.
4670   */
4671
4672 -int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
4673 +static inline int do_dcache_readdir_filter(struct file *filp,
4674 +       void *dirent, filldir_t filldir, int (*filter)(struct dentry *dentry))
4675  {
4676         struct dentry *dentry = filp->f_path.dentry;
4677         struct dentry *cursor = filp->private_data;
4678 @@ -157,6 +158,8 @@ int dcache_readdir(struct file * filp, v
4679                                 next = list_entry(p, struct dentry, d_u.d_child);
4680                                 if (d_unhashed(next) || !next->d_inode)
4681                                         continue;
4682 +                               if (filter && !filter(next))
4683 +                                       continue;
4684
4685                                 spin_unlock(&dcache_lock);
4686                                 if (filldir(dirent, next->d_name.name,
4687 @@ -175,6 +178,18 @@ int dcache_readdir(struct file * filp, v
4688         return 0;
4689  }
4690
4691 +int dcache_readdir(struct file *filp, void *dirent, filldir_t filldir)
4692 +{
4693 +       return do_dcache_readdir_filter(filp, dirent, filldir, NULL);
4694 +}
4695 +
4696 +int dcache_readdir_filter(struct file *filp, void *dirent, filldir_t filldir,
4697 +       int (*filter)(struct dentry *))
4698 +{
4699 +       return do_dcache_readdir_filter(filp, dirent, filldir, filter);
4700 +}
4701 +
4702 +
4703  ssize_t generic_read_dir(struct file *filp, char __user *buf, size_t siz, loff_t *ppos)
4704  {
4705         return -EISDIR;
4706 @@ -925,6 +940,7 @@ EXPORT_SYMBOL(dcache_dir_close);
4707  EXPORT_SYMBOL(dcache_dir_lseek);
4708  EXPORT_SYMBOL(dcache_dir_open);
4709  EXPORT_SYMBOL(dcache_readdir);
4710 +EXPORT_SYMBOL(dcache_readdir_filter);
4711  EXPORT_SYMBOL(generic_read_dir);
4712  EXPORT_SYMBOL(get_sb_pseudo);
4713  EXPORT_SYMBOL(simple_write_begin);
4714 diff -NurpP --minimal linux-2.6.36/fs/locks.c linux-2.6.36-vs2.3.0.36.38/fs/locks.c
4715 --- linux-2.6.36/fs/locks.c     2010-07-07 18:31:52.000000000 +0200
4716 +++ linux-2.6.36-vs2.3.0.36.38/fs/locks.c       2010-10-21 13:09:36.000000000 +0200
4717 @@ -127,6 +127,8 @@
4718  #include <linux/time.h>
4719  #include <linux/rcupdate.h>
4720  #include <linux/pid_namespace.h>
4721 +#include <linux/vs_base.h>
4722 +#include <linux/vs_limit.h>
4723
4724  #include <asm/uaccess.h>
4725
4726 @@ -148,6 +150,8 @@ static struct kmem_cache *filelock_cache
4727  /* Allocate an empty lock structure. */
4728  static struct file_lock *locks_alloc_lock(void)
4729  {
4730 +       if (!vx_locks_avail(1))
4731 +               return NULL;
4732         return kmem_cache_alloc(filelock_cache, GFP_KERNEL);
4733  }
4734
4735 @@ -174,6 +178,7 @@ static void locks_free_lock(struct file_
4736         BUG_ON(!list_empty(&fl->fl_block));
4737         BUG_ON(!list_empty(&fl->fl_link));
4738
4739 +       vx_locks_dec(fl);
4740         locks_release_private(fl);
4741         kmem_cache_free(filelock_cache, fl);
4742  }
4743 @@ -194,6 +199,7 @@ void locks_init_lock(struct file_lock *f
4744         fl->fl_start = fl->fl_end = 0;
4745         fl->fl_ops = NULL;
4746         fl->fl_lmops = NULL;
4747 +       fl->fl_xid = -1;
4748  }
4749
4750  EXPORT_SYMBOL(locks_init_lock);
4751 @@ -248,6 +254,7 @@ void locks_copy_lock(struct file_lock *n
4752         new->fl_file = fl->fl_file;
4753         new->fl_ops = fl->fl_ops;
4754         new->fl_lmops = fl->fl_lmops;
4755 +       new->fl_xid = fl->fl_xid;
4756
4757         locks_copy_private(new, fl);
4758  }
4759 @@ -286,6 +293,11 @@ static int flock_make_lock(struct file *
4760         fl->fl_flags = FL_FLOCK;
4761         fl->fl_type = type;
4762         fl->fl_end = OFFSET_MAX;
4763 +
4764 +       vxd_assert(filp->f_xid == vx_current_xid(),
4765 +               "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4766 +       fl->fl_xid = filp->f_xid;
4767 +       vx_locks_inc(fl);
4768
4769         *lock = fl;
4770         return 0;
4771 @@ -451,6 +463,7 @@ static int lease_init(struct file *filp,
4772
4773         fl->fl_owner = current->files;
4774         fl->fl_pid = current->tgid;
4775 +       fl->fl_xid = vx_current_xid();
4776
4777         fl->fl_file = filp;
4778         fl->fl_flags = FL_LEASE;
4779 @@ -470,6 +483,11 @@ static struct file_lock *lease_alloc(str
4780         if (fl == NULL)
4781                 return ERR_PTR(error);
4782
4783 +       fl->fl_xid = vx_current_xid();
4784 +       if (filp)
4785 +               vxd_assert(filp->f_xid == fl->fl_xid,
4786 +                       "f_xid(%d) == fl_xid(%d)", filp->f_xid, fl->fl_xid);
4787 +       vx_locks_inc(fl);
4788         error = lease_init(filp, type, fl);
4789         if (error) {
4790                 locks_free_lock(fl);
4791 @@ -770,6 +788,7 @@ static int flock_lock_file(struct file *
4792         if (found)
4793                 cond_resched();
4794
4795 +       new_fl->fl_xid = -1;
4796  find_conflict:
4797         for_each_lock(inode, before) {
4798                 struct file_lock *fl = *before;
4799 @@ -790,6 +809,7 @@ find_conflict:
4800                 goto out;
4801         locks_copy_lock(new_fl, request);
4802         locks_insert_lock(before, new_fl);
4803 +       vx_locks_inc(new_fl);
4804         new_fl = NULL;
4805         error = 0;
4806
4807 @@ -800,7 +820,8 @@ out:
4808         return error;
4809  }
4810
4811 -static int __posix_lock_file(struct inode *inode, struct file_lock *request, struct file_lock *conflock)
4812 +static int __posix_lock_file(struct inode *inode, struct file_lock *request,
4813 +       struct file_lock *conflock, xid_t xid)
4814  {
4815         struct file_lock *fl;
4816         struct file_lock *new_fl = NULL;
4817 @@ -810,6 +831,8 @@ static int __posix_lock_file(struct inod
4818         struct file_lock **before;
4819         int error, added = 0;
4820
4821 +       vxd_assert(xid == vx_current_xid(),
4822 +               "xid(%d) == current(%d)", xid, vx_current_xid());
4823         /*
4824          * We may need two file_lock structures for this operation,
4825          * so we get them in advance to avoid races.
4826 @@ -820,7 +843,11 @@ static int __posix_lock_file(struct inod
4827             (request->fl_type != F_UNLCK ||
4828              request->fl_start != 0 || request->fl_end != OFFSET_MAX)) {
4829                 new_fl = locks_alloc_lock();
4830 +               new_fl->fl_xid = xid;
4831 +               vx_locks_inc(new_fl);
4832                 new_fl2 = locks_alloc_lock();
4833 +               new_fl2->fl_xid = xid;
4834 +               vx_locks_inc(new_fl2);
4835         }
4836
4837         lock_kernel();
4838 @@ -1019,7 +1046,8 @@ static int __posix_lock_file(struct inod
4839  int posix_lock_file(struct file *filp, struct file_lock *fl,
4840                         struct file_lock *conflock)
4841  {
4842 -       return __posix_lock_file(filp->f_path.dentry->d_inode, fl, conflock);
4843 +       return __posix_lock_file(filp->f_path.dentry->d_inode,
4844 +               fl, conflock, filp->f_xid);
4845  }
4846  EXPORT_SYMBOL(posix_lock_file);
4847
4848 @@ -1109,7 +1137,7 @@ int locks_mandatory_area(int read_write,
4849         fl.fl_end = offset + count - 1;
4850
4851         for (;;) {
4852 -               error = __posix_lock_file(inode, &fl, NULL);
4853 +               error = __posix_lock_file(inode, &fl, NULL, filp->f_xid);
4854                 if (error != FILE_LOCK_DEFERRED)
4855                         break;
4856                 error = wait_event_interruptible(fl.fl_wait, !fl.fl_next);
4857 @@ -1425,6 +1453,7 @@ int generic_setlease(struct file *filp,
4858
4859         locks_copy_lock(new_fl, lease);
4860         locks_insert_lock(before, new_fl);
4861 +       vx_locks_inc(new_fl);
4862
4863         *flp = new_fl;
4864         return 0;
4865 @@ -1780,6 +1809,11 @@ int fcntl_setlk(unsigned int fd, struct
4866         if (file_lock == NULL)
4867                 return -ENOLCK;
4868
4869 +       vxd_assert(filp->f_xid == vx_current_xid(),
4870 +               "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4871 +       file_lock->fl_xid = filp->f_xid;
4872 +       vx_locks_inc(file_lock);
4873 +
4874         /*
4875          * This might block, so we do it before checking the inode.
4876          */
4877 @@ -1898,6 +1932,11 @@ int fcntl_setlk64(unsigned int fd, struc
4878         if (file_lock == NULL)
4879                 return -ENOLCK;
4880
4881 +       vxd_assert(filp->f_xid == vx_current_xid(),
4882 +               "f_xid(%d) == current(%d)", filp->f_xid, vx_current_xid());
4883 +       file_lock->fl_xid = filp->f_xid;
4884 +       vx_locks_inc(file_lock);
4885 +
4886         /*
4887          * This might block, so we do it before checking the inode.
4888          */
4889 @@ -2163,8 +2202,11 @@ static int locks_show(struct seq_file *f
4890
4891         lock_get_status(f, fl, (long)f->private, "");
4892
4893 -       list_for_each_entry(bfl, &fl->fl_block, fl_block)
4894 +       list_for_each_entry(bfl, &fl->fl_block, fl_block) {
4895 +               if (!vx_check(fl->fl_xid, VS_WATCH_P | VS_IDENT))
4896 +                       continue;
4897                 lock_get_status(f, bfl, (long)f->private, " ->");
4898 +       }
4899
4900         f->private++;
4901         return 0;
4902 diff -NurpP --minimal linux-2.6.36/fs/namei.c linux-2.6.36-vs2.3.0.36.38/fs/namei.c
4903 --- linux-2.6.36/fs/namei.c     2010-10-21 13:07:50.000000000 +0200
4904 +++ linux-2.6.36-vs2.3.0.36.38/fs/namei.c       2010-10-21 22:29:08.000000000 +0200
4905 @@ -32,6 +32,14 @@
4906  #include <linux/fcntl.h>
4907  #include <linux/device_cgroup.h>
4908  #include <linux/fs_struct.h>
4909 +#include <linux/proc_fs.h>
4910 +#include <linux/vserver/inode.h>
4911 +#include <linux/vs_base.h>
4912 +#include <linux/vs_tag.h>
4913 +#include <linux/vs_cowbl.h>
4914 +#include <linux/vs_device.h>
4915 +#include <linux/vs_context.h>
4916 +#include <linux/pid_namespace.h>
4917  #include <asm/uaccess.h>
4918
4919  #include "internal.h"
4920 @@ -166,6 +174,84 @@ void putname(const char *name)
4921  EXPORT_SYMBOL(putname);
4922  #endif
4923
4924 +static inline int dx_barrier(const struct inode *inode)
4925 +{
4926 +       if (IS_BARRIER(inode) && !vx_check(0, VS_ADMIN | VS_WATCH)) {
4927 +               vxwprintk_task(1, "did hit the barrier.");
4928 +               return 1;
4929 +       }
4930 +       return 0;
4931 +}
4932 +
4933 +static int __dx_permission(const struct inode *inode, int mask)
4934 +{
4935 +       if (dx_barrier(inode))
4936 +               return -EACCES;
4937 +
4938 +       if (inode->i_sb->s_magic == DEVPTS_SUPER_MAGIC) {
4939 +               /* devpts is xid tagged */
4940 +               if (S_ISDIR(inode->i_mode) ||
4941 +                   vx_check((xid_t)inode->i_tag, VS_IDENT | VS_WATCH_P))
4942 +                       return 0;
4943 +
4944 +               /* just pretend we didn't find anything */
4945 +               return -ENOENT;
4946 +       }
4947 +       else if (inode->i_sb->s_magic == PROC_SUPER_MAGIC) {
4948 +               struct proc_dir_entry *de = PDE(inode);
4949 +
4950 +               if (de && !vx_hide_check(0, de->vx_flags))
4951 +                       goto out;
4952 +
4953 +               if ((mask & (MAY_WRITE | MAY_APPEND))) {
4954 +                       struct pid *pid;
4955 +                       struct task_struct *tsk;
4956 +
4957 +                       if (vx_check(0, VS_ADMIN | VS_WATCH_P) ||
4958 +                           vx_flags(VXF_STATE_SETUP, 0))
4959 +                               return 0;
4960 +
4961 +                       pid = PROC_I(inode)->pid;
4962 +                       if (!pid)
4963 +                               goto out;
4964 +
4965 +                       tsk = pid_task(pid, PIDTYPE_PID);
4966 +                       vxdprintk(VXD_CBIT(tag, 0), "accessing %p[#%u]",
4967 +                                 tsk, (tsk ? vx_task_xid(tsk) : 0));
4968 +                       if (tsk && vx_check(vx_task_xid(tsk), VS_IDENT | VS_WATCH_P))
4969 +                               return 0;
4970 +               }
4971 +               else {
4972 +                       /* FIXME: Should we block some entries here? */
4973 +                       return 0;
4974 +               }
4975 +       }
4976 +       else {
4977 +               if (dx_notagcheck(inode->i_sb) ||
4978 +                   dx_check(inode->i_tag, DX_HOSTID | DX_ADMIN | DX_WATCH |
4979 +                            DX_IDENT))
4980 +                       return 0;
4981 +       }
4982 +
4983 +out:
4984 +       return -EACCES;
4985 +}
4986 +
4987 +int dx_permission(const struct inode *inode, int mask)
4988 +{
4989 +       int ret = __dx_permission(inode, mask);
4990 +       if (unlikely(ret)) {
4991 +#ifndef        CONFIG_VSERVER_WARN_DEVPTS
4992 +               if (inode->i_sb->s_magic != DEVPTS_SUPER_MAGIC)
4993 +#endif
4994 +                   vxwprintk_task(1,
4995 +                       "denied [0x%x] access to inode %s:%p[#%d,%lu]",
4996 +                       mask, inode->i_sb->s_id, inode, inode->i_tag,
4997 +                       inode->i_ino);
4998 +       }
4999 +       return ret;
5000 +}
5001 +
5002  /*
5003   * This does basic POSIX ACL permission checking
5004   */
5005 @@ -266,10 +352,14 @@ int inode_permission(struct inode *inode
5006                 /*
5007                  * Nobody gets write access to an immutable file.
5008                  */
5009 -               if (IS_IMMUTABLE(inode))
5010 +               if (IS_IMMUTABLE(inode) && !IS_COW(inode))
5011                         return -EACCES;
5012         }
5013
5014 +       retval = dx_permission(inode, mask);
5015 +       if (retval)
5016 +               return retval;
5017 +
5018         if (inode->i_op->permission)
5019                 retval = inode->i_op->permission(inode, mask);
5020         else
5021 @@ -463,6 +553,9 @@ static int exec_permission(struct inode
5022  {
5023         int ret;
5024
5025 +       if (dx_barrier(inode))
5026 +               return -EACCES;
5027 +
5028         if (inode->i_op->permission) {
5029                 ret = inode->i_op->permission(inode, MAY_EXEC);
5030                 if (!ret)
5031 @@ -672,7 +765,8 @@ static __always_inline void follow_dotdo
5032
5033                 if (nd->path.dentry == nd->root.dentry &&
5034                     nd->path.mnt == nd->root.mnt) {
5035 -                       break;
5036 +                       /* for sane '/' avoid follow_mount() */
5037 +                       return;
5038                 }
5039                 if (nd->path.dentry != nd->path.mnt->mnt_root) {
5040                         /* rare case of legitimate dget_parent()... */
5041 @@ -725,7 +819,7 @@ static int do_lookup(struct nameidata *n
5042  {
5043         struct vfsmount *mnt = nd->path.mnt;
5044         struct dentry *dentry, *parent;
5045 -       struct inode *dir;
5046 +       struct inode *dir, *inode;
5047         /*
5048          * See if the low-level filesystem might want
5049          * to use its own hash..
5050 @@ -747,12 +841,31 @@ static int do_lookup(struct nameidata *n
5051  found:
5052         if (dentry->d_op && dentry->d_op->d_revalidate)
5053                 goto need_revalidate;
5054 +
5055 +       inode = dentry->d_inode;
5056 +       if (!inode)
5057 +               goto done;
5058 +
5059 +       if (__dx_permission(inode, MAY_ACCESS))
5060 +               goto hidden;
5061  done:
5062         path->mnt = mnt;
5063         path->dentry = dentry;
5064         __follow_mount(path);
5065         return 0;
5066
5067 +hidden:
5068 +#ifndef        CONFIG_VSERVER_WARN_DEVPTS
5069 +       if (inode->i_sb->s_magic != DEVPTS_SUPER_MAGIC)
5070 +#endif
5071 +           vxwprintk_task(1,
5072 +               "did lookup hidden %s:%p[#%d,%lu] " VS_Q("%s/%.*s") ".",
5073 +               inode->i_sb->s_id, inode, inode->i_tag, inode->i_ino,
5074 +               vxd_path(&nd->path), name->len, name->name);
5075 +
5076 +       dput(dentry);
5077 +       return -ENOENT;
5078 +
5079  need_lookup:
5080         parent = nd->path.dentry;
5081         dir = parent->d_inode;
5082 @@ -1307,7 +1420,7 @@ static int may_delete(struct inode *dir,
5083         if (IS_APPEND(dir))
5084                 return -EPERM;
5085         if (check_sticky(dir, victim->d_inode)||IS_APPEND(victim->d_inode)||
5086 -           IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5087 +               IS_IXORUNLINK(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
5088                 return -EPERM;
5089         if (isdir) {
5090                 if (!S_ISDIR(victim->d_inode->i_mode))
5091 @@ -1430,6 +1543,14 @@ int may_open(struct path *path, int acc_
5092                 break;
5093         }
5094
5095 +#ifdef CONFIG_VSERVER_COWBL
5096 +       if (IS_COW(inode) && (flag & FMODE_WRITE)) {
5097 +               if (IS_COW_LINK(inode))
5098 +                       return -EMLINK;
5099 +               inode->i_flags &= ~(S_IXUNLINK|S_IMMUTABLE);
5100 +               mark_inode_dirty(inode);
5101 +       }
5102 +#endif
5103         error = inode_permission(inode, acc_mode);
5104         if (error)
5105                 return error;
5106 @@ -1538,7 +1659,8 @@ static int open_will_truncate(int flag,
5107  }
5108
5109  static struct file *finish_open(struct nameidata *nd,
5110 -                               int open_flag, int acc_mode)
5111 +                               int open_flag, int acc_mode,
5112 +                               const char *pathname)
5113  {
5114         struct file *filp;
5115         int will_truncate;
5116 @@ -1551,6 +1673,23 @@ static struct file *finish_open(struct n
5117                         goto exit;
5118         }
5119         error = may_open(&nd->path, acc_mode, open_flag);
5120 +#ifdef CONFIG_VSERVER_COWBL
5121 +       if (error == -EMLINK) {
5122 +               struct dentry *dentry;
5123 +               dentry = cow_break_link(pathname);
5124 +               if (IS_ERR(dentry)) {
5125 +                       error = PTR_ERR(dentry);
5126 +                       goto exit_cow;
5127 +               }
5128 +               dput(dentry);
5129 +               if (will_truncate)
5130 +                       mnt_drop_write(nd->path.mnt);
5131 +               release_open_intent(nd);
5132 +               path_put(&nd->path);
5133 +               return ERR_PTR(-EMLINK);
5134 +       }
5135 +exit_cow:
5136 +#endif
5137         if (error) {
5138                 if (will_truncate)
5139                         mnt_drop_write(nd->path.mnt);
5140 @@ -1719,7 +1858,7 @@ static struct file *do_last(struct namei
5141         if (S_ISDIR(path->dentry->d_inode->i_mode))
5142                 goto exit;
5143  ok:
5144 -       filp = finish_open(nd, open_flag, acc_mode);
5145 +       filp = finish_open(nd, open_flag, acc_mode, pathname);
5146         return filp;
5147
5148  exit_mutex_unlock:
5149 @@ -1748,7 +1887,11 @@ struct file *do_filp_open(int dfd, const
5150         int count = 0;
5151         int flag = open_to_namei_flags(open_flag);
5152         int force_reval = 0;
5153 -
5154 +#ifdef CONFIG_VSERVER_COWBL
5155 +       int rflag = flag;
5156 +       int rmode = mode;
5157 +restart:
5158 +#endif
5159         if (!(open_flag & O_CREAT))
5160                 mode = 0;
5161
5162 @@ -1814,6 +1957,13 @@ reval:
5163         if (!(open_flag & O_NOFOLLOW))
5164                 nd.flags |= LOOKUP_FOLLOW;
5165         filp = do_last(&nd, &path, open_flag, acc_mode, mode, pathname);
5166 +#ifdef CONFIG_VSERVER_COWBL
5167 +       if (unlikely(IS_ERR(filp) && PTR_ERR(filp) == -EMLINK)) {
5168 +               flag = rflag;
5169 +               mode = rmode;
5170 +               goto restart;
5171 +       }
5172 +#endif
5173         while (unlikely(!filp)) { /* trailing symlink */
5174                 struct path holder;
5175                 struct inode *inode = path.dentry->d_inode;
5176 @@ -1852,6 +2002,13 @@ reval:
5177                 holder = path;
5178                 nd.flags &= ~LOOKUP_PARENT;
5179                 filp = do_last(&nd, &path, open_flag, acc_mode, mode, pathname);
5180 +#ifdef CONFIG_VSERVER_COWBL
5181 +               if (unlikely(IS_ERR(filp) && PTR_ERR(filp) == -EMLINK)) {
5182 +                       flag = rflag;
5183 +                       mode = rmode;
5184 +                       goto restart;
5185 +               }
5186 +#endif
5187                 if (inode->i_op->put_link)
5188                         inode->i_op->put_link(holder.dentry, &nd, cookie);
5189                 path_put(&holder);
5190 @@ -1952,9 +2109,17 @@ int vfs_mknod(struct inode *dir, struct
5191         if (error)
5192                 return error;
5193
5194 -       if ((S_ISCHR(mode) || S_ISBLK(mode)) && !capable(CAP_MKNOD))
5195 +       if (!(S_ISCHR(mode) || S_ISBLK(mode)))
5196 +               goto okay;
5197 +
5198 +       if (!capable(CAP_MKNOD))
5199                 return -EPERM;
5200
5201 +       if (S_ISCHR(mode) && !vs_chrdev_perm(dev, DATTR_CREATE))
5202 +               return -EPERM;
5203 +       if (S_ISBLK(mode) && !vs_blkdev_perm(dev, DATTR_CREATE))
5204 +               return -EPERM;
5205 +okay:
5206         if (!dir->i_op->mknod)
5207                 return -EPERM;
5208
5209 @@ -2419,7 +2584,7 @@ int vfs_link(struct dentry *old_dentry,
5210         /*
5211          * A link to an append-only or immutable file cannot be created.
5212          */
5213 -       if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
5214 +       if (IS_APPEND(inode) || IS_IXORUNLINK(inode))
5215                 return -EPERM;
5216         if (!dir->i_op->link)
5217                 return -EPERM;
5218 @@ -2791,6 +2956,222 @@ int vfs_follow_link(struct nameidata *nd
5219         return __vfs_follow_link(nd, link);
5220  }
5221
5222 +
5223 +#ifdef CONFIG_VSERVER_COWBL
5224 +
5225 +#include <linux/file.h>
5226 +
5227 +static inline
5228 +long do_cow_splice(struct file *in, struct file *out, size_t len)
5229 +{
5230 +       loff_t ppos = 0;
5231 +
5232 +       return do_splice_direct(in, &ppos, out, len, 0);
5233 +}
5234 +
5235 +struct dentry *cow_break_link(const char *pathname)
5236 +{
5237 +       int ret, mode, pathlen, redo = 0;
5238 +       struct nameidata old_nd, dir_nd;
5239 +       struct path old_path, new_path;
5240 +       struct dentry *dir, *res = NULL;
5241 +       struct file *old_file;
5242 +       struct file *new_file;
5243 +       char *to, *path, pad='\251';
5244 +       loff_t size;
5245 +
5246 +       vxdprintk(VXD_CBIT(misc, 1),
5247 +               "cow_break_link(" VS_Q("%s") ")", pathname);
5248 +       path = kmalloc(PATH_MAX, GFP_KERNEL);
5249 +       ret = -ENOMEM;
5250 +       if (!path)
5251 +               goto out;
5252 +
5253 +       /* old_nd will have refs to dentry and mnt */
5254 +       ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5255 +       vxdprintk(VXD_CBIT(misc, 2), "path_lookup(old): %d", ret);
5256 +       if (ret < 0)
5257 +               goto out_free_path;
5258 +
5259 +       old_path = old_nd.path;
5260 +       mode = old_path.dentry->d_inode->i_mode;
5261 +
5262 +       to = d_path(&old_path, path, PATH_MAX-2);
5263 +       pathlen = strlen(to);
5264 +       vxdprintk(VXD_CBIT(misc, 2),
5265 +               "old path " VS_Q("%s") " [" VS_Q("%.*s") ":%d]", to,
5266 +               old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5267 +               old_path.dentry->d_name.len);
5268 +
5269 +       to[pathlen + 1] = 0;
5270 +retry:
5271 +       to[pathlen] = pad--;
5272 +       ret = -EMLINK;
5273 +       if (pad <= '\240')
5274 +               goto out_rel_old;
5275 +
5276 +       vxdprintk(VXD_CBIT(misc, 1), "temp copy " VS_Q("%s"), to);
5277 +       /* dir_nd will have refs to dentry and mnt */
5278 +       ret = path_lookup(to,
5279 +               LOOKUP_PARENT | LOOKUP_OPEN | LOOKUP_CREATE, &dir_nd);
5280 +       vxdprintk(VXD_CBIT(misc, 2),
5281 +               "path_lookup(new): %d", ret);
5282 +       if (ret < 0)
5283 +               goto retry;
5284 +
5285 +       /* this puppy downs the inode mutex */
5286 +       new_path.dentry = lookup_create(&dir_nd, 0);
5287 +       if (!new_path.dentry || IS_ERR(new_path.dentry)) {
5288 +               vxdprintk(VXD_CBIT(misc, 2),
5289 +                       "lookup_create(new): %p", new_path.dentry);
5290 +               mutex_unlock(&dir_nd.path.dentry->d_inode->i_mutex);
5291 +               path_put(&dir_nd.path);
5292 +               goto retry;
5293 +       }
5294 +       vxdprintk(VXD_CBIT(misc, 2),
5295 +               "lookup_create(new): %p [" VS_Q("%.*s") ":%d]",
5296 +               new_path.dentry,
5297 +               new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5298 +               new_path.dentry->d_name.len);
5299 +       dir = dir_nd.path.dentry;
5300 +
5301 +       ret = vfs_create(dir_nd.path.dentry->d_inode, new_path.dentry, mode, &dir_nd);
5302 +       vxdprintk(VXD_CBIT(misc, 2),
5303 +               "vfs_create(new): %d", ret);
5304 +       if (ret == -EEXIST) {
5305 +               mutex_unlock(&dir->d_inode->i_mutex);
5306 +               dput(new_path.dentry);
5307 +               path_put(&dir_nd.path);
5308 +               goto retry;
5309 +       }
5310 +       else if (ret < 0)
5311 +               goto out_unlock_new;
5312 +
5313 +       /* drop out early, ret passes ENOENT */
5314 +       ret = -ENOENT;
5315 +       if ((redo = d_unhashed(old_path.dentry)))
5316 +               goto out_unlock_new;
5317 +
5318 +       new_path.mnt = dir_nd.path.mnt;
5319 +       dget(old_path.dentry);
5320 +       mntget(old_path.mnt);
5321 +       /* this one cleans up the dentry/mnt in case of failure */
5322 +       old_file = dentry_open(old_path.dentry, old_path.mnt,
5323 +               O_RDONLY, current_cred());
5324 +       vxdprintk(VXD_CBIT(misc, 2),
5325 +               "dentry_open(old): %p", old_file);
5326 +       if (!old_file || IS_ERR(old_file)) {
5327 +               res = IS_ERR(old_file) ? (void *) old_file : res;
5328 +               goto out_unlock_new;
5329 +       }
5330 +
5331 +       dget(new_path.dentry);
5332 +       mntget(new_path.mnt);
5333 +       /* this one cleans up the dentry/mnt in case of failure */
5334 +       new_file = dentry_open(new_path.dentry, new_path.mnt,
5335 +               O_WRONLY, current_cred());
5336 +       vxdprintk(VXD_CBIT(misc, 2),
5337 +               "dentry_open(new): %p", new_file);
5338 +
5339 +       ret = IS_ERR(new_file) ? PTR_ERR(new_file) : -ENOENT;
5340 +       if (!new_file || IS_ERR(new_file))
5341 +               goto out_fput_old;
5342 +
5343 +       size = i_size_read(old_file->f_dentry->d_inode);
5344 +       ret = do_cow_splice(old_file, new_file, size);
5345 +       vxdprintk(VXD_CBIT(misc, 2), "do_splice_direct: %d", ret);
5346 +       if (ret < 0) {
5347 +               goto out_fput_both;
5348 +       } else if (ret < size) {
5349 +               ret = -ENOSPC;
5350 +               goto out_fput_both;
5351 +       } else {
5352 +               struct inode *old_inode = old_path.dentry->d_inode;
5353 +               struct inode *new_inode = new_path.dentry->d_inode;
5354 +               struct iattr attr = {
5355 +                       .ia_uid = old_inode->i_uid,
5356 +                       .ia_gid = old_inode->i_gid,
5357 +                       .ia_valid = ATTR_UID | ATTR_GID
5358 +                       };
5359 +
5360 +               setattr_copy(new_inode, &attr);
5361 +               mark_inode_dirty(new_inode);
5362 +       }
5363 +
5364 +       mutex_lock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5365 +
5366 +       /* drop out late */
5367 +       ret = -ENOENT;
5368 +       if ((redo = d_unhashed(old_path.dentry)))
5369 +               goto out_unlock;
5370 +
5371 +       vxdprintk(VXD_CBIT(misc, 2),
5372 +               "vfs_rename: [" VS_Q("%*s") ":%d] -> [" VS_Q("%*s") ":%d]",
5373 +               new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5374 +               new_path.dentry->d_name.len,
5375 +               old_path.dentry->d_name.len, old_path.dentry->d_name.name,
5376 +               old_path.dentry->d_name.len);
5377 +       ret = vfs_rename(dir_nd.path.dentry->d_inode, new_path.dentry,
5378 +               old_nd.path.dentry->d_parent->d_inode, old_path.dentry);
5379 +       vxdprintk(VXD_CBIT(misc, 2), "vfs_rename: %d", ret);
5380 +       res = new_path.dentry;
5381 +
5382 +out_unlock:
5383 +       mutex_unlock(&old_path.dentry->d_inode->i_sb->s_vfs_rename_mutex);
5384 +
5385 +out_fput_both:
5386 +       vxdprintk(VXD_CBIT(misc, 3),
5387 +               "fput(new_file=%p[#%ld])", new_file,
5388 +               atomic_long_read(&new_file->f_count));
5389 +       fput(new_file);
5390 +
5391 +out_fput_old:
5392 +       vxdprintk(VXD_CBIT(misc, 3),
5393 +               "fput(old_file=%p[#%ld])", old_file,
5394 +               atomic_long_read(&old_file->f_count));
5395 +       fput(old_file);
5396 +
5397 +out_unlock_new:
5398 +       mutex_unlock(&dir->d_inode->i_mutex);
5399 +       if (!ret)
5400 +               goto out_redo;
5401 +
5402 +       /* error path cleanup */
5403 +       vfs_unlink(dir->d_inode, new_path.dentry);
5404 +       dput(new_path.dentry);
5405 +
5406 +out_redo:
5407 +       if (!redo)
5408 +               goto out_rel_both;
5409 +       /* lookup dentry once again */
5410 +       path_put(&old_nd.path);
5411 +       ret = path_lookup(pathname, LOOKUP_FOLLOW, &old_nd);
5412 +       if (ret)
5413 +               goto out_rel_both;
5414 +
5415 +       new_path.dentry = old_nd.path.dentry;
5416 +       vxdprintk(VXD_CBIT(misc, 2),
5417 +               "path_lookup(redo): %p [" VS_Q("%.*s") ":%d]",
5418 +               new_path.dentry,
5419 +               new_path.dentry->d_name.len, new_path.dentry->d_name.name,
5420 +               new_path.dentry->d_name.len);
5421 +       dget(new_path.dentry);
5422 +       res = new_path.dentry;
5423 +
5424 +out_rel_both:
5425 +       path_put(&dir_nd.path);
5426 +out_rel_old:
5427 +       path_put(&old_nd.path);
5428 +out_free_path:
5429 +       kfree(path);
5430 +out:
5431 +       if (ret)
5432 +               res = ERR_PTR(ret);
5433 +       return res;
5434 +}
5435 +
5436 +#endif
5437 +
5438  /* get the link contents into pagecache */
5439  static char *page_getlink(struct dentry * dentry, struct page **ppage)
5440  {
5441 diff -NurpP --minimal linux-2.6.36/fs/namespace.c linux-2.6.36-vs2.3.0.36.38/fs/namespace.c
5442 --- linux-2.6.36/fs/namespace.c 2010-10-21 13:07:50.000000000 +0200
5443 +++ linux-2.6.36-vs2.3.0.36.38/fs/namespace.c   2010-10-21 15:11:02.000000000 +0200
5444 @@ -32,6 +32,11 @@
5445  #include <linux/idr.h>
5446  #include <linux/fs_struct.h>
5447  #include <linux/fsnotify.h>
5448 +#include <linux/vs_base.h>
5449 +#include <linux/vs_context.h>
5450 +#include <linux/vs_tag.h>
5451 +#include <linux/vserver/space.h>
5452 +#include <linux/vserver/global.h>
5453  #include <asm/uaccess.h>
5454  #include <asm/unistd.h>
5455  #include "pnode.h"
5456 @@ -601,6 +606,7 @@ static struct vfsmount *clone_mnt(struct
5457                 mnt->mnt_root = dget(root);
5458                 mnt->mnt_mountpoint = mnt->mnt_root;
5459                 mnt->mnt_parent = mnt;
5460 +               mnt->mnt_tag = old->mnt_tag;
5461
5462                 if (flag & CL_SLAVE) {
5463                         list_add(&mnt->mnt_slave, &old->mnt_slave_list);
5464 @@ -699,6 +705,31 @@ static inline void mangle(struct seq_fil
5465         seq_escape(m, s, " \t\n\\");
5466  }
5467
5468 +static int mnt_is_reachable(struct vfsmount *mnt)
5469 +{
5470 +       struct path root;
5471 +       struct dentry *point;
5472 +       int ret;
5473 +
5474 +       if (mnt == mnt->mnt_ns->root)
5475 +               return 1;
5476 +
5477 +       br_read_lock(vfsmount_lock);
5478 +       root = current->fs->root;
5479 +       point = root.dentry;
5480 +
5481 +       while ((mnt != mnt->mnt_parent) && (mnt != root.mnt)) {
5482 +               point = mnt->mnt_mountpoint;
5483 +               mnt = mnt->mnt_parent;
5484 +       }
5485 +
5486 +       ret = (mnt == root.mnt) && is_subdir(point, root.dentry);
5487 +
5488 +       br_read_unlock(vfsmount_lock);
5489 +
5490 +       return ret;
5491 +}
5492 +
5493  /*
5494   * Simple .show_options callback for filesystems which don't want to
5495   * implement more complex mount option showing.
5496 @@ -801,6 +832,8 @@ static int show_sb_opts(struct seq_file
5497                 { MS_SYNCHRONOUS, ",sync" },
5498                 { MS_DIRSYNC, ",dirsync" },
5499                 { MS_MANDLOCK, ",mand" },
5500 +               { MS_TAGGED, ",tag" },
5501 +               { MS_NOTAGCHECK, ",notagcheck" },
5502                 { 0, NULL }
5503         };
5504         const struct proc_fs_info *fs_infop;
5505 @@ -847,10 +880,20 @@ static int show_vfsmnt(struct seq_file *
5506         int err = 0;
5507         struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5508
5509 -       mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5510 -       seq_putc(m, ' ');
5511 -       seq_path(m, &mnt_path, " \t\n\\");
5512 -       seq_putc(m, ' ');
5513 +       if (vx_flags(VXF_HIDE_MOUNT, 0))
5514 +               return SEQ_SKIP;
5515 +       if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5516 +               return SEQ_SKIP;
5517 +
5518 +       if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5519 +               mnt == current->fs->root.mnt) {
5520 +               seq_puts(m, "/dev/root / ");
5521 +       } else {
5522 +               mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
5523 +               seq_putc(m, ' ');
5524 +               seq_path(m, &mnt_path, " \t\n\\");
5525 +               seq_putc(m, ' ');
5526 +       }
5527         show_type(m, mnt->mnt_sb);
5528         seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
5529         err = show_sb_opts(m, mnt->mnt_sb);
5530 @@ -880,6 +923,11 @@ static int show_mountinfo(struct seq_fil
5531         struct path root = p->root;
5532         int err = 0;
5533
5534 +       if (vx_flags(VXF_HIDE_MOUNT, 0))
5535 +               return SEQ_SKIP;
5536 +       if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5537 +               return SEQ_SKIP;
5538 +
5539         seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
5540                    MAJOR(sb->s_dev), MINOR(sb->s_dev));
5541         seq_dentry(m, mnt->mnt_root, " \t\n\\");
5542 @@ -938,17 +986,27 @@ static int show_vfsstat(struct seq_file
5543         struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
5544         int err = 0;
5545
5546 -       /* device */
5547 -       if (mnt->mnt_devname) {
5548 -               seq_puts(m, "device ");
5549 -               mangle(m, mnt->mnt_devname);
5550 -       } else
5551 -               seq_puts(m, "no device");
5552 +       if (vx_flags(VXF_HIDE_MOUNT, 0))
5553 +               return SEQ_SKIP;
5554 +       if (!mnt_is_reachable(mnt) && !vx_check(0, VS_WATCH_P))
5555 +               return SEQ_SKIP;
5556
5557 -       /* mount point */
5558 -       seq_puts(m, " mounted on ");
5559 -       seq_path(m, &mnt_path, " \t\n\\");
5560 -       seq_putc(m, ' ');
5561 +       if (!vx_check(0, VS_ADMIN|VS_WATCH) &&
5562 +               mnt == current->fs->root.mnt) {
5563 +               seq_puts(m, "device /dev/root mounted on / ");
5564 +       } else {
5565 +               /* device */
5566 +               if (mnt->mnt_devname) {
5567 +                       seq_puts(m, "device ");
5568 +                       mangle(m, mnt->mnt_devname);
5569 +               } else
5570 +                       seq_puts(m, "no device");
5571 +
5572 +               /* mount point */
5573 +               seq_puts(m, " mounted on ");
5574 +               seq_path(m, &mnt_path, " \t\n\\");
5575 +               seq_putc(m, ' ');
5576 +       }
5577
5578         /* file system type */
5579         seq_puts(m, "with fstype ");
5580 @@ -1194,7 +1252,7 @@ SYSCALL_DEFINE2(umount, char __user *, n
5581                 goto dput_and_out;
5582
5583         retval = -EPERM;
5584 -       if (!capable(CAP_SYS_ADMIN))
5585 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5586                 goto dput_and_out;
5587
5588         retval = do_umount(path.mnt, flags);
5589 @@ -1220,7 +1278,7 @@ SYSCALL_DEFINE1(oldumount, char __user *
5590
5591  static int mount_is_safe(struct path *path)
5592  {
5593 -       if (capable(CAP_SYS_ADMIN))
5594 +       if (vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5595                 return 0;
5596         return -EPERM;
5597  #ifdef notyet
5598 @@ -1510,7 +1568,7 @@ static int do_change_type(struct path *p
5599         int type;
5600         int err = 0;
5601
5602 -       if (!capable(CAP_SYS_ADMIN))
5603 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_NAMESPACE))
5604                 return -EPERM;
5605
5606         if (path->dentry != path->mnt->mnt_root)
5607 @@ -1541,11 +1599,13 @@ static int do_change_type(struct path *p
5608   * do loopback mount.
5609   */
5610  static int do_loopback(struct path *path, char *old_name,
5611 -                               int recurse)
5612 +       tag_t tag, unsigned long flags, int mnt_flags)
5613  {
5614         struct path old_path;
5615         struct vfsmount *mnt = NULL;
5616         int err = mount_is_safe(path);
5617 +       int recurse = flags & MS_REC;
5618 +
5619         if (err)
5620                 return err;
5621         if (!old_name || !*old_name)
5622 @@ -1580,6 +1640,7 @@ static int do_loopback(struct path *path
5623                 br_write_unlock(vfsmount_lock);
5624                 release_mounts(&umount_list);
5625         }
5626 +       mnt->mnt_flags = mnt_flags;
5627
5628  out:
5629         up_write(&namespace_sem);
5630 @@ -1610,12 +1671,12 @@ static int change_mount_flags(struct vfs
5631   * on it - tough luck.
5632   */
5633  static int do_remount(struct path *path, int flags, int mnt_flags,
5634 -                     void *data)
5635 +       void *data, xid_t xid)
5636  {
5637         int err;
5638         struct super_block *sb = path->mnt->mnt_sb;
5639
5640 -       if (!capable(CAP_SYS_ADMIN))
5641 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_REMOUNT))
5642                 return -EPERM;
5643
5644         if (!check_mnt(path->mnt))
5645 @@ -1659,7 +1720,7 @@ static int do_move_mount(struct path *pa
5646         struct path old_path, parent_path;
5647         struct vfsmount *p;
5648         int err = 0;
5649 -       if (!capable(CAP_SYS_ADMIN))
5650 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5651                 return -EPERM;
5652         if (!old_name || !*old_name)
5653                 return -EINVAL;
5654 @@ -1741,7 +1802,7 @@ static int do_new_mount(struct path *pat
5655                 return -EINVAL;
5656
5657         /* we need capabilities... */
5658 -       if (!capable(CAP_SYS_ADMIN))
5659 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SECURE_MOUNT))
5660                 return -EPERM;
5661
5662         lock_kernel();
5663 @@ -2009,6 +2070,7 @@ long do_mount(char *dev_name, char *dir_
5664         struct path path;
5665         int retval = 0;
5666         int mnt_flags = 0;
5667 +       tag_t tag = 0;
5668
5669         /* Discard magic */
5670         if ((flags & MS_MGC_MSK) == MS_MGC_VAL)
5671 @@ -2036,6 +2098,12 @@ long do_mount(char *dev_name, char *dir_
5672         if (!(flags & MS_NOATIME))
5673                 mnt_flags |= MNT_RELATIME;
5674
5675 +       if (dx_parse_tag(data_page, &tag, 1, &mnt_flags, &flags)) {
5676 +               /* FIXME: bind and re-mounts get the tag flag? */
5677 +               if (flags & (MS_BIND|MS_REMOUNT))
5678 +                       flags |= MS_TAGID;
5679 +       }
5680 +
5681         /* Separate the per-mountpoint flags */
5682         if (flags & MS_NOSUID)
5683                 mnt_flags |= MNT_NOSUID;
5684 @@ -2052,15 +2120,17 @@ long do_mount(char *dev_name, char *dir_
5685         if (flags & MS_RDONLY)
5686                 mnt_flags |= MNT_READONLY;
5687
5688 +       if (!capable(CAP_SYS_ADMIN))
5689 +               mnt_flags |= MNT_NODEV;
5690         flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE | MS_BORN |
5691                    MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT |
5692                    MS_STRICTATIME);
5693
5694         if (flags & MS_REMOUNT)
5695                 retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
5696 -                                   data_page);
5697 +                                   data_page, tag);
5698         else if (flags & MS_BIND)
5699 -               retval = do_loopback(&path, dev_name, flags & MS_REC);
5700 +               retval = do_loopback(&path, dev_name, tag, flags, mnt_flags);
5701         else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
5702                 retval = do_change_type(&path, flags);
5703         else if (flags & MS_MOVE)
5704 @@ -2139,6 +2209,7 @@ static struct mnt_namespace *dup_mnt_ns(
5705                 q = next_mnt(q, new_ns->root);
5706         }
5707         up_write(&namespace_sem);
5708 +       atomic_inc(&vs_global_mnt_ns);
5709
5710         if (rootmnt)
5711                 mntput(rootmnt);
5712 @@ -2280,9 +2351,10 @@ SYSCALL_DEFINE2(pivot_root, const char _
5713         down_write(&namespace_sem);
5714         mutex_lock(&old.dentry->d_inode->i_mutex);
5715         error = -EINVAL;
5716 -       if (IS_MNT_SHARED(old.mnt) ||
5717 +       if ((IS_MNT_SHARED(old.mnt) ||
5718                 IS_MNT_SHARED(new.mnt->mnt_parent) ||
5719 -               IS_MNT_SHARED(root.mnt->mnt_parent))
5720 +               IS_MNT_SHARED(root.mnt->mnt_parent)) &&
5721 +               !vx_flags(VXF_STATE_SETUP, 0))
5722                 goto out2;
5723         if (!check_mnt(root.mnt))
5724                 goto out2;
5725 @@ -2415,6 +2487,7 @@ void put_mnt_ns(struct mnt_namespace *ns
5726         br_write_unlock(vfsmount_lock);
5727         up_write(&namespace_sem);
5728         release_mounts(&umount_list);
5729 +       atomic_dec(&vs_global_mnt_ns);
5730         kfree(ns);
5731  }
5732  EXPORT_SYMBOL(put_mnt_ns);
5733 diff -NurpP --minimal linux-2.6.36/fs/nfs/client.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/client.c
5734 --- linux-2.6.36/fs/nfs/client.c        2010-10-21 13:07:50.000000000 +0200
5735 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/client.c  2010-10-21 13:09:36.000000000 +0200
5736 @@ -740,6 +740,9 @@ static int nfs_init_server_rpcclient(str
5737         if (server->flags & NFS_MOUNT_SOFT)
5738                 server->client->cl_softrtry = 1;
5739
5740 +       server->client->cl_tag = 0;
5741 +       if (server->flags & NFS_MOUNT_TAGGED)
5742 +               server->client->cl_tag = 1;
5743         return 0;
5744  }
5745
5746 @@ -911,6 +914,10 @@ static void nfs_server_set_fsinfo(struct
5747                 server->acdirmin = server->acdirmax = 0;
5748         }
5749
5750 +       /* FIXME: needs fsinfo
5751 +       if (server->flags & NFS_MOUNT_TAGGED)
5752 +               sb->s_flags |= MS_TAGGED;       */
5753 +
5754         server->maxfilesize = fsinfo->maxfilesize;
5755
5756         /* We're airborne Set socket buffersize */
5757 diff -NurpP --minimal linux-2.6.36/fs/nfs/dir.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/dir.c
5758 --- linux-2.6.36/fs/nfs/dir.c   2010-10-21 13:07:50.000000000 +0200
5759 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/dir.c     2010-10-21 13:09:36.000000000 +0200
5760 @@ -33,6 +33,7 @@
5761  #include <linux/namei.h>
5762  #include <linux/mount.h>
5763  #include <linux/sched.h>
5764 +#include <linux/vs_tag.h>
5765
5766  #include "nfs4_fs.h"
5767  #include "delegation.h"
5768 @@ -986,6 +987,7 @@ static struct dentry *nfs_lookup(struct
5769         if (IS_ERR(res))
5770                 goto out_unblock_sillyrename;
5771
5772 +       dx_propagate_tag(nd, inode);
5773  no_entry:
5774         res = d_materialise_unique(dentry, inode);
5775         if (res != NULL) {
5776 diff -NurpP --minimal linux-2.6.36/fs/nfs/inode.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/inode.c
5777 --- linux-2.6.36/fs/nfs/inode.c 2010-10-21 13:07:50.000000000 +0200
5778 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/inode.c   2010-10-21 13:09:36.000000000 +0200
5779 @@ -37,6 +37,7 @@
5780  #include <linux/inet.h>
5781  #include <linux/nfs_xdr.h>
5782  #include <linux/slab.h>
5783 +#include <linux/vs_tag.h>
5784
5785  #include <asm/system.h>
5786  #include <asm/uaccess.h>
5787 @@ -267,6 +268,8 @@ nfs_fhget(struct super_block *sb, struct
5788         if (inode->i_state & I_NEW) {
5789                 struct nfs_inode *nfsi = NFS_I(inode);
5790                 unsigned long now = jiffies;
5791 +               uid_t uid;
5792 +               gid_t gid;
5793
5794                 /* We set i_ino for the few things that still rely on it,
5795                  * such as stat(2) */
5796 @@ -315,8 +318,8 @@ nfs_fhget(struct super_block *sb, struct
5797                 nfsi->change_attr = 0;
5798                 inode->i_size = 0;
5799                 inode->i_nlink = 0;
5800 -               inode->i_uid = -2;
5801 -               inode->i_gid = -2;
5802 +               uid = -2;
5803 +               gid = -2;
5804                 inode->i_blocks = 0;
5805                 memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
5806
5807 @@ -353,13 +356,13 @@ nfs_fhget(struct super_block *sb, struct
5808                 else if (nfs_server_capable(inode, NFS_CAP_NLINK))
5809                         nfsi->cache_validity |= NFS_INO_INVALID_ATTR;
5810                 if (fattr->valid & NFS_ATTR_FATTR_OWNER)
5811 -                       inode->i_uid = fattr->uid;
5812 +                       uid = fattr->uid;
5813                 else if (nfs_server_capable(inode, NFS_CAP_OWNER))
5814                         nfsi->cache_validity |= NFS_INO_INVALID_ATTR
5815                                 | NFS_INO_INVALID_ACCESS
5816                                 | NFS_INO_INVALID_ACL;
5817                 if (fattr->valid & NFS_ATTR_FATTR_GROUP)
5818 -                       inode->i_gid = fattr->gid;
5819 +                       gid = fattr->gid;
5820                 else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP))
5821                         nfsi->cache_validity |= NFS_INO_INVALID_ATTR
5822                                 | NFS_INO_INVALID_ACCESS
5823 @@ -372,6 +375,11 @@ nfs_fhget(struct super_block *sb, struct
5824                          */
5825                         inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
5826                 }
5827 +               inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
5828 +               inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
5829 +               inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
5830 +                               /* maybe fattr->xid someday */
5831 +
5832                 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
5833                 nfsi->attrtimeo_timestamp = now;
5834                 nfsi->access_cache = RB_ROOT;
5835 @@ -488,6 +496,8 @@ void nfs_setattr_update_inode(struct ino
5836                         inode->i_uid = attr->ia_uid;
5837                 if ((attr->ia_valid & ATTR_GID) != 0)
5838                         inode->i_gid = attr->ia_gid;
5839 +               if ((attr->ia_valid & ATTR_TAG) && IS_TAGGED(inode))
5840 +                       inode->i_tag = attr->ia_tag;
5841                 NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5842                 spin_unlock(&inode->i_lock);
5843         }
5844 @@ -923,6 +933,9 @@ static int nfs_check_inode_attributes(st
5845         struct nfs_inode *nfsi = NFS_I(inode);
5846         loff_t cur_size, new_isize;
5847         unsigned long invalid = 0;
5848 +       uid_t uid;
5849 +       gid_t gid;
5850 +       tag_t tag;
5851
5852
5853         /* Has the inode gone and changed behind our back? */
5854 @@ -946,13 +959,18 @@ static int nfs_check_inode_attributes(st
5855                         invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
5856         }
5857
5858 +       uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5859 +       gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5860 +       tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5861 +
5862         /* Have any file permissions changed? */
5863         if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
5864                 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5865 -       if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && inode->i_uid != fattr->uid)
5866 +       if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && uid != fattr->uid)
5867                 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5868 -       if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && inode->i_gid != fattr->gid)
5869 +       if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && gid != fattr->gid)
5870                 invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
5871 +               /* maybe check for tag too? */
5872
5873         /* Has the link count changed? */
5874         if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink)
5875 @@ -1187,6 +1205,9 @@ static int nfs_update_inode(struct inode
5876         unsigned long invalid = 0;
5877         unsigned long now = jiffies;
5878         unsigned long save_cache_validity;
5879 +       uid_t uid;
5880 +       gid_t gid;
5881 +       tag_t tag;
5882
5883         dfprintk(VFS, "NFS: %s(%s/%ld ct=%d info=0x%x)\n",
5884                         __func__, inode->i_sb->s_id, inode->i_ino,
5885 @@ -1289,6 +1310,9 @@ static int nfs_update_inode(struct inode
5886                                 | NFS_INO_REVAL_PAGECACHE
5887                                 | NFS_INO_REVAL_FORCED);
5888
5889 +       uid = INOTAG_UID(DX_TAG(inode), fattr->uid, fattr->gid);
5890 +       gid = INOTAG_GID(DX_TAG(inode), fattr->uid, fattr->gid);
5891 +       tag = INOTAG_TAG(DX_TAG(inode), fattr->uid, fattr->gid, 0);
5892
5893         if (fattr->valid & NFS_ATTR_FATTR_ATIME)
5894                 memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
5895 @@ -1310,9 +1334,9 @@ static int nfs_update_inode(struct inode
5896                                 | NFS_INO_REVAL_FORCED);
5897
5898         if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
5899 -               if (inode->i_uid != fattr->uid) {
5900 +               if (uid != fattr->uid) {
5901                         invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5902 -                       inode->i_uid = fattr->uid;
5903 +                       uid = fattr->uid;
5904                 }
5905         } else if (server->caps & NFS_CAP_OWNER)
5906                 invalid |= save_cache_validity & (NFS_INO_INVALID_ATTR
5907 @@ -1321,9 +1345,9 @@ static int nfs_update_inode(struct inode
5908                                 | NFS_INO_REVAL_FORCED);
5909
5910         if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
5911 -               if (inode->i_gid != fattr->gid) {
5912 +               if (gid != fattr->gid) {
5913                         invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
5914 -                       inode->i_gid = fattr->gid;
5915 +                       gid = fattr->gid;
5916                 }
5917         } else if (server->caps & NFS_CAP_OWNER_GROUP)
5918                 invalid |= save_cache_validity & (NFS_INO_INVALID_ATTR
5919 @@ -1331,6 +1355,10 @@ static int nfs_update_inode(struct inode
5920                                 | NFS_INO_INVALID_ACL
5921                                 | NFS_INO_REVAL_FORCED);
5922
5923 +       inode->i_uid = uid;
5924 +       inode->i_gid = gid;
5925 +       inode->i_tag = tag;
5926 +
5927         if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
5928                 if (inode->i_nlink != fattr->nlink) {
5929                         invalid |= NFS_INO_INVALID_ATTR;
5930 diff -NurpP --minimal linux-2.6.36/fs/nfs/nfs3xdr.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/nfs3xdr.c
5931 --- linux-2.6.36/fs/nfs/nfs3xdr.c       2010-10-21 13:07:50.000000000 +0200
5932 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/nfs3xdr.c 2010-10-21 13:09:36.000000000 +0200
5933 @@ -20,6 +20,7 @@
5934  #include <linux/nfs3.h>
5935  #include <linux/nfs_fs.h>
5936  #include <linux/nfsacl.h>
5937 +#include <linux/vs_tag.h>
5938  #include "internal.h"
5939
5940  #define NFSDBG_FACILITY                NFSDBG_XDR
5941 @@ -175,7 +176,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_f
5942  }
5943
5944  static inline __be32 *
5945 -xdr_encode_sattr(__be32 *p, struct iattr *attr)
5946 +xdr_encode_sattr(__be32 *p, struct iattr *attr, int tag)
5947  {
5948         if (attr->ia_valid & ATTR_MODE) {
5949                 *p++ = xdr_one;
5950 @@ -183,15 +184,17 @@ xdr_encode_sattr(__be32 *p, struct iattr
5951         } else {
5952                 *p++ = xdr_zero;
5953         }
5954 -       if (attr->ia_valid & ATTR_UID) {
5955 +       if (attr->ia_valid & ATTR_UID ||
5956 +               (tag && (attr->ia_valid & ATTR_TAG))) {
5957                 *p++ = xdr_one;
5958 -               *p++ = htonl(attr->ia_uid);
5959 +               *p++ = htonl(TAGINO_UID(tag, attr->ia_uid, attr->ia_tag));
5960         } else {
5961                 *p++ = xdr_zero;
5962         }
5963 -       if (attr->ia_valid & ATTR_GID) {
5964 +       if (attr->ia_valid & ATTR_GID ||
5965 +               (tag && (attr->ia_valid & ATTR_TAG))) {
5966                 *p++ = xdr_one;
5967 -               *p++ = htonl(attr->ia_gid);
5968 +               *p++ = htonl(TAGINO_GID(tag, attr->ia_gid, attr->ia_tag));
5969         } else {
5970                 *p++ = xdr_zero;
5971         }
5972 @@ -278,7 +281,8 @@ static int
5973  nfs3_xdr_sattrargs(struct rpc_rqst *req, __be32 *p, struct nfs3_sattrargs *args)
5974  {
5975         p = xdr_encode_fhandle(p, args->fh);
5976 -       p = xdr_encode_sattr(p, args->sattr);
5977 +       p = xdr_encode_sattr(p, args->sattr,
5978 +               req->rq_task->tk_client->cl_tag);
5979         *p++ = htonl(args->guard);
5980         if (args->guard)
5981                 p = xdr_encode_time3(p, &args->guardtime);
5982 @@ -383,7 +387,8 @@ nfs3_xdr_createargs(struct rpc_rqst *req
5983                 *p++ = args->verifier[0];
5984                 *p++ = args->verifier[1];
5985         } else
5986 -               p = xdr_encode_sattr(p, args->sattr);
5987 +               p = xdr_encode_sattr(p, args->sattr,
5988 +                       req->rq_task->tk_client->cl_tag);
5989
5990         req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
5991         return 0;
5992 @@ -397,7 +402,8 @@ nfs3_xdr_mkdirargs(struct rpc_rqst *req,
5993  {
5994         p = xdr_encode_fhandle(p, args->fh);
5995         p = xdr_encode_array(p, args->name, args->len);
5996 -       p = xdr_encode_sattr(p, args->sattr);
5997 +       p = xdr_encode_sattr(p, args->sattr,
5998 +               req->rq_task->tk_client->cl_tag);
5999         req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
6000         return 0;
6001  }
6002 @@ -410,7 +416,8 @@ nfs3_xdr_symlinkargs(struct rpc_rqst *re
6003  {
6004         p = xdr_encode_fhandle(p, args->fromfh);
6005         p = xdr_encode_array(p, args->fromname, args->fromlen);
6006 -       p = xdr_encode_sattr(p, args->sattr);
6007 +       p = xdr_encode_sattr(p, args->sattr,
6008 +               req->rq_task->tk_client->cl_tag);
6009         *p++ = htonl(args->pathlen);
6010         req->rq_slen = xdr_adjust_iovec(req->rq_svec, p);
6011
6012 @@ -428,7 +435,8 @@ nfs3_xdr_mknodargs(struct rpc_rqst *req,
6013         p = xdr_encode_fhandle(p, args->fh);
6014         p = xdr_encode_array(p, args->name, args->len);
6015         *p++ = htonl(args->type);
6016 -       p = xdr_encode_sattr(p, args->sattr);
6017 +       p = xdr_encode_sattr(p, args->sattr,
6018 +               req->rq_task->tk_client->cl_tag);
6019         if (args->type == NF3CHR || args->type == NF3BLK) {
6020                 *p++ = htonl(MAJOR(args->rdev));
6021                 *p++ = htonl(MINOR(args->rdev));
6022 diff -NurpP --minimal linux-2.6.36/fs/nfs/nfsroot.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/nfsroot.c
6023 --- linux-2.6.36/fs/nfs/nfsroot.c       2010-08-02 16:52:50.000000000 +0200
6024 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/nfsroot.c 2010-10-21 13:09:36.000000000 +0200
6025 @@ -122,12 +122,12 @@ static int mount_port __initdata = 0;             /
6026  enum {
6027         /* Options that take integer arguments */
6028         Opt_port, Opt_rsize, Opt_wsize, Opt_timeo, Opt_retrans, Opt_acregmin,
6029 -       Opt_acregmax, Opt_acdirmin, Opt_acdirmax,
6030 +       Opt_acregmax, Opt_acdirmin, Opt_acdirmax, Opt_tagid,
6031         /* Options that take no arguments */
6032         Opt_soft, Opt_hard, Opt_intr,
6033         Opt_nointr, Opt_posix, Opt_noposix, Opt_cto, Opt_nocto, Opt_ac,
6034         Opt_noac, Opt_lock, Opt_nolock, Opt_v2, Opt_v3, Opt_udp, Opt_tcp,
6035 -       Opt_acl, Opt_noacl,
6036 +       Opt_acl, Opt_noacl, Opt_tag, Opt_notag,
6037         /* Error token */
6038         Opt_err
6039  };
6040 @@ -164,6 +164,9 @@ static const match_table_t tokens __init
6041         {Opt_tcp, "tcp"},
6042         {Opt_acl, "acl"},
6043         {Opt_noacl, "noacl"},
6044 +       {Opt_tag, "tag"},
6045 +       {Opt_notag, "notag"},
6046 +       {Opt_tagid, "tagid=%u"},
6047         {Opt_err, NULL}
6048
6049  };
6050 @@ -275,6 +278,20 @@ static int __init root_nfs_parse(char *n
6051                         case Opt_noacl:
6052                                 nfs_data.flags |= NFS_MOUNT_NOACL;
6053                                 break;
6054 +#ifndef CONFIG_TAGGING_NONE
6055 +                       case Opt_tag:
6056 +                               nfs_data.flags |= NFS_MOUNT_TAGGED;
6057 +                               break;
6058 +                       case Opt_notag:
6059 +                               nfs_data.flags &= ~NFS_MOUNT_TAGGED;
6060 +                               break;
6061 +#endif
6062 +#ifdef CONFIG_PROPAGATE
6063 +                       case Opt_tagid:
6064 +                               /* use args[0] */
6065 +                               nfs_data.flags |= NFS_MOUNT_TAGGED;
6066 +                               break;
6067 +#endif
6068                         default:
6069                                 printk(KERN_WARNING "Root-NFS: unknown "
6070                                         "option: %s\n", p);
6071 diff -NurpP --minimal linux-2.6.36/fs/nfs/super.c linux-2.6.36-vs2.3.0.36.38/fs/nfs/super.c
6072 --- linux-2.6.36/fs/nfs/super.c 2010-10-21 13:07:50.000000000 +0200
6073 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfs/super.c   2010-10-21 13:09:36.000000000 +0200
6074 @@ -54,6 +54,7 @@
6075  #include <linux/nfs_xdr.h>
6076  #include <linux/magic.h>
6077  #include <linux/parser.h>
6078 +#include <linux/vs_tag.h>
6079
6080  #include <asm/system.h>
6081  #include <asm/uaccess.h>
6082 @@ -617,6 +618,7 @@ static void nfs_show_mount_options(struc
6083                 { NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
6084                 { NFS_MOUNT_UNSHARED, ",nosharecache", "" },
6085                 { NFS_MOUNT_NORESVPORT, ",noresvport", "" },
6086 +               { NFS_MOUNT_TAGGED, ",tag", "" },
6087                 { 0, NULL, NULL }
6088         };
6089         const struct proc_nfs_info *nfs_infop;
6090 diff -NurpP --minimal linux-2.6.36/fs/nfsd/auth.c linux-2.6.36-vs2.3.0.36.38/fs/nfsd/auth.c
6091 --- linux-2.6.36/fs/nfsd/auth.c 2010-02-25 11:52:05.000000000 +0100
6092 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfsd/auth.c   2010-10-21 13:09:36.000000000 +0200
6093 @@ -1,6 +1,7 @@
6094  /* Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de> */
6095
6096  #include <linux/sched.h>
6097 +#include <linux/vs_tag.h>
6098  #include "nfsd.h"
6099  #include "auth.h"
6100
6101 @@ -36,6 +37,9 @@ int nfsd_setuser(struct svc_rqst *rqstp,
6102
6103         new->fsuid = rqstp->rq_cred.cr_uid;
6104         new->fsgid = rqstp->rq_cred.cr_gid;
6105 +       /* FIXME: this desperately needs a tag :)
6106 +       new->xid = (xid_t)INOTAG_TAG(DX_TAG_NFSD, cred.cr_uid, cred.cr_gid, 0);
6107 +                       */
6108
6109         rqgi = rqstp->rq_cred.cr_group_info;
6110
6111 diff -NurpP --minimal linux-2.6.36/fs/nfsd/nfs3xdr.c linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfs3xdr.c
6112 --- linux-2.6.36/fs/nfsd/nfs3xdr.c      2010-02-25 11:52:05.000000000 +0100
6113 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfs3xdr.c        2010-10-21 13:09:36.000000000 +0200
6114 @@ -7,6 +7,7 @@
6115   */
6116
6117  #include <linux/namei.h>
6118 +#include <linux/vs_tag.h>
6119  #include "xdr3.h"
6120  #include "auth.h"
6121
6122 @@ -95,6 +96,8 @@ static __be32 *
6123  decode_sattr3(__be32 *p, struct iattr *iap)
6124  {
6125         u32     tmp;
6126 +       uid_t   uid = 0;
6127 +       gid_t   gid = 0;
6128
6129         iap->ia_valid = 0;
6130
6131 @@ -104,12 +107,15 @@ decode_sattr3(__be32 *p, struct iattr *i
6132         }
6133         if (*p++) {
6134                 iap->ia_valid |= ATTR_UID;
6135 -               iap->ia_uid = ntohl(*p++);
6136 +               uid = ntohl(*p++);
6137         }
6138         if (*p++) {
6139                 iap->ia_valid |= ATTR_GID;
6140 -               iap->ia_gid = ntohl(*p++);
6141 +               gid = ntohl(*p++);
6142         }
6143 +       iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
6144 +       iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
6145 +       iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
6146         if (*p++) {
6147                 u64     newsize;
6148
6149 @@ -165,8 +171,12 @@ encode_fattr3(struct svc_rqst *rqstp, __
6150         *p++ = htonl(nfs3_ftypes[(stat->mode & S_IFMT) >> 12]);
6151         *p++ = htonl((u32) stat->mode);
6152         *p++ = htonl((u32) stat->nlink);
6153 -       *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
6154 -       *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
6155 +       *p++ = htonl((u32) nfsd_ruid(rqstp,
6156 +               TAGINO_UID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
6157 +               stat->uid, stat->tag)));
6158 +       *p++ = htonl((u32) nfsd_rgid(rqstp,
6159 +               TAGINO_GID(0 /* FIXME: DX_TAG(dentry->d_inode) */,
6160 +               stat->gid, stat->tag)));
6161         if (S_ISLNK(stat->mode) && stat->size > NFS3_MAXPATHLEN) {
6162                 p = xdr_encode_hyper(p, (u64) NFS3_MAXPATHLEN);
6163         } else {
6164 diff -NurpP --minimal linux-2.6.36/fs/nfsd/nfs4xdr.c linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfs4xdr.c
6165 --- linux-2.6.36/fs/nfsd/nfs4xdr.c      2010-10-21 13:07:50.000000000 +0200
6166 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfs4xdr.c        2010-10-21 13:09:36.000000000 +0200
6167 @@ -47,6 +47,7 @@
6168  #include <linux/nfsd_idmap.h>
6169  #include <linux/nfs4_acl.h>
6170  #include <linux/sunrpc/svcauth_gss.h>
6171 +#include <linux/vs_tag.h>
6172
6173  #include "xdr4.h"
6174  #include "vfs.h"
6175 @@ -2057,14 +2058,18 @@ out_acl:
6176                 WRITE32(stat.nlink);
6177         }
6178         if (bmval1 & FATTR4_WORD1_OWNER) {
6179 -               status = nfsd4_encode_user(rqstp, stat.uid, &p, &buflen);
6180 +               status = nfsd4_encode_user(rqstp,
6181 +                       TAGINO_UID(DX_TAG(dentry->d_inode),
6182 +                       stat.uid, stat.tag), &p, &buflen);
6183                 if (status == nfserr_resource)
6184                         goto out_resource;
6185                 if (status)
6186                         goto out;
6187         }
6188         if (bmval1 & FATTR4_WORD1_OWNER_GROUP) {
6189 -               status = nfsd4_encode_group(rqstp, stat.gid, &p, &buflen);
6190 +               status = nfsd4_encode_group(rqstp,
6191 +                       TAGINO_GID(DX_TAG(dentry->d_inode),
6192 +                       stat.gid, stat.tag), &p, &buflen);
6193                 if (status == nfserr_resource)
6194                         goto out_resource;
6195                 if (status)
6196 diff -NurpP --minimal linux-2.6.36/fs/nfsd/nfsxdr.c linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfsxdr.c
6197 --- linux-2.6.36/fs/nfsd/nfsxdr.c       2010-02-25 11:52:05.000000000 +0100
6198 +++ linux-2.6.36-vs2.3.0.36.38/fs/nfsd/nfsxdr.c 2010-10-21 13:09:36.000000000 +0200
6199 @@ -6,6 +6,7 @@
6200
6201  #include "xdr.h"
6202  #include "auth.h"
6203 +#include <linux/vs_tag.h>
6204
6205  #define NFSDDBG_FACILITY               NFSDDBG_XDR
6206
6207 @@ -88,6 +89,8 @@ static __be32 *
6208  decode_sattr(__be32 *p, struct iattr *iap)
6209  {
6210         u32     tmp, tmp1;
6211 +       uid_t   uid = 0;
6212 +       gid_t   gid = 0;
6213
6214         iap->ia_valid = 0;
6215
6216 @@ -101,12 +104,15 @@ decode_sattr(__be32 *p, struct iattr *ia
6217         }
6218         if ((tmp = ntohl(*p++)) != (u32)-1) {
6219                 iap->ia_valid |= ATTR_UID;
6220 -               iap->ia_uid = tmp;
6221 +               uid = tmp;
6222         }
6223         if ((tmp = ntohl(*p++)) != (u32)-1) {
6224                 iap->ia_valid |= ATTR_GID;
6225 -               iap->ia_gid = tmp;
6226 +               gid = tmp;
6227         }
6228 +       iap->ia_uid = INOTAG_UID(DX_TAG_NFSD, uid, gid);
6229 +       iap->ia_gid = INOTAG_GID(DX_TAG_NFSD, uid, gid);
6230 +       iap->ia_tag = INOTAG_TAG(DX_TAG_NFSD, uid, gid, 0);
6231         if ((tmp = ntohl(*p++)) != (u32)-1) {
6232                 iap->ia_valid |= ATTR_SIZE;
6233                 iap->ia_size = tmp;
6234 @@ -151,8 +157,10 @@ encode_fattr(struct svc_rqst *rqstp, __b
6235         *p++ = htonl(nfs_ftypes[type >> 12]);
6236         *p++ = htonl((u32) stat->mode);
6237         *p++ = htonl((u32) stat->nlink);
6238 -       *p++ = htonl((u32) nfsd_ruid(rqstp, stat->uid));
6239 -       *p++ = htonl((u32) nfsd_rgid(rqstp, stat->gid));
6240 +       *p++ = htonl((u32) nfsd_ruid(rqstp,
6241 +               TAGINO_UID(DX_TAG(dentry->d_inode), stat->uid, stat->tag)));
6242 +       *p++ = htonl((u32) nfsd_rgid(rqstp,
6243 +               TAGINO_GID(DX_TAG(dentry->d_inode), stat->gid, stat->tag)));
6244
6245         if (S_ISLNK(type) && stat->size > NFS_MAXPATHLEN) {
6246                 *p++ = htonl(NFS_MAXPATHLEN);
6247 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/dlmglue.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/dlmglue.c
6248 --- linux-2.6.36/fs/ocfs2/dlmglue.c     2010-10-21 13:07:50.000000000 +0200
6249 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/dlmglue.c       2010-10-21 13:09:36.000000000 +0200
6250 @@ -2114,6 +2114,7 @@ static void __ocfs2_stuff_meta_lvb(struc
6251         lvb->lvb_iclusters = cpu_to_be32(oi->ip_clusters);
6252         lvb->lvb_iuid      = cpu_to_be32(inode->i_uid);
6253         lvb->lvb_igid      = cpu_to_be32(inode->i_gid);
6254 +       lvb->lvb_itag      = cpu_to_be16(inode->i_tag);
6255         lvb->lvb_imode     = cpu_to_be16(inode->i_mode);
6256         lvb->lvb_inlink    = cpu_to_be16(inode->i_nlink);
6257         lvb->lvb_iatime_packed  =
6258 @@ -2168,6 +2169,7 @@ static void ocfs2_refresh_inode_from_lvb
6259
6260         inode->i_uid     = be32_to_cpu(lvb->lvb_iuid);
6261         inode->i_gid     = be32_to_cpu(lvb->lvb_igid);
6262 +       inode->i_tag     = be16_to_cpu(lvb->lvb_itag);
6263         inode->i_mode    = be16_to_cpu(lvb->lvb_imode);
6264         inode->i_nlink   = be16_to_cpu(lvb->lvb_inlink);
6265         ocfs2_unpack_timespec(&inode->i_atime,
6266 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/dlmglue.h linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/dlmglue.h
6267 --- linux-2.6.36/fs/ocfs2/dlmglue.h     2010-10-21 13:07:50.000000000 +0200
6268 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/dlmglue.h       2010-10-21 13:09:36.000000000 +0200
6269 @@ -46,7 +46,8 @@ struct ocfs2_meta_lvb {
6270         __be16       lvb_inlink;
6271         __be32       lvb_iattr;
6272         __be32       lvb_igeneration;
6273 -       __be32       lvb_reserved2;
6274 +       __be16       lvb_itag;
6275 +       __be16       lvb_reserved2;
6276  };
6277
6278  #define OCFS2_QINFO_LVB_VERSION 1
6279 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/file.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/file.c
6280 --- linux-2.6.36/fs/ocfs2/file.c        2010-10-21 13:07:50.000000000 +0200
6281 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/file.c  2010-10-21 13:09:36.000000000 +0200
6282 @@ -1138,13 +1138,15 @@ int ocfs2_setattr(struct dentry *dentry,
6283                 mlog(0, "uid change: %d\n", attr->ia_uid);
6284         if (attr->ia_valid & ATTR_GID)
6285                 mlog(0, "gid change: %d\n", attr->ia_gid);
6286 +       if (attr->ia_valid & ATTR_TAG)
6287 +               mlog(0, "tag change: %d\n", attr->ia_tag);
6288         if (attr->ia_valid & ATTR_SIZE)
6289                 mlog(0, "size change...\n");
6290         if (attr->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME))
6291                 mlog(0, "time change...\n");
6292
6293  #define OCFS2_VALID_ATTRS (ATTR_ATIME | ATTR_MTIME | ATTR_CTIME | ATTR_SIZE \
6294 -                          | ATTR_GID | ATTR_UID | ATTR_MODE)
6295 +                          | ATTR_GID | ATTR_UID | ATTR_TAG | ATTR_MODE)
6296         if (!(attr->ia_valid & OCFS2_VALID_ATTRS)) {
6297                 mlog(0, "can't handle attrs: 0x%x\n", attr->ia_valid);
6298                 return 0;
6299 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/inode.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/inode.c
6300 --- linux-2.6.36/fs/ocfs2/inode.c       2010-10-21 13:07:50.000000000 +0200
6301 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/inode.c 2010-10-21 13:09:36.000000000 +0200
6302 @@ -28,6 +28,7 @@
6303  #include <linux/highmem.h>
6304  #include <linux/pagemap.h>
6305  #include <linux/quotaops.h>
6306 +#include <linux/vs_tag.h>
6307
6308  #include <asm/byteorder.h>
6309
6310 @@ -78,11 +79,13 @@ void ocfs2_set_inode_flags(struct inode
6311  {
6312         unsigned int flags = OCFS2_I(inode)->ip_attr;
6313
6314 -       inode->i_flags &= ~(S_IMMUTABLE |
6315 +       inode->i_flags &= ~(S_IMMUTABLE | S_IXUNLINK |
6316                 S_SYNC | S_APPEND | S_NOATIME | S_DIRSYNC);
6317
6318         if (flags & OCFS2_IMMUTABLE_FL)
6319                 inode->i_flags |= S_IMMUTABLE;
6320 +       if (flags & OCFS2_IXUNLINK_FL)
6321 +               inode->i_flags |= S_IXUNLINK;
6322
6323         if (flags & OCFS2_SYNC_FL)
6324                 inode->i_flags |= S_SYNC;
6325 @@ -92,25 +95,44 @@ void ocfs2_set_inode_flags(struct inode
6326                 inode->i_flags |= S_NOATIME;
6327         if (flags & OCFS2_DIRSYNC_FL)
6328                 inode->i_flags |= S_DIRSYNC;
6329 +
6330 +       inode->i_vflags &= ~(V_BARRIER | V_COW);
6331 +
6332 +       if (flags & OCFS2_BARRIER_FL)
6333 +               inode->i_vflags |= V_BARRIER;
6334 +       if (flags & OCFS2_COW_FL)
6335 +               inode->i_vflags |= V_COW;
6336  }
6337
6338  /* Propagate flags from i_flags to OCFS2_I(inode)->ip_attr */
6339  void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi)
6340  {
6341         unsigned int flags = oi->vfs_inode.i_flags;
6342 +       unsigned int vflags = oi->vfs_inode.i_vflags;
6343 +
6344 +       oi->ip_attr &= ~(OCFS2_SYNC_FL | OCFS2_APPEND_FL |
6345 +                       OCFS2_IMMUTABLE_FL | OCFS2_IXUNLINK_FL |
6346 +                       OCFS2_NOATIME_FL | OCFS2_DIRSYNC_FL |
6347 +                       OCFS2_BARRIER_FL | OCFS2_COW_FL);
6348 +
6349 +       if (flags & S_IMMUTABLE)
6350 +               oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6351 +       if (flags & S_IXUNLINK)
6352 +               oi->ip_attr |= OCFS2_IXUNLINK_FL;
6353
6354 -       oi->ip_attr &= ~(OCFS2_SYNC_FL|OCFS2_APPEND_FL|
6355 -                       OCFS2_IMMUTABLE_FL|OCFS2_NOATIME_FL|OCFS2_DIRSYNC_FL);
6356         if (flags & S_SYNC)
6357                 oi->ip_attr |= OCFS2_SYNC_FL;
6358         if (flags & S_APPEND)
6359                 oi->ip_attr |= OCFS2_APPEND_FL;
6360 -       if (flags & S_IMMUTABLE)
6361 -               oi->ip_attr |= OCFS2_IMMUTABLE_FL;
6362         if (flags & S_NOATIME)
6363                 oi->ip_attr |= OCFS2_NOATIME_FL;
6364         if (flags & S_DIRSYNC)
6365                 oi->ip_attr |= OCFS2_DIRSYNC_FL;
6366 +
6367 +       if (vflags & V_BARRIER)
6368 +               oi->ip_attr |= OCFS2_BARRIER_FL;
6369 +       if (vflags & V_COW)
6370 +               oi->ip_attr |= OCFS2_COW_FL;
6371  }
6372
6373  struct inode *ocfs2_ilookup(struct super_block *sb, u64 blkno)
6374 @@ -245,6 +267,8 @@ void ocfs2_populate_inode(struct inode *
6375         struct super_block *sb;
6376         struct ocfs2_super *osb;
6377         int use_plocks = 1;
6378 +       uid_t uid;
6379 +       gid_t gid;
6380
6381         mlog_entry("(0x%p, size:%llu)\n", inode,
6382                    (unsigned long long)le64_to_cpu(fe->i_size));
6383 @@ -276,8 +300,12 @@ void ocfs2_populate_inode(struct inode *
6384         inode->i_generation = le32_to_cpu(fe->i_generation);
6385         inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
6386         inode->i_mode = le16_to_cpu(fe->i_mode);
6387 -       inode->i_uid = le32_to_cpu(fe->i_uid);
6388 -       inode->i_gid = le32_to_cpu(fe->i_gid);
6389 +       uid = le32_to_cpu(fe->i_uid);
6390 +       gid = le32_to_cpu(fe->i_gid);
6391 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
6392 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
6393 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid,
6394 +               /* le16_to_cpu(raw_inode->i_raw_tag)i */ 0);
6395
6396         /* Fast symlinks will have i_size but no allocated clusters. */
6397         if (S_ISLNK(inode->i_mode) && !fe->i_clusters)
6398 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/inode.h linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/inode.h
6399 --- linux-2.6.36/fs/ocfs2/inode.h       2010-10-21 13:07:50.000000000 +0200
6400 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/inode.h 2010-10-21 13:09:36.000000000 +0200
6401 @@ -153,6 +153,7 @@ struct buffer_head *ocfs2_bread(struct i
6402
6403  void ocfs2_set_inode_flags(struct inode *inode);
6404  void ocfs2_get_inode_flags(struct ocfs2_inode_info *oi);
6405 +int ocfs2_sync_flags(struct inode *inode, int, int);
6406
6407  static inline blkcnt_t ocfs2_inode_sector_count(struct inode *inode)
6408  {
6409 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ioctl.c
6410 --- linux-2.6.36/fs/ocfs2/ioctl.c       2010-02-25 11:52:06.000000000 +0100
6411 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ioctl.c 2010-10-21 13:09:36.000000000 +0200
6412 @@ -43,7 +43,41 @@ static int ocfs2_get_inode_attr(struct i
6413         return status;
6414  }
6415
6416 -static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6417 +int ocfs2_sync_flags(struct inode *inode, int flags, int vflags)
6418 +{
6419 +       struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
6420 +       struct buffer_head *bh = NULL;
6421 +       handle_t *handle = NULL;
6422 +       int status;
6423 +
6424 +       status = ocfs2_inode_lock(inode, &bh, 1);
6425 +       if (status < 0) {
6426 +               mlog_errno(status);
6427 +               return status;
6428 +       }
6429 +       handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6430 +       if (IS_ERR(handle)) {
6431 +               status = PTR_ERR(handle);
6432 +               mlog_errno(status);
6433 +               goto bail_unlock;
6434 +       }
6435 +
6436 +       inode->i_flags = flags;
6437 +       inode->i_vflags = vflags;
6438 +       ocfs2_get_inode_flags(OCFS2_I(inode));
6439 +
6440 +       status = ocfs2_mark_inode_dirty(handle, inode, bh);
6441 +       if (status < 0)
6442 +               mlog_errno(status);
6443 +
6444 +       ocfs2_commit_trans(osb, handle);
6445 +bail_unlock:
6446 +       ocfs2_inode_unlock(inode, 1);
6447 +       brelse(bh);
6448 +       return status;
6449 +}
6450 +
6451 +int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
6452                                 unsigned mask)
6453  {
6454         struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
6455 @@ -68,6 +102,11 @@ static int ocfs2_set_inode_attr(struct i
6456         if (!S_ISDIR(inode->i_mode))
6457                 flags &= ~OCFS2_DIRSYNC_FL;
6458
6459 +       if (IS_BARRIER(inode)) {
6460 +               vxwprintk_task(1, "messing with the barrier.");
6461 +               goto bail_unlock;
6462 +       }
6463 +
6464         handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
6465         if (IS_ERR(handle)) {
6466                 status = PTR_ERR(handle);
6467 @@ -109,6 +148,7 @@ bail:
6468         return status;
6469  }
6470
6471 +
6472  long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
6473  {
6474         struct inode *inode = filp->f_path.dentry->d_inode;
6475 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/namei.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/namei.c
6476 --- linux-2.6.36/fs/ocfs2/namei.c       2010-10-21 13:07:50.000000000 +0200
6477 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/namei.c 2010-10-21 13:09:36.000000000 +0200
6478 @@ -41,6 +41,7 @@
6479  #include <linux/slab.h>
6480  #include <linux/highmem.h>
6481  #include <linux/quotaops.h>
6482 +#include <linux/vs_tag.h>
6483
6484  #define MLOG_MASK_PREFIX ML_NAMEI
6485  #include <cluster/masklog.h>
6486 @@ -486,6 +487,7 @@ static int __ocfs2_mknod_locked(struct i
6487         struct ocfs2_dinode *fe = NULL;
6488         struct ocfs2_extent_list *fel;
6489         u16 feat;
6490 +       tag_t tag;
6491
6492         *new_fe_bh = NULL;
6493
6494 @@ -523,8 +525,11 @@ static int __ocfs2_mknod_locked(struct i
6495         fe->i_suballoc_loc = cpu_to_le64(suballoc_loc);
6496         fe->i_suballoc_bit = cpu_to_le16(suballoc_bit);
6497         fe->i_suballoc_slot = cpu_to_le16(inode_ac->ac_alloc_slot);
6498 -       fe->i_uid = cpu_to_le32(inode->i_uid);
6499 -       fe->i_gid = cpu_to_le32(inode->i_gid);
6500 +
6501 +       tag = dx_current_fstag(osb->sb);
6502 +       fe->i_uid = cpu_to_le32(TAGINO_UID(DX_TAG(inode), inode->i_uid, tag));
6503 +       fe->i_gid = cpu_to_le32(TAGINO_GID(DX_TAG(inode), inode->i_gid, tag));
6504 +       inode->i_tag = tag;
6505         fe->i_mode = cpu_to_le16(inode->i_mode);
6506         if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode))
6507                 fe->id1.dev1.i_rdev = cpu_to_le64(huge_encode_dev(dev));
6508 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/ocfs2.h linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ocfs2.h
6509 --- linux-2.6.36/fs/ocfs2/ocfs2.h       2010-08-02 16:52:51.000000000 +0200
6510 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ocfs2.h 2010-10-21 13:09:36.000000000 +0200
6511 @@ -256,6 +256,7 @@ enum ocfs2_mount_options
6512                                                    control lists */
6513         OCFS2_MOUNT_USRQUOTA = 1 << 10, /* We support user quotas */
6514         OCFS2_MOUNT_GRPQUOTA = 1 << 11, /* We support group quotas */
6515 +       OCFS2_MOUNT_TAGGED = 1 << 12, /* use tagging */
6516  };
6517
6518  #define OCFS2_OSB_SOFT_RO                      0x0001
6519 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/ocfs2_fs.h linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ocfs2_fs.h
6520 --- linux-2.6.36/fs/ocfs2/ocfs2_fs.h    2010-10-21 13:07:51.000000000 +0200
6521 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/ocfs2_fs.h      2010-10-21 14:00:41.000000000 +0200
6522 @@ -258,6 +258,11 @@
6523  #define OCFS2_TOPDIR_FL                        FS_TOPDIR_FL    /* Top of directory hierarchies*/
6524  #define OCFS2_RESERVED_FL              FS_RESERVED_FL  /* reserved for ext2 lib */
6525
6526 +#define OCFS2_IXUNLINK_FL              FS_IXUNLINK_FL  /* Immutable invert on unlink */
6527 +
6528 +#define OCFS2_BARRIER_FL               FS_BARRIER_FL   /* Barrier for chroot() */
6529 +#define OCFS2_COW_FL                   FS_COW_FL       /* Copy on Write marker */
6530 +
6531  #define OCFS2_FL_VISIBLE               FS_FL_USER_VISIBLE      /* User visible flags */
6532  #define OCFS2_FL_MODIFIABLE            FS_FL_USER_MODIFIABLE   /* User modifiable flags */
6533
6534 diff -NurpP --minimal linux-2.6.36/fs/ocfs2/super.c linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/super.c
6535 --- linux-2.6.36/fs/ocfs2/super.c       2010-10-21 13:07:51.000000000 +0200
6536 +++ linux-2.6.36-vs2.3.0.36.38/fs/ocfs2/super.c 2010-10-21 13:09:36.000000000 +0200
6537 @@ -179,6 +179,7 @@ enum {
6538         Opt_grpquota,
6539         Opt_resv_level,
6540         Opt_dir_resv_level,
6541 +       Opt_tag, Opt_notag, Opt_tagid,
6542         Opt_err,
6543  };
6544
6545 @@ -207,6 +208,9 @@ static const match_table_t tokens = {
6546         {Opt_grpquota, "grpquota"},
6547         {Opt_resv_level, "resv_level=%u"},
6548         {Opt_dir_resv_level, "dir_resv_level=%u"},
6549 +       {Opt_tag, "tag"},
6550 +       {Opt_notag, "notag"},
6551 +       {Opt_tagid, "tagid=%u"},
6552         {Opt_err, NULL}
6553  };
6554
6555 @@ -617,6 +621,13 @@ static int ocfs2_remount(struct super_bl
6556                 goto out;
6557         }
6558
6559 +       if ((osb->s_mount_opt & OCFS2_MOUNT_TAGGED) !=
6560 +           (parsed_options.mount_opt & OCFS2_MOUNT_TAGGED)) {
6561 +               ret = -EINVAL;
6562 +               mlog(ML_ERROR, "Cannot change tagging on remount\n");
6563 +               goto out;
6564 +       }
6565 +
6566         if ((osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) !=
6567             (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
6568                 ret = -EINVAL;
6569 @@ -1153,6 +1164,9 @@ static int ocfs2_fill_super(struct super
6570
6571         ocfs2_complete_mount_recovery(osb);
6572
6573 +       if (osb->s_mount_opt & OCFS2_MOUNT_TAGGED)
6574 +               sb->s_flags |= MS_TAGGED;
6575 +
6576         if (ocfs2_mount_local(osb))
6577                 snprintf(nodestr, sizeof(nodestr), "local");
6578         else
6579 @@ -1468,6 +1482,20 @@ static int ocfs2_parse_options(struct su
6580                             option < OCFS2_MAX_RESV_LEVEL)
6581                                 mopt->dir_resv_level = option;
6582                         break;
6583 +#ifndef CONFIG_TAGGING_NONE
6584 +               case Opt_tag:
6585 +                       mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6586 +                       break;
6587 +               case Opt_notag:
6588 +                       mopt->mount_opt &= ~OCFS2_MOUNT_TAGGED;
6589 +                       break;
6590 +#endif
6591 +#ifdef CONFIG_PROPAGATE
6592 +               case Opt_tagid:
6593 +                       /* use args[0] */
6594 +                       mopt->mount_opt |= OCFS2_MOUNT_TAGGED;
6595 +                       break;
6596 +#endif
6597                 default:
6598                         mlog(ML_ERROR,
6599                              "Unrecognized mount option \"%s\" "
6600 diff -NurpP --minimal linux-2.6.36/fs/open.c linux-2.6.36-vs2.3.0.36.38/fs/open.c
6601 --- linux-2.6.36/fs/open.c      2010-10-21 13:07:51.000000000 +0200
6602 +++ linux-2.6.36-vs2.3.0.36.38/fs/open.c        2010-10-21 13:55:41.000000000 +0200
6603 @@ -30,6 +30,11 @@
6604  #include <linux/fs_struct.h>
6605  #include <linux/ima.h>
6606  #include <linux/dnotify.h>
6607 +#include <linux/vs_base.h>
6608 +#include <linux/vs_limit.h>
6609 +#include <linux/vs_tag.h>
6610 +#include <linux/vs_cowbl.h>
6611 +#include <linux/vserver/dlimit.h>
6612
6613  #include "internal.h"
6614
6615 @@ -481,6 +486,12 @@ SYSCALL_DEFINE3(fchmodat, int, dfd, cons
6616         error = user_path_at(dfd, filename, LOOKUP_FOLLOW, &path);
6617         if (error)
6618                 goto out;
6619 +
6620 +#ifdef CONFIG_VSERVER_COWBL
6621 +       error = cow_check_and_break(&path);
6622 +       if (error)
6623 +               goto dput_and_out;
6624 +#endif
6625         inode = path.dentry->d_inode;
6626
6627         error = mnt_want_write(path.mnt);
6628 @@ -518,11 +529,11 @@ static int chown_common(struct path *pat
6629         newattrs.ia_valid =  ATTR_CTIME;
6630         if (user != (uid_t) -1) {
6631                 newattrs.ia_valid |= ATTR_UID;
6632 -               newattrs.ia_uid = user;
6633 +               newattrs.ia_uid = dx_map_uid(user);
6634         }
6635         if (group != (gid_t) -1) {
6636                 newattrs.ia_valid |= ATTR_GID;
6637 -               newattrs.ia_gid = group;
6638 +               newattrs.ia_gid = dx_map_gid(group);
6639         }
6640         if (!S_ISDIR(inode->i_mode))
6641                 newattrs.ia_valid |=
6642 @@ -547,6 +558,10 @@ SYSCALL_DEFINE3(chown, const char __user
6643         error = mnt_want_write(path.mnt);
6644         if (error)
6645                 goto out_release;
6646 +#ifdef CONFIG_VSERVER_COWBL
6647 +       error = cow_check_and_break(&path);
6648 +       if (!error)
6649 +#endif
6650         error = chown_common(&path, user, group);
6651         mnt_drop_write(path.mnt);
6652  out_release:
6653 @@ -572,6 +587,10 @@ SYSCALL_DEFINE5(fchownat, int, dfd, cons
6654         error = mnt_want_write(path.mnt);
6655         if (error)
6656                 goto out_release;
6657 +#ifdef CONFIG_VSERVER_COWBL
6658 +       error = cow_check_and_break(&path);
6659 +       if (!error)
6660 +#endif
6661         error = chown_common(&path, user, group);
6662         mnt_drop_write(path.mnt);
6663  out_release:
6664 @@ -591,6 +610,10 @@ SYSCALL_DEFINE3(lchown, const char __use
6665         error = mnt_want_write(path.mnt);
6666         if (error)
6667                 goto out_release;
6668 +#ifdef CONFIG_VSERVER_COWBL
6669 +       error = cow_check_and_break(&path);
6670 +       if (!error)
6671 +#endif
6672         error = chown_common(&path, user, group);
6673         mnt_drop_write(path.mnt);
6674  out_release:
6675 @@ -837,6 +860,7 @@ static void __put_unused_fd(struct files
6676         __FD_CLR(fd, fdt->open_fds);
6677         if (fd < files->next_fd)
6678                 files->next_fd = fd;
6679 +       vx_openfd_dec(fd);
6680  }
6681
6682  void put_unused_fd(unsigned int fd)
6683 diff -NurpP --minimal linux-2.6.36/fs/proc/array.c linux-2.6.36-vs2.3.0.36.38/fs/proc/array.c
6684 --- linux-2.6.36/fs/proc/array.c        2010-08-02 16:52:51.000000000 +0200
6685 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/array.c  2010-10-21 13:09:36.000000000 +0200
6686 @@ -81,6 +81,8 @@
6687  #include <linux/pid_namespace.h>
6688  #include <linux/ptrace.h>
6689  #include <linux/tracehook.h>
6690 +#include <linux/vs_context.h>
6691 +#include <linux/vs_network.h>
6692
6693  #include <asm/pgtable.h>
6694  #include <asm/processor.h>
6695 @@ -170,6 +172,9 @@ static inline void task_state(struct seq
6696         rcu_read_lock();
6697         ppid = pid_alive(p) ?
6698                 task_tgid_nr_ns(rcu_dereference(p->real_parent), ns) : 0;
6699 +       if (unlikely(vx_current_initpid(p->pid)))
6700 +               ppid = 0;
6701 +
6702         tpid = 0;
6703         if (pid_alive(p)) {
6704                 struct task_struct *tracer = tracehook_tracer_task(p);
6705 @@ -287,7 +292,7 @@ static inline void task_sig(struct seq_f
6706  }
6707
6708  static void render_cap_t(struct seq_file *m, const char *header,
6709 -                       kernel_cap_t *a)
6710 +                       struct vx_info *vxi, kernel_cap_t *a)
6711  {
6712         unsigned __capi;
6713
6714 @@ -312,10 +317,11 @@ static inline void task_cap(struct seq_f
6715         cap_bset        = cred->cap_bset;
6716         rcu_read_unlock();
6717
6718 -       render_cap_t(m, "CapInh:\t", &cap_inheritable);
6719 -       render_cap_t(m, "CapPrm:\t", &cap_permitted);
6720 -       render_cap_t(m, "CapEff:\t", &cap_effective);
6721 -       render_cap_t(m, "CapBnd:\t", &cap_bset);
6722 +       /* FIXME: maybe move the p->vx_info masking to __task_cred() ? */
6723 +       render_cap_t(m, "CapInh:\t", p->vx_info, &cap_inheritable);
6724 +       render_cap_t(m, "CapPrm:\t", p->vx_info, &cap_permitted);
6725 +       render_cap_t(m, "CapEff:\t", p->vx_info, &cap_effective);
6726 +       render_cap_t(m, "CapBnd:\t", p->vx_info, &cap_bset);
6727  }
6728
6729  static inline void task_context_switch_counts(struct seq_file *m,
6730 @@ -337,6 +343,42 @@ static void task_cpus_allowed(struct seq
6731         seq_printf(m, "\n");
6732  }
6733
6734 +int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
6735 +                       struct pid *pid, struct task_struct *task)
6736 +{
6737 +       seq_printf(m,   "Proxy:\t%p(%c)\n"
6738 +                       "Count:\t%u\n"
6739 +                       "uts:\t%p(%c)\n"
6740 +                       "ipc:\t%p(%c)\n"
6741 +                       "mnt:\t%p(%c)\n"
6742 +                       "pid:\t%p(%c)\n"
6743 +                       "net:\t%p(%c)\n",
6744 +                       task->nsproxy,
6745 +                       (task->nsproxy == init_task.nsproxy ? 'I' : '-'),
6746 +                       atomic_read(&task->nsproxy->count),
6747 +                       task->nsproxy->uts_ns,
6748 +                       (task->nsproxy->uts_ns == init_task.nsproxy->uts_ns ? 'I' : '-'),
6749 +                       task->nsproxy->ipc_ns,
6750 +                       (task->nsproxy->ipc_ns == init_task.nsproxy->ipc_ns ? 'I' : '-'),
6751 +                       task->nsproxy->mnt_ns,
6752 +                       (task->nsproxy->mnt_ns == init_task.nsproxy->mnt_ns ? 'I' : '-'),
6753 +                       task->nsproxy->pid_ns,
6754 +                       (task->nsproxy->pid_ns == init_task.nsproxy->pid_ns ? 'I' : '-'),
6755 +                       task->nsproxy->net_ns,
6756 +                       (task->nsproxy->net_ns == init_task.nsproxy->net_ns ? 'I' : '-'));
6757 +       return 0;
6758 +}
6759 +
6760 +void task_vs_id(struct seq_file *m, struct task_struct *task)
6761 +{
6762 +       if (task_vx_flags(task, VXF_HIDE_VINFO, 0))
6763 +               return;
6764 +
6765 +       seq_printf(m, "VxID: %d\n", vx_task_xid(task));
6766 +       seq_printf(m, "NxID: %d\n", nx_task_nid(task));
6767 +}
6768 +
6769 +
6770  int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
6771                         struct pid *pid, struct task_struct *task)
6772  {
6773 @@ -353,6 +395,7 @@ int proc_pid_status(struct seq_file *m,
6774         task_cap(m, task);
6775         task_cpus_allowed(m, task);
6776         cpuset_task_status_allowed(m, task);
6777 +       task_vs_id(m, task);
6778  #if defined(CONFIG_S390)
6779         task_show_regs(m, task);
6780  #endif
6781 @@ -465,6 +508,17 @@ static int do_task_stat(struct seq_file
6782         /* convert nsec -> ticks */
6783         start_time = nsec_to_clock_t(start_time);
6784
6785 +       /* fixup start time for virt uptime */
6786 +       if (vx_flags(VXF_VIRT_UPTIME, 0)) {
6787 +               unsigned long long bias =
6788 +                       current->vx_info->cvirt.bias_clock;
6789 +
6790 +               if (start_time > bias)
6791 +                       start_time -= bias;
6792 +               else
6793 +                       start_time = 0;
6794 +       }
6795 +
6796         seq_printf(m, "%d (%s) %c %d %d %d %d %d %u %lu \
6797  %lu %lu %lu %lu %lu %ld %ld %ld %ld %d 0 %llu %lu %ld %lu %lu %lu %lu %lu \
6798  %lu %lu %lu %lu %lu %lu %lu %lu %d %d %u %u %llu %lu %ld\n",
6799 diff -NurpP --minimal linux-2.6.36/fs/proc/base.c linux-2.6.36-vs2.3.0.36.38/fs/proc/base.c
6800 --- linux-2.6.36/fs/proc/base.c 2010-10-21 13:07:51.000000000 +0200
6801 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/base.c   2010-10-21 13:58:32.000000000 +0200
6802 @@ -83,6 +83,8 @@
6803  #include <linux/pid_namespace.h>
6804  #include <linux/fs_struct.h>
6805  #include <linux/slab.h>
6806 +#include <linux/vs_context.h>
6807 +#include <linux/vs_network.h>
6808  #include "internal.h"
6809
6810  /* NOTE:
6811 @@ -1041,12 +1043,17 @@ static ssize_t oom_adjust_write(struct f
6812                 return -ESRCH;
6813         }
6814
6815 -       if (oom_adjust < task->signal->oom_adj && !capable(CAP_SYS_RESOURCE)) {
6816 +       if (oom_adjust < task->signal->oom_adj &&
6817 +               !vx_capable(CAP_SYS_RESOURCE, VXC_OOM_ADJUST)) {
6818                 unlock_task_sighand(task, &flags);
6819                 put_task_struct(task);
6820                 return -EACCES;
6821         }
6822
6823 +       /* prevent guest processes from circumventing the oom killer */
6824 +       if (vx_current_xid() && (oom_adjust == OOM_DISABLE))
6825 +               oom_adjust = OOM_ADJUST_MIN;
6826 +
6827         /*
6828          * Warn that /proc/pid/oom_adj is deprecated, see
6829          * Documentation/feature-removal-schedule.txt.
6830 @@ -1179,7 +1186,7 @@ static ssize_t proc_loginuid_write(struc
6831         ssize_t length;
6832         uid_t loginuid;
6833
6834 -       if (!capable(CAP_AUDIT_CONTROL))
6835 +       if (!vx_capable(CAP_AUDIT_CONTROL, VXC_AUDIT_CONTROL))
6836                 return -EPERM;
6837
6838         rcu_read_lock();
6839 @@ -1617,6 +1624,8 @@ static struct inode *proc_pid_make_inode
6840                 inode->i_gid = cred->egid;
6841                 rcu_read_unlock();
6842         }
6843 +       /* procfs is xid tagged */
6844 +       inode->i_tag = (tag_t)vx_task_xid(task);
6845         security_task_to_inode(task, inode);
6846
6847  out:
6848 @@ -2167,6 +2176,13 @@ static struct dentry *proc_pident_lookup
6849         if (!task)
6850                 goto out_no_task;
6851
6852 +       /* TODO: maybe we can come up with a generic approach? */
6853 +       if (task_vx_flags(task, VXF_HIDE_VINFO, 0) &&
6854 +               (dentry->d_name.len == 5) &&
6855 +               (!memcmp(dentry->d_name.name, "vinfo", 5) ||
6856 +               !memcmp(dentry->d_name.name, "ninfo", 5)))
6857 +               goto out;
6858 +
6859         /*
6860          * Yes, it does not scale. And it should not. Don't add
6861          * new entries into /proc/<tgid>/ without very good reasons.
6862 @@ -2574,7 +2590,7 @@ out_iput:
6863  static struct dentry *proc_base_lookup(struct inode *dir, struct dentry *dentry)
6864  {
6865         struct dentry *error;
6866 -       struct task_struct *task = get_proc_task(dir);
6867 +       struct task_struct *task = get_proc_task_real(dir);
6868         const struct pid_entry *p, *last;
6869
6870         error = ERR_PTR(-ENOENT);
6871 @@ -2664,6 +2680,9 @@ static int proc_pid_personality(struct s
6872  static const struct file_operations proc_task_operations;
6873  static const struct inode_operations proc_task_inode_operations;
6874
6875 +extern int proc_pid_vx_info(struct task_struct *, char *);
6876 +extern int proc_pid_nx_info(struct task_struct *, char *);
6877 +
6878  static const struct pid_entry tgid_base_stuff[] = {
6879         DIR("task",       S_IRUGO|S_IXUGO, proc_task_inode_operations, proc_task_operations),
6880         DIR("fd",         S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations),
6881 @@ -2723,6 +2742,8 @@ static const struct pid_entry tgid_base_
6882  #ifdef CONFIG_CGROUPS
6883         REG("cgroup",  S_IRUGO, proc_cgroup_operations),
6884  #endif
6885 +       INF("vinfo",      S_IRUGO, proc_pid_vx_info),
6886 +       INF("ninfo",      S_IRUGO, proc_pid_nx_info),
6887         INF("oom_score",  S_IRUGO, proc_oom_score),
6888         REG("oom_adj",    S_IRUGO|S_IWUSR, proc_oom_adjust_operations),
6889         REG("oom_score_adj", S_IRUGO|S_IWUSR, proc_oom_score_adj_operations),
6890 @@ -2739,6 +2760,7 @@ static const struct pid_entry tgid_base_
6891  #ifdef CONFIG_TASK_IO_ACCOUNTING
6892         INF("io",       S_IRUGO, proc_tgid_io_accounting),
6893  #endif
6894 +       ONE("nsproxy",  S_IRUGO, proc_pid_nsproxy),
6895  };
6896
6897  static int proc_tgid_base_readdir(struct file * filp,
6898 @@ -2930,7 +2952,7 @@ retry:
6899         iter.task = NULL;
6900         pid = find_ge_pid(iter.tgid, ns);
6901         if (pid) {
6902 -               iter.tgid = pid_nr_ns(pid, ns);
6903 +               iter.tgid = pid_unmapped_nr_ns(pid, ns);
6904                 iter.task = pid_task(pid, PIDTYPE_PID);
6905                 /* What we to know is if the pid we have find is the
6906                  * pid of a thread_group_leader.  Testing for task
6907 @@ -2960,7 +2982,7 @@ static int proc_pid_fill_cache(struct fi
6908         struct tgid_iter iter)
6909  {
6910         char name[PROC_NUMBUF];
6911 -       int len = snprintf(name, sizeof(name), "%d", iter.tgid);
6912 +       int len = snprintf(name, sizeof(name), "%d", vx_map_tgid(iter.tgid));
6913         return proc_fill_cache(filp, dirent, filldir, name, len,
6914                                 proc_pid_instantiate, iter.task, NULL);
6915  }
6916 @@ -2969,7 +2991,7 @@ static int proc_pid_fill_cache(struct fi
6917  int proc_pid_readdir(struct file * filp, void * dirent, filldir_t filldir)
6918  {
6919         unsigned int nr = filp->f_pos - FIRST_PROCESS_ENTRY;
6920 -       struct task_struct *reaper = get_proc_task(filp->f_path.dentry->d_inode);
6921 +       struct task_struct *reaper = get_proc_task_real(filp->f_path.dentry->d_inode);
6922         struct tgid_iter iter;
6923         struct pid_namespace *ns;
6924
6925 @@ -2989,6 +3011,8 @@ int proc_pid_readdir(struct file * filp,
6926              iter.task;
6927              iter.tgid += 1, iter = next_tgid(ns, iter)) {
6928                 filp->f_pos = iter.tgid + TGID_OFFSET;
6929 +               if (!vx_proc_task_visible(iter.task))
6930 +                       continue;
6931                 if (proc_pid_fill_cache(filp, dirent, filldir, iter) < 0) {
6932                         put_task_struct(iter.task);
6933                         goto out;
6934 @@ -3137,6 +3161,8 @@ static struct dentry *proc_task_lookup(s
6935         tid = name_to_int(dentry);
6936         if (tid == ~0U)
6937                 goto out;
6938 +       if (vx_current_initpid(tid))
6939 +               goto out;
6940
6941         ns = dentry->d_sb->s_fs_info;
6942         rcu_read_lock();
6943 diff -NurpP --minimal linux-2.6.36/fs/proc/generic.c linux-2.6.36-vs2.3.0.36.38/fs/proc/generic.c
6944 --- linux-2.6.36/fs/proc/generic.c      2010-10-21 13:07:51.000000000 +0200
6945 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/generic.c        2010-10-21 21:43:28.000000000 +0200
6946 @@ -22,6 +22,7 @@
6947  #include <linux/bitops.h>
6948  #include <linux/spinlock.h>
6949  #include <linux/completion.h>
6950 +#include <linux/vserver/inode.h>
6951  #include <asm/uaccess.h>
6952
6953  #include "internal.h"
6954 @@ -424,6 +425,8 @@ struct dentry *proc_lookup_de(struct pro
6955         for (de = de->subdir; de ; de = de->next) {
6956                 if (de->namelen != dentry->d_name.len)
6957                         continue;
6958 +               if (!vx_hide_check(0, de->vx_flags))
6959 +                       continue;
6960                 if (!memcmp(dentry->d_name.name, de->name, de->namelen)) {
6961                         unsigned int ino;
6962
6963 @@ -432,6 +435,8 @@ struct dentry *proc_lookup_de(struct pro
6964                         spin_unlock(&proc_subdir_lock);
6965                         error = -EINVAL;
6966                         inode = proc_get_inode(dir->i_sb, ino, de);
6967 +                       /* generic proc entries belong to the host */
6968 +                       inode->i_tag = 0;
6969                         goto out_unlock;
6970                 }
6971         }
6972 @@ -509,6 +514,8 @@ int proc_readdir_de(struct proc_dir_entr
6973
6974                                 /* filldir passes info to user space */
6975                                 pde_get(de);
6976 +                               if (!vx_hide_check(0, de->vx_flags))
6977 +                                       goto skip;
6978                                 spin_unlock(&proc_subdir_lock);
6979                                 if (filldir(dirent, de->name, de->namelen, filp->f_pos,
6980                                             de->low_ino, de->mode >> 12) < 0) {
6981 @@ -516,6 +523,7 @@ int proc_readdir_de(struct proc_dir_entr
6982                                         goto out;
6983                                 }
6984                                 spin_lock(&proc_subdir_lock);
6985 +                       skip:
6986                                 filp->f_pos++;
6987                                 next = de->next;
6988                                 pde_put(de);
6989 @@ -630,6 +638,7 @@ static struct proc_dir_entry *__proc_cre
6990         ent->nlink = nlink;
6991         atomic_set(&ent->count, 1);
6992         ent->pde_users = 0;
6993 +       ent->vx_flags = IATTR_PROC_DEFAULT;
6994         spin_lock_init(&ent->pde_unload_lock);
6995         ent->pde_unload_completion = NULL;
6996         INIT_LIST_HEAD(&ent->pde_openers);
6997 @@ -653,7 +662,8 @@ struct proc_dir_entry *proc_symlink(cons
6998                                 kfree(ent->data);
6999                                 kfree(ent);
7000                                 ent = NULL;
7001 -                       }
7002 +                       } else
7003 +                               ent->vx_flags = IATTR_PROC_SYMLINK;
7004                 } else {
7005                         kfree(ent);
7006                         ent = NULL;
7007 diff -NurpP --minimal linux-2.6.36/fs/proc/inode.c linux-2.6.36-vs2.3.0.36.38/fs/proc/inode.c
7008 --- linux-2.6.36/fs/proc/inode.c        2010-10-21 13:07:51.000000000 +0200
7009 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/inode.c  2010-10-21 13:09:36.000000000 +0200
7010 @@ -428,6 +428,8 @@ struct inode *proc_get_inode(struct supe
7011                         inode->i_uid = de->uid;
7012                         inode->i_gid = de->gid;
7013                 }
7014 +               if (de->vx_flags)
7015 +                       PROC_I(inode)->vx_flags = de->vx_flags;
7016                 if (de->size)
7017                         inode->i_size = de->size;
7018                 if (de->nlink)
7019 diff -NurpP --minimal linux-2.6.36/fs/proc/internal.h linux-2.6.36-vs2.3.0.36.38/fs/proc/internal.h
7020 --- linux-2.6.36/fs/proc/internal.h     2010-02-25 11:52:06.000000000 +0100
7021 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/internal.h       2010-10-21 13:09:36.000000000 +0200
7022 @@ -10,6 +10,7 @@
7023   */
7024
7025  #include <linux/proc_fs.h>
7026 +#include <linux/vs_pid.h>
7027
7028  extern struct proc_dir_entry proc_root;
7029  #ifdef CONFIG_PROC_SYSCTL
7030 @@ -51,6 +52,9 @@ extern int proc_pid_status(struct seq_fi
7031                                 struct pid *pid, struct task_struct *task);
7032  extern int proc_pid_statm(struct seq_file *m, struct pid_namespace *ns,
7033                                 struct pid *pid, struct task_struct *task);
7034 +extern int proc_pid_nsproxy(struct seq_file *m, struct pid_namespace *ns,
7035 +                               struct pid *pid, struct task_struct *task);
7036 +
7037  extern loff_t mem_lseek(struct file *file, loff_t offset, int orig);
7038
7039  extern const struct file_operations proc_maps_operations;
7040 @@ -68,11 +72,16 @@ static inline struct pid *proc_pid(struc
7041         return PROC_I(inode)->pid;
7042  }
7043
7044 -static inline struct task_struct *get_proc_task(struct inode *inode)
7045 +static inline struct task_struct *get_proc_task_real(struct inode *inode)
7046  {
7047         return get_pid_task(proc_pid(inode), PIDTYPE_PID);
7048  }
7049
7050 +static inline struct task_struct *get_proc_task(struct inode *inode)
7051 +{
7052 +       return vx_get_proc_task(inode, proc_pid(inode));
7053 +}
7054 +
7055  static inline int proc_fd(struct inode *inode)
7056  {
7057         return PROC_I(inode)->fd;
7058 diff -NurpP --minimal linux-2.6.36/fs/proc/loadavg.c linux-2.6.36-vs2.3.0.36.38/fs/proc/loadavg.c
7059 --- linux-2.6.36/fs/proc/loadavg.c      2009-09-10 15:26:23.000000000 +0200
7060 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/loadavg.c        2010-10-21 13:09:36.000000000 +0200
7061 @@ -12,15 +12,27 @@
7062
7063  static int loadavg_proc_show(struct seq_file *m, void *v)
7064  {
7065 +       unsigned long running;
7066 +       unsigned int threads;
7067         unsigned long avnrun[3];
7068
7069         get_avenrun(avnrun, FIXED_1/200, 0);
7070
7071 +       if (vx_flags(VXF_VIRT_LOAD, 0)) {
7072 +               struct vx_info *vxi = current_vx_info();
7073 +
7074 +               running = atomic_read(&vxi->cvirt.nr_running);
7075 +               threads = atomic_read(&vxi->cvirt.nr_threads);
7076 +       } else {
7077 +               running = nr_running();
7078 +               threads = nr_threads;
7079 +       }
7080 +
7081         seq_printf(m, "%lu.%02lu %lu.%02lu %lu.%02lu %ld/%d %d\n",
7082                 LOAD_INT(avnrun[0]), LOAD_FRAC(avnrun[0]),
7083                 LOAD_INT(avnrun[1]), LOAD_FRAC(avnrun[1]),
7084                 LOAD_INT(avnrun[2]), LOAD_FRAC(avnrun[2]),
7085 -               nr_running(), nr_threads,
7086 +               running, threads,
7087                 task_active_pid_ns(current)->last_pid);
7088         return 0;
7089  }
7090 diff -NurpP --minimal linux-2.6.36/fs/proc/meminfo.c linux-2.6.36-vs2.3.0.36.38/fs/proc/meminfo.c
7091 --- linux-2.6.36/fs/proc/meminfo.c      2009-12-03 20:02:53.000000000 +0100
7092 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/meminfo.c        2010-10-21 13:09:36.000000000 +0200
7093 @@ -39,7 +39,8 @@ static int meminfo_proc_show(struct seq_
7094         allowed = ((totalram_pages - hugetlb_total_pages())
7095                 * sysctl_overcommit_ratio / 100) + total_swap_pages;
7096
7097 -       cached = global_page_state(NR_FILE_PAGES) -
7098 +       cached = vx_flags(VXF_VIRT_MEM, 0) ?
7099 +               vx_vsi_cached(&i) : global_page_state(NR_FILE_PAGES) -
7100                         total_swapcache_pages - i.bufferram;
7101         if (cached < 0)
7102                 cached = 0;
7103 diff -NurpP --minimal linux-2.6.36/fs/proc/root.c linux-2.6.36-vs2.3.0.36.38/fs/proc/root.c
7104 --- linux-2.6.36/fs/proc/root.c 2010-08-02 16:52:51.000000000 +0200
7105 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/root.c   2010-10-21 13:09:36.000000000 +0200
7106 @@ -18,9 +18,14 @@
7107  #include <linux/bitops.h>
7108  #include <linux/mount.h>
7109  #include <linux/pid_namespace.h>
7110 +#include <linux/vserver/inode.h>
7111
7112  #include "internal.h"
7113
7114 +struct proc_dir_entry *proc_virtual;
7115 +
7116 +extern void proc_vx_init(void);
7117 +
7118  static int proc_test_super(struct super_block *sb, void *data)
7119  {
7120         return sb->s_fs_info == data;
7121 @@ -135,6 +140,7 @@ void __init proc_root_init(void)
7122  #endif
7123         proc_mkdir("bus", NULL);
7124         proc_sys_init();
7125 +       proc_vx_init();
7126  }
7127
7128  static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat
7129 @@ -202,6 +208,7 @@ struct proc_dir_entry proc_root = {
7130         .proc_iops      = &proc_root_inode_operations,
7131         .proc_fops      = &proc_root_operations,
7132         .parent         = &proc_root,
7133 +       .vx_flags       = IATTR_ADMIN | IATTR_WATCH,
7134  };
7135
7136  int pid_ns_prepare_proc(struct pid_namespace *ns)
7137 diff -NurpP --minimal linux-2.6.36/fs/proc/uptime.c linux-2.6.36-vs2.3.0.36.38/fs/proc/uptime.c
7138 --- linux-2.6.36/fs/proc/uptime.c       2009-12-03 20:02:53.000000000 +0100
7139 +++ linux-2.6.36-vs2.3.0.36.38/fs/proc/uptime.c 2010-10-21 13:09:36.000000000 +0200
7140 @@ -4,22 +4,22 @@
7141  #include <linux/sched.h>
7142  #include <linux/seq_file.h>
7143  #include <linux/time.h>
7144 -#include <linux/kernel_stat.h>
7145 +#include <linux/vserver/cvirt.h>
7146  #include <asm/cputime.h>
7147
7148  static int uptime_proc_show(struct seq_file *m, void *v)
7149  {
7150         struct timespec uptime;
7151         struct timespec idle;
7152 -       int i;
7153 -       cputime_t idletime = cputime_zero;
7154 -
7155 -       for_each_possible_cpu(i)
7156 -               idletime = cputime64_add(idletime, kstat_cpu(i).cpustat.idle);
7157 +       cputime_t idletime = cputime_add(init_task.utime, init_task.stime);
7158
7159         do_posix_clock_monotonic_gettime(&uptime);
7160         monotonic_to_bootbased(&uptime);
7161         cputime_to_timespec(idletime, &idle);
7162 +
7163 +       if (vx_flags(VXF_VIRT_UPTIME, 0))
7164 +               vx_vsi_uptime(&uptime, &idle);
7165 +
7166         seq_printf(m, "%lu.%02lu %lu.%02lu\n",
7167                         (unsigned long) uptime.tv_sec,
7168                         (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
7169 diff -NurpP --minimal linux-2.6.36/fs/quota/dquot.c linux-2.6.36-vs2.3.0.36.38/fs/quota/dquot.c
7170 --- linux-2.6.36/fs/quota/dquot.c       2010-10-21 13:07:51.000000000 +0200
7171 +++ linux-2.6.36-vs2.3.0.36.38/fs/quota/dquot.c 2010-10-21 13:09:36.000000000 +0200
7172 @@ -1538,6 +1538,9 @@ int __dquot_alloc_space(struct inode *in
7173         int reserve = flags & DQUOT_SPACE_RESERVE;
7174         int nofail = flags & DQUOT_SPACE_NOFAIL;
7175
7176 +       if ((ret = dl_alloc_space(inode, number)))
7177 +               return ret;
7178 +
7179         /*
7180          * First test before acquiring mutex - solves deadlocks when we
7181          * re-enter the quota code and are already holding the mutex
7182 @@ -1592,6 +1595,9 @@ int dquot_alloc_inode(const struct inode
7183         int cnt, ret = 0;
7184         char warntype[MAXQUOTAS];
7185
7186 +       if ((ret = dl_alloc_inode(inode)))
7187 +               return ret;
7188 +
7189         /* First test before acquiring mutex - solves deadlocks when we
7190           * re-enter the quota code and are already holding the mutex */
7191         if (!dquot_active(inode))
7192 @@ -1662,6 +1668,8 @@ void __dquot_free_space(struct inode *in
7193         char warntype[MAXQUOTAS];
7194         int reserve = flags & DQUOT_SPACE_RESERVE;
7195
7196 +       dl_free_space(inode, number);
7197 +
7198         /* First test before acquiring mutex - solves deadlocks when we
7199           * re-enter the quota code and are already holding the mutex */
7200         if (!dquot_active(inode)) {
7201 @@ -1700,6 +1708,8 @@ void dquot_free_inode(const struct inode
7202         unsigned int cnt;
7203         char warntype[MAXQUOTAS];
7204
7205 +       dl_free_inode(inode);
7206 +
7207         /* First test before acquiring mutex - solves deadlocks when we
7208           * re-enter the quota code and are already holding the mutex */
7209         if (!dquot_active(inode))
7210 diff -NurpP --minimal linux-2.6.36/fs/quota/quota.c linux-2.6.36-vs2.3.0.36.38/fs/quota/quota.c
7211 --- linux-2.6.36/fs/quota/quota.c       2010-08-02 16:52:51.000000000 +0200
7212 +++ linux-2.6.36-vs2.3.0.36.38/fs/quota/quota.c 2010-10-21 13:09:36.000000000 +0200
7213 @@ -8,6 +8,7 @@
7214  #include <linux/fs.h>
7215  #include <linux/namei.h>
7216  #include <linux/slab.h>
7217 +#include <linux/vs_context.h>
7218  #include <asm/current.h>
7219  #include <asm/uaccess.h>
7220  #include <linux/kernel.h>
7221 @@ -38,7 +39,7 @@ static int check_quotactl_permission(str
7222                         break;
7223                 /*FALLTHROUGH*/
7224         default:
7225 -               if (!capable(CAP_SYS_ADMIN))
7226 +               if (!vx_capable(CAP_SYS_ADMIN, VXC_QUOTA_CTL))
7227                         return -EPERM;
7228         }
7229
7230 @@ -296,6 +297,46 @@ static int do_quotactl(struct super_bloc
7231         }
7232  }
7233
7234 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7235 +
7236 +#include <linux/vroot.h>
7237 +#include <linux/major.h>
7238 +#include <linux/module.h>
7239 +#include <linux/kallsyms.h>
7240 +#include <linux/vserver/debug.h>
7241 +
7242 +static vroot_grb_func *vroot_get_real_bdev = NULL;
7243 +
7244 +static spinlock_t vroot_grb_lock = SPIN_LOCK_UNLOCKED;
7245 +
7246 +int register_vroot_grb(vroot_grb_func *func) {
7247 +       int ret = -EBUSY;
7248 +
7249 +       spin_lock(&vroot_grb_lock);
7250 +       if (!vroot_get_real_bdev) {
7251 +               vroot_get_real_bdev = func;
7252 +               ret = 0;
7253 +       }
7254 +       spin_unlock(&vroot_grb_lock);
7255 +       return ret;
7256 +}
7257 +EXPORT_SYMBOL(register_vroot_grb);
7258 +
7259 +int unregister_vroot_grb(vroot_grb_func *func) {
7260 +       int ret = -EINVAL;
7261 +
7262 +       spin_lock(&vroot_grb_lock);
7263 +       if (vroot_get_real_bdev) {
7264 +               vroot_get_real_bdev = NULL;
7265 +               ret = 0;
7266 +       }
7267 +       spin_unlock(&vroot_grb_lock);
7268 +       return ret;
7269 +}
7270 +EXPORT_SYMBOL(unregister_vroot_grb);
7271 +
7272 +#endif
7273 +
7274  /*
7275   * look up a superblock on which quota ops will be performed
7276   * - use the name of a block device to find the superblock thereon
7277 @@ -313,6 +354,22 @@ static struct super_block *quotactl_bloc
7278         putname(tmp);
7279         if (IS_ERR(bdev))
7280                 return ERR_CAST(bdev);
7281 +#if defined(CONFIG_BLK_DEV_VROOT) || defined(CONFIG_BLK_DEV_VROOT_MODULE)
7282 +       if (bdev && bdev->bd_inode &&
7283 +                       imajor(bdev->bd_inode) == VROOT_MAJOR) {
7284 +               struct block_device *bdnew = (void *)-EINVAL;
7285 +
7286 +               if (vroot_get_real_bdev)
7287 +                       bdnew = vroot_get_real_bdev(bdev);
7288 +               else
7289 +                       vxdprintk(VXD_CBIT(misc, 0),
7290 +                                       "vroot_get_real_bdev not set");
7291 +               bdput(bdev);
7292 +               if (IS_ERR(bdnew))
7293 +                       return ERR_PTR(PTR_ERR(bdnew));
7294 +               bdev = bdnew;
7295 +       }
7296 +#endif
7297         sb = get_super(bdev);
7298         bdput(bdev);
7299         if (!sb)
7300 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/file.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/file.c
7301 --- linux-2.6.36/fs/reiserfs/file.c     2010-10-21 13:07:51.000000000 +0200
7302 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/file.c       2010-10-21 13:09:36.000000000 +0200
7303 @@ -313,4 +313,5 @@ const struct inode_operations reiserfs_f
7304         .listxattr = reiserfs_listxattr,
7305         .removexattr = reiserfs_removexattr,
7306         .permission = reiserfs_permission,
7307 +       .sync_flags = reiserfs_sync_flags,
7308  };
7309 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/inode.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/inode.c
7310 --- linux-2.6.36/fs/reiserfs/inode.c    2010-10-21 13:07:51.000000000 +0200
7311 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/inode.c      2010-10-21 13:59:16.000000000 +0200
7312 @@ -19,6 +19,7 @@
7313  #include <linux/writeback.h>
7314  #include <linux/quotaops.h>
7315  #include <linux/swap.h>
7316 +#include <linux/vs_tag.h>
7317
7318  int reiserfs_commit_write(struct file *f, struct page *page,
7319                           unsigned from, unsigned to);
7320 @@ -1133,6 +1134,8 @@ static void init_inode(struct inode *ino
7321         struct buffer_head *bh;
7322         struct item_head *ih;
7323         __u32 rdev;
7324 +       uid_t uid;
7325 +       gid_t gid;
7326         //int version = ITEM_VERSION_1;
7327
7328         bh = PATH_PLAST_BUFFER(path);
7329 @@ -1153,12 +1156,13 @@ static void init_inode(struct inode *ino
7330                     (struct stat_data_v1 *)B_I_PITEM(bh, ih);
7331                 unsigned long blocks;
7332
7333 +               uid = sd_v1_uid(sd);
7334 +               gid = sd_v1_gid(sd);
7335 +
7336                 set_inode_item_key_version(inode, KEY_FORMAT_3_5);
7337                 set_inode_sd_version(inode, STAT_DATA_V1);
7338                 inode->i_mode = sd_v1_mode(sd);
7339                 inode->i_nlink = sd_v1_nlink(sd);
7340 -               inode->i_uid = sd_v1_uid(sd);
7341 -               inode->i_gid = sd_v1_gid(sd);
7342                 inode->i_size = sd_v1_size(sd);
7343                 inode->i_atime.tv_sec = sd_v1_atime(sd);
7344                 inode->i_mtime.tv_sec = sd_v1_mtime(sd);
7345 @@ -1200,11 +1204,12 @@ static void init_inode(struct inode *ino
7346                 // (directories and symlinks)
7347                 struct stat_data *sd = (struct stat_data *)B_I_PITEM(bh, ih);
7348
7349 +               uid    = sd_v2_uid(sd);
7350 +               gid    = sd_v2_gid(sd);
7351 +
7352                 inode->i_mode = sd_v2_mode(sd);
7353                 inode->i_nlink = sd_v2_nlink(sd);
7354 -               inode->i_uid = sd_v2_uid(sd);
7355                 inode->i_size = sd_v2_size(sd);
7356 -               inode->i_gid = sd_v2_gid(sd);
7357                 inode->i_mtime.tv_sec = sd_v2_mtime(sd);
7358                 inode->i_atime.tv_sec = sd_v2_atime(sd);
7359                 inode->i_ctime.tv_sec = sd_v2_ctime(sd);
7360 @@ -1234,6 +1239,10 @@ static void init_inode(struct inode *ino
7361                 sd_attrs_to_i_attrs(sd_v2_attrs(sd), inode);
7362         }
7363
7364 +       inode->i_uid = INOTAG_UID(DX_TAG(inode), uid, gid);
7365 +       inode->i_gid = INOTAG_GID(DX_TAG(inode), uid, gid);
7366 +       inode->i_tag = INOTAG_TAG(DX_TAG(inode), uid, gid, 0);
7367 +
7368         pathrelse(path);
7369         if (S_ISREG(inode->i_mode)) {
7370                 inode->i_op = &reiserfs_file_inode_operations;
7371 @@ -1256,13 +1265,15 @@ static void init_inode(struct inode *ino
7372  static void inode2sd(void *sd, struct inode *inode, loff_t size)
7373  {
7374         struct stat_data *sd_v2 = (struct stat_data *)sd;
7375 +       uid_t uid = TAGINO_UID(DX_TAG(inode), inode->i_uid, inode->i_tag);
7376 +       gid_t gid = TAGINO_GID(DX_TAG(inode), inode->i_gid, inode->i_tag);
7377         __u16 flags;
7378
7379 +       set_sd_v2_uid(sd_v2, uid);
7380 +       set_sd_v2_gid(sd_v2, gid);
7381         set_sd_v2_mode(sd_v2, inode->i_mode);
7382         set_sd_v2_nlink(sd_v2, inode->i_nlink);
7383 -       set_sd_v2_uid(sd_v2, inode->i_uid);
7384         set_sd_v2_size(sd_v2, size);
7385 -       set_sd_v2_gid(sd_v2, inode->i_gid);
7386         set_sd_v2_mtime(sd_v2, inode->i_mtime.tv_sec);
7387         set_sd_v2_atime(sd_v2, inode->i_atime.tv_sec);
7388         set_sd_v2_ctime(sd_v2, inode->i_ctime.tv_sec);
7389 @@ -2861,14 +2872,19 @@ int reiserfs_commit_write(struct file *f
7390  void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode)
7391  {
7392         if (reiserfs_attrs(inode->i_sb)) {
7393 -               if (sd_attrs & REISERFS_SYNC_FL)
7394 -                       inode->i_flags |= S_SYNC;
7395 -               else
7396 -                       inode->i_flags &= ~S_SYNC;
7397                 if (sd_attrs & REISERFS_IMMUTABLE_FL)
7398                         inode->i_flags |= S_IMMUTABLE;
7399                 else
7400                         inode->i_flags &= ~S_IMMUTABLE;
7401 +               if (sd_attrs & REISERFS_IXUNLINK_FL)
7402 +                       inode->i_flags |= S_IXUNLINK;
7403 +               else
7404 +                       inode->i_flags &= ~S_IXUNLINK;
7405 +
7406 +               if (sd_attrs & REISERFS_SYNC_FL)
7407 +                       inode->i_flags |= S_SYNC;
7408 +               else
7409 +                       inode->i_flags &= ~S_SYNC;
7410                 if (sd_attrs & REISERFS_APPEND_FL)
7411                         inode->i_flags |= S_APPEND;
7412                 else
7413 @@ -2881,6 +2897,15 @@ void sd_attrs_to_i_attrs(__u16 sd_attrs,
7414                         REISERFS_I(inode)->i_flags |= i_nopack_mask;
7415                 else
7416                         REISERFS_I(inode)->i_flags &= ~i_nopack_mask;
7417 +
7418 +               if (sd_attrs & REISERFS_BARRIER_FL)
7419 +                       inode->i_vflags |= V_BARRIER;
7420 +               else
7421 +                       inode->i_vflags &= ~V_BARRIER;
7422 +               if (sd_attrs & REISERFS_COW_FL)
7423 +                       inode->i_vflags |= V_COW;
7424 +               else
7425 +                       inode->i_vflags &= ~V_COW;
7426         }
7427  }
7428
7429 @@ -2891,6 +2916,11 @@ void i_attrs_to_sd_attrs(struct inode *i
7430                         *sd_attrs |= REISERFS_IMMUTABLE_FL;
7431                 else
7432                         *sd_attrs &= ~REISERFS_IMMUTABLE_FL;
7433 +               if (inode->i_flags & S_IXUNLINK)
7434 +                       *sd_attrs |= REISERFS_IXUNLINK_FL;
7435 +               else
7436 +                       *sd_attrs &= ~REISERFS_IXUNLINK_FL;
7437 +
7438                 if (inode->i_flags & S_SYNC)
7439                         *sd_attrs |= REISERFS_SYNC_FL;
7440                 else
7441 @@ -2903,6 +2933,15 @@ void i_attrs_to_sd_attrs(struct inode *i
7442                         *sd_attrs |= REISERFS_NOTAIL_FL;
7443                 else
7444                         *sd_attrs &= ~REISERFS_NOTAIL_FL;
7445 +
7446 +               if (inode->i_vflags & V_BARRIER)
7447 +                       *sd_attrs |= REISERFS_BARRIER_FL;
7448 +               else
7449 +                       *sd_attrs &= ~REISERFS_BARRIER_FL;
7450 +               if (inode->i_vflags & V_COW)
7451 +                       *sd_attrs |= REISERFS_COW_FL;
7452 +               else
7453 +                       *sd_attrs &= ~REISERFS_COW_FL;
7454         }
7455  }
7456
7457 @@ -3146,7 +3185,8 @@ int reiserfs_setattr(struct dentry *dent
7458         }
7459
7460         if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
7461 -           (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
7462 +           (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid) ||
7463 +           (ia_valid & ATTR_TAG && attr->ia_tag != inode->i_tag)) {
7464                 struct reiserfs_transaction_handle th;
7465                 int jbegin_count =
7466                     2 *
7467 @@ -3175,6 +3215,9 @@ int reiserfs_setattr(struct dentry *dent
7468                         inode->i_uid = attr->ia_uid;
7469                 if (attr->ia_valid & ATTR_GID)
7470                         inode->i_gid = attr->ia_gid;
7471 +                               if ((attr->ia_valid & ATTR_TAG) &&
7472 +                                       IS_TAGGED(inode))
7473 +                                       inode->i_tag = attr->ia_tag;
7474                 mark_inode_dirty(inode);
7475                 error = journal_end(&th, inode->i_sb, jbegin_count);
7476                 if (error)
7477 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/ioctl.c
7478 --- linux-2.6.36/fs/reiserfs/ioctl.c    2010-10-21 13:07:51.000000000 +0200
7479 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/ioctl.c      2010-10-21 13:09:36.000000000 +0200
7480 @@ -7,11 +7,27 @@
7481  #include <linux/mount.h>
7482  #include <linux/reiserfs_fs.h>
7483  #include <linux/time.h>
7484 +#include <linux/mount.h>
7485  #include <asm/uaccess.h>
7486  #include <linux/pagemap.h>
7487  #include <linux/smp_lock.h>
7488  #include <linux/compat.h>
7489
7490 +
7491 +int reiserfs_sync_flags(struct inode *inode, int flags, int vflags)
7492 +{
7493 +       __u16 sd_attrs = 0;
7494 +
7495 +       inode->i_flags = flags;
7496 +       inode->i_vflags = vflags;
7497 +
7498 +       i_attrs_to_sd_attrs(inode, &sd_attrs);
7499 +       REISERFS_I(inode)->i_attrs = sd_attrs;
7500 +       inode->i_ctime = CURRENT_TIME_SEC;
7501 +       mark_inode_dirty(inode);
7502 +       return 0;
7503 +}
7504 +
7505  /*
7506   * reiserfs_ioctl - handler for ioctl for inode
7507   * supported commands:
7508 @@ -23,7 +39,7 @@
7509  long reiserfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
7510  {
7511         struct inode *inode = filp->f_path.dentry->d_inode;
7512 -       unsigned int flags;
7513 +       unsigned int flags, oldflags;
7514         int err = 0;
7515
7516         reiserfs_write_lock(inode->i_sb);
7517 @@ -48,6 +64,7 @@ long reiserfs_ioctl(struct file *filp, u
7518
7519                 flags = REISERFS_I(inode)->i_attrs;
7520                 i_attrs_to_sd_attrs(inode, (__u16 *) & flags);
7521 +               flags &= REISERFS_FL_USER_VISIBLE;
7522                 err = put_user(flags, (int __user *)arg);
7523                 break;
7524         case REISERFS_IOC_SETFLAGS:{
7525 @@ -68,6 +85,10 @@ long reiserfs_ioctl(struct file *filp, u
7526                                 err = -EFAULT;
7527                                 goto setflags_out;
7528                         }
7529 +                       if (IS_BARRIER(inode)) {
7530 +                               vxwprintk_task(1, "messing with the barrier.");
7531 +                               return -EACCES;
7532 +                       }
7533                         /*
7534                          * Is it quota file? Do not allow user to mess with it
7535                          */
7536 @@ -92,6 +113,10 @@ long reiserfs_ioctl(struct file *filp, u
7537                                         goto setflags_out;
7538                                 }
7539                         }
7540 +
7541 +                       oldflags = REISERFS_I(inode)->i_attrs;
7542 +                       flags &= REISERFS_FL_USER_MODIFIABLE;
7543 +                       flags |= oldflags & ~REISERFS_FL_USER_MODIFIABLE;
7544                         sd_attrs_to_i_attrs(flags, inode);
7545                         REISERFS_I(inode)->i_attrs = flags;
7546                         inode->i_ctime = CURRENT_TIME_SEC;
7547 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/namei.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/namei.c
7548 --- linux-2.6.36/fs/reiserfs/namei.c    2010-08-02 16:52:52.000000000 +0200
7549 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/namei.c      2010-10-21 13:09:36.000000000 +0200
7550 @@ -18,6 +18,7 @@
7551  #include <linux/reiserfs_acl.h>
7552  #include <linux/reiserfs_xattr.h>
7553  #include <linux/quotaops.h>
7554 +#include <linux/vs_tag.h>
7555
7556  #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; }
7557  #define DEC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) drop_nlink(i);
7558 @@ -362,6 +363,7 @@ static struct dentry *reiserfs_lookup(st
7559         if (retval == IO_ERROR) {
7560                 return ERR_PTR(-EIO);
7561         }
7562 +               dx_propagate_tag(nd, inode);
7563
7564         return d_splice_alias(inode, dentry);
7565  }
7566 @@ -1532,6 +1534,7 @@ const struct inode_operations reiserfs_d
7567         .listxattr = reiserfs_listxattr,
7568         .removexattr = reiserfs_removexattr,
7569         .permission = reiserfs_permission,
7570 +       .sync_flags = reiserfs_sync_flags,
7571  };
7572
7573  /*
7574 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/super.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/super.c
7575 --- linux-2.6.36/fs/reiserfs/super.c    2010-10-21 13:07:51.000000000 +0200
7576 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/super.c      2010-10-21 13:09:36.000000000 +0200
7577 @@ -893,6 +893,14 @@ static int reiserfs_parse_options(struct
7578                 {"user_xattr",.setmask = 1 << REISERFS_UNSUPPORTED_OPT},
7579                 {"nouser_xattr",.clrmask = 1 << REISERFS_UNSUPPORTED_OPT},
7580  #endif
7581 +#ifndef CONFIG_TAGGING_NONE
7582 +               {"tagxid",.setmask = 1 << REISERFS_TAGGED},
7583 +               {"tag",.setmask = 1 << REISERFS_TAGGED},
7584 +               {"notag",.clrmask = 1 << REISERFS_TAGGED},
7585 +#endif
7586 +#ifdef CONFIG_PROPAGATE
7587 +               {"tag",.arg_required = 'T',.values = NULL},
7588 +#endif
7589  #ifdef CONFIG_REISERFS_FS_POSIX_ACL
7590                 {"acl",.setmask = 1 << REISERFS_POSIXACL},
7591                 {"noacl",.clrmask = 1 << REISERFS_POSIXACL},
7592 @@ -1202,6 +1210,14 @@ static int reiserfs_remount(struct super
7593         handle_quota_files(s, qf_names, &qfmt);
7594  #endif
7595
7596 +       if ((mount_options & (1 << REISERFS_TAGGED)) &&
7597 +               !(s->s_flags & MS_TAGGED)) {
7598 +               reiserfs_warning(s, "super-vs01",
7599 +                       "reiserfs: tagging not permitted on remount.");
7600 +               err = -EINVAL;
7601 +               goto out_err;
7602 +       }
7603 +
7604         handle_attrs(s);
7605
7606         /* Add options that are safe here */
7607 @@ -1684,6 +1700,10 @@ static int reiserfs_fill_super(struct su
7608                 goto error;
7609         }
7610
7611 +       /* map mount option tagxid */
7612 +       if (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TAGGED))
7613 +               s->s_flags |= MS_TAGGED;
7614 +
7615         rs = SB_DISK_SUPER_BLOCK(s);
7616         /* Let's do basic sanity check to verify that underlying device is not
7617            smaller than the filesystem. If the check fails then abort and scream,
7618 diff -NurpP --minimal linux-2.6.36/fs/reiserfs/xattr.c linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/xattr.c
7619 --- linux-2.6.36/fs/reiserfs/xattr.c    2010-08-02 16:52:52.000000000 +0200
7620 +++ linux-2.6.36-vs2.3.0.36.38/fs/reiserfs/xattr.c      2010-10-21 13:09:36.000000000 +0200
7621 @@ -40,6 +40,7 @@
7622  #include <linux/errno.h>
7623  #include <linux/gfp.h>
7624  #include <linux/fs.h>
7625 +#include <linux/mount.h>
7626  #include <linux/file.h>
7627  #include <linux/pagemap.h>
7628  #include <linux/xattr.h>
7629 diff -NurpP --minimal linux-2.6.36/fs/stat.c linux-2.6.36-vs2.3.0.36.38/fs/stat.c
7630 --- linux-2.6.36/fs/stat.c      2010-10-21 13:07:51.000000000 +0200
7631 +++ linux-2.6.36-vs2.3.0.36.38/fs/stat.c        2010-10-21 13:09:36.000000000 +0200
7632 @@ -26,6 +26,7 @@ void generic_fillattr(struct inode *inod
7633         stat->nlink = inode->i_nlink;
7634         stat->uid = inode->i_uid;
7635         stat->gid = inode->i_gid;
7636 +       stat->tag = inode->i_tag;
7637         stat->rdev = inode->i_rdev;
7638         stat->atime = inode->i_atime;
7639         stat->mtime = inode->i_mtime;
7640 diff -NurpP --minimal linux-2.6.36/fs/statfs.c linux-2.6.36-vs2.3.0.36.38/fs/statfs.c
7641 --- linux-2.6.36/fs/statfs.c    2010-10-21 13:07:51.000000000 +0200
7642 +++ linux-2.6.36-vs2.3.0.36.38/fs/statfs.c      2010-10-21 14:07:35.000000000 +0200
7643 @@ -7,6 +7,8 @@
7644  #include <linux/statfs.h>
7645  #include <linux/security.h>
7646  #include <linux/uaccess.h>
7647 +#include <linux/vs_base.h>
7648 +#include <linux/vs_dlimit.h>
7649
7650  static int flags_by_mnt(int mnt_flags)
7651  {
7652 @@ -59,6 +61,8 @@ int statfs_by_dentry(struct dentry *dent
7653         retval = dentry->d_sb->s_op->statfs(dentry, buf);
7654         if (retval == 0 && buf->f_frsize == 0)
7655                 buf->f_frsize = buf->f_bsize;
7656 +       if (!vx_check(0, VS_ADMIN|VS_WATCH))
7657 +               vx_vsi_statfs(dentry->d_sb, buf);
7658         return retval;
7659  }
7660
7661 diff -NurpP --minimal linux-2.6.36/fs/super.c linux-2.6.36-vs2.3.0.36.38/fs/super.c
7662 --- linux-2.6.36/fs/super.c     2010-10-21 13:07:51.000000000 +0200
7663 +++ linux-2.6.36-vs2.3.0.36.38/fs/super.c       2010-10-21 13:09:36.000000000 +0200
7664 @@ -30,6 +30,9 @@
7665  #include <linux/idr.h>
7666  #include <linux/mutex.h>
7667  #include <linux/backing-dev.h>
7668 +#include <linux/devpts_fs.h>
7669 +#include <linux/proc_fs.h>
7670 +#include <linux/vs_context.h>
7671  #include "internal.h"
7672
7673
7674 @@ -918,12 +921,18 @@ struct vfsmount *
7675  vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
7676  {
7677         struct vfsmount *mnt;
7678 +       struct super_block *sb;
7679         char *secdata = NULL;
7680         int error;
7681
7682         if (!type)
7683                 return ERR_PTR(-ENODEV);
7684
7685 +       error = -EPERM;
7686 +       if ((type->fs_flags & FS_BINARY_MOUNTDATA) &&
7687 +               !vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT))
7688 +               goto out;
7689 +
7690         error = -ENOMEM;
7691         mnt = alloc_vfsmnt(name);
7692         if (!mnt)
7693 @@ -945,11 +954,19 @@ vfs_kern_mount(struct file_system_type *
7694         error = type->get_sb(type, flags, name, data, mnt);
7695         if (error < 0)
7696                 goto out_free_secdata;
7697 -       BUG_ON(!mnt->mnt_sb);
7698 -       WARN_ON(!mnt->mnt_sb->s_bdi);
7699 +
7700 +       sb = mnt->mnt_sb;
7701 +       BUG_ON(!sb);
7702 +       WARN_ON(!sb->s_bdi);
7703         mnt->mnt_sb->s_flags |= MS_BORN;
7704
7705 -       error = security_sb_kern_mount(mnt->mnt_sb, flags, secdata);
7706 +       error = -EPERM;
7707 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_BINARY_MOUNT) && !sb->s_bdev &&
7708 +               (sb->s_magic != PROC_SUPER_MAGIC) &&
7709 +               (sb->s_magic != DEVPTS_SUPER_MAGIC))
7710 +               goto out_sb;
7711 +
7712 +       error = security_sb_kern_mount(sb, flags, secdata);
7713         if (error)
7714                 goto out_sb;
7715
7716 diff -NurpP --minimal linux-2.6.36/fs/sysfs/mount.c linux-2.6.36-vs2.3.0.36.38/fs/sysfs/mount.c
7717 --- linux-2.6.36/fs/sysfs/mount.c       2010-10-21 13:07:51.000000000 +0200
7718 +++ linux-2.6.36-vs2.3.0.36.38/fs/sysfs/mount.c 2010-10-21 13:09:36.000000000 +0200
7719 @@ -47,7 +47,7 @@ static int sysfs_fill_super(struct super
7720
7721         sb->s_blocksize = PAGE_CACHE_SIZE;
7722         sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
7723 -       sb->s_magic = SYSFS_MAGIC;
7724 +       sb->s_magic = SYSFS_SUPER_MAGIC;
7725         sb->s_op = &sysfs_ops;
7726         sb->s_time_gran = 1;
7727
7728 diff -NurpP --minimal linux-2.6.36/fs/utimes.c linux-2.6.36-vs2.3.0.36.38/fs/utimes.c
7729 --- linux-2.6.36/fs/utimes.c    2010-10-21 13:07:51.000000000 +0200
7730 +++ linux-2.6.36-vs2.3.0.36.38/fs/utimes.c      2010-10-21 13:09:36.000000000 +0200
7731 @@ -8,6 +8,8 @@
7732  #include <linux/stat.h>
7733  #include <linux/utime.h>
7734  #include <linux/syscalls.h>
7735 +#include <linux/mount.h>
7736 +#include <linux/vs_cowbl.h>
7737  #include <asm/uaccess.h>
7738  #include <asm/unistd.h>
7739
7740 @@ -52,12 +54,18 @@ static int utimes_common(struct path *pa
7741  {
7742         int error;
7743         struct iattr newattrs;
7744 -       struct inode *inode = path->dentry->d_inode;
7745 +       struct inode *inode;
7746
7747         error = mnt_want_write(path->mnt);
7748         if (error)
7749                 goto out;
7750
7751 +       error = cow_check_and_break(path);
7752 +       if (error)
7753 +               goto mnt_drop_write_and_out;
7754 +
7755 +       inode = path->dentry->d_inode;
7756 +
7757         if (times && times[0].tv_nsec == UTIME_NOW &&
7758                      times[1].tv_nsec == UTIME_NOW)
7759                 times = NULL;
7760 diff -NurpP --minimal linux-2.6.36/fs/xattr.c linux-2.6.36-vs2.3.0.36.38/fs/xattr.c
7761 --- linux-2.6.36/fs/xattr.c     2010-08-02 16:52:52.000000000 +0200
7762 +++ linux-2.6.36-vs2.3.0.36.38/fs/xattr.c       2010-10-21 13:09:36.000000000 +0200
7763 @@ -18,6 +18,7 @@
7764  #include <linux/module.h>
7765  #include <linux/fsnotify.h>
7766  #include <linux/audit.h>
7767 +#include <linux/mount.h>
7768  #include <asm/uaccess.h>
7769
7770
7771 @@ -49,7 +50,7 @@ xattr_permission(struct inode *inode, co
7772          * The trusted.* namespace can only be accessed by a privileged user.
7773          */
7774         if (!strncmp(name, XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN))
7775 -               return (capable(CAP_SYS_ADMIN) ? 0 : -EPERM);
7776 +               return (vx_capable(CAP_SYS_ADMIN, VXC_FS_TRUSTED) ? 0 : -EPERM);
7777
7778         /* In user.* namespace, only regular files and directories can have
7779          * extended attributes. For sticky directories, only the owner and
7780 diff -NurpP --minimal linux-2.6.36/fs/xfs/linux-2.6/xfs_ioctl.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_ioctl.c
7781 --- linux-2.6.36/fs/xfs/linux-2.6/xfs_ioctl.c   2010-10-21 13:07:51.000000000 +0200
7782 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_ioctl.c     2010-10-21 14:02:37.000000000 +0200
7783 @@ -28,7 +28,7 @@
7784  #include "xfs_bmap_btree.h"
7785  #include "xfs_dinode.h"
7786  #include "xfs_inode.h"
7787 -#include "xfs_ioctl.h"
7788 +// #include "xfs_ioctl.h"
7789  #include "xfs_rtalloc.h"
7790  #include "xfs_itable.h"
7791  #include "xfs_error.h"
7792 @@ -738,6 +738,10 @@ xfs_merge_ioc_xflags(
7793                 xflags |= XFS_XFLAG_IMMUTABLE;
7794         else
7795                 xflags &= ~XFS_XFLAG_IMMUTABLE;
7796 +       if (flags & FS_IXUNLINK_FL)
7797 +               xflags |= XFS_XFLAG_IXUNLINK;
7798 +       else
7799 +               xflags &= ~XFS_XFLAG_IXUNLINK;
7800         if (flags & FS_APPEND_FL)
7801                 xflags |= XFS_XFLAG_APPEND;
7802         else
7803 @@ -766,6 +770,8 @@ xfs_di2lxflags(
7804
7805         if (di_flags & XFS_DIFLAG_IMMUTABLE)
7806                 flags |= FS_IMMUTABLE_FL;
7807 +       if (di_flags & XFS_DIFLAG_IXUNLINK)
7808 +               flags |= FS_IXUNLINK_FL;
7809         if (di_flags & XFS_DIFLAG_APPEND)
7810                 flags |= FS_APPEND_FL;
7811         if (di_flags & XFS_DIFLAG_SYNC)
7812 @@ -826,6 +832,8 @@ xfs_set_diflags(
7813         di_flags = (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC);
7814         if (xflags & XFS_XFLAG_IMMUTABLE)
7815                 di_flags |= XFS_DIFLAG_IMMUTABLE;
7816 +       if (xflags & XFS_XFLAG_IXUNLINK)
7817 +               di_flags |= XFS_DIFLAG_IXUNLINK;
7818         if (xflags & XFS_XFLAG_APPEND)
7819                 di_flags |= XFS_DIFLAG_APPEND;
7820         if (xflags & XFS_XFLAG_SYNC)
7821 @@ -868,6 +876,10 @@ xfs_diflags_to_linux(
7822                 inode->i_flags |= S_IMMUTABLE;
7823         else
7824                 inode->i_flags &= ~S_IMMUTABLE;
7825 +       if (xflags & XFS_XFLAG_IXUNLINK)
7826 +               inode->i_flags |= S_IXUNLINK;
7827 +       else
7828 +               inode->i_flags &= ~S_IXUNLINK;
7829         if (xflags & XFS_XFLAG_APPEND)
7830                 inode->i_flags |= S_APPEND;
7831         else
7832 @@ -1341,10 +1353,18 @@ xfs_file_ioctl(
7833         case XFS_IOC_FSGETXATTRA:
7834                 return xfs_ioc_fsgetxattr(ip, 1, arg);
7835         case XFS_IOC_FSSETXATTR:
7836 +               if (IS_BARRIER(inode)) {
7837 +                       vxwprintk_task(1, "messing with the barrier.");
7838 +                       return -XFS_ERROR(EACCES);
7839 +               }
7840                 return xfs_ioc_fssetxattr(ip, filp, arg);
7841         case XFS_IOC_GETXFLAGS:
7842                 return xfs_ioc_getxflags(ip, arg);
7843         case XFS_IOC_SETXFLAGS:
7844 +               if (IS_BARRIER(inode)) {
7845 +                       vxwprintk_task(1, "messing with the barrier.");
7846 +                       return -XFS_ERROR(EACCES);
7847 +               }
7848                 return xfs_ioc_setxflags(ip, filp, arg);
7849
7850         case XFS_IOC_FSSETDM: {
7851 diff -NurpP --minimal linux-2.6.36/fs/xfs/linux-2.6/xfs_ioctl.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_ioctl.h
7852 --- linux-2.6.36/fs/xfs/linux-2.6/xfs_ioctl.h   2010-07-07 18:31:54.000000000 +0200
7853 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_ioctl.h     2010-10-21 13:09:36.000000000 +0200
7854 @@ -70,6 +70,12 @@ xfs_handle_to_dentry(
7855         void __user             *uhandle,
7856         u32                     hlen);
7857
7858 +extern int
7859 +xfs_sync_flags(
7860 +       struct inode            *inode,
7861 +       int                     flags,
7862 +       int                     vflags);
7863 +
7864  extern long
7865  xfs_file_ioctl(
7866         struct file             *filp,
7867 diff -NurpP --minimal linux-2.6.36/fs/xfs/linux-2.6/xfs_iops.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_iops.c
7868 --- linux-2.6.36/fs/xfs/linux-2.6/xfs_iops.c    2010-10-21 13:07:51.000000000 +0200
7869 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_iops.c      2010-10-21 13:09:36.000000000 +0200
7870 @@ -30,6 +30,7 @@
7871  #include "xfs_bmap_btree.h"
7872  #include "xfs_dinode.h"
7873  #include "xfs_inode.h"
7874 +#include "xfs_ioctl.h"
7875  #include "xfs_bmap.h"
7876  #include "xfs_rtalloc.h"
7877  #include "xfs_error.h"
7878 @@ -49,6 +50,7 @@
7879  #include <linux/falloc.h>
7880  #include <linux/fiemap.h>
7881  #include <linux/slab.h>
7882 +#include <linux/vs_tag.h>
7883
7884  /*
7885   * Bring the timestamps in the XFS inode uptodate.
7886 @@ -499,6 +501,7 @@ xfs_vn_getattr(
7887         stat->nlink = ip->i_d.di_nlink;
7888         stat->uid = ip->i_d.di_uid;
7889         stat->gid = ip->i_d.di_gid;
7890 +       stat->tag = ip->i_d.di_tag;
7891         stat->ino = ip->i_ino;
7892         stat->atime = inode->i_atime;
7893         stat->mtime = inode->i_mtime;
7894 @@ -687,6 +690,7 @@ static const struct inode_operations xfs
7895         .listxattr              = xfs_vn_listxattr,
7896         .fallocate              = xfs_vn_fallocate,
7897         .fiemap                 = xfs_vn_fiemap,
7898 +       .sync_flags             = xfs_sync_flags,
7899  };
7900
7901  static const struct inode_operations xfs_dir_inode_operations = {
7902 @@ -712,6 +716,7 @@ static const struct inode_operations xfs
7903         .getxattr               = generic_getxattr,
7904         .removexattr            = generic_removexattr,
7905         .listxattr              = xfs_vn_listxattr,
7906 +       .sync_flags             = xfs_sync_flags,
7907  };
7908
7909  static const struct inode_operations xfs_dir_ci_inode_operations = {
7910 @@ -761,6 +766,10 @@ xfs_diflags_to_iflags(
7911                 inode->i_flags |= S_IMMUTABLE;
7912         else
7913                 inode->i_flags &= ~S_IMMUTABLE;
7914 +       if (ip->i_d.di_flags & XFS_DIFLAG_IXUNLINK)
7915 +               inode->i_flags |= S_IXUNLINK;
7916 +       else
7917 +               inode->i_flags &= ~S_IXUNLINK;
7918         if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
7919                 inode->i_flags |= S_APPEND;
7920         else
7921 @@ -773,6 +782,15 @@ xfs_diflags_to_iflags(
7922                 inode->i_flags |= S_NOATIME;
7923         else
7924                 inode->i_flags &= ~S_NOATIME;
7925 +
7926 +       if (ip->i_d.di_vflags & XFS_DIVFLAG_BARRIER)
7927 +               inode->i_vflags |= V_BARRIER;
7928 +       else
7929 +               inode->i_vflags &= ~V_BARRIER;
7930 +       if (ip->i_d.di_vflags & XFS_DIVFLAG_COW)
7931 +               inode->i_vflags |= V_COW;
7932 +       else
7933 +               inode->i_vflags &= ~V_COW;
7934  }
7935
7936  /*
7937 @@ -801,6 +819,7 @@ xfs_setup_inode(
7938         inode->i_nlink  = ip->i_d.di_nlink;
7939         inode->i_uid    = ip->i_d.di_uid;
7940         inode->i_gid    = ip->i_d.di_gid;
7941 +       inode->i_tag    = ip->i_d.di_tag;
7942
7943         switch (inode->i_mode & S_IFMT) {
7944         case S_IFBLK:
7945 diff -NurpP --minimal linux-2.6.36/fs/xfs/linux-2.6/xfs_linux.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_linux.h
7946 --- linux-2.6.36/fs/xfs/linux-2.6/xfs_linux.h   2010-10-21 13:07:51.000000000 +0200
7947 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_linux.h     2010-10-21 13:09:36.000000000 +0200
7948 @@ -116,6 +116,7 @@
7949
7950  #define current_cpu()          (raw_smp_processor_id())
7951  #define current_pid()          (current->pid)
7952 +#define current_fstag(cred,vp) (dx_current_fstag((vp)->i_sb))
7953  #define current_test_flags(f)  (current->flags & (f))
7954  #define current_set_flags_nested(sp, f)                \
7955                 (*(sp) = current->flags, current->flags |= (f))
7956 diff -NurpP --minimal linux-2.6.36/fs/xfs/linux-2.6/xfs_super.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_super.c
7957 --- linux-2.6.36/fs/xfs/linux-2.6/xfs_super.c   2010-10-21 13:07:51.000000000 +0200
7958 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/linux-2.6/xfs_super.c     2010-10-21 13:09:36.000000000 +0200
7959 @@ -113,6 +113,9 @@ mempool_t *xfs_ioend_pool;
7960  #define MNTOPT_QUOTANOENF  "qnoenforce"        /* same as uqnoenforce */
7961  #define MNTOPT_DELAYLOG   "delaylog"   /* Delayed loging enabled */
7962  #define MNTOPT_NODELAYLOG "nodelaylog" /* Delayed loging disabled */
7963 +#define MNTOPT_TAGXID  "tagxid"        /* context tagging for inodes */
7964 +#define MNTOPT_TAGGED  "tag"           /* context tagging for inodes */
7965 +#define MNTOPT_NOTAGTAG        "notag"         /* do not use context tagging */
7966
7967  /*
7968   * Table driven mount option parser.
7969 @@ -121,10 +124,14 @@ mempool_t *xfs_ioend_pool;
7970   * in the future, too.
7971   */
7972  enum {
7973 +       Opt_tag, Opt_notag,
7974         Opt_barrier, Opt_nobarrier, Opt_err
7975  };
7976
7977  static const match_table_t tokens = {
7978 +       {Opt_tag, "tagxid"},
7979 +       {Opt_tag, "tag"},
7980 +       {Opt_notag, "notag"},
7981         {Opt_barrier, "barrier"},
7982         {Opt_nobarrier, "nobarrier"},
7983         {Opt_err, NULL}
7984 @@ -371,6 +378,19 @@ xfs_parseargs(
7985                 } else if (!strcmp(this_char, "irixsgid")) {
7986                         cmn_err(CE_WARN,
7987         "XFS: irixsgid is now a sysctl(2) variable, option is deprecated.");
7988 +#ifndef CONFIG_TAGGING_NONE
7989 +               } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
7990 +                       mp->m_flags |= XFS_MOUNT_TAGGED;
7991 +               } else if (!strcmp(this_char, MNTOPT_NOTAGTAG)) {
7992 +                       mp->m_flags &= ~XFS_MOUNT_TAGGED;
7993 +               } else if (!strcmp(this_char, MNTOPT_TAGXID)) {
7994 +                       mp->m_flags |= XFS_MOUNT_TAGGED;
7995 +#endif
7996 +#ifdef CONFIG_PROPAGATE
7997 +               } else if (!strcmp(this_char, MNTOPT_TAGGED)) {
7998 +                       /* use value */
7999 +                       mp->m_flags |= XFS_MOUNT_TAGGED;
8000 +#endif
8001                 } else {
8002                         cmn_err(CE_WARN,
8003                                 "XFS: unknown mount option [%s].", this_char);
8004 @@ -1320,6 +1340,16 @@ xfs_fs_remount(
8005                 case Opt_nobarrier:
8006                         mp->m_flags &= ~XFS_MOUNT_BARRIER;
8007                         break;
8008 +               case Opt_tag:
8009 +                       if (!(sb->s_flags & MS_TAGGED)) {
8010 +                               printk(KERN_INFO
8011 +                                       "XFS: %s: tagging not permitted on remount.\n",
8012 +                                       sb->s_id);
8013 +                               return -EINVAL;
8014 +                       }
8015 +                       break;
8016 +               case Opt_notag:
8017 +                       break;
8018                 default:
8019                         /*
8020                          * Logically we would return an error here to prevent
8021 @@ -1547,6 +1577,9 @@ xfs_fs_fill_super(
8022         if (error)
8023                 goto out_filestream_unmount;
8024
8025 +       if (mp->m_flags & XFS_MOUNT_TAGGED)
8026 +               sb->s_flags |= MS_TAGGED;
8027 +
8028         sb->s_magic = XFS_SB_MAGIC;
8029         sb->s_blocksize = mp->m_sb.sb_blocksize;
8030         sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1;
8031 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_dinode.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_dinode.h
8032 --- linux-2.6.36/fs/xfs/xfs_dinode.h    2009-06-11 17:13:09.000000000 +0200
8033 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_dinode.h      2010-10-21 13:09:36.000000000 +0200
8034 @@ -50,7 +50,9 @@ typedef struct xfs_dinode {
8035         __be32          di_gid;         /* owner's group id */
8036         __be32          di_nlink;       /* number of links to file */
8037         __be16          di_projid;      /* owner's project id */
8038 -       __u8            di_pad[8];      /* unused, zeroed space */
8039 +       __be16          di_tag;         /* context tagging */
8040 +       __be16          di_vflags;      /* vserver specific flags */
8041 +       __u8            di_pad[4];      /* unused, zeroed space */
8042         __be16          di_flushiter;   /* incremented on flush */
8043         xfs_timestamp_t di_atime;       /* time last accessed */
8044         xfs_timestamp_t di_mtime;       /* time last modified */
8045 @@ -183,6 +185,8 @@ static inline void xfs_dinode_put_rdev(s
8046  #define XFS_DIFLAG_EXTSZINHERIT_BIT 12 /* inherit inode extent size */
8047  #define XFS_DIFLAG_NODEFRAG_BIT     13 /* do not reorganize/defragment */
8048  #define XFS_DIFLAG_FILESTREAM_BIT   14  /* use filestream allocator */
8049 +#define XFS_DIFLAG_IXUNLINK_BIT     15 /* Immutable inver on unlink */
8050 +
8051  #define XFS_DIFLAG_REALTIME      (1 << XFS_DIFLAG_REALTIME_BIT)
8052  #define XFS_DIFLAG_PREALLOC      (1 << XFS_DIFLAG_PREALLOC_BIT)
8053  #define XFS_DIFLAG_NEWRTBM       (1 << XFS_DIFLAG_NEWRTBM_BIT)
8054 @@ -198,6 +202,7 @@ static inline void xfs_dinode_put_rdev(s
8055  #define XFS_DIFLAG_EXTSZINHERIT  (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
8056  #define XFS_DIFLAG_NODEFRAG      (1 << XFS_DIFLAG_NODEFRAG_BIT)
8057  #define XFS_DIFLAG_FILESTREAM    (1 << XFS_DIFLAG_FILESTREAM_BIT)
8058 +#define XFS_DIFLAG_IXUNLINK      (1 << XFS_DIFLAG_IXUNLINK_BIT)
8059
8060  #ifdef CONFIG_XFS_RT
8061  #define XFS_IS_REALTIME_INODE(ip) ((ip)->i_d.di_flags & XFS_DIFLAG_REALTIME)
8062 @@ -210,6 +215,10 @@ static inline void xfs_dinode_put_rdev(s
8063          XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
8064          XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
8065          XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
8066 -        XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
8067 +        XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM | \
8068 +        XFS_DIFLAG_IXUNLINK)
8069 +
8070 +#define XFS_DIVFLAG_BARRIER    0x01
8071 +#define XFS_DIVFLAG_COW                0x02
8072
8073  #endif /* __XFS_DINODE_H__ */
8074 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_fs.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_fs.h
8075 --- linux-2.6.36/fs/xfs/xfs_fs.h        2010-10-21 13:07:52.000000000 +0200
8076 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_fs.h  2010-10-21 13:09:36.000000000 +0200
8077 @@ -67,6 +67,9 @@ struct fsxattr {
8078  #define XFS_XFLAG_EXTSZINHERIT 0x00001000      /* inherit inode extent size */
8079  #define XFS_XFLAG_NODEFRAG     0x00002000      /* do not defragment */
8080  #define XFS_XFLAG_FILESTREAM   0x00004000      /* use filestream allocator */
8081 +#define XFS_XFLAG_IXUNLINK     0x00008000      /* immutable invert on unlink */
8082 +#define XFS_XFLAG_BARRIER      0x10000000      /* chroot() barrier */
8083 +#define XFS_XFLAG_COW          0x20000000      /* copy on write mark */
8084  #define XFS_XFLAG_HASATTR      0x80000000      /* no DIFLAG for this   */
8085
8086  /*
8087 @@ -295,7 +298,8 @@ typedef struct xfs_bstat {
8088         __u32           bs_gen;         /* generation count             */
8089         __u16           bs_projid;      /* project id                   */
8090         __u16           bs_forkoff;     /* inode fork offset in bytes   */
8091 -       unsigned char   bs_pad[12];     /* pad space, unused            */
8092 +       __u16           bs_tag;         /* context tagging              */
8093 +       unsigned char   bs_pad[10];     /* pad space, unused            */
8094         __u32           bs_dmevmask;    /* DMIG event mask              */
8095         __u16           bs_dmstate;     /* DMIG state info              */
8096         __u16           bs_aextents;    /* attribute number of extents  */
8097 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_ialloc.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_ialloc.c
8098 --- linux-2.6.36/fs/xfs/xfs_ialloc.c    2010-10-21 13:07:52.000000000 +0200
8099 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_ialloc.c      2010-10-21 13:09:36.000000000 +0200
8100 @@ -37,7 +37,6 @@
8101  #include "xfs_error.h"
8102  #include "xfs_bmap.h"
8103
8104 -
8105  /*
8106   * Allocation group level functions.
8107   */
8108 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_inode.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_inode.c
8109 --- linux-2.6.36/fs/xfs/xfs_inode.c     2010-10-21 13:07:52.000000000 +0200
8110 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_inode.c       2010-10-21 13:09:36.000000000 +0200
8111 @@ -245,6 +245,7 @@ xfs_inotobp(
8112         return 0;
8113  }
8114
8115 +#include <linux/vs_tag.h>
8116
8117  /*
8118   * This routine is called to map an inode to the buffer containing
8119 @@ -650,15 +651,25 @@ xfs_iformat_btree(
8120  STATIC void
8121  xfs_dinode_from_disk(
8122         xfs_icdinode_t          *to,
8123 -       xfs_dinode_t            *from)
8124 +       xfs_dinode_t            *from,
8125 +       int tagged)
8126  {
8127 +       uint32_t uid, gid, tag;
8128 +
8129         to->di_magic = be16_to_cpu(from->di_magic);
8130         to->di_mode = be16_to_cpu(from->di_mode);
8131         to->di_version = from ->di_version;
8132         to->di_format = from->di_format;
8133         to->di_onlink = be16_to_cpu(from->di_onlink);
8134 -       to->di_uid = be32_to_cpu(from->di_uid);
8135 -       to->di_gid = be32_to_cpu(from->di_gid);
8136 +
8137 +       uid = be32_to_cpu(from->di_uid);
8138 +       gid = be32_to_cpu(from->di_gid);
8139 +       tag = be16_to_cpu(from->di_tag);
8140 +
8141 +       to->di_uid = INOTAG_UID(tagged, uid, gid);
8142 +       to->di_gid = INOTAG_GID(tagged, uid, gid);
8143 +       to->di_tag = INOTAG_TAG(tagged, uid, gid, tag);
8144 +
8145         to->di_nlink = be32_to_cpu(from->di_nlink);
8146         to->di_projid = be16_to_cpu(from->di_projid);
8147         memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8148 @@ -679,21 +690,26 @@ xfs_dinode_from_disk(
8149         to->di_dmevmask = be32_to_cpu(from->di_dmevmask);
8150         to->di_dmstate  = be16_to_cpu(from->di_dmstate);
8151         to->di_flags    = be16_to_cpu(from->di_flags);
8152 +       to->di_vflags   = be16_to_cpu(from->di_vflags);
8153         to->di_gen      = be32_to_cpu(from->di_gen);
8154  }
8155
8156  void
8157  xfs_dinode_to_disk(
8158         xfs_dinode_t            *to,
8159 -       xfs_icdinode_t          *from)
8160 +       xfs_icdinode_t          *from,
8161 +       int tagged)
8162  {
8163         to->di_magic = cpu_to_be16(from->di_magic);
8164         to->di_mode = cpu_to_be16(from->di_mode);
8165         to->di_version = from ->di_version;
8166         to->di_format = from->di_format;
8167         to->di_onlink = cpu_to_be16(from->di_onlink);
8168 -       to->di_uid = cpu_to_be32(from->di_uid);
8169 -       to->di_gid = cpu_to_be32(from->di_gid);
8170 +
8171 +       to->di_uid = cpu_to_be32(TAGINO_UID(tagged, from->di_uid, from->di_tag));
8172 +       to->di_gid = cpu_to_be32(TAGINO_GID(tagged, from->di_gid, from->di_tag));
8173 +       to->di_tag = cpu_to_be16(TAGINO_TAG(tagged, from->di_tag));
8174 +
8175         to->di_nlink = cpu_to_be32(from->di_nlink);
8176         to->di_projid = cpu_to_be16(from->di_projid);
8177         memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
8178 @@ -714,12 +730,14 @@ xfs_dinode_to_disk(
8179         to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
8180         to->di_dmstate = cpu_to_be16(from->di_dmstate);
8181         to->di_flags = cpu_to_be16(from->di_flags);
8182 +       to->di_vflags = cpu_to_be16(from->di_vflags);
8183         to->di_gen = cpu_to_be32(from->di_gen);
8184  }
8185
8186  STATIC uint
8187  _xfs_dic2xflags(
8188 -       __uint16_t              di_flags)
8189 +       __uint16_t              di_flags,
8190 +       __uint16_t              di_vflags)
8191  {
8192         uint                    flags = 0;
8193
8194 @@ -730,6 +748,8 @@ _xfs_dic2xflags(
8195                         flags |= XFS_XFLAG_PREALLOC;
8196                 if (di_flags & XFS_DIFLAG_IMMUTABLE)
8197                         flags |= XFS_XFLAG_IMMUTABLE;
8198 +               if (di_flags & XFS_DIFLAG_IXUNLINK)
8199 +                       flags |= XFS_XFLAG_IXUNLINK;
8200                 if (di_flags & XFS_DIFLAG_APPEND)
8201                         flags |= XFS_XFLAG_APPEND;
8202                 if (di_flags & XFS_DIFLAG_SYNC)
8203 @@ -754,6 +774,10 @@ _xfs_dic2xflags(
8204                         flags |= XFS_XFLAG_FILESTREAM;
8205         }
8206
8207 +       if (di_vflags & XFS_DIVFLAG_BARRIER)
8208 +               flags |= FS_BARRIER_FL;
8209 +       if (di_vflags & XFS_DIVFLAG_COW)
8210 +               flags |= FS_COW_FL;
8211         return flags;
8212  }
8213
8214 @@ -763,7 +787,7 @@ xfs_ip2xflags(
8215  {
8216         xfs_icdinode_t          *dic = &ip->i_d;
8217
8218 -       return _xfs_dic2xflags(dic->di_flags) |
8219 +       return _xfs_dic2xflags(dic->di_flags, dic->di_vflags) |
8220                                 (XFS_IFORK_Q(ip) ? XFS_XFLAG_HASATTR : 0);
8221  }
8222
8223 @@ -771,7 +795,8 @@ uint
8224  xfs_dic2xflags(
8225         xfs_dinode_t            *dip)
8226  {
8227 -       return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
8228 +       return _xfs_dic2xflags(be16_to_cpu(dip->di_flags),
8229 +                               be16_to_cpu(dip->di_vflags)) |
8230                                 (XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
8231  }
8232
8233 @@ -804,7 +829,6 @@ xfs_iread(
8234         if (error)
8235                 return error;
8236         dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
8237 -
8238         /*
8239          * If we got something that isn't an inode it means someone
8240          * (nfs or dmi) has a stale handle.
8241 @@ -829,7 +853,8 @@ xfs_iread(
8242          * Otherwise, just get the truly permanent information.
8243          */
8244         if (dip->di_mode) {
8245 -               xfs_dinode_from_disk(&ip->i_d, dip);
8246 +               xfs_dinode_from_disk(&ip->i_d, dip,
8247 +                       mp->m_flags & XFS_MOUNT_TAGGED);
8248                 error = xfs_iformat(ip, dip);
8249                 if (error)  {
8250  #ifdef DEBUG
8251 @@ -1027,6 +1052,7 @@ xfs_ialloc(
8252         ASSERT(ip->i_d.di_nlink == nlink);
8253         ip->i_d.di_uid = current_fsuid();
8254         ip->i_d.di_gid = current_fsgid();
8255 +       ip->i_d.di_tag = current_fstag(cr, &ip->i_vnode);
8256         ip->i_d.di_projid = prid;
8257         memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
8258
8259 @@ -1087,6 +1113,7 @@ xfs_ialloc(
8260         ip->i_d.di_dmevmask = 0;
8261         ip->i_d.di_dmstate = 0;
8262         ip->i_d.di_flags = 0;
8263 +       ip->i_d.di_vflags = 0;
8264         flags = XFS_ILOG_CORE;
8265         switch (mode & S_IFMT) {
8266         case S_IFIFO:
8267 @@ -2105,6 +2132,7 @@ xfs_ifree(
8268         }
8269         ip->i_d.di_mode = 0;            /* mark incore inode as free */
8270         ip->i_d.di_flags = 0;
8271 +       ip->i_d.di_vflags = 0;
8272         ip->i_d.di_dmevmask = 0;
8273         ip->i_d.di_forkoff = 0;         /* mark the attr fork not in use */
8274         ip->i_df.if_ext_max =
8275 @@ -2975,7 +3003,8 @@ xfs_iflush_int(
8276          * because if the inode is dirty at all the core must
8277          * be.
8278          */
8279 -       xfs_dinode_to_disk(dip, &ip->i_d);
8280 +       xfs_dinode_to_disk(dip, &ip->i_d,
8281 +               mp->m_flags & XFS_MOUNT_TAGGED);
8282
8283         /* Wrap, we never let the log put out DI_MAX_FLUSH */
8284         if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
8285 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_inode.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_inode.h
8286 --- linux-2.6.36/fs/xfs/xfs_inode.h     2010-10-21 13:07:52.000000000 +0200
8287 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_inode.h       2010-10-21 13:09:36.000000000 +0200
8288 @@ -135,7 +135,9 @@ typedef struct xfs_icdinode {
8289         __uint32_t      di_gid;         /* owner's group id */
8290         __uint32_t      di_nlink;       /* number of links to file */
8291         __uint16_t      di_projid;      /* owner's project id */
8292 -       __uint8_t       di_pad[8];      /* unused, zeroed space */
8293 +       __uint16_t      di_tag;         /* context tagging */
8294 +       __uint16_t      di_vflags;      /* vserver specific flags */
8295 +       __uint8_t       di_pad[4];      /* unused, zeroed space */
8296         __uint16_t      di_flushiter;   /* incremented on flush */
8297         xfs_ictimestamp_t di_atime;     /* time last accessed */
8298         xfs_ictimestamp_t di_mtime;     /* time last modified */
8299 @@ -509,7 +511,7 @@ int         xfs_itobp(struct xfs_mount *, struc
8300  int            xfs_iread(struct xfs_mount *, struct xfs_trans *,
8301                           struct xfs_inode *, uint);
8302  void           xfs_dinode_to_disk(struct xfs_dinode *,
8303 -                                  struct xfs_icdinode *);
8304 +                                  struct xfs_icdinode *, int);
8305  void           xfs_idestroy_fork(struct xfs_inode *, int);
8306  void           xfs_idata_realloc(struct xfs_inode *, int, int);
8307  void           xfs_iroot_realloc(struct xfs_inode *, int, int);
8308 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_itable.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_itable.c
8309 --- linux-2.6.36/fs/xfs/xfs_itable.c    2010-10-21 13:07:52.000000000 +0200
8310 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_itable.c      2010-10-21 13:09:36.000000000 +0200
8311 @@ -97,6 +97,7 @@ xfs_bulkstat_one_int(
8312         buf->bs_mode = dic->di_mode;
8313         buf->bs_uid = dic->di_uid;
8314         buf->bs_gid = dic->di_gid;
8315 +       buf->bs_tag = dic->di_tag;
8316         buf->bs_size = dic->di_size;
8317
8318         /*
8319 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_log_recover.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_log_recover.c
8320 --- linux-2.6.36/fs/xfs/xfs_log_recover.c       2010-10-21 13:07:52.000000000 +0200
8321 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_log_recover.c 2010-10-21 14:02:02.000000000 +0200
8322 @@ -2450,7 +2450,8 @@ xlog_recover_do_inode_trans(
8323         }
8324
8325         /* The core is in in-core format */
8326 -       xfs_dinode_to_disk(dip, item->ri_buf[1].i_addr);
8327 +       xfs_dinode_to_disk(dip, item->ri_buf[1].i_addr,
8328 +               mp->m_flags & XFS_MOUNT_TAGGED);
8329
8330         /* the rest is in on-disk format */
8331         if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
8332 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_mount.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_mount.h
8333 --- linux-2.6.36/fs/xfs/xfs_mount.h     2010-10-21 13:07:52.000000000 +0200
8334 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_mount.h       2010-10-21 13:09:36.000000000 +0200
8335 @@ -238,6 +238,7 @@ typedef struct xfs_mount {
8336                                                    allocator */
8337  #define XFS_MOUNT_NOATTR2      (1ULL << 25)    /* disable use of attr2 format */
8338
8339 +#define XFS_MOUNT_TAGGED       (1ULL << 31)    /* context tagging */
8340
8341  /*
8342   * Default minimum read and write sizes.
8343 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_vnodeops.c linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_vnodeops.c
8344 --- linux-2.6.36/fs/xfs/xfs_vnodeops.c  2010-10-21 13:07:52.000000000 +0200
8345 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_vnodeops.c    2010-10-21 15:17:26.000000000 +0200
8346 @@ -50,6 +50,78 @@
8347  #include "xfs_vnodeops.h"
8348  #include "xfs_trace.h"
8349
8350 +
8351 +STATIC void
8352 +xfs_get_inode_flags(
8353 +       xfs_inode_t     *ip)
8354 +{
8355 +       struct inode    *inode = VFS_I(ip);
8356 +       unsigned int    flags = inode->i_flags;
8357 +       unsigned int    vflags = inode->i_vflags;
8358 +
8359 +       if (flags & S_IMMUTABLE)
8360 +               ip->i_d.di_flags |= XFS_DIFLAG_IMMUTABLE;
8361 +       else
8362 +               ip->i_d.di_flags &= ~XFS_DIFLAG_IMMUTABLE;
8363 +       if (flags & S_IXUNLINK)
8364 +               ip->i_d.di_flags |= XFS_DIFLAG_IXUNLINK;
8365 +       else
8366 +               ip->i_d.di_flags &= ~XFS_DIFLAG_IXUNLINK;
8367 +
8368 +       if (vflags & V_BARRIER)
8369 +               ip->i_d.di_vflags |= XFS_DIVFLAG_BARRIER;
8370 +       else
8371 +               ip->i_d.di_vflags &= ~XFS_DIVFLAG_BARRIER;
8372 +       if (vflags & V_COW)
8373 +               ip->i_d.di_vflags |= XFS_DIVFLAG_COW;
8374 +       else
8375 +               ip->i_d.di_vflags &= ~XFS_DIVFLAG_COW;
8376 +}
8377 +
8378 +int
8379 +xfs_sync_flags(
8380 +       struct inode            *inode,
8381 +       int                     flags,
8382 +       int                     vflags)
8383 +{
8384 +       struct xfs_inode        *ip = XFS_I(inode);
8385 +       struct xfs_mount        *mp = ip->i_mount;
8386 +       struct xfs_trans        *tp;
8387 +       unsigned int            lock_flags = 0;
8388 +       int                     code;
8389 +
8390 +       tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
8391 +       code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
8392 +       if (code)
8393 +               goto error_out;
8394 +
8395 +       xfs_ilock(ip, XFS_ILOCK_EXCL);
8396 +
8397 +       xfs_trans_ijoin(tp, ip);
8398 +
8399 +       inode->i_flags = flags;
8400 +       inode->i_vflags = vflags;
8401 +       xfs_get_inode_flags(ip);
8402 +
8403 +       xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
8404 +       xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
8405 +
8406 +       XFS_STATS_INC(xs_ig_attrchg);
8407 +
8408 +       if (mp->m_flags & XFS_MOUNT_WSYNC)
8409 +               xfs_trans_set_sync(tp);
8410 +       code = xfs_trans_commit(tp, 0);
8411 +       xfs_iunlock(ip, XFS_ILOCK_EXCL);
8412 +       return code;
8413 +
8414 +error_out:
8415 +       xfs_trans_cancel(tp, 0);
8416 +       if (lock_flags)
8417 +               xfs_iunlock(ip, XFS_ILOCK_EXCL);
8418 +       return code;
8419 +}
8420 +
8421 +
8422  int
8423  xfs_setattr(
8424         struct xfs_inode        *ip,
8425 @@ -65,6 +137,7 @@ xfs_setattr(
8426         uint                    commit_flags=0;
8427         uid_t                   uid=0, iuid=0;
8428         gid_t                   gid=0, igid=0;
8429 +       tag_t                   tag=0, itag=0;
8430         struct xfs_dquot        *udqp, *gdqp, *olddquot1, *olddquot2;
8431         int                     need_iolock = 1;
8432
8433 @@ -147,7 +220,7 @@ xfs_setattr(
8434         /*
8435          * Change file ownership.  Must be the owner or privileged.
8436          */
8437 -       if (mask & (ATTR_UID|ATTR_GID)) {
8438 +       if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8439                 /*
8440                  * These IDs could have changed since we last looked at them.
8441                  * But, we're assured that if the ownership did change
8442 @@ -156,8 +229,10 @@ xfs_setattr(
8443                  */
8444                 iuid = ip->i_d.di_uid;
8445                 igid = ip->i_d.di_gid;
8446 +               itag = ip->i_d.di_tag;
8447                 gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
8448                 uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
8449 +               tag = (mask & ATTR_TAG) ? iattr->ia_tag : itag;
8450
8451                 /*
8452                  * Do a quota reservation only if uid/gid is actually
8453 @@ -165,7 +240,8 @@ xfs_setattr(
8454                  */
8455                 if (XFS_IS_QUOTA_RUNNING(mp) &&
8456                     ((XFS_IS_UQUOTA_ON(mp) && iuid != uid) ||
8457 -                    (XFS_IS_GQUOTA_ON(mp) && igid != gid))) {
8458 +                    (XFS_IS_GQUOTA_ON(mp) && igid != gid) ||
8459 +                    (XFS_IS_GQUOTA_ON(mp) && itag != tag))) {
8460                         ASSERT(tp);
8461                         code = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
8462                                                 capable(CAP_FOWNER) ?
8463 @@ -326,7 +402,7 @@ xfs_setattr(
8464         /*
8465          * Change file ownership.  Must be the owner or privileged.
8466          */
8467 -       if (mask & (ATTR_UID|ATTR_GID)) {
8468 +       if (mask & (ATTR_UID|ATTR_GID|ATTR_TAG)) {
8469                 /*
8470                  * CAP_FSETID overrides the following restrictions:
8471                  *
8472 @@ -342,6 +418,10 @@ xfs_setattr(
8473                  * Change the ownerships and register quota modifications
8474                  * in the transaction.
8475                  */
8476 +               if (itag != tag) {
8477 +                       ip->i_d.di_tag = tag;
8478 +                       inode->i_tag = tag;
8479 +               }
8480                 if (iuid != uid) {
8481                         if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_UQUOTA_ON(mp)) {
8482                                 ASSERT(mask & ATTR_UID);
8483 diff -NurpP --minimal linux-2.6.36/fs/xfs/xfs_vnodeops.h linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_vnodeops.h
8484 --- linux-2.6.36/fs/xfs/xfs_vnodeops.h  2010-07-07 18:31:54.000000000 +0200
8485 +++ linux-2.6.36-vs2.3.0.36.38/fs/xfs/xfs_vnodeops.h    2010-10-21 13:09:36.000000000 +0200
8486 @@ -14,6 +14,7 @@ struct xfs_inode;
8487  struct xfs_iomap;
8488
8489
8490 +int xfs_sync_xflags(struct xfs_inode *ip);
8491  int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
8492  #define        XFS_ATTR_DMI            0x01    /* invocation from a DMI function */
8493  #define        XFS_ATTR_NONBLOCK       0x02    /* return EAGAIN if operation would block */
8494 diff -NurpP --minimal linux-2.6.36/include/asm-generic/tlb.h linux-2.6.36-vs2.3.0.36.38/include/asm-generic/tlb.h
8495 --- linux-2.6.36/include/asm-generic/tlb.h      2009-09-10 15:26:24.000000000 +0200
8496 +++ linux-2.6.36-vs2.3.0.36.38/include/asm-generic/tlb.h        2010-10-21 13:09:36.000000000 +0200
8497 @@ -14,6 +14,7 @@
8498  #define _ASM_GENERIC__TLB_H
8499
8500  #include <linux/swap.h>
8501 +#include <linux/vs_memory.h>
8502  #include <asm/pgalloc.h>
8503  #include <asm/tlbflush.h>
8504
8505 diff -NurpP --minimal linux-2.6.36/include/linux/Kbuild linux-2.6.36-vs2.3.0.36.38/include/linux/Kbuild
8506 --- linux-2.6.36/include/linux/Kbuild   2010-10-21 13:07:52.000000000 +0200
8507 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/Kbuild     2010-10-21 14:12:01.000000000 +0200
8508 @@ -15,6 +15,7 @@ header-y += netfilter_bridge/
8509  header-y += netfilter_ipv4/
8510  header-y += netfilter_ipv6/
8511  header-y += usb/
8512 +header-y += vserver/
8513  header-y += wimax/
8514
8515  objhdr-y += version.h
8516 diff -NurpP --minimal linux-2.6.36/include/linux/capability.h linux-2.6.36-vs2.3.0.36.38/include/linux/capability.h
8517 --- linux-2.6.36/include/linux/capability.h     2010-10-21 13:07:52.000000000 +0200
8518 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/capability.h       2010-10-21 13:09:36.000000000 +0200
8519 @@ -280,6 +280,7 @@ struct cpu_vfs_cap_data {
8520     arbitrary SCSI commands */
8521  /* Allow setting encryption key on loopback filesystem */
8522  /* Allow setting zone reclaim policy */
8523 +/* Allow the selection of a security context */
8524
8525  #define CAP_SYS_ADMIN        21
8526
8527 @@ -352,7 +353,13 @@ struct cpu_vfs_cap_data {
8528
8529  #define CAP_MAC_ADMIN        33
8530
8531 -#define CAP_LAST_CAP         CAP_MAC_ADMIN
8532 +/* Allow context manipulations */
8533 +/* Allow changing context info on files */
8534 +
8535 +#define CAP_CONTEXT         34
8536 +
8537 +
8538 +#define CAP_LAST_CAP         CAP_CONTEXT
8539
8540  #define cap_valid(x) ((x) >= 0 && (x) <= CAP_LAST_CAP)
8541
8542 diff -NurpP --minimal linux-2.6.36/include/linux/devpts_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/devpts_fs.h
8543 --- linux-2.6.36/include/linux/devpts_fs.h      2008-12-25 00:26:37.000000000 +0100
8544 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/devpts_fs.h        2010-10-21 13:09:36.000000000 +0200
8545 @@ -45,5 +45,4 @@ static inline void devpts_pty_kill(struc
8546
8547  #endif
8548
8549 -
8550  #endif /* _LINUX_DEVPTS_FS_H */
8551 diff -NurpP --minimal linux-2.6.36/include/linux/ext2_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/ext2_fs.h
8552 --- linux-2.6.36/include/linux/ext2_fs.h        2010-02-25 11:52:07.000000000 +0100
8553 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/ext2_fs.h  2010-10-21 13:09:36.000000000 +0200
8554 @@ -189,8 +189,12 @@ struct ext2_group_desc
8555  #define EXT2_NOTAIL_FL                 FS_NOTAIL_FL    /* file tail should not be merged */
8556  #define EXT2_DIRSYNC_FL                        FS_DIRSYNC_FL   /* dirsync behaviour (directories only) */
8557  #define EXT2_TOPDIR_FL                 FS_TOPDIR_FL    /* Top of directory hierarchies*/
8558 +#define EXT2_IXUNLINK_FL               FS_IXUNLINK_FL  /* Immutable invert on unlink */
8559  #define EXT2_RESERVED_FL               FS_RESERVED_FL  /* reserved for ext2 lib */
8560
8561 +#define EXT2_BARRIER_FL                        FS_BARRIER_FL   /* Barrier for chroot() */
8562 +#define EXT2_COW_FL                    FS_COW_FL       /* Copy on Write marker */
8563 +
8564  #define EXT2_FL_USER_VISIBLE           FS_FL_USER_VISIBLE      /* User visible flags */
8565  #define EXT2_FL_USER_MODIFIABLE                FS_FL_USER_MODIFIABLE   /* User modifiable flags */
8566
8567 @@ -274,7 +278,8 @@ struct ext2_inode {
8568                         __u16   i_pad1;
8569                         __le16  l_i_uid_high;   /* these 2 fields    */
8570                         __le16  l_i_gid_high;   /* were reserved2[0] */
8571 -                       __u32   l_i_reserved2;
8572 +                       __le16  l_i_tag;        /* Context Tag */
8573 +                       __u16   l_i_reserved2;
8574                 } linux2;
8575                 struct {
8576                         __u8    h_i_frag;       /* Fragment number */
8577 @@ -303,6 +308,7 @@ struct ext2_inode {
8578  #define i_gid_low      i_gid
8579  #define i_uid_high     osd2.linux2.l_i_uid_high
8580  #define i_gid_high     osd2.linux2.l_i_gid_high
8581 +#define i_raw_tag      osd2.linux2.l_i_tag
8582  #define i_reserved2    osd2.linux2.l_i_reserved2
8583  #endif
8584
8585 @@ -347,6 +353,7 @@ struct ext2_inode {
8586  #define EXT2_MOUNT_USRQUOTA            0x020000  /* user quota */
8587  #define EXT2_MOUNT_GRPQUOTA            0x040000  /* group quota */
8588  #define EXT2_MOUNT_RESERVATION         0x080000  /* Preallocation */
8589 +#define EXT2_MOUNT_TAGGED              (1<<24)   /* Enable Context Tags */
8590
8591
8592  #define clear_opt(o, opt)              o &= ~EXT2_MOUNT_##opt
8593 diff -NurpP --minimal linux-2.6.36/include/linux/ext3_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/ext3_fs.h
8594 --- linux-2.6.36/include/linux/ext3_fs.h        2010-10-21 13:07:53.000000000 +0200
8595 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/ext3_fs.h  2010-10-21 13:09:36.000000000 +0200
8596 @@ -173,10 +173,14 @@ struct ext3_group_desc
8597  #define EXT3_NOTAIL_FL                 0x00008000 /* file tail should not be merged */
8598  #define EXT3_DIRSYNC_FL                        0x00010000 /* dirsync behaviour (directories only) */
8599  #define EXT3_TOPDIR_FL                 0x00020000 /* Top of directory hierarchies*/
8600 +#define EXT3_IXUNLINK_FL               0x08000000 /* Immutable invert on unlink */
8601  #define EXT3_RESERVED_FL               0x80000000 /* reserved for ext3 lib */
8602
8603 -#define EXT3_FL_USER_VISIBLE           0x0003DFFF /* User visible flags */
8604 -#define EXT3_FL_USER_MODIFIABLE                0x000380FF /* User modifiable flags */
8605 +#define EXT3_BARRIER_FL                        0x04000000 /* Barrier for chroot() */
8606 +#define EXT3_COW_FL                    0x20000000 /* Copy on Write marker */
8607 +
8608 +#define EXT3_FL_USER_VISIBLE           0x0103DFFF /* User visible flags */
8609 +#define EXT3_FL_USER_MODIFIABLE                0x010380FF /* User modifiable flags */
8610
8611  /* Flags that should be inherited by new inodes from their parent. */
8612  #define EXT3_FL_INHERITED (EXT3_SECRM_FL | EXT3_UNRM_FL | EXT3_COMPR_FL |\
8613 @@ -312,7 +316,8 @@ struct ext3_inode {
8614                         __u16   i_pad1;
8615                         __le16  l_i_uid_high;   /* these 2 fields    */
8616                         __le16  l_i_gid_high;   /* were reserved2[0] */
8617 -                       __u32   l_i_reserved2;
8618 +                       __le16  l_i_tag;        /* Context Tag */
8619 +                       __u16   l_i_reserved2;
8620                 } linux2;
8621                 struct {
8622                         __u8    h_i_frag;       /* Fragment number */
8623 @@ -343,6 +348,7 @@ struct ext3_inode {
8624  #define i_gid_low      i_gid
8625  #define i_uid_high     osd2.linux2.l_i_uid_high
8626  #define i_gid_high     osd2.linux2.l_i_gid_high
8627 +#define i_raw_tag      osd2.linux2.l_i_tag
8628  #define i_reserved2    osd2.linux2.l_i_reserved2
8629
8630  #elif defined(__GNU__)
8631 @@ -405,6 +411,7 @@ struct ext3_inode {
8632  #define EXT3_MOUNT_GRPQUOTA            0x200000 /* "old" group quota */
8633  #define EXT3_MOUNT_DATA_ERR_ABORT      0x400000 /* Abort on file data write
8634                                                   * error in ordered mode */
8635 +#define EXT3_MOUNT_TAGGED              (1<<24) /* Enable Context Tags */
8636
8637  /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
8638  #ifndef _LINUX_EXT2_FS_H
8639 @@ -908,6 +915,7 @@ extern void ext3_get_inode_flags(struct
8640  extern void ext3_set_aops(struct inode *inode);
8641  extern int ext3_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
8642                        u64 start, u64 len);
8643 +extern int ext3_sync_flags(struct inode *, int, int);
8644
8645  /* ioctl.c */
8646  extern long ext3_ioctl(struct file *, unsigned int, unsigned long);
8647 diff -NurpP --minimal linux-2.6.36/include/linux/fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/fs.h
8648 --- linux-2.6.36/include/linux/fs.h     2010-10-21 13:07:53.000000000 +0200
8649 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/fs.h       2010-10-21 13:09:36.000000000 +0200
8650 @@ -208,6 +208,9 @@ struct inodes_stat_t {
8651  #define MS_KERNMOUNT   (1<<22) /* this is a kern_mount call */
8652  #define MS_I_VERSION   (1<<23) /* Update inode I_version field */
8653  #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
8654 +#define MS_TAGGED      (1<<25) /* use generic inode tagging */
8655 +#define MS_TAGID       (1<<26) /* use specific tag for this mount */
8656 +#define MS_NOTAGCHECK  (1<<27) /* don't check tags */
8657  #define MS_BORN                (1<<29)
8658  #define MS_ACTIVE      (1<<30)
8659  #define MS_NOUSER      (1<<31)
8660 @@ -235,6 +238,14 @@ struct inodes_stat_t {
8661  #define S_NOCMTIME     128     /* Do not update file c/mtime */
8662  #define S_SWAPFILE     256     /* Do not truncate: swapon got its bmaps */
8663  #define S_PRIVATE      512     /* Inode is fs-internal */
8664 +#define S_IXUNLINK     1024    /* Immutable Invert on unlink */
8665 +
8666 +/* Linux-VServer related Inode flags */
8667 +
8668 +#define V_VALID                1
8669 +#define V_XATTR                2
8670 +#define V_BARRIER      4       /* Barrier for chroot() */
8671 +#define V_COW          8       /* Copy on Write */
8672
8673  /*
8674   * Note that nosuid etc flags are inode-specific: setting some file-system
8675 @@ -257,12 +268,15 @@ struct inodes_stat_t {
8676  #define IS_DIRSYNC(inode)      (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
8677                                         ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
8678  #define IS_MANDLOCK(inode)     __IS_FLG(inode, MS_MANDLOCK)
8679 -#define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8680 -#define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
8681 +#define IS_NOATIME(inode)      __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
8682 +#define IS_I_VERSION(inode)    __IS_FLG(inode, MS_I_VERSION)
8683 +#define IS_TAGGED(inode)       __IS_FLG(inode, MS_TAGGED)
8684
8685  #define IS_NOQUOTA(inode)      ((inode)->i_flags & S_NOQUOTA)
8686  #define IS_APPEND(inode)       ((inode)->i_flags & S_APPEND)
8687  #define IS_IMMUTABLE(inode)    ((inode)->i_flags & S_IMMUTABLE)
8688 +#define IS_IXUNLINK(inode)     ((inode)->i_flags & S_IXUNLINK)
8689 +#define IS_IXORUNLINK(inode)   ((IS_IXUNLINK(inode) ? S_IMMUTABLE : 0) ^ IS_IMMUTABLE(inode))
8690  #define IS_POSIXACL(inode)     __IS_FLG(inode, MS_POSIXACL)
8691
8692  #define IS_DEADDIR(inode)      ((inode)->i_flags & S_DEAD)
8693 @@ -270,6 +284,16 @@ struct inodes_stat_t {
8694  #define IS_SWAPFILE(inode)     ((inode)->i_flags & S_SWAPFILE)
8695  #define IS_PRIVATE(inode)      ((inode)->i_flags & S_PRIVATE)
8696
8697 +#define IS_BARRIER(inode)      (S_ISDIR((inode)->i_mode) && ((inode)->i_vflags & V_BARRIER))
8698 +
8699 +#ifdef CONFIG_VSERVER_COWBL
8700 +#  define IS_COW(inode)                (IS_IXUNLINK(inode) && IS_IMMUTABLE(inode))
8701 +#  define IS_COW_LINK(inode)   (S_ISREG((inode)->i_mode) && ((inode)->i_nlink > 1))
8702 +#else
8703 +#  define IS_COW(inode)                (0)
8704 +#  define IS_COW_LINK(inode)   (0)
8705 +#endif
8706 +
8707  /* the read-only stuff doesn't really belong here, but any other place is
8708     probably as bad and I don't want to create yet another include file. */
8709
8710 @@ -353,11 +377,14 @@ struct inodes_stat_t {
8711  #define FS_TOPDIR_FL                   0x00020000 /* Top of directory hierarchies*/
8712  #define FS_EXTENT_FL                   0x00080000 /* Extents */
8713  #define FS_DIRECTIO_FL                 0x00100000 /* Use direct i/o */
8714 +#define FS_IXUNLINK_FL                 0x08000000 /* Immutable invert on unlink */
8715  #define FS_RESERVED_FL                 0x80000000 /* reserved for ext2 lib */
8716
8717 -#define FS_FL_USER_VISIBLE             0x0003DFFF /* User visible flags */
8718 -#define FS_FL_USER_MODIFIABLE          0x000380FF /* User modifiable flags */
8719 +#define FS_BARRIER_FL                  0x04000000 /* Barrier for chroot() */
8720 +#define FS_COW_FL                      0x20000000 /* Copy on Write marker */
8721
8722 +#define FS_FL_USER_VISIBLE             0x0103DFFF /* User visible flags */
8723 +#define FS_FL_USER_MODIFIABLE          0x010380FF /* User modifiable flags */
8724
8725  #define SYNC_FILE_RANGE_WAIT_BEFORE    1
8726  #define SYNC_FILE_RANGE_WRITE          2
8727 @@ -437,6 +464,7 @@ typedef void (dio_iodone_t)(struct kiocb
8728  #define ATTR_KILL_PRIV (1 << 14)
8729  #define ATTR_OPEN      (1 << 15) /* Truncating from open(O_TRUNC) */
8730  #define ATTR_TIMES_SET (1 << 16)
8731 +#define ATTR_TAG       (1 << 17)
8732
8733  /*
8734   * This is the Inode Attributes structure, used for notify_change().  It
8735 @@ -452,6 +480,7 @@ struct iattr {
8736         umode_t         ia_mode;
8737         uid_t           ia_uid;
8738         gid_t           ia_gid;
8739 +       tag_t           ia_tag;
8740         loff_t          ia_size;
8741         struct timespec ia_atime;
8742         struct timespec ia_mtime;
8743 @@ -465,6 +494,9 @@ struct iattr {
8744         struct file     *ia_file;
8745  };
8746
8747 +#define ATTR_FLAG_BARRIER      512     /* Barrier for chroot() */
8748 +#define ATTR_FLAG_IXUNLINK     1024    /* Immutable invert on unlink */
8749 +
8750  /*
8751   * Includes for diskquotas.
8752   */
8753 @@ -732,7 +764,9 @@ struct inode {
8754         unsigned int            i_nlink;
8755         uid_t                   i_uid;
8756         gid_t                   i_gid;
8757 +       tag_t                   i_tag;
8758         dev_t                   i_rdev;
8759 +       dev_t                   i_mdev;
8760         unsigned int            i_blkbits;
8761         u64                     i_version;
8762         loff_t                  i_size;
8763 @@ -774,7 +808,8 @@ struct inode {
8764         unsigned long           i_state;
8765         unsigned long           dirtied_when;   /* jiffies of first dirtying */
8766
8767 -       unsigned int            i_flags;
8768 +       unsigned short          i_flags;
8769 +       unsigned short          i_vflags;
8770
8771         atomic_t                i_writecount;
8772  #ifdef CONFIG_SECURITY
8773 @@ -862,12 +897,12 @@ static inline void i_size_write(struct i
8774
8775  static inline unsigned iminor(const struct inode *inode)
8776  {
8777 -       return MINOR(inode->i_rdev);
8778 +       return MINOR(inode->i_mdev);
8779  }
8780
8781  static inline unsigned imajor(const struct inode *inode)
8782  {
8783 -       return MAJOR(inode->i_rdev);
8784 +       return MAJOR(inode->i_mdev);
8785  }
8786
8787  extern struct block_device *I_BDEV(struct inode *inode);
8788 @@ -929,6 +964,7 @@ struct file {
8789         loff_t                  f_pos;
8790         struct fown_struct      f_owner;
8791         const struct cred       *f_cred;
8792 +       xid_t                   f_xid;
8793         struct file_ra_state    f_ra;
8794
8795         u64                     f_version;
8796 @@ -1068,6 +1104,7 @@ struct file_lock {
8797         struct file *fl_file;
8798         loff_t fl_start;
8799         loff_t fl_end;
8800 +       xid_t fl_xid;
8801
8802         struct fasync_struct *  fl_fasync; /* for lease break notifications */
8803         unsigned long fl_break_time;    /* for nonblocking lease breaks */
8804 @@ -1536,6 +1573,7 @@ struct inode_operations {
8805         ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
8806         ssize_t (*listxattr) (struct dentry *, char *, size_t);
8807         int (*removexattr) (struct dentry *, const char *);
8808 +       int (*sync_flags) (struct inode *, int, int);
8809         void (*truncate_range)(struct inode *, loff_t, loff_t);
8810         long (*fallocate)(struct inode *inode, int mode, loff_t offset,
8811                           loff_t len);
8812 @@ -1556,6 +1594,7 @@ extern ssize_t vfs_readv(struct file *,
8813                 unsigned long, loff_t *);
8814  extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
8815                 unsigned long, loff_t *);
8816 +ssize_t vfs_sendfile(struct file *, struct file *, loff_t *, size_t, loff_t);
8817
8818  struct super_operations {
8819         struct inode *(*alloc_inode)(struct super_block *sb);
8820 @@ -2344,6 +2383,7 @@ extern int dcache_dir_open(struct inode
8821  extern int dcache_dir_close(struct inode *, struct file *);
8822  extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
8823  extern int dcache_readdir(struct file *, void *, filldir_t);
8824 +extern int dcache_readdir_filter(struct file *, void *, filldir_t, int (*)(struct dentry *));
8825  extern int simple_setattr(struct dentry *, struct iattr *);
8826  extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
8827  extern int simple_statfs(struct dentry *, struct kstatfs *);
8828 diff -NurpP --minimal linux-2.6.36/include/linux/gfs2_ondisk.h linux-2.6.36-vs2.3.0.36.38/include/linux/gfs2_ondisk.h
8829 --- linux-2.6.36/include/linux/gfs2_ondisk.h    2010-07-07 18:31:55.000000000 +0200
8830 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/gfs2_ondisk.h      2010-10-21 13:09:36.000000000 +0200
8831 @@ -211,6 +211,9 @@ enum {
8832         gfs2fl_NoAtime          = 7,
8833         gfs2fl_Sync             = 8,
8834         gfs2fl_System           = 9,
8835 +       gfs2fl_IXUnlink         = 16,
8836 +       gfs2fl_Barrier          = 17,
8837 +       gfs2fl_Cow              = 18,
8838         gfs2fl_TruncInProg      = 29,
8839         gfs2fl_InheritDirectio  = 30,
8840         gfs2fl_InheritJdata     = 31,
8841 @@ -227,6 +230,9 @@ enum {
8842  #define GFS2_DIF_NOATIME               0x00000080
8843  #define GFS2_DIF_SYNC                  0x00000100
8844  #define GFS2_DIF_SYSTEM                        0x00000200 /* New in gfs2 */
8845 +#define GFS2_DIF_IXUNLINK              0x00010000
8846 +#define GFS2_DIF_BARRIER               0x00020000
8847 +#define GFS2_DIF_COW                   0x00040000
8848  #define GFS2_DIF_TRUNC_IN_PROG         0x20000000 /* New in gfs2 */
8849  #define GFS2_DIF_INHERIT_DIRECTIO      0x40000000
8850  #define GFS2_DIF_INHERIT_JDATA         0x80000000
8851 diff -NurpP --minimal linux-2.6.36/include/linux/if_tun.h linux-2.6.36-vs2.3.0.36.38/include/linux/if_tun.h
8852 --- linux-2.6.36/include/linux/if_tun.h 2010-08-02 16:52:54.000000000 +0200
8853 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/if_tun.h   2010-10-21 13:09:36.000000000 +0200
8854 @@ -53,6 +53,7 @@
8855  #define TUNDETACHFILTER _IOW('T', 214, struct sock_fprog)
8856  #define TUNGETVNETHDRSZ _IOR('T', 215, int)
8857  #define TUNSETVNETHDRSZ _IOW('T', 216, int)
8858 +#define TUNSETNID     _IOW('T', 217, int)
8859
8860  /* TUNSETIFF ifr flags */
8861  #define IFF_TUN                0x0001
8862 diff -NurpP --minimal linux-2.6.36/include/linux/init_task.h linux-2.6.36-vs2.3.0.36.38/include/linux/init_task.h
8863 --- linux-2.6.36/include/linux/init_task.h      2010-08-02 16:52:54.000000000 +0200
8864 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/init_task.h        2010-10-21 13:09:36.000000000 +0200
8865 @@ -172,6 +172,10 @@ extern struct cred init_cred;
8866         INIT_FTRACE_GRAPH                                               \
8867         INIT_TRACE_RECURSION                                            \
8868         INIT_TASK_RCU_PREEMPT(tsk)                                      \
8869 +       .xid            = 0,                                            \
8870 +       .vx_info        = NULL,                                         \
8871 +       .nid            = 0,                                            \
8872 +       .nx_info        = NULL,                                         \
8873  }
8874
8875
8876 diff -NurpP --minimal linux-2.6.36/include/linux/ipc.h linux-2.6.36-vs2.3.0.36.38/include/linux/ipc.h
8877 --- linux-2.6.36/include/linux/ipc.h    2009-12-03 20:02:55.000000000 +0100
8878 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/ipc.h      2010-10-21 13:09:36.000000000 +0200
8879 @@ -91,6 +91,7 @@ struct kern_ipc_perm
8880         key_t           key;
8881         uid_t           uid;
8882         gid_t           gid;
8883 +       xid_t           xid;
8884         uid_t           cuid;
8885         gid_t           cgid;
8886         mode_t          mode;
8887 diff -NurpP --minimal linux-2.6.36/include/linux/loop.h linux-2.6.36-vs2.3.0.36.38/include/linux/loop.h
8888 --- linux-2.6.36/include/linux/loop.h   2009-09-10 15:26:25.000000000 +0200
8889 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/loop.h     2010-10-21 13:09:36.000000000 +0200
8890 @@ -45,6 +45,7 @@ struct loop_device {
8891         struct loop_func_table *lo_encryption;
8892         __u32           lo_init[2];
8893         uid_t           lo_key_owner;   /* Who set the key */
8894 +       xid_t           lo_xid;
8895         int             (*ioctl)(struct loop_device *, int cmd,
8896                                  unsigned long arg);
8897
8898 diff -NurpP --minimal linux-2.6.36/include/linux/magic.h linux-2.6.36-vs2.3.0.36.38/include/linux/magic.h
8899 --- linux-2.6.36/include/linux/magic.h  2010-07-07 18:31:55.000000000 +0200
8900 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/magic.h    2010-10-21 13:09:36.000000000 +0200
8901 @@ -3,7 +3,7 @@
8902
8903  #define ADFS_SUPER_MAGIC       0xadf5
8904  #define AFFS_SUPER_MAGIC       0xadff
8905 -#define AFS_SUPER_MAGIC                0x5346414F
8906 +#define AFS_SUPER_MAGIC                0x5346414F
8907  #define AUTOFS_SUPER_MAGIC     0x0187
8908  #define CODA_SUPER_MAGIC       0x73757245
8909  #define CRAMFS_MAGIC           0x28cd3d45      /* some random number */
8910 @@ -38,6 +38,7 @@
8911  #define NFS_SUPER_MAGIC                0x6969
8912  #define OPENPROM_SUPER_MAGIC   0x9fa1
8913  #define PROC_SUPER_MAGIC       0x9fa0
8914 +#define DEVPTS_SUPER_MAGIC     0x1cd1
8915  #define QNX4_SUPER_MAGIC       0x002f          /* qnx4 fs detection */
8916
8917  #define REISERFS_SUPER_MAGIC   0x52654973      /* used by gcc */
8918 diff -NurpP --minimal linux-2.6.36/include/linux/major.h linux-2.6.36-vs2.3.0.36.38/include/linux/major.h
8919 --- linux-2.6.36/include/linux/major.h  2009-09-10 15:26:25.000000000 +0200
8920 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/major.h    2010-10-21 13:09:36.000000000 +0200
8921 @@ -15,6 +15,7 @@
8922  #define HD_MAJOR               IDE0_MAJOR
8923  #define PTY_SLAVE_MAJOR                3
8924  #define TTY_MAJOR              4
8925 +#define VROOT_MAJOR            4
8926  #define TTYAUX_MAJOR           5
8927  #define LP_MAJOR               6
8928  #define VCS_MAJOR              7
8929 diff -NurpP --minimal linux-2.6.36/include/linux/memcontrol.h linux-2.6.36-vs2.3.0.36.38/include/linux/memcontrol.h
8930 --- linux-2.6.36/include/linux/memcontrol.h     2010-10-21 13:07:53.000000000 +0200
8931 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/memcontrol.h       2010-10-28 19:36:06.000000000 +0200
8932 @@ -77,6 +77,13 @@ int task_in_mem_cgroup(struct task_struc
8933  extern struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page);
8934  extern struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p);
8935
8936 +extern u64 mem_cgroup_res_read_u64(struct mem_cgroup *mem, int member);
8937 +extern u64 mem_cgroup_memsw_read_u64(struct mem_cgroup *mem, int member);
8938 +
8939 +extern s64 mem_cgroup_stat_read_cache(struct mem_cgroup *mem);
8940 +extern s64 mem_cgroup_stat_read_anon(struct mem_cgroup *mem);
8941 +extern s64 mem_cgroup_stat_read_mapped(struct mem_cgroup *mem);
8942 +
8943  static inline
8944  int mm_match_cgroup(const struct mm_struct *mm, const struct mem_cgroup *cgroup)
8945  {
8946 diff -NurpP --minimal linux-2.6.36/include/linux/mm_types.h linux-2.6.36-vs2.3.0.36.38/include/linux/mm_types.h
8947 --- linux-2.6.36/include/linux/mm_types.h       2010-10-21 13:07:53.000000000 +0200
8948 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/mm_types.h 2010-10-21 13:09:36.000000000 +0200
8949 @@ -269,6 +269,7 @@ struct mm_struct {
8950
8951         /* Architecture-specific MM context */
8952         mm_context_t context;
8953 +       struct vx_info *mm_vx_info;
8954
8955         /* Swap token stuff */
8956         /*
8957 diff -NurpP --minimal linux-2.6.36/include/linux/mount.h linux-2.6.36-vs2.3.0.36.38/include/linux/mount.h
8958 --- linux-2.6.36/include/linux/mount.h  2010-10-21 13:07:53.000000000 +0200
8959 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/mount.h    2010-10-21 13:09:36.000000000 +0200
8960 @@ -46,6 +46,9 @@ struct mnt_namespace;
8961
8962  #define MNT_INTERNAL   0x4000
8963
8964 +#define MNT_TAGID      0x10000
8965 +#define MNT_NOTAG      0x20000
8966 +
8967  struct vfsmount {
8968         struct list_head mnt_hash;
8969         struct vfsmount *mnt_parent;    /* fs we are mounted on */
8970 @@ -84,6 +87,7 @@ struct vfsmount {
8971  #else
8972         int mnt_writers;
8973  #endif
8974 +       tag_t mnt_tag;                  /* tagging used for vfsmount */
8975  };
8976
8977  static inline int *get_mnt_writers_ptr(struct vfsmount *mnt)
8978 diff -NurpP --minimal linux-2.6.36/include/linux/net.h linux-2.6.36-vs2.3.0.36.38/include/linux/net.h
8979 --- linux-2.6.36/include/linux/net.h    2010-08-02 16:52:55.000000000 +0200
8980 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/net.h      2010-10-21 13:09:36.000000000 +0200
8981 @@ -71,6 +71,7 @@ struct net;
8982  #define SOCK_NOSPACE           2
8983  #define SOCK_PASSCRED          3
8984  #define SOCK_PASSSEC           4
8985 +#define SOCK_USER_SOCKET       5
8986
8987  #ifndef ARCH_HAS_SOCKET_TYPES
8988  /**
8989 diff -NurpP --minimal linux-2.6.36/include/linux/nfs_mount.h linux-2.6.36-vs2.3.0.36.38/include/linux/nfs_mount.h
8990 --- linux-2.6.36/include/linux/nfs_mount.h      2010-10-21 13:07:54.000000000 +0200
8991 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/nfs_mount.h        2010-10-21 13:09:36.000000000 +0200
8992 @@ -63,7 +63,8 @@ struct nfs_mount_data {
8993  #define NFS_MOUNT_SECFLAVOUR   0x2000  /* 5 */
8994  #define NFS_MOUNT_NORDIRPLUS   0x4000  /* 5 */
8995  #define NFS_MOUNT_UNSHARED     0x8000  /* 5 */
8996 -#define NFS_MOUNT_FLAGMASK     0xFFFF
8997 +#define NFS_MOUNT_TAGGED       0x10000 /* context tagging */
8998 +#define NFS_MOUNT_FLAGMASK     0x1FFFF
8999
9000  /* The following are for internal use only */
9001  #define NFS_MOUNT_LOOKUP_CACHE_NONEG   0x10000
9002 diff -NurpP --minimal linux-2.6.36/include/linux/nsproxy.h linux-2.6.36-vs2.3.0.36.38/include/linux/nsproxy.h
9003 --- linux-2.6.36/include/linux/nsproxy.h        2009-06-11 17:13:17.000000000 +0200
9004 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/nsproxy.h  2010-10-21 13:09:36.000000000 +0200
9005 @@ -3,6 +3,7 @@
9006
9007  #include <linux/spinlock.h>
9008  #include <linux/sched.h>
9009 +#include <linux/vserver/debug.h>
9010
9011  struct mnt_namespace;
9012  struct uts_namespace;
9013 @@ -63,22 +64,33 @@ static inline struct nsproxy *task_nspro
9014  }
9015
9016  int copy_namespaces(unsigned long flags, struct task_struct *tsk);
9017 +struct nsproxy *copy_nsproxy(struct nsproxy *orig);
9018  void exit_task_namespaces(struct task_struct *tsk);
9019  void switch_task_namespaces(struct task_struct *tsk, struct nsproxy *new);
9020  void free_nsproxy(struct nsproxy *ns);
9021  int unshare_nsproxy_namespaces(unsigned long, struct nsproxy **,
9022         struct fs_struct *);
9023
9024 -static inline void put_nsproxy(struct nsproxy *ns)
9025 +#define        get_nsproxy(n)  __get_nsproxy(n, __FILE__, __LINE__)
9026 +
9027 +static inline void __get_nsproxy(struct nsproxy *ns,
9028 +       const char *_file, int _line)
9029  {
9030 -       if (atomic_dec_and_test(&ns->count)) {
9031 -               free_nsproxy(ns);
9032 -       }
9033 +       vxlprintk(VXD_CBIT(space, 0), "get_nsproxy(%p[%u])",
9034 +               ns, atomic_read(&ns->count), _file, _line);
9035 +       atomic_inc(&ns->count);
9036  }
9037
9038 -static inline void get_nsproxy(struct nsproxy *ns)
9039 +#define        put_nsproxy(n)  __put_nsproxy(n, __FILE__, __LINE__)
9040 +
9041 +static inline void __put_nsproxy(struct nsproxy *ns,
9042 +       const char *_file, int _line)
9043  {
9044 -       atomic_inc(&ns->count);
9045 +       vxlprintk(VXD_CBIT(space, 0), "put_nsproxy(%p[%u])",
9046 +               ns, atomic_read(&ns->count), _file, _line);
9047 +       if (atomic_dec_and_test(&ns->count)) {
9048 +               free_nsproxy(ns);
9049 +       }
9050  }
9051
9052  #ifdef CONFIG_CGROUP_NS
9053 diff -NurpP --minimal linux-2.6.36/include/linux/pid.h linux-2.6.36-vs2.3.0.36.38/include/linux/pid.h
9054 --- linux-2.6.36/include/linux/pid.h    2009-03-24 14:22:43.000000000 +0100
9055 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/pid.h      2010-10-21 13:09:36.000000000 +0200
9056 @@ -8,7 +8,8 @@ enum pid_type
9057         PIDTYPE_PID,
9058         PIDTYPE_PGID,
9059         PIDTYPE_SID,
9060 -       PIDTYPE_MAX
9061 +       PIDTYPE_MAX,
9062 +       PIDTYPE_REALPID
9063  };
9064
9065  /*
9066 @@ -160,6 +161,7 @@ static inline pid_t pid_nr(struct pid *p
9067  }
9068
9069  pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns);
9070 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns);
9071  pid_t pid_vnr(struct pid *pid);
9072
9073  #define do_each_pid_task(pid, type, task)                              \
9074 diff -NurpP --minimal linux-2.6.36/include/linux/proc_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/proc_fs.h
9075 --- linux-2.6.36/include/linux/proc_fs.h        2009-12-03 20:02:56.000000000 +0100
9076 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/proc_fs.h  2010-10-21 13:09:36.000000000 +0200
9077 @@ -56,6 +56,7 @@ struct proc_dir_entry {
9078         nlink_t nlink;
9079         uid_t uid;
9080         gid_t gid;
9081 +       int vx_flags;
9082         loff_t size;
9083         const struct inode_operations *proc_iops;
9084         /*
9085 @@ -250,12 +251,18 @@ kclist_add(struct kcore_list *new, void
9086  extern void kclist_add(struct kcore_list *, void *, size_t, int type);
9087  #endif
9088
9089 +struct vx_info;
9090 +struct nx_info;
9091 +
9092  union proc_op {
9093         int (*proc_get_link)(struct inode *, struct path *);
9094         int (*proc_read)(struct task_struct *task, char *page);
9095         int (*proc_show)(struct seq_file *m,
9096                 struct pid_namespace *ns, struct pid *pid,
9097                 struct task_struct *task);
9098 +       int (*proc_vs_read)(char *page);
9099 +       int (*proc_vxi_read)(struct vx_info *vxi, char *page);
9100 +       int (*proc_nxi_read)(struct nx_info *nxi, char *page);
9101  };
9102
9103  struct ctl_table_header;
9104 @@ -263,6 +270,7 @@ struct ctl_table;
9105
9106  struct proc_inode {
9107         struct pid *pid;
9108 +       int vx_flags;
9109         int fd;
9110         union proc_op op;
9111         struct proc_dir_entry *pde;
9112 diff -NurpP --minimal linux-2.6.36/include/linux/quotaops.h linux-2.6.36-vs2.3.0.36.38/include/linux/quotaops.h
9113 --- linux-2.6.36/include/linux/quotaops.h       2010-10-21 13:07:54.000000000 +0200
9114 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/quotaops.h 2010-10-21 13:09:36.000000000 +0200
9115 @@ -8,6 +8,7 @@
9116  #define _LINUX_QUOTAOPS_
9117
9118  #include <linux/fs.h>
9119 +#include <linux/vs_dlimit.h>
9120
9121  #define DQUOT_SPACE_WARN       0x1
9122  #define DQUOT_SPACE_RESERVE    0x2
9123 @@ -205,11 +206,12 @@ static inline void dquot_drop(struct ino
9124
9125  static inline int dquot_alloc_inode(const struct inode *inode)
9126  {
9127 -       return 0;
9128 +       return dl_alloc_inode(inode);
9129  }
9130
9131  static inline void dquot_free_inode(const struct inode *inode)
9132  {
9133 +       dl_free_inode(inode);
9134  }
9135
9136  static inline int dquot_transfer(struct inode *inode, struct iattr *iattr)
9137 @@ -220,6 +222,10 @@ static inline int dquot_transfer(struct
9138  static inline int __dquot_alloc_space(struct inode *inode, qsize_t number,
9139                 int flags)
9140  {
9141 +       int ret = 0;
9142 +
9143 +       if ((ret = dl_alloc_space(inode, number)))
9144 +               return ret;
9145         if (!(flags & DQUOT_SPACE_RESERVE))
9146                 inode_add_bytes(inode, number);
9147         return 0;
9148 @@ -230,6 +236,7 @@ static inline void __dquot_free_space(st
9149  {
9150         if (!(flags & DQUOT_SPACE_RESERVE))
9151                 inode_sub_bytes(inode, number);
9152 +       dl_free_space(inode, number);
9153  }
9154
9155  static inline int dquot_claim_space_nodirty(struct inode *inode, qsize_t number)
9156 diff -NurpP --minimal linux-2.6.36/include/linux/reboot.h linux-2.6.36-vs2.3.0.36.38/include/linux/reboot.h
9157 --- linux-2.6.36/include/linux/reboot.h 2010-07-07 18:31:56.000000000 +0200
9158 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/reboot.h   2010-10-21 13:09:36.000000000 +0200
9159 @@ -33,6 +33,7 @@
9160  #define        LINUX_REBOOT_CMD_RESTART2       0xA1B2C3D4
9161  #define        LINUX_REBOOT_CMD_SW_SUSPEND     0xD000FCE2
9162  #define        LINUX_REBOOT_CMD_KEXEC          0x45584543
9163 +#define        LINUX_REBOOT_CMD_OOM            0xDEADBEEF
9164
9165
9166  #ifdef __KERNEL__
9167 diff -NurpP --minimal linux-2.6.36/include/linux/reiserfs_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/reiserfs_fs.h
9168 --- linux-2.6.36/include/linux/reiserfs_fs.h    2010-10-21 13:07:54.000000000 +0200
9169 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/reiserfs_fs.h      2010-10-21 13:09:36.000000000 +0200
9170 @@ -977,6 +977,11 @@ struct stat_data_v1 {
9171  #define REISERFS_COMPR_FL     FS_COMPR_FL
9172  #define REISERFS_NOTAIL_FL    FS_NOTAIL_FL
9173
9174 +/* unfortunately reiserfs sdattr is only 16 bit */
9175 +#define REISERFS_IXUNLINK_FL  (FS_IXUNLINK_FL >> 16)
9176 +#define REISERFS_BARRIER_FL   (FS_BARRIER_FL >> 16)
9177 +#define REISERFS_COW_FL       (FS_COW_FL >> 16)
9178 +
9179  /* persistent flags that file inherits from the parent directory */
9180  #define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL |        \
9181                                 REISERFS_SYNC_FL |      \
9182 @@ -986,6 +991,9 @@ struct stat_data_v1 {
9183                                 REISERFS_COMPR_FL |     \
9184                                 REISERFS_NOTAIL_FL )
9185
9186 +#define REISERFS_FL_USER_VISIBLE       0x80FF
9187 +#define REISERFS_FL_USER_MODIFIABLE    0x80FF
9188 +
9189  /* Stat Data on disk (reiserfs version of UFS disk inode minus the
9190     address blocks) */
9191  struct stat_data {
9192 @@ -2071,6 +2079,7 @@ static inline void reiserfs_update_sd(st
9193  void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
9194  void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
9195  int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);
9196 +int reiserfs_sync_flags(struct inode *inode, int, int);
9197
9198  /* namei.c */
9199  void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
9200 diff -NurpP --minimal linux-2.6.36/include/linux/reiserfs_fs_sb.h linux-2.6.36-vs2.3.0.36.38/include/linux/reiserfs_fs_sb.h
9201 --- linux-2.6.36/include/linux/reiserfs_fs_sb.h 2010-02-25 11:52:07.000000000 +0100
9202 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/reiserfs_fs_sb.h   2010-10-21 13:09:36.000000000 +0200
9203 @@ -476,6 +476,7 @@ enum reiserfs_mount_options {
9204         REISERFS_EXPOSE_PRIVROOT,
9205         REISERFS_BARRIER_NONE,
9206         REISERFS_BARRIER_FLUSH,
9207 +       REISERFS_TAGGED,
9208
9209         /* Actions on error */
9210         REISERFS_ERROR_PANIC,
9211 diff -NurpP --minimal linux-2.6.36/include/linux/sched.h linux-2.6.36-vs2.3.0.36.38/include/linux/sched.h
9212 --- linux-2.6.36/include/linux/sched.h  2010-10-21 13:07:54.000000000 +0200
9213 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/sched.h    2010-10-21 13:09:36.000000000 +0200
9214 @@ -1339,6 +1339,14 @@ struct task_struct {
9215  #endif
9216         seccomp_t seccomp;
9217
9218 +/* vserver context data */
9219 +       struct vx_info *vx_info;
9220 +       struct nx_info *nx_info;
9221 +
9222 +       xid_t xid;
9223 +       nid_t nid;
9224 +       tag_t tag;
9225 +
9226  /* Thread group tracking */
9227         u32 parent_exec_id;
9228         u32 self_exec_id;
9229 @@ -1573,6 +1581,11 @@ struct pid_namespace;
9230  pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
9231                         struct pid_namespace *ns);
9232
9233 +#include <linux/vserver/base.h>
9234 +#include <linux/vserver/context.h>
9235 +#include <linux/vserver/debug.h>
9236 +#include <linux/vserver/pid.h>
9237 +
9238  static inline pid_t task_pid_nr(struct task_struct *tsk)
9239  {
9240         return tsk->pid;
9241 @@ -1586,7 +1599,8 @@ static inline pid_t task_pid_nr_ns(struc
9242
9243  static inline pid_t task_pid_vnr(struct task_struct *tsk)
9244  {
9245 -       return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
9246 +       // return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
9247 +       return vx_map_pid(__task_pid_nr_ns(tsk, PIDTYPE_PID, NULL));
9248  }
9249
9250
9251 @@ -1599,7 +1613,7 @@ pid_t task_tgid_nr_ns(struct task_struct
9252
9253  static inline pid_t task_tgid_vnr(struct task_struct *tsk)
9254  {
9255 -       return pid_vnr(task_tgid(tsk));
9256 +       return vx_map_tgid(pid_vnr(task_tgid(tsk)));
9257  }
9258
9259
9260 diff -NurpP --minimal linux-2.6.36/include/linux/shmem_fs.h linux-2.6.36-vs2.3.0.36.38/include/linux/shmem_fs.h
9261 --- linux-2.6.36/include/linux/shmem_fs.h       2010-10-21 13:07:55.000000000 +0200
9262 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/shmem_fs.h 2010-10-21 13:09:36.000000000 +0200
9263 @@ -9,6 +9,9 @@
9264
9265  #define SHMEM_NR_DIRECT 16
9266
9267 +#define TMPFS_SUPER_MAGIC      0x01021994
9268 +
9269 +
9270  struct shmem_inode_info {
9271         spinlock_t              lock;
9272         unsigned long           flags;
9273 diff -NurpP --minimal linux-2.6.36/include/linux/stat.h linux-2.6.36-vs2.3.0.36.38/include/linux/stat.h
9274 --- linux-2.6.36/include/linux/stat.h   2008-12-25 00:26:37.000000000 +0100
9275 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/stat.h     2010-10-21 13:09:36.000000000 +0200
9276 @@ -66,6 +66,7 @@ struct kstat {
9277         unsigned int    nlink;
9278         uid_t           uid;
9279         gid_t           gid;
9280 +       tag_t           tag;
9281         dev_t           rdev;
9282         loff_t          size;
9283         struct timespec  atime;
9284 diff -NurpP --minimal linux-2.6.36/include/linux/sunrpc/auth.h linux-2.6.36-vs2.3.0.36.38/include/linux/sunrpc/auth.h
9285 --- linux-2.6.36/include/linux/sunrpc/auth.h    2010-10-21 13:07:55.000000000 +0200
9286 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/sunrpc/auth.h      2010-10-21 13:09:36.000000000 +0200
9287 @@ -25,6 +25,7 @@
9288  struct auth_cred {
9289         uid_t   uid;
9290         gid_t   gid;
9291 +       tag_t   tag;
9292         struct group_info *group_info;
9293         unsigned char machine_cred : 1;
9294  };
9295 diff -NurpP --minimal linux-2.6.36/include/linux/sunrpc/clnt.h linux-2.6.36-vs2.3.0.36.38/include/linux/sunrpc/clnt.h
9296 --- linux-2.6.36/include/linux/sunrpc/clnt.h    2010-10-21 13:07:55.000000000 +0200
9297 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/sunrpc/clnt.h      2010-10-21 13:09:36.000000000 +0200
9298 @@ -49,7 +49,8 @@ struct rpc_clnt {
9299         unsigned int            cl_softrtry : 1,/* soft timeouts */
9300                                 cl_discrtry : 1,/* disconnect before retry */
9301                                 cl_autobind : 1,/* use getport() */
9302 -                               cl_chatty   : 1;/* be verbose */
9303 +                               cl_chatty   : 1,/* be verbose */
9304 +                               cl_tag      : 1;/* context tagging */
9305
9306         struct rpc_rtt *        cl_rtt;         /* RTO estimator data */
9307         const struct rpc_timeout *cl_timeout;   /* Timeout strategy */
9308 diff -NurpP --minimal linux-2.6.36/include/linux/syscalls.h linux-2.6.36-vs2.3.0.36.38/include/linux/syscalls.h
9309 --- linux-2.6.36/include/linux/syscalls.h       2010-10-21 13:07:55.000000000 +0200
9310 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/syscalls.h 2010-10-21 13:09:36.000000000 +0200
9311 @@ -478,6 +478,8 @@ asmlinkage long sys_symlink(const char _
9312  asmlinkage long sys_unlink(const char __user *pathname);
9313  asmlinkage long sys_rename(const char __user *oldname,
9314                                 const char __user *newname);
9315 +asmlinkage long sys_copyfile(const char __user *from, const char __user *to,
9316 +                               umode_t mode);
9317  asmlinkage long sys_chmod(const char __user *filename, mode_t mode);
9318  asmlinkage long sys_fchmod(unsigned int fd, mode_t mode);
9319
9320 diff -NurpP --minimal linux-2.6.36/include/linux/sysctl.h linux-2.6.36-vs2.3.0.36.38/include/linux/sysctl.h
9321 --- linux-2.6.36/include/linux/sysctl.h 2010-08-02 16:52:56.000000000 +0200
9322 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/sysctl.h   2010-10-21 13:09:36.000000000 +0200
9323 @@ -60,6 +60,7 @@ enum
9324         CTL_ABI=9,              /* Binary emulation */
9325         CTL_CPU=10,             /* CPU stuff (speed scaling, etc) */
9326         CTL_ARLAN=254,          /* arlan wireless driver */
9327 +       CTL_VSERVER=4242,       /* Linux-VServer debug */
9328         CTL_S390DBF=5677,       /* s390 debug */
9329         CTL_SUNRPC=7249,        /* sunrpc debug */
9330         CTL_PM=9899,            /* frv power management */
9331 @@ -94,6 +95,7 @@ enum
9332
9333         KERN_PANIC=15,          /* int: panic timeout */
9334         KERN_REALROOTDEV=16,    /* real root device to mount after initrd */
9335 +       KERN_VSHELPER=17,       /* string: path to vshelper policy agent */
9336
9337         KERN_SPARC_REBOOT=21,   /* reboot command on Sparc */
9338         KERN_CTLALTDEL=22,      /* int: allow ctl-alt-del to reboot */
9339 diff -NurpP --minimal linux-2.6.36/include/linux/sysfs.h linux-2.6.36-vs2.3.0.36.38/include/linux/sysfs.h
9340 --- linux-2.6.36/include/linux/sysfs.h  2010-10-21 13:07:55.000000000 +0200
9341 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/sysfs.h    2010-10-21 13:09:36.000000000 +0200
9342 @@ -19,6 +19,8 @@
9343  #include <linux/kobject_ns.h>
9344  #include <asm/atomic.h>
9345
9346 +#define SYSFS_SUPER_MAGIC      0x62656572
9347 +
9348  struct kobject;
9349  struct module;
9350  enum kobj_ns_type;
9351 diff -NurpP --minimal linux-2.6.36/include/linux/time.h linux-2.6.36-vs2.3.0.36.38/include/linux/time.h
9352 --- linux-2.6.36/include/linux/time.h   2010-10-21 13:07:55.000000000 +0200
9353 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/time.h     2010-10-21 13:09:36.000000000 +0200
9354 @@ -252,6 +252,9 @@ static __always_inline void timespec_add
9355         a->tv_sec += __iter_div_u64_rem(a->tv_nsec + ns, NSEC_PER_SEC, &ns);
9356         a->tv_nsec = ns;
9357  }
9358 +
9359 +#include <linux/vs_time.h>
9360 +
9361  #endif /* __KERNEL__ */
9362
9363  #define NFDBITS                        __NFDBITS
9364 diff -NurpP --minimal linux-2.6.36/include/linux/types.h linux-2.6.36-vs2.3.0.36.38/include/linux/types.h
9365 --- linux-2.6.36/include/linux/types.h  2010-10-21 13:07:55.000000000 +0200
9366 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/types.h    2010-10-21 13:09:36.000000000 +0200
9367 @@ -40,6 +40,9 @@ typedef __kernel_uid32_t      uid_t;
9368  typedef __kernel_gid32_t       gid_t;
9369  typedef __kernel_uid16_t        uid16_t;
9370  typedef __kernel_gid16_t        gid16_t;
9371 +typedef unsigned int           xid_t;
9372 +typedef unsigned int           nid_t;
9373 +typedef unsigned int           tag_t;
9374
9375  typedef unsigned long          uintptr_t;
9376
9377 diff -NurpP --minimal linux-2.6.36/include/linux/vroot.h linux-2.6.36-vs2.3.0.36.38/include/linux/vroot.h
9378 --- linux-2.6.36/include/linux/vroot.h  1970-01-01 01:00:00.000000000 +0100
9379 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vroot.h    2010-10-21 13:09:36.000000000 +0200
9380 @@ -0,0 +1,51 @@
9381 +
9382 +/*
9383 + * include/linux/vroot.h
9384 + *
9385 + * written by Herbert Pötzl, 9/11/2002
9386 + * ported to 2.6 by Herbert Pötzl, 30/12/2004
9387 + *
9388 + * Copyright (C) 2002-2007 by Herbert Pötzl.
9389 + * Redistribution of this file is permitted under the
9390 + * GNU General Public License.
9391 + */
9392 +
9393 +#ifndef _LINUX_VROOT_H
9394 +#define _LINUX_VROOT_H
9395 +
9396 +
9397 +#ifdef __KERNEL__
9398 +
9399 +/* Possible states of device */
9400 +enum {
9401 +       Vr_unbound,
9402 +       Vr_bound,
9403 +};
9404 +
9405 +struct vroot_device {
9406 +       int             vr_number;
9407 +       int             vr_refcnt;
9408 +
9409 +       struct semaphore        vr_ctl_mutex;
9410 +       struct block_device    *vr_device;
9411 +       int                     vr_state;
9412 +};
9413 +
9414 +
9415 +typedef struct block_device *(vroot_grb_func)(struct block_device *);
9416 +
9417 +extern int register_vroot_grb(vroot_grb_func *);
9418 +extern int unregister_vroot_grb(vroot_grb_func *);
9419 +
9420 +#endif /* __KERNEL__ */
9421 +
9422 +#define MAX_VROOT_DEFAULT      8
9423 +
9424 +/*
9425 + * IOCTL commands --- we will commandeer 0x56 ('V')
9426 + */
9427 +
9428 +#define VROOT_SET_DEV          0x5600
9429 +#define VROOT_CLR_DEV          0x5601
9430 +
9431 +#endif /* _LINUX_VROOT_H */
9432 diff -NurpP --minimal linux-2.6.36/include/linux/vs_base.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_base.h
9433 --- linux-2.6.36/include/linux/vs_base.h        1970-01-01 01:00:00.000000000 +0100
9434 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_base.h  2010-10-21 13:09:36.000000000 +0200
9435 @@ -0,0 +1,10 @@
9436 +#ifndef _VS_BASE_H
9437 +#define _VS_BASE_H
9438 +
9439 +#include "vserver/base.h"
9440 +#include "vserver/check.h"
9441 +#include "vserver/debug.h"
9442 +
9443 +#else
9444 +#warning duplicate inclusion
9445 +#endif
9446 diff -NurpP --minimal linux-2.6.36/include/linux/vs_context.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_context.h
9447 --- linux-2.6.36/include/linux/vs_context.h     1970-01-01 01:00:00.000000000 +0100
9448 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_context.h       2010-10-21 13:09:36.000000000 +0200
9449 @@ -0,0 +1,242 @@
9450 +#ifndef _VS_CONTEXT_H
9451 +#define _VS_CONTEXT_H
9452 +
9453 +#include "vserver/base.h"
9454 +#include "vserver/check.h"
9455 +#include "vserver/context.h"
9456 +#include "vserver/history.h"
9457 +#include "vserver/debug.h"
9458 +
9459 +#include <linux/sched.h>
9460 +
9461 +
9462 +#define get_vx_info(i) __get_vx_info(i, __FILE__, __LINE__, __HERE__)
9463 +
9464 +static inline struct vx_info *__get_vx_info(struct vx_info *vxi,
9465 +       const char *_file, int _line, void *_here)
9466 +{
9467 +       if (!vxi)
9468 +               return NULL;
9469 +
9470 +       vxlprintk(VXD_CBIT(xid, 2), "get_vx_info(%p[#%d.%d])",
9471 +               vxi, vxi ? vxi->vx_id : 0,
9472 +               vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9473 +               _file, _line);
9474 +       __vxh_get_vx_info(vxi, _here);
9475 +
9476 +       atomic_inc(&vxi->vx_usecnt);
9477 +       return vxi;
9478 +}
9479 +
9480 +
9481 +extern void free_vx_info(struct vx_info *);
9482 +
9483 +#define put_vx_info(i) __put_vx_info(i, __FILE__, __LINE__, __HERE__)
9484 +
9485 +static inline void __put_vx_info(struct vx_info *vxi,
9486 +       const char *_file, int _line, void *_here)
9487 +{
9488 +       if (!vxi)
9489 +               return;
9490 +
9491 +       vxlprintk(VXD_CBIT(xid, 2), "put_vx_info(%p[#%d.%d])",
9492 +               vxi, vxi ? vxi->vx_id : 0,
9493 +               vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9494 +               _file, _line);
9495 +       __vxh_put_vx_info(vxi, _here);
9496 +
9497 +       if (atomic_dec_and_test(&vxi->vx_usecnt))
9498 +               free_vx_info(vxi);
9499 +}
9500 +
9501 +
9502 +#define init_vx_info(p, i) \
9503 +       __init_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9504 +
9505 +static inline void __init_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9506 +       const char *_file, int _line, void *_here)
9507 +{
9508 +       if (vxi) {
9509 +               vxlprintk(VXD_CBIT(xid, 3),
9510 +                       "init_vx_info(%p[#%d.%d])",
9511 +                       vxi, vxi ? vxi->vx_id : 0,
9512 +                       vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9513 +                       _file, _line);
9514 +               __vxh_init_vx_info(vxi, vxp, _here);
9515 +
9516 +               atomic_inc(&vxi->vx_usecnt);
9517 +       }
9518 +       *vxp = vxi;
9519 +}
9520 +
9521 +
9522 +#define set_vx_info(p, i) \
9523 +       __set_vx_info(p, i, __FILE__, __LINE__, __HERE__)
9524 +
9525 +static inline void __set_vx_info(struct vx_info **vxp, struct vx_info *vxi,
9526 +       const char *_file, int _line, void *_here)
9527 +{
9528 +       struct vx_info *vxo;
9529 +
9530 +       if (!vxi)
9531 +               return;
9532 +
9533 +       vxlprintk(VXD_CBIT(xid, 3), "set_vx_info(%p[#%d.%d])",
9534 +               vxi, vxi ? vxi->vx_id : 0,
9535 +               vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9536 +               _file, _line);
9537 +       __vxh_set_vx_info(vxi, vxp, _here);
9538 +
9539 +       atomic_inc(&vxi->vx_usecnt);
9540 +       vxo = xchg(vxp, vxi);
9541 +       BUG_ON(vxo);
9542 +}
9543 +
9544 +
9545 +#define clr_vx_info(p) __clr_vx_info(p, __FILE__, __LINE__, __HERE__)
9546 +
9547 +static inline void __clr_vx_info(struct vx_info **vxp,
9548 +       const char *_file, int _line, void *_here)
9549 +{
9550 +       struct vx_info *vxo;
9551 +
9552 +       vxo = xchg(vxp, NULL);
9553 +       if (!vxo)
9554 +               return;
9555 +
9556 +       vxlprintk(VXD_CBIT(xid, 3), "clr_vx_info(%p[#%d.%d])",
9557 +               vxo, vxo ? vxo->vx_id : 0,
9558 +               vxo ? atomic_read(&vxo->vx_usecnt) : 0,
9559 +               _file, _line);
9560 +       __vxh_clr_vx_info(vxo, vxp, _here);
9561 +
9562 +       if (atomic_dec_and_test(&vxo->vx_usecnt))
9563 +               free_vx_info(vxo);
9564 +}
9565 +
9566 +
9567 +#define claim_vx_info(v, p) \
9568 +       __claim_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9569 +
9570 +static inline void __claim_vx_info(struct vx_info *vxi,
9571 +       struct task_struct *task,
9572 +       const char *_file, int _line, void *_here)
9573 +{
9574 +       vxlprintk(VXD_CBIT(xid, 3), "claim_vx_info(%p[#%d.%d.%d]) %p",
9575 +               vxi, vxi ? vxi->vx_id : 0,
9576 +               vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9577 +               vxi ? atomic_read(&vxi->vx_tasks) : 0,
9578 +               task, _file, _line);
9579 +       __vxh_claim_vx_info(vxi, task, _here);
9580 +
9581 +       atomic_inc(&vxi->vx_tasks);
9582 +}
9583 +
9584 +
9585 +extern void unhash_vx_info(struct vx_info *);
9586 +
9587 +#define release_vx_info(v, p) \
9588 +       __release_vx_info(v, p, __FILE__, __LINE__, __HERE__)
9589 +
9590 +static inline void __release_vx_info(struct vx_info *vxi,
9591 +       struct task_struct *task,
9592 +       const char *_file, int _line, void *_here)
9593 +{
9594 +       vxlprintk(VXD_CBIT(xid, 3), "release_vx_info(%p[#%d.%d.%d]) %p",
9595 +               vxi, vxi ? vxi->vx_id : 0,
9596 +               vxi ? atomic_read(&vxi->vx_usecnt) : 0,
9597 +               vxi ? atomic_read(&vxi->vx_tasks) : 0,
9598 +               task, _file, _line);
9599 +       __vxh_release_vx_info(vxi, task, _here);
9600 +
9601 +       might_sleep();
9602 +
9603 +       if (atomic_dec_and_test(&vxi->vx_tasks))
9604 +               unhash_vx_info(vxi);
9605 +}
9606 +
9607 +
9608 +#define task_get_vx_info(p) \
9609 +       __task_get_vx_info(p, __FILE__, __LINE__, __HERE__)
9610 +
9611 +static inline struct vx_info *__task_get_vx_info(struct task_struct *p,
9612 +       const char *_file, int _line, void *_here)
9613 +{
9614 +       struct vx_info *vxi;
9615 +
9616 +       task_lock(p);
9617 +       vxlprintk(VXD_CBIT(xid, 5), "task_get_vx_info(%p)",
9618 +               p, _file, _line);
9619 +       vxi = __get_vx_info(p->vx_info, _file, _line, _here);
9620 +       task_unlock(p);
9621 +       return vxi;
9622 +}
9623 +
9624 +
9625 +static inline void __wakeup_vx_info(struct vx_info *vxi)
9626 +{
9627 +       if (waitqueue_active(&vxi->vx_wait))
9628 +               wake_up_interruptible(&vxi->vx_wait);
9629 +}
9630 +
9631 +
9632 +#define enter_vx_info(v, s) __enter_vx_info(v, s, __FILE__, __LINE__)
9633 +
9634 +static inline void __enter_vx_info(struct vx_info *vxi,
9635 +       struct vx_info_save *vxis, const char *_file, int _line)
9636 +{
9637 +       vxlprintk(VXD_CBIT(xid, 5), "enter_vx_info(%p[#%d],%p) %p[#%d,%p]",
9638 +               vxi, vxi ? vxi->vx_id : 0, vxis, current,
9639 +               current->xid, current->vx_info, _file, _line);
9640 +       vxis->vxi = xchg(&current->vx_info, vxi);
9641 +       vxis->xid = current->xid;
9642 +       current->xid = vxi ? vxi->vx_id : 0;
9643 +}
9644 +
9645 +#define leave_vx_info(s) __leave_vx_info(s, __FILE__, __LINE__)
9646 +
9647 +static inline void __leave_vx_info(struct vx_info_save *vxis,
9648 +       const char *_file, int _line)
9649 +{
9650 +       vxlprintk(VXD_CBIT(xid, 5), "leave_vx_info(%p[#%d,%p]) %p[#%d,%p]",
9651 +               vxis, vxis->xid, vxis->vxi, current,
9652 +               current->xid, current->vx_info, _file, _line);
9653 +       (void)xchg(&current->vx_info, vxis->vxi);
9654 +       current->xid = vxis->xid;
9655 +}
9656 +
9657 +
9658 +static inline void __enter_vx_admin(struct vx_info_save *vxis)
9659 +{
9660 +       vxis->vxi = xchg(&current->vx_info, NULL);
9661 +       vxis->xid = xchg(&current->xid, (xid_t)0);
9662 +}
9663 +
9664 +static inline void __leave_vx_admin(struct vx_info_save *vxis)
9665 +{
9666 +       (void)xchg(&current->xid, vxis->xid);
9667 +       (void)xchg(&current->vx_info, vxis->vxi);
9668 +}
9669 +
9670 +#define task_is_init(p) \
9671 +       __task_is_init(p, __FILE__, __LINE__, __HERE__)
9672 +
9673 +static inline int __task_is_init(struct task_struct *p,
9674 +       const char *_file, int _line, void *_here)
9675 +{
9676 +       int is_init = is_global_init(p);
9677 +
9678 +       task_lock(p);
9679 +       if (p->vx_info)
9680 +               is_init = p->vx_info->vx_initpid == p->pid;
9681 +       task_unlock(p);
9682 +       return is_init;
9683 +}
9684 +
9685 +extern void exit_vx_info(struct task_struct *, int);
9686 +extern void exit_vx_info_early(struct task_struct *, int);
9687 +
9688 +
9689 +#else
9690 +#warning duplicate inclusion
9691 +#endif
9692 diff -NurpP --minimal linux-2.6.36/include/linux/vs_cowbl.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_cowbl.h
9693 --- linux-2.6.36/include/linux/vs_cowbl.h       1970-01-01 01:00:00.000000000 +0100
9694 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_cowbl.h 2010-10-21 13:09:36.000000000 +0200
9695 @@ -0,0 +1,48 @@
9696 +#ifndef _VS_COWBL_H
9697 +#define _VS_COWBL_H
9698 +
9699 +#include <linux/fs.h>
9700 +#include <linux/dcache.h>
9701 +#include <linux/namei.h>
9702 +#include <linux/slab.h>
9703 +
9704 +extern struct dentry *cow_break_link(const char *pathname);
9705 +
9706 +static inline int cow_check_and_break(struct path *path)
9707 +{
9708 +       struct inode *inode = path->dentry->d_inode;
9709 +       int error = 0;
9710 +
9711 +       /* do we need this check? */
9712 +       if (IS_RDONLY(inode))
9713 +               return -EROFS;
9714 +
9715 +       if (IS_COW(inode)) {
9716 +               if (IS_COW_LINK(inode)) {
9717 +                       struct dentry *new_dentry, *old_dentry = path->dentry;
9718 +                       char *pp, *buf;
9719 +
9720 +                       buf = kmalloc(PATH_MAX, GFP_KERNEL);
9721 +                       if (!buf) {
9722 +                               return -ENOMEM;
9723 +                       }
9724 +                       pp = d_path(path, buf, PATH_MAX);
9725 +                       new_dentry = cow_break_link(pp);
9726 +                       kfree(buf);
9727 +                       if (!IS_ERR(new_dentry)) {
9728 +                               path->dentry = new_dentry;
9729 +                               dput(old_dentry);
9730 +                       } else
9731 +                               error = PTR_ERR(new_dentry);
9732 +               } else {
9733 +                       inode->i_flags &= ~(S_IXUNLINK | S_IMMUTABLE);
9734 +                       inode->i_ctime = CURRENT_TIME;
9735 +                       mark_inode_dirty(inode);
9736 +               }
9737 +       }
9738 +       return error;
9739 +}
9740 +
9741 +#else
9742 +#warning duplicate inclusion
9743 +#endif
9744 diff -NurpP --minimal linux-2.6.36/include/linux/vs_cvirt.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_cvirt.h
9745 --- linux-2.6.36/include/linux/vs_cvirt.h       1970-01-01 01:00:00.000000000 +0100
9746 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_cvirt.h 2010-10-21 13:09:36.000000000 +0200
9747 @@ -0,0 +1,50 @@
9748 +#ifndef _VS_CVIRT_H
9749 +#define _VS_CVIRT_H
9750 +
9751 +#include "vserver/cvirt.h"
9752 +#include "vserver/context.h"
9753 +#include "vserver/base.h"
9754 +#include "vserver/check.h"
9755 +#include "vserver/debug.h"
9756 +
9757 +
9758 +static inline void vx_activate_task(struct task_struct *p)
9759 +{
9760 +       struct vx_info *vxi;
9761 +
9762 +       if ((vxi = p->vx_info)) {
9763 +               vx_update_load(vxi);
9764 +               atomic_inc(&vxi->cvirt.nr_running);
9765 +       }
9766 +}
9767 +
9768 +static inline void vx_deactivate_task(struct task_struct *p)
9769 +{
9770 +       struct vx_info *vxi;
9771 +
9772 +       if ((vxi = p->vx_info)) {
9773 +               vx_update_load(vxi);
9774 +               atomic_dec(&vxi->cvirt.nr_running);
9775 +       }
9776 +}
9777 +
9778 +static inline void vx_uninterruptible_inc(struct task_struct *p)
9779 +{
9780 +       struct vx_info *vxi;
9781 +
9782 +       if ((vxi = p->vx_info))
9783 +               atomic_inc(&vxi->cvirt.nr_uninterruptible);
9784 +}
9785 +
9786 +static inline void vx_uninterruptible_dec(struct task_struct *p)
9787 +{
9788 +       struct vx_info *vxi;
9789 +
9790 +       if ((vxi = p->vx_info))
9791 +               atomic_dec(&vxi->cvirt.nr_uninterruptible);
9792 +}
9793 +
9794 +
9795 +#else
9796 +#warning duplicate inclusion
9797 +#endif
9798 diff -NurpP --minimal linux-2.6.36/include/linux/vs_device.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_device.h
9799 --- linux-2.6.36/include/linux/vs_device.h      1970-01-01 01:00:00.000000000 +0100
9800 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_device.h        2010-10-21 13:09:36.000000000 +0200
9801 @@ -0,0 +1,45 @@
9802 +#ifndef _VS_DEVICE_H
9803 +#define _VS_DEVICE_H
9804 +
9805 +#include "vserver/base.h"
9806 +#include "vserver/device.h"
9807 +#include "vserver/debug.h"
9808 +
9809 +
9810 +#ifdef CONFIG_VSERVER_DEVICE
9811 +
9812 +int vs_map_device(struct vx_info *, dev_t, dev_t *, umode_t);
9813 +
9814 +#define vs_device_perm(v, d, m, p) \
9815 +       ((vs_map_device(current_vx_info(), d, NULL, m) & (p)) == (p))
9816 +
9817 +#else
9818 +
9819 +static inline
9820 +int vs_map_device(struct vx_info *vxi,
9821 +       dev_t device, dev_t *target, umode_t mode)
9822 +{
9823 +       if (target)
9824 +               *target = device;
9825 +       return ~0;
9826 +}
9827 +
9828 +#define vs_device_perm(v, d, m, p) ((p) == (p))
9829 +
9830 +#endif
9831 +
9832 +
9833 +#define vs_map_chrdev(d, t, p) \
9834 +       ((vs_map_device(current_vx_info(), d, t, S_IFCHR) & (p)) == (p))
9835 +#define vs_map_blkdev(d, t, p) \
9836 +       ((vs_map_device(current_vx_info(), d, t, S_IFBLK) & (p)) == (p))
9837 +
9838 +#define vs_chrdev_perm(d, p) \
9839 +       vs_device_perm(current_vx_info(), d, S_IFCHR, p)
9840 +#define vs_blkdev_perm(d, p) \
9841 +       vs_device_perm(current_vx_info(), d, S_IFBLK, p)
9842 +
9843 +
9844 +#else
9845 +#warning duplicate inclusion
9846 +#endif
9847 diff -NurpP --minimal linux-2.6.36/include/linux/vs_dlimit.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_dlimit.h
9848 --- linux-2.6.36/include/linux/vs_dlimit.h      1970-01-01 01:00:00.000000000 +0100
9849 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_dlimit.h        2010-10-21 13:09:36.000000000 +0200
9850 @@ -0,0 +1,215 @@
9851 +#ifndef _VS_DLIMIT_H
9852 +#define _VS_DLIMIT_H
9853 +
9854 +#include <linux/fs.h>
9855 +
9856 +#include "vserver/dlimit.h"
9857 +#include "vserver/base.h"
9858 +#include "vserver/debug.h"
9859 +
9860 +
9861 +#define get_dl_info(i) __get_dl_info(i, __FILE__, __LINE__)
9862 +
9863 +static inline struct dl_info *__get_dl_info(struct dl_info *dli,
9864 +       const char *_file, int _line)
9865 +{
9866 +       if (!dli)
9867 +               return NULL;
9868 +       vxlprintk(VXD_CBIT(dlim, 4), "get_dl_info(%p[#%d.%d])",
9869 +               dli, dli ? dli->dl_tag : 0,
9870 +               dli ? atomic_read(&dli->dl_usecnt) : 0,
9871 +               _file, _line);
9872 +       atomic_inc(&dli->dl_usecnt);
9873 +       return dli;
9874 +}
9875 +
9876 +
9877 +#define free_dl_info(i) \
9878 +       call_rcu(&(i)->dl_rcu, rcu_free_dl_info)
9879 +
9880 +#define put_dl_info(i) __put_dl_info(i, __FILE__, __LINE__)
9881 +
9882 +static inline void __put_dl_info(struct dl_info *dli,
9883 +       const char *_file, int _line)
9884 +{
9885 +       if (!dli)
9886 +               return;
9887 +       vxlprintk(VXD_CBIT(dlim, 4), "put_dl_info(%p[#%d.%d])",
9888 +               dli, dli ? dli->dl_tag : 0,
9889 +               dli ? atomic_read(&dli->dl_usecnt) : 0,
9890 +               _file, _line);
9891 +       if (atomic_dec_and_test(&dli->dl_usecnt))
9892 +               free_dl_info(dli);
9893 +}
9894 +
9895 +
9896 +#define __dlimit_char(d)       ((d) ? '*' : ' ')
9897 +
9898 +static inline int __dl_alloc_space(struct super_block *sb,
9899 +       tag_t tag, dlsize_t nr, const char *file, int line)
9900 +{
9901 +       struct dl_info *dli = NULL;
9902 +       int ret = 0;
9903 +
9904 +       if (nr == 0)
9905 +               goto out;
9906 +       dli = locate_dl_info(sb, tag);
9907 +       if (!dli)
9908 +               goto out;
9909 +
9910 +       spin_lock(&dli->dl_lock);
9911 +       ret = (dli->dl_space_used + nr > dli->dl_space_total);
9912 +       if (!ret)
9913 +               dli->dl_space_used += nr;
9914 +       spin_unlock(&dli->dl_lock);
9915 +       put_dl_info(dli);
9916 +out:
9917 +       vxlprintk(VXD_CBIT(dlim, 1),
9918 +               "ALLOC (%p,#%d)%c %lld bytes (%d)",
9919 +               sb, tag, __dlimit_char(dli), (long long)nr,
9920 +               ret, file, line);
9921 +       return ret ? -ENOSPC : 0;
9922 +}
9923 +
9924 +static inline void __dl_free_space(struct super_block *sb,
9925 +       tag_t tag, dlsize_t nr, const char *_file, int _line)
9926 +{
9927 +       struct dl_info *dli = NULL;
9928 +
9929 +       if (nr == 0)
9930 +               goto out;
9931 +       dli = locate_dl_info(sb, tag);
9932 +       if (!dli)
9933 +               goto out;
9934 +
9935 +       spin_lock(&dli->dl_lock);
9936 +       if (dli->dl_space_used > nr)
9937 +               dli->dl_space_used -= nr;
9938 +       else
9939 +               dli->dl_space_used = 0;
9940 +       spin_unlock(&dli->dl_lock);
9941 +       put_dl_info(dli);
9942 +out:
9943 +       vxlprintk(VXD_CBIT(dlim, 1),
9944 +               "FREE  (%p,#%d)%c %lld bytes",
9945 +               sb, tag, __dlimit_char(dli), (long long)nr,
9946 +               _file, _line);
9947 +}
9948 +
9949 +static inline int __dl_alloc_inode(struct super_block *sb,
9950 +       tag_t tag, const char *_file, int _line)
9951 +{
9952 +       struct dl_info *dli;
9953 +       int ret = 0;
9954 +
9955 +       dli = locate_dl_info(sb, tag);
9956 +       if (!dli)
9957 +               goto out;
9958 +
9959 +       spin_lock(&dli->dl_lock);
9960 +       dli->dl_inodes_used++;
9961 +       ret = (dli->dl_inodes_used > dli->dl_inodes_total);
9962 +       spin_unlock(&dli->dl_lock);
9963 +       put_dl_info(dli);
9964 +out:
9965 +       vxlprintk(VXD_CBIT(dlim, 0),
9966 +               "ALLOC (%p,#%d)%c inode (%d)",
9967 +               sb, tag, __dlimit_char(dli), ret, _file, _line);
9968 +       return ret ? -ENOSPC : 0;
9969 +}
9970 +
9971 +static inline void __dl_free_inode(struct super_block *sb,
9972 +       tag_t tag, const char *_file, int _line)
9973 +{
9974 +       struct dl_info *dli;
9975 +
9976 +       dli = locate_dl_info(sb, tag);
9977 +       if (!dli)
9978 +               goto out;
9979 +
9980 +       spin_lock(&dli->dl_lock);
9981 +       if (dli->dl_inodes_used > 1)
9982 +               dli->dl_inodes_used--;
9983 +       else
9984 +               dli->dl_inodes_used = 0;
9985 +       spin_unlock(&dli->dl_lock);
9986 +       put_dl_info(dli);
9987 +out:
9988 +       vxlprintk(VXD_CBIT(dlim, 0),
9989 +               "FREE  (%p,#%d)%c inode",
9990 +               sb, tag, __dlimit_char(dli), _file, _line);
9991 +}
9992 +
9993 +static inline void __dl_adjust_block(struct super_block *sb, tag_t tag,
9994 +       unsigned long long *free_blocks, unsigned long long *root_blocks,
9995 +       const char *_file, int _line)
9996 +{
9997 +       struct dl_info *dli;
9998 +       uint64_t broot, bfree;
9999 +
10000 +       dli = locate_dl_info(sb, tag);
10001 +       if (!dli)
10002 +               return;
10003 +
10004 +       spin_lock(&dli->dl_lock);
10005 +       broot = (dli->dl_space_total -
10006 +               (dli->dl_space_total >> 10) * dli->dl_nrlmult)
10007 +               >> sb->s_blocksize_bits;
10008 +       bfree = (dli->dl_space_total - dli->dl_space_used)
10009 +                       >> sb->s_blocksize_bits;
10010 +       spin_unlock(&dli->dl_lock);
10011 +
10012 +       vxlprintk(VXD_CBIT(dlim, 2),
10013 +               "ADJUST: %lld,%lld on %lld,%lld [mult=%d]",
10014 +               (long long)bfree, (long long)broot,
10015 +               *free_blocks, *root_blocks, dli->dl_nrlmult,
10016 +               _file, _line);
10017 +       if (free_blocks) {
10018 +               if (*free_blocks > bfree)
10019 +                       *free_blocks = bfree;
10020 +       }
10021 +       if (root_blocks) {
10022 +               if (*root_blocks > broot)
10023 +                       *root_blocks = broot;
10024 +       }
10025 +       put_dl_info(dli);
10026 +}
10027 +
10028 +#define dl_prealloc_space(in, bytes) \
10029 +       __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10030 +               __FILE__, __LINE__ )
10031 +
10032 +#define dl_alloc_space(in, bytes) \
10033 +       __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10034 +               __FILE__, __LINE__ )
10035 +
10036 +#define dl_reserve_space(in, bytes) \
10037 +       __dl_alloc_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10038 +               __FILE__, __LINE__ )
10039 +
10040 +#define dl_claim_space(in, bytes) (0)
10041 +
10042 +#define dl_release_space(in, bytes) \
10043 +       __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10044 +               __FILE__, __LINE__ )
10045 +
10046 +#define dl_free_space(in, bytes) \
10047 +       __dl_free_space((in)->i_sb, (in)->i_tag, (dlsize_t)(bytes), \
10048 +               __FILE__, __LINE__ )
10049 +
10050 +
10051 +
10052 +#define dl_alloc_inode(in) \
10053 +       __dl_alloc_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10054 +
10055 +#define dl_free_inode(in) \
10056 +       __dl_free_inode((in)->i_sb, (in)->i_tag, __FILE__, __LINE__ )
10057 +
10058 +
10059 +#define dl_adjust_block(sb, tag, fb, rb) \
10060 +       __dl_adjust_block(sb, tag, fb, rb, __FILE__, __LINE__ )
10061 +
10062 +
10063 +#else
10064 +#warning duplicate inclusion
10065 +#endif
10066 diff -NurpP --minimal linux-2.6.36/include/linux/vs_inet.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_inet.h
10067 --- linux-2.6.36/include/linux/vs_inet.h        1970-01-01 01:00:00.000000000 +0100
10068 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_inet.h  2010-10-21 17:41:35.000000000 +0200
10069 @@ -0,0 +1,351 @@
10070 +#ifndef _VS_INET_H
10071 +#define _VS_INET_H
10072 +
10073 +#include "vserver/base.h"
10074 +#include "vserver/network.h"
10075 +#include "vserver/debug.h"
10076 +
10077 +#define IPI_LOOPBACK   htonl(INADDR_LOOPBACK)
10078 +
10079 +#define NXAV4(a)       NIPQUAD((a)->ip[0]), NIPQUAD((a)->ip[1]), \
10080 +                       NIPQUAD((a)->mask), (a)->type
10081 +#define NXAV4_FMT      "[" NIPQUAD_FMT "-" NIPQUAD_FMT "/" NIPQUAD_FMT ":%04x]"
10082 +
10083 +#define NIPQUAD(addr) \
10084 +       ((unsigned char *)&addr)[0], \
10085 +       ((unsigned char *)&addr)[1], \
10086 +       ((unsigned char *)&addr)[2], \
10087 +       ((unsigned char *)&addr)[3]
10088 +
10089 +#define NIPQUAD_FMT "%u.%u.%u.%u"
10090 +
10091 +
10092 +static inline
10093 +int v4_addr_match(struct nx_addr_v4 *nxa, __be32 addr, uint16_t tmask)
10094 +{
10095 +       __be32 ip = nxa->ip[0].s_addr;
10096 +       __be32 mask = nxa->mask.s_addr;
10097 +       __be32 bcast = ip | ~mask;
10098 +       int ret = 0;
10099 +
10100 +       switch (nxa->type & tmask) {
10101 +       case NXA_TYPE_MASK:
10102 +               ret = (ip == (addr & mask));
10103 +               break;
10104 +       case NXA_TYPE_ADDR:
10105 +               ret = 3;
10106 +               if (addr == ip)
10107 +                       break;
10108 +               /* fall through to broadcast */
10109 +       case NXA_MOD_BCAST:
10110 +               ret = ((tmask & NXA_MOD_BCAST) && (addr == bcast));
10111 +               break;
10112 +       case NXA_TYPE_RANGE:
10113 +               ret = ((nxa->ip[0].s_addr <= addr) &&
10114 +                       (nxa->ip[1].s_addr > addr));
10115 +               break;
10116 +       case NXA_TYPE_ANY:
10117 +               ret = 2;
10118 +               break;
10119 +       }
10120 +
10121 +       vxdprintk(VXD_CBIT(net, 0),
10122 +               "v4_addr_match(%p" NXAV4_FMT "," NIPQUAD_FMT ",%04x) = %d",
10123 +               nxa, NXAV4(nxa), NIPQUAD(addr), tmask, ret);
10124 +       return ret;
10125 +}
10126 +
10127 +static inline
10128 +int v4_addr_in_nx_info(struct nx_info *nxi, __be32 addr, uint16_t tmask)
10129 +{
10130 +       struct nx_addr_v4 *nxa;
10131 +       int ret = 1;
10132 +
10133 +       if (!nxi)
10134 +               goto out;
10135 +
10136 +       ret = 2;
10137 +       /* allow 127.0.0.1 when remapping lback */
10138 +       if ((tmask & NXA_LOOPBACK) &&
10139 +               (addr == IPI_LOOPBACK) &&
10140 +               nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
10141 +               goto out;
10142 +       ret = 3;
10143 +       /* check for lback address */
10144 +       if ((tmask & NXA_MOD_LBACK) &&
10145 +               (nxi->v4_lback.s_addr == addr))
10146 +               goto out;
10147 +       ret = 4;
10148 +       /* check for broadcast address */
10149 +       if ((tmask & NXA_MOD_BCAST) &&
10150 +               (nxi->v4_bcast.s_addr == addr))
10151 +               goto out;
10152 +       ret = 5;
10153 +       /* check for v4 addresses */
10154 +       for (nxa = &nxi->v4; nxa; nxa = nxa->next)
10155 +               if (v4_addr_match(nxa, addr, tmask))
10156 +                       goto out;
10157 +       ret = 0;
10158 +out:
10159 +       vxdprintk(VXD_CBIT(net, 0),
10160 +               "v4_addr_in_nx_info(%p[#%u]," NIPQUAD_FMT ",%04x) = %d",
10161 +               nxi, nxi ? nxi->nx_id : 0, NIPQUAD(addr), tmask, ret);
10162 +       return ret;
10163 +}
10164 +
10165 +static inline
10166 +int v4_nx_addr_match(struct nx_addr_v4 *nxa, struct nx_addr_v4 *addr, uint16_t mask)
10167 +{
10168 +       /* FIXME: needs full range checks */
10169 +       return v4_addr_match(nxa, addr->ip[0].s_addr, mask);
10170 +}
10171 +
10172 +static inline
10173 +int v4_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v4 *nxa, uint16_t mask)
10174 +{
10175 +       struct nx_addr_v4 *ptr;
10176 +
10177 +       for (ptr = &nxi->v4; ptr; ptr = ptr->next)
10178 +               if (v4_nx_addr_match(ptr, nxa, mask))
10179 +                       return 1;
10180 +       return 0;
10181 +}
10182 +
10183 +#include <net/inet_sock.h>
10184 +
10185 +/*
10186 + *     Check if a given address matches for a socket
10187 + *
10188 + *     nxi:            the socket's nx_info if any
10189 + *     addr:           to be verified address
10190 + */
10191 +static inline
10192 +int v4_sock_addr_match (
10193 +       struct nx_info *nxi,
10194 +       struct inet_sock *inet,
10195 +       __be32 addr)
10196 +{
10197 +       __be32 saddr = inet->inet_rcv_saddr;
10198 +       __be32 bcast = nxi ? nxi->v4_bcast.s_addr : INADDR_BROADCAST;
10199 +
10200 +       if (addr && (saddr == addr || bcast == addr))
10201 +               return 1;
10202 +       if (!saddr)
10203 +               return v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND);
10204 +       return 0;
10205 +}
10206 +
10207 +
10208 +/* inet related checks and helpers */
10209 +
10210 +
10211 +struct in_ifaddr;
10212 +struct net_device;
10213 +struct sock;
10214 +
10215 +#ifdef CONFIG_INET
10216 +
10217 +#include <linux/netdevice.h>
10218 +#include <linux/inetdevice.h>
10219 +#include <net/inet_sock.h>
10220 +#include <net/inet_timewait_sock.h>
10221 +
10222 +
10223 +int dev_in_nx_info(struct net_device *, struct nx_info *);
10224 +int v4_dev_in_nx_info(struct net_device *, struct nx_info *);
10225 +int nx_v4_addr_conflict(struct nx_info *, struct nx_info *);
10226 +
10227 +
10228 +/*
10229 + *     check if address is covered by socket
10230 + *
10231 + *     sk:     the socket to check against
10232 + *     addr:   the address in question (must be != 0)
10233 + */
10234 +
10235 +static inline
10236 +int __v4_addr_match_socket(const struct sock *sk, struct nx_addr_v4 *nxa)
10237 +{
10238 +       struct nx_info *nxi = sk->sk_nx_info;
10239 +       __be32 saddr = inet_rcv_saddr(sk);
10240 +
10241 +       vxdprintk(VXD_CBIT(net, 5),
10242 +               "__v4_addr_in_socket(%p," NXAV4_FMT ") %p:" NIPQUAD_FMT " %p;%lx",
10243 +               sk, NXAV4(nxa), nxi, NIPQUAD(saddr), sk->sk_socket,
10244 +               (sk->sk_socket?sk->sk_socket->flags:0));
10245 +
10246 +       if (saddr) {            /* direct address match */
10247 +               return v4_addr_match(nxa, saddr, -1);
10248 +       } else if (nxi) {       /* match against nx_info */
10249 +               return v4_nx_addr_in_nx_info(nxi, nxa, -1);
10250 +       } else {                /* unrestricted any socket */
10251 +               return 1;
10252 +       }
10253 +}
10254 +
10255 +
10256 +
10257 +static inline
10258 +int nx_dev_visible(struct nx_info *nxi, struct net_device *dev)
10259 +{
10260 +       vxdprintk(VXD_CBIT(net, 1),
10261 +               "nx_dev_visible(%p[#%u],%p " VS_Q("%s") ") %d",
10262 +               nxi, nxi ? nxi->nx_id : 0, dev, dev->name,
10263 +               nxi ? dev_in_nx_info(dev, nxi) : 0);
10264 +
10265 +       if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
10266 +               return 1;
10267 +       if (dev_in_nx_info(dev, nxi))
10268 +               return 1;
10269 +       return 0;
10270 +}
10271 +
10272 +
10273 +static inline
10274 +int v4_ifa_in_nx_info(struct in_ifaddr *ifa, struct nx_info *nxi)
10275 +{
10276 +       if (!nxi)
10277 +               return 1;
10278 +       if (!ifa)
10279 +               return 0;
10280 +       return v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW);
10281 +}
10282 +
10283 +static inline
10284 +int nx_v4_ifa_visible(struct nx_info *nxi, struct in_ifaddr *ifa)
10285 +{
10286 +       vxdprintk(VXD_CBIT(net, 1), "nx_v4_ifa_visible(%p[#%u],%p) %d",
10287 +               nxi, nxi ? nxi->nx_id : 0, ifa,
10288 +               nxi ? v4_ifa_in_nx_info(ifa, nxi) : 0);
10289 +
10290 +       if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
10291 +               return 1;
10292 +       if (v4_ifa_in_nx_info(ifa, nxi))
10293 +               return 1;
10294 +       return 0;
10295 +}
10296 +
10297 +
10298 +struct nx_v4_sock_addr {
10299 +       __be32 saddr;   /* Address used for validation */
10300 +       __be32 baddr;   /* Address used for socket bind */
10301 +};
10302 +
10303 +static inline
10304 +int v4_map_sock_addr(struct inet_sock *inet, struct sockaddr_in *addr,
10305 +       struct nx_v4_sock_addr *nsa)
10306 +{
10307 +       struct sock *sk = &inet->sk;
10308 +       struct nx_info *nxi = sk->sk_nx_info;
10309 +       __be32 saddr = addr->sin_addr.s_addr;
10310 +       __be32 baddr = saddr;
10311 +
10312 +       vxdprintk(VXD_CBIT(net, 3),
10313 +               "inet_bind(%p)* %p,%p;%lx " NIPQUAD_FMT,
10314 +               sk, sk->sk_nx_info, sk->sk_socket,
10315 +               (sk->sk_socket ? sk->sk_socket->flags : 0),
10316 +               NIPQUAD(saddr));
10317 +
10318 +       if (nxi) {
10319 +               if (saddr == INADDR_ANY) {
10320 +                       if (nx_info_flags(nxi, NXF_SINGLE_IP, 0))
10321 +                               baddr = nxi->v4.ip[0].s_addr;
10322 +               } else if (saddr == IPI_LOOPBACK) {
10323 +                       if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
10324 +                               baddr = nxi->v4_lback.s_addr;
10325 +               } else {        /* normal address bind */
10326 +                       if (!v4_addr_in_nx_info(nxi, saddr, NXA_MASK_BIND))
10327 +                               return -EADDRNOTAVAIL;
10328 +               }
10329 +       }
10330 +
10331 +       vxdprintk(VXD_CBIT(net, 3),
10332 +               "inet_bind(%p) " NIPQUAD_FMT ", " NIPQUAD_FMT,
10333 +               sk, NIPQUAD(saddr), NIPQUAD(baddr));
10334 +
10335 +       nsa->saddr = saddr;
10336 +       nsa->baddr = baddr;
10337 +       return 0;
10338 +}
10339 +
10340 +static inline
10341 +void v4_set_sock_addr(struct inet_sock *inet, struct nx_v4_sock_addr *nsa)
10342 +{
10343 +       inet->inet_saddr = nsa->baddr;
10344 +       inet->inet_rcv_saddr = nsa->baddr;
10345 +}
10346 +
10347 +
10348 +/*
10349 + *      helper to simplify inet_lookup_listener
10350 + *
10351 + *      nxi:   the socket's nx_info if any
10352 + *      addr:  to be verified address
10353 + *      saddr: socket address
10354 + */
10355 +static inline int v4_inet_addr_match (
10356 +       struct nx_info *nxi,
10357 +       __be32 addr,
10358 +       __be32 saddr)
10359 +{
10360 +       if (addr && (saddr == addr))
10361 +               return 1;
10362 +       if (!saddr)
10363 +               return nxi ? v4_addr_in_nx_info(nxi, addr, NXA_MASK_BIND) : 1;
10364 +       return 0;
10365 +}
10366 +
10367 +static inline __be32 nx_map_sock_lback(struct nx_info *nxi, __be32 addr)
10368 +{
10369 +       if (nx_info_flags(nxi, NXF_HIDE_LBACK, 0) &&
10370 +               (addr == nxi->v4_lback.s_addr))
10371 +               return IPI_LOOPBACK;
10372 +       return addr;
10373 +}
10374 +
10375 +static inline
10376 +int nx_info_has_v4(struct nx_info *nxi)
10377 +{
10378 +       if (!nxi)
10379 +               return 1;
10380 +       if (NX_IPV4(nxi))
10381 +               return 1;
10382 +       if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0))
10383 +               return 1;
10384 +       return 0;
10385 +}
10386 +
10387 +#else /* CONFIG_INET */
10388 +
10389 +static inline
10390 +int nx_dev_visible(struct nx_info *n, struct net_device *d)
10391 +{
10392 +       return 1;
10393 +}
10394 +
10395 +static inline
10396 +int nx_v4_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
10397 +{
10398 +       return 1;
10399 +}
10400 +
10401 +static inline
10402 +int v4_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
10403 +{
10404 +       return 1;
10405 +}
10406 +
10407 +static inline
10408 +int nx_info_has_v4(struct nx_info *nxi)
10409 +{
10410 +       return 0;
10411 +}
10412 +
10413 +#endif /* CONFIG_INET */
10414 +
10415 +#define current_nx_info_has_v4() \
10416 +       nx_info_has_v4(current_nx_info())
10417 +
10418 +#else
10419 +// #warning duplicate inclusion
10420 +#endif
10421 diff -NurpP --minimal linux-2.6.36/include/linux/vs_inet6.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_inet6.h
10422 --- linux-2.6.36/include/linux/vs_inet6.h       1970-01-01 01:00:00.000000000 +0100
10423 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_inet6.h 2010-10-21 13:09:36.000000000 +0200
10424 @@ -0,0 +1,246 @@
10425 +#ifndef _VS_INET6_H
10426 +#define _VS_INET6_H
10427 +
10428 +#include "vserver/base.h"
10429 +#include "vserver/network.h"
10430 +#include "vserver/debug.h"
10431 +
10432 +#include <net/ipv6.h>
10433 +
10434 +#define NXAV6(a)       &(a)->ip, &(a)->mask, (a)->prefix, (a)->type
10435 +#define NXAV6_FMT      "[%pI6/%pI6/%d:%04x]"
10436 +
10437 +
10438 +#ifdef CONFIG_IPV6
10439 +
10440 +static inline
10441 +int v6_addr_match(struct nx_addr_v6 *nxa,
10442 +       const struct in6_addr *addr, uint16_t mask)
10443 +{
10444 +       int ret = 0;
10445 +
10446 +       switch (nxa->type & mask) {
10447 +       case NXA_TYPE_MASK:
10448 +               ret = ipv6_masked_addr_cmp(&nxa->ip, &nxa->mask, addr);
10449 +               break;
10450 +       case NXA_TYPE_ADDR:
10451 +               ret = ipv6_addr_equal(&nxa->ip, addr);
10452 +               break;
10453 +       case NXA_TYPE_ANY:
10454 +               ret = 1;
10455 +               break;
10456 +       }
10457 +       vxdprintk(VXD_CBIT(net, 0),
10458 +               "v6_addr_match(%p" NXAV6_FMT ",%pI6,%04x) = %d",
10459 +               nxa, NXAV6(nxa), addr, mask, ret);
10460 +       return ret;
10461 +}
10462 +
10463 +static inline
10464 +int v6_addr_in_nx_info(struct nx_info *nxi,
10465 +       const struct in6_addr *addr, uint16_t mask)
10466 +{
10467 +       struct nx_addr_v6 *nxa;
10468 +       int ret = 1;
10469 +
10470 +       if (!nxi)
10471 +               goto out;
10472 +       for (nxa = &nxi->v6; nxa; nxa = nxa->next)
10473 +               if (v6_addr_match(nxa, addr, mask))
10474 +                       goto out;
10475 +       ret = 0;
10476 +out:
10477 +       vxdprintk(VXD_CBIT(net, 0),
10478 +               "v6_addr_in_nx_info(%p[#%u],%pI6,%04x) = %d",
10479 +               nxi, nxi ? nxi->nx_id : 0, addr, mask, ret);
10480 +       return ret;
10481 +}
10482 +
10483 +static inline
10484 +int v6_nx_addr_match(struct nx_addr_v6 *nxa, struct nx_addr_v6 *addr, uint16_t mask)
10485 +{
10486 +       /* FIXME: needs full range checks */
10487 +       return v6_addr_match(nxa, &addr->ip, mask);
10488 +}
10489 +
10490 +static inline
10491 +int v6_nx_addr_in_nx_info(struct nx_info *nxi, struct nx_addr_v6 *nxa, uint16_t mask)
10492 +{
10493 +       struct nx_addr_v6 *ptr;
10494 +
10495 +       for (ptr = &nxi->v6; ptr; ptr = ptr->next)
10496 +               if (v6_nx_addr_match(ptr, nxa, mask))
10497 +                       return 1;
10498 +       return 0;
10499 +}
10500 +
10501 +
10502 +/*
10503 + *     Check if a given address matches for a socket
10504 + *
10505 + *     nxi:            the socket's nx_info if any
10506 + *     addr:           to be verified address
10507 + */
10508 +static inline
10509 +int v6_sock_addr_match (
10510 +       struct nx_info *nxi,
10511 +       struct inet_sock *inet,
10512 +       struct in6_addr *addr)
10513 +{
10514 +       struct sock *sk = &inet->sk;
10515 +       struct in6_addr *saddr = inet6_rcv_saddr(sk);
10516 +
10517 +       if (!ipv6_addr_any(addr) &&
10518 +               ipv6_addr_equal(saddr, addr))
10519 +               return 1;
10520 +       if (ipv6_addr_any(saddr))
10521 +               return v6_addr_in_nx_info(nxi, addr, -1);
10522 +       return 0;
10523 +}
10524 +
10525 +/*
10526 + *     check if address is covered by socket
10527 + *
10528 + *     sk:     the socket to check against
10529 + *     addr:   the address in question (must be != 0)
10530 + */
10531 +
10532 +static inline
10533 +int __v6_addr_match_socket(const struct sock *sk, struct nx_addr_v6 *nxa)
10534 +{
10535 +       struct nx_info *nxi = sk->sk_nx_info;
10536 +       struct in6_addr *saddr = inet6_rcv_saddr(sk);
10537 +
10538 +       vxdprintk(VXD_CBIT(net, 5),
10539 +               "__v6_addr_in_socket(%p," NXAV6_FMT ") %p:%pI6 %p;%lx",
10540 +               sk, NXAV6(nxa), nxi, saddr, sk->sk_socket,
10541 +               (sk->sk_socket?sk->sk_socket->flags:0));
10542 +
10543 +       if (!ipv6_addr_any(saddr)) {    /* direct address match */
10544 +               return v6_addr_match(nxa, saddr, -1);
10545 +       } else if (nxi) {               /* match against nx_info */
10546 +               return v6_nx_addr_in_nx_info(nxi, nxa, -1);
10547 +       } else {                        /* unrestricted any socket */
10548 +               return 1;
10549 +       }
10550 +}
10551 +
10552 +
10553 +/* inet related checks and helpers */
10554 +
10555 +
10556 +struct in_ifaddr;
10557 +struct net_device;
10558 +struct sock;
10559 +
10560 +
10561 +#include <linux/netdevice.h>
10562 +#include <linux/inetdevice.h>
10563 +#include <net/inet_timewait_sock.h>
10564 +
10565 +
10566 +int dev_in_nx_info(struct net_device *, struct nx_info *);
10567 +int v6_dev_in_nx_info(struct net_device *, struct nx_info *);
10568 +int nx_v6_addr_conflict(struct nx_info *, struct nx_info *);
10569 +
10570 +
10571 +
10572 +static inline
10573 +int v6_ifa_in_nx_info(struct inet6_ifaddr *ifa, struct nx_info *nxi)
10574 +{
10575 +       if (!nxi)
10576 +               return 1;
10577 +       if (!ifa)
10578 +               return 0;
10579 +       return v6_addr_in_nx_info(nxi, &ifa->addr, -1);
10580 +}
10581 +
10582 +static inline
10583 +int nx_v6_ifa_visible(struct nx_info *nxi, struct inet6_ifaddr *ifa)
10584 +{
10585 +       vxdprintk(VXD_CBIT(net, 1), "nx_v6_ifa_visible(%p[#%u],%p) %d",
10586 +               nxi, nxi ? nxi->nx_id : 0, ifa,
10587 +               nxi ? v6_ifa_in_nx_info(ifa, nxi) : 0);
10588 +
10589 +       if (!nx_info_flags(nxi, NXF_HIDE_NETIF, 0))
10590 +               return 1;
10591 +       if (v6_ifa_in_nx_info(ifa, nxi))
10592 +               return 1;
10593 +       return 0;
10594 +}
10595 +
10596 +
10597 +struct nx_v6_sock_addr {
10598 +       struct in6_addr saddr;  /* Address used for validation */
10599 +       struct in6_addr baddr;  /* Address used for socket bind */
10600 +};
10601 +
10602 +static inline
10603 +int v6_map_sock_addr(struct inet_sock *inet, struct sockaddr_in6 *addr,
10604 +       struct nx_v6_sock_addr *nsa)
10605 +{
10606 +       // struct sock *sk = &inet->sk;
10607 +       // struct nx_info *nxi = sk->sk_nx_info;
10608 +       struct in6_addr saddr = addr->sin6_addr;
10609 +       struct in6_addr baddr = saddr;
10610 +
10611 +       nsa->saddr = saddr;
10612 +       nsa->baddr = baddr;
10613 +       return 0;
10614 +}
10615 +
10616 +static inline
10617 +void v6_set_sock_addr(struct inet_sock *inet, struct nx_v6_sock_addr *nsa)
10618 +{
10619 +       // struct sock *sk = &inet->sk;
10620 +       // struct in6_addr *saddr = inet6_rcv_saddr(sk);
10621 +
10622 +       // *saddr = nsa->baddr;
10623 +       // inet->inet_saddr = nsa->baddr;
10624 +}
10625 +
10626 +static inline
10627 +int nx_info_has_v6(struct nx_info *nxi)
10628 +{
10629 +       if (!nxi)
10630 +               return 1;
10631 +       if (NX_IPV6(nxi))
10632 +               return 1;
10633 +       return 0;
10634 +}
10635 +
10636 +#else /* CONFIG_IPV6 */
10637 +
10638 +static inline
10639 +int nx_v6_dev_visible(struct nx_info *n, struct net_device *d)
10640 +{
10641 +       return 1;
10642 +}
10643 +
10644 +
10645 +static inline
10646 +int nx_v6_addr_conflict(struct nx_info *n, uint32_t a, const struct sock *s)
10647 +{
10648 +       return 1;
10649 +}
10650 +
10651 +static inline
10652 +int v6_ifa_in_nx_info(struct in_ifaddr *a, struct nx_info *n)
10653 +{
10654 +       return 1;
10655 +}
10656 +
10657 +static inline
10658 +int nx_info_has_v6(struct nx_info *nxi)
10659 +{
10660 +       return 0;
10661 +}
10662 +
10663 +#endif /* CONFIG_IPV6 */
10664 +
10665 +#define current_nx_info_has_v6() \
10666 +       nx_info_has_v6(current_nx_info())
10667 +
10668 +#else
10669 +#warning duplicate inclusion
10670 +#endif
10671 diff -NurpP --minimal linux-2.6.36/include/linux/vs_limit.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_limit.h
10672 --- linux-2.6.36/include/linux/vs_limit.h       1970-01-01 01:00:00.000000000 +0100
10673 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_limit.h 2010-10-21 13:09:36.000000000 +0200
10674 @@ -0,0 +1,140 @@
10675 +#ifndef _VS_LIMIT_H
10676 +#define _VS_LIMIT_H
10677 +
10678 +#include "vserver/limit.h"
10679 +#include "vserver/base.h"
10680 +#include "vserver/context.h"
10681 +#include "vserver/debug.h"
10682 +#include "vserver/context.h"
10683 +#include "vserver/limit_int.h"
10684 +
10685 +
10686 +#define vx_acc_cres(v, d, p, r) \
10687 +       __vx_acc_cres(v, r, d, p, __FILE__, __LINE__)
10688 +
10689 +#define vx_acc_cres_cond(x, d, p, r) \
10690 +       __vx_acc_cres(((x) == vx_current_xid()) ? current_vx_info() : 0, \
10691 +       r, d, p, __FILE__, __LINE__)
10692 +
10693 +
10694 +#define vx_add_cres(v, a, p, r) \
10695 +       __vx_add_cres(v, r, a, p, __FILE__, __LINE__)
10696 +#define vx_sub_cres(v, a, p, r)                vx_add_cres(v, -(a), p, r)
10697 +
10698 +#define vx_add_cres_cond(x, a, p, r) \
10699 +       __vx_add_cres(((x) == vx_current_xid()) ? current_vx_info() : 0, \
10700 +       r, a, p, __FILE__, __LINE__)
10701 +#define vx_sub_cres_cond(x, a, p, r)   vx_add_cres_cond(x, -(a), p, r)
10702 +
10703 +
10704 +/* process and file limits */
10705 +
10706 +#define vx_nproc_inc(p) \
10707 +       vx_acc_cres((p)->vx_info, 1, p, RLIMIT_NPROC)
10708 +
10709 +#define vx_nproc_dec(p) \
10710 +       vx_acc_cres((p)->vx_info,-1, p, RLIMIT_NPROC)
10711 +
10712 +#define vx_files_inc(f) \
10713 +       vx_acc_cres_cond((f)->f_xid, 1, f, RLIMIT_NOFILE)
10714 +
10715 +#define vx_files_dec(f) \
10716 +       vx_acc_cres_cond((f)->f_xid,-1, f, RLIMIT_NOFILE)
10717 +
10718 +#define vx_locks_inc(l) \
10719 +       vx_acc_cres_cond((l)->fl_xid, 1, l, RLIMIT_LOCKS)
10720 +
10721 +#define vx_locks_dec(l) \
10722 +       vx_acc_cres_cond((l)->fl_xid,-1, l, RLIMIT_LOCKS)
10723 +
10724 +#define vx_openfd_inc(f) \
10725 +       vx_acc_cres(current_vx_info(), 1, (void *)(long)(f), VLIMIT_OPENFD)
10726 +
10727 +#define vx_openfd_dec(f) \
10728 +       vx_acc_cres(current_vx_info(),-1, (void *)(long)(f), VLIMIT_OPENFD)
10729 +
10730 +
10731 +#define vx_cres_avail(v, n, r) \
10732 +       __vx_cres_avail(v, r, n, __FILE__, __LINE__)
10733 +
10734 +
10735 +#define vx_nproc_avail(n) \
10736 +       vx_cres_avail(current_vx_info(), n, RLIMIT_NPROC)
10737 +
10738 +#define vx_files_avail(n) \
10739 +       vx_cres_avail(current_vx_info(), n, RLIMIT_NOFILE)
10740 +
10741 +#define vx_locks_avail(n) \
10742 +       vx_cres_avail(current_vx_info(), n, RLIMIT_LOCKS)
10743 +
10744 +#define vx_openfd_avail(n) \
10745 +       vx_cres_avail(current_vx_info(), n, VLIMIT_OPENFD)
10746 +
10747 +
10748 +/* dentry limits */
10749 +
10750 +#define vx_dentry_inc(d) do {                                          \
10751 +       if (atomic_read(&d->d_count) == 1)                              \
10752 +               vx_acc_cres(current_vx_info(), 1, d, VLIMIT_DENTRY);    \
10753 +       } while (0)
10754 +
10755 +#define vx_dentry_dec(d) do {                                          \
10756 +       if (atomic_read(&d->d_count) == 0)                              \
10757 +               vx_acc_cres(current_vx_info(),-1, d, VLIMIT_DENTRY);    \
10758 +       } while (0)
10759 +
10760 +#define vx_dentry_avail(n) \
10761 +       vx_cres_avail(current_vx_info(), n, VLIMIT_DENTRY)
10762 +
10763 +
10764 +/* socket limits */
10765 +
10766 +#define vx_sock_inc(s) \
10767 +       vx_acc_cres((s)->sk_vx_info, 1, s, VLIMIT_NSOCK)
10768 +
10769 +#define vx_sock_dec(s) \
10770 +       vx_acc_cres((s)->sk_vx_info,-1, s, VLIMIT_NSOCK)
10771 +
10772 +#define vx_sock_avail(n) \
10773 +       vx_cres_avail(current_vx_info(), n, VLIMIT_NSOCK)
10774 +
10775 +
10776 +/* ipc resource limits */
10777 +
10778 +#define vx_ipcmsg_add(v, u, a) \
10779 +       vx_add_cres(v, a, u, RLIMIT_MSGQUEUE)
10780 +
10781 +#define vx_ipcmsg_sub(v, u, a) \
10782 +       vx_sub_cres(v, a, u, RLIMIT_MSGQUEUE)
10783 +
10784 +#define vx_ipcmsg_avail(v, a) \
10785 +       vx_cres_avail(v, a, RLIMIT_MSGQUEUE)
10786 +
10787 +
10788 +#define vx_ipcshm_add(v, k, a) \
10789 +       vx_add_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
10790 +
10791 +#define vx_ipcshm_sub(v, k, a) \
10792 +       vx_sub_cres(v, a, (void *)(long)(k), VLIMIT_SHMEM)
10793 +
10794 +#define vx_ipcshm_avail(v, a) \
10795 +       vx_cres_avail(v, a, VLIMIT_SHMEM)
10796 +
10797 +
10798 +#define vx_semary_inc(a) \
10799 +       vx_acc_cres(current_vx_info(), 1, a, VLIMIT_SEMARY)
10800 +
10801 +#define vx_semary_dec(a) \
10802 +       vx_acc_cres(current_vx_info(), -1, a, VLIMIT_SEMARY)
10803 +
10804 +
10805 +#define vx_nsems_add(a,n) \
10806 +       vx_add_cres(current_vx_info(), n, a, VLIMIT_NSEMS)
10807 +
10808 +#define vx_nsems_sub(a,n) \
10809 +       vx_sub_cres(current_vx_info(), n, a, VLIMIT_NSEMS)
10810 +
10811 +
10812 +#else
10813 +#warning duplicate inclusion
10814 +#endif
10815 diff -NurpP --minimal linux-2.6.36/include/linux/vs_memory.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_memory.h
10816 --- linux-2.6.36/include/linux/vs_memory.h      1970-01-01 01:00:00.000000000 +0100
10817 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_memory.h        2010-10-21 13:09:36.000000000 +0200
10818 @@ -0,0 +1,58 @@
10819 +#ifndef _VS_MEMORY_H
10820 +#define _VS_MEMORY_H
10821 +
10822 +#include "vserver/limit.h"
10823 +#include "vserver/base.h"
10824 +#include "vserver/context.h"
10825 +#include "vserver/debug.h"
10826 +#include "vserver/context.h"
10827 +#include "vserver/limit_int.h"
10828 +
10829 +enum {
10830 +       VXPT_UNKNOWN = 0,
10831 +       VXPT_ANON,
10832 +       VXPT_NONE,
10833 +       VXPT_FILE,
10834 +       VXPT_SWAP,
10835 +       VXPT_WRITE
10836 +};
10837 +
10838 +#if 0
10839 +#define        vx_page_fault(mm, vma, type, ret)
10840 +#else
10841 +
10842 +static inline
10843 +void __vx_page_fault(struct mm_struct *mm,
10844 +       struct vm_area_struct *vma, int type, int ret)
10845 +{
10846 +       struct vx_info *vxi = mm->mm_vx_info;
10847 +       int what;
10848 +/*
10849 +       static char *page_type[6] =
10850 +               { "UNKNOWN", "ANON", "NONE", "FILE", "SWAP", "WRITE" };
10851 +       static char *page_what[4] =
10852 +               { "FAULT_OOM", "FAULT_SIGBUS", "FAULT_MINOR", "FAULT_MAJOR" };
10853 +*/
10854 +
10855 +       if (!vxi)
10856 +               return;
10857 +
10858 +       what = (ret & 0x3);
10859 +
10860 +/*     printk("[%d] page[%d][%d] %2x %s %s\n", vxi->vx_id,
10861 +               type, what, ret, page_type[type], page_what[what]);
10862 +*/
10863 +       if (ret & VM_FAULT_WRITE)
10864 +               what |= 0x4;
10865 +       atomic_inc(&vxi->cacct.page[type][what]);
10866 +}
10867 +
10868 +#define        vx_page_fault(mm, vma, type, ret)       __vx_page_fault(mm, vma, type, ret)
10869 +#endif
10870 +
10871 +
10872 +extern unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm);
10873 +
10874 +#else
10875 +#warning duplicate inclusion
10876 +#endif
10877 diff -NurpP --minimal linux-2.6.36/include/linux/vs_network.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_network.h
10878 --- linux-2.6.36/include/linux/vs_network.h     1970-01-01 01:00:00.000000000 +0100
10879 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_network.h       2010-10-21 13:09:36.000000000 +0200
10880 @@ -0,0 +1,169 @@
10881 +#ifndef _NX_VS_NETWORK_H
10882 +#define _NX_VS_NETWORK_H
10883 +
10884 +#include "vserver/context.h"
10885 +#include "vserver/network.h"
10886 +#include "vserver/base.h"
10887 +#include "vserver/check.h"
10888 +#include "vserver/debug.h"
10889 +
10890 +#include <linux/sched.h>
10891 +
10892 +
10893 +#define get_nx_info(i) __get_nx_info(i, __FILE__, __LINE__)
10894 +
10895 +static inline struct nx_info *__get_nx_info(struct nx_info *nxi,
10896 +       const char *_file, int _line)
10897 +{
10898 +       if (!nxi)
10899 +               return NULL;
10900 +
10901 +       vxlprintk(VXD_CBIT(nid, 2), "get_nx_info(%p[#%d.%d])",
10902 +               nxi, nxi ? nxi->nx_id : 0,
10903 +               nxi ? atomic_read(&nxi->nx_usecnt) : 0,
10904 +               _file, _line);
10905 +
10906 +       atomic_inc(&nxi->nx_usecnt);
10907 +       return nxi;
10908 +}
10909 +
10910 +
10911 +extern void free_nx_info(struct nx_info *);
10912 +
10913 +#define put_nx_info(i) __put_nx_info(i, __FILE__, __LINE__)
10914 +
10915 +static inline void __put_nx_info(struct nx_info *nxi, const char *_file, int _line)
10916 +{
10917 +       if (!nxi)
10918 +               return;
10919 +
10920 +       vxlprintk(VXD_CBIT(nid, 2), "put_nx_info(%p[#%d.%d])",
10921 +               nxi, nxi ? nxi->nx_id : 0,
10922 +               nxi ? atomic_read(&nxi->nx_usecnt) : 0,
10923 +               _file, _line);
10924 +
10925 +       if (atomic_dec_and_test(&nxi->nx_usecnt))
10926 +               free_nx_info(nxi);
10927 +}
10928 +
10929 +
10930 +#define init_nx_info(p, i) __init_nx_info(p, i, __FILE__, __LINE__)
10931 +
10932 +static inline void __init_nx_info(struct nx_info **nxp, struct nx_info *nxi,
10933 +               const char *_file, int _line)
10934 +{
10935 +       if (nxi) {
10936 +               vxlprintk(VXD_CBIT(nid, 3),
10937 +                       "init_nx_info(%p[#%d.%d])",
10938 +                       nxi, nxi ? nxi->nx_id : 0,
10939 +                       nxi ? atomic_read(&nxi->nx_usecnt) : 0,
10940 +                       _file, _line);
10941 +
10942 +               atomic_inc(&nxi->nx_usecnt);
10943 +       }
10944 +       *nxp = nxi;
10945 +}
10946 +
10947 +
10948 +#define set_nx_info(p, i) __set_nx_info(p, i, __FILE__, __LINE__)
10949 +
10950 +static inline void __set_nx_info(struct nx_info **nxp, struct nx_info *nxi,
10951 +       const char *_file, int _line)
10952 +{
10953 +       struct nx_info *nxo;
10954 +
10955 +       if (!nxi)
10956 +               return;
10957 +
10958 +       vxlprintk(VXD_CBIT(nid, 3), "set_nx_info(%p[#%d.%d])",
10959 +               nxi, nxi ? nxi->nx_id : 0,
10960 +               nxi ? atomic_read(&nxi->nx_usecnt) : 0,
10961 +               _file, _line);
10962 +
10963 +       atomic_inc(&nxi->nx_usecnt);
10964 +       nxo = xchg(nxp, nxi);
10965 +       BUG_ON(nxo);
10966 +}
10967 +
10968 +#define clr_nx_info(p) __clr_nx_info(p, __FILE__, __LINE__)
10969 +
10970 +static inline void __clr_nx_info(struct nx_info **nxp,
10971 +       const char *_file, int _line)
10972 +{
10973 +       struct nx_info *nxo;
10974 +
10975 +       nxo = xchg(nxp, NULL);
10976 +       if (!nxo)
10977 +               return;
10978 +
10979 +       vxlprintk(VXD_CBIT(nid, 3), "clr_nx_info(%p[#%d.%d])",
10980 +               nxo, nxo ? nxo->nx_id : 0,
10981 +               nxo ? atomic_read(&nxo->nx_usecnt) : 0,
10982 +               _file, _line);
10983 +
10984 +       if (atomic_dec_and_test(&nxo->nx_usecnt))
10985 +               free_nx_info(nxo);
10986 +}
10987 +
10988 +
10989 +#define claim_nx_info(v, p) __claim_nx_info(v, p, __FILE__, __LINE__)
10990 +
10991 +static inline void __claim_nx_info(struct nx_info *nxi,
10992 +       struct task_struct *task, const char *_file, int _line)
10993 +{
10994 +       vxlprintk(VXD_CBIT(nid, 3), "claim_nx_info(%p[#%d.%d.%d]) %p",
10995 +               nxi, nxi ? nxi->nx_id : 0,
10996 +               nxi?atomic_read(&nxi->nx_usecnt):0,
10997 +               nxi?atomic_read(&nxi->nx_tasks):0,
10998 +               task, _file, _line);
10999 +
11000 +       atomic_inc(&nxi->nx_tasks);
11001 +}
11002 +
11003 +
11004 +extern void unhash_nx_info(struct nx_info *);
11005 +
11006 +#define release_nx_info(v, p) __release_nx_info(v, p, __FILE__, __LINE__)
11007 +
11008 +static inline void __release_nx_info(struct nx_info *nxi,
11009 +       struct task_struct *task, const char *_file, int _line)
11010 +{
11011 +       vxlprintk(VXD_CBIT(nid, 3), "release_nx_info(%p[#%d.%d.%d]) %p",
11012 +               nxi, nxi ? nxi->nx_id : 0,
11013 +               nxi ? atomic_read(&nxi->nx_usecnt) : 0,
11014 +               nxi ? atomic_read(&nxi->nx_tasks) : 0,
11015 +               task, _file, _line);
11016 +
11017 +       might_sleep();
11018 +
11019 +       if (atomic_dec_and_test(&nxi->nx_tasks))
11020 +               unhash_nx_info(nxi);
11021 +}
11022 +
11023 +
11024 +#define task_get_nx_info(i)    __task_get_nx_info(i, __FILE__, __LINE__)
11025 +
11026 +static __inline__ struct nx_info *__task_get_nx_info(struct task_struct *p,
11027 +       const char *_file, int _line)
11028 +{
11029 +       struct nx_info *nxi;
11030 +
11031 +       task_lock(p);
11032 +       vxlprintk(VXD_CBIT(nid, 5), "task_get_nx_info(%p)",
11033 +               p, _file, _line);
11034 +       nxi = __get_nx_info(p->nx_info, _file, _line);
11035 +       task_unlock(p);
11036 +       return nxi;
11037 +}
11038 +
11039 +
11040 +static inline void exit_nx_info(struct task_struct *p)
11041 +{
11042 +       if (p->nx_info)
11043 +               release_nx_info(p->nx_info, p);
11044 +}
11045 +
11046 +
11047 +#else
11048 +#warning duplicate inclusion
11049 +#endif
11050 diff -NurpP --minimal linux-2.6.36/include/linux/vs_pid.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_pid.h
11051 --- linux-2.6.36/include/linux/vs_pid.h 1970-01-01 01:00:00.000000000 +0100
11052 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_pid.h   2010-10-21 13:09:36.000000000 +0200
11053 @@ -0,0 +1,95 @@
11054 +#ifndef _VS_PID_H
11055 +#define _VS_PID_H
11056 +
11057 +#include "vserver/base.h"
11058 +#include "vserver/check.h"
11059 +#include "vserver/context.h"
11060 +#include "vserver/debug.h"
11061 +#include "vserver/pid.h"
11062 +#include <linux/pid_namespace.h>
11063 +
11064 +
11065 +#define VXF_FAKE_INIT  (VXF_INFO_INIT | VXF_STATE_INIT)
11066 +
11067 +static inline
11068 +int vx_proc_task_visible(struct task_struct *task)
11069 +{
11070 +       if ((task->pid == 1) &&
11071 +               !vx_flags(VXF_FAKE_INIT, VXF_FAKE_INIT))
11072 +               /* show a blend through init */
11073 +               goto visible;
11074 +       if (vx_check(vx_task_xid(task), VS_WATCH | VS_IDENT))
11075 +               goto visible;
11076 +       return 0;
11077 +visible:
11078 +       return 1;
11079 +}
11080 +
11081 +#define find_task_by_real_pid(pid) find_task_by_pid_ns(pid, &init_pid_ns)
11082 +
11083 +#if 0
11084 +
11085 +static inline
11086 +struct task_struct *vx_find_proc_task_by_pid(int pid)
11087 +{
11088 +       struct task_struct *task = find_task_by_real_pid(pid);
11089 +
11090 +       if (task && !vx_proc_task_visible(task)) {
11091 +               vxdprintk(VXD_CBIT(misc, 6),
11092 +                       "dropping task (find) %p[#%u,%u] for %p[#%u,%u]",
11093 +                       task, task->xid, task->pid,
11094 +                       current, current->xid, current->pid);
11095 +               task = NULL;
11096 +       }
11097 +       return task;
11098 +}
11099 +
11100 +#endif
11101 +
11102 +static inline
11103 +struct task_struct *vx_get_proc_task(struct inode *inode, struct pid *pid)
11104 +{
11105 +       struct task_struct *task = get_pid_task(pid, PIDTYPE_PID);
11106 +
11107 +       if (task && !vx_proc_task_visible(task)) {
11108 +               vxdprintk(VXD_CBIT(misc, 6),
11109 +                       "dropping task (get) %p[#%u,%u] for %p[#%u,%u]",
11110 +                       task, task->xid, task->pid,
11111 +                       current, current->xid, current->pid);
11112 +               put_task_struct(task);
11113 +               task = NULL;
11114 +       }
11115 +       return task;
11116 +}
11117 +
11118 +#if 0
11119 +
11120 +static inline
11121 +struct task_struct *vx_child_reaper(struct task_struct *p)
11122 +{
11123 +       struct vx_info *vxi = p->vx_info;
11124 +       struct task_struct *reaper = child_reaper(p);
11125 +
11126 +       if (!vxi)
11127 +               goto out;
11128 +
11129 +       BUG_ON(!p->vx_info->vx_reaper);
11130 +
11131 +       /* child reaper for the guest reaper */
11132 +       if (vxi->vx_reaper == p)
11133 +               goto out;
11134 +
11135 +       reaper = vxi->vx_reaper;
11136 +out:
11137 +       vxdprintk(VXD_CBIT(xid, 7),
11138 +               "vx_child_reaper(%p[#%u,%u]) = %p[#%u,%u]",
11139 +               p, p->xid, p->pid, reaper, reaper->xid, reaper->pid);
11140 +       return reaper;
11141 +}
11142 +
11143 +#endif
11144 +
11145 +
11146 +#else
11147 +#warning duplicate inclusion
11148 +#endif
11149 diff -NurpP --minimal linux-2.6.36/include/linux/vs_sched.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_sched.h
11150 --- linux-2.6.36/include/linux/vs_sched.h       1970-01-01 01:00:00.000000000 +0100
11151 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_sched.h 2010-10-24 15:13:57.000000000 +0200
11152 @@ -0,0 +1,40 @@
11153 +#ifndef _VS_SCHED_H
11154 +#define _VS_SCHED_H
11155 +
11156 +#include "vserver/base.h"
11157 +#include "vserver/context.h"
11158 +#include "vserver/sched.h"
11159 +
11160 +
11161 +#define MAX_PRIO_BIAS           20
11162 +#define MIN_PRIO_BIAS          -20
11163 +
11164 +static inline
11165 +int vx_adjust_prio(struct task_struct *p, int prio, int max_user)
11166 +{
11167 +       struct vx_info *vxi = p->vx_info;
11168 +
11169 +       if (vxi)
11170 +               prio += vx_cpu(vxi, sched_pc).prio_bias;
11171 +       return prio;
11172 +}
11173 +
11174 +static inline void vx_account_user(struct vx_info *vxi,
11175 +       cputime_t cputime, int nice)
11176 +{
11177 +       if (!vxi)
11178 +               return;
11179 +       vx_cpu(vxi, sched_pc).user_ticks += cputime;
11180 +}
11181 +
11182 +static inline void vx_account_system(struct vx_info *vxi,
11183 +       cputime_t cputime, int idle)
11184 +{
11185 +       if (!vxi)
11186 +               return;
11187 +       vx_cpu(vxi, sched_pc).sys_ticks += cputime;
11188 +}
11189 +
11190 +#else
11191 +#warning duplicate inclusion
11192 +#endif
11193 diff -NurpP --minimal linux-2.6.36/include/linux/vs_socket.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_socket.h
11194 --- linux-2.6.36/include/linux/vs_socket.h      1970-01-01 01:00:00.000000000 +0100
11195 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_socket.h        2010-10-21 13:09:36.000000000 +0200
11196 @@ -0,0 +1,67 @@
11197 +#ifndef _VS_SOCKET_H
11198 +#define _VS_SOCKET_H
11199 +
11200 +#include "vserver/debug.h"
11201 +#include "vserver/base.h"
11202 +#include "vserver/cacct.h"
11203 +#include "vserver/context.h"
11204 +#include "vserver/tag.h"
11205 +
11206 +
11207 +/* socket accounting */
11208 +
11209 +#include <linux/socket.h>
11210 +
11211 +static inline int vx_sock_type(int family)
11212 +{
11213 +       switch (family) {
11214 +       case PF_UNSPEC:
11215 +               return VXA_SOCK_UNSPEC;
11216 +       case PF_UNIX:
11217 +               return VXA_SOCK_UNIX;
11218 +       case PF_INET:
11219 +               return VXA_SOCK_INET;
11220 +       case PF_INET6:
11221 +               return VXA_SOCK_INET6;
11222 +       case PF_PACKET:
11223 +               return VXA_SOCK_PACKET;
11224 +       default:
11225 +               return VXA_SOCK_OTHER;
11226 +       }
11227 +}
11228 +
11229 +#define vx_acc_sock(v, f, p, s) \
11230 +       __vx_acc_sock(v, f, p, s, __FILE__, __LINE__)
11231 +
11232 +static inline void __vx_acc_sock(struct vx_info *vxi,
11233 +       int family, int pos, int size, char *file, int line)
11234 +{
11235 +       if (vxi) {
11236 +               int type = vx_sock_type(family);
11237 +
11238 +               atomic_long_inc(&vxi->cacct.sock[type][pos].count);
11239 +               atomic_long_add(size, &vxi->cacct.sock[type][pos].total);
11240 +       }
11241 +}
11242 +
11243 +#define vx_sock_recv(sk, s) \
11244 +       vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 0, s)
11245 +#define vx_sock_send(sk, s) \
11246 +       vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 1, s)
11247 +#define vx_sock_fail(sk, s) \
11248 +       vx_acc_sock((sk)->sk_vx_info, (sk)->sk_family, 2, s)
11249 +
11250 +
11251 +#define sock_vx_init(s) do {           \
11252 +       (s)->sk_xid = 0;                \
11253 +       (s)->sk_vx_info = NULL;         \
11254 +       } while (0)
11255 +
11256 +#define sock_nx_init(s) do {           \
11257 +       (s)->sk_nid = 0;                \
11258 +       (s)->sk_nx_info = NULL;         \
11259 +       } while (0)
11260 +
11261 +#else
11262 +#warning duplicate inclusion
11263 +#endif
11264 diff -NurpP --minimal linux-2.6.36/include/linux/vs_tag.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_tag.h
11265 --- linux-2.6.36/include/linux/vs_tag.h 1970-01-01 01:00:00.000000000 +0100
11266 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_tag.h   2010-10-21 13:09:36.000000000 +0200
11267 @@ -0,0 +1,47 @@
11268 +#ifndef _VS_TAG_H
11269 +#define _VS_TAG_H
11270 +
11271 +#include <linux/vserver/tag.h>
11272 +
11273 +/* check conditions */
11274 +
11275 +#define DX_ADMIN       0x0001
11276 +#define DX_WATCH       0x0002
11277 +#define DX_HOSTID      0x0008
11278 +
11279 +#define DX_IDENT       0x0010
11280 +
11281 +#define DX_ARG_MASK    0x0010
11282 +
11283 +
11284 +#define dx_task_tag(t) ((t)->tag)
11285 +
11286 +#define dx_current_tag() dx_task_tag(current)
11287 +
11288 +#define dx_check(c, m) __dx_check(dx_current_tag(), c, m)
11289 +
11290 +#define dx_weak_check(c, m)    ((m) ? dx_check(c, m) : 1)
11291 +
11292 +
11293 +/*
11294 + * check current context for ADMIN/WATCH and
11295 + * optionally against supplied argument
11296 + */
11297 +static inline int __dx_check(tag_t cid, tag_t id, unsigned int mode)
11298 +{
11299 +       if (mode & DX_ARG_MASK) {
11300 +               if ((mode & DX_IDENT) && (id == cid))
11301 +                       return 1;
11302 +       }
11303 +       return (((mode & DX_ADMIN) && (cid == 0)) ||
11304 +               ((mode & DX_WATCH) && (cid == 1)) ||
11305 +               ((mode & DX_HOSTID) && (id == 0)));
11306 +}
11307 +
11308 +struct inode;
11309 +int dx_permission(const struct inode *inode, int mask);
11310 +
11311 +
11312 +#else
11313 +#warning duplicate inclusion
11314 +#endif
11315 diff -NurpP --minimal linux-2.6.36/include/linux/vs_time.h linux-2.6.36-vs2.3.0.36.38/include/linux/vs_time.h
11316 --- linux-2.6.36/include/linux/vs_time.h        1970-01-01 01:00:00.000000000 +0100
11317 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vs_time.h  2010-10-21 14:25:37.000000000 +0200
11318 @@ -0,0 +1,19 @@
11319 +#ifndef _VS_TIME_H
11320 +#define _VS_TIME_H
11321 +
11322 +
11323 +/* time faking stuff */
11324 +
11325 +#ifdef CONFIG_VSERVER_VTIME
11326 +
11327 +extern void vx_adjust_timespec(struct timespec *ts);
11328 +extern int vx_settimeofday(struct timespec *ts);
11329 +
11330 +#else
11331 +#define        vx_adjust_timespec(t)   do { } while (0)
11332 +#define        vx_settimeofday(t)      do_settimeofday(t)
11333 +#endif
11334 +
11335 +#else
11336 +#warning duplicate inclusion
11337 +#endif
11338 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/Kbuild linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/Kbuild
11339 --- linux-2.6.36/include/linux/vserver/Kbuild   1970-01-01 01:00:00.000000000 +0100
11340 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/Kbuild     2010-10-21 13:09:36.000000000 +0200
11341 @@ -0,0 +1,8 @@
11342 +
11343 +unifdef-y += context_cmd.h network_cmd.h space_cmd.h \
11344 +       cacct_cmd.h cvirt_cmd.h limit_cmd.h dlimit_cmd.h \
11345 +       inode_cmd.h tag_cmd.h sched_cmd.h signal_cmd.h \
11346 +       debug_cmd.h device_cmd.h
11347 +
11348 +unifdef-y += switch.h network.h monitor.h inode.h device.h
11349 +
11350 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/base.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/base.h
11351 --- linux-2.6.36/include/linux/vserver/base.h   1970-01-01 01:00:00.000000000 +0100
11352 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/base.h     2010-10-21 13:09:36.000000000 +0200
11353 @@ -0,0 +1,170 @@
11354 +#ifndef _VX_BASE_H
11355 +#define _VX_BASE_H
11356 +
11357 +
11358 +/* context state changes */
11359 +
11360 +enum {
11361 +       VSC_STARTUP = 1,
11362 +       VSC_SHUTDOWN,
11363 +
11364 +       VSC_NETUP,
11365 +       VSC_NETDOWN,
11366 +};
11367 +
11368 +
11369 +
11370 +#define vx_task_xid(t) ((t)->xid)
11371 +
11372 +#define vx_current_xid() vx_task_xid(current)
11373 +
11374 +#define current_vx_info() (current->vx_info)
11375 +
11376 +
11377 +#define nx_task_nid(t) ((t)->nid)
11378 +
11379 +#define nx_current_nid() nx_task_nid(current)
11380 +
11381 +#define current_nx_info() (current->nx_info)
11382 +
11383 +
11384 +/* generic flag merging */
11385 +
11386 +#define vs_check_flags(v, m, f)        (((v) & (m)) ^ (f))
11387 +
11388 +#define vs_mask_flags(v, f, m) (((v) & ~(m)) | ((f) & (m)))
11389 +
11390 +#define vs_mask_mask(v, f, m)  (((v) & ~(m)) | ((v) & (f) & (m)))
11391 +
11392 +#define vs_check_bit(v, n)     ((v) & (1LL << (n)))
11393 +
11394 +
11395 +/* context flags */
11396 +
11397 +#define __vx_flags(v)  ((v) ? (v)->vx_flags : 0)
11398 +
11399 +#define vx_current_flags()     __vx_flags(current_vx_info())
11400 +
11401 +#define vx_info_flags(v, m, f) \
11402 +       vs_check_flags(__vx_flags(v), m, f)
11403 +
11404 +#define task_vx_flags(t, m, f) \
11405 +       ((t) && vx_info_flags((t)->vx_info, m, f))
11406 +
11407 +#define vx_flags(m, f) vx_info_flags(current_vx_info(), m, f)
11408 +
11409 +
11410 +/* context caps */
11411 +
11412 +#define __vx_ccaps(v)  ((v) ? (v)->vx_ccaps : 0)
11413 +
11414 +#define vx_current_ccaps()     __vx_ccaps(current_vx_info())
11415 +
11416 +#define vx_info_ccaps(v, c)    (__vx_ccaps(v) & (c))
11417 +
11418 +#define vx_ccaps(c)    vx_info_ccaps(current_vx_info(), (c))
11419 +
11420 +
11421 +
11422 +/* network flags */
11423 +
11424 +#define __nx_flags(n)  ((n) ? (n)->nx_flags : 0)
11425 +
11426 +#define nx_current_flags()     __nx_flags(current_nx_info())
11427 +
11428 +#define nx_info_flags(n, m, f) \
11429 +       vs_check_flags(__nx_flags(n), m, f)
11430 +
11431 +#define task_nx_flags(t, m, f) \
11432 +       ((t) && nx_info_flags((t)->nx_info, m, f))
11433 +
11434 +#define nx_flags(m, f) nx_info_flags(current_nx_info(), m, f)
11435 +
11436 +
11437 +/* network caps */
11438 +
11439 +#define __nx_ncaps(n)  ((n) ? (n)->nx_ncaps : 0)
11440 +
11441 +#define nx_current_ncaps()     __nx_ncaps(current_nx_info())
11442 +
11443 +#define nx_info_ncaps(n, c)    (__nx_ncaps(n) & (c))
11444 +
11445 +#define nx_ncaps(c)    nx_info_ncaps(current_nx_info(), c)
11446 +
11447 +
11448 +/* context mask capabilities */
11449 +
11450 +#define __vx_mcaps(v)  ((v) ? (v)->vx_ccaps >> 32UL : ~0 )
11451 +
11452 +#define vx_info_mcaps(v, c)    (__vx_mcaps(v) & (c))
11453 +
11454 +#define vx_mcaps(c)    vx_info_mcaps(current_vx_info(), c)
11455 +
11456 +
11457 +/* context bcap mask */
11458 +
11459 +#define __vx_bcaps(v)          ((v)->vx_bcaps)
11460 +
11461 +#define vx_current_bcaps()     __vx_bcaps(current_vx_info())
11462 +
11463 +
11464 +/* mask given bcaps */
11465 +
11466 +#define vx_info_mbcaps(v, c)   ((v) ? cap_intersect(__vx_bcaps(v), c) : c)
11467 +
11468 +#define vx_mbcaps(c)           vx_info_mbcaps(current_vx_info(), c)
11469 +
11470 +
11471 +/* masked cap_bset */
11472 +
11473 +#define vx_info_cap_bset(v)    vx_info_mbcaps(v, current->cap_bset)
11474 +
11475 +#define vx_current_cap_bset()  vx_info_cap_bset(current_vx_info())
11476 +
11477 +#if 0
11478 +#define vx_info_mbcap(v, b) \
11479 +       (!vx_info_flags(v, VXF_STATE_SETUP, 0) ? \
11480 +       vx_info_bcaps(v, b) : (b))
11481 +
11482 +#define task_vx_mbcap(t, b) \
11483 +       vx_info_mbcap((t)->vx_info, (t)->b)
11484 +
11485 +#define vx_mbcap(b)    task_vx_mbcap(current, b)
11486 +#endif
11487 +
11488 +#define vx_cap_raised(v, c, f) cap_raised(vx_info_mbcaps(v, c), f)
11489 +
11490 +#define vx_capable(b, c) (capable(b) || \
11491 +       (cap_raised(current_cap(), b) && vx_ccaps(c)))
11492 +
11493 +#define nx_capable(b, c) (capable(b) || \
11494 +       (cap_raised(current_cap(), b) && nx_ncaps(c)))
11495 +
11496 +#define vx_task_initpid(t, n) \
11497 +       ((t)->vx_info && \
11498 +       ((t)->vx_info->vx_initpid == (n)))
11499 +
11500 +#define vx_current_initpid(n)  vx_task_initpid(current, n)
11501 +
11502 +
11503 +/* context unshare mask */
11504 +
11505 +#define __vx_umask(v)          ((v)->vx_umask)
11506 +
11507 +#define vx_current_umask()     __vx_umask(current_vx_info())
11508 +
11509 +#define vx_can_unshare(b, f) (capable(b) || \
11510 +       (cap_raised(current_cap(), b) && \
11511 +       !((f) & ~vx_current_umask())))
11512 +
11513 +
11514 +#define __vx_state(v)  ((v) ? ((v)->vx_state) : 0)
11515 +
11516 +#define vx_info_state(v, m)    (__vx_state(v) & (m))
11517 +
11518 +
11519 +#define __nx_state(n)  ((n) ? ((n)->nx_state) : 0)
11520 +
11521 +#define nx_info_state(n, m)    (__nx_state(n) & (m))
11522 +
11523 +#endif
11524 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cacct.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct.h
11525 --- linux-2.6.36/include/linux/vserver/cacct.h  1970-01-01 01:00:00.000000000 +0100
11526 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct.h    2010-10-21 13:09:36.000000000 +0200
11527 @@ -0,0 +1,15 @@
11528 +#ifndef _VX_CACCT_H
11529 +#define _VX_CACCT_H
11530 +
11531 +
11532 +enum sock_acc_field {
11533 +       VXA_SOCK_UNSPEC = 0,
11534 +       VXA_SOCK_UNIX,
11535 +       VXA_SOCK_INET,
11536 +       VXA_SOCK_INET6,
11537 +       VXA_SOCK_PACKET,
11538 +       VXA_SOCK_OTHER,
11539 +       VXA_SOCK_SIZE   /* array size */
11540 +};
11541 +
11542 +#endif /* _VX_CACCT_H */
11543 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cacct_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_cmd.h
11544 --- linux-2.6.36/include/linux/vserver/cacct_cmd.h      1970-01-01 01:00:00.000000000 +0100
11545 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_cmd.h        2010-10-21 13:09:36.000000000 +0200
11546 @@ -0,0 +1,23 @@
11547 +#ifndef _VX_CACCT_CMD_H
11548 +#define _VX_CACCT_CMD_H
11549 +
11550 +
11551 +/* virtual host info name commands */
11552 +
11553 +#define VCMD_sock_stat         VC_CMD(VSTAT, 5, 0)
11554 +
11555 +struct vcmd_sock_stat_v0 {
11556 +       uint32_t field;
11557 +       uint32_t count[3];
11558 +       uint64_t total[3];
11559 +};
11560 +
11561 +
11562 +#ifdef __KERNEL__
11563 +
11564 +#include <linux/compiler.h>
11565 +
11566 +extern int vc_sock_stat(struct vx_info *, void __user *);
11567 +
11568 +#endif /* __KERNEL__ */
11569 +#endif /* _VX_CACCT_CMD_H */
11570 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cacct_def.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_def.h
11571 --- linux-2.6.36/include/linux/vserver/cacct_def.h      1970-01-01 01:00:00.000000000 +0100
11572 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_def.h        2010-10-21 13:09:36.000000000 +0200
11573 @@ -0,0 +1,43 @@
11574 +#ifndef _VX_CACCT_DEF_H
11575 +#define _VX_CACCT_DEF_H
11576 +
11577 +#include <asm/atomic.h>
11578 +#include <linux/vserver/cacct.h>
11579 +
11580 +
11581 +struct _vx_sock_acc {
11582 +       atomic_long_t count;
11583 +       atomic_long_t total;
11584 +};
11585 +
11586 +/* context sub struct */
11587 +
11588 +struct _vx_cacct {
11589 +       struct _vx_sock_acc sock[VXA_SOCK_SIZE][3];
11590 +       atomic_t slab[8];
11591 +       atomic_t page[6][8];
11592 +};
11593 +
11594 +#ifdef CONFIG_VSERVER_DEBUG
11595 +
11596 +static inline void __dump_vx_cacct(struct _vx_cacct *cacct)
11597 +{
11598 +       int i, j;
11599 +
11600 +       printk("\t_vx_cacct:");
11601 +       for (i = 0; i < 6; i++) {
11602 +               struct _vx_sock_acc *ptr = cacct->sock[i];
11603 +
11604 +               printk("\t [%d] =", i);
11605 +               for (j = 0; j < 3; j++) {
11606 +                       printk(" [%d] = %8lu, %8lu", j,
11607 +                               atomic_long_read(&ptr[j].count),
11608 +                               atomic_long_read(&ptr[j].total));
11609 +               }
11610 +               printk("\n");
11611 +       }
11612 +}
11613 +
11614 +#endif
11615 +
11616 +#endif /* _VX_CACCT_DEF_H */
11617 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cacct_int.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_int.h
11618 --- linux-2.6.36/include/linux/vserver/cacct_int.h      1970-01-01 01:00:00.000000000 +0100
11619 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cacct_int.h        2010-10-21 13:09:36.000000000 +0200
11620 @@ -0,0 +1,21 @@
11621 +#ifndef _VX_CACCT_INT_H
11622 +#define _VX_CACCT_INT_H
11623 +
11624 +
11625 +#ifdef __KERNEL__
11626 +
11627 +static inline
11628 +unsigned long vx_sock_count(struct _vx_cacct *cacct, int type, int pos)
11629 +{
11630 +       return atomic_long_read(&cacct->sock[type][pos].count);
11631 +}
11632 +
11633 +
11634 +static inline
11635 +unsigned long vx_sock_total(struct _vx_cacct *cacct, int type, int pos)
11636 +{
11637 +       return atomic_long_read(&cacct->sock[type][pos].total);
11638 +}
11639 +
11640 +#endif /* __KERNEL__ */
11641 +#endif /* _VX_CACCT_INT_H */
11642 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/check.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/check.h
11643 --- linux-2.6.36/include/linux/vserver/check.h  1970-01-01 01:00:00.000000000 +0100
11644 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/check.h    2010-10-21 13:09:36.000000000 +0200
11645 @@ -0,0 +1,89 @@
11646 +#ifndef _VS_CHECK_H
11647 +#define _VS_CHECK_H
11648 +
11649 +
11650 +#define MAX_S_CONTEXT  65535   /* Arbitrary limit */
11651 +
11652 +#ifdef CONFIG_VSERVER_DYNAMIC_IDS
11653 +#define MIN_D_CONTEXT  49152   /* dynamic contexts start here */
11654 +#else
11655 +#define MIN_D_CONTEXT  65536
11656 +#endif
11657 +
11658 +/* check conditions */
11659 +
11660 +#define VS_ADMIN       0x0001
11661 +#define VS_WATCH       0x0002
11662 +#define VS_HIDE                0x0004
11663 +#define VS_HOSTID      0x0008
11664 +
11665 +#define VS_IDENT       0x0010
11666 +#define VS_EQUIV       0x0020
11667 +#define VS_PARENT      0x0040
11668 +#define VS_CHILD       0x0080
11669 +
11670 +#define VS_ARG_MASK    0x00F0
11671 +
11672 +#define VS_DYNAMIC     0x0100
11673 +#define VS_STATIC      0x0200
11674 +
11675 +#define VS_ATR_MASK    0x0F00
11676 +
11677 +#ifdef CONFIG_VSERVER_PRIVACY
11678 +#define VS_ADMIN_P     (0)
11679 +#define VS_WATCH_P     (0)
11680 +#else
11681 +#define VS_ADMIN_P     VS_ADMIN
11682 +#define VS_WATCH_P     VS_WATCH
11683 +#endif
11684 +
11685 +#define VS_HARDIRQ     0x1000
11686 +#define VS_SOFTIRQ     0x2000
11687 +#define VS_IRQ         0x4000
11688 +
11689 +#define VS_IRQ_MASK    0xF000
11690 +
11691 +#include <linux/hardirq.h>
11692 +
11693 +/*
11694 + * check current context for ADMIN/WATCH and
11695 + * optionally against supplied argument
11696 + */
11697 +static inline int __vs_check(int cid, int id, unsigned int mode)
11698 +{
11699 +       if (mode & VS_ARG_MASK) {
11700 +               if ((mode & VS_IDENT) && (id == cid))
11701 +                       return 1;
11702 +       }
11703 +       if (mode & VS_ATR_MASK) {
11704 +               if ((mode & VS_DYNAMIC) &&
11705 +                       (id >= MIN_D_CONTEXT) &&
11706 +                       (id <= MAX_S_CONTEXT))
11707 +                       return 1;
11708 +               if ((mode & VS_STATIC) &&
11709 +                       (id > 1) && (id < MIN_D_CONTEXT))
11710 +                       return 1;
11711 +       }
11712 +       if (mode & VS_IRQ_MASK) {
11713 +               if ((mode & VS_IRQ) && unlikely(in_interrupt()))
11714 +                       return 1;
11715 +               if ((mode & VS_HARDIRQ) && unlikely(in_irq()))
11716 +                       return 1;
11717 +               if ((mode & VS_SOFTIRQ) && unlikely(in_softirq()))
11718 +                       return 1;
11719 +       }
11720 +       return (((mode & VS_ADMIN) && (cid == 0)) ||
11721 +               ((mode & VS_WATCH) && (cid == 1)) ||
11722 +               ((mode & VS_HOSTID) && (id == 0)));
11723 +}
11724 +
11725 +#define vx_check(c, m) __vs_check(vx_current_xid(), c, (m) | VS_IRQ)
11726 +
11727 +#define vx_weak_check(c, m)    ((m) ? vx_check(c, m) : 1)
11728 +
11729 +
11730 +#define nx_check(c, m) __vs_check(nx_current_nid(), c, m)
11731 +
11732 +#define nx_weak_check(c, m)    ((m) ? nx_check(c, m) : 1)
11733 +
11734 +#endif
11735 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/context.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/context.h
11736 --- linux-2.6.36/include/linux/vserver/context.h        1970-01-01 01:00:00.000000000 +0100
11737 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/context.h  2010-10-21 13:09:36.000000000 +0200
11738 @@ -0,0 +1,184 @@
11739 +#ifndef _VX_CONTEXT_H
11740 +#define _VX_CONTEXT_H
11741 +
11742 +#include <linux/types.h>
11743 +#include <linux/capability.h>
11744 +
11745 +
11746 +/* context flags */
11747 +
11748 +#define VXF_INFO_SCHED         0x00000002
11749 +#define VXF_INFO_NPROC         0x00000004
11750 +#define VXF_INFO_PRIVATE       0x00000008
11751 +
11752 +#define VXF_INFO_INIT          0x00000010
11753 +#define VXF_INFO_HIDE          0x00000020
11754 +#define VXF_INFO_ULIMIT                0x00000040
11755 +#define VXF_INFO_NSPACE                0x00000080
11756 +
11757 +#define VXF_SCHED_HARD         0x00000100
11758 +#define VXF_SCHED_PRIO         0x00000200
11759 +#define VXF_SCHED_PAUSE                0x00000400
11760 +
11761 +#define VXF_VIRT_MEM           0x00010000
11762 +#define VXF_VIRT_UPTIME                0x00020000
11763 +#define VXF_VIRT_CPU           0x00040000
11764 +#define VXF_VIRT_LOAD          0x00080000
11765 +#define VXF_VIRT_TIME          0x00100000
11766 +
11767 +#define VXF_HIDE_MOUNT         0x01000000
11768 +/* was VXF_HIDE_NETIF          0x02000000 */
11769 +#define VXF_HIDE_VINFO         0x04000000
11770 +
11771 +#define VXF_STATE_SETUP                (1ULL << 32)
11772 +#define VXF_STATE_INIT         (1ULL << 33)
11773 +#define VXF_STATE_ADMIN                (1ULL << 34)
11774 +
11775 +#define VXF_SC_HELPER          (1ULL << 36)
11776 +#define VXF_REBOOT_KILL                (1ULL << 37)
11777 +#define VXF_PERSISTENT         (1ULL << 38)
11778 +
11779 +#define VXF_FORK_RSS           (1ULL << 48)
11780 +#define VXF_PROLIFIC           (1ULL << 49)
11781 +
11782 +#define VXF_IGNEG_NICE         (1ULL << 52)
11783 +
11784 +#define VXF_ONE_TIME           (0x0007ULL << 32)
11785 +
11786 +#define VXF_INIT_SET           (VXF_STATE_SETUP | VXF_STATE_INIT | VXF_STATE_ADMIN)
11787 +
11788 +
11789 +/* context migration */
11790 +
11791 +#define VXM_SET_INIT           0x00000001
11792 +#define VXM_SET_REAPER         0x00000002
11793 +
11794 +/* context caps */
11795 +
11796 +#define VXC_CAP_MASK           0x00000000
11797 +
11798 +#define VXC_SET_UTSNAME                0x00000001
11799 +#define VXC_SET_RLIMIT         0x00000002
11800 +#define VXC_FS_SECURITY                0x00000004
11801 +#define VXC_FS_TRUSTED         0x00000008
11802 +#define VXC_TIOCSTI            0x00000010
11803 +
11804 +/* was VXC_RAW_ICMP            0x00000100 */
11805 +#define VXC_SYSLOG             0x00001000
11806 +#define VXC_OOM_ADJUST         0x00002000
11807 +#define VXC_AUDIT_CONTROL      0x00004000
11808 +
11809 +#define VXC_SECURE_MOUNT       0x00010000
11810 +#define VXC_SECURE_REMOUNT     0x00020000
11811 +#define VXC_BINARY_MOUNT       0x00040000
11812 +
11813 +#define VXC_QUOTA_CTL          0x00100000
11814 +#define VXC_ADMIN_MAPPER       0x00200000
11815 +#define VXC_ADMIN_CLOOP                0x00400000
11816 +
11817 +#define VXC_KTHREAD            0x01000000
11818 +#define VXC_NAMESPACE          0x02000000
11819 +
11820 +
11821 +#ifdef __KERNEL__
11822 +
11823 +#include <linux/list.h>
11824 +#include <linux/spinlock.h>
11825 +#include <linux/rcupdate.h>
11826 +
11827 +#include "limit_def.h"
11828 +#include "sched_def.h"
11829 +#include "cvirt_def.h"
11830 +#include "cacct_def.h"
11831 +#include "device_def.h"
11832 +
11833 +#define VX_SPACES      2
11834 +
11835 +struct _vx_info_pc {
11836 +       struct _vx_sched_pc sched_pc;
11837 +       struct _vx_cvirt_pc cvirt_pc;
11838 +};
11839 +
11840 +struct vx_info {
11841 +       struct hlist_node vx_hlist;             /* linked list of contexts */
11842 +       xid_t vx_id;                            /* context id */
11843 +       atomic_t vx_usecnt;                     /* usage count */
11844 +       atomic_t vx_tasks;                      /* tasks count */
11845 +       struct vx_info *vx_parent;              /* parent context */
11846 +       int vx_state;                           /* context state */
11847 +
11848 +       unsigned long vx_nsmask[VX_SPACES];     /* assignment mask */
11849 +       struct nsproxy *vx_nsproxy[VX_SPACES];  /* private namespaces */
11850 +       struct fs_struct *vx_fs[VX_SPACES];     /* private namespace fs */
11851 +
11852 +       uint64_t vx_flags;                      /* context flags */
11853 +       uint64_t vx_ccaps;                      /* context caps (vserver) */
11854 +       kernel_cap_t vx_bcaps;                  /* bounding caps (system) */
11855 +       unsigned long vx_umask;                 /* unshare mask (guest) */
11856 +
11857 +       struct task_struct *vx_reaper;          /* guest reaper process */
11858 +       pid_t vx_initpid;                       /* PID of guest init */
11859 +       int64_t vx_badness_bias;                /* OOM points bias */
11860 +
11861 +       struct _vx_limit limit;                 /* vserver limits */
11862 +       struct _vx_sched sched;                 /* vserver scheduler */
11863 +       struct _vx_cvirt cvirt;                 /* virtual/bias stuff */
11864 +       struct _vx_cacct cacct;                 /* context accounting */
11865 +
11866 +       struct _vx_device dmap;                 /* default device map targets */
11867 +
11868 +#ifndef CONFIG_SMP
11869 +       struct _vx_info_pc info_pc;             /* per cpu data */
11870 +#else
11871 +       struct _vx_info_pc *ptr_pc;             /* per cpu array */
11872 +#endif
11873 +
11874 +       wait_queue_head_t vx_wait;              /* context exit waitqueue */
11875 +       int reboot_cmd;                         /* last sys_reboot() cmd */
11876 +       int exit_code;                          /* last process exit code */
11877 +
11878 +       char vx_name[65];                       /* vserver name */
11879 +};
11880 +
11881 +#ifndef CONFIG_SMP
11882 +#define        vx_ptr_pc(vxi)          (&(vxi)->info_pc)
11883 +#define        vx_per_cpu(vxi, v, id)  vx_ptr_pc(vxi)->v
11884 +#else
11885 +#define        vx_ptr_pc(vxi)          ((vxi)->ptr_pc)
11886 +#define        vx_per_cpu(vxi, v, id)  per_cpu_ptr(vx_ptr_pc(vxi), id)->v
11887 +#endif
11888 +
11889 +#define        vx_cpu(vxi, v)          vx_per_cpu(vxi, v, smp_processor_id())
11890 +
11891 +
11892 +struct vx_info_save {
11893 +       struct vx_info *vxi;
11894 +       xid_t xid;
11895 +};
11896 +
11897 +
11898 +/* status flags */
11899 +
11900 +#define VXS_HASHED     0x0001
11901 +#define VXS_PAUSED     0x0010
11902 +#define VXS_SHUTDOWN   0x0100
11903 +#define VXS_HELPER     0x1000
11904 +#define VXS_RELEASED   0x8000
11905 +
11906 +
11907 +extern void claim_vx_info(struct vx_info *, struct task_struct *);
11908 +extern void release_vx_info(struct vx_info *, struct task_struct *);
11909 +
11910 +extern struct vx_info *lookup_vx_info(int);
11911 +extern struct vx_info *lookup_or_create_vx_info(int);
11912 +
11913 +extern int get_xid_list(int, unsigned int *, int);
11914 +extern int xid_is_hashed(xid_t);
11915 +
11916 +extern int vx_migrate_task(struct task_struct *, struct vx_info *, int);
11917 +
11918 +extern long vs_state_change(struct vx_info *, unsigned int);
11919 +
11920 +
11921 +#endif /* __KERNEL__ */
11922 +#endif /* _VX_CONTEXT_H */
11923 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/context_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/context_cmd.h
11924 --- linux-2.6.36/include/linux/vserver/context_cmd.h    1970-01-01 01:00:00.000000000 +0100
11925 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/context_cmd.h      2010-10-21 13:09:36.000000000 +0200
11926 @@ -0,0 +1,145 @@
11927 +#ifndef _VX_CONTEXT_CMD_H
11928 +#define _VX_CONTEXT_CMD_H
11929 +
11930 +
11931 +/* vinfo commands */
11932 +
11933 +#define VCMD_task_xid          VC_CMD(VINFO, 1, 0)
11934 +
11935 +#ifdef __KERNEL__
11936 +extern int vc_task_xid(uint32_t);
11937 +
11938 +#endif /* __KERNEL__ */
11939 +
11940 +#define VCMD_vx_info           VC_CMD(VINFO, 5, 0)
11941 +
11942 +struct vcmd_vx_info_v0 {
11943 +       uint32_t xid;
11944 +       uint32_t initpid;
11945 +       /* more to come */
11946 +};
11947 +
11948 +#ifdef __KERNEL__
11949 +extern int vc_vx_info(struct vx_info *, void __user *);
11950 +
11951 +#endif /* __KERNEL__ */
11952 +
11953 +#define VCMD_ctx_stat          VC_CMD(VSTAT, 0, 0)
11954 +
11955 +struct vcmd_ctx_stat_v0 {
11956 +       uint32_t usecnt;
11957 +       uint32_t tasks;
11958 +       /* more to come */
11959 +};
11960 +
11961 +#ifdef __KERNEL__
11962 +extern int vc_ctx_stat(struct vx_info *, void __user *);
11963 +
11964 +#endif /* __KERNEL__ */
11965 +
11966 +/* context commands */
11967 +
11968 +#define VCMD_ctx_create_v0     VC_CMD(VPROC, 1, 0)
11969 +#define VCMD_ctx_create                VC_CMD(VPROC, 1, 1)
11970 +
11971 +struct vcmd_ctx_create {
11972 +       uint64_t flagword;
11973 +};
11974 +
11975 +#define VCMD_ctx_migrate_v0    VC_CMD(PROCMIG, 1, 0)
11976 +#define VCMD_ctx_migrate       VC_CMD(PROCMIG, 1, 1)
11977 +
11978 +struct vcmd_ctx_migrate {
11979 +       uint64_t flagword;
11980 +};
11981 +
11982 +#ifdef __KERNEL__
11983 +extern int vc_ctx_create(uint32_t, void __user *);
11984 +extern int vc_ctx_migrate(struct vx_info *, void __user *);
11985 +
11986 +#endif /* __KERNEL__ */
11987 +
11988 +
11989 +/* flag commands */
11990 +
11991 +#define VCMD_get_cflags                VC_CMD(FLAGS, 1, 0)
11992 +#define VCMD_set_cflags                VC_CMD(FLAGS, 2, 0)
11993 +
11994 +struct vcmd_ctx_flags_v0 {
11995 +       uint64_t flagword;
11996 +       uint64_t mask;
11997 +};
11998 +
11999 +#ifdef __KERNEL__
12000 +extern int vc_get_cflags(struct vx_info *, void __user *);
12001 +extern int vc_set_cflags(struct vx_info *, void __user *);
12002 +
12003 +#endif /* __KERNEL__ */
12004 +
12005 +
12006 +/* context caps commands */
12007 +
12008 +#define VCMD_get_ccaps         VC_CMD(FLAGS, 3, 1)
12009 +#define VCMD_set_ccaps         VC_CMD(FLAGS, 4, 1)
12010 +
12011 +struct vcmd_ctx_caps_v1 {
12012 +       uint64_t ccaps;
12013 +       uint64_t cmask;
12014 +};
12015 +
12016 +#ifdef __KERNEL__
12017 +extern int vc_get_ccaps(struct vx_info *, void __user *);
12018 +extern int vc_set_ccaps(struct vx_info *, void __user *);
12019 +
12020 +#endif /* __KERNEL__ */
12021 +
12022 +
12023 +/* bcaps commands */
12024 +
12025 +#define VCMD_get_bcaps         VC_CMD(FLAGS, 9, 0)
12026 +#define VCMD_set_bcaps         VC_CMD(FLAGS, 10, 0)
12027 +
12028 +struct vcmd_bcaps {
12029 +       uint64_t bcaps;
12030 +       uint64_t bmask;
12031 +};
12032 +
12033 +#ifdef __KERNEL__
12034 +extern int vc_get_bcaps(struct vx_info *, void __user *);
12035 +extern int vc_set_bcaps(struct vx_info *, void __user *);
12036 +
12037 +#endif /* __KERNEL__ */
12038 +
12039 +
12040 +/* umask commands */
12041 +
12042 +#define VCMD_get_umask         VC_CMD(FLAGS, 13, 0)
12043 +#define VCMD_set_umask         VC_CMD(FLAGS, 14, 0)
12044 +
12045 +struct vcmd_umask {
12046 +       uint64_t umask;
12047 +       uint64_t mask;
12048 +};
12049 +
12050 +#ifdef __KERNEL__
12051 +extern int vc_get_umask(struct vx_info *, void __user *);
12052 +extern int vc_set_umask(struct vx_info *, void __user *);
12053 +
12054 +#endif /* __KERNEL__ */
12055 +
12056 +
12057 +/* OOM badness */
12058 +
12059 +#define VCMD_get_badness       VC_CMD(MEMCTRL, 5, 0)
12060 +#define VCMD_set_badness       VC_CMD(MEMCTRL, 6, 0)
12061 +
12062 +struct vcmd_badness_v0 {
12063 +       int64_t bias;
12064 +};
12065 +
12066 +#ifdef __KERNEL__
12067 +extern int vc_get_badness(struct vx_info *, void __user *);
12068 +extern int vc_set_badness(struct vx_info *, void __user *);
12069 +
12070 +#endif /* __KERNEL__ */
12071 +#endif /* _VX_CONTEXT_CMD_H */
12072 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cvirt.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt.h
12073 --- linux-2.6.36/include/linux/vserver/cvirt.h  1970-01-01 01:00:00.000000000 +0100
12074 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt.h    2010-10-21 13:09:36.000000000 +0200
12075 @@ -0,0 +1,20 @@
12076 +#ifndef _VX_CVIRT_H
12077 +#define _VX_CVIRT_H
12078 +
12079 +
12080 +#ifdef __KERNEL__
12081 +
12082 +struct timespec;
12083 +
12084 +void vx_vsi_uptime(struct timespec *, struct timespec *);
12085 +
12086 +
12087 +struct vx_info;
12088 +
12089 +void vx_update_load(struct vx_info *);
12090 +
12091 +
12092 +int vx_do_syslog(int, char __user *, int);
12093 +
12094 +#endif /* __KERNEL__ */
12095 +#endif /* _VX_CVIRT_H */
12096 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cvirt_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt_cmd.h
12097 --- linux-2.6.36/include/linux/vserver/cvirt_cmd.h      1970-01-01 01:00:00.000000000 +0100
12098 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt_cmd.h        2010-10-21 13:09:36.000000000 +0200
12099 @@ -0,0 +1,53 @@
12100 +#ifndef _VX_CVIRT_CMD_H
12101 +#define _VX_CVIRT_CMD_H
12102 +
12103 +
12104 +/* virtual host info name commands */
12105 +
12106 +#define VCMD_set_vhi_name      VC_CMD(VHOST, 1, 0)
12107 +#define VCMD_get_vhi_name      VC_CMD(VHOST, 2, 0)
12108 +
12109 +struct vcmd_vhi_name_v0 {
12110 +       uint32_t field;
12111 +       char name[65];
12112 +};
12113 +
12114 +
12115 +enum vhi_name_field {
12116 +       VHIN_CONTEXT = 0,
12117 +       VHIN_SYSNAME,
12118 +       VHIN_NODENAME,
12119 +       VHIN_RELEASE,
12120 +       VHIN_VERSION,
12121 +       VHIN_MACHINE,
12122 +       VHIN_DOMAINNAME,
12123 +};
12124 +
12125 +
12126 +#ifdef __KERNEL__
12127 +
12128 +#include <linux/compiler.h>
12129 +
12130 +extern int vc_set_vhi_name(struct vx_info *, void __user *);
12131 +extern int vc_get_vhi_name(struct vx_info *, void __user *);
12132 +
12133 +#endif /* __KERNEL__ */
12134 +
12135 +#define VCMD_virt_stat         VC_CMD(VSTAT, 3, 0)
12136 +
12137 +struct vcmd_virt_stat_v0 {
12138 +       uint64_t offset;
12139 +       uint64_t uptime;
12140 +       uint32_t nr_threads;
12141 +       uint32_t nr_running;
12142 +       uint32_t nr_uninterruptible;
12143 +       uint32_t nr_onhold;
12144 +       uint32_t nr_forks;
12145 +       uint32_t load[3];
12146 +};
12147 +
12148 +#ifdef __KERNEL__
12149 +extern int vc_virt_stat(struct vx_info *, void __user *);
12150 +
12151 +#endif /* __KERNEL__ */
12152 +#endif /* _VX_CVIRT_CMD_H */
12153 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/cvirt_def.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt_def.h
12154 --- linux-2.6.36/include/linux/vserver/cvirt_def.h      1970-01-01 01:00:00.000000000 +0100
12155 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/cvirt_def.h        2010-10-21 14:26:47.000000000 +0200
12156 @@ -0,0 +1,80 @@
12157 +#ifndef _VX_CVIRT_DEF_H
12158 +#define _VX_CVIRT_DEF_H
12159 +
12160 +#include <linux/jiffies.h>
12161 +#include <linux/spinlock.h>
12162 +#include <linux/wait.h>
12163 +#include <linux/time.h>
12164 +#include <asm/atomic.h>
12165 +
12166 +
12167 +struct _vx_usage_stat {
12168 +       uint64_t user;
12169 +       uint64_t nice;
12170 +       uint64_t system;
12171 +       uint64_t softirq;
12172 +       uint64_t irq;
12173 +       uint64_t idle;
12174 +       uint64_t iowait;
12175 +};
12176 +
12177 +struct _vx_syslog {
12178 +       wait_queue_head_t log_wait;
12179 +       spinlock_t logbuf_lock;         /* lock for the log buffer */
12180 +
12181 +       unsigned long log_start;        /* next char to be read by syslog() */
12182 +       unsigned long con_start;        /* next char to be sent to consoles */
12183 +       unsigned long log_end;  /* most-recently-written-char + 1 */
12184 +       unsigned long logged_chars;     /* #chars since last read+clear operation */
12185 +
12186 +       char log_buf[1024];
12187 +};
12188 +
12189 +
12190 +/* context sub struct */
12191 +
12192 +struct _vx_cvirt {
12193 +       atomic_t nr_threads;            /* number of current threads */
12194 +       atomic_t nr_running;            /* number of running threads */
12195 +       atomic_t nr_uninterruptible;    /* number of uninterruptible threads */
12196 +
12197 +       atomic_t nr_onhold;             /* processes on hold */
12198 +       uint32_t onhold_last;           /* jiffies when put on hold */
12199 +
12200 +       struct timespec bias_ts;        /* time offset to the host */
12201 +       struct timespec bias_idle;
12202 +       struct timespec bias_uptime;    /* context creation point */
12203 +       uint64_t bias_clock;            /* offset in clock_t */
12204 +
12205 +       spinlock_t load_lock;           /* lock for the load averages */
12206 +       atomic_t load_updates;          /* nr of load updates done so far */
12207 +       uint32_t load_last;             /* last time load was calculated */
12208 +       uint32_t load[3];               /* load averages 1,5,15 */
12209 +
12210 +       atomic_t total_forks;           /* number of forks so far */
12211 +
12212 +       struct _vx_syslog syslog;
12213 +};
12214 +
12215 +struct _vx_cvirt_pc {
12216 +       struct _vx_usage_stat cpustat;
12217 +};
12218 +
12219 +
12220 +#ifdef CONFIG_VSERVER_DEBUG
12221 +
12222 +static inline void __dump_vx_cvirt(struct _vx_cvirt *cvirt)
12223 +{
12224 +       printk("\t_vx_cvirt:\n");
12225 +       printk("\t threads: %4d, %4d, %4d, %4d\n",
12226 +               atomic_read(&cvirt->nr_threads),
12227 +               atomic_read(&cvirt->nr_running),
12228 +               atomic_read(&cvirt->nr_uninterruptible),
12229 +               atomic_read(&cvirt->nr_onhold));
12230 +       /* add rest here */
12231 +       printk("\t total_forks = %d\n", atomic_read(&cvirt->total_forks));
12232 +}
12233 +
12234 +#endif
12235 +
12236 +#endif /* _VX_CVIRT_DEF_H */
12237 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/debug.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/debug.h
12238 --- linux-2.6.36/include/linux/vserver/debug.h  1970-01-01 01:00:00.000000000 +0100
12239 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/debug.h    2010-10-21 18:35:25.000000000 +0200
12240 @@ -0,0 +1,140 @@
12241 +#ifndef _VX_DEBUG_H
12242 +#define _VX_DEBUG_H
12243 +
12244 +
12245 +#define VXD_CBIT(n, m) (vx_debug_ ## n & (1 << (m)))
12246 +#define VXD_CMIN(n, m) (vx_debug_ ## n > (m))
12247 +#define VXD_MASK(n, m) (vx_debug_ ## n & (m))
12248 +
12249 +#define VXD_DEV(d)     (d), (d)->bd_inode->i_ino,              \
12250 +                       imajor((d)->bd_inode), iminor((d)->bd_inode)
12251 +#define VXF_DEV                "%p[%lu,%d:%d]"
12252 +
12253 +#if    defined(CONFIG_QUOTES_UTF8)
12254 +#define        VS_Q_LQM        "\xc2\xbb"
12255 +#define        VS_Q_RQM        "\xc2\xab"
12256 +#elif  defined(CONFIG_QUOTES_ASCII)
12257 +#define        VS_Q_LQM        "\x27"
12258 +#define        VS_Q_RQM        "\x27"
12259 +#else
12260 +#define        VS_Q_LQM        "\xbb"
12261 +#define        VS_Q_RQM        "\xab"
12262 +#endif
12263 +
12264 +#define        VS_Q(f)         VS_Q_LQM f VS_Q_RQM
12265 +
12266 +
12267 +#define vxd_path(p)                                            \
12268 +       ({ static char _buffer[PATH_MAX];                       \
12269 +          d_path(p, _buffer, sizeof(_buffer)); })
12270 +
12271 +#define vxd_cond_path(n)                                       \
12272 +       ((n) ? vxd_path(&(n)->path) : "<null>" )
12273 +
12274 +
12275 +#ifdef CONFIG_VSERVER_DEBUG
12276 +
12277 +extern unsigned int vx_debug_switch;
12278 +extern unsigned int vx_debug_xid;
12279 +extern unsigned int vx_debug_nid;
12280 +extern unsigned int vx_debug_tag;
12281 +extern unsigned int vx_debug_net;
12282 +extern unsigned int vx_debug_limit;
12283 +extern unsigned int vx_debug_cres;
12284 +extern unsigned int vx_debug_dlim;
12285 +extern unsigned int vx_debug_quota;
12286 +extern unsigned int vx_debug_cvirt;
12287 +extern unsigned int vx_debug_space;
12288 +extern unsigned int vx_debug_misc;
12289 +
12290 +
12291 +#define VX_LOGLEVEL    "vxD: "
12292 +#define VX_PROC_FMT    "%p: "
12293 +#define VX_PROCESS     current
12294 +
12295 +#define vxdprintk(c, f, x...)                                  \
12296 +       do {                                                    \
12297 +               if (c)                                          \
12298 +                       printk(VX_LOGLEVEL VX_PROC_FMT f "\n",  \
12299 +                               VX_PROCESS , ##x);              \
12300 +       } while (0)
12301 +
12302 +#define vxlprintk(c, f, x...)                                  \
12303 +       do {                                                    \
12304 +               if (c)                                          \
12305 +                       printk(VX_LOGLEVEL f " @%s:%d\n", x);   \
12306 +       } while (0)
12307 +
12308 +#define vxfprintk(c, f, x...)                                  \
12309 +       do {                                                    \
12310 +               if (c)                                          \
12311 +                       printk(VX_LOGLEVEL f " %s@%s:%d\n", x); \
12312 +       } while (0)
12313 +
12314 +
12315 +struct vx_info;
12316 +
12317 +void dump_vx_info(struct vx_info *, int);
12318 +void dump_vx_info_inactive(int);
12319 +
12320 +#else  /* CONFIG_VSERVER_DEBUG */
12321 +
12322 +#define vx_debug_switch 0
12323 +#define vx_debug_xid   0
12324 +#define vx_debug_nid   0
12325 +#define vx_debug_tag   0
12326 +#define vx_debug_net   0
12327 +#define vx_debug_limit 0
12328 +#define vx_debug_cres  0
12329 +#define vx_debug_dlim  0
12330 +#define vx_debug_cvirt 0
12331 +
12332 +#define vxdprintk(x...) do { } while (0)
12333 +#define vxlprintk(x...) do { } while (0)
12334 +#define vxfprintk(x...) do { } while (0)
12335 +
12336 +#endif /* CONFIG_VSERVER_DEBUG */
12337 +
12338 +
12339 +#ifdef CONFIG_VSERVER_WARN
12340 +
12341 +#define VX_WARNLEVEL   KERN_WARNING "vxW: "
12342 +#define VX_WARN_TASK   "[" VS_Q("%s") ",%u:#%u|%u|%u] "
12343 +#define VX_WARN_XID    "[xid #%u] "
12344 +#define VX_WARN_NID    "[nid #%u] "
12345 +#define VX_WARN_TAG    "[tag #%u] "
12346 +
12347 +#define vxwprintk(c, f, x...)                                  \
12348 +       do {                                                    \
12349 +               if (c)                                          \
12350 +                       printk(VX_WARNLEVEL f "\n", ##x);       \
12351 +       } while (0)
12352 +
12353 +#else  /* CONFIG_VSERVER_WARN */
12354 +
12355 +#define vxwprintk(x...) do { } while (0)
12356 +
12357 +#endif /* CONFIG_VSERVER_WARN */
12358 +
12359 +#define vxwprintk_task(c, f, x...)                             \
12360 +       vxwprintk(c, VX_WARN_TASK f,                            \
12361 +               current->comm, current->pid,                    \
12362 +               current->xid, current->nid, current->tag, ##x)
12363 +#define vxwprintk_xid(c, f, x...)                              \
12364 +       vxwprintk(c, VX_WARN_XID f, current->xid, x)
12365 +#define vxwprintk_nid(c, f, x...)                              \
12366 +       vxwprintk(c, VX_WARN_NID f, current->nid, x)
12367 +#define vxwprintk_tag(c, f, x...)                              \
12368 +       vxwprintk(c, VX_WARN_TAG f, current->tag, x)
12369 +
12370 +#ifdef CONFIG_VSERVER_DEBUG
12371 +#define vxd_assert_lock(l)     assert_spin_locked(l)
12372 +#define vxd_assert(c, f, x...) vxlprintk(!(c), \
12373 +       "assertion [" f "] failed.", ##x, __FILE__, __LINE__)
12374 +#else
12375 +#define vxd_assert_lock(l)     do { } while (0)
12376 +#define vxd_assert(c, f, x...) do { } while (0)
12377 +#endif
12378 +
12379 +
12380 +#endif /* _VX_DEBUG_H */
12381 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/debug_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/debug_cmd.h
12382 --- linux-2.6.36/include/linux/vserver/debug_cmd.h      1970-01-01 01:00:00.000000000 +0100
12383 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/debug_cmd.h        2010-10-21 13:09:36.000000000 +0200
12384 @@ -0,0 +1,58 @@
12385 +#ifndef _VX_DEBUG_CMD_H
12386 +#define _VX_DEBUG_CMD_H
12387 +
12388 +
12389 +/* debug commands */
12390 +
12391 +#define VCMD_dump_history      VC_CMD(DEBUG, 1, 0)
12392 +
12393 +#define VCMD_read_history      VC_CMD(DEBUG, 5, 0)
12394 +#define VCMD_read_monitor      VC_CMD(DEBUG, 6, 0)
12395 +
12396 +struct  vcmd_read_history_v0 {
12397 +       uint32_t index;
12398 +       uint32_t count;
12399 +       char __user *data;
12400 +};
12401 +
12402 +struct  vcmd_read_monitor_v0 {
12403 +       uint32_t index;
12404 +       uint32_t count;
12405 +       char __user *data;
12406 +};
12407 +
12408 +
12409 +#ifdef __KERNEL__
12410 +
12411 +#ifdef CONFIG_COMPAT
12412 +
12413 +#include <asm/compat.h>
12414 +
12415 +struct vcmd_read_history_v0_x32 {
12416 +       uint32_t index;
12417 +       uint32_t count;
12418 +       compat_uptr_t data_ptr;
12419 +};
12420 +
12421 +struct vcmd_read_monitor_v0_x32 {
12422 +       uint32_t index;
12423 +       uint32_t count;
12424 +       compat_uptr_t data_ptr;
12425 +};
12426 +
12427 +#endif  /* CONFIG_COMPAT */
12428 +
12429 +extern int vc_dump_history(uint32_t);
12430 +
12431 +extern int vc_read_history(uint32_t, void __user *);
12432 +extern int vc_read_monitor(uint32_t, void __user *);
12433 +
12434 +#ifdef CONFIG_COMPAT
12435 +
12436 +extern int vc_read_history_x32(uint32_t, void __user *);
12437 +extern int vc_read_monitor_x32(uint32_t, void __user *);
12438 +
12439 +#endif  /* CONFIG_COMPAT */
12440 +
12441 +#endif /* __KERNEL__ */
12442 +#endif /* _VX_DEBUG_CMD_H */
12443 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/device.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device.h
12444 --- linux-2.6.36/include/linux/vserver/device.h 1970-01-01 01:00:00.000000000 +0100
12445 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device.h   2010-10-21 13:09:36.000000000 +0200
12446 @@ -0,0 +1,15 @@
12447 +#ifndef _VX_DEVICE_H
12448 +#define _VX_DEVICE_H
12449 +
12450 +
12451 +#define DATTR_CREATE   0x00000001
12452 +#define DATTR_OPEN     0x00000002
12453 +
12454 +#define DATTR_REMAP    0x00000010
12455 +
12456 +#define DATTR_MASK     0x00000013
12457 +
12458 +
12459 +#else  /* _VX_DEVICE_H */
12460 +#warning duplicate inclusion
12461 +#endif /* _VX_DEVICE_H */
12462 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/device_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device_cmd.h
12463 --- linux-2.6.36/include/linux/vserver/device_cmd.h     1970-01-01 01:00:00.000000000 +0100
12464 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device_cmd.h       2010-10-21 13:09:36.000000000 +0200
12465 @@ -0,0 +1,44 @@
12466 +#ifndef _VX_DEVICE_CMD_H
12467 +#define _VX_DEVICE_CMD_H
12468 +
12469 +
12470 +/*  device vserver commands */
12471 +
12472 +#define VCMD_set_mapping       VC_CMD(DEVICE, 1, 0)
12473 +#define VCMD_unset_mapping     VC_CMD(DEVICE, 2, 0)
12474 +
12475 +struct vcmd_set_mapping_v0 {
12476 +       const char __user *device;
12477 +       const char __user *target;
12478 +       uint32_t flags;
12479 +};
12480 +
12481 +
12482 +#ifdef __KERNEL__
12483 +
12484 +#ifdef CONFIG_COMPAT
12485 +
12486 +#include <asm/compat.h>
12487 +
12488 +struct vcmd_set_mapping_v0_x32 {
12489 +       compat_uptr_t device_ptr;
12490 +       compat_uptr_t target_ptr;
12491 +       uint32_t flags;
12492 +};
12493 +
12494 +#endif /* CONFIG_COMPAT */
12495 +
12496 +#include <linux/compiler.h>
12497 +
12498 +extern int vc_set_mapping(struct vx_info *, void __user *);
12499 +extern int vc_unset_mapping(struct vx_info *, void __user *);
12500 +
12501 +#ifdef CONFIG_COMPAT
12502 +
12503 +extern int vc_set_mapping_x32(struct vx_info *, void __user *);
12504 +extern int vc_unset_mapping_x32(struct vx_info *, void __user *);
12505 +
12506 +#endif /* CONFIG_COMPAT */
12507 +
12508 +#endif /* __KERNEL__ */
12509 +#endif /* _VX_DEVICE_CMD_H */
12510 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/device_def.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device_def.h
12511 --- linux-2.6.36/include/linux/vserver/device_def.h     1970-01-01 01:00:00.000000000 +0100
12512 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/device_def.h       2010-10-21 13:09:36.000000000 +0200
12513 @@ -0,0 +1,17 @@
12514 +#ifndef _VX_DEVICE_DEF_H
12515 +#define _VX_DEVICE_DEF_H
12516 +
12517 +#include <linux/types.h>
12518 +
12519 +struct vx_dmap_target {
12520 +       dev_t target;
12521 +       uint32_t flags;
12522 +};
12523 +
12524 +struct _vx_device {
12525 +#ifdef CONFIG_VSERVER_DEVICE
12526 +       struct vx_dmap_target targets[2];
12527 +#endif
12528 +};
12529 +
12530 +#endif /* _VX_DEVICE_DEF_H */
12531 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/dlimit.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/dlimit.h
12532 --- linux-2.6.36/include/linux/vserver/dlimit.h 1970-01-01 01:00:00.000000000 +0100
12533 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/dlimit.h   2010-10-21 13:09:36.000000000 +0200
12534 @@ -0,0 +1,54 @@
12535 +#ifndef _VX_DLIMIT_H
12536 +#define _VX_DLIMIT_H
12537 +
12538 +#include "switch.h"
12539 +
12540 +
12541 +#ifdef __KERNEL__
12542 +
12543 +/*      keep in sync with CDLIM_INFINITY       */
12544 +
12545 +#define DLIM_INFINITY          (~0ULL)
12546 +
12547 +#include <linux/spinlock.h>
12548 +#include <linux/rcupdate.h>
12549 +
12550 +struct super_block;
12551 +
12552 +struct dl_info {
12553 +       struct hlist_node dl_hlist;             /* linked list of contexts */
12554 +       struct rcu_head dl_rcu;                 /* the rcu head */
12555 +       tag_t dl_tag;                           /* context tag */
12556 +       atomic_t dl_usecnt;                     /* usage count */
12557 +       atomic_t dl_refcnt;                     /* reference count */
12558 +
12559 +       struct super_block *dl_sb;              /* associated superblock */
12560 +
12561 +       spinlock_t dl_lock;                     /* protect the values */
12562 +
12563 +       unsigned long long dl_space_used;       /* used space in bytes */
12564 +       unsigned long long dl_space_total;      /* maximum space in bytes */
12565 +       unsigned long dl_inodes_used;           /* used inodes */
12566 +       unsigned long dl_inodes_total;          /* maximum inodes */
12567 +
12568 +       unsigned int dl_nrlmult;                /* non root limit mult */
12569 +};
12570 +
12571 +struct rcu_head;
12572 +
12573 +extern void rcu_free_dl_info(struct rcu_head *);
12574 +extern void unhash_dl_info(struct dl_info *);
12575 +
12576 +extern struct dl_info *locate_dl_info(struct super_block *, tag_t);
12577 +
12578 +
12579 +struct kstatfs;
12580 +
12581 +extern void vx_vsi_statfs(struct super_block *, struct kstatfs *);
12582 +
12583 +typedef uint64_t dlsize_t;
12584 +
12585 +#endif /* __KERNEL__ */
12586 +#else  /* _VX_DLIMIT_H */
12587 +#warning duplicate inclusion
12588 +#endif /* _VX_DLIMIT_H */
12589 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/dlimit_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/dlimit_cmd.h
12590 --- linux-2.6.36/include/linux/vserver/dlimit_cmd.h     1970-01-01 01:00:00.000000000 +0100
12591 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/dlimit_cmd.h       2010-10-21 13:09:36.000000000 +0200
12592 @@ -0,0 +1,109 @@
12593 +#ifndef _VX_DLIMIT_CMD_H
12594 +#define _VX_DLIMIT_CMD_H
12595 +
12596 +
12597 +/*  dlimit vserver commands */
12598 +
12599 +#define VCMD_add_dlimit                VC_CMD(DLIMIT, 1, 0)
12600 +#define VCMD_rem_dlimit                VC_CMD(DLIMIT, 2, 0)
12601 +
12602 +#define VCMD_set_dlimit                VC_CMD(DLIMIT, 5, 0)
12603 +#define VCMD_get_dlimit                VC_CMD(DLIMIT, 6, 0)
12604 +
12605 +struct vcmd_ctx_dlimit_base_v0 {
12606 +       const char __user *name;
12607 +       uint32_t flags;
12608 +};
12609 +
12610 +struct vcmd_ctx_dlimit_v0 {
12611 +       const char __user *name;
12612 +       uint32_t space_used;                    /* used space in kbytes */
12613 +       uint32_t space_total;                   /* maximum space in kbytes */
12614 +       uint32_t inodes_used;                   /* used inodes */
12615 +       uint32_t inodes_total;                  /* maximum inodes */
12616 +       uint32_t reserved;                      /* reserved for root in % */
12617 +       uint32_t flags;
12618 +};
12619 +
12620 +#define CDLIM_UNSET            ((uint32_t)0UL)
12621 +#define CDLIM_INFINITY         ((uint32_t)~0UL)
12622 +#define CDLIM_KEEP             ((uint32_t)~1UL)
12623 +
12624 +#define DLIME_UNIT     0
12625 +#define DLIME_KILO     1
12626 +#define DLIME_MEGA     2
12627 +#define DLIME_GIGA     3
12628 +
12629 +#define DLIMF_SHIFT    0x10
12630 +
12631 +#define DLIMS_USED     0
12632 +#define DLIMS_TOTAL    2
12633 +
12634 +static inline
12635 +uint64_t dlimit_space_32to64(uint32_t val, uint32_t flags, int shift)
12636 +{
12637 +       int exp = (flags & DLIMF_SHIFT) ?
12638 +               (flags >> shift) & DLIME_GIGA : DLIME_KILO;
12639 +       return ((uint64_t)val) << (10 * exp);
12640 +}
12641 +
12642 +static inline
12643 +uint32_t dlimit_space_64to32(uint64_t val, uint32_t *flags, int shift)
12644 +{
12645 +       int exp = 0;
12646 +
12647 +       if (*flags & DLIMF_SHIFT) {
12648 +               while (val > (1LL << 32) && (exp < 3)) {
12649 +                       val >>= 10;
12650 +                       exp++;
12651 +               }
12652 +               *flags &= ~(DLIME_GIGA << shift);
12653 +               *flags |= exp << shift;
12654 +       } else
12655 +               val >>= 10;
12656 +       return val;
12657 +}
12658 +
12659 +#ifdef __KERNEL__
12660 +
12661 +#ifdef CONFIG_COMPAT
12662 +
12663 +#include <asm/compat.h>
12664 +
12665 +struct vcmd_ctx_dlimit_base_v0_x32 {
12666 +       compat_uptr_t name_ptr;
12667 +       uint32_t flags;
12668 +};
12669 +
12670 +struct vcmd_ctx_dlimit_v0_x32 {
12671 +       compat_uptr_t name_ptr;
12672 +       uint32_t space_used;                    /* used space in kbytes */
12673 +       uint32_t space_total;                   /* maximum space in kbytes */
12674 +       uint32_t inodes_used;                   /* used inodes */
12675 +       uint32_t inodes_total;                  /* maximum inodes */
12676 +       uint32_t reserved;                      /* reserved for root in % */
12677 +       uint32_t flags;
12678 +};
12679 +
12680 +#endif /* CONFIG_COMPAT */
12681 +
12682 +#include <linux/compiler.h>
12683 +
12684 +extern int vc_add_dlimit(uint32_t, void __user *);
12685 +extern int vc_rem_dlimit(uint32_t, void __user *);
12686 +
12687 +extern int vc_set_dlimit(uint32_t, void __user *);
12688 +extern int vc_get_dlimit(uint32_t, void __user *);
12689 +
12690 +#ifdef CONFIG_COMPAT
12691 +
12692 +extern int vc_add_dlimit_x32(uint32_t, void __user *);
12693 +extern int vc_rem_dlimit_x32(uint32_t, void __user *);
12694 +
12695 +extern int vc_set_dlimit_x32(uint32_t, void __user *);
12696 +extern int vc_get_dlimit_x32(uint32_t, void __user *);
12697 +
12698 +#endif /* CONFIG_COMPAT */
12699 +
12700 +#endif /* __KERNEL__ */
12701 +#endif /* _VX_DLIMIT_CMD_H */
12702 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/global.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/global.h
12703 --- linux-2.6.36/include/linux/vserver/global.h 1970-01-01 01:00:00.000000000 +0100
12704 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/global.h   2010-10-21 13:09:36.000000000 +0200
12705 @@ -0,0 +1,19 @@
12706 +#ifndef _VX_GLOBAL_H
12707 +#define _VX_GLOBAL_H
12708 +
12709 +
12710 +extern atomic_t vx_global_ctotal;
12711 +extern atomic_t vx_global_cactive;
12712 +
12713 +extern atomic_t nx_global_ctotal;
12714 +extern atomic_t nx_global_cactive;
12715 +
12716 +extern atomic_t vs_global_nsproxy;
12717 +extern atomic_t vs_global_fs;
12718 +extern atomic_t vs_global_mnt_ns;
12719 +extern atomic_t vs_global_uts_ns;
12720 +extern atomic_t vs_global_user_ns;
12721 +extern atomic_t vs_global_pid_ns;
12722 +
12723 +
12724 +#endif /* _VX_GLOBAL_H */
12725 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/history.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/history.h
12726 --- linux-2.6.36/include/linux/vserver/history.h        1970-01-01 01:00:00.000000000 +0100
12727 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/history.h  2010-10-21 13:09:36.000000000 +0200
12728 @@ -0,0 +1,197 @@
12729 +#ifndef _VX_HISTORY_H
12730 +#define _VX_HISTORY_H
12731 +
12732 +
12733 +enum {
12734 +       VXH_UNUSED = 0,
12735 +       VXH_THROW_OOPS = 1,
12736 +
12737 +       VXH_GET_VX_INFO,
12738 +       VXH_PUT_VX_INFO,
12739 +       VXH_INIT_VX_INFO,
12740 +       VXH_SET_VX_INFO,
12741 +       VXH_CLR_VX_INFO,
12742 +       VXH_CLAIM_VX_INFO,
12743 +       VXH_RELEASE_VX_INFO,
12744 +       VXH_ALLOC_VX_INFO,
12745 +       VXH_DEALLOC_VX_INFO,
12746 +       VXH_HASH_VX_INFO,
12747 +       VXH_UNHASH_VX_INFO,
12748 +       VXH_LOC_VX_INFO,
12749 +       VXH_LOOKUP_VX_INFO,
12750 +       VXH_CREATE_VX_INFO,
12751 +};
12752 +
12753 +struct _vxhe_vxi {
12754 +       struct vx_info *ptr;
12755 +       unsigned xid;
12756 +       unsigned usecnt;
12757 +       unsigned tasks;
12758 +};
12759 +
12760 +struct _vxhe_set_clr {
12761 +       void *data;
12762 +};
12763 +
12764 +struct _vxhe_loc_lookup {
12765 +       unsigned arg;
12766 +};
12767 +
12768 +struct _vx_hist_entry {
12769 +       void *loc;
12770 +       unsigned short seq;
12771 +       unsigned short type;
12772 +       struct _vxhe_vxi vxi;
12773 +       union {
12774 +               struct _vxhe_set_clr sc;
12775 +               struct _vxhe_loc_lookup ll;
12776 +       };
12777 +};
12778 +
12779 +#ifdef CONFIG_VSERVER_HISTORY
12780 +
12781 +extern unsigned volatile int vxh_active;
12782 +
12783 +struct _vx_hist_entry *vxh_advance(void *loc);
12784 +
12785 +
12786 +static inline
12787 +void   __vxh_copy_vxi(struct _vx_hist_entry *entry, struct vx_info *vxi)
12788 +{
12789 +       entry->vxi.ptr = vxi;
12790 +       if (vxi) {
12791 +               entry->vxi.usecnt = atomic_read(&vxi->vx_usecnt);
12792 +               entry->vxi.tasks = atomic_read(&vxi->vx_tasks);
12793 +               entry->vxi.xid = vxi->vx_id;
12794 +       }
12795 +}
12796 +
12797 +
12798 +#define        __HERE__ current_text_addr()
12799 +
12800 +#define __VXH_BODY(__type, __data, __here)     \
12801 +       struct _vx_hist_entry *entry;           \
12802 +                                               \
12803 +       preempt_disable();                      \
12804 +       entry = vxh_advance(__here);            \
12805 +       __data;                                 \
12806 +       entry->type = __type;                   \
12807 +       preempt_enable();
12808 +
12809 +
12810 +       /* pass vxi only */
12811 +
12812 +#define __VXH_SMPL                             \
12813 +       __vxh_copy_vxi(entry, vxi)
12814 +
12815 +static inline
12816 +void   __vxh_smpl(struct vx_info *vxi, int __type, void *__here)
12817 +{
12818 +       __VXH_BODY(__type, __VXH_SMPL, __here)
12819 +}
12820 +
12821 +       /* pass vxi and data (void *) */
12822 +
12823 +#define __VXH_DATA                             \
12824 +       __vxh_copy_vxi(entry, vxi);             \
12825 +       entry->sc.data = data
12826 +
12827 +static inline
12828 +void   __vxh_data(struct vx_info *vxi, void *data,
12829 +                       int __type, void *__here)
12830 +{
12831 +       __VXH_BODY(__type, __VXH_DATA, __here)
12832 +}
12833 +
12834 +       /* pass vxi and arg (long) */
12835 +
12836 +#define __VXH_LONG                             \
12837 +       __vxh_copy_vxi(entry, vxi);             \
12838 +       entry->ll.arg = arg
12839 +
12840 +static inline
12841 +void   __vxh_long(struct vx_info *vxi, long arg,
12842 +                       int __type, void *__here)
12843 +{
12844 +       __VXH_BODY(__type, __VXH_LONG, __here)
12845 +}
12846 +
12847 +
12848 +static inline
12849 +void   __vxh_throw_oops(void *__here)
12850 +{
12851 +       __VXH_BODY(VXH_THROW_OOPS, {}, __here);
12852 +       /* prevent further acquisition */
12853 +       vxh_active = 0;
12854 +}
12855 +
12856 +
12857 +#define vxh_throw_oops()       __vxh_throw_oops(__HERE__);
12858 +
12859 +#define __vxh_get_vx_info(v, h)        __vxh_smpl(v, VXH_GET_VX_INFO, h);
12860 +#define __vxh_put_vx_info(v, h)        __vxh_smpl(v, VXH_PUT_VX_INFO, h);
12861 +
12862 +#define __vxh_init_vx_info(v, d, h) \
12863 +       __vxh_data(v, d, VXH_INIT_VX_INFO, h);
12864 +#define __vxh_set_vx_info(v, d, h) \
12865 +       __vxh_data(v, d, VXH_SET_VX_INFO, h);
12866 +#define __vxh_clr_vx_info(v, d, h) \
12867 +       __vxh_data(v, d, VXH_CLR_VX_INFO, h);
12868 +
12869 +#define __vxh_claim_vx_info(v, d, h) \
12870 +       __vxh_data(v, d, VXH_CLAIM_VX_INFO, h);
12871 +#define __vxh_release_vx_info(v, d, h) \
12872 +       __vxh_data(v, d, VXH_RELEASE_VX_INFO, h);
12873 +
12874 +#define vxh_alloc_vx_info(v) \
12875 +       __vxh_smpl(v, VXH_ALLOC_VX_INFO, __HERE__);
12876 +#define vxh_dealloc_vx_info(v) \
12877 +       __vxh_smpl(v, VXH_DEALLOC_VX_INFO, __HERE__);
12878 +
12879 +#define vxh_hash_vx_info(v) \
12880 +       __vxh_smpl(v, VXH_HASH_VX_INFO, __HERE__);
12881 +#define vxh_unhash_vx_info(v) \
12882 +       __vxh_smpl(v, VXH_UNHASH_VX_INFO, __HERE__);
12883 +
12884 +#define vxh_loc_vx_info(v, l) \
12885 +       __vxh_long(v, l, VXH_LOC_VX_INFO, __HERE__);
12886 +#define vxh_lookup_vx_info(v, l) \
12887 +       __vxh_long(v, l, VXH_LOOKUP_VX_INFO, __HERE__);
12888 +#define vxh_create_vx_info(v, l) \
12889 +       __vxh_long(v, l, VXH_CREATE_VX_INFO, __HERE__);
12890 +
12891 +extern void vxh_dump_history(void);
12892 +
12893 +
12894 +#else  /* CONFIG_VSERVER_HISTORY */
12895 +
12896 +#define        __HERE__        0
12897 +
12898 +#define vxh_throw_oops()               do { } while (0)
12899 +
12900 +#define __vxh_get_vx_info(v, h)                do { } while (0)
12901 +#define __vxh_put_vx_info(v, h)                do { } while (0)
12902 +
12903 +#define __vxh_init_vx_info(v, d, h)    do { } while (0)
12904 +#define __vxh_set_vx_info(v, d, h)     do { } while (0)
12905 +#define __vxh_clr_vx_info(v, d, h)     do { } while (0)
12906 +
12907 +#define __vxh_claim_vx_info(v, d, h)   do { } while (0)
12908 +#define __vxh_release_vx_info(v, d, h) do { } while (0)
12909 +
12910 +#define vxh_alloc_vx_info(v)           do { } while (0)
12911 +#define vxh_dealloc_vx_info(v)         do { } while (0)
12912 +
12913 +#define vxh_hash_vx_info(v)            do { } while (0)
12914 +#define vxh_unhash_vx_info(v)          do { } while (0)
12915 +
12916 +#define vxh_loc_vx_info(v, l)          do { } while (0)
12917 +#define vxh_lookup_vx_info(v, l)       do { } while (0)
12918 +#define vxh_create_vx_info(v, l)       do { } while (0)
12919 +
12920 +#define vxh_dump_history()             do { } while (0)
12921 +
12922 +
12923 +#endif /* CONFIG_VSERVER_HISTORY */
12924 +
12925 +#endif /* _VX_HISTORY_H */
12926 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/inode.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/inode.h
12927 --- linux-2.6.36/include/linux/vserver/inode.h  1970-01-01 01:00:00.000000000 +0100
12928 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/inode.h    2010-10-21 13:09:36.000000000 +0200
12929 @@ -0,0 +1,39 @@
12930 +#ifndef _VX_INODE_H
12931 +#define _VX_INODE_H
12932 +
12933 +
12934 +#define IATTR_TAG      0x01000000
12935 +
12936 +#define IATTR_ADMIN    0x00000001
12937 +#define IATTR_WATCH    0x00000002
12938 +#define IATTR_HIDE     0x00000004
12939 +#define IATTR_FLAGS    0x00000007
12940 +
12941 +#define IATTR_BARRIER  0x00010000
12942 +#define IATTR_IXUNLINK 0x00020000
12943 +#define IATTR_IMMUTABLE 0x00040000
12944 +#define IATTR_COW      0x00080000
12945 +
12946 +#ifdef __KERNEL__
12947 +
12948 +
12949 +#ifdef CONFIG_VSERVER_PROC_SECURE
12950 +#define IATTR_PROC_DEFAULT     ( IATTR_ADMIN | IATTR_HIDE )
12951 +#define IATTR_PROC_SYMLINK     ( IATTR_ADMIN )
12952 +#else
12953 +#define IATTR_PROC_DEFAULT     ( IATTR_ADMIN )
12954 +#define IATTR_PROC_SYMLINK     ( IATTR_ADMIN )
12955 +#endif
12956 +
12957 +#define vx_hide_check(c, m)    (((m) & IATTR_HIDE) ? vx_check(c, m) : 1)
12958 +
12959 +#endif /* __KERNEL__ */
12960 +
12961 +/* inode ioctls */
12962 +
12963 +#define FIOC_GETXFLG   _IOR('x', 5, long)
12964 +#define FIOC_SETXFLG   _IOW('x', 6, long)
12965 +
12966 +#else  /* _VX_INODE_H */
12967 +#warning duplicate inclusion
12968 +#endif /* _VX_INODE_H */
12969 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/inode_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/inode_cmd.h
12970 --- linux-2.6.36/include/linux/vserver/inode_cmd.h      1970-01-01 01:00:00.000000000 +0100
12971 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/inode_cmd.h        2010-10-21 13:09:36.000000000 +0200
12972 @@ -0,0 +1,59 @@
12973 +#ifndef _VX_INODE_CMD_H
12974 +#define _VX_INODE_CMD_H
12975 +
12976 +
12977 +/*  inode vserver commands */
12978 +
12979 +#define VCMD_get_iattr         VC_CMD(INODE, 1, 1)
12980 +#define VCMD_set_iattr         VC_CMD(INODE, 2, 1)
12981 +
12982 +#define VCMD_fget_iattr                VC_CMD(INODE, 3, 0)
12983 +#define VCMD_fset_iattr                VC_CMD(INODE, 4, 0)
12984 +
12985 +struct vcmd_ctx_iattr_v1 {
12986 +       const char __user *name;
12987 +       uint32_t tag;
12988 +       uint32_t flags;
12989 +       uint32_t mask;
12990 +};
12991 +
12992 +struct vcmd_ctx_fiattr_v0 {
12993 +       uint32_t tag;
12994 +       uint32_t flags;
12995 +       uint32_t mask;
12996 +};
12997 +
12998 +
12999 +#ifdef __KERNEL__
13000 +
13001 +
13002 +#ifdef CONFIG_COMPAT
13003 +
13004 +#include <asm/compat.h>
13005 +
13006 +struct vcmd_ctx_iattr_v1_x32 {
13007 +       compat_uptr_t name_ptr;
13008 +       uint32_t tag;
13009 +       uint32_t flags;
13010 +       uint32_t mask;
13011 +};
13012 +
13013 +#endif /* CONFIG_COMPAT */
13014 +
13015 +#include <linux/compiler.h>
13016 +
13017 +extern int vc_get_iattr(void __user *);
13018 +extern int vc_set_iattr(void __user *);
13019 +
13020 +extern int vc_fget_iattr(uint32_t, void __user *);
13021 +extern int vc_fset_iattr(uint32_t, void __user *);
13022 +
13023 +#ifdef CONFIG_COMPAT
13024 +
13025 +extern int vc_get_iattr_x32(void __user *);
13026 +extern int vc_set_iattr_x32(void __user *);
13027 +
13028 +#endif /* CONFIG_COMPAT */
13029 +
13030 +#endif /* __KERNEL__ */
13031 +#endif /* _VX_INODE_CMD_H */
13032 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/limit.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit.h
13033 --- linux-2.6.36/include/linux/vserver/limit.h  1970-01-01 01:00:00.000000000 +0100
13034 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit.h    2010-10-21 13:09:36.000000000 +0200
13035 @@ -0,0 +1,71 @@
13036 +#ifndef _VX_LIMIT_H
13037 +#define _VX_LIMIT_H
13038 +
13039 +#define VLIMIT_NSOCK   16
13040 +#define VLIMIT_OPENFD  17
13041 +#define VLIMIT_ANON    18
13042 +#define VLIMIT_SHMEM   19
13043 +#define VLIMIT_SEMARY  20
13044 +#define VLIMIT_NSEMS   21
13045 +#define VLIMIT_DENTRY  22
13046 +#define VLIMIT_MAPPED  23
13047 +
13048 +
13049 +#ifdef __KERNEL__
13050 +
13051 +#define        VLIM_NOCHECK    ((1L << VLIMIT_DENTRY) | (1L << RLIMIT_RSS))
13052 +
13053 +/*     keep in sync with CRLIM_INFINITY */
13054 +
13055 +#define        VLIM_INFINITY   (~0ULL)
13056 +
13057 +#include <asm/atomic.h>
13058 +#include <asm/resource.h>
13059 +
13060 +#ifndef RLIM_INFINITY
13061 +#warning RLIM_INFINITY is undefined
13062 +#endif
13063 +
13064 +#define __rlim_val(l, r, v)    ((l)->res[r].v)
13065 +
13066 +#define __rlim_soft(l, r)      __rlim_val(l, r, soft)
13067 +#define __rlim_hard(l, r)      __rlim_val(l, r, hard)
13068 +
13069 +#define __rlim_rcur(l, r)      __rlim_val(l, r, rcur)
13070 +#define __rlim_rmin(l, r)      __rlim_val(l, r, rmin)
13071 +#define __rlim_rmax(l, r)      __rlim_val(l, r, rmax)
13072 +
13073 +#define __rlim_lhit(l, r)      __rlim_val(l, r, lhit)
13074 +#define __rlim_hit(l, r)       atomic_inc(&__rlim_lhit(l, r))
13075 +
13076 +typedef atomic_long_t rlim_atomic_t;
13077 +typedef unsigned long rlim_t;
13078 +
13079 +#define __rlim_get(l, r)       atomic_long_read(&__rlim_rcur(l, r))
13080 +#define __rlim_set(l, r, v)    atomic_long_set(&__rlim_rcur(l, r), v)
13081 +#define __rlim_inc(l, r)       atomic_long_inc(&__rlim_rcur(l, r))
13082 +#define __rlim_dec(l, r)       atomic_long_dec(&__rlim_rcur(l, r))
13083 +#define __rlim_add(l, r, v)    atomic_long_add(v, &__rlim_rcur(l, r))
13084 +#define __rlim_sub(l, r, v)    atomic_long_sub(v, &__rlim_rcur(l, r))
13085 +
13086 +
13087 +#if    (RLIM_INFINITY == VLIM_INFINITY)
13088 +#define        VX_VLIM(r) ((long long)(long)(r))
13089 +#define        VX_RLIM(v) ((rlim_t)(v))
13090 +#else
13091 +#define        VX_VLIM(r) (((r) == RLIM_INFINITY) \
13092 +               ? VLIM_INFINITY : (long long)(r))
13093 +#define        VX_RLIM(v) (((v) == VLIM_INFINITY) \
13094 +               ? RLIM_INFINITY : (rlim_t)(v))
13095 +#endif
13096 +
13097 +struct sysinfo;
13098 +
13099 +void vx_vsi_meminfo(struct sysinfo *);
13100 +void vx_vsi_swapinfo(struct sysinfo *);
13101 +long vx_vsi_cached(struct sysinfo *);
13102 +
13103 +#define NUM_LIMITS     24
13104 +
13105 +#endif /* __KERNEL__ */
13106 +#endif /* _VX_LIMIT_H */
13107 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/limit_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_cmd.h
13108 --- linux-2.6.36/include/linux/vserver/limit_cmd.h      1970-01-01 01:00:00.000000000 +0100
13109 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_cmd.h        2010-10-21 13:09:36.000000000 +0200
13110 @@ -0,0 +1,71 @@
13111 +#ifndef _VX_LIMIT_CMD_H
13112 +#define _VX_LIMIT_CMD_H
13113 +
13114 +
13115 +/*  rlimit vserver commands */
13116 +
13117 +#define VCMD_get_rlimit                VC_CMD(RLIMIT, 1, 0)
13118 +#define VCMD_set_rlimit                VC_CMD(RLIMIT, 2, 0)
13119 +#define VCMD_get_rlimit_mask   VC_CMD(RLIMIT, 3, 0)
13120 +#define VCMD_reset_hits                VC_CMD(RLIMIT, 7, 0)
13121 +#define VCMD_reset_minmax      VC_CMD(RLIMIT, 9, 0)
13122 +
13123 +struct vcmd_ctx_rlimit_v0 {
13124 +       uint32_t id;
13125 +       uint64_t minimum;
13126 +       uint64_t softlimit;
13127 +       uint64_t maximum;
13128 +};
13129 +
13130 +struct vcmd_ctx_rlimit_mask_v0 {
13131 +       uint32_t minimum;
13132 +       uint32_t softlimit;
13133 +       uint32_t maximum;
13134 +};
13135 +
13136 +#define VCMD_rlimit_stat       VC_CMD(VSTAT, 1, 0)
13137 +
13138 +struct vcmd_rlimit_stat_v0 {
13139 +       uint32_t id;
13140 +       uint32_t hits;
13141 +       uint64_t value;
13142 +       uint64_t minimum;
13143 +       uint64_t maximum;
13144 +};
13145 +
13146 +#define CRLIM_UNSET            (0ULL)
13147 +#define CRLIM_INFINITY         (~0ULL)
13148 +#define CRLIM_KEEP             (~1ULL)
13149 +
13150 +#ifdef __KERNEL__
13151 +
13152 +#ifdef CONFIG_IA32_EMULATION
13153 +
13154 +struct vcmd_ctx_rlimit_v0_x32 {
13155 +       uint32_t id;
13156 +       uint64_t minimum;
13157 +       uint64_t softlimit;
13158 +       uint64_t maximum;
13159 +} __attribute__ ((packed));
13160 +
13161 +#endif /* CONFIG_IA32_EMULATION */
13162 +
13163 +#include <linux/compiler.h>
13164 +
13165 +extern int vc_get_rlimit_mask(uint32_t, void __user *);
13166 +extern int vc_get_rlimit(struct vx_info *, void __user *);
13167 +extern int vc_set_rlimit(struct vx_info *, void __user *);
13168 +extern int vc_reset_hits(struct vx_info *, void __user *);
13169 +extern int vc_reset_minmax(struct vx_info *, void __user *);
13170 +
13171 +extern int vc_rlimit_stat(struct vx_info *, void __user *);
13172 +
13173 +#ifdef CONFIG_IA32_EMULATION
13174 +
13175 +extern int vc_get_rlimit_x32(struct vx_info *, void __user *);
13176 +extern int vc_set_rlimit_x32(struct vx_info *, void __user *);
13177 +
13178 +#endif /* CONFIG_IA32_EMULATION */
13179 +
13180 +#endif /* __KERNEL__ */
13181 +#endif /* _VX_LIMIT_CMD_H */
13182 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/limit_def.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_def.h
13183 --- linux-2.6.36/include/linux/vserver/limit_def.h      1970-01-01 01:00:00.000000000 +0100
13184 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_def.h        2010-10-21 13:09:36.000000000 +0200
13185 @@ -0,0 +1,47 @@
13186 +#ifndef _VX_LIMIT_DEF_H
13187 +#define _VX_LIMIT_DEF_H
13188 +
13189 +#include <asm/atomic.h>
13190 +#include <asm/resource.h>
13191 +
13192 +#include "limit.h"
13193 +
13194 +
13195 +struct _vx_res_limit {
13196 +       rlim_t soft;            /* Context soft limit */
13197 +       rlim_t hard;            /* Context hard limit */
13198 +
13199 +       rlim_atomic_t rcur;     /* Current value */
13200 +       rlim_t rmin;            /* Context minimum */
13201 +       rlim_t rmax;            /* Context maximum */
13202 +
13203 +       atomic_t lhit;          /* Limit hits */
13204 +};
13205 +
13206 +/* context sub struct */
13207 +
13208 +struct _vx_limit {
13209 +       struct _vx_res_limit res[NUM_LIMITS];
13210 +};
13211 +
13212 +#ifdef CONFIG_VSERVER_DEBUG
13213 +
13214 +static inline void __dump_vx_limit(struct _vx_limit *limit)
13215 +{
13216 +       int i;
13217 +
13218 +       printk("\t_vx_limit:");
13219 +       for (i = 0; i < NUM_LIMITS; i++) {
13220 +               printk("\t [%2d] = %8lu %8lu/%8lu, %8ld/%8ld, %8d\n",
13221 +                       i, (unsigned long)__rlim_get(limit, i),
13222 +                       (unsigned long)__rlim_rmin(limit, i),
13223 +                       (unsigned long)__rlim_rmax(limit, i),
13224 +                       (long)__rlim_soft(limit, i),
13225 +                       (long)__rlim_hard(limit, i),
13226 +                       atomic_read(&__rlim_lhit(limit, i)));
13227 +       }
13228 +}
13229 +
13230 +#endif
13231 +
13232 +#endif /* _VX_LIMIT_DEF_H */
13233 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/limit_int.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_int.h
13234 --- linux-2.6.36/include/linux/vserver/limit_int.h      1970-01-01 01:00:00.000000000 +0100
13235 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/limit_int.h        2010-10-21 13:09:36.000000000 +0200
13236 @@ -0,0 +1,198 @@
13237 +#ifndef _VX_LIMIT_INT_H
13238 +#define _VX_LIMIT_INT_H
13239 +
13240 +#include "context.h"
13241 +
13242 +#ifdef __KERNEL__
13243 +
13244 +#define VXD_RCRES_COND(r)      VXD_CBIT(cres, r)
13245 +#define VXD_RLIMIT_COND(r)     VXD_CBIT(limit, r)
13246 +
13247 +extern const char *vlimit_name[NUM_LIMITS];
13248 +
13249 +static inline void __vx_acc_cres(struct vx_info *vxi,
13250 +       int res, int dir, void *_data, char *_file, int _line)
13251 +{
13252 +       if (VXD_RCRES_COND(res))
13253 +               vxlprintk(1, "vx_acc_cres[%5d,%s,%2d]: %5ld%s (%p)",
13254 +                       (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
13255 +                       (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
13256 +                       (dir > 0) ? "++" : "--", _data, _file, _line);
13257 +       if (!vxi)
13258 +               return;
13259 +
13260 +       if (dir > 0)
13261 +               __rlim_inc(&vxi->limit, res);
13262 +       else
13263 +               __rlim_dec(&vxi->limit, res);
13264 +}
13265 +
13266 +static inline void __vx_add_cres(struct vx_info *vxi,
13267 +       int res, int amount, void *_data, char *_file, int _line)
13268 +{
13269 +       if (VXD_RCRES_COND(res))
13270 +               vxlprintk(1, "vx_add_cres[%5d,%s,%2d]: %5ld += %5d (%p)",
13271 +                       (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
13272 +                       (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
13273 +                       amount, _data, _file, _line);
13274 +       if (amount == 0)
13275 +               return;
13276 +       if (!vxi)
13277 +               return;
13278 +       __rlim_add(&vxi->limit, res, amount);
13279 +}
13280 +
13281 +static inline
13282 +int __vx_cres_adjust_max(struct _vx_limit *limit, int res, rlim_t value)
13283 +{
13284 +       int cond = (value > __rlim_rmax(limit, res));
13285 +
13286 +       if (cond)
13287 +               __rlim_rmax(limit, res) = value;
13288 +       return cond;
13289 +}
13290 +
13291 +static inline
13292 +int __vx_cres_adjust_min(struct _vx_limit *limit, int res, rlim_t value)
13293 +{
13294 +       int cond = (value < __rlim_rmin(limit, res));
13295 +
13296 +       if (cond)
13297 +               __rlim_rmin(limit, res) = value;
13298 +       return cond;
13299 +}
13300 +
13301 +static inline
13302 +void __vx_cres_fixup(struct _vx_limit *limit, int res, rlim_t value)
13303 +{
13304 +       if (!__vx_cres_adjust_max(limit, res, value))
13305 +               __vx_cres_adjust_min(limit, res, value);
13306 +}
13307 +
13308 +
13309 +/*     return values:
13310 +        +1 ... no limit hit
13311 +        -1 ... over soft limit
13312 +         0 ... over hard limit         */
13313 +
13314 +static inline int __vx_cres_avail(struct vx_info *vxi,
13315 +       int res, int num, char *_file, int _line)
13316 +{
13317 +       struct _vx_limit *limit;
13318 +       rlim_t value;
13319 +
13320 +       if (VXD_RLIMIT_COND(res))
13321 +               vxlprintk(1, "vx_cres_avail[%5d,%s,%2d]: %5ld/%5ld > %5ld + %5d",
13322 +                       (vxi ? vxi->vx_id : -1), vlimit_name[res], res,
13323 +                       (vxi ? (long)__rlim_soft(&vxi->limit, res) : -1),
13324 +                       (vxi ? (long)__rlim_hard(&vxi->limit, res) : -1),
13325 +                       (vxi ? (long)__rlim_get(&vxi->limit, res) : 0),
13326 +                       num, _file, _line);
13327 +       if (!vxi)
13328 +               return 1;
13329 +
13330 +       limit = &vxi->limit;
13331 +       value = __rlim_get(limit, res);
13332 +
13333 +       if (!__vx_cres_adjust_max(limit, res, value))
13334 +               __vx_cres_adjust_min(limit, res, value);
13335 +
13336 +       if (num == 0)
13337 +               return 1;
13338 +
13339 +       if (__rlim_soft(limit, res) == RLIM_INFINITY)
13340 +               return -1;
13341 +       if (value + num <= __rlim_soft(limit, res))
13342 +               return -1;
13343 +
13344 +       if (__rlim_hard(limit, res) == RLIM_INFINITY)
13345 +               return 1;
13346 +       if (value + num <= __rlim_hard(limit, res))
13347 +               return 1;
13348 +
13349 +       __rlim_hit(limit, res);
13350 +       return 0;
13351 +}
13352 +
13353 +
13354 +static const int VLA_RSS[] = { RLIMIT_RSS, VLIMIT_ANON, VLIMIT_MAPPED, 0 };
13355 +
13356 +static inline
13357 +rlim_t __vx_cres_array_sum(struct _vx_limit *limit, const int *array)
13358 +{
13359 +       rlim_t value, sum = 0;
13360 +       int res;
13361 +
13362 +       while ((res = *array++)) {
13363 +               value = __rlim_get(limit, res);
13364 +               __vx_cres_fixup(limit, res, value);
13365 +               sum += value;
13366 +       }
13367 +       return sum;
13368 +}
13369 +
13370 +static inline
13371 +rlim_t __vx_cres_array_fixup(struct _vx_limit *limit, const int *array)
13372 +{
13373 +       rlim_t value = __vx_cres_array_sum(limit, array + 1);
13374 +       int res = *array;
13375 +
13376 +       if (value == __rlim_get(limit, res))
13377 +               return value;
13378 +
13379 +       __rlim_set(limit, res, value);
13380 +       /* now adjust min/max */
13381 +       if (!__vx_cres_adjust_max(limit, res, value))
13382 +               __vx_cres_adjust_min(limit, res, value);
13383 +
13384 +       return value;
13385 +}
13386 +
13387 +static inline int __vx_cres_array_avail(struct vx_info *vxi,
13388 +       const int *array, int num, char *_file, int _line)
13389 +{
13390 +       struct _vx_limit *limit;
13391 +       rlim_t value = 0;
13392 +       int res;
13393 +
13394 +       if (num == 0)
13395 +               return 1;
13396 +       if (!vxi)
13397 +               return 1;
13398 +
13399 +       limit = &vxi->limit;
13400 +       res = *array;
13401 +       value = __vx_cres_array_sum(limit, array + 1);
13402 +
13403 +       __rlim_set(limit, res, value);
13404 +       __vx_cres_fixup(limit, res, value);
13405 +
13406 +       return __vx_cres_avail(vxi, res, num, _file, _line);
13407 +}
13408 +
13409 +
13410 +static inline void vx_limit_fixup(struct _vx_limit *limit, int id)
13411 +{
13412 +       rlim_t value;
13413 +       int res;
13414 +
13415 +       /* complex resources first */
13416 +       if ((id < 0) || (id == RLIMIT_RSS))
13417 +               __vx_cres_array_fixup(limit, VLA_RSS);
13418 +
13419 +       for (res = 0; res < NUM_LIMITS; res++) {
13420 +               if ((id > 0) && (res != id))
13421 +                       continue;
13422 +
13423 +               value = __rlim_get(limit, res);
13424 +               __vx_cres_fixup(limit, res, value);
13425 +
13426 +               /* not supposed to happen, maybe warn? */
13427 +               if (__rlim_rmax(limit, res) > __rlim_hard(limit, res))
13428 +                       __rlim_rmax(limit, res) = __rlim_hard(limit, res);
13429 +       }
13430 +}
13431 +
13432 +
13433 +#endif /* __KERNEL__ */
13434 +#endif /* _VX_LIMIT_INT_H */
13435 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/monitor.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/monitor.h
13436 --- linux-2.6.36/include/linux/vserver/monitor.h        1970-01-01 01:00:00.000000000 +0100
13437 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/monitor.h  2010-10-21 13:09:36.000000000 +0200
13438 @@ -0,0 +1,96 @@
13439 +#ifndef _VX_MONITOR_H
13440 +#define _VX_MONITOR_H
13441 +
13442 +#include <linux/types.h>
13443 +
13444 +enum {
13445 +       VXM_UNUSED = 0,
13446 +
13447 +       VXM_SYNC = 0x10,
13448 +
13449 +       VXM_UPDATE = 0x20,
13450 +       VXM_UPDATE_1,
13451 +       VXM_UPDATE_2,
13452 +
13453 +       VXM_RQINFO_1 = 0x24,
13454 +       VXM_RQINFO_2,
13455 +
13456 +       VXM_ACTIVATE = 0x40,
13457 +       VXM_DEACTIVATE,
13458 +       VXM_IDLE,
13459 +
13460 +       VXM_HOLD = 0x44,
13461 +       VXM_UNHOLD,
13462 +
13463 +       VXM_MIGRATE = 0x48,
13464 +       VXM_RESCHED,
13465 +
13466 +       /* all other bits are flags */
13467 +       VXM_SCHED = 0x80,
13468 +};
13469 +
13470 +struct _vxm_update_1 {
13471 +       uint32_t tokens_max;
13472 +       uint32_t fill_rate;
13473 +       uint32_t interval;
13474 +};
13475 +
13476 +struct _vxm_update_2 {
13477 +       uint32_t tokens_min;
13478 +       uint32_t fill_rate;
13479 +       uint32_t interval;
13480 +};
13481 +
13482 +struct _vxm_rqinfo_1 {
13483 +       uint16_t running;
13484 +       uint16_t onhold;
13485 +       uint16_t iowait;
13486 +       uint16_t uintr;
13487 +       uint32_t idle_tokens;
13488 +};
13489 +
13490 +struct _vxm_rqinfo_2 {
13491 +       uint32_t norm_time;
13492 +       uint32_t idle_time;
13493 +       uint32_t idle_skip;
13494 +};
13495 +
13496 +struct _vxm_sched {
13497 +       uint32_t tokens;
13498 +       uint32_t norm_time;
13499 +       uint32_t idle_time;
13500 +};
13501 +
13502 +struct _vxm_task {
13503 +       uint16_t pid;
13504 +       uint16_t state;
13505 +};
13506 +
13507 +struct _vxm_event {
13508 +       uint32_t jif;
13509 +       union {
13510 +               uint32_t seq;
13511 +               uint32_t sec;
13512 +       };
13513 +       union {
13514 +               uint32_t tokens;
13515 +               uint32_t nsec;
13516 +               struct _vxm_task tsk;
13517 +       };
13518 +};
13519 +
13520 +struct _vx_mon_entry {
13521 +       uint16_t type;
13522 +       uint16_t xid;
13523 +       union {
13524 +               struct _vxm_event ev;
13525 +               struct _vxm_sched sd;
13526 +               struct _vxm_update_1 u1;
13527 +               struct _vxm_update_2 u2;
13528 +               struct _vxm_rqinfo_1 q1;
13529 +               struct _vxm_rqinfo_2 q2;
13530 +       };
13531 +};
13532 +
13533 +
13534 +#endif /* _VX_MONITOR_H */
13535 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/network.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/network.h
13536 --- linux-2.6.36/include/linux/vserver/network.h        1970-01-01 01:00:00.000000000 +0100
13537 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/network.h  2010-10-21 13:09:36.000000000 +0200
13538 @@ -0,0 +1,146 @@
13539 +#ifndef _VX_NETWORK_H
13540 +#define _VX_NETWORK_H
13541 +
13542 +#include <linux/types.h>
13543 +
13544 +
13545 +#define MAX_N_CONTEXT  65535   /* Arbitrary limit */
13546 +
13547 +
13548 +/* network flags */
13549 +
13550 +#define NXF_INFO_PRIVATE       0x00000008
13551 +
13552 +#define NXF_SINGLE_IP          0x00000100
13553 +#define NXF_LBACK_REMAP                0x00000200
13554 +#define NXF_LBACK_ALLOW                0x00000400
13555 +
13556 +#define NXF_HIDE_NETIF         0x02000000
13557 +#define NXF_HIDE_LBACK         0x04000000
13558 +
13559 +#define NXF_STATE_SETUP                (1ULL << 32)
13560 +#define NXF_STATE_ADMIN                (1ULL << 34)
13561 +
13562 +#define NXF_SC_HELPER          (1ULL << 36)
13563 +#define NXF_PERSISTENT         (1ULL << 38)
13564 +
13565 +#define NXF_ONE_TIME           (0x0005ULL << 32)
13566 +
13567 +
13568 +#define        NXF_INIT_SET            (__nxf_init_set())
13569 +
13570 +static inline uint64_t __nxf_init_set(void) {
13571 +       return    NXF_STATE_ADMIN
13572 +#ifdef CONFIG_VSERVER_AUTO_LBACK
13573 +               | NXF_LBACK_REMAP
13574 +               | NXF_HIDE_LBACK
13575 +#endif
13576 +#ifdef CONFIG_VSERVER_AUTO_SINGLE
13577 +               | NXF_SINGLE_IP
13578 +#endif
13579 +               | NXF_HIDE_NETIF;
13580 +}
13581 +
13582 +
13583 +/* network caps */
13584 +
13585 +#define NXC_TUN_CREATE         0x00000001
13586 +
13587 +#define NXC_RAW_ICMP           0x00000100
13588 +
13589 +
13590 +/* address types */
13591 +
13592 +#define NXA_TYPE_IPV4          0x0001
13593 +#define NXA_TYPE_IPV6          0x0002
13594 +
13595 +#define NXA_TYPE_NONE          0x0000
13596 +#define NXA_TYPE_ANY           0x00FF
13597 +
13598 +#define NXA_TYPE_ADDR          0x0010
13599 +#define NXA_TYPE_MASK          0x0020
13600 +#define NXA_TYPE_RANGE         0x0040
13601 +
13602 +#define NXA_MASK_ALL           (NXA_TYPE_ADDR | NXA_TYPE_MASK | NXA_TYPE_RANGE)
13603 +
13604 +#define NXA_MOD_BCAST          0x0100
13605 +#define NXA_MOD_LBACK          0x0200
13606 +
13607 +#define NXA_LOOPBACK           0x1000
13608 +
13609 +#define NXA_MASK_BIND          (NXA_MASK_ALL | NXA_MOD_BCAST | NXA_MOD_LBACK)
13610 +#define NXA_MASK_SHOW          (NXA_MASK_ALL | NXA_LOOPBACK)
13611 +
13612 +#ifdef __KERNEL__
13613 +
13614 +#include <linux/list.h>
13615 +#include <linux/spinlock.h>
13616 +#include <linux/rcupdate.h>
13617 +#include <linux/in.h>
13618 +#include <linux/in6.h>
13619 +#include <asm/atomic.h>
13620 +
13621 +struct nx_addr_v4 {
13622 +       struct nx_addr_v4 *next;
13623 +       struct in_addr ip[2];
13624 +       struct in_addr mask;
13625 +       uint16_t type;
13626 +       uint16_t flags;
13627 +};
13628 +
13629 +struct nx_addr_v6 {
13630 +       struct nx_addr_v6 *next;
13631 +       struct in6_addr ip;
13632 +       struct in6_addr mask;
13633 +       uint32_t prefix;
13634 +       uint16_t type;
13635 +       uint16_t flags;
13636 +};
13637 +
13638 +struct nx_info {
13639 +       struct hlist_node nx_hlist;     /* linked list of nxinfos */
13640 +       nid_t nx_id;                    /* vnet id */
13641 +       atomic_t nx_usecnt;             /* usage count */
13642 +       atomic_t nx_tasks;              /* tasks count */
13643 +       int nx_state;                   /* context state */
13644 +
13645 +       uint64_t nx_flags;              /* network flag word */
13646 +       uint64_t nx_ncaps;              /* network capabilities */
13647 +
13648 +       struct in_addr v4_lback;        /* Loopback address */
13649 +       struct in_addr v4_bcast;        /* Broadcast address */
13650 +       struct nx_addr_v4 v4;           /* First/Single ipv4 address */
13651 +#ifdef CONFIG_IPV6
13652 +       struct nx_addr_v6 v6;           /* First/Single ipv6 address */
13653 +#endif
13654 +       char nx_name[65];               /* network context name */
13655 +};
13656 +
13657 +
13658 +/* status flags */
13659 +
13660 +#define NXS_HASHED      0x0001
13661 +#define NXS_SHUTDOWN    0x0100
13662 +#define NXS_RELEASED    0x8000
13663 +
13664 +extern struct nx_info *lookup_nx_info(int);
13665 +
13666 +extern int get_nid_list(int, unsigned int *, int);
13667 +extern int nid_is_hashed(nid_t);
13668 +
13669 +extern int nx_migrate_task(struct task_struct *, struct nx_info *);
13670 +
13671 +extern long vs_net_change(struct nx_info *, unsigned int);
13672 +
13673 +struct sock;
13674 +
13675 +
13676 +#define NX_IPV4(n)     ((n)->v4.type != NXA_TYPE_NONE)
13677 +#ifdef  CONFIG_IPV6
13678 +#define NX_IPV6(n)     ((n)->v6.type != NXA_TYPE_NONE)
13679 +#else
13680 +#define NX_IPV6(n)     (0)
13681 +#endif
13682 +
13683 +#endif /* __KERNEL__ */
13684 +#endif /* _VX_NETWORK_H */
13685 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/network_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/network_cmd.h
13686 --- linux-2.6.36/include/linux/vserver/network_cmd.h    1970-01-01 01:00:00.000000000 +0100
13687 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/network_cmd.h      2010-10-21 13:09:36.000000000 +0200
13688 @@ -0,0 +1,150 @@
13689 +#ifndef _VX_NETWORK_CMD_H
13690 +#define _VX_NETWORK_CMD_H
13691 +
13692 +
13693 +/* vinfo commands */
13694 +
13695 +#define VCMD_task_nid          VC_CMD(VINFO, 2, 0)
13696 +
13697 +#ifdef __KERNEL__
13698 +extern int vc_task_nid(uint32_t);
13699 +
13700 +#endif /* __KERNEL__ */
13701 +
13702 +#define VCMD_nx_info           VC_CMD(VINFO, 6, 0)
13703 +
13704 +struct vcmd_nx_info_v0 {
13705 +       uint32_t nid;
13706 +       /* more to come */
13707 +};
13708 +
13709 +#ifdef __KERNEL__
13710 +extern int vc_nx_info(struct nx_info *, void __user *);
13711 +
13712 +#endif /* __KERNEL__ */
13713 +
13714 +#include <linux/in.h>
13715 +#include <linux/in6.h>
13716 +
13717 +#define VCMD_net_create_v0     VC_CMD(VNET, 1, 0)
13718 +#define VCMD_net_create                VC_CMD(VNET, 1, 1)
13719 +
13720 +struct  vcmd_net_create {
13721 +       uint64_t flagword;
13722 +};
13723 +
13724 +#define VCMD_net_migrate       VC_CMD(NETMIG, 1, 0)
13725 +
13726 +#define VCMD_net_add           VC_CMD(NETALT, 1, 0)
13727 +#define VCMD_net_remove                VC_CMD(NETALT, 2, 0)
13728 +
13729 +struct vcmd_net_addr_v0 {
13730 +       uint16_t type;
13731 +       uint16_t count;
13732 +       struct in_addr ip[4];
13733 +       struct in_addr mask[4];
13734 +};
13735 +
13736 +#define VCMD_net_add_ipv4      VC_CMD(NETALT, 1, 1)
13737 +#define VCMD_net_remove_ipv4   VC_CMD(NETALT, 2, 1)
13738 +
13739 +struct vcmd_net_addr_ipv4_v1 {
13740 +       uint16_t type;
13741 +       uint16_t flags;
13742 +       struct in_addr ip;
13743 +       struct in_addr mask;
13744 +};
13745 +
13746 +#define VCMD_net_add_ipv6      VC_CMD(NETALT, 3, 1)
13747 +#define VCMD_net_remove_ipv6   VC_CMD(NETALT, 4, 1)
13748 +
13749 +struct vcmd_net_addr_ipv6_v1 {
13750 +       uint16_t type;
13751 +       uint16_t flags;
13752 +       uint32_t prefix;
13753 +       struct in6_addr ip;
13754 +       struct in6_addr mask;
13755 +};
13756 +
13757 +#define VCMD_add_match_ipv4    VC_CMD(NETALT, 5, 0)
13758 +#define VCMD_get_match_ipv4    VC_CMD(NETALT, 6, 0)
13759 +
13760 +struct vcmd_match_ipv4_v0 {
13761 +       uint16_t type;
13762 +       uint16_t flags;
13763 +       uint16_t parent;
13764 +       uint16_t prefix;
13765 +       struct in_addr ip;
13766 +       struct in_addr ip2;
13767 +       struct in_addr mask;
13768 +};
13769 +
13770 +#define VCMD_add_match_ipv6    VC_CMD(NETALT, 7, 0)
13771 +#define VCMD_get_match_ipv6    VC_CMD(NETALT, 8, 0)
13772 +
13773 +struct vcmd_match_ipv6_v0 {
13774 +       uint16_t type;
13775 +       uint16_t flags;
13776 +       uint16_t parent;
13777 +       uint16_t prefix;
13778 +       struct in6_addr ip;
13779 +       struct in6_addr ip2;
13780 +       struct in6_addr mask;
13781 +};
13782 +
13783 +
13784 +#ifdef __KERNEL__
13785 +extern int vc_net_create(uint32_t, void __user *);
13786 +extern int vc_net_migrate(struct nx_info *, void __user *);
13787 +
13788 +extern int vc_net_add(struct nx_info *, void __user *);
13789 +extern int vc_net_remove(struct nx_info *, void __user *);
13790 +
13791 +extern int vc_net_add_ipv4(struct nx_info *, void __user *);
13792 +extern int vc_net_remove_ipv4(struct nx_info *, void __user *);
13793 +
13794 +extern int vc_net_add_ipv6(struct nx_info *, void __user *);
13795 +extern int vc_net_remove_ipv6(struct nx_info *, void __user *);
13796 +
13797 +extern int vc_add_match_ipv4(struct nx_info *, void __user *);
13798 +extern int vc_get_match_ipv4(struct nx_info *, void __user *);
13799 +
13800 +extern int vc_add_match_ipv6(struct nx_info *, void __user *);
13801 +extern int vc_get_match_ipv6(struct nx_info *, void __user *);
13802 +
13803 +#endif /* __KERNEL__ */
13804 +
13805 +
13806 +/* flag commands */
13807 +
13808 +#define VCMD_get_nflags                VC_CMD(FLAGS, 5, 0)
13809 +#define VCMD_set_nflags                VC_CMD(FLAGS, 6, 0)
13810 +
13811 +struct vcmd_net_flags_v0 {
13812 +       uint64_t flagword;
13813 +       uint64_t mask;
13814 +};
13815 +
13816 +#ifdef __KERNEL__
13817 +extern int vc_get_nflags(struct nx_info *, void __user *);
13818 +extern int vc_set_nflags(struct nx_info *, void __user *);
13819 +
13820 +#endif /* __KERNEL__ */
13821 +
13822 +
13823 +/* network caps commands */
13824 +
13825 +#define VCMD_get_ncaps         VC_CMD(FLAGS, 7, 0)
13826 +#define VCMD_set_ncaps         VC_CMD(FLAGS, 8, 0)
13827 +
13828 +struct vcmd_net_caps_v0 {
13829 +       uint64_t ncaps;
13830 +       uint64_t cmask;
13831 +};
13832 +
13833 +#ifdef __KERNEL__
13834 +extern int vc_get_ncaps(struct nx_info *, void __user *);
13835 +extern int vc_set_ncaps(struct nx_info *, void __user *);
13836 +
13837 +#endif /* __KERNEL__ */
13838 +#endif /* _VX_CONTEXT_CMD_H */
13839 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/percpu.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/percpu.h
13840 --- linux-2.6.36/include/linux/vserver/percpu.h 1970-01-01 01:00:00.000000000 +0100
13841 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/percpu.h   2010-10-21 13:09:36.000000000 +0200
13842 @@ -0,0 +1,14 @@
13843 +#ifndef _VX_PERCPU_H
13844 +#define _VX_PERCPU_H
13845 +
13846 +#include "cvirt_def.h"
13847 +#include "sched_def.h"
13848 +
13849 +struct _vx_percpu {
13850 +       struct _vx_cvirt_pc cvirt;
13851 +       struct _vx_sched_pc sched;
13852 +};
13853 +
13854 +#define        PERCPU_PERCTX   (sizeof(struct _vx_percpu))
13855 +
13856 +#endif /* _VX_PERCPU_H */
13857 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/pid.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/pid.h
13858 --- linux-2.6.36/include/linux/vserver/pid.h    1970-01-01 01:00:00.000000000 +0100
13859 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/pid.h      2010-10-21 13:09:36.000000000 +0200
13860 @@ -0,0 +1,51 @@
13861 +#ifndef _VSERVER_PID_H
13862 +#define _VSERVER_PID_H
13863 +
13864 +/* pid faking stuff */
13865 +
13866 +#define vx_info_map_pid(v, p) \
13867 +       __vx_info_map_pid((v), (p), __func__, __FILE__, __LINE__)
13868 +#define vx_info_map_tgid(v,p)  vx_info_map_pid(v,p)
13869 +#define vx_map_pid(p) vx_info_map_pid(current_vx_info(), p)
13870 +#define vx_map_tgid(p) vx_map_pid(p)
13871 +
13872 +static inline int __vx_info_map_pid(struct vx_info *vxi, int pid,
13873 +       const char *func, const char *file, int line)
13874 +{
13875 +       if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
13876 +               vxfprintk(VXD_CBIT(cvirt, 2),
13877 +                       "vx_map_tgid: %p/%llx: %d -> %d",
13878 +                       vxi, (long long)vxi->vx_flags, pid,
13879 +                       (pid && pid == vxi->vx_initpid) ? 1 : pid,
13880 +                       func, file, line);
13881 +               if (pid == 0)
13882 +                       return 0;
13883 +               if (pid == vxi->vx_initpid)
13884 +                       return 1;
13885 +       }
13886 +       return pid;
13887 +}
13888 +
13889 +#define vx_info_rmap_pid(v, p) \
13890 +       __vx_info_rmap_pid((v), (p), __func__, __FILE__, __LINE__)
13891 +#define vx_rmap_pid(p) vx_info_rmap_pid(current_vx_info(), p)
13892 +#define vx_rmap_tgid(p) vx_rmap_pid(p)
13893 +
13894 +static inline int __vx_info_rmap_pid(struct vx_info *vxi, int pid,
13895 +       const char *func, const char *file, int line)
13896 +{
13897 +       if (vx_info_flags(vxi, VXF_INFO_INIT, 0)) {
13898 +               vxfprintk(VXD_CBIT(cvirt, 2),
13899 +                       "vx_rmap_tgid: %p/%llx: %d -> %d",
13900 +                       vxi, (long long)vxi->vx_flags, pid,
13901 +                       (pid == 1) ? vxi->vx_initpid : pid,
13902 +                       func, file, line);
13903 +               if ((pid == 1) && vxi->vx_initpid)
13904 +                       return vxi->vx_initpid;
13905 +               if (pid == vxi->vx_initpid)
13906 +                       return ~0U;
13907 +       }
13908 +       return pid;
13909 +}
13910 +
13911 +#endif
13912 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/sched.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched.h
13913 --- linux-2.6.36/include/linux/vserver/sched.h  1970-01-01 01:00:00.000000000 +0100
13914 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched.h    2010-10-28 18:16:35.000000000 +0200
13915 @@ -0,0 +1,23 @@
13916 +#ifndef _VX_SCHED_H
13917 +#define _VX_SCHED_H
13918 +
13919 +
13920 +#ifdef __KERNEL__
13921 +
13922 +struct timespec;
13923 +
13924 +void vx_vsi_uptime(struct timespec *, struct timespec *);
13925 +
13926 +
13927 +struct vx_info;
13928 +
13929 +void vx_update_load(struct vx_info *);
13930 +
13931 +
13932 +void vx_update_sched_param(struct _vx_sched *sched,
13933 +       struct _vx_sched_pc *sched_pc);
13934 +
13935 +#endif /* __KERNEL__ */
13936 +#else  /* _VX_SCHED_H */
13937 +#warning duplicate inclusion
13938 +#endif /* _VX_SCHED_H */
13939 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/sched_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched_cmd.h
13940 --- linux-2.6.36/include/linux/vserver/sched_cmd.h      1970-01-01 01:00:00.000000000 +0100
13941 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched_cmd.h        2010-10-28 19:02:11.000000000 +0200
13942 @@ -0,0 +1,21 @@
13943 +#ifndef _VX_SCHED_CMD_H
13944 +#define _VX_SCHED_CMD_H
13945 +
13946 +
13947 +struct vcmd_prio_bias {
13948 +       int32_t cpu_id;
13949 +       int32_t prio_bias;
13950 +};
13951 +
13952 +#define VCMD_set_prio_bias     VC_CMD(SCHED, 4, 0)
13953 +#define VCMD_get_prio_bias     VC_CMD(SCHED, 5, 0)
13954 +
13955 +#ifdef __KERNEL__
13956 +
13957 +#include <linux/compiler.h>
13958 +
13959 +extern int vc_set_prio_bias(struct vx_info *, void __user *);
13960 +extern int vc_get_prio_bias(struct vx_info *, void __user *);
13961 +
13962 +#endif /* __KERNEL__ */
13963 +#endif /* _VX_SCHED_CMD_H */
13964 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/sched_def.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched_def.h
13965 --- linux-2.6.36/include/linux/vserver/sched_def.h      1970-01-01 01:00:00.000000000 +0100
13966 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/sched_def.h        2010-10-28 18:13:49.000000000 +0200
13967 @@ -0,0 +1,38 @@
13968 +#ifndef _VX_SCHED_DEF_H
13969 +#define _VX_SCHED_DEF_H
13970 +
13971 +#include <linux/spinlock.h>
13972 +#include <linux/jiffies.h>
13973 +#include <linux/cpumask.h>
13974 +#include <asm/atomic.h>
13975 +#include <asm/param.h>
13976 +
13977 +
13978 +/* context sub struct */
13979 +
13980 +struct _vx_sched {
13981 +       int prio_bias;                  /* bias offset for priority */
13982 +
13983 +       cpumask_t update;               /* CPUs which should update */
13984 +};
13985 +
13986 +struct _vx_sched_pc {
13987 +       int prio_bias;                  /* bias offset for priority */
13988 +
13989 +       uint64_t user_ticks;            /* token tick events */
13990 +       uint64_t sys_ticks;             /* token tick events */
13991 +       uint64_t hold_ticks;            /* token ticks paused */
13992 +};
13993 +
13994 +
13995 +#ifdef CONFIG_VSERVER_DEBUG
13996 +
13997 +static inline void __dump_vx_sched(struct _vx_sched *sched)
13998 +{
13999 +       printk("\t_vx_sched:\n");
14000 +       printk("\t priority = %4d\n", sched->prio_bias);
14001 +}
14002 +
14003 +#endif
14004 +
14005 +#endif /* _VX_SCHED_DEF_H */
14006 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/signal.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/signal.h
14007 --- linux-2.6.36/include/linux/vserver/signal.h 1970-01-01 01:00:00.000000000 +0100
14008 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/signal.h   2010-10-21 13:09:36.000000000 +0200
14009 @@ -0,0 +1,14 @@
14010 +#ifndef _VX_SIGNAL_H
14011 +#define _VX_SIGNAL_H
14012 +
14013 +
14014 +#ifdef __KERNEL__
14015 +
14016 +struct vx_info;
14017 +
14018 +int vx_info_kill(struct vx_info *, int, int);
14019 +
14020 +#endif /* __KERNEL__ */
14021 +#else  /* _VX_SIGNAL_H */
14022 +#warning duplicate inclusion
14023 +#endif /* _VX_SIGNAL_H */
14024 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/signal_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/signal_cmd.h
14025 --- linux-2.6.36/include/linux/vserver/signal_cmd.h     1970-01-01 01:00:00.000000000 +0100
14026 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/signal_cmd.h       2010-10-21 13:09:36.000000000 +0200
14027 @@ -0,0 +1,43 @@
14028 +#ifndef _VX_SIGNAL_CMD_H
14029 +#define _VX_SIGNAL_CMD_H
14030 +
14031 +
14032 +/*  signalling vserver commands */
14033 +
14034 +#define VCMD_ctx_kill          VC_CMD(PROCTRL, 1, 0)
14035 +#define VCMD_wait_exit         VC_CMD(EVENT, 99, 0)
14036 +
14037 +struct vcmd_ctx_kill_v0 {
14038 +       int32_t pid;
14039 +       int32_t sig;
14040 +};
14041 +
14042 +struct vcmd_wait_exit_v0 {
14043 +       int32_t reboot_cmd;
14044 +       int32_t exit_code;
14045 +};
14046 +
14047 +#ifdef __KERNEL__
14048 +
14049 +extern int vc_ctx_kill(struct vx_info *, void __user *);
14050 +extern int vc_wait_exit(struct vx_info *, void __user *);
14051 +
14052 +#endif /* __KERNEL__ */
14053 +
14054 +/*  process alteration commands */
14055 +
14056 +#define VCMD_get_pflags                VC_CMD(PROCALT, 5, 0)
14057 +#define VCMD_set_pflags                VC_CMD(PROCALT, 6, 0)
14058 +
14059 +struct vcmd_pflags_v0 {
14060 +       uint32_t flagword;
14061 +       uint32_t mask;
14062 +};
14063 +
14064 +#ifdef __KERNEL__
14065 +
14066 +extern int vc_get_pflags(uint32_t pid, void __user *);
14067 +extern int vc_set_pflags(uint32_t pid, void __user *);
14068 +
14069 +#endif /* __KERNEL__ */
14070 +#endif /* _VX_SIGNAL_CMD_H */
14071 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/space.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/space.h
14072 --- linux-2.6.36/include/linux/vserver/space.h  1970-01-01 01:00:00.000000000 +0100
14073 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/space.h    2010-10-21 13:09:36.000000000 +0200
14074 @@ -0,0 +1,12 @@
14075 +#ifndef _VX_SPACE_H
14076 +#define _VX_SPACE_H
14077 +
14078 +#include <linux/types.h>
14079 +
14080 +struct vx_info;
14081 +
14082 +int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index);
14083 +
14084 +#else  /* _VX_SPACE_H */
14085 +#warning duplicate inclusion
14086 +#endif /* _VX_SPACE_H */
14087 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/space_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/space_cmd.h
14088 --- linux-2.6.36/include/linux/vserver/space_cmd.h      1970-01-01 01:00:00.000000000 +0100
14089 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/space_cmd.h        2010-10-21 13:09:36.000000000 +0200
14090 @@ -0,0 +1,38 @@
14091 +#ifndef _VX_SPACE_CMD_H
14092 +#define _VX_SPACE_CMD_H
14093 +
14094 +
14095 +#define VCMD_enter_space_v0    VC_CMD(PROCALT, 1, 0)
14096 +#define VCMD_enter_space_v1    VC_CMD(PROCALT, 1, 1)
14097 +#define VCMD_enter_space       VC_CMD(PROCALT, 1, 2)
14098 +
14099 +#define VCMD_set_space_v0      VC_CMD(PROCALT, 3, 0)
14100 +#define VCMD_set_space_v1      VC_CMD(PROCALT, 3, 1)
14101 +#define VCMD_set_space         VC_CMD(PROCALT, 3, 2)
14102 +
14103 +#define VCMD_get_space_mask_v0 VC_CMD(PROCALT, 4, 0)
14104 +
14105 +#define VCMD_get_space_mask    VC_CMD(VSPACE, 0, 1)
14106 +#define VCMD_get_space_default VC_CMD(VSPACE, 1, 0)
14107 +
14108 +
14109 +struct vcmd_space_mask_v1 {
14110 +       uint64_t mask;
14111 +};
14112 +
14113 +struct vcmd_space_mask_v2 {
14114 +       uint64_t mask;
14115 +       uint32_t index;
14116 +};
14117 +
14118 +
14119 +#ifdef __KERNEL__
14120 +
14121 +extern int vc_enter_space_v1(struct vx_info *, void __user *);
14122 +extern int vc_set_space_v1(struct vx_info *, void __user *);
14123 +extern int vc_enter_space(struct vx_info *, void __user *);
14124 +extern int vc_set_space(struct vx_info *, void __user *);
14125 +extern int vc_get_space_mask(void __user *, int);
14126 +
14127 +#endif /* __KERNEL__ */
14128 +#endif /* _VX_SPACE_CMD_H */
14129 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/switch.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/switch.h
14130 --- linux-2.6.36/include/linux/vserver/switch.h 1970-01-01 01:00:00.000000000 +0100
14131 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/switch.h   2010-10-21 13:09:36.000000000 +0200
14132 @@ -0,0 +1,98 @@
14133 +#ifndef _VX_SWITCH_H
14134 +#define _VX_SWITCH_H
14135 +
14136 +#include <linux/types.h>
14137 +
14138 +
14139 +#define VC_CATEGORY(c)         (((c) >> 24) & 0x3F)
14140 +#define VC_COMMAND(c)          (((c) >> 16) & 0xFF)
14141 +#define VC_VERSION(c)          ((c) & 0xFFF)
14142 +
14143 +#define VC_CMD(c, i, v)                ((((VC_CAT_ ## c) & 0x3F) << 24) \
14144 +                               | (((i) & 0xFF) << 16) | ((v) & 0xFFF))
14145 +
14146 +/*
14147 +
14148 +  Syscall Matrix V2.8
14149 +
14150 +        |VERSION|CREATE |MODIFY |MIGRATE|CONTROL|EXPERIM| |SPECIAL|SPECIAL|
14151 +        |STATS  |DESTROY|ALTER  |CHANGE |LIMIT  |TEST   | |       |       |
14152 +        |INFO   |SETUP  |       |MOVE   |       |       | |       |       |
14153 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14154 +  SYSTEM |VERSION|VSETUP |VHOST  |       |       |       | |DEVICE |       |
14155 +  HOST   |     00|     01|     02|     03|     04|     05| |     06|     07|
14156 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14157 +  CPU    |       |VPROC  |PROCALT|PROCMIG|PROCTRL|       | |SCHED. |       |
14158 +  PROCESS|     08|     09|     10|     11|     12|     13| |     14|     15|
14159 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14160 +  MEMORY |       |       |       |       |MEMCTRL|       | |SWAP   |       |
14161 +        |     16|     17|     18|     19|     20|     21| |     22|     23|
14162 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14163 +  NETWORK|       |VNET   |NETALT |NETMIG |NETCTL |       | |SERIAL |       |
14164 +        |     24|     25|     26|     27|     28|     29| |     30|     31|
14165 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14166 +  DISK   |       |       |       |TAGMIG |DLIMIT |       | |INODE  |       |
14167 +  VFS    |     32|     33|     34|     35|     36|     37| |     38|     39|
14168 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14169 +  OTHER  |VSTAT  |       |       |       |       |       | |VINFO  |       |
14170 +        |     40|     41|     42|     43|     44|     45| |     46|     47|
14171 +  =======+=======+=======+=======+=======+=======+=======+ +=======+=======+
14172 +  SPECIAL|EVENT  |       |       |       |FLAGS  |       | |VSPACE |       |
14173 +        |     48|     49|     50|     51|     52|     53| |     54|     55|
14174 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14175 +  SPECIAL|DEBUG  |       |       |       |RLIMIT |SYSCALL| |       |COMPAT |
14176 +        |     56|     57|     58|     59|     60|TEST 61| |     62|     63|
14177 +  -------+-------+-------+-------+-------+-------+-------+ +-------+-------+
14178 +
14179 +*/
14180 +
14181 +#define VC_CAT_VERSION         0
14182 +
14183 +#define VC_CAT_VSETUP          1
14184 +#define VC_CAT_VHOST           2
14185 +
14186 +#define VC_CAT_DEVICE          6
14187 +
14188 +#define VC_CAT_VPROC           9
14189 +#define VC_CAT_PROCALT         10
14190 +#define VC_CAT_PROCMIG         11
14191 +#define VC_CAT_PROCTRL         12
14192 +
14193 +#define VC_CAT_SCHED           14
14194 +#define VC_CAT_MEMCTRL         20
14195 +
14196 +#define VC_CAT_VNET            25
14197 +#define VC_CAT_NETALT          26
14198 +#define VC_CAT_NETMIG          27
14199 +#define VC_CAT_NETCTRL         28
14200 +
14201 +#define VC_CAT_TAGMIG          35
14202 +#define VC_CAT_DLIMIT          36
14203 +#define VC_CAT_INODE           38
14204 +
14205 +#define VC_CAT_VSTAT           40
14206 +#define VC_CAT_VINFO           46
14207 +#define VC_CAT_EVENT           48
14208 +
14209 +#define VC_CAT_FLAGS           52
14210 +#define VC_CAT_VSPACE          54
14211 +#define VC_CAT_DEBUG           56
14212 +#define VC_CAT_RLIMIT          60
14213 +
14214 +#define VC_CAT_SYSTEST         61
14215 +#define VC_CAT_COMPAT          63
14216 +
14217 +/*  query version */
14218 +
14219 +#define VCMD_get_version       VC_CMD(VERSION, 0, 0)
14220 +#define VCMD_get_vci           VC_CMD(VERSION, 1, 0)
14221 +
14222 +
14223 +#ifdef __KERNEL__
14224 +
14225 +#include <linux/errno.h>
14226 +
14227 +#endif /* __KERNEL__ */
14228 +
14229 +#endif /* _VX_SWITCH_H */
14230 +
14231 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/tag.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/tag.h
14232 --- linux-2.6.36/include/linux/vserver/tag.h    1970-01-01 01:00:00.000000000 +0100
14233 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/tag.h      2010-10-21 13:09:36.000000000 +0200
14234 @@ -0,0 +1,143 @@
14235 +#ifndef _DX_TAG_H
14236 +#define _DX_TAG_H
14237 +
14238 +#include <linux/types.h>
14239 +
14240 +
14241 +#define DX_TAG(in)     (IS_TAGGED(in))
14242 +
14243 +
14244 +#ifdef CONFIG_TAG_NFSD
14245 +#define DX_TAG_NFSD    1
14246 +#else
14247 +#define DX_TAG_NFSD    0
14248 +#endif
14249 +
14250 +
14251 +#ifdef CONFIG_TAGGING_NONE
14252 +
14253 +#define MAX_UID                0xFFFFFFFF
14254 +#define MAX_GID                0xFFFFFFFF
14255 +
14256 +#define INOTAG_TAG(cond, uid, gid, tag)        (0)
14257 +
14258 +#define TAGINO_UID(cond, uid, tag)     (uid)
14259 +#define TAGINO_GID(cond, gid, tag)     (gid)
14260 +
14261 +#endif
14262 +
14263 +
14264 +#ifdef CONFIG_TAGGING_GID16
14265 +
14266 +#define MAX_UID                0xFFFFFFFF
14267 +#define MAX_GID                0x0000FFFF
14268 +
14269 +#define INOTAG_TAG(cond, uid, gid, tag)        \
14270 +       ((cond) ? (((gid) >> 16) & 0xFFFF) : 0)
14271 +
14272 +#define TAGINO_UID(cond, uid, tag)     (uid)
14273 +#define TAGINO_GID(cond, gid, tag)     \
14274 +       ((cond) ? (((gid) & 0xFFFF) | ((tag) << 16)) : (gid))
14275 +
14276 +#endif
14277 +
14278 +
14279 +#ifdef CONFIG_TAGGING_ID24
14280 +
14281 +#define MAX_UID                0x00FFFFFF
14282 +#define MAX_GID                0x00FFFFFF
14283 +
14284 +#define INOTAG_TAG(cond, uid, gid, tag)        \
14285 +       ((cond) ? ((((uid) >> 16) & 0xFF00) | (((gid) >> 24) & 0xFF)) : 0)
14286 +
14287 +#define TAGINO_UID(cond, uid, tag)     \
14288 +       ((cond) ? (((uid) & 0xFFFFFF) | (((tag) & 0xFF00) << 16)) : (uid))
14289 +#define TAGINO_GID(cond, gid, tag)     \
14290 +       ((cond) ? (((gid) & 0xFFFFFF) | (((tag) & 0x00FF) << 24)) : (gid))
14291 +
14292 +#endif
14293 +
14294 +
14295 +#ifdef CONFIG_TAGGING_UID16
14296 +
14297 +#define MAX_UID                0x0000FFFF
14298 +#define MAX_GID                0xFFFFFFFF
14299 +
14300 +#define INOTAG_TAG(cond, uid, gid, tag)        \
14301 +       ((cond) ? (((uid) >> 16) & 0xFFFF) : 0)
14302 +
14303 +#define TAGINO_UID(cond, uid, tag)     \
14304 +       ((cond) ? (((uid) & 0xFFFF) | ((tag) << 16)) : (uid))
14305 +#define TAGINO_GID(cond, gid, tag)     (gid)
14306 +
14307 +#endif
14308 +
14309 +
14310 +#ifdef CONFIG_TAGGING_INTERN
14311 +
14312 +#define MAX_UID                0xFFFFFFFF
14313 +#define MAX_GID                0xFFFFFFFF
14314 +
14315 +#define INOTAG_TAG(cond, uid, gid, tag)        \
14316 +       ((cond) ? (tag) : 0)
14317 +
14318 +#define TAGINO_UID(cond, uid, tag)     (uid)
14319 +#define TAGINO_GID(cond, gid, tag)     (gid)
14320 +
14321 +#endif
14322 +
14323 +
14324 +#ifndef CONFIG_TAGGING_NONE
14325 +#define dx_current_fstag(sb)   \
14326 +       ((sb)->s_flags & MS_TAGGED ? dx_current_tag() : 0)
14327 +#else
14328 +#define dx_current_fstag(sb)   (0)
14329 +#endif
14330 +
14331 +#ifndef CONFIG_TAGGING_INTERN
14332 +#define TAGINO_TAG(cond, tag)  (0)
14333 +#else
14334 +#define TAGINO_TAG(cond, tag)  ((cond) ? (tag) : 0)
14335 +#endif
14336 +
14337 +#define INOTAG_UID(cond, uid, gid)     \
14338 +       ((cond) ? ((uid) & MAX_UID) : (uid))
14339 +#define INOTAG_GID(cond, uid, gid)     \
14340 +       ((cond) ? ((gid) & MAX_GID) : (gid))
14341 +
14342 +
14343 +static inline uid_t dx_map_uid(uid_t uid)
14344 +{
14345 +       if ((uid > MAX_UID) && (uid != -1))
14346 +               uid = -2;
14347 +       return (uid & MAX_UID);
14348 +}
14349 +
14350 +static inline gid_t dx_map_gid(gid_t gid)
14351 +{
14352 +       if ((gid > MAX_GID) && (gid != -1))
14353 +               gid = -2;
14354 +       return (gid & MAX_GID);
14355 +}
14356 +
14357 +struct peer_tag {
14358 +       int32_t xid;
14359 +       int32_t nid;
14360 +};
14361 +
14362 +#define dx_notagcheck(sb) ((sb) && ((sb)->s_flags & MS_NOTAGCHECK))
14363 +
14364 +int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
14365 +                unsigned long *flags);
14366 +
14367 +#ifdef CONFIG_PROPAGATE
14368 +
14369 +void __dx_propagate_tag(struct nameidata *nd, struct inode *inode);
14370 +
14371 +#define dx_propagate_tag(n, i) __dx_propagate_tag(n, i)
14372 +
14373 +#else
14374 +#define dx_propagate_tag(n, i) do { } while (0)
14375 +#endif
14376 +
14377 +#endif /* _DX_TAG_H */
14378 diff -NurpP --minimal linux-2.6.36/include/linux/vserver/tag_cmd.h linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/tag_cmd.h
14379 --- linux-2.6.36/include/linux/vserver/tag_cmd.h        1970-01-01 01:00:00.000000000 +0100
14380 +++ linux-2.6.36-vs2.3.0.36.38/include/linux/vserver/tag_cmd.h  2010-10-21 13:09:36.000000000 +0200
14381 @@ -0,0 +1,22 @@
14382 +#ifndef _VX_TAG_CMD_H
14383 +#define _VX_TAG_CMD_H
14384 +
14385 +
14386 +/* vinfo commands */
14387 +
14388 +#define VCMD_task_tag          VC_CMD(VINFO, 3, 0)
14389 +
14390 +#ifdef __KERNEL__
14391 +extern int vc_task_tag(uint32_t);
14392 +
14393 +#endif /* __KERNEL__ */
14394 +
14395 +/* context commands */
14396 +
14397 +#define VCMD_tag_migrate       VC_CMD(TAGMIG, 1, 0)
14398 +
14399 +#ifdef __KERNEL__
14400 +extern int vc_tag_migrate(uint32_t);
14401 +
14402 +#endif /* __KERNEL__ */
14403 +#endif /* _VX_TAG_CMD_H */
14404 diff -NurpP --minimal linux-2.6.36/include/net/addrconf.h linux-2.6.36-vs2.3.0.36.38/include/net/addrconf.h
14405 --- linux-2.6.36/include/net/addrconf.h 2010-10-21 13:07:56.000000000 +0200
14406 +++ linux-2.6.36-vs2.3.0.36.38/include/net/addrconf.h   2010-10-21 13:09:36.000000000 +0200
14407 @@ -84,7 +84,8 @@ extern int                    ipv6_dev_get_saddr(struct n
14408                                                struct net_device *dev,
14409                                                const struct in6_addr *daddr,
14410                                                unsigned int srcprefs,
14411 -                                              struct in6_addr *saddr);
14412 +                                              struct in6_addr *saddr,
14413 +                                              struct nx_info *nxi);
14414  extern int                     ipv6_get_lladdr(struct net_device *dev,
14415                                                 struct in6_addr *addr,
14416                                                 unsigned char banned_flags);
14417 diff -NurpP --minimal linux-2.6.36/include/net/af_unix.h linux-2.6.36-vs2.3.0.36.38/include/net/af_unix.h
14418 --- linux-2.6.36/include/net/af_unix.h  2010-10-21 13:07:56.000000000 +0200
14419 +++ linux-2.6.36-vs2.3.0.36.38/include/net/af_unix.h    2010-10-21 13:09:36.000000000 +0200
14420 @@ -4,6 +4,7 @@
14421  #include <linux/socket.h>
14422  #include <linux/un.h>
14423  #include <linux/mutex.h>
14424 +#include <linux/vs_base.h>
14425  #include <net/sock.h>
14426
14427  extern void unix_inflight(struct file *fp);
14428 diff -NurpP --minimal linux-2.6.36/include/net/inet_timewait_sock.h linux-2.6.36-vs2.3.0.36.38/include/net/inet_timewait_sock.h
14429 --- linux-2.6.36/include/net/inet_timewait_sock.h       2010-08-02 16:52:56.000000000 +0200
14430 +++ linux-2.6.36-vs2.3.0.36.38/include/net/inet_timewait_sock.h 2010-10-21 13:09:36.000000000 +0200
14431 @@ -117,6 +117,10 @@ struct inet_timewait_sock {
14432  #define tw_hash                        __tw_common.skc_hash
14433  #define tw_prot                        __tw_common.skc_prot
14434  #define tw_net                 __tw_common.skc_net
14435 +#define tw_xid                 __tw_common.skc_xid
14436 +#define tw_vx_info             __tw_common.skc_vx_info
14437 +#define tw_nid                 __tw_common.skc_nid
14438 +#define tw_nx_info             __tw_common.skc_nx_info
14439         int                     tw_timeout;
14440         volatile unsigned char  tw_substate;
14441         /* 3 bits hole, try to pack */
14442 diff -NurpP --minimal linux-2.6.36/include/net/route.h linux-2.6.36-vs2.3.0.36.38/include/net/route.h
14443 --- linux-2.6.36/include/net/route.h    2010-10-21 13:07:56.000000000 +0200
14444 +++ linux-2.6.36-vs2.3.0.36.38/include/net/route.h      2010-10-21 13:09:36.000000000 +0200
14445 @@ -145,6 +145,9 @@ static inline void ip_rt_put(struct rtab
14446                 dst_release(&rt->dst);
14447  }
14448
14449 +#include <linux/vs_base.h>
14450 +#include <linux/vs_inet.h>
14451 +
14452  #define IPTOS_RT_MASK  (IPTOS_TOS_MASK & ~3)
14453
14454  extern const __u8 ip_tos2prio[16];
14455 @@ -154,6 +157,9 @@ static inline char rt_tos2priority(u8 to
14456         return ip_tos2prio[IPTOS_TOS(tos)>>1];
14457  }
14458
14459 +extern int ip_v4_find_src(struct net *net, struct nx_info *,
14460 +       struct rtable **, struct flowi *);
14461 +
14462  static inline int ip_route_connect(struct rtable **rp, __be32 dst,
14463                                    __be32 src, u32 tos, int oif, u8 protocol,
14464                                    __be16 sport, __be16 dport, struct sock *sk,
14465 @@ -171,11 +177,24 @@ static inline int ip_route_connect(struc
14466
14467         int err;
14468         struct net *net = sock_net(sk);
14469 +       struct nx_info *nx_info = current_nx_info();
14470
14471         if (inet_sk(sk)->transparent)
14472                 fl.flags |= FLOWI_FLAG_ANYSRC;
14473
14474 -       if (!dst || !src) {
14475 +       if (sk)
14476 +               nx_info = sk->sk_nx_info;
14477 +
14478 +       vxdprintk(VXD_CBIT(net, 4),
14479 +               "ip_route_connect(%p) %p,%p;%lx",
14480 +               sk, nx_info, sk->sk_socket,
14481 +               (sk->sk_socket?sk->sk_socket->flags:0));
14482 +
14483 +       err = ip_v4_find_src(net, nx_info, rp, &fl);
14484 +       if (err)
14485 +               return err;
14486 +
14487 +       if (!fl.fl4_dst || !fl.fl4_src) {
14488                 err = __ip_route_output_key(net, rp, &fl);
14489                 if (err)
14490                         return err;
14491 diff -NurpP --minimal linux-2.6.36/include/net/sock.h linux-2.6.36-vs2.3.0.36.38/include/net/sock.h
14492 --- linux-2.6.36/include/net/sock.h     2010-10-21 13:07:56.000000000 +0200
14493 +++ linux-2.6.36-vs2.3.0.36.38/include/net/sock.h       2010-10-21 13:09:36.000000000 +0200
14494 @@ -150,6 +150,10 @@ struct sock_common {
14495  #ifdef CONFIG_NET_NS
14496         struct net              *skc_net;
14497  #endif
14498 +       xid_t                   skc_xid;
14499 +       struct vx_info          *skc_vx_info;
14500 +       nid_t                   skc_nid;
14501 +       struct nx_info          *skc_nx_info;
14502  };
14503
14504  /**
14505 @@ -241,6 +245,10 @@ struct sock {
14506  #define sk_bind_node           __sk_common.skc_bind_node
14507  #define sk_prot                        __sk_common.skc_prot
14508  #define sk_net                 __sk_common.skc_net
14509 +#define sk_xid                 __sk_common.skc_xid
14510 +#define sk_vx_info             __sk_common.skc_vx_info
14511 +#define sk_nid                 __sk_common.skc_nid
14512 +#define sk_nx_info             __sk_common.skc_nx_info
14513         kmemcheck_bitfield_begin(flags);
14514         unsigned int            sk_shutdown  : 2,
14515                                 sk_no_check  : 2,
14516 diff -NurpP --minimal linux-2.6.36/init/Kconfig linux-2.6.36-vs2.3.0.36.38/init/Kconfig
14517 --- linux-2.6.36/init/Kconfig   2010-10-21 13:07:56.000000000 +0200
14518 +++ linux-2.6.36-vs2.3.0.36.38/init/Kconfig     2010-10-21 13:09:36.000000000 +0200
14519 @@ -476,6 +476,7 @@ config HAVE_UNSTABLE_SCHED_CLOCK
14520  menuconfig CGROUPS
14521         boolean "Control Group support"
14522         depends on EVENTFD
14523 +       default y
14524         help
14525           This option adds support for grouping sets of processes together, for
14526           use with process control subsystems such as Cpusets, CFS, memory
14527 @@ -503,6 +504,7 @@ config CGROUP_DEBUG
14528  config CGROUP_NS
14529         bool "Namespace cgroup subsystem"
14530         depends on CGROUPS
14531 +       default n
14532         help
14533           Provides a simple namespace cgroup subsystem to
14534           provide hierarchical naming of sets of namespaces,
14535 diff -NurpP --minimal linux-2.6.36/init/main.c linux-2.6.36-vs2.3.0.36.38/init/main.c
14536 --- linux-2.6.36/init/main.c    2010-10-21 13:07:56.000000000 +0200
14537 +++ linux-2.6.36-vs2.3.0.36.38/init/main.c      2010-10-21 13:10:12.000000000 +0200
14538 @@ -68,6 +68,7 @@
14539  #include <linux/sfi.h>
14540  #include <linux/shmem_fs.h>
14541  #include <linux/slab.h>
14542 +#include <linux/vserver/percpu.h>
14543
14544  #include <asm/io.h>
14545  #include <asm/bugs.h>
14546 diff -NurpP --minimal linux-2.6.36/ipc/mqueue.c linux-2.6.36-vs2.3.0.36.38/ipc/mqueue.c
14547 --- linux-2.6.36/ipc/mqueue.c   2010-10-21 13:07:56.000000000 +0200
14548 +++ linux-2.6.36-vs2.3.0.36.38/ipc/mqueue.c     2010-10-21 13:28:42.000000000 +0200
14549 @@ -33,6 +33,8 @@
14550  #include <linux/pid.h>
14551  #include <linux/ipc_namespace.h>
14552  #include <linux/slab.h>
14553 +#include <linux/vs_context.h>
14554 +#include <linux/vs_limit.h>
14555
14556  #include <net/sock.h>
14557  #include "util.h"
14558 @@ -66,6 +68,7 @@ struct mqueue_inode_info {
14559         struct sigevent notify;
14560         struct pid* notify_owner;
14561         struct user_struct *user;       /* user who created, for accounting */
14562 +       struct vx_info *vxi;
14563         struct sock *notify_sock;
14564         struct sk_buff *notify_cookie;
14565
14566 @@ -125,6 +128,7 @@ static struct inode *mqueue_get_inode(st
14567                 if (S_ISREG(mode)) {
14568                         struct mqueue_inode_info *info;
14569                         struct task_struct *p = current;
14570 +                       struct vx_info *vxi = p->vx_info;
14571                         unsigned long mq_bytes, mq_msg_tblsz;
14572
14573                         inode->i_fop = &mqueue_file_operations;
14574 @@ -138,6 +142,7 @@ static struct inode *mqueue_get_inode(st
14575                         info->notify_owner = NULL;
14576                         info->qsize = 0;
14577                         info->user = NULL;      /* set when all is ok */
14578 +                       info->vxi = NULL;
14579                         memset(&info->attr, 0, sizeof(info->attr));
14580                         info->attr.mq_maxmsg = ipc_ns->mq_msg_max;
14581                         info->attr.mq_msgsize = ipc_ns->mq_msgsize_max;
14582 @@ -156,16 +161,19 @@ static struct inode *mqueue_get_inode(st
14583                         spin_lock(&mq_lock);
14584                         if (u->mq_bytes + mq_bytes < u->mq_bytes ||
14585                             u->mq_bytes + mq_bytes >
14586 -                           task_rlimit(p, RLIMIT_MSGQUEUE)) {
14587 +                           task_rlimit(p, RLIMIT_MSGQUEUE) ||
14588 +                           !vx_ipcmsg_avail(vxi, mq_bytes)) {
14589                                 spin_unlock(&mq_lock);
14590                                 /* mqueue_evict_inode() releases info->messages */
14591                                 goto out_inode;
14592                         }
14593                         u->mq_bytes += mq_bytes;
14594 +                       vx_ipcmsg_add(vxi, u, mq_bytes);
14595                         spin_unlock(&mq_lock);
14596
14597                         /* all is ok */
14598                         info->user = get_uid(u);
14599 +                       info->vxi = get_vx_info(vxi);
14600                 } else if (S_ISDIR(mode)) {
14601                         inc_nlink(inode);
14602                         /* Some things misbehave if size == 0 on a directory */
14603 @@ -267,8 +275,11 @@ static void mqueue_evict_inode(struct in
14604             + info->attr.mq_msgsize);
14605         user = info->user;
14606         if (user) {
14607 +               struct vx_info *vxi = info->vxi;
14608 +
14609                 spin_lock(&mq_lock);
14610                 user->mq_bytes -= mq_bytes;
14611 +               vx_ipcmsg_sub(vxi, user, mq_bytes);
14612                 /*
14613                  * get_ns_from_inode() ensures that the
14614                  * (ipc_ns = sb->s_fs_info) is either a valid ipc_ns
14615 @@ -278,6 +289,7 @@ static void mqueue_evict_inode(struct in
14616                 if (ipc_ns)
14617                         ipc_ns->mq_queues_count--;
14618                 spin_unlock(&mq_lock);
14619 +               put_vx_info(vxi);
14620                 free_uid(user);
14621         }
14622         if (ipc_ns)
14623 diff -NurpP --minimal linux-2.6.36/ipc/msg.c linux-2.6.36-vs2.3.0.36.38/ipc/msg.c
14624 --- linux-2.6.36/ipc/msg.c      2010-08-02 16:52:57.000000000 +0200
14625 +++ linux-2.6.36-vs2.3.0.36.38/ipc/msg.c        2010-10-21 13:09:36.000000000 +0200
14626 @@ -37,6 +37,7 @@
14627  #include <linux/rwsem.h>
14628  #include <linux/nsproxy.h>
14629  #include <linux/ipc_namespace.h>
14630 +#include <linux/vs_base.h>
14631
14632  #include <asm/current.h>
14633  #include <asm/uaccess.h>
14634 @@ -190,6 +191,7 @@ static int newque(struct ipc_namespace *
14635
14636         msq->q_perm.mode = msgflg & S_IRWXUGO;
14637         msq->q_perm.key = key;
14638 +       msq->q_perm.xid = vx_current_xid();
14639
14640         msq->q_perm.security = NULL;
14641         retval = security_msg_queue_alloc(msq);
14642 diff -NurpP --minimal linux-2.6.36/ipc/namespace.c linux-2.6.36-vs2.3.0.36.38/ipc/namespace.c
14643 --- linux-2.6.36/ipc/namespace.c        2009-09-10 15:26:27.000000000 +0200
14644 +++ linux-2.6.36-vs2.3.0.36.38/ipc/namespace.c  2010-10-21 13:09:36.000000000 +0200
14645 @@ -11,6 +11,8 @@
14646  #include <linux/slab.h>
14647  #include <linux/fs.h>
14648  #include <linux/mount.h>
14649 +#include <linux/vs_base.h>
14650 +#include <linux/vserver/global.h>
14651
14652  #include "util.h"
14653
14654 diff -NurpP --minimal linux-2.6.36/ipc/sem.c linux-2.6.36-vs2.3.0.36.38/ipc/sem.c
14655 --- linux-2.6.36/ipc/sem.c      2010-10-21 13:07:56.000000000 +0200
14656 +++ linux-2.6.36-vs2.3.0.36.38/ipc/sem.c        2010-10-21 13:09:36.000000000 +0200
14657 @@ -86,6 +86,8 @@
14658  #include <linux/rwsem.h>
14659  #include <linux/nsproxy.h>
14660  #include <linux/ipc_namespace.h>
14661 +#include <linux/vs_base.h>
14662 +#include <linux/vs_limit.h>
14663
14664  #include <asm/uaccess.h>
14665  #include "util.h"
14666 @@ -260,6 +262,7 @@ static int newary(struct ipc_namespace *
14667
14668         sma->sem_perm.mode = (semflg & S_IRWXUGO);
14669         sma->sem_perm.key = key;
14670 +       sma->sem_perm.xid = vx_current_xid();
14671
14672         sma->sem_perm.security = NULL;
14673         retval = security_sem_alloc(sma);
14674 @@ -275,6 +278,9 @@ static int newary(struct ipc_namespace *
14675                 return id;
14676         }
14677         ns->used_sems += nsems;
14678 +       /* FIXME: obsoleted? */
14679 +       vx_semary_inc(sma);
14680 +       vx_nsems_add(sma, nsems);
14681
14682         sma->sem_base = (struct sem *) &sma[1];
14683
14684 @@ -730,6 +736,9 @@ static void freeary(struct ipc_namespace
14685
14686         wake_up_sem_queue_do(&tasks);
14687         ns->used_sems -= sma->sem_nsems;
14688 +       /* FIXME: obsoleted? */
14689 +       vx_nsems_sub(sma, sma->sem_nsems);
14690 +       vx_semary_dec(sma);
14691         security_sem_free(sma);
14692         ipc_rcu_putref(sma);
14693  }
14694 diff -NurpP --minimal linux-2.6.36/ipc/shm.c linux-2.6.36-vs2.3.0.36.38/ipc/shm.c
14695 --- linux-2.6.36/ipc/shm.c      2010-08-02 16:52:57.000000000 +0200
14696 +++ linux-2.6.36-vs2.3.0.36.38/ipc/shm.c        2010-10-21 13:09:36.000000000 +0200
14697 @@ -39,6 +39,8 @@
14698  #include <linux/nsproxy.h>
14699  #include <linux/mount.h>
14700  #include <linux/ipc_namespace.h>
14701 +#include <linux/vs_context.h>
14702 +#include <linux/vs_limit.h>
14703
14704  #include <asm/uaccess.h>
14705
14706 @@ -169,7 +171,12 @@ static void shm_open(struct vm_area_stru
14707   */
14708  static void shm_destroy(struct ipc_namespace *ns, struct shmid_kernel *shp)
14709  {
14710 -       ns->shm_tot -= (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
14711 +       struct vx_info *vxi = lookup_vx_info(shp->shm_perm.xid);
14712 +       int numpages = (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT;
14713 +
14714 +       vx_ipcshm_sub(vxi, shp, numpages);
14715 +       ns->shm_tot -= numpages;
14716 +
14717         shm_rmid(ns, shp);
14718         shm_unlock(shp);
14719         if (!is_file_hugepages(shp->shm_file))
14720 @@ -179,6 +186,7 @@ static void shm_destroy(struct ipc_names
14721                                                 shp->mlock_user);
14722         fput (shp->shm_file);
14723         security_shm_free(shp);
14724 +       put_vx_info(vxi);
14725         ipc_rcu_putref(shp);
14726  }
14727
14728 @@ -349,11 +357,15 @@ static int newseg(struct ipc_namespace *
14729         if (ns->shm_tot + numpages > ns->shm_ctlall)
14730                 return -ENOSPC;
14731
14732 +       if (!vx_ipcshm_avail(current_vx_info(), numpages))
14733 +               return -ENOSPC;
14734 +
14735         shp = ipc_rcu_alloc(sizeof(*shp));
14736         if (!shp)
14737                 return -ENOMEM;
14738
14739         shp->shm_perm.key = key;
14740 +       shp->shm_perm.xid = vx_current_xid();
14741         shp->shm_perm.mode = (shmflg & S_IRWXUGO);
14742         shp->mlock_user = NULL;
14743
14744 @@ -407,6 +419,7 @@ static int newseg(struct ipc_namespace *
14745         ns->shm_tot += numpages;
14746         error = shp->shm_perm.id;
14747         shm_unlock(shp);
14748 +       vx_ipcshm_add(current_vx_info(), key, numpages);
14749         return error;
14750
14751  no_id:
14752 diff -NurpP --minimal linux-2.6.36/kernel/Makefile linux-2.6.36-vs2.3.0.36.38/kernel/Makefile
14753 --- linux-2.6.36/kernel/Makefile        2010-10-21 13:07:56.000000000 +0200
14754 +++ linux-2.6.36-vs2.3.0.36.38/kernel/Makefile  2010-10-21 13:09:36.000000000 +0200
14755 @@ -25,6 +25,7 @@ CFLAGS_REMOVE_sched_clock.o = -pg
14756  CFLAGS_REMOVE_perf_event.o = -pg
14757  endif
14758
14759 +obj-y += vserver/
14760  obj-$(CONFIG_FREEZER) += freezer.o
14761  obj-$(CONFIG_PROFILING) += profile.o
14762  obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o
14763 diff -NurpP --minimal linux-2.6.36/kernel/capability.c linux-2.6.36-vs2.3.0.36.38/kernel/capability.c
14764 --- linux-2.6.36/kernel/capability.c    2010-08-02 16:52:57.000000000 +0200
14765 +++ linux-2.6.36-vs2.3.0.36.38/kernel/capability.c      2010-10-21 13:09:36.000000000 +0200
14766 @@ -14,6 +14,7 @@
14767  #include <linux/security.h>
14768  #include <linux/syscalls.h>
14769  #include <linux/pid_namespace.h>
14770 +#include <linux/vs_context.h>
14771  #include <asm/uaccess.h>
14772
14773  /*
14774 @@ -119,6 +120,7 @@ static int cap_validate_magic(cap_user_h
14775         return 0;
14776  }
14777
14778 +
14779  /*
14780   * The only thing that can change the capabilities of the current
14781   * process is the current process. As such, we can't be in this code
14782 @@ -289,6 +291,8 @@ error:
14783         return ret;
14784  }
14785
14786 +#include <linux/vserver/base.h>
14787 +
14788  /**
14789   * capable - Determine if the current task has a superior capability in effect
14790   * @cap: The capability to be tested for
14791 @@ -301,6 +305,9 @@ error:
14792   */
14793  int capable(int cap)
14794  {
14795 +       /* here for now so we don't require task locking */
14796 +       if (vs_check_bit(VXC_CAP_MASK, cap) && !vx_mcaps(1L << cap))
14797 +               return 0;
14798         if (unlikely(!cap_valid(cap))) {
14799                 printk(KERN_CRIT "capable() called with invalid cap=%u\n", cap);
14800                 BUG();
14801 diff -NurpP --minimal linux-2.6.36/kernel/compat.c linux-2.6.36-vs2.3.0.36.38/kernel/compat.c
14802 --- linux-2.6.36/kernel/compat.c        2010-10-21 13:07:56.000000000 +0200
14803 +++ linux-2.6.36-vs2.3.0.36.38/kernel/compat.c  2010-10-21 14:22:43.000000000 +0200
14804 @@ -914,7 +914,7 @@ asmlinkage long compat_sys_stime(compat_
14805         if (err)
14806                 return err;
14807
14808 -       do_settimeofday(&tv);
14809 +       vx_settimeofday(&tv);
14810         return 0;
14811  }
14812
14813 diff -NurpP --minimal linux-2.6.36/kernel/exit.c linux-2.6.36-vs2.3.0.36.38/kernel/exit.c
14814 --- linux-2.6.36/kernel/exit.c  2010-10-21 13:07:56.000000000 +0200
14815 +++ linux-2.6.36-vs2.3.0.36.38/kernel/exit.c    2010-10-21 13:09:36.000000000 +0200
14816 @@ -48,6 +48,10 @@
14817  #include <linux/fs_struct.h>
14818  #include <linux/init_task.h>
14819  #include <linux/perf_event.h>
14820 +#include <linux/vs_limit.h>
14821 +#include <linux/vs_context.h>
14822 +#include <linux/vs_network.h>
14823 +#include <linux/vs_pid.h>
14824  #include <trace/events/sched.h>
14825  #include <linux/hw_breakpoint.h>
14826
14827 @@ -487,9 +491,11 @@ static void close_files(struct files_str
14828                                         filp_close(file, files);
14829                                         cond_resched();
14830                                 }
14831 +                               vx_openfd_dec(i);
14832                         }
14833                         i++;
14834                         set >>= 1;
14835 +                       cond_resched();
14836                 }
14837         }
14838  }
14839 @@ -1023,11 +1029,16 @@ NORET_TYPE void do_exit(long code)
14840
14841         validate_creds_for_do_exit(tsk);
14842
14843 +       /* needs to stay after exit_notify() */
14844 +       exit_vx_info(tsk, code);
14845 +       exit_nx_info(tsk);
14846 +
14847         preempt_disable();
14848         exit_rcu();
14849         /* causes final put_task_struct in finish_task_switch(). */
14850         tsk->state = TASK_DEAD;
14851         schedule();
14852 +       printk("bad task: %p [%lx]\n", current, current->state);
14853         BUG();
14854         /* Avoid "noreturn function does return".  */
14855         for (;;)
14856 diff -NurpP --minimal linux-2.6.36/kernel/fork.c linux-2.6.36-vs2.3.0.36.38/kernel/fork.c
14857 --- linux-2.6.36/kernel/fork.c  2010-10-21 13:07:56.000000000 +0200
14858 +++ linux-2.6.36-vs2.3.0.36.38/kernel/fork.c    2010-10-21 13:09:36.000000000 +0200
14859 @@ -65,6 +65,10 @@
14860  #include <linux/perf_event.h>
14861  #include <linux/posix-timers.h>
14862  #include <linux/user-return-notifier.h>
14863 +#include <linux/vs_context.h>
14864 +#include <linux/vs_network.h>
14865 +#include <linux/vs_limit.h>
14866 +#include <linux/vs_memory.h>
14867
14868  #include <asm/pgtable.h>
14869  #include <asm/pgalloc.h>
14870 @@ -160,6 +164,8 @@ void free_task(struct task_struct *tsk)
14871         account_kernel_stack(tsk->stack, -1);
14872         free_thread_info(tsk->stack);
14873         rt_mutex_debug_task_free(tsk);
14874 +       clr_vx_info(&tsk->vx_info);
14875 +       clr_nx_info(&tsk->nx_info);
14876         ftrace_graph_exit_task(tsk);
14877         free_task_struct(tsk);
14878  }
14879 @@ -492,6 +498,7 @@ static struct mm_struct * mm_init(struct
14880         if (likely(!mm_alloc_pgd(mm))) {
14881                 mm->def_flags = 0;
14882                 mmu_notifier_mm_init(mm);
14883 +               set_vx_info(&mm->mm_vx_info, p->vx_info);
14884                 return mm;
14885         }
14886
14887 @@ -525,6 +532,7 @@ void __mmdrop(struct mm_struct *mm)
14888         mm_free_pgd(mm);
14889         destroy_context(mm);
14890         mmu_notifier_mm_destroy(mm);
14891 +       clr_vx_info(&mm->mm_vx_info);
14892         free_mm(mm);
14893  }
14894  EXPORT_SYMBOL_GPL(__mmdrop);
14895 @@ -660,6 +668,7 @@ struct mm_struct *dup_mm(struct task_str
14896                 goto fail_nomem;
14897
14898         memcpy(mm, oldmm, sizeof(*mm));
14899 +       mm->mm_vx_info = NULL;
14900
14901         /* Initializing for Swap token stuff */
14902         mm->token_priority = 0;
14903 @@ -698,6 +707,7 @@ fail_nocontext:
14904          * If init_new_context() failed, we cannot use mmput() to free the mm
14905          * because it calls destroy_context()
14906          */
14907 +       clr_vx_info(&mm->mm_vx_info);
14908         mm_free_pgd(mm);
14909         free_mm(mm);
14910         return NULL;
14911 @@ -973,6 +983,8 @@ static struct task_struct *copy_process(
14912         int retval;
14913         struct task_struct *p;
14914         int cgroup_callbacks_done = 0;
14915 +       struct vx_info *vxi;
14916 +       struct nx_info *nxi;
14917
14918         if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
14919                 return ERR_PTR(-EINVAL);
14920 @@ -1019,7 +1031,12 @@ static struct task_struct *copy_process(
14921         DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
14922         DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
14923  #endif
14924 +       init_vx_info(&p->vx_info, current_vx_info());
14925 +       init_nx_info(&p->nx_info, current_nx_info());
14926 +
14927         retval = -EAGAIN;
14928 +       if (!vx_nproc_avail(1))
14929 +               goto bad_fork_free;
14930         if (atomic_read(&p->real_cred->user->processes) >=
14931                         task_rlimit(p, RLIMIT_NPROC)) {
14932                 if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
14933 @@ -1284,6 +1301,18 @@ static struct task_struct *copy_process(
14934
14935         total_forks++;
14936         spin_unlock(&current->sighand->siglock);
14937 +
14938 +       /* p is copy of current */
14939 +       vxi = p->vx_info;
14940 +       if (vxi) {
14941 +               claim_vx_info(vxi, p);
14942 +               atomic_inc(&vxi->cvirt.nr_threads);
14943 +               atomic_inc(&vxi->cvirt.total_forks);
14944 +               vx_nproc_inc(p);
14945 +       }
14946 +       nxi = p->nx_info;
14947 +       if (nxi)
14948 +               claim_nx_info(nxi, p);
14949         write_unlock_irq(&tasklist_lock);
14950         proc_fork_connector(p);
14951         cgroup_post_fork(p);
14952 diff -NurpP --minimal linux-2.6.36/kernel/kthread.c linux-2.6.36-vs2.3.0.36.38/kernel/kthread.c
14953 --- linux-2.6.36/kernel/kthread.c       2010-10-21 13:07:57.000000000 +0200
14954 +++ linux-2.6.36-vs2.3.0.36.38/kernel/kthread.c 2010-10-21 13:42:40.000000000 +0200
14955 @@ -16,6 +16,7 @@
14956  #include <linux/mutex.h>
14957  #include <linux/slab.h>
14958  #include <linux/freezer.h>
14959 +#include <linux/vs_pid.h>
14960  #include <trace/events/sched.h>
14961
14962  static DEFINE_SPINLOCK(kthread_create_lock);
14963 diff -NurpP --minimal linux-2.6.36/kernel/nsproxy.c linux-2.6.36-vs2.3.0.36.38/kernel/nsproxy.c
14964 --- linux-2.6.36/kernel/nsproxy.c       2010-07-07 18:31:57.000000000 +0200
14965 +++ linux-2.6.36-vs2.3.0.36.38/kernel/nsproxy.c 2010-10-21 13:09:36.000000000 +0200
14966 @@ -20,6 +20,8 @@
14967  #include <linux/mnt_namespace.h>
14968  #include <linux/utsname.h>
14969  #include <linux/pid_namespace.h>
14970 +#include <linux/vserver/global.h>
14971 +#include <linux/vserver/debug.h>
14972  #include <net/net_namespace.h>
14973  #include <linux/ipc_namespace.h>
14974
14975 @@ -43,8 +45,11 @@ static inline struct nsproxy *create_nsp
14976         struct nsproxy *nsproxy;
14977
14978         nsproxy = kmem_cache_alloc(nsproxy_cachep, GFP_KERNEL);
14979 -       if (nsproxy)
14980 +       if (nsproxy) {
14981                 atomic_set(&nsproxy->count, 1);
14982 +               atomic_inc(&vs_global_nsproxy);
14983 +       }
14984 +       vxdprintk(VXD_CBIT(space, 2), "create_nsproxy = %p[1]", nsproxy);
14985         return nsproxy;
14986  }
14987
14988 @@ -53,41 +58,52 @@ static inline struct nsproxy *create_nsp
14989   * Return the newly created nsproxy.  Do not attach this to the task,
14990   * leave it to the caller to do proper locking and attach it to task.
14991   */
14992 -static struct nsproxy *create_new_namespaces(unsigned long flags,
14993 -                       struct task_struct *tsk, struct fs_struct *new_fs)
14994 +static struct nsproxy *unshare_namespaces(unsigned long flags,
14995 +                       struct nsproxy *orig, struct fs_struct *new_fs)
14996  {
14997         struct nsproxy *new_nsp;
14998         int err;
14999
15000 +       vxdprintk(VXD_CBIT(space, 4),
15001 +               "unshare_namespaces(0x%08lx,%p,%p)",
15002 +               flags, orig, new_fs);
15003 +
15004         new_nsp = create_nsproxy();
15005         if (!new_nsp)
15006                 return ERR_PTR(-ENOMEM);
15007
15008 -       new_nsp->mnt_ns = copy_mnt_ns(flags, tsk->nsproxy->mnt_ns, new_fs);
15009 +       new_nsp->mnt_ns = copy_mnt_ns(flags, orig->mnt_ns, new_fs);
15010         if (IS_ERR(new_nsp->mnt_ns)) {
15011                 err = PTR_ERR(new_nsp->mnt_ns);
15012                 goto out_ns;
15013         }
15014
15015 -       new_nsp->uts_ns = copy_utsname(flags, tsk->nsproxy->uts_ns);
15016 +       new_nsp->uts_ns = copy_utsname(flags, orig->uts_ns);
15017         if (IS_ERR(new_nsp->uts_ns)) {
15018                 err = PTR_ERR(new_nsp->uts_ns);
15019                 goto out_uts;
15020         }
15021
15022 -       new_nsp->ipc_ns = copy_ipcs(flags, tsk->nsproxy->ipc_ns);
15023 +       new_nsp->ipc_ns = copy_ipcs(flags, orig->ipc_ns);
15024         if (IS_ERR(new_nsp->ipc_ns)) {
15025                 err = PTR_ERR(new_nsp->ipc_ns);
15026                 goto out_ipc;
15027         }
15028
15029 -       new_nsp->pid_ns = copy_pid_ns(flags, task_active_pid_ns(tsk));
15030 +       new_nsp->pid_ns = copy_pid_ns(flags, orig->pid_ns);
15031         if (IS_ERR(new_nsp->pid_ns)) {
15032                 err = PTR_ERR(new_nsp->pid_ns);
15033                 goto out_pid;
15034         }
15035
15036 -       new_nsp->net_ns = copy_net_ns(flags, tsk->nsproxy->net_ns);
15037 +       /* disabled now?
15038 +       new_nsp->user_ns = copy_user_ns(flags, orig->user_ns);
15039 +       if (IS_ERR(new_nsp->user_ns)) {
15040 +               err = PTR_ERR(new_nsp->user_ns);
15041 +               goto out_user;
15042 +       } */
15043 +
15044 +       new_nsp->net_ns = copy_net_ns(flags, orig->net_ns);
15045         if (IS_ERR(new_nsp->net_ns)) {
15046                 err = PTR_ERR(new_nsp->net_ns);
15047                 goto out_net;
15048 @@ -112,6 +128,38 @@ out_ns:
15049         return ERR_PTR(err);
15050  }
15051
15052 +static struct nsproxy *create_new_namespaces(int flags, struct task_struct *tsk,
15053 +                       struct fs_struct *new_fs)
15054 +{
15055 +       return unshare_namespaces(flags, tsk->nsproxy, new_fs);
15056 +}
15057 +
15058 +/*
15059 + * copies the nsproxy, setting refcount to 1, and grabbing a
15060 + * reference to all contained namespaces.
15061 + */
15062 +struct nsproxy *copy_nsproxy(struct nsproxy *orig)
15063 +{
15064 +       struct nsproxy *ns = create_nsproxy();
15065 +
15066 +       if (ns) {
15067 +               memcpy(ns, orig, sizeof(struct nsproxy));
15068 +               atomic_set(&ns->count, 1);
15069 +
15070 +               if (ns->mnt_ns)
15071 +                       get_mnt_ns(ns->mnt_ns);
15072 +               if (ns->uts_ns)
15073 +                       get_uts_ns(ns->uts_ns);
15074 +               if (ns->ipc_ns)
15075 +                       get_ipc_ns(ns->ipc_ns);
15076 +               if (ns->pid_ns)
15077 +                       get_pid_ns(ns->pid_ns);
15078 +               if (ns->net_ns)
15079 +                       get_net(ns->net_ns);
15080 +       }
15081 +       return ns;
15082 +}
15083 +
15084  /*
15085   * called from clone.  This now handles copy for nsproxy and all
15086   * namespaces therein.
15087 @@ -119,9 +167,12 @@ out_ns:
15088  int copy_namespaces(unsigned long flags, struct task_struct *tsk)
15089  {
15090         struct nsproxy *old_ns = tsk->nsproxy;
15091 -       struct nsproxy *new_ns;
15092 +       struct nsproxy *new_ns = NULL;
15093         int err = 0;
15094
15095 +       vxdprintk(VXD_CBIT(space, 7), "copy_namespaces(0x%08lx,%p[%p])",
15096 +               flags, tsk, old_ns);
15097 +
15098         if (!old_ns)
15099                 return 0;
15100
15101 @@ -131,7 +182,7 @@ int copy_namespaces(unsigned long flags,
15102                                 CLONE_NEWPID | CLONE_NEWNET)))
15103                 return 0;
15104
15105 -       if (!capable(CAP_SYS_ADMIN)) {
15106 +       if (!vx_can_unshare(CAP_SYS_ADMIN, flags)) {
15107                 err = -EPERM;
15108                 goto out;
15109         }
15110 @@ -158,6 +209,9 @@ int copy_namespaces(unsigned long flags,
15111
15112  out:
15113         put_nsproxy(old_ns);
15114 +       vxdprintk(VXD_CBIT(space, 3),
15115 +               "copy_namespaces(0x%08lx,%p[%p]) = %d [%p]",
15116 +               flags, tsk, old_ns, err, new_ns);
15117         return err;
15118  }
15119
15120 @@ -171,7 +225,9 @@ void free_nsproxy(struct nsproxy *ns)
15121                 put_ipc_ns(ns->ipc_ns);
15122         if (ns->pid_ns)
15123                 put_pid_ns(ns->pid_ns);
15124 -       put_net(ns->net_ns);
15125 +       if (ns->net_ns)
15126 +               put_net(ns->net_ns);
15127 +       atomic_dec(&vs_global_nsproxy);
15128         kmem_cache_free(nsproxy_cachep, ns);
15129  }
15130
15131 @@ -184,11 +240,15 @@ int unshare_nsproxy_namespaces(unsigned
15132  {
15133         int err = 0;
15134
15135 +       vxdprintk(VXD_CBIT(space, 4),
15136 +               "unshare_nsproxy_namespaces(0x%08lx,[%p])",
15137 +               unshare_flags, current->nsproxy);
15138 +
15139         if (!(unshare_flags & (CLONE_NEWNS | CLONE_NEWUTS | CLONE_NEWIPC |
15140                                CLONE_NEWNET)))
15141                 return 0;
15142
15143 -       if (!capable(CAP_SYS_ADMIN))
15144 +       if (!vx_can_unshare(CAP_SYS_ADMIN, unshare_flags))
15145                 return -EPERM;
15146
15147         *new_nsp = create_new_namespaces(unshare_flags, current,
15148 diff -NurpP --minimal linux-2.6.36/kernel/pid.c linux-2.6.36-vs2.3.0.36.38/kernel/pid.c
15149 --- linux-2.6.36/kernel/pid.c   2010-10-21 13:07:57.000000000 +0200
15150 +++ linux-2.6.36-vs2.3.0.36.38/kernel/pid.c     2010-10-21 13:09:36.000000000 +0200
15151 @@ -36,6 +36,7 @@
15152  #include <linux/pid_namespace.h>
15153  #include <linux/init_task.h>
15154  #include <linux/syscalls.h>
15155 +#include <linux/vs_pid.h>
15156
15157  #define pid_hashfn(nr, ns)     \
15158         hash_long((unsigned long)nr + (unsigned long)ns, pidhash_shift)
15159 @@ -339,7 +340,7 @@ EXPORT_SYMBOL_GPL(find_pid_ns);
15160
15161  struct pid *find_vpid(int nr)
15162  {
15163 -       return find_pid_ns(nr, current->nsproxy->pid_ns);
15164 +       return find_pid_ns(vx_rmap_pid(nr), current->nsproxy->pid_ns);
15165  }
15166  EXPORT_SYMBOL_GPL(find_vpid);
15167
15168 @@ -399,6 +400,9 @@ void transfer_pid(struct task_struct *ol
15169  struct task_struct *pid_task(struct pid *pid, enum pid_type type)
15170  {
15171         struct task_struct *result = NULL;
15172 +
15173 +       if (type == PIDTYPE_REALPID)
15174 +               type = PIDTYPE_PID;
15175         if (pid) {
15176                 struct hlist_node *first;
15177                 first = rcu_dereference_check(pid->tasks[type].first,
15178 @@ -416,7 +420,7 @@ EXPORT_SYMBOL(pid_task);
15179   */
15180  struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns)
15181  {
15182 -       return pid_task(find_pid_ns(nr, ns), PIDTYPE_PID);
15183 +       return pid_task(find_pid_ns(vx_rmap_pid(nr), ns), PIDTYPE_PID);
15184  }
15185
15186  struct task_struct *find_task_by_vpid(pid_t vnr)
15187 @@ -458,7 +462,7 @@ struct pid *find_get_pid(pid_t nr)
15188  }
15189  EXPORT_SYMBOL_GPL(find_get_pid);
15190
15191 -pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15192 +pid_t pid_unmapped_nr_ns(struct pid *pid, struct pid_namespace *ns)
15193  {
15194         struct upid *upid;
15195         pid_t nr = 0;
15196 @@ -471,6 +475,11 @@ pid_t pid_nr_ns(struct pid *pid, struct
15197         return nr;
15198  }
15199
15200 +pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
15201 +{
15202 +       return vx_map_pid(pid_unmapped_nr_ns(pid, ns));
15203 +}
15204 +
15205  pid_t pid_vnr(struct pid *pid)
15206  {
15207         return pid_nr_ns(pid, current->nsproxy->pid_ns);
15208 diff -NurpP --minimal linux-2.6.36/kernel/pid_namespace.c linux-2.6.36-vs2.3.0.36.38/kernel/pid_namespace.c
15209 --- linux-2.6.36/kernel/pid_namespace.c 2010-07-07 18:31:57.000000000 +0200
15210 +++ linux-2.6.36-vs2.3.0.36.38/kernel/pid_namespace.c   2010-10-21 13:09:36.000000000 +0200
15211 @@ -14,6 +14,7 @@
15212  #include <linux/err.h>
15213  #include <linux/acct.h>
15214  #include <linux/slab.h>
15215 +#include <linux/vserver/global.h>
15216
15217  #define BITS_PER_PAGE          (PAGE_SIZE*8)
15218
15219 @@ -87,6 +88,7 @@ static struct pid_namespace *create_pid_
15220                 goto out_free_map;
15221
15222         kref_init(&ns->kref);
15223 +       atomic_inc(&vs_global_pid_ns);
15224         ns->level = level;
15225         ns->parent = get_pid_ns(parent_pid_ns);
15226
15227 @@ -112,6 +114,7 @@ static void destroy_pid_namespace(struct
15228
15229         for (i = 0; i < PIDMAP_ENTRIES; i++)
15230                 kfree(ns->pidmap[i].page);
15231 +       atomic_dec(&vs_global_pid_ns);
15232         kmem_cache_free(pid_ns_cachep, ns);
15233  }
15234
15235 diff -NurpP --minimal linux-2.6.36/kernel/posix-timers.c linux-2.6.36-vs2.3.0.36.38/kernel/posix-timers.c
15236 --- linux-2.6.36/kernel/posix-timers.c  2010-10-21 13:07:57.000000000 +0200
15237 +++ linux-2.6.36-vs2.3.0.36.38/kernel/posix-timers.c    2010-10-21 13:09:36.000000000 +0200
15238 @@ -46,6 +46,7 @@
15239  #include <linux/wait.h>
15240  #include <linux/workqueue.h>
15241  #include <linux/module.h>
15242 +#include <linux/vs_context.h>
15243
15244  /*
15245   * Management arrays for POSIX timers.  Timers are kept in slab memory
15246 @@ -363,6 +364,7 @@ int posix_timer_event(struct k_itimer *t
15247  {
15248         struct task_struct *task;
15249         int shared, ret = -1;
15250 +
15251         /*
15252          * FIXME: if ->sigq is queued we can race with
15253          * dequeue_signal()->do_schedule_next_timer().
15254 @@ -379,10 +381,18 @@ int posix_timer_event(struct k_itimer *t
15255         rcu_read_lock();
15256         task = pid_task(timr->it_pid, PIDTYPE_PID);
15257         if (task) {
15258 +               struct vx_info_save vxis;
15259 +               struct vx_info *vxi;
15260 +
15261 +               vxi = get_vx_info(task->vx_info);
15262 +               enter_vx_info(vxi, &vxis);
15263                 shared = !(timr->it_sigev_notify & SIGEV_THREAD_ID);
15264                 ret = send_sigqueue(timr->sigq, task, shared);
15265 +               leave_vx_info(&vxis);
15266 +               put_vx_info(vxi);
15267         }
15268         rcu_read_unlock();
15269 +
15270         /* If we failed to send the signal the timer stops. */
15271         return ret > 0;
15272  }
15273 diff -NurpP --minimal linux-2.6.36/kernel/printk.c linux-2.6.36-vs2.3.0.36.38/kernel/printk.c
15274 --- linux-2.6.36/kernel/printk.c        2010-10-21 13:07:57.000000000 +0200
15275 +++ linux-2.6.36-vs2.3.0.36.38/kernel/printk.c  2010-10-21 13:43:06.000000000 +0200
15276 @@ -39,6 +39,7 @@
15277  #include <linux/syslog.h>
15278  #include <linux/cpu.h>
15279  #include <linux/notifier.h>
15280 +#include <linux/vs_cvirt.h>
15281
15282  #include <asm/uaccess.h>
15283
15284 @@ -266,18 +267,15 @@ int do_syslog(int type, char __user *buf
15285         unsigned i, j, limit, count;
15286         int do_clear = 0;
15287         char c;
15288 -       int error = 0;
15289 +       int error;
15290
15291         error = security_syslog(type, from_file);
15292         if (error)
15293                 return error;
15294
15295 -       switch (type) {
15296 -       case SYSLOG_ACTION_CLOSE:       /* Close log */
15297 -               break;
15298 -       case SYSLOG_ACTION_OPEN:        /* Open log */
15299 -               break;
15300 -       case SYSLOG_ACTION_READ:        /* Read from log */
15301 +       if ((type == SYSLOG_ACTION_READ) ||
15302 +           (type == SYSLOG_ACTION_READ_ALL) ||
15303 +           (type == SYSLOG_ACTION_READ_CLEAR)) {
15304                 error = -EINVAL;
15305                 if (!buf || len < 0)
15306                         goto out;
15307 @@ -288,6 +286,16 @@ int do_syslog(int type, char __user *buf
15308                         error = -EFAULT;
15309                         goto out;
15310                 }
15311 +       }
15312 +       if (!vx_check(0, VS_ADMIN|VS_WATCH))
15313 +               return vx_do_syslog(type, buf, len);
15314 +
15315 +       switch (type) {
15316 +       case SYSLOG_ACTION_CLOSE:       /* Close log */
15317 +               break;
15318 +       case SYSLOG_ACTION_OPEN:        /* Open log */
15319 +               break;
15320 +       case SYSLOG_ACTION_READ:        /* Read from log */
15321                 error = wait_event_interruptible(log_wait,
15322                                                         (log_start - log_end));
15323                 if (error)
15324 @@ -314,16 +322,6 @@ int do_syslog(int type, char __user *buf
15325                 /* FALL THRU */
15326         /* Read last kernel messages */
15327         case SYSLOG_ACTION_READ_ALL:
15328 -               error = -EINVAL;
15329 -               if (!buf || len < 0)
15330 -                       goto out;
15331 -               error = 0;
15332 -               if (!len)
15333 -                       goto out;
15334 -               if (!access_ok(VERIFY_WRITE, buf, len)) {
15335 -                       error = -EFAULT;
15336 -                       goto out;
15337 -               }
15338                 count = len;
15339                 if (count > log_buf_len)
15340                         count = log_buf_len;
15341 diff -NurpP --minimal linux-2.6.36/kernel/ptrace.c linux-2.6.36-vs2.3.0.36.38/kernel/ptrace.c
15342 --- linux-2.6.36/kernel/ptrace.c        2010-10-21 13:07:57.000000000 +0200
15343 +++ linux-2.6.36-vs2.3.0.36.38/kernel/ptrace.c  2010-10-21 13:09:36.000000000 +0200
15344 @@ -22,6 +22,7 @@
15345  #include <linux/syscalls.h>
15346  #include <linux/uaccess.h>
15347  #include <linux/regset.h>
15348 +#include <linux/vs_context.h>
15349
15350
15351  /*
15352 @@ -150,6 +151,11 @@ int __ptrace_may_access(struct task_stru
15353                 dumpable = get_dumpable(task->mm);
15354         if (!dumpable && !capable(CAP_SYS_PTRACE))
15355                 return -EPERM;
15356 +       if (!vx_check(task->xid, VS_ADMIN_P|VS_IDENT))
15357 +               return -EPERM;
15358 +       if (!vx_check(task->xid, VS_IDENT) &&
15359 +               !task_vx_flags(task, VXF_STATE_ADMIN, 0))
15360 +               return -EACCES;
15361
15362         return security_ptrace_access_check(task, mode);
15363  }
15364 @@ -709,6 +715,10 @@ SYSCALL_DEFINE4(ptrace, long, request, l
15365                 goto out;
15366         }
15367
15368 +       ret = -EPERM;
15369 +       if (!vx_check(vx_task_xid(child), VS_WATCH_P | VS_IDENT))
15370 +               goto out_put_task_struct;
15371 +
15372         if (request == PTRACE_ATTACH) {
15373                 ret = ptrace_attach(child);
15374                 /*
15375 diff -NurpP --minimal linux-2.6.36/kernel/sched.c linux-2.6.36-vs2.3.0.36.38/kernel/sched.c
15376 --- linux-2.6.36/kernel/sched.c 2010-10-21 13:07:57.000000000 +0200
15377 +++ linux-2.6.36-vs2.3.0.36.38/kernel/sched.c   2010-10-21 13:09:36.000000000 +0200
15378 @@ -72,6 +72,8 @@
15379  #include <linux/ctype.h>
15380  #include <linux/ftrace.h>
15381  #include <linux/slab.h>
15382 +#include <linux/vs_sched.h>
15383 +#include <linux/vs_cvirt.h>
15384
15385  #include <asm/tlb.h>
15386  #include <asm/irq_regs.h>
15387 @@ -3024,9 +3026,17 @@ static inline long calc_load_fold_idle(v
15388   */
15389  void get_avenrun(unsigned long *loads, unsigned long offset, int shift)
15390  {
15391 -       loads[0] = (avenrun[0] + offset) << shift;
15392 -       loads[1] = (avenrun[1] + offset) << shift;
15393 -       loads[2] = (avenrun[2] + offset) << shift;
15394 +       if (vx_flags(VXF_VIRT_LOAD, 0)) {
15395 +               struct vx_info *vxi = current_vx_info();
15396 +
15397 +               loads[0] = (vxi->cvirt.load[0] + offset) << shift;
15398 +               loads[1] = (vxi->cvirt.load[1] + offset) << shift;
15399 +               loads[2] = (vxi->cvirt.load[2] + offset) << shift;
15400 +       } else {
15401 +               loads[0] = (avenrun[0] + offset) << shift;
15402 +               loads[1] = (avenrun[1] + offset) << shift;
15403 +               loads[2] = (avenrun[2] + offset) << shift;
15404 +       }
15405  }
15406
15407  static unsigned long
15408 @@ -3321,16 +3331,19 @@ void account_user_time(struct task_struc
15409                        cputime_t cputime_scaled)
15410  {
15411         struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15412 +       struct vx_info *vxi = p->vx_info;  /* p is _always_ current */
15413         cputime64_t tmp;
15414 +       int nice = (TASK_NICE(p) > 0);
15415
15416         /* Add user time to process. */
15417         p->utime = cputime_add(p->utime, cputime);
15418         p->utimescaled = cputime_add(p->utimescaled, cputime_scaled);
15419 +       vx_account_user(vxi, cputime, nice);
15420         account_group_user_time(p, cputime);
15421
15422         /* Add user time to cpustat. */
15423         tmp = cputime_to_cputime64(cputime);
15424 -       if (TASK_NICE(p) > 0)
15425 +       if (nice)
15426                 cpustat->nice = cputime64_add(cpustat->nice, tmp);
15427         else
15428                 cpustat->user = cputime64_add(cpustat->user, tmp);
15429 @@ -3381,6 +3394,7 @@ void account_system_time(struct task_str
15430                          cputime_t cputime, cputime_t cputime_scaled)
15431  {
15432         struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
15433 +       struct vx_info *vxi = p->vx_info;  /* p is _always_ current */
15434         cputime64_t tmp;
15435
15436         if ((p->flags & PF_VCPU) && (irq_count() - hardirq_offset == 0)) {
15437 @@ -3391,6 +3405,7 @@ void account_system_time(struct task_str
15438         /* Add system time to process. */
15439         p->stime = cputime_add(p->stime, cputime);
15440         p->stimescaled = cputime_add(p->stimescaled, cputime_scaled);
15441 +       vx_account_system(vxi, cputime, 0 /* do we have idle time? */);
15442         account_group_system_time(p, cputime);
15443
15444         /* Add system time to cpustat. */
15445 @@ -4477,7 +4492,7 @@ SYSCALL_DEFINE1(nice, int, increment)
15446                 nice = 19;
15447
15448         if (increment < 0 && !can_nice(current, nice))
15449 -               return -EPERM;
15450 +               return vx_flags(VXF_IGNEG_NICE, 0) ? 0 : -EPERM;
15451
15452         retval = security_task_setnice(current, nice);
15453         if (retval)
15454 diff -NurpP --minimal linux-2.6.36/kernel/sched_fair.c linux-2.6.36-vs2.3.0.36.38/kernel/sched_fair.c
15455 --- linux-2.6.36/kernel/sched_fair.c    2010-10-21 13:07:57.000000000 +0200
15456 +++ linux-2.6.36-vs2.3.0.36.38/kernel/sched_fair.c      2010-10-21 13:09:36.000000000 +0200
15457 @@ -782,6 +782,9 @@ enqueue_entity(struct cfs_rq *cfs_rq, st
15458         check_spread(cfs_rq, se);
15459         if (se != cfs_rq->curr)
15460                 __enqueue_entity(cfs_rq, se);
15461 +
15462 +       if (entity_is_task(se))
15463 +               vx_activate_task(task_of(se));
15464  }
15465
15466  static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
15467 @@ -825,6 +828,8 @@ dequeue_entity(struct cfs_rq *cfs_rq, st
15468
15469         if (se != cfs_rq->curr)
15470                 __dequeue_entity(cfs_rq, se);
15471 +       if (entity_is_task(se))
15472 +               vx_deactivate_task(task_of(se));
15473         account_entity_dequeue(cfs_rq, se);
15474         update_min_vruntime(cfs_rq);
15475
15476 diff -NurpP --minimal linux-2.6.36/kernel/signal.c linux-2.6.36-vs2.3.0.36.38/kernel/signal.c
15477 --- linux-2.6.36/kernel/signal.c        2010-10-21 13:07:57.000000000 +0200
15478 +++ linux-2.6.36-vs2.3.0.36.38/kernel/signal.c  2010-10-21 13:09:36.000000000 +0200
15479 @@ -28,6 +28,8 @@
15480  #include <linux/freezer.h>
15481  #include <linux/pid_namespace.h>
15482  #include <linux/nsproxy.h>
15483 +#include <linux/vs_context.h>
15484 +#include <linux/vs_pid.h>
15485  #define CREATE_TRACE_POINTS
15486  #include <trace/events/signal.h>
15487
15488 @@ -646,9 +648,18 @@ static int check_kill_permission(int sig
15489         struct pid *sid;
15490         int error;
15491
15492 +       vxdprintk(VXD_CBIT(misc, 7),
15493 +               "check_kill_permission(%d,%p,%p[#%u,%u])",
15494 +               sig, info, t, vx_task_xid(t), t->pid);
15495 +
15496         if (!valid_signal(sig))
15497                 return -EINVAL;
15498
15499 +/*     FIXME: needed? if so, why?
15500 +       if ((info != SEND_SIG_NOINFO) &&
15501 +               (is_si_special(info) || !si_fromuser(info)))
15502 +               goto skip;      */
15503 +
15504         if (!si_fromuser(info))
15505                 return 0;
15506
15507 @@ -678,6 +689,20 @@ static int check_kill_permission(int sig
15508                 }
15509         }
15510
15511 +       error = -EPERM;
15512 +       if (t->pid == 1 && current->xid)
15513 +               return error;
15514 +
15515 +       error = -ESRCH;
15516 +       /* FIXME: we shouldn't return ESRCH ever, to avoid
15517 +                 loops, maybe ENOENT or EACCES? */
15518 +       if (!vx_check(vx_task_xid(t), VS_WATCH_P | VS_IDENT)) {
15519 +               vxdprintk(current->xid || VXD_CBIT(misc, 7),
15520 +                       "signal %d[%p] xid mismatch %p[#%u,%u] xid=#%u",
15521 +                       sig, info, t, vx_task_xid(t), t->pid, current->xid);
15522 +               return error;
15523 +       }
15524 +/* skip: */
15525         return security_task_kill(t, info, sig, 0);
15526  }
15527
15528 @@ -1170,7 +1195,7 @@ int kill_pid_info(int sig, struct siginf
15529         rcu_read_lock();
15530  retry:
15531         p = pid_task(pid, PIDTYPE_PID);
15532 -       if (p) {
15533 +       if (p && vx_check(vx_task_xid(p), VS_IDENT)) {
15534                 error = group_send_sig_info(sig, info, p);
15535                 if (unlikely(error == -ESRCH))
15536                         /*
15537 @@ -1210,7 +1235,7 @@ int kill_pid_info_as_uid(int sig, struct
15538
15539         rcu_read_lock();
15540         p = pid_task(pid, PIDTYPE_PID);
15541 -       if (!p) {
15542 +       if (!p || !vx_check(vx_task_xid(p), VS_IDENT)) {
15543                 ret = -ESRCH;
15544                 goto out_unlock;
15545         }
15546 @@ -1265,8 +1290,10 @@ static int kill_something_info(int sig,
15547                 struct task_struct * p;
15548
15549                 for_each_process(p) {
15550 -                       if (task_pid_vnr(p) > 1 &&
15551 -                                       !same_thread_group(p, current)) {
15552 +                       if (vx_check(vx_task_xid(p), VS_ADMIN|VS_IDENT) &&
15553 +                               task_pid_vnr(p) > 1 &&
15554 +                               !same_thread_group(p, current) &&
15555 +                               !vx_current_initpid(p->pid)) {
15556                                 int err = group_send_sig_info(sig, info, p);
15557                                 ++count;
15558                                 if (err != -EPERM)
15559 @@ -1933,6 +1960,11 @@ relock:
15560                                 !sig_kernel_only(signr))
15561                         continue;
15562
15563 +               /* virtual init is protected against user signals */
15564 +               if ((info->si_code == SI_USER) &&
15565 +                       vx_current_initpid(current->pid))
15566 +                       continue;
15567 +
15568                 if (sig_kernel_stop(signr)) {
15569                         /*
15570                          * The default action is to stop all threads in
15571 diff -NurpP --minimal linux-2.6.36/kernel/softirq.c linux-2.6.36-vs2.3.0.36.38/kernel/softirq.c
15572 --- linux-2.6.36/kernel/softirq.c       2010-08-02 16:52:57.000000000 +0200
15573 +++ linux-2.6.36-vs2.3.0.36.38/kernel/softirq.c 2010-10-21 13:09:36.000000000 +0200
15574 @@ -24,6 +24,7 @@
15575  #include <linux/ftrace.h>
15576  #include <linux/smp.h>
15577  #include <linux/tick.h>
15578 +#include <linux/vs_context.h>
15579
15580  #define CREATE_TRACE_POINTS
15581  #include <trace/events/irq.h>
15582 diff -NurpP --minimal linux-2.6.36/kernel/sys.c linux-2.6.36-vs2.3.0.36.38/kernel/sys.c
15583 --- linux-2.6.36/kernel/sys.c   2010-10-21 13:07:57.000000000 +0200
15584 +++ linux-2.6.36-vs2.3.0.36.38/kernel/sys.c     2010-10-21 13:45:48.000000000 +0200
15585 @@ -42,6 +42,7 @@
15586  #include <linux/syscalls.h>
15587  #include <linux/kprobes.h>
15588  #include <linux/user_namespace.h>
15589 +#include <linux/vs_pid.h>
15590
15591  #include <asm/uaccess.h>
15592  #include <asm/io.h>
15593 @@ -131,7 +132,10 @@ static int set_one_prio(struct task_stru
15594                 goto out;
15595         }
15596         if (niceval < task_nice(p) && !can_nice(p, niceval)) {
15597 -               error = -EACCES;
15598 +               if (vx_flags(VXF_IGNEG_NICE, 0))
15599 +                       error = 0;
15600 +               else
15601 +                       error = -EACCES;
15602                 goto out;
15603         }
15604         no_nice = security_task_setnice(p, niceval);
15605 @@ -181,6 +185,8 @@ SYSCALL_DEFINE3(setpriority, int, which,
15606                         else
15607                                 pgrp = task_pgrp(current);
15608                         do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
15609 +                               if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
15610 +                                       continue;
15611                                 error = set_one_prio(p, niceval, error);
15612                         } while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
15613                         break;
15614 @@ -244,6 +250,8 @@ SYSCALL_DEFINE2(getpriority, int, which,
15615                         else
15616                                 pgrp = task_pgrp(current);
15617                         do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
15618 +                               if (!vx_check(p->xid, VS_ADMIN_P | VS_IDENT))
15619 +                                       continue;
15620                                 niceval = 20 - task_nice(p);
15621                                 if (niceval > retval)
15622                                         retval = niceval;
15623 @@ -357,6 +365,8 @@ EXPORT_SYMBOL_GPL(kernel_power_off);
15624
15625  static DEFINE_MUTEX(reboot_mutex);
15626
15627 +long vs_reboot(unsigned int, void __user *);
15628 +
15629  /*
15630   * Reboot system call: for obvious reasons only root may call it,
15631   * and even root needs to set up some magic numbers in the registers
15632 @@ -389,6 +399,9 @@ SYSCALL_DEFINE4(reboot, int, magic1, int
15633         if ((cmd == LINUX_REBOOT_CMD_POWER_OFF) && !pm_power_off)
15634                 cmd = LINUX_REBOOT_CMD_HALT;
15635
15636 +       if (!vx_check(0, VS_ADMIN|VS_WATCH))
15637 +               return vs_reboot(cmd, arg);
15638 +
15639         mutex_lock(&reboot_mutex);
15640         switch (cmd) {
15641         case LINUX_REBOOT_CMD_RESTART:
15642 @@ -1169,7 +1182,7 @@ SYSCALL_DEFINE2(sethostname, char __user
15643         int errno;
15644         char tmp[__NEW_UTS_LEN];
15645
15646 -       if (!capable(CAP_SYS_ADMIN))
15647 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
15648                 return -EPERM;
15649         if (len < 0 || len > __NEW_UTS_LEN)
15650                 return -EINVAL;
15651 @@ -1218,7 +1231,7 @@ SYSCALL_DEFINE2(setdomainname, char __us
15652         int errno;
15653         char tmp[__NEW_UTS_LEN];
15654
15655 -       if (!capable(CAP_SYS_ADMIN))
15656 +       if (!vx_capable(CAP_SYS_ADMIN, VXC_SET_UTSNAME))
15657                 return -EPERM;
15658         if (len < 0 || len > __NEW_UTS_LEN)
15659                 return -EINVAL;
15660 @@ -1334,7 +1347,7 @@ int do_prlimit(struct task_struct *tsk,
15661         task_lock(tsk->group_leader);
15662         if (new_rlim) {
15663                 if (new_rlim->rlim_max > rlim->rlim_max &&
15664 -                               !capable(CAP_SYS_RESOURCE))
15665 +                       !vx_capable(CAP_SYS_RESOURCE, VXC_SET_RLIMIT))
15666                         retval = -EPERM;
15667                 if (!retval)
15668                         retval = security_task_setrlimit(tsk->group_leader,
15669 @@ -1383,7 +1396,7 @@ static int check_prlimit_permission(stru
15670              cred->gid != tcred->egid ||
15671              cred->gid != tcred->sgid ||
15672              cred->gid != tcred->gid) &&
15673 -            !capable(CAP_SYS_RESOURCE)) {
15674 +            !vx_capable(CAP_SYS_RESOURCE, VXC_SET_RLIMIT)) {
15675                 return -EPERM;
15676         }
15677
15678 diff -NurpP --minimal linux-2.6.36/kernel/sysctl.c linux-2.6.36-vs2.3.0.36.38/kernel/sysctl.c
15679 --- linux-2.6.36/kernel/sysctl.c        2010-10-21 13:07:57.000000000 +0200
15680 +++ linux-2.6.36-vs2.3.0.36.38/kernel/sysctl.c  2010-10-21 13:09:36.000000000 +0200
15681 @@ -73,6 +73,7 @@
15682  #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
15683  #include <linux/lockdep.h>
15684  #endif
15685 +extern char vshelper_path[];
15686  #ifdef CONFIG_CHR_DEV_SG
15687  #include <scsi/sg.h>
15688  #endif
15689 @@ -573,6 +574,13 @@ static struct ctl_table kern_table[] = {
15690                 .proc_handler   = proc_dostring,
15691         },
15692  #endif
15693 +       {
15694 +               .procname       = "vshelper",
15695 +               .data           = &vshelper_path,
15696 +               .maxlen         = 256,
15697 +               .mode           = 0644,
15698 +               .proc_handler   = &proc_dostring,
15699 +       },
15700  #ifdef CONFIG_CHR_DEV_SG
15701         {
15702                 .procname       = "sg-big-buff",
15703 diff -NurpP --minimal linux-2.6.36/kernel/sysctl_binary.c linux-2.6.36-vs2.3.0.36.38/kernel/sysctl_binary.c
15704 --- linux-2.6.36/kernel/sysctl_binary.c 2010-08-02 16:52:57.000000000 +0200
15705 +++ linux-2.6.36-vs2.3.0.36.38/kernel/sysctl_binary.c   2010-10-21 13:09:36.000000000 +0200
15706 @@ -73,6 +73,7 @@ static const struct bin_table bin_kern_t
15707
15708         { CTL_INT,      KERN_PANIC,                     "panic" },
15709         { CTL_INT,      KERN_REALROOTDEV,               "real-root-dev" },
15710 +       { CTL_STR,      KERN_VSHELPER,                  "vshelper" },
15711
15712         { CTL_STR,      KERN_SPARC_REBOOT,              "reboot-cmd" },
15713         { CTL_INT,      KERN_CTLALTDEL,                 "ctrl-alt-del" },
15714 diff -NurpP --minimal linux-2.6.36/kernel/time/timekeeping.c linux-2.6.36-vs2.3.0.36.38/kernel/time/timekeeping.c
15715 --- linux-2.6.36/kernel/time/timekeeping.c      2010-10-21 13:07:57.000000000 +0200
15716 +++ linux-2.6.36-vs2.3.0.36.38/kernel/time/timekeeping.c        2010-10-21 14:17:59.000000000 +0200
15717 @@ -229,6 +229,7 @@ void getnstimeofday(struct timespec *ts)
15718         } while (read_seqretry(&xtime_lock, seq));
15719
15720         timespec_add_ns(ts, nsecs);
15721 +       vx_adjust_timespec(ts);
15722  }
15723
15724  EXPORT_SYMBOL(getnstimeofday);
15725 diff -NurpP --minimal linux-2.6.36/kernel/time.c linux-2.6.36-vs2.3.0.36.38/kernel/time.c
15726 --- linux-2.6.36/kernel/time.c  2010-10-21 13:07:57.000000000 +0200
15727 +++ linux-2.6.36-vs2.3.0.36.38/kernel/time.c    2010-10-21 14:22:43.000000000 +0200
15728 @@ -92,7 +92,7 @@ SYSCALL_DEFINE1(stime, time_t __user *,
15729         if (err)
15730                 return err;
15731
15732 -       do_settimeofday(&tv);
15733 +       vx_settimeofday(&tv);
15734         return 0;
15735  }
15736
15737 @@ -177,7 +177,7 @@ int do_sys_settimeofday(struct timespec
15738                 /* SMP safe, again the code in arch/foo/time.c should
15739                  * globally block out interrupts when it runs.
15740                  */
15741 -               return do_settimeofday(tv);
15742 +               return vx_settimeofday(tv);
15743         }
15744         return 0;
15745  }
15746 diff -NurpP --minimal linux-2.6.36/kernel/timer.c linux-2.6.36-vs2.3.0.36.38/kernel/timer.c
15747 --- linux-2.6.36/kernel/timer.c 2010-10-21 13:07:57.000000000 +0200
15748 +++ linux-2.6.36-vs2.3.0.36.38/kernel/timer.c   2010-10-21 13:09:36.000000000 +0200
15749 @@ -40,6 +40,10 @@
15750  #include <linux/perf_event.h>
15751  #include <linux/sched.h>
15752  #include <linux/slab.h>
15753 +#include <linux/vs_base.h>
15754 +#include <linux/vs_cvirt.h>
15755 +#include <linux/vs_pid.h>
15756 +#include <linux/vserver/sched.h>
15757
15758  #include <asm/uaccess.h>
15759  #include <asm/unistd.h>
15760 @@ -1332,12 +1336,6 @@ SYSCALL_DEFINE1(alarm, unsigned int, sec
15761
15762  #endif
15763
15764 -#ifndef __alpha__
15765 -
15766 -/*
15767 - * The Alpha uses getxpid, getxuid, and getxgid instead.  Maybe this
15768 - * should be moved into arch/i386 instead?
15769 - */
15770
15771  /**
15772   * sys_getpid - return the thread group id of the current process
15773 @@ -1366,10 +1364,23 @@ SYSCALL_DEFINE0(getppid)
15774         rcu_read_lock();
15775         pid = task_tgid_vnr(current->real_parent);
15776         rcu_read_unlock();
15777 +       return vx_map_pid(pid);
15778 +}
15779
15780 -       return pid;
15781 +#ifdef __alpha__
15782 +
15783 +/*
15784 + * The Alpha uses getxpid, getxuid, and getxgid instead.
15785 + */
15786 +
15787 +asmlinkage long do_getxpid(long *ppid)
15788 +{
15789 +       *ppid = sys_getppid();
15790 +       return sys_getpid();
15791  }
15792
15793 +#else /* _alpha_ */
15794 +
15795  SYSCALL_DEFINE0(getuid)
15796  {
15797         /* Only we change this so SMP safe */
15798 diff -NurpP --minimal linux-2.6.36/kernel/user_namespace.c linux-2.6.36-vs2.3.0.36.38/kernel/user_namespace.c
15799 --- linux-2.6.36/kernel/user_namespace.c        2010-10-21 13:07:57.000000000 +0200
15800 +++ linux-2.6.36-vs2.3.0.36.38/kernel/user_namespace.c  2010-10-21 13:09:36.000000000 +0200
15801 @@ -11,6 +11,7 @@
15802  #include <linux/user_namespace.h>
15803  #include <linux/highuid.h>
15804  #include <linux/cred.h>
15805 +#include <linux/vserver/global.h>
15806
15807  /*
15808   * Create a new user namespace, deriving the creator from the user in the
15809 @@ -31,6 +32,7 @@ int create_user_ns(struct cred *new)
15810                 return -ENOMEM;
15811
15812         kref_init(&ns->kref);
15813 +       atomic_inc(&vs_global_user_ns);
15814
15815         for (n = 0; n < UIDHASH_SZ; ++n)
15816                 INIT_HLIST_HEAD(ns->uidhash_table + n);
15817 @@ -79,6 +81,8 @@ void free_user_ns(struct kref *kref)
15818         struct user_namespace *ns =
15819                 container_of(kref, struct user_namespace, kref);
15820
15821 +       /* FIXME: maybe move into destroyer? */
15822 +       atomic_dec(&vs_global_user_ns);
15823         INIT_WORK(&ns->destroyer, free_user_ns_work);
15824         schedule_work(&ns->destroyer);
15825  }
15826 diff -NurpP --minimal linux-2.6.36/kernel/utsname.c linux-2.6.36-vs2.3.0.36.38/kernel/utsname.c
15827 --- linux-2.6.36/kernel/utsname.c       2009-09-10 15:26:28.000000000 +0200
15828 +++ linux-2.6.36-vs2.3.0.36.38/kernel/utsname.c 2010-10-21 13:09:36.000000000 +0200
15829 @@ -14,14 +14,17 @@
15830  #include <linux/utsname.h>
15831  #include <linux/err.h>
15832  #include <linux/slab.h>
15833 +#include <linux/vserver/global.h>
15834
15835  static struct uts_namespace *create_uts_ns(void)
15836  {
15837         struct uts_namespace *uts_ns;
15838
15839         uts_ns = kmalloc(sizeof(struct uts_namespace), GFP_KERNEL);
15840 -       if (uts_ns)
15841 +       if (uts_ns) {
15842                 kref_init(&uts_ns->kref);
15843 +               atomic_inc(&vs_global_uts_ns);
15844 +       }
15845         return uts_ns;
15846  }
15847
15848 @@ -71,5 +74,6 @@ void free_uts_ns(struct kref *kref)
15849         struct uts_namespace *ns;
15850
15851         ns = container_of(kref, struct uts_namespace, kref);
15852 +       atomic_dec(&vs_global_uts_ns);
15853         kfree(ns);
15854  }
15855 diff -NurpP --minimal linux-2.6.36/kernel/vserver/Kconfig linux-2.6.36-vs2.3.0.36.38/kernel/vserver/Kconfig
15856 --- linux-2.6.36/kernel/vserver/Kconfig 1970-01-01 01:00:00.000000000 +0100
15857 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/Kconfig   2010-10-21 21:49:28.000000000 +0200
15858 @@ -0,0 +1,265 @@
15859 +#
15860 +# Linux VServer configuration
15861 +#
15862 +
15863 +menu "Linux VServer"
15864 +
15865 +config VSERVER_AUTO_LBACK
15866 +       bool    "Automatically Assign Loopback IP"
15867 +       default y
15868 +       help
15869 +         Automatically assign a guest specific loopback
15870 +         IP and add it to the kernel network stack on
15871 +         startup.
15872 +
15873 +config VSERVER_AUTO_SINGLE
15874 +       bool    "Automatic Single IP Special Casing"
15875 +       depends on EXPERIMENTAL
15876 +       default y
15877 +       help
15878 +         This allows network contexts with a single IP to
15879 +         automatically remap 0.0.0.0 bindings to that IP,
15880 +         avoiding further network checks and improving
15881 +         performance.
15882 +
15883 +         (note: such guests do not allow to change the ip
15884 +          on the fly and do not show loopback addresses)
15885 +
15886 +config VSERVER_COWBL
15887 +       bool    "Enable COW Immutable Link Breaking"
15888 +       default y
15889 +       help
15890 +         This enables the COW (Copy-On-Write) link break code.
15891 +         It allows you to treat unified files like normal files
15892 +         when writing to them (which will implicitely break the
15893 +         link and create a copy of the unified file)
15894 +
15895 +config VSERVER_VTIME
15896 +       bool    "Enable Virtualized Guest Time"
15897 +       depends on EXPERIMENTAL
15898 +       default n
15899 +       help
15900 +         This enables per guest time offsets to allow for
15901 +         adjusting the system clock individually per guest.
15902 +         this adds some overhead to the time functions and
15903 +         therefore should not be enabled without good reason.
15904 +
15905 +config VSERVER_DEVICE
15906 +       bool    "Enable Guest Device Mapping"
15907 +       depends on EXPERIMENTAL
15908 +       default n
15909 +       help
15910 +         This enables generic device remapping.
15911 +
15912 +config VSERVER_PROC_SECURE
15913 +       bool    "Enable Proc Security"
15914 +       depends on PROC_FS
15915 +       default y
15916 +       help
15917 +         This configures ProcFS security to initially hide
15918 +         non-process entries for all contexts except the main and
15919 +         spectator context (i.e. for all guests), which is a secure
15920 +         default.
15921 +
15922 +         (note: on 1.2x the entries were visible by default)
15923 +
15924 +config VSERVER_HARDCPU
15925 +       bool    "Enable Hard CPU Limits"
15926 +       default y
15927 +       help
15928 +         Activate the Hard CPU Limits
15929 +
15930 +         This will compile in code that allows the Token Bucket
15931 +         Scheduler to put processes on hold when a context's
15932 +         tokens are depleted (provided that its per-context
15933 +         sched_hard flag is set).
15934 +
15935 +         Processes belonging to that context will not be able
15936 +         to consume CPU resources again until a per-context
15937 +         configured minimum of tokens has been reached.
15938 +
15939 +config VSERVER_IDLETIME
15940 +       bool    "Avoid idle CPUs by skipping Time"
15941 +       depends on VSERVER_HARDCPU
15942 +       default y
15943 +       help
15944 +         This option allows the scheduler to artificially
15945 +         advance time (per cpu) when otherwise the idle
15946 +         task would be scheduled, thus keeping the cpu
15947 +         busy and sharing the available resources among
15948 +         certain contexts.
15949 +
15950 +config VSERVER_IDLELIMIT
15951 +       bool    "Limit the IDLE task"
15952 +       depends on VSERVER_HARDCPU
15953 +       default n
15954 +       help
15955 +         Limit the idle slices, so the the next context
15956 +         will be scheduled as soon as possible.
15957 +
15958 +         This might improve interactivity and latency, but
15959 +         will also marginally increase scheduling overhead.
15960 +
15961 +choice
15962 +       prompt  "Persistent Inode Tagging"
15963 +       default TAGGING_ID24
15964 +       help
15965 +         This adds persistent context information to filesystems
15966 +         mounted with the tagxid option. Tagging is a requirement
15967 +         for per-context disk limits and per-context quota.
15968 +
15969 +
15970 +config TAGGING_NONE
15971 +       bool    "Disabled"
15972 +       help
15973 +         do not store per-context information in inodes.
15974 +
15975 +config TAGGING_UID16
15976 +       bool    "UID16/GID32"
15977 +       help
15978 +         reduces UID to 16 bit, but leaves GID at 32 bit.
15979 +
15980 +config TAGGING_GID16
15981 +       bool    "UID32/GID16"
15982 +       help
15983 +         reduces GID to 16 bit, but leaves UID at 32 bit.
15984 +
15985 +config TAGGING_ID24
15986 +       bool    "UID24/GID24"
15987 +       help
15988 +         uses the upper 8bit from UID and GID for XID tagging
15989 +         which leaves 24bit for UID/GID each, which should be
15990 +         more than sufficient for normal use.
15991 +
15992 +config TAGGING_INTERN
15993 +       bool    "UID32/GID32"
15994 +       help
15995 +         this uses otherwise reserved inode fields in the on
15996 +         disk representation, which limits the use to a few
15997 +         filesystems (currently ext2 and ext3)
15998 +
15999 +endchoice
16000 +
16001 +config TAG_NFSD
16002 +       bool    "Tag NFSD User Auth and Files"
16003 +       default n
16004 +       help
16005 +         Enable this if you do want the in-kernel NFS
16006 +         Server to use the tagging specified above.
16007 +         (will require patched clients too)
16008 +
16009 +config VSERVER_PRIVACY
16010 +       bool    "Honor Privacy Aspects of Guests"
16011 +       default n
16012 +       help
16013 +         When enabled, most context checks will disallow
16014 +         access to structures assigned to a specific context,
16015 +         like ptys or loop devices.
16016 +
16017 +config VSERVER_CONTEXTS
16018 +       int     "Maximum number of Contexts (1-65533)"  if EMBEDDED
16019 +       range 1 65533
16020 +       default "768"   if 64BIT
16021 +       default "256"
16022 +       help
16023 +         This setting will optimize certain data structures
16024 +         and memory allocations according to the expected
16025 +         maximum.
16026 +
16027 +         note: this is not a strict upper limit.
16028 +
16029 +config VSERVER_WARN
16030 +       bool    "VServer Warnings"
16031 +       default y
16032 +       help
16033 +         This enables various runtime warnings, which will
16034 +         notify about potential manipulation attempts or
16035 +         resource shortage. It is generally considered to
16036 +         be a good idea to have that enabled.
16037 +
16038 +config VSERVER_WARN_DEVPTS
16039 +       bool    "VServer DevPTS Warnings"
16040 +       depends on VSERVER_WARN
16041 +       default y
16042 +       help
16043 +         This enables DevPTS related warnings, issued when a
16044 +         process inside a context tries to lookup or access
16045 +         a dynamic pts from the host or a different context.
16046 +
16047 +config VSERVER_DEBUG
16048 +       bool    "VServer Debugging Code"
16049 +       default n
16050 +       help
16051 +         Set this to yes if you want to be able to activate
16052 +         debugging output at runtime. It adds a very small
16053 +         overhead to all vserver related functions and
16054 +         increases the kernel size by about 20k.
16055 +
16056 +config VSERVER_HISTORY
16057 +       bool    "VServer History Tracing"
16058 +       depends on VSERVER_DEBUG
16059 +       default n
16060 +       help
16061 +         Set this to yes if you want to record the history of
16062 +         linux-vserver activities, so they can be replayed in
16063 +         the event of a kernel panic or oops.
16064 +
16065 +config VSERVER_HISTORY_SIZE
16066 +       int     "Per-CPU History Size (32-65536)"
16067 +       depends on VSERVER_HISTORY
16068 +       range 32 65536
16069 +       default 64
16070 +       help
16071 +         This allows you to specify the number of entries in
16072 +         the per-CPU history buffer.
16073 +
16074 +config VSERVER_LEGACY_MEM
16075 +       bool    "Legacy Memory Limits"
16076 +       default n
16077 +       help
16078 +         This provides fake memory limits to keep
16079 +         older tools happy in the face of memory
16080 +         cgroups
16081 +
16082 +choice
16083 +       prompt  "Quotes used in debug and warn messages"
16084 +       default QUOTES_ISO8859
16085 +
16086 +config QUOTES_ISO8859
16087 +       bool    "Extended ASCII (ISO 8859) angle quotes"
16088 +       help
16089 +         This uses the extended ASCII characters \xbb
16090 +         and \xab for quoting file and process names.
16091 +
16092 +config QUOTES_UTF8
16093 +       bool    "UTF-8 angle quotes"
16094 +       help
16095 +         This uses the the UTF-8 sequences for angle
16096 +         quotes to quote file and process names.
16097 +
16098 +config QUOTES_ASCII
16099 +       bool    "ASCII single quotes"
16100 +       help
16101 +         This uses the ASCII single quote character
16102 +         (\x27) to quote file and process names.
16103 +
16104 +endchoice
16105 +
16106 +endmenu
16107 +
16108 +
16109 +config VSERVER
16110 +       bool
16111 +       default y
16112 +       select NAMESPACES
16113 +       select UTS_NS
16114 +       select IPC_NS
16115 +       select USER_NS
16116 +       select SYSVIPC
16117 +
16118 +config VSERVER_SECURITY
16119 +       bool
16120 +       depends on SECURITY
16121 +       default y
16122 +       select SECURITY_CAPABILITIES
16123 +
16124 diff -NurpP --minimal linux-2.6.36/kernel/vserver/Makefile linux-2.6.36-vs2.3.0.36.38/kernel/vserver/Makefile
16125 --- linux-2.6.36/kernel/vserver/Makefile        1970-01-01 01:00:00.000000000 +0100
16126 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/Makefile  2010-10-21 13:09:36.000000000 +0200
16127 @@ -0,0 +1,18 @@
16128 +#
16129 +# Makefile for the Linux vserver routines.
16130 +#
16131 +
16132 +
16133 +obj-y          += vserver.o
16134 +
16135 +vserver-y      := switch.o context.o space.o sched.o network.o inode.o \
16136 +                  limit.o cvirt.o cacct.o signal.o helper.o init.o \
16137 +                  dlimit.o tag.o
16138 +
16139 +vserver-$(CONFIG_INET) += inet.o
16140 +vserver-$(CONFIG_PROC_FS) += proc.o
16141 +vserver-$(CONFIG_VSERVER_DEBUG) += sysctl.o debug.o
16142 +vserver-$(CONFIG_VSERVER_HISTORY) += history.o
16143 +vserver-$(CONFIG_VSERVER_MONITOR) += monitor.o
16144 +vserver-$(CONFIG_VSERVER_DEVICE) += device.o
16145 +
16146 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cacct.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct.c
16147 --- linux-2.6.36/kernel/vserver/cacct.c 1970-01-01 01:00:00.000000000 +0100
16148 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct.c   2010-10-21 13:09:36.000000000 +0200
16149 @@ -0,0 +1,42 @@
16150 +/*
16151 + *  linux/kernel/vserver/cacct.c
16152 + *
16153 + *  Virtual Server: Context Accounting
16154 + *
16155 + *  Copyright (C) 2006-2007 Herbert Pötzl
16156 + *
16157 + *  V0.01  added accounting stats
16158 + *
16159 + */
16160 +
16161 +#include <linux/types.h>
16162 +#include <linux/vs_context.h>
16163 +#include <linux/vserver/cacct_cmd.h>
16164 +#include <linux/vserver/cacct_int.h>
16165 +
16166 +#include <asm/errno.h>
16167 +#include <asm/uaccess.h>
16168 +
16169 +
16170 +int vc_sock_stat(struct vx_info *vxi, void __user *data)
16171 +{
16172 +       struct vcmd_sock_stat_v0 vc_data;
16173 +       int j, field;
16174 +
16175 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
16176 +               return -EFAULT;
16177 +
16178 +       field = vc_data.field;
16179 +       if ((field < 0) || (field >= VXA_SOCK_SIZE))
16180 +               return -EINVAL;
16181 +
16182 +       for (j = 0; j < 3; j++) {
16183 +               vc_data.count[j] = vx_sock_count(&vxi->cacct, field, j);
16184 +               vc_data.total[j] = vx_sock_total(&vxi->cacct, field, j);
16185 +       }
16186 +
16187 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
16188 +               return -EFAULT;
16189 +       return 0;
16190 +}
16191 +
16192 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cacct_init.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct_init.h
16193 --- linux-2.6.36/kernel/vserver/cacct_init.h    1970-01-01 01:00:00.000000000 +0100
16194 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct_init.h      2010-10-21 13:09:36.000000000 +0200
16195 @@ -0,0 +1,25 @@
16196 +
16197 +
16198 +static inline void vx_info_init_cacct(struct _vx_cacct *cacct)
16199 +{
16200 +       int i, j;
16201 +
16202 +
16203 +       for (i = 0; i < VXA_SOCK_SIZE; i++) {
16204 +               for (j = 0; j < 3; j++) {
16205 +                       atomic_long_set(&cacct->sock[i][j].count, 0);
16206 +                       atomic_long_set(&cacct->sock[i][j].total, 0);
16207 +               }
16208 +       }
16209 +       for (i = 0; i < 8; i++)
16210 +               atomic_set(&cacct->slab[i], 0);
16211 +       for (i = 0; i < 5; i++)
16212 +               for (j = 0; j < 4; j++)
16213 +                       atomic_set(&cacct->page[i][j], 0);
16214 +}
16215 +
16216 +static inline void vx_info_exit_cacct(struct _vx_cacct *cacct)
16217 +{
16218 +       return;
16219 +}
16220 +
16221 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cacct_proc.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct_proc.h
16222 --- linux-2.6.36/kernel/vserver/cacct_proc.h    1970-01-01 01:00:00.000000000 +0100
16223 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cacct_proc.h      2010-10-21 13:09:36.000000000 +0200
16224 @@ -0,0 +1,53 @@
16225 +#ifndef _VX_CACCT_PROC_H
16226 +#define _VX_CACCT_PROC_H
16227 +
16228 +#include <linux/vserver/cacct_int.h>
16229 +
16230 +
16231 +#define VX_SOCKA_TOP   \
16232 +       "Type\t    recv #/bytes\t\t   send #/bytes\t\t    fail #/bytes\n"
16233 +
16234 +static inline int vx_info_proc_cacct(struct _vx_cacct *cacct, char *buffer)
16235 +{
16236 +       int i, j, length = 0;
16237 +       static char *type[VXA_SOCK_SIZE] = {
16238 +               "UNSPEC", "UNIX", "INET", "INET6", "PACKET", "OTHER"
16239 +       };
16240 +
16241 +       length += sprintf(buffer + length, VX_SOCKA_TOP);
16242 +       for (i = 0; i < VXA_SOCK_SIZE; i++) {
16243 +               length += sprintf(buffer + length, "%s:", type[i]);
16244 +               for (j = 0; j < 3; j++) {
16245 +                       length += sprintf(buffer + length,
16246 +                               "\t%10lu/%-10lu",
16247 +                               vx_sock_count(cacct, i, j),
16248 +                               vx_sock_total(cacct, i, j));
16249 +               }
16250 +               buffer[length++] = '\n';
16251 +       }
16252 +
16253 +       length += sprintf(buffer + length, "\n");
16254 +       length += sprintf(buffer + length,
16255 +               "slab:\t %8u %8u %8u %8u\n",
16256 +               atomic_read(&cacct->slab[1]),
16257 +               atomic_read(&cacct->slab[4]),
16258 +               atomic_read(&cacct->slab[0]),
16259 +               atomic_read(&cacct->slab[2]));
16260 +
16261 +       length += sprintf(buffer + length, "\n");
16262 +       for (i = 0; i < 5; i++) {
16263 +               length += sprintf(buffer + length,
16264 +                       "page[%d]: %8u %8u %8u %8u\t %8u %8u %8u %8u\n", i,
16265 +                       atomic_read(&cacct->page[i][0]),
16266 +                       atomic_read(&cacct->page[i][1]),
16267 +                       atomic_read(&cacct->page[i][2]),
16268 +                       atomic_read(&cacct->page[i][3]),
16269 +                       atomic_read(&cacct->page[i][4]),
16270 +                       atomic_read(&cacct->page[i][5]),
16271 +                       atomic_read(&cacct->page[i][6]),
16272 +                       atomic_read(&cacct->page[i][7]));
16273 +       }
16274 +       return length;
16275 +}
16276 +
16277 +#endif /* _VX_CACCT_PROC_H */
16278 diff -NurpP --minimal linux-2.6.36/kernel/vserver/context.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/context.c
16279 --- linux-2.6.36/kernel/vserver/context.c       1970-01-01 01:00:00.000000000 +0100
16280 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/context.c 2010-10-21 14:39:59.000000000 +0200
16281 @@ -0,0 +1,1058 @@
16282 +/*
16283 + *  linux/kernel/vserver/context.c
16284 + *
16285 + *  Virtual Server: Context Support
16286 + *
16287 + *  Copyright (C) 2003-2007  Herbert Pötzl
16288 + *
16289 + *  V0.01  context helper
16290 + *  V0.02  vx_ctx_kill syscall command
16291 + *  V0.03  replaced context_info calls
16292 + *  V0.04  redesign of struct (de)alloc
16293 + *  V0.05  rlimit basic implementation
16294 + *  V0.06  task_xid and info commands
16295 + *  V0.07  context flags and caps
16296 + *  V0.08  switch to RCU based hash
16297 + *  V0.09  revert to non RCU for now
16298 + *  V0.10  and back to working RCU hash
16299 + *  V0.11  and back to locking again
16300 + *  V0.12  referenced context store
16301 + *  V0.13  separate per cpu data
16302 + *  V0.14  changed vcmds to vxi arg
16303 + *  V0.15  added context stat
16304 + *  V0.16  have __create claim() the vxi
16305 + *  V0.17  removed older and legacy stuff
16306 + *
16307 + */
16308 +
16309 +#include <linux/slab.h>
16310 +#include <linux/types.h>
16311 +#include <linux/security.h>
16312 +#include <linux/pid_namespace.h>
16313 +
16314 +#include <linux/vserver/context.h>
16315 +#include <linux/vserver/network.h>
16316 +#include <linux/vserver/debug.h>
16317 +#include <linux/vserver/limit.h>
16318 +#include <linux/vserver/limit_int.h>
16319 +#include <linux/vserver/space.h>
16320 +#include <linux/init_task.h>
16321 +#include <linux/fs_struct.h>
16322 +
16323 +#include <linux/vs_context.h>
16324 +#include <linux/vs_limit.h>
16325 +#include <linux/vs_pid.h>
16326 +#include <linux/vserver/context_cmd.h>
16327 +
16328 +#include "cvirt_init.h"
16329 +#include "cacct_init.h"
16330 +#include "limit_init.h"
16331 +#include "sched_init.h"
16332 +
16333 +
16334 +atomic_t vx_global_ctotal      = ATOMIC_INIT(0);
16335 +atomic_t vx_global_cactive     = ATOMIC_INIT(0);
16336 +
16337 +
16338 +/*     now inactive context structures */
16339 +
16340 +static struct hlist_head vx_info_inactive = HLIST_HEAD_INIT;
16341 +
16342 +static spinlock_t vx_info_inactive_lock = SPIN_LOCK_UNLOCKED;
16343 +
16344 +
16345 +/*     __alloc_vx_info()
16346 +
16347 +       * allocate an initialized vx_info struct
16348 +       * doesn't make it visible (hash)                        */
16349 +
16350 +static struct vx_info *__alloc_vx_info(xid_t xid)
16351 +{
16352 +       struct vx_info *new = NULL;
16353 +       int cpu, index;
16354 +
16355 +       vxdprintk(VXD_CBIT(xid, 0), "alloc_vx_info(%d)*", xid);
16356 +
16357 +       /* would this benefit from a slab cache? */
16358 +       new = kmalloc(sizeof(struct vx_info), GFP_KERNEL);
16359 +       if (!new)
16360 +               return 0;
16361 +
16362 +       memset(new, 0, sizeof(struct vx_info));
16363 +#ifdef CONFIG_SMP
16364 +       new->ptr_pc = alloc_percpu(struct _vx_info_pc);
16365 +       if (!new->ptr_pc)
16366 +               goto error;
16367 +#endif
16368 +       new->vx_id = xid;
16369 +       INIT_HLIST_NODE(&new->vx_hlist);
16370 +       atomic_set(&new->vx_usecnt, 0);
16371 +       atomic_set(&new->vx_tasks, 0);
16372 +       new->vx_parent = NULL;
16373 +       new->vx_state = 0;
16374 +       init_waitqueue_head(&new->vx_wait);
16375 +
16376 +       /* prepare reaper */
16377 +       get_task_struct(init_pid_ns.child_reaper);
16378 +       new->vx_reaper = init_pid_ns.child_reaper;
16379 +       new->vx_badness_bias = 0;
16380 +
16381 +       /* rest of init goes here */
16382 +       vx_info_init_limit(&new->limit);
16383 +       vx_info_init_sched(&new->sched);
16384 +       vx_info_init_cvirt(&new->cvirt);
16385 +       vx_info_init_cacct(&new->cacct);
16386 +
16387 +       /* per cpu data structures */
16388 +       for_each_possible_cpu(cpu) {
16389 +               vx_info_init_sched_pc(
16390 +                       &vx_per_cpu(new, sched_pc, cpu), cpu);
16391 +               vx_info_init_cvirt_pc(
16392 +                       &vx_per_cpu(new, cvirt_pc, cpu), cpu);
16393 +       }
16394 +
16395 +       new->vx_flags = VXF_INIT_SET;
16396 +       cap_set_init_eff(new->vx_bcaps);
16397 +       new->vx_ccaps = 0;
16398 +       new->vx_umask = 0;
16399 +
16400 +       new->reboot_cmd = 0;
16401 +       new->exit_code = 0;
16402 +
16403 +       // preconfig fs entries
16404 +       for (index = 0; index < VX_SPACES; index++) {
16405 +               spin_lock(&init_fs.lock);
16406 +               init_fs.users++;
16407 +               spin_unlock(&init_fs.lock);
16408 +               new->vx_fs[index] = &init_fs;
16409 +       }
16410 +
16411 +       vxdprintk(VXD_CBIT(xid, 0),
16412 +               "alloc_vx_info(%d) = %p", xid, new);
16413 +       vxh_alloc_vx_info(new);
16414 +       atomic_inc(&vx_global_ctotal);
16415 +       return new;
16416 +#ifdef CONFIG_SMP
16417 +error:
16418 +       kfree(new);
16419 +       return 0;
16420 +#endif
16421 +}
16422 +
16423 +/*     __dealloc_vx_info()
16424 +
16425 +       * final disposal of vx_info                             */
16426 +
16427 +static void __dealloc_vx_info(struct vx_info *vxi)
16428 +{
16429 +#ifdef CONFIG_VSERVER_WARN
16430 +       struct vx_info_save vxis;
16431 +       int cpu;
16432 +#endif
16433 +       vxdprintk(VXD_CBIT(xid, 0),
16434 +               "dealloc_vx_info(%p)", vxi);
16435 +       vxh_dealloc_vx_info(vxi);
16436 +
16437 +#ifdef CONFIG_VSERVER_WARN
16438 +       enter_vx_info(vxi, &vxis);
16439 +       vx_info_exit_limit(&vxi->limit);
16440 +       vx_info_exit_sched(&vxi->sched);
16441 +       vx_info_exit_cvirt(&vxi->cvirt);
16442 +       vx_info_exit_cacct(&vxi->cacct);
16443 +
16444 +       for_each_possible_cpu(cpu) {
16445 +               vx_info_exit_sched_pc(
16446 +                       &vx_per_cpu(vxi, sched_pc, cpu), cpu);
16447 +               vx_info_exit_cvirt_pc(
16448 +                       &vx_per_cpu(vxi, cvirt_pc, cpu), cpu);
16449 +       }
16450 +       leave_vx_info(&vxis);
16451 +#endif
16452 +
16453 +       vxi->vx_id = -1;
16454 +       vxi->vx_state |= VXS_RELEASED;
16455 +
16456 +#ifdef CONFIG_SMP
16457 +       free_percpu(vxi->ptr_pc);
16458 +#endif
16459 +       kfree(vxi);
16460 +       atomic_dec(&vx_global_ctotal);
16461 +}
16462 +
16463 +static void __shutdown_vx_info(struct vx_info *vxi)
16464 +{
16465 +       struct nsproxy *nsproxy;
16466 +       struct fs_struct *fs;
16467 +       int index, kill;
16468 +
16469 +       might_sleep();
16470 +
16471 +       vxi->vx_state |= VXS_SHUTDOWN;
16472 +       vs_state_change(vxi, VSC_SHUTDOWN);
16473 +
16474 +       for (index = 0; index < VX_SPACES; index++) {
16475 +               nsproxy = xchg(&vxi->vx_nsproxy[index], NULL);
16476 +               if (nsproxy)
16477 +                       put_nsproxy(nsproxy);
16478 +
16479 +               fs = xchg(&vxi->vx_fs[index], NULL);
16480 +               spin_lock(&fs->lock);
16481 +               kill = !--fs->users;
16482 +               spin_unlock(&fs->lock);
16483 +               if (kill)
16484 +                       free_fs_struct(fs);
16485 +       }
16486 +}
16487 +
16488 +/* exported stuff */
16489 +
16490 +void free_vx_info(struct vx_info *vxi)
16491 +{
16492 +       unsigned long flags;
16493 +       unsigned index;
16494 +
16495 +       /* check for reference counts first */
16496 +       BUG_ON(atomic_read(&vxi->vx_usecnt));
16497 +       BUG_ON(atomic_read(&vxi->vx_tasks));
16498 +
16499 +       /* context must not be hashed */
16500 +       BUG_ON(vx_info_state(vxi, VXS_HASHED));
16501 +
16502 +       /* context shutdown is mandatory */
16503 +       BUG_ON(!vx_info_state(vxi, VXS_SHUTDOWN));
16504 +
16505 +       /* nsproxy and fs check */
16506 +       for (index = 0; index < VX_SPACES; index++) {
16507 +               BUG_ON(vxi->vx_nsproxy[index]);
16508 +               BUG_ON(vxi->vx_fs[index]);
16509 +       }
16510 +
16511 +       spin_lock_irqsave(&vx_info_inactive_lock, flags);
16512 +       hlist_del(&vxi->vx_hlist);
16513 +       spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
16514 +
16515 +       __dealloc_vx_info(vxi);
16516 +}
16517 +
16518 +
16519 +/*     hash table for vx_info hash */
16520 +
16521 +#define VX_HASH_SIZE   13
16522 +
16523 +static struct hlist_head vx_info_hash[VX_HASH_SIZE] =
16524 +       { [0 ... VX_HASH_SIZE-1] = HLIST_HEAD_INIT };
16525 +
16526 +static spinlock_t vx_info_hash_lock = SPIN_LOCK_UNLOCKED;
16527 +
16528 +
16529 +static inline unsigned int __hashval(xid_t xid)
16530 +{
16531 +       return (xid % VX_HASH_SIZE);
16532 +}
16533 +
16534 +
16535 +
16536 +/*     __hash_vx_info()
16537 +
16538 +       * add the vxi to the global hash table
16539 +       * requires the hash_lock to be held                     */
16540 +
16541 +static inline void __hash_vx_info(struct vx_info *vxi)
16542 +{
16543 +       struct hlist_head *head;
16544 +
16545 +       vxd_assert_lock(&vx_info_hash_lock);
16546 +       vxdprintk(VXD_CBIT(xid, 4),
16547 +               "__hash_vx_info: %p[#%d]", vxi, vxi->vx_id);
16548 +       vxh_hash_vx_info(vxi);
16549 +
16550 +       /* context must not be hashed */
16551 +       BUG_ON(vx_info_state(vxi, VXS_HASHED));
16552 +
16553 +       vxi->vx_state |= VXS_HASHED;
16554 +       head = &vx_info_hash[__hashval(vxi->vx_id)];
16555 +       hlist_add_head(&vxi->vx_hlist, head);
16556 +       atomic_inc(&vx_global_cactive);
16557 +}
16558 +
16559 +/*     __unhash_vx_info()
16560 +
16561 +       * remove the vxi from the global hash table
16562 +       * requires the hash_lock to be held                     */
16563 +
16564 +static inline void __unhash_vx_info(struct vx_info *vxi)
16565 +{
16566 +       unsigned long flags;
16567 +
16568 +       vxd_assert_lock(&vx_info_hash_lock);
16569 +       vxdprintk(VXD_CBIT(xid, 4),
16570 +               "__unhash_vx_info: %p[#%d.%d.%d]", vxi, vxi->vx_id,
16571 +               atomic_read(&vxi->vx_usecnt), atomic_read(&vxi->vx_tasks));
16572 +       vxh_unhash_vx_info(vxi);
16573 +
16574 +       /* context must be hashed */
16575 +       BUG_ON(!vx_info_state(vxi, VXS_HASHED));
16576 +       /* but without tasks */
16577 +       BUG_ON(atomic_read(&vxi->vx_tasks));
16578 +
16579 +       vxi->vx_state &= ~VXS_HASHED;
16580 +       hlist_del_init(&vxi->vx_hlist);
16581 +       spin_lock_irqsave(&vx_info_inactive_lock, flags);
16582 +       hlist_add_head(&vxi->vx_hlist, &vx_info_inactive);
16583 +       spin_unlock_irqrestore(&vx_info_inactive_lock, flags);
16584 +       atomic_dec(&vx_global_cactive);
16585 +}
16586 +
16587 +
16588 +/*     __lookup_vx_info()
16589 +
16590 +       * requires the hash_lock to be held
16591 +       * doesn't increment the vx_refcnt                       */
16592 +
16593 +static inline struct vx_info *__lookup_vx_info(xid_t xid)
16594 +{
16595 +       struct hlist_head *head = &vx_info_hash[__hashval(xid)];
16596 +       struct hlist_node *pos;
16597 +       struct vx_info *vxi;
16598 +
16599 +       vxd_assert_lock(&vx_info_hash_lock);
16600 +       hlist_for_each(pos, head) {
16601 +               vxi = hlist_entry(pos, struct vx_info, vx_hlist);
16602 +
16603 +               if (vxi->vx_id == xid)
16604 +                       goto found;
16605 +       }
16606 +       vxi = NULL;
16607 +found:
16608 +       vxdprintk(VXD_CBIT(xid, 0),
16609 +               "__lookup_vx_info(#%u): %p[#%u]",
16610 +               xid, vxi, vxi ? vxi->vx_id : 0);
16611 +       vxh_lookup_vx_info(vxi, xid);
16612 +       return vxi;
16613 +}
16614 +
16615 +
16616 +/*     __create_vx_info()
16617 +
16618 +       * create the requested context
16619 +       * get(), claim() and hash it                            */
16620 +
16621 +static struct vx_info *__create_vx_info(int id)
16622 +{
16623 +       struct vx_info *new, *vxi = NULL;
16624 +
16625 +       vxdprintk(VXD_CBIT(xid, 1), "create_vx_info(%d)*", id);
16626 +
16627 +       if (!(new = __alloc_vx_info(id)))
16628 +               return ERR_PTR(-ENOMEM);
16629 +
16630 +       /* required to make dynamic xids unique */
16631 +       spin_lock(&vx_info_hash_lock);
16632 +
16633 +       /* static context requested */
16634 +       if ((vxi = __lookup_vx_info(id))) {
16635 +               vxdprintk(VXD_CBIT(xid, 0),
16636 +                       "create_vx_info(%d) = %p (already there)", id, vxi);
16637 +               if (vx_info_flags(vxi, VXF_STATE_SETUP, 0))
16638 +                       vxi = ERR_PTR(-EBUSY);
16639 +               else
16640 +                       vxi = ERR_PTR(-EEXIST);
16641 +               goto out_unlock;
16642 +       }
16643 +       /* new context */
16644 +       vxdprintk(VXD_CBIT(xid, 0),
16645 +               "create_vx_info(%d) = %p (new)", id, new);
16646 +       claim_vx_info(new, NULL);
16647 +       __hash_vx_info(get_vx_info(new));
16648 +       vxi = new, new = NULL;
16649 +
16650 +out_unlock:
16651 +       spin_unlock(&vx_info_hash_lock);
16652 +       vxh_create_vx_info(IS_ERR(vxi) ? NULL : vxi, id);
16653 +       if (new)
16654 +               __dealloc_vx_info(new);
16655 +       return vxi;
16656 +}
16657 +
16658 +
16659 +/*     exported stuff                                          */
16660 +
16661 +
16662 +void unhash_vx_info(struct vx_info *vxi)
16663 +{
16664 +       __shutdown_vx_info(vxi);
16665 +       spin_lock(&vx_info_hash_lock);
16666 +       __unhash_vx_info(vxi);
16667 +       spin_unlock(&vx_info_hash_lock);
16668 +       __wakeup_vx_info(vxi);
16669 +}
16670 +
16671 +
16672 +/*     lookup_vx_info()
16673 +
16674 +       * search for a vx_info and get() it
16675 +       * negative id means current                             */
16676 +
16677 +struct vx_info *lookup_vx_info(int id)
16678 +{
16679 +       struct vx_info *vxi = NULL;
16680 +
16681 +       if (id < 0) {
16682 +               vxi = get_vx_info(current_vx_info());
16683 +       } else if (id > 1) {
16684 +               spin_lock(&vx_info_hash_lock);
16685 +               vxi = get_vx_info(__lookup_vx_info(id));
16686 +               spin_unlock(&vx_info_hash_lock);
16687 +       }
16688 +       return vxi;
16689 +}
16690 +
16691 +/*     xid_is_hashed()
16692 +
16693 +       * verify that xid is still hashed                       */
16694 +
16695 +int xid_is_hashed(xid_t xid)
16696 +{
16697 +       int hashed;
16698 +
16699 +       spin_lock(&vx_info_hash_lock);
16700 +       hashed = (__lookup_vx_info(xid) != NULL);
16701 +       spin_unlock(&vx_info_hash_lock);
16702 +       return hashed;
16703 +}
16704 +
16705 +#ifdef CONFIG_PROC_FS
16706 +
16707 +/*     get_xid_list()
16708 +
16709 +       * get a subset of hashed xids for proc
16710 +       * assumes size is at least one                          */
16711 +
16712 +int get_xid_list(int index, unsigned int *xids, int size)
16713 +{
16714 +       int hindex, nr_xids = 0;
16715 +
16716 +       /* only show current and children */
16717 +       if (!vx_check(0, VS_ADMIN | VS_WATCH)) {
16718 +               if (index > 0)
16719 +                       return 0;
16720 +               xids[nr_xids] = vx_current_xid();
16721 +               return 1;
16722 +       }
16723 +
16724 +       for (hindex = 0; hindex < VX_HASH_SIZE; hindex++) {
16725 +               struct hlist_head *head = &vx_info_hash[hindex];
16726 +               struct hlist_node *pos;
16727 +
16728 +               spin_lock(&vx_info_hash_lock);
16729 +               hlist_for_each(pos, head) {
16730 +                       struct vx_info *vxi;
16731 +
16732 +                       if (--index > 0)
16733 +                               continue;
16734 +
16735 +                       vxi = hlist_entry(pos, struct vx_info, vx_hlist);
16736 +                       xids[nr_xids] = vxi->vx_id;
16737 +                       if (++nr_xids >= size) {
16738 +                               spin_unlock(&vx_info_hash_lock);
16739 +                               goto out;
16740 +                       }
16741 +               }
16742 +               /* keep the lock time short */
16743 +               spin_unlock(&vx_info_hash_lock);
16744 +       }
16745 +out:
16746 +       return nr_xids;
16747 +}
16748 +#endif
16749 +
16750 +#ifdef CONFIG_VSERVER_DEBUG
16751 +
16752 +void   dump_vx_info_inactive(int level)
16753 +{
16754 +       struct hlist_node *entry, *next;
16755 +
16756 +       hlist_for_each_safe(entry, next, &vx_info_inactive) {
16757 +               struct vx_info *vxi =
16758 +                       list_entry(entry, struct vx_info, vx_hlist);
16759 +
16760 +               dump_vx_info(vxi, level);
16761 +       }
16762 +}
16763 +
16764 +#endif
16765 +
16766 +#if 0
16767 +int vx_migrate_user(struct task_struct *p, struct vx_info *vxi)
16768 +{
16769 +       struct user_struct *new_user, *old_user;
16770 +
16771 +       if (!p || !vxi)
16772 +               BUG();
16773 +
16774 +       if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
16775 +               return -EACCES;
16776 +
16777 +       new_user = alloc_uid(vxi->vx_id, p->uid);
16778 +       if (!new_user)
16779 +               return -ENOMEM;
16780 +
16781 +       old_user = p->user;
16782 +       if (new_user != old_user) {
16783 +               atomic_inc(&new_user->processes);
16784 +               atomic_dec(&old_user->processes);
16785 +               p->user = new_user;
16786 +       }
16787 +       free_uid(old_user);
16788 +       return 0;
16789 +}
16790 +#endif
16791 +
16792 +#if 0
16793 +void vx_mask_cap_bset(struct vx_info *vxi, struct task_struct *p)
16794 +{
16795 +       // p->cap_effective &= vxi->vx_cap_bset;
16796 +       p->cap_effective =
16797 +               cap_intersect(p->cap_effective, vxi->cap_bset);
16798 +       // p->cap_inheritable &= vxi->vx_cap_bset;
16799 +       p->cap_inheritable =
16800 +               cap_intersect(p->cap_inheritable, vxi->cap_bset);
16801 +       // p->cap_permitted &= vxi->vx_cap_bset;
16802 +       p->cap_permitted =
16803 +               cap_intersect(p->cap_permitted, vxi->cap_bset);
16804 +}
16805 +#endif
16806 +
16807 +
16808 +#include <linux/file.h>
16809 +#include <linux/fdtable.h>
16810 +
16811 +static int vx_openfd_task(struct task_struct *tsk)
16812 +{
16813 +       struct files_struct *files = tsk->files;
16814 +       struct fdtable *fdt;
16815 +       const unsigned long *bptr;
16816 +       int count, total;
16817 +
16818 +       /* no rcu_read_lock() because of spin_lock() */
16819 +       spin_lock(&files->file_lock);
16820 +       fdt = files_fdtable(files);
16821 +       bptr = fdt->open_fds->fds_bits;
16822 +       count = fdt->max_fds / (sizeof(unsigned long) * 8);
16823 +       for (total = 0; count > 0; count--) {
16824 +               if (*bptr)
16825 +                       total += hweight_long(*bptr);
16826 +               bptr++;
16827 +       }
16828 +       spin_unlock(&files->file_lock);
16829 +       return total;
16830 +}
16831 +
16832 +
16833 +/*     for *space compatibility */
16834 +
16835 +asmlinkage long sys_unshare(unsigned long);
16836 +
16837 +/*
16838 + *     migrate task to new context
16839 + *     gets vxi, puts old_vxi on change
16840 + *     optionally unshares namespaces (hack)
16841 + */
16842 +
16843 +int vx_migrate_task(struct task_struct *p, struct vx_info *vxi, int unshare)
16844 +{
16845 +       struct vx_info *old_vxi;
16846 +       int ret = 0;
16847 +
16848 +       if (!p || !vxi)
16849 +               BUG();
16850 +
16851 +       vxdprintk(VXD_CBIT(xid, 5),
16852 +               "vx_migrate_task(%p,%p[#%d.%d])", p, vxi,
16853 +               vxi->vx_id, atomic_read(&vxi->vx_usecnt));
16854 +
16855 +       if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0) &&
16856 +               !vx_info_flags(vxi, VXF_STATE_SETUP, 0))
16857 +               return -EACCES;
16858 +
16859 +       if (vx_info_state(vxi, VXS_SHUTDOWN))
16860 +               return -EFAULT;
16861 +
16862 +       old_vxi = task_get_vx_info(p);
16863 +       if (old_vxi == vxi)
16864 +               goto out;
16865 +
16866 +//     if (!(ret = vx_migrate_user(p, vxi))) {
16867 +       {
16868 +               int openfd;
16869 +
16870 +               task_lock(p);
16871 +               openfd = vx_openfd_task(p);
16872 +
16873 +               if (old_vxi) {
16874 +                       atomic_dec(&old_vxi->cvirt.nr_threads);
16875 +                       atomic_dec(&old_vxi->cvirt.nr_running);
16876 +                       __rlim_dec(&old_vxi->limit, RLIMIT_NPROC);
16877 +                       /* FIXME: what about the struct files here? */
16878 +                       __rlim_sub(&old_vxi->limit, VLIMIT_OPENFD, openfd);
16879 +                       /* account for the executable */
16880 +                       __rlim_dec(&old_vxi->limit, VLIMIT_DENTRY);
16881 +               }
16882 +               atomic_inc(&vxi->cvirt.nr_threads);
16883 +               atomic_inc(&vxi->cvirt.nr_running);
16884 +               __rlim_inc(&vxi->limit, RLIMIT_NPROC);
16885 +               /* FIXME: what about the struct files here? */
16886 +               __rlim_add(&vxi->limit, VLIMIT_OPENFD, openfd);
16887 +               /* account for the executable */
16888 +               __rlim_inc(&vxi->limit, VLIMIT_DENTRY);
16889 +
16890 +               if (old_vxi) {
16891 +                       release_vx_info(old_vxi, p);
16892 +                       clr_vx_info(&p->vx_info);
16893 +               }
16894 +               claim_vx_info(vxi, p);
16895 +               set_vx_info(&p->vx_info, vxi);
16896 +               p->xid = vxi->vx_id;
16897 +
16898 +               vxdprintk(VXD_CBIT(xid, 5),
16899 +                       "moved task %p into vxi:%p[#%d]",
16900 +                       p, vxi, vxi->vx_id);
16901 +
16902 +               // vx_mask_cap_bset(vxi, p);
16903 +               task_unlock(p);
16904 +
16905 +               /* hack for *spaces to provide compatibility */
16906 +               if (unshare) {
16907 +                       struct nsproxy *old_nsp, *new_nsp;
16908 +
16909 +                       ret = unshare_nsproxy_namespaces(
16910 +                               CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER,
16911 +                               &new_nsp, NULL);
16912 +                       if (ret)
16913 +                               goto out;
16914 +
16915 +                       old_nsp = xchg(&p->nsproxy, new_nsp);
16916 +                       vx_set_space(vxi,
16917 +                               CLONE_NEWUTS | CLONE_NEWIPC | CLONE_NEWUSER, 0);
16918 +                       put_nsproxy(old_nsp);
16919 +               }
16920 +       }
16921 +out:
16922 +       put_vx_info(old_vxi);
16923 +       return ret;
16924 +}
16925 +
16926 +int vx_set_reaper(struct vx_info *vxi, struct task_struct *p)
16927 +{
16928 +       struct task_struct *old_reaper;
16929 +
16930 +       if (!vxi)
16931 +               return -EINVAL;
16932 +
16933 +       vxdprintk(VXD_CBIT(xid, 6),
16934 +               "vx_set_reaper(%p[#%d],%p[#%d,%d])",
16935 +               vxi, vxi->vx_id, p, p->xid, p->pid);
16936 +
16937 +       old_reaper = vxi->vx_reaper;
16938 +       if (old_reaper == p)
16939 +               return 0;
16940 +
16941 +       /* set new child reaper */
16942 +       get_task_struct(p);
16943 +       vxi->vx_reaper = p;
16944 +       put_task_struct(old_reaper);
16945 +       return 0;
16946 +}
16947 +
16948 +int vx_set_init(struct vx_info *vxi, struct task_struct *p)
16949 +{
16950 +       if (!vxi)
16951 +               return -EINVAL;
16952 +
16953 +       vxdprintk(VXD_CBIT(xid, 6),
16954 +               "vx_set_init(%p[#%d],%p[#%d,%d,%d])",
16955 +               vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
16956 +
16957 +       vxi->vx_flags &= ~VXF_STATE_INIT;
16958 +       // vxi->vx_initpid = p->tgid;
16959 +       vxi->vx_initpid = p->pid;
16960 +       return 0;
16961 +}
16962 +
16963 +void vx_exit_init(struct vx_info *vxi, struct task_struct *p, int code)
16964 +{
16965 +       vxdprintk(VXD_CBIT(xid, 6),
16966 +               "vx_exit_init(%p[#%d],%p[#%d,%d,%d])",
16967 +               vxi, vxi->vx_id, p, p->xid, p->pid, p->tgid);
16968 +
16969 +       vxi->exit_code = code;
16970 +       vxi->vx_initpid = 0;
16971 +}
16972 +
16973 +
16974 +void vx_set_persistent(struct vx_info *vxi)
16975 +{
16976 +       vxdprintk(VXD_CBIT(xid, 6),
16977 +               "vx_set_persistent(%p[#%d])", vxi, vxi->vx_id);
16978 +
16979 +       get_vx_info(vxi);
16980 +       claim_vx_info(vxi, NULL);
16981 +}
16982 +
16983 +void vx_clear_persistent(struct vx_info *vxi)
16984 +{
16985 +       vxdprintk(VXD_CBIT(xid, 6),
16986 +               "vx_clear_persistent(%p[#%d])", vxi, vxi->vx_id);
16987 +
16988 +       release_vx_info(vxi, NULL);
16989 +       put_vx_info(vxi);
16990 +}
16991 +
16992 +void vx_update_persistent(struct vx_info *vxi)
16993 +{
16994 +       if (vx_info_flags(vxi, VXF_PERSISTENT, 0))
16995 +               vx_set_persistent(vxi);
16996 +       else
16997 +               vx_clear_persistent(vxi);
16998 +}
16999 +
17000 +
17001 +/*     task must be current or locked          */
17002 +
17003 +void   exit_vx_info(struct task_struct *p, int code)
17004 +{
17005 +       struct vx_info *vxi = p->vx_info;
17006 +
17007 +       if (vxi) {
17008 +               atomic_dec(&vxi->cvirt.nr_threads);
17009 +               vx_nproc_dec(p);
17010 +
17011 +               vxi->exit_code = code;
17012 +               release_vx_info(vxi, p);
17013 +       }
17014 +}
17015 +
17016 +void   exit_vx_info_early(struct task_struct *p, int code)
17017 +{
17018 +       struct vx_info *vxi = p->vx_info;
17019 +
17020 +       if (vxi) {
17021 +               if (vxi->vx_initpid == p->pid)
17022 +                       vx_exit_init(vxi, p, code);
17023 +               if (vxi->vx_reaper == p)
17024 +                       vx_set_reaper(vxi, init_pid_ns.child_reaper);
17025 +       }
17026 +}
17027 +
17028 +
17029 +/* vserver syscall commands below here */
17030 +
17031 +/* taks xid and vx_info functions */
17032 +
17033 +#include <asm/uaccess.h>
17034 +
17035 +
17036 +int vc_task_xid(uint32_t id)
17037 +{
17038 +       xid_t xid;
17039 +
17040 +       if (id) {
17041 +               struct task_struct *tsk;
17042 +
17043 +               read_lock(&tasklist_lock);
17044 +               tsk = find_task_by_real_pid(id);
17045 +               xid = (tsk) ? tsk->xid : -ESRCH;
17046 +               read_unlock(&tasklist_lock);
17047 +       } else
17048 +               xid = vx_current_xid();
17049 +       return xid;
17050 +}
17051 +
17052 +
17053 +int vc_vx_info(struct vx_info *vxi, void __user *data)
17054 +{
17055 +       struct vcmd_vx_info_v0 vc_data;
17056 +
17057 +       vc_data.xid = vxi->vx_id;
17058 +       vc_data.initpid = vxi->vx_initpid;
17059 +
17060 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17061 +               return -EFAULT;
17062 +       return 0;
17063 +}
17064 +
17065 +
17066 +int vc_ctx_stat(struct vx_info *vxi, void __user *data)
17067 +{
17068 +       struct vcmd_ctx_stat_v0 vc_data;
17069 +
17070 +       vc_data.usecnt = atomic_read(&vxi->vx_usecnt);
17071 +       vc_data.tasks = atomic_read(&vxi->vx_tasks);
17072 +
17073 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17074 +               return -EFAULT;
17075 +       return 0;
17076 +}
17077 +
17078 +
17079 +/* context functions */
17080 +
17081 +int vc_ctx_create(uint32_t xid, void __user *data)
17082 +{
17083 +       struct vcmd_ctx_create vc_data = { .flagword = VXF_INIT_SET };
17084 +       struct vx_info *new_vxi;
17085 +       int ret;
17086 +
17087 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
17088 +               return -EFAULT;
17089 +
17090 +       if ((xid > MAX_S_CONTEXT) || (xid < 2))
17091 +               return -EINVAL;
17092 +
17093 +       new_vxi = __create_vx_info(xid);
17094 +       if (IS_ERR(new_vxi))
17095 +               return PTR_ERR(new_vxi);
17096 +
17097 +       /* initial flags */
17098 +       new_vxi->vx_flags = vc_data.flagword;
17099 +
17100 +       ret = -ENOEXEC;
17101 +       if (vs_state_change(new_vxi, VSC_STARTUP))
17102 +               goto out;
17103 +
17104 +       ret = vx_migrate_task(current, new_vxi, (!data));
17105 +       if (ret)
17106 +               goto out;
17107 +
17108 +       /* return context id on success */
17109 +       ret = new_vxi->vx_id;
17110 +
17111 +       /* get a reference for persistent contexts */
17112 +       if ((vc_data.flagword & VXF_PERSISTENT))
17113 +               vx_set_persistent(new_vxi);
17114 +out:
17115 +       release_vx_info(new_vxi, NULL);
17116 +       put_vx_info(new_vxi);
17117 +       return ret;
17118 +}
17119 +
17120 +
17121 +int vc_ctx_migrate(struct vx_info *vxi, void __user *data)
17122 +{
17123 +       struct vcmd_ctx_migrate vc_data = { .flagword = 0 };
17124 +       int ret;
17125 +
17126 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
17127 +               return -EFAULT;
17128 +
17129 +       ret = vx_migrate_task(current, vxi, 0);
17130 +       if (ret)
17131 +               return ret;
17132 +       if (vc_data.flagword & VXM_SET_INIT)
17133 +               ret = vx_set_init(vxi, current);
17134 +       if (ret)
17135 +               return ret;
17136 +       if (vc_data.flagword & VXM_SET_REAPER)
17137 +               ret = vx_set_reaper(vxi, current);
17138 +       return ret;
17139 +}
17140 +
17141 +
17142 +int vc_get_cflags(struct vx_info *vxi, void __user *data)
17143 +{
17144 +       struct vcmd_ctx_flags_v0 vc_data;
17145 +
17146 +       vc_data.flagword = vxi->vx_flags;
17147 +
17148 +       /* special STATE flag handling */
17149 +       vc_data.mask = vs_mask_flags(~0ULL, vxi->vx_flags, VXF_ONE_TIME);
17150 +
17151 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17152 +               return -EFAULT;
17153 +       return 0;
17154 +}
17155 +
17156 +int vc_set_cflags(struct vx_info *vxi, void __user *data)
17157 +{
17158 +       struct vcmd_ctx_flags_v0 vc_data;
17159 +       uint64_t mask, trigger;
17160 +
17161 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17162 +               return -EFAULT;
17163 +
17164 +       /* special STATE flag handling */
17165 +       mask = vs_mask_mask(vc_data.mask, vxi->vx_flags, VXF_ONE_TIME);
17166 +       trigger = (mask & vxi->vx_flags) ^ (mask & vc_data.flagword);
17167 +
17168 +       if (vxi == current_vx_info()) {
17169 +               /* if (trigger & VXF_STATE_SETUP)
17170 +                       vx_mask_cap_bset(vxi, current); */
17171 +               if (trigger & VXF_STATE_INIT) {
17172 +                       int ret;
17173 +
17174 +                       ret = vx_set_init(vxi, current);
17175 +                       if (ret)
17176 +                               return ret;
17177 +                       ret = vx_set_reaper(vxi, current);
17178 +                       if (ret)
17179 +                               return ret;
17180 +               }
17181 +       }
17182 +
17183 +       vxi->vx_flags = vs_mask_flags(vxi->vx_flags,
17184 +               vc_data.flagword, mask);
17185 +       if (trigger & VXF_PERSISTENT)
17186 +               vx_update_persistent(vxi);
17187 +
17188 +       return 0;
17189 +}
17190 +
17191 +
17192 +static inline uint64_t caps_from_cap_t(kernel_cap_t c)
17193 +{
17194 +       uint64_t v = c.cap[0] | ((uint64_t)c.cap[1] << 32);
17195 +
17196 +       // printk("caps_from_cap_t(%08x:%08x) = %016llx\n", c.cap[1], c.cap[0], v);
17197 +       return v;
17198 +}
17199 +
17200 +static inline kernel_cap_t cap_t_from_caps(uint64_t v)
17201 +{
17202 +       kernel_cap_t c = __cap_empty_set;
17203 +
17204 +       c.cap[0] = v & 0xFFFFFFFF;
17205 +       c.cap[1] = (v >> 32) & 0xFFFFFFFF;
17206 +
17207 +       // printk("cap_t_from_caps(%016llx) = %08x:%08x\n", v, c.cap[1], c.cap[0]);
17208 +       return c;
17209 +}
17210 +
17211 +
17212 +static int do_get_caps(struct vx_info *vxi, uint64_t *bcaps, uint64_t *ccaps)
17213 +{
17214 +       if (bcaps)
17215 +               *bcaps = caps_from_cap_t(vxi->vx_bcaps);
17216 +       if (ccaps)
17217 +               *ccaps = vxi->vx_ccaps;
17218 +
17219 +       return 0;
17220 +}
17221 +
17222 +int vc_get_ccaps(struct vx_info *vxi, void __user *data)
17223 +{
17224 +       struct vcmd_ctx_caps_v1 vc_data;
17225 +       int ret;
17226 +
17227 +       ret = do_get_caps(vxi, NULL, &vc_data.ccaps);
17228 +       if (ret)
17229 +               return ret;
17230 +       vc_data.cmask = ~0ULL;
17231 +
17232 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17233 +               return -EFAULT;
17234 +       return 0;
17235 +}
17236 +
17237 +static int do_set_caps(struct vx_info *vxi,
17238 +       uint64_t bcaps, uint64_t bmask, uint64_t ccaps, uint64_t cmask)
17239 +{
17240 +       uint64_t bcold = caps_from_cap_t(vxi->vx_bcaps);
17241 +
17242 +#if 0
17243 +       printk("do_set_caps(%16llx, %16llx, %16llx, %16llx)\n",
17244 +               bcaps, bmask, ccaps, cmask);
17245 +#endif
17246 +       vxi->vx_bcaps = cap_t_from_caps(
17247 +               vs_mask_flags(bcold, bcaps, bmask));
17248 +       vxi->vx_ccaps = vs_mask_flags(vxi->vx_ccaps, ccaps, cmask);
17249 +
17250 +       return 0;
17251 +}
17252 +
17253 +int vc_set_ccaps(struct vx_info *vxi, void __user *data)
17254 +{
17255 +       struct vcmd_ctx_caps_v1 vc_data;
17256 +
17257 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17258 +               return -EFAULT;
17259 +
17260 +       return do_set_caps(vxi, 0, 0, vc_data.ccaps, vc_data.cmask);
17261 +}
17262 +
17263 +int vc_get_bcaps(struct vx_info *vxi, void __user *data)
17264 +{
17265 +       struct vcmd_bcaps vc_data;
17266 +       int ret;
17267 +
17268 +       ret = do_get_caps(vxi, &vc_data.bcaps, NULL);
17269 +       if (ret)
17270 +               return ret;
17271 +       vc_data.bmask = ~0ULL;
17272 +
17273 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17274 +               return -EFAULT;
17275 +       return 0;
17276 +}
17277 +
17278 +int vc_set_bcaps(struct vx_info *vxi, void __user *data)
17279 +{
17280 +       struct vcmd_bcaps vc_data;
17281 +
17282 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17283 +               return -EFAULT;
17284 +
17285 +       return do_set_caps(vxi, vc_data.bcaps, vc_data.bmask, 0, 0);
17286 +}
17287 +
17288 +
17289 +int vc_get_umask(struct vx_info *vxi, void __user *data)
17290 +{
17291 +       struct vcmd_umask vc_data;
17292 +
17293 +       vc_data.umask = vxi->vx_umask;
17294 +       vc_data.mask = ~0ULL;
17295 +
17296 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17297 +               return -EFAULT;
17298 +       return 0;
17299 +}
17300 +
17301 +int vc_set_umask(struct vx_info *vxi, void __user *data)
17302 +{
17303 +       struct vcmd_umask vc_data;
17304 +
17305 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17306 +               return -EFAULT;
17307 +
17308 +       vxi->vx_umask = vs_mask_flags(vxi->vx_umask,
17309 +               vc_data.umask, vc_data.mask);
17310 +       return 0;
17311 +}
17312 +
17313 +
17314 +int vc_get_badness(struct vx_info *vxi, void __user *data)
17315 +{
17316 +       struct vcmd_badness_v0 vc_data;
17317 +
17318 +       vc_data.bias = vxi->vx_badness_bias;
17319 +
17320 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17321 +               return -EFAULT;
17322 +       return 0;
17323 +}
17324 +
17325 +int vc_set_badness(struct vx_info *vxi, void __user *data)
17326 +{
17327 +       struct vcmd_badness_v0 vc_data;
17328 +
17329 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17330 +               return -EFAULT;
17331 +
17332 +       vxi->vx_badness_bias = vc_data.bias;
17333 +       return 0;
17334 +}
17335 +
17336 +#include <linux/module.h>
17337 +
17338 +EXPORT_SYMBOL_GPL(free_vx_info);
17339 +
17340 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cvirt.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt.c
17341 --- linux-2.6.36/kernel/vserver/cvirt.c 1970-01-01 01:00:00.000000000 +0100
17342 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt.c   2010-10-24 12:39:42.000000000 +0200
17343 @@ -0,0 +1,303 @@
17344 +/*
17345 + *  linux/kernel/vserver/cvirt.c
17346 + *
17347 + *  Virtual Server: Context Virtualization
17348 + *
17349 + *  Copyright (C) 2004-2007  Herbert Pötzl
17350 + *
17351 + *  V0.01  broken out from limit.c
17352 + *  V0.02  added utsname stuff
17353 + *  V0.03  changed vcmds to vxi arg
17354 + *
17355 + */
17356 +
17357 +#include <linux/types.h>
17358 +#include <linux/utsname.h>
17359 +#include <linux/vs_cvirt.h>
17360 +#include <linux/vserver/switch.h>
17361 +#include <linux/vserver/cvirt_cmd.h>
17362 +
17363 +#include <asm/uaccess.h>
17364 +
17365 +
17366 +void vx_vsi_uptime(struct timespec *uptime, struct timespec *idle)
17367 +{
17368 +       struct vx_info *vxi = current_vx_info();
17369 +
17370 +       set_normalized_timespec(uptime,
17371 +               uptime->tv_sec - vxi->cvirt.bias_uptime.tv_sec,
17372 +               uptime->tv_nsec - vxi->cvirt.bias_uptime.tv_nsec);
17373 +       if (!idle)
17374 +               return;
17375 +       set_normalized_timespec(idle,
17376 +               idle->tv_sec - vxi->cvirt.bias_idle.tv_sec,
17377 +               idle->tv_nsec - vxi->cvirt.bias_idle.tv_nsec);
17378 +       return;
17379 +}
17380 +
17381 +uint64_t vx_idle_jiffies(void)
17382 +{
17383 +       return init_task.utime + init_task.stime;
17384 +}
17385 +
17386 +
17387 +
17388 +static inline uint32_t __update_loadavg(uint32_t load,
17389 +       int wsize, int delta, int n)
17390 +{
17391 +       unsigned long long calc, prev;
17392 +
17393 +       /* just set it to n */
17394 +       if (unlikely(delta >= wsize))
17395 +               return (n << FSHIFT);
17396 +
17397 +       calc = delta * n;
17398 +       calc <<= FSHIFT;
17399 +       prev = (wsize - delta);
17400 +       prev *= load;
17401 +       calc += prev;
17402 +       do_div(calc, wsize);
17403 +       return calc;
17404 +}
17405 +
17406 +
17407 +void vx_update_load(struct vx_info *vxi)
17408 +{
17409 +       uint32_t now, last, delta;
17410 +       unsigned int nr_running, nr_uninterruptible;
17411 +       unsigned int total;
17412 +       unsigned long flags;
17413 +
17414 +       spin_lock_irqsave(&vxi->cvirt.load_lock, flags);
17415 +
17416 +       now = jiffies;
17417 +       last = vxi->cvirt.load_last;
17418 +       delta = now - last;
17419 +
17420 +       if (delta < 5*HZ)
17421 +               goto out;
17422 +
17423 +       nr_running = atomic_read(&vxi->cvirt.nr_running);
17424 +       nr_uninterruptible = atomic_read(&vxi->cvirt.nr_uninterruptible);
17425 +       total = nr_running + nr_uninterruptible;
17426 +
17427 +       vxi->cvirt.load[0] = __update_loadavg(vxi->cvirt.load[0],
17428 +               60*HZ, delta, total);
17429 +       vxi->cvirt.load[1] = __update_loadavg(vxi->cvirt.load[1],
17430 +               5*60*HZ, delta, total);
17431 +       vxi->cvirt.load[2] = __update_loadavg(vxi->cvirt.load[2],
17432 +               15*60*HZ, delta, total);
17433 +
17434 +       vxi->cvirt.load_last = now;
17435 +out:
17436 +       atomic_inc(&vxi->cvirt.load_updates);
17437 +       spin_unlock_irqrestore(&vxi->cvirt.load_lock, flags);
17438 +}
17439 +
17440 +
17441 +/*
17442 + * Commands to do_syslog:
17443 + *
17444 + *      0 -- Close the log.  Currently a NOP.
17445 + *      1 -- Open the log. Currently a NOP.
17446 + *      2 -- Read from the log.
17447 + *      3 -- Read all messages remaining in the ring buffer.
17448 + *      4 -- Read and clear all messages remaining in the ring buffer
17449 + *      5 -- Clear ring buffer.
17450 + *      6 -- Disable printk's to console
17451 + *      7 -- Enable printk's to console
17452 + *      8 -- Set level of messages printed to console
17453 + *      9 -- Return number of unread characters in the log buffer
17454 + *     10 -- Return size of the log buffer
17455 + */
17456 +int vx_do_syslog(int type, char __user *buf, int len)
17457 +{
17458 +       int error = 0;
17459 +       int do_clear = 0;
17460 +       struct vx_info *vxi = current_vx_info();
17461 +       struct _vx_syslog *log;
17462 +
17463 +       if (!vxi)
17464 +               return -EINVAL;
17465 +       log = &vxi->cvirt.syslog;
17466 +
17467 +       switch (type) {
17468 +       case 0:         /* Close log */
17469 +       case 1:         /* Open log */
17470 +               break;
17471 +       case 2:         /* Read from log */
17472 +               error = wait_event_interruptible(log->log_wait,
17473 +                       (log->log_start - log->log_end));
17474 +               if (error)
17475 +                       break;
17476 +               spin_lock_irq(&log->logbuf_lock);
17477 +               spin_unlock_irq(&log->logbuf_lock);
17478 +               break;
17479 +       case 4:         /* Read/clear last kernel messages */
17480 +               do_clear = 1;
17481 +               /* fall through */
17482 +       case 3:         /* Read last kernel messages */
17483 +               return 0;
17484 +
17485 +       case 5:         /* Clear ring buffer */
17486 +               return 0;
17487 +
17488 +       case 6:         /* Disable logging to console */
17489 +       case 7:         /* Enable logging to console */
17490 +       case 8:         /* Set level of messages printed to console */
17491 +               break;
17492 +
17493 +       case 9:         /* Number of chars in the log buffer */
17494 +               return 0;
17495 +       case 10:        /* Size of the log buffer */
17496 +               return 0;
17497 +       default:
17498 +               error = -EINVAL;
17499 +               break;
17500 +       }
17501 +       return error;
17502 +}
17503 +
17504 +
17505 +/* virtual host info names */
17506 +
17507 +static char *vx_vhi_name(struct vx_info *vxi, int id)
17508 +{
17509 +       struct nsproxy *nsproxy;
17510 +       struct uts_namespace *uts;
17511 +
17512 +       if (id == VHIN_CONTEXT)
17513 +               return vxi->vx_name;
17514 +
17515 +       nsproxy = vxi->vx_nsproxy[0];
17516 +       if (!nsproxy)
17517 +               return NULL;
17518 +
17519 +       uts = nsproxy->uts_ns;
17520 +       if (!uts)
17521 +               return NULL;
17522 +
17523 +       switch (id) {
17524 +       case VHIN_SYSNAME:
17525 +               return uts->name.sysname;
17526 +       case VHIN_NODENAME:
17527 +               return uts->name.nodename;
17528 +       case VHIN_RELEASE:
17529 +               return uts->name.release;
17530 +       case VHIN_VERSION:
17531 +               return uts->name.version;
17532 +       case VHIN_MACHINE:
17533 +               return uts->name.machine;
17534 +       case VHIN_DOMAINNAME:
17535 +               return uts->name.domainname;
17536 +       default:
17537 +               return NULL;
17538 +       }
17539 +       return NULL;
17540 +}
17541 +
17542 +int vc_set_vhi_name(struct vx_info *vxi, void __user *data)
17543 +{
17544 +       struct vcmd_vhi_name_v0 vc_data;
17545 +       char *name;
17546 +
17547 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17548 +               return -EFAULT;
17549 +
17550 +       name = vx_vhi_name(vxi, vc_data.field);
17551 +       if (!name)
17552 +               return -EINVAL;
17553 +
17554 +       memcpy(name, vc_data.name, 65);
17555 +       return 0;
17556 +}
17557 +
17558 +int vc_get_vhi_name(struct vx_info *vxi, void __user *data)
17559 +{
17560 +       struct vcmd_vhi_name_v0 vc_data;
17561 +       char *name;
17562 +
17563 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
17564 +               return -EFAULT;
17565 +
17566 +       name = vx_vhi_name(vxi, vc_data.field);
17567 +       if (!name)
17568 +               return -EINVAL;
17569 +
17570 +       memcpy(vc_data.name, name, 65);
17571 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17572 +               return -EFAULT;
17573 +       return 0;
17574 +}
17575 +
17576 +
17577 +int vc_virt_stat(struct vx_info *vxi, void __user *data)
17578 +{
17579 +       struct vcmd_virt_stat_v0 vc_data;
17580 +       struct _vx_cvirt *cvirt = &vxi->cvirt;
17581 +       struct timespec uptime;
17582 +
17583 +       do_posix_clock_monotonic_gettime(&uptime);
17584 +       set_normalized_timespec(&uptime,
17585 +               uptime.tv_sec - cvirt->bias_uptime.tv_sec,
17586 +               uptime.tv_nsec - cvirt->bias_uptime.tv_nsec);
17587 +
17588 +       vc_data.offset = timespec_to_ns(&cvirt->bias_ts);
17589 +       vc_data.uptime = timespec_to_ns(&uptime);
17590 +       vc_data.nr_threads = atomic_read(&cvirt->nr_threads);
17591 +       vc_data.nr_running = atomic_read(&cvirt->nr_running);
17592 +       vc_data.nr_uninterruptible = atomic_read(&cvirt->nr_uninterruptible);
17593 +       vc_data.nr_onhold = atomic_read(&cvirt->nr_onhold);
17594 +       vc_data.nr_forks = atomic_read(&cvirt->total_forks);
17595 +       vc_data.load[0] = cvirt->load[0];
17596 +       vc_data.load[1] = cvirt->load[1];
17597 +       vc_data.load[2] = cvirt->load[2];
17598 +
17599 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
17600 +               return -EFAULT;
17601 +       return 0;
17602 +}
17603 +
17604 +
17605 +#ifdef CONFIG_VSERVER_VTIME
17606 +
17607 +/* virtualized time base */
17608 +
17609 +void vx_adjust_timespec(struct timespec *ts)
17610 +{
17611 +       struct vx_info *vxi;
17612 +
17613 +       if (!vx_flags(VXF_VIRT_TIME, 0))
17614 +               return;
17615 +
17616 +       vxi = current_vx_info();
17617 +       ts->tv_sec += vxi->cvirt.bias_ts.tv_sec;
17618 +       ts->tv_nsec += vxi->cvirt.bias_ts.tv_nsec;
17619 +
17620 +       if (ts->tv_nsec >= NSEC_PER_SEC) {
17621 +               ts->tv_sec++;
17622 +               ts->tv_nsec -= NSEC_PER_SEC;
17623 +       } else if (ts->tv_nsec < 0) {
17624 +               ts->tv_sec--;
17625 +               ts->tv_nsec += NSEC_PER_SEC;
17626 +       }
17627 +}
17628 +
17629 +int vx_settimeofday(struct timespec *ts)
17630 +{
17631 +       struct timespec ats, delta;
17632 +       struct vx_info *vxi;
17633 +
17634 +       if (!vx_flags(VXF_VIRT_TIME, 0))
17635 +               return do_settimeofday(ts);
17636 +
17637 +       getnstimeofday(&ats);
17638 +       delta = timespec_sub(*ts, ats);
17639 +
17640 +       vxi = current_vx_info();
17641 +       vxi->cvirt.bias_ts = timespec_add(vxi->cvirt.bias_ts, delta);
17642 +       return 0;
17643 +}
17644 +
17645 +#endif
17646 +
17647 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cvirt_init.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt_init.h
17648 --- linux-2.6.36/kernel/vserver/cvirt_init.h    1970-01-01 01:00:00.000000000 +0100
17649 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt_init.h      2010-10-21 14:40:17.000000000 +0200
17650 @@ -0,0 +1,69 @@
17651 +
17652 +
17653 +extern uint64_t vx_idle_jiffies(void);
17654 +
17655 +static inline void vx_info_init_cvirt(struct _vx_cvirt *cvirt)
17656 +{
17657 +       uint64_t idle_jiffies = vx_idle_jiffies();
17658 +       uint64_t nsuptime;
17659 +
17660 +       do_posix_clock_monotonic_gettime(&cvirt->bias_uptime);
17661 +       nsuptime = (unsigned long long)cvirt->bias_uptime.tv_sec
17662 +               * NSEC_PER_SEC + cvirt->bias_uptime.tv_nsec;
17663 +       cvirt->bias_clock = nsec_to_clock_t(nsuptime);
17664 +       cvirt->bias_ts.tv_sec = 0;
17665 +       cvirt->bias_ts.tv_nsec = 0;
17666 +
17667 +       jiffies_to_timespec(idle_jiffies, &cvirt->bias_idle);
17668 +       atomic_set(&cvirt->nr_threads, 0);
17669 +       atomic_set(&cvirt->nr_running, 0);
17670 +       atomic_set(&cvirt->nr_uninterruptible, 0);
17671 +       atomic_set(&cvirt->nr_onhold, 0);
17672 +
17673 +       spin_lock_init(&cvirt->load_lock);
17674 +       cvirt->load_last = jiffies;
17675 +       atomic_set(&cvirt->load_updates, 0);
17676 +       cvirt->load[0] = 0;
17677 +       cvirt->load[1] = 0;
17678 +       cvirt->load[2] = 0;
17679 +       atomic_set(&cvirt->total_forks, 0);
17680 +
17681 +       spin_lock_init(&cvirt->syslog.logbuf_lock);
17682 +       init_waitqueue_head(&cvirt->syslog.log_wait);
17683 +       cvirt->syslog.log_start = 0;
17684 +       cvirt->syslog.log_end = 0;
17685 +       cvirt->syslog.con_start = 0;
17686 +       cvirt->syslog.logged_chars = 0;
17687 +}
17688 +
17689 +static inline
17690 +void vx_info_init_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
17691 +{
17692 +       // cvirt_pc->cpustat = { 0 };
17693 +}
17694 +
17695 +static inline void vx_info_exit_cvirt(struct _vx_cvirt *cvirt)
17696 +{
17697 +       int value;
17698 +
17699 +       vxwprintk_xid((value = atomic_read(&cvirt->nr_threads)),
17700 +               "!!! cvirt: %p[nr_threads] = %d on exit.",
17701 +               cvirt, value);
17702 +       vxwprintk_xid((value = atomic_read(&cvirt->nr_running)),
17703 +               "!!! cvirt: %p[nr_running] = %d on exit.",
17704 +               cvirt, value);
17705 +       vxwprintk_xid((value = atomic_read(&cvirt->nr_uninterruptible)),
17706 +               "!!! cvirt: %p[nr_uninterruptible] = %d on exit.",
17707 +               cvirt, value);
17708 +       vxwprintk_xid((value = atomic_read(&cvirt->nr_onhold)),
17709 +               "!!! cvirt: %p[nr_onhold] = %d on exit.",
17710 +               cvirt, value);
17711 +       return;
17712 +}
17713 +
17714 +static inline
17715 +void vx_info_exit_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc, int cpu)
17716 +{
17717 +       return;
17718 +}
17719 +
17720 diff -NurpP --minimal linux-2.6.36/kernel/vserver/cvirt_proc.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt_proc.h
17721 --- linux-2.6.36/kernel/vserver/cvirt_proc.h    1970-01-01 01:00:00.000000000 +0100
17722 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/cvirt_proc.h      2010-10-21 13:09:36.000000000 +0200
17723 @@ -0,0 +1,135 @@
17724 +#ifndef _VX_CVIRT_PROC_H
17725 +#define _VX_CVIRT_PROC_H
17726 +
17727 +#include <linux/nsproxy.h>
17728 +#include <linux/mnt_namespace.h>
17729 +#include <linux/ipc_namespace.h>
17730 +#include <linux/utsname.h>
17731 +#include <linux/ipc.h>
17732 +
17733 +
17734 +static inline
17735 +int vx_info_proc_nsproxy(struct nsproxy *nsproxy, char *buffer)
17736 +{
17737 +       struct mnt_namespace *ns;
17738 +       struct uts_namespace *uts;
17739 +       struct ipc_namespace *ipc;
17740 +       struct path path;
17741 +       char *pstr, *root;
17742 +       int length = 0;
17743 +
17744 +       if (!nsproxy)
17745 +               goto out;
17746 +
17747 +       length += sprintf(buffer + length,
17748 +               "NSProxy:\t%p [%p,%p,%p]\n",
17749 +               nsproxy, nsproxy->mnt_ns,
17750 +               nsproxy->uts_ns, nsproxy->ipc_ns);
17751 +
17752 +       ns = nsproxy->mnt_ns;
17753 +       if (!ns)
17754 +               goto skip_ns;
17755 +
17756 +       pstr = kmalloc(PATH_MAX, GFP_KERNEL);
17757 +       if (!pstr)
17758 +               goto skip_ns;
17759 +
17760 +       path.mnt = ns->root;
17761 +       path.dentry = ns->root->mnt_root;
17762 +       root = d_path(&path, pstr, PATH_MAX - 2);
17763 +       length += sprintf(buffer + length,
17764 +               "Namespace:\t%p [#%u]\n"
17765 +               "RootPath:\t%s\n",
17766 +               ns, atomic_read(&ns->count),
17767 +               root);
17768 +       kfree(pstr);
17769 +skip_ns:
17770 +
17771 +       uts = nsproxy->uts_ns;
17772 +       if (!uts)
17773 +               goto skip_uts;
17774 +
17775 +       length += sprintf(buffer + length,
17776 +               "SysName:\t%.*s\n"
17777 +               "NodeName:\t%.*s\n"
17778 +               "Release:\t%.*s\n"
17779 +               "Version:\t%.*s\n"
17780 +               "Machine:\t%.*s\n"
17781 +               "DomainName:\t%.*s\n",
17782 +               __NEW_UTS_LEN, uts->name.sysname,
17783 +               __NEW_UTS_LEN, uts->name.nodename,
17784 +               __NEW_UTS_LEN, uts->name.release,
17785 +               __NEW_UTS_LEN, uts->name.version,
17786 +               __NEW_UTS_LEN, uts->name.machine,
17787 +               __NEW_UTS_LEN, uts->name.domainname);
17788 +skip_uts:
17789 +
17790 +       ipc = nsproxy->ipc_ns;
17791 +       if (!ipc)
17792 +               goto skip_ipc;
17793 +
17794 +       length += sprintf(buffer + length,
17795 +               "SEMS:\t\t%d %d %d %d  %d\n"
17796 +               "MSG:\t\t%d %d %d\n"
17797 +               "SHM:\t\t%lu %lu  %d %d\n",
17798 +               ipc->sem_ctls[0], ipc->sem_ctls[1],
17799 +               ipc->sem_ctls[2], ipc->sem_ctls[3],
17800 +               ipc->used_sems,
17801 +               ipc->msg_ctlmax, ipc->msg_ctlmnb, ipc->msg_ctlmni,
17802 +               (unsigned long)ipc->shm_ctlmax,
17803 +               (unsigned long)ipc->shm_ctlall,
17804 +               ipc->shm_ctlmni, ipc->shm_tot);
17805 +skip_ipc:
17806 +out:
17807 +       return length;
17808 +}
17809 +
17810 +
17811 +#include <linux/sched.h>
17812 +
17813 +#define LOAD_INT(x) ((x) >> FSHIFT)
17814 +#define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1 - 1)) * 100)
17815 +
17816 +static inline
17817 +int vx_info_proc_cvirt(struct _vx_cvirt *cvirt, char *buffer)
17818 +{
17819 +       int length = 0;
17820 +       int a, b, c;
17821 +
17822 +       length += sprintf(buffer + length,
17823 +               "BiasUptime:\t%lu.%02lu\n",
17824 +               (unsigned long)cvirt->bias_uptime.tv_sec,
17825 +               (cvirt->bias_uptime.tv_nsec / (NSEC_PER_SEC / 100)));
17826 +
17827 +       a = cvirt->load[0] + (FIXED_1 / 200);
17828 +       b = cvirt->load[1] + (FIXED_1 / 200);
17829 +       c = cvirt->load[2] + (FIXED_1 / 200);
17830 +       length += sprintf(buffer + length,
17831 +               "nr_threads:\t%d\n"
17832 +               "nr_running:\t%d\n"
17833 +               "nr_unintr:\t%d\n"
17834 +               "nr_onhold:\t%d\n"
17835 +               "load_updates:\t%d\n"
17836 +               "loadavg:\t%d.%02d %d.%02d %d.%02d\n"
17837 +               "total_forks:\t%d\n",
17838 +               atomic_read(&cvirt->nr_threads),
17839 +               atomic_read(&cvirt->nr_running),
17840 +               atomic_read(&cvirt->nr_uninterruptible),
17841 +               atomic_read(&cvirt->nr_onhold),
17842 +               atomic_read(&cvirt->load_updates),
17843 +               LOAD_INT(a), LOAD_FRAC(a),
17844 +               LOAD_INT(b), LOAD_FRAC(b),
17845 +               LOAD_INT(c), LOAD_FRAC(c),
17846 +               atomic_read(&cvirt->total_forks));
17847 +       return length;
17848 +}
17849 +
17850 +static inline
17851 +int vx_info_proc_cvirt_pc(struct _vx_cvirt_pc *cvirt_pc,
17852 +       char *buffer, int cpu)
17853 +{
17854 +       int length = 0;
17855 +       return length;
17856 +}
17857 +
17858 +#endif /* _VX_CVIRT_PROC_H */
17859 diff -NurpP --minimal linux-2.6.36/kernel/vserver/debug.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/debug.c
17860 --- linux-2.6.36/kernel/vserver/debug.c 1970-01-01 01:00:00.000000000 +0100
17861 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/debug.c   2010-10-21 13:09:36.000000000 +0200
17862 @@ -0,0 +1,32 @@
17863 +/*
17864 + *  kernel/vserver/debug.c
17865 + *
17866 + *  Copyright (C) 2005-2007 Herbert Pötzl
17867 + *
17868 + *  V0.01  vx_info dump support
17869 + *
17870 + */
17871 +
17872 +#include <linux/module.h>
17873 +
17874 +#include <linux/vserver/context.h>
17875 +
17876 +
17877 +void   dump_vx_info(struct vx_info *vxi, int level)
17878 +{
17879 +       printk("vx_info %p[#%d, %d.%d, %4x]\n", vxi, vxi->vx_id,
17880 +               atomic_read(&vxi->vx_usecnt),
17881 +               atomic_read(&vxi->vx_tasks),
17882 +               vxi->vx_state);
17883 +       if (level > 0) {
17884 +               __dump_vx_limit(&vxi->limit);
17885 +               __dump_vx_sched(&vxi->sched);
17886 +               __dump_vx_cvirt(&vxi->cvirt);
17887 +               __dump_vx_cacct(&vxi->cacct);
17888 +       }
17889 +       printk("---\n");
17890 +}
17891 +
17892 +
17893 +EXPORT_SYMBOL_GPL(dump_vx_info);
17894 +
17895 diff -NurpP --minimal linux-2.6.36/kernel/vserver/device.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/device.c
17896 --- linux-2.6.36/kernel/vserver/device.c        1970-01-01 01:00:00.000000000 +0100
17897 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/device.c  2010-10-21 13:09:36.000000000 +0200
17898 @@ -0,0 +1,443 @@
17899 +/*
17900 + *  linux/kernel/vserver/device.c
17901 + *
17902 + *  Linux-VServer: Device Support
17903 + *
17904 + *  Copyright (C) 2006  Herbert Pötzl
17905 + *  Copyright (C) 2007  Daniel Hokka Zakrisson
17906 + *
17907 + *  V0.01  device mapping basics
17908 + *  V0.02  added defaults
17909 + *
17910 + */
17911 +
17912 +#include <linux/slab.h>
17913 +#include <linux/rcupdate.h>
17914 +#include <linux/fs.h>
17915 +#include <linux/namei.h>
17916 +#include <linux/hash.h>
17917 +
17918 +#include <asm/errno.h>
17919 +#include <asm/uaccess.h>
17920 +#include <linux/vserver/base.h>
17921 +#include <linux/vserver/debug.h>
17922 +#include <linux/vserver/context.h>
17923 +#include <linux/vserver/device.h>
17924 +#include <linux/vserver/device_cmd.h>
17925 +
17926 +
17927 +#define DMAP_HASH_BITS 4
17928 +
17929 +
17930 +struct vs_mapping {
17931 +       union {
17932 +               struct hlist_node hlist;
17933 +               struct list_head list;
17934 +       } u;
17935 +#define dm_hlist       u.hlist
17936 +#define dm_list                u.list
17937 +       xid_t xid;
17938 +       dev_t device;
17939 +       struct vx_dmap_target target;
17940 +};
17941 +
17942 +
17943 +static struct hlist_head dmap_main_hash[1 << DMAP_HASH_BITS];
17944 +
17945 +static spinlock_t dmap_main_hash_lock = SPIN_LOCK_UNLOCKED;
17946 +
17947 +static struct vx_dmap_target dmap_defaults[2] = {
17948 +       { .flags = DATTR_OPEN },
17949 +       { .flags = DATTR_OPEN },
17950 +};
17951 +
17952 +
17953 +struct kmem_cache *dmap_cachep __read_mostly;
17954 +
17955 +int __init dmap_cache_init(void)
17956 +{
17957 +       dmap_cachep = kmem_cache_create("dmap_cache",
17958 +               sizeof(struct vs_mapping), 0,
17959 +               SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
17960 +       return 0;
17961 +}
17962 +
17963 +__initcall(dmap_cache_init);
17964 +
17965 +
17966 +static inline unsigned int __hashval(dev_t dev, int bits)
17967 +{
17968 +       return hash_long((unsigned long)dev, bits);
17969 +}
17970 +
17971 +
17972 +/*     __hash_mapping()
17973 + *     add the mapping to the hash table
17974 + */
17975 +static inline void __hash_mapping(struct vx_info *vxi, struct vs_mapping *vdm)
17976 +{
17977 +       spinlock_t *hash_lock = &dmap_main_hash_lock;
17978 +       struct hlist_head *head, *hash = dmap_main_hash;
17979 +       int device = vdm->device;
17980 +
17981 +       spin_lock(hash_lock);
17982 +       vxdprintk(VXD_CBIT(misc, 8), "__hash_mapping: %p[#%d] %08x:%08x",
17983 +               vxi, vxi ? vxi->vx_id : 0, device, vdm->target.target);
17984 +
17985 +       head = &hash[__hashval(device, DMAP_HASH_BITS)];
17986 +       hlist_add_head(&vdm->dm_hlist, head);
17987 +       spin_unlock(hash_lock);
17988 +}
17989 +
17990 +
17991 +static inline int __mode_to_default(umode_t mode)
17992 +{
17993 +       switch (mode) {
17994 +       case S_IFBLK:
17995 +               return 0;
17996 +       case S_IFCHR:
17997 +               return 1;
17998 +       default:
17999 +               BUG();
18000 +       }
18001 +}
18002 +
18003 +
18004 +/*     __set_default()
18005 + *     set a default
18006 + */
18007 +static inline void __set_default(struct vx_info *vxi, umode_t mode,
18008 +       struct vx_dmap_target *vdmt)
18009 +{
18010 +       spinlock_t *hash_lock = &dmap_main_hash_lock;
18011 +       spin_lock(hash_lock);
18012 +
18013 +       if (vxi)
18014 +               vxi->dmap.targets[__mode_to_default(mode)] = *vdmt;
18015 +       else
18016 +               dmap_defaults[__mode_to_default(mode)] = *vdmt;
18017 +
18018 +
18019 +       spin_unlock(hash_lock);
18020 +
18021 +       vxdprintk(VXD_CBIT(misc, 8), "__set_default: %p[#%u] %08x %04x",
18022 +                 vxi, vxi ? vxi->vx_id : 0, vdmt->target, vdmt->flags);
18023 +}
18024 +
18025 +
18026 +/*     __remove_default()
18027 + *     remove a default
18028 + */
18029 +static inline int __remove_default(struct vx_info *vxi, umode_t mode)
18030 +{
18031 +       spinlock_t *hash_lock = &dmap_main_hash_lock;
18032 +       spin_lock(hash_lock);
18033 +
18034 +       if (vxi)
18035 +               vxi->dmap.targets[__mode_to_default(mode)].flags = 0;
18036 +       else    /* remove == reset */
18037 +               dmap_defaults[__mode_to_default(mode)].flags = DATTR_OPEN | mode;
18038 +
18039 +       spin_unlock(hash_lock);
18040 +       return 0;
18041 +}
18042 +
18043 +
18044 +/*     __find_mapping()
18045 + *     find a mapping in the hash table
18046 + *
18047 + *     caller must hold hash_lock
18048 + */
18049 +static inline int __find_mapping(xid_t xid, dev_t device, umode_t mode,
18050 +       struct vs_mapping **local, struct vs_mapping **global)
18051 +{
18052 +       struct hlist_head *hash = dmap_main_hash;
18053 +       struct hlist_head *head = &hash[__hashval(device, DMAP_HASH_BITS)];
18054 +       struct hlist_node *pos;
18055 +       struct vs_mapping *vdm;
18056 +
18057 +       *local = NULL;
18058 +       if (global)
18059 +               *global = NULL;
18060 +
18061 +       hlist_for_each(pos, head) {
18062 +               vdm = hlist_entry(pos, struct vs_mapping, dm_hlist);
18063 +
18064 +               if ((vdm->device == device) &&
18065 +                       !((vdm->target.flags ^ mode) & S_IFMT)) {
18066 +                       if (vdm->xid == xid) {
18067 +                               *local = vdm;
18068 +                               return 1;
18069 +                       } else if (global && vdm->xid == 0)
18070 +                               *global = vdm;
18071 +               }
18072 +       }
18073 +
18074 +       if (global && *global)
18075 +               return 0;
18076 +       else
18077 +               return -ENOENT;
18078 +}
18079 +
18080 +
18081 +/*     __lookup_mapping()
18082 + *     find a mapping and store the result in target and flags
18083 + */
18084 +static inline int __lookup_mapping(struct vx_info *vxi,
18085 +       dev_t device, dev_t *target, int *flags, umode_t mode)
18086 +{
18087 +       spinlock_t *hash_lock = &dmap_main_hash_lock;
18088 +       struct vs_mapping *vdm, *global;
18089 +       struct vx_dmap_target *vdmt;
18090 +       int ret = 0;
18091 +       xid_t xid = vxi->vx_id;
18092 +       int index;
18093 +
18094 +       spin_lock(hash_lock);
18095 +       if (__find_mapping(xid, device, mode, &vdm, &global) > 0) {
18096 +               ret = 1;
18097 +               vdmt = &vdm->target;
18098 +               goto found;
18099 +       }
18100 +
18101 +       index = __mode_to_default(mode);
18102 +       if (vxi && vxi->dmap.targets[index].flags) {
18103 +               ret = 2;
18104 +               vdmt = &vxi->dmap.targets[index];
18105 +       } else if (global) {
18106 +               ret = 3;
18107 +               vdmt = &global->target;
18108 +               goto found;
18109 +       } else {
18110 +               ret = 4;
18111 +               vdmt = &dmap_defaults[index];
18112 +       }
18113 +
18114 +found:
18115 +       if (target && (vdmt->flags & DATTR_REMAP))
18116 +               *target = vdmt->target;
18117 +       else if (target)
18118 +               *target = device;
18119 +       if (flags)
18120 +               *flags = vdmt->flags;
18121 +
18122 +       spin_unlock(hash_lock);
18123 +
18124 +       return ret;
18125 +}
18126 +
18127 +
18128 +/*     __remove_mapping()
18129 + *     remove a mapping from the hash table
18130 + */
18131 +static inline int __remove_mapping(struct vx_info *vxi, dev_t device,
18132 +       umode_t mode)
18133 +{
18134 +       spinlock_t *hash_lock = &dmap_main_hash_lock;
18135 +       struct vs_mapping *vdm = NULL;
18136 +       int ret = 0;
18137 +
18138 +       spin_lock(hash_lock);
18139 +
18140 +       ret = __find_mapping((vxi ? vxi->vx_id : 0), device, mode, &vdm,
18141 +               NULL);
18142 +       vxdprintk(VXD_CBIT(misc, 8), "__remove_mapping: %p[#%d] %08x %04x",
18143 +               vxi, vxi ? vxi->vx_id : 0, device, mode);
18144 +       if (ret < 0)
18145 +               goto out;
18146 +       hlist_del(&vdm->dm_hlist);
18147 +
18148 +out:
18149 +       spin_unlock(hash_lock);
18150 +       if (vdm)
18151 +               kmem_cache_free(dmap_cachep, vdm);
18152 +       return ret;
18153 +}
18154 +
18155 +
18156 +
18157 +int vs_map_device(struct vx_info *vxi,
18158 +       dev_t device, dev_t *target, umode_t mode)
18159 +{
18160 +       int ret, flags = DATTR_MASK;
18161 +
18162 +       if (!vxi) {
18163 +               if (target)
18164 +                       *target = device;
18165 +               goto out;
18166 +       }
18167 +       ret = __lookup_mapping(vxi, device, target, &flags, mode);
18168 +       vxdprintk(VXD_CBIT(misc, 8), "vs_map_device: %08x target: %08x flags: %04x mode: %04x mapped=%d",
18169 +               device, target ? *target : 0, flags, mode, ret);
18170 +out:
18171 +       return (flags & DATTR_MASK);
18172 +}
18173 +
18174 +
18175 +
18176 +static int do_set_mapping(struct vx_info *vxi,
18177 +       dev_t device, dev_t target, int flags, umode_t mode)
18178 +{
18179 +       if (device) {
18180 +               struct vs_mapping *new;
18181 +
18182 +               new = kmem_cache_alloc(dmap_cachep, GFP_KERNEL);
18183 +               if (!new)
18184 +                       return -ENOMEM;
18185 +
18186 +               INIT_HLIST_NODE(&new->dm_hlist);
18187 +               new->device = device;
18188 +               new->target.target = target;
18189 +               new->target.flags = flags | mode;
18190 +               new->xid = (vxi ? vxi->vx_id : 0);
18191 +
18192 +               vxdprintk(VXD_CBIT(misc, 8), "do_set_mapping: %08x target: %08x flags: %04x", device, target, flags);
18193 +               __hash_mapping(vxi, new);
18194 +       } else {
18195 +               struct vx_dmap_target new = {
18196 +                       .target = target,
18197 +                       .flags = flags | mode,
18198 +               };
18199 +               __set_default(vxi, mode, &new);
18200 +       }
18201 +       return 0;
18202 +}
18203 +
18204 +
18205 +static int do_unset_mapping(struct vx_info *vxi,
18206 +       dev_t device, dev_t target, int flags, umode_t mode)
18207 +{
18208 +       int ret = -EINVAL;
18209 +
18210 +       if (device) {
18211 +               ret = __remove_mapping(vxi, device, mode);
18212 +               if (ret < 0)
18213 +                       goto out;
18214 +       } else {
18215 +               ret = __remove_default(vxi, mode);
18216 +               if (ret < 0)
18217 +                       goto out;
18218 +       }
18219 +
18220 +out:
18221 +       return ret;
18222 +}
18223 +
18224 +
18225 +static inline int __user_device(const char __user *name, dev_t *dev,
18226 +       umode_t *mode)
18227 +{
18228 +       struct nameidata nd;
18229 +       int ret;
18230 +
18231 +       if (!name) {
18232 +               *dev = 0;
18233 +               return 0;
18234 +       }
18235 +       ret = user_lpath(name, &nd.path);
18236 +       if (ret)
18237 +               return ret;
18238 +       if (nd.path.dentry->d_inode) {
18239 +               *dev = nd.path.dentry->d_inode->i_rdev;
18240 +               *mode = nd.path.dentry->d_inode->i_mode;
18241 +       }
18242 +       path_put(&nd.path);
18243 +       return 0;
18244 +}
18245 +
18246 +static inline int __mapping_mode(dev_t device, dev_t target,
18247 +       umode_t device_mode, umode_t target_mode, umode_t *mode)
18248 +{
18249 +       if (device)
18250 +               *mode = device_mode & S_IFMT;
18251 +       else if (target)
18252 +               *mode = target_mode & S_IFMT;
18253 +       else
18254 +               return -EINVAL;
18255 +
18256 +       /* if both given, device and target mode have to match */
18257 +       if (device && target &&
18258 +               ((device_mode ^ target_mode) & S_IFMT))
18259 +               return -EINVAL;
18260 +       return 0;
18261 +}
18262 +
18263 +
18264 +static inline int do_mapping(struct vx_info *vxi, const char __user *device_path,
18265 +       const char __user *target_path, int flags, int set)
18266 +{
18267 +       dev_t device = ~0, target = ~0;
18268 +       umode_t device_mode = 0, target_mode = 0, mode;
18269 +       int ret;
18270 +
18271 +       ret = __user_device(device_path, &device, &device_mode);
18272 +       if (ret)
18273 +               return ret;
18274 +       ret = __user_device(target_path, &target, &target_mode);
18275 +       if (ret)
18276 +               return ret;
18277 +
18278 +       ret = __mapping_mode(device, target,
18279 +               device_mode, target_mode, &mode);
18280 +       if (ret)
18281 +               return ret;
18282 +
18283 +       if (set)
18284 +               return do_set_mapping(vxi, device, target,
18285 +                       flags, mode);
18286 +       else
18287 +               return do_unset_mapping(vxi, device, target,
18288 +                       flags, mode);
18289 +}
18290 +
18291 +
18292 +int vc_set_mapping(struct vx_info *vxi, void __user *data)
18293 +{
18294 +       struct vcmd_set_mapping_v0 vc_data;
18295 +
18296 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18297 +               return -EFAULT;
18298 +
18299 +       return do_mapping(vxi, vc_data.device, vc_data.target,
18300 +               vc_data.flags, 1);
18301 +}
18302 +
18303 +int vc_unset_mapping(struct vx_info *vxi, void __user *data)
18304 +{
18305 +       struct vcmd_set_mapping_v0 vc_data;
18306 +
18307 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18308 +               return -EFAULT;
18309 +
18310 +       return do_mapping(vxi, vc_data.device, vc_data.target,
18311 +               vc_data.flags, 0);
18312 +}
18313 +
18314 +
18315 +#ifdef CONFIG_COMPAT
18316 +
18317 +int vc_set_mapping_x32(struct vx_info *vxi, void __user *data)
18318 +{
18319 +       struct vcmd_set_mapping_v0_x32 vc_data;
18320 +
18321 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18322 +               return -EFAULT;
18323 +
18324 +       return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
18325 +               compat_ptr(vc_data.target_ptr), vc_data.flags, 1);
18326 +}
18327 +
18328 +int vc_unset_mapping_x32(struct vx_info *vxi, void __user *data)
18329 +{
18330 +       struct vcmd_set_mapping_v0_x32 vc_data;
18331 +
18332 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18333 +               return -EFAULT;
18334 +
18335 +       return do_mapping(vxi, compat_ptr(vc_data.device_ptr),
18336 +               compat_ptr(vc_data.target_ptr), vc_data.flags, 0);
18337 +}
18338 +
18339 +#endif /* CONFIG_COMPAT */
18340 +
18341 +
18342 diff -NurpP --minimal linux-2.6.36/kernel/vserver/dlimit.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/dlimit.c
18343 --- linux-2.6.36/kernel/vserver/dlimit.c        1970-01-01 01:00:00.000000000 +0100
18344 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/dlimit.c  2010-10-21 13:09:36.000000000 +0200
18345 @@ -0,0 +1,531 @@
18346 +/*
18347 + *  linux/kernel/vserver/dlimit.c
18348 + *
18349 + *  Virtual Server: Context Disk Limits
18350 + *
18351 + *  Copyright (C) 2004-2009  Herbert Pötzl
18352 + *
18353 + *  V0.01  initial version
18354 + *  V0.02  compat32 splitup
18355 + *  V0.03  extended interface
18356 + *
18357 + */
18358 +
18359 +#include <linux/statfs.h>
18360 +#include <linux/sched.h>
18361 +#include <linux/namei.h>
18362 +#include <linux/vs_tag.h>
18363 +#include <linux/vs_dlimit.h>
18364 +#include <linux/vserver/dlimit_cmd.h>
18365 +#include <linux/slab.h>
18366 +// #include <linux/gfp.h>
18367 +
18368 +#include <asm/uaccess.h>
18369 +
18370 +/*     __alloc_dl_info()
18371 +
18372 +       * allocate an initialized dl_info struct
18373 +       * doesn't make it visible (hash)                        */
18374 +
18375 +static struct dl_info *__alloc_dl_info(struct super_block *sb, tag_t tag)
18376 +{
18377 +       struct dl_info *new = NULL;
18378 +
18379 +       vxdprintk(VXD_CBIT(dlim, 5),
18380 +               "alloc_dl_info(%p,%d)*", sb, tag);
18381 +
18382 +       /* would this benefit from a slab cache? */
18383 +       new = kmalloc(sizeof(struct dl_info), GFP_KERNEL);
18384 +       if (!new)
18385 +               return 0;
18386 +
18387 +       memset(new, 0, sizeof(struct dl_info));
18388 +       new->dl_tag = tag;
18389 +       new->dl_sb = sb;
18390 +       INIT_RCU_HEAD(&new->dl_rcu);
18391 +       INIT_HLIST_NODE(&new->dl_hlist);
18392 +       spin_lock_init(&new->dl_lock);
18393 +       atomic_set(&new->dl_refcnt, 0);
18394 +       atomic_set(&new->dl_usecnt, 0);
18395 +
18396 +       /* rest of init goes here */
18397 +
18398 +       vxdprintk(VXD_CBIT(dlim, 4),
18399 +               "alloc_dl_info(%p,%d) = %p", sb, tag, new);
18400 +       return new;
18401 +}
18402 +
18403 +/*     __dealloc_dl_info()
18404 +
18405 +       * final disposal of dl_info                             */
18406 +
18407 +static void __dealloc_dl_info(struct dl_info *dli)
18408 +{
18409 +       vxdprintk(VXD_CBIT(dlim, 4),
18410 +               "dealloc_dl_info(%p)", dli);
18411 +
18412 +       dli->dl_hlist.next = LIST_POISON1;
18413 +       dli->dl_tag = -1;
18414 +       dli->dl_sb = 0;
18415 +
18416 +       BUG_ON(atomic_read(&dli->dl_usecnt));
18417 +       BUG_ON(atomic_read(&dli->dl_refcnt));
18418 +
18419 +       kfree(dli);
18420 +}
18421 +
18422 +
18423 +/*     hash table for dl_info hash */
18424 +
18425 +#define DL_HASH_SIZE   13
18426 +
18427 +struct hlist_head dl_info_hash[DL_HASH_SIZE];
18428 +
18429 +static spinlock_t dl_info_hash_lock = SPIN_LOCK_UNLOCKED;
18430 +
18431 +
18432 +static inline unsigned int __hashval(struct super_block *sb, tag_t tag)
18433 +{
18434 +       return ((tag ^ (unsigned long)sb) % DL_HASH_SIZE);
18435 +}
18436 +
18437 +
18438 +
18439 +/*     __hash_dl_info()
18440 +
18441 +       * add the dli to the global hash table
18442 +       * requires the hash_lock to be held                     */
18443 +
18444 +static inline void __hash_dl_info(struct dl_info *dli)
18445 +{
18446 +       struct hlist_head *head;
18447 +
18448 +       vxdprintk(VXD_CBIT(dlim, 6),
18449 +               "__hash_dl_info: %p[#%d]", dli, dli->dl_tag);
18450 +       get_dl_info(dli);
18451 +       head = &dl_info_hash[__hashval(dli->dl_sb, dli->dl_tag)];
18452 +       hlist_add_head_rcu(&dli->dl_hlist, head);
18453 +}
18454 +
18455 +/*     __unhash_dl_info()
18456 +
18457 +       * remove the dli from the global hash table
18458 +       * requires the hash_lock to be held                     */
18459 +
18460 +static inline void __unhash_dl_info(struct dl_info *dli)
18461 +{
18462 +       vxdprintk(VXD_CBIT(dlim, 6),
18463 +               "__unhash_dl_info: %p[#%d]", dli, dli->dl_tag);
18464 +       hlist_del_rcu(&dli->dl_hlist);
18465 +       put_dl_info(dli);
18466 +}
18467 +
18468 +
18469 +/*     __lookup_dl_info()
18470 +
18471 +       * requires the rcu_read_lock()
18472 +       * doesn't increment the dl_refcnt                       */
18473 +
18474 +static inline struct dl_info *__lookup_dl_info(struct super_block *sb, tag_t tag)
18475 +{
18476 +       struct hlist_head *head = &dl_info_hash[__hashval(sb, tag)];
18477 +       struct hlist_node *pos;
18478 +       struct dl_info *dli;
18479 +
18480 +       hlist_for_each_entry_rcu(dli, pos, head, dl_hlist) {
18481 +
18482 +               if (dli->dl_tag == tag && dli->dl_sb == sb) {
18483 +                       return dli;
18484 +               }
18485 +       }
18486 +       return NULL;
18487 +}
18488 +
18489 +
18490 +struct dl_info *locate_dl_info(struct super_block *sb, tag_t tag)
18491 +{
18492 +       struct dl_info *dli;
18493 +
18494 +       rcu_read_lock();
18495 +       dli = get_dl_info(__lookup_dl_info(sb, tag));
18496 +       vxdprintk(VXD_CBIT(dlim, 7),
18497 +               "locate_dl_info(%p,#%d) = %p", sb, tag, dli);
18498 +       rcu_read_unlock();
18499 +       return dli;
18500 +}
18501 +
18502 +void rcu_free_dl_info(struct rcu_head *head)
18503 +{
18504 +       struct dl_info *dli = container_of(head, struct dl_info, dl_rcu);
18505 +       int usecnt, refcnt;
18506 +
18507 +       BUG_ON(!dli || !head);
18508 +
18509 +       usecnt = atomic_read(&dli->dl_usecnt);
18510 +       BUG_ON(usecnt < 0);
18511 +
18512 +       refcnt = atomic_read(&dli->dl_refcnt);
18513 +       BUG_ON(refcnt < 0);
18514 +
18515 +       vxdprintk(VXD_CBIT(dlim, 3),
18516 +               "rcu_free_dl_info(%p)", dli);
18517 +       if (!usecnt)
18518 +               __dealloc_dl_info(dli);
18519 +       else
18520 +               printk("!!! rcu didn't free\n");
18521 +}
18522 +
18523 +
18524 +
18525 +
18526 +static int do_addrem_dlimit(uint32_t id, const char __user *name,
18527 +       uint32_t flags, int add)
18528 +{
18529 +       struct path path;
18530 +       int ret;
18531 +
18532 +       ret = user_lpath(name, &path);
18533 +       if (!ret) {
18534 +               struct super_block *sb;
18535 +               struct dl_info *dli;
18536 +
18537 +               ret = -EINVAL;
18538 +               if (!path.dentry->d_inode)
18539 +                       goto out_release;
18540 +               if (!(sb = path.dentry->d_inode->i_sb))
18541 +                       goto out_release;
18542 +
18543 +               if (add) {
18544 +                       dli = __alloc_dl_info(sb, id);
18545 +                       spin_lock(&dl_info_hash_lock);
18546 +
18547 +                       ret = -EEXIST;
18548 +                       if (__lookup_dl_info(sb, id))
18549 +                               goto out_unlock;
18550 +                       __hash_dl_info(dli);
18551 +                       dli = NULL;
18552 +               } else {
18553 +                       spin_lock(&dl_info_hash_lock);
18554 +                       dli = __lookup_dl_info(sb, id);
18555 +
18556 +                       ret = -ESRCH;
18557 +                       if (!dli)
18558 +                               goto out_unlock;
18559 +                       __unhash_dl_info(dli);
18560 +               }
18561 +               ret = 0;
18562 +       out_unlock:
18563 +               spin_unlock(&dl_info_hash_lock);
18564 +               if (add && dli)
18565 +                       __dealloc_dl_info(dli);
18566 +       out_release:
18567 +               path_put(&path);
18568 +       }
18569 +       return ret;
18570 +}
18571 +
18572 +int vc_add_dlimit(uint32_t id, void __user *data)
18573 +{
18574 +       struct vcmd_ctx_dlimit_base_v0 vc_data;
18575 +
18576 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18577 +               return -EFAULT;
18578 +
18579 +       return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 1);
18580 +}
18581 +
18582 +int vc_rem_dlimit(uint32_t id, void __user *data)
18583 +{
18584 +       struct vcmd_ctx_dlimit_base_v0 vc_data;
18585 +
18586 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18587 +               return -EFAULT;
18588 +
18589 +       return do_addrem_dlimit(id, vc_data.name, vc_data.flags, 0);
18590 +}
18591 +
18592 +#ifdef CONFIG_COMPAT
18593 +
18594 +int vc_add_dlimit_x32(uint32_t id, void __user *data)
18595 +{
18596 +       struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
18597 +
18598 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18599 +               return -EFAULT;
18600 +
18601 +       return do_addrem_dlimit(id,
18602 +               compat_ptr(vc_data.name_ptr), vc_data.flags, 1);
18603 +}
18604 +
18605 +int vc_rem_dlimit_x32(uint32_t id, void __user *data)
18606 +{
18607 +       struct vcmd_ctx_dlimit_base_v0_x32 vc_data;
18608 +
18609 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18610 +               return -EFAULT;
18611 +
18612 +       return do_addrem_dlimit(id,
18613 +               compat_ptr(vc_data.name_ptr), vc_data.flags, 0);
18614 +}
18615 +
18616 +#endif /* CONFIG_COMPAT */
18617 +
18618 +
18619 +static inline
18620 +int do_set_dlimit(uint32_t id, const char __user *name,
18621 +       uint32_t space_used, uint32_t space_total,
18622 +       uint32_t inodes_used, uint32_t inodes_total,
18623 +       uint32_t reserved, uint32_t flags)
18624 +{
18625 +       struct path path;
18626 +       int ret;
18627 +
18628 +       ret = user_lpath(name, &path);
18629 +       if (!ret) {
18630 +               struct super_block *sb;
18631 +               struct dl_info *dli;
18632 +
18633 +               ret = -EINVAL;
18634 +               if (!path.dentry->d_inode)
18635 +                       goto out_release;
18636 +               if (!(sb = path.dentry->d_inode->i_sb))
18637 +                       goto out_release;
18638 +
18639 +               /* sanity checks */
18640 +               if ((reserved != CDLIM_KEEP &&
18641 +                       reserved > 100) ||
18642 +                       (inodes_used != CDLIM_KEEP &&
18643 +                       inodes_used > inodes_total) ||
18644 +                       (space_used != CDLIM_KEEP &&
18645 +                       space_used > space_total))
18646 +                       goto out_release;
18647 +
18648 +               ret = -ESRCH;
18649 +               dli = locate_dl_info(sb, id);
18650 +               if (!dli)
18651 +                       goto out_release;
18652 +
18653 +               spin_lock(&dli->dl_lock);
18654 +
18655 +               if (inodes_used != CDLIM_KEEP)
18656 +                       dli->dl_inodes_used = inodes_used;
18657 +               if (inodes_total != CDLIM_KEEP)
18658 +                       dli->dl_inodes_total = inodes_total;
18659 +               if (space_used != CDLIM_KEEP)
18660 +                       dli->dl_space_used = dlimit_space_32to64(
18661 +                               space_used, flags, DLIMS_USED);
18662 +
18663 +               if (space_total == CDLIM_INFINITY)
18664 +                       dli->dl_space_total = DLIM_INFINITY;
18665 +               else if (space_total != CDLIM_KEEP)
18666 +                       dli->dl_space_total = dlimit_space_32to64(
18667 +                               space_total, flags, DLIMS_TOTAL);
18668 +
18669 +               if (reserved != CDLIM_KEEP)
18670 +                       dli->dl_nrlmult = (1 << 10) * (100 - reserved) / 100;
18671 +
18672 +               spin_unlock(&dli->dl_lock);
18673 +
18674 +               put_dl_info(dli);
18675 +               ret = 0;
18676 +
18677 +       out_release:
18678 +               path_put(&path);
18679 +       }
18680 +       return ret;
18681 +}
18682 +
18683 +int vc_set_dlimit(uint32_t id, void __user *data)
18684 +{
18685 +       struct vcmd_ctx_dlimit_v0 vc_data;
18686 +
18687 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18688 +               return -EFAULT;
18689 +
18690 +       return do_set_dlimit(id, vc_data.name,
18691 +               vc_data.space_used, vc_data.space_total,
18692 +               vc_data.inodes_used, vc_data.inodes_total,
18693 +               vc_data.reserved, vc_data.flags);
18694 +}
18695 +
18696 +#ifdef CONFIG_COMPAT
18697 +
18698 +int vc_set_dlimit_x32(uint32_t id, void __user *data)
18699 +{
18700 +       struct vcmd_ctx_dlimit_v0_x32 vc_data;
18701 +
18702 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18703 +               return -EFAULT;
18704 +
18705 +       return do_set_dlimit(id, compat_ptr(vc_data.name_ptr),
18706 +               vc_data.space_used, vc_data.space_total,
18707 +               vc_data.inodes_used, vc_data.inodes_total,
18708 +               vc_data.reserved, vc_data.flags);
18709 +}
18710 +
18711 +#endif /* CONFIG_COMPAT */
18712 +
18713 +
18714 +static inline
18715 +int do_get_dlimit(uint32_t id, const char __user *name,
18716 +       uint32_t *space_used, uint32_t *space_total,
18717 +       uint32_t *inodes_used, uint32_t *inodes_total,
18718 +       uint32_t *reserved, uint32_t *flags)
18719 +{
18720 +       struct path path;
18721 +       int ret;
18722 +
18723 +       ret = user_lpath(name, &path);
18724 +       if (!ret) {
18725 +               struct super_block *sb;
18726 +               struct dl_info *dli;
18727 +
18728 +               ret = -EINVAL;
18729 +               if (!path.dentry->d_inode)
18730 +                       goto out_release;
18731 +               if (!(sb = path.dentry->d_inode->i_sb))
18732 +                       goto out_release;
18733 +
18734 +               ret = -ESRCH;
18735 +               dli = locate_dl_info(sb, id);
18736 +               if (!dli)
18737 +                       goto out_release;
18738 +
18739 +               spin_lock(&dli->dl_lock);
18740 +               *inodes_used = dli->dl_inodes_used;
18741 +               *inodes_total = dli->dl_inodes_total;
18742 +
18743 +               *space_used = dlimit_space_64to32(
18744 +                       dli->dl_space_used, flags, DLIMS_USED);
18745 +
18746 +               if (dli->dl_space_total == DLIM_INFINITY)
18747 +                       *space_total = CDLIM_INFINITY;
18748 +               else
18749 +                       *space_total = dlimit_space_64to32(
18750 +                               dli->dl_space_total, flags, DLIMS_TOTAL);
18751 +
18752 +               *reserved = 100 - ((dli->dl_nrlmult * 100 + 512) >> 10);
18753 +               spin_unlock(&dli->dl_lock);
18754 +
18755 +               put_dl_info(dli);
18756 +               ret = -EFAULT;
18757 +
18758 +               ret = 0;
18759 +       out_release:
18760 +               path_put(&path);
18761 +       }
18762 +       return ret;
18763 +}
18764 +
18765 +
18766 +int vc_get_dlimit(uint32_t id, void __user *data)
18767 +{
18768 +       struct vcmd_ctx_dlimit_v0 vc_data;
18769 +       int ret;
18770 +
18771 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18772 +               return -EFAULT;
18773 +
18774 +       ret = do_get_dlimit(id, vc_data.name,
18775 +               &vc_data.space_used, &vc_data.space_total,
18776 +               &vc_data.inodes_used, &vc_data.inodes_total,
18777 +               &vc_data.reserved, &vc_data.flags);
18778 +       if (ret)
18779 +               return ret;
18780 +
18781 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18782 +               return -EFAULT;
18783 +       return 0;
18784 +}
18785 +
18786 +#ifdef CONFIG_COMPAT
18787 +
18788 +int vc_get_dlimit_x32(uint32_t id, void __user *data)
18789 +{
18790 +       struct vcmd_ctx_dlimit_v0_x32 vc_data;
18791 +       int ret;
18792 +
18793 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
18794 +               return -EFAULT;
18795 +
18796 +       ret = do_get_dlimit(id, compat_ptr(vc_data.name_ptr),
18797 +               &vc_data.space_used, &vc_data.space_total,
18798 +               &vc_data.inodes_used, &vc_data.inodes_total,
18799 +               &vc_data.reserved, &vc_data.flags);
18800 +       if (ret)
18801 +               return ret;
18802 +
18803 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
18804 +               return -EFAULT;
18805 +       return 0;
18806 +}
18807 +
18808 +#endif /* CONFIG_COMPAT */
18809 +
18810 +
18811 +void vx_vsi_statfs(struct super_block *sb, struct kstatfs *buf)
18812 +{
18813 +       struct dl_info *dli;
18814 +       __u64 blimit, bfree, bavail;
18815 +       __u32 ifree;
18816 +
18817 +       dli = locate_dl_info(sb, dx_current_tag());
18818 +       if (!dli)
18819 +               return;
18820 +
18821 +       spin_lock(&dli->dl_lock);
18822 +       if (dli->dl_inodes_total == (unsigned long)DLIM_INFINITY)
18823 +               goto no_ilim;
18824 +
18825 +       /* reduce max inodes available to limit */
18826 +       if (buf->f_files > dli->dl_inodes_total)
18827 +               buf->f_files = dli->dl_inodes_total;
18828 +
18829 +       ifree = dli->dl_inodes_total - dli->dl_inodes_used;
18830 +       /* reduce free inodes to min */
18831 +       if (ifree < buf->f_ffree)
18832 +               buf->f_ffree = ifree;
18833 +
18834 +no_ilim:
18835 +       if (dli->dl_space_total == DLIM_INFINITY)
18836 +               goto no_blim;
18837 +
18838 +       blimit = dli->dl_space_total >> sb->s_blocksize_bits;
18839 +
18840 +       if (dli->dl_space_total < dli->dl_space_used)
18841 +               bfree = 0;
18842 +       else
18843 +               bfree = (dli->dl_space_total - dli->dl_space_used)
18844 +                       >> sb->s_blocksize_bits;
18845 +
18846 +       bavail = ((dli->dl_space_total >> 10) * dli->dl_nrlmult);
18847 +       if (bavail < dli->dl_space_used)
18848 +               bavail = 0;
18849 +       else
18850 +               bavail = (bavail - dli->dl_space_used)
18851 +                       >> sb->s_blocksize_bits;
18852 +
18853 +       /* reduce max space available to limit */
18854 +       if (buf->f_blocks > blimit)
18855 +               buf->f_blocks = blimit;
18856 +
18857 +       /* reduce free space to min */
18858 +       if (bfree < buf->f_bfree)
18859 +               buf->f_bfree = bfree;
18860 +
18861 +       /* reduce avail space to min */
18862 +       if (bavail < buf->f_bavail)
18863 +               buf->f_bavail = bavail;
18864 +
18865 +no_blim:
18866 +       spin_unlock(&dli->dl_lock);
18867 +       put_dl_info(dli);
18868 +
18869 +       return;
18870 +}
18871 +
18872 +#include <linux/module.h>
18873 +
18874 +EXPORT_SYMBOL_GPL(locate_dl_info);
18875 +EXPORT_SYMBOL_GPL(rcu_free_dl_info);
18876 +
18877 diff -NurpP --minimal linux-2.6.36/kernel/vserver/helper.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/helper.c
18878 --- linux-2.6.36/kernel/vserver/helper.c        1970-01-01 01:00:00.000000000 +0100
18879 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/helper.c  2010-10-21 13:09:36.000000000 +0200
18880 @@ -0,0 +1,223 @@
18881 +/*
18882 + *  linux/kernel/vserver/helper.c
18883 + *
18884 + *  Virtual Context Support
18885 + *
18886 + *  Copyright (C) 2004-2007  Herbert Pötzl
18887 + *
18888 + *  V0.01  basic helper
18889 + *
18890 + */
18891 +
18892 +#include <linux/kmod.h>
18893 +#include <linux/reboot.h>
18894 +#include <linux/vs_context.h>
18895 +#include <linux/vs_network.h>
18896 +#include <linux/vserver/signal.h>
18897 +
18898 +
18899 +char vshelper_path[255] = "/sbin/vshelper";
18900 +
18901 +
18902 +static int do_vshelper(char *name, char *argv[], char *envp[], int sync)
18903 +{
18904 +       int ret;
18905 +
18906 +       if ((ret = call_usermodehelper(name, argv, envp, sync))) {
18907 +               printk( KERN_WARNING
18908 +                       "%s: (%s %s) returned %s with %d\n",
18909 +                       name, argv[1], argv[2],
18910 +                       sync ? "sync" : "async", ret);
18911 +       }
18912 +       vxdprintk(VXD_CBIT(switch, 4),
18913 +               "%s: (%s %s) returned %s with %d",
18914 +               name, argv[1], argv[2], sync ? "sync" : "async", ret);
18915 +       return ret;
18916 +}
18917 +
18918 +/*
18919 + *      vshelper path is set via /proc/sys
18920 + *      invoked by vserver sys_reboot(), with
18921 + *      the following arguments
18922 + *
18923 + *      argv [0] = vshelper_path;
18924 + *      argv [1] = action: "restart", "halt", "poweroff", ...
18925 + *      argv [2] = context identifier
18926 + *
18927 + *      envp [*] = type-specific parameters
18928 + */
18929 +
18930 +long vs_reboot_helper(struct vx_info *vxi, int cmd, void __user *arg)
18931 +{
18932 +       char id_buf[8], cmd_buf[16];
18933 +       char uid_buf[16], pid_buf[16];
18934 +       int ret;
18935 +
18936 +       char *argv[] = {vshelper_path, NULL, id_buf, 0};
18937 +       char *envp[] = {"HOME=/", "TERM=linux",
18938 +                       "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
18939 +                       uid_buf, pid_buf, cmd_buf, 0};
18940 +
18941 +       if (vx_info_state(vxi, VXS_HELPER))
18942 +               return -EAGAIN;
18943 +       vxi->vx_state |= VXS_HELPER;
18944 +
18945 +       snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
18946 +
18947 +       snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
18948 +       snprintf(uid_buf, sizeof(uid_buf)-1, "VS_UID=%d", current_uid());
18949 +       snprintf(pid_buf, sizeof(pid_buf)-1, "VS_PID=%d", current->pid);
18950 +
18951 +       switch (cmd) {
18952 +       case LINUX_REBOOT_CMD_RESTART:
18953 +               argv[1] = "restart";
18954 +               break;
18955 +
18956 +       case LINUX_REBOOT_CMD_HALT:
18957 +               argv[1] = "halt";
18958 +               break;
18959 +
18960 +       case LINUX_REBOOT_CMD_POWER_OFF:
18961 +               argv[1] = "poweroff";
18962 +               break;
18963 +
18964 +       case LINUX_REBOOT_CMD_SW_SUSPEND:
18965 +               argv[1] = "swsusp";
18966 +               break;
18967 +
18968 +       case LINUX_REBOOT_CMD_OOM:
18969 +               argv[1] = "oom";
18970 +               break;
18971 +
18972 +       default:
18973 +               vxi->vx_state &= ~VXS_HELPER;
18974 +               return 0;
18975 +       }
18976 +
18977 +       ret = do_vshelper(vshelper_path, argv, envp, 0);
18978 +       vxi->vx_state &= ~VXS_HELPER;
18979 +       __wakeup_vx_info(vxi);
18980 +       return (ret) ? -EPERM : 0;
18981 +}
18982 +
18983 +
18984 +long vs_reboot(unsigned int cmd, void __user *arg)
18985 +{
18986 +       struct vx_info *vxi = current_vx_info();
18987 +       long ret = 0;
18988 +
18989 +       vxdprintk(VXD_CBIT(misc, 5),
18990 +               "vs_reboot(%p[#%d],%u)",
18991 +               vxi, vxi ? vxi->vx_id : 0, cmd);
18992 +
18993 +       ret = vs_reboot_helper(vxi, cmd, arg);
18994 +       if (ret)
18995 +               return ret;
18996 +
18997 +       vxi->reboot_cmd = cmd;
18998 +       if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
18999 +               switch (cmd) {
19000 +               case LINUX_REBOOT_CMD_RESTART:
19001 +               case LINUX_REBOOT_CMD_HALT:
19002 +               case LINUX_REBOOT_CMD_POWER_OFF:
19003 +                       vx_info_kill(vxi, 0, SIGKILL);
19004 +                       vx_info_kill(vxi, 1, SIGKILL);
19005 +               default:
19006 +                       break;
19007 +               }
19008 +       }
19009 +       return 0;
19010 +}
19011 +
19012 +long vs_oom_action(unsigned int cmd)
19013 +{
19014 +       struct vx_info *vxi = current_vx_info();
19015 +       long ret = 0;
19016 +
19017 +       vxdprintk(VXD_CBIT(misc, 5),
19018 +               "vs_oom_action(%p[#%d],%u)",
19019 +               vxi, vxi ? vxi->vx_id : 0, cmd);
19020 +
19021 +       ret = vs_reboot_helper(vxi, cmd, NULL);
19022 +       if (ret)
19023 +               return ret;
19024 +
19025 +       vxi->reboot_cmd = cmd;
19026 +       if (vx_info_flags(vxi, VXF_REBOOT_KILL, 0)) {
19027 +               vx_info_kill(vxi, 0, SIGKILL);
19028 +               vx_info_kill(vxi, 1, SIGKILL);
19029 +       }
19030 +       return 0;
19031 +}
19032 +
19033 +/*
19034 + *      argv [0] = vshelper_path;
19035 + *      argv [1] = action: "startup", "shutdown"
19036 + *      argv [2] = context identifier
19037 + *
19038 + *      envp [*] = type-specific parameters
19039 + */
19040 +
19041 +long vs_state_change(struct vx_info *vxi, unsigned int cmd)
19042 +{
19043 +       char id_buf[8], cmd_buf[16];
19044 +       char *argv[] = {vshelper_path, NULL, id_buf, 0};
19045 +       char *envp[] = {"HOME=/", "TERM=linux",
19046 +                       "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19047 +
19048 +       if (!vx_info_flags(vxi, VXF_SC_HELPER, 0))
19049 +               return 0;
19050 +
19051 +       snprintf(id_buf, sizeof(id_buf)-1, "%d", vxi->vx_id);
19052 +       snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19053 +
19054 +       switch (cmd) {
19055 +       case VSC_STARTUP:
19056 +               argv[1] = "startup";
19057 +               break;
19058 +       case VSC_SHUTDOWN:
19059 +               argv[1] = "shutdown";
19060 +               break;
19061 +       default:
19062 +               return 0;
19063 +       }
19064 +
19065 +       return do_vshelper(vshelper_path, argv, envp, 1);
19066 +}
19067 +
19068 +
19069 +/*
19070 + *      argv [0] = vshelper_path;
19071 + *      argv [1] = action: "netup", "netdown"
19072 + *      argv [2] = context identifier
19073 + *
19074 + *      envp [*] = type-specific parameters
19075 + */
19076 +
19077 +long vs_net_change(struct nx_info *nxi, unsigned int cmd)
19078 +{
19079 +       char id_buf[8], cmd_buf[16];
19080 +       char *argv[] = {vshelper_path, NULL, id_buf, 0};
19081 +       char *envp[] = {"HOME=/", "TERM=linux",
19082 +                       "PATH=/sbin:/usr/sbin:/bin:/usr/bin", cmd_buf, 0};
19083 +
19084 +       if (!nx_info_flags(nxi, NXF_SC_HELPER, 0))
19085 +               return 0;
19086 +
19087 +       snprintf(id_buf, sizeof(id_buf)-1, "%d", nxi->nx_id);
19088 +       snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd);
19089 +
19090 +       switch (cmd) {
19091 +       case VSC_NETUP:
19092 +               argv[1] = "netup";
19093 +               break;
19094 +       case VSC_NETDOWN:
19095 +               argv[1] = "netdown";
19096 +               break;
19097 +       default:
19098 +               return 0;
19099 +       }
19100 +
19101 +       return do_vshelper(vshelper_path, argv, envp, 1);
19102 +}
19103 +
19104 diff -NurpP --minimal linux-2.6.36/kernel/vserver/history.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/history.c
19105 --- linux-2.6.36/kernel/vserver/history.c       1970-01-01 01:00:00.000000000 +0100
19106 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/history.c 2010-10-21 13:09:36.000000000 +0200
19107 @@ -0,0 +1,258 @@
19108 +/*
19109 + *  kernel/vserver/history.c
19110 + *
19111 + *  Virtual Context History Backtrace
19112 + *
19113 + *  Copyright (C) 2004-2007  Herbert Pötzl
19114 + *
19115 + *  V0.01  basic structure
19116 + *  V0.02  hash/unhash and trace
19117 + *  V0.03  preemption fixes
19118 + *
19119 + */
19120 +
19121 +#include <linux/module.h>
19122 +#include <asm/uaccess.h>
19123 +
19124 +#include <linux/vserver/context.h>
19125 +#include <linux/vserver/debug.h>
19126 +#include <linux/vserver/debug_cmd.h>
19127 +#include <linux/vserver/history.h>
19128 +
19129 +
19130 +#ifdef CONFIG_VSERVER_HISTORY
19131 +#define VXH_SIZE       CONFIG_VSERVER_HISTORY_SIZE
19132 +#else
19133 +#define VXH_SIZE       64
19134 +#endif
19135 +
19136 +struct _vx_history {
19137 +       unsigned int counter;
19138 +
19139 +       struct _vx_hist_entry entry[VXH_SIZE + 1];
19140 +};
19141 +
19142 +
19143 +DEFINE_PER_CPU(struct _vx_history, vx_history_buffer);
19144 +
19145 +unsigned volatile int vxh_active = 1;
19146 +
19147 +static atomic_t sequence = ATOMIC_INIT(0);
19148 +
19149 +
19150 +/*     vxh_advance()
19151 +
19152 +       * requires disabled preemption                          */
19153 +
19154 +struct _vx_hist_entry *vxh_advance(void *loc)
19155 +{
19156 +       unsigned int cpu = smp_processor_id();
19157 +       struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
19158 +       struct _vx_hist_entry *entry;
19159 +       unsigned int index;
19160 +
19161 +       index = vxh_active ? (hist->counter++ % VXH_SIZE) : VXH_SIZE;
19162 +       entry = &hist->entry[index];
19163 +
19164 +       entry->seq = atomic_inc_return(&sequence);
19165 +       entry->loc = loc;
19166 +       return entry;
19167 +}
19168 +
19169 +EXPORT_SYMBOL_GPL(vxh_advance);
19170 +
19171 +
19172 +#define VXH_LOC_FMTS   "(#%04x,*%d):%p"
19173 +
19174 +#define VXH_LOC_ARGS(e)        (e)->seq, cpu, (e)->loc
19175 +
19176 +
19177 +#define VXH_VXI_FMTS   "%p[#%d,%d.%d]"
19178 +
19179 +#define VXH_VXI_ARGS(e)        (e)->vxi.ptr,                           \
19180 +                       (e)->vxi.ptr ? (e)->vxi.xid : 0,        \
19181 +                       (e)->vxi.ptr ? (e)->vxi.usecnt : 0,     \
19182 +                       (e)->vxi.ptr ? (e)->vxi.tasks : 0
19183 +
19184 +void   vxh_dump_entry(struct _vx_hist_entry *e, unsigned cpu)
19185 +{
19186 +       switch (e->type) {
19187 +       case VXH_THROW_OOPS:
19188 +               printk( VXH_LOC_FMTS " oops \n", VXH_LOC_ARGS(e));
19189 +               break;
19190 +
19191 +       case VXH_GET_VX_INFO:
19192 +       case VXH_PUT_VX_INFO:
19193 +               printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
19194 +                       VXH_LOC_ARGS(e),
19195 +                       (e->type == VXH_GET_VX_INFO) ? "get" : "put",
19196 +                       VXH_VXI_ARGS(e));
19197 +               break;
19198 +
19199 +       case VXH_INIT_VX_INFO:
19200 +       case VXH_SET_VX_INFO:
19201 +       case VXH_CLR_VX_INFO:
19202 +               printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
19203 +                       VXH_LOC_ARGS(e),
19204 +                       (e->type == VXH_INIT_VX_INFO) ? "init" :
19205 +                       ((e->type == VXH_SET_VX_INFO) ? "set" : "clr"),
19206 +                       VXH_VXI_ARGS(e), e->sc.data);
19207 +               break;
19208 +
19209 +       case VXH_CLAIM_VX_INFO:
19210 +       case VXH_RELEASE_VX_INFO:
19211 +               printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS " @%p\n",
19212 +                       VXH_LOC_ARGS(e),
19213 +                       (e->type == VXH_CLAIM_VX_INFO) ? "claim" : "release",
19214 +                       VXH_VXI_ARGS(e), e->sc.data);
19215 +               break;
19216 +
19217 +       case VXH_ALLOC_VX_INFO:
19218 +       case VXH_DEALLOC_VX_INFO:
19219 +               printk( VXH_LOC_FMTS " %s_vx_info " VXH_VXI_FMTS "\n",
19220 +                       VXH_LOC_ARGS(e),
19221 +                       (e->type == VXH_ALLOC_VX_INFO) ? "alloc" : "dealloc",
19222 +                       VXH_VXI_ARGS(e));
19223 +               break;
19224 +
19225 +       case VXH_HASH_VX_INFO:
19226 +       case VXH_UNHASH_VX_INFO:
19227 +               printk( VXH_LOC_FMTS " __%s_vx_info " VXH_VXI_FMTS "\n",
19228 +                       VXH_LOC_ARGS(e),
19229 +                       (e->type == VXH_HASH_VX_INFO) ? "hash" : "unhash",
19230 +                       VXH_VXI_ARGS(e));
19231 +               break;
19232 +
19233 +       case VXH_LOC_VX_INFO:
19234 +       case VXH_LOOKUP_VX_INFO:
19235 +       case VXH_CREATE_VX_INFO:
19236 +               printk( VXH_LOC_FMTS " __%s_vx_info [#%d] -> " VXH_VXI_FMTS "\n",
19237 +                       VXH_LOC_ARGS(e),
19238 +                       (e->type == VXH_CREATE_VX_INFO) ? "create" :
19239 +                       ((e->type == VXH_LOC_VX_INFO) ? "loc" : "lookup"),
19240 +                       e->ll.arg, VXH_VXI_ARGS(e));
19241 +               break;
19242 +       }
19243 +}
19244 +
19245 +static void __vxh_dump_history(void)
19246 +{
19247 +       unsigned int i, cpu;
19248 +
19249 +       printk("History:\tSEQ: %8x\tNR_CPUS: %d\n",
19250 +               atomic_read(&sequence), NR_CPUS);
19251 +
19252 +       for (i = 0; i < VXH_SIZE; i++) {
19253 +               for_each_online_cpu(cpu) {
19254 +                       struct _vx_history *hist =
19255 +                               &per_cpu(vx_history_buffer, cpu);
19256 +                       unsigned int index = (hist->counter - i) % VXH_SIZE;
19257 +                       struct _vx_hist_entry *entry = &hist->entry[index];
19258 +
19259 +                       vxh_dump_entry(entry, cpu);
19260 +               }
19261 +       }
19262 +}
19263 +
19264 +void   vxh_dump_history(void)
19265 +{
19266 +       vxh_active = 0;
19267 +#ifdef CONFIG_SMP
19268 +       local_irq_enable();
19269 +       smp_send_stop();
19270 +       local_irq_disable();
19271 +#endif
19272 +       __vxh_dump_history();
19273 +}
19274 +
19275 +
19276 +/* vserver syscall commands below here */
19277 +
19278 +
19279 +int vc_dump_history(uint32_t id)
19280 +{
19281 +       vxh_active = 0;
19282 +       __vxh_dump_history();
19283 +       vxh_active = 1;
19284 +
19285 +       return 0;
19286 +}
19287 +
19288 +
19289 +int do_read_history(struct __user _vx_hist_entry *data,
19290 +       int cpu, uint32_t *index, uint32_t *count)
19291 +{
19292 +       int pos, ret = 0;
19293 +       struct _vx_history *hist = &per_cpu(vx_history_buffer, cpu);
19294 +       int end = hist->counter;
19295 +       int start = end - VXH_SIZE + 2;
19296 +       int idx = *index;
19297 +
19298 +       /* special case: get current pos */
19299 +       if (!*count) {
19300 +               *index = end;
19301 +               return 0;
19302 +       }
19303 +
19304 +       /* have we lost some data? */
19305 +       if (idx < start)
19306 +               idx = start;
19307 +
19308 +       for (pos = 0; (pos < *count) && (idx < end); pos++, idx++) {
19309 +               struct _vx_hist_entry *entry =
19310 +                       &hist->entry[idx % VXH_SIZE];
19311 +
19312 +               /* send entry to userspace */
19313 +               ret = copy_to_user(&data[pos], entry, sizeof(*entry));
19314 +               if (ret)
19315 +                       break;
19316 +       }
19317 +       /* save new index and count */
19318 +       *index = idx;
19319 +       *count = pos;
19320 +       return ret ? ret : (*index < end);
19321 +}
19322 +
19323 +int vc_read_history(uint32_t id, void __user *data)
19324 +{
19325 +       struct vcmd_read_history_v0 vc_data;
19326 +       int ret;
19327 +
19328 +       if (id >= NR_CPUS)
19329 +               return -EINVAL;
19330 +
19331 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19332 +               return -EFAULT;
19333 +
19334 +       ret = do_read_history((struct __user _vx_hist_entry *)vc_data.data,
19335 +               id, &vc_data.index, &vc_data.count);
19336 +
19337 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19338 +               return -EFAULT;
19339 +       return ret;
19340 +}
19341 +
19342 +#ifdef CONFIG_COMPAT
19343 +
19344 +int vc_read_history_x32(uint32_t id, void __user *data)
19345 +{
19346 +       struct vcmd_read_history_v0_x32 vc_data;
19347 +       int ret;
19348 +
19349 +       if (id >= NR_CPUS)
19350 +               return -EINVAL;
19351 +
19352 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19353 +               return -EFAULT;
19354 +
19355 +       ret = do_read_history((struct __user _vx_hist_entry *)
19356 +               compat_ptr(vc_data.data_ptr),
19357 +               id, &vc_data.index, &vc_data.count);
19358 +
19359 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19360 +               return -EFAULT;
19361 +       return ret;
19362 +}
19363 +
19364 +#endif /* CONFIG_COMPAT */
19365 +
19366 diff -NurpP --minimal linux-2.6.36/kernel/vserver/inet.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/inet.c
19367 --- linux-2.6.36/kernel/vserver/inet.c  1970-01-01 01:00:00.000000000 +0100
19368 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/inet.c    2010-10-21 13:09:36.000000000 +0200
19369 @@ -0,0 +1,224 @@
19370 +
19371 +#include <linux/in.h>
19372 +#include <linux/inetdevice.h>
19373 +#include <linux/vs_inet.h>
19374 +#include <linux/vs_inet6.h>
19375 +#include <linux/vserver/debug.h>
19376 +#include <net/route.h>
19377 +#include <net/addrconf.h>
19378 +
19379 +
19380 +int nx_v4_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
19381 +{
19382 +       int ret = 0;
19383 +
19384 +       if (!nxi1 || !nxi2 || nxi1 == nxi2)
19385 +               ret = 1;
19386 +       else {
19387 +               struct nx_addr_v4 *ptr;
19388 +
19389 +               for (ptr = &nxi1->v4; ptr; ptr = ptr->next) {
19390 +                       if (v4_nx_addr_in_nx_info(nxi2, ptr, -1)) {
19391 +                               ret = 1;
19392 +                               break;
19393 +                       }
19394 +               }
19395 +       }
19396 +
19397 +       vxdprintk(VXD_CBIT(net, 2),
19398 +               "nx_v4_addr_conflict(%p,%p): %d",
19399 +               nxi1, nxi2, ret);
19400 +
19401 +       return ret;
19402 +}
19403 +
19404 +
19405 +#ifdef CONFIG_IPV6
19406 +
19407 +int nx_v6_addr_conflict(struct nx_info *nxi1, struct nx_info *nxi2)
19408 +{
19409 +       int ret = 0;
19410 +
19411 +       if (!nxi1 || !nxi2 || nxi1 == nxi2)
19412 +               ret = 1;
19413 +       else {
19414 +               struct nx_addr_v6 *ptr;
19415 +
19416 +               for (ptr = &nxi1->v6; ptr; ptr = ptr->next) {
19417 +                       if (v6_nx_addr_in_nx_info(nxi2, ptr, -1)) {
19418 +                               ret = 1;
19419 +                               break;
19420 +                       }
19421 +               }
19422 +       }
19423 +
19424 +       vxdprintk(VXD_CBIT(net, 2),
19425 +               "nx_v6_addr_conflict(%p,%p): %d",
19426 +               nxi1, nxi2, ret);
19427 +
19428 +       return ret;
19429 +}
19430 +
19431 +#endif
19432 +
19433 +int v4_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
19434 +{
19435 +       struct in_device *in_dev;
19436 +       struct in_ifaddr **ifap;
19437 +       struct in_ifaddr *ifa;
19438 +       int ret = 0;
19439 +
19440 +       if (!dev)
19441 +               goto out;
19442 +       in_dev = in_dev_get(dev);
19443 +       if (!in_dev)
19444 +               goto out;
19445 +
19446 +       for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
19447 +               ifap = &ifa->ifa_next) {
19448 +               if (v4_addr_in_nx_info(nxi, ifa->ifa_local, NXA_MASK_SHOW)) {
19449 +                       ret = 1;
19450 +                       break;
19451 +               }
19452 +       }
19453 +       in_dev_put(in_dev);
19454 +out:
19455 +       return ret;
19456 +}
19457 +
19458 +
19459 +#ifdef CONFIG_IPV6
19460 +
19461 +int v6_dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
19462 +{
19463 +       struct inet6_dev *in_dev;
19464 +       struct inet6_ifaddr *ifa;
19465 +       int ret = 0;
19466 +
19467 +       if (!dev)
19468 +               goto out;
19469 +       in_dev = in6_dev_get(dev);
19470 +       if (!in_dev)
19471 +               goto out;
19472 +
19473 +       // for (ifap = &in_dev->addr_list; (ifa = *ifap) != NULL;
19474 +       list_for_each_entry(ifa, &in_dev->addr_list, if_list) {
19475 +               if (v6_addr_in_nx_info(nxi, &ifa->addr, -1)) {
19476 +                       ret = 1;
19477 +                       break;
19478 +               }
19479 +       }
19480 +       in6_dev_put(in_dev);
19481 +out:
19482 +       return ret;
19483 +}
19484 +
19485 +#endif
19486 +
19487 +int dev_in_nx_info(struct net_device *dev, struct nx_info *nxi)
19488 +{
19489 +       int ret = 1;
19490 +
19491 +       if (!nxi)
19492 +               goto out;
19493 +       if (nxi->v4.type && v4_dev_in_nx_info(dev, nxi))
19494 +               goto out;
19495 +#ifdef CONFIG_IPV6
19496 +       ret = 2;
19497 +       if (nxi->v6.type && v6_dev_in_nx_info(dev, nxi))
19498 +               goto out;
19499 +#endif
19500 +       ret = 0;
19501 +out:
19502 +       vxdprintk(VXD_CBIT(net, 3),
19503 +               "dev_in_nx_info(%p,%p[#%d]) = %d",
19504 +               dev, nxi, nxi ? nxi->nx_id : 0, ret);
19505 +       return ret;
19506 +}
19507 +
19508 +int ip_v4_find_src(struct net *net, struct nx_info *nxi,
19509 +       struct rtable **rp, struct flowi *fl)
19510 +{
19511 +       if (!nxi)
19512 +               return 0;
19513 +
19514 +       /* FIXME: handle lback only case */
19515 +       if (!NX_IPV4(nxi))
19516 +               return -EPERM;
19517 +
19518 +       vxdprintk(VXD_CBIT(net, 4),
19519 +               "ip_v4_find_src(%p[#%u]) " NIPQUAD_FMT " -> " NIPQUAD_FMT,
19520 +               nxi, nxi ? nxi->nx_id : 0,
19521 +               NIPQUAD(fl->fl4_src), NIPQUAD(fl->fl4_dst));
19522 +
19523 +       /* single IP is unconditional */
19524 +       if (nx_info_flags(nxi, NXF_SINGLE_IP, 0) &&
19525 +               (fl->fl4_src == INADDR_ANY))
19526 +               fl->fl4_src = nxi->v4.ip[0].s_addr;
19527 +
19528 +       if (fl->fl4_src == INADDR_ANY) {
19529 +               struct nx_addr_v4 *ptr;
19530 +               __be32 found = 0;
19531 +               int err;
19532 +
19533 +               err = __ip_route_output_key(net, rp, fl);
19534 +               if (!err) {
19535 +                       found = (*rp)->rt_src;
19536 +                       ip_rt_put(*rp);
19537 +                       vxdprintk(VXD_CBIT(net, 4),
19538 +                               "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
19539 +                               nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(found));
19540 +                       if (v4_addr_in_nx_info(nxi, found, NXA_MASK_BIND))
19541 +                               goto found;
19542 +               }
19543 +
19544 +               for (ptr = &nxi->v4; ptr; ptr = ptr->next) {
19545 +                       __be32 primary = ptr->ip[0].s_addr;
19546 +                       __be32 mask = ptr->mask.s_addr;
19547 +                       __be32 neta = primary & mask;
19548 +
19549 +                       vxdprintk(VXD_CBIT(net, 4), "ip_v4_find_src(%p[#%u]) chk: "
19550 +                               NIPQUAD_FMT "/" NIPQUAD_FMT "/" NIPQUAD_FMT,
19551 +                               nxi, nxi ? nxi->nx_id : 0, NIPQUAD(primary),
19552 +                               NIPQUAD(mask), NIPQUAD(neta));
19553 +                       if ((found & mask) != neta)
19554 +                               continue;
19555 +
19556 +                       fl->fl4_src = primary;
19557 +                       err = __ip_route_output_key(net, rp, fl);
19558 +                       vxdprintk(VXD_CBIT(net, 4),
19559 +                               "ip_v4_find_src(%p[#%u]) rok[%u]: " NIPQUAD_FMT,
19560 +                               nxi, nxi ? nxi->nx_id : 0, fl->oif, NIPQUAD(primary));
19561 +                       if (!err) {
19562 +                               found = (*rp)->rt_src;
19563 +                               ip_rt_put(*rp);
19564 +                               if (found == primary)
19565 +                                       goto found;
19566 +                       }
19567 +               }
19568 +               /* still no source ip? */
19569 +               found = ipv4_is_loopback(fl->fl4_dst)
19570 +                       ? IPI_LOOPBACK : nxi->v4.ip[0].s_addr;
19571 +       found:
19572 +               /* assign src ip to flow */
19573 +               fl->fl4_src = found;
19574 +
19575 +       } else {
19576 +               if (!v4_addr_in_nx_info(nxi, fl->fl4_src, NXA_MASK_BIND))
19577 +                       return -EPERM;
19578 +       }
19579 +
19580 +       if (nx_info_flags(nxi, NXF_LBACK_REMAP, 0)) {
19581 +               if (ipv4_is_loopback(fl->fl4_dst))
19582 +                       fl->fl4_dst = nxi->v4_lback.s_addr;
19583 +               if (ipv4_is_loopback(fl->fl4_src))
19584 +                       fl->fl4_src = nxi->v4_lback.s_addr;
19585 +       } else if (ipv4_is_loopback(fl->fl4_dst) &&
19586 +               !nx_info_flags(nxi, NXF_LBACK_ALLOW, 0))
19587 +               return -EPERM;
19588 +
19589 +       return 0;
19590 +}
19591 +
19592 +EXPORT_SYMBOL_GPL(ip_v4_find_src);
19593 +
19594 diff -NurpP --minimal linux-2.6.36/kernel/vserver/init.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/init.c
19595 --- linux-2.6.36/kernel/vserver/init.c  1970-01-01 01:00:00.000000000 +0100
19596 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/init.c    2010-10-21 13:09:36.000000000 +0200
19597 @@ -0,0 +1,45 @@
19598 +/*
19599 + *  linux/kernel/init.c
19600 + *
19601 + *  Virtual Server Init
19602 + *
19603 + *  Copyright (C) 2004-2007  Herbert Pötzl
19604 + *
19605 + *  V0.01  basic structure
19606 + *
19607 + */
19608 +
19609 +#include <linux/init.h>
19610 +
19611 +int    vserver_register_sysctl(void);
19612 +void   vserver_unregister_sysctl(void);
19613 +
19614 +
19615 +static int __init init_vserver(void)
19616 +{
19617 +       int ret = 0;
19618 +
19619 +#ifdef CONFIG_VSERVER_DEBUG
19620 +       vserver_register_sysctl();
19621 +#endif
19622 +       return ret;
19623 +}
19624 +
19625 +
19626 +static void __exit exit_vserver(void)
19627 +{
19628 +
19629 +#ifdef CONFIG_VSERVER_DEBUG
19630 +       vserver_unregister_sysctl();
19631 +#endif
19632 +       return;
19633 +}
19634 +
19635 +/* FIXME: GFP_ZONETYPES gone
19636 +long vx_slab[GFP_ZONETYPES]; */
19637 +long vx_area;
19638 +
19639 +
19640 +module_init(init_vserver);
19641 +module_exit(exit_vserver);
19642 +
19643 diff -NurpP --minimal linux-2.6.36/kernel/vserver/inode.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/inode.c
19644 --- linux-2.6.36/kernel/vserver/inode.c 1970-01-01 01:00:00.000000000 +0100
19645 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/inode.c   2010-10-21 17:40:44.000000000 +0200
19646 @@ -0,0 +1,435 @@
19647 +/*
19648 + *  linux/kernel/vserver/inode.c
19649 + *
19650 + *  Virtual Server: File System Support
19651 + *
19652 + *  Copyright (C) 2004-2007  Herbert Pötzl
19653 + *
19654 + *  V0.01  separated from vcontext V0.05
19655 + *  V0.02  moved to tag (instead of xid)
19656 + *
19657 + */
19658 +
19659 +#include <linux/tty.h>
19660 +#include <linux/proc_fs.h>
19661 +#include <linux/devpts_fs.h>
19662 +#include <linux/fs.h>
19663 +#include <linux/file.h>
19664 +#include <linux/mount.h>
19665 +#include <linux/parser.h>
19666 +#include <linux/namei.h>
19667 +#include <linux/vserver/inode.h>
19668 +#include <linux/vserver/inode_cmd.h>
19669 +#include <linux/vs_base.h>
19670 +#include <linux/vs_tag.h>
19671 +
19672 +#include <asm/uaccess.h>
19673 +
19674 +
19675 +static int __vc_get_iattr(struct inode *in, uint32_t *tag, uint32_t *flags, uint32_t *mask)
19676 +{
19677 +       struct proc_dir_entry *entry;
19678 +
19679 +       if (!in || !in->i_sb)
19680 +               return -ESRCH;
19681 +
19682 +       *flags = IATTR_TAG
19683 +               | (IS_IMMUTABLE(in) ? IATTR_IMMUTABLE : 0)
19684 +               | (IS_IXUNLINK(in) ? IATTR_IXUNLINK : 0)
19685 +               | (IS_BARRIER(in) ? IATTR_BARRIER : 0)
19686 +               | (IS_COW(in) ? IATTR_COW : 0);
19687 +       *mask = IATTR_IXUNLINK | IATTR_IMMUTABLE | IATTR_COW;
19688 +
19689 +       if (S_ISDIR(in->i_mode))
19690 +               *mask |= IATTR_BARRIER;
19691 +
19692 +       if (IS_TAGGED(in)) {
19693 +               *tag = in->i_tag;
19694 +               *mask |= IATTR_TAG;
19695 +       }
19696 +
19697 +       switch (in->i_sb->s_magic) {
19698 +       case PROC_SUPER_MAGIC:
19699 +               entry = PROC_I(in)->pde;
19700 +
19701 +               /* check for specific inodes? */
19702 +               if (entry)
19703 +                       *mask |= IATTR_FLAGS;
19704 +               if (entry)
19705 +                       *flags |= (entry->vx_flags & IATTR_FLAGS);
19706 +               else
19707 +                       *flags |= (PROC_I(in)->vx_flags & IATTR_FLAGS);
19708 +               break;
19709 +
19710 +       case DEVPTS_SUPER_MAGIC:
19711 +               *tag = in->i_tag;
19712 +               *mask |= IATTR_TAG;
19713 +               break;
19714 +
19715 +       default:
19716 +               break;
19717 +       }
19718 +       return 0;
19719 +}
19720 +
19721 +int vc_get_iattr(void __user *data)
19722 +{
19723 +       struct path path;
19724 +       struct vcmd_ctx_iattr_v1 vc_data = { .tag = -1 };
19725 +       int ret;
19726 +
19727 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19728 +               return -EFAULT;
19729 +
19730 +       ret = user_lpath(vc_data.name, &path);
19731 +       if (!ret) {
19732 +               ret = __vc_get_iattr(path.dentry->d_inode,
19733 +                       &vc_data.tag, &vc_data.flags, &vc_data.mask);
19734 +               path_put(&path);
19735 +       }
19736 +       if (ret)
19737 +               return ret;
19738 +
19739 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19740 +               ret = -EFAULT;
19741 +       return ret;
19742 +}
19743 +
19744 +#ifdef CONFIG_COMPAT
19745 +
19746 +int vc_get_iattr_x32(void __user *data)
19747 +{
19748 +       struct path path;
19749 +       struct vcmd_ctx_iattr_v1_x32 vc_data = { .tag = -1 };
19750 +       int ret;
19751 +
19752 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19753 +               return -EFAULT;
19754 +
19755 +       ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
19756 +       if (!ret) {
19757 +               ret = __vc_get_iattr(path.dentry->d_inode,
19758 +                       &vc_data.tag, &vc_data.flags, &vc_data.mask);
19759 +               path_put(&path);
19760 +       }
19761 +       if (ret)
19762 +               return ret;
19763 +
19764 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19765 +               ret = -EFAULT;
19766 +       return ret;
19767 +}
19768 +
19769 +#endif /* CONFIG_COMPAT */
19770 +
19771 +
19772 +int vc_fget_iattr(uint32_t fd, void __user *data)
19773 +{
19774 +       struct file *filp;
19775 +       struct vcmd_ctx_fiattr_v0 vc_data = { .tag = -1 };
19776 +       int ret;
19777 +
19778 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19779 +               return -EFAULT;
19780 +
19781 +       filp = fget(fd);
19782 +       if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
19783 +               return -EBADF;
19784 +
19785 +       ret = __vc_get_iattr(filp->f_dentry->d_inode,
19786 +               &vc_data.tag, &vc_data.flags, &vc_data.mask);
19787 +
19788 +       fput(filp);
19789 +
19790 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19791 +               ret = -EFAULT;
19792 +       return ret;
19793 +}
19794 +
19795 +
19796 +static int __vc_set_iattr(struct dentry *de, uint32_t *tag, uint32_t *flags, uint32_t *mask)
19797 +{
19798 +       struct inode *in = de->d_inode;
19799 +       int error = 0, is_proc = 0, has_tag = 0;
19800 +       struct iattr attr = { 0 };
19801 +
19802 +       if (!in || !in->i_sb)
19803 +               return -ESRCH;
19804 +
19805 +       is_proc = (in->i_sb->s_magic == PROC_SUPER_MAGIC);
19806 +       if ((*mask & IATTR_FLAGS) && !is_proc)
19807 +               return -EINVAL;
19808 +
19809 +       has_tag = IS_TAGGED(in) ||
19810 +               (in->i_sb->s_magic == DEVPTS_SUPER_MAGIC);
19811 +       if ((*mask & IATTR_TAG) && !has_tag)
19812 +               return -EINVAL;
19813 +
19814 +       mutex_lock(&in->i_mutex);
19815 +       if (*mask & IATTR_TAG) {
19816 +               attr.ia_tag = *tag;
19817 +               attr.ia_valid |= ATTR_TAG;
19818 +       }
19819 +
19820 +       if (*mask & IATTR_FLAGS) {
19821 +               struct proc_dir_entry *entry = PROC_I(in)->pde;
19822 +               unsigned int iflags = PROC_I(in)->vx_flags;
19823 +
19824 +               iflags = (iflags & ~(*mask & IATTR_FLAGS))
19825 +                       | (*flags & IATTR_FLAGS);
19826 +               PROC_I(in)->vx_flags = iflags;
19827 +               if (entry)
19828 +                       entry->vx_flags = iflags;
19829 +       }
19830 +
19831 +       if (*mask & (IATTR_IMMUTABLE | IATTR_IXUNLINK |
19832 +               IATTR_BARRIER | IATTR_COW)) {
19833 +               int iflags = in->i_flags;
19834 +               int vflags = in->i_vflags;
19835 +
19836 +               if (*mask & IATTR_IMMUTABLE) {
19837 +                       if (*flags & IATTR_IMMUTABLE)
19838 +                               iflags |= S_IMMUTABLE;
19839 +                       else
19840 +                               iflags &= ~S_IMMUTABLE;
19841 +               }
19842 +               if (*mask & IATTR_IXUNLINK) {
19843 +                       if (*flags & IATTR_IXUNLINK)
19844 +                               iflags |= S_IXUNLINK;
19845 +                       else
19846 +                               iflags &= ~S_IXUNLINK;
19847 +               }
19848 +               if (S_ISDIR(in->i_mode) && (*mask & IATTR_BARRIER)) {
19849 +                       if (*flags & IATTR_BARRIER)
19850 +                               vflags |= V_BARRIER;
19851 +                       else
19852 +                               vflags &= ~V_BARRIER;
19853 +               }
19854 +               if (S_ISREG(in->i_mode) && (*mask & IATTR_COW)) {
19855 +                       if (*flags & IATTR_COW)
19856 +                               vflags |= V_COW;
19857 +                       else
19858 +                               vflags &= ~V_COW;
19859 +               }
19860 +               if (in->i_op && in->i_op->sync_flags) {
19861 +                       error = in->i_op->sync_flags(in, iflags, vflags);
19862 +                       if (error)
19863 +                               goto out;
19864 +               }
19865 +       }
19866 +
19867 +       if (attr.ia_valid) {
19868 +               if (in->i_op && in->i_op->setattr)
19869 +                       error = in->i_op->setattr(de, &attr);
19870 +               else {
19871 +                       error = inode_change_ok(in, &attr);
19872 +                       if (!error) {
19873 +                               setattr_copy(in, &attr);
19874 +                               mark_inode_dirty(in);
19875 +                       }
19876 +               }
19877 +       }
19878 +
19879 +out:
19880 +       mutex_unlock(&in->i_mutex);
19881 +       return error;
19882 +}
19883 +
19884 +int vc_set_iattr(void __user *data)
19885 +{
19886 +       struct path path;
19887 +       struct vcmd_ctx_iattr_v1 vc_data;
19888 +       int ret;
19889 +
19890 +       if (!capable(CAP_LINUX_IMMUTABLE))
19891 +               return -EPERM;
19892 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19893 +               return -EFAULT;
19894 +
19895 +       ret = user_lpath(vc_data.name, &path);
19896 +       if (!ret) {
19897 +               ret = __vc_set_iattr(path.dentry,
19898 +                       &vc_data.tag, &vc_data.flags, &vc_data.mask);
19899 +               path_put(&path);
19900 +       }
19901 +
19902 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19903 +               ret = -EFAULT;
19904 +       return ret;
19905 +}
19906 +
19907 +#ifdef CONFIG_COMPAT
19908 +
19909 +int vc_set_iattr_x32(void __user *data)
19910 +{
19911 +       struct path path;
19912 +       struct vcmd_ctx_iattr_v1_x32 vc_data;
19913 +       int ret;
19914 +
19915 +       if (!capable(CAP_LINUX_IMMUTABLE))
19916 +               return -EPERM;
19917 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19918 +               return -EFAULT;
19919 +
19920 +       ret = user_lpath(compat_ptr(vc_data.name_ptr), &path);
19921 +       if (!ret) {
19922 +               ret = __vc_set_iattr(path.dentry,
19923 +                       &vc_data.tag, &vc_data.flags, &vc_data.mask);
19924 +               path_put(&path);
19925 +       }
19926 +
19927 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19928 +               ret = -EFAULT;
19929 +       return ret;
19930 +}
19931 +
19932 +#endif /* CONFIG_COMPAT */
19933 +
19934 +int vc_fset_iattr(uint32_t fd, void __user *data)
19935 +{
19936 +       struct file *filp;
19937 +       struct vcmd_ctx_fiattr_v0 vc_data;
19938 +       int ret;
19939 +
19940 +       if (!capable(CAP_LINUX_IMMUTABLE))
19941 +               return -EPERM;
19942 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
19943 +               return -EFAULT;
19944 +
19945 +       filp = fget(fd);
19946 +       if (!filp || !filp->f_dentry || !filp->f_dentry->d_inode)
19947 +               return -EBADF;
19948 +
19949 +       ret = __vc_set_iattr(filp->f_dentry, &vc_data.tag,
19950 +               &vc_data.flags, &vc_data.mask);
19951 +
19952 +       fput(filp);
19953 +
19954 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
19955 +               return -EFAULT;
19956 +       return ret;
19957 +}
19958 +
19959 +
19960 +enum { Opt_notagcheck, Opt_tag, Opt_notag, Opt_tagid, Opt_err };
19961 +
19962 +static match_table_t tokens = {
19963 +       {Opt_notagcheck, "notagcheck"},
19964 +#ifdef CONFIG_PROPAGATE
19965 +       {Opt_notag, "notag"},
19966 +       {Opt_tag, "tag"},
19967 +       {Opt_tagid, "tagid=%u"},
19968 +#endif
19969 +       {Opt_err, NULL}
19970 +};
19971 +
19972 +
19973 +static void __dx_parse_remove(char *string, char *opt)
19974 +{
19975 +       char *p = strstr(string, opt);
19976 +       char *q = p;
19977 +
19978 +       if (p) {
19979 +               while (*q != '\0' && *q != ',')
19980 +                       q++;
19981 +               while (*q)
19982 +                       *p++ = *q++;
19983 +               while (*p)
19984 +                       *p++ = '\0';
19985 +       }
19986 +}
19987 +
19988 +int dx_parse_tag(char *string, tag_t *tag, int remove, int *mnt_flags,
19989 +                unsigned long *flags)
19990 +{
19991 +       int set = 0;
19992 +       substring_t args[MAX_OPT_ARGS];
19993 +       int token, option = 0;
19994 +       char *s, *p, *opts;
19995 +
19996 +       if (!string)
19997 +               return 0;
19998 +       s = kstrdup(string, GFP_KERNEL | GFP_ATOMIC);
19999 +       if (!s)
20000 +               return 0;
20001 +
20002 +       opts = s;
20003 +       while ((p = strsep(&opts, ",")) != NULL) {
20004 +               token = match_token(p, tokens, args);
20005 +
20006 +               vxdprintk(VXD_CBIT(tag, 7),
20007 +                       "dx_parse_tag(" VS_Q("%s") "): %d:#%d",
20008 +                       p, token, option);
20009 +
20010 +               switch (token) {
20011 +#ifdef CONFIG_PROPAGATE
20012 +               case Opt_tag:
20013 +                       if (tag)
20014 +                               *tag = 0;
20015 +                       if (remove)
20016 +                               __dx_parse_remove(s, "tag");
20017 +                       *mnt_flags |= MNT_TAGID;
20018 +                       set |= MNT_TAGID;
20019 +                       break;
20020 +               case Opt_notag:
20021 +                       if (remove)
20022 +                               __dx_parse_remove(s, "notag");
20023 +                       *mnt_flags |= MNT_NOTAG;
20024 +                       set |= MNT_NOTAG;
20025 +                       break;
20026 +               case Opt_tagid:
20027 +                       if (tag && !match_int(args, &option))
20028 +                               *tag = option;
20029 +                       if (remove)
20030 +                               __dx_parse_remove(s, "tagid");
20031 +                       *mnt_flags |= MNT_TAGID;
20032 +                       set |= MNT_TAGID;
20033 +                       break;
20034 +#endif
20035 +               case Opt_notagcheck:
20036 +                       if (remove)
20037 +                               __dx_parse_remove(s, "notagcheck");
20038 +                       *flags |= MS_NOTAGCHECK;
20039 +                       set |= MS_NOTAGCHECK;
20040 +                       break;
20041 +               }
20042 +       }
20043 +       if (set)
20044 +               strcpy(string, s);
20045 +       kfree(s);
20046 +       return set;
20047 +}
20048 +
20049 +#ifdef CONFIG_PROPAGATE
20050 +
20051 +void __dx_propagate_tag(struct nameidata *nd, struct inode *inode)
20052 +{
20053 +       tag_t new_tag = 0;
20054 +       struct vfsmount *mnt;
20055 +       int propagate;
20056 +
20057 +       if (!nd)
20058 +               return;
20059 +       mnt = nd->path.mnt;
20060 +       if (!mnt)
20061 +               return;
20062 +
20063 +       propagate = (mnt->mnt_flags & MNT_TAGID);
20064 +       if (propagate)
20065 +               new_tag = mnt->mnt_tag;
20066 +
20067 +       vxdprintk(VXD_CBIT(tag, 7),
20068 +               "dx_propagate_tag(%p[#%lu.%d]): %d,%d",
20069 +               inode, inode->i_ino, inode->i_tag,
20070 +               new_tag, (propagate) ? 1 : 0);
20071 +
20072 +       if (propagate)
20073 +               inode->i_tag = new_tag;
20074 +}
20075 +
20076 +#include <linux/module.h>
20077 +
20078 +EXPORT_SYMBOL_GPL(__dx_propagate_tag);
20079 +
20080 +#endif /* CONFIG_PROPAGATE */
20081 +
20082 diff -NurpP --minimal linux-2.6.36/kernel/vserver/limit.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit.c
20083 --- linux-2.6.36/kernel/vserver/limit.c 1970-01-01 01:00:00.000000000 +0100
20084 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit.c   2010-10-21 13:09:36.000000000 +0200
20085 @@ -0,0 +1,354 @@
20086 +/*
20087 + *  linux/kernel/vserver/limit.c
20088 + *
20089 + *  Virtual Server: Context Limits
20090 + *
20091 + *  Copyright (C) 2004-2010  Herbert Pötzl
20092 + *
20093 + *  V0.01  broken out from vcontext V0.05
20094 + *  V0.02  changed vcmds to vxi arg
20095 + *  V0.03  added memory cgroup support
20096 + *
20097 + */
20098 +
20099 +#include <linux/sched.h>
20100 +#include <linux/module.h>
20101 +#include <linux/memcontrol.h>
20102 +#include <linux/res_counter.h>
20103 +#include <linux/vs_limit.h>
20104 +#include <linux/vserver/limit.h>
20105 +#include <linux/vserver/limit_cmd.h>
20106 +
20107 +#include <asm/uaccess.h>
20108 +
20109 +
20110 +const char *vlimit_name[NUM_LIMITS] = {
20111 +#ifdef CONFIG_VSERVER_LEGACY_MEM
20112 +       [RLIMIT_RSS]            = "RSS",
20113 +       [RLIMIT_AS]             = "VM",
20114 +#endif /* CONFIG_VSERVER_LEGACY_MEM */
20115 +       [RLIMIT_CPU]            = "CPU",
20116 +       [RLIMIT_NPROC]          = "NPROC",
20117 +       [RLIMIT_NOFILE]         = "NOFILE",
20118 +       [RLIMIT_LOCKS]          = "LOCKS",
20119 +       [RLIMIT_SIGPENDING]     = "SIGP",
20120 +       [RLIMIT_MSGQUEUE]       = "MSGQ",
20121 +
20122 +       [VLIMIT_NSOCK]          = "NSOCK",
20123 +       [VLIMIT_OPENFD]         = "OPENFD",
20124 +       [VLIMIT_SHMEM]          = "SHMEM",
20125 +       [VLIMIT_DENTRY]         = "DENTRY",
20126 +};
20127 +
20128 +EXPORT_SYMBOL_GPL(vlimit_name);
20129 +
20130 +#define MASK_ENTRY(x)  (1 << (x))
20131 +
20132 +const struct vcmd_ctx_rlimit_mask_v0 vlimit_mask = {
20133 +               /* minimum */
20134 +       0
20135 +       ,       /* softlimit */
20136 +#ifdef CONFIG_VSERVER_LEGACY_MEM
20137 +       MASK_ENTRY( RLIMIT_RSS          ) |
20138 +#endif /* CONFIG_VSERVER_LEGACY_MEM */
20139 +       0
20140 +       ,       /* maximum */
20141 +#ifdef CONFIG_VSERVER_LEGACY_MEM
20142 +       MASK_ENTRY( RLIMIT_RSS          ) |
20143 +       MASK_ENTRY( RLIMIT_AS           ) |
20144 +#endif /* CONFIG_VSERVER_LEGACY_MEM */
20145 +       MASK_ENTRY( RLIMIT_NPROC        ) |
20146 +       MASK_ENTRY( RLIMIT_NOFILE       ) |
20147 +       MASK_ENTRY( RLIMIT_LOCKS        ) |
20148 +       MASK_ENTRY( RLIMIT_MSGQUEUE     ) |
20149 +
20150 +       MASK_ENTRY( VLIMIT_NSOCK        ) |
20151 +       MASK_ENTRY( VLIMIT_OPENFD       ) |
20152 +       MASK_ENTRY( VLIMIT_SHMEM        ) |
20153 +       MASK_ENTRY( VLIMIT_DENTRY       ) |
20154 +       0
20155 +};
20156 +               /* accounting only */
20157 +uint32_t account_mask =
20158 +       MASK_ENTRY( VLIMIT_SEMARY       ) |
20159 +       MASK_ENTRY( VLIMIT_NSEMS        ) |
20160 +       MASK_ENTRY( VLIMIT_MAPPED       ) |
20161 +       0;
20162 +
20163 +
20164 +static int is_valid_vlimit(int id)
20165 +{
20166 +       uint32_t mask = vlimit_mask.minimum |
20167 +               vlimit_mask.softlimit | vlimit_mask.maximum;
20168 +       return mask & (1 << id);
20169 +}
20170 +
20171 +static int is_accounted_vlimit(int id)
20172 +{
20173 +       if (is_valid_vlimit(id))
20174 +               return 1;
20175 +       return account_mask & (1 << id);
20176 +}
20177 +
20178 +
20179 +static inline uint64_t vc_get_soft(struct vx_info *vxi, int id)
20180 +{
20181 +       rlim_t limit = __rlim_soft(&vxi->limit, id);
20182 +       return VX_VLIM(limit);
20183 +}
20184 +
20185 +static inline uint64_t vc_get_hard(struct vx_info *vxi, int id)
20186 +{
20187 +       rlim_t limit = __rlim_hard(&vxi->limit, id);
20188 +       return VX_VLIM(limit);
20189 +}
20190 +
20191 +static int do_get_rlimit(struct vx_info *vxi, uint32_t id,
20192 +       uint64_t *minimum, uint64_t *softlimit, uint64_t *maximum)
20193 +{
20194 +       if (!is_valid_vlimit(id))
20195 +               return -EINVAL;
20196 +
20197 +       if (minimum)
20198 +               *minimum = CRLIM_UNSET;
20199 +       if (softlimit)
20200 +               *softlimit = vc_get_soft(vxi, id);
20201 +       if (maximum)
20202 +               *maximum = vc_get_hard(vxi, id);
20203 +       return 0;
20204 +}
20205 +
20206 +int vc_get_rlimit(struct vx_info *vxi, void __user *data)
20207 +{
20208 +       struct vcmd_ctx_rlimit_v0 vc_data;
20209 +       int ret;
20210 +
20211 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20212 +               return -EFAULT;
20213 +
20214 +       ret = do_get_rlimit(vxi, vc_data.id,
20215 +               &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
20216 +       if (ret)
20217 +               return ret;
20218 +
20219 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20220 +               return -EFAULT;
20221 +       return 0;
20222 +}
20223 +
20224 +static int do_set_rlimit(struct vx_info *vxi, uint32_t id,
20225 +       uint64_t minimum, uint64_t softlimit, uint64_t maximum)
20226 +{
20227 +       if (!is_valid_vlimit(id))
20228 +               return -EINVAL;
20229 +
20230 +       if (maximum != CRLIM_KEEP)
20231 +               __rlim_hard(&vxi->limit, id) = VX_RLIM(maximum);
20232 +       if (softlimit != CRLIM_KEEP)
20233 +               __rlim_soft(&vxi->limit, id) = VX_RLIM(softlimit);
20234 +
20235 +       /* clamp soft limit */
20236 +       if (__rlim_soft(&vxi->limit, id) > __rlim_hard(&vxi->limit, id))
20237 +               __rlim_soft(&vxi->limit, id) = __rlim_hard(&vxi->limit, id);
20238 +
20239 +       return 0;
20240 +}
20241 +
20242 +int vc_set_rlimit(struct vx_info *vxi, void __user *data)
20243 +{
20244 +       struct vcmd_ctx_rlimit_v0 vc_data;
20245 +
20246 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20247 +               return -EFAULT;
20248 +
20249 +       return do_set_rlimit(vxi, vc_data.id,
20250 +               vc_data.minimum, vc_data.softlimit, vc_data.maximum);
20251 +}
20252 +
20253 +#ifdef CONFIG_IA32_EMULATION
20254 +
20255 +int vc_set_rlimit_x32(struct vx_info *vxi, void __user *data)
20256 +{
20257 +       struct vcmd_ctx_rlimit_v0_x32 vc_data;
20258 +
20259 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20260 +               return -EFAULT;
20261 +
20262 +       return do_set_rlimit(vxi, vc_data.id,
20263 +               vc_data.minimum, vc_data.softlimit, vc_data.maximum);
20264 +}
20265 +
20266 +int vc_get_rlimit_x32(struct vx_info *vxi, void __user *data)
20267 +{
20268 +       struct vcmd_ctx_rlimit_v0_x32 vc_data;
20269 +       int ret;
20270 +
20271 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20272 +               return -EFAULT;
20273 +
20274 +       ret = do_get_rlimit(vxi, vc_data.id,
20275 +               &vc_data.minimum, &vc_data.softlimit, &vc_data.maximum);
20276 +       if (ret)
20277 +               return ret;
20278 +
20279 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20280 +               return -EFAULT;
20281 +       return 0;
20282 +}
20283 +
20284 +#endif /* CONFIG_IA32_EMULATION */
20285 +
20286 +
20287 +int vc_get_rlimit_mask(uint32_t id, void __user *data)
20288 +{
20289 +       if (copy_to_user(data, &vlimit_mask, sizeof(vlimit_mask)))
20290 +               return -EFAULT;
20291 +       return 0;
20292 +}
20293 +
20294 +
20295 +static inline void vx_reset_hits(struct _vx_limit *limit)
20296 +{
20297 +       int lim;
20298 +
20299 +       for (lim = 0; lim < NUM_LIMITS; lim++) {
20300 +               atomic_set(&__rlim_lhit(limit, lim), 0);
20301 +       }
20302 +}
20303 +
20304 +int vc_reset_hits(struct vx_info *vxi, void __user *data)
20305 +{
20306 +       vx_reset_hits(&vxi->limit);
20307 +       return 0;
20308 +}
20309 +
20310 +static inline void vx_reset_minmax(struct _vx_limit *limit)
20311 +{
20312 +       rlim_t value;
20313 +       int lim;
20314 +
20315 +       for (lim = 0; lim < NUM_LIMITS; lim++) {
20316 +               value = __rlim_get(limit, lim);
20317 +               __rlim_rmax(limit, lim) = value;
20318 +               __rlim_rmin(limit, lim) = value;
20319 +       }
20320 +}
20321 +
20322 +int vc_reset_minmax(struct vx_info *vxi, void __user *data)
20323 +{
20324 +       vx_reset_minmax(&vxi->limit);
20325 +       return 0;
20326 +}
20327 +
20328 +
20329 +int vc_rlimit_stat(struct vx_info *vxi, void __user *data)
20330 +{
20331 +       struct vcmd_rlimit_stat_v0 vc_data;
20332 +       struct _vx_limit *limit = &vxi->limit;
20333 +       int id;
20334 +
20335 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
20336 +               return -EFAULT;
20337 +
20338 +       id = vc_data.id;
20339 +       if (!is_accounted_vlimit(id))
20340 +               return -EINVAL;
20341 +
20342 +       vx_limit_fixup(limit, id);
20343 +       vc_data.hits = atomic_read(&__rlim_lhit(limit, id));
20344 +       vc_data.value = __rlim_get(limit, id);
20345 +       vc_data.minimum = __rlim_rmin(limit, id);
20346 +       vc_data.maximum = __rlim_rmax(limit, id);
20347 +
20348 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
20349 +               return -EFAULT;
20350 +       return 0;
20351 +}
20352 +
20353 +
20354 +void vx_vsi_meminfo(struct sysinfo *val)
20355 +{
20356 +#ifdef CONFIG_CGROUP_MEM_RES_CTLR
20357 +       struct mem_cgroup *mcg = mem_cgroup_from_task(current);
20358 +       u64 res_limit, res_usage;
20359 +
20360 +       if (!mcg)
20361 +               return;
20362 +
20363 +       res_limit = mem_cgroup_res_read_u64(mcg, RES_LIMIT);
20364 +       res_usage = mem_cgroup_res_read_u64(mcg, RES_USAGE);
20365 +
20366 +       if (res_limit != RESOURCE_MAX)
20367 +               val->totalram = (res_limit >> PAGE_SHIFT);
20368 +       val->freeram = val->totalram - (res_usage >> PAGE_SHIFT);
20369 +       val->bufferram = 0;
20370 +       val->totalhigh = 0;
20371 +       val->freehigh = 0;
20372 +#endif /* CONFIG_CGROUP_MEM_RES_CTLR */
20373 +       return;
20374 +}
20375 +
20376 +void vx_vsi_swapinfo(struct sysinfo *val)
20377 +{
20378 +#ifdef CONFIG_CGROUP_MEM_RES_CTLR
20379 +#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
20380 +       struct mem_cgroup *mcg = mem_cgroup_from_task(current);
20381 +       u64 res_limit, res_usage, memsw_limit, memsw_usage;
20382 +       s64 swap_limit, swap_usage;
20383 +
20384 +       if (!mcg)
20385 +               return;
20386 +
20387 +       res_limit = mem_cgroup_res_read_u64(mcg, RES_LIMIT);
20388 +       res_usage = mem_cgroup_res_read_u64(mcg, RES_USAGE);
20389 +       memsw_limit = mem_cgroup_memsw_read_u64(mcg, RES_LIMIT);
20390 +       memsw_usage = mem_cgroup_memsw_read_u64(mcg, RES_USAGE);
20391 +
20392 +       if (res_limit == RESOURCE_MAX)
20393 +               return;
20394 +
20395 +       swap_limit = memsw_limit - res_limit;
20396 +       if (memsw_limit != RESOURCE_MAX)
20397 +               val->totalswap = swap_limit >> PAGE_SHIFT;
20398 +
20399 +       swap_usage = memsw_usage - res_usage;
20400 +       val->freeswap = (swap_usage < swap_limit) ?
20401 +               val->totalswap - (swap_usage >> PAGE_SHIFT) : 0;
20402 +#else  /* !CONFIG_CGROUP_MEM_RES_CTLR_SWAP */
20403 +       val->totalswap = 0;
20404 +       val->freeswap = 0;
20405 +#endif /* !CONFIG_CGROUP_MEM_RES_CTLR_SWAP */
20406 +#endif /* CONFIG_CGROUP_MEM_RES_CTLR */
20407 +       return;
20408 +}
20409 +
20410 +long vx_vsi_cached(struct sysinfo *val)
20411 +{
20412 +#ifdef CONFIG_CGROUP_MEM_RES_CTLR
20413 +       struct mem_cgroup *mcg = mem_cgroup_from_task(current);
20414 +
20415 +       return mem_cgroup_stat_read_cache(mcg);
20416 +#else
20417 +       return 0;
20418 +#endif
20419 +}
20420 +
20421 +
20422 +unsigned long vx_badness(struct task_struct *task, struct mm_struct *mm)
20423 +{
20424 +       struct vx_info *vxi = mm->mm_vx_info;
20425 +       unsigned long points;
20426 +       rlim_t v, w;
20427 +
20428 +       if (!vxi)
20429 +               return 0;
20430 +
20431 +       points = vxi->vx_badness_bias;
20432 +
20433 +       v = __vx_cres_array_fixup(&vxi->limit, VLA_RSS);
20434 +       w = __rlim_soft(&vxi->limit, RLIMIT_RSS);
20435 +       points += (v > w) ? (v - w) : 0;
20436 +
20437 +       return points;
20438 +}
20439 +
20440 diff -NurpP --minimal linux-2.6.36/kernel/vserver/limit_init.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit_init.h
20441 --- linux-2.6.36/kernel/vserver/limit_init.h    1970-01-01 01:00:00.000000000 +0100
20442 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit_init.h      2010-10-21 13:09:36.000000000 +0200
20443 @@ -0,0 +1,31 @@
20444 +
20445 +
20446 +static inline void vx_info_init_limit(struct _vx_limit *limit)
20447 +{
20448 +       int lim;
20449 +
20450 +       for (lim = 0; lim < NUM_LIMITS; lim++) {
20451 +               __rlim_soft(limit, lim) = RLIM_INFINITY;
20452 +               __rlim_hard(limit, lim) = RLIM_INFINITY;
20453 +               __rlim_set(limit, lim, 0);
20454 +               atomic_set(&__rlim_lhit(limit, lim), 0);
20455 +               __rlim_rmin(limit, lim) = 0;
20456 +               __rlim_rmax(limit, lim) = 0;
20457 +       }
20458 +}
20459 +
20460 +static inline void vx_info_exit_limit(struct _vx_limit *limit)
20461 +{
20462 +       rlim_t value;
20463 +       int lim;
20464 +
20465 +       for (lim = 0; lim < NUM_LIMITS; lim++) {
20466 +               if ((1 << lim) & VLIM_NOCHECK)
20467 +                       continue;
20468 +               value = __rlim_get(limit, lim);
20469 +               vxwprintk_xid(value,
20470 +                       "!!! limit: %p[%s,%d] = %ld on exit.",
20471 +                       limit, vlimit_name[lim], lim, (long)value);
20472 +       }
20473 +}
20474 +
20475 diff -NurpP --minimal linux-2.6.36/kernel/vserver/limit_proc.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit_proc.h
20476 --- linux-2.6.36/kernel/vserver/limit_proc.h    1970-01-01 01:00:00.000000000 +0100
20477 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/limit_proc.h      2010-10-21 13:09:36.000000000 +0200
20478 @@ -0,0 +1,57 @@
20479 +#ifndef _VX_LIMIT_PROC_H
20480 +#define _VX_LIMIT_PROC_H
20481 +
20482 +#include <linux/vserver/limit_int.h>
20483 +
20484 +
20485 +#define VX_LIMIT_FMT   ":\t%8ld\t%8ld/%8ld\t%8lld/%8lld\t%6d\n"
20486 +#define VX_LIMIT_TOP   \
20487 +       "Limit\t current\t     min/max\t\t    soft/hard\t\thits\n"
20488 +
20489 +#define VX_LIMIT_ARG(r)                                \
20490 +       (unsigned long)__rlim_get(limit, r),    \
20491 +       (unsigned long)__rlim_rmin(limit, r),   \
20492 +       (unsigned long)__rlim_rmax(limit, r),   \
20493 +       VX_VLIM(__rlim_soft(limit, r)),         \
20494 +       VX_VLIM(__rlim_hard(limit, r)),         \
20495 +       atomic_read(&__rlim_lhit(limit, r))
20496 +
20497 +static inline int vx_info_proc_limit(struct _vx_limit *limit, char *buffer)
20498 +{
20499 +       vx_limit_fixup(limit, -1);
20500 +       return sprintf(buffer, VX_LIMIT_TOP
20501 +               "PROC"  VX_LIMIT_FMT
20502 +               "VM"    VX_LIMIT_FMT
20503 +               "VML"   VX_LIMIT_FMT
20504 +               "RSS"   VX_LIMIT_FMT
20505 +               "ANON"  VX_LIMIT_FMT
20506 +               "RMAP"  VX_LIMIT_FMT
20507 +               "FILES" VX_LIMIT_FMT
20508 +               "OFD"   VX_LIMIT_FMT
20509 +               "LOCKS" VX_LIMIT_FMT
20510 +               "SOCK"  VX_LIMIT_FMT
20511 +               "MSGQ"  VX_LIMIT_FMT
20512 +               "SHM"   VX_LIMIT_FMT
20513 +               "SEMA"  VX_LIMIT_FMT
20514 +               "SEMS"  VX_LIMIT_FMT
20515 +               "DENT"  VX_LIMIT_FMT,
20516 +               VX_LIMIT_ARG(RLIMIT_NPROC),
20517 +               VX_LIMIT_ARG(RLIMIT_AS),
20518 +               VX_LIMIT_ARG(RLIMIT_MEMLOCK),
20519 +               VX_LIMIT_ARG(RLIMIT_RSS),
20520 +               VX_LIMIT_ARG(VLIMIT_ANON),
20521 +               VX_LIMIT_ARG(VLIMIT_MAPPED),
20522 +               VX_LIMIT_ARG(RLIMIT_NOFILE),
20523 +               VX_LIMIT_ARG(VLIMIT_OPENFD),
20524 +               VX_LIMIT_ARG(RLIMIT_LOCKS),
20525 +               VX_LIMIT_ARG(VLIMIT_NSOCK),
20526 +               VX_LIMIT_ARG(RLIMIT_MSGQUEUE),
20527 +               VX_LIMIT_ARG(VLIMIT_SHMEM),
20528 +               VX_LIMIT_ARG(VLIMIT_SEMARY),
20529 +               VX_LIMIT_ARG(VLIMIT_NSEMS),
20530 +               VX_LIMIT_ARG(VLIMIT_DENTRY));
20531 +}
20532 +
20533 +#endif /* _VX_LIMIT_PROC_H */
20534 +
20535 +
20536 diff -NurpP --minimal linux-2.6.36/kernel/vserver/network.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/network.c
20537 --- linux-2.6.36/kernel/vserver/network.c       1970-01-01 01:00:00.000000000 +0100
20538 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/network.c 2010-10-21 13:09:36.000000000 +0200
20539 @@ -0,0 +1,864 @@
20540 +/*
20541 + *  linux/kernel/vserver/network.c
20542 + *
20543 + *  Virtual Server: Network Support
20544 + *
20545 + *  Copyright (C) 2003-2007  Herbert Pötzl
20546 + *
20547 + *  V0.01  broken out from vcontext V0.05
20548 + *  V0.02  cleaned up implementation
20549 + *  V0.03  added equiv nx commands
20550 + *  V0.04  switch to RCU based hash
20551 + *  V0.05  and back to locking again
20552 + *  V0.06  changed vcmds to nxi arg
20553 + *  V0.07  have __create claim() the nxi
20554 + *
20555 + */
20556 +
20557 +#include <linux/err.h>
20558 +#include <linux/slab.h>
20559 +#include <linux/rcupdate.h>
20560 +
20561 +#include <linux/vs_network.h>
20562 +#include <linux/vs_pid.h>
20563 +#include <linux/vserver/network_cmd.h>
20564 +
20565 +
20566 +atomic_t nx_global_ctotal      = ATOMIC_INIT(0);
20567 +atomic_t nx_global_cactive     = ATOMIC_INIT(0);
20568 +
20569 +static struct kmem_cache *nx_addr_v4_cachep = NULL;
20570 +static struct kmem_cache *nx_addr_v6_cachep = NULL;
20571 +
20572 +
20573 +static int __init init_network(void)
20574 +{
20575 +       nx_addr_v4_cachep = kmem_cache_create("nx_v4_addr_cache",
20576 +               sizeof(struct nx_addr_v4), 0,
20577 +               SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
20578 +       nx_addr_v6_cachep = kmem_cache_create("nx_v6_addr_cache",
20579 +               sizeof(struct nx_addr_v6), 0,
20580 +               SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
20581 +       return 0;
20582 +}
20583 +
20584 +
20585 +/*     __alloc_nx_addr_v4()                                    */
20586 +
20587 +static inline struct nx_addr_v4 *__alloc_nx_addr_v4(void)
20588 +{
20589 +       struct nx_addr_v4 *nxa = kmem_cache_alloc(
20590 +               nx_addr_v4_cachep, GFP_KERNEL);
20591 +
20592 +       if (!IS_ERR(nxa))
20593 +               memset(nxa, 0, sizeof(*nxa));
20594 +       return nxa;
20595 +}
20596 +
20597 +/*     __dealloc_nx_addr_v4()                                  */
20598 +
20599 +static inline void __dealloc_nx_addr_v4(struct nx_addr_v4 *nxa)
20600 +{
20601 +       kmem_cache_free(nx_addr_v4_cachep, nxa);
20602 +}
20603 +
20604 +/*     __dealloc_nx_addr_v4_all()                              */
20605 +
20606 +static inline void __dealloc_nx_addr_v4_all(struct nx_addr_v4 *nxa)
20607 +{
20608 +       while (nxa) {
20609 +               struct nx_addr_v4 *next = nxa->next;
20610 +
20611 +               __dealloc_nx_addr_v4(nxa);
20612 +               nxa = next;
20613 +       }
20614 +}
20615 +
20616 +
20617 +#ifdef CONFIG_IPV6
20618 +
20619 +/*     __alloc_nx_addr_v6()                                    */
20620 +
20621 +static inline struct nx_addr_v6 *__alloc_nx_addr_v6(void)
20622 +{
20623 +       struct nx_addr_v6 *nxa = kmem_cache_alloc(
20624 +               nx_addr_v6_cachep, GFP_KERNEL);
20625 +
20626 +       if (!IS_ERR(nxa))
20627 +               memset(nxa, 0, sizeof(*nxa));
20628 +       return nxa;
20629 +}
20630 +
20631 +/*     __dealloc_nx_addr_v6()                                  */
20632 +
20633 +static inline void __dealloc_nx_addr_v6(struct nx_addr_v6 *nxa)
20634 +{
20635 +       kmem_cache_free(nx_addr_v6_cachep, nxa);
20636 +}
20637 +
20638 +/*     __dealloc_nx_addr_v6_all()                              */
20639 +
20640 +static inline void __dealloc_nx_addr_v6_all(struct nx_addr_v6 *nxa)
20641 +{
20642 +       while (nxa) {
20643 +               struct nx_addr_v6 *next = nxa->next;
20644 +
20645 +               __dealloc_nx_addr_v6(nxa);
20646 +               nxa = next;
20647 +       }
20648 +}
20649 +
20650 +#endif /* CONFIG_IPV6 */
20651 +
20652 +/*     __alloc_nx_info()
20653 +
20654 +       * allocate an initialized nx_info struct
20655 +       * doesn't make it visible (hash)                        */
20656 +
20657 +static struct nx_info *__alloc_nx_info(nid_t nid)
20658 +{
20659 +       struct nx_info *new = NULL;
20660 +
20661 +       vxdprintk(VXD_CBIT(nid, 1), "alloc_nx_info(%d)*", nid);
20662 +
20663 +       /* would this benefit from a slab cache? */
20664 +       new = kmalloc(sizeof(struct nx_info), GFP_KERNEL);
20665 +       if (!new)
20666 +               return 0;
20667 +
20668 +       memset(new, 0, sizeof(struct nx_info));
20669 +       new->nx_id = nid;
20670 +       INIT_HLIST_NODE(&new->nx_hlist);
20671 +       atomic_set(&new->nx_usecnt, 0);
20672 +       atomic_set(&new->nx_tasks, 0);
20673 +       new->nx_state = 0;
20674 +
20675 +       new->nx_flags = NXF_INIT_SET;
20676 +
20677 +       /* rest of init goes here */
20678 +
20679 +       new->v4_lback.s_addr = htonl(INADDR_LOOPBACK);
20680 +       new->v4_bcast.s_addr = htonl(INADDR_BROADCAST);
20681 +
20682 +       vxdprintk(VXD_CBIT(nid, 0),
20683 +               "alloc_nx_info(%d) = %p", nid, new);
20684 +       atomic_inc(&nx_global_ctotal);
20685 +       return new;
20686 +}
20687 +
20688 +/*     __dealloc_nx_info()
20689 +
20690 +       * final disposal of nx_info                             */
20691 +
20692 +static void __dealloc_nx_info(struct nx_info *nxi)
20693 +{
20694 +       vxdprintk(VXD_CBIT(nid, 0),
20695 +               "dealloc_nx_info(%p)", nxi);
20696 +
20697 +       nxi->nx_hlist.next = LIST_POISON1;
20698 +       nxi->nx_id = -1;
20699 +
20700 +       BUG_ON(atomic_read(&nxi->nx_usecnt));
20701 +       BUG_ON(atomic_read(&nxi->nx_tasks));
20702 +
20703 +       __dealloc_nx_addr_v4_all(nxi->v4.next);
20704 +
20705 +       nxi->nx_state |= NXS_RELEASED;
20706 +       kfree(nxi);
20707 +       atomic_dec(&nx_global_ctotal);
20708 +}
20709 +
20710 +static void __shutdown_nx_info(struct nx_info *nxi)
20711 +{
20712 +       nxi->nx_state |= NXS_SHUTDOWN;
20713 +       vs_net_change(nxi, VSC_NETDOWN);
20714 +}
20715 +
20716 +/*     exported stuff                                          */
20717 +
20718 +void free_nx_info(struct nx_info *nxi)
20719 +{
20720 +       /* context shutdown is mandatory */
20721 +       BUG_ON(nxi->nx_state != NXS_SHUTDOWN);
20722 +
20723 +       /* context must not be hashed */
20724 +       BUG_ON(nxi->nx_state & NXS_HASHED);
20725 +
20726 +       BUG_ON(atomic_read(&nxi->nx_usecnt));
20727 +       BUG_ON(atomic_read(&nxi->nx_tasks));
20728 +
20729 +       __dealloc_nx_info(nxi);
20730 +}
20731 +
20732 +
20733 +void __nx_set_lback(struct nx_info *nxi)
20734 +{
20735 +       int nid = nxi->nx_id;
20736 +       __be32 lback = htonl(INADDR_LOOPBACK ^ ((nid & 0xFFFF) << 8));
20737 +
20738 +       nxi->v4_lback.s_addr = lback;
20739 +}
20740 +
20741 +extern int __nx_inet_add_lback(__be32 addr);
20742 +extern int __nx_inet_del_lback(__be32 addr);
20743 +
20744 +
20745 +/*     hash table for nx_info hash */
20746 +
20747 +#define NX_HASH_SIZE   13
20748 +
20749 +struct hlist_head nx_info_hash[NX_HASH_SIZE];
20750 +
20751 +static spinlock_t nx_info_hash_lock = SPIN_LOCK_UNLOCKED;
20752 +
20753 +
20754 +static inline unsigned int __hashval(nid_t nid)
20755 +{
20756 +       return (nid % NX_HASH_SIZE);
20757 +}
20758 +
20759 +
20760 +
20761 +/*     __hash_nx_info()
20762 +
20763 +       * add the nxi to the global hash table
20764 +       * requires the hash_lock to be held                     */
20765 +
20766 +static inline void __hash_nx_info(struct nx_info *nxi)
20767 +{
20768 +       struct hlist_head *head;
20769 +
20770 +       vxd_assert_lock(&nx_info_hash_lock);
20771 +       vxdprintk(VXD_CBIT(nid, 4),
20772 +               "__hash_nx_info: %p[#%d]", nxi, nxi->nx_id);
20773 +
20774 +       /* context must not be hashed */
20775 +       BUG_ON(nx_info_state(nxi, NXS_HASHED));
20776 +
20777 +       nxi->nx_state |= NXS_HASHED;
20778 +       head = &nx_info_hash[__hashval(nxi->nx_id)];
20779 +       hlist_add_head(&nxi->nx_hlist, head);
20780 +       atomic_inc(&nx_global_cactive);
20781 +}
20782 +
20783 +/*     __unhash_nx_info()
20784 +
20785 +       * remove the nxi from the global hash table
20786 +       * requires the hash_lock to be held                     */
20787 +
20788 +static inline void __unhash_nx_info(struct nx_info *nxi)
20789 +{
20790 +       vxd_assert_lock(&nx_info_hash_lock);
20791 +       vxdprintk(VXD_CBIT(nid, 4),
20792 +               "__unhash_nx_info: %p[#%d.%d.%d]", nxi, nxi->nx_id,
20793 +               atomic_read(&nxi->nx_usecnt), atomic_read(&nxi->nx_tasks));
20794 +
20795 +       /* context must be hashed */
20796 +       BUG_ON(!nx_info_state(nxi, NXS_HASHED));
20797 +       /* but without tasks */
20798 +       BUG_ON(atomic_read(&nxi->nx_tasks));
20799 +
20800 +       nxi->nx_state &= ~NXS_HASHED;
20801 +       hlist_del(&nxi->nx_hlist);
20802 +       atomic_dec(&nx_global_cactive);
20803 +}
20804 +
20805 +
20806 +/*     __lookup_nx_info()
20807 +
20808 +       * requires the hash_lock to be held
20809 +       * doesn't increment the nx_refcnt                       */
20810 +
20811 +static inline struct nx_info *__lookup_nx_info(nid_t nid)
20812 +{
20813 +       struct hlist_head *head = &nx_info_hash[__hashval(nid)];
20814 +       struct hlist_node *pos;
20815 +       struct nx_info *nxi;
20816 +
20817 +       vxd_assert_lock(&nx_info_hash_lock);
20818 +       hlist_for_each(pos, head) {
20819 +               nxi = hlist_entry(pos, struct nx_info, nx_hlist);
20820 +
20821 +               if (nxi->nx_id == nid)
20822 +                       goto found;
20823 +       }
20824 +       nxi = NULL;
20825 +found:
20826 +       vxdprintk(VXD_CBIT(nid, 0),
20827 +               "__lookup_nx_info(#%u): %p[#%u]",
20828 +               nid, nxi, nxi ? nxi->nx_id : 0);
20829 +       return nxi;
20830 +}
20831 +
20832 +
20833 +/*     __create_nx_info()
20834 +
20835 +       * create the requested context
20836 +       * get(), claim() and hash it                            */
20837 +
20838 +static struct nx_info *__create_nx_info(int id)
20839 +{
20840 +       struct nx_info *new, *nxi = NULL;
20841 +
20842 +       vxdprintk(VXD_CBIT(nid, 1), "create_nx_info(%d)*", id);
20843 +
20844 +       if (!(new = __alloc_nx_info(id)))
20845 +               return ERR_PTR(-ENOMEM);
20846 +
20847 +       /* required to make dynamic xids unique */
20848 +       spin_lock(&nx_info_hash_lock);
20849 +
20850 +       /* static context requested */
20851 +       if ((nxi = __lookup_nx_info(id))) {
20852 +               vxdprintk(VXD_CBIT(nid, 0),
20853 +                       "create_nx_info(%d) = %p (already there)", id, nxi);
20854 +               if (nx_info_flags(nxi, NXF_STATE_SETUP, 0))
20855 +                       nxi = ERR_PTR(-EBUSY);
20856 +               else
20857 +                       nxi = ERR_PTR(-EEXIST);
20858 +               goto out_unlock;
20859 +       }
20860 +       /* new context */
20861 +       vxdprintk(VXD_CBIT(nid, 0),
20862 +               "create_nx_info(%d) = %p (new)", id, new);
20863 +       claim_nx_info(new, NULL);
20864 +       __nx_set_lback(new);
20865 +       __hash_nx_info(get_nx_info(new));
20866 +       nxi = new, new = NULL;
20867 +
20868 +out_unlock:
20869 +       spin_unlock(&nx_info_hash_lock);
20870 +       if (new)
20871 +               __dealloc_nx_info(new);
20872 +       return nxi;
20873 +}
20874 +
20875 +
20876 +
20877 +/*     exported stuff                                          */
20878 +
20879 +
20880 +void unhash_nx_info(struct nx_info *nxi)
20881 +{
20882 +       __shutdown_nx_info(nxi);
20883 +       spin_lock(&nx_info_hash_lock);
20884 +       __unhash_nx_info(nxi);
20885 +       spin_unlock(&nx_info_hash_lock);
20886 +}
20887 +
20888 +/*     lookup_nx_info()
20889 +
20890 +       * search for a nx_info and get() it
20891 +       * negative id means current                             */
20892 +
20893 +struct nx_info *lookup_nx_info(int id)
20894 +{
20895 +       struct nx_info *nxi = NULL;
20896 +
20897 +       if (id < 0) {
20898 +               nxi = get_nx_info(current_nx_info());
20899 +       } else if (id > 1) {
20900 +               spin_lock(&nx_info_hash_lock);
20901 +               nxi = get_nx_info(__lookup_nx_info(id));
20902 +               spin_unlock(&nx_info_hash_lock);
20903 +       }
20904 +       return nxi;
20905 +}
20906 +
20907 +/*     nid_is_hashed()
20908 +
20909 +       * verify that nid is still hashed                       */
20910 +
20911 +int nid_is_hashed(nid_t nid)
20912 +{
20913 +       int hashed;
20914 +
20915 +       spin_lock(&nx_info_hash_lock);
20916 +       hashed = (__lookup_nx_info(nid) != NULL);
20917 +       spin_unlock(&nx_info_hash_lock);
20918 +       return hashed;
20919 +}
20920 +
20921 +
20922 +#ifdef CONFIG_PROC_FS
20923 +
20924 +/*     get_nid_list()
20925 +
20926 +       * get a subset of hashed nids for proc
20927 +       * assumes size is at least one                          */
20928 +
20929 +int get_nid_list(int index, unsigned int *nids, int size)
20930 +{
20931 +       int hindex, nr_nids = 0;
20932 +
20933 +       /* only show current and children */
20934 +       if (!nx_check(0, VS_ADMIN | VS_WATCH)) {
20935 +               if (index > 0)
20936 +                       return 0;
20937 +               nids[nr_nids] = nx_current_nid();
20938 +               return 1;
20939 +       }
20940 +
20941 +       for (hindex = 0; hindex < NX_HASH_SIZE; hindex++) {
20942 +               struct hlist_head *head = &nx_info_hash[hindex];
20943 +               struct hlist_node *pos;
20944 +
20945 +               spin_lock(&nx_info_hash_lock);
20946 +               hlist_for_each(pos, head) {
20947 +                       struct nx_info *nxi;
20948 +
20949 +                       if (--index > 0)
20950 +                               continue;
20951 +
20952 +                       nxi = hlist_entry(pos, struct nx_info, nx_hlist);
20953 +                       nids[nr_nids] = nxi->nx_id;
20954 +                       if (++nr_nids >= size) {
20955 +                               spin_unlock(&nx_info_hash_lock);
20956 +                               goto out;
20957 +                       }
20958 +               }
20959 +               /* keep the lock time short */
20960 +               spin_unlock(&nx_info_hash_lock);
20961 +       }
20962 +out:
20963 +       return nr_nids;
20964 +}
20965 +#endif
20966 +
20967 +
20968 +/*
20969 + *     migrate task to new network
20970 + *     gets nxi, puts old_nxi on change
20971 + */
20972 +
20973 +int nx_migrate_task(struct task_struct *p, struct nx_info *nxi)
20974 +{
20975 +       struct nx_info *old_nxi;
20976 +       int ret = 0;
20977 +
20978 +       if (!p || !nxi)
20979 +               BUG();
20980 +
20981 +       vxdprintk(VXD_CBIT(nid, 5),
20982 +               "nx_migrate_task(%p,%p[#%d.%d.%d])",
20983 +               p, nxi, nxi->nx_id,
20984 +               atomic_read(&nxi->nx_usecnt),
20985 +               atomic_read(&nxi->nx_tasks));
20986 +
20987 +       if (nx_info_flags(nxi, NXF_INFO_PRIVATE, 0) &&
20988 +               !nx_info_flags(nxi, NXF_STATE_SETUP, 0))
20989 +               return -EACCES;
20990 +
20991 +       if (nx_info_state(nxi, NXS_SHUTDOWN))
20992 +               return -EFAULT;
20993 +
20994 +       /* maybe disallow this completely? */
20995 +       old_nxi = task_get_nx_info(p);
20996 +       if (old_nxi == nxi)
20997 +               goto out;
20998 +
20999 +       task_lock(p);
21000 +       if (old_nxi)
21001 +               clr_nx_info(&p->nx_info);
21002 +       claim_nx_info(nxi, p);
21003 +       set_nx_info(&p->nx_info, nxi);
21004 +       p->nid = nxi->nx_id;
21005 +       task_unlock(p);
21006 +
21007 +       vxdprintk(VXD_CBIT(nid, 5),
21008 +               "moved task %p into nxi:%p[#%d]",
21009 +               p, nxi, nxi->nx_id);
21010 +
21011 +       if (old_nxi)
21012 +               release_nx_info(old_nxi, p);
21013 +       ret = 0;
21014 +out:
21015 +       put_nx_info(old_nxi);
21016 +       return ret;
21017 +}
21018 +
21019 +
21020 +void nx_set_persistent(struct nx_info *nxi)
21021 +{
21022 +       vxdprintk(VXD_CBIT(nid, 6),
21023 +               "nx_set_persistent(%p[#%d])", nxi, nxi->nx_id);
21024 +
21025 +       get_nx_info(nxi);
21026 +       claim_nx_info(nxi, NULL);
21027 +}
21028 +
21029 +void nx_clear_persistent(struct nx_info *nxi)
21030 +{
21031 +       vxdprintk(VXD_CBIT(nid, 6),
21032 +               "nx_clear_persistent(%p[#%d])", nxi, nxi->nx_id);
21033 +
21034 +       release_nx_info(nxi, NULL);
21035 +       put_nx_info(nxi);
21036 +}
21037 +
21038 +void nx_update_persistent(struct nx_info *nxi)
21039 +{
21040 +       if (nx_info_flags(nxi, NXF_PERSISTENT, 0))
21041 +               nx_set_persistent(nxi);
21042 +       else
21043 +               nx_clear_persistent(nxi);
21044 +}
21045 +
21046 +/* vserver syscall commands below here */
21047 +
21048 +/* taks nid and nx_info functions */
21049 +
21050 +#include <asm/uaccess.h>
21051 +
21052 +
21053 +int vc_task_nid(uint32_t id)
21054 +{
21055 +       nid_t nid;
21056 +
21057 +       if (id) {
21058 +               struct task_struct *tsk;
21059 +
21060 +               read_lock(&tasklist_lock);
21061 +               tsk = find_task_by_real_pid(id);
21062 +               nid = (tsk) ? tsk->nid : -ESRCH;
21063 +               read_unlock(&tasklist_lock);
21064 +       } else
21065 +               nid = nx_current_nid();
21066 +       return nid;
21067 +}
21068 +
21069 +
21070 +int vc_nx_info(struct nx_info *nxi, void __user *data)
21071 +{
21072 +       struct vcmd_nx_info_v0 vc_data;
21073 +
21074 +       vc_data.nid = nxi->nx_id;
21075 +
21076 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21077 +               return -EFAULT;
21078 +       return 0;
21079 +}
21080 +
21081 +
21082 +/* network functions */
21083 +
21084 +int vc_net_create(uint32_t nid, void __user *data)
21085 +{
21086 +       struct vcmd_net_create vc_data = { .flagword = NXF_INIT_SET };
21087 +       struct nx_info *new_nxi;
21088 +       int ret;
21089 +
21090 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21091 +               return -EFAULT;
21092 +
21093 +       if ((nid > MAX_S_CONTEXT) || (nid < 2))
21094 +               return -EINVAL;
21095 +
21096 +       new_nxi = __create_nx_info(nid);
21097 +       if (IS_ERR(new_nxi))
21098 +               return PTR_ERR(new_nxi);
21099 +
21100 +       /* initial flags */
21101 +       new_nxi->nx_flags = vc_data.flagword;
21102 +
21103 +       ret = -ENOEXEC;
21104 +       if (vs_net_change(new_nxi, VSC_NETUP))
21105 +               goto out;
21106 +
21107 +       ret = nx_migrate_task(current, new_nxi);
21108 +       if (ret)
21109 +               goto out;
21110 +
21111 +       /* return context id on success */
21112 +       ret = new_nxi->nx_id;
21113 +
21114 +       /* get a reference for persistent contexts */
21115 +       if ((vc_data.flagword & NXF_PERSISTENT))
21116 +               nx_set_persistent(new_nxi);
21117 +out:
21118 +       release_nx_info(new_nxi, NULL);
21119 +       put_nx_info(new_nxi);
21120 +       return ret;
21121 +}
21122 +
21123 +
21124 +int vc_net_migrate(struct nx_info *nxi, void __user *data)
21125 +{
21126 +       return nx_migrate_task(current, nxi);
21127 +}
21128 +
21129 +
21130 +
21131 +int do_add_v4_addr(struct nx_info *nxi, __be32 ip, __be32 ip2, __be32 mask,
21132 +       uint16_t type, uint16_t flags)
21133 +{
21134 +       struct nx_addr_v4 *nxa = &nxi->v4;
21135 +
21136 +       if (NX_IPV4(nxi)) {
21137 +               /* locate last entry */
21138 +               for (; nxa->next; nxa = nxa->next);
21139 +               nxa->next = __alloc_nx_addr_v4();
21140 +               nxa = nxa->next;
21141 +
21142 +               if (IS_ERR(nxa))
21143 +                       return PTR_ERR(nxa);
21144 +       }
21145 +
21146 +       if (nxi->v4.next)
21147 +               /* remove single ip for ip list */
21148 +               nxi->nx_flags &= ~NXF_SINGLE_IP;
21149 +
21150 +       nxa->ip[0].s_addr = ip;
21151 +       nxa->ip[1].s_addr = ip2;
21152 +       nxa->mask.s_addr = mask;
21153 +       nxa->type = type;
21154 +       nxa->flags = flags;
21155 +       return 0;
21156 +}
21157 +
21158 +
21159 +int vc_net_add(struct nx_info *nxi, void __user *data)
21160 +{
21161 +       struct vcmd_net_addr_v0 vc_data;
21162 +       int index, ret = 0;
21163 +
21164 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21165 +               return -EFAULT;
21166 +
21167 +       switch (vc_data.type) {
21168 +       case NXA_TYPE_IPV4:
21169 +               if ((vc_data.count < 1) || (vc_data.count > 4))
21170 +                       return -EINVAL;
21171 +
21172 +               index = 0;
21173 +               while (index < vc_data.count) {
21174 +                       ret = do_add_v4_addr(nxi, vc_data.ip[index].s_addr, 0,
21175 +                               vc_data.mask[index].s_addr, NXA_TYPE_ADDR, 0);
21176 +                       if (ret)
21177 +                               return ret;
21178 +                       index++;
21179 +               }
21180 +               ret = index;
21181 +               break;
21182 +
21183 +       case NXA_TYPE_IPV4|NXA_MOD_BCAST:
21184 +               nxi->v4_bcast = vc_data.ip[0];
21185 +               ret = 1;
21186 +               break;
21187 +
21188 +       case NXA_TYPE_IPV4|NXA_MOD_LBACK:
21189 +               nxi->v4_lback = vc_data.ip[0];
21190 +               ret = 1;
21191 +               break;
21192 +
21193 +       default:
21194 +               ret = -EINVAL;
21195 +               break;
21196 +       }
21197 +       return ret;
21198 +}
21199 +
21200 +int vc_net_remove(struct nx_info *nxi, void __user *data)
21201 +{
21202 +       struct vcmd_net_addr_v0 vc_data;
21203 +
21204 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21205 +               return -EFAULT;
21206 +
21207 +       switch (vc_data.type) {
21208 +       case NXA_TYPE_ANY:
21209 +               __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
21210 +               memset(&nxi->v4, 0, sizeof(nxi->v4));
21211 +               break;
21212 +
21213 +       default:
21214 +               return -EINVAL;
21215 +       }
21216 +       return 0;
21217 +}
21218 +
21219 +
21220 +int vc_net_add_ipv4(struct nx_info *nxi, void __user *data)
21221 +{
21222 +       struct vcmd_net_addr_ipv4_v1 vc_data;
21223 +
21224 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21225 +               return -EFAULT;
21226 +
21227 +       switch (vc_data.type) {
21228 +       case NXA_TYPE_ADDR:
21229 +       case NXA_TYPE_RANGE:
21230 +       case NXA_TYPE_MASK:
21231 +               return do_add_v4_addr(nxi, vc_data.ip.s_addr, 0,
21232 +                       vc_data.mask.s_addr, vc_data.type, vc_data.flags);
21233 +
21234 +       case NXA_TYPE_ADDR | NXA_MOD_BCAST:
21235 +               nxi->v4_bcast = vc_data.ip;
21236 +               break;
21237 +
21238 +       case NXA_TYPE_ADDR | NXA_MOD_LBACK:
21239 +               nxi->v4_lback = vc_data.ip;
21240 +               break;
21241 +
21242 +       default:
21243 +               return -EINVAL;
21244 +       }
21245 +       return 0;
21246 +}
21247 +
21248 +int vc_net_remove_ipv4(struct nx_info *nxi, void __user *data)
21249 +{
21250 +       struct vcmd_net_addr_ipv4_v1 vc_data;
21251 +
21252 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21253 +               return -EFAULT;
21254 +
21255 +       switch (vc_data.type) {
21256 +/*     case NXA_TYPE_ADDR:
21257 +               break;          */
21258 +
21259 +       case NXA_TYPE_ANY:
21260 +               __dealloc_nx_addr_v4_all(xchg(&nxi->v4.next, NULL));
21261 +               memset(&nxi->v4, 0, sizeof(nxi->v4));
21262 +               break;
21263 +
21264 +       default:
21265 +               return -EINVAL;
21266 +       }
21267 +       return 0;
21268 +}
21269 +
21270 +
21271 +#ifdef CONFIG_IPV6
21272 +
21273 +int do_add_v6_addr(struct nx_info *nxi,
21274 +       struct in6_addr *ip, struct in6_addr *mask,
21275 +       uint32_t prefix, uint16_t type, uint16_t flags)
21276 +{
21277 +       struct nx_addr_v6 *nxa = &nxi->v6;
21278 +
21279 +       if (NX_IPV6(nxi)) {
21280 +               /* locate last entry */
21281 +               for (; nxa->next; nxa = nxa->next);
21282 +               nxa->next = __alloc_nx_addr_v6();
21283 +               nxa = nxa->next;
21284 +
21285 +               if (IS_ERR(nxa))
21286 +                       return PTR_ERR(nxa);
21287 +       }
21288 +
21289 +       nxa->ip = *ip;
21290 +       nxa->mask = *mask;
21291 +       nxa->prefix = prefix;
21292 +       nxa->type = type;
21293 +       nxa->flags = flags;
21294 +       return 0;
21295 +}
21296 +
21297 +
21298 +int vc_net_add_ipv6(struct nx_info *nxi, void __user *data)
21299 +{
21300 +       struct vcmd_net_addr_ipv6_v1 vc_data;
21301 +
21302 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21303 +               return -EFAULT;
21304 +
21305 +       switch (vc_data.type) {
21306 +       case NXA_TYPE_ADDR:
21307 +       case NXA_TYPE_MASK:
21308 +               return do_add_v6_addr(nxi, &vc_data.ip, &vc_data.mask,
21309 +                       vc_data.prefix, vc_data.type, vc_data.flags);
21310 +       default:
21311 +               return -EINVAL;
21312 +       }
21313 +       return 0;
21314 +}
21315 +
21316 +int vc_net_remove_ipv6(struct nx_info *nxi, void __user *data)
21317 +{
21318 +       struct vcmd_net_addr_ipv6_v1 vc_data;
21319 +
21320 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
21321 +               return -EFAULT;
21322 +
21323 +       switch (vc_data.type) {
21324 +       case NXA_TYPE_ANY:
21325 +               __dealloc_nx_addr_v6_all(xchg(&nxi->v6.next, NULL));
21326 +               memset(&nxi->v6, 0, sizeof(nxi->v6));
21327 +               break;
21328 +
21329 +       default:
21330 +               return -EINVAL;
21331 +       }
21332 +       return 0;
21333 +}
21334 +
21335 +#endif /* CONFIG_IPV6 */
21336 +
21337 +
21338 +int vc_get_nflags(struct nx_info *nxi, void __user *data)
21339 +{
21340 +       struct vcmd_net_flags_v0 vc_data;
21341 +
21342 +       vc_data.flagword = nxi->nx_flags;
21343 +
21344 +       /* special STATE flag handling */
21345 +       vc_data.mask = vs_mask_flags(~0ULL, nxi->nx_flags, NXF_ONE_TIME);
21346 +
21347 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21348 +               return -EFAULT;
21349 +       return 0;
21350 +}
21351 +
21352 +int vc_set_nflags(struct nx_info *nxi, void __user *data)
21353 +{
21354 +       struct vcmd_net_flags_v0 vc_data;
21355 +       uint64_t mask, trigger;
21356 +
21357 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21358 +               return -EFAULT;
21359 +
21360 +       /* special STATE flag handling */
21361 +       mask = vs_mask_mask(vc_data.mask, nxi->nx_flags, NXF_ONE_TIME);
21362 +       trigger = (mask & nxi->nx_flags) ^ (mask & vc_data.flagword);
21363 +
21364 +       nxi->nx_flags = vs_mask_flags(nxi->nx_flags,
21365 +               vc_data.flagword, mask);
21366 +       if (trigger & NXF_PERSISTENT)
21367 +               nx_update_persistent(nxi);
21368 +
21369 +       return 0;
21370 +}
21371 +
21372 +int vc_get_ncaps(struct nx_info *nxi, void __user *data)
21373 +{
21374 +       struct vcmd_net_caps_v0 vc_data;
21375 +
21376 +       vc_data.ncaps = nxi->nx_ncaps;
21377 +       vc_data.cmask = ~0ULL;
21378 +
21379 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
21380 +               return -EFAULT;
21381 +       return 0;
21382 +}
21383 +
21384 +int vc_set_ncaps(struct nx_info *nxi, void __user *data)
21385 +{
21386 +       struct vcmd_net_caps_v0 vc_data;
21387 +
21388 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
21389 +               return -EFAULT;
21390 +
21391 +       nxi->nx_ncaps = vs_mask_flags(nxi->nx_ncaps,
21392 +               vc_data.ncaps, vc_data.cmask);
21393 +       return 0;
21394 +}
21395 +
21396 +
21397 +#include <linux/module.h>
21398 +
21399 +module_init(init_network);
21400 +
21401 +EXPORT_SYMBOL_GPL(free_nx_info);
21402 +EXPORT_SYMBOL_GPL(unhash_nx_info);
21403 +
21404 diff -NurpP --minimal linux-2.6.36/kernel/vserver/proc.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/proc.c
21405 --- linux-2.6.36/kernel/vserver/proc.c  1970-01-01 01:00:00.000000000 +0100
21406 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/proc.c    2010-10-21 13:09:36.000000000 +0200
21407 @@ -0,0 +1,1098 @@
21408 +/*
21409 + *  linux/kernel/vserver/proc.c
21410 + *
21411 + *  Virtual Context Support
21412 + *
21413 + *  Copyright (C) 2003-2007  Herbert Pötzl
21414 + *
21415 + *  V0.01  basic structure
21416 + *  V0.02  adaptation vs1.3.0
21417 + *  V0.03  proc permissions
21418 + *  V0.04  locking/generic
21419 + *  V0.05  next generation procfs
21420 + *  V0.06  inode validation
21421 + *  V0.07  generic rewrite vid
21422 + *  V0.08  remove inode type
21423 + *
21424 + */
21425 +
21426 +#include <linux/proc_fs.h>
21427 +#include <linux/fs_struct.h>
21428 +#include <linux/mount.h>
21429 +#include <asm/unistd.h>
21430 +
21431 +#include <linux/vs_context.h>
21432 +#include <linux/vs_network.h>
21433 +#include <linux/vs_cvirt.h>
21434 +
21435 +#include <linux/in.h>
21436 +#include <linux/inetdevice.h>
21437 +#include <linux/vs_inet.h>
21438 +#include <linux/vs_inet6.h>
21439 +
21440 +#include <linux/vserver/global.h>
21441 +
21442 +#include "cvirt_proc.h"
21443 +#include "cacct_proc.h"
21444 +#include "limit_proc.h"
21445 +#include "sched_proc.h"
21446 +#include "vci_config.h"
21447 +
21448 +
21449 +static inline char *print_cap_t(char *buffer, kernel_cap_t *c)
21450 +{
21451 +       unsigned __capi;
21452 +
21453 +       CAP_FOR_EACH_U32(__capi) {
21454 +               buffer += sprintf(buffer, "%08x",
21455 +                       c->cap[(_KERNEL_CAPABILITY_U32S-1) - __capi]);
21456 +       }
21457 +       return buffer;
21458 +}
21459 +
21460 +
21461 +static struct proc_dir_entry *proc_virtual;
21462 +
21463 +static struct proc_dir_entry *proc_virtnet;
21464 +
21465 +
21466 +/* first the actual feeds */
21467 +
21468 +
21469 +static int proc_vci(char *buffer)
21470 +{
21471 +       return sprintf(buffer,
21472 +               "VCIVersion:\t%04x:%04x\n"
21473 +               "VCISyscall:\t%d\n"
21474 +               "VCIKernel:\t%08x\n",
21475 +               VCI_VERSION >> 16,
21476 +               VCI_VERSION & 0xFFFF,
21477 +               __NR_vserver,
21478 +               vci_kernel_config());
21479 +}
21480 +
21481 +static int proc_virtual_info(char *buffer)
21482 +{
21483 +       return proc_vci(buffer);
21484 +}
21485 +
21486 +static int proc_virtual_status(char *buffer)
21487 +{
21488 +       return sprintf(buffer,
21489 +               "#CTotal:\t%d\n"
21490 +               "#CActive:\t%d\n"
21491 +               "#NSProxy:\t%d\t%d %d %d %d %d %d\n"
21492 +               "#InitTask:\t%d\t%d %d\n",
21493 +               atomic_read(&vx_global_ctotal),
21494 +               atomic_read(&vx_global_cactive),
21495 +               atomic_read(&vs_global_nsproxy),
21496 +               atomic_read(&vs_global_fs),
21497 +               atomic_read(&vs_global_mnt_ns),
21498 +               atomic_read(&vs_global_uts_ns),
21499 +               atomic_read(&nr_ipc_ns),
21500 +               atomic_read(&vs_global_user_ns),
21501 +               atomic_read(&vs_global_pid_ns),
21502 +               atomic_read(&init_task.usage),
21503 +               atomic_read(&init_task.nsproxy->count),
21504 +               init_task.fs->users);
21505 +}
21506 +
21507 +
21508 +int proc_vxi_info(struct vx_info *vxi, char *buffer)
21509 +{
21510 +       int length;
21511 +
21512 +       length = sprintf(buffer,
21513 +               "ID:\t%d\n"
21514 +               "Info:\t%p\n"
21515 +               "Init:\t%d\n"
21516 +               "OOM:\t%lld\n",
21517 +               vxi->vx_id,
21518 +               vxi,
21519 +               vxi->vx_initpid,
21520 +               vxi->vx_badness_bias);
21521 +       return length;
21522 +}
21523 +
21524 +int proc_vxi_status(struct vx_info *vxi, char *buffer)
21525 +{
21526 +       char *orig = buffer;
21527 +
21528 +       buffer += sprintf(buffer,
21529 +               "UseCnt:\t%d\n"
21530 +               "Tasks:\t%d\n"
21531 +               "Flags:\t%016llx\n",
21532 +               atomic_read(&vxi->vx_usecnt),
21533 +               atomic_read(&vxi->vx_tasks),
21534 +               (unsigned long long)vxi->vx_flags);
21535 +
21536 +       buffer += sprintf(buffer, "BCaps:\t");
21537 +       buffer = print_cap_t(buffer, &vxi->vx_bcaps);
21538 +       buffer += sprintf(buffer, "\n");
21539 +
21540 +       buffer += sprintf(buffer,
21541 +               "CCaps:\t%016llx\n"
21542 +               "Spaces:\t%08lx %08lx\n",
21543 +               (unsigned long long)vxi->vx_ccaps,
21544 +               vxi->vx_nsmask[0], vxi->vx_nsmask[1]);
21545 +       return buffer - orig;
21546 +}
21547 +
21548 +int proc_vxi_limit(struct vx_info *vxi, char *buffer)
21549 +{
21550 +       return vx_info_proc_limit(&vxi->limit, buffer);
21551 +}
21552 +
21553 +int proc_vxi_sched(struct vx_info *vxi, char *buffer)
21554 +{
21555 +       int cpu, length;
21556 +
21557 +       length = vx_info_proc_sched(&vxi->sched, buffer);
21558 +       for_each_online_cpu(cpu) {
21559 +               length += vx_info_proc_sched_pc(
21560 +                       &vx_per_cpu(vxi, sched_pc, cpu),
21561 +                       buffer + length, cpu);
21562 +       }
21563 +       return length;
21564 +}
21565 +
21566 +int proc_vxi_nsproxy0(struct vx_info *vxi, char *buffer)
21567 +{
21568 +       return vx_info_proc_nsproxy(vxi->vx_nsproxy[0], buffer);
21569 +}
21570 +
21571 +int proc_vxi_nsproxy1(struct vx_info *vxi, char *buffer)
21572 +{
21573 +       return vx_info_proc_nsproxy(vxi->vx_nsproxy[1], buffer);
21574 +}
21575 +
21576 +int proc_vxi_cvirt(struct vx_info *vxi, char *buffer)
21577 +{
21578 +       int cpu, length;
21579 +
21580 +       vx_update_load(vxi);
21581 +       length = vx_info_proc_cvirt(&vxi->cvirt, buffer);
21582 +       for_each_online_cpu(cpu) {
21583 +               length += vx_info_proc_cvirt_pc(
21584 +                       &vx_per_cpu(vxi, cvirt_pc, cpu),
21585 +                       buffer + length, cpu);
21586 +       }
21587 +       return length;
21588 +}
21589 +
21590 +int proc_vxi_cacct(struct vx_info *vxi, char *buffer)
21591 +{
21592 +       return vx_info_proc_cacct(&vxi->cacct, buffer);
21593 +}
21594 +
21595 +
21596 +static int proc_virtnet_info(char *buffer)
21597 +{
21598 +       return proc_vci(buffer);
21599 +}
21600 +
21601 +static int proc_virtnet_status(char *buffer)
21602 +{
21603 +       return sprintf(buffer,
21604 +               "#CTotal:\t%d\n"
21605 +               "#CActive:\t%d\n",
21606 +               atomic_read(&nx_global_ctotal),
21607 +               atomic_read(&nx_global_cactive));
21608 +}
21609 +
21610 +int proc_nxi_info(struct nx_info *nxi, char *buffer)
21611 +{
21612 +       struct nx_addr_v4 *v4a;
21613 +#ifdef CONFIG_IPV6
21614 +       struct nx_addr_v6 *v6a;
21615 +#endif
21616 +       int length, i;
21617 +
21618 +       length = sprintf(buffer,
21619 +               "ID:\t%d\n"
21620 +               "Info:\t%p\n"
21621 +               "Bcast:\t" NIPQUAD_FMT "\n"
21622 +               "Lback:\t" NIPQUAD_FMT "\n",
21623 +               nxi->nx_id,
21624 +               nxi,
21625 +               NIPQUAD(nxi->v4_bcast.s_addr),
21626 +               NIPQUAD(nxi->v4_lback.s_addr));
21627 +
21628 +       if (!NX_IPV4(nxi))
21629 +               goto skip_v4;
21630 +       for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
21631 +               length += sprintf(buffer + length, "%d:\t" NXAV4_FMT "\n",
21632 +                       i, NXAV4(v4a));
21633 +skip_v4:
21634 +#ifdef CONFIG_IPV6
21635 +       if (!NX_IPV6(nxi))
21636 +               goto skip_v6;
21637 +       for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
21638 +               length += sprintf(buffer + length, "%d:\t" NXAV6_FMT "\n",
21639 +                       i, NXAV6(v6a));
21640 +skip_v6:
21641 +#endif
21642 +       return length;
21643 +}
21644 +
21645 +int proc_nxi_status(struct nx_info *nxi, char *buffer)
21646 +{
21647 +       int length;
21648 +
21649 +       length = sprintf(buffer,
21650 +               "UseCnt:\t%d\n"
21651 +               "Tasks:\t%d\n"
21652 +               "Flags:\t%016llx\n"
21653 +               "NCaps:\t%016llx\n",
21654 +               atomic_read(&nxi->nx_usecnt),
21655 +               atomic_read(&nxi->nx_tasks),
21656 +               (unsigned long long)nxi->nx_flags,
21657 +               (unsigned long long)nxi->nx_ncaps);
21658 +       return length;
21659 +}
21660 +
21661 +
21662 +
21663 +/* here the inode helpers */
21664 +
21665 +struct vs_entry {
21666 +       int len;
21667 +       char *name;
21668 +       mode_t mode;
21669 +       struct inode_operations *iop;
21670 +       struct file_operations *fop;
21671 +       union proc_op op;
21672 +};
21673 +
21674 +static struct inode *vs_proc_make_inode(struct super_block *sb, struct vs_entry *p)
21675 +{
21676 +       struct inode *inode = new_inode(sb);
21677 +
21678 +       if (!inode)
21679 +               goto out;
21680 +
21681 +       inode->i_mode = p->mode;
21682 +       if (p->iop)
21683 +               inode->i_op = p->iop;
21684 +       if (p->fop)
21685 +               inode->i_fop = p->fop;
21686 +
21687 +       inode->i_nlink = (p->mode & S_IFDIR) ? 2 : 1;
21688 +       inode->i_flags |= S_IMMUTABLE;
21689 +
21690 +       inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
21691 +
21692 +       inode->i_uid = 0;
21693 +       inode->i_gid = 0;
21694 +       inode->i_tag = 0;
21695 +out:
21696 +       return inode;
21697 +}
21698 +
21699 +static struct dentry *vs_proc_instantiate(struct inode *dir,
21700 +       struct dentry *dentry, int id, void *ptr)
21701 +{
21702 +       struct vs_entry *p = ptr;
21703 +       struct inode *inode = vs_proc_make_inode(dir->i_sb, p);
21704 +       struct dentry *error = ERR_PTR(-EINVAL);
21705 +
21706 +       if (!inode)
21707 +               goto out;
21708 +
21709 +       PROC_I(inode)->op = p->op;
21710 +       PROC_I(inode)->fd = id;
21711 +       d_add(dentry, inode);
21712 +       error = NULL;
21713 +out:
21714 +       return error;
21715 +}
21716 +
21717 +/* Lookups */
21718 +
21719 +typedef struct dentry *instantiate_t(struct inode *, struct dentry *, int, void *);
21720 +
21721 +/*
21722 + * Fill a directory entry.
21723 + *
21724 + * If possible create the dcache entry and derive our inode number and
21725 + * file type from dcache entry.
21726 + *
21727 + * Since all of the proc inode numbers are dynamically generated, the inode
21728 + * numbers do not exist until the inode is cache.  This means creating the
21729 + * the dcache entry in readdir is necessary to keep the inode numbers
21730 + * reported by readdir in sync with the inode numbers reported
21731 + * by stat.
21732 + */
21733 +static int proc_fill_cache(struct file *filp, void *dirent, filldir_t filldir,
21734 +       char *name, int len, instantiate_t instantiate, int id, void *ptr)
21735 +{
21736 +       struct dentry *child, *dir = filp->f_dentry;
21737 +       struct inode *inode;
21738 +       struct qstr qname;
21739 +       ino_t ino = 0;
21740 +       unsigned type = DT_UNKNOWN;
21741 +
21742 +       qname.name = name;
21743 +       qname.len  = len;
21744 +       qname.hash = full_name_hash(name, len);
21745 +
21746 +       child = d_lookup(dir, &qname);
21747 +       if (!child) {
21748 +               struct dentry *new;
21749 +               new = d_alloc(dir, &qname);
21750 +               if (new) {
21751 +                       child = instantiate(dir->d_inode, new, id, ptr);
21752 +                       if (child)
21753 +                               dput(new);
21754 +                       else
21755 +                               child = new;
21756 +               }
21757 +       }
21758 +       if (!child || IS_ERR(child) || !child->d_inode)
21759 +               goto end_instantiate;
21760 +       inode = child->d_inode;
21761 +       if (inode) {
21762 +               ino = inode->i_ino;
21763 +               type = inode->i_mode >> 12;
21764 +       }
21765 +       dput(child);
21766 +end_instantiate:
21767 +       if (!ino)
21768 +               ino = find_inode_number(dir, &qname);
21769 +       if (!ino)
21770 +               ino = 1;
21771 +       return filldir(dirent, name, len, filp->f_pos, ino, type);
21772 +}
21773 +
21774 +
21775 +
21776 +/* get and revalidate vx_info/xid */
21777 +
21778 +static inline
21779 +struct vx_info *get_proc_vx_info(struct inode *inode)
21780 +{
21781 +       return lookup_vx_info(PROC_I(inode)->fd);
21782 +}
21783 +
21784 +static int proc_xid_revalidate(struct dentry *dentry, struct nameidata *nd)
21785 +{
21786 +       struct inode *inode = dentry->d_inode;
21787 +       xid_t xid = PROC_I(inode)->fd;
21788 +
21789 +       if (!xid || xid_is_hashed(xid))
21790 +               return 1;
21791 +       d_drop(dentry);
21792 +       return 0;
21793 +}
21794 +
21795 +
21796 +/* get and revalidate nx_info/nid */
21797 +
21798 +static int proc_nid_revalidate(struct dentry *dentry, struct nameidata *nd)
21799 +{
21800 +       struct inode *inode = dentry->d_inode;
21801 +       nid_t nid = PROC_I(inode)->fd;
21802 +
21803 +       if (!nid || nid_is_hashed(nid))
21804 +               return 1;
21805 +       d_drop(dentry);
21806 +       return 0;
21807 +}
21808 +
21809 +
21810 +
21811 +#define PROC_BLOCK_SIZE (PAGE_SIZE - 1024)
21812 +
21813 +static ssize_t proc_vs_info_read(struct file *file, char __user *buf,
21814 +                         size_t count, loff_t *ppos)
21815 +{
21816 +       struct inode *inode = file->f_dentry->d_inode;
21817 +       unsigned long page;
21818 +       ssize_t length = 0;
21819 +
21820 +       if (count > PROC_BLOCK_SIZE)
21821 +               count = PROC_BLOCK_SIZE;
21822 +
21823 +       /* fade that out as soon as stable */
21824 +       WARN_ON(PROC_I(inode)->fd);
21825 +
21826 +       if (!(page = __get_free_page(GFP_KERNEL)))
21827 +               return -ENOMEM;
21828 +
21829 +       BUG_ON(!PROC_I(inode)->op.proc_vs_read);
21830 +       length = PROC_I(inode)->op.proc_vs_read((char *)page);
21831 +
21832 +       if (length >= 0)
21833 +               length = simple_read_from_buffer(buf, count, ppos,
21834 +                       (char *)page, length);
21835 +
21836 +       free_page(page);
21837 +       return length;
21838 +}
21839 +
21840 +static ssize_t proc_vx_info_read(struct file *file, char __user *buf,
21841 +                         size_t count, loff_t *ppos)
21842 +{
21843 +       struct inode *inode = file->f_dentry->d_inode;
21844 +       struct vx_info *vxi = NULL;
21845 +       xid_t xid = PROC_I(inode)->fd;
21846 +       unsigned long page;
21847 +       ssize_t length = 0;
21848 +
21849 +       if (count > PROC_BLOCK_SIZE)
21850 +               count = PROC_BLOCK_SIZE;
21851 +
21852 +       /* fade that out as soon as stable */
21853 +       WARN_ON(!xid);
21854 +       vxi = lookup_vx_info(xid);
21855 +       if (!vxi)
21856 +               goto out;
21857 +
21858 +       length = -ENOMEM;
21859 +       if (!(page = __get_free_page(GFP_KERNEL)))
21860 +               goto out_put;
21861 +
21862 +       BUG_ON(!PROC_I(inode)->op.proc_vxi_read);
21863 +       length = PROC_I(inode)->op.proc_vxi_read(vxi, (char *)page);
21864 +
21865 +       if (length >= 0)
21866 +               length = simple_read_from_buffer(buf, count, ppos,
21867 +                       (char *)page, length);
21868 +
21869 +       free_page(page);
21870 +out_put:
21871 +       put_vx_info(vxi);
21872 +out:
21873 +       return length;
21874 +}
21875 +
21876 +static ssize_t proc_nx_info_read(struct file *file, char __user *buf,
21877 +                         size_t count, loff_t *ppos)
21878 +{
21879 +       struct inode *inode = file->f_dentry->d_inode;
21880 +       struct nx_info *nxi = NULL;
21881 +       nid_t nid = PROC_I(inode)->fd;
21882 +       unsigned long page;
21883 +       ssize_t length = 0;
21884 +
21885 +       if (count > PROC_BLOCK_SIZE)
21886 +               count = PROC_BLOCK_SIZE;
21887 +
21888 +       /* fade that out as soon as stable */
21889 +       WARN_ON(!nid);
21890 +       nxi = lookup_nx_info(nid);
21891 +       if (!nxi)
21892 +               goto out;
21893 +
21894 +       length = -ENOMEM;
21895 +       if (!(page = __get_free_page(GFP_KERNEL)))
21896 +               goto out_put;
21897 +
21898 +       BUG_ON(!PROC_I(inode)->op.proc_nxi_read);
21899 +       length = PROC_I(inode)->op.proc_nxi_read(nxi, (char *)page);
21900 +
21901 +       if (length >= 0)
21902 +               length = simple_read_from_buffer(buf, count, ppos,
21903 +                       (char *)page, length);
21904 +
21905 +       free_page(page);
21906 +out_put:
21907 +       put_nx_info(nxi);
21908 +out:
21909 +       return length;
21910 +}
21911 +
21912 +
21913 +
21914 +/* here comes the lower level */
21915 +
21916 +
21917 +#define NOD(NAME, MODE, IOP, FOP, OP) {        \
21918 +       .len  = sizeof(NAME) - 1,       \
21919 +       .name = (NAME),                 \
21920 +       .mode = MODE,                   \
21921 +       .iop  = IOP,                    \
21922 +       .fop  = FOP,                    \
21923 +       .op   = OP,                     \
21924 +}
21925 +
21926 +
21927 +#define DIR(NAME, MODE, OTYPE)                         \
21928 +       NOD(NAME, (S_IFDIR | (MODE)),                   \
21929 +               &proc_ ## OTYPE ## _inode_operations,   \
21930 +               &proc_ ## OTYPE ## _file_operations, { } )
21931 +
21932 +#define INF(NAME, MODE, OTYPE)                         \
21933 +       NOD(NAME, (S_IFREG | (MODE)), NULL,             \
21934 +               &proc_vs_info_file_operations,          \
21935 +               { .proc_vs_read = &proc_##OTYPE } )
21936 +
21937 +#define VINF(NAME, MODE, OTYPE)                                \
21938 +       NOD(NAME, (S_IFREG | (MODE)), NULL,             \
21939 +               &proc_vx_info_file_operations,          \
21940 +               { .proc_vxi_read = &proc_##OTYPE } )
21941 +
21942 +#define NINF(NAME, MODE, OTYPE)                                \
21943 +       NOD(NAME, (S_IFREG | (MODE)), NULL,             \
21944 +               &proc_nx_info_file_operations,          \
21945 +               { .proc_nxi_read = &proc_##OTYPE } )
21946 +
21947 +
21948 +static struct file_operations proc_vs_info_file_operations = {
21949 +       .read =         proc_vs_info_read,
21950 +};
21951 +
21952 +static struct file_operations proc_vx_info_file_operations = {
21953 +       .read =         proc_vx_info_read,
21954 +};
21955 +
21956 +static struct dentry_operations proc_xid_dentry_operations = {
21957 +       .d_revalidate = proc_xid_revalidate,
21958 +};
21959 +
21960 +static struct vs_entry vx_base_stuff[] = {
21961 +       VINF("info",    S_IRUGO, vxi_info),
21962 +       VINF("status",  S_IRUGO, vxi_status),
21963 +       VINF("limit",   S_IRUGO, vxi_limit),
21964 +       VINF("sched",   S_IRUGO, vxi_sched),
21965 +       VINF("nsproxy", S_IRUGO, vxi_nsproxy0),
21966 +       VINF("nsproxy1",S_IRUGO, vxi_nsproxy1),
21967 +       VINF("cvirt",   S_IRUGO, vxi_cvirt),
21968 +       VINF("cacct",   S_IRUGO, vxi_cacct),
21969 +       {}
21970 +};
21971 +
21972 +
21973 +
21974 +
21975 +static struct dentry *proc_xid_instantiate(struct inode *dir,
21976 +       struct dentry *dentry, int id, void *ptr)
21977 +{
21978 +       dentry->d_op = &proc_xid_dentry_operations;
21979 +       return vs_proc_instantiate(dir, dentry, id, ptr);
21980 +}
21981 +
21982 +static struct dentry *proc_xid_lookup(struct inode *dir,
21983 +       struct dentry *dentry, struct nameidata *nd)
21984 +{
21985 +       struct vs_entry *p = vx_base_stuff;
21986 +       struct dentry *error = ERR_PTR(-ENOENT);
21987 +
21988 +       for (; p->name; p++) {
21989 +               if (p->len != dentry->d_name.len)
21990 +                       continue;
21991 +               if (!memcmp(dentry->d_name.name, p->name, p->len))
21992 +                       break;
21993 +       }
21994 +       if (!p->name)
21995 +               goto out;
21996 +
21997 +       error = proc_xid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
21998 +out:
21999 +       return error;
22000 +}
22001 +
22002 +static int proc_xid_readdir(struct file *filp,
22003 +       void *dirent, filldir_t filldir)
22004 +{
22005 +       struct dentry *dentry = filp->f_dentry;
22006 +       struct inode *inode = dentry->d_inode;
22007 +       struct vs_entry *p = vx_base_stuff;
22008 +       int size = sizeof(vx_base_stuff) / sizeof(struct vs_entry);
22009 +       int pos, index;
22010 +       u64 ino;
22011 +
22012 +       pos = filp->f_pos;
22013 +       switch (pos) {
22014 +       case 0:
22015 +               ino = inode->i_ino;
22016 +               if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
22017 +                       goto out;
22018 +               pos++;
22019 +               /* fall through */
22020 +       case 1:
22021 +               ino = parent_ino(dentry);
22022 +               if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
22023 +                       goto out;
22024 +               pos++;
22025 +               /* fall through */
22026 +       default:
22027 +               index = pos - 2;
22028 +               if (index >= size)
22029 +                       goto out;
22030 +               for (p += index; p->name; p++) {
22031 +                       if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
22032 +                               vs_proc_instantiate, PROC_I(inode)->fd, p))
22033 +                               goto out;
22034 +                       pos++;
22035 +               }
22036 +       }
22037 +out:
22038 +       filp->f_pos = pos;
22039 +       return 1;
22040 +}
22041 +
22042 +
22043 +
22044 +static struct file_operations proc_nx_info_file_operations = {
22045 +       .read =         proc_nx_info_read,
22046 +};
22047 +
22048 +static struct dentry_operations proc_nid_dentry_operations = {
22049 +       .d_revalidate = proc_nid_revalidate,
22050 +};
22051 +
22052 +static struct vs_entry nx_base_stuff[] = {
22053 +       NINF("info",    S_IRUGO, nxi_info),
22054 +       NINF("status",  S_IRUGO, nxi_status),
22055 +       {}
22056 +};
22057 +
22058 +
22059 +static struct dentry *proc_nid_instantiate(struct inode *dir,
22060 +       struct dentry *dentry, int id, void *ptr)
22061 +{
22062 +       dentry->d_op = &proc_nid_dentry_operations;
22063 +       return vs_proc_instantiate(dir, dentry, id, ptr);
22064 +}
22065 +
22066 +static struct dentry *proc_nid_lookup(struct inode *dir,
22067 +       struct dentry *dentry, struct nameidata *nd)
22068 +{
22069 +       struct vs_entry *p = nx_base_stuff;
22070 +       struct dentry *error = ERR_PTR(-ENOENT);
22071 +
22072 +       for (; p->name; p++) {
22073 +               if (p->len != dentry->d_name.len)
22074 +                       continue;
22075 +               if (!memcmp(dentry->d_name.name, p->name, p->len))
22076 +                       break;
22077 +       }
22078 +       if (!p->name)
22079 +               goto out;
22080 +
22081 +       error = proc_nid_instantiate(dir, dentry, PROC_I(dir)->fd, p);
22082 +out:
22083 +       return error;
22084 +}
22085 +
22086 +static int proc_nid_readdir(struct file *filp,
22087 +       void *dirent, filldir_t filldir)
22088 +{
22089 +       struct dentry *dentry = filp->f_dentry;
22090 +       struct inode *inode = dentry->d_inode;
22091 +       struct vs_entry *p = nx_base_stuff;
22092 +       int size = sizeof(nx_base_stuff) / sizeof(struct vs_entry);
22093 +       int pos, index;
22094 +       u64 ino;
22095 +
22096 +       pos = filp->f_pos;
22097 +       switch (pos) {
22098 +       case 0:
22099 +               ino = inode->i_ino;
22100 +               if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
22101 +                       goto out;
22102 +               pos++;
22103 +               /* fall through */
22104 +       case 1:
22105 +               ino = parent_ino(dentry);
22106 +               if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
22107 +                       goto out;
22108 +               pos++;
22109 +               /* fall through */
22110 +       default:
22111 +               index = pos - 2;
22112 +               if (index >= size)
22113 +                       goto out;
22114 +               for (p += index; p->name; p++) {
22115 +                       if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
22116 +                               vs_proc_instantiate, PROC_I(inode)->fd, p))
22117 +                               goto out;
22118 +                       pos++;
22119 +               }
22120 +       }
22121 +out:
22122 +       filp->f_pos = pos;
22123 +       return 1;
22124 +}
22125 +
22126 +
22127 +#define MAX_MULBY10    ((~0U - 9) / 10)
22128 +
22129 +static inline int atovid(const char *str, int len)
22130 +{
22131 +       int vid, c;
22132 +
22133 +       vid = 0;
22134 +       while (len-- > 0) {
22135 +               c = *str - '0';
22136 +               str++;
22137 +               if (c > 9)
22138 +                       return -1;
22139 +               if (vid >= MAX_MULBY10)
22140 +                       return -1;
22141 +               vid *= 10;
22142 +               vid += c;
22143 +               if (!vid)
22144 +                       return -1;
22145 +       }
22146 +       return vid;
22147 +}
22148 +
22149 +/* now the upper level (virtual) */
22150 +
22151 +
22152 +static struct file_operations proc_xid_file_operations = {
22153 +       .read =         generic_read_dir,
22154 +       .readdir =      proc_xid_readdir,
22155 +};
22156 +
22157 +static struct inode_operations proc_xid_inode_operations = {
22158 +       .lookup =       proc_xid_lookup,
22159 +};
22160 +
22161 +static struct vs_entry vx_virtual_stuff[] = {
22162 +       INF("info",     S_IRUGO, virtual_info),
22163 +       INF("status",   S_IRUGO, virtual_status),
22164 +       DIR(NULL,       S_IRUGO | S_IXUGO, xid),
22165 +};
22166 +
22167 +
22168 +static struct dentry *proc_virtual_lookup(struct inode *dir,
22169 +       struct dentry *dentry, struct nameidata *nd)
22170 +{
22171 +       struct vs_entry *p = vx_virtual_stuff;
22172 +       struct dentry *error = ERR_PTR(-ENOENT);
22173 +       int id = 0;
22174 +
22175 +       for (; p->name; p++) {
22176 +               if (p->len != dentry->d_name.len)
22177 +                       continue;
22178 +               if (!memcmp(dentry->d_name.name, p->name, p->len))
22179 +                       break;
22180 +       }
22181 +       if (p->name)
22182 +               goto instantiate;
22183 +
22184 +       id = atovid(dentry->d_name.name, dentry->d_name.len);
22185 +       if ((id < 0) || !xid_is_hashed(id))
22186 +               goto out;
22187 +
22188 +instantiate:
22189 +       error = proc_xid_instantiate(dir, dentry, id, p);
22190 +out:
22191 +       return error;
22192 +}
22193 +
22194 +static struct file_operations proc_nid_file_operations = {
22195 +       .read =         generic_read_dir,
22196 +       .readdir =      proc_nid_readdir,
22197 +};
22198 +
22199 +static struct inode_operations proc_nid_inode_operations = {
22200 +       .lookup =       proc_nid_lookup,
22201 +};
22202 +
22203 +static struct vs_entry nx_virtnet_stuff[] = {
22204 +       INF("info",     S_IRUGO, virtnet_info),
22205 +       INF("status",   S_IRUGO, virtnet_status),
22206 +       DIR(NULL,       S_IRUGO | S_IXUGO, nid),
22207 +};
22208 +
22209 +
22210 +static struct dentry *proc_virtnet_lookup(struct inode *dir,
22211 +       struct dentry *dentry, struct nameidata *nd)
22212 +{
22213 +       struct vs_entry *p = nx_virtnet_stuff;
22214 +       struct dentry *error = ERR_PTR(-ENOENT);
22215 +       int id = 0;
22216 +
22217 +       for (; p->name; p++) {
22218 +               if (p->len != dentry->d_name.len)
22219 +                       continue;
22220 +               if (!memcmp(dentry->d_name.name, p->name, p->len))
22221 +                       break;
22222 +       }
22223 +       if (p->name)
22224 +               goto instantiate;
22225 +
22226 +       id = atovid(dentry->d_name.name, dentry->d_name.len);
22227 +       if ((id < 0) || !nid_is_hashed(id))
22228 +               goto out;
22229 +
22230 +instantiate:
22231 +       error = proc_nid_instantiate(dir, dentry, id, p);
22232 +out:
22233 +       return error;
22234 +}
22235 +
22236 +
22237 +#define PROC_MAXVIDS 32
22238 +
22239 +int proc_virtual_readdir(struct file *filp,
22240 +       void *dirent, filldir_t filldir)
22241 +{
22242 +       struct dentry *dentry = filp->f_dentry;
22243 +       struct inode *inode = dentry->d_inode;
22244 +       struct vs_entry *p = vx_virtual_stuff;
22245 +       int size = sizeof(vx_virtual_stuff) / sizeof(struct vs_entry);
22246 +       int pos, index;
22247 +       unsigned int xid_array[PROC_MAXVIDS];
22248 +       char buf[PROC_NUMBUF];
22249 +       unsigned int nr_xids, i;
22250 +       u64 ino;
22251 +
22252 +       pos = filp->f_pos;
22253 +       switch (pos) {
22254 +       case 0:
22255 +               ino = inode->i_ino;
22256 +               if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
22257 +                       goto out;
22258 +               pos++;
22259 +               /* fall through */
22260 +       case 1:
22261 +               ino = parent_ino(dentry);
22262 +               if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
22263 +                       goto out;
22264 +               pos++;
22265 +               /* fall through */
22266 +       default:
22267 +               index = pos - 2;
22268 +               if (index >= size)
22269 +                       goto entries;
22270 +               for (p += index; p->name; p++) {
22271 +                       if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
22272 +                               vs_proc_instantiate, 0, p))
22273 +                               goto out;
22274 +                       pos++;
22275 +               }
22276 +       entries:
22277 +               index = pos - size;
22278 +               p = &vx_virtual_stuff[size - 1];
22279 +               nr_xids = get_xid_list(index, xid_array, PROC_MAXVIDS);
22280 +               for (i = 0; i < nr_xids; i++) {
22281 +                       int n, xid = xid_array[i];
22282 +                       unsigned int j = PROC_NUMBUF;
22283 +
22284 +                       n = xid;
22285 +                       do
22286 +                               buf[--j] = '0' + (n % 10);
22287 +                       while (n /= 10);
22288 +
22289 +                       if (proc_fill_cache(filp, dirent, filldir,
22290 +                               buf + j, PROC_NUMBUF - j,
22291 +                               vs_proc_instantiate, xid, p))
22292 +                               goto out;
22293 +                       pos++;
22294 +               }
22295 +       }
22296 +out:
22297 +       filp->f_pos = pos;
22298 +       return 0;
22299 +}
22300 +
22301 +static int proc_virtual_getattr(struct vfsmount *mnt,
22302 +       struct dentry *dentry, struct kstat *stat)
22303 +{
22304 +       struct inode *inode = dentry->d_inode;
22305 +
22306 +       generic_fillattr(inode, stat);
22307 +       stat->nlink = 2 + atomic_read(&vx_global_cactive);
22308 +       return 0;
22309 +}
22310 +
22311 +static struct file_operations proc_virtual_dir_operations = {
22312 +       .read =         generic_read_dir,
22313 +       .readdir =      proc_virtual_readdir,
22314 +};
22315 +
22316 +static struct inode_operations proc_virtual_dir_inode_operations = {
22317 +       .getattr =      proc_virtual_getattr,
22318 +       .lookup =       proc_virtual_lookup,
22319 +};
22320 +
22321 +
22322 +
22323 +
22324 +
22325 +int proc_virtnet_readdir(struct file *filp,
22326 +       void *dirent, filldir_t filldir)
22327 +{
22328 +       struct dentry *dentry = filp->f_dentry;
22329 +       struct inode *inode = dentry->d_inode;
22330 +       struct vs_entry *p = nx_virtnet_stuff;
22331 +       int size = sizeof(nx_virtnet_stuff) / sizeof(struct vs_entry);
22332 +       int pos, index;
22333 +       unsigned int nid_array[PROC_MAXVIDS];
22334 +       char buf[PROC_NUMBUF];
22335 +       unsigned int nr_nids, i;
22336 +       u64 ino;
22337 +
22338 +       pos = filp->f_pos;
22339 +       switch (pos) {
22340 +       case 0:
22341 +               ino = inode->i_ino;
22342 +               if (filldir(dirent, ".", 1, pos, ino, DT_DIR) < 0)
22343 +                       goto out;
22344 +               pos++;
22345 +               /* fall through */
22346 +       case 1:
22347 +               ino = parent_ino(dentry);
22348 +               if (filldir(dirent, "..", 2, pos, ino, DT_DIR) < 0)
22349 +                       goto out;
22350 +               pos++;
22351 +               /* fall through */
22352 +       default:
22353 +               index = pos - 2;
22354 +               if (index >= size)
22355 +                       goto entries;
22356 +               for (p += index; p->name; p++) {
22357 +                       if (proc_fill_cache(filp, dirent, filldir, p->name, p->len,
22358 +                               vs_proc_instantiate, 0, p))
22359 +                               goto out;
22360 +                       pos++;
22361 +               }
22362 +       entries:
22363 +               index = pos - size;
22364 +               p = &nx_virtnet_stuff[size - 1];
22365 +               nr_nids = get_nid_list(index, nid_array, PROC_MAXVIDS);
22366 +               for (i = 0; i < nr_nids; i++) {
22367 +                       int n, nid = nid_array[i];
22368 +                       unsigned int j = PROC_NUMBUF;
22369 +
22370 +                       n = nid;
22371 +                       do
22372 +                               buf[--j] = '0' + (n % 10);
22373 +                       while (n /= 10);
22374 +
22375 +                       if (proc_fill_cache(filp, dirent, filldir,
22376 +                               buf + j, PROC_NUMBUF - j,
22377 +                               vs_proc_instantiate, nid, p))
22378 +                               goto out;
22379 +                       pos++;
22380 +               }
22381 +       }
22382 +out:
22383 +       filp->f_pos = pos;
22384 +       return 0;
22385 +}
22386 +
22387 +static int proc_virtnet_getattr(struct vfsmount *mnt,
22388 +       struct dentry *dentry, struct kstat *stat)
22389 +{
22390 +       struct inode *inode = dentry->d_inode;
22391 +
22392 +       generic_fillattr(inode, stat);
22393 +       stat->nlink = 2 + atomic_read(&nx_global_cactive);
22394 +       return 0;
22395 +}
22396 +
22397 +static struct file_operations proc_virtnet_dir_operations = {
22398 +       .read =         generic_read_dir,
22399 +       .readdir =      proc_virtnet_readdir,
22400 +};
22401 +
22402 +static struct inode_operations proc_virtnet_dir_inode_operations = {
22403 +       .getattr =      proc_virtnet_getattr,
22404 +       .lookup =       proc_virtnet_lookup,
22405 +};
22406 +
22407 +
22408 +
22409 +void proc_vx_init(void)
22410 +{
22411 +       struct proc_dir_entry *ent;
22412 +
22413 +       ent = proc_mkdir("virtual", 0);
22414 +       if (ent) {
22415 +               ent->proc_fops = &proc_virtual_dir_operations;
22416 +               ent->proc_iops = &proc_virtual_dir_inode_operations;
22417 +       }
22418 +       proc_virtual = ent;
22419 +
22420 +       ent = proc_mkdir("virtnet", 0);
22421 +       if (ent) {
22422 +               ent->proc_fops = &proc_virtnet_dir_operations;
22423 +               ent->proc_iops = &proc_virtnet_dir_inode_operations;
22424 +       }
22425 +       proc_virtnet = ent;
22426 +}
22427 +
22428 +
22429 +
22430 +
22431 +/* per pid info */
22432 +
22433 +
22434 +int proc_pid_vx_info(struct task_struct *p, char *buffer)
22435 +{
22436 +       struct vx_info *vxi;
22437 +       char *orig = buffer;
22438 +
22439 +       buffer += sprintf(buffer, "XID:\t%d\n", vx_task_xid(p));
22440 +
22441 +       vxi = task_get_vx_info(p);
22442 +       if (!vxi)
22443 +               goto out;
22444 +
22445 +       buffer += sprintf(buffer, "BCaps:\t");
22446 +       buffer = print_cap_t(buffer, &vxi->vx_bcaps);
22447 +       buffer += sprintf(buffer, "\n");
22448 +       buffer += sprintf(buffer, "CCaps:\t%016llx\n",
22449 +               (unsigned long long)vxi->vx_ccaps);
22450 +       buffer += sprintf(buffer, "CFlags:\t%016llx\n",
22451 +               (unsigned long long)vxi->vx_flags);
22452 +       buffer += sprintf(buffer, "CIPid:\t%d\n", vxi->vx_initpid);
22453 +
22454 +       put_vx_info(vxi);
22455 +out:
22456 +       return buffer - orig;
22457 +}
22458 +
22459 +
22460 +int proc_pid_nx_info(struct task_struct *p, char *buffer)
22461 +{
22462 +       struct nx_info *nxi;
22463 +       struct nx_addr_v4 *v4a;
22464 +#ifdef CONFIG_IPV6
22465 +       struct nx_addr_v6 *v6a;
22466 +#endif
22467 +       char *orig = buffer;
22468 +       int i;
22469 +
22470 +       buffer += sprintf(buffer, "NID:\t%d\n", nx_task_nid(p));
22471 +
22472 +       nxi = task_get_nx_info(p);
22473 +       if (!nxi)
22474 +               goto out;
22475 +
22476 +       buffer += sprintf(buffer, "NCaps:\t%016llx\n",
22477 +               (unsigned long long)nxi->nx_ncaps);
22478 +       buffer += sprintf(buffer, "NFlags:\t%016llx\n",
22479 +               (unsigned long long)nxi->nx_flags);
22480 +
22481 +       buffer += sprintf(buffer,
22482 +               "V4Root[bcast]:\t" NIPQUAD_FMT "\n",
22483 +               NIPQUAD(nxi->v4_bcast.s_addr));
22484 +       buffer += sprintf (buffer,
22485 +               "V4Root[lback]:\t" NIPQUAD_FMT "\n",
22486 +               NIPQUAD(nxi->v4_lback.s_addr));
22487 +       if (!NX_IPV4(nxi))
22488 +               goto skip_v4;
22489 +       for (i = 0, v4a = &nxi->v4; v4a; i++, v4a = v4a->next)
22490 +               buffer += sprintf(buffer, "V4Root[%d]:\t" NXAV4_FMT "\n",
22491 +                       i, NXAV4(v4a));
22492 +skip_v4:
22493 +#ifdef CONFIG_IPV6
22494 +       if (!NX_IPV6(nxi))
22495 +               goto skip_v6;
22496 +       for (i = 0, v6a = &nxi->v6; v6a; i++, v6a = v6a->next)
22497 +               buffer += sprintf(buffer, "V6Root[%d]:\t" NXAV6_FMT "\n",
22498 +                       i, NXAV6(v6a));
22499 +skip_v6:
22500 +#endif
22501 +       put_nx_info(nxi);
22502 +out:
22503 +       return buffer - orig;
22504 +}
22505 +
22506 diff -NurpP --minimal linux-2.6.36/kernel/vserver/sched.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched.c
22507 --- linux-2.6.36/kernel/vserver/sched.c 1970-01-01 01:00:00.000000000 +0100
22508 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched.c   2010-10-28 19:13:19.000000000 +0200
22509 @@ -0,0 +1,82 @@
22510 +/*
22511 + *  linux/kernel/vserver/sched.c
22512 + *
22513 + *  Virtual Server: Scheduler Support
22514 + *
22515 + *  Copyright (C) 2004-2010  Herbert Pötzl
22516 + *
22517 + *  V0.01  adapted Sam Vilains version to 2.6.3
22518 + *  V0.02  removed legacy interface
22519 + *  V0.03  changed vcmds to vxi arg
22520 + *  V0.04  removed older and legacy interfaces
22521 + *  V0.05  removed scheduler code/commands
22522 + *
22523 + */
22524 +
22525 +#include <linux/vs_context.h>
22526 +#include <linux/vs_sched.h>
22527 +#include <linux/vserver/sched_cmd.h>
22528 +
22529 +#include <asm/uaccess.h>
22530 +
22531 +
22532 +void vx_update_sched_param(struct _vx_sched *sched,
22533 +       struct _vx_sched_pc *sched_pc)
22534 +{
22535 +       sched_pc->prio_bias = sched->prio_bias;
22536 +}
22537 +
22538 +static int do_set_prio_bias(struct vx_info *vxi, struct vcmd_prio_bias *data)
22539 +{
22540 +       int cpu;
22541 +
22542 +       if (data->prio_bias > MAX_PRIO_BIAS)
22543 +               data->prio_bias = MAX_PRIO_BIAS;
22544 +       if (data->prio_bias < MIN_PRIO_BIAS)
22545 +               data->prio_bias = MIN_PRIO_BIAS;
22546 +
22547 +       if (data->cpu_id != ~0) {
22548 +               vxi->sched.update = cpumask_of_cpu(data->cpu_id);
22549 +               cpus_and(vxi->sched.update, cpu_online_map,
22550 +                       vxi->sched.update);
22551 +       } else
22552 +               vxi->sched.update = cpu_online_map;
22553 +
22554 +       for_each_cpu_mask(cpu, vxi->sched.update)
22555 +               vx_update_sched_param(&vxi->sched,
22556 +                       &vx_per_cpu(vxi, sched_pc, cpu));
22557 +       return 0;
22558 +}
22559 +
22560 +int vc_set_prio_bias(struct vx_info *vxi, void __user *data)
22561 +{
22562 +       struct vcmd_prio_bias vc_data;
22563 +
22564 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22565 +               return -EFAULT;
22566 +
22567 +       return do_set_prio_bias(vxi, &vc_data);
22568 +}
22569 +
22570 +int vc_get_prio_bias(struct vx_info *vxi, void __user *data)
22571 +{
22572 +       struct vcmd_prio_bias vc_data;
22573 +       struct _vx_sched_pc *pcd;
22574 +       int cpu;
22575 +
22576 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22577 +               return -EFAULT;
22578 +
22579 +       cpu = vc_data.cpu_id;
22580 +
22581 +       if (!cpu_possible(cpu))
22582 +               return -EINVAL;
22583 +
22584 +       pcd = &vx_per_cpu(vxi, sched_pc, cpu);
22585 +       vc_data.prio_bias = pcd->prio_bias;
22586 +
22587 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22588 +               return -EFAULT;
22589 +       return 0;
22590 +}
22591 +
22592 diff -NurpP --minimal linux-2.6.36/kernel/vserver/sched_init.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched_init.h
22593 --- linux-2.6.36/kernel/vserver/sched_init.h    1970-01-01 01:00:00.000000000 +0100
22594 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched_init.h      2010-10-28 18:17:11.000000000 +0200
22595 @@ -0,0 +1,27 @@
22596 +
22597 +static inline void vx_info_init_sched(struct _vx_sched *sched)
22598 +{
22599 +       /* scheduling; hard code starting values as constants */
22600 +       sched->prio_bias = 0;
22601 +}
22602 +
22603 +static inline
22604 +void vx_info_init_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
22605 +{
22606 +       sched_pc->prio_bias = 0;
22607 +
22608 +       sched_pc->user_ticks = 0;
22609 +       sched_pc->sys_ticks = 0;
22610 +       sched_pc->hold_ticks = 0;
22611 +}
22612 +
22613 +static inline void vx_info_exit_sched(struct _vx_sched *sched)
22614 +{
22615 +       return;
22616 +}
22617 +
22618 +static inline
22619 +void vx_info_exit_sched_pc(struct _vx_sched_pc *sched_pc, int cpu)
22620 +{
22621 +       return;
22622 +}
22623 diff -NurpP --minimal linux-2.6.36/kernel/vserver/sched_proc.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched_proc.h
22624 --- linux-2.6.36/kernel/vserver/sched_proc.h    1970-01-01 01:00:00.000000000 +0100
22625 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sched_proc.h      2010-10-28 18:15:30.000000000 +0200
22626 @@ -0,0 +1,32 @@
22627 +#ifndef _VX_SCHED_PROC_H
22628 +#define _VX_SCHED_PROC_H
22629 +
22630 +
22631 +static inline
22632 +int vx_info_proc_sched(struct _vx_sched *sched, char *buffer)
22633 +{
22634 +       int length = 0;
22635 +
22636 +       length += sprintf(buffer,
22637 +               "PrioBias:\t%8d\n",
22638 +               sched->prio_bias);
22639 +       return length;
22640 +}
22641 +
22642 +static inline
22643 +int vx_info_proc_sched_pc(struct _vx_sched_pc *sched_pc,
22644 +       char *buffer, int cpu)
22645 +{
22646 +       int length = 0;
22647 +
22648 +       length += sprintf(buffer + length,
22649 +               "cpu %d: %lld %lld %lld", cpu,
22650 +               (unsigned long long)sched_pc->user_ticks,
22651 +               (unsigned long long)sched_pc->sys_ticks,
22652 +               (unsigned long long)sched_pc->hold_ticks);
22653 +       length += sprintf(buffer + length,
22654 +               " %d\n", sched_pc->prio_bias);
22655 +       return length;
22656 +}
22657 +
22658 +#endif /* _VX_SCHED_PROC_H */
22659 diff -NurpP --minimal linux-2.6.36/kernel/vserver/signal.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/signal.c
22660 --- linux-2.6.36/kernel/vserver/signal.c        1970-01-01 01:00:00.000000000 +0100
22661 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/signal.c  2010-10-21 13:09:36.000000000 +0200
22662 @@ -0,0 +1,132 @@
22663 +/*
22664 + *  linux/kernel/vserver/signal.c
22665 + *
22666 + *  Virtual Server: Signal Support
22667 + *
22668 + *  Copyright (C) 2003-2007  Herbert Pötzl
22669 + *
22670 + *  V0.01  broken out from vcontext V0.05
22671 + *  V0.02  changed vcmds to vxi arg
22672 + *  V0.03  adjusted siginfo for kill
22673 + *
22674 + */
22675 +
22676 +#include <asm/uaccess.h>
22677 +
22678 +#include <linux/vs_context.h>
22679 +#include <linux/vs_pid.h>
22680 +#include <linux/vserver/signal_cmd.h>
22681 +
22682 +
22683 +int vx_info_kill(struct vx_info *vxi, int pid, int sig)
22684 +{
22685 +       int retval, count = 0;
22686 +       struct task_struct *p;
22687 +       struct siginfo *sip = SEND_SIG_PRIV;
22688 +
22689 +       retval = -ESRCH;
22690 +       vxdprintk(VXD_CBIT(misc, 4),
22691 +               "vx_info_kill(%p[#%d],%d,%d)*",
22692 +               vxi, vxi->vx_id, pid, sig);
22693 +       read_lock(&tasklist_lock);
22694 +       switch (pid) {
22695 +       case  0:
22696 +       case -1:
22697 +               for_each_process(p) {
22698 +                       int err = 0;
22699 +
22700 +                       if (vx_task_xid(p) != vxi->vx_id || p->pid <= 1 ||
22701 +                               (pid && vxi->vx_initpid == p->pid))
22702 +                               continue;
22703 +
22704 +                       err = group_send_sig_info(sig, sip, p);
22705 +                       ++count;
22706 +                       if (err != -EPERM)
22707 +                               retval = err;
22708 +               }
22709 +               break;
22710 +
22711 +       case 1:
22712 +               if (vxi->vx_initpid) {
22713 +                       pid = vxi->vx_initpid;
22714 +                       /* for now, only SIGINT to private init ... */
22715 +                       if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
22716 +                               /* ... as long as there are tasks left */
22717 +                               (atomic_read(&vxi->vx_tasks) > 1))
22718 +                               sig = SIGINT;
22719 +               }
22720 +               /* fallthrough */
22721 +       default:
22722 +               p = find_task_by_real_pid(pid);
22723 +               if (p) {
22724 +                       if (vx_task_xid(p) == vxi->vx_id)
22725 +                               retval = group_send_sig_info(sig, sip, p);
22726 +               }
22727 +               break;
22728 +       }
22729 +       read_unlock(&tasklist_lock);
22730 +       vxdprintk(VXD_CBIT(misc, 4),
22731 +               "vx_info_kill(%p[#%d],%d,%d,%ld) = %d",
22732 +               vxi, vxi->vx_id, pid, sig, (long)sip, retval);
22733 +       return retval;
22734 +}
22735 +
22736 +int vc_ctx_kill(struct vx_info *vxi, void __user *data)
22737 +{
22738 +       struct vcmd_ctx_kill_v0 vc_data;
22739 +
22740 +       if (copy_from_user(&vc_data, data, sizeof(vc_data)))
22741 +               return -EFAULT;
22742 +
22743 +       /* special check to allow guest shutdown */
22744 +       if (!vx_info_flags(vxi, VXF_STATE_ADMIN, 0) &&
22745 +               /* forbid killall pid=0 when init is present */
22746 +               (((vc_data.pid < 1) && vxi->vx_initpid) ||
22747 +               (vc_data.pid > 1)))
22748 +               return -EACCES;
22749 +
22750 +       return vx_info_kill(vxi, vc_data.pid, vc_data.sig);
22751 +}
22752 +
22753 +
22754 +static int __wait_exit(struct vx_info *vxi)
22755 +{
22756 +       DECLARE_WAITQUEUE(wait, current);
22757 +       int ret = 0;
22758 +
22759 +       add_wait_queue(&vxi->vx_wait, &wait);
22760 +       set_current_state(TASK_INTERRUPTIBLE);
22761 +
22762 +wait:
22763 +       if (vx_info_state(vxi,
22764 +               VXS_SHUTDOWN | VXS_HASHED | VXS_HELPER) == VXS_SHUTDOWN)
22765 +               goto out;
22766 +       if (signal_pending(current)) {
22767 +               ret = -ERESTARTSYS;
22768 +               goto out;
22769 +       }
22770 +       schedule();
22771 +       goto wait;
22772 +
22773 +out:
22774 +       set_current_state(TASK_RUNNING);
22775 +       remove_wait_queue(&vxi->vx_wait, &wait);
22776 +       return ret;
22777 +}
22778 +
22779 +
22780 +
22781 +int vc_wait_exit(struct vx_info *vxi, void __user *data)
22782 +{
22783 +       struct vcmd_wait_exit_v0 vc_data;
22784 +       int ret;
22785 +
22786 +       ret = __wait_exit(vxi);
22787 +       vc_data.reboot_cmd = vxi->reboot_cmd;
22788 +       vc_data.exit_code = vxi->exit_code;
22789 +
22790 +       if (copy_to_user(data, &vc_data, sizeof(vc_data)))
22791 +               ret = -EFAULT;
22792 +       return ret;
22793 +}
22794 +
22795 diff -NurpP --minimal linux-2.6.36/kernel/vserver/space.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/space.c
22796 --- linux-2.6.36/kernel/vserver/space.c 1970-01-01 01:00:00.000000000 +0100
22797 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/space.c   2010-10-21 14:41:06.000000000 +0200
22798 @@ -0,0 +1,375 @@
22799 +/*
22800 + *  linux/kernel/vserver/space.c
22801 + *
22802 + *  Virtual Server: Context Space Support
22803 + *
22804 + *  Copyright (C) 2003-2007  Herbert Pötzl
22805 + *
22806 + *  V0.01  broken out from context.c 0.07
22807 + *  V0.02  added task locking for namespace
22808 + *  V0.03  broken out vx_enter_namespace
22809 + *  V0.04  added *space support and commands
22810 + *
22811 + */
22812 +
22813 +#include <linux/utsname.h>
22814 +#include <linux/nsproxy.h>
22815 +#include <linux/err.h>
22816 +#include <linux/fs_struct.h>
22817 +#include <asm/uaccess.h>
22818 +
22819 +#include <linux/vs_context.h>
22820 +#include <linux/vserver/space.h>
22821 +#include <linux/vserver/space_cmd.h>
22822 +
22823 +atomic_t vs_global_nsproxy     = ATOMIC_INIT(0);
22824 +atomic_t vs_global_fs          = ATOMIC_INIT(0);
22825 +atomic_t vs_global_mnt_ns      = ATOMIC_INIT(0);
22826 +atomic_t vs_global_uts_ns      = ATOMIC_INIT(0);
22827 +atomic_t vs_global_user_ns     = ATOMIC_INIT(0);
22828 +atomic_t vs_global_pid_ns      = ATOMIC_INIT(0);
22829 +
22830 +
22831 +/* namespace functions */
22832 +
22833 +#include <linux/mnt_namespace.h>
22834 +#include <linux/user_namespace.h>
22835 +#include <linux/pid_namespace.h>
22836 +#include <linux/ipc_namespace.h>
22837 +#include <net/net_namespace.h>
22838 +
22839 +
22840 +static const struct vcmd_space_mask_v1 space_mask_v0 = {
22841 +       .mask = CLONE_FS |
22842 +               CLONE_NEWNS |
22843 +               CLONE_NEWUTS |
22844 +               CLONE_NEWIPC |
22845 +               CLONE_NEWUSER |
22846 +               0
22847 +};
22848 +
22849 +static const struct vcmd_space_mask_v1 space_mask = {
22850 +       .mask = CLONE_FS |
22851 +               CLONE_NEWNS |
22852 +               CLONE_NEWUTS |
22853 +               CLONE_NEWIPC |
22854 +               CLONE_NEWUSER |
22855 +#ifdef CONFIG_PID_NS
22856 +               CLONE_NEWPID |
22857 +#endif
22858 +#ifdef CONFIG_NET_NS
22859 +               CLONE_NEWNET |
22860 +#endif
22861 +               0
22862 +};
22863 +
22864 +static const struct vcmd_space_mask_v1 default_space_mask = {
22865 +       .mask = CLONE_FS |
22866 +               CLONE_NEWNS |
22867 +               CLONE_NEWUTS |
22868 +               CLONE_NEWIPC |
22869 +               CLONE_NEWUSER |
22870 +#ifdef CONFIG_PID_NS
22871 +//             CLONE_NEWPID |
22872 +#endif
22873 +               0
22874 +};
22875 +
22876 +/*
22877 + *     build a new nsproxy mix
22878 + *      assumes that both proxies are 'const'
22879 + *     does not touch nsproxy refcounts
22880 + *     will hold a reference on the result.
22881 + */
22882 +
22883 +struct nsproxy *vs_mix_nsproxy(struct nsproxy *old_nsproxy,
22884 +       struct nsproxy *new_nsproxy, unsigned long mask)
22885 +{
22886 +       struct mnt_namespace *old_ns;
22887 +       struct uts_namespace *old_uts;
22888 +       struct ipc_namespace *old_ipc;
22889 +#ifdef CONFIG_PID_NS
22890 +       struct pid_namespace *old_pid;
22891 +#endif
22892 +#ifdef CONFIG_NET_NS
22893 +       struct net *old_net;
22894 +#endif
22895 +       struct nsproxy *nsproxy;
22896 +
22897 +       nsproxy = copy_nsproxy(old_nsproxy);
22898 +       if (!nsproxy)
22899 +               goto out;
22900 +
22901 +       if (mask & CLONE_NEWNS) {
22902 +               old_ns = nsproxy->mnt_ns;
22903 +               nsproxy->mnt_ns = new_nsproxy->mnt_ns;
22904 +               if (nsproxy->mnt_ns)
22905 +                       get_mnt_ns(nsproxy->mnt_ns);
22906 +       } else
22907 +               old_ns = NULL;
22908 +
22909 +       if (mask & CLONE_NEWUTS) {
22910 +               old_uts = nsproxy->uts_ns;
22911 +               nsproxy->uts_ns = new_nsproxy->uts_ns;
22912 +               if (nsproxy->uts_ns)
22913 +                       get_uts_ns(nsproxy->uts_ns);
22914 +       } else
22915 +               old_uts = NULL;
22916 +
22917 +       if (mask & CLONE_NEWIPC) {
22918 +               old_ipc = nsproxy->ipc_ns;
22919 +               nsproxy->ipc_ns = new_nsproxy->ipc_ns;
22920 +               if (nsproxy->ipc_ns)
22921 +                       get_ipc_ns(nsproxy->ipc_ns);
22922 +       } else
22923 +               old_ipc = NULL;
22924 +
22925 +#ifdef CONFIG_PID_NS
22926 +       if (mask & CLONE_NEWPID) {
22927 +               old_pid = nsproxy->pid_ns;
22928 +               nsproxy->pid_ns = new_nsproxy->pid_ns;
22929 +               if (nsproxy->pid_ns)
22930 +                       get_pid_ns(nsproxy->pid_ns);
22931 +       } else
22932 +               old_pid = NULL;
22933 +#endif
22934 +#ifdef CONFIG_NET_NS
22935 +       if (mask & CLONE_NEWNET) {
22936 +               old_net = nsproxy->net_ns;
22937 +               nsproxy->net_ns = new_nsproxy->net_ns;
22938 +               if (nsproxy->net_ns)
22939 +                       get_net(nsproxy->net_ns);
22940 +       } else
22941 +               old_net = NULL;
22942 +#endif
22943 +       if (old_ns)
22944 +               put_mnt_ns(old_ns);
22945 +       if (old_uts)
22946 +               put_uts_ns(old_uts);
22947 +       if (old_ipc)
22948 +               put_ipc_ns(old_ipc);
22949 +#ifdef CONFIG_PID_NS
22950 +       if (old_pid)
22951 +               put_pid_ns(old_pid);
22952 +#endif
22953 +#ifdef CONFIG_NET_NS
22954 +       if (old_net)
22955 +               put_net(old_net);
22956 +#endif
22957 +out:
22958 +       return nsproxy;
22959 +}
22960 +
22961 +
22962 +/*
22963 + *     merge two nsproxy structs into a new one.
22964 + *     will hold a reference on the result.
22965 + */
22966 +
22967 +static inline
22968 +struct nsproxy *__vs_merge_nsproxy(struct nsproxy *old,
22969 +       struct nsproxy *proxy, unsigned long mask)
22970 +{
22971 +       struct nsproxy null_proxy = { .mnt_ns = NULL };
22972 +
22973 +       if (!proxy)
22974 +               return NULL;
22975 +
22976 +       if (mask) {
22977 +               /* vs_mix_nsproxy returns with reference */
22978 +               return vs_mix_nsproxy(old ? old : &null_proxy,
22979 +                       proxy, mask);
22980 +       }
22981 +       get_nsproxy(proxy);
22982 +       return proxy;
22983 +}
22984 +
22985 +
22986 +int vx_enter_space(struct vx_info *vxi, unsigned long mask, unsigned index)
22987 +{
22988 +       struct nsproxy *proxy, *proxy_cur, *proxy_new;
22989 +       struct fs_struct *fs_cur, *fs = NULL;
22990 +       int ret, kill = 0;
22991 +
22992 +       vxdprintk(VXD_CBIT(space, 8), "vx_enter_space(%p[#%u],0x%08lx,%d)",
22993 +               vxi, vxi->vx_id, mask, index);
22994 +
22995 +       if (vx_info_flags(vxi, VXF_INFO_PRIVATE, 0))
22996 +               return -EACCES;
22997 +
22998 +       if (!mask)
22999 +               mask = vxi->vx_nsmask[index];
23000 +
23001 +       if ((mask & vxi->vx_nsmask[index]) != mask)
23002 +               return -EINVAL;
23003 +
23004 +       if (mask & CLONE_FS) {
23005 +               fs = copy_fs_struct(vxi->vx_fs[index]);
23006 +               if (!fs)
23007 +                       return -ENOMEM;
23008 +       }
23009 +       proxy = vxi->vx_nsproxy[index];
23010 +
23011 +       vxdprintk(VXD_CBIT(space, 9),
23012 +               "vx_enter_space(%p[#%u],0x%08lx,%d) -> (%p,%p)",
23013 +               vxi, vxi->vx_id, mask, index, proxy, fs);
23014 +
23015 +       task_lock(current);
23016 +       fs_cur = current->fs;
23017 +
23018 +       if (mask & CLONE_FS) {
23019 +               spin_lock(&fs_cur->lock);
23020 +               current->fs = fs;
23021 +               kill = !--fs_cur->users;
23022 +               spin_unlock(&fs_cur->lock);
23023 +       }
23024 +
23025 +       proxy_cur = current->nsproxy;
23026 +       get_nsproxy(proxy_cur);
23027 +       task_unlock(current);
23028 +
23029 +       if (kill)
23030 +               free_fs_struct(fs_cur);
23031 +
23032 +       proxy_new = __vs_merge_nsproxy(proxy_cur, proxy, mask);
23033 +       if (IS_ERR(proxy_new)) {
23034 +               ret = PTR_ERR(proxy_new);
23035 +               goto out_put;
23036 +       }
23037 +
23038 +       proxy_new = xchg(&current->nsproxy, proxy_new);
23039 +       ret = 0;
23040 +
23041 +       if (proxy_new)
23042 +               put_nsproxy(proxy_new);
23043 +out_put:
23044 +       if (proxy_cur)
23045 +               put_nsproxy(proxy_cur);
23046 +       return ret;
23047 +}
23048 +
23049 +
23050 +int vx_set_space(struct vx_info *vxi, unsigned long mask, unsigned index)
23051 +{
23052 +       struct nsproxy *proxy_vxi, *proxy_cur, *proxy_new;
23053 +       struct fs_struct *fs_vxi, *fs;
23054 +       int ret, kill = 0;
23055 +
23056 +       vxdprintk(VXD_CBIT(space, 8), "vx_set_space(%p[#%u],0x%08lx,%d)",
23057 +               vxi, vxi->vx_id, mask, index);
23058 +#if 0
23059 +       if (!mask)
23060 +               mask = default_space_mask.mask;
23061 +#endif
23062 +       if ((mask & space_mask.mask) != mask)
23063 +               return -EINVAL;
23064 +
23065 +       proxy_vxi = vxi->vx_nsproxy[index];
23066 +       fs_vxi = vxi->vx_fs[index];
23067 +
23068 +       if (mask & CLONE_FS) {
23069 +               fs = copy_fs_struct(current->fs);
23070 +               if (!fs)
23071 +                       return -ENOMEM;
23072 +       }
23073 +
23074 +       task_lock(current);
23075 +
23076 +       if (mask & CLONE_FS) {
23077 +               spin_lock(&fs_vxi->lock);
23078 +               vxi->vx_fs[index] = fs;
23079 +               kill = !--fs_vxi->users;
23080 +               spin_unlock(&fs_vxi->lock);
23081 +       }
23082 +
23083 +       proxy_cur = current->nsproxy;
23084 +       get_nsproxy(proxy_cur);
23085 +       task_unlock(current);
23086 +
23087 +       if (kill)
23088 +               free_fs_struct(fs_vxi);
23089 +
23090 +       proxy_new = __vs_merge_nsproxy(proxy_vxi, proxy_cur, mask);
23091 +       if (IS_ERR(proxy_new)) {
23092 +               ret = PTR_ERR(proxy_new);
23093 +               goto out_put;
23094 +       }
23095 +
23096 +       proxy_new = xchg(&vxi->vx_nsproxy[index], proxy_new);
23097 +       vxi->vx_nsmask[index] |= mask;
23098 +       ret = 0;
23099 +
23100 +       if (proxy_new)
23101 +               put_nsproxy(proxy_new);
23102 +out_put:
23103 +       if (proxy_cur)
23104 +               put_nsproxy(proxy_cur);
23105 +       return ret;
23106 +}
23107 +
23108 +
23109 +int vc_enter_space_v1(struct vx_info *vxi, void __user *data)
23110 +{
23111 +       struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
23112 +
23113 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
23114 +               return -EFAULT;
23115 +
23116 +       return vx_enter_space(vxi, vc_data.mask, 0);
23117 +}
23118 +
23119 +int vc_enter_space(struct vx_info *vxi, void __user *data)
23120 +{
23121 +       struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
23122 +
23123 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
23124 +               return -EFAULT;
23125 +
23126 +       if (vc_data.index >= VX_SPACES)
23127 +               return -EINVAL;
23128 +
23129 +       return vx_enter_space(vxi, vc_data.mask, vc_data.index);
23130 +}
23131 +
23132 +int vc_set_space_v1(struct vx_info *vxi, void __user *data)
23133 +{
23134 +       struct vcmd_space_mask_v1 vc_data = { .mask = 0 };
23135 +
23136 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
23137 +               return -EFAULT;
23138 +
23139 +       return vx_set_space(vxi, vc_data.mask, 0);
23140 +}
23141 +
23142 +int vc_set_space(struct vx_info *vxi, void __user *data)
23143 +{
23144 +       struct vcmd_space_mask_v2 vc_data = { .mask = 0 };
23145 +
23146 +       if (data && copy_from_user(&vc_data, data, sizeof(vc_data)))
23147 +               return -EFAULT;
23148 +
23149 +       if (vc_data.index >= VX_SPACES)
23150 +               return -EINVAL;
23151 +
23152 +       return vx_set_space(vxi, vc_data.mask, vc_data.index);
23153 +}
23154 +
23155 +int vc_get_space_mask(void __user *data, int type)
23156 +{
23157 +       const struct vcmd_space_mask_v1 *mask;
23158 +
23159 +       if (type == 0)
23160 +               mask = &space_mask_v0;
23161 +       else if (type == 1)
23162 +               mask = &space_mask;
23163 +       else
23164 +               mask = &default_space_mask;
23165 +
23166 +       vxdprintk(VXD_CBIT(space, 10),
23167 +               "vc_get_space_mask(%d) = %08llx", type, mask->mask);
23168 +
23169 +       if (copy_to_user(data, mask, sizeof(*mask)))
23170 +               return -EFAULT;
23171 +       return 0;
23172 +}
23173 +
23174 diff -NurpP --minimal linux-2.6.36/kernel/vserver/switch.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/switch.c
23175 --- linux-2.6.36/kernel/vserver/switch.c        1970-01-01 01:00:00.000000000 +0100
23176 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/switch.c  2010-10-28 19:10:47.000000000 +0200
23177 @@ -0,0 +1,531 @@
23178 +/*
23179 + *  linux/kernel/vserver/switch.c
23180 + *
23181 + *  Virtual Server: Syscall Switch
23182 + *
23183 + *  Copyright (C) 2003-2007  Herbert Pötzl
23184 + *
23185 + *  V0.01  syscall switch
23186 + *  V0.02  added signal to context
23187 + *  V0.03  added rlimit functions
23188 + *  V0.04  added iattr, task/xid functions
23189 + *  V0.05  added debug/history stuff
23190 + *  V0.06  added compat32 layer
23191 + *  V0.07  vcmd args and perms
23192 + *  V0.08  added status commands
23193 + *  V0.09  added tag commands
23194 + *  V0.10  added oom bias
23195 + *  V0.11  added device commands
23196 + *
23197 + */
23198 +
23199 +#include <linux/vs_context.h>
23200 +#include <linux/vs_network.h>
23201 +#include <linux/vserver/switch.h>
23202 +
23203 +#include "vci_config.h"
23204 +
23205 +
23206 +static inline
23207 +int vc_get_version(uint32_t id)
23208 +{
23209 +       return VCI_VERSION;
23210 +}
23211 +
23212 +static inline
23213 +int vc_get_vci(uint32_t id)
23214 +{
23215 +       return vci_kernel_config();
23216 +}
23217 +
23218 +#include <linux/vserver/context_cmd.h>
23219 +#include <linux/vserver/cvirt_cmd.h>
23220 +#include <linux/vserver/cacct_cmd.h>
23221 +#include <linux/vserver/limit_cmd.h>
23222 +#include <linux/vserver/network_cmd.h>
23223 +#include <linux/vserver/sched_cmd.h>
23224 +#include <linux/vserver/debug_cmd.h>
23225 +#include <linux/vserver/inode_cmd.h>
23226 +#include <linux/vserver/dlimit_cmd.h>
23227 +#include <linux/vserver/signal_cmd.h>
23228 +#include <linux/vserver/space_cmd.h>
23229 +#include <linux/vserver/tag_cmd.h>
23230 +#include <linux/vserver/device_cmd.h>
23231 +
23232 +#include <linux/vserver/inode.h>
23233 +#include <linux/vserver/dlimit.h>
23234 +
23235 +
23236 +#ifdef CONFIG_COMPAT
23237 +#define __COMPAT(name, id, data, compat)       \
23238 +       (compat) ? name ## _x32(id, data) : name(id, data)
23239 +#define __COMPAT_NO_ID(name, data, compat)     \
23240 +       (compat) ? name ## _x32(data) : name(data)
23241 +#else
23242 +#define __COMPAT(name, id, data, compat)       \
23243 +       name(id, data)
23244 +#define __COMPAT_NO_ID(name, data, compat)     \
23245 +       name(data)
23246 +#endif
23247 +
23248 +
23249 +static inline
23250 +long do_vcmd(uint32_t cmd, uint32_t id,
23251 +       struct vx_info *vxi, struct nx_info *nxi,
23252 +       void __user *data, int compat)
23253 +{
23254 +       switch (cmd) {
23255 +
23256 +       case VCMD_get_version:
23257 +               return vc_get_version(id);
23258 +       case VCMD_get_vci:
23259 +               return vc_get_vci(id);
23260 +
23261 +       case VCMD_task_xid:
23262 +               return vc_task_xid(id);
23263 +       case VCMD_vx_info:
23264 +               return vc_vx_info(vxi, data);
23265 +
23266 +       case VCMD_task_nid:
23267 +               return vc_task_nid(id);
23268 +       case VCMD_nx_info:
23269 +               return vc_nx_info(nxi, data);
23270 +
23271 +       case VCMD_task_tag:
23272 +               return vc_task_tag(id);
23273 +
23274 +       case VCMD_set_space_v1:
23275 +               return vc_set_space_v1(vxi, data);
23276 +       /* this is version 2 */
23277 +       case VCMD_set_space:
23278 +               return vc_set_space(vxi, data);
23279 +
23280 +       case VCMD_get_space_mask_v0:
23281 +               return vc_get_space_mask(data, 0);
23282 +       /* this is version 1 */
23283 +       case VCMD_get_space_mask:
23284 +               return vc_get_space_mask(data, 1);
23285 +
23286 +       case VCMD_get_space_default:
23287 +               return vc_get_space_mask(data, -1);
23288 +
23289 +#ifdef CONFIG_IA32_EMULATION
23290 +       case VCMD_get_rlimit:
23291 +               return __COMPAT(vc_get_rlimit, vxi, data, compat);
23292 +       case VCMD_set_rlimit:
23293 +               return __COMPAT(vc_set_rlimit, vxi, data, compat);
23294 +#else
23295 +       case VCMD_get_rlimit:
23296 +               return vc_get_rlimit(vxi, data);
23297 +       case VCMD_set_rlimit:
23298 +               return vc_set_rlimit(vxi, data);
23299 +#endif
23300 +       case VCMD_get_rlimit_mask:
23301 +               return vc_get_rlimit_mask(id, data);
23302 +       case VCMD_reset_hits:
23303 +               return vc_reset_hits(vxi, data);
23304 +       case VCMD_reset_minmax:
23305 +               return vc_reset_minmax(vxi, data);
23306 +
23307 +       case VCMD_get_vhi_name:
23308 +               return vc_get_vhi_name(vxi, data);
23309 +       case VCMD_set_vhi_name:
23310 +               return vc_set_vhi_name(vxi, data);
23311 +
23312 +       case VCMD_ctx_stat:
23313 +               return vc_ctx_stat(vxi, data);
23314 +       case VCMD_virt_stat:
23315 +               return vc_virt_stat(vxi, data);
23316 +       case VCMD_sock_stat:
23317 +               return vc_sock_stat(vxi, data);
23318 +       case VCMD_rlimit_stat:
23319 +               return vc_rlimit_stat(vxi, data);
23320 +
23321 +       case VCMD_set_cflags:
23322 +               return vc_set_cflags(vxi, data);
23323 +       case VCMD_get_cflags:
23324 +               return vc_get_cflags(vxi, data);
23325 +
23326 +       /* this is version 1 */
23327 +       case VCMD_set_ccaps:
23328 +               return vc_set_ccaps(vxi, data);
23329 +       /* this is version 1 */
23330 +       case VCMD_get_ccaps:
23331 +               return vc_get_ccaps(vxi, data);
23332 +       case VCMD_set_bcaps:
23333 +               return vc_set_bcaps(vxi, data);
23334 +       case VCMD_get_bcaps:
23335 +               return vc_get_bcaps(vxi, data);
23336 +
23337 +       case VCMD_set_badness:
23338 +               return vc_set_badness(vxi, data);
23339 +       case VCMD_get_badness:
23340 +               return vc_get_badness(vxi, data);
23341 +
23342 +       case VCMD_set_nflags:
23343 +               return vc_set_nflags(nxi, data);
23344 +       case VCMD_get_nflags:
23345 +               return vc_get_nflags(nxi, data);
23346 +
23347 +       case VCMD_set_ncaps:
23348 +               return vc_set_ncaps(nxi, data);
23349 +       case VCMD_get_ncaps:
23350 +               return vc_get_ncaps(nxi, data);
23351 +
23352 +       case VCMD_set_prio_bias:
23353 +               return vc_set_prio_bias(vxi, data);
23354 +       case VCMD_get_prio_bias:
23355 +               return vc_get_prio_bias(vxi, data);
23356 +       case VCMD_add_dlimit:
23357 +               return __COMPAT(vc_add_dlimit, id, data, compat);
23358 +       case VCMD_rem_dlimit:
23359 +               return __COMPAT(vc_rem_dlimit, id, data, compat);
23360 +       case VCMD_set_dlimit:
23361 +               return __COMPAT(vc_set_dlimit, id, data, compat);
23362 +       case VCMD_get_dlimit:
23363 +               return __COMPAT(vc_get_dlimit, id, data, compat);
23364 +
23365 +       case VCMD_ctx_kill:
23366 +               return vc_ctx_kill(vxi, data);
23367 +
23368 +       case VCMD_wait_exit:
23369 +               return vc_wait_exit(vxi, data);
23370 +
23371 +       case VCMD_get_iattr:
23372 +               return __COMPAT_NO_ID(vc_get_iattr, data, compat);
23373 +       case VCMD_set_iattr:
23374 +               return __COMPAT_NO_ID(vc_set_iattr, data, compat);
23375 +
23376 +       case VCMD_fget_iattr:
23377 +               return vc_fget_iattr(id, data);
23378 +       case VCMD_fset_iattr:
23379 +               return vc_fset_iattr(id, data);
23380 +
23381 +       case VCMD_enter_space_v0:
23382 +               return vc_enter_space_v1(vxi, NULL);
23383 +       case VCMD_enter_space_v1:
23384 +               return vc_enter_space_v1(vxi, data);
23385 +       /* this is version 2 */
23386 +       case VCMD_enter_space:
23387 +               return vc_enter_space(vxi, data);
23388 +
23389 +       case VCMD_ctx_create_v0:
23390 +               return vc_ctx_create(id, NULL);
23391 +       case VCMD_ctx_create:
23392 +               return vc_ctx_create(id, data);
23393 +       case VCMD_ctx_migrate_v0:
23394 +               return vc_ctx_migrate(vxi, NULL);
23395 +       case VCMD_ctx_migrate:
23396 +               return vc_ctx_migrate(vxi, data);
23397 +
23398 +       case VCMD_net_create_v0:
23399 +               return vc_net_create(id, NULL);
23400 +       case VCMD_net_create:
23401 +               return vc_net_create(id, data);
23402 +       case VCMD_net_migrate:
23403 +               return vc_net_migrate(nxi, data);
23404 +
23405 +       case VCMD_tag_migrate:
23406 +               return vc_tag_migrate(id);
23407 +
23408 +       case VCMD_net_add:
23409 +               return vc_net_add(nxi, data);
23410 +       case VCMD_net_remove:
23411 +               return vc_net_remove(nxi, data);
23412 +
23413 +       case VCMD_net_add_ipv4:
23414 +               return vc_net_add_ipv4(nxi, data);
23415 +       case VCMD_net_remove_ipv4:
23416 +               return vc_net_remove_ipv4(nxi, data);
23417 +#ifdef CONFIG_IPV6
23418 +       case VCMD_net_add_ipv6:
23419 +               return vc_net_add_ipv6(nxi, data);
23420 +       case VCMD_net_remove_ipv6:
23421 +               return vc_net_remove_ipv6(nxi, data);
23422 +#endif
23423 +/*     case VCMD_add_match_ipv4:
23424 +               return vc_add_match_ipv4(nxi, data);
23425 +       case VCMD_get_match_ipv4:
23426 +               return vc_get_match_ipv4(nxi, data);
23427 +#ifdef CONFIG_IPV6
23428 +       case VCMD_add_match_ipv6:
23429 +               return vc_add_match_ipv6(nxi, data);
23430 +       case VCMD_get_match_ipv6:
23431 +               return vc_get_match_ipv6(nxi, data);
23432 +#endif */
23433 +
23434 +#ifdef CONFIG_VSERVER_DEVICE
23435 +       case VCMD_set_mapping:
23436 +               return __COMPAT(vc_set_mapping, vxi, data, compat);
23437 +       case VCMD_unset_mapping:
23438 +               return __COMPAT(vc_unset_mapping, vxi, data, compat);
23439 +#endif
23440 +#ifdef CONFIG_VSERVER_HISTORY
23441 +       case VCMD_dump_history:
23442 +               return vc_dump_history(id);
23443 +       case VCMD_read_history:
23444 +               return __COMPAT(vc_read_history, id, data, compat);
23445 +#endif
23446 +       default:
23447 +               vxwprintk_task(1, "unimplemented VCMD_%02d_%d[%d]",
23448 +                       VC_CATEGORY(cmd), VC_COMMAND(cmd), VC_VERSION(cmd));
23449 +       }
23450 +       return -ENOSYS;
23451 +}
23452 +
23453 +
23454 +#define        __VCMD(vcmd, _perm, _args, _flags)              \
23455 +       case VCMD_ ## vcmd: perm = _perm;               \
23456 +               args = _args; flags = _flags; break
23457 +
23458 +
23459 +#define VCA_NONE       0x00
23460 +#define VCA_VXI                0x01
23461 +#define VCA_NXI                0x02
23462 +
23463 +#define VCF_NONE       0x00
23464 +#define VCF_INFO       0x01
23465 +#define VCF_ADMIN      0x02
23466 +#define VCF_ARES       0x06    /* includes admin */
23467 +#define VCF_SETUP      0x08
23468 +
23469 +#define VCF_ZIDOK      0x10    /* zero id okay */
23470 +
23471 +
23472 +static inline
23473 +long do_vserver(uint32_t cmd, uint32_t id, void __user *data, int compat)
23474 +{
23475 +       long ret;
23476 +       int permit = -1, state = 0;
23477 +       int perm = -1, args = 0, flags = 0;
23478 +       struct vx_info *vxi = NULL;
23479 +       struct nx_info *nxi = NULL;
23480 +
23481 +       switch (cmd) {
23482 +       /* unpriviledged commands */
23483 +       __VCMD(get_version,      0, VCA_NONE,   0);
23484 +       __VCMD(get_vci,          0, VCA_NONE,   0);
23485 +       __VCMD(get_rlimit_mask,  0, VCA_NONE,   0);
23486 +       __VCMD(get_space_mask_v0,0, VCA_NONE,   0);
23487 +       __VCMD(get_space_mask,   0, VCA_NONE,   0);
23488 +       __VCMD(get_space_default,0, VCA_NONE,   0);
23489 +
23490 +       /* info commands */
23491 +       __VCMD(task_xid,         2, VCA_NONE,   0);
23492 +       __VCMD(reset_hits,       2, VCA_VXI,    0);
23493 +       __VCMD(reset_minmax,     2, VCA_VXI,    0);
23494 +       __VCMD(vx_info,          3, VCA_VXI,    VCF_INFO);
23495 +       __VCMD(get_bcaps,        3, VCA_VXI,    VCF_INFO);
23496 +       __VCMD(get_ccaps,        3, VCA_VXI,    VCF_INFO);
23497 +       __VCMD(get_cflags,       3, VCA_VXI,    VCF_INFO);
23498 +       __VCMD(get_badness,      3, VCA_VXI,    VCF_INFO);
23499 +       __VCMD(get_vhi_name,     3, VCA_VXI,    VCF_INFO);
23500 +       __VCMD(get_rlimit,       3, VCA_VXI,    VCF_INFO);
23501 +
23502 +       __VCMD(ctx_stat,         3, VCA_VXI,    VCF_INFO);
23503 +       __VCMD(virt_stat,        3, VCA_VXI,    VCF_INFO);
23504 +       __VCMD(sock_stat,        3, VCA_VXI,    VCF_INFO);
23505 +       __VCMD(rlimit_stat,      3, VCA_VXI,    VCF_INFO);
23506 +
23507 +       __VCMD(task_nid,         2, VCA_NONE,   0);
23508 +       __VCMD(nx_info,          3, VCA_NXI,    VCF_INFO);
23509 +       __VCMD(get_ncaps,        3, VCA_NXI,    VCF_INFO);
23510 +       __VCMD(get_nflags,       3, VCA_NXI,    VCF_INFO);
23511 +
23512 +       __VCMD(task_tag,         2, VCA_NONE,   0);
23513 +
23514 +       __VCMD(get_iattr,        2, VCA_NONE,   0);
23515 +       __VCMD(fget_iattr,       2, VCA_NONE,   0);
23516 +       __VCMD(get_dlimit,       3, VCA_NONE,   VCF_INFO);
23517 +       __VCMD(get_prio_bias,    3, VCA_VXI,    VCF_INFO);
23518 +
23519 +       /* lower admin commands */
23520 +       __VCMD(wait_exit,        4, VCA_VXI,    VCF_INFO);
23521 +       __VCMD(ctx_create_v0,    5, VCA_NONE,   0);
23522 +       __VCMD(ctx_create,       5, VCA_NONE,   0);
23523 +       __VCMD(ctx_migrate_v0,   5, VCA_VXI,    VCF_ADMIN);
23524 +       __VCMD(ctx_migrate,      5, VCA_VXI,    VCF_ADMIN);
23525 +       __VCMD(enter_space_v0,   5, VCA_VXI,    VCF_ADMIN);
23526 +       __VCMD(enter_space_v1,   5, VCA_VXI,    VCF_ADMIN);
23527 +       __VCMD(enter_space,      5, VCA_VXI,    VCF_ADMIN);
23528 +
23529 +       __VCMD(net_create_v0,    5, VCA_NONE,   0);
23530 +       __VCMD(net_create,       5, VCA_NONE,   0);
23531 +       __VCMD(net_migrate,      5, VCA_NXI,    VCF_ADMIN);
23532 +
23533 +       __VCMD(tag_migrate,      5, VCA_NONE,   VCF_ADMIN);
23534 +
23535 +       /* higher admin commands */
23536 +       __VCMD(ctx_kill,         6, VCA_VXI,    VCF_ARES);
23537 +       __VCMD(set_space_v1,     7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23538 +       __VCMD(set_space,        7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23539 +
23540 +       __VCMD(set_ccaps,        7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23541 +       __VCMD(set_bcaps,        7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23542 +       __VCMD(set_cflags,       7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23543 +       __VCMD(set_badness,      7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23544 +
23545 +       __VCMD(set_vhi_name,     7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23546 +       __VCMD(set_rlimit,       7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23547 +       __VCMD(set_prio_bias,    7, VCA_VXI,    VCF_ARES | VCF_SETUP);
23548 +
23549 +       __VCMD(set_ncaps,        7, VCA_NXI,    VCF_ARES | VCF_SETUP);
23550 +       __VCMD(set_nflags,       7, VCA_NXI,    VCF_ARES | VCF_SETUP);
23551 +       __VCMD(net_add,          8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23552 +       __VCMD(net_remove,       8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23553 +       __VCMD(net_add_ipv4,     8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23554 +       __VCMD(net_remove_ipv4,  8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23555 +#ifdef CONFIG_IPV6
23556 +       __VCMD(net_add_ipv6,     8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23557 +       __VCMD(net_remove_ipv6,  8, VCA_NXI,    VCF_ARES | VCF_SETUP);
23558 +#endif
23559 +       __VCMD(set_iattr,        7, VCA_NONE,   0);
23560 +       __VCMD(fset_iattr,       7, VCA_NONE,   0);
23561 +       __VCMD(set_dlimit,       7, VCA_NONE,   VCF_ARES);
23562 +       __VCMD(add_dlimit,       8, VCA_NONE,   VCF_ARES);
23563 +       __VCMD(rem_dlimit,       8, VCA_NONE,   VCF_ARES);
23564 +
23565 +#ifdef CONFIG_VSERVER_DEVICE
23566 +       __VCMD(set_mapping,      8, VCA_VXI,    VCF_ARES|VCF_ZIDOK);
23567 +       __VCMD(unset_mapping,    8, VCA_VXI,    VCF_ARES|VCF_ZIDOK);
23568 +#endif
23569 +       /* debug level admin commands */
23570 +#ifdef CONFIG_VSERVER_HISTORY
23571 +       __VCMD(dump_history,     9, VCA_NONE,   0);
23572 +       __VCMD(read_history,     9, VCA_NONE,   0);
23573 +#endif
23574 +
23575 +       default:
23576 +               perm = -1;
23577 +       }
23578 +
23579 +       vxdprintk(VXD_CBIT(switch, 0),
23580 +               "vc: VCMD_%02d_%d[%d], %d,%p [%d,%d,%x,%x]",
23581 +               VC_CATEGORY(cmd), VC_COMMAND(cmd),
23582 +               VC_VERSION(cmd), id, data, compat,
23583 +               perm, args, flags);
23584 +
23585 +       ret = -ENOSYS;
23586 +       if (perm < 0)
23587 +               goto out;
23588 +
23589 +       state = 1;
23590 +       if (!capable(CAP_CONTEXT))
23591 +               goto out;
23592 +
23593 +       state = 2;
23594 +       /* moved here from the individual commands */
23595 +       ret = -EPERM;
23596 +       if ((perm > 1) && !capable(CAP_SYS_ADMIN))
23597 +               goto out;
23598 +
23599 +       state = 3;
23600 +       /* vcmd involves resource management  */
23601 +       ret = -EPERM;
23602 +       if ((flags & VCF_ARES) && !capable(CAP_SYS_RESOURCE))
23603 +               goto out;
23604 +
23605 +       state = 4;
23606 +       /* various legacy exceptions */
23607 +       switch (cmd) {
23608 +       /* will go away when spectator is a cap */
23609 +       case VCMD_ctx_migrate_v0:
23610 +       case VCMD_ctx_migrate:
23611 +               if (id == 1) {
23612 +                       current->xid = 1;
23613 +                       ret = 1;
23614 +                       goto out;
23615 +               }
23616 +               break;
23617 +
23618 +       /* will go away when spectator is a cap */
23619 +       case VCMD_net_migrate:
23620 +               if (id == 1) {
23621 +                       current->nid = 1;
23622 +                       ret = 1;
23623 +                       goto out;
23624 +               }
23625 +               break;
23626 +       }
23627 +
23628 +       /* vcmds are fine by default */
23629 +       permit = 1;
23630 +
23631 +       /* admin type vcmds require admin ... */
23632 +       if (flags & VCF_ADMIN)
23633 +               permit = vx_check(0, VS_ADMIN) ? 1 : 0;
23634 +
23635 +       /* ... but setup type vcmds override that */
23636 +       if (!permit && (flags & VCF_SETUP))
23637 +               permit = vx_flags(VXF_STATE_SETUP, 0) ? 2 : 0;
23638 +
23639 +       state = 5;
23640 +       ret = -EPERM;
23641 +       if (!permit)
23642 +               goto out;
23643 +
23644 +       state = 6;
23645 +       if (!id && (flags & VCF_ZIDOK))
23646 +               goto skip_id;
23647 +
23648 +       ret = -ESRCH;
23649 +       if (args & VCA_VXI) {
23650 +               vxi = lookup_vx_info(id);
23651 +               if (!vxi)
23652 +                       goto out;
23653 +
23654 +               if ((flags & VCF_ADMIN) &&
23655 +                       /* special case kill for shutdown */
23656 +                       (cmd != VCMD_ctx_kill) &&
23657 +                       /* can context be administrated? */
23658 +                       !vx_info_flags(vxi, VXF_STATE_ADMIN, 0)) {
23659 +                       ret = -EACCES;
23660 +                       goto out_vxi;
23661 +               }
23662 +       }
23663 +       state = 7;
23664 +       if (args & VCA_NXI) {
23665 +               nxi = lookup_nx_info(id);
23666 +               if (!nxi)
23667 +                       goto out_vxi;
23668 +
23669 +               if ((flags & VCF_ADMIN) &&
23670 +                       /* can context be administrated? */
23671 +                       !nx_info_flags(nxi, NXF_STATE_ADMIN, 0)) {
23672 +                       ret = -EACCES;
23673 +                       goto out_nxi;
23674 +               }
23675 +       }
23676 +skip_id:
23677 +       state = 8;
23678 +       ret = do_vcmd(cmd, id, vxi, nxi, data, compat);
23679 +
23680 +out_nxi:
23681 +       if ((args & VCA_NXI) && nxi)
23682 +               put_nx_info(nxi);
23683 +out_vxi:
23684 +       if ((args & VCA_VXI) && vxi)
23685 +               put_vx_info(vxi);
23686 +out:
23687 +       vxdprintk(VXD_CBIT(switch, 1),
23688 +               "vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]",
23689 +               VC_CATEGORY(cmd), VC_COMMAND(cmd),
23690 +               VC_VERSION(cmd), ret, ret, state, permit);
23691 +       return ret;
23692 +}
23693 +
23694 +asmlinkage long
23695 +sys_vserver(uint32_t cmd, uint32_t id, void __user *data)
23696 +{
23697 +       return do_vserver(cmd, id, data, 0);
23698 +}
23699 +
23700 +#ifdef CONFIG_COMPAT
23701 +
23702 +asmlinkage long
23703 +sys32_vserver(uint32_t cmd, uint32_t id, void __user *data)
23704 +{
23705 +       return do_vserver(cmd, id, data, 1);
23706 +}
23707 +
23708 +#endif /* CONFIG_COMPAT */
23709 diff -NurpP --minimal linux-2.6.36/kernel/vserver/sysctl.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sysctl.c
23710 --- linux-2.6.36/kernel/vserver/sysctl.c        1970-01-01 01:00:00.000000000 +0100
23711 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/sysctl.c  2010-10-21 13:09:36.000000000 +0200
23712 @@ -0,0 +1,241 @@
23713 +/*
23714 + *  kernel/vserver/sysctl.c
23715 + *
23716 + *  Virtual Context Support
23717 + *
23718 + *  Copyright (C) 2004-2007  Herbert Pötzl
23719 + *
23720 + *  V0.01  basic structure
23721 + *
23722 + */
23723 +
23724 +#include <linux/module.h>
23725 +#include <linux/ctype.h>
23726 +#include <linux/sysctl.h>
23727 +#include <linux/parser.h>
23728 +#include <asm/uaccess.h>
23729 +
23730 +enum {
23731 +       CTL_DEBUG_ERROR         = 0,
23732 +       CTL_DEBUG_SWITCH        = 1,
23733 +       CTL_DEBUG_XID,
23734 +       CTL_DEBUG_NID,
23735 +       CTL_DEBUG_TAG,
23736 +       CTL_DEBUG_NET,
23737 +       CTL_DEBUG_LIMIT,
23738 +       CTL_DEBUG_CRES,
23739 +       CTL_DEBUG_DLIM,
23740 +       CTL_DEBUG_QUOTA,
23741 +       CTL_DEBUG_CVIRT,
23742 +       CTL_DEBUG_SPACE,
23743 +       CTL_DEBUG_MISC,
23744 +};
23745 +
23746 +
23747 +unsigned int vx_debug_switch   = 0;
23748 +unsigned int vx_debug_xid      = 0;
23749 +unsigned int vx_debug_nid      = 0;
23750 +unsigned int vx_debug_tag      = 0;
23751 +unsigned int vx_debug_net      = 0;
23752 +unsigned int vx_debug_limit    = 0;
23753 +unsigned int vx_debug_cres     = 0;
23754 +unsigned int vx_debug_dlim     = 0;
23755 +unsigned int vx_debug_quota    = 0;
23756 +unsigned int vx_debug_cvirt    = 0;
23757 +unsigned int vx_debug_space    = 0;
23758 +unsigned int vx_debug_misc     = 0;
23759 +
23760 +
23761 +static struct ctl_table_header *vserver_table_header;
23762 +static ctl_table vserver_root_table[];
23763 +
23764 +
23765 +void vserver_register_sysctl(void)
23766 +{
23767 +       if (!vserver_table_header) {
23768 +               vserver_table_header = register_sysctl_table(vserver_root_table);
23769 +       }
23770 +
23771 +}
23772 +
23773 +void vserver_unregister_sysctl(void)
23774 +{
23775 +       if (vserver_table_header) {
23776 +               unregister_sysctl_table(vserver_table_header);
23777 +               vserver_table_header = NULL;
23778 +       }
23779 +}
23780 +
23781 +
23782 +static int proc_dodebug(ctl_table *table, int write,
23783 +       void __user *buffer, size_t *lenp, loff_t *ppos)
23784 +{
23785 +       char            tmpbuf[20], *p, c;
23786 +       unsigned int    value;
23787 +       size_t          left, len;
23788 +
23789 +       if ((*ppos && !write) || !*lenp) {
23790 +               *lenp = 0;
23791 +               return 0;
23792 +       }
23793 +
23794 +       left = *lenp;
23795 +
23796 +       if (write) {
23797 +               if (!access_ok(VERIFY_READ, buffer, left))
23798 +                       return -EFAULT;
23799 +               p = (char *)buffer;
23800 +               while (left && __get_user(c, p) >= 0 && isspace(c))
23801 +                       left--, p++;
23802 +               if (!left)
23803 +                       goto done;
23804 +
23805 +               if (left > sizeof(tmpbuf) - 1)
23806 +                       return -EINVAL;
23807 +               if (copy_from_user(tmpbuf, p, left))
23808 +                       return -EFAULT;
23809 +               tmpbuf[left] = '\0';
23810 +
23811 +               for (p = tmpbuf, value = 0; '0' <= *p && *p <= '9'; p++, left--)
23812 +                       value = 10 * value + (*p - '0');
23813 +               if (*p && !isspace(*p))
23814 +                       return -EINVAL;
23815 +               while (left && isspace(*p))
23816 +                       left--, p++;
23817 +               *(unsigned int *)table->data = value;
23818 +       } else {
23819 +               if (!access_ok(VERIFY_WRITE, buffer, left))
23820 +                       return -EFAULT;
23821 +               len = sprintf(tmpbuf, "%d", *(unsigned int *)table->data);
23822 +               if (len > left)
23823 +                       len = left;
23824 +               if (__copy_to_user(buffer, tmpbuf, len))
23825 +                       return -EFAULT;
23826 +               if ((left -= len) > 0) {
23827 +                       if (put_user('\n', (char *)buffer + len))
23828 +                               return -EFAULT;
23829 +                       left--;
23830 +               }
23831 +       }
23832 +
23833 +done:
23834 +       *lenp -= left;
23835 +       *ppos += *lenp;
23836 +       return 0;
23837 +}
23838 +
23839 +static int zero;
23840 +
23841 +#define        CTL_ENTRY(ctl, name)                            \
23842 +       {                                               \
23843 +               .procname       = #name,                \
23844 +               .data           = &vx_ ## name,         \
23845 +               .maxlen         = sizeof(int),          \
23846 +               .mode           = 0644,                 \
23847 +               .proc_handler   = &proc_dodebug,        \
23848 +               .extra1         = &zero,                \
23849 +               .extra2         = &zero,                \
23850 +       }
23851 +
23852 +static ctl_table vserver_debug_table[] = {
23853 +       CTL_ENTRY(CTL_DEBUG_SWITCH,     debug_switch),
23854 +       CTL_ENTRY(CTL_DEBUG_XID,        debug_xid),
23855 +       CTL_ENTRY(CTL_DEBUG_NID,        debug_nid),
23856 +       CTL_ENTRY(CTL_DEBUG_TAG,        debug_tag),
23857 +       CTL_ENTRY(CTL_DEBUG_NET,        debug_net),
23858 +       CTL_ENTRY(CTL_DEBUG_LIMIT,      debug_limit),
23859 +       CTL_ENTRY(CTL_DEBUG_CRES,       debug_cres),
23860 +       CTL_ENTRY(CTL_DEBUG_DLIM,       debug_dlim),
23861 +       CTL_ENTRY(CTL_DEBUG_QUOTA,      debug_quota),
23862 +       CTL_ENTRY(CTL_DEBUG_CVIRT,      debug_cvirt),
23863 +       CTL_ENTRY(CTL_DEBUG_SPACE,      debug_space),
23864 +       CTL_ENTRY(CTL_DEBUG_MISC,       debug_misc),
23865 +       { 0 }
23866 +};
23867 +
23868 +static ctl_table vserver_root_table[] = {
23869 +       {
23870 +               .procname       = "vserver",
23871 +               .mode           = 0555,
23872 +               .child          = vserver_debug_table
23873 +       },
23874 +       { 0 }
23875 +};
23876 +
23877 +
23878 +static match_table_t tokens = {
23879 +       { CTL_DEBUG_SWITCH,     "switch=%x"     },
23880 +       { CTL_DEBUG_XID,        "xid=%x"        },
23881 +       { CTL_DEBUG_NID,        "nid=%x"        },
23882 +       { CTL_DEBUG_TAG,        "tag=%x"        },
23883 +       { CTL_DEBUG_NET,        "net=%x"        },
23884 +       { CTL_DEBUG_LIMIT,      "limit=%x"      },
23885 +       { CTL_DEBUG_CRES,       "cres=%x"       },
23886 +       { CTL_DEBUG_DLIM,       "dlim=%x"       },
23887 +       { CTL_DEBUG_QUOTA,      "quota=%x"      },
23888 +       { CTL_DEBUG_CVIRT,      "cvirt=%x"      },
23889 +       { CTL_DEBUG_SPACE,      "space=%x"      },
23890 +       { CTL_DEBUG_MISC,       "misc=%x"       },
23891 +       { CTL_DEBUG_ERROR,      NULL            }
23892 +};
23893 +
23894 +#define        HANDLE_CASE(id, name, val)                              \
23895 +       case CTL_DEBUG_ ## id:                                  \
23896 +               vx_debug_ ## name = val;                        \
23897 +               printk("vs_debug_" #name "=0x%x\n", val);       \
23898 +               break
23899 +
23900 +
23901 +static int __init vs_debug_setup(char *str)
23902 +{
23903 +       char *p;
23904 +       int token;
23905 +
23906 +       printk("vs_debug_setup(%s)\n", str);
23907 +       while ((p = strsep(&str, ",")) != NULL) {
23908 +               substring_t args[MAX_OPT_ARGS];
23909 +               unsigned int value;
23910 +
23911 +               if (!*p)
23912 +                       continue;
23913 +
23914 +               token = match_token(p, tokens, args);
23915 +               value = (token > 0) ? simple_strtoul(args[0].from, NULL, 0) : 0;
23916 +
23917 +               switch (token) {
23918 +               HANDLE_CASE(SWITCH, switch, value);
23919 +               HANDLE_CASE(XID,    xid,    value);
23920 +               HANDLE_CASE(NID,    nid,    value);
23921 +               HANDLE_CASE(TAG,    tag,    value);
23922 +               HANDLE_CASE(NET,    net,    value);
23923 +               HANDLE_CASE(LIMIT,  limit,  value);
23924 +               HANDLE_CASE(CRES,   cres,   value);
23925 +               HANDLE_CASE(DLIM,   dlim,   value);
23926 +               HANDLE_CASE(QUOTA,  quota,  value);
23927 +               HANDLE_CASE(CVIRT,  cvirt,  value);
23928 +               HANDLE_CASE(SPACE,  space,  value);
23929 +               HANDLE_CASE(MISC,   misc,   value);
23930 +               default:
23931 +                       return -EINVAL;
23932 +                       break;
23933 +               }
23934 +       }
23935 +       return 1;
23936 +}
23937 +
23938 +__setup("vsdebug=", vs_debug_setup);
23939 +
23940 +
23941 +
23942 +EXPORT_SYMBOL_GPL(vx_debug_switch);
23943 +EXPORT_SYMBOL_GPL(vx_debug_xid);
23944 +EXPORT_SYMBOL_GPL(vx_debug_nid);
23945 +EXPORT_SYMBOL_GPL(vx_debug_net);
23946 +EXPORT_SYMBOL_GPL(vx_debug_limit);
23947 +EXPORT_SYMBOL_GPL(vx_debug_cres);
23948 +EXPORT_SYMBOL_GPL(vx_debug_dlim);
23949 +EXPORT_SYMBOL_GPL(vx_debug_quota);
23950 +EXPORT_SYMBOL_GPL(vx_debug_cvirt);
23951 +EXPORT_SYMBOL_GPL(vx_debug_space);
23952 +EXPORT_SYMBOL_GPL(vx_debug_misc);
23953 +
23954 diff -NurpP --minimal linux-2.6.36/kernel/vserver/tag.c linux-2.6.36-vs2.3.0.36.38/kernel/vserver/tag.c
23955 --- linux-2.6.36/kernel/vserver/tag.c   1970-01-01 01:00:00.000000000 +0100
23956 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/tag.c     2010-10-21 13:09:36.000000000 +0200
23957 @@ -0,0 +1,63 @@
23958 +/*
23959 + *  linux/kernel/vserver/tag.c
23960 + *
23961 + *  Virtual Server: Shallow Tag Space
23962 + *
23963 + *  Copyright (C) 2007  Herbert Pötzl
23964 + *
23965 + *  V0.01  basic implementation
23966 + *
23967 + */
23968 +
23969 +#include <linux/sched.h>
23970 +#include <linux/vserver/debug.h>
23971 +#include <linux/vs_pid.h>
23972 +#include <linux/vs_tag.h>
23973 +
23974 +#include <linux/vserver/tag_cmd.h>
23975 +
23976 +
23977 +int dx_migrate_task(struct task_struct *p, tag_t tag)
23978 +{
23979 +       if (!p)
23980 +               BUG();
23981 +
23982 +       vxdprintk(VXD_CBIT(tag, 5),
23983 +               "dx_migrate_task(%p[#%d],#%d)", p, p->tag, tag);
23984 +
23985 +       task_lock(p);
23986 +       p->tag = tag;
23987 +       task_unlock(p);
23988 +
23989 +       vxdprintk(VXD_CBIT(tag, 5),
23990 +               "moved task %p into [#%d]", p, tag);
23991 +       return 0;
23992 +}
23993 +
23994 +/* vserver syscall commands below here */
23995 +
23996 +/* taks xid and vx_info functions */
23997 +
23998 +
23999 +int vc_task_tag(uint32_t id)
24000 +{
24001 +       tag_t tag;
24002 +
24003 +       if (id) {
24004 +               struct task_struct *tsk;
24005 +               read_lock(&tasklist_lock);
24006 +               tsk = find_task_by_real_pid(id);
24007 +               tag = (tsk) ? tsk->tag : -ESRCH;
24008 +               read_unlock(&tasklist_lock);
24009 +       } else
24010 +               tag = dx_current_tag();
24011 +       return tag;
24012 +}
24013 +
24014 +
24015 +int vc_tag_migrate(uint32_t tag)
24016 +{
24017 +       return dx_migrate_task(current, tag & 0xFFFF);
24018 +}
24019 +
24020 +
24021 diff -NurpP --minimal linux-2.6.36/kernel/vserver/vci_config.h linux-2.6.36-vs2.3.0.36.38/kernel/vserver/vci_config.h
24022 --- linux-2.6.36/kernel/vserver/vci_config.h    1970-01-01 01:00:00.000000000 +0100
24023 +++ linux-2.6.36-vs2.3.0.36.38/kernel/vserver/vci_config.h      2010-10-28 20:45:52.000000000 +0200
24024 @@ -0,0 +1,82 @@
24025 +
24026 +/*  interface version */
24027 +
24028 +#define VCI_VERSION            0x00020307
24029 +
24030 +
24031 +enum {
24032 +       VCI_KCBIT_NO_DYNAMIC = 0,
24033 +
24034 +       VCI_KCBIT_PROC_SECURE = 4,
24035 +       VCI_KCBIT_HARDCPU = 5,
24036 +       VCI_KCBIT_IDLELIMIT = 6,
24037 +       VCI_KCBIT_IDLETIME = 7,
24038 +
24039 +       VCI_KCBIT_COWBL = 8,
24040 +       VCI_KCBIT_FULLCOWBL = 9,
24041 +       VCI_KCBIT_SPACES = 10,
24042 +       VCI_KCBIT_NETV2 = 11,
24043 +       VCI_KCBIT_MEMCG = 12,
24044 +
24045 +       VCI_KCBIT_DEBUG = 16,
24046 +       VCI_KCBIT_HISTORY = 20,
24047 +       VCI_KCBIT_TAGGED = 24,
24048 +       VCI_KCBIT_PPTAG = 28,
24049 +
24050 +       VCI_KCBIT_MORE = 31,
24051 +};
24052 +
24053 +
24054 +static inline uint32_t vci_kernel_config(void)
24055 +{
24056 +       return
24057 +       (1 << VCI_KCBIT_NO_DYNAMIC) |
24058 +
24059 +       /* configured features */
24060 +#ifdef CONFIG_VSERVER_PROC_SECURE
24061 +       (1 << VCI_KCBIT_PROC_SECURE) |
24062 +#endif
24063 +#ifdef CONFIG_VSERVER_IDLELIMIT
24064 +       (1 << VCI_KCBIT_IDLELIMIT) |
24065 +#endif
24066 +#ifdef CONFIG_VSERVER_IDLETIME
24067 +       (1 << VCI_KCBIT_IDLETIME) |
24068 +#endif
24069 +#ifdef CONFIG_VSERVER_COWBL
24070 +       (1 << VCI_KCBIT_COWBL) |
24071 +       (1 << VCI_KCBIT_FULLCOWBL) |
24072 +#endif
24073 +       (1 << VCI_KCBIT_SPACES) |
24074 +       (1 << VCI_KCBIT_NETV2) |
24075 +#ifdef CONFIG_CGROUP_MEM_RES_CTLR
24076 +       (1 << VCI_KCBIT_MEMCG) |
24077 +#endif
24078 +
24079 +       /* debug options */
24080 +#ifdef CONFIG_VSERVER_DEBUG
24081 +       (1 << VCI_KCBIT_DEBUG) |
24082 +#endif
24083 +#ifdef CONFIG_VSERVER_HISTORY
24084 +       (1 << VCI_KCBIT_HISTORY) |
24085 +#endif
24086 +
24087 +       /* inode context tagging */
24088 +#if    defined(CONFIG_TAGGING_NONE)
24089 +       (0 << VCI_KCBIT_TAGGED) |
24090 +#elif  defined(CONFIG_TAGGING_UID16)
24091 +       (1 << VCI_KCBIT_TAGGED) |
24092 +#elif  defined(CONFIG_TAGGING_GID16)
24093 +       (2 << VCI_KCBIT_TAGGED) |
24094 +#elif  defined(CONFIG_TAGGING_ID24)
24095 +       (3 << VCI_KCBIT_TAGGED) |
24096 +#elif  defined(CONFIG_TAGGING_INTERN)
24097 +       (4 << VCI_KCBIT_TAGGED) |
24098 +#elif  defined(CONFIG_TAGGING_RUNTIME)
24099 +       (5 << VCI_KCBIT_TAGGED) |
24100 +#else
24101 +       (7 << VCI_KCBIT_TAGGED) |
24102 +#endif
24103 +       (1 << VCI_KCBIT_PPTAG) |
24104 +       0;
24105 +}
24106 +
24107 diff -NurpP --minimal linux-2.6.36/mm/filemap_xip.c linux-2.6.36-vs2.3.0.36.38/mm/filemap_xip.c
24108 --- linux-2.6.36/mm/filemap_xip.c       2010-07-07 18:31:58.000000000 +0200
24109 +++ linux-2.6.36-vs2.3.0.36.38/mm/filemap_xip.c 2010-10-21 13:09:36.000000000 +0200
24110 @@ -18,6 +18,7 @@
24111  #include <linux/seqlock.h>
24112  #include <linux/mutex.h>
24113  #include <linux/gfp.h>
24114 +#include <linux/vs_memory.h>
24115  #include <asm/tlbflush.h>
24116  #include <asm/io.h>
24117
24118 diff -NurpP --minimal linux-2.6.36/mm/fremap.c linux-2.6.36-vs2.3.0.36.38/mm/fremap.c
24119 --- linux-2.6.36/mm/fremap.c    2010-10-21 13:07:57.000000000 +0200
24120 +++ linux-2.6.36-vs2.3.0.36.38/mm/fremap.c      2010-10-21 13:09:36.000000000 +0200
24121 @@ -16,6 +16,7 @@
24122  #include <linux/module.h>
24123  #include <linux/syscalls.h>
24124  #include <linux/mmu_notifier.h>
24125 +#include <linux/vs_memory.h>
24126
24127  #include <asm/mmu_context.h>
24128  #include <asm/cacheflush.h>
24129 diff -NurpP --minimal linux-2.6.36/mm/hugetlb.c linux-2.6.36-vs2.3.0.36.38/mm/hugetlb.c
24130 --- linux-2.6.36/mm/hugetlb.c   2010-10-21 13:07:57.000000000 +0200
24131 +++ linux-2.6.36-vs2.3.0.36.38/mm/hugetlb.c     2010-10-21 13:09:36.000000000 +0200
24132 @@ -28,6 +28,7 @@
24133
24134  #include <linux/hugetlb.h>
24135  #include <linux/node.h>
24136 +#include <linux/vs_memory.h>
24137  #include "internal.h"
24138
24139  const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
24140 diff -NurpP --minimal linux-2.6.36/mm/memcontrol.c linux-2.6.36-vs2.3.0.36.38/mm/memcontrol.c
24141 --- linux-2.6.36/mm/memcontrol.c        2010-10-21 13:07:57.000000000 +0200
24142 +++ linux-2.6.36-vs2.3.0.36.38/mm/memcontrol.c  2010-10-28 19:36:06.000000000 +0200
24143 @@ -638,6 +638,31 @@ struct mem_cgroup *mem_cgroup_from_task(
24144                                 struct mem_cgroup, css);
24145  }
24146
24147 +u64 mem_cgroup_res_read_u64(struct mem_cgroup *mem, int member)
24148 +{
24149 +       return res_counter_read_u64(&mem->res, member);
24150 +}
24151 +
24152 +u64 mem_cgroup_memsw_read_u64(struct mem_cgroup *mem, int member)
24153 +{
24154 +       return res_counter_read_u64(&mem->memsw, member);
24155 +}
24156 +
24157 +s64 mem_cgroup_stat_read_cache(struct mem_cgroup *mem)
24158 +{
24159 +       return mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_CACHE);
24160 +}
24161 +
24162 +s64 mem_cgroup_stat_read_anon(struct mem_cgroup *mem)
24163 +{
24164 +       return mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_RSS);
24165 +}
24166 +
24167 +s64 mem_cgroup_stat_read_mapped(struct mem_cgroup *mem)
24168 +{
24169 +       return mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_FILE_MAPPED);
24170 +}
24171 +
24172  static struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm)
24173  {
24174         struct mem_cgroup *mem = NULL;
24175 diff -NurpP --minimal linux-2.6.36/mm/memory.c linux-2.6.36-vs2.3.0.36.38/mm/memory.c
24176 --- linux-2.6.36/mm/memory.c    2010-10-21 13:07:57.000000000 +0200
24177 +++ linux-2.6.36-vs2.3.0.36.38/mm/memory.c      2010-10-21 13:09:36.000000000 +0200
24178 @@ -3145,6 +3145,7 @@ static inline int handle_pte_fault(struc
24179  {
24180         pte_t entry;
24181         spinlock_t *ptl;
24182 +       int ret = 0, type = VXPT_UNKNOWN;
24183
24184         entry = *pte;
24185         if (!pte_present(entry)) {
24186 @@ -3169,9 +3170,12 @@ static inline int handle_pte_fault(struc
24187         if (unlikely(!pte_same(*pte, entry)))
24188                 goto unlock;
24189         if (flags & FAULT_FLAG_WRITE) {
24190 -               if (!pte_write(entry))
24191 -                       return do_wp_page(mm, vma, address,
24192 +               if (!pte_write(entry)) {
24193 +                       ret = do_wp_page(mm, vma, address,
24194                                         pte, pmd, ptl, entry);
24195 +                       type = VXPT_WRITE;
24196 +                       goto out;
24197 +               }
24198                 entry = pte_mkdirty(entry);
24199         }
24200         entry = pte_mkyoung(entry);
24201 @@ -3189,7 +3193,10 @@ static inline int handle_pte_fault(struc
24202         }
24203  unlock:
24204         pte_unmap_unlock(pte, ptl);
24205 -       return 0;
24206 +       ret = 0;
24207 +out:
24208 +       vx_page_fault(mm, vma, type, ret);
24209 +       return ret;
24210  }
24211
24212  /*
24213 diff -NurpP --minimal linux-2.6.36/mm/mremap.c linux-2.6.36-vs2.3.0.36.38/mm/mremap.c
24214 --- linux-2.6.36/mm/mremap.c    2010-07-07 18:31:58.000000000 +0200
24215 +++ linux-2.6.36-vs2.3.0.36.38/mm/mremap.c      2010-10-21 13:09:36.000000000 +0200
24216 @@ -19,6 +19,7 @@
24217  #include <linux/security.h>
24218  #include <linux/syscalls.h>
24219  #include <linux/mmu_notifier.h>
24220 +#include <linux/vs_memory.h>
24221
24222  #include <asm/uaccess.h>
24223  #include <asm/cacheflush.h>
24224 diff -NurpP --minimal linux-2.6.36/mm/oom_kill.c linux-2.6.36-vs2.3.0.36.38/mm/oom_kill.c
24225 --- linux-2.6.36/mm/oom_kill.c  2010-10-21 13:07:57.000000000 +0200
24226 +++ linux-2.6.36-vs2.3.0.36.38/mm/oom_kill.c    2010-10-28 19:36:06.000000000 +0200
24227 @@ -31,6 +31,9 @@
24228  #include <linux/memcontrol.h>
24229  #include <linux/mempolicy.h>
24230  #include <linux/security.h>
24231 +#include <linux/reboot.h>
24232 +#include <linux/vs_memory.h>
24233 +#include <linux/vs_context.h>
24234
24235  int sysctl_panic_on_oom;
24236  int sysctl_oom_kill_allocating_task;
24237 @@ -124,11 +127,18 @@ struct task_struct *find_lock_task_mm(st
24238  static bool oom_unkillable_task(struct task_struct *p,
24239                 const struct mem_cgroup *mem, const nodemask_t *nodemask)
24240  {
24241 -       if (is_global_init(p))
24242 +       unsigned xid = vx_current_xid();
24243 +
24244 +       /* skip the init task, global and per guest */
24245 +       if (task_is_init(p))
24246                 return true;
24247         if (p->flags & PF_KTHREAD)
24248                 return true;
24249
24250 +       /* skip other guest and host processes if oom in guest */
24251 +       if (xid && vx_task_xid(p) != xid)
24252 +               return true;
24253 +
24254         /* When mem_cgroup_out_of_memory() and p is not member of the group */
24255         if (mem && !task_in_mem_cgroup(p, mem))
24256                 return true;
24257 @@ -209,6 +219,18 @@ unsigned int oom_badness(struct task_str
24258         points += p->signal->oom_score_adj;
24259
24260         /*
24261 +        * add points for context badness and
24262 +        * reduce badness for processes belonging to
24263 +        * a different context
24264 +        */
24265 +
24266 +       points += vx_badness(p, p->mm);
24267 +
24268 +       if ((vx_current_xid() > 1) &&
24269 +               vx_current_xid() != vx_task_xid(p))
24270 +               points /= 16;
24271 +
24272 +       /*
24273          * Never return 0 for an eligible task that may be killed since it's
24274          * possible that no single user task uses more than 0.1% of memory and
24275          * no single admin tasks uses more than 3.0%.
24276 @@ -407,8 +429,8 @@ static int oom_kill_task(struct task_str
24277         if (!p)
24278                 return 1;
24279
24280 -       pr_err("Killed process %d (%s) total-vm:%lukB, anon-rss:%lukB, file-rss:%lukB\n",
24281 -               task_pid_nr(p), p->comm, K(p->mm->total_vm),
24282 +       pr_err("Killed process %d:#%u (%s) total-vm:%lukB, anon-rss:%lukB, file-rss:%lukB\n",
24283 +               task_pid_nr(p), p->xid, p->comm, K(p->mm->total_vm),
24284                 K(get_mm_counter(p->mm, MM_ANONPAGES)),
24285                 K(get_mm_counter(p->mm, MM_FILEPAGES)));
24286         task_unlock(p);
24287 @@ -452,8 +474,8 @@ static int oom_kill_process(struct task_
24288         }
24289
24290         task_lock(p);
24291 -       pr_err("%s: Kill process %d (%s) score %d or sacrifice child\n",
24292 -               message, task_pid_nr(p), p->comm, points);
24293 +       pr_err("%s: Kill process %d:#%u (%s) score %d or sacrifice child\n",
24294 +               message, task_pid_nr(p), p->xid, p->comm, points);
24295         task_unlock(p);
24296
24297         /*
24298 @@ -542,6 +564,8 @@ int unregister_oom_notifier(struct notif
24299  }
24300  EXPORT_SYMBOL_GPL(unregister_oom_notifier);
24301
24302 +long vs_oom_action(unsigned int);
24303 +
24304  /*
24305   * Try to acquire the OOM killer lock for the zones in zonelist.  Returns zero
24306   * if a parallel OOM killing is already taking place that includes a zone in
24307 @@ -701,7 +725,12 @@ retry:
24308         if (!p) {
24309                 dump_header(NULL, gfp_mask, order, NULL, mpol_mask);
24310                 read_unlock(&tasklist_lock);
24311 -               panic("Out of memory and no killable processes...\n");
24312 +
24313 +               /* avoid panic for guest OOM */
24314 +               if (current->xid)
24315 +                       vs_oom_action(LINUX_REBOOT_CMD_OOM);
24316 +               else
24317 +                       panic("Out of memory and no killable processes...\n");
24318         }
24319
24320         if (oom_kill_process(p, gfp_mask, order, points, totalpages, NULL,
24321 diff -NurpP --minimal linux-2.6.36/mm/page_alloc.c linux-2.6.36-vs2.3.0.36.38/mm/page_alloc.c
24322 --- linux-2.6.36/mm/page_alloc.c        2010-10-21 13:07:57.000000000 +0200
24323 +++ linux-2.6.36-vs2.3.0.36.38/mm/page_alloc.c  2010-10-21 13:09:36.000000000 +0200
24324 @@ -52,6 +52,8 @@
24325  #include <linux/compaction.h>
24326  #include <trace/events/kmem.h>
24327  #include <linux/ftrace_event.h>
24328 +#include <linux/vs_base.h>
24329 +#include <linux/vs_limit.h>
24330
24331  #include <asm/tlbflush.h>
24332  #include <asm/div64.h>
24333 @@ -2329,6 +2331,9 @@ void si_meminfo(struct sysinfo *val)
24334         val->totalhigh = totalhigh_pages;
24335         val->freehigh = nr_free_highpages();
24336         val->mem_unit = PAGE_SIZE;
24337 +
24338 +       if (vx_flags(VXF_VIRT_MEM, 0))
24339 +               vx_vsi_meminfo(val);
24340  }
24341
24342  EXPORT_SYMBOL(si_meminfo);
24343 @@ -2349,6 +2354,9 @@ void si_meminfo_node(struct sysinfo *val
24344         val->freehigh = 0;
24345  #endif
24346         val->mem_unit = PAGE_SIZE;
24347 +
24348 +       if (vx_flags(VXF_VIRT_MEM, 0))
24349 +               vx_vsi_meminfo(val);
24350  }
24351  #endif
24352
24353 diff -NurpP --minimal linux-2.6.36/mm/rmap.c linux-2.6.36-vs2.3.0.36.38/mm/rmap.c
24354 --- linux-2.6.36/mm/rmap.c      2010-10-21 13:07:57.000000000 +0200
24355 +++ linux-2.6.36-vs2.3.0.36.38/mm/rmap.c        2010-10-21 13:19:11.000000000 +0200
24356 @@ -57,6 +57,7 @@
24357  #include <linux/mmu_notifier.h>
24358  #include <linux/migrate.h>
24359  #include <linux/hugetlb.h>
24360 +#include <linux/vs_memory.h>
24361
24362  #include <asm/tlbflush.h>
24363
24364 diff -NurpP --minimal linux-2.6.36/mm/shmem.c linux-2.6.36-vs2.3.0.36.38/mm/shmem.c
24365 --- linux-2.6.36/mm/shmem.c     2010-10-21 13:07:57.000000000 +0200
24366 +++ linux-2.6.36-vs2.3.0.36.38/mm/shmem.c       2010-10-21 13:09:36.000000000 +0200
24367 @@ -1815,7 +1815,7 @@ static int shmem_statfs(struct dentry *d
24368  {
24369         struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
24370
24371 -       buf->f_type = TMPFS_MAGIC;
24372 +       buf->f_type = TMPFS_SUPER_MAGIC;
24373         buf->f_bsize = PAGE_CACHE_SIZE;
24374         buf->f_namelen = NAME_MAX;
24375         if (sbinfo->max_blocks) {
24376 @@ -2377,7 +2377,7 @@ int shmem_fill_super(struct super_block
24377         sb->s_maxbytes = SHMEM_MAX_BYTES;
24378         sb->s_blocksize = PAGE_CACHE_SIZE;
24379         sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
24380 -       sb->s_magic = TMPFS_MAGIC;
24381 +       sb->s_magic = TMPFS_SUPER_MAGIC;
24382         sb->s_op = &shmem_ops;
24383         sb->s_time_gran = 1;
24384  #ifdef CONFIG_TMPFS_POSIX_ACL
24385 diff -NurpP --minimal linux-2.6.36/mm/slab.c linux-2.6.36-vs2.3.0.36.38/mm/slab.c
24386 --- linux-2.6.36/mm/slab.c      2010-10-21 13:07:57.000000000 +0200
24387 +++ linux-2.6.36-vs2.3.0.36.38/mm/slab.c        2010-10-21 13:09:36.000000000 +0200
24388 @@ -407,6 +407,8 @@ static void kmem_list3_init(struct kmem_
24389  #define STATS_INC_FREEMISS(x)  do { } while (0)
24390  #endif
24391
24392 +#include "slab_vs.h"
24393 +
24394  #if DEBUG
24395
24396  /*
24397 @@ -3346,6 +3348,7 @@ retry:
24398
24399         obj = slab_get_obj(cachep, slabp, nodeid);
24400         check_slabp(cachep, slabp);
24401 +       vx_slab_alloc(cachep, flags);
24402         l3->free_objects--;
24403         /* move slabp to correct slabp list: */
24404         list_del(&slabp->list);
24405 @@ -3423,6 +3426,7 @@ __cache_alloc_node(struct kmem_cache *ca
24406         /* ___cache_alloc_node can fall back to other nodes */
24407         ptr = ____cache_alloc_node(cachep, flags, nodeid);
24408    out:
24409 +       vx_slab_alloc(cachep, flags);
24410         local_irq_restore(save_flags);
24411         ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller);
24412         kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags,
24413 @@ -3609,6 +3613,7 @@ static inline void __cache_free(struct k
24414         check_irq_off();
24415         kmemleak_free_recursive(objp, cachep->flags);
24416         objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0));
24417 +       vx_slab_free(cachep);
24418
24419         kmemcheck_slab_free(cachep, objp, obj_size(cachep));
24420
24421 diff -NurpP --minimal linux-2.6.36/mm/slab_vs.h linux-2.6.36-vs2.3.0.36.38/mm/slab_vs.h
24422 --- linux-2.6.36/mm/slab_vs.h   1970-01-01 01:00:00.000000000 +0100
24423 +++ linux-2.6.36-vs2.3.0.36.38/mm/slab_vs.h     2010-10-21 13:09:36.000000000 +0200
24424 @@ -0,0 +1,29 @@
24425 +
24426 +#include <linux/vserver/context.h>
24427 +
24428 +#include <linux/vs_context.h>
24429 +
24430 +static inline
24431 +void vx_slab_alloc(struct kmem_cache *cachep, gfp_t flags)
24432 +{
24433 +       int what = gfp_zone(cachep->gfpflags);
24434 +       struct vx_info *vxi = current_vx_info();
24435 +
24436 +       if (!vxi)
24437 +               return;
24438 +
24439 +       atomic_add(cachep->buffer_size, &vxi->cacct.slab[what]);
24440 +}
24441 +
24442 +static inline
24443 +void vx_slab_free(struct kmem_cache *cachep)
24444 +{
24445 +       int what = gfp_zone(cachep->gfpflags);
24446 +       struct vx_info *vxi = current_vx_info();
24447 +
24448 +       if (!vxi)
24449 +               return;
24450 +
24451 +       atomic_sub(cachep->buffer_size, &vxi->cacct.slab[what]);
24452 +}
24453 +
24454 diff -NurpP --minimal linux-2.6.36/mm/swapfile.c linux-2.6.36-vs2.3.0.36.38/mm/swapfile.c
24455 --- linux-2.6.36/mm/swapfile.c  2010-10-21 13:07:57.000000000 +0200
24456 +++ linux-2.6.36-vs2.3.0.36.38/mm/swapfile.c    2010-10-21 13:09:36.000000000 +0200
24457 @@ -35,6 +35,8 @@
24458  #include <asm/tlbflush.h>
24459  #include <linux/swapops.h>
24460  #include <linux/page_cgroup.h>
24461 +#include <linux/vs_base.h>
24462 +#include <linux/vs_memory.h>
24463
24464  static bool swap_count_continued(struct swap_info_struct *, pgoff_t,
24465                                  unsigned char);
24466 @@ -1747,6 +1749,16 @@ static int swap_show(struct seq_file *sw
24467
24468         if (si == SEQ_START_TOKEN) {
24469                 seq_puts(swap,"Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
24470 +               if (vx_flags(VXF_VIRT_MEM, 0)) {
24471 +                       struct sysinfo si;
24472 +
24473 +                       vx_vsi_swapinfo(&si);
24474 +                       if (si.totalswap < (1 << 10))
24475 +                               return 0;
24476 +                       seq_printf(swap, "%s\t\t\t\t\t%s\t%lu\t%lu\t%d\n",
24477 +                               "hdv0", "partition", si.totalswap >> 10,
24478 +                               (si.totalswap - si.freeswap) >> 10, -1);
24479 +               }
24480                 return 0;
24481         }
24482
24483 @@ -2131,6 +2143,8 @@ void si_swapinfo(struct sysinfo *val)
24484         val->freeswap = nr_swap_pages + nr_to_be_unused;
24485         val->totalswap = total_swap_pages + nr_to_be_unused;
24486         spin_unlock(&swap_lock);
24487 +       if (vx_flags(VXF_VIRT_MEM, 0))
24488 +               vx_vsi_swapinfo(val);
24489  }
24490
24491  /*
24492 diff -NurpP --minimal linux-2.6.36/net/core/dev.c linux-2.6.36-vs2.3.0.36.38/net/core/dev.c
24493 --- linux-2.6.36/net/core/dev.c 2010-10-21 13:07:58.000000000 +0200
24494 +++ linux-2.6.36-vs2.3.0.36.38/net/core/dev.c   2010-10-21 14:12:56.000000000 +0200
24495 @@ -127,6 +127,7 @@
24496  #include <linux/in.h>
24497  #include <linux/jhash.h>
24498  #include <linux/random.h>
24499 +#include <linux/vs_inet.h>
24500  #include <trace/events/napi.h>
24501  #include <linux/pci.h>
24502
24503 @@ -607,7 +608,8 @@ struct net_device *__dev_get_by_name(str
24504         struct hlist_head *head = dev_name_hash(net, name);
24505
24506         hlist_for_each_entry(dev, p, head, name_hlist)
24507 -               if (!strncmp(dev->name, name, IFNAMSIZ))
24508 +               if (!strncmp(dev->name, name, IFNAMSIZ) &&
24509 +                   nx_dev_visible(current_nx_info(), dev))
24510                         return dev;
24511
24512         return NULL;
24513 @@ -633,7 +635,8 @@ struct net_device *dev_get_by_name_rcu(s
24514         struct hlist_head *head = dev_name_hash(net, name);
24515
24516         hlist_for_each_entry_rcu(dev, p, head, name_hlist)
24517 -               if (!strncmp(dev->name, name, IFNAMSIZ))
24518 +               if (!strncmp(dev->name, name, IFNAMSIZ) &&
24519 +                   nx_dev_visible(current_nx_info(), dev))
24520                         return dev;
24521
24522         return NULL;
24523 @@ -684,7 +687,8 @@ struct net_device *__dev_get_by_index(st
24524         struct hlist_head *head = dev_index_hash(net, ifindex);
24525
24526         hlist_for_each_entry(dev, p, head, index_hlist)
24527 -               if (dev->ifindex == ifindex)
24528 +               if ((dev->ifindex == ifindex) &&
24529 +                   nx_dev_visible(current_nx_info(), dev))
24530                         return dev;
24531
24532         return NULL;
24533 @@ -709,7 +713,8 @@ struct net_device *dev_get_by_index_rcu(
24534         struct hlist_head *head = dev_index_hash(net, ifindex);
24535
24536         hlist_for_each_entry_rcu(dev, p, head, index_hlist)
24537 -               if (dev->ifindex == ifindex)
24538 +               if ((dev->ifindex == ifindex) &&
24539 +                   nx_dev_visible(current_nx_info(), dev))
24540                         return dev;
24541
24542         return NULL;
24543 @@ -762,10 +767,12 @@ struct net_device *dev_getbyhwaddr(struc
24544
24545         ASSERT_RTNL();
24546
24547 -       for_each_netdev(net, dev)
24548 +       for_each_netdev(net, dev) {
24549                 if (dev->type == type &&
24550 -                   !memcmp(dev->dev_addr, ha, dev->addr_len))
24551 +                   !memcmp(dev->dev_addr, ha, dev->addr_len) &&
24552 +                   nx_dev_visible(current_nx_info(), dev))
24553                         return dev;
24554 +       }
24555
24556         return NULL;
24557  }
24558 @@ -776,9 +783,11 @@ struct net_device *__dev_getfirstbyhwtyp
24559         struct net_device *dev;
24560
24561         ASSERT_RTNL();
24562 -       for_each_netdev(net, dev)
24563 -               if (dev->type == type)
24564 +       for_each_netdev(net, dev) {
24565 +               if ((dev->type == type) &&
24566 +                   nx_dev_visible(current_nx_info(), dev))
24567                         return dev;
24568 +       }
24569
24570         return NULL;
24571  }
24572 @@ -896,6 +905,8 @@ static int __dev_alloc_name(struct net *
24573                                 continue;
24574                         if (i < 0 || i >= max_netdevices)
24575                                 continue;
24576 +                       if (!nx_dev_visible(current_nx_info(), d))
24577 +                               continue;
24578
24579                         /*  avoid cases where sscanf is not exact inverse of printf */
24580                         snprintf(buf, IFNAMSIZ, name, i);
24581 @@ -3658,6 +3669,8 @@ static int dev_ifconf(struct net *net, c
24582
24583         total = 0;
24584         for_each_netdev(net, dev) {
24585 +               if (!nx_dev_visible(current_nx_info(), dev))
24586 +                       continue;
24587                 for (i = 0; i < NPROTO; i++) {
24588                         if (gifconf_list[i]) {
24589                                 int done;
24590 @@ -3729,6 +3742,10 @@ static void dev_seq_printf_stats(struct
24591         struct rtnl_link_stats64 temp;
24592         const struct rtnl_link_stats64 *stats = dev_get_stats(dev, &temp);
24593
24594 +       /* device visible inside network context? */
24595 +       if (!nx_dev_visible(current_nx_info(), dev))
24596 +               return;
24597 +
24598         seq_printf(seq, "%6s: %7llu %7llu %4llu %4llu %4llu %5llu %10llu %9llu "
24599                    "%8llu %7llu %4llu %4llu %4llu %5llu %7llu %10llu\n",
24600                    dev->name, stats->rx_bytes, stats->rx_packets,
24601 diff -NurpP --minimal linux-2.6.36/net/core/rtnetlink.c linux-2.6.36-vs2.3.0.36.38/net/core/rtnetlink.c
24602 --- linux-2.6.36/net/core/rtnetlink.c   2010-10-21 13:07:58.000000000 +0200
24603 +++ linux-2.6.36-vs2.3.0.36.38/net/core/rtnetlink.c     2010-10-21 13:09:36.000000000 +0200
24604 @@ -927,6 +927,8 @@ static int rtnl_dump_ifinfo(struct sk_bu
24605                 hlist_for_each_entry(dev, node, head, index_hlist) {
24606                         if (idx < s_idx)
24607                                 goto cont;
24608 +                       if (!nx_dev_visible(skb->sk->sk_nx_info, dev))
24609 +                               continue;
24610                         if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK,
24611                                              NETLINK_CB(cb->skb).pid,
24612                                              cb->nlh->nlmsg_seq, 0,
24613 @@ -1643,6 +1645,9 @@ void rtmsg_ifinfo(int type, struct net_d
24614         struct sk_buff *skb;
24615         int err = -ENOBUFS;
24616
24617 +       if (!nx_dev_visible(current_nx_info(), dev))
24618 +               return;
24619 +
24620         skb = nlmsg_new(if_nlmsg_size(dev), GFP_KERNEL);
24621         if (skb == NULL)
24622                 goto errout;
24623 diff -NurpP --minimal linux-2.6.36/net/core/sock.c linux-2.6.36-vs2.3.0.36.38/net/core/sock.c
24624 --- linux-2.6.36/net/core/sock.c        2010-10-21 13:07:58.000000000 +0200
24625 +++ linux-2.6.36-vs2.3.0.36.38/net/core/sock.c  2010-10-21 13:09:36.000000000 +0200
24626 @@ -127,6 +127,10 @@
24627  #include <net/cls_cgroup.h>
24628
24629  #include <linux/filter.h>
24630 +#include <linux/vs_socket.h>
24631 +#include <linux/vs_limit.h>
24632 +#include <linux/vs_context.h>
24633 +#include <linux/vs_network.h>
24634
24635  #ifdef CONFIG_INET
24636  #include <net/tcp.h>
24637 @@ -1046,6 +1050,8 @@ static struct sock *sk_prot_alloc(struct
24638                         goto out_free_sec;
24639                 sk_tx_queue_clear(sk);
24640         }
24641 +               sock_vx_init(sk);
24642 +               sock_nx_init(sk);
24643
24644         return sk;
24645
24646 @@ -1142,6 +1148,11 @@ static void __sk_free(struct sock *sk)
24647                 put_cred(sk->sk_peer_cred);
24648         put_pid(sk->sk_peer_pid);
24649         put_net(sock_net(sk));
24650 +       vx_sock_dec(sk);
24651 +       clr_vx_info(&sk->sk_vx_info);
24652 +       sk->sk_xid = -1;
24653 +       clr_nx_info(&sk->sk_nx_info);
24654 +       sk->sk_nid = -1;
24655         sk_prot_free(sk->sk_prot_creator, sk);
24656  }
24657
24658 @@ -1189,6 +1200,8 @@ struct sock *sk_clone(const struct sock
24659
24660                 /* SANITY */
24661                 get_net(sock_net(newsk));
24662 +               sock_vx_init(newsk);
24663 +               sock_nx_init(newsk);
24664                 sk_node_init(&newsk->sk_node);
24665                 sock_lock_init(newsk);
24666                 bh_lock_sock(newsk);
24667 @@ -1244,6 +1257,12 @@ struct sock *sk_clone(const struct sock
24668                 smp_wmb();
24669                 atomic_set(&newsk->sk_refcnt, 2);
24670
24671 +               set_vx_info(&newsk->sk_vx_info, sk->sk_vx_info);
24672 +               newsk->sk_xid = sk->sk_xid;
24673 +               vx_sock_inc(newsk);
24674 +               set_nx_info(&newsk->sk_nx_info, sk->sk_nx_info);
24675 +               newsk->sk_nid = sk->sk_nid;
24676 +
24677                 /*
24678                  * Increment the counter in the same struct proto as the master
24679                  * sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
24680 @@ -1986,6 +2005,12 @@ void sock_init_data(struct socket *sock,
24681
24682         sk->sk_stamp = ktime_set(-1L, 0);
24683
24684 +       set_vx_info(&sk->sk_vx_info, current_vx_info());
24685 +       sk->sk_xid = vx_current_xid();
24686 +       vx_sock_inc(sk);
24687 +       set_nx_info(&sk->sk_nx_info, current_nx_info());
24688 +       sk->sk_nid = nx_current_nid();
24689 +
24690         /*
24691          * Before updating sk_refcnt, we must commit prior changes to memory
24692          * (Documentation/RCU/rculist_nulls.txt for details)
24693 diff -NurpP --minimal linux-2.6.36/net/ipv4/af_inet.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/af_inet.c
24694 --- linux-2.6.36/net/ipv4/af_inet.c     2010-10-21 13:07:58.000000000 +0200
24695 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/af_inet.c       2010-10-21 13:09:36.000000000 +0200
24696 @@ -116,6 +116,7 @@
24697  #ifdef CONFIG_IP_MROUTE
24698  #include <linux/mroute.h>
24699  #endif
24700 +#include <linux/vs_limit.h>
24701
24702
24703  /* The inetsw table contains everything that inet_create needs to
24704 @@ -327,9 +328,13 @@ lookup_protocol:
24705         }
24706
24707         err = -EPERM;
24708 +       if ((protocol == IPPROTO_ICMP) &&
24709 +               nx_capable(CAP_NET_RAW, NXC_RAW_ICMP))
24710 +               goto override;
24711 +
24712         if (sock->type == SOCK_RAW && !kern && !capable(CAP_NET_RAW))
24713                 goto out_rcu_unlock;
24714 -
24715 +override:
24716         err = -EAFNOSUPPORT;
24717         if (!inet_netns_ok(net, protocol))
24718                 goto out_rcu_unlock;
24719 @@ -453,6 +458,7 @@ int inet_bind(struct socket *sock, struc
24720         struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
24721         struct sock *sk = sock->sk;
24722         struct inet_sock *inet = inet_sk(sk);
24723 +       struct nx_v4_sock_addr nsa;
24724         unsigned short snum;
24725         int chk_addr_ret;
24726         int err;
24727 @@ -466,7 +472,11 @@ int inet_bind(struct socket *sock, struc
24728         if (addr_len < sizeof(struct sockaddr_in))
24729                 goto out;
24730
24731 -       chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
24732 +       err = v4_map_sock_addr(inet, addr, &nsa);
24733 +       if (err)
24734 +               goto out;
24735 +
24736 +       chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
24737
24738         /* Not specified by any standard per-se, however it breaks too
24739          * many applications when removed.  It is unfortunate since
24740 @@ -478,7 +488,7 @@ int inet_bind(struct socket *sock, struc
24741         err = -EADDRNOTAVAIL;
24742         if (!sysctl_ip_nonlocal_bind &&
24743             !(inet->freebind || inet->transparent) &&
24744 -           addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
24745 +           nsa.saddr != htonl(INADDR_ANY) &&
24746             chk_addr_ret != RTN_LOCAL &&
24747             chk_addr_ret != RTN_MULTICAST &&
24748             chk_addr_ret != RTN_BROADCAST)
24749 @@ -503,7 +513,7 @@ int inet_bind(struct socket *sock, struc
24750         if (sk->sk_state != TCP_CLOSE || inet->inet_num)
24751                 goto out_release_sock;
24752
24753 -       inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
24754 +       v4_set_sock_addr(inet, &nsa);
24755         if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
24756                 inet->inet_saddr = 0;  /* Use device */
24757
24758 @@ -705,11 +715,13 @@ int inet_getname(struct socket *sock, st
24759                      peer == 1))
24760                         return -ENOTCONN;
24761                 sin->sin_port = inet->inet_dport;
24762 -               sin->sin_addr.s_addr = inet->inet_daddr;
24763 +               sin->sin_addr.s_addr =
24764 +                       nx_map_sock_lback(sk->sk_nx_info, inet->inet_daddr);
24765         } else {
24766                 __be32 addr = inet->inet_rcv_saddr;
24767                 if (!addr)
24768                         addr = inet->inet_saddr;
24769 +               addr = nx_map_sock_lback(sk->sk_nx_info, addr);
24770                 sin->sin_port = inet->inet_sport;
24771                 sin->sin_addr.s_addr = addr;
24772         }
24773 diff -NurpP --minimal linux-2.6.36/net/ipv4/devinet.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/devinet.c
24774 --- linux-2.6.36/net/ipv4/devinet.c     2010-10-21 13:07:58.000000000 +0200
24775 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/devinet.c       2010-10-21 13:09:36.000000000 +0200
24776 @@ -417,6 +417,7 @@ struct in_device *inetdev_by_index(struc
24777  }
24778  EXPORT_SYMBOL(inetdev_by_index);
24779
24780 +
24781  /* Called only from RTNL semaphored context. No locks. */
24782
24783  struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix,
24784 @@ -659,6 +660,8 @@ int devinet_ioctl(struct net *net, unsig
24785
24786         in_dev = __in_dev_get_rtnl(dev);
24787         if (in_dev) {
24788 +               struct nx_info *nxi = current_nx_info();
24789 +
24790                 if (tryaddrmatch) {
24791                         /* Matthias Andree */
24792                         /* compare label and address (4.4BSD style) */
24793 @@ -667,6 +670,8 @@ int devinet_ioctl(struct net *net, unsig
24794                            This is checked above. */
24795                         for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
24796                              ifap = &ifa->ifa_next) {
24797 +                               if (!nx_v4_ifa_visible(nxi, ifa))
24798 +                                       continue;
24799                                 if (!strcmp(ifr.ifr_name, ifa->ifa_label) &&
24800                                     sin_orig.sin_addr.s_addr ==
24801                                                         ifa->ifa_address) {
24802 @@ -679,9 +684,12 @@ int devinet_ioctl(struct net *net, unsig
24803                    comparing just the label */
24804                 if (!ifa) {
24805                         for (ifap = &in_dev->ifa_list; (ifa = *ifap) != NULL;
24806 -                            ifap = &ifa->ifa_next)
24807 +                            ifap = &ifa->ifa_next) {
24808 +                               if (!nx_v4_ifa_visible(nxi, ifa))
24809 +                                       continue;
24810                                 if (!strcmp(ifr.ifr_name, ifa->ifa_label))
24811                                         break;
24812 +                       }
24813                 }
24814         }
24815
24816 @@ -833,6 +841,8 @@ static int inet_gifconf(struct net_devic
24817                 goto out;
24818
24819         for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
24820 +               if (!nx_v4_ifa_visible(current_nx_info(), ifa))
24821 +                       continue;
24822                 if (!buf) {
24823                         done += sizeof(ifr);
24824                         continue;
24825 @@ -1182,6 +1192,7 @@ static int inet_dump_ifaddr(struct sk_bu
24826         struct net_device *dev;
24827         struct in_device *in_dev;
24828         struct in_ifaddr *ifa;
24829 +       struct sock *sk = skb->sk;
24830         struct hlist_head *head;
24831         struct hlist_node *node;
24832
24833 @@ -1204,6 +1215,8 @@ static int inet_dump_ifaddr(struct sk_bu
24834
24835                         for (ifa = in_dev->ifa_list, ip_idx = 0; ifa;
24836                              ifa = ifa->ifa_next, ip_idx++) {
24837 +                       if (sk && !nx_v4_ifa_visible(sk->sk_nx_info, ifa))
24838 +                               continue;
24839                                 if (ip_idx < s_ip_idx)
24840                                         continue;
24841                                 if (inet_fill_ifaddr(skb, ifa,
24842 diff -NurpP --minimal linux-2.6.36/net/ipv4/fib_hash.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/fib_hash.c
24843 --- linux-2.6.36/net/ipv4/fib_hash.c    2010-07-07 18:31:59.000000000 +0200
24844 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/fib_hash.c      2010-10-21 13:09:36.000000000 +0200
24845 @@ -1017,7 +1017,7 @@ static int fib_seq_show(struct seq_file
24846         prefix  = f->fn_key;
24847         mask    = FZ_MASK(iter->zone);
24848         flags   = fib_flag_trans(fa->fa_type, mask, fi);
24849 -       if (fi)
24850 +       if (fi && nx_dev_visible(current_nx_info(), fi->fib_dev))
24851                 seq_printf(seq,
24852                          "%s\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u%n",
24853                          fi->fib_dev ? fi->fib_dev->name : "*", prefix,
24854 diff -NurpP --minimal linux-2.6.36/net/ipv4/inet_connection_sock.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_connection_sock.c
24855 --- linux-2.6.36/net/ipv4/inet_connection_sock.c        2010-10-21 13:07:58.000000000 +0200
24856 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_connection_sock.c  2010-10-21 13:09:36.000000000 +0200
24857 @@ -52,10 +52,40 @@ void inet_get_local_port_range(int *low,
24858  }
24859  EXPORT_SYMBOL(inet_get_local_port_range);
24860
24861 +int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
24862 +{
24863 +       __be32  sk1_rcv_saddr = inet_rcv_saddr(sk1),
24864 +               sk2_rcv_saddr = inet_rcv_saddr(sk2);
24865 +
24866 +       if (inet_v6_ipv6only(sk2))
24867 +               return 0;
24868 +
24869 +       if (sk1_rcv_saddr &&
24870 +           sk2_rcv_saddr &&
24871 +           sk1_rcv_saddr == sk2_rcv_saddr)
24872 +               return 1;
24873 +
24874 +       if (sk1_rcv_saddr &&
24875 +           !sk2_rcv_saddr &&
24876 +           v4_addr_in_nx_info(sk2->sk_nx_info, sk1_rcv_saddr, NXA_MASK_BIND))
24877 +               return 1;
24878 +
24879 +       if (sk2_rcv_saddr &&
24880 +           !sk1_rcv_saddr &&
24881 +           v4_addr_in_nx_info(sk1->sk_nx_info, sk2_rcv_saddr, NXA_MASK_BIND))
24882 +               return 1;
24883 +
24884 +       if (!sk1_rcv_saddr &&
24885 +           !sk2_rcv_saddr &&
24886 +           nx_v4_addr_conflict(sk1->sk_nx_info, sk2->sk_nx_info))
24887 +               return 1;
24888 +
24889 +       return 0;
24890 +}
24891 +
24892  int inet_csk_bind_conflict(const struct sock *sk,
24893                            const struct inet_bind_bucket *tb)
24894  {
24895 -       const __be32 sk_rcv_saddr = inet_rcv_saddr(sk);
24896         struct sock *sk2;
24897         struct hlist_node *node;
24898         int reuse = sk->sk_reuse;
24899 @@ -75,9 +105,7 @@ int inet_csk_bind_conflict(const struct
24900                      sk->sk_bound_dev_if == sk2->sk_bound_dev_if)) {
24901                         if (!reuse || !sk2->sk_reuse ||
24902                             sk2->sk_state == TCP_LISTEN) {
24903 -                               const __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
24904 -                               if (!sk2_rcv_saddr || !sk_rcv_saddr ||
24905 -                                   sk2_rcv_saddr == sk_rcv_saddr)
24906 +                               if (ipv4_rcv_saddr_equal(sk, sk2))
24907                                         break;
24908                         }
24909                 }
24910 diff -NurpP --minimal linux-2.6.36/net/ipv4/inet_diag.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_diag.c
24911 --- linux-2.6.36/net/ipv4/inet_diag.c   2010-07-07 18:31:59.000000000 +0200
24912 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_diag.c     2010-10-21 13:09:36.000000000 +0200
24913 @@ -33,6 +33,8 @@
24914  #include <linux/stddef.h>
24915
24916  #include <linux/inet_diag.h>
24917 +#include <linux/vs_network.h>
24918 +#include <linux/vs_inet.h>
24919
24920  static const struct inet_diag_handler **inet_diag_table;
24921
24922 @@ -119,8 +121,10 @@ static int inet_csk_diag_fill(struct soc
24923
24924         r->id.idiag_sport = inet->inet_sport;
24925         r->id.idiag_dport = inet->inet_dport;
24926 -       r->id.idiag_src[0] = inet->inet_rcv_saddr;
24927 -       r->id.idiag_dst[0] = inet->inet_daddr;
24928 +       r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info,
24929 +               inet->inet_rcv_saddr);
24930 +       r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info,
24931 +               inet->inet_daddr);
24932
24933  #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
24934         if (r->idiag_family == AF_INET6) {
24935 @@ -205,8 +209,8 @@ static int inet_twsk_diag_fill(struct in
24936         r->id.idiag_cookie[1] = (u32)(((unsigned long)tw >> 31) >> 1);
24937         r->id.idiag_sport     = tw->tw_sport;
24938         r->id.idiag_dport     = tw->tw_dport;
24939 -       r->id.idiag_src[0]    = tw->tw_rcv_saddr;
24940 -       r->id.idiag_dst[0]    = tw->tw_daddr;
24941 +       r->id.idiag_src[0]    = nx_map_sock_lback(tw->tw_nx_info, tw->tw_rcv_saddr);
24942 +       r->id.idiag_dst[0]    = nx_map_sock_lback(tw->tw_nx_info, tw->tw_daddr);
24943         r->idiag_state        = tw->tw_substate;
24944         r->idiag_timer        = 3;
24945         r->idiag_expires      = DIV_ROUND_UP(tmo * 1000, HZ);
24946 @@ -263,6 +267,7 @@ static int inet_diag_get_exact(struct sk
24947         err = -EINVAL;
24948
24949         if (req->idiag_family == AF_INET) {
24950 +               /* TODO: lback */
24951                 sk = inet_lookup(&init_net, hashinfo, req->id.idiag_dst[0],
24952                                  req->id.idiag_dport, req->id.idiag_src[0],
24953                                  req->id.idiag_sport, req->id.idiag_if);
24954 @@ -505,6 +510,7 @@ static int inet_csk_diag_dump(struct soc
24955                 } else
24956  #endif
24957                 {
24958 +                       /* TODO: lback */
24959                         entry.saddr = &inet->inet_rcv_saddr;
24960                         entry.daddr = &inet->inet_daddr;
24961                 }
24962 @@ -541,6 +547,7 @@ static int inet_twsk_diag_dump(struct in
24963                 } else
24964  #endif
24965                 {
24966 +                       /* TODO: lback */
24967                         entry.saddr = &tw->tw_rcv_saddr;
24968                         entry.daddr = &tw->tw_daddr;
24969                 }
24970 @@ -587,8 +594,8 @@ static int inet_diag_fill_req(struct sk_
24971
24972         r->id.idiag_sport = inet->inet_sport;
24973         r->id.idiag_dport = ireq->rmt_port;
24974 -       r->id.idiag_src[0] = ireq->loc_addr;
24975 -       r->id.idiag_dst[0] = ireq->rmt_addr;
24976 +       r->id.idiag_src[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->loc_addr);
24977 +       r->id.idiag_dst[0] = nx_map_sock_lback(sk->sk_nx_info, ireq->rmt_addr);
24978         r->idiag_expires = jiffies_to_msecs(tmo);
24979         r->idiag_rqueue = 0;
24980         r->idiag_wqueue = 0;
24981 @@ -658,6 +665,7 @@ static int inet_diag_dump_reqs(struct sk
24982                                 continue;
24983
24984                         if (bc) {
24985 +                               /* TODO: lback */
24986                                 entry.saddr =
24987  #if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
24988                                         (entry.family == AF_INET6) ?
24989 @@ -728,6 +736,8 @@ static int inet_diag_dump(struct sk_buff
24990                         sk_nulls_for_each(sk, node, &ilb->head) {
24991                                 struct inet_sock *inet = inet_sk(sk);
24992
24993 +                               if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
24994 +                                       continue;
24995                                 if (num < s_num) {
24996                                         num++;
24997                                         continue;
24998 @@ -794,6 +804,8 @@ skip_listen_ht:
24999                 sk_nulls_for_each(sk, node, &head->chain) {
25000                         struct inet_sock *inet = inet_sk(sk);
25001
25002 +                       if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25003 +                               continue;
25004                         if (num < s_num)
25005                                 goto next_normal;
25006                         if (!(r->idiag_states & (1 << sk->sk_state)))
25007 @@ -818,6 +830,8 @@ next_normal:
25008                         inet_twsk_for_each(tw, node,
25009                                     &head->twchain) {
25010
25011 +                               if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
25012 +                                       continue;
25013                                 if (num < s_num)
25014                                         goto next_dying;
25015                                 if (r->id.idiag_sport != tw->tw_sport &&
25016 diff -NurpP --minimal linux-2.6.36/net/ipv4/inet_hashtables.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_hashtables.c
25017 --- linux-2.6.36/net/ipv4/inet_hashtables.c     2010-10-21 13:07:58.000000000 +0200
25018 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/inet_hashtables.c       2010-10-21 13:09:36.000000000 +0200
25019 @@ -21,6 +21,7 @@
25020
25021  #include <net/inet_connection_sock.h>
25022  #include <net/inet_hashtables.h>
25023 +#include <net/route.h>
25024  #include <net/ip.h>
25025
25026  /*
25027 @@ -132,6 +133,11 @@ static inline int compute_score(struct s
25028                         if (rcv_saddr != daddr)
25029                                 return -1;
25030                         score += 2;
25031 +               } else {
25032 +                       /* block non nx_info ips */
25033 +                       if (!v4_addr_in_nx_info(sk->sk_nx_info,
25034 +                               daddr, NXA_MASK_BIND))
25035 +                               return -1;
25036                 }
25037                 if (sk->sk_bound_dev_if) {
25038                         if (sk->sk_bound_dev_if != dif)
25039 @@ -149,7 +155,6 @@ static inline int compute_score(struct s
25040   * wildcarded during the search since they can never be otherwise.
25041   */
25042
25043 -
25044  struct sock *__inet_lookup_listener(struct net *net,
25045                                     struct inet_hashinfo *hashinfo,
25046                                     const __be32 daddr, const unsigned short hnum,
25047 @@ -172,6 +177,7 @@ begin:
25048                         hiscore = score;
25049                 }
25050         }
25051 +
25052         /*
25053          * if the nulls value we got at the end of this lookup is
25054          * not the expected one, we must restart lookup.
25055 diff -NurpP --minimal linux-2.6.36/net/ipv4/netfilter/nf_nat_helper.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/netfilter/nf_nat_helper.c
25056 --- linux-2.6.36/net/ipv4/netfilter/nf_nat_helper.c     2010-07-07 18:31:59.000000000 +0200
25057 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/netfilter/nf_nat_helper.c       2010-10-21 13:09:36.000000000 +0200
25058 @@ -20,6 +20,7 @@
25059  #include <net/route.h>
25060
25061  #include <linux/netfilter_ipv4.h>
25062 +#include <net/route.h>
25063  #include <net/netfilter/nf_conntrack.h>
25064  #include <net/netfilter/nf_conntrack_helper.h>
25065  #include <net/netfilter/nf_conntrack_ecache.h>
25066 diff -NurpP --minimal linux-2.6.36/net/ipv4/netfilter.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/netfilter.c
25067 --- linux-2.6.36/net/ipv4/netfilter.c   2010-10-21 13:07:58.000000000 +0200
25068 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/netfilter.c     2010-10-21 13:09:36.000000000 +0200
25069 @@ -5,7 +5,7 @@
25070  #include <linux/ip.h>
25071  #include <linux/skbuff.h>
25072  #include <linux/gfp.h>
25073 -#include <net/route.h>
25074 +// #include <net/route.h>
25075  #include <net/xfrm.h>
25076  #include <net/ip.h>
25077  #include <net/netfilter/nf_queue.h>
25078 diff -NurpP --minimal linux-2.6.36/net/ipv4/raw.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/raw.c
25079 --- linux-2.6.36/net/ipv4/raw.c 2010-10-21 13:07:58.000000000 +0200
25080 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/raw.c   2010-10-21 13:09:36.000000000 +0200
25081 @@ -116,7 +116,7 @@ static struct sock *__raw_v4_lookup(stru
25082
25083                 if (net_eq(sock_net(sk), net) && inet->inet_num == num  &&
25084                     !(inet->inet_daddr && inet->inet_daddr != raddr)    &&
25085 -                   !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
25086 +                   v4_sock_addr_match(sk->sk_nx_info, inet, laddr)     &&
25087                     !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
25088                         goto found; /* gotcha */
25089         }
25090 @@ -383,6 +383,12 @@ static int raw_send_hdrinc(struct sock *
25091                 icmp_out_count(net, ((struct icmphdr *)
25092                         skb_transport_header(skb))->type);
25093
25094 +       err = -EPERM;
25095 +       if (!nx_check(0, VS_ADMIN) && !capable(CAP_NET_RAW) &&
25096 +               sk->sk_nx_info &&
25097 +               !v4_addr_in_nx_info(sk->sk_nx_info, iph->saddr, NXA_MASK_BIND))
25098 +               goto error_free;
25099 +
25100         err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
25101                       rt->dst.dev, dst_output);
25102         if (err > 0)
25103 @@ -563,6 +569,13 @@ static int raw_sendmsg(struct kiocb *ioc
25104                 }
25105
25106                 security_sk_classify_flow(sk, &fl);
25107 +               if (sk->sk_nx_info) {
25108 +                       err = ip_v4_find_src(sock_net(sk),
25109 +                               sk->sk_nx_info, &rt, &fl);
25110 +
25111 +                       if (err)
25112 +                               goto done;
25113 +               }
25114                 err = ip_route_output_flow(sock_net(sk), &rt, &fl, sk, 1);
25115         }
25116         if (err)
25117 @@ -635,17 +648,19 @@ static int raw_bind(struct sock *sk, str
25118  {
25119         struct inet_sock *inet = inet_sk(sk);
25120         struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
25121 +       struct nx_v4_sock_addr nsa = { 0 };
25122         int ret = -EINVAL;
25123         int chk_addr_ret;
25124
25125         if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
25126                 goto out;
25127 -       chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
25128 +       v4_map_sock_addr(inet, addr, &nsa);
25129 +       chk_addr_ret = inet_addr_type(sock_net(sk), nsa.saddr);
25130         ret = -EADDRNOTAVAIL;
25131 -       if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
25132 +       if (nsa.saddr && chk_addr_ret != RTN_LOCAL &&
25133             chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
25134                 goto out;
25135 -       inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
25136 +       v4_set_sock_addr(inet, &nsa);
25137         if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
25138                 inet->inet_saddr = 0;  /* Use device */
25139         sk_dst_reset(sk);
25140 @@ -697,7 +712,8 @@ static int raw_recvmsg(struct kiocb *ioc
25141         /* Copy the address. */
25142         if (sin) {
25143                 sin->sin_family = AF_INET;
25144 -               sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
25145 +               sin->sin_addr.s_addr =
25146 +                       nx_map_sock_lback(sk->sk_nx_info, ip_hdr(skb)->saddr);
25147                 sin->sin_port = 0;
25148                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
25149         }
25150 @@ -875,7 +891,8 @@ static struct sock *raw_get_first(struct
25151                 struct hlist_node *node;
25152
25153                 sk_for_each(sk, node, &state->h->ht[state->bucket])
25154 -                       if (sock_net(sk) == seq_file_net(seq))
25155 +                       if ((sock_net(sk) == seq_file_net(seq)) &&
25156 +                               nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25157                                 goto found;
25158         }
25159         sk = NULL;
25160 @@ -891,7 +908,8 @@ static struct sock *raw_get_next(struct
25161                 sk = sk_next(sk);
25162  try_again:
25163                 ;
25164 -       } while (sk && sock_net(sk) != seq_file_net(seq));
25165 +       } while (sk && ((sock_net(sk) != seq_file_net(seq)) ||
25166 +               !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
25167
25168         if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
25169                 sk = sk_head(&state->h->ht[state->bucket]);
25170 @@ -950,7 +968,10 @@ static void raw_sock_seq_show(struct seq
25171
25172         seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
25173                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
25174 -               i, src, srcp, dest, destp, sp->sk_state,
25175 +               i,
25176 +               nx_map_sock_lback(current_nx_info(), src), srcp,
25177 +               nx_map_sock_lback(current_nx_info(), dest), destp,
25178 +               sp->sk_state,
25179                 sk_wmem_alloc_get(sp),
25180                 sk_rmem_alloc_get(sp),
25181                 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
25182 diff -NurpP --minimal linux-2.6.36/net/ipv4/tcp.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp.c
25183 --- linux-2.6.36/net/ipv4/tcp.c 2010-10-21 13:07:58.000000000 +0200
25184 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp.c   2010-10-21 13:09:36.000000000 +0200
25185 @@ -266,6 +266,7 @@
25186  #include <linux/crypto.h>
25187  #include <linux/time.h>
25188  #include <linux/slab.h>
25189 +#include <linux/in.h>
25190
25191  #include <net/icmp.h>
25192  #include <net/tcp.h>
25193 diff -NurpP --minimal linux-2.6.36/net/ipv4/tcp_ipv4.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp_ipv4.c
25194 --- linux-2.6.36/net/ipv4/tcp_ipv4.c    2010-10-21 13:07:58.000000000 +0200
25195 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp_ipv4.c      2010-10-21 13:09:36.000000000 +0200
25196 @@ -2016,6 +2016,12 @@ static void *listening_get_next(struct s
25197                 req = req->dl_next;
25198                 while (1) {
25199                         while (req) {
25200 +                               vxdprintk(VXD_CBIT(net, 6),
25201 +                                       "sk,req: %p [#%d] (from %d)", req->sk,
25202 +                                       (req->sk)?req->sk->sk_nid:0, nx_current_nid());
25203 +                               if (req->sk &&
25204 +                                       !nx_check(req->sk->sk_nid, VS_WATCH_P | VS_IDENT))
25205 +                                       continue;
25206                                 if (req->rsk_ops->family == st->family) {
25207                                         cur = req;
25208                                         goto out;
25209 @@ -2041,6 +2047,10 @@ get_req:
25210         }
25211  get_sk:
25212         sk_nulls_for_each_from(sk, node) {
25213 +               vxdprintk(VXD_CBIT(net, 6), "sk: %p [#%d] (from %d)",
25214 +                       sk, sk->sk_nid, nx_current_nid());
25215 +               if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25216 +                       continue;
25217                 if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) {
25218                         cur = sk;
25219                         goto out;
25220 @@ -2115,6 +2125,11 @@ static void *established_get_first(struc
25221
25222                 spin_lock_bh(lock);
25223                 sk_nulls_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
25224 +                       vxdprintk(VXD_CBIT(net, 6),
25225 +                               "sk,egf: %p [#%d] (from %d)",
25226 +                               sk, sk->sk_nid, nx_current_nid());
25227 +                       if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25228 +                               continue;
25229                         if (sk->sk_family != st->family ||
25230                             !net_eq(sock_net(sk), net)) {
25231                                 continue;
25232 @@ -2125,6 +2140,11 @@ static void *established_get_first(struc
25233                 st->state = TCP_SEQ_STATE_TIME_WAIT;
25234                 inet_twsk_for_each(tw, node,
25235                                    &tcp_hashinfo.ehash[st->bucket].twchain) {
25236 +                       vxdprintk(VXD_CBIT(net, 6),
25237 +                               "tw: %p [#%d] (from %d)",
25238 +                               tw, tw->tw_nid, nx_current_nid());
25239 +                       if (!nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))
25240 +                               continue;
25241                         if (tw->tw_family != st->family ||
25242                             !net_eq(twsk_net(tw), net)) {
25243                                 continue;
25244 @@ -2154,7 +2174,9 @@ static void *established_get_next(struct
25245                 tw = cur;
25246                 tw = tw_next(tw);
25247  get_tw:
25248 -               while (tw && (tw->tw_family != st->family || !net_eq(twsk_net(tw), net))) {
25249 +               while (tw && (tw->tw_family != st->family ||
25250 +                       !net_eq(twsk_net(tw), net) ||
25251 +                       !nx_check(tw->tw_nid, VS_WATCH_P | VS_IDENT))) {
25252                         tw = tw_next(tw);
25253                 }
25254                 if (tw) {
25255 @@ -2178,6 +2200,11 @@ get_tw:
25256                 sk = sk_nulls_next(sk);
25257
25258         sk_nulls_for_each_from(sk, node) {
25259 +               vxdprintk(VXD_CBIT(net, 6),
25260 +                       "sk,egn: %p [#%d] (from %d)",
25261 +                       sk, sk->sk_nid, nx_current_nid());
25262 +               if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25263 +                       continue;
25264                 if (sk->sk_family == st->family && net_eq(sock_net(sk), net))
25265                         goto found;
25266         }
25267 @@ -2387,9 +2414,9 @@ static void get_openreq4(struct sock *sk
25268         seq_printf(f, "%4d: %08X:%04X %08X:%04X"
25269                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %u %d %p%n",
25270                 i,
25271 -               ireq->loc_addr,
25272 +               nx_map_sock_lback(current_nx_info(), ireq->loc_addr),
25273                 ntohs(inet_sk(sk)->inet_sport),
25274 -               ireq->rmt_addr,
25275 +               nx_map_sock_lback(current_nx_info(), ireq->rmt_addr),
25276                 ntohs(ireq->rmt_port),
25277                 TCP_SYN_RECV,
25278                 0, 0, /* could print option size, but that is af dependent. */
25279 @@ -2441,7 +2468,10 @@ static void get_tcp4_sock(struct sock *s
25280
25281         seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
25282                         "%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
25283 -               i, src, srcp, dest, destp, sk->sk_state,
25284 +               i,
25285 +               nx_map_sock_lback(current_nx_info(), src), srcp,
25286 +               nx_map_sock_lback(current_nx_info(), dest), destp,
25287 +               sk->sk_state,
25288                 tp->write_seq - tp->snd_una,
25289                 rx_queue,
25290                 timer_active,
25291 @@ -2476,7 +2506,10 @@ static void get_timewait4_sock(struct in
25292
25293         seq_printf(f, "%4d: %08X:%04X %08X:%04X"
25294                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %d %d %p%n",
25295 -               i, src, srcp, dest, destp, tw->tw_substate, 0, 0,
25296 +               i,
25297 +               nx_map_sock_lback(current_nx_info(), src), srcp,
25298 +               nx_map_sock_lback(current_nx_info(), dest), destp,
25299 +               tw->tw_substate, 0, 0,
25300                 3, jiffies_to_clock_t(ttd), 0, 0, 0, 0,
25301                 atomic_read(&tw->tw_refcnt), tw, len);
25302  }
25303 diff -NurpP --minimal linux-2.6.36/net/ipv4/tcp_minisocks.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp_minisocks.c
25304 --- linux-2.6.36/net/ipv4/tcp_minisocks.c       2010-10-21 13:07:58.000000000 +0200
25305 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/tcp_minisocks.c 2010-10-21 13:09:36.000000000 +0200
25306 @@ -23,6 +23,9 @@
25307  #include <linux/slab.h>
25308  #include <linux/sysctl.h>
25309  #include <linux/workqueue.h>
25310 +#include <linux/vs_limit.h>
25311 +#include <linux/vs_socket.h>
25312 +#include <linux/vs_context.h>
25313  #include <net/tcp.h>
25314  #include <net/inet_common.h>
25315  #include <net/xfrm.h>
25316 @@ -290,6 +293,11 @@ void tcp_time_wait(struct sock *sk, int
25317                 tcptw->tw_ts_recent     = tp->rx_opt.ts_recent;
25318                 tcptw->tw_ts_recent_stamp = tp->rx_opt.ts_recent_stamp;
25319
25320 +               tw->tw_xid              = sk->sk_xid;
25321 +               tw->tw_vx_info          = NULL;
25322 +               tw->tw_nid              = sk->sk_nid;
25323 +               tw->tw_nx_info          = NULL;
25324 +
25325  #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
25326                 if (tw->tw_family == PF_INET6) {
25327                         struct ipv6_pinfo *np = inet6_sk(sk);
25328 diff -NurpP --minimal linux-2.6.36/net/ipv4/udp.c linux-2.6.36-vs2.3.0.36.38/net/ipv4/udp.c
25329 --- linux-2.6.36/net/ipv4/udp.c 2010-10-21 13:07:59.000000000 +0200
25330 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv4/udp.c   2010-10-21 13:09:36.000000000 +0200
25331 @@ -296,14 +296,7 @@ fail:
25332  }
25333  EXPORT_SYMBOL(udp_lib_get_port);
25334
25335 -static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
25336 -{
25337 -       struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
25338 -
25339 -       return  (!ipv6_only_sock(sk2)  &&
25340 -                (!inet1->inet_rcv_saddr || !inet2->inet_rcv_saddr ||
25341 -                  inet1->inet_rcv_saddr == inet2->inet_rcv_saddr));
25342 -}
25343 +extern int ipv4_rcv_saddr_equal(const struct sock *, const struct sock *);
25344
25345  static unsigned int udp4_portaddr_hash(struct net *net, __be32 saddr,
25346                                        unsigned int port)
25347 @@ -338,6 +331,11 @@ static inline int compute_score(struct s
25348                         if (inet->inet_rcv_saddr != daddr)
25349                                 return -1;
25350                         score += 2;
25351 +               } else {
25352 +                       /* block non nx_info ips */
25353 +                       if (!v4_addr_in_nx_info(sk->sk_nx_info,
25354 +                               daddr, NXA_MASK_BIND))
25355 +                               return -1;
25356                 }
25357                 if (inet->inet_daddr) {
25358                         if (inet->inet_daddr != saddr)
25359 @@ -441,6 +439,7 @@ exact_match:
25360         return result;
25361  }
25362
25363 +
25364  /* UDP is nearly always wildcards out the wazoo, it makes no sense to try
25365   * harder than this. -DaveM
25366   */
25367 @@ -486,6 +485,11 @@ begin:
25368         sk_nulls_for_each_rcu(sk, node, &hslot->head) {
25369                 score = compute_score(sk, net, saddr, hnum, sport,
25370                                       daddr, dport, dif);
25371 +               /* FIXME: disabled?
25372 +               if (score == 9) {
25373 +                       result = sk;
25374 +                       break;
25375 +               } else */
25376                 if (score > badness) {
25377                         result = sk;
25378                         badness = score;
25379 @@ -499,6 +503,7 @@ begin:
25380         if (get_nulls_value(node) != slot)
25381                 goto begin;
25382
25383 +
25384         if (result) {
25385                 if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
25386                         result = NULL;
25387 @@ -508,6 +513,7 @@ begin:
25388                         goto begin;
25389                 }
25390         }
25391 +
25392         rcu_read_unlock();
25393         return result;
25394  }
25395 @@ -550,8 +556,7 @@ static inline struct sock *udp_v4_mcast_
25396                     udp_sk(s)->udp_port_hash != hnum ||
25397                     (inet->inet_daddr && inet->inet_daddr != rmt_addr) ||
25398                     (inet->inet_dport != rmt_port && inet->inet_dport) ||
25399 -                   (inet->inet_rcv_saddr &&
25400 -                    inet->inet_rcv_saddr != loc_addr) ||
25401 +                   !v4_sock_addr_match(sk->sk_nx_info, inet, loc_addr) ||
25402                     ipv6_only_sock(s) ||
25403                     (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
25404                         continue;
25405 @@ -900,8 +905,13 @@ int udp_sendmsg(struct kiocb *iocb, stru
25406                                                { .sport = inet->inet_sport,
25407                                                  .dport = dport } } };
25408                 struct net *net = sock_net(sk);
25409 +               struct nx_info *nxi = sk->sk_nx_info;
25410
25411                 security_sk_classify_flow(sk, &fl);
25412 +               err = ip_v4_find_src(net, nxi, &rt, &fl);
25413 +               if (err)
25414 +                       goto out;
25415 +
25416                 err = ip_route_output_flow(net, &rt, &fl, sk, 1);
25417                 if (err) {
25418                         if (err == -ENETUNREACH)
25419 @@ -1183,7 +1193,8 @@ try_again:
25420         if (sin) {
25421                 sin->sin_family = AF_INET;
25422                 sin->sin_port = udp_hdr(skb)->source;
25423 -               sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
25424 +               sin->sin_addr.s_addr = nx_map_sock_lback(
25425 +                       skb->sk->sk_nx_info, ip_hdr(skb)->saddr);
25426                 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
25427         }
25428         if (inet->cmsg_flags)
25429 @@ -1923,6 +1934,8 @@ static struct sock *udp_get_first(struct
25430                 sk_nulls_for_each(sk, node, &hslot->head) {
25431                         if (!net_eq(sock_net(sk), net))
25432                                 continue;
25433 +                       if (!nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT))
25434 +                               continue;
25435                         if (sk->sk_family == state->family)
25436                                 goto found;
25437                 }
25438 @@ -1940,7 +1953,9 @@ static struct sock *udp_get_next(struct
25439
25440         do {
25441                 sk = sk_nulls_next(sk);
25442 -       } while (sk && (!net_eq(sock_net(sk), net) || sk->sk_family != state->family));
25443 +       } while (sk && (!net_eq(sock_net(sk), net) ||
25444 +               sk->sk_family != state->family ||
25445 +               !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)));
25446
25447         if (!sk) {
25448                 if (state->bucket <= state->udp_table->mask)
25449 @@ -2047,7 +2062,10 @@ static void udp4_format_sock(struct sock
25450
25451         seq_printf(f, "%5d: %08X:%04X %08X:%04X"
25452                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d%n",
25453 -               bucket, src, srcp, dest, destp, sp->sk_state,
25454 +               bucket,
25455 +               nx_map_sock_lback(current_nx_info(), src), srcp,
25456 +               nx_map_sock_lback(current_nx_info(), dest), destp,
25457 +               sp->sk_state,
25458                 sk_wmem_alloc_get(sp),
25459                 sk_rmem_alloc_get(sp),
25460                 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
25461 diff -NurpP --minimal linux-2.6.36/net/ipv6/Kconfig linux-2.6.36-vs2.3.0.36.38/net/ipv6/Kconfig
25462 --- linux-2.6.36/net/ipv6/Kconfig       2010-08-02 16:52:59.000000000 +0200
25463 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/Kconfig 2010-10-21 13:09:36.000000000 +0200
25464 @@ -4,8 +4,8 @@
25465
25466  #   IPv6 as module will cause a CRASH if you try to unload it
25467  menuconfig IPV6
25468 -       tristate "The IPv6 protocol"
25469 -       default m
25470 +       bool "The IPv6 protocol"
25471 +       default n
25472         ---help---
25473           This is complemental support for the IP version 6.
25474           You will still be able to do traditional IPv4 networking as well.
25475 diff -NurpP --minimal linux-2.6.36/net/ipv6/addrconf.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/addrconf.c
25476 --- linux-2.6.36/net/ipv6/addrconf.c    2010-10-21 13:07:59.000000000 +0200
25477 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/addrconf.c      2010-10-21 13:09:36.000000000 +0200
25478 @@ -87,6 +87,8 @@
25479
25480  #include <linux/proc_fs.h>
25481  #include <linux/seq_file.h>
25482 +#include <linux/vs_network.h>
25483 +#include <linux/vs_inet6.h>
25484
25485  /* Set to 3 to get tracing... */
25486  #define ACONF_DEBUG 2
25487 @@ -1119,7 +1121,7 @@ out:
25488
25489  int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
25490                        const struct in6_addr *daddr, unsigned int prefs,
25491 -                      struct in6_addr *saddr)
25492 +                      struct in6_addr *saddr, struct nx_info *nxi)
25493  {
25494         struct ipv6_saddr_score scores[2],
25495                                 *score = &scores[0], *hiscore = &scores[1];
25496 @@ -1191,6 +1193,8 @@ int ipv6_dev_get_saddr(struct net *net,
25497                                                dev->name);
25498                                 continue;
25499                         }
25500 +                       if (!v6_addr_in_nx_info(nxi, &score->ifa->addr, -1))
25501 +                               continue;
25502
25503                         score->rule = -1;
25504                         bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
25505 @@ -3077,7 +3081,10 @@ static void if6_seq_stop(struct seq_file
25506  static int if6_seq_show(struct seq_file *seq, void *v)
25507  {
25508         struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
25509 -       seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
25510 +
25511 +       if (nx_check(0, VS_ADMIN|VS_WATCH) ||
25512 +           v6_addr_in_nx_info(current_nx_info(), &ifp->addr, -1))
25513 +               seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
25514                    &ifp->addr,
25515                    ifp->idev->dev->ifindex,
25516                    ifp->prefix_len,
25517 @@ -3585,6 +3592,11 @@ static int in6_dump_addrs(struct inet6_d
25518         struct ifacaddr6 *ifaca;
25519         int err = 1;
25520         int ip_idx = *p_ip_idx;
25521 +       struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
25522 +
25523 +       /* disable ipv6 on non v6 guests */
25524 +       if (nxi && !nx_info_has_v6(nxi))
25525 +               return skb->len;
25526
25527         read_lock_bh(&idev->lock);
25528         switch (type) {
25529 @@ -3595,6 +3607,8 @@ static int in6_dump_addrs(struct inet6_d
25530                 list_for_each_entry(ifa, &idev->addr_list, if_list) {
25531                         if (++ip_idx < s_ip_idx)
25532                                 continue;
25533 +                               if (!v6_addr_in_nx_info(nxi, &ifa->addr, -1))
25534 +                                       continue;
25535                         err = inet6_fill_ifaddr(skb, ifa,
25536                                                 NETLINK_CB(cb->skb).pid,
25537                                                 cb->nlh->nlmsg_seq,
25538 @@ -3611,6 +3625,8 @@ static int in6_dump_addrs(struct inet6_d
25539                      ifmca = ifmca->next, ip_idx++) {
25540                         if (ip_idx < s_ip_idx)
25541                                 continue;
25542 +                               if (!v6_addr_in_nx_info(nxi, &ifmca->mca_addr, -1))
25543 +                                       continue;
25544                         err = inet6_fill_ifmcaddr(skb, ifmca,
25545                                                   NETLINK_CB(cb->skb).pid,
25546                                                   cb->nlh->nlmsg_seq,
25547 @@ -3626,6 +3642,8 @@ static int in6_dump_addrs(struct inet6_d
25548                      ifaca = ifaca->aca_next, ip_idx++) {
25549                         if (ip_idx < s_ip_idx)
25550                                 continue;
25551 +                               if (!v6_addr_in_nx_info(nxi, &ifaca->aca_addr, -1))
25552 +                                       continue;
25553                         err = inet6_fill_ifacaddr(skb, ifaca,
25554                                                   NETLINK_CB(cb->skb).pid,
25555                                                   cb->nlh->nlmsg_seq,
25556 @@ -3973,6 +3991,11 @@ static int inet6_dump_ifinfo(struct sk_b
25557         struct inet6_dev *idev;
25558         struct hlist_head *head;
25559         struct hlist_node *node;
25560 +       struct nx_info *nxi = skb->sk ? skb->sk->sk_nx_info : NULL;
25561 +
25562 +       /* FIXME: maybe disable ipv6 on non v6 guests?
25563 +       if (skb->sk && skb->sk->sk_vx_info)
25564 +               return skb->len; */
25565
25566         s_h = cb->args[0];
25567         s_idx = cb->args[1];
25568 @@ -3984,6 +4007,8 @@ static int inet6_dump_ifinfo(struct sk_b
25569                 hlist_for_each_entry_rcu(dev, node, head, index_hlist) {
25570                         if (idx < s_idx)
25571                                 goto cont;
25572 +                       if (!v6_dev_in_nx_info(dev, nxi))
25573 +                               goto cont;
25574                         idev = __in6_dev_get(dev);
25575                         if (!idev)
25576                                 goto cont;
25577 diff -NurpP --minimal linux-2.6.36/net/ipv6/af_inet6.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/af_inet6.c
25578 --- linux-2.6.36/net/ipv6/af_inet6.c    2010-10-21 13:07:59.000000000 +0200
25579 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/af_inet6.c      2010-10-21 13:09:36.000000000 +0200
25580 @@ -42,6 +42,8 @@
25581  #include <linux/netdevice.h>
25582  #include <linux/icmpv6.h>
25583  #include <linux/netfilter_ipv6.h>
25584 +#include <linux/vs_inet.h>
25585 +#include <linux/vs_inet6.h>
25586
25587  #include <net/ip.h>
25588  #include <net/ipv6.h>
25589 @@ -160,9 +162,12 @@ lookup_protocol:
25590         }
25591
25592         err = -EPERM;
25593 +       if ((protocol == IPPROTO_ICMPV6) &&
25594 +               nx_capable(CAP_NET_RAW, NXC_RAW_ICMP))
25595 +               goto override;
25596         if (sock->type == SOCK_RAW && !kern && !capable(CAP_NET_RAW))
25597                 goto out_rcu_unlock;
25598 -
25599 +override:
25600         sock->ops = answer->ops;
25601         answer_prot = answer->prot;
25602         answer_no_check = answer->no_check;
25603 @@ -261,6 +266,7 @@ int inet6_bind(struct socket *sock, stru
25604         struct inet_sock *inet = inet_sk(sk);
25605         struct ipv6_pinfo *np = inet6_sk(sk);
25606         struct net *net = sock_net(sk);
25607 +       struct nx_v6_sock_addr nsa;
25608         __be32 v4addr = 0;
25609         unsigned short snum;
25610         int addr_type = 0;
25611 @@ -272,6 +278,11 @@ int inet6_bind(struct socket *sock, stru
25612
25613         if (addr_len < SIN6_LEN_RFC2133)
25614                 return -EINVAL;
25615 +
25616 +       err = v6_map_sock_addr(inet, addr, &nsa);
25617 +       if (err)
25618 +               return err;
25619 +
25620         addr_type = ipv6_addr_type(&addr->sin6_addr);
25621         if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
25622                 return -EINVAL;
25623 @@ -303,6 +314,7 @@ int inet6_bind(struct socket *sock, stru
25624                 /* Reproduce AF_INET checks to make the bindings consitant */
25625                 v4addr = addr->sin6_addr.s6_addr32[3];
25626                 chk_addr_ret = inet_addr_type(net, v4addr);
25627 +
25628                 if (!sysctl_ip_nonlocal_bind &&
25629                     !(inet->freebind || inet->transparent) &&
25630                     v4addr != htonl(INADDR_ANY) &&
25631 @@ -312,6 +324,10 @@ int inet6_bind(struct socket *sock, stru
25632                         err = -EADDRNOTAVAIL;
25633                         goto out;
25634                 }
25635 +               if (!v4_addr_in_nx_info(sk->sk_nx_info, v4addr, NXA_MASK_BIND)) {
25636 +                       err = -EADDRNOTAVAIL;
25637 +                       goto out;
25638 +               }
25639         } else {
25640                 if (addr_type != IPV6_ADDR_ANY) {
25641                         struct net_device *dev = NULL;
25642 @@ -338,6 +354,11 @@ int inet6_bind(struct socket *sock, stru
25643                                 }
25644                         }
25645
25646 +                       if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
25647 +                               err = -EADDRNOTAVAIL;
25648 +                               goto out;
25649 +                       }
25650 +
25651                         /* ipv4 addr of the socket is invalid.  Only the
25652                          * unspecified and mapped address have a v4 equivalent.
25653                          */
25654 @@ -353,6 +374,9 @@ int inet6_bind(struct socket *sock, stru
25655                 }
25656         }
25657
25658 +       /* what's that for? */
25659 +       v6_set_sock_addr(inet, &nsa);
25660 +
25661         inet->inet_rcv_saddr = v4addr;
25662         inet->inet_saddr = v4addr;
25663
25664 @@ -454,9 +478,11 @@ int inet6_getname(struct socket *sock, s
25665                         return -ENOTCONN;
25666                 sin->sin6_port = inet->inet_dport;
25667                 ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
25668 +               /* FIXME: remap lback? */
25669                 if (np->sndflow)
25670                         sin->sin6_flowinfo = np->flow_label;
25671         } else {
25672 +               /* FIXME: remap lback? */
25673                 if (ipv6_addr_any(&np->rcv_saddr))
25674                         ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
25675                 else
25676 diff -NurpP --minimal linux-2.6.36/net/ipv6/fib6_rules.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/fib6_rules.c
25677 --- linux-2.6.36/net/ipv6/fib6_rules.c  2010-10-21 13:07:59.000000000 +0200
25678 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/fib6_rules.c    2010-10-21 13:09:36.000000000 +0200
25679 @@ -89,7 +89,7 @@ static int fib6_rule_action(struct fib_r
25680                                                ip6_dst_idev(&rt->dst)->dev,
25681                                                &flp->fl6_dst,
25682                                                rt6_flags2srcprefs(flags),
25683 -                                              &saddr))
25684 +                                              &saddr, NULL))
25685                                 goto again;
25686                         if (!ipv6_prefix_equal(&saddr, &r->src.addr,
25687                                                r->src.plen))
25688 diff -NurpP --minimal linux-2.6.36/net/ipv6/inet6_hashtables.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/inet6_hashtables.c
25689 --- linux-2.6.36/net/ipv6/inet6_hashtables.c    2010-02-25 11:52:10.000000000 +0100
25690 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/inet6_hashtables.c      2010-10-21 13:09:36.000000000 +0200
25691 @@ -16,6 +16,7 @@
25692
25693  #include <linux/module.h>
25694  #include <linux/random.h>
25695 +#include <linux/vs_inet6.h>
25696
25697  #include <net/inet_connection_sock.h>
25698  #include <net/inet_hashtables.h>
25699 @@ -82,7 +83,6 @@ struct sock *__inet6_lookup_established(
25700         unsigned int slot = hash & hashinfo->ehash_mask;
25701         struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
25702
25703 -
25704         rcu_read_lock();
25705  begin:
25706         sk_nulls_for_each_rcu(sk, node, &head->chain) {
25707 @@ -94,7 +94,7 @@ begin:
25708                                 sock_put(sk);
25709                                 goto begin;
25710                         }
25711 -               goto out;
25712 +                       goto out;
25713                 }
25714         }
25715         if (get_nulls_value(node) != slot)
25716 @@ -140,6 +140,9 @@ static int inline compute_score(struct s
25717                         if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
25718                                 return -1;
25719                         score++;
25720 +               } else {
25721 +                       if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
25722 +                               return -1;
25723                 }
25724                 if (sk->sk_bound_dev_if) {
25725                         if (sk->sk_bound_dev_if != dif)
25726 diff -NurpP --minimal linux-2.6.36/net/ipv6/ip6_output.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/ip6_output.c
25727 --- linux-2.6.36/net/ipv6/ip6_output.c  2010-10-21 13:07:59.000000000 +0200
25728 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/ip6_output.c    2010-10-21 13:09:36.000000000 +0200
25729 @@ -938,7 +938,7 @@ static int ip6_dst_lookup_tail(struct so
25730                 err = ipv6_dev_get_saddr(net, ip6_dst_idev(*dst)->dev,
25731                                          &fl->fl6_dst,
25732                                          sk ? inet6_sk(sk)->srcprefs : 0,
25733 -                                        &fl->fl6_src);
25734 +                                        &fl->fl6_src, sk->sk_nx_info);
25735                 if (err)
25736                         goto out_err_release;
25737         }
25738 diff -NurpP --minimal linux-2.6.36/net/ipv6/ndisc.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/ndisc.c
25739 --- linux-2.6.36/net/ipv6/ndisc.c       2010-10-21 13:07:59.000000000 +0200
25740 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/ndisc.c 2010-10-21 13:09:36.000000000 +0200
25741 @@ -591,7 +591,7 @@ static void ndisc_send_na(struct net_dev
25742         } else {
25743                 if (ipv6_dev_get_saddr(dev_net(dev), dev, daddr,
25744                                        inet6_sk(dev_net(dev)->ipv6.ndisc_sk)->srcprefs,
25745 -                                      &tmpaddr))
25746 +                                      &tmpaddr, NULL /* FIXME: ? */ ))
25747                         return;
25748                 src_addr = &tmpaddr;
25749         }
25750 diff -NurpP --minimal linux-2.6.36/net/ipv6/raw.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/raw.c
25751 --- linux-2.6.36/net/ipv6/raw.c 2010-10-21 13:07:59.000000000 +0200
25752 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/raw.c   2010-10-21 13:09:36.000000000 +0200
25753 @@ -30,6 +30,7 @@
25754  #include <linux/icmpv6.h>
25755  #include <linux/netfilter.h>
25756  #include <linux/netfilter_ipv6.h>
25757 +#include <linux/vs_inet6.h>
25758  #include <linux/skbuff.h>
25759  #include <asm/uaccess.h>
25760  #include <asm/ioctls.h>
25761 @@ -283,6 +284,13 @@ static int rawv6_bind(struct sock *sk, s
25762                                 goto out_unlock;
25763                 }
25764
25765 +               if (!v6_addr_in_nx_info(sk->sk_nx_info, &addr->sin6_addr, -1)) {
25766 +                       err = -EADDRNOTAVAIL;
25767 +                       if (dev)
25768 +                               dev_put(dev);
25769 +                       goto out;
25770 +               }
25771 +
25772                 /* ipv4 addr of the socket is invalid.  Only the
25773                  * unspecified and mapped address have a v4 equivalent.
25774                  */
25775 diff -NurpP --minimal linux-2.6.36/net/ipv6/route.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/route.c
25776 --- linux-2.6.36/net/ipv6/route.c       2010-10-21 13:07:59.000000000 +0200
25777 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/route.c 2010-10-21 13:09:36.000000000 +0200
25778 @@ -2272,7 +2272,8 @@ static int rt6_fill_node(struct net *net
25779                 struct inet6_dev *idev = ip6_dst_idev(&rt->dst);
25780                 struct in6_addr saddr_buf;
25781                 if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
25782 -                                      dst, 0, &saddr_buf) == 0)
25783 +                       dst, 0, &saddr_buf,
25784 +                       (skb->sk ? skb->sk->sk_nx_info : NULL)) == 0)
25785                         NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
25786         }
25787
25788 diff -NurpP --minimal linux-2.6.36/net/ipv6/tcp_ipv6.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/tcp_ipv6.c
25789 --- linux-2.6.36/net/ipv6/tcp_ipv6.c    2010-10-21 13:07:59.000000000 +0200
25790 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/tcp_ipv6.c      2010-10-21 13:09:36.000000000 +0200
25791 @@ -69,6 +69,7 @@
25792
25793  #include <linux/crypto.h>
25794  #include <linux/scatterlist.h>
25795 +#include <linux/vs_inet6.h>
25796
25797  static void    tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
25798  static void    tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
25799 @@ -160,8 +161,15 @@ static int tcp_v6_connect(struct sock *s
25800          *      connect() to INADDR_ANY means loopback (BSD'ism).
25801          */
25802
25803 -       if(ipv6_addr_any(&usin->sin6_addr))
25804 -               usin->sin6_addr.s6_addr[15] = 0x1;
25805 +       if(ipv6_addr_any(&usin->sin6_addr)) {
25806 +               struct nx_info *nxi =  sk->sk_nx_info;
25807 +
25808 +               if (nxi && nx_info_has_v6(nxi))
25809 +                       /* FIXME: remap lback? */
25810 +                       usin->sin6_addr = nxi->v6.ip;
25811 +               else
25812 +                       usin->sin6_addr.s6_addr[15] = 0x1;
25813 +       }
25814
25815         addr_type = ipv6_addr_type(&usin->sin6_addr);
25816
25817 diff -NurpP --minimal linux-2.6.36/net/ipv6/udp.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/udp.c
25818 --- linux-2.6.36/net/ipv6/udp.c 2010-10-21 13:07:59.000000000 +0200
25819 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/udp.c   2010-10-21 13:09:36.000000000 +0200
25820 @@ -48,13 +48,14 @@
25821
25822  #include <linux/proc_fs.h>
25823  #include <linux/seq_file.h>
25824 +#include <linux/vs_inet6.h>
25825  #include "udp_impl.h"
25826
25827  int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
25828  {
25829         const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
25830         const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
25831 -       __be32 sk1_rcv_saddr = inet_sk(sk)->inet_rcv_saddr;
25832 +       __be32 sk_rcv_saddr = inet_sk(sk)->inet_rcv_saddr;
25833         __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
25834         int sk_ipv6only = ipv6_only_sock(sk);
25835         int sk2_ipv6only = inet_v6_ipv6only(sk2);
25836 @@ -62,24 +63,49 @@ int ipv6_rcv_saddr_equal(const struct so
25837         int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
25838
25839         /* if both are mapped, treat as IPv4 */
25840 -       if (addr_type == IPV6_ADDR_MAPPED && addr_type2 == IPV6_ADDR_MAPPED)
25841 -               return (!sk2_ipv6only &&
25842 -                       (!sk1_rcv_saddr || !sk2_rcv_saddr ||
25843 -                         sk1_rcv_saddr == sk2_rcv_saddr));
25844 +       if (addr_type == IPV6_ADDR_MAPPED && addr_type2 == IPV6_ADDR_MAPPED) {
25845 +               if (!sk2_ipv6only &&
25846 +                       (!sk_rcv_saddr || !sk2_rcv_saddr ||
25847 +                         sk_rcv_saddr == sk2_rcv_saddr))
25848 +                       goto vs_v4;
25849 +               else
25850 +                       return 0;
25851 +       }
25852
25853         if (addr_type2 == IPV6_ADDR_ANY &&
25854             !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
25855 -               return 1;
25856 +               goto vs;
25857
25858         if (addr_type == IPV6_ADDR_ANY &&
25859             !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
25860 -               return 1;
25861 +               goto vs;
25862
25863         if (sk2_rcv_saddr6 &&
25864             ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
25865 -               return 1;
25866 +               goto vs;
25867
25868         return 0;
25869 +
25870 +vs_v4:
25871 +       if (!sk_rcv_saddr && !sk2_rcv_saddr)
25872 +               return nx_v4_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
25873 +       if (!sk2_rcv_saddr)
25874 +               return v4_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr, -1);
25875 +       if (!sk_rcv_saddr)
25876 +               return v4_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr, -1);
25877 +       return 1;
25878 +vs:
25879 +       if (addr_type2 == IPV6_ADDR_ANY && addr_type == IPV6_ADDR_ANY)
25880 +               return nx_v6_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
25881 +       else if (addr_type2 == IPV6_ADDR_ANY)
25882 +               return v6_addr_in_nx_info(sk2->sk_nx_info, sk_rcv_saddr6, -1);
25883 +       else if (addr_type == IPV6_ADDR_ANY) {
25884 +               if (addr_type2 == IPV6_ADDR_MAPPED)
25885 +                       return nx_v4_addr_conflict(sk->sk_nx_info, sk2->sk_nx_info);
25886 +               else
25887 +                       return v6_addr_in_nx_info(sk->sk_nx_info, sk2_rcv_saddr6, -1);
25888 +       }
25889 +       return 1;
25890  }
25891
25892  static unsigned int udp6_portaddr_hash(struct net *net,
25893 @@ -143,6 +169,10 @@ static inline int compute_score(struct s
25894                         if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
25895                                 return -1;
25896                         score++;
25897 +               } else {
25898 +                       /* block non nx_info ips */
25899 +                       if (!v6_addr_in_nx_info(sk->sk_nx_info, daddr, -1))
25900 +                               return -1;
25901                 }
25902                 if (!ipv6_addr_any(&np->daddr)) {
25903                         if (!ipv6_addr_equal(&np->daddr, saddr))
25904 diff -NurpP --minimal linux-2.6.36/net/ipv6/xfrm6_policy.c linux-2.6.36-vs2.3.0.36.38/net/ipv6/xfrm6_policy.c
25905 --- linux-2.6.36/net/ipv6/xfrm6_policy.c        2010-08-02 16:53:00.000000000 +0200
25906 +++ linux-2.6.36-vs2.3.0.36.38/net/ipv6/xfrm6_policy.c  2010-10-21 13:09:36.000000000 +0200
25907 @@ -62,7 +62,7 @@ static int xfrm6_get_saddr(struct net *n
25908         dev = ip6_dst_idev(dst)->dev;
25909         ipv6_dev_get_saddr(dev_net(dev), dev,
25910                            (struct in6_addr *)&daddr->a6, 0,
25911 -                          (struct in6_addr *)&saddr->a6);
25912 +                          (struct in6_addr *)&saddr->a6, NULL);
25913         dst_release(dst);
25914         return 0;
25915  }
25916 diff -NurpP --minimal linux-2.6.36/net/netlink/af_netlink.c linux-2.6.36-vs2.3.0.36.38/net/netlink/af_netlink.c
25917 --- linux-2.6.36/net/netlink/af_netlink.c       2010-10-21 13:08:01.000000000 +0200
25918 +++ linux-2.6.36-vs2.3.0.36.38/net/netlink/af_netlink.c 2010-10-21 13:09:36.000000000 +0200
25919 @@ -55,6 +55,9 @@
25920  #include <linux/types.h>
25921  #include <linux/audit.h>
25922  #include <linux/mutex.h>
25923 +#include <linux/vs_context.h>
25924 +#include <linux/vs_network.h>
25925 +#include <linux/vs_limit.h>
25926
25927  #include <net/net_namespace.h>
25928  #include <net/sock.h>
25929 @@ -1928,6 +1931,8 @@ static struct sock *netlink_seq_socket_i
25930                         sk_for_each(s, node, &hash->table[j]) {
25931                                 if (sock_net(s) != seq_file_net(seq))
25932                                         continue;
25933 +                               if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
25934 +                                       continue;
25935                                 if (off == pos) {
25936                                         iter->link = i;
25937                                         iter->hash_idx = j;
25938 @@ -1962,7 +1967,8 @@ static void *netlink_seq_next(struct seq
25939         s = v;
25940         do {
25941                 s = sk_next(s);
25942 -       } while (s && sock_net(s) != seq_file_net(seq));
25943 +       } while (s && (sock_net(s) != seq_file_net(seq) ||
25944 +               !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)));
25945         if (s)
25946                 return s;
25947
25948 @@ -1974,7 +1980,8 @@ static void *netlink_seq_next(struct seq
25949
25950                 for (; j <= hash->mask; j++) {
25951                         s = sk_head(&hash->table[j]);
25952 -                       while (s && sock_net(s) != seq_file_net(seq))
25953 +                       while (s && (sock_net(s) != seq_file_net(seq) ||
25954 +                               !nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT)))
25955                                 s = sk_next(s);
25956                         if (s) {
25957                                 iter->link = i;
25958 diff -NurpP --minimal linux-2.6.36/net/sctp/ipv6.c linux-2.6.36-vs2.3.0.36.38/net/sctp/ipv6.c
25959 --- linux-2.6.36/net/sctp/ipv6.c        2010-08-02 16:53:01.000000000 +0200
25960 +++ linux-2.6.36-vs2.3.0.36.38/net/sctp/ipv6.c  2010-10-21 13:09:36.000000000 +0200
25961 @@ -304,7 +304,8 @@ static void sctp_v6_get_saddr(struct sct
25962                                    dst ? ip6_dst_idev(dst)->dev : NULL,
25963                                    &daddr->v6.sin6_addr,
25964                                    inet6_sk(&sk->inet.sk)->srcprefs,
25965 -                                  &saddr->v6.sin6_addr);
25966 +                                  &saddr->v6.sin6_addr,
25967 +                                  asoc->base.sk->sk_nx_info);
25968                 SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: %pI6\n",
25969                                   &saddr->v6.sin6_addr);
25970                 return;
25971 diff -NurpP --minimal linux-2.6.36/net/socket.c linux-2.6.36-vs2.3.0.36.38/net/socket.c
25972 --- linux-2.6.36/net/socket.c   2010-10-21 13:08:01.000000000 +0200
25973 +++ linux-2.6.36-vs2.3.0.36.38/net/socket.c     2010-10-21 13:09:36.000000000 +0200
25974 @@ -98,6 +98,10 @@
25975
25976  #include <net/sock.h>
25977  #include <linux/netfilter.h>
25978 +#include <linux/vs_base.h>
25979 +#include <linux/vs_socket.h>
25980 +#include <linux/vs_inet.h>
25981 +#include <linux/vs_inet6.h>
25982
25983  #include <linux/if_tun.h>
25984  #include <linux/ipv6_route.h>
25985 @@ -551,7 +555,7 @@ static inline int __sock_sendmsg(struct
25986                                  struct msghdr *msg, size_t size)
25987  {
25988         struct sock_iocb *si = kiocb_to_siocb(iocb);
25989 -       int err;
25990 +       int err, len;
25991
25992         sock_update_classid(sock->sk);
25993
25994 @@ -564,7 +568,22 @@ static inline int __sock_sendmsg(struct
25995         if (err)
25996                 return err;
25997
25998 -       return sock->ops->sendmsg(iocb, sock, msg, size);
25999 +       len = sock->ops->sendmsg(iocb, sock, msg, size);
26000 +       if (sock->sk) {
26001 +               if (len == size)
26002 +                       vx_sock_send(sock->sk, size);
26003 +               else
26004 +                       vx_sock_fail(sock->sk, size);
26005 +       }
26006 +       vxdprintk(VXD_CBIT(net, 7),
26007 +               "__sock_sendmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
26008 +               sock, sock->sk,
26009 +               (sock->sk)?sock->sk->sk_nx_info:0,
26010 +               (sock->sk)?sock->sk->sk_vx_info:0,
26011 +               (sock->sk)?sock->sk->sk_xid:0,
26012 +               (sock->sk)?sock->sk->sk_nid:0,
26013 +               (unsigned int)size, len);
26014 +       return len;
26015  }
26016
26017  int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
26018 @@ -681,6 +700,7 @@ static inline int __sock_recvmsg_nosec(s
26019                                        struct msghdr *msg, size_t size, int flags)
26020  {
26021         struct sock_iocb *si = kiocb_to_siocb(iocb);
26022 +       int len;
26023
26024         sock_update_classid(sock->sk);
26025
26026 @@ -690,7 +710,18 @@ static inline int __sock_recvmsg_nosec(s
26027         si->size = size;
26028         si->flags = flags;
26029
26030 -       return sock->ops->recvmsg(iocb, sock, msg, size, flags);
26031 +       len = sock->ops->recvmsg(iocb, sock, msg, size, flags);
26032 +       if ((len >= 0) && sock->sk)
26033 +               vx_sock_recv(sock->sk, len);
26034 +       vxdprintk(VXD_CBIT(net, 7),
26035 +               "__sock_recvmsg: %p[%p,%p,%p;%d/%d]:%d/%d",
26036 +               sock, sock->sk,
26037 +               (sock->sk)?sock->sk->sk_nx_info:0,
26038 +               (sock->sk)?sock->sk->sk_vx_info:0,
26039 +               (sock->sk)?sock->sk->sk_xid:0,
26040 +               (sock->sk)?sock->sk->sk_nid:0,
26041 +               (unsigned int)size, len);
26042 +       return len;
26043  }
26044
26045  static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock,
26046 @@ -1159,6 +1190,13 @@ static int __sock_create(struct net *net
26047         if (type < 0 || type >= SOCK_MAX)
26048                 return -EINVAL;
26049
26050 +       if (!nx_check(0, VS_ADMIN)) {
26051 +               if (family == PF_INET && !current_nx_info_has_v4())
26052 +                       return -EAFNOSUPPORT;
26053 +               if (family == PF_INET6 && !current_nx_info_has_v6())
26054 +                       return -EAFNOSUPPORT;
26055 +       }
26056 +
26057         /* Compatibility.
26058
26059            This uglymoron is moved from INET layer to here to avoid
26060 @@ -1293,6 +1331,7 @@ SYSCALL_DEFINE3(socket, int, family, int
26061         if (retval < 0)
26062                 goto out;
26063
26064 +       set_bit(SOCK_USER_SOCKET, &sock->flags);
26065         retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
26066         if (retval < 0)
26067                 goto out_release;
26068 @@ -1334,10 +1373,12 @@ SYSCALL_DEFINE4(socketpair, int, family,
26069         err = sock_create(family, type, protocol, &sock1);
26070         if (err < 0)
26071                 goto out;
26072 +       set_bit(SOCK_USER_SOCKET, &sock1->flags);
26073
26074         err = sock_create(family, type, protocol, &sock2);
26075         if (err < 0)
26076                 goto out_release_1;
26077 +       set_bit(SOCK_USER_SOCKET, &sock2->flags);
26078
26079         err = sock1->ops->socketpair(sock1, sock2);
26080         if (err < 0)
26081 diff -NurpP --minimal linux-2.6.36/net/sunrpc/auth.c linux-2.6.36-vs2.3.0.36.38/net/sunrpc/auth.c
26082 --- linux-2.6.36/net/sunrpc/auth.c      2010-10-21 13:08:01.000000000 +0200
26083 +++ linux-2.6.36-vs2.3.0.36.38/net/sunrpc/auth.c        2010-10-21 13:09:36.000000000 +0200
26084 @@ -14,6 +14,7 @@
26085  #include <linux/hash.h>
26086  #include <linux/sunrpc/clnt.h>
26087  #include <linux/spinlock.h>
26088 +#include <linux/vs_tag.h>
26089
26090  #ifdef RPC_DEBUG
26091  # define RPCDBG_FACILITY       RPCDBG_AUTH
26092 @@ -425,6 +426,7 @@ rpcauth_lookupcred(struct rpc_auth *auth
26093         memset(&acred, 0, sizeof(acred));
26094         acred.uid = cred->fsuid;
26095         acred.gid = cred->fsgid;
26096 +       acred.tag = dx_current_tag();
26097         acred.group_info = get_group_info(((struct cred *)cred)->group_info);
26098
26099         ret = auth->au_ops->lookup_cred(auth, &acred, flags);
26100 @@ -465,6 +467,7 @@ rpcauth_bind_root_cred(struct rpc_task *
26101         struct auth_cred acred = {
26102                 .uid = 0,
26103                 .gid = 0,
26104 +               .tag = dx_current_tag(),
26105         };
26106
26107         dprintk("RPC: %5u looking up %s cred\n",
26108 diff -NurpP --minimal linux-2.6.36/net/sunrpc/auth_unix.c linux-2.6.36-vs2.3.0.36.38/net/sunrpc/auth_unix.c
26109 --- linux-2.6.36/net/sunrpc/auth_unix.c 2010-10-21 13:08:01.000000000 +0200
26110 +++ linux-2.6.36-vs2.3.0.36.38/net/sunrpc/auth_unix.c   2010-10-21 13:09:36.000000000 +0200
26111 @@ -12,12 +12,14 @@
26112  #include <linux/module.h>
26113  #include <linux/sunrpc/clnt.h>
26114  #include <linux/sunrpc/auth.h>
26115 +#include <linux/vs_tag.h>
26116
26117  #define NFS_NGROUPS    16
26118
26119  struct unx_cred {
26120         struct rpc_cred         uc_base;
26121         gid_t                   uc_gid;
26122 +       tag_t                   uc_tag;
26123         gid_t                   uc_gids[NFS_NGROUPS];
26124  };
26125  #define uc_uid                 uc_base.cr_uid
26126 @@ -78,6 +80,7 @@ unx_create_cred(struct rpc_auth *auth, s
26127                 groups = NFS_NGROUPS;
26128
26129         cred->uc_gid = acred->gid;
26130 +       cred->uc_tag = acred->tag;
26131         for (i = 0; i < groups; i++)
26132                 cred->uc_gids[i] = GROUP_AT(acred->group_info, i);
26133         if (i < NFS_NGROUPS)
26134 @@ -119,7 +122,9 @@ unx_match(struct auth_cred *acred, struc
26135         unsigned int i;
26136
26137
26138 -       if (cred->uc_uid != acred->uid || cred->uc_gid != acred->gid)
26139 +       if (cred->uc_uid != acred->uid ||
26140 +               cred->uc_gid != acred->gid ||
26141 +               cred->uc_tag != acred->tag)
26142                 return 0;
26143
26144         if (acred->group_info != NULL)
26145 @@ -142,7 +147,7 @@ unx_marshal(struct rpc_task *task, __be3
26146         struct rpc_clnt *clnt = task->tk_client;
26147         struct unx_cred *cred = container_of(task->tk_rqstp->rq_cred, struct unx_cred, uc_base);
26148         __be32          *base, *hold;
26149 -       int             i;
26150 +       int             i, tag;
26151
26152         *p++ = htonl(RPC_AUTH_UNIX);
26153         base = p++;
26154 @@ -152,9 +157,12 @@ unx_marshal(struct rpc_task *task, __be3
26155          * Copy the UTS nodename captured when the client was created.
26156          */
26157         p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen);
26158 +       tag = task->tk_client->cl_tag;
26159
26160 -       *p++ = htonl((u32) cred->uc_uid);
26161 -       *p++ = htonl((u32) cred->uc_gid);
26162 +       *p++ = htonl((u32) TAGINO_UID(tag,
26163 +               cred->uc_uid, cred->uc_tag));
26164 +       *p++ = htonl((u32) TAGINO_GID(tag,
26165 +               cred->uc_gid, cred->uc_tag));
26166         hold = p++;
26167         for (i = 0; i < 16 && cred->uc_gids[i] != (gid_t) NOGROUP; i++)
26168                 *p++ = htonl((u32) cred->uc_gids[i]);
26169 diff -NurpP --minimal linux-2.6.36/net/sunrpc/clnt.c linux-2.6.36-vs2.3.0.36.38/net/sunrpc/clnt.c
26170 --- linux-2.6.36/net/sunrpc/clnt.c      2010-10-21 13:08:01.000000000 +0200
26171 +++ linux-2.6.36-vs2.3.0.36.38/net/sunrpc/clnt.c        2010-10-21 13:09:36.000000000 +0200
26172 @@ -33,6 +33,7 @@
26173  #include <linux/utsname.h>
26174  #include <linux/workqueue.h>
26175  #include <linux/in6.h>
26176 +#include <linux/vs_cvirt.h>
26177
26178  #include <linux/sunrpc/clnt.h>
26179  #include <linux/sunrpc/rpc_pipe_fs.h>
26180 @@ -358,6 +359,9 @@ struct rpc_clnt *rpc_create(struct rpc_c
26181         if (!(args->flags & RPC_CLNT_CREATE_QUIET))
26182                 clnt->cl_chatty = 1;
26183
26184 +       /* TODO: handle RPC_CLNT_CREATE_TAGGED
26185 +       if (args->flags & RPC_CLNT_CREATE_TAGGED)
26186 +               clnt->cl_tag = 1; */
26187         return clnt;
26188  }
26189  EXPORT_SYMBOL_GPL(rpc_create);
26190 diff -NurpP --minimal linux-2.6.36/net/unix/af_unix.c linux-2.6.36-vs2.3.0.36.38/net/unix/af_unix.c
26191 --- linux-2.6.36/net/unix/af_unix.c     2010-10-21 13:08:01.000000000 +0200
26192 +++ linux-2.6.36-vs2.3.0.36.38/net/unix/af_unix.c       2010-10-21 13:09:36.000000000 +0200
26193 @@ -114,6 +114,8 @@
26194  #include <linux/mount.h>
26195  #include <net/checksum.h>
26196  #include <linux/security.h>
26197 +#include <linux/vs_context.h>
26198 +#include <linux/vs_limit.h>
26199
26200  static struct hlist_head unix_socket_table[UNIX_HASH_SIZE + 1];
26201  static DEFINE_SPINLOCK(unix_table_lock);
26202 @@ -258,6 +260,8 @@ static struct sock *__unix_find_socket_b
26203                 if (!net_eq(sock_net(s), net))
26204                         continue;
26205
26206 +               if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
26207 +                       continue;
26208                 if (u->addr->len == len &&
26209                     !memcmp(u->addr->name, sunname, len))
26210                         goto found;
26211 @@ -2148,6 +2152,8 @@ static struct sock *unix_seq_idx(struct
26212         for (s = first_unix_socket(&iter->i); s; s = next_unix_socket(&iter->i, s)) {
26213                 if (sock_net(s) != seq_file_net(seq))
26214                         continue;
26215 +               if (!nx_check(s->sk_nid, VS_WATCH_P | VS_IDENT))
26216 +                       continue;
26217                 if (off == pos)
26218                         return s;
26219                 ++off;
26220 @@ -2172,7 +2178,8 @@ static void *unix_seq_next(struct seq_fi
26221                 sk = first_unix_socket(&iter->i);
26222         else
26223                 sk = next_unix_socket(&iter->i, sk);
26224 -       while (sk && (sock_net(sk) != seq_file_net(seq)))
26225 +       while (sk && (sock_net(sk) != seq_file_net(seq) ||
26226 +               !nx_check(sk->sk_nid, VS_WATCH_P | VS_IDENT)))
26227                 sk = next_unix_socket(&iter->i, sk);
26228         return sk;
26229  }
26230 diff -NurpP --minimal linux-2.6.36/scripts/checksyscalls.sh linux-2.6.36-vs2.3.0.36.38/scripts/checksyscalls.sh
26231 --- linux-2.6.36/scripts/checksyscalls.sh       2010-10-21 13:08:01.000000000 +0200
26232 +++ linux-2.6.36-vs2.3.0.36.38/scripts/checksyscalls.sh 2010-10-21 13:09:36.000000000 +0200
26233 @@ -193,7 +193,6 @@ cat << EOF
26234  #define __IGNORE_afs_syscall
26235  #define __IGNORE_getpmsg
26236  #define __IGNORE_putpmsg
26237 -#define __IGNORE_vserver
26238  EOF
26239  }
26240
26241 diff -NurpP --minimal linux-2.6.36/security/commoncap.c linux-2.6.36-vs2.3.0.36.38/security/commoncap.c
26242 --- linux-2.6.36/security/commoncap.c   2010-10-21 13:08:02.000000000 +0200
26243 +++ linux-2.6.36-vs2.3.0.36.38/security/commoncap.c     2010-10-21 13:09:36.000000000 +0200
26244 @@ -28,6 +28,7 @@
26245  #include <linux/prctl.h>
26246  #include <linux/securebits.h>
26247  #include <linux/syslog.h>
26248 +#include <linux/vs_context.h>
26249
26250  /*
26251   * If a non-root user executes a setuid-root binary in
26252 @@ -53,7 +54,7 @@ static void warn_setuid_and_fcaps_mixed(
26253
26254  int cap_netlink_send(struct sock *sk, struct sk_buff *skb)
26255  {
26256 -       NETLINK_CB(skb).eff_cap = current_cap();
26257 +       NETLINK_CB(skb).eff_cap = vx_mbcaps(current_cap());
26258         return 0;
26259  }
26260
26261 @@ -63,6 +64,7 @@ int cap_netlink_recv(struct sk_buff *skb
26262                 return -EPERM;
26263         return 0;
26264  }
26265 +
26266  EXPORT_SYMBOL(cap_netlink_recv);
26267
26268  /**
26269 @@ -83,7 +85,22 @@ EXPORT_SYMBOL(cap_netlink_recv);
26270  int cap_capable(struct task_struct *tsk, const struct cred *cred, int cap,
26271                 int audit)
26272  {
26273 -       return cap_raised(cred->cap_effective, cap) ? 0 : -EPERM;
26274 +       struct vx_info *vxi = tsk->vx_info;
26275 +
26276 +#if 0
26277 +       printk("cap_capable() VXF_STATE_SETUP = %llx, raised = %x, eff = %08x:%08x\n",
26278 +               vx_info_flags(vxi, VXF_STATE_SETUP, 0),
26279 +               cap_raised(tsk->cap_effective, cap),
26280 +               tsk->cap_effective.cap[1], tsk->cap_effective.cap[0]);
26281 +#endif
26282 +
26283 +       /* special case SETUP */
26284 +       if (vx_info_flags(vxi, VXF_STATE_SETUP, 0) &&
26285 +               /* FIXME: maybe use cred instead? */
26286 +               cap_raised(tsk->cred->cap_effective, cap))
26287 +               return 0;
26288 +
26289 +       return vx_cap_raised(vxi, cred->cap_effective, cap) ? 0 : -EPERM;
26290  }
26291
26292  /**
26293 @@ -571,7 +588,7 @@ int cap_inode_setxattr(struct dentry *de
26294
26295         if (!strncmp(name, XATTR_SECURITY_PREFIX,
26296                      sizeof(XATTR_SECURITY_PREFIX) - 1) &&
26297 -           !capable(CAP_SYS_ADMIN))
26298 +               !vx_capable(CAP_SYS_ADMIN, VXC_FS_SECURITY))
26299                 return -EPERM;
26300         return 0;
26301  }
26302 @@ -597,7 +614,7 @@ int cap_inode_removexattr(struct dentry
26303
26304         if (!strncmp(name, XATTR_SECURITY_PREFIX,
26305                      sizeof(XATTR_SECURITY_PREFIX) - 1) &&
26306 -           !capable(CAP_SYS_ADMIN))
26307 +               !vx_capable(CAP_SYS_ADMIN, VXC_FS_SECURITY))
26308                 return -EPERM;
26309         return 0;
26310  }
26311 @@ -899,7 +916,8 @@ int cap_syslog(int type, bool from_file)
26312         if (type != SYSLOG_ACTION_OPEN && from_file)
26313                 return 0;
26314         if ((type != SYSLOG_ACTION_READ_ALL &&
26315 -            type != SYSLOG_ACTION_SIZE_BUFFER) && !capable(CAP_SYS_ADMIN))
26316 +            type != SYSLOG_ACTION_SIZE_BUFFER) &&
26317 +               !vx_capable(CAP_SYS_ADMIN, VXC_SYSLOG))
26318                 return -EPERM;
26319         return 0;
26320  }
26321 @@ -951,3 +969,4 @@ int cap_file_mmap(struct file *file, uns
26322         }
26323         return ret;
26324  }
26325 +
26326 diff -NurpP --minimal linux-2.6.36/security/selinux/hooks.c linux-2.6.36-vs2.3.0.36.38/security/selinux/hooks.c
26327 --- linux-2.6.36/security/selinux/hooks.c       2010-10-21 13:08:02.000000000 +0200
26328 +++ linux-2.6.36-vs2.3.0.36.38/security/selinux/hooks.c 2010-10-21 13:09:36.000000000 +0200
26329 @@ -64,7 +64,6 @@
26330  #include <linux/dccp.h>
26331  #include <linux/quota.h>
26332  #include <linux/un.h>          /* for Unix socket types */
26333 -#include <net/af_unix.h>       /* for Unix socket types */
26334  #include <linux/parser.h>
26335  #include <linux/nfs_mount.h>
26336  #include <net/ipv6.h>