kernel/sysctl.c

   1 /*
   2  * sysctl.c: General linux system control interface
   3  *
   4  * Begun 24 March 1995, Stephen Tweedie
   5  * Added /proc support, Dec 1995
   6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
   7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
   8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
   9  * Dynamic registration fixes, Stephen Tweedie.
  10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
  11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
  12  *  Horn.
  13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
  14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
  15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
  16  *  Wendling.
  17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
  18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
  19  */
  20
  21 #include <linux/module.h>
  22 #include <linux/mm.h>
  23 #include <linux/swap.h>
  24 #include <linux/slab.h>
  25 #include <linux/sysctl.h>
  26 #include <linux/bitmap.h>
  27 #include <linux/signal.h>
  28 #include <linux/printk.h>
  29 #include <linux/proc_fs.h>
  30 #include <linux/security.h>
  31 #include <linux/ctype.h>
  32 #include <linux/kmemcheck.h>
  33 #include <linux/fs.h>
  34 #include <linux/init.h>
  35 #include <linux/kernel.h>
  36 #include <linux/kobject.h>
  37 #include <linux/net.h>
  38 #include <linux/sysrq.h>
  39 #include <linux/highuid.h>
  40 #include <linux/writeback.h>
  41 #include <linux/ratelimit.h>
  42 #include <linux/compaction.h>
  43 #include <linux/hugetlb.h>
  44 #include <linux/initrd.h>
  45 #include <linux/key.h>
  46 #include <linux/times.h>
  47 #include <linux/limits.h>
  48 #include <linux/dcache.h>
  49 #include <linux/dnotify.h>
  50 #include <linux/syscalls.h>
  51 #include <linux/vmstat.h>
  52 #include <linux/nfs_fs.h>
  53 #include <linux/acpi.h>
  54 #include <linux/reboot.h>
  55 #include <linux/ftrace.h>
  56 #include <linux/perf_event.h>
  57 #include <linux/kprobes.h>
  58 #include <linux/pipe_fs_i.h>
  59 #include <linux/oom.h>
  60 #include <linux/kmod.h>
  61 #include <linux/capability.h>
  62 #include <linux/binfmts.h>
  63
  64 #include <asm/uaccess.h>
  65 #include <asm/processor.h>
  66
  67 #ifdef CONFIG_X86
  68 #include <asm/nmi.h>
  69 #include <asm/stacktrace.h>
  70 #include <asm/io.h>
  71 #endif
  72 #ifdef CONFIG_SPARC
  73 #include <asm/setup.h>
  74 #endif
  75 #ifdef CONFIG_BSD_PROCESS_ACCT
  76 #include <linux/acct.h>
  77 #endif
  78 #ifdef CONFIG_RT_MUTEXES
  79 #include <linux/rtmutex.h>
  80 #endif
  81 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
  82 #include <linux/lockdep.h>
  83 #endif
  84 #ifdef CONFIG_CHR_DEV_SG
  85 #include <scsi/sg.h>
  86 #endif
  87
  88 #ifdef CONFIG_LOCKUP_DETECTOR
  89 #include <linux/nmi.h>
  90 #endif
  91
  92
  93 #if defined(CONFIG_SYSCTL)
  94
  95 /* External variables not in a header file. */
  96 extern int sysctl_overcommit_memory;
  97 extern int sysctl_overcommit_ratio;
  98 extern int max_threads;
  99 extern int core_uses_pid;
 100 extern int suid_dumpable;
 101 extern char core_pattern[];
 102 extern unsigned int core_pipe_limit;
 103 extern int pid_max;
 104 extern int min_free_kbytes;
 105 extern int pid_max_min, pid_max_max;
 106 extern int sysctl_drop_caches;
 107 extern int percpu_pagelist_fraction;
 108 extern int compat_log;
 109 extern int latencytop_enabled;
 110 extern int sysctl_nr_open_min, sysctl_nr_open_max;
 111 #ifndef CONFIG_MMU
 112 extern int sysctl_nr_trim_pages;
 113 #endif
 114 #ifdef CONFIG_BLOCK
 115 extern int blk_iopoll_enabled;
 116 #endif
 117
 118 /* Constants used for minimum and  maximum */
 119 #ifdef CONFIG_LOCKUP_DETECTOR
 120 static int sixty = 60;
 121 static int neg_one = -1;
 122 #endif
 123
 124 static int zero;
 125 static int __maybe_unused one = 1;
 126 static int __maybe_unused two = 2;
 127 static int __maybe_unused three = 3;
 128 static unsigned long one_ul = 1;
 129 static int one_hundred = 100;
 130 #ifdef CONFIG_PRINTK
 131 static int ten_thousand = 10000;
 132 #endif
 133
 134 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 135 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 136
 137 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 138 static int maxolduid = 65535;
 139 static int minolduid;
 140 static int min_percpu_pagelist_fract = 8;
 141
 142 static int ngroups_max = NGROUPS_MAX;
 143 static const int cap_last_cap = CAP_LAST_CAP;
 144
 145 #ifdef CONFIG_INOTIFY_USER
 146 #include <linux/inotify.h>
 147 #endif
 148 #ifdef CONFIG_SPARC
 149 #endif
 150
 151 #ifdef CONFIG_SPARC64
 152 extern int sysctl_tsb_ratio;
 153 #endif
 154
 155 #ifdef __hppa__
 156 extern int pwrsw_enabled;
 157 extern int unaligned_enabled;
 158 #endif
 159
 160 #ifdef CONFIG_IA64
 161 extern int no_unaligned_warning;
 162 extern int unaligned_dump_stack;
 163 #endif
 164
 165 #ifdef CONFIG_PROC_SYSCTL
 166 static int proc_do_cad_pid(struct ctl_table *table, int write,
 167                   void __user *buffer, size_t *lenp, loff_t *ppos);
 168 static int proc_taint(struct ctl_table *table, int write,
 169                                void __user *buffer, size_t *lenp, loff_t *ppos);
 170 #endif
 171
 172 #ifdef CONFIG_PRINTK
 173 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
 174                                 void __user *buffer, size_t *lenp, loff_t *ppos);
 175 #endif
 176
 177 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
 178                 void __user *buffer, size_t *lenp, loff_t *ppos);
 179 static int proc_dostring_coredump(struct ctl_table *table, int write,
 180                 void __user *buffer, size_t *lenp, loff_t *ppos);
 181
 182 #ifdef CONFIG_MAGIC_SYSRQ
 183 /* Note: sysrq code uses it's own private copy */
 184 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
 185
 186 static int sysrq_sysctl_handler(ctl_table *table, int write,
 187                                 void __user *buffer, size_t *lenp,
 188                                 loff_t *ppos)
 189 {
 190         int error;
 191
 192         error = proc_dointvec(table, write, buffer, lenp, ppos);
 193         if (error)
 194                 return error;
 195
 196         if (write)
 197                 sysrq_toggle_support(__sysrq_enabled);
 198
 199         return 0;
 200 }
 201
 202 #endif
 203
 204 static struct ctl_table kern_table[];
 205 static struct ctl_table vm_table[];
 206 static struct ctl_table fs_table[];
 207 static struct ctl_table debug_table[];
 208 static struct ctl_table dev_table[];
 209 extern struct ctl_table random_table[];
 210 #ifdef CONFIG_EPOLL
 211 extern struct ctl_table epoll_table[];
 212 #endif
 213
 214 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 215 int sysctl_legacy_va_layout;
 216 #endif
 217
 218 /* The default sysctl tables: */
 219
 220 static struct ctl_table sysctl_base_table[] = {
 221         {
 222                 .procname       = "kernel",
 223                 .mode           = 0555,
 224                 .child          = kern_table,
 225         },
 226         {
 227                 .procname       = "vm",
 228                 .mode           = 0555,
 229                 .child          = vm_table,
 230         },
 231         {
 232                 .procname       = "fs",
 233                 .mode           = 0555,
 234                 .child          = fs_table,
 235         },
 236         {
 237                 .procname       = "debug",
 238                 .mode           = 0555,
 239                 .child          = debug_table,
 240         },
 241         {
 242                 .procname       = "dev",
 243                 .mode           = 0555,
 244                 .child          = dev_table,
 245         },
 246         { }
 247 };
 248
 249 #ifdef CONFIG_SCHED_DEBUG
 250 static int min_sched_granularity_ns = 100000;           /* 100 usecs */
 251 static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
 252 static int min_wakeup_granularity_ns;                   /* 0 usecs */
 253 static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
 254 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 255 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
 256 #endif
 257
 258 #ifdef CONFIG_COMPACTION
 259 static int min_extfrag_threshold;
 260 static int max_extfrag_threshold = 1000;
 261 #endif
 262
 263 static struct ctl_table kern_table[] = {
 264         {
 265                 .procname       = "sched_child_runs_first",
 266                 .data           = &sysctl_sched_child_runs_first,
 267                 .maxlen         = sizeof(unsigned int),
 268                 .mode           = 0644,
 269                 .proc_handler   = proc_dointvec,
 270         },
 271 #ifdef CONFIG_SCHED_DEBUG
 272         {
 273                 .procname       = "sched_min_granularity_ns",
 274                 .data           = &sysctl_sched_min_granularity,
 275                 .maxlen         = sizeof(unsigned int),
 276                 .mode           = 0644,
 277                 .proc_handler   = sched_proc_update_handler,
 278                 .extra1         = &min_sched_granularity_ns,
 279                 .extra2         = &max_sched_granularity_ns,
 280         },
 281         {
 282                 .procname       = "sched_latency_ns",
 283                 .data           = &sysctl_sched_latency,
 284                 .maxlen         = sizeof(unsigned int),
 285                 .mode           = 0644,
 286                 .proc_handler   = sched_proc_update_handler,
 287                 .extra1         = &min_sched_granularity_ns,
 288                 .extra2         = &max_sched_granularity_ns,
 289         },
 290         {
 291                 .procname       = "sched_wakeup_granularity_ns",
 292                 .data           = &sysctl_sched_wakeup_granularity,
 293                 .maxlen         = sizeof(unsigned int),
 294                 .mode           = 0644,
 295                 .proc_handler   = sched_proc_update_handler,
 296                 .extra1         = &min_wakeup_granularity_ns,
 297                 .extra2         = &max_wakeup_granularity_ns,
 298         },
 299         {
 300                 .procname       = "sched_tunable_scaling",
 301                 .data           = &sysctl_sched_tunable_scaling,
 302                 .maxlen         = sizeof(enum sched_tunable_scaling),
 303                 .mode           = 0644,
 304                 .proc_handler   = sched_proc_update_handler,
 305                 .extra1         = &min_sched_tunable_scaling,
 306                 .extra2         = &max_sched_tunable_scaling,
 307         },
 308         {
 309                 .procname       = "sched_migration_cost",
 310                 .data           = &sysctl_sched_migration_cost,
 311                 .maxlen         = sizeof(unsigned int),
 312                 .mode           = 0644,
 313                 .proc_handler   = proc_dointvec,
 314         },
 315         {
 316                 .procname       = "sched_nr_migrate",
 317                 .data           = &sysctl_sched_nr_migrate,
 318                 .maxlen         = sizeof(unsigned int),
 319                 .mode           = 0644,
 320                 .proc_handler   = proc_dointvec,
 321         },
 322         {
 323                 .procname       = "sched_time_avg",
 324                 .data           = &sysctl_sched_time_avg,
 325                 .maxlen         = sizeof(unsigned int),
 326                 .mode           = 0644,
 327                 .proc_handler   = proc_dointvec,
 328         },
 329         {
 330                 .procname       = "sched_shares_window",
 331                 .data           = &sysctl_sched_shares_window,
 332                 .maxlen         = sizeof(unsigned int),
 333                 .mode           = 0644,
 334                 .proc_handler   = proc_dointvec,
 335         },
 336         {
 337                 .procname       = "timer_migration",
 338                 .data           = &sysctl_timer_migration,
 339                 .maxlen         = sizeof(unsigned int),
 340                 .mode           = 0644,
 341                 .proc_handler   = proc_dointvec_minmax,
 342                 .extra1         = &zero,
 343                 .extra2         = &one,
 344         },
 345 #endif
 346         {
 347                 .procname       = "sched_rt_period_us",
 348                 .data           = &sysctl_sched_rt_period,
 349                 .maxlen         = sizeof(unsigned int),
 350                 .mode           = 0644,
 351                 .proc_handler   = sched_rt_handler,
 352         },
 353         {
 354                 .procname       = "sched_rt_runtime_us",
 355                 .data           = &sysctl_sched_rt_runtime,
 356                 .maxlen         = sizeof(int),
 357                 .mode           = 0644,
 358                 .proc_handler   = sched_rt_handler,
 359         },
 360 #ifdef CONFIG_SCHED_AUTOGROUP
 361         {
 362                 .procname       = "sched_autogroup_enabled",
 363                 .data           = &sysctl_sched_autogroup_enabled,
 364                 .maxlen         = sizeof(unsigned int),
 365                 .mode           = 0644,
 366                 .proc_handler   = proc_dointvec_minmax,
 367                 .extra1         = &zero,
 368                 .extra2         = &one,
 369         },
 370 #endif
 371 #ifdef CONFIG_CFS_BANDWIDTH
 372         {
 373                 .procname       = "sched_cfs_bandwidth_slice_us",
 374                 .data           = &sysctl_sched_cfs_bandwidth_slice,
 375                 .maxlen         = sizeof(unsigned int),
 376                 .mode           = 0644,
 377                 .proc_handler   = proc_dointvec_minmax,
 378                 .extra1         = &one,
 379         },
 380 #endif
 381 #ifdef CONFIG_PROVE_LOCKING
 382         {
 383                 .procname       = "prove_locking",
 384                 .data           = &prove_locking,
 385                 .maxlen         = sizeof(int),
 386                 .mode           = 0644,
 387                 .proc_handler   = proc_dointvec,
 388         },
 389 #endif
 390 #ifdef CONFIG_LOCK_STAT
 391         {
 392                 .procname       = "lock_stat",
 393                 .data           = &lock_stat,
 394                 .maxlen         = sizeof(int),
 395                 .mode           = 0644,
 396                 .proc_handler   = proc_dointvec,
 397         },
 398 #endif
 399         {
 400                 .procname       = "panic",
 401                 .data           = &panic_timeout,
 402                 .maxlen         = sizeof(int),
 403                 .mode           = 0644,
 404                 .proc_handler   = proc_dointvec,
 405         },
 406         {
 407                 .procname       = "core_uses_pid",
 408                 .data           = &core_uses_pid,
 409                 .maxlen         = sizeof(int),
 410                 .mode           = 0644,
 411                 .proc_handler   = proc_dointvec,
 412         },
 413         {
 414                 .procname       = "core_pattern",
 415                 .data           = core_pattern,
 416                 .maxlen         = CORENAME_MAX_SIZE,
 417                 .mode           = 0644,
 418                 .proc_handler   = proc_dostring_coredump,
 419         },
 420         {
 421                 .procname       = "core_pipe_limit",
 422                 .data           = &core_pipe_limit,
 423                 .maxlen         = sizeof(unsigned int),
 424                 .mode           = 0644,
 425                 .proc_handler   = proc_dointvec,
 426         },
 427 #ifdef CONFIG_PROC_SYSCTL
 428         {
 429                 .procname       = "tainted",
 430                 .maxlen         = sizeof(long),
 431                 .mode           = 0644,
 432                 .proc_handler   = proc_taint,
 433         },
 434 #endif
 435 #ifdef CONFIG_LATENCYTOP
 436         {
 437                 .procname       = "latencytop",
 438                 .data           = &latencytop_enabled,
 439                 .maxlen         = sizeof(int),
 440                 .mode           = 0644,
 441                 .proc_handler   = proc_dointvec,
 442         },
 443 #endif
 444 #ifdef CONFIG_BLK_DEV_INITRD
 445         {
 446                 .procname       = "real-root-dev",
 447                 .data           = &real_root_dev,
 448                 .maxlen         = sizeof(int),
 449                 .mode           = 0644,
 450                 .proc_handler   = proc_dointvec,
 451         },
 452 #endif
 453         {
 454                 .procname       = "print-fatal-signals",
 455                 .data           = &print_fatal_signals,
 456                 .maxlen         = sizeof(int),
 457                 .mode           = 0644,
 458                 .proc_handler   = proc_dointvec,
 459         },
 460 #ifdef CONFIG_SPARC
 461         {
 462                 .procname       = "reboot-cmd",
 463                 .data           = reboot_command,
 464                 .maxlen         = 256,
 465                 .mode           = 0644,
 466                 .proc_handler   = proc_dostring,
 467         },
 468         {
 469                 .procname       = "stop-a",
 470                 .data           = &stop_a_enabled,
 471                 .maxlen         = sizeof (int),
 472                 .mode           = 0644,
 473                 .proc_handler   = proc_dointvec,
 474         },
 475         {
 476                 .procname       = "scons-poweroff",
 477                 .data           = &scons_pwroff,
 478                 .maxlen         = sizeof (int),
 479                 .mode           = 0644,
 480                 .proc_handler   = proc_dointvec,
 481         },
 482 #endif
 483 #ifdef CONFIG_SPARC64
 484         {
 485                 .procname       = "tsb-ratio",
 486                 .data           = &sysctl_tsb_ratio,
 487                 .maxlen         = sizeof (int),
 488                 .mode           = 0644,
 489                 .proc_handler   = proc_dointvec,
 490         },
 491 #endif
 492 #ifdef __hppa__
 493         {
 494                 .procname       = "soft-power",
 495                 .data           = &pwrsw_enabled,
 496                 .maxlen         = sizeof (int),
 497                 .mode           = 0644,
 498                 .proc_handler   = proc_dointvec,
 499         },
 500         {
 501                 .procname       = "unaligned-trap",
 502                 .data           = &unaligned_enabled,
 503                 .maxlen         = sizeof (int),
 504                 .mode           = 0644,
 505                 .proc_handler   = proc_dointvec,
 506         },
 507 #endif
 508         {
 509                 .procname       = "ctrl-alt-del",
 510                 .data           = &C_A_D,
 511                 .maxlen         = sizeof(int),
 512                 .mode           = 0644,
 513                 .proc_handler   = proc_dointvec,
 514         },
 515 #ifdef CONFIG_FUNCTION_TRACER
 516         {
 517                 .procname       = "ftrace_enabled",
 518                 .data           = &ftrace_enabled,
 519                 .maxlen         = sizeof(int),
 520                 .mode           = 0644,
 521                 .proc_handler   = ftrace_enable_sysctl,
 522         },
 523 #endif
 524 #ifdef CONFIG_STACK_TRACER
 525         {
 526                 .procname       = "stack_tracer_enabled",
 527                 .data           = &stack_tracer_enabled,
 528                 .maxlen         = sizeof(int),
 529                 .mode           = 0644,
 530                 .proc_handler   = stack_trace_sysctl,
 531         },
 532 #endif
 533 #ifdef CONFIG_TRACING
 534         {
 535                 .procname       = "ftrace_dump_on_oops",
 536                 .data           = &ftrace_dump_on_oops,
 537                 .maxlen         = sizeof(int),
 538                 .mode           = 0644,
 539                 .proc_handler   = proc_dointvec,
 540         },
 541 #endif
 542 #ifdef CONFIG_MODULES
 543         {
 544                 .procname       = "modprobe",
 545                 .data           = &modprobe_path,
 546                 .maxlen         = KMOD_PATH_LEN,
 547                 .mode           = 0644,
 548                 .proc_handler   = proc_dostring,
 549         },
 550         {
 551                 .procname       = "modules_disabled",
 552                 .data           = &modules_disabled,
 553                 .maxlen         = sizeof(int),
 554                 .mode           = 0644,
 555                 /* only handle a transition from default "0" to "1" */
 556                 .proc_handler   = proc_dointvec_minmax,
 557                 .extra1         = &one,
 558                 .extra2         = &one,
 559         },
 560 #endif
 561 #ifdef CONFIG_HOTPLUG
 562         {
 563                 .procname       = "hotplug",
 564                 .data           = &uevent_helper,
 565                 .maxlen         = UEVENT_HELPER_PATH_LEN,
 566                 .mode           = 0644,
 567                 .proc_handler   = proc_dostring,
 568         },
 569 #endif
 570 #ifdef CONFIG_CHR_DEV_SG
 571         {
 572                 .procname       = "sg-big-buff",
 573                 .data           = &sg_big_buff,
 574                 .maxlen         = sizeof (int),
 575                 .mode           = 0444,
 576                 .proc_handler   = proc_dointvec,
 577         },
 578 #endif
 579 #ifdef CONFIG_BSD_PROCESS_ACCT
 580         {
 581                 .procname       = "acct",
 582                 .data           = &acct_parm,
 583                 .maxlen         = 3*sizeof(int),
 584                 .mode           = 0644,
 585                 .proc_handler   = proc_dointvec,
 586         },
 587 #endif
 588 #ifdef CONFIG_MAGIC_SYSRQ
 589         {
 590                 .procname       = "sysrq",
 591                 .data           = &__sysrq_enabled,
 592                 .maxlen         = sizeof (int),
 593                 .mode           = 0644,
 594                 .proc_handler   = sysrq_sysctl_handler,
 595         },
 596 #endif
 597 #ifdef CONFIG_PROC_SYSCTL
 598         {
 599                 .procname       = "cad_pid",
 600                 .data           = NULL,
 601                 .maxlen         = sizeof (int),
 602                 .mode           = 0600,
 603                 .proc_handler   = proc_do_cad_pid,
 604         },
 605 #endif
 606         {
 607                 .procname       = "threads-max",
 608                 .data           = &max_threads,
 609                 .maxlen         = sizeof(int),
 610                 .mode           = 0644,
 611                 .proc_handler   = proc_dointvec,
 612         },
 613         {
 614                 .procname       = "random",
 615                 .mode           = 0555,
 616                 .child          = random_table,
 617         },
 618         {
 619                 .procname       = "usermodehelper",
 620                 .mode           = 0555,
 621                 .child          = usermodehelper_table,
 622         },
 623         {
 624                 .procname       = "overflowuid",
 625                 .data           = &overflowuid,
 626                 .maxlen         = sizeof(int),
 627                 .mode           = 0644,
 628                 .proc_handler   = proc_dointvec_minmax,
 629                 .extra1         = &minolduid,
 630                 .extra2         = &maxolduid,
 631         },
 632         {
 633                 .procname       = "overflowgid",
 634                 .data           = &overflowgid,
 635                 .maxlen         = sizeof(int),
 636                 .mode           = 0644,
 637                 .proc_handler   = proc_dointvec_minmax,
 638                 .extra1         = &minolduid,
 639                 .extra2         = &maxolduid,
 640         },
 641 #ifdef CONFIG_S390
 642 #ifdef CONFIG_MATHEMU
 643         {
 644                 .procname       = "ieee_emulation_warnings",
 645                 .data           = &sysctl_ieee_emulation_warnings,
 646                 .maxlen         = sizeof(int),
 647                 .mode           = 0644,
 648                 .proc_handler   = proc_dointvec,
 649         },
 650 #endif
 651         {
 652                 .procname       = "userprocess_debug",
 653                 .data           = &show_unhandled_signals,
 654                 .maxlen         = sizeof(int),
 655                 .mode           = 0644,
 656                 .proc_handler   = proc_dointvec,
 657         },
 658 #endif
 659         {
 660                 .procname       = "pid_max",
 661                 .data           = &pid_max,
 662                 .maxlen         = sizeof (int),
 663                 .mode           = 0644,
 664                 .proc_handler   = proc_dointvec_minmax,
 665                 .extra1         = &pid_max_min,
 666                 .extra2         = &pid_max_max,
 667         },
 668         {
 669                 .procname       = "panic_on_oops",
 670                 .data           = &panic_on_oops,
 671                 .maxlen         = sizeof(int),
 672                 .mode           = 0644,
 673                 .proc_handler   = proc_dointvec,
 674         },
 675 #if defined CONFIG_PRINTK
 676         {
 677                 .procname       = "printk",
 678                 .data           = &console_loglevel,
 679                 .maxlen         = 4*sizeof(int),
 680                 .mode           = 0644,
 681                 .proc_handler   = proc_dointvec,
 682         },
 683         {
 684                 .procname       = "printk_ratelimit",
 685                 .data           = &printk_ratelimit_state.interval,
 686                 .maxlen         = sizeof(int),
 687                 .mode           = 0644,
 688                 .proc_handler   = proc_dointvec_jiffies,
 689         },
 690         {
 691                 .procname       = "printk_ratelimit_burst",
 692                 .data           = &printk_ratelimit_state.burst,
 693                 .maxlen         = sizeof(int),
 694                 .mode           = 0644,
 695                 .proc_handler   = proc_dointvec,
 696         },
 697         {
 698                 .procname       = "printk_delay",
 699                 .data           = &printk_delay_msec,
 700                 .maxlen         = sizeof(int),
 701                 .mode           = 0644,
 702                 .proc_handler   = proc_dointvec_minmax,
 703                 .extra1         = &zero,
 704                 .extra2         = &ten_thousand,
 705         },
 706         {
 707                 .procname       = "dmesg_restrict",
 708                 .data           = &dmesg_restrict,
 709                 .maxlen         = sizeof(int),
 710                 .mode           = 0644,
 711                 .proc_handler   = proc_dointvec_minmax_sysadmin,
 712                 .extra1         = &zero,
 713                 .extra2         = &one,
 714         },
 715         {
 716                 .procname       = "kptr_restrict",
 717                 .data           = &kptr_restrict,
 718                 .maxlen         = sizeof(int),
 719                 .mode           = 0644,
 720                 .proc_handler   = proc_dointvec_minmax_sysadmin,
 721                 .extra1         = &zero,
 722                 .extra2         = &two,
 723         },
 724 #endif
 725         {
 726                 .procname       = "ngroups_max",
 727                 .data           = &ngroups_max,
 728                 .maxlen         = sizeof (int),
 729                 .mode           = 0444,
 730                 .proc_handler   = proc_dointvec,
 731         },
 732         {
 733                 .procname       = "cap_last_cap",
 734                 .data           = (void *)&cap_last_cap,
 735                 .maxlen         = sizeof(int),
 736                 .mode           = 0444,
 737                 .proc_handler   = proc_dointvec,
 738         },
 739 #if defined(CONFIG_LOCKUP_DETECTOR)
 740         {
 741                 .procname       = "watchdog",
 742                 .data           = &watchdog_enabled,
 743                 .maxlen         = sizeof (int),
 744                 .mode           = 0644,
 745                 .proc_handler   = proc_dowatchdog,
 746                 .extra1         = &zero,
 747                 .extra2         = &one,
 748         },
 749         {
 750                 .procname       = "watchdog_thresh",
 751                 .data           = &watchdog_thresh,
 752                 .maxlen         = sizeof(int),
 753                 .mode           = 0644,
 754                 .proc_handler   = proc_dowatchdog,
 755                 .extra1         = &neg_one,
 756                 .extra2         = &sixty,
 757         },
 758         {
 759                 .procname       = "softlockup_panic",
 760                 .data           = &softlockup_panic,
 761                 .maxlen         = sizeof(int),
 762                 .mode           = 0644,
 763                 .proc_handler   = proc_dointvec_minmax,
 764                 .extra1         = &zero,
 765                 .extra2         = &one,
 766         },
 767         {
 768                 .procname       = "nmi_watchdog",
 769                 .data           = &watchdog_enabled,
 770                 .maxlen         = sizeof (int),
 771                 .mode           = 0644,
 772                 .proc_handler   = proc_dowatchdog,
 773                 .extra1         = &zero,
 774                 .extra2         = &one,
 775         },
 776 #endif
 777 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 778         {
 779                 .procname       = "unknown_nmi_panic",
 780                 .data           = &unknown_nmi_panic,
 781                 .maxlen         = sizeof (int),
 782                 .mode           = 0644,
 783                 .proc_handler   = proc_dointvec,
 784         },
 785 #endif
 786 #if defined(CONFIG_X86)
 787         {
 788                 .procname       = "panic_on_unrecovered_nmi",
 789                 .data           = &panic_on_unrecovered_nmi,
 790                 .maxlen         = sizeof(int),
 791                 .mode           = 0644,
 792                 .proc_handler   = proc_dointvec,
 793         },
 794         {
 795                 .procname       = "panic_on_io_nmi",
 796                 .data           = &panic_on_io_nmi,
 797                 .maxlen         = sizeof(int),
 798                 .mode           = 0644,
 799                 .proc_handler   = proc_dointvec,
 800         },
 801 #ifdef CONFIG_DEBUG_STACKOVERFLOW
 802         {
 803                 .procname       = "panic_on_stackoverflow",
 804                 .data           = &sysctl_panic_on_stackoverflow,
 805                 .maxlen         = sizeof(int),
 806                 .mode           = 0644,
 807                 .proc_handler   = proc_dointvec,
 808         },
 809 #endif
 810         {
 811                 .procname       = "bootloader_type",
 812                 .data           = &bootloader_type,
 813                 .maxlen         = sizeof (int),
 814                 .mode           = 0444,
 815                 .proc_handler   = proc_dointvec,
 816         },
 817         {
 818                 .procname       = "bootloader_version",
 819                 .data           = &bootloader_version,
 820                 .maxlen         = sizeof (int),
 821                 .mode           = 0444,
 822                 .proc_handler   = proc_dointvec,
 823         },
 824         {
 825                 .procname       = "kstack_depth_to_print",
 826                 .data           = &kstack_depth_to_print,
 827                 .maxlen         = sizeof(int),
 828                 .mode           = 0644,
 829                 .proc_handler   = proc_dointvec,
 830         },
 831         {
 832                 .procname       = "io_delay_type",
 833                 .data           = &io_delay_type,
 834                 .maxlen         = sizeof(int),
 835                 .mode           = 0644,
 836                 .proc_handler   = proc_dointvec,
 837         },
 838 #endif
 839 #if defined(CONFIG_MMU)
 840         {
 841                 .procname       = "randomize_va_space",
 842                 .data           = &randomize_va_space,
 843                 .maxlen         = sizeof(int),
 844                 .mode           = 0644,
 845                 .proc_handler   = proc_dointvec,
 846         },
 847 #endif
 848 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
 849         {
 850                 .procname       = "spin_retry",
 851                 .data           = &spin_retry,
 852                 .maxlen         = sizeof (int),
 853                 .mode           = 0644,
 854                 .proc_handler   = proc_dointvec,
 855         },
 856 #endif
 857 #if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
 858         {
 859                 .procname       = "acpi_video_flags",
 860                 .data           = &acpi_realmode_flags,
 861                 .maxlen         = sizeof (unsigned long),
 862                 .mode           = 0644,
 863                 .proc_handler   = proc_doulongvec_minmax,
 864         },
 865 #endif
 866 #ifdef CONFIG_IA64
 867         {
 868                 .procname       = "ignore-unaligned-usertrap",
 869                 .data           = &no_unaligned_warning,
 870                 .maxlen         = sizeof (int),
 871                 .mode           = 0644,
 872                 .proc_handler   = proc_dointvec,
 873         },
 874         {
 875                 .procname       = "unaligned-dump-stack",
 876                 .data           = &unaligned_dump_stack,
 877                 .maxlen         = sizeof (int),
 878                 .mode           = 0644,
 879                 .proc_handler   = proc_dointvec,
 880         },
 881 #endif
 882 #ifdef CONFIG_DETECT_HUNG_TASK
 883         {
 884                 .procname       = "hung_task_panic",
 885                 .data           = &sysctl_hung_task_panic,
 886                 .maxlen         = sizeof(int),
 887                 .mode           = 0644,
 888                 .proc_handler   = proc_dointvec_minmax,
 889                 .extra1         = &zero,
 890                 .extra2         = &one,
 891         },
 892         {
 893                 .procname       = "hung_task_check_count",
 894                 .data           = &sysctl_hung_task_check_count,
 895                 .maxlen         = sizeof(unsigned long),
 896                 .mode           = 0644,
 897                 .proc_handler   = proc_doulongvec_minmax,
 898         },
 899         {
 900                 .procname       = "hung_task_timeout_secs",
 901                 .data           = &sysctl_hung_task_timeout_secs,
 902                 .maxlen         = sizeof(unsigned long),
 903                 .mode           = 0644,
 904                 .proc_handler   = proc_dohung_task_timeout_secs,
 905         },
 906         {
 907                 .procname       = "hung_task_warnings",
 908                 .data           = &sysctl_hung_task_warnings,
 909                 .maxlen         = sizeof(unsigned long),
 910                 .mode           = 0644,
 911                 .proc_handler   = proc_doulongvec_minmax,
 912         },
 913 #endif
 914 #ifdef CONFIG_COMPAT
 915         {
 916                 .procname       = "compat-log",
 917                 .data           = &compat_log,
 918                 .maxlen         = sizeof (int),
 919                 .mode           = 0644,
 920                 .proc_handler   = proc_dointvec,
 921         },
 922 #endif
 923 #ifdef CONFIG_RT_MUTEXES
 924         {
 925                 .procname       = "max_lock_depth",
 926                 .data           = &max_lock_depth,
 927                 .maxlen         = sizeof(int),
 928                 .mode           = 0644,
 929                 .proc_handler   = proc_dointvec,
 930         },
 931 #endif
 932         {
 933                 .procname       = "poweroff_cmd",
 934                 .data           = &poweroff_cmd,
 935                 .maxlen         = POWEROFF_CMD_PATH_LEN,
 936                 .mode           = 0644,
 937                 .proc_handler   = proc_dostring,
 938         },
 939 #ifdef CONFIG_KEYS
 940         {
 941                 .procname       = "keys",
 942                 .mode           = 0555,
 943                 .child          = key_sysctls,
 944         },
 945 #endif
 946 #ifdef CONFIG_RCU_TORTURE_TEST
 947         {
 948                 .procname       = "rcutorture_runnable",
 949                 .data           = &rcutorture_runnable,
 950                 .maxlen         = sizeof(int),
 951                 .mode           = 0644,
 952                 .proc_handler   = proc_dointvec,
 953         },
 954 #endif
 955 #ifdef CONFIG_PERF_EVENTS
 956         /*
 957          * User-space scripts rely on the existence of this file
 958          * as a feature check for perf_events being enabled.
 959          *
 960          * So it's an ABI, do not remove!
 961          */
 962         {
 963                 .procname       = "perf_event_paranoid",
 964                 .data           = &sysctl_perf_event_paranoid,
 965                 .maxlen         = sizeof(sysctl_perf_event_paranoid),
 966                 .mode           = 0644,
 967                 .proc_handler   = proc_dointvec,
 968         },
 969         {
 970                 .procname       = "perf_event_mlock_kb",
 971                 .data           = &sysctl_perf_event_mlock,
 972                 .maxlen         = sizeof(sysctl_perf_event_mlock),
 973                 .mode           = 0644,
 974                 .proc_handler   = proc_dointvec,
 975         },
 976         {
 977                 .procname       = "perf_event_max_sample_rate",
 978                 .data           = &sysctl_perf_event_sample_rate,
 979                 .maxlen         = sizeof(sysctl_perf_event_sample_rate),
 980                 .mode           = 0644,
 981                 .proc_handler   = perf_proc_update_handler,
 982         },
 983 #endif
 984 #ifdef CONFIG_KMEMCHECK
 985         {
 986                 .procname       = "kmemcheck",
 987                 .data           = &kmemcheck_enabled,
 988                 .maxlen         = sizeof(int),
 989                 .mode           = 0644,
 990                 .proc_handler   = proc_dointvec,
 991         },
 992 #endif
 993 #ifdef CONFIG_BLOCK
 994         {
 995                 .procname       = "blk_iopoll",
 996                 .data           = &blk_iopoll_enabled,
 997                 .maxlen         = sizeof(int),
 998                 .mode           = 0644,
 999                 .proc_handler   = proc_dointvec,
1000         },
1001 #endif
1002         { }
1003 };
1004
1005 static struct ctl_table vm_table[] = {
1006         {
1007                 .procname       = "overcommit_memory",
1008                 .data           = &sysctl_overcommit_memory,
1009                 .maxlen         = sizeof(sysctl_overcommit_memory),
1010                 .mode           = 0644,
1011                 .proc_handler   = proc_dointvec_minmax,
1012                 .extra1         = &zero,
1013                 .extra2         = &two,
1014         },
1015         {
1016                 .procname       = "panic_on_oom",
1017                 .data           = &sysctl_panic_on_oom,
1018                 .maxlen         = sizeof(sysctl_panic_on_oom),
1019                 .mode           = 0644,
1020                 .proc_handler   = proc_dointvec_minmax,
1021                 .extra1         = &zero,
1022                 .extra2         = &two,
1023         },
1024         {
1025                 .procname       = "oom_kill_allocating_task",
1026                 .data           = &sysctl_oom_kill_allocating_task,
1027                 .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
1028                 .mode           = 0644,
1029                 .proc_handler   = proc_dointvec,
1030         },
1031         {
1032                 .procname       = "oom_dump_tasks",
1033                 .data           = &sysctl_oom_dump_tasks,
1034                 .maxlen         = sizeof(sysctl_oom_dump_tasks),
1035                 .mode           = 0644,
1036                 .proc_handler   = proc_dointvec,
1037         },
1038         {
1039                 .procname       = "overcommit_ratio",
1040                 .data           = &sysctl_overcommit_ratio,
1041                 .maxlen         = sizeof(sysctl_overcommit_ratio),
1042                 .mode           = 0644,
1043                 .proc_handler   = proc_dointvec,
1044         },
1045         {
1046                 .procname       = "page-cluster",
1047                 .data           = &page_cluster,
1048                 .maxlen         = sizeof(int),
1049                 .mode           = 0644,
1050                 .proc_handler   = proc_dointvec_minmax,
1051                 .extra1         = &zero,
1052         },
1053         {
1054                 .procname       = "dirty_background_ratio",
1055                 .data           = &dirty_background_ratio,
1056                 .maxlen         = sizeof(dirty_background_ratio),
1057                 .mode           = 0644,
1058                 .proc_handler   = dirty_background_ratio_handler,
1059                 .extra1         = &zero,
1060                 .extra2         = &one_hundred,
1061         },
1062         {
1063                 .procname       = "dirty_background_bytes",
1064                 .data           = &dirty_background_bytes,
1065                 .maxlen         = sizeof(dirty_background_bytes),
1066                 .mode           = 0644,
1067                 .proc_handler   = dirty_background_bytes_handler,
1068                 .extra1         = &one_ul,
1069         },
1070         {
1071                 .procname       = "dirty_ratio",
1072                 .data           = &vm_dirty_ratio,
1073                 .maxlen         = sizeof(vm_dirty_ratio),
1074                 .mode           = 0644,
1075                 .proc_handler   = dirty_ratio_handler,
1076                 .extra1         = &zero,
1077                 .extra2         = &one_hundred,
1078         },
1079         {
1080                 .procname       = "dirty_bytes",
1081                 .data           = &vm_dirty_bytes,
1082                 .maxlen         = sizeof(vm_dirty_bytes),
1083                 .mode           = 0644,
1084                 .proc_handler   = dirty_bytes_handler,
1085                 .extra1         = &dirty_bytes_min,
1086         },
1087         {
1088                 .procname       = "dirty_writeback_centisecs",
1089                 .data           = &dirty_writeback_interval,
1090                 .maxlen         = sizeof(dirty_writeback_interval),
1091                 .mode           = 0644,
1092                 .proc_handler   = dirty_writeback_centisecs_handler,
1093         },
1094         {
1095                 .procname       = "dirty_expire_centisecs",
1096                 .data           = &dirty_expire_interval,
1097                 .maxlen         = sizeof(dirty_expire_interval),
1098                 .mode           = 0644,
1099                 .proc_handler   = proc_dointvec_minmax,
1100                 .extra1         = &zero,
1101         },
1102         {
1103                 .procname       = "nr_pdflush_threads",
1104                 .data           = &nr_pdflush_threads,
1105                 .maxlen         = sizeof nr_pdflush_threads,
1106                 .mode           = 0444 /* read-only*/,
1107                 .proc_handler   = proc_dointvec,
1108         },
1109         {
1110                 .procname       = "swappiness",
1111                 .data           = &vm_swappiness,
1112                 .maxlen         = sizeof(vm_swappiness),
1113                 .mode           = 0644,
1114                 .proc_handler   = proc_dointvec_minmax,
1115                 .extra1         = &zero,
1116                 .extra2         = &one_hundred,
1117         },
1118 #ifdef CONFIG_HUGETLB_PAGE
1119         {
1120                 .procname       = "nr_hugepages",
1121                 .data           = NULL,
1122                 .maxlen         = sizeof(unsigned long),
1123                 .mode           = 0644,
1124                 .proc_handler   = hugetlb_sysctl_handler,
1125                 .extra1         = (void *)&hugetlb_zero,
1126                 .extra2         = (void *)&hugetlb_infinity,
1127         },
1128 #ifdef CONFIG_NUMA
1129         {
1130                 .procname       = "nr_hugepages_mempolicy",
1131                 .data           = NULL,
1132                 .maxlen         = sizeof(unsigned long),
1133                 .mode           = 0644,
1134                 .proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1135                 .extra1         = (void *)&hugetlb_zero,
1136                 .extra2         = (void *)&hugetlb_infinity,
1137         },
1138 #endif
1139          {
1140                 .procname       = "hugetlb_shm_group",
1141                 .data           = &sysctl_hugetlb_shm_group,
1142                 .maxlen         = sizeof(gid_t),
1143                 .mode           = 0644,
1144                 .proc_handler   = proc_dointvec,
1145          },
1146          {
1147                 .procname       = "hugepages_treat_as_movable",
1148                 .data           = &hugepages_treat_as_movable,
1149                 .maxlen         = sizeof(int),
1150                 .mode           = 0644,
1151                 .proc_handler   = hugetlb_treat_movable_handler,
1152         },
1153         {
1154                 .procname       = "nr_overcommit_hugepages",
1155                 .data           = NULL,
1156                 .maxlen         = sizeof(unsigned long),
1157                 .mode           = 0644,
1158                 .proc_handler   = hugetlb_overcommit_handler,
1159                 .extra1         = (void *)&hugetlb_zero,
1160                 .extra2         = (void *)&hugetlb_infinity,
1161         },
1162 #endif
1163         {
1164                 .procname       = "lowmem_reserve_ratio",
1165                 .data           = &sysctl_lowmem_reserve_ratio,
1166                 .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1167                 .mode           = 0644,
1168                 .proc_handler   = lowmem_reserve_ratio_sysctl_handler,
1169         },
1170         {
1171                 .procname       = "drop_caches",
1172                 .data           = &sysctl_drop_caches,
1173                 .maxlen         = sizeof(int),
1174                 .mode           = 0644,
1175                 .proc_handler   = drop_caches_sysctl_handler,
1176                 .extra1         = &one,
1177                 .extra2         = &three,
1178         },
1179 #ifdef CONFIG_COMPACTION
1180         {
1181                 .procname       = "compact_memory",
1182                 .data           = &sysctl_compact_memory,
1183                 .maxlen         = sizeof(int),
1184                 .mode           = 0200,
1185                 .proc_handler   = sysctl_compaction_handler,
1186         },
1187         {
1188                 .procname       = "extfrag_threshold",
1189                 .data           = &sysctl_extfrag_threshold,
1190                 .maxlen         = sizeof(int),
1191                 .mode           = 0644,
1192                 .proc_handler   = sysctl_extfrag_handler,
1193                 .extra1         = &min_extfrag_threshold,
1194                 .extra2         = &max_extfrag_threshold,
1195         },
1196
1197 #endif /* CONFIG_COMPACTION */
1198         {
1199                 .procname       = "min_free_kbytes",
1200                 .data           = &min_free_kbytes,
1201                 .maxlen         = sizeof(min_free_kbytes),
1202                 .mode           = 0644,
1203                 .proc_handler   = min_free_kbytes_sysctl_handler,
1204                 .extra1         = &zero,
1205         },
1206         {
1207                 .procname       = "percpu_pagelist_fraction",
1208                 .data           = &percpu_pagelist_fraction,
1209                 .maxlen         = sizeof(percpu_pagelist_fraction),
1210                 .mode           = 0644,
1211                 .proc_handler   = percpu_pagelist_fraction_sysctl_handler,
1212                 .extra1         = &min_percpu_pagelist_fract,
1213         },
1214 #ifdef CONFIG_MMU
1215         {
1216                 .procname       = "max_map_count",
1217                 .data           = &sysctl_max_map_count,
1218                 .maxlen         = sizeof(sysctl_max_map_count),
1219                 .mode           = 0644,
1220                 .proc_handler   = proc_dointvec_minmax,
1221                 .extra1         = &zero,
1222         },
1223 #else
1224         {
1225                 .procname       = "nr_trim_pages",
1226                 .data           = &sysctl_nr_trim_pages,
1227                 .maxlen         = sizeof(sysctl_nr_trim_pages),
1228                 .mode           = 0644,
1229                 .proc_handler   = proc_dointvec_minmax,
1230                 .extra1         = &zero,
1231         },
1232 #endif
1233         {
1234                 .procname       = "laptop_mode",
1235                 .data           = &laptop_mode,
1236                 .maxlen         = sizeof(laptop_mode),
1237                 .mode           = 0644,
1238                 .proc_handler   = proc_dointvec_jiffies,
1239         },
1240         {
1241                 .procname       = "block_dump",
1242                 .data           = &block_dump,
1243                 .maxlen         = sizeof(block_dump),
1244                 .mode           = 0644,
1245                 .proc_handler   = proc_dointvec,
1246                 .extra1         = &zero,
1247         },
1248         {
1249                 .procname       = "vfs_cache_pressure",
1250                 .data           = &sysctl_vfs_cache_pressure,
1251                 .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1252                 .mode           = 0644,
1253                 .proc_handler   = proc_dointvec,
1254                 .extra1         = &zero,
1255         },
1256 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1257         {
1258                 .procname       = "legacy_va_layout",
1259                 .data           = &sysctl_legacy_va_layout,
1260                 .maxlen         = sizeof(sysctl_legacy_va_layout),
1261                 .mode           = 0644,
1262                 .proc_handler   = proc_dointvec,
1263                 .extra1         = &zero,
1264         },
1265 #endif
1266 #ifdef CONFIG_NUMA
1267         {
1268                 .procname       = "zone_reclaim_mode",
1269                 .data           = &zone_reclaim_mode,
1270                 .maxlen         = sizeof(zone_reclaim_mode),
1271                 .mode           = 0644,
1272                 .proc_handler   = proc_dointvec,
1273                 .extra1         = &zero,
1274         },
1275         {
1276                 .procname       = "min_unmapped_ratio",
1277                 .data           = &sysctl_min_unmapped_ratio,
1278                 .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1279                 .mode           = 0644,
1280                 .proc_handler   = sysctl_min_unmapped_ratio_sysctl_handler,
1281                 .extra1         = &zero,
1282                 .extra2         = &one_hundred,
1283         },
1284         {
1285                 .procname       = "min_slab_ratio",
1286                 .data           = &sysctl_min_slab_ratio,
1287                 .maxlen         = sizeof(sysctl_min_slab_ratio),
1288                 .mode           = 0644,
1289                 .proc_handler   = sysctl_min_slab_ratio_sysctl_handler,
1290                 .extra1         = &zero,
1291                 .extra2         = &one_hundred,
1292         },
1293 #endif
1294 #ifdef CONFIG_SMP
1295         {
1296                 .procname       = "stat_interval",
1297                 .data           = &sysctl_stat_interval,
1298                 .maxlen         = sizeof(sysctl_stat_interval),
1299                 .mode           = 0644,
1300                 .proc_handler   = proc_dointvec_jiffies,
1301         },
1302 #endif
1303 #ifdef CONFIG_MMU
1304         {
1305                 .procname       = "mmap_min_addr",
1306                 .data           = &dac_mmap_min_addr,
1307                 .maxlen         = sizeof(unsigned long),
1308                 .mode           = 0644,
1309                 .proc_handler   = mmap_min_addr_handler,
1310         },
1311 #endif
1312 #ifdef CONFIG_NUMA
1313         {
1314                 .procname       = "numa_zonelist_order",
1315                 .data           = &numa_zonelist_order,
1316                 .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1317                 .mode           = 0644,
1318                 .proc_handler   = numa_zonelist_order_handler,
1319         },
1320 #endif
1321 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1322    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1323         {
1324                 .procname       = "vdso_enabled",
1325                 .data           = &vdso_enabled,
1326                 .maxlen         = sizeof(vdso_enabled),
1327                 .mode           = 0644,
1328                 .proc_handler   = proc_dointvec,
1329                 .extra1         = &zero,
1330         },
1331 #endif
1332 #ifdef CONFIG_HIGHMEM
1333         {
1334                 .procname       = "highmem_is_dirtyable",
1335                 .data           = &vm_highmem_is_dirtyable,
1336                 .maxlen         = sizeof(vm_highmem_is_dirtyable),
1337                 .mode           = 0644,
1338                 .proc_handler   = proc_dointvec_minmax,
1339                 .extra1         = &zero,
1340                 .extra2         = &one,
1341         },
1342 #endif
1343         {
1344                 .procname       = "scan_unevictable_pages",
1345                 .data           = &scan_unevictable_pages,
1346                 .maxlen         = sizeof(scan_unevictable_pages),
1347                 .mode           = 0644,
1348                 .proc_handler   = scan_unevictable_handler,
1349         },
1350 #ifdef CONFIG_MEMORY_FAILURE
1351         {
1352                 .procname       = "memory_failure_early_kill",
1353                 .data           = &sysctl_memory_failure_early_kill,
1354                 .maxlen         = sizeof(sysctl_memory_failure_early_kill),
1355                 .mode           = 0644,
1356                 .proc_handler   = proc_dointvec_minmax,
1357                 .extra1         = &zero,
1358                 .extra2         = &one,
1359         },
1360         {
1361                 .procname       = "memory_failure_recovery",
1362                 .data           = &sysctl_memory_failure_recovery,
1363                 .maxlen         = sizeof(sysctl_memory_failure_recovery),
1364                 .mode           = 0644,
1365                 .proc_handler   = proc_dointvec_minmax,
1366                 .extra1         = &zero,
1367                 .extra2         = &one,
1368         },
1369 #endif
1370         { }
1371 };
1372
1373 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1374 static struct ctl_table binfmt_misc_table[] = {
1375         { }
1376 };
1377 #endif
1378
1379 static struct ctl_table fs_table[] = {
1380         {
1381                 .procname       = "inode-nr",
1382                 .data           = &inodes_stat,
1383                 .maxlen         = 2*sizeof(int),
1384                 .mode           = 0444,
1385                 .proc_handler   = proc_nr_inodes,
1386         },
1387         {
1388                 .procname       = "inode-state",
1389                 .data           = &inodes_stat,
1390                 .maxlen         = 7*sizeof(int),
1391                 .mode           = 0444,
1392                 .proc_handler   = proc_nr_inodes,
1393         },
1394         {
1395                 .procname       = "file-nr",
1396                 .data           = &files_stat,
1397                 .maxlen         = sizeof(files_stat),
1398                 .mode           = 0444,
1399                 .proc_handler   = proc_nr_files,
1400         },
1401         {
1402                 .procname       = "file-max",
1403                 .data           = &files_stat.max_files,
1404                 .maxlen         = sizeof(files_stat.max_files),
1405                 .mode           = 0644,
1406                 .proc_handler   = proc_doulongvec_minmax,
1407         },
1408         {
1409                 .procname       = "nr_open",
1410                 .data           = &sysctl_nr_open,
1411                 .maxlen         = sizeof(int),
1412                 .mode           = 0644,
1413                 .proc_handler   = proc_dointvec_minmax,
1414                 .extra1         = &sysctl_nr_open_min,
1415                 .extra2         = &sysctl_nr_open_max,
1416         },
1417         {
1418                 .procname       = "dentry-state",
1419                 .data           = &dentry_stat,
1420                 .maxlen         = 6*sizeof(int),
1421                 .mode           = 0444,
1422                 .proc_handler   = proc_nr_dentry,
1423         },
1424         {
1425                 .procname       = "overflowuid",
1426                 .data           = &fs_overflowuid,
1427                 .maxlen         = sizeof(int),
1428                 .mode           = 0644,
1429                 .proc_handler   = proc_dointvec_minmax,
1430                 .extra1         = &minolduid,
1431                 .extra2         = &maxolduid,
1432         },
1433         {
1434                 .procname       = "overflowgid",
1435                 .data           = &fs_overflowgid,
1436                 .maxlen         = sizeof(int),
1437                 .mode           = 0644,
1438                 .proc_handler   = proc_dointvec_minmax,
1439                 .extra1         = &minolduid,
1440                 .extra2         = &maxolduid,
1441         },
1442 #ifdef CONFIG_FILE_LOCKING
1443         {
1444                 .procname       = "leases-enable",
1445                 .data           = &leases_enable,
1446                 .maxlen         = sizeof(int),
1447                 .mode           = 0644,
1448                 .proc_handler   = proc_dointvec,
1449         },
1450 #endif
1451 #ifdef CONFIG_DNOTIFY
1452         {
1453                 .procname       = "dir-notify-enable",
1454                 .data           = &dir_notify_enable,
1455                 .maxlen         = sizeof(int),
1456                 .mode           = 0644,
1457                 .proc_handler   = proc_dointvec,
1458         },
1459 #endif
1460 #ifdef CONFIG_MMU
1461 #ifdef CONFIG_FILE_LOCKING
1462         {
1463                 .procname       = "lease-break-time",
1464                 .data           = &lease_break_time,
1465                 .maxlen         = sizeof(int),
1466                 .mode           = 0644,
1467                 .proc_handler   = proc_dointvec,
1468         },
1469 #endif
1470 #ifdef CONFIG_AIO
1471         {
1472                 .procname       = "aio-nr",
1473                 .data           = &aio_nr,
1474                 .maxlen         = sizeof(aio_nr),
1475                 .mode           = 0444,
1476                 .proc_handler   = proc_doulongvec_minmax,
1477         },
1478         {
1479                 .procname       = "aio-max-nr",
1480                 .data           = &aio_max_nr,
1481                 .maxlen         = sizeof(aio_max_nr),
1482                 .mode           = 0644,
1483                 .proc_handler   = proc_doulongvec_minmax,
1484         },
1485 #endif /* CONFIG_AIO */
1486 #ifdef CONFIG_INOTIFY_USER
1487         {
1488                 .procname       = "inotify",
1489                 .mode           = 0555,
1490                 .child          = inotify_table,
1491         },
1492 #endif
1493 #ifdef CONFIG_EPOLL
1494         {
1495                 .procname       = "epoll",
1496                 .mode           = 0555,
1497                 .child          = epoll_table,
1498         },
1499 #endif
1500 #endif
1501         {
1502                 .procname       = "suid_dumpable",
1503                 .data           = &suid_dumpable,
1504                 .maxlen         = sizeof(int),
1505                 .mode           = 0644,
1506                 .proc_handler   = proc_dointvec_minmax_coredump,
1507                 .extra1         = &zero,
1508                 .extra2         = &two,
1509         },
1510 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1511         {
1512                 .procname       = "binfmt_misc",
1513                 .mode           = 0555,
1514                 .child          = binfmt_misc_table,
1515         },
1516 #endif
1517         {
1518                 .procname       = "pipe-max-size",
1519                 .data           = &pipe_max_size,
1520                 .maxlen         = sizeof(int),
1521                 .mode           = 0644,
1522                 .proc_handler   = &pipe_proc_fn,
1523                 .extra1         = &pipe_min_size,
1524         },
1525         { }
1526 };
1527
1528 static struct ctl_table debug_table[] = {
1529 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1530     defined(CONFIG_S390) || defined(CONFIG_TILE)
1531         {
1532                 .procname       = "exception-trace",
1533                 .data           = &show_unhandled_signals,
1534                 .maxlen         = sizeof(int),
1535                 .mode           = 0644,
1536                 .proc_handler   = proc_dointvec
1537         },
1538 #endif
1539 #if defined(CONFIG_OPTPROBES)
1540         {
1541                 .procname       = "kprobes-optimization",
1542                 .data           = &sysctl_kprobes_optimization,
1543                 .maxlen         = sizeof(int),
1544                 .mode           = 0644,
1545                 .proc_handler   = proc_kprobes_optimization_handler,
1546                 .extra1         = &zero,
1547                 .extra2         = &one,
1548         },
1549 #endif
1550         { }
1551 };
1552
1553 static struct ctl_table dev_table[] = {
1554         { }
1555 };
1556
1557 int __init sysctl_init(void)
1558 {
1559         register_sysctl_table(sysctl_base_table);
1560         return 0;
1561 }
1562
1563 #endif /* CONFIG_SYSCTL */
1564
1565 /*
1566  * /proc/sys support
1567  */
1568
1569 #ifdef CONFIG_PROC_SYSCTL
1570
1571 static int _proc_do_string(void* data, int maxlen, int write,
1572                            void __user *buffer,
1573                            size_t *lenp, loff_t *ppos)
1574 {
1575         size_t len;
1576         char __user *p;
1577         char c;
1578
1579         if (!data || !maxlen || !*lenp) {
1580                 *lenp = 0;
1581                 return 0;
1582         }
1583
1584         if (write) {
1585                 len = 0;
1586                 p = buffer;
1587                 while (len < *lenp) {
1588                         if (get_user(c, p++))
1589                                 return -EFAULT;
1590                         if (c == 0 || c == '\n')
1591                                 break;
1592                         len++;
1593                 }
1594                 if (len >= maxlen)
1595                         len = maxlen-1;
1596                 if(copy_from_user(data, buffer, len))
1597                         return -EFAULT;
1598                 ((char *) data)[len] = 0;
1599                 *ppos += *lenp;
1600         } else {
1601                 len = strlen(data);
1602                 if (len > maxlen)
1603                         len = maxlen;
1604
1605                 if (*ppos > len) {
1606                         *lenp = 0;
1607                         return 0;
1608                 }
1609
1610                 data += *ppos;
1611                 len  -= *ppos;
1612
1613                 if (len > *lenp)
1614                         len = *lenp;
1615                 if (len)
1616                         if(copy_to_user(buffer, data, len))
1617                                 return -EFAULT;
1618                 if (len < *lenp) {
1619                         if(put_user('\n', ((char __user *) buffer) + len))
1620                                 return -EFAULT;
1621                         len++;
1622                 }
1623                 *lenp = len;
1624                 *ppos += len;
1625         }
1626         return 0;
1627 }
1628
1629 /**
1630  * proc_dostring - read a string sysctl
1631  * @table: the sysctl table
1632  * @write: %TRUE if this is a write to the sysctl file
1633  * @buffer: the user buffer
1634  * @lenp: the size of the user buffer
1635  * @ppos: file position
1636  *
1637  * Reads/writes a string from/to the user buffer. If the kernel
1638  * buffer provided is not large enough to hold the string, the
1639  * string is truncated. The copied string is %NULL-terminated.
1640  * If the string is being read by the user process, it is copied
1641  * and a newline '\n' is added. It is truncated if the buffer is
1642  * not large enough.
1643  *
1644  * Returns 0 on success.
1645  */
1646 int proc_dostring(struct ctl_table *table, int write,
1647                   void __user *buffer, size_t *lenp, loff_t *ppos)
1648 {
1649         return _proc_do_string(table->data, table->maxlen, write,
1650                                buffer, lenp, ppos);
1651 }
1652
1653 static size_t proc_skip_spaces(char **buf)
1654 {
1655         size_t ret;
1656         char *tmp = skip_spaces(*buf);
1657         ret = tmp - *buf;
1658         *buf = tmp;
1659         return ret;
1660 }
1661
1662 static void proc_skip_char(char **buf, size_t *size, const char v)
1663 {
1664         while (*size) {
1665                 if (**buf != v)
1666                         break;
1667                 (*size)--;
1668                 (*buf)++;
1669         }
1670 }
1671
1672 #define TMPBUFLEN 22
1673 /**
1674  * proc_get_long - reads an ASCII formatted integer from a user buffer
1675  *
1676  * @buf: a kernel buffer
1677  * @size: size of the kernel buffer
1678  * @val: this is where the number will be stored
1679  * @neg: set to %TRUE if number is negative
1680  * @perm_tr: a vector which contains the allowed trailers
1681  * @perm_tr_len: size of the perm_tr vector
1682  * @tr: pointer to store the trailer character
1683  *
1684  * In case of success %0 is returned and @buf and @size are updated with
1685  * the amount of bytes read. If @tr is non-NULL and a trailing
1686  * character exists (size is non-zero after returning from this
1687  * function), @tr is updated with the trailing character.
1688  */
1689 static int proc_get_long(char **buf, size_t *size,
1690                           unsigned long *val, bool *neg,
1691                           const char *perm_tr, unsigned perm_tr_len, char *tr)
1692 {
1693         int len;
1694         char *p, tmp[TMPBUFLEN];
1695
1696         if (!*size)
1697                 return -EINVAL;
1698
1699         len = *size;
1700         if (len > TMPBUFLEN - 1)
1701                 len = TMPBUFLEN - 1;
1702
1703         memcpy(tmp, *buf, len);
1704
1705         tmp[len] = 0;
1706         p = tmp;
1707         if (*p == '-' && *size > 1) {
1708                 *neg = true;
1709                 p++;
1710         } else
1711                 *neg = false;
1712         if (!isdigit(*p))
1713                 return -EINVAL;
1714
1715         *val = simple_strtoul(p, &p, 0);
1716
1717         len = p - tmp;
1718
1719         /* We don't know if the next char is whitespace thus we may accept
1720          * invalid integers (e.g. 1234...a) or two integers instead of one
1721          * (e.g. 123...1). So lets not allow such large numbers. */
1722         if (len == TMPBUFLEN - 1)
1723                 return -EINVAL;
1724
1725         if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1726                 return -EINVAL;
1727
1728         if (tr && (len < *size))
1729                 *tr = *p;
1730
1731         *buf += len;
1732         *size -= len;
1733
1734         return 0;
1735 }
1736
1737 /**
1738  * proc_put_long - converts an integer to a decimal ASCII formatted string
1739  *
1740  * @buf: the user buffer
1741  * @size: the size of the user buffer
1742  * @val: the integer to be converted
1743  * @neg: sign of the number, %TRUE for negative
1744  *
1745  * In case of success %0 is returned and @buf and @size are updated with
1746  * the amount of bytes written.
1747  */
1748 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1749                           bool neg)
1750 {
1751         int len;
1752         char tmp[TMPBUFLEN], *p = tmp;
1753
1754         sprintf(p, "%s%lu", neg ? "-" : "", val);
1755         len = strlen(tmp);
1756         if (len > *size)
1757                 len = *size;
1758         if (copy_to_user(*buf, tmp, len))
1759                 return -EFAULT;
1760         *size -= len;
1761         *buf += len;
1762         return 0;
1763 }
1764 #undef TMPBUFLEN
1765
1766 static int proc_put_char(void __user **buf, size_t *size, char c)
1767 {
1768         if (*size) {
1769                 char __user **buffer = (char __user **)buf;
1770                 if (put_user(c, *buffer))
1771                         return -EFAULT;
1772                 (*size)--, (*buffer)++;
1773                 *buf = *buffer;
1774         }
1775         return 0;
1776 }
1777
1778 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1779                                  int *valp,
1780                                  int write, void *data)
1781 {
1782         if (write) {
1783                 *valp = *negp ? -*lvalp : *lvalp;
1784         } else {
1785                 int val = *valp;
1786                 if (val < 0) {
1787                         *negp = true;
1788                         *lvalp = (unsigned long)-val;
1789                 } else {
1790                         *negp = false;
1791                         *lvalp = (unsigned long)val;
1792                 }
1793         }
1794         return 0;
1795 }
1796
1797 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1798
1799 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1800                   int write, void __user *buffer,
1801                   size_t *lenp, loff_t *ppos,
1802                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1803                               int write, void *data),
1804                   void *data)
1805 {
1806         int *i, vleft, first = 1, err = 0;
1807         unsigned long page = 0;
1808         size_t left;
1809         char *kbuf;
1810
1811         if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1812                 *lenp = 0;
1813                 return 0;
1814         }
1815
1816         i = (int *) tbl_data;
1817         vleft = table->maxlen / sizeof(*i);
1818         left = *lenp;
1819
1820         if (!conv)
1821                 conv = do_proc_dointvec_conv;
1822
1823         if (write) {
1824                 if (left > PAGE_SIZE - 1)
1825                         left = PAGE_SIZE - 1;
1826                 page = __get_free_page(GFP_TEMPORARY);
1827                 kbuf = (char *) page;
1828                 if (!kbuf)
1829                         return -ENOMEM;
1830                 if (copy_from_user(kbuf, buffer, left)) {
1831                         err = -EFAULT;
1832                         goto free;
1833                 }
1834                 kbuf[left] = 0;
1835         }
1836
1837         for (; left && vleft--; i++, first=0) {
1838                 unsigned long lval;
1839                 bool neg;
1840
1841                 if (write) {
1842                         left -= proc_skip_spaces(&kbuf);
1843
1844                         if (!left)
1845                                 break;
1846                         err = proc_get_long(&kbuf, &left, &lval, &neg,
1847                                              proc_wspace_sep,
1848                                              sizeof(proc_wspace_sep), NULL);
1849                         if (err)
1850                                 break;
1851                         if (conv(&neg, &lval, i, 1, data)) {
1852                                 err = -EINVAL;
1853                                 break;
1854                         }
1855                 } else {
1856                         if (conv(&neg, &lval, i, 0, data)) {
1857                                 err = -EINVAL;
1858                                 break;
1859                         }
1860                         if (!first)
1861                                 err = proc_put_char(&buffer, &left, '\t');
1862                         if (err)
1863                                 break;
1864                         err = proc_put_long(&buffer, &left, lval, neg);
1865                         if (err)
1866                                 break;
1867                 }
1868         }
1869
1870         if (!write && !first && left && !err)
1871                 err = proc_put_char(&buffer, &left, '\n');
1872         if (write && !err && left)
1873                 left -= proc_skip_spaces(&kbuf);
1874 free:
1875         if (write) {
1876                 free_page(page);
1877                 if (first)
1878                         return err ? : -EINVAL;
1879         }
1880         *lenp -= left;
1881         *ppos += *lenp;
1882         return err;
1883 }
1884
1885 static int do_proc_dointvec(struct ctl_table *table, int write,
1886                   void __user *buffer, size_t *lenp, loff_t *ppos,
1887                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1888                               int write, void *data),
1889                   void *data)
1890 {
1891         return __do_proc_dointvec(table->data, table, write,
1892                         buffer, lenp, ppos, conv, data);
1893 }
1894
1895 /**
1896  * proc_dointvec - read a vector of integers
1897  * @table: the sysctl table
1898  * @write: %TRUE if this is a write to the sysctl file
1899  * @buffer: the user buffer
1900  * @lenp: the size of the user buffer
1901  * @ppos: file position
1902  *
1903  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1904  * values from/to the user buffer, treated as an ASCII string.
1905  *
1906  * Returns 0 on success.
1907  */
1908 int proc_dointvec(struct ctl_table *table, int write,
1909                      void __user *buffer, size_t *lenp, loff_t *ppos)
1910 {
1911     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1912                             NULL,NULL);
1913 }
1914
1915 /*
1916  * Taint values can only be increased
1917  * This means we can safely use a temporary.
1918  */
1919 static int proc_taint(struct ctl_table *table, int write,
1920                                void __user *buffer, size_t *lenp, loff_t *ppos)
1921 {
1922         struct ctl_table t;
1923         unsigned long tmptaint = get_taint();
1924         int err;
1925
1926         if (write && !capable(CAP_SYS_ADMIN))
1927                 return -EPERM;
1928
1929         t = *table;
1930         t.data = &tmptaint;
1931         err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1932         if (err < 0)
1933                 return err;
1934
1935         if (write) {
1936                 /*
1937                  * Poor man's atomic or. Not worth adding a primitive
1938                  * to everyone's atomic.h for this
1939                  */
1940                 int i;
1941                 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1942                         if ((tmptaint >> i) & 1)
1943                                 add_taint(i);
1944                 }
1945         }
1946
1947         return err;
1948 }
1949
1950 #ifdef CONFIG_PRINTK
1951 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1952                                 void __user *buffer, size_t *lenp, loff_t *ppos)
1953 {
1954         if (write && !capable(CAP_SYS_ADMIN))
1955                 return -EPERM;
1956
1957         return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1958 }
1959 #endif
1960
1961 struct do_proc_dointvec_minmax_conv_param {
1962         int *min;
1963         int *max;
1964 };
1965
1966 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1967                                         int *valp,
1968                                         int write, void *data)
1969 {
1970         struct do_proc_dointvec_minmax_conv_param *param = data;
1971         if (write) {
1972                 int val = *negp ? -*lvalp : *lvalp;
1973                 if ((param->min && *param->min > val) ||
1974                     (param->max && *param->max < val))
1975                         return -EINVAL;
1976                 *valp = val;
1977         } else {
1978                 int val = *valp;
1979                 if (val < 0) {
1980                         *negp = true;
1981                         *lvalp = (unsigned long)-val;
1982                 } else {
1983                         *negp = false;
1984                         *lvalp = (unsigned long)val;
1985                 }
1986         }
1987         return 0;
1988 }
1989
1990 /**
1991  * proc_dointvec_minmax - read a vector of integers with min/max values
1992  * @table: the sysctl table
1993  * @write: %TRUE if this is a write to the sysctl file
1994  * @buffer: the user buffer
1995  * @lenp: the size of the user buffer
1996  * @ppos: file position
1997  *
1998  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1999  * values from/to the user buffer, treated as an ASCII string.
2000  *
2001  * This routine will ensure the values are within the range specified by
2002  * table->extra1 (min) and table->extra2 (max).
2003  *
2004  * Returns 0 on success.
2005  */
2006 int proc_dointvec_minmax(struct ctl_table *table, int write,
2007                   void __user *buffer, size_t *lenp, loff_t *ppos)
2008 {
2009         struct do_proc_dointvec_minmax_conv_param param = {
2010                 .min = (int *) table->extra1,
2011                 .max = (int *) table->extra2,
2012         };
2013         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2014                                 do_proc_dointvec_minmax_conv, &param);
2015 }
2016
2017 static void validate_coredump_safety(void)
2018 {
2019         if (suid_dumpable == SUID_DUMPABLE_SAFE &&
2020             core_pattern[0] != '/' && core_pattern[0] != '|') {
2021                 printk(KERN_WARNING "Unsafe core_pattern used with "\
2022                         "suid_dumpable=2. Pipe handler or fully qualified "\
2023                         "core dump path required.\n");
2024         }
2025 }
2026
2027 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2028                 void __user *buffer, size_t *lenp, loff_t *ppos)
2029 {
2030         int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2031         if (!error)
2032                 validate_coredump_safety();
2033         return error;
2034 }
2035
2036 static int proc_dostring_coredump(struct ctl_table *table, int write,
2037                   void __user *buffer, size_t *lenp, loff_t *ppos)
2038 {
2039         int error = proc_dostring(table, write, buffer, lenp, ppos);
2040         if (!error)
2041                 validate_coredump_safety();
2042         return error;
2043 }
2044
2045 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2046                                      void __user *buffer,
2047                                      size_t *lenp, loff_t *ppos,
2048                                      unsigned long convmul,
2049                                      unsigned long convdiv)
2050 {
2051         unsigned long *i, *min, *max;
2052         int vleft, first = 1, err = 0;
2053         unsigned long page = 0;
2054         size_t left;
2055         char *kbuf;
2056
2057         if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2058                 *lenp = 0;
2059                 return 0;
2060         }
2061
2062         i = (unsigned long *) data;
2063         min = (unsigned long *) table->extra1;
2064         max = (unsigned long *) table->extra2;
2065         vleft = table->maxlen / sizeof(unsigned long);
2066         left = *lenp;
2067
2068         if (write) {
2069                 if (left > PAGE_SIZE - 1)
2070                         left = PAGE_SIZE - 1;
2071                 page = __get_free_page(GFP_TEMPORARY);
2072                 kbuf = (char *) page;
2073                 if (!kbuf)
2074                         return -ENOMEM;
2075                 if (copy_from_user(kbuf, buffer, left)) {
2076                         err = -EFAULT;
2077                         goto free;
2078                 }
2079                 kbuf[left] = 0;
2080         }
2081
2082         for (; left && vleft--; i++, first = 0) {
2083                 unsigned long val;
2084
2085                 if (write) {
2086                         bool neg;
2087
2088                         left -= proc_skip_spaces(&kbuf);
2089
2090                         err = proc_get_long(&kbuf, &left, &val, &neg,
2091                                              proc_wspace_sep,
2092                                              sizeof(proc_wspace_sep), NULL);
2093                         if (err)
2094                                 break;
2095                         if (neg)
2096                                 continue;
2097                         if ((min && val < *min) || (max && val > *max))
2098                                 continue;
2099                         *i = val;
2100                 } else {
2101                         val = convdiv * (*i) / convmul;
2102                         if (!first)
2103                                 err = proc_put_char(&buffer, &left, '\t');
2104                         err = proc_put_long(&buffer, &left, val, false);
2105                         if (err)
2106                                 break;
2107                 }
2108         }
2109
2110         if (!write && !first && left && !err)
2111                 err = proc_put_char(&buffer, &left, '\n');
2112         if (write && !err)
2113                 left -= proc_skip_spaces(&kbuf);
2114 free:
2115         if (write) {
2116                 free_page(page);
2117                 if (first)
2118                         return err ? : -EINVAL;
2119         }
2120         *lenp -= left;
2121         *ppos += *lenp;
2122         return err;
2123 }
2124
2125 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2126                                      void __user *buffer,
2127                                      size_t *lenp, loff_t *ppos,
2128                                      unsigned long convmul,
2129                                      unsigned long convdiv)
2130 {
2131         return __do_proc_doulongvec_minmax(table->data, table, write,
2132                         buffer, lenp, ppos, convmul, convdiv);
2133 }
2134
2135 /**
2136  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2137  * @table: the sysctl table
2138  * @write: %TRUE if this is a write to the sysctl file
2139  * @buffer: the user buffer
2140  * @lenp: the size of the user buffer
2141  * @ppos: file position
2142  *
2143  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2144  * values from/to the user buffer, treated as an ASCII string.
2145  *
2146  * This routine will ensure the values are within the range specified by
2147  * table->extra1 (min) and table->extra2 (max).
2148  *
2149  * Returns 0 on success.
2150  */
2151 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2152                            void __user *buffer, size_t *lenp, loff_t *ppos)
2153 {
2154     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2155 }
2156
2157 /**
2158  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2159  * @table: the sysctl table
2160  * @write: %TRUE if this is a write to the sysctl file
2161  * @buffer: the user buffer
2162  * @lenp: the size of the user buffer
2163  * @ppos: file position
2164  *
2165  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2166  * values from/to the user buffer, treated as an ASCII string. The values
2167  * are treated as milliseconds, and converted to jiffies when they are stored.
2168  *
2169  * This routine will ensure the values are within the range specified by
2170  * table->extra1 (min) and table->extra2 (max).
2171  *
2172  * Returns 0 on success.
2173  */
2174 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2175                                       void __user *buffer,
2176                                       size_t *lenp, loff_t *ppos)
2177 {
2178     return do_proc_doulongvec_minmax(table, write, buffer,
2179                                      lenp, ppos, HZ, 1000l);
2180 }
2181
2182
2183 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2184                                          int *valp,
2185                                          int write, void *data)
2186 {
2187         if (write) {
2188                 if (*lvalp > LONG_MAX / HZ)
2189                         return 1;
2190                 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2191         } else {
2192                 int val = *valp;
2193                 unsigned long lval;
2194                 if (val < 0) {
2195                         *negp = true;
2196                         lval = (unsigned long)-val;
2197                 } else {
2198                         *negp = false;
2199                         lval = (unsigned long)val;
2200                 }
2201                 *lvalp = lval / HZ;
2202         }
2203         return 0;
2204 }
2205
2206 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2207                                                 int *valp,
2208                                                 int write, void *data)
2209 {
2210         if (write) {
2211                 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2212                         return 1;
2213                 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2214         } else {
2215                 int val = *valp;
2216                 unsigned long lval;
2217                 if (val < 0) {
2218                         *negp = true;
2219                         lval = (unsigned long)-val;
2220                 } else {
2221                         *negp = false;
2222                         lval = (unsigned long)val;
2223                 }
2224                 *lvalp = jiffies_to_clock_t(lval);
2225         }
2226         return 0;
2227 }
2228
2229 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2230                                             int *valp,
2231                                             int write, void *data)
2232 {
2233         if (write) {
2234                 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2235         } else {
2236                 int val = *valp;
2237                 unsigned long lval;
2238                 if (val < 0) {
2239                         *negp = true;
2240                         lval = (unsigned long)-val;
2241                 } else {
2242                         *negp = false;
2243                         lval = (unsigned long)val;
2244                 }
2245                 *lvalp = jiffies_to_msecs(lval);
2246         }
2247         return 0;
2248 }
2249
2250 /**
2251  * proc_dointvec_jiffies - read a vector of integers as seconds
2252  * @table: the sysctl table
2253  * @write: %TRUE if this is a write to the sysctl file
2254  * @buffer: the user buffer
2255  * @lenp: the size of the user buffer
2256  * @ppos: file position
2257  *
2258  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2259  * values from/to the user buffer, treated as an ASCII string.
2260  * The values read are assumed to be in seconds, and are converted into
2261  * jiffies.
2262  *
2263  * Returns 0 on success.
2264  */
2265 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2266                           void __user *buffer, size_t *lenp, loff_t *ppos)
2267 {
2268     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2269                             do_proc_dointvec_jiffies_conv,NULL);
2270 }
2271
2272 /**
2273  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2274  * @table: the sysctl table
2275  * @write: %TRUE if this is a write to the sysctl file
2276  * @buffer: the user buffer
2277  * @lenp: the size of the user buffer
2278  * @ppos: pointer to the file position
2279  *
2280  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2281  * values from/to the user buffer, treated as an ASCII string.
2282  * The values read are assumed to be in 1/USER_HZ seconds, and
2283  * are converted into jiffies.
2284  *
2285  * Returns 0 on success.
2286  */
2287 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2288                                  void __user *buffer, size_t *lenp, loff_t *ppos)
2289 {
2290     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2291                             do_proc_dointvec_userhz_jiffies_conv,NULL);
2292 }
2293
2294 /**
2295  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2296  * @table: the sysctl table
2297  * @write: %TRUE if this is a write to the sysctl file
2298  * @buffer: the user buffer
2299  * @lenp: the size of the user buffer
2300  * @ppos: file position
2301  * @ppos: the current position in the file
2302  *
2303  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2304  * values from/to the user buffer, treated as an ASCII string.
2305  * The values read are assumed to be in 1/1000 seconds, and
2306  * are converted into jiffies.
2307  *
2308  * Returns 0 on success.
2309  */
2310 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2311                              void __user *buffer, size_t *lenp, loff_t *ppos)
2312 {
2313         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2314                                 do_proc_dointvec_ms_jiffies_conv, NULL);
2315 }
2316
2317 static int proc_do_cad_pid(struct ctl_table *table, int write,
2318                            void __user *buffer, size_t *lenp, loff_t *ppos)
2319 {
2320         struct pid *new_pid;
2321         pid_t tmp;
2322         int r;
2323
2324         tmp = pid_vnr(cad_pid);
2325
2326         r = __do_proc_dointvec(&tmp, table, write, buffer,
2327                                lenp, ppos, NULL, NULL);
2328         if (r || !write)
2329                 return r;
2330
2331         new_pid = find_get_pid(tmp);
2332         if (!new_pid)
2333                 return -ESRCH;
2334
2335         put_pid(xchg(&cad_pid, new_pid));
2336         return 0;
2337 }
2338
2339 /**
2340  * proc_do_large_bitmap - read/write from/to a large bitmap
2341  * @table: the sysctl table
2342  * @write: %TRUE if this is a write to the sysctl file
2343  * @buffer: the user buffer
2344  * @lenp: the size of the user buffer
2345  * @ppos: file position
2346  *
2347  * The bitmap is stored at table->data and the bitmap length (in bits)
2348  * in table->maxlen.
2349  *
2350  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2351  * large bitmaps may be represented in a compact manner. Writing into
2352  * the file will clear the bitmap then update it with the given input.
2353  *
2354  * Returns 0 on success.
2355  */
2356 int proc_do_large_bitmap(struct ctl_table *table, int write,
2357                          void __user *buffer, size_t *lenp, loff_t *ppos)
2358 {
2359         int err = 0;
2360         bool first = 1;
2361         size_t left = *lenp;
2362         unsigned long bitmap_len = table->maxlen;
2363         unsigned long *bitmap = (unsigned long *) table->data;
2364         unsigned long *tmp_bitmap = NULL;
2365         char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2366
2367         if (!bitmap_len || !left || (*ppos && !write)) {
2368                 *lenp = 0;
2369                 return 0;
2370         }
2371
2372         if (write) {
2373                 unsigned long page = 0;
2374                 char *kbuf;
2375
2376                 if (left > PAGE_SIZE - 1)
2377                         left = PAGE_SIZE - 1;
2378
2379                 page = __get_free_page(GFP_TEMPORARY);
2380                 kbuf = (char *) page;
2381                 if (!kbuf)
2382                         return -ENOMEM;
2383                 if (copy_from_user(kbuf, buffer, left)) {
2384                         free_page(page);
2385                         return -EFAULT;
2386                 }
2387                 kbuf[left] = 0;
2388
2389                 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2390                                      GFP_KERNEL);
2391                 if (!tmp_bitmap) {
2392                         free_page(page);
2393                         return -ENOMEM;
2394                 }
2395                 proc_skip_char(&kbuf, &left, '\n');
2396                 while (!err && left) {
2397                         unsigned long val_a, val_b;
2398                         bool neg;
2399
2400                         err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2401                                              sizeof(tr_a), &c);
2402                         if (err)
2403                                 break;
2404                         if (val_a >= bitmap_len || neg) {
2405                                 err = -EINVAL;
2406                                 break;
2407                         }
2408
2409                         val_b = val_a;
2410                         if (left) {
2411                                 kbuf++;
2412                                 left--;
2413                         }
2414
2415                         if (c == '-') {
2416                                 err = proc_get_long(&kbuf, &left, &val_b,
2417                                                      &neg, tr_b, sizeof(tr_b),
2418                                                      &c);
2419                                 if (err)
2420                                         break;
2421                                 if (val_b >= bitmap_len || neg ||
2422                                     val_a > val_b) {
2423                                         err = -EINVAL;
2424                                         break;
2425                                 }
2426                                 if (left) {
2427                                         kbuf++;
2428                                         left--;
2429                                 }
2430                         }
2431
2432                         bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2433                         first = 0;
2434                         proc_skip_char(&kbuf, &left, '\n');
2435                 }
2436                 free_page(page);
2437         } else {
2438                 unsigned long bit_a, bit_b = 0;
2439
2440                 while (left) {
2441                         bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2442                         if (bit_a >= bitmap_len)
2443                                 break;
2444                         bit_b = find_next_zero_bit(bitmap, bitmap_len,
2445                                                    bit_a + 1) - 1;
2446
2447                         if (!first) {
2448                                 err = proc_put_char(&buffer, &left, ',');
2449                                 if (err)
2450                                         break;
2451                         }
2452                         err = proc_put_long(&buffer, &left, bit_a, false);
2453                         if (err)
2454                                 break;
2455                         if (bit_a != bit_b) {
2456                                 err = proc_put_char(&buffer, &left, '-');
2457                                 if (err)
2458                                         break;
2459                                 err = proc_put_long(&buffer, &left, bit_b, false);
2460                                 if (err)
2461                                         break;
2462                         }
2463
2464                         first = 0; bit_b++;
2465                 }
2466                 if (!err)
2467                         err = proc_put_char(&buffer, &left, '\n');
2468         }
2469
2470         if (!err) {
2471                 if (write) {
2472                         if (*ppos)
2473                                 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2474                         else
2475                                 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2476                 }
2477                 kfree(tmp_bitmap);
2478                 *lenp -= left;
2479                 *ppos += *lenp;
2480                 return 0;
2481         } else {
2482                 kfree(tmp_bitmap);
2483                 return err;
2484         }
2485 }
2486
2487 #else /* CONFIG_PROC_SYSCTL */
2488
2489 int proc_dostring(struct ctl_table *table, int write,
2490                   void __user *buffer, size_t *lenp, loff_t *ppos)
2491 {
2492         return -ENOSYS;
2493 }
2494
2495 int proc_dointvec(struct ctl_table *table, int write,
2496                   void __user *buffer, size_t *lenp, loff_t *ppos)
2497 {
2498         return -ENOSYS;
2499 }
2500
2501 int proc_dointvec_minmax(struct ctl_table *table, int write,
2502                     void __user *buffer, size_t *lenp, loff_t *ppos)
2503 {
2504         return -ENOSYS;
2505 }
2506
2507 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2508                     void __user *buffer, size_t *lenp, loff_t *ppos)
2509 {
2510         return -ENOSYS;
2511 }
2512
2513 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2514                     void __user *buffer, size_t *lenp, loff_t *ppos)
2515 {
2516         return -ENOSYS;
2517 }
2518
2519 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2520                              void __user *buffer, size_t *lenp, loff_t *ppos)
2521 {
2522         return -ENOSYS;
2523 }
2524
2525 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2526                     void __user *buffer, size_t *lenp, loff_t *ppos)
2527 {
2528         return -ENOSYS;
2529 }
2530
2531 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2532                                       void __user *buffer,
2533                                       size_t *lenp, loff_t *ppos)
2534 {
2535     return -ENOSYS;
2536 }
2537
2538
2539 #endif /* CONFIG_PROC_SYSCTL */
2540
2541 /*
2542  * No sense putting this after each symbol definition, twice,
2543  * exception granted :-)
2544  */
2545 EXPORT_SYMBOL(proc_dointvec);
2546 EXPORT_SYMBOL(proc_dointvec_jiffies);
2547 EXPORT_SYMBOL(proc_dointvec_minmax);
2548 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2549 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2550 EXPORT_SYMBOL(proc_dostring);
2551 EXPORT_SYMBOL(proc_doulongvec_minmax);
2552 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);