linux/kernel/sysctl.c
<<
>>
Prefs
   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/proc_fs.h>
  27#include <linux/security.h>
  28#include <linux/ctype.h>
  29#include <linux/utsname.h>
  30#include <linux/smp_lock.h>
  31#include <linux/fs.h>
  32#include <linux/init.h>
  33#include <linux/kernel.h>
  34#include <linux/kobject.h>
  35#include <linux/net.h>
  36#include <linux/sysrq.h>
  37#include <linux/highuid.h>
  38#include <linux/writeback.h>
  39#include <linux/hugetlb.h>
  40#include <linux/initrd.h>
  41#include <linux/key.h>
  42#include <linux/times.h>
  43#include <linux/limits.h>
  44#include <linux/dcache.h>
  45#include <linux/syscalls.h>
  46#include <linux/vmstat.h>
  47#include <linux/nfs_fs.h>
  48#include <linux/acpi.h>
  49#include <linux/reboot.h>
  50#include <linux/ftrace.h>
  51#include <linux/slow-work.h>
  52
  53#include <asm/uaccess.h>
  54#include <asm/processor.h>
  55
  56#ifdef CONFIG_X86
  57#include <asm/nmi.h>
  58#include <asm/stacktrace.h>
  59#include <asm/io.h>
  60#endif
  61
  62static int deprecated_sysctl_warning(struct __sysctl_args *args);
  63
  64#if defined(CONFIG_SYSCTL)
  65
  66/* External variables not in a header file. */
  67extern int C_A_D;
  68extern int print_fatal_signals;
  69extern int sysctl_overcommit_memory;
  70extern int sysctl_overcommit_ratio;
  71extern int sysctl_panic_on_oom;
  72extern int sysctl_oom_kill_allocating_task;
  73extern int sysctl_oom_dump_tasks;
  74extern int max_threads;
  75extern int core_uses_pid;
  76extern int suid_dumpable;
  77extern char core_pattern[];
  78extern int pid_max;
  79extern int min_free_kbytes;
  80extern int pid_max_min, pid_max_max;
  81extern int sysctl_drop_caches;
  82extern int percpu_pagelist_fraction;
  83extern int compat_log;
  84extern int latencytop_enabled;
  85extern int sysctl_nr_open_min, sysctl_nr_open_max;
  86#ifndef CONFIG_MMU
  87extern int sysctl_nr_trim_pages;
  88#endif
  89#ifdef CONFIG_RCU_TORTURE_TEST
  90extern int rcutorture_runnable;
  91#endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
  92
  93/* Constants used for minimum and  maximum */
  94#ifdef CONFIG_DETECT_SOFTLOCKUP
  95static int sixty = 60;
  96static int neg_one = -1;
  97#endif
  98
  99static int zero;
 100static int __maybe_unused one = 1;
 101static int __maybe_unused two = 2;
 102static unsigned long one_ul = 1;
 103static int one_hundred = 100;
 104
 105/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 106static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 107
 108/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 109static int maxolduid = 65535;
 110static int minolduid;
 111static int min_percpu_pagelist_fract = 8;
 112
 113static int ngroups_max = NGROUPS_MAX;
 114
 115#ifdef CONFIG_MODULES
 116extern char modprobe_path[];
 117#endif
 118#ifdef CONFIG_CHR_DEV_SG
 119extern int sg_big_buff;
 120#endif
 121
 122#ifdef CONFIG_SPARC
 123#include <asm/system.h>
 124#endif
 125
 126#ifdef CONFIG_SPARC64
 127extern int sysctl_tsb_ratio;
 128#endif
 129
 130#ifdef __hppa__
 131extern int pwrsw_enabled;
 132extern int unaligned_enabled;
 133#endif
 134
 135#ifdef CONFIG_S390
 136#ifdef CONFIG_MATHEMU
 137extern int sysctl_ieee_emulation_warnings;
 138#endif
 139extern int sysctl_userprocess_debug;
 140extern int spin_retry;
 141#endif
 142
 143#ifdef CONFIG_BSD_PROCESS_ACCT
 144extern int acct_parm[];
 145#endif
 146
 147#ifdef CONFIG_IA64
 148extern int no_unaligned_warning;
 149extern int unaligned_dump_stack;
 150#endif
 151
 152#ifdef CONFIG_RT_MUTEXES
 153extern int max_lock_depth;
 154#endif
 155
 156#ifdef CONFIG_PROC_SYSCTL
 157static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
 158                  void __user *buffer, size_t *lenp, loff_t *ppos);
 159static int proc_taint(struct ctl_table *table, int write, struct file *filp,
 160                               void __user *buffer, size_t *lenp, loff_t *ppos);
 161#endif
 162
 163static struct ctl_table root_table[];
 164static struct ctl_table_root sysctl_table_root;
 165static struct ctl_table_header root_table_header = {
 166        .count = 1,
 167        .ctl_table = root_table,
 168        .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
 169        .root = &sysctl_table_root,
 170        .set = &sysctl_table_root.default_set,
 171};
 172static struct ctl_table_root sysctl_table_root = {
 173        .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
 174        .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
 175};
 176
 177static struct ctl_table kern_table[];
 178static struct ctl_table vm_table[];
 179static struct ctl_table fs_table[];
 180static struct ctl_table debug_table[];
 181static struct ctl_table dev_table[];
 182extern struct ctl_table random_table[];
 183#ifdef CONFIG_INOTIFY_USER
 184extern struct ctl_table inotify_table[];
 185#endif
 186#ifdef CONFIG_EPOLL
 187extern struct ctl_table epoll_table[];
 188#endif
 189
 190#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 191int sysctl_legacy_va_layout;
 192#endif
 193
 194extern int prove_locking;
 195extern int lock_stat;
 196
 197/* The default sysctl tables: */
 198
 199static struct ctl_table root_table[] = {
 200        {
 201                .ctl_name       = CTL_KERN,
 202                .procname       = "kernel",
 203                .mode           = 0555,
 204                .child          = kern_table,
 205        },
 206        {
 207                .ctl_name       = CTL_VM,
 208                .procname       = "vm",
 209                .mode           = 0555,
 210                .child          = vm_table,
 211        },
 212        {
 213                .ctl_name       = CTL_FS,
 214                .procname       = "fs",
 215                .mode           = 0555,
 216                .child          = fs_table,
 217        },
 218        {
 219                .ctl_name       = CTL_DEBUG,
 220                .procname       = "debug",
 221                .mode           = 0555,
 222                .child          = debug_table,
 223        },
 224        {
 225                .ctl_name       = CTL_DEV,
 226                .procname       = "dev",
 227                .mode           = 0555,
 228                .child          = dev_table,
 229        },
 230/*
 231 * NOTE: do not add new entries to this table unless you have read
 232 * Documentation/sysctl/ctl_unnumbered.txt
 233 */
 234        { .ctl_name = 0 }
 235};
 236
 237#ifdef CONFIG_SCHED_DEBUG
 238static int min_sched_granularity_ns = 100000;           /* 100 usecs */
 239static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
 240static int min_wakeup_granularity_ns;                   /* 0 usecs */
 241static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
 242#endif
 243
 244static struct ctl_table kern_table[] = {
 245#ifdef CONFIG_SCHED_DEBUG
 246        {
 247                .ctl_name       = CTL_UNNUMBERED,
 248                .procname       = "sched_min_granularity_ns",
 249                .data           = &sysctl_sched_min_granularity,
 250                .maxlen         = sizeof(unsigned int),
 251                .mode           = 0644,
 252                .proc_handler   = &sched_nr_latency_handler,
 253                .strategy       = &sysctl_intvec,
 254                .extra1         = &min_sched_granularity_ns,
 255                .extra2         = &max_sched_granularity_ns,
 256        },
 257        {
 258                .ctl_name       = CTL_UNNUMBERED,
 259                .procname       = "sched_latency_ns",
 260                .data           = &sysctl_sched_latency,
 261                .maxlen         = sizeof(unsigned int),
 262                .mode           = 0644,
 263                .proc_handler   = &sched_nr_latency_handler,
 264                .strategy       = &sysctl_intvec,
 265                .extra1         = &min_sched_granularity_ns,
 266                .extra2         = &max_sched_granularity_ns,
 267        },
 268        {
 269                .ctl_name       = CTL_UNNUMBERED,
 270                .procname       = "sched_wakeup_granularity_ns",
 271                .data           = &sysctl_sched_wakeup_granularity,
 272                .maxlen         = sizeof(unsigned int),
 273                .mode           = 0644,
 274                .proc_handler   = &proc_dointvec_minmax,
 275                .strategy       = &sysctl_intvec,
 276                .extra1         = &min_wakeup_granularity_ns,
 277                .extra2         = &max_wakeup_granularity_ns,
 278        },
 279        {
 280                .ctl_name       = CTL_UNNUMBERED,
 281                .procname       = "sched_shares_ratelimit",
 282                .data           = &sysctl_sched_shares_ratelimit,
 283                .maxlen         = sizeof(unsigned int),
 284                .mode           = 0644,
 285                .proc_handler   = &proc_dointvec,
 286        },
 287        {
 288                .ctl_name       = CTL_UNNUMBERED,
 289                .procname       = "sched_shares_thresh",
 290                .data           = &sysctl_sched_shares_thresh,
 291                .maxlen         = sizeof(unsigned int),
 292                .mode           = 0644,
 293                .proc_handler   = &proc_dointvec_minmax,
 294                .strategy       = &sysctl_intvec,
 295                .extra1         = &zero,
 296        },
 297        {
 298                .ctl_name       = CTL_UNNUMBERED,
 299                .procname       = "sched_child_runs_first",
 300                .data           = &sysctl_sched_child_runs_first,
 301                .maxlen         = sizeof(unsigned int),
 302                .mode           = 0644,
 303                .proc_handler   = &proc_dointvec,
 304        },
 305        {
 306                .ctl_name       = CTL_UNNUMBERED,
 307                .procname       = "sched_features",
 308                .data           = &sysctl_sched_features,
 309                .maxlen         = sizeof(unsigned int),
 310                .mode           = 0644,
 311                .proc_handler   = &proc_dointvec,
 312        },
 313        {
 314                .ctl_name       = CTL_UNNUMBERED,
 315                .procname       = "sched_migration_cost",
 316                .data           = &sysctl_sched_migration_cost,
 317                .maxlen         = sizeof(unsigned int),
 318                .mode           = 0644,
 319                .proc_handler   = &proc_dointvec,
 320        },
 321        {
 322                .ctl_name       = CTL_UNNUMBERED,
 323                .procname       = "sched_nr_migrate",
 324                .data           = &sysctl_sched_nr_migrate,
 325                .maxlen         = sizeof(unsigned int),
 326                .mode           = 0644,
 327                .proc_handler   = &proc_dointvec,
 328        },
 329#endif
 330        {
 331                .ctl_name       = CTL_UNNUMBERED,
 332                .procname       = "sched_rt_period_us",
 333                .data           = &sysctl_sched_rt_period,
 334                .maxlen         = sizeof(unsigned int),
 335                .mode           = 0644,
 336                .proc_handler   = &sched_rt_handler,
 337        },
 338        {
 339                .ctl_name       = CTL_UNNUMBERED,
 340                .procname       = "sched_rt_runtime_us",
 341                .data           = &sysctl_sched_rt_runtime,
 342                .maxlen         = sizeof(int),
 343                .mode           = 0644,
 344                .proc_handler   = &sched_rt_handler,
 345        },
 346        {
 347                .ctl_name       = CTL_UNNUMBERED,
 348                .procname       = "sched_compat_yield",
 349                .data           = &sysctl_sched_compat_yield,
 350                .maxlen         = sizeof(unsigned int),
 351                .mode           = 0644,
 352                .proc_handler   = &proc_dointvec,
 353        },
 354#ifdef CONFIG_PROVE_LOCKING
 355        {
 356                .ctl_name       = CTL_UNNUMBERED,
 357                .procname       = "prove_locking",
 358                .data           = &prove_locking,
 359                .maxlen         = sizeof(int),
 360                .mode           = 0644,
 361                .proc_handler   = &proc_dointvec,
 362        },
 363#endif
 364#ifdef CONFIG_LOCK_STAT
 365        {
 366                .ctl_name       = CTL_UNNUMBERED,
 367                .procname       = "lock_stat",
 368                .data           = &lock_stat,
 369                .maxlen         = sizeof(int),
 370                .mode           = 0644,
 371                .proc_handler   = &proc_dointvec,
 372        },
 373#endif
 374        {
 375                .ctl_name       = KERN_PANIC,
 376                .procname       = "panic",
 377                .data           = &panic_timeout,
 378                .maxlen         = sizeof(int),
 379                .mode           = 0644,
 380                .proc_handler   = &proc_dointvec,
 381        },
 382        {
 383                .ctl_name       = KERN_CORE_USES_PID,
 384                .procname       = "core_uses_pid",
 385                .data           = &core_uses_pid,
 386                .maxlen         = sizeof(int),
 387                .mode           = 0644,
 388                .proc_handler   = &proc_dointvec,
 389        },
 390        {
 391                .ctl_name       = KERN_CORE_PATTERN,
 392                .procname       = "core_pattern",
 393                .data           = core_pattern,
 394                .maxlen         = CORENAME_MAX_SIZE,
 395                .mode           = 0644,
 396                .proc_handler   = &proc_dostring,
 397                .strategy       = &sysctl_string,
 398        },
 399#ifdef CONFIG_PROC_SYSCTL
 400        {
 401                .procname       = "tainted",
 402                .maxlen         = sizeof(long),
 403                .mode           = 0644,
 404                .proc_handler   = &proc_taint,
 405        },
 406#endif
 407#ifdef CONFIG_LATENCYTOP
 408        {
 409                .procname       = "latencytop",
 410                .data           = &latencytop_enabled,
 411                .maxlen         = sizeof(int),
 412                .mode           = 0644,
 413                .proc_handler   = &proc_dointvec,
 414        },
 415#endif
 416#ifdef CONFIG_BLK_DEV_INITRD
 417        {
 418                .ctl_name       = KERN_REALROOTDEV,
 419                .procname       = "real-root-dev",
 420                .data           = &real_root_dev,
 421                .maxlen         = sizeof(int),
 422                .mode           = 0644,
 423                .proc_handler   = &proc_dointvec,
 424        },
 425#endif
 426        {
 427                .ctl_name       = CTL_UNNUMBERED,
 428                .procname       = "print-fatal-signals",
 429                .data           = &print_fatal_signals,
 430                .maxlen         = sizeof(int),
 431                .mode           = 0644,
 432                .proc_handler   = &proc_dointvec,
 433        },
 434#ifdef CONFIG_SPARC
 435        {
 436                .ctl_name       = KERN_SPARC_REBOOT,
 437                .procname       = "reboot-cmd",
 438                .data           = reboot_command,
 439                .maxlen         = 256,
 440                .mode           = 0644,
 441                .proc_handler   = &proc_dostring,
 442                .strategy       = &sysctl_string,
 443        },
 444        {
 445                .ctl_name       = KERN_SPARC_STOP_A,
 446                .procname       = "stop-a",
 447                .data           = &stop_a_enabled,
 448                .maxlen         = sizeof (int),
 449                .mode           = 0644,
 450                .proc_handler   = &proc_dointvec,
 451        },
 452        {
 453                .ctl_name       = KERN_SPARC_SCONS_PWROFF,
 454                .procname       = "scons-poweroff",
 455                .data           = &scons_pwroff,
 456                .maxlen         = sizeof (int),
 457                .mode           = 0644,
 458                .proc_handler   = &proc_dointvec,
 459        },
 460#endif
 461#ifdef CONFIG_SPARC64
 462        {
 463                .ctl_name       = CTL_UNNUMBERED,
 464                .procname       = "tsb-ratio",
 465                .data           = &sysctl_tsb_ratio,
 466                .maxlen         = sizeof (int),
 467                .mode           = 0644,
 468                .proc_handler   = &proc_dointvec,
 469        },
 470#endif
 471#ifdef __hppa__
 472        {
 473                .ctl_name       = KERN_HPPA_PWRSW,
 474                .procname       = "soft-power",
 475                .data           = &pwrsw_enabled,
 476                .maxlen         = sizeof (int),
 477                .mode           = 0644,
 478                .proc_handler   = &proc_dointvec,
 479        },
 480        {
 481                .ctl_name       = KERN_HPPA_UNALIGNED,
 482                .procname       = "unaligned-trap",
 483                .data           = &unaligned_enabled,
 484                .maxlen         = sizeof (int),
 485                .mode           = 0644,
 486                .proc_handler   = &proc_dointvec,
 487        },
 488#endif
 489        {
 490                .ctl_name       = KERN_CTLALTDEL,
 491                .procname       = "ctrl-alt-del",
 492                .data           = &C_A_D,
 493                .maxlen         = sizeof(int),
 494                .mode           = 0644,
 495                .proc_handler   = &proc_dointvec,
 496        },
 497#ifdef CONFIG_FUNCTION_TRACER
 498        {
 499                .ctl_name       = CTL_UNNUMBERED,
 500                .procname       = "ftrace_enabled",
 501                .data           = &ftrace_enabled,
 502                .maxlen         = sizeof(int),
 503                .mode           = 0644,
 504                .proc_handler   = &ftrace_enable_sysctl,
 505        },
 506#endif
 507#ifdef CONFIG_STACK_TRACER
 508        {
 509                .ctl_name       = CTL_UNNUMBERED,
 510                .procname       = "stack_tracer_enabled",
 511                .data           = &stack_tracer_enabled,
 512                .maxlen         = sizeof(int),
 513                .mode           = 0644,
 514                .proc_handler   = &stack_trace_sysctl,
 515        },
 516#endif
 517#ifdef CONFIG_TRACING
 518        {
 519                .ctl_name       = CTL_UNNUMBERED,
 520                .procname       = "ftrace_dump_on_oops",
 521                .data           = &ftrace_dump_on_oops,
 522                .maxlen         = sizeof(int),
 523                .mode           = 0644,
 524                .proc_handler   = &proc_dointvec,
 525        },
 526#endif
 527#ifdef CONFIG_MODULES
 528        {
 529                .ctl_name       = KERN_MODPROBE,
 530                .procname       = "modprobe",
 531                .data           = &modprobe_path,
 532                .maxlen         = KMOD_PATH_LEN,
 533                .mode           = 0644,
 534                .proc_handler   = &proc_dostring,
 535                .strategy       = &sysctl_string,
 536        },
 537#endif
 538#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
 539        {
 540                .ctl_name       = KERN_HOTPLUG,
 541                .procname       = "hotplug",
 542                .data           = &uevent_helper,
 543                .maxlen         = UEVENT_HELPER_PATH_LEN,
 544                .mode           = 0644,
 545                .proc_handler   = &proc_dostring,
 546                .strategy       = &sysctl_string,
 547        },
 548#endif
 549#ifdef CONFIG_CHR_DEV_SG
 550        {
 551                .ctl_name       = KERN_SG_BIG_BUFF,
 552                .procname       = "sg-big-buff",
 553                .data           = &sg_big_buff,
 554                .maxlen         = sizeof (int),
 555                .mode           = 0444,
 556                .proc_handler   = &proc_dointvec,
 557        },
 558#endif
 559#ifdef CONFIG_BSD_PROCESS_ACCT
 560        {
 561                .ctl_name       = KERN_ACCT,
 562                .procname       = "acct",
 563                .data           = &acct_parm,
 564                .maxlen         = 3*sizeof(int),
 565                .mode           = 0644,
 566                .proc_handler   = &proc_dointvec,
 567        },
 568#endif
 569#ifdef CONFIG_MAGIC_SYSRQ
 570        {
 571                .ctl_name       = KERN_SYSRQ,
 572                .procname       = "sysrq",
 573                .data           = &__sysrq_enabled,
 574                .maxlen         = sizeof (int),
 575                .mode           = 0644,
 576                .proc_handler   = &proc_dointvec,
 577        },
 578#endif
 579#ifdef CONFIG_PROC_SYSCTL
 580        {
 581                .procname       = "cad_pid",
 582                .data           = NULL,
 583                .maxlen         = sizeof (int),
 584                .mode           = 0600,
 585                .proc_handler   = &proc_do_cad_pid,
 586        },
 587#endif
 588        {
 589                .ctl_name       = KERN_MAX_THREADS,
 590                .procname       = "threads-max",
 591                .data           = &max_threads,
 592                .maxlen         = sizeof(int),
 593                .mode           = 0644,
 594                .proc_handler   = &proc_dointvec,
 595        },
 596        {
 597                .ctl_name       = KERN_RANDOM,
 598                .procname       = "random",
 599                .mode           = 0555,
 600                .child          = random_table,
 601        },
 602        {
 603                .ctl_name       = KERN_OVERFLOWUID,
 604                .procname       = "overflowuid",
 605                .data           = &overflowuid,
 606                .maxlen         = sizeof(int),
 607                .mode           = 0644,
 608                .proc_handler   = &proc_dointvec_minmax,
 609                .strategy       = &sysctl_intvec,
 610                .extra1         = &minolduid,
 611                .extra2         = &maxolduid,
 612        },
 613        {
 614                .ctl_name       = KERN_OVERFLOWGID,
 615                .procname       = "overflowgid",
 616                .data           = &overflowgid,
 617                .maxlen         = sizeof(int),
 618                .mode           = 0644,
 619                .proc_handler   = &proc_dointvec_minmax,
 620                .strategy       = &sysctl_intvec,
 621                .extra1         = &minolduid,
 622                .extra2         = &maxolduid,
 623        },
 624#ifdef CONFIG_S390
 625#ifdef CONFIG_MATHEMU
 626        {
 627                .ctl_name       = KERN_IEEE_EMULATION_WARNINGS,
 628                .procname       = "ieee_emulation_warnings",
 629                .data           = &sysctl_ieee_emulation_warnings,
 630                .maxlen         = sizeof(int),
 631                .mode           = 0644,
 632                .proc_handler   = &proc_dointvec,
 633        },
 634#endif
 635        {
 636                .ctl_name       = KERN_S390_USER_DEBUG_LOGGING,
 637                .procname       = "userprocess_debug",
 638                .data           = &sysctl_userprocess_debug,
 639                .maxlen         = sizeof(int),
 640                .mode           = 0644,
 641                .proc_handler   = &proc_dointvec,
 642        },
 643#endif
 644        {
 645                .ctl_name       = KERN_PIDMAX,
 646                .procname       = "pid_max",
 647                .data           = &pid_max,
 648                .maxlen         = sizeof (int),
 649                .mode           = 0644,
 650                .proc_handler   = &proc_dointvec_minmax,
 651                .strategy       = sysctl_intvec,
 652                .extra1         = &pid_max_min,
 653                .extra2         = &pid_max_max,
 654        },
 655        {
 656                .ctl_name       = KERN_PANIC_ON_OOPS,
 657                .procname       = "panic_on_oops",
 658                .data           = &panic_on_oops,
 659                .maxlen         = sizeof(int),
 660                .mode           = 0644,
 661                .proc_handler   = &proc_dointvec,
 662        },
 663#if defined CONFIG_PRINTK
 664        {
 665                .ctl_name       = KERN_PRINTK,
 666                .procname       = "printk",
 667                .data           = &console_loglevel,
 668                .maxlen         = 4*sizeof(int),
 669                .mode           = 0644,
 670                .proc_handler   = &proc_dointvec,
 671        },
 672        {
 673                .ctl_name       = KERN_PRINTK_RATELIMIT,
 674                .procname       = "printk_ratelimit",
 675                .data           = &printk_ratelimit_state.interval,
 676                .maxlen         = sizeof(int),
 677                .mode           = 0644,
 678                .proc_handler   = &proc_dointvec_jiffies,
 679                .strategy       = &sysctl_jiffies,
 680        },
 681        {
 682                .ctl_name       = KERN_PRINTK_RATELIMIT_BURST,
 683                .procname       = "printk_ratelimit_burst",
 684                .data           = &printk_ratelimit_state.burst,
 685                .maxlen         = sizeof(int),
 686                .mode           = 0644,
 687                .proc_handler   = &proc_dointvec,
 688        },
 689#endif
 690        {
 691                .ctl_name       = KERN_NGROUPS_MAX,
 692                .procname       = "ngroups_max",
 693                .data           = &ngroups_max,
 694                .maxlen         = sizeof (int),
 695                .mode           = 0444,
 696                .proc_handler   = &proc_dointvec,
 697        },
 698#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 699        {
 700                .ctl_name       = KERN_UNKNOWN_NMI_PANIC,
 701                .procname       = "unknown_nmi_panic",
 702                .data           = &unknown_nmi_panic,
 703                .maxlen         = sizeof (int),
 704                .mode           = 0644,
 705                .proc_handler   = &proc_dointvec,
 706        },
 707        {
 708                .procname       = "nmi_watchdog",
 709                .data           = &nmi_watchdog_enabled,
 710                .maxlen         = sizeof (int),
 711                .mode           = 0644,
 712                .proc_handler   = &proc_nmi_enabled,
 713        },
 714#endif
 715#if defined(CONFIG_X86)
 716        {
 717                .ctl_name       = KERN_PANIC_ON_NMI,
 718                .procname       = "panic_on_unrecovered_nmi",
 719                .data           = &panic_on_unrecovered_nmi,
 720                .maxlen         = sizeof(int),
 721                .mode           = 0644,
 722                .proc_handler   = &proc_dointvec,
 723        },
 724        {
 725                .ctl_name       = KERN_BOOTLOADER_TYPE,
 726                .procname       = "bootloader_type",
 727                .data           = &bootloader_type,
 728                .maxlen         = sizeof (int),
 729                .mode           = 0444,
 730                .proc_handler   = &proc_dointvec,
 731        },
 732        {
 733                .ctl_name       = CTL_UNNUMBERED,
 734                .procname       = "kstack_depth_to_print",
 735                .data           = &kstack_depth_to_print,
 736                .maxlen         = sizeof(int),
 737                .mode           = 0644,
 738                .proc_handler   = &proc_dointvec,
 739        },
 740        {
 741                .ctl_name       = CTL_UNNUMBERED,
 742                .procname       = "io_delay_type",
 743                .data           = &io_delay_type,
 744                .maxlen         = sizeof(int),
 745                .mode           = 0644,
 746                .proc_handler   = &proc_dointvec,
 747        },
 748#endif
 749#if defined(CONFIG_MMU)
 750        {
 751                .ctl_name       = KERN_RANDOMIZE,
 752                .procname       = "randomize_va_space",
 753                .data           = &randomize_va_space,
 754                .maxlen         = sizeof(int),
 755                .mode           = 0644,
 756                .proc_handler   = &proc_dointvec,
 757        },
 758#endif
 759#if defined(CONFIG_S390) && defined(CONFIG_SMP)
 760        {
 761                .ctl_name       = KERN_SPIN_RETRY,
 762                .procname       = "spin_retry",
 763                .data           = &spin_retry,
 764                .maxlen         = sizeof (int),
 765                .mode           = 0644,
 766                .proc_handler   = &proc_dointvec,
 767        },
 768#endif
 769#if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
 770        {
 771                .procname       = "acpi_video_flags",
 772                .data           = &acpi_realmode_flags,
 773                .maxlen         = sizeof (unsigned long),
 774                .mode           = 0644,
 775                .proc_handler   = &proc_doulongvec_minmax,
 776        },
 777#endif
 778#ifdef CONFIG_IA64
 779        {
 780                .ctl_name       = KERN_IA64_UNALIGNED,
 781                .procname       = "ignore-unaligned-usertrap",
 782                .data           = &no_unaligned_warning,
 783                .maxlen         = sizeof (int),
 784                .mode           = 0644,
 785                .proc_handler   = &proc_dointvec,
 786        },
 787        {
 788                .ctl_name       = CTL_UNNUMBERED,
 789                .procname       = "unaligned-dump-stack",
 790                .data           = &unaligned_dump_stack,
 791                .maxlen         = sizeof (int),
 792                .mode           = 0644,
 793                .proc_handler   = &proc_dointvec,
 794        },
 795#endif
 796#ifdef CONFIG_DETECT_SOFTLOCKUP
 797        {
 798                .ctl_name       = CTL_UNNUMBERED,
 799                .procname       = "softlockup_panic",
 800                .data           = &softlockup_panic,
 801                .maxlen         = sizeof(int),
 802                .mode           = 0644,
 803                .proc_handler   = &proc_dointvec_minmax,
 804                .strategy       = &sysctl_intvec,
 805                .extra1         = &zero,
 806                .extra2         = &one,
 807        },
 808        {
 809                .ctl_name       = CTL_UNNUMBERED,
 810                .procname       = "softlockup_thresh",
 811                .data           = &softlockup_thresh,
 812                .maxlen         = sizeof(int),
 813                .mode           = 0644,
 814                .proc_handler   = &proc_dosoftlockup_thresh,
 815                .strategy       = &sysctl_intvec,
 816                .extra1         = &neg_one,
 817                .extra2         = &sixty,
 818        },
 819#endif
 820#ifdef CONFIG_DETECT_HUNG_TASK
 821        {
 822                .ctl_name       = CTL_UNNUMBERED,
 823                .procname       = "hung_task_panic",
 824                .data           = &sysctl_hung_task_panic,
 825                .maxlen         = sizeof(int),
 826                .mode           = 0644,
 827                .proc_handler   = &proc_dointvec_minmax,
 828                .strategy       = &sysctl_intvec,
 829                .extra1         = &zero,
 830                .extra2         = &one,
 831        },
 832        {
 833                .ctl_name       = CTL_UNNUMBERED,
 834                .procname       = "hung_task_check_count",
 835                .data           = &sysctl_hung_task_check_count,
 836                .maxlen         = sizeof(unsigned long),
 837                .mode           = 0644,
 838                .proc_handler   = &proc_doulongvec_minmax,
 839                .strategy       = &sysctl_intvec,
 840        },
 841        {
 842                .ctl_name       = CTL_UNNUMBERED,
 843                .procname       = "hung_task_timeout_secs",
 844                .data           = &sysctl_hung_task_timeout_secs,
 845                .maxlen         = sizeof(unsigned long),
 846                .mode           = 0644,
 847                .proc_handler   = &proc_dohung_task_timeout_secs,
 848                .strategy       = &sysctl_intvec,
 849        },
 850        {
 851                .ctl_name       = CTL_UNNUMBERED,
 852                .procname       = "hung_task_warnings",
 853                .data           = &sysctl_hung_task_warnings,
 854                .maxlen         = sizeof(unsigned long),
 855                .mode           = 0644,
 856                .proc_handler   = &proc_doulongvec_minmax,
 857                .strategy       = &sysctl_intvec,
 858        },
 859#endif
 860#ifdef CONFIG_COMPAT
 861        {
 862                .ctl_name       = KERN_COMPAT_LOG,
 863                .procname       = "compat-log",
 864                .data           = &compat_log,
 865                .maxlen         = sizeof (int),
 866                .mode           = 0644,
 867                .proc_handler   = &proc_dointvec,
 868        },
 869#endif
 870#ifdef CONFIG_RT_MUTEXES
 871        {
 872                .ctl_name       = KERN_MAX_LOCK_DEPTH,
 873                .procname       = "max_lock_depth",
 874                .data           = &max_lock_depth,
 875                .maxlen         = sizeof(int),
 876                .mode           = 0644,
 877                .proc_handler   = &proc_dointvec,
 878        },
 879#endif
 880        {
 881                .ctl_name       = CTL_UNNUMBERED,
 882                .procname       = "poweroff_cmd",
 883                .data           = &poweroff_cmd,
 884                .maxlen         = POWEROFF_CMD_PATH_LEN,
 885                .mode           = 0644,
 886                .proc_handler   = &proc_dostring,
 887                .strategy       = &sysctl_string,
 888        },
 889#ifdef CONFIG_KEYS
 890        {
 891                .ctl_name       = CTL_UNNUMBERED,
 892                .procname       = "keys",
 893                .mode           = 0555,
 894                .child          = key_sysctls,
 895        },
 896#endif
 897#ifdef CONFIG_RCU_TORTURE_TEST
 898        {
 899                .ctl_name       = CTL_UNNUMBERED,
 900                .procname       = "rcutorture_runnable",
 901                .data           = &rcutorture_runnable,
 902                .maxlen         = sizeof(int),
 903                .mode           = 0644,
 904                .proc_handler   = &proc_dointvec,
 905        },
 906#endif
 907#ifdef CONFIG_SLOW_WORK
 908        {
 909                .ctl_name       = CTL_UNNUMBERED,
 910                .procname       = "slow-work",
 911                .mode           = 0555,
 912                .child          = slow_work_sysctls,
 913        },
 914#endif
 915/*
 916 * NOTE: do not add new entries to this table unless you have read
 917 * Documentation/sysctl/ctl_unnumbered.txt
 918 */
 919        { .ctl_name = 0 }
 920};
 921
 922static struct ctl_table vm_table[] = {
 923        {
 924                .ctl_name       = VM_OVERCOMMIT_MEMORY,
 925                .procname       = "overcommit_memory",
 926                .data           = &sysctl_overcommit_memory,
 927                .maxlen         = sizeof(sysctl_overcommit_memory),
 928                .mode           = 0644,
 929                .proc_handler   = &proc_dointvec,
 930        },
 931        {
 932                .ctl_name       = VM_PANIC_ON_OOM,
 933                .procname       = "panic_on_oom",
 934                .data           = &sysctl_panic_on_oom,
 935                .maxlen         = sizeof(sysctl_panic_on_oom),
 936                .mode           = 0644,
 937                .proc_handler   = &proc_dointvec,
 938        },
 939        {
 940                .ctl_name       = CTL_UNNUMBERED,
 941                .procname       = "oom_kill_allocating_task",
 942                .data           = &sysctl_oom_kill_allocating_task,
 943                .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
 944                .mode           = 0644,
 945                .proc_handler   = &proc_dointvec,
 946        },
 947        {
 948                .ctl_name       = CTL_UNNUMBERED,
 949                .procname       = "oom_dump_tasks",
 950                .data           = &sysctl_oom_dump_tasks,
 951                .maxlen         = sizeof(sysctl_oom_dump_tasks),
 952                .mode           = 0644,
 953                .proc_handler   = &proc_dointvec,
 954        },
 955        {
 956                .ctl_name       = VM_OVERCOMMIT_RATIO,
 957                .procname       = "overcommit_ratio",
 958                .data           = &sysctl_overcommit_ratio,
 959                .maxlen         = sizeof(sysctl_overcommit_ratio),
 960                .mode           = 0644,
 961                .proc_handler   = &proc_dointvec,
 962        },
 963        {
 964                .ctl_name       = VM_PAGE_CLUSTER,
 965                .procname       = "page-cluster", 
 966                .data           = &page_cluster,
 967                .maxlen         = sizeof(int),
 968                .mode           = 0644,
 969                .proc_handler   = &proc_dointvec,
 970        },
 971        {
 972                .ctl_name       = VM_DIRTY_BACKGROUND,
 973                .procname       = "dirty_background_ratio",
 974                .data           = &dirty_background_ratio,
 975                .maxlen         = sizeof(dirty_background_ratio),
 976                .mode           = 0644,
 977                .proc_handler   = &dirty_background_ratio_handler,
 978                .strategy       = &sysctl_intvec,
 979                .extra1         = &zero,
 980                .extra2         = &one_hundred,
 981        },
 982        {
 983                .ctl_name       = CTL_UNNUMBERED,
 984                .procname       = "dirty_background_bytes",
 985                .data           = &dirty_background_bytes,
 986                .maxlen         = sizeof(dirty_background_bytes),
 987                .mode           = 0644,
 988                .proc_handler   = &dirty_background_bytes_handler,
 989                .strategy       = &sysctl_intvec,
 990                .extra1         = &one_ul,
 991        },
 992        {
 993                .ctl_name       = VM_DIRTY_RATIO,
 994                .procname       = "dirty_ratio",
 995                .data           = &vm_dirty_ratio,
 996                .maxlen         = sizeof(vm_dirty_ratio),
 997                .mode           = 0644,
 998                .proc_handler   = &dirty_ratio_handler,
 999                .strategy       = &sysctl_intvec,
1000                .extra1         = &zero,
1001                .extra2         = &one_hundred,
1002        },
1003        {
1004                .ctl_name       = CTL_UNNUMBERED,
1005                .procname       = "dirty_bytes",
1006                .data           = &vm_dirty_bytes,
1007                .maxlen         = sizeof(vm_dirty_bytes),
1008                .mode           = 0644,
1009                .proc_handler   = &dirty_bytes_handler,
1010                .strategy       = &sysctl_intvec,
1011                .extra1         = &dirty_bytes_min,
1012        },
1013        {
1014                .procname       = "dirty_writeback_centisecs",
1015                .data           = &dirty_writeback_interval,
1016                .maxlen         = sizeof(dirty_writeback_interval),
1017                .mode           = 0644,
1018                .proc_handler   = &dirty_writeback_centisecs_handler,
1019        },
1020        {
1021                .procname       = "dirty_expire_centisecs",
1022                .data           = &dirty_expire_interval,
1023                .maxlen         = sizeof(dirty_expire_interval),
1024                .mode           = 0644,
1025                .proc_handler   = &proc_dointvec,
1026        },
1027        {
1028                .ctl_name       = VM_NR_PDFLUSH_THREADS,
1029                .procname       = "nr_pdflush_threads",
1030                .data           = &nr_pdflush_threads,
1031                .maxlen         = sizeof nr_pdflush_threads,
1032                .mode           = 0444 /* read-only*/,
1033                .proc_handler   = &proc_dointvec,
1034        },
1035        {
1036                .ctl_name       = VM_SWAPPINESS,
1037                .procname       = "swappiness",
1038                .data           = &vm_swappiness,
1039                .maxlen         = sizeof(vm_swappiness),
1040                .mode           = 0644,
1041                .proc_handler   = &proc_dointvec_minmax,
1042                .strategy       = &sysctl_intvec,
1043                .extra1         = &zero,
1044                .extra2         = &one_hundred,
1045        },
1046#ifdef CONFIG_HUGETLB_PAGE
1047         {
1048                .procname       = "nr_hugepages",
1049                .data           = NULL,
1050                .maxlen         = sizeof(unsigned long),
1051                .mode           = 0644,
1052                .proc_handler   = &hugetlb_sysctl_handler,
1053                .extra1         = (void *)&hugetlb_zero,
1054                .extra2         = (void *)&hugetlb_infinity,
1055         },
1056         {
1057                .ctl_name       = VM_HUGETLB_GROUP,
1058                .procname       = "hugetlb_shm_group",
1059                .data           = &sysctl_hugetlb_shm_group,
1060                .maxlen         = sizeof(gid_t),
1061                .mode           = 0644,
1062                .proc_handler   = &proc_dointvec,
1063         },
1064         {
1065                .ctl_name       = CTL_UNNUMBERED,
1066                .procname       = "hugepages_treat_as_movable",
1067                .data           = &hugepages_treat_as_movable,
1068                .maxlen         = sizeof(int),
1069                .mode           = 0644,
1070                .proc_handler   = &hugetlb_treat_movable_handler,
1071        },
1072        {
1073                .ctl_name       = CTL_UNNUMBERED,
1074                .procname       = "nr_overcommit_hugepages",
1075                .data           = NULL,
1076                .maxlen         = sizeof(unsigned long),
1077                .mode           = 0644,
1078                .proc_handler   = &hugetlb_overcommit_handler,
1079                .extra1         = (void *)&hugetlb_zero,
1080                .extra2         = (void *)&hugetlb_infinity,
1081        },
1082#endif
1083        {
1084                .ctl_name       = VM_LOWMEM_RESERVE_RATIO,
1085                .procname       = "lowmem_reserve_ratio",
1086                .data           = &sysctl_lowmem_reserve_ratio,
1087                .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1088                .mode           = 0644,
1089                .proc_handler   = &lowmem_reserve_ratio_sysctl_handler,
1090                .strategy       = &sysctl_intvec,
1091        },
1092        {
1093                .ctl_name       = VM_DROP_PAGECACHE,
1094                .procname       = "drop_caches",
1095                .data           = &sysctl_drop_caches,
1096                .maxlen         = sizeof(int),
1097                .mode           = 0644,
1098                .proc_handler   = drop_caches_sysctl_handler,
1099                .strategy       = &sysctl_intvec,
1100        },
1101        {
1102                .ctl_name       = VM_MIN_FREE_KBYTES,
1103                .procname       = "min_free_kbytes",
1104                .data           = &min_free_kbytes,
1105                .maxlen         = sizeof(min_free_kbytes),
1106                .mode           = 0644,
1107                .proc_handler   = &min_free_kbytes_sysctl_handler,
1108                .strategy       = &sysctl_intvec,
1109                .extra1         = &zero,
1110        },
1111        {
1112                .ctl_name       = VM_PERCPU_PAGELIST_FRACTION,
1113                .procname       = "percpu_pagelist_fraction",
1114                .data           = &percpu_pagelist_fraction,
1115                .maxlen         = sizeof(percpu_pagelist_fraction),
1116                .mode           = 0644,
1117                .proc_handler   = &percpu_pagelist_fraction_sysctl_handler,
1118                .strategy       = &sysctl_intvec,
1119                .extra1         = &min_percpu_pagelist_fract,
1120        },
1121#ifdef CONFIG_MMU
1122        {
1123                .ctl_name       = VM_MAX_MAP_COUNT,
1124                .procname       = "max_map_count",
1125                .data           = &sysctl_max_map_count,
1126                .maxlen         = sizeof(sysctl_max_map_count),
1127                .mode           = 0644,
1128                .proc_handler   = &proc_dointvec
1129        },
1130#else
1131        {
1132                .ctl_name       = CTL_UNNUMBERED,
1133                .procname       = "nr_trim_pages",
1134                .data           = &sysctl_nr_trim_pages,
1135                .maxlen         = sizeof(sysctl_nr_trim_pages),
1136                .mode           = 0644,
1137                .proc_handler   = &proc_dointvec_minmax,
1138                .strategy       = &sysctl_intvec,
1139                .extra1         = &zero,
1140        },
1141#endif
1142        {
1143                .ctl_name       = VM_LAPTOP_MODE,
1144                .procname       = "laptop_mode",
1145                .data           = &laptop_mode,
1146                .maxlen         = sizeof(laptop_mode),
1147                .mode           = 0644,
1148                .proc_handler   = &proc_dointvec_jiffies,
1149                .strategy       = &sysctl_jiffies,
1150        },
1151        {
1152                .ctl_name       = VM_BLOCK_DUMP,
1153                .procname       = "block_dump",
1154                .data           = &block_dump,
1155                .maxlen         = sizeof(block_dump),
1156                .mode           = 0644,
1157                .proc_handler   = &proc_dointvec,
1158                .strategy       = &sysctl_intvec,
1159                .extra1         = &zero,
1160        },
1161        {
1162                .ctl_name       = VM_VFS_CACHE_PRESSURE,
1163                .procname       = "vfs_cache_pressure",
1164                .data           = &sysctl_vfs_cache_pressure,
1165                .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1166                .mode           = 0644,
1167                .proc_handler   = &proc_dointvec,
1168                .strategy       = &sysctl_intvec,
1169                .extra1         = &zero,
1170        },
1171#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1172        {
1173                .ctl_name       = VM_LEGACY_VA_LAYOUT,
1174                .procname       = "legacy_va_layout",
1175                .data           = &sysctl_legacy_va_layout,
1176                .maxlen         = sizeof(sysctl_legacy_va_layout),
1177                .mode           = 0644,
1178                .proc_handler   = &proc_dointvec,
1179                .strategy       = &sysctl_intvec,
1180                .extra1         = &zero,
1181        },
1182#endif
1183#ifdef CONFIG_NUMA
1184        {
1185                .ctl_name       = VM_ZONE_RECLAIM_MODE,
1186                .procname       = "zone_reclaim_mode",
1187                .data           = &zone_reclaim_mode,
1188                .maxlen         = sizeof(zone_reclaim_mode),
1189                .mode           = 0644,
1190                .proc_handler   = &proc_dointvec,
1191                .strategy       = &sysctl_intvec,
1192                .extra1         = &zero,
1193        },
1194        {
1195                .ctl_name       = VM_MIN_UNMAPPED,
1196                .procname       = "min_unmapped_ratio",
1197                .data           = &sysctl_min_unmapped_ratio,
1198                .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1199                .mode           = 0644,
1200                .proc_handler   = &sysctl_min_unmapped_ratio_sysctl_handler,
1201                .strategy       = &sysctl_intvec,
1202                .extra1         = &zero,
1203                .extra2         = &one_hundred,
1204        },
1205        {
1206                .ctl_name       = VM_MIN_SLAB,
1207                .procname       = "min_slab_ratio",
1208                .data           = &sysctl_min_slab_ratio,
1209                .maxlen         = sizeof(sysctl_min_slab_ratio),
1210                .mode           = 0644,
1211                .proc_handler   = &sysctl_min_slab_ratio_sysctl_handler,
1212                .strategy       = &sysctl_intvec,
1213                .extra1         = &zero,
1214                .extra2         = &one_hundred,
1215        },
1216#endif
1217#ifdef CONFIG_SMP
1218        {
1219                .ctl_name       = CTL_UNNUMBERED,
1220                .procname       = "stat_interval",
1221                .data           = &sysctl_stat_interval,
1222                .maxlen         = sizeof(sysctl_stat_interval),
1223                .mode           = 0644,
1224                .proc_handler   = &proc_dointvec_jiffies,
1225                .strategy       = &sysctl_jiffies,
1226        },
1227#endif
1228        {
1229                .ctl_name       = CTL_UNNUMBERED,
1230                .procname       = "mmap_min_addr",
1231                .data           = &mmap_min_addr,
1232                .maxlen         = sizeof(unsigned long),
1233                .mode           = 0644,
1234                .proc_handler   = &proc_doulongvec_minmax,
1235        },
1236#ifdef CONFIG_NUMA
1237        {
1238                .ctl_name       = CTL_UNNUMBERED,
1239                .procname       = "numa_zonelist_order",
1240                .data           = &numa_zonelist_order,
1241                .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1242                .mode           = 0644,
1243                .proc_handler   = &numa_zonelist_order_handler,
1244                .strategy       = &sysctl_string,
1245        },
1246#endif
1247#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1248   (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1249        {
1250                .ctl_name       = VM_VDSO_ENABLED,
1251                .procname       = "vdso_enabled",
1252                .data           = &vdso_enabled,
1253                .maxlen         = sizeof(vdso_enabled),
1254                .mode           = 0644,
1255                .proc_handler   = &proc_dointvec,
1256                .strategy       = &sysctl_intvec,
1257                .extra1         = &zero,
1258        },
1259#endif
1260#ifdef CONFIG_HIGHMEM
1261        {
1262                .ctl_name       = CTL_UNNUMBERED,
1263                .procname       = "highmem_is_dirtyable",
1264                .data           = &vm_highmem_is_dirtyable,
1265                .maxlen         = sizeof(vm_highmem_is_dirtyable),
1266                .mode           = 0644,
1267                .proc_handler   = &proc_dointvec_minmax,
1268                .strategy       = &sysctl_intvec,
1269                .extra1         = &zero,
1270                .extra2         = &one,
1271        },
1272#endif
1273#ifdef CONFIG_UNEVICTABLE_LRU
1274        {
1275                .ctl_name       = CTL_UNNUMBERED,
1276                .procname       = "scan_unevictable_pages",
1277                .data           = &scan_unevictable_pages,
1278                .maxlen         = sizeof(scan_unevictable_pages),
1279                .mode           = 0644,
1280                .proc_handler   = &scan_unevictable_handler,
1281        },
1282#endif
1283/*
1284 * NOTE: do not add new entries to this table unless you have read
1285 * Documentation/sysctl/ctl_unnumbered.txt
1286 */
1287        { .ctl_name = 0 }
1288};
1289
1290#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1291static struct ctl_table binfmt_misc_table[] = {
1292        { .ctl_name = 0 }
1293};
1294#endif
1295
1296static struct ctl_table fs_table[] = {
1297        {
1298                .ctl_name       = FS_NRINODE,
1299                .procname       = "inode-nr",
1300                .data           = &inodes_stat,
1301                .maxlen         = 2*sizeof(int),
1302                .mode           = 0444,
1303                .proc_handler   = &proc_dointvec,
1304        },
1305        {
1306                .ctl_name       = FS_STATINODE,
1307                .procname       = "inode-state",
1308                .data           = &inodes_stat,
1309                .maxlen         = 7*sizeof(int),
1310                .mode           = 0444,
1311                .proc_handler   = &proc_dointvec,
1312        },
1313        {
1314                .procname       = "file-nr",
1315                .data           = &files_stat,
1316                .maxlen         = 3*sizeof(int),
1317                .mode           = 0444,
1318                .proc_handler   = &proc_nr_files,
1319        },
1320        {
1321                .ctl_name       = FS_MAXFILE,
1322                .procname       = "file-max",
1323                .data           = &files_stat.max_files,
1324                .maxlen         = sizeof(int),
1325                .mode           = 0644,
1326                .proc_handler   = &proc_dointvec,
1327        },
1328        {
1329                .ctl_name       = CTL_UNNUMBERED,
1330                .procname       = "nr_open",
1331                .data           = &sysctl_nr_open,
1332                .maxlen         = sizeof(int),
1333                .mode           = 0644,
1334                .proc_handler   = &proc_dointvec_minmax,
1335                .extra1         = &sysctl_nr_open_min,
1336                .extra2         = &sysctl_nr_open_max,
1337        },
1338        {
1339                .ctl_name       = FS_DENTRY,
1340                .procname       = "dentry-state",
1341                .data           = &dentry_stat,
1342                .maxlen         = 6*sizeof(int),
1343                .mode           = 0444,
1344                .proc_handler   = &proc_dointvec,
1345        },
1346        {
1347                .ctl_name       = FS_OVERFLOWUID,
1348                .procname       = "overflowuid",
1349                .data           = &fs_overflowuid,
1350                .maxlen         = sizeof(int),
1351                .mode           = 0644,
1352                .proc_handler   = &proc_dointvec_minmax,
1353                .strategy       = &sysctl_intvec,
1354                .extra1         = &minolduid,
1355                .extra2         = &maxolduid,
1356        },
1357        {
1358                .ctl_name       = FS_OVERFLOWGID,
1359                .procname       = "overflowgid",
1360                .data           = &fs_overflowgid,
1361                .maxlen         = sizeof(int),
1362                .mode           = 0644,
1363                .proc_handler   = &proc_dointvec_minmax,
1364                .strategy       = &sysctl_intvec,
1365                .extra1         = &minolduid,
1366                .extra2         = &maxolduid,
1367        },
1368#ifdef CONFIG_FILE_LOCKING
1369        {
1370                .ctl_name       = FS_LEASES,
1371                .procname       = "leases-enable",
1372                .data           = &leases_enable,
1373                .maxlen         = sizeof(int),
1374                .mode           = 0644,
1375                .proc_handler   = &proc_dointvec,
1376        },
1377#endif
1378#ifdef CONFIG_DNOTIFY
1379        {
1380                .ctl_name       = FS_DIR_NOTIFY,
1381                .procname       = "dir-notify-enable",
1382                .data           = &dir_notify_enable,
1383                .maxlen         = sizeof(int),
1384                .mode           = 0644,
1385                .proc_handler   = &proc_dointvec,
1386        },
1387#endif
1388#ifdef CONFIG_MMU
1389#ifdef CONFIG_FILE_LOCKING
1390        {
1391                .ctl_name       = FS_LEASE_TIME,
1392                .procname       = "lease-break-time",
1393                .data           = &lease_break_time,
1394                .maxlen         = sizeof(int),
1395                .mode           = 0644,
1396                .proc_handler   = &proc_dointvec,
1397        },
1398#endif
1399#ifdef CONFIG_AIO
1400        {
1401                .procname       = "aio-nr",
1402                .data           = &aio_nr,
1403                .maxlen         = sizeof(aio_nr),
1404                .mode           = 0444,
1405                .proc_handler   = &proc_doulongvec_minmax,
1406        },
1407        {
1408                .procname       = "aio-max-nr",
1409                .data           = &aio_max_nr,
1410                .maxlen         = sizeof(aio_max_nr),
1411                .mode           = 0644,
1412                .proc_handler   = &proc_doulongvec_minmax,
1413        },
1414#endif /* CONFIG_AIO */
1415#ifdef CONFIG_INOTIFY_USER
1416        {
1417                .ctl_name       = FS_INOTIFY,
1418                .procname       = "inotify",
1419                .mode           = 0555,
1420                .child          = inotify_table,
1421        },
1422#endif  
1423#ifdef CONFIG_EPOLL
1424        {
1425                .procname       = "epoll",
1426                .mode           = 0555,
1427                .child          = epoll_table,
1428        },
1429#endif
1430#endif
1431        {
1432                .ctl_name       = KERN_SETUID_DUMPABLE,
1433                .procname       = "suid_dumpable",
1434                .data           = &suid_dumpable,
1435                .maxlen         = sizeof(int),
1436                .mode           = 0644,
1437                .proc_handler   = &proc_dointvec_minmax,
1438                .strategy       = &sysctl_intvec,
1439                .extra1         = &zero,
1440                .extra2         = &two,
1441        },
1442#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1443        {
1444                .ctl_name       = CTL_UNNUMBERED,
1445                .procname       = "binfmt_misc",
1446                .mode           = 0555,
1447                .child          = binfmt_misc_table,
1448        },
1449#endif
1450/*
1451 * NOTE: do not add new entries to this table unless you have read
1452 * Documentation/sysctl/ctl_unnumbered.txt
1453 */
1454        { .ctl_name = 0 }
1455};
1456
1457static struct ctl_table debug_table[] = {
1458#if defined(CONFIG_X86) || defined(CONFIG_PPC)
1459        {
1460                .ctl_name       = CTL_UNNUMBERED,
1461                .procname       = "exception-trace",
1462                .data           = &show_unhandled_signals,
1463                .maxlen         = sizeof(int),
1464                .mode           = 0644,
1465                .proc_handler   = proc_dointvec
1466        },
1467#endif
1468        { .ctl_name = 0 }
1469};
1470
1471static struct ctl_table dev_table[] = {
1472        { .ctl_name = 0 }
1473};
1474
1475static DEFINE_SPINLOCK(sysctl_lock);
1476
1477/* called under sysctl_lock */
1478static int use_table(struct ctl_table_header *p)
1479{
1480        if (unlikely(p->unregistering))
1481                return 0;
1482        p->used++;
1483        return 1;
1484}
1485
1486/* called under sysctl_lock */
1487static void unuse_table(struct ctl_table_header *p)
1488{
1489        if (!--p->used)
1490                if (unlikely(p->unregistering))
1491                        complete(p->unregistering);
1492}
1493
1494/* called under sysctl_lock, will reacquire if has to wait */
1495static void start_unregistering(struct ctl_table_header *p)
1496{
1497        /*
1498         * if p->used is 0, nobody will ever touch that entry again;
1499         * we'll eliminate all paths to it before dropping sysctl_lock
1500         */
1501        if (unlikely(p->used)) {
1502                struct completion wait;
1503                init_completion(&wait);
1504                p->unregistering = &wait;
1505                spin_unlock(&sysctl_lock);
1506                wait_for_completion(&wait);
1507                spin_lock(&sysctl_lock);
1508        } else {
1509                /* anything non-NULL; we'll never dereference it */
1510                p->unregistering = ERR_PTR(-EINVAL);
1511        }
1512        /*
1513         * do not remove from the list until nobody holds it; walking the
1514         * list in do_sysctl() relies on that.
1515         */
1516        list_del_init(&p->ctl_entry);
1517}
1518
1519void sysctl_head_get(struct ctl_table_header *head)
1520{
1521        spin_lock(&sysctl_lock);
1522        head->count++;
1523        spin_unlock(&sysctl_lock);
1524}
1525
1526void sysctl_head_put(struct ctl_table_header *head)
1527{
1528        spin_lock(&sysctl_lock);
1529        if (!--head->count)
1530                kfree(head);
1531        spin_unlock(&sysctl_lock);
1532}
1533
1534struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1535{
1536        if (!head)
1537                BUG();
1538        spin_lock(&sysctl_lock);
1539        if (!use_table(head))
1540                head = ERR_PTR(-ENOENT);
1541        spin_unlock(&sysctl_lock);
1542        return head;
1543}
1544
1545void sysctl_head_finish(struct ctl_table_header *head)
1546{
1547        if (!head)
1548                return;
1549        spin_lock(&sysctl_lock);
1550        unuse_table(head);
1551        spin_unlock(&sysctl_lock);
1552}
1553
1554static struct ctl_table_set *
1555lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1556{
1557        struct ctl_table_set *set = &root->default_set;
1558        if (root->lookup)
1559                set = root->lookup(root, namespaces);
1560        return set;
1561}
1562
1563static struct list_head *
1564lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1565{
1566        struct ctl_table_set *set = lookup_header_set(root, namespaces);
1567        return &set->list;
1568}
1569
1570struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1571                                            struct ctl_table_header *prev)
1572{
1573        struct ctl_table_root *root;
1574        struct list_head *header_list;
1575        struct ctl_table_header *head;
1576        struct list_head *tmp;
1577
1578        spin_lock(&sysctl_lock);
1579        if (prev) {
1580                head = prev;
1581                tmp = &prev->ctl_entry;
1582                unuse_table(prev);
1583                goto next;
1584        }
1585        tmp = &root_table_header.ctl_entry;
1586        for (;;) {
1587                head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1588
1589                if (!use_table(head))
1590                        goto next;
1591                spin_unlock(&sysctl_lock);
1592                return head;
1593        next:
1594                root = head->root;
1595                tmp = tmp->next;
1596                header_list = lookup_header_list(root, namespaces);
1597                if (tmp != header_list)
1598                        continue;
1599
1600                do {
1601                        root = list_entry(root->root_list.next,
1602                                        struct ctl_table_root, root_list);
1603                        if (root == &sysctl_table_root)
1604                                goto out;
1605                        header_list = lookup_header_list(root, namespaces);
1606                } while (list_empty(header_list));
1607                tmp = header_list->next;
1608        }
1609out:
1610        spin_unlock(&sysctl_lock);
1611        return NULL;
1612}
1613
1614struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1615{
1616        return __sysctl_head_next(current->nsproxy, prev);
1617}
1618
1619void register_sysctl_root(struct ctl_table_root *root)
1620{
1621        spin_lock(&sysctl_lock);
1622        list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1623        spin_unlock(&sysctl_lock);
1624}
1625
1626#ifdef CONFIG_SYSCTL_SYSCALL
1627/* Perform the actual read/write of a sysctl table entry. */
1628static int do_sysctl_strategy(struct ctl_table_root *root,
1629                        struct ctl_table *table,
1630                        void __user *oldval, size_t __user *oldlenp,
1631                        void __user *newval, size_t newlen)
1632{
1633        int op = 0, rc;
1634
1635        if (oldval)
1636                op |= MAY_READ;
1637        if (newval)
1638                op |= MAY_WRITE;
1639        if (sysctl_perm(root, table, op))
1640                return -EPERM;
1641
1642        if (table->strategy) {
1643                rc = table->strategy(table, oldval, oldlenp, newval, newlen);
1644                if (rc < 0)
1645                        return rc;
1646                if (rc > 0)
1647                        return 0;
1648        }
1649
1650        /* If there is no strategy routine, or if the strategy returns
1651         * zero, proceed with automatic r/w */
1652        if (table->data && table->maxlen) {
1653                rc = sysctl_data(table, oldval, oldlenp, newval, newlen);
1654                if (rc < 0)
1655                        return rc;
1656        }
1657        return 0;
1658}
1659
1660static int parse_table(int __user *name, int nlen,
1661                       void __user *oldval, size_t __user *oldlenp,
1662                       void __user *newval, size_t newlen,
1663                       struct ctl_table_root *root,
1664                       struct ctl_table *table)
1665{
1666        int n;
1667repeat:
1668        if (!nlen)
1669                return -ENOTDIR;
1670        if (get_user(n, name))
1671                return -EFAULT;
1672        for ( ; table->ctl_name || table->procname; table++) {
1673                if (!table->ctl_name)
1674                        continue;
1675                if (n == table->ctl_name) {
1676                        int error;
1677                        if (table->child) {
1678                                if (sysctl_perm(root, table, MAY_EXEC))
1679                                        return -EPERM;
1680                                name++;
1681                                nlen--;
1682                                table = table->child;
1683                                goto repeat;
1684                        }
1685                        error = do_sysctl_strategy(root, table,
1686                                                   oldval, oldlenp,
1687                                                   newval, newlen);
1688                        return error;
1689                }
1690        }
1691        return -ENOTDIR;
1692}
1693
1694int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1695               void __user *newval, size_t newlen)
1696{
1697        struct ctl_table_header *head;
1698        int error = -ENOTDIR;
1699
1700        if (nlen <= 0 || nlen >= CTL_MAXNAME)
1701                return -ENOTDIR;
1702        if (oldval) {
1703                int old_len;
1704                if (!oldlenp || get_user(old_len, oldlenp))
1705                        return -EFAULT;
1706        }
1707
1708        for (head = sysctl_head_next(NULL); head;
1709                        head = sysctl_head_next(head)) {
1710                error = parse_table(name, nlen, oldval, oldlenp, 
1711                                        newval, newlen,
1712                                        head->root, head->ctl_table);
1713                if (error != -ENOTDIR) {
1714                        sysctl_head_finish(head);
1715                        break;
1716                }
1717        }
1718        return error;
1719}
1720
1721SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
1722{
1723        struct __sysctl_args tmp;
1724        int error;
1725
1726        if (copy_from_user(&tmp, args, sizeof(tmp)))
1727                return -EFAULT;
1728
1729        error = deprecated_sysctl_warning(&tmp);
1730        if (error)
1731                goto out;
1732
1733        lock_kernel();
1734        error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1735                          tmp.newval, tmp.newlen);
1736        unlock_kernel();
1737out:
1738        return error;
1739}
1740#endif /* CONFIG_SYSCTL_SYSCALL */
1741
1742/*
1743 * sysctl_perm does NOT grant the superuser all rights automatically, because
1744 * some sysctl variables are readonly even to root.
1745 */
1746
1747static int test_perm(int mode, int op)
1748{
1749        if (!current_euid())
1750                mode >>= 6;
1751        else if (in_egroup_p(0))
1752                mode >>= 3;
1753        if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1754                return 0;
1755        return -EACCES;
1756}
1757
1758int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1759{
1760        int error;
1761        int mode;
1762
1763        error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1764        if (error)
1765                return error;
1766
1767        if (root->permissions)
1768                mode = root->permissions(root, current->nsproxy, table);
1769        else
1770                mode = table->mode;
1771
1772        return test_perm(mode, op);
1773}
1774
1775static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1776{
1777        for (; table->ctl_name || table->procname; table++) {
1778                table->parent = parent;
1779                if (table->child)
1780                        sysctl_set_parent(table, table->child);
1781        }
1782}
1783
1784static __init int sysctl_init(void)
1785{
1786        sysctl_set_parent(NULL, root_table);
1787#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1788        {
1789                int err;
1790                err = sysctl_check_table(current->nsproxy, root_table);
1791        }
1792#endif
1793        return 0;
1794}
1795
1796core_initcall(sysctl_init);
1797
1798static struct ctl_table *is_branch_in(struct ctl_table *branch,
1799                                      struct ctl_table *table)
1800{
1801        struct ctl_table *p;
1802        const char *s = branch->procname;
1803
1804        /* branch should have named subdirectory as its first element */
1805        if (!s || !branch->child)
1806                return NULL;
1807
1808        /* ... and nothing else */
1809        if (branch[1].procname || branch[1].ctl_name)
1810                return NULL;
1811
1812        /* table should contain subdirectory with the same name */
1813        for (p = table; p->procname || p->ctl_name; p++) {
1814                if (!p->child)
1815                        continue;
1816                if (p->procname && strcmp(p->procname, s) == 0)
1817                        return p;
1818        }
1819        return NULL;
1820}
1821
1822/* see if attaching q to p would be an improvement */
1823static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1824{
1825        struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1826        struct ctl_table *next;
1827        int is_better = 0;
1828        int not_in_parent = !p->attached_by;
1829
1830        while ((next = is_branch_in(by, to)) != NULL) {
1831                if (by == q->attached_by)
1832                        is_better = 1;
1833                if (to == p->attached_by)
1834                        not_in_parent = 1;
1835                by = by->child;
1836                to = next->child;
1837        }
1838
1839        if (is_better && not_in_parent) {
1840                q->attached_by = by;
1841                q->attached_to = to;
1842                q->parent = p;
1843        }
1844}
1845
1846/**
1847 * __register_sysctl_paths - register a sysctl hierarchy
1848 * @root: List of sysctl headers to register on
1849 * @namespaces: Data to compute which lists of sysctl entries are visible
1850 * @path: The path to the directory the sysctl table is in.
1851 * @table: the top-level table structure
1852 *
1853 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1854 * array. A completely 0 filled entry terminates the table.
1855 *
1856 * The members of the &struct ctl_table structure are used as follows:
1857 *
1858 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1859 *            must be unique within that level of sysctl
1860 *
1861 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1862 *            enter a sysctl file
1863 *
1864 * data - a pointer to data for use by proc_handler
1865 *
1866 * maxlen - the maximum size in bytes of the data
1867 *
1868 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1869 *
1870 * child - a pointer to the child sysctl table if this entry is a directory, or
1871 *         %NULL.
1872 *
1873 * proc_handler - the text handler routine (described below)
1874 *
1875 * strategy - the strategy routine (described below)
1876 *
1877 * de - for internal use by the sysctl routines
1878 *
1879 * extra1, extra2 - extra pointers usable by the proc handler routines
1880 *
1881 * Leaf nodes in the sysctl tree will be represented by a single file
1882 * under /proc; non-leaf nodes will be represented by directories.
1883 *
1884 * sysctl(2) can automatically manage read and write requests through
1885 * the sysctl table.  The data and maxlen fields of the ctl_table
1886 * struct enable minimal validation of the values being written to be
1887 * performed, and the mode field allows minimal authentication.
1888 *
1889 * More sophisticated management can be enabled by the provision of a
1890 * strategy routine with the table entry.  This will be called before
1891 * any automatic read or write of the data is performed.
1892 *
1893 * The strategy routine may return
1894 *
1895 * < 0 - Error occurred (error is passed to user process)
1896 *
1897 * 0   - OK - proceed with automatic read or write.
1898 *
1899 * > 0 - OK - read or write has been done by the strategy routine, so
1900 *       return immediately.
1901 *
1902 * There must be a proc_handler routine for any terminal nodes
1903 * mirrored under /proc/sys (non-terminals are handled by a built-in
1904 * directory handler).  Several default handlers are available to
1905 * cover common cases -
1906 *
1907 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1908 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(), 
1909 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1910 *
1911 * It is the handler's job to read the input buffer from user memory
1912 * and process it. The handler should return 0 on success.
1913 *
1914 * This routine returns %NULL on a failure to register, and a pointer
1915 * to the table header on success.
1916 */
1917struct ctl_table_header *__register_sysctl_paths(
1918        struct ctl_table_root *root,
1919        struct nsproxy *namespaces,
1920        const struct ctl_path *path, struct ctl_table *table)
1921{
1922        struct ctl_table_header *header;
1923        struct ctl_table *new, **prevp;
1924        unsigned int n, npath;
1925        struct ctl_table_set *set;
1926
1927        /* Count the path components */
1928        for (npath = 0; path[npath].ctl_name || path[npath].procname; ++npath)
1929                ;
1930
1931        /*
1932         * For each path component, allocate a 2-element ctl_table array.
1933         * The first array element will be filled with the sysctl entry
1934         * for this, the second will be the sentinel (ctl_name == 0).
1935         *
1936         * We allocate everything in one go so that we don't have to
1937         * worry about freeing additional memory in unregister_sysctl_table.
1938         */
1939        header = kzalloc(sizeof(struct ctl_table_header) +
1940                         (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1941        if (!header)
1942                return NULL;
1943
1944        new = (struct ctl_table *) (header + 1);
1945
1946        /* Now connect the dots */
1947        prevp = &header->ctl_table;
1948        for (n = 0; n < npath; ++n, ++path) {
1949                /* Copy the procname */
1950                new->procname = path->procname;
1951                new->ctl_name = path->ctl_name;
1952                new->mode     = 0555;
1953
1954                *prevp = new;
1955                prevp = &new->child;
1956
1957                new += 2;
1958        }
1959        *prevp = table;
1960        header->ctl_table_arg = table;
1961
1962        INIT_LIST_HEAD(&header->ctl_entry);
1963        header->used = 0;
1964        header->unregistering = NULL;
1965        header->root = root;
1966        sysctl_set_parent(NULL, header->ctl_table);
1967        header->count = 1;
1968#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1969        if (sysctl_check_table(namespaces, header->ctl_table)) {
1970                kfree(header);
1971                return NULL;
1972        }
1973#endif
1974        spin_lock(&sysctl_lock);
1975        header->set = lookup_header_set(root, namespaces);
1976        header->attached_by = header->ctl_table;
1977        header->attached_to = root_table;
1978        header->parent = &root_table_header;
1979        for (set = header->set; set; set = set->parent) {
1980                struct ctl_table_header *p;
1981                list_for_each_entry(p, &set->list, ctl_entry) {
1982                        if (p->unregistering)
1983                                continue;
1984                        try_attach(p, header);
1985                }
1986        }
1987        header->parent->count++;
1988        list_add_tail(&header->ctl_entry, &header->set->list);
1989        spin_unlock(&sysctl_lock);
1990
1991        return header;
1992}
1993
1994/**
1995 * register_sysctl_table_path - register a sysctl table hierarchy
1996 * @path: The path to the directory the sysctl table is in.
1997 * @table: the top-level table structure
1998 *
1999 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2000 * array. A completely 0 filled entry terminates the table.
2001 *
2002 * See __register_sysctl_paths for more details.
2003 */
2004struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2005                                                struct ctl_table *table)
2006{
2007        return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
2008                                        path, table);
2009}
2010
2011/**
2012 * register_sysctl_table - register a sysctl table hierarchy
2013 * @table: the top-level table structure
2014 *
2015 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
2016 * array. A completely 0 filled entry terminates the table.
2017 *
2018 * See register_sysctl_paths for more details.
2019 */
2020struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
2021{
2022        static const struct ctl_path null_path[] = { {} };
2023
2024        return register_sysctl_paths(null_path, table);
2025}
2026
2027/**
2028 * unregister_sysctl_table - unregister a sysctl table hierarchy
2029 * @header: the header returned from register_sysctl_table
2030 *
2031 * Unregisters the sysctl table and all children. proc entries may not
2032 * actually be removed until they are no longer used by anyone.
2033 */
2034void unregister_sysctl_table(struct ctl_table_header * header)
2035{
2036        might_sleep();
2037
2038        if (header == NULL)
2039                return;
2040
2041        spin_lock(&sysctl_lock);
2042        start_unregistering(header);
2043        if (!--header->parent->count) {
2044                WARN_ON(1);
2045                kfree(header->parent);
2046        }
2047        if (!--header->count)
2048                kfree(header);
2049        spin_unlock(&sysctl_lock);
2050}
2051
2052int sysctl_is_seen(struct ctl_table_header *p)
2053{
2054        struct ctl_table_set *set = p->set;
2055        int res;
2056        spin_lock(&sysctl_lock);
2057        if (p->unregistering)
2058                res = 0;
2059        else if (!set->is_seen)
2060                res = 1;
2061        else
2062                res = set->is_seen(set);
2063        spin_unlock(&sysctl_lock);
2064        return res;
2065}
2066
2067void setup_sysctl_set(struct ctl_table_set *p,
2068        struct ctl_table_set *parent,
2069        int (*is_seen)(struct ctl_table_set *))
2070{
2071        INIT_LIST_HEAD(&p->list);
2072        p->parent = parent ? parent : &sysctl_table_root.default_set;
2073        p->is_seen = is_seen;
2074}
2075
2076#else /* !CONFIG_SYSCTL */
2077struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2078{
2079        return NULL;
2080}
2081
2082struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2083                                                    struct ctl_table *table)
2084{
2085        return NULL;
2086}
2087
2088void unregister_sysctl_table(struct ctl_table_header * table)
2089{
2090}
2091
2092void setup_sysctl_set(struct ctl_table_set *p,
2093        struct ctl_table_set *parent,
2094        int (*is_seen)(struct ctl_table_set *))
2095{
2096}
2097
2098void sysctl_head_put(struct ctl_table_header *head)
2099{
2100}
2101
2102#endif /* CONFIG_SYSCTL */
2103
2104/*
2105 * /proc/sys support
2106 */
2107
2108#ifdef CONFIG_PROC_SYSCTL
2109
2110static int _proc_do_string(void* data, int maxlen, int write,
2111                           struct file *filp, void __user *buffer,
2112                           size_t *lenp, loff_t *ppos)
2113{
2114        size_t len;
2115        char __user *p;
2116        char c;
2117
2118        if (!data || !maxlen || !*lenp) {
2119                *lenp = 0;
2120                return 0;
2121        }
2122
2123        if (write) {
2124                len = 0;
2125                p = buffer;
2126                while (len < *lenp) {
2127                        if (get_user(c, p++))
2128                                return -EFAULT;
2129                        if (c == 0 || c == '\n')
2130                                break;
2131                        len++;
2132                }
2133                if (len >= maxlen)
2134                        len = maxlen-1;
2135                if(copy_from_user(data, buffer, len))
2136                        return -EFAULT;
2137                ((char *) data)[len] = 0;
2138                *ppos += *lenp;
2139        } else {
2140                len = strlen(data);
2141                if (len > maxlen)
2142                        len = maxlen;
2143
2144                if (*ppos > len) {
2145                        *lenp = 0;
2146                        return 0;
2147                }
2148
2149                data += *ppos;
2150                len  -= *ppos;
2151
2152                if (len > *lenp)
2153                        len = *lenp;
2154                if (len)
2155                        if(copy_to_user(buffer, data, len))
2156                                return -EFAULT;
2157                if (len < *lenp) {
2158                        if(put_user('\n', ((char __user *) buffer) + len))
2159                                return -EFAULT;
2160                        len++;
2161                }
2162                *lenp = len;
2163                *ppos += len;
2164        }
2165        return 0;
2166}
2167
2168/**
2169 * proc_dostring - read a string sysctl
2170 * @table: the sysctl table
2171 * @write: %TRUE if this is a write to the sysctl file
2172 * @filp: the file structure
2173 * @buffer: the user buffer
2174 * @lenp: the size of the user buffer
2175 * @ppos: file position
2176 *
2177 * Reads/writes a string from/to the user buffer. If the kernel
2178 * buffer provided is not large enough to hold the string, the
2179 * string is truncated. The copied string is %NULL-terminated.
2180 * If the string is being read by the user process, it is copied
2181 * and a newline '\n' is added. It is truncated if the buffer is
2182 * not large enough.
2183 *
2184 * Returns 0 on success.
2185 */
2186int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2187                  void __user *buffer, size_t *lenp, loff_t *ppos)
2188{
2189        return _proc_do_string(table->data, table->maxlen, write, filp,
2190                               buffer, lenp, ppos);
2191}
2192
2193
2194static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2195                                 int *valp,
2196                                 int write, void *data)
2197{
2198        if (write) {
2199                *valp = *negp ? -*lvalp : *lvalp;
2200        } else {
2201                int val = *valp;
2202                if (val < 0) {
2203                        *negp = -1;
2204                        *lvalp = (unsigned long)-val;
2205                } else {
2206                        *negp = 0;
2207                        *lvalp = (unsigned long)val;
2208                }
2209        }
2210        return 0;
2211}
2212
2213static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2214                  int write, struct file *filp, void __user *buffer,
2215                  size_t *lenp, loff_t *ppos,
2216                  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2217                              int write, void *data),
2218                  void *data)
2219{
2220#define TMPBUFLEN 21
2221        int *i, vleft, first=1, neg, val;
2222        unsigned long lval;
2223        size_t left, len;
2224        
2225        char buf[TMPBUFLEN], *p;
2226        char __user *s = buffer;
2227        
2228        if (!tbl_data || !table->maxlen || !*lenp ||
2229            (*ppos && !write)) {
2230                *lenp = 0;
2231                return 0;
2232        }
2233        
2234        i = (int *) tbl_data;
2235        vleft = table->maxlen / sizeof(*i);
2236        left = *lenp;
2237
2238        if (!conv)
2239                conv = do_proc_dointvec_conv;
2240
2241        for (; left && vleft--; i++, first=0) {
2242                if (write) {
2243                        while (left) {
2244                                char c;
2245                                if (get_user(c, s))
2246                                        return -EFAULT;
2247                                if (!isspace(c))
2248                                        break;
2249                                left--;
2250                                s++;
2251                        }
2252                        if (!left)
2253                                break;
2254                        neg = 0;
2255                        len = left;
2256                        if (len > sizeof(buf) - 1)
2257                                len = sizeof(buf) - 1;
2258                        if (copy_from_user(buf, s, len))
2259                                return -EFAULT;
2260                        buf[len] = 0;
2261                        p = buf;
2262                        if (*p == '-' && left > 1) {
2263                                neg = 1;
2264                                p++;
2265                        }
2266                        if (*p < '0' || *p > '9')
2267                                break;
2268
2269                        lval = simple_strtoul(p, &p, 0);
2270
2271                        len = p-buf;
2272                        if ((len < left) && *p && !isspace(*p))
2273                                break;
2274                        if (neg)
2275                                val = -val;
2276                        s += len;
2277                        left -= len;
2278
2279                        if (conv(&neg, &lval, i, 1, data))
2280                                break;
2281                } else {
2282                        p = buf;
2283                        if (!first)
2284                                *p++ = '\t';
2285        
2286                        if (conv(&neg, &lval, i, 0, data))
2287                                break;
2288
2289                        sprintf(p, "%s%lu", neg ? "-" : "", lval);
2290                        len = strlen(buf);
2291                        if (len > left)
2292                                len = left;
2293                        if(copy_to_user(s, buf, len))
2294                                return -EFAULT;
2295                        left -= len;
2296                        s += len;
2297                }
2298        }
2299
2300        if (!write && !first && left) {
2301                if(put_user('\n', s))
2302                        return -EFAULT;
2303                left--, s++;
2304        }
2305        if (write) {
2306                while (left) {
2307                        char c;
2308                        if (get_user(c, s++))
2309                                return -EFAULT;
2310                        if (!isspace(c))
2311                                break;
2312                        left--;
2313                }
2314        }
2315        if (write && first)
2316                return -EINVAL;
2317        *lenp -= left;
2318        *ppos += *lenp;
2319        return 0;
2320#undef TMPBUFLEN
2321}
2322
2323static int do_proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2324                  void __user *buffer, size_t *lenp, loff_t *ppos,
2325                  int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2326                              int write, void *data),
2327                  void *data)
2328{
2329        return __do_proc_dointvec(table->data, table, write, filp,
2330                        buffer, lenp, ppos, conv, data);
2331}
2332
2333/**
2334 * proc_dointvec - read a vector of integers
2335 * @table: the sysctl table
2336 * @write: %TRUE if this is a write to the sysctl file
2337 * @filp: the file structure
2338 * @buffer: the user buffer
2339 * @lenp: the size of the user buffer
2340 * @ppos: file position
2341 *
2342 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2343 * values from/to the user buffer, treated as an ASCII string. 
2344 *
2345 * Returns 0 on success.
2346 */
2347int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2348                     void __user *buffer, size_t *lenp, loff_t *ppos)
2349{
2350    return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2351                            NULL,NULL);
2352}
2353
2354/*
2355 * Taint values can only be increased
2356 * This means we can safely use a temporary.
2357 */
2358static int proc_taint(struct ctl_table *table, int write, struct file *filp,
2359                               void __user *buffer, size_t *lenp, loff_t *ppos)
2360{
2361        struct ctl_table t;
2362        unsigned long tmptaint = get_taint();
2363        int err;
2364
2365        if (write && !capable(CAP_SYS_ADMIN))
2366                return -EPERM;
2367
2368        t = *table;
2369        t.data = &tmptaint;
2370        err = proc_doulongvec_minmax(&t, write, filp, buffer, lenp, ppos);
2371        if (err < 0)
2372                return err;
2373
2374        if (write) {
2375                /*
2376                 * Poor man's atomic or. Not worth adding a primitive
2377                 * to everyone's atomic.h for this
2378                 */
2379                int i;
2380                for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2381                        if ((tmptaint >> i) & 1)
2382                                add_taint(i);
2383                }
2384        }
2385
2386        return err;
2387}
2388
2389struct do_proc_dointvec_minmax_conv_param {
2390        int *min;
2391        int *max;
2392};
2393
2394static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp, 
2395                                        int *valp, 
2396                                        int write, void *data)
2397{
2398        struct do_proc_dointvec_minmax_conv_param *param = data;
2399        if (write) {
2400                int val = *negp ? -*lvalp : *lvalp;
2401                if ((param->min && *param->min > val) ||
2402                    (param->max && *param->max < val))
2403                        return -EINVAL;
2404                *valp = val;
2405        } else {
2406                int val = *valp;
2407                if (val < 0) {
2408                        *negp = -1;
2409                        *lvalp = (unsigned long)-val;
2410                } else {
2411                        *negp = 0;
2412                        *lvalp = (unsigned long)val;
2413                }
2414        }
2415        return 0;
2416}
2417
2418/**
2419 * proc_dointvec_minmax - read a vector of integers with min/max values
2420 * @table: the sysctl table
2421 * @write: %TRUE if this is a write to the sysctl file
2422 * @filp: the file structure
2423 * @buffer: the user buffer
2424 * @lenp: the size of the user buffer
2425 * @ppos: file position
2426 *
2427 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2428 * values from/to the user buffer, treated as an ASCII string.
2429 *
2430 * This routine will ensure the values are within the range specified by
2431 * table->extra1 (min) and table->extra2 (max).
2432 *
2433 * Returns 0 on success.
2434 */
2435int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2436                  void __user *buffer, size_t *lenp, loff_t *ppos)
2437{
2438        struct do_proc_dointvec_minmax_conv_param param = {
2439                .min = (int *) table->extra1,
2440                .max = (int *) table->extra2,
2441        };
2442        return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2443                                do_proc_dointvec_minmax_conv, &param);
2444}
2445
2446static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2447                                     struct file *filp,
2448                                     void __user *buffer,
2449                                     size_t *lenp, loff_t *ppos,
2450                                     unsigned long convmul,
2451                                     unsigned long convdiv)
2452{
2453#define TMPBUFLEN 21
2454        unsigned long *i, *min, *max, val;
2455        int vleft, first=1, neg;
2456        size_t len, left;
2457        char buf[TMPBUFLEN], *p;
2458        char __user *s = buffer;
2459        
2460        if (!data || !table->maxlen || !*lenp ||
2461            (*ppos && !write)) {
2462                *lenp = 0;
2463                return 0;
2464        }
2465        
2466        i = (unsigned long *) data;
2467        min = (unsigned long *) table->extra1;
2468        max = (unsigned long *) table->extra2;
2469        vleft = table->maxlen / sizeof(unsigned long);
2470        left = *lenp;
2471        
2472        for (; left && vleft--; i++, min++, max++, first=0) {
2473                if (write) {
2474                        while (left) {
2475                                char c;
2476                                if (get_user(c, s))
2477                                        return -EFAULT;
2478                                if (!isspace(c))
2479                                        break;
2480                                left--;
2481                                s++;
2482                        }
2483                        if (!left)
2484                                break;
2485                        neg = 0;
2486                        len = left;
2487                        if (len > TMPBUFLEN-1)
2488                                len = TMPBUFLEN-1;
2489                        if (copy_from_user(buf, s, len))
2490                                return -EFAULT;
2491                        buf[len] = 0;
2492                        p = buf;
2493                        if (*p == '-' && left > 1) {
2494                                neg = 1;
2495                                p++;
2496                        }
2497                        if (*p < '0' || *p > '9')
2498                                break;
2499                        val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2500                        len = p-buf;
2501                        if ((len < left) && *p && !isspace(*p))
2502                                break;
2503                        if (neg)
2504                                val = -val;
2505                        s += len;
2506                        left -= len;
2507
2508                        if(neg)
2509                                continue;
2510                        if ((min && val < *min) || (max && val > *max))
2511                                continue;
2512                        *i = val;
2513                } else {
2514                        p = buf;
2515                        if (!first)
2516                                *p++ = '\t';
2517                        sprintf(p, "%lu", convdiv * (*i) / convmul);
2518                        len = strlen(buf);
2519                        if (len > left)
2520                                len = left;
2521                        if(copy_to_user(s, buf, len))
2522                                return -EFAULT;
2523                        left -= len;
2524                        s += len;
2525                }
2526        }
2527
2528        if (!write && !first && left) {
2529                if(put_user('\n', s))
2530                        return -EFAULT;
2531                left--, s++;
2532        }
2533        if (write) {
2534                while (left) {
2535                        char c;
2536                        if (get_user(c, s++))
2537                                return -EFAULT;
2538                        if (!isspace(c))
2539                                break;
2540                        left--;
2541                }
2542        }
2543        if (write && first)
2544                return -EINVAL;
2545        *lenp -= left;
2546        *ppos += *lenp;
2547        return 0;
2548#undef TMPBUFLEN
2549}
2550
2551static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2552                                     struct file *filp,
2553                                     void __user *buffer,
2554                                     size_t *lenp, loff_t *ppos,
2555                                     unsigned long convmul,
2556                                     unsigned long convdiv)
2557{
2558        return __do_proc_doulongvec_minmax(table->data, table, write,
2559                        filp, buffer, lenp, ppos, convmul, convdiv);
2560}
2561
2562/**
2563 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2564 * @table: the sysctl table
2565 * @write: %TRUE if this is a write to the sysctl file
2566 * @filp: the file structure
2567 * @buffer: the user buffer
2568 * @lenp: the size of the user buffer
2569 * @ppos: file position
2570 *
2571 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2572 * values from/to the user buffer, treated as an ASCII string.
2573 *
2574 * This routine will ensure the values are within the range specified by
2575 * table->extra1 (min) and table->extra2 (max).
2576 *
2577 * Returns 0 on success.
2578 */
2579int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2580                           void __user *buffer, size_t *lenp, loff_t *ppos)
2581{
2582    return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2583}
2584
2585/**
2586 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2587 * @table: the sysctl table
2588 * @write: %TRUE if this is a write to the sysctl file
2589 * @filp: the file structure
2590 * @buffer: the user buffer
2591 * @lenp: the size of the user buffer
2592 * @ppos: file position
2593 *
2594 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2595 * values from/to the user buffer, treated as an ASCII string. The values
2596 * are treated as milliseconds, and converted to jiffies when they are stored.
2597 *
2598 * This routine will ensure the values are within the range specified by
2599 * table->extra1 (min) and table->extra2 (max).
2600 *
2601 * Returns 0 on success.
2602 */
2603int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2604                                      struct file *filp,
2605                                      void __user *buffer,
2606                                      size_t *lenp, loff_t *ppos)
2607{
2608    return do_proc_doulongvec_minmax(table, write, filp, buffer,
2609                                     lenp, ppos, HZ, 1000l);
2610}
2611
2612
2613static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2614                                         int *valp,
2615                                         int write, void *data)
2616{
2617        if (write) {
2618                if (*lvalp > LONG_MAX / HZ)
2619                        return 1;
2620                *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2621        } else {
2622                int val = *valp;
2623                unsigned long lval;
2624                if (val < 0) {
2625                        *negp = -1;
2626                        lval = (unsigned long)-val;
2627                } else {
2628                        *negp = 0;
2629                        lval = (unsigned long)val;
2630                }
2631                *lvalp = lval / HZ;
2632        }
2633        return 0;
2634}
2635
2636static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2637                                                int *valp,
2638                                                int write, void *data)
2639{
2640        if (write) {
2641                if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2642                        return 1;
2643                *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2644        } else {
2645                int val = *valp;
2646                unsigned long lval;
2647                if (val < 0) {
2648                        *negp = -1;
2649                        lval = (unsigned long)-val;
2650                } else {
2651                        *negp = 0;
2652                        lval = (unsigned long)val;
2653                }
2654                *lvalp = jiffies_to_clock_t(lval);
2655        }
2656        return 0;
2657}
2658
2659static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2660                                            int *valp,
2661                                            int write, void *data)
2662{
2663        if (write) {
2664                *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2665        } else {
2666                int val = *valp;
2667                unsigned long lval;
2668                if (val < 0) {
2669                        *negp = -1;
2670                        lval = (unsigned long)-val;
2671                } else {
2672                        *negp = 0;
2673                        lval = (unsigned long)val;
2674                }
2675                *lvalp = jiffies_to_msecs(lval);
2676        }
2677        return 0;
2678}
2679
2680/**
2681 * proc_dointvec_jiffies - read a vector of integers as seconds
2682 * @table: the sysctl table
2683 * @write: %TRUE if this is a write to the sysctl file
2684 * @filp: the file structure
2685 * @buffer: the user buffer
2686 * @lenp: the size of the user buffer
2687 * @ppos: file position
2688 *
2689 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2690 * values from/to the user buffer, treated as an ASCII string. 
2691 * The values read are assumed to be in seconds, and are converted into
2692 * jiffies.
2693 *
2694 * Returns 0 on success.
2695 */
2696int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2697                          void __user *buffer, size_t *lenp, loff_t *ppos)
2698{
2699    return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2700                            do_proc_dointvec_jiffies_conv,NULL);
2701}
2702
2703/**
2704 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2705 * @table: the sysctl table
2706 * @write: %TRUE if this is a write to the sysctl file
2707 * @filp: the file structure
2708 * @buffer: the user buffer
2709 * @lenp: the size of the user buffer
2710 * @ppos: pointer to the file position
2711 *
2712 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2713 * values from/to the user buffer, treated as an ASCII string. 
2714 * The values read are assumed to be in 1/USER_HZ seconds, and 
2715 * are converted into jiffies.
2716 *
2717 * Returns 0 on success.
2718 */
2719int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2720                                 void __user *buffer, size_t *lenp, loff_t *ppos)
2721{
2722    return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2723                            do_proc_dointvec_userhz_jiffies_conv,NULL);
2724}
2725
2726/**
2727 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2728 * @table: the sysctl table
2729 * @write: %TRUE if this is a write to the sysctl file
2730 * @filp: the file structure
2731 * @buffer: the user buffer
2732 * @lenp: the size of the user buffer
2733 * @ppos: file position
2734 * @ppos: the current position in the file
2735 *
2736 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2737 * values from/to the user buffer, treated as an ASCII string. 
2738 * The values read are assumed to be in 1/1000 seconds, and 
2739 * are converted into jiffies.
2740 *
2741 * Returns 0 on success.
2742 */
2743int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2744                             void __user *buffer, size_t *lenp, loff_t *ppos)
2745{
2746        return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2747                                do_proc_dointvec_ms_jiffies_conv, NULL);
2748}
2749
2750static int proc_do_cad_pid(struct ctl_table *table, int write, struct file *filp,
2751                           void __user *buffer, size_t *lenp, loff_t *ppos)
2752{
2753        struct pid *new_pid;
2754        pid_t tmp;
2755        int r;
2756
2757        tmp = pid_vnr(cad_pid);
2758
2759        r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2760                               lenp, ppos, NULL, NULL);
2761        if (r || !write)
2762                return r;
2763
2764        new_pid = find_get_pid(tmp);
2765        if (!new_pid)
2766                return -ESRCH;
2767
2768        put_pid(xchg(&cad_pid, new_pid));
2769        return 0;
2770}
2771
2772#else /* CONFIG_PROC_FS */
2773
2774int proc_dostring(struct ctl_table *table, int write, struct file *filp,
2775                  void __user *buffer, size_t *lenp, loff_t *ppos)
2776{
2777        return -ENOSYS;
2778}
2779
2780int proc_dointvec(struct ctl_table *table, int write, struct file *filp,
2781                  void __user *buffer, size_t *lenp, loff_t *ppos)
2782{
2783        return -ENOSYS;
2784}
2785
2786int proc_dointvec_minmax(struct ctl_table *table, int write, struct file *filp,
2787                    void __user *buffer, size_t *lenp, loff_t *ppos)
2788{
2789        return -ENOSYS;
2790}
2791
2792int proc_dointvec_jiffies(struct ctl_table *table, int write, struct file *filp,
2793                    void __user *buffer, size_t *lenp, loff_t *ppos)
2794{
2795        return -ENOSYS;
2796}
2797
2798int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write, struct file *filp,
2799                    void __user *buffer, size_t *lenp, loff_t *ppos)
2800{
2801        return -ENOSYS;
2802}
2803
2804int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, struct file *filp,
2805                             void __user *buffer, size_t *lenp, loff_t *ppos)
2806{
2807        return -ENOSYS;
2808}
2809
2810int proc_doulongvec_minmax(struct ctl_table *table, int write, struct file *filp,
2811                    void __user *buffer, size_t *lenp, loff_t *ppos)
2812{
2813        return -ENOSYS;
2814}
2815
2816int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2817                                      struct file *filp,
2818                                      void __user *buffer,
2819                                      size_t *lenp, loff_t *ppos)
2820{
2821    return -ENOSYS;
2822}
2823
2824
2825#endif /* CONFIG_PROC_FS */
2826
2827
2828#ifdef CONFIG_SYSCTL_SYSCALL
2829/*
2830 * General sysctl support routines 
2831 */
2832
2833/* The generic sysctl data routine (used if no strategy routine supplied) */
2834int sysctl_data(struct ctl_table *table,
2835                void __user *oldval, size_t __user *oldlenp,
2836                void __user *newval, size_t newlen)
2837{
2838        size_t len;
2839
2840        /* Get out of I don't have a variable */
2841        if (!table->data || !table->maxlen)
2842                return -ENOTDIR;
2843
2844        if (oldval && oldlenp) {
2845                if (get_user(len, oldlenp))
2846                        return -EFAULT;
2847                if (len) {
2848                        if (len > table->maxlen)
2849                                len = table->maxlen;
2850                        if (copy_to_user(oldval, table->data, len))
2851                                return -EFAULT;
2852                        if (put_user(len, oldlenp))
2853                                return -EFAULT;
2854                }
2855        }
2856
2857        if (newval && newlen) {
2858                if (newlen > table->maxlen)
2859                        newlen = table->maxlen;
2860
2861                if (copy_from_user(table->data, newval, newlen))
2862                        return -EFAULT;
2863        }
2864        return 1;
2865}
2866
2867/* The generic string strategy routine: */
2868int sysctl_string(struct ctl_table *table,
2869                  void __user *oldval, size_t __user *oldlenp,
2870                  void __user *newval, size_t newlen)
2871{
2872        if (!table->data || !table->maxlen) 
2873                return -ENOTDIR;
2874        
2875        if (oldval && oldlenp) {
2876                size_t bufsize;
2877                if (get_user(bufsize, oldlenp))
2878                        return -EFAULT;
2879                if (bufsize) {
2880                        size_t len = strlen(table->data), copied;
2881
2882                        /* This shouldn't trigger for a well-formed sysctl */
2883                        if (len > table->maxlen)
2884                                len = table->maxlen;
2885
2886                        /* Copy up to a max of bufsize-1 bytes of the string */
2887                        copied = (len >= bufsize) ? bufsize - 1 : len;
2888
2889                        if (copy_to_user(oldval, table->data, copied) ||
2890                            put_user(0, (char __user *)(oldval + copied)))
2891                                return -EFAULT;
2892                        if (put_user(len, oldlenp))
2893                                return -EFAULT;
2894                }
2895        }
2896        if (newval && newlen) {
2897                size_t len = newlen;
2898                if (len > table->maxlen)
2899                        len = table->maxlen;
2900                if(copy_from_user(table->data, newval, len))
2901                        return -EFAULT;
2902                if (len == table->maxlen)
2903                        len--;
2904                ((char *) table->data)[len] = 0;
2905        }
2906        return 1;
2907}
2908
2909/*
2910 * This function makes sure that all of the integers in the vector
2911 * are between the minimum and maximum values given in the arrays
2912 * table->extra1 and table->extra2, respectively.
2913 */
2914int sysctl_intvec(struct ctl_table *table,
2915                void __user *oldval, size_t __user *oldlenp,
2916                void __user *newval, size_t newlen)
2917{
2918
2919        if (newval && newlen) {
2920                int __user *vec = (int __user *) newval;
2921                int *min = (int *) table->extra1;
2922                int *max = (int *) table->extra2;
2923                size_t length;
2924                int i;
2925
2926                if (newlen % sizeof(int) != 0)
2927                        return -EINVAL;
2928
2929                if (!table->extra1 && !table->extra2)
2930                        return 0;
2931
2932                if (newlen > table->maxlen)
2933                        newlen = table->maxlen;
2934                length = newlen / sizeof(int);
2935
2936                for (i = 0; i < length; i++) {
2937                        int value;
2938                        if (get_user(value, vec + i))
2939                                return -EFAULT;
2940                        if (min && value < min[i])
2941                                return -EINVAL;
2942                        if (max && value > max[i])
2943                                return -EINVAL;
2944                }
2945        }
2946        return 0;
2947}
2948
2949/* Strategy function to convert jiffies to seconds */ 
2950int sysctl_jiffies(struct ctl_table *table,
2951                void __user *oldval, size_t __user *oldlenp,
2952                void __user *newval, size_t newlen)
2953{
2954        if (oldval && oldlenp) {
2955                size_t olen;
2956
2957                if (get_user(olen, oldlenp))
2958                        return -EFAULT;
2959                if (olen) {
2960                        int val;
2961
2962                        if (olen < sizeof(int))
2963                                return -EINVAL;
2964
2965                        val = *(int *)(table->data) / HZ;
2966                        if (put_user(val, (int __user *)oldval))
2967                                return -EFAULT;
2968                        if (put_user(sizeof(int), oldlenp))
2969                                return -EFAULT;
2970                }
2971        }
2972        if (newval && newlen) { 
2973                int new;
2974                if (newlen != sizeof(int))
2975                        return -EINVAL; 
2976                if (get_user(new, (int __user *)newval))
2977                        return -EFAULT;
2978                *(int *)(table->data) = new*HZ; 
2979        }
2980        return 1;
2981}
2982
2983/* Strategy function to convert jiffies to seconds */ 
2984int sysctl_ms_jiffies(struct ctl_table *table,
2985                void __user *oldval, size_t __user *oldlenp,
2986                void __user *newval, size_t newlen)
2987{
2988        if (oldval && oldlenp) {
2989                size_t olen;
2990
2991                if (get_user(olen, oldlenp))
2992                        return -EFAULT;
2993                if (olen) {
2994                        int val;
2995
2996                        if (olen < sizeof(int))
2997                                return -EINVAL;
2998
2999                        val = jiffies_to_msecs(*(int *)(table->data));
3000                        if (put_user(val, (int __user *)oldval))
3001                                return -EFAULT;
3002                        if (put_user(sizeof(int), oldlenp))
3003                                return -EFAULT;
3004                }
3005        }
3006        if (newval && newlen) { 
3007                int new;
3008                if (newlen != sizeof(int))
3009                        return -EINVAL; 
3010                if (get_user(new, (int __user *)newval))
3011                        return -EFAULT;
3012                *(int *)(table->data) = msecs_to_jiffies(new);
3013        }
3014        return 1;
3015}
3016
3017
3018
3019#else /* CONFIG_SYSCTL_SYSCALL */
3020
3021
3022SYSCALL_DEFINE1(sysctl, struct __sysctl_args __user *, args)
3023{
3024        struct __sysctl_args tmp;
3025        int error;
3026
3027        if (copy_from_user(&tmp, args, sizeof(tmp)))
3028                return -EFAULT;
3029
3030        error = deprecated_sysctl_warning(&tmp);
3031
3032        /* If no error reading the parameters then just -ENOSYS ... */
3033        if (!error)
3034                error = -ENOSYS;
3035
3036        return error;
3037}
3038
3039int sysctl_data(struct ctl_table *table,
3040                  void __user *oldval, size_t __user *oldlenp,
3041                  void __user *newval, size_t newlen)
3042{
3043        return -ENOSYS;
3044}
3045
3046int sysctl_string(struct ctl_table *table,
3047                  void __user *oldval, size_t __user *oldlenp,
3048                  void __user *newval, size_t newlen)
3049{
3050        return -ENOSYS;
3051}
3052
3053int sysctl_intvec(struct ctl_table *table,
3054                void __user *oldval, size_t __user *oldlenp,
3055                void __user *newval, size_t newlen)
3056{
3057        return -ENOSYS;
3058}
3059
3060int sysctl_jiffies(struct ctl_table *table,
3061                void __user *oldval, size_t __user *oldlenp,
3062                void __user *newval, size_t newlen)
3063{
3064        return -ENOSYS;
3065}
3066
3067int sysctl_ms_jiffies(struct ctl_table *table,
3068                void __user *oldval, size_t __user *oldlenp,
3069                void __user *newval, size_t newlen)
3070{
3071        return -ENOSYS;
3072}
3073
3074#endif /* CONFIG_SYSCTL_SYSCALL */
3075
3076static int deprecated_sysctl_warning(struct __sysctl_args *args)
3077{
3078        static int msg_count;
3079        int name[CTL_MAXNAME];
3080        int i;
3081
3082        /* Check args->nlen. */
3083        if (args->nlen < 0 || args->nlen > CTL_MAXNAME)
3084                return -ENOTDIR;
3085
3086        /* Read in the sysctl name for better debug message logging */
3087        for (i = 0; i < args->nlen; i++)
3088                if (get_user(name[i], args->name + i))
3089                        return -EFAULT;
3090
3091        /* Ignore accesses to kernel.version */
3092        if ((args->nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
3093                return 0;
3094
3095        if (msg_count < 5) {
3096                msg_count++;
3097                printk(KERN_INFO
3098                        "warning: process `%s' used the deprecated sysctl "
3099                        "system call with ", current->comm);
3100                for (i = 0; i < args->nlen; i++)
3101                        printk("%d.", name[i]);
3102                printk("\n");
3103        }
3104        return 0;
3105}
3106
3107/*
3108 * No sense putting this after each symbol definition, twice,
3109 * exception granted :-)
3110 */
3111EXPORT_SYMBOL(proc_dointvec);
3112EXPORT_SYMBOL(proc_dointvec_jiffies);
3113EXPORT_SYMBOL(proc_dointvec_minmax);
3114EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3115EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3116EXPORT_SYMBOL(proc_dostring);
3117EXPORT_SYMBOL(proc_doulongvec_minmax);
3118EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3119EXPORT_SYMBOL(register_sysctl_table);
3120EXPORT_SYMBOL(register_sysctl_paths);
3121EXPORT_SYMBOL(sysctl_intvec);
3122EXPORT_SYMBOL(sysctl_jiffies);
3123EXPORT_SYMBOL(sysctl_ms_jiffies);
3124EXPORT_SYMBOL(sysctl_string);
3125EXPORT_SYMBOL(sysctl_data);
3126EXPORT_SYMBOL(unregister_sysctl_table);
3127