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/aio.h>
  23#include <linux/mm.h>
  24#include <linux/swap.h>
  25#include <linux/slab.h>
  26#include <linux/sysctl.h>
  27#include <linux/bitmap.h>
  28#include <linux/signal.h>
  29#include <linux/printk.h>
  30#include <linux/proc_fs.h>
  31#include <linux/security.h>
  32#include <linux/ctype.h>
  33#include <linux/kmemcheck.h>
  34#include <linux/kmemleak.h>
  35#include <linux/fs.h>
  36#include <linux/init.h>
  37#include <linux/kernel.h>
  38#include <linux/kobject.h>
  39#include <linux/net.h>
  40#include <linux/sysrq.h>
  41#include <linux/highuid.h>
  42#include <linux/writeback.h>
  43#include <linux/ratelimit.h>
  44#include <linux/compaction.h>
  45#include <linux/hugetlb.h>
  46#include <linux/initrd.h>
  47#include <linux/key.h>
  48#include <linux/times.h>
  49#include <linux/limits.h>
  50#include <linux/dcache.h>
  51#include <linux/dnotify.h>
  52#include <linux/syscalls.h>
  53#include <linux/vmstat.h>
  54#include <linux/nfs_fs.h>
  55#include <linux/acpi.h>
  56#include <linux/reboot.h>
  57#include <linux/ftrace.h>
  58#include <linux/perf_event.h>
  59#include <linux/kprobes.h>
  60#include <linux/pipe_fs_i.h>
  61#include <linux/oom.h>
  62#include <linux/kmod.h>
  63#include <linux/capability.h>
  64#include <linux/binfmts.h>
  65#include <linux/sched/sysctl.h>
  66#include <linux/kexec.h>
  67#include <linux/bpf.h>
  68#include <linux/mount.h>
  69
  70#include <asm/uaccess.h>
  71#include <asm/processor.h>
  72
  73#ifdef CONFIG_X86
  74#include <asm/nmi.h>
  75#include <asm/stacktrace.h>
  76#include <asm/io.h>
  77#endif
  78#ifdef CONFIG_SPARC
  79#include <asm/setup.h>
  80#endif
  81#ifdef CONFIG_BSD_PROCESS_ACCT
  82#include <linux/acct.h>
  83#endif
  84#ifdef CONFIG_RT_MUTEXES
  85#include <linux/rtmutex.h>
  86#endif
  87#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
  88#include <linux/lockdep.h>
  89#endif
  90#ifdef CONFIG_CHR_DEV_SG
  91#include <scsi/sg.h>
  92#endif
  93
  94#ifdef CONFIG_LOCKUP_DETECTOR
  95#include <linux/nmi.h>
  96#endif
  97
  98#if defined(CONFIG_SYSCTL)
  99
 100/* External variables not in a header file. */
 101extern int suid_dumpable;
 102#ifdef CONFIG_COREDUMP
 103extern int core_uses_pid;
 104extern char core_pattern[];
 105extern unsigned int core_pipe_limit;
 106#endif
 107extern int pid_max;
 108extern int pid_max_min, pid_max_max;
 109extern int percpu_pagelist_fraction;
 110extern int latencytop_enabled;
 111extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
 112#ifndef CONFIG_MMU
 113extern int sysctl_nr_trim_pages;
 114#endif
 115
 116/* Constants used for minimum and  maximum */
 117#ifdef CONFIG_LOCKUP_DETECTOR
 118static int sixty = 60;
 119#endif
 120
 121static int __maybe_unused neg_one = -1;
 122
 123static int zero;
 124static int __maybe_unused one = 1;
 125static int __maybe_unused two = 2;
 126static int __maybe_unused four = 4;
 127static unsigned long one_ul = 1;
 128static int one_hundred = 100;
 129static int one_thousand = 1000;
 130#ifdef CONFIG_PRINTK
 131static int ten_thousand = 10000;
 132#endif
 133#ifdef CONFIG_PERF_EVENTS
 134static int six_hundred_forty_kb = 640 * 1024;
 135#endif
 136
 137/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 138static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 139
 140/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 141static int maxolduid = 65535;
 142static int minolduid;
 143
 144static int ngroups_max = NGROUPS_MAX;
 145static const int cap_last_cap = CAP_LAST_CAP;
 146
 147/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
 148#ifdef CONFIG_DETECT_HUNG_TASK
 149static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
 150#endif
 151
 152#ifdef CONFIG_INOTIFY_USER
 153#include <linux/inotify.h>
 154#endif
 155#ifdef CONFIG_SPARC
 156#endif
 157
 158#ifdef __hppa__
 159extern int pwrsw_enabled;
 160#endif
 161
 162#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 163extern int unaligned_enabled;
 164#endif
 165
 166#ifdef CONFIG_IA64
 167extern int unaligned_dump_stack;
 168#endif
 169
 170#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
 171extern int no_unaligned_warning;
 172#endif
 173
 174#ifdef CONFIG_PROC_SYSCTL
 175
 176#define SYSCTL_WRITES_LEGACY    -1
 177#define SYSCTL_WRITES_WARN       0
 178#define SYSCTL_WRITES_STRICT     1
 179
 180static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
 181
 182static int proc_do_cad_pid(struct ctl_table *table, int write,
 183                  void __user *buffer, size_t *lenp, loff_t *ppos);
 184static int proc_taint(struct ctl_table *table, int write,
 185                               void __user *buffer, size_t *lenp, loff_t *ppos);
 186#endif
 187
 188#ifdef CONFIG_PRINTK
 189static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
 190                                void __user *buffer, size_t *lenp, loff_t *ppos);
 191#endif
 192
 193static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
 194                void __user *buffer, size_t *lenp, loff_t *ppos);
 195#ifdef CONFIG_COREDUMP
 196static int proc_dostring_coredump(struct ctl_table *table, int write,
 197                void __user *buffer, size_t *lenp, loff_t *ppos);
 198#endif
 199
 200#ifdef CONFIG_MAGIC_SYSRQ
 201/* Note: sysrq code uses it's own private copy */
 202static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
 203
 204static int sysrq_sysctl_handler(struct ctl_table *table, int write,
 205                                void __user *buffer, size_t *lenp,
 206                                loff_t *ppos)
 207{
 208        int error;
 209
 210        error = proc_dointvec(table, write, buffer, lenp, ppos);
 211        if (error)
 212                return error;
 213
 214        if (write)
 215                sysrq_toggle_support(__sysrq_enabled);
 216
 217        return 0;
 218}
 219
 220#endif
 221
 222static struct ctl_table kern_table[];
 223static struct ctl_table vm_table[];
 224static struct ctl_table fs_table[];
 225static struct ctl_table debug_table[];
 226static struct ctl_table dev_table[];
 227extern struct ctl_table random_table[];
 228#ifdef CONFIG_EPOLL
 229extern struct ctl_table epoll_table[];
 230#endif
 231
 232#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 233int sysctl_legacy_va_layout;
 234#endif
 235
 236/* The default sysctl tables: */
 237
 238static struct ctl_table sysctl_base_table[] = {
 239        {
 240                .procname       = "kernel",
 241                .mode           = 0555,
 242                .child          = kern_table,
 243        },
 244        {
 245                .procname       = "vm",
 246                .mode           = 0555,
 247                .child          = vm_table,
 248        },
 249        {
 250                .procname       = "fs",
 251                .mode           = 0555,
 252                .child          = fs_table,
 253        },
 254        {
 255                .procname       = "debug",
 256                .mode           = 0555,
 257                .child          = debug_table,
 258        },
 259        {
 260                .procname       = "dev",
 261                .mode           = 0555,
 262                .child          = dev_table,
 263        },
 264        { }
 265};
 266
 267#ifdef CONFIG_SCHED_DEBUG
 268static int min_sched_granularity_ns = 100000;           /* 100 usecs */
 269static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
 270static int min_wakeup_granularity_ns;                   /* 0 usecs */
 271static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
 272#ifdef CONFIG_SMP
 273static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 274static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
 275#endif /* CONFIG_SMP */
 276#endif /* CONFIG_SCHED_DEBUG */
 277
 278#ifdef CONFIG_COMPACTION
 279static int min_extfrag_threshold;
 280static int max_extfrag_threshold = 1000;
 281#endif
 282
 283static struct ctl_table kern_table[] = {
 284        {
 285                .procname       = "sched_child_runs_first",
 286                .data           = &sysctl_sched_child_runs_first,
 287                .maxlen         = sizeof(unsigned int),
 288                .mode           = 0644,
 289                .proc_handler   = proc_dointvec,
 290        },
 291#ifdef CONFIG_SCHED_DEBUG
 292        {
 293                .procname       = "sched_min_granularity_ns",
 294                .data           = &sysctl_sched_min_granularity,
 295                .maxlen         = sizeof(unsigned int),
 296                .mode           = 0644,
 297                .proc_handler   = sched_proc_update_handler,
 298                .extra1         = &min_sched_granularity_ns,
 299                .extra2         = &max_sched_granularity_ns,
 300        },
 301        {
 302                .procname       = "sched_latency_ns",
 303                .data           = &sysctl_sched_latency,
 304                .maxlen         = sizeof(unsigned int),
 305                .mode           = 0644,
 306                .proc_handler   = sched_proc_update_handler,
 307                .extra1         = &min_sched_granularity_ns,
 308                .extra2         = &max_sched_granularity_ns,
 309        },
 310        {
 311                .procname       = "sched_wakeup_granularity_ns",
 312                .data           = &sysctl_sched_wakeup_granularity,
 313                .maxlen         = sizeof(unsigned int),
 314                .mode           = 0644,
 315                .proc_handler   = sched_proc_update_handler,
 316                .extra1         = &min_wakeup_granularity_ns,
 317                .extra2         = &max_wakeup_granularity_ns,
 318        },
 319#ifdef CONFIG_SMP
 320        {
 321                .procname       = "sched_tunable_scaling",
 322                .data           = &sysctl_sched_tunable_scaling,
 323                .maxlen         = sizeof(enum sched_tunable_scaling),
 324                .mode           = 0644,
 325                .proc_handler   = sched_proc_update_handler,
 326                .extra1         = &min_sched_tunable_scaling,
 327                .extra2         = &max_sched_tunable_scaling,
 328        },
 329        {
 330                .procname       = "sched_migration_cost_ns",
 331                .data           = &sysctl_sched_migration_cost,
 332                .maxlen         = sizeof(unsigned int),
 333                .mode           = 0644,
 334                .proc_handler   = proc_dointvec,
 335        },
 336        {
 337                .procname       = "sched_nr_migrate",
 338                .data           = &sysctl_sched_nr_migrate,
 339                .maxlen         = sizeof(unsigned int),
 340                .mode           = 0644,
 341                .proc_handler   = proc_dointvec,
 342        },
 343        {
 344                .procname       = "sched_time_avg_ms",
 345                .data           = &sysctl_sched_time_avg,
 346                .maxlen         = sizeof(unsigned int),
 347                .mode           = 0644,
 348                .proc_handler   = proc_dointvec,
 349        },
 350        {
 351                .procname       = "sched_shares_window_ns",
 352                .data           = &sysctl_sched_shares_window,
 353                .maxlen         = sizeof(unsigned int),
 354                .mode           = 0644,
 355                .proc_handler   = proc_dointvec,
 356        },
 357#ifdef CONFIG_SCHEDSTATS
 358        {
 359                .procname       = "sched_schedstats",
 360                .data           = NULL,
 361                .maxlen         = sizeof(unsigned int),
 362                .mode           = 0644,
 363                .proc_handler   = sysctl_schedstats,
 364                .extra1         = &zero,
 365                .extra2         = &one,
 366        },
 367#endif /* CONFIG_SCHEDSTATS */
 368#endif /* CONFIG_SMP */
 369#ifdef CONFIG_NUMA_BALANCING
 370        {
 371                .procname       = "numa_balancing_scan_delay_ms",
 372                .data           = &sysctl_numa_balancing_scan_delay,
 373                .maxlen         = sizeof(unsigned int),
 374                .mode           = 0644,
 375                .proc_handler   = proc_dointvec,
 376        },
 377        {
 378                .procname       = "numa_balancing_scan_period_min_ms",
 379                .data           = &sysctl_numa_balancing_scan_period_min,
 380                .maxlen         = sizeof(unsigned int),
 381                .mode           = 0644,
 382                .proc_handler   = proc_dointvec,
 383        },
 384        {
 385                .procname       = "numa_balancing_scan_period_max_ms",
 386                .data           = &sysctl_numa_balancing_scan_period_max,
 387                .maxlen         = sizeof(unsigned int),
 388                .mode           = 0644,
 389                .proc_handler   = proc_dointvec,
 390        },
 391        {
 392                .procname       = "numa_balancing_scan_size_mb",
 393                .data           = &sysctl_numa_balancing_scan_size,
 394                .maxlen         = sizeof(unsigned int),
 395                .mode           = 0644,
 396                .proc_handler   = proc_dointvec_minmax,
 397                .extra1         = &one,
 398        },
 399        {
 400                .procname       = "numa_balancing",
 401                .data           = NULL, /* filled in by handler */
 402                .maxlen         = sizeof(unsigned int),
 403                .mode           = 0644,
 404                .proc_handler   = sysctl_numa_balancing,
 405                .extra1         = &zero,
 406                .extra2         = &one,
 407        },
 408#endif /* CONFIG_NUMA_BALANCING */
 409#endif /* CONFIG_SCHED_DEBUG */
 410        {
 411                .procname       = "sched_rt_period_us",
 412                .data           = &sysctl_sched_rt_period,
 413                .maxlen         = sizeof(unsigned int),
 414                .mode           = 0644,
 415                .proc_handler   = sched_rt_handler,
 416        },
 417        {
 418                .procname       = "sched_rt_runtime_us",
 419                .data           = &sysctl_sched_rt_runtime,
 420                .maxlen         = sizeof(int),
 421                .mode           = 0644,
 422                .proc_handler   = sched_rt_handler,
 423        },
 424        {
 425                .procname       = "sched_rr_timeslice_ms",
 426                .data           = &sched_rr_timeslice,
 427                .maxlen         = sizeof(int),
 428                .mode           = 0644,
 429                .proc_handler   = sched_rr_handler,
 430        },
 431#ifdef CONFIG_SCHED_AUTOGROUP
 432        {
 433                .procname       = "sched_autogroup_enabled",
 434                .data           = &sysctl_sched_autogroup_enabled,
 435                .maxlen         = sizeof(unsigned int),
 436                .mode           = 0644,
 437                .proc_handler   = proc_dointvec_minmax,
 438                .extra1         = &zero,
 439                .extra2         = &one,
 440        },
 441#endif
 442#ifdef CONFIG_CFS_BANDWIDTH
 443        {
 444                .procname       = "sched_cfs_bandwidth_slice_us",
 445                .data           = &sysctl_sched_cfs_bandwidth_slice,
 446                .maxlen         = sizeof(unsigned int),
 447                .mode           = 0644,
 448                .proc_handler   = proc_dointvec_minmax,
 449                .extra1         = &one,
 450        },
 451#endif
 452#ifdef CONFIG_PROVE_LOCKING
 453        {
 454                .procname       = "prove_locking",
 455                .data           = &prove_locking,
 456                .maxlen         = sizeof(int),
 457                .mode           = 0644,
 458                .proc_handler   = proc_dointvec,
 459        },
 460#endif
 461#ifdef CONFIG_LOCK_STAT
 462        {
 463                .procname       = "lock_stat",
 464                .data           = &lock_stat,
 465                .maxlen         = sizeof(int),
 466                .mode           = 0644,
 467                .proc_handler   = proc_dointvec,
 468        },
 469#endif
 470        {
 471                .procname       = "panic",
 472                .data           = &panic_timeout,
 473                .maxlen         = sizeof(int),
 474                .mode           = 0644,
 475                .proc_handler   = proc_dointvec,
 476        },
 477#ifdef CONFIG_COREDUMP
 478        {
 479                .procname       = "core_uses_pid",
 480                .data           = &core_uses_pid,
 481                .maxlen         = sizeof(int),
 482                .mode           = 0644,
 483                .proc_handler   = proc_dointvec,
 484        },
 485        {
 486                .procname       = "core_pattern",
 487                .data           = core_pattern,
 488                .maxlen         = CORENAME_MAX_SIZE,
 489                .mode           = 0644,
 490                .proc_handler   = proc_dostring_coredump,
 491        },
 492        {
 493                .procname       = "core_pipe_limit",
 494                .data           = &core_pipe_limit,
 495                .maxlen         = sizeof(unsigned int),
 496                .mode           = 0644,
 497                .proc_handler   = proc_dointvec,
 498        },
 499#endif
 500#ifdef CONFIG_PROC_SYSCTL
 501        {
 502                .procname       = "tainted",
 503                .maxlen         = sizeof(long),
 504                .mode           = 0644,
 505                .proc_handler   = proc_taint,
 506        },
 507        {
 508                .procname       = "sysctl_writes_strict",
 509                .data           = &sysctl_writes_strict,
 510                .maxlen         = sizeof(int),
 511                .mode           = 0644,
 512                .proc_handler   = proc_dointvec_minmax,
 513                .extra1         = &neg_one,
 514                .extra2         = &one,
 515        },
 516#endif
 517#ifdef CONFIG_LATENCYTOP
 518        {
 519                .procname       = "latencytop",
 520                .data           = &latencytop_enabled,
 521                .maxlen         = sizeof(int),
 522                .mode           = 0644,
 523                .proc_handler   = sysctl_latencytop,
 524        },
 525#endif
 526#ifdef CONFIG_BLK_DEV_INITRD
 527        {
 528                .procname       = "real-root-dev",
 529                .data           = &real_root_dev,
 530                .maxlen         = sizeof(int),
 531                .mode           = 0644,
 532                .proc_handler   = proc_dointvec,
 533        },
 534#endif
 535        {
 536                .procname       = "print-fatal-signals",
 537                .data           = &print_fatal_signals,
 538                .maxlen         = sizeof(int),
 539                .mode           = 0644,
 540                .proc_handler   = proc_dointvec,
 541        },
 542#ifdef CONFIG_SPARC
 543        {
 544                .procname       = "reboot-cmd",
 545                .data           = reboot_command,
 546                .maxlen         = 256,
 547                .mode           = 0644,
 548                .proc_handler   = proc_dostring,
 549        },
 550        {
 551                .procname       = "stop-a",
 552                .data           = &stop_a_enabled,
 553                .maxlen         = sizeof (int),
 554                .mode           = 0644,
 555                .proc_handler   = proc_dointvec,
 556        },
 557        {
 558                .procname       = "scons-poweroff",
 559                .data           = &scons_pwroff,
 560                .maxlen         = sizeof (int),
 561                .mode           = 0644,
 562                .proc_handler   = proc_dointvec,
 563        },
 564#endif
 565#ifdef CONFIG_SPARC64
 566        {
 567                .procname       = "tsb-ratio",
 568                .data           = &sysctl_tsb_ratio,
 569                .maxlen         = sizeof (int),
 570                .mode           = 0644,
 571                .proc_handler   = proc_dointvec,
 572        },
 573#endif
 574#ifdef __hppa__
 575        {
 576                .procname       = "soft-power",
 577                .data           = &pwrsw_enabled,
 578                .maxlen         = sizeof (int),
 579                .mode           = 0644,
 580                .proc_handler   = proc_dointvec,
 581        },
 582#endif
 583#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 584        {
 585                .procname       = "unaligned-trap",
 586                .data           = &unaligned_enabled,
 587                .maxlen         = sizeof (int),
 588                .mode           = 0644,
 589                .proc_handler   = proc_dointvec,
 590        },
 591#endif
 592        {
 593                .procname       = "ctrl-alt-del",
 594                .data           = &C_A_D,
 595                .maxlen         = sizeof(int),
 596                .mode           = 0644,
 597                .proc_handler   = proc_dointvec,
 598        },
 599#ifdef CONFIG_FUNCTION_TRACER
 600        {
 601                .procname       = "ftrace_enabled",
 602                .data           = &ftrace_enabled,
 603                .maxlen         = sizeof(int),
 604                .mode           = 0644,
 605                .proc_handler   = ftrace_enable_sysctl,
 606        },
 607#endif
 608#ifdef CONFIG_STACK_TRACER
 609        {
 610                .procname       = "stack_tracer_enabled",
 611                .data           = &stack_tracer_enabled,
 612                .maxlen         = sizeof(int),
 613                .mode           = 0644,
 614                .proc_handler   = stack_trace_sysctl,
 615        },
 616#endif
 617#ifdef CONFIG_TRACING
 618        {
 619                .procname       = "ftrace_dump_on_oops",
 620                .data           = &ftrace_dump_on_oops,
 621                .maxlen         = sizeof(int),
 622                .mode           = 0644,
 623                .proc_handler   = proc_dointvec,
 624        },
 625        {
 626                .procname       = "traceoff_on_warning",
 627                .data           = &__disable_trace_on_warning,
 628                .maxlen         = sizeof(__disable_trace_on_warning),
 629                .mode           = 0644,
 630                .proc_handler   = proc_dointvec,
 631        },
 632        {
 633                .procname       = "tracepoint_printk",
 634                .data           = &tracepoint_printk,
 635                .maxlen         = sizeof(tracepoint_printk),
 636                .mode           = 0644,
 637                .proc_handler   = proc_dointvec,
 638        },
 639#endif
 640#ifdef CONFIG_KEXEC_CORE
 641        {
 642                .procname       = "kexec_load_disabled",
 643                .data           = &kexec_load_disabled,
 644                .maxlen         = sizeof(int),
 645                .mode           = 0644,
 646                /* only handle a transition from default "0" to "1" */
 647                .proc_handler   = proc_dointvec_minmax,
 648                .extra1         = &one,
 649                .extra2         = &one,
 650        },
 651#endif
 652#ifdef CONFIG_MODULES
 653        {
 654                .procname       = "modprobe",
 655                .data           = &modprobe_path,
 656                .maxlen         = KMOD_PATH_LEN,
 657                .mode           = 0644,
 658                .proc_handler   = proc_dostring,
 659        },
 660        {
 661                .procname       = "modules_disabled",
 662                .data           = &modules_disabled,
 663                .maxlen         = sizeof(int),
 664                .mode           = 0644,
 665                /* only handle a transition from default "0" to "1" */
 666                .proc_handler   = proc_dointvec_minmax,
 667                .extra1         = &one,
 668                .extra2         = &one,
 669        },
 670#endif
 671#ifdef CONFIG_UEVENT_HELPER
 672        {
 673                .procname       = "hotplug",
 674                .data           = &uevent_helper,
 675                .maxlen         = UEVENT_HELPER_PATH_LEN,
 676                .mode           = 0644,
 677                .proc_handler   = proc_dostring,
 678        },
 679#endif
 680#ifdef CONFIG_CHR_DEV_SG
 681        {
 682                .procname       = "sg-big-buff",
 683                .data           = &sg_big_buff,
 684                .maxlen         = sizeof (int),
 685                .mode           = 0444,
 686                .proc_handler   = proc_dointvec,
 687        },
 688#endif
 689#ifdef CONFIG_BSD_PROCESS_ACCT
 690        {
 691                .procname       = "acct",
 692                .data           = &acct_parm,
 693                .maxlen         = 3*sizeof(int),
 694                .mode           = 0644,
 695                .proc_handler   = proc_dointvec,
 696        },
 697#endif
 698#ifdef CONFIG_MAGIC_SYSRQ
 699        {
 700                .procname       = "sysrq",
 701                .data           = &__sysrq_enabled,
 702                .maxlen         = sizeof (int),
 703                .mode           = 0644,
 704                .proc_handler   = sysrq_sysctl_handler,
 705        },
 706#endif
 707#ifdef CONFIG_PROC_SYSCTL
 708        {
 709                .procname       = "cad_pid",
 710                .data           = NULL,
 711                .maxlen         = sizeof (int),
 712                .mode           = 0600,
 713                .proc_handler   = proc_do_cad_pid,
 714        },
 715#endif
 716        {
 717                .procname       = "threads-max",
 718                .data           = NULL,
 719                .maxlen         = sizeof(int),
 720                .mode           = 0644,
 721                .proc_handler   = sysctl_max_threads,
 722        },
 723        {
 724                .procname       = "random",
 725                .mode           = 0555,
 726                .child          = random_table,
 727        },
 728        {
 729                .procname       = "usermodehelper",
 730                .mode           = 0555,
 731                .child          = usermodehelper_table,
 732        },
 733        {
 734                .procname       = "overflowuid",
 735                .data           = &overflowuid,
 736                .maxlen         = sizeof(int),
 737                .mode           = 0644,
 738                .proc_handler   = proc_dointvec_minmax,
 739                .extra1         = &minolduid,
 740                .extra2         = &maxolduid,
 741        },
 742        {
 743                .procname       = "overflowgid",
 744                .data           = &overflowgid,
 745                .maxlen         = sizeof(int),
 746                .mode           = 0644,
 747                .proc_handler   = proc_dointvec_minmax,
 748                .extra1         = &minolduid,
 749                .extra2         = &maxolduid,
 750        },
 751#ifdef CONFIG_S390
 752#ifdef CONFIG_MATHEMU
 753        {
 754                .procname       = "ieee_emulation_warnings",
 755                .data           = &sysctl_ieee_emulation_warnings,
 756                .maxlen         = sizeof(int),
 757                .mode           = 0644,
 758                .proc_handler   = proc_dointvec,
 759        },
 760#endif
 761        {
 762                .procname       = "userprocess_debug",
 763                .data           = &show_unhandled_signals,
 764                .maxlen         = sizeof(int),
 765                .mode           = 0644,
 766                .proc_handler   = proc_dointvec,
 767        },
 768#endif
 769        {
 770                .procname       = "pid_max",
 771                .data           = &pid_max,
 772                .maxlen         = sizeof (int),
 773                .mode           = 0644,
 774                .proc_handler   = proc_dointvec_minmax,
 775                .extra1         = &pid_max_min,
 776                .extra2         = &pid_max_max,
 777        },
 778        {
 779                .procname       = "panic_on_oops",
 780                .data           = &panic_on_oops,
 781                .maxlen         = sizeof(int),
 782                .mode           = 0644,
 783                .proc_handler   = proc_dointvec,
 784        },
 785#if defined CONFIG_PRINTK
 786        {
 787                .procname       = "printk",
 788                .data           = &console_loglevel,
 789                .maxlen         = 4*sizeof(int),
 790                .mode           = 0644,
 791                .proc_handler   = proc_dointvec,
 792        },
 793        {
 794                .procname       = "printk_ratelimit",
 795                .data           = &printk_ratelimit_state.interval,
 796                .maxlen         = sizeof(int),
 797                .mode           = 0644,
 798                .proc_handler   = proc_dointvec_jiffies,
 799        },
 800        {
 801                .procname       = "printk_ratelimit_burst",
 802                .data           = &printk_ratelimit_state.burst,
 803                .maxlen         = sizeof(int),
 804                .mode           = 0644,
 805                .proc_handler   = proc_dointvec,
 806        },
 807        {
 808                .procname       = "printk_delay",
 809                .data           = &printk_delay_msec,
 810                .maxlen         = sizeof(int),
 811                .mode           = 0644,
 812                .proc_handler   = proc_dointvec_minmax,
 813                .extra1         = &zero,
 814                .extra2         = &ten_thousand,
 815        },
 816        {
 817                .procname       = "printk_devkmsg",
 818                .data           = devkmsg_log_str,
 819                .maxlen         = DEVKMSG_STR_MAX_SIZE,
 820                .mode           = 0644,
 821                .proc_handler   = devkmsg_sysctl_set_loglvl,
 822        },
 823        {
 824                .procname       = "dmesg_restrict",
 825                .data           = &dmesg_restrict,
 826                .maxlen         = sizeof(int),
 827                .mode           = 0644,
 828                .proc_handler   = proc_dointvec_minmax_sysadmin,
 829                .extra1         = &zero,
 830                .extra2         = &one,
 831        },
 832        {
 833                .procname       = "kptr_restrict",
 834                .data           = &kptr_restrict,
 835                .maxlen         = sizeof(int),
 836                .mode           = 0644,
 837                .proc_handler   = proc_dointvec_minmax_sysadmin,
 838                .extra1         = &zero,
 839                .extra2         = &two,
 840        },
 841#endif
 842        {
 843                .procname       = "ngroups_max",
 844                .data           = &ngroups_max,
 845                .maxlen         = sizeof (int),
 846                .mode           = 0444,
 847                .proc_handler   = proc_dointvec,
 848        },
 849        {
 850                .procname       = "cap_last_cap",
 851                .data           = (void *)&cap_last_cap,
 852                .maxlen         = sizeof(int),
 853                .mode           = 0444,
 854                .proc_handler   = proc_dointvec,
 855        },
 856#if defined(CONFIG_LOCKUP_DETECTOR)
 857        {
 858                .procname       = "watchdog",
 859                .data           = &watchdog_user_enabled,
 860                .maxlen         = sizeof (int),
 861                .mode           = 0644,
 862                .proc_handler   = proc_watchdog,
 863                .extra1         = &zero,
 864                .extra2         = &one,
 865        },
 866        {
 867                .procname       = "watchdog_thresh",
 868                .data           = &watchdog_thresh,
 869                .maxlen         = sizeof(int),
 870                .mode           = 0644,
 871                .proc_handler   = proc_watchdog_thresh,
 872                .extra1         = &zero,
 873                .extra2         = &sixty,
 874        },
 875        {
 876                .procname       = "nmi_watchdog",
 877                .data           = &nmi_watchdog_enabled,
 878                .maxlen         = sizeof (int),
 879                .mode           = 0644,
 880                .proc_handler   = proc_nmi_watchdog,
 881                .extra1         = &zero,
 882#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
 883                .extra2         = &one,
 884#else
 885                .extra2         = &zero,
 886#endif
 887        },
 888        {
 889                .procname       = "soft_watchdog",
 890                .data           = &soft_watchdog_enabled,
 891                .maxlen         = sizeof (int),
 892                .mode           = 0644,
 893                .proc_handler   = proc_soft_watchdog,
 894                .extra1         = &zero,
 895                .extra2         = &one,
 896        },
 897        {
 898                .procname       = "watchdog_cpumask",
 899                .data           = &watchdog_cpumask_bits,
 900                .maxlen         = NR_CPUS,
 901                .mode           = 0644,
 902                .proc_handler   = proc_watchdog_cpumask,
 903        },
 904        {
 905                .procname       = "softlockup_panic",
 906                .data           = &softlockup_panic,
 907                .maxlen         = sizeof(int),
 908                .mode           = 0644,
 909                .proc_handler   = proc_dointvec_minmax,
 910                .extra1         = &zero,
 911                .extra2         = &one,
 912        },
 913#ifdef CONFIG_HARDLOCKUP_DETECTOR
 914        {
 915                .procname       = "hardlockup_panic",
 916                .data           = &hardlockup_panic,
 917                .maxlen         = sizeof(int),
 918                .mode           = 0644,
 919                .proc_handler   = proc_dointvec_minmax,
 920                .extra1         = &zero,
 921                .extra2         = &one,
 922        },
 923#endif
 924#ifdef CONFIG_SMP
 925        {
 926                .procname       = "softlockup_all_cpu_backtrace",
 927                .data           = &sysctl_softlockup_all_cpu_backtrace,
 928                .maxlen         = sizeof(int),
 929                .mode           = 0644,
 930                .proc_handler   = proc_dointvec_minmax,
 931                .extra1         = &zero,
 932                .extra2         = &one,
 933        },
 934        {
 935                .procname       = "hardlockup_all_cpu_backtrace",
 936                .data           = &sysctl_hardlockup_all_cpu_backtrace,
 937                .maxlen         = sizeof(int),
 938                .mode           = 0644,
 939                .proc_handler   = proc_dointvec_minmax,
 940                .extra1         = &zero,
 941                .extra2         = &one,
 942        },
 943#endif /* CONFIG_SMP */
 944#endif
 945#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 946        {
 947                .procname       = "unknown_nmi_panic",
 948                .data           = &unknown_nmi_panic,
 949                .maxlen         = sizeof (int),
 950                .mode           = 0644,
 951                .proc_handler   = proc_dointvec,
 952        },
 953#endif
 954#if defined(CONFIG_X86)
 955        {
 956                .procname       = "panic_on_unrecovered_nmi",
 957                .data           = &panic_on_unrecovered_nmi,
 958                .maxlen         = sizeof(int),
 959                .mode           = 0644,
 960                .proc_handler   = proc_dointvec,
 961        },
 962        {
 963                .procname       = "panic_on_io_nmi",
 964                .data           = &panic_on_io_nmi,
 965                .maxlen         = sizeof(int),
 966                .mode           = 0644,
 967                .proc_handler   = proc_dointvec,
 968        },
 969#ifdef CONFIG_DEBUG_STACKOVERFLOW
 970        {
 971                .procname       = "panic_on_stackoverflow",
 972                .data           = &sysctl_panic_on_stackoverflow,
 973                .maxlen         = sizeof(int),
 974                .mode           = 0644,
 975                .proc_handler   = proc_dointvec,
 976        },
 977#endif
 978        {
 979                .procname       = "bootloader_type",
 980                .data           = &bootloader_type,
 981                .maxlen         = sizeof (int),
 982                .mode           = 0444,
 983                .proc_handler   = proc_dointvec,
 984        },
 985        {
 986                .procname       = "bootloader_version",
 987                .data           = &bootloader_version,
 988                .maxlen         = sizeof (int),
 989                .mode           = 0444,
 990                .proc_handler   = proc_dointvec,
 991        },
 992        {
 993                .procname       = "kstack_depth_to_print",
 994                .data           = &kstack_depth_to_print,
 995                .maxlen         = sizeof(int),
 996                .mode           = 0644,
 997                .proc_handler   = proc_dointvec,
 998        },
 999        {
1000                .procname       = "io_delay_type",
1001                .data           = &io_delay_type,
1002                .maxlen         = sizeof(int),
1003                .mode           = 0644,
1004                .proc_handler   = proc_dointvec,
1005        },
1006#endif
1007#if defined(CONFIG_MMU)
1008        {
1009                .procname       = "randomize_va_space",
1010                .data           = &randomize_va_space,
1011                .maxlen         = sizeof(int),
1012                .mode           = 0644,
1013                .proc_handler   = proc_dointvec,
1014        },
1015#endif
1016#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1017        {
1018                .procname       = "spin_retry",
1019                .data           = &spin_retry,
1020                .maxlen         = sizeof (int),
1021                .mode           = 0644,
1022                .proc_handler   = proc_dointvec,
1023        },
1024#endif
1025#if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1026        {
1027                .procname       = "acpi_video_flags",
1028                .data           = &acpi_realmode_flags,
1029                .maxlen         = sizeof (unsigned long),
1030                .mode           = 0644,
1031                .proc_handler   = proc_doulongvec_minmax,
1032        },
1033#endif
1034#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1035        {
1036                .procname       = "ignore-unaligned-usertrap",
1037                .data           = &no_unaligned_warning,
1038                .maxlen         = sizeof (int),
1039                .mode           = 0644,
1040                .proc_handler   = proc_dointvec,
1041        },
1042#endif
1043#ifdef CONFIG_IA64
1044        {
1045                .procname       = "unaligned-dump-stack",
1046                .data           = &unaligned_dump_stack,
1047                .maxlen         = sizeof (int),
1048                .mode           = 0644,
1049                .proc_handler   = proc_dointvec,
1050        },
1051#endif
1052#ifdef CONFIG_DETECT_HUNG_TASK
1053        {
1054                .procname       = "hung_task_panic",
1055                .data           = &sysctl_hung_task_panic,
1056                .maxlen         = sizeof(int),
1057                .mode           = 0644,
1058                .proc_handler   = proc_dointvec_minmax,
1059                .extra1         = &zero,
1060                .extra2         = &one,
1061        },
1062        {
1063                .procname       = "hung_task_check_count",
1064                .data           = &sysctl_hung_task_check_count,
1065                .maxlen         = sizeof(int),
1066                .mode           = 0644,
1067                .proc_handler   = proc_dointvec_minmax,
1068                .extra1         = &zero,
1069        },
1070        {
1071                .procname       = "hung_task_timeout_secs",
1072                .data           = &sysctl_hung_task_timeout_secs,
1073                .maxlen         = sizeof(unsigned long),
1074                .mode           = 0644,
1075                .proc_handler   = proc_dohung_task_timeout_secs,
1076                .extra2         = &hung_task_timeout_max,
1077        },
1078        {
1079                .procname       = "hung_task_warnings",
1080                .data           = &sysctl_hung_task_warnings,
1081                .maxlen         = sizeof(int),
1082                .mode           = 0644,
1083                .proc_handler   = proc_dointvec_minmax,
1084                .extra1         = &neg_one,
1085        },
1086#endif
1087#ifdef CONFIG_RT_MUTEXES
1088        {
1089                .procname       = "max_lock_depth",
1090                .data           = &max_lock_depth,
1091                .maxlen         = sizeof(int),
1092                .mode           = 0644,
1093                .proc_handler   = proc_dointvec,
1094        },
1095#endif
1096        {
1097                .procname       = "poweroff_cmd",
1098                .data           = &poweroff_cmd,
1099                .maxlen         = POWEROFF_CMD_PATH_LEN,
1100                .mode           = 0644,
1101                .proc_handler   = proc_dostring,
1102        },
1103#ifdef CONFIG_KEYS
1104        {
1105                .procname       = "keys",
1106                .mode           = 0555,
1107                .child          = key_sysctls,
1108        },
1109#endif
1110#ifdef CONFIG_PERF_EVENTS
1111        /*
1112         * User-space scripts rely on the existence of this file
1113         * as a feature check for perf_events being enabled.
1114         *
1115         * So it's an ABI, do not remove!
1116         */
1117        {
1118                .procname       = "perf_event_paranoid",
1119                .data           = &sysctl_perf_event_paranoid,
1120                .maxlen         = sizeof(sysctl_perf_event_paranoid),
1121                .mode           = 0644,
1122                .proc_handler   = proc_dointvec,
1123        },
1124        {
1125                .procname       = "perf_event_mlock_kb",
1126                .data           = &sysctl_perf_event_mlock,
1127                .maxlen         = sizeof(sysctl_perf_event_mlock),
1128                .mode           = 0644,
1129                .proc_handler   = proc_dointvec,
1130        },
1131        {
1132                .procname       = "perf_event_max_sample_rate",
1133                .data           = &sysctl_perf_event_sample_rate,
1134                .maxlen         = sizeof(sysctl_perf_event_sample_rate),
1135                .mode           = 0644,
1136                .proc_handler   = perf_proc_update_handler,
1137                .extra1         = &one,
1138        },
1139        {
1140                .procname       = "perf_cpu_time_max_percent",
1141                .data           = &sysctl_perf_cpu_time_max_percent,
1142                .maxlen         = sizeof(sysctl_perf_cpu_time_max_percent),
1143                .mode           = 0644,
1144                .proc_handler   = perf_cpu_time_max_percent_handler,
1145                .extra1         = &zero,
1146                .extra2         = &one_hundred,
1147        },
1148        {
1149                .procname       = "perf_event_max_stack",
1150                .data           = &sysctl_perf_event_max_stack,
1151                .maxlen         = sizeof(sysctl_perf_event_max_stack),
1152                .mode           = 0644,
1153                .proc_handler   = perf_event_max_stack_handler,
1154                .extra1         = &zero,
1155                .extra2         = &six_hundred_forty_kb,
1156        },
1157        {
1158                .procname       = "perf_event_max_contexts_per_stack",
1159                .data           = &sysctl_perf_event_max_contexts_per_stack,
1160                .maxlen         = sizeof(sysctl_perf_event_max_contexts_per_stack),
1161                .mode           = 0644,
1162                .proc_handler   = perf_event_max_stack_handler,
1163                .extra1         = &zero,
1164                .extra2         = &one_thousand,
1165        },
1166#endif
1167#ifdef CONFIG_KMEMCHECK
1168        {
1169                .procname       = "kmemcheck",
1170                .data           = &kmemcheck_enabled,
1171                .maxlen         = sizeof(int),
1172                .mode           = 0644,
1173                .proc_handler   = proc_dointvec,
1174        },
1175#endif
1176        {
1177                .procname       = "panic_on_warn",
1178                .data           = &panic_on_warn,
1179                .maxlen         = sizeof(int),
1180                .mode           = 0644,
1181                .proc_handler   = proc_dointvec_minmax,
1182                .extra1         = &zero,
1183                .extra2         = &one,
1184        },
1185#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1186        {
1187                .procname       = "timer_migration",
1188                .data           = &sysctl_timer_migration,
1189                .maxlen         = sizeof(unsigned int),
1190                .mode           = 0644,
1191                .proc_handler   = timer_migration_handler,
1192        },
1193#endif
1194#ifdef CONFIG_BPF_SYSCALL
1195        {
1196                .procname       = "unprivileged_bpf_disabled",
1197                .data           = &sysctl_unprivileged_bpf_disabled,
1198                .maxlen         = sizeof(sysctl_unprivileged_bpf_disabled),
1199                .mode           = 0644,
1200                /* only handle a transition from default "0" to "1" */
1201                .proc_handler   = proc_dointvec_minmax,
1202                .extra1         = &one,
1203                .extra2         = &one,
1204        },
1205#endif
1206#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1207        {
1208                .procname       = "panic_on_rcu_stall",
1209                .data           = &sysctl_panic_on_rcu_stall,
1210                .maxlen         = sizeof(sysctl_panic_on_rcu_stall),
1211                .mode           = 0644,
1212                .proc_handler   = proc_dointvec_minmax,
1213                .extra1         = &zero,
1214                .extra2         = &one,
1215        },
1216#endif
1217        { }
1218};
1219
1220static struct ctl_table vm_table[] = {
1221        {
1222                .procname       = "overcommit_memory",
1223                .data           = &sysctl_overcommit_memory,
1224                .maxlen         = sizeof(sysctl_overcommit_memory),
1225                .mode           = 0644,
1226                .proc_handler   = proc_dointvec_minmax,
1227                .extra1         = &zero,
1228                .extra2         = &two,
1229        },
1230        {
1231                .procname       = "panic_on_oom",
1232                .data           = &sysctl_panic_on_oom,
1233                .maxlen         = sizeof(sysctl_panic_on_oom),
1234                .mode           = 0644,
1235                .proc_handler   = proc_dointvec_minmax,
1236                .extra1         = &zero,
1237                .extra2         = &two,
1238        },
1239        {
1240                .procname       = "oom_kill_allocating_task",
1241                .data           = &sysctl_oom_kill_allocating_task,
1242                .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
1243                .mode           = 0644,
1244                .proc_handler   = proc_dointvec,
1245        },
1246        {
1247                .procname       = "oom_dump_tasks",
1248                .data           = &sysctl_oom_dump_tasks,
1249                .maxlen         = sizeof(sysctl_oom_dump_tasks),
1250                .mode           = 0644,
1251                .proc_handler   = proc_dointvec,
1252        },
1253        {
1254                .procname       = "overcommit_ratio",
1255                .data           = &sysctl_overcommit_ratio,
1256                .maxlen         = sizeof(sysctl_overcommit_ratio),
1257                .mode           = 0644,
1258                .proc_handler   = overcommit_ratio_handler,
1259        },
1260        {
1261                .procname       = "overcommit_kbytes",
1262                .data           = &sysctl_overcommit_kbytes,
1263                .maxlen         = sizeof(sysctl_overcommit_kbytes),
1264                .mode           = 0644,
1265                .proc_handler   = overcommit_kbytes_handler,
1266        },
1267        {
1268                .procname       = "page-cluster", 
1269                .data           = &page_cluster,
1270                .maxlen         = sizeof(int),
1271                .mode           = 0644,
1272                .proc_handler   = proc_dointvec_minmax,
1273                .extra1         = &zero,
1274        },
1275        {
1276                .procname       = "dirty_background_ratio",
1277                .data           = &dirty_background_ratio,
1278                .maxlen         = sizeof(dirty_background_ratio),
1279                .mode           = 0644,
1280                .proc_handler   = dirty_background_ratio_handler,
1281                .extra1         = &zero,
1282                .extra2         = &one_hundred,
1283        },
1284        {
1285                .procname       = "dirty_background_bytes",
1286                .data           = &dirty_background_bytes,
1287                .maxlen         = sizeof(dirty_background_bytes),
1288                .mode           = 0644,
1289                .proc_handler   = dirty_background_bytes_handler,
1290                .extra1         = &one_ul,
1291        },
1292        {
1293                .procname       = "dirty_ratio",
1294                .data           = &vm_dirty_ratio,
1295                .maxlen         = sizeof(vm_dirty_ratio),
1296                .mode           = 0644,
1297                .proc_handler   = dirty_ratio_handler,
1298                .extra1         = &zero,
1299                .extra2         = &one_hundred,
1300        },
1301        {
1302                .procname       = "dirty_bytes",
1303                .data           = &vm_dirty_bytes,
1304                .maxlen         = sizeof(vm_dirty_bytes),
1305                .mode           = 0644,
1306                .proc_handler   = dirty_bytes_handler,
1307                .extra1         = &dirty_bytes_min,
1308        },
1309        {
1310                .procname       = "dirty_writeback_centisecs",
1311                .data           = &dirty_writeback_interval,
1312                .maxlen         = sizeof(dirty_writeback_interval),
1313                .mode           = 0644,
1314                .proc_handler   = dirty_writeback_centisecs_handler,
1315        },
1316        {
1317                .procname       = "dirty_expire_centisecs",
1318                .data           = &dirty_expire_interval,
1319                .maxlen         = sizeof(dirty_expire_interval),
1320                .mode           = 0644,
1321                .proc_handler   = proc_dointvec_minmax,
1322                .extra1         = &zero,
1323        },
1324        {
1325                .procname       = "dirtytime_expire_seconds",
1326                .data           = &dirtytime_expire_interval,
1327                .maxlen         = sizeof(dirty_expire_interval),
1328                .mode           = 0644,
1329                .proc_handler   = dirtytime_interval_handler,
1330                .extra1         = &zero,
1331        },
1332        {
1333                .procname       = "nr_pdflush_threads",
1334                .mode           = 0444 /* read-only */,
1335                .proc_handler   = pdflush_proc_obsolete,
1336        },
1337        {
1338                .procname       = "swappiness",
1339                .data           = &vm_swappiness,
1340                .maxlen         = sizeof(vm_swappiness),
1341                .mode           = 0644,
1342                .proc_handler   = proc_dointvec_minmax,
1343                .extra1         = &zero,
1344                .extra2         = &one_hundred,
1345        },
1346#ifdef CONFIG_HUGETLB_PAGE
1347        {
1348                .procname       = "nr_hugepages",
1349                .data           = NULL,
1350                .maxlen         = sizeof(unsigned long),
1351                .mode           = 0644,
1352                .proc_handler   = hugetlb_sysctl_handler,
1353        },
1354#ifdef CONFIG_NUMA
1355        {
1356                .procname       = "nr_hugepages_mempolicy",
1357                .data           = NULL,
1358                .maxlen         = sizeof(unsigned long),
1359                .mode           = 0644,
1360                .proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1361        },
1362#endif
1363         {
1364                .procname       = "hugetlb_shm_group",
1365                .data           = &sysctl_hugetlb_shm_group,
1366                .maxlen         = sizeof(gid_t),
1367                .mode           = 0644,
1368                .proc_handler   = proc_dointvec,
1369         },
1370         {
1371                .procname       = "hugepages_treat_as_movable",
1372                .data           = &hugepages_treat_as_movable,
1373                .maxlen         = sizeof(int),
1374                .mode           = 0644,
1375                .proc_handler   = proc_dointvec,
1376        },
1377        {
1378                .procname       = "nr_overcommit_hugepages",
1379                .data           = NULL,
1380                .maxlen         = sizeof(unsigned long),
1381                .mode           = 0644,
1382                .proc_handler   = hugetlb_overcommit_handler,
1383        },
1384#endif
1385        {
1386                .procname       = "lowmem_reserve_ratio",
1387                .data           = &sysctl_lowmem_reserve_ratio,
1388                .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1389                .mode           = 0644,
1390                .proc_handler   = lowmem_reserve_ratio_sysctl_handler,
1391        },
1392        {
1393                .procname       = "drop_caches",
1394                .data           = &sysctl_drop_caches,
1395                .maxlen         = sizeof(int),
1396                .mode           = 0644,
1397                .proc_handler   = drop_caches_sysctl_handler,
1398                .extra1         = &one,
1399                .extra2         = &four,
1400        },
1401#ifdef CONFIG_COMPACTION
1402        {
1403                .procname       = "compact_memory",
1404                .data           = &sysctl_compact_memory,
1405                .maxlen         = sizeof(int),
1406                .mode           = 0200,
1407                .proc_handler   = sysctl_compaction_handler,
1408        },
1409        {
1410                .procname       = "extfrag_threshold",
1411                .data           = &sysctl_extfrag_threshold,
1412                .maxlen         = sizeof(int),
1413                .mode           = 0644,
1414                .proc_handler   = sysctl_extfrag_handler,
1415                .extra1         = &min_extfrag_threshold,
1416                .extra2         = &max_extfrag_threshold,
1417        },
1418        {
1419                .procname       = "compact_unevictable_allowed",
1420                .data           = &sysctl_compact_unevictable_allowed,
1421                .maxlen         = sizeof(int),
1422                .mode           = 0644,
1423                .proc_handler   = proc_dointvec,
1424                .extra1         = &zero,
1425                .extra2         = &one,
1426        },
1427
1428#endif /* CONFIG_COMPACTION */
1429        {
1430                .procname       = "min_free_kbytes",
1431                .data           = &min_free_kbytes,
1432                .maxlen         = sizeof(min_free_kbytes),
1433                .mode           = 0644,
1434                .proc_handler   = min_free_kbytes_sysctl_handler,
1435                .extra1         = &zero,
1436        },
1437        {
1438                .procname       = "watermark_scale_factor",
1439                .data           = &watermark_scale_factor,
1440                .maxlen         = sizeof(watermark_scale_factor),
1441                .mode           = 0644,
1442                .proc_handler   = watermark_scale_factor_sysctl_handler,
1443                .extra1         = &one,
1444                .extra2         = &one_thousand,
1445        },
1446        {
1447                .procname       = "percpu_pagelist_fraction",
1448                .data           = &percpu_pagelist_fraction,
1449                .maxlen         = sizeof(percpu_pagelist_fraction),
1450                .mode           = 0644,
1451                .proc_handler   = percpu_pagelist_fraction_sysctl_handler,
1452                .extra1         = &zero,
1453        },
1454#ifdef CONFIG_MMU
1455        {
1456                .procname       = "max_map_count",
1457                .data           = &sysctl_max_map_count,
1458                .maxlen         = sizeof(sysctl_max_map_count),
1459                .mode           = 0644,
1460                .proc_handler   = proc_dointvec_minmax,
1461                .extra1         = &zero,
1462        },
1463#else
1464        {
1465                .procname       = "nr_trim_pages",
1466                .data           = &sysctl_nr_trim_pages,
1467                .maxlen         = sizeof(sysctl_nr_trim_pages),
1468                .mode           = 0644,
1469                .proc_handler   = proc_dointvec_minmax,
1470                .extra1         = &zero,
1471        },
1472#endif
1473        {
1474                .procname       = "laptop_mode",
1475                .data           = &laptop_mode,
1476                .maxlen         = sizeof(laptop_mode),
1477                .mode           = 0644,
1478                .proc_handler   = proc_dointvec_jiffies,
1479        },
1480        {
1481                .procname       = "block_dump",
1482                .data           = &block_dump,
1483                .maxlen         = sizeof(block_dump),
1484                .mode           = 0644,
1485                .proc_handler   = proc_dointvec,
1486                .extra1         = &zero,
1487        },
1488        {
1489                .procname       = "vfs_cache_pressure",
1490                .data           = &sysctl_vfs_cache_pressure,
1491                .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1492                .mode           = 0644,
1493                .proc_handler   = proc_dointvec,
1494                .extra1         = &zero,
1495        },
1496#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1497        {
1498                .procname       = "legacy_va_layout",
1499                .data           = &sysctl_legacy_va_layout,
1500                .maxlen         = sizeof(sysctl_legacy_va_layout),
1501                .mode           = 0644,
1502                .proc_handler   = proc_dointvec,
1503                .extra1         = &zero,
1504        },
1505#endif
1506#ifdef CONFIG_NUMA
1507        {
1508                .procname       = "zone_reclaim_mode",
1509                .data           = &node_reclaim_mode,
1510                .maxlen         = sizeof(node_reclaim_mode),
1511                .mode           = 0644,
1512                .proc_handler   = proc_dointvec,
1513                .extra1         = &zero,
1514        },
1515        {
1516                .procname       = "min_unmapped_ratio",
1517                .data           = &sysctl_min_unmapped_ratio,
1518                .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1519                .mode           = 0644,
1520                .proc_handler   = sysctl_min_unmapped_ratio_sysctl_handler,
1521                .extra1         = &zero,
1522                .extra2         = &one_hundred,
1523        },
1524        {
1525                .procname       = "min_slab_ratio",
1526                .data           = &sysctl_min_slab_ratio,
1527                .maxlen         = sizeof(sysctl_min_slab_ratio),
1528                .mode           = 0644,
1529                .proc_handler   = sysctl_min_slab_ratio_sysctl_handler,
1530                .extra1         = &zero,
1531                .extra2         = &one_hundred,
1532        },
1533#endif
1534#ifdef CONFIG_SMP
1535        {
1536                .procname       = "stat_interval",
1537                .data           = &sysctl_stat_interval,
1538                .maxlen         = sizeof(sysctl_stat_interval),
1539                .mode           = 0644,
1540                .proc_handler   = proc_dointvec_jiffies,
1541        },
1542        {
1543                .procname       = "stat_refresh",
1544                .data           = NULL,
1545                .maxlen         = 0,
1546                .mode           = 0600,
1547                .proc_handler   = vmstat_refresh,
1548        },
1549#endif
1550#ifdef CONFIG_MMU
1551        {
1552                .procname       = "mmap_min_addr",
1553                .data           = &dac_mmap_min_addr,
1554                .maxlen         = sizeof(unsigned long),
1555                .mode           = 0644,
1556                .proc_handler   = mmap_min_addr_handler,
1557        },
1558#endif
1559#ifdef CONFIG_NUMA
1560        {
1561                .procname       = "numa_zonelist_order",
1562                .data           = &numa_zonelist_order,
1563                .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1564                .mode           = 0644,
1565                .proc_handler   = numa_zonelist_order_handler,
1566        },
1567#endif
1568#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1569   (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1570        {
1571                .procname       = "vdso_enabled",
1572#ifdef CONFIG_X86_32
1573                .data           = &vdso32_enabled,
1574                .maxlen         = sizeof(vdso32_enabled),
1575#else
1576                .data           = &vdso_enabled,
1577                .maxlen         = sizeof(vdso_enabled),
1578#endif
1579                .mode           = 0644,
1580                .proc_handler   = proc_dointvec,
1581                .extra1         = &zero,
1582        },
1583#endif
1584#ifdef CONFIG_HIGHMEM
1585        {
1586                .procname       = "highmem_is_dirtyable",
1587                .data           = &vm_highmem_is_dirtyable,
1588                .maxlen         = sizeof(vm_highmem_is_dirtyable),
1589                .mode           = 0644,
1590                .proc_handler   = proc_dointvec_minmax,
1591                .extra1         = &zero,
1592                .extra2         = &one,
1593        },
1594#endif
1595#ifdef CONFIG_MEMORY_FAILURE
1596        {
1597                .procname       = "memory_failure_early_kill",
1598                .data           = &sysctl_memory_failure_early_kill,
1599                .maxlen         = sizeof(sysctl_memory_failure_early_kill),
1600                .mode           = 0644,
1601                .proc_handler   = proc_dointvec_minmax,
1602                .extra1         = &zero,
1603                .extra2         = &one,
1604        },
1605        {
1606                .procname       = "memory_failure_recovery",
1607                .data           = &sysctl_memory_failure_recovery,
1608                .maxlen         = sizeof(sysctl_memory_failure_recovery),
1609                .mode           = 0644,
1610                .proc_handler   = proc_dointvec_minmax,
1611                .extra1         = &zero,
1612                .extra2         = &one,
1613        },
1614#endif
1615        {
1616                .procname       = "user_reserve_kbytes",
1617                .data           = &sysctl_user_reserve_kbytes,
1618                .maxlen         = sizeof(sysctl_user_reserve_kbytes),
1619                .mode           = 0644,
1620                .proc_handler   = proc_doulongvec_minmax,
1621        },
1622        {
1623                .procname       = "admin_reserve_kbytes",
1624                .data           = &sysctl_admin_reserve_kbytes,
1625                .maxlen         = sizeof(sysctl_admin_reserve_kbytes),
1626                .mode           = 0644,
1627                .proc_handler   = proc_doulongvec_minmax,
1628        },
1629#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1630        {
1631                .procname       = "mmap_rnd_bits",
1632                .data           = &mmap_rnd_bits,
1633                .maxlen         = sizeof(mmap_rnd_bits),
1634                .mode           = 0600,
1635                .proc_handler   = proc_dointvec_minmax,
1636                .extra1         = (void *)&mmap_rnd_bits_min,
1637                .extra2         = (void *)&mmap_rnd_bits_max,
1638        },
1639#endif
1640#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1641        {
1642                .procname       = "mmap_rnd_compat_bits",
1643                .data           = &mmap_rnd_compat_bits,
1644                .maxlen         = sizeof(mmap_rnd_compat_bits),
1645                .mode           = 0600,
1646                .proc_handler   = proc_dointvec_minmax,
1647                .extra1         = (void *)&mmap_rnd_compat_bits_min,
1648                .extra2         = (void *)&mmap_rnd_compat_bits_max,
1649        },
1650#endif
1651        { }
1652};
1653
1654static struct ctl_table fs_table[] = {
1655        {
1656                .procname       = "inode-nr",
1657                .data           = &inodes_stat,
1658                .maxlen         = 2*sizeof(long),
1659                .mode           = 0444,
1660                .proc_handler   = proc_nr_inodes,
1661        },
1662        {
1663                .procname       = "inode-state",
1664                .data           = &inodes_stat,
1665                .maxlen         = 7*sizeof(long),
1666                .mode           = 0444,
1667                .proc_handler   = proc_nr_inodes,
1668        },
1669        {
1670                .procname       = "file-nr",
1671                .data           = &files_stat,
1672                .maxlen         = sizeof(files_stat),
1673                .mode           = 0444,
1674                .proc_handler   = proc_nr_files,
1675        },
1676        {
1677                .procname       = "file-max",
1678                .data           = &files_stat.max_files,
1679                .maxlen         = sizeof(files_stat.max_files),
1680                .mode           = 0644,
1681                .proc_handler   = proc_doulongvec_minmax,
1682        },
1683        {
1684                .procname       = "nr_open",
1685                .data           = &sysctl_nr_open,
1686                .maxlen         = sizeof(unsigned int),
1687                .mode           = 0644,
1688                .proc_handler   = proc_dointvec_minmax,
1689                .extra1         = &sysctl_nr_open_min,
1690                .extra2         = &sysctl_nr_open_max,
1691        },
1692        {
1693                .procname       = "dentry-state",
1694                .data           = &dentry_stat,
1695                .maxlen         = 6*sizeof(long),
1696                .mode           = 0444,
1697                .proc_handler   = proc_nr_dentry,
1698        },
1699        {
1700                .procname       = "overflowuid",
1701                .data           = &fs_overflowuid,
1702                .maxlen         = sizeof(int),
1703                .mode           = 0644,
1704                .proc_handler   = proc_dointvec_minmax,
1705                .extra1         = &minolduid,
1706                .extra2         = &maxolduid,
1707        },
1708        {
1709                .procname       = "overflowgid",
1710                .data           = &fs_overflowgid,
1711                .maxlen         = sizeof(int),
1712                .mode           = 0644,
1713                .proc_handler   = proc_dointvec_minmax,
1714                .extra1         = &minolduid,
1715                .extra2         = &maxolduid,
1716        },
1717#ifdef CONFIG_FILE_LOCKING
1718        {
1719                .procname       = "leases-enable",
1720                .data           = &leases_enable,
1721                .maxlen         = sizeof(int),
1722                .mode           = 0644,
1723                .proc_handler   = proc_dointvec,
1724        },
1725#endif
1726#ifdef CONFIG_DNOTIFY
1727        {
1728                .procname       = "dir-notify-enable",
1729                .data           = &dir_notify_enable,
1730                .maxlen         = sizeof(int),
1731                .mode           = 0644,
1732                .proc_handler   = proc_dointvec,
1733        },
1734#endif
1735#ifdef CONFIG_MMU
1736#ifdef CONFIG_FILE_LOCKING
1737        {
1738                .procname       = "lease-break-time",
1739                .data           = &lease_break_time,
1740                .maxlen         = sizeof(int),
1741                .mode           = 0644,
1742                .proc_handler   = proc_dointvec,
1743        },
1744#endif
1745#ifdef CONFIG_AIO
1746        {
1747                .procname       = "aio-nr",
1748                .data           = &aio_nr,
1749                .maxlen         = sizeof(aio_nr),
1750                .mode           = 0444,
1751                .proc_handler   = proc_doulongvec_minmax,
1752        },
1753        {
1754                .procname       = "aio-max-nr",
1755                .data           = &aio_max_nr,
1756                .maxlen         = sizeof(aio_max_nr),
1757                .mode           = 0644,
1758                .proc_handler   = proc_doulongvec_minmax,
1759        },
1760#endif /* CONFIG_AIO */
1761#ifdef CONFIG_INOTIFY_USER
1762        {
1763                .procname       = "inotify",
1764                .mode           = 0555,
1765                .child          = inotify_table,
1766        },
1767#endif  
1768#ifdef CONFIG_EPOLL
1769        {
1770                .procname       = "epoll",
1771                .mode           = 0555,
1772                .child          = epoll_table,
1773        },
1774#endif
1775#endif
1776        {
1777                .procname       = "protected_symlinks",
1778                .data           = &sysctl_protected_symlinks,
1779                .maxlen         = sizeof(int),
1780                .mode           = 0600,
1781                .proc_handler   = proc_dointvec_minmax,
1782                .extra1         = &zero,
1783                .extra2         = &one,
1784        },
1785        {
1786                .procname       = "protected_hardlinks",
1787                .data           = &sysctl_protected_hardlinks,
1788                .maxlen         = sizeof(int),
1789                .mode           = 0600,
1790                .proc_handler   = proc_dointvec_minmax,
1791                .extra1         = &zero,
1792                .extra2         = &one,
1793        },
1794        {
1795                .procname       = "suid_dumpable",
1796                .data           = &suid_dumpable,
1797                .maxlen         = sizeof(int),
1798                .mode           = 0644,
1799                .proc_handler   = proc_dointvec_minmax_coredump,
1800                .extra1         = &zero,
1801                .extra2         = &two,
1802        },
1803#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1804        {
1805                .procname       = "binfmt_misc",
1806                .mode           = 0555,
1807                .child          = sysctl_mount_point,
1808        },
1809#endif
1810        {
1811                .procname       = "pipe-max-size",
1812                .data           = &pipe_max_size,
1813                .maxlen         = sizeof(int),
1814                .mode           = 0644,
1815                .proc_handler   = &pipe_proc_fn,
1816                .extra1         = &pipe_min_size,
1817        },
1818        {
1819                .procname       = "pipe-user-pages-hard",
1820                .data           = &pipe_user_pages_hard,
1821                .maxlen         = sizeof(pipe_user_pages_hard),
1822                .mode           = 0644,
1823                .proc_handler   = proc_doulongvec_minmax,
1824        },
1825        {
1826                .procname       = "pipe-user-pages-soft",
1827                .data           = &pipe_user_pages_soft,
1828                .maxlen         = sizeof(pipe_user_pages_soft),
1829                .mode           = 0644,
1830                .proc_handler   = proc_doulongvec_minmax,
1831        },
1832        {
1833                .procname       = "mount-max",
1834                .data           = &sysctl_mount_max,
1835                .maxlen         = sizeof(unsigned int),
1836                .mode           = 0644,
1837                .proc_handler   = proc_dointvec_minmax,
1838                .extra1         = &one,
1839        },
1840        { }
1841};
1842
1843static struct ctl_table debug_table[] = {
1844#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1845        {
1846                .procname       = "exception-trace",
1847                .data           = &show_unhandled_signals,
1848                .maxlen         = sizeof(int),
1849                .mode           = 0644,
1850                .proc_handler   = proc_dointvec
1851        },
1852#endif
1853#if defined(CONFIG_OPTPROBES)
1854        {
1855                .procname       = "kprobes-optimization",
1856                .data           = &sysctl_kprobes_optimization,
1857                .maxlen         = sizeof(int),
1858                .mode           = 0644,
1859                .proc_handler   = proc_kprobes_optimization_handler,
1860                .extra1         = &zero,
1861                .extra2         = &one,
1862        },
1863#endif
1864        { }
1865};
1866
1867static struct ctl_table dev_table[] = {
1868        { }
1869};
1870
1871int __init sysctl_init(void)
1872{
1873        struct ctl_table_header *hdr;
1874
1875        hdr = register_sysctl_table(sysctl_base_table);
1876        kmemleak_not_leak(hdr);
1877        return 0;
1878}
1879
1880#endif /* CONFIG_SYSCTL */
1881
1882/*
1883 * /proc/sys support
1884 */
1885
1886#ifdef CONFIG_PROC_SYSCTL
1887
1888static int _proc_do_string(char *data, int maxlen, int write,
1889                           char __user *buffer,
1890                           size_t *lenp, loff_t *ppos)
1891{
1892        size_t len;
1893        char __user *p;
1894        char c;
1895
1896        if (!data || !maxlen || !*lenp) {
1897                *lenp = 0;
1898                return 0;
1899        }
1900
1901        if (write) {
1902                if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1903                        /* Only continue writes not past the end of buffer. */
1904                        len = strlen(data);
1905                        if (len > maxlen - 1)
1906                                len = maxlen - 1;
1907
1908                        if (*ppos > len)
1909                                return 0;
1910                        len = *ppos;
1911                } else {
1912                        /* Start writing from beginning of buffer. */
1913                        len = 0;
1914                }
1915
1916                *ppos += *lenp;
1917                p = buffer;
1918                while ((p - buffer) < *lenp && len < maxlen - 1) {
1919                        if (get_user(c, p++))
1920                                return -EFAULT;
1921                        if (c == 0 || c == '\n')
1922                                break;
1923                        data[len++] = c;
1924                }
1925                data[len] = 0;
1926        } else {
1927                len = strlen(data);
1928                if (len > maxlen)
1929                        len = maxlen;
1930
1931                if (*ppos > len) {
1932                        *lenp = 0;
1933                        return 0;
1934                }
1935
1936                data += *ppos;
1937                len  -= *ppos;
1938
1939                if (len > *lenp)
1940                        len = *lenp;
1941                if (len)
1942                        if (copy_to_user(buffer, data, len))
1943                                return -EFAULT;
1944                if (len < *lenp) {
1945                        if (put_user('\n', buffer + len))
1946                                return -EFAULT;
1947                        len++;
1948                }
1949                *lenp = len;
1950                *ppos += len;
1951        }
1952        return 0;
1953}
1954
1955static void warn_sysctl_write(struct ctl_table *table)
1956{
1957        pr_warn_once("%s wrote to %s when file position was not 0!\n"
1958                "This will not be supported in the future. To silence this\n"
1959                "warning, set kernel.sysctl_writes_strict = -1\n",
1960                current->comm, table->procname);
1961}
1962
1963/**
1964 * proc_dostring - read a string sysctl
1965 * @table: the sysctl table
1966 * @write: %TRUE if this is a write to the sysctl file
1967 * @buffer: the user buffer
1968 * @lenp: the size of the user buffer
1969 * @ppos: file position
1970 *
1971 * Reads/writes a string from/to the user buffer. If the kernel
1972 * buffer provided is not large enough to hold the string, the
1973 * string is truncated. The copied string is %NULL-terminated.
1974 * If the string is being read by the user process, it is copied
1975 * and a newline '\n' is added. It is truncated if the buffer is
1976 * not large enough.
1977 *
1978 * Returns 0 on success.
1979 */
1980int proc_dostring(struct ctl_table *table, int write,
1981                  void __user *buffer, size_t *lenp, loff_t *ppos)
1982{
1983        if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1984                warn_sysctl_write(table);
1985
1986        return _proc_do_string((char *)(table->data), table->maxlen, write,
1987                               (char __user *)buffer, lenp, ppos);
1988}
1989
1990static size_t proc_skip_spaces(char **buf)
1991{
1992        size_t ret;
1993        char *tmp = skip_spaces(*buf);
1994        ret = tmp - *buf;
1995        *buf = tmp;
1996        return ret;
1997}
1998
1999static void proc_skip_char(char **buf, size_t *size, const char v)
2000{
2001        while (*size) {
2002                if (**buf != v)
2003                        break;
2004                (*size)--;
2005                (*buf)++;
2006        }
2007}
2008
2009#define TMPBUFLEN 22
2010/**
2011 * proc_get_long - reads an ASCII formatted integer from a user buffer
2012 *
2013 * @buf: a kernel buffer
2014 * @size: size of the kernel buffer
2015 * @val: this is where the number will be stored
2016 * @neg: set to %TRUE if number is negative
2017 * @perm_tr: a vector which contains the allowed trailers
2018 * @perm_tr_len: size of the perm_tr vector
2019 * @tr: pointer to store the trailer character
2020 *
2021 * In case of success %0 is returned and @buf and @size are updated with
2022 * the amount of bytes read. If @tr is non-NULL and a trailing
2023 * character exists (size is non-zero after returning from this
2024 * function), @tr is updated with the trailing character.
2025 */
2026static int proc_get_long(char **buf, size_t *size,
2027                          unsigned long *val, bool *neg,
2028                          const char *perm_tr, unsigned perm_tr_len, char *tr)
2029{
2030        int len;
2031        char *p, tmp[TMPBUFLEN];
2032
2033        if (!*size)
2034                return -EINVAL;
2035
2036        len = *size;
2037        if (len > TMPBUFLEN - 1)
2038                len = TMPBUFLEN - 1;
2039
2040        memcpy(tmp, *buf, len);
2041
2042        tmp[len] = 0;
2043        p = tmp;
2044        if (*p == '-' && *size > 1) {
2045                *neg = true;
2046                p++;
2047        } else
2048                *neg = false;
2049        if (!isdigit(*p))
2050                return -EINVAL;
2051
2052        *val = simple_strtoul(p, &p, 0);
2053
2054        len = p - tmp;
2055
2056        /* We don't know if the next char is whitespace thus we may accept
2057         * invalid integers (e.g. 1234...a) or two integers instead of one
2058         * (e.g. 123...1). So lets not allow such large numbers. */
2059        if (len == TMPBUFLEN - 1)
2060                return -EINVAL;
2061
2062        if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2063                return -EINVAL;
2064
2065        if (tr && (len < *size))
2066                *tr = *p;
2067
2068        *buf += len;
2069        *size -= len;
2070
2071        return 0;
2072}
2073
2074/**
2075 * proc_put_long - converts an integer to a decimal ASCII formatted string
2076 *
2077 * @buf: the user buffer
2078 * @size: the size of the user buffer
2079 * @val: the integer to be converted
2080 * @neg: sign of the number, %TRUE for negative
2081 *
2082 * In case of success %0 is returned and @buf and @size are updated with
2083 * the amount of bytes written.
2084 */
2085static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2086                          bool neg)
2087{
2088        int len;
2089        char tmp[TMPBUFLEN], *p = tmp;
2090
2091        sprintf(p, "%s%lu", neg ? "-" : "", val);
2092        len = strlen(tmp);
2093        if (len > *size)
2094                len = *size;
2095        if (copy_to_user(*buf, tmp, len))
2096                return -EFAULT;
2097        *size -= len;
2098        *buf += len;
2099        return 0;
2100}
2101#undef TMPBUFLEN
2102
2103static int proc_put_char(void __user **buf, size_t *size, char c)
2104{
2105        if (*size) {
2106                char __user **buffer = (char __user **)buf;
2107                if (put_user(c, *buffer))
2108                        return -EFAULT;
2109                (*size)--, (*buffer)++;
2110                *buf = *buffer;
2111        }
2112        return 0;
2113}
2114
2115static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2116                                 int *valp,
2117                                 int write, void *data)
2118{
2119        if (write) {
2120                if (*negp) {
2121                        if (*lvalp > (unsigned long) INT_MAX + 1)
2122                                return -EINVAL;
2123                        *valp = -*lvalp;
2124                } else {
2125                        if (*lvalp > (unsigned long) INT_MAX)
2126                                return -EINVAL;
2127                        *valp = *lvalp;
2128                }
2129        } else {
2130                int val = *valp;
2131                if (val < 0) {
2132                        *negp = true;
2133                        *lvalp = -(unsigned long)val;
2134                } else {
2135                        *negp = false;
2136                        *lvalp = (unsigned long)val;
2137                }
2138        }
2139        return 0;
2140}
2141
2142static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2143                                 int *valp,
2144                                 int write, void *data)
2145{
2146        if (write) {
2147                if (*negp)
2148                        return -EINVAL;
2149                *valp = *lvalp;
2150        } else {
2151                unsigned int val = *valp;
2152                *lvalp = (unsigned long)val;
2153        }
2154        return 0;
2155}
2156
2157static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2158
2159static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2160                  int write, void __user *buffer,
2161                  size_t *lenp, loff_t *ppos,
2162                  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2163                              int write, void *data),
2164                  void *data)
2165{
2166        int *i, vleft, first = 1, err = 0;
2167        size_t left;
2168        char *kbuf = NULL, *p;
2169        
2170        if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2171                *lenp = 0;
2172                return 0;
2173        }
2174        
2175        i = (int *) tbl_data;
2176        vleft = table->maxlen / sizeof(*i);
2177        left = *lenp;
2178
2179        if (!conv)
2180                conv = do_proc_dointvec_conv;
2181
2182        if (write) {
2183                if (*ppos) {
2184                        switch (sysctl_writes_strict) {
2185                        case SYSCTL_WRITES_STRICT:
2186                                goto out;
2187                        case SYSCTL_WRITES_WARN:
2188                                warn_sysctl_write(table);
2189                                break;
2190                        default:
2191                                break;
2192                        }
2193                }
2194
2195                if (left > PAGE_SIZE - 1)
2196                        left = PAGE_SIZE - 1;
2197                p = kbuf = memdup_user_nul(buffer, left);
2198                if (IS_ERR(kbuf))
2199                        return PTR_ERR(kbuf);
2200        }
2201
2202        for (; left && vleft--; i++, first=0) {
2203                unsigned long lval;
2204                bool neg;
2205
2206                if (write) {
2207                        left -= proc_skip_spaces(&p);
2208
2209                        if (!left)
2210                                break;
2211                        err = proc_get_long(&p, &left, &lval, &neg,
2212                                             proc_wspace_sep,
2213                                             sizeof(proc_wspace_sep), NULL);
2214                        if (err)
2215                                break;
2216                        if (conv(&neg, &lval, i, 1, data)) {
2217                                err = -EINVAL;
2218                                break;
2219                        }
2220                } else {
2221                        if (conv(&neg, &lval, i, 0, data)) {
2222                                err = -EINVAL;
2223                                break;
2224                        }
2225                        if (!first)
2226                                err = proc_put_char(&buffer, &left, '\t');
2227                        if (err)
2228                                break;
2229                        err = proc_put_long(&buffer, &left, lval, neg);
2230                        if (err)
2231                                break;
2232                }
2233        }
2234
2235        if (!write && !first && left && !err)
2236                err = proc_put_char(&buffer, &left, '\n');
2237        if (write && !err && left)
2238                left -= proc_skip_spaces(&p);
2239        if (write) {
2240                kfree(kbuf);
2241                if (first)
2242                        return err ? : -EINVAL;
2243        }
2244        *lenp -= left;
2245out:
2246        *ppos += *lenp;
2247        return err;
2248}
2249
2250static int do_proc_dointvec(struct ctl_table *table, int write,
2251                  void __user *buffer, size_t *lenp, loff_t *ppos,
2252                  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2253                              int write, void *data),
2254                  void *data)
2255{
2256        return __do_proc_dointvec(table->data, table, write,
2257                        buffer, lenp, ppos, conv, data);
2258}
2259
2260/**
2261 * proc_dointvec - read a vector of integers
2262 * @table: the sysctl table
2263 * @write: %TRUE if this is a write to the sysctl file
2264 * @buffer: the user buffer
2265 * @lenp: the size of the user buffer
2266 * @ppos: file position
2267 *
2268 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2269 * values from/to the user buffer, treated as an ASCII string. 
2270 *
2271 * Returns 0 on success.
2272 */
2273int proc_dointvec(struct ctl_table *table, int write,
2274                     void __user *buffer, size_t *lenp, loff_t *ppos)
2275{
2276        return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2277}
2278
2279/**
2280 * proc_douintvec - read a vector of unsigned integers
2281 * @table: the sysctl table
2282 * @write: %TRUE if this is a write to the sysctl file
2283 * @buffer: the user buffer
2284 * @lenp: the size of the user buffer
2285 * @ppos: file position
2286 *
2287 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2288 * values from/to the user buffer, treated as an ASCII string.
2289 *
2290 * Returns 0 on success.
2291 */
2292int proc_douintvec(struct ctl_table *table, int write,
2293                     void __user *buffer, size_t *lenp, loff_t *ppos)
2294{
2295        return do_proc_dointvec(table, write, buffer, lenp, ppos,
2296                                do_proc_douintvec_conv, NULL);
2297}
2298
2299/*
2300 * Taint values can only be increased
2301 * This means we can safely use a temporary.
2302 */
2303static int proc_taint(struct ctl_table *table, int write,
2304                               void __user *buffer, size_t *lenp, loff_t *ppos)
2305{
2306        struct ctl_table t;
2307        unsigned long tmptaint = get_taint();
2308        int err;
2309
2310        if (write && !capable(CAP_SYS_ADMIN))
2311                return -EPERM;
2312
2313        t = *table;
2314        t.data = &tmptaint;
2315        err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2316        if (err < 0)
2317                return err;
2318
2319        if (write) {
2320                /*
2321                 * Poor man's atomic or. Not worth adding a primitive
2322                 * to everyone's atomic.h for this
2323                 */
2324                int i;
2325                for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2326                        if ((tmptaint >> i) & 1)
2327                                add_taint(i, LOCKDEP_STILL_OK);
2328                }
2329        }
2330
2331        return err;
2332}
2333
2334#ifdef CONFIG_PRINTK
2335static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2336                                void __user *buffer, size_t *lenp, loff_t *ppos)
2337{
2338        if (write && !capable(CAP_SYS_ADMIN))
2339                return -EPERM;
2340
2341        return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2342}
2343#endif
2344
2345struct do_proc_dointvec_minmax_conv_param {
2346        int *min;
2347        int *max;
2348};
2349
2350static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2351                                        int *valp,
2352                                        int write, void *data)
2353{
2354        struct do_proc_dointvec_minmax_conv_param *param = data;
2355        if (write) {
2356                int val = *negp ? -*lvalp : *lvalp;
2357                if ((param->min && *param->min > val) ||
2358                    (param->max && *param->max < val))
2359                        return -EINVAL;
2360                *valp = val;
2361        } else {
2362                int val = *valp;
2363                if (val < 0) {
2364                        *negp = true;
2365                        *lvalp = -(unsigned long)val;
2366                } else {
2367                        *negp = false;
2368                        *lvalp = (unsigned long)val;
2369                }
2370        }
2371        return 0;
2372}
2373
2374/**
2375 * proc_dointvec_minmax - read a vector of integers with min/max values
2376 * @table: the sysctl table
2377 * @write: %TRUE if this is a write to the sysctl file
2378 * @buffer: the user buffer
2379 * @lenp: the size of the user buffer
2380 * @ppos: file position
2381 *
2382 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2383 * values from/to the user buffer, treated as an ASCII string.
2384 *
2385 * This routine will ensure the values are within the range specified by
2386 * table->extra1 (min) and table->extra2 (max).
2387 *
2388 * Returns 0 on success.
2389 */
2390int proc_dointvec_minmax(struct ctl_table *table, int write,
2391                  void __user *buffer, size_t *lenp, loff_t *ppos)
2392{
2393        struct do_proc_dointvec_minmax_conv_param param = {
2394                .min = (int *) table->extra1,
2395                .max = (int *) table->extra2,
2396        };
2397        return do_proc_dointvec(table, write, buffer, lenp, ppos,
2398                                do_proc_dointvec_minmax_conv, &param);
2399}
2400
2401static void validate_coredump_safety(void)
2402{
2403#ifdef CONFIG_COREDUMP
2404        if (suid_dumpable == SUID_DUMP_ROOT &&
2405            core_pattern[0] != '/' && core_pattern[0] != '|') {
2406                printk(KERN_WARNING "Unsafe core_pattern used with "\
2407                        "suid_dumpable=2. Pipe handler or fully qualified "\
2408                        "core dump path required.\n");
2409        }
2410#endif
2411}
2412
2413static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2414                void __user *buffer, size_t *lenp, loff_t *ppos)
2415{
2416        int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2417        if (!error)
2418                validate_coredump_safety();
2419        return error;
2420}
2421
2422#ifdef CONFIG_COREDUMP
2423static int proc_dostring_coredump(struct ctl_table *table, int write,
2424                  void __user *buffer, size_t *lenp, loff_t *ppos)
2425{
2426        int error = proc_dostring(table, write, buffer, lenp, ppos);
2427        if (!error)
2428                validate_coredump_safety();
2429        return error;
2430}
2431#endif
2432
2433static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2434                                     void __user *buffer,
2435                                     size_t *lenp, loff_t *ppos,
2436                                     unsigned long convmul,
2437                                     unsigned long convdiv)
2438{
2439        unsigned long *i, *min, *max;
2440        int vleft, first = 1, err = 0;
2441        size_t left;
2442        char *kbuf = NULL, *p;
2443
2444        if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2445                *lenp = 0;
2446                return 0;
2447        }
2448
2449        i = (unsigned long *) data;
2450        min = (unsigned long *) table->extra1;
2451        max = (unsigned long *) table->extra2;
2452        vleft = table->maxlen / sizeof(unsigned long);
2453        left = *lenp;
2454
2455        if (write) {
2456                if (*ppos) {
2457                        switch (sysctl_writes_strict) {
2458                        case SYSCTL_WRITES_STRICT:
2459                                goto out;
2460                        case SYSCTL_WRITES_WARN:
2461                                warn_sysctl_write(table);
2462                                break;
2463                        default:
2464                                break;
2465                        }
2466                }
2467
2468                if (left > PAGE_SIZE - 1)
2469                        left = PAGE_SIZE - 1;
2470                p = kbuf = memdup_user_nul(buffer, left);
2471                if (IS_ERR(kbuf))
2472                        return PTR_ERR(kbuf);
2473        }
2474
2475        for (; left && vleft--; i++, first = 0) {
2476                unsigned long val;
2477
2478                if (write) {
2479                        bool neg;
2480
2481                        left -= proc_skip_spaces(&p);
2482
2483                        err = proc_get_long(&p, &left, &val, &neg,
2484                                             proc_wspace_sep,
2485                                             sizeof(proc_wspace_sep), NULL);
2486                        if (err)
2487                                break;
2488                        if (neg)
2489                                continue;
2490                        if ((min && val < *min) || (max && val > *max))
2491                                continue;
2492                        *i = val;
2493                } else {
2494                        val = convdiv * (*i) / convmul;
2495                        if (!first) {
2496                                err = proc_put_char(&buffer, &left, '\t');
2497                                if (err)
2498                                        break;
2499                        }
2500                        err = proc_put_long(&buffer, &left, val, false);
2501                        if (err)
2502                                break;
2503                }
2504        }
2505
2506        if (!write && !first && left && !err)
2507                err = proc_put_char(&buffer, &left, '\n');
2508        if (write && !err)
2509                left -= proc_skip_spaces(&p);
2510        if (write) {
2511                kfree(kbuf);
2512                if (first)
2513                        return err ? : -EINVAL;
2514        }
2515        *lenp -= left;
2516out:
2517        *ppos += *lenp;
2518        return err;
2519}
2520
2521static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2522                                     void __user *buffer,
2523                                     size_t *lenp, loff_t *ppos,
2524                                     unsigned long convmul,
2525                                     unsigned long convdiv)
2526{
2527        return __do_proc_doulongvec_minmax(table->data, table, write,
2528                        buffer, lenp, ppos, convmul, convdiv);
2529}
2530
2531/**
2532 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2533 * @table: the sysctl table
2534 * @write: %TRUE if this is a write to the sysctl file
2535 * @buffer: the user buffer
2536 * @lenp: the size of the user buffer
2537 * @ppos: file position
2538 *
2539 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2540 * values from/to the user buffer, treated as an ASCII string.
2541 *
2542 * This routine will ensure the values are within the range specified by
2543 * table->extra1 (min) and table->extra2 (max).
2544 *
2545 * Returns 0 on success.
2546 */
2547int proc_doulongvec_minmax(struct ctl_table *table, int write,
2548                           void __user *buffer, size_t *lenp, loff_t *ppos)
2549{
2550    return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2551}
2552
2553/**
2554 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2555 * @table: the sysctl table
2556 * @write: %TRUE if this is a write to the sysctl file
2557 * @buffer: the user buffer
2558 * @lenp: the size of the user buffer
2559 * @ppos: file position
2560 *
2561 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2562 * values from/to the user buffer, treated as an ASCII string. The values
2563 * are treated as milliseconds, and converted to jiffies when they are stored.
2564 *
2565 * This routine will ensure the values are within the range specified by
2566 * table->extra1 (min) and table->extra2 (max).
2567 *
2568 * Returns 0 on success.
2569 */
2570int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2571                                      void __user *buffer,
2572                                      size_t *lenp, loff_t *ppos)
2573{
2574    return do_proc_doulongvec_minmax(table, write, buffer,
2575                                     lenp, ppos, HZ, 1000l);
2576}
2577
2578
2579static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2580                                         int *valp,
2581                                         int write, void *data)
2582{
2583        if (write) {
2584                if (*lvalp > LONG_MAX / HZ)
2585                        return 1;
2586                *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2587        } else {
2588                int val = *valp;
2589                unsigned long lval;
2590                if (val < 0) {
2591                        *negp = true;
2592                        lval = -(unsigned long)val;
2593                } else {
2594                        *negp = false;
2595                        lval = (unsigned long)val;
2596                }
2597                *lvalp = lval / HZ;
2598        }
2599        return 0;
2600}
2601
2602static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2603                                                int *valp,
2604                                                int write, void *data)
2605{
2606        if (write) {
2607                if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2608                        return 1;
2609                *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2610        } else {
2611                int val = *valp;
2612                unsigned long lval;
2613                if (val < 0) {
2614                        *negp = true;
2615                        lval = -(unsigned long)val;
2616                } else {
2617                        *negp = false;
2618                        lval = (unsigned long)val;
2619                }
2620                *lvalp = jiffies_to_clock_t(lval);
2621        }
2622        return 0;
2623}
2624
2625static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2626                                            int *valp,
2627                                            int write, void *data)
2628{
2629        if (write) {
2630                unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2631
2632                if (jif > INT_MAX)
2633                        return 1;
2634                *valp = (int)jif;
2635        } else {
2636                int val = *valp;
2637                unsigned long lval;
2638                if (val < 0) {
2639                        *negp = true;
2640                        lval = -(unsigned long)val;
2641                } else {
2642                        *negp = false;
2643                        lval = (unsigned long)val;
2644                }
2645                *lvalp = jiffies_to_msecs(lval);
2646        }
2647        return 0;
2648}
2649
2650/**
2651 * proc_dointvec_jiffies - read a vector of integers as seconds
2652 * @table: the sysctl table
2653 * @write: %TRUE if this is a write to the sysctl file
2654 * @buffer: the user buffer
2655 * @lenp: the size of the user buffer
2656 * @ppos: file position
2657 *
2658 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2659 * values from/to the user buffer, treated as an ASCII string. 
2660 * The values read are assumed to be in seconds, and are converted into
2661 * jiffies.
2662 *
2663 * Returns 0 on success.
2664 */
2665int proc_dointvec_jiffies(struct ctl_table *table, int write,
2666                          void __user *buffer, size_t *lenp, loff_t *ppos)
2667{
2668    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2669                            do_proc_dointvec_jiffies_conv,NULL);
2670}
2671
2672/**
2673 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2674 * @table: the sysctl table
2675 * @write: %TRUE if this is a write to the sysctl file
2676 * @buffer: the user buffer
2677 * @lenp: the size of the user buffer
2678 * @ppos: pointer to the file position
2679 *
2680 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2681 * values from/to the user buffer, treated as an ASCII string. 
2682 * The values read are assumed to be in 1/USER_HZ seconds, and 
2683 * are converted into jiffies.
2684 *
2685 * Returns 0 on success.
2686 */
2687int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2688                                 void __user *buffer, size_t *lenp, loff_t *ppos)
2689{
2690    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2691                            do_proc_dointvec_userhz_jiffies_conv,NULL);
2692}
2693
2694/**
2695 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2696 * @table: the sysctl table
2697 * @write: %TRUE if this is a write to the sysctl file
2698 * @buffer: the user buffer
2699 * @lenp: the size of the user buffer
2700 * @ppos: file position
2701 * @ppos: the current position in the file
2702 *
2703 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2704 * values from/to the user buffer, treated as an ASCII string. 
2705 * The values read are assumed to be in 1/1000 seconds, and 
2706 * are converted into jiffies.
2707 *
2708 * Returns 0 on success.
2709 */
2710int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2711                             void __user *buffer, size_t *lenp, loff_t *ppos)
2712{
2713        return do_proc_dointvec(table, write, buffer, lenp, ppos,
2714                                do_proc_dointvec_ms_jiffies_conv, NULL);
2715}
2716
2717static int proc_do_cad_pid(struct ctl_table *table, int write,
2718                           void __user *buffer, size_t *lenp, loff_t *ppos)
2719{
2720        struct pid *new_pid;
2721        pid_t tmp;
2722        int r;
2723
2724        tmp = pid_vnr(cad_pid);
2725
2726        r = __do_proc_dointvec(&tmp, table, write, buffer,
2727                               lenp, ppos, NULL, NULL);
2728        if (r || !write)
2729                return r;
2730
2731        new_pid = find_get_pid(tmp);
2732        if (!new_pid)
2733                return -ESRCH;
2734
2735        put_pid(xchg(&cad_pid, new_pid));
2736        return 0;
2737}
2738
2739/**
2740 * proc_do_large_bitmap - read/write from/to a large bitmap
2741 * @table: the sysctl table
2742 * @write: %TRUE if this is a write to the sysctl file
2743 * @buffer: the user buffer
2744 * @lenp: the size of the user buffer
2745 * @ppos: file position
2746 *
2747 * The bitmap is stored at table->data and the bitmap length (in bits)
2748 * in table->maxlen.
2749 *
2750 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2751 * large bitmaps may be represented in a compact manner. Writing into
2752 * the file will clear the bitmap then update it with the given input.
2753 *
2754 * Returns 0 on success.
2755 */
2756int proc_do_large_bitmap(struct ctl_table *table, int write,
2757                         void __user *buffer, size_t *lenp, loff_t *ppos)
2758{
2759        int err = 0;
2760        bool first = 1;
2761        size_t left = *lenp;
2762        unsigned long bitmap_len = table->maxlen;
2763        unsigned long *bitmap = *(unsigned long **) table->data;
2764        unsigned long *tmp_bitmap = NULL;
2765        char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2766
2767        if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2768                *lenp = 0;
2769                return 0;
2770        }
2771
2772        if (write) {
2773                char *kbuf, *p;
2774
2775                if (left > PAGE_SIZE - 1)
2776                        left = PAGE_SIZE - 1;
2777
2778                p = kbuf = memdup_user_nul(buffer, left);
2779                if (IS_ERR(kbuf))
2780                        return PTR_ERR(kbuf);
2781
2782                tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2783                                     GFP_KERNEL);
2784                if (!tmp_bitmap) {
2785                        kfree(kbuf);
2786                        return -ENOMEM;
2787                }
2788                proc_skip_char(&p, &left, '\n');
2789                while (!err && left) {
2790                        unsigned long val_a, val_b;
2791                        bool neg;
2792
2793                        err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2794                                             sizeof(tr_a), &c);
2795                        if (err)
2796                                break;
2797                        if (val_a >= bitmap_len || neg) {
2798                                err = -EINVAL;
2799                                break;
2800                        }
2801
2802                        val_b = val_a;
2803                        if (left) {
2804                                p++;
2805                                left--;
2806                        }
2807
2808                        if (c == '-') {
2809                                err = proc_get_long(&p, &left, &val_b,
2810                                                     &neg, tr_b, sizeof(tr_b),
2811                                                     &c);
2812                                if (err)
2813                                        break;
2814                                if (val_b >= bitmap_len || neg ||
2815                                    val_a > val_b) {
2816                                        err = -EINVAL;
2817                                        break;
2818                                }
2819                                if (left) {
2820                                        p++;
2821                                        left--;
2822                                }
2823                        }
2824
2825                        bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2826                        first = 0;
2827                        proc_skip_char(&p, &left, '\n');
2828                }
2829                kfree(kbuf);
2830        } else {
2831                unsigned long bit_a, bit_b = 0;
2832
2833                while (left) {
2834                        bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2835                        if (bit_a >= bitmap_len)
2836                                break;
2837                        bit_b = find_next_zero_bit(bitmap, bitmap_len,
2838                                                   bit_a + 1) - 1;
2839
2840                        if (!first) {
2841                                err = proc_put_char(&buffer, &left, ',');
2842                                if (err)
2843                                        break;
2844                        }
2845                        err = proc_put_long(&buffer, &left, bit_a, false);
2846                        if (err)
2847                                break;
2848                        if (bit_a != bit_b) {
2849                                err = proc_put_char(&buffer, &left, '-');
2850                                if (err)
2851                                        break;
2852                                err = proc_put_long(&buffer, &left, bit_b, false);
2853                                if (err)
2854                                        break;
2855                        }
2856
2857                        first = 0; bit_b++;
2858                }
2859                if (!err)
2860                        err = proc_put_char(&buffer, &left, '\n');
2861        }
2862
2863        if (!err) {
2864                if (write) {
2865                        if (*ppos)
2866                                bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2867                        else
2868                                bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2869                }
2870                kfree(tmp_bitmap);
2871                *lenp -= left;
2872                *ppos += *lenp;
2873                return 0;
2874        } else {
2875                kfree(tmp_bitmap);
2876                return err;
2877        }
2878}
2879
2880#else /* CONFIG_PROC_SYSCTL */
2881
2882int proc_dostring(struct ctl_table *table, int write,
2883                  void __user *buffer, size_t *lenp, loff_t *ppos)
2884{
2885        return -ENOSYS;
2886}
2887
2888int proc_dointvec(struct ctl_table *table, int write,
2889                  void __user *buffer, size_t *lenp, loff_t *ppos)
2890{
2891        return -ENOSYS;
2892}
2893
2894int proc_douintvec(struct ctl_table *table, int write,
2895                  void __user *buffer, size_t *lenp, loff_t *ppos)
2896{
2897        return -ENOSYS;
2898}
2899
2900int proc_dointvec_minmax(struct ctl_table *table, int write,
2901                    void __user *buffer, size_t *lenp, loff_t *ppos)
2902{
2903        return -ENOSYS;
2904}
2905
2906int proc_dointvec_jiffies(struct ctl_table *table, int write,
2907                    void __user *buffer, size_t *lenp, loff_t *ppos)
2908{
2909        return -ENOSYS;
2910}
2911
2912int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2913                    void __user *buffer, size_t *lenp, loff_t *ppos)
2914{
2915        return -ENOSYS;
2916}
2917
2918int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2919                             void __user *buffer, size_t *lenp, loff_t *ppos)
2920{
2921        return -ENOSYS;
2922}
2923
2924int proc_doulongvec_minmax(struct ctl_table *table, int write,
2925                    void __user *buffer, size_t *lenp, loff_t *ppos)
2926{
2927        return -ENOSYS;
2928}
2929
2930int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2931                                      void __user *buffer,
2932                                      size_t *lenp, loff_t *ppos)
2933{
2934    return -ENOSYS;
2935}
2936
2937
2938#endif /* CONFIG_PROC_SYSCTL */
2939
2940/*
2941 * No sense putting this after each symbol definition, twice,
2942 * exception granted :-)
2943 */
2944EXPORT_SYMBOL(proc_dointvec);
2945EXPORT_SYMBOL(proc_douintvec);
2946EXPORT_SYMBOL(proc_dointvec_jiffies);
2947EXPORT_SYMBOL(proc_dointvec_minmax);
2948EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2949EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2950EXPORT_SYMBOL(proc_dostring);
2951EXPORT_SYMBOL(proc_doulongvec_minmax);
2952EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2953