kernel/sysctl.c

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