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/kmemleak.h>
  34 #include <linux/fs.h>
  35 #include <linux/init.h>
  36 #include <linux/kernel.h>
  37 #include <linux/kobject.h>
  38 #include <linux/net.h>
  39 #include <linux/sysrq.h>
  40 #include <linux/highuid.h>
  41 #include <linux/writeback.h>
  42 #include <linux/ratelimit.h>
  43 #include <linux/compaction.h>
  44 #include <linux/hugetlb.h>
  45 #include <linux/initrd.h>
  46 #include <linux/key.h>
  47 #include <linux/times.h>
  48 #include <linux/limits.h>
  49 #include <linux/dcache.h>
  50 #include <linux/dnotify.h>
  51 #include <linux/syscalls.h>
  52 #include <linux/vmstat.h>
  53 #include <linux/nfs_fs.h>
  54 #include <linux/acpi.h>
  55 #include <linux/reboot.h>
  56 #include <linux/ftrace.h>
  57 #include <linux/perf_event.h>
  58 #include <linux/kprobes.h>
  59 #include <linux/pipe_fs_i.h>
  60 #include <linux/oom.h>
  61 #include <linux/kmod.h>
  62 #include <linux/capability.h>
  63 #include <linux/binfmts.h>
  64 #include <linux/sched/sysctl.h>
  65
  66 #include <asm/uaccess.h>
  67 #include <asm/processor.h>
  68
  69 #ifdef CONFIG_X86
  70 #include <asm/nmi.h>
  71 #include <asm/stacktrace.h>
  72 #include <asm/io.h>
  73 #endif
  74 #ifdef CONFIG_SPARC
  75 #include <asm/setup.h>
  76 #endif
  77 #ifdef CONFIG_BSD_PROCESS_ACCT
  78 #include <linux/acct.h>
  79 #endif
  80 #ifdef CONFIG_RT_MUTEXES
  81 #include <linux/rtmutex.h>
  82 #endif
  83 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
  84 #include <linux/lockdep.h>
  85 #endif
  86 #ifdef CONFIG_CHR_DEV_SG
  87 #include <scsi/sg.h>
  88 #endif
  89
  90 #ifdef CONFIG_LOCKUP_DETECTOR
  91 #include <linux/nmi.h>
  92 #endif
  93
  94
  95 #if defined(CONFIG_SYSCTL)
  96
  97 /* External variables not in a header file. */
  98 extern int sysctl_overcommit_memory;
  99 extern int sysctl_overcommit_ratio;
 100 extern int max_threads;
 101 extern int suid_dumpable;
 102 #ifdef CONFIG_COREDUMP
 103 extern int core_uses_pid;
 104 extern char core_pattern[];
 105 extern unsigned int core_pipe_limit;
 106 #endif
 107 extern int pid_max;
 108 extern int pid_max_min, pid_max_max;
 109 extern int percpu_pagelist_fraction;
 110 extern int compat_log;
 111 extern int latencytop_enabled;
 112 extern int sysctl_nr_open_min, sysctl_nr_open_max;
 113 #ifndef CONFIG_MMU
 114 extern int sysctl_nr_trim_pages;
 115 #endif
 116 #ifdef CONFIG_BLOCK
 117 extern int blk_iopoll_enabled;
 118 #endif
 119
 120 /* Constants used for minimum and  maximum */
 121 #ifdef CONFIG_LOCKUP_DETECTOR
 122 static int sixty = 60;
 123 static int neg_one = -1;
 124 #endif
 125
 126 static int zero;
 127 static int __maybe_unused one = 1;
 128 static int __maybe_unused two = 2;
 129 static int __maybe_unused three = 3;
 130 static unsigned long one_ul = 1;
 131 static int one_hundred = 100;
 132 #ifdef CONFIG_PRINTK
 133 static int ten_thousand = 10000;
 134 #endif
 135
 136 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 137 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 138
 139 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 140 static int maxolduid = 65535;
 141 static int minolduid;
 142 static int min_percpu_pagelist_fract = 8;
 143
 144 static int ngroups_max = NGROUPS_MAX;
 145 static const int cap_last_cap = CAP_LAST_CAP;
 146
 147 #ifdef CONFIG_INOTIFY_USER
 148 #include <linux/inotify.h>
 149 #endif
 150 #ifdef CONFIG_SPARC
 151 #endif
 152
 153 #ifdef CONFIG_SPARC64
 154 extern int sysctl_tsb_ratio;
 155 #endif
 156
 157 #ifdef __hppa__
 158 extern int pwrsw_enabled;
 159 #endif
 160
 161 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 162 extern int unaligned_enabled;
 163 #endif
 164
 165 #ifdef CONFIG_IA64
 166 extern int unaligned_dump_stack;
 167 #endif
 168
 169 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
 170 extern int no_unaligned_warning;
 171 #endif
 172
 173 #ifdef CONFIG_PROC_SYSCTL
 174 static int proc_do_cad_pid(struct ctl_table *table, int write,
 175                   void __user *buffer, size_t *lenp, loff_t *ppos);
 176 static int proc_taint(struct ctl_table *table, int write,
 177                                void __user *buffer, size_t *lenp, loff_t *ppos);
 178 #endif
 179
 180 #ifdef CONFIG_PRINTK
 181 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
 182                                 void __user *buffer, size_t *lenp, loff_t *ppos);
 183 #endif
 184
 185 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
 186                 void __user *buffer, size_t *lenp, loff_t *ppos);
 187 #ifdef CONFIG_COREDUMP
 188 static int proc_dostring_coredump(struct ctl_table *table, int write,
 189                 void __user *buffer, size_t *lenp, loff_t *ppos);
 190 #endif
 191
 192 #ifdef CONFIG_MAGIC_SYSRQ
 193 /* Note: sysrq code uses it's own private copy */
 194 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
 195
 196 static int sysrq_sysctl_handler(ctl_table *table, int write,
 197                                 void __user *buffer, size_t *lenp,
 198                                 loff_t *ppos)
 199 {
 200         int error;
 201
 202         error = proc_dointvec(table, write, buffer, lenp, ppos);
 203         if (error)
 204                 return error;
 205
 206         if (write)
 207                 sysrq_toggle_support(__sysrq_enabled);
 208
 209         return 0;
 210 }
 211
 212 #endif
 213
 214 static struct ctl_table kern_table[];
 215 static struct ctl_table vm_table[];
 216 static struct ctl_table fs_table[];
 217 static struct ctl_table debug_table[];
 218 static struct ctl_table dev_table[];
 219 extern struct ctl_table random_table[];
 220 #ifdef CONFIG_EPOLL
 221 extern struct ctl_table epoll_table[];
 222 #endif
 223
 224 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 225 int sysctl_legacy_va_layout;
 226 #endif
 227
 228 /* The default sysctl tables: */
 229
 230 static struct ctl_table sysctl_base_table[] = {
 231         {
 232                 .procname       = "kernel",
 233                 .mode           = 0555,
 234                 .child          = kern_table,
 235         },
 236         {
 237                 .procname       = "vm",
 238                 .mode           = 0555,
 239                 .child          = vm_table,
 240         },
 241         {
 242                 .procname       = "fs",
 243                 .mode           = 0555,
 244                 .child          = fs_table,
 245         },
 246         {
 247                 .procname       = "debug",
 248                 .mode           = 0555,
 249                 .child          = debug_table,
 250         },
 251         {
 252                 .procname       = "dev",
 253                 .mode           = 0555,
 254                 .child          = dev_table,
 255         },
 256         { }
 257 };
 258
 259 #ifdef CONFIG_SCHED_DEBUG
 260 static int min_sched_granularity_ns = 100000;           /* 100 usecs */
 261 static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
 262 static int min_wakeup_granularity_ns;                   /* 0 usecs */
 263 static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
 264 #ifdef CONFIG_SMP
 265 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 266 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
 267 #endif /* CONFIG_SMP */
 268 #endif /* CONFIG_SCHED_DEBUG */
 269
 270 #ifdef CONFIG_COMPACTION
 271 static int min_extfrag_threshold;
 272 static int max_extfrag_threshold = 1000;
 273 #endif
 274
 275 static struct ctl_table kern_table[] = {
 276         {
 277                 .procname       = "sched_child_runs_first",
 278                 .data           = &sysctl_sched_child_runs_first,
 279                 .maxlen         = sizeof(unsigned int),
 280                 .mode           = 0644,
 281                 .proc_handler   = proc_dointvec,
 282         },
 283 #ifdef CONFIG_SCHED_DEBUG
 284         {
 285                 .procname       = "sched_min_granularity_ns",
 286                 .data           = &sysctl_sched_min_granularity,
 287                 .maxlen         = sizeof(unsigned int),
 288                 .mode           = 0644,
 289                 .proc_handler   = sched_proc_update_handler,
 290                 .extra1         = &min_sched_granularity_ns,
 291                 .extra2         = &max_sched_granularity_ns,
 292         },
 293         {
 294                 .procname       = "sched_latency_ns",
 295                 .data           = &sysctl_sched_latency,
 296                 .maxlen         = sizeof(unsigned int),
 297                 .mode           = 0644,
 298                 .proc_handler   = sched_proc_update_handler,
 299                 .extra1         = &min_sched_granularity_ns,
 300                 .extra2         = &max_sched_granularity_ns,
 301         },
 302         {
 303                 .procname       = "sched_wakeup_granularity_ns",
 304                 .data           = &sysctl_sched_wakeup_granularity,
 305                 .maxlen         = sizeof(unsigned int),
 306                 .mode           = 0644,
 307                 .proc_handler   = sched_proc_update_handler,
 308                 .extra1         = &min_wakeup_granularity_ns,
 309                 .extra2         = &max_wakeup_granularity_ns,
 310         },
 311 #ifdef CONFIG_SMP
 312         {
 313                 .procname       = "sched_tunable_scaling",
 314                 .data           = &sysctl_sched_tunable_scaling,
 315                 .maxlen         = sizeof(enum sched_tunable_scaling),
 316                 .mode           = 0644,
 317                 .proc_handler   = sched_proc_update_handler,
 318                 .extra1         = &min_sched_tunable_scaling,
 319                 .extra2         = &max_sched_tunable_scaling,
 320         },
 321         {
 322                 .procname       = "sched_migration_cost_ns",
 323                 .data           = &sysctl_sched_migration_cost,
 324                 .maxlen         = sizeof(unsigned int),
 325                 .mode           = 0644,
 326                 .proc_handler   = proc_dointvec,
 327         },
 328         {
 329                 .procname       = "sched_nr_migrate",
 330                 .data           = &sysctl_sched_nr_migrate,
 331                 .maxlen         = sizeof(unsigned int),
 332                 .mode           = 0644,
 333                 .proc_handler   = proc_dointvec,
 334         },
 335         {
 336                 .procname       = "sched_time_avg_ms",
 337                 .data           = &sysctl_sched_time_avg,
 338                 .maxlen         = sizeof(unsigned int),
 339                 .mode           = 0644,
 340                 .proc_handler   = proc_dointvec,
 341         },
 342         {
 343                 .procname       = "sched_shares_window_ns",
 344                 .data           = &sysctl_sched_shares_window,
 345                 .maxlen         = sizeof(unsigned int),
 346                 .mode           = 0644,
 347                 .proc_handler   = proc_dointvec,
 348         },
 349         {
 350                 .procname       = "timer_migration",
 351                 .data           = &sysctl_timer_migration,
 352                 .maxlen         = sizeof(unsigned int),
 353                 .mode           = 0644,
 354                 .proc_handler   = proc_dointvec_minmax,
 355                 .extra1         = &zero,
 356                 .extra2         = &one,
 357         },
 358 #endif /* CONFIG_SMP */
 359 #ifdef CONFIG_NUMA_BALANCING
 360         {
 361                 .procname       = "numa_balancing_scan_delay_ms",
 362                 .data           = &sysctl_numa_balancing_scan_delay,
 363                 .maxlen         = sizeof(unsigned int),
 364                 .mode           = 0644,
 365                 .proc_handler   = proc_dointvec,
 366         },
 367         {
 368                 .procname       = "numa_balancing_scan_period_min_ms",
 369                 .data           = &sysctl_numa_balancing_scan_period_min,
 370                 .maxlen         = sizeof(unsigned int),
 371                 .mode           = 0644,
 372                 .proc_handler   = proc_dointvec,
 373         },
 374         {
 375                 .procname       = "numa_balancing_scan_period_reset",
 376                 .data           = &sysctl_numa_balancing_scan_period_reset,
 377                 .maxlen         = sizeof(unsigned int),
 378                 .mode           = 0644,
 379                 .proc_handler   = proc_dointvec,
 380         },
 381         {
 382                 .procname       = "numa_balancing_scan_period_max_ms",
 383                 .data           = &sysctl_numa_balancing_scan_period_max,
 384                 .maxlen         = sizeof(unsigned int),
 385                 .mode           = 0644,
 386                 .proc_handler   = proc_dointvec,
 387         },
 388         {
 389                 .procname       = "numa_balancing_scan_size_mb",
 390                 .data           = &sysctl_numa_balancing_scan_size,
 391                 .maxlen         = sizeof(unsigned int),
 392                 .mode           = 0644,
 393                 .proc_handler   = proc_dointvec,
 394         },
 395 #endif /* CONFIG_NUMA_BALANCING */
 396 #endif /* CONFIG_SCHED_DEBUG */
 397         {
 398                 .procname       = "sched_rt_period_us",
 399                 .data           = &sysctl_sched_rt_period,
 400                 .maxlen         = sizeof(unsigned int),
 401                 .mode           = 0644,
 402                 .proc_handler   = sched_rt_handler,
 403         },
 404         {
 405                 .procname       = "sched_rt_runtime_us",
 406                 .data           = &sysctl_sched_rt_runtime,
 407                 .maxlen         = sizeof(int),
 408                 .mode           = 0644,
 409                 .proc_handler   = sched_rt_handler,
 410         },
 411         {
 412                 .procname       = "sched_rr_timeslice_ms",
 413                 .data           = &sched_rr_timeslice,
 414                 .maxlen         = sizeof(int),
 415                 .mode           = 0644,
 416                 .proc_handler   = sched_rr_handler,
 417         },
 418 #ifdef CONFIG_SCHED_AUTOGROUP
 419         {
 420                 .procname       = "sched_autogroup_enabled",
 421                 .data           = &sysctl_sched_autogroup_enabled,
 422                 .maxlen         = sizeof(unsigned int),
 423                 .mode           = 0644,
 424                 .proc_handler   = proc_dointvec_minmax,
 425                 .extra1         = &zero,
 426                 .extra2         = &one,
 427         },
 428 #endif
 429 #ifdef CONFIG_CFS_BANDWIDTH
 430         {
 431                 .procname       = "sched_cfs_bandwidth_slice_us",
 432                 .data           = &sysctl_sched_cfs_bandwidth_slice,
 433                 .maxlen         = sizeof(unsigned int),
 434                 .mode           = 0644,
 435                 .proc_handler   = proc_dointvec_minmax,
 436                 .extra1         = &one,
 437         },
 438 #endif
 439 #ifdef CONFIG_PROVE_LOCKING
 440         {
 441                 .procname       = "prove_locking",
 442                 .data           = &prove_locking,
 443                 .maxlen         = sizeof(int),
 444                 .mode           = 0644,
 445                 .proc_handler   = proc_dointvec,
 446         },
 447 #endif
 448 #ifdef CONFIG_LOCK_STAT
 449         {
 450                 .procname       = "lock_stat",
 451                 .data           = &lock_stat,
 452                 .maxlen         = sizeof(int),
 453                 .mode           = 0644,
 454                 .proc_handler   = proc_dointvec,
 455         },
 456 #endif
 457         {
 458                 .procname       = "panic",
 459                 .data           = &panic_timeout,
 460                 .maxlen         = sizeof(int),
 461                 .mode           = 0644,
 462                 .proc_handler   = proc_dointvec,
 463         },
 464 #ifdef CONFIG_COREDUMP
 465         {
 466                 .procname       = "core_uses_pid",
 467                 .data           = &core_uses_pid,
 468                 .maxlen         = sizeof(int),
 469                 .mode           = 0644,
 470                 .proc_handler   = proc_dointvec,
 471         },
 472         {
 473                 .procname       = "core_pattern",
 474                 .data           = core_pattern,
 475                 .maxlen         = CORENAME_MAX_SIZE,
 476                 .mode           = 0644,
 477                 .proc_handler   = proc_dostring_coredump,
 478         },
 479         {
 480                 .procname       = "core_pipe_limit",
 481                 .data           = &core_pipe_limit,
 482                 .maxlen         = sizeof(unsigned int),
 483                 .mode           = 0644,
 484                 .proc_handler   = proc_dointvec,
 485         },
 486 #endif
 487 #ifdef CONFIG_PROC_SYSCTL
 488         {
 489                 .procname       = "tainted",
 490                 .maxlen         = sizeof(long),
 491                 .mode           = 0644,
 492                 .proc_handler   = proc_taint,
 493         },
 494 #endif
 495 #ifdef CONFIG_LATENCYTOP
 496         {
 497                 .procname       = "latencytop",
 498                 .data           = &latencytop_enabled,
 499                 .maxlen         = sizeof(int),
 500                 .mode           = 0644,
 501                 .proc_handler   = proc_dointvec,
 502         },
 503 #endif
 504 #ifdef CONFIG_BLK_DEV_INITRD
 505         {
 506                 .procname       = "real-root-dev",
 507                 .data           = &real_root_dev,
 508                 .maxlen         = sizeof(int),
 509                 .mode           = 0644,
 510                 .proc_handler   = proc_dointvec,
 511         },
 512 #endif
 513         {
 514                 .procname       = "print-fatal-signals",
 515                 .data           = &print_fatal_signals,
 516                 .maxlen         = sizeof(int),
 517                 .mode           = 0644,
 518                 .proc_handler   = proc_dointvec,
 519         },
 520 #ifdef CONFIG_SPARC
 521         {
 522                 .procname       = "reboot-cmd",
 523                 .data           = reboot_command,
 524                 .maxlen         = 256,
 525                 .mode           = 0644,
 526                 .proc_handler   = proc_dostring,
 527         },
 528         {
 529                 .procname       = "stop-a",
 530                 .data           = &stop_a_enabled,
 531                 .maxlen         = sizeof (int),
 532                 .mode           = 0644,
 533                 .proc_handler   = proc_dointvec,
 534         },
 535         {
 536                 .procname       = "scons-poweroff",
 537                 .data           = &scons_pwroff,
 538                 .maxlen         = sizeof (int),
 539                 .mode           = 0644,
 540                 .proc_handler   = proc_dointvec,
 541         },
 542 #endif
 543 #ifdef CONFIG_SPARC64
 544         {
 545                 .procname       = "tsb-ratio",
 546                 .data           = &sysctl_tsb_ratio,
 547                 .maxlen         = sizeof (int),
 548                 .mode           = 0644,
 549                 .proc_handler   = proc_dointvec,
 550         },
 551 #endif
 552 #ifdef __hppa__
 553         {
 554                 .procname       = "soft-power",
 555                 .data           = &pwrsw_enabled,
 556                 .maxlen         = sizeof (int),
 557                 .mode           = 0644,
 558                 .proc_handler   = proc_dointvec,
 559         },
 560 #endif
 561 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 562         {
 563                 .procname       = "unaligned-trap",
 564                 .data           = &unaligned_enabled,
 565                 .maxlen         = sizeof (int),
 566                 .mode           = 0644,
 567                 .proc_handler   = proc_dointvec,
 568         },
 569 #endif
 570         {
 571                 .procname       = "ctrl-alt-del",
 572                 .data           = &C_A_D,
 573                 .maxlen         = sizeof(int),
 574                 .mode           = 0644,
 575                 .proc_handler   = proc_dointvec,
 576         },
 577 #ifdef CONFIG_FUNCTION_TRACER
 578         {
 579                 .procname       = "ftrace_enabled",
 580                 .data           = &ftrace_enabled,
 581                 .maxlen         = sizeof(int),
 582                 .mode           = 0644,
 583                 .proc_handler   = ftrace_enable_sysctl,
 584         },
 585 #endif
 586 #ifdef CONFIG_STACK_TRACER
 587         {
 588                 .procname       = "stack_tracer_enabled",
 589                 .data           = &stack_tracer_enabled,
 590                 .maxlen         = sizeof(int),
 591                 .mode           = 0644,
 592                 .proc_handler   = stack_trace_sysctl,
 593         },
 594 #endif
 595 #ifdef CONFIG_TRACING
 596         {
 597                 .procname       = "ftrace_dump_on_oops",
 598                 .data           = &ftrace_dump_on_oops,
 599                 .maxlen         = sizeof(int),
 600                 .mode           = 0644,
 601                 .proc_handler   = proc_dointvec,
 602         },
 603 #endif
 604 #ifdef CONFIG_MODULES
 605         {
 606                 .procname       = "modprobe",
 607                 .data           = &modprobe_path,
 608                 .maxlen         = KMOD_PATH_LEN,
 609                 .mode           = 0644,
 610                 .proc_handler   = proc_dostring,
 611         },
 612         {
 613                 .procname       = "modules_disabled",
 614                 .data           = &modules_disabled,
 615                 .maxlen         = sizeof(int),
 616                 .mode           = 0644,
 617                 /* only handle a transition from default "0" to "1" */
 618                 .proc_handler   = proc_dointvec_minmax,
 619                 .extra1         = &one,
 620                 .extra2         = &one,
 621         },
 622 #endif
 623
 624         {
 625                 .procname       = "hotplug",
 626                 .data           = &uevent_helper,
 627                 .maxlen         = UEVENT_HELPER_PATH_LEN,
 628                 .mode           = 0644,
 629                 .proc_handler   = proc_dostring,
 630         },
 631
 632 #ifdef CONFIG_CHR_DEV_SG
 633         {
 634                 .procname       = "sg-big-buff",
 635                 .data           = &sg_big_buff,
 636                 .maxlen         = sizeof (int),
 637                 .mode           = 0444,
 638                 .proc_handler   = proc_dointvec,
 639         },
 640 #endif
 641 #ifdef CONFIG_BSD_PROCESS_ACCT
 642         {
 643                 .procname       = "acct",
 644                 .data           = &acct_parm,
 645                 .maxlen         = 3*sizeof(int),
 646                 .mode           = 0644,
 647                 .proc_handler   = proc_dointvec,
 648         },
 649 #endif
 650 #ifdef CONFIG_MAGIC_SYSRQ
 651         {
 652                 .procname       = "sysrq",
 653                 .data           = &__sysrq_enabled,
 654                 .maxlen         = sizeof (int),
 655                 .mode           = 0644,
 656                 .proc_handler   = sysrq_sysctl_handler,
 657         },
 658 #endif
 659 #ifdef CONFIG_PROC_SYSCTL
 660         {
 661                 .procname       = "cad_pid",
 662                 .data           = NULL,
 663                 .maxlen         = sizeof (int),
 664                 .mode           = 0600,
 665                 .proc_handler   = proc_do_cad_pid,
 666         },
 667 #endif
 668         {
 669                 .procname       = "threads-max",
 670                 .data           = &max_threads,
 671                 .maxlen         = sizeof(int),
 672                 .mode           = 0644,
 673                 .proc_handler   = proc_dointvec,
 674         },
 675         {
 676                 .procname       = "random",
 677                 .mode           = 0555,
 678                 .child          = random_table,
 679         },
 680         {
 681                 .procname       = "usermodehelper",
 682                 .mode           = 0555,
 683                 .child          = usermodehelper_table,
 684         },
 685         {
 686                 .procname       = "overflowuid",
 687                 .data           = &overflowuid,
 688                 .maxlen         = sizeof(int),
 689                 .mode           = 0644,
 690                 .proc_handler   = proc_dointvec_minmax,
 691                 .extra1         = &minolduid,
 692                 .extra2         = &maxolduid,
 693         },
 694         {
 695                 .procname       = "overflowgid",
 696                 .data           = &overflowgid,
 697                 .maxlen         = sizeof(int),
 698                 .mode           = 0644,
 699                 .proc_handler   = proc_dointvec_minmax,
 700                 .extra1         = &minolduid,
 701                 .extra2         = &maxolduid,
 702         },
 703 #ifdef CONFIG_S390
 704 #ifdef CONFIG_MATHEMU
 705         {
 706                 .procname       = "ieee_emulation_warnings",
 707                 .data           = &sysctl_ieee_emulation_warnings,
 708                 .maxlen         = sizeof(int),
 709                 .mode           = 0644,
 710                 .proc_handler   = proc_dointvec,
 711         },
 712 #endif
 713         {
 714                 .procname       = "userprocess_debug",
 715                 .data           = &show_unhandled_signals,
 716                 .maxlen         = sizeof(int),
 717                 .mode           = 0644,
 718                 .proc_handler   = proc_dointvec,
 719         },
 720 #endif
 721         {
 722                 .procname       = "pid_max",
 723                 .data           = &pid_max,
 724                 .maxlen         = sizeof (int),
 725                 .mode           = 0644,
 726                 .proc_handler   = proc_dointvec_minmax,
 727                 .extra1         = &pid_max_min,
 728                 .extra2         = &pid_max_max,
 729         },
 730         {
 731                 .procname       = "panic_on_oops",
 732                 .data           = &panic_on_oops,
 733                 .maxlen         = sizeof(int),
 734                 .mode           = 0644,
 735                 .proc_handler   = proc_dointvec,
 736         },
 737 #if defined CONFIG_PRINTK
 738         {
 739                 .procname       = "printk",
 740                 .data           = &console_loglevel,
 741                 .maxlen         = 4*sizeof(int),
 742                 .mode           = 0644,
 743                 .proc_handler   = proc_dointvec,
 744         },
 745         {
 746                 .procname       = "printk_ratelimit",
 747                 .data           = &printk_ratelimit_state.interval,
 748                 .maxlen         = sizeof(int),
 749                 .mode           = 0644,
 750                 .proc_handler   = proc_dointvec_jiffies,
 751         },
 752         {
 753                 .procname       = "printk_ratelimit_burst",
 754                 .data           = &printk_ratelimit_state.burst,
 755                 .maxlen         = sizeof(int),
 756                 .mode           = 0644,
 757                 .proc_handler   = proc_dointvec,
 758         },
 759         {
 760                 .procname       = "printk_delay",
 761                 .data           = &printk_delay_msec,
 762                 .maxlen         = sizeof(int),
 763                 .mode           = 0644,
 764                 .proc_handler   = proc_dointvec_minmax,
 765                 .extra1         = &zero,
 766                 .extra2         = &ten_thousand,
 767         },
 768         {
 769                 .procname       = "dmesg_restrict",
 770                 .data           = &dmesg_restrict,
 771                 .maxlen         = sizeof(int),
 772                 .mode           = 0644,
 773                 .proc_handler   = proc_dointvec_minmax_sysadmin,
 774                 .extra1         = &zero,
 775                 .extra2         = &one,
 776         },
 777         {
 778                 .procname       = "kptr_restrict",
 779                 .data           = &kptr_restrict,
 780                 .maxlen         = sizeof(int),
 781                 .mode           = 0644,
 782                 .proc_handler   = proc_dointvec_minmax_sysadmin,
 783                 .extra1         = &zero,
 784                 .extra2         = &two,
 785         },
 786 #endif
 787         {
 788                 .procname       = "ngroups_max",
 789                 .data           = &ngroups_max,
 790                 .maxlen         = sizeof (int),
 791                 .mode           = 0444,
 792                 .proc_handler   = proc_dointvec,
 793         },
 794         {
 795                 .procname       = "cap_last_cap",
 796                 .data           = (void *)&cap_last_cap,
 797                 .maxlen         = sizeof(int),
 798                 .mode           = 0444,
 799                 .proc_handler   = proc_dointvec,
 800         },
 801 #if defined(CONFIG_LOCKUP_DETECTOR)
 802         {
 803                 .procname       = "watchdog",
 804                 .data           = &watchdog_enabled,
 805                 .maxlen         = sizeof (int),
 806                 .mode           = 0644,
 807                 .proc_handler   = proc_dowatchdog,
 808                 .extra1         = &zero,
 809                 .extra2         = &one,
 810         },
 811         {
 812                 .procname       = "watchdog_thresh",
 813                 .data           = &watchdog_thresh,
 814                 .maxlen         = sizeof(int),
 815                 .mode           = 0644,
 816                 .proc_handler   = proc_dowatchdog,
 817                 .extra1         = &neg_one,
 818                 .extra2         = &sixty,
 819         },
 820         {
 821                 .procname       = "softlockup_panic",
 822                 .data           = &softlockup_panic,
 823                 .maxlen         = sizeof(int),
 824                 .mode           = 0644,
 825                 .proc_handler   = proc_dointvec_minmax,
 826                 .extra1         = &zero,
 827                 .extra2         = &one,
 828         },
 829         {
 830                 .procname       = "nmi_watchdog",
 831                 .data           = &watchdog_enabled,
 832                 .maxlen         = sizeof (int),
 833                 .mode           = 0644,
 834                 .proc_handler   = proc_dowatchdog,
 835                 .extra1         = &zero,
 836                 .extra2         = &one,
 837         },
 838 #endif
 839 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 840         {
 841                 .procname       = "unknown_nmi_panic",
 842                 .data           = &unknown_nmi_panic,
 843                 .maxlen         = sizeof (int),
 844                 .mode           = 0644,
 845                 .proc_handler   = proc_dointvec,
 846         },
 847 #endif
 848 #if defined(CONFIG_X86)
 849         {
 850                 .procname       = "panic_on_unrecovered_nmi",
 851                 .data           = &panic_on_unrecovered_nmi,
 852                 .maxlen         = sizeof(int),
 853                 .mode           = 0644,
 854                 .proc_handler   = proc_dointvec,
 855         },
 856         {
 857                 .procname       = "panic_on_io_nmi",
 858                 .data           = &panic_on_io_nmi,
 859                 .maxlen         = sizeof(int),
 860                 .mode           = 0644,
 861                 .proc_handler   = proc_dointvec,
 862         },
 863 #ifdef CONFIG_DEBUG_STACKOVERFLOW
 864         {
 865                 .procname       = "panic_on_stackoverflow",
 866                 .data           = &sysctl_panic_on_stackoverflow,
 867                 .maxlen         = sizeof(int),
 868                 .mode           = 0644,
 869                 .proc_handler   = proc_dointvec,
 870         },
 871 #endif
 872         {
 873                 .procname       = "bootloader_type",
 874                 .data           = &bootloader_type,
 875                 .maxlen         = sizeof (int),
 876                 .mode           = 0444,
 877                 .proc_handler   = proc_dointvec,
 878         },
 879         {
 880                 .procname       = "bootloader_version",
 881                 .data           = &bootloader_version,
 882                 .maxlen         = sizeof (int),
 883                 .mode           = 0444,
 884                 .proc_handler   = proc_dointvec,
 885         },
 886         {
 887                 .procname       = "kstack_depth_to_print",
 888                 .data           = &kstack_depth_to_print,
 889                 .maxlen         = sizeof(int),
 890                 .mode           = 0644,
 891                 .proc_handler   = proc_dointvec,
 892         },
 893         {
 894                 .procname       = "io_delay_type",
 895                 .data           = &io_delay_type,
 896                 .maxlen         = sizeof(int),
 897                 .mode           = 0644,
 898                 .proc_handler   = proc_dointvec,
 899         },
 900 #endif
 901 #if defined(CONFIG_MMU)
 902         {
 903                 .procname       = "randomize_va_space",
 904                 .data           = &randomize_va_space,
 905                 .maxlen         = sizeof(int),
 906                 .mode           = 0644,
 907                 .proc_handler   = proc_dointvec,
 908         },
 909 #endif
 910 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
 911         {
 912                 .procname       = "spin_retry",
 913                 .data           = &spin_retry,
 914                 .maxlen         = sizeof (int),
 915                 .mode           = 0644,
 916                 .proc_handler   = proc_dointvec,
 917         },
 918 #endif
 919 #if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
 920         {
 921                 .procname       = "acpi_video_flags",
 922                 .data           = &acpi_realmode_flags,
 923                 .maxlen         = sizeof (unsigned long),
 924                 .mode           = 0644,
 925                 .proc_handler   = proc_doulongvec_minmax,
 926         },
 927 #endif
 928 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
 929         {
 930                 .procname       = "ignore-unaligned-usertrap",
 931                 .data           = &no_unaligned_warning,
 932                 .maxlen         = sizeof (int),
 933                 .mode           = 0644,
 934                 .proc_handler   = proc_dointvec,
 935         },
 936 #endif
 937 #ifdef CONFIG_IA64
 938         {
 939                 .procname       = "unaligned-dump-stack",
 940                 .data           = &unaligned_dump_stack,
 941                 .maxlen         = sizeof (int),
 942                 .mode           = 0644,
 943                 .proc_handler   = proc_dointvec,
 944         },
 945 #endif
 946 #ifdef CONFIG_DETECT_HUNG_TASK
 947         {
 948                 .procname       = "hung_task_panic",
 949                 .data           = &sysctl_hung_task_panic,
 950                 .maxlen         = sizeof(int),
 951                 .mode           = 0644,
 952                 .proc_handler   = proc_dointvec_minmax,
 953                 .extra1         = &zero,
 954                 .extra2         = &one,
 955         },
 956         {
 957                 .procname       = "hung_task_check_count",
 958                 .data           = &sysctl_hung_task_check_count,
 959                 .maxlen         = sizeof(unsigned long),
 960                 .mode           = 0644,
 961                 .proc_handler   = proc_doulongvec_minmax,
 962         },
 963         {
 964                 .procname       = "hung_task_timeout_secs",
 965                 .data           = &sysctl_hung_task_timeout_secs,
 966                 .maxlen         = sizeof(unsigned long),
 967                 .mode           = 0644,
 968                 .proc_handler   = proc_dohung_task_timeout_secs,
 969         },
 970         {
 971                 .procname       = "hung_task_warnings",
 972                 .data           = &sysctl_hung_task_warnings,
 973                 .maxlen         = sizeof(unsigned long),
 974                 .mode           = 0644,
 975                 .proc_handler   = proc_doulongvec_minmax,
 976         },
 977 #endif
 978 #ifdef CONFIG_COMPAT
 979         {
 980                 .procname       = "compat-log",
 981                 .data           = &compat_log,
 982                 .maxlen         = sizeof (int),
 983                 .mode           = 0644,
 984                 .proc_handler   = proc_dointvec,
 985         },
 986 #endif
 987 #ifdef CONFIG_RT_MUTEXES
 988         {
 989                 .procname       = "max_lock_depth",
 990                 .data           = &max_lock_depth,
 991                 .maxlen         = sizeof(int),
 992                 .mode           = 0644,
 993                 .proc_handler   = proc_dointvec,
 994         },
 995 #endif
 996         {
 997                 .procname       = "poweroff_cmd",
 998                 .data           = &poweroff_cmd,
 999                 .maxlen         = POWEROFF_CMD_PATH_LEN,
1000                 .mode           = 0644,
1001                 .proc_handler   = proc_dostring,
1002         },
1003 #ifdef CONFIG_KEYS
1004         {
1005                 .procname       = "keys",
1006                 .mode           = 0555,
1007                 .child          = key_sysctls,
1008         },
1009 #endif
1010 #ifdef CONFIG_RCU_TORTURE_TEST
1011         {
1012                 .procname       = "rcutorture_runnable",
1013                 .data           = &rcutorture_runnable,
1014                 .maxlen         = sizeof(int),
1015                 .mode           = 0644,
1016                 .proc_handler   = proc_dointvec,
1017         },
1018 #endif
1019 #ifdef CONFIG_PERF_EVENTS
1020         /*
1021          * User-space scripts rely on the existence of this file
1022          * as a feature check for perf_events being enabled.
1023          *
1024          * So it's an ABI, do not remove!
1025          */
1026         {
1027                 .procname       = "perf_event_paranoid",
1028                 .data           = &sysctl_perf_event_paranoid,
1029                 .maxlen         = sizeof(sysctl_perf_event_paranoid),
1030                 .mode           = 0644,
1031                 .proc_handler   = proc_dointvec,
1032         },
1033         {
1034                 .procname       = "perf_event_mlock_kb",
1035                 .data           = &sysctl_perf_event_mlock,
1036                 .maxlen         = sizeof(sysctl_perf_event_mlock),
1037                 .mode           = 0644,
1038                 .proc_handler   = proc_dointvec,
1039         },
1040         {
1041                 .procname       = "perf_event_max_sample_rate",
1042                 .data           = &sysctl_perf_event_sample_rate,
1043                 .maxlen         = sizeof(sysctl_perf_event_sample_rate),
1044                 .mode           = 0644,
1045                 .proc_handler   = perf_proc_update_handler,
1046         },
1047 #endif
1048 #ifdef CONFIG_KMEMCHECK
1049         {
1050                 .procname       = "kmemcheck",
1051                 .data           = &kmemcheck_enabled,
1052                 .maxlen         = sizeof(int),
1053                 .mode           = 0644,
1054                 .proc_handler   = proc_dointvec,
1055         },
1056 #endif
1057 #ifdef CONFIG_BLOCK
1058         {
1059                 .procname       = "blk_iopoll",
1060                 .data           = &blk_iopoll_enabled,
1061                 .maxlen         = sizeof(int),
1062                 .mode           = 0644,
1063                 .proc_handler   = proc_dointvec,
1064         },
1065 #endif
1066         { }
1067 };
1068
1069 static struct ctl_table vm_table[] = {
1070         {
1071                 .procname       = "overcommit_memory",
1072                 .data           = &sysctl_overcommit_memory,
1073                 .maxlen         = sizeof(sysctl_overcommit_memory),
1074                 .mode           = 0644,
1075                 .proc_handler   = proc_dointvec_minmax,
1076                 .extra1         = &zero,
1077                 .extra2         = &two,
1078         },
1079         {
1080                 .procname       = "panic_on_oom",
1081                 .data           = &sysctl_panic_on_oom,
1082                 .maxlen         = sizeof(sysctl_panic_on_oom),
1083                 .mode           = 0644,
1084                 .proc_handler   = proc_dointvec_minmax,
1085                 .extra1         = &zero,
1086                 .extra2         = &two,
1087         },
1088         {
1089                 .procname       = "oom_kill_allocating_task",
1090                 .data           = &sysctl_oom_kill_allocating_task,
1091                 .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
1092                 .mode           = 0644,
1093                 .proc_handler   = proc_dointvec,
1094         },
1095         {
1096                 .procname       = "oom_dump_tasks",
1097                 .data           = &sysctl_oom_dump_tasks,
1098                 .maxlen         = sizeof(sysctl_oom_dump_tasks),
1099                 .mode           = 0644,
1100                 .proc_handler   = proc_dointvec,
1101         },
1102         {
1103                 .procname       = "overcommit_ratio",
1104                 .data           = &sysctl_overcommit_ratio,
1105                 .maxlen         = sizeof(sysctl_overcommit_ratio),
1106                 .mode           = 0644,
1107                 .proc_handler   = proc_dointvec,
1108         },
1109         {
1110                 .procname       = "page-cluster",
1111                 .data           = &page_cluster,
1112                 .maxlen         = sizeof(int),
1113                 .mode           = 0644,
1114                 .proc_handler   = proc_dointvec_minmax,
1115                 .extra1         = &zero,
1116         },
1117         {
1118                 .procname       = "dirty_background_ratio",
1119                 .data           = &dirty_background_ratio,
1120                 .maxlen         = sizeof(dirty_background_ratio),
1121                 .mode           = 0644,
1122                 .proc_handler   = dirty_background_ratio_handler,
1123                 .extra1         = &zero,
1124                 .extra2         = &one_hundred,
1125         },
1126         {
1127                 .procname       = "dirty_background_bytes",
1128                 .data           = &dirty_background_bytes,
1129                 .maxlen         = sizeof(dirty_background_bytes),
1130                 .mode           = 0644,
1131                 .proc_handler   = dirty_background_bytes_handler,
1132                 .extra1         = &one_ul,
1133         },
1134         {
1135                 .procname       = "dirty_ratio",
1136                 .data           = &vm_dirty_ratio,
1137                 .maxlen         = sizeof(vm_dirty_ratio),
1138                 .mode           = 0644,
1139                 .proc_handler   = dirty_ratio_handler,
1140                 .extra1         = &zero,
1141                 .extra2         = &one_hundred,
1142         },
1143         {
1144                 .procname       = "dirty_bytes",
1145                 .data           = &vm_dirty_bytes,
1146                 .maxlen         = sizeof(vm_dirty_bytes),
1147                 .mode           = 0644,
1148                 .proc_handler   = dirty_bytes_handler,
1149                 .extra1         = &dirty_bytes_min,
1150         },
1151         {
1152                 .procname       = "dirty_writeback_centisecs",
1153                 .data           = &dirty_writeback_interval,
1154                 .maxlen         = sizeof(dirty_writeback_interval),
1155                 .mode           = 0644,
1156                 .proc_handler   = dirty_writeback_centisecs_handler,
1157         },
1158         {
1159                 .procname       = "dirty_expire_centisecs",
1160                 .data           = &dirty_expire_interval,
1161                 .maxlen         = sizeof(dirty_expire_interval),
1162                 .mode           = 0644,
1163                 .proc_handler   = proc_dointvec_minmax,
1164                 .extra1         = &zero,
1165         },
1166         {
1167                 .procname       = "nr_pdflush_threads",
1168                 .mode           = 0444 /* read-only */,
1169                 .proc_handler   = pdflush_proc_obsolete,
1170         },
1171         {
1172                 .procname       = "swappiness",
1173                 .data           = &vm_swappiness,
1174                 .maxlen         = sizeof(vm_swappiness),
1175                 .mode           = 0644,
1176                 .proc_handler   = proc_dointvec_minmax,
1177                 .extra1         = &zero,
1178                 .extra2         = &one_hundred,
1179         },
1180 #ifdef CONFIG_HUGETLB_PAGE
1181         {
1182                 .procname       = "nr_hugepages",
1183                 .data           = NULL,
1184                 .maxlen         = sizeof(unsigned long),
1185                 .mode           = 0644,
1186                 .proc_handler   = hugetlb_sysctl_handler,
1187                 .extra1         = (void *)&hugetlb_zero,
1188                 .extra2         = (void *)&hugetlb_infinity,
1189         },
1190 #ifdef CONFIG_NUMA
1191         {
1192                 .procname       = "nr_hugepages_mempolicy",
1193                 .data           = NULL,
1194                 .maxlen         = sizeof(unsigned long),
1195                 .mode           = 0644,
1196                 .proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1197                 .extra1         = (void *)&hugetlb_zero,
1198                 .extra2         = (void *)&hugetlb_infinity,
1199         },
1200 #endif
1201          {
1202                 .procname       = "hugetlb_shm_group",
1203                 .data           = &sysctl_hugetlb_shm_group,
1204                 .maxlen         = sizeof(gid_t),
1205                 .mode           = 0644,
1206                 .proc_handler   = proc_dointvec,
1207          },
1208          {
1209                 .procname       = "hugepages_treat_as_movable",
1210                 .data           = &hugepages_treat_as_movable,
1211                 .maxlen         = sizeof(int),
1212                 .mode           = 0644,
1213                 .proc_handler   = hugetlb_treat_movable_handler,
1214         },
1215         {
1216                 .procname       = "nr_overcommit_hugepages",
1217                 .data           = NULL,
1218                 .maxlen         = sizeof(unsigned long),
1219                 .mode           = 0644,
1220                 .proc_handler   = hugetlb_overcommit_handler,
1221                 .extra1         = (void *)&hugetlb_zero,
1222                 .extra2         = (void *)&hugetlb_infinity,
1223         },
1224 #endif
1225         {
1226                 .procname       = "lowmem_reserve_ratio",
1227                 .data           = &sysctl_lowmem_reserve_ratio,
1228                 .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1229                 .mode           = 0644,
1230                 .proc_handler   = lowmem_reserve_ratio_sysctl_handler,
1231         },
1232         {
1233                 .procname       = "drop_caches",
1234                 .data           = &sysctl_drop_caches,
1235                 .maxlen         = sizeof(int),
1236                 .mode           = 0644,
1237                 .proc_handler   = drop_caches_sysctl_handler,
1238                 .extra1         = &one,
1239                 .extra2         = &three,
1240         },
1241 #ifdef CONFIG_COMPACTION
1242         {
1243                 .procname       = "compact_memory",
1244                 .data           = &sysctl_compact_memory,
1245                 .maxlen         = sizeof(int),
1246                 .mode           = 0200,
1247                 .proc_handler   = sysctl_compaction_handler,
1248         },
1249         {
1250                 .procname       = "extfrag_threshold",
1251                 .data           = &sysctl_extfrag_threshold,
1252                 .maxlen         = sizeof(int),
1253                 .mode           = 0644,
1254                 .proc_handler   = sysctl_extfrag_handler,
1255                 .extra1         = &min_extfrag_threshold,
1256                 .extra2         = &max_extfrag_threshold,
1257         },
1258
1259 #endif /* CONFIG_COMPACTION */
1260         {
1261                 .procname       = "min_free_kbytes",
1262                 .data           = &min_free_kbytes,
1263                 .maxlen         = sizeof(min_free_kbytes),
1264                 .mode           = 0644,
1265                 .proc_handler   = min_free_kbytes_sysctl_handler,
1266                 .extra1         = &zero,
1267         },
1268         {
1269                 .procname       = "percpu_pagelist_fraction",
1270                 .data           = &percpu_pagelist_fraction,
1271                 .maxlen         = sizeof(percpu_pagelist_fraction),
1272                 .mode           = 0644,
1273                 .proc_handler   = percpu_pagelist_fraction_sysctl_handler,
1274                 .extra1         = &min_percpu_pagelist_fract,
1275         },
1276 #ifdef CONFIG_MMU
1277         {
1278                 .procname       = "max_map_count",
1279                 .data           = &sysctl_max_map_count,
1280                 .maxlen         = sizeof(sysctl_max_map_count),
1281                 .mode           = 0644,
1282                 .proc_handler   = proc_dointvec_minmax,
1283                 .extra1         = &zero,
1284         },
1285 #else
1286         {
1287                 .procname       = "nr_trim_pages",
1288                 .data           = &sysctl_nr_trim_pages,
1289                 .maxlen         = sizeof(sysctl_nr_trim_pages),
1290                 .mode           = 0644,
1291                 .proc_handler   = proc_dointvec_minmax,
1292                 .extra1         = &zero,
1293         },
1294 #endif
1295         {
1296                 .procname       = "laptop_mode",
1297                 .data           = &laptop_mode,
1298                 .maxlen         = sizeof(laptop_mode),
1299                 .mode           = 0644,
1300                 .proc_handler   = proc_dointvec_jiffies,
1301         },
1302         {
1303                 .procname       = "block_dump",
1304                 .data           = &block_dump,
1305                 .maxlen         = sizeof(block_dump),
1306                 .mode           = 0644,
1307                 .proc_handler   = proc_dointvec,
1308                 .extra1         = &zero,
1309         },
1310         {
1311                 .procname       = "vfs_cache_pressure",
1312                 .data           = &sysctl_vfs_cache_pressure,
1313                 .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1314                 .mode           = 0644,
1315                 .proc_handler   = proc_dointvec,
1316                 .extra1         = &zero,
1317         },
1318 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1319         {
1320                 .procname       = "legacy_va_layout",
1321                 .data           = &sysctl_legacy_va_layout,
1322                 .maxlen         = sizeof(sysctl_legacy_va_layout),
1323                 .mode           = 0644,
1324                 .proc_handler   = proc_dointvec,
1325                 .extra1         = &zero,
1326         },
1327 #endif
1328 #ifdef CONFIG_NUMA
1329         {
1330                 .procname       = "zone_reclaim_mode",
1331                 .data           = &zone_reclaim_mode,
1332                 .maxlen         = sizeof(zone_reclaim_mode),
1333                 .mode           = 0644,
1334                 .proc_handler   = proc_dointvec,
1335                 .extra1         = &zero,
1336         },
1337         {
1338                 .procname       = "min_unmapped_ratio",
1339                 .data           = &sysctl_min_unmapped_ratio,
1340                 .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1341                 .mode           = 0644,
1342                 .proc_handler   = sysctl_min_unmapped_ratio_sysctl_handler,
1343                 .extra1         = &zero,
1344                 .extra2         = &one_hundred,
1345         },
1346         {
1347                 .procname       = "min_slab_ratio",
1348                 .data           = &sysctl_min_slab_ratio,
1349                 .maxlen         = sizeof(sysctl_min_slab_ratio),
1350                 .mode           = 0644,
1351                 .proc_handler   = sysctl_min_slab_ratio_sysctl_handler,
1352                 .extra1         = &zero,
1353                 .extra2         = &one_hundred,
1354         },
1355 #endif
1356 #ifdef CONFIG_SMP
1357         {
1358                 .procname       = "stat_interval",
1359                 .data           = &sysctl_stat_interval,
1360                 .maxlen         = sizeof(sysctl_stat_interval),
1361                 .mode           = 0644,
1362                 .proc_handler   = proc_dointvec_jiffies,
1363         },
1364 #endif
1365 #ifdef CONFIG_MMU
1366         {
1367                 .procname       = "mmap_min_addr",
1368                 .data           = &dac_mmap_min_addr,
1369                 .maxlen         = sizeof(unsigned long),
1370                 .mode           = 0644,
1371                 .proc_handler   = mmap_min_addr_handler,
1372         },
1373 #endif
1374 #ifdef CONFIG_NUMA
1375         {
1376                 .procname       = "numa_zonelist_order",
1377                 .data           = &numa_zonelist_order,
1378                 .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1379                 .mode           = 0644,
1380                 .proc_handler   = numa_zonelist_order_handler,
1381         },
1382 #endif
1383 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1384    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1385         {
1386                 .procname       = "vdso_enabled",
1387                 .data           = &vdso_enabled,
1388                 .maxlen         = sizeof(vdso_enabled),
1389                 .mode           = 0644,
1390                 .proc_handler   = proc_dointvec,
1391                 .extra1         = &zero,
1392         },
1393 #endif
1394 #ifdef CONFIG_HIGHMEM
1395         {
1396                 .procname       = "highmem_is_dirtyable",
1397                 .data           = &vm_highmem_is_dirtyable,
1398                 .maxlen         = sizeof(vm_highmem_is_dirtyable),
1399                 .mode           = 0644,
1400                 .proc_handler   = proc_dointvec_minmax,
1401                 .extra1         = &zero,
1402                 .extra2         = &one,
1403         },
1404 #endif
1405         {
1406                 .procname       = "scan_unevictable_pages",
1407                 .data           = &scan_unevictable_pages,
1408                 .maxlen         = sizeof(scan_unevictable_pages),
1409                 .mode           = 0644,
1410                 .proc_handler   = scan_unevictable_handler,
1411         },
1412 #ifdef CONFIG_MEMORY_FAILURE
1413         {
1414                 .procname       = "memory_failure_early_kill",
1415                 .data           = &sysctl_memory_failure_early_kill,
1416                 .maxlen         = sizeof(sysctl_memory_failure_early_kill),
1417                 .mode           = 0644,
1418                 .proc_handler   = proc_dointvec_minmax,
1419                 .extra1         = &zero,
1420                 .extra2         = &one,
1421         },
1422         {
1423                 .procname       = "memory_failure_recovery",
1424                 .data           = &sysctl_memory_failure_recovery,
1425                 .maxlen         = sizeof(sysctl_memory_failure_recovery),
1426                 .mode           = 0644,
1427                 .proc_handler   = proc_dointvec_minmax,
1428                 .extra1         = &zero,
1429                 .extra2         = &one,
1430         },
1431 #endif
1432         {
1433                 .procname       = "user_reserve_kbytes",
1434                 .data           = &sysctl_user_reserve_kbytes,
1435                 .maxlen         = sizeof(sysctl_user_reserve_kbytes),
1436                 .mode           = 0644,
1437                 .proc_handler   = proc_doulongvec_minmax,
1438         },
1439         { }
1440 };
1441
1442 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1443 static struct ctl_table binfmt_misc_table[] = {
1444         { }
1445 };
1446 #endif
1447
1448 static struct ctl_table fs_table[] = {
1449         {
1450                 .procname       = "inode-nr",
1451                 .data           = &inodes_stat,
1452                 .maxlen         = 2*sizeof(int),
1453                 .mode           = 0444,
1454                 .proc_handler   = proc_nr_inodes,
1455         },
1456         {
1457                 .procname       = "inode-state",
1458                 .data           = &inodes_stat,
1459                 .maxlen         = 7*sizeof(int),
1460                 .mode           = 0444,
1461                 .proc_handler   = proc_nr_inodes,
1462         },
1463         {
1464                 .procname       = "file-nr",
1465                 .data           = &files_stat,
1466                 .maxlen         = sizeof(files_stat),
1467                 .mode           = 0444,
1468                 .proc_handler   = proc_nr_files,
1469         },
1470         {
1471                 .procname       = "file-max",
1472                 .data           = &files_stat.max_files,
1473                 .maxlen         = sizeof(files_stat.max_files),
1474                 .mode           = 0644,
1475                 .proc_handler   = proc_doulongvec_minmax,
1476         },
1477         {
1478                 .procname       = "nr_open",
1479                 .data           = &sysctl_nr_open,
1480                 .maxlen         = sizeof(int),
1481                 .mode           = 0644,
1482                 .proc_handler   = proc_dointvec_minmax,
1483                 .extra1         = &sysctl_nr_open_min,
1484                 .extra2         = &sysctl_nr_open_max,
1485         },
1486         {
1487                 .procname       = "dentry-state",
1488                 .data           = &dentry_stat,
1489                 .maxlen         = 6*sizeof(int),
1490                 .mode           = 0444,
1491                 .proc_handler   = proc_nr_dentry,
1492         },
1493         {
1494                 .procname       = "overflowuid",
1495                 .data           = &fs_overflowuid,
1496                 .maxlen         = sizeof(int),
1497                 .mode           = 0644,
1498                 .proc_handler   = proc_dointvec_minmax,
1499                 .extra1         = &minolduid,
1500                 .extra2         = &maxolduid,
1501         },
1502         {
1503                 .procname       = "overflowgid",
1504                 .data           = &fs_overflowgid,
1505                 .maxlen         = sizeof(int),
1506                 .mode           = 0644,
1507                 .proc_handler   = proc_dointvec_minmax,
1508                 .extra1         = &minolduid,
1509                 .extra2         = &maxolduid,
1510         },
1511 #ifdef CONFIG_FILE_LOCKING
1512         {
1513                 .procname       = "leases-enable",
1514                 .data           = &leases_enable,
1515                 .maxlen         = sizeof(int),
1516                 .mode           = 0644,
1517                 .proc_handler   = proc_dointvec,
1518         },
1519 #endif
1520 #ifdef CONFIG_DNOTIFY
1521         {
1522                 .procname       = "dir-notify-enable",
1523                 .data           = &dir_notify_enable,
1524                 .maxlen         = sizeof(int),
1525                 .mode           = 0644,
1526                 .proc_handler   = proc_dointvec,
1527         },
1528 #endif
1529 #ifdef CONFIG_MMU
1530 #ifdef CONFIG_FILE_LOCKING
1531         {
1532                 .procname       = "lease-break-time",
1533                 .data           = &lease_break_time,
1534                 .maxlen         = sizeof(int),
1535                 .mode           = 0644,
1536                 .proc_handler   = proc_dointvec,
1537         },
1538 #endif
1539 #ifdef CONFIG_AIO
1540         {
1541                 .procname       = "aio-nr",
1542                 .data           = &aio_nr,
1543                 .maxlen         = sizeof(aio_nr),
1544                 .mode           = 0444,
1545                 .proc_handler   = proc_doulongvec_minmax,
1546         },
1547         {
1548                 .procname       = "aio-max-nr",
1549                 .data           = &aio_max_nr,
1550                 .maxlen         = sizeof(aio_max_nr),
1551                 .mode           = 0644,
1552                 .proc_handler   = proc_doulongvec_minmax,
1553         },
1554 #endif /* CONFIG_AIO */
1555 #ifdef CONFIG_INOTIFY_USER
1556         {
1557                 .procname       = "inotify",
1558                 .mode           = 0555,
1559                 .child          = inotify_table,
1560         },
1561 #endif
1562 #ifdef CONFIG_EPOLL
1563         {
1564                 .procname       = "epoll",
1565                 .mode           = 0555,
1566                 .child          = epoll_table,
1567         },
1568 #endif
1569 #endif
1570         {
1571                 .procname       = "protected_symlinks",
1572                 .data           = &sysctl_protected_symlinks,
1573                 .maxlen         = sizeof(int),
1574                 .mode           = 0600,
1575                 .proc_handler   = proc_dointvec_minmax,
1576                 .extra1         = &zero,
1577                 .extra2         = &one,
1578         },
1579         {
1580                 .procname       = "protected_hardlinks",
1581                 .data           = &sysctl_protected_hardlinks,
1582                 .maxlen         = sizeof(int),
1583                 .mode           = 0600,
1584                 .proc_handler   = proc_dointvec_minmax,
1585                 .extra1         = &zero,
1586                 .extra2         = &one,
1587         },
1588         {
1589                 .procname       = "suid_dumpable",
1590                 .data           = &suid_dumpable,
1591                 .maxlen         = sizeof(int),
1592                 .mode           = 0644,
1593                 .proc_handler   = proc_dointvec_minmax_coredump,
1594                 .extra1         = &zero,
1595                 .extra2         = &two,
1596         },
1597 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1598         {
1599                 .procname       = "binfmt_misc",
1600                 .mode           = 0555,
1601                 .child          = binfmt_misc_table,
1602         },
1603 #endif
1604         {
1605                 .procname       = "pipe-max-size",
1606                 .data           = &pipe_max_size,
1607                 .maxlen         = sizeof(int),
1608                 .mode           = 0644,
1609                 .proc_handler   = &pipe_proc_fn,
1610                 .extra1         = &pipe_min_size,
1611         },
1612         { }
1613 };
1614
1615 static struct ctl_table debug_table[] = {
1616 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1617         {
1618                 .procname       = "exception-trace",
1619                 .data           = &show_unhandled_signals,
1620                 .maxlen         = sizeof(int),
1621                 .mode           = 0644,
1622                 .proc_handler   = proc_dointvec
1623         },
1624 #endif
1625 #if defined(CONFIG_OPTPROBES)
1626         {
1627                 .procname       = "kprobes-optimization",
1628                 .data           = &sysctl_kprobes_optimization,
1629                 .maxlen         = sizeof(int),
1630                 .mode           = 0644,
1631                 .proc_handler   = proc_kprobes_optimization_handler,
1632                 .extra1         = &zero,
1633                 .extra2         = &one,
1634         },
1635 #endif
1636         { }
1637 };
1638
1639 static struct ctl_table dev_table[] = {
1640         { }
1641 };
1642
1643 int __init sysctl_init(void)
1644 {
1645         struct ctl_table_header *hdr;
1646
1647         hdr = register_sysctl_table(sysctl_base_table);
1648         kmemleak_not_leak(hdr);
1649         return 0;
1650 }
1651
1652 #endif /* CONFIG_SYSCTL */
1653
1654 /*
1655  * /proc/sys support
1656  */
1657
1658 #ifdef CONFIG_PROC_SYSCTL
1659
1660 static int _proc_do_string(void* data, int maxlen, int write,
1661                            void __user *buffer,
1662                            size_t *lenp, loff_t *ppos)
1663 {
1664         size_t len;
1665         char __user *p;
1666         char c;
1667
1668         if (!data || !maxlen || !*lenp) {
1669                 *lenp = 0;
1670                 return 0;
1671         }
1672
1673         if (write) {
1674                 len = 0;
1675                 p = buffer;
1676                 while (len < *lenp) {
1677                         if (get_user(c, p++))
1678                                 return -EFAULT;
1679                         if (c == 0 || c == '\n')
1680                                 break;
1681                         len++;
1682                 }
1683                 if (len >= maxlen)
1684                         len = maxlen-1;
1685                 if(copy_from_user(data, buffer, len))
1686                         return -EFAULT;
1687                 ((char *) data)[len] = 0;
1688                 *ppos += *lenp;
1689         } else {
1690                 len = strlen(data);
1691                 if (len > maxlen)
1692                         len = maxlen;
1693
1694                 if (*ppos > len) {
1695                         *lenp = 0;
1696                         return 0;
1697                 }
1698
1699                 data += *ppos;
1700                 len  -= *ppos;
1701
1702                 if (len > *lenp)
1703                         len = *lenp;
1704                 if (len)
1705                         if(copy_to_user(buffer, data, len))
1706                                 return -EFAULT;
1707                 if (len < *lenp) {
1708                         if(put_user('\n', ((char __user *) buffer) + len))
1709                                 return -EFAULT;
1710                         len++;
1711                 }
1712                 *lenp = len;
1713                 *ppos += len;
1714         }
1715         return 0;
1716 }
1717
1718 /**
1719  * proc_dostring - read a string sysctl
1720  * @table: the sysctl table
1721  * @write: %TRUE if this is a write to the sysctl file
1722  * @buffer: the user buffer
1723  * @lenp: the size of the user buffer
1724  * @ppos: file position
1725  *
1726  * Reads/writes a string from/to the user buffer. If the kernel
1727  * buffer provided is not large enough to hold the string, the
1728  * string is truncated. The copied string is %NULL-terminated.
1729  * If the string is being read by the user process, it is copied
1730  * and a newline '\n' is added. It is truncated if the buffer is
1731  * not large enough.
1732  *
1733  * Returns 0 on success.
1734  */
1735 int proc_dostring(struct ctl_table *table, int write,
1736                   void __user *buffer, size_t *lenp, loff_t *ppos)
1737 {
1738         return _proc_do_string(table->data, table->maxlen, write,
1739                                buffer, lenp, ppos);
1740 }
1741
1742 static size_t proc_skip_spaces(char **buf)
1743 {
1744         size_t ret;
1745         char *tmp = skip_spaces(*buf);
1746         ret = tmp - *buf;
1747         *buf = tmp;
1748         return ret;
1749 }
1750
1751 static void proc_skip_char(char **buf, size_t *size, const char v)
1752 {
1753         while (*size) {
1754                 if (**buf != v)
1755                         break;
1756                 (*size)--;
1757                 (*buf)++;
1758         }
1759 }
1760
1761 #define TMPBUFLEN 22
1762 /**
1763  * proc_get_long - reads an ASCII formatted integer from a user buffer
1764  *
1765  * @buf: a kernel buffer
1766  * @size: size of the kernel buffer
1767  * @val: this is where the number will be stored
1768  * @neg: set to %TRUE if number is negative
1769  * @perm_tr: a vector which contains the allowed trailers
1770  * @perm_tr_len: size of the perm_tr vector
1771  * @tr: pointer to store the trailer character
1772  *
1773  * In case of success %0 is returned and @buf and @size are updated with
1774  * the amount of bytes read. If @tr is non-NULL and a trailing
1775  * character exists (size is non-zero after returning from this
1776  * function), @tr is updated with the trailing character.
1777  */
1778 static int proc_get_long(char **buf, size_t *size,
1779                           unsigned long *val, bool *neg,
1780                           const char *perm_tr, unsigned perm_tr_len, char *tr)
1781 {
1782         int len;
1783         char *p, tmp[TMPBUFLEN];
1784
1785         if (!*size)
1786                 return -EINVAL;
1787
1788         len = *size;
1789         if (len > TMPBUFLEN - 1)
1790                 len = TMPBUFLEN - 1;
1791
1792         memcpy(tmp, *buf, len);
1793
1794         tmp[len] = 0;
1795         p = tmp;
1796         if (*p == '-' && *size > 1) {
1797                 *neg = true;
1798                 p++;
1799         } else
1800                 *neg = false;
1801         if (!isdigit(*p))
1802                 return -EINVAL;
1803
1804         *val = simple_strtoul(p, &p, 0);
1805
1806         len = p - tmp;
1807
1808         /* We don't know if the next char is whitespace thus we may accept
1809          * invalid integers (e.g. 1234...a) or two integers instead of one
1810          * (e.g. 123...1). So lets not allow such large numbers. */
1811         if (len == TMPBUFLEN - 1)
1812                 return -EINVAL;
1813
1814         if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1815                 return -EINVAL;
1816
1817         if (tr && (len < *size))
1818                 *tr = *p;
1819
1820         *buf += len;
1821         *size -= len;
1822
1823         return 0;
1824 }
1825
1826 /**
1827  * proc_put_long - converts an integer to a decimal ASCII formatted string
1828  *
1829  * @buf: the user buffer
1830  * @size: the size of the user buffer
1831  * @val: the integer to be converted
1832  * @neg: sign of the number, %TRUE for negative
1833  *
1834  * In case of success %0 is returned and @buf and @size are updated with
1835  * the amount of bytes written.
1836  */
1837 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1838                           bool neg)
1839 {
1840         int len;
1841         char tmp[TMPBUFLEN], *p = tmp;
1842
1843         sprintf(p, "%s%lu", neg ? "-" : "", val);
1844         len = strlen(tmp);
1845         if (len > *size)
1846                 len = *size;
1847         if (copy_to_user(*buf, tmp, len))
1848                 return -EFAULT;
1849         *size -= len;
1850         *buf += len;
1851         return 0;
1852 }
1853 #undef TMPBUFLEN
1854
1855 static int proc_put_char(void __user **buf, size_t *size, char c)
1856 {
1857         if (*size) {
1858                 char __user **buffer = (char __user **)buf;
1859                 if (put_user(c, *buffer))
1860                         return -EFAULT;
1861                 (*size)--, (*buffer)++;
1862                 *buf = *buffer;
1863         }
1864         return 0;
1865 }
1866
1867 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1868                                  int *valp,
1869                                  int write, void *data)
1870 {
1871         if (write) {
1872                 *valp = *negp ? -*lvalp : *lvalp;
1873         } else {
1874                 int val = *valp;
1875                 if (val < 0) {
1876                         *negp = true;
1877                         *lvalp = (unsigned long)-val;
1878                 } else {
1879                         *negp = false;
1880                         *lvalp = (unsigned long)val;
1881                 }
1882         }
1883         return 0;
1884 }
1885
1886 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1887
1888 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1889                   int write, void __user *buffer,
1890                   size_t *lenp, loff_t *ppos,
1891                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1892                               int write, void *data),
1893                   void *data)
1894 {
1895         int *i, vleft, first = 1, err = 0;
1896         unsigned long page = 0;
1897         size_t left;
1898         char *kbuf;
1899
1900         if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1901                 *lenp = 0;
1902                 return 0;
1903         }
1904
1905         i = (int *) tbl_data;
1906         vleft = table->maxlen / sizeof(*i);
1907         left = *lenp;
1908
1909         if (!conv)
1910                 conv = do_proc_dointvec_conv;
1911
1912         if (write) {
1913                 if (left > PAGE_SIZE - 1)
1914                         left = PAGE_SIZE - 1;
1915                 page = __get_free_page(GFP_TEMPORARY);
1916                 kbuf = (char *) page;
1917                 if (!kbuf)
1918                         return -ENOMEM;
1919                 if (copy_from_user(kbuf, buffer, left)) {
1920                         err = -EFAULT;
1921                         goto free;
1922                 }
1923                 kbuf[left] = 0;
1924         }
1925
1926         for (; left && vleft--; i++, first=0) {
1927                 unsigned long lval;
1928                 bool neg;
1929
1930                 if (write) {
1931                         left -= proc_skip_spaces(&kbuf);
1932
1933                         if (!left)
1934                                 break;
1935                         err = proc_get_long(&kbuf, &left, &lval, &neg,
1936                                              proc_wspace_sep,
1937                                              sizeof(proc_wspace_sep), NULL);
1938                         if (err)
1939                                 break;
1940                         if (conv(&neg, &lval, i, 1, data)) {
1941                                 err = -EINVAL;
1942                                 break;
1943                         }
1944                 } else {
1945                         if (conv(&neg, &lval, i, 0, data)) {
1946                                 err = -EINVAL;
1947                                 break;
1948                         }
1949                         if (!first)
1950                                 err = proc_put_char(&buffer, &left, '\t');
1951                         if (err)
1952                                 break;
1953                         err = proc_put_long(&buffer, &left, lval, neg);
1954                         if (err)
1955                                 break;
1956                 }
1957         }
1958
1959         if (!write && !first && left && !err)
1960                 err = proc_put_char(&buffer, &left, '\n');
1961         if (write && !err && left)
1962                 left -= proc_skip_spaces(&kbuf);
1963 free:
1964         if (write) {
1965                 free_page(page);
1966                 if (first)
1967                         return err ? : -EINVAL;
1968         }
1969         *lenp -= left;
1970         *ppos += *lenp;
1971         return err;
1972 }
1973
1974 static int do_proc_dointvec(struct ctl_table *table, int write,
1975                   void __user *buffer, size_t *lenp, loff_t *ppos,
1976                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1977                               int write, void *data),
1978                   void *data)
1979 {
1980         return __do_proc_dointvec(table->data, table, write,
1981                         buffer, lenp, ppos, conv, data);
1982 }
1983
1984 /**
1985  * proc_dointvec - read a vector of integers
1986  * @table: the sysctl table
1987  * @write: %TRUE if this is a write to the sysctl file
1988  * @buffer: the user buffer
1989  * @lenp: the size of the user buffer
1990  * @ppos: file position
1991  *
1992  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1993  * values from/to the user buffer, treated as an ASCII string.
1994  *
1995  * Returns 0 on success.
1996  */
1997 int proc_dointvec(struct ctl_table *table, int write,
1998                      void __user *buffer, size_t *lenp, loff_t *ppos)
1999 {
2000     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2001                             NULL,NULL);
2002 }
2003
2004 /*
2005  * Taint values can only be increased
2006  * This means we can safely use a temporary.
2007  */
2008 static int proc_taint(struct ctl_table *table, int write,
2009                                void __user *buffer, size_t *lenp, loff_t *ppos)
2010 {
2011         struct ctl_table t;
2012         unsigned long tmptaint = get_taint();
2013         int err;
2014
2015         if (write && !capable(CAP_SYS_ADMIN))
2016                 return -EPERM;
2017
2018         t = *table;
2019         t.data = &tmptaint;
2020         err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2021         if (err < 0)
2022                 return err;
2023
2024         if (write) {
2025                 /*
2026                  * Poor man's atomic or. Not worth adding a primitive
2027                  * to everyone's atomic.h for this
2028                  */
2029                 int i;
2030                 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2031                         if ((tmptaint >> i) & 1)
2032                                 add_taint(i, LOCKDEP_STILL_OK);
2033                 }
2034         }
2035
2036         return err;
2037 }
2038
2039 #ifdef CONFIG_PRINTK
2040 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2041                                 void __user *buffer, size_t *lenp, loff_t *ppos)
2042 {
2043         if (write && !capable(CAP_SYS_ADMIN))
2044                 return -EPERM;
2045
2046         return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2047 }
2048 #endif
2049
2050 struct do_proc_dointvec_minmax_conv_param {
2051         int *min;
2052         int *max;
2053 };
2054
2055 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2056                                         int *valp,
2057                                         int write, void *data)
2058 {
2059         struct do_proc_dointvec_minmax_conv_param *param = data;
2060         if (write) {
2061                 int val = *negp ? -*lvalp : *lvalp;
2062                 if ((param->min && *param->min > val) ||
2063                     (param->max && *param->max < val))
2064                         return -EINVAL;
2065                 *valp = val;
2066         } else {
2067                 int val = *valp;
2068                 if (val < 0) {
2069                         *negp = true;
2070                         *lvalp = (unsigned long)-val;
2071                 } else {
2072                         *negp = false;
2073                         *lvalp = (unsigned long)val;
2074                 }
2075         }
2076         return 0;
2077 }
2078
2079 /**
2080  * proc_dointvec_minmax - read a vector of integers with min/max values
2081  * @table: the sysctl table
2082  * @write: %TRUE if this is a write to the sysctl file
2083  * @buffer: the user buffer
2084  * @lenp: the size of the user buffer
2085  * @ppos: file position
2086  *
2087  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2088  * values from/to the user buffer, treated as an ASCII string.
2089  *
2090  * This routine will ensure the values are within the range specified by
2091  * table->extra1 (min) and table->extra2 (max).
2092  *
2093  * Returns 0 on success.
2094  */
2095 int proc_dointvec_minmax(struct ctl_table *table, int write,
2096                   void __user *buffer, size_t *lenp, loff_t *ppos)
2097 {
2098         struct do_proc_dointvec_minmax_conv_param param = {
2099                 .min = (int *) table->extra1,
2100                 .max = (int *) table->extra2,
2101         };
2102         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2103                                 do_proc_dointvec_minmax_conv, &param);
2104 }
2105
2106 static void validate_coredump_safety(void)
2107 {
2108 #ifdef CONFIG_COREDUMP
2109         if (suid_dumpable == SUID_DUMP_ROOT &&
2110             core_pattern[0] != '/' && core_pattern[0] != '|') {
2111                 printk(KERN_WARNING "Unsafe core_pattern used with "\
2112                         "suid_dumpable=2. Pipe handler or fully qualified "\
2113                         "core dump path required.\n");
2114         }
2115 #endif
2116 }
2117
2118 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2119                 void __user *buffer, size_t *lenp, loff_t *ppos)
2120 {
2121         int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2122         if (!error)
2123                 validate_coredump_safety();
2124         return error;
2125 }
2126
2127 #ifdef CONFIG_COREDUMP
2128 static int proc_dostring_coredump(struct ctl_table *table, int write,
2129                   void __user *buffer, size_t *lenp, loff_t *ppos)
2130 {
2131         int error = proc_dostring(table, write, buffer, lenp, ppos);
2132         if (!error)
2133                 validate_coredump_safety();
2134         return error;
2135 }
2136 #endif
2137
2138 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2139                                      void __user *buffer,
2140                                      size_t *lenp, loff_t *ppos,
2141                                      unsigned long convmul,
2142                                      unsigned long convdiv)
2143 {
2144         unsigned long *i, *min, *max;
2145         int vleft, first = 1, err = 0;
2146         unsigned long page = 0;
2147         size_t left;
2148         char *kbuf;
2149
2150         if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2151                 *lenp = 0;
2152                 return 0;
2153         }
2154
2155         i = (unsigned long *) data;
2156         min = (unsigned long *) table->extra1;
2157         max = (unsigned long *) table->extra2;
2158         vleft = table->maxlen / sizeof(unsigned long);
2159         left = *lenp;
2160
2161         if (write) {
2162                 if (left > PAGE_SIZE - 1)
2163                         left = PAGE_SIZE - 1;
2164                 page = __get_free_page(GFP_TEMPORARY);
2165                 kbuf = (char *) page;
2166                 if (!kbuf)
2167                         return -ENOMEM;
2168                 if (copy_from_user(kbuf, buffer, left)) {
2169                         err = -EFAULT;
2170                         goto free;
2171                 }
2172                 kbuf[left] = 0;
2173         }
2174
2175         for (; left && vleft--; i++, first = 0) {
2176                 unsigned long val;
2177
2178                 if (write) {
2179                         bool neg;
2180
2181                         left -= proc_skip_spaces(&kbuf);
2182
2183                         err = proc_get_long(&kbuf, &left, &val, &neg,
2184                                              proc_wspace_sep,
2185                                              sizeof(proc_wspace_sep), NULL);
2186                         if (err)
2187                                 break;
2188                         if (neg)
2189                                 continue;
2190                         if ((min && val < *min) || (max && val > *max))
2191                                 continue;
2192                         *i = val;
2193                 } else {
2194                         val = convdiv * (*i) / convmul;
2195                         if (!first)
2196                                 err = proc_put_char(&buffer, &left, '\t');
2197                         err = proc_put_long(&buffer, &left, val, false);
2198                         if (err)
2199                                 break;
2200                 }
2201         }
2202
2203         if (!write && !first && left && !err)
2204                 err = proc_put_char(&buffer, &left, '\n');
2205         if (write && !err)
2206                 left -= proc_skip_spaces(&kbuf);
2207 free:
2208         if (write) {
2209                 free_page(page);
2210                 if (first)
2211                         return err ? : -EINVAL;
2212         }
2213         *lenp -= left;
2214         *ppos += *lenp;
2215         return err;
2216 }
2217
2218 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2219                                      void __user *buffer,
2220                                      size_t *lenp, loff_t *ppos,
2221                                      unsigned long convmul,
2222                                      unsigned long convdiv)
2223 {
2224         return __do_proc_doulongvec_minmax(table->data, table, write,
2225                         buffer, lenp, ppos, convmul, convdiv);
2226 }
2227
2228 /**
2229  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2230  * @table: the sysctl table
2231  * @write: %TRUE if this is a write to the sysctl file
2232  * @buffer: the user buffer
2233  * @lenp: the size of the user buffer
2234  * @ppos: file position
2235  *
2236  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2237  * values from/to the user buffer, treated as an ASCII string.
2238  *
2239  * This routine will ensure the values are within the range specified by
2240  * table->extra1 (min) and table->extra2 (max).
2241  *
2242  * Returns 0 on success.
2243  */
2244 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2245                            void __user *buffer, size_t *lenp, loff_t *ppos)
2246 {
2247     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2248 }
2249
2250 /**
2251  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2252  * @table: the sysctl table
2253  * @write: %TRUE if this is a write to the sysctl file
2254  * @buffer: the user buffer
2255  * @lenp: the size of the user buffer
2256  * @ppos: file position
2257  *
2258  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2259  * values from/to the user buffer, treated as an ASCII string. The values
2260  * are treated as milliseconds, and converted to jiffies when they are stored.
2261  *
2262  * This routine will ensure the values are within the range specified by
2263  * table->extra1 (min) and table->extra2 (max).
2264  *
2265  * Returns 0 on success.
2266  */
2267 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2268                                       void __user *buffer,
2269                                       size_t *lenp, loff_t *ppos)
2270 {
2271     return do_proc_doulongvec_minmax(table, write, buffer,
2272                                      lenp, ppos, HZ, 1000l);
2273 }
2274
2275
2276 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2277                                          int *valp,
2278                                          int write, void *data)
2279 {
2280         if (write) {
2281                 if (*lvalp > LONG_MAX / HZ)
2282                         return 1;
2283                 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2284         } else {
2285                 int val = *valp;
2286                 unsigned long lval;
2287                 if (val < 0) {
2288                         *negp = true;
2289                         lval = (unsigned long)-val;
2290                 } else {
2291                         *negp = false;
2292                         lval = (unsigned long)val;
2293                 }
2294                 *lvalp = lval / HZ;
2295         }
2296         return 0;
2297 }
2298
2299 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2300                                                 int *valp,
2301                                                 int write, void *data)
2302 {
2303         if (write) {
2304                 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2305                         return 1;
2306                 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2307         } else {
2308                 int val = *valp;
2309                 unsigned long lval;
2310                 if (val < 0) {
2311                         *negp = true;
2312                         lval = (unsigned long)-val;
2313                 } else {
2314                         *negp = false;
2315                         lval = (unsigned long)val;
2316                 }
2317                 *lvalp = jiffies_to_clock_t(lval);
2318         }
2319         return 0;
2320 }
2321
2322 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2323                                             int *valp,
2324                                             int write, void *data)
2325 {
2326         if (write) {
2327                 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2328         } else {
2329                 int val = *valp;
2330                 unsigned long lval;
2331                 if (val < 0) {
2332                         *negp = true;
2333                         lval = (unsigned long)-val;
2334                 } else {
2335                         *negp = false;
2336                         lval = (unsigned long)val;
2337                 }
2338                 *lvalp = jiffies_to_msecs(lval);
2339         }
2340         return 0;
2341 }
2342
2343 /**
2344  * proc_dointvec_jiffies - read a vector of integers as seconds
2345  * @table: the sysctl table
2346  * @write: %TRUE if this is a write to the sysctl file
2347  * @buffer: the user buffer
2348  * @lenp: the size of the user buffer
2349  * @ppos: file position
2350  *
2351  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2352  * values from/to the user buffer, treated as an ASCII string.
2353  * The values read are assumed to be in seconds, and are converted into
2354  * jiffies.
2355  *
2356  * Returns 0 on success.
2357  */
2358 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2359                           void __user *buffer, size_t *lenp, loff_t *ppos)
2360 {
2361     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2362                             do_proc_dointvec_jiffies_conv,NULL);
2363 }
2364
2365 /**
2366  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2367  * @table: the sysctl table
2368  * @write: %TRUE if this is a write to the sysctl file
2369  * @buffer: the user buffer
2370  * @lenp: the size of the user buffer
2371  * @ppos: pointer to the file position
2372  *
2373  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2374  * values from/to the user buffer, treated as an ASCII string.
2375  * The values read are assumed to be in 1/USER_HZ seconds, and
2376  * are converted into jiffies.
2377  *
2378  * Returns 0 on success.
2379  */
2380 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2381                                  void __user *buffer, size_t *lenp, loff_t *ppos)
2382 {
2383     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2384                             do_proc_dointvec_userhz_jiffies_conv,NULL);
2385 }
2386
2387 /**
2388  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2389  * @table: the sysctl table
2390  * @write: %TRUE if this is a write to the sysctl file
2391  * @buffer: the user buffer
2392  * @lenp: the size of the user buffer
2393  * @ppos: file position
2394  * @ppos: the current position in the file
2395  *
2396  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2397  * values from/to the user buffer, treated as an ASCII string.
2398  * The values read are assumed to be in 1/1000 seconds, and
2399  * are converted into jiffies.
2400  *
2401  * Returns 0 on success.
2402  */
2403 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2404                              void __user *buffer, size_t *lenp, loff_t *ppos)
2405 {
2406         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2407                                 do_proc_dointvec_ms_jiffies_conv, NULL);
2408 }
2409
2410 static int proc_do_cad_pid(struct ctl_table *table, int write,
2411                            void __user *buffer, size_t *lenp, loff_t *ppos)
2412 {
2413         struct pid *new_pid;
2414         pid_t tmp;
2415         int r;
2416
2417         tmp = pid_vnr(cad_pid);
2418
2419         r = __do_proc_dointvec(&tmp, table, write, buffer,
2420                                lenp, ppos, NULL, NULL);
2421         if (r || !write)
2422                 return r;
2423
2424         new_pid = find_get_pid(tmp);
2425         if (!new_pid)
2426                 return -ESRCH;
2427
2428         put_pid(xchg(&cad_pid, new_pid));
2429         return 0;
2430 }
2431
2432 /**
2433  * proc_do_large_bitmap - read/write from/to a large bitmap
2434  * @table: the sysctl table
2435  * @write: %TRUE if this is a write to the sysctl file
2436  * @buffer: the user buffer
2437  * @lenp: the size of the user buffer
2438  * @ppos: file position
2439  *
2440  * The bitmap is stored at table->data and the bitmap length (in bits)
2441  * in table->maxlen.
2442  *
2443  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2444  * large bitmaps may be represented in a compact manner. Writing into
2445  * the file will clear the bitmap then update it with the given input.
2446  *
2447  * Returns 0 on success.
2448  */
2449 int proc_do_large_bitmap(struct ctl_table *table, int write,
2450                          void __user *buffer, size_t *lenp, loff_t *ppos)
2451 {
2452         int err = 0;
2453         bool first = 1;
2454         size_t left = *lenp;
2455         unsigned long bitmap_len = table->maxlen;
2456         unsigned long *bitmap = (unsigned long *) table->data;
2457         unsigned long *tmp_bitmap = NULL;
2458         char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2459
2460         if (!bitmap_len || !left || (*ppos && !write)) {
2461                 *lenp = 0;
2462                 return 0;
2463         }
2464
2465         if (write) {
2466                 unsigned long page = 0;
2467                 char *kbuf;
2468
2469                 if (left > PAGE_SIZE - 1)
2470                         left = PAGE_SIZE - 1;
2471
2472                 page = __get_free_page(GFP_TEMPORARY);
2473                 kbuf = (char *) page;
2474                 if (!kbuf)
2475                         return -ENOMEM;
2476                 if (copy_from_user(kbuf, buffer, left)) {
2477                         free_page(page);
2478                         return -EFAULT;
2479                 }
2480                 kbuf[left] = 0;
2481
2482                 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2483                                      GFP_KERNEL);
2484                 if (!tmp_bitmap) {
2485                         free_page(page);
2486                         return -ENOMEM;
2487                 }
2488                 proc_skip_char(&kbuf, &left, '\n');
2489                 while (!err && left) {
2490                         unsigned long val_a, val_b;
2491                         bool neg;
2492
2493                         err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2494                                              sizeof(tr_a), &c);
2495                         if (err)
2496                                 break;
2497                         if (val_a >= bitmap_len || neg) {
2498                                 err = -EINVAL;
2499                                 break;
2500                         }
2501
2502                         val_b = val_a;
2503                         if (left) {
2504                                 kbuf++;
2505                                 left--;
2506                         }
2507
2508                         if (c == '-') {
2509                                 err = proc_get_long(&kbuf, &left, &val_b,
2510                                                      &neg, tr_b, sizeof(tr_b),
2511                                                      &c);
2512                                 if (err)
2513                                         break;
2514                                 if (val_b >= bitmap_len || neg ||
2515                                     val_a > val_b) {
2516                                         err = -EINVAL;
2517                                         break;
2518                                 }
2519                                 if (left) {
2520                                         kbuf++;
2521                                         left--;
2522                                 }
2523                         }
2524
2525                         bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2526                         first = 0;
2527                         proc_skip_char(&kbuf, &left, '\n');
2528                 }
2529                 free_page(page);
2530         } else {
2531                 unsigned long bit_a, bit_b = 0;
2532
2533                 while (left) {
2534                         bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2535                         if (bit_a >= bitmap_len)
2536                                 break;
2537                         bit_b = find_next_zero_bit(bitmap, bitmap_len,
2538                                                    bit_a + 1) - 1;
2539
2540                         if (!first) {
2541                                 err = proc_put_char(&buffer, &left, ',');
2542                                 if (err)
2543                                         break;
2544                         }
2545                         err = proc_put_long(&buffer, &left, bit_a, false);
2546                         if (err)
2547                                 break;
2548                         if (bit_a != bit_b) {
2549                                 err = proc_put_char(&buffer, &left, '-');
2550                                 if (err)
2551                                         break;
2552                                 err = proc_put_long(&buffer, &left, bit_b, false);
2553                                 if (err)
2554                                         break;
2555                         }
2556
2557                         first = 0; bit_b++;
2558                 }
2559                 if (!err)
2560                         err = proc_put_char(&buffer, &left, '\n');
2561         }
2562
2563         if (!err) {
2564                 if (write) {
2565                         if (*ppos)
2566                                 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2567                         else
2568                                 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2569                 }
2570                 kfree(tmp_bitmap);
2571                 *lenp -= left;
2572                 *ppos += *lenp;
2573                 return 0;
2574         } else {
2575                 kfree(tmp_bitmap);
2576                 return err;
2577         }
2578 }
2579
2580 #else /* CONFIG_PROC_SYSCTL */
2581
2582 int proc_dostring(struct ctl_table *table, int write,
2583                   void __user *buffer, size_t *lenp, loff_t *ppos)
2584 {
2585         return -ENOSYS;
2586 }
2587
2588 int proc_dointvec(struct ctl_table *table, int write,
2589                   void __user *buffer, size_t *lenp, loff_t *ppos)
2590 {
2591         return -ENOSYS;
2592 }
2593
2594 int proc_dointvec_minmax(struct ctl_table *table, int write,
2595                     void __user *buffer, size_t *lenp, loff_t *ppos)
2596 {
2597         return -ENOSYS;
2598 }
2599
2600 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2601                     void __user *buffer, size_t *lenp, loff_t *ppos)
2602 {
2603         return -ENOSYS;
2604 }
2605
2606 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2607                     void __user *buffer, size_t *lenp, loff_t *ppos)
2608 {
2609         return -ENOSYS;
2610 }
2611
2612 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2613                              void __user *buffer, size_t *lenp, loff_t *ppos)
2614 {
2615         return -ENOSYS;
2616 }
2617
2618 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2619                     void __user *buffer, size_t *lenp, loff_t *ppos)
2620 {
2621         return -ENOSYS;
2622 }
2623
2624 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2625                                       void __user *buffer,
2626                                       size_t *lenp, loff_t *ppos)
2627 {
2628     return -ENOSYS;
2629 }
2630
2631
2632 #endif /* CONFIG_PROC_SYSCTL */
2633
2634 /*
2635  * No sense putting this after each symbol definition, twice,
2636  * exception granted :-)
2637  */
2638 EXPORT_SYMBOL(proc_dointvec);
2639 EXPORT_SYMBOL(proc_dointvec_jiffies);
2640 EXPORT_SYMBOL(proc_dointvec_minmax);
2641 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2642 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2643 EXPORT_SYMBOL(proc_dostring);
2644 EXPORT_SYMBOL(proc_doulongvec_minmax);
2645 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);