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