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