Linux Audio

Check our new training course

Loading...
v4.10.11
   1/*
   2 * sysctl.c: General linux system control interface
   3 *
   4 * Begun 24 March 1995, Stephen Tweedie
   5 * Added /proc support, Dec 1995
   6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
   7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
   8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
   9 * Dynamic registration fixes, Stephen Tweedie.
  10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
  11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
  12 *  Horn.
  13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
  14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
  15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
  16 *  Wendling.
  17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
  18 *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
  19 */
  20
  21#include <linux/module.h>
  22#include <linux/aio.h>
  23#include <linux/mm.h>
  24#include <linux/swap.h>
  25#include <linux/slab.h>
  26#include <linux/sysctl.h>
  27#include <linux/bitmap.h>
  28#include <linux/signal.h>
  29#include <linux/printk.h>
  30#include <linux/proc_fs.h>
  31#include <linux/security.h>
  32#include <linux/ctype.h>
  33#include <linux/kmemcheck.h>
  34#include <linux/kmemleak.h>
  35#include <linux/fs.h>
  36#include <linux/init.h>
  37#include <linux/kernel.h>
  38#include <linux/kobject.h>
  39#include <linux/net.h>
  40#include <linux/sysrq.h>
  41#include <linux/highuid.h>
  42#include <linux/writeback.h>
  43#include <linux/ratelimit.h>
  44#include <linux/compaction.h>
  45#include <linux/hugetlb.h>
  46#include <linux/initrd.h>
  47#include <linux/key.h>
  48#include <linux/times.h>
  49#include <linux/limits.h>
  50#include <linux/dcache.h>
  51#include <linux/dnotify.h>
  52#include <linux/syscalls.h>
  53#include <linux/vmstat.h>
  54#include <linux/nfs_fs.h>
  55#include <linux/acpi.h>
  56#include <linux/reboot.h>
  57#include <linux/ftrace.h>
  58#include <linux/perf_event.h>
  59#include <linux/kprobes.h>
  60#include <linux/pipe_fs_i.h>
  61#include <linux/oom.h>
  62#include <linux/kmod.h>
  63#include <linux/capability.h>
  64#include <linux/binfmts.h>
  65#include <linux/sched/sysctl.h>
  66#include <linux/kexec.h>
  67#include <linux/bpf.h>
  68#include <linux/mount.h>
  69
  70#include <linux/uaccess.h>
  71#include <asm/processor.h>
  72
  73#ifdef CONFIG_X86
  74#include <asm/nmi.h>
  75#include <asm/stacktrace.h>
  76#include <asm/io.h>
  77#endif
  78#ifdef CONFIG_SPARC
  79#include <asm/setup.h>
  80#endif
  81#ifdef CONFIG_BSD_PROCESS_ACCT
  82#include <linux/acct.h>
  83#endif
  84#ifdef CONFIG_RT_MUTEXES
  85#include <linux/rtmutex.h>
  86#endif
  87#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
  88#include <linux/lockdep.h>
  89#endif
  90#ifdef CONFIG_CHR_DEV_SG
  91#include <scsi/sg.h>
  92#endif
  93
  94#ifdef CONFIG_LOCKUP_DETECTOR
  95#include <linux/nmi.h>
  96#endif
  97
 
  98#if defined(CONFIG_SYSCTL)
  99
 100/* External variables not in a header file. */
 101extern int suid_dumpable;
 102#ifdef CONFIG_COREDUMP
 
 103extern int core_uses_pid;
 
 104extern char core_pattern[];
 105extern unsigned int core_pipe_limit;
 106#endif
 107extern int pid_max;
 
 108extern int pid_max_min, pid_max_max;
 
 109extern int percpu_pagelist_fraction;
 
 110extern int latencytop_enabled;
 111extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
 112#ifndef CONFIG_MMU
 113extern int sysctl_nr_trim_pages;
 114#endif
 
 
 
 115
 116/* Constants used for minimum and  maximum */
 117#ifdef CONFIG_LOCKUP_DETECTOR
 118static int sixty = 60;
 
 119#endif
 120
 121static int __maybe_unused neg_one = -1;
 122
 123static int zero;
 124static int __maybe_unused one = 1;
 125static int __maybe_unused two = 2;
 126static int __maybe_unused four = 4;
 127static unsigned long one_ul = 1;
 128static int one_hundred = 100;
 129static int one_thousand = 1000;
 130#ifdef CONFIG_PRINTK
 131static int ten_thousand = 10000;
 132#endif
 133#ifdef CONFIG_PERF_EVENTS
 134static int six_hundred_forty_kb = 640 * 1024;
 135#endif
 136
 137/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 138static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 139
 140/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 141static int maxolduid = 65535;
 142static int minolduid;
 
 143
 144static int ngroups_max = NGROUPS_MAX;
 145static const int cap_last_cap = CAP_LAST_CAP;
 146
 147/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
 148#ifdef CONFIG_DETECT_HUNG_TASK
 149static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
 150#endif
 151
 152#ifdef CONFIG_INOTIFY_USER
 153#include <linux/inotify.h>
 154#endif
 155#ifdef CONFIG_SPARC
 
 156#endif
 157
 158#ifdef __hppa__
 159extern int pwrsw_enabled;
 160#endif
 161
 162#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 
 163extern int unaligned_enabled;
 164#endif
 165
 166#ifdef CONFIG_IA64
 167extern int unaligned_dump_stack;
 
 
 
 
 168#endif
 169
 170#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
 171extern int no_unaligned_warning;
 
 172#endif
 173
 174#ifdef CONFIG_PROC_SYSCTL
 175
 176#define SYSCTL_WRITES_LEGACY	-1
 177#define SYSCTL_WRITES_WARN	 0
 178#define SYSCTL_WRITES_STRICT	 1
 179
 180static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
 181
 182static int proc_do_cad_pid(struct ctl_table *table, int write,
 183		  void __user *buffer, size_t *lenp, loff_t *ppos);
 184static int proc_taint(struct ctl_table *table, int write,
 185			       void __user *buffer, size_t *lenp, loff_t *ppos);
 186#endif
 187
 188#ifdef CONFIG_PRINTK
 189static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
 190				void __user *buffer, size_t *lenp, loff_t *ppos);
 191#endif
 192
 193static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
 194		void __user *buffer, size_t *lenp, loff_t *ppos);
 195#ifdef CONFIG_COREDUMP
 196static int proc_dostring_coredump(struct ctl_table *table, int write,
 197		void __user *buffer, size_t *lenp, loff_t *ppos);
 198#endif
 199
 200#ifdef CONFIG_MAGIC_SYSRQ
 201/* Note: sysrq code uses it's own private copy */
 202static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
 203
 204static int sysrq_sysctl_handler(struct ctl_table *table, int write,
 205				void __user *buffer, size_t *lenp,
 206				loff_t *ppos)
 207{
 208	int error;
 209
 210	error = proc_dointvec(table, write, buffer, lenp, ppos);
 211	if (error)
 212		return error;
 213
 214	if (write)
 215		sysrq_toggle_support(__sysrq_enabled);
 216
 217	return 0;
 218}
 219
 220#endif
 221
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 222static struct ctl_table kern_table[];
 223static struct ctl_table vm_table[];
 224static struct ctl_table fs_table[];
 225static struct ctl_table debug_table[];
 226static struct ctl_table dev_table[];
 227extern struct ctl_table random_table[];
 228#ifdef CONFIG_EPOLL
 229extern struct ctl_table epoll_table[];
 230#endif
 231
 232#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 233int sysctl_legacy_va_layout;
 234#endif
 235
 236/* The default sysctl tables: */
 237
 238static struct ctl_table sysctl_base_table[] = {
 239	{
 240		.procname	= "kernel",
 241		.mode		= 0555,
 242		.child		= kern_table,
 243	},
 244	{
 245		.procname	= "vm",
 246		.mode		= 0555,
 247		.child		= vm_table,
 248	},
 249	{
 250		.procname	= "fs",
 251		.mode		= 0555,
 252		.child		= fs_table,
 253	},
 254	{
 255		.procname	= "debug",
 256		.mode		= 0555,
 257		.child		= debug_table,
 258	},
 259	{
 260		.procname	= "dev",
 261		.mode		= 0555,
 262		.child		= dev_table,
 263	},
 264	{ }
 265};
 266
 267#ifdef CONFIG_SCHED_DEBUG
 268static int min_sched_granularity_ns = 100000;		/* 100 usecs */
 269static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
 270static int min_wakeup_granularity_ns;			/* 0 usecs */
 271static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
 272#ifdef CONFIG_SMP
 273static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 274static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
 275#endif /* CONFIG_SMP */
 276#endif /* CONFIG_SCHED_DEBUG */
 277
 278#ifdef CONFIG_COMPACTION
 279static int min_extfrag_threshold;
 280static int max_extfrag_threshold = 1000;
 281#endif
 282
 283static struct ctl_table kern_table[] = {
 284	{
 285		.procname	= "sched_child_runs_first",
 286		.data		= &sysctl_sched_child_runs_first,
 287		.maxlen		= sizeof(unsigned int),
 288		.mode		= 0644,
 289		.proc_handler	= proc_dointvec,
 290	},
 291#ifdef CONFIG_SCHED_DEBUG
 292	{
 293		.procname	= "sched_min_granularity_ns",
 294		.data		= &sysctl_sched_min_granularity,
 295		.maxlen		= sizeof(unsigned int),
 296		.mode		= 0644,
 297		.proc_handler	= sched_proc_update_handler,
 298		.extra1		= &min_sched_granularity_ns,
 299		.extra2		= &max_sched_granularity_ns,
 300	},
 301	{
 302		.procname	= "sched_latency_ns",
 303		.data		= &sysctl_sched_latency,
 304		.maxlen		= sizeof(unsigned int),
 305		.mode		= 0644,
 306		.proc_handler	= sched_proc_update_handler,
 307		.extra1		= &min_sched_granularity_ns,
 308		.extra2		= &max_sched_granularity_ns,
 309	},
 310	{
 311		.procname	= "sched_wakeup_granularity_ns",
 312		.data		= &sysctl_sched_wakeup_granularity,
 313		.maxlen		= sizeof(unsigned int),
 314		.mode		= 0644,
 315		.proc_handler	= sched_proc_update_handler,
 316		.extra1		= &min_wakeup_granularity_ns,
 317		.extra2		= &max_wakeup_granularity_ns,
 318	},
 319#ifdef CONFIG_SMP
 320	{
 321		.procname	= "sched_tunable_scaling",
 322		.data		= &sysctl_sched_tunable_scaling,
 323		.maxlen		= sizeof(enum sched_tunable_scaling),
 324		.mode		= 0644,
 325		.proc_handler	= sched_proc_update_handler,
 326		.extra1		= &min_sched_tunable_scaling,
 327		.extra2		= &max_sched_tunable_scaling,
 328	},
 329	{
 330		.procname	= "sched_migration_cost_ns",
 331		.data		= &sysctl_sched_migration_cost,
 332		.maxlen		= sizeof(unsigned int),
 333		.mode		= 0644,
 334		.proc_handler	= proc_dointvec,
 335	},
 336	{
 337		.procname	= "sched_nr_migrate",
 338		.data		= &sysctl_sched_nr_migrate,
 339		.maxlen		= sizeof(unsigned int),
 340		.mode		= 0644,
 341		.proc_handler	= proc_dointvec,
 342	},
 343	{
 344		.procname	= "sched_time_avg_ms",
 345		.data		= &sysctl_sched_time_avg,
 346		.maxlen		= sizeof(unsigned int),
 347		.mode		= 0644,
 348		.proc_handler	= proc_dointvec,
 349	},
 350#ifdef CONFIG_SCHEDSTATS
 351	{
 352		.procname	= "sched_schedstats",
 353		.data		= NULL,
 354		.maxlen		= sizeof(unsigned int),
 355		.mode		= 0644,
 356		.proc_handler	= sysctl_schedstats,
 357		.extra1		= &zero,
 358		.extra2		= &one,
 359	},
 360#endif /* CONFIG_SCHEDSTATS */
 361#endif /* CONFIG_SMP */
 362#ifdef CONFIG_NUMA_BALANCING
 363	{
 364		.procname	= "numa_balancing_scan_delay_ms",
 365		.data		= &sysctl_numa_balancing_scan_delay,
 366		.maxlen		= sizeof(unsigned int),
 367		.mode		= 0644,
 368		.proc_handler	= proc_dointvec,
 369	},
 370	{
 371		.procname	= "numa_balancing_scan_period_min_ms",
 372		.data		= &sysctl_numa_balancing_scan_period_min,
 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_minmax,
 390		.extra1		= &one,
 391	},
 392	{
 393		.procname	= "numa_balancing",
 394		.data		= NULL, /* filled in by handler */
 395		.maxlen		= sizeof(unsigned int),
 396		.mode		= 0644,
 397		.proc_handler	= sysctl_numa_balancing,
 398		.extra1		= &zero,
 399		.extra2		= &one,
 400	},
 401#endif /* CONFIG_NUMA_BALANCING */
 402#endif /* CONFIG_SCHED_DEBUG */
 403	{
 404		.procname	= "sched_rt_period_us",
 405		.data		= &sysctl_sched_rt_period,
 406		.maxlen		= sizeof(unsigned int),
 407		.mode		= 0644,
 408		.proc_handler	= sched_rt_handler,
 409	},
 410	{
 411		.procname	= "sched_rt_runtime_us",
 412		.data		= &sysctl_sched_rt_runtime,
 413		.maxlen		= sizeof(int),
 414		.mode		= 0644,
 415		.proc_handler	= sched_rt_handler,
 416	},
 417	{
 418		.procname	= "sched_rr_timeslice_ms",
 419		.data		= &sched_rr_timeslice,
 420		.maxlen		= sizeof(int),
 421		.mode		= 0644,
 422		.proc_handler	= sched_rr_handler,
 423	},
 424#ifdef CONFIG_SCHED_AUTOGROUP
 425	{
 426		.procname	= "sched_autogroup_enabled",
 427		.data		= &sysctl_sched_autogroup_enabled,
 428		.maxlen		= sizeof(unsigned int),
 429		.mode		= 0644,
 430		.proc_handler	= proc_dointvec_minmax,
 431		.extra1		= &zero,
 432		.extra2		= &one,
 433	},
 434#endif
 435#ifdef CONFIG_CFS_BANDWIDTH
 436	{
 437		.procname	= "sched_cfs_bandwidth_slice_us",
 438		.data		= &sysctl_sched_cfs_bandwidth_slice,
 439		.maxlen		= sizeof(unsigned int),
 440		.mode		= 0644,
 441		.proc_handler	= proc_dointvec_minmax,
 442		.extra1		= &one,
 443	},
 444#endif
 445#ifdef CONFIG_PROVE_LOCKING
 446	{
 447		.procname	= "prove_locking",
 448		.data		= &prove_locking,
 449		.maxlen		= sizeof(int),
 450		.mode		= 0644,
 451		.proc_handler	= proc_dointvec,
 452	},
 453#endif
 454#ifdef CONFIG_LOCK_STAT
 455	{
 456		.procname	= "lock_stat",
 457		.data		= &lock_stat,
 458		.maxlen		= sizeof(int),
 459		.mode		= 0644,
 460		.proc_handler	= proc_dointvec,
 461	},
 462#endif
 463	{
 464		.procname	= "panic",
 465		.data		= &panic_timeout,
 466		.maxlen		= sizeof(int),
 467		.mode		= 0644,
 468		.proc_handler	= proc_dointvec,
 469	},
 470#ifdef CONFIG_COREDUMP
 471	{
 472		.procname	= "core_uses_pid",
 473		.data		= &core_uses_pid,
 474		.maxlen		= sizeof(int),
 475		.mode		= 0644,
 476		.proc_handler	= proc_dointvec,
 477	},
 478	{
 479		.procname	= "core_pattern",
 480		.data		= core_pattern,
 481		.maxlen		= CORENAME_MAX_SIZE,
 482		.mode		= 0644,
 483		.proc_handler	= proc_dostring_coredump,
 484	},
 485	{
 486		.procname	= "core_pipe_limit",
 487		.data		= &core_pipe_limit,
 488		.maxlen		= sizeof(unsigned int),
 489		.mode		= 0644,
 490		.proc_handler	= proc_dointvec,
 491	},
 492#endif
 493#ifdef CONFIG_PROC_SYSCTL
 494	{
 495		.procname	= "tainted",
 496		.maxlen 	= sizeof(long),
 497		.mode		= 0644,
 498		.proc_handler	= proc_taint,
 499	},
 500	{
 501		.procname	= "sysctl_writes_strict",
 502		.data		= &sysctl_writes_strict,
 503		.maxlen		= sizeof(int),
 504		.mode		= 0644,
 505		.proc_handler	= proc_dointvec_minmax,
 506		.extra1		= &neg_one,
 507		.extra2		= &one,
 508	},
 509#endif
 510#ifdef CONFIG_LATENCYTOP
 511	{
 512		.procname	= "latencytop",
 513		.data		= &latencytop_enabled,
 514		.maxlen		= sizeof(int),
 515		.mode		= 0644,
 516		.proc_handler	= sysctl_latencytop,
 517	},
 518#endif
 519#ifdef CONFIG_BLK_DEV_INITRD
 520	{
 521		.procname	= "real-root-dev",
 522		.data		= &real_root_dev,
 523		.maxlen		= sizeof(int),
 524		.mode		= 0644,
 525		.proc_handler	= proc_dointvec,
 526	},
 527#endif
 528	{
 529		.procname	= "print-fatal-signals",
 530		.data		= &print_fatal_signals,
 531		.maxlen		= sizeof(int),
 532		.mode		= 0644,
 533		.proc_handler	= proc_dointvec,
 534	},
 535#ifdef CONFIG_SPARC
 536	{
 537		.procname	= "reboot-cmd",
 538		.data		= reboot_command,
 539		.maxlen		= 256,
 540		.mode		= 0644,
 541		.proc_handler	= proc_dostring,
 542	},
 543	{
 544		.procname	= "stop-a",
 545		.data		= &stop_a_enabled,
 546		.maxlen		= sizeof (int),
 547		.mode		= 0644,
 548		.proc_handler	= proc_dointvec,
 549	},
 550	{
 551		.procname	= "scons-poweroff",
 552		.data		= &scons_pwroff,
 553		.maxlen		= sizeof (int),
 554		.mode		= 0644,
 555		.proc_handler	= proc_dointvec,
 556	},
 557#endif
 558#ifdef CONFIG_SPARC64
 559	{
 560		.procname	= "tsb-ratio",
 561		.data		= &sysctl_tsb_ratio,
 562		.maxlen		= sizeof (int),
 563		.mode		= 0644,
 564		.proc_handler	= proc_dointvec,
 565	},
 566#endif
 567#ifdef __hppa__
 568	{
 569		.procname	= "soft-power",
 570		.data		= &pwrsw_enabled,
 571		.maxlen		= sizeof (int),
 572	 	.mode		= 0644,
 573		.proc_handler	= proc_dointvec,
 574	},
 575#endif
 576#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
 577	{
 578		.procname	= "unaligned-trap",
 579		.data		= &unaligned_enabled,
 580		.maxlen		= sizeof (int),
 581		.mode		= 0644,
 582		.proc_handler	= proc_dointvec,
 583	},
 584#endif
 585	{
 586		.procname	= "ctrl-alt-del",
 587		.data		= &C_A_D,
 588		.maxlen		= sizeof(int),
 589		.mode		= 0644,
 590		.proc_handler	= proc_dointvec,
 591	},
 592#ifdef CONFIG_FUNCTION_TRACER
 593	{
 594		.procname	= "ftrace_enabled",
 595		.data		= &ftrace_enabled,
 596		.maxlen		= sizeof(int),
 597		.mode		= 0644,
 598		.proc_handler	= ftrace_enable_sysctl,
 599	},
 600#endif
 601#ifdef CONFIG_STACK_TRACER
 602	{
 603		.procname	= "stack_tracer_enabled",
 604		.data		= &stack_tracer_enabled,
 605		.maxlen		= sizeof(int),
 606		.mode		= 0644,
 607		.proc_handler	= stack_trace_sysctl,
 608	},
 609#endif
 610#ifdef CONFIG_TRACING
 611	{
 612		.procname	= "ftrace_dump_on_oops",
 613		.data		= &ftrace_dump_on_oops,
 614		.maxlen		= sizeof(int),
 615		.mode		= 0644,
 616		.proc_handler	= proc_dointvec,
 617	},
 618	{
 619		.procname	= "traceoff_on_warning",
 620		.data		= &__disable_trace_on_warning,
 621		.maxlen		= sizeof(__disable_trace_on_warning),
 622		.mode		= 0644,
 623		.proc_handler	= proc_dointvec,
 624	},
 625	{
 626		.procname	= "tracepoint_printk",
 627		.data		= &tracepoint_printk,
 628		.maxlen		= sizeof(tracepoint_printk),
 629		.mode		= 0644,
 630		.proc_handler	= tracepoint_printk_sysctl,
 631	},
 632#endif
 633#ifdef CONFIG_KEXEC_CORE
 634	{
 635		.procname	= "kexec_load_disabled",
 636		.data		= &kexec_load_disabled,
 637		.maxlen		= sizeof(int),
 638		.mode		= 0644,
 639		/* only handle a transition from default "0" to "1" */
 640		.proc_handler	= proc_dointvec_minmax,
 641		.extra1		= &one,
 642		.extra2		= &one,
 643	},
 644#endif
 645#ifdef CONFIG_MODULES
 646	{
 647		.procname	= "modprobe",
 648		.data		= &modprobe_path,
 649		.maxlen		= KMOD_PATH_LEN,
 650		.mode		= 0644,
 651		.proc_handler	= proc_dostring,
 652	},
 653	{
 654		.procname	= "modules_disabled",
 655		.data		= &modules_disabled,
 656		.maxlen		= sizeof(int),
 657		.mode		= 0644,
 658		/* only handle a transition from default "0" to "1" */
 659		.proc_handler	= proc_dointvec_minmax,
 660		.extra1		= &one,
 661		.extra2		= &one,
 662	},
 663#endif
 664#ifdef CONFIG_UEVENT_HELPER
 665	{
 666		.procname	= "hotplug",
 667		.data		= &uevent_helper,
 668		.maxlen		= UEVENT_HELPER_PATH_LEN,
 669		.mode		= 0644,
 670		.proc_handler	= proc_dostring,
 671	},
 672#endif
 673#ifdef CONFIG_CHR_DEV_SG
 674	{
 675		.procname	= "sg-big-buff",
 676		.data		= &sg_big_buff,
 677		.maxlen		= sizeof (int),
 678		.mode		= 0444,
 679		.proc_handler	= proc_dointvec,
 680	},
 681#endif
 682#ifdef CONFIG_BSD_PROCESS_ACCT
 683	{
 684		.procname	= "acct",
 685		.data		= &acct_parm,
 686		.maxlen		= 3*sizeof(int),
 687		.mode		= 0644,
 688		.proc_handler	= proc_dointvec,
 689	},
 690#endif
 691#ifdef CONFIG_MAGIC_SYSRQ
 692	{
 693		.procname	= "sysrq",
 694		.data		= &__sysrq_enabled,
 695		.maxlen		= sizeof (int),
 696		.mode		= 0644,
 697		.proc_handler	= sysrq_sysctl_handler,
 698	},
 699#endif
 700#ifdef CONFIG_PROC_SYSCTL
 701	{
 702		.procname	= "cad_pid",
 703		.data		= NULL,
 704		.maxlen		= sizeof (int),
 705		.mode		= 0600,
 706		.proc_handler	= proc_do_cad_pid,
 707	},
 708#endif
 709	{
 710		.procname	= "threads-max",
 711		.data		= NULL,
 712		.maxlen		= sizeof(int),
 713		.mode		= 0644,
 714		.proc_handler	= sysctl_max_threads,
 715	},
 716	{
 717		.procname	= "random",
 718		.mode		= 0555,
 719		.child		= random_table,
 720	},
 721	{
 722		.procname	= "usermodehelper",
 723		.mode		= 0555,
 724		.child		= usermodehelper_table,
 725	},
 726	{
 727		.procname	= "overflowuid",
 728		.data		= &overflowuid,
 729		.maxlen		= sizeof(int),
 730		.mode		= 0644,
 731		.proc_handler	= proc_dointvec_minmax,
 732		.extra1		= &minolduid,
 733		.extra2		= &maxolduid,
 734	},
 735	{
 736		.procname	= "overflowgid",
 737		.data		= &overflowgid,
 738		.maxlen		= sizeof(int),
 739		.mode		= 0644,
 740		.proc_handler	= proc_dointvec_minmax,
 741		.extra1		= &minolduid,
 742		.extra2		= &maxolduid,
 743	},
 744#ifdef CONFIG_S390
 745#ifdef CONFIG_MATHEMU
 746	{
 747		.procname	= "ieee_emulation_warnings",
 748		.data		= &sysctl_ieee_emulation_warnings,
 749		.maxlen		= sizeof(int),
 750		.mode		= 0644,
 751		.proc_handler	= proc_dointvec,
 752	},
 753#endif
 754	{
 755		.procname	= "userprocess_debug",
 756		.data		= &show_unhandled_signals,
 757		.maxlen		= sizeof(int),
 758		.mode		= 0644,
 759		.proc_handler	= proc_dointvec,
 760	},
 761#endif
 762	{
 763		.procname	= "pid_max",
 764		.data		= &pid_max,
 765		.maxlen		= sizeof (int),
 766		.mode		= 0644,
 767		.proc_handler	= proc_dointvec_minmax,
 768		.extra1		= &pid_max_min,
 769		.extra2		= &pid_max_max,
 770	},
 771	{
 772		.procname	= "panic_on_oops",
 773		.data		= &panic_on_oops,
 774		.maxlen		= sizeof(int),
 775		.mode		= 0644,
 776		.proc_handler	= proc_dointvec,
 777	},
 778#if defined CONFIG_PRINTK
 779	{
 780		.procname	= "printk",
 781		.data		= &console_loglevel,
 782		.maxlen		= 4*sizeof(int),
 783		.mode		= 0644,
 784		.proc_handler	= proc_dointvec,
 785	},
 786	{
 787		.procname	= "printk_ratelimit",
 788		.data		= &printk_ratelimit_state.interval,
 789		.maxlen		= sizeof(int),
 790		.mode		= 0644,
 791		.proc_handler	= proc_dointvec_jiffies,
 792	},
 793	{
 794		.procname	= "printk_ratelimit_burst",
 795		.data		= &printk_ratelimit_state.burst,
 796		.maxlen		= sizeof(int),
 797		.mode		= 0644,
 798		.proc_handler	= proc_dointvec,
 799	},
 800	{
 801		.procname	= "printk_delay",
 802		.data		= &printk_delay_msec,
 803		.maxlen		= sizeof(int),
 804		.mode		= 0644,
 805		.proc_handler	= proc_dointvec_minmax,
 806		.extra1		= &zero,
 807		.extra2		= &ten_thousand,
 808	},
 809	{
 810		.procname	= "printk_devkmsg",
 811		.data		= devkmsg_log_str,
 812		.maxlen		= DEVKMSG_STR_MAX_SIZE,
 813		.mode		= 0644,
 814		.proc_handler	= devkmsg_sysctl_set_loglvl,
 815	},
 816	{
 817		.procname	= "dmesg_restrict",
 818		.data		= &dmesg_restrict,
 819		.maxlen		= sizeof(int),
 820		.mode		= 0644,
 821		.proc_handler	= proc_dointvec_minmax_sysadmin,
 822		.extra1		= &zero,
 823		.extra2		= &one,
 824	},
 825	{
 826		.procname	= "kptr_restrict",
 827		.data		= &kptr_restrict,
 828		.maxlen		= sizeof(int),
 829		.mode		= 0644,
 830		.proc_handler	= proc_dointvec_minmax_sysadmin,
 831		.extra1		= &zero,
 832		.extra2		= &two,
 833	},
 834#endif
 835	{
 836		.procname	= "ngroups_max",
 837		.data		= &ngroups_max,
 838		.maxlen		= sizeof (int),
 839		.mode		= 0444,
 840		.proc_handler	= proc_dointvec,
 841	},
 842	{
 843		.procname	= "cap_last_cap",
 844		.data		= (void *)&cap_last_cap,
 845		.maxlen		= sizeof(int),
 846		.mode		= 0444,
 847		.proc_handler	= proc_dointvec,
 848	},
 849#if defined(CONFIG_LOCKUP_DETECTOR)
 850	{
 851		.procname       = "watchdog",
 852		.data           = &watchdog_user_enabled,
 853		.maxlen         = sizeof (int),
 854		.mode           = 0644,
 855		.proc_handler   = proc_watchdog,
 856		.extra1		= &zero,
 857		.extra2		= &one,
 858	},
 859	{
 860		.procname	= "watchdog_thresh",
 861		.data		= &watchdog_thresh,
 862		.maxlen		= sizeof(int),
 863		.mode		= 0644,
 864		.proc_handler	= proc_watchdog_thresh,
 865		.extra1		= &zero,
 866		.extra2		= &sixty,
 867	},
 868	{
 869		.procname       = "nmi_watchdog",
 870		.data           = &nmi_watchdog_enabled,
 871		.maxlen         = sizeof (int),
 872		.mode           = 0644,
 873		.proc_handler   = proc_nmi_watchdog,
 874		.extra1		= &zero,
 875#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
 876		.extra2		= &one,
 877#else
 878		.extra2		= &zero,
 879#endif
 880	},
 881	{
 882		.procname       = "soft_watchdog",
 883		.data           = &soft_watchdog_enabled,
 884		.maxlen         = sizeof (int),
 885		.mode           = 0644,
 886		.proc_handler   = proc_soft_watchdog,
 887		.extra1		= &zero,
 888		.extra2		= &one,
 889	},
 890	{
 891		.procname	= "watchdog_cpumask",
 892		.data		= &watchdog_cpumask_bits,
 893		.maxlen		= NR_CPUS,
 894		.mode		= 0644,
 895		.proc_handler	= proc_watchdog_cpumask,
 896	},
 897	{
 898		.procname	= "softlockup_panic",
 899		.data		= &softlockup_panic,
 900		.maxlen		= sizeof(int),
 901		.mode		= 0644,
 902		.proc_handler	= proc_dointvec_minmax,
 903		.extra1		= &zero,
 904		.extra2		= &one,
 905	},
 906#ifdef CONFIG_HARDLOCKUP_DETECTOR
 907	{
 908		.procname	= "hardlockup_panic",
 909		.data		= &hardlockup_panic,
 910		.maxlen		= sizeof(int),
 911		.mode		= 0644,
 912		.proc_handler	= proc_dointvec_minmax,
 913		.extra1		= &zero,
 914		.extra2		= &one,
 915	},
 916#endif
 917#ifdef CONFIG_SMP
 918	{
 919		.procname	= "softlockup_all_cpu_backtrace",
 920		.data		= &sysctl_softlockup_all_cpu_backtrace,
 921		.maxlen		= sizeof(int),
 922		.mode		= 0644,
 923		.proc_handler	= proc_dointvec_minmax,
 924		.extra1		= &zero,
 925		.extra2		= &one,
 926	},
 927	{
 928		.procname	= "hardlockup_all_cpu_backtrace",
 929		.data		= &sysctl_hardlockup_all_cpu_backtrace,
 930		.maxlen		= sizeof(int),
 931		.mode		= 0644,
 932		.proc_handler	= proc_dointvec_minmax,
 933		.extra1		= &zero,
 934		.extra2		= &one,
 935	},
 936#endif /* CONFIG_SMP */
 937#endif
 938#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 939	{
 940		.procname       = "unknown_nmi_panic",
 941		.data           = &unknown_nmi_panic,
 942		.maxlen         = sizeof (int),
 943		.mode           = 0644,
 944		.proc_handler   = proc_dointvec,
 945	},
 946#endif
 947#if defined(CONFIG_X86)
 948	{
 949		.procname	= "panic_on_unrecovered_nmi",
 950		.data		= &panic_on_unrecovered_nmi,
 951		.maxlen		= sizeof(int),
 952		.mode		= 0644,
 953		.proc_handler	= proc_dointvec,
 954	},
 955	{
 956		.procname	= "panic_on_io_nmi",
 957		.data		= &panic_on_io_nmi,
 958		.maxlen		= sizeof(int),
 959		.mode		= 0644,
 960		.proc_handler	= proc_dointvec,
 961	},
 962#ifdef CONFIG_DEBUG_STACKOVERFLOW
 963	{
 964		.procname	= "panic_on_stackoverflow",
 965		.data		= &sysctl_panic_on_stackoverflow,
 966		.maxlen		= sizeof(int),
 967		.mode		= 0644,
 968		.proc_handler	= proc_dointvec,
 969	},
 970#endif
 971	{
 972		.procname	= "bootloader_type",
 973		.data		= &bootloader_type,
 974		.maxlen		= sizeof (int),
 975		.mode		= 0444,
 976		.proc_handler	= proc_dointvec,
 977	},
 978	{
 979		.procname	= "bootloader_version",
 980		.data		= &bootloader_version,
 981		.maxlen		= sizeof (int),
 982		.mode		= 0444,
 983		.proc_handler	= proc_dointvec,
 984	},
 985	{
 
 
 
 
 
 
 
 986		.procname	= "io_delay_type",
 987		.data		= &io_delay_type,
 988		.maxlen		= sizeof(int),
 989		.mode		= 0644,
 990		.proc_handler	= proc_dointvec,
 991	},
 992#endif
 993#if defined(CONFIG_MMU)
 994	{
 995		.procname	= "randomize_va_space",
 996		.data		= &randomize_va_space,
 997		.maxlen		= sizeof(int),
 998		.mode		= 0644,
 999		.proc_handler	= proc_dointvec,
1000	},
1001#endif
1002#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1003	{
1004		.procname	= "spin_retry",
1005		.data		= &spin_retry,
1006		.maxlen		= sizeof (int),
1007		.mode		= 0644,
1008		.proc_handler	= proc_dointvec,
1009	},
1010#endif
1011#if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1012	{
1013		.procname	= "acpi_video_flags",
1014		.data		= &acpi_realmode_flags,
1015		.maxlen		= sizeof (unsigned long),
1016		.mode		= 0644,
1017		.proc_handler	= proc_doulongvec_minmax,
1018	},
1019#endif
1020#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1021	{
1022		.procname	= "ignore-unaligned-usertrap",
1023		.data		= &no_unaligned_warning,
1024		.maxlen		= sizeof (int),
1025	 	.mode		= 0644,
1026		.proc_handler	= proc_dointvec,
1027	},
1028#endif
1029#ifdef CONFIG_IA64
1030	{
1031		.procname	= "unaligned-dump-stack",
1032		.data		= &unaligned_dump_stack,
1033		.maxlen		= sizeof (int),
1034		.mode		= 0644,
1035		.proc_handler	= proc_dointvec,
1036	},
1037#endif
1038#ifdef CONFIG_DETECT_HUNG_TASK
1039	{
1040		.procname	= "hung_task_panic",
1041		.data		= &sysctl_hung_task_panic,
1042		.maxlen		= sizeof(int),
1043		.mode		= 0644,
1044		.proc_handler	= proc_dointvec_minmax,
1045		.extra1		= &zero,
1046		.extra2		= &one,
1047	},
1048	{
1049		.procname	= "hung_task_check_count",
1050		.data		= &sysctl_hung_task_check_count,
1051		.maxlen		= sizeof(int),
1052		.mode		= 0644,
1053		.proc_handler	= proc_dointvec_minmax,
1054		.extra1		= &zero,
1055	},
1056	{
1057		.procname	= "hung_task_timeout_secs",
1058		.data		= &sysctl_hung_task_timeout_secs,
1059		.maxlen		= sizeof(unsigned long),
1060		.mode		= 0644,
1061		.proc_handler	= proc_dohung_task_timeout_secs,
1062		.extra2		= &hung_task_timeout_max,
1063	},
1064	{
1065		.procname	= "hung_task_warnings",
1066		.data		= &sysctl_hung_task_warnings,
1067		.maxlen		= sizeof(int),
1068		.mode		= 0644,
1069		.proc_handler	= proc_dointvec_minmax,
1070		.extra1		= &neg_one,
 
 
 
 
 
 
 
 
1071	},
1072#endif
1073#ifdef CONFIG_RT_MUTEXES
1074	{
1075		.procname	= "max_lock_depth",
1076		.data		= &max_lock_depth,
1077		.maxlen		= sizeof(int),
1078		.mode		= 0644,
1079		.proc_handler	= proc_dointvec,
1080	},
1081#endif
1082	{
1083		.procname	= "poweroff_cmd",
1084		.data		= &poweroff_cmd,
1085		.maxlen		= POWEROFF_CMD_PATH_LEN,
1086		.mode		= 0644,
1087		.proc_handler	= proc_dostring,
1088	},
1089#ifdef CONFIG_KEYS
1090	{
1091		.procname	= "keys",
1092		.mode		= 0555,
1093		.child		= key_sysctls,
1094	},
1095#endif
 
 
 
 
 
 
 
 
 
1096#ifdef CONFIG_PERF_EVENTS
1097	/*
1098	 * User-space scripts rely on the existence of this file
1099	 * as a feature check for perf_events being enabled.
1100	 *
1101	 * So it's an ABI, do not remove!
1102	 */
1103	{
1104		.procname	= "perf_event_paranoid",
1105		.data		= &sysctl_perf_event_paranoid,
1106		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1107		.mode		= 0644,
1108		.proc_handler	= proc_dointvec,
1109	},
1110	{
1111		.procname	= "perf_event_mlock_kb",
1112		.data		= &sysctl_perf_event_mlock,
1113		.maxlen		= sizeof(sysctl_perf_event_mlock),
1114		.mode		= 0644,
1115		.proc_handler	= proc_dointvec,
1116	},
1117	{
1118		.procname	= "perf_event_max_sample_rate",
1119		.data		= &sysctl_perf_event_sample_rate,
1120		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1121		.mode		= 0644,
1122		.proc_handler	= perf_proc_update_handler,
1123		.extra1		= &one,
1124	},
1125	{
1126		.procname	= "perf_cpu_time_max_percent",
1127		.data		= &sysctl_perf_cpu_time_max_percent,
1128		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1129		.mode		= 0644,
1130		.proc_handler	= perf_cpu_time_max_percent_handler,
1131		.extra1		= &zero,
1132		.extra2		= &one_hundred,
1133	},
1134	{
1135		.procname	= "perf_event_max_stack",
1136		.data		= &sysctl_perf_event_max_stack,
1137		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1138		.mode		= 0644,
1139		.proc_handler	= perf_event_max_stack_handler,
1140		.extra1		= &zero,
1141		.extra2		= &six_hundred_forty_kb,
1142	},
1143	{
1144		.procname	= "perf_event_max_contexts_per_stack",
1145		.data		= &sysctl_perf_event_max_contexts_per_stack,
1146		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1147		.mode		= 0644,
1148		.proc_handler	= perf_event_max_stack_handler,
1149		.extra1		= &zero,
1150		.extra2		= &one_thousand,
1151	},
1152#endif
1153#ifdef CONFIG_KMEMCHECK
1154	{
1155		.procname	= "kmemcheck",
1156		.data		= &kmemcheck_enabled,
1157		.maxlen		= sizeof(int),
1158		.mode		= 0644,
1159		.proc_handler	= proc_dointvec,
1160	},
1161#endif
 
1162	{
1163		.procname	= "panic_on_warn",
1164		.data		= &panic_on_warn,
1165		.maxlen		= sizeof(int),
1166		.mode		= 0644,
1167		.proc_handler	= proc_dointvec_minmax,
1168		.extra1		= &zero,
1169		.extra2		= &one,
1170	},
1171#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1172	{
1173		.procname	= "timer_migration",
1174		.data		= &sysctl_timer_migration,
1175		.maxlen		= sizeof(unsigned int),
1176		.mode		= 0644,
1177		.proc_handler	= timer_migration_handler,
1178	},
1179#endif
1180#ifdef CONFIG_BPF_SYSCALL
1181	{
1182		.procname	= "unprivileged_bpf_disabled",
1183		.data		= &sysctl_unprivileged_bpf_disabled,
1184		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1185		.mode		= 0644,
1186		/* only handle a transition from default "0" to "1" */
1187		.proc_handler	= proc_dointvec_minmax,
1188		.extra1		= &one,
1189		.extra2		= &one,
1190	},
1191#endif
1192#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1193	{
1194		.procname	= "panic_on_rcu_stall",
1195		.data		= &sysctl_panic_on_rcu_stall,
1196		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1197		.mode		= 0644,
1198		.proc_handler	= proc_dointvec_minmax,
1199		.extra1		= &zero,
1200		.extra2		= &one,
1201	},
1202#endif
1203	{ }
1204};
1205
1206static struct ctl_table vm_table[] = {
1207	{
1208		.procname	= "overcommit_memory",
1209		.data		= &sysctl_overcommit_memory,
1210		.maxlen		= sizeof(sysctl_overcommit_memory),
1211		.mode		= 0644,
1212		.proc_handler	= proc_dointvec_minmax,
1213		.extra1		= &zero,
1214		.extra2		= &two,
1215	},
1216	{
1217		.procname	= "panic_on_oom",
1218		.data		= &sysctl_panic_on_oom,
1219		.maxlen		= sizeof(sysctl_panic_on_oom),
1220		.mode		= 0644,
1221		.proc_handler	= proc_dointvec_minmax,
1222		.extra1		= &zero,
1223		.extra2		= &two,
1224	},
1225	{
1226		.procname	= "oom_kill_allocating_task",
1227		.data		= &sysctl_oom_kill_allocating_task,
1228		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1229		.mode		= 0644,
1230		.proc_handler	= proc_dointvec,
1231	},
1232	{
1233		.procname	= "oom_dump_tasks",
1234		.data		= &sysctl_oom_dump_tasks,
1235		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1236		.mode		= 0644,
1237		.proc_handler	= proc_dointvec,
1238	},
1239	{
1240		.procname	= "overcommit_ratio",
1241		.data		= &sysctl_overcommit_ratio,
1242		.maxlen		= sizeof(sysctl_overcommit_ratio),
1243		.mode		= 0644,
1244		.proc_handler	= overcommit_ratio_handler,
1245	},
1246	{
1247		.procname	= "overcommit_kbytes",
1248		.data		= &sysctl_overcommit_kbytes,
1249		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1250		.mode		= 0644,
1251		.proc_handler	= overcommit_kbytes_handler,
1252	},
1253	{
1254		.procname	= "page-cluster", 
1255		.data		= &page_cluster,
1256		.maxlen		= sizeof(int),
1257		.mode		= 0644,
1258		.proc_handler	= proc_dointvec_minmax,
1259		.extra1		= &zero,
1260	},
1261	{
1262		.procname	= "dirty_background_ratio",
1263		.data		= &dirty_background_ratio,
1264		.maxlen		= sizeof(dirty_background_ratio),
1265		.mode		= 0644,
1266		.proc_handler	= dirty_background_ratio_handler,
1267		.extra1		= &zero,
1268		.extra2		= &one_hundred,
1269	},
1270	{
1271		.procname	= "dirty_background_bytes",
1272		.data		= &dirty_background_bytes,
1273		.maxlen		= sizeof(dirty_background_bytes),
1274		.mode		= 0644,
1275		.proc_handler	= dirty_background_bytes_handler,
1276		.extra1		= &one_ul,
1277	},
1278	{
1279		.procname	= "dirty_ratio",
1280		.data		= &vm_dirty_ratio,
1281		.maxlen		= sizeof(vm_dirty_ratio),
1282		.mode		= 0644,
1283		.proc_handler	= dirty_ratio_handler,
1284		.extra1		= &zero,
1285		.extra2		= &one_hundred,
1286	},
1287	{
1288		.procname	= "dirty_bytes",
1289		.data		= &vm_dirty_bytes,
1290		.maxlen		= sizeof(vm_dirty_bytes),
1291		.mode		= 0644,
1292		.proc_handler	= dirty_bytes_handler,
1293		.extra1		= &dirty_bytes_min,
1294	},
1295	{
1296		.procname	= "dirty_writeback_centisecs",
1297		.data		= &dirty_writeback_interval,
1298		.maxlen		= sizeof(dirty_writeback_interval),
1299		.mode		= 0644,
1300		.proc_handler	= dirty_writeback_centisecs_handler,
1301	},
1302	{
1303		.procname	= "dirty_expire_centisecs",
1304		.data		= &dirty_expire_interval,
1305		.maxlen		= sizeof(dirty_expire_interval),
1306		.mode		= 0644,
1307		.proc_handler	= proc_dointvec_minmax,
1308		.extra1		= &zero,
1309	},
1310	{
1311		.procname	= "dirtytime_expire_seconds",
1312		.data		= &dirtytime_expire_interval,
1313		.maxlen		= sizeof(dirty_expire_interval),
1314		.mode		= 0644,
1315		.proc_handler	= dirtytime_interval_handler,
1316		.extra1		= &zero,
1317	},
1318	{
1319		.procname       = "nr_pdflush_threads",
1320		.mode           = 0444 /* read-only */,
1321		.proc_handler   = pdflush_proc_obsolete,
1322	},
1323	{
1324		.procname	= "swappiness",
1325		.data		= &vm_swappiness,
1326		.maxlen		= sizeof(vm_swappiness),
1327		.mode		= 0644,
1328		.proc_handler	= proc_dointvec_minmax,
1329		.extra1		= &zero,
1330		.extra2		= &one_hundred,
1331	},
1332#ifdef CONFIG_HUGETLB_PAGE
1333	{
1334		.procname	= "nr_hugepages",
1335		.data		= NULL,
1336		.maxlen		= sizeof(unsigned long),
1337		.mode		= 0644,
1338		.proc_handler	= hugetlb_sysctl_handler,
 
 
1339	},
1340#ifdef CONFIG_NUMA
1341	{
1342		.procname       = "nr_hugepages_mempolicy",
1343		.data           = NULL,
1344		.maxlen         = sizeof(unsigned long),
1345		.mode           = 0644,
1346		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
 
 
1347	},
1348#endif
1349	 {
1350		.procname	= "hugetlb_shm_group",
1351		.data		= &sysctl_hugetlb_shm_group,
1352		.maxlen		= sizeof(gid_t),
1353		.mode		= 0644,
1354		.proc_handler	= proc_dointvec,
1355	 },
1356	 {
1357		.procname	= "hugepages_treat_as_movable",
1358		.data		= &hugepages_treat_as_movable,
1359		.maxlen		= sizeof(int),
1360		.mode		= 0644,
1361		.proc_handler	= proc_dointvec,
1362	},
1363	{
1364		.procname	= "nr_overcommit_hugepages",
1365		.data		= NULL,
1366		.maxlen		= sizeof(unsigned long),
1367		.mode		= 0644,
1368		.proc_handler	= hugetlb_overcommit_handler,
 
 
1369	},
1370#endif
1371	{
1372		.procname	= "lowmem_reserve_ratio",
1373		.data		= &sysctl_lowmem_reserve_ratio,
1374		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1375		.mode		= 0644,
1376		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1377	},
1378	{
1379		.procname	= "drop_caches",
1380		.data		= &sysctl_drop_caches,
1381		.maxlen		= sizeof(int),
1382		.mode		= 0644,
1383		.proc_handler	= drop_caches_sysctl_handler,
1384		.extra1		= &one,
1385		.extra2		= &four,
1386	},
1387#ifdef CONFIG_COMPACTION
1388	{
1389		.procname	= "compact_memory",
1390		.data		= &sysctl_compact_memory,
1391		.maxlen		= sizeof(int),
1392		.mode		= 0200,
1393		.proc_handler	= sysctl_compaction_handler,
1394	},
1395	{
1396		.procname	= "extfrag_threshold",
1397		.data		= &sysctl_extfrag_threshold,
1398		.maxlen		= sizeof(int),
1399		.mode		= 0644,
1400		.proc_handler	= sysctl_extfrag_handler,
1401		.extra1		= &min_extfrag_threshold,
1402		.extra2		= &max_extfrag_threshold,
1403	},
1404	{
1405		.procname	= "compact_unevictable_allowed",
1406		.data		= &sysctl_compact_unevictable_allowed,
1407		.maxlen		= sizeof(int),
1408		.mode		= 0644,
1409		.proc_handler	= proc_dointvec,
1410		.extra1		= &zero,
1411		.extra2		= &one,
1412	},
1413
1414#endif /* CONFIG_COMPACTION */
1415	{
1416		.procname	= "min_free_kbytes",
1417		.data		= &min_free_kbytes,
1418		.maxlen		= sizeof(min_free_kbytes),
1419		.mode		= 0644,
1420		.proc_handler	= min_free_kbytes_sysctl_handler,
1421		.extra1		= &zero,
1422	},
1423	{
1424		.procname	= "watermark_scale_factor",
1425		.data		= &watermark_scale_factor,
1426		.maxlen		= sizeof(watermark_scale_factor),
1427		.mode		= 0644,
1428		.proc_handler	= watermark_scale_factor_sysctl_handler,
1429		.extra1		= &one,
1430		.extra2		= &one_thousand,
1431	},
1432	{
1433		.procname	= "percpu_pagelist_fraction",
1434		.data		= &percpu_pagelist_fraction,
1435		.maxlen		= sizeof(percpu_pagelist_fraction),
1436		.mode		= 0644,
1437		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1438		.extra1		= &zero,
1439	},
1440#ifdef CONFIG_MMU
1441	{
1442		.procname	= "max_map_count",
1443		.data		= &sysctl_max_map_count,
1444		.maxlen		= sizeof(sysctl_max_map_count),
1445		.mode		= 0644,
1446		.proc_handler	= proc_dointvec_minmax,
1447		.extra1		= &zero,
1448	},
1449#else
1450	{
1451		.procname	= "nr_trim_pages",
1452		.data		= &sysctl_nr_trim_pages,
1453		.maxlen		= sizeof(sysctl_nr_trim_pages),
1454		.mode		= 0644,
1455		.proc_handler	= proc_dointvec_minmax,
1456		.extra1		= &zero,
1457	},
1458#endif
1459	{
1460		.procname	= "laptop_mode",
1461		.data		= &laptop_mode,
1462		.maxlen		= sizeof(laptop_mode),
1463		.mode		= 0644,
1464		.proc_handler	= proc_dointvec_jiffies,
1465	},
1466	{
1467		.procname	= "block_dump",
1468		.data		= &block_dump,
1469		.maxlen		= sizeof(block_dump),
1470		.mode		= 0644,
1471		.proc_handler	= proc_dointvec,
1472		.extra1		= &zero,
1473	},
1474	{
1475		.procname	= "vfs_cache_pressure",
1476		.data		= &sysctl_vfs_cache_pressure,
1477		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1478		.mode		= 0644,
1479		.proc_handler	= proc_dointvec,
1480		.extra1		= &zero,
1481	},
1482#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1483	{
1484		.procname	= "legacy_va_layout",
1485		.data		= &sysctl_legacy_va_layout,
1486		.maxlen		= sizeof(sysctl_legacy_va_layout),
1487		.mode		= 0644,
1488		.proc_handler	= proc_dointvec,
1489		.extra1		= &zero,
1490	},
1491#endif
1492#ifdef CONFIG_NUMA
1493	{
1494		.procname	= "zone_reclaim_mode",
1495		.data		= &node_reclaim_mode,
1496		.maxlen		= sizeof(node_reclaim_mode),
1497		.mode		= 0644,
1498		.proc_handler	= proc_dointvec,
1499		.extra1		= &zero,
1500	},
1501	{
1502		.procname	= "min_unmapped_ratio",
1503		.data		= &sysctl_min_unmapped_ratio,
1504		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1505		.mode		= 0644,
1506		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1507		.extra1		= &zero,
1508		.extra2		= &one_hundred,
1509	},
1510	{
1511		.procname	= "min_slab_ratio",
1512		.data		= &sysctl_min_slab_ratio,
1513		.maxlen		= sizeof(sysctl_min_slab_ratio),
1514		.mode		= 0644,
1515		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1516		.extra1		= &zero,
1517		.extra2		= &one_hundred,
1518	},
1519#endif
1520#ifdef CONFIG_SMP
1521	{
1522		.procname	= "stat_interval",
1523		.data		= &sysctl_stat_interval,
1524		.maxlen		= sizeof(sysctl_stat_interval),
1525		.mode		= 0644,
1526		.proc_handler	= proc_dointvec_jiffies,
1527	},
1528	{
1529		.procname	= "stat_refresh",
1530		.data		= NULL,
1531		.maxlen		= 0,
1532		.mode		= 0600,
1533		.proc_handler	= vmstat_refresh,
1534	},
1535#endif
1536#ifdef CONFIG_MMU
1537	{
1538		.procname	= "mmap_min_addr",
1539		.data		= &dac_mmap_min_addr,
1540		.maxlen		= sizeof(unsigned long),
1541		.mode		= 0644,
1542		.proc_handler	= mmap_min_addr_handler,
1543	},
1544#endif
1545#ifdef CONFIG_NUMA
1546	{
1547		.procname	= "numa_zonelist_order",
1548		.data		= &numa_zonelist_order,
1549		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1550		.mode		= 0644,
1551		.proc_handler	= numa_zonelist_order_handler,
1552	},
1553#endif
1554#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1555   (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1556	{
1557		.procname	= "vdso_enabled",
1558#ifdef CONFIG_X86_32
1559		.data		= &vdso32_enabled,
1560		.maxlen		= sizeof(vdso32_enabled),
1561#else
1562		.data		= &vdso_enabled,
1563		.maxlen		= sizeof(vdso_enabled),
1564#endif
1565		.mode		= 0644,
1566		.proc_handler	= proc_dointvec,
1567		.extra1		= &zero,
1568	},
1569#endif
1570#ifdef CONFIG_HIGHMEM
1571	{
1572		.procname	= "highmem_is_dirtyable",
1573		.data		= &vm_highmem_is_dirtyable,
1574		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1575		.mode		= 0644,
1576		.proc_handler	= proc_dointvec_minmax,
1577		.extra1		= &zero,
1578		.extra2		= &one,
1579	},
1580#endif
 
 
 
 
 
 
 
1581#ifdef CONFIG_MEMORY_FAILURE
1582	{
1583		.procname	= "memory_failure_early_kill",
1584		.data		= &sysctl_memory_failure_early_kill,
1585		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1586		.mode		= 0644,
1587		.proc_handler	= proc_dointvec_minmax,
1588		.extra1		= &zero,
1589		.extra2		= &one,
1590	},
1591	{
1592		.procname	= "memory_failure_recovery",
1593		.data		= &sysctl_memory_failure_recovery,
1594		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1595		.mode		= 0644,
1596		.proc_handler	= proc_dointvec_minmax,
1597		.extra1		= &zero,
1598		.extra2		= &one,
1599	},
1600#endif
1601	{
1602		.procname	= "user_reserve_kbytes",
1603		.data		= &sysctl_user_reserve_kbytes,
1604		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1605		.mode		= 0644,
1606		.proc_handler	= proc_doulongvec_minmax,
1607	},
1608	{
1609		.procname	= "admin_reserve_kbytes",
1610		.data		= &sysctl_admin_reserve_kbytes,
1611		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1612		.mode		= 0644,
1613		.proc_handler	= proc_doulongvec_minmax,
1614	},
1615#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1616	{
1617		.procname	= "mmap_rnd_bits",
1618		.data		= &mmap_rnd_bits,
1619		.maxlen		= sizeof(mmap_rnd_bits),
1620		.mode		= 0600,
1621		.proc_handler	= proc_dointvec_minmax,
1622		.extra1		= (void *)&mmap_rnd_bits_min,
1623		.extra2		= (void *)&mmap_rnd_bits_max,
1624	},
1625#endif
1626#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1627	{
1628		.procname	= "mmap_rnd_compat_bits",
1629		.data		= &mmap_rnd_compat_bits,
1630		.maxlen		= sizeof(mmap_rnd_compat_bits),
1631		.mode		= 0600,
1632		.proc_handler	= proc_dointvec_minmax,
1633		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1634		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1635	},
1636#endif
1637	{ }
1638};
1639
 
 
 
 
 
 
1640static struct ctl_table fs_table[] = {
1641	{
1642		.procname	= "inode-nr",
1643		.data		= &inodes_stat,
1644		.maxlen		= 2*sizeof(long),
1645		.mode		= 0444,
1646		.proc_handler	= proc_nr_inodes,
1647	},
1648	{
1649		.procname	= "inode-state",
1650		.data		= &inodes_stat,
1651		.maxlen		= 7*sizeof(long),
1652		.mode		= 0444,
1653		.proc_handler	= proc_nr_inodes,
1654	},
1655	{
1656		.procname	= "file-nr",
1657		.data		= &files_stat,
1658		.maxlen		= sizeof(files_stat),
1659		.mode		= 0444,
1660		.proc_handler	= proc_nr_files,
1661	},
1662	{
1663		.procname	= "file-max",
1664		.data		= &files_stat.max_files,
1665		.maxlen		= sizeof(files_stat.max_files),
1666		.mode		= 0644,
1667		.proc_handler	= proc_doulongvec_minmax,
1668	},
1669	{
1670		.procname	= "nr_open",
1671		.data		= &sysctl_nr_open,
1672		.maxlen		= sizeof(unsigned int),
1673		.mode		= 0644,
1674		.proc_handler	= proc_dointvec_minmax,
1675		.extra1		= &sysctl_nr_open_min,
1676		.extra2		= &sysctl_nr_open_max,
1677	},
1678	{
1679		.procname	= "dentry-state",
1680		.data		= &dentry_stat,
1681		.maxlen		= 6*sizeof(long),
1682		.mode		= 0444,
1683		.proc_handler	= proc_nr_dentry,
1684	},
1685	{
1686		.procname	= "overflowuid",
1687		.data		= &fs_overflowuid,
1688		.maxlen		= sizeof(int),
1689		.mode		= 0644,
1690		.proc_handler	= proc_dointvec_minmax,
1691		.extra1		= &minolduid,
1692		.extra2		= &maxolduid,
1693	},
1694	{
1695		.procname	= "overflowgid",
1696		.data		= &fs_overflowgid,
1697		.maxlen		= sizeof(int),
1698		.mode		= 0644,
1699		.proc_handler	= proc_dointvec_minmax,
1700		.extra1		= &minolduid,
1701		.extra2		= &maxolduid,
1702	},
1703#ifdef CONFIG_FILE_LOCKING
1704	{
1705		.procname	= "leases-enable",
1706		.data		= &leases_enable,
1707		.maxlen		= sizeof(int),
1708		.mode		= 0644,
1709		.proc_handler	= proc_dointvec,
1710	},
1711#endif
1712#ifdef CONFIG_DNOTIFY
1713	{
1714		.procname	= "dir-notify-enable",
1715		.data		= &dir_notify_enable,
1716		.maxlen		= sizeof(int),
1717		.mode		= 0644,
1718		.proc_handler	= proc_dointvec,
1719	},
1720#endif
1721#ifdef CONFIG_MMU
1722#ifdef CONFIG_FILE_LOCKING
1723	{
1724		.procname	= "lease-break-time",
1725		.data		= &lease_break_time,
1726		.maxlen		= sizeof(int),
1727		.mode		= 0644,
1728		.proc_handler	= proc_dointvec,
1729	},
1730#endif
1731#ifdef CONFIG_AIO
1732	{
1733		.procname	= "aio-nr",
1734		.data		= &aio_nr,
1735		.maxlen		= sizeof(aio_nr),
1736		.mode		= 0444,
1737		.proc_handler	= proc_doulongvec_minmax,
1738	},
1739	{
1740		.procname	= "aio-max-nr",
1741		.data		= &aio_max_nr,
1742		.maxlen		= sizeof(aio_max_nr),
1743		.mode		= 0644,
1744		.proc_handler	= proc_doulongvec_minmax,
1745	},
1746#endif /* CONFIG_AIO */
1747#ifdef CONFIG_INOTIFY_USER
1748	{
1749		.procname	= "inotify",
1750		.mode		= 0555,
1751		.child		= inotify_table,
1752	},
1753#endif	
1754#ifdef CONFIG_EPOLL
1755	{
1756		.procname	= "epoll",
1757		.mode		= 0555,
1758		.child		= epoll_table,
1759	},
1760#endif
1761#endif
1762	{
1763		.procname	= "protected_symlinks",
1764		.data		= &sysctl_protected_symlinks,
1765		.maxlen		= sizeof(int),
1766		.mode		= 0600,
1767		.proc_handler	= proc_dointvec_minmax,
1768		.extra1		= &zero,
1769		.extra2		= &one,
1770	},
1771	{
1772		.procname	= "protected_hardlinks",
1773		.data		= &sysctl_protected_hardlinks,
1774		.maxlen		= sizeof(int),
1775		.mode		= 0600,
1776		.proc_handler	= proc_dointvec_minmax,
1777		.extra1		= &zero,
1778		.extra2		= &one,
1779	},
1780	{
1781		.procname	= "suid_dumpable",
1782		.data		= &suid_dumpable,
1783		.maxlen		= sizeof(int),
1784		.mode		= 0644,
1785		.proc_handler	= proc_dointvec_minmax_coredump,
1786		.extra1		= &zero,
1787		.extra2		= &two,
1788	},
1789#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1790	{
1791		.procname	= "binfmt_misc",
1792		.mode		= 0555,
1793		.child		= sysctl_mount_point,
1794	},
1795#endif
1796	{
1797		.procname	= "pipe-max-size",
1798		.data		= &pipe_max_size,
1799		.maxlen		= sizeof(int),
1800		.mode		= 0644,
1801		.proc_handler	= &pipe_proc_fn,
1802		.extra1		= &pipe_min_size,
1803	},
1804	{
1805		.procname	= "pipe-user-pages-hard",
1806		.data		= &pipe_user_pages_hard,
1807		.maxlen		= sizeof(pipe_user_pages_hard),
1808		.mode		= 0644,
1809		.proc_handler	= proc_doulongvec_minmax,
1810	},
1811	{
1812		.procname	= "pipe-user-pages-soft",
1813		.data		= &pipe_user_pages_soft,
1814		.maxlen		= sizeof(pipe_user_pages_soft),
1815		.mode		= 0644,
1816		.proc_handler	= proc_doulongvec_minmax,
1817	},
1818	{
1819		.procname	= "mount-max",
1820		.data		= &sysctl_mount_max,
1821		.maxlen		= sizeof(unsigned int),
1822		.mode		= 0644,
1823		.proc_handler	= proc_dointvec_minmax,
1824		.extra1		= &one,
1825	},
1826	{ }
1827};
1828
1829static struct ctl_table debug_table[] = {
1830#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
 
1831	{
1832		.procname	= "exception-trace",
1833		.data		= &show_unhandled_signals,
1834		.maxlen		= sizeof(int),
1835		.mode		= 0644,
1836		.proc_handler	= proc_dointvec
1837	},
1838#endif
1839#if defined(CONFIG_OPTPROBES)
1840	{
1841		.procname	= "kprobes-optimization",
1842		.data		= &sysctl_kprobes_optimization,
1843		.maxlen		= sizeof(int),
1844		.mode		= 0644,
1845		.proc_handler	= proc_kprobes_optimization_handler,
1846		.extra1		= &zero,
1847		.extra2		= &one,
1848	},
1849#endif
1850	{ }
1851};
1852
1853static struct ctl_table dev_table[] = {
1854	{ }
1855};
1856
1857int __init sysctl_init(void)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1858{
1859	struct ctl_table_header *hdr;
 
 
 
 
 
 
 
 
1860
1861	hdr = register_sysctl_table(sysctl_base_table);
1862	kmemleak_not_leak(hdr);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1863	return 0;
1864}
1865
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1866#endif /* CONFIG_SYSCTL */
1867
1868/*
1869 * /proc/sys support
1870 */
1871
1872#ifdef CONFIG_PROC_SYSCTL
1873
1874static int _proc_do_string(char *data, int maxlen, int write,
1875			   char __user *buffer,
1876			   size_t *lenp, loff_t *ppos)
1877{
1878	size_t len;
1879	char __user *p;
1880	char c;
1881
1882	if (!data || !maxlen || !*lenp) {
1883		*lenp = 0;
1884		return 0;
1885	}
1886
1887	if (write) {
1888		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1889			/* Only continue writes not past the end of buffer. */
1890			len = strlen(data);
1891			if (len > maxlen - 1)
1892				len = maxlen - 1;
1893
1894			if (*ppos > len)
1895				return 0;
1896			len = *ppos;
1897		} else {
1898			/* Start writing from beginning of buffer. */
1899			len = 0;
1900		}
1901
1902		*ppos += *lenp;
1903		p = buffer;
1904		while ((p - buffer) < *lenp && len < maxlen - 1) {
1905			if (get_user(c, p++))
1906				return -EFAULT;
1907			if (c == 0 || c == '\n')
1908				break;
1909			data[len++] = c;
1910		}
1911		data[len] = 0;
 
 
 
 
 
1912	} else {
1913		len = strlen(data);
1914		if (len > maxlen)
1915			len = maxlen;
1916
1917		if (*ppos > len) {
1918			*lenp = 0;
1919			return 0;
1920		}
1921
1922		data += *ppos;
1923		len  -= *ppos;
1924
1925		if (len > *lenp)
1926			len = *lenp;
1927		if (len)
1928			if (copy_to_user(buffer, data, len))
1929				return -EFAULT;
1930		if (len < *lenp) {
1931			if (put_user('\n', buffer + len))
1932				return -EFAULT;
1933			len++;
1934		}
1935		*lenp = len;
1936		*ppos += len;
1937	}
1938	return 0;
1939}
1940
1941static void warn_sysctl_write(struct ctl_table *table)
1942{
1943	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1944		"This will not be supported in the future. To silence this\n"
1945		"warning, set kernel.sysctl_writes_strict = -1\n",
1946		current->comm, table->procname);
1947}
1948
1949/**
1950 * proc_dostring - read a string sysctl
1951 * @table: the sysctl table
1952 * @write: %TRUE if this is a write to the sysctl file
1953 * @buffer: the user buffer
1954 * @lenp: the size of the user buffer
1955 * @ppos: file position
1956 *
1957 * Reads/writes a string from/to the user buffer. If the kernel
1958 * buffer provided is not large enough to hold the string, the
1959 * string is truncated. The copied string is %NULL-terminated.
1960 * If the string is being read by the user process, it is copied
1961 * and a newline '\n' is added. It is truncated if the buffer is
1962 * not large enough.
1963 *
1964 * Returns 0 on success.
1965 */
1966int proc_dostring(struct ctl_table *table, int write,
1967		  void __user *buffer, size_t *lenp, loff_t *ppos)
1968{
1969	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1970		warn_sysctl_write(table);
1971
1972	return _proc_do_string((char *)(table->data), table->maxlen, write,
1973			       (char __user *)buffer, lenp, ppos);
1974}
1975
1976static size_t proc_skip_spaces(char **buf)
1977{
1978	size_t ret;
1979	char *tmp = skip_spaces(*buf);
1980	ret = tmp - *buf;
1981	*buf = tmp;
1982	return ret;
1983}
1984
1985static void proc_skip_char(char **buf, size_t *size, const char v)
1986{
1987	while (*size) {
1988		if (**buf != v)
1989			break;
1990		(*size)--;
1991		(*buf)++;
1992	}
1993}
1994
1995#define TMPBUFLEN 22
1996/**
1997 * proc_get_long - reads an ASCII formatted integer from a user buffer
1998 *
1999 * @buf: a kernel buffer
2000 * @size: size of the kernel buffer
2001 * @val: this is where the number will be stored
2002 * @neg: set to %TRUE if number is negative
2003 * @perm_tr: a vector which contains the allowed trailers
2004 * @perm_tr_len: size of the perm_tr vector
2005 * @tr: pointer to store the trailer character
2006 *
2007 * In case of success %0 is returned and @buf and @size are updated with
2008 * the amount of bytes read. If @tr is non-NULL and a trailing
2009 * character exists (size is non-zero after returning from this
2010 * function), @tr is updated with the trailing character.
2011 */
2012static int proc_get_long(char **buf, size_t *size,
2013			  unsigned long *val, bool *neg,
2014			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2015{
2016	int len;
2017	char *p, tmp[TMPBUFLEN];
2018
2019	if (!*size)
2020		return -EINVAL;
2021
2022	len = *size;
2023	if (len > TMPBUFLEN - 1)
2024		len = TMPBUFLEN - 1;
2025
2026	memcpy(tmp, *buf, len);
2027
2028	tmp[len] = 0;
2029	p = tmp;
2030	if (*p == '-' && *size > 1) {
2031		*neg = true;
2032		p++;
2033	} else
2034		*neg = false;
2035	if (!isdigit(*p))
2036		return -EINVAL;
2037
2038	*val = simple_strtoul(p, &p, 0);
2039
2040	len = p - tmp;
2041
2042	/* We don't know if the next char is whitespace thus we may accept
2043	 * invalid integers (e.g. 1234...a) or two integers instead of one
2044	 * (e.g. 123...1). So lets not allow such large numbers. */
2045	if (len == TMPBUFLEN - 1)
2046		return -EINVAL;
2047
2048	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2049		return -EINVAL;
2050
2051	if (tr && (len < *size))
2052		*tr = *p;
2053
2054	*buf += len;
2055	*size -= len;
2056
2057	return 0;
2058}
2059
2060/**
2061 * proc_put_long - converts an integer to a decimal ASCII formatted string
2062 *
2063 * @buf: the user buffer
2064 * @size: the size of the user buffer
2065 * @val: the integer to be converted
2066 * @neg: sign of the number, %TRUE for negative
2067 *
2068 * In case of success %0 is returned and @buf and @size are updated with
2069 * the amount of bytes written.
2070 */
2071static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2072			  bool neg)
2073{
2074	int len;
2075	char tmp[TMPBUFLEN], *p = tmp;
2076
2077	sprintf(p, "%s%lu", neg ? "-" : "", val);
2078	len = strlen(tmp);
2079	if (len > *size)
2080		len = *size;
2081	if (copy_to_user(*buf, tmp, len))
2082		return -EFAULT;
2083	*size -= len;
2084	*buf += len;
2085	return 0;
2086}
2087#undef TMPBUFLEN
2088
2089static int proc_put_char(void __user **buf, size_t *size, char c)
2090{
2091	if (*size) {
2092		char __user **buffer = (char __user **)buf;
2093		if (put_user(c, *buffer))
2094			return -EFAULT;
2095		(*size)--, (*buffer)++;
2096		*buf = *buffer;
2097	}
2098	return 0;
2099}
2100
2101static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2102				 int *valp,
2103				 int write, void *data)
2104{
2105	if (write) {
2106		if (*negp) {
2107			if (*lvalp > (unsigned long) INT_MAX + 1)
2108				return -EINVAL;
2109			*valp = -*lvalp;
2110		} else {
2111			if (*lvalp > (unsigned long) INT_MAX)
2112				return -EINVAL;
2113			*valp = *lvalp;
2114		}
2115	} else {
2116		int val = *valp;
2117		if (val < 0) {
2118			*negp = true;
2119			*lvalp = -(unsigned long)val;
2120		} else {
2121			*negp = false;
2122			*lvalp = (unsigned long)val;
2123		}
2124	}
2125	return 0;
2126}
2127
2128static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2129				 int *valp,
2130				 int write, void *data)
2131{
2132	if (write) {
2133		if (*negp)
2134			return -EINVAL;
2135		*valp = *lvalp;
2136	} else {
2137		unsigned int val = *valp;
2138		*lvalp = (unsigned long)val;
2139	}
2140	return 0;
2141}
2142
2143static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2144
2145static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2146		  int write, void __user *buffer,
2147		  size_t *lenp, loff_t *ppos,
2148		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2149			      int write, void *data),
2150		  void *data)
2151{
2152	int *i, vleft, first = 1, err = 0;
 
2153	size_t left;
2154	char *kbuf = NULL, *p;
2155	
2156	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2157		*lenp = 0;
2158		return 0;
2159	}
2160	
2161	i = (int *) tbl_data;
2162	vleft = table->maxlen / sizeof(*i);
2163	left = *lenp;
2164
2165	if (!conv)
2166		conv = do_proc_dointvec_conv;
2167
2168	if (write) {
2169		if (*ppos) {
2170			switch (sysctl_writes_strict) {
2171			case SYSCTL_WRITES_STRICT:
2172				goto out;
2173			case SYSCTL_WRITES_WARN:
2174				warn_sysctl_write(table);
2175				break;
2176			default:
2177				break;
2178			}
2179		}
2180
2181		if (left > PAGE_SIZE - 1)
2182			left = PAGE_SIZE - 1;
2183		p = kbuf = memdup_user_nul(buffer, left);
2184		if (IS_ERR(kbuf))
2185			return PTR_ERR(kbuf);
 
 
 
 
 
 
2186	}
2187
2188	for (; left && vleft--; i++, first=0) {
2189		unsigned long lval;
2190		bool neg;
2191
2192		if (write) {
2193			left -= proc_skip_spaces(&p);
2194
2195			if (!left)
2196				break;
2197			err = proc_get_long(&p, &left, &lval, &neg,
2198					     proc_wspace_sep,
2199					     sizeof(proc_wspace_sep), NULL);
2200			if (err)
2201				break;
2202			if (conv(&neg, &lval, i, 1, data)) {
2203				err = -EINVAL;
2204				break;
2205			}
2206		} else {
2207			if (conv(&neg, &lval, i, 0, data)) {
2208				err = -EINVAL;
2209				break;
2210			}
2211			if (!first)
2212				err = proc_put_char(&buffer, &left, '\t');
2213			if (err)
2214				break;
2215			err = proc_put_long(&buffer, &left, lval, neg);
2216			if (err)
2217				break;
2218		}
2219	}
2220
2221	if (!write && !first && left && !err)
2222		err = proc_put_char(&buffer, &left, '\n');
2223	if (write && !err && left)
2224		left -= proc_skip_spaces(&p);
 
2225	if (write) {
2226		kfree(kbuf);
2227		if (first)
2228			return err ? : -EINVAL;
2229	}
2230	*lenp -= left;
2231out:
2232	*ppos += *lenp;
2233	return err;
2234}
2235
2236static int do_proc_dointvec(struct ctl_table *table, int write,
2237		  void __user *buffer, size_t *lenp, loff_t *ppos,
2238		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2239			      int write, void *data),
2240		  void *data)
2241{
2242	return __do_proc_dointvec(table->data, table, write,
2243			buffer, lenp, ppos, conv, data);
2244}
2245
2246/**
2247 * proc_dointvec - read a vector of integers
2248 * @table: the sysctl table
2249 * @write: %TRUE if this is a write to the sysctl file
2250 * @buffer: the user buffer
2251 * @lenp: the size of the user buffer
2252 * @ppos: file position
2253 *
2254 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2255 * values from/to the user buffer, treated as an ASCII string. 
2256 *
2257 * Returns 0 on success.
2258 */
2259int proc_dointvec(struct ctl_table *table, int write,
2260		     void __user *buffer, size_t *lenp, loff_t *ppos)
2261{
2262	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2263}
2264
2265/**
2266 * proc_douintvec - read a vector of unsigned integers
2267 * @table: the sysctl table
2268 * @write: %TRUE if this is a write to the sysctl file
2269 * @buffer: the user buffer
2270 * @lenp: the size of the user buffer
2271 * @ppos: file position
2272 *
2273 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2274 * values from/to the user buffer, treated as an ASCII string.
2275 *
2276 * Returns 0 on success.
2277 */
2278int proc_douintvec(struct ctl_table *table, int write,
2279		     void __user *buffer, size_t *lenp, loff_t *ppos)
2280{
2281	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2282				do_proc_douintvec_conv, NULL);
2283}
2284
2285/*
2286 * Taint values can only be increased
2287 * This means we can safely use a temporary.
2288 */
2289static int proc_taint(struct ctl_table *table, int write,
2290			       void __user *buffer, size_t *lenp, loff_t *ppos)
2291{
2292	struct ctl_table t;
2293	unsigned long tmptaint = get_taint();
2294	int err;
2295
2296	if (write && !capable(CAP_SYS_ADMIN))
2297		return -EPERM;
2298
2299	t = *table;
2300	t.data = &tmptaint;
2301	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2302	if (err < 0)
2303		return err;
2304
2305	if (write) {
2306		/*
2307		 * Poor man's atomic or. Not worth adding a primitive
2308		 * to everyone's atomic.h for this
2309		 */
2310		int i;
2311		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2312			if ((tmptaint >> i) & 1)
2313				add_taint(i, LOCKDEP_STILL_OK);
2314		}
2315	}
2316
2317	return err;
2318}
2319
2320#ifdef CONFIG_PRINTK
2321static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2322				void __user *buffer, size_t *lenp, loff_t *ppos)
2323{
2324	if (write && !capable(CAP_SYS_ADMIN))
2325		return -EPERM;
2326
2327	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2328}
2329#endif
2330
2331struct do_proc_dointvec_minmax_conv_param {
2332	int *min;
2333	int *max;
2334};
2335
2336static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2337					int *valp,
2338					int write, void *data)
2339{
2340	struct do_proc_dointvec_minmax_conv_param *param = data;
2341	if (write) {
2342		int val = *negp ? -*lvalp : *lvalp;
2343		if ((param->min && *param->min > val) ||
2344		    (param->max && *param->max < val))
2345			return -EINVAL;
2346		*valp = val;
2347	} else {
2348		int val = *valp;
2349		if (val < 0) {
2350			*negp = true;
2351			*lvalp = -(unsigned long)val;
2352		} else {
2353			*negp = false;
2354			*lvalp = (unsigned long)val;
2355		}
2356	}
2357	return 0;
2358}
2359
2360/**
2361 * proc_dointvec_minmax - read a vector of integers with min/max values
2362 * @table: the sysctl table
2363 * @write: %TRUE if this is a write to the sysctl file
2364 * @buffer: the user buffer
2365 * @lenp: the size of the user buffer
2366 * @ppos: file position
2367 *
2368 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2369 * values from/to the user buffer, treated as an ASCII string.
2370 *
2371 * This routine will ensure the values are within the range specified by
2372 * table->extra1 (min) and table->extra2 (max).
2373 *
2374 * Returns 0 on success.
2375 */
2376int proc_dointvec_minmax(struct ctl_table *table, int write,
2377		  void __user *buffer, size_t *lenp, loff_t *ppos)
2378{
2379	struct do_proc_dointvec_minmax_conv_param param = {
2380		.min = (int *) table->extra1,
2381		.max = (int *) table->extra2,
2382	};
2383	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2384				do_proc_dointvec_minmax_conv, &param);
2385}
2386
2387static void validate_coredump_safety(void)
2388{
2389#ifdef CONFIG_COREDUMP
2390	if (suid_dumpable == SUID_DUMP_ROOT &&
2391	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2392		printk(KERN_WARNING
2393"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2394"Pipe handler or fully qualified core dump path required.\n"
2395"Set kernel.core_pattern before fs.suid_dumpable.\n"
2396		);
2397	}
2398#endif
2399}
2400
2401static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2402		void __user *buffer, size_t *lenp, loff_t *ppos)
2403{
2404	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2405	if (!error)
2406		validate_coredump_safety();
2407	return error;
2408}
2409
2410#ifdef CONFIG_COREDUMP
2411static int proc_dostring_coredump(struct ctl_table *table, int write,
2412		  void __user *buffer, size_t *lenp, loff_t *ppos)
2413{
2414	int error = proc_dostring(table, write, buffer, lenp, ppos);
2415	if (!error)
2416		validate_coredump_safety();
2417	return error;
2418}
2419#endif
2420
2421static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2422				     void __user *buffer,
2423				     size_t *lenp, loff_t *ppos,
2424				     unsigned long convmul,
2425				     unsigned long convdiv)
2426{
2427	unsigned long *i, *min, *max;
2428	int vleft, first = 1, err = 0;
 
2429	size_t left;
2430	char *kbuf = NULL, *p;
2431
2432	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2433		*lenp = 0;
2434		return 0;
2435	}
2436
2437	i = (unsigned long *) data;
2438	min = (unsigned long *) table->extra1;
2439	max = (unsigned long *) table->extra2;
2440	vleft = table->maxlen / sizeof(unsigned long);
2441	left = *lenp;
2442
2443	if (write) {
2444		if (*ppos) {
2445			switch (sysctl_writes_strict) {
2446			case SYSCTL_WRITES_STRICT:
2447				goto out;
2448			case SYSCTL_WRITES_WARN:
2449				warn_sysctl_write(table);
2450				break;
2451			default:
2452				break;
2453			}
2454		}
2455
2456		if (left > PAGE_SIZE - 1)
2457			left = PAGE_SIZE - 1;
2458		p = kbuf = memdup_user_nul(buffer, left);
2459		if (IS_ERR(kbuf))
2460			return PTR_ERR(kbuf);
 
 
 
 
 
 
2461	}
2462
2463	for (; left && vleft--; i++, first = 0) {
2464		unsigned long val;
2465
2466		if (write) {
2467			bool neg;
2468
2469			left -= proc_skip_spaces(&p);
2470
2471			err = proc_get_long(&p, &left, &val, &neg,
2472					     proc_wspace_sep,
2473					     sizeof(proc_wspace_sep), NULL);
2474			if (err)
2475				break;
2476			if (neg)
2477				continue;
2478			val = convmul * val / convdiv;
2479			if ((min && val < *min) || (max && val > *max))
2480				continue;
2481			*i = val;
2482		} else {
2483			val = convdiv * (*i) / convmul;
2484			if (!first) {
2485				err = proc_put_char(&buffer, &left, '\t');
2486				if (err)
2487					break;
2488			}
2489			err = proc_put_long(&buffer, &left, val, false);
2490			if (err)
2491				break;
2492		}
2493	}
2494
2495	if (!write && !first && left && !err)
2496		err = proc_put_char(&buffer, &left, '\n');
2497	if (write && !err)
2498		left -= proc_skip_spaces(&p);
 
2499	if (write) {
2500		kfree(kbuf);
2501		if (first)
2502			return err ? : -EINVAL;
2503	}
2504	*lenp -= left;
2505out:
2506	*ppos += *lenp;
2507	return err;
2508}
2509
2510static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2511				     void __user *buffer,
2512				     size_t *lenp, loff_t *ppos,
2513				     unsigned long convmul,
2514				     unsigned long convdiv)
2515{
2516	return __do_proc_doulongvec_minmax(table->data, table, write,
2517			buffer, lenp, ppos, convmul, convdiv);
2518}
2519
2520/**
2521 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2522 * @table: the sysctl table
2523 * @write: %TRUE if this is a write to the sysctl file
2524 * @buffer: the user buffer
2525 * @lenp: the size of the user buffer
2526 * @ppos: file position
2527 *
2528 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2529 * values from/to the user buffer, treated as an ASCII string.
2530 *
2531 * This routine will ensure the values are within the range specified by
2532 * table->extra1 (min) and table->extra2 (max).
2533 *
2534 * Returns 0 on success.
2535 */
2536int proc_doulongvec_minmax(struct ctl_table *table, int write,
2537			   void __user *buffer, size_t *lenp, loff_t *ppos)
2538{
2539    return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2540}
2541
2542/**
2543 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2544 * @table: the sysctl table
2545 * @write: %TRUE if this is a write to the sysctl file
2546 * @buffer: the user buffer
2547 * @lenp: the size of the user buffer
2548 * @ppos: file position
2549 *
2550 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551 * values from/to the user buffer, treated as an ASCII string. The values
2552 * are treated as milliseconds, and converted to jiffies when they are stored.
2553 *
2554 * This routine will ensure the values are within the range specified by
2555 * table->extra1 (min) and table->extra2 (max).
2556 *
2557 * Returns 0 on success.
2558 */
2559int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2560				      void __user *buffer,
2561				      size_t *lenp, loff_t *ppos)
2562{
2563    return do_proc_doulongvec_minmax(table, write, buffer,
2564				     lenp, ppos, HZ, 1000l);
2565}
2566
2567
2568static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2569					 int *valp,
2570					 int write, void *data)
2571{
2572	if (write) {
2573		if (*lvalp > LONG_MAX / HZ)
2574			return 1;
2575		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2576	} else {
2577		int val = *valp;
2578		unsigned long lval;
2579		if (val < 0) {
2580			*negp = true;
2581			lval = -(unsigned long)val;
2582		} else {
2583			*negp = false;
2584			lval = (unsigned long)val;
2585		}
2586		*lvalp = lval / HZ;
2587	}
2588	return 0;
2589}
2590
2591static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2592						int *valp,
2593						int write, void *data)
2594{
2595	if (write) {
2596		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2597			return 1;
2598		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2599	} else {
2600		int val = *valp;
2601		unsigned long lval;
2602		if (val < 0) {
2603			*negp = true;
2604			lval = -(unsigned long)val;
2605		} else {
2606			*negp = false;
2607			lval = (unsigned long)val;
2608		}
2609		*lvalp = jiffies_to_clock_t(lval);
2610	}
2611	return 0;
2612}
2613
2614static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2615					    int *valp,
2616					    int write, void *data)
2617{
2618	if (write) {
2619		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2620
2621		if (jif > INT_MAX)
2622			return 1;
2623		*valp = (int)jif;
2624	} else {
2625		int val = *valp;
2626		unsigned long lval;
2627		if (val < 0) {
2628			*negp = true;
2629			lval = -(unsigned long)val;
2630		} else {
2631			*negp = false;
2632			lval = (unsigned long)val;
2633		}
2634		*lvalp = jiffies_to_msecs(lval);
2635	}
2636	return 0;
2637}
2638
2639/**
2640 * proc_dointvec_jiffies - read a vector of integers as seconds
2641 * @table: the sysctl table
2642 * @write: %TRUE if this is a write to the sysctl file
2643 * @buffer: the user buffer
2644 * @lenp: the size of the user buffer
2645 * @ppos: file position
2646 *
2647 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2648 * values from/to the user buffer, treated as an ASCII string. 
2649 * The values read are assumed to be in seconds, and are converted into
2650 * jiffies.
2651 *
2652 * Returns 0 on success.
2653 */
2654int proc_dointvec_jiffies(struct ctl_table *table, int write,
2655			  void __user *buffer, size_t *lenp, loff_t *ppos)
2656{
2657    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2658		    	    do_proc_dointvec_jiffies_conv,NULL);
2659}
2660
2661/**
2662 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2663 * @table: the sysctl table
2664 * @write: %TRUE if this is a write to the sysctl file
2665 * @buffer: the user buffer
2666 * @lenp: the size of the user buffer
2667 * @ppos: pointer to the file position
2668 *
2669 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2670 * values from/to the user buffer, treated as an ASCII string. 
2671 * The values read are assumed to be in 1/USER_HZ seconds, and 
2672 * are converted into jiffies.
2673 *
2674 * Returns 0 on success.
2675 */
2676int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2677				 void __user *buffer, size_t *lenp, loff_t *ppos)
2678{
2679    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2680		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2681}
2682
2683/**
2684 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2685 * @table: the sysctl table
2686 * @write: %TRUE if this is a write to the sysctl file
2687 * @buffer: the user buffer
2688 * @lenp: the size of the user buffer
2689 * @ppos: file position
2690 * @ppos: the current position in the file
2691 *
2692 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693 * values from/to the user buffer, treated as an ASCII string. 
2694 * The values read are assumed to be in 1/1000 seconds, and 
2695 * are converted into jiffies.
2696 *
2697 * Returns 0 on success.
2698 */
2699int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2700			     void __user *buffer, size_t *lenp, loff_t *ppos)
2701{
2702	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2703				do_proc_dointvec_ms_jiffies_conv, NULL);
2704}
2705
2706static int proc_do_cad_pid(struct ctl_table *table, int write,
2707			   void __user *buffer, size_t *lenp, loff_t *ppos)
2708{
2709	struct pid *new_pid;
2710	pid_t tmp;
2711	int r;
2712
2713	tmp = pid_vnr(cad_pid);
2714
2715	r = __do_proc_dointvec(&tmp, table, write, buffer,
2716			       lenp, ppos, NULL, NULL);
2717	if (r || !write)
2718		return r;
2719
2720	new_pid = find_get_pid(tmp);
2721	if (!new_pid)
2722		return -ESRCH;
2723
2724	put_pid(xchg(&cad_pid, new_pid));
2725	return 0;
2726}
2727
2728/**
2729 * proc_do_large_bitmap - read/write from/to a large bitmap
2730 * @table: the sysctl table
2731 * @write: %TRUE if this is a write to the sysctl file
2732 * @buffer: the user buffer
2733 * @lenp: the size of the user buffer
2734 * @ppos: file position
2735 *
2736 * The bitmap is stored at table->data and the bitmap length (in bits)
2737 * in table->maxlen.
2738 *
2739 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2740 * large bitmaps may be represented in a compact manner. Writing into
2741 * the file will clear the bitmap then update it with the given input.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_do_large_bitmap(struct ctl_table *table, int write,
2746			 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748	int err = 0;
2749	bool first = 1;
2750	size_t left = *lenp;
2751	unsigned long bitmap_len = table->maxlen;
2752	unsigned long *bitmap = *(unsigned long **) table->data;
2753	unsigned long *tmp_bitmap = NULL;
2754	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2755
2756	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2757		*lenp = 0;
2758		return 0;
2759	}
2760
2761	if (write) {
2762		char *kbuf, *p;
 
2763
2764		if (left > PAGE_SIZE - 1)
2765			left = PAGE_SIZE - 1;
2766
2767		p = kbuf = memdup_user_nul(buffer, left);
2768		if (IS_ERR(kbuf))
2769			return PTR_ERR(kbuf);
 
 
 
 
 
 
2770
2771		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2772				     GFP_KERNEL);
2773		if (!tmp_bitmap) {
2774			kfree(kbuf);
2775			return -ENOMEM;
2776		}
2777		proc_skip_char(&p, &left, '\n');
2778		while (!err && left) {
2779			unsigned long val_a, val_b;
2780			bool neg;
2781
2782			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2783					     sizeof(tr_a), &c);
2784			if (err)
2785				break;
2786			if (val_a >= bitmap_len || neg) {
2787				err = -EINVAL;
2788				break;
2789			}
2790
2791			val_b = val_a;
2792			if (left) {
2793				p++;
2794				left--;
2795			}
2796
2797			if (c == '-') {
2798				err = proc_get_long(&p, &left, &val_b,
2799						     &neg, tr_b, sizeof(tr_b),
2800						     &c);
2801				if (err)
2802					break;
2803				if (val_b >= bitmap_len || neg ||
2804				    val_a > val_b) {
2805					err = -EINVAL;
2806					break;
2807				}
2808				if (left) {
2809					p++;
2810					left--;
2811				}
2812			}
2813
2814			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
 
 
2815			first = 0;
2816			proc_skip_char(&p, &left, '\n');
2817		}
2818		kfree(kbuf);
2819	} else {
2820		unsigned long bit_a, bit_b = 0;
2821
2822		while (left) {
2823			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2824			if (bit_a >= bitmap_len)
2825				break;
2826			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2827						   bit_a + 1) - 1;
2828
2829			if (!first) {
2830				err = proc_put_char(&buffer, &left, ',');
2831				if (err)
2832					break;
2833			}
2834			err = proc_put_long(&buffer, &left, bit_a, false);
2835			if (err)
2836				break;
2837			if (bit_a != bit_b) {
2838				err = proc_put_char(&buffer, &left, '-');
2839				if (err)
2840					break;
2841				err = proc_put_long(&buffer, &left, bit_b, false);
2842				if (err)
2843					break;
2844			}
2845
2846			first = 0; bit_b++;
2847		}
2848		if (!err)
2849			err = proc_put_char(&buffer, &left, '\n');
2850	}
2851
2852	if (!err) {
2853		if (write) {
2854			if (*ppos)
2855				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2856			else
2857				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
 
2858		}
2859		kfree(tmp_bitmap);
2860		*lenp -= left;
2861		*ppos += *lenp;
2862		return 0;
2863	} else {
2864		kfree(tmp_bitmap);
2865		return err;
2866	}
2867}
2868
2869#else /* CONFIG_PROC_SYSCTL */
2870
2871int proc_dostring(struct ctl_table *table, int write,
2872		  void __user *buffer, size_t *lenp, loff_t *ppos)
2873{
2874	return -ENOSYS;
2875}
2876
2877int proc_dointvec(struct ctl_table *table, int write,
2878		  void __user *buffer, size_t *lenp, loff_t *ppos)
2879{
2880	return -ENOSYS;
2881}
2882
2883int proc_douintvec(struct ctl_table *table, int write,
2884		  void __user *buffer, size_t *lenp, loff_t *ppos)
2885{
2886	return -ENOSYS;
2887}
2888
2889int proc_dointvec_minmax(struct ctl_table *table, int write,
2890		    void __user *buffer, size_t *lenp, loff_t *ppos)
2891{
2892	return -ENOSYS;
2893}
2894
2895int proc_dointvec_jiffies(struct ctl_table *table, int write,
2896		    void __user *buffer, size_t *lenp, loff_t *ppos)
2897{
2898	return -ENOSYS;
2899}
2900
2901int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2902		    void __user *buffer, size_t *lenp, loff_t *ppos)
2903{
2904	return -ENOSYS;
2905}
2906
2907int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2908			     void __user *buffer, size_t *lenp, loff_t *ppos)
2909{
2910	return -ENOSYS;
2911}
2912
2913int proc_doulongvec_minmax(struct ctl_table *table, int write,
2914		    void __user *buffer, size_t *lenp, loff_t *ppos)
2915{
2916	return -ENOSYS;
2917}
2918
2919int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2920				      void __user *buffer,
2921				      size_t *lenp, loff_t *ppos)
2922{
2923    return -ENOSYS;
2924}
2925
2926
2927#endif /* CONFIG_PROC_SYSCTL */
2928
2929/*
2930 * No sense putting this after each symbol definition, twice,
2931 * exception granted :-)
2932 */
2933EXPORT_SYMBOL(proc_dointvec);
2934EXPORT_SYMBOL(proc_douintvec);
2935EXPORT_SYMBOL(proc_dointvec_jiffies);
2936EXPORT_SYMBOL(proc_dointvec_minmax);
2937EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2938EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2939EXPORT_SYMBOL(proc_dostring);
2940EXPORT_SYMBOL(proc_doulongvec_minmax);
2941EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
v3.1
   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/signal.h>
  27#include <linux/printk.h>
  28#include <linux/proc_fs.h>
  29#include <linux/security.h>
  30#include <linux/ctype.h>
  31#include <linux/kmemcheck.h>
 
  32#include <linux/fs.h>
  33#include <linux/init.h>
  34#include <linux/kernel.h>
  35#include <linux/kobject.h>
  36#include <linux/net.h>
  37#include <linux/sysrq.h>
  38#include <linux/highuid.h>
  39#include <linux/writeback.h>
  40#include <linux/ratelimit.h>
  41#include <linux/compaction.h>
  42#include <linux/hugetlb.h>
  43#include <linux/initrd.h>
  44#include <linux/key.h>
  45#include <linux/times.h>
  46#include <linux/limits.h>
  47#include <linux/dcache.h>
  48#include <linux/dnotify.h>
  49#include <linux/syscalls.h>
  50#include <linux/vmstat.h>
  51#include <linux/nfs_fs.h>
  52#include <linux/acpi.h>
  53#include <linux/reboot.h>
  54#include <linux/ftrace.h>
  55#include <linux/perf_event.h>
  56#include <linux/kprobes.h>
  57#include <linux/pipe_fs_i.h>
  58#include <linux/oom.h>
  59#include <linux/kmod.h>
 
 
 
 
 
 
  60
  61#include <asm/uaccess.h>
  62#include <asm/processor.h>
  63
  64#ifdef CONFIG_X86
  65#include <asm/nmi.h>
  66#include <asm/stacktrace.h>
  67#include <asm/io.h>
  68#endif
 
 
 
  69#ifdef CONFIG_BSD_PROCESS_ACCT
  70#include <linux/acct.h>
  71#endif
  72#ifdef CONFIG_RT_MUTEXES
  73#include <linux/rtmutex.h>
  74#endif
  75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
  76#include <linux/lockdep.h>
  77#endif
  78#ifdef CONFIG_CHR_DEV_SG
  79#include <scsi/sg.h>
  80#endif
  81
  82#ifdef CONFIG_LOCKUP_DETECTOR
  83#include <linux/nmi.h>
  84#endif
  85
  86
  87#if defined(CONFIG_SYSCTL)
  88
  89/* External variables not in a header file. */
  90extern int sysctl_overcommit_memory;
  91extern int sysctl_overcommit_ratio;
  92extern int max_threads;
  93extern int core_uses_pid;
  94extern int suid_dumpable;
  95extern char core_pattern[];
  96extern unsigned int core_pipe_limit;
 
  97extern int pid_max;
  98extern int min_free_kbytes;
  99extern int pid_max_min, pid_max_max;
 100extern int sysctl_drop_caches;
 101extern int percpu_pagelist_fraction;
 102extern int compat_log;
 103extern int latencytop_enabled;
 104extern int sysctl_nr_open_min, sysctl_nr_open_max;
 105#ifndef CONFIG_MMU
 106extern int sysctl_nr_trim_pages;
 107#endif
 108#ifdef CONFIG_BLOCK
 109extern int blk_iopoll_enabled;
 110#endif
 111
 112/* Constants used for minimum and  maximum */
 113#ifdef CONFIG_LOCKUP_DETECTOR
 114static int sixty = 60;
 115static int neg_one = -1;
 116#endif
 117
 
 
 118static int zero;
 119static int __maybe_unused one = 1;
 120static int __maybe_unused two = 2;
 121static int __maybe_unused three = 3;
 122static unsigned long one_ul = 1;
 123static int one_hundred = 100;
 
 124#ifdef CONFIG_PRINTK
 125static int ten_thousand = 10000;
 126#endif
 
 
 
 127
 128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
 130
 131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
 132static int maxolduid = 65535;
 133static int minolduid;
 134static int min_percpu_pagelist_fract = 8;
 135
 136static int ngroups_max = NGROUPS_MAX;
 
 
 
 
 
 
 137
 138#ifdef CONFIG_INOTIFY_USER
 139#include <linux/inotify.h>
 140#endif
 141#ifdef CONFIG_SPARC
 142#include <asm/system.h>
 143#endif
 144
 145#ifdef CONFIG_SPARC64
 146extern int sysctl_tsb_ratio;
 147#endif
 148
 149#ifdef __hppa__
 150extern int pwrsw_enabled;
 151extern int unaligned_enabled;
 152#endif
 153
 154#ifdef CONFIG_S390
 155#ifdef CONFIG_MATHEMU
 156extern int sysctl_ieee_emulation_warnings;
 157#endif
 158extern int sysctl_userprocess_debug;
 159extern int spin_retry;
 160#endif
 161
 162#ifdef CONFIG_IA64
 163extern int no_unaligned_warning;
 164extern int unaligned_dump_stack;
 165#endif
 166
 167#ifdef CONFIG_PROC_SYSCTL
 
 
 
 
 
 
 
 168static int proc_do_cad_pid(struct ctl_table *table, int write,
 169		  void __user *buffer, size_t *lenp, loff_t *ppos);
 170static int proc_taint(struct ctl_table *table, int write,
 171			       void __user *buffer, size_t *lenp, loff_t *ppos);
 172#endif
 173
 174#ifdef CONFIG_PRINTK
 175static int proc_dmesg_restrict(struct ctl_table *table, int write,
 176				void __user *buffer, size_t *lenp, loff_t *ppos);
 177#endif
 178
 
 
 
 
 
 
 
 179#ifdef CONFIG_MAGIC_SYSRQ
 180/* Note: sysrq code uses it's own private copy */
 181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
 182
 183static int sysrq_sysctl_handler(ctl_table *table, int write,
 184				void __user *buffer, size_t *lenp,
 185				loff_t *ppos)
 186{
 187	int error;
 188
 189	error = proc_dointvec(table, write, buffer, lenp, ppos);
 190	if (error)
 191		return error;
 192
 193	if (write)
 194		sysrq_toggle_support(__sysrq_enabled);
 195
 196	return 0;
 197}
 198
 199#endif
 200
 201static struct ctl_table root_table[];
 202static struct ctl_table_root sysctl_table_root;
 203static struct ctl_table_header root_table_header = {
 204	{{.count = 1,
 205	.ctl_table = root_table,
 206	.ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
 207	.root = &sysctl_table_root,
 208	.set = &sysctl_table_root.default_set,
 209};
 210static struct ctl_table_root sysctl_table_root = {
 211	.root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
 212	.default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
 213};
 214
 215static struct ctl_table kern_table[];
 216static struct ctl_table vm_table[];
 217static struct ctl_table fs_table[];
 218static struct ctl_table debug_table[];
 219static struct ctl_table dev_table[];
 220extern struct ctl_table random_table[];
 221#ifdef CONFIG_EPOLL
 222extern struct ctl_table epoll_table[];
 223#endif
 224
 225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
 226int sysctl_legacy_va_layout;
 227#endif
 228
 229/* The default sysctl tables: */
 230
 231static struct ctl_table root_table[] = {
 232	{
 233		.procname	= "kernel",
 234		.mode		= 0555,
 235		.child		= kern_table,
 236	},
 237	{
 238		.procname	= "vm",
 239		.mode		= 0555,
 240		.child		= vm_table,
 241	},
 242	{
 243		.procname	= "fs",
 244		.mode		= 0555,
 245		.child		= fs_table,
 246	},
 247	{
 248		.procname	= "debug",
 249		.mode		= 0555,
 250		.child		= debug_table,
 251	},
 252	{
 253		.procname	= "dev",
 254		.mode		= 0555,
 255		.child		= dev_table,
 256	},
 257	{ }
 258};
 259
 260#ifdef CONFIG_SCHED_DEBUG
 261static int min_sched_granularity_ns = 100000;		/* 100 usecs */
 262static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
 263static int min_wakeup_granularity_ns;			/* 0 usecs */
 264static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
 
 265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
 267#endif
 
 268
 269#ifdef CONFIG_COMPACTION
 270static int min_extfrag_threshold;
 271static int max_extfrag_threshold = 1000;
 272#endif
 273
 274static struct ctl_table kern_table[] = {
 275	{
 276		.procname	= "sched_child_runs_first",
 277		.data		= &sysctl_sched_child_runs_first,
 278		.maxlen		= sizeof(unsigned int),
 279		.mode		= 0644,
 280		.proc_handler	= proc_dointvec,
 281	},
 282#ifdef CONFIG_SCHED_DEBUG
 283	{
 284		.procname	= "sched_min_granularity_ns",
 285		.data		= &sysctl_sched_min_granularity,
 286		.maxlen		= sizeof(unsigned int),
 287		.mode		= 0644,
 288		.proc_handler	= sched_proc_update_handler,
 289		.extra1		= &min_sched_granularity_ns,
 290		.extra2		= &max_sched_granularity_ns,
 291	},
 292	{
 293		.procname	= "sched_latency_ns",
 294		.data		= &sysctl_sched_latency,
 295		.maxlen		= sizeof(unsigned int),
 296		.mode		= 0644,
 297		.proc_handler	= sched_proc_update_handler,
 298		.extra1		= &min_sched_granularity_ns,
 299		.extra2		= &max_sched_granularity_ns,
 300	},
 301	{
 302		.procname	= "sched_wakeup_granularity_ns",
 303		.data		= &sysctl_sched_wakeup_granularity,
 304		.maxlen		= sizeof(unsigned int),
 305		.mode		= 0644,
 306		.proc_handler	= sched_proc_update_handler,
 307		.extra1		= &min_wakeup_granularity_ns,
 308		.extra2		= &max_wakeup_granularity_ns,
 309	},
 
 310	{
 311		.procname	= "sched_tunable_scaling",
 312		.data		= &sysctl_sched_tunable_scaling,
 313		.maxlen		= sizeof(enum sched_tunable_scaling),
 314		.mode		= 0644,
 315		.proc_handler	= sched_proc_update_handler,
 316		.extra1		= &min_sched_tunable_scaling,
 317		.extra2		= &max_sched_tunable_scaling,
 318	},
 319	{
 320		.procname	= "sched_migration_cost",
 321		.data		= &sysctl_sched_migration_cost,
 322		.maxlen		= sizeof(unsigned int),
 323		.mode		= 0644,
 324		.proc_handler	= proc_dointvec,
 325	},
 326	{
 327		.procname	= "sched_nr_migrate",
 328		.data		= &sysctl_sched_nr_migrate,
 329		.maxlen		= sizeof(unsigned int),
 330		.mode		= 0644,
 331		.proc_handler	= proc_dointvec,
 332	},
 333	{
 334		.procname	= "sched_time_avg",
 335		.data		= &sysctl_sched_time_avg,
 336		.maxlen		= sizeof(unsigned int),
 337		.mode		= 0644,
 338		.proc_handler	= proc_dointvec,
 339	},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 340	{
 341		.procname	= "sched_shares_window",
 342		.data		= &sysctl_sched_shares_window,
 343		.maxlen		= sizeof(unsigned int),
 344		.mode		= 0644,
 345		.proc_handler	= proc_dointvec,
 346	},
 347	{
 348		.procname	= "timer_migration",
 349		.data		= &sysctl_timer_migration,
 350		.maxlen		= sizeof(unsigned int),
 351		.mode		= 0644,
 352		.proc_handler	= proc_dointvec_minmax,
 
 
 
 
 
 
 
 
 353		.extra1		= &zero,
 354		.extra2		= &one,
 355	},
 356#endif
 
 357	{
 358		.procname	= "sched_rt_period_us",
 359		.data		= &sysctl_sched_rt_period,
 360		.maxlen		= sizeof(unsigned int),
 361		.mode		= 0644,
 362		.proc_handler	= sched_rt_handler,
 363	},
 364	{
 365		.procname	= "sched_rt_runtime_us",
 366		.data		= &sysctl_sched_rt_runtime,
 367		.maxlen		= sizeof(int),
 368		.mode		= 0644,
 369		.proc_handler	= sched_rt_handler,
 370	},
 
 
 
 
 
 
 
 371#ifdef CONFIG_SCHED_AUTOGROUP
 372	{
 373		.procname	= "sched_autogroup_enabled",
 374		.data		= &sysctl_sched_autogroup_enabled,
 375		.maxlen		= sizeof(unsigned int),
 376		.mode		= 0644,
 377		.proc_handler	= proc_dointvec_minmax,
 378		.extra1		= &zero,
 379		.extra2		= &one,
 380	},
 381#endif
 
 
 
 
 
 
 
 
 
 
 382#ifdef CONFIG_PROVE_LOCKING
 383	{
 384		.procname	= "prove_locking",
 385		.data		= &prove_locking,
 386		.maxlen		= sizeof(int),
 387		.mode		= 0644,
 388		.proc_handler	= proc_dointvec,
 389	},
 390#endif
 391#ifdef CONFIG_LOCK_STAT
 392	{
 393		.procname	= "lock_stat",
 394		.data		= &lock_stat,
 395		.maxlen		= sizeof(int),
 396		.mode		= 0644,
 397		.proc_handler	= proc_dointvec,
 398	},
 399#endif
 400	{
 401		.procname	= "panic",
 402		.data		= &panic_timeout,
 403		.maxlen		= sizeof(int),
 404		.mode		= 0644,
 405		.proc_handler	= proc_dointvec,
 406	},
 
 407	{
 408		.procname	= "core_uses_pid",
 409		.data		= &core_uses_pid,
 410		.maxlen		= sizeof(int),
 411		.mode		= 0644,
 412		.proc_handler	= proc_dointvec,
 413	},
 414	{
 415		.procname	= "core_pattern",
 416		.data		= core_pattern,
 417		.maxlen		= CORENAME_MAX_SIZE,
 418		.mode		= 0644,
 419		.proc_handler	= proc_dostring,
 420	},
 421	{
 422		.procname	= "core_pipe_limit",
 423		.data		= &core_pipe_limit,
 424		.maxlen		= sizeof(unsigned int),
 425		.mode		= 0644,
 426		.proc_handler	= proc_dointvec,
 427	},
 
 428#ifdef CONFIG_PROC_SYSCTL
 429	{
 430		.procname	= "tainted",
 431		.maxlen 	= sizeof(long),
 432		.mode		= 0644,
 433		.proc_handler	= proc_taint,
 434	},
 
 
 
 
 
 
 
 
 
 435#endif
 436#ifdef CONFIG_LATENCYTOP
 437	{
 438		.procname	= "latencytop",
 439		.data		= &latencytop_enabled,
 440		.maxlen		= sizeof(int),
 441		.mode		= 0644,
 442		.proc_handler	= proc_dointvec,
 443	},
 444#endif
 445#ifdef CONFIG_BLK_DEV_INITRD
 446	{
 447		.procname	= "real-root-dev",
 448		.data		= &real_root_dev,
 449		.maxlen		= sizeof(int),
 450		.mode		= 0644,
 451		.proc_handler	= proc_dointvec,
 452	},
 453#endif
 454	{
 455		.procname	= "print-fatal-signals",
 456		.data		= &print_fatal_signals,
 457		.maxlen		= sizeof(int),
 458		.mode		= 0644,
 459		.proc_handler	= proc_dointvec,
 460	},
 461#ifdef CONFIG_SPARC
 462	{
 463		.procname	= "reboot-cmd",
 464		.data		= reboot_command,
 465		.maxlen		= 256,
 466		.mode		= 0644,
 467		.proc_handler	= proc_dostring,
 468	},
 469	{
 470		.procname	= "stop-a",
 471		.data		= &stop_a_enabled,
 472		.maxlen		= sizeof (int),
 473		.mode		= 0644,
 474		.proc_handler	= proc_dointvec,
 475	},
 476	{
 477		.procname	= "scons-poweroff",
 478		.data		= &scons_pwroff,
 479		.maxlen		= sizeof (int),
 480		.mode		= 0644,
 481		.proc_handler	= proc_dointvec,
 482	},
 483#endif
 484#ifdef CONFIG_SPARC64
 485	{
 486		.procname	= "tsb-ratio",
 487		.data		= &sysctl_tsb_ratio,
 488		.maxlen		= sizeof (int),
 489		.mode		= 0644,
 490		.proc_handler	= proc_dointvec,
 491	},
 492#endif
 493#ifdef __hppa__
 494	{
 495		.procname	= "soft-power",
 496		.data		= &pwrsw_enabled,
 497		.maxlen		= sizeof (int),
 498	 	.mode		= 0644,
 499		.proc_handler	= proc_dointvec,
 500	},
 
 
 501	{
 502		.procname	= "unaligned-trap",
 503		.data		= &unaligned_enabled,
 504		.maxlen		= sizeof (int),
 505		.mode		= 0644,
 506		.proc_handler	= proc_dointvec,
 507	},
 508#endif
 509	{
 510		.procname	= "ctrl-alt-del",
 511		.data		= &C_A_D,
 512		.maxlen		= sizeof(int),
 513		.mode		= 0644,
 514		.proc_handler	= proc_dointvec,
 515	},
 516#ifdef CONFIG_FUNCTION_TRACER
 517	{
 518		.procname	= "ftrace_enabled",
 519		.data		= &ftrace_enabled,
 520		.maxlen		= sizeof(int),
 521		.mode		= 0644,
 522		.proc_handler	= ftrace_enable_sysctl,
 523	},
 524#endif
 525#ifdef CONFIG_STACK_TRACER
 526	{
 527		.procname	= "stack_tracer_enabled",
 528		.data		= &stack_tracer_enabled,
 529		.maxlen		= sizeof(int),
 530		.mode		= 0644,
 531		.proc_handler	= stack_trace_sysctl,
 532	},
 533#endif
 534#ifdef CONFIG_TRACING
 535	{
 536		.procname	= "ftrace_dump_on_oops",
 537		.data		= &ftrace_dump_on_oops,
 538		.maxlen		= sizeof(int),
 539		.mode		= 0644,
 540		.proc_handler	= proc_dointvec,
 541	},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 542#endif
 543#ifdef CONFIG_MODULES
 544	{
 545		.procname	= "modprobe",
 546		.data		= &modprobe_path,
 547		.maxlen		= KMOD_PATH_LEN,
 548		.mode		= 0644,
 549		.proc_handler	= proc_dostring,
 550	},
 551	{
 552		.procname	= "modules_disabled",
 553		.data		= &modules_disabled,
 554		.maxlen		= sizeof(int),
 555		.mode		= 0644,
 556		/* only handle a transition from default "0" to "1" */
 557		.proc_handler	= proc_dointvec_minmax,
 558		.extra1		= &one,
 559		.extra2		= &one,
 560	},
 561#endif
 562#ifdef CONFIG_HOTPLUG
 563	{
 564		.procname	= "hotplug",
 565		.data		= &uevent_helper,
 566		.maxlen		= UEVENT_HELPER_PATH_LEN,
 567		.mode		= 0644,
 568		.proc_handler	= proc_dostring,
 569	},
 570#endif
 571#ifdef CONFIG_CHR_DEV_SG
 572	{
 573		.procname	= "sg-big-buff",
 574		.data		= &sg_big_buff,
 575		.maxlen		= sizeof (int),
 576		.mode		= 0444,
 577		.proc_handler	= proc_dointvec,
 578	},
 579#endif
 580#ifdef CONFIG_BSD_PROCESS_ACCT
 581	{
 582		.procname	= "acct",
 583		.data		= &acct_parm,
 584		.maxlen		= 3*sizeof(int),
 585		.mode		= 0644,
 586		.proc_handler	= proc_dointvec,
 587	},
 588#endif
 589#ifdef CONFIG_MAGIC_SYSRQ
 590	{
 591		.procname	= "sysrq",
 592		.data		= &__sysrq_enabled,
 593		.maxlen		= sizeof (int),
 594		.mode		= 0644,
 595		.proc_handler	= sysrq_sysctl_handler,
 596	},
 597#endif
 598#ifdef CONFIG_PROC_SYSCTL
 599	{
 600		.procname	= "cad_pid",
 601		.data		= NULL,
 602		.maxlen		= sizeof (int),
 603		.mode		= 0600,
 604		.proc_handler	= proc_do_cad_pid,
 605	},
 606#endif
 607	{
 608		.procname	= "threads-max",
 609		.data		= &max_threads,
 610		.maxlen		= sizeof(int),
 611		.mode		= 0644,
 612		.proc_handler	= proc_dointvec,
 613	},
 614	{
 615		.procname	= "random",
 616		.mode		= 0555,
 617		.child		= random_table,
 618	},
 619	{
 620		.procname	= "usermodehelper",
 621		.mode		= 0555,
 622		.child		= usermodehelper_table,
 623	},
 624	{
 625		.procname	= "overflowuid",
 626		.data		= &overflowuid,
 627		.maxlen		= sizeof(int),
 628		.mode		= 0644,
 629		.proc_handler	= proc_dointvec_minmax,
 630		.extra1		= &minolduid,
 631		.extra2		= &maxolduid,
 632	},
 633	{
 634		.procname	= "overflowgid",
 635		.data		= &overflowgid,
 636		.maxlen		= sizeof(int),
 637		.mode		= 0644,
 638		.proc_handler	= proc_dointvec_minmax,
 639		.extra1		= &minolduid,
 640		.extra2		= &maxolduid,
 641	},
 642#ifdef CONFIG_S390
 643#ifdef CONFIG_MATHEMU
 644	{
 645		.procname	= "ieee_emulation_warnings",
 646		.data		= &sysctl_ieee_emulation_warnings,
 647		.maxlen		= sizeof(int),
 648		.mode		= 0644,
 649		.proc_handler	= proc_dointvec,
 650	},
 651#endif
 652	{
 653		.procname	= "userprocess_debug",
 654		.data		= &show_unhandled_signals,
 655		.maxlen		= sizeof(int),
 656		.mode		= 0644,
 657		.proc_handler	= proc_dointvec,
 658	},
 659#endif
 660	{
 661		.procname	= "pid_max",
 662		.data		= &pid_max,
 663		.maxlen		= sizeof (int),
 664		.mode		= 0644,
 665		.proc_handler	= proc_dointvec_minmax,
 666		.extra1		= &pid_max_min,
 667		.extra2		= &pid_max_max,
 668	},
 669	{
 670		.procname	= "panic_on_oops",
 671		.data		= &panic_on_oops,
 672		.maxlen		= sizeof(int),
 673		.mode		= 0644,
 674		.proc_handler	= proc_dointvec,
 675	},
 676#if defined CONFIG_PRINTK
 677	{
 678		.procname	= "printk",
 679		.data		= &console_loglevel,
 680		.maxlen		= 4*sizeof(int),
 681		.mode		= 0644,
 682		.proc_handler	= proc_dointvec,
 683	},
 684	{
 685		.procname	= "printk_ratelimit",
 686		.data		= &printk_ratelimit_state.interval,
 687		.maxlen		= sizeof(int),
 688		.mode		= 0644,
 689		.proc_handler	= proc_dointvec_jiffies,
 690	},
 691	{
 692		.procname	= "printk_ratelimit_burst",
 693		.data		= &printk_ratelimit_state.burst,
 694		.maxlen		= sizeof(int),
 695		.mode		= 0644,
 696		.proc_handler	= proc_dointvec,
 697	},
 698	{
 699		.procname	= "printk_delay",
 700		.data		= &printk_delay_msec,
 701		.maxlen		= sizeof(int),
 702		.mode		= 0644,
 703		.proc_handler	= proc_dointvec_minmax,
 704		.extra1		= &zero,
 705		.extra2		= &ten_thousand,
 706	},
 707	{
 
 
 
 
 
 
 
 708		.procname	= "dmesg_restrict",
 709		.data		= &dmesg_restrict,
 710		.maxlen		= sizeof(int),
 711		.mode		= 0644,
 712		.proc_handler	= proc_dointvec_minmax,
 713		.extra1		= &zero,
 714		.extra2		= &one,
 715	},
 716	{
 717		.procname	= "kptr_restrict",
 718		.data		= &kptr_restrict,
 719		.maxlen		= sizeof(int),
 720		.mode		= 0644,
 721		.proc_handler	= proc_dmesg_restrict,
 722		.extra1		= &zero,
 723		.extra2		= &two,
 724	},
 725#endif
 726	{
 727		.procname	= "ngroups_max",
 728		.data		= &ngroups_max,
 729		.maxlen		= sizeof (int),
 730		.mode		= 0444,
 731		.proc_handler	= proc_dointvec,
 732	},
 
 
 
 
 
 
 
 733#if defined(CONFIG_LOCKUP_DETECTOR)
 734	{
 735		.procname       = "watchdog",
 736		.data           = &watchdog_enabled,
 737		.maxlen         = sizeof (int),
 738		.mode           = 0644,
 739		.proc_handler   = proc_dowatchdog,
 740		.extra1		= &zero,
 741		.extra2		= &one,
 742	},
 743	{
 744		.procname	= "watchdog_thresh",
 745		.data		= &watchdog_thresh,
 746		.maxlen		= sizeof(int),
 747		.mode		= 0644,
 748		.proc_handler	= proc_dowatchdog,
 749		.extra1		= &neg_one,
 750		.extra2		= &sixty,
 751	},
 752	{
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 753		.procname	= "softlockup_panic",
 754		.data		= &softlockup_panic,
 755		.maxlen		= sizeof(int),
 756		.mode		= 0644,
 757		.proc_handler	= proc_dointvec_minmax,
 758		.extra1		= &zero,
 759		.extra2		= &one,
 760	},
 
 
 
 
 
 
 
 
 
 
 
 
 761	{
 762		.procname       = "nmi_watchdog",
 763		.data           = &watchdog_enabled,
 764		.maxlen         = sizeof (int),
 765		.mode           = 0644,
 766		.proc_handler   = proc_dowatchdog,
 
 
 
 
 
 
 
 
 
 767		.extra1		= &zero,
 768		.extra2		= &one,
 769	},
 
 770#endif
 771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
 772	{
 773		.procname       = "unknown_nmi_panic",
 774		.data           = &unknown_nmi_panic,
 775		.maxlen         = sizeof (int),
 776		.mode           = 0644,
 777		.proc_handler   = proc_dointvec,
 778	},
 779#endif
 780#if defined(CONFIG_X86)
 781	{
 782		.procname	= "panic_on_unrecovered_nmi",
 783		.data		= &panic_on_unrecovered_nmi,
 784		.maxlen		= sizeof(int),
 785		.mode		= 0644,
 786		.proc_handler	= proc_dointvec,
 787	},
 788	{
 789		.procname	= "panic_on_io_nmi",
 790		.data		= &panic_on_io_nmi,
 791		.maxlen		= sizeof(int),
 792		.mode		= 0644,
 793		.proc_handler	= proc_dointvec,
 794	},
 
 
 
 
 
 
 
 
 
 795	{
 796		.procname	= "bootloader_type",
 797		.data		= &bootloader_type,
 798		.maxlen		= sizeof (int),
 799		.mode		= 0444,
 800		.proc_handler	= proc_dointvec,
 801	},
 802	{
 803		.procname	= "bootloader_version",
 804		.data		= &bootloader_version,
 805		.maxlen		= sizeof (int),
 806		.mode		= 0444,
 807		.proc_handler	= proc_dointvec,
 808	},
 809	{
 810		.procname	= "kstack_depth_to_print",
 811		.data		= &kstack_depth_to_print,
 812		.maxlen		= sizeof(int),
 813		.mode		= 0644,
 814		.proc_handler	= proc_dointvec,
 815	},
 816	{
 817		.procname	= "io_delay_type",
 818		.data		= &io_delay_type,
 819		.maxlen		= sizeof(int),
 820		.mode		= 0644,
 821		.proc_handler	= proc_dointvec,
 822	},
 823#endif
 824#if defined(CONFIG_MMU)
 825	{
 826		.procname	= "randomize_va_space",
 827		.data		= &randomize_va_space,
 828		.maxlen		= sizeof(int),
 829		.mode		= 0644,
 830		.proc_handler	= proc_dointvec,
 831	},
 832#endif
 833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
 834	{
 835		.procname	= "spin_retry",
 836		.data		= &spin_retry,
 837		.maxlen		= sizeof (int),
 838		.mode		= 0644,
 839		.proc_handler	= proc_dointvec,
 840	},
 841#endif
 842#if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
 843	{
 844		.procname	= "acpi_video_flags",
 845		.data		= &acpi_realmode_flags,
 846		.maxlen		= sizeof (unsigned long),
 847		.mode		= 0644,
 848		.proc_handler	= proc_doulongvec_minmax,
 849	},
 850#endif
 851#ifdef CONFIG_IA64
 852	{
 853		.procname	= "ignore-unaligned-usertrap",
 854		.data		= &no_unaligned_warning,
 855		.maxlen		= sizeof (int),
 856	 	.mode		= 0644,
 857		.proc_handler	= proc_dointvec,
 858	},
 
 
 859	{
 860		.procname	= "unaligned-dump-stack",
 861		.data		= &unaligned_dump_stack,
 862		.maxlen		= sizeof (int),
 863		.mode		= 0644,
 864		.proc_handler	= proc_dointvec,
 865	},
 866#endif
 867#ifdef CONFIG_DETECT_HUNG_TASK
 868	{
 869		.procname	= "hung_task_panic",
 870		.data		= &sysctl_hung_task_panic,
 871		.maxlen		= sizeof(int),
 872		.mode		= 0644,
 873		.proc_handler	= proc_dointvec_minmax,
 874		.extra1		= &zero,
 875		.extra2		= &one,
 876	},
 877	{
 878		.procname	= "hung_task_check_count",
 879		.data		= &sysctl_hung_task_check_count,
 880		.maxlen		= sizeof(unsigned long),
 881		.mode		= 0644,
 882		.proc_handler	= proc_doulongvec_minmax,
 
 883	},
 884	{
 885		.procname	= "hung_task_timeout_secs",
 886		.data		= &sysctl_hung_task_timeout_secs,
 887		.maxlen		= sizeof(unsigned long),
 888		.mode		= 0644,
 889		.proc_handler	= proc_dohung_task_timeout_secs,
 
 890	},
 891	{
 892		.procname	= "hung_task_warnings",
 893		.data		= &sysctl_hung_task_warnings,
 894		.maxlen		= sizeof(unsigned long),
 895		.mode		= 0644,
 896		.proc_handler	= proc_doulongvec_minmax,
 897	},
 898#endif
 899#ifdef CONFIG_COMPAT
 900	{
 901		.procname	= "compat-log",
 902		.data		= &compat_log,
 903		.maxlen		= sizeof (int),
 904	 	.mode		= 0644,
 905		.proc_handler	= proc_dointvec,
 906	},
 907#endif
 908#ifdef CONFIG_RT_MUTEXES
 909	{
 910		.procname	= "max_lock_depth",
 911		.data		= &max_lock_depth,
 912		.maxlen		= sizeof(int),
 913		.mode		= 0644,
 914		.proc_handler	= proc_dointvec,
 915	},
 916#endif
 917	{
 918		.procname	= "poweroff_cmd",
 919		.data		= &poweroff_cmd,
 920		.maxlen		= POWEROFF_CMD_PATH_LEN,
 921		.mode		= 0644,
 922		.proc_handler	= proc_dostring,
 923	},
 924#ifdef CONFIG_KEYS
 925	{
 926		.procname	= "keys",
 927		.mode		= 0555,
 928		.child		= key_sysctls,
 929	},
 930#endif
 931#ifdef CONFIG_RCU_TORTURE_TEST
 932	{
 933		.procname       = "rcutorture_runnable",
 934		.data           = &rcutorture_runnable,
 935		.maxlen         = sizeof(int),
 936		.mode           = 0644,
 937		.proc_handler	= proc_dointvec,
 938	},
 939#endif
 940#ifdef CONFIG_PERF_EVENTS
 941	/*
 942	 * User-space scripts rely on the existence of this file
 943	 * as a feature check for perf_events being enabled.
 944	 *
 945	 * So it's an ABI, do not remove!
 946	 */
 947	{
 948		.procname	= "perf_event_paranoid",
 949		.data		= &sysctl_perf_event_paranoid,
 950		.maxlen		= sizeof(sysctl_perf_event_paranoid),
 951		.mode		= 0644,
 952		.proc_handler	= proc_dointvec,
 953	},
 954	{
 955		.procname	= "perf_event_mlock_kb",
 956		.data		= &sysctl_perf_event_mlock,
 957		.maxlen		= sizeof(sysctl_perf_event_mlock),
 958		.mode		= 0644,
 959		.proc_handler	= proc_dointvec,
 960	},
 961	{
 962		.procname	= "perf_event_max_sample_rate",
 963		.data		= &sysctl_perf_event_sample_rate,
 964		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
 965		.mode		= 0644,
 966		.proc_handler	= perf_proc_update_handler,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 967	},
 968#endif
 969#ifdef CONFIG_KMEMCHECK
 970	{
 971		.procname	= "kmemcheck",
 972		.data		= &kmemcheck_enabled,
 973		.maxlen		= sizeof(int),
 974		.mode		= 0644,
 975		.proc_handler	= proc_dointvec,
 976	},
 977#endif
 978#ifdef CONFIG_BLOCK
 979	{
 980		.procname	= "blk_iopoll",
 981		.data		= &blk_iopoll_enabled,
 982		.maxlen		= sizeof(int),
 983		.mode		= 0644,
 984		.proc_handler	= proc_dointvec,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 985	},
 986#endif
 987	{ }
 988};
 989
 990static struct ctl_table vm_table[] = {
 991	{
 992		.procname	= "overcommit_memory",
 993		.data		= &sysctl_overcommit_memory,
 994		.maxlen		= sizeof(sysctl_overcommit_memory),
 995		.mode		= 0644,
 996		.proc_handler	= proc_dointvec_minmax,
 997		.extra1		= &zero,
 998		.extra2		= &two,
 999	},
1000	{
1001		.procname	= "panic_on_oom",
1002		.data		= &sysctl_panic_on_oom,
1003		.maxlen		= sizeof(sysctl_panic_on_oom),
1004		.mode		= 0644,
1005		.proc_handler	= proc_dointvec_minmax,
1006		.extra1		= &zero,
1007		.extra2		= &two,
1008	},
1009	{
1010		.procname	= "oom_kill_allocating_task",
1011		.data		= &sysctl_oom_kill_allocating_task,
1012		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1013		.mode		= 0644,
1014		.proc_handler	= proc_dointvec,
1015	},
1016	{
1017		.procname	= "oom_dump_tasks",
1018		.data		= &sysctl_oom_dump_tasks,
1019		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1020		.mode		= 0644,
1021		.proc_handler	= proc_dointvec,
1022	},
1023	{
1024		.procname	= "overcommit_ratio",
1025		.data		= &sysctl_overcommit_ratio,
1026		.maxlen		= sizeof(sysctl_overcommit_ratio),
1027		.mode		= 0644,
1028		.proc_handler	= proc_dointvec,
 
 
 
 
 
 
 
1029	},
1030	{
1031		.procname	= "page-cluster", 
1032		.data		= &page_cluster,
1033		.maxlen		= sizeof(int),
1034		.mode		= 0644,
1035		.proc_handler	= proc_dointvec_minmax,
1036		.extra1		= &zero,
1037	},
1038	{
1039		.procname	= "dirty_background_ratio",
1040		.data		= &dirty_background_ratio,
1041		.maxlen		= sizeof(dirty_background_ratio),
1042		.mode		= 0644,
1043		.proc_handler	= dirty_background_ratio_handler,
1044		.extra1		= &zero,
1045		.extra2		= &one_hundred,
1046	},
1047	{
1048		.procname	= "dirty_background_bytes",
1049		.data		= &dirty_background_bytes,
1050		.maxlen		= sizeof(dirty_background_bytes),
1051		.mode		= 0644,
1052		.proc_handler	= dirty_background_bytes_handler,
1053		.extra1		= &one_ul,
1054	},
1055	{
1056		.procname	= "dirty_ratio",
1057		.data		= &vm_dirty_ratio,
1058		.maxlen		= sizeof(vm_dirty_ratio),
1059		.mode		= 0644,
1060		.proc_handler	= dirty_ratio_handler,
1061		.extra1		= &zero,
1062		.extra2		= &one_hundred,
1063	},
1064	{
1065		.procname	= "dirty_bytes",
1066		.data		= &vm_dirty_bytes,
1067		.maxlen		= sizeof(vm_dirty_bytes),
1068		.mode		= 0644,
1069		.proc_handler	= dirty_bytes_handler,
1070		.extra1		= &dirty_bytes_min,
1071	},
1072	{
1073		.procname	= "dirty_writeback_centisecs",
1074		.data		= &dirty_writeback_interval,
1075		.maxlen		= sizeof(dirty_writeback_interval),
1076		.mode		= 0644,
1077		.proc_handler	= dirty_writeback_centisecs_handler,
1078	},
1079	{
1080		.procname	= "dirty_expire_centisecs",
1081		.data		= &dirty_expire_interval,
1082		.maxlen		= sizeof(dirty_expire_interval),
1083		.mode		= 0644,
1084		.proc_handler	= proc_dointvec_minmax,
1085		.extra1		= &zero,
1086	},
1087	{
1088		.procname	= "nr_pdflush_threads",
1089		.data		= &nr_pdflush_threads,
1090		.maxlen		= sizeof nr_pdflush_threads,
1091		.mode		= 0444 /* read-only*/,
1092		.proc_handler	= proc_dointvec,
 
 
 
 
 
 
1093	},
1094	{
1095		.procname	= "swappiness",
1096		.data		= &vm_swappiness,
1097		.maxlen		= sizeof(vm_swappiness),
1098		.mode		= 0644,
1099		.proc_handler	= proc_dointvec_minmax,
1100		.extra1		= &zero,
1101		.extra2		= &one_hundred,
1102	},
1103#ifdef CONFIG_HUGETLB_PAGE
1104	{
1105		.procname	= "nr_hugepages",
1106		.data		= NULL,
1107		.maxlen		= sizeof(unsigned long),
1108		.mode		= 0644,
1109		.proc_handler	= hugetlb_sysctl_handler,
1110		.extra1		= (void *)&hugetlb_zero,
1111		.extra2		= (void *)&hugetlb_infinity,
1112	},
1113#ifdef CONFIG_NUMA
1114	{
1115		.procname       = "nr_hugepages_mempolicy",
1116		.data           = NULL,
1117		.maxlen         = sizeof(unsigned long),
1118		.mode           = 0644,
1119		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1120		.extra1		= (void *)&hugetlb_zero,
1121		.extra2		= (void *)&hugetlb_infinity,
1122	},
1123#endif
1124	 {
1125		.procname	= "hugetlb_shm_group",
1126		.data		= &sysctl_hugetlb_shm_group,
1127		.maxlen		= sizeof(gid_t),
1128		.mode		= 0644,
1129		.proc_handler	= proc_dointvec,
1130	 },
1131	 {
1132		.procname	= "hugepages_treat_as_movable",
1133		.data		= &hugepages_treat_as_movable,
1134		.maxlen		= sizeof(int),
1135		.mode		= 0644,
1136		.proc_handler	= hugetlb_treat_movable_handler,
1137	},
1138	{
1139		.procname	= "nr_overcommit_hugepages",
1140		.data		= NULL,
1141		.maxlen		= sizeof(unsigned long),
1142		.mode		= 0644,
1143		.proc_handler	= hugetlb_overcommit_handler,
1144		.extra1		= (void *)&hugetlb_zero,
1145		.extra2		= (void *)&hugetlb_infinity,
1146	},
1147#endif
1148	{
1149		.procname	= "lowmem_reserve_ratio",
1150		.data		= &sysctl_lowmem_reserve_ratio,
1151		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1152		.mode		= 0644,
1153		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1154	},
1155	{
1156		.procname	= "drop_caches",
1157		.data		= &sysctl_drop_caches,
1158		.maxlen		= sizeof(int),
1159		.mode		= 0644,
1160		.proc_handler	= drop_caches_sysctl_handler,
1161		.extra1		= &one,
1162		.extra2		= &three,
1163	},
1164#ifdef CONFIG_COMPACTION
1165	{
1166		.procname	= "compact_memory",
1167		.data		= &sysctl_compact_memory,
1168		.maxlen		= sizeof(int),
1169		.mode		= 0200,
1170		.proc_handler	= sysctl_compaction_handler,
1171	},
1172	{
1173		.procname	= "extfrag_threshold",
1174		.data		= &sysctl_extfrag_threshold,
1175		.maxlen		= sizeof(int),
1176		.mode		= 0644,
1177		.proc_handler	= sysctl_extfrag_handler,
1178		.extra1		= &min_extfrag_threshold,
1179		.extra2		= &max_extfrag_threshold,
1180	},
 
 
 
 
 
 
 
 
 
1181
1182#endif /* CONFIG_COMPACTION */
1183	{
1184		.procname	= "min_free_kbytes",
1185		.data		= &min_free_kbytes,
1186		.maxlen		= sizeof(min_free_kbytes),
1187		.mode		= 0644,
1188		.proc_handler	= min_free_kbytes_sysctl_handler,
1189		.extra1		= &zero,
1190	},
1191	{
 
 
 
 
 
 
 
 
 
1192		.procname	= "percpu_pagelist_fraction",
1193		.data		= &percpu_pagelist_fraction,
1194		.maxlen		= sizeof(percpu_pagelist_fraction),
1195		.mode		= 0644,
1196		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1197		.extra1		= &min_percpu_pagelist_fract,
1198	},
1199#ifdef CONFIG_MMU
1200	{
1201		.procname	= "max_map_count",
1202		.data		= &sysctl_max_map_count,
1203		.maxlen		= sizeof(sysctl_max_map_count),
1204		.mode		= 0644,
1205		.proc_handler	= proc_dointvec_minmax,
1206		.extra1		= &zero,
1207	},
1208#else
1209	{
1210		.procname	= "nr_trim_pages",
1211		.data		= &sysctl_nr_trim_pages,
1212		.maxlen		= sizeof(sysctl_nr_trim_pages),
1213		.mode		= 0644,
1214		.proc_handler	= proc_dointvec_minmax,
1215		.extra1		= &zero,
1216	},
1217#endif
1218	{
1219		.procname	= "laptop_mode",
1220		.data		= &laptop_mode,
1221		.maxlen		= sizeof(laptop_mode),
1222		.mode		= 0644,
1223		.proc_handler	= proc_dointvec_jiffies,
1224	},
1225	{
1226		.procname	= "block_dump",
1227		.data		= &block_dump,
1228		.maxlen		= sizeof(block_dump),
1229		.mode		= 0644,
1230		.proc_handler	= proc_dointvec,
1231		.extra1		= &zero,
1232	},
1233	{
1234		.procname	= "vfs_cache_pressure",
1235		.data		= &sysctl_vfs_cache_pressure,
1236		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1237		.mode		= 0644,
1238		.proc_handler	= proc_dointvec,
1239		.extra1		= &zero,
1240	},
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242	{
1243		.procname	= "legacy_va_layout",
1244		.data		= &sysctl_legacy_va_layout,
1245		.maxlen		= sizeof(sysctl_legacy_va_layout),
1246		.mode		= 0644,
1247		.proc_handler	= proc_dointvec,
1248		.extra1		= &zero,
1249	},
1250#endif
1251#ifdef CONFIG_NUMA
1252	{
1253		.procname	= "zone_reclaim_mode",
1254		.data		= &zone_reclaim_mode,
1255		.maxlen		= sizeof(zone_reclaim_mode),
1256		.mode		= 0644,
1257		.proc_handler	= proc_dointvec,
1258		.extra1		= &zero,
1259	},
1260	{
1261		.procname	= "min_unmapped_ratio",
1262		.data		= &sysctl_min_unmapped_ratio,
1263		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1264		.mode		= 0644,
1265		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1266		.extra1		= &zero,
1267		.extra2		= &one_hundred,
1268	},
1269	{
1270		.procname	= "min_slab_ratio",
1271		.data		= &sysctl_min_slab_ratio,
1272		.maxlen		= sizeof(sysctl_min_slab_ratio),
1273		.mode		= 0644,
1274		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1275		.extra1		= &zero,
1276		.extra2		= &one_hundred,
1277	},
1278#endif
1279#ifdef CONFIG_SMP
1280	{
1281		.procname	= "stat_interval",
1282		.data		= &sysctl_stat_interval,
1283		.maxlen		= sizeof(sysctl_stat_interval),
1284		.mode		= 0644,
1285		.proc_handler	= proc_dointvec_jiffies,
1286	},
 
 
 
 
 
 
 
1287#endif
1288#ifdef CONFIG_MMU
1289	{
1290		.procname	= "mmap_min_addr",
1291		.data		= &dac_mmap_min_addr,
1292		.maxlen		= sizeof(unsigned long),
1293		.mode		= 0644,
1294		.proc_handler	= mmap_min_addr_handler,
1295	},
1296#endif
1297#ifdef CONFIG_NUMA
1298	{
1299		.procname	= "numa_zonelist_order",
1300		.data		= &numa_zonelist_order,
1301		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1302		.mode		= 0644,
1303		.proc_handler	= numa_zonelist_order_handler,
1304	},
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307   (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308	{
1309		.procname	= "vdso_enabled",
 
 
 
 
1310		.data		= &vdso_enabled,
1311		.maxlen		= sizeof(vdso_enabled),
 
1312		.mode		= 0644,
1313		.proc_handler	= proc_dointvec,
1314		.extra1		= &zero,
1315	},
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318	{
1319		.procname	= "highmem_is_dirtyable",
1320		.data		= &vm_highmem_is_dirtyable,
1321		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1322		.mode		= 0644,
1323		.proc_handler	= proc_dointvec_minmax,
1324		.extra1		= &zero,
1325		.extra2		= &one,
1326	},
1327#endif
1328	{
1329		.procname	= "scan_unevictable_pages",
1330		.data		= &scan_unevictable_pages,
1331		.maxlen		= sizeof(scan_unevictable_pages),
1332		.mode		= 0644,
1333		.proc_handler	= scan_unevictable_handler,
1334	},
1335#ifdef CONFIG_MEMORY_FAILURE
1336	{
1337		.procname	= "memory_failure_early_kill",
1338		.data		= &sysctl_memory_failure_early_kill,
1339		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1340		.mode		= 0644,
1341		.proc_handler	= proc_dointvec_minmax,
1342		.extra1		= &zero,
1343		.extra2		= &one,
1344	},
1345	{
1346		.procname	= "memory_failure_recovery",
1347		.data		= &sysctl_memory_failure_recovery,
1348		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1349		.mode		= 0644,
1350		.proc_handler	= proc_dointvec_minmax,
1351		.extra1		= &zero,
1352		.extra2		= &one,
1353	},
1354#endif
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1355	{ }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360	{ }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365	{
1366		.procname	= "inode-nr",
1367		.data		= &inodes_stat,
1368		.maxlen		= 2*sizeof(int),
1369		.mode		= 0444,
1370		.proc_handler	= proc_nr_inodes,
1371	},
1372	{
1373		.procname	= "inode-state",
1374		.data		= &inodes_stat,
1375		.maxlen		= 7*sizeof(int),
1376		.mode		= 0444,
1377		.proc_handler	= proc_nr_inodes,
1378	},
1379	{
1380		.procname	= "file-nr",
1381		.data		= &files_stat,
1382		.maxlen		= sizeof(files_stat),
1383		.mode		= 0444,
1384		.proc_handler	= proc_nr_files,
1385	},
1386	{
1387		.procname	= "file-max",
1388		.data		= &files_stat.max_files,
1389		.maxlen		= sizeof(files_stat.max_files),
1390		.mode		= 0644,
1391		.proc_handler	= proc_doulongvec_minmax,
1392	},
1393	{
1394		.procname	= "nr_open",
1395		.data		= &sysctl_nr_open,
1396		.maxlen		= sizeof(int),
1397		.mode		= 0644,
1398		.proc_handler	= proc_dointvec_minmax,
1399		.extra1		= &sysctl_nr_open_min,
1400		.extra2		= &sysctl_nr_open_max,
1401	},
1402	{
1403		.procname	= "dentry-state",
1404		.data		= &dentry_stat,
1405		.maxlen		= 6*sizeof(int),
1406		.mode		= 0444,
1407		.proc_handler	= proc_nr_dentry,
1408	},
1409	{
1410		.procname	= "overflowuid",
1411		.data		= &fs_overflowuid,
1412		.maxlen		= sizeof(int),
1413		.mode		= 0644,
1414		.proc_handler	= proc_dointvec_minmax,
1415		.extra1		= &minolduid,
1416		.extra2		= &maxolduid,
1417	},
1418	{
1419		.procname	= "overflowgid",
1420		.data		= &fs_overflowgid,
1421		.maxlen		= sizeof(int),
1422		.mode		= 0644,
1423		.proc_handler	= proc_dointvec_minmax,
1424		.extra1		= &minolduid,
1425		.extra2		= &maxolduid,
1426	},
1427#ifdef CONFIG_FILE_LOCKING
1428	{
1429		.procname	= "leases-enable",
1430		.data		= &leases_enable,
1431		.maxlen		= sizeof(int),
1432		.mode		= 0644,
1433		.proc_handler	= proc_dointvec,
1434	},
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437	{
1438		.procname	= "dir-notify-enable",
1439		.data		= &dir_notify_enable,
1440		.maxlen		= sizeof(int),
1441		.mode		= 0644,
1442		.proc_handler	= proc_dointvec,
1443	},
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447	{
1448		.procname	= "lease-break-time",
1449		.data		= &lease_break_time,
1450		.maxlen		= sizeof(int),
1451		.mode		= 0644,
1452		.proc_handler	= proc_dointvec,
1453	},
1454#endif
1455#ifdef CONFIG_AIO
1456	{
1457		.procname	= "aio-nr",
1458		.data		= &aio_nr,
1459		.maxlen		= sizeof(aio_nr),
1460		.mode		= 0444,
1461		.proc_handler	= proc_doulongvec_minmax,
1462	},
1463	{
1464		.procname	= "aio-max-nr",
1465		.data		= &aio_max_nr,
1466		.maxlen		= sizeof(aio_max_nr),
1467		.mode		= 0644,
1468		.proc_handler	= proc_doulongvec_minmax,
1469	},
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472	{
1473		.procname	= "inotify",
1474		.mode		= 0555,
1475		.child		= inotify_table,
1476	},
1477#endif	
1478#ifdef CONFIG_EPOLL
1479	{
1480		.procname	= "epoll",
1481		.mode		= 0555,
1482		.child		= epoll_table,
1483	},
1484#endif
1485#endif
1486	{
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1487		.procname	= "suid_dumpable",
1488		.data		= &suid_dumpable,
1489		.maxlen		= sizeof(int),
1490		.mode		= 0644,
1491		.proc_handler	= proc_dointvec_minmax,
1492		.extra1		= &zero,
1493		.extra2		= &two,
1494	},
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496	{
1497		.procname	= "binfmt_misc",
1498		.mode		= 0555,
1499		.child		= binfmt_misc_table,
1500	},
1501#endif
1502	{
1503		.procname	= "pipe-max-size",
1504		.data		= &pipe_max_size,
1505		.maxlen		= sizeof(int),
1506		.mode		= 0644,
1507		.proc_handler	= &pipe_proc_fn,
1508		.extra1		= &pipe_min_size,
1509	},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1510	{ }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515    defined(CONFIG_S390) || defined(CONFIG_TILE)
1516	{
1517		.procname	= "exception-trace",
1518		.data		= &show_unhandled_signals,
1519		.maxlen		= sizeof(int),
1520		.mode		= 0644,
1521		.proc_handler	= proc_dointvec
1522	},
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525	{
1526		.procname	= "kprobes-optimization",
1527		.data		= &sysctl_kprobes_optimization,
1528		.maxlen		= sizeof(int),
1529		.mode		= 0644,
1530		.proc_handler	= proc_kprobes_optimization_handler,
1531		.extra1		= &zero,
1532		.extra2		= &one,
1533	},
1534#endif
1535	{ }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539	{ }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547	if (unlikely(p->unregistering))
1548		return 0;
1549	p->used++;
1550	return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556	if (!--p->used)
1557		if (unlikely(p->unregistering))
1558			complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564	/*
1565	 * if p->used is 0, nobody will ever touch that entry again;
1566	 * we'll eliminate all paths to it before dropping sysctl_lock
1567	 */
1568	if (unlikely(p->used)) {
1569		struct completion wait;
1570		init_completion(&wait);
1571		p->unregistering = &wait;
1572		spin_unlock(&sysctl_lock);
1573		wait_for_completion(&wait);
1574		spin_lock(&sysctl_lock);
1575	} else {
1576		/* anything non-NULL; we'll never dereference it */
1577		p->unregistering = ERR_PTR(-EINVAL);
1578	}
1579	/*
1580	 * do not remove from the list until nobody holds it; walking the
1581	 * list in do_sysctl() relies on that.
1582	 */
1583	list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588	spin_lock(&sysctl_lock);
1589	head->count++;
1590	spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595	spin_lock(&sysctl_lock);
1596	if (!--head->count)
1597		kfree_rcu(head, rcu);
1598	spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603	if (!head)
1604		BUG();
1605	spin_lock(&sysctl_lock);
1606	if (!use_table(head))
1607		head = ERR_PTR(-ENOENT);
1608	spin_unlock(&sysctl_lock);
1609	return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614	if (!head)
1615		return;
1616	spin_lock(&sysctl_lock);
1617	unuse_table(head);
1618	spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624	struct ctl_table_set *set = &root->default_set;
1625	if (root->lookup)
1626		set = root->lookup(root, namespaces);
1627	return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633	struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634	return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638					    struct ctl_table_header *prev)
1639{
1640	struct ctl_table_root *root;
1641	struct list_head *header_list;
1642	struct ctl_table_header *head;
1643	struct list_head *tmp;
1644
1645	spin_lock(&sysctl_lock);
1646	if (prev) {
1647		head = prev;
1648		tmp = &prev->ctl_entry;
1649		unuse_table(prev);
1650		goto next;
1651	}
1652	tmp = &root_table_header.ctl_entry;
1653	for (;;) {
1654		head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656		if (!use_table(head))
1657			goto next;
1658		spin_unlock(&sysctl_lock);
1659		return head;
1660	next:
1661		root = head->root;
1662		tmp = tmp->next;
1663		header_list = lookup_header_list(root, namespaces);
1664		if (tmp != header_list)
1665			continue;
1666
1667		do {
1668			root = list_entry(root->root_list.next,
1669					struct ctl_table_root, root_list);
1670			if (root == &sysctl_table_root)
1671				goto out;
1672			header_list = lookup_header_list(root, namespaces);
1673		} while (list_empty(header_list));
1674		tmp = header_list->next;
1675	}
1676out:
1677	spin_unlock(&sysctl_lock);
1678	return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683	return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688	spin_lock(&sysctl_lock);
1689	list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690	spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700	if (!current_euid())
1701		mode >>= 6;
1702	else if (in_egroup_p(0))
1703		mode >>= 3;
1704	if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705		return 0;
1706	return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711	int mode;
1712
1713	if (root->permissions)
1714		mode = root->permissions(root, current->nsproxy, table);
1715	else
1716		mode = table->mode;
1717
1718	return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723	for (; table->procname; table++) {
1724		table->parent = parent;
1725		if (table->child)
1726			sysctl_set_parent(table, table->child);
1727	}
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732	sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734	sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736	return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742				      struct ctl_table *table)
1743{
1744	struct ctl_table *p;
1745	const char *s = branch->procname;
1746
1747	/* branch should have named subdirectory as its first element */
1748	if (!s || !branch->child)
1749		return NULL;
1750
1751	/* ... and nothing else */
1752	if (branch[1].procname)
1753		return NULL;
1754
1755	/* table should contain subdirectory with the same name */
1756	for (p = table; p->procname; p++) {
1757		if (!p->child)
1758			continue;
1759		if (p->procname && strcmp(p->procname, s) == 0)
1760			return p;
1761	}
1762	return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768	struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769	struct ctl_table *next;
1770	int is_better = 0;
1771	int not_in_parent = !p->attached_by;
1772
1773	while ((next = is_branch_in(by, to)) != NULL) {
1774		if (by == q->attached_by)
1775			is_better = 1;
1776		if (to == p->attached_by)
1777			not_in_parent = 1;
1778		by = by->child;
1779		to = next->child;
1780	}
1781
1782	if (is_better && not_in_parent) {
1783		q->attached_by = by;
1784		q->attached_to = to;
1785		q->parent = p;
1786	}
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 *            enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 *         %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table.  The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler).  Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(), 
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843	struct ctl_table_root *root,
1844	struct nsproxy *namespaces,
1845	const struct ctl_path *path, struct ctl_table *table)
1846{
1847	struct ctl_table_header *header;
1848	struct ctl_table *new, **prevp;
1849	unsigned int n, npath;
1850	struct ctl_table_set *set;
1851
1852	/* Count the path components */
1853	for (npath = 0; path[npath].procname; ++npath)
1854		;
1855
1856	/*
1857	 * For each path component, allocate a 2-element ctl_table array.
1858	 * The first array element will be filled with the sysctl entry
1859	 * for this, the second will be the sentinel (procname == 0).
1860	 *
1861	 * We allocate everything in one go so that we don't have to
1862	 * worry about freeing additional memory in unregister_sysctl_table.
1863	 */
1864	header = kzalloc(sizeof(struct ctl_table_header) +
1865			 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866	if (!header)
1867		return NULL;
1868
1869	new = (struct ctl_table *) (header + 1);
1870
1871	/* Now connect the dots */
1872	prevp = &header->ctl_table;
1873	for (n = 0; n < npath; ++n, ++path) {
1874		/* Copy the procname */
1875		new->procname = path->procname;
1876		new->mode     = 0555;
1877
1878		*prevp = new;
1879		prevp = &new->child;
1880
1881		new += 2;
1882	}
1883	*prevp = table;
1884	header->ctl_table_arg = table;
1885
1886	INIT_LIST_HEAD(&header->ctl_entry);
1887	header->used = 0;
1888	header->unregistering = NULL;
1889	header->root = root;
1890	sysctl_set_parent(NULL, header->ctl_table);
1891	header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893	if (sysctl_check_table(namespaces, header->ctl_table)) {
1894		kfree(header);
1895		return NULL;
1896	}
1897#endif
1898	spin_lock(&sysctl_lock);
1899	header->set = lookup_header_set(root, namespaces);
1900	header->attached_by = header->ctl_table;
1901	header->attached_to = root_table;
1902	header->parent = &root_table_header;
1903	for (set = header->set; set; set = set->parent) {
1904		struct ctl_table_header *p;
1905		list_for_each_entry(p, &set->list, ctl_entry) {
1906			if (p->unregistering)
1907				continue;
1908			try_attach(p, header);
1909		}
1910	}
1911	header->parent->count++;
1912	list_add_tail(&header->ctl_entry, &header->set->list);
1913	spin_unlock(&sysctl_lock);
1914
1915	return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929						struct ctl_table *table)
1930{
1931	return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932					path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946	static const struct ctl_path null_path[] = { {} };
1947
1948	return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960	might_sleep();
1961
1962	if (header == NULL)
1963		return;
1964
1965	spin_lock(&sysctl_lock);
1966	start_unregistering(header);
1967	if (!--header->parent->count) {
1968		WARN_ON(1);
1969		kfree_rcu(header->parent, rcu);
1970	}
1971	if (!--header->count)
1972		kfree_rcu(header, rcu);
1973	spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978	struct ctl_table_set *set = p->set;
1979	int res;
1980	spin_lock(&sysctl_lock);
1981	if (p->unregistering)
1982		res = 0;
1983	else if (!set->is_seen)
1984		res = 1;
1985	else
1986		res = set->is_seen(set);
1987	spin_unlock(&sysctl_lock);
1988	return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992	struct ctl_table_set *parent,
1993	int (*is_seen)(struct ctl_table_set *))
1994{
1995	INIT_LIST_HEAD(&p->list);
1996	p->parent = parent ? parent : &sysctl_table_root.default_set;
1997	p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003	return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007						    struct ctl_table *table)
2008{
2009	return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017	struct ctl_table_set *parent,
2018	int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035			   void __user *buffer,
2036			   size_t *lenp, loff_t *ppos)
2037{
2038	size_t len;
2039	char __user *p;
2040	char c;
2041
2042	if (!data || !maxlen || !*lenp) {
2043		*lenp = 0;
2044		return 0;
2045	}
2046
2047	if (write) {
2048		len = 0;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2049		p = buffer;
2050		while (len < *lenp) {
2051			if (get_user(c, p++))
2052				return -EFAULT;
2053			if (c == 0 || c == '\n')
2054				break;
2055			len++;
2056		}
2057		if (len >= maxlen)
2058			len = maxlen-1;
2059		if(copy_from_user(data, buffer, len))
2060			return -EFAULT;
2061		((char *) data)[len] = 0;
2062		*ppos += *lenp;
2063	} else {
2064		len = strlen(data);
2065		if (len > maxlen)
2066			len = maxlen;
2067
2068		if (*ppos > len) {
2069			*lenp = 0;
2070			return 0;
2071		}
2072
2073		data += *ppos;
2074		len  -= *ppos;
2075
2076		if (len > *lenp)
2077			len = *lenp;
2078		if (len)
2079			if(copy_to_user(buffer, data, len))
2080				return -EFAULT;
2081		if (len < *lenp) {
2082			if(put_user('\n', ((char __user *) buffer) + len))
2083				return -EFAULT;
2084			len++;
2085		}
2086		*lenp = len;
2087		*ppos += len;
2088	}
2089	return 0;
2090}
2091
 
 
 
 
 
 
 
 
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110		  void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112	return _proc_do_string(table->data, table->maxlen, write,
2113			       buffer, lenp, ppos);
 
 
 
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118	size_t ret;
2119	char *tmp = skip_spaces(*buf);
2120	ret = tmp - *buf;
2121	*buf = tmp;
2122	return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127	while (*size) {
2128		if (**buf != v)
2129			break;
2130		(*size)--;
2131		(*buf)++;
2132	}
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153			  unsigned long *val, bool *neg,
2154			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156	int len;
2157	char *p, tmp[TMPBUFLEN];
2158
2159	if (!*size)
2160		return -EINVAL;
2161
2162	len = *size;
2163	if (len > TMPBUFLEN - 1)
2164		len = TMPBUFLEN - 1;
2165
2166	memcpy(tmp, *buf, len);
2167
2168	tmp[len] = 0;
2169	p = tmp;
2170	if (*p == '-' && *size > 1) {
2171		*neg = true;
2172		p++;
2173	} else
2174		*neg = false;
2175	if (!isdigit(*p))
2176		return -EINVAL;
2177
2178	*val = simple_strtoul(p, &p, 0);
2179
2180	len = p - tmp;
2181
2182	/* We don't know if the next char is whitespace thus we may accept
2183	 * invalid integers (e.g. 1234...a) or two integers instead of one
2184	 * (e.g. 123...1). So lets not allow such large numbers. */
2185	if (len == TMPBUFLEN - 1)
2186		return -EINVAL;
2187
2188	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189		return -EINVAL;
2190
2191	if (tr && (len < *size))
2192		*tr = *p;
2193
2194	*buf += len;
2195	*size -= len;
2196
2197	return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212			  bool neg)
2213{
2214	int len;
2215	char tmp[TMPBUFLEN], *p = tmp;
2216
2217	sprintf(p, "%s%lu", neg ? "-" : "", val);
2218	len = strlen(tmp);
2219	if (len > *size)
2220		len = *size;
2221	if (copy_to_user(*buf, tmp, len))
2222		return -EFAULT;
2223	*size -= len;
2224	*buf += len;
2225	return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231	if (*size) {
2232		char __user **buffer = (char __user **)buf;
2233		if (put_user(c, *buffer))
2234			return -EFAULT;
2235		(*size)--, (*buffer)++;
2236		*buf = *buffer;
2237	}
2238	return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242				 int *valp,
2243				 int write, void *data)
2244{
2245	if (write) {
2246		*valp = *negp ? -*lvalp : *lvalp;
 
 
 
 
 
 
 
 
2247	} else {
2248		int val = *valp;
2249		if (val < 0) {
2250			*negp = true;
2251			*lvalp = (unsigned long)-val;
2252		} else {
2253			*negp = false;
2254			*lvalp = (unsigned long)val;
2255		}
2256	}
2257	return 0;
2258}
2259
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263		  int write, void __user *buffer,
2264		  size_t *lenp, loff_t *ppos,
2265		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266			      int write, void *data),
2267		  void *data)
2268{
2269	int *i, vleft, first = 1, err = 0;
2270	unsigned long page = 0;
2271	size_t left;
2272	char *kbuf;
2273	
2274	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275		*lenp = 0;
2276		return 0;
2277	}
2278	
2279	i = (int *) tbl_data;
2280	vleft = table->maxlen / sizeof(*i);
2281	left = *lenp;
2282
2283	if (!conv)
2284		conv = do_proc_dointvec_conv;
2285
2286	if (write) {
 
 
 
 
 
 
 
 
 
 
 
 
2287		if (left > PAGE_SIZE - 1)
2288			left = PAGE_SIZE - 1;
2289		page = __get_free_page(GFP_TEMPORARY);
2290		kbuf = (char *) page;
2291		if (!kbuf)
2292			return -ENOMEM;
2293		if (copy_from_user(kbuf, buffer, left)) {
2294			err = -EFAULT;
2295			goto free;
2296		}
2297		kbuf[left] = 0;
2298	}
2299
2300	for (; left && vleft--; i++, first=0) {
2301		unsigned long lval;
2302		bool neg;
2303
2304		if (write) {
2305			left -= proc_skip_spaces(&kbuf);
2306
2307			if (!left)
2308				break;
2309			err = proc_get_long(&kbuf, &left, &lval, &neg,
2310					     proc_wspace_sep,
2311					     sizeof(proc_wspace_sep), NULL);
2312			if (err)
2313				break;
2314			if (conv(&neg, &lval, i, 1, data)) {
2315				err = -EINVAL;
2316				break;
2317			}
2318		} else {
2319			if (conv(&neg, &lval, i, 0, data)) {
2320				err = -EINVAL;
2321				break;
2322			}
2323			if (!first)
2324				err = proc_put_char(&buffer, &left, '\t');
2325			if (err)
2326				break;
2327			err = proc_put_long(&buffer, &left, lval, neg);
2328			if (err)
2329				break;
2330		}
2331	}
2332
2333	if (!write && !first && left && !err)
2334		err = proc_put_char(&buffer, &left, '\n');
2335	if (write && !err && left)
2336		left -= proc_skip_spaces(&kbuf);
2337free:
2338	if (write) {
2339		free_page(page);
2340		if (first)
2341			return err ? : -EINVAL;
2342	}
2343	*lenp -= left;
 
2344	*ppos += *lenp;
2345	return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349		  void __user *buffer, size_t *lenp, loff_t *ppos,
2350		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351			      int write, void *data),
2352		  void *data)
2353{
2354	return __do_proc_dointvec(table->data, table, write,
2355			buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string. 
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372		     void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375		    	    NULL,NULL);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383			       void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385	struct ctl_table t;
2386	unsigned long tmptaint = get_taint();
2387	int err;
2388
2389	if (write && !capable(CAP_SYS_ADMIN))
2390		return -EPERM;
2391
2392	t = *table;
2393	t.data = &tmptaint;
2394	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395	if (err < 0)
2396		return err;
2397
2398	if (write) {
2399		/*
2400		 * Poor man's atomic or. Not worth adding a primitive
2401		 * to everyone's atomic.h for this
2402		 */
2403		int i;
2404		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405			if ((tmptaint >> i) & 1)
2406				add_taint(i);
2407		}
2408	}
2409
2410	return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415				void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417	if (write && !capable(CAP_SYS_ADMIN))
2418		return -EPERM;
2419
2420	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425	int *min;
2426	int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430					int *valp,
2431					int write, void *data)
2432{
2433	struct do_proc_dointvec_minmax_conv_param *param = data;
2434	if (write) {
2435		int val = *negp ? -*lvalp : *lvalp;
2436		if ((param->min && *param->min > val) ||
2437		    (param->max && *param->max < val))
2438			return -EINVAL;
2439		*valp = val;
2440	} else {
2441		int val = *valp;
2442		if (val < 0) {
2443			*negp = true;
2444			*lvalp = (unsigned long)-val;
2445		} else {
2446			*negp = false;
2447			*lvalp = (unsigned long)val;
2448		}
2449	}
2450	return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470		  void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472	struct do_proc_dointvec_minmax_conv_param param = {
2473		.min = (int *) table->extra1,
2474		.max = (int *) table->extra2,
2475	};
2476	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477				do_proc_dointvec_minmax_conv, &param);
2478}
2479
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481				     void __user *buffer,
2482				     size_t *lenp, loff_t *ppos,
2483				     unsigned long convmul,
2484				     unsigned long convdiv)
2485{
2486	unsigned long *i, *min, *max;
2487	int vleft, first = 1, err = 0;
2488	unsigned long page = 0;
2489	size_t left;
2490	char *kbuf;
2491
2492	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493		*lenp = 0;
2494		return 0;
2495	}
2496
2497	i = (unsigned long *) data;
2498	min = (unsigned long *) table->extra1;
2499	max = (unsigned long *) table->extra2;
2500	vleft = table->maxlen / sizeof(unsigned long);
2501	left = *lenp;
2502
2503	if (write) {
 
 
 
 
 
 
 
 
 
 
 
 
2504		if (left > PAGE_SIZE - 1)
2505			left = PAGE_SIZE - 1;
2506		page = __get_free_page(GFP_TEMPORARY);
2507		kbuf = (char *) page;
2508		if (!kbuf)
2509			return -ENOMEM;
2510		if (copy_from_user(kbuf, buffer, left)) {
2511			err = -EFAULT;
2512			goto free;
2513		}
2514		kbuf[left] = 0;
2515	}
2516
2517	for (; left && vleft--; i++, first = 0) {
2518		unsigned long val;
2519
2520		if (write) {
2521			bool neg;
2522
2523			left -= proc_skip_spaces(&kbuf);
2524
2525			err = proc_get_long(&kbuf, &left, &val, &neg,
2526					     proc_wspace_sep,
2527					     sizeof(proc_wspace_sep), NULL);
2528			if (err)
2529				break;
2530			if (neg)
2531				continue;
 
2532			if ((min && val < *min) || (max && val > *max))
2533				continue;
2534			*i = val;
2535		} else {
2536			val = convdiv * (*i) / convmul;
2537			if (!first)
2538				err = proc_put_char(&buffer, &left, '\t');
 
 
 
2539			err = proc_put_long(&buffer, &left, val, false);
2540			if (err)
2541				break;
2542		}
2543	}
2544
2545	if (!write && !first && left && !err)
2546		err = proc_put_char(&buffer, &left, '\n');
2547	if (write && !err)
2548		left -= proc_skip_spaces(&kbuf);
2549free:
2550	if (write) {
2551		free_page(page);
2552		if (first)
2553			return err ? : -EINVAL;
2554	}
2555	*lenp -= left;
 
2556	*ppos += *lenp;
2557	return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561				     void __user *buffer,
2562				     size_t *lenp, loff_t *ppos,
2563				     unsigned long convmul,
2564				     unsigned long convdiv)
2565{
2566	return __do_proc_doulongvec_minmax(table->data, table, write,
2567			buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587			   void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589    return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int 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 do_proc_doulongvec_minmax(table, write, buffer,
2614				     lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619					 int *valp,
2620					 int write, void *data)
2621{
2622	if (write) {
2623		if (*lvalp > LONG_MAX / HZ)
2624			return 1;
2625		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626	} else {
2627		int val = *valp;
2628		unsigned long lval;
2629		if (val < 0) {
2630			*negp = true;
2631			lval = (unsigned long)-val;
2632		} else {
2633			*negp = false;
2634			lval = (unsigned long)val;
2635		}
2636		*lvalp = lval / HZ;
2637	}
2638	return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642						int *valp,
2643						int write, void *data)
2644{
2645	if (write) {
2646		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647			return 1;
2648		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649	} else {
2650		int val = *valp;
2651		unsigned long lval;
2652		if (val < 0) {
2653			*negp = true;
2654			lval = (unsigned long)-val;
2655		} else {
2656			*negp = false;
2657			lval = (unsigned long)val;
2658		}
2659		*lvalp = jiffies_to_clock_t(lval);
2660	}
2661	return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665					    int *valp,
2666					    int write, void *data)
2667{
2668	if (write) {
2669		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
 
 
 
 
2670	} else {
2671		int val = *valp;
2672		unsigned long lval;
2673		if (val < 0) {
2674			*negp = true;
2675			lval = (unsigned long)-val;
2676		} else {
2677			*negp = false;
2678			lval = (unsigned long)val;
2679		}
2680		*lvalp = jiffies_to_msecs(lval);
2681	}
2682	return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string. 
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701			  void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704		    	    do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string. 
2717 * The values read are assumed to be in 1/USER_HZ seconds, and 
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
2721 */
2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2723				 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
2725    return do_proc_dointvec(table,write,buffer,lenp,ppos,
2726		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string. 
2740 * The values read are assumed to be in 1/1000 seconds, and 
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746			     void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749				do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753			   void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755	struct pid *new_pid;
2756	pid_t tmp;
2757	int r;
2758
2759	tmp = pid_vnr(cad_pid);
2760
2761	r = __do_proc_dointvec(&tmp, table, write, buffer,
2762			       lenp, ppos, NULL, NULL);
2763	if (r || !write)
2764		return r;
2765
2766	new_pid = find_get_pid(tmp);
2767	if (!new_pid)
2768		return -ESRCH;
2769
2770	put_pid(xchg(&cad_pid, new_pid));
2771	return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792			 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794	int err = 0;
2795	bool first = 1;
2796	size_t left = *lenp;
2797	unsigned long bitmap_len = table->maxlen;
2798	unsigned long *bitmap = (unsigned long *) table->data;
2799	unsigned long *tmp_bitmap = NULL;
2800	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802	if (!bitmap_len || !left || (*ppos && !write)) {
2803		*lenp = 0;
2804		return 0;
2805	}
2806
2807	if (write) {
2808		unsigned long page = 0;
2809		char *kbuf;
2810
2811		if (left > PAGE_SIZE - 1)
2812			left = PAGE_SIZE - 1;
2813
2814		page = __get_free_page(GFP_TEMPORARY);
2815		kbuf = (char *) page;
2816		if (!kbuf)
2817			return -ENOMEM;
2818		if (copy_from_user(kbuf, buffer, left)) {
2819			free_page(page);
2820			return -EFAULT;
2821                }
2822		kbuf[left] = 0;
2823
2824		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825				     GFP_KERNEL);
2826		if (!tmp_bitmap) {
2827			free_page(page);
2828			return -ENOMEM;
2829		}
2830		proc_skip_char(&kbuf, &left, '\n');
2831		while (!err && left) {
2832			unsigned long val_a, val_b;
2833			bool neg;
2834
2835			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836					     sizeof(tr_a), &c);
2837			if (err)
2838				break;
2839			if (val_a >= bitmap_len || neg) {
2840				err = -EINVAL;
2841				break;
2842			}
2843
2844			val_b = val_a;
2845			if (left) {
2846				kbuf++;
2847				left--;
2848			}
2849
2850			if (c == '-') {
2851				err = proc_get_long(&kbuf, &left, &val_b,
2852						     &neg, tr_b, sizeof(tr_b),
2853						     &c);
2854				if (err)
2855					break;
2856				if (val_b >= bitmap_len || neg ||
2857				    val_a > val_b) {
2858					err = -EINVAL;
2859					break;
2860				}
2861				if (left) {
2862					kbuf++;
2863					left--;
2864				}
2865			}
2866
2867			while (val_a <= val_b)
2868				set_bit(val_a++, tmp_bitmap);
2869
2870			first = 0;
2871			proc_skip_char(&kbuf, &left, '\n');
2872		}
2873		free_page(page);
2874	} else {
2875		unsigned long bit_a, bit_b = 0;
2876
2877		while (left) {
2878			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879			if (bit_a >= bitmap_len)
2880				break;
2881			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882						   bit_a + 1) - 1;
2883
2884			if (!first) {
2885				err = proc_put_char(&buffer, &left, ',');
2886				if (err)
2887					break;
2888			}
2889			err = proc_put_long(&buffer, &left, bit_a, false);
2890			if (err)
2891				break;
2892			if (bit_a != bit_b) {
2893				err = proc_put_char(&buffer, &left, '-');
2894				if (err)
2895					break;
2896				err = proc_put_long(&buffer, &left, bit_b, false);
2897				if (err)
2898					break;
2899			}
2900
2901			first = 0; bit_b++;
2902		}
2903		if (!err)
2904			err = proc_put_char(&buffer, &left, '\n');
2905	}
2906
2907	if (!err) {
2908		if (write) {
2909			if (*ppos)
2910				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911			else
2912				memcpy(bitmap, tmp_bitmap,
2913					BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914		}
2915		kfree(tmp_bitmap);
2916		*lenp -= left;
2917		*ppos += *lenp;
2918		return 0;
2919	} else {
2920		kfree(tmp_bitmap);
2921		return err;
2922	}
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928		  void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930	return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934		  void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936	return -ENOSYS;
2937}
2938
 
 
 
 
 
 
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940		    void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942	return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946		    void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948	return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952		    void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954	return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958			     void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960	return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964		    void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966	return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970				      void __user *buffer,
2971				      size_t *lenp, loff_t *ppos)
2972{
2973    return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
 
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);