projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
sched: Validate CFS quota hierarchies
[linux-2.6-block.git] / kernel / sysctl.c
CommitLineData
1da177e4
LT		 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
1da177e4
LT		 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>
d33ed52d 26#include <linux/signal.h>
455cd5ab 27#include <linux/printk.h>
1da177e4 28#include <linux/proc_fs.h>
72c2d582 29#include <linux/security.h>
1da177e4 30#include <linux/ctype.h>
dfec072e 31#include <linux/kmemcheck.h>
62239ac2 32#include <linux/fs.h>
1da177e4
LT		 33#include <linux/init.h>
34#include <linux/kernel.h>
0296b228 35#include <linux/kobject.h>
20380731 36#include <linux/net.h>
1da177e4
LT		 37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
3fff4c42 40#include <linux/ratelimit.h>
76ab0f53 41#include <linux/compaction.h>
1da177e4 42#include <linux/hugetlb.h>
1da177e4 43#include <linux/initrd.h>
0b77f5bf 44#include <linux/key.h>
1da177e4
LT		 45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
6e006701 48#include <linux/dnotify.h>
1da177e4 49#include <linux/syscalls.h>
c748e134 50#include <linux/vmstat.h>
c255d844
PM		 51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
10a0a8d4 53#include <linux/reboot.h>
b0fc494f 54#include <linux/ftrace.h>
cdd6c482 55#include <linux/perf_event.h>
b2be84df 56#include <linux/kprobes.h>
b492e95b 57#include <linux/pipe_fs_i.h>
8e4228e1 58#include <linux/oom.h>
17f60a7d 59#include <linux/kmod.h>
1da177e4
LT		 60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
29cbc78b
AK		 64#ifdef CONFIG_X86
65#include <asm/nmi.h>
0741f4d2 66#include <asm/stacktrace.h>
6e7c4025 67#include <asm/io.h>
29cbc78b 68#endif
c55b7c3e
DY		 69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
4f0e056f
DY		 72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
2edf5e49
DY		 75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
15485a46
DY		 78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
29cbc78b 81
58687acb 82#ifdef CONFIG_LOCKUP_DETECTOR
504d7cf1
DZ		 83#include <linux/nmi.h>
84#endif
85
7058cb02 86
1da177e4
LT		 87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
1da177e4
LT		 90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
1da177e4 93extern int core_uses_pid;
d6e71144 94extern int suid_dumpable;
1da177e4 95extern char core_pattern[];
a293980c 96extern unsigned int core_pipe_limit;
1da177e4
LT		 97extern int pid_max;
98extern int min_free_kbytes;
1da177e4 99extern int pid_max_min, pid_max_max;
9d0243bc 100extern int sysctl_drop_caches;
8ad4b1fb 101extern int percpu_pagelist_fraction;
bebfa101 102extern int compat_log;
9745512c 103extern int latencytop_enabled;
eceea0b3 104extern int sysctl_nr_open_min, sysctl_nr_open_max;
dd8632a1
PM		 105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
cb684b5b 108#ifdef CONFIG_BLOCK
5e605b64 109extern int blk_iopoll_enabled;
cb684b5b 110#endif
1da177e4 111
c4f3b63f 112/* Constants used for minimum and maximum */
2508ce18 113#ifdef CONFIG_LOCKUP_DETECTOR
c4f3b63f 114static int sixty = 60;
9383d967 115static int neg_one = -1;
c4f3b63f
RT		 116#endif
117
c4f3b63f 118static int zero;
cd5f9a4c
LT		 119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
cb16e95f 121static int __maybe_unused three = 3;
fc3501d4 122static unsigned long one_ul = 1;
c4f3b63f 123static int one_hundred = 100;
af91322e
DY		 124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
c4f3b63f 127
9e4a5bda
AR		 128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
1da177e4
LT		 131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
8ad4b1fb 134static int min_percpu_pagelist_fract = 8;
1da177e4
LT		 135
136static int ngroups_max = NGROUPS_MAX;
137
d14f1729
DY		 138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
72c57ed5 141#ifdef CONFIG_SPARC
17f04fbb 142#include <asm/system.h>
1da177e4
LT		 143#endif
144
0871420f
DM		 145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
1da177e4
LT		 149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
347a8dc3 154#ifdef CONFIG_S390
1da177e4
LT		 155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
951f22d5 159extern int spin_retry;
1da177e4
LT		 160#endif
161
d2b176ed
JS		 162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
88fc241f 164extern int unaligned_dump_stack;
d2b176ed
JS		 165#endif
166
d6f8ff73 167#ifdef CONFIG_PROC_SYSCTL
8d65af78 168static int proc_do_cad_pid(struct ctl_table *table, int write,
9ec52099 169 void __user *buffer, size_t *lenp, loff_t *ppos);
8d65af78 170static int proc_taint(struct ctl_table *table, int write,
34f5a398 171 void __user *buffer, size_t *lenp, loff_t *ppos);
d6f8ff73 172#endif
9ec52099 173
bfdc0b49
RW		 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
97f5f0cd 179#ifdef CONFIG_MAGIC_SYSRQ
8c6a98b2
AW		 180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
97f5f0cd
DT		 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
d8217f07 201static struct ctl_table root_table[];
e51b6ba0
EB		 202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
dfef6dcd 204 {{.count = 1,
e51b6ba0 205 .ctl_table = root_table,
dfef6dcd 206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
e51b6ba0 207 .root = &sysctl_table_root,
73455092 208 .set = &sysctl_table_root.default_set,
e51b6ba0
EB		 209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
73455092 212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
e51b6ba0 213};
1da177e4 214
d8217f07
EB		 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[];
7ef9964e
DL		 221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
1da177e4
LT		 224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
1da177e4
LT		 229/* The default sysctl tables: */
230
d8217f07 231static struct ctl_table root_table[] = {
1da177e4 232 {
1da177e4
LT		 233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
1da177e4
LT		 238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
1da177e4 242 {
1da177e4
LT		 243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
1da177e4
LT		 248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
1da177e4
LT		 253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
6fce56ec 257 { }
1da177e4
LT		 258};
259
77e54a1f 260#ifdef CONFIG_SCHED_DEBUG
73c4efd2
ED		 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 */
1983a922
CE		 265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
77e54a1f
IM		 267#endif
268
5e771905
MG		 269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
d8217f07 274static struct ctl_table kern_table[] = {
2bba22c5 275 {
2bba22c5
MG		 276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
6d456111 280 .proc_handler = proc_dointvec,
2bba22c5 281 },
77e54a1f
IM		 282#ifdef CONFIG_SCHED_DEBUG
283 {
b2be5e96
PZ		 284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
77e54a1f
IM		 286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
702a7c76 288 .proc_handler = sched_proc_update_handler,
b2be5e96
PZ		 289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
77e54a1f 291 },
21805085 292 {
21805085
PZ		 293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
702a7c76 297 .proc_handler = sched_proc_update_handler,
21805085
PZ		 298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
77e54a1f 301 {
77e54a1f
IM		 302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
702a7c76 306 .proc_handler = sched_proc_update_handler,
77e54a1f
IM		 307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
1983a922 310 {
1983a922
CE		 311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
702a7c76 315 .proc_handler = sched_proc_update_handler,
1983a922
CE		 316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
2398f2c6 318 },
da84d961 319 {
da84d961
IM		 320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
6d456111 324 .proc_handler = proc_dointvec,
da84d961 325 },
b82d9fdd 326 {
b82d9fdd
PZ		 327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
fa85ae24 330 .mode = 0644,
6d456111 331 .proc_handler = proc_dointvec,
fa85ae24 332 },
e9e9250b 333 {
e9e9250b
PZ		 334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
6d456111 338 .proc_handler = proc_dointvec,
e9e9250b 339 },
a7a4f8a7
PT		 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 },
cd1bb94b 347 {
cd1bb94b
AB		 348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
6d456111 352 .proc_handler = proc_dointvec_minmax,
bfdb4d9f
AB		 353 .extra1 = &zero,
354 .extra2 = &one,
fa85ae24 355 },
1fc84aaa 356#endif
9f0c1e56 357 {
9f0c1e56
PZ		 358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
6d456111 362 .proc_handler = sched_rt_handler,
9f0c1e56
PZ		 363 },
364 {
9f0c1e56
PZ		 365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
6d456111 369 .proc_handler = sched_rt_handler,
9f0c1e56 370 },
5091faa4
MG		 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,
1747b21f 377 .proc_handler = proc_dointvec_minmax,
5091faa4
MG		 378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
f20786ff
PZ		 382#ifdef CONFIG_PROVE_LOCKING
383 {
f20786ff
PZ		 384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
6d456111 388 .proc_handler = proc_dointvec,
f20786ff
PZ		 389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
f20786ff
PZ		 393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
6d456111 397 .proc_handler = proc_dointvec,
f20786ff 398 },
77e54a1f 399#endif
1da177e4 400 {
1da177e4
LT		 401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
6d456111 405 .proc_handler = proc_dointvec,
1da177e4
LT		 406 },
407 {
1da177e4
LT		 408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
6d456111 412 .proc_handler = proc_dointvec,
1da177e4
LT		 413 },
414 {
1da177e4
LT		 415 .procname = "core_pattern",
416 .data = core_pattern,
71ce92f3 417 .maxlen = CORENAME_MAX_SIZE,
1da177e4 418 .mode = 0644,
6d456111 419 .proc_handler = proc_dostring,
1da177e4 420 },
a293980c 421 {
a293980c
NH		 422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
6d456111 426 .proc_handler = proc_dointvec,
a293980c 427 },
34f5a398 428#ifdef CONFIG_PROC_SYSCTL
1da177e4 429 {
1da177e4 430 .procname = "tainted",
25ddbb18 431 .maxlen = sizeof(long),
34f5a398 432 .mode = 0644,
6d456111 433 .proc_handler = proc_taint,
1da177e4 434 },
34f5a398 435#endif
9745512c
AV		 436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
6d456111 442 .proc_handler = proc_dointvec,
9745512c
AV		 443 },
444#endif
1da177e4
LT		 445#ifdef CONFIG_BLK_DEV_INITRD
446 {
1da177e4
LT		 447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
6d456111 451 .proc_handler = proc_dointvec,
1da177e4
LT		 452 },
453#endif
45807a1d 454 {
45807a1d
IM		 455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
6d456111 459 .proc_handler = proc_dointvec,
45807a1d 460 },
72c57ed5 461#ifdef CONFIG_SPARC
1da177e4 462 {
1da177e4
LT		 463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
6d456111 467 .proc_handler = proc_dostring,
1da177e4
LT		 468 },
469 {
1da177e4
LT		 470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
6d456111 474 .proc_handler = proc_dointvec,
1da177e4
LT		 475 },
476 {
1da177e4
LT		 477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
6d456111 481 .proc_handler = proc_dointvec,
1da177e4
LT		 482 },
483#endif
0871420f
DM		 484#ifdef CONFIG_SPARC64
485 {
0871420f
DM		 486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
6d456111 490 .proc_handler = proc_dointvec,
0871420f
DM		 491 },
492#endif
1da177e4
LT		 493#ifdef __hppa__
494 {
1da177e4
LT		 495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
6d456111 499 .proc_handler = proc_dointvec,
1da177e4
LT		 500 },
501 {
1da177e4
LT		 502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
6d456111 506 .proc_handler = proc_dointvec,
1da177e4
LT		 507 },
508#endif
509 {
1da177e4
LT		 510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
6d456111 514 .proc_handler = proc_dointvec,
1da177e4 515 },
606576ce 516#ifdef CONFIG_FUNCTION_TRACER
b0fc494f 517 {
b0fc494f
SR		 518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
6d456111 522 .proc_handler = ftrace_enable_sysctl,
b0fc494f
SR		 523 },
524#endif
f38f1d2a
SR		 525#ifdef CONFIG_STACK_TRACER
526 {
f38f1d2a
SR		 527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
6d456111 531 .proc_handler = stack_trace_sysctl,
f38f1d2a
SR		 532 },
533#endif
944ac425
SR		 534#ifdef CONFIG_TRACING
535 {
3299b4dd 536 .procname = "ftrace_dump_on_oops",
944ac425
SR		 537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
6d456111 540 .proc_handler = proc_dointvec,
944ac425
SR		 541 },
542#endif
a1ef5adb 543#ifdef CONFIG_MODULES
1da177e4 544 {
1da177e4
LT		 545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
6d456111 549 .proc_handler = proc_dostring,
1da177e4 550 },
3d43321b 551 {
3d43321b
KC		 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" */
6d456111 557 .proc_handler = proc_dointvec_minmax,
3d43321b
KC		 558 .extra1 = &one,
559 .extra2 = &one,
560 },
1da177e4 561#endif
94f17cd7 562#ifdef CONFIG_HOTPLUG
1da177e4 563 {
1da177e4 564 .procname = "hotplug",
312c004d
KS		 565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4 567 .mode = 0644,
6d456111 568 .proc_handler = proc_dostring,
1da177e4
LT		 569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
1da177e4
LT		 573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
6d456111 577 .proc_handler = proc_dointvec,
1da177e4
LT		 578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
1da177e4
LT		 582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
6d456111 586 .proc_handler = proc_dointvec,
1da177e4
LT		 587 },
588#endif
1da177e4
LT		 589#ifdef CONFIG_MAGIC_SYSRQ
590 {
1da177e4 591 .procname = "sysrq",
5d6f647f 592 .data = &__sysrq_enabled,
1da177e4
LT		 593 .maxlen = sizeof (int),
594 .mode = 0644,
97f5f0cd 595 .proc_handler = sysrq_sysctl_handler,
1da177e4
LT		 596 },
597#endif
d6f8ff73 598#ifdef CONFIG_PROC_SYSCTL
1da177e4 599 {
1da177e4 600 .procname = "cad_pid",
9ec52099 601 .data = NULL,
1da177e4
LT		 602 .maxlen = sizeof (int),
603 .mode = 0600,
6d456111 604 .proc_handler = proc_do_cad_pid,
1da177e4 605 },
d6f8ff73 606#endif
1da177e4 607 {
1da177e4
LT		 608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
6d456111 612 .proc_handler = proc_dointvec,
1da177e4
LT		 613 },
614 {
1da177e4
LT		 615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
17f60a7d
EP		 619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
1da177e4 624 {
1da177e4
LT		 625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
6d456111 629 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
1da177e4
LT		 634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
6d456111 638 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
347a8dc3 642#ifdef CONFIG_S390
1da177e4
LT		 643#ifdef CONFIG_MATHEMU
644 {
1da177e4
LT		 645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
6d456111 649 .proc_handler = proc_dointvec,
1da177e4 650 },
1da177e4
LT		 651#endif
652 {
1da177e4 653 .procname = "userprocess_debug",
ab3c68ee 654 .data = &show_unhandled_signals,
1da177e4
LT		 655 .maxlen = sizeof(int),
656 .mode = 0644,
6d456111 657 .proc_handler = proc_dointvec,
1da177e4
LT		 658 },
659#endif
660 {
1da177e4
LT		 661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
6d456111 665 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
1da177e4
LT		 670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
6d456111 674 .proc_handler = proc_dointvec,
1da177e4 675 },
7ef3d2fd
JP		 676#if defined CONFIG_PRINTK
677 {
7ef3d2fd
JP		 678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
6d456111 682 .proc_handler = proc_dointvec,
7ef3d2fd 683 },
1da177e4 684 {
1da177e4 685 .procname = "printk_ratelimit",
717115e1 686 .data = &printk_ratelimit_state.interval,
1da177e4
LT		 687 .maxlen = sizeof(int),
688 .mode = 0644,
6d456111 689 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 690 },
691 {
1da177e4 692 .procname = "printk_ratelimit_burst",
717115e1 693 .data = &printk_ratelimit_state.burst,
1da177e4
LT		 694 .maxlen = sizeof(int),
695 .mode = 0644,
6d456111 696 .proc_handler = proc_dointvec,
1da177e4 697 },
af91322e 698 {
af91322e
DY		 699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
6d456111 703 .proc_handler = proc_dointvec_minmax,
af91322e
DY		 704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
eaf06b24
DR		 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 },
455cd5ab
DR		 716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
bfdc0b49 721 .proc_handler = proc_dmesg_restrict,
455cd5ab
DR		 722 .extra1 = &zero,
723 .extra2 = &two,
724 },
df6e61d4 725#endif
1da177e4 726 {
1da177e4
LT		 727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
6d456111 731 .proc_handler = proc_dointvec,
1da177e4 732 },
58687acb 733#if defined(CONFIG_LOCKUP_DETECTOR)
504d7cf1 734 {
58687acb
DZ		 735 .procname = "watchdog",
736 .data = &watchdog_enabled,
504d7cf1
DZ		 737 .maxlen = sizeof (int),
738 .mode = 0644,
586692a5
MSB		 739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
58687acb
DZ		 742 },
743 {
744 .procname = "watchdog_thresh",
586692a5 745 .data = &watchdog_thresh,
58687acb
DZ		 746 .maxlen = sizeof(int),
747 .mode = 0644,
586692a5 748 .proc_handler = proc_dowatchdog,
58687acb
DZ		 749 .extra1 = &neg_one,
750 .extra2 = &sixty,
504d7cf1 751 },
2508ce18
DZ		 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 },
5dc30558
DZ		 761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
586692a5
MSB		 766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
5dc30558
DZ		 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 },
504d7cf1 779#endif
1da177e4 780#if defined(CONFIG_X86)
8da5adda 781 {
8da5adda
DZ		 782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
6d456111 786 .proc_handler = proc_dointvec,
8da5adda 787 },
5211a242 788 {
5211a242
KG		 789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
6d456111 793 .proc_handler = proc_dointvec,
5211a242 794 },
1da177e4 795 {
1da177e4
LT		 796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
6d456111 800 .proc_handler = proc_dointvec,
1da177e4 801 },
5031296c 802 {
5031296c
PA		 803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
6d456111 807 .proc_handler = proc_dointvec,
5031296c 808 },
0741f4d2 809 {
0741f4d2
CE		 810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
6d456111 814 .proc_handler = proc_dointvec,
0741f4d2 815 },
6e7c4025 816 {
6e7c4025
IM		 817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
6d456111 821 .proc_handler = proc_dointvec,
6e7c4025 822 },
1da177e4 823#endif
7a9166e3 824#if defined(CONFIG_MMU)
1da177e4 825 {
1da177e4
LT		 826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
6d456111 830 .proc_handler = proc_dointvec,
1da177e4 831 },
7a9166e3 832#endif
0152fb37 833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5 834 {
951f22d5
MS		 835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
6d456111 839 .proc_handler = proc_dointvec,
951f22d5 840 },
c255d844 841#endif
673d5b43 842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
c255d844 843 {
c255d844 844 .procname = "acpi_video_flags",
77afcf78 845 .data = &acpi_realmode_flags,
c255d844
PM		 846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
6d456111 848 .proc_handler = proc_doulongvec_minmax,
c255d844 849 },
d2b176ed
JS		 850#endif
851#ifdef CONFIG_IA64
852 {
d2b176ed
JS		 853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
6d456111 857 .proc_handler = proc_dointvec,
d2b176ed 858 },
88fc241f 859 {
88fc241f
DC		 860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
6d456111 864 .proc_handler = proc_dointvec,
88fc241f 865 },
bebfa101 866#endif
e162b39a
MSB		 867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
e162b39a
MSB		 869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
6d456111 873 .proc_handler = proc_dointvec_minmax,
e162b39a
MSB		 874 .extra1 = &zero,
875 .extra2 = &one,
876 },
82a1fcb9 877 {
82a1fcb9
IM		 878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
90739081 880 .maxlen = sizeof(unsigned long),
82a1fcb9 881 .mode = 0644,
6d456111 882 .proc_handler = proc_doulongvec_minmax,
82a1fcb9
IM		 883 },
884 {
82a1fcb9
IM		 885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
90739081 887 .maxlen = sizeof(unsigned long),
82a1fcb9 888 .mode = 0644,
6d456111 889 .proc_handler = proc_dohung_task_timeout_secs,
82a1fcb9
IM		 890 },
891 {
82a1fcb9
IM		 892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
90739081 894 .maxlen = sizeof(unsigned long),
82a1fcb9 895 .mode = 0644,
6d456111 896 .proc_handler = proc_doulongvec_minmax,
82a1fcb9 897 },
c4f3b63f 898#endif
bebfa101
AK		 899#ifdef CONFIG_COMPAT
900 {
bebfa101
AK		 901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
6d456111 905 .proc_handler = proc_dointvec,
bebfa101 906 },
951f22d5 907#endif
23f78d4a
IM		 908#ifdef CONFIG_RT_MUTEXES
909 {
23f78d4a
IM		 910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
6d456111 914 .proc_handler = proc_dointvec,
23f78d4a 915 },
5096add8 916#endif
10a0a8d4 917 {
10a0a8d4
JF		 918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
6d456111 922 .proc_handler = proc_dostring,
10a0a8d4 923 },
0b77f5bf
DH		 924#ifdef CONFIG_KEYS
925 {
0b77f5bf
DH		 926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
31a72bce
PM		 931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
31a72bce
PM		 933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
6d456111 937 .proc_handler = proc_dointvec,
31a72bce
PM		 938 },
939#endif
cdd6c482 940#ifdef CONFIG_PERF_EVENTS
aa4a2218
VW		 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 */
1ccd1549 947 {
cdd6c482
IM		 948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
1ccd1549 951 .mode = 0644,
6d456111 952 .proc_handler = proc_dointvec,
1ccd1549 953 },
c5078f78 954 {
cdd6c482
IM		 955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
c5078f78 958 .mode = 0644,
6d456111 959 .proc_handler = proc_dointvec,
c5078f78 960 },
a78ac325 961 {
cdd6c482
IM		 962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
a78ac325 965 .mode = 0644,
163ec435 966 .proc_handler = perf_proc_update_handler,
a78ac325 967 },
1ccd1549 968#endif
dfec072e
VN		 969#ifdef CONFIG_KMEMCHECK
970 {
dfec072e
VN		 971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
6d456111 975 .proc_handler = proc_dointvec,
dfec072e
VN		 976 },
977#endif
cb684b5b 978#ifdef CONFIG_BLOCK
5e605b64 979 {
5e605b64
JA		 980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
6d456111 984 .proc_handler = proc_dointvec,
5e605b64 985 },
cb684b5b 986#endif
6fce56ec 987 { }
1da177e4
LT		 988};
989
d8217f07 990static struct ctl_table vm_table[] = {
1da177e4 991 {
1da177e4
LT		 992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
cb16e95f
PH		 996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
1da177e4 999 },
fadd8fbd 1000 {
fadd8fbd
KH		 1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
cb16e95f
PH		 1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
fadd8fbd 1008 },
fe071d7e 1009 {
fe071d7e
DR		 1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
6d456111 1014 .proc_handler = proc_dointvec,
fe071d7e 1015 },
fef1bdd6 1016 {
fef1bdd6
DR		 1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
6d456111 1021 .proc_handler = proc_dointvec,
fef1bdd6 1022 },
1da177e4 1023 {
1da177e4
LT		 1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
6d456111 1028 .proc_handler = proc_dointvec,
1da177e4
LT		 1029 },
1030 {
1da177e4
LT		 1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
cb16e95f
PH		 1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1da177e4
LT		 1037 },
1038 {
1da177e4
LT		 1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
6d456111 1043 .proc_handler = dirty_background_ratio_handler,
1da177e4
LT		 1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
2da02997 1047 {
2da02997
DR		 1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
6d456111 1052 .proc_handler = dirty_background_bytes_handler,
fc3501d4 1053 .extra1 = &one_ul,
2da02997 1054 },
1da177e4 1055 {
1da177e4
LT		 1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
6d456111 1060 .proc_handler = dirty_ratio_handler,
1da177e4
LT		 1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
2da02997 1064 {
2da02997
DR		 1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
6d456111 1069 .proc_handler = dirty_bytes_handler,
9e4a5bda 1070 .extra1 = &dirty_bytes_min,
2da02997 1071 },
1da177e4 1072 {
1da177e4 1073 .procname = "dirty_writeback_centisecs",
f6ef9438
BS		 1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1da177e4 1076 .mode = 0644,
6d456111 1077 .proc_handler = dirty_writeback_centisecs_handler,
1da177e4
LT		 1078 },
1079 {
1da177e4 1080 .procname = "dirty_expire_centisecs",
f6ef9438
BS		 1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1da177e4 1083 .mode = 0644,
cb16e95f
PH		 1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1da177e4
LT		 1086 },
1087 {
1da177e4
LT		 1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
6d456111 1092 .proc_handler = proc_dointvec,
1da177e4
LT		 1093 },
1094 {
1da177e4
LT		 1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
6d456111 1099 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
06808b08 1104 {
1da177e4 1105 .procname = "nr_hugepages",
e5ff2159 1106 .data = NULL,
1da177e4
LT		 1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
6d456111 1109 .proc_handler = hugetlb_sysctl_handler,
1da177e4
LT		 1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
06808b08
LS		 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
1da177e4 1124 {
1da177e4
LT		 1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
6d456111 1129 .proc_handler = proc_dointvec,
1da177e4 1130 },
396faf03 1131 {
396faf03
MG		 1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
6d456111 1136 .proc_handler = hugetlb_treat_movable_handler,
396faf03 1137 },
d1c3fb1f 1138 {
d1c3fb1f 1139 .procname = "nr_overcommit_hugepages",
e5ff2159
AK		 1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
d1c3fb1f 1142 .mode = 0644,
6d456111 1143 .proc_handler = hugetlb_overcommit_handler,
e5ff2159
AK		 1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
d1c3fb1f 1146 },
1da177e4
LT		 1147#endif
1148 {
1da177e4
LT		 1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
6d456111 1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1da177e4 1154 },
9d0243bc 1155 {
9d0243bc
AM		 1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
cb16e95f
PH		 1161 .extra1 = &one,
1162 .extra2 = &three,
9d0243bc 1163 },
76ab0f53
MG		 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 },
5e771905
MG		 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
76ab0f53 1182#endif /* CONFIG_COMPACTION */
1da177e4 1183 {
1da177e4
LT		 1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
6d456111 1188 .proc_handler = min_free_kbytes_sysctl_handler,
1da177e4
LT		 1189 .extra1 = &zero,
1190 },
8ad4b1fb 1191 {
8ad4b1fb
RS		 1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
6d456111 1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
8ad4b1fb
RS		 1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1da177e4
LT		 1199#ifdef CONFIG_MMU
1200 {
1da177e4
LT		 1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
3e26120c 1205 .proc_handler = proc_dointvec_minmax,
70da2340 1206 .extra1 = &zero,
1da177e4 1207 },
dd8632a1
PM		 1208#else
1209 {
dd8632a1
PM		 1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
6d456111 1214 .proc_handler = proc_dointvec_minmax,
dd8632a1
PM		 1215 .extra1 = &zero,
1216 },
1da177e4
LT		 1217#endif
1218 {
1da177e4
LT		 1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
6d456111 1223 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 1224 },
1225 {
1da177e4
LT		 1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
6d456111 1230 .proc_handler = proc_dointvec,
1da177e4
LT		 1231 .extra1 = &zero,
1232 },
1233 {
1da177e4
LT		 1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
6d456111 1238 .proc_handler = proc_dointvec,
1da177e4
LT		 1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1da177e4
LT		 1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
6d456111 1247 .proc_handler = proc_dointvec,
1da177e4
LT		 1248 .extra1 = &zero,
1249 },
1250#endif
1743660b
CL		 1251#ifdef CONFIG_NUMA
1252 {
1743660b
CL		 1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
6d456111 1257 .proc_handler = proc_dointvec,
c84db23c 1258 .extra1 = &zero,
1743660b 1259 },
9614634f 1260 {
9614634f
CL		 1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
6d456111 1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
9614634f
CL		 1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
0ff38490 1269 {
0ff38490
CL		 1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
6d456111 1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
0ff38490
CL		 1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
e6e5494c 1278#endif
77461ab3
CL		 1279#ifdef CONFIG_SMP
1280 {
77461ab3
CL		 1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
6d456111 1285 .proc_handler = proc_dointvec_jiffies,
77461ab3
CL		 1286 },
1287#endif
6e141546 1288#ifdef CONFIG_MMU
ed032189 1289 {
ed032189 1290 .procname = "mmap_min_addr",
788084ab
EP		 1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
ed032189 1293 .mode = 0644,
6d456111 1294 .proc_handler = mmap_min_addr_handler,
ed032189 1295 },
6e141546 1296#endif
f0c0b2b8
KH		 1297#ifdef CONFIG_NUMA
1298 {
f0c0b2b8
KH		 1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
6d456111 1303 .proc_handler = numa_zonelist_order_handler,
f0c0b2b8
KH		 1304 },
1305#endif
2b8232ce 1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
5c36e657 1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
e6e5494c 1308 {
e6e5494c
IM		 1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
6d456111 1313 .proc_handler = proc_dointvec,
e6e5494c
IM		 1314 .extra1 = &zero,
1315 },
1da177e4 1316#endif
195cf453
BG		 1317#ifdef CONFIG_HIGHMEM
1318 {
195cf453
BG		 1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
6d456111 1323 .proc_handler = proc_dointvec_minmax,
195cf453
BG		 1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
4be6f6bb 1328 {
4be6f6bb
PZ		 1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
6d456111 1333 .proc_handler = scan_unevictable_handler,
4be6f6bb 1334 },
6a46079c
AK		 1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
6a46079c
AK		 1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
6d456111 1341 .proc_handler = proc_dointvec_minmax,
6a46079c
AK		 1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
6a46079c
AK		 1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
6d456111 1350 .proc_handler = proc_dointvec_minmax,
6a46079c
AK		 1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
6fce56ec 1355 { }
1da177e4
LT		 1356};
1357
2abc26fc 1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
d8217f07 1359static struct ctl_table binfmt_misc_table[] = {
6fce56ec 1360 { }
2abc26fc
EB		 1361};
1362#endif
1363
d8217f07 1364static struct ctl_table fs_table[] = {
1da177e4 1365 {
1da177e4
LT		 1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
cffbc8aa 1370 .proc_handler = proc_nr_inodes,
1da177e4
LT		 1371 },
1372 {
1da177e4
LT		 1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
cffbc8aa 1377 .proc_handler = proc_nr_inodes,
1da177e4
LT		 1378 },
1379 {
1da177e4
LT		 1380 .procname = "file-nr",
1381 .data = &files_stat,
518de9b3 1382 .maxlen = sizeof(files_stat),
1da177e4 1383 .mode = 0444,
6d456111 1384 .proc_handler = proc_nr_files,
1da177e4
LT		 1385 },
1386 {
1da177e4
LT		 1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
518de9b3 1389 .maxlen = sizeof(files_stat.max_files),
1da177e4 1390 .mode = 0644,
518de9b3 1391 .proc_handler = proc_doulongvec_minmax,
1da177e4 1392 },
9cfe015a 1393 {
9cfe015a
ED		 1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
6d456111 1398 .proc_handler = proc_dointvec_minmax,
eceea0b3
AV		 1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
9cfe015a 1401 },
1da177e4 1402 {
1da177e4
LT		 1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
312d3ca8 1407 .proc_handler = proc_nr_dentry,
1da177e4
LT		 1408 },
1409 {
1da177e4
LT		 1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
6d456111 1414 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1da177e4
LT		 1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
6d456111 1423 .proc_handler = proc_dointvec_minmax,
1da177e4
LT		 1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
bfcd17a6 1427#ifdef CONFIG_FILE_LOCKING
1da177e4 1428 {
1da177e4
LT		 1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
6d456111 1433 .proc_handler = proc_dointvec,
1da177e4 1434 },
bfcd17a6 1435#endif
1da177e4
LT		 1436#ifdef CONFIG_DNOTIFY
1437 {
1da177e4
LT		 1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
6d456111 1442 .proc_handler = proc_dointvec,
1da177e4
LT		 1443 },
1444#endif
1445#ifdef CONFIG_MMU
bfcd17a6 1446#ifdef CONFIG_FILE_LOCKING
1da177e4 1447 {
1da177e4
LT		 1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
6d456111 1452 .proc_handler = proc_dointvec,
1da177e4 1453 },
bfcd17a6 1454#endif
ebf3f09c 1455#ifdef CONFIG_AIO
1da177e4 1456 {
1da177e4
LT		 1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
6d456111 1461 .proc_handler = proc_doulongvec_minmax,
1da177e4
LT		 1462 },
1463 {
1da177e4
LT		 1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
6d456111 1468 .proc_handler = proc_doulongvec_minmax,
1da177e4 1469 },
ebf3f09c 1470#endif /* CONFIG_AIO */
2d9048e2 1471#ifdef CONFIG_INOTIFY_USER
0399cb08 1472 {
0399cb08
RL		 1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
7ef9964e
DL		 1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1da177e4 1485#endif
d6e71144 1486 {
d6e71144
AC		 1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
6d456111 1491 .proc_handler = proc_dointvec_minmax,
8e654fba
MW		 1492 .extra1 = &zero,
1493 .extra2 = &two,
d6e71144 1494 },
2abc26fc
EB		 1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
2abc26fc
EB		 1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
b492e95b 1502 {
ff9da691
JA		 1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
b492e95b
JA		 1505 .maxlen = sizeof(int),
1506 .mode = 0644,
ff9da691
JA		 1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
b492e95b 1509 },
6fce56ec 1510 { }
1da177e4
LT		 1511};
1512
d8217f07 1513static struct ctl_table debug_table[] = {
ab3c68ee 1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
571d76ac 1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
abd4f750 1516 {
abd4f750
MAS		 1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
b2be84df
MH		 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 },
abd4f750 1534#endif
6fce56ec 1535 { }
1da177e4
LT		 1536};
1537
d8217f07 1538static struct ctl_table dev_table[] = {
6fce56ec 1539 { }
0eeca283 1540};
1da177e4 1541
330d57fb
AV		 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);
f7e6ced4
AV		 1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
330d57fb
AV		 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
f7e6ced4
AV		 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)
a95cded3 1597 kfree_rcu(head, rcu);
f7e6ced4
AV		 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
805b5d5e
EB		 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
73455092
AV		 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
e51b6ba0
EB		 1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
805b5d5e 1632{
73455092
AV		 1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
e51b6ba0
EB		 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;
805b5d5e
EB		 1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
e51b6ba0 1644
805b5d5e
EB		 1645 spin_lock(&sysctl_lock);
1646 if (prev) {
e51b6ba0 1647 head = prev;
805b5d5e
EB		 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:
e51b6ba0 1661 root = head->root;
805b5d5e 1662 tmp = tmp->next;
e51b6ba0
EB		 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;
805b5d5e 1675 }
e51b6ba0 1676out:
805b5d5e
EB		 1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
e51b6ba0
EB		 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
1da177e4 1693/*
1ff007eb 1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1da177e4
LT		 1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
76aac0e9 1700 if (!current_euid())
1da177e4
LT		 1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
e6305c43 1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1da177e4
LT		 1705 return 0;
1706 return -EACCES;
1707}
1708
d7321cd6 1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1da177e4 1710{
d7321cd6
PE		 1711 int mode;
1712
d7321cd6
PE		 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);
1da177e4
LT		 1719}
1720
d912b0cc
EB		 1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
2315ffa0 1723 for (; table->procname; table++) {
d912b0cc
EB		 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);
88f458e4 1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
b3bd3de6 1734 sysctl_check_table(current->nsproxy, root_table);
88f458e4 1735#endif
d912b0cc
EB		 1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
bfbcf034
AV		 1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
ae7edecc
AV		 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)
bfbcf034 1749 return NULL;
ae7edecc
AV		 1750
1751 /* ... and nothing else */
2315ffa0 1752 if (branch[1].procname)
bfbcf034 1753 return NULL;
ae7edecc
AV		 1754
1755 /* table should contain subdirectory with the same name */
2315ffa0 1756 for (p = table; p->procname; p++) {
ae7edecc
AV		 1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
bfbcf034 1760 return p;
ae7edecc 1761 }
bfbcf034 1762 return NULL;
ae7edecc
AV		 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;
bfbcf034 1769 struct ctl_table *next;
ae7edecc
AV		 1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
bfbcf034 1773 while ((next = is_branch_in(by, to)) != NULL) {
ae7edecc
AV		 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;
bfbcf034 1779 to = next->child;
ae7edecc
AV		 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
1da177e4 1789/**
e51b6ba0
EB		 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
29e796fd 1793 * @path: The path to the directory the sysctl table is in.
1da177e4 1794 * @table: the top-level table structure
1da177e4
LT		 1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
29e796fd 1797 * array. A completely 0 filled entry terminates the table.
1da177e4 1798 *
d8217f07 1799 * The members of the &struct ctl_table structure are used as follows:
1da177e4 1800 *
1da177e4
LT		 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 *
1da177e4
LT		 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 *
1da177e4
LT		 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 */
e51b6ba0
EB		 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)
1da177e4 1846{
29e796fd
EB		 1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
ae7edecc 1850 struct ctl_table_set *set;
29e796fd
EB		 1851
1852 /* Count the path components */
2315ffa0 1853 for (npath = 0; path[npath].procname; ++npath)
29e796fd
EB		 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
2315ffa0 1859 * for this, the second will be the sentinel (procname == 0).
29e796fd
EB		 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)
1da177e4 1867 return NULL;
29e796fd
EB		 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;
29e796fd
EB		 1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
23eb06de 1884 header->ctl_table_arg = table;
29e796fd
EB		 1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
e51b6ba0 1889 header->root = root;
29e796fd 1890 sysctl_set_parent(NULL, header->ctl_table);
f7e6ced4 1891 header->count = 1;
88f458e4 1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
e51b6ba0 1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
29e796fd 1894 kfree(header);
fc6cd25b
EB		 1895 return NULL;
1896 }
88f458e4 1897#endif
330d57fb 1898 spin_lock(&sysctl_lock);
73455092 1899 header->set = lookup_header_set(root, namespaces);
ae7edecc
AV		 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++;
73455092 1912 list_add_tail(&header->ctl_entry, &header->set->list);
330d57fb 1913 spin_unlock(&sysctl_lock);
29e796fd
EB		 1914
1915 return header;
1916}
1917
e51b6ba0
EB		 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
29e796fd
EB		 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);
1da177e4
LT		 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{
330d57fb 1960 might_sleep();
f1dad166
PE		 1961
1962 if (header == NULL)
1963 return;
1964
330d57fb
AV		 1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
ae7edecc
AV		 1967 if (!--header->parent->count) {
1968 WARN_ON(1);
a95cded3 1969 kfree_rcu(header->parent, rcu);
ae7edecc 1970 }
f7e6ced4 1971 if (!--header->count)
a95cded3 1972 kfree_rcu(header, rcu);
330d57fb 1973 spin_unlock(&sysctl_lock);
1da177e4
LT		 1974}
1975
9043476f
AV		 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
73455092
AV		 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
b89a8171 2000#else /* !CONFIG_SYSCTL */
d8217f07 2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
b89a8171
EB		 2002{
2003 return NULL;
2004}
2005
29e796fd
EB		 2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
b89a8171
EB		 2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
73455092
AV		 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
f7e6ced4
AV		 2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
b89a8171
EB		 2026#endif /* CONFIG_SYSCTL */
2027
1da177e4
LT		 2028/*
2029 * /proc/sys support
2030 */
2031
b89a8171 2032#ifdef CONFIG_PROC_SYSCTL
1da177e4 2033
b1ba4ddd 2034static int _proc_do_string(void* data, int maxlen, int write,
8d65af78 2035 void __user *buffer,
b1ba4ddd 2036 size_t *lenp, loff_t *ppos)
1da177e4
LT		 2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
8d060877
ON		 2041
2042 if (!data || !maxlen || !*lenp) {
1da177e4
LT		 2043 *lenp = 0;
2044 return 0;
2045 }
8d060877 2046
1da177e4
LT		 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 }
f5dd3d6f
SV		 2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
1da177e4 2060 return -EFAULT;
f5dd3d6f 2061 ((char *) data)[len] = 0;
1da177e4
LT		 2062 *ppos += *lenp;
2063 } else {
f5dd3d6f
SV		 2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
8d060877
ON		 2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
1da177e4
LT		 2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
f5dd3d6f 2079 if(copy_to_user(buffer, data, len))
1da177e4
LT		 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
f5dd3d6f
SV		 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
f5dd3d6f
SV		 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 */
8d65af78 2109int proc_dostring(struct ctl_table *table, int write,
f5dd3d6f
SV		 2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
8d65af78 2112 return _proc_do_string(table->data, table->maxlen, write,
f5dd3d6f
SV		 2113 buffer, lenp, ppos);
2114}
2115
00b7c339
AW		 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
9f977fb7
OP		 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
00b7c339
AW		 2135#define TMPBUFLEN 22
2136/**
0fc377bd 2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
00b7c339 2138 *
0fc377bd
RD		 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
00b7c339 2146 *
0fc377bd
RD		 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.
00b7c339
AW		 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;
1da177e4 2190
00b7c339
AW		 2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
0fc377bd 2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
00b7c339 2202 *
0fc377bd
RD		 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
00b7c339 2207 *
0fc377bd
RD		 2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
00b7c339
AW		 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}
1da177e4 2240
00b7c339 2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT		 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) {
00b7c339 2250 *negp = true;
1da177e4
LT		 2251 *lvalp = (unsigned long)-val;
2252 } else {
00b7c339 2253 *negp = false;
1da177e4
LT		 2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
00b7c339
AW		 2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
d8217f07 2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
8d65af78 2263 int write, void __user *buffer,
fcfbd547 2264 size_t *lenp, loff_t *ppos,
00b7c339 2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1da177e4
LT		 2266 int write, void *data),
2267 void *data)
2268{
00b7c339
AW		 2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
1da177e4 2273
00b7c339 2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1da177e4
LT		 2275 *lenp = 0;
2276 return 0;
2277 }
2278
fcfbd547 2279 i = (int *) tbl_data;
1da177e4
LT		 2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
00b7c339
AW		 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
1da177e4 2300 for (; left && vleft--; i++, first=0) {
00b7c339
AW		 2301 unsigned long lval;
2302 bool neg;
1da177e4 2303
00b7c339
AW		 2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
1da177e4 2306
563b0467
O		 2307 if (!left)
2308 break;
00b7c339
AW		 2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
1da177e4 2313 break;
00b7c339
AW		 2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
1da177e4 2316 break;
00b7c339 2317 }
1da177e4 2318 } else {
00b7c339
AW		 2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
1da177e4 2323 if (!first)
00b7c339
AW		 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)
1da177e4 2329 break;
1da177e4
LT		 2330 }
2331 }
2332
00b7c339
AW		 2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
563b0467 2335 if (write && !err && left)
00b7c339
AW		 2336 left -= proc_skip_spaces(&kbuf);
2337free:
1da177e4 2338 if (write) {
00b7c339
AW		 2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
1da177e4 2342 }
1da177e4
LT		 2343 *lenp -= left;
2344 *ppos += *lenp;
00b7c339 2345 return err;
1da177e4
LT		 2346}
2347
8d65af78 2348static int do_proc_dointvec(struct ctl_table *table, int write,
fcfbd547 2349 void __user *buffer, size_t *lenp, loff_t *ppos,
00b7c339 2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
fcfbd547
KK		 2351 int write, void *data),
2352 void *data)
2353{
8d65af78 2354 return __do_proc_dointvec(table->data, table, write,
fcfbd547
KK		 2355 buffer, lenp, ppos, conv, data);
2356}
2357
1da177e4
LT		 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
1da177e4
LT		 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 */
8d65af78 2371int proc_dointvec(struct ctl_table *table, int write,
1da177e4
LT		 2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
8d65af78 2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT		 2375 NULL,NULL);
2376}
2377
34f5a398 2378/*
25ddbb18
AK		 2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
34f5a398 2381 */
8d65af78 2382static int proc_taint(struct ctl_table *table, int write,
34f5a398
TT		 2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
25ddbb18
AK		 2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
34f5a398 2388
91fcd412 2389 if (write && !capable(CAP_SYS_ADMIN))
34f5a398
TT		 2390 return -EPERM;
2391
25ddbb18
AK		 2392 t = *table;
2393 t.data = &tmptaint;
8d65af78 2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
25ddbb18
AK		 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;
34f5a398
TT		 2411}
2412
bfdc0b49
RW		 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
1da177e4
LT		 2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
00b7c339
AW		 2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
1da177e4
LT		 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) {
00b7c339 2443 *negp = true;
1da177e4
LT		 2444 *lvalp = (unsigned long)-val;
2445 } else {
00b7c339 2446 *negp = false;
1da177e4
LT		 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
1da177e4
LT		 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 */
8d65af78 2469int proc_dointvec_minmax(struct ctl_table *table, int write,
1da177e4
LT		 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 };
8d65af78 2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1da177e4
LT		 2477 do_proc_dointvec_minmax_conv, &param);
2478}
2479
d8217f07 2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
1da177e4
LT		 2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
00b7c339
AW		 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)) {
1da177e4
LT		 2493 *lenp = 0;
2494 return 0;
2495 }
00b7c339 2496
fcfbd547 2497 i = (unsigned long *) data;
1da177e4
LT		 2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
00b7c339
AW		 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
27b3d80a 2517 for (; left && vleft--; i++, first = 0) {
00b7c339
AW		 2518 unsigned long val;
2519
1da177e4 2520 if (write) {
00b7c339
AW		 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)
1da177e4
LT		 2529 break;
2530 if (neg)
1da177e4
LT		 2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
00b7c339 2536 val = convdiv * (*i) / convmul;
1da177e4 2537 if (!first)
00b7c339
AW		 2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
1da177e4
LT		 2542 }
2543 }
2544
00b7c339
AW		 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:
1da177e4 2550 if (write) {
00b7c339
AW		 2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
1da177e4 2554 }
1da177e4
LT		 2555 *lenp -= left;
2556 *ppos += *lenp;
00b7c339 2557 return err;
1da177e4
LT		 2558}
2559
d8217f07 2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
fcfbd547
KK		 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,
8d65af78 2567 buffer, lenp, ppos, convmul, convdiv);
fcfbd547
KK		 2568}
2569
1da177e4
LT		 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
1da177e4
LT		 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 */
8d65af78 2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
1da177e4
LT		 2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
8d65af78 2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1da177e4
LT		 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
1da177e4
LT		 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 */
d8217f07 2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1da177e4
LT		 2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
8d65af78 2613 return do_proc_doulongvec_minmax(table, write, buffer,
1da177e4
LT		 2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
00b7c339 2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT		 2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
cba9f33d
BS		 2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
1da177e4
LT		 2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
00b7c339 2630 *negp = true;
1da177e4
LT		 2631 lval = (unsigned long)-val;
2632 } else {
00b7c339 2633 *negp = false;
1da177e4
LT		 2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
00b7c339 2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT		 2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
cba9f33d
BS		 2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
1da177e4
LT		 2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
00b7c339 2653 *negp = true;
1da177e4
LT		 2654 lval = (unsigned long)-val;
2655 } else {
00b7c339 2656 *negp = false;
1da177e4
LT		 2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
00b7c339 2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT		 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) {
00b7c339 2674 *negp = true;
1da177e4
LT		 2675 lval = (unsigned long)-val;
2676 } else {
00b7c339 2677 *negp = false;
1da177e4
LT		 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
1da177e4
LT		 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 */
8d65af78 2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
8d65af78 2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT		 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
1da177e4
LT		 2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
1e5d5331 2713 * @ppos: pointer to the file position
1da177e4
LT		 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 */
8d65af78 2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2723 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
8d65af78 2725 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT		 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
1da177e4
LT		 2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
67be2dd1
MW		 2735 * @ppos: file position
2736 * @ppos: the current position in the file
1da177e4
LT		 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 */
8d65af78 2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
8d65af78 2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1da177e4
LT		 2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
8d65af78 2752static int proc_do_cad_pid(struct ctl_table *table, int write,
9ec52099
CLG		 2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
6c5f3e7b 2759 tmp = pid_vnr(cad_pid);
9ec52099 2760
8d65af78 2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
9ec52099
CLG		 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
9f977fb7
OP		 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
55610500 2925#else /* CONFIG_PROC_SYSCTL */
1da177e4 2926
8d65af78 2927int proc_dostring(struct ctl_table *table, int write,
1da177e4
LT		 2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
8d65af78 2933int proc_dointvec(struct ctl_table *table, int write,
1da177e4 2934 void __user *buffer, size_t *lenp, loff_t *ppos)
1da177e4
LT		 2935{
2936 return -ENOSYS;
2937}
2938
8d65af78 2939int proc_dointvec_minmax(struct ctl_table *table, int write,
1da177e4
LT		 2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
8d65af78 2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
8d65af78 2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
8d65af78 2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1da177e4
LT		 2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
8d65af78 2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
1da177e4
LT		 2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
d8217f07 2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1da177e4
LT		 2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
55610500 2977#endif /* CONFIG_PROC_SYSCTL */
1da177e4 2978
1da177e4
LT		 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);
29e796fd 2992EXPORT_SYMBOL(register_sysctl_paths);
1da177e4 2993EXPORT_SYMBOL(unregister_sysctl_table);
Linux 6.x block layer and io_uring tree(s)