]> bbs.cooldavid.org Git - net-next-2.6.git/blame - kernel/sysctl.c
dnotify: move dir_notify_enable declaration
[net-next-2.6.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>
1da177e4 27#include <linux/proc_fs.h>
72c2d582 28#include <linux/security.h>
1da177e4 29#include <linux/ctype.h>
dfec072e 30#include <linux/kmemcheck.h>
62239ac2 31#include <linux/fs.h>
1da177e4
LT
32#include <linux/init.h>
33#include <linux/kernel.h>
0296b228 34#include <linux/kobject.h>
20380731 35#include <linux/net.h>
1da177e4
LT
36#include <linux/sysrq.h>
37#include <linux/highuid.h>
38#include <linux/writeback.h>
3fff4c42 39#include <linux/ratelimit.h>
76ab0f53 40#include <linux/compaction.h>
1da177e4 41#include <linux/hugetlb.h>
1da177e4 42#include <linux/initrd.h>
0b77f5bf 43#include <linux/key.h>
1da177e4
LT
44#include <linux/times.h>
45#include <linux/limits.h>
46#include <linux/dcache.h>
6e006701 47#include <linux/dnotify.h>
1da177e4 48#include <linux/syscalls.h>
c748e134 49#include <linux/vmstat.h>
c255d844
PM
50#include <linux/nfs_fs.h>
51#include <linux/acpi.h>
10a0a8d4 52#include <linux/reboot.h>
b0fc494f 53#include <linux/ftrace.h>
12e22c5e 54#include <linux/slow-work.h>
cdd6c482 55#include <linux/perf_event.h>
b2be84df 56#include <linux/kprobes.h>
b492e95b 57#include <linux/pipe_fs_i.h>
1da177e4
LT
58
59#include <asm/uaccess.h>
60#include <asm/processor.h>
61
29cbc78b
AK
62#ifdef CONFIG_X86
63#include <asm/nmi.h>
0741f4d2 64#include <asm/stacktrace.h>
6e7c4025 65#include <asm/io.h>
29cbc78b 66#endif
c55b7c3e
DY
67#ifdef CONFIG_BSD_PROCESS_ACCT
68#include <linux/acct.h>
69#endif
4f0e056f
DY
70#ifdef CONFIG_RT_MUTEXES
71#include <linux/rtmutex.h>
72#endif
2edf5e49
DY
73#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
74#include <linux/lockdep.h>
75#endif
15485a46
DY
76#ifdef CONFIG_CHR_DEV_SG
77#include <scsi/sg.h>
78#endif
29cbc78b 79
7058cb02 80
1da177e4
LT
81#if defined(CONFIG_SYSCTL)
82
83/* External variables not in a header file. */
1da177e4
LT
84extern int sysctl_overcommit_memory;
85extern int sysctl_overcommit_ratio;
fadd8fbd 86extern int sysctl_panic_on_oom;
fe071d7e 87extern int sysctl_oom_kill_allocating_task;
fef1bdd6 88extern int sysctl_oom_dump_tasks;
1da177e4 89extern int max_threads;
1da177e4 90extern int core_uses_pid;
d6e71144 91extern int suid_dumpable;
1da177e4 92extern char core_pattern[];
a293980c 93extern unsigned int core_pipe_limit;
1da177e4
LT
94extern int pid_max;
95extern int min_free_kbytes;
1da177e4 96extern int pid_max_min, pid_max_max;
9d0243bc 97extern int sysctl_drop_caches;
8ad4b1fb 98extern int percpu_pagelist_fraction;
bebfa101 99extern int compat_log;
9745512c 100extern int latencytop_enabled;
eceea0b3 101extern int sysctl_nr_open_min, sysctl_nr_open_max;
dd8632a1
PM
102#ifndef CONFIG_MMU
103extern int sysctl_nr_trim_pages;
104#endif
cb684b5b 105#ifdef CONFIG_BLOCK
5e605b64 106extern int blk_iopoll_enabled;
cb684b5b 107#endif
1da177e4 108
c4f3b63f 109/* Constants used for minimum and maximum */
195cf453 110#ifdef CONFIG_DETECT_SOFTLOCKUP
c4f3b63f 111static int sixty = 60;
9383d967 112static int neg_one = -1;
c4f3b63f
RT
113#endif
114
c4f3b63f 115static int zero;
cd5f9a4c
LT
116static int __maybe_unused one = 1;
117static int __maybe_unused two = 2;
fc3501d4 118static unsigned long one_ul = 1;
c4f3b63f 119static int one_hundred = 100;
af91322e
DY
120#ifdef CONFIG_PRINTK
121static int ten_thousand = 10000;
122#endif
c4f3b63f 123
9e4a5bda
AR
124/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
125static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
126
1da177e4
LT
127/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
128static int maxolduid = 65535;
129static int minolduid;
8ad4b1fb 130static int min_percpu_pagelist_fract = 8;
1da177e4
LT
131
132static int ngroups_max = NGROUPS_MAX;
133
72c57ed5 134#ifdef CONFIG_SPARC
17f04fbb 135#include <asm/system.h>
1da177e4
LT
136#endif
137
0871420f
DM
138#ifdef CONFIG_SPARC64
139extern int sysctl_tsb_ratio;
140#endif
141
1da177e4
LT
142#ifdef __hppa__
143extern int pwrsw_enabled;
144extern int unaligned_enabled;
145#endif
146
347a8dc3 147#ifdef CONFIG_S390
1da177e4
LT
148#ifdef CONFIG_MATHEMU
149extern int sysctl_ieee_emulation_warnings;
150#endif
151extern int sysctl_userprocess_debug;
951f22d5 152extern int spin_retry;
1da177e4
LT
153#endif
154
d2b176ed
JS
155#ifdef CONFIG_IA64
156extern int no_unaligned_warning;
88fc241f 157extern int unaligned_dump_stack;
d2b176ed
JS
158#endif
159
3fff4c42
IM
160extern struct ratelimit_state printk_ratelimit_state;
161
d6f8ff73 162#ifdef CONFIG_PROC_SYSCTL
8d65af78 163static int proc_do_cad_pid(struct ctl_table *table, int write,
9ec52099 164 void __user *buffer, size_t *lenp, loff_t *ppos);
8d65af78 165static int proc_taint(struct ctl_table *table, int write,
34f5a398 166 void __user *buffer, size_t *lenp, loff_t *ppos);
d6f8ff73 167#endif
9ec52099 168
97f5f0cd
DT
169#ifdef CONFIG_MAGIC_SYSRQ
170static int __sysrq_enabled; /* Note: sysrq code ises it's own private copy */
171
172static int sysrq_sysctl_handler(ctl_table *table, int write,
173 void __user *buffer, size_t *lenp,
174 loff_t *ppos)
175{
176 int error;
177
178 error = proc_dointvec(table, write, buffer, lenp, ppos);
179 if (error)
180 return error;
181
182 if (write)
183 sysrq_toggle_support(__sysrq_enabled);
184
185 return 0;
186}
187
188#endif
189
d8217f07 190static struct ctl_table root_table[];
e51b6ba0
EB
191static struct ctl_table_root sysctl_table_root;
192static struct ctl_table_header root_table_header = {
b380b0d4 193 .count = 1,
e51b6ba0 194 .ctl_table = root_table,
73455092 195 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
e51b6ba0 196 .root = &sysctl_table_root,
73455092 197 .set = &sysctl_table_root.default_set,
e51b6ba0
EB
198};
199static struct ctl_table_root sysctl_table_root = {
200 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
73455092 201 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
e51b6ba0 202};
1da177e4 203
d8217f07
EB
204static struct ctl_table kern_table[];
205static struct ctl_table vm_table[];
206static struct ctl_table fs_table[];
207static struct ctl_table debug_table[];
208static struct ctl_table dev_table[];
209extern struct ctl_table random_table[];
2d9048e2 210#ifdef CONFIG_INOTIFY_USER
d8217f07 211extern struct ctl_table inotify_table[];
0399cb08 212#endif
7ef9964e
DL
213#ifdef CONFIG_EPOLL
214extern struct ctl_table epoll_table[];
215#endif
1da177e4
LT
216
217#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
218int sysctl_legacy_va_layout;
219#endif
220
1da177e4
LT
221/* The default sysctl tables: */
222
d8217f07 223static struct ctl_table root_table[] = {
1da177e4 224 {
1da177e4
LT
225 .procname = "kernel",
226 .mode = 0555,
227 .child = kern_table,
228 },
229 {
1da177e4
LT
230 .procname = "vm",
231 .mode = 0555,
232 .child = vm_table,
233 },
1da177e4 234 {
1da177e4
LT
235 .procname = "fs",
236 .mode = 0555,
237 .child = fs_table,
238 },
239 {
1da177e4
LT
240 .procname = "debug",
241 .mode = 0555,
242 .child = debug_table,
243 },
244 {
1da177e4
LT
245 .procname = "dev",
246 .mode = 0555,
247 .child = dev_table,
248 },
2be7fe07
AM
249/*
250 * NOTE: do not add new entries to this table unless you have read
251 * Documentation/sysctl/ctl_unnumbered.txt
252 */
6fce56ec 253 { }
1da177e4
LT
254};
255
77e54a1f 256#ifdef CONFIG_SCHED_DEBUG
73c4efd2
ED
257static int min_sched_granularity_ns = 100000; /* 100 usecs */
258static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
259static int min_wakeup_granularity_ns; /* 0 usecs */
260static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
1983a922
CE
261static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
262static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
acb4a848
CE
263static int min_sched_shares_ratelimit = 100000; /* 100 usec */
264static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
77e54a1f
IM
265#endif
266
5e771905
MG
267#ifdef CONFIG_COMPACTION
268static int min_extfrag_threshold;
269static int max_extfrag_threshold = 1000;
270#endif
271
d8217f07 272static struct ctl_table kern_table[] = {
2bba22c5 273 {
2bba22c5
MG
274 .procname = "sched_child_runs_first",
275 .data = &sysctl_sched_child_runs_first,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
6d456111 278 .proc_handler = proc_dointvec,
2bba22c5 279 },
77e54a1f
IM
280#ifdef CONFIG_SCHED_DEBUG
281 {
b2be5e96
PZ
282 .procname = "sched_min_granularity_ns",
283 .data = &sysctl_sched_min_granularity,
77e54a1f
IM
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
702a7c76 286 .proc_handler = sched_proc_update_handler,
b2be5e96
PZ
287 .extra1 = &min_sched_granularity_ns,
288 .extra2 = &max_sched_granularity_ns,
77e54a1f 289 },
21805085 290 {
21805085
PZ
291 .procname = "sched_latency_ns",
292 .data = &sysctl_sched_latency,
293 .maxlen = sizeof(unsigned int),
294 .mode = 0644,
702a7c76 295 .proc_handler = sched_proc_update_handler,
21805085
PZ
296 .extra1 = &min_sched_granularity_ns,
297 .extra2 = &max_sched_granularity_ns,
298 },
77e54a1f 299 {
77e54a1f
IM
300 .procname = "sched_wakeup_granularity_ns",
301 .data = &sysctl_sched_wakeup_granularity,
302 .maxlen = sizeof(unsigned int),
303 .mode = 0644,
702a7c76 304 .proc_handler = sched_proc_update_handler,
77e54a1f
IM
305 .extra1 = &min_wakeup_granularity_ns,
306 .extra2 = &max_wakeup_granularity_ns,
307 },
2398f2c6 308 {
2398f2c6
PZ
309 .procname = "sched_shares_ratelimit",
310 .data = &sysctl_sched_shares_ratelimit,
311 .maxlen = sizeof(unsigned int),
312 .mode = 0644,
702a7c76 313 .proc_handler = sched_proc_update_handler,
acb4a848
CE
314 .extra1 = &min_sched_shares_ratelimit,
315 .extra2 = &max_sched_shares_ratelimit,
2398f2c6 316 },
1983a922 317 {
1983a922
CE
318 .procname = "sched_tunable_scaling",
319 .data = &sysctl_sched_tunable_scaling,
320 .maxlen = sizeof(enum sched_tunable_scaling),
321 .mode = 0644,
702a7c76 322 .proc_handler = sched_proc_update_handler,
1983a922
CE
323 .extra1 = &min_sched_tunable_scaling,
324 .extra2 = &max_sched_tunable_scaling,
2398f2c6 325 },
ffda12a1 326 {
ffda12a1
PZ
327 .procname = "sched_shares_thresh",
328 .data = &sysctl_sched_shares_thresh,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
6d456111 331 .proc_handler = proc_dointvec_minmax,
ffda12a1
PZ
332 .extra1 = &zero,
333 },
da84d961 334 {
da84d961
IM
335 .procname = "sched_migration_cost",
336 .data = &sysctl_sched_migration_cost,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
6d456111 339 .proc_handler = proc_dointvec,
da84d961 340 },
b82d9fdd 341 {
b82d9fdd
PZ
342 .procname = "sched_nr_migrate",
343 .data = &sysctl_sched_nr_migrate,
344 .maxlen = sizeof(unsigned int),
fa85ae24 345 .mode = 0644,
6d456111 346 .proc_handler = proc_dointvec,
fa85ae24 347 },
e9e9250b 348 {
e9e9250b
PZ
349 .procname = "sched_time_avg",
350 .data = &sysctl_sched_time_avg,
351 .maxlen = sizeof(unsigned int),
352 .mode = 0644,
6d456111 353 .proc_handler = proc_dointvec,
e9e9250b 354 },
cd1bb94b 355 {
cd1bb94b
AB
356 .procname = "timer_migration",
357 .data = &sysctl_timer_migration,
358 .maxlen = sizeof(unsigned int),
359 .mode = 0644,
6d456111 360 .proc_handler = proc_dointvec_minmax,
bfdb4d9f
AB
361 .extra1 = &zero,
362 .extra2 = &one,
fa85ae24 363 },
1fc84aaa 364#endif
9f0c1e56 365 {
9f0c1e56
PZ
366 .procname = "sched_rt_period_us",
367 .data = &sysctl_sched_rt_period,
368 .maxlen = sizeof(unsigned int),
369 .mode = 0644,
6d456111 370 .proc_handler = sched_rt_handler,
9f0c1e56
PZ
371 },
372 {
9f0c1e56
PZ
373 .procname = "sched_rt_runtime_us",
374 .data = &sysctl_sched_rt_runtime,
375 .maxlen = sizeof(int),
376 .mode = 0644,
6d456111 377 .proc_handler = sched_rt_handler,
9f0c1e56 378 },
1799e35d 379 {
1799e35d
IM
380 .procname = "sched_compat_yield",
381 .data = &sysctl_sched_compat_yield,
382 .maxlen = sizeof(unsigned int),
383 .mode = 0644,
6d456111 384 .proc_handler = proc_dointvec,
1799e35d 385 },
f20786ff
PZ
386#ifdef CONFIG_PROVE_LOCKING
387 {
f20786ff
PZ
388 .procname = "prove_locking",
389 .data = &prove_locking,
390 .maxlen = sizeof(int),
391 .mode = 0644,
6d456111 392 .proc_handler = proc_dointvec,
f20786ff
PZ
393 },
394#endif
395#ifdef CONFIG_LOCK_STAT
396 {
f20786ff
PZ
397 .procname = "lock_stat",
398 .data = &lock_stat,
399 .maxlen = sizeof(int),
400 .mode = 0644,
6d456111 401 .proc_handler = proc_dointvec,
f20786ff 402 },
77e54a1f 403#endif
1da177e4 404 {
1da177e4
LT
405 .procname = "panic",
406 .data = &panic_timeout,
407 .maxlen = sizeof(int),
408 .mode = 0644,
6d456111 409 .proc_handler = proc_dointvec,
1da177e4
LT
410 },
411 {
1da177e4
LT
412 .procname = "core_uses_pid",
413 .data = &core_uses_pid,
414 .maxlen = sizeof(int),
415 .mode = 0644,
6d456111 416 .proc_handler = proc_dointvec,
1da177e4
LT
417 },
418 {
1da177e4
LT
419 .procname = "core_pattern",
420 .data = core_pattern,
71ce92f3 421 .maxlen = CORENAME_MAX_SIZE,
1da177e4 422 .mode = 0644,
6d456111 423 .proc_handler = proc_dostring,
1da177e4 424 },
a293980c 425 {
a293980c
NH
426 .procname = "core_pipe_limit",
427 .data = &core_pipe_limit,
428 .maxlen = sizeof(unsigned int),
429 .mode = 0644,
6d456111 430 .proc_handler = proc_dointvec,
a293980c 431 },
34f5a398 432#ifdef CONFIG_PROC_SYSCTL
1da177e4 433 {
1da177e4 434 .procname = "tainted",
25ddbb18 435 .maxlen = sizeof(long),
34f5a398 436 .mode = 0644,
6d456111 437 .proc_handler = proc_taint,
1da177e4 438 },
34f5a398 439#endif
9745512c
AV
440#ifdef CONFIG_LATENCYTOP
441 {
442 .procname = "latencytop",
443 .data = &latencytop_enabled,
444 .maxlen = sizeof(int),
445 .mode = 0644,
6d456111 446 .proc_handler = proc_dointvec,
9745512c
AV
447 },
448#endif
1da177e4
LT
449#ifdef CONFIG_BLK_DEV_INITRD
450 {
1da177e4
LT
451 .procname = "real-root-dev",
452 .data = &real_root_dev,
453 .maxlen = sizeof(int),
454 .mode = 0644,
6d456111 455 .proc_handler = proc_dointvec,
1da177e4
LT
456 },
457#endif
45807a1d 458 {
45807a1d
IM
459 .procname = "print-fatal-signals",
460 .data = &print_fatal_signals,
461 .maxlen = sizeof(int),
462 .mode = 0644,
6d456111 463 .proc_handler = proc_dointvec,
45807a1d 464 },
72c57ed5 465#ifdef CONFIG_SPARC
1da177e4 466 {
1da177e4
LT
467 .procname = "reboot-cmd",
468 .data = reboot_command,
469 .maxlen = 256,
470 .mode = 0644,
6d456111 471 .proc_handler = proc_dostring,
1da177e4
LT
472 },
473 {
1da177e4
LT
474 .procname = "stop-a",
475 .data = &stop_a_enabled,
476 .maxlen = sizeof (int),
477 .mode = 0644,
6d456111 478 .proc_handler = proc_dointvec,
1da177e4
LT
479 },
480 {
1da177e4
LT
481 .procname = "scons-poweroff",
482 .data = &scons_pwroff,
483 .maxlen = sizeof (int),
484 .mode = 0644,
6d456111 485 .proc_handler = proc_dointvec,
1da177e4
LT
486 },
487#endif
0871420f
DM
488#ifdef CONFIG_SPARC64
489 {
0871420f
DM
490 .procname = "tsb-ratio",
491 .data = &sysctl_tsb_ratio,
492 .maxlen = sizeof (int),
493 .mode = 0644,
6d456111 494 .proc_handler = proc_dointvec,
0871420f
DM
495 },
496#endif
1da177e4
LT
497#ifdef __hppa__
498 {
1da177e4
LT
499 .procname = "soft-power",
500 .data = &pwrsw_enabled,
501 .maxlen = sizeof (int),
502 .mode = 0644,
6d456111 503 .proc_handler = proc_dointvec,
1da177e4
LT
504 },
505 {
1da177e4
LT
506 .procname = "unaligned-trap",
507 .data = &unaligned_enabled,
508 .maxlen = sizeof (int),
509 .mode = 0644,
6d456111 510 .proc_handler = proc_dointvec,
1da177e4
LT
511 },
512#endif
513 {
1da177e4
LT
514 .procname = "ctrl-alt-del",
515 .data = &C_A_D,
516 .maxlen = sizeof(int),
517 .mode = 0644,
6d456111 518 .proc_handler = proc_dointvec,
1da177e4 519 },
606576ce 520#ifdef CONFIG_FUNCTION_TRACER
b0fc494f 521 {
b0fc494f
SR
522 .procname = "ftrace_enabled",
523 .data = &ftrace_enabled,
524 .maxlen = sizeof(int),
525 .mode = 0644,
6d456111 526 .proc_handler = ftrace_enable_sysctl,
b0fc494f
SR
527 },
528#endif
f38f1d2a
SR
529#ifdef CONFIG_STACK_TRACER
530 {
f38f1d2a
SR
531 .procname = "stack_tracer_enabled",
532 .data = &stack_tracer_enabled,
533 .maxlen = sizeof(int),
534 .mode = 0644,
6d456111 535 .proc_handler = stack_trace_sysctl,
f38f1d2a
SR
536 },
537#endif
944ac425
SR
538#ifdef CONFIG_TRACING
539 {
3299b4dd 540 .procname = "ftrace_dump_on_oops",
944ac425
SR
541 .data = &ftrace_dump_on_oops,
542 .maxlen = sizeof(int),
543 .mode = 0644,
6d456111 544 .proc_handler = proc_dointvec,
944ac425
SR
545 },
546#endif
a1ef5adb 547#ifdef CONFIG_MODULES
1da177e4 548 {
1da177e4
LT
549 .procname = "modprobe",
550 .data = &modprobe_path,
551 .maxlen = KMOD_PATH_LEN,
552 .mode = 0644,
6d456111 553 .proc_handler = proc_dostring,
1da177e4 554 },
3d43321b 555 {
3d43321b
KC
556 .procname = "modules_disabled",
557 .data = &modules_disabled,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 /* only handle a transition from default "0" to "1" */
6d456111 561 .proc_handler = proc_dointvec_minmax,
3d43321b
KC
562 .extra1 = &one,
563 .extra2 = &one,
564 },
1da177e4 565#endif
57ae2508 566#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
1da177e4 567 {
1da177e4 568 .procname = "hotplug",
312c004d
KS
569 .data = &uevent_helper,
570 .maxlen = UEVENT_HELPER_PATH_LEN,
1da177e4 571 .mode = 0644,
6d456111 572 .proc_handler = proc_dostring,
1da177e4
LT
573 },
574#endif
575#ifdef CONFIG_CHR_DEV_SG
576 {
1da177e4
LT
577 .procname = "sg-big-buff",
578 .data = &sg_big_buff,
579 .maxlen = sizeof (int),
580 .mode = 0444,
6d456111 581 .proc_handler = proc_dointvec,
1da177e4
LT
582 },
583#endif
584#ifdef CONFIG_BSD_PROCESS_ACCT
585 {
1da177e4
LT
586 .procname = "acct",
587 .data = &acct_parm,
588 .maxlen = 3*sizeof(int),
589 .mode = 0644,
6d456111 590 .proc_handler = proc_dointvec,
1da177e4
LT
591 },
592#endif
1da177e4
LT
593#ifdef CONFIG_MAGIC_SYSRQ
594 {
1da177e4 595 .procname = "sysrq",
5d6f647f 596 .data = &__sysrq_enabled,
1da177e4
LT
597 .maxlen = sizeof (int),
598 .mode = 0644,
97f5f0cd 599 .proc_handler = sysrq_sysctl_handler,
1da177e4
LT
600 },
601#endif
d6f8ff73 602#ifdef CONFIG_PROC_SYSCTL
1da177e4 603 {
1da177e4 604 .procname = "cad_pid",
9ec52099 605 .data = NULL,
1da177e4
LT
606 .maxlen = sizeof (int),
607 .mode = 0600,
6d456111 608 .proc_handler = proc_do_cad_pid,
1da177e4 609 },
d6f8ff73 610#endif
1da177e4 611 {
1da177e4
LT
612 .procname = "threads-max",
613 .data = &max_threads,
614 .maxlen = sizeof(int),
615 .mode = 0644,
6d456111 616 .proc_handler = proc_dointvec,
1da177e4
LT
617 },
618 {
1da177e4
LT
619 .procname = "random",
620 .mode = 0555,
621 .child = random_table,
622 },
1da177e4 623 {
1da177e4
LT
624 .procname = "overflowuid",
625 .data = &overflowuid,
626 .maxlen = sizeof(int),
627 .mode = 0644,
6d456111 628 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
629 .extra1 = &minolduid,
630 .extra2 = &maxolduid,
631 },
632 {
1da177e4
LT
633 .procname = "overflowgid",
634 .data = &overflowgid,
635 .maxlen = sizeof(int),
636 .mode = 0644,
6d456111 637 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
638 .extra1 = &minolduid,
639 .extra2 = &maxolduid,
640 },
347a8dc3 641#ifdef CONFIG_S390
1da177e4
LT
642#ifdef CONFIG_MATHEMU
643 {
1da177e4
LT
644 .procname = "ieee_emulation_warnings",
645 .data = &sysctl_ieee_emulation_warnings,
646 .maxlen = sizeof(int),
647 .mode = 0644,
6d456111 648 .proc_handler = proc_dointvec,
1da177e4 649 },
1da177e4
LT
650#endif
651 {
1da177e4 652 .procname = "userprocess_debug",
ab3c68ee 653 .data = &show_unhandled_signals,
1da177e4
LT
654 .maxlen = sizeof(int),
655 .mode = 0644,
6d456111 656 .proc_handler = proc_dointvec,
1da177e4
LT
657 },
658#endif
659 {
1da177e4
LT
660 .procname = "pid_max",
661 .data = &pid_max,
662 .maxlen = sizeof (int),
663 .mode = 0644,
6d456111 664 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
665 .extra1 = &pid_max_min,
666 .extra2 = &pid_max_max,
667 },
668 {
1da177e4
LT
669 .procname = "panic_on_oops",
670 .data = &panic_on_oops,
671 .maxlen = sizeof(int),
672 .mode = 0644,
6d456111 673 .proc_handler = proc_dointvec,
1da177e4 674 },
7ef3d2fd
JP
675#if defined CONFIG_PRINTK
676 {
7ef3d2fd
JP
677 .procname = "printk",
678 .data = &console_loglevel,
679 .maxlen = 4*sizeof(int),
680 .mode = 0644,
6d456111 681 .proc_handler = proc_dointvec,
7ef3d2fd 682 },
1da177e4 683 {
1da177e4 684 .procname = "printk_ratelimit",
717115e1 685 .data = &printk_ratelimit_state.interval,
1da177e4
LT
686 .maxlen = sizeof(int),
687 .mode = 0644,
6d456111 688 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT
689 },
690 {
1da177e4 691 .procname = "printk_ratelimit_burst",
717115e1 692 .data = &printk_ratelimit_state.burst,
1da177e4
LT
693 .maxlen = sizeof(int),
694 .mode = 0644,
6d456111 695 .proc_handler = proc_dointvec,
1da177e4 696 },
af91322e 697 {
af91322e
DY
698 .procname = "printk_delay",
699 .data = &printk_delay_msec,
700 .maxlen = sizeof(int),
701 .mode = 0644,
6d456111 702 .proc_handler = proc_dointvec_minmax,
af91322e
DY
703 .extra1 = &zero,
704 .extra2 = &ten_thousand,
705 },
7ef3d2fd 706#endif
1da177e4 707 {
1da177e4
LT
708 .procname = "ngroups_max",
709 .data = &ngroups_max,
710 .maxlen = sizeof (int),
711 .mode = 0444,
6d456111 712 .proc_handler = proc_dointvec,
1da177e4
LT
713 },
714#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
715 {
1da177e4
LT
716 .procname = "unknown_nmi_panic",
717 .data = &unknown_nmi_panic,
718 .maxlen = sizeof (int),
719 .mode = 0644,
6d456111 720 .proc_handler = proc_dointvec,
1da177e4 721 },
407984f1 722 {
407984f1
DZ
723 .procname = "nmi_watchdog",
724 .data = &nmi_watchdog_enabled,
725 .maxlen = sizeof (int),
726 .mode = 0644,
6d456111 727 .proc_handler = proc_nmi_enabled,
1da177e4
LT
728 },
729#endif
730#if defined(CONFIG_X86)
8da5adda 731 {
8da5adda
DZ
732 .procname = "panic_on_unrecovered_nmi",
733 .data = &panic_on_unrecovered_nmi,
734 .maxlen = sizeof(int),
735 .mode = 0644,
6d456111 736 .proc_handler = proc_dointvec,
8da5adda 737 },
5211a242 738 {
5211a242
KG
739 .procname = "panic_on_io_nmi",
740 .data = &panic_on_io_nmi,
741 .maxlen = sizeof(int),
742 .mode = 0644,
6d456111 743 .proc_handler = proc_dointvec,
5211a242 744 },
1da177e4 745 {
1da177e4
LT
746 .procname = "bootloader_type",
747 .data = &bootloader_type,
748 .maxlen = sizeof (int),
749 .mode = 0444,
6d456111 750 .proc_handler = proc_dointvec,
1da177e4 751 },
5031296c 752 {
5031296c
PA
753 .procname = "bootloader_version",
754 .data = &bootloader_version,
755 .maxlen = sizeof (int),
756 .mode = 0444,
6d456111 757 .proc_handler = proc_dointvec,
5031296c 758 },
0741f4d2 759 {
0741f4d2
CE
760 .procname = "kstack_depth_to_print",
761 .data = &kstack_depth_to_print,
762 .maxlen = sizeof(int),
763 .mode = 0644,
6d456111 764 .proc_handler = proc_dointvec,
0741f4d2 765 },
6e7c4025 766 {
6e7c4025
IM
767 .procname = "io_delay_type",
768 .data = &io_delay_type,
769 .maxlen = sizeof(int),
770 .mode = 0644,
6d456111 771 .proc_handler = proc_dointvec,
6e7c4025 772 },
1da177e4 773#endif
7a9166e3 774#if defined(CONFIG_MMU)
1da177e4 775 {
1da177e4
LT
776 .procname = "randomize_va_space",
777 .data = &randomize_va_space,
778 .maxlen = sizeof(int),
779 .mode = 0644,
6d456111 780 .proc_handler = proc_dointvec,
1da177e4 781 },
7a9166e3 782#endif
0152fb37 783#if defined(CONFIG_S390) && defined(CONFIG_SMP)
951f22d5 784 {
951f22d5
MS
785 .procname = "spin_retry",
786 .data = &spin_retry,
787 .maxlen = sizeof (int),
788 .mode = 0644,
6d456111 789 .proc_handler = proc_dointvec,
951f22d5 790 },
c255d844 791#endif
673d5b43 792#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
c255d844 793 {
c255d844 794 .procname = "acpi_video_flags",
77afcf78 795 .data = &acpi_realmode_flags,
c255d844
PM
796 .maxlen = sizeof (unsigned long),
797 .mode = 0644,
6d456111 798 .proc_handler = proc_doulongvec_minmax,
c255d844 799 },
d2b176ed
JS
800#endif
801#ifdef CONFIG_IA64
802 {
d2b176ed
JS
803 .procname = "ignore-unaligned-usertrap",
804 .data = &no_unaligned_warning,
805 .maxlen = sizeof (int),
806 .mode = 0644,
6d456111 807 .proc_handler = proc_dointvec,
d2b176ed 808 },
88fc241f 809 {
88fc241f
DC
810 .procname = "unaligned-dump-stack",
811 .data = &unaligned_dump_stack,
812 .maxlen = sizeof (int),
813 .mode = 0644,
6d456111 814 .proc_handler = proc_dointvec,
88fc241f 815 },
bebfa101 816#endif
c4f3b63f 817#ifdef CONFIG_DETECT_SOFTLOCKUP
9c44bc03 818 {
9c44bc03
IM
819 .procname = "softlockup_panic",
820 .data = &softlockup_panic,
821 .maxlen = sizeof(int),
822 .mode = 0644,
6d456111 823 .proc_handler = proc_dointvec_minmax,
9c44bc03
IM
824 .extra1 = &zero,
825 .extra2 = &one,
826 },
c4f3b63f 827 {
c4f3b63f
RT
828 .procname = "softlockup_thresh",
829 .data = &softlockup_thresh,
9383d967 830 .maxlen = sizeof(int),
c4f3b63f 831 .mode = 0644,
6d456111 832 .proc_handler = proc_dosoftlockup_thresh,
9383d967 833 .extra1 = &neg_one,
c4f3b63f
RT
834 .extra2 = &sixty,
835 },
e162b39a
MSB
836#endif
837#ifdef CONFIG_DETECT_HUNG_TASK
838 {
e162b39a
MSB
839 .procname = "hung_task_panic",
840 .data = &sysctl_hung_task_panic,
841 .maxlen = sizeof(int),
842 .mode = 0644,
6d456111 843 .proc_handler = proc_dointvec_minmax,
e162b39a
MSB
844 .extra1 = &zero,
845 .extra2 = &one,
846 },
82a1fcb9 847 {
82a1fcb9
IM
848 .procname = "hung_task_check_count",
849 .data = &sysctl_hung_task_check_count,
90739081 850 .maxlen = sizeof(unsigned long),
82a1fcb9 851 .mode = 0644,
6d456111 852 .proc_handler = proc_doulongvec_minmax,
82a1fcb9
IM
853 },
854 {
82a1fcb9
IM
855 .procname = "hung_task_timeout_secs",
856 .data = &sysctl_hung_task_timeout_secs,
90739081 857 .maxlen = sizeof(unsigned long),
82a1fcb9 858 .mode = 0644,
6d456111 859 .proc_handler = proc_dohung_task_timeout_secs,
82a1fcb9
IM
860 },
861 {
82a1fcb9
IM
862 .procname = "hung_task_warnings",
863 .data = &sysctl_hung_task_warnings,
90739081 864 .maxlen = sizeof(unsigned long),
82a1fcb9 865 .mode = 0644,
6d456111 866 .proc_handler = proc_doulongvec_minmax,
82a1fcb9 867 },
c4f3b63f 868#endif
bebfa101
AK
869#ifdef CONFIG_COMPAT
870 {
bebfa101
AK
871 .procname = "compat-log",
872 .data = &compat_log,
873 .maxlen = sizeof (int),
874 .mode = 0644,
6d456111 875 .proc_handler = proc_dointvec,
bebfa101 876 },
951f22d5 877#endif
23f78d4a
IM
878#ifdef CONFIG_RT_MUTEXES
879 {
23f78d4a
IM
880 .procname = "max_lock_depth",
881 .data = &max_lock_depth,
882 .maxlen = sizeof(int),
883 .mode = 0644,
6d456111 884 .proc_handler = proc_dointvec,
23f78d4a 885 },
5096add8 886#endif
10a0a8d4 887 {
10a0a8d4
JF
888 .procname = "poweroff_cmd",
889 .data = &poweroff_cmd,
890 .maxlen = POWEROFF_CMD_PATH_LEN,
891 .mode = 0644,
6d456111 892 .proc_handler = proc_dostring,
10a0a8d4 893 },
0b77f5bf
DH
894#ifdef CONFIG_KEYS
895 {
0b77f5bf
DH
896 .procname = "keys",
897 .mode = 0555,
898 .child = key_sysctls,
899 },
900#endif
31a72bce
PM
901#ifdef CONFIG_RCU_TORTURE_TEST
902 {
31a72bce
PM
903 .procname = "rcutorture_runnable",
904 .data = &rcutorture_runnable,
905 .maxlen = sizeof(int),
906 .mode = 0644,
6d456111 907 .proc_handler = proc_dointvec,
31a72bce
PM
908 },
909#endif
12e22c5e
DH
910#ifdef CONFIG_SLOW_WORK
911 {
12e22c5e
DH
912 .procname = "slow-work",
913 .mode = 0555,
914 .child = slow_work_sysctls,
915 },
916#endif
cdd6c482 917#ifdef CONFIG_PERF_EVENTS
1ccd1549 918 {
cdd6c482
IM
919 .procname = "perf_event_paranoid",
920 .data = &sysctl_perf_event_paranoid,
921 .maxlen = sizeof(sysctl_perf_event_paranoid),
1ccd1549 922 .mode = 0644,
6d456111 923 .proc_handler = proc_dointvec,
1ccd1549 924 },
c5078f78 925 {
cdd6c482
IM
926 .procname = "perf_event_mlock_kb",
927 .data = &sysctl_perf_event_mlock,
928 .maxlen = sizeof(sysctl_perf_event_mlock),
c5078f78 929 .mode = 0644,
6d456111 930 .proc_handler = proc_dointvec,
c5078f78 931 },
a78ac325 932 {
cdd6c482
IM
933 .procname = "perf_event_max_sample_rate",
934 .data = &sysctl_perf_event_sample_rate,
935 .maxlen = sizeof(sysctl_perf_event_sample_rate),
a78ac325 936 .mode = 0644,
6d456111 937 .proc_handler = proc_dointvec,
a78ac325 938 },
1ccd1549 939#endif
dfec072e
VN
940#ifdef CONFIG_KMEMCHECK
941 {
dfec072e
VN
942 .procname = "kmemcheck",
943 .data = &kmemcheck_enabled,
944 .maxlen = sizeof(int),
945 .mode = 0644,
6d456111 946 .proc_handler = proc_dointvec,
dfec072e
VN
947 },
948#endif
cb684b5b 949#ifdef CONFIG_BLOCK
5e605b64 950 {
5e605b64
JA
951 .procname = "blk_iopoll",
952 .data = &blk_iopoll_enabled,
953 .maxlen = sizeof(int),
954 .mode = 0644,
6d456111 955 .proc_handler = proc_dointvec,
5e605b64 956 },
cb684b5b 957#endif
ed2c12f3
AM
958/*
959 * NOTE: do not add new entries to this table unless you have read
960 * Documentation/sysctl/ctl_unnumbered.txt
961 */
6fce56ec 962 { }
1da177e4
LT
963};
964
d8217f07 965static struct ctl_table vm_table[] = {
1da177e4 966 {
1da177e4
LT
967 .procname = "overcommit_memory",
968 .data = &sysctl_overcommit_memory,
969 .maxlen = sizeof(sysctl_overcommit_memory),
970 .mode = 0644,
6d456111 971 .proc_handler = proc_dointvec,
1da177e4 972 },
fadd8fbd 973 {
fadd8fbd
KH
974 .procname = "panic_on_oom",
975 .data = &sysctl_panic_on_oom,
976 .maxlen = sizeof(sysctl_panic_on_oom),
977 .mode = 0644,
6d456111 978 .proc_handler = proc_dointvec,
fadd8fbd 979 },
fe071d7e 980 {
fe071d7e
DR
981 .procname = "oom_kill_allocating_task",
982 .data = &sysctl_oom_kill_allocating_task,
983 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
984 .mode = 0644,
6d456111 985 .proc_handler = proc_dointvec,
fe071d7e 986 },
fef1bdd6 987 {
fef1bdd6
DR
988 .procname = "oom_dump_tasks",
989 .data = &sysctl_oom_dump_tasks,
990 .maxlen = sizeof(sysctl_oom_dump_tasks),
991 .mode = 0644,
6d456111 992 .proc_handler = proc_dointvec,
fef1bdd6 993 },
1da177e4 994 {
1da177e4
LT
995 .procname = "overcommit_ratio",
996 .data = &sysctl_overcommit_ratio,
997 .maxlen = sizeof(sysctl_overcommit_ratio),
998 .mode = 0644,
6d456111 999 .proc_handler = proc_dointvec,
1da177e4
LT
1000 },
1001 {
1da177e4
LT
1002 .procname = "page-cluster",
1003 .data = &page_cluster,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
6d456111 1006 .proc_handler = proc_dointvec,
1da177e4
LT
1007 },
1008 {
1da177e4
LT
1009 .procname = "dirty_background_ratio",
1010 .data = &dirty_background_ratio,
1011 .maxlen = sizeof(dirty_background_ratio),
1012 .mode = 0644,
6d456111 1013 .proc_handler = dirty_background_ratio_handler,
1da177e4
LT
1014 .extra1 = &zero,
1015 .extra2 = &one_hundred,
1016 },
2da02997 1017 {
2da02997
DR
1018 .procname = "dirty_background_bytes",
1019 .data = &dirty_background_bytes,
1020 .maxlen = sizeof(dirty_background_bytes),
1021 .mode = 0644,
6d456111 1022 .proc_handler = dirty_background_bytes_handler,
fc3501d4 1023 .extra1 = &one_ul,
2da02997 1024 },
1da177e4 1025 {
1da177e4
LT
1026 .procname = "dirty_ratio",
1027 .data = &vm_dirty_ratio,
1028 .maxlen = sizeof(vm_dirty_ratio),
1029 .mode = 0644,
6d456111 1030 .proc_handler = dirty_ratio_handler,
1da177e4
LT
1031 .extra1 = &zero,
1032 .extra2 = &one_hundred,
1033 },
2da02997 1034 {
2da02997
DR
1035 .procname = "dirty_bytes",
1036 .data = &vm_dirty_bytes,
1037 .maxlen = sizeof(vm_dirty_bytes),
1038 .mode = 0644,
6d456111 1039 .proc_handler = dirty_bytes_handler,
9e4a5bda 1040 .extra1 = &dirty_bytes_min,
2da02997 1041 },
1da177e4 1042 {
1da177e4 1043 .procname = "dirty_writeback_centisecs",
f6ef9438
BS
1044 .data = &dirty_writeback_interval,
1045 .maxlen = sizeof(dirty_writeback_interval),
1da177e4 1046 .mode = 0644,
6d456111 1047 .proc_handler = dirty_writeback_centisecs_handler,
1da177e4
LT
1048 },
1049 {
1da177e4 1050 .procname = "dirty_expire_centisecs",
f6ef9438
BS
1051 .data = &dirty_expire_interval,
1052 .maxlen = sizeof(dirty_expire_interval),
1da177e4 1053 .mode = 0644,
6d456111 1054 .proc_handler = proc_dointvec,
1da177e4
LT
1055 },
1056 {
1da177e4
LT
1057 .procname = "nr_pdflush_threads",
1058 .data = &nr_pdflush_threads,
1059 .maxlen = sizeof nr_pdflush_threads,
1060 .mode = 0444 /* read-only*/,
6d456111 1061 .proc_handler = proc_dointvec,
1da177e4
LT
1062 },
1063 {
1da177e4
LT
1064 .procname = "swappiness",
1065 .data = &vm_swappiness,
1066 .maxlen = sizeof(vm_swappiness),
1067 .mode = 0644,
6d456111 1068 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
1069 .extra1 = &zero,
1070 .extra2 = &one_hundred,
1071 },
1072#ifdef CONFIG_HUGETLB_PAGE
06808b08 1073 {
1da177e4 1074 .procname = "nr_hugepages",
e5ff2159 1075 .data = NULL,
1da177e4
LT
1076 .maxlen = sizeof(unsigned long),
1077 .mode = 0644,
6d456111 1078 .proc_handler = hugetlb_sysctl_handler,
1da177e4
LT
1079 .extra1 = (void *)&hugetlb_zero,
1080 .extra2 = (void *)&hugetlb_infinity,
06808b08
LS
1081 },
1082#ifdef CONFIG_NUMA
1083 {
1084 .procname = "nr_hugepages_mempolicy",
1085 .data = NULL,
1086 .maxlen = sizeof(unsigned long),
1087 .mode = 0644,
1088 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1089 .extra1 = (void *)&hugetlb_zero,
1090 .extra2 = (void *)&hugetlb_infinity,
1091 },
1092#endif
1da177e4 1093 {
1da177e4
LT
1094 .procname = "hugetlb_shm_group",
1095 .data = &sysctl_hugetlb_shm_group,
1096 .maxlen = sizeof(gid_t),
1097 .mode = 0644,
6d456111 1098 .proc_handler = proc_dointvec,
1da177e4 1099 },
396faf03 1100 {
396faf03
MG
1101 .procname = "hugepages_treat_as_movable",
1102 .data = &hugepages_treat_as_movable,
1103 .maxlen = sizeof(int),
1104 .mode = 0644,
6d456111 1105 .proc_handler = hugetlb_treat_movable_handler,
396faf03 1106 },
d1c3fb1f 1107 {
d1c3fb1f 1108 .procname = "nr_overcommit_hugepages",
e5ff2159
AK
1109 .data = NULL,
1110 .maxlen = sizeof(unsigned long),
d1c3fb1f 1111 .mode = 0644,
6d456111 1112 .proc_handler = hugetlb_overcommit_handler,
e5ff2159
AK
1113 .extra1 = (void *)&hugetlb_zero,
1114 .extra2 = (void *)&hugetlb_infinity,
d1c3fb1f 1115 },
1da177e4
LT
1116#endif
1117 {
1da177e4
LT
1118 .procname = "lowmem_reserve_ratio",
1119 .data = &sysctl_lowmem_reserve_ratio,
1120 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1121 .mode = 0644,
6d456111 1122 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1da177e4 1123 },
9d0243bc 1124 {
9d0243bc
AM
1125 .procname = "drop_caches",
1126 .data = &sysctl_drop_caches,
1127 .maxlen = sizeof(int),
1128 .mode = 0644,
1129 .proc_handler = drop_caches_sysctl_handler,
9d0243bc 1130 },
76ab0f53
MG
1131#ifdef CONFIG_COMPACTION
1132 {
1133 .procname = "compact_memory",
1134 .data = &sysctl_compact_memory,
1135 .maxlen = sizeof(int),
1136 .mode = 0200,
1137 .proc_handler = sysctl_compaction_handler,
1138 },
5e771905
MG
1139 {
1140 .procname = "extfrag_threshold",
1141 .data = &sysctl_extfrag_threshold,
1142 .maxlen = sizeof(int),
1143 .mode = 0644,
1144 .proc_handler = sysctl_extfrag_handler,
1145 .extra1 = &min_extfrag_threshold,
1146 .extra2 = &max_extfrag_threshold,
1147 },
1148
76ab0f53 1149#endif /* CONFIG_COMPACTION */
1da177e4 1150 {
1da177e4
LT
1151 .procname = "min_free_kbytes",
1152 .data = &min_free_kbytes,
1153 .maxlen = sizeof(min_free_kbytes),
1154 .mode = 0644,
6d456111 1155 .proc_handler = min_free_kbytes_sysctl_handler,
1da177e4
LT
1156 .extra1 = &zero,
1157 },
8ad4b1fb 1158 {
8ad4b1fb
RS
1159 .procname = "percpu_pagelist_fraction",
1160 .data = &percpu_pagelist_fraction,
1161 .maxlen = sizeof(percpu_pagelist_fraction),
1162 .mode = 0644,
6d456111 1163 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
8ad4b1fb
RS
1164 .extra1 = &min_percpu_pagelist_fract,
1165 },
1da177e4
LT
1166#ifdef CONFIG_MMU
1167 {
1da177e4
LT
1168 .procname = "max_map_count",
1169 .data = &sysctl_max_map_count,
1170 .maxlen = sizeof(sysctl_max_map_count),
1171 .mode = 0644,
3e26120c 1172 .proc_handler = proc_dointvec_minmax,
70da2340 1173 .extra1 = &zero,
1da177e4 1174 },
dd8632a1
PM
1175#else
1176 {
dd8632a1
PM
1177 .procname = "nr_trim_pages",
1178 .data = &sysctl_nr_trim_pages,
1179 .maxlen = sizeof(sysctl_nr_trim_pages),
1180 .mode = 0644,
6d456111 1181 .proc_handler = proc_dointvec_minmax,
dd8632a1
PM
1182 .extra1 = &zero,
1183 },
1da177e4
LT
1184#endif
1185 {
1da177e4
LT
1186 .procname = "laptop_mode",
1187 .data = &laptop_mode,
1188 .maxlen = sizeof(laptop_mode),
1189 .mode = 0644,
6d456111 1190 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT
1191 },
1192 {
1da177e4
LT
1193 .procname = "block_dump",
1194 .data = &block_dump,
1195 .maxlen = sizeof(block_dump),
1196 .mode = 0644,
6d456111 1197 .proc_handler = proc_dointvec,
1da177e4
LT
1198 .extra1 = &zero,
1199 },
1200 {
1da177e4
LT
1201 .procname = "vfs_cache_pressure",
1202 .data = &sysctl_vfs_cache_pressure,
1203 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1204 .mode = 0644,
6d456111 1205 .proc_handler = proc_dointvec,
1da177e4
LT
1206 .extra1 = &zero,
1207 },
1208#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1209 {
1da177e4
LT
1210 .procname = "legacy_va_layout",
1211 .data = &sysctl_legacy_va_layout,
1212 .maxlen = sizeof(sysctl_legacy_va_layout),
1213 .mode = 0644,
6d456111 1214 .proc_handler = proc_dointvec,
1da177e4
LT
1215 .extra1 = &zero,
1216 },
1217#endif
1743660b
CL
1218#ifdef CONFIG_NUMA
1219 {
1743660b
CL
1220 .procname = "zone_reclaim_mode",
1221 .data = &zone_reclaim_mode,
1222 .maxlen = sizeof(zone_reclaim_mode),
1223 .mode = 0644,
6d456111 1224 .proc_handler = proc_dointvec,
c84db23c 1225 .extra1 = &zero,
1743660b 1226 },
9614634f 1227 {
9614634f
CL
1228 .procname = "min_unmapped_ratio",
1229 .data = &sysctl_min_unmapped_ratio,
1230 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1231 .mode = 0644,
6d456111 1232 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
9614634f
CL
1233 .extra1 = &zero,
1234 .extra2 = &one_hundred,
1235 },
0ff38490 1236 {
0ff38490
CL
1237 .procname = "min_slab_ratio",
1238 .data = &sysctl_min_slab_ratio,
1239 .maxlen = sizeof(sysctl_min_slab_ratio),
1240 .mode = 0644,
6d456111 1241 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
0ff38490
CL
1242 .extra1 = &zero,
1243 .extra2 = &one_hundred,
1244 },
e6e5494c 1245#endif
77461ab3
CL
1246#ifdef CONFIG_SMP
1247 {
77461ab3
CL
1248 .procname = "stat_interval",
1249 .data = &sysctl_stat_interval,
1250 .maxlen = sizeof(sysctl_stat_interval),
1251 .mode = 0644,
6d456111 1252 .proc_handler = proc_dointvec_jiffies,
77461ab3
CL
1253 },
1254#endif
6e141546 1255#ifdef CONFIG_MMU
ed032189 1256 {
ed032189 1257 .procname = "mmap_min_addr",
788084ab
EP
1258 .data = &dac_mmap_min_addr,
1259 .maxlen = sizeof(unsigned long),
ed032189 1260 .mode = 0644,
6d456111 1261 .proc_handler = mmap_min_addr_handler,
ed032189 1262 },
6e141546 1263#endif
f0c0b2b8
KH
1264#ifdef CONFIG_NUMA
1265 {
f0c0b2b8
KH
1266 .procname = "numa_zonelist_order",
1267 .data = &numa_zonelist_order,
1268 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1269 .mode = 0644,
6d456111 1270 .proc_handler = numa_zonelist_order_handler,
f0c0b2b8
KH
1271 },
1272#endif
2b8232ce 1273#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
5c36e657 1274 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
e6e5494c 1275 {
e6e5494c
IM
1276 .procname = "vdso_enabled",
1277 .data = &vdso_enabled,
1278 .maxlen = sizeof(vdso_enabled),
1279 .mode = 0644,
6d456111 1280 .proc_handler = proc_dointvec,
e6e5494c
IM
1281 .extra1 = &zero,
1282 },
1da177e4 1283#endif
195cf453
BG
1284#ifdef CONFIG_HIGHMEM
1285 {
195cf453
BG
1286 .procname = "highmem_is_dirtyable",
1287 .data = &vm_highmem_is_dirtyable,
1288 .maxlen = sizeof(vm_highmem_is_dirtyable),
1289 .mode = 0644,
6d456111 1290 .proc_handler = proc_dointvec_minmax,
195cf453
BG
1291 .extra1 = &zero,
1292 .extra2 = &one,
1293 },
1294#endif
4be6f6bb 1295 {
4be6f6bb
PZ
1296 .procname = "scan_unevictable_pages",
1297 .data = &scan_unevictable_pages,
1298 .maxlen = sizeof(scan_unevictable_pages),
1299 .mode = 0644,
6d456111 1300 .proc_handler = scan_unevictable_handler,
4be6f6bb 1301 },
6a46079c
AK
1302#ifdef CONFIG_MEMORY_FAILURE
1303 {
6a46079c
AK
1304 .procname = "memory_failure_early_kill",
1305 .data = &sysctl_memory_failure_early_kill,
1306 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1307 .mode = 0644,
6d456111 1308 .proc_handler = proc_dointvec_minmax,
6a46079c
AK
1309 .extra1 = &zero,
1310 .extra2 = &one,
1311 },
1312 {
6a46079c
AK
1313 .procname = "memory_failure_recovery",
1314 .data = &sysctl_memory_failure_recovery,
1315 .maxlen = sizeof(sysctl_memory_failure_recovery),
1316 .mode = 0644,
6d456111 1317 .proc_handler = proc_dointvec_minmax,
6a46079c
AK
1318 .extra1 = &zero,
1319 .extra2 = &one,
1320 },
1321#endif
1322
2be7fe07
AM
1323/*
1324 * NOTE: do not add new entries to this table unless you have read
1325 * Documentation/sysctl/ctl_unnumbered.txt
1326 */
6fce56ec 1327 { }
1da177e4
LT
1328};
1329
2abc26fc 1330#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
d8217f07 1331static struct ctl_table binfmt_misc_table[] = {
6fce56ec 1332 { }
2abc26fc
EB
1333};
1334#endif
1335
d8217f07 1336static struct ctl_table fs_table[] = {
1da177e4 1337 {
1da177e4
LT
1338 .procname = "inode-nr",
1339 .data = &inodes_stat,
1340 .maxlen = 2*sizeof(int),
1341 .mode = 0444,
6d456111 1342 .proc_handler = proc_dointvec,
1da177e4
LT
1343 },
1344 {
1da177e4
LT
1345 .procname = "inode-state",
1346 .data = &inodes_stat,
1347 .maxlen = 7*sizeof(int),
1348 .mode = 0444,
6d456111 1349 .proc_handler = proc_dointvec,
1da177e4
LT
1350 },
1351 {
1da177e4
LT
1352 .procname = "file-nr",
1353 .data = &files_stat,
1354 .maxlen = 3*sizeof(int),
1355 .mode = 0444,
6d456111 1356 .proc_handler = proc_nr_files,
1da177e4
LT
1357 },
1358 {
1da177e4
LT
1359 .procname = "file-max",
1360 .data = &files_stat.max_files,
1361 .maxlen = sizeof(int),
1362 .mode = 0644,
6d456111 1363 .proc_handler = proc_dointvec,
1da177e4 1364 },
9cfe015a 1365 {
9cfe015a
ED
1366 .procname = "nr_open",
1367 .data = &sysctl_nr_open,
1368 .maxlen = sizeof(int),
1369 .mode = 0644,
6d456111 1370 .proc_handler = proc_dointvec_minmax,
eceea0b3
AV
1371 .extra1 = &sysctl_nr_open_min,
1372 .extra2 = &sysctl_nr_open_max,
9cfe015a 1373 },
1da177e4 1374 {
1da177e4
LT
1375 .procname = "dentry-state",
1376 .data = &dentry_stat,
1377 .maxlen = 6*sizeof(int),
1378 .mode = 0444,
6d456111 1379 .proc_handler = proc_dointvec,
1da177e4
LT
1380 },
1381 {
1da177e4
LT
1382 .procname = "overflowuid",
1383 .data = &fs_overflowuid,
1384 .maxlen = sizeof(int),
1385 .mode = 0644,
6d456111 1386 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
1387 .extra1 = &minolduid,
1388 .extra2 = &maxolduid,
1389 },
1390 {
1da177e4
LT
1391 .procname = "overflowgid",
1392 .data = &fs_overflowgid,
1393 .maxlen = sizeof(int),
1394 .mode = 0644,
6d456111 1395 .proc_handler = proc_dointvec_minmax,
1da177e4
LT
1396 .extra1 = &minolduid,
1397 .extra2 = &maxolduid,
1398 },
bfcd17a6 1399#ifdef CONFIG_FILE_LOCKING
1da177e4 1400 {
1da177e4
LT
1401 .procname = "leases-enable",
1402 .data = &leases_enable,
1403 .maxlen = sizeof(int),
1404 .mode = 0644,
6d456111 1405 .proc_handler = proc_dointvec,
1da177e4 1406 },
bfcd17a6 1407#endif
1da177e4
LT
1408#ifdef CONFIG_DNOTIFY
1409 {
1da177e4
LT
1410 .procname = "dir-notify-enable",
1411 .data = &dir_notify_enable,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
6d456111 1414 .proc_handler = proc_dointvec,
1da177e4
LT
1415 },
1416#endif
1417#ifdef CONFIG_MMU
bfcd17a6 1418#ifdef CONFIG_FILE_LOCKING
1da177e4 1419 {
1da177e4
LT
1420 .procname = "lease-break-time",
1421 .data = &lease_break_time,
1422 .maxlen = sizeof(int),
1423 .mode = 0644,
6d456111 1424 .proc_handler = proc_dointvec,
1da177e4 1425 },
bfcd17a6 1426#endif
ebf3f09c 1427#ifdef CONFIG_AIO
1da177e4 1428 {
1da177e4
LT
1429 .procname = "aio-nr",
1430 .data = &aio_nr,
1431 .maxlen = sizeof(aio_nr),
1432 .mode = 0444,
6d456111 1433 .proc_handler = proc_doulongvec_minmax,
1da177e4
LT
1434 },
1435 {
1da177e4
LT
1436 .procname = "aio-max-nr",
1437 .data = &aio_max_nr,
1438 .maxlen = sizeof(aio_max_nr),
1439 .mode = 0644,
6d456111 1440 .proc_handler = proc_doulongvec_minmax,
1da177e4 1441 },
ebf3f09c 1442#endif /* CONFIG_AIO */
2d9048e2 1443#ifdef CONFIG_INOTIFY_USER
0399cb08 1444 {
0399cb08
RL
1445 .procname = "inotify",
1446 .mode = 0555,
1447 .child = inotify_table,
1448 },
1449#endif
7ef9964e
DL
1450#ifdef CONFIG_EPOLL
1451 {
1452 .procname = "epoll",
1453 .mode = 0555,
1454 .child = epoll_table,
1455 },
1456#endif
1da177e4 1457#endif
d6e71144 1458 {
d6e71144
AC
1459 .procname = "suid_dumpable",
1460 .data = &suid_dumpable,
1461 .maxlen = sizeof(int),
1462 .mode = 0644,
6d456111 1463 .proc_handler = proc_dointvec_minmax,
8e654fba
MW
1464 .extra1 = &zero,
1465 .extra2 = &two,
d6e71144 1466 },
2abc26fc
EB
1467#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1468 {
2abc26fc
EB
1469 .procname = "binfmt_misc",
1470 .mode = 0555,
1471 .child = binfmt_misc_table,
1472 },
1473#endif
b492e95b 1474 {
ff9da691
JA
1475 .procname = "pipe-max-size",
1476 .data = &pipe_max_size,
b492e95b
JA
1477 .maxlen = sizeof(int),
1478 .mode = 0644,
ff9da691
JA
1479 .proc_handler = &pipe_proc_fn,
1480 .extra1 = &pipe_min_size,
b492e95b 1481 },
2be7fe07
AM
1482/*
1483 * NOTE: do not add new entries to this table unless you have read
1484 * Documentation/sysctl/ctl_unnumbered.txt
2be7fe07 1485 */
6fce56ec 1486 { }
1da177e4
LT
1487};
1488
d8217f07 1489static struct ctl_table debug_table[] = {
ab3c68ee
HC
1490#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1491 defined(CONFIG_S390)
abd4f750 1492 {
abd4f750
MAS
1493 .procname = "exception-trace",
1494 .data = &show_unhandled_signals,
1495 .maxlen = sizeof(int),
1496 .mode = 0644,
1497 .proc_handler = proc_dointvec
1498 },
b2be84df
MH
1499#endif
1500#if defined(CONFIG_OPTPROBES)
1501 {
1502 .procname = "kprobes-optimization",
1503 .data = &sysctl_kprobes_optimization,
1504 .maxlen = sizeof(int),
1505 .mode = 0644,
1506 .proc_handler = proc_kprobes_optimization_handler,
1507 .extra1 = &zero,
1508 .extra2 = &one,
1509 },
abd4f750 1510#endif
6fce56ec 1511 { }
1da177e4
LT
1512};
1513
d8217f07 1514static struct ctl_table dev_table[] = {
6fce56ec 1515 { }
0eeca283 1516};
1da177e4 1517
330d57fb
AV
1518static DEFINE_SPINLOCK(sysctl_lock);
1519
1520/* called under sysctl_lock */
1521static int use_table(struct ctl_table_header *p)
1522{
1523 if (unlikely(p->unregistering))
1524 return 0;
1525 p->used++;
1526 return 1;
1527}
1528
1529/* called under sysctl_lock */
1530static void unuse_table(struct ctl_table_header *p)
1531{
1532 if (!--p->used)
1533 if (unlikely(p->unregistering))
1534 complete(p->unregistering);
1535}
1536
1537/* called under sysctl_lock, will reacquire if has to wait */
1538static void start_unregistering(struct ctl_table_header *p)
1539{
1540 /*
1541 * if p->used is 0, nobody will ever touch that entry again;
1542 * we'll eliminate all paths to it before dropping sysctl_lock
1543 */
1544 if (unlikely(p->used)) {
1545 struct completion wait;
1546 init_completion(&wait);
1547 p->unregistering = &wait;
1548 spin_unlock(&sysctl_lock);
1549 wait_for_completion(&wait);
1550 spin_lock(&sysctl_lock);
f7e6ced4
AV
1551 } else {
1552 /* anything non-NULL; we'll never dereference it */
1553 p->unregistering = ERR_PTR(-EINVAL);
330d57fb
AV
1554 }
1555 /*
1556 * do not remove from the list until nobody holds it; walking the
1557 * list in do_sysctl() relies on that.
1558 */
1559 list_del_init(&p->ctl_entry);
1560}
1561
f7e6ced4
AV
1562void sysctl_head_get(struct ctl_table_header *head)
1563{
1564 spin_lock(&sysctl_lock);
1565 head->count++;
1566 spin_unlock(&sysctl_lock);
1567}
1568
1569void sysctl_head_put(struct ctl_table_header *head)
1570{
1571 spin_lock(&sysctl_lock);
1572 if (!--head->count)
1573 kfree(head);
1574 spin_unlock(&sysctl_lock);
1575}
1576
1577struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1578{
1579 if (!head)
1580 BUG();
1581 spin_lock(&sysctl_lock);
1582 if (!use_table(head))
1583 head = ERR_PTR(-ENOENT);
1584 spin_unlock(&sysctl_lock);
1585 return head;
1586}
1587
805b5d5e
EB
1588void sysctl_head_finish(struct ctl_table_header *head)
1589{
1590 if (!head)
1591 return;
1592 spin_lock(&sysctl_lock);
1593 unuse_table(head);
1594 spin_unlock(&sysctl_lock);
1595}
1596
73455092
AV
1597static struct ctl_table_set *
1598lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1599{
1600 struct ctl_table_set *set = &root->default_set;
1601 if (root->lookup)
1602 set = root->lookup(root, namespaces);
1603 return set;
1604}
1605
e51b6ba0
EB
1606static struct list_head *
1607lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
805b5d5e 1608{
73455092
AV
1609 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1610 return &set->list;
e51b6ba0
EB
1611}
1612
1613struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1614 struct ctl_table_header *prev)
1615{
1616 struct ctl_table_root *root;
1617 struct list_head *header_list;
805b5d5e
EB
1618 struct ctl_table_header *head;
1619 struct list_head *tmp;
e51b6ba0 1620
805b5d5e
EB
1621 spin_lock(&sysctl_lock);
1622 if (prev) {
e51b6ba0 1623 head = prev;
805b5d5e
EB
1624 tmp = &prev->ctl_entry;
1625 unuse_table(prev);
1626 goto next;
1627 }
1628 tmp = &root_table_header.ctl_entry;
1629 for (;;) {
1630 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1631
1632 if (!use_table(head))
1633 goto next;
1634 spin_unlock(&sysctl_lock);
1635 return head;
1636 next:
e51b6ba0 1637 root = head->root;
805b5d5e 1638 tmp = tmp->next;
e51b6ba0
EB
1639 header_list = lookup_header_list(root, namespaces);
1640 if (tmp != header_list)
1641 continue;
1642
1643 do {
1644 root = list_entry(root->root_list.next,
1645 struct ctl_table_root, root_list);
1646 if (root == &sysctl_table_root)
1647 goto out;
1648 header_list = lookup_header_list(root, namespaces);
1649 } while (list_empty(header_list));
1650 tmp = header_list->next;
805b5d5e 1651 }
e51b6ba0 1652out:
805b5d5e
EB
1653 spin_unlock(&sysctl_lock);
1654 return NULL;
1655}
1656
e51b6ba0
EB
1657struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1658{
1659 return __sysctl_head_next(current->nsproxy, prev);
1660}
1661
1662void register_sysctl_root(struct ctl_table_root *root)
1663{
1664 spin_lock(&sysctl_lock);
1665 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1666 spin_unlock(&sysctl_lock);
1667}
1668
1da177e4 1669/*
1ff007eb 1670 * sysctl_perm does NOT grant the superuser all rights automatically, because
1da177e4
LT
1671 * some sysctl variables are readonly even to root.
1672 */
1673
1674static int test_perm(int mode, int op)
1675{
76aac0e9 1676 if (!current_euid())
1da177e4
LT
1677 mode >>= 6;
1678 else if (in_egroup_p(0))
1679 mode >>= 3;
e6305c43 1680 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1da177e4
LT
1681 return 0;
1682 return -EACCES;
1683}
1684
d7321cd6 1685int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1da177e4
LT
1686{
1687 int error;
d7321cd6
PE
1688 int mode;
1689
e6305c43 1690 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1da177e4
LT
1691 if (error)
1692 return error;
d7321cd6
PE
1693
1694 if (root->permissions)
1695 mode = root->permissions(root, current->nsproxy, table);
1696 else
1697 mode = table->mode;
1698
1699 return test_perm(mode, op);
1da177e4
LT
1700}
1701
d912b0cc
EB
1702static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1703{
2315ffa0 1704 for (; table->procname; table++) {
d912b0cc
EB
1705 table->parent = parent;
1706 if (table->child)
1707 sysctl_set_parent(table, table->child);
1708 }
1709}
1710
1711static __init int sysctl_init(void)
1712{
1713 sysctl_set_parent(NULL, root_table);
88f458e4
HS
1714#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1715 {
1716 int err;
1717 err = sysctl_check_table(current->nsproxy, root_table);
1718 }
1719#endif
d912b0cc
EB
1720 return 0;
1721}
1722
1723core_initcall(sysctl_init);
1724
bfbcf034
AV
1725static struct ctl_table *is_branch_in(struct ctl_table *branch,
1726 struct ctl_table *table)
ae7edecc
AV
1727{
1728 struct ctl_table *p;
1729 const char *s = branch->procname;
1730
1731 /* branch should have named subdirectory as its first element */
1732 if (!s || !branch->child)
bfbcf034 1733 return NULL;
ae7edecc
AV
1734
1735 /* ... and nothing else */
2315ffa0 1736 if (branch[1].procname)
bfbcf034 1737 return NULL;
ae7edecc
AV
1738
1739 /* table should contain subdirectory with the same name */
2315ffa0 1740 for (p = table; p->procname; p++) {
ae7edecc
AV
1741 if (!p->child)
1742 continue;
1743 if (p->procname && strcmp(p->procname, s) == 0)
bfbcf034 1744 return p;
ae7edecc 1745 }
bfbcf034 1746 return NULL;
ae7edecc
AV
1747}
1748
1749/* see if attaching q to p would be an improvement */
1750static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1751{
1752 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
bfbcf034 1753 struct ctl_table *next;
ae7edecc
AV
1754 int is_better = 0;
1755 int not_in_parent = !p->attached_by;
1756
bfbcf034 1757 while ((next = is_branch_in(by, to)) != NULL) {
ae7edecc
AV
1758 if (by == q->attached_by)
1759 is_better = 1;
1760 if (to == p->attached_by)
1761 not_in_parent = 1;
1762 by = by->child;
bfbcf034 1763 to = next->child;
ae7edecc
AV
1764 }
1765
1766 if (is_better && not_in_parent) {
1767 q->attached_by = by;
1768 q->attached_to = to;
1769 q->parent = p;
1770 }
1771}
1772
1da177e4 1773/**
e51b6ba0
EB
1774 * __register_sysctl_paths - register a sysctl hierarchy
1775 * @root: List of sysctl headers to register on
1776 * @namespaces: Data to compute which lists of sysctl entries are visible
29e796fd 1777 * @path: The path to the directory the sysctl table is in.
1da177e4 1778 * @table: the top-level table structure
1da177e4
LT
1779 *
1780 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
29e796fd 1781 * array. A completely 0 filled entry terminates the table.
1da177e4 1782 *
d8217f07 1783 * The members of the &struct ctl_table structure are used as follows:
1da177e4 1784 *
1da177e4
LT
1785 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1786 * enter a sysctl file
1787 *
1788 * data - a pointer to data for use by proc_handler
1789 *
1790 * maxlen - the maximum size in bytes of the data
1791 *
1792 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1793 *
1794 * child - a pointer to the child sysctl table if this entry is a directory, or
1795 * %NULL.
1796 *
1797 * proc_handler - the text handler routine (described below)
1798 *
1da177e4
LT
1799 * de - for internal use by the sysctl routines
1800 *
1801 * extra1, extra2 - extra pointers usable by the proc handler routines
1802 *
1803 * Leaf nodes in the sysctl tree will be represented by a single file
1804 * under /proc; non-leaf nodes will be represented by directories.
1805 *
1806 * sysctl(2) can automatically manage read and write requests through
1807 * the sysctl table. The data and maxlen fields of the ctl_table
1808 * struct enable minimal validation of the values being written to be
1809 * performed, and the mode field allows minimal authentication.
1810 *
1da177e4
LT
1811 * There must be a proc_handler routine for any terminal nodes
1812 * mirrored under /proc/sys (non-terminals are handled by a built-in
1813 * directory handler). Several default handlers are available to
1814 * cover common cases -
1815 *
1816 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1817 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1818 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1819 *
1820 * It is the handler's job to read the input buffer from user memory
1821 * and process it. The handler should return 0 on success.
1822 *
1823 * This routine returns %NULL on a failure to register, and a pointer
1824 * to the table header on success.
1825 */
e51b6ba0
EB
1826struct ctl_table_header *__register_sysctl_paths(
1827 struct ctl_table_root *root,
1828 struct nsproxy *namespaces,
1829 const struct ctl_path *path, struct ctl_table *table)
1da177e4 1830{
29e796fd
EB
1831 struct ctl_table_header *header;
1832 struct ctl_table *new, **prevp;
1833 unsigned int n, npath;
ae7edecc 1834 struct ctl_table_set *set;
29e796fd
EB
1835
1836 /* Count the path components */
2315ffa0 1837 for (npath = 0; path[npath].procname; ++npath)
29e796fd
EB
1838 ;
1839
1840 /*
1841 * For each path component, allocate a 2-element ctl_table array.
1842 * The first array element will be filled with the sysctl entry
2315ffa0 1843 * for this, the second will be the sentinel (procname == 0).
29e796fd
EB
1844 *
1845 * We allocate everything in one go so that we don't have to
1846 * worry about freeing additional memory in unregister_sysctl_table.
1847 */
1848 header = kzalloc(sizeof(struct ctl_table_header) +
1849 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1850 if (!header)
1da177e4 1851 return NULL;
29e796fd
EB
1852
1853 new = (struct ctl_table *) (header + 1);
1854
1855 /* Now connect the dots */
1856 prevp = &header->ctl_table;
1857 for (n = 0; n < npath; ++n, ++path) {
1858 /* Copy the procname */
1859 new->procname = path->procname;
29e796fd
EB
1860 new->mode = 0555;
1861
1862 *prevp = new;
1863 prevp = &new->child;
1864
1865 new += 2;
1866 }
1867 *prevp = table;
23eb06de 1868 header->ctl_table_arg = table;
29e796fd
EB
1869
1870 INIT_LIST_HEAD(&header->ctl_entry);
1871 header->used = 0;
1872 header->unregistering = NULL;
e51b6ba0 1873 header->root = root;
29e796fd 1874 sysctl_set_parent(NULL, header->ctl_table);
f7e6ced4 1875 header->count = 1;
88f458e4 1876#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
e51b6ba0 1877 if (sysctl_check_table(namespaces, header->ctl_table)) {
29e796fd 1878 kfree(header);
fc6cd25b
EB
1879 return NULL;
1880 }
88f458e4 1881#endif
330d57fb 1882 spin_lock(&sysctl_lock);
73455092 1883 header->set = lookup_header_set(root, namespaces);
ae7edecc
AV
1884 header->attached_by = header->ctl_table;
1885 header->attached_to = root_table;
1886 header->parent = &root_table_header;
1887 for (set = header->set; set; set = set->parent) {
1888 struct ctl_table_header *p;
1889 list_for_each_entry(p, &set->list, ctl_entry) {
1890 if (p->unregistering)
1891 continue;
1892 try_attach(p, header);
1893 }
1894 }
1895 header->parent->count++;
73455092 1896 list_add_tail(&header->ctl_entry, &header->set->list);
330d57fb 1897 spin_unlock(&sysctl_lock);
29e796fd
EB
1898
1899 return header;
1900}
1901
e51b6ba0
EB
1902/**
1903 * register_sysctl_table_path - register a sysctl table hierarchy
1904 * @path: The path to the directory the sysctl table is in.
1905 * @table: the top-level table structure
1906 *
1907 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1908 * array. A completely 0 filled entry terminates the table.
1909 *
1910 * See __register_sysctl_paths for more details.
1911 */
1912struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1913 struct ctl_table *table)
1914{
1915 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1916 path, table);
1917}
1918
29e796fd
EB
1919/**
1920 * register_sysctl_table - register a sysctl table hierarchy
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_table(struct ctl_table *table)
1929{
1930 static const struct ctl_path null_path[] = { {} };
1931
1932 return register_sysctl_paths(null_path, table);
1da177e4
LT
1933}
1934
1935/**
1936 * unregister_sysctl_table - unregister a sysctl table hierarchy
1937 * @header: the header returned from register_sysctl_table
1938 *
1939 * Unregisters the sysctl table and all children. proc entries may not
1940 * actually be removed until they are no longer used by anyone.
1941 */
1942void unregister_sysctl_table(struct ctl_table_header * header)
1943{
330d57fb 1944 might_sleep();
f1dad166
PE
1945
1946 if (header == NULL)
1947 return;
1948
330d57fb
AV
1949 spin_lock(&sysctl_lock);
1950 start_unregistering(header);
ae7edecc
AV
1951 if (!--header->parent->count) {
1952 WARN_ON(1);
1953 kfree(header->parent);
1954 }
f7e6ced4
AV
1955 if (!--header->count)
1956 kfree(header);
330d57fb 1957 spin_unlock(&sysctl_lock);
1da177e4
LT
1958}
1959
9043476f
AV
1960int sysctl_is_seen(struct ctl_table_header *p)
1961{
1962 struct ctl_table_set *set = p->set;
1963 int res;
1964 spin_lock(&sysctl_lock);
1965 if (p->unregistering)
1966 res = 0;
1967 else if (!set->is_seen)
1968 res = 1;
1969 else
1970 res = set->is_seen(set);
1971 spin_unlock(&sysctl_lock);
1972 return res;
1973}
1974
73455092
AV
1975void setup_sysctl_set(struct ctl_table_set *p,
1976 struct ctl_table_set *parent,
1977 int (*is_seen)(struct ctl_table_set *))
1978{
1979 INIT_LIST_HEAD(&p->list);
1980 p->parent = parent ? parent : &sysctl_table_root.default_set;
1981 p->is_seen = is_seen;
1982}
1983
b89a8171 1984#else /* !CONFIG_SYSCTL */
d8217f07 1985struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
b89a8171
EB
1986{
1987 return NULL;
1988}
1989
29e796fd
EB
1990struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1991 struct ctl_table *table)
1992{
1993 return NULL;
1994}
1995
b89a8171
EB
1996void unregister_sysctl_table(struct ctl_table_header * table)
1997{
1998}
1999
73455092
AV
2000void setup_sysctl_set(struct ctl_table_set *p,
2001 struct ctl_table_set *parent,
2002 int (*is_seen)(struct ctl_table_set *))
2003{
2004}
2005
f7e6ced4
AV
2006void sysctl_head_put(struct ctl_table_header *head)
2007{
2008}
2009
b89a8171
EB
2010#endif /* CONFIG_SYSCTL */
2011
1da177e4
LT
2012/*
2013 * /proc/sys support
2014 */
2015
b89a8171 2016#ifdef CONFIG_PROC_SYSCTL
1da177e4 2017
b1ba4ddd 2018static int _proc_do_string(void* data, int maxlen, int write,
8d65af78 2019 void __user *buffer,
b1ba4ddd 2020 size_t *lenp, loff_t *ppos)
1da177e4
LT
2021{
2022 size_t len;
2023 char __user *p;
2024 char c;
8d060877
ON
2025
2026 if (!data || !maxlen || !*lenp) {
1da177e4
LT
2027 *lenp = 0;
2028 return 0;
2029 }
8d060877 2030
1da177e4
LT
2031 if (write) {
2032 len = 0;
2033 p = buffer;
2034 while (len < *lenp) {
2035 if (get_user(c, p++))
2036 return -EFAULT;
2037 if (c == 0 || c == '\n')
2038 break;
2039 len++;
2040 }
f5dd3d6f
SV
2041 if (len >= maxlen)
2042 len = maxlen-1;
2043 if(copy_from_user(data, buffer, len))
1da177e4 2044 return -EFAULT;
f5dd3d6f 2045 ((char *) data)[len] = 0;
1da177e4
LT
2046 *ppos += *lenp;
2047 } else {
f5dd3d6f
SV
2048 len = strlen(data);
2049 if (len > maxlen)
2050 len = maxlen;
8d060877
ON
2051
2052 if (*ppos > len) {
2053 *lenp = 0;
2054 return 0;
2055 }
2056
2057 data += *ppos;
2058 len -= *ppos;
2059
1da177e4
LT
2060 if (len > *lenp)
2061 len = *lenp;
2062 if (len)
f5dd3d6f 2063 if(copy_to_user(buffer, data, len))
1da177e4
LT
2064 return -EFAULT;
2065 if (len < *lenp) {
2066 if(put_user('\n', ((char __user *) buffer) + len))
2067 return -EFAULT;
2068 len++;
2069 }
2070 *lenp = len;
2071 *ppos += len;
2072 }
2073 return 0;
2074}
2075
f5dd3d6f
SV
2076/**
2077 * proc_dostring - read a string sysctl
2078 * @table: the sysctl table
2079 * @write: %TRUE if this is a write to the sysctl file
f5dd3d6f
SV
2080 * @buffer: the user buffer
2081 * @lenp: the size of the user buffer
2082 * @ppos: file position
2083 *
2084 * Reads/writes a string from/to the user buffer. If the kernel
2085 * buffer provided is not large enough to hold the string, the
2086 * string is truncated. The copied string is %NULL-terminated.
2087 * If the string is being read by the user process, it is copied
2088 * and a newline '\n' is added. It is truncated if the buffer is
2089 * not large enough.
2090 *
2091 * Returns 0 on success.
2092 */
8d65af78 2093int proc_dostring(struct ctl_table *table, int write,
f5dd3d6f
SV
2094 void __user *buffer, size_t *lenp, loff_t *ppos)
2095{
8d65af78 2096 return _proc_do_string(table->data, table->maxlen, write,
f5dd3d6f
SV
2097 buffer, lenp, ppos);
2098}
2099
00b7c339
AW
2100static size_t proc_skip_spaces(char **buf)
2101{
2102 size_t ret;
2103 char *tmp = skip_spaces(*buf);
2104 ret = tmp - *buf;
2105 *buf = tmp;
2106 return ret;
2107}
2108
9f977fb7
OP
2109static void proc_skip_char(char **buf, size_t *size, const char v)
2110{
2111 while (*size) {
2112 if (**buf != v)
2113 break;
2114 (*size)--;
2115 (*buf)++;
2116 }
2117}
2118
00b7c339
AW
2119#define TMPBUFLEN 22
2120/**
0fc377bd 2121 * proc_get_long - reads an ASCII formatted integer from a user buffer
00b7c339 2122 *
0fc377bd
RD
2123 * @buf: a kernel buffer
2124 * @size: size of the kernel buffer
2125 * @val: this is where the number will be stored
2126 * @neg: set to %TRUE if number is negative
2127 * @perm_tr: a vector which contains the allowed trailers
2128 * @perm_tr_len: size of the perm_tr vector
2129 * @tr: pointer to store the trailer character
00b7c339 2130 *
0fc377bd
RD
2131 * In case of success %0 is returned and @buf and @size are updated with
2132 * the amount of bytes read. If @tr is non-NULL and a trailing
2133 * character exists (size is non-zero after returning from this
2134 * function), @tr is updated with the trailing character.
00b7c339
AW
2135 */
2136static int proc_get_long(char **buf, size_t *size,
2137 unsigned long *val, bool *neg,
2138 const char *perm_tr, unsigned perm_tr_len, char *tr)
2139{
2140 int len;
2141 char *p, tmp[TMPBUFLEN];
2142
2143 if (!*size)
2144 return -EINVAL;
2145
2146 len = *size;
2147 if (len > TMPBUFLEN - 1)
2148 len = TMPBUFLEN - 1;
2149
2150 memcpy(tmp, *buf, len);
2151
2152 tmp[len] = 0;
2153 p = tmp;
2154 if (*p == '-' && *size > 1) {
2155 *neg = true;
2156 p++;
2157 } else
2158 *neg = false;
2159 if (!isdigit(*p))
2160 return -EINVAL;
2161
2162 *val = simple_strtoul(p, &p, 0);
2163
2164 len = p - tmp;
2165
2166 /* We don't know if the next char is whitespace thus we may accept
2167 * invalid integers (e.g. 1234...a) or two integers instead of one
2168 * (e.g. 123...1). So lets not allow such large numbers. */
2169 if (len == TMPBUFLEN - 1)
2170 return -EINVAL;
2171
2172 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2173 return -EINVAL;
1da177e4 2174
00b7c339
AW
2175 if (tr && (len < *size))
2176 *tr = *p;
2177
2178 *buf += len;
2179 *size -= len;
2180
2181 return 0;
2182}
2183
2184/**
0fc377bd 2185 * proc_put_long - converts an integer to a decimal ASCII formatted string
00b7c339 2186 *
0fc377bd
RD
2187 * @buf: the user buffer
2188 * @size: the size of the user buffer
2189 * @val: the integer to be converted
2190 * @neg: sign of the number, %TRUE for negative
00b7c339 2191 *
0fc377bd
RD
2192 * In case of success %0 is returned and @buf and @size are updated with
2193 * the amount of bytes written.
00b7c339
AW
2194 */
2195static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2196 bool neg)
2197{
2198 int len;
2199 char tmp[TMPBUFLEN], *p = tmp;
2200
2201 sprintf(p, "%s%lu", neg ? "-" : "", val);
2202 len = strlen(tmp);
2203 if (len > *size)
2204 len = *size;
2205 if (copy_to_user(*buf, tmp, len))
2206 return -EFAULT;
2207 *size -= len;
2208 *buf += len;
2209 return 0;
2210}
2211#undef TMPBUFLEN
2212
2213static int proc_put_char(void __user **buf, size_t *size, char c)
2214{
2215 if (*size) {
2216 char __user **buffer = (char __user **)buf;
2217 if (put_user(c, *buffer))
2218 return -EFAULT;
2219 (*size)--, (*buffer)++;
2220 *buf = *buffer;
2221 }
2222 return 0;
2223}
1da177e4 2224
00b7c339 2225static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT
2226 int *valp,
2227 int write, void *data)
2228{
2229 if (write) {
2230 *valp = *negp ? -*lvalp : *lvalp;
2231 } else {
2232 int val = *valp;
2233 if (val < 0) {
00b7c339 2234 *negp = true;
1da177e4
LT
2235 *lvalp = (unsigned long)-val;
2236 } else {
00b7c339 2237 *negp = false;
1da177e4
LT
2238 *lvalp = (unsigned long)val;
2239 }
2240 }
2241 return 0;
2242}
2243
00b7c339
AW
2244static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2245
d8217f07 2246static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
8d65af78 2247 int write, void __user *buffer,
fcfbd547 2248 size_t *lenp, loff_t *ppos,
00b7c339 2249 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1da177e4
LT
2250 int write, void *data),
2251 void *data)
2252{
00b7c339
AW
2253 int *i, vleft, first = 1, err = 0;
2254 unsigned long page = 0;
2255 size_t left;
2256 char *kbuf;
1da177e4 2257
00b7c339 2258 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1da177e4
LT
2259 *lenp = 0;
2260 return 0;
2261 }
2262
fcfbd547 2263 i = (int *) tbl_data;
1da177e4
LT
2264 vleft = table->maxlen / sizeof(*i);
2265 left = *lenp;
2266
2267 if (!conv)
2268 conv = do_proc_dointvec_conv;
2269
00b7c339
AW
2270 if (write) {
2271 if (left > PAGE_SIZE - 1)
2272 left = PAGE_SIZE - 1;
2273 page = __get_free_page(GFP_TEMPORARY);
2274 kbuf = (char *) page;
2275 if (!kbuf)
2276 return -ENOMEM;
2277 if (copy_from_user(kbuf, buffer, left)) {
2278 err = -EFAULT;
2279 goto free;
2280 }
2281 kbuf[left] = 0;
2282 }
2283
1da177e4 2284 for (; left && vleft--; i++, first=0) {
00b7c339
AW
2285 unsigned long lval;
2286 bool neg;
1da177e4 2287
00b7c339
AW
2288 if (write) {
2289 left -= proc_skip_spaces(&kbuf);
1da177e4 2290
563b0467
O
2291 if (!left)
2292 break;
00b7c339
AW
2293 err = proc_get_long(&kbuf, &left, &lval, &neg,
2294 proc_wspace_sep,
2295 sizeof(proc_wspace_sep), NULL);
2296 if (err)
1da177e4 2297 break;
00b7c339
AW
2298 if (conv(&neg, &lval, i, 1, data)) {
2299 err = -EINVAL;
1da177e4 2300 break;
00b7c339 2301 }
1da177e4 2302 } else {
00b7c339
AW
2303 if (conv(&neg, &lval, i, 0, data)) {
2304 err = -EINVAL;
2305 break;
2306 }
1da177e4 2307 if (!first)
00b7c339
AW
2308 err = proc_put_char(&buffer, &left, '\t');
2309 if (err)
2310 break;
2311 err = proc_put_long(&buffer, &left, lval, neg);
2312 if (err)
1da177e4 2313 break;
1da177e4
LT
2314 }
2315 }
2316
00b7c339
AW
2317 if (!write && !first && left && !err)
2318 err = proc_put_char(&buffer, &left, '\n');
563b0467 2319 if (write && !err && left)
00b7c339
AW
2320 left -= proc_skip_spaces(&kbuf);
2321free:
1da177e4 2322 if (write) {
00b7c339
AW
2323 free_page(page);
2324 if (first)
2325 return err ? : -EINVAL;
1da177e4 2326 }
1da177e4
LT
2327 *lenp -= left;
2328 *ppos += *lenp;
00b7c339 2329 return err;
1da177e4
LT
2330}
2331
8d65af78 2332static int do_proc_dointvec(struct ctl_table *table, int write,
fcfbd547 2333 void __user *buffer, size_t *lenp, loff_t *ppos,
00b7c339 2334 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
fcfbd547
KK
2335 int write, void *data),
2336 void *data)
2337{
8d65af78 2338 return __do_proc_dointvec(table->data, table, write,
fcfbd547
KK
2339 buffer, lenp, ppos, conv, data);
2340}
2341
1da177e4
LT
2342/**
2343 * proc_dointvec - read a vector of integers
2344 * @table: the sysctl table
2345 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2346 * @buffer: the user buffer
2347 * @lenp: the size of the user buffer
2348 * @ppos: file position
2349 *
2350 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2351 * values from/to the user buffer, treated as an ASCII string.
2352 *
2353 * Returns 0 on success.
2354 */
8d65af78 2355int proc_dointvec(struct ctl_table *table, int write,
1da177e4
LT
2356 void __user *buffer, size_t *lenp, loff_t *ppos)
2357{
8d65af78 2358 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT
2359 NULL,NULL);
2360}
2361
34f5a398 2362/*
25ddbb18
AK
2363 * Taint values can only be increased
2364 * This means we can safely use a temporary.
34f5a398 2365 */
8d65af78 2366static int proc_taint(struct ctl_table *table, int write,
34f5a398
TT
2367 void __user *buffer, size_t *lenp, loff_t *ppos)
2368{
25ddbb18
AK
2369 struct ctl_table t;
2370 unsigned long tmptaint = get_taint();
2371 int err;
34f5a398 2372
91fcd412 2373 if (write && !capable(CAP_SYS_ADMIN))
34f5a398
TT
2374 return -EPERM;
2375
25ddbb18
AK
2376 t = *table;
2377 t.data = &tmptaint;
8d65af78 2378 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
25ddbb18
AK
2379 if (err < 0)
2380 return err;
2381
2382 if (write) {
2383 /*
2384 * Poor man's atomic or. Not worth adding a primitive
2385 * to everyone's atomic.h for this
2386 */
2387 int i;
2388 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2389 if ((tmptaint >> i) & 1)
2390 add_taint(i);
2391 }
2392 }
2393
2394 return err;
34f5a398
TT
2395}
2396
1da177e4
LT
2397struct do_proc_dointvec_minmax_conv_param {
2398 int *min;
2399 int *max;
2400};
2401
00b7c339
AW
2402static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2403 int *valp,
1da177e4
LT
2404 int write, void *data)
2405{
2406 struct do_proc_dointvec_minmax_conv_param *param = data;
2407 if (write) {
2408 int val = *negp ? -*lvalp : *lvalp;
2409 if ((param->min && *param->min > val) ||
2410 (param->max && *param->max < val))
2411 return -EINVAL;
2412 *valp = val;
2413 } else {
2414 int val = *valp;
2415 if (val < 0) {
00b7c339 2416 *negp = true;
1da177e4
LT
2417 *lvalp = (unsigned long)-val;
2418 } else {
00b7c339 2419 *negp = false;
1da177e4
LT
2420 *lvalp = (unsigned long)val;
2421 }
2422 }
2423 return 0;
2424}
2425
2426/**
2427 * proc_dointvec_minmax - read a vector of integers with min/max values
2428 * @table: the sysctl table
2429 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2430 * @buffer: the user buffer
2431 * @lenp: the size of the user buffer
2432 * @ppos: file position
2433 *
2434 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2435 * values from/to the user buffer, treated as an ASCII string.
2436 *
2437 * This routine will ensure the values are within the range specified by
2438 * table->extra1 (min) and table->extra2 (max).
2439 *
2440 * Returns 0 on success.
2441 */
8d65af78 2442int proc_dointvec_minmax(struct ctl_table *table, int write,
1da177e4
LT
2443 void __user *buffer, size_t *lenp, loff_t *ppos)
2444{
2445 struct do_proc_dointvec_minmax_conv_param param = {
2446 .min = (int *) table->extra1,
2447 .max = (int *) table->extra2,
2448 };
8d65af78 2449 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1da177e4
LT
2450 do_proc_dointvec_minmax_conv, &param);
2451}
2452
d8217f07 2453static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
1da177e4
LT
2454 void __user *buffer,
2455 size_t *lenp, loff_t *ppos,
2456 unsigned long convmul,
2457 unsigned long convdiv)
2458{
00b7c339
AW
2459 unsigned long *i, *min, *max;
2460 int vleft, first = 1, err = 0;
2461 unsigned long page = 0;
2462 size_t left;
2463 char *kbuf;
2464
2465 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1da177e4
LT
2466 *lenp = 0;
2467 return 0;
2468 }
00b7c339 2469
fcfbd547 2470 i = (unsigned long *) data;
1da177e4
LT
2471 min = (unsigned long *) table->extra1;
2472 max = (unsigned long *) table->extra2;
2473 vleft = table->maxlen / sizeof(unsigned long);
2474 left = *lenp;
00b7c339
AW
2475
2476 if (write) {
2477 if (left > PAGE_SIZE - 1)
2478 left = PAGE_SIZE - 1;
2479 page = __get_free_page(GFP_TEMPORARY);
2480 kbuf = (char *) page;
2481 if (!kbuf)
2482 return -ENOMEM;
2483 if (copy_from_user(kbuf, buffer, left)) {
2484 err = -EFAULT;
2485 goto free;
2486 }
2487 kbuf[left] = 0;
2488 }
2489
1da177e4 2490 for (; left && vleft--; i++, min++, max++, first=0) {
00b7c339
AW
2491 unsigned long val;
2492
1da177e4 2493 if (write) {
00b7c339
AW
2494 bool neg;
2495
2496 left -= proc_skip_spaces(&kbuf);
2497
2498 err = proc_get_long(&kbuf, &left, &val, &neg,
2499 proc_wspace_sep,
2500 sizeof(proc_wspace_sep), NULL);
2501 if (err)
1da177e4
LT
2502 break;
2503 if (neg)
1da177e4
LT
2504 continue;
2505 if ((min && val < *min) || (max && val > *max))
2506 continue;
2507 *i = val;
2508 } else {
00b7c339 2509 val = convdiv * (*i) / convmul;
1da177e4 2510 if (!first)
00b7c339
AW
2511 err = proc_put_char(&buffer, &left, '\t');
2512 err = proc_put_long(&buffer, &left, val, false);
2513 if (err)
2514 break;
1da177e4
LT
2515 }
2516 }
2517
00b7c339
AW
2518 if (!write && !first && left && !err)
2519 err = proc_put_char(&buffer, &left, '\n');
2520 if (write && !err)
2521 left -= proc_skip_spaces(&kbuf);
2522free:
1da177e4 2523 if (write) {
00b7c339
AW
2524 free_page(page);
2525 if (first)
2526 return err ? : -EINVAL;
1da177e4 2527 }
1da177e4
LT
2528 *lenp -= left;
2529 *ppos += *lenp;
00b7c339 2530 return err;
1da177e4
LT
2531}
2532
d8217f07 2533static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
fcfbd547
KK
2534 void __user *buffer,
2535 size_t *lenp, loff_t *ppos,
2536 unsigned long convmul,
2537 unsigned long convdiv)
2538{
2539 return __do_proc_doulongvec_minmax(table->data, table, write,
8d65af78 2540 buffer, lenp, ppos, convmul, convdiv);
fcfbd547
KK
2541}
2542
1da177e4
LT
2543/**
2544 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2545 * @table: the sysctl table
2546 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2547 * @buffer: the user buffer
2548 * @lenp: the size of the user buffer
2549 * @ppos: file position
2550 *
2551 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2552 * values from/to the user buffer, treated as an ASCII string.
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 */
8d65af78 2559int proc_doulongvec_minmax(struct ctl_table *table, int write,
1da177e4
LT
2560 void __user *buffer, size_t *lenp, loff_t *ppos)
2561{
8d65af78 2562 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1da177e4
LT
2563}
2564
2565/**
2566 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2567 * @table: the sysctl table
2568 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2569 * @buffer: the user buffer
2570 * @lenp: the size of the user buffer
2571 * @ppos: file position
2572 *
2573 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2574 * values from/to the user buffer, treated as an ASCII string. The values
2575 * are treated as milliseconds, and converted to jiffies when they are stored.
2576 *
2577 * This routine will ensure the values are within the range specified by
2578 * table->extra1 (min) and table->extra2 (max).
2579 *
2580 * Returns 0 on success.
2581 */
d8217f07 2582int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1da177e4
LT
2583 void __user *buffer,
2584 size_t *lenp, loff_t *ppos)
2585{
8d65af78 2586 return do_proc_doulongvec_minmax(table, write, buffer,
1da177e4
LT
2587 lenp, ppos, HZ, 1000l);
2588}
2589
2590
00b7c339 2591static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT
2592 int *valp,
2593 int write, void *data)
2594{
2595 if (write) {
cba9f33d
BS
2596 if (*lvalp > LONG_MAX / HZ)
2597 return 1;
1da177e4
LT
2598 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2599 } else {
2600 int val = *valp;
2601 unsigned long lval;
2602 if (val < 0) {
00b7c339 2603 *negp = true;
1da177e4
LT
2604 lval = (unsigned long)-val;
2605 } else {
00b7c339 2606 *negp = false;
1da177e4
LT
2607 lval = (unsigned long)val;
2608 }
2609 *lvalp = lval / HZ;
2610 }
2611 return 0;
2612}
2613
00b7c339 2614static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT
2615 int *valp,
2616 int write, void *data)
2617{
2618 if (write) {
cba9f33d
BS
2619 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2620 return 1;
1da177e4
LT
2621 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2622 } else {
2623 int val = *valp;
2624 unsigned long lval;
2625 if (val < 0) {
00b7c339 2626 *negp = true;
1da177e4
LT
2627 lval = (unsigned long)-val;
2628 } else {
00b7c339 2629 *negp = false;
1da177e4
LT
2630 lval = (unsigned long)val;
2631 }
2632 *lvalp = jiffies_to_clock_t(lval);
2633 }
2634 return 0;
2635}
2636
00b7c339 2637static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1da177e4
LT
2638 int *valp,
2639 int write, void *data)
2640{
2641 if (write) {
2642 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2643 } else {
2644 int val = *valp;
2645 unsigned long lval;
2646 if (val < 0) {
00b7c339 2647 *negp = true;
1da177e4
LT
2648 lval = (unsigned long)-val;
2649 } else {
00b7c339 2650 *negp = false;
1da177e4
LT
2651 lval = (unsigned long)val;
2652 }
2653 *lvalp = jiffies_to_msecs(lval);
2654 }
2655 return 0;
2656}
2657
2658/**
2659 * proc_dointvec_jiffies - read a vector of integers as seconds
2660 * @table: the sysctl table
2661 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2662 * @buffer: the user buffer
2663 * @lenp: the size of the user buffer
2664 * @ppos: file position
2665 *
2666 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2667 * values from/to the user buffer, treated as an ASCII string.
2668 * The values read are assumed to be in seconds, and are converted into
2669 * jiffies.
2670 *
2671 * Returns 0 on success.
2672 */
8d65af78 2673int proc_dointvec_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2674 void __user *buffer, size_t *lenp, loff_t *ppos)
2675{
8d65af78 2676 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT
2677 do_proc_dointvec_jiffies_conv,NULL);
2678}
2679
2680/**
2681 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2682 * @table: the sysctl table
2683 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2684 * @buffer: the user buffer
2685 * @lenp: the size of the user buffer
1e5d5331 2686 * @ppos: pointer to the file position
1da177e4
LT
2687 *
2688 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2689 * values from/to the user buffer, treated as an ASCII string.
2690 * The values read are assumed to be in 1/USER_HZ seconds, and
2691 * are converted into jiffies.
2692 *
2693 * Returns 0 on success.
2694 */
8d65af78 2695int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2696 void __user *buffer, size_t *lenp, loff_t *ppos)
2697{
8d65af78 2698 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1da177e4
LT
2699 do_proc_dointvec_userhz_jiffies_conv,NULL);
2700}
2701
2702/**
2703 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2704 * @table: the sysctl table
2705 * @write: %TRUE if this is a write to the sysctl file
1da177e4
LT
2706 * @buffer: the user buffer
2707 * @lenp: the size of the user buffer
67be2dd1
MW
2708 * @ppos: file position
2709 * @ppos: the current position in the file
1da177e4
LT
2710 *
2711 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2712 * values from/to the user buffer, treated as an ASCII string.
2713 * The values read are assumed to be in 1/1000 seconds, and
2714 * are converted into jiffies.
2715 *
2716 * Returns 0 on success.
2717 */
8d65af78 2718int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2719 void __user *buffer, size_t *lenp, loff_t *ppos)
2720{
8d65af78 2721 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1da177e4
LT
2722 do_proc_dointvec_ms_jiffies_conv, NULL);
2723}
2724
8d65af78 2725static int proc_do_cad_pid(struct ctl_table *table, int write,
9ec52099
CLG
2726 void __user *buffer, size_t *lenp, loff_t *ppos)
2727{
2728 struct pid *new_pid;
2729 pid_t tmp;
2730 int r;
2731
6c5f3e7b 2732 tmp = pid_vnr(cad_pid);
9ec52099 2733
8d65af78 2734 r = __do_proc_dointvec(&tmp, table, write, buffer,
9ec52099
CLG
2735 lenp, ppos, NULL, NULL);
2736 if (r || !write)
2737 return r;
2738
2739 new_pid = find_get_pid(tmp);
2740 if (!new_pid)
2741 return -ESRCH;
2742
2743 put_pid(xchg(&cad_pid, new_pid));
2744 return 0;
2745}
2746
9f977fb7
OP
2747/**
2748 * proc_do_large_bitmap - read/write from/to a large bitmap
2749 * @table: the sysctl table
2750 * @write: %TRUE if this is a write to the sysctl file
2751 * @buffer: the user buffer
2752 * @lenp: the size of the user buffer
2753 * @ppos: file position
2754 *
2755 * The bitmap is stored at table->data and the bitmap length (in bits)
2756 * in table->maxlen.
2757 *
2758 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2759 * large bitmaps may be represented in a compact manner. Writing into
2760 * the file will clear the bitmap then update it with the given input.
2761 *
2762 * Returns 0 on success.
2763 */
2764int proc_do_large_bitmap(struct ctl_table *table, int write,
2765 void __user *buffer, size_t *lenp, loff_t *ppos)
2766{
2767 int err = 0;
2768 bool first = 1;
2769 size_t left = *lenp;
2770 unsigned long bitmap_len = table->maxlen;
2771 unsigned long *bitmap = (unsigned long *) table->data;
2772 unsigned long *tmp_bitmap = NULL;
2773 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2774
2775 if (!bitmap_len || !left || (*ppos && !write)) {
2776 *lenp = 0;
2777 return 0;
2778 }
2779
2780 if (write) {
2781 unsigned long page = 0;
2782 char *kbuf;
2783
2784 if (left > PAGE_SIZE - 1)
2785 left = PAGE_SIZE - 1;
2786
2787 page = __get_free_page(GFP_TEMPORARY);
2788 kbuf = (char *) page;
2789 if (!kbuf)
2790 return -ENOMEM;
2791 if (copy_from_user(kbuf, buffer, left)) {
2792 free_page(page);
2793 return -EFAULT;
2794 }
2795 kbuf[left] = 0;
2796
2797 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2798 GFP_KERNEL);
2799 if (!tmp_bitmap) {
2800 free_page(page);
2801 return -ENOMEM;
2802 }
2803 proc_skip_char(&kbuf, &left, '\n');
2804 while (!err && left) {
2805 unsigned long val_a, val_b;
2806 bool neg;
2807
2808 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2809 sizeof(tr_a), &c);
2810 if (err)
2811 break;
2812 if (val_a >= bitmap_len || neg) {
2813 err = -EINVAL;
2814 break;
2815 }
2816
2817 val_b = val_a;
2818 if (left) {
2819 kbuf++;
2820 left--;
2821 }
2822
2823 if (c == '-') {
2824 err = proc_get_long(&kbuf, &left, &val_b,
2825 &neg, tr_b, sizeof(tr_b),
2826 &c);
2827 if (err)
2828 break;
2829 if (val_b >= bitmap_len || neg ||
2830 val_a > val_b) {
2831 err = -EINVAL;
2832 break;
2833 }
2834 if (left) {
2835 kbuf++;
2836 left--;
2837 }
2838 }
2839
2840 while (val_a <= val_b)
2841 set_bit(val_a++, tmp_bitmap);
2842
2843 first = 0;
2844 proc_skip_char(&kbuf, &left, '\n');
2845 }
2846 free_page(page);
2847 } else {
2848 unsigned long bit_a, bit_b = 0;
2849
2850 while (left) {
2851 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2852 if (bit_a >= bitmap_len)
2853 break;
2854 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2855 bit_a + 1) - 1;
2856
2857 if (!first) {
2858 err = proc_put_char(&buffer, &left, ',');
2859 if (err)
2860 break;
2861 }
2862 err = proc_put_long(&buffer, &left, bit_a, false);
2863 if (err)
2864 break;
2865 if (bit_a != bit_b) {
2866 err = proc_put_char(&buffer, &left, '-');
2867 if (err)
2868 break;
2869 err = proc_put_long(&buffer, &left, bit_b, false);
2870 if (err)
2871 break;
2872 }
2873
2874 first = 0; bit_b++;
2875 }
2876 if (!err)
2877 err = proc_put_char(&buffer, &left, '\n');
2878 }
2879
2880 if (!err) {
2881 if (write) {
2882 if (*ppos)
2883 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2884 else
2885 memcpy(bitmap, tmp_bitmap,
2886 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2887 }
2888 kfree(tmp_bitmap);
2889 *lenp -= left;
2890 *ppos += *lenp;
2891 return 0;
2892 } else {
2893 kfree(tmp_bitmap);
2894 return err;
2895 }
2896}
2897
1da177e4
LT
2898#else /* CONFIG_PROC_FS */
2899
8d65af78 2900int proc_dostring(struct ctl_table *table, int write,
1da177e4
LT
2901 void __user *buffer, size_t *lenp, loff_t *ppos)
2902{
2903 return -ENOSYS;
2904}
2905
8d65af78 2906int proc_dointvec(struct ctl_table *table, int write,
1da177e4 2907 void __user *buffer, size_t *lenp, loff_t *ppos)
1da177e4
LT
2908{
2909 return -ENOSYS;
2910}
2911
8d65af78 2912int proc_dointvec_minmax(struct ctl_table *table, int write,
1da177e4
LT
2913 void __user *buffer, size_t *lenp, loff_t *ppos)
2914{
2915 return -ENOSYS;
2916}
2917
8d65af78 2918int proc_dointvec_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2919 void __user *buffer, size_t *lenp, loff_t *ppos)
2920{
2921 return -ENOSYS;
2922}
2923
8d65af78 2924int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2925 void __user *buffer, size_t *lenp, loff_t *ppos)
2926{
2927 return -ENOSYS;
2928}
2929
8d65af78 2930int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1da177e4
LT
2931 void __user *buffer, size_t *lenp, loff_t *ppos)
2932{
2933 return -ENOSYS;
2934}
2935
8d65af78 2936int proc_doulongvec_minmax(struct ctl_table *table, int write,
1da177e4
LT
2937 void __user *buffer, size_t *lenp, loff_t *ppos)
2938{
2939 return -ENOSYS;
2940}
2941
d8217f07 2942int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1da177e4
LT
2943 void __user *buffer,
2944 size_t *lenp, loff_t *ppos)
2945{
2946 return -ENOSYS;
2947}
2948
2949
2950#endif /* CONFIG_PROC_FS */
2951
1da177e4
LT
2952/*
2953 * No sense putting this after each symbol definition, twice,
2954 * exception granted :-)
2955 */
2956EXPORT_SYMBOL(proc_dointvec);
2957EXPORT_SYMBOL(proc_dointvec_jiffies);
2958EXPORT_SYMBOL(proc_dointvec_minmax);
2959EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2960EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2961EXPORT_SYMBOL(proc_dostring);
2962EXPORT_SYMBOL(proc_doulongvec_minmax);
2963EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2964EXPORT_SYMBOL(register_sysctl_table);
29e796fd 2965EXPORT_SYMBOL(register_sysctl_paths);
1da177e4 2966EXPORT_SYMBOL(unregister_sysctl_table);