projects / linux-2.6-block.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski...
[linux-2.6-block.git] / kernel / sysctl.c
... / ...
CommitLineData
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/aio.h>
23#include <linux/mm.h>
24#include <linux/swap.h>
25#include <linux/slab.h>
26#include <linux/sysctl.h>
27#include <linux/bitmap.h>
28#include <linux/signal.h>
29#include <linux/printk.h>
30#include <linux/proc_fs.h>
31#include <linux/security.h>
32#include <linux/ctype.h>
33#include <linux/kmemcheck.h>
34#include <linux/kmemleak.h>
35#include <linux/fs.h>
36#include <linux/init.h>
37#include <linux/kernel.h>
38#include <linux/kobject.h>
39#include <linux/net.h>
40#include <linux/sysrq.h>
41#include <linux/highuid.h>
42#include <linux/writeback.h>
43#include <linux/ratelimit.h>
44#include <linux/compaction.h>
45#include <linux/hugetlb.h>
46#include <linux/initrd.h>
47#include <linux/key.h>
48#include <linux/times.h>
49#include <linux/limits.h>
50#include <linux/dcache.h>
51#include <linux/dnotify.h>
52#include <linux/syscalls.h>
53#include <linux/vmstat.h>
54#include <linux/nfs_fs.h>
55#include <linux/acpi.h>
56#include <linux/reboot.h>
57#include <linux/ftrace.h>
58#include <linux/perf_event.h>
59#include <linux/kprobes.h>
60#include <linux/pipe_fs_i.h>
61#include <linux/oom.h>
62#include <linux/kmod.h>
63#include <linux/capability.h>
64#include <linux/binfmts.h>
65#include <linux/sched/sysctl.h>
66#include <linux/kexec.h>
67#include <linux/bpf.h>
68
69#include <asm/uaccess.h>
70#include <asm/processor.h>
71
72#ifdef CONFIG_X86
73#include <asm/nmi.h>
74#include <asm/stacktrace.h>
75#include <asm/io.h>
76#endif
77#ifdef CONFIG_SPARC
78#include <asm/setup.h>
79#endif
80#ifdef CONFIG_BSD_PROCESS_ACCT
81#include <linux/acct.h>
82#endif
83#ifdef CONFIG_RT_MUTEXES
84#include <linux/rtmutex.h>
85#endif
86#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
87#include <linux/lockdep.h>
88#endif
89#ifdef CONFIG_CHR_DEV_SG
90#include <scsi/sg.h>
91#endif
92
93#ifdef CONFIG_LOCKUP_DETECTOR
94#include <linux/nmi.h>
95#endif
96
97#if defined(CONFIG_SYSCTL)
98
99/* External variables not in a header file. */
100extern int suid_dumpable;
101#ifdef CONFIG_COREDUMP
102extern int core_uses_pid;
103extern char core_pattern[];
104extern unsigned int core_pipe_limit;
105#endif
106extern int pid_max;
107extern int pid_max_min, pid_max_max;
108extern int percpu_pagelist_fraction;
109extern int compat_log;
110extern int latencytop_enabled;
111extern int sysctl_nr_open_min, sysctl_nr_open_max;
112#ifndef CONFIG_MMU
113extern int sysctl_nr_trim_pages;
114#endif
115
116/* Constants used for minimum and maximum */
117#ifdef CONFIG_LOCKUP_DETECTOR
118static int sixty = 60;
119#endif
120
121static int __maybe_unused neg_one = -1;
122
123static int zero;
124static int __maybe_unused one = 1;
125static int __maybe_unused two = 2;
126static int __maybe_unused four = 4;
127static unsigned long one_ul = 1;
128static int one_hundred = 100;
129static int one_thousand = 1000;
130#ifdef CONFIG_PRINTK
131static int ten_thousand = 10000;
132#endif
133#ifdef CONFIG_PERF_EVENTS
134static int six_hundred_forty_kb = 640 * 1024;
135#endif
136
137/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139
140/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141static int maxolduid = 65535;
142static int minolduid;
143
144static int ngroups_max = NGROUPS_MAX;
145static const int cap_last_cap = CAP_LAST_CAP;
146
147/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
148#ifdef CONFIG_DETECT_HUNG_TASK
149static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
150#endif
151
152#ifdef CONFIG_INOTIFY_USER
153#include <linux/inotify.h>
154#endif
155#ifdef CONFIG_SPARC
156#endif
157
158#ifdef __hppa__
159extern int pwrsw_enabled;
160#endif
161
162#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
163extern int unaligned_enabled;
164#endif
165
166#ifdef CONFIG_IA64
167extern int unaligned_dump_stack;
168#endif
169
170#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
171extern int no_unaligned_warning;
172#endif
173
174#ifdef CONFIG_PROC_SYSCTL
175
176#define SYSCTL_WRITES_LEGACY -1
177#define SYSCTL_WRITES_WARN 0
178#define SYSCTL_WRITES_STRICT 1
179
180static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
181
182static int proc_do_cad_pid(struct ctl_table *table, int write,
183 void __user *buffer, size_t *lenp, loff_t *ppos);
184static int proc_taint(struct ctl_table *table, int write,
185 void __user *buffer, size_t *lenp, loff_t *ppos);
186#endif
187
188#ifdef CONFIG_PRINTK
189static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
190 void __user *buffer, size_t *lenp, loff_t *ppos);
191#endif
192
193static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
194 void __user *buffer, size_t *lenp, loff_t *ppos);
195#ifdef CONFIG_COREDUMP
196static int proc_dostring_coredump(struct ctl_table *table, int write,
197 void __user *buffer, size_t *lenp, loff_t *ppos);
198#endif
199
200#ifdef CONFIG_MAGIC_SYSRQ
201/* Note: sysrq code uses it's own private copy */
202static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
203
204static int sysrq_sysctl_handler(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp,
206 loff_t *ppos)
207{
208 int error;
209
210 error = proc_dointvec(table, write, buffer, lenp, ppos);
211 if (error)
212 return error;
213
214 if (write)
215 sysrq_toggle_support(__sysrq_enabled);
216
217 return 0;
218}
219
220#endif
221
222static struct ctl_table kern_table[];
223static struct ctl_table vm_table[];
224static struct ctl_table fs_table[];
225static struct ctl_table debug_table[];
226static struct ctl_table dev_table[];
227extern struct ctl_table random_table[];
228#ifdef CONFIG_EPOLL
229extern struct ctl_table epoll_table[];
230#endif
231
232#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
233int sysctl_legacy_va_layout;
234#endif
235
236/* The default sysctl tables: */
237
238static struct ctl_table sysctl_base_table[] = {
239 {
240 .procname = "kernel",
241 .mode = 0555,
242 .child = kern_table,
243 },
244 {
245 .procname = "vm",
246 .mode = 0555,
247 .child = vm_table,
248 },
249 {
250 .procname = "fs",
251 .mode = 0555,
252 .child = fs_table,
253 },
254 {
255 .procname = "debug",
256 .mode = 0555,
257 .child = debug_table,
258 },
259 {
260 .procname = "dev",
261 .mode = 0555,
262 .child = dev_table,
263 },
264 { }
265};
266
267#ifdef CONFIG_SCHED_DEBUG
268static int min_sched_granularity_ns = 100000; /* 100 usecs */
269static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
270static int min_wakeup_granularity_ns; /* 0 usecs */
271static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
272#ifdef CONFIG_SMP
273static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
274static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
275#endif /* CONFIG_SMP */
276#endif /* CONFIG_SCHED_DEBUG */
277
278#ifdef CONFIG_COMPACTION
279static int min_extfrag_threshold;
280static int max_extfrag_threshold = 1000;
281#endif
282
283static struct ctl_table kern_table[] = {
284 {
285 .procname = "sched_child_runs_first",
286 .data = &sysctl_sched_child_runs_first,
287 .maxlen = sizeof(unsigned int),
288 .mode = 0644,
289 .proc_handler = proc_dointvec,
290 },
291#ifdef CONFIG_SCHED_DEBUG
292 {
293 .procname = "sched_min_granularity_ns",
294 .data = &sysctl_sched_min_granularity,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_latency_ns",
303 .data = &sysctl_sched_latency,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_sched_granularity_ns,
308 .extra2 = &max_sched_granularity_ns,
309 },
310 {
311 .procname = "sched_wakeup_granularity_ns",
312 .data = &sysctl_sched_wakeup_granularity,
313 .maxlen = sizeof(unsigned int),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_wakeup_granularity_ns,
317 .extra2 = &max_wakeup_granularity_ns,
318 },
319#ifdef CONFIG_SMP
320 {
321 .procname = "sched_tunable_scaling",
322 .data = &sysctl_sched_tunable_scaling,
323 .maxlen = sizeof(enum sched_tunable_scaling),
324 .mode = 0644,
325 .proc_handler = sched_proc_update_handler,
326 .extra1 = &min_sched_tunable_scaling,
327 .extra2 = &max_sched_tunable_scaling,
328 },
329 {
330 .procname = "sched_migration_cost_ns",
331 .data = &sysctl_sched_migration_cost,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = proc_dointvec,
335 },
336 {
337 .procname = "sched_nr_migrate",
338 .data = &sysctl_sched_nr_migrate,
339 .maxlen = sizeof(unsigned int),
340 .mode = 0644,
341 .proc_handler = proc_dointvec,
342 },
343 {
344 .procname = "sched_time_avg_ms",
345 .data = &sysctl_sched_time_avg,
346 .maxlen = sizeof(unsigned int),
347 .mode = 0644,
348 .proc_handler = proc_dointvec,
349 },
350 {
351 .procname = "sched_shares_window_ns",
352 .data = &sysctl_sched_shares_window,
353 .maxlen = sizeof(unsigned int),
354 .mode = 0644,
355 .proc_handler = proc_dointvec,
356 },
357#ifdef CONFIG_SCHEDSTATS
358 {
359 .procname = "sched_schedstats",
360 .data = NULL,
361 .maxlen = sizeof(unsigned int),
362 .mode = 0644,
363 .proc_handler = sysctl_schedstats,
364 .extra1 = &zero,
365 .extra2 = &one,
366 },
367#endif /* CONFIG_SCHEDSTATS */
368#endif /* CONFIG_SMP */
369#ifdef CONFIG_NUMA_BALANCING
370 {
371 .procname = "numa_balancing_scan_delay_ms",
372 .data = &sysctl_numa_balancing_scan_delay,
373 .maxlen = sizeof(unsigned int),
374 .mode = 0644,
375 .proc_handler = proc_dointvec,
376 },
377 {
378 .procname = "numa_balancing_scan_period_min_ms",
379 .data = &sysctl_numa_balancing_scan_period_min,
380 .maxlen = sizeof(unsigned int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec,
383 },
384 {
385 .procname = "numa_balancing_scan_period_max_ms",
386 .data = &sysctl_numa_balancing_scan_period_max,
387 .maxlen = sizeof(unsigned int),
388 .mode = 0644,
389 .proc_handler = proc_dointvec,
390 },
391 {
392 .procname = "numa_balancing_scan_size_mb",
393 .data = &sysctl_numa_balancing_scan_size,
394 .maxlen = sizeof(unsigned int),
395 .mode = 0644,
396 .proc_handler = proc_dointvec_minmax,
397 .extra1 = &one,
398 },
399 {
400 .procname = "numa_balancing",
401 .data = NULL, /* filled in by handler */
402 .maxlen = sizeof(unsigned int),
403 .mode = 0644,
404 .proc_handler = sysctl_numa_balancing,
405 .extra1 = &zero,
406 .extra2 = &one,
407 },
408#endif /* CONFIG_NUMA_BALANCING */
409#endif /* CONFIG_SCHED_DEBUG */
410 {
411 .procname = "sched_rt_period_us",
412 .data = &sysctl_sched_rt_period,
413 .maxlen = sizeof(unsigned int),
414 .mode = 0644,
415 .proc_handler = sched_rt_handler,
416 },
417 {
418 .procname = "sched_rt_runtime_us",
419 .data = &sysctl_sched_rt_runtime,
420 .maxlen = sizeof(int),
421 .mode = 0644,
422 .proc_handler = sched_rt_handler,
423 },
424 {
425 .procname = "sched_rr_timeslice_ms",
426 .data = &sched_rr_timeslice,
427 .maxlen = sizeof(int),
428 .mode = 0644,
429 .proc_handler = sched_rr_handler,
430 },
431#ifdef CONFIG_SCHED_AUTOGROUP
432 {
433 .procname = "sched_autogroup_enabled",
434 .data = &sysctl_sched_autogroup_enabled,
435 .maxlen = sizeof(unsigned int),
436 .mode = 0644,
437 .proc_handler = proc_dointvec_minmax,
438 .extra1 = &zero,
439 .extra2 = &one,
440 },
441#endif
442#ifdef CONFIG_CFS_BANDWIDTH
443 {
444 .procname = "sched_cfs_bandwidth_slice_us",
445 .data = &sysctl_sched_cfs_bandwidth_slice,
446 .maxlen = sizeof(unsigned int),
447 .mode = 0644,
448 .proc_handler = proc_dointvec_minmax,
449 .extra1 = &one,
450 },
451#endif
452#ifdef CONFIG_PROVE_LOCKING
453 {
454 .procname = "prove_locking",
455 .data = &prove_locking,
456 .maxlen = sizeof(int),
457 .mode = 0644,
458 .proc_handler = proc_dointvec,
459 },
460#endif
461#ifdef CONFIG_LOCK_STAT
462 {
463 .procname = "lock_stat",
464 .data = &lock_stat,
465 .maxlen = sizeof(int),
466 .mode = 0644,
467 .proc_handler = proc_dointvec,
468 },
469#endif
470 {
471 .procname = "panic",
472 .data = &panic_timeout,
473 .maxlen = sizeof(int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477#ifdef CONFIG_COREDUMP
478 {
479 .procname = "core_uses_pid",
480 .data = &core_uses_pid,
481 .maxlen = sizeof(int),
482 .mode = 0644,
483 .proc_handler = proc_dointvec,
484 },
485 {
486 .procname = "core_pattern",
487 .data = core_pattern,
488 .maxlen = CORENAME_MAX_SIZE,
489 .mode = 0644,
490 .proc_handler = proc_dostring_coredump,
491 },
492 {
493 .procname = "core_pipe_limit",
494 .data = &core_pipe_limit,
495 .maxlen = sizeof(unsigned int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499#endif
500#ifdef CONFIG_PROC_SYSCTL
501 {
502 .procname = "tainted",
503 .maxlen = sizeof(long),
504 .mode = 0644,
505 .proc_handler = proc_taint,
506 },
507 {
508 .procname = "sysctl_writes_strict",
509 .data = &sysctl_writes_strict,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = proc_dointvec_minmax,
513 .extra1 = &neg_one,
514 .extra2 = &one,
515 },
516#endif
517#ifdef CONFIG_LATENCYTOP
518 {
519 .procname = "latencytop",
520 .data = &latencytop_enabled,
521 .maxlen = sizeof(int),
522 .mode = 0644,
523 .proc_handler = sysctl_latencytop,
524 },
525#endif
526#ifdef CONFIG_BLK_DEV_INITRD
527 {
528 .procname = "real-root-dev",
529 .data = &real_root_dev,
530 .maxlen = sizeof(int),
531 .mode = 0644,
532 .proc_handler = proc_dointvec,
533 },
534#endif
535 {
536 .procname = "print-fatal-signals",
537 .data = &print_fatal_signals,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#ifdef CONFIG_SPARC
543 {
544 .procname = "reboot-cmd",
545 .data = reboot_command,
546 .maxlen = 256,
547 .mode = 0644,
548 .proc_handler = proc_dostring,
549 },
550 {
551 .procname = "stop-a",
552 .data = &stop_a_enabled,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = proc_dointvec,
556 },
557 {
558 .procname = "scons-poweroff",
559 .data = &scons_pwroff,
560 .maxlen = sizeof (int),
561 .mode = 0644,
562 .proc_handler = proc_dointvec,
563 },
564#endif
565#ifdef CONFIG_SPARC64
566 {
567 .procname = "tsb-ratio",
568 .data = &sysctl_tsb_ratio,
569 .maxlen = sizeof (int),
570 .mode = 0644,
571 .proc_handler = proc_dointvec,
572 },
573#endif
574#ifdef __hppa__
575 {
576 .procname = "soft-power",
577 .data = &pwrsw_enabled,
578 .maxlen = sizeof (int),
579 .mode = 0644,
580 .proc_handler = proc_dointvec,
581 },
582#endif
583#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
584 {
585 .procname = "unaligned-trap",
586 .data = &unaligned_enabled,
587 .maxlen = sizeof (int),
588 .mode = 0644,
589 .proc_handler = proc_dointvec,
590 },
591#endif
592 {
593 .procname = "ctrl-alt-del",
594 .data = &C_A_D,
595 .maxlen = sizeof(int),
596 .mode = 0644,
597 .proc_handler = proc_dointvec,
598 },
599#ifdef CONFIG_FUNCTION_TRACER
600 {
601 .procname = "ftrace_enabled",
602 .data = &ftrace_enabled,
603 .maxlen = sizeof(int),
604 .mode = 0644,
605 .proc_handler = ftrace_enable_sysctl,
606 },
607#endif
608#ifdef CONFIG_STACK_TRACER
609 {
610 .procname = "stack_tracer_enabled",
611 .data = &stack_tracer_enabled,
612 .maxlen = sizeof(int),
613 .mode = 0644,
614 .proc_handler = stack_trace_sysctl,
615 },
616#endif
617#ifdef CONFIG_TRACING
618 {
619 .procname = "ftrace_dump_on_oops",
620 .data = &ftrace_dump_on_oops,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = proc_dointvec,
624 },
625 {
626 .procname = "traceoff_on_warning",
627 .data = &__disable_trace_on_warning,
628 .maxlen = sizeof(__disable_trace_on_warning),
629 .mode = 0644,
630 .proc_handler = proc_dointvec,
631 },
632 {
633 .procname = "tracepoint_printk",
634 .data = &tracepoint_printk,
635 .maxlen = sizeof(tracepoint_printk),
636 .mode = 0644,
637 .proc_handler = proc_dointvec,
638 },
639#endif
640#ifdef CONFIG_KEXEC_CORE
641 {
642 .procname = "kexec_load_disabled",
643 .data = &kexec_load_disabled,
644 .maxlen = sizeof(int),
645 .mode = 0644,
646 /* only handle a transition from default "0" to "1" */
647 .proc_handler = proc_dointvec_minmax,
648 .extra1 = &one,
649 .extra2 = &one,
650 },
651#endif
652#ifdef CONFIG_MODULES
653 {
654 .procname = "modprobe",
655 .data = &modprobe_path,
656 .maxlen = KMOD_PATH_LEN,
657 .mode = 0644,
658 .proc_handler = proc_dostring,
659 },
660 {
661 .procname = "modules_disabled",
662 .data = &modules_disabled,
663 .maxlen = sizeof(int),
664 .mode = 0644,
665 /* only handle a transition from default "0" to "1" */
666 .proc_handler = proc_dointvec_minmax,
667 .extra1 = &one,
668 .extra2 = &one,
669 },
670#endif
671#ifdef CONFIG_UEVENT_HELPER
672 {
673 .procname = "hotplug",
674 .data = &uevent_helper,
675 .maxlen = UEVENT_HELPER_PATH_LEN,
676 .mode = 0644,
677 .proc_handler = proc_dostring,
678 },
679#endif
680#ifdef CONFIG_CHR_DEV_SG
681 {
682 .procname = "sg-big-buff",
683 .data = &sg_big_buff,
684 .maxlen = sizeof (int),
685 .mode = 0444,
686 .proc_handler = proc_dointvec,
687 },
688#endif
689#ifdef CONFIG_BSD_PROCESS_ACCT
690 {
691 .procname = "acct",
692 .data = &acct_parm,
693 .maxlen = 3*sizeof(int),
694 .mode = 0644,
695 .proc_handler = proc_dointvec,
696 },
697#endif
698#ifdef CONFIG_MAGIC_SYSRQ
699 {
700 .procname = "sysrq",
701 .data = &__sysrq_enabled,
702 .maxlen = sizeof (int),
703 .mode = 0644,
704 .proc_handler = sysrq_sysctl_handler,
705 },
706#endif
707#ifdef CONFIG_PROC_SYSCTL
708 {
709 .procname = "cad_pid",
710 .data = NULL,
711 .maxlen = sizeof (int),
712 .mode = 0600,
713 .proc_handler = proc_do_cad_pid,
714 },
715#endif
716 {
717 .procname = "threads-max",
718 .data = NULL,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = sysctl_max_threads,
722 },
723 {
724 .procname = "random",
725 .mode = 0555,
726 .child = random_table,
727 },
728 {
729 .procname = "usermodehelper",
730 .mode = 0555,
731 .child = usermodehelper_table,
732 },
733 {
734 .procname = "overflowuid",
735 .data = &overflowuid,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = proc_dointvec_minmax,
739 .extra1 = &minolduid,
740 .extra2 = &maxolduid,
741 },
742 {
743 .procname = "overflowgid",
744 .data = &overflowgid,
745 .maxlen = sizeof(int),
746 .mode = 0644,
747 .proc_handler = proc_dointvec_minmax,
748 .extra1 = &minolduid,
749 .extra2 = &maxolduid,
750 },
751#ifdef CONFIG_S390
752#ifdef CONFIG_MATHEMU
753 {
754 .procname = "ieee_emulation_warnings",
755 .data = &sysctl_ieee_emulation_warnings,
756 .maxlen = sizeof(int),
757 .mode = 0644,
758 .proc_handler = proc_dointvec,
759 },
760#endif
761 {
762 .procname = "userprocess_debug",
763 .data = &show_unhandled_signals,
764 .maxlen = sizeof(int),
765 .mode = 0644,
766 .proc_handler = proc_dointvec,
767 },
768#endif
769 {
770 .procname = "pid_max",
771 .data = &pid_max,
772 .maxlen = sizeof (int),
773 .mode = 0644,
774 .proc_handler = proc_dointvec_minmax,
775 .extra1 = &pid_max_min,
776 .extra2 = &pid_max_max,
777 },
778 {
779 .procname = "panic_on_oops",
780 .data = &panic_on_oops,
781 .maxlen = sizeof(int),
782 .mode = 0644,
783 .proc_handler = proc_dointvec,
784 },
785#if defined CONFIG_PRINTK
786 {
787 .procname = "printk",
788 .data = &console_loglevel,
789 .maxlen = 4*sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec,
792 },
793 {
794 .procname = "printk_ratelimit",
795 .data = &printk_ratelimit_state.interval,
796 .maxlen = sizeof(int),
797 .mode = 0644,
798 .proc_handler = proc_dointvec_jiffies,
799 },
800 {
801 .procname = "printk_ratelimit_burst",
802 .data = &printk_ratelimit_state.burst,
803 .maxlen = sizeof(int),
804 .mode = 0644,
805 .proc_handler = proc_dointvec,
806 },
807 {
808 .procname = "printk_delay",
809 .data = &printk_delay_msec,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec_minmax,
813 .extra1 = &zero,
814 .extra2 = &ten_thousand,
815 },
816 {
817 .procname = "dmesg_restrict",
818 .data = &dmesg_restrict,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec_minmax_sysadmin,
822 .extra1 = &zero,
823 .extra2 = &one,
824 },
825 {
826 .procname = "kptr_restrict",
827 .data = &kptr_restrict,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec_minmax_sysadmin,
831 .extra1 = &zero,
832 .extra2 = &two,
833 },
834#endif
835 {
836 .procname = "ngroups_max",
837 .data = &ngroups_max,
838 .maxlen = sizeof (int),
839 .mode = 0444,
840 .proc_handler = proc_dointvec,
841 },
842 {
843 .procname = "cap_last_cap",
844 .data = (void *)&cap_last_cap,
845 .maxlen = sizeof(int),
846 .mode = 0444,
847 .proc_handler = proc_dointvec,
848 },
849#if defined(CONFIG_LOCKUP_DETECTOR)
850 {
851 .procname = "watchdog",
852 .data = &watchdog_user_enabled,
853 .maxlen = sizeof (int),
854 .mode = 0644,
855 .proc_handler = proc_watchdog,
856 .extra1 = &zero,
857 .extra2 = &one,
858 },
859 {
860 .procname = "watchdog_thresh",
861 .data = &watchdog_thresh,
862 .maxlen = sizeof(int),
863 .mode = 0644,
864 .proc_handler = proc_watchdog_thresh,
865 .extra1 = &zero,
866 .extra2 = &sixty,
867 },
868 {
869 .procname = "nmi_watchdog",
870 .data = &nmi_watchdog_enabled,
871 .maxlen = sizeof (int),
872 .mode = 0644,
873 .proc_handler = proc_nmi_watchdog,
874 .extra1 = &zero,
875#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
876 .extra2 = &one,
877#else
878 .extra2 = &zero,
879#endif
880 },
881 {
882 .procname = "soft_watchdog",
883 .data = &soft_watchdog_enabled,
884 .maxlen = sizeof (int),
885 .mode = 0644,
886 .proc_handler = proc_soft_watchdog,
887 .extra1 = &zero,
888 .extra2 = &one,
889 },
890 {
891 .procname = "watchdog_cpumask",
892 .data = &watchdog_cpumask_bits,
893 .maxlen = NR_CPUS,
894 .mode = 0644,
895 .proc_handler = proc_watchdog_cpumask,
896 },
897 {
898 .procname = "softlockup_panic",
899 .data = &softlockup_panic,
900 .maxlen = sizeof(int),
901 .mode = 0644,
902 .proc_handler = proc_dointvec_minmax,
903 .extra1 = &zero,
904 .extra2 = &one,
905 },
906#ifdef CONFIG_HARDLOCKUP_DETECTOR
907 {
908 .procname = "hardlockup_panic",
909 .data = &hardlockup_panic,
910 .maxlen = sizeof(int),
911 .mode = 0644,
912 .proc_handler = proc_dointvec_minmax,
913 .extra1 = &zero,
914 .extra2 = &one,
915 },
916#endif
917#ifdef CONFIG_SMP
918 {
919 .procname = "softlockup_all_cpu_backtrace",
920 .data = &sysctl_softlockup_all_cpu_backtrace,
921 .maxlen = sizeof(int),
922 .mode = 0644,
923 .proc_handler = proc_dointvec_minmax,
924 .extra1 = &zero,
925 .extra2 = &one,
926 },
927 {
928 .procname = "hardlockup_all_cpu_backtrace",
929 .data = &sysctl_hardlockup_all_cpu_backtrace,
930 .maxlen = sizeof(int),
931 .mode = 0644,
932 .proc_handler = proc_dointvec_minmax,
933 .extra1 = &zero,
934 .extra2 = &one,
935 },
936#endif /* CONFIG_SMP */
937#endif
938#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
939 {
940 .procname = "unknown_nmi_panic",
941 .data = &unknown_nmi_panic,
942 .maxlen = sizeof (int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec,
945 },
946#endif
947#if defined(CONFIG_X86)
948 {
949 .procname = "panic_on_unrecovered_nmi",
950 .data = &panic_on_unrecovered_nmi,
951 .maxlen = sizeof(int),
952 .mode = 0644,
953 .proc_handler = proc_dointvec,
954 },
955 {
956 .procname = "panic_on_io_nmi",
957 .data = &panic_on_io_nmi,
958 .maxlen = sizeof(int),
959 .mode = 0644,
960 .proc_handler = proc_dointvec,
961 },
962#ifdef CONFIG_DEBUG_STACKOVERFLOW
963 {
964 .procname = "panic_on_stackoverflow",
965 .data = &sysctl_panic_on_stackoverflow,
966 .maxlen = sizeof(int),
967 .mode = 0644,
968 .proc_handler = proc_dointvec,
969 },
970#endif
971 {
972 .procname = "bootloader_type",
973 .data = &bootloader_type,
974 .maxlen = sizeof (int),
975 .mode = 0444,
976 .proc_handler = proc_dointvec,
977 },
978 {
979 .procname = "bootloader_version",
980 .data = &bootloader_version,
981 .maxlen = sizeof (int),
982 .mode = 0444,
983 .proc_handler = proc_dointvec,
984 },
985 {
986 .procname = "kstack_depth_to_print",
987 .data = &kstack_depth_to_print,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992 {
993 .procname = "io_delay_type",
994 .data = &io_delay_type,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999#endif
1000#if defined(CONFIG_MMU)
1001 {
1002 .procname = "randomize_va_space",
1003 .data = &randomize_va_space,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec,
1007 },
1008#endif
1009#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1010 {
1011 .procname = "spin_retry",
1012 .data = &spin_retry,
1013 .maxlen = sizeof (int),
1014 .mode = 0644,
1015 .proc_handler = proc_dointvec,
1016 },
1017#endif
1018#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1019 {
1020 .procname = "acpi_video_flags",
1021 .data = &acpi_realmode_flags,
1022 .maxlen = sizeof (unsigned long),
1023 .mode = 0644,
1024 .proc_handler = proc_doulongvec_minmax,
1025 },
1026#endif
1027#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1028 {
1029 .procname = "ignore-unaligned-usertrap",
1030 .data = &no_unaligned_warning,
1031 .maxlen = sizeof (int),
1032 .mode = 0644,
1033 .proc_handler = proc_dointvec,
1034 },
1035#endif
1036#ifdef CONFIG_IA64
1037 {
1038 .procname = "unaligned-dump-stack",
1039 .data = &unaligned_dump_stack,
1040 .maxlen = sizeof (int),
1041 .mode = 0644,
1042 .proc_handler = proc_dointvec,
1043 },
1044#endif
1045#ifdef CONFIG_DETECT_HUNG_TASK
1046 {
1047 .procname = "hung_task_panic",
1048 .data = &sysctl_hung_task_panic,
1049 .maxlen = sizeof(int),
1050 .mode = 0644,
1051 .proc_handler = proc_dointvec_minmax,
1052 .extra1 = &zero,
1053 .extra2 = &one,
1054 },
1055 {
1056 .procname = "hung_task_check_count",
1057 .data = &sysctl_hung_task_check_count,
1058 .maxlen = sizeof(int),
1059 .mode = 0644,
1060 .proc_handler = proc_dointvec_minmax,
1061 .extra1 = &zero,
1062 },
1063 {
1064 .procname = "hung_task_timeout_secs",
1065 .data = &sysctl_hung_task_timeout_secs,
1066 .maxlen = sizeof(unsigned long),
1067 .mode = 0644,
1068 .proc_handler = proc_dohung_task_timeout_secs,
1069 .extra2 = &hung_task_timeout_max,
1070 },
1071 {
1072 .procname = "hung_task_warnings",
1073 .data = &sysctl_hung_task_warnings,
1074 .maxlen = sizeof(int),
1075 .mode = 0644,
1076 .proc_handler = proc_dointvec_minmax,
1077 .extra1 = &neg_one,
1078 },
1079#endif
1080#ifdef CONFIG_COMPAT
1081 {
1082 .procname = "compat-log",
1083 .data = &compat_log,
1084 .maxlen = sizeof (int),
1085 .mode = 0644,
1086 .proc_handler = proc_dointvec,
1087 },
1088#endif
1089#ifdef CONFIG_RT_MUTEXES
1090 {
1091 .procname = "max_lock_depth",
1092 .data = &max_lock_depth,
1093 .maxlen = sizeof(int),
1094 .mode = 0644,
1095 .proc_handler = proc_dointvec,
1096 },
1097#endif
1098 {
1099 .procname = "poweroff_cmd",
1100 .data = &poweroff_cmd,
1101 .maxlen = POWEROFF_CMD_PATH_LEN,
1102 .mode = 0644,
1103 .proc_handler = proc_dostring,
1104 },
1105#ifdef CONFIG_KEYS
1106 {
1107 .procname = "keys",
1108 .mode = 0555,
1109 .child = key_sysctls,
1110 },
1111#endif
1112#ifdef CONFIG_PERF_EVENTS
1113 /*
1114 * User-space scripts rely on the existence of this file
1115 * as a feature check for perf_events being enabled.
1116 *
1117 * So it's an ABI, do not remove!
1118 */
1119 {
1120 .procname = "perf_event_paranoid",
1121 .data = &sysctl_perf_event_paranoid,
1122 .maxlen = sizeof(sysctl_perf_event_paranoid),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec,
1125 },
1126 {
1127 .procname = "perf_event_mlock_kb",
1128 .data = &sysctl_perf_event_mlock,
1129 .maxlen = sizeof(sysctl_perf_event_mlock),
1130 .mode = 0644,
1131 .proc_handler = proc_dointvec,
1132 },
1133 {
1134 .procname = "perf_event_max_sample_rate",
1135 .data = &sysctl_perf_event_sample_rate,
1136 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1137 .mode = 0644,
1138 .proc_handler = perf_proc_update_handler,
1139 .extra1 = &one,
1140 },
1141 {
1142 .procname = "perf_cpu_time_max_percent",
1143 .data = &sysctl_perf_cpu_time_max_percent,
1144 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1145 .mode = 0644,
1146 .proc_handler = perf_cpu_time_max_percent_handler,
1147 .extra1 = &zero,
1148 .extra2 = &one_hundred,
1149 },
1150 {
1151 .procname = "perf_event_max_stack",
1152 .data = NULL, /* filled in by handler */
1153 .maxlen = sizeof(sysctl_perf_event_max_stack),
1154 .mode = 0644,
1155 .proc_handler = perf_event_max_stack_handler,
1156 .extra1 = &zero,
1157 .extra2 = &six_hundred_forty_kb,
1158 },
1159#endif
1160#ifdef CONFIG_KMEMCHECK
1161 {
1162 .procname = "kmemcheck",
1163 .data = &kmemcheck_enabled,
1164 .maxlen = sizeof(int),
1165 .mode = 0644,
1166 .proc_handler = proc_dointvec,
1167 },
1168#endif
1169 {
1170 .procname = "panic_on_warn",
1171 .data = &panic_on_warn,
1172 .maxlen = sizeof(int),
1173 .mode = 0644,
1174 .proc_handler = proc_dointvec_minmax,
1175 .extra1 = &zero,
1176 .extra2 = &one,
1177 },
1178#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1179 {
1180 .procname = "timer_migration",
1181 .data = &sysctl_timer_migration,
1182 .maxlen = sizeof(unsigned int),
1183 .mode = 0644,
1184 .proc_handler = timer_migration_handler,
1185 },
1186#endif
1187#ifdef CONFIG_BPF_SYSCALL
1188 {
1189 .procname = "unprivileged_bpf_disabled",
1190 .data = &sysctl_unprivileged_bpf_disabled,
1191 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1192 .mode = 0644,
1193 /* only handle a transition from default "0" to "1" */
1194 .proc_handler = proc_dointvec_minmax,
1195 .extra1 = &one,
1196 .extra2 = &one,
1197 },
1198#endif
1199 { }
1200};
1201
1202static struct ctl_table vm_table[] = {
1203 {
1204 .procname = "overcommit_memory",
1205 .data = &sysctl_overcommit_memory,
1206 .maxlen = sizeof(sysctl_overcommit_memory),
1207 .mode = 0644,
1208 .proc_handler = proc_dointvec_minmax,
1209 .extra1 = &zero,
1210 .extra2 = &two,
1211 },
1212 {
1213 .procname = "panic_on_oom",
1214 .data = &sysctl_panic_on_oom,
1215 .maxlen = sizeof(sysctl_panic_on_oom),
1216 .mode = 0644,
1217 .proc_handler = proc_dointvec_minmax,
1218 .extra1 = &zero,
1219 .extra2 = &two,
1220 },
1221 {
1222 .procname = "oom_kill_allocating_task",
1223 .data = &sysctl_oom_kill_allocating_task,
1224 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1225 .mode = 0644,
1226 .proc_handler = proc_dointvec,
1227 },
1228 {
1229 .procname = "oom_dump_tasks",
1230 .data = &sysctl_oom_dump_tasks,
1231 .maxlen = sizeof(sysctl_oom_dump_tasks),
1232 .mode = 0644,
1233 .proc_handler = proc_dointvec,
1234 },
1235 {
1236 .procname = "overcommit_ratio",
1237 .data = &sysctl_overcommit_ratio,
1238 .maxlen = sizeof(sysctl_overcommit_ratio),
1239 .mode = 0644,
1240 .proc_handler = overcommit_ratio_handler,
1241 },
1242 {
1243 .procname = "overcommit_kbytes",
1244 .data = &sysctl_overcommit_kbytes,
1245 .maxlen = sizeof(sysctl_overcommit_kbytes),
1246 .mode = 0644,
1247 .proc_handler = overcommit_kbytes_handler,
1248 },
1249 {
1250 .procname = "page-cluster",
1251 .data = &page_cluster,
1252 .maxlen = sizeof(int),
1253 .mode = 0644,
1254 .proc_handler = proc_dointvec_minmax,
1255 .extra1 = &zero,
1256 },
1257 {
1258 .procname = "dirty_background_ratio",
1259 .data = &dirty_background_ratio,
1260 .maxlen = sizeof(dirty_background_ratio),
1261 .mode = 0644,
1262 .proc_handler = dirty_background_ratio_handler,
1263 .extra1 = &zero,
1264 .extra2 = &one_hundred,
1265 },
1266 {
1267 .procname = "dirty_background_bytes",
1268 .data = &dirty_background_bytes,
1269 .maxlen = sizeof(dirty_background_bytes),
1270 .mode = 0644,
1271 .proc_handler = dirty_background_bytes_handler,
1272 .extra1 = &one_ul,
1273 },
1274 {
1275 .procname = "dirty_ratio",
1276 .data = &vm_dirty_ratio,
1277 .maxlen = sizeof(vm_dirty_ratio),
1278 .mode = 0644,
1279 .proc_handler = dirty_ratio_handler,
1280 .extra1 = &zero,
1281 .extra2 = &one_hundred,
1282 },
1283 {
1284 .procname = "dirty_bytes",
1285 .data = &vm_dirty_bytes,
1286 .maxlen = sizeof(vm_dirty_bytes),
1287 .mode = 0644,
1288 .proc_handler = dirty_bytes_handler,
1289 .extra1 = &dirty_bytes_min,
1290 },
1291 {
1292 .procname = "dirty_writeback_centisecs",
1293 .data = &dirty_writeback_interval,
1294 .maxlen = sizeof(dirty_writeback_interval),
1295 .mode = 0644,
1296 .proc_handler = dirty_writeback_centisecs_handler,
1297 },
1298 {
1299 .procname = "dirty_expire_centisecs",
1300 .data = &dirty_expire_interval,
1301 .maxlen = sizeof(dirty_expire_interval),
1302 .mode = 0644,
1303 .proc_handler = proc_dointvec_minmax,
1304 .extra1 = &zero,
1305 },
1306 {
1307 .procname = "dirtytime_expire_seconds",
1308 .data = &dirtytime_expire_interval,
1309 .maxlen = sizeof(dirty_expire_interval),
1310 .mode = 0644,
1311 .proc_handler = dirtytime_interval_handler,
1312 .extra1 = &zero,
1313 },
1314 {
1315 .procname = "nr_pdflush_threads",
1316 .mode = 0444 /* read-only */,
1317 .proc_handler = pdflush_proc_obsolete,
1318 },
1319 {
1320 .procname = "swappiness",
1321 .data = &vm_swappiness,
1322 .maxlen = sizeof(vm_swappiness),
1323 .mode = 0644,
1324 .proc_handler = proc_dointvec_minmax,
1325 .extra1 = &zero,
1326 .extra2 = &one_hundred,
1327 },
1328#ifdef CONFIG_HUGETLB_PAGE
1329 {
1330 .procname = "nr_hugepages",
1331 .data = NULL,
1332 .maxlen = sizeof(unsigned long),
1333 .mode = 0644,
1334 .proc_handler = hugetlb_sysctl_handler,
1335 },
1336#ifdef CONFIG_NUMA
1337 {
1338 .procname = "nr_hugepages_mempolicy",
1339 .data = NULL,
1340 .maxlen = sizeof(unsigned long),
1341 .mode = 0644,
1342 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1343 },
1344#endif
1345 {
1346 .procname = "hugetlb_shm_group",
1347 .data = &sysctl_hugetlb_shm_group,
1348 .maxlen = sizeof(gid_t),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec,
1351 },
1352 {
1353 .procname = "hugepages_treat_as_movable",
1354 .data = &hugepages_treat_as_movable,
1355 .maxlen = sizeof(int),
1356 .mode = 0644,
1357 .proc_handler = proc_dointvec,
1358 },
1359 {
1360 .procname = "nr_overcommit_hugepages",
1361 .data = NULL,
1362 .maxlen = sizeof(unsigned long),
1363 .mode = 0644,
1364 .proc_handler = hugetlb_overcommit_handler,
1365 },
1366#endif
1367 {
1368 .procname = "lowmem_reserve_ratio",
1369 .data = &sysctl_lowmem_reserve_ratio,
1370 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1371 .mode = 0644,
1372 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1373 },
1374 {
1375 .procname = "drop_caches",
1376 .data = &sysctl_drop_caches,
1377 .maxlen = sizeof(int),
1378 .mode = 0644,
1379 .proc_handler = drop_caches_sysctl_handler,
1380 .extra1 = &one,
1381 .extra2 = &four,
1382 },
1383#ifdef CONFIG_COMPACTION
1384 {
1385 .procname = "compact_memory",
1386 .data = &sysctl_compact_memory,
1387 .maxlen = sizeof(int),
1388 .mode = 0200,
1389 .proc_handler = sysctl_compaction_handler,
1390 },
1391 {
1392 .procname = "extfrag_threshold",
1393 .data = &sysctl_extfrag_threshold,
1394 .maxlen = sizeof(int),
1395 .mode = 0644,
1396 .proc_handler = sysctl_extfrag_handler,
1397 .extra1 = &min_extfrag_threshold,
1398 .extra2 = &max_extfrag_threshold,
1399 },
1400 {
1401 .procname = "compact_unevictable_allowed",
1402 .data = &sysctl_compact_unevictable_allowed,
1403 .maxlen = sizeof(int),
1404 .mode = 0644,
1405 .proc_handler = proc_dointvec,
1406 .extra1 = &zero,
1407 .extra2 = &one,
1408 },
1409
1410#endif /* CONFIG_COMPACTION */
1411 {
1412 .procname = "min_free_kbytes",
1413 .data = &min_free_kbytes,
1414 .maxlen = sizeof(min_free_kbytes),
1415 .mode = 0644,
1416 .proc_handler = min_free_kbytes_sysctl_handler,
1417 .extra1 = &zero,
1418 },
1419 {
1420 .procname = "watermark_scale_factor",
1421 .data = &watermark_scale_factor,
1422 .maxlen = sizeof(watermark_scale_factor),
1423 .mode = 0644,
1424 .proc_handler = watermark_scale_factor_sysctl_handler,
1425 .extra1 = &one,
1426 .extra2 = &one_thousand,
1427 },
1428 {
1429 .procname = "percpu_pagelist_fraction",
1430 .data = &percpu_pagelist_fraction,
1431 .maxlen = sizeof(percpu_pagelist_fraction),
1432 .mode = 0644,
1433 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1434 .extra1 = &zero,
1435 },
1436#ifdef CONFIG_MMU
1437 {
1438 .procname = "max_map_count",
1439 .data = &sysctl_max_map_count,
1440 .maxlen = sizeof(sysctl_max_map_count),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec_minmax,
1443 .extra1 = &zero,
1444 },
1445#else
1446 {
1447 .procname = "nr_trim_pages",
1448 .data = &sysctl_nr_trim_pages,
1449 .maxlen = sizeof(sysctl_nr_trim_pages),
1450 .mode = 0644,
1451 .proc_handler = proc_dointvec_minmax,
1452 .extra1 = &zero,
1453 },
1454#endif
1455 {
1456 .procname = "laptop_mode",
1457 .data = &laptop_mode,
1458 .maxlen = sizeof(laptop_mode),
1459 .mode = 0644,
1460 .proc_handler = proc_dointvec_jiffies,
1461 },
1462 {
1463 .procname = "block_dump",
1464 .data = &block_dump,
1465 .maxlen = sizeof(block_dump),
1466 .mode = 0644,
1467 .proc_handler = proc_dointvec,
1468 .extra1 = &zero,
1469 },
1470 {
1471 .procname = "vfs_cache_pressure",
1472 .data = &sysctl_vfs_cache_pressure,
1473 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1474 .mode = 0644,
1475 .proc_handler = proc_dointvec,
1476 .extra1 = &zero,
1477 },
1478#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1479 {
1480 .procname = "legacy_va_layout",
1481 .data = &sysctl_legacy_va_layout,
1482 .maxlen = sizeof(sysctl_legacy_va_layout),
1483 .mode = 0644,
1484 .proc_handler = proc_dointvec,
1485 .extra1 = &zero,
1486 },
1487#endif
1488#ifdef CONFIG_NUMA
1489 {
1490 .procname = "zone_reclaim_mode",
1491 .data = &zone_reclaim_mode,
1492 .maxlen = sizeof(zone_reclaim_mode),
1493 .mode = 0644,
1494 .proc_handler = proc_dointvec,
1495 .extra1 = &zero,
1496 },
1497 {
1498 .procname = "min_unmapped_ratio",
1499 .data = &sysctl_min_unmapped_ratio,
1500 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1501 .mode = 0644,
1502 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1503 .extra1 = &zero,
1504 .extra2 = &one_hundred,
1505 },
1506 {
1507 .procname = "min_slab_ratio",
1508 .data = &sysctl_min_slab_ratio,
1509 .maxlen = sizeof(sysctl_min_slab_ratio),
1510 .mode = 0644,
1511 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1512 .extra1 = &zero,
1513 .extra2 = &one_hundred,
1514 },
1515#endif
1516#ifdef CONFIG_SMP
1517 {
1518 .procname = "stat_interval",
1519 .data = &sysctl_stat_interval,
1520 .maxlen = sizeof(sysctl_stat_interval),
1521 .mode = 0644,
1522 .proc_handler = proc_dointvec_jiffies,
1523 },
1524#endif
1525#ifdef CONFIG_MMU
1526 {
1527 .procname = "mmap_min_addr",
1528 .data = &dac_mmap_min_addr,
1529 .maxlen = sizeof(unsigned long),
1530 .mode = 0644,
1531 .proc_handler = mmap_min_addr_handler,
1532 },
1533#endif
1534#ifdef CONFIG_NUMA
1535 {
1536 .procname = "numa_zonelist_order",
1537 .data = &numa_zonelist_order,
1538 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1539 .mode = 0644,
1540 .proc_handler = numa_zonelist_order_handler,
1541 },
1542#endif
1543#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1544 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1545 {
1546 .procname = "vdso_enabled",
1547#ifdef CONFIG_X86_32
1548 .data = &vdso32_enabled,
1549 .maxlen = sizeof(vdso32_enabled),
1550#else
1551 .data = &vdso_enabled,
1552 .maxlen = sizeof(vdso_enabled),
1553#endif
1554 .mode = 0644,
1555 .proc_handler = proc_dointvec,
1556 .extra1 = &zero,
1557 },
1558#endif
1559#ifdef CONFIG_HIGHMEM
1560 {
1561 .procname = "highmem_is_dirtyable",
1562 .data = &vm_highmem_is_dirtyable,
1563 .maxlen = sizeof(vm_highmem_is_dirtyable),
1564 .mode = 0644,
1565 .proc_handler = proc_dointvec_minmax,
1566 .extra1 = &zero,
1567 .extra2 = &one,
1568 },
1569#endif
1570#ifdef CONFIG_MEMORY_FAILURE
1571 {
1572 .procname = "memory_failure_early_kill",
1573 .data = &sysctl_memory_failure_early_kill,
1574 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1575 .mode = 0644,
1576 .proc_handler = proc_dointvec_minmax,
1577 .extra1 = &zero,
1578 .extra2 = &one,
1579 },
1580 {
1581 .procname = "memory_failure_recovery",
1582 .data = &sysctl_memory_failure_recovery,
1583 .maxlen = sizeof(sysctl_memory_failure_recovery),
1584 .mode = 0644,
1585 .proc_handler = proc_dointvec_minmax,
1586 .extra1 = &zero,
1587 .extra2 = &one,
1588 },
1589#endif
1590 {
1591 .procname = "user_reserve_kbytes",
1592 .data = &sysctl_user_reserve_kbytes,
1593 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1594 .mode = 0644,
1595 .proc_handler = proc_doulongvec_minmax,
1596 },
1597 {
1598 .procname = "admin_reserve_kbytes",
1599 .data = &sysctl_admin_reserve_kbytes,
1600 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1601 .mode = 0644,
1602 .proc_handler = proc_doulongvec_minmax,
1603 },
1604#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1605 {
1606 .procname = "mmap_rnd_bits",
1607 .data = &mmap_rnd_bits,
1608 .maxlen = sizeof(mmap_rnd_bits),
1609 .mode = 0600,
1610 .proc_handler = proc_dointvec_minmax,
1611 .extra1 = (void *)&mmap_rnd_bits_min,
1612 .extra2 = (void *)&mmap_rnd_bits_max,
1613 },
1614#endif
1615#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1616 {
1617 .procname = "mmap_rnd_compat_bits",
1618 .data = &mmap_rnd_compat_bits,
1619 .maxlen = sizeof(mmap_rnd_compat_bits),
1620 .mode = 0600,
1621 .proc_handler = proc_dointvec_minmax,
1622 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1623 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1624 },
1625#endif
1626 { }
1627};
1628
1629static struct ctl_table fs_table[] = {
1630 {
1631 .procname = "inode-nr",
1632 .data = &inodes_stat,
1633 .maxlen = 2*sizeof(long),
1634 .mode = 0444,
1635 .proc_handler = proc_nr_inodes,
1636 },
1637 {
1638 .procname = "inode-state",
1639 .data = &inodes_stat,
1640 .maxlen = 7*sizeof(long),
1641 .mode = 0444,
1642 .proc_handler = proc_nr_inodes,
1643 },
1644 {
1645 .procname = "file-nr",
1646 .data = &files_stat,
1647 .maxlen = sizeof(files_stat),
1648 .mode = 0444,
1649 .proc_handler = proc_nr_files,
1650 },
1651 {
1652 .procname = "file-max",
1653 .data = &files_stat.max_files,
1654 .maxlen = sizeof(files_stat.max_files),
1655 .mode = 0644,
1656 .proc_handler = proc_doulongvec_minmax,
1657 },
1658 {
1659 .procname = "nr_open",
1660 .data = &sysctl_nr_open,
1661 .maxlen = sizeof(int),
1662 .mode = 0644,
1663 .proc_handler = proc_dointvec_minmax,
1664 .extra1 = &sysctl_nr_open_min,
1665 .extra2 = &sysctl_nr_open_max,
1666 },
1667 {
1668 .procname = "dentry-state",
1669 .data = &dentry_stat,
1670 .maxlen = 6*sizeof(long),
1671 .mode = 0444,
1672 .proc_handler = proc_nr_dentry,
1673 },
1674 {
1675 .procname = "overflowuid",
1676 .data = &fs_overflowuid,
1677 .maxlen = sizeof(int),
1678 .mode = 0644,
1679 .proc_handler = proc_dointvec_minmax,
1680 .extra1 = &minolduid,
1681 .extra2 = &maxolduid,
1682 },
1683 {
1684 .procname = "overflowgid",
1685 .data = &fs_overflowgid,
1686 .maxlen = sizeof(int),
1687 .mode = 0644,
1688 .proc_handler = proc_dointvec_minmax,
1689 .extra1 = &minolduid,
1690 .extra2 = &maxolduid,
1691 },
1692#ifdef CONFIG_FILE_LOCKING
1693 {
1694 .procname = "leases-enable",
1695 .data = &leases_enable,
1696 .maxlen = sizeof(int),
1697 .mode = 0644,
1698 .proc_handler = proc_dointvec,
1699 },
1700#endif
1701#ifdef CONFIG_DNOTIFY
1702 {
1703 .procname = "dir-notify-enable",
1704 .data = &dir_notify_enable,
1705 .maxlen = sizeof(int),
1706 .mode = 0644,
1707 .proc_handler = proc_dointvec,
1708 },
1709#endif
1710#ifdef CONFIG_MMU
1711#ifdef CONFIG_FILE_LOCKING
1712 {
1713 .procname = "lease-break-time",
1714 .data = &lease_break_time,
1715 .maxlen = sizeof(int),
1716 .mode = 0644,
1717 .proc_handler = proc_dointvec,
1718 },
1719#endif
1720#ifdef CONFIG_AIO
1721 {
1722 .procname = "aio-nr",
1723 .data = &aio_nr,
1724 .maxlen = sizeof(aio_nr),
1725 .mode = 0444,
1726 .proc_handler = proc_doulongvec_minmax,
1727 },
1728 {
1729 .procname = "aio-max-nr",
1730 .data = &aio_max_nr,
1731 .maxlen = sizeof(aio_max_nr),
1732 .mode = 0644,
1733 .proc_handler = proc_doulongvec_minmax,
1734 },
1735#endif /* CONFIG_AIO */
1736#ifdef CONFIG_INOTIFY_USER
1737 {
1738 .procname = "inotify",
1739 .mode = 0555,
1740 .child = inotify_table,
1741 },
1742#endif
1743#ifdef CONFIG_EPOLL
1744 {
1745 .procname = "epoll",
1746 .mode = 0555,
1747 .child = epoll_table,
1748 },
1749#endif
1750#endif
1751 {
1752 .procname = "protected_symlinks",
1753 .data = &sysctl_protected_symlinks,
1754 .maxlen = sizeof(int),
1755 .mode = 0600,
1756 .proc_handler = proc_dointvec_minmax,
1757 .extra1 = &zero,
1758 .extra2 = &one,
1759 },
1760 {
1761 .procname = "protected_hardlinks",
1762 .data = &sysctl_protected_hardlinks,
1763 .maxlen = sizeof(int),
1764 .mode = 0600,
1765 .proc_handler = proc_dointvec_minmax,
1766 .extra1 = &zero,
1767 .extra2 = &one,
1768 },
1769 {
1770 .procname = "suid_dumpable",
1771 .data = &suid_dumpable,
1772 .maxlen = sizeof(int),
1773 .mode = 0644,
1774 .proc_handler = proc_dointvec_minmax_coredump,
1775 .extra1 = &zero,
1776 .extra2 = &two,
1777 },
1778#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1779 {
1780 .procname = "binfmt_misc",
1781 .mode = 0555,
1782 .child = sysctl_mount_point,
1783 },
1784#endif
1785 {
1786 .procname = "pipe-max-size",
1787 .data = &pipe_max_size,
1788 .maxlen = sizeof(int),
1789 .mode = 0644,
1790 .proc_handler = &pipe_proc_fn,
1791 .extra1 = &pipe_min_size,
1792 },
1793 {
1794 .procname = "pipe-user-pages-hard",
1795 .data = &pipe_user_pages_hard,
1796 .maxlen = sizeof(pipe_user_pages_hard),
1797 .mode = 0644,
1798 .proc_handler = proc_doulongvec_minmax,
1799 },
1800 {
1801 .procname = "pipe-user-pages-soft",
1802 .data = &pipe_user_pages_soft,
1803 .maxlen = sizeof(pipe_user_pages_soft),
1804 .mode = 0644,
1805 .proc_handler = proc_doulongvec_minmax,
1806 },
1807 { }
1808};
1809
1810static struct ctl_table debug_table[] = {
1811#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1812 {
1813 .procname = "exception-trace",
1814 .data = &show_unhandled_signals,
1815 .maxlen = sizeof(int),
1816 .mode = 0644,
1817 .proc_handler = proc_dointvec
1818 },
1819#endif
1820#if defined(CONFIG_OPTPROBES)
1821 {
1822 .procname = "kprobes-optimization",
1823 .data = &sysctl_kprobes_optimization,
1824 .maxlen = sizeof(int),
1825 .mode = 0644,
1826 .proc_handler = proc_kprobes_optimization_handler,
1827 .extra1 = &zero,
1828 .extra2 = &one,
1829 },
1830#endif
1831 { }
1832};
1833
1834static struct ctl_table dev_table[] = {
1835 { }
1836};
1837
1838int __init sysctl_init(void)
1839{
1840 struct ctl_table_header *hdr;
1841
1842 hdr = register_sysctl_table(sysctl_base_table);
1843 kmemleak_not_leak(hdr);
1844 return 0;
1845}
1846
1847#endif /* CONFIG_SYSCTL */
1848
1849/*
1850 * /proc/sys support
1851 */
1852
1853#ifdef CONFIG_PROC_SYSCTL
1854
1855static int _proc_do_string(char *data, int maxlen, int write,
1856 char __user *buffer,
1857 size_t *lenp, loff_t *ppos)
1858{
1859 size_t len;
1860 char __user *p;
1861 char c;
1862
1863 if (!data || !maxlen || !*lenp) {
1864 *lenp = 0;
1865 return 0;
1866 }
1867
1868 if (write) {
1869 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1870 /* Only continue writes not past the end of buffer. */
1871 len = strlen(data);
1872 if (len > maxlen - 1)
1873 len = maxlen - 1;
1874
1875 if (*ppos > len)
1876 return 0;
1877 len = *ppos;
1878 } else {
1879 /* Start writing from beginning of buffer. */
1880 len = 0;
1881 }
1882
1883 *ppos += *lenp;
1884 p = buffer;
1885 while ((p - buffer) < *lenp && len < maxlen - 1) {
1886 if (get_user(c, p++))
1887 return -EFAULT;
1888 if (c == 0 || c == '\n')
1889 break;
1890 data[len++] = c;
1891 }
1892 data[len] = 0;
1893 } else {
1894 len = strlen(data);
1895 if (len > maxlen)
1896 len = maxlen;
1897
1898 if (*ppos > len) {
1899 *lenp = 0;
1900 return 0;
1901 }
1902
1903 data += *ppos;
1904 len -= *ppos;
1905
1906 if (len > *lenp)
1907 len = *lenp;
1908 if (len)
1909 if (copy_to_user(buffer, data, len))
1910 return -EFAULT;
1911 if (len < *lenp) {
1912 if (put_user('\n', buffer + len))
1913 return -EFAULT;
1914 len++;
1915 }
1916 *lenp = len;
1917 *ppos += len;
1918 }
1919 return 0;
1920}
1921
1922static void warn_sysctl_write(struct ctl_table *table)
1923{
1924 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1925 "This will not be supported in the future. To silence this\n"
1926 "warning, set kernel.sysctl_writes_strict = -1\n",
1927 current->comm, table->procname);
1928}
1929
1930/**
1931 * proc_dostring - read a string sysctl
1932 * @table: the sysctl table
1933 * @write: %TRUE if this is a write to the sysctl file
1934 * @buffer: the user buffer
1935 * @lenp: the size of the user buffer
1936 * @ppos: file position
1937 *
1938 * Reads/writes a string from/to the user buffer. If the kernel
1939 * buffer provided is not large enough to hold the string, the
1940 * string is truncated. The copied string is %NULL-terminated.
1941 * If the string is being read by the user process, it is copied
1942 * and a newline '\n' is added. It is truncated if the buffer is
1943 * not large enough.
1944 *
1945 * Returns 0 on success.
1946 */
1947int proc_dostring(struct ctl_table *table, int write,
1948 void __user *buffer, size_t *lenp, loff_t *ppos)
1949{
1950 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1951 warn_sysctl_write(table);
1952
1953 return _proc_do_string((char *)(table->data), table->maxlen, write,
1954 (char __user *)buffer, lenp, ppos);
1955}
1956
1957static size_t proc_skip_spaces(char **buf)
1958{
1959 size_t ret;
1960 char *tmp = skip_spaces(*buf);
1961 ret = tmp - *buf;
1962 *buf = tmp;
1963 return ret;
1964}
1965
1966static void proc_skip_char(char **buf, size_t *size, const char v)
1967{
1968 while (*size) {
1969 if (**buf != v)
1970 break;
1971 (*size)--;
1972 (*buf)++;
1973 }
1974}
1975
1976#define TMPBUFLEN 22
1977/**
1978 * proc_get_long - reads an ASCII formatted integer from a user buffer
1979 *
1980 * @buf: a kernel buffer
1981 * @size: size of the kernel buffer
1982 * @val: this is where the number will be stored
1983 * @neg: set to %TRUE if number is negative
1984 * @perm_tr: a vector which contains the allowed trailers
1985 * @perm_tr_len: size of the perm_tr vector
1986 * @tr: pointer to store the trailer character
1987 *
1988 * In case of success %0 is returned and @buf and @size are updated with
1989 * the amount of bytes read. If @tr is non-NULL and a trailing
1990 * character exists (size is non-zero after returning from this
1991 * function), @tr is updated with the trailing character.
1992 */
1993static int proc_get_long(char **buf, size_t *size,
1994 unsigned long *val, bool *neg,
1995 const char *perm_tr, unsigned perm_tr_len, char *tr)
1996{
1997 int len;
1998 char *p, tmp[TMPBUFLEN];
1999
2000 if (!*size)
2001 return -EINVAL;
2002
2003 len = *size;
2004 if (len > TMPBUFLEN - 1)
2005 len = TMPBUFLEN - 1;
2006
2007 memcpy(tmp, *buf, len);
2008
2009 tmp[len] = 0;
2010 p = tmp;
2011 if (*p == '-' && *size > 1) {
2012 *neg = true;
2013 p++;
2014 } else
2015 *neg = false;
2016 if (!isdigit(*p))
2017 return -EINVAL;
2018
2019 *val = simple_strtoul(p, &p, 0);
2020
2021 len = p - tmp;
2022
2023 /* We don't know if the next char is whitespace thus we may accept
2024 * invalid integers (e.g. 1234...a) or two integers instead of one
2025 * (e.g. 123...1). So lets not allow such large numbers. */
2026 if (len == TMPBUFLEN - 1)
2027 return -EINVAL;
2028
2029 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2030 return -EINVAL;
2031
2032 if (tr && (len < *size))
2033 *tr = *p;
2034
2035 *buf += len;
2036 *size -= len;
2037
2038 return 0;
2039}
2040
2041/**
2042 * proc_put_long - converts an integer to a decimal ASCII formatted string
2043 *
2044 * @buf: the user buffer
2045 * @size: the size of the user buffer
2046 * @val: the integer to be converted
2047 * @neg: sign of the number, %TRUE for negative
2048 *
2049 * In case of success %0 is returned and @buf and @size are updated with
2050 * the amount of bytes written.
2051 */
2052static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2053 bool neg)
2054{
2055 int len;
2056 char tmp[TMPBUFLEN], *p = tmp;
2057
2058 sprintf(p, "%s%lu", neg ? "-" : "", val);
2059 len = strlen(tmp);
2060 if (len > *size)
2061 len = *size;
2062 if (copy_to_user(*buf, tmp, len))
2063 return -EFAULT;
2064 *size -= len;
2065 *buf += len;
2066 return 0;
2067}
2068#undef TMPBUFLEN
2069
2070static int proc_put_char(void __user **buf, size_t *size, char c)
2071{
2072 if (*size) {
2073 char __user **buffer = (char __user **)buf;
2074 if (put_user(c, *buffer))
2075 return -EFAULT;
2076 (*size)--, (*buffer)++;
2077 *buf = *buffer;
2078 }
2079 return 0;
2080}
2081
2082static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2083 int *valp,
2084 int write, void *data)
2085{
2086 if (write) {
2087 if (*negp) {
2088 if (*lvalp > (unsigned long) INT_MAX + 1)
2089 return -EINVAL;
2090 *valp = -*lvalp;
2091 } else {
2092 if (*lvalp > (unsigned long) INT_MAX)
2093 return -EINVAL;
2094 *valp = *lvalp;
2095 }
2096 } else {
2097 int val = *valp;
2098 if (val < 0) {
2099 *negp = true;
2100 *lvalp = -(unsigned long)val;
2101 } else {
2102 *negp = false;
2103 *lvalp = (unsigned long)val;
2104 }
2105 }
2106 return 0;
2107}
2108
2109static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2110
2111static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2112 int write, void __user *buffer,
2113 size_t *lenp, loff_t *ppos,
2114 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2115 int write, void *data),
2116 void *data)
2117{
2118 int *i, vleft, first = 1, err = 0;
2119 size_t left;
2120 char *kbuf = NULL, *p;
2121
2122 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2123 *lenp = 0;
2124 return 0;
2125 }
2126
2127 i = (int *) tbl_data;
2128 vleft = table->maxlen / sizeof(*i);
2129 left = *lenp;
2130
2131 if (!conv)
2132 conv = do_proc_dointvec_conv;
2133
2134 if (write) {
2135 if (*ppos) {
2136 switch (sysctl_writes_strict) {
2137 case SYSCTL_WRITES_STRICT:
2138 goto out;
2139 case SYSCTL_WRITES_WARN:
2140 warn_sysctl_write(table);
2141 break;
2142 default:
2143 break;
2144 }
2145 }
2146
2147 if (left > PAGE_SIZE - 1)
2148 left = PAGE_SIZE - 1;
2149 p = kbuf = memdup_user_nul(buffer, left);
2150 if (IS_ERR(kbuf))
2151 return PTR_ERR(kbuf);
2152 }
2153
2154 for (; left && vleft--; i++, first=0) {
2155 unsigned long lval;
2156 bool neg;
2157
2158 if (write) {
2159 left -= proc_skip_spaces(&p);
2160
2161 if (!left)
2162 break;
2163 err = proc_get_long(&p, &left, &lval, &neg,
2164 proc_wspace_sep,
2165 sizeof(proc_wspace_sep), NULL);
2166 if (err)
2167 break;
2168 if (conv(&neg, &lval, i, 1, data)) {
2169 err = -EINVAL;
2170 break;
2171 }
2172 } else {
2173 if (conv(&neg, &lval, i, 0, data)) {
2174 err = -EINVAL;
2175 break;
2176 }
2177 if (!first)
2178 err = proc_put_char(&buffer, &left, '\t');
2179 if (err)
2180 break;
2181 err = proc_put_long(&buffer, &left, lval, neg);
2182 if (err)
2183 break;
2184 }
2185 }
2186
2187 if (!write && !first && left && !err)
2188 err = proc_put_char(&buffer, &left, '\n');
2189 if (write && !err && left)
2190 left -= proc_skip_spaces(&p);
2191 if (write) {
2192 kfree(kbuf);
2193 if (first)
2194 return err ? : -EINVAL;
2195 }
2196 *lenp -= left;
2197out:
2198 *ppos += *lenp;
2199 return err;
2200}
2201
2202static int do_proc_dointvec(struct ctl_table *table, int write,
2203 void __user *buffer, size_t *lenp, loff_t *ppos,
2204 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2205 int write, void *data),
2206 void *data)
2207{
2208 return __do_proc_dointvec(table->data, table, write,
2209 buffer, lenp, ppos, conv, data);
2210}
2211
2212/**
2213 * proc_dointvec - read a vector of integers
2214 * @table: the sysctl table
2215 * @write: %TRUE if this is a write to the sysctl file
2216 * @buffer: the user buffer
2217 * @lenp: the size of the user buffer
2218 * @ppos: file position
2219 *
2220 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2221 * values from/to the user buffer, treated as an ASCII string.
2222 *
2223 * Returns 0 on success.
2224 */
2225int proc_dointvec(struct ctl_table *table, int write,
2226 void __user *buffer, size_t *lenp, loff_t *ppos)
2227{
2228 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2229 NULL,NULL);
2230}
2231
2232/*
2233 * Taint values can only be increased
2234 * This means we can safely use a temporary.
2235 */
2236static int proc_taint(struct ctl_table *table, int write,
2237 void __user *buffer, size_t *lenp, loff_t *ppos)
2238{
2239 struct ctl_table t;
2240 unsigned long tmptaint = get_taint();
2241 int err;
2242
2243 if (write && !capable(CAP_SYS_ADMIN))
2244 return -EPERM;
2245
2246 t = *table;
2247 t.data = &tmptaint;
2248 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2249 if (err < 0)
2250 return err;
2251
2252 if (write) {
2253 /*
2254 * Poor man's atomic or. Not worth adding a primitive
2255 * to everyone's atomic.h for this
2256 */
2257 int i;
2258 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2259 if ((tmptaint >> i) & 1)
2260 add_taint(i, LOCKDEP_STILL_OK);
2261 }
2262 }
2263
2264 return err;
2265}
2266
2267#ifdef CONFIG_PRINTK
2268static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2269 void __user *buffer, size_t *lenp, loff_t *ppos)
2270{
2271 if (write && !capable(CAP_SYS_ADMIN))
2272 return -EPERM;
2273
2274 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2275}
2276#endif
2277
2278struct do_proc_dointvec_minmax_conv_param {
2279 int *min;
2280 int *max;
2281};
2282
2283static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2284 int *valp,
2285 int write, void *data)
2286{
2287 struct do_proc_dointvec_minmax_conv_param *param = data;
2288 if (write) {
2289 int val = *negp ? -*lvalp : *lvalp;
2290 if ((param->min && *param->min > val) ||
2291 (param->max && *param->max < val))
2292 return -EINVAL;
2293 *valp = val;
2294 } else {
2295 int val = *valp;
2296 if (val < 0) {
2297 *negp = true;
2298 *lvalp = -(unsigned long)val;
2299 } else {
2300 *negp = false;
2301 *lvalp = (unsigned long)val;
2302 }
2303 }
2304 return 0;
2305}
2306
2307/**
2308 * proc_dointvec_minmax - read a vector of integers with min/max values
2309 * @table: the sysctl table
2310 * @write: %TRUE if this is a write to the sysctl file
2311 * @buffer: the user buffer
2312 * @lenp: the size of the user buffer
2313 * @ppos: file position
2314 *
2315 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2316 * values from/to the user buffer, treated as an ASCII string.
2317 *
2318 * This routine will ensure the values are within the range specified by
2319 * table->extra1 (min) and table->extra2 (max).
2320 *
2321 * Returns 0 on success.
2322 */
2323int proc_dointvec_minmax(struct ctl_table *table, int write,
2324 void __user *buffer, size_t *lenp, loff_t *ppos)
2325{
2326 struct do_proc_dointvec_minmax_conv_param param = {
2327 .min = (int *) table->extra1,
2328 .max = (int *) table->extra2,
2329 };
2330 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2331 do_proc_dointvec_minmax_conv, &param);
2332}
2333
2334static void validate_coredump_safety(void)
2335{
2336#ifdef CONFIG_COREDUMP
2337 if (suid_dumpable == SUID_DUMP_ROOT &&
2338 core_pattern[0] != '/' && core_pattern[0] != '|') {
2339 printk(KERN_WARNING "Unsafe core_pattern used with "\
2340 "suid_dumpable=2. Pipe handler or fully qualified "\
2341 "core dump path required.\n");
2342 }
2343#endif
2344}
2345
2346static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2347 void __user *buffer, size_t *lenp, loff_t *ppos)
2348{
2349 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2350 if (!error)
2351 validate_coredump_safety();
2352 return error;
2353}
2354
2355#ifdef CONFIG_COREDUMP
2356static int proc_dostring_coredump(struct ctl_table *table, int write,
2357 void __user *buffer, size_t *lenp, loff_t *ppos)
2358{
2359 int error = proc_dostring(table, write, buffer, lenp, ppos);
2360 if (!error)
2361 validate_coredump_safety();
2362 return error;
2363}
2364#endif
2365
2366static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2367 void __user *buffer,
2368 size_t *lenp, loff_t *ppos,
2369 unsigned long convmul,
2370 unsigned long convdiv)
2371{
2372 unsigned long *i, *min, *max;
2373 int vleft, first = 1, err = 0;
2374 size_t left;
2375 char *kbuf = NULL, *p;
2376
2377 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2378 *lenp = 0;
2379 return 0;
2380 }
2381
2382 i = (unsigned long *) data;
2383 min = (unsigned long *) table->extra1;
2384 max = (unsigned long *) table->extra2;
2385 vleft = table->maxlen / sizeof(unsigned long);
2386 left = *lenp;
2387
2388 if (write) {
2389 if (*ppos) {
2390 switch (sysctl_writes_strict) {
2391 case SYSCTL_WRITES_STRICT:
2392 goto out;
2393 case SYSCTL_WRITES_WARN:
2394 warn_sysctl_write(table);
2395 break;
2396 default:
2397 break;
2398 }
2399 }
2400
2401 if (left > PAGE_SIZE - 1)
2402 left = PAGE_SIZE - 1;
2403 p = kbuf = memdup_user_nul(buffer, left);
2404 if (IS_ERR(kbuf))
2405 return PTR_ERR(kbuf);
2406 }
2407
2408 for (; left && vleft--; i++, first = 0) {
2409 unsigned long val;
2410
2411 if (write) {
2412 bool neg;
2413
2414 left -= proc_skip_spaces(&p);
2415
2416 err = proc_get_long(&p, &left, &val, &neg,
2417 proc_wspace_sep,
2418 sizeof(proc_wspace_sep), NULL);
2419 if (err)
2420 break;
2421 if (neg)
2422 continue;
2423 if ((min && val < *min) || (max && val > *max))
2424 continue;
2425 *i = val;
2426 } else {
2427 val = convdiv * (*i) / convmul;
2428 if (!first) {
2429 err = proc_put_char(&buffer, &left, '\t');
2430 if (err)
2431 break;
2432 }
2433 err = proc_put_long(&buffer, &left, val, false);
2434 if (err)
2435 break;
2436 }
2437 }
2438
2439 if (!write && !first && left && !err)
2440 err = proc_put_char(&buffer, &left, '\n');
2441 if (write && !err)
2442 left -= proc_skip_spaces(&p);
2443 if (write) {
2444 kfree(kbuf);
2445 if (first)
2446 return err ? : -EINVAL;
2447 }
2448 *lenp -= left;
2449out:
2450 *ppos += *lenp;
2451 return err;
2452}
2453
2454static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2455 void __user *buffer,
2456 size_t *lenp, loff_t *ppos,
2457 unsigned long convmul,
2458 unsigned long convdiv)
2459{
2460 return __do_proc_doulongvec_minmax(table->data, table, write,
2461 buffer, lenp, ppos, convmul, convdiv);
2462}
2463
2464/**
2465 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2466 * @table: the sysctl table
2467 * @write: %TRUE if this is a write to the sysctl file
2468 * @buffer: the user buffer
2469 * @lenp: the size of the user buffer
2470 * @ppos: file position
2471 *
2472 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2473 * values from/to the user buffer, treated as an ASCII string.
2474 *
2475 * This routine will ensure the values are within the range specified by
2476 * table->extra1 (min) and table->extra2 (max).
2477 *
2478 * Returns 0 on success.
2479 */
2480int proc_doulongvec_minmax(struct ctl_table *table, int write,
2481 void __user *buffer, size_t *lenp, loff_t *ppos)
2482{
2483 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2484}
2485
2486/**
2487 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2488 * @table: the sysctl table
2489 * @write: %TRUE if this is a write to the sysctl file
2490 * @buffer: the user buffer
2491 * @lenp: the size of the user buffer
2492 * @ppos: file position
2493 *
2494 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2495 * values from/to the user buffer, treated as an ASCII string. The values
2496 * are treated as milliseconds, and converted to jiffies when they are stored.
2497 *
2498 * This routine will ensure the values are within the range specified by
2499 * table->extra1 (min) and table->extra2 (max).
2500 *
2501 * Returns 0 on success.
2502 */
2503int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2504 void __user *buffer,
2505 size_t *lenp, loff_t *ppos)
2506{
2507 return do_proc_doulongvec_minmax(table, write, buffer,
2508 lenp, ppos, HZ, 1000l);
2509}
2510
2511
2512static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2513 int *valp,
2514 int write, void *data)
2515{
2516 if (write) {
2517 if (*lvalp > LONG_MAX / HZ)
2518 return 1;
2519 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2520 } else {
2521 int val = *valp;
2522 unsigned long lval;
2523 if (val < 0) {
2524 *negp = true;
2525 lval = -(unsigned long)val;
2526 } else {
2527 *negp = false;
2528 lval = (unsigned long)val;
2529 }
2530 *lvalp = lval / HZ;
2531 }
2532 return 0;
2533}
2534
2535static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2536 int *valp,
2537 int write, void *data)
2538{
2539 if (write) {
2540 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2541 return 1;
2542 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2543 } else {
2544 int val = *valp;
2545 unsigned long lval;
2546 if (val < 0) {
2547 *negp = true;
2548 lval = -(unsigned long)val;
2549 } else {
2550 *negp = false;
2551 lval = (unsigned long)val;
2552 }
2553 *lvalp = jiffies_to_clock_t(lval);
2554 }
2555 return 0;
2556}
2557
2558static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2559 int *valp,
2560 int write, void *data)
2561{
2562 if (write) {
2563 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2564
2565 if (jif > INT_MAX)
2566 return 1;
2567 *valp = (int)jif;
2568 } else {
2569 int val = *valp;
2570 unsigned long lval;
2571 if (val < 0) {
2572 *negp = true;
2573 lval = -(unsigned long)val;
2574 } else {
2575 *negp = false;
2576 lval = (unsigned long)val;
2577 }
2578 *lvalp = jiffies_to_msecs(lval);
2579 }
2580 return 0;
2581}
2582
2583/**
2584 * proc_dointvec_jiffies - read a vector of integers as seconds
2585 * @table: the sysctl table
2586 * @write: %TRUE if this is a write to the sysctl file
2587 * @buffer: the user buffer
2588 * @lenp: the size of the user buffer
2589 * @ppos: file position
2590 *
2591 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2592 * values from/to the user buffer, treated as an ASCII string.
2593 * The values read are assumed to be in seconds, and are converted into
2594 * jiffies.
2595 *
2596 * Returns 0 on success.
2597 */
2598int proc_dointvec_jiffies(struct ctl_table *table, int write,
2599 void __user *buffer, size_t *lenp, loff_t *ppos)
2600{
2601 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2602 do_proc_dointvec_jiffies_conv,NULL);
2603}
2604
2605/**
2606 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2607 * @table: the sysctl table
2608 * @write: %TRUE if this is a write to the sysctl file
2609 * @buffer: the user buffer
2610 * @lenp: the size of the user buffer
2611 * @ppos: pointer to the file position
2612 *
2613 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2614 * values from/to the user buffer, treated as an ASCII string.
2615 * The values read are assumed to be in 1/USER_HZ seconds, and
2616 * are converted into jiffies.
2617 *
2618 * Returns 0 on success.
2619 */
2620int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2621 void __user *buffer, size_t *lenp, loff_t *ppos)
2622{
2623 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2624 do_proc_dointvec_userhz_jiffies_conv,NULL);
2625}
2626
2627/**
2628 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2629 * @table: the sysctl table
2630 * @write: %TRUE if this is a write to the sysctl file
2631 * @buffer: the user buffer
2632 * @lenp: the size of the user buffer
2633 * @ppos: file position
2634 * @ppos: the current position in the file
2635 *
2636 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2637 * values from/to the user buffer, treated as an ASCII string.
2638 * The values read are assumed to be in 1/1000 seconds, and
2639 * are converted into jiffies.
2640 *
2641 * Returns 0 on success.
2642 */
2643int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2644 void __user *buffer, size_t *lenp, loff_t *ppos)
2645{
2646 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2647 do_proc_dointvec_ms_jiffies_conv, NULL);
2648}
2649
2650static int proc_do_cad_pid(struct ctl_table *table, int write,
2651 void __user *buffer, size_t *lenp, loff_t *ppos)
2652{
2653 struct pid *new_pid;
2654 pid_t tmp;
2655 int r;
2656
2657 tmp = pid_vnr(cad_pid);
2658
2659 r = __do_proc_dointvec(&tmp, table, write, buffer,
2660 lenp, ppos, NULL, NULL);
2661 if (r || !write)
2662 return r;
2663
2664 new_pid = find_get_pid(tmp);
2665 if (!new_pid)
2666 return -ESRCH;
2667
2668 put_pid(xchg(&cad_pid, new_pid));
2669 return 0;
2670}
2671
2672/**
2673 * proc_do_large_bitmap - read/write from/to a large bitmap
2674 * @table: the sysctl table
2675 * @write: %TRUE if this is a write to the sysctl file
2676 * @buffer: the user buffer
2677 * @lenp: the size of the user buffer
2678 * @ppos: file position
2679 *
2680 * The bitmap is stored at table->data and the bitmap length (in bits)
2681 * in table->maxlen.
2682 *
2683 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2684 * large bitmaps may be represented in a compact manner. Writing into
2685 * the file will clear the bitmap then update it with the given input.
2686 *
2687 * Returns 0 on success.
2688 */
2689int proc_do_large_bitmap(struct ctl_table *table, int write,
2690 void __user *buffer, size_t *lenp, loff_t *ppos)
2691{
2692 int err = 0;
2693 bool first = 1;
2694 size_t left = *lenp;
2695 unsigned long bitmap_len = table->maxlen;
2696 unsigned long *bitmap = *(unsigned long **) table->data;
2697 unsigned long *tmp_bitmap = NULL;
2698 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2699
2700 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2701 *lenp = 0;
2702 return 0;
2703 }
2704
2705 if (write) {
2706 char *kbuf, *p;
2707
2708 if (left > PAGE_SIZE - 1)
2709 left = PAGE_SIZE - 1;
2710
2711 p = kbuf = memdup_user_nul(buffer, left);
2712 if (IS_ERR(kbuf))
2713 return PTR_ERR(kbuf);
2714
2715 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2716 GFP_KERNEL);
2717 if (!tmp_bitmap) {
2718 kfree(kbuf);
2719 return -ENOMEM;
2720 }
2721 proc_skip_char(&p, &left, '\n');
2722 while (!err && left) {
2723 unsigned long val_a, val_b;
2724 bool neg;
2725
2726 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2727 sizeof(tr_a), &c);
2728 if (err)
2729 break;
2730 if (val_a >= bitmap_len || neg) {
2731 err = -EINVAL;
2732 break;
2733 }
2734
2735 val_b = val_a;
2736 if (left) {
2737 p++;
2738 left--;
2739 }
2740
2741 if (c == '-') {
2742 err = proc_get_long(&p, &left, &val_b,
2743 &neg, tr_b, sizeof(tr_b),
2744 &c);
2745 if (err)
2746 break;
2747 if (val_b >= bitmap_len || neg ||
2748 val_a > val_b) {
2749 err = -EINVAL;
2750 break;
2751 }
2752 if (left) {
2753 p++;
2754 left--;
2755 }
2756 }
2757
2758 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2759 first = 0;
2760 proc_skip_char(&p, &left, '\n');
2761 }
2762 kfree(kbuf);
2763 } else {
2764 unsigned long bit_a, bit_b = 0;
2765
2766 while (left) {
2767 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2768 if (bit_a >= bitmap_len)
2769 break;
2770 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2771 bit_a + 1) - 1;
2772
2773 if (!first) {
2774 err = proc_put_char(&buffer, &left, ',');
2775 if (err)
2776 break;
2777 }
2778 err = proc_put_long(&buffer, &left, bit_a, false);
2779 if (err)
2780 break;
2781 if (bit_a != bit_b) {
2782 err = proc_put_char(&buffer, &left, '-');
2783 if (err)
2784 break;
2785 err = proc_put_long(&buffer, &left, bit_b, false);
2786 if (err)
2787 break;
2788 }
2789
2790 first = 0; bit_b++;
2791 }
2792 if (!err)
2793 err = proc_put_char(&buffer, &left, '\n');
2794 }
2795
2796 if (!err) {
2797 if (write) {
2798 if (*ppos)
2799 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2800 else
2801 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2802 }
2803 kfree(tmp_bitmap);
2804 *lenp -= left;
2805 *ppos += *lenp;
2806 return 0;
2807 } else {
2808 kfree(tmp_bitmap);
2809 return err;
2810 }
2811}
2812
2813#else /* CONFIG_PROC_SYSCTL */
2814
2815int proc_dostring(struct ctl_table *table, int write,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2817{
2818 return -ENOSYS;
2819}
2820
2821int proc_dointvec(struct ctl_table *table, int write,
2822 void __user *buffer, size_t *lenp, loff_t *ppos)
2823{
2824 return -ENOSYS;
2825}
2826
2827int proc_dointvec_minmax(struct ctl_table *table, int write,
2828 void __user *buffer, size_t *lenp, loff_t *ppos)
2829{
2830 return -ENOSYS;
2831}
2832
2833int proc_dointvec_jiffies(struct ctl_table *table, int write,
2834 void __user *buffer, size_t *lenp, loff_t *ppos)
2835{
2836 return -ENOSYS;
2837}
2838
2839int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2840 void __user *buffer, size_t *lenp, loff_t *ppos)
2841{
2842 return -ENOSYS;
2843}
2844
2845int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2846 void __user *buffer, size_t *lenp, loff_t *ppos)
2847{
2848 return -ENOSYS;
2849}
2850
2851int proc_doulongvec_minmax(struct ctl_table *table, int write,
2852 void __user *buffer, size_t *lenp, loff_t *ppos)
2853{
2854 return -ENOSYS;
2855}
2856
2857int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2858 void __user *buffer,
2859 size_t *lenp, loff_t *ppos)
2860{
2861 return -ENOSYS;
2862}
2863
2864
2865#endif /* CONFIG_PROC_SYSCTL */
2866
2867/*
2868 * No sense putting this after each symbol definition, twice,
2869 * exception granted :-)
2870 */
2871EXPORT_SYMBOL(proc_dointvec);
2872EXPORT_SYMBOL(proc_dointvec_jiffies);
2873EXPORT_SYMBOL(proc_dointvec_minmax);
2874EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2875EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2876EXPORT_SYMBOL(proc_dostring);
2877EXPORT_SYMBOL(proc_doulongvec_minmax);
2878EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux 6.x block layer and io_uring tree(s)