projects / linux-2.6-block.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
Merge tag 'pm+acpi-4.6-rc1-1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
[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;
129#ifdef CONFIG_PRINTK
130static int ten_thousand = 10000;
131#endif
132
133/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
134static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
135
136/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
137static int maxolduid = 65535;
138static int minolduid;
139
140static int ngroups_max = NGROUPS_MAX;
141static const int cap_last_cap = CAP_LAST_CAP;
142
143/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
144#ifdef CONFIG_DETECT_HUNG_TASK
145static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146#endif
147
148#ifdef CONFIG_INOTIFY_USER
149#include <linux/inotify.h>
150#endif
151#ifdef CONFIG_SPARC
152#endif
153
154#ifdef __hppa__
155extern int pwrsw_enabled;
156#endif
157
158#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
159extern int unaligned_enabled;
160#endif
161
162#ifdef CONFIG_IA64
163extern int unaligned_dump_stack;
164#endif
165
166#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
167extern int no_unaligned_warning;
168#endif
169
170#ifdef CONFIG_PROC_SYSCTL
171
172#define SYSCTL_WRITES_LEGACY -1
173#define SYSCTL_WRITES_WARN 0
174#define SYSCTL_WRITES_STRICT 1
175
176static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
177
178static int proc_do_cad_pid(struct ctl_table *table, int write,
179 void __user *buffer, size_t *lenp, loff_t *ppos);
180static int proc_taint(struct ctl_table *table, int write,
181 void __user *buffer, size_t *lenp, loff_t *ppos);
182#endif
183
184#ifdef CONFIG_PRINTK
185static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
186 void __user *buffer, size_t *lenp, loff_t *ppos);
187#endif
188
189static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
190 void __user *buffer, size_t *lenp, loff_t *ppos);
191#ifdef CONFIG_COREDUMP
192static int proc_dostring_coredump(struct ctl_table *table, int write,
193 void __user *buffer, size_t *lenp, loff_t *ppos);
194#endif
195
196#ifdef CONFIG_MAGIC_SYSRQ
197/* Note: sysrq code uses it's own private copy */
198static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
199
200static int sysrq_sysctl_handler(struct ctl_table *table, int write,
201 void __user *buffer, size_t *lenp,
202 loff_t *ppos)
203{
204 int error;
205
206 error = proc_dointvec(table, write, buffer, lenp, ppos);
207 if (error)
208 return error;
209
210 if (write)
211 sysrq_toggle_support(__sysrq_enabled);
212
213 return 0;
214}
215
216#endif
217
218static struct ctl_table kern_table[];
219static struct ctl_table vm_table[];
220static struct ctl_table fs_table[];
221static struct ctl_table debug_table[];
222static struct ctl_table dev_table[];
223extern struct ctl_table random_table[];
224#ifdef CONFIG_EPOLL
225extern struct ctl_table epoll_table[];
226#endif
227
228#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
229int sysctl_legacy_va_layout;
230#endif
231
232/* The default sysctl tables: */
233
234static struct ctl_table sysctl_base_table[] = {
235 {
236 .procname = "kernel",
237 .mode = 0555,
238 .child = kern_table,
239 },
240 {
241 .procname = "vm",
242 .mode = 0555,
243 .child = vm_table,
244 },
245 {
246 .procname = "fs",
247 .mode = 0555,
248 .child = fs_table,
249 },
250 {
251 .procname = "debug",
252 .mode = 0555,
253 .child = debug_table,
254 },
255 {
256 .procname = "dev",
257 .mode = 0555,
258 .child = dev_table,
259 },
260 { }
261};
262
263#ifdef CONFIG_SCHED_DEBUG
264static int min_sched_granularity_ns = 100000; /* 100 usecs */
265static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
266static int min_wakeup_granularity_ns; /* 0 usecs */
267static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
268#ifdef CONFIG_SMP
269static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
270static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
271#endif /* CONFIG_SMP */
272#endif /* CONFIG_SCHED_DEBUG */
273
274#ifdef CONFIG_COMPACTION
275static int min_extfrag_threshold;
276static int max_extfrag_threshold = 1000;
277#endif
278
279static struct ctl_table kern_table[] = {
280 {
281 .procname = "sched_child_runs_first",
282 .data = &sysctl_sched_child_runs_first,
283 .maxlen = sizeof(unsigned int),
284 .mode = 0644,
285 .proc_handler = proc_dointvec,
286 },
287#ifdef CONFIG_SCHED_DEBUG
288 {
289 .procname = "sched_min_granularity_ns",
290 .data = &sysctl_sched_min_granularity,
291 .maxlen = sizeof(unsigned int),
292 .mode = 0644,
293 .proc_handler = sched_proc_update_handler,
294 .extra1 = &min_sched_granularity_ns,
295 .extra2 = &max_sched_granularity_ns,
296 },
297 {
298 .procname = "sched_latency_ns",
299 .data = &sysctl_sched_latency,
300 .maxlen = sizeof(unsigned int),
301 .mode = 0644,
302 .proc_handler = sched_proc_update_handler,
303 .extra1 = &min_sched_granularity_ns,
304 .extra2 = &max_sched_granularity_ns,
305 },
306 {
307 .procname = "sched_wakeup_granularity_ns",
308 .data = &sysctl_sched_wakeup_granularity,
309 .maxlen = sizeof(unsigned int),
310 .mode = 0644,
311 .proc_handler = sched_proc_update_handler,
312 .extra1 = &min_wakeup_granularity_ns,
313 .extra2 = &max_wakeup_granularity_ns,
314 },
315#ifdef CONFIG_SMP
316 {
317 .procname = "sched_tunable_scaling",
318 .data = &sysctl_sched_tunable_scaling,
319 .maxlen = sizeof(enum sched_tunable_scaling),
320 .mode = 0644,
321 .proc_handler = sched_proc_update_handler,
322 .extra1 = &min_sched_tunable_scaling,
323 .extra2 = &max_sched_tunable_scaling,
324 },
325 {
326 .procname = "sched_migration_cost_ns",
327 .data = &sysctl_sched_migration_cost,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0644,
330 .proc_handler = proc_dointvec,
331 },
332 {
333 .procname = "sched_nr_migrate",
334 .data = &sysctl_sched_nr_migrate,
335 .maxlen = sizeof(unsigned int),
336 .mode = 0644,
337 .proc_handler = proc_dointvec,
338 },
339 {
340 .procname = "sched_time_avg_ms",
341 .data = &sysctl_sched_time_avg,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = proc_dointvec,
345 },
346 {
347 .procname = "sched_shares_window_ns",
348 .data = &sysctl_sched_shares_window,
349 .maxlen = sizeof(unsigned int),
350 .mode = 0644,
351 .proc_handler = proc_dointvec,
352 },
353#ifdef CONFIG_SCHEDSTATS
354 {
355 .procname = "sched_schedstats",
356 .data = NULL,
357 .maxlen = sizeof(unsigned int),
358 .mode = 0644,
359 .proc_handler = sysctl_schedstats,
360 .extra1 = &zero,
361 .extra2 = &one,
362 },
363#endif /* CONFIG_SCHEDSTATS */
364#endif /* CONFIG_SMP */
365#ifdef CONFIG_NUMA_BALANCING
366 {
367 .procname = "numa_balancing_scan_delay_ms",
368 .data = &sysctl_numa_balancing_scan_delay,
369 .maxlen = sizeof(unsigned int),
370 .mode = 0644,
371 .proc_handler = proc_dointvec,
372 },
373 {
374 .procname = "numa_balancing_scan_period_min_ms",
375 .data = &sysctl_numa_balancing_scan_period_min,
376 .maxlen = sizeof(unsigned int),
377 .mode = 0644,
378 .proc_handler = proc_dointvec,
379 },
380 {
381 .procname = "numa_balancing_scan_period_max_ms",
382 .data = &sysctl_numa_balancing_scan_period_max,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = proc_dointvec,
386 },
387 {
388 .procname = "numa_balancing_scan_size_mb",
389 .data = &sysctl_numa_balancing_scan_size,
390 .maxlen = sizeof(unsigned int),
391 .mode = 0644,
392 .proc_handler = proc_dointvec_minmax,
393 .extra1 = &one,
394 },
395 {
396 .procname = "numa_balancing",
397 .data = NULL, /* filled in by handler */
398 .maxlen = sizeof(unsigned int),
399 .mode = 0644,
400 .proc_handler = sysctl_numa_balancing,
401 .extra1 = &zero,
402 .extra2 = &one,
403 },
404#endif /* CONFIG_NUMA_BALANCING */
405#endif /* CONFIG_SCHED_DEBUG */
406 {
407 .procname = "sched_rt_period_us",
408 .data = &sysctl_sched_rt_period,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = sched_rt_handler,
412 },
413 {
414 .procname = "sched_rt_runtime_us",
415 .data = &sysctl_sched_rt_runtime,
416 .maxlen = sizeof(int),
417 .mode = 0644,
418 .proc_handler = sched_rt_handler,
419 },
420 {
421 .procname = "sched_rr_timeslice_ms",
422 .data = &sched_rr_timeslice,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = sched_rr_handler,
426 },
427#ifdef CONFIG_SCHED_AUTOGROUP
428 {
429 .procname = "sched_autogroup_enabled",
430 .data = &sysctl_sched_autogroup_enabled,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec_minmax,
434 .extra1 = &zero,
435 .extra2 = &one,
436 },
437#endif
438#ifdef CONFIG_CFS_BANDWIDTH
439 {
440 .procname = "sched_cfs_bandwidth_slice_us",
441 .data = &sysctl_sched_cfs_bandwidth_slice,
442 .maxlen = sizeof(unsigned int),
443 .mode = 0644,
444 .proc_handler = proc_dointvec_minmax,
445 .extra1 = &one,
446 },
447#endif
448#ifdef CONFIG_PROVE_LOCKING
449 {
450 .procname = "prove_locking",
451 .data = &prove_locking,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec,
455 },
456#endif
457#ifdef CONFIG_LOCK_STAT
458 {
459 .procname = "lock_stat",
460 .data = &lock_stat,
461 .maxlen = sizeof(int),
462 .mode = 0644,
463 .proc_handler = proc_dointvec,
464 },
465#endif
466 {
467 .procname = "panic",
468 .data = &panic_timeout,
469 .maxlen = sizeof(int),
470 .mode = 0644,
471 .proc_handler = proc_dointvec,
472 },
473#ifdef CONFIG_COREDUMP
474 {
475 .procname = "core_uses_pid",
476 .data = &core_uses_pid,
477 .maxlen = sizeof(int),
478 .mode = 0644,
479 .proc_handler = proc_dointvec,
480 },
481 {
482 .procname = "core_pattern",
483 .data = core_pattern,
484 .maxlen = CORENAME_MAX_SIZE,
485 .mode = 0644,
486 .proc_handler = proc_dostring_coredump,
487 },
488 {
489 .procname = "core_pipe_limit",
490 .data = &core_pipe_limit,
491 .maxlen = sizeof(unsigned int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495#endif
496#ifdef CONFIG_PROC_SYSCTL
497 {
498 .procname = "tainted",
499 .maxlen = sizeof(long),
500 .mode = 0644,
501 .proc_handler = proc_taint,
502 },
503 {
504 .procname = "sysctl_writes_strict",
505 .data = &sysctl_writes_strict,
506 .maxlen = sizeof(int),
507 .mode = 0644,
508 .proc_handler = proc_dointvec_minmax,
509 .extra1 = &neg_one,
510 .extra2 = &one,
511 },
512#endif
513#ifdef CONFIG_LATENCYTOP
514 {
515 .procname = "latencytop",
516 .data = &latencytop_enabled,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = sysctl_latencytop,
520 },
521#endif
522#ifdef CONFIG_BLK_DEV_INITRD
523 {
524 .procname = "real-root-dev",
525 .data = &real_root_dev,
526 .maxlen = sizeof(int),
527 .mode = 0644,
528 .proc_handler = proc_dointvec,
529 },
530#endif
531 {
532 .procname = "print-fatal-signals",
533 .data = &print_fatal_signals,
534 .maxlen = sizeof(int),
535 .mode = 0644,
536 .proc_handler = proc_dointvec,
537 },
538#ifdef CONFIG_SPARC
539 {
540 .procname = "reboot-cmd",
541 .data = reboot_command,
542 .maxlen = 256,
543 .mode = 0644,
544 .proc_handler = proc_dostring,
545 },
546 {
547 .procname = "stop-a",
548 .data = &stop_a_enabled,
549 .maxlen = sizeof (int),
550 .mode = 0644,
551 .proc_handler = proc_dointvec,
552 },
553 {
554 .procname = "scons-poweroff",
555 .data = &scons_pwroff,
556 .maxlen = sizeof (int),
557 .mode = 0644,
558 .proc_handler = proc_dointvec,
559 },
560#endif
561#ifdef CONFIG_SPARC64
562 {
563 .procname = "tsb-ratio",
564 .data = &sysctl_tsb_ratio,
565 .maxlen = sizeof (int),
566 .mode = 0644,
567 .proc_handler = proc_dointvec,
568 },
569#endif
570#ifdef __hppa__
571 {
572 .procname = "soft-power",
573 .data = &pwrsw_enabled,
574 .maxlen = sizeof (int),
575 .mode = 0644,
576 .proc_handler = proc_dointvec,
577 },
578#endif
579#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
580 {
581 .procname = "unaligned-trap",
582 .data = &unaligned_enabled,
583 .maxlen = sizeof (int),
584 .mode = 0644,
585 .proc_handler = proc_dointvec,
586 },
587#endif
588 {
589 .procname = "ctrl-alt-del",
590 .data = &C_A_D,
591 .maxlen = sizeof(int),
592 .mode = 0644,
593 .proc_handler = proc_dointvec,
594 },
595#ifdef CONFIG_FUNCTION_TRACER
596 {
597 .procname = "ftrace_enabled",
598 .data = &ftrace_enabled,
599 .maxlen = sizeof(int),
600 .mode = 0644,
601 .proc_handler = ftrace_enable_sysctl,
602 },
603#endif
604#ifdef CONFIG_STACK_TRACER
605 {
606 .procname = "stack_tracer_enabled",
607 .data = &stack_tracer_enabled,
608 .maxlen = sizeof(int),
609 .mode = 0644,
610 .proc_handler = stack_trace_sysctl,
611 },
612#endif
613#ifdef CONFIG_TRACING
614 {
615 .procname = "ftrace_dump_on_oops",
616 .data = &ftrace_dump_on_oops,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621 {
622 .procname = "traceoff_on_warning",
623 .data = &__disable_trace_on_warning,
624 .maxlen = sizeof(__disable_trace_on_warning),
625 .mode = 0644,
626 .proc_handler = proc_dointvec,
627 },
628 {
629 .procname = "tracepoint_printk",
630 .data = &tracepoint_printk,
631 .maxlen = sizeof(tracepoint_printk),
632 .mode = 0644,
633 .proc_handler = proc_dointvec,
634 },
635#endif
636#ifdef CONFIG_KEXEC_CORE
637 {
638 .procname = "kexec_load_disabled",
639 .data = &kexec_load_disabled,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 /* only handle a transition from default "0" to "1" */
643 .proc_handler = proc_dointvec_minmax,
644 .extra1 = &one,
645 .extra2 = &one,
646 },
647#endif
648#ifdef CONFIG_MODULES
649 {
650 .procname = "modprobe",
651 .data = &modprobe_path,
652 .maxlen = KMOD_PATH_LEN,
653 .mode = 0644,
654 .proc_handler = proc_dostring,
655 },
656 {
657 .procname = "modules_disabled",
658 .data = &modules_disabled,
659 .maxlen = sizeof(int),
660 .mode = 0644,
661 /* only handle a transition from default "0" to "1" */
662 .proc_handler = proc_dointvec_minmax,
663 .extra1 = &one,
664 .extra2 = &one,
665 },
666#endif
667#ifdef CONFIG_UEVENT_HELPER
668 {
669 .procname = "hotplug",
670 .data = &uevent_helper,
671 .maxlen = UEVENT_HELPER_PATH_LEN,
672 .mode = 0644,
673 .proc_handler = proc_dostring,
674 },
675#endif
676#ifdef CONFIG_CHR_DEV_SG
677 {
678 .procname = "sg-big-buff",
679 .data = &sg_big_buff,
680 .maxlen = sizeof (int),
681 .mode = 0444,
682 .proc_handler = proc_dointvec,
683 },
684#endif
685#ifdef CONFIG_BSD_PROCESS_ACCT
686 {
687 .procname = "acct",
688 .data = &acct_parm,
689 .maxlen = 3*sizeof(int),
690 .mode = 0644,
691 .proc_handler = proc_dointvec,
692 },
693#endif
694#ifdef CONFIG_MAGIC_SYSRQ
695 {
696 .procname = "sysrq",
697 .data = &__sysrq_enabled,
698 .maxlen = sizeof (int),
699 .mode = 0644,
700 .proc_handler = sysrq_sysctl_handler,
701 },
702#endif
703#ifdef CONFIG_PROC_SYSCTL
704 {
705 .procname = "cad_pid",
706 .data = NULL,
707 .maxlen = sizeof (int),
708 .mode = 0600,
709 .proc_handler = proc_do_cad_pid,
710 },
711#endif
712 {
713 .procname = "threads-max",
714 .data = NULL,
715 .maxlen = sizeof(int),
716 .mode = 0644,
717 .proc_handler = sysctl_max_threads,
718 },
719 {
720 .procname = "random",
721 .mode = 0555,
722 .child = random_table,
723 },
724 {
725 .procname = "usermodehelper",
726 .mode = 0555,
727 .child = usermodehelper_table,
728 },
729 {
730 .procname = "overflowuid",
731 .data = &overflowuid,
732 .maxlen = sizeof(int),
733 .mode = 0644,
734 .proc_handler = proc_dointvec_minmax,
735 .extra1 = &minolduid,
736 .extra2 = &maxolduid,
737 },
738 {
739 .procname = "overflowgid",
740 .data = &overflowgid,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = proc_dointvec_minmax,
744 .extra1 = &minolduid,
745 .extra2 = &maxolduid,
746 },
747#ifdef CONFIG_S390
748#ifdef CONFIG_MATHEMU
749 {
750 .procname = "ieee_emulation_warnings",
751 .data = &sysctl_ieee_emulation_warnings,
752 .maxlen = sizeof(int),
753 .mode = 0644,
754 .proc_handler = proc_dointvec,
755 },
756#endif
757 {
758 .procname = "userprocess_debug",
759 .data = &show_unhandled_signals,
760 .maxlen = sizeof(int),
761 .mode = 0644,
762 .proc_handler = proc_dointvec,
763 },
764#endif
765 {
766 .procname = "pid_max",
767 .data = &pid_max,
768 .maxlen = sizeof (int),
769 .mode = 0644,
770 .proc_handler = proc_dointvec_minmax,
771 .extra1 = &pid_max_min,
772 .extra2 = &pid_max_max,
773 },
774 {
775 .procname = "panic_on_oops",
776 .data = &panic_on_oops,
777 .maxlen = sizeof(int),
778 .mode = 0644,
779 .proc_handler = proc_dointvec,
780 },
781#if defined CONFIG_PRINTK
782 {
783 .procname = "printk",
784 .data = &console_loglevel,
785 .maxlen = 4*sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789 {
790 .procname = "printk_ratelimit",
791 .data = &printk_ratelimit_state.interval,
792 .maxlen = sizeof(int),
793 .mode = 0644,
794 .proc_handler = proc_dointvec_jiffies,
795 },
796 {
797 .procname = "printk_ratelimit_burst",
798 .data = &printk_ratelimit_state.burst,
799 .maxlen = sizeof(int),
800 .mode = 0644,
801 .proc_handler = proc_dointvec,
802 },
803 {
804 .procname = "printk_delay",
805 .data = &printk_delay_msec,
806 .maxlen = sizeof(int),
807 .mode = 0644,
808 .proc_handler = proc_dointvec_minmax,
809 .extra1 = &zero,
810 .extra2 = &ten_thousand,
811 },
812 {
813 .procname = "dmesg_restrict",
814 .data = &dmesg_restrict,
815 .maxlen = sizeof(int),
816 .mode = 0644,
817 .proc_handler = proc_dointvec_minmax_sysadmin,
818 .extra1 = &zero,
819 .extra2 = &one,
820 },
821 {
822 .procname = "kptr_restrict",
823 .data = &kptr_restrict,
824 .maxlen = sizeof(int),
825 .mode = 0644,
826 .proc_handler = proc_dointvec_minmax_sysadmin,
827 .extra1 = &zero,
828 .extra2 = &two,
829 },
830#endif
831 {
832 .procname = "ngroups_max",
833 .data = &ngroups_max,
834 .maxlen = sizeof (int),
835 .mode = 0444,
836 .proc_handler = proc_dointvec,
837 },
838 {
839 .procname = "cap_last_cap",
840 .data = (void *)&cap_last_cap,
841 .maxlen = sizeof(int),
842 .mode = 0444,
843 .proc_handler = proc_dointvec,
844 },
845#if defined(CONFIG_LOCKUP_DETECTOR)
846 {
847 .procname = "watchdog",
848 .data = &watchdog_user_enabled,
849 .maxlen = sizeof (int),
850 .mode = 0644,
851 .proc_handler = proc_watchdog,
852 .extra1 = &zero,
853 .extra2 = &one,
854 },
855 {
856 .procname = "watchdog_thresh",
857 .data = &watchdog_thresh,
858 .maxlen = sizeof(int),
859 .mode = 0644,
860 .proc_handler = proc_watchdog_thresh,
861 .extra1 = &zero,
862 .extra2 = &sixty,
863 },
864 {
865 .procname = "nmi_watchdog",
866 .data = &nmi_watchdog_enabled,
867 .maxlen = sizeof (int),
868 .mode = 0644,
869 .proc_handler = proc_nmi_watchdog,
870 .extra1 = &zero,
871#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
872 .extra2 = &one,
873#else
874 .extra2 = &zero,
875#endif
876 },
877 {
878 .procname = "soft_watchdog",
879 .data = &soft_watchdog_enabled,
880 .maxlen = sizeof (int),
881 .mode = 0644,
882 .proc_handler = proc_soft_watchdog,
883 .extra1 = &zero,
884 .extra2 = &one,
885 },
886 {
887 .procname = "watchdog_cpumask",
888 .data = &watchdog_cpumask_bits,
889 .maxlen = NR_CPUS,
890 .mode = 0644,
891 .proc_handler = proc_watchdog_cpumask,
892 },
893 {
894 .procname = "softlockup_panic",
895 .data = &softlockup_panic,
896 .maxlen = sizeof(int),
897 .mode = 0644,
898 .proc_handler = proc_dointvec_minmax,
899 .extra1 = &zero,
900 .extra2 = &one,
901 },
902#ifdef CONFIG_HARDLOCKUP_DETECTOR
903 {
904 .procname = "hardlockup_panic",
905 .data = &hardlockup_panic,
906 .maxlen = sizeof(int),
907 .mode = 0644,
908 .proc_handler = proc_dointvec_minmax,
909 .extra1 = &zero,
910 .extra2 = &one,
911 },
912#endif
913#ifdef CONFIG_SMP
914 {
915 .procname = "softlockup_all_cpu_backtrace",
916 .data = &sysctl_softlockup_all_cpu_backtrace,
917 .maxlen = sizeof(int),
918 .mode = 0644,
919 .proc_handler = proc_dointvec_minmax,
920 .extra1 = &zero,
921 .extra2 = &one,
922 },
923 {
924 .procname = "hardlockup_all_cpu_backtrace",
925 .data = &sysctl_hardlockup_all_cpu_backtrace,
926 .maxlen = sizeof(int),
927 .mode = 0644,
928 .proc_handler = proc_dointvec_minmax,
929 .extra1 = &zero,
930 .extra2 = &one,
931 },
932#endif /* CONFIG_SMP */
933#endif
934#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
935 {
936 .procname = "unknown_nmi_panic",
937 .data = &unknown_nmi_panic,
938 .maxlen = sizeof (int),
939 .mode = 0644,
940 .proc_handler = proc_dointvec,
941 },
942#endif
943#if defined(CONFIG_X86)
944 {
945 .procname = "panic_on_unrecovered_nmi",
946 .data = &panic_on_unrecovered_nmi,
947 .maxlen = sizeof(int),
948 .mode = 0644,
949 .proc_handler = proc_dointvec,
950 },
951 {
952 .procname = "panic_on_io_nmi",
953 .data = &panic_on_io_nmi,
954 .maxlen = sizeof(int),
955 .mode = 0644,
956 .proc_handler = proc_dointvec,
957 },
958#ifdef CONFIG_DEBUG_STACKOVERFLOW
959 {
960 .procname = "panic_on_stackoverflow",
961 .data = &sysctl_panic_on_stackoverflow,
962 .maxlen = sizeof(int),
963 .mode = 0644,
964 .proc_handler = proc_dointvec,
965 },
966#endif
967 {
968 .procname = "bootloader_type",
969 .data = &bootloader_type,
970 .maxlen = sizeof (int),
971 .mode = 0444,
972 .proc_handler = proc_dointvec,
973 },
974 {
975 .procname = "bootloader_version",
976 .data = &bootloader_version,
977 .maxlen = sizeof (int),
978 .mode = 0444,
979 .proc_handler = proc_dointvec,
980 },
981 {
982 .procname = "kstack_depth_to_print",
983 .data = &kstack_depth_to_print,
984 .maxlen = sizeof(int),
985 .mode = 0644,
986 .proc_handler = proc_dointvec,
987 },
988 {
989 .procname = "io_delay_type",
990 .data = &io_delay_type,
991 .maxlen = sizeof(int),
992 .mode = 0644,
993 .proc_handler = proc_dointvec,
994 },
995#endif
996#if defined(CONFIG_MMU)
997 {
998 .procname = "randomize_va_space",
999 .data = &randomize_va_space,
1000 .maxlen = sizeof(int),
1001 .mode = 0644,
1002 .proc_handler = proc_dointvec,
1003 },
1004#endif
1005#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1006 {
1007 .procname = "spin_retry",
1008 .data = &spin_retry,
1009 .maxlen = sizeof (int),
1010 .mode = 0644,
1011 .proc_handler = proc_dointvec,
1012 },
1013#endif
1014#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1015 {
1016 .procname = "acpi_video_flags",
1017 .data = &acpi_realmode_flags,
1018 .maxlen = sizeof (unsigned long),
1019 .mode = 0644,
1020 .proc_handler = proc_doulongvec_minmax,
1021 },
1022#endif
1023#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1024 {
1025 .procname = "ignore-unaligned-usertrap",
1026 .data = &no_unaligned_warning,
1027 .maxlen = sizeof (int),
1028 .mode = 0644,
1029 .proc_handler = proc_dointvec,
1030 },
1031#endif
1032#ifdef CONFIG_IA64
1033 {
1034 .procname = "unaligned-dump-stack",
1035 .data = &unaligned_dump_stack,
1036 .maxlen = sizeof (int),
1037 .mode = 0644,
1038 .proc_handler = proc_dointvec,
1039 },
1040#endif
1041#ifdef CONFIG_DETECT_HUNG_TASK
1042 {
1043 .procname = "hung_task_panic",
1044 .data = &sysctl_hung_task_panic,
1045 .maxlen = sizeof(int),
1046 .mode = 0644,
1047 .proc_handler = proc_dointvec_minmax,
1048 .extra1 = &zero,
1049 .extra2 = &one,
1050 },
1051 {
1052 .procname = "hung_task_check_count",
1053 .data = &sysctl_hung_task_check_count,
1054 .maxlen = sizeof(int),
1055 .mode = 0644,
1056 .proc_handler = proc_dointvec_minmax,
1057 .extra1 = &zero,
1058 },
1059 {
1060 .procname = "hung_task_timeout_secs",
1061 .data = &sysctl_hung_task_timeout_secs,
1062 .maxlen = sizeof(unsigned long),
1063 .mode = 0644,
1064 .proc_handler = proc_dohung_task_timeout_secs,
1065 .extra2 = &hung_task_timeout_max,
1066 },
1067 {
1068 .procname = "hung_task_warnings",
1069 .data = &sysctl_hung_task_warnings,
1070 .maxlen = sizeof(int),
1071 .mode = 0644,
1072 .proc_handler = proc_dointvec_minmax,
1073 .extra1 = &neg_one,
1074 },
1075#endif
1076#ifdef CONFIG_COMPAT
1077 {
1078 .procname = "compat-log",
1079 .data = &compat_log,
1080 .maxlen = sizeof (int),
1081 .mode = 0644,
1082 .proc_handler = proc_dointvec,
1083 },
1084#endif
1085#ifdef CONFIG_RT_MUTEXES
1086 {
1087 .procname = "max_lock_depth",
1088 .data = &max_lock_depth,
1089 .maxlen = sizeof(int),
1090 .mode = 0644,
1091 .proc_handler = proc_dointvec,
1092 },
1093#endif
1094 {
1095 .procname = "poweroff_cmd",
1096 .data = &poweroff_cmd,
1097 .maxlen = POWEROFF_CMD_PATH_LEN,
1098 .mode = 0644,
1099 .proc_handler = proc_dostring,
1100 },
1101#ifdef CONFIG_KEYS
1102 {
1103 .procname = "keys",
1104 .mode = 0555,
1105 .child = key_sysctls,
1106 },
1107#endif
1108#ifdef CONFIG_PERF_EVENTS
1109 /*
1110 * User-space scripts rely on the existence of this file
1111 * as a feature check for perf_events being enabled.
1112 *
1113 * So it's an ABI, do not remove!
1114 */
1115 {
1116 .procname = "perf_event_paranoid",
1117 .data = &sysctl_perf_event_paranoid,
1118 .maxlen = sizeof(sysctl_perf_event_paranoid),
1119 .mode = 0644,
1120 .proc_handler = proc_dointvec,
1121 },
1122 {
1123 .procname = "perf_event_mlock_kb",
1124 .data = &sysctl_perf_event_mlock,
1125 .maxlen = sizeof(sysctl_perf_event_mlock),
1126 .mode = 0644,
1127 .proc_handler = proc_dointvec,
1128 },
1129 {
1130 .procname = "perf_event_max_sample_rate",
1131 .data = &sysctl_perf_event_sample_rate,
1132 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1133 .mode = 0644,
1134 .proc_handler = perf_proc_update_handler,
1135 .extra1 = &one,
1136 },
1137 {
1138 .procname = "perf_cpu_time_max_percent",
1139 .data = &sysctl_perf_cpu_time_max_percent,
1140 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1141 .mode = 0644,
1142 .proc_handler = perf_cpu_time_max_percent_handler,
1143 .extra1 = &zero,
1144 .extra2 = &one_hundred,
1145 },
1146#endif
1147#ifdef CONFIG_KMEMCHECK
1148 {
1149 .procname = "kmemcheck",
1150 .data = &kmemcheck_enabled,
1151 .maxlen = sizeof(int),
1152 .mode = 0644,
1153 .proc_handler = proc_dointvec,
1154 },
1155#endif
1156 {
1157 .procname = "panic_on_warn",
1158 .data = &panic_on_warn,
1159 .maxlen = sizeof(int),
1160 .mode = 0644,
1161 .proc_handler = proc_dointvec_minmax,
1162 .extra1 = &zero,
1163 .extra2 = &one,
1164 },
1165#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1166 {
1167 .procname = "timer_migration",
1168 .data = &sysctl_timer_migration,
1169 .maxlen = sizeof(unsigned int),
1170 .mode = 0644,
1171 .proc_handler = timer_migration_handler,
1172 },
1173#endif
1174#ifdef CONFIG_BPF_SYSCALL
1175 {
1176 .procname = "unprivileged_bpf_disabled",
1177 .data = &sysctl_unprivileged_bpf_disabled,
1178 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1179 .mode = 0644,
1180 /* only handle a transition from default "0" to "1" */
1181 .proc_handler = proc_dointvec_minmax,
1182 .extra1 = &one,
1183 .extra2 = &one,
1184 },
1185#endif
1186 { }
1187};
1188
1189static struct ctl_table vm_table[] = {
1190 {
1191 .procname = "overcommit_memory",
1192 .data = &sysctl_overcommit_memory,
1193 .maxlen = sizeof(sysctl_overcommit_memory),
1194 .mode = 0644,
1195 .proc_handler = proc_dointvec_minmax,
1196 .extra1 = &zero,
1197 .extra2 = &two,
1198 },
1199 {
1200 .procname = "panic_on_oom",
1201 .data = &sysctl_panic_on_oom,
1202 .maxlen = sizeof(sysctl_panic_on_oom),
1203 .mode = 0644,
1204 .proc_handler = proc_dointvec_minmax,
1205 .extra1 = &zero,
1206 .extra2 = &two,
1207 },
1208 {
1209 .procname = "oom_kill_allocating_task",
1210 .data = &sysctl_oom_kill_allocating_task,
1211 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1212 .mode = 0644,
1213 .proc_handler = proc_dointvec,
1214 },
1215 {
1216 .procname = "oom_dump_tasks",
1217 .data = &sysctl_oom_dump_tasks,
1218 .maxlen = sizeof(sysctl_oom_dump_tasks),
1219 .mode = 0644,
1220 .proc_handler = proc_dointvec,
1221 },
1222 {
1223 .procname = "overcommit_ratio",
1224 .data = &sysctl_overcommit_ratio,
1225 .maxlen = sizeof(sysctl_overcommit_ratio),
1226 .mode = 0644,
1227 .proc_handler = overcommit_ratio_handler,
1228 },
1229 {
1230 .procname = "overcommit_kbytes",
1231 .data = &sysctl_overcommit_kbytes,
1232 .maxlen = sizeof(sysctl_overcommit_kbytes),
1233 .mode = 0644,
1234 .proc_handler = overcommit_kbytes_handler,
1235 },
1236 {
1237 .procname = "page-cluster",
1238 .data = &page_cluster,
1239 .maxlen = sizeof(int),
1240 .mode = 0644,
1241 .proc_handler = proc_dointvec_minmax,
1242 .extra1 = &zero,
1243 },
1244 {
1245 .procname = "dirty_background_ratio",
1246 .data = &dirty_background_ratio,
1247 .maxlen = sizeof(dirty_background_ratio),
1248 .mode = 0644,
1249 .proc_handler = dirty_background_ratio_handler,
1250 .extra1 = &zero,
1251 .extra2 = &one_hundred,
1252 },
1253 {
1254 .procname = "dirty_background_bytes",
1255 .data = &dirty_background_bytes,
1256 .maxlen = sizeof(dirty_background_bytes),
1257 .mode = 0644,
1258 .proc_handler = dirty_background_bytes_handler,
1259 .extra1 = &one_ul,
1260 },
1261 {
1262 .procname = "dirty_ratio",
1263 .data = &vm_dirty_ratio,
1264 .maxlen = sizeof(vm_dirty_ratio),
1265 .mode = 0644,
1266 .proc_handler = dirty_ratio_handler,
1267 .extra1 = &zero,
1268 .extra2 = &one_hundred,
1269 },
1270 {
1271 .procname = "dirty_bytes",
1272 .data = &vm_dirty_bytes,
1273 .maxlen = sizeof(vm_dirty_bytes),
1274 .mode = 0644,
1275 .proc_handler = dirty_bytes_handler,
1276 .extra1 = &dirty_bytes_min,
1277 },
1278 {
1279 .procname = "dirty_writeback_centisecs",
1280 .data = &dirty_writeback_interval,
1281 .maxlen = sizeof(dirty_writeback_interval),
1282 .mode = 0644,
1283 .proc_handler = dirty_writeback_centisecs_handler,
1284 },
1285 {
1286 .procname = "dirty_expire_centisecs",
1287 .data = &dirty_expire_interval,
1288 .maxlen = sizeof(dirty_expire_interval),
1289 .mode = 0644,
1290 .proc_handler = proc_dointvec_minmax,
1291 .extra1 = &zero,
1292 },
1293 {
1294 .procname = "dirtytime_expire_seconds",
1295 .data = &dirtytime_expire_interval,
1296 .maxlen = sizeof(dirty_expire_interval),
1297 .mode = 0644,
1298 .proc_handler = dirtytime_interval_handler,
1299 .extra1 = &zero,
1300 },
1301 {
1302 .procname = "nr_pdflush_threads",
1303 .mode = 0444 /* read-only */,
1304 .proc_handler = pdflush_proc_obsolete,
1305 },
1306 {
1307 .procname = "swappiness",
1308 .data = &vm_swappiness,
1309 .maxlen = sizeof(vm_swappiness),
1310 .mode = 0644,
1311 .proc_handler = proc_dointvec_minmax,
1312 .extra1 = &zero,
1313 .extra2 = &one_hundred,
1314 },
1315#ifdef CONFIG_HUGETLB_PAGE
1316 {
1317 .procname = "nr_hugepages",
1318 .data = NULL,
1319 .maxlen = sizeof(unsigned long),
1320 .mode = 0644,
1321 .proc_handler = hugetlb_sysctl_handler,
1322 },
1323#ifdef CONFIG_NUMA
1324 {
1325 .procname = "nr_hugepages_mempolicy",
1326 .data = NULL,
1327 .maxlen = sizeof(unsigned long),
1328 .mode = 0644,
1329 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1330 },
1331#endif
1332 {
1333 .procname = "hugetlb_shm_group",
1334 .data = &sysctl_hugetlb_shm_group,
1335 .maxlen = sizeof(gid_t),
1336 .mode = 0644,
1337 .proc_handler = proc_dointvec,
1338 },
1339 {
1340 .procname = "hugepages_treat_as_movable",
1341 .data = &hugepages_treat_as_movable,
1342 .maxlen = sizeof(int),
1343 .mode = 0644,
1344 .proc_handler = proc_dointvec,
1345 },
1346 {
1347 .procname = "nr_overcommit_hugepages",
1348 .data = NULL,
1349 .maxlen = sizeof(unsigned long),
1350 .mode = 0644,
1351 .proc_handler = hugetlb_overcommit_handler,
1352 },
1353#endif
1354 {
1355 .procname = "lowmem_reserve_ratio",
1356 .data = &sysctl_lowmem_reserve_ratio,
1357 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1358 .mode = 0644,
1359 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1360 },
1361 {
1362 .procname = "drop_caches",
1363 .data = &sysctl_drop_caches,
1364 .maxlen = sizeof(int),
1365 .mode = 0644,
1366 .proc_handler = drop_caches_sysctl_handler,
1367 .extra1 = &one,
1368 .extra2 = &four,
1369 },
1370#ifdef CONFIG_COMPACTION
1371 {
1372 .procname = "compact_memory",
1373 .data = &sysctl_compact_memory,
1374 .maxlen = sizeof(int),
1375 .mode = 0200,
1376 .proc_handler = sysctl_compaction_handler,
1377 },
1378 {
1379 .procname = "extfrag_threshold",
1380 .data = &sysctl_extfrag_threshold,
1381 .maxlen = sizeof(int),
1382 .mode = 0644,
1383 .proc_handler = sysctl_extfrag_handler,
1384 .extra1 = &min_extfrag_threshold,
1385 .extra2 = &max_extfrag_threshold,
1386 },
1387 {
1388 .procname = "compact_unevictable_allowed",
1389 .data = &sysctl_compact_unevictable_allowed,
1390 .maxlen = sizeof(int),
1391 .mode = 0644,
1392 .proc_handler = proc_dointvec,
1393 .extra1 = &zero,
1394 .extra2 = &one,
1395 },
1396
1397#endif /* CONFIG_COMPACTION */
1398 {
1399 .procname = "min_free_kbytes",
1400 .data = &min_free_kbytes,
1401 .maxlen = sizeof(min_free_kbytes),
1402 .mode = 0644,
1403 .proc_handler = min_free_kbytes_sysctl_handler,
1404 .extra1 = &zero,
1405 },
1406 {
1407 .procname = "percpu_pagelist_fraction",
1408 .data = &percpu_pagelist_fraction,
1409 .maxlen = sizeof(percpu_pagelist_fraction),
1410 .mode = 0644,
1411 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1412 .extra1 = &zero,
1413 },
1414#ifdef CONFIG_MMU
1415 {
1416 .procname = "max_map_count",
1417 .data = &sysctl_max_map_count,
1418 .maxlen = sizeof(sysctl_max_map_count),
1419 .mode = 0644,
1420 .proc_handler = proc_dointvec_minmax,
1421 .extra1 = &zero,
1422 },
1423#else
1424 {
1425 .procname = "nr_trim_pages",
1426 .data = &sysctl_nr_trim_pages,
1427 .maxlen = sizeof(sysctl_nr_trim_pages),
1428 .mode = 0644,
1429 .proc_handler = proc_dointvec_minmax,
1430 .extra1 = &zero,
1431 },
1432#endif
1433 {
1434 .procname = "laptop_mode",
1435 .data = &laptop_mode,
1436 .maxlen = sizeof(laptop_mode),
1437 .mode = 0644,
1438 .proc_handler = proc_dointvec_jiffies,
1439 },
1440 {
1441 .procname = "block_dump",
1442 .data = &block_dump,
1443 .maxlen = sizeof(block_dump),
1444 .mode = 0644,
1445 .proc_handler = proc_dointvec,
1446 .extra1 = &zero,
1447 },
1448 {
1449 .procname = "vfs_cache_pressure",
1450 .data = &sysctl_vfs_cache_pressure,
1451 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1452 .mode = 0644,
1453 .proc_handler = proc_dointvec,
1454 .extra1 = &zero,
1455 },
1456#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1457 {
1458 .procname = "legacy_va_layout",
1459 .data = &sysctl_legacy_va_layout,
1460 .maxlen = sizeof(sysctl_legacy_va_layout),
1461 .mode = 0644,
1462 .proc_handler = proc_dointvec,
1463 .extra1 = &zero,
1464 },
1465#endif
1466#ifdef CONFIG_NUMA
1467 {
1468 .procname = "zone_reclaim_mode",
1469 .data = &zone_reclaim_mode,
1470 .maxlen = sizeof(zone_reclaim_mode),
1471 .mode = 0644,
1472 .proc_handler = proc_dointvec,
1473 .extra1 = &zero,
1474 },
1475 {
1476 .procname = "min_unmapped_ratio",
1477 .data = &sysctl_min_unmapped_ratio,
1478 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1479 .mode = 0644,
1480 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1481 .extra1 = &zero,
1482 .extra2 = &one_hundred,
1483 },
1484 {
1485 .procname = "min_slab_ratio",
1486 .data = &sysctl_min_slab_ratio,
1487 .maxlen = sizeof(sysctl_min_slab_ratio),
1488 .mode = 0644,
1489 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1490 .extra1 = &zero,
1491 .extra2 = &one_hundred,
1492 },
1493#endif
1494#ifdef CONFIG_SMP
1495 {
1496 .procname = "stat_interval",
1497 .data = &sysctl_stat_interval,
1498 .maxlen = sizeof(sysctl_stat_interval),
1499 .mode = 0644,
1500 .proc_handler = proc_dointvec_jiffies,
1501 },
1502#endif
1503#ifdef CONFIG_MMU
1504 {
1505 .procname = "mmap_min_addr",
1506 .data = &dac_mmap_min_addr,
1507 .maxlen = sizeof(unsigned long),
1508 .mode = 0644,
1509 .proc_handler = mmap_min_addr_handler,
1510 },
1511#endif
1512#ifdef CONFIG_NUMA
1513 {
1514 .procname = "numa_zonelist_order",
1515 .data = &numa_zonelist_order,
1516 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1517 .mode = 0644,
1518 .proc_handler = numa_zonelist_order_handler,
1519 },
1520#endif
1521#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1522 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1523 {
1524 .procname = "vdso_enabled",
1525#ifdef CONFIG_X86_32
1526 .data = &vdso32_enabled,
1527 .maxlen = sizeof(vdso32_enabled),
1528#else
1529 .data = &vdso_enabled,
1530 .maxlen = sizeof(vdso_enabled),
1531#endif
1532 .mode = 0644,
1533 .proc_handler = proc_dointvec,
1534 .extra1 = &zero,
1535 },
1536#endif
1537#ifdef CONFIG_HIGHMEM
1538 {
1539 .procname = "highmem_is_dirtyable",
1540 .data = &vm_highmem_is_dirtyable,
1541 .maxlen = sizeof(vm_highmem_is_dirtyable),
1542 .mode = 0644,
1543 .proc_handler = proc_dointvec_minmax,
1544 .extra1 = &zero,
1545 .extra2 = &one,
1546 },
1547#endif
1548#ifdef CONFIG_MEMORY_FAILURE
1549 {
1550 .procname = "memory_failure_early_kill",
1551 .data = &sysctl_memory_failure_early_kill,
1552 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1553 .mode = 0644,
1554 .proc_handler = proc_dointvec_minmax,
1555 .extra1 = &zero,
1556 .extra2 = &one,
1557 },
1558 {
1559 .procname = "memory_failure_recovery",
1560 .data = &sysctl_memory_failure_recovery,
1561 .maxlen = sizeof(sysctl_memory_failure_recovery),
1562 .mode = 0644,
1563 .proc_handler = proc_dointvec_minmax,
1564 .extra1 = &zero,
1565 .extra2 = &one,
1566 },
1567#endif
1568 {
1569 .procname = "user_reserve_kbytes",
1570 .data = &sysctl_user_reserve_kbytes,
1571 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1572 .mode = 0644,
1573 .proc_handler = proc_doulongvec_minmax,
1574 },
1575 {
1576 .procname = "admin_reserve_kbytes",
1577 .data = &sysctl_admin_reserve_kbytes,
1578 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1579 .mode = 0644,
1580 .proc_handler = proc_doulongvec_minmax,
1581 },
1582#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1583 {
1584 .procname = "mmap_rnd_bits",
1585 .data = &mmap_rnd_bits,
1586 .maxlen = sizeof(mmap_rnd_bits),
1587 .mode = 0600,
1588 .proc_handler = proc_dointvec_minmax,
1589 .extra1 = (void *)&mmap_rnd_bits_min,
1590 .extra2 = (void *)&mmap_rnd_bits_max,
1591 },
1592#endif
1593#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1594 {
1595 .procname = "mmap_rnd_compat_bits",
1596 .data = &mmap_rnd_compat_bits,
1597 .maxlen = sizeof(mmap_rnd_compat_bits),
1598 .mode = 0600,
1599 .proc_handler = proc_dointvec_minmax,
1600 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1601 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1602 },
1603#endif
1604 { }
1605};
1606
1607static struct ctl_table fs_table[] = {
1608 {
1609 .procname = "inode-nr",
1610 .data = &inodes_stat,
1611 .maxlen = 2*sizeof(long),
1612 .mode = 0444,
1613 .proc_handler = proc_nr_inodes,
1614 },
1615 {
1616 .procname = "inode-state",
1617 .data = &inodes_stat,
1618 .maxlen = 7*sizeof(long),
1619 .mode = 0444,
1620 .proc_handler = proc_nr_inodes,
1621 },
1622 {
1623 .procname = "file-nr",
1624 .data = &files_stat,
1625 .maxlen = sizeof(files_stat),
1626 .mode = 0444,
1627 .proc_handler = proc_nr_files,
1628 },
1629 {
1630 .procname = "file-max",
1631 .data = &files_stat.max_files,
1632 .maxlen = sizeof(files_stat.max_files),
1633 .mode = 0644,
1634 .proc_handler = proc_doulongvec_minmax,
1635 },
1636 {
1637 .procname = "nr_open",
1638 .data = &sysctl_nr_open,
1639 .maxlen = sizeof(int),
1640 .mode = 0644,
1641 .proc_handler = proc_dointvec_minmax,
1642 .extra1 = &sysctl_nr_open_min,
1643 .extra2 = &sysctl_nr_open_max,
1644 },
1645 {
1646 .procname = "dentry-state",
1647 .data = &dentry_stat,
1648 .maxlen = 6*sizeof(long),
1649 .mode = 0444,
1650 .proc_handler = proc_nr_dentry,
1651 },
1652 {
1653 .procname = "overflowuid",
1654 .data = &fs_overflowuid,
1655 .maxlen = sizeof(int),
1656 .mode = 0644,
1657 .proc_handler = proc_dointvec_minmax,
1658 .extra1 = &minolduid,
1659 .extra2 = &maxolduid,
1660 },
1661 {
1662 .procname = "overflowgid",
1663 .data = &fs_overflowgid,
1664 .maxlen = sizeof(int),
1665 .mode = 0644,
1666 .proc_handler = proc_dointvec_minmax,
1667 .extra1 = &minolduid,
1668 .extra2 = &maxolduid,
1669 },
1670#ifdef CONFIG_FILE_LOCKING
1671 {
1672 .procname = "leases-enable",
1673 .data = &leases_enable,
1674 .maxlen = sizeof(int),
1675 .mode = 0644,
1676 .proc_handler = proc_dointvec,
1677 },
1678#endif
1679#ifdef CONFIG_DNOTIFY
1680 {
1681 .procname = "dir-notify-enable",
1682 .data = &dir_notify_enable,
1683 .maxlen = sizeof(int),
1684 .mode = 0644,
1685 .proc_handler = proc_dointvec,
1686 },
1687#endif
1688#ifdef CONFIG_MMU
1689#ifdef CONFIG_FILE_LOCKING
1690 {
1691 .procname = "lease-break-time",
1692 .data = &lease_break_time,
1693 .maxlen = sizeof(int),
1694 .mode = 0644,
1695 .proc_handler = proc_dointvec,
1696 },
1697#endif
1698#ifdef CONFIG_AIO
1699 {
1700 .procname = "aio-nr",
1701 .data = &aio_nr,
1702 .maxlen = sizeof(aio_nr),
1703 .mode = 0444,
1704 .proc_handler = proc_doulongvec_minmax,
1705 },
1706 {
1707 .procname = "aio-max-nr",
1708 .data = &aio_max_nr,
1709 .maxlen = sizeof(aio_max_nr),
1710 .mode = 0644,
1711 .proc_handler = proc_doulongvec_minmax,
1712 },
1713#endif /* CONFIG_AIO */
1714#ifdef CONFIG_INOTIFY_USER
1715 {
1716 .procname = "inotify",
1717 .mode = 0555,
1718 .child = inotify_table,
1719 },
1720#endif
1721#ifdef CONFIG_EPOLL
1722 {
1723 .procname = "epoll",
1724 .mode = 0555,
1725 .child = epoll_table,
1726 },
1727#endif
1728#endif
1729 {
1730 .procname = "protected_symlinks",
1731 .data = &sysctl_protected_symlinks,
1732 .maxlen = sizeof(int),
1733 .mode = 0600,
1734 .proc_handler = proc_dointvec_minmax,
1735 .extra1 = &zero,
1736 .extra2 = &one,
1737 },
1738 {
1739 .procname = "protected_hardlinks",
1740 .data = &sysctl_protected_hardlinks,
1741 .maxlen = sizeof(int),
1742 .mode = 0600,
1743 .proc_handler = proc_dointvec_minmax,
1744 .extra1 = &zero,
1745 .extra2 = &one,
1746 },
1747 {
1748 .procname = "suid_dumpable",
1749 .data = &suid_dumpable,
1750 .maxlen = sizeof(int),
1751 .mode = 0644,
1752 .proc_handler = proc_dointvec_minmax_coredump,
1753 .extra1 = &zero,
1754 .extra2 = &two,
1755 },
1756#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1757 {
1758 .procname = "binfmt_misc",
1759 .mode = 0555,
1760 .child = sysctl_mount_point,
1761 },
1762#endif
1763 {
1764 .procname = "pipe-max-size",
1765 .data = &pipe_max_size,
1766 .maxlen = sizeof(int),
1767 .mode = 0644,
1768 .proc_handler = &pipe_proc_fn,
1769 .extra1 = &pipe_min_size,
1770 },
1771 {
1772 .procname = "pipe-user-pages-hard",
1773 .data = &pipe_user_pages_hard,
1774 .maxlen = sizeof(pipe_user_pages_hard),
1775 .mode = 0644,
1776 .proc_handler = proc_doulongvec_minmax,
1777 },
1778 {
1779 .procname = "pipe-user-pages-soft",
1780 .data = &pipe_user_pages_soft,
1781 .maxlen = sizeof(pipe_user_pages_soft),
1782 .mode = 0644,
1783 .proc_handler = proc_doulongvec_minmax,
1784 },
1785 { }
1786};
1787
1788static struct ctl_table debug_table[] = {
1789#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1790 {
1791 .procname = "exception-trace",
1792 .data = &show_unhandled_signals,
1793 .maxlen = sizeof(int),
1794 .mode = 0644,
1795 .proc_handler = proc_dointvec
1796 },
1797#endif
1798#if defined(CONFIG_OPTPROBES)
1799 {
1800 .procname = "kprobes-optimization",
1801 .data = &sysctl_kprobes_optimization,
1802 .maxlen = sizeof(int),
1803 .mode = 0644,
1804 .proc_handler = proc_kprobes_optimization_handler,
1805 .extra1 = &zero,
1806 .extra2 = &one,
1807 },
1808#endif
1809 { }
1810};
1811
1812static struct ctl_table dev_table[] = {
1813 { }
1814};
1815
1816int __init sysctl_init(void)
1817{
1818 struct ctl_table_header *hdr;
1819
1820 hdr = register_sysctl_table(sysctl_base_table);
1821 kmemleak_not_leak(hdr);
1822 return 0;
1823}
1824
1825#endif /* CONFIG_SYSCTL */
1826
1827/*
1828 * /proc/sys support
1829 */
1830
1831#ifdef CONFIG_PROC_SYSCTL
1832
1833static int _proc_do_string(char *data, int maxlen, int write,
1834 char __user *buffer,
1835 size_t *lenp, loff_t *ppos)
1836{
1837 size_t len;
1838 char __user *p;
1839 char c;
1840
1841 if (!data || !maxlen || !*lenp) {
1842 *lenp = 0;
1843 return 0;
1844 }
1845
1846 if (write) {
1847 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1848 /* Only continue writes not past the end of buffer. */
1849 len = strlen(data);
1850 if (len > maxlen - 1)
1851 len = maxlen - 1;
1852
1853 if (*ppos > len)
1854 return 0;
1855 len = *ppos;
1856 } else {
1857 /* Start writing from beginning of buffer. */
1858 len = 0;
1859 }
1860
1861 *ppos += *lenp;
1862 p = buffer;
1863 while ((p - buffer) < *lenp && len < maxlen - 1) {
1864 if (get_user(c, p++))
1865 return -EFAULT;
1866 if (c == 0 || c == '\n')
1867 break;
1868 data[len++] = c;
1869 }
1870 data[len] = 0;
1871 } else {
1872 len = strlen(data);
1873 if (len > maxlen)
1874 len = maxlen;
1875
1876 if (*ppos > len) {
1877 *lenp = 0;
1878 return 0;
1879 }
1880
1881 data += *ppos;
1882 len -= *ppos;
1883
1884 if (len > *lenp)
1885 len = *lenp;
1886 if (len)
1887 if (copy_to_user(buffer, data, len))
1888 return -EFAULT;
1889 if (len < *lenp) {
1890 if (put_user('\n', buffer + len))
1891 return -EFAULT;
1892 len++;
1893 }
1894 *lenp = len;
1895 *ppos += len;
1896 }
1897 return 0;
1898}
1899
1900static void warn_sysctl_write(struct ctl_table *table)
1901{
1902 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1903 "This will not be supported in the future. To silence this\n"
1904 "warning, set kernel.sysctl_writes_strict = -1\n",
1905 current->comm, table->procname);
1906}
1907
1908/**
1909 * proc_dostring - read a string sysctl
1910 * @table: the sysctl table
1911 * @write: %TRUE if this is a write to the sysctl file
1912 * @buffer: the user buffer
1913 * @lenp: the size of the user buffer
1914 * @ppos: file position
1915 *
1916 * Reads/writes a string from/to the user buffer. If the kernel
1917 * buffer provided is not large enough to hold the string, the
1918 * string is truncated. The copied string is %NULL-terminated.
1919 * If the string is being read by the user process, it is copied
1920 * and a newline '\n' is added. It is truncated if the buffer is
1921 * not large enough.
1922 *
1923 * Returns 0 on success.
1924 */
1925int proc_dostring(struct ctl_table *table, int write,
1926 void __user *buffer, size_t *lenp, loff_t *ppos)
1927{
1928 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1929 warn_sysctl_write(table);
1930
1931 return _proc_do_string((char *)(table->data), table->maxlen, write,
1932 (char __user *)buffer, lenp, ppos);
1933}
1934
1935static size_t proc_skip_spaces(char **buf)
1936{
1937 size_t ret;
1938 char *tmp = skip_spaces(*buf);
1939 ret = tmp - *buf;
1940 *buf = tmp;
1941 return ret;
1942}
1943
1944static void proc_skip_char(char **buf, size_t *size, const char v)
1945{
1946 while (*size) {
1947 if (**buf != v)
1948 break;
1949 (*size)--;
1950 (*buf)++;
1951 }
1952}
1953
1954#define TMPBUFLEN 22
1955/**
1956 * proc_get_long - reads an ASCII formatted integer from a user buffer
1957 *
1958 * @buf: a kernel buffer
1959 * @size: size of the kernel buffer
1960 * @val: this is where the number will be stored
1961 * @neg: set to %TRUE if number is negative
1962 * @perm_tr: a vector which contains the allowed trailers
1963 * @perm_tr_len: size of the perm_tr vector
1964 * @tr: pointer to store the trailer character
1965 *
1966 * In case of success %0 is returned and @buf and @size are updated with
1967 * the amount of bytes read. If @tr is non-NULL and a trailing
1968 * character exists (size is non-zero after returning from this
1969 * function), @tr is updated with the trailing character.
1970 */
1971static int proc_get_long(char **buf, size_t *size,
1972 unsigned long *val, bool *neg,
1973 const char *perm_tr, unsigned perm_tr_len, char *tr)
1974{
1975 int len;
1976 char *p, tmp[TMPBUFLEN];
1977
1978 if (!*size)
1979 return -EINVAL;
1980
1981 len = *size;
1982 if (len > TMPBUFLEN - 1)
1983 len = TMPBUFLEN - 1;
1984
1985 memcpy(tmp, *buf, len);
1986
1987 tmp[len] = 0;
1988 p = tmp;
1989 if (*p == '-' && *size > 1) {
1990 *neg = true;
1991 p++;
1992 } else
1993 *neg = false;
1994 if (!isdigit(*p))
1995 return -EINVAL;
1996
1997 *val = simple_strtoul(p, &p, 0);
1998
1999 len = p - tmp;
2000
2001 /* We don't know if the next char is whitespace thus we may accept
2002 * invalid integers (e.g. 1234...a) or two integers instead of one
2003 * (e.g. 123...1). So lets not allow such large numbers. */
2004 if (len == TMPBUFLEN - 1)
2005 return -EINVAL;
2006
2007 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2008 return -EINVAL;
2009
2010 if (tr && (len < *size))
2011 *tr = *p;
2012
2013 *buf += len;
2014 *size -= len;
2015
2016 return 0;
2017}
2018
2019/**
2020 * proc_put_long - converts an integer to a decimal ASCII formatted string
2021 *
2022 * @buf: the user buffer
2023 * @size: the size of the user buffer
2024 * @val: the integer to be converted
2025 * @neg: sign of the number, %TRUE for negative
2026 *
2027 * In case of success %0 is returned and @buf and @size are updated with
2028 * the amount of bytes written.
2029 */
2030static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2031 bool neg)
2032{
2033 int len;
2034 char tmp[TMPBUFLEN], *p = tmp;
2035
2036 sprintf(p, "%s%lu", neg ? "-" : "", val);
2037 len = strlen(tmp);
2038 if (len > *size)
2039 len = *size;
2040 if (copy_to_user(*buf, tmp, len))
2041 return -EFAULT;
2042 *size -= len;
2043 *buf += len;
2044 return 0;
2045}
2046#undef TMPBUFLEN
2047
2048static int proc_put_char(void __user **buf, size_t *size, char c)
2049{
2050 if (*size) {
2051 char __user **buffer = (char __user **)buf;
2052 if (put_user(c, *buffer))
2053 return -EFAULT;
2054 (*size)--, (*buffer)++;
2055 *buf = *buffer;
2056 }
2057 return 0;
2058}
2059
2060static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2061 int *valp,
2062 int write, void *data)
2063{
2064 if (write) {
2065 if (*negp) {
2066 if (*lvalp > (unsigned long) INT_MAX + 1)
2067 return -EINVAL;
2068 *valp = -*lvalp;
2069 } else {
2070 if (*lvalp > (unsigned long) INT_MAX)
2071 return -EINVAL;
2072 *valp = *lvalp;
2073 }
2074 } else {
2075 int val = *valp;
2076 if (val < 0) {
2077 *negp = true;
2078 *lvalp = -(unsigned long)val;
2079 } else {
2080 *negp = false;
2081 *lvalp = (unsigned long)val;
2082 }
2083 }
2084 return 0;
2085}
2086
2087static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2088
2089static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2090 int write, void __user *buffer,
2091 size_t *lenp, loff_t *ppos,
2092 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2093 int write, void *data),
2094 void *data)
2095{
2096 int *i, vleft, first = 1, err = 0;
2097 size_t left;
2098 char *kbuf = NULL, *p;
2099
2100 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2101 *lenp = 0;
2102 return 0;
2103 }
2104
2105 i = (int *) tbl_data;
2106 vleft = table->maxlen / sizeof(*i);
2107 left = *lenp;
2108
2109 if (!conv)
2110 conv = do_proc_dointvec_conv;
2111
2112 if (write) {
2113 if (*ppos) {
2114 switch (sysctl_writes_strict) {
2115 case SYSCTL_WRITES_STRICT:
2116 goto out;
2117 case SYSCTL_WRITES_WARN:
2118 warn_sysctl_write(table);
2119 break;
2120 default:
2121 break;
2122 }
2123 }
2124
2125 if (left > PAGE_SIZE - 1)
2126 left = PAGE_SIZE - 1;
2127 p = kbuf = memdup_user_nul(buffer, left);
2128 if (IS_ERR(kbuf))
2129 return PTR_ERR(kbuf);
2130 }
2131
2132 for (; left && vleft--; i++, first=0) {
2133 unsigned long lval;
2134 bool neg;
2135
2136 if (write) {
2137 left -= proc_skip_spaces(&p);
2138
2139 if (!left)
2140 break;
2141 err = proc_get_long(&p, &left, &lval, &neg,
2142 proc_wspace_sep,
2143 sizeof(proc_wspace_sep), NULL);
2144 if (err)
2145 break;
2146 if (conv(&neg, &lval, i, 1, data)) {
2147 err = -EINVAL;
2148 break;
2149 }
2150 } else {
2151 if (conv(&neg, &lval, i, 0, data)) {
2152 err = -EINVAL;
2153 break;
2154 }
2155 if (!first)
2156 err = proc_put_char(&buffer, &left, '\t');
2157 if (err)
2158 break;
2159 err = proc_put_long(&buffer, &left, lval, neg);
2160 if (err)
2161 break;
2162 }
2163 }
2164
2165 if (!write && !first && left && !err)
2166 err = proc_put_char(&buffer, &left, '\n');
2167 if (write && !err && left)
2168 left -= proc_skip_spaces(&p);
2169 if (write) {
2170 kfree(kbuf);
2171 if (first)
2172 return err ? : -EINVAL;
2173 }
2174 *lenp -= left;
2175out:
2176 *ppos += *lenp;
2177 return err;
2178}
2179
2180static int do_proc_dointvec(struct ctl_table *table, int write,
2181 void __user *buffer, size_t *lenp, loff_t *ppos,
2182 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2183 int write, void *data),
2184 void *data)
2185{
2186 return __do_proc_dointvec(table->data, table, write,
2187 buffer, lenp, ppos, conv, data);
2188}
2189
2190/**
2191 * proc_dointvec - read a vector of integers
2192 * @table: the sysctl table
2193 * @write: %TRUE if this is a write to the sysctl file
2194 * @buffer: the user buffer
2195 * @lenp: the size of the user buffer
2196 * @ppos: file position
2197 *
2198 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2199 * values from/to the user buffer, treated as an ASCII string.
2200 *
2201 * Returns 0 on success.
2202 */
2203int proc_dointvec(struct ctl_table *table, int write,
2204 void __user *buffer, size_t *lenp, loff_t *ppos)
2205{
2206 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2207 NULL,NULL);
2208}
2209
2210/*
2211 * Taint values can only be increased
2212 * This means we can safely use a temporary.
2213 */
2214static int proc_taint(struct ctl_table *table, int write,
2215 void __user *buffer, size_t *lenp, loff_t *ppos)
2216{
2217 struct ctl_table t;
2218 unsigned long tmptaint = get_taint();
2219 int err;
2220
2221 if (write && !capable(CAP_SYS_ADMIN))
2222 return -EPERM;
2223
2224 t = *table;
2225 t.data = &tmptaint;
2226 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2227 if (err < 0)
2228 return err;
2229
2230 if (write) {
2231 /*
2232 * Poor man's atomic or. Not worth adding a primitive
2233 * to everyone's atomic.h for this
2234 */
2235 int i;
2236 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2237 if ((tmptaint >> i) & 1)
2238 add_taint(i, LOCKDEP_STILL_OK);
2239 }
2240 }
2241
2242 return err;
2243}
2244
2245#ifdef CONFIG_PRINTK
2246static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2247 void __user *buffer, size_t *lenp, loff_t *ppos)
2248{
2249 if (write && !capable(CAP_SYS_ADMIN))
2250 return -EPERM;
2251
2252 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2253}
2254#endif
2255
2256struct do_proc_dointvec_minmax_conv_param {
2257 int *min;
2258 int *max;
2259};
2260
2261static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2262 int *valp,
2263 int write, void *data)
2264{
2265 struct do_proc_dointvec_minmax_conv_param *param = data;
2266 if (write) {
2267 int val = *negp ? -*lvalp : *lvalp;
2268 if ((param->min && *param->min > val) ||
2269 (param->max && *param->max < val))
2270 return -EINVAL;
2271 *valp = val;
2272 } else {
2273 int val = *valp;
2274 if (val < 0) {
2275 *negp = true;
2276 *lvalp = -(unsigned long)val;
2277 } else {
2278 *negp = false;
2279 *lvalp = (unsigned long)val;
2280 }
2281 }
2282 return 0;
2283}
2284
2285/**
2286 * proc_dointvec_minmax - read a vector of integers with min/max values
2287 * @table: the sysctl table
2288 * @write: %TRUE if this is a write to the sysctl file
2289 * @buffer: the user buffer
2290 * @lenp: the size of the user buffer
2291 * @ppos: file position
2292 *
2293 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2294 * values from/to the user buffer, treated as an ASCII string.
2295 *
2296 * This routine will ensure the values are within the range specified by
2297 * table->extra1 (min) and table->extra2 (max).
2298 *
2299 * Returns 0 on success.
2300 */
2301int proc_dointvec_minmax(struct ctl_table *table, int write,
2302 void __user *buffer, size_t *lenp, loff_t *ppos)
2303{
2304 struct do_proc_dointvec_minmax_conv_param param = {
2305 .min = (int *) table->extra1,
2306 .max = (int *) table->extra2,
2307 };
2308 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2309 do_proc_dointvec_minmax_conv, &param);
2310}
2311
2312static void validate_coredump_safety(void)
2313{
2314#ifdef CONFIG_COREDUMP
2315 if (suid_dumpable == SUID_DUMP_ROOT &&
2316 core_pattern[0] != '/' && core_pattern[0] != '|') {
2317 printk(KERN_WARNING "Unsafe core_pattern used with "\
2318 "suid_dumpable=2. Pipe handler or fully qualified "\
2319 "core dump path required.\n");
2320 }
2321#endif
2322}
2323
2324static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2325 void __user *buffer, size_t *lenp, loff_t *ppos)
2326{
2327 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2328 if (!error)
2329 validate_coredump_safety();
2330 return error;
2331}
2332
2333#ifdef CONFIG_COREDUMP
2334static int proc_dostring_coredump(struct ctl_table *table, int write,
2335 void __user *buffer, size_t *lenp, loff_t *ppos)
2336{
2337 int error = proc_dostring(table, write, buffer, lenp, ppos);
2338 if (!error)
2339 validate_coredump_safety();
2340 return error;
2341}
2342#endif
2343
2344static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2345 void __user *buffer,
2346 size_t *lenp, loff_t *ppos,
2347 unsigned long convmul,
2348 unsigned long convdiv)
2349{
2350 unsigned long *i, *min, *max;
2351 int vleft, first = 1, err = 0;
2352 size_t left;
2353 char *kbuf = NULL, *p;
2354
2355 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2356 *lenp = 0;
2357 return 0;
2358 }
2359
2360 i = (unsigned long *) data;
2361 min = (unsigned long *) table->extra1;
2362 max = (unsigned long *) table->extra2;
2363 vleft = table->maxlen / sizeof(unsigned long);
2364 left = *lenp;
2365
2366 if (write) {
2367 if (*ppos) {
2368 switch (sysctl_writes_strict) {
2369 case SYSCTL_WRITES_STRICT:
2370 goto out;
2371 case SYSCTL_WRITES_WARN:
2372 warn_sysctl_write(table);
2373 break;
2374 default:
2375 break;
2376 }
2377 }
2378
2379 if (left > PAGE_SIZE - 1)
2380 left = PAGE_SIZE - 1;
2381 p = kbuf = memdup_user_nul(buffer, left);
2382 if (IS_ERR(kbuf))
2383 return PTR_ERR(kbuf);
2384 }
2385
2386 for (; left && vleft--; i++, first = 0) {
2387 unsigned long val;
2388
2389 if (write) {
2390 bool neg;
2391
2392 left -= proc_skip_spaces(&p);
2393
2394 err = proc_get_long(&p, &left, &val, &neg,
2395 proc_wspace_sep,
2396 sizeof(proc_wspace_sep), NULL);
2397 if (err)
2398 break;
2399 if (neg)
2400 continue;
2401 if ((min && val < *min) || (max && val > *max))
2402 continue;
2403 *i = val;
2404 } else {
2405 val = convdiv * (*i) / convmul;
2406 if (!first) {
2407 err = proc_put_char(&buffer, &left, '\t');
2408 if (err)
2409 break;
2410 }
2411 err = proc_put_long(&buffer, &left, val, false);
2412 if (err)
2413 break;
2414 }
2415 }
2416
2417 if (!write && !first && left && !err)
2418 err = proc_put_char(&buffer, &left, '\n');
2419 if (write && !err)
2420 left -= proc_skip_spaces(&p);
2421 if (write) {
2422 kfree(kbuf);
2423 if (first)
2424 return err ? : -EINVAL;
2425 }
2426 *lenp -= left;
2427out:
2428 *ppos += *lenp;
2429 return err;
2430}
2431
2432static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2433 void __user *buffer,
2434 size_t *lenp, loff_t *ppos,
2435 unsigned long convmul,
2436 unsigned long convdiv)
2437{
2438 return __do_proc_doulongvec_minmax(table->data, table, write,
2439 buffer, lenp, ppos, convmul, convdiv);
2440}
2441
2442/**
2443 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2444 * @table: the sysctl table
2445 * @write: %TRUE if this is a write to the sysctl file
2446 * @buffer: the user buffer
2447 * @lenp: the size of the user buffer
2448 * @ppos: file position
2449 *
2450 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2451 * values from/to the user buffer, treated as an ASCII string.
2452 *
2453 * This routine will ensure the values are within the range specified by
2454 * table->extra1 (min) and table->extra2 (max).
2455 *
2456 * Returns 0 on success.
2457 */
2458int proc_doulongvec_minmax(struct ctl_table *table, int write,
2459 void __user *buffer, size_t *lenp, loff_t *ppos)
2460{
2461 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2462}
2463
2464/**
2465 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values 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. The values
2474 * are treated as milliseconds, and converted to jiffies when they are stored.
2475 *
2476 * This routine will ensure the values are within the range specified by
2477 * table->extra1 (min) and table->extra2 (max).
2478 *
2479 * Returns 0 on success.
2480 */
2481int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2482 void __user *buffer,
2483 size_t *lenp, loff_t *ppos)
2484{
2485 return do_proc_doulongvec_minmax(table, write, buffer,
2486 lenp, ppos, HZ, 1000l);
2487}
2488
2489
2490static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2491 int *valp,
2492 int write, void *data)
2493{
2494 if (write) {
2495 if (*lvalp > LONG_MAX / HZ)
2496 return 1;
2497 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2498 } else {
2499 int val = *valp;
2500 unsigned long lval;
2501 if (val < 0) {
2502 *negp = true;
2503 lval = -(unsigned long)val;
2504 } else {
2505 *negp = false;
2506 lval = (unsigned long)val;
2507 }
2508 *lvalp = lval / HZ;
2509 }
2510 return 0;
2511}
2512
2513static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2514 int *valp,
2515 int write, void *data)
2516{
2517 if (write) {
2518 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2519 return 1;
2520 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2521 } else {
2522 int val = *valp;
2523 unsigned long lval;
2524 if (val < 0) {
2525 *negp = true;
2526 lval = -(unsigned long)val;
2527 } else {
2528 *negp = false;
2529 lval = (unsigned long)val;
2530 }
2531 *lvalp = jiffies_to_clock_t(lval);
2532 }
2533 return 0;
2534}
2535
2536static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2537 int *valp,
2538 int write, void *data)
2539{
2540 if (write) {
2541 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2542
2543 if (jif > INT_MAX)
2544 return 1;
2545 *valp = (int)jif;
2546 } else {
2547 int val = *valp;
2548 unsigned long lval;
2549 if (val < 0) {
2550 *negp = true;
2551 lval = -(unsigned long)val;
2552 } else {
2553 *negp = false;
2554 lval = (unsigned long)val;
2555 }
2556 *lvalp = jiffies_to_msecs(lval);
2557 }
2558 return 0;
2559}
2560
2561/**
2562 * proc_dointvec_jiffies - read a vector of integers as seconds
2563 * @table: the sysctl table
2564 * @write: %TRUE if this is a write to the sysctl file
2565 * @buffer: the user buffer
2566 * @lenp: the size of the user buffer
2567 * @ppos: file position
2568 *
2569 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2570 * values from/to the user buffer, treated as an ASCII string.
2571 * The values read are assumed to be in seconds, and are converted into
2572 * jiffies.
2573 *
2574 * Returns 0 on success.
2575 */
2576int proc_dointvec_jiffies(struct ctl_table *table, int write,
2577 void __user *buffer, size_t *lenp, loff_t *ppos)
2578{
2579 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2580 do_proc_dointvec_jiffies_conv,NULL);
2581}
2582
2583/**
2584 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ 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: pointer to the 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 1/USER_HZ seconds, and
2594 * are converted into jiffies.
2595 *
2596 * Returns 0 on success.
2597 */
2598int proc_dointvec_userhz_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_userhz_jiffies_conv,NULL);
2603}
2604
2605/**
2606 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
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: file position
2612 * @ppos: the current position in the file
2613 *
2614 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2615 * values from/to the user buffer, treated as an ASCII string.
2616 * The values read are assumed to be in 1/1000 seconds, and
2617 * are converted into jiffies.
2618 *
2619 * Returns 0 on success.
2620 */
2621int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2622 void __user *buffer, size_t *lenp, loff_t *ppos)
2623{
2624 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2625 do_proc_dointvec_ms_jiffies_conv, NULL);
2626}
2627
2628static int proc_do_cad_pid(struct ctl_table *table, int write,
2629 void __user *buffer, size_t *lenp, loff_t *ppos)
2630{
2631 struct pid *new_pid;
2632 pid_t tmp;
2633 int r;
2634
2635 tmp = pid_vnr(cad_pid);
2636
2637 r = __do_proc_dointvec(&tmp, table, write, buffer,
2638 lenp, ppos, NULL, NULL);
2639 if (r || !write)
2640 return r;
2641
2642 new_pid = find_get_pid(tmp);
2643 if (!new_pid)
2644 return -ESRCH;
2645
2646 put_pid(xchg(&cad_pid, new_pid));
2647 return 0;
2648}
2649
2650/**
2651 * proc_do_large_bitmap - read/write from/to a large bitmap
2652 * @table: the sysctl table
2653 * @write: %TRUE if this is a write to the sysctl file
2654 * @buffer: the user buffer
2655 * @lenp: the size of the user buffer
2656 * @ppos: file position
2657 *
2658 * The bitmap is stored at table->data and the bitmap length (in bits)
2659 * in table->maxlen.
2660 *
2661 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2662 * large bitmaps may be represented in a compact manner. Writing into
2663 * the file will clear the bitmap then update it with the given input.
2664 *
2665 * Returns 0 on success.
2666 */
2667int proc_do_large_bitmap(struct ctl_table *table, int write,
2668 void __user *buffer, size_t *lenp, loff_t *ppos)
2669{
2670 int err = 0;
2671 bool first = 1;
2672 size_t left = *lenp;
2673 unsigned long bitmap_len = table->maxlen;
2674 unsigned long *bitmap = *(unsigned long **) table->data;
2675 unsigned long *tmp_bitmap = NULL;
2676 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2677
2678 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2679 *lenp = 0;
2680 return 0;
2681 }
2682
2683 if (write) {
2684 char *kbuf, *p;
2685
2686 if (left > PAGE_SIZE - 1)
2687 left = PAGE_SIZE - 1;
2688
2689 p = kbuf = memdup_user_nul(buffer, left);
2690 if (IS_ERR(kbuf))
2691 return PTR_ERR(kbuf);
2692
2693 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2694 GFP_KERNEL);
2695 if (!tmp_bitmap) {
2696 kfree(kbuf);
2697 return -ENOMEM;
2698 }
2699 proc_skip_char(&p, &left, '\n');
2700 while (!err && left) {
2701 unsigned long val_a, val_b;
2702 bool neg;
2703
2704 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2705 sizeof(tr_a), &c);
2706 if (err)
2707 break;
2708 if (val_a >= bitmap_len || neg) {
2709 err = -EINVAL;
2710 break;
2711 }
2712
2713 val_b = val_a;
2714 if (left) {
2715 p++;
2716 left--;
2717 }
2718
2719 if (c == '-') {
2720 err = proc_get_long(&p, &left, &val_b,
2721 &neg, tr_b, sizeof(tr_b),
2722 &c);
2723 if (err)
2724 break;
2725 if (val_b >= bitmap_len || neg ||
2726 val_a > val_b) {
2727 err = -EINVAL;
2728 break;
2729 }
2730 if (left) {
2731 p++;
2732 left--;
2733 }
2734 }
2735
2736 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2737 first = 0;
2738 proc_skip_char(&p, &left, '\n');
2739 }
2740 kfree(kbuf);
2741 } else {
2742 unsigned long bit_a, bit_b = 0;
2743
2744 while (left) {
2745 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2746 if (bit_a >= bitmap_len)
2747 break;
2748 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2749 bit_a + 1) - 1;
2750
2751 if (!first) {
2752 err = proc_put_char(&buffer, &left, ',');
2753 if (err)
2754 break;
2755 }
2756 err = proc_put_long(&buffer, &left, bit_a, false);
2757 if (err)
2758 break;
2759 if (bit_a != bit_b) {
2760 err = proc_put_char(&buffer, &left, '-');
2761 if (err)
2762 break;
2763 err = proc_put_long(&buffer, &left, bit_b, false);
2764 if (err)
2765 break;
2766 }
2767
2768 first = 0; bit_b++;
2769 }
2770 if (!err)
2771 err = proc_put_char(&buffer, &left, '\n');
2772 }
2773
2774 if (!err) {
2775 if (write) {
2776 if (*ppos)
2777 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2778 else
2779 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2780 }
2781 kfree(tmp_bitmap);
2782 *lenp -= left;
2783 *ppos += *lenp;
2784 return 0;
2785 } else {
2786 kfree(tmp_bitmap);
2787 return err;
2788 }
2789}
2790
2791#else /* CONFIG_PROC_SYSCTL */
2792
2793int proc_dostring(struct ctl_table *table, int write,
2794 void __user *buffer, size_t *lenp, loff_t *ppos)
2795{
2796 return -ENOSYS;
2797}
2798
2799int proc_dointvec(struct ctl_table *table, int write,
2800 void __user *buffer, size_t *lenp, loff_t *ppos)
2801{
2802 return -ENOSYS;
2803}
2804
2805int proc_dointvec_minmax(struct ctl_table *table, int write,
2806 void __user *buffer, size_t *lenp, loff_t *ppos)
2807{
2808 return -ENOSYS;
2809}
2810
2811int proc_dointvec_jiffies(struct ctl_table *table, int write,
2812 void __user *buffer, size_t *lenp, loff_t *ppos)
2813{
2814 return -ENOSYS;
2815}
2816
2817int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2818 void __user *buffer, size_t *lenp, loff_t *ppos)
2819{
2820 return -ENOSYS;
2821}
2822
2823int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2824 void __user *buffer, size_t *lenp, loff_t *ppos)
2825{
2826 return -ENOSYS;
2827}
2828
2829int proc_doulongvec_minmax(struct ctl_table *table, int write,
2830 void __user *buffer, size_t *lenp, loff_t *ppos)
2831{
2832 return -ENOSYS;
2833}
2834
2835int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2836 void __user *buffer,
2837 size_t *lenp, loff_t *ppos)
2838{
2839 return -ENOSYS;
2840}
2841
2842
2843#endif /* CONFIG_PROC_SYSCTL */
2844
2845/*
2846 * No sense putting this after each symbol definition, twice,
2847 * exception granted :-)
2848 */
2849EXPORT_SYMBOL(proc_dointvec);
2850EXPORT_SYMBOL(proc_dointvec_jiffies);
2851EXPORT_SYMBOL(proc_dointvec_minmax);
2852EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2853EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2854EXPORT_SYMBOL(proc_dostring);
2855EXPORT_SYMBOL(proc_doulongvec_minmax);
2856EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux 6.x block layer and io_uring tree(s)