Loading...
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/aio.h>
23#include <linux/mm.h>
24#include <linux/swap.h>
25#include <linux/slab.h>
26#include <linux/sysctl.h>
27#include <linux/bitmap.h>
28#include <linux/signal.h>
29#include <linux/printk.h>
30#include <linux/proc_fs.h>
31#include <linux/security.h>
32#include <linux/ctype.h>
33#include <linux/kmemcheck.h>
34#include <linux/kmemleak.h>
35#include <linux/fs.h>
36#include <linux/init.h>
37#include <linux/kernel.h>
38#include <linux/kobject.h>
39#include <linux/net.h>
40#include <linux/sysrq.h>
41#include <linux/highuid.h>
42#include <linux/writeback.h>
43#include <linux/ratelimit.h>
44#include <linux/compaction.h>
45#include <linux/hugetlb.h>
46#include <linux/initrd.h>
47#include <linux/key.h>
48#include <linux/times.h>
49#include <linux/limits.h>
50#include <linux/dcache.h>
51#include <linux/dnotify.h>
52#include <linux/syscalls.h>
53#include <linux/vmstat.h>
54#include <linux/nfs_fs.h>
55#include <linux/acpi.h>
56#include <linux/reboot.h>
57#include <linux/ftrace.h>
58#include <linux/perf_event.h>
59#include <linux/kprobes.h>
60#include <linux/pipe_fs_i.h>
61#include <linux/oom.h>
62#include <linux/kmod.h>
63#include <linux/capability.h>
64#include <linux/binfmts.h>
65#include <linux/sched/sysctl.h>
66#include <linux/kexec.h>
67#include <linux/bpf.h>
68#include <linux/mount.h>
69
70#include <linux/uaccess.h>
71#include <asm/processor.h>
72
73#ifdef CONFIG_X86
74#include <asm/nmi.h>
75#include <asm/stacktrace.h>
76#include <asm/io.h>
77#endif
78#ifdef CONFIG_SPARC
79#include <asm/setup.h>
80#endif
81#ifdef CONFIG_BSD_PROCESS_ACCT
82#include <linux/acct.h>
83#endif
84#ifdef CONFIG_RT_MUTEXES
85#include <linux/rtmutex.h>
86#endif
87#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88#include <linux/lockdep.h>
89#endif
90#ifdef CONFIG_CHR_DEV_SG
91#include <scsi/sg.h>
92#endif
93
94#ifdef CONFIG_LOCKUP_DETECTOR
95#include <linux/nmi.h>
96#endif
97
98#if defined(CONFIG_SYSCTL)
99
100/* External variables not in a header file. */
101extern int suid_dumpable;
102#ifdef CONFIG_COREDUMP
103extern int core_uses_pid;
104extern char core_pattern[];
105extern unsigned int core_pipe_limit;
106#endif
107extern int pid_max;
108extern int pid_max_min, pid_max_max;
109extern int percpu_pagelist_fraction;
110extern int latencytop_enabled;
111extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
112#ifndef CONFIG_MMU
113extern int sysctl_nr_trim_pages;
114#endif
115
116/* Constants used for minimum and maximum */
117#ifdef CONFIG_LOCKUP_DETECTOR
118static int sixty = 60;
119#endif
120
121static int __maybe_unused neg_one = -1;
122
123static int zero;
124static int __maybe_unused one = 1;
125static int __maybe_unused two = 2;
126static int __maybe_unused four = 4;
127static unsigned long one_ul = 1;
128static int one_hundred = 100;
129static int one_thousand = 1000;
130#ifdef CONFIG_PRINTK
131static int ten_thousand = 10000;
132#endif
133#ifdef CONFIG_PERF_EVENTS
134static int six_hundred_forty_kb = 640 * 1024;
135#endif
136
137/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139
140/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141static int maxolduid = 65535;
142static int minolduid;
143
144static int ngroups_max = NGROUPS_MAX;
145static const int cap_last_cap = CAP_LAST_CAP;
146
147/*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
148#ifdef CONFIG_DETECT_HUNG_TASK
149static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
150#endif
151
152#ifdef CONFIG_INOTIFY_USER
153#include <linux/inotify.h>
154#endif
155#ifdef CONFIG_SPARC
156#endif
157
158#ifdef __hppa__
159extern int pwrsw_enabled;
160#endif
161
162#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
163extern int unaligned_enabled;
164#endif
165
166#ifdef CONFIG_IA64
167extern int unaligned_dump_stack;
168#endif
169
170#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
171extern int no_unaligned_warning;
172#endif
173
174#ifdef CONFIG_PROC_SYSCTL
175
176#define SYSCTL_WRITES_LEGACY -1
177#define SYSCTL_WRITES_WARN 0
178#define SYSCTL_WRITES_STRICT 1
179
180static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
181
182static int proc_do_cad_pid(struct ctl_table *table, int write,
183 void __user *buffer, size_t *lenp, loff_t *ppos);
184static int proc_taint(struct ctl_table *table, int write,
185 void __user *buffer, size_t *lenp, loff_t *ppos);
186#endif
187
188#ifdef CONFIG_PRINTK
189static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
190 void __user *buffer, size_t *lenp, loff_t *ppos);
191#endif
192
193static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
194 void __user *buffer, size_t *lenp, loff_t *ppos);
195#ifdef CONFIG_COREDUMP
196static int proc_dostring_coredump(struct ctl_table *table, int write,
197 void __user *buffer, size_t *lenp, loff_t *ppos);
198#endif
199
200#ifdef CONFIG_MAGIC_SYSRQ
201/* Note: sysrq code uses it's own private copy */
202static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
203
204static int sysrq_sysctl_handler(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp,
206 loff_t *ppos)
207{
208 int error;
209
210 error = proc_dointvec(table, write, buffer, lenp, ppos);
211 if (error)
212 return error;
213
214 if (write)
215 sysrq_toggle_support(__sysrq_enabled);
216
217 return 0;
218}
219
220#endif
221
222static struct ctl_table kern_table[];
223static struct ctl_table vm_table[];
224static struct ctl_table fs_table[];
225static struct ctl_table debug_table[];
226static struct ctl_table dev_table[];
227extern struct ctl_table random_table[];
228#ifdef CONFIG_EPOLL
229extern struct ctl_table epoll_table[];
230#endif
231
232#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
233int sysctl_legacy_va_layout;
234#endif
235
236/* The default sysctl tables: */
237
238static struct ctl_table sysctl_base_table[] = {
239 {
240 .procname = "kernel",
241 .mode = 0555,
242 .child = kern_table,
243 },
244 {
245 .procname = "vm",
246 .mode = 0555,
247 .child = vm_table,
248 },
249 {
250 .procname = "fs",
251 .mode = 0555,
252 .child = fs_table,
253 },
254 {
255 .procname = "debug",
256 .mode = 0555,
257 .child = debug_table,
258 },
259 {
260 .procname = "dev",
261 .mode = 0555,
262 .child = dev_table,
263 },
264 { }
265};
266
267#ifdef CONFIG_SCHED_DEBUG
268static int min_sched_granularity_ns = 100000; /* 100 usecs */
269static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
270static int min_wakeup_granularity_ns; /* 0 usecs */
271static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
272#ifdef CONFIG_SMP
273static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
274static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
275#endif /* CONFIG_SMP */
276#endif /* CONFIG_SCHED_DEBUG */
277
278#ifdef CONFIG_COMPACTION
279static int min_extfrag_threshold;
280static int max_extfrag_threshold = 1000;
281#endif
282
283static struct ctl_table kern_table[] = {
284 {
285 .procname = "sched_child_runs_first",
286 .data = &sysctl_sched_child_runs_first,
287 .maxlen = sizeof(unsigned int),
288 .mode = 0644,
289 .proc_handler = proc_dointvec,
290 },
291#ifdef CONFIG_SCHED_DEBUG
292 {
293 .procname = "sched_min_granularity_ns",
294 .data = &sysctl_sched_min_granularity,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_latency_ns",
303 .data = &sysctl_sched_latency,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_sched_granularity_ns,
308 .extra2 = &max_sched_granularity_ns,
309 },
310 {
311 .procname = "sched_wakeup_granularity_ns",
312 .data = &sysctl_sched_wakeup_granularity,
313 .maxlen = sizeof(unsigned int),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_wakeup_granularity_ns,
317 .extra2 = &max_wakeup_granularity_ns,
318 },
319#ifdef CONFIG_SMP
320 {
321 .procname = "sched_tunable_scaling",
322 .data = &sysctl_sched_tunable_scaling,
323 .maxlen = sizeof(enum sched_tunable_scaling),
324 .mode = 0644,
325 .proc_handler = sched_proc_update_handler,
326 .extra1 = &min_sched_tunable_scaling,
327 .extra2 = &max_sched_tunable_scaling,
328 },
329 {
330 .procname = "sched_migration_cost_ns",
331 .data = &sysctl_sched_migration_cost,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = proc_dointvec,
335 },
336 {
337 .procname = "sched_nr_migrate",
338 .data = &sysctl_sched_nr_migrate,
339 .maxlen = sizeof(unsigned int),
340 .mode = 0644,
341 .proc_handler = proc_dointvec,
342 },
343 {
344 .procname = "sched_time_avg_ms",
345 .data = &sysctl_sched_time_avg,
346 .maxlen = sizeof(unsigned int),
347 .mode = 0644,
348 .proc_handler = proc_dointvec,
349 },
350#ifdef CONFIG_SCHEDSTATS
351 {
352 .procname = "sched_schedstats",
353 .data = NULL,
354 .maxlen = sizeof(unsigned int),
355 .mode = 0644,
356 .proc_handler = sysctl_schedstats,
357 .extra1 = &zero,
358 .extra2 = &one,
359 },
360#endif /* CONFIG_SCHEDSTATS */
361#endif /* CONFIG_SMP */
362#ifdef CONFIG_NUMA_BALANCING
363 {
364 .procname = "numa_balancing_scan_delay_ms",
365 .data = &sysctl_numa_balancing_scan_delay,
366 .maxlen = sizeof(unsigned int),
367 .mode = 0644,
368 .proc_handler = proc_dointvec,
369 },
370 {
371 .procname = "numa_balancing_scan_period_min_ms",
372 .data = &sysctl_numa_balancing_scan_period_min,
373 .maxlen = sizeof(unsigned int),
374 .mode = 0644,
375 .proc_handler = proc_dointvec,
376 },
377 {
378 .procname = "numa_balancing_scan_period_max_ms",
379 .data = &sysctl_numa_balancing_scan_period_max,
380 .maxlen = sizeof(unsigned int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec,
383 },
384 {
385 .procname = "numa_balancing_scan_size_mb",
386 .data = &sysctl_numa_balancing_scan_size,
387 .maxlen = sizeof(unsigned int),
388 .mode = 0644,
389 .proc_handler = proc_dointvec_minmax,
390 .extra1 = &one,
391 },
392 {
393 .procname = "numa_balancing",
394 .data = NULL, /* filled in by handler */
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = sysctl_numa_balancing,
398 .extra1 = &zero,
399 .extra2 = &one,
400 },
401#endif /* CONFIG_NUMA_BALANCING */
402#endif /* CONFIG_SCHED_DEBUG */
403 {
404 .procname = "sched_rt_period_us",
405 .data = &sysctl_sched_rt_period,
406 .maxlen = sizeof(unsigned int),
407 .mode = 0644,
408 .proc_handler = sched_rt_handler,
409 },
410 {
411 .procname = "sched_rt_runtime_us",
412 .data = &sysctl_sched_rt_runtime,
413 .maxlen = sizeof(int),
414 .mode = 0644,
415 .proc_handler = sched_rt_handler,
416 },
417 {
418 .procname = "sched_rr_timeslice_ms",
419 .data = &sched_rr_timeslice,
420 .maxlen = sizeof(int),
421 .mode = 0644,
422 .proc_handler = sched_rr_handler,
423 },
424#ifdef CONFIG_SCHED_AUTOGROUP
425 {
426 .procname = "sched_autogroup_enabled",
427 .data = &sysctl_sched_autogroup_enabled,
428 .maxlen = sizeof(unsigned int),
429 .mode = 0644,
430 .proc_handler = proc_dointvec_minmax,
431 .extra1 = &zero,
432 .extra2 = &one,
433 },
434#endif
435#ifdef CONFIG_CFS_BANDWIDTH
436 {
437 .procname = "sched_cfs_bandwidth_slice_us",
438 .data = &sysctl_sched_cfs_bandwidth_slice,
439 .maxlen = sizeof(unsigned int),
440 .mode = 0644,
441 .proc_handler = proc_dointvec_minmax,
442 .extra1 = &one,
443 },
444#endif
445#ifdef CONFIG_PROVE_LOCKING
446 {
447 .procname = "prove_locking",
448 .data = &prove_locking,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454#ifdef CONFIG_LOCK_STAT
455 {
456 .procname = "lock_stat",
457 .data = &lock_stat,
458 .maxlen = sizeof(int),
459 .mode = 0644,
460 .proc_handler = proc_dointvec,
461 },
462#endif
463 {
464 .procname = "panic",
465 .data = &panic_timeout,
466 .maxlen = sizeof(int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470#ifdef CONFIG_COREDUMP
471 {
472 .procname = "core_uses_pid",
473 .data = &core_uses_pid,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec,
477 },
478 {
479 .procname = "core_pattern",
480 .data = core_pattern,
481 .maxlen = CORENAME_MAX_SIZE,
482 .mode = 0644,
483 .proc_handler = proc_dostring_coredump,
484 },
485 {
486 .procname = "core_pipe_limit",
487 .data = &core_pipe_limit,
488 .maxlen = sizeof(unsigned int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef CONFIG_PROC_SYSCTL
494 {
495 .procname = "tainted",
496 .maxlen = sizeof(long),
497 .mode = 0644,
498 .proc_handler = proc_taint,
499 },
500 {
501 .procname = "sysctl_writes_strict",
502 .data = &sysctl_writes_strict,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec_minmax,
506 .extra1 = &neg_one,
507 .extra2 = &one,
508 },
509#endif
510#ifdef CONFIG_LATENCYTOP
511 {
512 .procname = "latencytop",
513 .data = &latencytop_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = sysctl_latencytop,
517 },
518#endif
519#ifdef CONFIG_BLK_DEV_INITRD
520 {
521 .procname = "real-root-dev",
522 .data = &real_root_dev,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527#endif
528 {
529 .procname = "print-fatal-signals",
530 .data = &print_fatal_signals,
531 .maxlen = sizeof(int),
532 .mode = 0644,
533 .proc_handler = proc_dointvec,
534 },
535#ifdef CONFIG_SPARC
536 {
537 .procname = "reboot-cmd",
538 .data = reboot_command,
539 .maxlen = 256,
540 .mode = 0644,
541 .proc_handler = proc_dostring,
542 },
543 {
544 .procname = "stop-a",
545 .data = &stop_a_enabled,
546 .maxlen = sizeof (int),
547 .mode = 0644,
548 .proc_handler = proc_dointvec,
549 },
550 {
551 .procname = "scons-poweroff",
552 .data = &scons_pwroff,
553 .maxlen = sizeof (int),
554 .mode = 0644,
555 .proc_handler = proc_dointvec,
556 },
557#endif
558#ifdef CONFIG_SPARC64
559 {
560 .procname = "tsb-ratio",
561 .data = &sysctl_tsb_ratio,
562 .maxlen = sizeof (int),
563 .mode = 0644,
564 .proc_handler = proc_dointvec,
565 },
566#endif
567#ifdef __hppa__
568 {
569 .procname = "soft-power",
570 .data = &pwrsw_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575#endif
576#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
577 {
578 .procname = "unaligned-trap",
579 .data = &unaligned_enabled,
580 .maxlen = sizeof (int),
581 .mode = 0644,
582 .proc_handler = proc_dointvec,
583 },
584#endif
585 {
586 .procname = "ctrl-alt-del",
587 .data = &C_A_D,
588 .maxlen = sizeof(int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592#ifdef CONFIG_FUNCTION_TRACER
593 {
594 .procname = "ftrace_enabled",
595 .data = &ftrace_enabled,
596 .maxlen = sizeof(int),
597 .mode = 0644,
598 .proc_handler = ftrace_enable_sysctl,
599 },
600#endif
601#ifdef CONFIG_STACK_TRACER
602 {
603 .procname = "stack_tracer_enabled",
604 .data = &stack_tracer_enabled,
605 .maxlen = sizeof(int),
606 .mode = 0644,
607 .proc_handler = stack_trace_sysctl,
608 },
609#endif
610#ifdef CONFIG_TRACING
611 {
612 .procname = "ftrace_dump_on_oops",
613 .data = &ftrace_dump_on_oops,
614 .maxlen = sizeof(int),
615 .mode = 0644,
616 .proc_handler = proc_dointvec,
617 },
618 {
619 .procname = "traceoff_on_warning",
620 .data = &__disable_trace_on_warning,
621 .maxlen = sizeof(__disable_trace_on_warning),
622 .mode = 0644,
623 .proc_handler = proc_dointvec,
624 },
625 {
626 .procname = "tracepoint_printk",
627 .data = &tracepoint_printk,
628 .maxlen = sizeof(tracepoint_printk),
629 .mode = 0644,
630 .proc_handler = tracepoint_printk_sysctl,
631 },
632#endif
633#ifdef CONFIG_KEXEC_CORE
634 {
635 .procname = "kexec_load_disabled",
636 .data = &kexec_load_disabled,
637 .maxlen = sizeof(int),
638 .mode = 0644,
639 /* only handle a transition from default "0" to "1" */
640 .proc_handler = proc_dointvec_minmax,
641 .extra1 = &one,
642 .extra2 = &one,
643 },
644#endif
645#ifdef CONFIG_MODULES
646 {
647 .procname = "modprobe",
648 .data = &modprobe_path,
649 .maxlen = KMOD_PATH_LEN,
650 .mode = 0644,
651 .proc_handler = proc_dostring,
652 },
653 {
654 .procname = "modules_disabled",
655 .data = &modules_disabled,
656 .maxlen = sizeof(int),
657 .mode = 0644,
658 /* only handle a transition from default "0" to "1" */
659 .proc_handler = proc_dointvec_minmax,
660 .extra1 = &one,
661 .extra2 = &one,
662 },
663#endif
664#ifdef CONFIG_UEVENT_HELPER
665 {
666 .procname = "hotplug",
667 .data = &uevent_helper,
668 .maxlen = UEVENT_HELPER_PATH_LEN,
669 .mode = 0644,
670 .proc_handler = proc_dostring,
671 },
672#endif
673#ifdef CONFIG_CHR_DEV_SG
674 {
675 .procname = "sg-big-buff",
676 .data = &sg_big_buff,
677 .maxlen = sizeof (int),
678 .mode = 0444,
679 .proc_handler = proc_dointvec,
680 },
681#endif
682#ifdef CONFIG_BSD_PROCESS_ACCT
683 {
684 .procname = "acct",
685 .data = &acct_parm,
686 .maxlen = 3*sizeof(int),
687 .mode = 0644,
688 .proc_handler = proc_dointvec,
689 },
690#endif
691#ifdef CONFIG_MAGIC_SYSRQ
692 {
693 .procname = "sysrq",
694 .data = &__sysrq_enabled,
695 .maxlen = sizeof (int),
696 .mode = 0644,
697 .proc_handler = sysrq_sysctl_handler,
698 },
699#endif
700#ifdef CONFIG_PROC_SYSCTL
701 {
702 .procname = "cad_pid",
703 .data = NULL,
704 .maxlen = sizeof (int),
705 .mode = 0600,
706 .proc_handler = proc_do_cad_pid,
707 },
708#endif
709 {
710 .procname = "threads-max",
711 .data = NULL,
712 .maxlen = sizeof(int),
713 .mode = 0644,
714 .proc_handler = sysctl_max_threads,
715 },
716 {
717 .procname = "random",
718 .mode = 0555,
719 .child = random_table,
720 },
721 {
722 .procname = "usermodehelper",
723 .mode = 0555,
724 .child = usermodehelper_table,
725 },
726 {
727 .procname = "overflowuid",
728 .data = &overflowuid,
729 .maxlen = sizeof(int),
730 .mode = 0644,
731 .proc_handler = proc_dointvec_minmax,
732 .extra1 = &minolduid,
733 .extra2 = &maxolduid,
734 },
735 {
736 .procname = "overflowgid",
737 .data = &overflowgid,
738 .maxlen = sizeof(int),
739 .mode = 0644,
740 .proc_handler = proc_dointvec_minmax,
741 .extra1 = &minolduid,
742 .extra2 = &maxolduid,
743 },
744#ifdef CONFIG_S390
745#ifdef CONFIG_MATHEMU
746 {
747 .procname = "ieee_emulation_warnings",
748 .data = &sysctl_ieee_emulation_warnings,
749 .maxlen = sizeof(int),
750 .mode = 0644,
751 .proc_handler = proc_dointvec,
752 },
753#endif
754 {
755 .procname = "userprocess_debug",
756 .data = &show_unhandled_signals,
757 .maxlen = sizeof(int),
758 .mode = 0644,
759 .proc_handler = proc_dointvec,
760 },
761#endif
762 {
763 .procname = "pid_max",
764 .data = &pid_max,
765 .maxlen = sizeof (int),
766 .mode = 0644,
767 .proc_handler = proc_dointvec_minmax,
768 .extra1 = &pid_max_min,
769 .extra2 = &pid_max_max,
770 },
771 {
772 .procname = "panic_on_oops",
773 .data = &panic_on_oops,
774 .maxlen = sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec,
777 },
778#if defined CONFIG_PRINTK
779 {
780 .procname = "printk",
781 .data = &console_loglevel,
782 .maxlen = 4*sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec,
785 },
786 {
787 .procname = "printk_ratelimit",
788 .data = &printk_ratelimit_state.interval,
789 .maxlen = sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec_jiffies,
792 },
793 {
794 .procname = "printk_ratelimit_burst",
795 .data = &printk_ratelimit_state.burst,
796 .maxlen = sizeof(int),
797 .mode = 0644,
798 .proc_handler = proc_dointvec,
799 },
800 {
801 .procname = "printk_delay",
802 .data = &printk_delay_msec,
803 .maxlen = sizeof(int),
804 .mode = 0644,
805 .proc_handler = proc_dointvec_minmax,
806 .extra1 = &zero,
807 .extra2 = &ten_thousand,
808 },
809 {
810 .procname = "printk_devkmsg",
811 .data = devkmsg_log_str,
812 .maxlen = DEVKMSG_STR_MAX_SIZE,
813 .mode = 0644,
814 .proc_handler = devkmsg_sysctl_set_loglvl,
815 },
816 {
817 .procname = "dmesg_restrict",
818 .data = &dmesg_restrict,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec_minmax_sysadmin,
822 .extra1 = &zero,
823 .extra2 = &one,
824 },
825 {
826 .procname = "kptr_restrict",
827 .data = &kptr_restrict,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec_minmax_sysadmin,
831 .extra1 = &zero,
832 .extra2 = &two,
833 },
834#endif
835 {
836 .procname = "ngroups_max",
837 .data = &ngroups_max,
838 .maxlen = sizeof (int),
839 .mode = 0444,
840 .proc_handler = proc_dointvec,
841 },
842 {
843 .procname = "cap_last_cap",
844 .data = (void *)&cap_last_cap,
845 .maxlen = sizeof(int),
846 .mode = 0444,
847 .proc_handler = proc_dointvec,
848 },
849#if defined(CONFIG_LOCKUP_DETECTOR)
850 {
851 .procname = "watchdog",
852 .data = &watchdog_user_enabled,
853 .maxlen = sizeof (int),
854 .mode = 0644,
855 .proc_handler = proc_watchdog,
856 .extra1 = &zero,
857 .extra2 = &one,
858 },
859 {
860 .procname = "watchdog_thresh",
861 .data = &watchdog_thresh,
862 .maxlen = sizeof(int),
863 .mode = 0644,
864 .proc_handler = proc_watchdog_thresh,
865 .extra1 = &zero,
866 .extra2 = &sixty,
867 },
868 {
869 .procname = "nmi_watchdog",
870 .data = &nmi_watchdog_enabled,
871 .maxlen = sizeof (int),
872 .mode = 0644,
873 .proc_handler = proc_nmi_watchdog,
874 .extra1 = &zero,
875#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
876 .extra2 = &one,
877#else
878 .extra2 = &zero,
879#endif
880 },
881 {
882 .procname = "soft_watchdog",
883 .data = &soft_watchdog_enabled,
884 .maxlen = sizeof (int),
885 .mode = 0644,
886 .proc_handler = proc_soft_watchdog,
887 .extra1 = &zero,
888 .extra2 = &one,
889 },
890 {
891 .procname = "watchdog_cpumask",
892 .data = &watchdog_cpumask_bits,
893 .maxlen = NR_CPUS,
894 .mode = 0644,
895 .proc_handler = proc_watchdog_cpumask,
896 },
897 {
898 .procname = "softlockup_panic",
899 .data = &softlockup_panic,
900 .maxlen = sizeof(int),
901 .mode = 0644,
902 .proc_handler = proc_dointvec_minmax,
903 .extra1 = &zero,
904 .extra2 = &one,
905 },
906#ifdef CONFIG_HARDLOCKUP_DETECTOR
907 {
908 .procname = "hardlockup_panic",
909 .data = &hardlockup_panic,
910 .maxlen = sizeof(int),
911 .mode = 0644,
912 .proc_handler = proc_dointvec_minmax,
913 .extra1 = &zero,
914 .extra2 = &one,
915 },
916#endif
917#ifdef CONFIG_SMP
918 {
919 .procname = "softlockup_all_cpu_backtrace",
920 .data = &sysctl_softlockup_all_cpu_backtrace,
921 .maxlen = sizeof(int),
922 .mode = 0644,
923 .proc_handler = proc_dointvec_minmax,
924 .extra1 = &zero,
925 .extra2 = &one,
926 },
927 {
928 .procname = "hardlockup_all_cpu_backtrace",
929 .data = &sysctl_hardlockup_all_cpu_backtrace,
930 .maxlen = sizeof(int),
931 .mode = 0644,
932 .proc_handler = proc_dointvec_minmax,
933 .extra1 = &zero,
934 .extra2 = &one,
935 },
936#endif /* CONFIG_SMP */
937#endif
938#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
939 {
940 .procname = "unknown_nmi_panic",
941 .data = &unknown_nmi_panic,
942 .maxlen = sizeof (int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec,
945 },
946#endif
947#if defined(CONFIG_X86)
948 {
949 .procname = "panic_on_unrecovered_nmi",
950 .data = &panic_on_unrecovered_nmi,
951 .maxlen = sizeof(int),
952 .mode = 0644,
953 .proc_handler = proc_dointvec,
954 },
955 {
956 .procname = "panic_on_io_nmi",
957 .data = &panic_on_io_nmi,
958 .maxlen = sizeof(int),
959 .mode = 0644,
960 .proc_handler = proc_dointvec,
961 },
962#ifdef CONFIG_DEBUG_STACKOVERFLOW
963 {
964 .procname = "panic_on_stackoverflow",
965 .data = &sysctl_panic_on_stackoverflow,
966 .maxlen = sizeof(int),
967 .mode = 0644,
968 .proc_handler = proc_dointvec,
969 },
970#endif
971 {
972 .procname = "bootloader_type",
973 .data = &bootloader_type,
974 .maxlen = sizeof (int),
975 .mode = 0444,
976 .proc_handler = proc_dointvec,
977 },
978 {
979 .procname = "bootloader_version",
980 .data = &bootloader_version,
981 .maxlen = sizeof (int),
982 .mode = 0444,
983 .proc_handler = proc_dointvec,
984 },
985 {
986 .procname = "io_delay_type",
987 .data = &io_delay_type,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992#endif
993#if defined(CONFIG_MMU)
994 {
995 .procname = "randomize_va_space",
996 .data = &randomize_va_space,
997 .maxlen = sizeof(int),
998 .mode = 0644,
999 .proc_handler = proc_dointvec,
1000 },
1001#endif
1002#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1003 {
1004 .procname = "spin_retry",
1005 .data = &spin_retry,
1006 .maxlen = sizeof (int),
1007 .mode = 0644,
1008 .proc_handler = proc_dointvec,
1009 },
1010#endif
1011#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1012 {
1013 .procname = "acpi_video_flags",
1014 .data = &acpi_realmode_flags,
1015 .maxlen = sizeof (unsigned long),
1016 .mode = 0644,
1017 .proc_handler = proc_doulongvec_minmax,
1018 },
1019#endif
1020#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1021 {
1022 .procname = "ignore-unaligned-usertrap",
1023 .data = &no_unaligned_warning,
1024 .maxlen = sizeof (int),
1025 .mode = 0644,
1026 .proc_handler = proc_dointvec,
1027 },
1028#endif
1029#ifdef CONFIG_IA64
1030 {
1031 .procname = "unaligned-dump-stack",
1032 .data = &unaligned_dump_stack,
1033 .maxlen = sizeof (int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec,
1036 },
1037#endif
1038#ifdef CONFIG_DETECT_HUNG_TASK
1039 {
1040 .procname = "hung_task_panic",
1041 .data = &sysctl_hung_task_panic,
1042 .maxlen = sizeof(int),
1043 .mode = 0644,
1044 .proc_handler = proc_dointvec_minmax,
1045 .extra1 = &zero,
1046 .extra2 = &one,
1047 },
1048 {
1049 .procname = "hung_task_check_count",
1050 .data = &sysctl_hung_task_check_count,
1051 .maxlen = sizeof(int),
1052 .mode = 0644,
1053 .proc_handler = proc_dointvec_minmax,
1054 .extra1 = &zero,
1055 },
1056 {
1057 .procname = "hung_task_timeout_secs",
1058 .data = &sysctl_hung_task_timeout_secs,
1059 .maxlen = sizeof(unsigned long),
1060 .mode = 0644,
1061 .proc_handler = proc_dohung_task_timeout_secs,
1062 .extra2 = &hung_task_timeout_max,
1063 },
1064 {
1065 .procname = "hung_task_warnings",
1066 .data = &sysctl_hung_task_warnings,
1067 .maxlen = sizeof(int),
1068 .mode = 0644,
1069 .proc_handler = proc_dointvec_minmax,
1070 .extra1 = &neg_one,
1071 },
1072#endif
1073#ifdef CONFIG_RT_MUTEXES
1074 {
1075 .procname = "max_lock_depth",
1076 .data = &max_lock_depth,
1077 .maxlen = sizeof(int),
1078 .mode = 0644,
1079 .proc_handler = proc_dointvec,
1080 },
1081#endif
1082 {
1083 .procname = "poweroff_cmd",
1084 .data = &poweroff_cmd,
1085 .maxlen = POWEROFF_CMD_PATH_LEN,
1086 .mode = 0644,
1087 .proc_handler = proc_dostring,
1088 },
1089#ifdef CONFIG_KEYS
1090 {
1091 .procname = "keys",
1092 .mode = 0555,
1093 .child = key_sysctls,
1094 },
1095#endif
1096#ifdef CONFIG_PERF_EVENTS
1097 /*
1098 * User-space scripts rely on the existence of this file
1099 * as a feature check for perf_events being enabled.
1100 *
1101 * So it's an ABI, do not remove!
1102 */
1103 {
1104 .procname = "perf_event_paranoid",
1105 .data = &sysctl_perf_event_paranoid,
1106 .maxlen = sizeof(sysctl_perf_event_paranoid),
1107 .mode = 0644,
1108 .proc_handler = proc_dointvec,
1109 },
1110 {
1111 .procname = "perf_event_mlock_kb",
1112 .data = &sysctl_perf_event_mlock,
1113 .maxlen = sizeof(sysctl_perf_event_mlock),
1114 .mode = 0644,
1115 .proc_handler = proc_dointvec,
1116 },
1117 {
1118 .procname = "perf_event_max_sample_rate",
1119 .data = &sysctl_perf_event_sample_rate,
1120 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1121 .mode = 0644,
1122 .proc_handler = perf_proc_update_handler,
1123 .extra1 = &one,
1124 },
1125 {
1126 .procname = "perf_cpu_time_max_percent",
1127 .data = &sysctl_perf_cpu_time_max_percent,
1128 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1129 .mode = 0644,
1130 .proc_handler = perf_cpu_time_max_percent_handler,
1131 .extra1 = &zero,
1132 .extra2 = &one_hundred,
1133 },
1134 {
1135 .procname = "perf_event_max_stack",
1136 .data = &sysctl_perf_event_max_stack,
1137 .maxlen = sizeof(sysctl_perf_event_max_stack),
1138 .mode = 0644,
1139 .proc_handler = perf_event_max_stack_handler,
1140 .extra1 = &zero,
1141 .extra2 = &six_hundred_forty_kb,
1142 },
1143 {
1144 .procname = "perf_event_max_contexts_per_stack",
1145 .data = &sysctl_perf_event_max_contexts_per_stack,
1146 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1147 .mode = 0644,
1148 .proc_handler = perf_event_max_stack_handler,
1149 .extra1 = &zero,
1150 .extra2 = &one_thousand,
1151 },
1152#endif
1153#ifdef CONFIG_KMEMCHECK
1154 {
1155 .procname = "kmemcheck",
1156 .data = &kmemcheck_enabled,
1157 .maxlen = sizeof(int),
1158 .mode = 0644,
1159 .proc_handler = proc_dointvec,
1160 },
1161#endif
1162 {
1163 .procname = "panic_on_warn",
1164 .data = &panic_on_warn,
1165 .maxlen = sizeof(int),
1166 .mode = 0644,
1167 .proc_handler = proc_dointvec_minmax,
1168 .extra1 = &zero,
1169 .extra2 = &one,
1170 },
1171#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1172 {
1173 .procname = "timer_migration",
1174 .data = &sysctl_timer_migration,
1175 .maxlen = sizeof(unsigned int),
1176 .mode = 0644,
1177 .proc_handler = timer_migration_handler,
1178 },
1179#endif
1180#ifdef CONFIG_BPF_SYSCALL
1181 {
1182 .procname = "unprivileged_bpf_disabled",
1183 .data = &sysctl_unprivileged_bpf_disabled,
1184 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1185 .mode = 0644,
1186 /* only handle a transition from default "0" to "1" */
1187 .proc_handler = proc_dointvec_minmax,
1188 .extra1 = &one,
1189 .extra2 = &one,
1190 },
1191#endif
1192#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1193 {
1194 .procname = "panic_on_rcu_stall",
1195 .data = &sysctl_panic_on_rcu_stall,
1196 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1197 .mode = 0644,
1198 .proc_handler = proc_dointvec_minmax,
1199 .extra1 = &zero,
1200 .extra2 = &one,
1201 },
1202#endif
1203 { }
1204};
1205
1206static struct ctl_table vm_table[] = {
1207 {
1208 .procname = "overcommit_memory",
1209 .data = &sysctl_overcommit_memory,
1210 .maxlen = sizeof(sysctl_overcommit_memory),
1211 .mode = 0644,
1212 .proc_handler = proc_dointvec_minmax,
1213 .extra1 = &zero,
1214 .extra2 = &two,
1215 },
1216 {
1217 .procname = "panic_on_oom",
1218 .data = &sysctl_panic_on_oom,
1219 .maxlen = sizeof(sysctl_panic_on_oom),
1220 .mode = 0644,
1221 .proc_handler = proc_dointvec_minmax,
1222 .extra1 = &zero,
1223 .extra2 = &two,
1224 },
1225 {
1226 .procname = "oom_kill_allocating_task",
1227 .data = &sysctl_oom_kill_allocating_task,
1228 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 },
1232 {
1233 .procname = "oom_dump_tasks",
1234 .data = &sysctl_oom_dump_tasks,
1235 .maxlen = sizeof(sysctl_oom_dump_tasks),
1236 .mode = 0644,
1237 .proc_handler = proc_dointvec,
1238 },
1239 {
1240 .procname = "overcommit_ratio",
1241 .data = &sysctl_overcommit_ratio,
1242 .maxlen = sizeof(sysctl_overcommit_ratio),
1243 .mode = 0644,
1244 .proc_handler = overcommit_ratio_handler,
1245 },
1246 {
1247 .procname = "overcommit_kbytes",
1248 .data = &sysctl_overcommit_kbytes,
1249 .maxlen = sizeof(sysctl_overcommit_kbytes),
1250 .mode = 0644,
1251 .proc_handler = overcommit_kbytes_handler,
1252 },
1253 {
1254 .procname = "page-cluster",
1255 .data = &page_cluster,
1256 .maxlen = sizeof(int),
1257 .mode = 0644,
1258 .proc_handler = proc_dointvec_minmax,
1259 .extra1 = &zero,
1260 },
1261 {
1262 .procname = "dirty_background_ratio",
1263 .data = &dirty_background_ratio,
1264 .maxlen = sizeof(dirty_background_ratio),
1265 .mode = 0644,
1266 .proc_handler = dirty_background_ratio_handler,
1267 .extra1 = &zero,
1268 .extra2 = &one_hundred,
1269 },
1270 {
1271 .procname = "dirty_background_bytes",
1272 .data = &dirty_background_bytes,
1273 .maxlen = sizeof(dirty_background_bytes),
1274 .mode = 0644,
1275 .proc_handler = dirty_background_bytes_handler,
1276 .extra1 = &one_ul,
1277 },
1278 {
1279 .procname = "dirty_ratio",
1280 .data = &vm_dirty_ratio,
1281 .maxlen = sizeof(vm_dirty_ratio),
1282 .mode = 0644,
1283 .proc_handler = dirty_ratio_handler,
1284 .extra1 = &zero,
1285 .extra2 = &one_hundred,
1286 },
1287 {
1288 .procname = "dirty_bytes",
1289 .data = &vm_dirty_bytes,
1290 .maxlen = sizeof(vm_dirty_bytes),
1291 .mode = 0644,
1292 .proc_handler = dirty_bytes_handler,
1293 .extra1 = &dirty_bytes_min,
1294 },
1295 {
1296 .procname = "dirty_writeback_centisecs",
1297 .data = &dirty_writeback_interval,
1298 .maxlen = sizeof(dirty_writeback_interval),
1299 .mode = 0644,
1300 .proc_handler = dirty_writeback_centisecs_handler,
1301 },
1302 {
1303 .procname = "dirty_expire_centisecs",
1304 .data = &dirty_expire_interval,
1305 .maxlen = sizeof(dirty_expire_interval),
1306 .mode = 0644,
1307 .proc_handler = proc_dointvec_minmax,
1308 .extra1 = &zero,
1309 },
1310 {
1311 .procname = "dirtytime_expire_seconds",
1312 .data = &dirtytime_expire_interval,
1313 .maxlen = sizeof(dirty_expire_interval),
1314 .mode = 0644,
1315 .proc_handler = dirtytime_interval_handler,
1316 .extra1 = &zero,
1317 },
1318 {
1319 .procname = "nr_pdflush_threads",
1320 .mode = 0444 /* read-only */,
1321 .proc_handler = pdflush_proc_obsolete,
1322 },
1323 {
1324 .procname = "swappiness",
1325 .data = &vm_swappiness,
1326 .maxlen = sizeof(vm_swappiness),
1327 .mode = 0644,
1328 .proc_handler = proc_dointvec_minmax,
1329 .extra1 = &zero,
1330 .extra2 = &one_hundred,
1331 },
1332#ifdef CONFIG_HUGETLB_PAGE
1333 {
1334 .procname = "nr_hugepages",
1335 .data = NULL,
1336 .maxlen = sizeof(unsigned long),
1337 .mode = 0644,
1338 .proc_handler = hugetlb_sysctl_handler,
1339 },
1340#ifdef CONFIG_NUMA
1341 {
1342 .procname = "nr_hugepages_mempolicy",
1343 .data = NULL,
1344 .maxlen = sizeof(unsigned long),
1345 .mode = 0644,
1346 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1347 },
1348#endif
1349 {
1350 .procname = "hugetlb_shm_group",
1351 .data = &sysctl_hugetlb_shm_group,
1352 .maxlen = sizeof(gid_t),
1353 .mode = 0644,
1354 .proc_handler = proc_dointvec,
1355 },
1356 {
1357 .procname = "hugepages_treat_as_movable",
1358 .data = &hugepages_treat_as_movable,
1359 .maxlen = sizeof(int),
1360 .mode = 0644,
1361 .proc_handler = proc_dointvec,
1362 },
1363 {
1364 .procname = "nr_overcommit_hugepages",
1365 .data = NULL,
1366 .maxlen = sizeof(unsigned long),
1367 .mode = 0644,
1368 .proc_handler = hugetlb_overcommit_handler,
1369 },
1370#endif
1371 {
1372 .procname = "lowmem_reserve_ratio",
1373 .data = &sysctl_lowmem_reserve_ratio,
1374 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1375 .mode = 0644,
1376 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1377 },
1378 {
1379 .procname = "drop_caches",
1380 .data = &sysctl_drop_caches,
1381 .maxlen = sizeof(int),
1382 .mode = 0644,
1383 .proc_handler = drop_caches_sysctl_handler,
1384 .extra1 = &one,
1385 .extra2 = &four,
1386 },
1387#ifdef CONFIG_COMPACTION
1388 {
1389 .procname = "compact_memory",
1390 .data = &sysctl_compact_memory,
1391 .maxlen = sizeof(int),
1392 .mode = 0200,
1393 .proc_handler = sysctl_compaction_handler,
1394 },
1395 {
1396 .procname = "extfrag_threshold",
1397 .data = &sysctl_extfrag_threshold,
1398 .maxlen = sizeof(int),
1399 .mode = 0644,
1400 .proc_handler = sysctl_extfrag_handler,
1401 .extra1 = &min_extfrag_threshold,
1402 .extra2 = &max_extfrag_threshold,
1403 },
1404 {
1405 .procname = "compact_unevictable_allowed",
1406 .data = &sysctl_compact_unevictable_allowed,
1407 .maxlen = sizeof(int),
1408 .mode = 0644,
1409 .proc_handler = proc_dointvec,
1410 .extra1 = &zero,
1411 .extra2 = &one,
1412 },
1413
1414#endif /* CONFIG_COMPACTION */
1415 {
1416 .procname = "min_free_kbytes",
1417 .data = &min_free_kbytes,
1418 .maxlen = sizeof(min_free_kbytes),
1419 .mode = 0644,
1420 .proc_handler = min_free_kbytes_sysctl_handler,
1421 .extra1 = &zero,
1422 },
1423 {
1424 .procname = "watermark_scale_factor",
1425 .data = &watermark_scale_factor,
1426 .maxlen = sizeof(watermark_scale_factor),
1427 .mode = 0644,
1428 .proc_handler = watermark_scale_factor_sysctl_handler,
1429 .extra1 = &one,
1430 .extra2 = &one_thousand,
1431 },
1432 {
1433 .procname = "percpu_pagelist_fraction",
1434 .data = &percpu_pagelist_fraction,
1435 .maxlen = sizeof(percpu_pagelist_fraction),
1436 .mode = 0644,
1437 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1438 .extra1 = &zero,
1439 },
1440#ifdef CONFIG_MMU
1441 {
1442 .procname = "max_map_count",
1443 .data = &sysctl_max_map_count,
1444 .maxlen = sizeof(sysctl_max_map_count),
1445 .mode = 0644,
1446 .proc_handler = proc_dointvec_minmax,
1447 .extra1 = &zero,
1448 },
1449#else
1450 {
1451 .procname = "nr_trim_pages",
1452 .data = &sysctl_nr_trim_pages,
1453 .maxlen = sizeof(sysctl_nr_trim_pages),
1454 .mode = 0644,
1455 .proc_handler = proc_dointvec_minmax,
1456 .extra1 = &zero,
1457 },
1458#endif
1459 {
1460 .procname = "laptop_mode",
1461 .data = &laptop_mode,
1462 .maxlen = sizeof(laptop_mode),
1463 .mode = 0644,
1464 .proc_handler = proc_dointvec_jiffies,
1465 },
1466 {
1467 .procname = "block_dump",
1468 .data = &block_dump,
1469 .maxlen = sizeof(block_dump),
1470 .mode = 0644,
1471 .proc_handler = proc_dointvec,
1472 .extra1 = &zero,
1473 },
1474 {
1475 .procname = "vfs_cache_pressure",
1476 .data = &sysctl_vfs_cache_pressure,
1477 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1478 .mode = 0644,
1479 .proc_handler = proc_dointvec,
1480 .extra1 = &zero,
1481 },
1482#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1483 {
1484 .procname = "legacy_va_layout",
1485 .data = &sysctl_legacy_va_layout,
1486 .maxlen = sizeof(sysctl_legacy_va_layout),
1487 .mode = 0644,
1488 .proc_handler = proc_dointvec,
1489 .extra1 = &zero,
1490 },
1491#endif
1492#ifdef CONFIG_NUMA
1493 {
1494 .procname = "zone_reclaim_mode",
1495 .data = &node_reclaim_mode,
1496 .maxlen = sizeof(node_reclaim_mode),
1497 .mode = 0644,
1498 .proc_handler = proc_dointvec,
1499 .extra1 = &zero,
1500 },
1501 {
1502 .procname = "min_unmapped_ratio",
1503 .data = &sysctl_min_unmapped_ratio,
1504 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1505 .mode = 0644,
1506 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1507 .extra1 = &zero,
1508 .extra2 = &one_hundred,
1509 },
1510 {
1511 .procname = "min_slab_ratio",
1512 .data = &sysctl_min_slab_ratio,
1513 .maxlen = sizeof(sysctl_min_slab_ratio),
1514 .mode = 0644,
1515 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1516 .extra1 = &zero,
1517 .extra2 = &one_hundred,
1518 },
1519#endif
1520#ifdef CONFIG_SMP
1521 {
1522 .procname = "stat_interval",
1523 .data = &sysctl_stat_interval,
1524 .maxlen = sizeof(sysctl_stat_interval),
1525 .mode = 0644,
1526 .proc_handler = proc_dointvec_jiffies,
1527 },
1528 {
1529 .procname = "stat_refresh",
1530 .data = NULL,
1531 .maxlen = 0,
1532 .mode = 0600,
1533 .proc_handler = vmstat_refresh,
1534 },
1535#endif
1536#ifdef CONFIG_MMU
1537 {
1538 .procname = "mmap_min_addr",
1539 .data = &dac_mmap_min_addr,
1540 .maxlen = sizeof(unsigned long),
1541 .mode = 0644,
1542 .proc_handler = mmap_min_addr_handler,
1543 },
1544#endif
1545#ifdef CONFIG_NUMA
1546 {
1547 .procname = "numa_zonelist_order",
1548 .data = &numa_zonelist_order,
1549 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1550 .mode = 0644,
1551 .proc_handler = numa_zonelist_order_handler,
1552 },
1553#endif
1554#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1555 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1556 {
1557 .procname = "vdso_enabled",
1558#ifdef CONFIG_X86_32
1559 .data = &vdso32_enabled,
1560 .maxlen = sizeof(vdso32_enabled),
1561#else
1562 .data = &vdso_enabled,
1563 .maxlen = sizeof(vdso_enabled),
1564#endif
1565 .mode = 0644,
1566 .proc_handler = proc_dointvec,
1567 .extra1 = &zero,
1568 },
1569#endif
1570#ifdef CONFIG_HIGHMEM
1571 {
1572 .procname = "highmem_is_dirtyable",
1573 .data = &vm_highmem_is_dirtyable,
1574 .maxlen = sizeof(vm_highmem_is_dirtyable),
1575 .mode = 0644,
1576 .proc_handler = proc_dointvec_minmax,
1577 .extra1 = &zero,
1578 .extra2 = &one,
1579 },
1580#endif
1581#ifdef CONFIG_MEMORY_FAILURE
1582 {
1583 .procname = "memory_failure_early_kill",
1584 .data = &sysctl_memory_failure_early_kill,
1585 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1586 .mode = 0644,
1587 .proc_handler = proc_dointvec_minmax,
1588 .extra1 = &zero,
1589 .extra2 = &one,
1590 },
1591 {
1592 .procname = "memory_failure_recovery",
1593 .data = &sysctl_memory_failure_recovery,
1594 .maxlen = sizeof(sysctl_memory_failure_recovery),
1595 .mode = 0644,
1596 .proc_handler = proc_dointvec_minmax,
1597 .extra1 = &zero,
1598 .extra2 = &one,
1599 },
1600#endif
1601 {
1602 .procname = "user_reserve_kbytes",
1603 .data = &sysctl_user_reserve_kbytes,
1604 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1605 .mode = 0644,
1606 .proc_handler = proc_doulongvec_minmax,
1607 },
1608 {
1609 .procname = "admin_reserve_kbytes",
1610 .data = &sysctl_admin_reserve_kbytes,
1611 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1612 .mode = 0644,
1613 .proc_handler = proc_doulongvec_minmax,
1614 },
1615#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1616 {
1617 .procname = "mmap_rnd_bits",
1618 .data = &mmap_rnd_bits,
1619 .maxlen = sizeof(mmap_rnd_bits),
1620 .mode = 0600,
1621 .proc_handler = proc_dointvec_minmax,
1622 .extra1 = (void *)&mmap_rnd_bits_min,
1623 .extra2 = (void *)&mmap_rnd_bits_max,
1624 },
1625#endif
1626#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1627 {
1628 .procname = "mmap_rnd_compat_bits",
1629 .data = &mmap_rnd_compat_bits,
1630 .maxlen = sizeof(mmap_rnd_compat_bits),
1631 .mode = 0600,
1632 .proc_handler = proc_dointvec_minmax,
1633 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1634 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1635 },
1636#endif
1637 { }
1638};
1639
1640static struct ctl_table fs_table[] = {
1641 {
1642 .procname = "inode-nr",
1643 .data = &inodes_stat,
1644 .maxlen = 2*sizeof(long),
1645 .mode = 0444,
1646 .proc_handler = proc_nr_inodes,
1647 },
1648 {
1649 .procname = "inode-state",
1650 .data = &inodes_stat,
1651 .maxlen = 7*sizeof(long),
1652 .mode = 0444,
1653 .proc_handler = proc_nr_inodes,
1654 },
1655 {
1656 .procname = "file-nr",
1657 .data = &files_stat,
1658 .maxlen = sizeof(files_stat),
1659 .mode = 0444,
1660 .proc_handler = proc_nr_files,
1661 },
1662 {
1663 .procname = "file-max",
1664 .data = &files_stat.max_files,
1665 .maxlen = sizeof(files_stat.max_files),
1666 .mode = 0644,
1667 .proc_handler = proc_doulongvec_minmax,
1668 },
1669 {
1670 .procname = "nr_open",
1671 .data = &sysctl_nr_open,
1672 .maxlen = sizeof(unsigned int),
1673 .mode = 0644,
1674 .proc_handler = proc_dointvec_minmax,
1675 .extra1 = &sysctl_nr_open_min,
1676 .extra2 = &sysctl_nr_open_max,
1677 },
1678 {
1679 .procname = "dentry-state",
1680 .data = &dentry_stat,
1681 .maxlen = 6*sizeof(long),
1682 .mode = 0444,
1683 .proc_handler = proc_nr_dentry,
1684 },
1685 {
1686 .procname = "overflowuid",
1687 .data = &fs_overflowuid,
1688 .maxlen = sizeof(int),
1689 .mode = 0644,
1690 .proc_handler = proc_dointvec_minmax,
1691 .extra1 = &minolduid,
1692 .extra2 = &maxolduid,
1693 },
1694 {
1695 .procname = "overflowgid",
1696 .data = &fs_overflowgid,
1697 .maxlen = sizeof(int),
1698 .mode = 0644,
1699 .proc_handler = proc_dointvec_minmax,
1700 .extra1 = &minolduid,
1701 .extra2 = &maxolduid,
1702 },
1703#ifdef CONFIG_FILE_LOCKING
1704 {
1705 .procname = "leases-enable",
1706 .data = &leases_enable,
1707 .maxlen = sizeof(int),
1708 .mode = 0644,
1709 .proc_handler = proc_dointvec,
1710 },
1711#endif
1712#ifdef CONFIG_DNOTIFY
1713 {
1714 .procname = "dir-notify-enable",
1715 .data = &dir_notify_enable,
1716 .maxlen = sizeof(int),
1717 .mode = 0644,
1718 .proc_handler = proc_dointvec,
1719 },
1720#endif
1721#ifdef CONFIG_MMU
1722#ifdef CONFIG_FILE_LOCKING
1723 {
1724 .procname = "lease-break-time",
1725 .data = &lease_break_time,
1726 .maxlen = sizeof(int),
1727 .mode = 0644,
1728 .proc_handler = proc_dointvec,
1729 },
1730#endif
1731#ifdef CONFIG_AIO
1732 {
1733 .procname = "aio-nr",
1734 .data = &aio_nr,
1735 .maxlen = sizeof(aio_nr),
1736 .mode = 0444,
1737 .proc_handler = proc_doulongvec_minmax,
1738 },
1739 {
1740 .procname = "aio-max-nr",
1741 .data = &aio_max_nr,
1742 .maxlen = sizeof(aio_max_nr),
1743 .mode = 0644,
1744 .proc_handler = proc_doulongvec_minmax,
1745 },
1746#endif /* CONFIG_AIO */
1747#ifdef CONFIG_INOTIFY_USER
1748 {
1749 .procname = "inotify",
1750 .mode = 0555,
1751 .child = inotify_table,
1752 },
1753#endif
1754#ifdef CONFIG_EPOLL
1755 {
1756 .procname = "epoll",
1757 .mode = 0555,
1758 .child = epoll_table,
1759 },
1760#endif
1761#endif
1762 {
1763 .procname = "protected_symlinks",
1764 .data = &sysctl_protected_symlinks,
1765 .maxlen = sizeof(int),
1766 .mode = 0600,
1767 .proc_handler = proc_dointvec_minmax,
1768 .extra1 = &zero,
1769 .extra2 = &one,
1770 },
1771 {
1772 .procname = "protected_hardlinks",
1773 .data = &sysctl_protected_hardlinks,
1774 .maxlen = sizeof(int),
1775 .mode = 0600,
1776 .proc_handler = proc_dointvec_minmax,
1777 .extra1 = &zero,
1778 .extra2 = &one,
1779 },
1780 {
1781 .procname = "suid_dumpable",
1782 .data = &suid_dumpable,
1783 .maxlen = sizeof(int),
1784 .mode = 0644,
1785 .proc_handler = proc_dointvec_minmax_coredump,
1786 .extra1 = &zero,
1787 .extra2 = &two,
1788 },
1789#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1790 {
1791 .procname = "binfmt_misc",
1792 .mode = 0555,
1793 .child = sysctl_mount_point,
1794 },
1795#endif
1796 {
1797 .procname = "pipe-max-size",
1798 .data = &pipe_max_size,
1799 .maxlen = sizeof(int),
1800 .mode = 0644,
1801 .proc_handler = &pipe_proc_fn,
1802 .extra1 = &pipe_min_size,
1803 },
1804 {
1805 .procname = "pipe-user-pages-hard",
1806 .data = &pipe_user_pages_hard,
1807 .maxlen = sizeof(pipe_user_pages_hard),
1808 .mode = 0644,
1809 .proc_handler = proc_doulongvec_minmax,
1810 },
1811 {
1812 .procname = "pipe-user-pages-soft",
1813 .data = &pipe_user_pages_soft,
1814 .maxlen = sizeof(pipe_user_pages_soft),
1815 .mode = 0644,
1816 .proc_handler = proc_doulongvec_minmax,
1817 },
1818 {
1819 .procname = "mount-max",
1820 .data = &sysctl_mount_max,
1821 .maxlen = sizeof(unsigned int),
1822 .mode = 0644,
1823 .proc_handler = proc_dointvec_minmax,
1824 .extra1 = &one,
1825 },
1826 { }
1827};
1828
1829static struct ctl_table debug_table[] = {
1830#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1831 {
1832 .procname = "exception-trace",
1833 .data = &show_unhandled_signals,
1834 .maxlen = sizeof(int),
1835 .mode = 0644,
1836 .proc_handler = proc_dointvec
1837 },
1838#endif
1839#if defined(CONFIG_OPTPROBES)
1840 {
1841 .procname = "kprobes-optimization",
1842 .data = &sysctl_kprobes_optimization,
1843 .maxlen = sizeof(int),
1844 .mode = 0644,
1845 .proc_handler = proc_kprobes_optimization_handler,
1846 .extra1 = &zero,
1847 .extra2 = &one,
1848 },
1849#endif
1850 { }
1851};
1852
1853static struct ctl_table dev_table[] = {
1854 { }
1855};
1856
1857int __init sysctl_init(void)
1858{
1859 struct ctl_table_header *hdr;
1860
1861 hdr = register_sysctl_table(sysctl_base_table);
1862 kmemleak_not_leak(hdr);
1863 return 0;
1864}
1865
1866#endif /* CONFIG_SYSCTL */
1867
1868/*
1869 * /proc/sys support
1870 */
1871
1872#ifdef CONFIG_PROC_SYSCTL
1873
1874static int _proc_do_string(char *data, int maxlen, int write,
1875 char __user *buffer,
1876 size_t *lenp, loff_t *ppos)
1877{
1878 size_t len;
1879 char __user *p;
1880 char c;
1881
1882 if (!data || !maxlen || !*lenp) {
1883 *lenp = 0;
1884 return 0;
1885 }
1886
1887 if (write) {
1888 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1889 /* Only continue writes not past the end of buffer. */
1890 len = strlen(data);
1891 if (len > maxlen - 1)
1892 len = maxlen - 1;
1893
1894 if (*ppos > len)
1895 return 0;
1896 len = *ppos;
1897 } else {
1898 /* Start writing from beginning of buffer. */
1899 len = 0;
1900 }
1901
1902 *ppos += *lenp;
1903 p = buffer;
1904 while ((p - buffer) < *lenp && len < maxlen - 1) {
1905 if (get_user(c, p++))
1906 return -EFAULT;
1907 if (c == 0 || c == '\n')
1908 break;
1909 data[len++] = c;
1910 }
1911 data[len] = 0;
1912 } else {
1913 len = strlen(data);
1914 if (len > maxlen)
1915 len = maxlen;
1916
1917 if (*ppos > len) {
1918 *lenp = 0;
1919 return 0;
1920 }
1921
1922 data += *ppos;
1923 len -= *ppos;
1924
1925 if (len > *lenp)
1926 len = *lenp;
1927 if (len)
1928 if (copy_to_user(buffer, data, len))
1929 return -EFAULT;
1930 if (len < *lenp) {
1931 if (put_user('\n', buffer + len))
1932 return -EFAULT;
1933 len++;
1934 }
1935 *lenp = len;
1936 *ppos += len;
1937 }
1938 return 0;
1939}
1940
1941static void warn_sysctl_write(struct ctl_table *table)
1942{
1943 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1944 "This will not be supported in the future. To silence this\n"
1945 "warning, set kernel.sysctl_writes_strict = -1\n",
1946 current->comm, table->procname);
1947}
1948
1949/**
1950 * proc_dostring - read a string sysctl
1951 * @table: the sysctl table
1952 * @write: %TRUE if this is a write to the sysctl file
1953 * @buffer: the user buffer
1954 * @lenp: the size of the user buffer
1955 * @ppos: file position
1956 *
1957 * Reads/writes a string from/to the user buffer. If the kernel
1958 * buffer provided is not large enough to hold the string, the
1959 * string is truncated. The copied string is %NULL-terminated.
1960 * If the string is being read by the user process, it is copied
1961 * and a newline '\n' is added. It is truncated if the buffer is
1962 * not large enough.
1963 *
1964 * Returns 0 on success.
1965 */
1966int proc_dostring(struct ctl_table *table, int write,
1967 void __user *buffer, size_t *lenp, loff_t *ppos)
1968{
1969 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1970 warn_sysctl_write(table);
1971
1972 return _proc_do_string((char *)(table->data), table->maxlen, write,
1973 (char __user *)buffer, lenp, ppos);
1974}
1975
1976static size_t proc_skip_spaces(char **buf)
1977{
1978 size_t ret;
1979 char *tmp = skip_spaces(*buf);
1980 ret = tmp - *buf;
1981 *buf = tmp;
1982 return ret;
1983}
1984
1985static void proc_skip_char(char **buf, size_t *size, const char v)
1986{
1987 while (*size) {
1988 if (**buf != v)
1989 break;
1990 (*size)--;
1991 (*buf)++;
1992 }
1993}
1994
1995#define TMPBUFLEN 22
1996/**
1997 * proc_get_long - reads an ASCII formatted integer from a user buffer
1998 *
1999 * @buf: a kernel buffer
2000 * @size: size of the kernel buffer
2001 * @val: this is where the number will be stored
2002 * @neg: set to %TRUE if number is negative
2003 * @perm_tr: a vector which contains the allowed trailers
2004 * @perm_tr_len: size of the perm_tr vector
2005 * @tr: pointer to store the trailer character
2006 *
2007 * In case of success %0 is returned and @buf and @size are updated with
2008 * the amount of bytes read. If @tr is non-NULL and a trailing
2009 * character exists (size is non-zero after returning from this
2010 * function), @tr is updated with the trailing character.
2011 */
2012static int proc_get_long(char **buf, size_t *size,
2013 unsigned long *val, bool *neg,
2014 const char *perm_tr, unsigned perm_tr_len, char *tr)
2015{
2016 int len;
2017 char *p, tmp[TMPBUFLEN];
2018
2019 if (!*size)
2020 return -EINVAL;
2021
2022 len = *size;
2023 if (len > TMPBUFLEN - 1)
2024 len = TMPBUFLEN - 1;
2025
2026 memcpy(tmp, *buf, len);
2027
2028 tmp[len] = 0;
2029 p = tmp;
2030 if (*p == '-' && *size > 1) {
2031 *neg = true;
2032 p++;
2033 } else
2034 *neg = false;
2035 if (!isdigit(*p))
2036 return -EINVAL;
2037
2038 *val = simple_strtoul(p, &p, 0);
2039
2040 len = p - tmp;
2041
2042 /* We don't know if the next char is whitespace thus we may accept
2043 * invalid integers (e.g. 1234...a) or two integers instead of one
2044 * (e.g. 123...1). So lets not allow such large numbers. */
2045 if (len == TMPBUFLEN - 1)
2046 return -EINVAL;
2047
2048 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2049 return -EINVAL;
2050
2051 if (tr && (len < *size))
2052 *tr = *p;
2053
2054 *buf += len;
2055 *size -= len;
2056
2057 return 0;
2058}
2059
2060/**
2061 * proc_put_long - converts an integer to a decimal ASCII formatted string
2062 *
2063 * @buf: the user buffer
2064 * @size: the size of the user buffer
2065 * @val: the integer to be converted
2066 * @neg: sign of the number, %TRUE for negative
2067 *
2068 * In case of success %0 is returned and @buf and @size are updated with
2069 * the amount of bytes written.
2070 */
2071static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2072 bool neg)
2073{
2074 int len;
2075 char tmp[TMPBUFLEN], *p = tmp;
2076
2077 sprintf(p, "%s%lu", neg ? "-" : "", val);
2078 len = strlen(tmp);
2079 if (len > *size)
2080 len = *size;
2081 if (copy_to_user(*buf, tmp, len))
2082 return -EFAULT;
2083 *size -= len;
2084 *buf += len;
2085 return 0;
2086}
2087#undef TMPBUFLEN
2088
2089static int proc_put_char(void __user **buf, size_t *size, char c)
2090{
2091 if (*size) {
2092 char __user **buffer = (char __user **)buf;
2093 if (put_user(c, *buffer))
2094 return -EFAULT;
2095 (*size)--, (*buffer)++;
2096 *buf = *buffer;
2097 }
2098 return 0;
2099}
2100
2101static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2102 int *valp,
2103 int write, void *data)
2104{
2105 if (write) {
2106 if (*negp) {
2107 if (*lvalp > (unsigned long) INT_MAX + 1)
2108 return -EINVAL;
2109 *valp = -*lvalp;
2110 } else {
2111 if (*lvalp > (unsigned long) INT_MAX)
2112 return -EINVAL;
2113 *valp = *lvalp;
2114 }
2115 } else {
2116 int val = *valp;
2117 if (val < 0) {
2118 *negp = true;
2119 *lvalp = -(unsigned long)val;
2120 } else {
2121 *negp = false;
2122 *lvalp = (unsigned long)val;
2123 }
2124 }
2125 return 0;
2126}
2127
2128static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2129 int *valp,
2130 int write, void *data)
2131{
2132 if (write) {
2133 if (*negp)
2134 return -EINVAL;
2135 *valp = *lvalp;
2136 } else {
2137 unsigned int val = *valp;
2138 *lvalp = (unsigned long)val;
2139 }
2140 return 0;
2141}
2142
2143static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2144
2145static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2146 int write, void __user *buffer,
2147 size_t *lenp, loff_t *ppos,
2148 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2149 int write, void *data),
2150 void *data)
2151{
2152 int *i, vleft, first = 1, err = 0;
2153 size_t left;
2154 char *kbuf = NULL, *p;
2155
2156 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2157 *lenp = 0;
2158 return 0;
2159 }
2160
2161 i = (int *) tbl_data;
2162 vleft = table->maxlen / sizeof(*i);
2163 left = *lenp;
2164
2165 if (!conv)
2166 conv = do_proc_dointvec_conv;
2167
2168 if (write) {
2169 if (*ppos) {
2170 switch (sysctl_writes_strict) {
2171 case SYSCTL_WRITES_STRICT:
2172 goto out;
2173 case SYSCTL_WRITES_WARN:
2174 warn_sysctl_write(table);
2175 break;
2176 default:
2177 break;
2178 }
2179 }
2180
2181 if (left > PAGE_SIZE - 1)
2182 left = PAGE_SIZE - 1;
2183 p = kbuf = memdup_user_nul(buffer, left);
2184 if (IS_ERR(kbuf))
2185 return PTR_ERR(kbuf);
2186 }
2187
2188 for (; left && vleft--; i++, first=0) {
2189 unsigned long lval;
2190 bool neg;
2191
2192 if (write) {
2193 left -= proc_skip_spaces(&p);
2194
2195 if (!left)
2196 break;
2197 err = proc_get_long(&p, &left, &lval, &neg,
2198 proc_wspace_sep,
2199 sizeof(proc_wspace_sep), NULL);
2200 if (err)
2201 break;
2202 if (conv(&neg, &lval, i, 1, data)) {
2203 err = -EINVAL;
2204 break;
2205 }
2206 } else {
2207 if (conv(&neg, &lval, i, 0, data)) {
2208 err = -EINVAL;
2209 break;
2210 }
2211 if (!first)
2212 err = proc_put_char(&buffer, &left, '\t');
2213 if (err)
2214 break;
2215 err = proc_put_long(&buffer, &left, lval, neg);
2216 if (err)
2217 break;
2218 }
2219 }
2220
2221 if (!write && !first && left && !err)
2222 err = proc_put_char(&buffer, &left, '\n');
2223 if (write && !err && left)
2224 left -= proc_skip_spaces(&p);
2225 if (write) {
2226 kfree(kbuf);
2227 if (first)
2228 return err ? : -EINVAL;
2229 }
2230 *lenp -= left;
2231out:
2232 *ppos += *lenp;
2233 return err;
2234}
2235
2236static int do_proc_dointvec(struct ctl_table *table, int write,
2237 void __user *buffer, size_t *lenp, loff_t *ppos,
2238 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2239 int write, void *data),
2240 void *data)
2241{
2242 return __do_proc_dointvec(table->data, table, write,
2243 buffer, lenp, ppos, conv, data);
2244}
2245
2246/**
2247 * proc_dointvec - read a vector of integers
2248 * @table: the sysctl table
2249 * @write: %TRUE if this is a write to the sysctl file
2250 * @buffer: the user buffer
2251 * @lenp: the size of the user buffer
2252 * @ppos: file position
2253 *
2254 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2255 * values from/to the user buffer, treated as an ASCII string.
2256 *
2257 * Returns 0 on success.
2258 */
2259int proc_dointvec(struct ctl_table *table, int write,
2260 void __user *buffer, size_t *lenp, loff_t *ppos)
2261{
2262 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2263}
2264
2265/**
2266 * proc_douintvec - read a vector of unsigned integers
2267 * @table: the sysctl table
2268 * @write: %TRUE if this is a write to the sysctl file
2269 * @buffer: the user buffer
2270 * @lenp: the size of the user buffer
2271 * @ppos: file position
2272 *
2273 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2274 * values from/to the user buffer, treated as an ASCII string.
2275 *
2276 * Returns 0 on success.
2277 */
2278int proc_douintvec(struct ctl_table *table, int write,
2279 void __user *buffer, size_t *lenp, loff_t *ppos)
2280{
2281 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2282 do_proc_douintvec_conv, NULL);
2283}
2284
2285/*
2286 * Taint values can only be increased
2287 * This means we can safely use a temporary.
2288 */
2289static int proc_taint(struct ctl_table *table, int write,
2290 void __user *buffer, size_t *lenp, loff_t *ppos)
2291{
2292 struct ctl_table t;
2293 unsigned long tmptaint = get_taint();
2294 int err;
2295
2296 if (write && !capable(CAP_SYS_ADMIN))
2297 return -EPERM;
2298
2299 t = *table;
2300 t.data = &tmptaint;
2301 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2302 if (err < 0)
2303 return err;
2304
2305 if (write) {
2306 /*
2307 * Poor man's atomic or. Not worth adding a primitive
2308 * to everyone's atomic.h for this
2309 */
2310 int i;
2311 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2312 if ((tmptaint >> i) & 1)
2313 add_taint(i, LOCKDEP_STILL_OK);
2314 }
2315 }
2316
2317 return err;
2318}
2319
2320#ifdef CONFIG_PRINTK
2321static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2322 void __user *buffer, size_t *lenp, loff_t *ppos)
2323{
2324 if (write && !capable(CAP_SYS_ADMIN))
2325 return -EPERM;
2326
2327 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2328}
2329#endif
2330
2331struct do_proc_dointvec_minmax_conv_param {
2332 int *min;
2333 int *max;
2334};
2335
2336static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2337 int *valp,
2338 int write, void *data)
2339{
2340 struct do_proc_dointvec_minmax_conv_param *param = data;
2341 if (write) {
2342 int val = *negp ? -*lvalp : *lvalp;
2343 if ((param->min && *param->min > val) ||
2344 (param->max && *param->max < val))
2345 return -EINVAL;
2346 *valp = val;
2347 } else {
2348 int val = *valp;
2349 if (val < 0) {
2350 *negp = true;
2351 *lvalp = -(unsigned long)val;
2352 } else {
2353 *negp = false;
2354 *lvalp = (unsigned long)val;
2355 }
2356 }
2357 return 0;
2358}
2359
2360/**
2361 * proc_dointvec_minmax - read a vector of integers with min/max values
2362 * @table: the sysctl table
2363 * @write: %TRUE if this is a write to the sysctl file
2364 * @buffer: the user buffer
2365 * @lenp: the size of the user buffer
2366 * @ppos: file position
2367 *
2368 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2369 * values from/to the user buffer, treated as an ASCII string.
2370 *
2371 * This routine will ensure the values are within the range specified by
2372 * table->extra1 (min) and table->extra2 (max).
2373 *
2374 * Returns 0 on success.
2375 */
2376int proc_dointvec_minmax(struct ctl_table *table, int write,
2377 void __user *buffer, size_t *lenp, loff_t *ppos)
2378{
2379 struct do_proc_dointvec_minmax_conv_param param = {
2380 .min = (int *) table->extra1,
2381 .max = (int *) table->extra2,
2382 };
2383 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2384 do_proc_dointvec_minmax_conv, ¶m);
2385}
2386
2387static void validate_coredump_safety(void)
2388{
2389#ifdef CONFIG_COREDUMP
2390 if (suid_dumpable == SUID_DUMP_ROOT &&
2391 core_pattern[0] != '/' && core_pattern[0] != '|') {
2392 printk(KERN_WARNING
2393"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2394"Pipe handler or fully qualified core dump path required.\n"
2395"Set kernel.core_pattern before fs.suid_dumpable.\n"
2396 );
2397 }
2398#endif
2399}
2400
2401static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2402 void __user *buffer, size_t *lenp, loff_t *ppos)
2403{
2404 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2405 if (!error)
2406 validate_coredump_safety();
2407 return error;
2408}
2409
2410#ifdef CONFIG_COREDUMP
2411static int proc_dostring_coredump(struct ctl_table *table, int write,
2412 void __user *buffer, size_t *lenp, loff_t *ppos)
2413{
2414 int error = proc_dostring(table, write, buffer, lenp, ppos);
2415 if (!error)
2416 validate_coredump_safety();
2417 return error;
2418}
2419#endif
2420
2421static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2422 void __user *buffer,
2423 size_t *lenp, loff_t *ppos,
2424 unsigned long convmul,
2425 unsigned long convdiv)
2426{
2427 unsigned long *i, *min, *max;
2428 int vleft, first = 1, err = 0;
2429 size_t left;
2430 char *kbuf = NULL, *p;
2431
2432 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2433 *lenp = 0;
2434 return 0;
2435 }
2436
2437 i = (unsigned long *) data;
2438 min = (unsigned long *) table->extra1;
2439 max = (unsigned long *) table->extra2;
2440 vleft = table->maxlen / sizeof(unsigned long);
2441 left = *lenp;
2442
2443 if (write) {
2444 if (*ppos) {
2445 switch (sysctl_writes_strict) {
2446 case SYSCTL_WRITES_STRICT:
2447 goto out;
2448 case SYSCTL_WRITES_WARN:
2449 warn_sysctl_write(table);
2450 break;
2451 default:
2452 break;
2453 }
2454 }
2455
2456 if (left > PAGE_SIZE - 1)
2457 left = PAGE_SIZE - 1;
2458 p = kbuf = memdup_user_nul(buffer, left);
2459 if (IS_ERR(kbuf))
2460 return PTR_ERR(kbuf);
2461 }
2462
2463 for (; left && vleft--; i++, first = 0) {
2464 unsigned long val;
2465
2466 if (write) {
2467 bool neg;
2468
2469 left -= proc_skip_spaces(&p);
2470
2471 err = proc_get_long(&p, &left, &val, &neg,
2472 proc_wspace_sep,
2473 sizeof(proc_wspace_sep), NULL);
2474 if (err)
2475 break;
2476 if (neg)
2477 continue;
2478 val = convmul * val / convdiv;
2479 if ((min && val < *min) || (max && val > *max))
2480 continue;
2481 *i = val;
2482 } else {
2483 val = convdiv * (*i) / convmul;
2484 if (!first) {
2485 err = proc_put_char(&buffer, &left, '\t');
2486 if (err)
2487 break;
2488 }
2489 err = proc_put_long(&buffer, &left, val, false);
2490 if (err)
2491 break;
2492 }
2493 }
2494
2495 if (!write && !first && left && !err)
2496 err = proc_put_char(&buffer, &left, '\n');
2497 if (write && !err)
2498 left -= proc_skip_spaces(&p);
2499 if (write) {
2500 kfree(kbuf);
2501 if (first)
2502 return err ? : -EINVAL;
2503 }
2504 *lenp -= left;
2505out:
2506 *ppos += *lenp;
2507 return err;
2508}
2509
2510static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2511 void __user *buffer,
2512 size_t *lenp, loff_t *ppos,
2513 unsigned long convmul,
2514 unsigned long convdiv)
2515{
2516 return __do_proc_doulongvec_minmax(table->data, table, write,
2517 buffer, lenp, ppos, convmul, convdiv);
2518}
2519
2520/**
2521 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2522 * @table: the sysctl table
2523 * @write: %TRUE if this is a write to the sysctl file
2524 * @buffer: the user buffer
2525 * @lenp: the size of the user buffer
2526 * @ppos: file position
2527 *
2528 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2529 * values from/to the user buffer, treated as an ASCII string.
2530 *
2531 * This routine will ensure the values are within the range specified by
2532 * table->extra1 (min) and table->extra2 (max).
2533 *
2534 * Returns 0 on success.
2535 */
2536int proc_doulongvec_minmax(struct ctl_table *table, int write,
2537 void __user *buffer, size_t *lenp, loff_t *ppos)
2538{
2539 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2540}
2541
2542/**
2543 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2544 * @table: the sysctl table
2545 * @write: %TRUE if this is a write to the sysctl file
2546 * @buffer: the user buffer
2547 * @lenp: the size of the user buffer
2548 * @ppos: file position
2549 *
2550 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551 * values from/to the user buffer, treated as an ASCII string. The values
2552 * are treated as milliseconds, and converted to jiffies when they are stored.
2553 *
2554 * This routine will ensure the values are within the range specified by
2555 * table->extra1 (min) and table->extra2 (max).
2556 *
2557 * Returns 0 on success.
2558 */
2559int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2560 void __user *buffer,
2561 size_t *lenp, loff_t *ppos)
2562{
2563 return do_proc_doulongvec_minmax(table, write, buffer,
2564 lenp, ppos, HZ, 1000l);
2565}
2566
2567
2568static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2569 int *valp,
2570 int write, void *data)
2571{
2572 if (write) {
2573 if (*lvalp > LONG_MAX / HZ)
2574 return 1;
2575 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2576 } else {
2577 int val = *valp;
2578 unsigned long lval;
2579 if (val < 0) {
2580 *negp = true;
2581 lval = -(unsigned long)val;
2582 } else {
2583 *negp = false;
2584 lval = (unsigned long)val;
2585 }
2586 *lvalp = lval / HZ;
2587 }
2588 return 0;
2589}
2590
2591static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2592 int *valp,
2593 int write, void *data)
2594{
2595 if (write) {
2596 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2597 return 1;
2598 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2599 } else {
2600 int val = *valp;
2601 unsigned long lval;
2602 if (val < 0) {
2603 *negp = true;
2604 lval = -(unsigned long)val;
2605 } else {
2606 *negp = false;
2607 lval = (unsigned long)val;
2608 }
2609 *lvalp = jiffies_to_clock_t(lval);
2610 }
2611 return 0;
2612}
2613
2614static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2615 int *valp,
2616 int write, void *data)
2617{
2618 if (write) {
2619 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2620
2621 if (jif > INT_MAX)
2622 return 1;
2623 *valp = (int)jif;
2624 } else {
2625 int val = *valp;
2626 unsigned long lval;
2627 if (val < 0) {
2628 *negp = true;
2629 lval = -(unsigned long)val;
2630 } else {
2631 *negp = false;
2632 lval = (unsigned long)val;
2633 }
2634 *lvalp = jiffies_to_msecs(lval);
2635 }
2636 return 0;
2637}
2638
2639/**
2640 * proc_dointvec_jiffies - read a vector of integers as seconds
2641 * @table: the sysctl table
2642 * @write: %TRUE if this is a write to the sysctl file
2643 * @buffer: the user buffer
2644 * @lenp: the size of the user buffer
2645 * @ppos: file position
2646 *
2647 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2648 * values from/to the user buffer, treated as an ASCII string.
2649 * The values read are assumed to be in seconds, and are converted into
2650 * jiffies.
2651 *
2652 * Returns 0 on success.
2653 */
2654int proc_dointvec_jiffies(struct ctl_table *table, int write,
2655 void __user *buffer, size_t *lenp, loff_t *ppos)
2656{
2657 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2658 do_proc_dointvec_jiffies_conv,NULL);
2659}
2660
2661/**
2662 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2663 * @table: the sysctl table
2664 * @write: %TRUE if this is a write to the sysctl file
2665 * @buffer: the user buffer
2666 * @lenp: the size of the user buffer
2667 * @ppos: pointer to the file position
2668 *
2669 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2670 * values from/to the user buffer, treated as an ASCII string.
2671 * The values read are assumed to be in 1/USER_HZ seconds, and
2672 * are converted into jiffies.
2673 *
2674 * Returns 0 on success.
2675 */
2676int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2677 void __user *buffer, size_t *lenp, loff_t *ppos)
2678{
2679 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2680 do_proc_dointvec_userhz_jiffies_conv,NULL);
2681}
2682
2683/**
2684 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2685 * @table: the sysctl table
2686 * @write: %TRUE if this is a write to the sysctl file
2687 * @buffer: the user buffer
2688 * @lenp: the size of the user buffer
2689 * @ppos: file position
2690 * @ppos: the current position in the file
2691 *
2692 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693 * values from/to the user buffer, treated as an ASCII string.
2694 * The values read are assumed to be in 1/1000 seconds, and
2695 * are converted into jiffies.
2696 *
2697 * Returns 0 on success.
2698 */
2699int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2700 void __user *buffer, size_t *lenp, loff_t *ppos)
2701{
2702 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2703 do_proc_dointvec_ms_jiffies_conv, NULL);
2704}
2705
2706static int proc_do_cad_pid(struct ctl_table *table, int write,
2707 void __user *buffer, size_t *lenp, loff_t *ppos)
2708{
2709 struct pid *new_pid;
2710 pid_t tmp;
2711 int r;
2712
2713 tmp = pid_vnr(cad_pid);
2714
2715 r = __do_proc_dointvec(&tmp, table, write, buffer,
2716 lenp, ppos, NULL, NULL);
2717 if (r || !write)
2718 return r;
2719
2720 new_pid = find_get_pid(tmp);
2721 if (!new_pid)
2722 return -ESRCH;
2723
2724 put_pid(xchg(&cad_pid, new_pid));
2725 return 0;
2726}
2727
2728/**
2729 * proc_do_large_bitmap - read/write from/to a large bitmap
2730 * @table: the sysctl table
2731 * @write: %TRUE if this is a write to the sysctl file
2732 * @buffer: the user buffer
2733 * @lenp: the size of the user buffer
2734 * @ppos: file position
2735 *
2736 * The bitmap is stored at table->data and the bitmap length (in bits)
2737 * in table->maxlen.
2738 *
2739 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2740 * large bitmaps may be represented in a compact manner. Writing into
2741 * the file will clear the bitmap then update it with the given input.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_do_large_bitmap(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 int err = 0;
2749 bool first = 1;
2750 size_t left = *lenp;
2751 unsigned long bitmap_len = table->maxlen;
2752 unsigned long *bitmap = *(unsigned long **) table->data;
2753 unsigned long *tmp_bitmap = NULL;
2754 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2755
2756 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2757 *lenp = 0;
2758 return 0;
2759 }
2760
2761 if (write) {
2762 char *kbuf, *p;
2763
2764 if (left > PAGE_SIZE - 1)
2765 left = PAGE_SIZE - 1;
2766
2767 p = kbuf = memdup_user_nul(buffer, left);
2768 if (IS_ERR(kbuf))
2769 return PTR_ERR(kbuf);
2770
2771 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2772 GFP_KERNEL);
2773 if (!tmp_bitmap) {
2774 kfree(kbuf);
2775 return -ENOMEM;
2776 }
2777 proc_skip_char(&p, &left, '\n');
2778 while (!err && left) {
2779 unsigned long val_a, val_b;
2780 bool neg;
2781
2782 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2783 sizeof(tr_a), &c);
2784 if (err)
2785 break;
2786 if (val_a >= bitmap_len || neg) {
2787 err = -EINVAL;
2788 break;
2789 }
2790
2791 val_b = val_a;
2792 if (left) {
2793 p++;
2794 left--;
2795 }
2796
2797 if (c == '-') {
2798 err = proc_get_long(&p, &left, &val_b,
2799 &neg, tr_b, sizeof(tr_b),
2800 &c);
2801 if (err)
2802 break;
2803 if (val_b >= bitmap_len || neg ||
2804 val_a > val_b) {
2805 err = -EINVAL;
2806 break;
2807 }
2808 if (left) {
2809 p++;
2810 left--;
2811 }
2812 }
2813
2814 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2815 first = 0;
2816 proc_skip_char(&p, &left, '\n');
2817 }
2818 kfree(kbuf);
2819 } else {
2820 unsigned long bit_a, bit_b = 0;
2821
2822 while (left) {
2823 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2824 if (bit_a >= bitmap_len)
2825 break;
2826 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2827 bit_a + 1) - 1;
2828
2829 if (!first) {
2830 err = proc_put_char(&buffer, &left, ',');
2831 if (err)
2832 break;
2833 }
2834 err = proc_put_long(&buffer, &left, bit_a, false);
2835 if (err)
2836 break;
2837 if (bit_a != bit_b) {
2838 err = proc_put_char(&buffer, &left, '-');
2839 if (err)
2840 break;
2841 err = proc_put_long(&buffer, &left, bit_b, false);
2842 if (err)
2843 break;
2844 }
2845
2846 first = 0; bit_b++;
2847 }
2848 if (!err)
2849 err = proc_put_char(&buffer, &left, '\n');
2850 }
2851
2852 if (!err) {
2853 if (write) {
2854 if (*ppos)
2855 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2856 else
2857 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2858 }
2859 kfree(tmp_bitmap);
2860 *lenp -= left;
2861 *ppos += *lenp;
2862 return 0;
2863 } else {
2864 kfree(tmp_bitmap);
2865 return err;
2866 }
2867}
2868
2869#else /* CONFIG_PROC_SYSCTL */
2870
2871int proc_dostring(struct ctl_table *table, int write,
2872 void __user *buffer, size_t *lenp, loff_t *ppos)
2873{
2874 return -ENOSYS;
2875}
2876
2877int proc_dointvec(struct ctl_table *table, int write,
2878 void __user *buffer, size_t *lenp, loff_t *ppos)
2879{
2880 return -ENOSYS;
2881}
2882
2883int proc_douintvec(struct ctl_table *table, int write,
2884 void __user *buffer, size_t *lenp, loff_t *ppos)
2885{
2886 return -ENOSYS;
2887}
2888
2889int proc_dointvec_minmax(struct ctl_table *table, int write,
2890 void __user *buffer, size_t *lenp, loff_t *ppos)
2891{
2892 return -ENOSYS;
2893}
2894
2895int proc_dointvec_jiffies(struct ctl_table *table, int write,
2896 void __user *buffer, size_t *lenp, loff_t *ppos)
2897{
2898 return -ENOSYS;
2899}
2900
2901int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2902 void __user *buffer, size_t *lenp, loff_t *ppos)
2903{
2904 return -ENOSYS;
2905}
2906
2907int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2908 void __user *buffer, size_t *lenp, loff_t *ppos)
2909{
2910 return -ENOSYS;
2911}
2912
2913int proc_doulongvec_minmax(struct ctl_table *table, int write,
2914 void __user *buffer, size_t *lenp, loff_t *ppos)
2915{
2916 return -ENOSYS;
2917}
2918
2919int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2920 void __user *buffer,
2921 size_t *lenp, loff_t *ppos)
2922{
2923 return -ENOSYS;
2924}
2925
2926
2927#endif /* CONFIG_PROC_SYSCTL */
2928
2929/*
2930 * No sense putting this after each symbol definition, twice,
2931 * exception granted :-)
2932 */
2933EXPORT_SYMBOL(proc_dointvec);
2934EXPORT_SYMBOL(proc_douintvec);
2935EXPORT_SYMBOL(proc_dointvec_jiffies);
2936EXPORT_SYMBOL(proc_dointvec_minmax);
2937EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2938EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2939EXPORT_SYMBOL(proc_dostring);
2940EXPORT_SYMBOL(proc_doulongvec_minmax);
2941EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * sysctl.c: General linux system control interface
4 *
5 * Begun 24 March 1995, Stephen Tweedie
6 * Added /proc support, Dec 1995
7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10 * Dynamic registration fixes, Stephen Tweedie.
11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13 * Horn.
14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17 * Wendling.
18 * The list_for_each() macro wasn't appropriate for the sysctl loop.
19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
20 */
21
22#include <linux/module.h>
23#include <linux/aio.h>
24#include <linux/mm.h>
25#include <linux/swap.h>
26#include <linux/slab.h>
27#include <linux/sysctl.h>
28#include <linux/bitmap.h>
29#include <linux/signal.h>
30#include <linux/printk.h>
31#include <linux/proc_fs.h>
32#include <linux/security.h>
33#include <linux/ctype.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/sched/coredump.h>
67#include <linux/kexec.h>
68#include <linux/bpf.h>
69#include <linux/mount.h>
70#include <linux/userfaultfd_k.h>
71#include <linux/coredump.h>
72#include <linux/latencytop.h>
73#include <linux/pid.h>
74
75#include "../lib/kstrtox.h"
76
77#include <linux/uaccess.h>
78#include <asm/processor.h>
79
80#ifdef CONFIG_X86
81#include <asm/nmi.h>
82#include <asm/stacktrace.h>
83#include <asm/io.h>
84#endif
85#ifdef CONFIG_SPARC
86#include <asm/setup.h>
87#endif
88#ifdef CONFIG_BSD_PROCESS_ACCT
89#include <linux/acct.h>
90#endif
91#ifdef CONFIG_RT_MUTEXES
92#include <linux/rtmutex.h>
93#endif
94#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
95#include <linux/lockdep.h>
96#endif
97#ifdef CONFIG_CHR_DEV_SG
98#include <scsi/sg.h>
99#endif
100#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
101#include <linux/stackleak.h>
102#endif
103#ifdef CONFIG_LOCKUP_DETECTOR
104#include <linux/nmi.h>
105#endif
106
107#if defined(CONFIG_SYSCTL)
108
109/* Constants used for minimum and maximum */
110#ifdef CONFIG_LOCKUP_DETECTOR
111static int sixty = 60;
112#endif
113
114static int __maybe_unused neg_one = -1;
115static int __maybe_unused two = 2;
116static int __maybe_unused four = 4;
117static unsigned long zero_ul;
118static unsigned long one_ul = 1;
119static unsigned long long_max = LONG_MAX;
120static int one_hundred = 100;
121static int two_hundred = 200;
122static int one_thousand = 1000;
123#ifdef CONFIG_PRINTK
124static int ten_thousand = 10000;
125#endif
126#ifdef CONFIG_PERF_EVENTS
127static int six_hundred_forty_kb = 640 * 1024;
128#endif
129
130/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
131static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
132
133/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
134static int maxolduid = 65535;
135static int minolduid;
136
137static int ngroups_max = NGROUPS_MAX;
138static const int cap_last_cap = CAP_LAST_CAP;
139
140/*
141 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
142 * and hung_task_check_interval_secs
143 */
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
152#ifdef CONFIG_PROC_SYSCTL
153
154/**
155 * enum sysctl_writes_mode - supported sysctl write modes
156 *
157 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
158 * to be written, and multiple writes on the same sysctl file descriptor
159 * will rewrite the sysctl value, regardless of file position. No warning
160 * is issued when the initial position is not 0.
161 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
162 * not 0.
163 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
164 * file position 0 and the value must be fully contained in the buffer
165 * sent to the write syscall. If dealing with strings respect the file
166 * position, but restrict this to the max length of the buffer, anything
167 * passed the max length will be ignored. Multiple writes will append
168 * to the buffer.
169 *
170 * These write modes control how current file position affects the behavior of
171 * updating sysctl values through the proc interface on each write.
172 */
173enum sysctl_writes_mode {
174 SYSCTL_WRITES_LEGACY = -1,
175 SYSCTL_WRITES_WARN = 0,
176 SYSCTL_WRITES_STRICT = 1,
177};
178
179static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
180#endif /* CONFIG_PROC_SYSCTL */
181
182#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
183 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
184int sysctl_legacy_va_layout;
185#endif
186
187#ifdef CONFIG_SCHED_DEBUG
188static int min_sched_granularity_ns = 100000; /* 100 usecs */
189static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
190static int min_wakeup_granularity_ns; /* 0 usecs */
191static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
192#ifdef CONFIG_SMP
193static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
194static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
195#endif /* CONFIG_SMP */
196#endif /* CONFIG_SCHED_DEBUG */
197
198#ifdef CONFIG_COMPACTION
199static int min_extfrag_threshold;
200static int max_extfrag_threshold = 1000;
201#endif
202
203#endif /* CONFIG_SYSCTL */
204
205#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
206static int bpf_stats_handler(struct ctl_table *table, int write,
207 void *buffer, size_t *lenp, loff_t *ppos)
208{
209 struct static_key *key = (struct static_key *)table->data;
210 static int saved_val;
211 int val, ret;
212 struct ctl_table tmp = {
213 .data = &val,
214 .maxlen = sizeof(val),
215 .mode = table->mode,
216 .extra1 = SYSCTL_ZERO,
217 .extra2 = SYSCTL_ONE,
218 };
219
220 if (write && !capable(CAP_SYS_ADMIN))
221 return -EPERM;
222
223 mutex_lock(&bpf_stats_enabled_mutex);
224 val = saved_val;
225 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
226 if (write && !ret && val != saved_val) {
227 if (val)
228 static_key_slow_inc(key);
229 else
230 static_key_slow_dec(key);
231 saved_val = val;
232 }
233 mutex_unlock(&bpf_stats_enabled_mutex);
234 return ret;
235}
236#endif
237
238/*
239 * /proc/sys support
240 */
241
242#ifdef CONFIG_PROC_SYSCTL
243
244static int _proc_do_string(char *data, int maxlen, int write,
245 char *buffer, size_t *lenp, loff_t *ppos)
246{
247 size_t len;
248 char c, *p;
249
250 if (!data || !maxlen || !*lenp) {
251 *lenp = 0;
252 return 0;
253 }
254
255 if (write) {
256 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
257 /* Only continue writes not past the end of buffer. */
258 len = strlen(data);
259 if (len > maxlen - 1)
260 len = maxlen - 1;
261
262 if (*ppos > len)
263 return 0;
264 len = *ppos;
265 } else {
266 /* Start writing from beginning of buffer. */
267 len = 0;
268 }
269
270 *ppos += *lenp;
271 p = buffer;
272 while ((p - buffer) < *lenp && len < maxlen - 1) {
273 c = *(p++);
274 if (c == 0 || c == '\n')
275 break;
276 data[len++] = c;
277 }
278 data[len] = 0;
279 } else {
280 len = strlen(data);
281 if (len > maxlen)
282 len = maxlen;
283
284 if (*ppos > len) {
285 *lenp = 0;
286 return 0;
287 }
288
289 data += *ppos;
290 len -= *ppos;
291
292 if (len > *lenp)
293 len = *lenp;
294 if (len)
295 memcpy(buffer, data, len);
296 if (len < *lenp) {
297 buffer[len] = '\n';
298 len++;
299 }
300 *lenp = len;
301 *ppos += len;
302 }
303 return 0;
304}
305
306static void warn_sysctl_write(struct ctl_table *table)
307{
308 pr_warn_once("%s wrote to %s when file position was not 0!\n"
309 "This will not be supported in the future. To silence this\n"
310 "warning, set kernel.sysctl_writes_strict = -1\n",
311 current->comm, table->procname);
312}
313
314/**
315 * proc_first_pos_non_zero_ignore - check if first position is allowed
316 * @ppos: file position
317 * @table: the sysctl table
318 *
319 * Returns true if the first position is non-zero and the sysctl_writes_strict
320 * mode indicates this is not allowed for numeric input types. String proc
321 * handlers can ignore the return value.
322 */
323static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
324 struct ctl_table *table)
325{
326 if (!*ppos)
327 return false;
328
329 switch (sysctl_writes_strict) {
330 case SYSCTL_WRITES_STRICT:
331 return true;
332 case SYSCTL_WRITES_WARN:
333 warn_sysctl_write(table);
334 return false;
335 default:
336 return false;
337 }
338}
339
340/**
341 * proc_dostring - read a string sysctl
342 * @table: the sysctl table
343 * @write: %TRUE if this is a write to the sysctl file
344 * @buffer: the user buffer
345 * @lenp: the size of the user buffer
346 * @ppos: file position
347 *
348 * Reads/writes a string from/to the user buffer. If the kernel
349 * buffer provided is not large enough to hold the string, the
350 * string is truncated. The copied string is %NULL-terminated.
351 * If the string is being read by the user process, it is copied
352 * and a newline '\n' is added. It is truncated if the buffer is
353 * not large enough.
354 *
355 * Returns 0 on success.
356 */
357int proc_dostring(struct ctl_table *table, int write,
358 void *buffer, size_t *lenp, loff_t *ppos)
359{
360 if (write)
361 proc_first_pos_non_zero_ignore(ppos, table);
362
363 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
364 ppos);
365}
366
367static size_t proc_skip_spaces(char **buf)
368{
369 size_t ret;
370 char *tmp = skip_spaces(*buf);
371 ret = tmp - *buf;
372 *buf = tmp;
373 return ret;
374}
375
376static void proc_skip_char(char **buf, size_t *size, const char v)
377{
378 while (*size) {
379 if (**buf != v)
380 break;
381 (*size)--;
382 (*buf)++;
383 }
384}
385
386/**
387 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
388 * fail on overflow
389 *
390 * @cp: kernel buffer containing the string to parse
391 * @endp: pointer to store the trailing characters
392 * @base: the base to use
393 * @res: where the parsed integer will be stored
394 *
395 * In case of success 0 is returned and @res will contain the parsed integer,
396 * @endp will hold any trailing characters.
397 * This function will fail the parse on overflow. If there wasn't an overflow
398 * the function will defer the decision what characters count as invalid to the
399 * caller.
400 */
401static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
402 unsigned long *res)
403{
404 unsigned long long result;
405 unsigned int rv;
406
407 cp = _parse_integer_fixup_radix(cp, &base);
408 rv = _parse_integer(cp, base, &result);
409 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
410 return -ERANGE;
411
412 cp += rv;
413
414 if (endp)
415 *endp = (char *)cp;
416
417 *res = (unsigned long)result;
418 return 0;
419}
420
421#define TMPBUFLEN 22
422/**
423 * proc_get_long - reads an ASCII formatted integer from a user buffer
424 *
425 * @buf: a kernel buffer
426 * @size: size of the kernel buffer
427 * @val: this is where the number will be stored
428 * @neg: set to %TRUE if number is negative
429 * @perm_tr: a vector which contains the allowed trailers
430 * @perm_tr_len: size of the perm_tr vector
431 * @tr: pointer to store the trailer character
432 *
433 * In case of success %0 is returned and @buf and @size are updated with
434 * the amount of bytes read. If @tr is non-NULL and a trailing
435 * character exists (size is non-zero after returning from this
436 * function), @tr is updated with the trailing character.
437 */
438static int proc_get_long(char **buf, size_t *size,
439 unsigned long *val, bool *neg,
440 const char *perm_tr, unsigned perm_tr_len, char *tr)
441{
442 int len;
443 char *p, tmp[TMPBUFLEN];
444
445 if (!*size)
446 return -EINVAL;
447
448 len = *size;
449 if (len > TMPBUFLEN - 1)
450 len = TMPBUFLEN - 1;
451
452 memcpy(tmp, *buf, len);
453
454 tmp[len] = 0;
455 p = tmp;
456 if (*p == '-' && *size > 1) {
457 *neg = true;
458 p++;
459 } else
460 *neg = false;
461 if (!isdigit(*p))
462 return -EINVAL;
463
464 if (strtoul_lenient(p, &p, 0, val))
465 return -EINVAL;
466
467 len = p - tmp;
468
469 /* We don't know if the next char is whitespace thus we may accept
470 * invalid integers (e.g. 1234...a) or two integers instead of one
471 * (e.g. 123...1). So lets not allow such large numbers. */
472 if (len == TMPBUFLEN - 1)
473 return -EINVAL;
474
475 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
476 return -EINVAL;
477
478 if (tr && (len < *size))
479 *tr = *p;
480
481 *buf += len;
482 *size -= len;
483
484 return 0;
485}
486
487/**
488 * proc_put_long - converts an integer to a decimal ASCII formatted string
489 *
490 * @buf: the user buffer
491 * @size: the size of the user buffer
492 * @val: the integer to be converted
493 * @neg: sign of the number, %TRUE for negative
494 *
495 * In case of success @buf and @size are updated with the amount of bytes
496 * written.
497 */
498static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
499{
500 int len;
501 char tmp[TMPBUFLEN], *p = tmp;
502
503 sprintf(p, "%s%lu", neg ? "-" : "", val);
504 len = strlen(tmp);
505 if (len > *size)
506 len = *size;
507 memcpy(*buf, tmp, len);
508 *size -= len;
509 *buf += len;
510}
511#undef TMPBUFLEN
512
513static void proc_put_char(void **buf, size_t *size, char c)
514{
515 if (*size) {
516 char **buffer = (char **)buf;
517 **buffer = c;
518
519 (*size)--;
520 (*buffer)++;
521 *buf = *buffer;
522 }
523}
524
525static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
526 int *valp,
527 int write, void *data)
528{
529 if (write) {
530 if (*negp) {
531 if (*lvalp > (unsigned long) INT_MAX + 1)
532 return -EINVAL;
533 *valp = -*lvalp;
534 } else {
535 if (*lvalp > (unsigned long) INT_MAX)
536 return -EINVAL;
537 *valp = *lvalp;
538 }
539 } else {
540 int val = *valp;
541 if (val < 0) {
542 *negp = true;
543 *lvalp = -(unsigned long)val;
544 } else {
545 *negp = false;
546 *lvalp = (unsigned long)val;
547 }
548 }
549 return 0;
550}
551
552static int do_proc_douintvec_conv(unsigned long *lvalp,
553 unsigned int *valp,
554 int write, void *data)
555{
556 if (write) {
557 if (*lvalp > UINT_MAX)
558 return -EINVAL;
559 *valp = *lvalp;
560 } else {
561 unsigned int val = *valp;
562 *lvalp = (unsigned long)val;
563 }
564 return 0;
565}
566
567static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
568
569static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
570 int write, void *buffer,
571 size_t *lenp, loff_t *ppos,
572 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
573 int write, void *data),
574 void *data)
575{
576 int *i, vleft, first = 1, err = 0;
577 size_t left;
578 char *p;
579
580 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
581 *lenp = 0;
582 return 0;
583 }
584
585 i = (int *) tbl_data;
586 vleft = table->maxlen / sizeof(*i);
587 left = *lenp;
588
589 if (!conv)
590 conv = do_proc_dointvec_conv;
591
592 if (write) {
593 if (proc_first_pos_non_zero_ignore(ppos, table))
594 goto out;
595
596 if (left > PAGE_SIZE - 1)
597 left = PAGE_SIZE - 1;
598 p = buffer;
599 }
600
601 for (; left && vleft--; i++, first=0) {
602 unsigned long lval;
603 bool neg;
604
605 if (write) {
606 left -= proc_skip_spaces(&p);
607
608 if (!left)
609 break;
610 err = proc_get_long(&p, &left, &lval, &neg,
611 proc_wspace_sep,
612 sizeof(proc_wspace_sep), NULL);
613 if (err)
614 break;
615 if (conv(&neg, &lval, i, 1, data)) {
616 err = -EINVAL;
617 break;
618 }
619 } else {
620 if (conv(&neg, &lval, i, 0, data)) {
621 err = -EINVAL;
622 break;
623 }
624 if (!first)
625 proc_put_char(&buffer, &left, '\t');
626 proc_put_long(&buffer, &left, lval, neg);
627 }
628 }
629
630 if (!write && !first && left && !err)
631 proc_put_char(&buffer, &left, '\n');
632 if (write && !err && left)
633 left -= proc_skip_spaces(&p);
634 if (write && first)
635 return err ? : -EINVAL;
636 *lenp -= left;
637out:
638 *ppos += *lenp;
639 return err;
640}
641
642static int do_proc_dointvec(struct ctl_table *table, int write,
643 void *buffer, size_t *lenp, loff_t *ppos,
644 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
645 int write, void *data),
646 void *data)
647{
648 return __do_proc_dointvec(table->data, table, write,
649 buffer, lenp, ppos, conv, data);
650}
651
652static int do_proc_douintvec_w(unsigned int *tbl_data,
653 struct ctl_table *table,
654 void *buffer,
655 size_t *lenp, loff_t *ppos,
656 int (*conv)(unsigned long *lvalp,
657 unsigned int *valp,
658 int write, void *data),
659 void *data)
660{
661 unsigned long lval;
662 int err = 0;
663 size_t left;
664 bool neg;
665 char *p = buffer;
666
667 left = *lenp;
668
669 if (proc_first_pos_non_zero_ignore(ppos, table))
670 goto bail_early;
671
672 if (left > PAGE_SIZE - 1)
673 left = PAGE_SIZE - 1;
674
675 left -= proc_skip_spaces(&p);
676 if (!left) {
677 err = -EINVAL;
678 goto out_free;
679 }
680
681 err = proc_get_long(&p, &left, &lval, &neg,
682 proc_wspace_sep,
683 sizeof(proc_wspace_sep), NULL);
684 if (err || neg) {
685 err = -EINVAL;
686 goto out_free;
687 }
688
689 if (conv(&lval, tbl_data, 1, data)) {
690 err = -EINVAL;
691 goto out_free;
692 }
693
694 if (!err && left)
695 left -= proc_skip_spaces(&p);
696
697out_free:
698 if (err)
699 return -EINVAL;
700
701 return 0;
702
703 /* This is in keeping with old __do_proc_dointvec() */
704bail_early:
705 *ppos += *lenp;
706 return err;
707}
708
709static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
710 size_t *lenp, loff_t *ppos,
711 int (*conv)(unsigned long *lvalp,
712 unsigned int *valp,
713 int write, void *data),
714 void *data)
715{
716 unsigned long lval;
717 int err = 0;
718 size_t left;
719
720 left = *lenp;
721
722 if (conv(&lval, tbl_data, 0, data)) {
723 err = -EINVAL;
724 goto out;
725 }
726
727 proc_put_long(&buffer, &left, lval, false);
728 if (!left)
729 goto out;
730
731 proc_put_char(&buffer, &left, '\n');
732
733out:
734 *lenp -= left;
735 *ppos += *lenp;
736
737 return err;
738}
739
740static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
741 int write, void *buffer,
742 size_t *lenp, loff_t *ppos,
743 int (*conv)(unsigned long *lvalp,
744 unsigned int *valp,
745 int write, void *data),
746 void *data)
747{
748 unsigned int *i, vleft;
749
750 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
751 *lenp = 0;
752 return 0;
753 }
754
755 i = (unsigned int *) tbl_data;
756 vleft = table->maxlen / sizeof(*i);
757
758 /*
759 * Arrays are not supported, keep this simple. *Do not* add
760 * support for them.
761 */
762 if (vleft != 1) {
763 *lenp = 0;
764 return -EINVAL;
765 }
766
767 if (!conv)
768 conv = do_proc_douintvec_conv;
769
770 if (write)
771 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
772 conv, data);
773 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
774}
775
776static int do_proc_douintvec(struct ctl_table *table, int write,
777 void *buffer, size_t *lenp, loff_t *ppos,
778 int (*conv)(unsigned long *lvalp,
779 unsigned int *valp,
780 int write, void *data),
781 void *data)
782{
783 return __do_proc_douintvec(table->data, table, write,
784 buffer, lenp, ppos, conv, data);
785}
786
787/**
788 * proc_dointvec - read a vector of integers
789 * @table: the sysctl table
790 * @write: %TRUE if this is a write to the sysctl file
791 * @buffer: the user buffer
792 * @lenp: the size of the user buffer
793 * @ppos: file position
794 *
795 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
796 * values from/to the user buffer, treated as an ASCII string.
797 *
798 * Returns 0 on success.
799 */
800int proc_dointvec(struct ctl_table *table, int write, void *buffer,
801 size_t *lenp, loff_t *ppos)
802{
803 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
804}
805
806#ifdef CONFIG_COMPACTION
807static int proc_dointvec_minmax_warn_RT_change(struct ctl_table *table,
808 int write, void *buffer, size_t *lenp, loff_t *ppos)
809{
810 int ret, old;
811
812 if (!IS_ENABLED(CONFIG_PREEMPT_RT) || !write)
813 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
814
815 old = *(int *)table->data;
816 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
817 if (ret)
818 return ret;
819 if (old != *(int *)table->data)
820 pr_warn_once("sysctl attribute %s changed by %s[%d]\n",
821 table->procname, current->comm,
822 task_pid_nr(current));
823 return ret;
824}
825#endif
826
827/**
828 * proc_douintvec - read a vector of unsigned integers
829 * @table: the sysctl table
830 * @write: %TRUE if this is a write to the sysctl file
831 * @buffer: the user buffer
832 * @lenp: the size of the user buffer
833 * @ppos: file position
834 *
835 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
836 * values from/to the user buffer, treated as an ASCII string.
837 *
838 * Returns 0 on success.
839 */
840int proc_douintvec(struct ctl_table *table, int write, void *buffer,
841 size_t *lenp, loff_t *ppos)
842{
843 return do_proc_douintvec(table, write, buffer, lenp, ppos,
844 do_proc_douintvec_conv, NULL);
845}
846
847/*
848 * Taint values can only be increased
849 * This means we can safely use a temporary.
850 */
851static int proc_taint(struct ctl_table *table, int write,
852 void *buffer, size_t *lenp, loff_t *ppos)
853{
854 struct ctl_table t;
855 unsigned long tmptaint = get_taint();
856 int err;
857
858 if (write && !capable(CAP_SYS_ADMIN))
859 return -EPERM;
860
861 t = *table;
862 t.data = &tmptaint;
863 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
864 if (err < 0)
865 return err;
866
867 if (write) {
868 int i;
869
870 /*
871 * If we are relying on panic_on_taint not producing
872 * false positives due to userspace input, bail out
873 * before setting the requested taint flags.
874 */
875 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
876 return -EINVAL;
877
878 /*
879 * Poor man's atomic or. Not worth adding a primitive
880 * to everyone's atomic.h for this
881 */
882 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
883 if ((1UL << i) & tmptaint)
884 add_taint(i, LOCKDEP_STILL_OK);
885 }
886
887 return err;
888}
889
890#ifdef CONFIG_PRINTK
891static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
892 void *buffer, size_t *lenp, loff_t *ppos)
893{
894 if (write && !capable(CAP_SYS_ADMIN))
895 return -EPERM;
896
897 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
898}
899#endif
900
901/**
902 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
903 * @min: pointer to minimum allowable value
904 * @max: pointer to maximum allowable value
905 *
906 * The do_proc_dointvec_minmax_conv_param structure provides the
907 * minimum and maximum values for doing range checking for those sysctl
908 * parameters that use the proc_dointvec_minmax() handler.
909 */
910struct do_proc_dointvec_minmax_conv_param {
911 int *min;
912 int *max;
913};
914
915static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
916 int *valp,
917 int write, void *data)
918{
919 int tmp, ret;
920 struct do_proc_dointvec_minmax_conv_param *param = data;
921 /*
922 * If writing, first do so via a temporary local int so we can
923 * bounds-check it before touching *valp.
924 */
925 int *ip = write ? &tmp : valp;
926
927 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
928 if (ret)
929 return ret;
930
931 if (write) {
932 if ((param->min && *param->min > tmp) ||
933 (param->max && *param->max < tmp))
934 return -EINVAL;
935 *valp = tmp;
936 }
937
938 return 0;
939}
940
941/**
942 * proc_dointvec_minmax - read a vector of integers with min/max values
943 * @table: the sysctl table
944 * @write: %TRUE if this is a write to the sysctl file
945 * @buffer: the user buffer
946 * @lenp: the size of the user buffer
947 * @ppos: file position
948 *
949 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
950 * values from/to the user buffer, treated as an ASCII string.
951 *
952 * This routine will ensure the values are within the range specified by
953 * table->extra1 (min) and table->extra2 (max).
954 *
955 * Returns 0 on success or -EINVAL on write when the range check fails.
956 */
957int proc_dointvec_minmax(struct ctl_table *table, int write,
958 void *buffer, size_t *lenp, loff_t *ppos)
959{
960 struct do_proc_dointvec_minmax_conv_param param = {
961 .min = (int *) table->extra1,
962 .max = (int *) table->extra2,
963 };
964 return do_proc_dointvec(table, write, buffer, lenp, ppos,
965 do_proc_dointvec_minmax_conv, ¶m);
966}
967
968/**
969 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
970 * @min: pointer to minimum allowable value
971 * @max: pointer to maximum allowable value
972 *
973 * The do_proc_douintvec_minmax_conv_param structure provides the
974 * minimum and maximum values for doing range checking for those sysctl
975 * parameters that use the proc_douintvec_minmax() handler.
976 */
977struct do_proc_douintvec_minmax_conv_param {
978 unsigned int *min;
979 unsigned int *max;
980};
981
982static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
983 unsigned int *valp,
984 int write, void *data)
985{
986 int ret;
987 unsigned int tmp;
988 struct do_proc_douintvec_minmax_conv_param *param = data;
989 /* write via temporary local uint for bounds-checking */
990 unsigned int *up = write ? &tmp : valp;
991
992 ret = do_proc_douintvec_conv(lvalp, up, write, data);
993 if (ret)
994 return ret;
995
996 if (write) {
997 if ((param->min && *param->min > tmp) ||
998 (param->max && *param->max < tmp))
999 return -ERANGE;
1000
1001 *valp = tmp;
1002 }
1003
1004 return 0;
1005}
1006
1007/**
1008 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
1009 * @table: the sysctl table
1010 * @write: %TRUE if this is a write to the sysctl file
1011 * @buffer: the user buffer
1012 * @lenp: the size of the user buffer
1013 * @ppos: file position
1014 *
1015 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
1016 * values from/to the user buffer, treated as an ASCII string. Negative
1017 * strings are not allowed.
1018 *
1019 * This routine will ensure the values are within the range specified by
1020 * table->extra1 (min) and table->extra2 (max). There is a final sanity
1021 * check for UINT_MAX to avoid having to support wrap around uses from
1022 * userspace.
1023 *
1024 * Returns 0 on success or -ERANGE on write when the range check fails.
1025 */
1026int proc_douintvec_minmax(struct ctl_table *table, int write,
1027 void *buffer, size_t *lenp, loff_t *ppos)
1028{
1029 struct do_proc_douintvec_minmax_conv_param param = {
1030 .min = (unsigned int *) table->extra1,
1031 .max = (unsigned int *) table->extra2,
1032 };
1033 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1034 do_proc_douintvec_minmax_conv, ¶m);
1035}
1036
1037static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
1038 unsigned int *valp,
1039 int write, void *data)
1040{
1041 if (write) {
1042 unsigned int val;
1043
1044 val = round_pipe_size(*lvalp);
1045 if (val == 0)
1046 return -EINVAL;
1047
1048 *valp = val;
1049 } else {
1050 unsigned int val = *valp;
1051 *lvalp = (unsigned long) val;
1052 }
1053
1054 return 0;
1055}
1056
1057static int proc_dopipe_max_size(struct ctl_table *table, int write,
1058 void *buffer, size_t *lenp, loff_t *ppos)
1059{
1060 return do_proc_douintvec(table, write, buffer, lenp, ppos,
1061 do_proc_dopipe_max_size_conv, NULL);
1062}
1063
1064static void validate_coredump_safety(void)
1065{
1066#ifdef CONFIG_COREDUMP
1067 if (suid_dumpable == SUID_DUMP_ROOT &&
1068 core_pattern[0] != '/' && core_pattern[0] != '|') {
1069 printk(KERN_WARNING
1070"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
1071"Pipe handler or fully qualified core dump path required.\n"
1072"Set kernel.core_pattern before fs.suid_dumpable.\n"
1073 );
1074 }
1075#endif
1076}
1077
1078static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
1079 void *buffer, size_t *lenp, loff_t *ppos)
1080{
1081 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1082 if (!error)
1083 validate_coredump_safety();
1084 return error;
1085}
1086
1087#ifdef CONFIG_COREDUMP
1088static int proc_dostring_coredump(struct ctl_table *table, int write,
1089 void *buffer, size_t *lenp, loff_t *ppos)
1090{
1091 int error = proc_dostring(table, write, buffer, lenp, ppos);
1092 if (!error)
1093 validate_coredump_safety();
1094 return error;
1095}
1096#endif
1097
1098#ifdef CONFIG_MAGIC_SYSRQ
1099static int sysrq_sysctl_handler(struct ctl_table *table, int write,
1100 void *buffer, size_t *lenp, loff_t *ppos)
1101{
1102 int tmp, ret;
1103
1104 tmp = sysrq_mask();
1105
1106 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1107 lenp, ppos, NULL, NULL);
1108 if (ret || !write)
1109 return ret;
1110
1111 if (write)
1112 sysrq_toggle_support(tmp);
1113
1114 return 0;
1115}
1116#endif
1117
1118static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table,
1119 int write, void *buffer, size_t *lenp, loff_t *ppos,
1120 unsigned long convmul, unsigned long convdiv)
1121{
1122 unsigned long *i, *min, *max;
1123 int vleft, first = 1, err = 0;
1124 size_t left;
1125 char *p;
1126
1127 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1128 *lenp = 0;
1129 return 0;
1130 }
1131
1132 i = (unsigned long *) data;
1133 min = (unsigned long *) table->extra1;
1134 max = (unsigned long *) table->extra2;
1135 vleft = table->maxlen / sizeof(unsigned long);
1136 left = *lenp;
1137
1138 if (write) {
1139 if (proc_first_pos_non_zero_ignore(ppos, table))
1140 goto out;
1141
1142 if (left > PAGE_SIZE - 1)
1143 left = PAGE_SIZE - 1;
1144 p = buffer;
1145 }
1146
1147 for (; left && vleft--; i++, first = 0) {
1148 unsigned long val;
1149
1150 if (write) {
1151 bool neg;
1152
1153 left -= proc_skip_spaces(&p);
1154 if (!left)
1155 break;
1156
1157 err = proc_get_long(&p, &left, &val, &neg,
1158 proc_wspace_sep,
1159 sizeof(proc_wspace_sep), NULL);
1160 if (err)
1161 break;
1162 if (neg)
1163 continue;
1164 val = convmul * val / convdiv;
1165 if ((min && val < *min) || (max && val > *max)) {
1166 err = -EINVAL;
1167 break;
1168 }
1169 *i = val;
1170 } else {
1171 val = convdiv * (*i) / convmul;
1172 if (!first)
1173 proc_put_char(&buffer, &left, '\t');
1174 proc_put_long(&buffer, &left, val, false);
1175 }
1176 }
1177
1178 if (!write && !first && left && !err)
1179 proc_put_char(&buffer, &left, '\n');
1180 if (write && !err)
1181 left -= proc_skip_spaces(&p);
1182 if (write && first)
1183 return err ? : -EINVAL;
1184 *lenp -= left;
1185out:
1186 *ppos += *lenp;
1187 return err;
1188}
1189
1190static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
1191 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1192 unsigned long convdiv)
1193{
1194 return __do_proc_doulongvec_minmax(table->data, table, write,
1195 buffer, lenp, ppos, convmul, convdiv);
1196}
1197
1198/**
1199 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1200 * @table: the sysctl table
1201 * @write: %TRUE if this is a write to the sysctl file
1202 * @buffer: the user buffer
1203 * @lenp: the size of the user buffer
1204 * @ppos: file position
1205 *
1206 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1207 * values from/to the user buffer, treated as an ASCII string.
1208 *
1209 * This routine will ensure the values are within the range specified by
1210 * table->extra1 (min) and table->extra2 (max).
1211 *
1212 * Returns 0 on success.
1213 */
1214int proc_doulongvec_minmax(struct ctl_table *table, int write,
1215 void *buffer, size_t *lenp, loff_t *ppos)
1216{
1217 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1218}
1219
1220/**
1221 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1222 * @table: the sysctl table
1223 * @write: %TRUE if this is a write to the sysctl file
1224 * @buffer: the user buffer
1225 * @lenp: the size of the user buffer
1226 * @ppos: file position
1227 *
1228 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1229 * values from/to the user buffer, treated as an ASCII string. The values
1230 * are treated as milliseconds, and converted to jiffies when they are stored.
1231 *
1232 * This routine will ensure the values are within the range specified by
1233 * table->extra1 (min) and table->extra2 (max).
1234 *
1235 * Returns 0 on success.
1236 */
1237int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1238 void *buffer, size_t *lenp, loff_t *ppos)
1239{
1240 return do_proc_doulongvec_minmax(table, write, buffer,
1241 lenp, ppos, HZ, 1000l);
1242}
1243
1244
1245static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1246 int *valp,
1247 int write, void *data)
1248{
1249 if (write) {
1250 if (*lvalp > INT_MAX / HZ)
1251 return 1;
1252 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
1253 } else {
1254 int val = *valp;
1255 unsigned long lval;
1256 if (val < 0) {
1257 *negp = true;
1258 lval = -(unsigned long)val;
1259 } else {
1260 *negp = false;
1261 lval = (unsigned long)val;
1262 }
1263 *lvalp = lval / HZ;
1264 }
1265 return 0;
1266}
1267
1268static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1269 int *valp,
1270 int write, void *data)
1271{
1272 if (write) {
1273 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1274 return 1;
1275 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1276 } else {
1277 int val = *valp;
1278 unsigned long lval;
1279 if (val < 0) {
1280 *negp = true;
1281 lval = -(unsigned long)val;
1282 } else {
1283 *negp = false;
1284 lval = (unsigned long)val;
1285 }
1286 *lvalp = jiffies_to_clock_t(lval);
1287 }
1288 return 0;
1289}
1290
1291static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1292 int *valp,
1293 int write, void *data)
1294{
1295 if (write) {
1296 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1297
1298 if (jif > INT_MAX)
1299 return 1;
1300 *valp = (int)jif;
1301 } else {
1302 int val = *valp;
1303 unsigned long lval;
1304 if (val < 0) {
1305 *negp = true;
1306 lval = -(unsigned long)val;
1307 } else {
1308 *negp = false;
1309 lval = (unsigned long)val;
1310 }
1311 *lvalp = jiffies_to_msecs(lval);
1312 }
1313 return 0;
1314}
1315
1316/**
1317 * proc_dointvec_jiffies - read a vector of integers as seconds
1318 * @table: the sysctl table
1319 * @write: %TRUE if this is a write to the sysctl file
1320 * @buffer: the user buffer
1321 * @lenp: the size of the user buffer
1322 * @ppos: file position
1323 *
1324 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1325 * values from/to the user buffer, treated as an ASCII string.
1326 * The values read are assumed to be in seconds, and are converted into
1327 * jiffies.
1328 *
1329 * Returns 0 on success.
1330 */
1331int proc_dointvec_jiffies(struct ctl_table *table, int write,
1332 void *buffer, size_t *lenp, loff_t *ppos)
1333{
1334 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1335 do_proc_dointvec_jiffies_conv,NULL);
1336}
1337
1338/**
1339 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1340 * @table: the sysctl table
1341 * @write: %TRUE if this is a write to the sysctl file
1342 * @buffer: the user buffer
1343 * @lenp: the size of the user buffer
1344 * @ppos: pointer to the file position
1345 *
1346 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1347 * values from/to the user buffer, treated as an ASCII string.
1348 * The values read are assumed to be in 1/USER_HZ seconds, and
1349 * are converted into jiffies.
1350 *
1351 * Returns 0 on success.
1352 */
1353int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1354 void *buffer, size_t *lenp, loff_t *ppos)
1355{
1356 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1357 do_proc_dointvec_userhz_jiffies_conv,NULL);
1358}
1359
1360/**
1361 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1362 * @table: the sysctl table
1363 * @write: %TRUE if this is a write to the sysctl file
1364 * @buffer: the user buffer
1365 * @lenp: the size of the user buffer
1366 * @ppos: file position
1367 * @ppos: the current position in the file
1368 *
1369 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1370 * values from/to the user buffer, treated as an ASCII string.
1371 * The values read are assumed to be in 1/1000 seconds, and
1372 * are converted into jiffies.
1373 *
1374 * Returns 0 on success.
1375 */
1376int proc_dointvec_ms_jiffies(struct ctl_table *table, int write, void *buffer,
1377 size_t *lenp, loff_t *ppos)
1378{
1379 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1380 do_proc_dointvec_ms_jiffies_conv, NULL);
1381}
1382
1383static int proc_do_cad_pid(struct ctl_table *table, int write, void *buffer,
1384 size_t *lenp, loff_t *ppos)
1385{
1386 struct pid *new_pid;
1387 pid_t tmp;
1388 int r;
1389
1390 tmp = pid_vnr(cad_pid);
1391
1392 r = __do_proc_dointvec(&tmp, table, write, buffer,
1393 lenp, ppos, NULL, NULL);
1394 if (r || !write)
1395 return r;
1396
1397 new_pid = find_get_pid(tmp);
1398 if (!new_pid)
1399 return -ESRCH;
1400
1401 put_pid(xchg(&cad_pid, new_pid));
1402 return 0;
1403}
1404
1405/**
1406 * proc_do_large_bitmap - read/write from/to a large bitmap
1407 * @table: the sysctl table
1408 * @write: %TRUE if this is a write to the sysctl file
1409 * @buffer: the user buffer
1410 * @lenp: the size of the user buffer
1411 * @ppos: file position
1412 *
1413 * The bitmap is stored at table->data and the bitmap length (in bits)
1414 * in table->maxlen.
1415 *
1416 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1417 * large bitmaps may be represented in a compact manner. Writing into
1418 * the file will clear the bitmap then update it with the given input.
1419 *
1420 * Returns 0 on success.
1421 */
1422int proc_do_large_bitmap(struct ctl_table *table, int write,
1423 void *buffer, size_t *lenp, loff_t *ppos)
1424{
1425 int err = 0;
1426 bool first = 1;
1427 size_t left = *lenp;
1428 unsigned long bitmap_len = table->maxlen;
1429 unsigned long *bitmap = *(unsigned long **) table->data;
1430 unsigned long *tmp_bitmap = NULL;
1431 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1432
1433 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1434 *lenp = 0;
1435 return 0;
1436 }
1437
1438 if (write) {
1439 char *p = buffer;
1440 size_t skipped = 0;
1441
1442 if (left > PAGE_SIZE - 1) {
1443 left = PAGE_SIZE - 1;
1444 /* How much of the buffer we'll skip this pass */
1445 skipped = *lenp - left;
1446 }
1447
1448 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1449 if (!tmp_bitmap)
1450 return -ENOMEM;
1451 proc_skip_char(&p, &left, '\n');
1452 while (!err && left) {
1453 unsigned long val_a, val_b;
1454 bool neg;
1455 size_t saved_left;
1456
1457 /* In case we stop parsing mid-number, we can reset */
1458 saved_left = left;
1459 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1460 sizeof(tr_a), &c);
1461 /*
1462 * If we consumed the entirety of a truncated buffer or
1463 * only one char is left (may be a "-"), then stop here,
1464 * reset, & come back for more.
1465 */
1466 if ((left <= 1) && skipped) {
1467 left = saved_left;
1468 break;
1469 }
1470
1471 if (err)
1472 break;
1473 if (val_a >= bitmap_len || neg) {
1474 err = -EINVAL;
1475 break;
1476 }
1477
1478 val_b = val_a;
1479 if (left) {
1480 p++;
1481 left--;
1482 }
1483
1484 if (c == '-') {
1485 err = proc_get_long(&p, &left, &val_b,
1486 &neg, tr_b, sizeof(tr_b),
1487 &c);
1488 /*
1489 * If we consumed all of a truncated buffer or
1490 * then stop here, reset, & come back for more.
1491 */
1492 if (!left && skipped) {
1493 left = saved_left;
1494 break;
1495 }
1496
1497 if (err)
1498 break;
1499 if (val_b >= bitmap_len || neg ||
1500 val_a > val_b) {
1501 err = -EINVAL;
1502 break;
1503 }
1504 if (left) {
1505 p++;
1506 left--;
1507 }
1508 }
1509
1510 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1511 first = 0;
1512 proc_skip_char(&p, &left, '\n');
1513 }
1514 left += skipped;
1515 } else {
1516 unsigned long bit_a, bit_b = 0;
1517
1518 while (left) {
1519 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1520 if (bit_a >= bitmap_len)
1521 break;
1522 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1523 bit_a + 1) - 1;
1524
1525 if (!first)
1526 proc_put_char(&buffer, &left, ',');
1527 proc_put_long(&buffer, &left, bit_a, false);
1528 if (bit_a != bit_b) {
1529 proc_put_char(&buffer, &left, '-');
1530 proc_put_long(&buffer, &left, bit_b, false);
1531 }
1532
1533 first = 0; bit_b++;
1534 }
1535 proc_put_char(&buffer, &left, '\n');
1536 }
1537
1538 if (!err) {
1539 if (write) {
1540 if (*ppos)
1541 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1542 else
1543 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1544 }
1545 *lenp -= left;
1546 *ppos += *lenp;
1547 }
1548
1549 bitmap_free(tmp_bitmap);
1550 return err;
1551}
1552
1553#else /* CONFIG_PROC_SYSCTL */
1554
1555int proc_dostring(struct ctl_table *table, int write,
1556 void *buffer, size_t *lenp, loff_t *ppos)
1557{
1558 return -ENOSYS;
1559}
1560
1561int proc_dointvec(struct ctl_table *table, int write,
1562 void *buffer, size_t *lenp, loff_t *ppos)
1563{
1564 return -ENOSYS;
1565}
1566
1567int proc_douintvec(struct ctl_table *table, int write,
1568 void *buffer, size_t *lenp, loff_t *ppos)
1569{
1570 return -ENOSYS;
1571}
1572
1573int proc_dointvec_minmax(struct ctl_table *table, int write,
1574 void *buffer, size_t *lenp, loff_t *ppos)
1575{
1576 return -ENOSYS;
1577}
1578
1579int proc_douintvec_minmax(struct ctl_table *table, int write,
1580 void *buffer, size_t *lenp, loff_t *ppos)
1581{
1582 return -ENOSYS;
1583}
1584
1585int proc_dointvec_jiffies(struct ctl_table *table, int write,
1586 void *buffer, size_t *lenp, loff_t *ppos)
1587{
1588 return -ENOSYS;
1589}
1590
1591int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
1592 void *buffer, size_t *lenp, loff_t *ppos)
1593{
1594 return -ENOSYS;
1595}
1596
1597int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
1598 void *buffer, size_t *lenp, loff_t *ppos)
1599{
1600 return -ENOSYS;
1601}
1602
1603int proc_doulongvec_minmax(struct ctl_table *table, int write,
1604 void *buffer, size_t *lenp, loff_t *ppos)
1605{
1606 return -ENOSYS;
1607}
1608
1609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
1610 void *buffer, size_t *lenp, loff_t *ppos)
1611{
1612 return -ENOSYS;
1613}
1614
1615int proc_do_large_bitmap(struct ctl_table *table, int write,
1616 void *buffer, size_t *lenp, loff_t *ppos)
1617{
1618 return -ENOSYS;
1619}
1620
1621#endif /* CONFIG_PROC_SYSCTL */
1622
1623#if defined(CONFIG_SYSCTL)
1624int proc_do_static_key(struct ctl_table *table, int write,
1625 void *buffer, size_t *lenp, loff_t *ppos)
1626{
1627 struct static_key *key = (struct static_key *)table->data;
1628 static DEFINE_MUTEX(static_key_mutex);
1629 int val, ret;
1630 struct ctl_table tmp = {
1631 .data = &val,
1632 .maxlen = sizeof(val),
1633 .mode = table->mode,
1634 .extra1 = SYSCTL_ZERO,
1635 .extra2 = SYSCTL_ONE,
1636 };
1637
1638 if (write && !capable(CAP_SYS_ADMIN))
1639 return -EPERM;
1640
1641 mutex_lock(&static_key_mutex);
1642 val = static_key_enabled(key);
1643 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1644 if (write && !ret) {
1645 if (val)
1646 static_key_enable(key);
1647 else
1648 static_key_disable(key);
1649 }
1650 mutex_unlock(&static_key_mutex);
1651 return ret;
1652}
1653
1654static struct ctl_table kern_table[] = {
1655 {
1656 .procname = "sched_child_runs_first",
1657 .data = &sysctl_sched_child_runs_first,
1658 .maxlen = sizeof(unsigned int),
1659 .mode = 0644,
1660 .proc_handler = proc_dointvec,
1661 },
1662#ifdef CONFIG_SCHED_DEBUG
1663 {
1664 .procname = "sched_min_granularity_ns",
1665 .data = &sysctl_sched_min_granularity,
1666 .maxlen = sizeof(unsigned int),
1667 .mode = 0644,
1668 .proc_handler = sched_proc_update_handler,
1669 .extra1 = &min_sched_granularity_ns,
1670 .extra2 = &max_sched_granularity_ns,
1671 },
1672 {
1673 .procname = "sched_latency_ns",
1674 .data = &sysctl_sched_latency,
1675 .maxlen = sizeof(unsigned int),
1676 .mode = 0644,
1677 .proc_handler = sched_proc_update_handler,
1678 .extra1 = &min_sched_granularity_ns,
1679 .extra2 = &max_sched_granularity_ns,
1680 },
1681 {
1682 .procname = "sched_wakeup_granularity_ns",
1683 .data = &sysctl_sched_wakeup_granularity,
1684 .maxlen = sizeof(unsigned int),
1685 .mode = 0644,
1686 .proc_handler = sched_proc_update_handler,
1687 .extra1 = &min_wakeup_granularity_ns,
1688 .extra2 = &max_wakeup_granularity_ns,
1689 },
1690#ifdef CONFIG_SMP
1691 {
1692 .procname = "sched_tunable_scaling",
1693 .data = &sysctl_sched_tunable_scaling,
1694 .maxlen = sizeof(enum sched_tunable_scaling),
1695 .mode = 0644,
1696 .proc_handler = sched_proc_update_handler,
1697 .extra1 = &min_sched_tunable_scaling,
1698 .extra2 = &max_sched_tunable_scaling,
1699 },
1700 {
1701 .procname = "sched_migration_cost_ns",
1702 .data = &sysctl_sched_migration_cost,
1703 .maxlen = sizeof(unsigned int),
1704 .mode = 0644,
1705 .proc_handler = proc_dointvec,
1706 },
1707 {
1708 .procname = "sched_nr_migrate",
1709 .data = &sysctl_sched_nr_migrate,
1710 .maxlen = sizeof(unsigned int),
1711 .mode = 0644,
1712 .proc_handler = proc_dointvec,
1713 },
1714#ifdef CONFIG_SCHEDSTATS
1715 {
1716 .procname = "sched_schedstats",
1717 .data = NULL,
1718 .maxlen = sizeof(unsigned int),
1719 .mode = 0644,
1720 .proc_handler = sysctl_schedstats,
1721 .extra1 = SYSCTL_ZERO,
1722 .extra2 = SYSCTL_ONE,
1723 },
1724#endif /* CONFIG_SCHEDSTATS */
1725#endif /* CONFIG_SMP */
1726#ifdef CONFIG_NUMA_BALANCING
1727 {
1728 .procname = "numa_balancing_scan_delay_ms",
1729 .data = &sysctl_numa_balancing_scan_delay,
1730 .maxlen = sizeof(unsigned int),
1731 .mode = 0644,
1732 .proc_handler = proc_dointvec,
1733 },
1734 {
1735 .procname = "numa_balancing_scan_period_min_ms",
1736 .data = &sysctl_numa_balancing_scan_period_min,
1737 .maxlen = sizeof(unsigned int),
1738 .mode = 0644,
1739 .proc_handler = proc_dointvec,
1740 },
1741 {
1742 .procname = "numa_balancing_scan_period_max_ms",
1743 .data = &sysctl_numa_balancing_scan_period_max,
1744 .maxlen = sizeof(unsigned int),
1745 .mode = 0644,
1746 .proc_handler = proc_dointvec,
1747 },
1748 {
1749 .procname = "numa_balancing_scan_size_mb",
1750 .data = &sysctl_numa_balancing_scan_size,
1751 .maxlen = sizeof(unsigned int),
1752 .mode = 0644,
1753 .proc_handler = proc_dointvec_minmax,
1754 .extra1 = SYSCTL_ONE,
1755 },
1756 {
1757 .procname = "numa_balancing",
1758 .data = NULL, /* filled in by handler */
1759 .maxlen = sizeof(unsigned int),
1760 .mode = 0644,
1761 .proc_handler = sysctl_numa_balancing,
1762 .extra1 = SYSCTL_ZERO,
1763 .extra2 = SYSCTL_ONE,
1764 },
1765#endif /* CONFIG_NUMA_BALANCING */
1766#endif /* CONFIG_SCHED_DEBUG */
1767 {
1768 .procname = "sched_rt_period_us",
1769 .data = &sysctl_sched_rt_period,
1770 .maxlen = sizeof(unsigned int),
1771 .mode = 0644,
1772 .proc_handler = sched_rt_handler,
1773 },
1774 {
1775 .procname = "sched_rt_runtime_us",
1776 .data = &sysctl_sched_rt_runtime,
1777 .maxlen = sizeof(int),
1778 .mode = 0644,
1779 .proc_handler = sched_rt_handler,
1780 },
1781 {
1782 .procname = "sched_deadline_period_max_us",
1783 .data = &sysctl_sched_dl_period_max,
1784 .maxlen = sizeof(unsigned int),
1785 .mode = 0644,
1786 .proc_handler = proc_dointvec,
1787 },
1788 {
1789 .procname = "sched_deadline_period_min_us",
1790 .data = &sysctl_sched_dl_period_min,
1791 .maxlen = sizeof(unsigned int),
1792 .mode = 0644,
1793 .proc_handler = proc_dointvec,
1794 },
1795 {
1796 .procname = "sched_rr_timeslice_ms",
1797 .data = &sysctl_sched_rr_timeslice,
1798 .maxlen = sizeof(int),
1799 .mode = 0644,
1800 .proc_handler = sched_rr_handler,
1801 },
1802#ifdef CONFIG_UCLAMP_TASK
1803 {
1804 .procname = "sched_util_clamp_min",
1805 .data = &sysctl_sched_uclamp_util_min,
1806 .maxlen = sizeof(unsigned int),
1807 .mode = 0644,
1808 .proc_handler = sysctl_sched_uclamp_handler,
1809 },
1810 {
1811 .procname = "sched_util_clamp_max",
1812 .data = &sysctl_sched_uclamp_util_max,
1813 .maxlen = sizeof(unsigned int),
1814 .mode = 0644,
1815 .proc_handler = sysctl_sched_uclamp_handler,
1816 },
1817 {
1818 .procname = "sched_util_clamp_min_rt_default",
1819 .data = &sysctl_sched_uclamp_util_min_rt_default,
1820 .maxlen = sizeof(unsigned int),
1821 .mode = 0644,
1822 .proc_handler = sysctl_sched_uclamp_handler,
1823 },
1824#endif
1825#ifdef CONFIG_SCHED_AUTOGROUP
1826 {
1827 .procname = "sched_autogroup_enabled",
1828 .data = &sysctl_sched_autogroup_enabled,
1829 .maxlen = sizeof(unsigned int),
1830 .mode = 0644,
1831 .proc_handler = proc_dointvec_minmax,
1832 .extra1 = SYSCTL_ZERO,
1833 .extra2 = SYSCTL_ONE,
1834 },
1835#endif
1836#ifdef CONFIG_CFS_BANDWIDTH
1837 {
1838 .procname = "sched_cfs_bandwidth_slice_us",
1839 .data = &sysctl_sched_cfs_bandwidth_slice,
1840 .maxlen = sizeof(unsigned int),
1841 .mode = 0644,
1842 .proc_handler = proc_dointvec_minmax,
1843 .extra1 = SYSCTL_ONE,
1844 },
1845#endif
1846#if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
1847 {
1848 .procname = "sched_energy_aware",
1849 .data = &sysctl_sched_energy_aware,
1850 .maxlen = sizeof(unsigned int),
1851 .mode = 0644,
1852 .proc_handler = sched_energy_aware_handler,
1853 .extra1 = SYSCTL_ZERO,
1854 .extra2 = SYSCTL_ONE,
1855 },
1856#endif
1857#ifdef CONFIG_PROVE_LOCKING
1858 {
1859 .procname = "prove_locking",
1860 .data = &prove_locking,
1861 .maxlen = sizeof(int),
1862 .mode = 0644,
1863 .proc_handler = proc_dointvec,
1864 },
1865#endif
1866#ifdef CONFIG_LOCK_STAT
1867 {
1868 .procname = "lock_stat",
1869 .data = &lock_stat,
1870 .maxlen = sizeof(int),
1871 .mode = 0644,
1872 .proc_handler = proc_dointvec,
1873 },
1874#endif
1875 {
1876 .procname = "panic",
1877 .data = &panic_timeout,
1878 .maxlen = sizeof(int),
1879 .mode = 0644,
1880 .proc_handler = proc_dointvec,
1881 },
1882#ifdef CONFIG_COREDUMP
1883 {
1884 .procname = "core_uses_pid",
1885 .data = &core_uses_pid,
1886 .maxlen = sizeof(int),
1887 .mode = 0644,
1888 .proc_handler = proc_dointvec,
1889 },
1890 {
1891 .procname = "core_pattern",
1892 .data = core_pattern,
1893 .maxlen = CORENAME_MAX_SIZE,
1894 .mode = 0644,
1895 .proc_handler = proc_dostring_coredump,
1896 },
1897 {
1898 .procname = "core_pipe_limit",
1899 .data = &core_pipe_limit,
1900 .maxlen = sizeof(unsigned int),
1901 .mode = 0644,
1902 .proc_handler = proc_dointvec,
1903 },
1904#endif
1905#ifdef CONFIG_PROC_SYSCTL
1906 {
1907 .procname = "tainted",
1908 .maxlen = sizeof(long),
1909 .mode = 0644,
1910 .proc_handler = proc_taint,
1911 },
1912 {
1913 .procname = "sysctl_writes_strict",
1914 .data = &sysctl_writes_strict,
1915 .maxlen = sizeof(int),
1916 .mode = 0644,
1917 .proc_handler = proc_dointvec_minmax,
1918 .extra1 = &neg_one,
1919 .extra2 = SYSCTL_ONE,
1920 },
1921#endif
1922#ifdef CONFIG_LATENCYTOP
1923 {
1924 .procname = "latencytop",
1925 .data = &latencytop_enabled,
1926 .maxlen = sizeof(int),
1927 .mode = 0644,
1928 .proc_handler = sysctl_latencytop,
1929 },
1930#endif
1931#ifdef CONFIG_BLK_DEV_INITRD
1932 {
1933 .procname = "real-root-dev",
1934 .data = &real_root_dev,
1935 .maxlen = sizeof(int),
1936 .mode = 0644,
1937 .proc_handler = proc_dointvec,
1938 },
1939#endif
1940 {
1941 .procname = "print-fatal-signals",
1942 .data = &print_fatal_signals,
1943 .maxlen = sizeof(int),
1944 .mode = 0644,
1945 .proc_handler = proc_dointvec,
1946 },
1947#ifdef CONFIG_SPARC
1948 {
1949 .procname = "reboot-cmd",
1950 .data = reboot_command,
1951 .maxlen = 256,
1952 .mode = 0644,
1953 .proc_handler = proc_dostring,
1954 },
1955 {
1956 .procname = "stop-a",
1957 .data = &stop_a_enabled,
1958 .maxlen = sizeof (int),
1959 .mode = 0644,
1960 .proc_handler = proc_dointvec,
1961 },
1962 {
1963 .procname = "scons-poweroff",
1964 .data = &scons_pwroff,
1965 .maxlen = sizeof (int),
1966 .mode = 0644,
1967 .proc_handler = proc_dointvec,
1968 },
1969#endif
1970#ifdef CONFIG_SPARC64
1971 {
1972 .procname = "tsb-ratio",
1973 .data = &sysctl_tsb_ratio,
1974 .maxlen = sizeof (int),
1975 .mode = 0644,
1976 .proc_handler = proc_dointvec,
1977 },
1978#endif
1979#ifdef CONFIG_PARISC
1980 {
1981 .procname = "soft-power",
1982 .data = &pwrsw_enabled,
1983 .maxlen = sizeof (int),
1984 .mode = 0644,
1985 .proc_handler = proc_dointvec,
1986 },
1987#endif
1988#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1989 {
1990 .procname = "unaligned-trap",
1991 .data = &unaligned_enabled,
1992 .maxlen = sizeof (int),
1993 .mode = 0644,
1994 .proc_handler = proc_dointvec,
1995 },
1996#endif
1997 {
1998 .procname = "ctrl-alt-del",
1999 .data = &C_A_D,
2000 .maxlen = sizeof(int),
2001 .mode = 0644,
2002 .proc_handler = proc_dointvec,
2003 },
2004#ifdef CONFIG_FUNCTION_TRACER
2005 {
2006 .procname = "ftrace_enabled",
2007 .data = &ftrace_enabled,
2008 .maxlen = sizeof(int),
2009 .mode = 0644,
2010 .proc_handler = ftrace_enable_sysctl,
2011 },
2012#endif
2013#ifdef CONFIG_STACK_TRACER
2014 {
2015 .procname = "stack_tracer_enabled",
2016 .data = &stack_tracer_enabled,
2017 .maxlen = sizeof(int),
2018 .mode = 0644,
2019 .proc_handler = stack_trace_sysctl,
2020 },
2021#endif
2022#ifdef CONFIG_TRACING
2023 {
2024 .procname = "ftrace_dump_on_oops",
2025 .data = &ftrace_dump_on_oops,
2026 .maxlen = sizeof(int),
2027 .mode = 0644,
2028 .proc_handler = proc_dointvec,
2029 },
2030 {
2031 .procname = "traceoff_on_warning",
2032 .data = &__disable_trace_on_warning,
2033 .maxlen = sizeof(__disable_trace_on_warning),
2034 .mode = 0644,
2035 .proc_handler = proc_dointvec,
2036 },
2037 {
2038 .procname = "tracepoint_printk",
2039 .data = &tracepoint_printk,
2040 .maxlen = sizeof(tracepoint_printk),
2041 .mode = 0644,
2042 .proc_handler = tracepoint_printk_sysctl,
2043 },
2044#endif
2045#ifdef CONFIG_KEXEC_CORE
2046 {
2047 .procname = "kexec_load_disabled",
2048 .data = &kexec_load_disabled,
2049 .maxlen = sizeof(int),
2050 .mode = 0644,
2051 /* only handle a transition from default "0" to "1" */
2052 .proc_handler = proc_dointvec_minmax,
2053 .extra1 = SYSCTL_ONE,
2054 .extra2 = SYSCTL_ONE,
2055 },
2056#endif
2057#ifdef CONFIG_MODULES
2058 {
2059 .procname = "modprobe",
2060 .data = &modprobe_path,
2061 .maxlen = KMOD_PATH_LEN,
2062 .mode = 0644,
2063 .proc_handler = proc_dostring,
2064 },
2065 {
2066 .procname = "modules_disabled",
2067 .data = &modules_disabled,
2068 .maxlen = sizeof(int),
2069 .mode = 0644,
2070 /* only handle a transition from default "0" to "1" */
2071 .proc_handler = proc_dointvec_minmax,
2072 .extra1 = SYSCTL_ONE,
2073 .extra2 = SYSCTL_ONE,
2074 },
2075#endif
2076#ifdef CONFIG_UEVENT_HELPER
2077 {
2078 .procname = "hotplug",
2079 .data = &uevent_helper,
2080 .maxlen = UEVENT_HELPER_PATH_LEN,
2081 .mode = 0644,
2082 .proc_handler = proc_dostring,
2083 },
2084#endif
2085#ifdef CONFIG_CHR_DEV_SG
2086 {
2087 .procname = "sg-big-buff",
2088 .data = &sg_big_buff,
2089 .maxlen = sizeof (int),
2090 .mode = 0444,
2091 .proc_handler = proc_dointvec,
2092 },
2093#endif
2094#ifdef CONFIG_BSD_PROCESS_ACCT
2095 {
2096 .procname = "acct",
2097 .data = &acct_parm,
2098 .maxlen = 3*sizeof(int),
2099 .mode = 0644,
2100 .proc_handler = proc_dointvec,
2101 },
2102#endif
2103#ifdef CONFIG_MAGIC_SYSRQ
2104 {
2105 .procname = "sysrq",
2106 .data = NULL,
2107 .maxlen = sizeof (int),
2108 .mode = 0644,
2109 .proc_handler = sysrq_sysctl_handler,
2110 },
2111#endif
2112#ifdef CONFIG_PROC_SYSCTL
2113 {
2114 .procname = "cad_pid",
2115 .data = NULL,
2116 .maxlen = sizeof (int),
2117 .mode = 0600,
2118 .proc_handler = proc_do_cad_pid,
2119 },
2120#endif
2121 {
2122 .procname = "threads-max",
2123 .data = NULL,
2124 .maxlen = sizeof(int),
2125 .mode = 0644,
2126 .proc_handler = sysctl_max_threads,
2127 },
2128 {
2129 .procname = "random",
2130 .mode = 0555,
2131 .child = random_table,
2132 },
2133 {
2134 .procname = "usermodehelper",
2135 .mode = 0555,
2136 .child = usermodehelper_table,
2137 },
2138#ifdef CONFIG_FW_LOADER_USER_HELPER
2139 {
2140 .procname = "firmware_config",
2141 .mode = 0555,
2142 .child = firmware_config_table,
2143 },
2144#endif
2145 {
2146 .procname = "overflowuid",
2147 .data = &overflowuid,
2148 .maxlen = sizeof(int),
2149 .mode = 0644,
2150 .proc_handler = proc_dointvec_minmax,
2151 .extra1 = &minolduid,
2152 .extra2 = &maxolduid,
2153 },
2154 {
2155 .procname = "overflowgid",
2156 .data = &overflowgid,
2157 .maxlen = sizeof(int),
2158 .mode = 0644,
2159 .proc_handler = proc_dointvec_minmax,
2160 .extra1 = &minolduid,
2161 .extra2 = &maxolduid,
2162 },
2163#ifdef CONFIG_S390
2164 {
2165 .procname = "userprocess_debug",
2166 .data = &show_unhandled_signals,
2167 .maxlen = sizeof(int),
2168 .mode = 0644,
2169 .proc_handler = proc_dointvec,
2170 },
2171#endif
2172#ifdef CONFIG_SMP
2173 {
2174 .procname = "oops_all_cpu_backtrace",
2175 .data = &sysctl_oops_all_cpu_backtrace,
2176 .maxlen = sizeof(int),
2177 .mode = 0644,
2178 .proc_handler = proc_dointvec_minmax,
2179 .extra1 = SYSCTL_ZERO,
2180 .extra2 = SYSCTL_ONE,
2181 },
2182#endif /* CONFIG_SMP */
2183 {
2184 .procname = "pid_max",
2185 .data = &pid_max,
2186 .maxlen = sizeof (int),
2187 .mode = 0644,
2188 .proc_handler = proc_dointvec_minmax,
2189 .extra1 = &pid_max_min,
2190 .extra2 = &pid_max_max,
2191 },
2192 {
2193 .procname = "panic_on_oops",
2194 .data = &panic_on_oops,
2195 .maxlen = sizeof(int),
2196 .mode = 0644,
2197 .proc_handler = proc_dointvec,
2198 },
2199 {
2200 .procname = "panic_print",
2201 .data = &panic_print,
2202 .maxlen = sizeof(unsigned long),
2203 .mode = 0644,
2204 .proc_handler = proc_doulongvec_minmax,
2205 },
2206#if defined CONFIG_PRINTK
2207 {
2208 .procname = "printk",
2209 .data = &console_loglevel,
2210 .maxlen = 4*sizeof(int),
2211 .mode = 0644,
2212 .proc_handler = proc_dointvec,
2213 },
2214 {
2215 .procname = "printk_ratelimit",
2216 .data = &printk_ratelimit_state.interval,
2217 .maxlen = sizeof(int),
2218 .mode = 0644,
2219 .proc_handler = proc_dointvec_jiffies,
2220 },
2221 {
2222 .procname = "printk_ratelimit_burst",
2223 .data = &printk_ratelimit_state.burst,
2224 .maxlen = sizeof(int),
2225 .mode = 0644,
2226 .proc_handler = proc_dointvec,
2227 },
2228 {
2229 .procname = "printk_delay",
2230 .data = &printk_delay_msec,
2231 .maxlen = sizeof(int),
2232 .mode = 0644,
2233 .proc_handler = proc_dointvec_minmax,
2234 .extra1 = SYSCTL_ZERO,
2235 .extra2 = &ten_thousand,
2236 },
2237 {
2238 .procname = "printk_devkmsg",
2239 .data = devkmsg_log_str,
2240 .maxlen = DEVKMSG_STR_MAX_SIZE,
2241 .mode = 0644,
2242 .proc_handler = devkmsg_sysctl_set_loglvl,
2243 },
2244 {
2245 .procname = "dmesg_restrict",
2246 .data = &dmesg_restrict,
2247 .maxlen = sizeof(int),
2248 .mode = 0644,
2249 .proc_handler = proc_dointvec_minmax_sysadmin,
2250 .extra1 = SYSCTL_ZERO,
2251 .extra2 = SYSCTL_ONE,
2252 },
2253 {
2254 .procname = "kptr_restrict",
2255 .data = &kptr_restrict,
2256 .maxlen = sizeof(int),
2257 .mode = 0644,
2258 .proc_handler = proc_dointvec_minmax_sysadmin,
2259 .extra1 = SYSCTL_ZERO,
2260 .extra2 = &two,
2261 },
2262#endif
2263 {
2264 .procname = "ngroups_max",
2265 .data = &ngroups_max,
2266 .maxlen = sizeof (int),
2267 .mode = 0444,
2268 .proc_handler = proc_dointvec,
2269 },
2270 {
2271 .procname = "cap_last_cap",
2272 .data = (void *)&cap_last_cap,
2273 .maxlen = sizeof(int),
2274 .mode = 0444,
2275 .proc_handler = proc_dointvec,
2276 },
2277#if defined(CONFIG_LOCKUP_DETECTOR)
2278 {
2279 .procname = "watchdog",
2280 .data = &watchdog_user_enabled,
2281 .maxlen = sizeof(int),
2282 .mode = 0644,
2283 .proc_handler = proc_watchdog,
2284 .extra1 = SYSCTL_ZERO,
2285 .extra2 = SYSCTL_ONE,
2286 },
2287 {
2288 .procname = "watchdog_thresh",
2289 .data = &watchdog_thresh,
2290 .maxlen = sizeof(int),
2291 .mode = 0644,
2292 .proc_handler = proc_watchdog_thresh,
2293 .extra1 = SYSCTL_ZERO,
2294 .extra2 = &sixty,
2295 },
2296 {
2297 .procname = "nmi_watchdog",
2298 .data = &nmi_watchdog_user_enabled,
2299 .maxlen = sizeof(int),
2300 .mode = NMI_WATCHDOG_SYSCTL_PERM,
2301 .proc_handler = proc_nmi_watchdog,
2302 .extra1 = SYSCTL_ZERO,
2303 .extra2 = SYSCTL_ONE,
2304 },
2305 {
2306 .procname = "watchdog_cpumask",
2307 .data = &watchdog_cpumask_bits,
2308 .maxlen = NR_CPUS,
2309 .mode = 0644,
2310 .proc_handler = proc_watchdog_cpumask,
2311 },
2312#ifdef CONFIG_SOFTLOCKUP_DETECTOR
2313 {
2314 .procname = "soft_watchdog",
2315 .data = &soft_watchdog_user_enabled,
2316 .maxlen = sizeof(int),
2317 .mode = 0644,
2318 .proc_handler = proc_soft_watchdog,
2319 .extra1 = SYSCTL_ZERO,
2320 .extra2 = SYSCTL_ONE,
2321 },
2322 {
2323 .procname = "softlockup_panic",
2324 .data = &softlockup_panic,
2325 .maxlen = sizeof(int),
2326 .mode = 0644,
2327 .proc_handler = proc_dointvec_minmax,
2328 .extra1 = SYSCTL_ZERO,
2329 .extra2 = SYSCTL_ONE,
2330 },
2331#ifdef CONFIG_SMP
2332 {
2333 .procname = "softlockup_all_cpu_backtrace",
2334 .data = &sysctl_softlockup_all_cpu_backtrace,
2335 .maxlen = sizeof(int),
2336 .mode = 0644,
2337 .proc_handler = proc_dointvec_minmax,
2338 .extra1 = SYSCTL_ZERO,
2339 .extra2 = SYSCTL_ONE,
2340 },
2341#endif /* CONFIG_SMP */
2342#endif
2343#ifdef CONFIG_HARDLOCKUP_DETECTOR
2344 {
2345 .procname = "hardlockup_panic",
2346 .data = &hardlockup_panic,
2347 .maxlen = sizeof(int),
2348 .mode = 0644,
2349 .proc_handler = proc_dointvec_minmax,
2350 .extra1 = SYSCTL_ZERO,
2351 .extra2 = SYSCTL_ONE,
2352 },
2353#ifdef CONFIG_SMP
2354 {
2355 .procname = "hardlockup_all_cpu_backtrace",
2356 .data = &sysctl_hardlockup_all_cpu_backtrace,
2357 .maxlen = sizeof(int),
2358 .mode = 0644,
2359 .proc_handler = proc_dointvec_minmax,
2360 .extra1 = SYSCTL_ZERO,
2361 .extra2 = SYSCTL_ONE,
2362 },
2363#endif /* CONFIG_SMP */
2364#endif
2365#endif
2366
2367#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
2368 {
2369 .procname = "unknown_nmi_panic",
2370 .data = &unknown_nmi_panic,
2371 .maxlen = sizeof (int),
2372 .mode = 0644,
2373 .proc_handler = proc_dointvec,
2374 },
2375#endif
2376
2377#if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
2378 defined(CONFIG_DEBUG_STACKOVERFLOW)
2379 {
2380 .procname = "panic_on_stackoverflow",
2381 .data = &sysctl_panic_on_stackoverflow,
2382 .maxlen = sizeof(int),
2383 .mode = 0644,
2384 .proc_handler = proc_dointvec,
2385 },
2386#endif
2387#if defined(CONFIG_X86)
2388 {
2389 .procname = "panic_on_unrecovered_nmi",
2390 .data = &panic_on_unrecovered_nmi,
2391 .maxlen = sizeof(int),
2392 .mode = 0644,
2393 .proc_handler = proc_dointvec,
2394 },
2395 {
2396 .procname = "panic_on_io_nmi",
2397 .data = &panic_on_io_nmi,
2398 .maxlen = sizeof(int),
2399 .mode = 0644,
2400 .proc_handler = proc_dointvec,
2401 },
2402 {
2403 .procname = "bootloader_type",
2404 .data = &bootloader_type,
2405 .maxlen = sizeof (int),
2406 .mode = 0444,
2407 .proc_handler = proc_dointvec,
2408 },
2409 {
2410 .procname = "bootloader_version",
2411 .data = &bootloader_version,
2412 .maxlen = sizeof (int),
2413 .mode = 0444,
2414 .proc_handler = proc_dointvec,
2415 },
2416 {
2417 .procname = "io_delay_type",
2418 .data = &io_delay_type,
2419 .maxlen = sizeof(int),
2420 .mode = 0644,
2421 .proc_handler = proc_dointvec,
2422 },
2423#endif
2424#if defined(CONFIG_MMU)
2425 {
2426 .procname = "randomize_va_space",
2427 .data = &randomize_va_space,
2428 .maxlen = sizeof(int),
2429 .mode = 0644,
2430 .proc_handler = proc_dointvec,
2431 },
2432#endif
2433#if defined(CONFIG_S390) && defined(CONFIG_SMP)
2434 {
2435 .procname = "spin_retry",
2436 .data = &spin_retry,
2437 .maxlen = sizeof (int),
2438 .mode = 0644,
2439 .proc_handler = proc_dointvec,
2440 },
2441#endif
2442#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
2443 {
2444 .procname = "acpi_video_flags",
2445 .data = &acpi_realmode_flags,
2446 .maxlen = sizeof (unsigned long),
2447 .mode = 0644,
2448 .proc_handler = proc_doulongvec_minmax,
2449 },
2450#endif
2451#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
2452 {
2453 .procname = "ignore-unaligned-usertrap",
2454 .data = &no_unaligned_warning,
2455 .maxlen = sizeof (int),
2456 .mode = 0644,
2457 .proc_handler = proc_dointvec,
2458 },
2459#endif
2460#ifdef CONFIG_IA64
2461 {
2462 .procname = "unaligned-dump-stack",
2463 .data = &unaligned_dump_stack,
2464 .maxlen = sizeof (int),
2465 .mode = 0644,
2466 .proc_handler = proc_dointvec,
2467 },
2468#endif
2469#ifdef CONFIG_DETECT_HUNG_TASK
2470#ifdef CONFIG_SMP
2471 {
2472 .procname = "hung_task_all_cpu_backtrace",
2473 .data = &sysctl_hung_task_all_cpu_backtrace,
2474 .maxlen = sizeof(int),
2475 .mode = 0644,
2476 .proc_handler = proc_dointvec_minmax,
2477 .extra1 = SYSCTL_ZERO,
2478 .extra2 = SYSCTL_ONE,
2479 },
2480#endif /* CONFIG_SMP */
2481 {
2482 .procname = "hung_task_panic",
2483 .data = &sysctl_hung_task_panic,
2484 .maxlen = sizeof(int),
2485 .mode = 0644,
2486 .proc_handler = proc_dointvec_minmax,
2487 .extra1 = SYSCTL_ZERO,
2488 .extra2 = SYSCTL_ONE,
2489 },
2490 {
2491 .procname = "hung_task_check_count",
2492 .data = &sysctl_hung_task_check_count,
2493 .maxlen = sizeof(int),
2494 .mode = 0644,
2495 .proc_handler = proc_dointvec_minmax,
2496 .extra1 = SYSCTL_ZERO,
2497 },
2498 {
2499 .procname = "hung_task_timeout_secs",
2500 .data = &sysctl_hung_task_timeout_secs,
2501 .maxlen = sizeof(unsigned long),
2502 .mode = 0644,
2503 .proc_handler = proc_dohung_task_timeout_secs,
2504 .extra2 = &hung_task_timeout_max,
2505 },
2506 {
2507 .procname = "hung_task_check_interval_secs",
2508 .data = &sysctl_hung_task_check_interval_secs,
2509 .maxlen = sizeof(unsigned long),
2510 .mode = 0644,
2511 .proc_handler = proc_dohung_task_timeout_secs,
2512 .extra2 = &hung_task_timeout_max,
2513 },
2514 {
2515 .procname = "hung_task_warnings",
2516 .data = &sysctl_hung_task_warnings,
2517 .maxlen = sizeof(int),
2518 .mode = 0644,
2519 .proc_handler = proc_dointvec_minmax,
2520 .extra1 = &neg_one,
2521 },
2522#endif
2523#ifdef CONFIG_RT_MUTEXES
2524 {
2525 .procname = "max_lock_depth",
2526 .data = &max_lock_depth,
2527 .maxlen = sizeof(int),
2528 .mode = 0644,
2529 .proc_handler = proc_dointvec,
2530 },
2531#endif
2532 {
2533 .procname = "poweroff_cmd",
2534 .data = &poweroff_cmd,
2535 .maxlen = POWEROFF_CMD_PATH_LEN,
2536 .mode = 0644,
2537 .proc_handler = proc_dostring,
2538 },
2539#ifdef CONFIG_KEYS
2540 {
2541 .procname = "keys",
2542 .mode = 0555,
2543 .child = key_sysctls,
2544 },
2545#endif
2546#ifdef CONFIG_PERF_EVENTS
2547 /*
2548 * User-space scripts rely on the existence of this file
2549 * as a feature check for perf_events being enabled.
2550 *
2551 * So it's an ABI, do not remove!
2552 */
2553 {
2554 .procname = "perf_event_paranoid",
2555 .data = &sysctl_perf_event_paranoid,
2556 .maxlen = sizeof(sysctl_perf_event_paranoid),
2557 .mode = 0644,
2558 .proc_handler = proc_dointvec,
2559 },
2560 {
2561 .procname = "perf_event_mlock_kb",
2562 .data = &sysctl_perf_event_mlock,
2563 .maxlen = sizeof(sysctl_perf_event_mlock),
2564 .mode = 0644,
2565 .proc_handler = proc_dointvec,
2566 },
2567 {
2568 .procname = "perf_event_max_sample_rate",
2569 .data = &sysctl_perf_event_sample_rate,
2570 .maxlen = sizeof(sysctl_perf_event_sample_rate),
2571 .mode = 0644,
2572 .proc_handler = perf_proc_update_handler,
2573 .extra1 = SYSCTL_ONE,
2574 },
2575 {
2576 .procname = "perf_cpu_time_max_percent",
2577 .data = &sysctl_perf_cpu_time_max_percent,
2578 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
2579 .mode = 0644,
2580 .proc_handler = perf_cpu_time_max_percent_handler,
2581 .extra1 = SYSCTL_ZERO,
2582 .extra2 = &one_hundred,
2583 },
2584 {
2585 .procname = "perf_event_max_stack",
2586 .data = &sysctl_perf_event_max_stack,
2587 .maxlen = sizeof(sysctl_perf_event_max_stack),
2588 .mode = 0644,
2589 .proc_handler = perf_event_max_stack_handler,
2590 .extra1 = SYSCTL_ZERO,
2591 .extra2 = &six_hundred_forty_kb,
2592 },
2593 {
2594 .procname = "perf_event_max_contexts_per_stack",
2595 .data = &sysctl_perf_event_max_contexts_per_stack,
2596 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
2597 .mode = 0644,
2598 .proc_handler = perf_event_max_stack_handler,
2599 .extra1 = SYSCTL_ZERO,
2600 .extra2 = &one_thousand,
2601 },
2602#endif
2603 {
2604 .procname = "panic_on_warn",
2605 .data = &panic_on_warn,
2606 .maxlen = sizeof(int),
2607 .mode = 0644,
2608 .proc_handler = proc_dointvec_minmax,
2609 .extra1 = SYSCTL_ZERO,
2610 .extra2 = SYSCTL_ONE,
2611 },
2612#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
2613 {
2614 .procname = "timer_migration",
2615 .data = &sysctl_timer_migration,
2616 .maxlen = sizeof(unsigned int),
2617 .mode = 0644,
2618 .proc_handler = timer_migration_handler,
2619 .extra1 = SYSCTL_ZERO,
2620 .extra2 = SYSCTL_ONE,
2621 },
2622#endif
2623#ifdef CONFIG_BPF_SYSCALL
2624 {
2625 .procname = "unprivileged_bpf_disabled",
2626 .data = &sysctl_unprivileged_bpf_disabled,
2627 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
2628 .mode = 0644,
2629 /* only handle a transition from default "0" to "1" */
2630 .proc_handler = proc_dointvec_minmax,
2631 .extra1 = SYSCTL_ONE,
2632 .extra2 = SYSCTL_ONE,
2633 },
2634 {
2635 .procname = "bpf_stats_enabled",
2636 .data = &bpf_stats_enabled_key.key,
2637 .maxlen = sizeof(bpf_stats_enabled_key),
2638 .mode = 0644,
2639 .proc_handler = bpf_stats_handler,
2640 },
2641#endif
2642#if defined(CONFIG_TREE_RCU)
2643 {
2644 .procname = "panic_on_rcu_stall",
2645 .data = &sysctl_panic_on_rcu_stall,
2646 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2647 .mode = 0644,
2648 .proc_handler = proc_dointvec_minmax,
2649 .extra1 = SYSCTL_ZERO,
2650 .extra2 = SYSCTL_ONE,
2651 },
2652#endif
2653#ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
2654 {
2655 .procname = "stack_erasing",
2656 .data = NULL,
2657 .maxlen = sizeof(int),
2658 .mode = 0600,
2659 .proc_handler = stack_erasing_sysctl,
2660 .extra1 = SYSCTL_ZERO,
2661 .extra2 = SYSCTL_ONE,
2662 },
2663#endif
2664 { }
2665};
2666
2667static struct ctl_table vm_table[] = {
2668 {
2669 .procname = "overcommit_memory",
2670 .data = &sysctl_overcommit_memory,
2671 .maxlen = sizeof(sysctl_overcommit_memory),
2672 .mode = 0644,
2673 .proc_handler = overcommit_policy_handler,
2674 .extra1 = SYSCTL_ZERO,
2675 .extra2 = &two,
2676 },
2677 {
2678 .procname = "panic_on_oom",
2679 .data = &sysctl_panic_on_oom,
2680 .maxlen = sizeof(sysctl_panic_on_oom),
2681 .mode = 0644,
2682 .proc_handler = proc_dointvec_minmax,
2683 .extra1 = SYSCTL_ZERO,
2684 .extra2 = &two,
2685 },
2686 {
2687 .procname = "oom_kill_allocating_task",
2688 .data = &sysctl_oom_kill_allocating_task,
2689 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
2690 .mode = 0644,
2691 .proc_handler = proc_dointvec,
2692 },
2693 {
2694 .procname = "oom_dump_tasks",
2695 .data = &sysctl_oom_dump_tasks,
2696 .maxlen = sizeof(sysctl_oom_dump_tasks),
2697 .mode = 0644,
2698 .proc_handler = proc_dointvec,
2699 },
2700 {
2701 .procname = "overcommit_ratio",
2702 .data = &sysctl_overcommit_ratio,
2703 .maxlen = sizeof(sysctl_overcommit_ratio),
2704 .mode = 0644,
2705 .proc_handler = overcommit_ratio_handler,
2706 },
2707 {
2708 .procname = "overcommit_kbytes",
2709 .data = &sysctl_overcommit_kbytes,
2710 .maxlen = sizeof(sysctl_overcommit_kbytes),
2711 .mode = 0644,
2712 .proc_handler = overcommit_kbytes_handler,
2713 },
2714 {
2715 .procname = "page-cluster",
2716 .data = &page_cluster,
2717 .maxlen = sizeof(int),
2718 .mode = 0644,
2719 .proc_handler = proc_dointvec_minmax,
2720 .extra1 = SYSCTL_ZERO,
2721 },
2722 {
2723 .procname = "dirty_background_ratio",
2724 .data = &dirty_background_ratio,
2725 .maxlen = sizeof(dirty_background_ratio),
2726 .mode = 0644,
2727 .proc_handler = dirty_background_ratio_handler,
2728 .extra1 = SYSCTL_ZERO,
2729 .extra2 = &one_hundred,
2730 },
2731 {
2732 .procname = "dirty_background_bytes",
2733 .data = &dirty_background_bytes,
2734 .maxlen = sizeof(dirty_background_bytes),
2735 .mode = 0644,
2736 .proc_handler = dirty_background_bytes_handler,
2737 .extra1 = &one_ul,
2738 },
2739 {
2740 .procname = "dirty_ratio",
2741 .data = &vm_dirty_ratio,
2742 .maxlen = sizeof(vm_dirty_ratio),
2743 .mode = 0644,
2744 .proc_handler = dirty_ratio_handler,
2745 .extra1 = SYSCTL_ZERO,
2746 .extra2 = &one_hundred,
2747 },
2748 {
2749 .procname = "dirty_bytes",
2750 .data = &vm_dirty_bytes,
2751 .maxlen = sizeof(vm_dirty_bytes),
2752 .mode = 0644,
2753 .proc_handler = dirty_bytes_handler,
2754 .extra1 = &dirty_bytes_min,
2755 },
2756 {
2757 .procname = "dirty_writeback_centisecs",
2758 .data = &dirty_writeback_interval,
2759 .maxlen = sizeof(dirty_writeback_interval),
2760 .mode = 0644,
2761 .proc_handler = dirty_writeback_centisecs_handler,
2762 },
2763 {
2764 .procname = "dirty_expire_centisecs",
2765 .data = &dirty_expire_interval,
2766 .maxlen = sizeof(dirty_expire_interval),
2767 .mode = 0644,
2768 .proc_handler = proc_dointvec_minmax,
2769 .extra1 = SYSCTL_ZERO,
2770 },
2771 {
2772 .procname = "dirtytime_expire_seconds",
2773 .data = &dirtytime_expire_interval,
2774 .maxlen = sizeof(dirtytime_expire_interval),
2775 .mode = 0644,
2776 .proc_handler = dirtytime_interval_handler,
2777 .extra1 = SYSCTL_ZERO,
2778 },
2779 {
2780 .procname = "swappiness",
2781 .data = &vm_swappiness,
2782 .maxlen = sizeof(vm_swappiness),
2783 .mode = 0644,
2784 .proc_handler = proc_dointvec_minmax,
2785 .extra1 = SYSCTL_ZERO,
2786 .extra2 = &two_hundred,
2787 },
2788#ifdef CONFIG_HUGETLB_PAGE
2789 {
2790 .procname = "nr_hugepages",
2791 .data = NULL,
2792 .maxlen = sizeof(unsigned long),
2793 .mode = 0644,
2794 .proc_handler = hugetlb_sysctl_handler,
2795 },
2796#ifdef CONFIG_NUMA
2797 {
2798 .procname = "nr_hugepages_mempolicy",
2799 .data = NULL,
2800 .maxlen = sizeof(unsigned long),
2801 .mode = 0644,
2802 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
2803 },
2804 {
2805 .procname = "numa_stat",
2806 .data = &sysctl_vm_numa_stat,
2807 .maxlen = sizeof(int),
2808 .mode = 0644,
2809 .proc_handler = sysctl_vm_numa_stat_handler,
2810 .extra1 = SYSCTL_ZERO,
2811 .extra2 = SYSCTL_ONE,
2812 },
2813#endif
2814 {
2815 .procname = "hugetlb_shm_group",
2816 .data = &sysctl_hugetlb_shm_group,
2817 .maxlen = sizeof(gid_t),
2818 .mode = 0644,
2819 .proc_handler = proc_dointvec,
2820 },
2821 {
2822 .procname = "nr_overcommit_hugepages",
2823 .data = NULL,
2824 .maxlen = sizeof(unsigned long),
2825 .mode = 0644,
2826 .proc_handler = hugetlb_overcommit_handler,
2827 },
2828#endif
2829 {
2830 .procname = "lowmem_reserve_ratio",
2831 .data = &sysctl_lowmem_reserve_ratio,
2832 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
2833 .mode = 0644,
2834 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
2835 },
2836 {
2837 .procname = "drop_caches",
2838 .data = &sysctl_drop_caches,
2839 .maxlen = sizeof(int),
2840 .mode = 0200,
2841 .proc_handler = drop_caches_sysctl_handler,
2842 .extra1 = SYSCTL_ONE,
2843 .extra2 = &four,
2844 },
2845#ifdef CONFIG_COMPACTION
2846 {
2847 .procname = "compact_memory",
2848 .data = &sysctl_compact_memory,
2849 .maxlen = sizeof(int),
2850 .mode = 0200,
2851 .proc_handler = sysctl_compaction_handler,
2852 },
2853 {
2854 .procname = "compaction_proactiveness",
2855 .data = &sysctl_compaction_proactiveness,
2856 .maxlen = sizeof(sysctl_compaction_proactiveness),
2857 .mode = 0644,
2858 .proc_handler = proc_dointvec_minmax,
2859 .extra1 = SYSCTL_ZERO,
2860 .extra2 = &one_hundred,
2861 },
2862 {
2863 .procname = "extfrag_threshold",
2864 .data = &sysctl_extfrag_threshold,
2865 .maxlen = sizeof(int),
2866 .mode = 0644,
2867 .proc_handler = proc_dointvec_minmax,
2868 .extra1 = &min_extfrag_threshold,
2869 .extra2 = &max_extfrag_threshold,
2870 },
2871 {
2872 .procname = "compact_unevictable_allowed",
2873 .data = &sysctl_compact_unevictable_allowed,
2874 .maxlen = sizeof(int),
2875 .mode = 0644,
2876 .proc_handler = proc_dointvec_minmax_warn_RT_change,
2877 .extra1 = SYSCTL_ZERO,
2878 .extra2 = SYSCTL_ONE,
2879 },
2880
2881#endif /* CONFIG_COMPACTION */
2882 {
2883 .procname = "min_free_kbytes",
2884 .data = &min_free_kbytes,
2885 .maxlen = sizeof(min_free_kbytes),
2886 .mode = 0644,
2887 .proc_handler = min_free_kbytes_sysctl_handler,
2888 .extra1 = SYSCTL_ZERO,
2889 },
2890 {
2891 .procname = "watermark_boost_factor",
2892 .data = &watermark_boost_factor,
2893 .maxlen = sizeof(watermark_boost_factor),
2894 .mode = 0644,
2895 .proc_handler = proc_dointvec_minmax,
2896 .extra1 = SYSCTL_ZERO,
2897 },
2898 {
2899 .procname = "watermark_scale_factor",
2900 .data = &watermark_scale_factor,
2901 .maxlen = sizeof(watermark_scale_factor),
2902 .mode = 0644,
2903 .proc_handler = watermark_scale_factor_sysctl_handler,
2904 .extra1 = SYSCTL_ONE,
2905 .extra2 = &one_thousand,
2906 },
2907 {
2908 .procname = "percpu_pagelist_fraction",
2909 .data = &percpu_pagelist_fraction,
2910 .maxlen = sizeof(percpu_pagelist_fraction),
2911 .mode = 0644,
2912 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
2913 .extra1 = SYSCTL_ZERO,
2914 },
2915 {
2916 .procname = "page_lock_unfairness",
2917 .data = &sysctl_page_lock_unfairness,
2918 .maxlen = sizeof(sysctl_page_lock_unfairness),
2919 .mode = 0644,
2920 .proc_handler = proc_dointvec_minmax,
2921 .extra1 = SYSCTL_ZERO,
2922 },
2923#ifdef CONFIG_MMU
2924 {
2925 .procname = "max_map_count",
2926 .data = &sysctl_max_map_count,
2927 .maxlen = sizeof(sysctl_max_map_count),
2928 .mode = 0644,
2929 .proc_handler = proc_dointvec_minmax,
2930 .extra1 = SYSCTL_ZERO,
2931 },
2932#else
2933 {
2934 .procname = "nr_trim_pages",
2935 .data = &sysctl_nr_trim_pages,
2936 .maxlen = sizeof(sysctl_nr_trim_pages),
2937 .mode = 0644,
2938 .proc_handler = proc_dointvec_minmax,
2939 .extra1 = SYSCTL_ZERO,
2940 },
2941#endif
2942 {
2943 .procname = "laptop_mode",
2944 .data = &laptop_mode,
2945 .maxlen = sizeof(laptop_mode),
2946 .mode = 0644,
2947 .proc_handler = proc_dointvec_jiffies,
2948 },
2949 {
2950 .procname = "block_dump",
2951 .data = &block_dump,
2952 .maxlen = sizeof(block_dump),
2953 .mode = 0644,
2954 .proc_handler = proc_dointvec,
2955 .extra1 = SYSCTL_ZERO,
2956 },
2957 {
2958 .procname = "vfs_cache_pressure",
2959 .data = &sysctl_vfs_cache_pressure,
2960 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2961 .mode = 0644,
2962 .proc_handler = proc_dointvec,
2963 .extra1 = SYSCTL_ZERO,
2964 },
2965#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2966 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2967 {
2968 .procname = "legacy_va_layout",
2969 .data = &sysctl_legacy_va_layout,
2970 .maxlen = sizeof(sysctl_legacy_va_layout),
2971 .mode = 0644,
2972 .proc_handler = proc_dointvec,
2973 .extra1 = SYSCTL_ZERO,
2974 },
2975#endif
2976#ifdef CONFIG_NUMA
2977 {
2978 .procname = "zone_reclaim_mode",
2979 .data = &node_reclaim_mode,
2980 .maxlen = sizeof(node_reclaim_mode),
2981 .mode = 0644,
2982 .proc_handler = proc_dointvec,
2983 .extra1 = SYSCTL_ZERO,
2984 },
2985 {
2986 .procname = "min_unmapped_ratio",
2987 .data = &sysctl_min_unmapped_ratio,
2988 .maxlen = sizeof(sysctl_min_unmapped_ratio),
2989 .mode = 0644,
2990 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
2991 .extra1 = SYSCTL_ZERO,
2992 .extra2 = &one_hundred,
2993 },
2994 {
2995 .procname = "min_slab_ratio",
2996 .data = &sysctl_min_slab_ratio,
2997 .maxlen = sizeof(sysctl_min_slab_ratio),
2998 .mode = 0644,
2999 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
3000 .extra1 = SYSCTL_ZERO,
3001 .extra2 = &one_hundred,
3002 },
3003#endif
3004#ifdef CONFIG_SMP
3005 {
3006 .procname = "stat_interval",
3007 .data = &sysctl_stat_interval,
3008 .maxlen = sizeof(sysctl_stat_interval),
3009 .mode = 0644,
3010 .proc_handler = proc_dointvec_jiffies,
3011 },
3012 {
3013 .procname = "stat_refresh",
3014 .data = NULL,
3015 .maxlen = 0,
3016 .mode = 0600,
3017 .proc_handler = vmstat_refresh,
3018 },
3019#endif
3020#ifdef CONFIG_MMU
3021 {
3022 .procname = "mmap_min_addr",
3023 .data = &dac_mmap_min_addr,
3024 .maxlen = sizeof(unsigned long),
3025 .mode = 0644,
3026 .proc_handler = mmap_min_addr_handler,
3027 },
3028#endif
3029#ifdef CONFIG_NUMA
3030 {
3031 .procname = "numa_zonelist_order",
3032 .data = &numa_zonelist_order,
3033 .maxlen = NUMA_ZONELIST_ORDER_LEN,
3034 .mode = 0644,
3035 .proc_handler = numa_zonelist_order_handler,
3036 },
3037#endif
3038#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
3039 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
3040 {
3041 .procname = "vdso_enabled",
3042#ifdef CONFIG_X86_32
3043 .data = &vdso32_enabled,
3044 .maxlen = sizeof(vdso32_enabled),
3045#else
3046 .data = &vdso_enabled,
3047 .maxlen = sizeof(vdso_enabled),
3048#endif
3049 .mode = 0644,
3050 .proc_handler = proc_dointvec,
3051 .extra1 = SYSCTL_ZERO,
3052 },
3053#endif
3054#ifdef CONFIG_HIGHMEM
3055 {
3056 .procname = "highmem_is_dirtyable",
3057 .data = &vm_highmem_is_dirtyable,
3058 .maxlen = sizeof(vm_highmem_is_dirtyable),
3059 .mode = 0644,
3060 .proc_handler = proc_dointvec_minmax,
3061 .extra1 = SYSCTL_ZERO,
3062 .extra2 = SYSCTL_ONE,
3063 },
3064#endif
3065#ifdef CONFIG_MEMORY_FAILURE
3066 {
3067 .procname = "memory_failure_early_kill",
3068 .data = &sysctl_memory_failure_early_kill,
3069 .maxlen = sizeof(sysctl_memory_failure_early_kill),
3070 .mode = 0644,
3071 .proc_handler = proc_dointvec_minmax,
3072 .extra1 = SYSCTL_ZERO,
3073 .extra2 = SYSCTL_ONE,
3074 },
3075 {
3076 .procname = "memory_failure_recovery",
3077 .data = &sysctl_memory_failure_recovery,
3078 .maxlen = sizeof(sysctl_memory_failure_recovery),
3079 .mode = 0644,
3080 .proc_handler = proc_dointvec_minmax,
3081 .extra1 = SYSCTL_ZERO,
3082 .extra2 = SYSCTL_ONE,
3083 },
3084#endif
3085 {
3086 .procname = "user_reserve_kbytes",
3087 .data = &sysctl_user_reserve_kbytes,
3088 .maxlen = sizeof(sysctl_user_reserve_kbytes),
3089 .mode = 0644,
3090 .proc_handler = proc_doulongvec_minmax,
3091 },
3092 {
3093 .procname = "admin_reserve_kbytes",
3094 .data = &sysctl_admin_reserve_kbytes,
3095 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
3096 .mode = 0644,
3097 .proc_handler = proc_doulongvec_minmax,
3098 },
3099#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
3100 {
3101 .procname = "mmap_rnd_bits",
3102 .data = &mmap_rnd_bits,
3103 .maxlen = sizeof(mmap_rnd_bits),
3104 .mode = 0600,
3105 .proc_handler = proc_dointvec_minmax,
3106 .extra1 = (void *)&mmap_rnd_bits_min,
3107 .extra2 = (void *)&mmap_rnd_bits_max,
3108 },
3109#endif
3110#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
3111 {
3112 .procname = "mmap_rnd_compat_bits",
3113 .data = &mmap_rnd_compat_bits,
3114 .maxlen = sizeof(mmap_rnd_compat_bits),
3115 .mode = 0600,
3116 .proc_handler = proc_dointvec_minmax,
3117 .extra1 = (void *)&mmap_rnd_compat_bits_min,
3118 .extra2 = (void *)&mmap_rnd_compat_bits_max,
3119 },
3120#endif
3121#ifdef CONFIG_USERFAULTFD
3122 {
3123 .procname = "unprivileged_userfaultfd",
3124 .data = &sysctl_unprivileged_userfaultfd,
3125 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
3126 .mode = 0644,
3127 .proc_handler = proc_dointvec_minmax,
3128 .extra1 = SYSCTL_ZERO,
3129 .extra2 = SYSCTL_ONE,
3130 },
3131#endif
3132 { }
3133};
3134
3135static struct ctl_table fs_table[] = {
3136 {
3137 .procname = "inode-nr",
3138 .data = &inodes_stat,
3139 .maxlen = 2*sizeof(long),
3140 .mode = 0444,
3141 .proc_handler = proc_nr_inodes,
3142 },
3143 {
3144 .procname = "inode-state",
3145 .data = &inodes_stat,
3146 .maxlen = 7*sizeof(long),
3147 .mode = 0444,
3148 .proc_handler = proc_nr_inodes,
3149 },
3150 {
3151 .procname = "file-nr",
3152 .data = &files_stat,
3153 .maxlen = sizeof(files_stat),
3154 .mode = 0444,
3155 .proc_handler = proc_nr_files,
3156 },
3157 {
3158 .procname = "file-max",
3159 .data = &files_stat.max_files,
3160 .maxlen = sizeof(files_stat.max_files),
3161 .mode = 0644,
3162 .proc_handler = proc_doulongvec_minmax,
3163 .extra1 = &zero_ul,
3164 .extra2 = &long_max,
3165 },
3166 {
3167 .procname = "nr_open",
3168 .data = &sysctl_nr_open,
3169 .maxlen = sizeof(unsigned int),
3170 .mode = 0644,
3171 .proc_handler = proc_dointvec_minmax,
3172 .extra1 = &sysctl_nr_open_min,
3173 .extra2 = &sysctl_nr_open_max,
3174 },
3175 {
3176 .procname = "dentry-state",
3177 .data = &dentry_stat,
3178 .maxlen = 6*sizeof(long),
3179 .mode = 0444,
3180 .proc_handler = proc_nr_dentry,
3181 },
3182 {
3183 .procname = "overflowuid",
3184 .data = &fs_overflowuid,
3185 .maxlen = sizeof(int),
3186 .mode = 0644,
3187 .proc_handler = proc_dointvec_minmax,
3188 .extra1 = &minolduid,
3189 .extra2 = &maxolduid,
3190 },
3191 {
3192 .procname = "overflowgid",
3193 .data = &fs_overflowgid,
3194 .maxlen = sizeof(int),
3195 .mode = 0644,
3196 .proc_handler = proc_dointvec_minmax,
3197 .extra1 = &minolduid,
3198 .extra2 = &maxolduid,
3199 },
3200#ifdef CONFIG_FILE_LOCKING
3201 {
3202 .procname = "leases-enable",
3203 .data = &leases_enable,
3204 .maxlen = sizeof(int),
3205 .mode = 0644,
3206 .proc_handler = proc_dointvec,
3207 },
3208#endif
3209#ifdef CONFIG_DNOTIFY
3210 {
3211 .procname = "dir-notify-enable",
3212 .data = &dir_notify_enable,
3213 .maxlen = sizeof(int),
3214 .mode = 0644,
3215 .proc_handler = proc_dointvec,
3216 },
3217#endif
3218#ifdef CONFIG_MMU
3219#ifdef CONFIG_FILE_LOCKING
3220 {
3221 .procname = "lease-break-time",
3222 .data = &lease_break_time,
3223 .maxlen = sizeof(int),
3224 .mode = 0644,
3225 .proc_handler = proc_dointvec,
3226 },
3227#endif
3228#ifdef CONFIG_AIO
3229 {
3230 .procname = "aio-nr",
3231 .data = &aio_nr,
3232 .maxlen = sizeof(aio_nr),
3233 .mode = 0444,
3234 .proc_handler = proc_doulongvec_minmax,
3235 },
3236 {
3237 .procname = "aio-max-nr",
3238 .data = &aio_max_nr,
3239 .maxlen = sizeof(aio_max_nr),
3240 .mode = 0644,
3241 .proc_handler = proc_doulongvec_minmax,
3242 },
3243#endif /* CONFIG_AIO */
3244#ifdef CONFIG_INOTIFY_USER
3245 {
3246 .procname = "inotify",
3247 .mode = 0555,
3248 .child = inotify_table,
3249 },
3250#endif
3251#ifdef CONFIG_EPOLL
3252 {
3253 .procname = "epoll",
3254 .mode = 0555,
3255 .child = epoll_table,
3256 },
3257#endif
3258#endif
3259 {
3260 .procname = "protected_symlinks",
3261 .data = &sysctl_protected_symlinks,
3262 .maxlen = sizeof(int),
3263 .mode = 0600,
3264 .proc_handler = proc_dointvec_minmax,
3265 .extra1 = SYSCTL_ZERO,
3266 .extra2 = SYSCTL_ONE,
3267 },
3268 {
3269 .procname = "protected_hardlinks",
3270 .data = &sysctl_protected_hardlinks,
3271 .maxlen = sizeof(int),
3272 .mode = 0600,
3273 .proc_handler = proc_dointvec_minmax,
3274 .extra1 = SYSCTL_ZERO,
3275 .extra2 = SYSCTL_ONE,
3276 },
3277 {
3278 .procname = "protected_fifos",
3279 .data = &sysctl_protected_fifos,
3280 .maxlen = sizeof(int),
3281 .mode = 0600,
3282 .proc_handler = proc_dointvec_minmax,
3283 .extra1 = SYSCTL_ZERO,
3284 .extra2 = &two,
3285 },
3286 {
3287 .procname = "protected_regular",
3288 .data = &sysctl_protected_regular,
3289 .maxlen = sizeof(int),
3290 .mode = 0600,
3291 .proc_handler = proc_dointvec_minmax,
3292 .extra1 = SYSCTL_ZERO,
3293 .extra2 = &two,
3294 },
3295 {
3296 .procname = "suid_dumpable",
3297 .data = &suid_dumpable,
3298 .maxlen = sizeof(int),
3299 .mode = 0644,
3300 .proc_handler = proc_dointvec_minmax_coredump,
3301 .extra1 = SYSCTL_ZERO,
3302 .extra2 = &two,
3303 },
3304#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
3305 {
3306 .procname = "binfmt_misc",
3307 .mode = 0555,
3308 .child = sysctl_mount_point,
3309 },
3310#endif
3311 {
3312 .procname = "pipe-max-size",
3313 .data = &pipe_max_size,
3314 .maxlen = sizeof(pipe_max_size),
3315 .mode = 0644,
3316 .proc_handler = proc_dopipe_max_size,
3317 },
3318 {
3319 .procname = "pipe-user-pages-hard",
3320 .data = &pipe_user_pages_hard,
3321 .maxlen = sizeof(pipe_user_pages_hard),
3322 .mode = 0644,
3323 .proc_handler = proc_doulongvec_minmax,
3324 },
3325 {
3326 .procname = "pipe-user-pages-soft",
3327 .data = &pipe_user_pages_soft,
3328 .maxlen = sizeof(pipe_user_pages_soft),
3329 .mode = 0644,
3330 .proc_handler = proc_doulongvec_minmax,
3331 },
3332 {
3333 .procname = "mount-max",
3334 .data = &sysctl_mount_max,
3335 .maxlen = sizeof(unsigned int),
3336 .mode = 0644,
3337 .proc_handler = proc_dointvec_minmax,
3338 .extra1 = SYSCTL_ONE,
3339 },
3340 { }
3341};
3342
3343static struct ctl_table debug_table[] = {
3344#ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
3345 {
3346 .procname = "exception-trace",
3347 .data = &show_unhandled_signals,
3348 .maxlen = sizeof(int),
3349 .mode = 0644,
3350 .proc_handler = proc_dointvec
3351 },
3352#endif
3353#if defined(CONFIG_OPTPROBES)
3354 {
3355 .procname = "kprobes-optimization",
3356 .data = &sysctl_kprobes_optimization,
3357 .maxlen = sizeof(int),
3358 .mode = 0644,
3359 .proc_handler = proc_kprobes_optimization_handler,
3360 .extra1 = SYSCTL_ZERO,
3361 .extra2 = SYSCTL_ONE,
3362 },
3363#endif
3364 { }
3365};
3366
3367static struct ctl_table dev_table[] = {
3368 { }
3369};
3370
3371static struct ctl_table sysctl_base_table[] = {
3372 {
3373 .procname = "kernel",
3374 .mode = 0555,
3375 .child = kern_table,
3376 },
3377 {
3378 .procname = "vm",
3379 .mode = 0555,
3380 .child = vm_table,
3381 },
3382 {
3383 .procname = "fs",
3384 .mode = 0555,
3385 .child = fs_table,
3386 },
3387 {
3388 .procname = "debug",
3389 .mode = 0555,
3390 .child = debug_table,
3391 },
3392 {
3393 .procname = "dev",
3394 .mode = 0555,
3395 .child = dev_table,
3396 },
3397 { }
3398};
3399
3400int __init sysctl_init(void)
3401{
3402 struct ctl_table_header *hdr;
3403
3404 hdr = register_sysctl_table(sysctl_base_table);
3405 kmemleak_not_leak(hdr);
3406 return 0;
3407}
3408#endif /* CONFIG_SYSCTL */
3409/*
3410 * No sense putting this after each symbol definition, twice,
3411 * exception granted :-)
3412 */
3413EXPORT_SYMBOL(proc_dointvec);
3414EXPORT_SYMBOL(proc_douintvec);
3415EXPORT_SYMBOL(proc_dointvec_jiffies);
3416EXPORT_SYMBOL(proc_dointvec_minmax);
3417EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3418EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3419EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3420EXPORT_SYMBOL(proc_dostring);
3421EXPORT_SYMBOL(proc_doulongvec_minmax);
3422EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3423EXPORT_SYMBOL(proc_do_large_bitmap);