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