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/mm.h>
24#include <linux/swap.h>
25#include <linux/slab.h>
26#include <linux/sysctl.h>
27#include <linux/bitmap.h>
28#include <linux/signal.h>
29#include <linux/panic.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/filter.h>
36#include <linux/fs.h>
37#include <linux/init.h>
38#include <linux/kernel.h>
39#include <linux/kobject.h>
40#include <linux/net.h>
41#include <linux/sysrq.h>
42#include <linux/highuid.h>
43#include <linux/writeback.h>
44#include <linux/ratelimit.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/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/oom.h>
59#include <linux/kmod.h>
60#include <linux/capability.h>
61#include <linux/binfmts.h>
62#include <linux/sched/sysctl.h>
63#include <linux/mount.h>
64#include <linux/userfaultfd_k.h>
65#include <linux/pid.h>
66
67#include "../lib/kstrtox.h"
68
69#include <linux/uaccess.h>
70#include <asm/processor.h>
71
72#ifdef CONFIG_X86
73#include <asm/nmi.h>
74#include <asm/stacktrace.h>
75#include <asm/io.h>
76#endif
77#ifdef CONFIG_SPARC
78#include <asm/setup.h>
79#endif
80#ifdef CONFIG_RT_MUTEXES
81#include <linux/rtmutex.h>
82#endif
83
84/* shared constants to be used in various sysctls */
85const int sysctl_vals[] = { 0, 1, 2, 3, 4, 100, 200, 1000, 3000, INT_MAX, 65535, -1 };
86EXPORT_SYMBOL(sysctl_vals);
87
88const unsigned long sysctl_long_vals[] = { 0, 1, LONG_MAX };
89EXPORT_SYMBOL_GPL(sysctl_long_vals);
90
91#if defined(CONFIG_SYSCTL)
92
93/* Constants used for minimum and maximum */
94
95#ifdef CONFIG_PERF_EVENTS
96static const int six_hundred_forty_kb = 640 * 1024;
97#endif
98
99
100static const int ngroups_max = NGROUPS_MAX;
101static const int cap_last_cap = CAP_LAST_CAP;
102
103#ifdef CONFIG_PROC_SYSCTL
104
105/**
106 * enum sysctl_writes_mode - supported sysctl write modes
107 *
108 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
109 * to be written, and multiple writes on the same sysctl file descriptor
110 * will rewrite the sysctl value, regardless of file position. No warning
111 * is issued when the initial position is not 0.
112 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
113 * not 0.
114 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
115 * file position 0 and the value must be fully contained in the buffer
116 * sent to the write syscall. If dealing with strings respect the file
117 * position, but restrict this to the max length of the buffer, anything
118 * passed the max length will be ignored. Multiple writes will append
119 * to the buffer.
120 *
121 * These write modes control how current file position affects the behavior of
122 * updating sysctl values through the proc interface on each write.
123 */
124enum sysctl_writes_mode {
125 SYSCTL_WRITES_LEGACY = -1,
126 SYSCTL_WRITES_WARN = 0,
127 SYSCTL_WRITES_STRICT = 1,
128};
129
130static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
131#endif /* CONFIG_PROC_SYSCTL */
132
133#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
134 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
135int sysctl_legacy_va_layout;
136#endif
137
138#endif /* CONFIG_SYSCTL */
139
140/*
141 * /proc/sys support
142 */
143
144#ifdef CONFIG_PROC_SYSCTL
145
146static int _proc_do_string(char *data, int maxlen, int write,
147 char *buffer, size_t *lenp, loff_t *ppos)
148{
149 size_t len;
150 char c, *p;
151
152 if (!data || !maxlen || !*lenp) {
153 *lenp = 0;
154 return 0;
155 }
156
157 if (write) {
158 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
159 /* Only continue writes not past the end of buffer. */
160 len = strlen(data);
161 if (len > maxlen - 1)
162 len = maxlen - 1;
163
164 if (*ppos > len)
165 return 0;
166 len = *ppos;
167 } else {
168 /* Start writing from beginning of buffer. */
169 len = 0;
170 }
171
172 *ppos += *lenp;
173 p = buffer;
174 while ((p - buffer) < *lenp && len < maxlen - 1) {
175 c = *(p++);
176 if (c == 0 || c == '\n')
177 break;
178 data[len++] = c;
179 }
180 data[len] = 0;
181 } else {
182 len = strlen(data);
183 if (len > maxlen)
184 len = maxlen;
185
186 if (*ppos > len) {
187 *lenp = 0;
188 return 0;
189 }
190
191 data += *ppos;
192 len -= *ppos;
193
194 if (len > *lenp)
195 len = *lenp;
196 if (len)
197 memcpy(buffer, data, len);
198 if (len < *lenp) {
199 buffer[len] = '\n';
200 len++;
201 }
202 *lenp = len;
203 *ppos += len;
204 }
205 return 0;
206}
207
208static void warn_sysctl_write(const struct ctl_table *table)
209{
210 pr_warn_once("%s wrote to %s when file position was not 0!\n"
211 "This will not be supported in the future. To silence this\n"
212 "warning, set kernel.sysctl_writes_strict = -1\n",
213 current->comm, table->procname);
214}
215
216/**
217 * proc_first_pos_non_zero_ignore - check if first position is allowed
218 * @ppos: file position
219 * @table: the sysctl table
220 *
221 * Returns true if the first position is non-zero and the sysctl_writes_strict
222 * mode indicates this is not allowed for numeric input types. String proc
223 * handlers can ignore the return value.
224 */
225static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
226 const struct ctl_table *table)
227{
228 if (!*ppos)
229 return false;
230
231 switch (sysctl_writes_strict) {
232 case SYSCTL_WRITES_STRICT:
233 return true;
234 case SYSCTL_WRITES_WARN:
235 warn_sysctl_write(table);
236 return false;
237 default:
238 return false;
239 }
240}
241
242/**
243 * proc_dostring - read a string sysctl
244 * @table: the sysctl table
245 * @write: %TRUE if this is a write to the sysctl file
246 * @buffer: the user buffer
247 * @lenp: the size of the user buffer
248 * @ppos: file position
249 *
250 * Reads/writes a string from/to the user buffer. If the kernel
251 * buffer provided is not large enough to hold the string, the
252 * string is truncated. The copied string is %NULL-terminated.
253 * If the string is being read by the user process, it is copied
254 * and a newline '\n' is added. It is truncated if the buffer is
255 * not large enough.
256 *
257 * Returns 0 on success.
258 */
259int proc_dostring(const struct ctl_table *table, int write,
260 void *buffer, size_t *lenp, loff_t *ppos)
261{
262 if (write)
263 proc_first_pos_non_zero_ignore(ppos, table);
264
265 return _proc_do_string(table->data, table->maxlen, write, buffer, lenp,
266 ppos);
267}
268
269static void proc_skip_spaces(char **buf, size_t *size)
270{
271 while (*size) {
272 if (!isspace(**buf))
273 break;
274 (*size)--;
275 (*buf)++;
276 }
277}
278
279static void proc_skip_char(char **buf, size_t *size, const char v)
280{
281 while (*size) {
282 if (**buf != v)
283 break;
284 (*size)--;
285 (*buf)++;
286 }
287}
288
289/**
290 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
291 * fail on overflow
292 *
293 * @cp: kernel buffer containing the string to parse
294 * @endp: pointer to store the trailing characters
295 * @base: the base to use
296 * @res: where the parsed integer will be stored
297 *
298 * In case of success 0 is returned and @res will contain the parsed integer,
299 * @endp will hold any trailing characters.
300 * This function will fail the parse on overflow. If there wasn't an overflow
301 * the function will defer the decision what characters count as invalid to the
302 * caller.
303 */
304static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
305 unsigned long *res)
306{
307 unsigned long long result;
308 unsigned int rv;
309
310 cp = _parse_integer_fixup_radix(cp, &base);
311 rv = _parse_integer(cp, base, &result);
312 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
313 return -ERANGE;
314
315 cp += rv;
316
317 if (endp)
318 *endp = (char *)cp;
319
320 *res = (unsigned long)result;
321 return 0;
322}
323
324#define TMPBUFLEN 22
325/**
326 * proc_get_long - reads an ASCII formatted integer from a user buffer
327 *
328 * @buf: a kernel buffer
329 * @size: size of the kernel buffer
330 * @val: this is where the number will be stored
331 * @neg: set to %TRUE if number is negative
332 * @perm_tr: a vector which contains the allowed trailers
333 * @perm_tr_len: size of the perm_tr vector
334 * @tr: pointer to store the trailer character
335 *
336 * In case of success %0 is returned and @buf and @size are updated with
337 * the amount of bytes read. If @tr is non-NULL and a trailing
338 * character exists (size is non-zero after returning from this
339 * function), @tr is updated with the trailing character.
340 */
341static int proc_get_long(char **buf, size_t *size,
342 unsigned long *val, bool *neg,
343 const char *perm_tr, unsigned perm_tr_len, char *tr)
344{
345 char *p, tmp[TMPBUFLEN];
346 ssize_t len = *size;
347
348 if (len <= 0)
349 return -EINVAL;
350
351 if (len > TMPBUFLEN - 1)
352 len = TMPBUFLEN - 1;
353
354 memcpy(tmp, *buf, len);
355
356 tmp[len] = 0;
357 p = tmp;
358 if (*p == '-' && *size > 1) {
359 *neg = true;
360 p++;
361 } else
362 *neg = false;
363 if (!isdigit(*p))
364 return -EINVAL;
365
366 if (strtoul_lenient(p, &p, 0, val))
367 return -EINVAL;
368
369 len = p - tmp;
370
371 /* We don't know if the next char is whitespace thus we may accept
372 * invalid integers (e.g. 1234...a) or two integers instead of one
373 * (e.g. 123...1). So lets not allow such large numbers. */
374 if (len == TMPBUFLEN - 1)
375 return -EINVAL;
376
377 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
378 return -EINVAL;
379
380 if (tr && (len < *size))
381 *tr = *p;
382
383 *buf += len;
384 *size -= len;
385
386 return 0;
387}
388
389/**
390 * proc_put_long - converts an integer to a decimal ASCII formatted string
391 *
392 * @buf: the user buffer
393 * @size: the size of the user buffer
394 * @val: the integer to be converted
395 * @neg: sign of the number, %TRUE for negative
396 *
397 * In case of success @buf and @size are updated with the amount of bytes
398 * written.
399 */
400static void proc_put_long(void **buf, size_t *size, unsigned long val, bool neg)
401{
402 int len;
403 char tmp[TMPBUFLEN], *p = tmp;
404
405 sprintf(p, "%s%lu", neg ? "-" : "", val);
406 len = strlen(tmp);
407 if (len > *size)
408 len = *size;
409 memcpy(*buf, tmp, len);
410 *size -= len;
411 *buf += len;
412}
413#undef TMPBUFLEN
414
415static void proc_put_char(void **buf, size_t *size, char c)
416{
417 if (*size) {
418 char **buffer = (char **)buf;
419 **buffer = c;
420
421 (*size)--;
422 (*buffer)++;
423 *buf = *buffer;
424 }
425}
426
427static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
428 int *valp,
429 int write, void *data)
430{
431 if (write) {
432 if (*negp) {
433 if (*lvalp > (unsigned long) INT_MAX + 1)
434 return -EINVAL;
435 WRITE_ONCE(*valp, -*lvalp);
436 } else {
437 if (*lvalp > (unsigned long) INT_MAX)
438 return -EINVAL;
439 WRITE_ONCE(*valp, *lvalp);
440 }
441 } else {
442 int val = READ_ONCE(*valp);
443 if (val < 0) {
444 *negp = true;
445 *lvalp = -(unsigned long)val;
446 } else {
447 *negp = false;
448 *lvalp = (unsigned long)val;
449 }
450 }
451 return 0;
452}
453
454static int do_proc_douintvec_conv(unsigned long *lvalp,
455 unsigned int *valp,
456 int write, void *data)
457{
458 if (write) {
459 if (*lvalp > UINT_MAX)
460 return -EINVAL;
461 WRITE_ONCE(*valp, *lvalp);
462 } else {
463 unsigned int val = READ_ONCE(*valp);
464 *lvalp = (unsigned long)val;
465 }
466 return 0;
467}
468
469static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
470
471static int __do_proc_dointvec(void *tbl_data, const struct ctl_table *table,
472 int write, void *buffer,
473 size_t *lenp, loff_t *ppos,
474 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
475 int write, void *data),
476 void *data)
477{
478 int *i, vleft, first = 1, err = 0;
479 size_t left;
480 char *p;
481
482 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
483 *lenp = 0;
484 return 0;
485 }
486
487 i = (int *) tbl_data;
488 vleft = table->maxlen / sizeof(*i);
489 left = *lenp;
490
491 if (!conv)
492 conv = do_proc_dointvec_conv;
493
494 if (write) {
495 if (proc_first_pos_non_zero_ignore(ppos, table))
496 goto out;
497
498 if (left > PAGE_SIZE - 1)
499 left = PAGE_SIZE - 1;
500 p = buffer;
501 }
502
503 for (; left && vleft--; i++, first=0) {
504 unsigned long lval;
505 bool neg;
506
507 if (write) {
508 proc_skip_spaces(&p, &left);
509
510 if (!left)
511 break;
512 err = proc_get_long(&p, &left, &lval, &neg,
513 proc_wspace_sep,
514 sizeof(proc_wspace_sep), NULL);
515 if (err)
516 break;
517 if (conv(&neg, &lval, i, 1, data)) {
518 err = -EINVAL;
519 break;
520 }
521 } else {
522 if (conv(&neg, &lval, i, 0, data)) {
523 err = -EINVAL;
524 break;
525 }
526 if (!first)
527 proc_put_char(&buffer, &left, '\t');
528 proc_put_long(&buffer, &left, lval, neg);
529 }
530 }
531
532 if (!write && !first && left && !err)
533 proc_put_char(&buffer, &left, '\n');
534 if (write && !err && left)
535 proc_skip_spaces(&p, &left);
536 if (write && first)
537 return err ? : -EINVAL;
538 *lenp -= left;
539out:
540 *ppos += *lenp;
541 return err;
542}
543
544static int do_proc_dointvec(const struct ctl_table *table, int write,
545 void *buffer, size_t *lenp, loff_t *ppos,
546 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
547 int write, void *data),
548 void *data)
549{
550 return __do_proc_dointvec(table->data, table, write,
551 buffer, lenp, ppos, conv, data);
552}
553
554static int do_proc_douintvec_w(unsigned int *tbl_data,
555 const struct ctl_table *table,
556 void *buffer,
557 size_t *lenp, loff_t *ppos,
558 int (*conv)(unsigned long *lvalp,
559 unsigned int *valp,
560 int write, void *data),
561 void *data)
562{
563 unsigned long lval;
564 int err = 0;
565 size_t left;
566 bool neg;
567 char *p = buffer;
568
569 left = *lenp;
570
571 if (proc_first_pos_non_zero_ignore(ppos, table))
572 goto bail_early;
573
574 if (left > PAGE_SIZE - 1)
575 left = PAGE_SIZE - 1;
576
577 proc_skip_spaces(&p, &left);
578 if (!left) {
579 err = -EINVAL;
580 goto out_free;
581 }
582
583 err = proc_get_long(&p, &left, &lval, &neg,
584 proc_wspace_sep,
585 sizeof(proc_wspace_sep), NULL);
586 if (err || neg) {
587 err = -EINVAL;
588 goto out_free;
589 }
590
591 if (conv(&lval, tbl_data, 1, data)) {
592 err = -EINVAL;
593 goto out_free;
594 }
595
596 if (!err && left)
597 proc_skip_spaces(&p, &left);
598
599out_free:
600 if (err)
601 return -EINVAL;
602
603 return 0;
604
605 /* This is in keeping with old __do_proc_dointvec() */
606bail_early:
607 *ppos += *lenp;
608 return err;
609}
610
611static int do_proc_douintvec_r(unsigned int *tbl_data, void *buffer,
612 size_t *lenp, loff_t *ppos,
613 int (*conv)(unsigned long *lvalp,
614 unsigned int *valp,
615 int write, void *data),
616 void *data)
617{
618 unsigned long lval;
619 int err = 0;
620 size_t left;
621
622 left = *lenp;
623
624 if (conv(&lval, tbl_data, 0, data)) {
625 err = -EINVAL;
626 goto out;
627 }
628
629 proc_put_long(&buffer, &left, lval, false);
630 if (!left)
631 goto out;
632
633 proc_put_char(&buffer, &left, '\n');
634
635out:
636 *lenp -= left;
637 *ppos += *lenp;
638
639 return err;
640}
641
642static int __do_proc_douintvec(void *tbl_data, const struct ctl_table *table,
643 int write, void *buffer,
644 size_t *lenp, loff_t *ppos,
645 int (*conv)(unsigned long *lvalp,
646 unsigned int *valp,
647 int write, void *data),
648 void *data)
649{
650 unsigned int *i, vleft;
651
652 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
653 *lenp = 0;
654 return 0;
655 }
656
657 i = (unsigned int *) tbl_data;
658 vleft = table->maxlen / sizeof(*i);
659
660 /*
661 * Arrays are not supported, keep this simple. *Do not* add
662 * support for them.
663 */
664 if (vleft != 1) {
665 *lenp = 0;
666 return -EINVAL;
667 }
668
669 if (!conv)
670 conv = do_proc_douintvec_conv;
671
672 if (write)
673 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
674 conv, data);
675 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
676}
677
678int do_proc_douintvec(const struct ctl_table *table, int write,
679 void *buffer, size_t *lenp, loff_t *ppos,
680 int (*conv)(unsigned long *lvalp,
681 unsigned int *valp,
682 int write, void *data),
683 void *data)
684{
685 return __do_proc_douintvec(table->data, table, write,
686 buffer, lenp, ppos, conv, data);
687}
688
689/**
690 * proc_dobool - read/write a bool
691 * @table: the sysctl table
692 * @write: %TRUE if this is a write to the sysctl file
693 * @buffer: the user buffer
694 * @lenp: the size of the user buffer
695 * @ppos: file position
696 *
697 * Reads/writes one integer value from/to the user buffer,
698 * treated as an ASCII string.
699 *
700 * table->data must point to a bool variable and table->maxlen must
701 * be sizeof(bool).
702 *
703 * Returns 0 on success.
704 */
705int proc_dobool(const struct ctl_table *table, int write, void *buffer,
706 size_t *lenp, loff_t *ppos)
707{
708 struct ctl_table tmp;
709 bool *data = table->data;
710 int res, val;
711
712 /* Do not support arrays yet. */
713 if (table->maxlen != sizeof(bool))
714 return -EINVAL;
715
716 tmp = *table;
717 tmp.maxlen = sizeof(val);
718 tmp.data = &val;
719
720 val = READ_ONCE(*data);
721 res = proc_dointvec(&tmp, write, buffer, lenp, ppos);
722 if (res)
723 return res;
724 if (write)
725 WRITE_ONCE(*data, val);
726 return 0;
727}
728
729/**
730 * proc_dointvec - read a vector of integers
731 * @table: the sysctl table
732 * @write: %TRUE if this is a write to the sysctl file
733 * @buffer: the user buffer
734 * @lenp: the size of the user buffer
735 * @ppos: file position
736 *
737 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
738 * values from/to the user buffer, treated as an ASCII string.
739 *
740 * Returns 0 on success.
741 */
742int proc_dointvec(const struct ctl_table *table, int write, void *buffer,
743 size_t *lenp, loff_t *ppos)
744{
745 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
746}
747
748/**
749 * proc_douintvec - read a vector of unsigned integers
750 * @table: the sysctl table
751 * @write: %TRUE if this is a write to the sysctl file
752 * @buffer: the user buffer
753 * @lenp: the size of the user buffer
754 * @ppos: file position
755 *
756 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
757 * values from/to the user buffer, treated as an ASCII string.
758 *
759 * Returns 0 on success.
760 */
761int proc_douintvec(const struct ctl_table *table, int write, void *buffer,
762 size_t *lenp, loff_t *ppos)
763{
764 return do_proc_douintvec(table, write, buffer, lenp, ppos,
765 do_proc_douintvec_conv, NULL);
766}
767
768/*
769 * Taint values can only be increased
770 * This means we can safely use a temporary.
771 */
772static int proc_taint(const struct ctl_table *table, int write,
773 void *buffer, size_t *lenp, loff_t *ppos)
774{
775 struct ctl_table t;
776 unsigned long tmptaint = get_taint();
777 int err;
778
779 if (write && !capable(CAP_SYS_ADMIN))
780 return -EPERM;
781
782 t = *table;
783 t.data = &tmptaint;
784 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
785 if (err < 0)
786 return err;
787
788 if (write) {
789 int i;
790
791 /*
792 * If we are relying on panic_on_taint not producing
793 * false positives due to userspace input, bail out
794 * before setting the requested taint flags.
795 */
796 if (panic_on_taint_nousertaint && (tmptaint & panic_on_taint))
797 return -EINVAL;
798
799 /*
800 * Poor man's atomic or. Not worth adding a primitive
801 * to everyone's atomic.h for this
802 */
803 for (i = 0; i < TAINT_FLAGS_COUNT; i++)
804 if ((1UL << i) & tmptaint)
805 add_taint(i, LOCKDEP_STILL_OK);
806 }
807
808 return err;
809}
810
811/**
812 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
813 * @min: pointer to minimum allowable value
814 * @max: pointer to maximum allowable value
815 *
816 * The do_proc_dointvec_minmax_conv_param structure provides the
817 * minimum and maximum values for doing range checking for those sysctl
818 * parameters that use the proc_dointvec_minmax() handler.
819 */
820struct do_proc_dointvec_minmax_conv_param {
821 int *min;
822 int *max;
823};
824
825static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
826 int *valp,
827 int write, void *data)
828{
829 int tmp, ret;
830 struct do_proc_dointvec_minmax_conv_param *param = data;
831 /*
832 * If writing, first do so via a temporary local int so we can
833 * bounds-check it before touching *valp.
834 */
835 int *ip = write ? &tmp : valp;
836
837 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
838 if (ret)
839 return ret;
840
841 if (write) {
842 if ((param->min && *param->min > tmp) ||
843 (param->max && *param->max < tmp))
844 return -EINVAL;
845 WRITE_ONCE(*valp, tmp);
846 }
847
848 return 0;
849}
850
851/**
852 * proc_dointvec_minmax - read a vector of integers with min/max values
853 * @table: the sysctl table
854 * @write: %TRUE if this is a write to the sysctl file
855 * @buffer: the user buffer
856 * @lenp: the size of the user buffer
857 * @ppos: file position
858 *
859 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
860 * values from/to the user buffer, treated as an ASCII string.
861 *
862 * This routine will ensure the values are within the range specified by
863 * table->extra1 (min) and table->extra2 (max).
864 *
865 * Returns 0 on success or -EINVAL on write when the range check fails.
866 */
867int proc_dointvec_minmax(const struct ctl_table *table, int write,
868 void *buffer, size_t *lenp, loff_t *ppos)
869{
870 struct do_proc_dointvec_minmax_conv_param param = {
871 .min = (int *) table->extra1,
872 .max = (int *) table->extra2,
873 };
874 return do_proc_dointvec(table, write, buffer, lenp, ppos,
875 do_proc_dointvec_minmax_conv, ¶m);
876}
877
878/**
879 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
880 * @min: pointer to minimum allowable value
881 * @max: pointer to maximum allowable value
882 *
883 * The do_proc_douintvec_minmax_conv_param structure provides the
884 * minimum and maximum values for doing range checking for those sysctl
885 * parameters that use the proc_douintvec_minmax() handler.
886 */
887struct do_proc_douintvec_minmax_conv_param {
888 unsigned int *min;
889 unsigned int *max;
890};
891
892static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
893 unsigned int *valp,
894 int write, void *data)
895{
896 int ret;
897 unsigned int tmp;
898 struct do_proc_douintvec_minmax_conv_param *param = data;
899 /* write via temporary local uint for bounds-checking */
900 unsigned int *up = write ? &tmp : valp;
901
902 ret = do_proc_douintvec_conv(lvalp, up, write, data);
903 if (ret)
904 return ret;
905
906 if (write) {
907 if ((param->min && *param->min > tmp) ||
908 (param->max && *param->max < tmp))
909 return -ERANGE;
910
911 WRITE_ONCE(*valp, tmp);
912 }
913
914 return 0;
915}
916
917/**
918 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
919 * @table: the sysctl table
920 * @write: %TRUE if this is a write to the sysctl file
921 * @buffer: the user buffer
922 * @lenp: the size of the user buffer
923 * @ppos: file position
924 *
925 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
926 * values from/to the user buffer, treated as an ASCII string. Negative
927 * strings are not allowed.
928 *
929 * This routine will ensure the values are within the range specified by
930 * table->extra1 (min) and table->extra2 (max). There is a final sanity
931 * check for UINT_MAX to avoid having to support wrap around uses from
932 * userspace.
933 *
934 * Returns 0 on success or -ERANGE on write when the range check fails.
935 */
936int proc_douintvec_minmax(const struct ctl_table *table, int write,
937 void *buffer, size_t *lenp, loff_t *ppos)
938{
939 struct do_proc_douintvec_minmax_conv_param param = {
940 .min = (unsigned int *) table->extra1,
941 .max = (unsigned int *) table->extra2,
942 };
943 return do_proc_douintvec(table, write, buffer, lenp, ppos,
944 do_proc_douintvec_minmax_conv, ¶m);
945}
946
947/**
948 * proc_dou8vec_minmax - read a vector of unsigned chars with min/max values
949 * @table: the sysctl table
950 * @write: %TRUE if this is a write to the sysctl file
951 * @buffer: the user buffer
952 * @lenp: the size of the user buffer
953 * @ppos: file position
954 *
955 * Reads/writes up to table->maxlen/sizeof(u8) unsigned chars
956 * values from/to the user buffer, treated as an ASCII string. Negative
957 * strings are not allowed.
958 *
959 * This routine will ensure the values are within the range specified by
960 * table->extra1 (min) and table->extra2 (max).
961 *
962 * Returns 0 on success or an error on write when the range check fails.
963 */
964int proc_dou8vec_minmax(const struct ctl_table *table, int write,
965 void *buffer, size_t *lenp, loff_t *ppos)
966{
967 struct ctl_table tmp;
968 unsigned int min = 0, max = 255U, val;
969 u8 *data = table->data;
970 struct do_proc_douintvec_minmax_conv_param param = {
971 .min = &min,
972 .max = &max,
973 };
974 int res;
975
976 /* Do not support arrays yet. */
977 if (table->maxlen != sizeof(u8))
978 return -EINVAL;
979
980 if (table->extra1)
981 min = *(unsigned int *) table->extra1;
982 if (table->extra2)
983 max = *(unsigned int *) table->extra2;
984
985 tmp = *table;
986
987 tmp.maxlen = sizeof(val);
988 tmp.data = &val;
989 val = READ_ONCE(*data);
990 res = do_proc_douintvec(&tmp, write, buffer, lenp, ppos,
991 do_proc_douintvec_minmax_conv, ¶m);
992 if (res)
993 return res;
994 if (write)
995 WRITE_ONCE(*data, val);
996 return 0;
997}
998EXPORT_SYMBOL_GPL(proc_dou8vec_minmax);
999
1000#ifdef CONFIG_MAGIC_SYSRQ
1001static int sysrq_sysctl_handler(const struct ctl_table *table, int write,
1002 void *buffer, size_t *lenp, loff_t *ppos)
1003{
1004 int tmp, ret;
1005
1006 tmp = sysrq_mask();
1007
1008 ret = __do_proc_dointvec(&tmp, table, write, buffer,
1009 lenp, ppos, NULL, NULL);
1010 if (ret || !write)
1011 return ret;
1012
1013 if (write)
1014 sysrq_toggle_support(tmp);
1015
1016 return 0;
1017}
1018#endif
1019
1020static int __do_proc_doulongvec_minmax(void *data,
1021 const struct ctl_table *table, int write,
1022 void *buffer, size_t *lenp, loff_t *ppos,
1023 unsigned long convmul, unsigned long convdiv)
1024{
1025 unsigned long *i, *min, *max;
1026 int vleft, first = 1, err = 0;
1027 size_t left;
1028 char *p;
1029
1030 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
1031 *lenp = 0;
1032 return 0;
1033 }
1034
1035 i = data;
1036 min = table->extra1;
1037 max = table->extra2;
1038 vleft = table->maxlen / sizeof(unsigned long);
1039 left = *lenp;
1040
1041 if (write) {
1042 if (proc_first_pos_non_zero_ignore(ppos, table))
1043 goto out;
1044
1045 if (left > PAGE_SIZE - 1)
1046 left = PAGE_SIZE - 1;
1047 p = buffer;
1048 }
1049
1050 for (; left && vleft--; i++, first = 0) {
1051 unsigned long val;
1052
1053 if (write) {
1054 bool neg;
1055
1056 proc_skip_spaces(&p, &left);
1057 if (!left)
1058 break;
1059
1060 err = proc_get_long(&p, &left, &val, &neg,
1061 proc_wspace_sep,
1062 sizeof(proc_wspace_sep), NULL);
1063 if (err || neg) {
1064 err = -EINVAL;
1065 break;
1066 }
1067
1068 val = convmul * val / convdiv;
1069 if ((min && val < *min) || (max && val > *max)) {
1070 err = -EINVAL;
1071 break;
1072 }
1073 WRITE_ONCE(*i, val);
1074 } else {
1075 val = convdiv * READ_ONCE(*i) / convmul;
1076 if (!first)
1077 proc_put_char(&buffer, &left, '\t');
1078 proc_put_long(&buffer, &left, val, false);
1079 }
1080 }
1081
1082 if (!write && !first && left && !err)
1083 proc_put_char(&buffer, &left, '\n');
1084 if (write && !err)
1085 proc_skip_spaces(&p, &left);
1086 if (write && first)
1087 return err ? : -EINVAL;
1088 *lenp -= left;
1089out:
1090 *ppos += *lenp;
1091 return err;
1092}
1093
1094static int do_proc_doulongvec_minmax(const struct ctl_table *table, int write,
1095 void *buffer, size_t *lenp, loff_t *ppos, unsigned long convmul,
1096 unsigned long convdiv)
1097{
1098 return __do_proc_doulongvec_minmax(table->data, table, write,
1099 buffer, lenp, ppos, convmul, convdiv);
1100}
1101
1102/**
1103 * proc_doulongvec_minmax - read a vector of long integers with min/max values
1104 * @table: the sysctl table
1105 * @write: %TRUE if this is a write to the sysctl file
1106 * @buffer: the user buffer
1107 * @lenp: the size of the user buffer
1108 * @ppos: file position
1109 *
1110 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1111 * values from/to the user buffer, treated as an ASCII string.
1112 *
1113 * This routine will ensure the values are within the range specified by
1114 * table->extra1 (min) and table->extra2 (max).
1115 *
1116 * Returns 0 on success.
1117 */
1118int proc_doulongvec_minmax(const struct ctl_table *table, int write,
1119 void *buffer, size_t *lenp, loff_t *ppos)
1120{
1121 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
1122}
1123
1124/**
1125 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
1126 * @table: the sysctl table
1127 * @write: %TRUE if this is a write to the sysctl file
1128 * @buffer: the user buffer
1129 * @lenp: the size of the user buffer
1130 * @ppos: file position
1131 *
1132 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
1133 * values from/to the user buffer, treated as an ASCII string. The values
1134 * are treated as milliseconds, and converted to jiffies when they are stored.
1135 *
1136 * This routine will ensure the values are within the range specified by
1137 * table->extra1 (min) and table->extra2 (max).
1138 *
1139 * Returns 0 on success.
1140 */
1141int proc_doulongvec_ms_jiffies_minmax(const struct ctl_table *table, int write,
1142 void *buffer, size_t *lenp, loff_t *ppos)
1143{
1144 return do_proc_doulongvec_minmax(table, write, buffer,
1145 lenp, ppos, HZ, 1000l);
1146}
1147
1148
1149static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
1150 int *valp,
1151 int write, void *data)
1152{
1153 if (write) {
1154 if (*lvalp > INT_MAX / HZ)
1155 return 1;
1156 if (*negp)
1157 WRITE_ONCE(*valp, -*lvalp * HZ);
1158 else
1159 WRITE_ONCE(*valp, *lvalp * HZ);
1160 } else {
1161 int val = READ_ONCE(*valp);
1162 unsigned long lval;
1163 if (val < 0) {
1164 *negp = true;
1165 lval = -(unsigned long)val;
1166 } else {
1167 *negp = false;
1168 lval = (unsigned long)val;
1169 }
1170 *lvalp = lval / HZ;
1171 }
1172 return 0;
1173}
1174
1175static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
1176 int *valp,
1177 int write, void *data)
1178{
1179 if (write) {
1180 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
1181 return 1;
1182 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
1183 } else {
1184 int val = *valp;
1185 unsigned long lval;
1186 if (val < 0) {
1187 *negp = true;
1188 lval = -(unsigned long)val;
1189 } else {
1190 *negp = false;
1191 lval = (unsigned long)val;
1192 }
1193 *lvalp = jiffies_to_clock_t(lval);
1194 }
1195 return 0;
1196}
1197
1198static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
1199 int *valp,
1200 int write, void *data)
1201{
1202 if (write) {
1203 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
1204
1205 if (jif > INT_MAX)
1206 return 1;
1207 WRITE_ONCE(*valp, (int)jif);
1208 } else {
1209 int val = READ_ONCE(*valp);
1210 unsigned long lval;
1211 if (val < 0) {
1212 *negp = true;
1213 lval = -(unsigned long)val;
1214 } else {
1215 *negp = false;
1216 lval = (unsigned long)val;
1217 }
1218 *lvalp = jiffies_to_msecs(lval);
1219 }
1220 return 0;
1221}
1222
1223static int do_proc_dointvec_ms_jiffies_minmax_conv(bool *negp, unsigned long *lvalp,
1224 int *valp, int write, void *data)
1225{
1226 int tmp, ret;
1227 struct do_proc_dointvec_minmax_conv_param *param = data;
1228 /*
1229 * If writing, first do so via a temporary local int so we can
1230 * bounds-check it before touching *valp.
1231 */
1232 int *ip = write ? &tmp : valp;
1233
1234 ret = do_proc_dointvec_ms_jiffies_conv(negp, lvalp, ip, write, data);
1235 if (ret)
1236 return ret;
1237
1238 if (write) {
1239 if ((param->min && *param->min > tmp) ||
1240 (param->max && *param->max < tmp))
1241 return -EINVAL;
1242 *valp = tmp;
1243 }
1244 return 0;
1245}
1246
1247/**
1248 * proc_dointvec_jiffies - read a vector of integers as seconds
1249 * @table: the sysctl table
1250 * @write: %TRUE if this is a write to the sysctl file
1251 * @buffer: the user buffer
1252 * @lenp: the size of the user buffer
1253 * @ppos: file position
1254 *
1255 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1256 * values from/to the user buffer, treated as an ASCII string.
1257 * The values read are assumed to be in seconds, and are converted into
1258 * jiffies.
1259 *
1260 * Returns 0 on success.
1261 */
1262int proc_dointvec_jiffies(const struct ctl_table *table, int write,
1263 void *buffer, size_t *lenp, loff_t *ppos)
1264{
1265 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1266 do_proc_dointvec_jiffies_conv,NULL);
1267}
1268
1269int proc_dointvec_ms_jiffies_minmax(const struct ctl_table *table, int write,
1270 void *buffer, size_t *lenp, loff_t *ppos)
1271{
1272 struct do_proc_dointvec_minmax_conv_param param = {
1273 .min = (int *) table->extra1,
1274 .max = (int *) table->extra2,
1275 };
1276 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1277 do_proc_dointvec_ms_jiffies_minmax_conv, ¶m);
1278}
1279
1280/**
1281 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
1282 * @table: the sysctl table
1283 * @write: %TRUE if this is a write to the sysctl file
1284 * @buffer: the user buffer
1285 * @lenp: the size of the user buffer
1286 * @ppos: pointer to the file position
1287 *
1288 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1289 * values from/to the user buffer, treated as an ASCII string.
1290 * The values read are assumed to be in 1/USER_HZ seconds, and
1291 * are converted into jiffies.
1292 *
1293 * Returns 0 on success.
1294 */
1295int proc_dointvec_userhz_jiffies(const struct ctl_table *table, int write,
1296 void *buffer, size_t *lenp, loff_t *ppos)
1297{
1298 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1299 do_proc_dointvec_userhz_jiffies_conv, NULL);
1300}
1301
1302/**
1303 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
1304 * @table: the sysctl table
1305 * @write: %TRUE if this is a write to the sysctl file
1306 * @buffer: the user buffer
1307 * @lenp: the size of the user buffer
1308 * @ppos: the current position in the file
1309 *
1310 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1311 * values from/to the user buffer, treated as an ASCII string.
1312 * The values read are assumed to be in 1/1000 seconds, and
1313 * are converted into jiffies.
1314 *
1315 * Returns 0 on success.
1316 */
1317int proc_dointvec_ms_jiffies(const struct ctl_table *table, int write, void *buffer,
1318 size_t *lenp, loff_t *ppos)
1319{
1320 return do_proc_dointvec(table, write, buffer, lenp, ppos,
1321 do_proc_dointvec_ms_jiffies_conv, NULL);
1322}
1323
1324static int proc_do_cad_pid(const struct ctl_table *table, int write, void *buffer,
1325 size_t *lenp, loff_t *ppos)
1326{
1327 struct pid *new_pid;
1328 pid_t tmp;
1329 int r;
1330
1331 tmp = pid_vnr(cad_pid);
1332
1333 r = __do_proc_dointvec(&tmp, table, write, buffer,
1334 lenp, ppos, NULL, NULL);
1335 if (r || !write)
1336 return r;
1337
1338 new_pid = find_get_pid(tmp);
1339 if (!new_pid)
1340 return -ESRCH;
1341
1342 put_pid(xchg(&cad_pid, new_pid));
1343 return 0;
1344}
1345
1346/**
1347 * proc_do_large_bitmap - read/write from/to a large bitmap
1348 * @table: the sysctl table
1349 * @write: %TRUE if this is a write to the sysctl file
1350 * @buffer: the user buffer
1351 * @lenp: the size of the user buffer
1352 * @ppos: file position
1353 *
1354 * The bitmap is stored at table->data and the bitmap length (in bits)
1355 * in table->maxlen.
1356 *
1357 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
1358 * large bitmaps may be represented in a compact manner. Writing into
1359 * the file will clear the bitmap then update it with the given input.
1360 *
1361 * Returns 0 on success.
1362 */
1363int proc_do_large_bitmap(const struct ctl_table *table, int write,
1364 void *buffer, size_t *lenp, loff_t *ppos)
1365{
1366 int err = 0;
1367 size_t left = *lenp;
1368 unsigned long bitmap_len = table->maxlen;
1369 unsigned long *bitmap = *(unsigned long **) table->data;
1370 unsigned long *tmp_bitmap = NULL;
1371 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
1372
1373 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
1374 *lenp = 0;
1375 return 0;
1376 }
1377
1378 if (write) {
1379 char *p = buffer;
1380 size_t skipped = 0;
1381
1382 if (left > PAGE_SIZE - 1) {
1383 left = PAGE_SIZE - 1;
1384 /* How much of the buffer we'll skip this pass */
1385 skipped = *lenp - left;
1386 }
1387
1388 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
1389 if (!tmp_bitmap)
1390 return -ENOMEM;
1391 proc_skip_char(&p, &left, '\n');
1392 while (!err && left) {
1393 unsigned long val_a, val_b;
1394 bool neg;
1395 size_t saved_left;
1396
1397 /* In case we stop parsing mid-number, we can reset */
1398 saved_left = left;
1399 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
1400 sizeof(tr_a), &c);
1401 /*
1402 * If we consumed the entirety of a truncated buffer or
1403 * only one char is left (may be a "-"), then stop here,
1404 * reset, & come back for more.
1405 */
1406 if ((left <= 1) && skipped) {
1407 left = saved_left;
1408 break;
1409 }
1410
1411 if (err)
1412 break;
1413 if (val_a >= bitmap_len || neg) {
1414 err = -EINVAL;
1415 break;
1416 }
1417
1418 val_b = val_a;
1419 if (left) {
1420 p++;
1421 left--;
1422 }
1423
1424 if (c == '-') {
1425 err = proc_get_long(&p, &left, &val_b,
1426 &neg, tr_b, sizeof(tr_b),
1427 &c);
1428 /*
1429 * If we consumed all of a truncated buffer or
1430 * then stop here, reset, & come back for more.
1431 */
1432 if (!left && skipped) {
1433 left = saved_left;
1434 break;
1435 }
1436
1437 if (err)
1438 break;
1439 if (val_b >= bitmap_len || neg ||
1440 val_a > val_b) {
1441 err = -EINVAL;
1442 break;
1443 }
1444 if (left) {
1445 p++;
1446 left--;
1447 }
1448 }
1449
1450 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
1451 proc_skip_char(&p, &left, '\n');
1452 }
1453 left += skipped;
1454 } else {
1455 unsigned long bit_a, bit_b = 0;
1456 bool first = 1;
1457
1458 while (left) {
1459 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
1460 if (bit_a >= bitmap_len)
1461 break;
1462 bit_b = find_next_zero_bit(bitmap, bitmap_len,
1463 bit_a + 1) - 1;
1464
1465 if (!first)
1466 proc_put_char(&buffer, &left, ',');
1467 proc_put_long(&buffer, &left, bit_a, false);
1468 if (bit_a != bit_b) {
1469 proc_put_char(&buffer, &left, '-');
1470 proc_put_long(&buffer, &left, bit_b, false);
1471 }
1472
1473 first = 0; bit_b++;
1474 }
1475 proc_put_char(&buffer, &left, '\n');
1476 }
1477
1478 if (!err) {
1479 if (write) {
1480 if (*ppos)
1481 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
1482 else
1483 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
1484 }
1485 *lenp -= left;
1486 *ppos += *lenp;
1487 }
1488
1489 bitmap_free(tmp_bitmap);
1490 return err;
1491}
1492
1493#else /* CONFIG_PROC_SYSCTL */
1494
1495int proc_dostring(const struct ctl_table *table, int write,
1496 void *buffer, size_t *lenp, loff_t *ppos)
1497{
1498 return -ENOSYS;
1499}
1500
1501int proc_dobool(const struct ctl_table *table, int write,
1502 void *buffer, size_t *lenp, loff_t *ppos)
1503{
1504 return -ENOSYS;
1505}
1506
1507int proc_dointvec(const struct ctl_table *table, int write,
1508 void *buffer, size_t *lenp, loff_t *ppos)
1509{
1510 return -ENOSYS;
1511}
1512
1513int proc_douintvec(const struct ctl_table *table, int write,
1514 void *buffer, size_t *lenp, loff_t *ppos)
1515{
1516 return -ENOSYS;
1517}
1518
1519int proc_dointvec_minmax(const struct ctl_table *table, int write,
1520 void *buffer, size_t *lenp, loff_t *ppos)
1521{
1522 return -ENOSYS;
1523}
1524
1525int proc_douintvec_minmax(const struct ctl_table *table, int write,
1526 void *buffer, size_t *lenp, loff_t *ppos)
1527{
1528 return -ENOSYS;
1529}
1530
1531int proc_dou8vec_minmax(const struct ctl_table *table, int write,
1532 void *buffer, size_t *lenp, loff_t *ppos)
1533{
1534 return -ENOSYS;
1535}
1536
1537int proc_dointvec_jiffies(const struct ctl_table *table, int write,
1538 void *buffer, size_t *lenp, loff_t *ppos)
1539{
1540 return -ENOSYS;
1541}
1542
1543int proc_dointvec_ms_jiffies_minmax(const struct ctl_table *table, int write,
1544 void *buffer, size_t *lenp, loff_t *ppos)
1545{
1546 return -ENOSYS;
1547}
1548
1549int proc_dointvec_userhz_jiffies(const struct ctl_table *table, int write,
1550 void *buffer, size_t *lenp, loff_t *ppos)
1551{
1552 return -ENOSYS;
1553}
1554
1555int proc_dointvec_ms_jiffies(const struct ctl_table *table, int write,
1556 void *buffer, size_t *lenp, loff_t *ppos)
1557{
1558 return -ENOSYS;
1559}
1560
1561int proc_doulongvec_minmax(const struct ctl_table *table, int write,
1562 void *buffer, size_t *lenp, loff_t *ppos)
1563{
1564 return -ENOSYS;
1565}
1566
1567int proc_doulongvec_ms_jiffies_minmax(const struct ctl_table *table, int write,
1568 void *buffer, size_t *lenp, loff_t *ppos)
1569{
1570 return -ENOSYS;
1571}
1572
1573int proc_do_large_bitmap(const struct ctl_table *table, int write,
1574 void *buffer, size_t *lenp, loff_t *ppos)
1575{
1576 return -ENOSYS;
1577}
1578
1579#endif /* CONFIG_PROC_SYSCTL */
1580
1581#if defined(CONFIG_SYSCTL)
1582int proc_do_static_key(const struct ctl_table *table, int write,
1583 void *buffer, size_t *lenp, loff_t *ppos)
1584{
1585 struct static_key *key = (struct static_key *)table->data;
1586 static DEFINE_MUTEX(static_key_mutex);
1587 int val, ret;
1588 struct ctl_table tmp = {
1589 .data = &val,
1590 .maxlen = sizeof(val),
1591 .mode = table->mode,
1592 .extra1 = SYSCTL_ZERO,
1593 .extra2 = SYSCTL_ONE,
1594 };
1595
1596 if (write && !capable(CAP_SYS_ADMIN))
1597 return -EPERM;
1598
1599 mutex_lock(&static_key_mutex);
1600 val = static_key_enabled(key);
1601 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
1602 if (write && !ret) {
1603 if (val)
1604 static_key_enable(key);
1605 else
1606 static_key_disable(key);
1607 }
1608 mutex_unlock(&static_key_mutex);
1609 return ret;
1610}
1611
1612static struct ctl_table kern_table[] = {
1613 {
1614 .procname = "panic",
1615 .data = &panic_timeout,
1616 .maxlen = sizeof(int),
1617 .mode = 0644,
1618 .proc_handler = proc_dointvec,
1619 },
1620#ifdef CONFIG_PROC_SYSCTL
1621 {
1622 .procname = "tainted",
1623 .maxlen = sizeof(long),
1624 .mode = 0644,
1625 .proc_handler = proc_taint,
1626 },
1627 {
1628 .procname = "sysctl_writes_strict",
1629 .data = &sysctl_writes_strict,
1630 .maxlen = sizeof(int),
1631 .mode = 0644,
1632 .proc_handler = proc_dointvec_minmax,
1633 .extra1 = SYSCTL_NEG_ONE,
1634 .extra2 = SYSCTL_ONE,
1635 },
1636#endif
1637 {
1638 .procname = "print-fatal-signals",
1639 .data = &print_fatal_signals,
1640 .maxlen = sizeof(int),
1641 .mode = 0644,
1642 .proc_handler = proc_dointvec,
1643 },
1644#ifdef CONFIG_SPARC
1645 {
1646 .procname = "reboot-cmd",
1647 .data = reboot_command,
1648 .maxlen = 256,
1649 .mode = 0644,
1650 .proc_handler = proc_dostring,
1651 },
1652 {
1653 .procname = "stop-a",
1654 .data = &stop_a_enabled,
1655 .maxlen = sizeof (int),
1656 .mode = 0644,
1657 .proc_handler = proc_dointvec,
1658 },
1659 {
1660 .procname = "scons-poweroff",
1661 .data = &scons_pwroff,
1662 .maxlen = sizeof (int),
1663 .mode = 0644,
1664 .proc_handler = proc_dointvec,
1665 },
1666#endif
1667#ifdef CONFIG_SPARC64
1668 {
1669 .procname = "tsb-ratio",
1670 .data = &sysctl_tsb_ratio,
1671 .maxlen = sizeof (int),
1672 .mode = 0644,
1673 .proc_handler = proc_dointvec,
1674 },
1675#endif
1676#ifdef CONFIG_PARISC
1677 {
1678 .procname = "soft-power",
1679 .data = &pwrsw_enabled,
1680 .maxlen = sizeof (int),
1681 .mode = 0644,
1682 .proc_handler = proc_dointvec,
1683 },
1684#endif
1685#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
1686 {
1687 .procname = "unaligned-trap",
1688 .data = &unaligned_enabled,
1689 .maxlen = sizeof (int),
1690 .mode = 0644,
1691 .proc_handler = proc_dointvec,
1692 },
1693#endif
1694#ifdef CONFIG_STACK_TRACER
1695 {
1696 .procname = "stack_tracer_enabled",
1697 .data = &stack_tracer_enabled,
1698 .maxlen = sizeof(int),
1699 .mode = 0644,
1700 .proc_handler = stack_trace_sysctl,
1701 },
1702#endif
1703#ifdef CONFIG_TRACING
1704 {
1705 .procname = "ftrace_dump_on_oops",
1706 .data = &ftrace_dump_on_oops,
1707 .maxlen = MAX_TRACER_SIZE,
1708 .mode = 0644,
1709 .proc_handler = proc_dostring,
1710 },
1711 {
1712 .procname = "traceoff_on_warning",
1713 .data = &__disable_trace_on_warning,
1714 .maxlen = sizeof(__disable_trace_on_warning),
1715 .mode = 0644,
1716 .proc_handler = proc_dointvec,
1717 },
1718 {
1719 .procname = "tracepoint_printk",
1720 .data = &tracepoint_printk,
1721 .maxlen = sizeof(tracepoint_printk),
1722 .mode = 0644,
1723 .proc_handler = tracepoint_printk_sysctl,
1724 },
1725#endif
1726#ifdef CONFIG_MODULES
1727 {
1728 .procname = "modprobe",
1729 .data = &modprobe_path,
1730 .maxlen = KMOD_PATH_LEN,
1731 .mode = 0644,
1732 .proc_handler = proc_dostring,
1733 },
1734 {
1735 .procname = "modules_disabled",
1736 .data = &modules_disabled,
1737 .maxlen = sizeof(int),
1738 .mode = 0644,
1739 /* only handle a transition from default "0" to "1" */
1740 .proc_handler = proc_dointvec_minmax,
1741 .extra1 = SYSCTL_ONE,
1742 .extra2 = SYSCTL_ONE,
1743 },
1744#endif
1745#ifdef CONFIG_UEVENT_HELPER
1746 {
1747 .procname = "hotplug",
1748 .data = &uevent_helper,
1749 .maxlen = UEVENT_HELPER_PATH_LEN,
1750 .mode = 0644,
1751 .proc_handler = proc_dostring,
1752 },
1753#endif
1754#ifdef CONFIG_MAGIC_SYSRQ
1755 {
1756 .procname = "sysrq",
1757 .data = NULL,
1758 .maxlen = sizeof (int),
1759 .mode = 0644,
1760 .proc_handler = sysrq_sysctl_handler,
1761 },
1762#endif
1763#ifdef CONFIG_PROC_SYSCTL
1764 {
1765 .procname = "cad_pid",
1766 .data = NULL,
1767 .maxlen = sizeof (int),
1768 .mode = 0600,
1769 .proc_handler = proc_do_cad_pid,
1770 },
1771#endif
1772 {
1773 .procname = "threads-max",
1774 .data = NULL,
1775 .maxlen = sizeof(int),
1776 .mode = 0644,
1777 .proc_handler = sysctl_max_threads,
1778 },
1779 {
1780 .procname = "overflowuid",
1781 .data = &overflowuid,
1782 .maxlen = sizeof(int),
1783 .mode = 0644,
1784 .proc_handler = proc_dointvec_minmax,
1785 .extra1 = SYSCTL_ZERO,
1786 .extra2 = SYSCTL_MAXOLDUID,
1787 },
1788 {
1789 .procname = "overflowgid",
1790 .data = &overflowgid,
1791 .maxlen = sizeof(int),
1792 .mode = 0644,
1793 .proc_handler = proc_dointvec_minmax,
1794 .extra1 = SYSCTL_ZERO,
1795 .extra2 = SYSCTL_MAXOLDUID,
1796 },
1797#ifdef CONFIG_S390
1798 {
1799 .procname = "userprocess_debug",
1800 .data = &show_unhandled_signals,
1801 .maxlen = sizeof(int),
1802 .mode = 0644,
1803 .proc_handler = proc_dointvec,
1804 },
1805#endif
1806 {
1807 .procname = "pid_max",
1808 .data = &pid_max,
1809 .maxlen = sizeof (int),
1810 .mode = 0644,
1811 .proc_handler = proc_dointvec_minmax,
1812 .extra1 = &pid_max_min,
1813 .extra2 = &pid_max_max,
1814 },
1815 {
1816 .procname = "panic_on_oops",
1817 .data = &panic_on_oops,
1818 .maxlen = sizeof(int),
1819 .mode = 0644,
1820 .proc_handler = proc_dointvec,
1821 },
1822 {
1823 .procname = "panic_print",
1824 .data = &panic_print,
1825 .maxlen = sizeof(unsigned long),
1826 .mode = 0644,
1827 .proc_handler = proc_doulongvec_minmax,
1828 },
1829 {
1830 .procname = "ngroups_max",
1831 .data = (void *)&ngroups_max,
1832 .maxlen = sizeof (int),
1833 .mode = 0444,
1834 .proc_handler = proc_dointvec,
1835 },
1836 {
1837 .procname = "cap_last_cap",
1838 .data = (void *)&cap_last_cap,
1839 .maxlen = sizeof(int),
1840 .mode = 0444,
1841 .proc_handler = proc_dointvec,
1842 },
1843#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1844 {
1845 .procname = "unknown_nmi_panic",
1846 .data = &unknown_nmi_panic,
1847 .maxlen = sizeof (int),
1848 .mode = 0644,
1849 .proc_handler = proc_dointvec,
1850 },
1851#endif
1852
1853#if (defined(CONFIG_X86_32) || defined(CONFIG_PARISC)) && \
1854 defined(CONFIG_DEBUG_STACKOVERFLOW)
1855 {
1856 .procname = "panic_on_stackoverflow",
1857 .data = &sysctl_panic_on_stackoverflow,
1858 .maxlen = sizeof(int),
1859 .mode = 0644,
1860 .proc_handler = proc_dointvec,
1861 },
1862#endif
1863#if defined(CONFIG_X86)
1864 {
1865 .procname = "panic_on_unrecovered_nmi",
1866 .data = &panic_on_unrecovered_nmi,
1867 .maxlen = sizeof(int),
1868 .mode = 0644,
1869 .proc_handler = proc_dointvec,
1870 },
1871 {
1872 .procname = "panic_on_io_nmi",
1873 .data = &panic_on_io_nmi,
1874 .maxlen = sizeof(int),
1875 .mode = 0644,
1876 .proc_handler = proc_dointvec,
1877 },
1878 {
1879 .procname = "bootloader_type",
1880 .data = &bootloader_type,
1881 .maxlen = sizeof (int),
1882 .mode = 0444,
1883 .proc_handler = proc_dointvec,
1884 },
1885 {
1886 .procname = "bootloader_version",
1887 .data = &bootloader_version,
1888 .maxlen = sizeof (int),
1889 .mode = 0444,
1890 .proc_handler = proc_dointvec,
1891 },
1892 {
1893 .procname = "io_delay_type",
1894 .data = &io_delay_type,
1895 .maxlen = sizeof(int),
1896 .mode = 0644,
1897 .proc_handler = proc_dointvec,
1898 },
1899#endif
1900#if defined(CONFIG_MMU)
1901 {
1902 .procname = "randomize_va_space",
1903 .data = &randomize_va_space,
1904 .maxlen = sizeof(int),
1905 .mode = 0644,
1906 .proc_handler = proc_dointvec,
1907 },
1908#endif
1909#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1910 {
1911 .procname = "spin_retry",
1912 .data = &spin_retry,
1913 .maxlen = sizeof (int),
1914 .mode = 0644,
1915 .proc_handler = proc_dointvec,
1916 },
1917#endif
1918#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1919 {
1920 .procname = "acpi_video_flags",
1921 .data = &acpi_realmode_flags,
1922 .maxlen = sizeof (unsigned long),
1923 .mode = 0644,
1924 .proc_handler = proc_doulongvec_minmax,
1925 },
1926#endif
1927#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1928 {
1929 .procname = "ignore-unaligned-usertrap",
1930 .data = &no_unaligned_warning,
1931 .maxlen = sizeof (int),
1932 .mode = 0644,
1933 .proc_handler = proc_dointvec,
1934 },
1935#endif
1936#ifdef CONFIG_RT_MUTEXES
1937 {
1938 .procname = "max_lock_depth",
1939 .data = &max_lock_depth,
1940 .maxlen = sizeof(int),
1941 .mode = 0644,
1942 .proc_handler = proc_dointvec,
1943 },
1944#endif
1945#ifdef CONFIG_PERF_EVENTS
1946 /*
1947 * User-space scripts rely on the existence of this file
1948 * as a feature check for perf_events being enabled.
1949 *
1950 * So it's an ABI, do not remove!
1951 */
1952 {
1953 .procname = "perf_event_paranoid",
1954 .data = &sysctl_perf_event_paranoid,
1955 .maxlen = sizeof(sysctl_perf_event_paranoid),
1956 .mode = 0644,
1957 .proc_handler = proc_dointvec,
1958 },
1959 {
1960 .procname = "perf_event_mlock_kb",
1961 .data = &sysctl_perf_event_mlock,
1962 .maxlen = sizeof(sysctl_perf_event_mlock),
1963 .mode = 0644,
1964 .proc_handler = proc_dointvec,
1965 },
1966 {
1967 .procname = "perf_event_max_sample_rate",
1968 .data = &sysctl_perf_event_sample_rate,
1969 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1970 .mode = 0644,
1971 .proc_handler = perf_event_max_sample_rate_handler,
1972 .extra1 = SYSCTL_ONE,
1973 },
1974 {
1975 .procname = "perf_cpu_time_max_percent",
1976 .data = &sysctl_perf_cpu_time_max_percent,
1977 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1978 .mode = 0644,
1979 .proc_handler = perf_cpu_time_max_percent_handler,
1980 .extra1 = SYSCTL_ZERO,
1981 .extra2 = SYSCTL_ONE_HUNDRED,
1982 },
1983 {
1984 .procname = "perf_event_max_stack",
1985 .data = &sysctl_perf_event_max_stack,
1986 .maxlen = sizeof(sysctl_perf_event_max_stack),
1987 .mode = 0644,
1988 .proc_handler = perf_event_max_stack_handler,
1989 .extra1 = SYSCTL_ZERO,
1990 .extra2 = (void *)&six_hundred_forty_kb,
1991 },
1992 {
1993 .procname = "perf_event_max_contexts_per_stack",
1994 .data = &sysctl_perf_event_max_contexts_per_stack,
1995 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1996 .mode = 0644,
1997 .proc_handler = perf_event_max_stack_handler,
1998 .extra1 = SYSCTL_ZERO,
1999 .extra2 = SYSCTL_ONE_THOUSAND,
2000 },
2001#endif
2002 {
2003 .procname = "panic_on_warn",
2004 .data = &panic_on_warn,
2005 .maxlen = sizeof(int),
2006 .mode = 0644,
2007 .proc_handler = proc_dointvec_minmax,
2008 .extra1 = SYSCTL_ZERO,
2009 .extra2 = SYSCTL_ONE,
2010 },
2011#ifdef CONFIG_TREE_RCU
2012 {
2013 .procname = "panic_on_rcu_stall",
2014 .data = &sysctl_panic_on_rcu_stall,
2015 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
2016 .mode = 0644,
2017 .proc_handler = proc_dointvec_minmax,
2018 .extra1 = SYSCTL_ZERO,
2019 .extra2 = SYSCTL_ONE,
2020 },
2021 {
2022 .procname = "max_rcu_stall_to_panic",
2023 .data = &sysctl_max_rcu_stall_to_panic,
2024 .maxlen = sizeof(sysctl_max_rcu_stall_to_panic),
2025 .mode = 0644,
2026 .proc_handler = proc_dointvec_minmax,
2027 .extra1 = SYSCTL_ONE,
2028 .extra2 = SYSCTL_INT_MAX,
2029 },
2030#endif
2031};
2032
2033static struct ctl_table vm_table[] = {
2034 {
2035 .procname = "overcommit_memory",
2036 .data = &sysctl_overcommit_memory,
2037 .maxlen = sizeof(sysctl_overcommit_memory),
2038 .mode = 0644,
2039 .proc_handler = overcommit_policy_handler,
2040 .extra1 = SYSCTL_ZERO,
2041 .extra2 = SYSCTL_TWO,
2042 },
2043 {
2044 .procname = "overcommit_ratio",
2045 .data = &sysctl_overcommit_ratio,
2046 .maxlen = sizeof(sysctl_overcommit_ratio),
2047 .mode = 0644,
2048 .proc_handler = overcommit_ratio_handler,
2049 },
2050 {
2051 .procname = "overcommit_kbytes",
2052 .data = &sysctl_overcommit_kbytes,
2053 .maxlen = sizeof(sysctl_overcommit_kbytes),
2054 .mode = 0644,
2055 .proc_handler = overcommit_kbytes_handler,
2056 },
2057 {
2058 .procname = "page-cluster",
2059 .data = &page_cluster,
2060 .maxlen = sizeof(int),
2061 .mode = 0644,
2062 .proc_handler = proc_dointvec_minmax,
2063 .extra1 = SYSCTL_ZERO,
2064 .extra2 = (void *)&page_cluster_max,
2065 },
2066 {
2067 .procname = "dirtytime_expire_seconds",
2068 .data = &dirtytime_expire_interval,
2069 .maxlen = sizeof(dirtytime_expire_interval),
2070 .mode = 0644,
2071 .proc_handler = dirtytime_interval_handler,
2072 .extra1 = SYSCTL_ZERO,
2073 },
2074 {
2075 .procname = "swappiness",
2076 .data = &vm_swappiness,
2077 .maxlen = sizeof(vm_swappiness),
2078 .mode = 0644,
2079 .proc_handler = proc_dointvec_minmax,
2080 .extra1 = SYSCTL_ZERO,
2081 .extra2 = SYSCTL_TWO_HUNDRED,
2082 },
2083#ifdef CONFIG_NUMA
2084 {
2085 .procname = "numa_stat",
2086 .data = &sysctl_vm_numa_stat,
2087 .maxlen = sizeof(int),
2088 .mode = 0644,
2089 .proc_handler = sysctl_vm_numa_stat_handler,
2090 .extra1 = SYSCTL_ZERO,
2091 .extra2 = SYSCTL_ONE,
2092 },
2093#endif
2094 {
2095 .procname = "drop_caches",
2096 .data = &sysctl_drop_caches,
2097 .maxlen = sizeof(int),
2098 .mode = 0200,
2099 .proc_handler = drop_caches_sysctl_handler,
2100 .extra1 = SYSCTL_ONE,
2101 .extra2 = SYSCTL_FOUR,
2102 },
2103 {
2104 .procname = "page_lock_unfairness",
2105 .data = &sysctl_page_lock_unfairness,
2106 .maxlen = sizeof(sysctl_page_lock_unfairness),
2107 .mode = 0644,
2108 .proc_handler = proc_dointvec_minmax,
2109 .extra1 = SYSCTL_ZERO,
2110 },
2111#ifdef CONFIG_MMU
2112 {
2113 .procname = "max_map_count",
2114 .data = &sysctl_max_map_count,
2115 .maxlen = sizeof(sysctl_max_map_count),
2116 .mode = 0644,
2117 .proc_handler = proc_dointvec_minmax,
2118 .extra1 = SYSCTL_ZERO,
2119 },
2120#else
2121 {
2122 .procname = "nr_trim_pages",
2123 .data = &sysctl_nr_trim_pages,
2124 .maxlen = sizeof(sysctl_nr_trim_pages),
2125 .mode = 0644,
2126 .proc_handler = proc_dointvec_minmax,
2127 .extra1 = SYSCTL_ZERO,
2128 },
2129#endif
2130 {
2131 .procname = "vfs_cache_pressure",
2132 .data = &sysctl_vfs_cache_pressure,
2133 .maxlen = sizeof(sysctl_vfs_cache_pressure),
2134 .mode = 0644,
2135 .proc_handler = proc_dointvec_minmax,
2136 .extra1 = SYSCTL_ZERO,
2137 },
2138#if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
2139 defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
2140 {
2141 .procname = "legacy_va_layout",
2142 .data = &sysctl_legacy_va_layout,
2143 .maxlen = sizeof(sysctl_legacy_va_layout),
2144 .mode = 0644,
2145 .proc_handler = proc_dointvec_minmax,
2146 .extra1 = SYSCTL_ZERO,
2147 },
2148#endif
2149#ifdef CONFIG_NUMA
2150 {
2151 .procname = "zone_reclaim_mode",
2152 .data = &node_reclaim_mode,
2153 .maxlen = sizeof(node_reclaim_mode),
2154 .mode = 0644,
2155 .proc_handler = proc_dointvec_minmax,
2156 .extra1 = SYSCTL_ZERO,
2157 },
2158#endif
2159#ifdef CONFIG_SMP
2160 {
2161 .procname = "stat_interval",
2162 .data = &sysctl_stat_interval,
2163 .maxlen = sizeof(sysctl_stat_interval),
2164 .mode = 0644,
2165 .proc_handler = proc_dointvec_jiffies,
2166 },
2167 {
2168 .procname = "stat_refresh",
2169 .data = NULL,
2170 .maxlen = 0,
2171 .mode = 0600,
2172 .proc_handler = vmstat_refresh,
2173 },
2174#endif
2175#ifdef CONFIG_MMU
2176 {
2177 .procname = "mmap_min_addr",
2178 .data = &dac_mmap_min_addr,
2179 .maxlen = sizeof(unsigned long),
2180 .mode = 0644,
2181 .proc_handler = mmap_min_addr_handler,
2182 },
2183#endif
2184#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
2185 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
2186 {
2187 .procname = "vdso_enabled",
2188#ifdef CONFIG_X86_32
2189 .data = &vdso32_enabled,
2190 .maxlen = sizeof(vdso32_enabled),
2191#else
2192 .data = &vdso_enabled,
2193 .maxlen = sizeof(vdso_enabled),
2194#endif
2195 .mode = 0644,
2196 .proc_handler = proc_dointvec,
2197 .extra1 = SYSCTL_ZERO,
2198 },
2199#endif
2200 {
2201 .procname = "user_reserve_kbytes",
2202 .data = &sysctl_user_reserve_kbytes,
2203 .maxlen = sizeof(sysctl_user_reserve_kbytes),
2204 .mode = 0644,
2205 .proc_handler = proc_doulongvec_minmax,
2206 },
2207 {
2208 .procname = "admin_reserve_kbytes",
2209 .data = &sysctl_admin_reserve_kbytes,
2210 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
2211 .mode = 0644,
2212 .proc_handler = proc_doulongvec_minmax,
2213 },
2214#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
2215 {
2216 .procname = "mmap_rnd_bits",
2217 .data = &mmap_rnd_bits,
2218 .maxlen = sizeof(mmap_rnd_bits),
2219 .mode = 0600,
2220 .proc_handler = proc_dointvec_minmax,
2221 .extra1 = (void *)&mmap_rnd_bits_min,
2222 .extra2 = (void *)&mmap_rnd_bits_max,
2223 },
2224#endif
2225#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
2226 {
2227 .procname = "mmap_rnd_compat_bits",
2228 .data = &mmap_rnd_compat_bits,
2229 .maxlen = sizeof(mmap_rnd_compat_bits),
2230 .mode = 0600,
2231 .proc_handler = proc_dointvec_minmax,
2232 .extra1 = (void *)&mmap_rnd_compat_bits_min,
2233 .extra2 = (void *)&mmap_rnd_compat_bits_max,
2234 },
2235#endif
2236};
2237
2238int __init sysctl_init_bases(void)
2239{
2240 register_sysctl_init("kernel", kern_table);
2241 register_sysctl_init("vm", vm_table);
2242
2243 return 0;
2244}
2245#endif /* CONFIG_SYSCTL */
2246/*
2247 * No sense putting this after each symbol definition, twice,
2248 * exception granted :-)
2249 */
2250EXPORT_SYMBOL(proc_dobool);
2251EXPORT_SYMBOL(proc_dointvec);
2252EXPORT_SYMBOL(proc_douintvec);
2253EXPORT_SYMBOL(proc_dointvec_jiffies);
2254EXPORT_SYMBOL(proc_dointvec_minmax);
2255EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
2256EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2257EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2258EXPORT_SYMBOL(proc_dostring);
2259EXPORT_SYMBOL(proc_doulongvec_minmax);
2260EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2261EXPORT_SYMBOL(proc_do_large_bitmap);