projects/linux·blob v4.0-rc1 · refs

2761 lines · 62.6 KiB

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