]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-bionic-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
lockup_detector: Combine nmi_watchdog and softlockup detector
[mirror_ubuntu-bionic-kernel.git] / kernel / sysctl.c
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/signal.h>
27 #include <linux/proc_fs.h>
28 #include <linux/security.h>
29 #include <linux/ctype.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/ratelimit.h>
40 #include <linux/hugetlb.h>
41 #include <linux/initrd.h>
42 #include <linux/key.h>
43 #include <linux/times.h>
44 #include <linux/limits.h>
45 #include <linux/dcache.h>
46 #include <linux/syscalls.h>
47 #include <linux/vmstat.h>
48 #include <linux/nfs_fs.h>
49 #include <linux/acpi.h>
50 #include <linux/reboot.h>
51 #include <linux/ftrace.h>
52 #include <linux/slow-work.h>
53 #include <linux/perf_event.h>
54 #include <linux/kprobes.h>
55
56 #include <asm/uaccess.h>
57 #include <asm/processor.h>
58
59 #ifdef CONFIG_X86
60 #include <asm/nmi.h>
61 #include <asm/stacktrace.h>
62 #include <asm/io.h>
63 #endif
64 #ifdef CONFIG_BSD_PROCESS_ACCT
65 #include <linux/acct.h>
66 #endif
67 #ifdef CONFIG_RT_MUTEXES
68 #include <linux/rtmutex.h>
69 #endif
70 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
71 #include <linux/lockdep.h>
72 #endif
73 #ifdef CONFIG_CHR_DEV_SG
74 #include <scsi/sg.h>
75 #endif
76
77 #ifdef CONFIG_LOCKUP_DETECTOR
78 #include <linux/nmi.h>
79 #endif
80
81
82 #if defined(CONFIG_SYSCTL)
83
84 /* External variables not in a header file. */
85 extern int sysctl_overcommit_memory;
86 extern int sysctl_overcommit_ratio;
87 extern int sysctl_panic_on_oom;
88 extern int sysctl_oom_kill_allocating_task;
89 extern int sysctl_oom_dump_tasks;
90 extern int max_threads;
91 extern int core_uses_pid;
92 extern int suid_dumpable;
93 extern char core_pattern[];
94 extern unsigned int core_pipe_limit;
95 extern int pid_max;
96 extern int min_free_kbytes;
97 extern int pid_max_min, pid_max_max;
98 extern int sysctl_drop_caches;
99 extern int percpu_pagelist_fraction;
100 extern int compat_log;
101 extern int latencytop_enabled;
102 extern int sysctl_nr_open_min, sysctl_nr_open_max;
103 #ifndef CONFIG_MMU
104 extern int sysctl_nr_trim_pages;
105 #endif
106 #ifdef CONFIG_BLOCK
107 extern int blk_iopoll_enabled;
108 #endif
109
110 /* Constants used for minimum and maximum */
111 #ifdef CONFIG_DETECT_SOFTLOCKUP
112 static int sixty = 60;
113 static int neg_one = -1;
114 #endif
115
116 static int zero;
117 static int __maybe_unused one = 1;
118 static int __maybe_unused two = 2;
119 static unsigned long one_ul = 1;
120 static int one_hundred = 100;
121 #ifdef CONFIG_PRINTK
122 static int ten_thousand = 10000;
123 #endif
124
125 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
126 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
127
128 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
129 static int maxolduid = 65535;
130 static int minolduid;
131 static int min_percpu_pagelist_fract = 8;
132
133 static int ngroups_max = NGROUPS_MAX;
134
135 #ifdef CONFIG_SPARC
136 #include <asm/system.h>
137 #endif
138
139 #ifdef CONFIG_SPARC64
140 extern int sysctl_tsb_ratio;
141 #endif
142
143 #ifdef __hppa__
144 extern int pwrsw_enabled;
145 extern int unaligned_enabled;
146 #endif
147
148 #ifdef CONFIG_S390
149 #ifdef CONFIG_MATHEMU
150 extern int sysctl_ieee_emulation_warnings;
151 #endif
152 extern int sysctl_userprocess_debug;
153 extern int spin_retry;
154 #endif
155
156 #ifdef CONFIG_IA64
157 extern int no_unaligned_warning;
158 extern int unaligned_dump_stack;
159 #endif
160
161 extern struct ratelimit_state printk_ratelimit_state;
162
163 #ifdef CONFIG_PROC_SYSCTL
164 static int proc_do_cad_pid(struct ctl_table *table, int write,
165 void __user *buffer, size_t *lenp, loff_t *ppos);
166 static int proc_taint(struct ctl_table *table, int write,
167 void __user *buffer, size_t *lenp, loff_t *ppos);
168 #endif
169
170 static struct ctl_table root_table[];
171 static struct ctl_table_root sysctl_table_root;
172 static struct ctl_table_header root_table_header = {
173 .count = 1,
174 .ctl_table = root_table,
175 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
176 .root = &sysctl_table_root,
177 .set = &sysctl_table_root.default_set,
178 };
179 static struct ctl_table_root sysctl_table_root = {
180 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
181 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
182 };
183
184 static struct ctl_table kern_table[];
185 static struct ctl_table vm_table[];
186 static struct ctl_table fs_table[];
187 static struct ctl_table debug_table[];
188 static struct ctl_table dev_table[];
189 extern struct ctl_table random_table[];
190 #ifdef CONFIG_INOTIFY_USER
191 extern struct ctl_table inotify_table[];
192 #endif
193 #ifdef CONFIG_EPOLL
194 extern struct ctl_table epoll_table[];
195 #endif
196
197 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
198 int sysctl_legacy_va_layout;
199 #endif
200
201 /* The default sysctl tables: */
202
203 static struct ctl_table root_table[] = {
204 {
205 .procname = "kernel",
206 .mode = 0555,
207 .child = kern_table,
208 },
209 {
210 .procname = "vm",
211 .mode = 0555,
212 .child = vm_table,
213 },
214 {
215 .procname = "fs",
216 .mode = 0555,
217 .child = fs_table,
218 },
219 {
220 .procname = "debug",
221 .mode = 0555,
222 .child = debug_table,
223 },
224 {
225 .procname = "dev",
226 .mode = 0555,
227 .child = dev_table,
228 },
229 /*
230 * NOTE: do not add new entries to this table unless you have read
231 * Documentation/sysctl/ctl_unnumbered.txt
232 */
233 { }
234 };
235
236 #ifdef CONFIG_SCHED_DEBUG
237 static int min_sched_granularity_ns = 100000; /* 100 usecs */
238 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
239 static int min_wakeup_granularity_ns; /* 0 usecs */
240 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
241 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
242 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
243 static int min_sched_shares_ratelimit = 100000; /* 100 usec */
244 static int max_sched_shares_ratelimit = NSEC_PER_SEC; /* 1 second */
245 #endif
246
247 static struct ctl_table kern_table[] = {
248 {
249 .procname = "sched_child_runs_first",
250 .data = &sysctl_sched_child_runs_first,
251 .maxlen = sizeof(unsigned int),
252 .mode = 0644,
253 .proc_handler = proc_dointvec,
254 },
255 #ifdef CONFIG_SCHED_DEBUG
256 {
257 .procname = "sched_min_granularity_ns",
258 .data = &sysctl_sched_min_granularity,
259 .maxlen = sizeof(unsigned int),
260 .mode = 0644,
261 .proc_handler = sched_proc_update_handler,
262 .extra1 = &min_sched_granularity_ns,
263 .extra2 = &max_sched_granularity_ns,
264 },
265 {
266 .procname = "sched_latency_ns",
267 .data = &sysctl_sched_latency,
268 .maxlen = sizeof(unsigned int),
269 .mode = 0644,
270 .proc_handler = sched_proc_update_handler,
271 .extra1 = &min_sched_granularity_ns,
272 .extra2 = &max_sched_granularity_ns,
273 },
274 {
275 .procname = "sched_wakeup_granularity_ns",
276 .data = &sysctl_sched_wakeup_granularity,
277 .maxlen = sizeof(unsigned int),
278 .mode = 0644,
279 .proc_handler = sched_proc_update_handler,
280 .extra1 = &min_wakeup_granularity_ns,
281 .extra2 = &max_wakeup_granularity_ns,
282 },
283 {
284 .procname = "sched_shares_ratelimit",
285 .data = &sysctl_sched_shares_ratelimit,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_shares_ratelimit,
290 .extra2 = &max_sched_shares_ratelimit,
291 },
292 {
293 .procname = "sched_tunable_scaling",
294 .data = &sysctl_sched_tunable_scaling,
295 .maxlen = sizeof(enum sched_tunable_scaling),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_tunable_scaling,
299 .extra2 = &max_sched_tunable_scaling,
300 },
301 {
302 .procname = "sched_shares_thresh",
303 .data = &sysctl_sched_shares_thresh,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = proc_dointvec_minmax,
307 .extra1 = &zero,
308 },
309 {
310 .procname = "sched_migration_cost",
311 .data = &sysctl_sched_migration_cost,
312 .maxlen = sizeof(unsigned int),
313 .mode = 0644,
314 .proc_handler = proc_dointvec,
315 },
316 {
317 .procname = "sched_nr_migrate",
318 .data = &sysctl_sched_nr_migrate,
319 .maxlen = sizeof(unsigned int),
320 .mode = 0644,
321 .proc_handler = proc_dointvec,
322 },
323 {
324 .procname = "sched_time_avg",
325 .data = &sysctl_sched_time_avg,
326 .maxlen = sizeof(unsigned int),
327 .mode = 0644,
328 .proc_handler = proc_dointvec,
329 },
330 {
331 .procname = "timer_migration",
332 .data = &sysctl_timer_migration,
333 .maxlen = sizeof(unsigned int),
334 .mode = 0644,
335 .proc_handler = proc_dointvec_minmax,
336 .extra1 = &zero,
337 .extra2 = &one,
338 },
339 #endif
340 {
341 .procname = "sched_rt_period_us",
342 .data = &sysctl_sched_rt_period,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = sched_rt_handler,
346 },
347 {
348 .procname = "sched_rt_runtime_us",
349 .data = &sysctl_sched_rt_runtime,
350 .maxlen = sizeof(int),
351 .mode = 0644,
352 .proc_handler = sched_rt_handler,
353 },
354 {
355 .procname = "sched_compat_yield",
356 .data = &sysctl_sched_compat_yield,
357 .maxlen = sizeof(unsigned int),
358 .mode = 0644,
359 .proc_handler = proc_dointvec,
360 },
361 #ifdef CONFIG_PROVE_LOCKING
362 {
363 .procname = "prove_locking",
364 .data = &prove_locking,
365 .maxlen = sizeof(int),
366 .mode = 0644,
367 .proc_handler = proc_dointvec,
368 },
369 #endif
370 #ifdef CONFIG_LOCK_STAT
371 {
372 .procname = "lock_stat",
373 .data = &lock_stat,
374 .maxlen = sizeof(int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec,
377 },
378 #endif
379 {
380 .procname = "panic",
381 .data = &panic_timeout,
382 .maxlen = sizeof(int),
383 .mode = 0644,
384 .proc_handler = proc_dointvec,
385 },
386 {
387 .procname = "core_uses_pid",
388 .data = &core_uses_pid,
389 .maxlen = sizeof(int),
390 .mode = 0644,
391 .proc_handler = proc_dointvec,
392 },
393 {
394 .procname = "core_pattern",
395 .data = core_pattern,
396 .maxlen = CORENAME_MAX_SIZE,
397 .mode = 0644,
398 .proc_handler = proc_dostring,
399 },
400 {
401 .procname = "core_pipe_limit",
402 .data = &core_pipe_limit,
403 .maxlen = sizeof(unsigned int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 #ifdef CONFIG_PROC_SYSCTL
408 {
409 .procname = "tainted",
410 .maxlen = sizeof(long),
411 .mode = 0644,
412 .proc_handler = proc_taint,
413 },
414 #endif
415 #ifdef CONFIG_LATENCYTOP
416 {
417 .procname = "latencytop",
418 .data = &latencytop_enabled,
419 .maxlen = sizeof(int),
420 .mode = 0644,
421 .proc_handler = proc_dointvec,
422 },
423 #endif
424 #ifdef CONFIG_BLK_DEV_INITRD
425 {
426 .procname = "real-root-dev",
427 .data = &real_root_dev,
428 .maxlen = sizeof(int),
429 .mode = 0644,
430 .proc_handler = proc_dointvec,
431 },
432 #endif
433 {
434 .procname = "print-fatal-signals",
435 .data = &print_fatal_signals,
436 .maxlen = sizeof(int),
437 .mode = 0644,
438 .proc_handler = proc_dointvec,
439 },
440 #ifdef CONFIG_SPARC
441 {
442 .procname = "reboot-cmd",
443 .data = reboot_command,
444 .maxlen = 256,
445 .mode = 0644,
446 .proc_handler = proc_dostring,
447 },
448 {
449 .procname = "stop-a",
450 .data = &stop_a_enabled,
451 .maxlen = sizeof (int),
452 .mode = 0644,
453 .proc_handler = proc_dointvec,
454 },
455 {
456 .procname = "scons-poweroff",
457 .data = &scons_pwroff,
458 .maxlen = sizeof (int),
459 .mode = 0644,
460 .proc_handler = proc_dointvec,
461 },
462 #endif
463 #ifdef CONFIG_SPARC64
464 {
465 .procname = "tsb-ratio",
466 .data = &sysctl_tsb_ratio,
467 .maxlen = sizeof (int),
468 .mode = 0644,
469 .proc_handler = proc_dointvec,
470 },
471 #endif
472 #ifdef __hppa__
473 {
474 .procname = "soft-power",
475 .data = &pwrsw_enabled,
476 .maxlen = sizeof (int),
477 .mode = 0644,
478 .proc_handler = proc_dointvec,
479 },
480 {
481 .procname = "unaligned-trap",
482 .data = &unaligned_enabled,
483 .maxlen = sizeof (int),
484 .mode = 0644,
485 .proc_handler = proc_dointvec,
486 },
487 #endif
488 {
489 .procname = "ctrl-alt-del",
490 .data = &C_A_D,
491 .maxlen = sizeof(int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495 #ifdef CONFIG_FUNCTION_TRACER
496 {
497 .procname = "ftrace_enabled",
498 .data = &ftrace_enabled,
499 .maxlen = sizeof(int),
500 .mode = 0644,
501 .proc_handler = ftrace_enable_sysctl,
502 },
503 #endif
504 #ifdef CONFIG_STACK_TRACER
505 {
506 .procname = "stack_tracer_enabled",
507 .data = &stack_tracer_enabled,
508 .maxlen = sizeof(int),
509 .mode = 0644,
510 .proc_handler = stack_trace_sysctl,
511 },
512 #endif
513 #ifdef CONFIG_TRACING
514 {
515 .procname = "ftrace_dump_on_oops",
516 .data = &ftrace_dump_on_oops,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521 #endif
522 #ifdef CONFIG_MODULES
523 {
524 .procname = "modprobe",
525 .data = &modprobe_path,
526 .maxlen = KMOD_PATH_LEN,
527 .mode = 0644,
528 .proc_handler = proc_dostring,
529 },
530 {
531 .procname = "modules_disabled",
532 .data = &modules_disabled,
533 .maxlen = sizeof(int),
534 .mode = 0644,
535 /* only handle a transition from default "0" to "1" */
536 .proc_handler = proc_dointvec_minmax,
537 .extra1 = &one,
538 .extra2 = &one,
539 },
540 #endif
541 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
542 {
543 .procname = "hotplug",
544 .data = &uevent_helper,
545 .maxlen = UEVENT_HELPER_PATH_LEN,
546 .mode = 0644,
547 .proc_handler = proc_dostring,
548 },
549 #endif
550 #ifdef CONFIG_CHR_DEV_SG
551 {
552 .procname = "sg-big-buff",
553 .data = &sg_big_buff,
554 .maxlen = sizeof (int),
555 .mode = 0444,
556 .proc_handler = proc_dointvec,
557 },
558 #endif
559 #ifdef CONFIG_BSD_PROCESS_ACCT
560 {
561 .procname = "acct",
562 .data = &acct_parm,
563 .maxlen = 3*sizeof(int),
564 .mode = 0644,
565 .proc_handler = proc_dointvec,
566 },
567 #endif
568 #ifdef CONFIG_MAGIC_SYSRQ
569 {
570 .procname = "sysrq",
571 .data = &__sysrq_enabled,
572 .maxlen = sizeof (int),
573 .mode = 0644,
574 .proc_handler = proc_dointvec,
575 },
576 #endif
577 #ifdef CONFIG_PROC_SYSCTL
578 {
579 .procname = "cad_pid",
580 .data = NULL,
581 .maxlen = sizeof (int),
582 .mode = 0600,
583 .proc_handler = proc_do_cad_pid,
584 },
585 #endif
586 {
587 .procname = "threads-max",
588 .data = &max_threads,
589 .maxlen = sizeof(int),
590 .mode = 0644,
591 .proc_handler = proc_dointvec,
592 },
593 {
594 .procname = "random",
595 .mode = 0555,
596 .child = random_table,
597 },
598 {
599 .procname = "overflowuid",
600 .data = &overflowuid,
601 .maxlen = sizeof(int),
602 .mode = 0644,
603 .proc_handler = proc_dointvec_minmax,
604 .extra1 = &minolduid,
605 .extra2 = &maxolduid,
606 },
607 {
608 .procname = "overflowgid",
609 .data = &overflowgid,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec_minmax,
613 .extra1 = &minolduid,
614 .extra2 = &maxolduid,
615 },
616 #ifdef CONFIG_S390
617 #ifdef CONFIG_MATHEMU
618 {
619 .procname = "ieee_emulation_warnings",
620 .data = &sysctl_ieee_emulation_warnings,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = proc_dointvec,
624 },
625 #endif
626 {
627 .procname = "userprocess_debug",
628 .data = &sysctl_userprocess_debug,
629 .maxlen = sizeof(int),
630 .mode = 0644,
631 .proc_handler = proc_dointvec,
632 },
633 #endif
634 {
635 .procname = "pid_max",
636 .data = &pid_max,
637 .maxlen = sizeof (int),
638 .mode = 0644,
639 .proc_handler = proc_dointvec_minmax,
640 .extra1 = &pid_max_min,
641 .extra2 = &pid_max_max,
642 },
643 {
644 .procname = "panic_on_oops",
645 .data = &panic_on_oops,
646 .maxlen = sizeof(int),
647 .mode = 0644,
648 .proc_handler = proc_dointvec,
649 },
650 #if defined CONFIG_PRINTK
651 {
652 .procname = "printk",
653 .data = &console_loglevel,
654 .maxlen = 4*sizeof(int),
655 .mode = 0644,
656 .proc_handler = proc_dointvec,
657 },
658 {
659 .procname = "printk_ratelimit",
660 .data = &printk_ratelimit_state.interval,
661 .maxlen = sizeof(int),
662 .mode = 0644,
663 .proc_handler = proc_dointvec_jiffies,
664 },
665 {
666 .procname = "printk_ratelimit_burst",
667 .data = &printk_ratelimit_state.burst,
668 .maxlen = sizeof(int),
669 .mode = 0644,
670 .proc_handler = proc_dointvec,
671 },
672 {
673 .procname = "printk_delay",
674 .data = &printk_delay_msec,
675 .maxlen = sizeof(int),
676 .mode = 0644,
677 .proc_handler = proc_dointvec_minmax,
678 .extra1 = &zero,
679 .extra2 = &ten_thousand,
680 },
681 #endif
682 {
683 .procname = "ngroups_max",
684 .data = &ngroups_max,
685 .maxlen = sizeof (int),
686 .mode = 0444,
687 .proc_handler = proc_dointvec,
688 },
689 #if defined(CONFIG_LOCKUP_DETECTOR)
690 {
691 .procname = "watchdog",
692 .data = &watchdog_enabled,
693 .maxlen = sizeof (int),
694 .mode = 0644,
695 .proc_handler = proc_dowatchdog_enabled,
696 },
697 {
698 .procname = "watchdog_thresh",
699 .data = &softlockup_thresh,
700 .maxlen = sizeof(int),
701 .mode = 0644,
702 .proc_handler = proc_dowatchdog_thresh,
703 .extra1 = &neg_one,
704 .extra2 = &sixty,
705 },
706 #endif
707 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86) && !defined(CONFIG_LOCKUP_DETECTOR)
708 {
709 .procname = "unknown_nmi_panic",
710 .data = &unknown_nmi_panic,
711 .maxlen = sizeof (int),
712 .mode = 0644,
713 .proc_handler = proc_dointvec,
714 },
715 {
716 .procname = "nmi_watchdog",
717 .data = &nmi_watchdog_enabled,
718 .maxlen = sizeof (int),
719 .mode = 0644,
720 .proc_handler = proc_nmi_enabled,
721 },
722 #endif
723 #if defined(CONFIG_X86)
724 {
725 .procname = "panic_on_unrecovered_nmi",
726 .data = &panic_on_unrecovered_nmi,
727 .maxlen = sizeof(int),
728 .mode = 0644,
729 .proc_handler = proc_dointvec,
730 },
731 {
732 .procname = "panic_on_io_nmi",
733 .data = &panic_on_io_nmi,
734 .maxlen = sizeof(int),
735 .mode = 0644,
736 .proc_handler = proc_dointvec,
737 },
738 {
739 .procname = "bootloader_type",
740 .data = &bootloader_type,
741 .maxlen = sizeof (int),
742 .mode = 0444,
743 .proc_handler = proc_dointvec,
744 },
745 {
746 .procname = "bootloader_version",
747 .data = &bootloader_version,
748 .maxlen = sizeof (int),
749 .mode = 0444,
750 .proc_handler = proc_dointvec,
751 },
752 {
753 .procname = "kstack_depth_to_print",
754 .data = &kstack_depth_to_print,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec,
758 },
759 {
760 .procname = "io_delay_type",
761 .data = &io_delay_type,
762 .maxlen = sizeof(int),
763 .mode = 0644,
764 .proc_handler = proc_dointvec,
765 },
766 #endif
767 #if defined(CONFIG_MMU)
768 {
769 .procname = "randomize_va_space",
770 .data = &randomize_va_space,
771 .maxlen = sizeof(int),
772 .mode = 0644,
773 .proc_handler = proc_dointvec,
774 },
775 #endif
776 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
777 {
778 .procname = "spin_retry",
779 .data = &spin_retry,
780 .maxlen = sizeof (int),
781 .mode = 0644,
782 .proc_handler = proc_dointvec,
783 },
784 #endif
785 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
786 {
787 .procname = "acpi_video_flags",
788 .data = &acpi_realmode_flags,
789 .maxlen = sizeof (unsigned long),
790 .mode = 0644,
791 .proc_handler = proc_doulongvec_minmax,
792 },
793 #endif
794 #ifdef CONFIG_IA64
795 {
796 .procname = "ignore-unaligned-usertrap",
797 .data = &no_unaligned_warning,
798 .maxlen = sizeof (int),
799 .mode = 0644,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "unaligned-dump-stack",
804 .data = &unaligned_dump_stack,
805 .maxlen = sizeof (int),
806 .mode = 0644,
807 .proc_handler = proc_dointvec,
808 },
809 #endif
810 #ifdef CONFIG_DETECT_SOFTLOCKUP
811 {
812 .procname = "softlockup_panic",
813 .data = &softlockup_panic,
814 .maxlen = sizeof(int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec_minmax,
817 .extra1 = &zero,
818 .extra2 = &one,
819 },
820 {
821 .procname = "softlockup_thresh",
822 .data = &softlockup_thresh,
823 .maxlen = sizeof(int),
824 .mode = 0644,
825 .proc_handler = proc_dosoftlockup_thresh,
826 .extra1 = &neg_one,
827 .extra2 = &sixty,
828 },
829 #endif
830 #ifdef CONFIG_DETECT_HUNG_TASK
831 {
832 .procname = "hung_task_panic",
833 .data = &sysctl_hung_task_panic,
834 .maxlen = sizeof(int),
835 .mode = 0644,
836 .proc_handler = proc_dointvec_minmax,
837 .extra1 = &zero,
838 .extra2 = &one,
839 },
840 {
841 .procname = "hung_task_check_count",
842 .data = &sysctl_hung_task_check_count,
843 .maxlen = sizeof(unsigned long),
844 .mode = 0644,
845 .proc_handler = proc_doulongvec_minmax,
846 },
847 {
848 .procname = "hung_task_timeout_secs",
849 .data = &sysctl_hung_task_timeout_secs,
850 .maxlen = sizeof(unsigned long),
851 .mode = 0644,
852 .proc_handler = proc_dohung_task_timeout_secs,
853 },
854 {
855 .procname = "hung_task_warnings",
856 .data = &sysctl_hung_task_warnings,
857 .maxlen = sizeof(unsigned long),
858 .mode = 0644,
859 .proc_handler = proc_doulongvec_minmax,
860 },
861 #endif
862 #ifdef CONFIG_COMPAT
863 {
864 .procname = "compat-log",
865 .data = &compat_log,
866 .maxlen = sizeof (int),
867 .mode = 0644,
868 .proc_handler = proc_dointvec,
869 },
870 #endif
871 #ifdef CONFIG_RT_MUTEXES
872 {
873 .procname = "max_lock_depth",
874 .data = &max_lock_depth,
875 .maxlen = sizeof(int),
876 .mode = 0644,
877 .proc_handler = proc_dointvec,
878 },
879 #endif
880 {
881 .procname = "poweroff_cmd",
882 .data = &poweroff_cmd,
883 .maxlen = POWEROFF_CMD_PATH_LEN,
884 .mode = 0644,
885 .proc_handler = proc_dostring,
886 },
887 #ifdef CONFIG_KEYS
888 {
889 .procname = "keys",
890 .mode = 0555,
891 .child = key_sysctls,
892 },
893 #endif
894 #ifdef CONFIG_RCU_TORTURE_TEST
895 {
896 .procname = "rcutorture_runnable",
897 .data = &rcutorture_runnable,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_dointvec,
901 },
902 #endif
903 #ifdef CONFIG_SLOW_WORK
904 {
905 .procname = "slow-work",
906 .mode = 0555,
907 .child = slow_work_sysctls,
908 },
909 #endif
910 #ifdef CONFIG_PERF_EVENTS
911 {
912 .procname = "perf_event_paranoid",
913 .data = &sysctl_perf_event_paranoid,
914 .maxlen = sizeof(sysctl_perf_event_paranoid),
915 .mode = 0644,
916 .proc_handler = proc_dointvec,
917 },
918 {
919 .procname = "perf_event_mlock_kb",
920 .data = &sysctl_perf_event_mlock,
921 .maxlen = sizeof(sysctl_perf_event_mlock),
922 .mode = 0644,
923 .proc_handler = proc_dointvec,
924 },
925 {
926 .procname = "perf_event_max_sample_rate",
927 .data = &sysctl_perf_event_sample_rate,
928 .maxlen = sizeof(sysctl_perf_event_sample_rate),
929 .mode = 0644,
930 .proc_handler = proc_dointvec,
931 },
932 #endif
933 #ifdef CONFIG_KMEMCHECK
934 {
935 .procname = "kmemcheck",
936 .data = &kmemcheck_enabled,
937 .maxlen = sizeof(int),
938 .mode = 0644,
939 .proc_handler = proc_dointvec,
940 },
941 #endif
942 #ifdef CONFIG_BLOCK
943 {
944 .procname = "blk_iopoll",
945 .data = &blk_iopoll_enabled,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = proc_dointvec,
949 },
950 #endif
951 /*
952 * NOTE: do not add new entries to this table unless you have read
953 * Documentation/sysctl/ctl_unnumbered.txt
954 */
955 { }
956 };
957
958 static struct ctl_table vm_table[] = {
959 {
960 .procname = "overcommit_memory",
961 .data = &sysctl_overcommit_memory,
962 .maxlen = sizeof(sysctl_overcommit_memory),
963 .mode = 0644,
964 .proc_handler = proc_dointvec,
965 },
966 {
967 .procname = "panic_on_oom",
968 .data = &sysctl_panic_on_oom,
969 .maxlen = sizeof(sysctl_panic_on_oom),
970 .mode = 0644,
971 .proc_handler = proc_dointvec,
972 },
973 {
974 .procname = "oom_kill_allocating_task",
975 .data = &sysctl_oom_kill_allocating_task,
976 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
977 .mode = 0644,
978 .proc_handler = proc_dointvec,
979 },
980 {
981 .procname = "oom_dump_tasks",
982 .data = &sysctl_oom_dump_tasks,
983 .maxlen = sizeof(sysctl_oom_dump_tasks),
984 .mode = 0644,
985 .proc_handler = proc_dointvec,
986 },
987 {
988 .procname = "overcommit_ratio",
989 .data = &sysctl_overcommit_ratio,
990 .maxlen = sizeof(sysctl_overcommit_ratio),
991 .mode = 0644,
992 .proc_handler = proc_dointvec,
993 },
994 {
995 .procname = "page-cluster",
996 .data = &page_cluster,
997 .maxlen = sizeof(int),
998 .mode = 0644,
999 .proc_handler = proc_dointvec,
1000 },
1001 {
1002 .procname = "dirty_background_ratio",
1003 .data = &dirty_background_ratio,
1004 .maxlen = sizeof(dirty_background_ratio),
1005 .mode = 0644,
1006 .proc_handler = dirty_background_ratio_handler,
1007 .extra1 = &zero,
1008 .extra2 = &one_hundred,
1009 },
1010 {
1011 .procname = "dirty_background_bytes",
1012 .data = &dirty_background_bytes,
1013 .maxlen = sizeof(dirty_background_bytes),
1014 .mode = 0644,
1015 .proc_handler = dirty_background_bytes_handler,
1016 .extra1 = &one_ul,
1017 },
1018 {
1019 .procname = "dirty_ratio",
1020 .data = &vm_dirty_ratio,
1021 .maxlen = sizeof(vm_dirty_ratio),
1022 .mode = 0644,
1023 .proc_handler = dirty_ratio_handler,
1024 .extra1 = &zero,
1025 .extra2 = &one_hundred,
1026 },
1027 {
1028 .procname = "dirty_bytes",
1029 .data = &vm_dirty_bytes,
1030 .maxlen = sizeof(vm_dirty_bytes),
1031 .mode = 0644,
1032 .proc_handler = dirty_bytes_handler,
1033 .extra1 = &dirty_bytes_min,
1034 },
1035 {
1036 .procname = "dirty_writeback_centisecs",
1037 .data = &dirty_writeback_interval,
1038 .maxlen = sizeof(dirty_writeback_interval),
1039 .mode = 0644,
1040 .proc_handler = dirty_writeback_centisecs_handler,
1041 },
1042 {
1043 .procname = "dirty_expire_centisecs",
1044 .data = &dirty_expire_interval,
1045 .maxlen = sizeof(dirty_expire_interval),
1046 .mode = 0644,
1047 .proc_handler = proc_dointvec,
1048 },
1049 {
1050 .procname = "nr_pdflush_threads",
1051 .data = &nr_pdflush_threads,
1052 .maxlen = sizeof nr_pdflush_threads,
1053 .mode = 0444 /* read-only*/,
1054 .proc_handler = proc_dointvec,
1055 },
1056 {
1057 .procname = "swappiness",
1058 .data = &vm_swappiness,
1059 .maxlen = sizeof(vm_swappiness),
1060 .mode = 0644,
1061 .proc_handler = proc_dointvec_minmax,
1062 .extra1 = &zero,
1063 .extra2 = &one_hundred,
1064 },
1065 #ifdef CONFIG_HUGETLB_PAGE
1066 {
1067 .procname = "nr_hugepages",
1068 .data = NULL,
1069 .maxlen = sizeof(unsigned long),
1070 .mode = 0644,
1071 .proc_handler = hugetlb_sysctl_handler,
1072 .extra1 = (void *)&hugetlb_zero,
1073 .extra2 = (void *)&hugetlb_infinity,
1074 },
1075 #ifdef CONFIG_NUMA
1076 {
1077 .procname = "nr_hugepages_mempolicy",
1078 .data = NULL,
1079 .maxlen = sizeof(unsigned long),
1080 .mode = 0644,
1081 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1082 .extra1 = (void *)&hugetlb_zero,
1083 .extra2 = (void *)&hugetlb_infinity,
1084 },
1085 #endif
1086 {
1087 .procname = "hugetlb_shm_group",
1088 .data = &sysctl_hugetlb_shm_group,
1089 .maxlen = sizeof(gid_t),
1090 .mode = 0644,
1091 .proc_handler = proc_dointvec,
1092 },
1093 {
1094 .procname = "hugepages_treat_as_movable",
1095 .data = &hugepages_treat_as_movable,
1096 .maxlen = sizeof(int),
1097 .mode = 0644,
1098 .proc_handler = hugetlb_treat_movable_handler,
1099 },
1100 {
1101 .procname = "nr_overcommit_hugepages",
1102 .data = NULL,
1103 .maxlen = sizeof(unsigned long),
1104 .mode = 0644,
1105 .proc_handler = hugetlb_overcommit_handler,
1106 .extra1 = (void *)&hugetlb_zero,
1107 .extra2 = (void *)&hugetlb_infinity,
1108 },
1109 #endif
1110 {
1111 .procname = "lowmem_reserve_ratio",
1112 .data = &sysctl_lowmem_reserve_ratio,
1113 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1114 .mode = 0644,
1115 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1116 },
1117 {
1118 .procname = "drop_caches",
1119 .data = &sysctl_drop_caches,
1120 .maxlen = sizeof(int),
1121 .mode = 0644,
1122 .proc_handler = drop_caches_sysctl_handler,
1123 },
1124 {
1125 .procname = "min_free_kbytes",
1126 .data = &min_free_kbytes,
1127 .maxlen = sizeof(min_free_kbytes),
1128 .mode = 0644,
1129 .proc_handler = min_free_kbytes_sysctl_handler,
1130 .extra1 = &zero,
1131 },
1132 {
1133 .procname = "percpu_pagelist_fraction",
1134 .data = &percpu_pagelist_fraction,
1135 .maxlen = sizeof(percpu_pagelist_fraction),
1136 .mode = 0644,
1137 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1138 .extra1 = &min_percpu_pagelist_fract,
1139 },
1140 #ifdef CONFIG_MMU
1141 {
1142 .procname = "max_map_count",
1143 .data = &sysctl_max_map_count,
1144 .maxlen = sizeof(sysctl_max_map_count),
1145 .mode = 0644,
1146 .proc_handler = proc_dointvec_minmax,
1147 .extra1 = &zero,
1148 },
1149 #else
1150 {
1151 .procname = "nr_trim_pages",
1152 .data = &sysctl_nr_trim_pages,
1153 .maxlen = sizeof(sysctl_nr_trim_pages),
1154 .mode = 0644,
1155 .proc_handler = proc_dointvec_minmax,
1156 .extra1 = &zero,
1157 },
1158 #endif
1159 {
1160 .procname = "laptop_mode",
1161 .data = &laptop_mode,
1162 .maxlen = sizeof(laptop_mode),
1163 .mode = 0644,
1164 .proc_handler = proc_dointvec_jiffies,
1165 },
1166 {
1167 .procname = "block_dump",
1168 .data = &block_dump,
1169 .maxlen = sizeof(block_dump),
1170 .mode = 0644,
1171 .proc_handler = proc_dointvec,
1172 .extra1 = &zero,
1173 },
1174 {
1175 .procname = "vfs_cache_pressure",
1176 .data = &sysctl_vfs_cache_pressure,
1177 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1178 .mode = 0644,
1179 .proc_handler = proc_dointvec,
1180 .extra1 = &zero,
1181 },
1182 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1183 {
1184 .procname = "legacy_va_layout",
1185 .data = &sysctl_legacy_va_layout,
1186 .maxlen = sizeof(sysctl_legacy_va_layout),
1187 .mode = 0644,
1188 .proc_handler = proc_dointvec,
1189 .extra1 = &zero,
1190 },
1191 #endif
1192 #ifdef CONFIG_NUMA
1193 {
1194 .procname = "zone_reclaim_mode",
1195 .data = &zone_reclaim_mode,
1196 .maxlen = sizeof(zone_reclaim_mode),
1197 .mode = 0644,
1198 .proc_handler = proc_dointvec,
1199 .extra1 = &zero,
1200 },
1201 {
1202 .procname = "min_unmapped_ratio",
1203 .data = &sysctl_min_unmapped_ratio,
1204 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1205 .mode = 0644,
1206 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1207 .extra1 = &zero,
1208 .extra2 = &one_hundred,
1209 },
1210 {
1211 .procname = "min_slab_ratio",
1212 .data = &sysctl_min_slab_ratio,
1213 .maxlen = sizeof(sysctl_min_slab_ratio),
1214 .mode = 0644,
1215 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1216 .extra1 = &zero,
1217 .extra2 = &one_hundred,
1218 },
1219 #endif
1220 #ifdef CONFIG_SMP
1221 {
1222 .procname = "stat_interval",
1223 .data = &sysctl_stat_interval,
1224 .maxlen = sizeof(sysctl_stat_interval),
1225 .mode = 0644,
1226 .proc_handler = proc_dointvec_jiffies,
1227 },
1228 #endif
1229 #ifdef CONFIG_MMU
1230 {
1231 .procname = "mmap_min_addr",
1232 .data = &dac_mmap_min_addr,
1233 .maxlen = sizeof(unsigned long),
1234 .mode = 0644,
1235 .proc_handler = mmap_min_addr_handler,
1236 },
1237 #endif
1238 #ifdef CONFIG_NUMA
1239 {
1240 .procname = "numa_zonelist_order",
1241 .data = &numa_zonelist_order,
1242 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1243 .mode = 0644,
1244 .proc_handler = numa_zonelist_order_handler,
1245 },
1246 #endif
1247 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1248 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1249 {
1250 .procname = "vdso_enabled",
1251 .data = &vdso_enabled,
1252 .maxlen = sizeof(vdso_enabled),
1253 .mode = 0644,
1254 .proc_handler = proc_dointvec,
1255 .extra1 = &zero,
1256 },
1257 #endif
1258 #ifdef CONFIG_HIGHMEM
1259 {
1260 .procname = "highmem_is_dirtyable",
1261 .data = &vm_highmem_is_dirtyable,
1262 .maxlen = sizeof(vm_highmem_is_dirtyable),
1263 .mode = 0644,
1264 .proc_handler = proc_dointvec_minmax,
1265 .extra1 = &zero,
1266 .extra2 = &one,
1267 },
1268 #endif
1269 {
1270 .procname = "scan_unevictable_pages",
1271 .data = &scan_unevictable_pages,
1272 .maxlen = sizeof(scan_unevictable_pages),
1273 .mode = 0644,
1274 .proc_handler = scan_unevictable_handler,
1275 },
1276 #ifdef CONFIG_MEMORY_FAILURE
1277 {
1278 .procname = "memory_failure_early_kill",
1279 .data = &sysctl_memory_failure_early_kill,
1280 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1281 .mode = 0644,
1282 .proc_handler = proc_dointvec_minmax,
1283 .extra1 = &zero,
1284 .extra2 = &one,
1285 },
1286 {
1287 .procname = "memory_failure_recovery",
1288 .data = &sysctl_memory_failure_recovery,
1289 .maxlen = sizeof(sysctl_memory_failure_recovery),
1290 .mode = 0644,
1291 .proc_handler = proc_dointvec_minmax,
1292 .extra1 = &zero,
1293 .extra2 = &one,
1294 },
1295 #endif
1296
1297 /*
1298 * NOTE: do not add new entries to this table unless you have read
1299 * Documentation/sysctl/ctl_unnumbered.txt
1300 */
1301 { }
1302 };
1303
1304 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1305 static struct ctl_table binfmt_misc_table[] = {
1306 { }
1307 };
1308 #endif
1309
1310 static struct ctl_table fs_table[] = {
1311 {
1312 .procname = "inode-nr",
1313 .data = &inodes_stat,
1314 .maxlen = 2*sizeof(int),
1315 .mode = 0444,
1316 .proc_handler = proc_dointvec,
1317 },
1318 {
1319 .procname = "inode-state",
1320 .data = &inodes_stat,
1321 .maxlen = 7*sizeof(int),
1322 .mode = 0444,
1323 .proc_handler = proc_dointvec,
1324 },
1325 {
1326 .procname = "file-nr",
1327 .data = &files_stat,
1328 .maxlen = 3*sizeof(int),
1329 .mode = 0444,
1330 .proc_handler = proc_nr_files,
1331 },
1332 {
1333 .procname = "file-max",
1334 .data = &files_stat.max_files,
1335 .maxlen = sizeof(int),
1336 .mode = 0644,
1337 .proc_handler = proc_dointvec,
1338 },
1339 {
1340 .procname = "nr_open",
1341 .data = &sysctl_nr_open,
1342 .maxlen = sizeof(int),
1343 .mode = 0644,
1344 .proc_handler = proc_dointvec_minmax,
1345 .extra1 = &sysctl_nr_open_min,
1346 .extra2 = &sysctl_nr_open_max,
1347 },
1348 {
1349 .procname = "dentry-state",
1350 .data = &dentry_stat,
1351 .maxlen = 6*sizeof(int),
1352 .mode = 0444,
1353 .proc_handler = proc_dointvec,
1354 },
1355 {
1356 .procname = "overflowuid",
1357 .data = &fs_overflowuid,
1358 .maxlen = sizeof(int),
1359 .mode = 0644,
1360 .proc_handler = proc_dointvec_minmax,
1361 .extra1 = &minolduid,
1362 .extra2 = &maxolduid,
1363 },
1364 {
1365 .procname = "overflowgid",
1366 .data = &fs_overflowgid,
1367 .maxlen = sizeof(int),
1368 .mode = 0644,
1369 .proc_handler = proc_dointvec_minmax,
1370 .extra1 = &minolduid,
1371 .extra2 = &maxolduid,
1372 },
1373 #ifdef CONFIG_FILE_LOCKING
1374 {
1375 .procname = "leases-enable",
1376 .data = &leases_enable,
1377 .maxlen = sizeof(int),
1378 .mode = 0644,
1379 .proc_handler = proc_dointvec,
1380 },
1381 #endif
1382 #ifdef CONFIG_DNOTIFY
1383 {
1384 .procname = "dir-notify-enable",
1385 .data = &dir_notify_enable,
1386 .maxlen = sizeof(int),
1387 .mode = 0644,
1388 .proc_handler = proc_dointvec,
1389 },
1390 #endif
1391 #ifdef CONFIG_MMU
1392 #ifdef CONFIG_FILE_LOCKING
1393 {
1394 .procname = "lease-break-time",
1395 .data = &lease_break_time,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec,
1399 },
1400 #endif
1401 #ifdef CONFIG_AIO
1402 {
1403 .procname = "aio-nr",
1404 .data = &aio_nr,
1405 .maxlen = sizeof(aio_nr),
1406 .mode = 0444,
1407 .proc_handler = proc_doulongvec_minmax,
1408 },
1409 {
1410 .procname = "aio-max-nr",
1411 .data = &aio_max_nr,
1412 .maxlen = sizeof(aio_max_nr),
1413 .mode = 0644,
1414 .proc_handler = proc_doulongvec_minmax,
1415 },
1416 #endif /* CONFIG_AIO */
1417 #ifdef CONFIG_INOTIFY_USER
1418 {
1419 .procname = "inotify",
1420 .mode = 0555,
1421 .child = inotify_table,
1422 },
1423 #endif
1424 #ifdef CONFIG_EPOLL
1425 {
1426 .procname = "epoll",
1427 .mode = 0555,
1428 .child = epoll_table,
1429 },
1430 #endif
1431 #endif
1432 {
1433 .procname = "suid_dumpable",
1434 .data = &suid_dumpable,
1435 .maxlen = sizeof(int),
1436 .mode = 0644,
1437 .proc_handler = proc_dointvec_minmax,
1438 .extra1 = &zero,
1439 .extra2 = &two,
1440 },
1441 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1442 {
1443 .procname = "binfmt_misc",
1444 .mode = 0555,
1445 .child = binfmt_misc_table,
1446 },
1447 #endif
1448 /*
1449 * NOTE: do not add new entries to this table unless you have read
1450 * Documentation/sysctl/ctl_unnumbered.txt
1451 */
1452 { }
1453 };
1454
1455 static struct ctl_table debug_table[] = {
1456 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC)
1457 {
1458 .procname = "exception-trace",
1459 .data = &show_unhandled_signals,
1460 .maxlen = sizeof(int),
1461 .mode = 0644,
1462 .proc_handler = proc_dointvec
1463 },
1464 #endif
1465 #if defined(CONFIG_OPTPROBES)
1466 {
1467 .procname = "kprobes-optimization",
1468 .data = &sysctl_kprobes_optimization,
1469 .maxlen = sizeof(int),
1470 .mode = 0644,
1471 .proc_handler = proc_kprobes_optimization_handler,
1472 .extra1 = &zero,
1473 .extra2 = &one,
1474 },
1475 #endif
1476 { }
1477 };
1478
1479 static struct ctl_table dev_table[] = {
1480 { }
1481 };
1482
1483 static DEFINE_SPINLOCK(sysctl_lock);
1484
1485 /* called under sysctl_lock */
1486 static int use_table(struct ctl_table_header *p)
1487 {
1488 if (unlikely(p->unregistering))
1489 return 0;
1490 p->used++;
1491 return 1;
1492 }
1493
1494 /* called under sysctl_lock */
1495 static void unuse_table(struct ctl_table_header *p)
1496 {
1497 if (!--p->used)
1498 if (unlikely(p->unregistering))
1499 complete(p->unregistering);
1500 }
1501
1502 /* called under sysctl_lock, will reacquire if has to wait */
1503 static void start_unregistering(struct ctl_table_header *p)
1504 {
1505 /*
1506 * if p->used is 0, nobody will ever touch that entry again;
1507 * we'll eliminate all paths to it before dropping sysctl_lock
1508 */
1509 if (unlikely(p->used)) {
1510 struct completion wait;
1511 init_completion(&wait);
1512 p->unregistering = &wait;
1513 spin_unlock(&sysctl_lock);
1514 wait_for_completion(&wait);
1515 spin_lock(&sysctl_lock);
1516 } else {
1517 /* anything non-NULL; we'll never dereference it */
1518 p->unregistering = ERR_PTR(-EINVAL);
1519 }
1520 /*
1521 * do not remove from the list until nobody holds it; walking the
1522 * list in do_sysctl() relies on that.
1523 */
1524 list_del_init(&p->ctl_entry);
1525 }
1526
1527 void sysctl_head_get(struct ctl_table_header *head)
1528 {
1529 spin_lock(&sysctl_lock);
1530 head->count++;
1531 spin_unlock(&sysctl_lock);
1532 }
1533
1534 void sysctl_head_put(struct ctl_table_header *head)
1535 {
1536 spin_lock(&sysctl_lock);
1537 if (!--head->count)
1538 kfree(head);
1539 spin_unlock(&sysctl_lock);
1540 }
1541
1542 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1543 {
1544 if (!head)
1545 BUG();
1546 spin_lock(&sysctl_lock);
1547 if (!use_table(head))
1548 head = ERR_PTR(-ENOENT);
1549 spin_unlock(&sysctl_lock);
1550 return head;
1551 }
1552
1553 void sysctl_head_finish(struct ctl_table_header *head)
1554 {
1555 if (!head)
1556 return;
1557 spin_lock(&sysctl_lock);
1558 unuse_table(head);
1559 spin_unlock(&sysctl_lock);
1560 }
1561
1562 static struct ctl_table_set *
1563 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1564 {
1565 struct ctl_table_set *set = &root->default_set;
1566 if (root->lookup)
1567 set = root->lookup(root, namespaces);
1568 return set;
1569 }
1570
1571 static struct list_head *
1572 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1573 {
1574 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1575 return &set->list;
1576 }
1577
1578 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1579 struct ctl_table_header *prev)
1580 {
1581 struct ctl_table_root *root;
1582 struct list_head *header_list;
1583 struct ctl_table_header *head;
1584 struct list_head *tmp;
1585
1586 spin_lock(&sysctl_lock);
1587 if (prev) {
1588 head = prev;
1589 tmp = &prev->ctl_entry;
1590 unuse_table(prev);
1591 goto next;
1592 }
1593 tmp = &root_table_header.ctl_entry;
1594 for (;;) {
1595 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1596
1597 if (!use_table(head))
1598 goto next;
1599 spin_unlock(&sysctl_lock);
1600 return head;
1601 next:
1602 root = head->root;
1603 tmp = tmp->next;
1604 header_list = lookup_header_list(root, namespaces);
1605 if (tmp != header_list)
1606 continue;
1607
1608 do {
1609 root = list_entry(root->root_list.next,
1610 struct ctl_table_root, root_list);
1611 if (root == &sysctl_table_root)
1612 goto out;
1613 header_list = lookup_header_list(root, namespaces);
1614 } while (list_empty(header_list));
1615 tmp = header_list->next;
1616 }
1617 out:
1618 spin_unlock(&sysctl_lock);
1619 return NULL;
1620 }
1621
1622 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1623 {
1624 return __sysctl_head_next(current->nsproxy, prev);
1625 }
1626
1627 void register_sysctl_root(struct ctl_table_root *root)
1628 {
1629 spin_lock(&sysctl_lock);
1630 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1631 spin_unlock(&sysctl_lock);
1632 }
1633
1634 /*
1635 * sysctl_perm does NOT grant the superuser all rights automatically, because
1636 * some sysctl variables are readonly even to root.
1637 */
1638
1639 static int test_perm(int mode, int op)
1640 {
1641 if (!current_euid())
1642 mode >>= 6;
1643 else if (in_egroup_p(0))
1644 mode >>= 3;
1645 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1646 return 0;
1647 return -EACCES;
1648 }
1649
1650 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1651 {
1652 int error;
1653 int mode;
1654
1655 error = security_sysctl(table, op & (MAY_READ | MAY_WRITE | MAY_EXEC));
1656 if (error)
1657 return error;
1658
1659 if (root->permissions)
1660 mode = root->permissions(root, current->nsproxy, table);
1661 else
1662 mode = table->mode;
1663
1664 return test_perm(mode, op);
1665 }
1666
1667 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1668 {
1669 for (; table->procname; table++) {
1670 table->parent = parent;
1671 if (table->child)
1672 sysctl_set_parent(table, table->child);
1673 }
1674 }
1675
1676 static __init int sysctl_init(void)
1677 {
1678 sysctl_set_parent(NULL, root_table);
1679 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1680 {
1681 int err;
1682 err = sysctl_check_table(current->nsproxy, root_table);
1683 }
1684 #endif
1685 return 0;
1686 }
1687
1688 core_initcall(sysctl_init);
1689
1690 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1691 struct ctl_table *table)
1692 {
1693 struct ctl_table *p;
1694 const char *s = branch->procname;
1695
1696 /* branch should have named subdirectory as its first element */
1697 if (!s || !branch->child)
1698 return NULL;
1699
1700 /* ... and nothing else */
1701 if (branch[1].procname)
1702 return NULL;
1703
1704 /* table should contain subdirectory with the same name */
1705 for (p = table; p->procname; p++) {
1706 if (!p->child)
1707 continue;
1708 if (p->procname && strcmp(p->procname, s) == 0)
1709 return p;
1710 }
1711 return NULL;
1712 }
1713
1714 /* see if attaching q to p would be an improvement */
1715 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1716 {
1717 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1718 struct ctl_table *next;
1719 int is_better = 0;
1720 int not_in_parent = !p->attached_by;
1721
1722 while ((next = is_branch_in(by, to)) != NULL) {
1723 if (by == q->attached_by)
1724 is_better = 1;
1725 if (to == p->attached_by)
1726 not_in_parent = 1;
1727 by = by->child;
1728 to = next->child;
1729 }
1730
1731 if (is_better && not_in_parent) {
1732 q->attached_by = by;
1733 q->attached_to = to;
1734 q->parent = p;
1735 }
1736 }
1737
1738 /**
1739 * __register_sysctl_paths - register a sysctl hierarchy
1740 * @root: List of sysctl headers to register on
1741 * @namespaces: Data to compute which lists of sysctl entries are visible
1742 * @path: The path to the directory the sysctl table is in.
1743 * @table: the top-level table structure
1744 *
1745 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1746 * array. A completely 0 filled entry terminates the table.
1747 *
1748 * The members of the &struct ctl_table structure are used as follows:
1749 *
1750 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1751 * enter a sysctl file
1752 *
1753 * data - a pointer to data for use by proc_handler
1754 *
1755 * maxlen - the maximum size in bytes of the data
1756 *
1757 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1758 *
1759 * child - a pointer to the child sysctl table if this entry is a directory, or
1760 * %NULL.
1761 *
1762 * proc_handler - the text handler routine (described below)
1763 *
1764 * de - for internal use by the sysctl routines
1765 *
1766 * extra1, extra2 - extra pointers usable by the proc handler routines
1767 *
1768 * Leaf nodes in the sysctl tree will be represented by a single file
1769 * under /proc; non-leaf nodes will be represented by directories.
1770 *
1771 * sysctl(2) can automatically manage read and write requests through
1772 * the sysctl table. The data and maxlen fields of the ctl_table
1773 * struct enable minimal validation of the values being written to be
1774 * performed, and the mode field allows minimal authentication.
1775 *
1776 * There must be a proc_handler routine for any terminal nodes
1777 * mirrored under /proc/sys (non-terminals are handled by a built-in
1778 * directory handler). Several default handlers are available to
1779 * cover common cases -
1780 *
1781 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1782 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1783 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1784 *
1785 * It is the handler's job to read the input buffer from user memory
1786 * and process it. The handler should return 0 on success.
1787 *
1788 * This routine returns %NULL on a failure to register, and a pointer
1789 * to the table header on success.
1790 */
1791 struct ctl_table_header *__register_sysctl_paths(
1792 struct ctl_table_root *root,
1793 struct nsproxy *namespaces,
1794 const struct ctl_path *path, struct ctl_table *table)
1795 {
1796 struct ctl_table_header *header;
1797 struct ctl_table *new, **prevp;
1798 unsigned int n, npath;
1799 struct ctl_table_set *set;
1800
1801 /* Count the path components */
1802 for (npath = 0; path[npath].procname; ++npath)
1803 ;
1804
1805 /*
1806 * For each path component, allocate a 2-element ctl_table array.
1807 * The first array element will be filled with the sysctl entry
1808 * for this, the second will be the sentinel (procname == 0).
1809 *
1810 * We allocate everything in one go so that we don't have to
1811 * worry about freeing additional memory in unregister_sysctl_table.
1812 */
1813 header = kzalloc(sizeof(struct ctl_table_header) +
1814 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1815 if (!header)
1816 return NULL;
1817
1818 new = (struct ctl_table *) (header + 1);
1819
1820 /* Now connect the dots */
1821 prevp = &header->ctl_table;
1822 for (n = 0; n < npath; ++n, ++path) {
1823 /* Copy the procname */
1824 new->procname = path->procname;
1825 new->mode = 0555;
1826
1827 *prevp = new;
1828 prevp = &new->child;
1829
1830 new += 2;
1831 }
1832 *prevp = table;
1833 header->ctl_table_arg = table;
1834
1835 INIT_LIST_HEAD(&header->ctl_entry);
1836 header->used = 0;
1837 header->unregistering = NULL;
1838 header->root = root;
1839 sysctl_set_parent(NULL, header->ctl_table);
1840 header->count = 1;
1841 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1842 if (sysctl_check_table(namespaces, header->ctl_table)) {
1843 kfree(header);
1844 return NULL;
1845 }
1846 #endif
1847 spin_lock(&sysctl_lock);
1848 header->set = lookup_header_set(root, namespaces);
1849 header->attached_by = header->ctl_table;
1850 header->attached_to = root_table;
1851 header->parent = &root_table_header;
1852 for (set = header->set; set; set = set->parent) {
1853 struct ctl_table_header *p;
1854 list_for_each_entry(p, &set->list, ctl_entry) {
1855 if (p->unregistering)
1856 continue;
1857 try_attach(p, header);
1858 }
1859 }
1860 header->parent->count++;
1861 list_add_tail(&header->ctl_entry, &header->set->list);
1862 spin_unlock(&sysctl_lock);
1863
1864 return header;
1865 }
1866
1867 /**
1868 * register_sysctl_table_path - register a sysctl table hierarchy
1869 * @path: The path to the directory the sysctl table is in.
1870 * @table: the top-level table structure
1871 *
1872 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1873 * array. A completely 0 filled entry terminates the table.
1874 *
1875 * See __register_sysctl_paths for more details.
1876 */
1877 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1878 struct ctl_table *table)
1879 {
1880 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1881 path, table);
1882 }
1883
1884 /**
1885 * register_sysctl_table - register a sysctl table hierarchy
1886 * @table: the top-level table structure
1887 *
1888 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1889 * array. A completely 0 filled entry terminates the table.
1890 *
1891 * See register_sysctl_paths for more details.
1892 */
1893 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1894 {
1895 static const struct ctl_path null_path[] = { {} };
1896
1897 return register_sysctl_paths(null_path, table);
1898 }
1899
1900 /**
1901 * unregister_sysctl_table - unregister a sysctl table hierarchy
1902 * @header: the header returned from register_sysctl_table
1903 *
1904 * Unregisters the sysctl table and all children. proc entries may not
1905 * actually be removed until they are no longer used by anyone.
1906 */
1907 void unregister_sysctl_table(struct ctl_table_header * header)
1908 {
1909 might_sleep();
1910
1911 if (header == NULL)
1912 return;
1913
1914 spin_lock(&sysctl_lock);
1915 start_unregistering(header);
1916 if (!--header->parent->count) {
1917 WARN_ON(1);
1918 kfree(header->parent);
1919 }
1920 if (!--header->count)
1921 kfree(header);
1922 spin_unlock(&sysctl_lock);
1923 }
1924
1925 int sysctl_is_seen(struct ctl_table_header *p)
1926 {
1927 struct ctl_table_set *set = p->set;
1928 int res;
1929 spin_lock(&sysctl_lock);
1930 if (p->unregistering)
1931 res = 0;
1932 else if (!set->is_seen)
1933 res = 1;
1934 else
1935 res = set->is_seen(set);
1936 spin_unlock(&sysctl_lock);
1937 return res;
1938 }
1939
1940 void setup_sysctl_set(struct ctl_table_set *p,
1941 struct ctl_table_set *parent,
1942 int (*is_seen)(struct ctl_table_set *))
1943 {
1944 INIT_LIST_HEAD(&p->list);
1945 p->parent = parent ? parent : &sysctl_table_root.default_set;
1946 p->is_seen = is_seen;
1947 }
1948
1949 #else /* !CONFIG_SYSCTL */
1950 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1951 {
1952 return NULL;
1953 }
1954
1955 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1956 struct ctl_table *table)
1957 {
1958 return NULL;
1959 }
1960
1961 void unregister_sysctl_table(struct ctl_table_header * table)
1962 {
1963 }
1964
1965 void setup_sysctl_set(struct ctl_table_set *p,
1966 struct ctl_table_set *parent,
1967 int (*is_seen)(struct ctl_table_set *))
1968 {
1969 }
1970
1971 void sysctl_head_put(struct ctl_table_header *head)
1972 {
1973 }
1974
1975 #endif /* CONFIG_SYSCTL */
1976
1977 /*
1978 * /proc/sys support
1979 */
1980
1981 #ifdef CONFIG_PROC_SYSCTL
1982
1983 static int _proc_do_string(void* data, int maxlen, int write,
1984 void __user *buffer,
1985 size_t *lenp, loff_t *ppos)
1986 {
1987 size_t len;
1988 char __user *p;
1989 char c;
1990
1991 if (!data || !maxlen || !*lenp) {
1992 *lenp = 0;
1993 return 0;
1994 }
1995
1996 if (write) {
1997 len = 0;
1998 p = buffer;
1999 while (len < *lenp) {
2000 if (get_user(c, p++))
2001 return -EFAULT;
2002 if (c == 0 || c == '\n')
2003 break;
2004 len++;
2005 }
2006 if (len >= maxlen)
2007 len = maxlen-1;
2008 if(copy_from_user(data, buffer, len))
2009 return -EFAULT;
2010 ((char *) data)[len] = 0;
2011 *ppos += *lenp;
2012 } else {
2013 len = strlen(data);
2014 if (len > maxlen)
2015 len = maxlen;
2016
2017 if (*ppos > len) {
2018 *lenp = 0;
2019 return 0;
2020 }
2021
2022 data += *ppos;
2023 len -= *ppos;
2024
2025 if (len > *lenp)
2026 len = *lenp;
2027 if (len)
2028 if(copy_to_user(buffer, data, len))
2029 return -EFAULT;
2030 if (len < *lenp) {
2031 if(put_user('\n', ((char __user *) buffer) + len))
2032 return -EFAULT;
2033 len++;
2034 }
2035 *lenp = len;
2036 *ppos += len;
2037 }
2038 return 0;
2039 }
2040
2041 /**
2042 * proc_dostring - read a string sysctl
2043 * @table: the sysctl table
2044 * @write: %TRUE if this is a write to the sysctl file
2045 * @buffer: the user buffer
2046 * @lenp: the size of the user buffer
2047 * @ppos: file position
2048 *
2049 * Reads/writes a string from/to the user buffer. If the kernel
2050 * buffer provided is not large enough to hold the string, the
2051 * string is truncated. The copied string is %NULL-terminated.
2052 * If the string is being read by the user process, it is copied
2053 * and a newline '\n' is added. It is truncated if the buffer is
2054 * not large enough.
2055 *
2056 * Returns 0 on success.
2057 */
2058 int proc_dostring(struct ctl_table *table, int write,
2059 void __user *buffer, size_t *lenp, loff_t *ppos)
2060 {
2061 return _proc_do_string(table->data, table->maxlen, write,
2062 buffer, lenp, ppos);
2063 }
2064
2065
2066 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
2067 int *valp,
2068 int write, void *data)
2069 {
2070 if (write) {
2071 *valp = *negp ? -*lvalp : *lvalp;
2072 } else {
2073 int val = *valp;
2074 if (val < 0) {
2075 *negp = -1;
2076 *lvalp = (unsigned long)-val;
2077 } else {
2078 *negp = 0;
2079 *lvalp = (unsigned long)val;
2080 }
2081 }
2082 return 0;
2083 }
2084
2085 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2086 int write, void __user *buffer,
2087 size_t *lenp, loff_t *ppos,
2088 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2089 int write, void *data),
2090 void *data)
2091 {
2092 #define TMPBUFLEN 21
2093 int *i, vleft, first = 1, neg;
2094 unsigned long lval;
2095 size_t left, len;
2096
2097 char buf[TMPBUFLEN], *p;
2098 char __user *s = buffer;
2099
2100 if (!tbl_data || !table->maxlen || !*lenp ||
2101 (*ppos && !write)) {
2102 *lenp = 0;
2103 return 0;
2104 }
2105
2106 i = (int *) tbl_data;
2107 vleft = table->maxlen / sizeof(*i);
2108 left = *lenp;
2109
2110 if (!conv)
2111 conv = do_proc_dointvec_conv;
2112
2113 for (; left && vleft--; i++, first=0) {
2114 if (write) {
2115 while (left) {
2116 char c;
2117 if (get_user(c, s))
2118 return -EFAULT;
2119 if (!isspace(c))
2120 break;
2121 left--;
2122 s++;
2123 }
2124 if (!left)
2125 break;
2126 neg = 0;
2127 len = left;
2128 if (len > sizeof(buf) - 1)
2129 len = sizeof(buf) - 1;
2130 if (copy_from_user(buf, s, len))
2131 return -EFAULT;
2132 buf[len] = 0;
2133 p = buf;
2134 if (*p == '-' && left > 1) {
2135 neg = 1;
2136 p++;
2137 }
2138 if (*p < '0' || *p > '9')
2139 break;
2140
2141 lval = simple_strtoul(p, &p, 0);
2142
2143 len = p-buf;
2144 if ((len < left) && *p && !isspace(*p))
2145 break;
2146 s += len;
2147 left -= len;
2148
2149 if (conv(&neg, &lval, i, 1, data))
2150 break;
2151 } else {
2152 p = buf;
2153 if (!first)
2154 *p++ = '\t';
2155
2156 if (conv(&neg, &lval, i, 0, data))
2157 break;
2158
2159 sprintf(p, "%s%lu", neg ? "-" : "", lval);
2160 len = strlen(buf);
2161 if (len > left)
2162 len = left;
2163 if(copy_to_user(s, buf, len))
2164 return -EFAULT;
2165 left -= len;
2166 s += len;
2167 }
2168 }
2169
2170 if (!write && !first && left) {
2171 if(put_user('\n', s))
2172 return -EFAULT;
2173 left--, s++;
2174 }
2175 if (write) {
2176 while (left) {
2177 char c;
2178 if (get_user(c, s++))
2179 return -EFAULT;
2180 if (!isspace(c))
2181 break;
2182 left--;
2183 }
2184 }
2185 if (write && first)
2186 return -EINVAL;
2187 *lenp -= left;
2188 *ppos += *lenp;
2189 return 0;
2190 #undef TMPBUFLEN
2191 }
2192
2193 static int do_proc_dointvec(struct ctl_table *table, int write,
2194 void __user *buffer, size_t *lenp, loff_t *ppos,
2195 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
2196 int write, void *data),
2197 void *data)
2198 {
2199 return __do_proc_dointvec(table->data, table, write,
2200 buffer, lenp, ppos, conv, data);
2201 }
2202
2203 /**
2204 * proc_dointvec - read a vector of integers
2205 * @table: the sysctl table
2206 * @write: %TRUE if this is a write to the sysctl file
2207 * @buffer: the user buffer
2208 * @lenp: the size of the user buffer
2209 * @ppos: file position
2210 *
2211 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2212 * values from/to the user buffer, treated as an ASCII string.
2213 *
2214 * Returns 0 on success.
2215 */
2216 int proc_dointvec(struct ctl_table *table, int write,
2217 void __user *buffer, size_t *lenp, loff_t *ppos)
2218 {
2219 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2220 NULL,NULL);
2221 }
2222
2223 /*
2224 * Taint values can only be increased
2225 * This means we can safely use a temporary.
2226 */
2227 static int proc_taint(struct ctl_table *table, int write,
2228 void __user *buffer, size_t *lenp, loff_t *ppos)
2229 {
2230 struct ctl_table t;
2231 unsigned long tmptaint = get_taint();
2232 int err;
2233
2234 if (write && !capable(CAP_SYS_ADMIN))
2235 return -EPERM;
2236
2237 t = *table;
2238 t.data = &tmptaint;
2239 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2240 if (err < 0)
2241 return err;
2242
2243 if (write) {
2244 /*
2245 * Poor man's atomic or. Not worth adding a primitive
2246 * to everyone's atomic.h for this
2247 */
2248 int i;
2249 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2250 if ((tmptaint >> i) & 1)
2251 add_taint(i);
2252 }
2253 }
2254
2255 return err;
2256 }
2257
2258 struct do_proc_dointvec_minmax_conv_param {
2259 int *min;
2260 int *max;
2261 };
2262
2263 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
2264 int *valp,
2265 int write, void *data)
2266 {
2267 struct do_proc_dointvec_minmax_conv_param *param = data;
2268 if (write) {
2269 int val = *negp ? -*lvalp : *lvalp;
2270 if ((param->min && *param->min > val) ||
2271 (param->max && *param->max < val))
2272 return -EINVAL;
2273 *valp = val;
2274 } else {
2275 int val = *valp;
2276 if (val < 0) {
2277 *negp = -1;
2278 *lvalp = (unsigned long)-val;
2279 } else {
2280 *negp = 0;
2281 *lvalp = (unsigned long)val;
2282 }
2283 }
2284 return 0;
2285 }
2286
2287 /**
2288 * proc_dointvec_minmax - read a vector of integers with min/max values
2289 * @table: the sysctl table
2290 * @write: %TRUE if this is a write to the sysctl file
2291 * @buffer: the user buffer
2292 * @lenp: the size of the user buffer
2293 * @ppos: file position
2294 *
2295 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2296 * values from/to the user buffer, treated as an ASCII string.
2297 *
2298 * This routine will ensure the values are within the range specified by
2299 * table->extra1 (min) and table->extra2 (max).
2300 *
2301 * Returns 0 on success.
2302 */
2303 int proc_dointvec_minmax(struct ctl_table *table, int write,
2304 void __user *buffer, size_t *lenp, loff_t *ppos)
2305 {
2306 struct do_proc_dointvec_minmax_conv_param param = {
2307 .min = (int *) table->extra1,
2308 .max = (int *) table->extra2,
2309 };
2310 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2311 do_proc_dointvec_minmax_conv, &param);
2312 }
2313
2314 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2315 void __user *buffer,
2316 size_t *lenp, loff_t *ppos,
2317 unsigned long convmul,
2318 unsigned long convdiv)
2319 {
2320 #define TMPBUFLEN 21
2321 unsigned long *i, *min, *max, val;
2322 int vleft, first=1, neg;
2323 size_t len, left;
2324 char buf[TMPBUFLEN], *p;
2325 char __user *s = buffer;
2326
2327 if (!data || !table->maxlen || !*lenp ||
2328 (*ppos && !write)) {
2329 *lenp = 0;
2330 return 0;
2331 }
2332
2333 i = (unsigned long *) data;
2334 min = (unsigned long *) table->extra1;
2335 max = (unsigned long *) table->extra2;
2336 vleft = table->maxlen / sizeof(unsigned long);
2337 left = *lenp;
2338
2339 for (; left && vleft--; i++, min++, max++, first=0) {
2340 if (write) {
2341 while (left) {
2342 char c;
2343 if (get_user(c, s))
2344 return -EFAULT;
2345 if (!isspace(c))
2346 break;
2347 left--;
2348 s++;
2349 }
2350 if (!left)
2351 break;
2352 neg = 0;
2353 len = left;
2354 if (len > TMPBUFLEN-1)
2355 len = TMPBUFLEN-1;
2356 if (copy_from_user(buf, s, len))
2357 return -EFAULT;
2358 buf[len] = 0;
2359 p = buf;
2360 if (*p == '-' && left > 1) {
2361 neg = 1;
2362 p++;
2363 }
2364 if (*p < '0' || *p > '9')
2365 break;
2366 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2367 len = p-buf;
2368 if ((len < left) && *p && !isspace(*p))
2369 break;
2370 if (neg)
2371 val = -val;
2372 s += len;
2373 left -= len;
2374
2375 if(neg)
2376 continue;
2377 if ((min && val < *min) || (max && val > *max))
2378 continue;
2379 *i = val;
2380 } else {
2381 p = buf;
2382 if (!first)
2383 *p++ = '\t';
2384 sprintf(p, "%lu", convdiv * (*i) / convmul);
2385 len = strlen(buf);
2386 if (len > left)
2387 len = left;
2388 if(copy_to_user(s, buf, len))
2389 return -EFAULT;
2390 left -= len;
2391 s += len;
2392 }
2393 }
2394
2395 if (!write && !first && left) {
2396 if(put_user('\n', s))
2397 return -EFAULT;
2398 left--, s++;
2399 }
2400 if (write) {
2401 while (left) {
2402 char c;
2403 if (get_user(c, s++))
2404 return -EFAULT;
2405 if (!isspace(c))
2406 break;
2407 left--;
2408 }
2409 }
2410 if (write && first)
2411 return -EINVAL;
2412 *lenp -= left;
2413 *ppos += *lenp;
2414 return 0;
2415 #undef TMPBUFLEN
2416 }
2417
2418 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2419 void __user *buffer,
2420 size_t *lenp, loff_t *ppos,
2421 unsigned long convmul,
2422 unsigned long convdiv)
2423 {
2424 return __do_proc_doulongvec_minmax(table->data, table, write,
2425 buffer, lenp, ppos, convmul, convdiv);
2426 }
2427
2428 /**
2429 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2430 * @table: the sysctl table
2431 * @write: %TRUE if this is a write to the sysctl file
2432 * @buffer: the user buffer
2433 * @lenp: the size of the user buffer
2434 * @ppos: file position
2435 *
2436 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2437 * values from/to the user buffer, treated as an ASCII string.
2438 *
2439 * This routine will ensure the values are within the range specified by
2440 * table->extra1 (min) and table->extra2 (max).
2441 *
2442 * Returns 0 on success.
2443 */
2444 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2445 void __user *buffer, size_t *lenp, loff_t *ppos)
2446 {
2447 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2448 }
2449
2450 /**
2451 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2452 * @table: the sysctl table
2453 * @write: %TRUE if this is a write to the sysctl file
2454 * @buffer: the user buffer
2455 * @lenp: the size of the user buffer
2456 * @ppos: file position
2457 *
2458 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2459 * values from/to the user buffer, treated as an ASCII string. The values
2460 * are treated as milliseconds, and converted to jiffies when they are stored.
2461 *
2462 * This routine will ensure the values are within the range specified by
2463 * table->extra1 (min) and table->extra2 (max).
2464 *
2465 * Returns 0 on success.
2466 */
2467 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2468 void __user *buffer,
2469 size_t *lenp, loff_t *ppos)
2470 {
2471 return do_proc_doulongvec_minmax(table, write, buffer,
2472 lenp, ppos, HZ, 1000l);
2473 }
2474
2475
2476 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2477 int *valp,
2478 int write, void *data)
2479 {
2480 if (write) {
2481 if (*lvalp > LONG_MAX / HZ)
2482 return 1;
2483 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2484 } else {
2485 int val = *valp;
2486 unsigned long lval;
2487 if (val < 0) {
2488 *negp = -1;
2489 lval = (unsigned long)-val;
2490 } else {
2491 *negp = 0;
2492 lval = (unsigned long)val;
2493 }
2494 *lvalp = lval / HZ;
2495 }
2496 return 0;
2497 }
2498
2499 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2500 int *valp,
2501 int write, void *data)
2502 {
2503 if (write) {
2504 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2505 return 1;
2506 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2507 } else {
2508 int val = *valp;
2509 unsigned long lval;
2510 if (val < 0) {
2511 *negp = -1;
2512 lval = (unsigned long)-val;
2513 } else {
2514 *negp = 0;
2515 lval = (unsigned long)val;
2516 }
2517 *lvalp = jiffies_to_clock_t(lval);
2518 }
2519 return 0;
2520 }
2521
2522 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2523 int *valp,
2524 int write, void *data)
2525 {
2526 if (write) {
2527 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2528 } else {
2529 int val = *valp;
2530 unsigned long lval;
2531 if (val < 0) {
2532 *negp = -1;
2533 lval = (unsigned long)-val;
2534 } else {
2535 *negp = 0;
2536 lval = (unsigned long)val;
2537 }
2538 *lvalp = jiffies_to_msecs(lval);
2539 }
2540 return 0;
2541 }
2542
2543 /**
2544 * proc_dointvec_jiffies - read a vector of integers as seconds
2545 * @table: the sysctl table
2546 * @write: %TRUE if this is a write to the sysctl file
2547 * @buffer: the user buffer
2548 * @lenp: the size of the user buffer
2549 * @ppos: file position
2550 *
2551 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2552 * values from/to the user buffer, treated as an ASCII string.
2553 * The values read are assumed to be in seconds, and are converted into
2554 * jiffies.
2555 *
2556 * Returns 0 on success.
2557 */
2558 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2559 void __user *buffer, size_t *lenp, loff_t *ppos)
2560 {
2561 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2562 do_proc_dointvec_jiffies_conv,NULL);
2563 }
2564
2565 /**
2566 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2567 * @table: the sysctl table
2568 * @write: %TRUE if this is a write to the sysctl file
2569 * @buffer: the user buffer
2570 * @lenp: the size of the user buffer
2571 * @ppos: pointer to the file position
2572 *
2573 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2574 * values from/to the user buffer, treated as an ASCII string.
2575 * The values read are assumed to be in 1/USER_HZ seconds, and
2576 * are converted into jiffies.
2577 *
2578 * Returns 0 on success.
2579 */
2580 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2581 void __user *buffer, size_t *lenp, loff_t *ppos)
2582 {
2583 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2584 do_proc_dointvec_userhz_jiffies_conv,NULL);
2585 }
2586
2587 /**
2588 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2589 * @table: the sysctl table
2590 * @write: %TRUE if this is a write to the sysctl file
2591 * @buffer: the user buffer
2592 * @lenp: the size of the user buffer
2593 * @ppos: file position
2594 * @ppos: the current position in the file
2595 *
2596 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2597 * values from/to the user buffer, treated as an ASCII string.
2598 * The values read are assumed to be in 1/1000 seconds, and
2599 * are converted into jiffies.
2600 *
2601 * Returns 0 on success.
2602 */
2603 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2604 void __user *buffer, size_t *lenp, loff_t *ppos)
2605 {
2606 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2607 do_proc_dointvec_ms_jiffies_conv, NULL);
2608 }
2609
2610 static int proc_do_cad_pid(struct ctl_table *table, int write,
2611 void __user *buffer, size_t *lenp, loff_t *ppos)
2612 {
2613 struct pid *new_pid;
2614 pid_t tmp;
2615 int r;
2616
2617 tmp = pid_vnr(cad_pid);
2618
2619 r = __do_proc_dointvec(&tmp, table, write, buffer,
2620 lenp, ppos, NULL, NULL);
2621 if (r || !write)
2622 return r;
2623
2624 new_pid = find_get_pid(tmp);
2625 if (!new_pid)
2626 return -ESRCH;
2627
2628 put_pid(xchg(&cad_pid, new_pid));
2629 return 0;
2630 }
2631
2632 #else /* CONFIG_PROC_FS */
2633
2634 int proc_dostring(struct ctl_table *table, int write,
2635 void __user *buffer, size_t *lenp, loff_t *ppos)
2636 {
2637 return -ENOSYS;
2638 }
2639
2640 int proc_dointvec(struct ctl_table *table, int write,
2641 void __user *buffer, size_t *lenp, loff_t *ppos)
2642 {
2643 return -ENOSYS;
2644 }
2645
2646 int proc_dointvec_minmax(struct ctl_table *table, int write,
2647 void __user *buffer, size_t *lenp, loff_t *ppos)
2648 {
2649 return -ENOSYS;
2650 }
2651
2652 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2653 void __user *buffer, size_t *lenp, loff_t *ppos)
2654 {
2655 return -ENOSYS;
2656 }
2657
2658 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2659 void __user *buffer, size_t *lenp, loff_t *ppos)
2660 {
2661 return -ENOSYS;
2662 }
2663
2664 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2665 void __user *buffer, size_t *lenp, loff_t *ppos)
2666 {
2667 return -ENOSYS;
2668 }
2669
2670 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2671 void __user *buffer, size_t *lenp, loff_t *ppos)
2672 {
2673 return -ENOSYS;
2674 }
2675
2676 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2677 void __user *buffer,
2678 size_t *lenp, loff_t *ppos)
2679 {
2680 return -ENOSYS;
2681 }
2682
2683
2684 #endif /* CONFIG_PROC_FS */
2685
2686 /*
2687 * No sense putting this after each symbol definition, twice,
2688 * exception granted :-)
2689 */
2690 EXPORT_SYMBOL(proc_dointvec);
2691 EXPORT_SYMBOL(proc_dointvec_jiffies);
2692 EXPORT_SYMBOL(proc_dointvec_minmax);
2693 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2694 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2695 EXPORT_SYMBOL(proc_dostring);
2696 EXPORT_SYMBOL(proc_doulongvec_minmax);
2697 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2698 EXPORT_SYMBOL(register_sysctl_table);
2699 EXPORT_SYMBOL(register_sysctl_paths);
2700 EXPORT_SYMBOL(unregister_sysctl_table);
ubuntu-bionic-kernel mirror