Loading...
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/bitmap.h>
27#include <linux/signal.h>
28#include <linux/printk.h>
29#include <linux/proc_fs.h>
30#include <linux/security.h>
31#include <linux/ctype.h>
32#include <linux/kmemcheck.h>
33#include <linux/fs.h>
34#include <linux/init.h>
35#include <linux/kernel.h>
36#include <linux/kobject.h>
37#include <linux/net.h>
38#include <linux/sysrq.h>
39#include <linux/highuid.h>
40#include <linux/writeback.h>
41#include <linux/ratelimit.h>
42#include <linux/compaction.h>
43#include <linux/hugetlb.h>
44#include <linux/initrd.h>
45#include <linux/key.h>
46#include <linux/times.h>
47#include <linux/limits.h>
48#include <linux/dcache.h>
49#include <linux/dnotify.h>
50#include <linux/syscalls.h>
51#include <linux/vmstat.h>
52#include <linux/nfs_fs.h>
53#include <linux/acpi.h>
54#include <linux/reboot.h>
55#include <linux/ftrace.h>
56#include <linux/perf_event.h>
57#include <linux/kprobes.h>
58#include <linux/pipe_fs_i.h>
59#include <linux/oom.h>
60#include <linux/kmod.h>
61#include <linux/capability.h>
62#include <linux/binfmts.h>
63
64#include <asm/uaccess.h>
65#include <asm/processor.h>
66
67#ifdef CONFIG_X86
68#include <asm/nmi.h>
69#include <asm/stacktrace.h>
70#include <asm/io.h>
71#endif
72#ifdef CONFIG_SPARC
73#include <asm/setup.h>
74#endif
75#ifdef CONFIG_BSD_PROCESS_ACCT
76#include <linux/acct.h>
77#endif
78#ifdef CONFIG_RT_MUTEXES
79#include <linux/rtmutex.h>
80#endif
81#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
82#include <linux/lockdep.h>
83#endif
84#ifdef CONFIG_CHR_DEV_SG
85#include <scsi/sg.h>
86#endif
87
88#ifdef CONFIG_LOCKUP_DETECTOR
89#include <linux/nmi.h>
90#endif
91
92
93#if defined(CONFIG_SYSCTL)
94
95/* External variables not in a header file. */
96extern int sysctl_overcommit_memory;
97extern int sysctl_overcommit_ratio;
98extern int max_threads;
99extern int core_uses_pid;
100extern int suid_dumpable;
101extern char core_pattern[];
102extern unsigned int core_pipe_limit;
103extern int pid_max;
104extern int min_free_kbytes;
105extern int pid_max_min, pid_max_max;
106extern int sysctl_drop_caches;
107extern int percpu_pagelist_fraction;
108extern int compat_log;
109extern int latencytop_enabled;
110extern int sysctl_nr_open_min, sysctl_nr_open_max;
111#ifndef CONFIG_MMU
112extern int sysctl_nr_trim_pages;
113#endif
114#ifdef CONFIG_BLOCK
115extern int blk_iopoll_enabled;
116#endif
117
118/* Constants used for minimum and maximum */
119#ifdef CONFIG_LOCKUP_DETECTOR
120static int sixty = 60;
121static int neg_one = -1;
122#endif
123
124static int zero;
125static int __maybe_unused one = 1;
126static int __maybe_unused two = 2;
127static int __maybe_unused three = 3;
128static unsigned long one_ul = 1;
129static int one_hundred = 100;
130#ifdef CONFIG_PRINTK
131static int ten_thousand = 10000;
132#endif
133
134/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
135static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
136
137/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
138static int maxolduid = 65535;
139static int minolduid;
140static int min_percpu_pagelist_fract = 8;
141
142static int ngroups_max = NGROUPS_MAX;
143static const int cap_last_cap = CAP_LAST_CAP;
144
145#ifdef CONFIG_INOTIFY_USER
146#include <linux/inotify.h>
147#endif
148#ifdef CONFIG_SPARC
149#endif
150
151#ifdef CONFIG_SPARC64
152extern int sysctl_tsb_ratio;
153#endif
154
155#ifdef __hppa__
156extern int pwrsw_enabled;
157extern int unaligned_enabled;
158#endif
159
160#ifdef CONFIG_IA64
161extern int no_unaligned_warning;
162extern int unaligned_dump_stack;
163#endif
164
165#ifdef CONFIG_PROC_SYSCTL
166static int proc_do_cad_pid(struct ctl_table *table, int write,
167 void __user *buffer, size_t *lenp, loff_t *ppos);
168static int proc_taint(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170#endif
171
172#ifdef CONFIG_PRINTK
173static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
174 void __user *buffer, size_t *lenp, loff_t *ppos);
175#endif
176
177#ifdef CONFIG_MAGIC_SYSRQ
178/* Note: sysrq code uses it's own private copy */
179static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
180
181static int sysrq_sysctl_handler(ctl_table *table, int write,
182 void __user *buffer, size_t *lenp,
183 loff_t *ppos)
184{
185 int error;
186
187 error = proc_dointvec(table, write, buffer, lenp, ppos);
188 if (error)
189 return error;
190
191 if (write)
192 sysrq_toggle_support(__sysrq_enabled);
193
194 return 0;
195}
196
197#endif
198
199static struct ctl_table kern_table[];
200static struct ctl_table vm_table[];
201static struct ctl_table fs_table[];
202static struct ctl_table debug_table[];
203static struct ctl_table dev_table[];
204extern struct ctl_table random_table[];
205#ifdef CONFIG_EPOLL
206extern struct ctl_table epoll_table[];
207#endif
208
209#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
210int sysctl_legacy_va_layout;
211#endif
212
213/* The default sysctl tables: */
214
215static struct ctl_table sysctl_base_table[] = {
216 {
217 .procname = "kernel",
218 .mode = 0555,
219 .child = kern_table,
220 },
221 {
222 .procname = "vm",
223 .mode = 0555,
224 .child = vm_table,
225 },
226 {
227 .procname = "fs",
228 .mode = 0555,
229 .child = fs_table,
230 },
231 {
232 .procname = "debug",
233 .mode = 0555,
234 .child = debug_table,
235 },
236 {
237 .procname = "dev",
238 .mode = 0555,
239 .child = dev_table,
240 },
241 { }
242};
243
244#ifdef CONFIG_SCHED_DEBUG
245static int min_sched_granularity_ns = 100000; /* 100 usecs */
246static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
247static int min_wakeup_granularity_ns; /* 0 usecs */
248static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
249static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
250static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
251#endif
252
253#ifdef CONFIG_COMPACTION
254static int min_extfrag_threshold;
255static int max_extfrag_threshold = 1000;
256#endif
257
258static struct ctl_table kern_table[] = {
259 {
260 .procname = "sched_child_runs_first",
261 .data = &sysctl_sched_child_runs_first,
262 .maxlen = sizeof(unsigned int),
263 .mode = 0644,
264 .proc_handler = proc_dointvec,
265 },
266#ifdef CONFIG_SCHED_DEBUG
267 {
268 .procname = "sched_min_granularity_ns",
269 .data = &sysctl_sched_min_granularity,
270 .maxlen = sizeof(unsigned int),
271 .mode = 0644,
272 .proc_handler = sched_proc_update_handler,
273 .extra1 = &min_sched_granularity_ns,
274 .extra2 = &max_sched_granularity_ns,
275 },
276 {
277 .procname = "sched_latency_ns",
278 .data = &sysctl_sched_latency,
279 .maxlen = sizeof(unsigned int),
280 .mode = 0644,
281 .proc_handler = sched_proc_update_handler,
282 .extra1 = &min_sched_granularity_ns,
283 .extra2 = &max_sched_granularity_ns,
284 },
285 {
286 .procname = "sched_wakeup_granularity_ns",
287 .data = &sysctl_sched_wakeup_granularity,
288 .maxlen = sizeof(unsigned int),
289 .mode = 0644,
290 .proc_handler = sched_proc_update_handler,
291 .extra1 = &min_wakeup_granularity_ns,
292 .extra2 = &max_wakeup_granularity_ns,
293 },
294 {
295 .procname = "sched_tunable_scaling",
296 .data = &sysctl_sched_tunable_scaling,
297 .maxlen = sizeof(enum sched_tunable_scaling),
298 .mode = 0644,
299 .proc_handler = sched_proc_update_handler,
300 .extra1 = &min_sched_tunable_scaling,
301 .extra2 = &max_sched_tunable_scaling,
302 },
303 {
304 .procname = "sched_migration_cost",
305 .data = &sysctl_sched_migration_cost,
306 .maxlen = sizeof(unsigned int),
307 .mode = 0644,
308 .proc_handler = proc_dointvec,
309 },
310 {
311 .procname = "sched_nr_migrate",
312 .data = &sysctl_sched_nr_migrate,
313 .maxlen = sizeof(unsigned int),
314 .mode = 0644,
315 .proc_handler = proc_dointvec,
316 },
317 {
318 .procname = "sched_time_avg",
319 .data = &sysctl_sched_time_avg,
320 .maxlen = sizeof(unsigned int),
321 .mode = 0644,
322 .proc_handler = proc_dointvec,
323 },
324 {
325 .procname = "sched_shares_window",
326 .data = &sysctl_sched_shares_window,
327 .maxlen = sizeof(unsigned int),
328 .mode = 0644,
329 .proc_handler = proc_dointvec,
330 },
331 {
332 .procname = "timer_migration",
333 .data = &sysctl_timer_migration,
334 .maxlen = sizeof(unsigned int),
335 .mode = 0644,
336 .proc_handler = proc_dointvec_minmax,
337 .extra1 = &zero,
338 .extra2 = &one,
339 },
340#endif
341 {
342 .procname = "sched_rt_period_us",
343 .data = &sysctl_sched_rt_period,
344 .maxlen = sizeof(unsigned int),
345 .mode = 0644,
346 .proc_handler = sched_rt_handler,
347 },
348 {
349 .procname = "sched_rt_runtime_us",
350 .data = &sysctl_sched_rt_runtime,
351 .maxlen = sizeof(int),
352 .mode = 0644,
353 .proc_handler = sched_rt_handler,
354 },
355#ifdef CONFIG_SCHED_AUTOGROUP
356 {
357 .procname = "sched_autogroup_enabled",
358 .data = &sysctl_sched_autogroup_enabled,
359 .maxlen = sizeof(unsigned int),
360 .mode = 0644,
361 .proc_handler = proc_dointvec_minmax,
362 .extra1 = &zero,
363 .extra2 = &one,
364 },
365#endif
366#ifdef CONFIG_CFS_BANDWIDTH
367 {
368 .procname = "sched_cfs_bandwidth_slice_us",
369 .data = &sysctl_sched_cfs_bandwidth_slice,
370 .maxlen = sizeof(unsigned int),
371 .mode = 0644,
372 .proc_handler = proc_dointvec_minmax,
373 .extra1 = &one,
374 },
375#endif
376#ifdef CONFIG_PROVE_LOCKING
377 {
378 .procname = "prove_locking",
379 .data = &prove_locking,
380 .maxlen = sizeof(int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec,
383 },
384#endif
385#ifdef CONFIG_LOCK_STAT
386 {
387 .procname = "lock_stat",
388 .data = &lock_stat,
389 .maxlen = sizeof(int),
390 .mode = 0644,
391 .proc_handler = proc_dointvec,
392 },
393#endif
394 {
395 .procname = "panic",
396 .data = &panic_timeout,
397 .maxlen = sizeof(int),
398 .mode = 0644,
399 .proc_handler = proc_dointvec,
400 },
401 {
402 .procname = "core_uses_pid",
403 .data = &core_uses_pid,
404 .maxlen = sizeof(int),
405 .mode = 0644,
406 .proc_handler = proc_dointvec,
407 },
408 {
409 .procname = "core_pattern",
410 .data = core_pattern,
411 .maxlen = CORENAME_MAX_SIZE,
412 .mode = 0644,
413 .proc_handler = proc_dostring,
414 },
415 {
416 .procname = "core_pipe_limit",
417 .data = &core_pipe_limit,
418 .maxlen = sizeof(unsigned int),
419 .mode = 0644,
420 .proc_handler = proc_dointvec,
421 },
422#ifdef CONFIG_PROC_SYSCTL
423 {
424 .procname = "tainted",
425 .maxlen = sizeof(long),
426 .mode = 0644,
427 .proc_handler = proc_taint,
428 },
429#endif
430#ifdef CONFIG_LATENCYTOP
431 {
432 .procname = "latencytop",
433 .data = &latencytop_enabled,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = proc_dointvec,
437 },
438#endif
439#ifdef CONFIG_BLK_DEV_INITRD
440 {
441 .procname = "real-root-dev",
442 .data = &real_root_dev,
443 .maxlen = sizeof(int),
444 .mode = 0644,
445 .proc_handler = proc_dointvec,
446 },
447#endif
448 {
449 .procname = "print-fatal-signals",
450 .data = &print_fatal_signals,
451 .maxlen = sizeof(int),
452 .mode = 0644,
453 .proc_handler = proc_dointvec,
454 },
455#ifdef CONFIG_SPARC
456 {
457 .procname = "reboot-cmd",
458 .data = reboot_command,
459 .maxlen = 256,
460 .mode = 0644,
461 .proc_handler = proc_dostring,
462 },
463 {
464 .procname = "stop-a",
465 .data = &stop_a_enabled,
466 .maxlen = sizeof (int),
467 .mode = 0644,
468 .proc_handler = proc_dointvec,
469 },
470 {
471 .procname = "scons-poweroff",
472 .data = &scons_pwroff,
473 .maxlen = sizeof (int),
474 .mode = 0644,
475 .proc_handler = proc_dointvec,
476 },
477#endif
478#ifdef CONFIG_SPARC64
479 {
480 .procname = "tsb-ratio",
481 .data = &sysctl_tsb_ratio,
482 .maxlen = sizeof (int),
483 .mode = 0644,
484 .proc_handler = proc_dointvec,
485 },
486#endif
487#ifdef __hppa__
488 {
489 .procname = "soft-power",
490 .data = &pwrsw_enabled,
491 .maxlen = sizeof (int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495 {
496 .procname = "unaligned-trap",
497 .data = &unaligned_enabled,
498 .maxlen = sizeof (int),
499 .mode = 0644,
500 .proc_handler = proc_dointvec,
501 },
502#endif
503 {
504 .procname = "ctrl-alt-del",
505 .data = &C_A_D,
506 .maxlen = sizeof(int),
507 .mode = 0644,
508 .proc_handler = proc_dointvec,
509 },
510#ifdef CONFIG_FUNCTION_TRACER
511 {
512 .procname = "ftrace_enabled",
513 .data = &ftrace_enabled,
514 .maxlen = sizeof(int),
515 .mode = 0644,
516 .proc_handler = ftrace_enable_sysctl,
517 },
518#endif
519#ifdef CONFIG_STACK_TRACER
520 {
521 .procname = "stack_tracer_enabled",
522 .data = &stack_tracer_enabled,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = stack_trace_sysctl,
526 },
527#endif
528#ifdef CONFIG_TRACING
529 {
530 .procname = "ftrace_dump_on_oops",
531 .data = &ftrace_dump_on_oops,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = proc_dointvec,
535 },
536#endif
537#ifdef CONFIG_MODULES
538 {
539 .procname = "modprobe",
540 .data = &modprobe_path,
541 .maxlen = KMOD_PATH_LEN,
542 .mode = 0644,
543 .proc_handler = proc_dostring,
544 },
545 {
546 .procname = "modules_disabled",
547 .data = &modules_disabled,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 /* only handle a transition from default "0" to "1" */
551 .proc_handler = proc_dointvec_minmax,
552 .extra1 = &one,
553 .extra2 = &one,
554 },
555#endif
556#ifdef CONFIG_HOTPLUG
557 {
558 .procname = "hotplug",
559 .data = &uevent_helper,
560 .maxlen = UEVENT_HELPER_PATH_LEN,
561 .mode = 0644,
562 .proc_handler = proc_dostring,
563 },
564#endif
565#ifdef CONFIG_CHR_DEV_SG
566 {
567 .procname = "sg-big-buff",
568 .data = &sg_big_buff,
569 .maxlen = sizeof (int),
570 .mode = 0444,
571 .proc_handler = proc_dointvec,
572 },
573#endif
574#ifdef CONFIG_BSD_PROCESS_ACCT
575 {
576 .procname = "acct",
577 .data = &acct_parm,
578 .maxlen = 3*sizeof(int),
579 .mode = 0644,
580 .proc_handler = proc_dointvec,
581 },
582#endif
583#ifdef CONFIG_MAGIC_SYSRQ
584 {
585 .procname = "sysrq",
586 .data = &__sysrq_enabled,
587 .maxlen = sizeof (int),
588 .mode = 0644,
589 .proc_handler = sysrq_sysctl_handler,
590 },
591#endif
592#ifdef CONFIG_PROC_SYSCTL
593 {
594 .procname = "cad_pid",
595 .data = NULL,
596 .maxlen = sizeof (int),
597 .mode = 0600,
598 .proc_handler = proc_do_cad_pid,
599 },
600#endif
601 {
602 .procname = "threads-max",
603 .data = &max_threads,
604 .maxlen = sizeof(int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec,
607 },
608 {
609 .procname = "random",
610 .mode = 0555,
611 .child = random_table,
612 },
613 {
614 .procname = "usermodehelper",
615 .mode = 0555,
616 .child = usermodehelper_table,
617 },
618 {
619 .procname = "overflowuid",
620 .data = &overflowuid,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = proc_dointvec_minmax,
624 .extra1 = &minolduid,
625 .extra2 = &maxolduid,
626 },
627 {
628 .procname = "overflowgid",
629 .data = &overflowgid,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = proc_dointvec_minmax,
633 .extra1 = &minolduid,
634 .extra2 = &maxolduid,
635 },
636#ifdef CONFIG_S390
637#ifdef CONFIG_MATHEMU
638 {
639 .procname = "ieee_emulation_warnings",
640 .data = &sysctl_ieee_emulation_warnings,
641 .maxlen = sizeof(int),
642 .mode = 0644,
643 .proc_handler = proc_dointvec,
644 },
645#endif
646 {
647 .procname = "userprocess_debug",
648 .data = &show_unhandled_signals,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 .proc_handler = proc_dointvec,
652 },
653#endif
654 {
655 .procname = "pid_max",
656 .data = &pid_max,
657 .maxlen = sizeof (int),
658 .mode = 0644,
659 .proc_handler = proc_dointvec_minmax,
660 .extra1 = &pid_max_min,
661 .extra2 = &pid_max_max,
662 },
663 {
664 .procname = "panic_on_oops",
665 .data = &panic_on_oops,
666 .maxlen = sizeof(int),
667 .mode = 0644,
668 .proc_handler = proc_dointvec,
669 },
670#if defined CONFIG_PRINTK
671 {
672 .procname = "printk",
673 .data = &console_loglevel,
674 .maxlen = 4*sizeof(int),
675 .mode = 0644,
676 .proc_handler = proc_dointvec,
677 },
678 {
679 .procname = "printk_ratelimit",
680 .data = &printk_ratelimit_state.interval,
681 .maxlen = sizeof(int),
682 .mode = 0644,
683 .proc_handler = proc_dointvec_jiffies,
684 },
685 {
686 .procname = "printk_ratelimit_burst",
687 .data = &printk_ratelimit_state.burst,
688 .maxlen = sizeof(int),
689 .mode = 0644,
690 .proc_handler = proc_dointvec,
691 },
692 {
693 .procname = "printk_delay",
694 .data = &printk_delay_msec,
695 .maxlen = sizeof(int),
696 .mode = 0644,
697 .proc_handler = proc_dointvec_minmax,
698 .extra1 = &zero,
699 .extra2 = &ten_thousand,
700 },
701 {
702 .procname = "dmesg_restrict",
703 .data = &dmesg_restrict,
704 .maxlen = sizeof(int),
705 .mode = 0644,
706 .proc_handler = proc_dointvec_minmax_sysadmin,
707 .extra1 = &zero,
708 .extra2 = &one,
709 },
710 {
711 .procname = "kptr_restrict",
712 .data = &kptr_restrict,
713 .maxlen = sizeof(int),
714 .mode = 0644,
715 .proc_handler = proc_dointvec_minmax_sysadmin,
716 .extra1 = &zero,
717 .extra2 = &two,
718 },
719#endif
720 {
721 .procname = "ngroups_max",
722 .data = &ngroups_max,
723 .maxlen = sizeof (int),
724 .mode = 0444,
725 .proc_handler = proc_dointvec,
726 },
727 {
728 .procname = "cap_last_cap",
729 .data = (void *)&cap_last_cap,
730 .maxlen = sizeof(int),
731 .mode = 0444,
732 .proc_handler = proc_dointvec,
733 },
734#if defined(CONFIG_LOCKUP_DETECTOR)
735 {
736 .procname = "watchdog",
737 .data = &watchdog_enabled,
738 .maxlen = sizeof (int),
739 .mode = 0644,
740 .proc_handler = proc_dowatchdog,
741 .extra1 = &zero,
742 .extra2 = &one,
743 },
744 {
745 .procname = "watchdog_thresh",
746 .data = &watchdog_thresh,
747 .maxlen = sizeof(int),
748 .mode = 0644,
749 .proc_handler = proc_dowatchdog,
750 .extra1 = &neg_one,
751 .extra2 = &sixty,
752 },
753 {
754 .procname = "softlockup_panic",
755 .data = &softlockup_panic,
756 .maxlen = sizeof(int),
757 .mode = 0644,
758 .proc_handler = proc_dointvec_minmax,
759 .extra1 = &zero,
760 .extra2 = &one,
761 },
762 {
763 .procname = "nmi_watchdog",
764 .data = &watchdog_enabled,
765 .maxlen = sizeof (int),
766 .mode = 0644,
767 .proc_handler = proc_dowatchdog,
768 .extra1 = &zero,
769 .extra2 = &one,
770 },
771#endif
772#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
773 {
774 .procname = "unknown_nmi_panic",
775 .data = &unknown_nmi_panic,
776 .maxlen = sizeof (int),
777 .mode = 0644,
778 .proc_handler = proc_dointvec,
779 },
780#endif
781#if defined(CONFIG_X86)
782 {
783 .procname = "panic_on_unrecovered_nmi",
784 .data = &panic_on_unrecovered_nmi,
785 .maxlen = sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789 {
790 .procname = "panic_on_io_nmi",
791 .data = &panic_on_io_nmi,
792 .maxlen = sizeof(int),
793 .mode = 0644,
794 .proc_handler = proc_dointvec,
795 },
796#ifdef CONFIG_DEBUG_STACKOVERFLOW
797 {
798 .procname = "panic_on_stackoverflow",
799 .data = &sysctl_panic_on_stackoverflow,
800 .maxlen = sizeof(int),
801 .mode = 0644,
802 .proc_handler = proc_dointvec,
803 },
804#endif
805 {
806 .procname = "bootloader_type",
807 .data = &bootloader_type,
808 .maxlen = sizeof (int),
809 .mode = 0444,
810 .proc_handler = proc_dointvec,
811 },
812 {
813 .procname = "bootloader_version",
814 .data = &bootloader_version,
815 .maxlen = sizeof (int),
816 .mode = 0444,
817 .proc_handler = proc_dointvec,
818 },
819 {
820 .procname = "kstack_depth_to_print",
821 .data = &kstack_depth_to_print,
822 .maxlen = sizeof(int),
823 .mode = 0644,
824 .proc_handler = proc_dointvec,
825 },
826 {
827 .procname = "io_delay_type",
828 .data = &io_delay_type,
829 .maxlen = sizeof(int),
830 .mode = 0644,
831 .proc_handler = proc_dointvec,
832 },
833#endif
834#if defined(CONFIG_MMU)
835 {
836 .procname = "randomize_va_space",
837 .data = &randomize_va_space,
838 .maxlen = sizeof(int),
839 .mode = 0644,
840 .proc_handler = proc_dointvec,
841 },
842#endif
843#if defined(CONFIG_S390) && defined(CONFIG_SMP)
844 {
845 .procname = "spin_retry",
846 .data = &spin_retry,
847 .maxlen = sizeof (int),
848 .mode = 0644,
849 .proc_handler = proc_dointvec,
850 },
851#endif
852#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
853 {
854 .procname = "acpi_video_flags",
855 .data = &acpi_realmode_flags,
856 .maxlen = sizeof (unsigned long),
857 .mode = 0644,
858 .proc_handler = proc_doulongvec_minmax,
859 },
860#endif
861#ifdef CONFIG_IA64
862 {
863 .procname = "ignore-unaligned-usertrap",
864 .data = &no_unaligned_warning,
865 .maxlen = sizeof (int),
866 .mode = 0644,
867 .proc_handler = proc_dointvec,
868 },
869 {
870 .procname = "unaligned-dump-stack",
871 .data = &unaligned_dump_stack,
872 .maxlen = sizeof (int),
873 .mode = 0644,
874 .proc_handler = proc_dointvec,
875 },
876#endif
877#ifdef CONFIG_DETECT_HUNG_TASK
878 {
879 .procname = "hung_task_panic",
880 .data = &sysctl_hung_task_panic,
881 .maxlen = sizeof(int),
882 .mode = 0644,
883 .proc_handler = proc_dointvec_minmax,
884 .extra1 = &zero,
885 .extra2 = &one,
886 },
887 {
888 .procname = "hung_task_check_count",
889 .data = &sysctl_hung_task_check_count,
890 .maxlen = sizeof(unsigned long),
891 .mode = 0644,
892 .proc_handler = proc_doulongvec_minmax,
893 },
894 {
895 .procname = "hung_task_timeout_secs",
896 .data = &sysctl_hung_task_timeout_secs,
897 .maxlen = sizeof(unsigned long),
898 .mode = 0644,
899 .proc_handler = proc_dohung_task_timeout_secs,
900 },
901 {
902 .procname = "hung_task_warnings",
903 .data = &sysctl_hung_task_warnings,
904 .maxlen = sizeof(unsigned long),
905 .mode = 0644,
906 .proc_handler = proc_doulongvec_minmax,
907 },
908#endif
909#ifdef CONFIG_COMPAT
910 {
911 .procname = "compat-log",
912 .data = &compat_log,
913 .maxlen = sizeof (int),
914 .mode = 0644,
915 .proc_handler = proc_dointvec,
916 },
917#endif
918#ifdef CONFIG_RT_MUTEXES
919 {
920 .procname = "max_lock_depth",
921 .data = &max_lock_depth,
922 .maxlen = sizeof(int),
923 .mode = 0644,
924 .proc_handler = proc_dointvec,
925 },
926#endif
927 {
928 .procname = "poweroff_cmd",
929 .data = &poweroff_cmd,
930 .maxlen = POWEROFF_CMD_PATH_LEN,
931 .mode = 0644,
932 .proc_handler = proc_dostring,
933 },
934#ifdef CONFIG_KEYS
935 {
936 .procname = "keys",
937 .mode = 0555,
938 .child = key_sysctls,
939 },
940#endif
941#ifdef CONFIG_RCU_TORTURE_TEST
942 {
943 .procname = "rcutorture_runnable",
944 .data = &rcutorture_runnable,
945 .maxlen = sizeof(int),
946 .mode = 0644,
947 .proc_handler = proc_dointvec,
948 },
949#endif
950#ifdef CONFIG_PERF_EVENTS
951 /*
952 * User-space scripts rely on the existence of this file
953 * as a feature check for perf_events being enabled.
954 *
955 * So it's an ABI, do not remove!
956 */
957 {
958 .procname = "perf_event_paranoid",
959 .data = &sysctl_perf_event_paranoid,
960 .maxlen = sizeof(sysctl_perf_event_paranoid),
961 .mode = 0644,
962 .proc_handler = proc_dointvec,
963 },
964 {
965 .procname = "perf_event_mlock_kb",
966 .data = &sysctl_perf_event_mlock,
967 .maxlen = sizeof(sysctl_perf_event_mlock),
968 .mode = 0644,
969 .proc_handler = proc_dointvec,
970 },
971 {
972 .procname = "perf_event_max_sample_rate",
973 .data = &sysctl_perf_event_sample_rate,
974 .maxlen = sizeof(sysctl_perf_event_sample_rate),
975 .mode = 0644,
976 .proc_handler = perf_proc_update_handler,
977 },
978#endif
979#ifdef CONFIG_KMEMCHECK
980 {
981 .procname = "kmemcheck",
982 .data = &kmemcheck_enabled,
983 .maxlen = sizeof(int),
984 .mode = 0644,
985 .proc_handler = proc_dointvec,
986 },
987#endif
988#ifdef CONFIG_BLOCK
989 {
990 .procname = "blk_iopoll",
991 .data = &blk_iopoll_enabled,
992 .maxlen = sizeof(int),
993 .mode = 0644,
994 .proc_handler = proc_dointvec,
995 },
996#endif
997 { }
998};
999
1000static struct ctl_table vm_table[] = {
1001 {
1002 .procname = "overcommit_memory",
1003 .data = &sysctl_overcommit_memory,
1004 .maxlen = sizeof(sysctl_overcommit_memory),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec_minmax,
1007 .extra1 = &zero,
1008 .extra2 = &two,
1009 },
1010 {
1011 .procname = "panic_on_oom",
1012 .data = &sysctl_panic_on_oom,
1013 .maxlen = sizeof(sysctl_panic_on_oom),
1014 .mode = 0644,
1015 .proc_handler = proc_dointvec_minmax,
1016 .extra1 = &zero,
1017 .extra2 = &two,
1018 },
1019 {
1020 .procname = "oom_kill_allocating_task",
1021 .data = &sysctl_oom_kill_allocating_task,
1022 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1023 .mode = 0644,
1024 .proc_handler = proc_dointvec,
1025 },
1026 {
1027 .procname = "oom_dump_tasks",
1028 .data = &sysctl_oom_dump_tasks,
1029 .maxlen = sizeof(sysctl_oom_dump_tasks),
1030 .mode = 0644,
1031 .proc_handler = proc_dointvec,
1032 },
1033 {
1034 .procname = "overcommit_ratio",
1035 .data = &sysctl_overcommit_ratio,
1036 .maxlen = sizeof(sysctl_overcommit_ratio),
1037 .mode = 0644,
1038 .proc_handler = proc_dointvec,
1039 },
1040 {
1041 .procname = "page-cluster",
1042 .data = &page_cluster,
1043 .maxlen = sizeof(int),
1044 .mode = 0644,
1045 .proc_handler = proc_dointvec_minmax,
1046 .extra1 = &zero,
1047 },
1048 {
1049 .procname = "dirty_background_ratio",
1050 .data = &dirty_background_ratio,
1051 .maxlen = sizeof(dirty_background_ratio),
1052 .mode = 0644,
1053 .proc_handler = dirty_background_ratio_handler,
1054 .extra1 = &zero,
1055 .extra2 = &one_hundred,
1056 },
1057 {
1058 .procname = "dirty_background_bytes",
1059 .data = &dirty_background_bytes,
1060 .maxlen = sizeof(dirty_background_bytes),
1061 .mode = 0644,
1062 .proc_handler = dirty_background_bytes_handler,
1063 .extra1 = &one_ul,
1064 },
1065 {
1066 .procname = "dirty_ratio",
1067 .data = &vm_dirty_ratio,
1068 .maxlen = sizeof(vm_dirty_ratio),
1069 .mode = 0644,
1070 .proc_handler = dirty_ratio_handler,
1071 .extra1 = &zero,
1072 .extra2 = &one_hundred,
1073 },
1074 {
1075 .procname = "dirty_bytes",
1076 .data = &vm_dirty_bytes,
1077 .maxlen = sizeof(vm_dirty_bytes),
1078 .mode = 0644,
1079 .proc_handler = dirty_bytes_handler,
1080 .extra1 = &dirty_bytes_min,
1081 },
1082 {
1083 .procname = "dirty_writeback_centisecs",
1084 .data = &dirty_writeback_interval,
1085 .maxlen = sizeof(dirty_writeback_interval),
1086 .mode = 0644,
1087 .proc_handler = dirty_writeback_centisecs_handler,
1088 },
1089 {
1090 .procname = "dirty_expire_centisecs",
1091 .data = &dirty_expire_interval,
1092 .maxlen = sizeof(dirty_expire_interval),
1093 .mode = 0644,
1094 .proc_handler = proc_dointvec_minmax,
1095 .extra1 = &zero,
1096 },
1097 {
1098 .procname = "nr_pdflush_threads",
1099 .data = &nr_pdflush_threads,
1100 .maxlen = sizeof nr_pdflush_threads,
1101 .mode = 0444 /* read-only*/,
1102 .proc_handler = proc_dointvec,
1103 },
1104 {
1105 .procname = "swappiness",
1106 .data = &vm_swappiness,
1107 .maxlen = sizeof(vm_swappiness),
1108 .mode = 0644,
1109 .proc_handler = proc_dointvec_minmax,
1110 .extra1 = &zero,
1111 .extra2 = &one_hundred,
1112 },
1113#ifdef CONFIG_HUGETLB_PAGE
1114 {
1115 .procname = "nr_hugepages",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = hugetlb_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#ifdef CONFIG_NUMA
1124 {
1125 .procname = "nr_hugepages_mempolicy",
1126 .data = NULL,
1127 .maxlen = sizeof(unsigned long),
1128 .mode = 0644,
1129 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1130 .extra1 = (void *)&hugetlb_zero,
1131 .extra2 = (void *)&hugetlb_infinity,
1132 },
1133#endif
1134 {
1135 .procname = "hugetlb_shm_group",
1136 .data = &sysctl_hugetlb_shm_group,
1137 .maxlen = sizeof(gid_t),
1138 .mode = 0644,
1139 .proc_handler = proc_dointvec,
1140 },
1141 {
1142 .procname = "hugepages_treat_as_movable",
1143 .data = &hugepages_treat_as_movable,
1144 .maxlen = sizeof(int),
1145 .mode = 0644,
1146 .proc_handler = hugetlb_treat_movable_handler,
1147 },
1148 {
1149 .procname = "nr_overcommit_hugepages",
1150 .data = NULL,
1151 .maxlen = sizeof(unsigned long),
1152 .mode = 0644,
1153 .proc_handler = hugetlb_overcommit_handler,
1154 .extra1 = (void *)&hugetlb_zero,
1155 .extra2 = (void *)&hugetlb_infinity,
1156 },
1157#endif
1158 {
1159 .procname = "lowmem_reserve_ratio",
1160 .data = &sysctl_lowmem_reserve_ratio,
1161 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1162 .mode = 0644,
1163 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1164 },
1165 {
1166 .procname = "drop_caches",
1167 .data = &sysctl_drop_caches,
1168 .maxlen = sizeof(int),
1169 .mode = 0644,
1170 .proc_handler = drop_caches_sysctl_handler,
1171 .extra1 = &one,
1172 .extra2 = &three,
1173 },
1174#ifdef CONFIG_COMPACTION
1175 {
1176 .procname = "compact_memory",
1177 .data = &sysctl_compact_memory,
1178 .maxlen = sizeof(int),
1179 .mode = 0200,
1180 .proc_handler = sysctl_compaction_handler,
1181 },
1182 {
1183 .procname = "extfrag_threshold",
1184 .data = &sysctl_extfrag_threshold,
1185 .maxlen = sizeof(int),
1186 .mode = 0644,
1187 .proc_handler = sysctl_extfrag_handler,
1188 .extra1 = &min_extfrag_threshold,
1189 .extra2 = &max_extfrag_threshold,
1190 },
1191
1192#endif /* CONFIG_COMPACTION */
1193 {
1194 .procname = "min_free_kbytes",
1195 .data = &min_free_kbytes,
1196 .maxlen = sizeof(min_free_kbytes),
1197 .mode = 0644,
1198 .proc_handler = min_free_kbytes_sysctl_handler,
1199 .extra1 = &zero,
1200 },
1201 {
1202 .procname = "percpu_pagelist_fraction",
1203 .data = &percpu_pagelist_fraction,
1204 .maxlen = sizeof(percpu_pagelist_fraction),
1205 .mode = 0644,
1206 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1207 .extra1 = &min_percpu_pagelist_fract,
1208 },
1209#ifdef CONFIG_MMU
1210 {
1211 .procname = "max_map_count",
1212 .data = &sysctl_max_map_count,
1213 .maxlen = sizeof(sysctl_max_map_count),
1214 .mode = 0644,
1215 .proc_handler = proc_dointvec_minmax,
1216 .extra1 = &zero,
1217 },
1218#else
1219 {
1220 .procname = "nr_trim_pages",
1221 .data = &sysctl_nr_trim_pages,
1222 .maxlen = sizeof(sysctl_nr_trim_pages),
1223 .mode = 0644,
1224 .proc_handler = proc_dointvec_minmax,
1225 .extra1 = &zero,
1226 },
1227#endif
1228 {
1229 .procname = "laptop_mode",
1230 .data = &laptop_mode,
1231 .maxlen = sizeof(laptop_mode),
1232 .mode = 0644,
1233 .proc_handler = proc_dointvec_jiffies,
1234 },
1235 {
1236 .procname = "block_dump",
1237 .data = &block_dump,
1238 .maxlen = sizeof(block_dump),
1239 .mode = 0644,
1240 .proc_handler = proc_dointvec,
1241 .extra1 = &zero,
1242 },
1243 {
1244 .procname = "vfs_cache_pressure",
1245 .data = &sysctl_vfs_cache_pressure,
1246 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1247 .mode = 0644,
1248 .proc_handler = proc_dointvec,
1249 .extra1 = &zero,
1250 },
1251#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1252 {
1253 .procname = "legacy_va_layout",
1254 .data = &sysctl_legacy_va_layout,
1255 .maxlen = sizeof(sysctl_legacy_va_layout),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260#endif
1261#ifdef CONFIG_NUMA
1262 {
1263 .procname = "zone_reclaim_mode",
1264 .data = &zone_reclaim_mode,
1265 .maxlen = sizeof(zone_reclaim_mode),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 .extra1 = &zero,
1269 },
1270 {
1271 .procname = "min_unmapped_ratio",
1272 .data = &sysctl_min_unmapped_ratio,
1273 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1274 .mode = 0644,
1275 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1276 .extra1 = &zero,
1277 .extra2 = &one_hundred,
1278 },
1279 {
1280 .procname = "min_slab_ratio",
1281 .data = &sysctl_min_slab_ratio,
1282 .maxlen = sizeof(sysctl_min_slab_ratio),
1283 .mode = 0644,
1284 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1285 .extra1 = &zero,
1286 .extra2 = &one_hundred,
1287 },
1288#endif
1289#ifdef CONFIG_SMP
1290 {
1291 .procname = "stat_interval",
1292 .data = &sysctl_stat_interval,
1293 .maxlen = sizeof(sysctl_stat_interval),
1294 .mode = 0644,
1295 .proc_handler = proc_dointvec_jiffies,
1296 },
1297#endif
1298#ifdef CONFIG_MMU
1299 {
1300 .procname = "mmap_min_addr",
1301 .data = &dac_mmap_min_addr,
1302 .maxlen = sizeof(unsigned long),
1303 .mode = 0644,
1304 .proc_handler = mmap_min_addr_handler,
1305 },
1306#endif
1307#ifdef CONFIG_NUMA
1308 {
1309 .procname = "numa_zonelist_order",
1310 .data = &numa_zonelist_order,
1311 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1312 .mode = 0644,
1313 .proc_handler = numa_zonelist_order_handler,
1314 },
1315#endif
1316#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1317 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1318 {
1319 .procname = "vdso_enabled",
1320 .data = &vdso_enabled,
1321 .maxlen = sizeof(vdso_enabled),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec,
1324 .extra1 = &zero,
1325 },
1326#endif
1327#ifdef CONFIG_HIGHMEM
1328 {
1329 .procname = "highmem_is_dirtyable",
1330 .data = &vm_highmem_is_dirtyable,
1331 .maxlen = sizeof(vm_highmem_is_dirtyable),
1332 .mode = 0644,
1333 .proc_handler = proc_dointvec_minmax,
1334 .extra1 = &zero,
1335 .extra2 = &one,
1336 },
1337#endif
1338 {
1339 .procname = "scan_unevictable_pages",
1340 .data = &scan_unevictable_pages,
1341 .maxlen = sizeof(scan_unevictable_pages),
1342 .mode = 0644,
1343 .proc_handler = scan_unevictable_handler,
1344 },
1345#ifdef CONFIG_MEMORY_FAILURE
1346 {
1347 .procname = "memory_failure_early_kill",
1348 .data = &sysctl_memory_failure_early_kill,
1349 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1350 .mode = 0644,
1351 .proc_handler = proc_dointvec_minmax,
1352 .extra1 = &zero,
1353 .extra2 = &one,
1354 },
1355 {
1356 .procname = "memory_failure_recovery",
1357 .data = &sysctl_memory_failure_recovery,
1358 .maxlen = sizeof(sysctl_memory_failure_recovery),
1359 .mode = 0644,
1360 .proc_handler = proc_dointvec_minmax,
1361 .extra1 = &zero,
1362 .extra2 = &one,
1363 },
1364#endif
1365 { }
1366};
1367
1368#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1369static struct ctl_table binfmt_misc_table[] = {
1370 { }
1371};
1372#endif
1373
1374static struct ctl_table fs_table[] = {
1375 {
1376 .procname = "inode-nr",
1377 .data = &inodes_stat,
1378 .maxlen = 2*sizeof(int),
1379 .mode = 0444,
1380 .proc_handler = proc_nr_inodes,
1381 },
1382 {
1383 .procname = "inode-state",
1384 .data = &inodes_stat,
1385 .maxlen = 7*sizeof(int),
1386 .mode = 0444,
1387 .proc_handler = proc_nr_inodes,
1388 },
1389 {
1390 .procname = "file-nr",
1391 .data = &files_stat,
1392 .maxlen = sizeof(files_stat),
1393 .mode = 0444,
1394 .proc_handler = proc_nr_files,
1395 },
1396 {
1397 .procname = "file-max",
1398 .data = &files_stat.max_files,
1399 .maxlen = sizeof(files_stat.max_files),
1400 .mode = 0644,
1401 .proc_handler = proc_doulongvec_minmax,
1402 },
1403 {
1404 .procname = "nr_open",
1405 .data = &sysctl_nr_open,
1406 .maxlen = sizeof(int),
1407 .mode = 0644,
1408 .proc_handler = proc_dointvec_minmax,
1409 .extra1 = &sysctl_nr_open_min,
1410 .extra2 = &sysctl_nr_open_max,
1411 },
1412 {
1413 .procname = "dentry-state",
1414 .data = &dentry_stat,
1415 .maxlen = 6*sizeof(int),
1416 .mode = 0444,
1417 .proc_handler = proc_nr_dentry,
1418 },
1419 {
1420 .procname = "overflowuid",
1421 .data = &fs_overflowuid,
1422 .maxlen = sizeof(int),
1423 .mode = 0644,
1424 .proc_handler = proc_dointvec_minmax,
1425 .extra1 = &minolduid,
1426 .extra2 = &maxolduid,
1427 },
1428 {
1429 .procname = "overflowgid",
1430 .data = &fs_overflowgid,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec_minmax,
1434 .extra1 = &minolduid,
1435 .extra2 = &maxolduid,
1436 },
1437#ifdef CONFIG_FILE_LOCKING
1438 {
1439 .procname = "leases-enable",
1440 .data = &leases_enable,
1441 .maxlen = sizeof(int),
1442 .mode = 0644,
1443 .proc_handler = proc_dointvec,
1444 },
1445#endif
1446#ifdef CONFIG_DNOTIFY
1447 {
1448 .procname = "dir-notify-enable",
1449 .data = &dir_notify_enable,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_MMU
1456#ifdef CONFIG_FILE_LOCKING
1457 {
1458 .procname = "lease-break-time",
1459 .data = &lease_break_time,
1460 .maxlen = sizeof(int),
1461 .mode = 0644,
1462 .proc_handler = proc_dointvec,
1463 },
1464#endif
1465#ifdef CONFIG_AIO
1466 {
1467 .procname = "aio-nr",
1468 .data = &aio_nr,
1469 .maxlen = sizeof(aio_nr),
1470 .mode = 0444,
1471 .proc_handler = proc_doulongvec_minmax,
1472 },
1473 {
1474 .procname = "aio-max-nr",
1475 .data = &aio_max_nr,
1476 .maxlen = sizeof(aio_max_nr),
1477 .mode = 0644,
1478 .proc_handler = proc_doulongvec_minmax,
1479 },
1480#endif /* CONFIG_AIO */
1481#ifdef CONFIG_INOTIFY_USER
1482 {
1483 .procname = "inotify",
1484 .mode = 0555,
1485 .child = inotify_table,
1486 },
1487#endif
1488#ifdef CONFIG_EPOLL
1489 {
1490 .procname = "epoll",
1491 .mode = 0555,
1492 .child = epoll_table,
1493 },
1494#endif
1495#endif
1496 {
1497 .procname = "suid_dumpable",
1498 .data = &suid_dumpable,
1499 .maxlen = sizeof(int),
1500 .mode = 0644,
1501 .proc_handler = proc_dointvec_minmax,
1502 .extra1 = &zero,
1503 .extra2 = &two,
1504 },
1505#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1506 {
1507 .procname = "binfmt_misc",
1508 .mode = 0555,
1509 .child = binfmt_misc_table,
1510 },
1511#endif
1512 {
1513 .procname = "pipe-max-size",
1514 .data = &pipe_max_size,
1515 .maxlen = sizeof(int),
1516 .mode = 0644,
1517 .proc_handler = &pipe_proc_fn,
1518 .extra1 = &pipe_min_size,
1519 },
1520 { }
1521};
1522
1523static struct ctl_table debug_table[] = {
1524#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1525 defined(CONFIG_S390) || defined(CONFIG_TILE)
1526 {
1527 .procname = "exception-trace",
1528 .data = &show_unhandled_signals,
1529 .maxlen = sizeof(int),
1530 .mode = 0644,
1531 .proc_handler = proc_dointvec
1532 },
1533#endif
1534#if defined(CONFIG_OPTPROBES)
1535 {
1536 .procname = "kprobes-optimization",
1537 .data = &sysctl_kprobes_optimization,
1538 .maxlen = sizeof(int),
1539 .mode = 0644,
1540 .proc_handler = proc_kprobes_optimization_handler,
1541 .extra1 = &zero,
1542 .extra2 = &one,
1543 },
1544#endif
1545 { }
1546};
1547
1548static struct ctl_table dev_table[] = {
1549 { }
1550};
1551
1552int __init sysctl_init(void)
1553{
1554 register_sysctl_table(sysctl_base_table);
1555 return 0;
1556}
1557
1558#endif /* CONFIG_SYSCTL */
1559
1560/*
1561 * /proc/sys support
1562 */
1563
1564#ifdef CONFIG_PROC_SYSCTL
1565
1566static int _proc_do_string(void* data, int maxlen, int write,
1567 void __user *buffer,
1568 size_t *lenp, loff_t *ppos)
1569{
1570 size_t len;
1571 char __user *p;
1572 char c;
1573
1574 if (!data || !maxlen || !*lenp) {
1575 *lenp = 0;
1576 return 0;
1577 }
1578
1579 if (write) {
1580 len = 0;
1581 p = buffer;
1582 while (len < *lenp) {
1583 if (get_user(c, p++))
1584 return -EFAULT;
1585 if (c == 0 || c == '\n')
1586 break;
1587 len++;
1588 }
1589 if (len >= maxlen)
1590 len = maxlen-1;
1591 if(copy_from_user(data, buffer, len))
1592 return -EFAULT;
1593 ((char *) data)[len] = 0;
1594 *ppos += *lenp;
1595 } else {
1596 len = strlen(data);
1597 if (len > maxlen)
1598 len = maxlen;
1599
1600 if (*ppos > len) {
1601 *lenp = 0;
1602 return 0;
1603 }
1604
1605 data += *ppos;
1606 len -= *ppos;
1607
1608 if (len > *lenp)
1609 len = *lenp;
1610 if (len)
1611 if(copy_to_user(buffer, data, len))
1612 return -EFAULT;
1613 if (len < *lenp) {
1614 if(put_user('\n', ((char __user *) buffer) + len))
1615 return -EFAULT;
1616 len++;
1617 }
1618 *lenp = len;
1619 *ppos += len;
1620 }
1621 return 0;
1622}
1623
1624/**
1625 * proc_dostring - read a string sysctl
1626 * @table: the sysctl table
1627 * @write: %TRUE if this is a write to the sysctl file
1628 * @buffer: the user buffer
1629 * @lenp: the size of the user buffer
1630 * @ppos: file position
1631 *
1632 * Reads/writes a string from/to the user buffer. If the kernel
1633 * buffer provided is not large enough to hold the string, the
1634 * string is truncated. The copied string is %NULL-terminated.
1635 * If the string is being read by the user process, it is copied
1636 * and a newline '\n' is added. It is truncated if the buffer is
1637 * not large enough.
1638 *
1639 * Returns 0 on success.
1640 */
1641int proc_dostring(struct ctl_table *table, int write,
1642 void __user *buffer, size_t *lenp, loff_t *ppos)
1643{
1644 return _proc_do_string(table->data, table->maxlen, write,
1645 buffer, lenp, ppos);
1646}
1647
1648static size_t proc_skip_spaces(char **buf)
1649{
1650 size_t ret;
1651 char *tmp = skip_spaces(*buf);
1652 ret = tmp - *buf;
1653 *buf = tmp;
1654 return ret;
1655}
1656
1657static void proc_skip_char(char **buf, size_t *size, const char v)
1658{
1659 while (*size) {
1660 if (**buf != v)
1661 break;
1662 (*size)--;
1663 (*buf)++;
1664 }
1665}
1666
1667#define TMPBUFLEN 22
1668/**
1669 * proc_get_long - reads an ASCII formatted integer from a user buffer
1670 *
1671 * @buf: a kernel buffer
1672 * @size: size of the kernel buffer
1673 * @val: this is where the number will be stored
1674 * @neg: set to %TRUE if number is negative
1675 * @perm_tr: a vector which contains the allowed trailers
1676 * @perm_tr_len: size of the perm_tr vector
1677 * @tr: pointer to store the trailer character
1678 *
1679 * In case of success %0 is returned and @buf and @size are updated with
1680 * the amount of bytes read. If @tr is non-NULL and a trailing
1681 * character exists (size is non-zero after returning from this
1682 * function), @tr is updated with the trailing character.
1683 */
1684static int proc_get_long(char **buf, size_t *size,
1685 unsigned long *val, bool *neg,
1686 const char *perm_tr, unsigned perm_tr_len, char *tr)
1687{
1688 int len;
1689 char *p, tmp[TMPBUFLEN];
1690
1691 if (!*size)
1692 return -EINVAL;
1693
1694 len = *size;
1695 if (len > TMPBUFLEN - 1)
1696 len = TMPBUFLEN - 1;
1697
1698 memcpy(tmp, *buf, len);
1699
1700 tmp[len] = 0;
1701 p = tmp;
1702 if (*p == '-' && *size > 1) {
1703 *neg = true;
1704 p++;
1705 } else
1706 *neg = false;
1707 if (!isdigit(*p))
1708 return -EINVAL;
1709
1710 *val = simple_strtoul(p, &p, 0);
1711
1712 len = p - tmp;
1713
1714 /* We don't know if the next char is whitespace thus we may accept
1715 * invalid integers (e.g. 1234...a) or two integers instead of one
1716 * (e.g. 123...1). So lets not allow such large numbers. */
1717 if (len == TMPBUFLEN - 1)
1718 return -EINVAL;
1719
1720 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1721 return -EINVAL;
1722
1723 if (tr && (len < *size))
1724 *tr = *p;
1725
1726 *buf += len;
1727 *size -= len;
1728
1729 return 0;
1730}
1731
1732/**
1733 * proc_put_long - converts an integer to a decimal ASCII formatted string
1734 *
1735 * @buf: the user buffer
1736 * @size: the size of the user buffer
1737 * @val: the integer to be converted
1738 * @neg: sign of the number, %TRUE for negative
1739 *
1740 * In case of success %0 is returned and @buf and @size are updated with
1741 * the amount of bytes written.
1742 */
1743static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1744 bool neg)
1745{
1746 int len;
1747 char tmp[TMPBUFLEN], *p = tmp;
1748
1749 sprintf(p, "%s%lu", neg ? "-" : "", val);
1750 len = strlen(tmp);
1751 if (len > *size)
1752 len = *size;
1753 if (copy_to_user(*buf, tmp, len))
1754 return -EFAULT;
1755 *size -= len;
1756 *buf += len;
1757 return 0;
1758}
1759#undef TMPBUFLEN
1760
1761static int proc_put_char(void __user **buf, size_t *size, char c)
1762{
1763 if (*size) {
1764 char __user **buffer = (char __user **)buf;
1765 if (put_user(c, *buffer))
1766 return -EFAULT;
1767 (*size)--, (*buffer)++;
1768 *buf = *buffer;
1769 }
1770 return 0;
1771}
1772
1773static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1774 int *valp,
1775 int write, void *data)
1776{
1777 if (write) {
1778 *valp = *negp ? -*lvalp : *lvalp;
1779 } else {
1780 int val = *valp;
1781 if (val < 0) {
1782 *negp = true;
1783 *lvalp = (unsigned long)-val;
1784 } else {
1785 *negp = false;
1786 *lvalp = (unsigned long)val;
1787 }
1788 }
1789 return 0;
1790}
1791
1792static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1793
1794static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1795 int write, void __user *buffer,
1796 size_t *lenp, loff_t *ppos,
1797 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1798 int write, void *data),
1799 void *data)
1800{
1801 int *i, vleft, first = 1, err = 0;
1802 unsigned long page = 0;
1803 size_t left;
1804 char *kbuf;
1805
1806 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1807 *lenp = 0;
1808 return 0;
1809 }
1810
1811 i = (int *) tbl_data;
1812 vleft = table->maxlen / sizeof(*i);
1813 left = *lenp;
1814
1815 if (!conv)
1816 conv = do_proc_dointvec_conv;
1817
1818 if (write) {
1819 if (left > PAGE_SIZE - 1)
1820 left = PAGE_SIZE - 1;
1821 page = __get_free_page(GFP_TEMPORARY);
1822 kbuf = (char *) page;
1823 if (!kbuf)
1824 return -ENOMEM;
1825 if (copy_from_user(kbuf, buffer, left)) {
1826 err = -EFAULT;
1827 goto free;
1828 }
1829 kbuf[left] = 0;
1830 }
1831
1832 for (; left && vleft--; i++, first=0) {
1833 unsigned long lval;
1834 bool neg;
1835
1836 if (write) {
1837 left -= proc_skip_spaces(&kbuf);
1838
1839 if (!left)
1840 break;
1841 err = proc_get_long(&kbuf, &left, &lval, &neg,
1842 proc_wspace_sep,
1843 sizeof(proc_wspace_sep), NULL);
1844 if (err)
1845 break;
1846 if (conv(&neg, &lval, i, 1, data)) {
1847 err = -EINVAL;
1848 break;
1849 }
1850 } else {
1851 if (conv(&neg, &lval, i, 0, data)) {
1852 err = -EINVAL;
1853 break;
1854 }
1855 if (!first)
1856 err = proc_put_char(&buffer, &left, '\t');
1857 if (err)
1858 break;
1859 err = proc_put_long(&buffer, &left, lval, neg);
1860 if (err)
1861 break;
1862 }
1863 }
1864
1865 if (!write && !first && left && !err)
1866 err = proc_put_char(&buffer, &left, '\n');
1867 if (write && !err && left)
1868 left -= proc_skip_spaces(&kbuf);
1869free:
1870 if (write) {
1871 free_page(page);
1872 if (first)
1873 return err ? : -EINVAL;
1874 }
1875 *lenp -= left;
1876 *ppos += *lenp;
1877 return err;
1878}
1879
1880static int do_proc_dointvec(struct ctl_table *table, int write,
1881 void __user *buffer, size_t *lenp, loff_t *ppos,
1882 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1883 int write, void *data),
1884 void *data)
1885{
1886 return __do_proc_dointvec(table->data, table, write,
1887 buffer, lenp, ppos, conv, data);
1888}
1889
1890/**
1891 * proc_dointvec - read a vector of integers
1892 * @table: the sysctl table
1893 * @write: %TRUE if this is a write to the sysctl file
1894 * @buffer: the user buffer
1895 * @lenp: the size of the user buffer
1896 * @ppos: file position
1897 *
1898 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1899 * values from/to the user buffer, treated as an ASCII string.
1900 *
1901 * Returns 0 on success.
1902 */
1903int proc_dointvec(struct ctl_table *table, int write,
1904 void __user *buffer, size_t *lenp, loff_t *ppos)
1905{
1906 return do_proc_dointvec(table,write,buffer,lenp,ppos,
1907 NULL,NULL);
1908}
1909
1910/*
1911 * Taint values can only be increased
1912 * This means we can safely use a temporary.
1913 */
1914static int proc_taint(struct ctl_table *table, int write,
1915 void __user *buffer, size_t *lenp, loff_t *ppos)
1916{
1917 struct ctl_table t;
1918 unsigned long tmptaint = get_taint();
1919 int err;
1920
1921 if (write && !capable(CAP_SYS_ADMIN))
1922 return -EPERM;
1923
1924 t = *table;
1925 t.data = &tmptaint;
1926 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1927 if (err < 0)
1928 return err;
1929
1930 if (write) {
1931 /*
1932 * Poor man's atomic or. Not worth adding a primitive
1933 * to everyone's atomic.h for this
1934 */
1935 int i;
1936 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1937 if ((tmptaint >> i) & 1)
1938 add_taint(i);
1939 }
1940 }
1941
1942 return err;
1943}
1944
1945#ifdef CONFIG_PRINTK
1946static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1947 void __user *buffer, size_t *lenp, loff_t *ppos)
1948{
1949 if (write && !capable(CAP_SYS_ADMIN))
1950 return -EPERM;
1951
1952 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1953}
1954#endif
1955
1956struct do_proc_dointvec_minmax_conv_param {
1957 int *min;
1958 int *max;
1959};
1960
1961static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1962 int *valp,
1963 int write, void *data)
1964{
1965 struct do_proc_dointvec_minmax_conv_param *param = data;
1966 if (write) {
1967 int val = *negp ? -*lvalp : *lvalp;
1968 if ((param->min && *param->min > val) ||
1969 (param->max && *param->max < val))
1970 return -EINVAL;
1971 *valp = val;
1972 } else {
1973 int val = *valp;
1974 if (val < 0) {
1975 *negp = true;
1976 *lvalp = (unsigned long)-val;
1977 } else {
1978 *negp = false;
1979 *lvalp = (unsigned long)val;
1980 }
1981 }
1982 return 0;
1983}
1984
1985/**
1986 * proc_dointvec_minmax - read a vector of integers with min/max values
1987 * @table: the sysctl table
1988 * @write: %TRUE if this is a write to the sysctl file
1989 * @buffer: the user buffer
1990 * @lenp: the size of the user buffer
1991 * @ppos: file position
1992 *
1993 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1994 * values from/to the user buffer, treated as an ASCII string.
1995 *
1996 * This routine will ensure the values are within the range specified by
1997 * table->extra1 (min) and table->extra2 (max).
1998 *
1999 * Returns 0 on success.
2000 */
2001int proc_dointvec_minmax(struct ctl_table *table, int write,
2002 void __user *buffer, size_t *lenp, loff_t *ppos)
2003{
2004 struct do_proc_dointvec_minmax_conv_param param = {
2005 .min = (int *) table->extra1,
2006 .max = (int *) table->extra2,
2007 };
2008 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2009 do_proc_dointvec_minmax_conv, ¶m);
2010}
2011
2012static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2013 void __user *buffer,
2014 size_t *lenp, loff_t *ppos,
2015 unsigned long convmul,
2016 unsigned long convdiv)
2017{
2018 unsigned long *i, *min, *max;
2019 int vleft, first = 1, err = 0;
2020 unsigned long page = 0;
2021 size_t left;
2022 char *kbuf;
2023
2024 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2025 *lenp = 0;
2026 return 0;
2027 }
2028
2029 i = (unsigned long *) data;
2030 min = (unsigned long *) table->extra1;
2031 max = (unsigned long *) table->extra2;
2032 vleft = table->maxlen / sizeof(unsigned long);
2033 left = *lenp;
2034
2035 if (write) {
2036 if (left > PAGE_SIZE - 1)
2037 left = PAGE_SIZE - 1;
2038 page = __get_free_page(GFP_TEMPORARY);
2039 kbuf = (char *) page;
2040 if (!kbuf)
2041 return -ENOMEM;
2042 if (copy_from_user(kbuf, buffer, left)) {
2043 err = -EFAULT;
2044 goto free;
2045 }
2046 kbuf[left] = 0;
2047 }
2048
2049 for (; left && vleft--; i++, first = 0) {
2050 unsigned long val;
2051
2052 if (write) {
2053 bool neg;
2054
2055 left -= proc_skip_spaces(&kbuf);
2056
2057 err = proc_get_long(&kbuf, &left, &val, &neg,
2058 proc_wspace_sep,
2059 sizeof(proc_wspace_sep), NULL);
2060 if (err)
2061 break;
2062 if (neg)
2063 continue;
2064 if ((min && val < *min) || (max && val > *max))
2065 continue;
2066 *i = val;
2067 } else {
2068 val = convdiv * (*i) / convmul;
2069 if (!first)
2070 err = proc_put_char(&buffer, &left, '\t');
2071 err = proc_put_long(&buffer, &left, val, false);
2072 if (err)
2073 break;
2074 }
2075 }
2076
2077 if (!write && !first && left && !err)
2078 err = proc_put_char(&buffer, &left, '\n');
2079 if (write && !err)
2080 left -= proc_skip_spaces(&kbuf);
2081free:
2082 if (write) {
2083 free_page(page);
2084 if (first)
2085 return err ? : -EINVAL;
2086 }
2087 *lenp -= left;
2088 *ppos += *lenp;
2089 return err;
2090}
2091
2092static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2093 void __user *buffer,
2094 size_t *lenp, loff_t *ppos,
2095 unsigned long convmul,
2096 unsigned long convdiv)
2097{
2098 return __do_proc_doulongvec_minmax(table->data, table, write,
2099 buffer, lenp, ppos, convmul, convdiv);
2100}
2101
2102/**
2103 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2104 * @table: the sysctl table
2105 * @write: %TRUE if this is a write to the sysctl file
2106 * @buffer: the user buffer
2107 * @lenp: the size of the user buffer
2108 * @ppos: file position
2109 *
2110 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2111 * values from/to the user buffer, treated as an ASCII string.
2112 *
2113 * This routine will ensure the values are within the range specified by
2114 * table->extra1 (min) and table->extra2 (max).
2115 *
2116 * Returns 0 on success.
2117 */
2118int proc_doulongvec_minmax(struct ctl_table *table, int write,
2119 void __user *buffer, size_t *lenp, loff_t *ppos)
2120{
2121 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2122}
2123
2124/**
2125 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2126 * @table: the sysctl table
2127 * @write: %TRUE if this is a write to the sysctl file
2128 * @buffer: the user buffer
2129 * @lenp: the size of the user buffer
2130 * @ppos: file position
2131 *
2132 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2133 * values from/to the user buffer, treated as an ASCII string. The values
2134 * are treated as milliseconds, and converted to jiffies when they are stored.
2135 *
2136 * This routine will ensure the values are within the range specified by
2137 * table->extra1 (min) and table->extra2 (max).
2138 *
2139 * Returns 0 on success.
2140 */
2141int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2142 void __user *buffer,
2143 size_t *lenp, loff_t *ppos)
2144{
2145 return do_proc_doulongvec_minmax(table, write, buffer,
2146 lenp, ppos, HZ, 1000l);
2147}
2148
2149
2150static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2151 int *valp,
2152 int write, void *data)
2153{
2154 if (write) {
2155 if (*lvalp > LONG_MAX / HZ)
2156 return 1;
2157 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2158 } else {
2159 int val = *valp;
2160 unsigned long lval;
2161 if (val < 0) {
2162 *negp = true;
2163 lval = (unsigned long)-val;
2164 } else {
2165 *negp = false;
2166 lval = (unsigned long)val;
2167 }
2168 *lvalp = lval / HZ;
2169 }
2170 return 0;
2171}
2172
2173static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2174 int *valp,
2175 int write, void *data)
2176{
2177 if (write) {
2178 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2179 return 1;
2180 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2181 } else {
2182 int val = *valp;
2183 unsigned long lval;
2184 if (val < 0) {
2185 *negp = true;
2186 lval = (unsigned long)-val;
2187 } else {
2188 *negp = false;
2189 lval = (unsigned long)val;
2190 }
2191 *lvalp = jiffies_to_clock_t(lval);
2192 }
2193 return 0;
2194}
2195
2196static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2197 int *valp,
2198 int write, void *data)
2199{
2200 if (write) {
2201 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2202 } else {
2203 int val = *valp;
2204 unsigned long lval;
2205 if (val < 0) {
2206 *negp = true;
2207 lval = (unsigned long)-val;
2208 } else {
2209 *negp = false;
2210 lval = (unsigned long)val;
2211 }
2212 *lvalp = jiffies_to_msecs(lval);
2213 }
2214 return 0;
2215}
2216
2217/**
2218 * proc_dointvec_jiffies - read a vector of integers as seconds
2219 * @table: the sysctl table
2220 * @write: %TRUE if this is a write to the sysctl file
2221 * @buffer: the user buffer
2222 * @lenp: the size of the user buffer
2223 * @ppos: file position
2224 *
2225 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2226 * values from/to the user buffer, treated as an ASCII string.
2227 * The values read are assumed to be in seconds, and are converted into
2228 * jiffies.
2229 *
2230 * Returns 0 on success.
2231 */
2232int proc_dointvec_jiffies(struct ctl_table *table, int write,
2233 void __user *buffer, size_t *lenp, loff_t *ppos)
2234{
2235 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2236 do_proc_dointvec_jiffies_conv,NULL);
2237}
2238
2239/**
2240 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2241 * @table: the sysctl table
2242 * @write: %TRUE if this is a write to the sysctl file
2243 * @buffer: the user buffer
2244 * @lenp: the size of the user buffer
2245 * @ppos: pointer to the file position
2246 *
2247 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2248 * values from/to the user buffer, treated as an ASCII string.
2249 * The values read are assumed to be in 1/USER_HZ seconds, and
2250 * are converted into jiffies.
2251 *
2252 * Returns 0 on success.
2253 */
2254int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2255 void __user *buffer, size_t *lenp, loff_t *ppos)
2256{
2257 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2258 do_proc_dointvec_userhz_jiffies_conv,NULL);
2259}
2260
2261/**
2262 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2263 * @table: the sysctl table
2264 * @write: %TRUE if this is a write to the sysctl file
2265 * @buffer: the user buffer
2266 * @lenp: the size of the user buffer
2267 * @ppos: file position
2268 * @ppos: the current position in the file
2269 *
2270 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2271 * values from/to the user buffer, treated as an ASCII string.
2272 * The values read are assumed to be in 1/1000 seconds, and
2273 * are converted into jiffies.
2274 *
2275 * Returns 0 on success.
2276 */
2277int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2278 void __user *buffer, size_t *lenp, loff_t *ppos)
2279{
2280 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2281 do_proc_dointvec_ms_jiffies_conv, NULL);
2282}
2283
2284static int proc_do_cad_pid(struct ctl_table *table, int write,
2285 void __user *buffer, size_t *lenp, loff_t *ppos)
2286{
2287 struct pid *new_pid;
2288 pid_t tmp;
2289 int r;
2290
2291 tmp = pid_vnr(cad_pid);
2292
2293 r = __do_proc_dointvec(&tmp, table, write, buffer,
2294 lenp, ppos, NULL, NULL);
2295 if (r || !write)
2296 return r;
2297
2298 new_pid = find_get_pid(tmp);
2299 if (!new_pid)
2300 return -ESRCH;
2301
2302 put_pid(xchg(&cad_pid, new_pid));
2303 return 0;
2304}
2305
2306/**
2307 * proc_do_large_bitmap - read/write from/to a large bitmap
2308 * @table: the sysctl table
2309 * @write: %TRUE if this is a write to the sysctl file
2310 * @buffer: the user buffer
2311 * @lenp: the size of the user buffer
2312 * @ppos: file position
2313 *
2314 * The bitmap is stored at table->data and the bitmap length (in bits)
2315 * in table->maxlen.
2316 *
2317 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2318 * large bitmaps may be represented in a compact manner. Writing into
2319 * the file will clear the bitmap then update it with the given input.
2320 *
2321 * Returns 0 on success.
2322 */
2323int proc_do_large_bitmap(struct ctl_table *table, int write,
2324 void __user *buffer, size_t *lenp, loff_t *ppos)
2325{
2326 int err = 0;
2327 bool first = 1;
2328 size_t left = *lenp;
2329 unsigned long bitmap_len = table->maxlen;
2330 unsigned long *bitmap = (unsigned long *) table->data;
2331 unsigned long *tmp_bitmap = NULL;
2332 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2333
2334 if (!bitmap_len || !left || (*ppos && !write)) {
2335 *lenp = 0;
2336 return 0;
2337 }
2338
2339 if (write) {
2340 unsigned long page = 0;
2341 char *kbuf;
2342
2343 if (left > PAGE_SIZE - 1)
2344 left = PAGE_SIZE - 1;
2345
2346 page = __get_free_page(GFP_TEMPORARY);
2347 kbuf = (char *) page;
2348 if (!kbuf)
2349 return -ENOMEM;
2350 if (copy_from_user(kbuf, buffer, left)) {
2351 free_page(page);
2352 return -EFAULT;
2353 }
2354 kbuf[left] = 0;
2355
2356 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2357 GFP_KERNEL);
2358 if (!tmp_bitmap) {
2359 free_page(page);
2360 return -ENOMEM;
2361 }
2362 proc_skip_char(&kbuf, &left, '\n');
2363 while (!err && left) {
2364 unsigned long val_a, val_b;
2365 bool neg;
2366
2367 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2368 sizeof(tr_a), &c);
2369 if (err)
2370 break;
2371 if (val_a >= bitmap_len || neg) {
2372 err = -EINVAL;
2373 break;
2374 }
2375
2376 val_b = val_a;
2377 if (left) {
2378 kbuf++;
2379 left--;
2380 }
2381
2382 if (c == '-') {
2383 err = proc_get_long(&kbuf, &left, &val_b,
2384 &neg, tr_b, sizeof(tr_b),
2385 &c);
2386 if (err)
2387 break;
2388 if (val_b >= bitmap_len || neg ||
2389 val_a > val_b) {
2390 err = -EINVAL;
2391 break;
2392 }
2393 if (left) {
2394 kbuf++;
2395 left--;
2396 }
2397 }
2398
2399 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2400 first = 0;
2401 proc_skip_char(&kbuf, &left, '\n');
2402 }
2403 free_page(page);
2404 } else {
2405 unsigned long bit_a, bit_b = 0;
2406
2407 while (left) {
2408 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2409 if (bit_a >= bitmap_len)
2410 break;
2411 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2412 bit_a + 1) - 1;
2413
2414 if (!first) {
2415 err = proc_put_char(&buffer, &left, ',');
2416 if (err)
2417 break;
2418 }
2419 err = proc_put_long(&buffer, &left, bit_a, false);
2420 if (err)
2421 break;
2422 if (bit_a != bit_b) {
2423 err = proc_put_char(&buffer, &left, '-');
2424 if (err)
2425 break;
2426 err = proc_put_long(&buffer, &left, bit_b, false);
2427 if (err)
2428 break;
2429 }
2430
2431 first = 0; bit_b++;
2432 }
2433 if (!err)
2434 err = proc_put_char(&buffer, &left, '\n');
2435 }
2436
2437 if (!err) {
2438 if (write) {
2439 if (*ppos)
2440 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2441 else
2442 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2443 }
2444 kfree(tmp_bitmap);
2445 *lenp -= left;
2446 *ppos += *lenp;
2447 return 0;
2448 } else {
2449 kfree(tmp_bitmap);
2450 return err;
2451 }
2452}
2453
2454#else /* CONFIG_PROC_SYSCTL */
2455
2456int proc_dostring(struct ctl_table *table, int write,
2457 void __user *buffer, size_t *lenp, loff_t *ppos)
2458{
2459 return -ENOSYS;
2460}
2461
2462int proc_dointvec(struct ctl_table *table, int write,
2463 void __user *buffer, size_t *lenp, loff_t *ppos)
2464{
2465 return -ENOSYS;
2466}
2467
2468int proc_dointvec_minmax(struct ctl_table *table, int write,
2469 void __user *buffer, size_t *lenp, loff_t *ppos)
2470{
2471 return -ENOSYS;
2472}
2473
2474int proc_dointvec_jiffies(struct ctl_table *table, int write,
2475 void __user *buffer, size_t *lenp, loff_t *ppos)
2476{
2477 return -ENOSYS;
2478}
2479
2480int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2481 void __user *buffer, size_t *lenp, loff_t *ppos)
2482{
2483 return -ENOSYS;
2484}
2485
2486int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2487 void __user *buffer, size_t *lenp, loff_t *ppos)
2488{
2489 return -ENOSYS;
2490}
2491
2492int proc_doulongvec_minmax(struct ctl_table *table, int write,
2493 void __user *buffer, size_t *lenp, loff_t *ppos)
2494{
2495 return -ENOSYS;
2496}
2497
2498int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2499 void __user *buffer,
2500 size_t *lenp, loff_t *ppos)
2501{
2502 return -ENOSYS;
2503}
2504
2505
2506#endif /* CONFIG_PROC_SYSCTL */
2507
2508/*
2509 * No sense putting this after each symbol definition, twice,
2510 * exception granted :-)
2511 */
2512EXPORT_SYMBOL(proc_dointvec);
2513EXPORT_SYMBOL(proc_dointvec_jiffies);
2514EXPORT_SYMBOL(proc_dointvec_minmax);
2515EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2516EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2517EXPORT_SYMBOL(proc_dostring);
2518EXPORT_SYMBOL(proc_doulongvec_minmax);
2519EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
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/printk.h>
28#include <linux/proc_fs.h>
29#include <linux/security.h>
30#include <linux/ctype.h>
31#include <linux/kmemcheck.h>
32#include <linux/fs.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
35#include <linux/kobject.h>
36#include <linux/net.h>
37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/ratelimit.h>
41#include <linux/compaction.h>
42#include <linux/hugetlb.h>
43#include <linux/initrd.h>
44#include <linux/key.h>
45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
48#include <linux/dnotify.h>
49#include <linux/syscalls.h>
50#include <linux/vmstat.h>
51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
53#include <linux/reboot.h>
54#include <linux/ftrace.h>
55#include <linux/perf_event.h>
56#include <linux/kprobes.h>
57#include <linux/pipe_fs_i.h>
58#include <linux/oom.h>
59#include <linux/kmod.h>
60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
64#ifdef CONFIG_X86
65#include <asm/nmi.h>
66#include <asm/stacktrace.h>
67#include <asm/io.h>
68#endif
69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
81
82#ifdef CONFIG_LOCKUP_DETECTOR
83#include <linux/nmi.h>
84#endif
85
86
87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
93extern int core_uses_pid;
94extern int suid_dumpable;
95extern char core_pattern[];
96extern unsigned int core_pipe_limit;
97extern int pid_max;
98extern int min_free_kbytes;
99extern int pid_max_min, pid_max_max;
100extern int sysctl_drop_caches;
101extern int percpu_pagelist_fraction;
102extern int compat_log;
103extern int latencytop_enabled;
104extern int sysctl_nr_open_min, sysctl_nr_open_max;
105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
108#ifdef CONFIG_BLOCK
109extern int blk_iopoll_enabled;
110#endif
111
112/* Constants used for minimum and maximum */
113#ifdef CONFIG_LOCKUP_DETECTOR
114static int sixty = 60;
115static int neg_one = -1;
116#endif
117
118static int zero;
119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
121static int __maybe_unused three = 3;
122static unsigned long one_ul = 1;
123static int one_hundred = 100;
124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
127
128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
134static int min_percpu_pagelist_fract = 8;
135
136static int ngroups_max = NGROUPS_MAX;
137
138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
141#ifdef CONFIG_SPARC
142#include <asm/system.h>
143#endif
144
145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
154#ifdef CONFIG_S390
155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
159extern int spin_retry;
160#endif
161
162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
164extern int unaligned_dump_stack;
165#endif
166
167#ifdef CONFIG_PROC_SYSCTL
168static int proc_do_cad_pid(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170static int proc_taint(struct ctl_table *table, int write,
171 void __user *buffer, size_t *lenp, loff_t *ppos);
172#endif
173
174#ifdef CONFIG_PRINTK
175static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177#endif
178
179#ifdef CONFIG_MAGIC_SYSRQ
180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182
183static int sysrq_sysctl_handler(ctl_table *table, int write,
184 void __user *buffer, size_t *lenp,
185 loff_t *ppos)
186{
187 int error;
188
189 error = proc_dointvec(table, write, buffer, lenp, ppos);
190 if (error)
191 return error;
192
193 if (write)
194 sysrq_toggle_support(__sysrq_enabled);
195
196 return 0;
197}
198
199#endif
200
201static struct ctl_table root_table[];
202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
204 {{.count = 1,
205 .ctl_table = root_table,
206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 .root = &sysctl_table_root,
208 .set = &sysctl_table_root.default_set,
209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213};
214
215static struct ctl_table kern_table[];
216static struct ctl_table vm_table[];
217static struct ctl_table fs_table[];
218static struct ctl_table debug_table[];
219static struct ctl_table dev_table[];
220extern struct ctl_table random_table[];
221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
229/* The default sysctl tables: */
230
231static struct ctl_table root_table[] = {
232 {
233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
242 {
243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
257 { }
258};
259
260#ifdef CONFIG_SCHED_DEBUG
261static int min_sched_granularity_ns = 100000; /* 100 usecs */
262static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
263static int min_wakeup_granularity_ns; /* 0 usecs */
264static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267#endif
268
269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
274static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282#ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 {
311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
318 },
319 {
320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_shares_window",
342 .data = &sysctl_sched_shares_window,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec_minmax,
353 .extra1 = &zero,
354 .extra2 = &one,
355 },
356#endif
357 {
358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = sched_rt_handler,
363 },
364 {
365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = sched_rt_handler,
370 },
371#ifdef CONFIG_SCHED_AUTOGROUP
372 {
373 .procname = "sched_autogroup_enabled",
374 .data = &sysctl_sched_autogroup_enabled,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec_minmax,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
382#ifdef CONFIG_PROVE_LOCKING
383 {
384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec,
389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399#endif
400 {
401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 {
408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = proc_dostring,
420 },
421 {
422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428#ifdef CONFIG_PROC_SYSCTL
429 {
430 .procname = "tainted",
431 .maxlen = sizeof(long),
432 .mode = 0644,
433 .proc_handler = proc_taint,
434 },
435#endif
436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
442 .proc_handler = proc_dointvec,
443 },
444#endif
445#ifdef CONFIG_BLK_DEV_INITRD
446 {
447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454 {
455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461#ifdef CONFIG_SPARC
462 {
463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
467 .proc_handler = proc_dostring,
468 },
469 {
470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483#endif
484#ifdef CONFIG_SPARC64
485 {
486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef __hppa__
494 {
495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 {
502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508#endif
509 {
510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516#ifdef CONFIG_FUNCTION_TRACER
517 {
518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = ftrace_enable_sysctl,
523 },
524#endif
525#ifdef CONFIG_STACK_TRACER
526 {
527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = stack_trace_sysctl,
532 },
533#endif
534#ifdef CONFIG_TRACING
535 {
536 .procname = "ftrace_dump_on_oops",
537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#endif
543#ifdef CONFIG_MODULES
544 {
545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
549 .proc_handler = proc_dostring,
550 },
551 {
552 .procname = "modules_disabled",
553 .data = &modules_disabled,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 /* only handle a transition from default "0" to "1" */
557 .proc_handler = proc_dointvec_minmax,
558 .extra1 = &one,
559 .extra2 = &one,
560 },
561#endif
562#ifdef CONFIG_HOTPLUG
563 {
564 .procname = "hotplug",
565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
567 .mode = 0644,
568 .proc_handler = proc_dostring,
569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
577 .proc_handler = proc_dointvec,
578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588#endif
589#ifdef CONFIG_MAGIC_SYSRQ
590 {
591 .procname = "sysrq",
592 .data = &__sysrq_enabled,
593 .maxlen = sizeof (int),
594 .mode = 0644,
595 .proc_handler = sysrq_sysctl_handler,
596 },
597#endif
598#ifdef CONFIG_PROC_SYSCTL
599 {
600 .procname = "cad_pid",
601 .data = NULL,
602 .maxlen = sizeof (int),
603 .mode = 0600,
604 .proc_handler = proc_do_cad_pid,
605 },
606#endif
607 {
608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec,
613 },
614 {
615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
624 {
625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec_minmax,
630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
642#ifdef CONFIG_S390
643#ifdef CONFIG_MATHEMU
644 {
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651#endif
652 {
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
658 },
659#endif
660 {
661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676#if defined CONFIG_PRINTK
677 {
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
683 },
684 {
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
690 },
691 {
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
697 },
698 {
699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
707 {
708 .procname = "dmesg_restrict",
709 .data = &dmesg_restrict,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dmesg_restrict,
722 .extra1 = &zero,
723 .extra2 = &two,
724 },
725#endif
726 {
727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733#if defined(CONFIG_LOCKUP_DETECTOR)
734 {
735 .procname = "watchdog",
736 .data = &watchdog_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
742 },
743 {
744 .procname = "watchdog_thresh",
745 .data = &watchdog_thresh,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dowatchdog,
749 .extra1 = &neg_one,
750 .extra2 = &sixty,
751 },
752 {
753 .procname = "softlockup_panic",
754 .data = &softlockup_panic,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &zero,
759 .extra2 = &one,
760 },
761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
769 },
770#endif
771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 {
773 .procname = "unknown_nmi_panic",
774 .data = &unknown_nmi_panic,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779#endif
780#if defined(CONFIG_X86)
781 {
782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
794 },
795 {
796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
807 .proc_handler = proc_dointvec,
808 },
809 {
810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 {
817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec,
822 },
823#endif
824#if defined(CONFIG_MMU)
825 {
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832#endif
833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841#endif
842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 {
844 .procname = "acpi_video_flags",
845 .data = &acpi_realmode_flags,
846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
848 .proc_handler = proc_doulongvec_minmax,
849 },
850#endif
851#ifdef CONFIG_IA64
852 {
853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
864 .proc_handler = proc_dointvec,
865 },
866#endif
867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec_minmax,
874 .extra1 = &zero,
875 .extra2 = &one,
876 },
877 {
878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
880 .maxlen = sizeof(unsigned long),
881 .mode = 0644,
882 .proc_handler = proc_doulongvec_minmax,
883 },
884 {
885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
887 .maxlen = sizeof(unsigned long),
888 .mode = 0644,
889 .proc_handler = proc_dohung_task_timeout_secs,
890 },
891 {
892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
894 .maxlen = sizeof(unsigned long),
895 .mode = 0644,
896 .proc_handler = proc_doulongvec_minmax,
897 },
898#endif
899#ifdef CONFIG_COMPAT
900 {
901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec,
906 },
907#endif
908#ifdef CONFIG_RT_MUTEXES
909 {
910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916#endif
917 {
918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
922 .proc_handler = proc_dostring,
923 },
924#ifdef CONFIG_KEYS
925 {
926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939#endif
940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
947 {
948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
965 .mode = 0644,
966 .proc_handler = perf_proc_update_handler,
967 },
968#endif
969#ifdef CONFIG_KMEMCHECK
970 {
971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
975 .proc_handler = proc_dointvec,
976 },
977#endif
978#ifdef CONFIG_BLOCK
979 {
980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = proc_dointvec,
985 },
986#endif
987 { }
988};
989
990static struct ctl_table vm_table[] = {
991 {
992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
999 },
1000 {
1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
1008 },
1009 {
1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 {
1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1037 },
1038 {
1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
1043 .proc_handler = dirty_background_ratio_handler,
1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
1047 {
1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
1052 .proc_handler = dirty_background_bytes_handler,
1053 .extra1 = &one_ul,
1054 },
1055 {
1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
1060 .proc_handler = dirty_ratio_handler,
1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
1064 {
1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
1069 .proc_handler = dirty_bytes_handler,
1070 .extra1 = &dirty_bytes_min,
1071 },
1072 {
1073 .procname = "dirty_writeback_centisecs",
1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1076 .mode = 0644,
1077 .proc_handler = dirty_writeback_centisecs_handler,
1078 },
1079 {
1080 .procname = "dirty_expire_centisecs",
1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1083 .mode = 0644,
1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec_minmax,
1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
1104 {
1105 .procname = "nr_hugepages",
1106 .data = NULL,
1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
1109 .proc_handler = hugetlb_sysctl_handler,
1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
1112 },
1113#ifdef CONFIG_NUMA
1114 {
1115 .procname = "nr_hugepages_mempolicy",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#endif
1124 {
1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
1129 .proc_handler = proc_dointvec,
1130 },
1131 {
1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
1136 .proc_handler = hugetlb_treat_movable_handler,
1137 },
1138 {
1139 .procname = "nr_overcommit_hugepages",
1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
1142 .mode = 0644,
1143 .proc_handler = hugetlb_overcommit_handler,
1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
1146 },
1147#endif
1148 {
1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1154 },
1155 {
1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
1161 .extra1 = &one,
1162 .extra2 = &three,
1163 },
1164#ifdef CONFIG_COMPACTION
1165 {
1166 .procname = "compact_memory",
1167 .data = &sysctl_compact_memory,
1168 .maxlen = sizeof(int),
1169 .mode = 0200,
1170 .proc_handler = sysctl_compaction_handler,
1171 },
1172 {
1173 .procname = "extfrag_threshold",
1174 .data = &sysctl_extfrag_threshold,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = sysctl_extfrag_handler,
1178 .extra1 = &min_extfrag_threshold,
1179 .extra2 = &max_extfrag_threshold,
1180 },
1181
1182#endif /* CONFIG_COMPACTION */
1183 {
1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
1188 .proc_handler = min_free_kbytes_sysctl_handler,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1199#ifdef CONFIG_MMU
1200 {
1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = &zero,
1207 },
1208#else
1209 {
1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = &zero,
1216 },
1217#endif
1218 {
1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_jiffies,
1224 },
1225 {
1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 .extra1 = &zero,
1232 },
1233 {
1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec,
1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
1247 .proc_handler = proc_dointvec,
1248 .extra1 = &zero,
1249 },
1250#endif
1251#ifdef CONFIG_NUMA
1252 {
1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260 {
1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
1269 {
1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
1278#endif
1279#ifdef CONFIG_SMP
1280 {
1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_jiffies,
1286 },
1287#endif
1288#ifdef CONFIG_MMU
1289 {
1290 .procname = "mmap_min_addr",
1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
1293 .mode = 0644,
1294 .proc_handler = mmap_min_addr_handler,
1295 },
1296#endif
1297#ifdef CONFIG_NUMA
1298 {
1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
1303 .proc_handler = numa_zonelist_order_handler,
1304 },
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308 {
1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 .extra1 = &zero,
1315 },
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318 {
1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
1328 {
1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
1333 .proc_handler = scan_unevictable_handler,
1334 },
1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
1355 { }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360 { }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365 {
1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
1370 .proc_handler = proc_nr_inodes,
1371 },
1372 {
1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
1377 .proc_handler = proc_nr_inodes,
1378 },
1379 {
1380 .procname = "file-nr",
1381 .data = &files_stat,
1382 .maxlen = sizeof(files_stat),
1383 .mode = 0444,
1384 .proc_handler = proc_nr_files,
1385 },
1386 {
1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
1389 .maxlen = sizeof(files_stat.max_files),
1390 .mode = 0644,
1391 .proc_handler = proc_doulongvec_minmax,
1392 },
1393 {
1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
1401 },
1402 {
1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
1407 .proc_handler = proc_nr_dentry,
1408 },
1409 {
1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec_minmax,
1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
1427#ifdef CONFIG_FILE_LOCKING
1428 {
1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec,
1434 },
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437 {
1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 },
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447 {
1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_AIO
1456 {
1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
1461 .proc_handler = proc_doulongvec_minmax,
1462 },
1463 {
1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
1468 .proc_handler = proc_doulongvec_minmax,
1469 },
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472 {
1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1485#endif
1486 {
1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_minmax,
1492 .extra1 = &zero,
1493 .extra2 = &two,
1494 },
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
1502 {
1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
1505 .maxlen = sizeof(int),
1506 .mode = 0644,
1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
1509 },
1510 { }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
1516 {
1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525 {
1526 .procname = "kprobes-optimization",
1527 .data = &sysctl_kprobes_optimization,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = proc_kprobes_optimization_handler,
1531 .extra1 = &zero,
1532 .extra2 = &one,
1533 },
1534#endif
1535 { }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539 { }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547 if (unlikely(p->unregistering))
1548 return 0;
1549 p->used++;
1550 return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556 if (!--p->used)
1557 if (unlikely(p->unregistering))
1558 complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564 /*
1565 * if p->used is 0, nobody will ever touch that entry again;
1566 * we'll eliminate all paths to it before dropping sysctl_lock
1567 */
1568 if (unlikely(p->used)) {
1569 struct completion wait;
1570 init_completion(&wait);
1571 p->unregistering = &wait;
1572 spin_unlock(&sysctl_lock);
1573 wait_for_completion(&wait);
1574 spin_lock(&sysctl_lock);
1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
1578 }
1579 /*
1580 * do not remove from the list until nobody holds it; walking the
1581 * list in do_sysctl() relies on that.
1582 */
1583 list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588 spin_lock(&sysctl_lock);
1589 head->count++;
1590 spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595 spin_lock(&sysctl_lock);
1596 if (!--head->count)
1597 kfree_rcu(head, rcu);
1598 spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603 if (!head)
1604 BUG();
1605 spin_lock(&sysctl_lock);
1606 if (!use_table(head))
1607 head = ERR_PTR(-ENOENT);
1608 spin_unlock(&sysctl_lock);
1609 return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614 if (!head)
1615 return;
1616 spin_lock(&sysctl_lock);
1617 unuse_table(head);
1618 spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624 struct ctl_table_set *set = &root->default_set;
1625 if (root->lookup)
1626 set = root->lookup(root, namespaces);
1627 return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638 struct ctl_table_header *prev)
1639{
1640 struct ctl_table_root *root;
1641 struct list_head *header_list;
1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
1644
1645 spin_lock(&sysctl_lock);
1646 if (prev) {
1647 head = prev;
1648 tmp = &prev->ctl_entry;
1649 unuse_table(prev);
1650 goto next;
1651 }
1652 tmp = &root_table_header.ctl_entry;
1653 for (;;) {
1654 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656 if (!use_table(head))
1657 goto next;
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 next:
1661 root = head->root;
1662 tmp = tmp->next;
1663 header_list = lookup_header_list(root, namespaces);
1664 if (tmp != header_list)
1665 continue;
1666
1667 do {
1668 root = list_entry(root->root_list.next,
1669 struct ctl_table_root, root_list);
1670 if (root == &sysctl_table_root)
1671 goto out;
1672 header_list = lookup_header_list(root, namespaces);
1673 } while (list_empty(header_list));
1674 tmp = header_list->next;
1675 }
1676out:
1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683 return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688 spin_lock(&sysctl_lock);
1689 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690 spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700 if (!current_euid())
1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705 return 0;
1706 return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711 int mode;
1712
1713 if (root->permissions)
1714 mode = root->permissions(root, current->nsproxy, table);
1715 else
1716 mode = table->mode;
1717
1718 return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723 for (; table->procname; table++) {
1724 table->parent = parent;
1725 if (table->child)
1726 sysctl_set_parent(table, table->child);
1727 }
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732 sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734 sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
1743{
1744 struct ctl_table *p;
1745 const char *s = branch->procname;
1746
1747 /* branch should have named subdirectory as its first element */
1748 if (!s || !branch->child)
1749 return NULL;
1750
1751 /* ... and nothing else */
1752 if (branch[1].procname)
1753 return NULL;
1754
1755 /* table should contain subdirectory with the same name */
1756 for (p = table; p->procname; p++) {
1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
1760 return p;
1761 }
1762 return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769 struct ctl_table *next;
1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
1773 while ((next = is_branch_in(by, to)) != NULL) {
1774 if (by == q->attached_by)
1775 is_better = 1;
1776 if (to == p->attached_by)
1777 not_in_parent = 1;
1778 by = by->child;
1779 to = next->child;
1780 }
1781
1782 if (is_better && not_in_parent) {
1783 q->attached_by = by;
1784 q->attached_to = to;
1785 q->parent = p;
1786 }
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 * enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 * %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table. The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler). Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843 struct ctl_table_root *root,
1844 struct nsproxy *namespaces,
1845 const struct ctl_path *path, struct ctl_table *table)
1846{
1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
1850 struct ctl_table_set *set;
1851
1852 /* Count the path components */
1853 for (npath = 0; path[npath].procname; ++npath)
1854 ;
1855
1856 /*
1857 * For each path component, allocate a 2-element ctl_table array.
1858 * The first array element will be filled with the sysctl entry
1859 * for this, the second will be the sentinel (procname == 0).
1860 *
1861 * We allocate everything in one go so that we don't have to
1862 * worry about freeing additional memory in unregister_sysctl_table.
1863 */
1864 header = kzalloc(sizeof(struct ctl_table_header) +
1865 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866 if (!header)
1867 return NULL;
1868
1869 new = (struct ctl_table *) (header + 1);
1870
1871 /* Now connect the dots */
1872 prevp = &header->ctl_table;
1873 for (n = 0; n < npath; ++n, ++path) {
1874 /* Copy the procname */
1875 new->procname = path->procname;
1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
1884 header->ctl_table_arg = table;
1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
1889 header->root = root;
1890 sysctl_set_parent(NULL, header->ctl_table);
1891 header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
1894 kfree(header);
1895 return NULL;
1896 }
1897#endif
1898 spin_lock(&sysctl_lock);
1899 header->set = lookup_header_set(root, namespaces);
1900 header->attached_by = header->ctl_table;
1901 header->attached_to = root_table;
1902 header->parent = &root_table_header;
1903 for (set = header->set; set; set = set->parent) {
1904 struct ctl_table_header *p;
1905 list_for_each_entry(p, &set->list, ctl_entry) {
1906 if (p->unregistering)
1907 continue;
1908 try_attach(p, header);
1909 }
1910 }
1911 header->parent->count++;
1912 list_add_tail(&header->ctl_entry, &header->set->list);
1913 spin_unlock(&sysctl_lock);
1914
1915 return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929 struct ctl_table *table)
1930{
1931 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932 path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946 static const struct ctl_path null_path[] = { {} };
1947
1948 return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960 might_sleep();
1961
1962 if (header == NULL)
1963 return;
1964
1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
1967 if (!--header->parent->count) {
1968 WARN_ON(1);
1969 kfree_rcu(header->parent, rcu);
1970 }
1971 if (!--header->count)
1972 kfree_rcu(header, rcu);
1973 spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978 struct ctl_table_set *set = p->set;
1979 int res;
1980 spin_lock(&sysctl_lock);
1981 if (p->unregistering)
1982 res = 0;
1983 else if (!set->is_seen)
1984 res = 1;
1985 else
1986 res = set->is_seen(set);
1987 spin_unlock(&sysctl_lock);
1988 return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992 struct ctl_table_set *parent,
1993 int (*is_seen)(struct ctl_table_set *))
1994{
1995 INIT_LIST_HEAD(&p->list);
1996 p->parent = parent ? parent : &sysctl_table_root.default_set;
1997 p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003 return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017 struct ctl_table_set *parent,
2018 int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035 void __user *buffer,
2036 size_t *lenp, loff_t *ppos)
2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
2041
2042 if (!data || !maxlen || !*lenp) {
2043 *lenp = 0;
2044 return 0;
2045 }
2046
2047 if (write) {
2048 len = 0;
2049 p = buffer;
2050 while (len < *lenp) {
2051 if (get_user(c, p++))
2052 return -EFAULT;
2053 if (c == 0 || c == '\n')
2054 break;
2055 len++;
2056 }
2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
2060 return -EFAULT;
2061 ((char *) data)[len] = 0;
2062 *ppos += *lenp;
2063 } else {
2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
2079 if(copy_to_user(buffer, data, len))
2080 return -EFAULT;
2081 if (len < *lenp) {
2082 if(put_user('\n', ((char __user *) buffer) + len))
2083 return -EFAULT;
2084 len++;
2085 }
2086 *lenp = len;
2087 *ppos += len;
2088 }
2089 return 0;
2090}
2091
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112 return _proc_do_string(table->data, table->maxlen, write,
2113 buffer, lenp, ppos);
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118 size_t ret;
2119 char *tmp = skip_spaces(*buf);
2120 ret = tmp - *buf;
2121 *buf = tmp;
2122 return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127 while (*size) {
2128 if (**buf != v)
2129 break;
2130 (*size)--;
2131 (*buf)++;
2132 }
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153 unsigned long *val, bool *neg,
2154 const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156 int len;
2157 char *p, tmp[TMPBUFLEN];
2158
2159 if (!*size)
2160 return -EINVAL;
2161
2162 len = *size;
2163 if (len > TMPBUFLEN - 1)
2164 len = TMPBUFLEN - 1;
2165
2166 memcpy(tmp, *buf, len);
2167
2168 tmp[len] = 0;
2169 p = tmp;
2170 if (*p == '-' && *size > 1) {
2171 *neg = true;
2172 p++;
2173 } else
2174 *neg = false;
2175 if (!isdigit(*p))
2176 return -EINVAL;
2177
2178 *val = simple_strtoul(p, &p, 0);
2179
2180 len = p - tmp;
2181
2182 /* We don't know if the next char is whitespace thus we may accept
2183 * invalid integers (e.g. 1234...a) or two integers instead of one
2184 * (e.g. 123...1). So lets not allow such large numbers. */
2185 if (len == TMPBUFLEN - 1)
2186 return -EINVAL;
2187
2188 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189 return -EINVAL;
2190
2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212 bool neg)
2213{
2214 int len;
2215 char tmp[TMPBUFLEN], *p = tmp;
2216
2217 sprintf(p, "%s%lu", neg ? "-" : "", val);
2218 len = strlen(tmp);
2219 if (len > *size)
2220 len = *size;
2221 if (copy_to_user(*buf, tmp, len))
2222 return -EFAULT;
2223 *size -= len;
2224 *buf += len;
2225 return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231 if (*size) {
2232 char __user **buffer = (char __user **)buf;
2233 if (put_user(c, *buffer))
2234 return -EFAULT;
2235 (*size)--, (*buffer)++;
2236 *buf = *buffer;
2237 }
2238 return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242 int *valp,
2243 int write, void *data)
2244{
2245 if (write) {
2246 *valp = *negp ? -*lvalp : *lvalp;
2247 } else {
2248 int val = *valp;
2249 if (val < 0) {
2250 *negp = true;
2251 *lvalp = (unsigned long)-val;
2252 } else {
2253 *negp = false;
2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263 int write, void __user *buffer,
2264 size_t *lenp, loff_t *ppos,
2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266 int write, void *data),
2267 void *data)
2268{
2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
2273
2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275 *lenp = 0;
2276 return 0;
2277 }
2278
2279 i = (int *) tbl_data;
2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
2286 if (write) {
2287 if (left > PAGE_SIZE - 1)
2288 left = PAGE_SIZE - 1;
2289 page = __get_free_page(GFP_TEMPORARY);
2290 kbuf = (char *) page;
2291 if (!kbuf)
2292 return -ENOMEM;
2293 if (copy_from_user(kbuf, buffer, left)) {
2294 err = -EFAULT;
2295 goto free;
2296 }
2297 kbuf[left] = 0;
2298 }
2299
2300 for (; left && vleft--; i++, first=0) {
2301 unsigned long lval;
2302 bool neg;
2303
2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
2306
2307 if (!left)
2308 break;
2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
2313 break;
2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
2316 break;
2317 }
2318 } else {
2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
2323 if (!first)
2324 err = proc_put_char(&buffer, &left, '\t');
2325 if (err)
2326 break;
2327 err = proc_put_long(&buffer, &left, lval, neg);
2328 if (err)
2329 break;
2330 }
2331 }
2332
2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
2335 if (write && !err && left)
2336 left -= proc_skip_spaces(&kbuf);
2337free:
2338 if (write) {
2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
2342 }
2343 *lenp -= left;
2344 *ppos += *lenp;
2345 return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349 void __user *buffer, size_t *lenp, loff_t *ppos,
2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351 int write, void *data),
2352 void *data)
2353{
2354 return __do_proc_dointvec(table->data, table, write,
2355 buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string.
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375 NULL,NULL);
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
2388
2389 if (write && !capable(CAP_SYS_ADMIN))
2390 return -EPERM;
2391
2392 t = *table;
2393 t.data = &tmptaint;
2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395 if (err < 0)
2396 return err;
2397
2398 if (write) {
2399 /*
2400 * Poor man's atomic or. Not worth adding a primitive
2401 * to everyone's atomic.h for this
2402 */
2403 int i;
2404 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405 if ((tmptaint >> i) & 1)
2406 add_taint(i);
2407 }
2408 }
2409
2410 return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417 if (write && !capable(CAP_SYS_ADMIN))
2418 return -EPERM;
2419
2420 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432{
2433 struct do_proc_dointvec_minmax_conv_param *param = data;
2434 if (write) {
2435 int val = *negp ? -*lvalp : *lvalp;
2436 if ((param->min && *param->min > val) ||
2437 (param->max && *param->max < val))
2438 return -EINVAL;
2439 *valp = val;
2440 } else {
2441 int val = *valp;
2442 if (val < 0) {
2443 *negp = true;
2444 *lvalp = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 *lvalp = (unsigned long)val;
2448 }
2449 }
2450 return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 struct do_proc_dointvec_minmax_conv_param param = {
2473 .min = (int *) table->extra1,
2474 .max = (int *) table->extra2,
2475 };
2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477 do_proc_dointvec_minmax_conv, ¶m);
2478}
2479
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
2486 unsigned long *i, *min, *max;
2487 int vleft, first = 1, err = 0;
2488 unsigned long page = 0;
2489 size_t left;
2490 char *kbuf;
2491
2492 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493 *lenp = 0;
2494 return 0;
2495 }
2496
2497 i = (unsigned long *) data;
2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
2502
2503 if (write) {
2504 if (left > PAGE_SIZE - 1)
2505 left = PAGE_SIZE - 1;
2506 page = __get_free_page(GFP_TEMPORARY);
2507 kbuf = (char *) page;
2508 if (!kbuf)
2509 return -ENOMEM;
2510 if (copy_from_user(kbuf, buffer, left)) {
2511 err = -EFAULT;
2512 goto free;
2513 }
2514 kbuf[left] = 0;
2515 }
2516
2517 for (; left && vleft--; i++, first = 0) {
2518 unsigned long val;
2519
2520 if (write) {
2521 bool neg;
2522
2523 left -= proc_skip_spaces(&kbuf);
2524
2525 err = proc_get_long(&kbuf, &left, &val, &neg,
2526 proc_wspace_sep,
2527 sizeof(proc_wspace_sep), NULL);
2528 if (err)
2529 break;
2530 if (neg)
2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
2536 val = convdiv * (*i) / convmul;
2537 if (!first)
2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
2542 }
2543 }
2544
2545 if (!write && !first && left && !err)
2546 err = proc_put_char(&buffer, &left, '\n');
2547 if (write && !err)
2548 left -= proc_skip_spaces(&kbuf);
2549free:
2550 if (write) {
2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
2554 }
2555 *lenp -= left;
2556 *ppos += *lenp;
2557 return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer,
2562 size_t *lenp, loff_t *ppos,
2563 unsigned long convmul,
2564 unsigned long convdiv)
2565{
2566 return __do_proc_doulongvec_minmax(table->data, table, write,
2567 buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
2613 return do_proc_doulongvec_minmax(table, write, buffer,
2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
2630 *negp = true;
2631 lval = (unsigned long)-val;
2632 } else {
2633 *negp = false;
2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
2653 *negp = true;
2654 lval = (unsigned long)-val;
2655 } else {
2656 *negp = false;
2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 if (write) {
2669 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2670 } else {
2671 int val = *valp;
2672 unsigned long lval;
2673 if (val < 0) {
2674 *negp = true;
2675 lval = (unsigned long)-val;
2676 } else {
2677 *negp = false;
2678 lval = (unsigned long)val;
2679 }
2680 *lvalp = jiffies_to_msecs(lval);
2681 }
2682 return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string.
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704 do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string.
2717 * The values read are assumed to be in 1/USER_HZ seconds, and
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
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 do_proc_dointvec(table,write,buffer,lenp,ppos,
2726 do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string.
2740 * The values read are assumed to be in 1/1000 seconds, and
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
2759 tmp = pid_vnr(cad_pid);
2760
2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
2762 lenp, ppos, NULL, NULL);
2763 if (r || !write)
2764 return r;
2765
2766 new_pid = find_get_pid(tmp);
2767 if (!new_pid)
2768 return -ESRCH;
2769
2770 put_pid(xchg(&cad_pid, new_pid));
2771 return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794 int err = 0;
2795 bool first = 1;
2796 size_t left = *lenp;
2797 unsigned long bitmap_len = table->maxlen;
2798 unsigned long *bitmap = (unsigned long *) table->data;
2799 unsigned long *tmp_bitmap = NULL;
2800 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802 if (!bitmap_len || !left || (*ppos && !write)) {
2803 *lenp = 0;
2804 return 0;
2805 }
2806
2807 if (write) {
2808 unsigned long page = 0;
2809 char *kbuf;
2810
2811 if (left > PAGE_SIZE - 1)
2812 left = PAGE_SIZE - 1;
2813
2814 page = __get_free_page(GFP_TEMPORARY);
2815 kbuf = (char *) page;
2816 if (!kbuf)
2817 return -ENOMEM;
2818 if (copy_from_user(kbuf, buffer, left)) {
2819 free_page(page);
2820 return -EFAULT;
2821 }
2822 kbuf[left] = 0;
2823
2824 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825 GFP_KERNEL);
2826 if (!tmp_bitmap) {
2827 free_page(page);
2828 return -ENOMEM;
2829 }
2830 proc_skip_char(&kbuf, &left, '\n');
2831 while (!err && left) {
2832 unsigned long val_a, val_b;
2833 bool neg;
2834
2835 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836 sizeof(tr_a), &c);
2837 if (err)
2838 break;
2839 if (val_a >= bitmap_len || neg) {
2840 err = -EINVAL;
2841 break;
2842 }
2843
2844 val_b = val_a;
2845 if (left) {
2846 kbuf++;
2847 left--;
2848 }
2849
2850 if (c == '-') {
2851 err = proc_get_long(&kbuf, &left, &val_b,
2852 &neg, tr_b, sizeof(tr_b),
2853 &c);
2854 if (err)
2855 break;
2856 if (val_b >= bitmap_len || neg ||
2857 val_a > val_b) {
2858 err = -EINVAL;
2859 break;
2860 }
2861 if (left) {
2862 kbuf++;
2863 left--;
2864 }
2865 }
2866
2867 while (val_a <= val_b)
2868 set_bit(val_a++, tmp_bitmap);
2869
2870 first = 0;
2871 proc_skip_char(&kbuf, &left, '\n');
2872 }
2873 free_page(page);
2874 } else {
2875 unsigned long bit_a, bit_b = 0;
2876
2877 while (left) {
2878 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879 if (bit_a >= bitmap_len)
2880 break;
2881 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882 bit_a + 1) - 1;
2883
2884 if (!first) {
2885 err = proc_put_char(&buffer, &left, ',');
2886 if (err)
2887 break;
2888 }
2889 err = proc_put_long(&buffer, &left, bit_a, false);
2890 if (err)
2891 break;
2892 if (bit_a != bit_b) {
2893 err = proc_put_char(&buffer, &left, '-');
2894 if (err)
2895 break;
2896 err = proc_put_long(&buffer, &left, bit_b, false);
2897 if (err)
2898 break;
2899 }
2900
2901 first = 0; bit_b++;
2902 }
2903 if (!err)
2904 err = proc_put_char(&buffer, &left, '\n');
2905 }
2906
2907 if (!err) {
2908 if (write) {
2909 if (*ppos)
2910 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911 else
2912 memcpy(bitmap, tmp_bitmap,
2913 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914 }
2915 kfree(tmp_bitmap);
2916 *lenp -= left;
2917 *ppos += *lenp;
2918 return 0;
2919 } else {
2920 kfree(tmp_bitmap);
2921 return err;
2922 }
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return -ENOSYS;
2937}
2938
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);